diff options
| author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
|---|---|---|
| committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
| commit | 26a029d407be480d791972afb5975cf62c9360a6 (patch) | |
| tree | f435a8308119effd964b339f76abb83a57c29483 /media/ffvpx/libavutil | |
| parent | Initial commit. (diff) | |
| download | firefox-26a029d407be480d791972afb5975cf62c9360a6.tar.xz firefox-26a029d407be480d791972afb5975cf62c9360a6.zip | |
Adding upstream version 124.0.1.upstream/124.0.1
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'media/ffvpx/libavutil')
176 files changed, 54108 insertions, 0 deletions
diff --git a/media/ffvpx/libavutil/aarch64/asm.S b/media/ffvpx/libavutil/aarch64/asm.S new file mode 100644 index 0000000000..1840f9fb01 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/asm.S @@ -0,0 +1,260 @@ +/* + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#ifdef __ELF__ +# define ELF +#else +# define ELF # +#endif + +#if HAVE_AS_FUNC +# define FUNC +#else +# define FUNC # +#endif + +#ifndef __has_feature +# define __has_feature(x) 0 +#endif + +#if HAVE_AS_ARCH_DIRECTIVE + .arch AS_ARCH_LEVEL +#endif + +#if HAVE_AS_ARCHEXT_DOTPROD_DIRECTIVE +#define ENABLE_DOTPROD .arch_extension dotprod +#define DISABLE_DOTPROD .arch_extension nodotprod +#else +#define ENABLE_DOTPROD +#define DISABLE_DOTPROD +#endif + +#if HAVE_AS_ARCHEXT_I8MM_DIRECTIVE +#define ENABLE_I8MM .arch_extension i8mm +#define DISABLE_I8MM .arch_extension noi8mm +#else +#define ENABLE_I8MM +#define DISABLE_I8MM +#endif + +DISABLE_DOTPROD +DISABLE_I8MM + + +/* Support macros for + * - Armv8.3-A Pointer Authentication and + * - Armv8.5-A Branch Target Identification + * features which require emitting a .note.gnu.property section with the + * appropriate architecture-dependent feature bits set. + * + * |AARCH64_SIGN_LINK_REGISTER| and |AARCH64_VALIDATE_LINK_REGISTER| expand to + * PACIxSP and AUTIxSP, respectively. |AARCH64_SIGN_LINK_REGISTER| should be + * used immediately before saving the LR register (x30) to the stack. + * |AARCH64_VALIDATE_LINK_REGISTER| should be used immediately after restoring + * it. Note |AARCH64_SIGN_LINK_REGISTER|'s modifications to LR must be undone + * with |AARCH64_VALIDATE_LINK_REGISTER| before RET. The SP register must also + * have the same value at the two points. For example: + * + * .global f + * f: + * AARCH64_SIGN_LINK_REGISTER + * stp x29, x30, [sp, #-96]! + * mov x29, sp + * ... + * ldp x29, x30, [sp], #96 + * AARCH64_VALIDATE_LINK_REGISTER + * ret + * + * |AARCH64_VALID_CALL_TARGET| expands to BTI 'c'. Either it, or + * |AARCH64_SIGN_LINK_REGISTER|, must be used at every point that may be an + * indirect call target. In particular, all symbols exported from a file must + * begin with one of these macros. For example, a leaf function that does not + * save LR can instead use |AARCH64_VALID_CALL_TARGET|: + * + * .globl return_zero + * return_zero: + * AARCH64_VALID_CALL_TARGET + * mov x0, #0 + * ret + * + * A non-leaf function which does not immediately save LR may need both macros + * because |AARCH64_SIGN_LINK_REGISTER| appears late. For example, the function + * may jump to an alternate implementation before setting up the stack: + * + * .globl with_early_jump + * with_early_jump: + * AARCH64_VALID_CALL_TARGET + * cmp x0, #128 + * b.lt .Lwith_early_jump_128 + * AARCH64_SIGN_LINK_REGISTER + * stp x29, x30, [sp, #-96]! + * mov x29, sp + * ... + * ldp x29, x30, [sp], #96 + * AARCH64_VALIDATE_LINK_REGISTER + * ret + * + * .Lwith_early_jump_128: + * ... + * ret + * + * These annotations are only required with indirect calls. Private symbols that + * are only the target of direct calls do not require annotations. Also note + * that |AARCH64_VALID_CALL_TARGET| is only valid for indirect calls (BLR), not + * indirect jumps (BR). Indirect jumps in assembly are supported through + * |AARCH64_VALID_JUMP_TARGET|. Landing Pads which shall serve for jumps and + * calls can be created using |AARCH64_VALID_JUMP_CALL_TARGET|. + * + * Although not necessary, it is safe to use these macros in 32-bit ARM + * assembly. This may be used to simplify dual 32-bit and 64-bit files. + * + * References: + * - "ELF for the Arm® 64-bit Architecture" + * https: *github.com/ARM-software/abi-aa/blob/master/aaelf64/aaelf64.rst + * - "Providing protection for complex software" + * https://developer.arm.com/architectures/learn-the-architecture/providing-protection-for-complex-software + */ +#if defined(__ARM_FEATURE_BTI_DEFAULT) && (__ARM_FEATURE_BTI_DEFAULT == 1) +# define GNU_PROPERTY_AARCH64_BTI (1 << 0) // Has BTI +# define AARCH64_VALID_CALL_TARGET hint #34 // BTI 'c' +# define AARCH64_VALID_JUMP_TARGET hint #38 // BTI 'j' +#else +# define GNU_PROPERTY_AARCH64_BTI 0 // No BTI +# define AARCH64_VALID_CALL_TARGET +# define AARCH64_VALID_JUMP_TARGET +#endif + +#if defined(__ARM_FEATURE_PAC_DEFAULT) +# if ((__ARM_FEATURE_PAC_DEFAULT & (1 << 0)) != 0) // authentication using key A +# define AARCH64_SIGN_LINK_REGISTER paciasp +# define AARCH64_VALIDATE_LINK_REGISTER autiasp +# elif ((__ARM_FEATURE_PAC_DEFAULT & (1 << 1)) != 0) // authentication using key B +# define AARCH64_SIGN_LINK_REGISTER pacibsp +# define AARCH64_VALIDATE_LINK_REGISTER autibsp +# else +# error Pointer authentication defines no valid key! +# endif +# if ((__ARM_FEATURE_PAC_DEFAULT & (1 << 2)) != 0) +# error Authentication of leaf functions is enabled but not supported in FFmpeg! +# endif +# define GNU_PROPERTY_AARCH64_PAC (1 << 1) +#else +# define GNU_PROPERTY_AARCH64_PAC 0 +# define AARCH64_SIGN_LINK_REGISTER +# define AARCH64_VALIDATE_LINK_REGISTER +#endif + + +#if (GNU_PROPERTY_AARCH64_BTI != 0 || GNU_PROPERTY_AARCH64_PAC != 0) && defined(__ELF__) + .pushsection .note.gnu.property, "a" + .balign 8 + .long 4 + .long 0x10 + .long 0x5 + .asciz "GNU" + .long 0xc0000000 /* GNU_PROPERTY_AARCH64_FEATURE_1_AND */ + .long 4 + .long (GNU_PROPERTY_AARCH64_BTI | GNU_PROPERTY_AARCH64_PAC) + .long 0 + .popsection +#endif + +.macro function name, export=0, align=2 + .macro endfunc +ELF .size \name, . - \name +FUNC .endfunc + .purgem endfunc + .endm + .text + .align \align + .if \export + .global EXTERN_ASM\name +ELF .type EXTERN_ASM\name, %function +FUNC .func EXTERN_ASM\name +EXTERN_ASM\name: + AARCH64_VALID_CALL_TARGET + .else +ELF .type \name, %function +FUNC .func \name +\name: + .endif +.endm + +.macro const name, align=2, relocate=0 + .macro endconst +ELF .size \name, . - \name + .purgem endconst + .endm +#if HAVE_SECTION_DATA_REL_RO +.if \relocate + .section .data.rel.ro +.else + .section .rodata +.endif +#elif defined(_WIN32) + .section .rdata +#elif !defined(__MACH__) + .section .rodata +#else + .const_data +#endif + .align \align +\name: +.endm + +.macro movrel rd, val, offset=0 +#if CONFIG_PIC && defined(__APPLE__) + .if \offset < 0 + adrp \rd, \val@PAGE + add \rd, \rd, \val@PAGEOFF + sub \rd, \rd, -(\offset) + .else + adrp \rd, \val+(\offset)@PAGE + add \rd, \rd, \val+(\offset)@PAGEOFF + .endif +#elif CONFIG_PIC && defined(_WIN32) + .if \offset < 0 + adrp \rd, \val + add \rd, \rd, :lo12:\val + sub \rd, \rd, -(\offset) + .else + adrp \rd, \val+(\offset) + add \rd, \rd, :lo12:\val+(\offset) + .endif +#elif CONFIG_PIC +# if __has_feature(hwaddress_sanitizer) + adrp \rd, :pg_hi21_nc:\val+(\offset) +# else + adrp \rd, \val+(\offset) +# endif + add \rd, \rd, :lo12:\val+(\offset) +#else + ldr \rd, =\val+\offset +#endif +.endm + +#define GLUE(a, b) a ## b +#define JOIN(a, b) GLUE(a, b) +#define X(s) JOIN(EXTERN_ASM, s) + +#define x18 do_not_use_x18 +#define w18 do_not_use_w18 diff --git a/media/ffvpx/libavutil/aarch64/bswap.h b/media/ffvpx/libavutil/aarch64/bswap.h new file mode 100644 index 0000000000..7abca657ba --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/bswap.h @@ -0,0 +1,56 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_AARCH64_BSWAP_H +#define AVUTIL_AARCH64_BSWAP_H + +#include <stdint.h> +#include "config.h" +#include "libavutil/attributes.h" + +#if HAVE_INLINE_ASM + +#define av_bswap16 av_bswap16 +static av_always_inline av_const unsigned av_bswap16(unsigned x) +{ + unsigned y; + + __asm__("rev16 %w0, %w1" : "=r"(y) : "r"(x)); + return y; +} + +#define av_bswap32 av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + uint32_t y; + + __asm__("rev %w0, %w1" : "=r"(y) : "r"(x)); + return y; +} + +#define av_bswap64 av_bswap64 +static av_always_inline av_const uint64_t av_bswap64(uint64_t x) +{ + uint64_t y; + + __asm__("rev %0, %1" : "=r"(y) : "r"(x)); + return y; +} + +#endif /* HAVE_INLINE_ASM */ +#endif /* AVUTIL_AARCH64_BSWAP_H */ diff --git a/media/ffvpx/libavutil/aarch64/cpu.c b/media/ffvpx/libavutil/aarch64/cpu.c new file mode 100644 index 0000000000..f27fef3992 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/cpu.c @@ -0,0 +1,127 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" +#include "config.h" + +#if (defined(__linux__) || defined(__ANDROID__)) && HAVE_GETAUXVAL +#include <stdint.h> +#include <sys/auxv.h> + +#define get_cpu_feature_reg(reg, val) \ + __asm__("mrs %0, " #reg : "=r" (val)) + +static int detect_flags(void) +{ + int flags = 0; + +#if defined(HWCAP_CPUID) && HAVE_INLINE_ASM + unsigned long hwcap = getauxval(AT_HWCAP); + // We can check for DOTPROD and I8MM using HWCAP_ASIMDDP and + // HWCAP2_I8MM too, avoiding to read the CPUID registers (which triggers + // a trap, handled by the kernel). However the HWCAP_* defines for these + // extensions are added much later than HWCAP_CPUID, so the userland + // headers might lack support for them even if the binary later is run + // on hardware that does support it (and where the kernel might support + // HWCAP_CPUID). + // See https://www.kernel.org/doc/html/latest/arm64/cpu-feature-registers.html + if (hwcap & HWCAP_CPUID) { + uint64_t tmp; + + get_cpu_feature_reg(ID_AA64ISAR0_EL1, tmp); + if (((tmp >> 44) & 0xf) == 0x1) + flags |= AV_CPU_FLAG_DOTPROD; + get_cpu_feature_reg(ID_AA64ISAR1_EL1, tmp); + if (((tmp >> 52) & 0xf) == 0x1) + flags |= AV_CPU_FLAG_I8MM; + } +#endif + + return flags; +} + +#elif defined(__APPLE__) && HAVE_SYSCTLBYNAME +#include <sys/sysctl.h> + +static int detect_flags(void) +{ + uint32_t value = 0; + size_t size; + int flags = 0; + + size = sizeof(value); + if (!sysctlbyname("hw.optional.arm.FEAT_DotProd", &value, &size, NULL, 0)) { + if (value) + flags |= AV_CPU_FLAG_DOTPROD; + } + size = sizeof(value); + if (!sysctlbyname("hw.optional.arm.FEAT_I8MM", &value, &size, NULL, 0)) { + if (value) + flags |= AV_CPU_FLAG_I8MM; + } + return flags; +} + +#elif defined(_WIN32) +#include <windows.h> + +static int detect_flags(void) +{ + int flags = 0; +#ifdef PF_ARM_V82_DP_INSTRUCTIONS_AVAILABLE + if (IsProcessorFeaturePresent(PF_ARM_V82_DP_INSTRUCTIONS_AVAILABLE)) + flags |= AV_CPU_FLAG_DOTPROD; +#endif + return flags; +} +#else + +static int detect_flags(void) +{ + return 0; +} + +#endif + +int ff_get_cpu_flags_aarch64(void) +{ + int flags = AV_CPU_FLAG_ARMV8 * HAVE_ARMV8 | + AV_CPU_FLAG_NEON * HAVE_NEON; + +#ifdef __ARM_FEATURE_DOTPROD + flags |= AV_CPU_FLAG_DOTPROD; +#endif +#ifdef __ARM_FEATURE_MATMUL_INT8 + flags |= AV_CPU_FLAG_I8MM; +#endif + + flags |= detect_flags(); + + return flags; +} + +size_t ff_get_cpu_max_align_aarch64(void) +{ + int flags = av_get_cpu_flags(); + + if (flags & AV_CPU_FLAG_NEON) + return 16; + + return 8; +} diff --git a/media/ffvpx/libavutil/aarch64/cpu.h b/media/ffvpx/libavutil/aarch64/cpu.h new file mode 100644 index 0000000000..64d703be37 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/cpu.h @@ -0,0 +1,31 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_AARCH64_CPU_H +#define AVUTIL_AARCH64_CPU_H + +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" + +#define have_armv8(flags) CPUEXT(flags, ARMV8) +#define have_neon(flags) CPUEXT(flags, NEON) +#define have_vfp(flags) CPUEXT(flags, VFP) +#define have_dotprod(flags) CPUEXT(flags, DOTPROD) +#define have_i8mm(flags) CPUEXT(flags, I8MM) + +#endif /* AVUTIL_AARCH64_CPU_H */ diff --git a/media/ffvpx/libavutil/aarch64/float_dsp_init.c b/media/ffvpx/libavutil/aarch64/float_dsp_init.c new file mode 100644 index 0000000000..4325071821 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/float_dsp_init.c @@ -0,0 +1,69 @@ +/* + * ARM NEON optimised Float DSP functions + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> + +#include "libavutil/attributes.h" +#include "libavutil/cpu.h" +#include "libavutil/float_dsp.h" +#include "cpu.h" + +void ff_vector_fmul_neon(float *dst, const float *src0, const float *src1, + int len); + +void ff_vector_fmac_scalar_neon(float *dst, const float *src, float mul, + int len); + +void ff_vector_fmul_scalar_neon(float *dst, const float *src, float mul, + int len); + +void ff_vector_dmul_scalar_neon(double *dst, const double *src, double mul, + int len); + +void ff_vector_fmul_window_neon(float *dst, const float *src0, + const float *src1, const float *win, int len); + +void ff_vector_fmul_add_neon(float *dst, const float *src0, const float *src1, + const float *src2, int len); + +void ff_vector_fmul_reverse_neon(float *dst, const float *src0, + const float *src1, int len); + +void ff_butterflies_float_neon(float *v1, float *v2, int len); + +float ff_scalarproduct_float_neon(const float *v1, const float *v2, int len); + +av_cold void ff_float_dsp_init_aarch64(AVFloatDSPContext *fdsp) +{ + int cpu_flags = av_get_cpu_flags(); + + if (have_neon(cpu_flags)) { + fdsp->butterflies_float = ff_butterflies_float_neon; + fdsp->scalarproduct_float = ff_scalarproduct_float_neon; + fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_neon; + fdsp->vector_fmul = ff_vector_fmul_neon; + fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_neon; + fdsp->vector_fmul_add = ff_vector_fmul_add_neon; + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_neon; + fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_neon; + fdsp->vector_fmul_window = ff_vector_fmul_window_neon; + } +} diff --git a/media/ffvpx/libavutil/aarch64/float_dsp_neon.S b/media/ffvpx/libavutil/aarch64/float_dsp_neon.S new file mode 100644 index 0000000000..35e2715b87 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/float_dsp_neon.S @@ -0,0 +1,202 @@ +/* + * ARM NEON optimised Float DSP functions + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * Copyright (c) 2014 Janne Grunau <janne-libav@jannau.net> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "asm.S" + +function ff_vector_fmul_neon, export=1 +1: subs w3, w3, #16 + ld1 {v0.4s, v1.4s}, [x1], #32 + ld1 {v2.4s, v3.4s}, [x1], #32 + ld1 {v4.4s, v5.4s}, [x2], #32 + ld1 {v6.4s, v7.4s}, [x2], #32 + fmul v16.4s, v0.4s, v4.4s + fmul v17.4s, v1.4s, v5.4s + fmul v18.4s, v2.4s, v6.4s + fmul v19.4s, v3.4s, v7.4s + st1 {v16.4s, v17.4s}, [x0], #32 + st1 {v18.4s, v19.4s}, [x0], #32 + b.ne 1b + ret +endfunc + +function ff_vector_fmac_scalar_neon, export=1 + mov x3, #-32 +1: subs w2, w2, #16 + ld1 {v16.4s, v17.4s}, [x0], #32 + ld1 {v18.4s, v19.4s}, [x0], x3 + ld1 {v4.4s, v5.4s}, [x1], #32 + ld1 {v6.4s, v7.4s}, [x1], #32 + fmla v16.4s, v4.4s, v0.s[0] + fmla v17.4s, v5.4s, v0.s[0] + fmla v18.4s, v6.4s, v0.s[0] + fmla v19.4s, v7.4s, v0.s[0] + st1 {v16.4s, v17.4s}, [x0], #32 + st1 {v18.4s, v19.4s}, [x0], #32 + b.ne 1b + ret +endfunc + +function ff_vector_fmul_scalar_neon, export=1 + mov w4, #15 + bics w3, w2, w4 + dup v16.4s, v0.s[0] + b.eq 3f + ld1 {v0.4s, v1.4s}, [x1], #32 +1: subs w3, w3, #16 + fmul v0.4s, v0.4s, v16.4s + ld1 {v2.4s, v3.4s}, [x1], #32 + fmul v1.4s, v1.4s, v16.4s + fmul v2.4s, v2.4s, v16.4s + st1 {v0.4s, v1.4s}, [x0], #32 + fmul v3.4s, v3.4s, v16.4s + b.eq 2f + ld1 {v0.4s, v1.4s}, [x1], #32 + st1 {v2.4s, v3.4s}, [x0], #32 + b 1b +2: ands w2, w2, #15 + st1 {v2.4s, v3.4s}, [x0], #32 + b.eq 4f +3: ld1 {v0.4s}, [x1], #16 + fmul v0.4s, v0.4s, v16.4s + st1 {v0.4s}, [x0], #16 + subs w2, w2, #4 + b.gt 3b +4: ret +endfunc + +function ff_vector_dmul_scalar_neon, export=1 + dup v16.2d, v0.d[0] + ld1 {v0.2d, v1.2d}, [x1], #32 +1: subs w2, w2, #8 + fmul v0.2d, v0.2d, v16.2d + ld1 {v2.2d, v3.2d}, [x1], #32 + fmul v1.2d, v1.2d, v16.2d + fmul v2.2d, v2.2d, v16.2d + st1 {v0.2d, v1.2d}, [x0], #32 + fmul v3.2d, v3.2d, v16.2d + ld1 {v0.2d, v1.2d}, [x1], #32 + st1 {v2.2d, v3.2d}, [x0], #32 + b.gt 1b + ret +endfunc + +function ff_vector_fmul_window_neon, export=1 + sxtw x4, w4 // len + sub x2, x2, #8 + sub x5, x4, #2 + add x2, x2, x5, lsl #2 // src1 + 4 * (len - 4) + add x6, x3, x5, lsl #3 // win + 8 * (len - 2) + add x5, x0, x5, lsl #3 // dst + 8 * (len - 2) + mov x7, #-16 + ld1 {v0.4s}, [x1], #16 // s0 + ld1 {v2.4s}, [x3], #16 // wi + ld1 {v1.4s}, [x2], x7 // s1 +1: ld1 {v3.4s}, [x6], x7 // wj + subs x4, x4, #4 + fmul v17.4s, v0.4s, v2.4s // s0 * wi + rev64 v4.4s, v1.4s + rev64 v5.4s, v3.4s + rev64 v17.4s, v17.4s + ext v4.16b, v4.16b, v4.16b, #8 // s1_r + ext v5.16b, v5.16b, v5.16b, #8 // wj_r + ext v17.16b, v17.16b, v17.16b, #8 // (s0 * wi)_rev + fmul v16.4s, v0.4s, v5.4s // s0 * wj_r + fmla v17.4s, v1.4s, v3.4s // (s0 * wi)_rev + s1 * wj + b.eq 2f + ld1 {v0.4s}, [x1], #16 + fmls v16.4s, v4.4s, v2.4s // s0 * wj_r - s1_r * wi + st1 {v17.4s}, [x5], x7 + ld1 {v2.4s}, [x3], #16 + ld1 {v1.4s}, [x2], x7 + st1 {v16.4s}, [x0], #16 + b 1b +2: + fmls v16.4s, v4.4s, v2.4s // s0 * wj_r - s1_r * wi + st1 {v17.4s}, [x5], x7 + st1 {v16.4s}, [x0], #16 + ret +endfunc + +function ff_vector_fmul_add_neon, export=1 + ld1 {v0.4s, v1.4s}, [x1], #32 + ld1 {v2.4s, v3.4s}, [x2], #32 + ld1 {v4.4s, v5.4s}, [x3], #32 +1: subs w4, w4, #8 + fmla v4.4s, v0.4s, v2.4s + fmla v5.4s, v1.4s, v3.4s + b.eq 2f + ld1 {v0.4s, v1.4s}, [x1], #32 + ld1 {v2.4s, v3.4s}, [x2], #32 + st1 {v4.4s, v5.4s}, [x0], #32 + ld1 {v4.4s, v5.4s}, [x3], #32 + b 1b +2: st1 {v4.4s, v5.4s}, [x0], #32 + ret +endfunc + +function ff_vector_fmul_reverse_neon, export=1 + sxtw x3, w3 + add x2, x2, x3, lsl #2 + sub x2, x2, #32 + mov x4, #-32 + ld1 {v2.4s, v3.4s}, [x2], x4 + ld1 {v0.4s, v1.4s}, [x1], #32 +1: subs x3, x3, #8 + rev64 v3.4s, v3.4s + rev64 v2.4s, v2.4s + ext v3.16b, v3.16b, v3.16b, #8 + ext v2.16b, v2.16b, v2.16b, #8 + fmul v16.4s, v0.4s, v3.4s + fmul v17.4s, v1.4s, v2.4s + b.eq 2f + ld1 {v2.4s, v3.4s}, [x2], x4 + ld1 {v0.4s, v1.4s}, [x1], #32 + st1 {v16.4s, v17.4s}, [x0], #32 + b 1b +2: st1 {v16.4s, v17.4s}, [x0], #32 + ret +endfunc + +function ff_butterflies_float_neon, export=1 +1: ld1 {v0.4s}, [x0] + ld1 {v1.4s}, [x1] + subs w2, w2, #4 + fsub v2.4s, v0.4s, v1.4s + fadd v3.4s, v0.4s, v1.4s + st1 {v2.4s}, [x1], #16 + st1 {v3.4s}, [x0], #16 + b.gt 1b + ret +endfunc + +function ff_scalarproduct_float_neon, export=1 + movi v2.4s, #0 +1: ld1 {v0.4s}, [x0], #16 + ld1 {v1.4s}, [x1], #16 + subs w2, w2, #4 + fmla v2.4s, v0.4s, v1.4s + b.gt 1b + faddp v0.4s, v2.4s, v2.4s + faddp s0, v0.2s + ret +endfunc diff --git a/media/ffvpx/libavutil/aarch64/moz.build b/media/ffvpx/libavutil/aarch64/moz.build new file mode 100644 index 0000000000..fa80cf8896 --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/moz.build @@ -0,0 +1,21 @@ +# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*- +# vim: set filetype=python: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +SOURCES += [ + 'cpu.c', + 'float_dsp_init.c', + 'float_dsp_neon.S', + 'tx_float_init.c', + 'tx_float_neon.S' +] + +if CONFIG['OS_ARCH'] == 'WINNT': + USE_INTEGRATED_CLANGCL_AS = True + DEFINES['EXTERN_ASM'] = '' + +FINAL_LIBRARY = 'mozavutil' + +include('/media/ffvpx/ffvpxcommon.mozbuild') diff --git a/media/ffvpx/libavutil/aarch64/timer.h b/media/ffvpx/libavutil/aarch64/timer.h new file mode 100644 index 0000000000..8b28fd354c --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/timer.h @@ -0,0 +1,50 @@ +/* + * Copyright (c) 2015 Janne Grunau <janne-libav@jannau.net> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_AARCH64_TIMER_H +#define AVUTIL_AARCH64_TIMER_H + +#include <stdint.h> +#include "config.h" + +#if defined(__APPLE__) + +#include <mach/mach_time.h> + +#define AV_READ_TIME mach_absolute_time + +#elif HAVE_INLINE_ASM + +#define AV_READ_TIME read_time + +static inline uint64_t read_time(void) +{ + uint64_t cycle_counter; + __asm__ volatile( + "isb \t\n" + "mrs %0, pmccntr_el0 " + : "=r"(cycle_counter) :: "memory" ); + + return cycle_counter; +} + +#endif /* HAVE_INLINE_ASM */ + +#endif /* AVUTIL_AARCH64_TIMER_H */ diff --git a/media/ffvpx/libavutil/aarch64/tx_float_init.c b/media/ffvpx/libavutil/aarch64/tx_float_init.c new file mode 100644 index 0000000000..8300472c4c --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/tx_float_init.c @@ -0,0 +1,64 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TX_FLOAT +#include "libavutil/tx_priv.h" +#include "libavutil/attributes.h" +#include "libavutil/aarch64/cpu.h" + +TX_DECL_FN(fft2, neon) +TX_DECL_FN(fft4_fwd, neon) +TX_DECL_FN(fft4_inv, neon) +TX_DECL_FN(fft8, neon) +TX_DECL_FN(fft8_ns, neon) +TX_DECL_FN(fft16, neon) +TX_DECL_FN(fft16_ns, neon) +TX_DECL_FN(fft32, neon) +TX_DECL_FN(fft32_ns, neon) +TX_DECL_FN(fft_sr, neon) +TX_DECL_FN(fft_sr_ns, neon) + +static av_cold int neon_init(AVTXContext *s, const FFTXCodelet *cd, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale) +{ + ff_tx_init_tabs_float(len); + if (cd->max_len == 2) + return ff_tx_gen_ptwo_revtab(s, opts); + else + return ff_tx_gen_split_radix_parity_revtab(s, len, inv, opts, 8, 0); +} + +const FFTXCodelet * const ff_tx_codelet_list_float_aarch64[] = { + TX_DEF(fft2, FFT, 2, 2, 2, 0, 128, NULL, neon, NEON, AV_TX_INPLACE, 0), + TX_DEF(fft2, FFT, 2, 2, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft4_fwd, FFT, 4, 4, 2, 0, 128, NULL, neon, NEON, AV_TX_INPLACE | FF_TX_FORWARD_ONLY, 0), + TX_DEF(fft4_fwd, FFT, 4, 4, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft4_inv, FFT, 4, 4, 2, 0, 128, NULL, neon, NEON, AV_TX_INPLACE | FF_TX_INVERSE_ONLY, 0), + TX_DEF(fft8, FFT, 8, 8, 2, 0, 128, neon_init, neon, NEON, AV_TX_INPLACE, 0), + TX_DEF(fft8_ns, FFT, 8, 8, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft16, FFT, 16, 16, 2, 0, 128, neon_init, neon, NEON, AV_TX_INPLACE, 0), + TX_DEF(fft16_ns, FFT, 16, 16, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft32, FFT, 32, 32, 2, 0, 128, neon_init, neon, NEON, AV_TX_INPLACE, 0), + TX_DEF(fft32_ns, FFT, 32, 32, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + + TX_DEF(fft_sr, FFT, 64, 131072, 2, 0, 128, neon_init, neon, NEON, 0, 0), + TX_DEF(fft_sr_ns, FFT, 64, 131072, 2, 0, 192, neon_init, neon, NEON, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + + NULL, +}; diff --git a/media/ffvpx/libavutil/aarch64/tx_float_neon.S b/media/ffvpx/libavutil/aarch64/tx_float_neon.S new file mode 100644 index 0000000000..78e4876d6c --- /dev/null +++ b/media/ffvpx/libavutil/aarch64/tx_float_neon.S @@ -0,0 +1,1294 @@ +/* + * Copyright (c) Lynne + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/aarch64/asm.S" + +/* Open `doc/transforms.md` to see the code upon which the transforms here were + * based upon. + * + * File conventions: + * GPRs: x0-x3 - arguments, untouched + * x4 - Lookup table base pointer + * x5-x6 - macro ld1 temps/function scratch + * x7-x9 - FFT table state + * x10-x17 - lookup table/macro scratch + * w19-w20 - current/target length when needed + * x21-x22 - len*2, len*6 + * + * Vectors: v0-v7 - coefficients + * v8-v15 - coefficients when needed, otherwise untouched + * v16-v30 - used as needed + * v31 - -1.0, +1.0, -1.0, +1.0. Never touched after loading. + * + * Stack: backup for v8-v15 and x19-x22 when needed, and transform lengths + */ + +#define M_SQRT1_2 0.707106781186547524401 +#define COS16_1 0.92387950420379638671875 +#define COS16_3 0.3826834261417388916015625 + +/* We only ever load this once at the start, and then live with losing an + * entire register as we need to lug this all the time everywhere. + * Clearly should be integrated into an fsadd and fmlsa, but "muh RISC!". */ +const subadd, align=4 + .float -1.0, 1.0, -1.0, 1.0 +endconst + +.macro LOAD_SUBADD + movrel x5, subadd + ld1 { v31.4s }, [x5] +.endm + +.macro SETUP_LUT no_lut=0 +.if \no_lut == 0 + ldr x4, [x0, #8] +.endif +.endm + +.macro LOAD_INPUT dst1, dst2, dst3, dst4, src, no_lut=0, discont=0 +.if \no_lut == 1 +.if \discont == 1 + ldp q\dst1\(), q\dst2\(), [\src\()] + ldp q\dst3\(), q\dst4\(), [\src\(), #32] + add \src\(), \src\(), #64 +.else + ld1 { v\dst1\().4s, v\dst2\().4s, v\dst3\().4s, v\dst4\().4s }, [\src], #64 +.endif +.else + ldp w10, w11, [x4, #0 ] + ldp w12, w13, [x4, #8 ] + ldp w14, w15, [x4, #16] + ldp w16, w17, [x4, #24] + + add x4, x4, #32 + + ldr d\dst1, [\src, x10, lsl #3] + add x11, \src, x11, lsl #3 + ldr d\dst2, [\src, x12, lsl #3] + add x13, \src, x13, lsl #3 + ldr d\dst3, [\src, x14, lsl #3] + add x15, \src, x15, lsl #3 + ldr d\dst4, [\src, x16, lsl #3] + add x17, \src, x17, lsl #3 + + ld1 { v\dst1\().d }[1], [x11] + ld1 { v\dst2\().d }[1], [x13] + ld1 { v\dst3\().d }[1], [x15] + ld1 { v\dst4\().d }[1], [x17] +.endif +.endm + +.macro FFT4 e0, o0, standalone + fadd v16.4s, \e0\().4s, \o0\().4s // r1..4 + fsub \e0\().4s, \e0\().4s, \o0\().4s // t1..4 + + rev64 v18.4s, \e0\().4s + + zip2 \o0\().2d, v16.2d, \e0\().2d + zip1 v17.2d, v16.2d, \e0\().2d + + mov \o0\().d[1], v18.d[1] + + fadd \e0\().4s, v17.4s, \o0\().4s // a1,2 b1,4 + fsub v16.4s, v17.4s, \o0\().4s // a3,4 b3,2 + + mov \o0\().16b, v16.16b // Swap once again... + mov \o0\().s[3], \e0\().s[3] + mov \e0\().s[3], v16.s[3] + +.if \standalone == 0 + uzp2 \o0\().2d, \e0\().2d, \o0\().2d + uzp1 \e0\().2d, \e0\().2d, v16.2d +.endif +.endm + +const shuf_4pt_x2, align=4 + .byte 24, 25, 26, 27 // reg2, 3 + .byte 12, 13, 14, 15 // reg1, 4 + .byte 8, 9, 10, 11 // reg1, 3 + .byte 28, 29, 30, 31 // reg2, 4 +endconst + +// Identical to FFT4, but does 2 transforms in parallel, with no deinterleaving +.macro FFT4_X2 e0, o0, e1, o1, \ + t0=v16, t1=v17, t2=v18, t3=v19, t4=v20, t5=v21, t6=v22 + + fadd \t0\().4s, \e0\().4s, \o0\().4s // r1234 + fadd \t2\().4s, \e1\().4s, \o1\().4s // r1234 + fsub \e0\().4s, \e0\().4s, \o0\().4s // t1234 + fsub \e1\().4s, \e1\().4s, \o1\().4s // t1234 + + movrel x5, shuf_4pt_x2 + + rev64 \t4\().4s, \e0\().4s + rev64 \t5\().4s, \e1\().4s + + zip2 \o0\().2d, \t0\().2d, \e0\().2d // t3,4 r3,4 + zip2 \o1\().2d, \t2\().2d, \e1\().2d // t3,4 r3,4 + + ld1 { \t6\().16b }, [x5] + + mov \o0\().d[1], \t4\().d[1] + mov \o1\().d[1], \t5\().d[1] + + zip1 \t1\().2d, \t0\().2d, \e0\().2d // t1,2 r1,2 + zip1 \t3\().2d, \t2\().2d, \e1\().2d // t1,2 r1,2 + + fsub \t4\().4s, \t1\().4s, \o0\().4s // a34 b32 + fadd \t5\().4s, \t1\().4s, \o0\().4s // a12 b14 + fsub \t2\().4s, \t3\().4s, \o1\().4s // a34 b32 + fadd \t3\().4s, \t3\().4s, \o1\().4s // a12 b14 + + // TODO: experiment with movs instead of tables here + tbl \o0\().16b, { \t4\().16b, \t5\().16b }, \t6\().16b // b1234 + tbl \o1\().16b, { \t2\().16b, \t3\().16b }, \t6\().16b // b1234 + + zip1 \e0\().2d, \t5\().2d, \t4\().2d // a1234 +// zip2 \o0\().2d, \t5\().2d, \t4\().2d // b1432 + zip1 \e1\().2d, \t3\().2d, \t2\().2d // a1234 +// zip2 \o1\().2d, \t3\().2d, \t2\().2d // b1432 +// rev64 \o0\().4s, \o0\().4s // b4123 +// rev64 \o1\().4s, \o1\().4s // b4123 +// ext \o0\().16b, \o0\().16b, \o0\().16b, #4 // b1234 +// ext \o1\().16b, \o1\().16b, \o1\().16b, #4 // b1234 +.endm + +const tab_8pt, align=4 + .float M_SQRT1_2, -M_SQRT1_2, -M_SQRT1_2, M_SQRT1_2 +endconst + +.macro FFT8 e0, e1, o0, o1, \ + t0=v16, t1=v17, t2=v18, t3=v19, t4=v20, t5=v21, t6=v22 + + movrel x5, tab_8pt + + fsub \t1\().4s, \e1\().4s, \o1\().4s // j1234 + fadd \o1\().4s, \e1\().4s, \o1\().4s // k1234 + fsub \t0\().4s, \e0\().4s, \o0\().4s // r1234 + fadd \o0\().4s, \e0\().4s, \o0\().4s // q1234 + + ld1 { \t5\().4s }, [x5] + + ext \t4\().16b, \o1\().16b, \o1\().16b, #12 + rev64 \t4\().4s, \t4\().4s + + ext \t2\().16b, \o0\().16b, \t4\().16b, #8 // o0[0,1], o1[3,2] + mov \o0\().d[1], \t4\().d[1] // o0[3, 4]; o1[1, 4] + + fsub \e1\().4s, \o0\().4s, \t2\().4s // s34, g43 + fadd \t2\().4s, \o0\().4s, \t2\().4s // s12, g12 + + rev64 \t6\().4s, v31.4s // 1, -1, 1, -1 + dup \o0\().2d, \t0\().d[0] // r1212 + dup \o1\().2d, \t0\().d[1] // r3434 + + rev64 \t4\().4s, \e1\().4s // xxg34 + rev64 \o1\().4s, \o1\().4s // r4343 + + ext \t6\().16b, v31.16b, \t6\().16b, #8 // -1, 1, 1, -1 + zip1 \t3\().2d, \t2\().2d, \e1\().2d // s1234 + zip2 \t2\().2d, \t2\().2d, \t4\().2d // g1234 + + fadd \e0\().4s, \t3\().4s, \t2\().4s // out_e1 + fsub \e1\().4s, \t3\().4s, \t2\().4s // out_e2 + + fmul \t1\().4s, \t1\().4s, \t5\().4s // j * +--+M_SQRT1_2 + fmls \o0\().4s, \o1\().4s, \t6\().4s // z1234 + + rev64 \t4\().4s, \t1\().4s // j2143 + fmla \t1\().4s, \t4\().4s, v31.4s // l2143 + + rev64 \t4\().4s, \t1\().4s // l1234 + ext \t4\().16b, \t4\().16b, \t4\().16b, #8 // l3412 + + fmla \t4\().4s, \t1\().4s, v31.4s // t1234 + + fadd \o1\().4s, \o0\().4s, \t4\().4s // out_o2 + fsub \o0\().4s, \o0\().4s, \t4\().4s // out_o1 +.endm + +// Identical as FFT8, but does 2 transforms in parallel +.macro FFT8_X2 e0, e1, o0, o1, e2, e3, o2, o3 + + movrel x5, tab_8pt + + fadd v19.4s, \e3\().4s, \o3\().4s // k1234 + fadd v17.4s, \e1\().4s, \o1\().4s // k1234 + fadd v18.4s, \e2\().4s, \o2\().4s // q1234 + fadd v16.4s, \e0\().4s, \o0\().4s // q1234 + + ld1 { v23.4s }, [x5] + + ext v22.16b, v19.16b, v19.16b, #12 + ext v21.16b, v17.16b, v17.16b, #12 + + rev64 v22.4s, v22.4s + rev64 v21.4s, v21.4s + + ext v19.16b, v18.16b, v22.16b, #8 + ext v17.16b, v16.16b, v21.16b, #8 + + mov v18.d[1], v22.d[1] + mov v21.d[0], v16.d[0] + + fadd v22.4s, v18.4s, v19.4s // s12, g12 + fsub v19.4s, v18.4s, v19.4s // s34, g43 + fsub v18.4s, v21.4s, v17.4s // s34, g43 + fadd v16.4s, v21.4s, v17.4s // s12, g12 + + fsub \e0\().4s, \e0\().4s, \o0\().4s // r1234 + fsub v20.4s, \e1\().4s, \o1\().4s // j1234 + fsub \e2\().4s, \e2\().4s, \o2\().4s // r1234 + fsub v21.4s, \e3\().4s, \o3\().4s // j1234 + + rev64 v24.4s, v31.4s // 1, -1, 1, -1 + zip1 v17.2d, v16.2d, v18.2d // s1234 + zip1 \e1\().2d, v22.2d, v19.2d // s1234 + + rev64 v18.4s, v18.4s // xxg34 + rev64 v19.4s, v19.4s // xxg34 + + zip2 v16.2d, v16.2d, v18.2d // g1234 + zip2 \e3\().2d, v22.2d, v19.2d // g1234 + + dup \o0\().2d, \e0\().d[0] // r1212 + dup \o1\().2d, \e0\().d[1] // r3434 + dup \o2\().2d, \e2\().d[0] // r1212 + dup \o3\().2d, \e2\().d[1] // r3434 + + fadd \e2\().4s, \e1\().4s, \e3\().4s // out_e1 + fsub \e3\().4s, \e1\().4s, \e3\().4s // out_e2 + fadd \e0\().4s, v17.4s, v16.4s // out_e1 + fsub \e1\().4s, v17.4s, v16.4s // out_e2 + + ext v24.16b, v31.16b, v24.16b, #8 // -1, 1, 1, -1 + rev64 \o1\().4s, \o1\().4s // r4343 + rev64 \o3\().4s, \o3\().4s // r4343 + + fmul v19.4s, v20.4s, v23.4s // j * +--+M_SQRT1_2 + fmul v21.4s, v21.4s, v23.4s // j * +--+M_SQRT1_2 + + rev64 v20.4s, v19.4s // j2143 + rev64 v18.4s, v21.4s // j2143 + + fmls \o0\().4s, \o1\().4s, v24.4s // z1234 + fmls \o2\().4s, \o3\().4s, v24.4s // z1234 + + fmla v19.4s, v20.4s, v31.4s // l2143 + fmla v21.4s, v18.4s, v31.4s // l2143 + + rev64 v20.4s, v19.4s // l1234 + rev64 v18.4s, v21.4s // l1234 + ext v20.16b, v20.16b, v20.16b, #8 // l3412 + ext v18.16b, v18.16b, v18.16b, #8 // l3412 + + fmla v20.4s, v19.4s, v31.4s // t1234 + fmla v18.4s, v21.4s, v31.4s // t1234 + + fadd \o1\().4s, \o0\().4s, v20.4s // out_o2 + fadd \o3\().4s, \o2\().4s, v18.4s // out_o2 + fsub \o0\().4s, \o0\().4s, v20.4s // out_o1 + fsub \o2\().4s, \o2\().4s, v18.4s // out_o1 +.endm + +const tab_16pt, align=4 + .float -COS16_1, COS16_1, -COS16_3, COS16_3 // Could be +-+- too + .float COS16_3, COS16_3, COS16_1, COS16_1 + .float 1.0, 1.0, M_SQRT1_2, M_SQRT1_2 +endconst + +// 16-point FFT +// t3, t4, t5, t6 must be sequential +.macro FFT16 e0, e1, e2, e3, o0, o1, o2, o3, \ + t0=v16, t1=v17, t2=v18, t3=v19, t4=v20, t5=v21, t6=v22 + + FFT8 \e0, \e1, \e2, \e3, \t0, \t1, \t2, \t3, \t4, \t5, \t6 + FFT4_X2 \o0, \o1, \o2, \o3, \t0, \t1, \t2, \t3, \t4, \t5, \t6 + + movrel x5, tab_16pt + + rev64 \t0\().4s, \o0\().4s // z[ 8, 9].imre + rev64 \t1\().4s, \o2\().4s // z[10,11].imre + + ins \t0\().d[0], xzr + ins \t1\().d[0], xzr + + ld1 { \t4\().4s, \t5\().4s, \t6\().4s }, [x5] + // TODO: We could derive \t4\() or \t5\() from either, but it seems cheaper to load + + fmla \o2\().4s, \t1\().4s, v31.4s // s[4567] + fmls \o0\().4s, \t0\().4s, v31.4s // s[0123] + + fmul \t2\().4s, \o1\().4s, \t4\().4s + fmul \t3\().4s, \o3\().4s, \t4\().4s + + rev64 \o3\().4s, \o3\().4s + rev64 \o1\().4s, \o1\().4s + + fmla \t3\().4s, \o3\().4s, \t5\().4s // s[12, 13, 14, 15] + fmls \t2\().4s, \o1\().4s, \t5\().4s // s[ 8, 9, 10, 11] + + fmul \t1\().4s, \o2\().4s, \t6\().4s // s[4567] * mult + fmul \t0\().4s, \o0\().4s, \t6\().4s // s[0123] * mult + + mov \o1\().16b, \t3\().16b + mov \o2\().16b, \t1\().16b + + fsub \t3\().4s, \t3\().4s, \t2\().4s // y34, u34 + fsub \t1\().4s, \t1\().4s, \t0\().4s // w34, x34 + + fadd \t2\().4s, \t2\().4s, \o1\().4s // y56, u56 + rev64 \t3\().4s, \t3\().4s + fadd \t0\().4s, \t0\().4s, \o2\().4s // w56, x56 + rev64 \t1\().4s, \t1\().4s + + fmul \t2\().4s, \t2\().4s, v31.4s + fmul \t1\().4s, \t1\().4s, v31.4s + + fadd \o3\().4s, \e3\().4s, \t3\().4s + fsub \o2\().4s, \e3\().4s, \t3\().4s + fsub \o1\().4s, \e2\().4s, \t2\().4s + fadd \o0\().4s, \e2\().4s, \t2\().4s + + fsub \e2\().4s, \e0\().4s, \t0\().4s + fadd \e0\().4s, \e0\().4s, \t0\().4s + fsub \e3\().4s, \e1\().4s, \t1\().4s + fadd \e1\().4s, \e1\().4s, \t1\().4s +.endm + +function ff_tx_fft2_float_neon, export=1 + ld2r { v0.2d, v1.2d }, [x2] + + fneg v2.2s, v1.2s + mov v2.d[1], v1.d[0] + + fsub v2.4s, v0.4s, v2.4s + + st1 { v2.4s }, [x1] + ret +endfunc + +.macro FFT4_FN name, inv +function ff_tx_fft4_\name\()_float_neon, export=1 + ld1 {v0.4s, v1.4s}, [x2] + +.if \inv == 1 + mov v2.d[0], v0.d[1] + mov v0.d[1], v1.d[1] + mov v1.d[1], v2.d[0] +.endif + + FFT4 v0, v1, 1 + + st1 { v0.4s, v1.4s }, [x1] + ret +endfunc +.endm + +FFT4_FN fwd, 0 +FFT4_FN inv, 1 + +.macro FFT8_FN name, no_perm +function ff_tx_fft8_\name\()_neon, export=1 + SETUP_LUT \no_perm + LOAD_INPUT 0, 1, 2, 3, x2, \no_perm + + LOAD_SUBADD + FFT8 v0, v1, v2, v3 + + zip1 v16.2d, v0.2d, v2.2d + zip2 v17.2d, v0.2d, v2.2d + zip1 v18.2d, v1.2d, v3.2d + zip2 v19.2d, v1.2d, v3.2d + st1 { v16.4s, v17.4s, v18.4s, v19.4s }, [x1] + + ret +endfunc +.endm + +FFT8_FN float, 0 +FFT8_FN ns_float, 1 + +.macro FFT16_FN name, no_perm +function ff_tx_fft16_\name\()_neon, export=1 + SETUP_LUT \no_perm + LOAD_INPUT 0, 1, 2, 3, x2, \no_perm + LOAD_INPUT 4, 5, 6, 7, x2, \no_perm + + LOAD_SUBADD + FFT16 v0, v1, v2, v3, v4, v5, v6, v7 + + zip1 v20.2d, v0.2d, v4.2d + zip2 v21.2d, v0.2d, v4.2d + zip1 v22.2d, v1.2d, v6.2d + zip2 v23.2d, v1.2d, v6.2d + st1 { v20.4s, v21.4s, v22.4s, v23.4s }, [x1], #64 + + zip1 v24.2d, v2.2d, v5.2d + zip2 v25.2d, v2.2d, v5.2d + zip1 v26.2d, v3.2d, v7.2d + zip2 v27.2d, v3.2d, v7.2d + st1 { v24.4s, v25.4s, v26.4s, v27.4s }, [x1] + + ret +endfunc +.endm + +FFT16_FN float, 0 +FFT16_FN ns_float, 1 + +.macro SETUP_SR_RECOMB len, re, im, dec + ldr w5, =(\len - 4*7) + movrel \re, X(ff_tx_tab_\len\()_float) + add \im, \re, x5 + mov \dec, #-32 + +.if \len > 32 + mov x21, #2*\len + add x22, x21, x21, lsl #1 +.endif +.endm + +.macro SR_COMBINE e0, e1, e2, e3, e4, e5, e6, e7, \ + o0, o1, o2, o3, o4, o5, o6, o7, \ + re, im, dec, swap_im, \ + t0=v16, t1=v17, t2=v18, t3=v19, t4=v20, t5=v21, \ + t6=v22, t7=v23, t8=v24, t9=v25, ta=v26, tb=v27 + + ld1 { \t8\().4s, \t9\().4s }, [\im], \dec + ld1 { \t0\().4s, \t1\().4s }, [\re], #32 + +.if \swap_im == 1 + ext \t2\().16b, \t9\().16b, \t9\().16b, #8 + ext \t3\().16b, \t8\().16b, \t8\().16b, #8 +.else + ext \t2\().16b, \t8\().16b, \t8\().16b, #8 + ext \t3\().16b, \t9\().16b, \t9\().16b, #8 +.endif + + trn1 \t4\().4s, \t0\().4s, \t0\().4s // cos0022 + trn2 \t0\().4s, \t0\().4s, \t0\().4s // cos4466 + trn1 \t5\().4s, \t1\().4s, \t1\().4s // cos1133 + trn2 \t1\().4s, \t1\().4s, \t1\().4s // cos5577 + + rev64 \t6\().4s, \o0\().4s // E m2[0,1].imre + rev64 \t7\().4s, \o2\().4s // O m2[0,1].imre + rev64 \t8\().4s, \o4\().4s // E m2[2,3].imre + rev64 \t9\().4s, \o6\().4s // O m2[2,3].imre + + fmul \t6\().4s, \t6\().4s, \t4\().4s // E m2[0,1].imre*t1[0,2] + fmul \t7\().4s, \t7\().4s, \t0\().4s // O m2[0,1].imre*t1[0,2] + fmul \t8\().4s, \t8\().4s, \t4\().4s // E m2[2,3].imre*t1[0,2] + fmul \t9\().4s, \t9\().4s, \t0\().4s // O m2[2,3].imre*t1[0,2] + + rev64 \ta\().4s, \o1\().4s // E m3[0,1].imre + rev64 \tb\().4s, \o3\().4s // O m3[0,1].imre + rev64 \t4\().4s, \o5\().4s // E m3[2,3].imre + rev64 \t0\().4s, \o7\().4s // O m3[2,3].imre + + fmul \ta\().4s, \ta\().4s, \t5\().4s // E m3[0,1].imre*t1[4,6] + fmul \tb\().4s, \tb\().4s, \t1\().4s // O m3[0,1].imre*t1[4,6] + fmul \t4\().4s, \t4\().4s, \t5\().4s // E m3[2,3].imre*t1[4,6] + fmul \t0\().4s, \t0\().4s, \t1\().4s // O m3[2,3].imre*t1[4,6] + + trn1 \t5\().4s, \t3\().4s, \t3\().4s // wim2200 + trn2 \t3\().4s, \t3\().4s, \t3\().4s // wim3311 + trn1 \t1\().4s, \t2\().4s, \t2\().4s // wim6644 + trn2 \t2\().4s, \t2\().4s, \t2\().4s // wim7755 + + fmul \t5\().4s, \t5\().4s, v31.4s + fmul \t3\().4s, \t3\().4s, v31.4s + fmul \t1\().4s, \t1\().4s, v31.4s + fmul \t2\().4s, \t2\().4s, v31.4s + + fmla \t7\().4s, \o2\().4s, \t5\().4s // O w0123 + fmls \t9\().4s, \o6\().4s, \t5\().4s // O j0123 + fmla \t6\().4s, \o0\().4s, \t3\().4s // E w0123 + fmls \t8\().4s, \o4\().4s, \t3\().4s // E j0123 + + fmla \ta\().4s, \o1\().4s, \t2\().4s // E w4567 + fmla \tb\().4s, \o3\().4s, \t1\().4s // O w4567 + fmls \t4\().4s, \o5\().4s, \t2\().4s // E j4567 + fmls \t0\().4s, \o7\().4s, \t1\().4s // O j4567 + + fsub \t2\().4s, \t7\().4s, \t9\().4s + fsub \t1\().4s, \t8\().4s, \t6\().4s + fsub \t3\().4s, \t4\().4s, \ta\().4s + fsub \t5\().4s, \t0\().4s, \tb\().4s + + fadd \t6\().4s, \t8\().4s, \t6\().4s + fadd \t7\().4s, \t9\().4s, \t7\().4s + fadd \t8\().4s, \t4\().4s, \ta\().4s + fadd \t9\().4s, \t0\().4s, \tb\().4s + + fmul \t1\().4s, \t1\().4s, v31.4s + fmul \t2\().4s, \t2\().4s, v31.4s + fmul \t3\().4s, \t3\().4s, v31.4s + fmul \t5\().4s, \t5\().4s, v31.4s + + rev64 \t6\().4s, \t6\().4s + rev64 \t8\().4s, \t8\().4s + rev64 \t7\().4s, \t7\().4s + rev64 \t9\().4s, \t9\().4s + + fsub \o0\().4s, \e0\().4s, \t6\().4s + fsub \o1\().4s, \e1\().4s, \t8\().4s + fsub \o2\().4s, \e2\().4s, \t1\().4s + fsub \o3\().4s, \e3\().4s, \t3\().4s + + fsub \o4\().4s, \e4\().4s, \t7\().4s + fsub \o5\().4s, \e6\().4s, \t9\().4s + fadd \o6\().4s, \e5\().4s, \t2\().4s + fsub \o7\().4s, \e7\().4s, \t5\().4s + + fadd \e0\().4s, \e0\().4s, \t6\().4s + fadd \e1\().4s, \e1\().4s, \t8\().4s + fadd \e2\().4s, \e2\().4s, \t1\().4s + fadd \e3\().4s, \e3\().4s, \t3\().4s + + fadd \e4\().4s, \e4\().4s, \t7\().4s + fsub \e5\().4s, \e5\().4s, \t2\().4s // swapped + fadd \e6\().4s, \e6\().4s, \t9\().4s // swapped + fadd \e7\().4s, \e7\().4s, \t5\().4s +.endm + +.macro SR_COMBINE_HALF e0, e1, e2, e3, \ + o0, o1, o2, o3, \ + c0, c1, c2, c3, \ + t0, t1, t2, t3, t4, t5, part + +.if \part == 0 + trn1 \t4\().4s, \c0\().4s, \c0\().4s // cos0022 + trn1 \c1\().4s, \c1\().4s, \c1\().4s // cos1133 +.else + trn2 \t4\().4s, \c0\().4s, \c0\().4s // cos0022 + trn2 \c1\().4s, \c1\().4s, \c1\().4s // cos1133 +.endif +.if \part == 0 + trn2 \t5\().4s, \c2\().4s, \c2\().4s // wim7755 + trn2 \c3\().4s, \c3\().4s, \c3\().4s // wim3311 +.else + trn1 \t5\().4s, \c2\().4s, \c2\().4s // wim7755 + trn1 \c3\().4s, \c3\().4s, \c3\().4s // wim3311 +.endif + + fmul \t5\().4s, \t5\().4s, v31.4s + fmul \c3\().4s, \c3\().4s, v31.4s + + rev64 \t0\().4s, \o0\().4s // E m2[0,1].imre + rev64 \t1\().4s, \o2\().4s // E m2[2,3].imre + rev64 \t2\().4s, \o1\().4s // E m3[0,1].imre + rev64 \t3\().4s, \o3\().4s // E m3[2,3].imre + + fmul \o0\().4s, \o0\().4s, \c3\().4s // E m2[0,1].imre*t1[0,2] + fmul \o1\().4s, \o1\().4s, \t5\().4s // E m3[0,1].imre*t1[4,6] + fmla \o0\().4s, \t0\().4s, \t4\().4s // E w0123 + fmla \o1\().4s, \t2\().4s, \c1\().4s // E w4567 + + fmul \t1\().4s, \t1\().4s, \t4\().4s // E m2[2,3].imre*t1[0,2] + fmul \t3\().4s, \t3\().4s, \c1\().4s // E m3[2,3].imre*t1[4,6] + fmls \t1\().4s, \o2\().4s, \c3\().4s // E j0123 + fmls \t3\().4s, \o3\().4s, \t5\().4s // E j4567 + + fsub \t0\().4s, \t1\().4s, \o0\().4s + fadd \t1\().4s, \t1\().4s, \o0\().4s + fadd \t2\().4s, \t3\().4s, \o1\().4s + fsub \t3\().4s, \t3\().4s, \o1\().4s + + fmul \t0\().4s, \t0\().4s, v31.4s + fmul \t3\().4s, \t3\().4s, v31.4s + + rev64 \t1\().4s, \t1\().4s + rev64 \t2\().4s, \t2\().4s + +.if \part == 0 + fsub \o0\().4s, \e0\().4s, \t1\().4s + fsub \o1\().4s, \e1\().4s, \t2\().4s + fsub \o2\().4s, \e2\().4s, \t0\().4s + fsub \o3\().4s, \e3\().4s, \t3\().4s +.else + fsub \o0\().4s, \e0\().4s, \t1\().4s + fadd \o2\().4s, \e1\().4s, \t2\().4s + fsub \o1\().4s, \e2\().4s, \t0\().4s + fadd \o3\().4s, \e3\().4s, \t3\().4s +.endif + +.if \part == 0 + fadd \e0\().4s, \e0\().4s, \t1\().4s + fadd \e1\().4s, \e1\().4s, \t2\().4s + fadd \e2\().4s, \e2\().4s, \t0\().4s + fadd \e3\().4s, \e3\().4s, \t3\().4s +.else + fadd \e0\().4s, \e0\().4s, \t1\().4s + fsub \e1\().4s, \e1\().4s, \t2\().4s // swapped + fadd \e2\().4s, \e2\().4s, \t0\().4s // swapped + fsub \e3\().4s, \e3\().4s, \t3\().4s +.endif +.endm + +/* Same as SR_COMBINE_HALF, but heroically tries to use 3 temporary registers + * without touching the tables. */ +.macro SR_COMBINE_LITE e0, e1, e2, e3, \ + o0, o1, o2, o3, \ + c0, c1, c2, c3, \ + t0, t1, t2, part + + rev64 \t0\().4s, \o0\().4s // E m2[0,1].imre + rev64 \t1\().4s, \o2\().4s // E m2[2,3].imre +.if \part == 0 + trn2 \t2\().4s, \c3\().4s, \c3\().4s // wim3311 +.else + trn1 \t2\().4s, \c3\().4s, \c3\().4s // wim3311 +.endif + fmul \t2\().4s, \t2\().4s, v31.4s + fmul \o2\().4s, \o2\().4s, \t2\().4s + fmul \o0\().4s, \o0\().4s, \t2\().4s // E m2[0,1].imre*t1[0,2] +.if \part == 0 + trn1 \t2\().4s, \c0\().4s, \c0\().4s // cos0022 +.else + trn2 \t2\().4s, \c0\().4s, \c0\().4s // cos0022 +.endif + fmul \t1\().4s, \t1\().4s, \t2\().4s // E m2[2,3].imre*t1[0,2] + fmla \o0\().4s, \t0\().4s, \t2\().4s // E w0123 + fsub \t1\().4s, \t1\().4s, \o2\().4s // E j0123 + + rev64 \t2\().4s, \o1\().4s // E m3[0,1].imre + rev64 \o2\().4s, \o3\().4s // E m3[2,3].imre + +.if \part == 0 + trn2 \t0\().4s, \c2\().4s, \c2\().4s // wim7755 +.else + trn1 \t0\().4s, \c2\().4s, \c2\().4s // wim7755 +.endif + fmul \t0\().4s, \t0\().4s, v31.4s + + fmul \o1\().4s, \o1\().4s, \t0\().4s // E m3[0,1].imre*t1[4,6] + fmul \o3\().4s, \o3\().4s, \t0\().4s + +.if \part == 0 + trn1 \t0\().4s, \c1\().4s, \c1\().4s // cos1133 +.else + trn2 \t0\().4s, \c1\().4s, \c1\().4s // cos1133 +.endif + fmul \o2\().4s, \o2\().4s, \t0\().4s // E m3[2,3].imre*t1[4,6] + fmla \o1\().4s, \t2\().4s, \t0\().4s // E w4567 + fsub \o2\().4s, \o2\().4s, \o3\().4s // E j4567 + + fsub \t0\().4s, \t1\().4s, \o0\().4s + fadd \o0\().4s, \t1\().4s, \o0\().4s + fadd \t2\().4s, \o2\().4s, \o1\().4s + fsub \t1\().4s, \o2\().4s, \o1\().4s + + fmul \t0\().4s, \t0\().4s, v31.4s + fmul \t1\().4s, \t1\().4s, v31.4s + + rev64 \t2\().4s, \t2\().4s + rev64 \o0\().4s, \o0\().4s + +.if \part == 0 + fsub \o1\().4s, \e1\().4s, \t2\().4s + fsub \o2\().4s, \e2\().4s, \t0\().4s + fsub \o3\().4s, \e3\().4s, \t1\().4s +.else + fadd \o2\().4s, \e1\().4s, \t0\().4s + fsub \o1\().4s, \e2\().4s, \t2\().4s + fadd \o3\().4s, \e3\().4s, \t1\().4s +.endif + +.if \part == 0 + fadd \e1\().4s, \e1\().4s, \t2\().4s + fadd \e2\().4s, \e2\().4s, \t0\().4s + fadd \e3\().4s, \e3\().4s, \t1\().4s +.else + fsub \e1\().4s, \e1\().4s, \t0\().4s // swapped + fadd \e2\().4s, \e2\().4s, \t2\().4s // swapped + fsub \e3\().4s, \e3\().4s, \t1\().4s +.endif + + mov \t1\().16b, \o0\().16b + + fsub \o0\().4s, \e0\().4s, \t1\().4s + fadd \e0\().4s, \e0\().4s, \t1\().4s +.endm + +.macro SR_COMBINE_4 len, part, off + add x10, x1, x21 + add x11, x1, x21, lsl #1 + add x12, x1, x22 + + ldp q0, q1, [x1, #((0 + \part)*32 + \off)] + ldp q4, q5, [x1, #((2 + \part)*32 + \off)] + ldp q2, q3, [x10, #((0 + \part)*32 + \off)] + ldp q6, q7, [x10, #((2 + \part)*32 + \off)] + + ldp q8, q9, [x11, #((0 + \part)*32 + \off)] + ldp q10, q11, [x11, #((2 + \part)*32 + \off)] + ldp q12, q13, [x12, #((0 + \part)*32 + \off)] + ldp q14, q15, [x12, #((2 + \part)*32 + \off)] + + SR_COMBINE v0, v1, v2, v3, v4, v6, v5, v7, \ + v8, v9, v10, v11, v12, v13, v14, v15, \ + x7, x8, x9, 0 + + stp q0, q1, [x1, #((0 + \part)*32 + \off)] + stp q4, q5, [x1, #((2 + \part)*32 + \off)] + stp q2, q3, [x10, #((0 + \part)*32 + \off)] + stp q6, q7, [x10, #((2 + \part)*32 + \off)] + + stp q8, q9, [x11, #((0 + \part)*32 + \off)] + stp q12, q13, [x11, #((2 + \part)*32 + \off)] + stp q10, q11, [x12, #((0 + \part)*32 + \off)] + stp q14, q15, [x12, #((2 + \part)*32 + \off)] +.endm + +.macro SR_COMBINE_FULL len, off=0 + add x10, x1, x21 + add x11, x1, x21, lsl #1 + add x12, x1, x22 + + SR_COMBINE_4 \len, 0, \off + SR_COMBINE_4 \len, 1, \off + SR_COMBINE_4 \len, 4, \off + SR_COMBINE_4 \len, 5, \off +.endm + +.macro SR_COMBINE_D2 part, off + add x10, x1, #((\part)*32 + \off) + add x11, x14, #((\part)*32 + \off) + add x12, x15, #((\part)*32 + \off) + add x13, x16, #((\part)*32 + \off) + + ldp q0, q1, [x10] + ldp q4, q5, [x10, #(2*32)] + ldp q2, q3, [x11] + ldp q6, q7, [x11, #(2*32)] + + ldp q8, q9, [x12] + ldp q10, q11, [x12, #(2*32)] + ldp q12, q13, [x13] + ldp q14, q15, [x13, #(2*32)] + + SR_COMBINE v0, v1, v2, v3, v4, v6, v5, v7, \ + v8, v9, v10, v11, v12, v13, v14, v15, \ + x7, x8, x9, 0, \ + v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27 + + zip1 v16.2d, v0.2d, v4.2d + zip2 v17.2d, v0.2d, v4.2d + zip1 v18.2d, v1.2d, v5.2d + zip2 v19.2d, v1.2d, v5.2d + + zip1 v20.2d, v2.2d, v6.2d + zip2 v21.2d, v2.2d, v6.2d + zip1 v22.2d, v3.2d, v7.2d + zip2 v23.2d, v3.2d, v7.2d + + ldp q0, q1, [x10, #(1*32)] + ldp q4, q5, [x10, #(3*32)] + ldp q2, q3, [x11, #(1*32)] + ldp q6, q7, [x11, #(3*32)] + + st1 { v16.4s, v17.4s, v18.4s, v19.4s }, [x10], #64 + st1 { v20.4s, v21.4s, v22.4s, v23.4s }, [x11], #64 + + zip1 v20.2d, v8.2d, v12.2d + zip2 v21.2d, v8.2d, v12.2d + zip1 v22.2d, v9.2d, v13.2d + zip2 v23.2d, v9.2d, v13.2d + zip1 v24.2d, v10.2d, v14.2d + zip2 v25.2d, v10.2d, v14.2d + zip1 v26.2d, v11.2d, v15.2d + zip2 v27.2d, v11.2d, v15.2d + + ldp q8, q9, [x12, #(1*32)] + ldp q10, q11, [x12, #(3*32)] + ldp q12, q13, [x13, #(1*32)] + ldp q14, q15, [x13, #(3*32)] + + st1 { v20.4s, v21.4s, v22.4s, v23.4s }, [x12], #64 + st1 { v24.4s, v25.4s, v26.4s, v27.4s }, [x13], #64 + + SR_COMBINE v0, v1, v2, v3, v4, v6, v5, v7, \ + v8, v9, v10, v11, v12, v13, v14, v15, \ + x7, x8, x9, 0, \ + v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27 + + zip1 v16.2d, v0.2d, v4.2d + zip2 v17.2d, v0.2d, v4.2d + zip1 v18.2d, v1.2d, v5.2d + zip2 v19.2d, v1.2d, v5.2d + st1 { v16.4s, v17.4s, v18.4s, v19.4s }, [x10] + + zip1 v16.2d, v2.2d, v6.2d + zip2 v17.2d, v2.2d, v6.2d + zip1 v18.2d, v3.2d, v7.2d + zip2 v19.2d, v3.2d, v7.2d + st1 { v16.4s, v17.4s, v18.4s, v19.4s }, [x11] + + zip1 v20.2d, v8.2d, v12.2d + zip2 v21.2d, v8.2d, v12.2d + zip1 v22.2d, v9.2d, v13.2d + zip2 v23.2d, v9.2d, v13.2d + st1 { v20.4s, v21.4s, v22.4s, v23.4s }, [x12] + + zip1 v24.2d, v10.2d, v14.2d + zip2 v25.2d, v10.2d, v14.2d + zip1 v26.2d, v11.2d, v15.2d + zip2 v27.2d, v11.2d, v15.2d + st1 { v24.4s, v25.4s, v26.4s, v27.4s }, [x13] +.endm + +.macro SR_COMBINE_DINT off=0 + add x14, x1, x21 + add x15, x1, x21, lsl #1 + add x16, x1, x22 + + SR_COMBINE_D2 0, \off + SR_COMBINE_D2 4, \off +.endm + +.macro FFT32_FN name, no_perm +function ff_tx_fft32_\name\()_neon, export=1 + stp d14, d15, [sp, #-16*4]! + stp d8, d9, [sp, #16*3] + stp d10, d11, [sp, #16*2] + stp d12, d13, [sp, #16] + + LOAD_SUBADD + SETUP_SR_RECOMB 32, x7, x8, x9 + + SETUP_LUT \no_perm + LOAD_INPUT 0, 1, 2, 3, x2, \no_perm + LOAD_INPUT 4, 5, 6, 7, x2, \no_perm + LOAD_INPUT 8, 9, 10, 11, x2, \no_perm + LOAD_INPUT 12, 13, 14, 15, x2, \no_perm + + FFT8_X2 v8, v9, v10, v11, v12, v13, v14, v15 + FFT16 v0, v1, v2, v3, v4, v5, v6, v7 + + SR_COMBINE v0, v1, v2, v3, v4, v5, v6, v7, \ + v8, v9, v10, v11, v12, v13, v14, v15, \ + x7, x8, x9, 0 + + zip1 v16.2d, v0.2d, v4.2d + zip2 v17.2d, v0.2d, v4.2d + zip1 v18.2d, v1.2d, v6.2d + zip2 v19.2d, v1.2d, v6.2d + st1 { v16.4s, v17.4s, v18.4s, v19.4s }, [x1], #64 + + zip1 v20.2d, v2.2d, v5.2d + zip2 v21.2d, v2.2d, v5.2d + zip1 v22.2d, v3.2d, v7.2d + zip2 v23.2d, v3.2d, v7.2d + st1 { v20.4s, v21.4s, v22.4s, v23.4s }, [x1], #64 + + zip1 v24.2d, v8.2d, v12.2d + zip2 v25.2d, v8.2d, v12.2d + zip1 v26.2d, v9.2d, v13.2d + zip2 v27.2d, v9.2d, v13.2d + st1 { v24.4s, v25.4s, v26.4s, v27.4s }, [x1], #64 + + zip1 v28.2d, v10.2d, v14.2d + zip2 v29.2d, v10.2d, v14.2d + zip1 v30.2d, v11.2d, v15.2d + zip2 v31.2d, v11.2d, v15.2d + st1 { v28.4s, v29.4s, v30.4s, v31.4s }, [x1] + + ldp d12, d13, [sp, #16] + ldp d10, d11, [sp, #16*2] + ldp d8, d9, [sp, #16*3] + ldp d14, d15, [sp], #16*4 + + ret +endfunc +.endm + +FFT32_FN float, 0 +FFT32_FN ns_float, 1 + +.macro cmp_imm reg, imm +.if \imm >= 4096 + cmp \reg, #((\imm)/4096), lsl #12 +.else + cmp \reg, #(\imm) +.endif +.endm + +.macro SR_TRANSFORM_DEF len, next=0 +\len: + stp x20, x30, [sp, #-16]! + mov w20, #(\len/4) + mov x5, #((\len*4) - (\len/1)) + add x1, x1, x5 + bl 32b + mov x5, #((\len*2) - (\len/2)) + add x1, x1, x5 + bl 32b + ldp x20, x30, [sp], #16 + ldr w5, =(\len*6 + \len/2) + sub x1, x1, x5 + + SETUP_SR_RECOMB \len, x7, x8, x9 + +.if \next\() != 0 + cmp_imm w19, \len + b.eq 0f + + mov w5, #(\len/128) +\len\()5: + SR_COMBINE_FULL \len + add x1, x1, 8*32 + subs w5, w5, 1 + b.gt \len\()5b + + cmp_imm w20, \len + b.gt \next\()f + ret +.endif +.endm + +.macro FFT_SPLIT_RADIX_FN name, no_perm +function ff_tx_fft_sr_\name\()_neon, export=1 + stp x21, x22, [sp, #-16*6]! + stp d8, d9, [sp, #16*5] + stp d10, d11, [sp, #16*4] + stp d12, d13, [sp, #16*3] + stp d14, d15, [sp, #16*2] + stp x19, x20, [sp, #16] + + ldr w19, [x0, #0] // global target + mov w20, w19 // local length + + LOAD_SUBADD + SETUP_LUT \no_perm + +32: + SETUP_SR_RECOMB 32, x7, x8, x9 + + LOAD_INPUT 0, 1, 2, 3, x2, \no_perm + LOAD_INPUT 4, 6, 5, 7, x2, \no_perm, 1 + LOAD_INPUT 8, 9, 10, 11, x2, \no_perm + LOAD_INPUT 12, 13, 14, 15, x2, \no_perm + + FFT8_X2 v8, v9, v10, v11, v12, v13, v14, v15 + FFT16 v0, v1, v2, v3, v4, v6, v5, v7 + + SR_COMBINE v0, v1, v2, v3, v4, v6, v5, v7, \ + v8, v9, v10, v11, v12, v13, v14, v15, \ + x7, x8, x9, 0 + + stp q2, q3, [x1, #32*1] + stp q6, q7, [x1, #32*3] + stp q10, q11, [x1, #32*5] + stp q14, q15, [x1, #32*7] + + cmp w20, #32 + b.gt 64f + + stp q0, q1, [x1, #32*0] + stp q4, q5, [x1, #32*2] + stp q8, q9, [x1, #32*4] + stp q12, q13, [x1, #32*6] + + ret +64: + SETUP_SR_RECOMB 64, x7, x8, x9 + + LOAD_INPUT 2, 3, 10, 11, x2, \no_perm, 1 + LOAD_INPUT 6, 14, 7, 15, x2, \no_perm, 1 + + FFT16 v2, v3, v10, v11, v6, v14, v7, v15 + + LOAD_INPUT 16, 17, 18, 19, x2, \no_perm + LOAD_INPUT 20, 22, 21, 23, x2, \no_perm, 1 + + FFT16 v16, v17, v18, v19, v20, v22, v21, v23, \ + v24, v25, v26, v27, v28, v29, v30 + + ld1 { v26.4s, v27.4s }, [x8], x9 + ldp q24, q25, [x7], #32 + + ext v26.16b, v26.16b, v26.16b, #8 + ext v27.16b, v27.16b, v27.16b, #8 + + cmp w19, #64 + b.eq 2f // custom deinterleave + + // TODO: investigate doing the 2 combines like in deinterleave + // TODO: experiment with spilling to gprs and converting to HALF or full + SR_COMBINE_LITE v0, v1, v8, v9, \ + v2, v3, v16, v17, \ + v24, v25, v26, v27, \ + v28, v29, v30, 0 + + stp q0, q1, [x1, #32* 0] + stp q8, q9, [x1, #32* 4] + stp q2, q3, [x1, #32* 8] + stp q16, q17, [x1, #32*12] + + SR_COMBINE_HALF v4, v5, v12, v13, \ + v6, v7, v20, v21, \ + v24, v25, v26, v27, \ + v28, v29, v30, v0, v1, v8, 1 + + stp q4, q20, [x1, #32* 2] + stp q12, q21, [x1, #32* 6] + stp q6, q5, [x1, #32*10] + stp q7, q13, [x1, #32*14] + + ldp q2, q3, [x1, #32*1] + ldp q6, q7, [x1, #32*3] + ldp q12, q13, [x1, #32*5] + ldp q16, q17, [x1, #32*7] + + SR_COMBINE v2, v3, v12, v13, v6, v16, v7, v17, \ + v10, v11, v14, v15, v18, v19, v22, v23, \ + x7, x8, x9, 0, \ + v24, v25, v26, v27, v28, v29, v30, v8, v0, v1, v4, v5 + + stp q2, q3, [x1, #32* 1] + stp q6, q7, [x1, #32* 3] + stp q12, q13, [x1, #32* 5] + stp q16, q17, [x1, #32* 7] + + stp q10, q11, [x1, #32* 9] + stp q18, q19, [x1, #32*11] + stp q14, q15, [x1, #32*13] + stp q22, q23, [x1, #32*15] + + cmp w20, #64 + b.gt 128f + ret +128: + stp x20, x30, [sp, #-16]! + mov w20, #32 + add x1, x1, #16*32 + bl 32b + add x1, x1, #8*32 + bl 32b + ldp x20, x30, [sp], #16 + sub x1, x1, #24*32 + + SETUP_SR_RECOMB 128, x7, x8, x9 + + cmp w19, #128 + b.eq 0f + + SR_COMBINE_FULL 128 + + cmp w20, #128 + b.gt 256f + ret +256: + stp x20, x30, [sp, #-16]! + mov w20, #64 + add x1, x1, #32*32 + bl 32b + add x1, x1, #16*32 + bl 32b + ldp x20, x30, [sp], #16 + sub x1, x1, #48*32 + + SETUP_SR_RECOMB 256, x7, x8, x9 + + cmp w19, #256 + b.eq 0f + + SR_COMBINE_FULL 256 + SR_COMBINE_FULL 256, 8*32 + + cmp w20, #256 + b.gt 512f + ret +512: + stp x20, x30, [sp, #-16]! + mov w20, #128 + add x1, x1, #64*32 + bl 32b + add x1, x1, #32*32 + bl 32b + ldp x20, x30, [sp], #16 + sub x1, x1, #96*32 + + SETUP_SR_RECOMB 512, x7, x8, x9 + + cmp w19, #512 + b.eq 0f + + mov x5, 4 +5125: + SR_COMBINE_FULL 512 + add x1, x1, 8*32 + subs w5, w5, 1 + b.gt 5125b + + cmp w20, #512 + b.gt 1024f + + ret +1024: + stp x20, x30, [sp, #-16]! + mov w20, #256 + add x1, x1, #96*32 + bl 32b + add x1, x1, #64*32 + bl 32b + ldp x20, x30, [sp], #16 + mov x5, #192*32 + sub x1, x1, x5 + + SETUP_SR_RECOMB 1024, x7, x8, x9 + + cmp w19, #1024 + b.eq 0f + + mov w5, 8 +10245: + SR_COMBINE_FULL 1024 + add x1, x1, 8*32 + subs w5, w5, 1 + b.gt 10245b + + cmp w20, #1024 + b.gt 2048f + + ret + +SR_TRANSFORM_DEF 2048, 4096 +SR_TRANSFORM_DEF 4096, 8192 +SR_TRANSFORM_DEF 8192, 16384 +SR_TRANSFORM_DEF 16384, 32768 +SR_TRANSFORM_DEF 32768, 65536 +SR_TRANSFORM_DEF 65536, 131072 +SR_TRANSFORM_DEF 131072 + +0: // general deinterleave loop + SR_COMBINE_DINT + add x1, x1, #32*8 + subs w19, w19, #32*4 + b.gt 0b + + ldp x19, x20, [sp, #16] + ldp d14, d15, [sp, #16*2] + ldp d12, d13, [sp, #16*3] + ldp d10, d11, [sp, #16*4] + ldp d8, d9, [sp, #16*5] + ldp x21, x22, [sp], #16*6 + + ret + +2: // special case for 64 point deinterleave + mov x10, v23.d[0] + mov x11, v23.d[1] + + SR_COMBINE_LITE v0, v1, v8, v9, \ + v2, v3, v16, v17, \ + v24, v25, v26, v27, \ + v28, v29, v30, 0 + + SR_COMBINE_HALF v4, v5, v12, v13, \ + v6, v7, v20, v21, \ + v24, v25, v26, v27, \ + v28, v29, v30, v23, v24, v26, 1 + + zip1 v23.2d, v0.2d, v4.2d + zip2 v24.2d, v0.2d, v4.2d + zip1 v25.2d, v1.2d, v20.2d + zip2 v26.2d, v1.2d, v20.2d + + zip1 v27.2d, v8.2d, v12.2d + zip2 v28.2d, v8.2d, v12.2d + zip1 v29.2d, v9.2d, v21.2d + zip2 v30.2d, v9.2d, v21.2d + + mov v20.16b, v5.16b + mov v21.16b, v7.16b + mov x12, x1 + add x13, x1, #32* 4 + add x14, x1, #32* 8 + add x15, x1, #32*12 + + zip1 v4.2d, v2.2d, v6.2d + zip2 v5.2d, v2.2d, v6.2d + zip1 v6.2d, v3.2d, v20.2d + zip2 v7.2d, v3.2d, v20.2d + + zip1 v0.2d, v16.2d, v21.2d + zip2 v1.2d, v16.2d, v21.2d + zip1 v2.2d, v17.2d, v13.2d + zip2 v3.2d, v17.2d, v13.2d + + // stp is faster by a little on A53, but this is faster on M1s (theory) + ldp q8, q9, [x1, #32*1] + ldp q12, q13, [x1, #32*5] + + st1 { v23.4s, v24.4s, v25.4s, v26.4s }, [x12], #64 // 32* 0...1 + st1 { v27.4s, v28.4s, v29.4s, v30.4s }, [x13], #64 // 32* 4...5 + st1 { v4.4s, v5.4s, v6.4s, v7.4s }, [x14], #64 // 32* 8...9 + st1 { v0.4s, v1.4s, v2.4s, v3.4s }, [x15], #64 // 32*12..13 + + mov v23.d[0], x10 + mov v23.d[1], x11 + + ldp q6, q7, [x1, #32*3] + ldp q16, q17, [x1, #32*7] + + SR_COMBINE v8, v9, v12, v13, v6, v16, v7, v17, \ + v10, v11, v14, v15, v18, v19, v22, v23, \ + x7, x8, x9, 0, \ + v24, v25, v26, v27, v28, v29, v30, v4, v0, v1, v5, v20 + + zip1 v0.2d, v8.2d, v6.2d + zip2 v1.2d, v8.2d, v6.2d + zip1 v2.2d, v9.2d, v7.2d + zip2 v3.2d, v9.2d, v7.2d + st1 { v0.4s, v1.4s, v2.4s, v3.4s }, [x12] + + zip1 v4.2d, v12.2d, v16.2d + zip2 v5.2d, v12.2d, v16.2d + zip1 v6.2d, v13.2d, v17.2d + zip2 v7.2d, v13.2d, v17.2d + st1 { v4.4s, v5.4s, v6.4s, v7.4s }, [x13] + + zip1 v0.2d, v10.2d, v18.2d + zip2 v1.2d, v10.2d, v18.2d + zip1 v2.2d, v11.2d, v19.2d + zip2 v3.2d, v11.2d, v19.2d + st1 { v0.4s, v1.4s, v2.4s, v3.4s }, [x14] + + zip1 v4.2d, v14.2d, v22.2d + zip2 v5.2d, v14.2d, v22.2d + zip1 v6.2d, v15.2d, v23.2d + zip2 v7.2d, v15.2d, v23.2d + st1 { v4.4s, v5.4s, v6.4s, v7.4s }, [x15] + + ldp x19, x20, [sp, #16] + ldp d14, d15, [sp, #16*2] + ldp d12, d13, [sp, #16*3] + ldp d10, d11, [sp, #16*4] + ldp d8, d9, [sp, #16*5] + ldp x21, x22, [sp], #16*6 + + ret +endfunc +.endm + +FFT_SPLIT_RADIX_FN float, 0 +FFT_SPLIT_RADIX_FN ns_float, 1 diff --git a/media/ffvpx/libavutil/adler32.c b/media/ffvpx/libavutil/adler32.c new file mode 100644 index 0000000000..7124f18802 --- /dev/null +++ b/media/ffvpx/libavutil/adler32.c @@ -0,0 +1,96 @@ +/* + * Compute the Adler-32 checksum of a data stream. + * This is a modified version based on adler32.c from the zlib library. + * + * Copyright (C) 1995 Mark Adler + * + * This software is provided 'as-is', without any express or implied + * warranty. In no event will the authors be held liable for any damages + * arising from the use of this software. + * + * Permission is granted to anyone to use this software for any purpose, + * including commercial applications, and to alter it and redistribute it + * freely, subject to the following restrictions: + * + * 1. The origin of this software must not be misrepresented; you must not + * claim that you wrote the original software. If you use this software + * in a product, an acknowledgment in the product documentation would be + * appreciated but is not required. + * 2. Altered source versions must be plainly marked as such, and must not be + * misrepresented as being the original software. + * 3. This notice may not be removed or altered from any source distribution. + */ + +/** + * @file + * Computes the Adler-32 checksum of a data stream + * + * This is a modified version based on adler32.c from the zlib library. + * @author Mark Adler + * @ingroup lavu_adler32 + */ + +#include "config.h" +#include "adler32.h" +#include "intreadwrite.h" +#include "macros.h" + +#define BASE 65521L /* largest prime smaller than 65536 */ + +#define DO1(buf) { s1 += *buf++; s2 += s1; } +#define DO4(buf) DO1(buf); DO1(buf); DO1(buf); DO1(buf); +#define DO16(buf) DO4(buf); DO4(buf); DO4(buf); DO4(buf); + +AVAdler av_adler32_update(AVAdler adler, const uint8_t *buf, size_t len) +{ + unsigned long s1 = adler & 0xffff; + unsigned long s2 = adler >> 16; + + while (len > 0) { +#if HAVE_FAST_64BIT && HAVE_FAST_UNALIGNED && !CONFIG_SMALL + unsigned len2 = FFMIN((len-1) & ~7, 23*8); + if (len2) { + uint64_t a1= 0; + uint64_t a2= 0; + uint64_t b1= 0; + uint64_t b2= 0; + len -= len2; + s2 += s1*len2; + while (len2 >= 8) { + uint64_t v = AV_RN64(buf); + a2 += a1; + b2 += b1; + a1 += v &0x00FF00FF00FF00FF; + b1 += (v>>8)&0x00FF00FF00FF00FF; + len2 -= 8; + buf+=8; + } + + //We combine the 8 interleaved adler32 checksums without overflows + //Decreasing the number of iterations would allow below code to be + //simplified but would likely be slower due to the fewer iterations + //of the inner loop + s1 += ((a1+b1)*0x1000100010001)>>48; + s2 += ((((a2&0xFFFF0000FFFF)+(b2&0xFFFF0000FFFF)+((a2>>16)&0xFFFF0000FFFF)+((b2>>16)&0xFFFF0000FFFF))*0x800000008)>>32) +#if HAVE_BIGENDIAN + + 2*((b1*0x1000200030004)>>48) + + ((a1*0x1000100010001)>>48) + + 2*((a1*0x0000100020003)>>48); +#else + + 2*((a1*0x4000300020001)>>48) + + ((b1*0x1000100010001)>>48) + + 2*((b1*0x3000200010000)>>48); +#endif + } +#else + while (len > 4 && s2 < (1U << 31)) { + DO4(buf); + len -= 4; + } +#endif + DO1(buf); len--; + s1 %= BASE; + s2 %= BASE; + } + return (s2 << 16) | s1; +} diff --git a/media/ffvpx/libavutil/adler32.h b/media/ffvpx/libavutil/adler32.h new file mode 100644 index 0000000000..232d07f5fe --- /dev/null +++ b/media/ffvpx/libavutil/adler32.h @@ -0,0 +1,63 @@ +/* + * copyright (c) 2006 Mans Rullgard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_adler32 + * Public header for Adler-32 hash function implementation. + */ + +#ifndef AVUTIL_ADLER32_H +#define AVUTIL_ADLER32_H + +#include <stddef.h> +#include <stdint.h> +#include "attributes.h" + +/** + * @defgroup lavu_adler32 Adler-32 + * @ingroup lavu_hash + * Adler-32 hash function implementation. + * + * @{ + */ + +typedef uint32_t AVAdler; + +/** + * Calculate the Adler32 checksum of a buffer. + * + * Passing the return value to a subsequent av_adler32_update() call + * allows the checksum of multiple buffers to be calculated as though + * they were concatenated. + * + * @param adler initial checksum value + * @param buf pointer to input buffer + * @param len size of input buffer + * @return updated checksum + */ +AVAdler av_adler32_update(AVAdler adler, const uint8_t *buf, + size_t len) av_pure; + +/** + * @} + */ + +#endif /* AVUTIL_ADLER32_H */ diff --git a/media/ffvpx/libavutil/arm/asm.S b/media/ffvpx/libavutil/arm/asm.S new file mode 100644 index 0000000000..e3a8c7f065 --- /dev/null +++ b/media/ffvpx/libavutil/arm/asm.S @@ -0,0 +1,367 @@ +/* + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#ifdef __ELF__ +# define ELF +#else +# define ELF @ +#endif + +#if CONFIG_THUMB +# define A @ +# define T +#else +# define A +# define T @ +#endif + +#if HAVE_AS_FUNC +# define FUNC +#else +# define FUNC @ +#endif + +#if HAVE_AS_FPU_DIRECTIVE +# define FPU +#else +# define FPU @ +#endif + +#if CONFIG_THUMB && defined(__APPLE__) +# define TFUNC +#else +# define TFUNC @ +#endif + +#if HAVE_AS_ARCH_DIRECTIVE +#if HAVE_NEON + .arch armv7-a +#elif HAVE_ARMV6T2 + .arch armv6t2 +#elif HAVE_ARMV6 + .arch armv6 +#elif HAVE_ARMV5TE + .arch armv5te +#endif +#endif +#if HAVE_AS_OBJECT_ARCH +ELF .object_arch armv4 +#endif + +#if HAVE_NEON +FPU .fpu neon +ELF .eabi_attribute 10, 0 @ suppress Tag_FP_arch +ELF .eabi_attribute 12, 0 @ suppress Tag_Advanced_SIMD_arch +#elif HAVE_VFP +FPU .fpu vfp +ELF .eabi_attribute 10, 0 @ suppress Tag_FP_arch +#endif + + .syntax unified +T .thumb +ELF .eabi_attribute 25, 1 @ Tag_ABI_align_preserved +ELF .section .note.GNU-stack,"",%progbits @ Mark stack as non-executable + +.macro function name, export=0, align=2 + .set .Lpic_idx, 0 + .set .Lpic_gp, 0 + .macro endfunc + .if .Lpic_idx + .align 2 + .altmacro + put_pic %(.Lpic_idx - 1) + .noaltmacro + .endif + .if .Lpic_gp + .unreq gp + .endif +ELF .size \name, . - \name +FUNC .endfunc + .purgem endfunc + .endm + .text + .align \align + .if \export + .global EXTERN_ASM\name +ELF .type EXTERN_ASM\name, %function +FUNC .func EXTERN_ASM\name +TFUNC .thumb_func EXTERN_ASM\name +EXTERN_ASM\name: + .else +ELF .type \name, %function +FUNC .func \name +TFUNC .thumb_func \name +\name: + .endif +.endm + +.macro const name, align=2, relocate=0 + .macro endconst +ELF .size \name, . - \name + .purgem endconst + .endm +#if HAVE_SECTION_DATA_REL_RO +.if \relocate + .section .data.rel.ro +.else + .section .rodata +.endif +#elif defined(_WIN32) + .section .rdata +#elif !defined(__MACH__) + .section .rodata +#else + .const_data +#endif + .align \align +\name: +.endm + +#if !HAVE_ARMV6T2_EXTERNAL +.macro movw rd, val + mov \rd, \val & 255 + orr \rd, \val & ~255 +.endm +#endif + +.macro mov32 rd, val +#if HAVE_ARMV6T2_EXTERNAL + movw \rd, #(\val) & 0xffff + .if (\val) >> 16 + movt \rd, #(\val) >> 16 + .endif +#else + ldr \rd, =\val +#endif +.endm + +.macro put_pic num + put_pic_\num +.endm + +.macro do_def_pic num, val, label + .macro put_pic_\num + .if \num + .altmacro + put_pic %(\num - 1) + .noaltmacro + .endif +\label: .word \val + .purgem put_pic_\num + .endm +.endm + +.macro def_pic val, label + .altmacro + do_def_pic %.Lpic_idx, \val, \label + .noaltmacro + .set .Lpic_idx, .Lpic_idx + 1 +.endm + +.macro ldpic rd, val, indir=0 + ldr \rd, .Lpicoff\@ +.Lpic\@: + .if \indir +A ldr \rd, [pc, \rd] +T add \rd, pc +T ldr \rd, [\rd] + .else + add \rd, pc + .endif + def_pic \val - (.Lpic\@ + (8 >> CONFIG_THUMB)), .Lpicoff\@ +.endm + +.macro movrel rd, val +#if CONFIG_PIC + ldpic \rd, \val +#elif HAVE_ARMV6T2_EXTERNAL && !defined(__APPLE__) + movw \rd, #:lower16:\val + movt \rd, #:upper16:\val +#else + ldr \rd, =\val +#endif +.endm + +.macro movrelx rd, val, gp + .ifc \rd,\gp + .error "movrelx needs two distinct registers" + .endif + .ifc \rd\()_\gp,r12_ + .warning "movrelx rd=\rd without explicit set gp" + .endif + .ifc \rd\()_\gp,ip_ + .warning "movrelx rd=\rd without explicit set gp" + .endif +#if CONFIG_PIC && defined(__ELF__) + .ifnb \gp + .if .Lpic_gp + .unreq gp + .endif + gp .req \gp + ldpic gp, _GLOBAL_OFFSET_TABLE_ + .elseif !.Lpic_gp + gp .req r12 + ldpic gp, _GLOBAL_OFFSET_TABLE_ + .endif + .set .Lpic_gp, 1 + ldr \rd, .Lpicoff\@ + ldr \rd, [gp, \rd] + def_pic \val(GOT), .Lpicoff\@ +#elif CONFIG_PIC && defined(__APPLE__) + ldpic \rd, .Lpic\@, indir=1 + .non_lazy_symbol_pointer +.Lpic\@: + .indirect_symbol \val + .word 0 + .text +#else + movrel \rd, \val +#endif +.endm + +.macro add_sh rd, rn, rm, sh:vararg +A add \rd, \rn, \rm, \sh +T mov \rm, \rm, \sh +T add \rd, \rn, \rm +.endm + +.macro ldr_pre rt, rn, rm:vararg +A ldr \rt, [\rn, \rm]! +T add \rn, \rn, \rm +T ldr \rt, [\rn] +.endm + +.macro ldr_dpre rt, rn, rm:vararg +A ldr \rt, [\rn, -\rm]! +T sub \rn, \rn, \rm +T ldr \rt, [\rn] +.endm + +.macro ldr_nreg rt, rn, rm:vararg +A ldr \rt, [\rn, -\rm] +T sub \rt, \rn, \rm +T ldr \rt, [\rt] +.endm + +.macro ldr_post rt, rn, rm:vararg +A ldr \rt, [\rn], \rm +T ldr \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro ldrc_pre cc, rt, rn, rm:vararg +A ldr\cc \rt, [\rn, \rm]! +T itt \cc +T add\cc \rn, \rn, \rm +T ldr\cc \rt, [\rn] +.endm + +.macro ldrd_reg rt, rt2, rn, rm +A ldrd \rt, \rt2, [\rn, \rm] +T add \rt, \rn, \rm +T ldrd \rt, \rt2, [\rt] +.endm + +.macro ldrd_post rt, rt2, rn, rm +A ldrd \rt, \rt2, [\rn], \rm +T ldrd \rt, \rt2, [\rn] +T add \rn, \rn, \rm +.endm + +.macro ldrh_pre rt, rn, rm +A ldrh \rt, [\rn, \rm]! +T add \rn, \rn, \rm +T ldrh \rt, [\rn] +.endm + +.macro ldrh_dpre rt, rn, rm +A ldrh \rt, [\rn, -\rm]! +T sub \rn, \rn, \rm +T ldrh \rt, [\rn] +.endm + +.macro ldrh_post rt, rn, rm +A ldrh \rt, [\rn], \rm +T ldrh \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro ldrb_post rt, rn, rm +A ldrb \rt, [\rn], \rm +T ldrb \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro str_post rt, rn, rm:vararg +A str \rt, [\rn], \rm +T str \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro strb_post rt, rn, rm:vararg +A strb \rt, [\rn], \rm +T strb \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro strd_post rt, rt2, rn, rm +A strd \rt, \rt2, [\rn], \rm +T strd \rt, \rt2, [\rn] +T add \rn, \rn, \rm +.endm + +.macro strh_pre rt, rn, rm +A strh \rt, [\rn, \rm]! +T add \rn, \rn, \rm +T strh \rt, [\rn] +.endm + +.macro strh_dpre rt, rn, rm +A strh \rt, [\rn, -\rm]! +T sub \rn, \rn, \rm +T strh \rt, [\rn] +.endm + +.macro strh_post rt, rn, rm +A strh \rt, [\rn], \rm +T strh \rt, [\rn] +T add \rn, \rn, \rm +.endm + +.macro strh_dpost rt, rn, rm +A strh \rt, [\rn], -\rm +T strh \rt, [\rn] +T sub \rn, \rn, \rm +.endm + +#if HAVE_VFP_ARGS +ELF .eabi_attribute 28, 1 +# define VFP +# define NOVFP @ +#else +# define VFP @ +# define NOVFP +#endif + +#define GLUE(a, b) a ## b +#define JOIN(a, b) GLUE(a, b) +#define X(s) JOIN(EXTERN_ASM, s) diff --git a/media/ffvpx/libavutil/arm/bswap.h b/media/ffvpx/libavutil/arm/bswap.h new file mode 100644 index 0000000000..c3460e035d --- /dev/null +++ b/media/ffvpx/libavutil/arm/bswap.h @@ -0,0 +1,70 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_BSWAP_H +#define AVUTIL_ARM_BSWAP_H + +#include <stdint.h> +#include "config.h" +#include "libavutil/attributes.h" + +#ifdef __ARMCC_VERSION + +#if HAVE_ARMV6 +#define av_bswap32 av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + return __rev(x); +} +#endif /* HAVE_ARMV6 */ + +#elif HAVE_INLINE_ASM + +#if HAVE_ARMV6_INLINE +#define av_bswap16 av_bswap16 +static av_always_inline av_const unsigned av_bswap16(unsigned x) +{ + unsigned y; + + __asm__("rev16 %0, %1" : "=r"(y) : "r"(x)); + return y; +} +#endif + +#if AV_GCC_VERSION_AT_MOST(4,4) +#define av_bswap32 av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + uint32_t y; +#if HAVE_ARMV6_INLINE + __asm__("rev %0, %1" : "=r"(y) : "r"(x)); +#else + uint32_t t; + __asm__ ("eor %1, %2, %2, ror #16 \n\t" + "bic %1, %1, #0xFF0000 \n\t" + "mov %0, %2, ror #8 \n\t" + "eor %0, %0, %1, lsr #8 \n\t" + : "=r"(y), "=&r"(t) : "r"(x)); +#endif /* HAVE_ARMV6_INLINE */ + return y; +} +#endif /* AV_GCC_VERSION_AT_MOST(4,4) */ + +#endif /* __ARMCC_VERSION */ + +#endif /* AVUTIL_ARM_BSWAP_H */ diff --git a/media/ffvpx/libavutil/arm/cpu.c b/media/ffvpx/libavutil/arm/cpu.c new file mode 100644 index 0000000000..c84a655c37 --- /dev/null +++ b/media/ffvpx/libavutil/arm/cpu.c @@ -0,0 +1,188 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" +#include "config.h" + +#define CORE_FLAG(f) \ + (AV_CPU_FLAG_ ## f * (HAVE_ ## f ## _EXTERNAL || HAVE_ ## f ## _INLINE)) + +#define CORE_CPU_FLAGS \ + (CORE_FLAG(ARMV5TE) | \ + CORE_FLAG(ARMV6) | \ + CORE_FLAG(ARMV6T2) | \ + CORE_FLAG(VFP) | \ + CORE_FLAG(VFPV3) | \ + CORE_FLAG(NEON)) + +#if defined __linux__ || defined __ANDROID__ + +#include <stdint.h> +#include <stdio.h> +#include <string.h> +#include "libavutil/avstring.h" + +#if HAVE_GETAUXVAL +#include <sys/auxv.h> +#endif + +#define AT_HWCAP 16 + +/* Relevant HWCAP values from kernel headers */ +#define HWCAP_VFP (1 << 6) +#define HWCAP_EDSP (1 << 7) +#define HWCAP_THUMBEE (1 << 11) +#define HWCAP_NEON (1 << 12) +#define HWCAP_VFPv3 (1 << 13) +#define HWCAP_TLS (1 << 15) + +static int get_auxval(uint32_t *hwcap) +{ +#if HAVE_GETAUXVAL + unsigned long ret = getauxval(AT_HWCAP); + if (ret == 0) + return -1; + *hwcap = ret; + return 0; +#else + return -1; +#endif +} + +static int get_hwcap(uint32_t *hwcap) +{ + struct { uint32_t a_type; uint32_t a_val; } auxv; + FILE *f = fopen("/proc/self/auxv", "r"); + int err = -1; + + if (!f) + return -1; + + while (fread(&auxv, sizeof(auxv), 1, f) > 0) { + if (auxv.a_type == AT_HWCAP) { + *hwcap = auxv.a_val; + err = 0; + break; + } + } + + fclose(f); + return err; +} + +static int get_cpuinfo(uint32_t *hwcap) +{ + FILE *f = fopen("/proc/cpuinfo", "r"); + char buf[200]; + + if (!f) + return -1; + + *hwcap = 0; + while (fgets(buf, sizeof(buf), f)) { + if (av_strstart(buf, "Features", NULL)) { + if (strstr(buf, " edsp ")) + *hwcap |= HWCAP_EDSP; + if (strstr(buf, " tls ")) + *hwcap |= HWCAP_TLS; + if (strstr(buf, " thumbee ")) + *hwcap |= HWCAP_THUMBEE; + if (strstr(buf, " vfp ")) + *hwcap |= HWCAP_VFP; + if (strstr(buf, " vfpv3 ")) + *hwcap |= HWCAP_VFPv3; + if (strstr(buf, " neon ") || strstr(buf, " asimd ")) + *hwcap |= HWCAP_NEON; + if (strstr(buf, " fp ")) // Listed on 64 bit ARMv8 kernels + *hwcap |= HWCAP_VFP | HWCAP_VFPv3; + break; + } + } + fclose(f); + return 0; +} + +int ff_get_cpu_flags_arm(void) +{ + int flags = CORE_CPU_FLAGS; + uint32_t hwcap; + + if (get_auxval(&hwcap) < 0) + if (get_hwcap(&hwcap) < 0) + if (get_cpuinfo(&hwcap) < 0) + return flags; + +#define check_cap(cap, flag) do { \ + if (hwcap & HWCAP_ ## cap) \ + flags |= AV_CPU_FLAG_ ## flag; \ + } while (0) + + /* No flags explicitly indicate v6 or v6T2 so check others which + imply support. */ + check_cap(EDSP, ARMV5TE); + check_cap(TLS, ARMV6); + check_cap(THUMBEE, ARMV6T2); + check_cap(VFP, VFP); + check_cap(VFPv3, VFPV3); + check_cap(NEON, NEON); + + /* The v6 checks above are not reliable so let higher flags + trickle down. */ + if (flags & (AV_CPU_FLAG_VFPV3 | AV_CPU_FLAG_NEON)) + flags |= AV_CPU_FLAG_ARMV6T2; + else if (flags & (AV_CPU_FLAG_ARMV6T2 | AV_CPU_FLAG_ARMV6)) + /* Some functions use the 'setend' instruction which is deprecated on ARMv8 + * and serializing on some ARMv7 cores. This ensures such functions + * are only enabled on ARMv6. */ + flags |= AV_CPU_FLAG_SETEND; + + if (flags & AV_CPU_FLAG_ARMV6T2) + flags |= AV_CPU_FLAG_ARMV6; + + /* set the virtual VFPv2 vector mode flag */ + if ((flags & AV_CPU_FLAG_VFP) && !(flags & (AV_CPU_FLAG_VFPV3 | AV_CPU_FLAG_NEON))) + flags |= AV_CPU_FLAG_VFP_VM; + + return flags; +} + +#else + +int ff_get_cpu_flags_arm(void) +{ + return AV_CPU_FLAG_ARMV5TE * HAVE_ARMV5TE | + AV_CPU_FLAG_ARMV6 * HAVE_ARMV6 | + AV_CPU_FLAG_ARMV6T2 * HAVE_ARMV6T2 | + AV_CPU_FLAG_VFP * HAVE_VFP | + AV_CPU_FLAG_VFPV3 * HAVE_VFPV3 | + AV_CPU_FLAG_NEON * HAVE_NEON | + AV_CPU_FLAG_SETEND * !(HAVE_NEON | HAVE_VFPV3); +} + +#endif + +size_t ff_get_cpu_max_align_arm(void) +{ + int flags = av_get_cpu_flags(); + + if (flags & AV_CPU_FLAG_NEON) + return 16; + + return 8; +} diff --git a/media/ffvpx/libavutil/arm/cpu.h b/media/ffvpx/libavutil/arm/cpu.h new file mode 100644 index 0000000000..1d6cc65dc4 --- /dev/null +++ b/media/ffvpx/libavutil/arm/cpu.h @@ -0,0 +1,38 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_CPU_H +#define AVUTIL_ARM_CPU_H + +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" + +#define have_armv5te(flags) CPUEXT(flags, ARMV5TE) +#define have_armv6(flags) CPUEXT(flags, ARMV6) +#define have_armv6t2(flags) CPUEXT(flags, ARMV6T2) +#define have_vfp(flags) CPUEXT(flags, VFP) +#define have_vfpv3(flags) CPUEXT(flags, VFPV3) +#define have_neon(flags) CPUEXT(flags, NEON) +#define have_setend(flags) CPUEXT(flags, SETEND) + +/* some functions use the VFPv2 vector mode which is deprecated in ARMv7-A + * and might trap on such CPU depending on the OS configuration */ +#define have_vfp_vm(flags) \ + (HAVE_VFP && ((flags) & AV_CPU_FLAG_VFP_VM)) + +#endif /* AVUTIL_ARM_CPU_H */ diff --git a/media/ffvpx/libavutil/arm/float_dsp_arm.h b/media/ffvpx/libavutil/arm/float_dsp_arm.h new file mode 100644 index 0000000000..fe311cc0d2 --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_arm.h @@ -0,0 +1,29 @@ +/* + * Copyright (c) 2009 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_FLOAT_DSP_ARM_H +#define AVUTIL_ARM_FLOAT_DSP_ARM_H + +#include "libavutil/float_dsp.h" + +void ff_float_dsp_init_vfp(AVFloatDSPContext *fdsp, int cpu_flags); +void ff_float_dsp_init_neon(AVFloatDSPContext *fdsp); + +#endif /* AVUTIL_ARM_FLOAT_DSP_ARM_H */ diff --git a/media/ffvpx/libavutil/arm/float_dsp_init_arm.c b/media/ffvpx/libavutil/arm/float_dsp_init_arm.c new file mode 100644 index 0000000000..678762862e --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_init_arm.c @@ -0,0 +1,32 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/attributes.h" +#include "libavutil/float_dsp.h" +#include "cpu.h" +#include "float_dsp_arm.h" + +av_cold void ff_float_dsp_init_arm(AVFloatDSPContext *fdsp) +{ + int cpu_flags = av_get_cpu_flags(); + + if (have_vfp(cpu_flags)) + ff_float_dsp_init_vfp(fdsp, cpu_flags); + if (have_neon(cpu_flags)) + ff_float_dsp_init_neon(fdsp); +} diff --git a/media/ffvpx/libavutil/arm/float_dsp_init_neon.c b/media/ffvpx/libavutil/arm/float_dsp_init_neon.c new file mode 100644 index 0000000000..689aa77c7b --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_init_neon.c @@ -0,0 +1,59 @@ +/* + * ARM NEON optimised Float DSP functions + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> + +#include "libavutil/attributes.h" +#include "libavutil/float_dsp.h" +#include "float_dsp_arm.h" + +void ff_vector_fmul_neon(float *dst, const float *src0, const float *src1, int len); + +void ff_vector_fmac_scalar_neon(float *dst, const float *src, float mul, + int len); + +void ff_vector_fmul_scalar_neon(float *dst, const float *src, float mul, + int len); + +void ff_vector_fmul_window_neon(float *dst, const float *src0, + const float *src1, const float *win, int len); + +void ff_vector_fmul_add_neon(float *dst, const float *src0, const float *src1, + const float *src2, int len); + +void ff_vector_fmul_reverse_neon(float *dst, const float *src0, + const float *src1, int len); + +void ff_butterflies_float_neon(float *v1, float *v2, int len); + +float ff_scalarproduct_float_neon(const float *v1, const float *v2, int len); + +av_cold void ff_float_dsp_init_neon(AVFloatDSPContext *fdsp) +{ + fdsp->vector_fmul = ff_vector_fmul_neon; + fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_neon; + fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_neon; + fdsp->vector_fmul_window = ff_vector_fmul_window_neon; + fdsp->vector_fmul_add = ff_vector_fmul_add_neon; + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_neon; + fdsp->butterflies_float = ff_butterflies_float_neon; + fdsp->scalarproduct_float = ff_scalarproduct_float_neon; +} diff --git a/media/ffvpx/libavutil/arm/float_dsp_init_vfp.c b/media/ffvpx/libavutil/arm/float_dsp_init_vfp.c new file mode 100644 index 0000000000..05873e7e37 --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_init_vfp.c @@ -0,0 +1,46 @@ +/* + * Copyright (c) 2008 Siarhei Siamashka <ssvb@users.sourceforge.net> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/attributes.h" +#include "libavutil/float_dsp.h" +#include "cpu.h" +#include "float_dsp_arm.h" + +void ff_vector_fmul_vfp(float *dst, const float *src0, const float *src1, + int len); + +void ff_vector_fmul_window_vfp(float *dst, const float *src0, + const float *src1, const float *win, int len); + +void ff_vector_fmul_reverse_vfp(float *dst, const float *src0, + const float *src1, int len); + +void ff_butterflies_float_vfp(float *av_restrict v1, float *av_restrict v2, int len); + +av_cold void ff_float_dsp_init_vfp(AVFloatDSPContext *fdsp, int cpu_flags) +{ + if (have_vfp_vm(cpu_flags)) { + fdsp->vector_fmul = ff_vector_fmul_vfp; + fdsp->vector_fmul_window = ff_vector_fmul_window_vfp; + } + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_vfp; + if (have_vfp_vm(cpu_flags)) + fdsp->butterflies_float = ff_butterflies_float_vfp; +} diff --git a/media/ffvpx/libavutil/arm/float_dsp_neon.S b/media/ffvpx/libavutil/arm/float_dsp_neon.S new file mode 100644 index 0000000000..3823227608 --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_neon.S @@ -0,0 +1,271 @@ +/* + * ARM NEON optimised Float DSP functions + * Copyright (c) 2008 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "asm.S" + +function ff_vector_fmul_neon, export=1 + subs r3, r3, #8 + vld1.32 {d0-d3}, [r1,:128]! + vld1.32 {d4-d7}, [r2,:128]! + vmul.f32 q8, q0, q2 + vmul.f32 q9, q1, q3 + beq 3f + bics ip, r3, #15 + beq 2f +1: subs ip, ip, #16 + vld1.32 {d0-d1}, [r1,:128]! + vld1.32 {d4-d5}, [r2,:128]! + vmul.f32 q10, q0, q2 + vld1.32 {d2-d3}, [r1,:128]! + vld1.32 {d6-d7}, [r2,:128]! + vmul.f32 q11, q1, q3 + vst1.32 {d16-d19},[r0,:128]! + vld1.32 {d0-d1}, [r1,:128]! + vld1.32 {d4-d5}, [r2,:128]! + vmul.f32 q8, q0, q2 + vld1.32 {d2-d3}, [r1,:128]! + vld1.32 {d6-d7}, [r2,:128]! + vmul.f32 q9, q1, q3 + vst1.32 {d20-d23},[r0,:128]! + bne 1b + ands r3, r3, #15 + beq 3f +2: vld1.32 {d0-d1}, [r1,:128]! + vld1.32 {d4-d5}, [r2,:128]! + vst1.32 {d16-d17},[r0,:128]! + vmul.f32 q8, q0, q2 + vld1.32 {d2-d3}, [r1,:128]! + vld1.32 {d6-d7}, [r2,:128]! + vst1.32 {d18-d19},[r0,:128]! + vmul.f32 q9, q1, q3 +3: vst1.32 {d16-d19},[r0,:128]! + bx lr +endfunc + +function ff_vector_fmac_scalar_neon, export=1 +VFP len .req r2 +VFP acc .req r3 +NOVFP len .req r3 +NOVFP acc .req r2 +VFP vdup.32 q15, d0[0] +NOVFP vdup.32 q15, r2 + bics r12, len, #15 + mov acc, r0 + beq 3f + vld1.32 {q0}, [r1,:128]! + vld1.32 {q8}, [acc,:128]! + vld1.32 {q1}, [r1,:128]! + vld1.32 {q9}, [acc,:128]! +1: vmla.f32 q8, q0, q15 + vld1.32 {q2}, [r1,:128]! + vld1.32 {q10}, [acc,:128]! + vmla.f32 q9, q1, q15 + vld1.32 {q3}, [r1,:128]! + vld1.32 {q11}, [acc,:128]! + vmla.f32 q10, q2, q15 + vst1.32 {q8}, [r0,:128]! + vmla.f32 q11, q3, q15 + vst1.32 {q9}, [r0,:128]! + subs r12, r12, #16 + beq 2f + vld1.32 {q0}, [r1,:128]! + vld1.32 {q8}, [acc,:128]! + vst1.32 {q10}, [r0,:128]! + vld1.32 {q1}, [r1,:128]! + vld1.32 {q9}, [acc,:128]! + vst1.32 {q11}, [r0,:128]! + b 1b +2: vst1.32 {q10}, [r0,:128]! + vst1.32 {q11}, [r0,:128]! + ands len, len, #15 + it eq + bxeq lr +3: vld1.32 {q0}, [r1,:128]! + vld1.32 {q8}, [acc,:128]! + vmla.f32 q8, q0, q15 + vst1.32 {q8}, [r0,:128]! + subs len, len, #4 + bgt 3b + bx lr + .unreq len +endfunc + +function ff_vector_fmul_scalar_neon, export=1 +VFP len .req r2 +NOVFP len .req r3 +VFP vdup.32 q8, d0[0] +NOVFP vdup.32 q8, r2 + bics r12, len, #15 + beq 3f + vld1.32 {q0},[r1,:128]! + vld1.32 {q1},[r1,:128]! +1: vmul.f32 q0, q0, q8 + vld1.32 {q2},[r1,:128]! + vmul.f32 q1, q1, q8 + vld1.32 {q3},[r1,:128]! + vmul.f32 q2, q2, q8 + vst1.32 {q0},[r0,:128]! + vmul.f32 q3, q3, q8 + vst1.32 {q1},[r0,:128]! + subs r12, r12, #16 + beq 2f + vld1.32 {q0},[r1,:128]! + vst1.32 {q2},[r0,:128]! + vld1.32 {q1},[r1,:128]! + vst1.32 {q3},[r0,:128]! + b 1b +2: vst1.32 {q2},[r0,:128]! + vst1.32 {q3},[r0,:128]! + ands len, len, #15 + it eq + bxeq lr +3: vld1.32 {q0},[r1,:128]! + vmul.f32 q0, q0, q8 + vst1.32 {q0},[r0,:128]! + subs len, len, #4 + bgt 3b + bx lr + .unreq len +endfunc + +function ff_vector_fmul_window_neon, export=1 + push {r4,r5,lr} + ldr lr, [sp, #12] + sub r2, r2, #8 + sub r5, lr, #2 + add r2, r2, r5, lsl #2 + add r4, r3, r5, lsl #3 + add ip, r0, r5, lsl #3 + mov r5, #-16 + vld1.32 {d0,d1}, [r1,:128]! + vld1.32 {d2,d3}, [r2,:128], r5 + vld1.32 {d4,d5}, [r3,:128]! + vld1.32 {d6,d7}, [r4,:128], r5 +1: subs lr, lr, #4 + vmul.f32 d22, d0, d4 + vrev64.32 q3, q3 + vmul.f32 d23, d1, d5 + vrev64.32 q1, q1 + vmul.f32 d20, d0, d7 + vmul.f32 d21, d1, d6 + beq 2f + vmla.f32 d22, d3, d7 + vld1.32 {d0,d1}, [r1,:128]! + vmla.f32 d23, d2, d6 + vld1.32 {d18,d19},[r2,:128], r5 + vmls.f32 d20, d3, d4 + vld1.32 {d24,d25},[r3,:128]! + vmls.f32 d21, d2, d5 + vld1.32 {d6,d7}, [r4,:128], r5 + vmov q1, q9 + vrev64.32 q11, q11 + vmov q2, q12 + vswp d22, d23 + vst1.32 {d20,d21},[r0,:128]! + vst1.32 {d22,d23},[ip,:128], r5 + b 1b +2: vmla.f32 d22, d3, d7 + vmla.f32 d23, d2, d6 + vmls.f32 d20, d3, d4 + vmls.f32 d21, d2, d5 + vrev64.32 q11, q11 + vswp d22, d23 + vst1.32 {d20,d21},[r0,:128]! + vst1.32 {d22,d23},[ip,:128], r5 + pop {r4,r5,pc} +endfunc + +function ff_vector_fmul_add_neon, export=1 + ldr r12, [sp] + vld1.32 {q0-q1}, [r1,:128]! + vld1.32 {q8-q9}, [r2,:128]! + vld1.32 {q2-q3}, [r3,:128]! + vmul.f32 q10, q0, q8 + vmul.f32 q11, q1, q9 +1: vadd.f32 q12, q2, q10 + vadd.f32 q13, q3, q11 + pld [r1, #16] + pld [r2, #16] + pld [r3, #16] + subs r12, r12, #8 + beq 2f + vld1.32 {q0}, [r1,:128]! + vld1.32 {q8}, [r2,:128]! + vmul.f32 q10, q0, q8 + vld1.32 {q1}, [r1,:128]! + vld1.32 {q9}, [r2,:128]! + vmul.f32 q11, q1, q9 + vld1.32 {q2-q3}, [r3,:128]! + vst1.32 {q12-q13},[r0,:128]! + b 1b +2: vst1.32 {q12-q13},[r0,:128]! + bx lr +endfunc + +function ff_vector_fmul_reverse_neon, export=1 + add r2, r2, r3, lsl #2 + sub r2, r2, #32 + mov r12, #-32 + vld1.32 {q0-q1}, [r1,:128]! + vld1.32 {q2-q3}, [r2,:128], r12 +1: pld [r1, #32] + vrev64.32 q3, q3 + vmul.f32 d16, d0, d7 + vmul.f32 d17, d1, d6 + pld [r2, #-32] + vrev64.32 q2, q2 + vmul.f32 d18, d2, d5 + vmul.f32 d19, d3, d4 + subs r3, r3, #8 + beq 2f + vld1.32 {q0-q1}, [r1,:128]! + vld1.32 {q2-q3}, [r2,:128], r12 + vst1.32 {q8-q9}, [r0,:128]! + b 1b +2: vst1.32 {q8-q9}, [r0,:128]! + bx lr +endfunc + +function ff_butterflies_float_neon, export=1 +1: vld1.32 {q0},[r0,:128] + vld1.32 {q1},[r1,:128] + vsub.f32 q2, q0, q1 + vadd.f32 q1, q0, q1 + vst1.32 {q2},[r1,:128]! + vst1.32 {q1},[r0,:128]! + subs r2, r2, #4 + bgt 1b + bx lr +endfunc + +function ff_scalarproduct_float_neon, export=1 + vmov.f32 q2, #0.0 +1: vld1.32 {q0},[r0,:128]! + vld1.32 {q1},[r1,:128]! + vmla.f32 q2, q0, q1 + subs r2, r2, #4 + bgt 1b + vadd.f32 d0, d4, d5 + vpadd.f32 d0, d0, d0 +NOVFP vmov.32 r0, d0[0] + bx lr +endfunc diff --git a/media/ffvpx/libavutil/arm/float_dsp_vfp.S b/media/ffvpx/libavutil/arm/float_dsp_vfp.S new file mode 100644 index 0000000000..7db2452284 --- /dev/null +++ b/media/ffvpx/libavutil/arm/float_dsp_vfp.S @@ -0,0 +1,457 @@ +/* + * Copyright (c) 2008 Siarhei Siamashka <ssvb@users.sourceforge.net> + * + * This file is part of FFmpeg + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "asm.S" + +/** + * Assume that len is a positive number and is multiple of 8 + */ +@ void ff_vector_fmul_vfp(float *dst, const float *src0, const float *src1, int len) +function ff_vector_fmul_vfp, export=1 + vpush {d8-d15} + fmrx r12, fpscr + orr r12, r12, #(3 << 16) /* set vector size to 4 */ + fmxr fpscr, r12 + + vldmia r1!, {s0-s3} + vldmia r2!, {s8-s11} + vldmia r1!, {s4-s7} + vldmia r2!, {s12-s15} + vmul.f32 s8, s0, s8 +1: + subs r3, r3, #16 + vmul.f32 s12, s4, s12 + itttt ge + vldmiage r1!, {s16-s19} + vldmiage r2!, {s24-s27} + vldmiage r1!, {s20-s23} + vldmiage r2!, {s28-s31} + it ge + vmulge.f32 s24, s16, s24 + vstmia r0!, {s8-s11} + vstmia r0!, {s12-s15} + it ge + vmulge.f32 s28, s20, s28 + itttt gt + vldmiagt r1!, {s0-s3} + vldmiagt r2!, {s8-s11} + vldmiagt r1!, {s4-s7} + vldmiagt r2!, {s12-s15} + ittt ge + vmulge.f32 s8, s0, s8 + vstmiage r0!, {s24-s27} + vstmiage r0!, {s28-s31} + bgt 1b + + bic r12, r12, #(7 << 16) /* set vector size back to 1 */ + fmxr fpscr, r12 + vpop {d8-d15} + bx lr +endfunc + +/** + * ARM VFP implementation of 'vector_fmul_window_c' function + * Assume that len is a positive non-zero number + */ +@ void ff_vector_fmul_window_vfp(float *dst, const float *src0, +@ const float *src1, const float *win, int len) +function ff_vector_fmul_window_vfp, export=1 +DST0 .req a1 +SRC0 .req a2 +SRC1 .req a3 +WIN0 .req a4 +LEN .req v1 +DST1 .req v2 +WIN1 .req v3 +OLDFPSCR .req ip + + push {v1-v3,lr} + ldr LEN, [sp, #4*4+0] + vpush {s16-s31} + fmrx OLDFPSCR, FPSCR + add DST1, DST0, LEN, lsl #3 + add SRC1, SRC1, LEN, lsl #2 + add WIN1, WIN0, LEN, lsl #3 + + tst LEN, #7 + beq 4f @ common case: len is a multiple of 8 + + ldr lr, =0x03000000 @ RunFast mode, scalar mode + fmxr FPSCR, lr + + tst LEN, #1 + beq 1f + vldmdb WIN1!, {s0} + vldmia SRC0!, {s8} + vldmia WIN0!, {s16} + vmul.f s24, s0, s8 + vldmdb SRC1!, {s20} + vmul.f s8, s16, s8 + vmls.f s24, s16, s20 + vmla.f s8, s0, s20 + vstmia DST0!, {s24} + vstmdb DST1!, {s8} +1: + tst LEN, #2 + beq 2f + vldmdb WIN1!, {s0} + vldmdb WIN1!, {s1} + vldmia SRC0!, {s8-s9} + vldmia WIN0!, {s16-s17} + vmul.f s24, s0, s8 + vmul.f s25, s1, s9 + vldmdb SRC1!, {s20} + vldmdb SRC1!, {s21} + vmul.f s8, s16, s8 + vmul.f s9, s17, s9 + vmls.f s24, s16, s20 + vmls.f s25, s17, s21 + vmla.f s8, s0, s20 + vmla.f s9, s1, s21 + vstmia DST0!, {s24-s25} + vstmdb DST1!, {s8} + vstmdb DST1!, {s9} +2: + tst LEN, #4 + beq 3f + vldmdb WIN1!, {s0} + vldmdb WIN1!, {s1} + vldmdb WIN1!, {s2} + vldmdb WIN1!, {s3} + vldmia SRC0!, {s8-s11} + vldmia WIN0!, {s16-s19} + vmul.f s24, s0, s8 + vmul.f s25, s1, s9 + vmul.f s26, s2, s10 + vmul.f s27, s3, s11 + vldmdb SRC1!, {s20} + vldmdb SRC1!, {s21} + vldmdb SRC1!, {s22} + vldmdb SRC1!, {s23} + vmul.f s8, s16, s8 + vmul.f s9, s17, s9 + vmul.f s10, s18, s10 + vmul.f s11, s19, s11 + vmls.f s24, s16, s20 + vmls.f s25, s17, s21 + vmls.f s26, s18, s22 + vmls.f s27, s19, s23 + vmla.f s8, s0, s20 + vmla.f s9, s1, s21 + vmla.f s10, s2, s22 + vmla.f s11, s3, s23 + vstmia DST0!, {s24-s27} + vstmdb DST1!, {s8} + vstmdb DST1!, {s9} + vstmdb DST1!, {s10} + vstmdb DST1!, {s11} +3: + bics LEN, LEN, #7 + beq 7f +4: + ldr lr, =0x03030000 @ RunFast mode, short vectors of length 4, stride 1 + fmxr FPSCR, lr + + vldmdb WIN1!, {s0} + vldmdb WIN1!, {s1} + vldmdb WIN1!, {s2} + vldmdb WIN1!, {s3} + vldmia SRC0!, {s8-s11} + vldmia WIN0!, {s16-s19} + vmul.f s24, s0, s8 @ vector * vector + vldmdb SRC1!, {s20} + vldmdb SRC1!, {s21} + vldmdb SRC1!, {s22} + vldmdb SRC1!, {s23} + vmul.f s8, s16, s8 @ vector * vector + vmls.f s24, s16, s20 @ vector * vector + vldmdb WIN1!, {s4} + vldmdb WIN1!, {s5} + vldmdb WIN1!, {s6} + vldmdb WIN1!, {s7} + vldmia SRC0!, {s12-s13} + vmla.f s8, s0, s20 @ vector * vector + vldmia SRC0!, {s14-s15} + subs LEN, LEN, #8 + beq 6f +5: vldmia WIN0!, {s20-s23} + vmul.f s28, s4, s12 @ vector * vector + vstmia DST0!, {s24-s25} + vldmdb SRC1!, {s16} + vldmdb SRC1!, {s17} + vldmdb SRC1!, {s18} + vldmdb SRC1!, {s19} + vmul.f s12, s20, s12 @ vector * vector + vstmia DST0!, {s26-s27} + vstmdb DST1!, {s8} + vstmdb DST1!, {s9} + vstmdb DST1!, {s10} + vstmdb DST1!, {s11} + vmls.f s28, s20, s16 @ vector * vector + vldmdb WIN1!, {s0} + vldmdb WIN1!, {s1} + vldmdb WIN1!, {s2} + vldmdb WIN1!, {s3} + vldmia SRC0!, {s8-s9} + vmla.f s12, s4, s16 @ vector * vector + vldmia SRC0!, {s10-s11} + subs LEN, LEN, #8 + vldmia WIN0!, {s16-s19} + vmul.f s24, s0, s8 @ vector * vector + vstmia DST0!, {s28-s29} + vldmdb SRC1!, {s20} + vldmdb SRC1!, {s21} + vldmdb SRC1!, {s22} + vldmdb SRC1!, {s23} + vmul.f s8, s16, s8 @ vector * vector + vstmia DST0!, {s30-s31} + vstmdb DST1!, {s12} + vstmdb DST1!, {s13} + vstmdb DST1!, {s14} + vstmdb DST1!, {s15} + vmls.f s24, s16, s20 @ vector * vector + vldmdb WIN1!, {s4} + vldmdb WIN1!, {s5} + vldmdb WIN1!, {s6} + vldmdb WIN1!, {s7} + vldmia SRC0!, {s12-s13} + vmla.f s8, s0, s20 @ vector * vector + vldmia SRC0!, {s14-s15} + bne 5b +6: vldmia WIN0!, {s20-s23} + vmul.f s28, s4, s12 @ vector * vector + vstmia DST0!, {s24-s25} + vldmdb SRC1!, {s16} + vldmdb SRC1!, {s17} + vldmdb SRC1!, {s18} + vldmdb SRC1!, {s19} + vmul.f s12, s20, s12 @ vector * vector + vstmia DST0!, {s26-s27} + vstmdb DST1!, {s8} + vstmdb DST1!, {s9} + vstmdb DST1!, {s10} + vstmdb DST1!, {s11} + vmls.f s28, s20, s16 @ vector * vector + vmla.f s12, s4, s16 @ vector * vector + vstmia DST0!, {s28-s31} + vstmdb DST1!, {s12} + vstmdb DST1!, {s13} + vstmdb DST1!, {s14} + vstmdb DST1!, {s15} +7: + fmxr FPSCR, OLDFPSCR + vpop {s16-s31} + pop {v1-v3,pc} + + .unreq DST0 + .unreq SRC0 + .unreq SRC1 + .unreq WIN0 + .unreq LEN + .unreq OLDFPSCR + .unreq DST1 + .unreq WIN1 +endfunc + +/** + * ARM VFP optimized implementation of 'vector_fmul_reverse_c' function. + * Assume that len is a positive number and is multiple of 8 + */ +@ void ff_vector_fmul_reverse_vfp(float *dst, const float *src0, +@ const float *src1, int len) +function ff_vector_fmul_reverse_vfp, export=1 + vpush {d8-d15} + add r2, r2, r3, lsl #2 + vldmdb r2!, {s0-s3} + vldmia r1!, {s8-s11} + vldmdb r2!, {s4-s7} + vldmia r1!, {s12-s15} + vmul.f32 s8, s3, s8 + vmul.f32 s9, s2, s9 + vmul.f32 s10, s1, s10 + vmul.f32 s11, s0, s11 +1: + subs r3, r3, #16 + it ge + vldmdbge r2!, {s16-s19} + vmul.f32 s12, s7, s12 + it ge + vldmiage r1!, {s24-s27} + vmul.f32 s13, s6, s13 + it ge + vldmdbge r2!, {s20-s23} + vmul.f32 s14, s5, s14 + it ge + vldmiage r1!, {s28-s31} + vmul.f32 s15, s4, s15 + it ge + vmulge.f32 s24, s19, s24 + it gt + vldmdbgt r2!, {s0-s3} + it ge + vmulge.f32 s25, s18, s25 + vstmia r0!, {s8-s13} + it ge + vmulge.f32 s26, s17, s26 + it gt + vldmiagt r1!, {s8-s11} + itt ge + vmulge.f32 s27, s16, s27 + vmulge.f32 s28, s23, s28 + it gt + vldmdbgt r2!, {s4-s7} + it ge + vmulge.f32 s29, s22, s29 + vstmia r0!, {s14-s15} + ittt ge + vmulge.f32 s30, s21, s30 + vmulge.f32 s31, s20, s31 + vmulge.f32 s8, s3, s8 + it gt + vldmiagt r1!, {s12-s15} + itttt ge + vmulge.f32 s9, s2, s9 + vmulge.f32 s10, s1, s10 + vstmiage r0!, {s24-s27} + vmulge.f32 s11, s0, s11 + it ge + vstmiage r0!, {s28-s31} + bgt 1b + + vpop {d8-d15} + bx lr +endfunc + +/** + * ARM VFP implementation of 'butterflies_float_c' function + * Assume that len is a positive non-zero number + */ +@ void ff_butterflies_float_vfp(float *restrict v1, float *restrict v2, int len) +function ff_butterflies_float_vfp, export=1 +BASE1 .req a1 +BASE2 .req a2 +LEN .req a3 +OLDFPSCR .req a4 + + vpush {s16-s31} + fmrx OLDFPSCR, FPSCR + + tst LEN, #7 + beq 4f @ common case: len is a multiple of 8 + + ldr ip, =0x03000000 @ RunFast mode, scalar mode + fmxr FPSCR, ip + + tst LEN, #1 + beq 1f + vldmia BASE1!, {s0} + vldmia BASE2!, {s8} + vadd.f s16, s0, s8 + vsub.f s24, s0, s8 + vstr s16, [BASE1, #0-4*1] + vstr s24, [BASE2, #0-4*1] +1: + tst LEN, #2 + beq 2f + vldmia BASE1!, {s0-s1} + vldmia BASE2!, {s8-s9} + vadd.f s16, s0, s8 + vadd.f s17, s1, s9 + vsub.f s24, s0, s8 + vsub.f s25, s1, s9 + vstr d8, [BASE1, #0-8*1] @ s16,s17 + vstr d12, [BASE2, #0-8*1] @ s24,s25 +2: + tst LEN, #4 + beq 3f + vldmia BASE1!, {s0-s1} + vldmia BASE2!, {s8-s9} + vldmia BASE1!, {s2-s3} + vldmia BASE2!, {s10-s11} + vadd.f s16, s0, s8 + vadd.f s17, s1, s9 + vsub.f s24, s0, s8 + vsub.f s25, s1, s9 + vadd.f s18, s2, s10 + vadd.f s19, s3, s11 + vsub.f s26, s2, s10 + vsub.f s27, s3, s11 + vstr d8, [BASE1, #0-16*1] @ s16,s17 + vstr d12, [BASE2, #0-16*1] @ s24,s25 + vstr d9, [BASE1, #8-16*1] @ s18,s19 + vstr d13, [BASE2, #8-16*1] @ s26,s27 +3: + bics LEN, LEN, #7 + beq 7f +4: + ldr ip, =0x03030000 @ RunFast mode, short vectors of length 4, stride 1 + fmxr FPSCR, ip + + vldmia BASE1!, {s0-s1} + vldmia BASE2!, {s8-s9} + vldmia BASE1!, {s2-s3} + vldmia BASE2!, {s10-s11} + vadd.f s16, s0, s8 + vldmia BASE1!, {s4-s5} + vldmia BASE2!, {s12-s13} + vldmia BASE1!, {s6-s7} + vldmia BASE2!, {s14-s15} + vsub.f s24, s0, s8 + vadd.f s20, s4, s12 + subs LEN, LEN, #8 + beq 6f +5: vldmia BASE1!, {s0-s3} + vldmia BASE2!, {s8-s11} + vsub.f s28, s4, s12 + vstr d8, [BASE1, #0-16*3] @ s16,s17 + vstr d9, [BASE1, #8-16*3] @ s18,s19 + vstr d12, [BASE2, #0-16*3] @ s24,s25 + vstr d13, [BASE2, #8-16*3] @ s26,s27 + vadd.f s16, s0, s8 + vldmia BASE1!, {s4-s7} + vldmia BASE2!, {s12-s15} + vsub.f s24, s0, s8 + vstr d10, [BASE1, #0-16*3] @ s20,s21 + vstr d11, [BASE1, #8-16*3] @ s22,s23 + vstr d14, [BASE2, #0-16*3] @ s28,s29 + vstr d15, [BASE2, #8-16*3] @ s30,s31 + vadd.f s20, s4, s12 + subs LEN, LEN, #8 + bne 5b +6: vsub.f s28, s4, s12 + vstr d8, [BASE1, #0-16*2] @ s16,s17 + vstr d9, [BASE1, #8-16*2] @ s18,s19 + vstr d12, [BASE2, #0-16*2] @ s24,s25 + vstr d13, [BASE2, #8-16*2] @ s26,s27 + vstr d10, [BASE1, #0-16*1] @ s20,s21 + vstr d11, [BASE1, #8-16*1] @ s22,s23 + vstr d14, [BASE2, #0-16*1] @ s28,s29 + vstr d15, [BASE2, #8-16*1] @ s30,s31 +7: + fmxr FPSCR, OLDFPSCR + vpop {s16-s31} + bx lr + + .unreq BASE1 + .unreq BASE2 + .unreq LEN + .unreq OLDFPSCR +endfunc diff --git a/media/ffvpx/libavutil/arm/intmath.h b/media/ffvpx/libavutil/arm/intmath.h new file mode 100644 index 0000000000..f19b21e98d --- /dev/null +++ b/media/ffvpx/libavutil/arm/intmath.h @@ -0,0 +1,146 @@ +/* + * Copyright (c) 2010 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_INTMATH_H +#define AVUTIL_ARM_INTMATH_H + +#include <stdint.h> + +#include "config.h" +#include "libavutil/attributes.h" + +#if HAVE_INLINE_ASM + +#if HAVE_ARMV6_INLINE + +#define av_clip_uint8 av_clip_uint8_arm +static av_always_inline av_const int av_clip_uint8_arm(int a) +{ + int x; + __asm__ ("usat %0, #8, %1" : "=r"(x) : "r"(a)); + return x; +} + +#define av_clip_int8 av_clip_int8_arm +static av_always_inline av_const int av_clip_int8_arm(int a) +{ + int x; + __asm__ ("ssat %0, #8, %1" : "=r"(x) : "r"(a)); + return x; +} + +#define av_clip_uint16 av_clip_uint16_arm +static av_always_inline av_const int av_clip_uint16_arm(int a) +{ + int x; + __asm__ ("usat %0, #16, %1" : "=r"(x) : "r"(a)); + return x; +} + +#define av_clip_int16 av_clip_int16_arm +static av_always_inline av_const int av_clip_int16_arm(int a) +{ + int x; + __asm__ ("ssat %0, #16, %1" : "=r"(x) : "r"(a)); + return x; +} + +#define av_clip_intp2 av_clip_intp2_arm +static av_always_inline av_const int av_clip_intp2_arm(int a, int p) +{ + if (av_builtin_constant_p(p)) { + unsigned x; + __asm__ ("ssat %0, %2, %1" : "=r"(x) : "r"(a), "i"(p+1)); + return x; + } else { + if (((unsigned)a + (1 << p)) & ~((2 << p) - 1)) + return (a >> 31) ^ ((1 << p) - 1); + else + return a; + } +} + +#define av_clip_uintp2 av_clip_uintp2_arm +static av_always_inline av_const unsigned av_clip_uintp2_arm(int a, int p) +{ + if (av_builtin_constant_p(p)) { + unsigned x; + __asm__ ("usat %0, %2, %1" : "=r"(x) : "r"(a), "i"(p)); + return x; + } else { + if (a & ~((1<<p) - 1)) return (~a) >> 31 & ((1<<p) - 1); + else return a; + } +} + +#define av_sat_add32 av_sat_add32_arm +static av_always_inline int av_sat_add32_arm(int a, int b) +{ + int r; + __asm__ ("qadd %0, %1, %2" : "=r"(r) : "r"(a), "r"(b)); + return r; +} + +#define av_sat_dadd32 av_sat_dadd32_arm +static av_always_inline int av_sat_dadd32_arm(int a, int b) +{ + int r; + __asm__ ("qdadd %0, %1, %2" : "=r"(r) : "r"(a), "r"(b)); + return r; +} + +#define av_sat_sub32 av_sat_sub32_arm +static av_always_inline int av_sat_sub32_arm(int a, int b) +{ + int r; + __asm__ ("qsub %0, %1, %2" : "=r"(r) : "r"(a), "r"(b)); + return r; +} + +#define av_sat_dsub32 av_sat_dsub32_arm +static av_always_inline int av_sat_dsub32_arm(int a, int b) +{ + int r; + __asm__ ("qdsub %0, %1, %2" : "=r"(r) : "r"(a), "r"(b)); + return r; +} + +#endif /* HAVE_ARMV6_INLINE */ + +#if HAVE_ASM_MOD_Q + +#define av_clipl_int32 av_clipl_int32_arm +static av_always_inline av_const int32_t av_clipl_int32_arm(int64_t a) +{ + int x, y; + __asm__ ("adds %1, %R2, %Q2, lsr #31 \n\t" + "itet ne \n\t" + "mvnne %1, #1<<31 \n\t" + "moveq %0, %Q2 \n\t" + "eorne %0, %1, %R2, asr #31 \n\t" + : "=r"(x), "=&r"(y) : "r"(a) : "cc"); + return x; +} + +#endif /* HAVE_ASM_MOD_Q */ + +#endif /* HAVE_INLINE_ASM */ + +#endif /* AVUTIL_ARM_INTMATH_H */ diff --git a/media/ffvpx/libavutil/arm/intreadwrite.h b/media/ffvpx/libavutil/arm/intreadwrite.h new file mode 100644 index 0000000000..60fc860cbb --- /dev/null +++ b/media/ffvpx/libavutil/arm/intreadwrite.h @@ -0,0 +1,91 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_INTREADWRITE_H +#define AVUTIL_ARM_INTREADWRITE_H + +#include <stdint.h> +#include "config.h" +#include "libavutil/attributes.h" + +#if HAVE_FAST_UNALIGNED && HAVE_INLINE_ASM && AV_GCC_VERSION_AT_MOST(4,6) + +#define AV_RN16 AV_RN16 +static av_always_inline unsigned AV_RN16(const void *p) +{ + const uint8_t *q = p; + unsigned v; +#if AV_GCC_VERSION_AT_MOST(4,5) + __asm__ ("ldrh %0, %1" : "=r"(v) : "m"(*(const uint16_t *)q)); +#elif defined __thumb__ + __asm__ ("ldrh %0, %1" : "=r"(v) : "m"(q[0]), "m"(q[1])); +#else + __asm__ ("ldrh %0, %1" : "=r"(v) : "Uq"(q[0]), "m"(q[1])); +#endif + return v; +} + +#define AV_WN16 AV_WN16 +static av_always_inline void AV_WN16(void *p, uint16_t v) +{ + __asm__ ("strh %1, %0" : "=m"(*(uint16_t *)p) : "r"(v)); +} + +#define AV_RN32 AV_RN32 +static av_always_inline uint32_t AV_RN32(const void *p) +{ + const struct __attribute__((packed)) { uint32_t v; } *q = p; + uint32_t v; + __asm__ ("ldr %0, %1" : "=r"(v) : "m"(*q)); + return v; +} + +#define AV_WN32 AV_WN32 +static av_always_inline void AV_WN32(void *p, uint32_t v) +{ + __asm__ ("str %1, %0" : "=m"(*(uint32_t *)p) : "r"(v)); +} + +#if HAVE_ASM_MOD_Q + +#define AV_RN64 AV_RN64 +static av_always_inline uint64_t AV_RN64(const void *p) +{ + const struct __attribute__((packed)) { uint32_t v; } *q = p; + uint64_t v; + __asm__ ("ldr %Q0, %1 \n\t" + "ldr %R0, %2 \n\t" + : "=&r"(v) + : "m"(q[0]), "m"(q[1])); + return v; +} + +#define AV_WN64 AV_WN64 +static av_always_inline void AV_WN64(void *p, uint64_t v) +{ + __asm__ ("str %Q2, %0 \n\t" + "str %R2, %1 \n\t" + : "=m"(*(uint32_t*)p), "=m"(*((uint32_t*)p+1)) + : "r"(v)); +} + +#endif /* HAVE_ASM_MOD_Q */ + +#endif /* HAVE_INLINE_ASM */ + +#endif /* AVUTIL_ARM_INTREADWRITE_H */ diff --git a/media/ffvpx/libavutil/arm/moz.build b/media/ffvpx/libavutil/arm/moz.build new file mode 100644 index 0000000000..e5f9fb23e3 --- /dev/null +++ b/media/ffvpx/libavutil/arm/moz.build @@ -0,0 +1,18 @@ +# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*- +# vim: set filetype=python: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +SOURCES += [ + 'cpu.c', + 'float_dsp_init_arm.c', + 'float_dsp_init_neon.c', + 'float_dsp_init_vfp.c', + 'float_dsp_neon.S', + 'float_dsp_vfp.S', +] + +FINAL_LIBRARY = 'mozavutil' + +include('/media/ffvpx/ffvpxcommon.mozbuild') diff --git a/media/ffvpx/libavutil/arm/timer.h b/media/ffvpx/libavutil/arm/timer.h new file mode 100644 index 0000000000..caf23e2a5a --- /dev/null +++ b/media/ffvpx/libavutil/arm/timer.h @@ -0,0 +1,46 @@ +/* + * Copyright (c) 2009 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ARM_TIMER_H +#define AVUTIL_ARM_TIMER_H + +#include <stdint.h> +#include "config.h" + +#if defined(__APPLE__) + +#include <mach/mach_time.h> + +#define AV_READ_TIME mach_absolute_time + +#elif HAVE_INLINE_ASM && defined(__ARM_ARCH_7A__) + +#define AV_READ_TIME read_time + +static inline uint64_t read_time(void) +{ + unsigned cc; + __asm__ volatile ("mrc p15, 0, %0, c9, c13, 0" : "=r"(cc)); + return cc; +} + +#endif /* HAVE_INLINE_ASM && __ARM_ARCH_7A__ */ + +#endif /* AVUTIL_ARM_TIMER_H */ diff --git a/media/ffvpx/libavutil/attributes.h b/media/ffvpx/libavutil/attributes.h new file mode 100644 index 0000000000..04c615c952 --- /dev/null +++ b/media/ffvpx/libavutil/attributes.h @@ -0,0 +1,173 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Macro definitions for various function/variable attributes + */ + +#ifndef AVUTIL_ATTRIBUTES_H +#define AVUTIL_ATTRIBUTES_H + +#ifdef __GNUC__ +# define AV_GCC_VERSION_AT_LEAST(x,y) (__GNUC__ > (x) || __GNUC__ == (x) && __GNUC_MINOR__ >= (y)) +# define AV_GCC_VERSION_AT_MOST(x,y) (__GNUC__ < (x) || __GNUC__ == (x) && __GNUC_MINOR__ <= (y)) +#else +# define AV_GCC_VERSION_AT_LEAST(x,y) 0 +# define AV_GCC_VERSION_AT_MOST(x,y) 0 +#endif + +#ifdef __has_builtin +# define AV_HAS_BUILTIN(x) __has_builtin(x) +#else +# define AV_HAS_BUILTIN(x) 0 +#endif + +#ifndef av_always_inline +#if AV_GCC_VERSION_AT_LEAST(3,1) +# define av_always_inline __attribute__((always_inline)) inline +#elif defined(_MSC_VER) +# define av_always_inline __forceinline +#else +# define av_always_inline inline +#endif +#endif + +#ifndef av_extern_inline +#if defined(__ICL) && __ICL >= 1210 || defined(__GNUC_STDC_INLINE__) +# define av_extern_inline extern inline +#else +# define av_extern_inline inline +#endif +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,4) +# define av_warn_unused_result __attribute__((warn_unused_result)) +#else +# define av_warn_unused_result +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,1) +# define av_noinline __attribute__((noinline)) +#elif defined(_MSC_VER) +# define av_noinline __declspec(noinline) +#else +# define av_noinline +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,1) || defined(__clang__) +# define av_pure __attribute__((pure)) +#else +# define av_pure +#endif + +#if AV_GCC_VERSION_AT_LEAST(2,6) || defined(__clang__) +# define av_const __attribute__((const)) +#else +# define av_const +#endif + +#if AV_GCC_VERSION_AT_LEAST(4,3) || defined(__clang__) +# define av_cold __attribute__((cold)) +#else +# define av_cold +#endif + +#if AV_GCC_VERSION_AT_LEAST(4,1) && !defined(__llvm__) +# define av_flatten __attribute__((flatten)) +#else +# define av_flatten +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,1) +# define attribute_deprecated __attribute__((deprecated)) +#elif defined(_MSC_VER) +# define attribute_deprecated __declspec(deprecated) +#else +# define attribute_deprecated +#endif + +/** + * Disable warnings about deprecated features + * This is useful for sections of code kept for backward compatibility and + * scheduled for removal. + */ +#ifndef AV_NOWARN_DEPRECATED +#if AV_GCC_VERSION_AT_LEAST(4,6) || defined(__clang__) +# define AV_NOWARN_DEPRECATED(code) \ + _Pragma("GCC diagnostic push") \ + _Pragma("GCC diagnostic ignored \"-Wdeprecated-declarations\"") \ + code \ + _Pragma("GCC diagnostic pop") +#elif defined(_MSC_VER) +# define AV_NOWARN_DEPRECATED(code) \ + __pragma(warning(push)) \ + __pragma(warning(disable : 4996)) \ + code; \ + __pragma(warning(pop)) +#else +# define AV_NOWARN_DEPRECATED(code) code +#endif +#endif + +#if defined(__GNUC__) || defined(__clang__) +# define av_unused __attribute__((unused)) +#else +# define av_unused +#endif + +/** + * Mark a variable as used and prevent the compiler from optimizing it + * away. This is useful for variables accessed only from inline + * assembler without the compiler being aware. + */ +#if AV_GCC_VERSION_AT_LEAST(3,1) || defined(__clang__) +# define av_used __attribute__((used)) +#else +# define av_used +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,3) || defined(__clang__) +# define av_alias __attribute__((may_alias)) +#else +# define av_alias +#endif + +#if (defined(__GNUC__) || defined(__clang__)) && !defined(__INTEL_COMPILER) +# define av_uninit(x) x=x +#else +# define av_uninit(x) x +#endif + +#if defined(__GNUC__) || defined(__clang__) +# define av_builtin_constant_p __builtin_constant_p +# define av_printf_format(fmtpos, attrpos) __attribute__((__format__(__printf__, fmtpos, attrpos))) +#else +# define av_builtin_constant_p(x) 0 +# define av_printf_format(fmtpos, attrpos) +#endif + +#if AV_GCC_VERSION_AT_LEAST(2,5) || defined(__clang__) +# define av_noreturn __attribute__((noreturn)) +#else +# define av_noreturn +#endif + +#endif /* AVUTIL_ATTRIBUTES_H */ diff --git a/media/ffvpx/libavutil/attributes_internal.h b/media/ffvpx/libavutil/attributes_internal.h new file mode 100644 index 0000000000..3df1ee6af3 --- /dev/null +++ b/media/ffvpx/libavutil/attributes_internal.h @@ -0,0 +1,34 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_ATTRIBUTES_INTERNAL_H +#define AVUTIL_ATTRIBUTES_INTERNAL_H + +#include "attributes.h" + +#if (AV_GCC_VERSION_AT_LEAST(4,0) || defined(__clang__)) && (defined(__ELF__) || defined(__MACH__)) +# define attribute_visibility_hidden __attribute__((visibility("hidden"))) +# define FF_VISIBILITY_PUSH_HIDDEN _Pragma("GCC visibility push(hidden)") +# define FF_VISIBILITY_POP_HIDDEN _Pragma("GCC visibility pop") +#else +# define attribute_visibility_hidden +# define FF_VISIBILITY_PUSH_HIDDEN +# define FF_VISIBILITY_POP_HIDDEN +#endif + +#endif /* AVUTIL_ATTRIBUTES_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/avassert.h b/media/ffvpx/libavutil/avassert.h new file mode 100644 index 0000000000..1895fb7551 --- /dev/null +++ b/media/ffvpx/libavutil/avassert.h @@ -0,0 +1,78 @@ +/* + * copyright (c) 2010 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * simple assert() macros that are a bit more flexible than ISO C assert(). + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#ifndef AVUTIL_AVASSERT_H +#define AVUTIL_AVASSERT_H + +#include <stdlib.h> +#ifdef HAVE_AV_CONFIG_H +# include "config.h" +#endif +#include "log.h" +#include "macros.h" + +/** + * assert() equivalent, that is always enabled. + */ +#define av_assert0(cond) do { \ + if (!(cond)) { \ + av_log(NULL, AV_LOG_PANIC, "Assertion %s failed at %s:%d\n", \ + AV_STRINGIFY(cond), __FILE__, __LINE__); \ + abort(); \ + } \ +} while (0) + + +/** + * assert() equivalent, that does not lie in speed critical code. + * These asserts() thus can be enabled without fearing speed loss. + */ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 0 +#define av_assert1(cond) av_assert0(cond) +#else +#define av_assert1(cond) ((void)0) +#endif + + +/** + * assert() equivalent, that does lie in speed critical code. + */ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1 +#define av_assert2(cond) av_assert0(cond) +#define av_assert2_fpu() av_assert0_fpu() +#else +#define av_assert2(cond) ((void)0) +#define av_assert2_fpu() ((void)0) +#endif + +/** + * Assert that floating point operations can be executed. + * + * This will av_assert0() that the cpu is not in MMX state on X86 + */ +void av_assert0_fpu(void); + +#endif /* AVUTIL_AVASSERT_H */ diff --git a/media/ffvpx/libavutil/avconfig.h b/media/ffvpx/libavutil/avconfig.h new file mode 100644 index 0000000000..c289fbb551 --- /dev/null +++ b/media/ffvpx/libavutil/avconfig.h @@ -0,0 +1,6 @@ +/* Generated by ffmpeg configure */ +#ifndef AVUTIL_AVCONFIG_H +#define AVUTIL_AVCONFIG_H +#define AV_HAVE_BIGENDIAN 0 +#define AV_HAVE_FAST_UNALIGNED 1 +#endif /* AVUTIL_AVCONFIG_H */ diff --git a/media/ffvpx/libavutil/avsscanf.c b/media/ffvpx/libavutil/avsscanf.c new file mode 100644 index 0000000000..7061e6d965 --- /dev/null +++ b/media/ffvpx/libavutil/avsscanf.c @@ -0,0 +1,970 @@ +/* + * Copyright (c) 2005-2014 Rich Felker, et al. + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <errno.h> +#include <limits.h> +#include <math.h> +#include <stdarg.h> +#include <stddef.h> +#include <stdint.h> +#include <stdio.h> +#include <string.h> +#include <float.h> + +#include "avstring.h" +#include "libm.h" + +typedef struct FFFILE { + size_t buf_size; + unsigned char *buf; + unsigned char *rpos, *rend; + unsigned char *shend; + ptrdiff_t shlim, shcnt; + void *cookie; + size_t (*read)(struct FFFILE *, unsigned char *, size_t); +} FFFILE; + +#define SIZE_hh -2 +#define SIZE_h -1 +#define SIZE_def 0 +#define SIZE_l 1 +#define SIZE_L 2 +#define SIZE_ll 3 + +#define shcnt(f) ((f)->shcnt + ((f)->rpos - (f)->buf)) + +static int fftoread(FFFILE *f) +{ + f->rpos = f->rend = f->buf + f->buf_size; + return 0; +} + +static size_t ffstring_read(FFFILE *f, unsigned char *buf, size_t len) +{ + char *src = f->cookie; + size_t k = len+256; + char *end = memchr(src, 0, k); + + if (end) k = end-src; + if (k < len) len = k; + memcpy(buf, src, len); + f->rpos = (void *)(src+len); + f->rend = (void *)(src+k); + f->cookie = src+k; + + return len; +} + +static int ffuflow(FFFILE *f) +{ + unsigned char c; + if (!fftoread(f) && f->read(f, &c, 1)==1) return c; + return EOF; +} + +static void ffshlim(FFFILE *f, ptrdiff_t lim) +{ + f->shlim = lim; + f->shcnt = f->buf - f->rpos; + /* If lim is nonzero, rend must be a valid pointer. */ + if (lim && f->rend - f->rpos > lim) + f->shend = f->rpos + lim; + else + f->shend = f->rend; +} + +static int ffshgetc(FFFILE *f) +{ + int c; + ptrdiff_t cnt = shcnt(f); + if (f->shlim && cnt >= f->shlim || (c=ffuflow(f)) < 0) { + f->shcnt = f->buf - f->rpos + cnt; + f->shend = 0; + return EOF; + } + cnt++; + if (f->shlim && f->rend - f->rpos > f->shlim - cnt) + f->shend = f->rpos + (f->shlim - cnt); + else + f->shend = f->rend; + f->shcnt = f->buf - f->rpos + cnt; + if (f->rpos[-1] != c) f->rpos[-1] = c; + return c; +} + +#define shlim(f, lim) ffshlim((f), (lim)) +#define shgetc(f) (((f)->rpos < (f)->shend) ? *(f)->rpos++ : ffshgetc(f)) +#define shunget(f) ((f)->shend ? (void)(f)->rpos-- : (void)0) + +static const unsigned char table[] = { -1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,-1,-1,-1,-1,-1,-1, + -1,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24, + 25,26,27,28,29,30,31,32,33,34,35,-1,-1,-1,-1,-1, + -1,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24, + 25,26,27,28,29,30,31,32,33,34,35,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, + -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, +}; + +static unsigned long long ffintscan(FFFILE *f, unsigned base, int pok, unsigned long long lim) +{ + const unsigned char *val = table+1; + int c, neg=0; + unsigned x; + unsigned long long y; + if (base > 36 || base == 1) { + errno = EINVAL; + return 0; + } + while (av_isspace((c=shgetc(f)))); + if (c=='+' || c=='-') { + neg = -(c=='-'); + c = shgetc(f); + } + if ((base == 0 || base == 16) && c=='0') { + c = shgetc(f); + if ((c|32)=='x') { + c = shgetc(f); + if (val[c]>=16) { + shunget(f); + if (pok) shunget(f); + else shlim(f, 0); + return 0; + } + base = 16; + } else if (base == 0) { + base = 8; + } + } else { + if (base == 0) base = 10; + if (val[c] >= base) { + shunget(f); + shlim(f, 0); + errno = EINVAL; + return 0; + } + } + if (base == 10) { + for (x=0; c-'0'<10U && x<=UINT_MAX/10-1; c=shgetc(f)) + x = x*10 + (c-'0'); + for (y=x; c-'0'<10U && y<=ULLONG_MAX/10 && 10*y<=ULLONG_MAX-(c-'0'); c=shgetc(f)) + y = y*10 + (c-'0'); + if (c-'0'>=10U) goto done; + } else if (!(base & base-1)) { + int bs = "\0\1\2\4\7\3\6\5"[(0x17*base)>>5&7]; + for (x=0; val[c]<base && x<=UINT_MAX/32; c=shgetc(f)) + x = x<<bs | val[c]; + for (y=x; val[c]<base && y<=ULLONG_MAX>>bs; c=shgetc(f)) + y = y<<bs | val[c]; + } else { + for (x=0; val[c]<base && x<=UINT_MAX/36-1; c=shgetc(f)) + x = x*base + val[c]; + for (y=x; val[c]<base && y<=ULLONG_MAX/base && base*y<=ULLONG_MAX-val[c]; c=shgetc(f)) + y = y*base + val[c]; + } + if (val[c]<base) { + for (; val[c]<base; c=shgetc(f)); + errno = ERANGE; + y = lim; + if (lim&1) neg = 0; + } +done: + shunget(f); + if (y>=lim) { + if (!(lim&1) && !neg) { + errno = ERANGE; + return lim-1; + } else if (y>lim) { + errno = ERANGE; + return lim; + } + } + return (y^neg)-neg; +} + +static long long scanexp(FFFILE *f, int pok) +{ + int c; + int x; + long long y; + int neg = 0; + + c = shgetc(f); + if (c=='+' || c=='-') { + neg = (c=='-'); + c = shgetc(f); + if (c-'0'>=10U && pok) shunget(f); + } + if (c-'0'>=10U) { + shunget(f); + return LLONG_MIN; + } + for (x=0; c-'0'<10U && x<INT_MAX/10; c = shgetc(f)) + x = 10*x + (c-'0'); + for (y=x; c-'0'<10U && y<LLONG_MAX/100; c = shgetc(f)) + y = 10*y + (c-'0'); + for (; c-'0'<10U; c = shgetc(f)); + shunget(f); + return neg ? -y : y; +} + +#define LD_B1B_DIG 2 +#define LD_B1B_MAX 9007199, 254740991 +#define KMAX 128 +#define MASK (KMAX-1) + +static double decfloat(FFFILE *f, int c, int bits, int emin, int sign, int pok) +{ + uint32_t x[KMAX]; + static const uint32_t th[] = { LD_B1B_MAX }; + int i, j, k, a, z; + long long lrp=0, dc=0; + long long e10=0; + int lnz = 0; + int gotdig = 0, gotrad = 0; + int rp; + int e2; + int emax = -emin-bits+3; + int denormal = 0; + double y; + double frac=0; + double bias=0; + static const int p10s[] = { 10, 100, 1000, 10000, + 100000, 1000000, 10000000, 100000000 }; + + j=0; + k=0; + + /* Don't let leading zeros consume buffer space */ + for (; c=='0'; c = shgetc(f)) gotdig=1; + if (c=='.') { + gotrad = 1; + for (c = shgetc(f); c=='0'; c = shgetc(f)) gotdig=1, lrp--; + } + + x[0] = 0; + for (; c-'0'<10U || c=='.'; c = shgetc(f)) { + if (c == '.') { + if (gotrad) break; + gotrad = 1; + lrp = dc; + } else if (k < KMAX-3) { + dc++; + if (c!='0') lnz = dc; + if (j) x[k] = x[k]*10 + c-'0'; + else x[k] = c-'0'; + if (++j==9) { + k++; + j=0; + } + gotdig=1; + } else { + dc++; + if (c!='0') { + lnz = (KMAX-4)*9; + x[KMAX-4] |= 1; + } + } + } + if (!gotrad) lrp=dc; + + if (gotdig && (c|32)=='e') { + e10 = scanexp(f, pok); + if (e10 == LLONG_MIN) { + if (pok) { + shunget(f); + } else { + shlim(f, 0); + return 0; + } + e10 = 0; + } + lrp += e10; + } else if (c>=0) { + shunget(f); + } + if (!gotdig) { + errno = EINVAL; + shlim(f, 0); + return 0; + } + + /* Handle zero specially to avoid nasty special cases later */ + if (!x[0]) return sign * 0.0; + + /* Optimize small integers (w/no exponent) and over/under-flow */ + if (lrp==dc && dc<10 && (bits>30 || x[0]>>bits==0)) + return sign * (double)x[0]; + if (lrp > -emin/2) { + errno = ERANGE; + return sign * DBL_MAX * DBL_MAX; + } + if (lrp < emin-2*DBL_MANT_DIG) { + errno = ERANGE; + return sign * DBL_MIN * DBL_MIN; + } + + /* Align incomplete final B1B digit */ + if (j) { + for (; j<9; j++) x[k]*=10; + k++; + j=0; + } + + a = 0; + z = k; + e2 = 0; + rp = lrp; + + /* Optimize small to mid-size integers (even in exp. notation) */ + if (lnz<9 && lnz<=rp && rp < 18) { + int bitlim; + if (rp == 9) return sign * (double)x[0]; + if (rp < 9) return sign * (double)x[0] / p10s[8-rp]; + bitlim = bits-3*(int)(rp-9); + if (bitlim>30 || x[0]>>bitlim==0) + return sign * (double)x[0] * p10s[rp-10]; + } + + /* Drop trailing zeros */ + for (; !x[z-1]; z--); + + /* Align radix point to B1B digit boundary */ + if (rp % 9) { + int rpm9 = rp>=0 ? rp%9 : rp%9+9; + int p10 = p10s[8-rpm9]; + uint32_t carry = 0; + for (k=a; k!=z; k++) { + uint32_t tmp = x[k] % p10; + x[k] = x[k]/p10 + carry; + carry = 1000000000/p10 * tmp; + if (k==a && !x[k]) { + a = (a+1 & MASK); + rp -= 9; + } + } + if (carry) x[z++] = carry; + rp += 9-rpm9; + } + + /* Upscale until desired number of bits are left of radix point */ + while (rp < 9*LD_B1B_DIG || (rp == 9*LD_B1B_DIG && x[a]<th[0])) { + uint32_t carry = 0; + e2 -= 29; + for (k=(z-1 & MASK); ; k=(k-1 & MASK)) { + uint64_t tmp = ((uint64_t)x[k] << 29) + carry; + if (tmp > 1000000000) { + carry = tmp / 1000000000; + x[k] = tmp % 1000000000; + } else { + carry = 0; + x[k] = tmp; + } + if (k==(z-1 & MASK) && k!=a && !x[k]) z = k; + if (k==a) break; + } + if (carry) { + rp += 9; + a = (a-1 & MASK); + if (a == z) { + z = (z-1 & MASK); + x[z-1 & MASK] |= x[z]; + } + x[a] = carry; + } + } + + /* Downscale until exactly number of bits are left of radix point */ + for (;;) { + uint32_t carry = 0; + int sh = 1; + for (i=0; i<LD_B1B_DIG; i++) { + k = (a+i & MASK); + if (k == z || x[k] < th[i]) { + i=LD_B1B_DIG; + break; + } + if (x[a+i & MASK] > th[i]) break; + } + if (i==LD_B1B_DIG && rp==9*LD_B1B_DIG) break; + /* FIXME: find a way to compute optimal sh */ + if (rp > 9+9*LD_B1B_DIG) sh = 9; + e2 += sh; + for (k=a; k!=z; k=(k+1 & MASK)) { + uint32_t tmp = x[k] & (1<<sh)-1; + x[k] = (x[k]>>sh) + carry; + carry = (1000000000>>sh) * tmp; + if (k==a && !x[k]) { + a = (a+1 & MASK); + i--; + rp -= 9; + } + } + if (carry) { + if ((z+1 & MASK) != a) { + x[z] = carry; + z = (z+1 & MASK); + } else x[z-1 & MASK] |= 1; + } + } + + /* Assemble desired bits into floating point variable */ + for (y=i=0; i<LD_B1B_DIG; i++) { + if ((a+i & MASK)==z) x[(z=(z+1 & MASK))-1] = 0; + y = 1000000000.0L * y + x[a+i & MASK]; + } + + y *= sign; + + /* Limit precision for denormal results */ + if (bits > DBL_MANT_DIG+e2-emin) { + bits = DBL_MANT_DIG+e2-emin; + if (bits<0) bits=0; + denormal = 1; + } + + /* Calculate bias term to force rounding, move out lower bits */ + if (bits < DBL_MANT_DIG) { + bias = copysign(scalbn(1, 2*DBL_MANT_DIG-bits-1), y); + frac = fmod(y, scalbn(1, DBL_MANT_DIG-bits)); + y -= frac; + y += bias; + } + + /* Process tail of decimal input so it can affect rounding */ + if ((a+i & MASK) != z) { + uint32_t t = x[a+i & MASK]; + if (t < 500000000 && (t || (a+i+1 & MASK) != z)) + frac += 0.25*sign; + else if (t > 500000000) + frac += 0.75*sign; + else if (t == 500000000) { + if ((a+i+1 & MASK) == z) + frac += 0.5*sign; + else + frac += 0.75*sign; + } + if (DBL_MANT_DIG-bits >= 2 && !fmod(frac, 1)) + frac++; + } + + y += frac; + y -= bias; + + if ((e2+DBL_MANT_DIG & INT_MAX) > emax-5) { + if (fabs(y) >= pow(2, DBL_MANT_DIG)) { + if (denormal && bits==DBL_MANT_DIG+e2-emin) + denormal = 0; + y *= 0.5; + e2++; + } + if (e2+DBL_MANT_DIG>emax || (denormal && frac)) + errno = ERANGE; + } + + return scalbn(y, e2); +} + +static double hexfloat(FFFILE *f, int bits, int emin, int sign, int pok) +{ + uint32_t x = 0; + double y = 0; + double scale = 1; + double bias = 0; + int gottail = 0, gotrad = 0, gotdig = 0; + long long rp = 0; + long long dc = 0; + long long e2 = 0; + int d; + int c; + + c = shgetc(f); + + /* Skip leading zeros */ + for (; c=='0'; c = shgetc(f)) + gotdig = 1; + + if (c=='.') { + gotrad = 1; + c = shgetc(f); + /* Count zeros after the radix point before significand */ + for (rp=0; c=='0'; c = shgetc(f), rp--) gotdig = 1; + } + + for (; c-'0'<10U || (c|32)-'a'<6U || c=='.'; c = shgetc(f)) { + if (c=='.') { + if (gotrad) break; + rp = dc; + gotrad = 1; + } else { + gotdig = 1; + if (c > '9') d = (c|32)+10-'a'; + else d = c-'0'; + if (dc<8) { + x = x*16 + d; + } else if (dc < DBL_MANT_DIG/4+1) { + y += d*(scale/=16); + } else if (d && !gottail) { + y += 0.5*scale; + gottail = 1; + } + dc++; + } + } + if (!gotdig) { + shunget(f); + if (pok) { + shunget(f); + if (gotrad) shunget(f); + } else { + shlim(f, 0); + } + return sign * 0.0; + } + if (!gotrad) rp = dc; + while (dc<8) x *= 16, dc++; + if ((c|32)=='p') { + e2 = scanexp(f, pok); + if (e2 == LLONG_MIN) { + if (pok) { + shunget(f); + } else { + shlim(f, 0); + return 0; + } + e2 = 0; + } + } else { + shunget(f); + } + e2 += 4*rp - 32; + + if (!x) return sign * 0.0; + if (e2 > -emin) { + errno = ERANGE; + return sign * DBL_MAX * DBL_MAX; + } + if (e2 < emin-2*DBL_MANT_DIG) { + errno = ERANGE; + return sign * DBL_MIN * DBL_MIN; + } + + while (x < 0x80000000) { + if (y>=0.5) { + x += x + 1; + y += y - 1; + } else { + x += x; + y += y; + } + e2--; + } + + if (bits > 32+e2-emin) { + bits = 32+e2-emin; + if (bits<0) bits=0; + } + + if (bits < DBL_MANT_DIG) + bias = copysign(scalbn(1, 32+DBL_MANT_DIG-bits-1), sign); + + if (bits<32 && y && !(x&1)) x++, y=0; + + y = bias + sign*(double)x + sign*y; + y -= bias; + + if (!y) errno = ERANGE; + + return scalbn(y, e2); +} + +static double fffloatscan(FFFILE *f, int prec, int pok) +{ + int sign = 1; + size_t i; + int bits; + int emin; + int c; + + switch (prec) { + case 0: + bits = FLT_MANT_DIG; + emin = FLT_MIN_EXP-bits; + break; + case 1: + bits = DBL_MANT_DIG; + emin = DBL_MIN_EXP-bits; + break; + case 2: + bits = DBL_MANT_DIG; + emin = DBL_MIN_EXP-bits; + break; + default: + return 0; + } + + while (av_isspace((c = shgetc(f)))); + + if (c=='+' || c=='-') { + sign -= 2*(c=='-'); + c = shgetc(f); + } + + for (i=0; i<8 && (c|32)=="infinity"[i]; i++) + if (i<7) c = shgetc(f); + if (i==3 || i==8 || (i>3 && pok)) { + if (i!=8) { + shunget(f); + if (pok) for (; i>3; i--) shunget(f); + } + return sign * INFINITY; + } + if (!i) for (i=0; i<3 && (c|32)=="nan"[i]; i++) + if (i<2) c = shgetc(f); + if (i==3) { + if (shgetc(f) != '(') { + shunget(f); + return NAN; + } + for (i=1; ; i++) { + c = shgetc(f); + if (c-'0'<10U || c-'A'<26U || c-'a'<26U || c=='_') + continue; + if (c==')') return NAN; + shunget(f); + if (!pok) { + errno = EINVAL; + shlim(f, 0); + return 0; + } + while (i--) shunget(f); + return NAN; + } + return NAN; + } + + if (i) { + shunget(f); + errno = EINVAL; + shlim(f, 0); + return 0; + } + + if (c=='0') { + c = shgetc(f); + if ((c|32) == 'x') + return hexfloat(f, bits, emin, sign, pok); + shunget(f); + c = '0'; + } + + return decfloat(f, c, bits, emin, sign, pok); +} + +static void *arg_n(va_list ap, unsigned int n) +{ + void *p; + unsigned int i; + va_list ap2; + va_copy(ap2, ap); + for (i=n; i>1; i--) va_arg(ap2, void *); + p = va_arg(ap2, void *); + va_end(ap2); + return p; +} + +static void store_int(void *dest, int size, unsigned long long i) +{ + if (!dest) return; + switch (size) { + case SIZE_hh: + *(char *)dest = i; + break; + case SIZE_h: + *(short *)dest = i; + break; + case SIZE_def: + *(int *)dest = i; + break; + case SIZE_l: + *(long *)dest = i; + break; + case SIZE_ll: + *(long long *)dest = i; + break; + } +} + +static int ff_vfscanf(FFFILE *f, const char *fmt, va_list ap) +{ + int width; + int size; + int base; + const unsigned char *p; + int c, t; + char *s; + void *dest=NULL; + int invert; + int matches=0; + unsigned long long x; + double y; + ptrdiff_t pos = 0; + unsigned char scanset[257]; + size_t i; + + for (p=(const unsigned char *)fmt; *p; p++) { + + if (av_isspace(*p)) { + while (av_isspace(p[1])) p++; + shlim(f, 0); + while (av_isspace(shgetc(f))); + shunget(f); + pos += shcnt(f); + continue; + } + if (*p != '%' || p[1] == '%') { + shlim(f, 0); + if (*p == '%') { + p++; + while (av_isspace((c=shgetc(f)))); + } else { + c = shgetc(f); + } + if (c!=*p) { + shunget(f); + if (c<0) goto input_fail; + goto match_fail; + } + pos += shcnt(f); + continue; + } + + p++; + if (*p=='*') { + dest = 0; p++; + } else if (av_isdigit(*p) && p[1]=='$') { + dest = arg_n(ap, *p-'0'); p+=2; + } else { + dest = va_arg(ap, void *); + } + + for (width=0; av_isdigit(*p); p++) { + width = 10*width + *p - '0'; + } + + if (*p=='m') { + s = 0; + p++; + } + + size = SIZE_def; + switch (*p++) { + case 'h': + if (*p == 'h') p++, size = SIZE_hh; + else size = SIZE_h; + break; + case 'l': + if (*p == 'l') p++, size = SIZE_ll; + else size = SIZE_l; + break; + case 'j': + size = SIZE_ll; + break; + case 'z': + case 't': + size = SIZE_l; + break; + case 'L': + size = SIZE_L; + break; + case 'd': case 'i': case 'o': case 'u': case 'x': + case 'a': case 'e': case 'f': case 'g': + case 'A': case 'E': case 'F': case 'G': case 'X': + case 's': case 'c': case '[': + case 'S': case 'C': + case 'p': case 'n': + p--; + break; + default: + goto fmt_fail; + } + + t = *p; + + /* C or S */ + if ((t&0x2f) == 3) { + t |= 32; + size = SIZE_l; + } + + switch (t) { + case 'c': + if (width < 1) width = 1; + case '[': + break; + case 'n': + store_int(dest, size, pos); + /* do not increment match count, etc! */ + continue; + default: + shlim(f, 0); + while (av_isspace(shgetc(f))); + shunget(f); + pos += shcnt(f); + } + + shlim(f, width); + if (shgetc(f) < 0) goto input_fail; + shunget(f); + + switch (t) { + case 's': + case 'c': + case '[': + if (t == 'c' || t == 's') { + memset(scanset, -1, sizeof scanset); + scanset[0] = 0; + if (t == 's') { + scanset[1 + '\t'] = 0; + scanset[1 + '\n'] = 0; + scanset[1 + '\v'] = 0; + scanset[1 + '\f'] = 0; + scanset[1 + '\r'] = 0; + scanset[1 + ' ' ] = 0; + } + } else { + if (*++p == '^') p++, invert = 1; + else invert = 0; + memset(scanset, invert, sizeof scanset); + scanset[0] = 0; + if (*p == '-') p++, scanset[1+'-'] = 1-invert; + else if (*p == ']') p++, scanset[1+']'] = 1-invert; + for (; *p != ']'; p++) { + if (!*p) goto fmt_fail; + if (*p=='-' && p[1] && p[1] != ']') + for (c=p++[-1]; c<*p; c++) + scanset[1+c] = 1-invert; + scanset[1+*p] = 1-invert; + } + } + s = 0; + i = 0; + if ((s = dest)) { + while (scanset[(c=shgetc(f))+1]) + s[i++] = c; + } else { + while (scanset[(c=shgetc(f))+1]); + } + shunget(f); + if (!shcnt(f)) goto match_fail; + if (t == 'c' && shcnt(f) != width) goto match_fail; + if (t != 'c') { + if (s) s[i] = 0; + } + break; + case 'p': + case 'X': + case 'x': + base = 16; + goto int_common; + case 'o': + base = 8; + goto int_common; + case 'd': + case 'u': + base = 10; + goto int_common; + case 'i': + base = 0; +int_common: + x = ffintscan(f, base, 0, ULLONG_MAX); + if (!shcnt(f)) + goto match_fail; + if (t=='p' && dest) + *(void **)dest = (void *)(uintptr_t)x; + else + store_int(dest, size, x); + break; + case 'a': case 'A': + case 'e': case 'E': + case 'f': case 'F': + case 'g': case 'G': + y = fffloatscan(f, size, 0); + if (!shcnt(f)) + goto match_fail; + if (dest) { + switch (size) { + case SIZE_def: + *(float *)dest = y; + break; + case SIZE_l: + *(double *)dest = y; + break; + case SIZE_L: + *(double *)dest = y; + break; + } + } + break; + } + + pos += shcnt(f); + if (dest) matches++; + } + if (0) { +fmt_fail: +input_fail: + if (!matches) matches--; + } +match_fail: + return matches; +} + +static int ff_vsscanf(const char *s, const char *fmt, va_list ap) +{ + FFFILE f = { + .buf = (void *)s, .cookie = (void *)s, + .read = ffstring_read, + }; + + return ff_vfscanf(&f, fmt, ap); +} + +int av_sscanf(const char *string, const char *format, ...) +{ + int ret; + va_list ap; + va_start(ap, format); + ret = ff_vsscanf(string, format, ap); + va_end(ap); + return ret; +} diff --git a/media/ffvpx/libavutil/avstring.c b/media/ffvpx/libavutil/avstring.c new file mode 100644 index 0000000000..8751ce5576 --- /dev/null +++ b/media/ffvpx/libavutil/avstring.c @@ -0,0 +1,462 @@ +/* + * Copyright (c) 2000, 2001, 2002 Fabrice Bellard + * Copyright (c) 2007 Mans Rullgard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <limits.h> +#include <stdarg.h> +#include <stdint.h> +#include <stdio.h> +#include <string.h> + +#include "config.h" +#include "mem.h" +#include "avassert.h" +#include "avstring.h" +#include "bprint.h" +#include "error.h" +#include "macros.h" + +int av_strstart(const char *str, const char *pfx, const char **ptr) +{ + while (*pfx && *pfx == *str) { + pfx++; + str++; + } + if (!*pfx && ptr) + *ptr = str; + return !*pfx; +} + +int av_stristart(const char *str, const char *pfx, const char **ptr) +{ + while (*pfx && av_toupper((unsigned)*pfx) == av_toupper((unsigned)*str)) { + pfx++; + str++; + } + if (!*pfx && ptr) + *ptr = str; + return !*pfx; +} + +char *av_stristr(const char *s1, const char *s2) +{ + if (!*s2) + return (char*)(intptr_t)s1; + + do + if (av_stristart(s1, s2, NULL)) + return (char*)(intptr_t)s1; + while (*s1++); + + return NULL; +} + +char *av_strnstr(const char *haystack, const char *needle, size_t hay_length) +{ + size_t needle_len = strlen(needle); + if (!needle_len) + return (char*)haystack; + while (hay_length >= needle_len) { + hay_length--; + if (!memcmp(haystack, needle, needle_len)) + return (char*)haystack; + haystack++; + } + return NULL; +} + +size_t av_strlcpy(char *dst, const char *src, size_t size) +{ + size_t len = 0; + while (++len < size && *src) + *dst++ = *src++; + if (len <= size) + *dst = 0; + return len + strlen(src) - 1; +} + +size_t av_strlcat(char *dst, const char *src, size_t size) +{ + size_t len = strlen(dst); + if (size <= len + 1) + return len + strlen(src); + return len + av_strlcpy(dst + len, src, size - len); +} + +size_t av_strlcatf(char *dst, size_t size, const char *fmt, ...) +{ + size_t len = strlen(dst); + va_list vl; + + va_start(vl, fmt); + len += vsnprintf(dst + len, size > len ? size - len : 0, fmt, vl); + va_end(vl); + + return len; +} + +char *av_asprintf(const char *fmt, ...) +{ + char *p = NULL; + va_list va; + int len; + + va_start(va, fmt); + len = vsnprintf(NULL, 0, fmt, va); + va_end(va); + if (len < 0) + goto end; + + p = av_malloc(len + 1); + if (!p) + goto end; + + va_start(va, fmt); + len = vsnprintf(p, len + 1, fmt, va); + va_end(va); + if (len < 0) + av_freep(&p); + +end: + return p; +} + +#define WHITESPACES " \n\t\r" + +char *av_get_token(const char **buf, const char *term) +{ + char *out = av_malloc(strlen(*buf) + 1); + char *ret = out, *end = out; + const char *p = *buf; + if (!out) + return NULL; + p += strspn(p, WHITESPACES); + + while (*p && !strspn(p, term)) { + char c = *p++; + if (c == '\\' && *p) { + *out++ = *p++; + end = out; + } else if (c == '\'') { + while (*p && *p != '\'') + *out++ = *p++; + if (*p) { + p++; + end = out; + } + } else { + *out++ = c; + } + } + + do + *out-- = 0; + while (out >= end && strspn(out, WHITESPACES)); + + *buf = p; + + return ret; +} + +char *av_strtok(char *s, const char *delim, char **saveptr) +{ + char *tok; + + if (!s && !(s = *saveptr)) + return NULL; + + /* skip leading delimiters */ + s += strspn(s, delim); + + /* s now points to the first non delimiter char, or to the end of the string */ + if (!*s) { + *saveptr = NULL; + return NULL; + } + tok = s++; + + /* skip non delimiters */ + s += strcspn(s, delim); + if (*s) { + *s = 0; + *saveptr = s+1; + } else { + *saveptr = NULL; + } + + return tok; +} + +int av_strcasecmp(const char *a, const char *b) +{ + uint8_t c1, c2; + do { + c1 = av_tolower(*a++); + c2 = av_tolower(*b++); + } while (c1 && c1 == c2); + return c1 - c2; +} + +int av_strncasecmp(const char *a, const char *b, size_t n) +{ + uint8_t c1, c2; + if (n <= 0) + return 0; + do { + c1 = av_tolower(*a++); + c2 = av_tolower(*b++); + } while (--n && c1 && c1 == c2); + return c1 - c2; +} + +char *av_strireplace(const char *str, const char *from, const char *to) +{ + char *ret = NULL; + const char *pstr2, *pstr = str; + size_t tolen = strlen(to), fromlen = strlen(from); + AVBPrint pbuf; + + av_bprint_init(&pbuf, 1, AV_BPRINT_SIZE_UNLIMITED); + while ((pstr2 = av_stristr(pstr, from))) { + av_bprint_append_data(&pbuf, pstr, pstr2 - pstr); + pstr = pstr2 + fromlen; + av_bprint_append_data(&pbuf, to, tolen); + } + av_bprint_append_data(&pbuf, pstr, strlen(pstr)); + if (!av_bprint_is_complete(&pbuf)) { + av_bprint_finalize(&pbuf, NULL); + } else { + av_bprint_finalize(&pbuf, &ret); + } + + return ret; +} + +const char *av_basename(const char *path) +{ + char *p; +#if HAVE_DOS_PATHS + char *q, *d; +#endif + + if (!path || *path == '\0') + return "."; + + p = strrchr(path, '/'); +#if HAVE_DOS_PATHS + q = strrchr(path, '\\'); + d = strchr(path, ':'); + p = FFMAX3(p, q, d); +#endif + + if (!p) + return path; + + return p + 1; +} + +const char *av_dirname(char *path) +{ + char *p = path ? strrchr(path, '/') : NULL; + +#if HAVE_DOS_PATHS + char *q = path ? strrchr(path, '\\') : NULL; + char *d = path ? strchr(path, ':') : NULL; + + d = d ? d + 1 : d; + + p = FFMAX3(p, q, d); +#endif + + if (!p) + return "."; + + *p = '\0'; + + return path; +} + +char *av_append_path_component(const char *path, const char *component) +{ + size_t p_len, c_len; + char *fullpath; + + if (!path) + return av_strdup(component); + if (!component) + return av_strdup(path); + + p_len = strlen(path); + c_len = strlen(component); + if (p_len > SIZE_MAX - c_len || p_len + c_len > SIZE_MAX - 2) + return NULL; + fullpath = av_malloc(p_len + c_len + 2); + if (fullpath) { + if (p_len) { + av_strlcpy(fullpath, path, p_len + 1); + if (c_len) { + if (fullpath[p_len - 1] != '/' && component[0] != '/') + fullpath[p_len++] = '/'; + else if (fullpath[p_len - 1] == '/' && component[0] == '/') + p_len--; + } + } + av_strlcpy(&fullpath[p_len], component, c_len + 1); + fullpath[p_len + c_len] = 0; + } + return fullpath; +} + +int av_escape(char **dst, const char *src, const char *special_chars, + enum AVEscapeMode mode, int flags) +{ + AVBPrint dstbuf; + int ret; + + av_bprint_init(&dstbuf, 1, INT_MAX); /* (int)dstbuf.len must be >= 0 */ + av_bprint_escape(&dstbuf, src, special_chars, mode, flags); + + if (!av_bprint_is_complete(&dstbuf)) { + av_bprint_finalize(&dstbuf, NULL); + return AVERROR(ENOMEM); + } + if ((ret = av_bprint_finalize(&dstbuf, dst)) < 0) + return ret; + return dstbuf.len; +} + +int av_match_name(const char *name, const char *names) +{ + const char *p; + int len, namelen; + + if (!name || !names) + return 0; + + namelen = strlen(name); + while (*names) { + int negate = '-' == *names; + p = strchr(names, ','); + if (!p) + p = names + strlen(names); + names += negate; + len = FFMAX(p - names, namelen); + if (!av_strncasecmp(name, names, len) || !strncmp("ALL", names, FFMAX(3, p - names))) + return !negate; + names = p + (*p == ','); + } + return 0; +} + +int av_utf8_decode(int32_t *codep, const uint8_t **bufp, const uint8_t *buf_end, + unsigned int flags) +{ + const uint8_t *p = *bufp; + uint32_t top; + uint64_t code; + int ret = 0, tail_len; + uint32_t overlong_encoding_mins[6] = { + 0x00000000, 0x00000080, 0x00000800, 0x00010000, 0x00200000, 0x04000000, + }; + + if (p >= buf_end) + return 0; + + code = *p++; + + /* first sequence byte starts with 10, or is 1111-1110 or 1111-1111, + which is not admitted */ + if ((code & 0xc0) == 0x80 || code >= 0xFE) { + ret = AVERROR(EILSEQ); + goto end; + } + top = (code & 128) >> 1; + + tail_len = 0; + while (code & top) { + int tmp; + tail_len++; + if (p >= buf_end) { + (*bufp) ++; + return AVERROR(EILSEQ); /* incomplete sequence */ + } + + /* we assume the byte to be in the form 10xx-xxxx */ + tmp = *p++ - 128; /* strip leading 1 */ + if (tmp>>6) { + (*bufp) ++; + return AVERROR(EILSEQ); + } + code = (code<<6) + tmp; + top <<= 5; + } + code &= (top << 1) - 1; + + /* check for overlong encodings */ + av_assert0(tail_len <= 5); + if (code < overlong_encoding_mins[tail_len]) { + ret = AVERROR(EILSEQ); + goto end; + } + + if (code >= 1U<<31) { + ret = AVERROR(EILSEQ); /* out-of-range value */ + goto end; + } + + *codep = code; + + if (code > 0x10FFFF && + !(flags & AV_UTF8_FLAG_ACCEPT_INVALID_BIG_CODES)) + ret = AVERROR(EILSEQ); + if (code < 0x20 && code != 0x9 && code != 0xA && code != 0xD && + flags & AV_UTF8_FLAG_EXCLUDE_XML_INVALID_CONTROL_CODES) + ret = AVERROR(EILSEQ); + if (code >= 0xD800 && code <= 0xDFFF && + !(flags & AV_UTF8_FLAG_ACCEPT_SURROGATES)) + ret = AVERROR(EILSEQ); + if ((code == 0xFFFE || code == 0xFFFF) && + !(flags & AV_UTF8_FLAG_ACCEPT_NON_CHARACTERS)) + ret = AVERROR(EILSEQ); + +end: + *bufp = p; + return ret; +} + +int av_match_list(const char *name, const char *list, char separator) +{ + const char *p, *q; + + for (p = name; p && *p; ) { + for (q = list; q && *q; ) { + int k; + for (k = 0; p[k] == q[k] || (p[k]*q[k] == 0 && p[k]+q[k] == separator); k++) + if (k && (!p[k] || p[k] == separator)) + return 1; + q = strchr(q, separator); + q += !!q; + } + p = strchr(p, separator); + p += !!p; + } + + return 0; +} diff --git a/media/ffvpx/libavutil/avstring.h b/media/ffvpx/libavutil/avstring.h new file mode 100644 index 0000000000..fc095349d2 --- /dev/null +++ b/media/ffvpx/libavutil/avstring.h @@ -0,0 +1,428 @@ +/* + * Copyright (c) 2007 Mans Rullgard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_AVSTRING_H +#define AVUTIL_AVSTRING_H + +#include <stddef.h> +#include <stdint.h> +#include "attributes.h" + +/** + * @addtogroup lavu_string + * @{ + */ + +/** + * Return non-zero if pfx is a prefix of str. If it is, *ptr is set to + * the address of the first character in str after the prefix. + * + * @param str input string + * @param pfx prefix to test + * @param ptr updated if the prefix is matched inside str + * @return non-zero if the prefix matches, zero otherwise + */ +int av_strstart(const char *str, const char *pfx, const char **ptr); + +/** + * Return non-zero if pfx is a prefix of str independent of case. If + * it is, *ptr is set to the address of the first character in str + * after the prefix. + * + * @param str input string + * @param pfx prefix to test + * @param ptr updated if the prefix is matched inside str + * @return non-zero if the prefix matches, zero otherwise + */ +int av_stristart(const char *str, const char *pfx, const char **ptr); + +/** + * Locate the first case-independent occurrence in the string haystack + * of the string needle. A zero-length string needle is considered to + * match at the start of haystack. + * + * This function is a case-insensitive version of the standard strstr(). + * + * @param haystack string to search in + * @param needle string to search for + * @return pointer to the located match within haystack + * or a null pointer if no match + */ +char *av_stristr(const char *haystack, const char *needle); + +/** + * Locate the first occurrence of the string needle in the string haystack + * where not more than hay_length characters are searched. A zero-length + * string needle is considered to match at the start of haystack. + * + * This function is a length-limited version of the standard strstr(). + * + * @param haystack string to search in + * @param needle string to search for + * @param hay_length length of string to search in + * @return pointer to the located match within haystack + * or a null pointer if no match + */ +char *av_strnstr(const char *haystack, const char *needle, size_t hay_length); + +/** + * Copy the string src to dst, but no more than size - 1 bytes, and + * null-terminate dst. + * + * This function is the same as BSD strlcpy(). + * + * @param dst destination buffer + * @param src source string + * @param size size of destination buffer + * @return the length of src + * + * @warning since the return value is the length of src, src absolutely + * _must_ be a properly 0-terminated string, otherwise this will read beyond + * the end of the buffer and possibly crash. + */ +size_t av_strlcpy(char *dst, const char *src, size_t size); + +/** + * Append the string src to the string dst, but to a total length of + * no more than size - 1 bytes, and null-terminate dst. + * + * This function is similar to BSD strlcat(), but differs when + * size <= strlen(dst). + * + * @param dst destination buffer + * @param src source string + * @param size size of destination buffer + * @return the total length of src and dst + * + * @warning since the return value use the length of src and dst, these + * absolutely _must_ be a properly 0-terminated strings, otherwise this + * will read beyond the end of the buffer and possibly crash. + */ +size_t av_strlcat(char *dst, const char *src, size_t size); + +/** + * Append output to a string, according to a format. Never write out of + * the destination buffer, and always put a terminating 0 within + * the buffer. + * @param dst destination buffer (string to which the output is + * appended) + * @param size total size of the destination buffer + * @param fmt printf-compatible format string, specifying how the + * following parameters are used + * @return the length of the string that would have been generated + * if enough space had been available + */ +size_t av_strlcatf(char *dst, size_t size, const char *fmt, ...) av_printf_format(3, 4); + +/** + * Get the count of continuous non zero chars starting from the beginning. + * + * @param s the string whose length to count + * @param len maximum number of characters to check in the string, that + * is the maximum value which is returned by the function + */ +static inline size_t av_strnlen(const char *s, size_t len) +{ + size_t i; + for (i = 0; i < len && s[i]; i++) + ; + return i; +} + +/** + * Print arguments following specified format into a large enough auto + * allocated buffer. It is similar to GNU asprintf(). + * @param fmt printf-compatible format string, specifying how the + * following parameters are used. + * @return the allocated string + * @note You have to free the string yourself with av_free(). + */ +char *av_asprintf(const char *fmt, ...) av_printf_format(1, 2); + +/** + * Unescape the given string until a non escaped terminating char, + * and return the token corresponding to the unescaped string. + * + * The normal \ and ' escaping is supported. Leading and trailing + * whitespaces are removed, unless they are escaped with '\' or are + * enclosed between ''. + * + * @param buf the buffer to parse, buf will be updated to point to the + * terminating char + * @param term a 0-terminated list of terminating chars + * @return the malloced unescaped string, which must be av_freed by + * the user, NULL in case of allocation failure + */ +char *av_get_token(const char **buf, const char *term); + +/** + * Split the string into several tokens which can be accessed by + * successive calls to av_strtok(). + * + * A token is defined as a sequence of characters not belonging to the + * set specified in delim. + * + * On the first call to av_strtok(), s should point to the string to + * parse, and the value of saveptr is ignored. In subsequent calls, s + * should be NULL, and saveptr should be unchanged since the previous + * call. + * + * This function is similar to strtok_r() defined in POSIX.1. + * + * @param s the string to parse, may be NULL + * @param delim 0-terminated list of token delimiters, must be non-NULL + * @param saveptr user-provided pointer which points to stored + * information necessary for av_strtok() to continue scanning the same + * string. saveptr is updated to point to the next character after the + * first delimiter found, or to NULL if the string was terminated + * @return the found token, or NULL when no token is found + */ +char *av_strtok(char *s, const char *delim, char **saveptr); + +/** + * Locale-independent conversion of ASCII isdigit. + */ +static inline av_const int av_isdigit(int c) +{ + return c >= '0' && c <= '9'; +} + +/** + * Locale-independent conversion of ASCII isgraph. + */ +static inline av_const int av_isgraph(int c) +{ + return c > 32 && c < 127; +} + +/** + * Locale-independent conversion of ASCII isspace. + */ +static inline av_const int av_isspace(int c) +{ + return c == ' ' || c == '\f' || c == '\n' || c == '\r' || c == '\t' || + c == '\v'; +} + +/** + * Locale-independent conversion of ASCII characters to uppercase. + */ +static inline av_const int av_toupper(int c) +{ + if (c >= 'a' && c <= 'z') + c ^= 0x20; + return c; +} + +/** + * Locale-independent conversion of ASCII characters to lowercase. + */ +static inline av_const int av_tolower(int c) +{ + if (c >= 'A' && c <= 'Z') + c ^= 0x20; + return c; +} + +/** + * Locale-independent conversion of ASCII isxdigit. + */ +static inline av_const int av_isxdigit(int c) +{ + c = av_tolower(c); + return av_isdigit(c) || (c >= 'a' && c <= 'f'); +} + +/** + * Locale-independent case-insensitive compare. + * @note This means only ASCII-range characters are case-insensitive + */ +int av_strcasecmp(const char *a, const char *b); + +/** + * Locale-independent case-insensitive compare. + * @note This means only ASCII-range characters are case-insensitive + */ +int av_strncasecmp(const char *a, const char *b, size_t n); + +/** + * Locale-independent strings replace. + * @note This means only ASCII-range characters are replaced. + */ +char *av_strireplace(const char *str, const char *from, const char *to); + +/** + * Thread safe basename. + * @param path the string to parse, on DOS both \ and / are considered separators. + * @return pointer to the basename substring. + * If path does not contain a slash, the function returns a copy of path. + * If path is a NULL pointer or points to an empty string, a pointer + * to a string "." is returned. + */ +const char *av_basename(const char *path); + +/** + * Thread safe dirname. + * @param path the string to parse, on DOS both \ and / are considered separators. + * @return A pointer to a string that's the parent directory of path. + * If path is a NULL pointer or points to an empty string, a pointer + * to a string "." is returned. + * @note the function may modify the contents of the path, so copies should be passed. + */ +const char *av_dirname(char *path); + +/** + * Match instances of a name in a comma-separated list of names. + * List entries are checked from the start to the end of the names list, + * the first match ends further processing. If an entry prefixed with '-' + * matches, then 0 is returned. The "ALL" list entry is considered to + * match all names. + * + * @param name Name to look for. + * @param names List of names. + * @return 1 on match, 0 otherwise. + */ +int av_match_name(const char *name, const char *names); + +/** + * Append path component to the existing path. + * Path separator '/' is placed between when needed. + * Resulting string have to be freed with av_free(). + * @param path base path + * @param component component to be appended + * @return new path or NULL on error. + */ +char *av_append_path_component(const char *path, const char *component); + +enum AVEscapeMode { + AV_ESCAPE_MODE_AUTO, ///< Use auto-selected escaping mode. + AV_ESCAPE_MODE_BACKSLASH, ///< Use backslash escaping. + AV_ESCAPE_MODE_QUOTE, ///< Use single-quote escaping. + AV_ESCAPE_MODE_XML, ///< Use XML non-markup character data escaping. +}; + +/** + * Consider spaces special and escape them even in the middle of the + * string. + * + * This is equivalent to adding the whitespace characters to the special + * characters lists, except it is guaranteed to use the exact same list + * of whitespace characters as the rest of libavutil. + */ +#define AV_ESCAPE_FLAG_WHITESPACE (1 << 0) + +/** + * Escape only specified special characters. + * Without this flag, escape also any characters that may be considered + * special by av_get_token(), such as the single quote. + */ +#define AV_ESCAPE_FLAG_STRICT (1 << 1) + +/** + * Within AV_ESCAPE_MODE_XML, additionally escape single quotes for single + * quoted attributes. + */ +#define AV_ESCAPE_FLAG_XML_SINGLE_QUOTES (1 << 2) + +/** + * Within AV_ESCAPE_MODE_XML, additionally escape double quotes for double + * quoted attributes. + */ +#define AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES (1 << 3) + + +/** + * Escape string in src, and put the escaped string in an allocated + * string in *dst, which must be freed with av_free(). + * + * @param dst pointer where an allocated string is put + * @param src string to escape, must be non-NULL + * @param special_chars string containing the special characters which + * need to be escaped, can be NULL + * @param mode escape mode to employ, see AV_ESCAPE_MODE_* macros. + * Any unknown value for mode will be considered equivalent to + * AV_ESCAPE_MODE_BACKSLASH, but this behaviour can change without + * notice. + * @param flags flags which control how to escape, see AV_ESCAPE_FLAG_ macros + * @return the length of the allocated string, or a negative error code in case of error + * @see av_bprint_escape() + */ +av_warn_unused_result +int av_escape(char **dst, const char *src, const char *special_chars, + enum AVEscapeMode mode, int flags); + +#define AV_UTF8_FLAG_ACCEPT_INVALID_BIG_CODES 1 ///< accept codepoints over 0x10FFFF +#define AV_UTF8_FLAG_ACCEPT_NON_CHARACTERS 2 ///< accept non-characters - 0xFFFE and 0xFFFF +#define AV_UTF8_FLAG_ACCEPT_SURROGATES 4 ///< accept UTF-16 surrogates codes +#define AV_UTF8_FLAG_EXCLUDE_XML_INVALID_CONTROL_CODES 8 ///< exclude control codes not accepted by XML + +#define AV_UTF8_FLAG_ACCEPT_ALL \ + AV_UTF8_FLAG_ACCEPT_INVALID_BIG_CODES|AV_UTF8_FLAG_ACCEPT_NON_CHARACTERS|AV_UTF8_FLAG_ACCEPT_SURROGATES + +/** + * Read and decode a single UTF-8 code point (character) from the + * buffer in *buf, and update *buf to point to the next byte to + * decode. + * + * In case of an invalid byte sequence, the pointer will be updated to + * the next byte after the invalid sequence and the function will + * return an error code. + * + * Depending on the specified flags, the function will also fail in + * case the decoded code point does not belong to a valid range. + * + * @note For speed-relevant code a carefully implemented use of + * GET_UTF8() may be preferred. + * + * @param codep pointer used to return the parsed code in case of success. + * The value in *codep is set even in case the range check fails. + * @param bufp pointer to the address the first byte of the sequence + * to decode, updated by the function to point to the + * byte next after the decoded sequence + * @param buf_end pointer to the end of the buffer, points to the next + * byte past the last in the buffer. This is used to + * avoid buffer overreads (in case of an unfinished + * UTF-8 sequence towards the end of the buffer). + * @param flags a collection of AV_UTF8_FLAG_* flags + * @return >= 0 in case a sequence was successfully read, a negative + * value in case of invalid sequence + */ +av_warn_unused_result +int av_utf8_decode(int32_t *codep, const uint8_t **bufp, const uint8_t *buf_end, + unsigned int flags); + +/** + * Check if a name is in a list. + * @returns 0 if not found, or the 1 based index where it has been found in the + * list. + */ +int av_match_list(const char *name, const char *list, char separator); + +/** + * See libc sscanf manual for more information. + * Locale-independent sscanf implementation. + */ +int av_sscanf(const char *string, const char *format, ...); + +/** + * @} + */ + +#endif /* AVUTIL_AVSTRING_H */ diff --git a/media/ffvpx/libavutil/avutil.h b/media/ffvpx/libavutil/avutil.h new file mode 100644 index 0000000000..a362c8baa8 --- /dev/null +++ b/media/ffvpx/libavutil/avutil.h @@ -0,0 +1,375 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_AVUTIL_H +#define AVUTIL_AVUTIL_H + +/** + * @file + * @ingroup lavu + * Convenience header that includes @ref lavu "libavutil"'s core. + */ + +/** + * @mainpage + * + * @section ffmpeg_intro Introduction + * + * This document describes the usage of the different libraries + * provided by FFmpeg. + * + * @li @ref libavc "libavcodec" encoding/decoding library + * @li @ref lavfi "libavfilter" graph-based frame editing library + * @li @ref libavf "libavformat" I/O and muxing/demuxing library + * @li @ref lavd "libavdevice" special devices muxing/demuxing library + * @li @ref lavu "libavutil" common utility library + * @li @ref lswr "libswresample" audio resampling, format conversion and mixing + * @li @ref lpp "libpostproc" post processing library + * @li @ref libsws "libswscale" color conversion and scaling library + * + * @section ffmpeg_versioning Versioning and compatibility + * + * Each of the FFmpeg libraries contains a version.h header, which defines a + * major, minor and micro version number with the + * <em>LIBRARYNAME_VERSION_{MAJOR,MINOR,MICRO}</em> macros. The major version + * number is incremented with backward incompatible changes - e.g. removing + * parts of the public API, reordering public struct members, etc. The minor + * version number is incremented for backward compatible API changes or major + * new features - e.g. adding a new public function or a new decoder. The micro + * version number is incremented for smaller changes that a calling program + * might still want to check for - e.g. changing behavior in a previously + * unspecified situation. + * + * FFmpeg guarantees backward API and ABI compatibility for each library as long + * as its major version number is unchanged. This means that no public symbols + * will be removed or renamed. Types and names of the public struct members and + * values of public macros and enums will remain the same (unless they were + * explicitly declared as not part of the public API). Documented behavior will + * not change. + * + * In other words, any correct program that works with a given FFmpeg snapshot + * should work just as well without any changes with any later snapshot with the + * same major versions. This applies to both rebuilding the program against new + * FFmpeg versions or to replacing the dynamic FFmpeg libraries that a program + * links against. + * + * However, new public symbols may be added and new members may be appended to + * public structs whose size is not part of public ABI (most public structs in + * FFmpeg). New macros and enum values may be added. Behavior in undocumented + * situations may change slightly (and be documented). All those are accompanied + * by an entry in doc/APIchanges and incrementing either the minor or micro + * version number. + */ + +/** + * @defgroup lavu libavutil + * Common code shared across all FFmpeg libraries. + * + * @note + * libavutil is designed to be modular. In most cases, in order to use the + * functions provided by one component of libavutil you must explicitly include + * the specific header containing that feature. If you are only using + * media-related components, you could simply include libavutil/avutil.h, which + * brings in most of the "core" components. + * + * @{ + * + * @defgroup lavu_crypto Crypto and Hashing + * + * @{ + * @} + * + * @defgroup lavu_math Mathematics + * @{ + * + * @} + * + * @defgroup lavu_string String Manipulation + * + * @{ + * + * @} + * + * @defgroup lavu_mem Memory Management + * + * @{ + * + * @} + * + * @defgroup lavu_data Data Structures + * @{ + * + * @} + * + * @defgroup lavu_video Video related + * + * @{ + * + * @} + * + * @defgroup lavu_audio Audio related + * + * @{ + * + * @} + * + * @defgroup lavu_error Error Codes + * + * @{ + * + * @} + * + * @defgroup lavu_log Logging Facility + * + * @{ + * + * @} + * + * @defgroup lavu_misc Other + * + * @{ + * + * @defgroup preproc_misc Preprocessor String Macros + * + * @{ + * + * @} + * + * @defgroup version_utils Library Version Macros + * + * @{ + * + * @} + */ + + +/** + * @addtogroup lavu_ver + * @{ + */ + +/** + * Return the LIBAVUTIL_VERSION_INT constant. + */ +unsigned avutil_version(void); + +/** + * Return an informative version string. This usually is the actual release + * version number or a git commit description. This string has no fixed format + * and can change any time. It should never be parsed by code. + */ +const char *av_version_info(void); + +/** + * Return the libavutil build-time configuration. + */ +const char *avutil_configuration(void); + +/** + * Return the libavutil license. + */ +const char *avutil_license(void); + +/** + * @} + */ + +/** + * @addtogroup lavu_media Media Type + * @brief Media Type + */ + +enum AVMediaType { + AVMEDIA_TYPE_UNKNOWN = -1, ///< Usually treated as AVMEDIA_TYPE_DATA + AVMEDIA_TYPE_VIDEO, + AVMEDIA_TYPE_AUDIO, + AVMEDIA_TYPE_DATA, ///< Opaque data information usually continuous + AVMEDIA_TYPE_SUBTITLE, + AVMEDIA_TYPE_ATTACHMENT, ///< Opaque data information usually sparse + AVMEDIA_TYPE_NB +}; + +/** + * Return a string describing the media_type enum, NULL if media_type + * is unknown. + */ +const char *av_get_media_type_string(enum AVMediaType media_type); + +/** + * @defgroup lavu_const Constants + * @{ + * + * @defgroup lavu_enc Encoding specific + * + * @note those definition should move to avcodec + * @{ + */ + +#define FF_LAMBDA_SHIFT 7 +#define FF_LAMBDA_SCALE (1<<FF_LAMBDA_SHIFT) +#define FF_QP2LAMBDA 118 ///< factor to convert from H.263 QP to lambda +#define FF_LAMBDA_MAX (256*128-1) + +#define FF_QUALITY_SCALE FF_LAMBDA_SCALE //FIXME maybe remove + +/** + * @} + * @defgroup lavu_time Timestamp specific + * + * FFmpeg internal timebase and timestamp definitions + * + * @{ + */ + +/** + * @brief Undefined timestamp value + * + * Usually reported by demuxer that work on containers that do not provide + * either pts or dts. + */ + +#define AV_NOPTS_VALUE ((int64_t)UINT64_C(0x8000000000000000)) + +/** + * Internal time base represented as integer + */ + +#define AV_TIME_BASE 1000000 + +/** + * Internal time base represented as fractional value + */ + +#ifdef __cplusplus +/* ISO C++ forbids compound-literals. */ +#define AV_TIME_BASE_Q av_make_q(1, AV_TIME_BASE) +#else +#define AV_TIME_BASE_Q (AVRational){1, AV_TIME_BASE} +#endif + +/** + * @} + * @} + * @defgroup lavu_picture Image related + * + * AVPicture types, pixel formats and basic image planes manipulation. + * + * @{ + */ + +enum AVPictureType { + AV_PICTURE_TYPE_NONE = 0, ///< Undefined + AV_PICTURE_TYPE_I, ///< Intra + AV_PICTURE_TYPE_P, ///< Predicted + AV_PICTURE_TYPE_B, ///< Bi-dir predicted + AV_PICTURE_TYPE_S, ///< S(GMC)-VOP MPEG-4 + AV_PICTURE_TYPE_SI, ///< Switching Intra + AV_PICTURE_TYPE_SP, ///< Switching Predicted + AV_PICTURE_TYPE_BI, ///< BI type +}; + +/** + * Return a single letter to describe the given picture type + * pict_type. + * + * @param[in] pict_type the picture type @return a single character + * representing the picture type, '?' if pict_type is unknown + */ +char av_get_picture_type_char(enum AVPictureType pict_type); + +/** + * @} + */ + +#include "common.h" +#include "rational.h" +#include "version.h" +#include "macros.h" +#include "mathematics.h" +#include "log.h" +#include "pixfmt.h" + +/** + * Return x default pointer in case p is NULL. + */ +static inline void *av_x_if_null(const void *p, const void *x) +{ + return (void *)(intptr_t)(p ? p : x); +} + +/** + * Compute the length of an integer list. + * + * @param elsize size in bytes of each list element (only 1, 2, 4 or 8) + * @param term list terminator (usually 0 or -1) + * @param list pointer to the list + * @return length of the list, in elements, not counting the terminator + */ +unsigned av_int_list_length_for_size(unsigned elsize, + const void *list, uint64_t term) av_pure; + +/** + * Compute the length of an integer list. + * + * @param term list terminator (usually 0 or -1) + * @param list pointer to the list + * @return length of the list, in elements, not counting the terminator + */ +#define av_int_list_length(list, term) \ + av_int_list_length_for_size(sizeof(*(list)), list, term) + +#if FF_API_AV_FOPEN_UTF8 +/** + * Open a file using a UTF-8 filename. + * The API of this function matches POSIX fopen(), errors are returned through + * errno. + * @deprecated Avoid using it, as on Windows, the FILE* allocated by this + * function may be allocated with a different CRT than the caller + * who uses the FILE*. No replacement provided in public API. + */ +attribute_deprecated +FILE *av_fopen_utf8(const char *path, const char *mode); +#endif + +/** + * Return the fractional representation of the internal time base. + */ +AVRational av_get_time_base_q(void); + +#define AV_FOURCC_MAX_STRING_SIZE 32 + +#define av_fourcc2str(fourcc) av_fourcc_make_string((char[AV_FOURCC_MAX_STRING_SIZE]){0}, fourcc) + +/** + * Fill the provided buffer with a string containing a FourCC (four-character + * code) representation. + * + * @param buf a buffer with size in bytes of at least AV_FOURCC_MAX_STRING_SIZE + * @param fourcc the fourcc to represent + * @return the buffer in input + */ +char *av_fourcc_make_string(char *buf, uint32_t fourcc); + +/** + * @} + * @} + */ + +#endif /* AVUTIL_AVUTIL_H */ diff --git a/media/ffvpx/libavutil/avutil.symbols b/media/ffvpx/libavutil/avutil.symbols new file mode 100644 index 0000000000..0ad6fad9cd --- /dev/null +++ b/media/ffvpx/libavutil/avutil.symbols @@ -0,0 +1,342 @@ +av_add_q +av_add_stable +av_append_path_component +av_asprintf +av_basename +#ifndef MOZ_FFVPX_AUDIOONLY +av_base64_decode +av_base64_encode +#endif +av_bprint_append_data +av_bprint_channel_layout +av_bprint_chars +av_bprint_clear +av_bprint_escape +av_bprint_finalize +av_bprint_get_buffer +av_bprint_init +av_bprint_init_for_buffer +av_bprint_strftime +av_bprintf +av_buffer_alloc +av_buffer_allocz +av_buffer_create +av_buffer_default_free +av_buffer_get_opaque +av_buffer_get_ref_count +av_buffer_is_writable +av_buffer_make_writable +av_buffer_pool_get +av_buffer_pool_init +av_buffer_pool_uninit +av_buffer_realloc +av_buffer_replace +av_buffer_ref +av_buffer_unref +av_calloc +av_channel_layout_check +av_channel_layout_compare +av_channel_layout_copy +av_channel_layout_default +av_channel_layout_describe +av_channel_layout_extract_channel +av_channel_layout_from_mask +av_channel_layout_from_mask +av_channel_layout_uninit +av_chroma_location_name +av_chroma_location_enum_to_pos +av_chroma_location_pos_to_enum +av_color_primaries_name +av_color_range_name +av_color_space_name +av_color_transfer_name +av_compare_mod +av_compare_ts +#ifndef MOZ_FFVPX_AUDIOONLY +av_content_light_metadata_create_side_data +#endif +av_cpu_count +av_crc +av_crc_get_table +av_crc_init +av_d2q +av_default_get_category +av_default_item_name +av_dict_copy +av_dict_count +av_dict_free +av_dict_get +av_dict_get_string +av_dict_iterate +av_dict_parse_string +av_dict_set +av_dict_set_int +av_dirname +av_div_q +#ifndef MOZ_FFVPX_AUDIOONLY +av_dynamic_hdr_plus_from_t35 +av_dynamic_hdr_plus_create_side_data +#endif +av_dynarray2_add +av_dynarray_add +av_dynarray_add_nofree +av_escape +av_expr_eval +av_expr_free +av_expr_parse +av_expr_parse_and_eval +av_fast_malloc +av_fast_mallocz +av_fast_realloc +av_fifo_alloc +av_fifo_alloc2 +av_fifo_alloc_array +av_fifo_can_read +av_fifo_drain +av_fifo_drain2 +av_fifo_free +av_fifo_freep +av_fifo_freep2 +av_fifo_generic_peek +av_fifo_generic_read +av_fifo_generic_write +av_fifo_grow +av_fifo_peek +av_fifo_realloc2 +av_fifo_read +av_fifo_reset +av_fifo_size +av_fifo_space +av_fifo_write +av_find_best_pix_fmt_of_2 +av_find_info_tag +av_find_nearest_q_idx +#ifndef MOZ_FFVPX_AUDIOONLY +av_film_grain_params_create_side_data +#endif +av_force_cpu_flags +av_fourcc_make_string +av_frame_alloc +av_frame_apply_cropping +av_frame_clone +av_frame_copy +av_frame_copy_props +av_frame_free +av_frame_get_buffer +av_frame_get_plane_buffer +av_frame_get_side_data +av_frame_is_writable +av_frame_make_writable +av_frame_move_ref +av_frame_new_side_data +av_frame_new_side_data_from_buf +av_frame_ref +av_frame_remove_side_data +av_frame_replace +av_frame_side_data_name +av_frame_unref +av_free +av_freep +av_gcd +av_get_alt_sample_fmt +av_get_bits_per_pixel +av_get_bytes_per_sample +av_get_channel_description +av_get_channel_layout +av_get_channel_layout_channel_index +av_get_channel_layout_nb_channels +av_get_channel_layout_string +av_get_channel_name +av_get_cpu_flags +av_get_default_channel_layout +av_get_known_color_name +av_get_media_type_string +av_get_packed_sample_fmt +av_get_padded_bits_per_pixel +av_get_picture_type_char +av_get_pix_fmt +av_get_pix_fmt_loss +av_get_pix_fmt_name +av_get_pix_fmt_string +av_get_planar_sample_fmt +av_get_sample_fmt +av_get_sample_fmt_string +av_get_standard_channel_layout +av_get_time_base_q +av_get_token +av_gettime +av_gettime_relative +av_gettime_relative_is_monotonic +av_hwdevice_get_hwframe_constraints +av_hwdevice_hwconfig_alloc +av_hwframe_constraints_free +av_hwframe_get_buffer +av_image_alloc +av_image_check_sar +av_image_check_size +av_image_check_size2 +av_image_copy +av_image_copy_plane +av_image_copy_to_buffer +av_image_fill_arrays +av_image_fill_linesizes +av_image_fill_max_pixsteps +av_image_fill_pointers +av_image_fill_plane_sizes +av_image_get_buffer_size +av_image_get_linesize +av_int_list_length_for_size +av_log +#ifndef MOZ_FFVPX_AUDIOONLY +av_log2 +av_log2_16bit +#endif +av_log_default_callback +av_log_format_line +av_log_get_flags +av_log_get_level +av_log_set_callback +av_log_set_flags +av_log_set_level +av_mallocz +av_malloc +#ifndef MOZ_FFVPX_AUDIOONLY +av_mastering_display_metadata_create_side_data +#endif +av_match_list +av_match_name +av_max_alloc +av_memcpy_backptr +av_memdup +av_mul_q +av_nearer_q +av_opt_child_next +av_opt_copy +av_opt_eval_double +av_opt_eval_flags +av_opt_eval_float +av_opt_eval_int +av_opt_eval_int64 +av_opt_eval_q +av_opt_find +av_opt_find2 +av_opt_flag_is_set +av_opt_free +av_opt_freep_ranges +av_opt_get +av_opt_get_channel_layout +av_opt_get_dict_val +av_opt_get_double +av_opt_get_image_size +av_opt_get_int +av_opt_get_key_value +av_opt_get_pixel_fmt +av_opt_get_q +av_opt_get_sample_fmt +av_opt_get_video_rate +av_opt_is_set_to_default +av_opt_is_set_to_default_by_name +av_opt_next +av_opt_ptr +av_opt_query_ranges +av_opt_query_ranges_default +av_opt_serialize +av_opt_set +av_opt_set_bin +av_opt_set_channel_layout +av_opt_set_defaults +av_opt_set_defaults2 +av_opt_set_dict +av_opt_set_dict2 +av_opt_set_dict_val +av_opt_set_double +av_opt_set_from_string +av_opt_set_image_size +av_opt_set_int +av_opt_set_pixel_fmt +av_opt_set_q +av_opt_set_sample_fmt +av_opt_set_video_rate +av_opt_show2 +av_parse_color +av_parse_cpu_caps +av_parse_ratio +av_parse_time +av_parse_video_rate +av_parse_video_size +av_pix_fmt_count_planes +av_pix_fmt_desc_get +av_pix_fmt_desc_get_id +av_pix_fmt_desc_next +av_pix_fmt_get_chroma_sub_sample +av_pix_fmt_swap_endianness +av_q2intfloat +av_read_image_line +av_realloc +av_realloc_array +av_realloc_f +av_reallocp +av_reallocp_array +av_reduce +av_rescale +av_rescale_delta +av_rescale_q +av_rescale_q_rnd +av_rescale_rnd +av_sample_fmt_is_planar +av_samples_alloc +av_samples_alloc_array_and_samples +av_samples_copy +av_samples_fill_arrays +av_samples_get_buffer_size +av_samples_set_silence +av_set_options_string +av_small_strptime +av_strcasecmp +av_strdup +av_strerror +av_stristart +av_stristr +av_strlcat +av_strlcatf +av_strlcpy +av_strncasecmp +av_strndup +av_strnstr +av_strstart +av_strtod +av_strtok +av_sub_q +av_timegm +av_usleep +av_utf8_decode +av_vbprintf +#ifndef MOZ_FFVPX_AUDIOONLY +av_video_enc_params_create_side_data +#endif +av_vlog +av_write_image_line +avpriv_alloc_fixed_dsp +avpriv_float_dsp_alloc +avpriv_report_missing_feature +avpriv_request_sample +avpriv_scalarproduct_float_c +avpriv_set_systematic_pal2 +#if defined(XP_WIN) && !defined(_ARM64_) +avpriv_emms_asm +#endif +avpriv_slicethread_create +avpriv_slicethread_execute +avpriv_slicethread_free +av_hwdevice_get_type_name +av_hwframe_ctx_alloc +av_hwframe_ctx_init +av_hwdevice_ctx_alloc +av_hwdevice_ctx_init +av_hwframe_transfer_get_formats +av_hwdevice_ctx_create_derived +av_malloc_array +av_get_sample_fmt_name +av_tx_uninit +av_tx_init +av_channel_layout_describe_bprint diff --git a/media/ffvpx/libavutil/avutilres.rc b/media/ffvpx/libavutil/avutilres.rc new file mode 100644 index 0000000000..40a75eb377 --- /dev/null +++ b/media/ffvpx/libavutil/avutilres.rc @@ -0,0 +1,55 @@ +/* + * Windows resource file for libavutil + * + * Copyright (C) 2012 James Almer + * Copyright (C) 2013 Tiancheng "Timothy" Gu + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <windows.h> +#include "libavutil/version.h" +#include "libavutil/ffversion.h" +#include "config.h" + +1 VERSIONINFO +FILEVERSION LIBAVUTIL_VERSION_MAJOR, LIBAVUTIL_VERSION_MINOR, LIBAVUTIL_VERSION_MICRO, 0 +PRODUCTVERSION LIBAVUTIL_VERSION_MAJOR, LIBAVUTIL_VERSION_MINOR, LIBAVUTIL_VERSION_MICRO, 0 +FILEFLAGSMASK VS_FFI_FILEFLAGSMASK +FILEOS VOS_NT_WINDOWS32 +FILETYPE VFT_DLL +{ + BLOCK "StringFileInfo" + { + BLOCK "040904B0" + { + VALUE "CompanyName", "FFmpeg Project" + VALUE "FileDescription", "FFmpeg utility library" + VALUE "FileVersion", AV_STRINGIFY(LIBAVUTIL_VERSION) + VALUE "InternalName", "libavutil" + VALUE "LegalCopyright", "Copyright (C) 2000-" AV_STRINGIFY(CONFIG_THIS_YEAR) " FFmpeg Project" + VALUE "OriginalFilename", "avutil" BUILDSUF "-" AV_STRINGIFY(LIBAVUTIL_VERSION_MAJOR) SLIBSUF + VALUE "ProductName", "FFmpeg" + VALUE "ProductVersion", FFMPEG_VERSION + } + } + + BLOCK "VarFileInfo" + { + VALUE "Translation", 0x0409, 0x04B0 + } +} diff --git a/media/ffvpx/libavutil/base64.c b/media/ffvpx/libavutil/base64.c new file mode 100644 index 0000000000..3e66f4fcbe --- /dev/null +++ b/media/ffvpx/libavutil/base64.c @@ -0,0 +1,181 @@ +/* + * Copyright (c) 2006 Ryan Martell. (rdm4@martellventures.com) + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @brief Base64 encode/decode + * @author Ryan Martell <rdm4@martellventures.com> (with lots of Michael) + */ + +#include <limits.h> +#include <stddef.h> + +#include "base64.h" +#include "error.h" +#include "intreadwrite.h" + +/* ---------------- private code */ +static const uint8_t map2[256] = +{ + 0xfe, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, + + 0x3e, 0xff, 0xff, 0xff, 0x3f, 0x34, 0x35, 0x36, + 0x37, 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0xff, + 0xff, 0xff, 0xfe, 0xff, 0xff, 0xff, 0x00, 0x01, + 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, + 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, + 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x1a, 0x1b, + 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 0x21, 0x22, 0x23, + 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, + 0x2c, 0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, + + 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, +}; + +#define BASE64_DEC_STEP(i) do { \ + bits = map2[in[i]]; \ + if (bits & 0x80) \ + goto out ## i; \ + v = i ? (v << 6) + bits : bits; \ +} while(0) + +int av_base64_decode(uint8_t *out, const char *in_str, int out_size) +{ + uint8_t *dst = out; + uint8_t *end; + // no sign extension + const uint8_t *in = in_str; + unsigned bits = 0xff; + unsigned v; + + if (!out) + goto validity_check; + + end = out + out_size; + while (end - dst > 3) { + BASE64_DEC_STEP(0); + BASE64_DEC_STEP(1); + BASE64_DEC_STEP(2); + BASE64_DEC_STEP(3); + // Using AV_WB32 directly confuses compiler + v = av_be2ne32(v << 8); + AV_WN32(dst, v); + dst += 3; + in += 4; + } + if (end - dst) { + BASE64_DEC_STEP(0); + BASE64_DEC_STEP(1); + BASE64_DEC_STEP(2); + BASE64_DEC_STEP(3); + *dst++ = v >> 16; + if (end - dst) + *dst++ = v >> 8; + if (end - dst) + *dst++ = v; + in += 4; + } +validity_check: + while (1) { + BASE64_DEC_STEP(0); + in++; + BASE64_DEC_STEP(0); + in++; + BASE64_DEC_STEP(0); + in++; + BASE64_DEC_STEP(0); + in++; + } + +out3: + *dst++ = v >> 10; + v <<= 2; +out2: + *dst++ = v >> 4; +out1: +out0: + return bits & 1 ? AVERROR_INVALIDDATA : out ? dst - out : 0; +} + +/***************************************************************************** +* b64_encode: Stolen from VLC's http.c. +* Simplified by Michael. +* Fixed edge cases and made it work from data (vs. strings) by Ryan. +*****************************************************************************/ + +char *av_base64_encode(char *out, int out_size, const uint8_t *in, int in_size) +{ + static const char b64[] = + "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; + char *ret, *dst; + unsigned i_bits = 0; + int i_shift = 0; + int bytes_remaining = in_size; + + if (in_size >= UINT_MAX / 4 || + out_size < AV_BASE64_SIZE(in_size)) + return NULL; + ret = dst = out; + while (bytes_remaining > 3) { + i_bits = AV_RB32(in); + in += 3; bytes_remaining -= 3; + *dst++ = b64[ i_bits>>26 ]; + *dst++ = b64[(i_bits>>20) & 0x3F]; + *dst++ = b64[(i_bits>>14) & 0x3F]; + *dst++ = b64[(i_bits>>8 ) & 0x3F]; + } + i_bits = 0; + while (bytes_remaining) { + i_bits = (i_bits << 8) + *in++; + bytes_remaining--; + i_shift += 8; + } + while (i_shift > 0) { + *dst++ = b64[(i_bits << 6 >> i_shift) & 0x3f]; + i_shift -= 6; + } + while ((dst - ret) & 3) + *dst++ = '='; + *dst = '\0'; + + return ret; +} diff --git a/media/ffvpx/libavutil/base64.h b/media/ffvpx/libavutil/base64.h new file mode 100644 index 0000000000..2954c12d42 --- /dev/null +++ b/media/ffvpx/libavutil/base64.h @@ -0,0 +1,72 @@ +/* + * Copyright (c) 2006 Ryan Martell. (rdm4@martellventures.com) + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_BASE64_H +#define AVUTIL_BASE64_H + +#include <stdint.h> + +/** + * @defgroup lavu_base64 Base64 + * @ingroup lavu_crypto + * @{ + */ + +/** + * Decode a base64-encoded string. + * + * @param out buffer for decoded data + * @param in null-terminated input string + * @param out_size size in bytes of the out buffer, must be at + * least 3/4 of the length of in, that is AV_BASE64_DECODE_SIZE(strlen(in)) + * @return number of bytes written, or a negative value in case of + * invalid input + */ +int av_base64_decode(uint8_t *out, const char *in, int out_size); + +/** + * Calculate the output size in bytes needed to decode a base64 string + * with length x to a data buffer. + */ +#define AV_BASE64_DECODE_SIZE(x) ((x) * 3LL / 4) + +/** + * Encode data to base64 and null-terminate. + * + * @param out buffer for encoded data + * @param out_size size in bytes of the out buffer (including the + * null terminator), must be at least AV_BASE64_SIZE(in_size) + * @param in input buffer containing the data to encode + * @param in_size size in bytes of the in buffer + * @return out or NULL in case of error + */ +char *av_base64_encode(char *out, int out_size, const uint8_t *in, int in_size); + +/** + * Calculate the output size needed to base64-encode x bytes to a + * null-terminated string. + */ +#define AV_BASE64_SIZE(x) (((x)+2) / 3 * 4 + 1) + + /** + * @} + */ + +#endif /* AVUTIL_BASE64_H */ diff --git a/media/ffvpx/libavutil/bprint.c b/media/ffvpx/libavutil/bprint.c new file mode 100644 index 0000000000..4e9571715c --- /dev/null +++ b/media/ffvpx/libavutil/bprint.c @@ -0,0 +1,337 @@ +/* + * Copyright (c) 2012 Nicolas George + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <limits.h> +#include <stdarg.h> +#include <stdio.h> +#include <string.h> +#include <time.h> +#include "avstring.h" +#include "bprint.h" +#include "compat/va_copy.h" +#include "error.h" +#include "macros.h" +#include "mem.h" + +#define av_bprint_room(buf) ((buf)->size - FFMIN((buf)->len, (buf)->size)) +#define av_bprint_is_allocated(buf) ((buf)->str != (buf)->reserved_internal_buffer) + +static int av_bprint_alloc(AVBPrint *buf, unsigned room) +{ + char *old_str, *new_str; + unsigned min_size, new_size; + + if (buf->size == buf->size_max) + return AVERROR(EIO); + if (!av_bprint_is_complete(buf)) + return AVERROR_INVALIDDATA; /* it is already truncated anyway */ + min_size = buf->len + 1 + FFMIN(UINT_MAX - buf->len - 1, room); + new_size = buf->size > buf->size_max / 2 ? buf->size_max : buf->size * 2; + if (new_size < min_size) + new_size = FFMIN(buf->size_max, min_size); + old_str = av_bprint_is_allocated(buf) ? buf->str : NULL; + new_str = av_realloc(old_str, new_size); + if (!new_str) + return AVERROR(ENOMEM); + if (!old_str) + memcpy(new_str, buf->str, buf->len + 1); + buf->str = new_str; + buf->size = new_size; + return 0; +} + +static void av_bprint_grow(AVBPrint *buf, unsigned extra_len) +{ + /* arbitrary margin to avoid small overflows */ + extra_len = FFMIN(extra_len, UINT_MAX - 5 - buf->len); + buf->len += extra_len; + if (buf->size) + buf->str[FFMIN(buf->len, buf->size - 1)] = 0; +} + +void av_bprint_init(AVBPrint *buf, unsigned size_init, unsigned size_max) +{ + unsigned size_auto = (char *)buf + sizeof(*buf) - + buf->reserved_internal_buffer; + + if (size_max == AV_BPRINT_SIZE_AUTOMATIC) + size_max = size_auto; + buf->str = buf->reserved_internal_buffer; + buf->len = 0; + buf->size = FFMIN(size_auto, size_max); + buf->size_max = size_max; + *buf->str = 0; + if (size_init > buf->size) + av_bprint_alloc(buf, size_init - 1); +} + +void av_bprint_init_for_buffer(AVBPrint *buf, char *buffer, unsigned size) +{ + if (size == 0) { + av_bprint_init(buf, 0, AV_BPRINT_SIZE_COUNT_ONLY); + return; + } + + buf->str = buffer; + buf->len = 0; + buf->size = size; + buf->size_max = size; + *buf->str = 0; +} + +void av_bprintf(AVBPrint *buf, const char *fmt, ...) +{ + unsigned room; + char *dst; + va_list vl; + int extra_len; + + while (1) { + room = av_bprint_room(buf); + dst = room ? buf->str + buf->len : NULL; + va_start(vl, fmt); + extra_len = vsnprintf(dst, room, fmt, vl); + va_end(vl); + if (extra_len <= 0) + return; + if (extra_len < room) + break; + if (av_bprint_alloc(buf, extra_len)) + break; + } + av_bprint_grow(buf, extra_len); +} + +void av_vbprintf(AVBPrint *buf, const char *fmt, va_list vl_arg) +{ + unsigned room; + char *dst; + int extra_len; + va_list vl; + + while (1) { + room = av_bprint_room(buf); + dst = room ? buf->str + buf->len : NULL; + va_copy(vl, vl_arg); + extra_len = vsnprintf(dst, room, fmt, vl); + va_end(vl); + if (extra_len <= 0) + return; + if (extra_len < room) + break; + if (av_bprint_alloc(buf, extra_len)) + break; + } + av_bprint_grow(buf, extra_len); +} + +void av_bprint_chars(AVBPrint *buf, char c, unsigned n) +{ + unsigned room, real_n; + + while (1) { + room = av_bprint_room(buf); + if (n < room) + break; + if (av_bprint_alloc(buf, n)) + break; + } + if (room) { + real_n = FFMIN(n, room - 1); + memset(buf->str + buf->len, c, real_n); + } + av_bprint_grow(buf, n); +} + +void av_bprint_append_data(AVBPrint *buf, const char *data, unsigned size) +{ + unsigned room, real_n; + + while (1) { + room = av_bprint_room(buf); + if (size < room) + break; + if (av_bprint_alloc(buf, size)) + break; + } + if (room) { + real_n = FFMIN(size, room - 1); + memcpy(buf->str + buf->len, data, real_n); + } + av_bprint_grow(buf, size); +} + +void av_bprint_strftime(AVBPrint *buf, const char *fmt, const struct tm *tm) +{ + unsigned room; + size_t l; + + if (!*fmt) + return; + while (1) { + room = av_bprint_room(buf); + if (room && (l = strftime(buf->str + buf->len, room, fmt, tm))) + break; + /* strftime does not tell us how much room it would need: let us + retry with twice as much until the buffer is large enough */ + room = !room ? strlen(fmt) + 1 : + room <= INT_MAX / 2 ? room * 2 : INT_MAX; + if (av_bprint_alloc(buf, room)) { + /* impossible to grow, try to manage something useful anyway */ + room = av_bprint_room(buf); + if (room < 1024) { + /* if strftime fails because the buffer has (almost) reached + its maximum size, let us try in a local buffer; 1k should + be enough to format any real date+time string */ + char buf2[1024]; + if ((l = strftime(buf2, sizeof(buf2), fmt, tm))) { + av_bprintf(buf, "%s", buf2); + return; + } + } + if (room) { + /* if anything else failed and the buffer is not already + truncated, let us add a stock string and force truncation */ + static const char txt[] = "[truncated strftime output]"; + memset(buf->str + buf->len, '!', room); + memcpy(buf->str + buf->len, txt, FFMIN(sizeof(txt) - 1, room)); + av_bprint_grow(buf, room); /* force truncation */ + } + return; + } + } + av_bprint_grow(buf, l); +} + +void av_bprint_get_buffer(AVBPrint *buf, unsigned size, + unsigned char **mem, unsigned *actual_size) +{ + if (size > av_bprint_room(buf)) + av_bprint_alloc(buf, size); + *actual_size = av_bprint_room(buf); + *mem = *actual_size ? buf->str + buf->len : NULL; +} + +void av_bprint_clear(AVBPrint *buf) +{ + if (buf->len) { + *buf->str = 0; + buf->len = 0; + } +} + +int av_bprint_finalize(AVBPrint *buf, char **ret_str) +{ + unsigned real_size = FFMIN(buf->len + 1, buf->size); + char *str; + int ret = 0; + + if (ret_str) { + if (av_bprint_is_allocated(buf)) { + str = av_realloc(buf->str, real_size); + if (!str) + str = buf->str; + buf->str = NULL; + } else { + str = av_memdup(buf->str, real_size); + if (!str) + ret = AVERROR(ENOMEM); + } + *ret_str = str; + } else { + if (av_bprint_is_allocated(buf)) + av_freep(&buf->str); + } + buf->size = real_size; + return ret; +} + +#define WHITESPACES " \n\t\r" + +void av_bprint_escape(AVBPrint *dstbuf, const char *src, const char *special_chars, + enum AVEscapeMode mode, int flags) +{ + const char *src0 = src; + + if (mode == AV_ESCAPE_MODE_AUTO) + mode = AV_ESCAPE_MODE_BACKSLASH; /* TODO: implement a heuristic */ + + switch (mode) { + case AV_ESCAPE_MODE_QUOTE: + /* enclose the string between '' */ + av_bprint_chars(dstbuf, '\'', 1); + for (; *src; src++) { + if (*src == '\'') + av_bprintf(dstbuf, "'\\''"); + else + av_bprint_chars(dstbuf, *src, 1); + } + av_bprint_chars(dstbuf, '\'', 1); + break; + + case AV_ESCAPE_MODE_XML: + /* escape XML non-markup character data as per 2.4 by default: */ + /* [^<&]* - ([^<&]* ']]>' [^<&]*) */ + + /* additionally, given one of the AV_ESCAPE_FLAG_XML_* flags, */ + /* escape those specific characters as required. */ + for (; *src; src++) { + switch (*src) { + case '&' : av_bprintf(dstbuf, "%s", "&"); break; + case '<' : av_bprintf(dstbuf, "%s", "<"); break; + case '>' : av_bprintf(dstbuf, "%s", ">"); break; + case '\'': + if (!(flags & AV_ESCAPE_FLAG_XML_SINGLE_QUOTES)) + goto XML_DEFAULT_HANDLING; + + av_bprintf(dstbuf, "%s", "'"); + break; + case '"' : + if (!(flags & AV_ESCAPE_FLAG_XML_DOUBLE_QUOTES)) + goto XML_DEFAULT_HANDLING; + + av_bprintf(dstbuf, "%s", """); + break; +XML_DEFAULT_HANDLING: + default: av_bprint_chars(dstbuf, *src, 1); + } + } + break; + + /* case AV_ESCAPE_MODE_BACKSLASH or unknown mode */ + default: + /* \-escape characters */ + for (; *src; src++) { + int is_first_last = src == src0 || !*(src+1); + int is_ws = !!strchr(WHITESPACES, *src); + int is_strictly_special = special_chars && strchr(special_chars, *src); + int is_special = + is_strictly_special || strchr("'\\", *src) || + (is_ws && (flags & AV_ESCAPE_FLAG_WHITESPACE)); + + if (is_strictly_special || + (!(flags & AV_ESCAPE_FLAG_STRICT) && + (is_special || (is_ws && is_first_last)))) + av_bprint_chars(dstbuf, '\\', 1); + av_bprint_chars(dstbuf, *src, 1); + } + break; + } +} diff --git a/media/ffvpx/libavutil/bprint.h b/media/ffvpx/libavutil/bprint.h new file mode 100644 index 0000000000..8559745478 --- /dev/null +++ b/media/ffvpx/libavutil/bprint.h @@ -0,0 +1,254 @@ +/* + * Copyright (c) 2012 Nicolas George + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_avbprint + * AVBPrint public header + */ + +#ifndef AVUTIL_BPRINT_H +#define AVUTIL_BPRINT_H + +#include <stdarg.h> + +#include "attributes.h" +#include "avstring.h" + +/** + * @defgroup lavu_avbprint AVBPrint + * @ingroup lavu_data + * + * A buffer to print data progressively + * @{ + */ + +/** + * Define a structure with extra padding to a fixed size + * This helps ensuring binary compatibility with future versions. + */ + +#define FF_PAD_STRUCTURE(name, size, ...) \ +struct ff_pad_helper_##name { __VA_ARGS__ }; \ +typedef struct name { \ + __VA_ARGS__ \ + char reserved_padding[size - sizeof(struct ff_pad_helper_##name)]; \ +} name; + +/** + * Buffer to print data progressively + * + * The string buffer grows as necessary and is always 0-terminated. + * The content of the string is never accessed, and thus is + * encoding-agnostic and can even hold binary data. + * + * Small buffers are kept in the structure itself, and thus require no + * memory allocation at all (unless the contents of the buffer is needed + * after the structure goes out of scope). This is almost as lightweight as + * declaring a local `char buf[512]`. + * + * The length of the string can go beyond the allocated size: the buffer is + * then truncated, but the functions still keep account of the actual total + * length. + * + * In other words, AVBPrint.len can be greater than AVBPrint.size and records + * the total length of what would have been to the buffer if there had been + * enough memory. + * + * Append operations do not need to be tested for failure: if a memory + * allocation fails, data stop being appended to the buffer, but the length + * is still updated. This situation can be tested with + * av_bprint_is_complete(). + * + * The AVBPrint.size_max field determines several possible behaviours: + * - `size_max = -1` (= `UINT_MAX`) or any large value will let the buffer be + * reallocated as necessary, with an amortized linear cost. + * - `size_max = 0` prevents writing anything to the buffer: only the total + * length is computed. The write operations can then possibly be repeated in + * a buffer with exactly the necessary size + * (using `size_init = size_max = len + 1`). + * - `size_max = 1` is automatically replaced by the exact size available in the + * structure itself, thus ensuring no dynamic memory allocation. The + * internal buffer is large enough to hold a reasonable paragraph of text, + * such as the current paragraph. + */ + +FF_PAD_STRUCTURE(AVBPrint, 1024, + char *str; /**< string so far */ + unsigned len; /**< length so far */ + unsigned size; /**< allocated memory */ + unsigned size_max; /**< maximum allocated memory */ + char reserved_internal_buffer[1]; +) + +/** + * @name Max size special values + * Convenience macros for special values for av_bprint_init() size_max + * parameter. + * @{ + */ + +/** + * Buffer will be reallocated as necessary, with an amortized linear cost. + */ +#define AV_BPRINT_SIZE_UNLIMITED ((unsigned)-1) +/** + * Use the exact size available in the AVBPrint structure itself. + * + * Thus ensuring no dynamic memory allocation. The internal buffer is large + * enough to hold a reasonable paragraph of text, such as the current paragraph. + */ +#define AV_BPRINT_SIZE_AUTOMATIC 1 +/** + * Do not write anything to the buffer, only calculate the total length. + * + * The write operations can then possibly be repeated in a buffer with + * exactly the necessary size (using `size_init = size_max = AVBPrint.len + 1`). + */ +#define AV_BPRINT_SIZE_COUNT_ONLY 0 +/** @} */ + +/** + * Init a print buffer. + * + * @param buf buffer to init + * @param size_init initial size (including the final 0) + * @param size_max maximum size; + * - `0` means do not write anything, just count the length + * - `1` is replaced by the maximum value for automatic storage + * any large value means that the internal buffer will be + * reallocated as needed up to that limit + * - `-1` is converted to `UINT_MAX`, the largest limit possible. + * Check also `AV_BPRINT_SIZE_*` macros. + */ +void av_bprint_init(AVBPrint *buf, unsigned size_init, unsigned size_max); + +/** + * Init a print buffer using a pre-existing buffer. + * + * The buffer will not be reallocated. + * In case size equals zero, the AVBPrint will be initialized to use + * the internal buffer as if using AV_BPRINT_SIZE_COUNT_ONLY with + * av_bprint_init(). + * + * @param buf buffer structure to init + * @param buffer byte buffer to use for the string data + * @param size size of buffer + */ +void av_bprint_init_for_buffer(AVBPrint *buf, char *buffer, unsigned size); + +/** + * Append a formatted string to a print buffer. + */ +void av_bprintf(AVBPrint *buf, const char *fmt, ...) av_printf_format(2, 3); + +/** + * Append a formatted string to a print buffer. + */ +void av_vbprintf(AVBPrint *buf, const char *fmt, va_list vl_arg); + +/** + * Append char c n times to a print buffer. + */ +void av_bprint_chars(AVBPrint *buf, char c, unsigned n); + +/** + * Append data to a print buffer. + * + * param buf bprint buffer to use + * param data pointer to data + * param size size of data + */ +void av_bprint_append_data(AVBPrint *buf, const char *data, unsigned size); + +struct tm; +/** + * Append a formatted date and time to a print buffer. + * + * param buf bprint buffer to use + * param fmt date and time format string, see strftime() + * param tm broken-down time structure to translate + * + * @note due to poor design of the standard strftime function, it may + * produce poor results if the format string expands to a very long text and + * the bprint buffer is near the limit stated by the size_max option. + */ +void av_bprint_strftime(AVBPrint *buf, const char *fmt, const struct tm *tm); + +/** + * Allocate bytes in the buffer for external use. + * + * @param[in] buf buffer structure + * @param[in] size required size + * @param[out] mem pointer to the memory area + * @param[out] actual_size size of the memory area after allocation; + * can be larger or smaller than size + */ +void av_bprint_get_buffer(AVBPrint *buf, unsigned size, + unsigned char **mem, unsigned *actual_size); + +/** + * Reset the string to "" but keep internal allocated data. + */ +void av_bprint_clear(AVBPrint *buf); + +/** + * Test if the print buffer is complete (not truncated). + * + * It may have been truncated due to a memory allocation failure + * or the size_max limit (compare size and size_max if necessary). + */ +static inline int av_bprint_is_complete(const AVBPrint *buf) +{ + return buf->len < buf->size; +} + +/** + * Finalize a print buffer. + * + * The print buffer can no longer be used afterwards, + * but the len and size fields are still valid. + * + * @arg[out] ret_str if not NULL, used to return a permanent copy of the + * buffer contents, or NULL if memory allocation fails; + * if NULL, the buffer is discarded and freed + * @return 0 for success or error code (probably AVERROR(ENOMEM)) + */ +int av_bprint_finalize(AVBPrint *buf, char **ret_str); + +/** + * Escape the content in src and append it to dstbuf. + * + * @param dstbuf already inited destination bprint buffer + * @param src string containing the text to escape + * @param special_chars string containing the special characters which + * need to be escaped, can be NULL + * @param mode escape mode to employ, see AV_ESCAPE_MODE_* macros. + * Any unknown value for mode will be considered equivalent to + * AV_ESCAPE_MODE_BACKSLASH, but this behaviour can change without + * notice. + * @param flags flags which control how to escape, see AV_ESCAPE_FLAG_* macros + */ +void av_bprint_escape(AVBPrint *dstbuf, const char *src, const char *special_chars, + enum AVEscapeMode mode, int flags); + +/** @} */ + +#endif /* AVUTIL_BPRINT_H */ diff --git a/media/ffvpx/libavutil/bswap.h b/media/ffvpx/libavutil/bswap.h new file mode 100644 index 0000000000..4840ab433f --- /dev/null +++ b/media/ffvpx/libavutil/bswap.h @@ -0,0 +1,111 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * byte swapping routines + */ + +#ifndef AVUTIL_BSWAP_H +#define AVUTIL_BSWAP_H + +#include <stdint.h> +#include "libavutil/avconfig.h" +#include "attributes.h" + +#ifdef HAVE_AV_CONFIG_H + +#include "config.h" + +#if ARCH_AARCH64 +# include "aarch64/bswap.h" +#elif ARCH_ARM +# include "arm/bswap.h" +#elif ARCH_AVR32 +# include "avr32/bswap.h" +#elif ARCH_RISCV +# include "riscv/bswap.h" +#elif ARCH_SH4 +# include "sh4/bswap.h" +#elif ARCH_X86 +# include "x86/bswap.h" +#endif + +#endif /* HAVE_AV_CONFIG_H */ + +#define AV_BSWAP16C(x) (((x) << 8 & 0xff00) | ((x) >> 8 & 0x00ff)) +#define AV_BSWAP32C(x) (AV_BSWAP16C(x) << 16 | AV_BSWAP16C((x) >> 16)) +#define AV_BSWAP64C(x) (AV_BSWAP32C(x) << 32 | AV_BSWAP32C((x) >> 32)) + +#define AV_BSWAPC(s, x) AV_BSWAP##s##C(x) + +#ifndef av_bswap16 +static av_always_inline av_const uint16_t av_bswap16(uint16_t x) +{ + x= (x>>8) | (x<<8); + return x; +} +#endif + +#ifndef av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + return AV_BSWAP32C(x); +} +#endif + +#ifndef av_bswap64 +static inline uint64_t av_const av_bswap64(uint64_t x) +{ + return (uint64_t)av_bswap32(x) << 32 | av_bswap32(x >> 32); +} +#endif + +// be2ne ... big-endian to native-endian +// le2ne ... little-endian to native-endian + +#if AV_HAVE_BIGENDIAN +#define av_be2ne16(x) (x) +#define av_be2ne32(x) (x) +#define av_be2ne64(x) (x) +#define av_le2ne16(x) av_bswap16(x) +#define av_le2ne32(x) av_bswap32(x) +#define av_le2ne64(x) av_bswap64(x) +#define AV_BE2NEC(s, x) (x) +#define AV_LE2NEC(s, x) AV_BSWAPC(s, x) +#else +#define av_be2ne16(x) av_bswap16(x) +#define av_be2ne32(x) av_bswap32(x) +#define av_be2ne64(x) av_bswap64(x) +#define av_le2ne16(x) (x) +#define av_le2ne32(x) (x) +#define av_le2ne64(x) (x) +#define AV_BE2NEC(s, x) AV_BSWAPC(s, x) +#define AV_LE2NEC(s, x) (x) +#endif + +#define AV_BE2NE16C(x) AV_BE2NEC(16, x) +#define AV_BE2NE32C(x) AV_BE2NEC(32, x) +#define AV_BE2NE64C(x) AV_BE2NEC(64, x) +#define AV_LE2NE16C(x) AV_LE2NEC(16, x) +#define AV_LE2NE32C(x) AV_LE2NEC(32, x) +#define AV_LE2NE64C(x) AV_LE2NEC(64, x) + +#endif /* AVUTIL_BSWAP_H */ diff --git a/media/ffvpx/libavutil/buffer.c b/media/ffvpx/libavutil/buffer.c new file mode 100644 index 0000000000..e4562a79b1 --- /dev/null +++ b/media/ffvpx/libavutil/buffer.c @@ -0,0 +1,416 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdatomic.h> +#include <stdint.h> +#include <string.h> + +#include "avassert.h" +#include "buffer_internal.h" +#include "common.h" +#include "mem.h" +#include "thread.h" + +static AVBufferRef *buffer_create(AVBuffer *buf, uint8_t *data, size_t size, + void (*free)(void *opaque, uint8_t *data), + void *opaque, int flags) +{ + AVBufferRef *ref = NULL; + + buf->data = data; + buf->size = size; + buf->free = free ? free : av_buffer_default_free; + buf->opaque = opaque; + + atomic_init(&buf->refcount, 1); + + buf->flags = flags; + + ref = av_mallocz(sizeof(*ref)); + if (!ref) + return NULL; + + ref->buffer = buf; + ref->data = data; + ref->size = size; + + return ref; +} + +AVBufferRef *av_buffer_create(uint8_t *data, size_t size, + void (*free)(void *opaque, uint8_t *data), + void *opaque, int flags) +{ + AVBufferRef *ret; + AVBuffer *buf = av_mallocz(sizeof(*buf)); + if (!buf) + return NULL; + + ret = buffer_create(buf, data, size, free, opaque, flags); + if (!ret) { + av_free(buf); + return NULL; + } + return ret; +} + +void av_buffer_default_free(void *opaque, uint8_t *data) +{ + av_free(data); +} + +AVBufferRef *av_buffer_alloc(size_t size) +{ + AVBufferRef *ret = NULL; + uint8_t *data = NULL; + + data = av_malloc(size); + if (!data) + return NULL; + + ret = av_buffer_create(data, size, av_buffer_default_free, NULL, 0); + if (!ret) + av_freep(&data); + + return ret; +} + +AVBufferRef *av_buffer_allocz(size_t size) +{ + AVBufferRef *ret = av_buffer_alloc(size); + if (!ret) + return NULL; + + memset(ret->data, 0, size); + return ret; +} + +AVBufferRef *av_buffer_ref(const AVBufferRef *buf) +{ + AVBufferRef *ret = av_mallocz(sizeof(*ret)); + + if (!ret) + return NULL; + + *ret = *buf; + + atomic_fetch_add_explicit(&buf->buffer->refcount, 1, memory_order_relaxed); + + return ret; +} + +static void buffer_replace(AVBufferRef **dst, AVBufferRef **src) +{ + AVBuffer *b; + + b = (*dst)->buffer; + + if (src) { + **dst = **src; + av_freep(src); + } else + av_freep(dst); + + if (atomic_fetch_sub_explicit(&b->refcount, 1, memory_order_acq_rel) == 1) { + /* b->free below might already free the structure containing *b, + * so we have to read the flag now to avoid use-after-free. */ + int free_avbuffer = !(b->flags_internal & BUFFER_FLAG_NO_FREE); + b->free(b->opaque, b->data); + if (free_avbuffer) + av_free(b); + } +} + +void av_buffer_unref(AVBufferRef **buf) +{ + if (!buf || !*buf) + return; + + buffer_replace(buf, NULL); +} + +int av_buffer_is_writable(const AVBufferRef *buf) +{ + if (buf->buffer->flags & AV_BUFFER_FLAG_READONLY) + return 0; + + return atomic_load(&buf->buffer->refcount) == 1; +} + +void *av_buffer_get_opaque(const AVBufferRef *buf) +{ + return buf->buffer->opaque; +} + +int av_buffer_get_ref_count(const AVBufferRef *buf) +{ + return atomic_load(&buf->buffer->refcount); +} + +int av_buffer_make_writable(AVBufferRef **pbuf) +{ + AVBufferRef *newbuf, *buf = *pbuf; + + if (av_buffer_is_writable(buf)) + return 0; + + newbuf = av_buffer_alloc(buf->size); + if (!newbuf) + return AVERROR(ENOMEM); + + memcpy(newbuf->data, buf->data, buf->size); + + buffer_replace(pbuf, &newbuf); + + return 0; +} + +int av_buffer_realloc(AVBufferRef **pbuf, size_t size) +{ + AVBufferRef *buf = *pbuf; + uint8_t *tmp; + int ret; + + if (!buf) { + /* allocate a new buffer with av_realloc(), so it will be reallocatable + * later */ + uint8_t *data = av_realloc(NULL, size); + if (!data) + return AVERROR(ENOMEM); + + buf = av_buffer_create(data, size, av_buffer_default_free, NULL, 0); + if (!buf) { + av_freep(&data); + return AVERROR(ENOMEM); + } + + buf->buffer->flags_internal |= BUFFER_FLAG_REALLOCATABLE; + *pbuf = buf; + + return 0; + } else if (buf->size == size) + return 0; + + if (!(buf->buffer->flags_internal & BUFFER_FLAG_REALLOCATABLE) || + !av_buffer_is_writable(buf) || buf->data != buf->buffer->data) { + /* cannot realloc, allocate a new reallocable buffer and copy data */ + AVBufferRef *new = NULL; + + ret = av_buffer_realloc(&new, size); + if (ret < 0) + return ret; + + memcpy(new->data, buf->data, FFMIN(size, buf->size)); + + buffer_replace(pbuf, &new); + return 0; + } + + tmp = av_realloc(buf->buffer->data, size); + if (!tmp) + return AVERROR(ENOMEM); + + buf->buffer->data = buf->data = tmp; + buf->buffer->size = buf->size = size; + return 0; +} + +int av_buffer_replace(AVBufferRef **pdst, const AVBufferRef *src) +{ + AVBufferRef *dst = *pdst; + AVBufferRef *tmp; + + if (!src) { + av_buffer_unref(pdst); + return 0; + } + + if (dst && dst->buffer == src->buffer) { + /* make sure the data pointers match */ + dst->data = src->data; + dst->size = src->size; + return 0; + } + + tmp = av_buffer_ref(src); + if (!tmp) + return AVERROR(ENOMEM); + + av_buffer_unref(pdst); + *pdst = tmp; + return 0; +} + +AVBufferPool *av_buffer_pool_init2(size_t size, void *opaque, + AVBufferRef* (*alloc)(void *opaque, size_t size), + void (*pool_free)(void *opaque)) +{ + AVBufferPool *pool = av_mallocz(sizeof(*pool)); + if (!pool) + return NULL; + + ff_mutex_init(&pool->mutex, NULL); + + pool->size = size; + pool->opaque = opaque; + pool->alloc2 = alloc; + pool->alloc = av_buffer_alloc; // fallback + pool->pool_free = pool_free; + + atomic_init(&pool->refcount, 1); + + return pool; +} + +AVBufferPool *av_buffer_pool_init(size_t size, AVBufferRef* (*alloc)(size_t size)) +{ + AVBufferPool *pool = av_mallocz(sizeof(*pool)); + if (!pool) + return NULL; + + ff_mutex_init(&pool->mutex, NULL); + + pool->size = size; + pool->alloc = alloc ? alloc : av_buffer_alloc; + + atomic_init(&pool->refcount, 1); + + return pool; +} + +static void buffer_pool_flush(AVBufferPool *pool) +{ + while (pool->pool) { + BufferPoolEntry *buf = pool->pool; + pool->pool = buf->next; + + buf->free(buf->opaque, buf->data); + av_freep(&buf); + } +} + +/* + * This function gets called when the pool has been uninited and + * all the buffers returned to it. + */ +static void buffer_pool_free(AVBufferPool *pool) +{ + buffer_pool_flush(pool); + ff_mutex_destroy(&pool->mutex); + + if (pool->pool_free) + pool->pool_free(pool->opaque); + + av_freep(&pool); +} + +void av_buffer_pool_uninit(AVBufferPool **ppool) +{ + AVBufferPool *pool; + + if (!ppool || !*ppool) + return; + pool = *ppool; + *ppool = NULL; + + ff_mutex_lock(&pool->mutex); + buffer_pool_flush(pool); + ff_mutex_unlock(&pool->mutex); + + if (atomic_fetch_sub_explicit(&pool->refcount, 1, memory_order_acq_rel) == 1) + buffer_pool_free(pool); +} + +static void pool_release_buffer(void *opaque, uint8_t *data) +{ + BufferPoolEntry *buf = opaque; + AVBufferPool *pool = buf->pool; + + ff_mutex_lock(&pool->mutex); + buf->next = pool->pool; + pool->pool = buf; + ff_mutex_unlock(&pool->mutex); + + if (atomic_fetch_sub_explicit(&pool->refcount, 1, memory_order_acq_rel) == 1) + buffer_pool_free(pool); +} + +/* allocate a new buffer and override its free() callback so that + * it is returned to the pool on free */ +static AVBufferRef *pool_alloc_buffer(AVBufferPool *pool) +{ + BufferPoolEntry *buf; + AVBufferRef *ret; + + av_assert0(pool->alloc || pool->alloc2); + + ret = pool->alloc2 ? pool->alloc2(pool->opaque, pool->size) : + pool->alloc(pool->size); + if (!ret) + return NULL; + + buf = av_mallocz(sizeof(*buf)); + if (!buf) { + av_buffer_unref(&ret); + return NULL; + } + + buf->data = ret->buffer->data; + buf->opaque = ret->buffer->opaque; + buf->free = ret->buffer->free; + buf->pool = pool; + + ret->buffer->opaque = buf; + ret->buffer->free = pool_release_buffer; + + return ret; +} + +AVBufferRef *av_buffer_pool_get(AVBufferPool *pool) +{ + AVBufferRef *ret; + BufferPoolEntry *buf; + + ff_mutex_lock(&pool->mutex); + buf = pool->pool; + if (buf) { + memset(&buf->buffer, 0, sizeof(buf->buffer)); + ret = buffer_create(&buf->buffer, buf->data, pool->size, + pool_release_buffer, buf, 0); + if (ret) { + pool->pool = buf->next; + buf->next = NULL; + buf->buffer.flags_internal |= BUFFER_FLAG_NO_FREE; + } + } else { + ret = pool_alloc_buffer(pool); + } + ff_mutex_unlock(&pool->mutex); + + if (ret) + atomic_fetch_add_explicit(&pool->refcount, 1, memory_order_relaxed); + + return ret; +} + +void *av_buffer_pool_buffer_get_opaque(const AVBufferRef *ref) +{ + BufferPoolEntry *buf = ref->buffer->opaque; + av_assert0(buf); + return buf->opaque; +} diff --git a/media/ffvpx/libavutil/buffer.h b/media/ffvpx/libavutil/buffer.h new file mode 100644 index 0000000000..e1ef5b7f07 --- /dev/null +++ b/media/ffvpx/libavutil/buffer.h @@ -0,0 +1,322 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_buffer + * refcounted data buffer API + */ + +#ifndef AVUTIL_BUFFER_H +#define AVUTIL_BUFFER_H + +#include <stddef.h> +#include <stdint.h> + +/** + * @defgroup lavu_buffer AVBuffer + * @ingroup lavu_data + * + * @{ + * AVBuffer is an API for reference-counted data buffers. + * + * There are two core objects in this API -- AVBuffer and AVBufferRef. AVBuffer + * represents the data buffer itself; it is opaque and not meant to be accessed + * by the caller directly, but only through AVBufferRef. However, the caller may + * e.g. compare two AVBuffer pointers to check whether two different references + * are describing the same data buffer. AVBufferRef represents a single + * reference to an AVBuffer and it is the object that may be manipulated by the + * caller directly. + * + * There are two functions provided for creating a new AVBuffer with a single + * reference -- av_buffer_alloc() to just allocate a new buffer, and + * av_buffer_create() to wrap an existing array in an AVBuffer. From an existing + * reference, additional references may be created with av_buffer_ref(). + * Use av_buffer_unref() to free a reference (this will automatically free the + * data once all the references are freed). + * + * The convention throughout this API and the rest of FFmpeg is such that the + * buffer is considered writable if there exists only one reference to it (and + * it has not been marked as read-only). The av_buffer_is_writable() function is + * provided to check whether this is true and av_buffer_make_writable() will + * automatically create a new writable buffer when necessary. + * Of course nothing prevents the calling code from violating this convention, + * however that is safe only when all the existing references are under its + * control. + * + * @note Referencing and unreferencing the buffers is thread-safe and thus + * may be done from multiple threads simultaneously without any need for + * additional locking. + * + * @note Two different references to the same buffer can point to different + * parts of the buffer (i.e. their AVBufferRef.data will not be equal). + */ + +/** + * A reference counted buffer type. It is opaque and is meant to be used through + * references (AVBufferRef). + */ +typedef struct AVBuffer AVBuffer; + +/** + * A reference to a data buffer. + * + * The size of this struct is not a part of the public ABI and it is not meant + * to be allocated directly. + */ +typedef struct AVBufferRef { + AVBuffer *buffer; + + /** + * The data buffer. It is considered writable if and only if + * this is the only reference to the buffer, in which case + * av_buffer_is_writable() returns 1. + */ + uint8_t *data; + /** + * Size of data in bytes. + */ + size_t size; +} AVBufferRef; + +/** + * Allocate an AVBuffer of the given size using av_malloc(). + * + * @return an AVBufferRef of given size or NULL when out of memory + */ +AVBufferRef *av_buffer_alloc(size_t size); + +/** + * Same as av_buffer_alloc(), except the returned buffer will be initialized + * to zero. + */ +AVBufferRef *av_buffer_allocz(size_t size); + +/** + * Always treat the buffer as read-only, even when it has only one + * reference. + */ +#define AV_BUFFER_FLAG_READONLY (1 << 0) + +/** + * Create an AVBuffer from an existing array. + * + * If this function is successful, data is owned by the AVBuffer. The caller may + * only access data through the returned AVBufferRef and references derived from + * it. + * If this function fails, data is left untouched. + * @param data data array + * @param size size of data in bytes + * @param free a callback for freeing this buffer's data + * @param opaque parameter to be got for processing or passed to free + * @param flags a combination of AV_BUFFER_FLAG_* + * + * @return an AVBufferRef referring to data on success, NULL on failure. + */ +AVBufferRef *av_buffer_create(uint8_t *data, size_t size, + void (*free)(void *opaque, uint8_t *data), + void *opaque, int flags); + +/** + * Default free callback, which calls av_free() on the buffer data. + * This function is meant to be passed to av_buffer_create(), not called + * directly. + */ +void av_buffer_default_free(void *opaque, uint8_t *data); + +/** + * Create a new reference to an AVBuffer. + * + * @return a new AVBufferRef referring to the same AVBuffer as buf or NULL on + * failure. + */ +AVBufferRef *av_buffer_ref(const AVBufferRef *buf); + +/** + * Free a given reference and automatically free the buffer if there are no more + * references to it. + * + * @param buf the reference to be freed. The pointer is set to NULL on return. + */ +void av_buffer_unref(AVBufferRef **buf); + +/** + * @return 1 if the caller may write to the data referred to by buf (which is + * true if and only if buf is the only reference to the underlying AVBuffer). + * Return 0 otherwise. + * A positive answer is valid until av_buffer_ref() is called on buf. + */ +int av_buffer_is_writable(const AVBufferRef *buf); + +/** + * @return the opaque parameter set by av_buffer_create. + */ +void *av_buffer_get_opaque(const AVBufferRef *buf); + +int av_buffer_get_ref_count(const AVBufferRef *buf); + +/** + * Create a writable reference from a given buffer reference, avoiding data copy + * if possible. + * + * @param buf buffer reference to make writable. On success, buf is either left + * untouched, or it is unreferenced and a new writable AVBufferRef is + * written in its place. On failure, buf is left untouched. + * @return 0 on success, a negative AVERROR on failure. + */ +int av_buffer_make_writable(AVBufferRef **buf); + +/** + * Reallocate a given buffer. + * + * @param buf a buffer reference to reallocate. On success, buf will be + * unreferenced and a new reference with the required size will be + * written in its place. On failure buf will be left untouched. *buf + * may be NULL, then a new buffer is allocated. + * @param size required new buffer size. + * @return 0 on success, a negative AVERROR on failure. + * + * @note the buffer is actually reallocated with av_realloc() only if it was + * initially allocated through av_buffer_realloc(NULL) and there is only one + * reference to it (i.e. the one passed to this function). In all other cases + * a new buffer is allocated and the data is copied. + */ +int av_buffer_realloc(AVBufferRef **buf, size_t size); + +/** + * Ensure dst refers to the same data as src. + * + * When *dst is already equivalent to src, do nothing. Otherwise unreference dst + * and replace it with a new reference to src. + * + * @param dst Pointer to either a valid buffer reference or NULL. On success, + * this will point to a buffer reference equivalent to src. On + * failure, dst will be left untouched. + * @param src A buffer reference to replace dst with. May be NULL, then this + * function is equivalent to av_buffer_unref(dst). + * @return 0 on success + * AVERROR(ENOMEM) on memory allocation failure. + */ +int av_buffer_replace(AVBufferRef **dst, const AVBufferRef *src); + +/** + * @} + */ + +/** + * @defgroup lavu_bufferpool AVBufferPool + * @ingroup lavu_data + * + * @{ + * AVBufferPool is an API for a lock-free thread-safe pool of AVBuffers. + * + * Frequently allocating and freeing large buffers may be slow. AVBufferPool is + * meant to solve this in cases when the caller needs a set of buffers of the + * same size (the most obvious use case being buffers for raw video or audio + * frames). + * + * At the beginning, the user must call av_buffer_pool_init() to create the + * buffer pool. Then whenever a buffer is needed, call av_buffer_pool_get() to + * get a reference to a new buffer, similar to av_buffer_alloc(). This new + * reference works in all aspects the same way as the one created by + * av_buffer_alloc(). However, when the last reference to this buffer is + * unreferenced, it is returned to the pool instead of being freed and will be + * reused for subsequent av_buffer_pool_get() calls. + * + * When the caller is done with the pool and no longer needs to allocate any new + * buffers, av_buffer_pool_uninit() must be called to mark the pool as freeable. + * Once all the buffers are released, it will automatically be freed. + * + * Allocating and releasing buffers with this API is thread-safe as long as + * either the default alloc callback is used, or the user-supplied one is + * thread-safe. + */ + +/** + * The buffer pool. This structure is opaque and not meant to be accessed + * directly. It is allocated with av_buffer_pool_init() and freed with + * av_buffer_pool_uninit(). + */ +typedef struct AVBufferPool AVBufferPool; + +/** + * Allocate and initialize a buffer pool. + * + * @param size size of each buffer in this pool + * @param alloc a function that will be used to allocate new buffers when the + * pool is empty. May be NULL, then the default allocator will be used + * (av_buffer_alloc()). + * @return newly created buffer pool on success, NULL on error. + */ +AVBufferPool *av_buffer_pool_init(size_t size, AVBufferRef* (*alloc)(size_t size)); + +/** + * Allocate and initialize a buffer pool with a more complex allocator. + * + * @param size size of each buffer in this pool + * @param opaque arbitrary user data used by the allocator + * @param alloc a function that will be used to allocate new buffers when the + * pool is empty. May be NULL, then the default allocator will be + * used (av_buffer_alloc()). + * @param pool_free a function that will be called immediately before the pool + * is freed. I.e. after av_buffer_pool_uninit() is called + * by the caller and all the frames are returned to the pool + * and freed. It is intended to uninitialize the user opaque + * data. May be NULL. + * @return newly created buffer pool on success, NULL on error. + */ +AVBufferPool *av_buffer_pool_init2(size_t size, void *opaque, + AVBufferRef* (*alloc)(void *opaque, size_t size), + void (*pool_free)(void *opaque)); + +/** + * Mark the pool as being available for freeing. It will actually be freed only + * once all the allocated buffers associated with the pool are released. Thus it + * is safe to call this function while some of the allocated buffers are still + * in use. + * + * @param pool pointer to the pool to be freed. It will be set to NULL. + */ +void av_buffer_pool_uninit(AVBufferPool **pool); + +/** + * Allocate a new AVBuffer, reusing an old buffer from the pool when available. + * This function may be called simultaneously from multiple threads. + * + * @return a reference to the new buffer on success, NULL on error. + */ +AVBufferRef *av_buffer_pool_get(AVBufferPool *pool); + +/** + * Query the original opaque parameter of an allocated buffer in the pool. + * + * @param ref a buffer reference to a buffer returned by av_buffer_pool_get. + * @return the opaque parameter set by the buffer allocator function of the + * buffer pool. + * + * @note the opaque parameter of ref is used by the buffer pool implementation, + * therefore you have to use this function to access the original opaque + * parameter of an allocated buffer. + */ +void *av_buffer_pool_buffer_get_opaque(const AVBufferRef *ref); + +/** + * @} + */ + +#endif /* AVUTIL_BUFFER_H */ diff --git a/media/ffvpx/libavutil/buffer_internal.h b/media/ffvpx/libavutil/buffer_internal.h new file mode 100644 index 0000000000..adb916aaa2 --- /dev/null +++ b/media/ffvpx/libavutil/buffer_internal.h @@ -0,0 +1,110 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_BUFFER_INTERNAL_H +#define AVUTIL_BUFFER_INTERNAL_H + +#include <stdatomic.h> +#include <stdint.h> + +#include "buffer.h" +#include "thread.h" + +/** + * The buffer was av_realloc()ed, so it is reallocatable. + */ +#define BUFFER_FLAG_REALLOCATABLE (1 << 0) +/** + * The AVBuffer structure is part of a larger structure + * and should not be freed. + */ +#define BUFFER_FLAG_NO_FREE (1 << 1) + +struct AVBuffer { + uint8_t *data; /**< data described by this buffer */ + size_t size; /**< size of data in bytes */ + + /** + * number of existing AVBufferRef instances referring to this buffer + */ + atomic_uint refcount; + + /** + * a callback for freeing the data + */ + void (*free)(void *opaque, uint8_t *data); + + /** + * an opaque pointer, to be used by the freeing callback + */ + void *opaque; + + /** + * A combination of AV_BUFFER_FLAG_* + */ + int flags; + + /** + * A combination of BUFFER_FLAG_* + */ + int flags_internal; +}; + +typedef struct BufferPoolEntry { + uint8_t *data; + + /* + * Backups of the original opaque/free of the AVBuffer corresponding to + * data. They will be used to free the buffer when the pool is freed. + */ + void *opaque; + void (*free)(void *opaque, uint8_t *data); + + AVBufferPool *pool; + struct BufferPoolEntry *next; + + /* + * An AVBuffer structure to (re)use as AVBuffer for subsequent uses + * of this BufferPoolEntry. + */ + AVBuffer buffer; +} BufferPoolEntry; + +struct AVBufferPool { + AVMutex mutex; + BufferPoolEntry *pool; + + /* + * This is used to track when the pool is to be freed. + * The pointer to the pool itself held by the caller is considered to + * be one reference. Each buffer requested by the caller increases refcount + * by one, returning the buffer to the pool decreases it by one. + * refcount reaches zero when the buffer has been uninited AND all the + * buffers have been released, then it's safe to free the pool and all + * the buffers in it. + */ + atomic_uint refcount; + + size_t size; + void *opaque; + AVBufferRef* (*alloc)(size_t size); + AVBufferRef* (*alloc2)(void *opaque, size_t size); + void (*pool_free)(void *opaque); +}; + +#endif /* AVUTIL_BUFFER_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/channel_layout.c b/media/ffvpx/libavutil/channel_layout.c new file mode 100644 index 0000000000..b59d798f29 --- /dev/null +++ b/media/ffvpx/libavutil/channel_layout.c @@ -0,0 +1,1018 @@ +/* + * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * audio channel layout utility functions + */ + +#include <stdint.h> +#include <stdlib.h> +#include <string.h> + +#include "avassert.h" +#include "channel_layout.h" +#include "bprint.h" +#include "common.h" +#include "error.h" +#include "macros.h" +#include "opt.h" + +#define CHAN_IS_AMBI(x) ((x) >= AV_CHAN_AMBISONIC_BASE &&\ + (x) <= AV_CHAN_AMBISONIC_END) + +struct channel_name { + const char *name; + const char *description; +}; + +static const struct channel_name channel_names[] = { + [AV_CHAN_FRONT_LEFT ] = { "FL", "front left" }, + [AV_CHAN_FRONT_RIGHT ] = { "FR", "front right" }, + [AV_CHAN_FRONT_CENTER ] = { "FC", "front center" }, + [AV_CHAN_LOW_FREQUENCY ] = { "LFE", "low frequency" }, + [AV_CHAN_BACK_LEFT ] = { "BL", "back left" }, + [AV_CHAN_BACK_RIGHT ] = { "BR", "back right" }, + [AV_CHAN_FRONT_LEFT_OF_CENTER ] = { "FLC", "front left-of-center" }, + [AV_CHAN_FRONT_RIGHT_OF_CENTER] = { "FRC", "front right-of-center" }, + [AV_CHAN_BACK_CENTER ] = { "BC", "back center" }, + [AV_CHAN_SIDE_LEFT ] = { "SL", "side left" }, + [AV_CHAN_SIDE_RIGHT ] = { "SR", "side right" }, + [AV_CHAN_TOP_CENTER ] = { "TC", "top center" }, + [AV_CHAN_TOP_FRONT_LEFT ] = { "TFL", "top front left" }, + [AV_CHAN_TOP_FRONT_CENTER ] = { "TFC", "top front center" }, + [AV_CHAN_TOP_FRONT_RIGHT ] = { "TFR", "top front right" }, + [AV_CHAN_TOP_BACK_LEFT ] = { "TBL", "top back left" }, + [AV_CHAN_TOP_BACK_CENTER ] = { "TBC", "top back center" }, + [AV_CHAN_TOP_BACK_RIGHT ] = { "TBR", "top back right" }, + [AV_CHAN_STEREO_LEFT ] = { "DL", "downmix left" }, + [AV_CHAN_STEREO_RIGHT ] = { "DR", "downmix right" }, + [AV_CHAN_WIDE_LEFT ] = { "WL", "wide left" }, + [AV_CHAN_WIDE_RIGHT ] = { "WR", "wide right" }, + [AV_CHAN_SURROUND_DIRECT_LEFT ] = { "SDL", "surround direct left" }, + [AV_CHAN_SURROUND_DIRECT_RIGHT] = { "SDR", "surround direct right" }, + [AV_CHAN_LOW_FREQUENCY_2 ] = { "LFE2", "low frequency 2" }, + [AV_CHAN_TOP_SIDE_LEFT ] = { "TSL", "top side left" }, + [AV_CHAN_TOP_SIDE_RIGHT ] = { "TSR", "top side right" }, + [AV_CHAN_BOTTOM_FRONT_CENTER ] = { "BFC", "bottom front center" }, + [AV_CHAN_BOTTOM_FRONT_LEFT ] = { "BFL", "bottom front left" }, + [AV_CHAN_BOTTOM_FRONT_RIGHT ] = { "BFR", "bottom front right" }, +}; + +static const char *get_channel_name(enum AVChannel channel_id) +{ + if ((unsigned) channel_id >= FF_ARRAY_ELEMS(channel_names) || + !channel_names[channel_id].name) + return NULL; + return channel_names[channel_id].name; +} + +void av_channel_name_bprint(AVBPrint *bp, enum AVChannel channel_id) +{ + if (channel_id >= AV_CHAN_AMBISONIC_BASE && + channel_id <= AV_CHAN_AMBISONIC_END) + av_bprintf(bp, "AMBI%d", channel_id - AV_CHAN_AMBISONIC_BASE); + else if ((unsigned)channel_id < FF_ARRAY_ELEMS(channel_names) && + channel_names[channel_id].name) + av_bprintf(bp, "%s", channel_names[channel_id].name); + else if (channel_id == AV_CHAN_NONE) + av_bprintf(bp, "NONE"); + else + av_bprintf(bp, "USR%d", channel_id); +} + +int av_channel_name(char *buf, size_t buf_size, enum AVChannel channel_id) +{ + AVBPrint bp; + + if (!buf && buf_size) + return AVERROR(EINVAL); + + av_bprint_init_for_buffer(&bp, buf, buf_size); + av_channel_name_bprint(&bp, channel_id); + + if (bp.len >= INT_MAX) + return AVERROR(ERANGE); + return bp.len + 1; +} + +void av_channel_description_bprint(AVBPrint *bp, enum AVChannel channel_id) +{ + if (channel_id >= AV_CHAN_AMBISONIC_BASE && + channel_id <= AV_CHAN_AMBISONIC_END) + av_bprintf(bp, "ambisonic ACN %d", channel_id - AV_CHAN_AMBISONIC_BASE); + else if ((unsigned)channel_id < FF_ARRAY_ELEMS(channel_names) && + channel_names[channel_id].description) + av_bprintf(bp, "%s", channel_names[channel_id].description); + else if (channel_id == AV_CHAN_NONE) + av_bprintf(bp, "none"); + else + av_bprintf(bp, "user %d", channel_id); +} + +int av_channel_description(char *buf, size_t buf_size, enum AVChannel channel_id) +{ + AVBPrint bp; + + if (!buf && buf_size) + return AVERROR(EINVAL); + + av_bprint_init_for_buffer(&bp, buf, buf_size); + av_channel_description_bprint(&bp, channel_id); + + if (bp.len >= INT_MAX) + return AVERROR(ERANGE); + return bp.len + 1; +} + +enum AVChannel av_channel_from_string(const char *str) +{ + int i; + char *endptr = (char *)str; + enum AVChannel id = AV_CHAN_NONE; + + if (!strncmp(str, "AMBI", 4)) { + i = strtol(str + 4, NULL, 0); + if (i < 0 || i > AV_CHAN_AMBISONIC_END - AV_CHAN_AMBISONIC_BASE) + return AV_CHAN_NONE; + return AV_CHAN_AMBISONIC_BASE + i; + } + + for (i = 0; i < FF_ARRAY_ELEMS(channel_names); i++) { + if (channel_names[i].name && !strcmp(str, channel_names[i].name)) + return i; + } + if (!strncmp(str, "USR", 3)) { + const char *p = str + 3; + id = strtol(p, &endptr, 0); + } + if (id >= 0 && !*endptr) + return id; + + return AV_CHAN_NONE; +} + +struct channel_layout_name { + const char *name; + AVChannelLayout layout; +}; + +static const struct channel_layout_name channel_layout_map[] = { + { "mono", AV_CHANNEL_LAYOUT_MONO }, + { "stereo", AV_CHANNEL_LAYOUT_STEREO }, + { "2.1", AV_CHANNEL_LAYOUT_2POINT1 }, + { "3.0", AV_CHANNEL_LAYOUT_SURROUND }, + { "3.0(back)", AV_CHANNEL_LAYOUT_2_1 }, + { "4.0", AV_CHANNEL_LAYOUT_4POINT0 }, + { "quad", AV_CHANNEL_LAYOUT_QUAD }, + { "quad(side)", AV_CHANNEL_LAYOUT_2_2 }, + { "3.1", AV_CHANNEL_LAYOUT_3POINT1 }, + { "5.0", AV_CHANNEL_LAYOUT_5POINT0_BACK }, + { "5.0(side)", AV_CHANNEL_LAYOUT_5POINT0 }, + { "4.1", AV_CHANNEL_LAYOUT_4POINT1 }, + { "5.1", AV_CHANNEL_LAYOUT_5POINT1_BACK }, + { "5.1(side)", AV_CHANNEL_LAYOUT_5POINT1 }, + { "6.0", AV_CHANNEL_LAYOUT_6POINT0 }, + { "6.0(front)", AV_CHANNEL_LAYOUT_6POINT0_FRONT }, + { "3.1.2", AV_CHANNEL_LAYOUT_3POINT1POINT2 }, + { "hexagonal", AV_CHANNEL_LAYOUT_HEXAGONAL }, + { "6.1", AV_CHANNEL_LAYOUT_6POINT1 }, + { "6.1(back)", AV_CHANNEL_LAYOUT_6POINT1_BACK }, + { "6.1(front)", AV_CHANNEL_LAYOUT_6POINT1_FRONT }, + { "7.0", AV_CHANNEL_LAYOUT_7POINT0 }, + { "7.0(front)", AV_CHANNEL_LAYOUT_7POINT0_FRONT }, + { "7.1", AV_CHANNEL_LAYOUT_7POINT1 }, + { "7.1(wide)", AV_CHANNEL_LAYOUT_7POINT1_WIDE_BACK }, + { "7.1(wide-side)", AV_CHANNEL_LAYOUT_7POINT1_WIDE }, + { "5.1.2", AV_CHANNEL_LAYOUT_5POINT1POINT2_BACK }, + { "octagonal", AV_CHANNEL_LAYOUT_OCTAGONAL }, + { "cube", AV_CHANNEL_LAYOUT_CUBE }, + { "5.1.4", AV_CHANNEL_LAYOUT_5POINT1POINT4_BACK }, + { "7.1.2", AV_CHANNEL_LAYOUT_7POINT1POINT2 }, + { "7.1.4", AV_CHANNEL_LAYOUT_7POINT1POINT4_BACK }, + { "7.2.3", AV_CHANNEL_LAYOUT_7POINT2POINT3 }, + { "9.1.4", AV_CHANNEL_LAYOUT_9POINT1POINT4_BACK }, + { "hexadecagonal", AV_CHANNEL_LAYOUT_HEXADECAGONAL }, + { "downmix", AV_CHANNEL_LAYOUT_STEREO_DOWNMIX, }, + { "22.2", AV_CHANNEL_LAYOUT_22POINT2, }, +}; + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS +static uint64_t get_channel_layout_single(const char *name, int name_len) +{ + int i; + char *end; + int64_t layout; + + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) { + if (strlen(channel_layout_map[i].name) == name_len && + !memcmp(channel_layout_map[i].name, name, name_len)) + return channel_layout_map[i].layout.u.mask; + } + for (i = 0; i < FF_ARRAY_ELEMS(channel_names); i++) + if (channel_names[i].name && + strlen(channel_names[i].name) == name_len && + !memcmp(channel_names[i].name, name, name_len)) + return (int64_t)1 << i; + + errno = 0; + i = strtol(name, &end, 10); + + if (!errno && (end + 1 - name == name_len && *end == 'c')) + return av_get_default_channel_layout(i); + + errno = 0; + layout = strtoll(name, &end, 0); + if (!errno && end - name == name_len) + return FFMAX(layout, 0); + return 0; +} + +uint64_t av_get_channel_layout(const char *name) +{ + const char *n, *e; + const char *name_end = name + strlen(name); + int64_t layout = 0, layout_single; + + for (n = name; n < name_end; n = e + 1) { + for (e = n; e < name_end && *e != '+' && *e != '|'; e++); + layout_single = get_channel_layout_single(n, e - n); + if (!layout_single) + return 0; + layout |= layout_single; + } + return layout; +} + +int av_get_extended_channel_layout(const char *name, uint64_t* channel_layout, int* nb_channels) +{ + int nb = 0; + char *end; + uint64_t layout = av_get_channel_layout(name); + + if (layout) { + *channel_layout = layout; + *nb_channels = av_get_channel_layout_nb_channels(layout); + return 0; + } + + nb = strtol(name, &end, 10); + if (!errno && *end == 'C' && *(end + 1) == '\0' && nb > 0 && nb < 64) { + *channel_layout = 0; + *nb_channels = nb; + return 0; + } + + return AVERROR(EINVAL); +} + +void av_bprint_channel_layout(struct AVBPrint *bp, + int nb_channels, uint64_t channel_layout) +{ + int i; + + if (nb_channels <= 0) + nb_channels = av_get_channel_layout_nb_channels(channel_layout); + + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) + if (nb_channels == channel_layout_map[i].layout.nb_channels && + channel_layout == channel_layout_map[i].layout.u.mask) { + av_bprintf(bp, "%s", channel_layout_map[i].name); + return; + } + + av_bprintf(bp, "%d channels", nb_channels); + if (channel_layout) { + int i, ch; + av_bprintf(bp, " ("); + for (i = 0, ch = 0; i < 64; i++) { + if ((channel_layout & (UINT64_C(1) << i))) { + const char *name = get_channel_name(i); + if (name) { + if (ch > 0) + av_bprintf(bp, "+"); + av_bprintf(bp, "%s", name); + } + ch++; + } + } + av_bprintf(bp, ")"); + } +} + +void av_get_channel_layout_string(char *buf, int buf_size, + int nb_channels, uint64_t channel_layout) +{ + AVBPrint bp; + + av_bprint_init_for_buffer(&bp, buf, buf_size); + av_bprint_channel_layout(&bp, nb_channels, channel_layout); +} + +int av_get_channel_layout_nb_channels(uint64_t channel_layout) +{ + return av_popcount64(channel_layout); +} + +int64_t av_get_default_channel_layout(int nb_channels) { + int i; + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) + if (nb_channels == channel_layout_map[i].layout.nb_channels) + return channel_layout_map[i].layout.u.mask; + return 0; +} + +int av_get_channel_layout_channel_index(uint64_t channel_layout, + uint64_t channel) +{ + if (!(channel_layout & channel) || + av_get_channel_layout_nb_channels(channel) != 1) + return AVERROR(EINVAL); + channel_layout &= channel - 1; + return av_get_channel_layout_nb_channels(channel_layout); +} + +const char *av_get_channel_name(uint64_t channel) +{ + int i; + if (av_get_channel_layout_nb_channels(channel) != 1) + return NULL; + for (i = 0; i < 64; i++) + if ((1ULL<<i) & channel) + return get_channel_name(i); + return NULL; +} + +const char *av_get_channel_description(uint64_t channel) +{ + int i; + if (av_get_channel_layout_nb_channels(channel) != 1) + return NULL; + for (i = 0; i < FF_ARRAY_ELEMS(channel_names); i++) + if ((1ULL<<i) & channel) + return channel_names[i].description; + return NULL; +} + +uint64_t av_channel_layout_extract_channel(uint64_t channel_layout, int index) +{ + int i; + + if (av_get_channel_layout_nb_channels(channel_layout) <= index) + return 0; + + for (i = 0; i < 64; i++) { + if ((1ULL << i) & channel_layout && !index--) + return 1ULL << i; + } + return 0; +} + +int av_get_standard_channel_layout(unsigned index, uint64_t *layout, + const char **name) +{ + if (index >= FF_ARRAY_ELEMS(channel_layout_map)) + return AVERROR_EOF; + if (layout) *layout = channel_layout_map[index].layout.u.mask; + if (name) *name = channel_layout_map[index].name; + return 0; +} +FF_ENABLE_DEPRECATION_WARNINGS +#endif + +int av_channel_layout_from_mask(AVChannelLayout *channel_layout, + uint64_t mask) +{ + if (!mask) + return AVERROR(EINVAL); + + channel_layout->order = AV_CHANNEL_ORDER_NATIVE; + channel_layout->nb_channels = av_popcount64(mask); + channel_layout->u.mask = mask; + + return 0; +} + +int av_channel_layout_from_string(AVChannelLayout *channel_layout, + const char *str) +{ + int i; + int channels = 0, nb_channels = 0, native = 1; + enum AVChannel highest_channel = AV_CHAN_NONE; + const char *dup; + char *chlist, *end; + uint64_t mask = 0; + + /* channel layout names */ + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) { + if (channel_layout_map[i].name && !strcmp(str, channel_layout_map[i].name)) { + *channel_layout = channel_layout_map[i].layout; + return 0; + } + } + + /* ambisonic */ + if (!strncmp(str, "ambisonic ", 10)) { + const char *p = str + 10; + char *endptr; + AVChannelLayout extra = {0}; + int order; + + order = strtol(p, &endptr, 0); + if (order < 0 || order + 1 > INT_MAX / (order + 1) || + (*endptr && *endptr != '+')) + return AVERROR(EINVAL); + + channel_layout->order = AV_CHANNEL_ORDER_AMBISONIC; + channel_layout->nb_channels = (order + 1) * (order + 1); + + if (*endptr) { + int ret = av_channel_layout_from_string(&extra, endptr + 1); + if (ret < 0) + return ret; + if (extra.nb_channels >= INT_MAX - channel_layout->nb_channels) { + av_channel_layout_uninit(&extra); + return AVERROR(EINVAL); + } + + if (extra.order == AV_CHANNEL_ORDER_NATIVE) { + channel_layout->u.mask = extra.u.mask; + } else { + channel_layout->order = AV_CHANNEL_ORDER_CUSTOM; + channel_layout->u.map = + av_calloc(channel_layout->nb_channels + extra.nb_channels, + sizeof(*channel_layout->u.map)); + if (!channel_layout->u.map) { + av_channel_layout_uninit(&extra); + return AVERROR(ENOMEM); + } + + for (i = 0; i < channel_layout->nb_channels; i++) + channel_layout->u.map[i].id = AV_CHAN_AMBISONIC_BASE + i; + for (i = 0; i < extra.nb_channels; i++) { + enum AVChannel ch = av_channel_layout_channel_from_index(&extra, i); + if (CHAN_IS_AMBI(ch)) { + av_channel_layout_uninit(&extra); + return AVERROR(EINVAL); + } + channel_layout->u.map[channel_layout->nb_channels + i].id = ch; + if (extra.order == AV_CHANNEL_ORDER_CUSTOM && + extra.u.map[i].name[0]) + av_strlcpy(channel_layout->u.map[channel_layout->nb_channels + i].name, + extra.u.map[i].name, + sizeof(channel_layout->u.map[channel_layout->nb_channels + i].name)); + } + } + channel_layout->nb_channels += extra.nb_channels; + av_channel_layout_uninit(&extra); + } + + return 0; + } + + chlist = av_strdup(str); + if (!chlist) + return AVERROR(ENOMEM); + + /* channel names */ + av_sscanf(str, "%d channels (%[^)]", &nb_channels, chlist); + end = strchr(str, ')'); + + dup = chlist; + while (*dup) { + char *channel, *chname; + int ret = av_opt_get_key_value(&dup, "@", "+", AV_OPT_FLAG_IMPLICIT_KEY, &channel, &chname); + if (ret < 0) { + av_free(chlist); + return ret; + } + if (*dup) + dup++; // skip separator + if (channel && !*channel) + av_freep(&channel); + for (i = 0; i < FF_ARRAY_ELEMS(channel_names); i++) { + if (channel_names[i].name && !strcmp(channel ? channel : chname, channel_names[i].name)) { + if (channel || i < highest_channel || mask & (1ULL << i)) + native = 0; // Not a native layout, use a custom one + highest_channel = i; + mask |= 1ULL << i; + break; + } + } + + if (!channel && i >= FF_ARRAY_ELEMS(channel_names)) { + char *endptr = chname; + enum AVChannel id = AV_CHAN_NONE; + + if (!strncmp(chname, "USR", 3)) { + const char *p = chname + 3; + id = strtol(p, &endptr, 0); + } + if (id < 0 || *endptr) { + native = 0; // Unknown channel name + channels = 0; + mask = 0; + av_free(chname); + break; + } + if (id > 63) + native = 0; // Not a native layout, use a custom one + else { + if (id < highest_channel || mask & (1ULL << id)) + native = 0; // Not a native layout, use a custom one + highest_channel = id; + mask |= 1ULL << id; + } + } + channels++; + av_free(channel); + av_free(chname); + } + + if (mask && native) { + av_free(chlist); + if (nb_channels && ((nb_channels != channels) || (!end || *++end))) + return AVERROR(EINVAL); + av_channel_layout_from_mask(channel_layout, mask); + return 0; + } + + /* custom layout of channel names */ + if (channels && !native) { + int idx = 0; + + if (nb_channels && ((nb_channels != channels) || (!end || *++end))) { + av_free(chlist); + return AVERROR(EINVAL); + } + + channel_layout->u.map = av_calloc(channels, sizeof(*channel_layout->u.map)); + if (!channel_layout->u.map) { + av_free(chlist); + return AVERROR(ENOMEM); + } + + channel_layout->order = AV_CHANNEL_ORDER_CUSTOM; + channel_layout->nb_channels = channels; + + dup = chlist; + while (*dup) { + char *channel, *chname; + int ret = av_opt_get_key_value(&dup, "@", "+", AV_OPT_FLAG_IMPLICIT_KEY, &channel, &chname); + if (ret < 0) { + av_freep(&channel_layout->u.map); + av_free(chlist); + return ret; + } + if (*dup) + dup++; // skip separator + for (i = 0; i < FF_ARRAY_ELEMS(channel_names); i++) { + if (channel_names[i].name && !strcmp(channel ? channel : chname, channel_names[i].name)) { + channel_layout->u.map[idx].id = i; + if (channel) + av_strlcpy(channel_layout->u.map[idx].name, chname, sizeof(channel_layout->u.map[idx].name)); + idx++; + break; + } + } + if (i >= FF_ARRAY_ELEMS(channel_names)) { + const char *p = (channel ? channel : chname) + 3; + channel_layout->u.map[idx].id = strtol(p, NULL, 0); + if (channel) + av_strlcpy(channel_layout->u.map[idx].name, chname, sizeof(channel_layout->u.map[idx].name)); + idx++; + } + av_free(channel); + av_free(chname); + } + av_free(chlist); + + return 0; + } + av_freep(&chlist); + + errno = 0; + mask = strtoull(str, &end, 0); + + /* channel layout mask */ + if (!errno && !*end && !strchr(str, '-') && mask) { + av_channel_layout_from_mask(channel_layout, mask); + return 0; + } + + errno = 0; + channels = strtol(str, &end, 10); + + /* number of channels */ + if (!errno && !strcmp(end, "c") && channels > 0) { + av_channel_layout_default(channel_layout, channels); + if (channel_layout->order == AV_CHANNEL_ORDER_NATIVE) + return 0; + } + + /* number of unordered channels */ + if (!errno && (!strcmp(end, "C") || !strcmp(end, " channels")) + && channels > 0) { + channel_layout->order = AV_CHANNEL_ORDER_UNSPEC; + channel_layout->nb_channels = channels; + return 0; + } + + return AVERROR(EINVAL); +} + +void av_channel_layout_uninit(AVChannelLayout *channel_layout) +{ + if (channel_layout->order == AV_CHANNEL_ORDER_CUSTOM) + av_freep(&channel_layout->u.map); + memset(channel_layout, 0, sizeof(*channel_layout)); +} + +int av_channel_layout_copy(AVChannelLayout *dst, const AVChannelLayout *src) +{ + av_channel_layout_uninit(dst); + *dst = *src; + if (src->order == AV_CHANNEL_ORDER_CUSTOM) { + dst->u.map = av_malloc_array(src->nb_channels, sizeof(*dst->u.map)); + if (!dst->u.map) + return AVERROR(ENOMEM); + memcpy(dst->u.map, src->u.map, src->nb_channels * sizeof(*src->u.map)); + } + return 0; +} + +/** + * If the layout is n-th order standard-order ambisonic, with optional + * extra non-diegetic channels at the end, return the order. + * Return a negative error code otherwise. + */ +static int ambisonic_order(const AVChannelLayout *channel_layout) +{ + int i, highest_ambi, order; + + highest_ambi = -1; + if (channel_layout->order == AV_CHANNEL_ORDER_AMBISONIC) + highest_ambi = channel_layout->nb_channels - av_popcount64(channel_layout->u.mask) - 1; + else { + const AVChannelCustom *map = channel_layout->u.map; + av_assert0(channel_layout->order == AV_CHANNEL_ORDER_CUSTOM); + + for (i = 0; i < channel_layout->nb_channels; i++) { + int is_ambi = CHAN_IS_AMBI(map[i].id); + + /* ambisonic following non-ambisonic */ + if (i > 0 && is_ambi && !CHAN_IS_AMBI(map[i - 1].id)) + return AVERROR(EINVAL); + + /* non-default ordering */ + if (is_ambi && map[i].id - AV_CHAN_AMBISONIC_BASE != i) + return AVERROR(EINVAL); + + if (CHAN_IS_AMBI(map[i].id)) + highest_ambi = i; + } + } + /* no ambisonic channels*/ + if (highest_ambi < 0) + return AVERROR(EINVAL); + + order = floor(sqrt(highest_ambi)); + /* incomplete order - some harmonics are missing */ + if ((order + 1) * (order + 1) != highest_ambi + 1) + return AVERROR(EINVAL); + + return order; +} + +/** + * If the custom layout is n-th order standard-order ambisonic, with optional + * extra non-diegetic channels at the end, write its string description in bp. + * Return a negative error code otherwise. + */ +static int try_describe_ambisonic(AVBPrint *bp, const AVChannelLayout *channel_layout) +{ + int nb_ambi_channels; + int order = ambisonic_order(channel_layout); + if (order < 0) + return order; + + av_bprintf(bp, "ambisonic %d", order); + + /* extra channels present */ + nb_ambi_channels = (order + 1) * (order + 1); + if (nb_ambi_channels < channel_layout->nb_channels) { + AVChannelLayout extra = { 0 }; + + if (channel_layout->order == AV_CHANNEL_ORDER_AMBISONIC) { + extra.order = AV_CHANNEL_ORDER_NATIVE; + extra.nb_channels = av_popcount64(channel_layout->u.mask); + extra.u.mask = channel_layout->u.mask; + } else { + extra.order = AV_CHANNEL_ORDER_CUSTOM; + extra.nb_channels = channel_layout->nb_channels - nb_ambi_channels; + extra.u.map = channel_layout->u.map + nb_ambi_channels; + } + + av_bprint_chars(bp, '+', 1); + av_channel_layout_describe_bprint(&extra, bp); + /* Not calling uninit here on extra because we don't own the u.map pointer */ + } + + return 0; +} + +int av_channel_layout_describe_bprint(const AVChannelLayout *channel_layout, + AVBPrint *bp) +{ + int i; + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_NATIVE: + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) + if (channel_layout->u.mask == channel_layout_map[i].layout.u.mask) { + av_bprintf(bp, "%s", channel_layout_map[i].name); + return 0; + } + // fall-through + case AV_CHANNEL_ORDER_CUSTOM: + if (channel_layout->order == AV_CHANNEL_ORDER_CUSTOM) { + int res = try_describe_ambisonic(bp, channel_layout); + if (res >= 0) + return 0; + } + if (channel_layout->nb_channels) + av_bprintf(bp, "%d channels (", channel_layout->nb_channels); + for (i = 0; i < channel_layout->nb_channels; i++) { + enum AVChannel ch = av_channel_layout_channel_from_index(channel_layout, i); + + if (i) + av_bprintf(bp, "+"); + av_channel_name_bprint(bp, ch); + if (channel_layout->order == AV_CHANNEL_ORDER_CUSTOM && + channel_layout->u.map[i].name[0]) + av_bprintf(bp, "@%s", channel_layout->u.map[i].name); + } + if (channel_layout->nb_channels) { + av_bprintf(bp, ")"); + return 0; + } + // fall-through + case AV_CHANNEL_ORDER_UNSPEC: + av_bprintf(bp, "%d channels", channel_layout->nb_channels); + return 0; + case AV_CHANNEL_ORDER_AMBISONIC: + return try_describe_ambisonic(bp, channel_layout); + default: + return AVERROR(EINVAL); + } +} + +int av_channel_layout_describe(const AVChannelLayout *channel_layout, + char *buf, size_t buf_size) +{ + AVBPrint bp; + int ret; + + if (!buf && buf_size) + return AVERROR(EINVAL); + + av_bprint_init_for_buffer(&bp, buf, buf_size); + ret = av_channel_layout_describe_bprint(channel_layout, &bp); + if (ret < 0) + return ret; + + if (bp.len >= INT_MAX) + return AVERROR(ERANGE); + return bp.len + 1; +} + +enum AVChannel +av_channel_layout_channel_from_index(const AVChannelLayout *channel_layout, + unsigned int idx) +{ + int i; + + if (idx >= channel_layout->nb_channels) + return AV_CHAN_NONE; + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_CUSTOM: + return channel_layout->u.map[idx].id; + case AV_CHANNEL_ORDER_AMBISONIC: { + int ambi_channels = channel_layout->nb_channels - av_popcount64(channel_layout->u.mask); + if (idx < ambi_channels) + return AV_CHAN_AMBISONIC_BASE + idx; + idx -= ambi_channels; + } + // fall-through + case AV_CHANNEL_ORDER_NATIVE: + for (i = 0; i < 64; i++) { + if ((1ULL << i) & channel_layout->u.mask && !idx--) + return i; + } + default: + return AV_CHAN_NONE; + } +} + +enum AVChannel +av_channel_layout_channel_from_string(const AVChannelLayout *channel_layout, + const char *str) +{ + int index = av_channel_layout_index_from_string(channel_layout, str); + + if (index < 0) + return AV_CHAN_NONE; + + return av_channel_layout_channel_from_index(channel_layout, index); +} + +int av_channel_layout_index_from_channel(const AVChannelLayout *channel_layout, + enum AVChannel channel) +{ + int i; + + if (channel == AV_CHAN_NONE) + return AVERROR(EINVAL); + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_CUSTOM: + for (i = 0; i < channel_layout->nb_channels; i++) + if (channel_layout->u.map[i].id == channel) + return i; + return AVERROR(EINVAL); + case AV_CHANNEL_ORDER_AMBISONIC: + case AV_CHANNEL_ORDER_NATIVE: { + uint64_t mask = channel_layout->u.mask; + int ambi_channels = channel_layout->nb_channels - av_popcount64(mask); + if (channel_layout->order == AV_CHANNEL_ORDER_AMBISONIC && + channel >= AV_CHAN_AMBISONIC_BASE) { + if (channel - AV_CHAN_AMBISONIC_BASE >= ambi_channels) + return AVERROR(EINVAL); + return channel - AV_CHAN_AMBISONIC_BASE; + } + if ((unsigned)channel > 63 || !(mask & (1ULL << channel))) + return AVERROR(EINVAL); + mask &= (1ULL << channel) - 1; + return av_popcount64(mask) + ambi_channels; + } + default: + return AVERROR(EINVAL); + } +} + +int av_channel_layout_index_from_string(const AVChannelLayout *channel_layout, + const char *str) +{ + char *chname; + enum AVChannel ch = AV_CHAN_NONE; + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_CUSTOM: + chname = strstr(str, "@"); + if (chname) { + char buf[16]; + chname++; + av_strlcpy(buf, str, FFMIN(sizeof(buf), chname - str)); + if (!*chname) + chname = NULL; + ch = av_channel_from_string(buf); + if (ch == AV_CHAN_NONE && *buf) + return AVERROR(EINVAL); + } + for (int i = 0; chname && i < channel_layout->nb_channels; i++) { + if (!strcmp(chname, channel_layout->u.map[i].name) && + (ch == AV_CHAN_NONE || ch == channel_layout->u.map[i].id)) + return i; + } + // fall-through + case AV_CHANNEL_ORDER_AMBISONIC: + case AV_CHANNEL_ORDER_NATIVE: + ch = av_channel_from_string(str); + if (ch == AV_CHAN_NONE) + return AVERROR(EINVAL); + return av_channel_layout_index_from_channel(channel_layout, ch); + } + + return AVERROR(EINVAL); +} + +int av_channel_layout_check(const AVChannelLayout *channel_layout) +{ + if (channel_layout->nb_channels <= 0) + return 0; + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_NATIVE: + return av_popcount64(channel_layout->u.mask) == channel_layout->nb_channels; + case AV_CHANNEL_ORDER_CUSTOM: + if (!channel_layout->u.map) + return 0; + for (int i = 0; i < channel_layout->nb_channels; i++) { + if (channel_layout->u.map[i].id == AV_CHAN_NONE) + return 0; + } + return 1; + case AV_CHANNEL_ORDER_AMBISONIC: + /* If non-diegetic channels are present, ensure they are taken into account */ + return av_popcount64(channel_layout->u.mask) < channel_layout->nb_channels; + case AV_CHANNEL_ORDER_UNSPEC: + return 1; + default: + return 0; + } +} + +int av_channel_layout_compare(const AVChannelLayout *chl, const AVChannelLayout *chl1) +{ + int i; + + /* different channel counts -> not equal */ + if (chl->nb_channels != chl1->nb_channels) + return 1; + + /* if only one is unspecified -> not equal */ + if ((chl->order == AV_CHANNEL_ORDER_UNSPEC) != + (chl1->order == AV_CHANNEL_ORDER_UNSPEC)) + return 1; + /* both are unspecified -> equal */ + else if (chl->order == AV_CHANNEL_ORDER_UNSPEC) + return 0; + + /* can compare masks directly */ + if ((chl->order == AV_CHANNEL_ORDER_NATIVE || + chl->order == AV_CHANNEL_ORDER_AMBISONIC) && + chl->order == chl1->order) + return chl->u.mask != chl1->u.mask; + + /* compare channel by channel */ + for (i = 0; i < chl->nb_channels; i++) + if (av_channel_layout_channel_from_index(chl, i) != + av_channel_layout_channel_from_index(chl1, i)) + return 1; + return 0; +} + +void av_channel_layout_default(AVChannelLayout *ch_layout, int nb_channels) +{ + int i; + for (i = 0; i < FF_ARRAY_ELEMS(channel_layout_map); i++) + if (nb_channels == channel_layout_map[i].layout.nb_channels) { + *ch_layout = channel_layout_map[i].layout; + return; + } + + ch_layout->order = AV_CHANNEL_ORDER_UNSPEC; + ch_layout->nb_channels = nb_channels; +} + +const AVChannelLayout *av_channel_layout_standard(void **opaque) +{ + uintptr_t i = (uintptr_t)*opaque; + const AVChannelLayout *ch_layout = NULL; + + if (i < FF_ARRAY_ELEMS(channel_layout_map)) { + ch_layout = &channel_layout_map[i].layout; + *opaque = (void*)(i + 1); + } + + return ch_layout; +} + +uint64_t av_channel_layout_subset(const AVChannelLayout *channel_layout, + uint64_t mask) +{ + uint64_t ret = 0; + int i; + + switch (channel_layout->order) { + case AV_CHANNEL_ORDER_NATIVE: + case AV_CHANNEL_ORDER_AMBISONIC: + return channel_layout->u.mask & mask; + case AV_CHANNEL_ORDER_CUSTOM: + for (i = 0; i < 64; i++) + if (mask & (1ULL << i) && av_channel_layout_index_from_channel(channel_layout, i) >= 0) + ret |= (1ULL << i); + break; + } + + return ret; +} diff --git a/media/ffvpx/libavutil/channel_layout.h b/media/ffvpx/libavutil/channel_layout.h new file mode 100644 index 0000000000..8dc1a91401 --- /dev/null +++ b/media/ffvpx/libavutil/channel_layout.h @@ -0,0 +1,811 @@ +/* + * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * Copyright (c) 2008 Peter Ross + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_CHANNEL_LAYOUT_H +#define AVUTIL_CHANNEL_LAYOUT_H + +#include <stdint.h> +#include <stdlib.h> + +#include "version.h" +#include "attributes.h" + +/** + * @file + * @ingroup lavu_audio_channels + * Public libavutil channel layout APIs header. + */ + + +/** + * @defgroup lavu_audio_channels Audio channels + * @ingroup lavu_audio + * + * Audio channel layout utility functions + * + * @{ + */ + +enum AVChannel { + ///< Invalid channel index + AV_CHAN_NONE = -1, + AV_CHAN_FRONT_LEFT, + AV_CHAN_FRONT_RIGHT, + AV_CHAN_FRONT_CENTER, + AV_CHAN_LOW_FREQUENCY, + AV_CHAN_BACK_LEFT, + AV_CHAN_BACK_RIGHT, + AV_CHAN_FRONT_LEFT_OF_CENTER, + AV_CHAN_FRONT_RIGHT_OF_CENTER, + AV_CHAN_BACK_CENTER, + AV_CHAN_SIDE_LEFT, + AV_CHAN_SIDE_RIGHT, + AV_CHAN_TOP_CENTER, + AV_CHAN_TOP_FRONT_LEFT, + AV_CHAN_TOP_FRONT_CENTER, + AV_CHAN_TOP_FRONT_RIGHT, + AV_CHAN_TOP_BACK_LEFT, + AV_CHAN_TOP_BACK_CENTER, + AV_CHAN_TOP_BACK_RIGHT, + /** Stereo downmix. */ + AV_CHAN_STEREO_LEFT = 29, + /** See above. */ + AV_CHAN_STEREO_RIGHT, + AV_CHAN_WIDE_LEFT, + AV_CHAN_WIDE_RIGHT, + AV_CHAN_SURROUND_DIRECT_LEFT, + AV_CHAN_SURROUND_DIRECT_RIGHT, + AV_CHAN_LOW_FREQUENCY_2, + AV_CHAN_TOP_SIDE_LEFT, + AV_CHAN_TOP_SIDE_RIGHT, + AV_CHAN_BOTTOM_FRONT_CENTER, + AV_CHAN_BOTTOM_FRONT_LEFT, + AV_CHAN_BOTTOM_FRONT_RIGHT, + + /** Channel is empty can be safely skipped. */ + AV_CHAN_UNUSED = 0x200, + + /** Channel contains data, but its position is unknown. */ + AV_CHAN_UNKNOWN = 0x300, + + /** + * Range of channels between AV_CHAN_AMBISONIC_BASE and + * AV_CHAN_AMBISONIC_END represent Ambisonic components using the ACN system. + * + * Given a channel id `<i>` between AV_CHAN_AMBISONIC_BASE and + * AV_CHAN_AMBISONIC_END (inclusive), the ACN index of the channel `<n>` is + * `<n> = <i> - AV_CHAN_AMBISONIC_BASE`. + * + * @note these values are only used for AV_CHANNEL_ORDER_CUSTOM channel + * orderings, the AV_CHANNEL_ORDER_AMBISONIC ordering orders the channels + * implicitly by their position in the stream. + */ + AV_CHAN_AMBISONIC_BASE = 0x400, + // leave space for 1024 ids, which correspond to maximum order-32 harmonics, + // which should be enough for the foreseeable use cases + AV_CHAN_AMBISONIC_END = 0x7ff, +}; + +enum AVChannelOrder { + /** + * Only the channel count is specified, without any further information + * about the channel order. + */ + AV_CHANNEL_ORDER_UNSPEC, + /** + * The native channel order, i.e. the channels are in the same order in + * which they are defined in the AVChannel enum. This supports up to 63 + * different channels. + */ + AV_CHANNEL_ORDER_NATIVE, + /** + * The channel order does not correspond to any other predefined order and + * is stored as an explicit map. For example, this could be used to support + * layouts with 64 or more channels, or with empty/skipped (AV_CHAN_SILENCE) + * channels at arbitrary positions. + */ + AV_CHANNEL_ORDER_CUSTOM, + /** + * The audio is represented as the decomposition of the sound field into + * spherical harmonics. Each channel corresponds to a single expansion + * component. Channels are ordered according to ACN (Ambisonic Channel + * Number). + * + * The channel with the index n in the stream contains the spherical + * harmonic of degree l and order m given by + * @code{.unparsed} + * l = floor(sqrt(n)), + * m = n - l * (l + 1). + * @endcode + * + * Conversely given a spherical harmonic of degree l and order m, the + * corresponding channel index n is given by + * @code{.unparsed} + * n = l * (l + 1) + m. + * @endcode + * + * Normalization is assumed to be SN3D (Schmidt Semi-Normalization) + * as defined in AmbiX format $ 2.1. + */ + AV_CHANNEL_ORDER_AMBISONIC, +}; + + +/** + * @defgroup channel_masks Audio channel masks + * + * A channel layout is a 64-bits integer with a bit set for every channel. + * The number of bits set must be equal to the number of channels. + * The value 0 means that the channel layout is not known. + * @note this data structure is not powerful enough to handle channels + * combinations that have the same channel multiple times, such as + * dual-mono. + * + * @{ + */ +#define AV_CH_FRONT_LEFT (1ULL << AV_CHAN_FRONT_LEFT ) +#define AV_CH_FRONT_RIGHT (1ULL << AV_CHAN_FRONT_RIGHT ) +#define AV_CH_FRONT_CENTER (1ULL << AV_CHAN_FRONT_CENTER ) +#define AV_CH_LOW_FREQUENCY (1ULL << AV_CHAN_LOW_FREQUENCY ) +#define AV_CH_BACK_LEFT (1ULL << AV_CHAN_BACK_LEFT ) +#define AV_CH_BACK_RIGHT (1ULL << AV_CHAN_BACK_RIGHT ) +#define AV_CH_FRONT_LEFT_OF_CENTER (1ULL << AV_CHAN_FRONT_LEFT_OF_CENTER ) +#define AV_CH_FRONT_RIGHT_OF_CENTER (1ULL << AV_CHAN_FRONT_RIGHT_OF_CENTER) +#define AV_CH_BACK_CENTER (1ULL << AV_CHAN_BACK_CENTER ) +#define AV_CH_SIDE_LEFT (1ULL << AV_CHAN_SIDE_LEFT ) +#define AV_CH_SIDE_RIGHT (1ULL << AV_CHAN_SIDE_RIGHT ) +#define AV_CH_TOP_CENTER (1ULL << AV_CHAN_TOP_CENTER ) +#define AV_CH_TOP_FRONT_LEFT (1ULL << AV_CHAN_TOP_FRONT_LEFT ) +#define AV_CH_TOP_FRONT_CENTER (1ULL << AV_CHAN_TOP_FRONT_CENTER ) +#define AV_CH_TOP_FRONT_RIGHT (1ULL << AV_CHAN_TOP_FRONT_RIGHT ) +#define AV_CH_TOP_BACK_LEFT (1ULL << AV_CHAN_TOP_BACK_LEFT ) +#define AV_CH_TOP_BACK_CENTER (1ULL << AV_CHAN_TOP_BACK_CENTER ) +#define AV_CH_TOP_BACK_RIGHT (1ULL << AV_CHAN_TOP_BACK_RIGHT ) +#define AV_CH_STEREO_LEFT (1ULL << AV_CHAN_STEREO_LEFT ) +#define AV_CH_STEREO_RIGHT (1ULL << AV_CHAN_STEREO_RIGHT ) +#define AV_CH_WIDE_LEFT (1ULL << AV_CHAN_WIDE_LEFT ) +#define AV_CH_WIDE_RIGHT (1ULL << AV_CHAN_WIDE_RIGHT ) +#define AV_CH_SURROUND_DIRECT_LEFT (1ULL << AV_CHAN_SURROUND_DIRECT_LEFT ) +#define AV_CH_SURROUND_DIRECT_RIGHT (1ULL << AV_CHAN_SURROUND_DIRECT_RIGHT) +#define AV_CH_LOW_FREQUENCY_2 (1ULL << AV_CHAN_LOW_FREQUENCY_2 ) +#define AV_CH_TOP_SIDE_LEFT (1ULL << AV_CHAN_TOP_SIDE_LEFT ) +#define AV_CH_TOP_SIDE_RIGHT (1ULL << AV_CHAN_TOP_SIDE_RIGHT ) +#define AV_CH_BOTTOM_FRONT_CENTER (1ULL << AV_CHAN_BOTTOM_FRONT_CENTER ) +#define AV_CH_BOTTOM_FRONT_LEFT (1ULL << AV_CHAN_BOTTOM_FRONT_LEFT ) +#define AV_CH_BOTTOM_FRONT_RIGHT (1ULL << AV_CHAN_BOTTOM_FRONT_RIGHT ) + +#if FF_API_OLD_CHANNEL_LAYOUT +/** Channel mask value used for AVCodecContext.request_channel_layout + to indicate that the user requests the channel order of the decoder output + to be the native codec channel order. + @deprecated channel order is now indicated in a special field in + AVChannelLayout + */ +#define AV_CH_LAYOUT_NATIVE 0x8000000000000000ULL +#endif + +/** + * @} + * @defgroup channel_mask_c Audio channel layouts + * @{ + * */ +#define AV_CH_LAYOUT_MONO (AV_CH_FRONT_CENTER) +#define AV_CH_LAYOUT_STEREO (AV_CH_FRONT_LEFT|AV_CH_FRONT_RIGHT) +#define AV_CH_LAYOUT_2POINT1 (AV_CH_LAYOUT_STEREO|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_2_1 (AV_CH_LAYOUT_STEREO|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_SURROUND (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) +#define AV_CH_LAYOUT_3POINT1 (AV_CH_LAYOUT_SURROUND|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_4POINT0 (AV_CH_LAYOUT_SURROUND|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_4POINT1 (AV_CH_LAYOUT_4POINT0|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_2_2 (AV_CH_LAYOUT_STEREO|AV_CH_SIDE_LEFT|AV_CH_SIDE_RIGHT) +#define AV_CH_LAYOUT_QUAD (AV_CH_LAYOUT_STEREO|AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT) +#define AV_CH_LAYOUT_5POINT0 (AV_CH_LAYOUT_SURROUND|AV_CH_SIDE_LEFT|AV_CH_SIDE_RIGHT) +#define AV_CH_LAYOUT_5POINT1 (AV_CH_LAYOUT_5POINT0|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_5POINT0_BACK (AV_CH_LAYOUT_SURROUND|AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT) +#define AV_CH_LAYOUT_5POINT1_BACK (AV_CH_LAYOUT_5POINT0_BACK|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_6POINT0 (AV_CH_LAYOUT_5POINT0|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_6POINT0_FRONT (AV_CH_LAYOUT_2_2|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER) +#define AV_CH_LAYOUT_HEXAGONAL (AV_CH_LAYOUT_5POINT0_BACK|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_3POINT1POINT2 (AV_CH_LAYOUT_3POINT1|AV_CH_TOP_FRONT_LEFT|AV_CH_TOP_FRONT_RIGHT) +#define AV_CH_LAYOUT_6POINT1 (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_6POINT1_BACK (AV_CH_LAYOUT_5POINT1_BACK|AV_CH_BACK_CENTER) +#define AV_CH_LAYOUT_6POINT1_FRONT (AV_CH_LAYOUT_6POINT0_FRONT|AV_CH_LOW_FREQUENCY) +#define AV_CH_LAYOUT_7POINT0 (AV_CH_LAYOUT_5POINT0|AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT) +#define AV_CH_LAYOUT_7POINT0_FRONT (AV_CH_LAYOUT_5POINT0|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER) +#define AV_CH_LAYOUT_7POINT1 (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_LEFT|AV_CH_BACK_RIGHT) +#define AV_CH_LAYOUT_7POINT1_WIDE (AV_CH_LAYOUT_5POINT1|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER) +#define AV_CH_LAYOUT_7POINT1_WIDE_BACK (AV_CH_LAYOUT_5POINT1_BACK|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER) +#define AV_CH_LAYOUT_5POINT1POINT2_BACK (AV_CH_LAYOUT_5POINT1_BACK|AV_CH_TOP_FRONT_LEFT|AV_CH_TOP_FRONT_RIGHT) +#define AV_CH_LAYOUT_OCTAGONAL (AV_CH_LAYOUT_5POINT0|AV_CH_BACK_LEFT|AV_CH_BACK_CENTER|AV_CH_BACK_RIGHT) +#define AV_CH_LAYOUT_CUBE (AV_CH_LAYOUT_QUAD|AV_CH_TOP_FRONT_LEFT|AV_CH_TOP_FRONT_RIGHT|AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_RIGHT) +#define AV_CH_LAYOUT_5POINT1POINT4_BACK (AV_CH_LAYOUT_5POINT1POINT2_BACK|AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_RIGHT) +#define AV_CH_LAYOUT_7POINT1POINT2 (AV_CH_LAYOUT_7POINT1|AV_CH_TOP_FRONT_LEFT|AV_CH_TOP_FRONT_RIGHT) +#define AV_CH_LAYOUT_7POINT1POINT4_BACK (AV_CH_LAYOUT_7POINT1POINT2|AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_RIGHT) +#define AV_CH_LAYOUT_7POINT2POINT3 (AV_CH_LAYOUT_7POINT1POINT2|AV_CH_TOP_BACK_CENTER|AV_CH_LOW_FREQUENCY_2) +#define AV_CH_LAYOUT_9POINT1POINT4_BACK (AV_CH_LAYOUT_7POINT1POINT4_BACK|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER) +#define AV_CH_LAYOUT_HEXADECAGONAL (AV_CH_LAYOUT_OCTAGONAL|AV_CH_WIDE_LEFT|AV_CH_WIDE_RIGHT|AV_CH_TOP_BACK_LEFT|AV_CH_TOP_BACK_RIGHT|AV_CH_TOP_BACK_CENTER|AV_CH_TOP_FRONT_CENTER|AV_CH_TOP_FRONT_LEFT|AV_CH_TOP_FRONT_RIGHT) +#define AV_CH_LAYOUT_STEREO_DOWNMIX (AV_CH_STEREO_LEFT|AV_CH_STEREO_RIGHT) +#define AV_CH_LAYOUT_22POINT2 (AV_CH_LAYOUT_7POINT1POINT4_BACK|AV_CH_FRONT_LEFT_OF_CENTER|AV_CH_FRONT_RIGHT_OF_CENTER|AV_CH_BACK_CENTER|AV_CH_LOW_FREQUENCY_2|AV_CH_TOP_FRONT_CENTER|AV_CH_TOP_CENTER|AV_CH_TOP_SIDE_LEFT|AV_CH_TOP_SIDE_RIGHT|AV_CH_TOP_BACK_CENTER|AV_CH_BOTTOM_FRONT_CENTER|AV_CH_BOTTOM_FRONT_LEFT|AV_CH_BOTTOM_FRONT_RIGHT) + +#define AV_CH_LAYOUT_7POINT1_TOP_BACK AV_CH_LAYOUT_5POINT1POINT2_BACK + +enum AVMatrixEncoding { + AV_MATRIX_ENCODING_NONE, + AV_MATRIX_ENCODING_DOLBY, + AV_MATRIX_ENCODING_DPLII, + AV_MATRIX_ENCODING_DPLIIX, + AV_MATRIX_ENCODING_DPLIIZ, + AV_MATRIX_ENCODING_DOLBYEX, + AV_MATRIX_ENCODING_DOLBYHEADPHONE, + AV_MATRIX_ENCODING_NB +}; + +/** + * @} + */ + +/** + * An AVChannelCustom defines a single channel within a custom order layout + * + * Unlike most structures in FFmpeg, sizeof(AVChannelCustom) is a part of the + * public ABI. + * + * No new fields may be added to it without a major version bump. + */ +typedef struct AVChannelCustom { + enum AVChannel id; + char name[16]; + void *opaque; +} AVChannelCustom; + +/** + * An AVChannelLayout holds information about the channel layout of audio data. + * + * A channel layout here is defined as a set of channels ordered in a specific + * way (unless the channel order is AV_CHANNEL_ORDER_UNSPEC, in which case an + * AVChannelLayout carries only the channel count). + * All orders may be treated as if they were AV_CHANNEL_ORDER_UNSPEC by + * ignoring everything but the channel count, as long as av_channel_layout_check() + * considers they are valid. + * + * Unlike most structures in FFmpeg, sizeof(AVChannelLayout) is a part of the + * public ABI and may be used by the caller. E.g. it may be allocated on stack + * or embedded in caller-defined structs. + * + * AVChannelLayout can be initialized as follows: + * - default initialization with {0}, followed by setting all used fields + * correctly; + * - by assigning one of the predefined AV_CHANNEL_LAYOUT_* initializers; + * - with a constructor function, such as av_channel_layout_default(), + * av_channel_layout_from_mask() or av_channel_layout_from_string(). + * + * The channel layout must be unitialized with av_channel_layout_uninit() + * + * Copying an AVChannelLayout via assigning is forbidden, + * av_channel_layout_copy() must be used instead (and its return value should + * be checked) + * + * No new fields may be added to it without a major version bump, except for + * new elements of the union fitting in sizeof(uint64_t). + */ +typedef struct AVChannelLayout { + /** + * Channel order used in this layout. + * This is a mandatory field. + */ + enum AVChannelOrder order; + + /** + * Number of channels in this layout. Mandatory field. + */ + int nb_channels; + + /** + * Details about which channels are present in this layout. + * For AV_CHANNEL_ORDER_UNSPEC, this field is undefined and must not be + * used. + */ + union { + /** + * This member must be used for AV_CHANNEL_ORDER_NATIVE, and may be used + * for AV_CHANNEL_ORDER_AMBISONIC to signal non-diegetic channels. + * It is a bitmask, where the position of each set bit means that the + * AVChannel with the corresponding value is present. + * + * I.e. when (mask & (1 << AV_CHAN_FOO)) is non-zero, then AV_CHAN_FOO + * is present in the layout. Otherwise it is not present. + * + * @note when a channel layout using a bitmask is constructed or + * modified manually (i.e. not using any of the av_channel_layout_* + * functions), the code doing it must ensure that the number of set bits + * is equal to nb_channels. + */ + uint64_t mask; + /** + * This member must be used when the channel order is + * AV_CHANNEL_ORDER_CUSTOM. It is a nb_channels-sized array, with each + * element signalling the presence of the AVChannel with the + * corresponding value in map[i].id. + * + * I.e. when map[i].id is equal to AV_CHAN_FOO, then AV_CH_FOO is the + * i-th channel in the audio data. + * + * When map[i].id is in the range between AV_CHAN_AMBISONIC_BASE and + * AV_CHAN_AMBISONIC_END (inclusive), the channel contains an ambisonic + * component with ACN index (as defined above) + * n = map[i].id - AV_CHAN_AMBISONIC_BASE. + * + * map[i].name may be filled with a 0-terminated string, in which case + * it will be used for the purpose of identifying the channel with the + * convenience functions below. Otherise it must be zeroed. + */ + AVChannelCustom *map; + } u; + + /** + * For some private data of the user. + */ + void *opaque; +} AVChannelLayout; + +/** + * Macro to define native channel layouts + * + * @note This doesn't use designated initializers for compatibility with C++ 17 and older. + */ +#define AV_CHANNEL_LAYOUT_MASK(nb, m) \ + { /* .order */ AV_CHANNEL_ORDER_NATIVE, \ + /* .nb_channels */ (nb), \ + /* .u.mask */ { m }, \ + /* .opaque */ NULL } + +/** + * @name Common pre-defined channel layouts + * @{ + */ +#define AV_CHANNEL_LAYOUT_MONO AV_CHANNEL_LAYOUT_MASK(1, AV_CH_LAYOUT_MONO) +#define AV_CHANNEL_LAYOUT_STEREO AV_CHANNEL_LAYOUT_MASK(2, AV_CH_LAYOUT_STEREO) +#define AV_CHANNEL_LAYOUT_2POINT1 AV_CHANNEL_LAYOUT_MASK(3, AV_CH_LAYOUT_2POINT1) +#define AV_CHANNEL_LAYOUT_2_1 AV_CHANNEL_LAYOUT_MASK(3, AV_CH_LAYOUT_2_1) +#define AV_CHANNEL_LAYOUT_SURROUND AV_CHANNEL_LAYOUT_MASK(3, AV_CH_LAYOUT_SURROUND) +#define AV_CHANNEL_LAYOUT_3POINT1 AV_CHANNEL_LAYOUT_MASK(4, AV_CH_LAYOUT_3POINT1) +#define AV_CHANNEL_LAYOUT_4POINT0 AV_CHANNEL_LAYOUT_MASK(4, AV_CH_LAYOUT_4POINT0) +#define AV_CHANNEL_LAYOUT_4POINT1 AV_CHANNEL_LAYOUT_MASK(5, AV_CH_LAYOUT_4POINT1) +#define AV_CHANNEL_LAYOUT_2_2 AV_CHANNEL_LAYOUT_MASK(4, AV_CH_LAYOUT_2_2) +#define AV_CHANNEL_LAYOUT_QUAD AV_CHANNEL_LAYOUT_MASK(4, AV_CH_LAYOUT_QUAD) +#define AV_CHANNEL_LAYOUT_5POINT0 AV_CHANNEL_LAYOUT_MASK(5, AV_CH_LAYOUT_5POINT0) +#define AV_CHANNEL_LAYOUT_5POINT1 AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_5POINT1) +#define AV_CHANNEL_LAYOUT_5POINT0_BACK AV_CHANNEL_LAYOUT_MASK(5, AV_CH_LAYOUT_5POINT0_BACK) +#define AV_CHANNEL_LAYOUT_5POINT1_BACK AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_5POINT1_BACK) +#define AV_CHANNEL_LAYOUT_6POINT0 AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_6POINT0) +#define AV_CHANNEL_LAYOUT_6POINT0_FRONT AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_6POINT0_FRONT) +#define AV_CHANNEL_LAYOUT_3POINT1POINT2 AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_3POINT1POINT2) +#define AV_CHANNEL_LAYOUT_HEXAGONAL AV_CHANNEL_LAYOUT_MASK(6, AV_CH_LAYOUT_HEXAGONAL) +#define AV_CHANNEL_LAYOUT_6POINT1 AV_CHANNEL_LAYOUT_MASK(7, AV_CH_LAYOUT_6POINT1) +#define AV_CHANNEL_LAYOUT_6POINT1_BACK AV_CHANNEL_LAYOUT_MASK(7, AV_CH_LAYOUT_6POINT1_BACK) +#define AV_CHANNEL_LAYOUT_6POINT1_FRONT AV_CHANNEL_LAYOUT_MASK(7, AV_CH_LAYOUT_6POINT1_FRONT) +#define AV_CHANNEL_LAYOUT_7POINT0 AV_CHANNEL_LAYOUT_MASK(7, AV_CH_LAYOUT_7POINT0) +#define AV_CHANNEL_LAYOUT_7POINT0_FRONT AV_CHANNEL_LAYOUT_MASK(7, AV_CH_LAYOUT_7POINT0_FRONT) +#define AV_CHANNEL_LAYOUT_7POINT1 AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_7POINT1) +#define AV_CHANNEL_LAYOUT_7POINT1_WIDE AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_7POINT1_WIDE) +#define AV_CHANNEL_LAYOUT_7POINT1_WIDE_BACK AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_7POINT1_WIDE_BACK) +#define AV_CHANNEL_LAYOUT_5POINT1POINT2_BACK AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_5POINT1POINT2_BACK) +#define AV_CHANNEL_LAYOUT_OCTAGONAL AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_OCTAGONAL) +#define AV_CHANNEL_LAYOUT_CUBE AV_CHANNEL_LAYOUT_MASK(8, AV_CH_LAYOUT_CUBE) +#define AV_CHANNEL_LAYOUT_5POINT1POINT4_BACK AV_CHANNEL_LAYOUT_MASK(10, AV_CH_LAYOUT_5POINT1POINT4_BACK) +#define AV_CHANNEL_LAYOUT_7POINT1POINT2 AV_CHANNEL_LAYOUT_MASK(10, AV_CH_LAYOUT_7POINT1POINT2) +#define AV_CHANNEL_LAYOUT_7POINT1POINT4_BACK AV_CHANNEL_LAYOUT_MASK(12, AV_CH_LAYOUT_7POINT1POINT4_BACK) +#define AV_CHANNEL_LAYOUT_7POINT2POINT3 AV_CHANNEL_LAYOUT_MASK(12, AV_CH_LAYOUT_7POINT2POINT3) +#define AV_CHANNEL_LAYOUT_9POINT1POINT4_BACK AV_CHANNEL_LAYOUT_MASK(14, AV_CH_LAYOUT_9POINT1POINT4_BACK) +#define AV_CHANNEL_LAYOUT_HEXADECAGONAL AV_CHANNEL_LAYOUT_MASK(16, AV_CH_LAYOUT_HEXADECAGONAL) +#define AV_CHANNEL_LAYOUT_STEREO_DOWNMIX AV_CHANNEL_LAYOUT_MASK(2, AV_CH_LAYOUT_STEREO_DOWNMIX) +#define AV_CHANNEL_LAYOUT_22POINT2 AV_CHANNEL_LAYOUT_MASK(24, AV_CH_LAYOUT_22POINT2) + +#define AV_CHANNEL_LAYOUT_7POINT1_TOP_BACK AV_CHANNEL_LAYOUT_5POINT1POINT2_BACK + +#define AV_CHANNEL_LAYOUT_AMBISONIC_FIRST_ORDER \ + { /* .order */ AV_CHANNEL_ORDER_AMBISONIC, \ + /* .nb_channels */ 4, \ + /* .u.mask */ { 0 }, \ + /* .opaque */ NULL } +/** @} */ + +struct AVBPrint; + +#if FF_API_OLD_CHANNEL_LAYOUT +/** + * @name Deprecated Functions + * @{ + */ + +/** + * Return a channel layout id that matches name, or 0 if no match is found. + * + * name can be one or several of the following notations, + * separated by '+' or '|': + * - the name of an usual channel layout (mono, stereo, 4.0, quad, 5.0, + * 5.0(side), 5.1, 5.1(side), 7.1, 7.1(wide), downmix); + * - the name of a single channel (FL, FR, FC, LFE, BL, BR, FLC, FRC, BC, + * SL, SR, TC, TFL, TFC, TFR, TBL, TBC, TBR, DL, DR); + * - a number of channels, in decimal, followed by 'c', yielding + * the default channel layout for that number of channels (@see + * av_get_default_channel_layout); + * - a channel layout mask, in hexadecimal starting with "0x" (see the + * AV_CH_* macros). + * + * Example: "stereo+FC" = "2c+FC" = "2c+1c" = "0x7" + * + * @deprecated use av_channel_layout_from_string() + */ +attribute_deprecated +uint64_t av_get_channel_layout(const char *name); + +/** + * Return a channel layout and the number of channels based on the specified name. + * + * This function is similar to (@see av_get_channel_layout), but can also parse + * unknown channel layout specifications. + * + * @param[in] name channel layout specification string + * @param[out] channel_layout parsed channel layout (0 if unknown) + * @param[out] nb_channels number of channels + * + * @return 0 on success, AVERROR(EINVAL) if the parsing fails. + * @deprecated use av_channel_layout_from_string() + */ +attribute_deprecated +int av_get_extended_channel_layout(const char *name, uint64_t* channel_layout, int* nb_channels); + +/** + * Return a description of a channel layout. + * If nb_channels is <= 0, it is guessed from the channel_layout. + * + * @param buf put here the string containing the channel layout + * @param buf_size size in bytes of the buffer + * @param nb_channels number of channels + * @param channel_layout channel layout bitset + * @deprecated use av_channel_layout_describe() + */ +attribute_deprecated +void av_get_channel_layout_string(char *buf, int buf_size, int nb_channels, uint64_t channel_layout); + +/** + * Append a description of a channel layout to a bprint buffer. + * @deprecated use av_channel_layout_describe() + */ +attribute_deprecated +void av_bprint_channel_layout(struct AVBPrint *bp, int nb_channels, uint64_t channel_layout); + +/** + * Return the number of channels in the channel layout. + * @deprecated use AVChannelLayout.nb_channels + */ +attribute_deprecated +int av_get_channel_layout_nb_channels(uint64_t channel_layout); + +/** + * Return default channel layout for a given number of channels. + * + * @deprecated use av_channel_layout_default() + */ +attribute_deprecated +int64_t av_get_default_channel_layout(int nb_channels); + +/** + * Get the index of a channel in channel_layout. + * + * @param channel_layout channel layout bitset + * @param channel a channel layout describing exactly one channel which must be + * present in channel_layout. + * + * @return index of channel in channel_layout on success, a negative AVERROR + * on error. + * + * @deprecated use av_channel_layout_index_from_channel() + */ +attribute_deprecated +int av_get_channel_layout_channel_index(uint64_t channel_layout, + uint64_t channel); + +/** + * Get the channel with the given index in channel_layout. + * @deprecated use av_channel_layout_channel_from_index() + */ +attribute_deprecated +uint64_t av_channel_layout_extract_channel(uint64_t channel_layout, int index); + +/** + * Get the name of a given channel. + * + * @return channel name on success, NULL on error. + * + * @deprecated use av_channel_name() + */ +attribute_deprecated +const char *av_get_channel_name(uint64_t channel); + +/** + * Get the description of a given channel. + * + * @param channel a channel layout with a single channel + * @return channel description on success, NULL on error + * @deprecated use av_channel_description() + */ +attribute_deprecated +const char *av_get_channel_description(uint64_t channel); + +/** + * Get the value and name of a standard channel layout. + * + * @param[in] index index in an internal list, starting at 0 + * @param[out] layout channel layout mask + * @param[out] name name of the layout + * @return 0 if the layout exists, + * <0 if index is beyond the limits + * @deprecated use av_channel_layout_standard() + */ +attribute_deprecated +int av_get_standard_channel_layout(unsigned index, uint64_t *layout, + const char **name); +/** + * @} + */ +#endif + +/** + * Get a human readable string in an abbreviated form describing a given channel. + * This is the inverse function of @ref av_channel_from_string(). + * + * @param buf pre-allocated buffer where to put the generated string + * @param buf_size size in bytes of the buffer. + * @param channel the AVChannel whose name to get + * @return amount of bytes needed to hold the output string, or a negative AVERROR + * on failure. If the returned value is bigger than buf_size, then the + * string was truncated. + */ +int av_channel_name(char *buf, size_t buf_size, enum AVChannel channel); + +/** + * bprint variant of av_channel_name(). + * + * @note the string will be appended to the bprint buffer. + */ +void av_channel_name_bprint(struct AVBPrint *bp, enum AVChannel channel_id); + +/** + * Get a human readable string describing a given channel. + * + * @param buf pre-allocated buffer where to put the generated string + * @param buf_size size in bytes of the buffer. + * @param channel the AVChannel whose description to get + * @return amount of bytes needed to hold the output string, or a negative AVERROR + * on failure. If the returned value is bigger than buf_size, then the + * string was truncated. + */ +int av_channel_description(char *buf, size_t buf_size, enum AVChannel channel); + +/** + * bprint variant of av_channel_description(). + * + * @note the string will be appended to the bprint buffer. + */ +void av_channel_description_bprint(struct AVBPrint *bp, enum AVChannel channel_id); + +/** + * This is the inverse function of @ref av_channel_name(). + * + * @return the channel with the given name + * AV_CHAN_NONE when name does not identify a known channel + */ +enum AVChannel av_channel_from_string(const char *name); + +/** + * Initialize a native channel layout from a bitmask indicating which channels + * are present. + * + * @param channel_layout the layout structure to be initialized + * @param mask bitmask describing the channel layout + * + * @return 0 on success + * AVERROR(EINVAL) for invalid mask values + */ +int av_channel_layout_from_mask(AVChannelLayout *channel_layout, uint64_t mask); + +/** + * Initialize a channel layout from a given string description. + * The input string can be represented by: + * - the formal channel layout name (returned by av_channel_layout_describe()) + * - single or multiple channel names (returned by av_channel_name(), eg. "FL", + * or concatenated with "+", each optionally containing a custom name after + * a "@", eg. "FL@Left+FR@Right+LFE") + * - a decimal or hexadecimal value of a native channel layout (eg. "4" or "0x4") + * - the number of channels with default layout (eg. "4c") + * - the number of unordered channels (eg. "4C" or "4 channels") + * - the ambisonic order followed by optional non-diegetic channels (eg. + * "ambisonic 2+stereo") + * + * @param channel_layout input channel layout + * @param str string describing the channel layout + * @return 0 channel layout was detected, AVERROR_INVALIDATATA otherwise + */ +int av_channel_layout_from_string(AVChannelLayout *channel_layout, + const char *str); + +/** + * Get the default channel layout for a given number of channels. + * + * @param ch_layout the layout structure to be initialized + * @param nb_channels number of channels + */ +void av_channel_layout_default(AVChannelLayout *ch_layout, int nb_channels); + +/** + * Iterate over all standard channel layouts. + * + * @param opaque a pointer where libavutil will store the iteration state. Must + * point to NULL to start the iteration. + * + * @return the standard channel layout or NULL when the iteration is + * finished + */ +const AVChannelLayout *av_channel_layout_standard(void **opaque); + +/** + * Free any allocated data in the channel layout and reset the channel + * count to 0. + * + * @param channel_layout the layout structure to be uninitialized + */ +void av_channel_layout_uninit(AVChannelLayout *channel_layout); + +/** + * Make a copy of a channel layout. This differs from just assigning src to dst + * in that it allocates and copies the map for AV_CHANNEL_ORDER_CUSTOM. + * + * @note the destination channel_layout will be always uninitialized before copy. + * + * @param dst destination channel layout + * @param src source channel layout + * @return 0 on success, a negative AVERROR on error. + */ +int av_channel_layout_copy(AVChannelLayout *dst, const AVChannelLayout *src); + +/** + * Get a human-readable string describing the channel layout properties. + * The string will be in the same format that is accepted by + * @ref av_channel_layout_from_string(), allowing to rebuild the same + * channel layout, except for opaque pointers. + * + * @param channel_layout channel layout to be described + * @param buf pre-allocated buffer where to put the generated string + * @param buf_size size in bytes of the buffer. + * @return amount of bytes needed to hold the output string, or a negative AVERROR + * on failure. If the returned value is bigger than buf_size, then the + * string was truncated. + */ +int av_channel_layout_describe(const AVChannelLayout *channel_layout, + char *buf, size_t buf_size); + +/** + * bprint variant of av_channel_layout_describe(). + * + * @note the string will be appended to the bprint buffer. + * @return 0 on success, or a negative AVERROR value on failure. + */ +int av_channel_layout_describe_bprint(const AVChannelLayout *channel_layout, + struct AVBPrint *bp); + +/** + * Get the channel with the given index in a channel layout. + * + * @param channel_layout input channel layout + * @param idx index of the channel + * @return channel with the index idx in channel_layout on success or + * AV_CHAN_NONE on failure (if idx is not valid or the channel order is + * unspecified) + */ +enum AVChannel +av_channel_layout_channel_from_index(const AVChannelLayout *channel_layout, unsigned int idx); + +/** + * Get the index of a given channel in a channel layout. In case multiple + * channels are found, only the first match will be returned. + * + * @param channel_layout input channel layout + * @param channel the channel whose index to obtain + * @return index of channel in channel_layout on success or a negative number if + * channel is not present in channel_layout. + */ +int av_channel_layout_index_from_channel(const AVChannelLayout *channel_layout, + enum AVChannel channel); + +/** + * Get the index in a channel layout of a channel described by the given string. + * In case multiple channels are found, only the first match will be returned. + * + * This function accepts channel names in the same format as + * @ref av_channel_from_string(). + * + * @param channel_layout input channel layout + * @param name string describing the channel whose index to obtain + * @return a channel index described by the given string, or a negative AVERROR + * value. + */ +int av_channel_layout_index_from_string(const AVChannelLayout *channel_layout, + const char *name); + +/** + * Get a channel described by the given string. + * + * This function accepts channel names in the same format as + * @ref av_channel_from_string(). + * + * @param channel_layout input channel layout + * @param name string describing the channel to obtain + * @return a channel described by the given string in channel_layout on success + * or AV_CHAN_NONE on failure (if the string is not valid or the channel + * order is unspecified) + */ +enum AVChannel +av_channel_layout_channel_from_string(const AVChannelLayout *channel_layout, + const char *name); + +/** + * Find out what channels from a given set are present in a channel layout, + * without regard for their positions. + * + * @param channel_layout input channel layout + * @param mask a combination of AV_CH_* representing a set of channels + * @return a bitfield representing all the channels from mask that are present + * in channel_layout + */ +uint64_t av_channel_layout_subset(const AVChannelLayout *channel_layout, + uint64_t mask); + +/** + * Check whether a channel layout is valid, i.e. can possibly describe audio + * data. + * + * @param channel_layout input channel layout + * @return 1 if channel_layout is valid, 0 otherwise. + */ +int av_channel_layout_check(const AVChannelLayout *channel_layout); + +/** + * Check whether two channel layouts are semantically the same, i.e. the same + * channels are present on the same positions in both. + * + * If one of the channel layouts is AV_CHANNEL_ORDER_UNSPEC, while the other is + * not, they are considered to be unequal. If both are AV_CHANNEL_ORDER_UNSPEC, + * they are considered equal iff the channel counts are the same in both. + * + * @param chl input channel layout + * @param chl1 input channel layout + * @return 0 if chl and chl1 are equal, 1 if they are not equal. A negative + * AVERROR code if one or both are invalid. + */ +int av_channel_layout_compare(const AVChannelLayout *chl, const AVChannelLayout *chl1); + +/** + * @} + */ + +#endif /* AVUTIL_CHANNEL_LAYOUT_H */ diff --git a/media/ffvpx/libavutil/color_utils.c b/media/ffvpx/libavutil/color_utils.c new file mode 100644 index 0000000000..5e221fb798 --- /dev/null +++ b/media/ffvpx/libavutil/color_utils.c @@ -0,0 +1,234 @@ +/* + * Copyright (c) 2015 Kevin Wheatley <kevin.j.wheatley@gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stddef.h> +#include <math.h> + +#include "libavutil/color_utils.h" +#include "libavutil/pixfmt.h" + +double avpriv_get_gamma_from_trc(enum AVColorTransferCharacteristic trc) +{ + double gamma; + switch (trc) { + case AVCOL_TRC_BT709: + case AVCOL_TRC_SMPTE170M: + case AVCOL_TRC_SMPTE240M: + case AVCOL_TRC_BT1361_ECG: + case AVCOL_TRC_BT2020_10: + case AVCOL_TRC_BT2020_12: + /* these share a segmented TRC, but gamma 1.961 is a close + approximation, and also more correct for decoding content */ + gamma = 1.961; + break; + case AVCOL_TRC_GAMMA22: + case AVCOL_TRC_IEC61966_2_1: + gamma = 2.2; + break; + case AVCOL_TRC_GAMMA28: + gamma = 2.8; + break; + case AVCOL_TRC_LINEAR: + gamma = 1.0; + break; + default: + gamma = 0.0; // Unknown value representation + } + return gamma; +} + +#define BT709_alpha 1.099296826809442 +#define BT709_beta 0.018053968510807 + +static double avpriv_trc_bt709(double Lc) +{ + const double a = BT709_alpha; + const double b = BT709_beta; + + return (0.0 > Lc) ? 0.0 + : ( b > Lc) ? 4.500 * Lc + : a * pow(Lc, 0.45) - (a - 1.0); +} + +static double avpriv_trc_gamma22(double Lc) +{ + return (0.0 > Lc) ? 0.0 : pow(Lc, 1.0/ 2.2); +} + +static double avpriv_trc_gamma28(double Lc) +{ + return (0.0 > Lc) ? 0.0 : pow(Lc, 1.0/ 2.8); +} + +static double avpriv_trc_smpte240M(double Lc) +{ + const double a = 1.1115; + const double b = 0.0228; + + return (0.0 > Lc) ? 0.0 + : ( b > Lc) ? 4.000 * Lc + : a * pow(Lc, 0.45) - (a - 1.0); +} + +static double avpriv_trc_linear(double Lc) +{ + return Lc; +} + +static double avpriv_trc_log(double Lc) +{ + return (0.01 > Lc) ? 0.0 : 1.0 + log10(Lc) / 2.0; +} + +static double avpriv_trc_log_sqrt(double Lc) +{ + // sqrt(10) / 1000 + return (0.00316227766 > Lc) ? 0.0 : 1.0 + log10(Lc) / 2.5; +} + +static double avpriv_trc_iec61966_2_4(double Lc) +{ + const double a = BT709_alpha; + const double b = BT709_beta; + + return (-b >= Lc) ? -a * pow(-Lc, 0.45) + (a - 1.0) + : ( b > Lc) ? 4.500 * Lc + : a * pow( Lc, 0.45) - (a - 1.0); +} + +static double avpriv_trc_bt1361(double Lc) +{ + const double a = BT709_alpha; + const double b = BT709_beta; + + return (-0.0045 >= Lc) ? -(a * pow(-4.0 * Lc, 0.45) + (a - 1.0)) / 4.0 + : ( b > Lc) ? 4.500 * Lc + : a * pow( Lc, 0.45) - (a - 1.0); +} + +static double avpriv_trc_iec61966_2_1(double Lc) +{ + const double a = 1.055; + const double b = 0.0031308; + + return (0.0 > Lc) ? 0.0 + : ( b > Lc) ? 12.92 * Lc + : a * pow(Lc, 1.0 / 2.4) - (a - 1.0); +} + +static double avpriv_trc_smpte_st2084(double Lc) +{ + const double c1 = 3424.0 / 4096.0; // c3-c2 + 1 + const double c2 = 32.0 * 2413.0 / 4096.0; + const double c3 = 32.0 * 2392.0 / 4096.0; + const double m = 128.0 * 2523.0 / 4096.0; + const double n = 0.25 * 2610.0 / 4096.0; + const double L = Lc / 10000.0; + const double Ln = pow(L, n); + + return (0.0 > Lc) ? 0.0 + : pow((c1 + c2 * Ln) / (1.0 + c3 * Ln), m); + +} + +static double avpriv_trc_smpte_st428_1(double Lc) +{ + return (0.0 > Lc) ? 0.0 + : pow(48.0 * Lc / 52.37, 1.0 / 2.6); +} + + +static double avpriv_trc_arib_std_b67(double Lc) { + // The function uses the definition from HEVC, which assumes that the peak + // white is input level = 1. (this is equivalent to scaling E = Lc * 12 and + // using the definition from the ARIB STD-B67 spec) + const double a = 0.17883277; + const double b = 0.28466892; + const double c = 0.55991073; + return (0.0 > Lc) ? 0.0 : + (Lc <= 1.0 / 12.0 ? sqrt(3.0 * Lc) : a * log(12.0 * Lc - b) + c); +} + +avpriv_trc_function avpriv_get_trc_function_from_trc(enum AVColorTransferCharacteristic trc) +{ + avpriv_trc_function func = NULL; + switch (trc) { + case AVCOL_TRC_BT709: + case AVCOL_TRC_SMPTE170M: + case AVCOL_TRC_BT2020_10: + case AVCOL_TRC_BT2020_12: + func = avpriv_trc_bt709; + break; + + case AVCOL_TRC_GAMMA22: + func = avpriv_trc_gamma22; + break; + case AVCOL_TRC_GAMMA28: + func = avpriv_trc_gamma28; + break; + + case AVCOL_TRC_SMPTE240M: + func = avpriv_trc_smpte240M; + break; + + case AVCOL_TRC_LINEAR: + func = avpriv_trc_linear; + break; + + case AVCOL_TRC_LOG: + func = avpriv_trc_log; + break; + + case AVCOL_TRC_LOG_SQRT: + func = avpriv_trc_log_sqrt; + break; + + case AVCOL_TRC_IEC61966_2_4: + func = avpriv_trc_iec61966_2_4; + break; + + case AVCOL_TRC_BT1361_ECG: + func = avpriv_trc_bt1361; + break; + + case AVCOL_TRC_IEC61966_2_1: + func = avpriv_trc_iec61966_2_1; + break; + + case AVCOL_TRC_SMPTEST2084: + func = avpriv_trc_smpte_st2084; + break; + + case AVCOL_TRC_SMPTEST428_1: + func = avpriv_trc_smpte_st428_1; + break; + + case AVCOL_TRC_ARIB_STD_B67: + func = avpriv_trc_arib_std_b67; + break; + + case AVCOL_TRC_RESERVED0: + case AVCOL_TRC_UNSPECIFIED: + case AVCOL_TRC_RESERVED: + default: + break; + } + return func; +} diff --git a/media/ffvpx/libavutil/color_utils.h b/media/ffvpx/libavutil/color_utils.h new file mode 100644 index 0000000000..9529006452 --- /dev/null +++ b/media/ffvpx/libavutil/color_utils.h @@ -0,0 +1,56 @@ +/* + * Copyright (c) 2015 Kevin Wheatley <kevin.j.wheatley@gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_COLOR_UTILS_H +#define AVUTIL_COLOR_UTILS_H + + +#include "libavutil/pixfmt.h" + +/** + * Determine a suitable 'gamma' value to match the supplied + * AVColorTransferCharacteristic. + * + * See Apple Technical Note TN2257 (https://developer.apple.com/library/mac/technotes/tn2257/_index.html) + * + * @return Will return an approximation to the simple gamma function matching + * the supplied Transfer Characteristic, Will return 0.0 for any + * we cannot reasonably match against. + */ +double avpriv_get_gamma_from_trc(enum AVColorTransferCharacteristic trc); + + +typedef double (*avpriv_trc_function)(double); + +/** + * Determine the function needed to apply the given + * AVColorTransferCharacteristic to linear input. + * + * The function returned should expect a nominal domain and range of [0.0-1.0] + * values outside of this range maybe valid depending on the chosen + * characteristic function. + * + * @return Will return pointer to the function matching the + * supplied Transfer Characteristic. If unspecified will + * return NULL: + */ +avpriv_trc_function avpriv_get_trc_function_from_trc(enum AVColorTransferCharacteristic trc); + +#endif diff --git a/media/ffvpx/libavutil/colorspace.h b/media/ffvpx/libavutil/colorspace.h new file mode 100644 index 0000000000..ef6f6107d6 --- /dev/null +++ b/media/ffvpx/libavutil/colorspace.h @@ -0,0 +1,150 @@ +/* + * Colorspace conversion defines + * Copyright (c) 2001, 2002, 2003 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Various defines for YUV<->RGB conversion + */ + +#ifndef AVUTIL_COLORSPACE_H +#define AVUTIL_COLORSPACE_H + +#define SCALEBITS 10 +#define ONE_HALF (1 << (SCALEBITS - 1)) +#define FIX(x) ((int) ((x) * (1<<SCALEBITS) + 0.5)) + +#define YUV_TO_RGB1_CCIR(cb1, cr1)\ +{\ + cb = (cb1) - 128;\ + cr = (cr1) - 128;\ + r_add = FIX(1.40200*255.0/224.0) * cr + ONE_HALF;\ + g_add = - FIX(0.34414*255.0/224.0) * cb - FIX(0.71414*255.0/224.0) * cr + \ + ONE_HALF;\ + b_add = FIX(1.77200*255.0/224.0) * cb + ONE_HALF;\ +} + +#define YUV_TO_RGB1_CCIR_BT709(cb1, cr1) \ + { \ + cb = (cb1) - 128; \ + cr = (cr1) - 128; \ + r_add = ONE_HALF + FIX(1.5747 * 255.0 / 224.0) * cr; \ + g_add = ONE_HALF - FIX(0.1873 * 255.0 / 224.0) * cb - \ + FIX(0.4682 * 255.0 / 224.0) * cr; \ + b_add = ONE_HALF + FIX(1.8556 * 255.0 / 224.0) * cb; \ + } + +// To be used for the BT709 variant as well +#define YUV_TO_RGB2_CCIR(r, g, b, y1)\ +{\ + y = ((y1) - 16) * FIX(255.0/219.0);\ + r = cm[(y + r_add) >> SCALEBITS];\ + g = cm[(y + g_add) >> SCALEBITS];\ + b = cm[(y + b_add) >> SCALEBITS];\ +} + +#define YUV_TO_RGB1(cb1, cr1)\ +{\ + cb = (cb1) - 128;\ + cr = (cr1) - 128;\ + r_add = FIX(1.40200) * cr + ONE_HALF;\ + g_add = - FIX(0.34414) * cb - FIX(0.71414) * cr + ONE_HALF;\ + b_add = FIX(1.77200) * cb + ONE_HALF;\ +} + +#define YUV_TO_RGB2(r, g, b, y1)\ +{\ + y = (y1) << SCALEBITS;\ + r = cm[(y + r_add) >> SCALEBITS];\ + g = cm[(y + g_add) >> SCALEBITS];\ + b = cm[(y + b_add) >> SCALEBITS];\ +} + +#define Y_CCIR_TO_JPEG(y)\ + cm[((y) * FIX(255.0/219.0) + (ONE_HALF - 16 * FIX(255.0/219.0))) >> SCALEBITS] + +#define Y_JPEG_TO_CCIR(y)\ + (((y) * FIX(219.0/255.0) + (ONE_HALF + (16 << SCALEBITS))) >> SCALEBITS) + +#define C_CCIR_TO_JPEG(y)\ + cm[(((y) - 128) * FIX(127.0/112.0) + (ONE_HALF + (128 << SCALEBITS))) >> SCALEBITS] + +/* NOTE: the clamp is really necessary! */ +static inline int C_JPEG_TO_CCIR(int y) { + y = (((y - 128) * FIX(112.0/127.0) + (ONE_HALF + (128 << SCALEBITS))) >> SCALEBITS); + if (y < 16) + y = 16; + return y; +} + + +#define RGB_TO_Y_CCIR(r, g, b) \ +((FIX(0.29900*219.0/255.0) * (r) + FIX(0.58700*219.0/255.0) * (g) + \ + FIX(0.11400*219.0/255.0) * (b) + (ONE_HALF + (16 << SCALEBITS))) >> SCALEBITS) + +#define RGB_TO_U_CCIR(r1, g1, b1, shift)\ +(((- FIX(0.16874*224.0/255.0) * r1 - FIX(0.33126*224.0/255.0) * g1 + \ + FIX(0.50000*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128) + +#define RGB_TO_V_CCIR(r1, g1, b1, shift)\ +(((FIX(0.50000*224.0/255.0) * r1 - FIX(0.41869*224.0/255.0) * g1 - \ + FIX(0.08131*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128) + +#define RGB_TO_Y_JPEG(r, g, b) \ +(FFMIN((FIX(0.29900) * (r) + FIX(0.58700) * (g) + \ + FIX(0.11400) * (b) + (ONE_HALF)) >> SCALEBITS, 255)) + +#define RGB_TO_U_JPEG(r1, g1, b1)\ +(((- FIX(0.16874) * r1 - FIX(0.33126) * g1 + \ + FIX(0.50000) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128) + +#define RGB_TO_V_JPEG(r1, g1, b1)\ +(((FIX(0.50000) * r1 - FIX(0.41869) * g1 - \ + FIX(0.08131) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128) + +// Conversion macros for 8-bit RGB to YUV +// Derived from ITU-R BT.709-6 (06/2015) Item 3.5 +// https://www.itu.int/rec/R-REC-BT.709-6-201506-I/en + +#define RGB_TO_Y_BT709(r, g, b) \ +((FIX(0.21260*219.0/255.0) * (r) + FIX(0.71520*219.0/255.0) * (g) + \ + FIX(0.07220*219.0/255.0) * (b) + (ONE_HALF + (16 << SCALEBITS))) >> SCALEBITS) + +#define RGB_TO_U_BT709(r1, g1, b1, shift)\ +(((- FIX(0.11457*224.0/255.0) * r1 - FIX(0.38543*224.0/255.0) * g1 + \ + FIX(0.50000*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128) + +#define RGB_TO_V_BT709(r1, g1, b1, shift)\ +(((FIX(0.50000*224.0/255.0) * r1 - FIX(0.45415*224.0/255.0) * g1 - \ + FIX(0.04585*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128) + +#define RGB_TO_Y_BT709_FULL(r, g, b) \ +(FFMIN((FIX(0.21260) * (r) + FIX(0.71520) * (g) + \ + FIX(0.07220) * (b) + (ONE_HALF)) >> SCALEBITS, 255)) + +#define RGB_TO_U_BT709_FULL(r1, g1, b1)\ +(((- FIX(0.11457) * r1 - FIX(0.38543) * g1 + \ + FIX(0.50000) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128) + +#define RGB_TO_V_BT709_FULL(r1, g1, b1)\ +(((FIX(0.50000) * r1 - FIX(0.45415) * g1 - \ + FIX(0.04585) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128) + +#endif /* AVUTIL_COLORSPACE_H */ diff --git a/media/ffvpx/libavutil/common.h b/media/ffvpx/libavutil/common.h new file mode 100644 index 0000000000..de2140a678 --- /dev/null +++ b/media/ffvpx/libavutil/common.h @@ -0,0 +1,579 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * common internal and external API header + */ + +#ifndef AVUTIL_COMMON_H +#define AVUTIL_COMMON_H + +#if defined(__cplusplus) && !defined(__STDC_CONSTANT_MACROS) && !defined(UINT64_C) +#error missing -D__STDC_CONSTANT_MACROS / #define __STDC_CONSTANT_MACROS +#endif + +#include <errno.h> +#include <inttypes.h> +#include <limits.h> +#include <math.h> +#include <stdint.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include "attributes.h" +#include "error.h" +#include "macros.h" + +//rounded division & shift +#define RSHIFT(a,b) ((a) > 0 ? ((a) + ((1<<(b))>>1))>>(b) : ((a) + ((1<<(b))>>1)-1)>>(b)) +/* assume b>0 */ +#define ROUNDED_DIV(a,b) (((a)>=0 ? (a) + ((b)>>1) : (a) - ((b)>>1))/(b)) +/* Fast a/(1<<b) rounded toward +inf. Assume a>=0 and b>=0 */ +#define AV_CEIL_RSHIFT(a,b) (!av_builtin_constant_p(b) ? -((-(a)) >> (b)) \ + : ((a) + (1<<(b)) - 1) >> (b)) +/* Backwards compat. */ +#define FF_CEIL_RSHIFT AV_CEIL_RSHIFT + +#define FFUDIV(a,b) (((a)>0 ?(a):(a)-(b)+1) / (b)) +#define FFUMOD(a,b) ((a)-(b)*FFUDIV(a,b)) + +/** + * Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they + * are not representable as absolute values of their type. This is the same + * as with *abs() + * @see FFNABS() + */ +#define FFABS(a) ((a) >= 0 ? (a) : (-(a))) +#define FFSIGN(a) ((a) > 0 ? 1 : -1) + +/** + * Negative Absolute value. + * this works for all integers of all types. + * As with many macros, this evaluates its argument twice, it thus must not have + * a sideeffect, that is FFNABS(x++) has undefined behavior. + */ +#define FFNABS(a) ((a) <= 0 ? (a) : (-(a))) + +/** + * Unsigned Absolute value. + * This takes the absolute value of a signed int and returns it as a unsigned. + * This also works with INT_MIN which would otherwise not be representable + * As with many macros, this evaluates its argument twice. + */ +#define FFABSU(a) ((a) <= 0 ? -(unsigned)(a) : (unsigned)(a)) +#define FFABS64U(a) ((a) <= 0 ? -(uint64_t)(a) : (uint64_t)(a)) + +/* misc math functions */ + +#ifdef HAVE_AV_CONFIG_H +# include "config.h" +# include "intmath.h" +#endif + +#ifndef av_ceil_log2 +# define av_ceil_log2 av_ceil_log2_c +#endif +#ifndef av_clip +# define av_clip av_clip_c +#endif +#ifndef av_clip64 +# define av_clip64 av_clip64_c +#endif +#ifndef av_clip_uint8 +# define av_clip_uint8 av_clip_uint8_c +#endif +#ifndef av_clip_int8 +# define av_clip_int8 av_clip_int8_c +#endif +#ifndef av_clip_uint16 +# define av_clip_uint16 av_clip_uint16_c +#endif +#ifndef av_clip_int16 +# define av_clip_int16 av_clip_int16_c +#endif +#ifndef av_clipl_int32 +# define av_clipl_int32 av_clipl_int32_c +#endif +#ifndef av_clip_intp2 +# define av_clip_intp2 av_clip_intp2_c +#endif +#ifndef av_clip_uintp2 +# define av_clip_uintp2 av_clip_uintp2_c +#endif +#ifndef av_mod_uintp2 +# define av_mod_uintp2 av_mod_uintp2_c +#endif +#ifndef av_sat_add32 +# define av_sat_add32 av_sat_add32_c +#endif +#ifndef av_sat_dadd32 +# define av_sat_dadd32 av_sat_dadd32_c +#endif +#ifndef av_sat_sub32 +# define av_sat_sub32 av_sat_sub32_c +#endif +#ifndef av_sat_dsub32 +# define av_sat_dsub32 av_sat_dsub32_c +#endif +#ifndef av_sat_add64 +# define av_sat_add64 av_sat_add64_c +#endif +#ifndef av_sat_sub64 +# define av_sat_sub64 av_sat_sub64_c +#endif +#ifndef av_clipf +# define av_clipf av_clipf_c +#endif +#ifndef av_clipd +# define av_clipd av_clipd_c +#endif +#ifndef av_popcount +# define av_popcount av_popcount_c +#endif +#ifndef av_popcount64 +# define av_popcount64 av_popcount64_c +#endif +#ifndef av_parity +# define av_parity av_parity_c +#endif + +#ifndef av_log2 +av_const int av_log2(unsigned v); +#endif + +#ifndef av_log2_16bit +av_const int av_log2_16bit(unsigned v); +#endif + +/** + * Clip a signed integer value into the amin-amax range. + * @param a value to clip + * @param amin minimum value of the clip range + * @param amax maximum value of the clip range + * @return clipped value + */ +static av_always_inline av_const int av_clip_c(int a, int amin, int amax) +{ +#if defined(HAVE_AV_CONFIG_H) && defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + if (a < amin) return amin; + else if (a > amax) return amax; + else return a; +} + +/** + * Clip a signed 64bit integer value into the amin-amax range. + * @param a value to clip + * @param amin minimum value of the clip range + * @param amax maximum value of the clip range + * @return clipped value + */ +static av_always_inline av_const int64_t av_clip64_c(int64_t a, int64_t amin, int64_t amax) +{ +#if defined(HAVE_AV_CONFIG_H) && defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + if (a < amin) return amin; + else if (a > amax) return amax; + else return a; +} + +/** + * Clip a signed integer value into the 0-255 range. + * @param a value to clip + * @return clipped value + */ +static av_always_inline av_const uint8_t av_clip_uint8_c(int a) +{ + if (a&(~0xFF)) return (~a)>>31; + else return a; +} + +/** + * Clip a signed integer value into the -128,127 range. + * @param a value to clip + * @return clipped value + */ +static av_always_inline av_const int8_t av_clip_int8_c(int a) +{ + if ((a+0x80U) & ~0xFF) return (a>>31) ^ 0x7F; + else return a; +} + +/** + * Clip a signed integer value into the 0-65535 range. + * @param a value to clip + * @return clipped value + */ +static av_always_inline av_const uint16_t av_clip_uint16_c(int a) +{ + if (a&(~0xFFFF)) return (~a)>>31; + else return a; +} + +/** + * Clip a signed integer value into the -32768,32767 range. + * @param a value to clip + * @return clipped value + */ +static av_always_inline av_const int16_t av_clip_int16_c(int a) +{ + if ((a+0x8000U) & ~0xFFFF) return (a>>31) ^ 0x7FFF; + else return a; +} + +/** + * Clip a signed 64-bit integer value into the -2147483648,2147483647 range. + * @param a value to clip + * @return clipped value + */ +static av_always_inline av_const int32_t av_clipl_int32_c(int64_t a) +{ + if ((a+0x80000000u) & ~UINT64_C(0xFFFFFFFF)) return (int32_t)((a>>63) ^ 0x7FFFFFFF); + else return (int32_t)a; +} + +/** + * Clip a signed integer into the -(2^p),(2^p-1) range. + * @param a value to clip + * @param p bit position to clip at + * @return clipped value + */ +static av_always_inline av_const int av_clip_intp2_c(int a, int p) +{ + if (((unsigned)a + (1 << p)) & ~((2 << p) - 1)) + return (a >> 31) ^ ((1 << p) - 1); + else + return a; +} + +/** + * Clip a signed integer to an unsigned power of two range. + * @param a value to clip + * @param p bit position to clip at + * @return clipped value + */ +static av_always_inline av_const unsigned av_clip_uintp2_c(int a, int p) +{ + if (a & ~((1<<p) - 1)) return (~a) >> 31 & ((1<<p) - 1); + else return a; +} + +/** + * Clear high bits from an unsigned integer starting with specific bit position + * @param a value to clip + * @param p bit position to clip at + * @return clipped value + */ +static av_always_inline av_const unsigned av_mod_uintp2_c(unsigned a, unsigned p) +{ + return a & ((1U << p) - 1); +} + +/** + * Add two signed 32-bit values with saturation. + * + * @param a one value + * @param b another value + * @return sum with signed saturation + */ +static av_always_inline int av_sat_add32_c(int a, int b) +{ + return av_clipl_int32((int64_t)a + b); +} + +/** + * Add a doubled value to another value with saturation at both stages. + * + * @param a first value + * @param b value doubled and added to a + * @return sum sat(a + sat(2*b)) with signed saturation + */ +static av_always_inline int av_sat_dadd32_c(int a, int b) +{ + return av_sat_add32(a, av_sat_add32(b, b)); +} + +/** + * Subtract two signed 32-bit values with saturation. + * + * @param a one value + * @param b another value + * @return difference with signed saturation + */ +static av_always_inline int av_sat_sub32_c(int a, int b) +{ + return av_clipl_int32((int64_t)a - b); +} + +/** + * Subtract a doubled value from another value with saturation at both stages. + * + * @param a first value + * @param b value doubled and subtracted from a + * @return difference sat(a - sat(2*b)) with signed saturation + */ +static av_always_inline int av_sat_dsub32_c(int a, int b) +{ + return av_sat_sub32(a, av_sat_add32(b, b)); +} + +/** + * Add two signed 64-bit values with saturation. + * + * @param a one value + * @param b another value + * @return sum with signed saturation + */ +static av_always_inline int64_t av_sat_add64_c(int64_t a, int64_t b) { +#if (!defined(__INTEL_COMPILER) && AV_GCC_VERSION_AT_LEAST(5,1)) || AV_HAS_BUILTIN(__builtin_add_overflow) + int64_t tmp; + return !__builtin_add_overflow(a, b, &tmp) ? tmp : (tmp < 0 ? INT64_MAX : INT64_MIN); +#else + int64_t s = a+(uint64_t)b; + if ((int64_t)(a^b | ~s^b) >= 0) + return INT64_MAX ^ (b >> 63); + return s; +#endif +} + +/** + * Subtract two signed 64-bit values with saturation. + * + * @param a one value + * @param b another value + * @return difference with signed saturation + */ +static av_always_inline int64_t av_sat_sub64_c(int64_t a, int64_t b) { +#if (!defined(__INTEL_COMPILER) && AV_GCC_VERSION_AT_LEAST(5,1)) || AV_HAS_BUILTIN(__builtin_sub_overflow) + int64_t tmp; + return !__builtin_sub_overflow(a, b, &tmp) ? tmp : (tmp < 0 ? INT64_MAX : INT64_MIN); +#else + if (b <= 0 && a >= INT64_MAX + b) + return INT64_MAX; + if (b >= 0 && a <= INT64_MIN + b) + return INT64_MIN; + return a - b; +#endif +} + +/** + * Clip a float value into the amin-amax range. + * If a is nan or -inf amin will be returned. + * If a is +inf amax will be returned. + * @param a value to clip + * @param amin minimum value of the clip range + * @param amax maximum value of the clip range + * @return clipped value + */ +static av_always_inline av_const float av_clipf_c(float a, float amin, float amax) +{ +#if defined(HAVE_AV_CONFIG_H) && defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + return FFMIN(FFMAX(a, amin), amax); +} + +/** + * Clip a double value into the amin-amax range. + * If a is nan or -inf amin will be returned. + * If a is +inf amax will be returned. + * @param a value to clip + * @param amin minimum value of the clip range + * @param amax maximum value of the clip range + * @return clipped value + */ +static av_always_inline av_const double av_clipd_c(double a, double amin, double amax) +{ +#if defined(HAVE_AV_CONFIG_H) && defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + return FFMIN(FFMAX(a, amin), amax); +} + +/** Compute ceil(log2(x)). + * @param x value used to compute ceil(log2(x)) + * @return computed ceiling of log2(x) + */ +static av_always_inline av_const int av_ceil_log2_c(int x) +{ + return av_log2((x - 1U) << 1); +} + +/** + * Count number of bits set to one in x + * @param x value to count bits of + * @return the number of bits set to one in x + */ +static av_always_inline av_const int av_popcount_c(uint32_t x) +{ + x -= (x >> 1) & 0x55555555; + x = (x & 0x33333333) + ((x >> 2) & 0x33333333); + x = (x + (x >> 4)) & 0x0F0F0F0F; + x += x >> 8; + return (x + (x >> 16)) & 0x3F; +} + +/** + * Count number of bits set to one in x + * @param x value to count bits of + * @return the number of bits set to one in x + */ +static av_always_inline av_const int av_popcount64_c(uint64_t x) +{ + return av_popcount((uint32_t)x) + av_popcount((uint32_t)(x >> 32)); +} + +static av_always_inline av_const int av_parity_c(uint32_t v) +{ + return av_popcount(v) & 1; +} + +/** + * Convert a UTF-8 character (up to 4 bytes) to its 32-bit UCS-4 encoded form. + * + * @param val Output value, must be an lvalue of type uint32_t. + * @param GET_BYTE Expression reading one byte from the input. + * Evaluated up to 7 times (4 for the currently + * assigned Unicode range). With a memory buffer + * input, this could be *ptr++, or if you want to make sure + * that *ptr stops at the end of a NULL terminated string then + * *ptr ? *ptr++ : 0 + * @param ERROR Expression to be evaluated on invalid input, + * typically a goto statement. + * + * @warning ERROR should not contain a loop control statement which + * could interact with the internal while loop, and should force an + * exit from the macro code (e.g. through a goto or a return) in order + * to prevent undefined results. + */ +#define GET_UTF8(val, GET_BYTE, ERROR)\ + val= (GET_BYTE);\ + {\ + uint32_t top = (val & 128) >> 1;\ + if ((val & 0xc0) == 0x80 || val >= 0xFE)\ + {ERROR}\ + while (val & top) {\ + unsigned int tmp = (GET_BYTE) - 128;\ + if(tmp>>6)\ + {ERROR}\ + val= (val<<6) + tmp;\ + top <<= 5;\ + }\ + val &= (top << 1) - 1;\ + } + +/** + * Convert a UTF-16 character (2 or 4 bytes) to its 32-bit UCS-4 encoded form. + * + * @param val Output value, must be an lvalue of type uint32_t. + * @param GET_16BIT Expression returning two bytes of UTF-16 data converted + * to native byte order. Evaluated one or two times. + * @param ERROR Expression to be evaluated on invalid input, + * typically a goto statement. + */ +#define GET_UTF16(val, GET_16BIT, ERROR)\ + val = (GET_16BIT);\ + {\ + unsigned int hi = val - 0xD800;\ + if (hi < 0x800) {\ + val = (GET_16BIT) - 0xDC00;\ + if (val > 0x3FFU || hi > 0x3FFU)\ + {ERROR}\ + val += (hi<<10) + 0x10000;\ + }\ + }\ + +/** + * @def PUT_UTF8(val, tmp, PUT_BYTE) + * Convert a 32-bit Unicode character to its UTF-8 encoded form (up to 4 bytes long). + * @param val is an input-only argument and should be of type uint32_t. It holds + * a UCS-4 encoded Unicode character that is to be converted to UTF-8. If + * val is given as a function it is executed only once. + * @param tmp is a temporary variable and should be of type uint8_t. It + * represents an intermediate value during conversion that is to be + * output by PUT_BYTE. + * @param PUT_BYTE writes the converted UTF-8 bytes to any proper destination. + * It could be a function or a statement, and uses tmp as the input byte. + * For example, PUT_BYTE could be "*output++ = tmp;" PUT_BYTE will be + * executed up to 4 times for values in the valid UTF-8 range and up to + * 7 times in the general case, depending on the length of the converted + * Unicode character. + */ +#define PUT_UTF8(val, tmp, PUT_BYTE)\ + {\ + int bytes, shift;\ + uint32_t in = val;\ + if (in < 0x80) {\ + tmp = in;\ + PUT_BYTE\ + } else {\ + bytes = (av_log2(in) + 4) / 5;\ + shift = (bytes - 1) * 6;\ + tmp = (256 - (256 >> bytes)) | (in >> shift);\ + PUT_BYTE\ + while (shift >= 6) {\ + shift -= 6;\ + tmp = 0x80 | ((in >> shift) & 0x3f);\ + PUT_BYTE\ + }\ + }\ + } + +/** + * @def PUT_UTF16(val, tmp, PUT_16BIT) + * Convert a 32-bit Unicode character to its UTF-16 encoded form (2 or 4 bytes). + * @param val is an input-only argument and should be of type uint32_t. It holds + * a UCS-4 encoded Unicode character that is to be converted to UTF-16. If + * val is given as a function it is executed only once. + * @param tmp is a temporary variable and should be of type uint16_t. It + * represents an intermediate value during conversion that is to be + * output by PUT_16BIT. + * @param PUT_16BIT writes the converted UTF-16 data to any proper destination + * in desired endianness. It could be a function or a statement, and uses tmp + * as the input byte. For example, PUT_BYTE could be "*output++ = tmp;" + * PUT_BYTE will be executed 1 or 2 times depending on input character. + */ +#define PUT_UTF16(val, tmp, PUT_16BIT)\ + {\ + uint32_t in = val;\ + if (in < 0x10000) {\ + tmp = in;\ + PUT_16BIT\ + } else {\ + tmp = 0xD800 | ((in - 0x10000) >> 10);\ + PUT_16BIT\ + tmp = 0xDC00 | ((in - 0x10000) & 0x3FF);\ + PUT_16BIT\ + }\ + }\ + + + +#include "mem.h" + +#ifdef HAVE_AV_CONFIG_H +# include "internal.h" +#endif /* HAVE_AV_CONFIG_H */ + +#endif /* AVUTIL_COMMON_H */ diff --git a/media/ffvpx/libavutil/cpu.c b/media/ffvpx/libavutil/cpu.c new file mode 100644 index 0000000000..48d195168c --- /dev/null +++ b/media/ffvpx/libavutil/cpu.c @@ -0,0 +1,285 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#if HAVE_SCHED_GETAFFINITY +#ifndef _GNU_SOURCE +# define _GNU_SOURCE +#endif +#include <sched.h> +#endif + +#include <stddef.h> +#include <stdint.h> +#include <stdatomic.h> + +#include "attributes.h" +#include "cpu.h" +#include "cpu_internal.h" +#include "opt.h" +#include "common.h" + +#if HAVE_GETPROCESSAFFINITYMASK || HAVE_WINRT +#include <windows.h> +#endif +#if HAVE_SYSCTL +#if HAVE_SYS_PARAM_H +#include <sys/param.h> +#endif +#include <sys/types.h> +#include <sys/sysctl.h> +#endif +#if HAVE_UNISTD_H +#include <unistd.h> +#endif + +static atomic_int cpu_flags = ATOMIC_VAR_INIT(-1); +static atomic_int cpu_count = ATOMIC_VAR_INIT(-1); + +static int get_cpu_flags(void) +{ +#if ARCH_MIPS + return ff_get_cpu_flags_mips(); +#elif ARCH_AARCH64 + return ff_get_cpu_flags_aarch64(); +#elif ARCH_ARM + return ff_get_cpu_flags_arm(); +#elif ARCH_PPC + return ff_get_cpu_flags_ppc(); +#elif ARCH_RISCV + return ff_get_cpu_flags_riscv(); +#elif ARCH_X86 + return ff_get_cpu_flags_x86(); +#elif ARCH_LOONGARCH + return ff_get_cpu_flags_loongarch(); +#endif + return 0; +} + +void av_force_cpu_flags(int arg){ + if (ARCH_X86 && + (arg & ( AV_CPU_FLAG_3DNOW | + AV_CPU_FLAG_3DNOWEXT | + AV_CPU_FLAG_MMXEXT | + AV_CPU_FLAG_SSE | + AV_CPU_FLAG_SSE2 | + AV_CPU_FLAG_SSE2SLOW | + AV_CPU_FLAG_SSE3 | + AV_CPU_FLAG_SSE3SLOW | + AV_CPU_FLAG_SSSE3 | + AV_CPU_FLAG_SSE4 | + AV_CPU_FLAG_SSE42 | + AV_CPU_FLAG_AVX | + AV_CPU_FLAG_AVXSLOW | + AV_CPU_FLAG_XOP | + AV_CPU_FLAG_FMA3 | + AV_CPU_FLAG_FMA4 | + AV_CPU_FLAG_AVX2 | + AV_CPU_FLAG_AVX512 )) + && !(arg & AV_CPU_FLAG_MMX)) { + av_log(NULL, AV_LOG_WARNING, "MMX implied by specified flags\n"); + arg |= AV_CPU_FLAG_MMX; + } + + atomic_store_explicit(&cpu_flags, arg, memory_order_relaxed); +} + +int av_get_cpu_flags(void) +{ + int flags = atomic_load_explicit(&cpu_flags, memory_order_relaxed); + if (flags == -1) { + flags = get_cpu_flags(); + atomic_store_explicit(&cpu_flags, flags, memory_order_relaxed); + } + return flags; +} + +int av_parse_cpu_caps(unsigned *flags, const char *s) +{ + static const AVOption cpuflags_opts[] = { + { "flags" , NULL, 0, AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT64_MIN, INT64_MAX, .unit = "flags" }, +#if ARCH_PPC + { "altivec" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ALTIVEC }, .unit = "flags" }, +#elif ARCH_X86 + { "mmx" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MMX }, .unit = "flags" }, + { "mmx2" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MMX2 }, .unit = "flags" }, + { "mmxext" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MMX2 }, .unit = "flags" }, + { "sse" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE }, .unit = "flags" }, + { "sse2" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE2 }, .unit = "flags" }, + { "sse2slow", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE2SLOW }, .unit = "flags" }, + { "sse3" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE3 }, .unit = "flags" }, + { "sse3slow", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE3SLOW }, .unit = "flags" }, + { "ssse3" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSSE3 }, .unit = "flags" }, + { "atom" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ATOM }, .unit = "flags" }, + { "sse4.1" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE4 }, .unit = "flags" }, + { "sse4.2" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SSE42 }, .unit = "flags" }, + { "avx" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVX }, .unit = "flags" }, + { "avxslow" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVXSLOW }, .unit = "flags" }, + { "xop" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_XOP }, .unit = "flags" }, + { "fma3" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_FMA3 }, .unit = "flags" }, + { "fma4" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_FMA4 }, .unit = "flags" }, + { "avx2" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVX2 }, .unit = "flags" }, + { "bmi1" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_BMI1 }, .unit = "flags" }, + { "bmi2" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_BMI2 }, .unit = "flags" }, + { "3dnow" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_3DNOW }, .unit = "flags" }, + { "3dnowext", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_3DNOWEXT }, .unit = "flags" }, + { "cmov", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_CMOV }, .unit = "flags" }, + { "aesni", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AESNI }, .unit = "flags" }, + { "avx512" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVX512 }, .unit = "flags" }, + { "avx512icl", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVX512ICL }, .unit = "flags" }, + { "slowgather", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SLOW_GATHER }, .unit = "flags" }, + +#define CPU_FLAG_P2 AV_CPU_FLAG_CMOV | AV_CPU_FLAG_MMX +#define CPU_FLAG_P3 CPU_FLAG_P2 | AV_CPU_FLAG_MMX2 | AV_CPU_FLAG_SSE +#define CPU_FLAG_P4 CPU_FLAG_P3| AV_CPU_FLAG_SSE2 + { "pentium2", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_P2 }, .unit = "flags" }, + { "pentium3", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_P3 }, .unit = "flags" }, + { "pentium4", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_P4 }, .unit = "flags" }, + +#define CPU_FLAG_K62 AV_CPU_FLAG_MMX | AV_CPU_FLAG_3DNOW +#define CPU_FLAG_ATHLON CPU_FLAG_K62 | AV_CPU_FLAG_CMOV | AV_CPU_FLAG_3DNOWEXT | AV_CPU_FLAG_MMX2 +#define CPU_FLAG_ATHLONXP CPU_FLAG_ATHLON | AV_CPU_FLAG_SSE +#define CPU_FLAG_K8 CPU_FLAG_ATHLONXP | AV_CPU_FLAG_SSE2 + { "k6", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MMX }, .unit = "flags" }, + { "k62", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_K62 }, .unit = "flags" }, + { "athlon", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_ATHLON }, .unit = "flags" }, + { "athlonxp", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_ATHLONXP }, .unit = "flags" }, + { "k8", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPU_FLAG_K8 }, .unit = "flags" }, +#elif ARCH_ARM + { "armv5te", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV5TE }, .unit = "flags" }, + { "armv6", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV6 }, .unit = "flags" }, + { "armv6t2", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV6T2 }, .unit = "flags" }, + { "vfp", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_VFP }, .unit = "flags" }, + { "vfp_vm", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_VFP_VM }, .unit = "flags" }, + { "vfpv3", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_VFPV3 }, .unit = "flags" }, + { "neon", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_NEON }, .unit = "flags" }, + { "setend", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_SETEND }, .unit = "flags" }, +#elif ARCH_AARCH64 + { "armv8", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV8 }, .unit = "flags" }, + { "neon", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_NEON }, .unit = "flags" }, + { "vfp", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_VFP }, .unit = "flags" }, + { "dotprod", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_DOTPROD }, .unit = "flags" }, + { "i8mm", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_I8MM }, .unit = "flags" }, +#elif ARCH_MIPS + { "mmi", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MMI }, .unit = "flags" }, + { "msa", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_MSA }, .unit = "flags" }, +#elif ARCH_LOONGARCH + { "lsx", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_LSX }, .unit = "flags" }, + { "lasx", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_LASX }, .unit = "flags" }, +#elif ARCH_RISCV + { "rvi", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVI }, .unit = "flags" }, + { "rvf", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVF }, .unit = "flags" }, + { "rvd", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVD }, .unit = "flags" }, + { "zve32x", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_I32 }, .unit = "flags" }, + { "zve32f", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F32 }, .unit = "flags" }, + { "zve64x", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_I64 }, .unit = "flags" }, + { "zve64d", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVV_F64 }, .unit = "flags" }, + { "zba", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVB_ADDR }, .unit = "flags" }, + { "zbb", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_RVB_BASIC }, .unit = "flags" }, +#endif + { NULL }, + }; + static const AVClass class = { + .class_name = "cpuflags", + .item_name = av_default_item_name, + .option = cpuflags_opts, + .version = LIBAVUTIL_VERSION_INT, + }; + const AVClass *pclass = &class; + + return av_opt_eval_flags(&pclass, &cpuflags_opts[0], s, flags); +} + +int av_cpu_count(void) +{ + static atomic_int printed = ATOMIC_VAR_INIT(0); + + int nb_cpus = 1; + int count = 0; +#if HAVE_WINRT + SYSTEM_INFO sysinfo; +#endif +#if HAVE_SCHED_GETAFFINITY && defined(CPU_COUNT) + cpu_set_t cpuset; + + CPU_ZERO(&cpuset); + + if (!sched_getaffinity(0, sizeof(cpuset), &cpuset)) + nb_cpus = CPU_COUNT(&cpuset); +#elif HAVE_GETPROCESSAFFINITYMASK + DWORD_PTR proc_aff, sys_aff; + if (GetProcessAffinityMask(GetCurrentProcess(), &proc_aff, &sys_aff)) + nb_cpus = av_popcount64(proc_aff); +#elif HAVE_SYSCTL && defined(HW_NCPUONLINE) + int mib[2] = { CTL_HW, HW_NCPUONLINE }; + size_t len = sizeof(nb_cpus); + + if (sysctl(mib, 2, &nb_cpus, &len, NULL, 0) == -1) + nb_cpus = 0; +#elif HAVE_SYSCTL && defined(HW_NCPU) + int mib[2] = { CTL_HW, HW_NCPU }; + size_t len = sizeof(nb_cpus); + + if (sysctl(mib, 2, &nb_cpus, &len, NULL, 0) == -1) + nb_cpus = 0; +#elif HAVE_SYSCONF && defined(_SC_NPROC_ONLN) + nb_cpus = sysconf(_SC_NPROC_ONLN); +#elif HAVE_SYSCONF && defined(_SC_NPROCESSORS_ONLN) + nb_cpus = sysconf(_SC_NPROCESSORS_ONLN); +#elif HAVE_WINRT + GetNativeSystemInfo(&sysinfo); + nb_cpus = sysinfo.dwNumberOfProcessors; +#endif + + if (!atomic_exchange_explicit(&printed, 1, memory_order_relaxed)) + av_log(NULL, AV_LOG_DEBUG, "detected %d logical cores\n", nb_cpus); + + count = atomic_load_explicit(&cpu_count, memory_order_relaxed); + + if (count > 0) { + nb_cpus = count; + av_log(NULL, AV_LOG_DEBUG, "overriding to %d logical cores\n", nb_cpus); + } + + return nb_cpus; +} + +void av_cpu_force_count(int count) +{ + atomic_store_explicit(&cpu_count, count, memory_order_relaxed); +} + +size_t av_cpu_max_align(void) +{ +#if ARCH_MIPS + return ff_get_cpu_max_align_mips(); +#elif ARCH_AARCH64 + return ff_get_cpu_max_align_aarch64(); +#elif ARCH_ARM + return ff_get_cpu_max_align_arm(); +#elif ARCH_PPC + return ff_get_cpu_max_align_ppc(); +#elif ARCH_X86 + return ff_get_cpu_max_align_x86(); +#elif ARCH_LOONGARCH + return ff_get_cpu_max_align_loongarch(); +#endif + + return 8; +} diff --git a/media/ffvpx/libavutil/cpu.h b/media/ffvpx/libavutil/cpu.h new file mode 100644 index 0000000000..8dff341886 --- /dev/null +++ b/media/ffvpx/libavutil/cpu.h @@ -0,0 +1,137 @@ +/* + * Copyright (c) 2000, 2001, 2002 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_CPU_H +#define AVUTIL_CPU_H + +#include <stddef.h> + +#define AV_CPU_FLAG_FORCE 0x80000000 /* force usage of selected flags (OR) */ + + /* lower 16 bits - CPU features */ +#define AV_CPU_FLAG_MMX 0x0001 ///< standard MMX +#define AV_CPU_FLAG_MMXEXT 0x0002 ///< SSE integer functions or AMD MMX ext +#define AV_CPU_FLAG_MMX2 0x0002 ///< SSE integer functions or AMD MMX ext +#define AV_CPU_FLAG_3DNOW 0x0004 ///< AMD 3DNOW +#define AV_CPU_FLAG_SSE 0x0008 ///< SSE functions +#define AV_CPU_FLAG_SSE2 0x0010 ///< PIV SSE2 functions +#define AV_CPU_FLAG_SSE2SLOW 0x40000000 ///< SSE2 supported, but usually not faster + ///< than regular MMX/SSE (e.g. Core1) +#define AV_CPU_FLAG_3DNOWEXT 0x0020 ///< AMD 3DNowExt +#define AV_CPU_FLAG_SSE3 0x0040 ///< Prescott SSE3 functions +#define AV_CPU_FLAG_SSE3SLOW 0x20000000 ///< SSE3 supported, but usually not faster + ///< than regular MMX/SSE (e.g. Core1) +#define AV_CPU_FLAG_SSSE3 0x0080 ///< Conroe SSSE3 functions +#define AV_CPU_FLAG_SSSE3SLOW 0x4000000 ///< SSSE3 supported, but usually not faster +#define AV_CPU_FLAG_ATOM 0x10000000 ///< Atom processor, some SSSE3 instructions are slower +#define AV_CPU_FLAG_SSE4 0x0100 ///< Penryn SSE4.1 functions +#define AV_CPU_FLAG_SSE42 0x0200 ///< Nehalem SSE4.2 functions +#define AV_CPU_FLAG_AESNI 0x80000 ///< Advanced Encryption Standard functions +#define AV_CPU_FLAG_AVX 0x4000 ///< AVX functions: requires OS support even if YMM registers aren't used +#define AV_CPU_FLAG_AVXSLOW 0x8000000 ///< AVX supported, but slow when using YMM registers (e.g. Bulldozer) +#define AV_CPU_FLAG_XOP 0x0400 ///< Bulldozer XOP functions +#define AV_CPU_FLAG_FMA4 0x0800 ///< Bulldozer FMA4 functions +#define AV_CPU_FLAG_CMOV 0x1000 ///< supports cmov instruction +#define AV_CPU_FLAG_AVX2 0x8000 ///< AVX2 functions: requires OS support even if YMM registers aren't used +#define AV_CPU_FLAG_FMA3 0x10000 ///< Haswell FMA3 functions +#define AV_CPU_FLAG_BMI1 0x20000 ///< Bit Manipulation Instruction Set 1 +#define AV_CPU_FLAG_BMI2 0x40000 ///< Bit Manipulation Instruction Set 2 +#define AV_CPU_FLAG_AVX512 0x100000 ///< AVX-512 functions: requires OS support even if YMM/ZMM registers aren't used +#define AV_CPU_FLAG_AVX512ICL 0x200000 ///< F/CD/BW/DQ/VL/VNNI/IFMA/VBMI/VBMI2/VPOPCNTDQ/BITALG/GFNI/VAES/VPCLMULQDQ +#define AV_CPU_FLAG_SLOW_GATHER 0x2000000 ///< CPU has slow gathers. + +#define AV_CPU_FLAG_ALTIVEC 0x0001 ///< standard +#define AV_CPU_FLAG_VSX 0x0002 ///< ISA 2.06 +#define AV_CPU_FLAG_POWER8 0x0004 ///< ISA 2.07 + +#define AV_CPU_FLAG_ARMV5TE (1 << 0) +#define AV_CPU_FLAG_ARMV6 (1 << 1) +#define AV_CPU_FLAG_ARMV6T2 (1 << 2) +#define AV_CPU_FLAG_VFP (1 << 3) +#define AV_CPU_FLAG_VFPV3 (1 << 4) +#define AV_CPU_FLAG_NEON (1 << 5) +#define AV_CPU_FLAG_ARMV8 (1 << 6) +#define AV_CPU_FLAG_VFP_VM (1 << 7) ///< VFPv2 vector mode, deprecated in ARMv7-A and unavailable in various CPUs implementations +#define AV_CPU_FLAG_DOTPROD (1 << 8) +#define AV_CPU_FLAG_I8MM (1 << 9) +#define AV_CPU_FLAG_SETEND (1 <<16) + +#define AV_CPU_FLAG_MMI (1 << 0) +#define AV_CPU_FLAG_MSA (1 << 1) + +//Loongarch SIMD extension. +#define AV_CPU_FLAG_LSX (1 << 0) +#define AV_CPU_FLAG_LASX (1 << 1) + +// RISC-V extensions +#define AV_CPU_FLAG_RVI (1 << 0) ///< I (full GPR bank) +#define AV_CPU_FLAG_RVF (1 << 1) ///< F (single precision FP) +#define AV_CPU_FLAG_RVD (1 << 2) ///< D (double precision FP) +#define AV_CPU_FLAG_RVV_I32 (1 << 3) ///< Vectors of 8/16/32-bit int's */ +#define AV_CPU_FLAG_RVV_F32 (1 << 4) ///< Vectors of float's */ +#define AV_CPU_FLAG_RVV_I64 (1 << 5) ///< Vectors of 64-bit int's */ +#define AV_CPU_FLAG_RVV_F64 (1 << 6) ///< Vectors of double's +#define AV_CPU_FLAG_RVB_BASIC (1 << 7) ///< Basic bit-manipulations +#define AV_CPU_FLAG_RVB_ADDR (1 << 8) ///< Address bit-manipulations + +/** + * Return the flags which specify extensions supported by the CPU. + * The returned value is affected by av_force_cpu_flags() if that was used + * before. So av_get_cpu_flags() can easily be used in an application to + * detect the enabled cpu flags. + */ +int av_get_cpu_flags(void); + +/** + * Disables cpu detection and forces the specified flags. + * -1 is a special case that disables forcing of specific flags. + */ +void av_force_cpu_flags(int flags); + +/** + * Parse CPU caps from a string and update the given AV_CPU_* flags based on that. + * + * @return negative on error. + */ +int av_parse_cpu_caps(unsigned *flags, const char *s); + +/** + * @return the number of logical CPU cores present. + */ +int av_cpu_count(void); + +/** + * Overrides cpu count detection and forces the specified count. + * Count < 1 disables forcing of specific count. + */ +void av_cpu_force_count(int count); + +/** + * Get the maximum data alignment that may be required by FFmpeg. + * + * Note that this is affected by the build configuration and the CPU flags mask, + * so e.g. if the CPU supports AVX, but libavutil has been built with + * --disable-avx or the AV_CPU_FLAG_AVX flag has been disabled through + * av_set_cpu_flags_mask(), then this function will behave as if AVX is not + * present. + */ +size_t av_cpu_max_align(void); + +#endif /* AVUTIL_CPU_H */ diff --git a/media/ffvpx/libavutil/cpu_internal.h b/media/ffvpx/libavutil/cpu_internal.h new file mode 100644 index 0000000000..634f28bac4 --- /dev/null +++ b/media/ffvpx/libavutil/cpu_internal.h @@ -0,0 +1,62 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_CPU_INTERNAL_H +#define AVUTIL_CPU_INTERNAL_H + +#include "config.h" + +#include "cpu.h" + +#define CPUEXT_SUFFIX(flags, suffix, cpuext) \ + (HAVE_ ## cpuext ## suffix && ((flags) & AV_CPU_FLAG_ ## cpuext)) + +#define CPUEXT_SUFFIX_FAST2(flags, suffix, cpuext, slow_cpuext) \ + (HAVE_ ## cpuext ## suffix && ((flags) & AV_CPU_FLAG_ ## cpuext) && \ + !((flags) & AV_CPU_FLAG_ ## slow_cpuext ## SLOW)) + +#define CPUEXT_SUFFIX_SLOW(flags, suffix, cpuext) \ + (HAVE_ ## cpuext ## suffix && \ + ((flags) & (AV_CPU_FLAG_ ## cpuext | AV_CPU_FLAG_ ## cpuext ## SLOW))) + +#define CPUEXT_SUFFIX_SLOW2(flags, suffix, cpuext, slow_cpuext) \ + (HAVE_ ## cpuext ## suffix && ((flags) & AV_CPU_FLAG_ ## cpuext) && \ + ((flags) & (AV_CPU_FLAG_ ## slow_cpuext | AV_CPU_FLAG_ ## slow_cpuext ## SLOW))) + +#define CPUEXT_SUFFIX_FAST(flags, suffix, cpuext) CPUEXT_SUFFIX_FAST2(flags, suffix, cpuext, cpuext) + +#define CPUEXT(flags, cpuext) CPUEXT_SUFFIX(flags, , cpuext) +#define CPUEXT_FAST(flags, cpuext) CPUEXT_SUFFIX_FAST(flags, , cpuext) +#define CPUEXT_SLOW(flags, cpuext) CPUEXT_SUFFIX_SLOW(flags, , cpuext) + +int ff_get_cpu_flags_mips(void); +int ff_get_cpu_flags_aarch64(void); +int ff_get_cpu_flags_arm(void); +int ff_get_cpu_flags_ppc(void); +int ff_get_cpu_flags_riscv(void); +int ff_get_cpu_flags_x86(void); +int ff_get_cpu_flags_loongarch(void); + +size_t ff_get_cpu_max_align_mips(void); +size_t ff_get_cpu_max_align_aarch64(void); +size_t ff_get_cpu_max_align_arm(void); +size_t ff_get_cpu_max_align_ppc(void); +size_t ff_get_cpu_max_align_x86(void); +size_t ff_get_cpu_max_align_loongarch(void); + +#endif /* AVUTIL_CPU_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/crc.c b/media/ffvpx/libavutil/crc.c new file mode 100644 index 0000000000..703b56f4e0 --- /dev/null +++ b/media/ffvpx/libavutil/crc.c @@ -0,0 +1,415 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include "thread.h" +#include "avassert.h" +#include "bswap.h" +#include "crc.h" +#include "error.h" + +#if CONFIG_HARDCODED_TABLES +static const AVCRC av_crc_table[AV_CRC_MAX][257] = { + [AV_CRC_8_ATM] = { + 0x00, 0x07, 0x0E, 0x09, 0x1C, 0x1B, 0x12, 0x15, 0x38, 0x3F, 0x36, 0x31, + 0x24, 0x23, 0x2A, 0x2D, 0x70, 0x77, 0x7E, 0x79, 0x6C, 0x6B, 0x62, 0x65, + 0x48, 0x4F, 0x46, 0x41, 0x54, 0x53, 0x5A, 0x5D, 0xE0, 0xE7, 0xEE, 0xE9, + 0xFC, 0xFB, 0xF2, 0xF5, 0xD8, 0xDF, 0xD6, 0xD1, 0xC4, 0xC3, 0xCA, 0xCD, + 0x90, 0x97, 0x9E, 0x99, 0x8C, 0x8B, 0x82, 0x85, 0xA8, 0xAF, 0xA6, 0xA1, + 0xB4, 0xB3, 0xBA, 0xBD, 0xC7, 0xC0, 0xC9, 0xCE, 0xDB, 0xDC, 0xD5, 0xD2, + 0xFF, 0xF8, 0xF1, 0xF6, 0xE3, 0xE4, 0xED, 0xEA, 0xB7, 0xB0, 0xB9, 0xBE, + 0xAB, 0xAC, 0xA5, 0xA2, 0x8F, 0x88, 0x81, 0x86, 0x93, 0x94, 0x9D, 0x9A, + 0x27, 0x20, 0x29, 0x2E, 0x3B, 0x3C, 0x35, 0x32, 0x1F, 0x18, 0x11, 0x16, + 0x03, 0x04, 0x0D, 0x0A, 0x57, 0x50, 0x59, 0x5E, 0x4B, 0x4C, 0x45, 0x42, + 0x6F, 0x68, 0x61, 0x66, 0x73, 0x74, 0x7D, 0x7A, 0x89, 0x8E, 0x87, 0x80, + 0x95, 0x92, 0x9B, 0x9C, 0xB1, 0xB6, 0xBF, 0xB8, 0xAD, 0xAA, 0xA3, 0xA4, + 0xF9, 0xFE, 0xF7, 0xF0, 0xE5, 0xE2, 0xEB, 0xEC, 0xC1, 0xC6, 0xCF, 0xC8, + 0xDD, 0xDA, 0xD3, 0xD4, 0x69, 0x6E, 0x67, 0x60, 0x75, 0x72, 0x7B, 0x7C, + 0x51, 0x56, 0x5F, 0x58, 0x4D, 0x4A, 0x43, 0x44, 0x19, 0x1E, 0x17, 0x10, + 0x05, 0x02, 0x0B, 0x0C, 0x21, 0x26, 0x2F, 0x28, 0x3D, 0x3A, 0x33, 0x34, + 0x4E, 0x49, 0x40, 0x47, 0x52, 0x55, 0x5C, 0x5B, 0x76, 0x71, 0x78, 0x7F, + 0x6A, 0x6D, 0x64, 0x63, 0x3E, 0x39, 0x30, 0x37, 0x22, 0x25, 0x2C, 0x2B, + 0x06, 0x01, 0x08, 0x0F, 0x1A, 0x1D, 0x14, 0x13, 0xAE, 0xA9, 0xA0, 0xA7, + 0xB2, 0xB5, 0xBC, 0xBB, 0x96, 0x91, 0x98, 0x9F, 0x8A, 0x8D, 0x84, 0x83, + 0xDE, 0xD9, 0xD0, 0xD7, 0xC2, 0xC5, 0xCC, 0xCB, 0xE6, 0xE1, 0xE8, 0xEF, + 0xFA, 0xFD, 0xF4, 0xF3, 0x01 + }, + [AV_CRC_8_EBU] = { + 0x00, 0x1D, 0x3A, 0x27, 0x74, 0x69, 0x4E, 0x53, 0xE8, 0xF5, 0xD2, 0xCF, + 0x9C, 0x81, 0xA6, 0xBB, 0xCD, 0xD0, 0xF7, 0xEA, 0xB9, 0xA4, 0x83, 0x9E, + 0x25, 0x38, 0x1F, 0x02, 0x51, 0x4C, 0x6B, 0x76, 0x87, 0x9A, 0xBD, 0xA0, + 0xF3, 0xEE, 0xC9, 0xD4, 0x6F, 0x72, 0x55, 0x48, 0x1B, 0x06, 0x21, 0x3C, + 0x4A, 0x57, 0x70, 0x6D, 0x3E, 0x23, 0x04, 0x19, 0xA2, 0xBF, 0x98, 0x85, + 0xD6, 0xCB, 0xEC, 0xF1, 0x13, 0x0E, 0x29, 0x34, 0x67, 0x7A, 0x5D, 0x40, + 0xFB, 0xE6, 0xC1, 0xDC, 0x8F, 0x92, 0xB5, 0xA8, 0xDE, 0xC3, 0xE4, 0xF9, + 0xAA, 0xB7, 0x90, 0x8D, 0x36, 0x2B, 0x0C, 0x11, 0x42, 0x5F, 0x78, 0x65, + 0x94, 0x89, 0xAE, 0xB3, 0xE0, 0xFD, 0xDA, 0xC7, 0x7C, 0x61, 0x46, 0x5B, + 0x08, 0x15, 0x32, 0x2F, 0x59, 0x44, 0x63, 0x7E, 0x2D, 0x30, 0x17, 0x0A, + 0xB1, 0xAC, 0x8B, 0x96, 0xC5, 0xD8, 0xFF, 0xE2, 0x26, 0x3B, 0x1C, 0x01, + 0x52, 0x4F, 0x68, 0x75, 0xCE, 0xD3, 0xF4, 0xE9, 0xBA, 0xA7, 0x80, 0x9D, + 0xEB, 0xF6, 0xD1, 0xCC, 0x9F, 0x82, 0xA5, 0xB8, 0x03, 0x1E, 0x39, 0x24, + 0x77, 0x6A, 0x4D, 0x50, 0xA1, 0xBC, 0x9B, 0x86, 0xD5, 0xC8, 0xEF, 0xF2, + 0x49, 0x54, 0x73, 0x6E, 0x3D, 0x20, 0x07, 0x1A, 0x6C, 0x71, 0x56, 0x4B, + 0x18, 0x05, 0x22, 0x3F, 0x84, 0x99, 0xBE, 0xA3, 0xF0, 0xED, 0xCA, 0xD7, + 0x35, 0x28, 0x0F, 0x12, 0x41, 0x5C, 0x7B, 0x66, 0xDD, 0xC0, 0xE7, 0xFA, + 0xA9, 0xB4, 0x93, 0x8E, 0xF8, 0xE5, 0xC2, 0xDF, 0x8C, 0x91, 0xB6, 0xAB, + 0x10, 0x0D, 0x2A, 0x37, 0x64, 0x79, 0x5E, 0x43, 0xB2, 0xAF, 0x88, 0x95, + 0xC6, 0xDB, 0xFC, 0xE1, 0x5A, 0x47, 0x60, 0x7D, 0x2E, 0x33, 0x14, 0x09, + 0x7F, 0x62, 0x45, 0x58, 0x0B, 0x16, 0x31, 0x2C, 0x97, 0x8A, 0xAD, 0xB0, + 0xE3, 0xFE, 0xD9, 0xC4, 0x01 + }, + [AV_CRC_16_ANSI] = { + 0x0000, 0x0580, 0x0F80, 0x0A00, 0x1B80, 0x1E00, 0x1400, 0x1180, + 0x3380, 0x3600, 0x3C00, 0x3980, 0x2800, 0x2D80, 0x2780, 0x2200, + 0x6380, 0x6600, 0x6C00, 0x6980, 0x7800, 0x7D80, 0x7780, 0x7200, + 0x5000, 0x5580, 0x5F80, 0x5A00, 0x4B80, 0x4E00, 0x4400, 0x4180, + 0xC380, 0xC600, 0xCC00, 0xC980, 0xD800, 0xDD80, 0xD780, 0xD200, + 0xF000, 0xF580, 0xFF80, 0xFA00, 0xEB80, 0xEE00, 0xE400, 0xE180, + 0xA000, 0xA580, 0xAF80, 0xAA00, 0xBB80, 0xBE00, 0xB400, 0xB180, + 0x9380, 0x9600, 0x9C00, 0x9980, 0x8800, 0x8D80, 0x8780, 0x8200, + 0x8381, 0x8601, 0x8C01, 0x8981, 0x9801, 0x9D81, 0x9781, 0x9201, + 0xB001, 0xB581, 0xBF81, 0xBA01, 0xAB81, 0xAE01, 0xA401, 0xA181, + 0xE001, 0xE581, 0xEF81, 0xEA01, 0xFB81, 0xFE01, 0xF401, 0xF181, + 0xD381, 0xD601, 0xDC01, 0xD981, 0xC801, 0xCD81, 0xC781, 0xC201, + 0x4001, 0x4581, 0x4F81, 0x4A01, 0x5B81, 0x5E01, 0x5401, 0x5181, + 0x7381, 0x7601, 0x7C01, 0x7981, 0x6801, 0x6D81, 0x6781, 0x6201, + 0x2381, 0x2601, 0x2C01, 0x2981, 0x3801, 0x3D81, 0x3781, 0x3201, + 0x1001, 0x1581, 0x1F81, 0x1A01, 0x0B81, 0x0E01, 0x0401, 0x0181, + 0x0383, 0x0603, 0x0C03, 0x0983, 0x1803, 0x1D83, 0x1783, 0x1203, + 0x3003, 0x3583, 0x3F83, 0x3A03, 0x2B83, 0x2E03, 0x2403, 0x2183, + 0x6003, 0x6583, 0x6F83, 0x6A03, 0x7B83, 0x7E03, 0x7403, 0x7183, + 0x5383, 0x5603, 0x5C03, 0x5983, 0x4803, 0x4D83, 0x4783, 0x4203, + 0xC003, 0xC583, 0xCF83, 0xCA03, 0xDB83, 0xDE03, 0xD403, 0xD183, + 0xF383, 0xF603, 0xFC03, 0xF983, 0xE803, 0xED83, 0xE783, 0xE203, + 0xA383, 0xA603, 0xAC03, 0xA983, 0xB803, 0xBD83, 0xB783, 0xB203, + 0x9003, 0x9583, 0x9F83, 0x9A03, 0x8B83, 0x8E03, 0x8403, 0x8183, + 0x8002, 0x8582, 0x8F82, 0x8A02, 0x9B82, 0x9E02, 0x9402, 0x9182, + 0xB382, 0xB602, 0xBC02, 0xB982, 0xA802, 0xAD82, 0xA782, 0xA202, + 0xE382, 0xE602, 0xEC02, 0xE982, 0xF802, 0xFD82, 0xF782, 0xF202, + 0xD002, 0xD582, 0xDF82, 0xDA02, 0xCB82, 0xCE02, 0xC402, 0xC182, + 0x4382, 0x4602, 0x4C02, 0x4982, 0x5802, 0x5D82, 0x5782, 0x5202, + 0x7002, 0x7582, 0x7F82, 0x7A02, 0x6B82, 0x6E02, 0x6402, 0x6182, + 0x2002, 0x2582, 0x2F82, 0x2A02, 0x3B82, 0x3E02, 0x3402, 0x3182, + 0x1382, 0x1602, 0x1C02, 0x1982, 0x0802, 0x0D82, 0x0782, 0x0202, + 0x0001 + }, + [AV_CRC_16_CCITT] = { + 0x0000, 0x2110, 0x4220, 0x6330, 0x8440, 0xA550, 0xC660, 0xE770, + 0x0881, 0x2991, 0x4AA1, 0x6BB1, 0x8CC1, 0xADD1, 0xCEE1, 0xEFF1, + 0x3112, 0x1002, 0x7332, 0x5222, 0xB552, 0x9442, 0xF772, 0xD662, + 0x3993, 0x1883, 0x7BB3, 0x5AA3, 0xBDD3, 0x9CC3, 0xFFF3, 0xDEE3, + 0x6224, 0x4334, 0x2004, 0x0114, 0xE664, 0xC774, 0xA444, 0x8554, + 0x6AA5, 0x4BB5, 0x2885, 0x0995, 0xEEE5, 0xCFF5, 0xACC5, 0x8DD5, + 0x5336, 0x7226, 0x1116, 0x3006, 0xD776, 0xF666, 0x9556, 0xB446, + 0x5BB7, 0x7AA7, 0x1997, 0x3887, 0xDFF7, 0xFEE7, 0x9DD7, 0xBCC7, + 0xC448, 0xE558, 0x8668, 0xA778, 0x4008, 0x6118, 0x0228, 0x2338, + 0xCCC9, 0xEDD9, 0x8EE9, 0xAFF9, 0x4889, 0x6999, 0x0AA9, 0x2BB9, + 0xF55A, 0xD44A, 0xB77A, 0x966A, 0x711A, 0x500A, 0x333A, 0x122A, + 0xFDDB, 0xDCCB, 0xBFFB, 0x9EEB, 0x799B, 0x588B, 0x3BBB, 0x1AAB, + 0xA66C, 0x877C, 0xE44C, 0xC55C, 0x222C, 0x033C, 0x600C, 0x411C, + 0xAEED, 0x8FFD, 0xECCD, 0xCDDD, 0x2AAD, 0x0BBD, 0x688D, 0x499D, + 0x977E, 0xB66E, 0xD55E, 0xF44E, 0x133E, 0x322E, 0x511E, 0x700E, + 0x9FFF, 0xBEEF, 0xDDDF, 0xFCCF, 0x1BBF, 0x3AAF, 0x599F, 0x788F, + 0x8891, 0xA981, 0xCAB1, 0xEBA1, 0x0CD1, 0x2DC1, 0x4EF1, 0x6FE1, + 0x8010, 0xA100, 0xC230, 0xE320, 0x0450, 0x2540, 0x4670, 0x6760, + 0xB983, 0x9893, 0xFBA3, 0xDAB3, 0x3DC3, 0x1CD3, 0x7FE3, 0x5EF3, + 0xB102, 0x9012, 0xF322, 0xD232, 0x3542, 0x1452, 0x7762, 0x5672, + 0xEAB5, 0xCBA5, 0xA895, 0x8985, 0x6EF5, 0x4FE5, 0x2CD5, 0x0DC5, + 0xE234, 0xC324, 0xA014, 0x8104, 0x6674, 0x4764, 0x2454, 0x0544, + 0xDBA7, 0xFAB7, 0x9987, 0xB897, 0x5FE7, 0x7EF7, 0x1DC7, 0x3CD7, + 0xD326, 0xF236, 0x9106, 0xB016, 0x5766, 0x7676, 0x1546, 0x3456, + 0x4CD9, 0x6DC9, 0x0EF9, 0x2FE9, 0xC899, 0xE989, 0x8AB9, 0xABA9, + 0x4458, 0x6548, 0x0678, 0x2768, 0xC018, 0xE108, 0x8238, 0xA328, + 0x7DCB, 0x5CDB, 0x3FEB, 0x1EFB, 0xF98B, 0xD89B, 0xBBAB, 0x9ABB, + 0x754A, 0x545A, 0x376A, 0x167A, 0xF10A, 0xD01A, 0xB32A, 0x923A, + 0x2EFD, 0x0FED, 0x6CDD, 0x4DCD, 0xAABD, 0x8BAD, 0xE89D, 0xC98D, + 0x267C, 0x076C, 0x645C, 0x454C, 0xA23C, 0x832C, 0xE01C, 0xC10C, + 0x1FEF, 0x3EFF, 0x5DCF, 0x7CDF, 0x9BAF, 0xBABF, 0xD98F, 0xF89F, + 0x176E, 0x367E, 0x554E, 0x745E, 0x932E, 0xB23E, 0xD10E, 0xF01E, + 0x0001 + }, + [AV_CRC_24_IEEE] = { + 0x000000, 0xFB4C86, 0x0DD58A, 0xF6990C, 0xE1E693, 0x1AAA15, 0xEC3319, + 0x177F9F, 0x3981A1, 0xC2CD27, 0x34542B, 0xCF18AD, 0xD86732, 0x232BB4, + 0xD5B2B8, 0x2EFE3E, 0x894EC5, 0x720243, 0x849B4F, 0x7FD7C9, 0x68A856, + 0x93E4D0, 0x657DDC, 0x9E315A, 0xB0CF64, 0x4B83E2, 0xBD1AEE, 0x465668, + 0x5129F7, 0xAA6571, 0x5CFC7D, 0xA7B0FB, 0xE9D10C, 0x129D8A, 0xE40486, + 0x1F4800, 0x08379F, 0xF37B19, 0x05E215, 0xFEAE93, 0xD050AD, 0x2B1C2B, + 0xDD8527, 0x26C9A1, 0x31B63E, 0xCAFAB8, 0x3C63B4, 0xC72F32, 0x609FC9, + 0x9BD34F, 0x6D4A43, 0x9606C5, 0x81795A, 0x7A35DC, 0x8CACD0, 0x77E056, + 0x591E68, 0xA252EE, 0x54CBE2, 0xAF8764, 0xB8F8FB, 0x43B47D, 0xB52D71, + 0x4E61F7, 0xD2A319, 0x29EF9F, 0xDF7693, 0x243A15, 0x33458A, 0xC8090C, + 0x3E9000, 0xC5DC86, 0xEB22B8, 0x106E3E, 0xE6F732, 0x1DBBB4, 0x0AC42B, + 0xF188AD, 0x0711A1, 0xFC5D27, 0x5BEDDC, 0xA0A15A, 0x563856, 0xAD74D0, + 0xBA0B4F, 0x4147C9, 0xB7DEC5, 0x4C9243, 0x626C7D, 0x9920FB, 0x6FB9F7, + 0x94F571, 0x838AEE, 0x78C668, 0x8E5F64, 0x7513E2, 0x3B7215, 0xC03E93, + 0x36A79F, 0xCDEB19, 0xDA9486, 0x21D800, 0xD7410C, 0x2C0D8A, 0x02F3B4, + 0xF9BF32, 0x0F263E, 0xF46AB8, 0xE31527, 0x1859A1, 0xEEC0AD, 0x158C2B, + 0xB23CD0, 0x497056, 0xBFE95A, 0x44A5DC, 0x53DA43, 0xA896C5, 0x5E0FC9, + 0xA5434F, 0x8BBD71, 0x70F1F7, 0x8668FB, 0x7D247D, 0x6A5BE2, 0x911764, + 0x678E68, 0x9CC2EE, 0xA44733, 0x5F0BB5, 0xA992B9, 0x52DE3F, 0x45A1A0, + 0xBEED26, 0x48742A, 0xB338AC, 0x9DC692, 0x668A14, 0x901318, 0x6B5F9E, + 0x7C2001, 0x876C87, 0x71F58B, 0x8AB90D, 0x2D09F6, 0xD64570, 0x20DC7C, + 0xDB90FA, 0xCCEF65, 0x37A3E3, 0xC13AEF, 0x3A7669, 0x148857, 0xEFC4D1, + 0x195DDD, 0xE2115B, 0xF56EC4, 0x0E2242, 0xF8BB4E, 0x03F7C8, 0x4D963F, + 0xB6DAB9, 0x4043B5, 0xBB0F33, 0xAC70AC, 0x573C2A, 0xA1A526, 0x5AE9A0, + 0x74179E, 0x8F5B18, 0x79C214, 0x828E92, 0x95F10D, 0x6EBD8B, 0x982487, + 0x636801, 0xC4D8FA, 0x3F947C, 0xC90D70, 0x3241F6, 0x253E69, 0xDE72EF, + 0x28EBE3, 0xD3A765, 0xFD595B, 0x0615DD, 0xF08CD1, 0x0BC057, 0x1CBFC8, + 0xE7F34E, 0x116A42, 0xEA26C4, 0x76E42A, 0x8DA8AC, 0x7B31A0, 0x807D26, + 0x9702B9, 0x6C4E3F, 0x9AD733, 0x619BB5, 0x4F658B, 0xB4290D, 0x42B001, + 0xB9FC87, 0xAE8318, 0x55CF9E, 0xA35692, 0x581A14, 0xFFAAEF, 0x04E669, + 0xF27F65, 0x0933E3, 0x1E4C7C, 0xE500FA, 0x1399F6, 0xE8D570, 0xC62B4E, + 0x3D67C8, 0xCBFEC4, 0x30B242, 0x27CDDD, 0xDC815B, 0x2A1857, 0xD154D1, + 0x9F3526, 0x6479A0, 0x92E0AC, 0x69AC2A, 0x7ED3B5, 0x859F33, 0x73063F, + 0x884AB9, 0xA6B487, 0x5DF801, 0xAB610D, 0x502D8B, 0x475214, 0xBC1E92, + 0x4A879E, 0xB1CB18, 0x167BE3, 0xED3765, 0x1BAE69, 0xE0E2EF, 0xF79D70, + 0x0CD1F6, 0xFA48FA, 0x01047C, 0x2FFA42, 0xD4B6C4, 0x222FC8, 0xD9634E, + 0xCE1CD1, 0x355057, 0xC3C95B, 0x3885DD, 0x000001, + }, + [AV_CRC_32_IEEE] = { + 0x00000000, 0xB71DC104, 0x6E3B8209, 0xD926430D, 0xDC760413, 0x6B6BC517, + 0xB24D861A, 0x0550471E, 0xB8ED0826, 0x0FF0C922, 0xD6D68A2F, 0x61CB4B2B, + 0x649B0C35, 0xD386CD31, 0x0AA08E3C, 0xBDBD4F38, 0x70DB114C, 0xC7C6D048, + 0x1EE09345, 0xA9FD5241, 0xACAD155F, 0x1BB0D45B, 0xC2969756, 0x758B5652, + 0xC836196A, 0x7F2BD86E, 0xA60D9B63, 0x11105A67, 0x14401D79, 0xA35DDC7D, + 0x7A7B9F70, 0xCD665E74, 0xE0B62398, 0x57ABE29C, 0x8E8DA191, 0x39906095, + 0x3CC0278B, 0x8BDDE68F, 0x52FBA582, 0xE5E66486, 0x585B2BBE, 0xEF46EABA, + 0x3660A9B7, 0x817D68B3, 0x842D2FAD, 0x3330EEA9, 0xEA16ADA4, 0x5D0B6CA0, + 0x906D32D4, 0x2770F3D0, 0xFE56B0DD, 0x494B71D9, 0x4C1B36C7, 0xFB06F7C3, + 0x2220B4CE, 0x953D75CA, 0x28803AF2, 0x9F9DFBF6, 0x46BBB8FB, 0xF1A679FF, + 0xF4F63EE1, 0x43EBFFE5, 0x9ACDBCE8, 0x2DD07DEC, 0x77708634, 0xC06D4730, + 0x194B043D, 0xAE56C539, 0xAB068227, 0x1C1B4323, 0xC53D002E, 0x7220C12A, + 0xCF9D8E12, 0x78804F16, 0xA1A60C1B, 0x16BBCD1F, 0x13EB8A01, 0xA4F64B05, + 0x7DD00808, 0xCACDC90C, 0x07AB9778, 0xB0B6567C, 0x69901571, 0xDE8DD475, + 0xDBDD936B, 0x6CC0526F, 0xB5E61162, 0x02FBD066, 0xBF469F5E, 0x085B5E5A, + 0xD17D1D57, 0x6660DC53, 0x63309B4D, 0xD42D5A49, 0x0D0B1944, 0xBA16D840, + 0x97C6A5AC, 0x20DB64A8, 0xF9FD27A5, 0x4EE0E6A1, 0x4BB0A1BF, 0xFCAD60BB, + 0x258B23B6, 0x9296E2B2, 0x2F2BAD8A, 0x98366C8E, 0x41102F83, 0xF60DEE87, + 0xF35DA999, 0x4440689D, 0x9D662B90, 0x2A7BEA94, 0xE71DB4E0, 0x500075E4, + 0x892636E9, 0x3E3BF7ED, 0x3B6BB0F3, 0x8C7671F7, 0x555032FA, 0xE24DF3FE, + 0x5FF0BCC6, 0xE8ED7DC2, 0x31CB3ECF, 0x86D6FFCB, 0x8386B8D5, 0x349B79D1, + 0xEDBD3ADC, 0x5AA0FBD8, 0xEEE00C69, 0x59FDCD6D, 0x80DB8E60, 0x37C64F64, + 0x3296087A, 0x858BC97E, 0x5CAD8A73, 0xEBB04B77, 0x560D044F, 0xE110C54B, + 0x38368646, 0x8F2B4742, 0x8A7B005C, 0x3D66C158, 0xE4408255, 0x535D4351, + 0x9E3B1D25, 0x2926DC21, 0xF0009F2C, 0x471D5E28, 0x424D1936, 0xF550D832, + 0x2C769B3F, 0x9B6B5A3B, 0x26D61503, 0x91CBD407, 0x48ED970A, 0xFFF0560E, + 0xFAA01110, 0x4DBDD014, 0x949B9319, 0x2386521D, 0x0E562FF1, 0xB94BEEF5, + 0x606DADF8, 0xD7706CFC, 0xD2202BE2, 0x653DEAE6, 0xBC1BA9EB, 0x0B0668EF, + 0xB6BB27D7, 0x01A6E6D3, 0xD880A5DE, 0x6F9D64DA, 0x6ACD23C4, 0xDDD0E2C0, + 0x04F6A1CD, 0xB3EB60C9, 0x7E8D3EBD, 0xC990FFB9, 0x10B6BCB4, 0xA7AB7DB0, + 0xA2FB3AAE, 0x15E6FBAA, 0xCCC0B8A7, 0x7BDD79A3, 0xC660369B, 0x717DF79F, + 0xA85BB492, 0x1F467596, 0x1A163288, 0xAD0BF38C, 0x742DB081, 0xC3307185, + 0x99908A5D, 0x2E8D4B59, 0xF7AB0854, 0x40B6C950, 0x45E68E4E, 0xF2FB4F4A, + 0x2BDD0C47, 0x9CC0CD43, 0x217D827B, 0x9660437F, 0x4F460072, 0xF85BC176, + 0xFD0B8668, 0x4A16476C, 0x93300461, 0x242DC565, 0xE94B9B11, 0x5E565A15, + 0x87701918, 0x306DD81C, 0x353D9F02, 0x82205E06, 0x5B061D0B, 0xEC1BDC0F, + 0x51A69337, 0xE6BB5233, 0x3F9D113E, 0x8880D03A, 0x8DD09724, 0x3ACD5620, + 0xE3EB152D, 0x54F6D429, 0x7926A9C5, 0xCE3B68C1, 0x171D2BCC, 0xA000EAC8, + 0xA550ADD6, 0x124D6CD2, 0xCB6B2FDF, 0x7C76EEDB, 0xC1CBA1E3, 0x76D660E7, + 0xAFF023EA, 0x18EDE2EE, 0x1DBDA5F0, 0xAAA064F4, 0x738627F9, 0xC49BE6FD, + 0x09FDB889, 0xBEE0798D, 0x67C63A80, 0xD0DBFB84, 0xD58BBC9A, 0x62967D9E, + 0xBBB03E93, 0x0CADFF97, 0xB110B0AF, 0x060D71AB, 0xDF2B32A6, 0x6836F3A2, + 0x6D66B4BC, 0xDA7B75B8, 0x035D36B5, 0xB440F7B1, 0x00000001 + }, + [AV_CRC_32_IEEE_LE] = { + 0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, + 0xE963A535, 0x9E6495A3, 0x0EDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988, + 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91, 0x1DB71064, 0x6AB020F2, + 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7, + 0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, + 0xFA0F3D63, 0x8D080DF5, 0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172, + 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B, 0x35B5A8FA, 0x42B2986C, + 0xDBBBC9D6, 0xACBCF940, 0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59, + 0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, + 0xCFBA9599, 0xB8BDA50F, 0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, + 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D, 0x76DC4190, 0x01DB7106, + 0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433, + 0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB, 0x086D3D2D, + 0x91646C97, 0xE6635C01, 0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, + 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457, 0x65B0D9C6, 0x12B7E950, + 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65, + 0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, + 0xA4D1C46D, 0xD3D6F4FB, 0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, + 0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9, 0x5005713C, 0x270241AA, + 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F, + 0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, + 0xB7BD5C3B, 0xC0BA6CAD, 0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A, + 0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683, 0xE3630B12, 0x94643B84, + 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1, + 0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, + 0x196C3671, 0x6E6B06E7, 0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC, + 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5, 0xD6D6A3E8, 0xA1D1937E, + 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B, + 0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, + 0x316E8EEF, 0x4669BE79, 0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, + 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F, 0xC5BA3BBE, 0xB2BD0B28, + 0x2BB45A92, 0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D, + 0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F, + 0x72076785, 0x05005713, 0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, + 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21, 0x86D3D2D4, 0xF1D4E242, + 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777, + 0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C, 0x8F659EFF, 0xF862AE69, + 0x616BFFD3, 0x166CCF45, 0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, + 0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB, 0xAED16A4A, 0xD9D65ADC, + 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9, + 0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, + 0x54DE5729, 0x23D967BF, 0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, + 0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D, 0x00000001 + }, + [AV_CRC_16_ANSI_LE] = { + 0x0000, 0xC0C1, 0xC181, 0x0140, 0xC301, 0x03C0, 0x0280, 0xC241, + 0xC601, 0x06C0, 0x0780, 0xC741, 0x0500, 0xC5C1, 0xC481, 0x0440, + 0xCC01, 0x0CC0, 0x0D80, 0xCD41, 0x0F00, 0xCFC1, 0xCE81, 0x0E40, + 0x0A00, 0xCAC1, 0xCB81, 0x0B40, 0xC901, 0x09C0, 0x0880, 0xC841, + 0xD801, 0x18C0, 0x1980, 0xD941, 0x1B00, 0xDBC1, 0xDA81, 0x1A40, + 0x1E00, 0xDEC1, 0xDF81, 0x1F40, 0xDD01, 0x1DC0, 0x1C80, 0xDC41, + 0x1400, 0xD4C1, 0xD581, 0x1540, 0xD701, 0x17C0, 0x1680, 0xD641, + 0xD201, 0x12C0, 0x1380, 0xD341, 0x1100, 0xD1C1, 0xD081, 0x1040, + 0xF001, 0x30C0, 0x3180, 0xF141, 0x3300, 0xF3C1, 0xF281, 0x3240, + 0x3600, 0xF6C1, 0xF781, 0x3740, 0xF501, 0x35C0, 0x3480, 0xF441, + 0x3C00, 0xFCC1, 0xFD81, 0x3D40, 0xFF01, 0x3FC0, 0x3E80, 0xFE41, + 0xFA01, 0x3AC0, 0x3B80, 0xFB41, 0x3900, 0xF9C1, 0xF881, 0x3840, + 0x2800, 0xE8C1, 0xE981, 0x2940, 0xEB01, 0x2BC0, 0x2A80, 0xEA41, + 0xEE01, 0x2EC0, 0x2F80, 0xEF41, 0x2D00, 0xEDC1, 0xEC81, 0x2C40, + 0xE401, 0x24C0, 0x2580, 0xE541, 0x2700, 0xE7C1, 0xE681, 0x2640, + 0x2200, 0xE2C1, 0xE381, 0x2340, 0xE101, 0x21C0, 0x2080, 0xE041, + 0xA001, 0x60C0, 0x6180, 0xA141, 0x6300, 0xA3C1, 0xA281, 0x6240, + 0x6600, 0xA6C1, 0xA781, 0x6740, 0xA501, 0x65C0, 0x6480, 0xA441, + 0x6C00, 0xACC1, 0xAD81, 0x6D40, 0xAF01, 0x6FC0, 0x6E80, 0xAE41, + 0xAA01, 0x6AC0, 0x6B80, 0xAB41, 0x6900, 0xA9C1, 0xA881, 0x6840, + 0x7800, 0xB8C1, 0xB981, 0x7940, 0xBB01, 0x7BC0, 0x7A80, 0xBA41, + 0xBE01, 0x7EC0, 0x7F80, 0xBF41, 0x7D00, 0xBDC1, 0xBC81, 0x7C40, + 0xB401, 0x74C0, 0x7580, 0xB541, 0x7700, 0xB7C1, 0xB681, 0x7640, + 0x7200, 0xB2C1, 0xB381, 0x7340, 0xB101, 0x71C0, 0x7080, 0xB041, + 0x5000, 0x90C1, 0x9181, 0x5140, 0x9301, 0x53C0, 0x5280, 0x9241, + 0x9601, 0x56C0, 0x5780, 0x9741, 0x5500, 0x95C1, 0x9481, 0x5440, + 0x9C01, 0x5CC0, 0x5D80, 0x9D41, 0x5F00, 0x9FC1, 0x9E81, 0x5E40, + 0x5A00, 0x9AC1, 0x9B81, 0x5B40, 0x9901, 0x59C0, 0x5880, 0x9841, + 0x8801, 0x48C0, 0x4980, 0x8941, 0x4B00, 0x8BC1, 0x8A81, 0x4A40, + 0x4E00, 0x8EC1, 0x8F81, 0x4F40, 0x8D01, 0x4DC0, 0x4C80, 0x8C41, + 0x4400, 0x84C1, 0x8581, 0x4540, 0x8701, 0x47C0, 0x4680, 0x8641, + 0x8201, 0x42C0, 0x4380, 0x8341, 0x4100, 0x81C1, 0x8081, 0x4040, + 0x0001 + }, +}; +#else +#if CONFIG_SMALL +#define CRC_TABLE_SIZE 257 +#else +#define CRC_TABLE_SIZE 1024 +#endif +static AVCRC av_crc_table[AV_CRC_MAX][CRC_TABLE_SIZE]; + +#define DECLARE_CRC_INIT_TABLE_ONCE(id, le, bits, poly) \ +static AVOnce id ## _once_control = AV_ONCE_INIT; \ +static void id ## _init_table_once(void) \ +{ \ + av_assert0(av_crc_init(av_crc_table[id], le, bits, poly, sizeof(av_crc_table[id])) >= 0); \ +} + +#define CRC_INIT_TABLE_ONCE(id) ff_thread_once(&id ## _once_control, id ## _init_table_once) + +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_8_ATM, 0, 8, 0x07) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_8_EBU, 0, 8, 0x1D) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_16_ANSI, 0, 16, 0x8005) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_16_CCITT, 0, 16, 0x1021) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_24_IEEE, 0, 24, 0x864CFB) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_32_IEEE, 0, 32, 0x04C11DB7) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_32_IEEE_LE, 1, 32, 0xEDB88320) +DECLARE_CRC_INIT_TABLE_ONCE(AV_CRC_16_ANSI_LE, 1, 16, 0xA001) +#endif + +int av_crc_init(AVCRC *ctx, int le, int bits, uint32_t poly, int ctx_size) +{ + unsigned i, j; + uint32_t c; + + if (bits < 8 || bits > 32 || poly >= (1LL << bits)) + return AVERROR(EINVAL); + if (ctx_size != sizeof(AVCRC) * 257 && ctx_size != sizeof(AVCRC) * 1024) + return AVERROR(EINVAL); + + for (i = 0; i < 256; i++) { + if (le) { + for (c = i, j = 0; j < 8; j++) + c = (c >> 1) ^ (poly & (-(c & 1))); + ctx[i] = c; + } else { + for (c = i << 24, j = 0; j < 8; j++) + c = (c << 1) ^ ((poly << (32 - bits)) & (((int32_t) c) >> 31)); + ctx[i] = av_bswap32(c); + } + } + ctx[256] = 1; +#if !CONFIG_SMALL + if (ctx_size >= sizeof(AVCRC) * 1024) + for (i = 0; i < 256; i++) + for (j = 0; j < 3; j++) + ctx[256 * (j + 1) + i] = + (ctx[256 * j + i] >> 8) ^ ctx[ctx[256 * j + i] & 0xFF]; +#endif + + return 0; +} + +const AVCRC *av_crc_get_table(AVCRCId crc_id) +{ +#if !CONFIG_HARDCODED_TABLES + switch (crc_id) { + case AV_CRC_8_ATM: CRC_INIT_TABLE_ONCE(AV_CRC_8_ATM); break; + case AV_CRC_8_EBU: CRC_INIT_TABLE_ONCE(AV_CRC_8_EBU); break; + case AV_CRC_16_ANSI: CRC_INIT_TABLE_ONCE(AV_CRC_16_ANSI); break; + case AV_CRC_16_CCITT: CRC_INIT_TABLE_ONCE(AV_CRC_16_CCITT); break; + case AV_CRC_24_IEEE: CRC_INIT_TABLE_ONCE(AV_CRC_24_IEEE); break; + case AV_CRC_32_IEEE: CRC_INIT_TABLE_ONCE(AV_CRC_32_IEEE); break; + case AV_CRC_32_IEEE_LE: CRC_INIT_TABLE_ONCE(AV_CRC_32_IEEE_LE); break; + case AV_CRC_16_ANSI_LE: CRC_INIT_TABLE_ONCE(AV_CRC_16_ANSI_LE); break; + default: av_assert0(0); + } +#endif + return av_crc_table[crc_id]; +} + +uint32_t av_crc(const AVCRC *ctx, uint32_t crc, + const uint8_t *buffer, size_t length) +{ + const uint8_t *end = buffer + length; + +#if !CONFIG_SMALL + if (!ctx[256]) { + while (((intptr_t) buffer & 3) && buffer < end) + crc = ctx[((uint8_t) crc) ^ *buffer++] ^ (crc >> 8); + + while (buffer < end - 3) { + crc ^= av_le2ne32(*(const uint32_t *) buffer); buffer += 4; + crc = ctx[3 * 256 + ( crc & 0xFF)] ^ + ctx[2 * 256 + ((crc >> 8 ) & 0xFF)] ^ + ctx[1 * 256 + ((crc >> 16) & 0xFF)] ^ + ctx[0 * 256 + ((crc >> 24) )]; + } + } +#endif + while (buffer < end) + crc = ctx[((uint8_t) crc) ^ *buffer++] ^ (crc >> 8); + + return crc; +} diff --git a/media/ffvpx/libavutil/crc.h b/media/ffvpx/libavutil/crc.h new file mode 100644 index 0000000000..7f59812a18 --- /dev/null +++ b/media/ffvpx/libavutil/crc.h @@ -0,0 +1,102 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_crc32 + * Public header for CRC hash function implementation. + */ + +#ifndef AVUTIL_CRC_H +#define AVUTIL_CRC_H + +#include <stdint.h> +#include <stddef.h> +#include "attributes.h" + +/** + * @defgroup lavu_crc32 CRC + * @ingroup lavu_hash + * CRC (Cyclic Redundancy Check) hash function implementation. + * + * This module supports numerous CRC polynomials, in addition to the most + * widely used CRC-32-IEEE. See @ref AVCRCId for a list of available + * polynomials. + * + * @{ + */ + +typedef uint32_t AVCRC; + +typedef enum { + AV_CRC_8_ATM, + AV_CRC_16_ANSI, + AV_CRC_16_CCITT, + AV_CRC_32_IEEE, + AV_CRC_32_IEEE_LE, /*< reversed bitorder version of AV_CRC_32_IEEE */ + AV_CRC_16_ANSI_LE, /*< reversed bitorder version of AV_CRC_16_ANSI */ + AV_CRC_24_IEEE, + AV_CRC_8_EBU, + AV_CRC_MAX, /*< Not part of public API! Do not use outside libavutil. */ +}AVCRCId; + +/** + * Initialize a CRC table. + * @param ctx must be an array of size sizeof(AVCRC)*257 or sizeof(AVCRC)*1024 + * @param le If 1, the lowest bit represents the coefficient for the highest + * exponent of the corresponding polynomial (both for poly and + * actual CRC). + * If 0, you must swap the CRC parameter and the result of av_crc + * if you need the standard representation (can be simplified in + * most cases to e.g. bswap16): + * av_bswap32(crc << (32-bits)) + * @param bits number of bits for the CRC + * @param poly generator polynomial without the x**bits coefficient, in the + * representation as specified by le + * @param ctx_size size of ctx in bytes + * @return <0 on failure + */ +int av_crc_init(AVCRC *ctx, int le, int bits, uint32_t poly, int ctx_size); + +/** + * Get an initialized standard CRC table. + * @param crc_id ID of a standard CRC + * @return a pointer to the CRC table or NULL on failure + */ +const AVCRC *av_crc_get_table(AVCRCId crc_id); + +/** + * Calculate the CRC of a block. + * @param ctx initialized AVCRC array (see av_crc_init()) + * @param crc CRC of previous blocks if any or initial value for CRC + * @param buffer buffer whose CRC to calculate + * @param length length of the buffer + * @return CRC updated with the data from the given block + * + * @see av_crc_init() "le" parameter + */ +uint32_t av_crc(const AVCRC *ctx, uint32_t crc, + const uint8_t *buffer, size_t length) av_pure; + +/** + * @} + */ + +#endif /* AVUTIL_CRC_H */ diff --git a/media/ffvpx/libavutil/dict.c b/media/ffvpx/libavutil/dict.c new file mode 100644 index 0000000000..7f23d5336a --- /dev/null +++ b/media/ffvpx/libavutil/dict.c @@ -0,0 +1,294 @@ +/* + * copyright (c) 2009 Michael Niedermayer + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <inttypes.h> +#include <stdio.h> +#include <string.h> + +#include "avassert.h" +#include "avstring.h" +#include "dict.h" +#include "dict_internal.h" +#include "error.h" +#include "mem.h" +#include "time_internal.h" +#include "bprint.h" + +struct AVDictionary { + int count; + AVDictionaryEntry *elems; +}; + +int av_dict_count(const AVDictionary *m) +{ + return m ? m->count : 0; +} + +const AVDictionaryEntry *av_dict_iterate(const AVDictionary *m, + const AVDictionaryEntry *prev) +{ + int i = 0; + + if (!m) + return NULL; + + if (prev) + i = prev - m->elems + 1; + + av_assert2(i >= 0); + if (i >= m->count) + return NULL; + + return &m->elems[i]; +} + +AVDictionaryEntry *av_dict_get(const AVDictionary *m, const char *key, + const AVDictionaryEntry *prev, int flags) +{ + const AVDictionaryEntry *entry = prev; + unsigned int j; + + if (!key) + return NULL; + + while ((entry = av_dict_iterate(m, entry))) { + const char *s = entry->key; + if (flags & AV_DICT_MATCH_CASE) + for (j = 0; s[j] == key[j] && key[j]; j++) + ; + else + for (j = 0; av_toupper(s[j]) == av_toupper(key[j]) && key[j]; j++) + ; + if (key[j]) + continue; + if (s[j] && !(flags & AV_DICT_IGNORE_SUFFIX)) + continue; + return (AVDictionaryEntry *)entry; + } + return NULL; +} + +int av_dict_set(AVDictionary **pm, const char *key, const char *value, + int flags) +{ + AVDictionary *m = *pm; + AVDictionaryEntry *tag = NULL; + char *copy_key = NULL, *copy_value = NULL; + int err; + + if (flags & AV_DICT_DONT_STRDUP_VAL) + copy_value = (void *)value; + else if (value) + copy_value = av_strdup(value); + if (!key) { + err = AVERROR(EINVAL); + goto err_out; + } + if (!(flags & AV_DICT_MULTIKEY)) { + tag = av_dict_get(m, key, NULL, flags); + } + if (flags & AV_DICT_DONT_STRDUP_KEY) + copy_key = (void *)key; + else + copy_key = av_strdup(key); + if (!m) + m = *pm = av_mallocz(sizeof(*m)); + if (!m || !copy_key || (value && !copy_value)) + goto enomem; + + if (tag) { + if (flags & AV_DICT_DONT_OVERWRITE) { + av_free(copy_key); + av_free(copy_value); + return 0; + } + if (copy_value && flags & AV_DICT_APPEND) { + size_t oldlen = strlen(tag->value); + size_t new_part_len = strlen(copy_value); + size_t len = oldlen + new_part_len + 1; + char *newval = av_realloc(tag->value, len); + if (!newval) + goto enomem; + memcpy(newval + oldlen, copy_value, new_part_len + 1); + av_freep(©_value); + copy_value = newval; + } else + av_free(tag->value); + av_free(tag->key); + *tag = m->elems[--m->count]; + } else if (copy_value) { + AVDictionaryEntry *tmp = av_realloc_array(m->elems, + m->count + 1, sizeof(*m->elems)); + if (!tmp) + goto enomem; + m->elems = tmp; + } + if (copy_value) { + m->elems[m->count].key = copy_key; + m->elems[m->count].value = copy_value; + m->count++; + } else { + if (!m->count) { + av_freep(&m->elems); + av_freep(pm); + } + av_freep(©_key); + } + + return 0; + +enomem: + err = AVERROR(ENOMEM); +err_out: + if (m && !m->count) { + av_freep(&m->elems); + av_freep(pm); + } + av_free(copy_key); + av_free(copy_value); + return err; +} + +int av_dict_set_int(AVDictionary **pm, const char *key, int64_t value, + int flags) +{ + char valuestr[22]; + snprintf(valuestr, sizeof(valuestr), "%"PRId64, value); + flags &= ~AV_DICT_DONT_STRDUP_VAL; + return av_dict_set(pm, key, valuestr, flags); +} + +static int parse_key_value_pair(AVDictionary **pm, const char **buf, + const char *key_val_sep, const char *pairs_sep, + int flags) +{ + char *key = av_get_token(buf, key_val_sep); + char *val = NULL; + int ret; + + if (key && *key && strspn(*buf, key_val_sep)) { + (*buf)++; + val = av_get_token(buf, pairs_sep); + } + + if (key && *key && val && *val) + ret = av_dict_set(pm, key, val, flags); + else + ret = AVERROR(EINVAL); + + av_freep(&key); + av_freep(&val); + + return ret; +} + +int av_dict_parse_string(AVDictionary **pm, const char *str, + const char *key_val_sep, const char *pairs_sep, + int flags) +{ + int ret; + + if (!str) + return 0; + + /* ignore STRDUP flags */ + flags &= ~(AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL); + + while (*str) { + if ((ret = parse_key_value_pair(pm, &str, key_val_sep, pairs_sep, flags)) < 0) + return ret; + + if (*str) + str++; + } + + return 0; +} + +void av_dict_free(AVDictionary **pm) +{ + AVDictionary *m = *pm; + + if (m) { + while (m->count--) { + av_freep(&m->elems[m->count].key); + av_freep(&m->elems[m->count].value); + } + av_freep(&m->elems); + } + av_freep(pm); +} + +int av_dict_copy(AVDictionary **dst, const AVDictionary *src, int flags) +{ + const AVDictionaryEntry *t = NULL; + + while ((t = av_dict_iterate(src, t))) { + int ret = av_dict_set(dst, t->key, t->value, flags); + if (ret < 0) + return ret; + } + + return 0; +} + +int av_dict_get_string(const AVDictionary *m, char **buffer, + const char key_val_sep, const char pairs_sep) +{ + const AVDictionaryEntry *t = NULL; + AVBPrint bprint; + int cnt = 0; + char special_chars[] = {pairs_sep, key_val_sep, '\0'}; + + if (!buffer || pairs_sep == '\0' || key_val_sep == '\0' || pairs_sep == key_val_sep || + pairs_sep == '\\' || key_val_sep == '\\') + return AVERROR(EINVAL); + + if (!av_dict_count(m)) { + *buffer = av_strdup(""); + return *buffer ? 0 : AVERROR(ENOMEM); + } + + av_bprint_init(&bprint, 64, AV_BPRINT_SIZE_UNLIMITED); + while ((t = av_dict_iterate(m, t))) { + if (cnt++) + av_bprint_append_data(&bprint, &pairs_sep, 1); + av_bprint_escape(&bprint, t->key, special_chars, AV_ESCAPE_MODE_BACKSLASH, 0); + av_bprint_append_data(&bprint, &key_val_sep, 1); + av_bprint_escape(&bprint, t->value, special_chars, AV_ESCAPE_MODE_BACKSLASH, 0); + } + return av_bprint_finalize(&bprint, buffer); +} + +int avpriv_dict_set_timestamp(AVDictionary **dict, const char *key, int64_t timestamp) +{ + time_t seconds = timestamp / 1000000; + struct tm *ptm, tmbuf; + ptm = gmtime_r(&seconds, &tmbuf); + if (ptm) { + char buf[32]; + if (!strftime(buf, sizeof(buf), "%Y-%m-%dT%H:%M:%S", ptm)) + return AVERROR_EXTERNAL; + av_strlcatf(buf, sizeof(buf), ".%06dZ", (int)(timestamp % 1000000)); + return av_dict_set(dict, key, buf, 0); + } else { + return AVERROR_EXTERNAL; + } +} diff --git a/media/ffvpx/libavutil/dict.h b/media/ffvpx/libavutil/dict.h new file mode 100644 index 0000000000..713c9e361a --- /dev/null +++ b/media/ffvpx/libavutil/dict.h @@ -0,0 +1,241 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Public dictionary API. + * @deprecated + * AVDictionary is provided for compatibility with libav. It is both in + * implementation as well as API inefficient. It does not scale and is + * extremely slow with large dictionaries. + * It is recommended that new code uses our tree container from tree.c/h + * where applicable, which uses AVL trees to achieve O(log n) performance. + */ + +#ifndef AVUTIL_DICT_H +#define AVUTIL_DICT_H + +#include <stdint.h> + +/** + * @addtogroup lavu_dict AVDictionary + * @ingroup lavu_data + * + * @brief Simple key:value store + * + * @{ + * Dictionaries are used for storing key-value pairs. + * + * - To **create an AVDictionary**, simply pass an address of a NULL + * pointer to av_dict_set(). NULL can be used as an empty dictionary + * wherever a pointer to an AVDictionary is required. + * - To **insert an entry**, use av_dict_set(). + * - Use av_dict_get() to **retrieve an entry**. + * - To **iterate over all entries**, use av_dict_iterate(). + * - In order to **free the dictionary and all its contents**, use av_dict_free(). + * + @code + AVDictionary *d = NULL; // "create" an empty dictionary + AVDictionaryEntry *t = NULL; + + av_dict_set(&d, "foo", "bar", 0); // add an entry + + char *k = av_strdup("key"); // if your strings are already allocated, + char *v = av_strdup("value"); // you can avoid copying them like this + av_dict_set(&d, k, v, AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL); + + while ((t = av_dict_iterate(d, t))) { + <....> // iterate over all entries in d + } + av_dict_free(&d); + @endcode + */ + +/** + * @name AVDictionary Flags + * Flags that influence behavior of the matching of keys or insertion to the dictionary. + * @{ + */ +#define AV_DICT_MATCH_CASE 1 /**< Only get an entry with exact-case key match. Only relevant in av_dict_get(). */ +#define AV_DICT_IGNORE_SUFFIX 2 /**< Return first entry in a dictionary whose first part corresponds to the search key, + ignoring the suffix of the found key string. Only relevant in av_dict_get(). */ +#define AV_DICT_DONT_STRDUP_KEY 4 /**< Take ownership of a key that's been + allocated with av_malloc() or another memory allocation function. */ +#define AV_DICT_DONT_STRDUP_VAL 8 /**< Take ownership of a value that's been + allocated with av_malloc() or another memory allocation function. */ +#define AV_DICT_DONT_OVERWRITE 16 /**< Don't overwrite existing entries. */ +#define AV_DICT_APPEND 32 /**< If the entry already exists, append to it. Note that no + delimiter is added, the strings are simply concatenated. */ +#define AV_DICT_MULTIKEY 64 /**< Allow to store several equal keys in the dictionary */ +/** + * @} + */ + +typedef struct AVDictionaryEntry { + char *key; + char *value; +} AVDictionaryEntry; + +typedef struct AVDictionary AVDictionary; + +/** + * Get a dictionary entry with matching key. + * + * The returned entry key or value must not be changed, or it will + * cause undefined behavior. + * + * @param prev Set to the previous matching element to find the next. + * If set to NULL the first matching element is returned. + * @param key Matching key + * @param flags A collection of AV_DICT_* flags controlling how the + * entry is retrieved + * + * @return Found entry or NULL in case no matching entry was found in the dictionary + */ +AVDictionaryEntry *av_dict_get(const AVDictionary *m, const char *key, + const AVDictionaryEntry *prev, int flags); + +/** + * Iterate over a dictionary + * + * Iterates through all entries in the dictionary. + * + * @warning The returned AVDictionaryEntry key/value must not be changed. + * + * @warning As av_dict_set() invalidates all previous entries returned + * by this function, it must not be called while iterating over the dict. + * + * Typical usage: + * @code + * const AVDictionaryEntry *e = NULL; + * while ((e = av_dict_iterate(m, e))) { + * // ... + * } + * @endcode + * + * @param m The dictionary to iterate over + * @param prev Pointer to the previous AVDictionaryEntry, NULL initially + * + * @retval AVDictionaryEntry* The next element in the dictionary + * @retval NULL No more elements in the dictionary + */ +const AVDictionaryEntry *av_dict_iterate(const AVDictionary *m, + const AVDictionaryEntry *prev); + +/** + * Get number of entries in dictionary. + * + * @param m dictionary + * @return number of entries in dictionary + */ +int av_dict_count(const AVDictionary *m); + +/** + * Set the given entry in *pm, overwriting an existing entry. + * + * Note: If AV_DICT_DONT_STRDUP_KEY or AV_DICT_DONT_STRDUP_VAL is set, + * these arguments will be freed on error. + * + * @warning Adding a new entry to a dictionary invalidates all existing entries + * previously returned with av_dict_get() or av_dict_iterate(). + * + * @param pm Pointer to a pointer to a dictionary struct. If *pm is NULL + * a dictionary struct is allocated and put in *pm. + * @param key Entry key to add to *pm (will either be av_strduped or added as a new key depending on flags) + * @param value Entry value to add to *pm (will be av_strduped or added as a new key depending on flags). + * Passing a NULL value will cause an existing entry to be deleted. + * + * @return >= 0 on success otherwise an error code <0 + */ +int av_dict_set(AVDictionary **pm, const char *key, const char *value, int flags); + +/** + * Convenience wrapper for av_dict_set() that converts the value to a string + * and stores it. + * + * Note: If ::AV_DICT_DONT_STRDUP_KEY is set, key will be freed on error. + */ +int av_dict_set_int(AVDictionary **pm, const char *key, int64_t value, int flags); + +/** + * Parse the key/value pairs list and add the parsed entries to a dictionary. + * + * In case of failure, all the successfully set entries are stored in + * *pm. You may need to manually free the created dictionary. + * + * @param key_val_sep A 0-terminated list of characters used to separate + * key from value + * @param pairs_sep A 0-terminated list of characters used to separate + * two pairs from each other + * @param flags Flags to use when adding to the dictionary. + * ::AV_DICT_DONT_STRDUP_KEY and ::AV_DICT_DONT_STRDUP_VAL + * are ignored since the key/value tokens will always + * be duplicated. + * + * @return 0 on success, negative AVERROR code on failure + */ +int av_dict_parse_string(AVDictionary **pm, const char *str, + const char *key_val_sep, const char *pairs_sep, + int flags); + +/** + * Copy entries from one AVDictionary struct into another. + * + * @note Metadata is read using the ::AV_DICT_IGNORE_SUFFIX flag + * + * @param dst Pointer to a pointer to a AVDictionary struct to copy into. If *dst is NULL, + * this function will allocate a struct for you and put it in *dst + * @param src Pointer to the source AVDictionary struct to copy items from. + * @param flags Flags to use when setting entries in *dst + * + * @return 0 on success, negative AVERROR code on failure. If dst was allocated + * by this function, callers should free the associated memory. + */ +int av_dict_copy(AVDictionary **dst, const AVDictionary *src, int flags); + +/** + * Free all the memory allocated for an AVDictionary struct + * and all keys and values. + */ +void av_dict_free(AVDictionary **m); + +/** + * Get dictionary entries as a string. + * + * Create a string containing dictionary's entries. + * Such string may be passed back to av_dict_parse_string(). + * @note String is escaped with backslashes ('\'). + * + * @warning Separators cannot be neither '\\' nor '\0'. They also cannot be the same. + * + * @param[in] m The dictionary + * @param[out] buffer Pointer to buffer that will be allocated with string containg entries. + * Buffer must be freed by the caller when is no longer needed. + * @param[in] key_val_sep Character used to separate key from value + * @param[in] pairs_sep Character used to separate two pairs from each other + * + * @return >= 0 on success, negative on error + */ +int av_dict_get_string(const AVDictionary *m, char **buffer, + const char key_val_sep, const char pairs_sep); + +/** + * @} + */ + +#endif /* AVUTIL_DICT_H */ diff --git a/media/ffvpx/libavutil/dict_internal.h b/media/ffvpx/libavutil/dict_internal.h new file mode 100644 index 0000000000..6d5b0dc2b0 --- /dev/null +++ b/media/ffvpx/libavutil/dict_internal.h @@ -0,0 +1,37 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_DICT_INTERNAL_H +#define AVUTIL_DICT_INTERNAL_H + +#include <stdint.h> + +#include "dict.h" + +/** + * Set a dictionary value to an ISO-8601 compliant timestamp string. + * + * @param dict pointer to a pointer to a dictionary struct. If *dict is NULL + * a dictionary struct is allocated and put in *dict. + * @param key metadata key + * @param timestamp unix timestamp in microseconds + * @return <0 on error + */ +int avpriv_dict_set_timestamp(AVDictionary **dict, const char *key, int64_t timestamp); + +#endif /* AVUTIL_DICT_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/dynarray.h b/media/ffvpx/libavutil/dynarray.h new file mode 100644 index 0000000000..3a7e146422 --- /dev/null +++ b/media/ffvpx/libavutil/dynarray.h @@ -0,0 +1,70 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public License + * as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public License + * along with FFmpeg; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_DYNARRAY_H +#define AVUTIL_DYNARRAY_H + +#include "log.h" +#include "mem.h" + +/** + * Add an element to a dynamic array. + * + * The array is reallocated when its number of elements reaches powers of 2. + * Therefore, the amortized cost of adding an element is constant. + * + * In case of success, the pointer to the array is updated in order to + * point to the new grown array, and the size is incremented. + * + * @param av_size_max maximum size of the array, usually the MAX macro of + * the type of the size + * @param av_elt_size size of the elements in the array, in bytes + * @param av_array pointer to the array, must be a lvalue + * @param av_size size of the array, must be an integer lvalue + * @param av_success statement to execute on success; at this point, the + * size variable is not yet incremented + * @param av_failure statement to execute on failure; if this happens, the + * array and size are not changed; the statement can end + * with a return or a goto + */ +#define FF_DYNARRAY_ADD(av_size_max, av_elt_size, av_array, av_size, \ + av_success, av_failure) \ + do { \ + size_t av_size_new = (av_size); \ + if (!((av_size) & ((av_size) - 1))) { \ + av_size_new = (av_size) ? (av_size) << 1 : 1; \ + if (av_size_new > (av_size_max) / (av_elt_size)) { \ + av_size_new = 0; \ + } else { \ + void *av_array_new = \ + av_realloc((av_array), av_size_new * (av_elt_size)); \ + if (!av_array_new) \ + av_size_new = 0; \ + else \ + (av_array) = av_array_new; \ + } \ + } \ + if (av_size_new) { \ + { av_success } \ + (av_size)++; \ + } else { \ + av_failure \ + } \ + } while (0) + +#endif /* AVUTIL_DYNARRAY_H */ diff --git a/media/ffvpx/libavutil/emms.h b/media/ffvpx/libavutil/emms.h new file mode 100644 index 0000000000..4ec0127457 --- /dev/null +++ b/media/ffvpx/libavutil/emms.h @@ -0,0 +1,66 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_EMMS_H +#define AVUTIL_EMMS_H + +#include "config.h" +#include "libavutil/attributes.h" + +#if ARCH_X86 + +void avpriv_emms_asm(void); + +#if HAVE_MMX_INLINE +#ifndef __MMX__ +#include "libavutil/cpu.h" +#endif + +# define emms_c emms_c +/** + * Empty mmx state. + * this must be called between any dsp function and float/double code. + * for example sin(); dsp->idct_put(); emms_c(); cos() + * Note, *alloc() and *free() also use float code in some libc implementations + * thus this also applies to them or any function using them. + */ +static av_always_inline void emms_c(void) +{ +/* Some inlined functions may also use mmx instructions regardless of + * runtime cpuflags. With that in mind, we unconditionally empty the + * mmx state if the target cpu chosen at configure time supports it. + */ +#if !defined(__MMX__) + if(av_get_cpu_flags() & AV_CPU_FLAG_MMX) +#endif + __asm__ volatile ("emms" ::: "memory"); +} +#elif HAVE_MMX && HAVE_MM_EMPTY +# include <mmintrin.h> +# define emms_c _mm_empty +#elif HAVE_MMX_EXTERNAL +# define emms_c avpriv_emms_asm +#endif /* HAVE_MMX_INLINE */ + +#endif /* ARCH_X86 */ + +#ifndef emms_c +# define emms_c() do {} while(0) +#endif + +#endif /* AVUTIL_EMMS_H */ diff --git a/media/ffvpx/libavutil/error.c b/media/ffvpx/libavutil/error.c new file mode 100644 index 0000000000..938a8bc000 --- /dev/null +++ b/media/ffvpx/libavutil/error.c @@ -0,0 +1,132 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#undef _GNU_SOURCE +#define _XOPEN_SOURCE 600 /* XSI-compliant version of strerror_r */ +#include <stdio.h> +#include <string.h> +#include "config.h" +#include "avstring.h" +#include "error.h" +#include "macros.h" + +struct error_entry { + int num; + const char *tag; + const char *str; +}; + +#define ERROR_TAG(tag) AVERROR_##tag, #tag +#define EERROR_TAG(tag) AVERROR(tag), #tag +#define AVERROR_INPUT_AND_OUTPUT_CHANGED (AVERROR_INPUT_CHANGED | AVERROR_OUTPUT_CHANGED) +static const struct error_entry error_entries[] = { + { ERROR_TAG(BSF_NOT_FOUND), "Bitstream filter not found" }, + { ERROR_TAG(BUG), "Internal bug, should not have happened" }, + { ERROR_TAG(BUG2), "Internal bug, should not have happened" }, + { ERROR_TAG(BUFFER_TOO_SMALL), "Buffer too small" }, + { ERROR_TAG(DECODER_NOT_FOUND), "Decoder not found" }, + { ERROR_TAG(DEMUXER_NOT_FOUND), "Demuxer not found" }, + { ERROR_TAG(ENCODER_NOT_FOUND), "Encoder not found" }, + { ERROR_TAG(EOF), "End of file" }, + { ERROR_TAG(EXIT), "Immediate exit requested" }, + { ERROR_TAG(EXTERNAL), "Generic error in an external library" }, + { ERROR_TAG(FILTER_NOT_FOUND), "Filter not found" }, + { ERROR_TAG(INPUT_CHANGED), "Input changed" }, + { ERROR_TAG(INVALIDDATA), "Invalid data found when processing input" }, + { ERROR_TAG(MUXER_NOT_FOUND), "Muxer not found" }, + { ERROR_TAG(OPTION_NOT_FOUND), "Option not found" }, + { ERROR_TAG(OUTPUT_CHANGED), "Output changed" }, + { ERROR_TAG(PATCHWELCOME), "Not yet implemented in FFmpeg, patches welcome" }, + { ERROR_TAG(PROTOCOL_NOT_FOUND), "Protocol not found" }, + { ERROR_TAG(STREAM_NOT_FOUND), "Stream not found" }, + { ERROR_TAG(UNKNOWN), "Unknown error occurred" }, + { ERROR_TAG(EXPERIMENTAL), "Experimental feature" }, + { ERROR_TAG(INPUT_AND_OUTPUT_CHANGED), "Input and output changed" }, + { ERROR_TAG(HTTP_BAD_REQUEST), "Server returned 400 Bad Request" }, + { ERROR_TAG(HTTP_UNAUTHORIZED), "Server returned 401 Unauthorized (authorization failed)" }, + { ERROR_TAG(HTTP_FORBIDDEN), "Server returned 403 Forbidden (access denied)" }, + { ERROR_TAG(HTTP_NOT_FOUND), "Server returned 404 Not Found" }, + { ERROR_TAG(HTTP_OTHER_4XX), "Server returned 4XX Client Error, but not one of 40{0,1,3,4}" }, + { ERROR_TAG(HTTP_SERVER_ERROR), "Server returned 5XX Server Error reply" }, +#if !HAVE_STRERROR_R + { EERROR_TAG(E2BIG), "Argument list too long" }, + { EERROR_TAG(EACCES), "Permission denied" }, + { EERROR_TAG(EAGAIN), "Resource temporarily unavailable" }, + { EERROR_TAG(EBADF), "Bad file descriptor" }, + { EERROR_TAG(EBUSY), "Device or resource busy" }, + { EERROR_TAG(ECHILD), "No child processes" }, + { EERROR_TAG(EDEADLK), "Resource deadlock avoided" }, + { EERROR_TAG(EDOM), "Numerical argument out of domain" }, + { EERROR_TAG(EEXIST), "File exists" }, + { EERROR_TAG(EFAULT), "Bad address" }, + { EERROR_TAG(EFBIG), "File too large" }, + { EERROR_TAG(EILSEQ), "Illegal byte sequence" }, + { EERROR_TAG(EINTR), "Interrupted system call" }, + { EERROR_TAG(EINVAL), "Invalid argument" }, + { EERROR_TAG(EIO), "I/O error" }, + { EERROR_TAG(EISDIR), "Is a directory" }, + { EERROR_TAG(EMFILE), "Too many open files" }, + { EERROR_TAG(EMLINK), "Too many links" }, + { EERROR_TAG(ENAMETOOLONG), "File name too long" }, + { EERROR_TAG(ENFILE), "Too many open files in system" }, + { EERROR_TAG(ENODEV), "No such device" }, + { EERROR_TAG(ENOENT), "No such file or directory" }, + { EERROR_TAG(ENOEXEC), "Exec format error" }, + { EERROR_TAG(ENOLCK), "No locks available" }, + { EERROR_TAG(ENOMEM), "Cannot allocate memory" }, + { EERROR_TAG(ENOSPC), "No space left on device" }, + { EERROR_TAG(ENOSYS), "Function not implemented" }, + { EERROR_TAG(ENOTDIR), "Not a directory" }, + { EERROR_TAG(ENOTEMPTY), "Directory not empty" }, + { EERROR_TAG(ENOTTY), "Inappropriate I/O control operation" }, + { EERROR_TAG(ENXIO), "No such device or address" }, + { EERROR_TAG(EPERM), "Operation not permitted" }, + { EERROR_TAG(EPIPE), "Broken pipe" }, + { EERROR_TAG(ERANGE), "Result too large" }, + { EERROR_TAG(EROFS), "Read-only file system" }, + { EERROR_TAG(ESPIPE), "Illegal seek" }, + { EERROR_TAG(ESRCH), "No such process" }, + { EERROR_TAG(EXDEV), "Cross-device link" }, +#endif +}; + +int av_strerror(int errnum, char *errbuf, size_t errbuf_size) +{ + int ret = 0, i; + const struct error_entry *entry = NULL; + + for (i = 0; i < FF_ARRAY_ELEMS(error_entries); i++) { + if (errnum == error_entries[i].num) { + entry = &error_entries[i]; + break; + } + } + if (entry) { + av_strlcpy(errbuf, entry->str, errbuf_size); + } else { +#if HAVE_STRERROR_R + ret = AVERROR(strerror_r(AVUNERROR(errnum), errbuf, errbuf_size)); +#else + ret = -1; +#endif + if (ret < 0) + snprintf(errbuf, errbuf_size, "Error number %d occurred", errnum); + } + + return ret; +} diff --git a/media/ffvpx/libavutil/error.h b/media/ffvpx/libavutil/error.h new file mode 100644 index 0000000000..0d3269aa6d --- /dev/null +++ b/media/ffvpx/libavutil/error.h @@ -0,0 +1,128 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * error code definitions + */ + +#ifndef AVUTIL_ERROR_H +#define AVUTIL_ERROR_H + +#include <errno.h> +#include <stddef.h> + +#include "macros.h" + +/** + * @addtogroup lavu_error + * + * @{ + */ + + +/* error handling */ +#if EDOM > 0 +#define AVERROR(e) (-(e)) ///< Returns a negative error code from a POSIX error code, to return from library functions. +#define AVUNERROR(e) (-(e)) ///< Returns a POSIX error code from a library function error return value. +#else +/* Some platforms have E* and errno already negated. */ +#define AVERROR(e) (e) +#define AVUNERROR(e) (e) +#endif + +#define FFERRTAG(a, b, c, d) (-(int)MKTAG(a, b, c, d)) + +#define AVERROR_BSF_NOT_FOUND FFERRTAG(0xF8,'B','S','F') ///< Bitstream filter not found +#define AVERROR_BUG FFERRTAG( 'B','U','G','!') ///< Internal bug, also see AVERROR_BUG2 +#define AVERROR_BUFFER_TOO_SMALL FFERRTAG( 'B','U','F','S') ///< Buffer too small +#define AVERROR_DECODER_NOT_FOUND FFERRTAG(0xF8,'D','E','C') ///< Decoder not found +#define AVERROR_DEMUXER_NOT_FOUND FFERRTAG(0xF8,'D','E','M') ///< Demuxer not found +#define AVERROR_ENCODER_NOT_FOUND FFERRTAG(0xF8,'E','N','C') ///< Encoder not found +#define AVERROR_EOF FFERRTAG( 'E','O','F',' ') ///< End of file +#define AVERROR_EXIT FFERRTAG( 'E','X','I','T') ///< Immediate exit was requested; the called function should not be restarted +#define AVERROR_EXTERNAL FFERRTAG( 'E','X','T',' ') ///< Generic error in an external library +#define AVERROR_FILTER_NOT_FOUND FFERRTAG(0xF8,'F','I','L') ///< Filter not found +#define AVERROR_INVALIDDATA FFERRTAG( 'I','N','D','A') ///< Invalid data found when processing input +#define AVERROR_MUXER_NOT_FOUND FFERRTAG(0xF8,'M','U','X') ///< Muxer not found +#define AVERROR_OPTION_NOT_FOUND FFERRTAG(0xF8,'O','P','T') ///< Option not found +#define AVERROR_PATCHWELCOME FFERRTAG( 'P','A','W','E') ///< Not yet implemented in FFmpeg, patches welcome +#define AVERROR_PROTOCOL_NOT_FOUND FFERRTAG(0xF8,'P','R','O') ///< Protocol not found + +#define AVERROR_STREAM_NOT_FOUND FFERRTAG(0xF8,'S','T','R') ///< Stream not found +/** + * This is semantically identical to AVERROR_BUG + * it has been introduced in Libav after our AVERROR_BUG and with a modified value. + */ +#define AVERROR_BUG2 FFERRTAG( 'B','U','G',' ') +#define AVERROR_UNKNOWN FFERRTAG( 'U','N','K','N') ///< Unknown error, typically from an external library +#define AVERROR_EXPERIMENTAL (-0x2bb2afa8) ///< Requested feature is flagged experimental. Set strict_std_compliance if you really want to use it. +#define AVERROR_INPUT_CHANGED (-0x636e6701) ///< Input changed between calls. Reconfiguration is required. (can be OR-ed with AVERROR_OUTPUT_CHANGED) +#define AVERROR_OUTPUT_CHANGED (-0x636e6702) ///< Output changed between calls. Reconfiguration is required. (can be OR-ed with AVERROR_INPUT_CHANGED) +/* HTTP & RTSP errors */ +#define AVERROR_HTTP_BAD_REQUEST FFERRTAG(0xF8,'4','0','0') +#define AVERROR_HTTP_UNAUTHORIZED FFERRTAG(0xF8,'4','0','1') +#define AVERROR_HTTP_FORBIDDEN FFERRTAG(0xF8,'4','0','3') +#define AVERROR_HTTP_NOT_FOUND FFERRTAG(0xF8,'4','0','4') +#define AVERROR_HTTP_OTHER_4XX FFERRTAG(0xF8,'4','X','X') +#define AVERROR_HTTP_SERVER_ERROR FFERRTAG(0xF8,'5','X','X') + +#define AV_ERROR_MAX_STRING_SIZE 64 + +/** + * Put a description of the AVERROR code errnum in errbuf. + * In case of failure the global variable errno is set to indicate the + * error. Even in case of failure av_strerror() will print a generic + * error message indicating the errnum provided to errbuf. + * + * @param errnum error code to describe + * @param errbuf buffer to which description is written + * @param errbuf_size the size in bytes of errbuf + * @return 0 on success, a negative value if a description for errnum + * cannot be found + */ +int av_strerror(int errnum, char *errbuf, size_t errbuf_size); + +/** + * Fill the provided buffer with a string containing an error string + * corresponding to the AVERROR code errnum. + * + * @param errbuf a buffer + * @param errbuf_size size in bytes of errbuf + * @param errnum error code to describe + * @return the buffer in input, filled with the error description + * @see av_strerror() + */ +static inline char *av_make_error_string(char *errbuf, size_t errbuf_size, int errnum) +{ + av_strerror(errnum, errbuf, errbuf_size); + return errbuf; +} + +/** + * Convenience macro, the return value should be used only directly in + * function arguments but never stand-alone. + */ +#define av_err2str(errnum) \ + av_make_error_string((char[AV_ERROR_MAX_STRING_SIZE]){0}, AV_ERROR_MAX_STRING_SIZE, errnum) + +/** + * @} + */ + +#endif /* AVUTIL_ERROR_H */ diff --git a/media/ffvpx/libavutil/eval.c b/media/ffvpx/libavutil/eval.c new file mode 100644 index 0000000000..95adb51609 --- /dev/null +++ b/media/ffvpx/libavutil/eval.c @@ -0,0 +1,820 @@ +/* + * Copyright (c) 2002-2006 Michael Niedermayer <michaelni@gmx.at> + * Copyright (c) 2006 Oded Shimon <ods15@ods15.dyndns.org> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * simple arithmetic expression evaluator. + * + * see http://joe.hotchkiss.com/programming/eval/eval.html + */ + +#include <float.h> +#include "attributes.h" +#include "avutil.h" +#include "common.h" +#include "eval.h" +#include "ffmath.h" +#include "internal.h" +#include "log.h" +#include "mathematics.h" +#include "sfc64.h" +#include "fftime.h" +#include "avstring.h" +#include "timer.h" +#include "reverse.h" + +typedef struct Parser { + const AVClass *class; + int stack_index; + char *s; + const double *const_values; + const char * const *const_names; // NULL terminated + double (* const *funcs1)(void *, double a); // NULL terminated + const char * const *func1_names; // NULL terminated + double (* const *funcs2)(void *, double a, double b); // NULL terminated + const char * const *func2_names; // NULL terminated + void *opaque; + int log_offset; + void *log_ctx; +#define VARS 10 + double *var; + FFSFC64 *prng_state; +} Parser; + +static const AVClass eval_class = { + .class_name = "Eval", + .item_name = av_default_item_name, + .option = NULL, + .version = LIBAVUTIL_VERSION_INT, + .log_level_offset_offset = offsetof(Parser, log_offset), + .parent_log_context_offset = offsetof(Parser, log_ctx), +}; + +static const struct { + double bin_val; + double dec_val; + int8_t exp; +} si_prefixes['z' - 'E' + 1] = { + ['y'-'E']= { 8.271806125530276749e-25, 1e-24, -24 }, + ['z'-'E']= { 8.4703294725430034e-22, 1e-21, -21 }, + ['a'-'E']= { 8.6736173798840355e-19, 1e-18, -18 }, + ['f'-'E']= { 8.8817841970012523e-16, 1e-15, -15 }, + ['p'-'E']= { 9.0949470177292824e-13, 1e-12, -12 }, + ['n'-'E']= { 9.3132257461547852e-10, 1e-9, -9 }, + ['u'-'E']= { 9.5367431640625e-7, 1e-6, -6 }, + ['m'-'E']= { 9.765625e-4, 1e-3, -3 }, + ['c'-'E']= { 9.8431332023036951e-3, 1e-2, -2 }, + ['d'-'E']= { 9.921256574801246e-2, 1e-1, -1 }, + ['h'-'E']= { 1.0159366732596479e2, 1e2, 2 }, + ['k'-'E']= { 1.024e3, 1e3, 3 }, + ['K'-'E']= { 1.024e3, 1e3, 3 }, + ['M'-'E']= { 1.048576e6, 1e6, 6 }, + ['G'-'E']= { 1.073741824e9, 1e9, 9 }, + ['T'-'E']= { 1.099511627776e12, 1e12, 12 }, + ['P'-'E']= { 1.125899906842624e15, 1e15, 15 }, + ['E'-'E']= { 1.152921504606847e18, 1e18, 18 }, + ['Z'-'E']= { 1.1805916207174113e21, 1e21, 21 }, + ['Y'-'E']= { 1.2089258196146292e24, 1e24, 24 }, +}; + +static const struct { + const char *name; + double value; +} constants[] = { + { "E", M_E }, + { "PI", M_PI }, + { "PHI", M_PHI }, + { "QP2LAMBDA", FF_QP2LAMBDA }, +}; + +double av_strtod(const char *numstr, char **tail) +{ + double d; + char *next; + if(numstr[0]=='0' && (numstr[1]|0x20)=='x') { + d = strtoul(numstr, &next, 16); + } else + d = strtod(numstr, &next); + /* if parsing succeeded, check for and interpret postfixes */ + if (next!=numstr) { + if (next[0] == 'd' && next[1] == 'B') { + /* treat dB as decibels instead of decibytes */ + d = ff_exp10(d / 20); + next += 2; + } else if (*next >= 'E' && *next <= 'z') { + int e= si_prefixes[*next - 'E'].exp; + if (e) { + if (next[1] == 'i') { + d*= si_prefixes[*next - 'E'].bin_val; + next+=2; + } else { + d*= si_prefixes[*next - 'E'].dec_val; + next++; + } + } + } + + if (*next=='B') { + d*=8; + next++; + } + } + /* if requested, fill in tail with the position after the last parsed + character */ + if (tail) + *tail = next; + return d; +} + +#define IS_IDENTIFIER_CHAR(c) ((c) - '0' <= 9U || (c) - 'a' <= 25U || (c) - 'A' <= 25U || (c) == '_') + +static int strmatch(const char *s, const char *prefix) +{ + int i; + for (i=0; prefix[i]; i++) { + if (prefix[i] != s[i]) return 0; + } + /* return 1 only if the s identifier is terminated */ + return !IS_IDENTIFIER_CHAR(s[i]); +} + +struct AVExpr { + enum { + e_value, e_const, e_func0, e_func1, e_func2, + e_squish, e_gauss, e_ld, e_isnan, e_isinf, + e_mod, e_max, e_min, e_eq, e_gt, e_gte, e_lte, e_lt, + e_pow, e_mul, e_div, e_add, + e_last, e_st, e_while, e_taylor, e_root, e_floor, e_ceil, e_trunc, e_round, + e_sqrt, e_not, e_random, e_hypot, e_gcd, + e_if, e_ifnot, e_print, e_bitand, e_bitor, e_between, e_clip, e_atan2, e_lerp, + e_sgn, e_randomi + } type; + double value; // is sign in other types + int const_index; + union { + double (*func0)(double); + double (*func1)(void *, double); + double (*func2)(void *, double, double); + } a; + struct AVExpr *param[3]; + double *var; + FFSFC64 *prng_state; +}; + +static double etime(double v) +{ + return av_gettime() * 0.000001; +} + +static double eval_expr(Parser *p, AVExpr *e) +{ + switch (e->type) { + case e_value: return e->value; + case e_const: return e->value * p->const_values[e->const_index]; + case e_func0: return e->value * e->a.func0(eval_expr(p, e->param[0])); + case e_func1: return e->value * e->a.func1(p->opaque, eval_expr(p, e->param[0])); + case e_func2: return e->value * e->a.func2(p->opaque, eval_expr(p, e->param[0]), eval_expr(p, e->param[1])); + case e_squish: return 1/(1+exp(4*eval_expr(p, e->param[0]))); + case e_gauss: { double d = eval_expr(p, e->param[0]); return exp(-d*d/2)/sqrt(2*M_PI); } + case e_ld: return e->value * p->var[av_clip(eval_expr(p, e->param[0]), 0, VARS-1)]; + case e_isnan: return e->value * !!isnan(eval_expr(p, e->param[0])); + case e_isinf: return e->value * !!isinf(eval_expr(p, e->param[0])); + case e_floor: return e->value * floor(eval_expr(p, e->param[0])); + case e_ceil : return e->value * ceil (eval_expr(p, e->param[0])); + case e_trunc: return e->value * trunc(eval_expr(p, e->param[0])); + case e_round: return e->value * round(eval_expr(p, e->param[0])); + case e_sgn: return e->value * FFDIFFSIGN(eval_expr(p, e->param[0]), 0); + case e_sqrt: return e->value * sqrt (eval_expr(p, e->param[0])); + case e_not: return e->value * (eval_expr(p, e->param[0]) == 0); + case e_if: return e->value * (eval_expr(p, e->param[0]) ? eval_expr(p, e->param[1]) : + e->param[2] ? eval_expr(p, e->param[2]) : 0); + case e_ifnot: return e->value * (!eval_expr(p, e->param[0]) ? eval_expr(p, e->param[1]) : + e->param[2] ? eval_expr(p, e->param[2]) : 0); + case e_clip: { + double x = eval_expr(p, e->param[0]); + double min = eval_expr(p, e->param[1]), max = eval_expr(p, e->param[2]); + if (isnan(min) || isnan(max) || isnan(x) || min > max) + return NAN; + return e->value * av_clipd(eval_expr(p, e->param[0]), min, max); + } + case e_between: { + double d = eval_expr(p, e->param[0]); + return e->value * (d >= eval_expr(p, e->param[1]) && + d <= eval_expr(p, e->param[2])); + } + case e_lerp: { + double v0 = eval_expr(p, e->param[0]); + double v1 = eval_expr(p, e->param[1]); + double f = eval_expr(p, e->param[2]); + return v0 + (v1 - v0) * f; + } + case e_print: { + double x = eval_expr(p, e->param[0]); + int level = e->param[1] ? av_clip(eval_expr(p, e->param[1]), INT_MIN, INT_MAX) : AV_LOG_INFO; + av_log(p, level, "%f\n", x); + return x; + } + +#define COMPUTE_NEXT_RANDOM() \ + int idx = av_clip(eval_expr(p, e->param[0]), 0, VARS-1); \ + FFSFC64 *s = p->prng_state + idx; \ + uint64_t r; \ + \ + if (!s->counter) { \ + r = isnan(p->var[idx]) ? 0 : p->var[idx]; \ + ff_sfc64_init(s, r, r, r, 12); \ + } \ + r = ff_sfc64_get(s); \ + p->var[idx] = r; \ + + case e_random: { + COMPUTE_NEXT_RANDOM(); + return r * (1.0/UINT64_MAX); + } + case e_randomi: { + double min = eval_expr(p, e->param[1]); + double max = eval_expr(p, e->param[2]); + COMPUTE_NEXT_RANDOM(); + return min + (max - min) * r / UINT64_MAX; + } + case e_while: { + double d = NAN; + while (eval_expr(p, e->param[0])) + d=eval_expr(p, e->param[1]); + return d; + } + case e_taylor: { + double t = 1, d = 0, v; + double x = eval_expr(p, e->param[1]); + int id = e->param[2] ? av_clip(eval_expr(p, e->param[2]), 0, VARS-1) : 0; + int i; + double var0 = p->var[id]; + for(i=0; i<1000; i++) { + double ld = d; + p->var[id] = i; + v = eval_expr(p, e->param[0]); + d += t*v; + if(ld==d && v) + break; + t *= x / (i+1); + } + p->var[id] = var0; + return d; + } + case e_root: { + int i, j; + double low = -1, high = -1, v, low_v = -DBL_MAX, high_v = DBL_MAX; + double var0 = p->var[0]; + double x_max = eval_expr(p, e->param[1]); + for(i=-1; i<1024; i++) { + if(i<255) { + p->var[0] = ff_reverse[i&255]*x_max/255; + } else { + p->var[0] = x_max*pow(0.9, i-255); + if (i&1) p->var[0] *= -1; + if (i&2) p->var[0] += low; + else p->var[0] += high; + } + v = eval_expr(p, e->param[0]); + if (v<=0 && v>low_v) { + low = p->var[0]; + low_v = v; + } + if (v>=0 && v<high_v) { + high = p->var[0]; + high_v = v; + } + if (low>=0 && high>=0){ + for (j=0; j<1000; j++) { + p->var[0] = (low+high)*0.5; + if (low == p->var[0] || high == p->var[0]) + break; + v = eval_expr(p, e->param[0]); + if (v<=0) low = p->var[0]; + if (v>=0) high= p->var[0]; + if (isnan(v)) { + low = high = v; + break; + } + } + break; + } + } + p->var[0] = var0; + return -low_v<high_v ? low : high; + } + default: { + double d = eval_expr(p, e->param[0]); + double d2 = eval_expr(p, e->param[1]); + switch (e->type) { + case e_mod: return e->value * (d - floor(d2 ? d / d2 : d * INFINITY) * d2); + case e_gcd: return e->value * av_gcd(d,d2); + case e_max: return e->value * (d > d2 ? d : d2); + case e_min: return e->value * (d < d2 ? d : d2); + case e_eq: return e->value * (d == d2 ? 1.0 : 0.0); + case e_gt: return e->value * (d > d2 ? 1.0 : 0.0); + case e_gte: return e->value * (d >= d2 ? 1.0 : 0.0); + case e_lt: return e->value * (d < d2 ? 1.0 : 0.0); + case e_lte: return e->value * (d <= d2 ? 1.0 : 0.0); + case e_pow: return e->value * pow(d, d2); + case e_mul: return e->value * (d * d2); + case e_div: return e->value * (d2 ? (d / d2) : d * INFINITY); + case e_add: return e->value * (d + d2); + case e_last:return e->value * d2; + case e_st : { + int index = av_clip(d, 0, VARS-1); + p->prng_state[index].counter = 0; + return e->value * (p->var[index]= d2); + } + case e_hypot:return e->value * hypot(d, d2); + case e_atan2:return e->value * atan2(d, d2); + case e_bitand: return isnan(d) || isnan(d2) ? NAN : e->value * ((long int)d & (long int)d2); + case e_bitor: return isnan(d) || isnan(d2) ? NAN : e->value * ((long int)d | (long int)d2); + } + } + } + return NAN; +} + +static int parse_expr(AVExpr **e, Parser *p); + +void av_expr_free(AVExpr *e) +{ + if (!e) return; + av_expr_free(e->param[0]); + av_expr_free(e->param[1]); + av_expr_free(e->param[2]); + av_freep(&e->var); + av_freep(&e->prng_state); + av_freep(&e); +} + +static int parse_primary(AVExpr **e, Parser *p) +{ + AVExpr *d = av_mallocz(sizeof(AVExpr)); + char *next = p->s, *s0 = p->s; + int ret, i; + + if (!d) + return AVERROR(ENOMEM); + + /* number */ + d->value = av_strtod(p->s, &next); + if (next != p->s) { + d->type = e_value; + p->s= next; + *e = d; + return 0; + } + d->value = 1; + + /* named constants */ + for (i=0; p->const_names && p->const_names[i]; i++) { + if (strmatch(p->s, p->const_names[i])) { + p->s+= strlen(p->const_names[i]); + d->type = e_const; + d->const_index = i; + *e = d; + return 0; + } + } + for (i = 0; i < FF_ARRAY_ELEMS(constants); i++) { + if (strmatch(p->s, constants[i].name)) { + p->s += strlen(constants[i].name); + d->type = e_value; + d->value = constants[i].value; + *e = d; + return 0; + } + } + + p->s= strchr(p->s, '('); + if (!p->s) { + av_log(p, AV_LOG_ERROR, "Undefined constant or missing '(' in '%s'\n", s0); + p->s= next; + av_expr_free(d); + return AVERROR(EINVAL); + } + p->s++; // "(" + if (*next == '(') { // special case do-nothing + av_freep(&d); + if ((ret = parse_expr(&d, p)) < 0) + return ret; + if (p->s[0] != ')') { + av_log(p, AV_LOG_ERROR, "Missing ')' in '%s'\n", s0); + av_expr_free(d); + return AVERROR(EINVAL); + } + p->s++; // ")" + *e = d; + return 0; + } + if ((ret = parse_expr(&(d->param[0]), p)) < 0) { + av_expr_free(d); + return ret; + } + if (p->s[0]== ',') { + p->s++; // "," + parse_expr(&d->param[1], p); + } + if (p->s[0]== ',') { + p->s++; // "," + parse_expr(&d->param[2], p); + } + if (p->s[0] != ')') { + av_log(p, AV_LOG_ERROR, "Missing ')' or too many args in '%s'\n", s0); + av_expr_free(d); + return AVERROR(EINVAL); + } + p->s++; // ")" + + d->type = e_func0; + if (strmatch(next, "sinh" )) d->a.func0 = sinh; + else if (strmatch(next, "cosh" )) d->a.func0 = cosh; + else if (strmatch(next, "tanh" )) d->a.func0 = tanh; + else if (strmatch(next, "sin" )) d->a.func0 = sin; + else if (strmatch(next, "cos" )) d->a.func0 = cos; + else if (strmatch(next, "tan" )) d->a.func0 = tan; + else if (strmatch(next, "atan" )) d->a.func0 = atan; + else if (strmatch(next, "asin" )) d->a.func0 = asin; + else if (strmatch(next, "acos" )) d->a.func0 = acos; + else if (strmatch(next, "exp" )) d->a.func0 = exp; + else if (strmatch(next, "log" )) d->a.func0 = log; + else if (strmatch(next, "abs" )) d->a.func0 = fabs; + else if (strmatch(next, "time" )) d->a.func0 = etime; + else if (strmatch(next, "squish")) d->type = e_squish; + else if (strmatch(next, "gauss" )) d->type = e_gauss; + else if (strmatch(next, "mod" )) d->type = e_mod; + else if (strmatch(next, "max" )) d->type = e_max; + else if (strmatch(next, "min" )) d->type = e_min; + else if (strmatch(next, "eq" )) d->type = e_eq; + else if (strmatch(next, "gte" )) d->type = e_gte; + else if (strmatch(next, "gt" )) d->type = e_gt; + else if (strmatch(next, "lte" )) d->type = e_lte; + else if (strmatch(next, "lt" )) d->type = e_lt; + else if (strmatch(next, "ld" )) d->type = e_ld; + else if (strmatch(next, "isnan" )) d->type = e_isnan; + else if (strmatch(next, "isinf" )) d->type = e_isinf; + else if (strmatch(next, "st" )) d->type = e_st; + else if (strmatch(next, "while" )) d->type = e_while; + else if (strmatch(next, "taylor")) d->type = e_taylor; + else if (strmatch(next, "root" )) d->type = e_root; + else if (strmatch(next, "floor" )) d->type = e_floor; + else if (strmatch(next, "ceil" )) d->type = e_ceil; + else if (strmatch(next, "trunc" )) d->type = e_trunc; + else if (strmatch(next, "round" )) d->type = e_round; + else if (strmatch(next, "sqrt" )) d->type = e_sqrt; + else if (strmatch(next, "not" )) d->type = e_not; + else if (strmatch(next, "pow" )) d->type = e_pow; + else if (strmatch(next, "print" )) d->type = e_print; + else if (strmatch(next, "random")) d->type = e_random; + else if (strmatch(next, "randomi")) d->type = e_randomi; + else if (strmatch(next, "hypot" )) d->type = e_hypot; + else if (strmatch(next, "gcd" )) d->type = e_gcd; + else if (strmatch(next, "if" )) d->type = e_if; + else if (strmatch(next, "ifnot" )) d->type = e_ifnot; + else if (strmatch(next, "bitand")) d->type = e_bitand; + else if (strmatch(next, "bitor" )) d->type = e_bitor; + else if (strmatch(next, "between"))d->type = e_between; + else if (strmatch(next, "clip" )) d->type = e_clip; + else if (strmatch(next, "atan2" )) d->type = e_atan2; + else if (strmatch(next, "lerp" )) d->type = e_lerp; + else if (strmatch(next, "sgn" )) d->type = e_sgn; + else { + for (i=0; p->func1_names && p->func1_names[i]; i++) { + if (strmatch(next, p->func1_names[i])) { + d->a.func1 = p->funcs1[i]; + d->type = e_func1; + d->const_index = i; + *e = d; + return 0; + } + } + + for (i=0; p->func2_names && p->func2_names[i]; i++) { + if (strmatch(next, p->func2_names[i])) { + d->a.func2 = p->funcs2[i]; + d->type = e_func2; + d->const_index = i; + *e = d; + return 0; + } + } + + av_log(p, AV_LOG_ERROR, "Unknown function in '%s'\n", s0); + av_expr_free(d); + return AVERROR(EINVAL); + } + + *e = d; + return 0; +} + +static AVExpr *make_eval_expr(int type, int value, AVExpr *p0, AVExpr *p1) +{ + AVExpr *e = av_mallocz(sizeof(AVExpr)); + if (!e) + return NULL; + e->type =type ; + e->value =value ; + e->param[0] =p0 ; + e->param[1] =p1 ; + return e; +} + +static int parse_pow(AVExpr **e, Parser *p, int *sign) +{ + *sign= (*p->s == '+') - (*p->s == '-'); + p->s += *sign&1; + return parse_primary(e, p); +} + +static int parse_dB(AVExpr **e, Parser *p, int *sign) +{ + /* do not filter out the negative sign when parsing a dB value. + for example, -3dB is not the same as -(3dB) */ + if (*p->s == '-') { + char *next; + double av_unused ignored = strtod(p->s, &next); + if (next != p->s && next[0] == 'd' && next[1] == 'B') { + *sign = 0; + return parse_primary(e, p); + } + } + return parse_pow(e, p, sign); +} + +static int parse_factor(AVExpr **e, Parser *p) +{ + int sign, sign2, ret; + AVExpr *e0, *e1, *e2; + if ((ret = parse_dB(&e0, p, &sign)) < 0) + return ret; + while(p->s[0]=='^'){ + e1 = e0; + p->s++; + if ((ret = parse_dB(&e2, p, &sign2)) < 0) { + av_expr_free(e1); + return ret; + } + e0 = make_eval_expr(e_pow, 1, e1, e2); + if (!e0) { + av_expr_free(e1); + av_expr_free(e2); + return AVERROR(ENOMEM); + } + if (e0->param[1]) e0->param[1]->value *= (sign2|1); + } + if (e0) e0->value *= (sign|1); + + *e = e0; + return 0; +} + +static int parse_term(AVExpr **e, Parser *p) +{ + int ret; + AVExpr *e0, *e1, *e2; + if ((ret = parse_factor(&e0, p)) < 0) + return ret; + while (p->s[0]=='*' || p->s[0]=='/') { + int c= *p->s++; + e1 = e0; + if ((ret = parse_factor(&e2, p)) < 0) { + av_expr_free(e1); + return ret; + } + e0 = make_eval_expr(c == '*' ? e_mul : e_div, 1, e1, e2); + if (!e0) { + av_expr_free(e1); + av_expr_free(e2); + return AVERROR(ENOMEM); + } + } + *e = e0; + return 0; +} + +static int parse_subexpr(AVExpr **e, Parser *p) +{ + int ret; + AVExpr *e0, *e1, *e2; + if ((ret = parse_term(&e0, p)) < 0) + return ret; + while (*p->s == '+' || *p->s == '-') { + e1 = e0; + if ((ret = parse_term(&e2, p)) < 0) { + av_expr_free(e1); + return ret; + } + e0 = make_eval_expr(e_add, 1, e1, e2); + if (!e0) { + av_expr_free(e1); + av_expr_free(e2); + return AVERROR(ENOMEM); + } + }; + + *e = e0; + return 0; +} + +static int parse_expr(AVExpr **e, Parser *p) +{ + int ret; + AVExpr *e0, *e1, *e2; + if (p->stack_index <= 0) //protect against stack overflows + return AVERROR(EINVAL); + p->stack_index--; + + if ((ret = parse_subexpr(&e0, p)) < 0) + return ret; + while (*p->s == ';') { + p->s++; + e1 = e0; + if ((ret = parse_subexpr(&e2, p)) < 0) { + av_expr_free(e1); + return ret; + } + e0 = make_eval_expr(e_last, 1, e1, e2); + if (!e0) { + av_expr_free(e1); + av_expr_free(e2); + return AVERROR(ENOMEM); + } + }; + + p->stack_index++; + *e = e0; + return 0; +} + +static int verify_expr(AVExpr *e) +{ + if (!e) return 0; + switch (e->type) { + case e_value: + case e_const: return 1; + case e_func0: + case e_func1: + case e_squish: + case e_ld: + case e_gauss: + case e_isnan: + case e_isinf: + case e_floor: + case e_ceil: + case e_trunc: + case e_round: + case e_sqrt: + case e_not: + case e_random: + case e_sgn: + return verify_expr(e->param[0]) && !e->param[1]; + case e_print: + return verify_expr(e->param[0]) + && (!e->param[1] || verify_expr(e->param[1])); + case e_if: + case e_ifnot: + case e_taylor: + return verify_expr(e->param[0]) && verify_expr(e->param[1]) + && (!e->param[2] || verify_expr(e->param[2])); + case e_between: + case e_clip: + case e_lerp: + case e_randomi: + return verify_expr(e->param[0]) && + verify_expr(e->param[1]) && + verify_expr(e->param[2]); + default: return verify_expr(e->param[0]) && verify_expr(e->param[1]) && !e->param[2]; + } +} + +int av_expr_parse(AVExpr **expr, const char *s, + const char * const *const_names, + const char * const *func1_names, double (* const *funcs1)(void *, double), + const char * const *func2_names, double (* const *funcs2)(void *, double, double), + int log_offset, void *log_ctx) +{ + Parser p = { 0 }; + AVExpr *e = NULL; + char *w = av_malloc(strlen(s) + 1); + char *wp = w; + const char *s0 = s; + int ret = 0; + + if (!w) + return AVERROR(ENOMEM); + + while (*s) + if (!av_isspace(*s++)) *wp++ = s[-1]; + *wp++ = 0; + + p.class = &eval_class; + p.stack_index=100; + p.s= w; + p.const_names = const_names; + p.funcs1 = funcs1; + p.func1_names = func1_names; + p.funcs2 = funcs2; + p.func2_names = func2_names; + p.log_offset = log_offset; + p.log_ctx = log_ctx; + + if ((ret = parse_expr(&e, &p)) < 0) + goto end; + if (*p.s) { + av_log(&p, AV_LOG_ERROR, "Invalid chars '%s' at the end of expression '%s'\n", p.s, s0); + ret = AVERROR(EINVAL); + goto end; + } + if (!verify_expr(e)) { + ret = AVERROR(EINVAL); + goto end; + } + e->var= av_mallocz(sizeof(double) *VARS); + e->prng_state = av_mallocz(sizeof(*e->prng_state) *VARS); + if (!e->var || !e->prng_state) { + ret = AVERROR(ENOMEM); + goto end; + } + *expr = e; + e = NULL; +end: + av_expr_free(e); + av_free(w); + return ret; +} + +static int expr_count(AVExpr *e, unsigned *counter, int size, int type) +{ + int i; + + if (!e || !counter || !size) + return AVERROR(EINVAL); + + for (i = 0; e->type != type && i < 3 && e->param[i]; i++) + expr_count(e->param[i], counter, size, type); + + if (e->type == type && e->const_index < size) + counter[e->const_index]++; + + return 0; +} + +int av_expr_count_vars(AVExpr *e, unsigned *counter, int size) +{ + return expr_count(e, counter, size, e_const); +} + +int av_expr_count_func(AVExpr *e, unsigned *counter, int size, int arg) +{ + return expr_count(e, counter, size, ((int[]){e_const, e_func1, e_func2})[arg]); +} + +double av_expr_eval(AVExpr *e, const double *const_values, void *opaque) +{ + Parser p = { 0 }; + p.var= e->var; + p.prng_state= e->prng_state; + + p.const_values = const_values; + p.opaque = opaque; + return eval_expr(&p, e); +} + +int av_expr_parse_and_eval(double *d, const char *s, + const char * const *const_names, const double *const_values, + const char * const *func1_names, double (* const *funcs1)(void *, double), + const char * const *func2_names, double (* const *funcs2)(void *, double, double), + void *opaque, int log_offset, void *log_ctx) +{ + AVExpr *e = NULL; + int ret = av_expr_parse(&e, s, const_names, func1_names, funcs1, func2_names, funcs2, log_offset, log_ctx); + + if (ret < 0) { + *d = NAN; + return ret; + } + *d = av_expr_eval(e, const_values, opaque); + av_expr_free(e); + return isnan(*d) ? AVERROR(EINVAL) : 0; +} diff --git a/media/ffvpx/libavutil/eval.h b/media/ffvpx/libavutil/eval.h new file mode 100644 index 0000000000..0d3eaeb3fb --- /dev/null +++ b/media/ffvpx/libavutil/eval.h @@ -0,0 +1,140 @@ +/* + * Copyright (c) 2002 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * simple arithmetic expression evaluator + */ + +#ifndef AVUTIL_EVAL_H +#define AVUTIL_EVAL_H + +typedef struct AVExpr AVExpr; + +/** + * Parse and evaluate an expression. + * Note, this is significantly slower than av_expr_eval(). + * + * @param res a pointer to a double where is put the result value of + * the expression, or NAN in case of error + * @param s expression as a zero terminated string, for example "1+2^3+5*5+sin(2/3)" + * @param const_names NULL terminated array of zero terminated strings of constant identifiers, for example {"PI", "E", 0} + * @param const_values a zero terminated array of values for the identifiers from const_names + * @param func1_names NULL terminated array of zero terminated strings of funcs1 identifiers + * @param funcs1 NULL terminated array of function pointers for functions which take 1 argument + * @param func2_names NULL terminated array of zero terminated strings of funcs2 identifiers + * @param funcs2 NULL terminated array of function pointers for functions which take 2 arguments + * @param opaque a pointer which will be passed to all functions from funcs1 and funcs2 + * @param log_offset log level offset, can be used to silence error messages + * @param log_ctx parent logging context + * @return >= 0 in case of success, a negative value corresponding to an + * AVERROR code otherwise + */ +int av_expr_parse_and_eval(double *res, const char *s, + const char * const *const_names, const double *const_values, + const char * const *func1_names, double (* const *funcs1)(void *, double), + const char * const *func2_names, double (* const *funcs2)(void *, double, double), + void *opaque, int log_offset, void *log_ctx); + +/** + * Parse an expression. + * + * @param expr a pointer where is put an AVExpr containing the parsed + * value in case of successful parsing, or NULL otherwise. + * The pointed to AVExpr must be freed with av_expr_free() by the user + * when it is not needed anymore. + * @param s expression as a zero terminated string, for example "1+2^3+5*5+sin(2/3)" + * @param const_names NULL terminated array of zero terminated strings of constant identifiers, for example {"PI", "E", 0} + * @param func1_names NULL terminated array of zero terminated strings of funcs1 identifiers + * @param funcs1 NULL terminated array of function pointers for functions which take 1 argument + * @param func2_names NULL terminated array of zero terminated strings of funcs2 identifiers + * @param funcs2 NULL terminated array of function pointers for functions which take 2 arguments + * @param log_offset log level offset, can be used to silence error messages + * @param log_ctx parent logging context + * @return >= 0 in case of success, a negative value corresponding to an + * AVERROR code otherwise + */ +int av_expr_parse(AVExpr **expr, const char *s, + const char * const *const_names, + const char * const *func1_names, double (* const *funcs1)(void *, double), + const char * const *func2_names, double (* const *funcs2)(void *, double, double), + int log_offset, void *log_ctx); + +/** + * Evaluate a previously parsed expression. + * + * @param e the AVExpr to evaluate + * @param const_values a zero terminated array of values for the identifiers from av_expr_parse() const_names + * @param opaque a pointer which will be passed to all functions from funcs1 and funcs2 + * @return the value of the expression + */ +double av_expr_eval(AVExpr *e, const double *const_values, void *opaque); + +/** + * Track the presence of variables and their number of occurrences in a parsed expression + * + * @param e the AVExpr to track variables in + * @param counter a zero-initialized array where the count of each variable will be stored + * @param size size of array + * @return 0 on success, a negative value indicates that no expression or array was passed + * or size was zero + */ +int av_expr_count_vars(AVExpr *e, unsigned *counter, int size); + +/** + * Track the presence of user provided functions and their number of occurrences + * in a parsed expression. + * + * @param e the AVExpr to track user provided functions in + * @param counter a zero-initialized array where the count of each function will be stored + * if you passed 5 functions with 2 arguments to av_expr_parse() + * then for arg=2 this will use up to 5 entries. + * @param size size of array + * @param arg number of arguments the counted functions have + * @return 0 on success, a negative value indicates that no expression or array was passed + * or size was zero + */ +int av_expr_count_func(AVExpr *e, unsigned *counter, int size, int arg); + +/** + * Free a parsed expression previously created with av_expr_parse(). + */ +void av_expr_free(AVExpr *e); + +/** + * Parse the string in numstr and return its value as a double. If + * the string is empty, contains only whitespaces, or does not contain + * an initial substring that has the expected syntax for a + * floating-point number, no conversion is performed. In this case, + * returns a value of zero and the value returned in tail is the value + * of numstr. + * + * @param numstr a string representing a number, may contain one of + * the International System number postfixes, for example 'K', 'M', + * 'G'. If 'i' is appended after the postfix, powers of 2 are used + * instead of powers of 10. The 'B' postfix multiplies the value by + * 8, and can be appended after another postfix or used alone. This + * allows using for example 'KB', 'MiB', 'G' and 'B' as postfix. + * @param tail if non-NULL puts here the pointer to the char next + * after the last parsed character + */ +double av_strtod(const char *numstr, char **tail); + +#endif /* AVUTIL_EVAL_H */ diff --git a/media/ffvpx/libavutil/ffmath.h b/media/ffvpx/libavutil/ffmath.h new file mode 100644 index 0000000000..aad1347f26 --- /dev/null +++ b/media/ffvpx/libavutil/ffmath.h @@ -0,0 +1,67 @@ +/* + * copyright (c) 2016 Ganesh Ajjanagadde <gajjanag@gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * internal math functions header + */ + +#ifndef AVUTIL_FFMATH_H +#define AVUTIL_FFMATH_H + +#include "attributes.h" +#include "libm.h" + +/** + * Compute 10^x for floating point values. Note: this function is by no means + * "correctly rounded", and is meant as a fast, reasonably accurate approximation. + * For instance, maximum relative error for the double precision variant is + * ~ 1e-13 for very small and very large values. + * This is ~2x faster than GNU libm's approach, which is still off by 2ulp on + * some inputs. + * @param x exponent + * @return 10^x + */ +static av_always_inline double ff_exp10(double x) +{ + return exp2(M_LOG2_10 * x); +} + +static av_always_inline float ff_exp10f(float x) +{ + return exp2f(M_LOG2_10 * x); +} + +/** + * Compute x^y for floating point x, y. Note: this function is faster than the + * libm variant due to mainly 2 reasons: + * 1. It does not handle any edge cases. In particular, this is only guaranteed + * to work correctly for x > 0. + * 2. It is not as accurate as a standard nearly "correctly rounded" libm variant. + * @param x base + * @param y exponent + * @return x^y + */ +static av_always_inline float ff_fast_powf(float x, float y) +{ + return expf(logf(x) * y); +} + +#endif /* AVUTIL_FFMATH_H */ diff --git a/media/ffvpx/libavutil/fftime.h b/media/ffvpx/libavutil/fftime.h new file mode 100644 index 0000000000..8f3b320e38 --- /dev/null +++ b/media/ffvpx/libavutil/fftime.h @@ -0,0 +1,57 @@ +/* + * Copyright (c) 2000-2003 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_TIME_H +#define AVUTIL_TIME_H + +#include <stdint.h> +#include <time.h> + +/** + * Get the current time in microseconds. + */ +int64_t av_gettime(void); + +/** + * Get the current time in microseconds since some unspecified starting point. + * On platforms that support it, the time comes from a monotonic clock + * This property makes this time source ideal for measuring relative time. + * The returned values may not be monotonic on platforms where a monotonic + * clock is not available. + */ +int64_t av_gettime_relative(void); + +/** + * Indicates with a boolean result if the av_gettime_relative() time source + * is monotonic. + */ +int av_gettime_relative_is_monotonic(void); + +/** + * Sleep for a period of time. Although the duration is expressed in + * microseconds, the actual delay may be rounded to the precision of the + * system timer. + * + * @param usec Number of microseconds to sleep. + * @return zero on success or (negative) error code. + */ +int av_usleep(unsigned usec); + +#endif /* AVUTIL_TIME_H */ diff --git a/media/ffvpx/libavutil/ffversion.h b/media/ffvpx/libavutil/ffversion.h new file mode 100644 index 0000000000..392808d726 --- /dev/null +++ b/media/ffvpx/libavutil/ffversion.h @@ -0,0 +1,5 @@ +/* Automatically generated by version.sh, do not manually edit! */ +#ifndef AVUTIL_FFVERSION_H +#define AVUTIL_FFVERSION_H +#define FFMPEG_VERSION "N-111736-gd9d5695390" +#endif /* AVUTIL_FFVERSION_H */ diff --git a/media/ffvpx/libavutil/fifo.c b/media/ffvpx/libavutil/fifo.c new file mode 100644 index 0000000000..b0807abbf7 --- /dev/null +++ b/media/ffvpx/libavutil/fifo.c @@ -0,0 +1,511 @@ +/* + * a very simple circular buffer FIFO implementation + * Copyright (c) 2000, 2001, 2002 Fabrice Bellard + * Copyright (c) 2006 Roman Shaposhnik + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> +#include <string.h> + +#include "avassert.h" +#include "error.h" +#include "fifo.h" +#include "macros.h" +#include "mem.h" + +// by default the FIFO can be auto-grown to 1MB +#define AUTO_GROW_DEFAULT_BYTES (1024 * 1024) + +struct AVFifo { + uint8_t *buffer; + + size_t elem_size, nb_elems; + size_t offset_r, offset_w; + // distinguishes the ambiguous situation offset_r == offset_w + int is_empty; + + unsigned int flags; + size_t auto_grow_limit; +}; + +AVFifo *av_fifo_alloc2(size_t nb_elems, size_t elem_size, + unsigned int flags) +{ + AVFifo *f; + void *buffer = NULL; + + if (!elem_size) + return NULL; + + if (nb_elems) { + buffer = av_realloc_array(NULL, nb_elems, elem_size); + if (!buffer) + return NULL; + } + f = av_mallocz(sizeof(*f)); + if (!f) { + av_free(buffer); + return NULL; + } + f->buffer = buffer; + f->nb_elems = nb_elems; + f->elem_size = elem_size; + f->is_empty = 1; + + f->flags = flags; + f->auto_grow_limit = FFMAX(AUTO_GROW_DEFAULT_BYTES / elem_size, 1); + + return f; +} + +void av_fifo_auto_grow_limit(AVFifo *f, size_t max_elems) +{ + f->auto_grow_limit = max_elems; +} + +size_t av_fifo_elem_size(const AVFifo *f) +{ + return f->elem_size; +} + +size_t av_fifo_can_read(const AVFifo *f) +{ + if (f->offset_w <= f->offset_r && !f->is_empty) + return f->nb_elems - f->offset_r + f->offset_w; + return f->offset_w - f->offset_r; +} + +size_t av_fifo_can_write(const AVFifo *f) +{ + return f->nb_elems - av_fifo_can_read(f); +} + +int av_fifo_grow2(AVFifo *f, size_t inc) +{ + uint8_t *tmp; + + if (inc > SIZE_MAX - f->nb_elems) + return AVERROR(EINVAL); + + tmp = av_realloc_array(f->buffer, f->nb_elems + inc, f->elem_size); + if (!tmp) + return AVERROR(ENOMEM); + f->buffer = tmp; + + // move the data from the beginning of the ring buffer + // to the newly allocated space + if (f->offset_w <= f->offset_r && !f->is_empty) { + const size_t copy = FFMIN(inc, f->offset_w); + memcpy(tmp + f->nb_elems * f->elem_size, tmp, copy * f->elem_size); + if (copy < f->offset_w) { + memmove(tmp, tmp + copy * f->elem_size, + (f->offset_w - copy) * f->elem_size); + f->offset_w -= copy; + } else + f->offset_w = copy == inc ? 0 : f->nb_elems + copy; + } + + f->nb_elems += inc; + + return 0; +} + +static int fifo_check_space(AVFifo *f, size_t to_write) +{ + const size_t can_write = av_fifo_can_write(f); + const size_t need_grow = to_write > can_write ? to_write - can_write : 0; + size_t can_grow; + + if (!need_grow) + return 0; + + can_grow = f->auto_grow_limit > f->nb_elems ? + f->auto_grow_limit - f->nb_elems : 0; + if ((f->flags & AV_FIFO_FLAG_AUTO_GROW) && need_grow <= can_grow) { + // allocate a bit more than necessary, if we can + const size_t inc = (need_grow < can_grow / 2 ) ? need_grow * 2 : can_grow; + return av_fifo_grow2(f, inc); + } + + return AVERROR(ENOSPC); +} + +static int fifo_write_common(AVFifo *f, const uint8_t *buf, size_t *nb_elems, + AVFifoCB read_cb, void *opaque) +{ + size_t to_write = *nb_elems; + size_t offset_w; + int ret = 0; + + ret = fifo_check_space(f, to_write); + if (ret < 0) + return ret; + + offset_w = f->offset_w; + + while (to_write > 0) { + size_t len = FFMIN(f->nb_elems - offset_w, to_write); + uint8_t *wptr = f->buffer + offset_w * f->elem_size; + + if (read_cb) { + ret = read_cb(opaque, wptr, &len); + if (ret < 0 || len == 0) + break; + } else { + memcpy(wptr, buf, len * f->elem_size); + buf += len * f->elem_size; + } + offset_w += len; + if (offset_w >= f->nb_elems) + offset_w = 0; + to_write -= len; + } + f->offset_w = offset_w; + + if (*nb_elems != to_write) + f->is_empty = 0; + *nb_elems -= to_write; + + return ret; +} + +int av_fifo_write(AVFifo *f, const void *buf, size_t nb_elems) +{ + return fifo_write_common(f, buf, &nb_elems, NULL, NULL); +} + +int av_fifo_write_from_cb(AVFifo *f, AVFifoCB read_cb, + void *opaque, size_t *nb_elems) +{ + return fifo_write_common(f, NULL, nb_elems, read_cb, opaque); +} + +static int fifo_peek_common(const AVFifo *f, uint8_t *buf, size_t *nb_elems, + size_t offset, AVFifoCB write_cb, void *opaque) +{ + size_t to_read = *nb_elems; + size_t offset_r = f->offset_r; + size_t can_read = av_fifo_can_read(f); + int ret = 0; + + if (offset > can_read || to_read > can_read - offset) { + *nb_elems = 0; + return AVERROR(EINVAL); + } + + if (offset_r >= f->nb_elems - offset) + offset_r -= f->nb_elems - offset; + else + offset_r += offset; + + while (to_read > 0) { + size_t len = FFMIN(f->nb_elems - offset_r, to_read); + uint8_t *rptr = f->buffer + offset_r * f->elem_size; + + if (write_cb) { + ret = write_cb(opaque, rptr, &len); + if (ret < 0 || len == 0) + break; + } else { + memcpy(buf, rptr, len * f->elem_size); + buf += len * f->elem_size; + } + offset_r += len; + if (offset_r >= f->nb_elems) + offset_r = 0; + to_read -= len; + } + + *nb_elems -= to_read; + + return ret; +} + +int av_fifo_read(AVFifo *f, void *buf, size_t nb_elems) +{ + int ret = fifo_peek_common(f, buf, &nb_elems, 0, NULL, NULL); + av_fifo_drain2(f, nb_elems); + return ret; +} + +int av_fifo_read_to_cb(AVFifo *f, AVFifoCB write_cb, + void *opaque, size_t *nb_elems) +{ + int ret = fifo_peek_common(f, NULL, nb_elems, 0, write_cb, opaque); + av_fifo_drain2(f, *nb_elems); + return ret; +} + +int av_fifo_peek(const AVFifo *f, void *buf, size_t nb_elems, size_t offset) +{ + return fifo_peek_common(f, buf, &nb_elems, offset, NULL, NULL); +} + +int av_fifo_peek_to_cb(const AVFifo *f, AVFifoCB write_cb, void *opaque, + size_t *nb_elems, size_t offset) +{ + return fifo_peek_common(f, NULL, nb_elems, offset, write_cb, opaque); +} + +void av_fifo_drain2(AVFifo *f, size_t size) +{ + const size_t cur_size = av_fifo_can_read(f); + + av_assert0(cur_size >= size); + if (cur_size == size) + f->is_empty = 1; + + if (f->offset_r >= f->nb_elems - size) + f->offset_r -= f->nb_elems - size; + else + f->offset_r += size; +} + +void av_fifo_reset2(AVFifo *f) +{ + f->offset_r = f->offset_w = 0; + f->is_empty = 1; +} + +void av_fifo_freep2(AVFifo **f) +{ + if (*f) { + av_freep(&(*f)->buffer); + av_freep(f); + } +} + + +#if FF_API_FIFO_OLD_API +#include "internal.h" +FF_DISABLE_DEPRECATION_WARNINGS +#define OLD_FIFO_SIZE_MAX (size_t)FFMIN3(INT_MAX, UINT32_MAX, SIZE_MAX) + +AVFifoBuffer *av_fifo_alloc_array(size_t nmemb, size_t size) +{ + AVFifoBuffer *f; + void *buffer; + + if (nmemb > OLD_FIFO_SIZE_MAX / size) + return NULL; + + buffer = av_realloc_array(NULL, nmemb, size); + if (!buffer) + return NULL; + f = av_mallocz(sizeof(AVFifoBuffer)); + if (!f) { + av_free(buffer); + return NULL; + } + f->buffer = buffer; + f->end = f->buffer + nmemb * size; + av_fifo_reset(f); + return f; +} + +AVFifoBuffer *av_fifo_alloc(unsigned int size) +{ + return av_fifo_alloc_array(size, 1); +} + +void av_fifo_free(AVFifoBuffer *f) +{ + if (f) { + av_freep(&f->buffer); + av_free(f); + } +} + +void av_fifo_freep(AVFifoBuffer **f) +{ + if (f) { + av_fifo_free(*f); + *f = NULL; + } +} + +void av_fifo_reset(AVFifoBuffer *f) +{ + f->wptr = f->rptr = f->buffer; + f->wndx = f->rndx = 0; +} + +int av_fifo_size(const AVFifoBuffer *f) +{ + return (uint32_t)(f->wndx - f->rndx); +} + +int av_fifo_space(const AVFifoBuffer *f) +{ + return f->end - f->buffer - av_fifo_size(f); +} + +int av_fifo_realloc2(AVFifoBuffer *f, unsigned int new_size) +{ + unsigned int old_size = f->end - f->buffer; + + if (new_size > OLD_FIFO_SIZE_MAX) + return AVERROR(EINVAL); + + if (old_size < new_size) { + size_t offset_r = f->rptr - f->buffer; + size_t offset_w = f->wptr - f->buffer; + uint8_t *tmp; + + tmp = av_realloc(f->buffer, new_size); + if (!tmp) + return AVERROR(ENOMEM); + + // move the data from the beginning of the ring buffer + // to the newly allocated space + // the second condition distinguishes full vs empty fifo + if (offset_w <= offset_r && av_fifo_size(f)) { + const size_t copy = FFMIN(new_size - old_size, offset_w); + memcpy(tmp + old_size, tmp, copy); + if (copy < offset_w) { + memmove(tmp, tmp + copy , offset_w - copy); + offset_w -= copy; + } else + offset_w = old_size + copy; + } + + f->buffer = tmp; + f->end = f->buffer + new_size; + f->rptr = f->buffer + offset_r; + f->wptr = f->buffer + offset_w; + } + return 0; +} + +int av_fifo_grow(AVFifoBuffer *f, unsigned int size) +{ + unsigned int old_size = f->end - f->buffer; + if(size + (unsigned)av_fifo_size(f) < size) + return AVERROR(EINVAL); + + size += av_fifo_size(f); + + if (old_size < size) + return av_fifo_realloc2(f, FFMAX(size, 2*old_size)); + return 0; +} + +/* src must NOT be const as it can be a context for func that may need + * updating (like a pointer or byte counter) */ +int av_fifo_generic_write(AVFifoBuffer *f, void *src, int size, + int (*func)(void *, void *, int)) +{ + int total = size; + uint32_t wndx= f->wndx; + uint8_t *wptr= f->wptr; + + if (size > av_fifo_space(f)) + return AVERROR(ENOSPC); + + do { + int len = FFMIN(f->end - wptr, size); + if (func) { + len = func(src, wptr, len); + if (len <= 0) + break; + } else { + memcpy(wptr, src, len); + src = (uint8_t *)src + len; + } + wptr += len; + if (wptr >= f->end) + wptr = f->buffer; + wndx += len; + size -= len; + } while (size > 0); + f->wndx= wndx; + f->wptr= wptr; + return total - size; +} + +int av_fifo_generic_peek_at(AVFifoBuffer *f, void *dest, int offset, int buf_size, void (*func)(void*, void*, int)) +{ + uint8_t *rptr = f->rptr; + + if (offset < 0 || buf_size > av_fifo_size(f) - offset) + return AVERROR(EINVAL); + + if (offset >= f->end - rptr) + rptr += offset - (f->end - f->buffer); + else + rptr += offset; + + while (buf_size > 0) { + int len; + + if (rptr >= f->end) + rptr -= f->end - f->buffer; + + len = FFMIN(f->end - rptr, buf_size); + if (func) + func(dest, rptr, len); + else { + memcpy(dest, rptr, len); + dest = (uint8_t *)dest + len; + } + + buf_size -= len; + rptr += len; + } + + return 0; +} + +int av_fifo_generic_peek(AVFifoBuffer *f, void *dest, int buf_size, + void (*func)(void *, void *, int)) +{ + return av_fifo_generic_peek_at(f, dest, 0, buf_size, func); +} + +int av_fifo_generic_read(AVFifoBuffer *f, void *dest, int buf_size, + void (*func)(void *, void *, int)) +{ + if (buf_size > av_fifo_size(f)) + return AVERROR(EINVAL); + + do { + int len = FFMIN(f->end - f->rptr, buf_size); + if (func) + func(dest, f->rptr, len); + else { + memcpy(dest, f->rptr, len); + dest = (uint8_t *)dest + len; + } + av_fifo_drain(f, len); + buf_size -= len; + } while (buf_size > 0); + return 0; +} + +/** Discard data from the FIFO. */ +void av_fifo_drain(AVFifoBuffer *f, int size) +{ + av_assert2(av_fifo_size(f) >= size); + f->rptr += size; + if (f->rptr >= f->end) + f->rptr -= f->end - f->buffer; + f->rndx += size; +} +FF_ENABLE_DEPRECATION_WARNINGS +#endif diff --git a/media/ffvpx/libavutil/fifo.h b/media/ffvpx/libavutil/fifo.h new file mode 100644 index 0000000000..ce3a2aed7c --- /dev/null +++ b/media/ffvpx/libavutil/fifo.h @@ -0,0 +1,448 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_fifo + * A generic FIFO API + */ + +#ifndef AVUTIL_FIFO_H +#define AVUTIL_FIFO_H + +#include <stddef.h> +#include <stdint.h> + +#include "attributes.h" +#include "version.h" + +/** + * @defgroup lavu_fifo AVFifo + * @ingroup lavu_data + * + * @{ + * A generic FIFO API + */ + +typedef struct AVFifo AVFifo; + +/** + * Callback for writing or reading from a FIFO, passed to (and invoked from) the + * av_fifo_*_cb() functions. It may be invoked multiple times from a single + * av_fifo_*_cb() call and may process less data than the maximum size indicated + * by nb_elems. + * + * @param opaque the opaque pointer provided to the av_fifo_*_cb() function + * @param buf the buffer for reading or writing the data, depending on which + * av_fifo_*_cb function is called + * @param nb_elems On entry contains the maximum number of elements that can be + * read from / written into buf. On success, the callback should + * update it to contain the number of elements actually written. + * + * @return 0 on success, a negative error code on failure (will be returned from + * the invoking av_fifo_*_cb() function) + */ +typedef int AVFifoCB(void *opaque, void *buf, size_t *nb_elems); + +/** + * Automatically resize the FIFO on writes, so that the data fits. This + * automatic resizing happens up to a limit that can be modified with + * av_fifo_auto_grow_limit(). + */ +#define AV_FIFO_FLAG_AUTO_GROW (1 << 0) + +/** + * Allocate and initialize an AVFifo with a given element size. + * + * @param elems initial number of elements that can be stored in the FIFO + * @param elem_size Size in bytes of a single element. Further operations on + * the returned FIFO will implicitly use this element size. + * @param flags a combination of AV_FIFO_FLAG_* + * + * @return newly-allocated AVFifo on success, a negative error code on failure + */ +AVFifo *av_fifo_alloc2(size_t elems, size_t elem_size, + unsigned int flags); + +/** + * @return Element size for FIFO operations. This element size is set at + * FIFO allocation and remains constant during its lifetime + */ +size_t av_fifo_elem_size(const AVFifo *f); + +/** + * Set the maximum size (in elements) to which the FIFO can be resized + * automatically. Has no effect unless AV_FIFO_FLAG_AUTO_GROW is used. + */ +void av_fifo_auto_grow_limit(AVFifo *f, size_t max_elems); + +/** + * @return number of elements available for reading from the given FIFO. + */ +size_t av_fifo_can_read(const AVFifo *f); + +/** + * @return Number of elements that can be written into the given FIFO without + * growing it. + * + * In other words, this number of elements or less is guaranteed to fit + * into the FIFO. More data may be written when the + * AV_FIFO_FLAG_AUTO_GROW flag was specified at FIFO creation, but this + * may involve memory allocation, which can fail. + */ +size_t av_fifo_can_write(const AVFifo *f); + +/** + * Enlarge an AVFifo. + * + * On success, the FIFO will be large enough to hold exactly + * inc + av_fifo_can_read() + av_fifo_can_write() + * elements. In case of failure, the old FIFO is kept unchanged. + * + * @param f AVFifo to resize + * @param inc number of elements to allocate for, in addition to the current + * allocated size + * @return a non-negative number on success, a negative error code on failure + */ +int av_fifo_grow2(AVFifo *f, size_t inc); + +/** + * Write data into a FIFO. + * + * In case nb_elems > av_fifo_can_write(f) and the AV_FIFO_FLAG_AUTO_GROW flag + * was not specified at FIFO creation, nothing is written and an error + * is returned. + * + * Calling function is guaranteed to succeed if nb_elems <= av_fifo_can_write(f). + * + * @param f the FIFO buffer + * @param buf Data to be written. nb_elems * av_fifo_elem_size(f) bytes will be + * read from buf on success. + * @param nb_elems number of elements to write into FIFO + * + * @return a non-negative number on success, a negative error code on failure + */ +int av_fifo_write(AVFifo *f, const void *buf, size_t nb_elems); + +/** + * Write data from a user-provided callback into a FIFO. + * + * @param f the FIFO buffer + * @param read_cb Callback supplying the data to the FIFO. May be called + * multiple times. + * @param opaque opaque user data to be provided to read_cb + * @param nb_elems Should point to the maximum number of elements that can be + * written. Will be updated to contain the number of elements + * actually written. + * + * @return non-negative number on success, a negative error code on failure + */ +int av_fifo_write_from_cb(AVFifo *f, AVFifoCB read_cb, + void *opaque, size_t *nb_elems); + +/** + * Read data from a FIFO. + * + * In case nb_elems > av_fifo_can_read(f), nothing is read and an error + * is returned. + * + * @param f the FIFO buffer + * @param buf Buffer to store the data. nb_elems * av_fifo_elem_size(f) bytes + * will be written into buf on success. + * @param nb_elems number of elements to read from FIFO + * + * @return a non-negative number on success, a negative error code on failure + */ +int av_fifo_read(AVFifo *f, void *buf, size_t nb_elems); + +/** + * Feed data from a FIFO into a user-provided callback. + * + * @param f the FIFO buffer + * @param write_cb Callback the data will be supplied to. May be called + * multiple times. + * @param opaque opaque user data to be provided to write_cb + * @param nb_elems Should point to the maximum number of elements that can be + * read. Will be updated to contain the total number of elements + * actually sent to the callback. + * + * @return non-negative number on success, a negative error code on failure + */ +int av_fifo_read_to_cb(AVFifo *f, AVFifoCB write_cb, + void *opaque, size_t *nb_elems); + +/** + * Read data from a FIFO without modifying FIFO state. + * + * Returns an error if an attempt is made to peek to nonexistent elements + * (i.e. if offset + nb_elems is larger than av_fifo_can_read(f)). + * + * @param f the FIFO buffer + * @param buf Buffer to store the data. nb_elems * av_fifo_elem_size(f) bytes + * will be written into buf. + * @param nb_elems number of elements to read from FIFO + * @param offset number of initial elements to skip. + * + * @return a non-negative number on success, a negative error code on failure + */ +int av_fifo_peek(const AVFifo *f, void *buf, size_t nb_elems, size_t offset); + +/** + * Feed data from a FIFO into a user-provided callback. + * + * @param f the FIFO buffer + * @param write_cb Callback the data will be supplied to. May be called + * multiple times. + * @param opaque opaque user data to be provided to write_cb + * @param nb_elems Should point to the maximum number of elements that can be + * read. Will be updated to contain the total number of elements + * actually sent to the callback. + * @param offset number of initial elements to skip; offset + *nb_elems must not + * be larger than av_fifo_can_read(f). + * + * @return a non-negative number on success, a negative error code on failure + */ +int av_fifo_peek_to_cb(const AVFifo *f, AVFifoCB write_cb, void *opaque, + size_t *nb_elems, size_t offset); + +/** + * Discard the specified amount of data from an AVFifo. + * @param size number of elements to discard, MUST NOT be larger than + * av_fifo_can_read(f) + */ +void av_fifo_drain2(AVFifo *f, size_t size); + +/* + * Empty the AVFifo. + * @param f AVFifo to reset + */ +void av_fifo_reset2(AVFifo *f); + +/** + * Free an AVFifo and reset pointer to NULL. + * @param f Pointer to an AVFifo to free. *f == NULL is allowed. + */ +void av_fifo_freep2(AVFifo **f); + + +#if FF_API_FIFO_OLD_API +typedef struct AVFifoBuffer { + uint8_t *buffer; + uint8_t *rptr, *wptr, *end; + uint32_t rndx, wndx; +} AVFifoBuffer; + +/** + * Initialize an AVFifoBuffer. + * @param size of FIFO + * @return AVFifoBuffer or NULL in case of memory allocation failure + * @deprecated use av_fifo_alloc2() + */ +attribute_deprecated +AVFifoBuffer *av_fifo_alloc(unsigned int size); + +/** + * Initialize an AVFifoBuffer. + * @param nmemb number of elements + * @param size size of the single element + * @return AVFifoBuffer or NULL in case of memory allocation failure + * @deprecated use av_fifo_alloc2() + */ +attribute_deprecated +AVFifoBuffer *av_fifo_alloc_array(size_t nmemb, size_t size); + +/** + * Free an AVFifoBuffer. + * @param f AVFifoBuffer to free + * @deprecated use the AVFifo API with av_fifo_freep2() + */ +attribute_deprecated +void av_fifo_free(AVFifoBuffer *f); + +/** + * Free an AVFifoBuffer and reset pointer to NULL. + * @param f AVFifoBuffer to free + * @deprecated use the AVFifo API with av_fifo_freep2() + */ +attribute_deprecated +void av_fifo_freep(AVFifoBuffer **f); + +/** + * Reset the AVFifoBuffer to the state right after av_fifo_alloc, in particular it is emptied. + * @param f AVFifoBuffer to reset + * @deprecated use av_fifo_reset2() with the new AVFifo-API + */ +attribute_deprecated +void av_fifo_reset(AVFifoBuffer *f); + +/** + * Return the amount of data in bytes in the AVFifoBuffer, that is the + * amount of data you can read from it. + * @param f AVFifoBuffer to read from + * @return size + * @deprecated use av_fifo_can_read() with the new AVFifo-API + */ +attribute_deprecated +int av_fifo_size(const AVFifoBuffer *f); + +/** + * Return the amount of space in bytes in the AVFifoBuffer, that is the + * amount of data you can write into it. + * @param f AVFifoBuffer to write into + * @return size + * @deprecated use av_fifo_can_write() with the new AVFifo-API + */ +attribute_deprecated +int av_fifo_space(const AVFifoBuffer *f); + +/** + * Feed data at specific position from an AVFifoBuffer to a user-supplied callback. + * Similar as av_fifo_gereric_read but without discarding data. + * @param f AVFifoBuffer to read from + * @param offset offset from current read position + * @param buf_size number of bytes to read + * @param func generic read function + * @param dest data destination + * + * @return a non-negative number on success, a negative error code on failure + * + * @deprecated use the new AVFifo-API with av_fifo_peek() when func == NULL, + * av_fifo_peek_to_cb() otherwise + */ +attribute_deprecated +int av_fifo_generic_peek_at(AVFifoBuffer *f, void *dest, int offset, int buf_size, void (*func)(void*, void*, int)); + +/** + * Feed data from an AVFifoBuffer to a user-supplied callback. + * Similar as av_fifo_gereric_read but without discarding data. + * @param f AVFifoBuffer to read from + * @param buf_size number of bytes to read + * @param func generic read function + * @param dest data destination + * + * @return a non-negative number on success, a negative error code on failure + * + * @deprecated use the new AVFifo-API with av_fifo_peek() when func == NULL, + * av_fifo_peek_to_cb() otherwise + */ +attribute_deprecated +int av_fifo_generic_peek(AVFifoBuffer *f, void *dest, int buf_size, void (*func)(void*, void*, int)); + +/** + * Feed data from an AVFifoBuffer to a user-supplied callback. + * @param f AVFifoBuffer to read from + * @param buf_size number of bytes to read + * @param func generic read function + * @param dest data destination + * + * @return a non-negative number on success, a negative error code on failure + * + * @deprecated use the new AVFifo-API with av_fifo_read() when func == NULL, + * av_fifo_read_to_cb() otherwise + */ +attribute_deprecated +int av_fifo_generic_read(AVFifoBuffer *f, void *dest, int buf_size, void (*func)(void*, void*, int)); + +/** + * Feed data from a user-supplied callback to an AVFifoBuffer. + * @param f AVFifoBuffer to write to + * @param src data source; non-const since it may be used as a + * modifiable context by the function defined in func + * @param size number of bytes to write + * @param func generic write function; the first parameter is src, + * the second is dest_buf, the third is dest_buf_size. + * func must return the number of bytes written to dest_buf, or <= 0 to + * indicate no more data available to write. + * If func is NULL, src is interpreted as a simple byte array for source data. + * @return the number of bytes written to the FIFO or a negative error code on failure + * + * @deprecated use the new AVFifo-API with av_fifo_write() when func == NULL, + * av_fifo_write_from_cb() otherwise + */ +attribute_deprecated +int av_fifo_generic_write(AVFifoBuffer *f, void *src, int size, int (*func)(void*, void*, int)); + +/** + * Resize an AVFifoBuffer. + * In case of reallocation failure, the old FIFO is kept unchanged. + * + * @param f AVFifoBuffer to resize + * @param size new AVFifoBuffer size in bytes + * @return <0 for failure, >=0 otherwise + * + * @deprecated use the new AVFifo-API with av_fifo_grow2() to increase FIFO size, + * decreasing FIFO size is not supported + */ +attribute_deprecated +int av_fifo_realloc2(AVFifoBuffer *f, unsigned int size); + +/** + * Enlarge an AVFifoBuffer. + * In case of reallocation failure, the old FIFO is kept unchanged. + * The new fifo size may be larger than the requested size. + * + * @param f AVFifoBuffer to resize + * @param additional_space the amount of space in bytes to allocate in addition to av_fifo_size() + * @return <0 for failure, >=0 otherwise + * + * @deprecated use the new AVFifo-API with av_fifo_grow2(); note that unlike + * this function it adds to the allocated size, rather than to the used size + */ +attribute_deprecated +int av_fifo_grow(AVFifoBuffer *f, unsigned int additional_space); + +/** + * Read and discard the specified amount of data from an AVFifoBuffer. + * @param f AVFifoBuffer to read from + * @param size amount of data to read in bytes + * + * @deprecated use the new AVFifo-API with av_fifo_drain2() + */ +attribute_deprecated +void av_fifo_drain(AVFifoBuffer *f, int size); + +#if FF_API_FIFO_PEEK2 +/** + * Return a pointer to the data stored in a FIFO buffer at a certain offset. + * The FIFO buffer is not modified. + * + * @param f AVFifoBuffer to peek at, f must be non-NULL + * @param offs an offset in bytes, its absolute value must be less + * than the used buffer size or the returned pointer will + * point outside to the buffer data. + * The used buffer size can be checked with av_fifo_size(). + * @deprecated use the new AVFifo-API with av_fifo_peek() or av_fifo_peek_to_cb() + */ +attribute_deprecated +static inline uint8_t *av_fifo_peek2(const AVFifoBuffer *f, int offs) +{ + uint8_t *ptr = f->rptr + offs; + if (ptr >= f->end) + ptr = f->buffer + (ptr - f->end); + else if (ptr < f->buffer) + ptr = f->end - (f->buffer - ptr); + return ptr; +} +#endif +#endif + +/** + * @} + */ + +#endif /* AVUTIL_FIFO_H */ diff --git a/media/ffvpx/libavutil/film_grain_params.c b/media/ffvpx/libavutil/film_grain_params.c new file mode 100644 index 0000000000..930d23c7fe --- /dev/null +++ b/media/ffvpx/libavutil/film_grain_params.c @@ -0,0 +1,42 @@ +/** + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "film_grain_params.h" + +AVFilmGrainParams *av_film_grain_params_alloc(size_t *size) +{ + AVFilmGrainParams *params = av_mallocz(sizeof(AVFilmGrainParams)); + + if (size) + *size = sizeof(*params); + + return params; +} + +AVFilmGrainParams *av_film_grain_params_create_side_data(AVFrame *frame) +{ + AVFrameSideData *side_data = av_frame_new_side_data(frame, + AV_FRAME_DATA_FILM_GRAIN_PARAMS, + sizeof(AVFilmGrainParams)); + if (!side_data) + return NULL; + + memset(side_data->data, 0, sizeof(AVFilmGrainParams)); + + return (AVFilmGrainParams *)side_data->data; +} diff --git a/media/ffvpx/libavutil/film_grain_params.h b/media/ffvpx/libavutil/film_grain_params.h new file mode 100644 index 0000000000..f3bd0a4a6a --- /dev/null +++ b/media/ffvpx/libavutil/film_grain_params.h @@ -0,0 +1,260 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_FILM_GRAIN_PARAMS_H +#define AVUTIL_FILM_GRAIN_PARAMS_H + +#include "frame.h" + +enum AVFilmGrainParamsType { + AV_FILM_GRAIN_PARAMS_NONE = 0, + + /** + * The union is valid when interpreted as AVFilmGrainAOMParams (codec.aom) + */ + AV_FILM_GRAIN_PARAMS_AV1, + + /** + * The union is valid when interpreted as AVFilmGrainH274Params (codec.h274) + */ + AV_FILM_GRAIN_PARAMS_H274, +}; + +/** + * This structure describes how to handle film grain synthesis for AOM codecs. + * + * @note The struct must be allocated as part of AVFilmGrainParams using + * av_film_grain_params_alloc(). Its size is not a part of the public ABI. + */ +typedef struct AVFilmGrainAOMParams { + /** + * Number of points, and the scale and value for each point of the + * piecewise linear scaling function for the uma plane. + */ + int num_y_points; + uint8_t y_points[14][2 /* value, scaling */]; + + /** + * Signals whether to derive the chroma scaling function from the luma. + * Not equivalent to copying the luma values and scales. + */ + int chroma_scaling_from_luma; + + /** + * If chroma_scaling_from_luma is set to 0, signals the chroma scaling + * function parameters. + */ + int num_uv_points[2 /* cb, cr */]; + uint8_t uv_points[2 /* cb, cr */][10][2 /* value, scaling */]; + + /** + * Specifies the shift applied to the chroma components. For AV1, its within + * [8; 11] and determines the range and quantization of the film grain. + */ + int scaling_shift; + + /** + * Specifies the auto-regression lag. + */ + int ar_coeff_lag; + + /** + * Luma auto-regression coefficients. The number of coefficients is given by + * 2 * ar_coeff_lag * (ar_coeff_lag + 1). + */ + int8_t ar_coeffs_y[24]; + + /** + * Chroma auto-regression coefficients. The number of coefficients is given by + * 2 * ar_coeff_lag * (ar_coeff_lag + 1) + !!num_y_points. + */ + int8_t ar_coeffs_uv[2 /* cb, cr */][25]; + + /** + * Specifies the range of the auto-regressive coefficients. Values of 6, + * 7, 8 and so on represent a range of [-2, 2), [-1, 1), [-0.5, 0.5) and + * so on. For AV1 must be between 6 and 9. + */ + int ar_coeff_shift; + + /** + * Signals the down shift applied to the generated gaussian numbers during + * synthesis. + */ + int grain_scale_shift; + + /** + * Specifies the luma/chroma multipliers for the index to the component + * scaling function. + */ + int uv_mult[2 /* cb, cr */]; + int uv_mult_luma[2 /* cb, cr */]; + + /** + * Offset used for component scaling function. For AV1 its a 9-bit value + * with a range [-256, 255] + */ + int uv_offset[2 /* cb, cr */]; + + /** + * Signals whether to overlap film grain blocks. + */ + int overlap_flag; + + /** + * Signals to clip to limited color levels after film grain application. + */ + int limit_output_range; +} AVFilmGrainAOMParams; + +/** + * This structure describes how to handle film grain synthesis for codecs using + * the ITU-T H.274 Versatile suplemental enhancement information message. + * + * @note The struct must be allocated as part of AVFilmGrainParams using + * av_film_grain_params_alloc(). Its size is not a part of the public ABI. + */ +typedef struct AVFilmGrainH274Params { + /** + * Specifies the film grain simulation mode. + * 0 = Frequency filtering, 1 = Auto-regression + */ + int model_id; + + /** + * Specifies the bit depth used for the luma component. + */ + int bit_depth_luma; + + /** + * Specifies the bit depth used for the chroma components. + */ + int bit_depth_chroma; + + enum AVColorRange color_range; + enum AVColorPrimaries color_primaries; + enum AVColorTransferCharacteristic color_trc; + enum AVColorSpace color_space; + + /** + * Specifies the blending mode used to blend the simulated film grain + * with the decoded images. + * + * 0 = Additive, 1 = Multiplicative + */ + int blending_mode_id; + + /** + * Specifies a scale factor used in the film grain characterization equations. + */ + int log2_scale_factor; + + /** + * Indicates if the modelling of film grain for a given component is present. + */ + int component_model_present[3 /* y, cb, cr */]; + + /** + * Specifies the number of intensity intervals for which a specific set of + * model values has been estimated, with a range of [1, 256]. + */ + uint16_t num_intensity_intervals[3 /* y, cb, cr */]; + + /** + * Specifies the number of model values present for each intensity interval + * in which the film grain has been modelled, with a range of [1, 6]. + */ + uint8_t num_model_values[3 /* y, cb, cr */]; + + /** + * Specifies the lower ounds of each intensity interval for whichthe set of + * model values applies for the component. + */ + uint8_t intensity_interval_lower_bound[3 /* y, cb, cr */][256 /* intensity interval */]; + + /** + * Specifies the upper bound of each intensity interval for which the set of + * model values applies for the component. + */ + uint8_t intensity_interval_upper_bound[3 /* y, cb, cr */][256 /* intensity interval */]; + + /** + * Specifies the model values for the component for each intensity interval. + * - When model_id == 0, the following applies: + * For comp_model_value[y], the range of values is [0, 2^bit_depth_luma - 1] + * For comp_model_value[cb..cr], the range of values is [0, 2^bit_depth_chroma - 1] + * - Otherwise, the following applies: + * For comp_model_value[y], the range of values is [-2^(bit_depth_luma - 1), 2^(bit_depth_luma - 1) - 1] + * For comp_model_value[cb..cr], the range of values is [-2^(bit_depth_chroma - 1), 2^(bit_depth_chroma - 1) - 1] + */ + int16_t comp_model_value[3 /* y, cb, cr */][256 /* intensity interval */][6 /* model value */]; +} AVFilmGrainH274Params; + +/** + * This structure describes how to handle film grain synthesis in video + * for specific codecs. Must be present on every frame where film grain is + * meant to be synthesised for correct presentation. + * + * @note The struct must be allocated with av_film_grain_params_alloc() and + * its size is not a part of the public ABI. + */ +typedef struct AVFilmGrainParams { + /** + * Specifies the codec for which this structure is valid. + */ + enum AVFilmGrainParamsType type; + + /** + * Seed to use for the synthesis process, if the codec allows for it. + * + * @note For H.264, this refers to `pic_offset` as defined in + * SMPTE RDD 5-2006. + */ + uint64_t seed; + + /** + * Additional fields may be added both here and in any structure included. + * If a codec's film grain structure differs slightly over another + * codec's, fields within may change meaning depending on the type. + */ + union { + AVFilmGrainAOMParams aom; + AVFilmGrainH274Params h274; + } codec; +} AVFilmGrainParams; + +/** + * Allocate an AVFilmGrainParams structure and set its fields to + * default values. The resulting struct can be freed using av_freep(). + * If size is not NULL it will be set to the number of bytes allocated. + * + * @return An AVFilmGrainParams filled with default values or NULL + * on failure. + */ +AVFilmGrainParams *av_film_grain_params_alloc(size_t *size); + +/** + * Allocate a complete AVFilmGrainParams and add it to the frame. + * + * @param frame The frame which side data is added to. + * + * @return The AVFilmGrainParams structure to be filled by caller. + */ +AVFilmGrainParams *av_film_grain_params_create_side_data(AVFrame *frame); + +#endif /* AVUTIL_FILM_GRAIN_PARAMS_H */ diff --git a/media/ffvpx/libavutil/fixed_dsp.c b/media/ffvpx/libavutil/fixed_dsp.c new file mode 100644 index 0000000000..5ab47d55d0 --- /dev/null +++ b/media/ffvpx/libavutil/fixed_dsp.c @@ -0,0 +1,172 @@ +/* + * Copyright (c) 2012 + * MIPS Technologies, Inc., California. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its + * contributors may be used to endorse or promote products derived from + * this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * Author: Nedeljko Babic (nedeljko.babic imgtec com) + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "common.h" +#include "fixed_dsp.h" + +static void vector_fmul_add_c(int *dst, const int *src0, const int *src1, const int *src2, int len){ + int i; + int64_t accu; + + for (i=0; i<len; i++) { + accu = (int64_t)src0[i] * src1[i]; + dst[i] = src2[i] + (int)((accu + 0x40000000) >> 31); + } +} + +static void vector_fmul_reverse_c(int *dst, const int *src0, const int *src1, int len) +{ + int i; + int64_t accu; + + src1 += len-1; + for (i=0; i<len; i++) { + accu = (int64_t)src0[i] * src1[-i]; + dst[i] = (int)((accu+0x40000000) >> 31); + } +} + +static void vector_fmul_window_scaled_c(int16_t *dst, const int32_t *src0, + const int32_t *src1, const int32_t *win, + int len, uint8_t bits) +{ + int32_t s0, s1, wi, wj, i,j, round; + + dst += len; + win += len; + src0+= len; + round = bits? 1 << (bits-1) : 0; + + for (i=-len, j=len-1; i<0; i++, j--) { + s0 = src0[i]; + s1 = src1[j]; + wi = win[i]; + wj = win[j]; + dst[i] = av_clip_int16(((((int64_t)s0*wj - (int64_t)s1*wi + 0x40000000) >> 31) + round) >> bits); + dst[j] = av_clip_int16(((((int64_t)s0*wi + (int64_t)s1*wj + 0x40000000) >> 31) + round) >> bits); + } +} + +static void vector_fmul_window_c(int32_t *dst, const int32_t *src0, + const int32_t *src1, const int32_t *win, + int len) +{ + int32_t s0, s1, wi, wj, i, j; + + dst += len; + win += len; + src0+= len; + + for (i=-len, j=len-1; i<0; i++, j--) { + s0 = src0[i]; + s1 = src1[j]; + wi = win[i]; + wj = win[j]; + dst[i] = ((int64_t)s0*wj - (int64_t)s1*wi + 0x40000000) >> 31; + dst[j] = ((int64_t)s0*wi + (int64_t)s1*wj + 0x40000000) >> 31; + } +} + +static void vector_fmul_c(int *dst, const int *src0, const int *src1, int len) +{ + int i; + int64_t accu; + + for (i = 0; i < len; i++){ + accu = (int64_t)src0[i] * src1[i]; + dst[i] = (int)((accu+0x40000000) >> 31); + } +} + +static int scalarproduct_fixed_c(const int *v1, const int *v2, int len) +{ + /** p is initialized with 0x40000000 so that the proper rounding will occur + * at the end */ + int64_t p = 0x40000000; + int i; + + for (i = 0; i < len; i++) + p += (int64_t)v1[i] * v2[i]; + + return (int)(p >> 31); +} + +static void butterflies_fixed_c(int *av_restrict v1s, int *av_restrict v2, int len) +{ + int i; + unsigned int *v1 = v1s; + + for (i = 0; i < len; i++){ + int t = v1[i] - v2[i]; + v1[i] += v2[i]; + v2[i] = t; + } +} + +AVFixedDSPContext * avpriv_alloc_fixed_dsp(int bit_exact) +{ + AVFixedDSPContext * fdsp = av_malloc(sizeof(AVFixedDSPContext)); + + if (!fdsp) + return NULL; + + fdsp->vector_fmul_window_scaled = vector_fmul_window_scaled_c; + fdsp->vector_fmul_window = vector_fmul_window_c; + fdsp->vector_fmul = vector_fmul_c; + fdsp->vector_fmul_add = vector_fmul_add_c; + fdsp->vector_fmul_reverse = vector_fmul_reverse_c; + fdsp->butterflies_fixed = butterflies_fixed_c; + fdsp->scalarproduct_fixed = scalarproduct_fixed_c; + +#if ARCH_RISCV + ff_fixed_dsp_init_riscv(fdsp); +#elif ARCH_X86 + ff_fixed_dsp_init_x86(fdsp); +#endif + + return fdsp; +} diff --git a/media/ffvpx/libavutil/fixed_dsp.h b/media/ffvpx/libavutil/fixed_dsp.h new file mode 100644 index 0000000000..1217d3a53b --- /dev/null +++ b/media/ffvpx/libavutil/fixed_dsp.h @@ -0,0 +1,206 @@ +/* + * Copyright (c) 2012 + * MIPS Technologies, Inc., California. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its + * contributors may be used to endorse or promote products derived from + * this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * Author: Nedeljko Babic (nbabic@mips.com) + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_FIXED_DSP_H +#define AVUTIL_FIXED_DSP_H + +#include <stdint.h> +#include "config.h" +#include "attributes.h" +#include "libavcodec/mathops.h" + +typedef struct AVFixedDSPContext { + /* Assume len is a multiple of 16, and arrays are 32-byte aligned */ + /* Results of multiplications are scaled down by 31 bit (and rounded) if not + * stated otherwise */ + + /** + * Overlap/add with window function. + * Result is scaled down by "bits" bits. + * Used primarily by MDCT-based audio codecs. + * Source and destination vectors must overlap exactly or not at all. + * + * @param dst result vector + * constraints: 16-byte aligned + * @param src0 first source vector + * constraints: 16-byte aligned + * @param src1 second source vector + * constraints: 16-byte aligned + * @param win half-window vector + * constraints: 16-byte aligned + * @param len length of vector + * constraints: multiple of 4 + * @param bits scaling parameter + * + */ + void (*vector_fmul_window_scaled)(int16_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len, uint8_t bits); + + /** + * Overlap/add with window function. + * Used primarily by MDCT-based audio codecs. + * Source and destination vectors must overlap exactly or not at all. + * + * @param dst result vector + * constraints: 32-byte aligned + * @param src0 first source vector + * constraints: 16-byte aligned + * @param src1 second source vector + * constraints: 16-byte aligned + * @param win half-window vector + * constraints: 16-byte aligned + * @param len length of vector + * constraints: multiple of 4 + */ + void (*vector_fmul_window)(int32_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len); + + /** + * Fixed-point multiplication that calculates the entry wise product of two + * vectors of integers and stores the result in a vector of integers. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_fmul)(int *dst, const int *src0, const int *src1, + int len); + + void (*vector_fmul_reverse)(int *dst, const int *src0, const int *src1, int len); + /** + * Calculate the entry wise product of two vectors of integers, add a third vector of + * integers and store the result in a vector of integers. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param src2 third input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_fmul_add)(int *dst, const int *src0, const int *src1, + const int *src2, int len); + + /** + * Calculate the scalar product of two vectors of integers. + * + * @param v1 first vector, 16-byte aligned + * @param v2 second vector, 16-byte aligned + * @param len length of vectors, multiple of 4 + * + * @return sum of elementwise products + */ + int (*scalarproduct_fixed)(const int *v1, const int *v2, int len); + + /** + * Calculate the sum and difference of two vectors of integers. + * + * @param v1 first input vector, sum output, 16-byte aligned + * @param v2 second input vector, difference output, 16-byte aligned + * @param len length of vectors, multiple of 4 + */ + void (*butterflies_fixed)(int *av_restrict v1, int *av_restrict v2, int len); +} AVFixedDSPContext; + +/** + * Allocate and initialize a fixed DSP context. + * note: should be freed with a av_free call when no longer needed. + * + * @param strict setting to non-zero avoids using functions which may not be IEEE-754 compliant + */ +AVFixedDSPContext * avpriv_alloc_fixed_dsp(int strict); + +void ff_fixed_dsp_init_riscv(AVFixedDSPContext *fdsp); +void ff_fixed_dsp_init_x86(AVFixedDSPContext *fdsp); + +/** + * Calculate the square root + * + * @param x input fixed point number + * + * @param bits format of fixed point number (32 - bits).bits + * + * note: input is normalized to (0, 1) fixed point value + */ + +static av_always_inline int fixed_sqrt(int x, int bits) +{ + int retval, bit_mask, guess, square, i; + int64_t accu; + int shift1 = 30 - bits; + int shift2 = bits - 15; + + if (shift1 > 0) retval = ff_sqrt(x << shift1); + else retval = ff_sqrt(x >> -shift1); + + if (shift2 > 0) { + retval = retval << shift2; + bit_mask = (1 << (shift2 - 1)); + + for (i=0; i<shift2; i++){ + guess = retval + bit_mask; + accu = (int64_t)guess * guess; + square = (int)((accu + bit_mask) >> bits); + if (x >= square) + retval += bit_mask; + bit_mask >>= 1; + } + + } + else retval >>= (-shift2); + + return retval; +} + +#endif /* AVUTIL_FIXED_DSP_H */ diff --git a/media/ffvpx/libavutil/float_dsp.c b/media/ffvpx/libavutil/float_dsp.c new file mode 100644 index 0000000000..742dd679d2 --- /dev/null +++ b/media/ffvpx/libavutil/float_dsp.c @@ -0,0 +1,167 @@ +/* + * Copyright 2005 Balatoni Denes + * Copyright 2006 Loren Merritt + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "attributes.h" +#include "float_dsp.h" +#include "mem.h" + +static void vector_fmul_c(float *dst, const float *src0, const float *src1, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] = src0[i] * src1[i]; +} + +static void vector_dmul_c(double *dst, const double *src0, const double *src1, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] = src0[i] * src1[i]; +} + +static void vector_fmac_scalar_c(float *dst, const float *src, float mul, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] += src[i] * mul; +} + +static void vector_dmac_scalar_c(double *dst, const double *src, double mul, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] += src[i] * mul; +} + +static void vector_fmul_scalar_c(float *dst, const float *src, float mul, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] = src[i] * mul; +} + +static void vector_dmul_scalar_c(double *dst, const double *src, double mul, + int len) +{ + int i; + for (i = 0; i < len; i++) + dst[i] = src[i] * mul; +} + +static void vector_fmul_window_c(float *dst, const float *src0, + const float *src1, const float *win, int len) +{ + int i, j; + + dst += len; + win += len; + src0 += len; + + for (i = -len, j = len - 1; i < 0; i++, j--) { + float s0 = src0[i]; + float s1 = src1[j]; + float wi = win[i]; + float wj = win[j]; + dst[i] = s0 * wj - s1 * wi; + dst[j] = s0 * wi + s1 * wj; + } +} + +static void vector_fmul_add_c(float *dst, const float *src0, const float *src1, + const float *src2, int len){ + int i; + + for (i = 0; i < len; i++) + dst[i] = src0[i] * src1[i] + src2[i]; +} + +static void vector_fmul_reverse_c(float *dst, const float *src0, + const float *src1, int len) +{ + int i; + + src1 += len-1; + for (i = 0; i < len; i++) + dst[i] = src0[i] * src1[-i]; +} + +static void butterflies_float_c(float *av_restrict v1, float *av_restrict v2, + int len) +{ + int i; + + for (i = 0; i < len; i++) { + float t = v1[i] - v2[i]; + v1[i] += v2[i]; + v2[i] = t; + } +} + +float avpriv_scalarproduct_float_c(const float *v1, const float *v2, int len) +{ + float p = 0.0; + int i; + + for (i = 0; i < len; i++) + p += v1[i] * v2[i]; + + return p; +} + +av_cold AVFloatDSPContext *avpriv_float_dsp_alloc(int bit_exact) +{ + AVFloatDSPContext *fdsp = av_mallocz(sizeof(AVFloatDSPContext)); + if (!fdsp) + return NULL; + + fdsp->vector_fmul = vector_fmul_c; + fdsp->vector_dmul = vector_dmul_c; + fdsp->vector_fmac_scalar = vector_fmac_scalar_c; + fdsp->vector_fmul_scalar = vector_fmul_scalar_c; + fdsp->vector_dmac_scalar = vector_dmac_scalar_c; + fdsp->vector_dmul_scalar = vector_dmul_scalar_c; + fdsp->vector_fmul_window = vector_fmul_window_c; + fdsp->vector_fmul_add = vector_fmul_add_c; + fdsp->vector_fmul_reverse = vector_fmul_reverse_c; + fdsp->butterflies_float = butterflies_float_c; + fdsp->scalarproduct_float = avpriv_scalarproduct_float_c; + +#if ARCH_AARCH64 + ff_float_dsp_init_aarch64(fdsp); +#elif ARCH_ARM + ff_float_dsp_init_arm(fdsp); +#elif ARCH_PPC + ff_float_dsp_init_ppc(fdsp, bit_exact); +#elif ARCH_RISCV + ff_float_dsp_init_riscv(fdsp); +#elif ARCH_X86 + ff_float_dsp_init_x86(fdsp); +#elif ARCH_MIPS + ff_float_dsp_init_mips(fdsp); +#endif + return fdsp; +} diff --git a/media/ffvpx/libavutil/float_dsp.h b/media/ffvpx/libavutil/float_dsp.h new file mode 100644 index 0000000000..7cad9fc622 --- /dev/null +++ b/media/ffvpx/libavutil/float_dsp.h @@ -0,0 +1,219 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_FLOAT_DSP_H +#define AVUTIL_FLOAT_DSP_H + +#include "config.h" + +typedef struct AVFloatDSPContext { + /** + * Calculate the entry wise product of two vectors of floats and store the result in + * a vector of floats. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_fmul)(float *dst, const float *src0, const float *src1, + int len); + + /** + * Multiply a vector of floats by a scalar float and add to + * destination vector. Source and destination vectors must + * overlap exactly or not at all. + * + * @param dst result vector + * constraints: 32-byte aligned + * @param src input vector + * constraints: 32-byte aligned + * @param mul scalar value + * @param len length of vector + * constraints: multiple of 16 + */ + void (*vector_fmac_scalar)(float *dst, const float *src, float mul, + int len); + + /** + * Multiply a vector of doubles by a scalar double and add to + * destination vector. Source and destination vectors must + * overlap exactly or not at all. + * + * @param dst result vector + * constraints: 32-byte aligned + * @param src input vector + * constraints: 32-byte aligned + * @param mul scalar value + * @param len length of vector + * constraints: multiple of 16 + */ + void (*vector_dmac_scalar)(double *dst, const double *src, double mul, + int len); + + /** + * Multiply a vector of floats by a scalar float. Source and + * destination vectors must overlap exactly or not at all. + * + * @param dst result vector + * constraints: 16-byte aligned + * @param src input vector + * constraints: 16-byte aligned + * @param mul scalar value + * @param len length of vector + * constraints: multiple of 4 + */ + void (*vector_fmul_scalar)(float *dst, const float *src, float mul, + int len); + + /** + * Multiply a vector of double by a scalar double. Source and + * destination vectors must overlap exactly or not at all. + * + * @param dst result vector + * constraints: 32-byte aligned + * @param src input vector + * constraints: 32-byte aligned + * @param mul scalar value + * @param len length of vector + * constraints: multiple of 8 + */ + void (*vector_dmul_scalar)(double *dst, const double *src, double mul, + int len); + + /** + * Overlap/add with window function. + * Used primarily by MDCT-based audio codecs. + * Source and destination vectors must overlap exactly or not at all. + * + * @param dst result vector + * constraints: 16-byte aligned + * @param src0 first source vector + * constraints: 16-byte aligned + * @param src1 second source vector + * constraints: 16-byte aligned + * @param win half-window vector + * constraints: 16-byte aligned + * @param len length of vector + * constraints: multiple of 4 + */ + void (*vector_fmul_window)(float *dst, const float *src0, + const float *src1, const float *win, int len); + + /** + * Calculate the entry wise product of two vectors of floats, add a third vector of + * floats and store the result in a vector of floats. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param src2 third input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_fmul_add)(float *dst, const float *src0, const float *src1, + const float *src2, int len); + + /** + * Calculate the entry wise product of two vectors of floats, and store the result + * in a vector of floats. The second vector of floats is iterated over + * in reverse order. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_fmul_reverse)(float *dst, const float *src0, + const float *src1, int len); + + /** + * Calculate the sum and difference of two vectors of floats. + * + * @param v1 first input vector, sum output, 16-byte aligned + * @param v2 second input vector, difference output, 16-byte aligned + * @param len length of vectors, multiple of 4 + */ + void (*butterflies_float)(float *av_restrict v1, float *av_restrict v2, int len); + + /** + * Calculate the scalar product of two vectors of floats. + * + * @param v1 first vector, 16-byte aligned + * @param v2 second vector, 16-byte aligned + * @param len length of vectors, multiple of 4 + * + * @return sum of elementwise products + */ + float (*scalarproduct_float)(const float *v1, const float *v2, int len); + + /** + * Calculate the entry wise product of two vectors of doubles and store the result in + * a vector of doubles. + * + * @param dst output vector + * constraints: 32-byte aligned + * @param src0 first input vector + * constraints: 32-byte aligned + * @param src1 second input vector + * constraints: 32-byte aligned + * @param len number of elements in the input + * constraints: multiple of 16 + */ + void (*vector_dmul)(double *dst, const double *src0, const double *src1, + int len); +} AVFloatDSPContext; + +/** + * Return the scalar product of two vectors. + * + * @param v1 first input vector + * @param v2 first input vector + * @param len number of elements + * + * @return sum of elementwise products + */ +float avpriv_scalarproduct_float_c(const float *v1, const float *v2, int len); + +void ff_float_dsp_init_aarch64(AVFloatDSPContext *fdsp); +void ff_float_dsp_init_arm(AVFloatDSPContext *fdsp); +void ff_float_dsp_init_ppc(AVFloatDSPContext *fdsp, int strict); +void ff_float_dsp_init_riscv(AVFloatDSPContext *fdsp); +void ff_float_dsp_init_x86(AVFloatDSPContext *fdsp); +void ff_float_dsp_init_mips(AVFloatDSPContext *fdsp); + +/** + * Allocate a float DSP context. + * + * @param strict setting to non-zero avoids using functions which may not be IEEE-754 compliant + */ +AVFloatDSPContext *avpriv_float_dsp_alloc(int strict); + +#endif /* AVUTIL_FLOAT_DSP_H */ diff --git a/media/ffvpx/libavutil/frame.c b/media/ffvpx/libavutil/frame.c new file mode 100644 index 0000000000..a3f07ca089 --- /dev/null +++ b/media/ffvpx/libavutil/frame.c @@ -0,0 +1,1060 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "channel_layout.h" +#include "avassert.h" +#include "buffer.h" +#include "common.h" +#include "cpu.h" +#include "dict.h" +#include "frame.h" +#include "imgutils.h" +#include "mem.h" +#include "samplefmt.h" +#include "hwcontext.h" + +#if FF_API_OLD_CHANNEL_LAYOUT +#define CHECK_CHANNELS_CONSISTENCY(frame) \ + av_assert2(!(frame)->channel_layout || \ + (frame)->channels == \ + av_get_channel_layout_nb_channels((frame)->channel_layout)) +#endif + +static void get_frame_defaults(AVFrame *frame) +{ + memset(frame, 0, sizeof(*frame)); + + frame->pts = + frame->pkt_dts = AV_NOPTS_VALUE; + frame->best_effort_timestamp = AV_NOPTS_VALUE; + frame->duration = 0; +#if FF_API_PKT_DURATION +FF_DISABLE_DEPRECATION_WARNINGS + frame->pkt_duration = 0; +FF_ENABLE_DEPRECATION_WARNINGS +#endif +#if FF_API_FRAME_PKT +FF_DISABLE_DEPRECATION_WARNINGS + frame->pkt_pos = -1; + frame->pkt_size = -1; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + frame->time_base = (AVRational){ 0, 1 }; + frame->sample_aspect_ratio = (AVRational){ 0, 1 }; + frame->format = -1; /* unknown */ + frame->extended_data = frame->data; + frame->color_primaries = AVCOL_PRI_UNSPECIFIED; + frame->color_trc = AVCOL_TRC_UNSPECIFIED; + frame->colorspace = AVCOL_SPC_UNSPECIFIED; + frame->color_range = AVCOL_RANGE_UNSPECIFIED; + frame->chroma_location = AVCHROMA_LOC_UNSPECIFIED; + frame->flags = 0; +} + +static void free_side_data(AVFrameSideData **ptr_sd) +{ + AVFrameSideData *sd = *ptr_sd; + + av_buffer_unref(&sd->buf); + av_dict_free(&sd->metadata); + av_freep(ptr_sd); +} + +static void wipe_side_data(AVFrame *frame) +{ + for (int i = 0; i < frame->nb_side_data; i++) { + free_side_data(&frame->side_data[i]); + } + frame->nb_side_data = 0; + + av_freep(&frame->side_data); +} + +AVFrame *av_frame_alloc(void) +{ + AVFrame *frame = av_malloc(sizeof(*frame)); + + if (!frame) + return NULL; + + get_frame_defaults(frame); + + return frame; +} + +void av_frame_free(AVFrame **frame) +{ + if (!frame || !*frame) + return; + + av_frame_unref(*frame); + av_freep(frame); +} + +static int get_video_buffer(AVFrame *frame, int align) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(frame->format); + int ret, padded_height, total_size; + int plane_padding = FFMAX(16 + 16/*STRIDE_ALIGN*/, align); + ptrdiff_t linesizes[4]; + size_t sizes[4]; + + if (!desc) + return AVERROR(EINVAL); + + if ((ret = av_image_check_size(frame->width, frame->height, 0, NULL)) < 0) + return ret; + + if (!frame->linesize[0]) { + if (align <= 0) + align = 32; /* STRIDE_ALIGN. Should be av_cpu_max_align() */ + + for (int i = 1; i <= align; i += i) { + ret = av_image_fill_linesizes(frame->linesize, frame->format, + FFALIGN(frame->width, i)); + if (ret < 0) + return ret; + if (!(frame->linesize[0] & (align-1))) + break; + } + + for (int i = 0; i < 4 && frame->linesize[i]; i++) + frame->linesize[i] = FFALIGN(frame->linesize[i], align); + } + + for (int i = 0; i < 4; i++) + linesizes[i] = frame->linesize[i]; + + padded_height = FFALIGN(frame->height, 32); + if ((ret = av_image_fill_plane_sizes(sizes, frame->format, + padded_height, linesizes)) < 0) + return ret; + + total_size = 4*plane_padding; + for (int i = 0; i < 4; i++) { + if (sizes[i] > INT_MAX - total_size) + return AVERROR(EINVAL); + total_size += sizes[i]; + } + + frame->buf[0] = av_buffer_alloc(total_size); + if (!frame->buf[0]) { + ret = AVERROR(ENOMEM); + goto fail; + } + + if ((ret = av_image_fill_pointers(frame->data, frame->format, padded_height, + frame->buf[0]->data, frame->linesize)) < 0) + goto fail; + + for (int i = 1; i < 4; i++) { + if (frame->data[i]) + frame->data[i] += i * plane_padding; + } + + frame->extended_data = frame->data; + + return 0; +fail: + av_frame_unref(frame); + return ret; +} + +static int get_audio_buffer(AVFrame *frame, int align) +{ + int planar = av_sample_fmt_is_planar(frame->format); + int channels, planes; + int ret; + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + if (!frame->ch_layout.nb_channels) { + if (frame->channel_layout) { + av_channel_layout_from_mask(&frame->ch_layout, frame->channel_layout); + } else { + frame->ch_layout.nb_channels = frame->channels; + frame->ch_layout.order = AV_CHANNEL_ORDER_UNSPEC; + } + } + frame->channels = frame->ch_layout.nb_channels; + frame->channel_layout = frame->ch_layout.order == AV_CHANNEL_ORDER_NATIVE ? + frame->ch_layout.u.mask : 0; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + channels = frame->ch_layout.nb_channels; + planes = planar ? channels : 1; + if (!frame->linesize[0]) { + ret = av_samples_get_buffer_size(&frame->linesize[0], channels, + frame->nb_samples, frame->format, + align); + if (ret < 0) + return ret; + } + + if (planes > AV_NUM_DATA_POINTERS) { + frame->extended_data = av_calloc(planes, + sizeof(*frame->extended_data)); + frame->extended_buf = av_calloc(planes - AV_NUM_DATA_POINTERS, + sizeof(*frame->extended_buf)); + if (!frame->extended_data || !frame->extended_buf) { + av_freep(&frame->extended_data); + av_freep(&frame->extended_buf); + return AVERROR(ENOMEM); + } + frame->nb_extended_buf = planes - AV_NUM_DATA_POINTERS; + } else + frame->extended_data = frame->data; + + for (int i = 0; i < FFMIN(planes, AV_NUM_DATA_POINTERS); i++) { + frame->buf[i] = av_buffer_alloc(frame->linesize[0]); + if (!frame->buf[i]) { + av_frame_unref(frame); + return AVERROR(ENOMEM); + } + frame->extended_data[i] = frame->data[i] = frame->buf[i]->data; + } + for (int i = 0; i < planes - AV_NUM_DATA_POINTERS; i++) { + frame->extended_buf[i] = av_buffer_alloc(frame->linesize[0]); + if (!frame->extended_buf[i]) { + av_frame_unref(frame); + return AVERROR(ENOMEM); + } + frame->extended_data[i + AV_NUM_DATA_POINTERS] = frame->extended_buf[i]->data; + } + return 0; + +} + +int av_frame_get_buffer(AVFrame *frame, int align) +{ + if (frame->format < 0) + return AVERROR(EINVAL); + +FF_DISABLE_DEPRECATION_WARNINGS + if (frame->width > 0 && frame->height > 0) + return get_video_buffer(frame, align); + else if (frame->nb_samples > 0 && + (av_channel_layout_check(&frame->ch_layout) +#if FF_API_OLD_CHANNEL_LAYOUT + || frame->channel_layout || frame->channels > 0 +#endif + )) + return get_audio_buffer(frame, align); +FF_ENABLE_DEPRECATION_WARNINGS + + return AVERROR(EINVAL); +} + +static int frame_copy_props(AVFrame *dst, const AVFrame *src, int force_copy) +{ + int ret; + +#if FF_API_FRAME_KEY +FF_DISABLE_DEPRECATION_WARNINGS + dst->key_frame = src->key_frame; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + dst->pict_type = src->pict_type; + dst->sample_aspect_ratio = src->sample_aspect_ratio; + dst->crop_top = src->crop_top; + dst->crop_bottom = src->crop_bottom; + dst->crop_left = src->crop_left; + dst->crop_right = src->crop_right; + dst->pts = src->pts; + dst->duration = src->duration; + dst->repeat_pict = src->repeat_pict; +#if FF_API_INTERLACED_FRAME +FF_DISABLE_DEPRECATION_WARNINGS + dst->interlaced_frame = src->interlaced_frame; + dst->top_field_first = src->top_field_first; +FF_ENABLE_DEPRECATION_WARNINGS +#endif +#if FF_API_PALETTE_HAS_CHANGED +FF_DISABLE_DEPRECATION_WARNINGS + dst->palette_has_changed = src->palette_has_changed; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + dst->sample_rate = src->sample_rate; + dst->opaque = src->opaque; + dst->pkt_dts = src->pkt_dts; +#if FF_API_FRAME_PKT +FF_DISABLE_DEPRECATION_WARNINGS + dst->pkt_pos = src->pkt_pos; + dst->pkt_size = src->pkt_size; +FF_ENABLE_DEPRECATION_WARNINGS +#endif +#if FF_API_PKT_DURATION +FF_DISABLE_DEPRECATION_WARNINGS + dst->pkt_duration = src->pkt_duration; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + dst->time_base = src->time_base; +#if FF_API_REORDERED_OPAQUE +FF_DISABLE_DEPRECATION_WARNINGS + dst->reordered_opaque = src->reordered_opaque; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + dst->quality = src->quality; + dst->best_effort_timestamp = src->best_effort_timestamp; +#if FF_API_FRAME_PICTURE_NUMBER +FF_DISABLE_DEPRECATION_WARNINGS + dst->coded_picture_number = src->coded_picture_number; + dst->display_picture_number = src->display_picture_number; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + dst->flags = src->flags; + dst->decode_error_flags = src->decode_error_flags; + dst->color_primaries = src->color_primaries; + dst->color_trc = src->color_trc; + dst->colorspace = src->colorspace; + dst->color_range = src->color_range; + dst->chroma_location = src->chroma_location; + + av_dict_copy(&dst->metadata, src->metadata, 0); + + for (int i = 0; i < src->nb_side_data; i++) { + const AVFrameSideData *sd_src = src->side_data[i]; + AVFrameSideData *sd_dst; + if ( sd_src->type == AV_FRAME_DATA_PANSCAN + && (src->width != dst->width || src->height != dst->height)) + continue; + if (force_copy) { + sd_dst = av_frame_new_side_data(dst, sd_src->type, + sd_src->size); + if (!sd_dst) { + wipe_side_data(dst); + return AVERROR(ENOMEM); + } + memcpy(sd_dst->data, sd_src->data, sd_src->size); + } else { + AVBufferRef *ref = av_buffer_ref(sd_src->buf); + sd_dst = av_frame_new_side_data_from_buf(dst, sd_src->type, ref); + if (!sd_dst) { + av_buffer_unref(&ref); + wipe_side_data(dst); + return AVERROR(ENOMEM); + } + } + av_dict_copy(&sd_dst->metadata, sd_src->metadata, 0); + } + + ret = av_buffer_replace(&dst->opaque_ref, src->opaque_ref); + ret |= av_buffer_replace(&dst->private_ref, src->private_ref); + return ret; +} + +int av_frame_ref(AVFrame *dst, const AVFrame *src) +{ + int ret = 0; + + av_assert1(dst->width == 0 && dst->height == 0); +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + av_assert1(dst->channels == 0); +FF_ENABLE_DEPRECATION_WARNINGS +#endif + av_assert1(dst->ch_layout.nb_channels == 0 && + dst->ch_layout.order == AV_CHANNEL_ORDER_UNSPEC); + + dst->format = src->format; + dst->width = src->width; + dst->height = src->height; + dst->nb_samples = src->nb_samples; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + dst->channels = src->channels; + dst->channel_layout = src->channel_layout; + if (!av_channel_layout_check(&src->ch_layout)) { + if (src->channel_layout) + av_channel_layout_from_mask(&dst->ch_layout, src->channel_layout); + else { + dst->ch_layout.nb_channels = src->channels; + dst->ch_layout.order = AV_CHANNEL_ORDER_UNSPEC; + } + } +FF_ENABLE_DEPRECATION_WARNINGS +#endif + + ret = frame_copy_props(dst, src, 0); + if (ret < 0) + goto fail; + + // this check is needed only until FF_API_OLD_CHANNEL_LAYOUT is out + if (av_channel_layout_check(&src->ch_layout)) { + ret = av_channel_layout_copy(&dst->ch_layout, &src->ch_layout); + if (ret < 0) + goto fail; + } + + /* duplicate the frame data if it's not refcounted */ + if (!src->buf[0]) { + ret = av_frame_get_buffer(dst, 0); + if (ret < 0) + goto fail; + + ret = av_frame_copy(dst, src); + if (ret < 0) + goto fail; + + return 0; + } + + /* ref the buffers */ + for (int i = 0; i < FF_ARRAY_ELEMS(src->buf); i++) { + if (!src->buf[i]) + continue; + dst->buf[i] = av_buffer_ref(src->buf[i]); + if (!dst->buf[i]) { + ret = AVERROR(ENOMEM); + goto fail; + } + } + + if (src->extended_buf) { + dst->extended_buf = av_calloc(src->nb_extended_buf, + sizeof(*dst->extended_buf)); + if (!dst->extended_buf) { + ret = AVERROR(ENOMEM); + goto fail; + } + dst->nb_extended_buf = src->nb_extended_buf; + + for (int i = 0; i < src->nb_extended_buf; i++) { + dst->extended_buf[i] = av_buffer_ref(src->extended_buf[i]); + if (!dst->extended_buf[i]) { + ret = AVERROR(ENOMEM); + goto fail; + } + } + } + + if (src->hw_frames_ctx) { + dst->hw_frames_ctx = av_buffer_ref(src->hw_frames_ctx); + if (!dst->hw_frames_ctx) { + ret = AVERROR(ENOMEM); + goto fail; + } + } + + /* duplicate extended data */ + if (src->extended_data != src->data) { + int ch = dst->ch_layout.nb_channels; + + if (!ch) { + ret = AVERROR(EINVAL); + goto fail; + } + + dst->extended_data = av_malloc_array(sizeof(*dst->extended_data), ch); + if (!dst->extended_data) { + ret = AVERROR(ENOMEM); + goto fail; + } + memcpy(dst->extended_data, src->extended_data, sizeof(*src->extended_data) * ch); + } else + dst->extended_data = dst->data; + + memcpy(dst->data, src->data, sizeof(src->data)); + memcpy(dst->linesize, src->linesize, sizeof(src->linesize)); + + return 0; + +fail: + av_frame_unref(dst); + return ret; +} + +int av_frame_replace(AVFrame *dst, const AVFrame *src) +{ + int ret = 0; + + if (dst == src) + return AVERROR(EINVAL); + + if (!src->buf[0]) { + av_frame_unref(dst); + + /* duplicate the frame data if it's not refcounted */ + if ( src->data[0] || src->data[1] + || src->data[2] || src->data[3]) + return av_frame_ref(dst, src); + + ret = frame_copy_props(dst, src, 0); + if (ret < 0) + goto fail; + } + + dst->format = src->format; + dst->width = src->width; + dst->height = src->height; + dst->nb_samples = src->nb_samples; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + dst->channels = src->channels; + dst->channel_layout = src->channel_layout; + if (!av_channel_layout_check(&src->ch_layout)) { + av_channel_layout_uninit(&dst->ch_layout); + if (src->channel_layout) + av_channel_layout_from_mask(&dst->ch_layout, src->channel_layout); + else { + dst->ch_layout.nb_channels = src->channels; + dst->ch_layout.order = AV_CHANNEL_ORDER_UNSPEC; + } + } else { +#endif + ret = av_channel_layout_copy(&dst->ch_layout, &src->ch_layout); + if (ret < 0) + goto fail; +#if FF_API_OLD_CHANNEL_LAYOUT + } +FF_ENABLE_DEPRECATION_WARNINGS +#endif + + wipe_side_data(dst); + av_dict_free(&dst->metadata); + ret = frame_copy_props(dst, src, 0); + if (ret < 0) + goto fail; + + /* replace the buffers */ + for (int i = 0; i < FF_ARRAY_ELEMS(src->buf); i++) { + ret = av_buffer_replace(&dst->buf[i], src->buf[i]); + if (ret < 0) + goto fail; + } + + if (src->extended_buf) { + if (dst->nb_extended_buf != src->nb_extended_buf) { + int nb_extended_buf = FFMIN(dst->nb_extended_buf, src->nb_extended_buf); + void *tmp; + + for (int i = nb_extended_buf; i < dst->nb_extended_buf; i++) + av_buffer_unref(&dst->extended_buf[i]); + + tmp = av_realloc_array(dst->extended_buf, sizeof(*dst->extended_buf), + src->nb_extended_buf); + if (!tmp) { + ret = AVERROR(ENOMEM); + goto fail; + } + dst->extended_buf = tmp; + dst->nb_extended_buf = src->nb_extended_buf; + + memset(&dst->extended_buf[nb_extended_buf], 0, + (src->nb_extended_buf - nb_extended_buf) * sizeof(*dst->extended_buf)); + } + + for (int i = 0; i < src->nb_extended_buf; i++) { + ret = av_buffer_replace(&dst->extended_buf[i], src->extended_buf[i]); + if (ret < 0) + goto fail; + } + } else if (dst->extended_buf) { + for (int i = 0; i < dst->nb_extended_buf; i++) + av_buffer_unref(&dst->extended_buf[i]); + av_freep(&dst->extended_buf); + } + + ret = av_buffer_replace(&dst->hw_frames_ctx, src->hw_frames_ctx); + if (ret < 0) + goto fail; + + if (dst->extended_data != dst->data) + av_freep(&dst->extended_data); + + if (src->extended_data != src->data) { + int ch = dst->ch_layout.nb_channels; + + if (!ch) { + ret = AVERROR(EINVAL); + goto fail; + } + + if (ch > SIZE_MAX / sizeof(*dst->extended_data)) + goto fail; + + dst->extended_data = av_memdup(src->extended_data, sizeof(*dst->extended_data) * ch); + if (!dst->extended_data) { + ret = AVERROR(ENOMEM); + goto fail; + } + } else + dst->extended_data = dst->data; + + memcpy(dst->data, src->data, sizeof(src->data)); + memcpy(dst->linesize, src->linesize, sizeof(src->linesize)); + + return 0; + +fail: + av_frame_unref(dst); + return ret; +} + +AVFrame *av_frame_clone(const AVFrame *src) +{ + AVFrame *ret = av_frame_alloc(); + + if (!ret) + return NULL; + + if (av_frame_ref(ret, src) < 0) + av_frame_free(&ret); + + return ret; +} + +void av_frame_unref(AVFrame *frame) +{ + if (!frame) + return; + + wipe_side_data(frame); + + for (int i = 0; i < FF_ARRAY_ELEMS(frame->buf); i++) + av_buffer_unref(&frame->buf[i]); + for (int i = 0; i < frame->nb_extended_buf; i++) + av_buffer_unref(&frame->extended_buf[i]); + av_freep(&frame->extended_buf); + av_dict_free(&frame->metadata); + + av_buffer_unref(&frame->hw_frames_ctx); + + av_buffer_unref(&frame->opaque_ref); + av_buffer_unref(&frame->private_ref); + + if (frame->extended_data != frame->data) + av_freep(&frame->extended_data); + + av_channel_layout_uninit(&frame->ch_layout); + + get_frame_defaults(frame); +} + +void av_frame_move_ref(AVFrame *dst, AVFrame *src) +{ + av_assert1(dst->width == 0 && dst->height == 0); +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + av_assert1(dst->channels == 0); +FF_ENABLE_DEPRECATION_WARNINGS +#endif + av_assert1(dst->ch_layout.nb_channels == 0 && + dst->ch_layout.order == AV_CHANNEL_ORDER_UNSPEC); + + *dst = *src; + if (src->extended_data == src->data) + dst->extended_data = dst->data; + get_frame_defaults(src); +} + +int av_frame_is_writable(AVFrame *frame) +{ + int ret = 1; + + /* assume non-refcounted frames are not writable */ + if (!frame->buf[0]) + return 0; + + for (int i = 0; i < FF_ARRAY_ELEMS(frame->buf); i++) + if (frame->buf[i]) + ret &= !!av_buffer_is_writable(frame->buf[i]); + for (int i = 0; i < frame->nb_extended_buf; i++) + ret &= !!av_buffer_is_writable(frame->extended_buf[i]); + + return ret; +} + +int av_frame_make_writable(AVFrame *frame) +{ + AVFrame tmp; + int ret; + + if (av_frame_is_writable(frame)) + return 0; + + memset(&tmp, 0, sizeof(tmp)); + tmp.format = frame->format; + tmp.width = frame->width; + tmp.height = frame->height; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + tmp.channels = frame->channels; + tmp.channel_layout = frame->channel_layout; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + tmp.nb_samples = frame->nb_samples; + ret = av_channel_layout_copy(&tmp.ch_layout, &frame->ch_layout); + if (ret < 0) { + av_frame_unref(&tmp); + return ret; + } + + if (frame->hw_frames_ctx) + ret = av_hwframe_get_buffer(frame->hw_frames_ctx, &tmp, 0); + else + ret = av_frame_get_buffer(&tmp, 0); + if (ret < 0) + return ret; + + ret = av_frame_copy(&tmp, frame); + if (ret < 0) { + av_frame_unref(&tmp); + return ret; + } + + ret = av_frame_copy_props(&tmp, frame); + if (ret < 0) { + av_frame_unref(&tmp); + return ret; + } + + av_frame_unref(frame); + + *frame = tmp; + if (tmp.data == tmp.extended_data) + frame->extended_data = frame->data; + + return 0; +} + +int av_frame_copy_props(AVFrame *dst, const AVFrame *src) +{ + return frame_copy_props(dst, src, 1); +} + +AVBufferRef *av_frame_get_plane_buffer(const AVFrame *frame, int plane) +{ + uint8_t *data; + int planes; + + if (frame->nb_samples) { + int channels = frame->ch_layout.nb_channels; + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + if (!channels) { + channels = frame->channels; + CHECK_CHANNELS_CONSISTENCY(frame); + } +FF_ENABLE_DEPRECATION_WARNINGS +#endif + if (!channels) + return NULL; + planes = av_sample_fmt_is_planar(frame->format) ? channels : 1; + } else + planes = 4; + + if (plane < 0 || plane >= planes || !frame->extended_data[plane]) + return NULL; + data = frame->extended_data[plane]; + + for (int i = 0; i < FF_ARRAY_ELEMS(frame->buf) && frame->buf[i]; i++) { + AVBufferRef *buf = frame->buf[i]; + if (data >= buf->data && data < buf->data + buf->size) + return buf; + } + for (int i = 0; i < frame->nb_extended_buf; i++) { + AVBufferRef *buf = frame->extended_buf[i]; + if (data >= buf->data && data < buf->data + buf->size) + return buf; + } + return NULL; +} + +AVFrameSideData *av_frame_new_side_data_from_buf(AVFrame *frame, + enum AVFrameSideDataType type, + AVBufferRef *buf) +{ + AVFrameSideData *ret, **tmp; + + if (!buf) + return NULL; + + if (frame->nb_side_data > INT_MAX / sizeof(*frame->side_data) - 1) + return NULL; + + tmp = av_realloc(frame->side_data, + (frame->nb_side_data + 1) * sizeof(*frame->side_data)); + if (!tmp) + return NULL; + frame->side_data = tmp; + + ret = av_mallocz(sizeof(*ret)); + if (!ret) + return NULL; + + ret->buf = buf; + ret->data = ret->buf->data; + ret->size = buf->size; + ret->type = type; + + frame->side_data[frame->nb_side_data++] = ret; + + return ret; +} + +AVFrameSideData *av_frame_new_side_data(AVFrame *frame, + enum AVFrameSideDataType type, + size_t size) +{ + AVFrameSideData *ret; + AVBufferRef *buf = av_buffer_alloc(size); + ret = av_frame_new_side_data_from_buf(frame, type, buf); + if (!ret) + av_buffer_unref(&buf); + return ret; +} + +AVFrameSideData *av_frame_get_side_data(const AVFrame *frame, + enum AVFrameSideDataType type) +{ + for (int i = 0; i < frame->nb_side_data; i++) { + if (frame->side_data[i]->type == type) + return frame->side_data[i]; + } + return NULL; +} + +static int frame_copy_video(AVFrame *dst, const AVFrame *src) +{ + int planes; + + if (dst->width < src->width || + dst->height < src->height) + return AVERROR(EINVAL); + + if (src->hw_frames_ctx || dst->hw_frames_ctx) + return av_hwframe_transfer_data(dst, src, 0); + + planes = av_pix_fmt_count_planes(dst->format); + for (int i = 0; i < planes; i++) + if (!dst->data[i] || !src->data[i]) + return AVERROR(EINVAL); + + av_image_copy2(dst->data, dst->linesize, + src->data, src->linesize, + dst->format, src->width, src->height); + + return 0; +} + +static int frame_copy_audio(AVFrame *dst, const AVFrame *src) +{ + int planar = av_sample_fmt_is_planar(dst->format); + int channels = dst->ch_layout.nb_channels; + int planes = planar ? channels : 1; + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + if (!channels || !src->ch_layout.nb_channels) { + if (dst->channels != src->channels || + dst->channel_layout != src->channel_layout) + return AVERROR(EINVAL); + CHECK_CHANNELS_CONSISTENCY(src); + } + if (!channels) { + channels = dst->channels; + planes = planar ? channels : 1; + } +FF_ENABLE_DEPRECATION_WARNINGS +#endif + + if (dst->nb_samples != src->nb_samples || +#if FF_API_OLD_CHANNEL_LAYOUT + (av_channel_layout_check(&dst->ch_layout) && + av_channel_layout_check(&src->ch_layout) && +#endif + av_channel_layout_compare(&dst->ch_layout, &src->ch_layout)) +#if FF_API_OLD_CHANNEL_LAYOUT + ) +#endif + return AVERROR(EINVAL); + + for (int i = 0; i < planes; i++) + if (!dst->extended_data[i] || !src->extended_data[i]) + return AVERROR(EINVAL); + + av_samples_copy(dst->extended_data, src->extended_data, 0, 0, + dst->nb_samples, channels, dst->format); + + return 0; +} + +int av_frame_copy(AVFrame *dst, const AVFrame *src) +{ + if (dst->format != src->format || dst->format < 0) + return AVERROR(EINVAL); + +FF_DISABLE_DEPRECATION_WARNINGS + if (dst->width > 0 && dst->height > 0) + return frame_copy_video(dst, src); + else if (dst->nb_samples > 0 && + (av_channel_layout_check(&dst->ch_layout) +#if FF_API_OLD_CHANNEL_LAYOUT + || dst->channels > 0 +#endif + )) + return frame_copy_audio(dst, src); +FF_ENABLE_DEPRECATION_WARNINGS + + return AVERROR(EINVAL); +} + +void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type) +{ + for (int i = frame->nb_side_data - 1; i >= 0; i--) { + AVFrameSideData *sd = frame->side_data[i]; + if (sd->type == type) { + free_side_data(&frame->side_data[i]); + frame->side_data[i] = frame->side_data[frame->nb_side_data - 1]; + frame->nb_side_data--; + } + } +} + +const char *av_frame_side_data_name(enum AVFrameSideDataType type) +{ + switch(type) { + case AV_FRAME_DATA_PANSCAN: return "AVPanScan"; + case AV_FRAME_DATA_A53_CC: return "ATSC A53 Part 4 Closed Captions"; + case AV_FRAME_DATA_STEREO3D: return "Stereo 3D"; + case AV_FRAME_DATA_MATRIXENCODING: return "AVMatrixEncoding"; + case AV_FRAME_DATA_DOWNMIX_INFO: return "Metadata relevant to a downmix procedure"; + case AV_FRAME_DATA_REPLAYGAIN: return "AVReplayGain"; + case AV_FRAME_DATA_DISPLAYMATRIX: return "3x3 displaymatrix"; + case AV_FRAME_DATA_AFD: return "Active format description"; + case AV_FRAME_DATA_MOTION_VECTORS: return "Motion vectors"; + case AV_FRAME_DATA_SKIP_SAMPLES: return "Skip samples"; + case AV_FRAME_DATA_AUDIO_SERVICE_TYPE: return "Audio service type"; + case AV_FRAME_DATA_MASTERING_DISPLAY_METADATA: return "Mastering display metadata"; + case AV_FRAME_DATA_CONTENT_LIGHT_LEVEL: return "Content light level metadata"; + case AV_FRAME_DATA_GOP_TIMECODE: return "GOP timecode"; + case AV_FRAME_DATA_S12M_TIMECODE: return "SMPTE 12-1 timecode"; + case AV_FRAME_DATA_SPHERICAL: return "Spherical Mapping"; + case AV_FRAME_DATA_ICC_PROFILE: return "ICC profile"; + case AV_FRAME_DATA_DYNAMIC_HDR_PLUS: return "HDR Dynamic Metadata SMPTE2094-40 (HDR10+)"; + case AV_FRAME_DATA_DYNAMIC_HDR_VIVID: return "HDR Dynamic Metadata CUVA 005.1 2021 (Vivid)"; + case AV_FRAME_DATA_REGIONS_OF_INTEREST: return "Regions Of Interest"; + case AV_FRAME_DATA_VIDEO_ENC_PARAMS: return "Video encoding parameters"; + case AV_FRAME_DATA_SEI_UNREGISTERED: return "H.26[45] User Data Unregistered SEI message"; + case AV_FRAME_DATA_FILM_GRAIN_PARAMS: return "Film grain parameters"; + case AV_FRAME_DATA_DETECTION_BBOXES: return "Bounding boxes for object detection and classification"; + case AV_FRAME_DATA_DOVI_RPU_BUFFER: return "Dolby Vision RPU Data"; + case AV_FRAME_DATA_DOVI_METADATA: return "Dolby Vision Metadata"; + case AV_FRAME_DATA_AMBIENT_VIEWING_ENVIRONMENT: return "Ambient viewing environment"; + } + return NULL; +} + +static int calc_cropping_offsets(size_t offsets[4], const AVFrame *frame, + const AVPixFmtDescriptor *desc) +{ + for (int i = 0; frame->data[i]; i++) { + const AVComponentDescriptor *comp = NULL; + int shift_x = (i == 1 || i == 2) ? desc->log2_chroma_w : 0; + int shift_y = (i == 1 || i == 2) ? desc->log2_chroma_h : 0; + + if (desc->flags & AV_PIX_FMT_FLAG_PAL && i == 1) { + offsets[i] = 0; + break; + } + + /* find any component descriptor for this plane */ + for (int j = 0; j < desc->nb_components; j++) { + if (desc->comp[j].plane == i) { + comp = &desc->comp[j]; + break; + } + } + if (!comp) + return AVERROR_BUG; + + offsets[i] = (frame->crop_top >> shift_y) * frame->linesize[i] + + (frame->crop_left >> shift_x) * comp->step; + } + + return 0; +} + +int av_frame_apply_cropping(AVFrame *frame, int flags) +{ + const AVPixFmtDescriptor *desc; + size_t offsets[4]; + + if (!(frame->width > 0 && frame->height > 0)) + return AVERROR(EINVAL); + + if (frame->crop_left >= INT_MAX - frame->crop_right || + frame->crop_top >= INT_MAX - frame->crop_bottom || + (frame->crop_left + frame->crop_right) >= frame->width || + (frame->crop_top + frame->crop_bottom) >= frame->height) + return AVERROR(ERANGE); + + desc = av_pix_fmt_desc_get(frame->format); + if (!desc) + return AVERROR_BUG; + + /* Apply just the right/bottom cropping for hwaccel formats. Bitstream + * formats cannot be easily handled here either (and corresponding decoders + * should not export any cropping anyway), so do the same for those as well. + * */ + if (desc->flags & (AV_PIX_FMT_FLAG_BITSTREAM | AV_PIX_FMT_FLAG_HWACCEL)) { + frame->width -= frame->crop_right; + frame->height -= frame->crop_bottom; + frame->crop_right = 0; + frame->crop_bottom = 0; + return 0; + } + + /* calculate the offsets for each plane */ + calc_cropping_offsets(offsets, frame, desc); + + /* adjust the offsets to avoid breaking alignment */ + if (!(flags & AV_FRAME_CROP_UNALIGNED)) { + int log2_crop_align = frame->crop_left ? ff_ctz(frame->crop_left) : INT_MAX; + int min_log2_align = INT_MAX; + + for (int i = 0; frame->data[i]; i++) { + int log2_align = offsets[i] ? ff_ctz(offsets[i]) : INT_MAX; + min_log2_align = FFMIN(log2_align, min_log2_align); + } + + /* we assume, and it should always be true, that the data alignment is + * related to the cropping alignment by a constant power-of-2 factor */ + if (log2_crop_align < min_log2_align) + return AVERROR_BUG; + + if (min_log2_align < 5) { + frame->crop_left &= ~((1 << (5 + log2_crop_align - min_log2_align)) - 1); + calc_cropping_offsets(offsets, frame, desc); + } + } + + for (int i = 0; frame->data[i]; i++) + frame->data[i] += offsets[i]; + + frame->width -= (frame->crop_left + frame->crop_right); + frame->height -= (frame->crop_top + frame->crop_bottom); + frame->crop_left = 0; + frame->crop_right = 0; + frame->crop_top = 0; + frame->crop_bottom = 0; + + return 0; +} diff --git a/media/ffvpx/libavutil/frame.h b/media/ffvpx/libavutil/frame.h new file mode 100644 index 0000000000..c0c1b23db7 --- /dev/null +++ b/media/ffvpx/libavutil/frame.h @@ -0,0 +1,1056 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_frame + * reference-counted frame API + */ + +#ifndef AVUTIL_FRAME_H +#define AVUTIL_FRAME_H + +#include <stddef.h> +#include <stdint.h> + +#include "avutil.h" +#include "buffer.h" +#include "channel_layout.h" +#include "dict.h" +#include "rational.h" +#include "samplefmt.h" +#include "pixfmt.h" +#include "version.h" + + +/** + * @defgroup lavu_frame AVFrame + * @ingroup lavu_data + * + * @{ + * AVFrame is an abstraction for reference-counted raw multimedia data. + */ + +enum AVFrameSideDataType { + /** + * The data is the AVPanScan struct defined in libavcodec. + */ + AV_FRAME_DATA_PANSCAN, + /** + * ATSC A53 Part 4 Closed Captions. + * A53 CC bitstream is stored as uint8_t in AVFrameSideData.data. + * The number of bytes of CC data is AVFrameSideData.size. + */ + AV_FRAME_DATA_A53_CC, + /** + * Stereoscopic 3d metadata. + * The data is the AVStereo3D struct defined in libavutil/stereo3d.h. + */ + AV_FRAME_DATA_STEREO3D, + /** + * The data is the AVMatrixEncoding enum defined in libavutil/channel_layout.h. + */ + AV_FRAME_DATA_MATRIXENCODING, + /** + * Metadata relevant to a downmix procedure. + * The data is the AVDownmixInfo struct defined in libavutil/downmix_info.h. + */ + AV_FRAME_DATA_DOWNMIX_INFO, + /** + * ReplayGain information in the form of the AVReplayGain struct. + */ + AV_FRAME_DATA_REPLAYGAIN, + /** + * This side data contains a 3x3 transformation matrix describing an affine + * transformation that needs to be applied to the frame for correct + * presentation. + * + * See libavutil/display.h for a detailed description of the data. + */ + AV_FRAME_DATA_DISPLAYMATRIX, + /** + * Active Format Description data consisting of a single byte as specified + * in ETSI TS 101 154 using AVActiveFormatDescription enum. + */ + AV_FRAME_DATA_AFD, + /** + * Motion vectors exported by some codecs (on demand through the export_mvs + * flag set in the libavcodec AVCodecContext flags2 option). + * The data is the AVMotionVector struct defined in + * libavutil/motion_vector.h. + */ + AV_FRAME_DATA_MOTION_VECTORS, + /** + * Recommmends skipping the specified number of samples. This is exported + * only if the "skip_manual" AVOption is set in libavcodec. + * This has the same format as AV_PKT_DATA_SKIP_SAMPLES. + * @code + * u32le number of samples to skip from start of this packet + * u32le number of samples to skip from end of this packet + * u8 reason for start skip + * u8 reason for end skip (0=padding silence, 1=convergence) + * @endcode + */ + AV_FRAME_DATA_SKIP_SAMPLES, + /** + * This side data must be associated with an audio frame and corresponds to + * enum AVAudioServiceType defined in avcodec.h. + */ + AV_FRAME_DATA_AUDIO_SERVICE_TYPE, + /** + * Mastering display metadata associated with a video frame. The payload is + * an AVMasteringDisplayMetadata type and contains information about the + * mastering display color volume. + */ + AV_FRAME_DATA_MASTERING_DISPLAY_METADATA, + /** + * The GOP timecode in 25 bit timecode format. Data format is 64-bit integer. + * This is set on the first frame of a GOP that has a temporal reference of 0. + */ + AV_FRAME_DATA_GOP_TIMECODE, + + /** + * The data represents the AVSphericalMapping structure defined in + * libavutil/spherical.h. + */ + AV_FRAME_DATA_SPHERICAL, + + /** + * Content light level (based on CTA-861.3). This payload contains data in + * the form of the AVContentLightMetadata struct. + */ + AV_FRAME_DATA_CONTENT_LIGHT_LEVEL, + + /** + * The data contains an ICC profile as an opaque octet buffer following the + * format described by ISO 15076-1 with an optional name defined in the + * metadata key entry "name". + */ + AV_FRAME_DATA_ICC_PROFILE, + + /** + * Timecode which conforms to SMPTE ST 12-1. The data is an array of 4 uint32_t + * where the first uint32_t describes how many (1-3) of the other timecodes are used. + * The timecode format is described in the documentation of av_timecode_get_smpte_from_framenum() + * function in libavutil/timecode.h. + */ + AV_FRAME_DATA_S12M_TIMECODE, + + /** + * HDR dynamic metadata associated with a video frame. The payload is + * an AVDynamicHDRPlus type and contains information for color + * volume transform - application 4 of SMPTE 2094-40:2016 standard. + */ + AV_FRAME_DATA_DYNAMIC_HDR_PLUS, + + /** + * Regions Of Interest, the data is an array of AVRegionOfInterest type, the number of + * array element is implied by AVFrameSideData.size / AVRegionOfInterest.self_size. + */ + AV_FRAME_DATA_REGIONS_OF_INTEREST, + + /** + * Encoding parameters for a video frame, as described by AVVideoEncParams. + */ + AV_FRAME_DATA_VIDEO_ENC_PARAMS, + + /** + * User data unregistered metadata associated with a video frame. + * This is the H.26[45] UDU SEI message, and shouldn't be used for any other purpose + * The data is stored as uint8_t in AVFrameSideData.data which is 16 bytes of + * uuid_iso_iec_11578 followed by AVFrameSideData.size - 16 bytes of user_data_payload_byte. + */ + AV_FRAME_DATA_SEI_UNREGISTERED, + + /** + * Film grain parameters for a frame, described by AVFilmGrainParams. + * Must be present for every frame which should have film grain applied. + */ + AV_FRAME_DATA_FILM_GRAIN_PARAMS, + + /** + * Bounding boxes for object detection and classification, + * as described by AVDetectionBBoxHeader. + */ + AV_FRAME_DATA_DETECTION_BBOXES, + + /** + * Dolby Vision RPU raw data, suitable for passing to x265 + * or other libraries. Array of uint8_t, with NAL emulation + * bytes intact. + */ + AV_FRAME_DATA_DOVI_RPU_BUFFER, + + /** + * Parsed Dolby Vision metadata, suitable for passing to a software + * implementation. The payload is the AVDOVIMetadata struct defined in + * libavutil/dovi_meta.h. + */ + AV_FRAME_DATA_DOVI_METADATA, + + /** + * HDR Vivid dynamic metadata associated with a video frame. The payload is + * an AVDynamicHDRVivid type and contains information for color + * volume transform - CUVA 005.1-2021. + */ + AV_FRAME_DATA_DYNAMIC_HDR_VIVID, + + /** + * Ambient viewing environment metadata, as defined by H.274. + */ + AV_FRAME_DATA_AMBIENT_VIEWING_ENVIRONMENT, + + /** + * Provide encoder-specific hinting information about changed/unchanged + * portions of a frame. It can be used to pass information about which + * macroblocks can be skipped because they didn't change from the + * corresponding ones in the previous frame. This could be useful for + * applications which know this information in advance to speed up + * encoding. + */ + AV_FRAME_DATA_VIDEO_HINT, +}; + +enum AVActiveFormatDescription { + AV_AFD_SAME = 8, + AV_AFD_4_3 = 9, + AV_AFD_16_9 = 10, + AV_AFD_14_9 = 11, + AV_AFD_4_3_SP_14_9 = 13, + AV_AFD_16_9_SP_14_9 = 14, + AV_AFD_SP_4_3 = 15, +}; + + +/** + * Structure to hold side data for an AVFrame. + * + * sizeof(AVFrameSideData) is not a part of the public ABI, so new fields may be added + * to the end with a minor bump. + */ +typedef struct AVFrameSideData { + enum AVFrameSideDataType type; + uint8_t *data; + size_t size; + AVDictionary *metadata; + AVBufferRef *buf; +} AVFrameSideData; + +/** + * Structure describing a single Region Of Interest. + * + * When multiple regions are defined in a single side-data block, they + * should be ordered from most to least important - some encoders are only + * capable of supporting a limited number of distinct regions, so will have + * to truncate the list. + * + * When overlapping regions are defined, the first region containing a given + * area of the frame applies. + */ +typedef struct AVRegionOfInterest { + /** + * Must be set to the size of this data structure (that is, + * sizeof(AVRegionOfInterest)). + */ + uint32_t self_size; + /** + * Distance in pixels from the top edge of the frame to the top and + * bottom edges and from the left edge of the frame to the left and + * right edges of the rectangle defining this region of interest. + * + * The constraints on a region are encoder dependent, so the region + * actually affected may be slightly larger for alignment or other + * reasons. + */ + int top; + int bottom; + int left; + int right; + /** + * Quantisation offset. + * + * Must be in the range -1 to +1. A value of zero indicates no quality + * change. A negative value asks for better quality (less quantisation), + * while a positive value asks for worse quality (greater quantisation). + * + * The range is calibrated so that the extreme values indicate the + * largest possible offset - if the rest of the frame is encoded with the + * worst possible quality, an offset of -1 indicates that this region + * should be encoded with the best possible quality anyway. Intermediate + * values are then interpolated in some codec-dependent way. + * + * For example, in 10-bit H.264 the quantisation parameter varies between + * -12 and 51. A typical qoffset value of -1/10 therefore indicates that + * this region should be encoded with a QP around one-tenth of the full + * range better than the rest of the frame. So, if most of the frame + * were to be encoded with a QP of around 30, this region would get a QP + * of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3). + * An extreme value of -1 would indicate that this region should be + * encoded with the best possible quality regardless of the treatment of + * the rest of the frame - that is, should be encoded at a QP of -12. + */ + AVRational qoffset; +} AVRegionOfInterest; + +/** + * This structure describes decoded (raw) audio or video data. + * + * AVFrame must be allocated using av_frame_alloc(). Note that this only + * allocates the AVFrame itself, the buffers for the data must be managed + * through other means (see below). + * AVFrame must be freed with av_frame_free(). + * + * AVFrame is typically allocated once and then reused multiple times to hold + * different data (e.g. a single AVFrame to hold frames received from a + * decoder). In such a case, av_frame_unref() will free any references held by + * the frame and reset it to its original clean state before it + * is reused again. + * + * The data described by an AVFrame is usually reference counted through the + * AVBuffer API. The underlying buffer references are stored in AVFrame.buf / + * AVFrame.extended_buf. An AVFrame is considered to be reference counted if at + * least one reference is set, i.e. if AVFrame.buf[0] != NULL. In such a case, + * every single data plane must be contained in one of the buffers in + * AVFrame.buf or AVFrame.extended_buf. + * There may be a single buffer for all the data, or one separate buffer for + * each plane, or anything in between. + * + * sizeof(AVFrame) is not a part of the public ABI, so new fields may be added + * to the end with a minor bump. + * + * Fields can be accessed through AVOptions, the name string used, matches the + * C structure field name for fields accessible through AVOptions. The AVClass + * for AVFrame can be obtained from avcodec_get_frame_class() + */ +typedef struct AVFrame { +#define AV_NUM_DATA_POINTERS 8 + /** + * pointer to the picture/channel planes. + * This might be different from the first allocated byte. For video, + * it could even point to the end of the image data. + * + * All pointers in data and extended_data must point into one of the + * AVBufferRef in buf or extended_buf. + * + * Some decoders access areas outside 0,0 - width,height, please + * see avcodec_align_dimensions2(). Some filters and swscale can read + * up to 16 bytes beyond the planes, if these filters are to be used, + * then 16 extra bytes must be allocated. + * + * NOTE: Pointers not needed by the format MUST be set to NULL. + * + * @attention In case of video, the data[] pointers can point to the + * end of image data in order to reverse line order, when used in + * combination with negative values in the linesize[] array. + */ + uint8_t *data[AV_NUM_DATA_POINTERS]; + + /** + * For video, a positive or negative value, which is typically indicating + * the size in bytes of each picture line, but it can also be: + * - the negative byte size of lines for vertical flipping + * (with data[n] pointing to the end of the data + * - a positive or negative multiple of the byte size as for accessing + * even and odd fields of a frame (possibly flipped) + * + * For audio, only linesize[0] may be set. For planar audio, each channel + * plane must be the same size. + * + * For video the linesizes should be multiples of the CPUs alignment + * preference, this is 16 or 32 for modern desktop CPUs. + * Some code requires such alignment other code can be slower without + * correct alignment, for yet other it makes no difference. + * + * @note The linesize may be larger than the size of usable data -- there + * may be extra padding present for performance reasons. + * + * @attention In case of video, line size values can be negative to achieve + * a vertically inverted iteration over image lines. + */ + int linesize[AV_NUM_DATA_POINTERS]; + + /** + * pointers to the data planes/channels. + * + * For video, this should simply point to data[]. + * + * For planar audio, each channel has a separate data pointer, and + * linesize[0] contains the size of each channel buffer. + * For packed audio, there is just one data pointer, and linesize[0] + * contains the total size of the buffer for all channels. + * + * Note: Both data and extended_data should always be set in a valid frame, + * but for planar audio with more channels that can fit in data, + * extended_data must be used in order to access all channels. + */ + uint8_t **extended_data; + + /** + * @name Video dimensions + * Video frames only. The coded dimensions (in pixels) of the video frame, + * i.e. the size of the rectangle that contains some well-defined values. + * + * @note The part of the frame intended for display/presentation is further + * restricted by the @ref cropping "Cropping rectangle". + * @{ + */ + int width, height; + /** + * @} + */ + + /** + * number of audio samples (per channel) described by this frame + */ + int nb_samples; + + /** + * format of the frame, -1 if unknown or unset + * Values correspond to enum AVPixelFormat for video frames, + * enum AVSampleFormat for audio) + */ + int format; + +#if FF_API_FRAME_KEY + /** + * 1 -> keyframe, 0-> not + * + * @deprecated Use AV_FRAME_FLAG_KEY instead + */ + attribute_deprecated + int key_frame; +#endif + + /** + * Picture type of the frame. + */ + enum AVPictureType pict_type; + + /** + * Sample aspect ratio for the video frame, 0/1 if unknown/unspecified. + */ + AVRational sample_aspect_ratio; + + /** + * Presentation timestamp in time_base units (time when frame should be shown to user). + */ + int64_t pts; + + /** + * DTS copied from the AVPacket that triggered returning this frame. (if frame threading isn't used) + * This is also the Presentation time of this AVFrame calculated from + * only AVPacket.dts values without pts values. + */ + int64_t pkt_dts; + + /** + * Time base for the timestamps in this frame. + * In the future, this field may be set on frames output by decoders or + * filters, but its value will be by default ignored on input to encoders + * or filters. + */ + AVRational time_base; + +#if FF_API_FRAME_PICTURE_NUMBER + /** + * picture number in bitstream order + */ + attribute_deprecated + int coded_picture_number; + /** + * picture number in display order + */ + attribute_deprecated + int display_picture_number; +#endif + + /** + * quality (between 1 (good) and FF_LAMBDA_MAX (bad)) + */ + int quality; + + /** + * Frame owner's private data. + * + * This field may be set by the code that allocates/owns the frame data. + * It is then not touched by any library functions, except: + * - it is copied to other references by av_frame_copy_props() (and hence by + * av_frame_ref()); + * - it is set to NULL when the frame is cleared by av_frame_unref() + * - on the caller's explicit request. E.g. libavcodec encoders/decoders + * will copy this field to/from @ref AVPacket "AVPackets" if the caller sets + * @ref AV_CODEC_FLAG_COPY_OPAQUE. + * + * @see opaque_ref the reference-counted analogue + */ + void *opaque; + + /** + * Number of fields in this frame which should be repeated, i.e. the total + * duration of this frame should be repeat_pict + 2 normal field durations. + * + * For interlaced frames this field may be set to 1, which signals that this + * frame should be presented as 3 fields: beginning with the first field (as + * determined by AV_FRAME_FLAG_TOP_FIELD_FIRST being set or not), followed + * by the second field, and then the first field again. + * + * For progressive frames this field may be set to a multiple of 2, which + * signals that this frame's duration should be (repeat_pict + 2) / 2 + * normal frame durations. + * + * @note This field is computed from MPEG2 repeat_first_field flag and its + * associated flags, H.264 pic_struct from picture timing SEI, and + * their analogues in other codecs. Typically it should only be used when + * higher-layer timing information is not available. + */ + int repeat_pict; + +#if FF_API_INTERLACED_FRAME + /** + * The content of the picture is interlaced. + * + * @deprecated Use AV_FRAME_FLAG_INTERLACED instead + */ + attribute_deprecated + int interlaced_frame; + + /** + * If the content is interlaced, is top field displayed first. + * + * @deprecated Use AV_FRAME_FLAG_TOP_FIELD_FIRST instead + */ + attribute_deprecated + int top_field_first; +#endif + +#if FF_API_PALETTE_HAS_CHANGED + /** + * Tell user application that palette has changed from previous frame. + */ + attribute_deprecated + int palette_has_changed; +#endif + +#if FF_API_REORDERED_OPAQUE + /** + * reordered opaque 64 bits (generally an integer or a double precision float + * PTS but can be anything). + * The user sets AVCodecContext.reordered_opaque to represent the input at + * that time, + * the decoder reorders values as needed and sets AVFrame.reordered_opaque + * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque + * + * @deprecated Use AV_CODEC_FLAG_COPY_OPAQUE instead + */ + attribute_deprecated + int64_t reordered_opaque; +#endif + + /** + * Sample rate of the audio data. + */ + int sample_rate; + +#if FF_API_OLD_CHANNEL_LAYOUT + /** + * Channel layout of the audio data. + * @deprecated use ch_layout instead + */ + attribute_deprecated + uint64_t channel_layout; +#endif + + /** + * AVBuffer references backing the data for this frame. All the pointers in + * data and extended_data must point inside one of the buffers in buf or + * extended_buf. This array must be filled contiguously -- if buf[i] is + * non-NULL then buf[j] must also be non-NULL for all j < i. + * + * There may be at most one AVBuffer per data plane, so for video this array + * always contains all the references. For planar audio with more than + * AV_NUM_DATA_POINTERS channels, there may be more buffers than can fit in + * this array. Then the extra AVBufferRef pointers are stored in the + * extended_buf array. + */ + AVBufferRef *buf[AV_NUM_DATA_POINTERS]; + + /** + * For planar audio which requires more than AV_NUM_DATA_POINTERS + * AVBufferRef pointers, this array will hold all the references which + * cannot fit into AVFrame.buf. + * + * Note that this is different from AVFrame.extended_data, which always + * contains all the pointers. This array only contains the extra pointers, + * which cannot fit into AVFrame.buf. + * + * This array is always allocated using av_malloc() by whoever constructs + * the frame. It is freed in av_frame_unref(). + */ + AVBufferRef **extended_buf; + /** + * Number of elements in extended_buf. + */ + int nb_extended_buf; + + AVFrameSideData **side_data; + int nb_side_data; + +/** + * @defgroup lavu_frame_flags AV_FRAME_FLAGS + * @ingroup lavu_frame + * Flags describing additional frame properties. + * + * @{ + */ + +/** + * The frame data may be corrupted, e.g. due to decoding errors. + */ +#define AV_FRAME_FLAG_CORRUPT (1 << 0) +/** + * A flag to mark frames that are keyframes. + */ +#define AV_FRAME_FLAG_KEY (1 << 1) +/** + * A flag to mark the frames which need to be decoded, but shouldn't be output. + */ +#define AV_FRAME_FLAG_DISCARD (1 << 2) +/** + * A flag to mark frames whose content is interlaced. + */ +#define AV_FRAME_FLAG_INTERLACED (1 << 3) +/** + * A flag to mark frames where the top field is displayed first if the content + * is interlaced. + */ +#define AV_FRAME_FLAG_TOP_FIELD_FIRST (1 << 4) +/** + * @} + */ + + /** + * Frame flags, a combination of @ref lavu_frame_flags + */ + int flags; + + /** + * MPEG vs JPEG YUV range. + * - encoding: Set by user + * - decoding: Set by libavcodec + */ + enum AVColorRange color_range; + + enum AVColorPrimaries color_primaries; + + enum AVColorTransferCharacteristic color_trc; + + /** + * YUV colorspace type. + * - encoding: Set by user + * - decoding: Set by libavcodec + */ + enum AVColorSpace colorspace; + + enum AVChromaLocation chroma_location; + + /** + * frame timestamp estimated using various heuristics, in stream time base + * - encoding: unused + * - decoding: set by libavcodec, read by user. + */ + int64_t best_effort_timestamp; + +#if FF_API_FRAME_PKT + /** + * reordered pos from the last AVPacket that has been input into the decoder + * - encoding: unused + * - decoding: Read by user. + * @deprecated use AV_CODEC_FLAG_COPY_OPAQUE to pass through arbitrary user + * data from packets to frames + */ + attribute_deprecated + int64_t pkt_pos; +#endif + +#if FF_API_PKT_DURATION + /** + * duration of the corresponding packet, expressed in + * AVStream->time_base units, 0 if unknown. + * - encoding: unused + * - decoding: Read by user. + * + * @deprecated use duration instead + */ + attribute_deprecated + int64_t pkt_duration; +#endif + + /** + * metadata. + * - encoding: Set by user. + * - decoding: Set by libavcodec. + */ + AVDictionary *metadata; + + /** + * decode error flags of the frame, set to a combination of + * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there + * were errors during the decoding. + * - encoding: unused + * - decoding: set by libavcodec, read by user. + */ + int decode_error_flags; +#define FF_DECODE_ERROR_INVALID_BITSTREAM 1 +#define FF_DECODE_ERROR_MISSING_REFERENCE 2 +#define FF_DECODE_ERROR_CONCEALMENT_ACTIVE 4 +#define FF_DECODE_ERROR_DECODE_SLICES 8 + +#if FF_API_OLD_CHANNEL_LAYOUT + /** + * number of audio channels, only used for audio. + * - encoding: unused + * - decoding: Read by user. + * @deprecated use ch_layout instead + */ + attribute_deprecated + int channels; +#endif + +#if FF_API_FRAME_PKT + /** + * size of the corresponding packet containing the compressed + * frame. + * It is set to a negative value if unknown. + * - encoding: unused + * - decoding: set by libavcodec, read by user. + * @deprecated use AV_CODEC_FLAG_COPY_OPAQUE to pass through arbitrary user + * data from packets to frames + */ + attribute_deprecated + int pkt_size; +#endif + + /** + * For hwaccel-format frames, this should be a reference to the + * AVHWFramesContext describing the frame. + */ + AVBufferRef *hw_frames_ctx; + + /** + * Frame owner's private data. + * + * This field may be set by the code that allocates/owns the frame data. + * It is then not touched by any library functions, except: + * - a new reference to the underlying buffer is propagated by + * av_frame_copy_props() (and hence by av_frame_ref()); + * - it is unreferenced in av_frame_unref(); + * - on the caller's explicit request. E.g. libavcodec encoders/decoders + * will propagate a new reference to/from @ref AVPacket "AVPackets" if the + * caller sets @ref AV_CODEC_FLAG_COPY_OPAQUE. + * + * @see opaque the plain pointer analogue + */ + AVBufferRef *opaque_ref; + + /** + * @anchor cropping + * @name Cropping + * Video frames only. The number of pixels to discard from the the + * top/bottom/left/right border of the frame to obtain the sub-rectangle of + * the frame intended for presentation. + * @{ + */ + size_t crop_top; + size_t crop_bottom; + size_t crop_left; + size_t crop_right; + /** + * @} + */ + + /** + * AVBufferRef for internal use by a single libav* library. + * Must not be used to transfer data between libraries. + * Has to be NULL when ownership of the frame leaves the respective library. + * + * Code outside the FFmpeg libs should never check or change the contents of the buffer ref. + * + * FFmpeg calls av_buffer_unref() on it when the frame is unreferenced. + * av_frame_copy_props() calls create a new reference with av_buffer_ref() + * for the target frame's private_ref field. + */ + AVBufferRef *private_ref; + + /** + * Channel layout of the audio data. + */ + AVChannelLayout ch_layout; + + /** + * Duration of the frame, in the same units as pts. 0 if unknown. + */ + int64_t duration; +} AVFrame; + + +/** + * Allocate an AVFrame and set its fields to default values. The resulting + * struct must be freed using av_frame_free(). + * + * @return An AVFrame filled with default values or NULL on failure. + * + * @note this only allocates the AVFrame itself, not the data buffers. Those + * must be allocated through other means, e.g. with av_frame_get_buffer() or + * manually. + */ +AVFrame *av_frame_alloc(void); + +/** + * Free the frame and any dynamically allocated objects in it, + * e.g. extended_data. If the frame is reference counted, it will be + * unreferenced first. + * + * @param frame frame to be freed. The pointer will be set to NULL. + */ +void av_frame_free(AVFrame **frame); + +/** + * Set up a new reference to the data described by the source frame. + * + * Copy frame properties from src to dst and create a new reference for each + * AVBufferRef from src. + * + * If src is not reference counted, new buffers are allocated and the data is + * copied. + * + * @warning: dst MUST have been either unreferenced with av_frame_unref(dst), + * or newly allocated with av_frame_alloc() before calling this + * function, or undefined behavior will occur. + * + * @return 0 on success, a negative AVERROR on error + */ +int av_frame_ref(AVFrame *dst, const AVFrame *src); + +/** + * Ensure the destination frame refers to the same data described by the source + * frame, either by creating a new reference for each AVBufferRef from src if + * they differ from those in dst, by allocating new buffers and copying data if + * src is not reference counted, or by unrefencing it if src is empty. + * + * Frame properties on dst will be replaced by those from src. + * + * @return 0 on success, a negative AVERROR on error. On error, dst is + * unreferenced. + */ +int av_frame_replace(AVFrame *dst, const AVFrame *src); + +/** + * Create a new frame that references the same data as src. + * + * This is a shortcut for av_frame_alloc()+av_frame_ref(). + * + * @return newly created AVFrame on success, NULL on error. + */ +AVFrame *av_frame_clone(const AVFrame *src); + +/** + * Unreference all the buffers referenced by frame and reset the frame fields. + */ +void av_frame_unref(AVFrame *frame); + +/** + * Move everything contained in src to dst and reset src. + * + * @warning: dst is not unreferenced, but directly overwritten without reading + * or deallocating its contents. Call av_frame_unref(dst) manually + * before calling this function to ensure that no memory is leaked. + */ +void av_frame_move_ref(AVFrame *dst, AVFrame *src); + +/** + * Allocate new buffer(s) for audio or video data. + * + * The following fields must be set on frame before calling this function: + * - format (pixel format for video, sample format for audio) + * - width and height for video + * - nb_samples and ch_layout for audio + * + * This function will fill AVFrame.data and AVFrame.buf arrays and, if + * necessary, allocate and fill AVFrame.extended_data and AVFrame.extended_buf. + * For planar formats, one buffer will be allocated for each plane. + * + * @warning: if frame already has been allocated, calling this function will + * leak memory. In addition, undefined behavior can occur in certain + * cases. + * + * @param frame frame in which to store the new buffers. + * @param align Required buffer size alignment. If equal to 0, alignment will be + * chosen automatically for the current CPU. It is highly + * recommended to pass 0 here unless you know what you are doing. + * + * @return 0 on success, a negative AVERROR on error. + */ +int av_frame_get_buffer(AVFrame *frame, int align); + +/** + * Check if the frame data is writable. + * + * @return A positive value if the frame data is writable (which is true if and + * only if each of the underlying buffers has only one reference, namely the one + * stored in this frame). Return 0 otherwise. + * + * If 1 is returned the answer is valid until av_buffer_ref() is called on any + * of the underlying AVBufferRefs (e.g. through av_frame_ref() or directly). + * + * @see av_frame_make_writable(), av_buffer_is_writable() + */ +int av_frame_is_writable(AVFrame *frame); + +/** + * Ensure that the frame data is writable, avoiding data copy if possible. + * + * Do nothing if the frame is writable, allocate new buffers and copy the data + * if it is not. Non-refcounted frames behave as non-writable, i.e. a copy + * is always made. + * + * @return 0 on success, a negative AVERROR on error. + * + * @see av_frame_is_writable(), av_buffer_is_writable(), + * av_buffer_make_writable() + */ +int av_frame_make_writable(AVFrame *frame); + +/** + * Copy the frame data from src to dst. + * + * This function does not allocate anything, dst must be already initialized and + * allocated with the same parameters as src. + * + * This function only copies the frame data (i.e. the contents of the data / + * extended data arrays), not any other properties. + * + * @return >= 0 on success, a negative AVERROR on error. + */ +int av_frame_copy(AVFrame *dst, const AVFrame *src); + +/** + * Copy only "metadata" fields from src to dst. + * + * Metadata for the purpose of this function are those fields that do not affect + * the data layout in the buffers. E.g. pts, sample rate (for audio) or sample + * aspect ratio (for video), but not width/height or channel layout. + * Side data is also copied. + */ +int av_frame_copy_props(AVFrame *dst, const AVFrame *src); + +/** + * Get the buffer reference a given data plane is stored in. + * + * @param frame the frame to get the plane's buffer from + * @param plane index of the data plane of interest in frame->extended_data. + * + * @return the buffer reference that contains the plane or NULL if the input + * frame is not valid. + */ +AVBufferRef *av_frame_get_plane_buffer(const AVFrame *frame, int plane); + +/** + * Add a new side data to a frame. + * + * @param frame a frame to which the side data should be added + * @param type type of the added side data + * @param size size of the side data + * + * @return newly added side data on success, NULL on error + */ +AVFrameSideData *av_frame_new_side_data(AVFrame *frame, + enum AVFrameSideDataType type, + size_t size); + +/** + * Add a new side data to a frame from an existing AVBufferRef + * + * @param frame a frame to which the side data should be added + * @param type the type of the added side data + * @param buf an AVBufferRef to add as side data. The ownership of + * the reference is transferred to the frame. + * + * @return newly added side data on success, NULL on error. On failure + * the frame is unchanged and the AVBufferRef remains owned by + * the caller. + */ +AVFrameSideData *av_frame_new_side_data_from_buf(AVFrame *frame, + enum AVFrameSideDataType type, + AVBufferRef *buf); + +/** + * @return a pointer to the side data of a given type on success, NULL if there + * is no side data with such type in this frame. + */ +AVFrameSideData *av_frame_get_side_data(const AVFrame *frame, + enum AVFrameSideDataType type); + +/** + * Remove and free all side data instances of the given type. + */ +void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type); + + +/** + * Flags for frame cropping. + */ +enum { + /** + * Apply the maximum possible cropping, even if it requires setting the + * AVFrame.data[] entries to unaligned pointers. Passing unaligned data + * to FFmpeg API is generally not allowed, and causes undefined behavior + * (such as crashes). You can pass unaligned data only to FFmpeg APIs that + * are explicitly documented to accept it. Use this flag only if you + * absolutely know what you are doing. + */ + AV_FRAME_CROP_UNALIGNED = 1 << 0, +}; + +/** + * Crop the given video AVFrame according to its crop_left/crop_top/crop_right/ + * crop_bottom fields. If cropping is successful, the function will adjust the + * data pointers and the width/height fields, and set the crop fields to 0. + * + * In all cases, the cropping boundaries will be rounded to the inherent + * alignment of the pixel format. In some cases, such as for opaque hwaccel + * formats, the left/top cropping is ignored. The crop fields are set to 0 even + * if the cropping was rounded or ignored. + * + * @param frame the frame which should be cropped + * @param flags Some combination of AV_FRAME_CROP_* flags, or 0. + * + * @return >= 0 on success, a negative AVERROR on error. If the cropping fields + * were invalid, AVERROR(ERANGE) is returned, and nothing is changed. + */ +int av_frame_apply_cropping(AVFrame *frame, int flags); + +/** + * @return a string identifying the side data type + */ +const char *av_frame_side_data_name(enum AVFrameSideDataType type); + +/** + * @} + */ + +#endif /* AVUTIL_FRAME_H */ diff --git a/media/ffvpx/libavutil/hdr_dynamic_metadata.c b/media/ffvpx/libavutil/hdr_dynamic_metadata.c new file mode 100644 index 0000000000..7033f060c0 --- /dev/null +++ b/media/ffvpx/libavutil/hdr_dynamic_metadata.c @@ -0,0 +1,397 @@ +/** + * Copyright (c) 2018 Mohammad Izadi <moh.izadi at gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "avassert.h" +#include "hdr_dynamic_metadata.h" +#include "mem.h" +#include "libavcodec/defs.h" +#include "libavcodec/get_bits.h" +#include "libavcodec/put_bits.h" + +static const int64_t luminance_den = 1; +static const int32_t peak_luminance_den = 15; +static const int64_t rgb_den = 100000; +static const int32_t fraction_pixel_den = 1000; +static const int32_t knee_point_den = 4095; +static const int32_t bezier_anchor_den = 1023; +static const int32_t saturation_weight_den = 8; + +AVDynamicHDRPlus *av_dynamic_hdr_plus_alloc(size_t *size) +{ + AVDynamicHDRPlus *hdr_plus = av_mallocz(sizeof(AVDynamicHDRPlus)); + if (!hdr_plus) + return NULL; + + if (size) + *size = sizeof(*hdr_plus); + + return hdr_plus; +} + +AVDynamicHDRPlus *av_dynamic_hdr_plus_create_side_data(AVFrame *frame) +{ + AVFrameSideData *side_data = av_frame_new_side_data(frame, + AV_FRAME_DATA_DYNAMIC_HDR_PLUS, + sizeof(AVDynamicHDRPlus)); + if (!side_data) + return NULL; + + memset(side_data->data, 0, sizeof(AVDynamicHDRPlus)); + + return (AVDynamicHDRPlus *)side_data->data; +} + +int av_dynamic_hdr_plus_from_t35(AVDynamicHDRPlus *s, const uint8_t *data, + size_t size) +{ + uint8_t padded_buf[AV_HDR_PLUS_MAX_PAYLOAD_SIZE + AV_INPUT_BUFFER_PADDING_SIZE]; + GetBitContext gbc, *gb = &gbc; + int ret; + + if (!s) + return AVERROR(ENOMEM); + + if (size > AV_HDR_PLUS_MAX_PAYLOAD_SIZE) + return AVERROR(EINVAL); + + memcpy(padded_buf, data, size); + // Zero-initialize the buffer padding to avoid overreads into uninitialized data. + memset(padded_buf + size, 0, AV_INPUT_BUFFER_PADDING_SIZE); + + ret = init_get_bits8(gb, padded_buf, size); + if (ret < 0) + return ret; + + if (get_bits_left(gb) < 10) + return AVERROR_INVALIDDATA; + + s->application_version = get_bits(gb, 8); + s->num_windows = get_bits(gb, 2); + + if (s->num_windows < 1 || s->num_windows > 3) { + return AVERROR_INVALIDDATA; + } + + if (get_bits_left(gb) < ((19 * 8 + 1) * (s->num_windows - 1))) + return AVERROR_INVALIDDATA; + + for (int w = 1; w < s->num_windows; w++) { + // The corners are set to absolute coordinates here. They should be + // converted to the relative coordinates (in [0, 1]) in the decoder. + AVHDRPlusColorTransformParams *params = &s->params[w]; + params->window_upper_left_corner_x = + (AVRational){get_bits(gb, 16), 1}; + params->window_upper_left_corner_y = + (AVRational){get_bits(gb, 16), 1}; + params->window_lower_right_corner_x = + (AVRational){get_bits(gb, 16), 1}; + params->window_lower_right_corner_y = + (AVRational){get_bits(gb, 16), 1}; + + params->center_of_ellipse_x = get_bits(gb, 16); + params->center_of_ellipse_y = get_bits(gb, 16); + params->rotation_angle = get_bits(gb, 8); + params->semimajor_axis_internal_ellipse = get_bits(gb, 16); + params->semimajor_axis_external_ellipse = get_bits(gb, 16); + params->semiminor_axis_external_ellipse = get_bits(gb, 16); + params->overlap_process_option = get_bits1(gb); + } + + if (get_bits_left(gb) < 28) + return AVERROR_INVALIDDATA; + + s->targeted_system_display_maximum_luminance = + (AVRational){get_bits_long(gb, 27), luminance_den}; + s->targeted_system_display_actual_peak_luminance_flag = get_bits1(gb); + + if (s->targeted_system_display_actual_peak_luminance_flag) { + int rows, cols; + if (get_bits_left(gb) < 10) + return AVERROR_INVALIDDATA; + rows = get_bits(gb, 5); + cols = get_bits(gb, 5); + if (((rows < 2) || (rows > 25)) || ((cols < 2) || (cols > 25))) { + return AVERROR_INVALIDDATA; + } + s->num_rows_targeted_system_display_actual_peak_luminance = rows; + s->num_cols_targeted_system_display_actual_peak_luminance = cols; + + if (get_bits_left(gb) < (rows * cols * 4)) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < rows; i++) { + for (int j = 0; j < cols; j++) { + s->targeted_system_display_actual_peak_luminance[i][j] = + (AVRational){get_bits(gb, 4), peak_luminance_den}; + } + } + } + for (int w = 0; w < s->num_windows; w++) { + AVHDRPlusColorTransformParams *params = &s->params[w]; + if (get_bits_left(gb) < (3 * 17 + 17 + 4)) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < 3; i++) { + params->maxscl[i] = + (AVRational){get_bits(gb, 17), rgb_den}; + } + params->average_maxrgb = + (AVRational){get_bits(gb, 17), rgb_den}; + params->num_distribution_maxrgb_percentiles = get_bits(gb, 4); + + if (get_bits_left(gb) < + (params->num_distribution_maxrgb_percentiles * 24)) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < params->num_distribution_maxrgb_percentiles; i++) { + params->distribution_maxrgb[i].percentage = get_bits(gb, 7); + params->distribution_maxrgb[i].percentile = + (AVRational){get_bits(gb, 17), rgb_den}; + } + + if (get_bits_left(gb) < 10) + return AVERROR_INVALIDDATA; + + params->fraction_bright_pixels = (AVRational){get_bits(gb, 10), fraction_pixel_den}; + } + if (get_bits_left(gb) < 1) + return AVERROR_INVALIDDATA; + s->mastering_display_actual_peak_luminance_flag = get_bits1(gb); + if (s->mastering_display_actual_peak_luminance_flag) { + int rows, cols; + if (get_bits_left(gb) < 10) + return AVERROR_INVALIDDATA; + rows = get_bits(gb, 5); + cols = get_bits(gb, 5); + if (((rows < 2) || (rows > 25)) || ((cols < 2) || (cols > 25))) { + return AVERROR_INVALIDDATA; + } + s->num_rows_mastering_display_actual_peak_luminance = rows; + s->num_cols_mastering_display_actual_peak_luminance = cols; + + if (get_bits_left(gb) < (rows * cols * 4)) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < rows; i++) { + for (int j = 0; j < cols; j++) { + s->mastering_display_actual_peak_luminance[i][j] = + (AVRational){get_bits(gb, 4), peak_luminance_den}; + } + } + } + + for (int w = 0; w < s->num_windows; w++) { + AVHDRPlusColorTransformParams *params = &s->params[w]; + if (get_bits_left(gb) < 1) + return AVERROR_INVALIDDATA; + + params->tone_mapping_flag = get_bits1(gb); + if (params->tone_mapping_flag) { + if (get_bits_left(gb) < 28) + return AVERROR_INVALIDDATA; + + params->knee_point_x = + (AVRational){get_bits(gb, 12), knee_point_den}; + params->knee_point_y = + (AVRational){get_bits(gb, 12), knee_point_den}; + params->num_bezier_curve_anchors = get_bits(gb, 4); + + if (get_bits_left(gb) < (params->num_bezier_curve_anchors * 10)) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < params->num_bezier_curve_anchors; i++) { + params->bezier_curve_anchors[i] = + (AVRational){get_bits(gb, 10), bezier_anchor_den}; + } + } + + if (get_bits_left(gb) < 1) + return AVERROR_INVALIDDATA; + params->color_saturation_mapping_flag = get_bits1(gb); + if (params->color_saturation_mapping_flag) { + if (get_bits_left(gb) < 6) + return AVERROR_INVALIDDATA; + params->color_saturation_weight = + (AVRational){get_bits(gb, 6), saturation_weight_den}; + } + } + + return 0; +} + +int av_dynamic_hdr_plus_to_t35(const AVDynamicHDRPlus *s, uint8_t **data, size_t *size) +{ + uint8_t *buf; + size_t size_bits, size_bytes; + PutBitContext pbc, *pb = &pbc; + + if (!s) + return AVERROR(EINVAL); + if ((!data || *data) && !size) + return AVERROR(EINVAL); + + /** + * Buffer size per CTA-861-H p.253-254: + * 48 header bits (excluded from the serialized payload) + * 8 bits for application_mode + * 2 bits for num_windows + * 153 bits for window geometry, for each window above 1 + * 27 bits for targeted_system_display_maximum_luminance + * 1-2511 bits for targeted system display peak luminance information + * 82-442 bits per window for pixel distribution information + * 1-2511 bits for mastering display peak luminance information + * 1-179 bits per window for tonemapping information + * 1-7 bits per window for color saturation mapping information + * Total: 123-7249 bits, excluding trimmed header bits + */ + size_bits = 8; + + size_bits += 2; + + for (int w = 1; w < s->num_windows; w++) + size_bits += 153; + + size_bits += 27; + + size_bits += 1; + if (s->targeted_system_display_actual_peak_luminance_flag) + size_bits += 10 + + s->num_rows_targeted_system_display_actual_peak_luminance * + s->num_cols_targeted_system_display_actual_peak_luminance * 4; + + for (int w = 0; w < s->num_windows; w++) + size_bits += 72 + s->params[w].num_distribution_maxrgb_percentiles * 24 + 10; + + size_bits += 1; + if (s->mastering_display_actual_peak_luminance_flag) + size_bits += 10 + + s->num_rows_mastering_display_actual_peak_luminance * + s->num_cols_mastering_display_actual_peak_luminance * 4; + + for (int w = 0; w < s->num_windows; w++) { + size_bits += 1; + if (s->params[w].tone_mapping_flag) + size_bits += 28 + s->params[w].num_bezier_curve_anchors * 10; + + size_bits += 1; + if (s->params[w].color_saturation_mapping_flag) + size_bits += 6; + } + + size_bytes = (size_bits + 7) / 8; + + av_assert0(size_bytes <= AV_HDR_PLUS_MAX_PAYLOAD_SIZE); + + if (!data) { + *size = size_bytes; + return 0; + } else if (*data) { + if (*size < size_bytes) + return AVERROR_BUFFER_TOO_SMALL; + buf = *data; + } else { + buf = av_malloc(size_bytes); + if (!buf) + return AVERROR(ENOMEM); + } + + init_put_bits(pb, buf, size_bytes); + + // application_mode is set to Application Version 1 + put_bits(pb, 8, 1); + + // Payload as per CTA-861-H p.253-254 + put_bits(pb, 2, s->num_windows); + + for (int w = 1; w < s->num_windows; w++) { + put_bits(pb, 16, s->params[w].window_upper_left_corner_x.num / s->params[w].window_upper_left_corner_x.den); + put_bits(pb, 16, s->params[w].window_upper_left_corner_y.num / s->params[w].window_upper_left_corner_y.den); + put_bits(pb, 16, s->params[w].window_lower_right_corner_x.num / s->params[w].window_lower_right_corner_x.den); + put_bits(pb, 16, s->params[w].window_lower_right_corner_y.num / s->params[w].window_lower_right_corner_y.den); + put_bits(pb, 16, s->params[w].center_of_ellipse_x); + put_bits(pb, 16, s->params[w].center_of_ellipse_y); + put_bits(pb, 8, s->params[w].rotation_angle); + put_bits(pb, 16, s->params[w].semimajor_axis_internal_ellipse); + put_bits(pb, 16, s->params[w].semimajor_axis_external_ellipse); + put_bits(pb, 16, s->params[w].semiminor_axis_external_ellipse); + put_bits(pb, 1, s->params[w].overlap_process_option); + } + + put_bits(pb, 27, s->targeted_system_display_maximum_luminance.num * luminance_den / + s->targeted_system_display_maximum_luminance.den); + put_bits(pb, 1, s->targeted_system_display_actual_peak_luminance_flag); + if (s->targeted_system_display_actual_peak_luminance_flag) { + put_bits(pb, 5, s->num_rows_targeted_system_display_actual_peak_luminance); + put_bits(pb, 5, s->num_cols_targeted_system_display_actual_peak_luminance); + for (int i = 0; i < s->num_rows_targeted_system_display_actual_peak_luminance; i++) { + for (int j = 0; j < s->num_cols_targeted_system_display_actual_peak_luminance; j++) + put_bits(pb, 4, s->targeted_system_display_actual_peak_luminance[i][j].num * peak_luminance_den / + s->targeted_system_display_actual_peak_luminance[i][j].den); + } + } + + for (int w = 0; w < s->num_windows; w++) { + for (int i = 0; i < 3; i++) + put_bits(pb, 17, s->params[w].maxscl[i].num * rgb_den / s->params[w].maxscl[i].den); + put_bits(pb, 17, s->params[w].average_maxrgb.num * rgb_den / s->params[w].average_maxrgb.den); + put_bits(pb, 4, s->params[w].num_distribution_maxrgb_percentiles); + for (int i = 0; i < s->params[w].num_distribution_maxrgb_percentiles; i++) { + put_bits(pb, 7, s->params[w].distribution_maxrgb[i].percentage); + put_bits(pb, 17, s->params[w].distribution_maxrgb[i].percentile.num * rgb_den / + s->params[w].distribution_maxrgb[i].percentile.den); + } + put_bits(pb, 10, s->params[w].fraction_bright_pixels.num * fraction_pixel_den / + s->params[w].fraction_bright_pixels.den); + } + + put_bits(pb, 1, s->mastering_display_actual_peak_luminance_flag); + if (s->mastering_display_actual_peak_luminance_flag) { + put_bits(pb, 5, s->num_rows_mastering_display_actual_peak_luminance); + put_bits(pb, 5, s->num_cols_mastering_display_actual_peak_luminance); + for (int i = 0; i < s->num_rows_mastering_display_actual_peak_luminance; i++) { + for (int j = 0; j < s->num_cols_mastering_display_actual_peak_luminance; j++) + put_bits(pb, 4, s->mastering_display_actual_peak_luminance[i][j].num * peak_luminance_den / + s->mastering_display_actual_peak_luminance[i][j].den); + } + } + + for (int w = 0; w < s->num_windows; w++) { + put_bits(pb, 1, s->params[w].tone_mapping_flag); + if (s->params[w].tone_mapping_flag) { + put_bits(pb, 12, s->params[w].knee_point_x.num * knee_point_den / s->params[w].knee_point_x.den); + put_bits(pb, 12, s->params[w].knee_point_y.num * knee_point_den / s->params[w].knee_point_y.den); + put_bits(pb, 4, s->params[w].num_bezier_curve_anchors); + for (int i = 0; i < s->params[w].num_bezier_curve_anchors; i++) + put_bits(pb, 10, s->params[w].bezier_curve_anchors[i].num * bezier_anchor_den / + s->params[w].bezier_curve_anchors[i].den); + put_bits(pb, 1, s->params[w].color_saturation_mapping_flag); + if (s->params[w].color_saturation_mapping_flag) + put_bits(pb, 6, s->params[w].color_saturation_weight.num * saturation_weight_den / + s->params[w].color_saturation_weight.den); + } + } + + flush_put_bits(pb); + + *data = buf; + if (size) + *size = size_bytes; + return 0; +} diff --git a/media/ffvpx/libavutil/hdr_dynamic_metadata.h b/media/ffvpx/libavutil/hdr_dynamic_metadata.h new file mode 100644 index 0000000000..09e9d8bbcc --- /dev/null +++ b/media/ffvpx/libavutil/hdr_dynamic_metadata.h @@ -0,0 +1,376 @@ +/* + * Copyright (c) 2018 Mohammad Izadi <moh.izadi at gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_HDR_DYNAMIC_METADATA_H +#define AVUTIL_HDR_DYNAMIC_METADATA_H + +#include "frame.h" +#include "rational.h" + +/** + * Option for overlapping elliptical pixel selectors in an image. + */ +enum AVHDRPlusOverlapProcessOption { + AV_HDR_PLUS_OVERLAP_PROCESS_WEIGHTED_AVERAGING = 0, + AV_HDR_PLUS_OVERLAP_PROCESS_LAYERING = 1, +}; + +/** + * Represents the percentile at a specific percentage in + * a distribution. + */ +typedef struct AVHDRPlusPercentile { + /** + * The percentage value corresponding to a specific percentile linearized + * RGB value in the processing window in the scene. The value shall be in + * the range of 0 to100, inclusive. + */ + uint8_t percentage; + + /** + * The linearized maxRGB value at a specific percentile in the processing + * window in the scene. The value shall be in the range of 0 to 1, inclusive + * and in multiples of 0.00001. + */ + AVRational percentile; +} AVHDRPlusPercentile; + +/** + * Color transform parameters at a processing window in a dynamic metadata for + * SMPTE 2094-40. + */ +typedef struct AVHDRPlusColorTransformParams { + /** + * The relative x coordinate of the top left pixel of the processing + * window. The value shall be in the range of 0 and 1, inclusive and + * in multiples of 1/(width of Picture - 1). The value 1 corresponds + * to the absolute coordinate of width of Picture - 1. The value for + * first processing window shall be 0. + */ + AVRational window_upper_left_corner_x; + + /** + * The relative y coordinate of the top left pixel of the processing + * window. The value shall be in the range of 0 and 1, inclusive and + * in multiples of 1/(height of Picture - 1). The value 1 corresponds + * to the absolute coordinate of height of Picture - 1. The value for + * first processing window shall be 0. + */ + AVRational window_upper_left_corner_y; + + /** + * The relative x coordinate of the bottom right pixel of the processing + * window. The value shall be in the range of 0 and 1, inclusive and + * in multiples of 1/(width of Picture - 1). The value 1 corresponds + * to the absolute coordinate of width of Picture - 1. The value for + * first processing window shall be 1. + */ + AVRational window_lower_right_corner_x; + + /** + * The relative y coordinate of the bottom right pixel of the processing + * window. The value shall be in the range of 0 and 1, inclusive and + * in multiples of 1/(height of Picture - 1). The value 1 corresponds + * to the absolute coordinate of height of Picture - 1. The value for + * first processing window shall be 1. + */ + AVRational window_lower_right_corner_y; + + /** + * The x coordinate of the center position of the concentric internal and + * external ellipses of the elliptical pixel selector in the processing + * window. The value shall be in the range of 0 to (width of Picture - 1), + * inclusive and in multiples of 1 pixel. + */ + uint16_t center_of_ellipse_x; + + /** + * The y coordinate of the center position of the concentric internal and + * external ellipses of the elliptical pixel selector in the processing + * window. The value shall be in the range of 0 to (height of Picture - 1), + * inclusive and in multiples of 1 pixel. + */ + uint16_t center_of_ellipse_y; + + /** + * The clockwise rotation angle in degree of arc with respect to the + * positive direction of the x-axis of the concentric internal and external + * ellipses of the elliptical pixel selector in the processing window. The + * value shall be in the range of 0 to 180, inclusive and in multiples of 1. + */ + uint8_t rotation_angle; + + /** + * The semi-major axis value of the internal ellipse of the elliptical pixel + * selector in amount of pixels in the processing window. The value shall be + * in the range of 1 to 65535, inclusive and in multiples of 1 pixel. + */ + uint16_t semimajor_axis_internal_ellipse; + + /** + * The semi-major axis value of the external ellipse of the elliptical pixel + * selector in amount of pixels in the processing window. The value + * shall not be less than semimajor_axis_internal_ellipse of the current + * processing window. The value shall be in the range of 1 to 65535, + * inclusive and in multiples of 1 pixel. + */ + uint16_t semimajor_axis_external_ellipse; + + /** + * The semi-minor axis value of the external ellipse of the elliptical pixel + * selector in amount of pixels in the processing window. The value shall be + * in the range of 1 to 65535, inclusive and in multiples of 1 pixel. + */ + uint16_t semiminor_axis_external_ellipse; + + /** + * Overlap process option indicates one of the two methods of combining + * rendered pixels in the processing window in an image with at least one + * elliptical pixel selector. For overlapping elliptical pixel selectors + * in an image, overlap_process_option shall have the same value. + */ + enum AVHDRPlusOverlapProcessOption overlap_process_option; + + /** + * The maximum of the color components of linearized RGB values in the + * processing window in the scene. The values should be in the range of 0 to + * 1, inclusive and in multiples of 0.00001. maxscl[ 0 ], maxscl[ 1 ], and + * maxscl[ 2 ] are corresponding to R, G, B color components respectively. + */ + AVRational maxscl[3]; + + /** + * The average of linearized maxRGB values in the processing window in the + * scene. The value should be in the range of 0 to 1, inclusive and in + * multiples of 0.00001. + */ + AVRational average_maxrgb; + + /** + * The number of linearized maxRGB values at given percentiles in the + * processing window in the scene. The maximum value shall be 15. + */ + uint8_t num_distribution_maxrgb_percentiles; + + /** + * The linearized maxRGB values at given percentiles in the + * processing window in the scene. + */ + AVHDRPlusPercentile distribution_maxrgb[15]; + + /** + * The fraction of selected pixels in the image that contains the brightest + * pixel in the scene. The value shall be in the range of 0 to 1, inclusive + * and in multiples of 0.001. + */ + AVRational fraction_bright_pixels; + + /** + * This flag indicates that the metadata for the tone mapping function in + * the processing window is present (for value of 1). + */ + uint8_t tone_mapping_flag; + + /** + * The x coordinate of the separation point between the linear part and the + * curved part of the tone mapping function. The value shall be in the range + * of 0 to 1, excluding 0 and in multiples of 1/4095. + */ + AVRational knee_point_x; + + /** + * The y coordinate of the separation point between the linear part and the + * curved part of the tone mapping function. The value shall be in the range + * of 0 to 1, excluding 0 and in multiples of 1/4095. + */ + AVRational knee_point_y; + + /** + * The number of the intermediate anchor parameters of the tone mapping + * function in the processing window. The maximum value shall be 15. + */ + uint8_t num_bezier_curve_anchors; + + /** + * The intermediate anchor parameters of the tone mapping function in the + * processing window in the scene. The values should be in the range of 0 + * to 1, inclusive and in multiples of 1/1023. + */ + AVRational bezier_curve_anchors[15]; + + /** + * This flag shall be equal to 0 in bitstreams conforming to this version of + * this Specification. Other values are reserved for future use. + */ + uint8_t color_saturation_mapping_flag; + + /** + * The color saturation gain in the processing window in the scene. The + * value shall be in the range of 0 to 63/8, inclusive and in multiples of + * 1/8. The default value shall be 1. + */ + AVRational color_saturation_weight; +} AVHDRPlusColorTransformParams; + +/** + * This struct represents dynamic metadata for color volume transform - + * application 4 of SMPTE 2094-40:2016 standard. + * + * To be used as payload of a AVFrameSideData or AVPacketSideData with the + * appropriate type. + * + * @note The struct should be allocated with + * av_dynamic_hdr_plus_alloc() and its size is not a part of + * the public ABI. + */ +typedef struct AVDynamicHDRPlus { + /** + * Country code by Rec. ITU-T T.35 Annex A. The value shall be 0xB5. + */ + uint8_t itu_t_t35_country_code; + + /** + * Application version in the application defining document in ST-2094 + * suite. The value shall be set to 0. + */ + uint8_t application_version; + + /** + * The number of processing windows. The value shall be in the range + * of 1 to 3, inclusive. + */ + uint8_t num_windows; + + /** + * The color transform parameters for every processing window. + */ + AVHDRPlusColorTransformParams params[3]; + + /** + * The nominal maximum display luminance of the targeted system display, + * in units of 0.0001 candelas per square metre. The value shall be in + * the range of 0 to 10000, inclusive. + */ + AVRational targeted_system_display_maximum_luminance; + + /** + * This flag shall be equal to 0 in bit streams conforming to this version + * of this Specification. The value 1 is reserved for future use. + */ + uint8_t targeted_system_display_actual_peak_luminance_flag; + + /** + * The number of rows in the targeted system_display_actual_peak_luminance + * array. The value shall be in the range of 2 to 25, inclusive. + */ + uint8_t num_rows_targeted_system_display_actual_peak_luminance; + + /** + * The number of columns in the + * targeted_system_display_actual_peak_luminance array. The value shall be + * in the range of 2 to 25, inclusive. + */ + uint8_t num_cols_targeted_system_display_actual_peak_luminance; + + /** + * The normalized actual peak luminance of the targeted system display. The + * values should be in the range of 0 to 1, inclusive and in multiples of + * 1/15. + */ + AVRational targeted_system_display_actual_peak_luminance[25][25]; + + /** + * This flag shall be equal to 0 in bitstreams conforming to this version of + * this Specification. The value 1 is reserved for future use. + */ + uint8_t mastering_display_actual_peak_luminance_flag; + + /** + * The number of rows in the mastering_display_actual_peak_luminance array. + * The value shall be in the range of 2 to 25, inclusive. + */ + uint8_t num_rows_mastering_display_actual_peak_luminance; + + /** + * The number of columns in the mastering_display_actual_peak_luminance + * array. The value shall be in the range of 2 to 25, inclusive. + */ + uint8_t num_cols_mastering_display_actual_peak_luminance; + + /** + * The normalized actual peak luminance of the mastering display used for + * mastering the image essence. The values should be in the range of 0 to 1, + * inclusive and in multiples of 1/15. + */ + AVRational mastering_display_actual_peak_luminance[25][25]; +} AVDynamicHDRPlus; + +/** + * Allocate an AVDynamicHDRPlus structure and set its fields to + * default values. The resulting struct can be freed using av_freep(). + * + * @return An AVDynamicHDRPlus filled with default values or NULL + * on failure. + */ +AVDynamicHDRPlus *av_dynamic_hdr_plus_alloc(size_t *size); + +/** + * Allocate a complete AVDynamicHDRPlus and add it to the frame. + * @param frame The frame which side data is added to. + * + * @return The AVDynamicHDRPlus structure to be filled by caller or NULL + * on failure. + */ +AVDynamicHDRPlus *av_dynamic_hdr_plus_create_side_data(AVFrame *frame); + +/** + * Parse the user data registered ITU-T T.35 to AVbuffer (AVDynamicHDRPlus). + * The T.35 buffer must begin with the application mode, skipping the + * country code, terminal provider codes, and application identifier. + * @param s A pointer containing the decoded AVDynamicHDRPlus structure. + * @param data The byte array containing the raw ITU-T T.35 data. + * @param size Size of the data array in bytes. + * + * @return >= 0 on success. Otherwise, returns the appropriate AVERROR. + */ +int av_dynamic_hdr_plus_from_t35(AVDynamicHDRPlus *s, const uint8_t *data, + size_t size); + +#define AV_HDR_PLUS_MAX_PAYLOAD_SIZE 907 + +/** + * Serialize dynamic HDR10+ metadata to a user data registered ITU-T T.35 buffer, + * excluding the first 48 bytes of the header, and beginning with the application mode. + * @param s A pointer containing the decoded AVDynamicHDRPlus structure. + * @param data[in,out] A pointer to pointer to a byte buffer to be filled with the + * serialized metadata. + * If *data is NULL, a buffer be will be allocated and a pointer to + * it stored in its place. The caller assumes ownership of the buffer. + * May be NULL, in which case the function will only store the + * required buffer size in *size. + * @param size[in,out] A pointer to a size to be set to the returned buffer's size. + * If *data is not NULL, *size must contain the size of the input + * buffer. May be NULL only if *data is NULL. + * + * @return >= 0 on success. Otherwise, returns the appropriate AVERROR. + */ +int av_dynamic_hdr_plus_to_t35(const AVDynamicHDRPlus *s, uint8_t **data, size_t *size); + +#endif /* AVUTIL_HDR_DYNAMIC_METADATA_H */ diff --git a/media/ffvpx/libavutil/hwcontext.c b/media/ffvpx/libavutil/hwcontext.c new file mode 100644 index 0000000000..e23bad230f --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext.c @@ -0,0 +1,957 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include "avassert.h" +#include "buffer.h" +#include "common.h" +#include "hwcontext.h" +#include "hwcontext_internal.h" +#include "imgutils.h" +#include "log.h" +#include "mem.h" +#include "pixdesc.h" +#include "pixfmt.h" + +static const HWContextType * const hw_table[] = { +#if CONFIG_CUDA + &ff_hwcontext_type_cuda, +#endif +#if CONFIG_D3D11VA + &ff_hwcontext_type_d3d11va, +#endif +#if CONFIG_D3D12VA + &ff_hwcontext_type_d3d12va, +#endif +#if CONFIG_LIBDRM + &ff_hwcontext_type_drm, +#endif +#if CONFIG_DXVA2 + &ff_hwcontext_type_dxva2, +#endif +#if CONFIG_OPENCL + &ff_hwcontext_type_opencl, +#endif +#if CONFIG_QSV + &ff_hwcontext_type_qsv, +#endif +#if CONFIG_VAAPI + &ff_hwcontext_type_vaapi, +#endif +#if CONFIG_VDPAU + &ff_hwcontext_type_vdpau, +#endif +#if CONFIG_VIDEOTOOLBOX + &ff_hwcontext_type_videotoolbox, +#endif +#if CONFIG_MEDIACODEC + &ff_hwcontext_type_mediacodec, +#endif +#if CONFIG_VULKAN + &ff_hwcontext_type_vulkan, +#endif + NULL, +}; + +static const char *const hw_type_names[] = { + [AV_HWDEVICE_TYPE_CUDA] = "cuda", + [AV_HWDEVICE_TYPE_DRM] = "drm", + [AV_HWDEVICE_TYPE_DXVA2] = "dxva2", + [AV_HWDEVICE_TYPE_D3D11VA] = "d3d11va", + [AV_HWDEVICE_TYPE_D3D12VA] = "d3d12va", + [AV_HWDEVICE_TYPE_OPENCL] = "opencl", + [AV_HWDEVICE_TYPE_QSV] = "qsv", + [AV_HWDEVICE_TYPE_VAAPI] = "vaapi", + [AV_HWDEVICE_TYPE_VDPAU] = "vdpau", + [AV_HWDEVICE_TYPE_VIDEOTOOLBOX] = "videotoolbox", + [AV_HWDEVICE_TYPE_MEDIACODEC] = "mediacodec", + [AV_HWDEVICE_TYPE_VULKAN] = "vulkan", +}; + +enum AVHWDeviceType av_hwdevice_find_type_by_name(const char *name) +{ + int type; + for (type = 0; type < FF_ARRAY_ELEMS(hw_type_names); type++) { + if (hw_type_names[type] && !strcmp(hw_type_names[type], name)) + return type; + } + return AV_HWDEVICE_TYPE_NONE; +} + +const char *av_hwdevice_get_type_name(enum AVHWDeviceType type) +{ + if (type > AV_HWDEVICE_TYPE_NONE && + type < FF_ARRAY_ELEMS(hw_type_names)) + return hw_type_names[type]; + else + return NULL; +} + +enum AVHWDeviceType av_hwdevice_iterate_types(enum AVHWDeviceType prev) +{ + enum AVHWDeviceType next; + int i, set = 0; + for (i = 0; hw_table[i]; i++) { + if (prev != AV_HWDEVICE_TYPE_NONE && hw_table[i]->type <= prev) + continue; + if (!set || hw_table[i]->type < next) { + next = hw_table[i]->type; + set = 1; + } + } + return set ? next : AV_HWDEVICE_TYPE_NONE; +} + +static const AVClass hwdevice_ctx_class = { + .class_name = "AVHWDeviceContext", + .item_name = av_default_item_name, + .version = LIBAVUTIL_VERSION_INT, +}; + +static void hwdevice_ctx_free(void *opaque, uint8_t *data) +{ + AVHWDeviceContext *ctx = (AVHWDeviceContext*)data; + + /* uninit might still want access the hw context and the user + * free() callback might destroy it, so uninit has to be called first */ + if (ctx->internal->hw_type->device_uninit) + ctx->internal->hw_type->device_uninit(ctx); + + if (ctx->free) + ctx->free(ctx); + + av_buffer_unref(&ctx->internal->source_device); + + av_freep(&ctx->hwctx); + av_freep(&ctx->internal->priv); + av_freep(&ctx->internal); + av_freep(&ctx); +} + +AVBufferRef *av_hwdevice_ctx_alloc(enum AVHWDeviceType type) +{ + AVHWDeviceContext *ctx; + AVBufferRef *buf; + const HWContextType *hw_type = NULL; + int i; + + for (i = 0; hw_table[i]; i++) { + if (hw_table[i]->type == type) { + hw_type = hw_table[i]; + break; + } + } + if (!hw_type) + return NULL; + + ctx = av_mallocz(sizeof(*ctx)); + if (!ctx) + return NULL; + + ctx->internal = av_mallocz(sizeof(*ctx->internal)); + if (!ctx->internal) + goto fail; + + if (hw_type->device_priv_size) { + ctx->internal->priv = av_mallocz(hw_type->device_priv_size); + if (!ctx->internal->priv) + goto fail; + } + + if (hw_type->device_hwctx_size) { + ctx->hwctx = av_mallocz(hw_type->device_hwctx_size); + if (!ctx->hwctx) + goto fail; + } + + buf = av_buffer_create((uint8_t*)ctx, sizeof(*ctx), + hwdevice_ctx_free, NULL, + AV_BUFFER_FLAG_READONLY); + if (!buf) + goto fail; + + ctx->type = type; + ctx->av_class = &hwdevice_ctx_class; + + ctx->internal->hw_type = hw_type; + + return buf; + +fail: + if (ctx->internal) + av_freep(&ctx->internal->priv); + av_freep(&ctx->internal); + av_freep(&ctx->hwctx); + av_freep(&ctx); + return NULL; +} + +int av_hwdevice_ctx_init(AVBufferRef *ref) +{ + AVHWDeviceContext *ctx = (AVHWDeviceContext*)ref->data; + int ret; + + if (ctx->internal->hw_type->device_init) { + ret = ctx->internal->hw_type->device_init(ctx); + if (ret < 0) + goto fail; + } + + return 0; +fail: + if (ctx->internal->hw_type->device_uninit) + ctx->internal->hw_type->device_uninit(ctx); + return ret; +} + +static const AVClass hwframe_ctx_class = { + .class_name = "AVHWFramesContext", + .item_name = av_default_item_name, + .version = LIBAVUTIL_VERSION_INT, +}; + +static void hwframe_ctx_free(void *opaque, uint8_t *data) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)data; + + if (ctx->internal->pool_internal) + av_buffer_pool_uninit(&ctx->internal->pool_internal); + + if (ctx->internal->hw_type->frames_uninit) + ctx->internal->hw_type->frames_uninit(ctx); + + if (ctx->free) + ctx->free(ctx); + + av_buffer_unref(&ctx->internal->source_frames); + + av_buffer_unref(&ctx->device_ref); + + av_freep(&ctx->hwctx); + av_freep(&ctx->internal->priv); + av_freep(&ctx->internal); + av_freep(&ctx); +} + +AVBufferRef *av_hwframe_ctx_alloc(AVBufferRef *device_ref_in) +{ + AVHWDeviceContext *device_ctx = (AVHWDeviceContext*)device_ref_in->data; + const HWContextType *hw_type = device_ctx->internal->hw_type; + AVHWFramesContext *ctx; + AVBufferRef *buf, *device_ref = NULL; + + ctx = av_mallocz(sizeof(*ctx)); + if (!ctx) + return NULL; + + ctx->internal = av_mallocz(sizeof(*ctx->internal)); + if (!ctx->internal) + goto fail; + + if (hw_type->frames_priv_size) { + ctx->internal->priv = av_mallocz(hw_type->frames_priv_size); + if (!ctx->internal->priv) + goto fail; + } + + if (hw_type->frames_hwctx_size) { + ctx->hwctx = av_mallocz(hw_type->frames_hwctx_size); + if (!ctx->hwctx) + goto fail; + } + + device_ref = av_buffer_ref(device_ref_in); + if (!device_ref) + goto fail; + + buf = av_buffer_create((uint8_t*)ctx, sizeof(*ctx), + hwframe_ctx_free, NULL, + AV_BUFFER_FLAG_READONLY); + if (!buf) + goto fail; + + ctx->av_class = &hwframe_ctx_class; + ctx->device_ref = device_ref; + ctx->device_ctx = device_ctx; + ctx->format = AV_PIX_FMT_NONE; + ctx->sw_format = AV_PIX_FMT_NONE; + + ctx->internal->hw_type = hw_type; + + return buf; + +fail: + if (device_ref) + av_buffer_unref(&device_ref); + if (ctx->internal) + av_freep(&ctx->internal->priv); + av_freep(&ctx->internal); + av_freep(&ctx->hwctx); + av_freep(&ctx); + return NULL; +} + +static int hwframe_pool_prealloc(AVBufferRef *ref) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)ref->data; + AVFrame **frames; + int i, ret = 0; + + frames = av_calloc(ctx->initial_pool_size, sizeof(*frames)); + if (!frames) + return AVERROR(ENOMEM); + + for (i = 0; i < ctx->initial_pool_size; i++) { + frames[i] = av_frame_alloc(); + if (!frames[i]) + goto fail; + + ret = av_hwframe_get_buffer(ref, frames[i], 0); + if (ret < 0) + goto fail; + } + +fail: + for (i = 0; i < ctx->initial_pool_size; i++) + av_frame_free(&frames[i]); + av_freep(&frames); + + return ret; +} + +int av_hwframe_ctx_init(AVBufferRef *ref) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)ref->data; + const enum AVPixelFormat *pix_fmt; + int ret; + + if (ctx->internal->source_frames) { + /* A derived frame context is already initialised. */ + return 0; + } + + /* validate the pixel format */ + for (pix_fmt = ctx->internal->hw_type->pix_fmts; *pix_fmt != AV_PIX_FMT_NONE; pix_fmt++) { + if (*pix_fmt == ctx->format) + break; + } + if (*pix_fmt == AV_PIX_FMT_NONE) { + av_log(ctx, AV_LOG_ERROR, + "The hardware pixel format '%s' is not supported by the device type '%s'\n", + av_get_pix_fmt_name(ctx->format), ctx->internal->hw_type->name); + return AVERROR(ENOSYS); + } + + /* validate the dimensions */ + ret = av_image_check_size(ctx->width, ctx->height, 0, ctx); + if (ret < 0) + return ret; + + /* format-specific init */ + if (ctx->internal->hw_type->frames_init) { + ret = ctx->internal->hw_type->frames_init(ctx); + if (ret < 0) + goto fail; + } + + if (ctx->internal->pool_internal && !ctx->pool) + ctx->pool = ctx->internal->pool_internal; + + /* preallocate the frames in the pool, if requested */ + if (ctx->initial_pool_size > 0) { + ret = hwframe_pool_prealloc(ref); + if (ret < 0) + goto fail; + } + + return 0; +fail: + if (ctx->internal->hw_type->frames_uninit) + ctx->internal->hw_type->frames_uninit(ctx); + return ret; +} + +int av_hwframe_transfer_get_formats(AVBufferRef *hwframe_ref, + enum AVHWFrameTransferDirection dir, + enum AVPixelFormat **formats, int flags) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)hwframe_ref->data; + + if (!ctx->internal->hw_type->transfer_get_formats) + return AVERROR(ENOSYS); + + return ctx->internal->hw_type->transfer_get_formats(ctx, dir, formats); +} + +static int transfer_data_alloc(AVFrame *dst, const AVFrame *src, int flags) +{ + AVHWFramesContext *ctx; + AVFrame *frame_tmp; + int ret = 0; + + if (!src->hw_frames_ctx) + return AVERROR(EINVAL); + ctx = (AVHWFramesContext*)src->hw_frames_ctx->data; + + frame_tmp = av_frame_alloc(); + if (!frame_tmp) + return AVERROR(ENOMEM); + + /* if the format is set, use that + * otherwise pick the first supported one */ + if (dst->format >= 0) { + frame_tmp->format = dst->format; + } else { + enum AVPixelFormat *formats; + + ret = av_hwframe_transfer_get_formats(src->hw_frames_ctx, + AV_HWFRAME_TRANSFER_DIRECTION_FROM, + &formats, 0); + if (ret < 0) + goto fail; + frame_tmp->format = formats[0]; + av_freep(&formats); + } + frame_tmp->width = ctx->width; + frame_tmp->height = ctx->height; + + ret = av_frame_get_buffer(frame_tmp, 0); + if (ret < 0) + goto fail; + + ret = av_hwframe_transfer_data(frame_tmp, src, flags); + if (ret < 0) + goto fail; + + frame_tmp->width = src->width; + frame_tmp->height = src->height; + + av_frame_move_ref(dst, frame_tmp); + +fail: + av_frame_free(&frame_tmp); + return ret; +} + +int av_hwframe_transfer_data(AVFrame *dst, const AVFrame *src, int flags) +{ + AVHWFramesContext *ctx; + int ret; + + if (!dst->buf[0]) + return transfer_data_alloc(dst, src, flags); + + /* + * Hardware -> Hardware Transfer. + * Unlike Software -> Hardware or Hardware -> Software, the transfer + * function could be provided by either the src or dst, depending on + * the specific combination of hardware. + */ + if (src->hw_frames_ctx && dst->hw_frames_ctx) { + AVHWFramesContext *src_ctx = + (AVHWFramesContext*)src->hw_frames_ctx->data; + AVHWFramesContext *dst_ctx = + (AVHWFramesContext*)dst->hw_frames_ctx->data; + + if (src_ctx->internal->source_frames) { + av_log(src_ctx, AV_LOG_ERROR, + "A device with a derived frame context cannot be used as " + "the source of a HW -> HW transfer."); + return AVERROR(ENOSYS); + } + + if (dst_ctx->internal->source_frames) { + av_log(src_ctx, AV_LOG_ERROR, + "A device with a derived frame context cannot be used as " + "the destination of a HW -> HW transfer."); + return AVERROR(ENOSYS); + } + + ret = src_ctx->internal->hw_type->transfer_data_from(src_ctx, dst, src); + if (ret == AVERROR(ENOSYS)) + ret = dst_ctx->internal->hw_type->transfer_data_to(dst_ctx, dst, src); + if (ret < 0) + return ret; + } else { + if (src->hw_frames_ctx) { + ctx = (AVHWFramesContext*)src->hw_frames_ctx->data; + + ret = ctx->internal->hw_type->transfer_data_from(ctx, dst, src); + if (ret < 0) + return ret; + } else if (dst->hw_frames_ctx) { + ctx = (AVHWFramesContext*)dst->hw_frames_ctx->data; + + ret = ctx->internal->hw_type->transfer_data_to(ctx, dst, src); + if (ret < 0) + return ret; + } else { + return AVERROR(ENOSYS); + } + } + return 0; +} + +int av_hwframe_get_buffer(AVBufferRef *hwframe_ref, AVFrame *frame, int flags) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)hwframe_ref->data; + int ret; + + if (ctx->internal->source_frames) { + // This is a derived frame context, so we allocate in the source + // and map the frame immediately. + AVFrame *src_frame; + + frame->format = ctx->format; + frame->hw_frames_ctx = av_buffer_ref(hwframe_ref); + if (!frame->hw_frames_ctx) + return AVERROR(ENOMEM); + + src_frame = av_frame_alloc(); + if (!src_frame) + return AVERROR(ENOMEM); + + ret = av_hwframe_get_buffer(ctx->internal->source_frames, + src_frame, 0); + if (ret < 0) { + av_frame_free(&src_frame); + return ret; + } + + ret = av_hwframe_map(frame, src_frame, + ctx->internal->source_allocation_map_flags); + if (ret) { + av_log(ctx, AV_LOG_ERROR, "Failed to map frame into derived " + "frame context: %d.\n", ret); + av_frame_free(&src_frame); + return ret; + } + + // Free the source frame immediately - the mapped frame still + // contains a reference to it. + av_frame_free(&src_frame); + + return 0; + } + + if (!ctx->internal->hw_type->frames_get_buffer) + return AVERROR(ENOSYS); + + if (!ctx->pool) + return AVERROR(EINVAL); + + frame->hw_frames_ctx = av_buffer_ref(hwframe_ref); + if (!frame->hw_frames_ctx) + return AVERROR(ENOMEM); + + ret = ctx->internal->hw_type->frames_get_buffer(ctx, frame); + if (ret < 0) { + av_buffer_unref(&frame->hw_frames_ctx); + return ret; + } + + frame->extended_data = frame->data; + + return 0; +} + +void *av_hwdevice_hwconfig_alloc(AVBufferRef *ref) +{ + AVHWDeviceContext *ctx = (AVHWDeviceContext*)ref->data; + const HWContextType *hw_type = ctx->internal->hw_type; + + if (hw_type->device_hwconfig_size == 0) + return NULL; + + return av_mallocz(hw_type->device_hwconfig_size); +} + +AVHWFramesConstraints *av_hwdevice_get_hwframe_constraints(AVBufferRef *ref, + const void *hwconfig) +{ + AVHWDeviceContext *ctx = (AVHWDeviceContext*)ref->data; + const HWContextType *hw_type = ctx->internal->hw_type; + AVHWFramesConstraints *constraints; + + if (!hw_type->frames_get_constraints) + return NULL; + + constraints = av_mallocz(sizeof(*constraints)); + if (!constraints) + return NULL; + + constraints->min_width = constraints->min_height = 0; + constraints->max_width = constraints->max_height = INT_MAX; + + if (hw_type->frames_get_constraints(ctx, hwconfig, constraints) >= 0) { + return constraints; + } else { + av_hwframe_constraints_free(&constraints); + return NULL; + } +} + +void av_hwframe_constraints_free(AVHWFramesConstraints **constraints) +{ + if (*constraints) { + av_freep(&(*constraints)->valid_hw_formats); + av_freep(&(*constraints)->valid_sw_formats); + } + av_freep(constraints); +} + +int av_hwdevice_ctx_create(AVBufferRef **pdevice_ref, enum AVHWDeviceType type, + const char *device, AVDictionary *opts, int flags) +{ + AVBufferRef *device_ref = NULL; + AVHWDeviceContext *device_ctx; + int ret = 0; + + device_ref = av_hwdevice_ctx_alloc(type); + if (!device_ref) { + ret = AVERROR(ENOMEM); + goto fail; + } + device_ctx = (AVHWDeviceContext*)device_ref->data; + + if (!device_ctx->internal->hw_type->device_create) { + ret = AVERROR(ENOSYS); + goto fail; + } + + ret = device_ctx->internal->hw_type->device_create(device_ctx, device, + opts, flags); + if (ret < 0) + goto fail; + + ret = av_hwdevice_ctx_init(device_ref); + if (ret < 0) + goto fail; + + *pdevice_ref = device_ref; + return 0; +fail: + av_buffer_unref(&device_ref); + *pdevice_ref = NULL; + return ret; +} + +int av_hwdevice_ctx_create_derived_opts(AVBufferRef **dst_ref_ptr, + enum AVHWDeviceType type, + AVBufferRef *src_ref, + AVDictionary *options, int flags) +{ + AVBufferRef *dst_ref = NULL, *tmp_ref; + AVHWDeviceContext *dst_ctx, *tmp_ctx; + int ret = 0; + + tmp_ref = src_ref; + while (tmp_ref) { + tmp_ctx = (AVHWDeviceContext*)tmp_ref->data; + if (tmp_ctx->type == type) { + dst_ref = av_buffer_ref(tmp_ref); + if (!dst_ref) { + ret = AVERROR(ENOMEM); + goto fail; + } + goto done; + } + tmp_ref = tmp_ctx->internal->source_device; + } + + dst_ref = av_hwdevice_ctx_alloc(type); + if (!dst_ref) { + ret = AVERROR(ENOMEM); + goto fail; + } + dst_ctx = (AVHWDeviceContext*)dst_ref->data; + + tmp_ref = src_ref; + while (tmp_ref) { + tmp_ctx = (AVHWDeviceContext*)tmp_ref->data; + if (dst_ctx->internal->hw_type->device_derive) { + ret = dst_ctx->internal->hw_type->device_derive(dst_ctx, + tmp_ctx, + options, + flags); + if (ret == 0) { + dst_ctx->internal->source_device = av_buffer_ref(src_ref); + if (!dst_ctx->internal->source_device) { + ret = AVERROR(ENOMEM); + goto fail; + } + ret = av_hwdevice_ctx_init(dst_ref); + if (ret < 0) + goto fail; + goto done; + } + if (ret != AVERROR(ENOSYS)) + goto fail; + } + tmp_ref = tmp_ctx->internal->source_device; + } + + ret = AVERROR(ENOSYS); + goto fail; + +done: + *dst_ref_ptr = dst_ref; + return 0; + +fail: + av_buffer_unref(&dst_ref); + *dst_ref_ptr = NULL; + return ret; +} + +int av_hwdevice_ctx_create_derived(AVBufferRef **dst_ref_ptr, + enum AVHWDeviceType type, + AVBufferRef *src_ref, int flags) +{ + return av_hwdevice_ctx_create_derived_opts(dst_ref_ptr, type, src_ref, + NULL, flags); +} + +static void ff_hwframe_unmap(void *opaque, uint8_t *data) +{ + HWMapDescriptor *hwmap = (HWMapDescriptor*)data; + AVHWFramesContext *ctx = opaque; + + if (hwmap->unmap) + hwmap->unmap(ctx, hwmap); + + av_frame_free(&hwmap->source); + + av_buffer_unref(&hwmap->hw_frames_ctx); + + av_free(hwmap); +} + +int ff_hwframe_map_create(AVBufferRef *hwframe_ref, + AVFrame *dst, const AVFrame *src, + void (*unmap)(AVHWFramesContext *ctx, + HWMapDescriptor *hwmap), + void *priv) +{ + AVHWFramesContext *ctx = (AVHWFramesContext*)hwframe_ref->data; + HWMapDescriptor *hwmap; + int ret; + + hwmap = av_mallocz(sizeof(*hwmap)); + if (!hwmap) { + ret = AVERROR(ENOMEM); + goto fail; + } + + hwmap->source = av_frame_alloc(); + if (!hwmap->source) { + ret = AVERROR(ENOMEM); + goto fail; + } + ret = av_frame_ref(hwmap->source, src); + if (ret < 0) + goto fail; + + hwmap->hw_frames_ctx = av_buffer_ref(hwframe_ref); + if (!hwmap->hw_frames_ctx) { + ret = AVERROR(ENOMEM); + goto fail; + } + + hwmap->unmap = unmap; + hwmap->priv = priv; + + dst->buf[0] = av_buffer_create((uint8_t*)hwmap, sizeof(*hwmap), + &ff_hwframe_unmap, ctx, 0); + if (!dst->buf[0]) { + ret = AVERROR(ENOMEM); + goto fail; + } + + return 0; + +fail: + if (hwmap) { + av_buffer_unref(&hwmap->hw_frames_ctx); + av_frame_free(&hwmap->source); + } + av_free(hwmap); + return ret; +} + +int av_hwframe_map(AVFrame *dst, const AVFrame *src, int flags) +{ + AVBufferRef *orig_dst_frames = dst->hw_frames_ctx; + enum AVPixelFormat orig_dst_fmt = dst->format; + AVHWFramesContext *src_frames, *dst_frames; + HWMapDescriptor *hwmap; + int ret; + + if (src->hw_frames_ctx && dst->hw_frames_ctx) { + src_frames = (AVHWFramesContext*)src->hw_frames_ctx->data; + dst_frames = (AVHWFramesContext*)dst->hw_frames_ctx->data; + + if ((src_frames == dst_frames && + src->format == dst_frames->sw_format && + dst->format == dst_frames->format) || + (src_frames->internal->source_frames && + src_frames->internal->source_frames->data == + (uint8_t*)dst_frames)) { + // This is an unmap operation. We don't need to directly + // do anything here other than fill in the original frame, + // because the real unmap will be invoked when the last + // reference to the mapped frame disappears. + if (!src->buf[0]) { + av_log(src_frames, AV_LOG_ERROR, "Invalid mapping " + "found when attempting unmap.\n"); + return AVERROR(EINVAL); + } + hwmap = (HWMapDescriptor*)src->buf[0]->data; + return av_frame_replace(dst, hwmap->source); + } + } + + if (src->hw_frames_ctx) { + src_frames = (AVHWFramesContext*)src->hw_frames_ctx->data; + + if (src_frames->format == src->format && + src_frames->internal->hw_type->map_from) { + ret = src_frames->internal->hw_type->map_from(src_frames, + dst, src, flags); + if (ret >= 0) + return ret; + else if (ret != AVERROR(ENOSYS)) + goto fail; + } + } + + if (dst->hw_frames_ctx) { + dst_frames = (AVHWFramesContext*)dst->hw_frames_ctx->data; + + if (dst_frames->format == dst->format && + dst_frames->internal->hw_type->map_to) { + ret = dst_frames->internal->hw_type->map_to(dst_frames, + dst, src, flags); + if (ret >= 0) + return ret; + else if (ret != AVERROR(ENOSYS)) + goto fail; + } + } + + return AVERROR(ENOSYS); + +fail: + // if the caller provided dst frames context, it should be preserved + // by this function + av_assert0(orig_dst_frames == NULL || + orig_dst_frames == dst->hw_frames_ctx); + + // preserve user-provided dst frame fields, but clean + // anything we might have set + dst->hw_frames_ctx = NULL; + av_frame_unref(dst); + + dst->hw_frames_ctx = orig_dst_frames; + dst->format = orig_dst_fmt; + + return ret; +} + +int av_hwframe_ctx_create_derived(AVBufferRef **derived_frame_ctx, + enum AVPixelFormat format, + AVBufferRef *derived_device_ctx, + AVBufferRef *source_frame_ctx, + int flags) +{ + AVBufferRef *dst_ref = NULL; + AVHWFramesContext *dst = NULL; + AVHWFramesContext *src = (AVHWFramesContext*)source_frame_ctx->data; + int ret; + + if (src->internal->source_frames) { + AVHWFramesContext *src_src = + (AVHWFramesContext*)src->internal->source_frames->data; + AVHWDeviceContext *dst_dev = + (AVHWDeviceContext*)derived_device_ctx->data; + + if (src_src->device_ctx == dst_dev) { + // This is actually an unmapping, so we just return a + // reference to the source frame context. + *derived_frame_ctx = + av_buffer_ref(src->internal->source_frames); + if (!*derived_frame_ctx) { + ret = AVERROR(ENOMEM); + goto fail; + } + return 0; + } + } + + dst_ref = av_hwframe_ctx_alloc(derived_device_ctx); + if (!dst_ref) { + ret = AVERROR(ENOMEM); + goto fail; + } + + dst = (AVHWFramesContext*)dst_ref->data; + + dst->format = format; + dst->sw_format = src->sw_format; + dst->width = src->width; + dst->height = src->height; + + dst->internal->source_frames = av_buffer_ref(source_frame_ctx); + if (!dst->internal->source_frames) { + ret = AVERROR(ENOMEM); + goto fail; + } + + dst->internal->source_allocation_map_flags = + flags & (AV_HWFRAME_MAP_READ | + AV_HWFRAME_MAP_WRITE | + AV_HWFRAME_MAP_OVERWRITE | + AV_HWFRAME_MAP_DIRECT); + + ret = AVERROR(ENOSYS); + if (src->internal->hw_type->frames_derive_from) + ret = src->internal->hw_type->frames_derive_from(dst, src, flags); + if (ret == AVERROR(ENOSYS) && + dst->internal->hw_type->frames_derive_to) + ret = dst->internal->hw_type->frames_derive_to(dst, src, flags); + if (ret == AVERROR(ENOSYS)) + ret = 0; + if (ret) + goto fail; + + *derived_frame_ctx = dst_ref; + return 0; + +fail: + if (dst) + av_buffer_unref(&dst->internal->source_frames); + av_buffer_unref(&dst_ref); + return ret; +} + +int ff_hwframe_map_replace(AVFrame *dst, const AVFrame *src) +{ + HWMapDescriptor *hwmap = (HWMapDescriptor*)dst->buf[0]->data; + return av_frame_replace(hwmap->source, src); +} diff --git a/media/ffvpx/libavutil/hwcontext.h b/media/ffvpx/libavutil/hwcontext.h new file mode 100644 index 0000000000..2b33721a97 --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext.h @@ -0,0 +1,611 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_HWCONTEXT_H +#define AVUTIL_HWCONTEXT_H + +#include "buffer.h" +#include "frame.h" +#include "log.h" +#include "pixfmt.h" + +enum AVHWDeviceType { + AV_HWDEVICE_TYPE_NONE, + AV_HWDEVICE_TYPE_VDPAU, + AV_HWDEVICE_TYPE_CUDA, + AV_HWDEVICE_TYPE_VAAPI, + AV_HWDEVICE_TYPE_DXVA2, + AV_HWDEVICE_TYPE_QSV, + AV_HWDEVICE_TYPE_VIDEOTOOLBOX, + AV_HWDEVICE_TYPE_D3D11VA, + AV_HWDEVICE_TYPE_DRM, + AV_HWDEVICE_TYPE_OPENCL, + AV_HWDEVICE_TYPE_MEDIACODEC, + AV_HWDEVICE_TYPE_VULKAN, + AV_HWDEVICE_TYPE_D3D12VA, +}; + +typedef struct AVHWDeviceInternal AVHWDeviceInternal; + +/** + * This struct aggregates all the (hardware/vendor-specific) "high-level" state, + * i.e. state that is not tied to a concrete processing configuration. + * E.g., in an API that supports hardware-accelerated encoding and decoding, + * this struct will (if possible) wrap the state that is common to both encoding + * and decoding and from which specific instances of encoders or decoders can be + * derived. + * + * This struct is reference-counted with the AVBuffer mechanism. The + * av_hwdevice_ctx_alloc() constructor yields a reference, whose data field + * points to the actual AVHWDeviceContext. Further objects derived from + * AVHWDeviceContext (such as AVHWFramesContext, describing a frame pool with + * specific properties) will hold an internal reference to it. After all the + * references are released, the AVHWDeviceContext itself will be freed, + * optionally invoking a user-specified callback for uninitializing the hardware + * state. + */ +typedef struct AVHWDeviceContext { + /** + * A class for logging. Set by av_hwdevice_ctx_alloc(). + */ + const AVClass *av_class; + + /** + * Private data used internally by libavutil. Must not be accessed in any + * way by the caller. + */ + AVHWDeviceInternal *internal; + + /** + * This field identifies the underlying API used for hardware access. + * + * This field is set when this struct is allocated and never changed + * afterwards. + */ + enum AVHWDeviceType type; + + /** + * The format-specific data, allocated and freed by libavutil along with + * this context. + * + * Should be cast by the user to the format-specific context defined in the + * corresponding header (hwcontext_*.h) and filled as described in the + * documentation before calling av_hwdevice_ctx_init(). + * + * After calling av_hwdevice_ctx_init() this struct should not be modified + * by the caller. + */ + void *hwctx; + + /** + * This field may be set by the caller before calling av_hwdevice_ctx_init(). + * + * If non-NULL, this callback will be called when the last reference to + * this context is unreferenced, immediately before it is freed. + * + * @note when other objects (e.g an AVHWFramesContext) are derived from this + * struct, this callback will be invoked after all such child objects + * are fully uninitialized and their respective destructors invoked. + */ + void (*free)(struct AVHWDeviceContext *ctx); + + /** + * Arbitrary user data, to be used e.g. by the free() callback. + */ + void *user_opaque; +} AVHWDeviceContext; + +typedef struct AVHWFramesInternal AVHWFramesInternal; + +/** + * This struct describes a set or pool of "hardware" frames (i.e. those with + * data not located in normal system memory). All the frames in the pool are + * assumed to be allocated in the same way and interchangeable. + * + * This struct is reference-counted with the AVBuffer mechanism and tied to a + * given AVHWDeviceContext instance. The av_hwframe_ctx_alloc() constructor + * yields a reference, whose data field points to the actual AVHWFramesContext + * struct. + */ +typedef struct AVHWFramesContext { + /** + * A class for logging. + */ + const AVClass *av_class; + + /** + * Private data used internally by libavutil. Must not be accessed in any + * way by the caller. + */ + AVHWFramesInternal *internal; + + /** + * A reference to the parent AVHWDeviceContext. This reference is owned and + * managed by the enclosing AVHWFramesContext, but the caller may derive + * additional references from it. + */ + AVBufferRef *device_ref; + + /** + * The parent AVHWDeviceContext. This is simply a pointer to + * device_ref->data provided for convenience. + * + * Set by libavutil in av_hwframe_ctx_init(). + */ + AVHWDeviceContext *device_ctx; + + /** + * The format-specific data, allocated and freed automatically along with + * this context. + * + * Should be cast by the user to the format-specific context defined in the + * corresponding header (hwframe_*.h) and filled as described in the + * documentation before calling av_hwframe_ctx_init(). + * + * After any frames using this context are created, the contents of this + * struct should not be modified by the caller. + */ + void *hwctx; + + /** + * This field may be set by the caller before calling av_hwframe_ctx_init(). + * + * If non-NULL, this callback will be called when the last reference to + * this context is unreferenced, immediately before it is freed. + */ + void (*free)(struct AVHWFramesContext *ctx); + + /** + * Arbitrary user data, to be used e.g. by the free() callback. + */ + void *user_opaque; + + /** + * A pool from which the frames are allocated by av_hwframe_get_buffer(). + * This field may be set by the caller before calling av_hwframe_ctx_init(). + * The buffers returned by calling av_buffer_pool_get() on this pool must + * have the properties described in the documentation in the corresponding hw + * type's header (hwcontext_*.h). The pool will be freed strictly before + * this struct's free() callback is invoked. + * + * This field may be NULL, then libavutil will attempt to allocate a pool + * internally. Note that certain device types enforce pools allocated at + * fixed size (frame count), which cannot be extended dynamically. In such a + * case, initial_pool_size must be set appropriately. + */ + AVBufferPool *pool; + + /** + * Initial size of the frame pool. If a device type does not support + * dynamically resizing the pool, then this is also the maximum pool size. + * + * May be set by the caller before calling av_hwframe_ctx_init(). Must be + * set if pool is NULL and the device type does not support dynamic pools. + */ + int initial_pool_size; + + /** + * The pixel format identifying the underlying HW surface type. + * + * Must be a hwaccel format, i.e. the corresponding descriptor must have the + * AV_PIX_FMT_FLAG_HWACCEL flag set. + * + * Must be set by the user before calling av_hwframe_ctx_init(). + */ + enum AVPixelFormat format; + + /** + * The pixel format identifying the actual data layout of the hardware + * frames. + * + * Must be set by the caller before calling av_hwframe_ctx_init(). + * + * @note when the underlying API does not provide the exact data layout, but + * only the colorspace/bit depth, this field should be set to the fully + * planar version of that format (e.g. for 8-bit 420 YUV it should be + * AV_PIX_FMT_YUV420P, not AV_PIX_FMT_NV12 or anything else). + */ + enum AVPixelFormat sw_format; + + /** + * The allocated dimensions of the frames in this pool. + * + * Must be set by the user before calling av_hwframe_ctx_init(). + */ + int width, height; +} AVHWFramesContext; + +/** + * Look up an AVHWDeviceType by name. + * + * @param name String name of the device type (case-insensitive). + * @return The type from enum AVHWDeviceType, or AV_HWDEVICE_TYPE_NONE if + * not found. + */ +enum AVHWDeviceType av_hwdevice_find_type_by_name(const char *name); + +/** Get the string name of an AVHWDeviceType. + * + * @param type Type from enum AVHWDeviceType. + * @return Pointer to a static string containing the name, or NULL if the type + * is not valid. + */ +const char *av_hwdevice_get_type_name(enum AVHWDeviceType type); + +/** + * Iterate over supported device types. + * + * @param prev AV_HWDEVICE_TYPE_NONE initially, then the previous type + * returned by this function in subsequent iterations. + * @return The next usable device type from enum AVHWDeviceType, or + * AV_HWDEVICE_TYPE_NONE if there are no more. + */ +enum AVHWDeviceType av_hwdevice_iterate_types(enum AVHWDeviceType prev); + +/** + * Allocate an AVHWDeviceContext for a given hardware type. + * + * @param type the type of the hardware device to allocate. + * @return a reference to the newly created AVHWDeviceContext on success or NULL + * on failure. + */ +AVBufferRef *av_hwdevice_ctx_alloc(enum AVHWDeviceType type); + +/** + * Finalize the device context before use. This function must be called after + * the context is filled with all the required information and before it is + * used in any way. + * + * @param ref a reference to the AVHWDeviceContext + * @return 0 on success, a negative AVERROR code on failure + */ +int av_hwdevice_ctx_init(AVBufferRef *ref); + +/** + * Open a device of the specified type and create an AVHWDeviceContext for it. + * + * This is a convenience function intended to cover the simple cases. Callers + * who need to fine-tune device creation/management should open the device + * manually and then wrap it in an AVHWDeviceContext using + * av_hwdevice_ctx_alloc()/av_hwdevice_ctx_init(). + * + * The returned context is already initialized and ready for use, the caller + * should not call av_hwdevice_ctx_init() on it. The user_opaque/free fields of + * the created AVHWDeviceContext are set by this function and should not be + * touched by the caller. + * + * @param device_ctx On success, a reference to the newly-created device context + * will be written here. The reference is owned by the caller + * and must be released with av_buffer_unref() when no longer + * needed. On failure, NULL will be written to this pointer. + * @param type The type of the device to create. + * @param device A type-specific string identifying the device to open. + * @param opts A dictionary of additional (type-specific) options to use in + * opening the device. The dictionary remains owned by the caller. + * @param flags currently unused + * + * @return 0 on success, a negative AVERROR code on failure. + */ +int av_hwdevice_ctx_create(AVBufferRef **device_ctx, enum AVHWDeviceType type, + const char *device, AVDictionary *opts, int flags); + +/** + * Create a new device of the specified type from an existing device. + * + * If the source device is a device of the target type or was originally + * derived from such a device (possibly through one or more intermediate + * devices of other types), then this will return a reference to the + * existing device of the same type as is requested. + * + * Otherwise, it will attempt to derive a new device from the given source + * device. If direct derivation to the new type is not implemented, it will + * attempt the same derivation from each ancestor of the source device in + * turn looking for an implemented derivation method. + * + * @param dst_ctx On success, a reference to the newly-created + * AVHWDeviceContext. + * @param type The type of the new device to create. + * @param src_ctx A reference to an existing AVHWDeviceContext which will be + * used to create the new device. + * @param flags Currently unused; should be set to zero. + * @return Zero on success, a negative AVERROR code on failure. + */ +int av_hwdevice_ctx_create_derived(AVBufferRef **dst_ctx, + enum AVHWDeviceType type, + AVBufferRef *src_ctx, int flags); + +/** + * Create a new device of the specified type from an existing device. + * + * This function performs the same action as av_hwdevice_ctx_create_derived, + * however, it is able to set options for the new device to be derived. + * + * @param dst_ctx On success, a reference to the newly-created + * AVHWDeviceContext. + * @param type The type of the new device to create. + * @param src_ctx A reference to an existing AVHWDeviceContext which will be + * used to create the new device. + * @param options Options for the new device to create, same format as in + * av_hwdevice_ctx_create. + * @param flags Currently unused; should be set to zero. + * @return Zero on success, a negative AVERROR code on failure. + */ +int av_hwdevice_ctx_create_derived_opts(AVBufferRef **dst_ctx, + enum AVHWDeviceType type, + AVBufferRef *src_ctx, + AVDictionary *options, int flags); + +/** + * Allocate an AVHWFramesContext tied to a given device context. + * + * @param device_ctx a reference to a AVHWDeviceContext. This function will make + * a new reference for internal use, the one passed to the + * function remains owned by the caller. + * @return a reference to the newly created AVHWFramesContext on success or NULL + * on failure. + */ +AVBufferRef *av_hwframe_ctx_alloc(AVBufferRef *device_ctx); + +/** + * Finalize the context before use. This function must be called after the + * context is filled with all the required information and before it is attached + * to any frames. + * + * @param ref a reference to the AVHWFramesContext + * @return 0 on success, a negative AVERROR code on failure + */ +int av_hwframe_ctx_init(AVBufferRef *ref); + +/** + * Allocate a new frame attached to the given AVHWFramesContext. + * + * @param hwframe_ctx a reference to an AVHWFramesContext + * @param frame an empty (freshly allocated or unreffed) frame to be filled with + * newly allocated buffers. + * @param flags currently unused, should be set to zero + * @return 0 on success, a negative AVERROR code on failure + */ +int av_hwframe_get_buffer(AVBufferRef *hwframe_ctx, AVFrame *frame, int flags); + +/** + * Copy data to or from a hw surface. At least one of dst/src must have an + * AVHWFramesContext attached. + * + * If src has an AVHWFramesContext attached, then the format of dst (if set) + * must use one of the formats returned by av_hwframe_transfer_get_formats(src, + * AV_HWFRAME_TRANSFER_DIRECTION_FROM). + * If dst has an AVHWFramesContext attached, then the format of src must use one + * of the formats returned by av_hwframe_transfer_get_formats(dst, + * AV_HWFRAME_TRANSFER_DIRECTION_TO) + * + * dst may be "clean" (i.e. with data/buf pointers unset), in which case the + * data buffers will be allocated by this function using av_frame_get_buffer(). + * If dst->format is set, then this format will be used, otherwise (when + * dst->format is AV_PIX_FMT_NONE) the first acceptable format will be chosen. + * + * The two frames must have matching allocated dimensions (i.e. equal to + * AVHWFramesContext.width/height), since not all device types support + * transferring a sub-rectangle of the whole surface. The display dimensions + * (i.e. AVFrame.width/height) may be smaller than the allocated dimensions, but + * also have to be equal for both frames. When the display dimensions are + * smaller than the allocated dimensions, the content of the padding in the + * destination frame is unspecified. + * + * @param dst the destination frame. dst is not touched on failure. + * @param src the source frame. + * @param flags currently unused, should be set to zero + * @return 0 on success, a negative AVERROR error code on failure. + */ +int av_hwframe_transfer_data(AVFrame *dst, const AVFrame *src, int flags); + +enum AVHWFrameTransferDirection { + /** + * Transfer the data from the queried hw frame. + */ + AV_HWFRAME_TRANSFER_DIRECTION_FROM, + + /** + * Transfer the data to the queried hw frame. + */ + AV_HWFRAME_TRANSFER_DIRECTION_TO, +}; + +/** + * Get a list of possible source or target formats usable in + * av_hwframe_transfer_data(). + * + * @param hwframe_ctx the frame context to obtain the information for + * @param dir the direction of the transfer + * @param formats the pointer to the output format list will be written here. + * The list is terminated with AV_PIX_FMT_NONE and must be freed + * by the caller when no longer needed using av_free(). + * If this function returns successfully, the format list will + * have at least one item (not counting the terminator). + * On failure, the contents of this pointer are unspecified. + * @param flags currently unused, should be set to zero + * @return 0 on success, a negative AVERROR code on failure. + */ +int av_hwframe_transfer_get_formats(AVBufferRef *hwframe_ctx, + enum AVHWFrameTransferDirection dir, + enum AVPixelFormat **formats, int flags); + + +/** + * This struct describes the constraints on hardware frames attached to + * a given device with a hardware-specific configuration. This is returned + * by av_hwdevice_get_hwframe_constraints() and must be freed by + * av_hwframe_constraints_free() after use. + */ +typedef struct AVHWFramesConstraints { + /** + * A list of possible values for format in the hw_frames_ctx, + * terminated by AV_PIX_FMT_NONE. This member will always be filled. + */ + enum AVPixelFormat *valid_hw_formats; + + /** + * A list of possible values for sw_format in the hw_frames_ctx, + * terminated by AV_PIX_FMT_NONE. Can be NULL if this information is + * not known. + */ + enum AVPixelFormat *valid_sw_formats; + + /** + * The minimum size of frames in this hw_frames_ctx. + * (Zero if not known.) + */ + int min_width; + int min_height; + + /** + * The maximum size of frames in this hw_frames_ctx. + * (INT_MAX if not known / no limit.) + */ + int max_width; + int max_height; +} AVHWFramesConstraints; + +/** + * Allocate a HW-specific configuration structure for a given HW device. + * After use, the user must free all members as required by the specific + * hardware structure being used, then free the structure itself with + * av_free(). + * + * @param device_ctx a reference to the associated AVHWDeviceContext. + * @return The newly created HW-specific configuration structure on + * success or NULL on failure. + */ +void *av_hwdevice_hwconfig_alloc(AVBufferRef *device_ctx); + +/** + * Get the constraints on HW frames given a device and the HW-specific + * configuration to be used with that device. If no HW-specific + * configuration is provided, returns the maximum possible capabilities + * of the device. + * + * @param ref a reference to the associated AVHWDeviceContext. + * @param hwconfig a filled HW-specific configuration structure, or NULL + * to return the maximum possible capabilities of the device. + * @return AVHWFramesConstraints structure describing the constraints + * on the device, or NULL if not available. + */ +AVHWFramesConstraints *av_hwdevice_get_hwframe_constraints(AVBufferRef *ref, + const void *hwconfig); + +/** + * Free an AVHWFrameConstraints structure. + * + * @param constraints The (filled or unfilled) AVHWFrameConstraints structure. + */ +void av_hwframe_constraints_free(AVHWFramesConstraints **constraints); + + +/** + * Flags to apply to frame mappings. + */ +enum { + /** + * The mapping must be readable. + */ + AV_HWFRAME_MAP_READ = 1 << 0, + /** + * The mapping must be writeable. + */ + AV_HWFRAME_MAP_WRITE = 1 << 1, + /** + * The mapped frame will be overwritten completely in subsequent + * operations, so the current frame data need not be loaded. Any values + * which are not overwritten are unspecified. + */ + AV_HWFRAME_MAP_OVERWRITE = 1 << 2, + /** + * The mapping must be direct. That is, there must not be any copying in + * the map or unmap steps. Note that performance of direct mappings may + * be much lower than normal memory. + */ + AV_HWFRAME_MAP_DIRECT = 1 << 3, +}; + +/** + * Map a hardware frame. + * + * This has a number of different possible effects, depending on the format + * and origin of the src and dst frames. On input, src should be a usable + * frame with valid buffers and dst should be blank (typically as just created + * by av_frame_alloc()). src should have an associated hwframe context, and + * dst may optionally have a format and associated hwframe context. + * + * If src was created by mapping a frame from the hwframe context of dst, + * then this function undoes the mapping - dst is replaced by a reference to + * the frame that src was originally mapped from. + * + * If both src and dst have an associated hwframe context, then this function + * attempts to map the src frame from its hardware context to that of dst and + * then fill dst with appropriate data to be usable there. This will only be + * possible if the hwframe contexts and associated devices are compatible - + * given compatible devices, av_hwframe_ctx_create_derived() can be used to + * create a hwframe context for dst in which mapping should be possible. + * + * If src has a hwframe context but dst does not, then the src frame is + * mapped to normal memory and should thereafter be usable as a normal frame. + * If the format is set on dst, then the mapping will attempt to create dst + * with that format and fail if it is not possible. If format is unset (is + * AV_PIX_FMT_NONE) then dst will be mapped with whatever the most appropriate + * format to use is (probably the sw_format of the src hwframe context). + * + * A return value of AVERROR(ENOSYS) indicates that the mapping is not + * possible with the given arguments and hwframe setup, while other return + * values indicate that it failed somehow. + * + * On failure, the destination frame will be left blank, except for the + * hw_frames_ctx/format fields thay may have been set by the caller - those will + * be preserved as they were. + * + * @param dst Destination frame, to contain the mapping. + * @param src Source frame, to be mapped. + * @param flags Some combination of AV_HWFRAME_MAP_* flags. + * @return Zero on success, negative AVERROR code on failure. + */ +int av_hwframe_map(AVFrame *dst, const AVFrame *src, int flags); + + +/** + * Create and initialise an AVHWFramesContext as a mapping of another existing + * AVHWFramesContext on a different device. + * + * av_hwframe_ctx_init() should not be called after this. + * + * @param derived_frame_ctx On success, a reference to the newly created + * AVHWFramesContext. + * @param format The AVPixelFormat for the derived context. + * @param derived_device_ctx A reference to the device to create the new + * AVHWFramesContext on. + * @param source_frame_ctx A reference to an existing AVHWFramesContext + * which will be mapped to the derived context. + * @param flags Some combination of AV_HWFRAME_MAP_* flags, defining the + * mapping parameters to apply to frames which are allocated + * in the derived device. + * @return Zero on success, negative AVERROR code on failure. + */ +int av_hwframe_ctx_create_derived(AVBufferRef **derived_frame_ctx, + enum AVPixelFormat format, + AVBufferRef *derived_device_ctx, + AVBufferRef *source_frame_ctx, + int flags); + +#endif /* AVUTIL_HWCONTEXT_H */ diff --git a/media/ffvpx/libavutil/hwcontext_drm.h b/media/ffvpx/libavutil/hwcontext_drm.h new file mode 100644 index 0000000000..42709f215e --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext_drm.h @@ -0,0 +1,169 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_HWCONTEXT_DRM_H +#define AVUTIL_HWCONTEXT_DRM_H + +#include <stddef.h> +#include <stdint.h> + +/** + * @file + * API-specific header for AV_HWDEVICE_TYPE_DRM. + * + * Internal frame allocation is not currently supported - all frames + * must be allocated by the user. Thus AVHWFramesContext is always + * NULL, though this may change if support for frame allocation is + * added in future. + */ + +enum { + /** + * The maximum number of layers/planes in a DRM frame. + */ + AV_DRM_MAX_PLANES = 4 +}; + +/** + * DRM object descriptor. + * + * Describes a single DRM object, addressing it as a PRIME file + * descriptor. + */ +typedef struct AVDRMObjectDescriptor { + /** + * DRM PRIME fd for the object. + */ + int fd; + /** + * Total size of the object. + * + * (This includes any parts not which do not contain image data.) + */ + size_t size; + /** + * Format modifier applied to the object (DRM_FORMAT_MOD_*). + * + * If the format modifier is unknown then this should be set to + * DRM_FORMAT_MOD_INVALID. + */ + uint64_t format_modifier; +} AVDRMObjectDescriptor; + +/** + * DRM plane descriptor. + * + * Describes a single plane of a layer, which is contained within + * a single object. + */ +typedef struct AVDRMPlaneDescriptor { + /** + * Index of the object containing this plane in the objects + * array of the enclosing frame descriptor. + */ + int object_index; + /** + * Offset within that object of this plane. + */ + ptrdiff_t offset; + /** + * Pitch (linesize) of this plane. + */ + ptrdiff_t pitch; +} AVDRMPlaneDescriptor; + +/** + * DRM layer descriptor. + * + * Describes a single layer within a frame. This has the structure + * defined by its format, and will contain one or more planes. + */ +typedef struct AVDRMLayerDescriptor { + /** + * Format of the layer (DRM_FORMAT_*). + */ + uint32_t format; + /** + * Number of planes in the layer. + * + * This must match the number of planes required by format. + */ + int nb_planes; + /** + * Array of planes in this layer. + */ + AVDRMPlaneDescriptor planes[AV_DRM_MAX_PLANES]; +} AVDRMLayerDescriptor; + +/** + * DRM frame descriptor. + * + * This is used as the data pointer for AV_PIX_FMT_DRM_PRIME frames. + * It is also used by user-allocated frame pools - allocating in + * AVHWFramesContext.pool must return AVBufferRefs which contain + * an object of this type. + * + * The fields of this structure should be set such it can be + * imported directly by EGL using the EGL_EXT_image_dma_buf_import + * and EGL_EXT_image_dma_buf_import_modifiers extensions. + * (Note that the exact layout of a particular format may vary between + * platforms - we only specify that the same platform should be able + * to import it.) + * + * The total number of planes must not exceed AV_DRM_MAX_PLANES, and + * the order of the planes by increasing layer index followed by + * increasing plane index must be the same as the order which would + * be used for the data pointers in the equivalent software format. + */ +typedef struct AVDRMFrameDescriptor { + /** + * Number of DRM objects making up this frame. + */ + int nb_objects; + /** + * Array of objects making up the frame. + */ + AVDRMObjectDescriptor objects[AV_DRM_MAX_PLANES]; + /** + * Number of layers in the frame. + */ + int nb_layers; + /** + * Array of layers in the frame. + */ + AVDRMLayerDescriptor layers[AV_DRM_MAX_PLANES]; +} AVDRMFrameDescriptor; + +/** + * DRM device. + * + * Allocated as AVHWDeviceContext.hwctx. + */ +typedef struct AVDRMDeviceContext { + /** + * File descriptor of DRM device. + * + * This is used as the device to create frames on, and may also be + * used in some derivation and mapping operations. + * + * If no device is required, set to -1. + */ + int fd; +} AVDRMDeviceContext; + +#endif /* AVUTIL_HWCONTEXT_DRM_H */ diff --git a/media/ffvpx/libavutil/hwcontext_internal.h b/media/ffvpx/libavutil/hwcontext_internal.h new file mode 100644 index 0000000000..4df516ee6a --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext_internal.h @@ -0,0 +1,179 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_HWCONTEXT_INTERNAL_H +#define AVUTIL_HWCONTEXT_INTERNAL_H + +#include <stddef.h> + +#include "buffer.h" +#include "hwcontext.h" +#include "frame.h" +#include "pixfmt.h" + +typedef struct HWContextType { + enum AVHWDeviceType type; + const char *name; + + /** + * An array of pixel formats supported by the AVHWFramesContext instances + * Terminated by AV_PIX_FMT_NONE. + */ + const enum AVPixelFormat *pix_fmts; + + /** + * size of the public hardware-specific context, + * i.e. AVHWDeviceContext.hwctx + */ + size_t device_hwctx_size; + /** + * size of the private data, i.e. + * AVHWDeviceInternal.priv + */ + size_t device_priv_size; + + /** + * Size of the hardware-specific device configuration. + * (Used to query hwframe constraints.) + */ + size_t device_hwconfig_size; + + /** + * size of the public frame pool hardware-specific context, + * i.e. AVHWFramesContext.hwctx + */ + size_t frames_hwctx_size; + /** + * size of the private data, i.e. + * AVHWFramesInternal.priv + */ + size_t frames_priv_size; + + int (*device_create)(AVHWDeviceContext *ctx, const char *device, + AVDictionary *opts, int flags); + int (*device_derive)(AVHWDeviceContext *dst_ctx, + AVHWDeviceContext *src_ctx, + AVDictionary *opts, int flags); + + int (*device_init)(AVHWDeviceContext *ctx); + void (*device_uninit)(AVHWDeviceContext *ctx); + + int (*frames_get_constraints)(AVHWDeviceContext *ctx, + const void *hwconfig, + AVHWFramesConstraints *constraints); + + int (*frames_init)(AVHWFramesContext *ctx); + void (*frames_uninit)(AVHWFramesContext *ctx); + + int (*frames_get_buffer)(AVHWFramesContext *ctx, AVFrame *frame); + int (*transfer_get_formats)(AVHWFramesContext *ctx, + enum AVHWFrameTransferDirection dir, + enum AVPixelFormat **formats); + int (*transfer_data_to)(AVHWFramesContext *ctx, AVFrame *dst, + const AVFrame *src); + int (*transfer_data_from)(AVHWFramesContext *ctx, AVFrame *dst, + const AVFrame *src); + + int (*map_to)(AVHWFramesContext *ctx, AVFrame *dst, + const AVFrame *src, int flags); + int (*map_from)(AVHWFramesContext *ctx, AVFrame *dst, + const AVFrame *src, int flags); + + int (*frames_derive_to)(AVHWFramesContext *dst_ctx, + AVHWFramesContext *src_ctx, int flags); + int (*frames_derive_from)(AVHWFramesContext *dst_ctx, + AVHWFramesContext *src_ctx, int flags); +} HWContextType; + +struct AVHWDeviceInternal { + const HWContextType *hw_type; + void *priv; + + /** + * For a derived device, a reference to the original device + * context it was derived from. + */ + AVBufferRef *source_device; +}; + +struct AVHWFramesInternal { + const HWContextType *hw_type; + void *priv; + + AVBufferPool *pool_internal; + + /** + * For a derived context, a reference to the original frames + * context it was derived from. + */ + AVBufferRef *source_frames; + /** + * Flags to apply to the mapping from the source to the derived + * frame context when trying to allocate in the derived context. + */ + int source_allocation_map_flags; +}; + +typedef struct HWMapDescriptor { + /** + * A reference to the original source of the mapping. + */ + AVFrame *source; + /** + * A reference to the hardware frames context in which this + * mapping was made. May be the same as source->hw_frames_ctx, + * but need not be. + */ + AVBufferRef *hw_frames_ctx; + /** + * Unmap function. + */ + void (*unmap)(AVHWFramesContext *ctx, + struct HWMapDescriptor *hwmap); + /** + * Hardware-specific private data associated with the mapping. + */ + void *priv; +} HWMapDescriptor; + +int ff_hwframe_map_create(AVBufferRef *hwframe_ref, + AVFrame *dst, const AVFrame *src, + void (*unmap)(AVHWFramesContext *ctx, + HWMapDescriptor *hwmap), + void *priv); + +/** + * Replace the current hwmap of dst with the one from src, used for indirect + * mappings like VAAPI->(DRM)->OpenCL/Vulkan where a direct interop is missing + */ +int ff_hwframe_map_replace(AVFrame *dst, const AVFrame *src); + +extern const HWContextType ff_hwcontext_type_cuda; +extern const HWContextType ff_hwcontext_type_d3d11va; +extern const HWContextType ff_hwcontext_type_d3d12va; +extern const HWContextType ff_hwcontext_type_drm; +extern const HWContextType ff_hwcontext_type_dxva2; +extern const HWContextType ff_hwcontext_type_opencl; +extern const HWContextType ff_hwcontext_type_qsv; +extern const HWContextType ff_hwcontext_type_vaapi; +extern const HWContextType ff_hwcontext_type_vdpau; +extern const HWContextType ff_hwcontext_type_videotoolbox; +extern const HWContextType ff_hwcontext_type_mediacodec; +extern const HWContextType ff_hwcontext_type_vulkan; + +#endif /* AVUTIL_HWCONTEXT_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/hwcontext_vaapi.c b/media/ffvpx/libavutil/hwcontext_vaapi.c new file mode 100644 index 0000000000..29fc8bd648 --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext_vaapi.c @@ -0,0 +1,2034 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#if HAVE_VAAPI_WIN32 +# include <windows.h> +#define COBJMACROS +# include <initguid.h> +# include <dxgi1_2.h> +# include "compat/w32dlfcn.h" +# include <va/va_win32.h> +typedef HRESULT (WINAPI *PFN_CREATE_DXGI_FACTORY)(REFIID riid, void **ppFactory); +#endif +#if HAVE_VAAPI_X11 +# include <va/va_x11.h> +#endif +#if HAVE_VAAPI_DRM +# include <va/va_drm.h> +#endif + +#if CONFIG_LIBDRM +# include <va/va_drmcommon.h> +# include <xf86drm.h> +# include <drm_fourcc.h> +# ifndef DRM_FORMAT_MOD_INVALID +# define DRM_FORMAT_MOD_INVALID ((1ULL << 56) - 1) +# endif +#endif + +#include <fcntl.h> +#if HAVE_UNISTD_H +# include <unistd.h> +#endif + + +#include "avassert.h" +#include "buffer.h" +#include "common.h" +#include "hwcontext.h" +#include "hwcontext_drm.h" +#include "hwcontext_internal.h" +#include "hwcontext_vaapi.h" +#include "mem.h" +#include "pixdesc.h" +#include "pixfmt.h" + + +typedef struct VAAPIDevicePriv { +#if HAVE_VAAPI_X11 + Display *x11_display; +#endif + + int drm_fd; +} VAAPIDevicePriv; + +typedef struct VAAPISurfaceFormat { + enum AVPixelFormat pix_fmt; + VAImageFormat image_format; +} VAAPISurfaceFormat; + +typedef struct VAAPIDeviceContext { + // Surface formats which can be used with this device. + VAAPISurfaceFormat *formats; + int nb_formats; +} VAAPIDeviceContext; + +typedef struct VAAPIFramesContext { + // Surface attributes set at create time. + VASurfaceAttrib *attributes; + int nb_attributes; + // RT format of the underlying surface (Intel driver ignores this anyway). + unsigned int rt_format; + // Whether vaDeriveImage works. + int derive_works; + // Caches whether VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2 is unsupported for + // surface imports. + int prime_2_import_unsupported; +} VAAPIFramesContext; + +typedef struct VAAPIMapping { + // Handle to the derived or copied image which is mapped. + VAImage image; + // The mapping flags actually used. + int flags; +} VAAPIMapping; + +typedef struct VAAPIFormat { + unsigned int fourcc; + unsigned int rt_format; + enum AVPixelFormat pix_fmt; + int chroma_planes_swapped; +} VAAPIFormatDescriptor; + +#define MAP(va, rt, av, swap_uv) { \ + VA_FOURCC_ ## va, \ + VA_RT_FORMAT_ ## rt, \ + AV_PIX_FMT_ ## av, \ + swap_uv, \ + } +// The map fourcc <-> pix_fmt isn't bijective because of the annoying U/V +// plane swap cases. The frame handling below tries to hide these. +static const VAAPIFormatDescriptor vaapi_format_map[] = { + MAP(NV12, YUV420, NV12, 0), +#ifdef VA_FOURCC_I420 + MAP(I420, YUV420, YUV420P, 0), +#endif + MAP(YV12, YUV420, YUV420P, 1), + MAP(IYUV, YUV420, YUV420P, 0), + MAP(422H, YUV422, YUV422P, 0), +#ifdef VA_FOURCC_YV16 + MAP(YV16, YUV422, YUV422P, 1), +#endif + MAP(UYVY, YUV422, UYVY422, 0), + MAP(YUY2, YUV422, YUYV422, 0), +#ifdef VA_FOURCC_Y210 + MAP(Y210, YUV422_10, Y210, 0), +#endif +#ifdef VA_FOURCC_Y212 + MAP(Y212, YUV422_12, Y212, 0), +#endif + MAP(411P, YUV411, YUV411P, 0), + MAP(422V, YUV422, YUV440P, 0), + MAP(444P, YUV444, YUV444P, 0), +#ifdef VA_FOURCC_XYUV + MAP(XYUV, YUV444, VUYX, 0), +#endif + MAP(Y800, YUV400, GRAY8, 0), +#ifdef VA_FOURCC_P010 + MAP(P010, YUV420_10BPP, P010, 0), +#endif +#ifdef VA_FOURCC_P012 + MAP(P012, YUV420_12, P012, 0), +#endif + MAP(BGRA, RGB32, BGRA, 0), + MAP(BGRX, RGB32, BGR0, 0), + MAP(RGBA, RGB32, RGBA, 0), + MAP(RGBX, RGB32, RGB0, 0), +#ifdef VA_FOURCC_ABGR + MAP(ABGR, RGB32, ABGR, 0), + MAP(XBGR, RGB32, 0BGR, 0), +#endif + MAP(ARGB, RGB32, ARGB, 0), + MAP(XRGB, RGB32, 0RGB, 0), +#ifdef VA_FOURCC_X2R10G10B10 + MAP(X2R10G10B10, RGB32_10, X2RGB10, 0), +#endif +#ifdef VA_FOURCC_Y410 + // libva doesn't include a fourcc for XV30 and the driver only declares + // support for Y410, so we must fudge the mapping here. + MAP(Y410, YUV444_10, XV30, 0), +#endif +#ifdef VA_FOURCC_Y412 + // libva doesn't include a fourcc for XV36 and the driver only declares + // support for Y412, so we must fudge the mapping here. + MAP(Y412, YUV444_12, XV36, 0), +#endif +}; +#undef MAP + +static const VAAPIFormatDescriptor * + vaapi_format_from_fourcc(unsigned int fourcc) +{ + int i; + for (i = 0; i < FF_ARRAY_ELEMS(vaapi_format_map); i++) + if (vaapi_format_map[i].fourcc == fourcc) + return &vaapi_format_map[i]; + return NULL; +} + +static const VAAPIFormatDescriptor * + vaapi_format_from_pix_fmt(enum AVPixelFormat pix_fmt) +{ + int i; + for (i = 0; i < FF_ARRAY_ELEMS(vaapi_format_map); i++) + if (vaapi_format_map[i].pix_fmt == pix_fmt) + return &vaapi_format_map[i]; + return NULL; +} + +static enum AVPixelFormat vaapi_pix_fmt_from_fourcc(unsigned int fourcc) +{ + const VAAPIFormatDescriptor *desc; + desc = vaapi_format_from_fourcc(fourcc); + if (desc) + return desc->pix_fmt; + else + return AV_PIX_FMT_NONE; +} + +static int vaapi_get_image_format(AVHWDeviceContext *hwdev, + enum AVPixelFormat pix_fmt, + VAImageFormat **image_format) +{ + VAAPIDeviceContext *ctx = hwdev->internal->priv; + int i; + + for (i = 0; i < ctx->nb_formats; i++) { + if (ctx->formats[i].pix_fmt == pix_fmt) { + if (image_format) + *image_format = &ctx->formats[i].image_format; + return 0; + } + } + return AVERROR(ENOSYS); +} + +static int vaapi_frames_get_constraints(AVHWDeviceContext *hwdev, + const void *hwconfig, + AVHWFramesConstraints *constraints) +{ + AVVAAPIDeviceContext *hwctx = hwdev->hwctx; + const AVVAAPIHWConfig *config = hwconfig; + VAAPIDeviceContext *ctx = hwdev->internal->priv; + VASurfaceAttrib *attr_list = NULL; + VAStatus vas; + enum AVPixelFormat pix_fmt; + unsigned int fourcc; + int err, i, j, attr_count, pix_fmt_count; + + if (config && + !(hwctx->driver_quirks & AV_VAAPI_DRIVER_QUIRK_SURFACE_ATTRIBUTES)) { + attr_count = 0; + vas = vaQuerySurfaceAttributes(hwctx->display, config->config_id, + 0, &attr_count); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwdev, AV_LOG_ERROR, "Failed to query surface attributes: " + "%d (%s).\n", vas, vaErrorStr(vas)); + err = AVERROR(ENOSYS); + goto fail; + } + + attr_list = av_malloc(attr_count * sizeof(*attr_list)); + if (!attr_list) { + err = AVERROR(ENOMEM); + goto fail; + } + + vas = vaQuerySurfaceAttributes(hwctx->display, config->config_id, + attr_list, &attr_count); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwdev, AV_LOG_ERROR, "Failed to query surface attributes: " + "%d (%s).\n", vas, vaErrorStr(vas)); + err = AVERROR(ENOSYS); + goto fail; + } + + pix_fmt_count = 0; + for (i = 0; i < attr_count; i++) { + switch (attr_list[i].type) { + case VASurfaceAttribPixelFormat: + fourcc = attr_list[i].value.value.i; + pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc); + if (pix_fmt != AV_PIX_FMT_NONE) { + ++pix_fmt_count; + } else { + // Something unsupported - ignore. + } + break; + case VASurfaceAttribMinWidth: + constraints->min_width = attr_list[i].value.value.i; + break; + case VASurfaceAttribMinHeight: + constraints->min_height = attr_list[i].value.value.i; + break; + case VASurfaceAttribMaxWidth: + constraints->max_width = attr_list[i].value.value.i; + break; + case VASurfaceAttribMaxHeight: + constraints->max_height = attr_list[i].value.value.i; + break; + } + } + if (pix_fmt_count == 0) { + // Nothing usable found. Presumably there exists something which + // works, so leave the set null to indicate unknown. + constraints->valid_sw_formats = NULL; + } else { + constraints->valid_sw_formats = av_malloc_array(pix_fmt_count + 1, + sizeof(pix_fmt)); + if (!constraints->valid_sw_formats) { + err = AVERROR(ENOMEM); + goto fail; + } + + for (i = j = 0; i < attr_count; i++) { + int k; + + if (attr_list[i].type != VASurfaceAttribPixelFormat) + continue; + fourcc = attr_list[i].value.value.i; + pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc); + + if (pix_fmt == AV_PIX_FMT_NONE) + continue; + + for (k = 0; k < j; k++) { + if (constraints->valid_sw_formats[k] == pix_fmt) + break; + } + + if (k == j) + constraints->valid_sw_formats[j++] = pix_fmt; + } + constraints->valid_sw_formats[j] = AV_PIX_FMT_NONE; + } + } else { + // No configuration supplied. + // Return the full set of image formats known by the implementation. + constraints->valid_sw_formats = av_malloc_array(ctx->nb_formats + 1, + sizeof(pix_fmt)); + if (!constraints->valid_sw_formats) { + err = AVERROR(ENOMEM); + goto fail; + } + for (i = j = 0; i < ctx->nb_formats; i++) { + int k; + + for (k = 0; k < j; k++) { + if (constraints->valid_sw_formats[k] == ctx->formats[i].pix_fmt) + break; + } + + if (k == j) + constraints->valid_sw_formats[j++] = ctx->formats[i].pix_fmt; + } + + constraints->valid_sw_formats[j] = AV_PIX_FMT_NONE; + } + + constraints->valid_hw_formats = av_malloc_array(2, sizeof(pix_fmt)); + if (!constraints->valid_hw_formats) { + err = AVERROR(ENOMEM); + goto fail; + } + constraints->valid_hw_formats[0] = AV_PIX_FMT_VAAPI; + constraints->valid_hw_formats[1] = AV_PIX_FMT_NONE; + + err = 0; +fail: + av_freep(&attr_list); + return err; +} + +static const struct { + const char *friendly_name; + const char *match_string; + unsigned int quirks; +} vaapi_driver_quirks_table[] = { +#if !VA_CHECK_VERSION(1, 0, 0) + // The i965 driver did not conform before version 2.0. + { + "Intel i965 (Quick Sync)", + "i965", + AV_VAAPI_DRIVER_QUIRK_RENDER_PARAM_BUFFERS, + }, +#endif + { + "Intel iHD", + "ubit", + AV_VAAPI_DRIVER_QUIRK_ATTRIB_MEMTYPE, + }, + { + "VDPAU wrapper", + "Splitted-Desktop Systems VDPAU backend for VA-API", + AV_VAAPI_DRIVER_QUIRK_SURFACE_ATTRIBUTES, + }, +}; + +static int vaapi_device_init(AVHWDeviceContext *hwdev) +{ + VAAPIDeviceContext *ctx = hwdev->internal->priv; + AVVAAPIDeviceContext *hwctx = hwdev->hwctx; + VAImageFormat *image_list = NULL; + VAStatus vas; + const char *vendor_string; + int err, i, image_count; + enum AVPixelFormat pix_fmt; + unsigned int fourcc; + + image_count = vaMaxNumImageFormats(hwctx->display); + if (image_count <= 0) { + err = AVERROR(EIO); + goto fail; + } + image_list = av_malloc(image_count * sizeof(*image_list)); + if (!image_list) { + err = AVERROR(ENOMEM); + goto fail; + } + vas = vaQueryImageFormats(hwctx->display, image_list, &image_count); + if (vas != VA_STATUS_SUCCESS) { + err = AVERROR(EIO); + goto fail; + } + + ctx->formats = av_malloc(image_count * sizeof(*ctx->formats)); + if (!ctx->formats) { + err = AVERROR(ENOMEM); + goto fail; + } + ctx->nb_formats = 0; + for (i = 0; i < image_count; i++) { + fourcc = image_list[i].fourcc; + pix_fmt = vaapi_pix_fmt_from_fourcc(fourcc); + if (pix_fmt == AV_PIX_FMT_NONE) { + av_log(hwdev, AV_LOG_DEBUG, "Format %#x -> unknown.\n", + fourcc); + } else { + av_log(hwdev, AV_LOG_DEBUG, "Format %#x -> %s.\n", + fourcc, av_get_pix_fmt_name(pix_fmt)); + ctx->formats[ctx->nb_formats].pix_fmt = pix_fmt; + ctx->formats[ctx->nb_formats].image_format = image_list[i]; + ++ctx->nb_formats; + } + } + + vendor_string = vaQueryVendorString(hwctx->display); + if (vendor_string) + av_log(hwdev, AV_LOG_VERBOSE, "VAAPI driver: %s.\n", vendor_string); + + if (hwctx->driver_quirks & AV_VAAPI_DRIVER_QUIRK_USER_SET) { + av_log(hwdev, AV_LOG_VERBOSE, "Using quirks set by user (%#x).\n", + hwctx->driver_quirks); + } else { + // Detect the driver in use and set quirk flags if necessary. + hwctx->driver_quirks = 0; + if (vendor_string) { + for (i = 0; i < FF_ARRAY_ELEMS(vaapi_driver_quirks_table); i++) { + if (strstr(vendor_string, + vaapi_driver_quirks_table[i].match_string)) { + av_log(hwdev, AV_LOG_VERBOSE, "Matched driver string " + "as known nonstandard driver \"%s\", setting " + "quirks (%#x).\n", + vaapi_driver_quirks_table[i].friendly_name, + vaapi_driver_quirks_table[i].quirks); + hwctx->driver_quirks |= + vaapi_driver_quirks_table[i].quirks; + break; + } + } + if (!(i < FF_ARRAY_ELEMS(vaapi_driver_quirks_table))) { + av_log(hwdev, AV_LOG_VERBOSE, "Driver not found in known " + "nonstandard list, using standard behaviour.\n"); + } + } else { + av_log(hwdev, AV_LOG_VERBOSE, "Driver has no vendor string, " + "assuming standard behaviour.\n"); + } + } + + av_free(image_list); + return 0; +fail: + av_freep(&ctx->formats); + av_free(image_list); + return err; +} + +static void vaapi_device_uninit(AVHWDeviceContext *hwdev) +{ + VAAPIDeviceContext *ctx = hwdev->internal->priv; + + av_freep(&ctx->formats); +} + +static void vaapi_buffer_free(void *opaque, uint8_t *data) +{ + AVHWFramesContext *hwfc = opaque; + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VASurfaceID surface_id; + VAStatus vas; + + surface_id = (VASurfaceID)(uintptr_t)data; + + vas = vaDestroySurfaces(hwctx->display, &surface_id, 1); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to destroy surface %#x: " + "%d (%s).\n", surface_id, vas, vaErrorStr(vas)); + } +} + +static AVBufferRef *vaapi_pool_alloc(void *opaque, size_t size) +{ + AVHWFramesContext *hwfc = opaque; + VAAPIFramesContext *ctx = hwfc->internal->priv; + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + AVVAAPIFramesContext *avfc = hwfc->hwctx; + VASurfaceID surface_id; + VAStatus vas; + AVBufferRef *ref; + + if (hwfc->initial_pool_size > 0 && + avfc->nb_surfaces >= hwfc->initial_pool_size) + return NULL; + + vas = vaCreateSurfaces(hwctx->display, ctx->rt_format, + hwfc->width, hwfc->height, + &surface_id, 1, + ctx->attributes, ctx->nb_attributes); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to create surface: " + "%d (%s).\n", vas, vaErrorStr(vas)); + return NULL; + } + av_log(hwfc, AV_LOG_DEBUG, "Created surface %#x.\n", surface_id); + + ref = av_buffer_create((uint8_t*)(uintptr_t)surface_id, + sizeof(surface_id), &vaapi_buffer_free, + hwfc, AV_BUFFER_FLAG_READONLY); + if (!ref) { + vaDestroySurfaces(hwctx->display, &surface_id, 1); + return NULL; + } + + if (hwfc->initial_pool_size > 0) { + // This is a fixed-size pool, so we must still be in the initial + // allocation sequence. + av_assert0(avfc->nb_surfaces < hwfc->initial_pool_size); + avfc->surface_ids[avfc->nb_surfaces] = surface_id; + ++avfc->nb_surfaces; + } + + return ref; +} + +static int vaapi_frames_init(AVHWFramesContext *hwfc) +{ + AVVAAPIFramesContext *avfc = hwfc->hwctx; + VAAPIFramesContext *ctx = hwfc->internal->priv; + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + const VAAPIFormatDescriptor *desc; + VAImageFormat *expected_format; + AVBufferRef *test_surface = NULL; + VASurfaceID test_surface_id; + VAImage test_image; + VAStatus vas; + int err, i; + + desc = vaapi_format_from_pix_fmt(hwfc->sw_format); + if (!desc) { + av_log(hwfc, AV_LOG_ERROR, "Unsupported format: %s.\n", + av_get_pix_fmt_name(hwfc->sw_format)); + return AVERROR(EINVAL); + } + + if (!hwfc->pool) { + if (!(hwctx->driver_quirks & AV_VAAPI_DRIVER_QUIRK_SURFACE_ATTRIBUTES)) { + int need_memory_type = !(hwctx->driver_quirks & AV_VAAPI_DRIVER_QUIRK_ATTRIB_MEMTYPE); + int need_pixel_format = 1; + for (i = 0; i < avfc->nb_attributes; i++) { + if (avfc->attributes[i].type == VASurfaceAttribMemoryType) + need_memory_type = 0; + if (avfc->attributes[i].type == VASurfaceAttribPixelFormat) + need_pixel_format = 0; + } + ctx->nb_attributes = + avfc->nb_attributes + need_memory_type + need_pixel_format; + + ctx->attributes = av_malloc(ctx->nb_attributes * + sizeof(*ctx->attributes)); + if (!ctx->attributes) { + err = AVERROR(ENOMEM); + goto fail; + } + + for (i = 0; i < avfc->nb_attributes; i++) + ctx->attributes[i] = avfc->attributes[i]; + if (need_memory_type) { + ctx->attributes[i++] = (VASurfaceAttrib) { + .type = VASurfaceAttribMemoryType, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypeInteger, + .value.value.i = VA_SURFACE_ATTRIB_MEM_TYPE_VA, + }; + } + if (need_pixel_format) { + ctx->attributes[i++] = (VASurfaceAttrib) { + .type = VASurfaceAttribPixelFormat, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypeInteger, + .value.value.i = desc->fourcc, + }; + } + av_assert0(i == ctx->nb_attributes); + } else { + ctx->attributes = NULL; + ctx->nb_attributes = 0; + } + + ctx->rt_format = desc->rt_format; + + if (hwfc->initial_pool_size > 0) { + // This pool will be usable as a render target, so we need to store + // all of the surface IDs somewhere that vaCreateContext() calls + // will be able to access them. + avfc->nb_surfaces = 0; + avfc->surface_ids = av_malloc(hwfc->initial_pool_size * + sizeof(*avfc->surface_ids)); + if (!avfc->surface_ids) { + err = AVERROR(ENOMEM); + goto fail; + } + } else { + // This pool allows dynamic sizing, and will not be usable as a + // render target. + avfc->nb_surfaces = 0; + avfc->surface_ids = NULL; + } + + hwfc->internal->pool_internal = + av_buffer_pool_init2(sizeof(VASurfaceID), hwfc, + &vaapi_pool_alloc, NULL); + if (!hwfc->internal->pool_internal) { + av_log(hwfc, AV_LOG_ERROR, "Failed to create VAAPI surface pool.\n"); + err = AVERROR(ENOMEM); + goto fail; + } + } + + // Allocate a single surface to test whether vaDeriveImage() is going + // to work for the specific configuration. + if (hwfc->pool) { + test_surface = av_buffer_pool_get(hwfc->pool); + if (!test_surface) { + av_log(hwfc, AV_LOG_ERROR, "Unable to allocate a surface from " + "user-configured buffer pool.\n"); + err = AVERROR(ENOMEM); + goto fail; + } + } else { + test_surface = av_buffer_pool_get(hwfc->internal->pool_internal); + if (!test_surface) { + av_log(hwfc, AV_LOG_ERROR, "Unable to allocate a surface from " + "internal buffer pool.\n"); + err = AVERROR(ENOMEM); + goto fail; + } + } + test_surface_id = (VASurfaceID)(uintptr_t)test_surface->data; + + ctx->derive_works = 0; + + err = vaapi_get_image_format(hwfc->device_ctx, + hwfc->sw_format, &expected_format); + if (err == 0) { + vas = vaDeriveImage(hwctx->display, test_surface_id, &test_image); + if (vas == VA_STATUS_SUCCESS) { + if (expected_format->fourcc == test_image.format.fourcc) { + av_log(hwfc, AV_LOG_DEBUG, "Direct mapping possible.\n"); + ctx->derive_works = 1; + } else { + av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: " + "derived image format %08x does not match " + "expected format %08x.\n", + expected_format->fourcc, test_image.format.fourcc); + } + vaDestroyImage(hwctx->display, test_image.image_id); + } else { + av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: " + "deriving image does not work: " + "%d (%s).\n", vas, vaErrorStr(vas)); + } + } else { + av_log(hwfc, AV_LOG_DEBUG, "Direct mapping disabled: " + "image format is not supported.\n"); + } + + av_buffer_unref(&test_surface); + return 0; + +fail: + av_buffer_unref(&test_surface); + av_freep(&avfc->surface_ids); + av_freep(&ctx->attributes); + return err; +} + +static void vaapi_frames_uninit(AVHWFramesContext *hwfc) +{ + AVVAAPIFramesContext *avfc = hwfc->hwctx; + VAAPIFramesContext *ctx = hwfc->internal->priv; + + av_freep(&avfc->surface_ids); + av_freep(&ctx->attributes); +} + +static int vaapi_get_buffer(AVHWFramesContext *hwfc, AVFrame *frame) +{ + frame->buf[0] = av_buffer_pool_get(hwfc->pool); + if (!frame->buf[0]) + return AVERROR(ENOMEM); + + frame->data[3] = frame->buf[0]->data; + frame->format = AV_PIX_FMT_VAAPI; + frame->width = hwfc->width; + frame->height = hwfc->height; + + return 0; +} + +static int vaapi_transfer_get_formats(AVHWFramesContext *hwfc, + enum AVHWFrameTransferDirection dir, + enum AVPixelFormat **formats) +{ + VAAPIDeviceContext *ctx = hwfc->device_ctx->internal->priv; + enum AVPixelFormat *pix_fmts; + int i, k, sw_format_available; + + sw_format_available = 0; + for (i = 0; i < ctx->nb_formats; i++) { + if (ctx->formats[i].pix_fmt == hwfc->sw_format) + sw_format_available = 1; + } + + pix_fmts = av_malloc((ctx->nb_formats + 1) * sizeof(*pix_fmts)); + if (!pix_fmts) + return AVERROR(ENOMEM); + + if (sw_format_available) { + pix_fmts[0] = hwfc->sw_format; + k = 1; + } else { + k = 0; + } + for (i = 0; i < ctx->nb_formats; i++) { + if (ctx->formats[i].pix_fmt == hwfc->sw_format) + continue; + av_assert0(k < ctx->nb_formats); + pix_fmts[k++] = ctx->formats[i].pix_fmt; + } + pix_fmts[k] = AV_PIX_FMT_NONE; + + *formats = pix_fmts; + return 0; +} + +static void vaapi_unmap_frame(AVHWFramesContext *hwfc, + HWMapDescriptor *hwmap) +{ + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VAAPIMapping *map = hwmap->priv; + VASurfaceID surface_id; + VAStatus vas; + + surface_id = (VASurfaceID)(uintptr_t)hwmap->source->data[3]; + av_log(hwfc, AV_LOG_DEBUG, "Unmap surface %#x.\n", surface_id); + + vas = vaUnmapBuffer(hwctx->display, map->image.buf); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to unmap image from surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + } + + if ((map->flags & AV_HWFRAME_MAP_WRITE) && + !(map->flags & AV_HWFRAME_MAP_DIRECT)) { + vas = vaPutImage(hwctx->display, surface_id, map->image.image_id, + 0, 0, hwfc->width, hwfc->height, + 0, 0, hwfc->width, hwfc->height); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to write image to surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + } + } + + vas = vaDestroyImage(hwctx->display, map->image.image_id); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to destroy image from surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + } + + av_free(map); +} + +static int vaapi_map_frame(AVHWFramesContext *hwfc, + AVFrame *dst, const AVFrame *src, int flags) +{ + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VAAPIFramesContext *ctx = hwfc->internal->priv; + VASurfaceID surface_id; + const VAAPIFormatDescriptor *desc; + VAImageFormat *image_format; + VAAPIMapping *map; + VAStatus vas; + void *address = NULL; + int err, i; + + surface_id = (VASurfaceID)(uintptr_t)src->data[3]; + av_log(hwfc, AV_LOG_DEBUG, "Map surface %#x.\n", surface_id); + + if (!ctx->derive_works && (flags & AV_HWFRAME_MAP_DIRECT)) { + // Requested direct mapping but it is not possible. + return AVERROR(EINVAL); + } + if (dst->format == AV_PIX_FMT_NONE) + dst->format = hwfc->sw_format; + if (dst->format != hwfc->sw_format && (flags & AV_HWFRAME_MAP_DIRECT)) { + // Requested direct mapping but the formats do not match. + return AVERROR(EINVAL); + } + + err = vaapi_get_image_format(hwfc->device_ctx, dst->format, &image_format); + if (err < 0) { + // Requested format is not a valid output format. + return err; + } + + map = av_malloc(sizeof(*map)); + if (!map) + return AVERROR(ENOMEM); + map->flags = flags; + map->image.image_id = VA_INVALID_ID; + + vas = vaSyncSurface(hwctx->display, surface_id); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to sync surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + + // The memory which we map using derive need not be connected to the CPU + // in a way conducive to fast access. On Gen7-Gen9 Intel graphics, the + // memory is mappable but not cached, so normal memcpy()-like access is + // very slow to read it (but writing is ok). It is possible to read much + // faster with a copy routine which is aware of the limitation, but we + // assume for now that the user is not aware of that and would therefore + // prefer not to be given direct-mapped memory if they request read access. + if (ctx->derive_works && dst->format == hwfc->sw_format && + ((flags & AV_HWFRAME_MAP_DIRECT) || !(flags & AV_HWFRAME_MAP_READ))) { + vas = vaDeriveImage(hwctx->display, surface_id, &map->image); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to derive image from " + "surface %#x: %d (%s).\n", + surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + if (map->image.format.fourcc != image_format->fourcc) { + av_log(hwfc, AV_LOG_ERROR, "Derive image of surface %#x " + "is in wrong format: expected %#08x, got %#08x.\n", + surface_id, image_format->fourcc, map->image.format.fourcc); + err = AVERROR(EIO); + goto fail; + } + map->flags |= AV_HWFRAME_MAP_DIRECT; + } else { + vas = vaCreateImage(hwctx->display, image_format, + hwfc->width, hwfc->height, &map->image); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to create image for " + "surface %#x: %d (%s).\n", + surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + if (!(flags & AV_HWFRAME_MAP_OVERWRITE)) { + vas = vaGetImage(hwctx->display, surface_id, 0, 0, + hwfc->width, hwfc->height, map->image.image_id); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to read image from " + "surface %#x: %d (%s).\n", + surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + } + } + + vas = vaMapBuffer(hwctx->display, map->image.buf, &address); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to map image from surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + + err = ff_hwframe_map_create(src->hw_frames_ctx, + dst, src, &vaapi_unmap_frame, map); + if (err < 0) + goto fail; + + dst->width = src->width; + dst->height = src->height; + + for (i = 0; i < map->image.num_planes; i++) { + dst->data[i] = (uint8_t*)address + map->image.offsets[i]; + dst->linesize[i] = map->image.pitches[i]; + } + + desc = vaapi_format_from_fourcc(map->image.format.fourcc); + if (desc && desc->chroma_planes_swapped) { + // Chroma planes are YVU rather than YUV, so swap them. + FFSWAP(uint8_t*, dst->data[1], dst->data[2]); + } + + return 0; + +fail: + if (map) { + if (address) + vaUnmapBuffer(hwctx->display, map->image.buf); + if (map->image.image_id != VA_INVALID_ID) + vaDestroyImage(hwctx->display, map->image.image_id); + av_free(map); + } + return err; +} + +static int vaapi_transfer_data_from(AVHWFramesContext *hwfc, + AVFrame *dst, const AVFrame *src) +{ + AVFrame *map; + int err; + + if (dst->width > hwfc->width || dst->height > hwfc->height) + return AVERROR(EINVAL); + + map = av_frame_alloc(); + if (!map) + return AVERROR(ENOMEM); + map->format = dst->format; + + err = vaapi_map_frame(hwfc, map, src, AV_HWFRAME_MAP_READ); + if (err) + goto fail; + + map->width = dst->width; + map->height = dst->height; + + err = av_frame_copy(dst, map); + if (err) + goto fail; + + err = 0; +fail: + av_frame_free(&map); + return err; +} + +static int vaapi_transfer_data_to(AVHWFramesContext *hwfc, + AVFrame *dst, const AVFrame *src) +{ + AVFrame *map; + int err; + + if (src->width > hwfc->width || src->height > hwfc->height) + return AVERROR(EINVAL); + + map = av_frame_alloc(); + if (!map) + return AVERROR(ENOMEM); + map->format = src->format; + + err = vaapi_map_frame(hwfc, map, dst, AV_HWFRAME_MAP_WRITE | AV_HWFRAME_MAP_OVERWRITE); + if (err) + goto fail; + + map->width = src->width; + map->height = src->height; + + err = av_frame_copy(map, src); + if (err) + goto fail; + + err = 0; +fail: + av_frame_free(&map); + return err; +} + +static int vaapi_map_to_memory(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ + int err; + + if (dst->format != AV_PIX_FMT_NONE) { + err = vaapi_get_image_format(hwfc->device_ctx, dst->format, NULL); + if (err < 0) + return err; + } + + err = vaapi_map_frame(hwfc, dst, src, flags); + if (err) + return err; + + err = av_frame_copy_props(dst, src); + if (err) + return err; + + return 0; +} + +#if CONFIG_LIBDRM + +#define DRM_MAP(va, layers, ...) { \ + VA_FOURCC_ ## va, \ + layers, \ + { __VA_ARGS__ } \ + } +static const struct { + uint32_t va_fourcc; + int nb_layer_formats; + uint32_t layer_formats[AV_DRM_MAX_PLANES]; +} vaapi_drm_format_map[] = { +#ifdef DRM_FORMAT_R8 + DRM_MAP(NV12, 2, DRM_FORMAT_R8, DRM_FORMAT_RG88), + DRM_MAP(NV12, 2, DRM_FORMAT_R8, DRM_FORMAT_GR88), +#endif + DRM_MAP(NV12, 1, DRM_FORMAT_NV12), +#if defined(VA_FOURCC_P010) && defined(DRM_FORMAT_R16) + DRM_MAP(P010, 2, DRM_FORMAT_R16, DRM_FORMAT_RG1616), +#endif +#if defined(VA_FOURCC_P012) && defined(DRM_FORMAT_R16) + DRM_MAP(P012, 2, DRM_FORMAT_R16, DRM_FORMAT_RG1616), +#endif + DRM_MAP(BGRA, 1, DRM_FORMAT_ARGB8888), + DRM_MAP(BGRX, 1, DRM_FORMAT_XRGB8888), + DRM_MAP(RGBA, 1, DRM_FORMAT_ABGR8888), + DRM_MAP(RGBX, 1, DRM_FORMAT_XBGR8888), +#ifdef VA_FOURCC_ABGR + DRM_MAP(ABGR, 1, DRM_FORMAT_RGBA8888), + DRM_MAP(XBGR, 1, DRM_FORMAT_RGBX8888), +#endif + DRM_MAP(ARGB, 1, DRM_FORMAT_BGRA8888), + DRM_MAP(XRGB, 1, DRM_FORMAT_BGRX8888), +#if defined(VA_FOURCC_XYUV) && defined(DRM_FORMAT_XYUV8888) + DRM_MAP(XYUV, 1, DRM_FORMAT_XYUV8888), +#endif +#if defined(VA_FOURCC_Y412) && defined(DRM_FORMAT_XVYU2101010) + DRM_MAP(Y410, 1, DRM_FORMAT_XVYU2101010), +#endif +#if defined(VA_FOURCC_Y412) && defined(DRM_FORMAT_XVYU12_16161616) + DRM_MAP(Y412, 1, DRM_FORMAT_XVYU12_16161616), +#endif +#if defined(VA_FOURCC_X2R10G10B10) && defined(DRM_FORMAT_XRGB2101010) + DRM_MAP(X2R10G10B10, 1, DRM_FORMAT_XRGB2101010), +#endif +}; +#undef DRM_MAP + +static void vaapi_unmap_from_drm(AVHWFramesContext *dst_fc, + HWMapDescriptor *hwmap) +{ + AVVAAPIDeviceContext *dst_dev = dst_fc->device_ctx->hwctx; + + VASurfaceID surface_id = (VASurfaceID)(uintptr_t)hwmap->priv; + + av_log(dst_fc, AV_LOG_DEBUG, "Destroy surface %#x.\n", surface_id); + + vaDestroySurfaces(dst_dev->display, &surface_id, 1); +} + +static int vaapi_map_from_drm(AVHWFramesContext *src_fc, AVFrame *dst, + const AVFrame *src, int flags) +{ +#if VA_CHECK_VERSION(1, 1, 0) + VAAPIFramesContext *src_vafc = src_fc->internal->priv; + int use_prime2; +#else + int k; +#endif + AVHWFramesContext *dst_fc = + (AVHWFramesContext*)dst->hw_frames_ctx->data; + AVVAAPIDeviceContext *dst_dev = dst_fc->device_ctx->hwctx; + const AVDRMFrameDescriptor *desc; + const VAAPIFormatDescriptor *format_desc; + VASurfaceID surface_id; + VAStatus vas = VA_STATUS_SUCCESS; + uint32_t va_fourcc; + int err, i, j; + +#if !VA_CHECK_VERSION(1, 1, 0) + unsigned long buffer_handle; + VASurfaceAttribExternalBuffers buffer_desc; + VASurfaceAttrib attrs[2] = { + { + .type = VASurfaceAttribMemoryType, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypeInteger, + .value.value.i = VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME, + }, + { + .type = VASurfaceAttribExternalBufferDescriptor, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypePointer, + .value.value.p = &buffer_desc, + } + }; +#endif + + desc = (AVDRMFrameDescriptor*)src->data[0]; + + if (desc->nb_objects != 1) { + av_log(dst_fc, AV_LOG_ERROR, "VAAPI can only map frames " + "made from a single DRM object.\n"); + return AVERROR(EINVAL); + } + + va_fourcc = 0; + for (i = 0; i < FF_ARRAY_ELEMS(vaapi_drm_format_map); i++) { + if (desc->nb_layers != vaapi_drm_format_map[i].nb_layer_formats) + continue; + for (j = 0; j < desc->nb_layers; j++) { + if (desc->layers[j].format != + vaapi_drm_format_map[i].layer_formats[j]) + break; + } + if (j != desc->nb_layers) + continue; + va_fourcc = vaapi_drm_format_map[i].va_fourcc; + break; + } + if (!va_fourcc) { + av_log(dst_fc, AV_LOG_ERROR, "DRM format not supported " + "by VAAPI.\n"); + return AVERROR(EINVAL); + } + + av_log(dst_fc, AV_LOG_DEBUG, "Map DRM object %d to VAAPI as " + "%08x.\n", desc->objects[0].fd, va_fourcc); + + format_desc = vaapi_format_from_fourcc(va_fourcc); + av_assert0(format_desc); + +#if VA_CHECK_VERSION(1, 1, 0) + use_prime2 = !src_vafc->prime_2_import_unsupported && + desc->objects[0].format_modifier != DRM_FORMAT_MOD_INVALID; + if (use_prime2) { + VADRMPRIMESurfaceDescriptor prime_desc; + VASurfaceAttrib prime_attrs[2] = { + { + .type = VASurfaceAttribMemoryType, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypeInteger, + .value.value.i = VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2, + }, + { + .type = VASurfaceAttribExternalBufferDescriptor, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypePointer, + .value.value.p = &prime_desc, + } + }; + prime_desc.fourcc = va_fourcc; + prime_desc.width = src_fc->width; + prime_desc.height = src_fc->height; + prime_desc.num_objects = desc->nb_objects; + for (i = 0; i < desc->nb_objects; ++i) { + prime_desc.objects[i].fd = desc->objects[i].fd; + prime_desc.objects[i].size = desc->objects[i].size; + prime_desc.objects[i].drm_format_modifier = + desc->objects[i].format_modifier; + } + + prime_desc.num_layers = desc->nb_layers; + for (i = 0; i < desc->nb_layers; ++i) { + prime_desc.layers[i].drm_format = desc->layers[i].format; + prime_desc.layers[i].num_planes = desc->layers[i].nb_planes; + for (j = 0; j < desc->layers[i].nb_planes; ++j) { + prime_desc.layers[i].object_index[j] = + desc->layers[i].planes[j].object_index; + prime_desc.layers[i].offset[j] = desc->layers[i].planes[j].offset; + prime_desc.layers[i].pitch[j] = desc->layers[i].planes[j].pitch; + } + + if (format_desc->chroma_planes_swapped && + desc->layers[i].nb_planes == 3) { + FFSWAP(uint32_t, prime_desc.layers[i].pitch[1], + prime_desc.layers[i].pitch[2]); + FFSWAP(uint32_t, prime_desc.layers[i].offset[1], + prime_desc.layers[i].offset[2]); + } + } + + /* + * We can query for PRIME_2 support with vaQuerySurfaceAttributes, but that + * that needs the config_id which we don't have here . Both Intel and + * Gallium seem to do the correct error checks, so lets just try the + * PRIME_2 import first. + */ + vas = vaCreateSurfaces(dst_dev->display, format_desc->rt_format, + src->width, src->height, &surface_id, 1, + prime_attrs, FF_ARRAY_ELEMS(prime_attrs)); + if (vas != VA_STATUS_SUCCESS) + src_vafc->prime_2_import_unsupported = 1; + } + + if (!use_prime2 || vas != VA_STATUS_SUCCESS) { + int k; + unsigned long buffer_handle; + VASurfaceAttribExternalBuffers buffer_desc; + VASurfaceAttrib buffer_attrs[2] = { + { + .type = VASurfaceAttribMemoryType, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypeInteger, + .value.value.i = VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME, + }, + { + .type = VASurfaceAttribExternalBufferDescriptor, + .flags = VA_SURFACE_ATTRIB_SETTABLE, + .value.type = VAGenericValueTypePointer, + .value.value.p = &buffer_desc, + } + }; + + buffer_handle = desc->objects[0].fd; + buffer_desc.pixel_format = va_fourcc; + buffer_desc.width = src_fc->width; + buffer_desc.height = src_fc->height; + buffer_desc.data_size = desc->objects[0].size; + buffer_desc.buffers = &buffer_handle; + buffer_desc.num_buffers = 1; + buffer_desc.flags = 0; + + k = 0; + for (i = 0; i < desc->nb_layers; i++) { + for (j = 0; j < desc->layers[i].nb_planes; j++) { + buffer_desc.pitches[k] = desc->layers[i].planes[j].pitch; + buffer_desc.offsets[k] = desc->layers[i].planes[j].offset; + ++k; + } + } + buffer_desc.num_planes = k; + + if (format_desc->chroma_planes_swapped && + buffer_desc.num_planes == 3) { + FFSWAP(uint32_t, buffer_desc.pitches[1], buffer_desc.pitches[2]); + FFSWAP(uint32_t, buffer_desc.offsets[1], buffer_desc.offsets[2]); + } + + vas = vaCreateSurfaces(dst_dev->display, format_desc->rt_format, + src->width, src->height, + &surface_id, 1, + buffer_attrs, FF_ARRAY_ELEMS(buffer_attrs)); + } +#else + buffer_handle = desc->objects[0].fd; + buffer_desc.pixel_format = va_fourcc; + buffer_desc.width = src_fc->width; + buffer_desc.height = src_fc->height; + buffer_desc.data_size = desc->objects[0].size; + buffer_desc.buffers = &buffer_handle; + buffer_desc.num_buffers = 1; + buffer_desc.flags = 0; + + k = 0; + for (i = 0; i < desc->nb_layers; i++) { + for (j = 0; j < desc->layers[i].nb_planes; j++) { + buffer_desc.pitches[k] = desc->layers[i].planes[j].pitch; + buffer_desc.offsets[k] = desc->layers[i].planes[j].offset; + ++k; + } + } + buffer_desc.num_planes = k; + + if (format_desc->chroma_planes_swapped && + buffer_desc.num_planes == 3) { + FFSWAP(uint32_t, buffer_desc.pitches[1], buffer_desc.pitches[2]); + FFSWAP(uint32_t, buffer_desc.offsets[1], buffer_desc.offsets[2]); + } + + vas = vaCreateSurfaces(dst_dev->display, format_desc->rt_format, + src->width, src->height, + &surface_id, 1, + attrs, FF_ARRAY_ELEMS(attrs)); +#endif + if (vas != VA_STATUS_SUCCESS) { + av_log(dst_fc, AV_LOG_ERROR, "Failed to create surface from DRM " + "object: %d (%s).\n", vas, vaErrorStr(vas)); + return AVERROR(EIO); + } + av_log(dst_fc, AV_LOG_DEBUG, "Create surface %#x.\n", surface_id); + + err = ff_hwframe_map_create(dst->hw_frames_ctx, dst, src, + &vaapi_unmap_from_drm, + (void*)(uintptr_t)surface_id); + if (err < 0) + return err; + + dst->width = src->width; + dst->height = src->height; + dst->data[3] = (uint8_t*)(uintptr_t)surface_id; + + av_log(dst_fc, AV_LOG_DEBUG, "Mapped DRM object %d to " + "surface %#x.\n", desc->objects[0].fd, surface_id); + + return 0; +} + +#if VA_CHECK_VERSION(1, 1, 0) +static void vaapi_unmap_to_drm_esh(AVHWFramesContext *hwfc, + HWMapDescriptor *hwmap) +{ + AVDRMFrameDescriptor *drm_desc = hwmap->priv; + int i; + + for (i = 0; i < drm_desc->nb_objects; i++) + close(drm_desc->objects[i].fd); + + av_freep(&drm_desc); +} + +static int vaapi_map_to_drm_esh(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VASurfaceID surface_id; + VAStatus vas; + VADRMPRIMESurfaceDescriptor va_desc; + AVDRMFrameDescriptor *drm_desc = NULL; + uint32_t export_flags; + int err, i, j; + + surface_id = (VASurfaceID)(uintptr_t)src->data[3]; + + export_flags = VA_EXPORT_SURFACE_SEPARATE_LAYERS; + if (flags & AV_HWFRAME_MAP_READ) { + export_flags |= VA_EXPORT_SURFACE_READ_ONLY; + + vas = vaSyncSurface(hwctx->display, surface_id); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to sync surface " + "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); + return AVERROR(EIO); + } + } + + if (flags & AV_HWFRAME_MAP_WRITE) + export_flags |= VA_EXPORT_SURFACE_WRITE_ONLY; + + vas = vaExportSurfaceHandle(hwctx->display, surface_id, + VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME_2, + export_flags, &va_desc); + if (vas != VA_STATUS_SUCCESS) { + if (vas == VA_STATUS_ERROR_UNIMPLEMENTED) + return AVERROR(ENOSYS); + av_log(hwfc, AV_LOG_ERROR, "Failed to export surface %#x: " + "%d (%s).\n", surface_id, vas, vaErrorStr(vas)); + return AVERROR(EIO); + } + + drm_desc = av_mallocz(sizeof(*drm_desc)); + if (!drm_desc) { + err = AVERROR(ENOMEM); + goto fail; + } + + // By some bizarre coincidence, these structures are very similar... + drm_desc->nb_objects = va_desc.num_objects; + for (i = 0; i < va_desc.num_objects; i++) { + drm_desc->objects[i].fd = va_desc.objects[i].fd; + drm_desc->objects[i].size = va_desc.objects[i].size; + drm_desc->objects[i].format_modifier = + va_desc.objects[i].drm_format_modifier; + } + drm_desc->nb_layers = va_desc.num_layers; + for (i = 0; i < va_desc.num_layers; i++) { + drm_desc->layers[i].format = va_desc.layers[i].drm_format; + drm_desc->layers[i].nb_planes = va_desc.layers[i].num_planes; + for (j = 0; j < va_desc.layers[i].num_planes; j++) { + drm_desc->layers[i].planes[j].object_index = + va_desc.layers[i].object_index[j]; + drm_desc->layers[i].planes[j].offset = + va_desc.layers[i].offset[j]; + drm_desc->layers[i].planes[j].pitch = + va_desc.layers[i].pitch[j]; + } + } + + err = ff_hwframe_map_create(src->hw_frames_ctx, dst, src, + &vaapi_unmap_to_drm_esh, drm_desc); + if (err < 0) + goto fail; + + dst->width = src->width; + dst->height = src->height; + dst->data[0] = (uint8_t*)drm_desc; + + return 0; + +fail: + for (i = 0; i < va_desc.num_objects; i++) + close(va_desc.objects[i].fd); + av_freep(&drm_desc); + return err; +} +#endif + +#if VA_CHECK_VERSION(0, 36, 0) +typedef struct VAAPIDRMImageBufferMapping { + VAImage image; + VABufferInfo buffer_info; + + AVDRMFrameDescriptor drm_desc; +} VAAPIDRMImageBufferMapping; + +static void vaapi_unmap_to_drm_abh(AVHWFramesContext *hwfc, + HWMapDescriptor *hwmap) +{ + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VAAPIDRMImageBufferMapping *mapping = hwmap->priv; + VASurfaceID surface_id; + VAStatus vas; + + surface_id = (VASurfaceID)(uintptr_t)hwmap->source->data[3]; + av_log(hwfc, AV_LOG_DEBUG, "Unmap VAAPI surface %#x from DRM.\n", + surface_id); + + // DRM PRIME file descriptors are closed by vaReleaseBufferHandle(), + // so we shouldn't close them separately. + + vas = vaReleaseBufferHandle(hwctx->display, mapping->image.buf); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to release buffer " + "handle of image %#x (derived from surface %#x): " + "%d (%s).\n", mapping->image.buf, surface_id, + vas, vaErrorStr(vas)); + } + + vas = vaDestroyImage(hwctx->display, mapping->image.image_id); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to destroy image " + "derived from surface %#x: %d (%s).\n", + surface_id, vas, vaErrorStr(vas)); + } + + av_free(mapping); +} + +static int vaapi_map_to_drm_abh(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ + AVVAAPIDeviceContext *hwctx = hwfc->device_ctx->hwctx; + VAAPIDRMImageBufferMapping *mapping = NULL; + VASurfaceID surface_id; + VAStatus vas; + int err, i, p; + + surface_id = (VASurfaceID)(uintptr_t)src->data[3]; + av_log(hwfc, AV_LOG_DEBUG, "Map VAAPI surface %#x to DRM.\n", + surface_id); + + mapping = av_mallocz(sizeof(*mapping)); + if (!mapping) + return AVERROR(ENOMEM); + + vas = vaDeriveImage(hwctx->display, surface_id, + &mapping->image); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to derive image from " + "surface %#x: %d (%s).\n", + surface_id, vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail; + } + + for (i = 0; i < FF_ARRAY_ELEMS(vaapi_drm_format_map); i++) { + if (vaapi_drm_format_map[i].va_fourcc == + mapping->image.format.fourcc) + break; + } + if (i >= FF_ARRAY_ELEMS(vaapi_drm_format_map)) { + av_log(hwfc, AV_LOG_ERROR, "No matching DRM format for " + "VAAPI format %#x.\n", mapping->image.format.fourcc); + err = AVERROR(EINVAL); + goto fail_derived; + } + + mapping->buffer_info.mem_type = + VA_SURFACE_ATTRIB_MEM_TYPE_DRM_PRIME; + + mapping->drm_desc.nb_layers = + vaapi_drm_format_map[i].nb_layer_formats; + if (mapping->drm_desc.nb_layers > 1) { + if (mapping->drm_desc.nb_layers != mapping->image.num_planes) { + av_log(hwfc, AV_LOG_ERROR, "Image properties do not match " + "expected format: got %d planes, but expected %d.\n", + mapping->image.num_planes, mapping->drm_desc.nb_layers); + err = AVERROR(EINVAL); + goto fail_derived; + } + + for(p = 0; p < mapping->drm_desc.nb_layers; p++) { + mapping->drm_desc.layers[p] = (AVDRMLayerDescriptor) { + .format = vaapi_drm_format_map[i].layer_formats[p], + .nb_planes = 1, + .planes[0] = { + .object_index = 0, + .offset = mapping->image.offsets[p], + .pitch = mapping->image.pitches[p], + }, + }; + } + } else { + mapping->drm_desc.layers[0].format = + vaapi_drm_format_map[i].layer_formats[0]; + mapping->drm_desc.layers[0].nb_planes = mapping->image.num_planes; + for (p = 0; p < mapping->image.num_planes; p++) { + mapping->drm_desc.layers[0].planes[p] = (AVDRMPlaneDescriptor) { + .object_index = 0, + .offset = mapping->image.offsets[p], + .pitch = mapping->image.pitches[p], + }; + } + } + + vas = vaAcquireBufferHandle(hwctx->display, mapping->image.buf, + &mapping->buffer_info); + if (vas != VA_STATUS_SUCCESS) { + av_log(hwfc, AV_LOG_ERROR, "Failed to get buffer " + "handle from image %#x (derived from surface %#x): " + "%d (%s).\n", mapping->image.buf, surface_id, + vas, vaErrorStr(vas)); + err = AVERROR(EIO); + goto fail_derived; + } + + av_log(hwfc, AV_LOG_DEBUG, "DRM PRIME fd is %"PRIdPTR".\n", + mapping->buffer_info.handle); + + mapping->drm_desc.nb_objects = 1; + mapping->drm_desc.objects[0] = (AVDRMObjectDescriptor) { + .fd = mapping->buffer_info.handle, + .size = mapping->image.data_size, + // There is no way to get the format modifier with this API. + .format_modifier = DRM_FORMAT_MOD_INVALID, + }; + + err = ff_hwframe_map_create(src->hw_frames_ctx, + dst, src, &vaapi_unmap_to_drm_abh, + mapping); + if (err < 0) + goto fail_mapped; + + dst->data[0] = (uint8_t*)&mapping->drm_desc; + dst->width = src->width; + dst->height = src->height; + + return 0; + +fail_mapped: + vaReleaseBufferHandle(hwctx->display, mapping->image.buf); +fail_derived: + vaDestroyImage(hwctx->display, mapping->image.image_id); +fail: + av_freep(&mapping); + return err; +} +#endif + +static int vaapi_map_to_drm(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ +#if VA_CHECK_VERSION(1, 1, 0) + int err; + err = vaapi_map_to_drm_esh(hwfc, dst, src, flags); + if (err != AVERROR(ENOSYS)) + return err; +#endif +#if VA_CHECK_VERSION(0, 36, 0) + return vaapi_map_to_drm_abh(hwfc, dst, src, flags); +#endif + return AVERROR(ENOSYS); +} + +#endif /* CONFIG_LIBDRM */ + +static int vaapi_map_to(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ + switch (src->format) { +#if CONFIG_LIBDRM + case AV_PIX_FMT_DRM_PRIME: + return vaapi_map_from_drm(hwfc, dst, src, flags); +#endif + default: + return AVERROR(ENOSYS); + } +} + +static int vaapi_map_from(AVHWFramesContext *hwfc, AVFrame *dst, + const AVFrame *src, int flags) +{ + switch (dst->format) { +#if CONFIG_LIBDRM + case AV_PIX_FMT_DRM_PRIME: + return vaapi_map_to_drm(hwfc, dst, src, flags); +#endif + default: + return vaapi_map_to_memory(hwfc, dst, src, flags); + } +} + +static void vaapi_device_free(AVHWDeviceContext *ctx) +{ + AVVAAPIDeviceContext *hwctx = ctx->hwctx; + VAAPIDevicePriv *priv = ctx->user_opaque; + + if (hwctx->display) + vaTerminate(hwctx->display); + +#if HAVE_VAAPI_X11 + if (priv->x11_display) + XCloseDisplay(priv->x11_display); +#endif + + if (priv->drm_fd >= 0) + close(priv->drm_fd); + + av_freep(&priv); +} + +#if CONFIG_VAAPI_1 +static void vaapi_device_log_error(void *context, const char *message) +{ + AVHWDeviceContext *ctx = context; + + av_log(ctx, AV_LOG_ERROR, "libva: %s", message); +} + +static void vaapi_device_log_info(void *context, const char *message) +{ + AVHWDeviceContext *ctx = context; + + av_log(ctx, AV_LOG_VERBOSE, "libva: %s", message); +} +#endif + +static int vaapi_device_connect(AVHWDeviceContext *ctx, + VADisplay display) +{ + AVVAAPIDeviceContext *hwctx = ctx->hwctx; + int major, minor; + VAStatus vas; + +#if CONFIG_VAAPI_1 + vaSetErrorCallback(display, &vaapi_device_log_error, ctx); + vaSetInfoCallback (display, &vaapi_device_log_info, ctx); +#endif + + hwctx->display = display; + + vas = vaInitialize(display, &major, &minor); + if (vas != VA_STATUS_SUCCESS) { + av_log(ctx, AV_LOG_ERROR, "Failed to initialise VAAPI " + "connection: %d (%s).\n", vas, vaErrorStr(vas)); + return AVERROR(EIO); + } + av_log(ctx, AV_LOG_VERBOSE, "Initialised VAAPI connection: " + "version %d.%d\n", major, minor); + + return 0; +} + +static int vaapi_device_create(AVHWDeviceContext *ctx, const char *device, + AVDictionary *opts, int flags) +{ + VAAPIDevicePriv *priv; + VADisplay display = NULL; + const AVDictionaryEntry *ent; + int try_drm, try_x11, try_win32, try_all; + + priv = av_mallocz(sizeof(*priv)); + if (!priv) + return AVERROR(ENOMEM); + + priv->drm_fd = -1; + + ctx->user_opaque = priv; + ctx->free = vaapi_device_free; + + ent = av_dict_get(opts, "connection_type", NULL, 0); + if (ent) { + try_all = try_drm = try_x11 = try_win32 = 0; + if (!strcmp(ent->value, "drm")) { + try_drm = 1; + } else if (!strcmp(ent->value, "x11")) { + try_x11 = 1; + } else if (!strcmp(ent->value, "win32")) { + try_win32 = 1; + } else { + av_log(ctx, AV_LOG_ERROR, "Invalid connection type %s.\n", + ent->value); + return AVERROR(EINVAL); + } + } else { + try_all = 1; + try_drm = HAVE_VAAPI_DRM; + try_x11 = HAVE_VAAPI_X11; + try_win32 = HAVE_VAAPI_WIN32; + } + +#if HAVE_VAAPI_DRM + while (!display && try_drm) { + // If the device is specified, try to open it as a DRM device node. + // If not, look for a usable render node, possibly restricted to those + // using a specified kernel driver. + int loglevel = try_all ? AV_LOG_VERBOSE : AV_LOG_ERROR; + if (device) { + priv->drm_fd = open(device, O_RDWR); + if (priv->drm_fd < 0) { + av_log(ctx, loglevel, "Failed to open %s as " + "DRM device node.\n", device); + break; + } + } else { + char path[64]; + int n, max_devices = 8; +#if CONFIG_LIBDRM + drmVersion *info; + const AVDictionaryEntry *kernel_driver; + kernel_driver = av_dict_get(opts, "kernel_driver", NULL, 0); +#endif + for (n = 0; n < max_devices; n++) { + snprintf(path, sizeof(path), + "/dev/dri/renderD%d", 128 + n); + priv->drm_fd = open(path, O_RDWR); + if (priv->drm_fd < 0) { + if (errno == ENOENT) { + if (n != max_devices - 1) { + av_log(ctx, AV_LOG_VERBOSE, + "No render device %s, try next device for " + "DRM render node.\n", path); + continue; + } + + av_log(ctx, AV_LOG_VERBOSE, "No available render device " + "for DRM render node.\n"); + } else + av_log(ctx, AV_LOG_VERBOSE, "Cannot open " + "DRM render node for device %d.\n", n); + break; + } +#if CONFIG_LIBDRM + info = drmGetVersion(priv->drm_fd); + if (!info) { + av_log(ctx, AV_LOG_VERBOSE, + "Failed to get DRM version for device %d.\n", n); + close(priv->drm_fd); + priv->drm_fd = -1; + continue; + } + if (kernel_driver) { + if (strcmp(kernel_driver->value, info->name)) { + av_log(ctx, AV_LOG_VERBOSE, "Ignoring device %d " + "with non-matching kernel driver (%s).\n", + n, info->name); + drmFreeVersion(info); + close(priv->drm_fd); + priv->drm_fd = -1; + continue; + } + av_log(ctx, AV_LOG_VERBOSE, "Trying to use " + "DRM render node for device %d, " + "with matching kernel driver (%s).\n", + n, info->name); + drmFreeVersion(info); + break; + // drmGetVersion() ensures |info->name| is 0-terminated. + } else if (!strcmp(info->name, "vgem")) { + av_log(ctx, AV_LOG_VERBOSE, + "Skipping vgem node for device %d.\n", n); + drmFreeVersion(info); + close(priv->drm_fd); + priv->drm_fd = -1; + continue; + } + drmFreeVersion(info); +#endif + av_log(ctx, AV_LOG_VERBOSE, "Trying to use " + "DRM render node for device %d.\n", n); + break; + } + if (n >= max_devices) + break; + } + + display = vaGetDisplayDRM(priv->drm_fd); + if (!display) { + av_log(ctx, AV_LOG_VERBOSE, "Cannot open a VA display " + "from DRM device %s.\n", device); + return AVERROR_EXTERNAL; + } + break; + } +#endif + +#if HAVE_VAAPI_X11 + if (!display && try_x11) { + // Try to open the device as an X11 display. + priv->x11_display = XOpenDisplay(device); + if (!priv->x11_display) { + av_log(ctx, AV_LOG_VERBOSE, "Cannot open X11 display " + "%s.\n", XDisplayName(device)); + } else { + display = vaGetDisplay(priv->x11_display); + if (!display) { + av_log(ctx, AV_LOG_ERROR, "Cannot open a VA display " + "from X11 display %s.\n", XDisplayName(device)); + return AVERROR_UNKNOWN; + } + + av_log(ctx, AV_LOG_VERBOSE, "Opened VA display via " + "X11 display %s.\n", XDisplayName(device)); + } + } +#endif + +#if HAVE_VAAPI_WIN32 + if (!display && try_win32) { + // Try to create a display from the specified device, if any. + if (!device) { + display = vaGetDisplayWin32(NULL); + } else { + IDXGIFactory2 *pDXGIFactory = NULL; + IDXGIAdapter *pAdapter = NULL; +#if !HAVE_UWP + HANDLE dxgi = dlopen("dxgi.dll", 0); + if (!dxgi) { + av_log(ctx, AV_LOG_ERROR, "Failed to load dxgi.dll\n"); + return AVERROR_UNKNOWN; + } + PFN_CREATE_DXGI_FACTORY pfnCreateDXGIFactory = + (PFN_CREATE_DXGI_FACTORY)dlsym(dxgi, "CreateDXGIFactory"); + if (!pfnCreateDXGIFactory) { + av_log(ctx, AV_LOG_ERROR, "CreateDXGIFactory load failed\n"); + dlclose(dxgi); + return AVERROR_UNKNOWN; + } +#else + // In UWP (which lacks LoadLibrary), CreateDXGIFactory isn't + // available, only CreateDXGIFactory1 + PFN_CREATE_DXGI_FACTORY pfnCreateDXGIFactory = + (PFN_CREATE_DXGI_FACTORY)CreateDXGIFactory1; +#endif + if (SUCCEEDED(pfnCreateDXGIFactory(&IID_IDXGIFactory2, + (void **)&pDXGIFactory))) { + int adapter = atoi(device); + if (SUCCEEDED(IDXGIFactory2_EnumAdapters(pDXGIFactory, + adapter, + &pAdapter))) { + DXGI_ADAPTER_DESC desc; + if (SUCCEEDED(IDXGIAdapter2_GetDesc(pAdapter, &desc))) { + av_log(ctx, AV_LOG_INFO, + "Using device %04x:%04x (%ls) - LUID %lu %ld.\n", + desc.VendorId, desc.DeviceId, desc.Description, + desc.AdapterLuid.LowPart, + desc.AdapterLuid.HighPart); + display = vaGetDisplayWin32(&desc.AdapterLuid); + } + IDXGIAdapter_Release(pAdapter); + } + IDXGIFactory2_Release(pDXGIFactory); + } +#if !HAVE_UWP + dlclose(dxgi); +#endif + } + + if (!display) { + av_log(ctx, AV_LOG_ERROR, "Cannot open a VA display " + "from Win32 display.\n"); + return AVERROR_UNKNOWN; + } + + av_log(ctx, AV_LOG_VERBOSE, "Opened VA display via " + "Win32 display.\n"); + } +#endif + + if (!display) { + if (device) + av_log(ctx, AV_LOG_ERROR, "No VA display found for " + "device %s.\n", device); + else + av_log(ctx, AV_LOG_ERROR, "No VA display found for " + "any default device.\n"); + return AVERROR(EINVAL); + } + + ent = av_dict_get(opts, "driver", NULL, 0); + if (ent) { +#if VA_CHECK_VERSION(0, 38, 0) + VAStatus vas; + vas = vaSetDriverName(display, ent->value); + if (vas != VA_STATUS_SUCCESS) { + av_log(ctx, AV_LOG_ERROR, "Failed to set driver name to " + "%s: %d (%s).\n", ent->value, vas, vaErrorStr(vas)); + vaTerminate(display); + return AVERROR_EXTERNAL; + } +#else + av_log(ctx, AV_LOG_WARNING, "Driver name setting is not " + "supported with this VAAPI version.\n"); +#endif + } + + return vaapi_device_connect(ctx, display); +} + +static int vaapi_device_derive(AVHWDeviceContext *ctx, + AVHWDeviceContext *src_ctx, + AVDictionary *opts, int flags) +{ +#if HAVE_VAAPI_DRM + if (src_ctx->type == AV_HWDEVICE_TYPE_DRM) { + AVDRMDeviceContext *src_hwctx = src_ctx->hwctx; + VADisplay *display; + VAAPIDevicePriv *priv; + int fd; + + if (src_hwctx->fd < 0) { + av_log(ctx, AV_LOG_ERROR, "DRM instance requires an associated " + "device to derive a VA display from.\n"); + return AVERROR(EINVAL); + } + +#if CONFIG_LIBDRM + { + int node_type = drmGetNodeTypeFromFd(src_hwctx->fd); + char *render_node; + if (node_type < 0) { + av_log(ctx, AV_LOG_ERROR, "DRM instance fd does not appear " + "to refer to a DRM device.\n"); + return AVERROR(EINVAL); + } + if (node_type == DRM_NODE_RENDER) { + fd = src_hwctx->fd; + } else { + render_node = drmGetRenderDeviceNameFromFd(src_hwctx->fd); + if (!render_node) { + av_log(ctx, AV_LOG_VERBOSE, "Using non-render node " + "because the device does not have an " + "associated render node.\n"); + fd = src_hwctx->fd; + } else { + fd = open(render_node, O_RDWR); + if (fd < 0) { + av_log(ctx, AV_LOG_VERBOSE, "Using non-render node " + "because the associated render node " + "could not be opened.\n"); + fd = src_hwctx->fd; + } else { + av_log(ctx, AV_LOG_VERBOSE, "Using render node %s " + "in place of non-render DRM device.\n", + render_node); + } + free(render_node); + } + } + } +#else + fd = src_hwctx->fd; +#endif + + priv = av_mallocz(sizeof(*priv)); + if (!priv) { + if (fd != src_hwctx->fd) { + // The fd was opened in this function. + close(fd); + } + return AVERROR(ENOMEM); + } + + if (fd == src_hwctx->fd) { + // The fd is inherited from the source context and we are holding + // a reference to that, we don't want to close it from here. + priv->drm_fd = -1; + } else { + priv->drm_fd = fd; + } + + ctx->user_opaque = priv; + ctx->free = &vaapi_device_free; + + display = vaGetDisplayDRM(fd); + if (!display) { + av_log(ctx, AV_LOG_ERROR, "Failed to open a VA display from " + "DRM device.\n"); + return AVERROR(EIO); + } + + return vaapi_device_connect(ctx, display); + } +#endif + return AVERROR(ENOSYS); +} + +const HWContextType ff_hwcontext_type_vaapi = { + .type = AV_HWDEVICE_TYPE_VAAPI, + .name = "VAAPI", + + .device_hwctx_size = sizeof(AVVAAPIDeviceContext), + .device_priv_size = sizeof(VAAPIDeviceContext), + .device_hwconfig_size = sizeof(AVVAAPIHWConfig), + .frames_hwctx_size = sizeof(AVVAAPIFramesContext), + .frames_priv_size = sizeof(VAAPIFramesContext), + + .device_create = &vaapi_device_create, + .device_derive = &vaapi_device_derive, + .device_init = &vaapi_device_init, + .device_uninit = &vaapi_device_uninit, + .frames_get_constraints = &vaapi_frames_get_constraints, + .frames_init = &vaapi_frames_init, + .frames_uninit = &vaapi_frames_uninit, + .frames_get_buffer = &vaapi_get_buffer, + .transfer_get_formats = &vaapi_transfer_get_formats, + .transfer_data_to = &vaapi_transfer_data_to, + .transfer_data_from = &vaapi_transfer_data_from, + .map_to = &vaapi_map_to, + .map_from = &vaapi_map_from, + + .pix_fmts = (const enum AVPixelFormat[]) { + AV_PIX_FMT_VAAPI, + AV_PIX_FMT_NONE + }, +}; diff --git a/media/ffvpx/libavutil/hwcontext_vaapi.h b/media/ffvpx/libavutil/hwcontext_vaapi.h new file mode 100644 index 0000000000..0b2e071cb3 --- /dev/null +++ b/media/ffvpx/libavutil/hwcontext_vaapi.h @@ -0,0 +1,117 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_HWCONTEXT_VAAPI_H +#define AVUTIL_HWCONTEXT_VAAPI_H + +#include <va/va.h> + +/** + * @file + * API-specific header for AV_HWDEVICE_TYPE_VAAPI. + * + * Dynamic frame pools are supported, but note that any pool used as a render + * target is required to be of fixed size in order to be be usable as an + * argument to vaCreateContext(). + * + * For user-allocated pools, AVHWFramesContext.pool must return AVBufferRefs + * with the data pointer set to a VASurfaceID. + */ + +enum { + /** + * The quirks field has been set by the user and should not be detected + * automatically by av_hwdevice_ctx_init(). + */ + AV_VAAPI_DRIVER_QUIRK_USER_SET = (1 << 0), + /** + * The driver does not destroy parameter buffers when they are used by + * vaRenderPicture(). Additional code will be required to destroy them + * separately afterwards. + */ + AV_VAAPI_DRIVER_QUIRK_RENDER_PARAM_BUFFERS = (1 << 1), + + /** + * The driver does not support the VASurfaceAttribMemoryType attribute, + * so the surface allocation code will not try to use it. + */ + AV_VAAPI_DRIVER_QUIRK_ATTRIB_MEMTYPE = (1 << 2), + + /** + * The driver does not support surface attributes at all. + * The surface allocation code will never pass them to surface allocation, + * and the results of the vaQuerySurfaceAttributes() call will be faked. + */ + AV_VAAPI_DRIVER_QUIRK_SURFACE_ATTRIBUTES = (1 << 3), +}; + +/** + * VAAPI connection details. + * + * Allocated as AVHWDeviceContext.hwctx + */ +typedef struct AVVAAPIDeviceContext { + /** + * The VADisplay handle, to be filled by the user. + */ + VADisplay display; + /** + * Driver quirks to apply - this is filled by av_hwdevice_ctx_init(), + * with reference to a table of known drivers, unless the + * AV_VAAPI_DRIVER_QUIRK_USER_SET bit is already present. The user + * may need to refer to this field when performing any later + * operations using VAAPI with the same VADisplay. + */ + unsigned int driver_quirks; +} AVVAAPIDeviceContext; + +/** + * VAAPI-specific data associated with a frame pool. + * + * Allocated as AVHWFramesContext.hwctx. + */ +typedef struct AVVAAPIFramesContext { + /** + * Set by the user to apply surface attributes to all surfaces in + * the frame pool. If null, default settings are used. + */ + VASurfaceAttrib *attributes; + int nb_attributes; + /** + * The surfaces IDs of all surfaces in the pool after creation. + * Only valid if AVHWFramesContext.initial_pool_size was positive. + * These are intended to be used as the render_targets arguments to + * vaCreateContext(). + */ + VASurfaceID *surface_ids; + int nb_surfaces; +} AVVAAPIFramesContext; + +/** + * VAAPI hardware pipeline configuration details. + * + * Allocated with av_hwdevice_hwconfig_alloc(). + */ +typedef struct AVVAAPIHWConfig { + /** + * ID of a VAAPI pipeline configuration. + */ + VAConfigID config_id; +} AVVAAPIHWConfig; + +#endif /* AVUTIL_HWCONTEXT_VAAPI_H */ diff --git a/media/ffvpx/libavutil/imgutils.c b/media/ffvpx/libavutil/imgutils.c new file mode 100644 index 0000000000..1e15f7c920 --- /dev/null +++ b/media/ffvpx/libavutil/imgutils.c @@ -0,0 +1,717 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * misc image utilities + */ + +#include "avassert.h" +#include "common.h" +#include "imgutils.h" +#include "imgutils_internal.h" +#include "internal.h" +#include "intreadwrite.h" +#include "log.h" +#include "mathematics.h" +#include "pixdesc.h" +#include "rational.h" + +void av_image_fill_max_pixsteps(int max_pixsteps[4], int max_pixstep_comps[4], + const AVPixFmtDescriptor *pixdesc) +{ + int i; + memset(max_pixsteps, 0, 4*sizeof(max_pixsteps[0])); + if (max_pixstep_comps) + memset(max_pixstep_comps, 0, 4*sizeof(max_pixstep_comps[0])); + + for (i = 0; i < 4; i++) { + const AVComponentDescriptor *comp = &(pixdesc->comp[i]); + if (comp->step > max_pixsteps[comp->plane]) { + max_pixsteps[comp->plane] = comp->step; + if (max_pixstep_comps) + max_pixstep_comps[comp->plane] = i; + } + } +} + +static inline +int image_get_linesize(int width, int plane, + int max_step, int max_step_comp, + const AVPixFmtDescriptor *desc) +{ + int s, shifted_w, linesize; + + if (!desc) + return AVERROR(EINVAL); + + if (width < 0) + return AVERROR(EINVAL); + s = (max_step_comp == 1 || max_step_comp == 2) ? desc->log2_chroma_w : 0; + shifted_w = ((width + (1 << s) - 1)) >> s; + if (shifted_w && max_step > INT_MAX / shifted_w) + return AVERROR(EINVAL); + linesize = max_step * shifted_w; + + if (desc->flags & AV_PIX_FMT_FLAG_BITSTREAM) + linesize = (linesize + 7) >> 3; + return linesize; +} + +int av_image_get_linesize(enum AVPixelFormat pix_fmt, int width, int plane) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int max_step [4]; /* max pixel step for each plane */ + int max_step_comp[4]; /* the component for each plane which has the max pixel step */ + + if (!desc || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return AVERROR(EINVAL); + + av_image_fill_max_pixsteps(max_step, max_step_comp, desc); + return image_get_linesize(width, plane, max_step[plane], max_step_comp[plane], desc); +} + +int av_image_fill_linesizes(int linesizes[4], enum AVPixelFormat pix_fmt, int width) +{ + int i, ret; + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int max_step [4]; /* max pixel step for each plane */ + int max_step_comp[4]; /* the component for each plane which has the max pixel step */ + + memset(linesizes, 0, 4*sizeof(linesizes[0])); + + if (!desc || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return AVERROR(EINVAL); + + av_image_fill_max_pixsteps(max_step, max_step_comp, desc); + for (i = 0; i < 4; i++) { + if ((ret = image_get_linesize(width, i, max_step[i], max_step_comp[i], desc)) < 0) + return ret; + linesizes[i] = ret; + } + + return 0; +} + +int av_image_fill_plane_sizes(size_t sizes[4], enum AVPixelFormat pix_fmt, + int height, const ptrdiff_t linesizes[4]) +{ + int i, has_plane[4] = { 0 }; + + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + memset(sizes , 0, sizeof(sizes[0])*4); + + if (!desc || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return AVERROR(EINVAL); + + if (linesizes[0] > SIZE_MAX / height) + return AVERROR(EINVAL); + sizes[0] = linesizes[0] * (size_t)height; + + if (desc->flags & AV_PIX_FMT_FLAG_PAL) { + sizes[1] = 256 * 4; /* palette is stored here as 256 32 bits words */ + return 0; + } + + for (i = 0; i < 4; i++) + has_plane[desc->comp[i].plane] = 1; + + for (i = 1; i < 4 && has_plane[i]; i++) { + int h, s = (i == 1 || i == 2) ? desc->log2_chroma_h : 0; + h = (height + (1 << s) - 1) >> s; + if (linesizes[i] > SIZE_MAX / h) + return AVERROR(EINVAL); + sizes[i] = (size_t)h * linesizes[i]; + } + + return 0; +} + +int av_image_fill_pointers(uint8_t *data[4], enum AVPixelFormat pix_fmt, int height, + uint8_t *ptr, const int linesizes[4]) +{ + int i, ret; + ptrdiff_t linesizes1[4]; + size_t sizes[4]; + + memset(data , 0, sizeof(data[0])*4); + + for (i = 0; i < 4; i++) + linesizes1[i] = linesizes[i]; + + ret = av_image_fill_plane_sizes(sizes, pix_fmt, height, linesizes1); + if (ret < 0) + return ret; + + ret = 0; + for (i = 0; i < 4; i++) { + if (sizes[i] > INT_MAX - ret) + return AVERROR(EINVAL); + ret += sizes[i]; + } + + if (!ptr) + return ret; + + data[0] = ptr; + for (i = 1; i < 4 && sizes[i]; i++) + data[i] = data[i - 1] + sizes[i - 1]; + + return ret; +} + +int avpriv_set_systematic_pal2(uint32_t pal[256], enum AVPixelFormat pix_fmt) +{ + int i; + + for (i = 0; i < 256; i++) { + int r, g, b; + + switch (pix_fmt) { + case AV_PIX_FMT_RGB8: + r = (i>>5 )*36; + g = ((i>>2)&7)*36; + b = (i&3 )*85; + break; + case AV_PIX_FMT_BGR8: + b = (i>>6 )*85; + g = ((i>>3)&7)*36; + r = (i&7 )*36; + break; + case AV_PIX_FMT_RGB4_BYTE: + r = (i>>3 )*255; + g = ((i>>1)&3)*85; + b = (i&1 )*255; + break; + case AV_PIX_FMT_BGR4_BYTE: + b = (i>>3 )*255; + g = ((i>>1)&3)*85; + r = (i&1 )*255; + break; + case AV_PIX_FMT_GRAY8: + r = b = g = i; + break; + default: + return AVERROR(EINVAL); + } + pal[i] = b + (g << 8) + (r << 16) + (0xFFU << 24); + } + + return 0; +} + +int av_image_alloc(uint8_t *pointers[4], int linesizes[4], + int w, int h, enum AVPixelFormat pix_fmt, int align) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int i, ret; + ptrdiff_t linesizes1[4]; + size_t total_size, sizes[4]; + uint8_t *buf; + + if (!desc) + return AVERROR(EINVAL); + + if ((ret = av_image_check_size(w, h, 0, NULL)) < 0) + return ret; + if ((ret = av_image_fill_linesizes(linesizes, pix_fmt, align>7 ? FFALIGN(w, 8) : w)) < 0) + return ret; + + for (i = 0; i < 4; i++) { + linesizes[i] = FFALIGN(linesizes[i], align); + linesizes1[i] = linesizes[i]; + } + + if ((ret = av_image_fill_plane_sizes(sizes, pix_fmt, h, linesizes1)) < 0) + return ret; + total_size = align; + for (i = 0; i < 4; i++) { + if (total_size > SIZE_MAX - sizes[i]) + return AVERROR(EINVAL); + total_size += sizes[i]; + } + buf = av_malloc(total_size); + if (!buf) + return AVERROR(ENOMEM); + if ((ret = av_image_fill_pointers(pointers, pix_fmt, h, buf, linesizes)) < 0) { + av_free(buf); + return ret; + } + if (desc->flags & AV_PIX_FMT_FLAG_PAL) { + avpriv_set_systematic_pal2((uint32_t*)pointers[1], pix_fmt); + if (align < 4) { + av_log(NULL, AV_LOG_ERROR, "Formats with a palette require a minimum alignment of 4\n"); + av_free(buf); + return AVERROR(EINVAL); + } + } + + if (desc->flags & AV_PIX_FMT_FLAG_PAL && pointers[1] && + pointers[1] - pointers[0] > linesizes[0] * h) { + /* zero-initialize the padding before the palette */ + memset(pointers[0] + linesizes[0] * h, 0, + pointers[1] - pointers[0] - linesizes[0] * h); + } + + return ret; +} + +typedef struct ImgUtils { + const AVClass *class; + int log_offset; + void *log_ctx; +} ImgUtils; + +static const AVClass imgutils_class = { + .class_name = "IMGUTILS", + .item_name = av_default_item_name, + .option = NULL, + .version = LIBAVUTIL_VERSION_INT, + .log_level_offset_offset = offsetof(ImgUtils, log_offset), + .parent_log_context_offset = offsetof(ImgUtils, log_ctx), +}; + +int av_image_check_size2(unsigned int w, unsigned int h, int64_t max_pixels, enum AVPixelFormat pix_fmt, int log_offset, void *log_ctx) +{ + ImgUtils imgutils = { + .class = &imgutils_class, + .log_offset = log_offset, + .log_ctx = log_ctx, + }; + int64_t stride = av_image_get_linesize(pix_fmt, w, 0); + if (stride <= 0) + stride = 8LL*w; + stride += 128*8; + + if ((int)w<=0 || (int)h<=0 || stride >= INT_MAX || stride*(uint64_t)(h+128) >= INT_MAX) { + av_log(&imgutils, AV_LOG_ERROR, "Picture size %ux%u is invalid\n", w, h); + return AVERROR(EINVAL); + } + + if (max_pixels < INT64_MAX) { + if (w*(int64_t)h > max_pixels) { + av_log(&imgutils, AV_LOG_ERROR, + "Picture size %ux%u exceeds specified max pixel count %"PRId64", see the documentation if you wish to increase it\n", + w, h, max_pixels); + return AVERROR(EINVAL); + } + } + + return 0; +} + +int av_image_check_size(unsigned int w, unsigned int h, int log_offset, void *log_ctx) +{ + return av_image_check_size2(w, h, INT64_MAX, AV_PIX_FMT_NONE, log_offset, log_ctx); +} + +int av_image_check_sar(unsigned int w, unsigned int h, AVRational sar) +{ + int64_t scaled_dim; + + if (sar.den <= 0 || sar.num < 0) + return AVERROR(EINVAL); + + if (!sar.num || sar.num == sar.den) + return 0; + + if (sar.num < sar.den) + scaled_dim = av_rescale_rnd(w, sar.num, sar.den, AV_ROUND_ZERO); + else + scaled_dim = av_rescale_rnd(h, sar.den, sar.num, AV_ROUND_ZERO); + + if (scaled_dim > 0) + return 0; + + return AVERROR(EINVAL); +} + +static void image_copy_plane(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height) +{ + if (!dst || !src) + return; + av_assert0(FFABS(src_linesize) >= bytewidth); + av_assert0(FFABS(dst_linesize) >= bytewidth); + for (;height > 0; height--) { + memcpy(dst, src, bytewidth); + dst += dst_linesize; + src += src_linesize; + } +} + +void av_image_copy_plane_uc_from(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height) +{ + int ret = -1; + +#if ARCH_X86 + ret = ff_image_copy_plane_uc_from_x86(dst, dst_linesize, src, src_linesize, + bytewidth, height); +#endif + + if (ret < 0) + image_copy_plane(dst, dst_linesize, src, src_linesize, bytewidth, height); +} + +void av_image_copy_plane(uint8_t *dst, int dst_linesize, + const uint8_t *src, int src_linesize, + int bytewidth, int height) +{ + image_copy_plane(dst, dst_linesize, src, src_linesize, bytewidth, height); +} + +static void image_copy(uint8_t *const dst_data[4], const ptrdiff_t dst_linesizes[4], + const uint8_t *const src_data[4], const ptrdiff_t src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height, + void (*copy_plane)(uint8_t *, ptrdiff_t, const uint8_t *, + ptrdiff_t, ptrdiff_t, int)) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + + if (!desc || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return; + + if (desc->flags & AV_PIX_FMT_FLAG_PAL) { + copy_plane(dst_data[0], dst_linesizes[0], + src_data[0], src_linesizes[0], + width, height); + /* copy the palette */ + if ((desc->flags & AV_PIX_FMT_FLAG_PAL) || (dst_data[1] && src_data[1])) + memcpy(dst_data[1], src_data[1], 4*256); + } else { + int i, planes_nb = 0; + + for (i = 0; i < desc->nb_components; i++) + planes_nb = FFMAX(planes_nb, desc->comp[i].plane + 1); + + for (i = 0; i < planes_nb; i++) { + int h = height; + ptrdiff_t bwidth = av_image_get_linesize(pix_fmt, width, i); + if (bwidth < 0) { + av_log(NULL, AV_LOG_ERROR, "av_image_get_linesize failed\n"); + return; + } + if (i == 1 || i == 2) { + h = AV_CEIL_RSHIFT(height, desc->log2_chroma_h); + } + copy_plane(dst_data[i], dst_linesizes[i], + src_data[i], src_linesizes[i], + bwidth, h); + } + } +} + +void av_image_copy(uint8_t *const dst_data[4], const int dst_linesizes[4], + const uint8_t * const src_data[4], const int src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height) +{ + ptrdiff_t dst_linesizes1[4], src_linesizes1[4]; + int i; + + for (i = 0; i < 4; i++) { + dst_linesizes1[i] = dst_linesizes[i]; + src_linesizes1[i] = src_linesizes[i]; + } + + image_copy(dst_data, dst_linesizes1, src_data, src_linesizes1, pix_fmt, + width, height, image_copy_plane); +} + +void av_image_copy_uc_from(uint8_t * const dst_data[4], const ptrdiff_t dst_linesizes[4], + const uint8_t * const src_data[4], const ptrdiff_t src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height) +{ + image_copy(dst_data, dst_linesizes, src_data, src_linesizes, pix_fmt, + width, height, av_image_copy_plane_uc_from); +} + +int av_image_fill_arrays(uint8_t *dst_data[4], int dst_linesize[4], + const uint8_t *src, enum AVPixelFormat pix_fmt, + int width, int height, int align) +{ + int ret, i; + + ret = av_image_check_size(width, height, 0, NULL); + if (ret < 0) + return ret; + + ret = av_image_fill_linesizes(dst_linesize, pix_fmt, width); + if (ret < 0) + return ret; + + for (i = 0; i < 4; i++) + dst_linesize[i] = FFALIGN(dst_linesize[i], align); + + return av_image_fill_pointers(dst_data, pix_fmt, height, (uint8_t *)src, dst_linesize); +} + +int av_image_get_buffer_size(enum AVPixelFormat pix_fmt, + int width, int height, int align) +{ + int ret, i; + int linesize[4]; + ptrdiff_t aligned_linesize[4]; + size_t sizes[4]; + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + if (!desc) + return AVERROR(EINVAL); + + ret = av_image_check_size(width, height, 0, NULL); + if (ret < 0) + return ret; + + ret = av_image_fill_linesizes(linesize, pix_fmt, width); + if (ret < 0) + return ret; + + for (i = 0; i < 4; i++) + aligned_linesize[i] = FFALIGN(linesize[i], align); + + ret = av_image_fill_plane_sizes(sizes, pix_fmt, height, aligned_linesize); + if (ret < 0) + return ret; + + ret = 0; + for (i = 0; i < 4; i++) { + if (sizes[i] > INT_MAX - ret) + return AVERROR(EINVAL); + ret += sizes[i]; + } + return ret; +} + +int av_image_copy_to_buffer(uint8_t *dst, int dst_size, + const uint8_t * const src_data[4], + const int src_linesize[4], + enum AVPixelFormat pix_fmt, + int width, int height, int align) +{ + int i, j, nb_planes = 0, linesize[4]; + int size = av_image_get_buffer_size(pix_fmt, width, height, align); + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int ret; + + if (size > dst_size || size < 0 || !desc) + return AVERROR(EINVAL); + + for (i = 0; i < desc->nb_components; i++) + nb_planes = FFMAX(desc->comp[i].plane, nb_planes); + + nb_planes++; + + ret = av_image_fill_linesizes(linesize, pix_fmt, width); + av_assert0(ret >= 0); // was checked previously + + for (i = 0; i < nb_planes; i++) { + int h, shift = (i == 1 || i == 2) ? desc->log2_chroma_h : 0; + const uint8_t *src = src_data[i]; + h = (height + (1 << shift) - 1) >> shift; + + for (j = 0; j < h; j++) { + memcpy(dst, src, linesize[i]); + dst += FFALIGN(linesize[i], align); + src += src_linesize[i]; + } + } + + if (desc->flags & AV_PIX_FMT_FLAG_PAL) { + uint32_t *d32 = (uint32_t *)dst; + + for (i = 0; i<256; i++) + AV_WL32(d32 + i, AV_RN32(src_data[1] + 4*i)); + } + + return size; +} + +// Fill dst[0..dst_size] with the bytes in clear[0..clear_size]. The clear +// bytes are repeated until dst_size is reached. If dst_size is unaligned (i.e. +// dst_size%clear_size!=0), the remaining data will be filled with the beginning +// of the clear data only. +static void memset_bytes(uint8_t *dst, size_t dst_size, uint8_t *clear, + size_t clear_size) +{ + int same = 1; + int i; + + if (!clear_size) + return; + + // Reduce to memset() if possible. + for (i = 0; i < clear_size; i++) { + if (clear[i] != clear[0]) { + same = 0; + break; + } + } + if (same) + clear_size = 1; + + if (clear_size == 1) { + memset(dst, clear[0], dst_size); + } else { + if (clear_size > dst_size) + clear_size = dst_size; + memcpy(dst, clear, clear_size); + av_memcpy_backptr(dst + clear_size, clear_size, dst_size - clear_size); + } +} + +// Maximum size in bytes of a plane element (usually a pixel, or multiple pixels +// if it's a subsampled packed format). +#define MAX_BLOCK_SIZE 32 + +int av_image_fill_color(uint8_t * const dst_data[4], const ptrdiff_t dst_linesize[4], + enum AVPixelFormat pix_fmt, const uint32_t color[4], + int width, int height, int flags) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int nb_planes = av_pix_fmt_count_planes(pix_fmt); + // A pixel or a group of pixels on each plane, with a value that represents the color. + // Consider e.g. AV_PIX_FMT_UYVY422 for non-trivial cases. + uint8_t clear_block[4][MAX_BLOCK_SIZE] = {{0}}; // clear padding with 0 + int clear_block_size[4] = {0}; + ptrdiff_t plane_line_bytes[4] = {0}; + int bitstream; + int plane, c; + + if (!desc || nb_planes < 1 || nb_planes > 4 || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return AVERROR(EINVAL); + + bitstream = !!(desc->flags & AV_PIX_FMT_FLAG_BITSTREAM); + + for (c = 0; c < desc->nb_components; c++) { + const AVComponentDescriptor comp = desc->comp[c]; + + // We try to operate on entire non-subsampled pixel groups (for + // AV_PIX_FMT_UYVY422 this would mean two consecutive pixels). + clear_block_size[comp.plane] = FFMAX(clear_block_size[comp.plane], comp.step); + + if (clear_block_size[comp.plane] > MAX_BLOCK_SIZE) + return AVERROR(EINVAL); + } + + // Create a byte array for clearing 1 pixel (sometimes several pixels). + for (c = 0; c < desc->nb_components; c++) { + const AVComponentDescriptor comp = desc->comp[c]; + // (Multiple pixels happen e.g. with AV_PIX_FMT_UYVY422.) + int w = (bitstream ? 8 : 1) * clear_block_size[comp.plane] / comp.step; + uint8_t *c_data[4]; + const int c_linesize[4] = {0}; + uint32_t src_array[MAX_BLOCK_SIZE]; + int x; + + if (comp.depth > 32) + return AVERROR(EINVAL); + if (w < 1) + return AVERROR(EINVAL); + + for (x = 0; x < w; x++) + src_array[x] = color[c]; + + for (x = 0; x < 4; x++) + c_data[x] = &clear_block[x][0]; + + av_write_image_line2(src_array, c_data, c_linesize, desc, 0, 0, c, w, 4); + } + + for (plane = 0; plane < nb_planes; plane++) { + plane_line_bytes[plane] = av_image_get_linesize(pix_fmt, width, plane); + if (plane_line_bytes[plane] < 0) + return AVERROR(EINVAL); + } + + if (!dst_data) + return 0; + + for (plane = 0; plane < nb_planes; plane++) { + size_t bytewidth = plane_line_bytes[plane]; + uint8_t *data = dst_data[plane]; + int chroma_div = plane == 1 || plane == 2 ? desc->log2_chroma_h : 0; + int plane_h = ((height + ( 1 << chroma_div) - 1)) >> chroma_div; + + for (; plane_h > 0; plane_h--) { + memset_bytes(data, bytewidth, &clear_block[plane][0], clear_block_size[plane]); + data += dst_linesize[plane]; + } + } + + return 0; +} + +int av_image_fill_black(uint8_t * const dst_data[4], const ptrdiff_t dst_linesize[4], + enum AVPixelFormat pix_fmt, enum AVColorRange range, + int width, int height) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int nb_planes = av_pix_fmt_count_planes(pix_fmt); + int rgb, xyz, pal, limited, alpha, fltp; + uint32_t colors[4] = {0}; + + if (!desc || nb_planes < 1 || nb_planes > 4 || desc->flags & AV_PIX_FMT_FLAG_HWACCEL) + return AVERROR(EINVAL); + + rgb = !!(desc->flags & AV_PIX_FMT_FLAG_RGB); + xyz = !!(desc->flags & AV_PIX_FMT_FLAG_XYZ); + pal = !!(desc->flags & AV_PIX_FMT_FLAG_PAL); + limited = !rgb && !xyz && !pal && range != AVCOL_RANGE_JPEG; + alpha = !pal && !!(desc->flags & AV_PIX_FMT_FLAG_ALPHA); + fltp = !!(desc->flags & AV_PIX_FMT_FLAG_FLOAT); + + for (int c = 0; c < desc->nb_components; c++) { + const AVComponentDescriptor comp = desc->comp[c]; + uint32_t color = 0; + + if (comp.depth > 32) + return AVERROR(EINVAL); + + if (pix_fmt == AV_PIX_FMT_MONOWHITE) { + color = 1; + } else if (c + 1 == desc->nb_components && alpha) { + // (Assume even limited YUV uses full range alpha.) + if (fltp) { + if (comp.depth != 16 && comp.depth != 32) + return AVERROR(EINVAL); + color = (comp.depth == 16 ? 0x3C00 : 0x3F800000); // 1.0 + } else { + color = (comp.depth == 32 ? 0 : (1 << comp.depth)) - 1; + } + } else if (c == 0 && limited && comp.depth > 1) { + if (comp.depth < 8 || (fltp && comp.depth != 16 && comp.depth != 32)) + return AVERROR(EINVAL); + if (fltp) + color = (comp.depth == 16 ? 0x2C00 : 0x3D800000); // 0.0625 + else + color = 16 << (comp.depth - 8); + } else if ((c == 1 || c == 2) && !rgb && !xyz) { + if (comp.depth < 8 || fltp && comp.depth != 16 && comp.depth != 32) + return AVERROR(EINVAL); + if (fltp) + color = (comp.depth == 16 ? 0x3800 : 0x3F000000); // 0.5 + else + color = 128 << (comp.depth - 8); + } + + colors[c] = color; + } + + return av_image_fill_color(dst_data, dst_linesize, pix_fmt, colors, width, height, 0); +} diff --git a/media/ffvpx/libavutil/imgutils.h b/media/ffvpx/libavutil/imgutils.h new file mode 100644 index 0000000000..123a9e5c0d --- /dev/null +++ b/media/ffvpx/libavutil/imgutils.h @@ -0,0 +1,377 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_IMGUTILS_H +#define AVUTIL_IMGUTILS_H + +/** + * @file + * misc image utilities + * + * @addtogroup lavu_picture + * @{ + */ + +#include <stddef.h> +#include <stdint.h> +#include "pixdesc.h" +#include "pixfmt.h" +#include "rational.h" + +/** + * Compute the max pixel step for each plane of an image with a + * format described by pixdesc. + * + * The pixel step is the distance in bytes between the first byte of + * the group of bytes which describe a pixel component and the first + * byte of the successive group in the same plane for the same + * component. + * + * @param max_pixsteps an array which is filled with the max pixel step + * for each plane. Since a plane may contain different pixel + * components, the computed max_pixsteps[plane] is relative to the + * component in the plane with the max pixel step. + * @param max_pixstep_comps an array which is filled with the component + * for each plane which has the max pixel step. May be NULL. + * @param pixdesc the AVPixFmtDescriptor for the image, describing its format + */ +void av_image_fill_max_pixsteps(int max_pixsteps[4], int max_pixstep_comps[4], + const AVPixFmtDescriptor *pixdesc); + +/** + * Compute the size of an image line with format pix_fmt and width + * width for the plane plane. + * + * @return the computed size in bytes + */ +int av_image_get_linesize(enum AVPixelFormat pix_fmt, int width, int plane); + +/** + * Fill plane linesizes for an image with pixel format pix_fmt and + * width width. + * + * @param linesizes array to be filled with the linesize for each plane + * @param pix_fmt the AVPixelFormat of the image + * @param width width of the image in pixels + * @return >= 0 in case of success, a negative error code otherwise + */ +int av_image_fill_linesizes(int linesizes[4], enum AVPixelFormat pix_fmt, int width); + +/** + * Fill plane sizes for an image with pixel format pix_fmt and height height. + * + * @param size the array to be filled with the size of each image plane + * @param pix_fmt the AVPixelFormat of the image + * @param height height of the image in pixels + * @param linesizes the array containing the linesize for each + * plane, should be filled by av_image_fill_linesizes() + * @return >= 0 in case of success, a negative error code otherwise + * + * @note The linesize parameters have the type ptrdiff_t here, while they are + * int for av_image_fill_linesizes(). + */ +int av_image_fill_plane_sizes(size_t size[4], enum AVPixelFormat pix_fmt, + int height, const ptrdiff_t linesizes[4]); + +/** + * Fill plane data pointers for an image with pixel format pix_fmt and + * height height. + * + * @param data pointers array to be filled with the pointer for each image plane + * @param pix_fmt the AVPixelFormat of the image + * @param height height of the image in pixels + * @param ptr the pointer to a buffer which will contain the image + * @param linesizes the array containing the linesize for each + * plane, should be filled by av_image_fill_linesizes() + * @return the size in bytes required for the image buffer, a negative + * error code in case of failure + */ +int av_image_fill_pointers(uint8_t *data[4], enum AVPixelFormat pix_fmt, int height, + uint8_t *ptr, const int linesizes[4]); + +/** + * Allocate an image with size w and h and pixel format pix_fmt, and + * fill pointers and linesizes accordingly. + * The allocated image buffer has to be freed by using + * av_freep(&pointers[0]). + * + * @param pointers array to be filled with the pointer for each image plane + * @param linesizes the array filled with the linesize for each plane + * @param w width of the image in pixels + * @param h height of the image in pixels + * @param pix_fmt the AVPixelFormat of the image + * @param align the value to use for buffer size alignment + * @return the size in bytes required for the image buffer, a negative + * error code in case of failure + */ +int av_image_alloc(uint8_t *pointers[4], int linesizes[4], + int w, int h, enum AVPixelFormat pix_fmt, int align); + +/** + * Copy image plane from src to dst. + * That is, copy "height" number of lines of "bytewidth" bytes each. + * The first byte of each successive line is separated by *_linesize + * bytes. + * + * bytewidth must be contained by both absolute values of dst_linesize + * and src_linesize, otherwise the function behavior is undefined. + * + * @param dst destination plane to copy to + * @param dst_linesize linesize for the image plane in dst + * @param src source plane to copy from + * @param src_linesize linesize for the image plane in src + * @param height height (number of lines) of the plane + */ +void av_image_copy_plane(uint8_t *dst, int dst_linesize, + const uint8_t *src, int src_linesize, + int bytewidth, int height); + +/** + * Copy image data located in uncacheable (e.g. GPU mapped) memory. Where + * available, this function will use special functionality for reading from such + * memory, which may result in greatly improved performance compared to plain + * av_image_copy_plane(). + * + * bytewidth must be contained by both absolute values of dst_linesize + * and src_linesize, otherwise the function behavior is undefined. + * + * @note The linesize parameters have the type ptrdiff_t here, while they are + * int for av_image_copy_plane(). + * @note On x86, the linesizes currently need to be aligned to the cacheline + * size (i.e. 64) to get improved performance. + */ +void av_image_copy_plane_uc_from(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height); + +/** + * Copy image in src_data to dst_data. + * + * @param dst_data destination image data buffer to copy to + * @param dst_linesizes linesizes for the image in dst_data + * @param src_data source image data buffer to copy from + * @param src_linesizes linesizes for the image in src_data + * @param pix_fmt the AVPixelFormat of the image + * @param width width of the image in pixels + * @param height height of the image in pixels + */ +void av_image_copy(uint8_t * const dst_data[4], const int dst_linesizes[4], + const uint8_t * const src_data[4], const int src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height); + +/** + * Wrapper around av_image_copy() to workaround the limitation + * that the conversion from uint8_t * const * to const uint8_t * const * + * is not performed automatically in C. + * @see av_image_copy() + */ +static inline +void av_image_copy2(uint8_t * const dst_data[4], const int dst_linesizes[4], + uint8_t * const src_data[4], const int src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height) +{ + av_image_copy(dst_data, dst_linesizes, + (const uint8_t * const *)src_data, src_linesizes, + pix_fmt, width, height); +} + +/** + * Copy image data located in uncacheable (e.g. GPU mapped) memory. Where + * available, this function will use special functionality for reading from such + * memory, which may result in greatly improved performance compared to plain + * av_image_copy(). + * + * The data pointers and the linesizes must be aligned to the maximum required + * by the CPU architecture. + * + * @note The linesize parameters have the type ptrdiff_t here, while they are + * int for av_image_copy(). + * @note On x86, the linesizes currently need to be aligned to the cacheline + * size (i.e. 64) to get improved performance. + */ +void av_image_copy_uc_from(uint8_t * const dst_data[4], const ptrdiff_t dst_linesizes[4], + const uint8_t * const src_data[4], const ptrdiff_t src_linesizes[4], + enum AVPixelFormat pix_fmt, int width, int height); + +/** + * Setup the data pointers and linesizes based on the specified image + * parameters and the provided array. + * + * The fields of the given image are filled in by using the src + * address which points to the image data buffer. Depending on the + * specified pixel format, one or multiple image data pointers and + * line sizes will be set. If a planar format is specified, several + * pointers will be set pointing to the different picture planes and + * the line sizes of the different planes will be stored in the + * lines_sizes array. Call with src == NULL to get the required + * size for the src buffer. + * + * To allocate the buffer and fill in the dst_data and dst_linesize in + * one call, use av_image_alloc(). + * + * @param dst_data data pointers to be filled in + * @param dst_linesize linesizes for the image in dst_data to be filled in + * @param src buffer which will contain or contains the actual image data, can be NULL + * @param pix_fmt the pixel format of the image + * @param width the width of the image in pixels + * @param height the height of the image in pixels + * @param align the value used in src for linesize alignment + * @return the size in bytes required for src, a negative error code + * in case of failure + */ +int av_image_fill_arrays(uint8_t *dst_data[4], int dst_linesize[4], + const uint8_t *src, + enum AVPixelFormat pix_fmt, int width, int height, int align); + +/** + * Return the size in bytes of the amount of data required to store an + * image with the given parameters. + * + * @param pix_fmt the pixel format of the image + * @param width the width of the image in pixels + * @param height the height of the image in pixels + * @param align the assumed linesize alignment + * @return the buffer size in bytes, a negative error code in case of failure + */ +int av_image_get_buffer_size(enum AVPixelFormat pix_fmt, int width, int height, int align); + +/** + * Copy image data from an image into a buffer. + * + * av_image_get_buffer_size() can be used to compute the required size + * for the buffer to fill. + * + * @param dst a buffer into which picture data will be copied + * @param dst_size the size in bytes of dst + * @param src_data pointers containing the source image data + * @param src_linesize linesizes for the image in src_data + * @param pix_fmt the pixel format of the source image + * @param width the width of the source image in pixels + * @param height the height of the source image in pixels + * @param align the assumed linesize alignment for dst + * @return the number of bytes written to dst, or a negative value + * (error code) on error + */ +int av_image_copy_to_buffer(uint8_t *dst, int dst_size, + const uint8_t * const src_data[4], const int src_linesize[4], + enum AVPixelFormat pix_fmt, int width, int height, int align); + +/** + * Check if the given dimension of an image is valid, meaning that all + * bytes of the image can be addressed with a signed int. + * + * @param w the width of the picture + * @param h the height of the picture + * @param log_offset the offset to sum to the log level for logging with log_ctx + * @param log_ctx the parent logging context, it may be NULL + * @return >= 0 if valid, a negative error code otherwise + */ +int av_image_check_size(unsigned int w, unsigned int h, int log_offset, void *log_ctx); + +/** + * Check if the given dimension of an image is valid, meaning that all + * bytes of a plane of an image with the specified pix_fmt can be addressed + * with a signed int. + * + * @param w the width of the picture + * @param h the height of the picture + * @param max_pixels the maximum number of pixels the user wants to accept + * @param pix_fmt the pixel format, can be AV_PIX_FMT_NONE if unknown. + * @param log_offset the offset to sum to the log level for logging with log_ctx + * @param log_ctx the parent logging context, it may be NULL + * @return >= 0 if valid, a negative error code otherwise + */ +int av_image_check_size2(unsigned int w, unsigned int h, int64_t max_pixels, enum AVPixelFormat pix_fmt, int log_offset, void *log_ctx); + +/** + * Check if the given sample aspect ratio of an image is valid. + * + * It is considered invalid if the denominator is 0 or if applying the ratio + * to the image size would make the smaller dimension less than 1. If the + * sar numerator is 0, it is considered unknown and will return as valid. + * + * @param w width of the image + * @param h height of the image + * @param sar sample aspect ratio of the image + * @return 0 if valid, a negative AVERROR code otherwise + */ +int av_image_check_sar(unsigned int w, unsigned int h, AVRational sar); + +/** + * Overwrite the image data with black. This is suitable for filling a + * sub-rectangle of an image, meaning the padding between the right most pixel + * and the left most pixel on the next line will not be overwritten. For some + * formats, the image size might be rounded up due to inherent alignment. + * + * If the pixel format has alpha, the alpha is cleared to opaque. + * + * This can return an error if the pixel format is not supported. Normally, all + * non-hwaccel pixel formats should be supported. + * + * Passing NULL for dst_data is allowed. Then the function returns whether the + * operation would have succeeded. (It can return an error if the pix_fmt is + * not supported.) + * + * @param dst_data data pointers to destination image + * @param dst_linesize linesizes for the destination image + * @param pix_fmt the pixel format of the image + * @param range the color range of the image (important for colorspaces such as YUV) + * @param width the width of the image in pixels + * @param height the height of the image in pixels + * @return 0 if the image data was cleared, a negative AVERROR code otherwise + */ +int av_image_fill_black(uint8_t * const dst_data[4], const ptrdiff_t dst_linesize[4], + enum AVPixelFormat pix_fmt, enum AVColorRange range, + int width, int height); + +/** + * Overwrite the image data with a color. This is suitable for filling a + * sub-rectangle of an image, meaning the padding between the right most pixel + * and the left most pixel on the next line will not be overwritten. For some + * formats, the image size might be rounded up due to inherent alignment. + * + * If the pixel format has alpha, it is also replaced. Color component values + * are interpreted as native integers (or intfloats) regardless of actual pixel + * format endianness. + * + * This can return an error if the pixel format is not supported. Normally, all + * non-hwaccel pixel formats should be supported. + * + * Passing NULL for dst_data is allowed. Then the function returns whether the + * operation would have succeeded. (It can return an error if the pix_fmt is + * not supported.) + * + * @param dst_data data pointers to destination image + * @param dst_linesize linesizes for the destination image + * @param pix_fmt the pixel format of the image + * @param color the color components to be used for the fill + * @param width the width of the image in pixels + * @param height the height of the image in pixels + * @param flags currently unused + * @return 0 if the image data was filled, a negative AVERROR code otherwise + */ +int av_image_fill_color(uint8_t * const dst_data[4], const ptrdiff_t dst_linesize[4], + enum AVPixelFormat pix_fmt, const uint32_t color[4], + int width, int height, int flags); + +/** + * @} + */ + + +#endif /* AVUTIL_IMGUTILS_H */ diff --git a/media/ffvpx/libavutil/imgutils_internal.h b/media/ffvpx/libavutil/imgutils_internal.h new file mode 100644 index 0000000000..d515858413 --- /dev/null +++ b/media/ffvpx/libavutil/imgutils_internal.h @@ -0,0 +1,30 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_IMGUTILS_INTERNAL_H +#define AVUTIL_IMGUTILS_INTERNAL_H + +#include <stddef.h> +#include <stdint.h> + +int ff_image_copy_plane_uc_from_x86(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height); + + +#endif /* AVUTIL_IMGUTILS_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/integer.c b/media/ffvpx/libavutil/integer.c new file mode 100644 index 0000000000..ae87c467b2 --- /dev/null +++ b/media/ffvpx/libavutil/integer.c @@ -0,0 +1,166 @@ +/* + * arbitrary precision integers + * Copyright (c) 2004 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * arbitrary precision integers + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#include <string.h> + +#include "integer.h" +#include "avassert.h" +#include "intmath.h" + +static const AVInteger zero_i; + +AVInteger av_add_i(AVInteger a, AVInteger b){ + int i, carry=0; + + for(i=0; i<AV_INTEGER_SIZE; i++){ + carry= (carry>>16) + a.v[i] + b.v[i]; + a.v[i]= carry; + } + return a; +} + +AVInteger av_sub_i(AVInteger a, AVInteger b){ + int i, carry=0; + + for(i=0; i<AV_INTEGER_SIZE; i++){ + carry= (carry>>16) + a.v[i] - b.v[i]; + a.v[i]= carry; + } + return a; +} + +int av_log2_i(AVInteger a){ + int i; + + for(i=AV_INTEGER_SIZE-1; i>=0; i--){ + if(a.v[i]) + return av_log2_16bit(a.v[i]) + 16*i; + } + return -1; +} + +AVInteger av_mul_i(AVInteger a, AVInteger b){ + AVInteger out; + int i, j; + int na= (av_log2_i(a)+16) >> 4; + int nb= (av_log2_i(b)+16) >> 4; + + memset(&out, 0, sizeof(out)); + + for(i=0; i<na; i++){ + unsigned int carry=0; + + if(a.v[i]) + for(j=i; j<AV_INTEGER_SIZE && j-i<=nb; j++){ + carry= (carry>>16) + out.v[j] + a.v[i]*(unsigned)b.v[j-i]; + out.v[j]= carry; + } + } + + return out; +} + +int av_cmp_i(AVInteger a, AVInteger b){ + int i; + int v= (int16_t)a.v[AV_INTEGER_SIZE-1] - (int16_t)b.v[AV_INTEGER_SIZE-1]; + if(v) return (v>>16)|1; + + for(i=AV_INTEGER_SIZE-2; i>=0; i--){ + int v= a.v[i] - b.v[i]; + if(v) return (v>>16)|1; + } + return 0; +} + +AVInteger av_shr_i(AVInteger a, int s){ + AVInteger out; + int i; + + for(i=0; i<AV_INTEGER_SIZE; i++){ + unsigned int index= i + (s>>4); + unsigned int v=0; + if (index + 1 < AV_INTEGER_SIZE) v = a.v[index + 1] * (1U << 16); + if (index < AV_INTEGER_SIZE) v |= a.v[index]; + out.v[i]= v >> (s&15); + } + return out; +} + +AVInteger av_mod_i(AVInteger *quot, AVInteger a, AVInteger b){ + int i= av_log2_i(a) - av_log2_i(b); + AVInteger quot_temp; + if(!quot) quot = "_temp; + + if ((int16_t)a.v[AV_INTEGER_SIZE-1] < 0) { + a = av_mod_i(quot, av_sub_i(zero_i, a), b); + *quot = av_sub_i(zero_i, *quot); + return av_sub_i(zero_i, a); + } + + av_assert2((int16_t)a.v[AV_INTEGER_SIZE-1] >= 0 && (int16_t)b.v[AV_INTEGER_SIZE-1] >= 0); + av_assert2(av_log2_i(b)>=0); + + if(i > 0) + b= av_shr_i(b, -i); + + memset(quot, 0, sizeof(AVInteger)); + + while(i-- >= 0){ + *quot= av_shr_i(*quot, -1); + if(av_cmp_i(a, b) >= 0){ + a= av_sub_i(a, b); + quot->v[0] += 1; + } + b= av_shr_i(b, 1); + } + return a; +} + +AVInteger av_div_i(AVInteger a, AVInteger b){ + AVInteger quot; + av_mod_i(", a, b); + return quot; +} + +AVInteger av_int2i(int64_t a){ + AVInteger out; + int i; + + for(i=0; i<AV_INTEGER_SIZE; i++){ + out.v[i]= a; + a>>=16; + } + return out; +} + +int64_t av_i2int(AVInteger a){ + uint64_t out = a.v[3]; + + for (int i = 2; i >= 0; i--) + out = (out << 16) | a.v[i]; + return out; +} diff --git a/media/ffvpx/libavutil/integer.h b/media/ffvpx/libavutil/integer.h new file mode 100644 index 0000000000..2d9b5bb10f --- /dev/null +++ b/media/ffvpx/libavutil/integer.h @@ -0,0 +1,86 @@ +/* + * arbitrary precision integers + * Copyright (c) 2004 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * arbitrary precision integers + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#ifndef AVUTIL_INTEGER_H +#define AVUTIL_INTEGER_H + +#include <stdint.h> +#include "attributes.h" + +#define AV_INTEGER_SIZE 8 + +typedef struct AVInteger{ + uint16_t v[AV_INTEGER_SIZE]; +} AVInteger; + +AVInteger av_add_i(AVInteger a, AVInteger b) av_const; +AVInteger av_sub_i(AVInteger a, AVInteger b) av_const; + +/** + * Return the rounded-down value of the base 2 logarithm of the given + * AVInteger. This is simply the index of the most significant bit + * which is 1, or 0 if all bits are 0. + */ +int av_log2_i(AVInteger a) av_const; +AVInteger av_mul_i(AVInteger a, AVInteger b) av_const; + +/** + * Return 0 if a==b, 1 if a>b and -1 if a<b. + */ +int av_cmp_i(AVInteger a, AVInteger b) av_const; + +/** + * bitwise shift + * @param s the number of bits by which the value should be shifted right, + may be negative for shifting left + */ +AVInteger av_shr_i(AVInteger a, int s) av_const; + +/** + * Return a % b. + * @param quot a/b will be stored here. + */ +AVInteger av_mod_i(AVInteger *quot, AVInteger a, AVInteger b); + +/** + * Return a/b. + */ +AVInteger av_div_i(AVInteger a, AVInteger b) av_const; + +/** + * Convert the given int64_t to an AVInteger. + */ +AVInteger av_int2i(int64_t a) av_const; + +/** + * Convert the given AVInteger to an int64_t. + * If the AVInteger is too large to fit into an int64_t, + * then only the least significant 64 bits will be used. + */ +int64_t av_i2int(AVInteger a) av_const; + +#endif /* AVUTIL_INTEGER_H */ diff --git a/media/ffvpx/libavutil/internal.h b/media/ffvpx/libavutil/internal.h new file mode 100644 index 0000000000..461c0df9ad --- /dev/null +++ b/media/ffvpx/libavutil/internal.h @@ -0,0 +1,181 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * common internal API header + */ + +#ifndef AVUTIL_INTERNAL_H +#define AVUTIL_INTERNAL_H + +#if !defined(DEBUG) && !defined(NDEBUG) +# define NDEBUG +#endif + +// This can be enabled to allow detection of additional integer overflows with ubsan +//#define CHECKED + +#include <limits.h> +#include <stdint.h> +#include <stddef.h> +#include <assert.h> +#include <stdio.h> +#include "config.h" +#include "attributes.h" +#include "macros.h" +#include "pixfmt.h" + +#ifndef attribute_align_arg +#if ARCH_X86_32 && AV_GCC_VERSION_AT_LEAST(4,2) +# define attribute_align_arg __attribute__((force_align_arg_pointer)) +#else +# define attribute_align_arg +#endif +#endif + +#if defined(_WIN32) && CONFIG_SHARED && !defined(BUILDING_avutil) +# define av_export_avutil __declspec(dllimport) +#else +# define av_export_avutil +#endif + +#if HAVE_PRAGMA_DEPRECATED +# if defined(__ICL) || defined (__INTEL_COMPILER) +# define FF_DISABLE_DEPRECATION_WARNINGS __pragma(warning(push)) __pragma(warning(disable:1478)) +# define FF_ENABLE_DEPRECATION_WARNINGS __pragma(warning(pop)) +# elif defined(_MSC_VER) +# define FF_DISABLE_DEPRECATION_WARNINGS __pragma(warning(push)) __pragma(warning(disable:4996)) +# define FF_ENABLE_DEPRECATION_WARNINGS __pragma(warning(pop)) +# else +# define FF_DISABLE_DEPRECATION_WARNINGS _Pragma("GCC diagnostic push") _Pragma("GCC diagnostic ignored \"-Wdeprecated-declarations\"") +# define FF_ENABLE_DEPRECATION_WARNINGS _Pragma("GCC diagnostic pop") +# endif +#else +# define FF_DISABLE_DEPRECATION_WARNINGS +# define FF_ENABLE_DEPRECATION_WARNINGS +#endif + + +#define FF_MEMORY_POISON 0x2a + +/* Check if the hard coded offset of a struct member still matches reality. + * Induce a compilation failure if not. + */ +#define AV_CHECK_OFFSET(s, m, o) struct check_##o { \ + int x_##o[offsetof(s, m) == o? 1: -1]; \ + } + + +#define FF_ALLOC_TYPED_ARRAY(p, nelem) (p = av_malloc_array(nelem, sizeof(*p))) +#define FF_ALLOCZ_TYPED_ARRAY(p, nelem) (p = av_calloc(nelem, sizeof(*p))) + +#define FF_PTR_ADD(ptr, off) ((off) ? (ptr) + (off) : (ptr)) + +/** + * Access a field in a structure by its offset. + */ +#define FF_FIELD_AT(type, off, obj) (*(type *)((char *)&(obj) + (off))) + +#include "libm.h" + +/** + * Return NULL if CONFIG_SMALL is true, otherwise the argument + * without modification. Used to disable the definition of strings. + */ +#if CONFIG_SMALL +# define NULL_IF_CONFIG_SMALL(x) NULL +#else +# define NULL_IF_CONFIG_SMALL(x) x +#endif + +/** + * Log a generic warning message about a missing feature. + * + * @param[in] avc a pointer to an arbitrary struct of which the first + * field is a pointer to an AVClass struct + * @param[in] msg string containing the name of the missing feature + */ +void avpriv_report_missing_feature(void *avc, + const char *msg, ...) av_printf_format(2, 3); + +/** + * Log a generic warning message about a missing feature. + * Additionally request that a sample showcasing the feature be uploaded. + * + * @param[in] avc a pointer to an arbitrary struct of which the first field is + * a pointer to an AVClass struct + * @param[in] msg string containing the name of the missing feature + */ +void avpriv_request_sample(void *avc, + const char *msg, ...) av_printf_format(2, 3); + +#if HAVE_LIBC_MSVCRT +#include <crtversion.h> +#if defined(_VC_CRT_MAJOR_VERSION) && _VC_CRT_MAJOR_VERSION < 14 +#pragma comment(linker, "/include:" EXTERN_PREFIX "avpriv_strtod") +#pragma comment(linker, "/include:" EXTERN_PREFIX "avpriv_snprintf") +#endif + +#define PTRDIFF_SPECIFIER "Id" +#define SIZE_SPECIFIER "Iu" +#else +#define PTRDIFF_SPECIFIER "td" +#define SIZE_SPECIFIER "zu" +#endif + +#ifdef DEBUG +# define ff_dlog(ctx, ...) av_log(ctx, AV_LOG_DEBUG, __VA_ARGS__) +#else +# define ff_dlog(ctx, ...) do { if (0) av_log(ctx, AV_LOG_DEBUG, __VA_ARGS__); } while (0) +#endif + +#ifdef TRACE +# define ff_tlog(ctx, ...) av_log(ctx, AV_LOG_TRACE, __VA_ARGS__) +#else +# define ff_tlog(ctx, ...) do { } while(0) +#endif + +// For debuging we use signed operations so overflows can be detected (by ubsan) +// For production we use unsigned so there are no undefined operations +#ifdef CHECKED +#define SUINT int +#define SUINT32 int32_t +#else +#define SUINT unsigned +#define SUINT32 uint32_t +#endif + +int avpriv_set_systematic_pal2(uint32_t pal[256], enum AVPixelFormat pix_fmt); + +static av_always_inline av_const int avpriv_mirror(int x, int w) +{ + if (!w) + return 0; + + while ((unsigned)x > (unsigned)w) { + x = -x; + if (x < 0) + x += 2 * w; + } + return x; +} + +#endif /* AVUTIL_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/intfloat.h b/media/ffvpx/libavutil/intfloat.h new file mode 100644 index 0000000000..fe3d7ec4a5 --- /dev/null +++ b/media/ffvpx/libavutil/intfloat.h @@ -0,0 +1,77 @@ +/* + * Copyright (c) 2011 Mans Rullgard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_INTFLOAT_H +#define AVUTIL_INTFLOAT_H + +#include <stdint.h> +#include "attributes.h" + +union av_intfloat32 { + uint32_t i; + float f; +}; + +union av_intfloat64 { + uint64_t i; + double f; +}; + +/** + * Reinterpret a 32-bit integer as a float. + */ +static av_always_inline float av_int2float(uint32_t i) +{ + union av_intfloat32 v; + v.i = i; + return v.f; +} + +/** + * Reinterpret a float as a 32-bit integer. + */ +static av_always_inline uint32_t av_float2int(float f) +{ + union av_intfloat32 v; + v.f = f; + return v.i; +} + +/** + * Reinterpret a 64-bit integer as a double. + */ +static av_always_inline double av_int2double(uint64_t i) +{ + union av_intfloat64 v; + v.i = i; + return v.f; +} + +/** + * Reinterpret a double as a 64-bit integer. + */ +static av_always_inline uint64_t av_double2int(double f) +{ + union av_intfloat64 v; + v.f = f; + return v.i; +} + +#endif /* AVUTIL_INTFLOAT_H */ diff --git a/media/ffvpx/libavutil/intmath.c b/media/ffvpx/libavutil/intmath.c new file mode 100644 index 0000000000..b0c00e1cad --- /dev/null +++ b/media/ffvpx/libavutil/intmath.c @@ -0,0 +1,34 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "intmath.h" + +/* undef these to get the function prototypes from common.h */ +#undef av_log2 +#undef av_log2_16bit +#include "common.h" + +int av_log2(unsigned v) +{ + return ff_log2(v); +} + +int av_log2_16bit(unsigned v) +{ + return ff_log2_16bit(v); +} diff --git a/media/ffvpx/libavutil/intmath.h b/media/ffvpx/libavutil/intmath.h new file mode 100644 index 0000000000..c54d23b7bf --- /dev/null +++ b/media/ffvpx/libavutil/intmath.h @@ -0,0 +1,166 @@ +/* + * Copyright (c) 2010 Mans Rullgard <mans@mansr.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_INTMATH_H +#define AVUTIL_INTMATH_H + +#include <stdint.h> + +#include "config.h" +#include "attributes.h" + +#if ARCH_ARM +# include "arm/intmath.h" +#elif ARCH_RISCV +# include "riscv/intmath.h" +#elif ARCH_X86 +# include "x86/intmath.h" +#endif + +#if HAVE_FAST_CLZ +#if AV_GCC_VERSION_AT_LEAST(3,4) +#ifndef ff_log2 +# define ff_log2(x) (31 - __builtin_clz((x)|1)) +# ifndef ff_log2_16bit +# define ff_log2_16bit av_log2 +# endif +#endif /* ff_log2 */ +#endif /* AV_GCC_VERSION_AT_LEAST(3,4) */ +#endif + +extern const uint8_t ff_log2_tab[256]; + +#ifndef ff_log2 +#define ff_log2 ff_log2_c +static av_always_inline av_const int ff_log2_c(unsigned int v) +{ + int n = 0; + if (v & 0xffff0000) { + v >>= 16; + n += 16; + } + if (v & 0xff00) { + v >>= 8; + n += 8; + } + n += ff_log2_tab[v]; + + return n; +} +#endif + +#ifndef ff_log2_16bit +#define ff_log2_16bit ff_log2_16bit_c +static av_always_inline av_const int ff_log2_16bit_c(unsigned int v) +{ + int n = 0; + if (v & 0xff00) { + v >>= 8; + n += 8; + } + n += ff_log2_tab[v]; + + return n; +} +#endif + +#define av_log2 ff_log2 +#define av_log2_16bit ff_log2_16bit + +/** + * @addtogroup lavu_math + * @{ + */ + +#if HAVE_FAST_CLZ +#if AV_GCC_VERSION_AT_LEAST(3,4) +#ifndef ff_ctz +#define ff_ctz(v) __builtin_ctz(v) +#endif +#ifndef ff_ctzll +#define ff_ctzll(v) __builtin_ctzll(v) +#endif +#ifndef ff_clz +#define ff_clz(v) __builtin_clz(v) +#endif +#endif +#endif + +#ifndef ff_ctz +#define ff_ctz ff_ctz_c +/** + * Trailing zero bit count. + * + * @param v input value. If v is 0, the result is undefined. + * @return the number of trailing 0-bits + */ +/* We use the De-Bruijn method outlined in: + * http://supertech.csail.mit.edu/papers/debruijn.pdf. */ +static av_always_inline av_const int ff_ctz_c(int v) +{ + static const uint8_t debruijn_ctz32[32] = { + 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8, + 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9 + }; + return debruijn_ctz32[(uint32_t)((v & -v) * 0x077CB531U) >> 27]; +} +#endif + +#ifndef ff_ctzll +#define ff_ctzll ff_ctzll_c +/* We use the De-Bruijn method outlined in: + * http://supertech.csail.mit.edu/papers/debruijn.pdf. */ +static av_always_inline av_const int ff_ctzll_c(long long v) +{ + static const uint8_t debruijn_ctz64[64] = { + 0, 1, 2, 53, 3, 7, 54, 27, 4, 38, 41, 8, 34, 55, 48, 28, + 62, 5, 39, 46, 44, 42, 22, 9, 24, 35, 59, 56, 49, 18, 29, 11, + 63, 52, 6, 26, 37, 40, 33, 47, 61, 45, 43, 21, 23, 58, 17, 10, + 51, 25, 36, 32, 60, 20, 57, 16, 50, 31, 19, 15, 30, 14, 13, 12 + }; + return debruijn_ctz64[(uint64_t)((v & -v) * 0x022FDD63CC95386DU) >> 58]; +} +#endif + +#ifndef ff_clz +#define ff_clz ff_clz_c +static av_always_inline av_const unsigned ff_clz_c(unsigned x) +{ + unsigned i = sizeof(x) * 8; + + while (x) { + x >>= 1; + i--; + } + + return i; +} +#endif + +#if AV_GCC_VERSION_AT_LEAST(3,4) +#ifndef av_parity +#define av_parity __builtin_parity +#endif +#endif + +/** + * @} + */ +#endif /* AVUTIL_INTMATH_H */ diff --git a/media/ffvpx/libavutil/intreadwrite.h b/media/ffvpx/libavutil/intreadwrite.h new file mode 100644 index 0000000000..21df7887f3 --- /dev/null +++ b/media/ffvpx/libavutil/intreadwrite.h @@ -0,0 +1,642 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_INTREADWRITE_H +#define AVUTIL_INTREADWRITE_H + +#include <stdint.h> +#include "libavutil/avconfig.h" +#include "attributes.h" +#include "bswap.h" + +typedef union { + uint64_t u64; + uint32_t u32[2]; + uint16_t u16[4]; + uint8_t u8 [8]; + double f64; + float f32[2]; +} av_alias av_alias64; + +typedef union { + uint32_t u32; + uint16_t u16[2]; + uint8_t u8 [4]; + float f32; +} av_alias av_alias32; + +typedef union { + uint16_t u16; + uint8_t u8 [2]; +} av_alias av_alias16; + +/* + * Arch-specific headers can provide any combination of + * AV_[RW][BLN](16|24|32|48|64) and AV_(COPY|SWAP|ZERO)(64|128) macros. + * Preprocessor symbols must be defined, even if these are implemented + * as inline functions. + * + * R/W means read/write, B/L/N means big/little/native endianness. + * The following macros require aligned access, compared to their + * unaligned variants: AV_(COPY|SWAP|ZERO)(64|128), AV_[RW]N[8-64]A. + * Incorrect usage may range from abysmal performance to crash + * depending on the platform. + * + * The unaligned variants are AV_[RW][BLN][8-64] and AV_COPY*U. + */ + +#ifdef HAVE_AV_CONFIG_H + +#include "config.h" + +#if ARCH_ARM +# include "arm/intreadwrite.h" +#elif ARCH_AVR32 +# include "avr32/intreadwrite.h" +#elif ARCH_MIPS +# include "mips/intreadwrite.h" +#elif ARCH_PPC +# include "ppc/intreadwrite.h" +#elif ARCH_X86 +# include "x86/intreadwrite.h" +#endif + +#endif /* HAVE_AV_CONFIG_H */ + +/* + * Map AV_RNXX <-> AV_R[BL]XX for all variants provided by per-arch headers. + */ + +#if AV_HAVE_BIGENDIAN + +# if defined(AV_RN16) && !defined(AV_RB16) +# define AV_RB16(p) AV_RN16(p) +# elif !defined(AV_RN16) && defined(AV_RB16) +# define AV_RN16(p) AV_RB16(p) +# endif + +# if defined(AV_WN16) && !defined(AV_WB16) +# define AV_WB16(p, v) AV_WN16(p, v) +# elif !defined(AV_WN16) && defined(AV_WB16) +# define AV_WN16(p, v) AV_WB16(p, v) +# endif + +# if defined(AV_RN24) && !defined(AV_RB24) +# define AV_RB24(p) AV_RN24(p) +# elif !defined(AV_RN24) && defined(AV_RB24) +# define AV_RN24(p) AV_RB24(p) +# endif + +# if defined(AV_WN24) && !defined(AV_WB24) +# define AV_WB24(p, v) AV_WN24(p, v) +# elif !defined(AV_WN24) && defined(AV_WB24) +# define AV_WN24(p, v) AV_WB24(p, v) +# endif + +# if defined(AV_RN32) && !defined(AV_RB32) +# define AV_RB32(p) AV_RN32(p) +# elif !defined(AV_RN32) && defined(AV_RB32) +# define AV_RN32(p) AV_RB32(p) +# endif + +# if defined(AV_WN32) && !defined(AV_WB32) +# define AV_WB32(p, v) AV_WN32(p, v) +# elif !defined(AV_WN32) && defined(AV_WB32) +# define AV_WN32(p, v) AV_WB32(p, v) +# endif + +# if defined(AV_RN48) && !defined(AV_RB48) +# define AV_RB48(p) AV_RN48(p) +# elif !defined(AV_RN48) && defined(AV_RB48) +# define AV_RN48(p) AV_RB48(p) +# endif + +# if defined(AV_WN48) && !defined(AV_WB48) +# define AV_WB48(p, v) AV_WN48(p, v) +# elif !defined(AV_WN48) && defined(AV_WB48) +# define AV_WN48(p, v) AV_WB48(p, v) +# endif + +# if defined(AV_RN64) && !defined(AV_RB64) +# define AV_RB64(p) AV_RN64(p) +# elif !defined(AV_RN64) && defined(AV_RB64) +# define AV_RN64(p) AV_RB64(p) +# endif + +# if defined(AV_WN64) && !defined(AV_WB64) +# define AV_WB64(p, v) AV_WN64(p, v) +# elif !defined(AV_WN64) && defined(AV_WB64) +# define AV_WN64(p, v) AV_WB64(p, v) +# endif + +#else /* AV_HAVE_BIGENDIAN */ + +# if defined(AV_RN16) && !defined(AV_RL16) +# define AV_RL16(p) AV_RN16(p) +# elif !defined(AV_RN16) && defined(AV_RL16) +# define AV_RN16(p) AV_RL16(p) +# endif + +# if defined(AV_WN16) && !defined(AV_WL16) +# define AV_WL16(p, v) AV_WN16(p, v) +# elif !defined(AV_WN16) && defined(AV_WL16) +# define AV_WN16(p, v) AV_WL16(p, v) +# endif + +# if defined(AV_RN24) && !defined(AV_RL24) +# define AV_RL24(p) AV_RN24(p) +# elif !defined(AV_RN24) && defined(AV_RL24) +# define AV_RN24(p) AV_RL24(p) +# endif + +# if defined(AV_WN24) && !defined(AV_WL24) +# define AV_WL24(p, v) AV_WN24(p, v) +# elif !defined(AV_WN24) && defined(AV_WL24) +# define AV_WN24(p, v) AV_WL24(p, v) +# endif + +# if defined(AV_RN32) && !defined(AV_RL32) +# define AV_RL32(p) AV_RN32(p) +# elif !defined(AV_RN32) && defined(AV_RL32) +# define AV_RN32(p) AV_RL32(p) +# endif + +# if defined(AV_WN32) && !defined(AV_WL32) +# define AV_WL32(p, v) AV_WN32(p, v) +# elif !defined(AV_WN32) && defined(AV_WL32) +# define AV_WN32(p, v) AV_WL32(p, v) +# endif + +# if defined(AV_RN48) && !defined(AV_RL48) +# define AV_RL48(p) AV_RN48(p) +# elif !defined(AV_RN48) && defined(AV_RL48) +# define AV_RN48(p) AV_RL48(p) +# endif + +# if defined(AV_WN48) && !defined(AV_WL48) +# define AV_WL48(p, v) AV_WN48(p, v) +# elif !defined(AV_WN48) && defined(AV_WL48) +# define AV_WN48(p, v) AV_WL48(p, v) +# endif + +# if defined(AV_RN64) && !defined(AV_RL64) +# define AV_RL64(p) AV_RN64(p) +# elif !defined(AV_RN64) && defined(AV_RL64) +# define AV_RN64(p) AV_RL64(p) +# endif + +# if defined(AV_WN64) && !defined(AV_WL64) +# define AV_WL64(p, v) AV_WN64(p, v) +# elif !defined(AV_WN64) && defined(AV_WL64) +# define AV_WN64(p, v) AV_WL64(p, v) +# endif + +#endif /* !AV_HAVE_BIGENDIAN */ + +/* + * Define AV_[RW]N helper macros to simplify definitions not provided + * by per-arch headers. + */ + +#if defined(__GNUC__) || defined(__clang__) + +union unaligned_64 { uint64_t l; } __attribute__((packed)) av_alias; +union unaligned_32 { uint32_t l; } __attribute__((packed)) av_alias; +union unaligned_16 { uint16_t l; } __attribute__((packed)) av_alias; + +# define AV_RN(s, p) (((const union unaligned_##s *) (p))->l) +# define AV_WN(s, p, v) ((((union unaligned_##s *) (p))->l) = (v)) + +#elif defined(_MSC_VER) && (defined(_M_ARM) || defined(_M_X64) || defined(_M_ARM64)) && AV_HAVE_FAST_UNALIGNED + +# define AV_RN(s, p) (*((const __unaligned uint##s##_t*)(p))) +# define AV_WN(s, p, v) (*((__unaligned uint##s##_t*)(p)) = (v)) + +#elif AV_HAVE_FAST_UNALIGNED + +# define AV_RN(s, p) (((const av_alias##s*)(p))->u##s) +# define AV_WN(s, p, v) (((av_alias##s*)(p))->u##s = (v)) + +#else + +#ifndef AV_RB16 +# define AV_RB16(x) \ + ((((const uint8_t*)(x))[0] << 8) | \ + ((const uint8_t*)(x))[1]) +#endif +#ifndef AV_WB16 +# define AV_WB16(p, val) do { \ + uint16_t d = (val); \ + ((uint8_t*)(p))[1] = (d); \ + ((uint8_t*)(p))[0] = (d)>>8; \ + } while(0) +#endif + +#ifndef AV_RL16 +# define AV_RL16(x) \ + ((((const uint8_t*)(x))[1] << 8) | \ + ((const uint8_t*)(x))[0]) +#endif +#ifndef AV_WL16 +# define AV_WL16(p, val) do { \ + uint16_t d = (val); \ + ((uint8_t*)(p))[0] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + } while(0) +#endif + +#ifndef AV_RB32 +# define AV_RB32(x) \ + (((uint32_t)((const uint8_t*)(x))[0] << 24) | \ + (((const uint8_t*)(x))[1] << 16) | \ + (((const uint8_t*)(x))[2] << 8) | \ + ((const uint8_t*)(x))[3]) +#endif +#ifndef AV_WB32 +# define AV_WB32(p, val) do { \ + uint32_t d = (val); \ + ((uint8_t*)(p))[3] = (d); \ + ((uint8_t*)(p))[2] = (d)>>8; \ + ((uint8_t*)(p))[1] = (d)>>16; \ + ((uint8_t*)(p))[0] = (d)>>24; \ + } while(0) +#endif + +#ifndef AV_RL32 +# define AV_RL32(x) \ + (((uint32_t)((const uint8_t*)(x))[3] << 24) | \ + (((const uint8_t*)(x))[2] << 16) | \ + (((const uint8_t*)(x))[1] << 8) | \ + ((const uint8_t*)(x))[0]) +#endif +#ifndef AV_WL32 +# define AV_WL32(p, val) do { \ + uint32_t d = (val); \ + ((uint8_t*)(p))[0] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + ((uint8_t*)(p))[2] = (d)>>16; \ + ((uint8_t*)(p))[3] = (d)>>24; \ + } while(0) +#endif + +#ifndef AV_RB64 +# define AV_RB64(x) \ + (((uint64_t)((const uint8_t*)(x))[0] << 56) | \ + ((uint64_t)((const uint8_t*)(x))[1] << 48) | \ + ((uint64_t)((const uint8_t*)(x))[2] << 40) | \ + ((uint64_t)((const uint8_t*)(x))[3] << 32) | \ + ((uint64_t)((const uint8_t*)(x))[4] << 24) | \ + ((uint64_t)((const uint8_t*)(x))[5] << 16) | \ + ((uint64_t)((const uint8_t*)(x))[6] << 8) | \ + (uint64_t)((const uint8_t*)(x))[7]) +#endif +#ifndef AV_WB64 +# define AV_WB64(p, val) do { \ + uint64_t d = (val); \ + ((uint8_t*)(p))[7] = (d); \ + ((uint8_t*)(p))[6] = (d)>>8; \ + ((uint8_t*)(p))[5] = (d)>>16; \ + ((uint8_t*)(p))[4] = (d)>>24; \ + ((uint8_t*)(p))[3] = (d)>>32; \ + ((uint8_t*)(p))[2] = (d)>>40; \ + ((uint8_t*)(p))[1] = (d)>>48; \ + ((uint8_t*)(p))[0] = (d)>>56; \ + } while(0) +#endif + +#ifndef AV_RL64 +# define AV_RL64(x) \ + (((uint64_t)((const uint8_t*)(x))[7] << 56) | \ + ((uint64_t)((const uint8_t*)(x))[6] << 48) | \ + ((uint64_t)((const uint8_t*)(x))[5] << 40) | \ + ((uint64_t)((const uint8_t*)(x))[4] << 32) | \ + ((uint64_t)((const uint8_t*)(x))[3] << 24) | \ + ((uint64_t)((const uint8_t*)(x))[2] << 16) | \ + ((uint64_t)((const uint8_t*)(x))[1] << 8) | \ + (uint64_t)((const uint8_t*)(x))[0]) +#endif +#ifndef AV_WL64 +# define AV_WL64(p, val) do { \ + uint64_t d = (val); \ + ((uint8_t*)(p))[0] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + ((uint8_t*)(p))[2] = (d)>>16; \ + ((uint8_t*)(p))[3] = (d)>>24; \ + ((uint8_t*)(p))[4] = (d)>>32; \ + ((uint8_t*)(p))[5] = (d)>>40; \ + ((uint8_t*)(p))[6] = (d)>>48; \ + ((uint8_t*)(p))[7] = (d)>>56; \ + } while(0) +#endif + +#if AV_HAVE_BIGENDIAN +# define AV_RN(s, p) AV_RB##s(p) +# define AV_WN(s, p, v) AV_WB##s(p, v) +#else +# define AV_RN(s, p) AV_RL##s(p) +# define AV_WN(s, p, v) AV_WL##s(p, v) +#endif + +#endif /* HAVE_FAST_UNALIGNED */ + +#ifndef AV_RN16 +# define AV_RN16(p) AV_RN(16, p) +#endif + +#ifndef AV_RN32 +# define AV_RN32(p) AV_RN(32, p) +#endif + +#ifndef AV_RN64 +# define AV_RN64(p) AV_RN(64, p) +#endif + +#ifndef AV_WN16 +# define AV_WN16(p, v) AV_WN(16, p, v) +#endif + +#ifndef AV_WN32 +# define AV_WN32(p, v) AV_WN(32, p, v) +#endif + +#ifndef AV_WN64 +# define AV_WN64(p, v) AV_WN(64, p, v) +#endif + +#if AV_HAVE_BIGENDIAN +# define AV_RB(s, p) AV_RN##s(p) +# define AV_WB(s, p, v) AV_WN##s(p, v) +# define AV_RL(s, p) av_bswap##s(AV_RN##s(p)) +# define AV_WL(s, p, v) AV_WN##s(p, av_bswap##s(v)) +#else +# define AV_RB(s, p) av_bswap##s(AV_RN##s(p)) +# define AV_WB(s, p, v) AV_WN##s(p, av_bswap##s(v)) +# define AV_RL(s, p) AV_RN##s(p) +# define AV_WL(s, p, v) AV_WN##s(p, v) +#endif + +#define AV_RB8(x) (((const uint8_t*)(x))[0]) +#define AV_WB8(p, d) do { ((uint8_t*)(p))[0] = (d); } while(0) + +#define AV_RL8(x) AV_RB8(x) +#define AV_WL8(p, d) AV_WB8(p, d) + +#ifndef AV_RB16 +# define AV_RB16(p) AV_RB(16, p) +#endif +#ifndef AV_WB16 +# define AV_WB16(p, v) AV_WB(16, p, v) +#endif + +#ifndef AV_RL16 +# define AV_RL16(p) AV_RL(16, p) +#endif +#ifndef AV_WL16 +# define AV_WL16(p, v) AV_WL(16, p, v) +#endif + +#ifndef AV_RB32 +# define AV_RB32(p) AV_RB(32, p) +#endif +#ifndef AV_WB32 +# define AV_WB32(p, v) AV_WB(32, p, v) +#endif + +#ifndef AV_RL32 +# define AV_RL32(p) AV_RL(32, p) +#endif +#ifndef AV_WL32 +# define AV_WL32(p, v) AV_WL(32, p, v) +#endif + +#ifndef AV_RB64 +# define AV_RB64(p) AV_RB(64, p) +#endif +#ifndef AV_WB64 +# define AV_WB64(p, v) AV_WB(64, p, v) +#endif + +#ifndef AV_RL64 +# define AV_RL64(p) AV_RL(64, p) +#endif +#ifndef AV_WL64 +# define AV_WL64(p, v) AV_WL(64, p, v) +#endif + +#ifndef AV_RB24 +# define AV_RB24(x) \ + ((((const uint8_t*)(x))[0] << 16) | \ + (((const uint8_t*)(x))[1] << 8) | \ + ((const uint8_t*)(x))[2]) +#endif +#ifndef AV_WB24 +# define AV_WB24(p, d) do { \ + ((uint8_t*)(p))[2] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + ((uint8_t*)(p))[0] = (d)>>16; \ + } while(0) +#endif + +#ifndef AV_RL24 +# define AV_RL24(x) \ + ((((const uint8_t*)(x))[2] << 16) | \ + (((const uint8_t*)(x))[1] << 8) | \ + ((const uint8_t*)(x))[0]) +#endif +#ifndef AV_WL24 +# define AV_WL24(p, d) do { \ + ((uint8_t*)(p))[0] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + ((uint8_t*)(p))[2] = (d)>>16; \ + } while(0) +#endif + +#ifndef AV_RB48 +# define AV_RB48(x) \ + (((uint64_t)((const uint8_t*)(x))[0] << 40) | \ + ((uint64_t)((const uint8_t*)(x))[1] << 32) | \ + ((uint64_t)((const uint8_t*)(x))[2] << 24) | \ + ((uint64_t)((const uint8_t*)(x))[3] << 16) | \ + ((uint64_t)((const uint8_t*)(x))[4] << 8) | \ + (uint64_t)((const uint8_t*)(x))[5]) +#endif +#ifndef AV_WB48 +# define AV_WB48(p, darg) do { \ + uint64_t d = (darg); \ + ((uint8_t*)(p))[5] = (d); \ + ((uint8_t*)(p))[4] = (d)>>8; \ + ((uint8_t*)(p))[3] = (d)>>16; \ + ((uint8_t*)(p))[2] = (d)>>24; \ + ((uint8_t*)(p))[1] = (d)>>32; \ + ((uint8_t*)(p))[0] = (d)>>40; \ + } while(0) +#endif + +#ifndef AV_RL48 +# define AV_RL48(x) \ + (((uint64_t)((const uint8_t*)(x))[5] << 40) | \ + ((uint64_t)((const uint8_t*)(x))[4] << 32) | \ + ((uint64_t)((const uint8_t*)(x))[3] << 24) | \ + ((uint64_t)((const uint8_t*)(x))[2] << 16) | \ + ((uint64_t)((const uint8_t*)(x))[1] << 8) | \ + (uint64_t)((const uint8_t*)(x))[0]) +#endif +#ifndef AV_WL48 +# define AV_WL48(p, darg) do { \ + uint64_t d = (darg); \ + ((uint8_t*)(p))[0] = (d); \ + ((uint8_t*)(p))[1] = (d)>>8; \ + ((uint8_t*)(p))[2] = (d)>>16; \ + ((uint8_t*)(p))[3] = (d)>>24; \ + ((uint8_t*)(p))[4] = (d)>>32; \ + ((uint8_t*)(p))[5] = (d)>>40; \ + } while(0) +#endif + +/* + * The AV_[RW]NA macros access naturally aligned data + * in a type-safe way. + */ + +#define AV_RNA(s, p) (((const av_alias##s*)(p))->u##s) +#define AV_WNA(s, p, v) (((av_alias##s*)(p))->u##s = (v)) + +#ifndef AV_RN16A +# define AV_RN16A(p) AV_RNA(16, p) +#endif + +#ifndef AV_RN32A +# define AV_RN32A(p) AV_RNA(32, p) +#endif + +#ifndef AV_RN64A +# define AV_RN64A(p) AV_RNA(64, p) +#endif + +#ifndef AV_WN16A +# define AV_WN16A(p, v) AV_WNA(16, p, v) +#endif + +#ifndef AV_WN32A +# define AV_WN32A(p, v) AV_WNA(32, p, v) +#endif + +#ifndef AV_WN64A +# define AV_WN64A(p, v) AV_WNA(64, p, v) +#endif + +#if AV_HAVE_BIGENDIAN +# define AV_RLA(s, p) av_bswap##s(AV_RN##s##A(p)) +# define AV_WLA(s, p, v) AV_WN##s##A(p, av_bswap##s(v)) +#else +# define AV_RLA(s, p) AV_RN##s##A(p) +# define AV_WLA(s, p, v) AV_WN##s##A(p, v) +#endif + +#ifndef AV_RL64A +# define AV_RL64A(p) AV_RLA(64, p) +#endif +#ifndef AV_WL64A +# define AV_WL64A(p, v) AV_WLA(64, p, v) +#endif + +/* + * The AV_COPYxxU macros are suitable for copying data to/from unaligned + * memory locations. + */ + +#define AV_COPYU(n, d, s) AV_WN##n(d, AV_RN##n(s)); + +#ifndef AV_COPY16U +# define AV_COPY16U(d, s) AV_COPYU(16, d, s) +#endif + +#ifndef AV_COPY32U +# define AV_COPY32U(d, s) AV_COPYU(32, d, s) +#endif + +#ifndef AV_COPY64U +# define AV_COPY64U(d, s) AV_COPYU(64, d, s) +#endif + +#ifndef AV_COPY128U +# define AV_COPY128U(d, s) \ + do { \ + AV_COPY64U(d, s); \ + AV_COPY64U((char *)(d) + 8, (const char *)(s) + 8); \ + } while(0) +#endif + +/* Parameters for AV_COPY*, AV_SWAP*, AV_ZERO* must be + * naturally aligned. They may be implemented using MMX, + * so emms_c() must be called before using any float code + * afterwards. + */ + +#define AV_COPY(n, d, s) \ + (((av_alias##n*)(d))->u##n = ((const av_alias##n*)(s))->u##n) + +#ifndef AV_COPY16 +# define AV_COPY16(d, s) AV_COPY(16, d, s) +#endif + +#ifndef AV_COPY32 +# define AV_COPY32(d, s) AV_COPY(32, d, s) +#endif + +#ifndef AV_COPY64 +# define AV_COPY64(d, s) AV_COPY(64, d, s) +#endif + +#ifndef AV_COPY128 +# define AV_COPY128(d, s) \ + do { \ + AV_COPY64(d, s); \ + AV_COPY64((char*)(d)+8, (char*)(s)+8); \ + } while(0) +#endif + +#define AV_SWAP(n, a, b) FFSWAP(av_alias##n, *(av_alias##n*)(a), *(av_alias##n*)(b)) + +#ifndef AV_SWAP64 +# define AV_SWAP64(a, b) AV_SWAP(64, a, b) +#endif + +#define AV_ZERO(n, d) (((av_alias##n*)(d))->u##n = 0) + +#ifndef AV_ZERO16 +# define AV_ZERO16(d) AV_ZERO(16, d) +#endif + +#ifndef AV_ZERO32 +# define AV_ZERO32(d) AV_ZERO(32, d) +#endif + +#ifndef AV_ZERO64 +# define AV_ZERO64(d) AV_ZERO(64, d) +#endif + +#ifndef AV_ZERO128 +# define AV_ZERO128(d) \ + do { \ + AV_ZERO64(d); \ + AV_ZERO64((char*)(d)+8); \ + } while(0) +#endif + +#endif /* AVUTIL_INTREADWRITE_H */ diff --git a/media/ffvpx/libavutil/libm.h b/media/ffvpx/libavutil/libm.h new file mode 100644 index 0000000000..a819962391 --- /dev/null +++ b/media/ffvpx/libavutil/libm.h @@ -0,0 +1,471 @@ +/* + * erf function: Copyright (c) 2006 John Maddock + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Replacements for frequently missing libm functions + */ + +#ifndef AVUTIL_LIBM_H +#define AVUTIL_LIBM_H + +#include <math.h> +#include "config.h" +#include "attributes.h" +#include "intfloat.h" +#include "mathematics.h" + +#if HAVE_MIPSFPU && HAVE_INLINE_ASM +#include "libavutil/mips/libm_mips.h" +#endif /* HAVE_MIPSFPU && HAVE_INLINE_ASM*/ + +#if !HAVE_ATANF +#undef atanf +#define atanf(x) ((float)atan(x)) +#endif /* HAVE_ATANF */ + +#if !HAVE_ATAN2F +#undef atan2f +#define atan2f(y, x) ((float)atan2(y, x)) +#endif /* HAVE_ATAN2F */ + +#if !HAVE_POWF +#undef powf +#define powf(x, y) ((float)pow(x, y)) +#endif /* HAVE_POWF */ + +#if !HAVE_CBRT +static av_always_inline double cbrt(double x) +{ + return x < 0 ? -pow(-x, 1.0 / 3.0) : pow(x, 1.0 / 3.0); +} +#endif /* HAVE_CBRT */ + +#if !HAVE_CBRTF +static av_always_inline float cbrtf(float x) +{ + return x < 0 ? -powf(-x, 1.0 / 3.0) : powf(x, 1.0 / 3.0); +} +#endif /* HAVE_CBRTF */ + +#if !HAVE_COPYSIGN +static av_always_inline double copysign(double x, double y) +{ + uint64_t vx = av_double2int(x); + uint64_t vy = av_double2int(y); + return av_int2double((vx & UINT64_C(0x7fffffffffffffff)) | (vy & UINT64_C(0x8000000000000000))); +} +#endif /* HAVE_COPYSIGN */ + +#if !HAVE_COSF +#undef cosf +#define cosf(x) ((float)cos(x)) +#endif /* HAVE_COSF */ + +#if !HAVE_ERF +static inline double ff_eval_poly(const double *coeff, int size, double x) { + double sum = coeff[size-1]; + int i; + for (i = size-2; i >= 0; --i) { + sum *= x; + sum += coeff[i]; + } + return sum; +} + +/** + * erf function + * Algorithm taken from the Boost project, source: + * http://www.boost.org/doc/libs/1_46_1/boost/math/special_functions/erf.hpp + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0 (see notice below). + * Boost Software License - Version 1.0 - August 17th, 2003 +Permission is hereby granted, free of charge, to any person or organization +obtaining a copy of the software and accompanying documentation covered by +this license (the "Software") to use, reproduce, display, distribute, +execute, and transmit the Software, and to prepare derivative works of the +Software, and to permit third-parties to whom the Software is furnished to +do so, all subject to the following: + +The copyright notices in the Software and this entire statement, including +the above license grant, this restriction and the following disclaimer, +must be included in all copies of the Software, in whole or in part, and +all derivative works of the Software, unless such copies or derivative +works are solely in the form of machine-executable object code generated by +a source language processor. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT +SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE +FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, +ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER +DEALINGS IN THE SOFTWARE. + */ +static inline double erf(double z) +{ +#ifndef FF_ARRAY_ELEMS +#define FF_ARRAY_ELEMS(a) (sizeof(a) / sizeof((a)[0])) +#endif + double result; + + /* handle the symmetry: erf(-x) = -erf(x) */ + if (z < 0) + return -erf(-z); + + /* branch based on range of z, and pick appropriate approximation */ + if (z == 0) + return 0; + else if (z < 1e-10) + return z * 1.125 + z * 0.003379167095512573896158903121545171688; + else if (z < 0.5) { + // Maximum Deviation Found: 1.561e-17 + // Expected Error Term: 1.561e-17 + // Maximum Relative Change in Control Points: 1.155e-04 + // Max Error found at double precision = 2.961182e-17 + + static const double y = 1.044948577880859375; + static const double p[] = { + 0.0834305892146531832907, + -0.338165134459360935041, + -0.0509990735146777432841, + -0.00772758345802133288487, + -0.000322780120964605683831, + }; + static const double q[] = { + 1, + 0.455004033050794024546, + 0.0875222600142252549554, + 0.00858571925074406212772, + 0.000370900071787748000569, + }; + double zz = z * z; + return z * (y + ff_eval_poly(p, FF_ARRAY_ELEMS(p), zz) / ff_eval_poly(q, FF_ARRAY_ELEMS(q), zz)); + } + /* here onwards compute erfc */ + else if (z < 1.5) { + // Maximum Deviation Found: 3.702e-17 + // Expected Error Term: 3.702e-17 + // Maximum Relative Change in Control Points: 2.845e-04 + // Max Error found at double precision = 4.841816e-17 + static const double y = 0.405935764312744140625; + static const double p[] = { + -0.098090592216281240205, + 0.178114665841120341155, + 0.191003695796775433986, + 0.0888900368967884466578, + 0.0195049001251218801359, + 0.00180424538297014223957, + }; + static const double q[] = { + 1, + 1.84759070983002217845, + 1.42628004845511324508, + 0.578052804889902404909, + 0.12385097467900864233, + 0.0113385233577001411017, + 0.337511472483094676155e-5, + }; + result = y + ff_eval_poly(p, FF_ARRAY_ELEMS(p), z - 0.5) / ff_eval_poly(q, FF_ARRAY_ELEMS(q), z - 0.5); + result *= exp(-z * z) / z; + return 1 - result; + } + else if (z < 2.5) { + // Max Error found at double precision = 6.599585e-18 + // Maximum Deviation Found: 3.909e-18 + // Expected Error Term: 3.909e-18 + // Maximum Relative Change in Control Points: 9.886e-05 + static const double y = 0.50672817230224609375; + static const double p[] = { + -0.0243500476207698441272, + 0.0386540375035707201728, + 0.04394818964209516296, + 0.0175679436311802092299, + 0.00323962406290842133584, + 0.000235839115596880717416, + }; + static const double q[] = { + 1, + 1.53991494948552447182, + 0.982403709157920235114, + 0.325732924782444448493, + 0.0563921837420478160373, + 0.00410369723978904575884, + }; + result = y + ff_eval_poly(p, FF_ARRAY_ELEMS(p), z - 1.5) / ff_eval_poly(q, FF_ARRAY_ELEMS(q), z - 1.5); + result *= exp(-z * z) / z; + return 1 - result; + } + else if (z < 4.5) { + // Maximum Deviation Found: 1.512e-17 + // Expected Error Term: 1.512e-17 + // Maximum Relative Change in Control Points: 2.222e-04 + // Max Error found at double precision = 2.062515e-17 + static const double y = 0.5405750274658203125; + static const double p[] = { + 0.00295276716530971662634, + 0.0137384425896355332126, + 0.00840807615555585383007, + 0.00212825620914618649141, + 0.000250269961544794627958, + 0.113212406648847561139e-4, + }; + static const double q[] = { + 1, + 1.04217814166938418171, + 0.442597659481563127003, + 0.0958492726301061423444, + 0.0105982906484876531489, + 0.000479411269521714493907, + }; + result = y + ff_eval_poly(p, FF_ARRAY_ELEMS(p), z - 3.5) / ff_eval_poly(q, FF_ARRAY_ELEMS(q), z - 3.5); + result *= exp(-z * z) / z; + return 1 - result; + } + /* differ from Boost here, the claim of underflow of erfc(x) past 5.8 is + * slightly incorrect, change to 5.92 + * (really somewhere between 5.9125 and 5.925 is when it saturates) */ + else if (z < 5.92) { + // Max Error found at double precision = 2.997958e-17 + // Maximum Deviation Found: 2.860e-17 + // Expected Error Term: 2.859e-17 + // Maximum Relative Change in Control Points: 1.357e-05 + static const double y = 0.5579090118408203125; + static const double p[] = { + 0.00628057170626964891937, + 0.0175389834052493308818, + -0.212652252872804219852, + -0.687717681153649930619, + -2.5518551727311523996, + -3.22729451764143718517, + -2.8175401114513378771, + }; + static const double q[] = { + 1, + 2.79257750980575282228, + 11.0567237927800161565, + 15.930646027911794143, + 22.9367376522880577224, + 13.5064170191802889145, + 5.48409182238641741584, + }; + result = y + ff_eval_poly(p, FF_ARRAY_ELEMS(p), 1 / z) / ff_eval_poly(q, FF_ARRAY_ELEMS(q), 1 / z); + result *= exp(-z * z) / z; + return 1 - result; + } + /* handle the nan case, but don't use isnan for max portability */ + else if (z != z) + return z; + /* finally return saturated result */ + else + return 1; +} +#endif /* HAVE_ERF */ + +#if !HAVE_EXPF +#undef expf +#define expf(x) ((float)exp(x)) +#endif /* HAVE_EXPF */ + +#if !HAVE_EXP2 +#undef exp2 +#define exp2(x) exp((x) * M_LN2) +#endif /* HAVE_EXP2 */ + +#if !HAVE_EXP2F +#undef exp2f +#define exp2f(x) ((float)exp2(x)) +#endif /* HAVE_EXP2F */ + +#if !HAVE_ISINF +#undef isinf +/* Note: these do not follow the BSD/Apple/GNU convention of returning -1 for +-Inf, +1 for Inf, 0 otherwise, but merely follow the POSIX/ISO mandated spec of +returning a non-zero value for +/-Inf, 0 otherwise. */ +static av_always_inline av_const int avpriv_isinff(float x) +{ + uint32_t v = av_float2int(x); + if ((v & 0x7f800000) != 0x7f800000) + return 0; + return !(v & 0x007fffff); +} + +static av_always_inline av_const int avpriv_isinf(double x) +{ + uint64_t v = av_double2int(x); + if ((v & 0x7ff0000000000000) != 0x7ff0000000000000) + return 0; + return !(v & 0x000fffffffffffff); +} + +#define isinf(x) \ + (sizeof(x) == sizeof(float) \ + ? avpriv_isinff(x) \ + : avpriv_isinf(x)) +#endif /* HAVE_ISINF */ + +#if !HAVE_ISNAN +static av_always_inline av_const int avpriv_isnanf(float x) +{ + uint32_t v = av_float2int(x); + if ((v & 0x7f800000) != 0x7f800000) + return 0; + return v & 0x007fffff; +} + +static av_always_inline av_const int avpriv_isnan(double x) +{ + uint64_t v = av_double2int(x); + if ((v & 0x7ff0000000000000) != 0x7ff0000000000000) + return 0; + return (v & 0x000fffffffffffff) && 1; +} + +#define isnan(x) \ + (sizeof(x) == sizeof(float) \ + ? avpriv_isnanf(x) \ + : avpriv_isnan(x)) +#endif /* HAVE_ISNAN */ + +#if !HAVE_ISFINITE +static av_always_inline av_const int avpriv_isfinitef(float x) +{ + uint32_t v = av_float2int(x); + return (v & 0x7f800000) != 0x7f800000; +} + +static av_always_inline av_const int avpriv_isfinite(double x) +{ + uint64_t v = av_double2int(x); + return (v & 0x7ff0000000000000) != 0x7ff0000000000000; +} + +#define isfinite(x) \ + (sizeof(x) == sizeof(float) \ + ? avpriv_isfinitef(x) \ + : avpriv_isfinite(x)) +#endif /* HAVE_ISFINITE */ + +#if !HAVE_HYPOT +static inline av_const double hypot(double x, double y) +{ + double ret, temp; + x = fabs(x); + y = fabs(y); + + if (isinf(x) || isinf(y)) + return av_int2double(0x7ff0000000000000); + if (x == 0 || y == 0) + return x + y; + if (x < y) { + temp = x; + x = y; + y = temp; + } + + y = y/x; + return x*sqrt(1 + y*y); +} +#endif /* HAVE_HYPOT */ + +#if !HAVE_LDEXPF +#undef ldexpf +#define ldexpf(x, exp) ((float)ldexp(x, exp)) +#endif /* HAVE_LDEXPF */ + +#if !HAVE_LLRINT +#undef llrint +#define llrint(x) ((long long)rint(x)) +#endif /* HAVE_LLRINT */ + +#if !HAVE_LLRINTF +#undef llrintf +#define llrintf(x) ((long long)rint(x)) +#endif /* HAVE_LLRINT */ + +#if !HAVE_LOG2 +#undef log2 +#define log2(x) (log(x) * 1.44269504088896340736) +#endif /* HAVE_LOG2 */ + +#if !HAVE_LOG2F +#undef log2f +#define log2f(x) ((float)log2(x)) +#endif /* HAVE_LOG2F */ + +#if !HAVE_LOG10F +#undef log10f +#define log10f(x) ((float)log10(x)) +#endif /* HAVE_LOG10F */ + +#if !HAVE_SINF +#undef sinf +#define sinf(x) ((float)sin(x)) +#endif /* HAVE_SINF */ + +#if !HAVE_RINT +static inline double rint(double x) +{ + return x >= 0 ? floor(x + 0.5) : ceil(x - 0.5); +} +#endif /* HAVE_RINT */ + +#if !HAVE_LRINT +static av_always_inline av_const long int lrint(double x) +{ + return rint(x); +} +#endif /* HAVE_LRINT */ + +#if !HAVE_LRINTF +static av_always_inline av_const long int lrintf(float x) +{ + return (int)(rint(x)); +} +#endif /* HAVE_LRINTF */ + +#if !HAVE_ROUND +static av_always_inline av_const double round(double x) +{ + return (x > 0) ? floor(x + 0.5) : ceil(x - 0.5); +} +#endif /* HAVE_ROUND */ + +#if !HAVE_ROUNDF +static av_always_inline av_const float roundf(float x) +{ + return (x > 0) ? floor(x + 0.5) : ceil(x - 0.5); +} +#endif /* HAVE_ROUNDF */ + +#if !HAVE_TRUNC +static av_always_inline av_const double trunc(double x) +{ + return (x > 0) ? floor(x) : ceil(x); +} +#endif /* HAVE_TRUNC */ + +#if !HAVE_TRUNCF +static av_always_inline av_const float truncf(float x) +{ + return (x > 0) ? floor(x) : ceil(x); +} +#endif /* HAVE_TRUNCF */ + +#endif /* AVUTIL_LIBM_H */ diff --git a/media/ffvpx/libavutil/lls.c b/media/ffvpx/libavutil/lls.c new file mode 100644 index 0000000000..c1e038daf1 --- /dev/null +++ b/media/ffvpx/libavutil/lls.c @@ -0,0 +1,123 @@ +/* + * linear least squares model + * + * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * linear least squares model + */ + +#include <math.h> +#include <string.h> + +#include "config.h" +#include "attributes.h" +#include "lls.h" + +static void update_lls(LLSModel *m, const double *var) +{ + int i, j; + + for (i = 0; i <= m->indep_count; i++) { + for (j = i; j <= m->indep_count; j++) { + m->covariance[i][j] += var[i] * var[j]; + } + } +} + +void avpriv_solve_lls(LLSModel *m, double threshold, unsigned short min_order) +{ + int i, j, k; + double (*factor)[MAX_VARS_ALIGN] = (void *) &m->covariance[1][0]; + double (*covar) [MAX_VARS_ALIGN] = (void *) &m->covariance[1][1]; + double *covar_y = m->covariance[0]; + int count = m->indep_count; + + for (i = 0; i < count; i++) { + for (j = i; j < count; j++) { + double sum = covar[i][j]; + + for (k = 0; k <= i-1; k++) + sum -= factor[i][k] * factor[j][k]; + + if (i == j) { + if (sum < threshold) + sum = 1.0; + factor[i][i] = sqrt(sum); + } else { + factor[j][i] = sum / factor[i][i]; + } + } + } + + for (i = 0; i < count; i++) { + double sum = covar_y[i + 1]; + + for (k = 0; k <= i-1; k++) + sum -= factor[i][k] * m->coeff[0][k]; + + m->coeff[0][i] = sum / factor[i][i]; + } + + for (j = count - 1; j >= min_order; j--) { + for (i = j; i >= 0; i--) { + double sum = m->coeff[0][i]; + + for (k = i + 1; k <= j; k++) + sum -= factor[k][i] * m->coeff[j][k]; + + m->coeff[j][i] = sum / factor[i][i]; + } + + m->variance[j] = covar_y[0]; + + for (i = 0; i <= j; i++) { + double sum = m->coeff[j][i] * covar[i][i] - 2 * covar_y[i + 1]; + + for (k = 0; k < i; k++) + sum += 2 * m->coeff[j][k] * covar[k][i]; + + m->variance[j] += m->coeff[j][i] * sum; + } + } +} + +static double evaluate_lls(LLSModel *m, const double *param, int order) +{ + int i; + double out = 0; + + for (i = 0; i <= order; i++) + out += param[i] * m->coeff[order][i]; + + return out; +} + +av_cold void avpriv_init_lls(LLSModel *m, int indep_count) +{ + memset(m, 0, sizeof(LLSModel)); + m->indep_count = indep_count; + m->update_lls = update_lls; + m->evaluate_lls = evaluate_lls; +#if ARCH_X86 + ff_init_lls_x86(m); +#endif +} diff --git a/media/ffvpx/libavutil/lls.h b/media/ffvpx/libavutil/lls.h new file mode 100644 index 0000000000..0709275822 --- /dev/null +++ b/media/ffvpx/libavutil/lls.h @@ -0,0 +1,63 @@ +/* + * linear least squares model + * + * Copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_LLS_H +#define AVUTIL_LLS_H + +#include "macros.h" +#include "mem_internal.h" + +#define MAX_VARS 32 +#define MAX_VARS_ALIGN FFALIGN(MAX_VARS+1,4) + +//FIXME avoid direct access to LLSModel from outside + +/** + * Linear least squares model. + */ +typedef struct LLSModel { + DECLARE_ALIGNED(32, double, covariance[MAX_VARS_ALIGN][MAX_VARS_ALIGN]); + DECLARE_ALIGNED(32, double, coeff[MAX_VARS][MAX_VARS]); + double variance[MAX_VARS]; + int indep_count; + /** + * Take the outer-product of var[] with itself, and add to the covariance matrix. + * @param m this context + * @param var training samples, starting with the value to be predicted + * 32-byte aligned, and any padding elements must be initialized + * (i.e not denormal/nan). + */ + void (*update_lls)(struct LLSModel *m, const double *var); + /** + * Inner product of var[] and the LPC coefs. + * @param m this context + * @param var training samples, excluding the value to be predicted. unaligned. + * @param order lpc order + */ + double (*evaluate_lls)(struct LLSModel *m, const double *var, int order); +} LLSModel; + +void avpriv_init_lls(LLSModel *m, int indep_count); +void ff_init_lls_x86(LLSModel *m); +void avpriv_solve_lls(LLSModel *m, double threshold, unsigned short min_order); + +#endif /* AVUTIL_LLS_H */ diff --git a/media/ffvpx/libavutil/log.c b/media/ffvpx/libavutil/log.c new file mode 100644 index 0000000000..46662f3db0 --- /dev/null +++ b/media/ffvpx/libavutil/log.c @@ -0,0 +1,497 @@ +/* + * log functions + * Copyright (c) 2003 Michel Bardiaux + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * logging functions + */ + +#include "config.h" + +#if HAVE_UNISTD_H +#include <unistd.h> +#endif +#if HAVE_IO_H +#include <io.h> +#endif +#include <inttypes.h> +#include <stdarg.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include "bprint.h" +#include "common.h" +#include "internal.h" +#include "log.h" +#include "thread.h" + +static AVMutex mutex = AV_MUTEX_INITIALIZER; + +#define LINE_SZ 1024 + +#if HAVE_VALGRIND_VALGRIND_H && CONFIG_VALGRIND_BACKTRACE +#include <valgrind/valgrind.h> +/* this is the log level at which valgrind will output a full backtrace */ +#define BACKTRACE_LOGLEVEL AV_LOG_ERROR +#endif + +static int av_log_level = AV_LOG_INFO; +static int flags; + +#define NB_LEVELS 8 +#if defined(_WIN32) && HAVE_SETCONSOLETEXTATTRIBUTE && HAVE_GETSTDHANDLE +#include <windows.h> +static const uint8_t color[16 + AV_CLASS_CATEGORY_NB] = { + [AV_LOG_PANIC /8] = 12, + [AV_LOG_FATAL /8] = 12, + [AV_LOG_ERROR /8] = 12, + [AV_LOG_WARNING/8] = 14, + [AV_LOG_INFO /8] = 7, + [AV_LOG_VERBOSE/8] = 10, + [AV_LOG_DEBUG /8] = 10, + [AV_LOG_TRACE /8] = 8, + [16+AV_CLASS_CATEGORY_NA ] = 7, + [16+AV_CLASS_CATEGORY_INPUT ] = 13, + [16+AV_CLASS_CATEGORY_OUTPUT ] = 5, + [16+AV_CLASS_CATEGORY_MUXER ] = 13, + [16+AV_CLASS_CATEGORY_DEMUXER ] = 5, + [16+AV_CLASS_CATEGORY_ENCODER ] = 11, + [16+AV_CLASS_CATEGORY_DECODER ] = 3, + [16+AV_CLASS_CATEGORY_FILTER ] = 10, + [16+AV_CLASS_CATEGORY_BITSTREAM_FILTER] = 9, + [16+AV_CLASS_CATEGORY_SWSCALER ] = 7, + [16+AV_CLASS_CATEGORY_SWRESAMPLER ] = 7, + [16+AV_CLASS_CATEGORY_DEVICE_VIDEO_OUTPUT ] = 13, + [16+AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT ] = 5, + [16+AV_CLASS_CATEGORY_DEVICE_AUDIO_OUTPUT ] = 13, + [16+AV_CLASS_CATEGORY_DEVICE_AUDIO_INPUT ] = 5, + [16+AV_CLASS_CATEGORY_DEVICE_OUTPUT ] = 13, + [16+AV_CLASS_CATEGORY_DEVICE_INPUT ] = 5, +}; + +static int16_t background, attr_orig; +static HANDLE con; +#else + +static const uint32_t color[16 + AV_CLASS_CATEGORY_NB] = { + [AV_LOG_PANIC /8] = 52 << 16 | 196 << 8 | 0x41, + [AV_LOG_FATAL /8] = 208 << 8 | 0x41, + [AV_LOG_ERROR /8] = 196 << 8 | 0x11, + [AV_LOG_WARNING/8] = 226 << 8 | 0x03, + [AV_LOG_INFO /8] = 253 << 8 | 0x09, + [AV_LOG_VERBOSE/8] = 40 << 8 | 0x02, + [AV_LOG_DEBUG /8] = 34 << 8 | 0x02, + [AV_LOG_TRACE /8] = 34 << 8 | 0x07, + [16+AV_CLASS_CATEGORY_NA ] = 250 << 8 | 0x09, + [16+AV_CLASS_CATEGORY_INPUT ] = 219 << 8 | 0x15, + [16+AV_CLASS_CATEGORY_OUTPUT ] = 201 << 8 | 0x05, + [16+AV_CLASS_CATEGORY_MUXER ] = 213 << 8 | 0x15, + [16+AV_CLASS_CATEGORY_DEMUXER ] = 207 << 8 | 0x05, + [16+AV_CLASS_CATEGORY_ENCODER ] = 51 << 8 | 0x16, + [16+AV_CLASS_CATEGORY_DECODER ] = 39 << 8 | 0x06, + [16+AV_CLASS_CATEGORY_FILTER ] = 155 << 8 | 0x12, + [16+AV_CLASS_CATEGORY_BITSTREAM_FILTER] = 192 << 8 | 0x14, + [16+AV_CLASS_CATEGORY_SWSCALER ] = 153 << 8 | 0x14, + [16+AV_CLASS_CATEGORY_SWRESAMPLER ] = 147 << 8 | 0x14, + [16+AV_CLASS_CATEGORY_DEVICE_VIDEO_OUTPUT ] = 213 << 8 | 0x15, + [16+AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT ] = 207 << 8 | 0x05, + [16+AV_CLASS_CATEGORY_DEVICE_AUDIO_OUTPUT ] = 213 << 8 | 0x15, + [16+AV_CLASS_CATEGORY_DEVICE_AUDIO_INPUT ] = 207 << 8 | 0x05, + [16+AV_CLASS_CATEGORY_DEVICE_OUTPUT ] = 213 << 8 | 0x15, + [16+AV_CLASS_CATEGORY_DEVICE_INPUT ] = 207 << 8 | 0x05, +}; + +#endif +static int use_color = -1; + +#if defined(_WIN32) && HAVE_SETCONSOLETEXTATTRIBUTE && HAVE_GETSTDHANDLE +static void win_console_puts(const char *str) +{ + const uint8_t *q = str; + uint16_t line[LINE_SZ]; + + while (*q) { + uint16_t *buf = line; + DWORD nb_chars = 0; + DWORD written; + + while (*q && nb_chars < LINE_SZ - 1) { + uint32_t ch; + uint16_t tmp; + + GET_UTF8(ch, *q ? *q++ : 0, ch = 0xfffd; goto continue_on_invalid;) +continue_on_invalid: + PUT_UTF16(ch, tmp, *buf++ = tmp; nb_chars++;) + } + + WriteConsoleW(con, line, nb_chars, &written, NULL); + } +} +#endif + +static void check_color_terminal(void) +{ + char *term = getenv("TERM"); + +#if defined(_WIN32) && HAVE_SETCONSOLETEXTATTRIBUTE && HAVE_GETSTDHANDLE + CONSOLE_SCREEN_BUFFER_INFO con_info; + DWORD dummy; + con = GetStdHandle(STD_ERROR_HANDLE); + if (con != INVALID_HANDLE_VALUE && !GetConsoleMode(con, &dummy)) + con = INVALID_HANDLE_VALUE; + if (con != INVALID_HANDLE_VALUE) { + GetConsoleScreenBufferInfo(con, &con_info); + attr_orig = con_info.wAttributes; + background = attr_orig & 0xF0; + } +#endif + + if (getenv("AV_LOG_FORCE_NOCOLOR")) { + use_color = 0; + } else if (getenv("AV_LOG_FORCE_COLOR")) { + use_color = 1; + } else { +#if defined(_WIN32) && HAVE_SETCONSOLETEXTATTRIBUTE && HAVE_GETSTDHANDLE + use_color = (con != INVALID_HANDLE_VALUE); +#elif HAVE_ISATTY + use_color = (term && isatty(2)); +#else + use_color = 0; +#endif + } + + if (getenv("AV_LOG_FORCE_256COLOR") || term && strstr(term, "256color")) + use_color *= 256; +} + +static void ansi_fputs(int level, int tint, const char *str, int local_use_color) +{ + if (local_use_color == 1) { + fprintf(stderr, + "\033[%"PRIu32";3%"PRIu32"m%s\033[0m", + (color[level] >> 4) & 15, + color[level] & 15, + str); + } else if (tint && use_color == 256) { + fprintf(stderr, + "\033[48;5;%"PRIu32"m\033[38;5;%dm%s\033[0m", + (color[level] >> 16) & 0xff, + tint, + str); + } else if (local_use_color == 256) { + fprintf(stderr, + "\033[48;5;%"PRIu32"m\033[38;5;%"PRIu32"m%s\033[0m", + (color[level] >> 16) & 0xff, + (color[level] >> 8) & 0xff, + str); + } else + fputs(str, stderr); +} + +static void colored_fputs(int level, int tint, const char *str) +{ + int local_use_color; + if (!*str) + return; + + if (use_color < 0) + check_color_terminal(); + + if (level == AV_LOG_INFO/8) local_use_color = 0; + else local_use_color = use_color; + +#if defined(_WIN32) && HAVE_SETCONSOLETEXTATTRIBUTE && HAVE_GETSTDHANDLE + if (con != INVALID_HANDLE_VALUE) { + if (local_use_color) + SetConsoleTextAttribute(con, background | color[level]); + win_console_puts(str); + if (local_use_color) + SetConsoleTextAttribute(con, attr_orig); + } else { + ansi_fputs(level, tint, str, local_use_color); + } +#else + ansi_fputs(level, tint, str, local_use_color); +#endif + +} + +const char *av_default_item_name(void *ptr) +{ + return (*(AVClass **) ptr)->class_name; +} + +AVClassCategory av_default_get_category(void *ptr) +{ + return (*(AVClass **) ptr)->category; +} + +static void sanitize(uint8_t *line){ + while(*line){ + if(*line < 0x08 || (*line > 0x0D && *line < 0x20)) + *line='?'; + line++; + } +} + +static int get_category(void *ptr){ + AVClass *avc = *(AVClass **) ptr; + if( !avc + || (avc->version&0xFF)<100 + || avc->version < (51 << 16 | 59 << 8) + || avc->category >= AV_CLASS_CATEGORY_NB) return AV_CLASS_CATEGORY_NA + 16; + + if(avc->get_category) + return avc->get_category(ptr) + 16; + + return avc->category + 16; +} + +static const char *get_level_str(int level) +{ + switch (level) { + case AV_LOG_QUIET: + return "quiet"; + case AV_LOG_DEBUG: + return "debug"; + case AV_LOG_TRACE: + return "trace"; + case AV_LOG_VERBOSE: + return "verbose"; + case AV_LOG_INFO: + return "info"; + case AV_LOG_WARNING: + return "warning"; + case AV_LOG_ERROR: + return "error"; + case AV_LOG_FATAL: + return "fatal"; + case AV_LOG_PANIC: + return "panic"; + default: + return ""; + } +} + +static const char *item_name(void *obj, const AVClass *cls) +{ + return (cls->item_name ? cls->item_name : av_default_item_name)(obj); +} + +static void format_line(void *avcl, int level, const char *fmt, va_list vl, + AVBPrint part[4], int *print_prefix, int type[2]) +{ + AVClass* avc = avcl ? *(AVClass **) avcl : NULL; + av_bprint_init(part+0, 0, AV_BPRINT_SIZE_AUTOMATIC); + av_bprint_init(part+1, 0, AV_BPRINT_SIZE_AUTOMATIC); + av_bprint_init(part+2, 0, AV_BPRINT_SIZE_AUTOMATIC); + av_bprint_init(part+3, 0, 65536); + + if(type) type[0] = type[1] = AV_CLASS_CATEGORY_NA + 16; + if (*print_prefix && avc) { + if (avc->parent_log_context_offset) { + AVClass** parent = *(AVClass ***) (((uint8_t *) avcl) + + avc->parent_log_context_offset); + if (parent && *parent) { + av_bprintf(part+0, "[%s @ %p] ", + item_name(parent, *parent), parent); + if(type) type[0] = get_category(parent); + } + } + av_bprintf(part+1, "[%s @ %p] ", + item_name(avcl, avc), avcl); + if(type) type[1] = get_category(avcl); + } + + if (*print_prefix && (level > AV_LOG_QUIET) && (flags & AV_LOG_PRINT_LEVEL)) + av_bprintf(part+2, "[%s] ", get_level_str(level)); + + av_vbprintf(part+3, fmt, vl); + + if(*part[0].str || *part[1].str || *part[2].str || *part[3].str) { + char lastc = part[3].len && part[3].len <= part[3].size ? part[3].str[part[3].len - 1] : 0; + *print_prefix = lastc == '\n' || lastc == '\r'; + } +} + +void av_log_format_line(void *ptr, int level, const char *fmt, va_list vl, + char *line, int line_size, int *print_prefix) +{ + av_log_format_line2(ptr, level, fmt, vl, line, line_size, print_prefix); +} + +int av_log_format_line2(void *ptr, int level, const char *fmt, va_list vl, + char *line, int line_size, int *print_prefix) +{ + AVBPrint part[4]; + int ret; + + format_line(ptr, level, fmt, vl, part, print_prefix, NULL); + ret = snprintf(line, line_size, "%s%s%s%s", part[0].str, part[1].str, part[2].str, part[3].str); + av_bprint_finalize(part+3, NULL); + return ret; +} + +void av_log_default_callback(void* ptr, int level, const char* fmt, va_list vl) +{ + static int print_prefix = 1; + static int count; + static char prev[LINE_SZ]; + AVBPrint part[4]; + char line[LINE_SZ]; + static int is_atty; + int type[2]; + unsigned tint = 0; + + if (level >= 0) { + tint = level & 0xff00; + level &= 0xff; + } + + if (level > av_log_level) + return; + ff_mutex_lock(&mutex); + + format_line(ptr, level, fmt, vl, part, &print_prefix, type); + snprintf(line, sizeof(line), "%s%s%s%s", part[0].str, part[1].str, part[2].str, part[3].str); + +#if HAVE_ISATTY + if (!is_atty) + is_atty = isatty(2) ? 1 : -1; +#endif + + if (print_prefix && (flags & AV_LOG_SKIP_REPEATED) && !strcmp(line, prev) && + *line && line[strlen(line) - 1] != '\r'){ + count++; + if (is_atty == 1) + fprintf(stderr, " Last message repeated %d times\r", count); + goto end; + } + if (count > 0) { + fprintf(stderr, " Last message repeated %d times\n", count); + count = 0; + } + strcpy(prev, line); + sanitize(part[0].str); + colored_fputs(type[0], 0, part[0].str); + sanitize(part[1].str); + colored_fputs(type[1], 0, part[1].str); + sanitize(part[2].str); + colored_fputs(av_clip(level >> 3, 0, NB_LEVELS - 1), tint >> 8, part[2].str); + sanitize(part[3].str); + colored_fputs(av_clip(level >> 3, 0, NB_LEVELS - 1), tint >> 8, part[3].str); + +#if CONFIG_VALGRIND_BACKTRACE + if (level <= BACKTRACE_LOGLEVEL) + VALGRIND_PRINTF_BACKTRACE("%s", ""); +#endif +end: + av_bprint_finalize(part+3, NULL); + ff_mutex_unlock(&mutex); +} + +static void (*av_log_callback)(void*, int, const char*, va_list) = + av_log_default_callback; + +void av_log(void* avcl, int level, const char *fmt, ...) +{ + va_list vl; + va_start(vl, fmt); + av_vlog(avcl, level, fmt, vl); + va_end(vl); +} + +void av_log_once(void* avcl, int initial_level, int subsequent_level, int *state, const char *fmt, ...) +{ + va_list vl; + va_start(vl, fmt); + av_vlog(avcl, *state ? subsequent_level : initial_level, fmt, vl); + va_end(vl); + *state = 1; +} + +void av_vlog(void* avcl, int level, const char *fmt, va_list vl) +{ + AVClass* avc = avcl ? *(AVClass **) avcl : NULL; + void (*log_callback)(void*, int, const char*, va_list) = av_log_callback; + if (avc && avc->version >= (50 << 16 | 15 << 8 | 2) && + avc->log_level_offset_offset && level >= AV_LOG_FATAL) + level += *(int *) (((uint8_t *) avcl) + avc->log_level_offset_offset); + if (log_callback) + log_callback(avcl, level, fmt, vl); +} + +int av_log_get_level(void) +{ + return av_log_level; +} + +void av_log_set_level(int level) +{ + av_log_level = level; +} + +void av_log_set_flags(int arg) +{ + flags = arg; +} + +int av_log_get_flags(void) +{ + return flags; +} + +void av_log_set_callback(void (*callback)(void*, int, const char*, va_list)) +{ + av_log_callback = callback; +} + +static void missing_feature_sample(int sample, void *avc, const char *msg, + va_list argument_list) +{ + av_vlog(avc, AV_LOG_WARNING, msg, argument_list); + av_log(avc, AV_LOG_WARNING, " is not implemented. Update your FFmpeg " + "version to the newest one from Git. If the problem still " + "occurs, it means that your file has a feature which has not " + "been implemented.\n"); + if (sample) + av_log(avc, AV_LOG_WARNING, "If you want to help, upload a sample " + "of this file to https://streams.videolan.org/upload/ " + "and contact the ffmpeg-devel mailing list. (ffmpeg-devel@ffmpeg.org)\n"); +} + +void avpriv_request_sample(void *avc, const char *msg, ...) +{ + va_list argument_list; + + va_start(argument_list, msg); + missing_feature_sample(1, avc, msg, argument_list); + va_end(argument_list); +} + +void avpriv_report_missing_feature(void *avc, const char *msg, ...) +{ + va_list argument_list; + + va_start(argument_list, msg); + missing_feature_sample(0, avc, msg, argument_list); + va_end(argument_list); +} diff --git a/media/ffvpx/libavutil/log.h b/media/ffvpx/libavutil/log.h new file mode 100644 index 0000000000..ab7ceabe22 --- /dev/null +++ b/media/ffvpx/libavutil/log.h @@ -0,0 +1,387 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_LOG_H +#define AVUTIL_LOG_H + +#include <stdarg.h> +#include "attributes.h" +#include "version.h" + +typedef enum { + AV_CLASS_CATEGORY_NA = 0, + AV_CLASS_CATEGORY_INPUT, + AV_CLASS_CATEGORY_OUTPUT, + AV_CLASS_CATEGORY_MUXER, + AV_CLASS_CATEGORY_DEMUXER, + AV_CLASS_CATEGORY_ENCODER, + AV_CLASS_CATEGORY_DECODER, + AV_CLASS_CATEGORY_FILTER, + AV_CLASS_CATEGORY_BITSTREAM_FILTER, + AV_CLASS_CATEGORY_SWSCALER, + AV_CLASS_CATEGORY_SWRESAMPLER, + AV_CLASS_CATEGORY_DEVICE_VIDEO_OUTPUT = 40, + AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT, + AV_CLASS_CATEGORY_DEVICE_AUDIO_OUTPUT, + AV_CLASS_CATEGORY_DEVICE_AUDIO_INPUT, + AV_CLASS_CATEGORY_DEVICE_OUTPUT, + AV_CLASS_CATEGORY_DEVICE_INPUT, + AV_CLASS_CATEGORY_NB ///< not part of ABI/API +}AVClassCategory; + +#define AV_IS_INPUT_DEVICE(category) \ + (((category) == AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT) || \ + ((category) == AV_CLASS_CATEGORY_DEVICE_AUDIO_INPUT) || \ + ((category) == AV_CLASS_CATEGORY_DEVICE_INPUT)) + +#define AV_IS_OUTPUT_DEVICE(category) \ + (((category) == AV_CLASS_CATEGORY_DEVICE_VIDEO_OUTPUT) || \ + ((category) == AV_CLASS_CATEGORY_DEVICE_AUDIO_OUTPUT) || \ + ((category) == AV_CLASS_CATEGORY_DEVICE_OUTPUT)) + +struct AVOptionRanges; + +/** + * Describe the class of an AVClass context structure. That is an + * arbitrary struct of which the first field is a pointer to an + * AVClass struct (e.g. AVCodecContext, AVFormatContext etc.). + */ +typedef struct AVClass { + /** + * The name of the class; usually it is the same name as the + * context structure type to which the AVClass is associated. + */ + const char* class_name; + + /** + * A pointer to a function which returns the name of a context + * instance ctx associated with the class. + */ + const char* (*item_name)(void* ctx); + + /** + * a pointer to the first option specified in the class if any or NULL + * + * @see av_set_default_options() + */ + const struct AVOption *option; + + /** + * LIBAVUTIL_VERSION with which this structure was created. + * This is used to allow fields to be added without requiring major + * version bumps everywhere. + */ + + int version; + + /** + * Offset in the structure where log_level_offset is stored. + * 0 means there is no such variable + */ + int log_level_offset_offset; + + /** + * Offset in the structure where a pointer to the parent context for + * logging is stored. For example a decoder could pass its AVCodecContext + * to eval as such a parent context, which an av_log() implementation + * could then leverage to display the parent context. + * The offset can be NULL. + */ + int parent_log_context_offset; + + /** + * Category used for visualization (like color) + * This is only set if the category is equal for all objects using this class. + * available since version (51 << 16 | 56 << 8 | 100) + */ + AVClassCategory category; + + /** + * Callback to return the category. + * available since version (51 << 16 | 59 << 8 | 100) + */ + AVClassCategory (*get_category)(void* ctx); + + /** + * Callback to return the supported/allowed ranges. + * available since version (52.12) + */ + int (*query_ranges)(struct AVOptionRanges **, void *obj, const char *key, int flags); + + /** + * Return next AVOptions-enabled child or NULL + */ + void* (*child_next)(void *obj, void *prev); + + /** + * Iterate over the AVClasses corresponding to potential AVOptions-enabled + * children. + * + * @param iter pointer to opaque iteration state. The caller must initialize + * *iter to NULL before the first call. + * @return AVClass for the next AVOptions-enabled child or NULL if there are + * no more such children. + * + * @note The difference between child_next and this is that child_next + * iterates over _already existing_ objects, while child_class_iterate + * iterates over _all possible_ children. + */ + const struct AVClass* (*child_class_iterate)(void **iter); +} AVClass; + +/** + * @addtogroup lavu_log + * + * @{ + * + * @defgroup lavu_log_constants Logging Constants + * + * @{ + */ + +/** + * Print no output. + */ +#define AV_LOG_QUIET -8 + +/** + * Something went really wrong and we will crash now. + */ +#define AV_LOG_PANIC 0 + +/** + * Something went wrong and recovery is not possible. + * For example, no header was found for a format which depends + * on headers or an illegal combination of parameters is used. + */ +#define AV_LOG_FATAL 8 + +/** + * Something went wrong and cannot losslessly be recovered. + * However, not all future data is affected. + */ +#define AV_LOG_ERROR 16 + +/** + * Something somehow does not look correct. This may or may not + * lead to problems. An example would be the use of '-vstrict -2'. + */ +#define AV_LOG_WARNING 24 + +/** + * Standard information. + */ +#define AV_LOG_INFO 32 + +/** + * Detailed information. + */ +#define AV_LOG_VERBOSE 40 + +/** + * Stuff which is only useful for libav* developers. + */ +#define AV_LOG_DEBUG 48 + +/** + * Extremely verbose debugging, useful for libav* development. + */ +#define AV_LOG_TRACE 56 + +#define AV_LOG_MAX_OFFSET (AV_LOG_TRACE - AV_LOG_QUIET) + +/** + * @} + */ + +/** + * Sets additional colors for extended debugging sessions. + * @code + av_log(ctx, AV_LOG_DEBUG|AV_LOG_C(134), "Message in purple\n"); + @endcode + * Requires 256color terminal support. Uses outside debugging is not + * recommended. + */ +#define AV_LOG_C(x) ((x) << 8) + +/** + * Send the specified message to the log if the level is less than or equal + * to the current av_log_level. By default, all logging messages are sent to + * stderr. This behavior can be altered by setting a different logging callback + * function. + * @see av_log_set_callback + * + * @param avcl A pointer to an arbitrary struct of which the first field is a + * pointer to an AVClass struct or NULL if general log. + * @param level The importance level of the message expressed using a @ref + * lavu_log_constants "Logging Constant". + * @param fmt The format string (printf-compatible) that specifies how + * subsequent arguments are converted to output. + */ +void av_log(void *avcl, int level, const char *fmt, ...) av_printf_format(3, 4); + +/** + * Send the specified message to the log once with the initial_level and then with + * the subsequent_level. By default, all logging messages are sent to + * stderr. This behavior can be altered by setting a different logging callback + * function. + * @see av_log + * + * @param avcl A pointer to an arbitrary struct of which the first field is a + * pointer to an AVClass struct or NULL if general log. + * @param initial_level importance level of the message expressed using a @ref + * lavu_log_constants "Logging Constant" for the first occurance. + * @param subsequent_level importance level of the message expressed using a @ref + * lavu_log_constants "Logging Constant" after the first occurance. + * @param fmt The format string (printf-compatible) that specifies how + * subsequent arguments are converted to output. + * @param state a variable to keep trak of if a message has already been printed + * this must be initialized to 0 before the first use. The same state + * must not be accessed by 2 Threads simultaneously. + */ +void av_log_once(void* avcl, int initial_level, int subsequent_level, int *state, const char *fmt, ...) av_printf_format(5, 6); + + +/** + * Send the specified message to the log if the level is less than or equal + * to the current av_log_level. By default, all logging messages are sent to + * stderr. This behavior can be altered by setting a different logging callback + * function. + * @see av_log_set_callback + * + * @param avcl A pointer to an arbitrary struct of which the first field is a + * pointer to an AVClass struct. + * @param level The importance level of the message expressed using a @ref + * lavu_log_constants "Logging Constant". + * @param fmt The format string (printf-compatible) that specifies how + * subsequent arguments are converted to output. + * @param vl The arguments referenced by the format string. + */ +void av_vlog(void *avcl, int level, const char *fmt, va_list vl); + +/** + * Get the current log level + * + * @see lavu_log_constants + * + * @return Current log level + */ +int av_log_get_level(void); + +/** + * Set the log level + * + * @see lavu_log_constants + * + * @param level Logging level + */ +void av_log_set_level(int level); + +/** + * Set the logging callback + * + * @note The callback must be thread safe, even if the application does not use + * threads itself as some codecs are multithreaded. + * + * @see av_log_default_callback + * + * @param callback A logging function with a compatible signature. + */ +void av_log_set_callback(void (*callback)(void*, int, const char*, va_list)); + +/** + * Default logging callback + * + * It prints the message to stderr, optionally colorizing it. + * + * @param avcl A pointer to an arbitrary struct of which the first field is a + * pointer to an AVClass struct. + * @param level The importance level of the message expressed using a @ref + * lavu_log_constants "Logging Constant". + * @param fmt The format string (printf-compatible) that specifies how + * subsequent arguments are converted to output. + * @param vl The arguments referenced by the format string. + */ +void av_log_default_callback(void *avcl, int level, const char *fmt, + va_list vl); + +/** + * Return the context name + * + * @param ctx The AVClass context + * + * @return The AVClass class_name + */ +const char* av_default_item_name(void* ctx); +AVClassCategory av_default_get_category(void *ptr); + +/** + * Format a line of log the same way as the default callback. + * @param line buffer to receive the formatted line + * @param line_size size of the buffer + * @param print_prefix used to store whether the prefix must be printed; + * must point to a persistent integer initially set to 1 + */ +void av_log_format_line(void *ptr, int level, const char *fmt, va_list vl, + char *line, int line_size, int *print_prefix); + +/** + * Format a line of log the same way as the default callback. + * @param line buffer to receive the formatted line; + * may be NULL if line_size is 0 + * @param line_size size of the buffer; at most line_size-1 characters will + * be written to the buffer, plus one null terminator + * @param print_prefix used to store whether the prefix must be printed; + * must point to a persistent integer initially set to 1 + * @return Returns a negative value if an error occurred, otherwise returns + * the number of characters that would have been written for a + * sufficiently large buffer, not including the terminating null + * character. If the return value is not less than line_size, it means + * that the log message was truncated to fit the buffer. + */ +int av_log_format_line2(void *ptr, int level, const char *fmt, va_list vl, + char *line, int line_size, int *print_prefix); + +/** + * Skip repeated messages, this requires the user app to use av_log() instead of + * (f)printf as the 2 would otherwise interfere and lead to + * "Last message repeated x times" messages below (f)printf messages with some + * bad luck. + * Also to receive the last, "last repeated" line if any, the user app must + * call av_log(NULL, AV_LOG_QUIET, "%s", ""); at the end + */ +#define AV_LOG_SKIP_REPEATED 1 + +/** + * Include the log severity in messages originating from codecs. + * + * Results in messages such as: + * [rawvideo @ 0xDEADBEEF] [error] encode did not produce valid pts + */ +#define AV_LOG_PRINT_LEVEL 2 + +void av_log_set_flags(int arg); +int av_log_get_flags(void); + +/** + * @} + */ + +#endif /* AVUTIL_LOG_H */ diff --git a/media/ffvpx/libavutil/log2_tab.c b/media/ffvpx/libavutil/log2_tab.c new file mode 100644 index 0000000000..0dbf07d74c --- /dev/null +++ b/media/ffvpx/libavutil/log2_tab.c @@ -0,0 +1,32 @@ +/* + * Copyright (c) 2003-2012 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> + +const uint8_t ff_log2_tab[256]={ + 0,0,1,1,2,2,2,2,3,3,3,3,3,3,3,3,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, + 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, + 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7 +}; diff --git a/media/ffvpx/libavutil/macros.h b/media/ffvpx/libavutil/macros.h new file mode 100644 index 0000000000..2a7567c3ea --- /dev/null +++ b/media/ffvpx/libavutil/macros.h @@ -0,0 +1,80 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu + * Utility Preprocessor macros + */ + +#ifndef AVUTIL_MACROS_H +#define AVUTIL_MACROS_H + +#include "libavutil/avconfig.h" + +#if AV_HAVE_BIGENDIAN +# define AV_NE(be, le) (be) +#else +# define AV_NE(be, le) (le) +#endif + +/** + * Comparator. + * For two numerical expressions x and y, gives 1 if x > y, -1 if x < y, and 0 + * if x == y. This is useful for instance in a qsort comparator callback. + * Furthermore, compilers are able to optimize this to branchless code, and + * there is no risk of overflow with signed types. + * As with many macros, this evaluates its argument multiple times, it thus + * must not have a side-effect. + */ +#define FFDIFFSIGN(x,y) (((x)>(y)) - ((x)<(y))) + +#define FFMAX(a,b) ((a) > (b) ? (a) : (b)) +#define FFMAX3(a,b,c) FFMAX(FFMAX(a,b),c) +#define FFMIN(a,b) ((a) > (b) ? (b) : (a)) +#define FFMIN3(a,b,c) FFMIN(FFMIN(a,b),c) + +#define FFSWAP(type,a,b) do{type SWAP_tmp= b; b= a; a= SWAP_tmp;}while(0) +#define FF_ARRAY_ELEMS(a) (sizeof(a) / sizeof((a)[0])) + +#define MKTAG(a,b,c,d) ((a) | ((b) << 8) | ((c) << 16) | ((unsigned)(d) << 24)) +#define MKBETAG(a,b,c,d) ((d) | ((c) << 8) | ((b) << 16) | ((unsigned)(a) << 24)) + +/** + * @addtogroup preproc_misc Preprocessor String Macros + * + * String manipulation macros + * + * @{ + */ + +#define AV_STRINGIFY(s) AV_TOSTRING(s) +#define AV_TOSTRING(s) #s + +#define AV_GLUE(a, b) a ## b +#define AV_JOIN(a, b) AV_GLUE(a, b) + +/** + * @} + */ + +#define AV_PRAGMA(s) _Pragma(#s) + +#define FFALIGN(x, a) (((x)+(a)-1)&~((a)-1)) + +#endif /* AVUTIL_MACROS_H */ diff --git a/media/ffvpx/libavutil/mastering_display_metadata.c b/media/ffvpx/libavutil/mastering_display_metadata.c new file mode 100644 index 0000000000..6069347617 --- /dev/null +++ b/media/ffvpx/libavutil/mastering_display_metadata.c @@ -0,0 +1,66 @@ +/** + * Copyright (c) 2016 Neil Birkbeck <neil.birkbeck@gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> +#include <string.h> + +#include "mastering_display_metadata.h" +#include "mem.h" + +AVMasteringDisplayMetadata *av_mastering_display_metadata_alloc(void) +{ + return av_mallocz(sizeof(AVMasteringDisplayMetadata)); +} + +AVMasteringDisplayMetadata *av_mastering_display_metadata_create_side_data(AVFrame *frame) +{ + AVFrameSideData *side_data = av_frame_new_side_data(frame, + AV_FRAME_DATA_MASTERING_DISPLAY_METADATA, + sizeof(AVMasteringDisplayMetadata)); + if (!side_data) + return NULL; + + memset(side_data->data, 0, sizeof(AVMasteringDisplayMetadata)); + + return (AVMasteringDisplayMetadata *)side_data->data; +} + +AVContentLightMetadata *av_content_light_metadata_alloc(size_t *size) +{ + AVContentLightMetadata *metadata = av_mallocz(sizeof(AVContentLightMetadata)); + + if (size) + *size = sizeof(*metadata); + + return metadata; +} + +AVContentLightMetadata *av_content_light_metadata_create_side_data(AVFrame *frame) +{ + AVFrameSideData *side_data = av_frame_new_side_data(frame, + AV_FRAME_DATA_CONTENT_LIGHT_LEVEL, + sizeof(AVContentLightMetadata)); + if (!side_data) + return NULL; + + memset(side_data->data, 0, sizeof(AVContentLightMetadata)); + + return (AVContentLightMetadata *)side_data->data; +} diff --git a/media/ffvpx/libavutil/mastering_display_metadata.h b/media/ffvpx/libavutil/mastering_display_metadata.h new file mode 100644 index 0000000000..c23b07c3cd --- /dev/null +++ b/media/ffvpx/libavutil/mastering_display_metadata.h @@ -0,0 +1,128 @@ +/* + * Copyright (c) 2016 Neil Birkbeck <neil.birkbeck@gmail.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_MASTERING_DISPLAY_METADATA_H +#define AVUTIL_MASTERING_DISPLAY_METADATA_H + +#include "frame.h" +#include "rational.h" + + +/** + * Mastering display metadata capable of representing the color volume of + * the display used to master the content (SMPTE 2086:2014). + * + * To be used as payload of a AVFrameSideData or AVPacketSideData with the + * appropriate type. + * + * @note The struct should be allocated with av_mastering_display_metadata_alloc() + * and its size is not a part of the public ABI. + */ +typedef struct AVMasteringDisplayMetadata { + /** + * CIE 1931 xy chromaticity coords of color primaries (r, g, b order). + */ + AVRational display_primaries[3][2]; + + /** + * CIE 1931 xy chromaticity coords of white point. + */ + AVRational white_point[2]; + + /** + * Min luminance of mastering display (cd/m^2). + */ + AVRational min_luminance; + + /** + * Max luminance of mastering display (cd/m^2). + */ + AVRational max_luminance; + + /** + * Flag indicating whether the display primaries (and white point) are set. + */ + int has_primaries; + + /** + * Flag indicating whether the luminance (min_ and max_) have been set. + */ + int has_luminance; + +} AVMasteringDisplayMetadata; + +/** + * Allocate an AVMasteringDisplayMetadata structure and set its fields to + * default values. The resulting struct can be freed using av_freep(). + * + * @return An AVMasteringDisplayMetadata filled with default values or NULL + * on failure. + */ +AVMasteringDisplayMetadata *av_mastering_display_metadata_alloc(void); + +/** + * Allocate a complete AVMasteringDisplayMetadata and add it to the frame. + * + * @param frame The frame which side data is added to. + * + * @return The AVMasteringDisplayMetadata structure to be filled by caller. + */ +AVMasteringDisplayMetadata *av_mastering_display_metadata_create_side_data(AVFrame *frame); + +/** + * Content light level needed by to transmit HDR over HDMI (CTA-861.3). + * + * To be used as payload of a AVFrameSideData or AVPacketSideData with the + * appropriate type. + * + * @note The struct should be allocated with av_content_light_metadata_alloc() + * and its size is not a part of the public ABI. + */ +typedef struct AVContentLightMetadata { + /** + * Max content light level (cd/m^2). + */ + unsigned MaxCLL; + + /** + * Max average light level per frame (cd/m^2). + */ + unsigned MaxFALL; +} AVContentLightMetadata; + +/** + * Allocate an AVContentLightMetadata structure and set its fields to + * default values. The resulting struct can be freed using av_freep(). + * + * @return An AVContentLightMetadata filled with default values or NULL + * on failure. + */ +AVContentLightMetadata *av_content_light_metadata_alloc(size_t *size); + +/** + * Allocate a complete AVContentLightMetadata and add it to the frame. + * + * @param frame The frame which side data is added to. + * + * @return The AVContentLightMetadata structure to be filled by caller. + */ +AVContentLightMetadata *av_content_light_metadata_create_side_data(AVFrame *frame); + +#endif /* AVUTIL_MASTERING_DISPLAY_METADATA_H */ diff --git a/media/ffvpx/libavutil/mathematics.c b/media/ffvpx/libavutil/mathematics.c new file mode 100644 index 0000000000..61aeb7c029 --- /dev/null +++ b/media/ffvpx/libavutil/mathematics.c @@ -0,0 +1,319 @@ +/* + * Copyright (c) 2005-2012 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * miscellaneous math routines and tables + */ + +#include <stdint.h> +#include <limits.h> + +#include "avutil.h" +#include "mathematics.h" +#include "libavutil/intmath.h" +#include "libavutil/common.h" +#include "avassert.h" + +/* Stein's binary GCD algorithm: + * https://en.wikipedia.org/wiki/Binary_GCD_algorithm */ +int64_t av_gcd(int64_t a, int64_t b) { + int za, zb, k; + int64_t u, v; + if (a == 0) + return b; + if (b == 0) + return a; + za = ff_ctzll(a); + zb = ff_ctzll(b); + k = FFMIN(za, zb); + u = llabs(a >> za); + v = llabs(b >> zb); + while (u != v) { + if (u > v) + FFSWAP(int64_t, v, u); + v -= u; + v >>= ff_ctzll(v); + } + return (uint64_t)u << k; +} + +int64_t av_rescale_rnd(int64_t a, int64_t b, int64_t c, enum AVRounding rnd) +{ + int64_t r = 0; + av_assert2(c > 0); + av_assert2(b >=0); + av_assert2((unsigned)(rnd&~AV_ROUND_PASS_MINMAX)<=5 && (rnd&~AV_ROUND_PASS_MINMAX)!=4); + + if (c <= 0 || b < 0 || !((unsigned)(rnd&~AV_ROUND_PASS_MINMAX)<=5 && (rnd&~AV_ROUND_PASS_MINMAX)!=4)) + return INT64_MIN; + + if (rnd & AV_ROUND_PASS_MINMAX) { + if (a == INT64_MIN || a == INT64_MAX) + return a; + rnd -= AV_ROUND_PASS_MINMAX; + } + + if (a < 0) + return -(uint64_t)av_rescale_rnd(-FFMAX(a, -INT64_MAX), b, c, rnd ^ ((rnd >> 1) & 1)); + + if (rnd == AV_ROUND_NEAR_INF) + r = c / 2; + else if (rnd & 1) + r = c - 1; + + if (b <= INT_MAX && c <= INT_MAX) { + if (a <= INT_MAX) + return (a * b + r) / c; + else { + int64_t ad = a / c; + int64_t a2 = (a % c * b + r) / c; + if (ad >= INT32_MAX && b && ad > (INT64_MAX - a2) / b) + return INT64_MIN; + return ad * b + a2; + } + } else { +#if 1 + uint64_t a0 = a & 0xFFFFFFFF; + uint64_t a1 = a >> 32; + uint64_t b0 = b & 0xFFFFFFFF; + uint64_t b1 = b >> 32; + uint64_t t1 = a0 * b1 + a1 * b0; + uint64_t t1a = t1 << 32; + int i; + + a0 = a0 * b0 + t1a; + a1 = a1 * b1 + (t1 >> 32) + (a0 < t1a); + a0 += r; + a1 += a0 < r; + + for (i = 63; i >= 0; i--) { + a1 += a1 + ((a0 >> i) & 1); + t1 += t1; + if (c <= a1) { + a1 -= c; + t1++; + } + } + if (t1 > INT64_MAX) + return INT64_MIN; + return t1; +#else + /* reference code doing (a*b + r) / c, requires libavutil/integer.h */ + AVInteger ai; + ai = av_mul_i(av_int2i(a), av_int2i(b)); + ai = av_add_i(ai, av_int2i(r)); + + return av_i2int(av_div_i(ai, av_int2i(c))); +#endif + } +} + +int64_t av_rescale(int64_t a, int64_t b, int64_t c) +{ + return av_rescale_rnd(a, b, c, AV_ROUND_NEAR_INF); +} + +int64_t av_rescale_q_rnd(int64_t a, AVRational bq, AVRational cq, + enum AVRounding rnd) +{ + int64_t b = bq.num * (int64_t)cq.den; + int64_t c = cq.num * (int64_t)bq.den; + return av_rescale_rnd(a, b, c, rnd); +} + +int64_t av_rescale_q(int64_t a, AVRational bq, AVRational cq) +{ + return av_rescale_q_rnd(a, bq, cq, AV_ROUND_NEAR_INF); +} + +int av_compare_ts(int64_t ts_a, AVRational tb_a, int64_t ts_b, AVRational tb_b) +{ + int64_t a = tb_a.num * (int64_t)tb_b.den; + int64_t b = tb_b.num * (int64_t)tb_a.den; + if ((FFABS64U(ts_a)|a|FFABS64U(ts_b)|b) <= INT_MAX) + return (ts_a*a > ts_b*b) - (ts_a*a < ts_b*b); + if (av_rescale_rnd(ts_a, a, b, AV_ROUND_DOWN) < ts_b) + return -1; + if (av_rescale_rnd(ts_b, b, a, AV_ROUND_DOWN) < ts_a) + return 1; + return 0; +} + +int64_t av_compare_mod(uint64_t a, uint64_t b, uint64_t mod) +{ + int64_t c = (a - b) & (mod - 1); + if (c > (mod >> 1)) + c -= mod; + return c; +} + +int64_t av_rescale_delta(AVRational in_tb, int64_t in_ts, AVRational fs_tb, int duration, int64_t *last, AVRational out_tb){ + int64_t a, b, this; + + av_assert0(in_ts != AV_NOPTS_VALUE); + av_assert0(duration >= 0); + + if (*last == AV_NOPTS_VALUE || !duration || in_tb.num*(int64_t)out_tb.den <= out_tb.num*(int64_t)in_tb.den) { +simple_round: + *last = av_rescale_q(in_ts, in_tb, fs_tb) + duration; + return av_rescale_q(in_ts, in_tb, out_tb); + } + + a = av_rescale_q_rnd(2*in_ts-1, in_tb, fs_tb, AV_ROUND_DOWN) >>1; + b = (av_rescale_q_rnd(2*in_ts+1, in_tb, fs_tb, AV_ROUND_UP )+1)>>1; + if (*last < 2*a - b || *last > 2*b - a) + goto simple_round; + + this = av_clip64(*last, a, b); + *last = this + duration; + + return av_rescale_q(this, fs_tb, out_tb); +} + +int64_t av_add_stable(AVRational ts_tb, int64_t ts, AVRational inc_tb, int64_t inc) +{ + int64_t m, d; + + if (inc != 1) + inc_tb = av_mul_q(inc_tb, (AVRational) {inc, 1}); + + m = inc_tb.num * (int64_t)ts_tb.den; + d = inc_tb.den * (int64_t)ts_tb.num; + + if (m % d == 0 && ts <= INT64_MAX - m / d) + return ts + m / d; + if (m < d) + return ts; + + { + int64_t old = av_rescale_q(ts, ts_tb, inc_tb); + int64_t old_ts = av_rescale_q(old, inc_tb, ts_tb); + + if (old == INT64_MAX || old == AV_NOPTS_VALUE || old_ts == AV_NOPTS_VALUE) + return ts; + + return av_sat_add64(av_rescale_q(old + 1, inc_tb, ts_tb), ts - old_ts); + } +} + +static inline double eval_poly(const double *coeff, int size, double x) { + double sum = coeff[size-1]; + int i; + for (i = size-2; i >= 0; --i) { + sum *= x; + sum += coeff[i]; + } + return sum; +} + +/** + * 0th order modified bessel function of the first kind. + * Algorithm taken from the Boost project, source: + * https://searchcode.com/codesearch/view/14918379/ + * Use, modification and distribution are subject to the + * Boost Software License, Version 1.0 (see notice below). + * Boost Software License - Version 1.0 - August 17th, 2003 +Permission is hereby granted, free of charge, to any person or organization +obtaining a copy of the software and accompanying documentation covered by +this license (the "Software") to use, reproduce, display, distribute, +execute, and transmit the Software, and to prepare derivative works of the +Software, and to permit third-parties to whom the Software is furnished to +do so, all subject to the following: + +The copyright notices in the Software and this entire statement, including +the above license grant, this restriction and the following disclaimer, +must be included in all copies of the Software, in whole or in part, and +all derivative works of the Software, unless such copies or derivative +works are solely in the form of machine-executable object code generated by +a source language processor. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT +SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE +FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE, +ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER +DEALINGS IN THE SOFTWARE. + */ + +double av_bessel_i0(double x) { +// Modified Bessel function of the first kind of order zero +// minimax rational approximations on intervals, see +// Blair and Edwards, Chalk River Report AECL-4928, 1974 + static const double p1[] = { + -2.2335582639474375249e+15, + -5.5050369673018427753e+14, + -3.2940087627407749166e+13, + -8.4925101247114157499e+11, + -1.1912746104985237192e+10, + -1.0313066708737980747e+08, + -5.9545626019847898221e+05, + -2.4125195876041896775e+03, + -7.0935347449210549190e+00, + -1.5453977791786851041e-02, + -2.5172644670688975051e-05, + -3.0517226450451067446e-08, + -2.6843448573468483278e-11, + -1.5982226675653184646e-14, + -5.2487866627945699800e-18, + }; + static const double q1[] = { + -2.2335582639474375245e+15, + 7.8858692566751002988e+12, + -1.2207067397808979846e+10, + 1.0377081058062166144e+07, + -4.8527560179962773045e+03, + 1.0, + }; + static const double p2[] = { + -2.2210262233306573296e-04, + 1.3067392038106924055e-02, + -4.4700805721174453923e-01, + 5.5674518371240761397e+00, + -2.3517945679239481621e+01, + 3.1611322818701131207e+01, + -9.6090021968656180000e+00, + }; + static const double q2[] = { + -5.5194330231005480228e-04, + 3.2547697594819615062e-02, + -1.1151759188741312645e+00, + 1.3982595353892851542e+01, + -6.0228002066743340583e+01, + 8.5539563258012929600e+01, + -3.1446690275135491500e+01, + 1.0, + }; + double y, r, factor; + if (x == 0) + return 1.0; + x = fabs(x); + if (x <= 15) { + y = x * x; + return eval_poly(p1, FF_ARRAY_ELEMS(p1), y) / eval_poly(q1, FF_ARRAY_ELEMS(q1), y); + } + else { + y = 1 / x - 1.0 / 15; + r = eval_poly(p2, FF_ARRAY_ELEMS(p2), y) / eval_poly(q2, FF_ARRAY_ELEMS(q2), y); + factor = exp(x) / sqrt(x); + return factor * r; + } +} diff --git a/media/ffvpx/libavutil/mathematics.h b/media/ffvpx/libavutil/mathematics.h new file mode 100644 index 0000000000..e213bab68c --- /dev/null +++ b/media/ffvpx/libavutil/mathematics.h @@ -0,0 +1,300 @@ +/* + * copyright (c) 2005-2012 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @addtogroup lavu_math + * Mathematical utilities for working with timestamp and time base. + */ + +#ifndef AVUTIL_MATHEMATICS_H +#define AVUTIL_MATHEMATICS_H + +#include <stdint.h> +#include <math.h> +#include "attributes.h" +#include "rational.h" +#include "intfloat.h" + +#ifndef M_E +#define M_E 2.7182818284590452354 /* e */ +#endif +#ifndef M_Ef +#define M_Ef 2.7182818284590452354f /* e */ +#endif +#ifndef M_LN2 +#define M_LN2 0.69314718055994530942 /* log_e 2 */ +#endif +#ifndef M_LN2f +#define M_LN2f 0.69314718055994530942f /* log_e 2 */ +#endif +#ifndef M_LN10 +#define M_LN10 2.30258509299404568402 /* log_e 10 */ +#endif +#ifndef M_LN10f +#define M_LN10f 2.30258509299404568402f /* log_e 10 */ +#endif +#ifndef M_LOG2_10 +#define M_LOG2_10 3.32192809488736234787 /* log_2 10 */ +#endif +#ifndef M_LOG2_10f +#define M_LOG2_10f 3.32192809488736234787f /* log_2 10 */ +#endif +#ifndef M_PHI +#define M_PHI 1.61803398874989484820 /* phi / golden ratio */ +#endif +#ifndef M_PHIf +#define M_PHIf 1.61803398874989484820f /* phi / golden ratio */ +#endif +#ifndef M_PI +#define M_PI 3.14159265358979323846 /* pi */ +#endif +#ifndef M_PIf +#define M_PIf 3.14159265358979323846f /* pi */ +#endif +#ifndef M_PI_2 +#define M_PI_2 1.57079632679489661923 /* pi/2 */ +#endif +#ifndef M_PI_2f +#define M_PI_2f 1.57079632679489661923f /* pi/2 */ +#endif +#ifndef M_PI_4 +#define M_PI_4 0.78539816339744830962 /* pi/4 */ +#endif +#ifndef M_PI_4f +#define M_PI_4f 0.78539816339744830962f /* pi/4 */ +#endif +#ifndef M_1_PI +#define M_1_PI 0.31830988618379067154 /* 1/pi */ +#endif +#ifndef M_1_PIf +#define M_1_PIf 0.31830988618379067154f /* 1/pi */ +#endif +#ifndef M_2_PI +#define M_2_PI 0.63661977236758134308 /* 2/pi */ +#endif +#ifndef M_2_PIf +#define M_2_PIf 0.63661977236758134308f /* 2/pi */ +#endif +#ifndef M_2_SQRTPI +#define M_2_SQRTPI 1.12837916709551257390 /* 2/sqrt(pi) */ +#endif +#ifndef M_2_SQRTPIf +#define M_2_SQRTPIf 1.12837916709551257390f /* 2/sqrt(pi) */ +#endif +#ifndef M_SQRT1_2 +#define M_SQRT1_2 0.70710678118654752440 /* 1/sqrt(2) */ +#endif +#ifndef M_SQRT1_2f +#define M_SQRT1_2f 0.70710678118654752440f /* 1/sqrt(2) */ +#endif +#ifndef M_SQRT2 +#define M_SQRT2 1.41421356237309504880 /* sqrt(2) */ +#endif +#ifndef M_SQRT2f +#define M_SQRT2f 1.41421356237309504880f /* sqrt(2) */ +#endif +#ifndef NAN +#define NAN av_int2float(0x7fc00000) +#endif +#ifndef INFINITY +#define INFINITY av_int2float(0x7f800000) +#endif + +/** + * @addtogroup lavu_math + * + * @{ + */ + +/** + * Rounding methods. + */ +enum AVRounding { + AV_ROUND_ZERO = 0, ///< Round toward zero. + AV_ROUND_INF = 1, ///< Round away from zero. + AV_ROUND_DOWN = 2, ///< Round toward -infinity. + AV_ROUND_UP = 3, ///< Round toward +infinity. + AV_ROUND_NEAR_INF = 5, ///< Round to nearest and halfway cases away from zero. + /** + * Flag telling rescaling functions to pass `INT64_MIN`/`MAX` through + * unchanged, avoiding special cases for #AV_NOPTS_VALUE. + * + * Unlike other values of the enumeration AVRounding, this value is a + * bitmask that must be used in conjunction with another value of the + * enumeration through a bitwise OR, in order to set behavior for normal + * cases. + * + * @code{.c} + * av_rescale_rnd(3, 1, 2, AV_ROUND_UP | AV_ROUND_PASS_MINMAX); + * // Rescaling 3: + * // Calculating 3 * 1 / 2 + * // 3 / 2 is rounded up to 2 + * // => 2 + * + * av_rescale_rnd(AV_NOPTS_VALUE, 1, 2, AV_ROUND_UP | AV_ROUND_PASS_MINMAX); + * // Rescaling AV_NOPTS_VALUE: + * // AV_NOPTS_VALUE == INT64_MIN + * // AV_NOPTS_VALUE is passed through + * // => AV_NOPTS_VALUE + * @endcode + */ + AV_ROUND_PASS_MINMAX = 8192, +}; + +/** + * Compute the greatest common divisor of two integer operands. + * + * @param a Operand + * @param b Operand + * @return GCD of a and b up to sign; if a >= 0 and b >= 0, return value is >= 0; + * if a == 0 and b == 0, returns 0. + */ +int64_t av_const av_gcd(int64_t a, int64_t b); + +/** + * Rescale a 64-bit integer with rounding to nearest. + * + * The operation is mathematically equivalent to `a * b / c`, but writing that + * directly can overflow. + * + * This function is equivalent to av_rescale_rnd() with #AV_ROUND_NEAR_INF. + * + * @see av_rescale_rnd(), av_rescale_q(), av_rescale_q_rnd() + */ +int64_t av_rescale(int64_t a, int64_t b, int64_t c) av_const; + +/** + * Rescale a 64-bit integer with specified rounding. + * + * The operation is mathematically equivalent to `a * b / c`, but writing that + * directly can overflow, and does not support different rounding methods. + * If the result is not representable then INT64_MIN is returned. + * + * @see av_rescale(), av_rescale_q(), av_rescale_q_rnd() + */ +int64_t av_rescale_rnd(int64_t a, int64_t b, int64_t c, enum AVRounding rnd) av_const; + +/** + * Rescale a 64-bit integer by 2 rational numbers. + * + * The operation is mathematically equivalent to `a * bq / cq`. + * + * This function is equivalent to av_rescale_q_rnd() with #AV_ROUND_NEAR_INF. + * + * @see av_rescale(), av_rescale_rnd(), av_rescale_q_rnd() + */ +int64_t av_rescale_q(int64_t a, AVRational bq, AVRational cq) av_const; + +/** + * Rescale a 64-bit integer by 2 rational numbers with specified rounding. + * + * The operation is mathematically equivalent to `a * bq / cq`. + * + * @see av_rescale(), av_rescale_rnd(), av_rescale_q() + */ +int64_t av_rescale_q_rnd(int64_t a, AVRational bq, AVRational cq, + enum AVRounding rnd) av_const; + +/** + * Compare two timestamps each in its own time base. + * + * @return One of the following values: + * - -1 if `ts_a` is before `ts_b` + * - 1 if `ts_a` is after `ts_b` + * - 0 if they represent the same position + * + * @warning + * The result of the function is undefined if one of the timestamps is outside + * the `int64_t` range when represented in the other's timebase. + */ +int av_compare_ts(int64_t ts_a, AVRational tb_a, int64_t ts_b, AVRational tb_b); + +/** + * Compare the remainders of two integer operands divided by a common divisor. + * + * In other words, compare the least significant `log2(mod)` bits of integers + * `a` and `b`. + * + * @code{.c} + * av_compare_mod(0x11, 0x02, 0x10) < 0 // since 0x11 % 0x10 (0x1) < 0x02 % 0x10 (0x2) + * av_compare_mod(0x11, 0x02, 0x20) > 0 // since 0x11 % 0x20 (0x11) > 0x02 % 0x20 (0x02) + * @endcode + * + * @param a Operand + * @param b Operand + * @param mod Divisor; must be a power of 2 + * @return + * - a negative value if `a % mod < b % mod` + * - a positive value if `a % mod > b % mod` + * - zero if `a % mod == b % mod` + */ +int64_t av_compare_mod(uint64_t a, uint64_t b, uint64_t mod); + +/** + * Rescale a timestamp while preserving known durations. + * + * This function is designed to be called per audio packet to scale the input + * timestamp to a different time base. Compared to a simple av_rescale_q() + * call, this function is robust against possible inconsistent frame durations. + * + * The `last` parameter is a state variable that must be preserved for all + * subsequent calls for the same stream. For the first call, `*last` should be + * initialized to #AV_NOPTS_VALUE. + * + * @param[in] in_tb Input time base + * @param[in] in_ts Input timestamp + * @param[in] fs_tb Duration time base; typically this is finer-grained + * (greater) than `in_tb` and `out_tb` + * @param[in] duration Duration till the next call to this function (i.e. + * duration of the current packet/frame) + * @param[in,out] last Pointer to a timestamp expressed in terms of + * `fs_tb`, acting as a state variable + * @param[in] out_tb Output timebase + * @return Timestamp expressed in terms of `out_tb` + * + * @note In the context of this function, "duration" is in term of samples, not + * seconds. + */ +int64_t av_rescale_delta(AVRational in_tb, int64_t in_ts, AVRational fs_tb, int duration, int64_t *last, AVRational out_tb); + +/** + * Add a value to a timestamp. + * + * This function guarantees that when the same value is repeatly added that + * no accumulation of rounding errors occurs. + * + * @param[in] ts Input timestamp + * @param[in] ts_tb Input timestamp time base + * @param[in] inc Value to be added + * @param[in] inc_tb Time base of `inc` + */ +int64_t av_add_stable(AVRational ts_tb, int64_t ts, AVRational inc_tb, int64_t inc); + +/** + * 0th order modified bessel function of the first kind. + */ +double av_bessel_i0(double x); + +/** + * @} + */ + +#endif /* AVUTIL_MATHEMATICS_H */ diff --git a/media/ffvpx/libavutil/mem.c b/media/ffvpx/libavutil/mem.c new file mode 100644 index 0000000000..36b8940a0c --- /dev/null +++ b/media/ffvpx/libavutil/mem.c @@ -0,0 +1,568 @@ +/* + * default memory allocator for libavutil + * Copyright (c) 2002 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * default memory allocator for libavutil + */ + +#define _XOPEN_SOURCE 600 + +#include "config.h" + +#include <limits.h> +#include <stdint.h> +#include <stdlib.h> +#include <stdatomic.h> +#include <string.h> +#if HAVE_MALLOC_H +#include <malloc.h> +#endif + +#include "attributes.h" +#include "avassert.h" +#include "dynarray.h" +#include "error.h" +#include "internal.h" +#include "intreadwrite.h" +#include "macros.h" +#include "mem.h" + +#ifdef MALLOC_PREFIX + +#define malloc AV_JOIN(MALLOC_PREFIX, malloc) +#define memalign AV_JOIN(MALLOC_PREFIX, memalign) +#define posix_memalign AV_JOIN(MALLOC_PREFIX, posix_memalign) +#define realloc AV_JOIN(MALLOC_PREFIX, realloc) +#define free AV_JOIN(MALLOC_PREFIX, free) + +void *malloc(size_t size); +void *memalign(size_t align, size_t size); +int posix_memalign(void **ptr, size_t align, size_t size); +void *realloc(void *ptr, size_t size); +void free(void *ptr); + +#endif /* MALLOC_PREFIX */ + +#define ALIGN (HAVE_AVX512 ? 64 : (HAVE_AVX ? 32 : 16)) + +/* NOTE: if you want to override these functions with your own + * implementations (not recommended) you have to link libav* as + * dynamic libraries and remove -Wl,-Bsymbolic from the linker flags. + * Note that this will cost performance. */ + +static atomic_size_t max_alloc_size = ATOMIC_VAR_INIT(INT_MAX); + +void av_max_alloc(size_t max){ + atomic_store_explicit(&max_alloc_size, max, memory_order_relaxed); +} + +static int size_mult(size_t a, size_t b, size_t *r) +{ + size_t t; + +#if (!defined(__INTEL_COMPILER) && AV_GCC_VERSION_AT_LEAST(5,1)) || AV_HAS_BUILTIN(__builtin_mul_overflow) + if (__builtin_mul_overflow(a, b, &t)) + return AVERROR(EINVAL); +#else + t = a * b; + /* Hack inspired from glibc: don't try the division if nelem and elsize + * are both less than sqrt(SIZE_MAX). */ + if ((a | b) >= ((size_t)1 << (sizeof(size_t) * 4)) && a && t / a != b) + return AVERROR(EINVAL); +#endif + *r = t; + return 0; +} + +void *av_malloc(size_t size) +{ + void *ptr = NULL; + + if (size > atomic_load_explicit(&max_alloc_size, memory_order_relaxed)) + return NULL; + +#if HAVE_POSIX_MEMALIGN + if (size) //OS X on SDK 10.6 has a broken posix_memalign implementation + if (posix_memalign(&ptr, ALIGN, size)) + ptr = NULL; +#elif HAVE_ALIGNED_MALLOC + ptr = _aligned_malloc(size, ALIGN); +#elif HAVE_MEMALIGN +#ifndef __DJGPP__ + ptr = memalign(ALIGN, size); +#else + ptr = memalign(size, ALIGN); +#endif + /* Why 64? + * Indeed, we should align it: + * on 4 for 386 + * on 16 for 486 + * on 32 for 586, PPro - K6-III + * on 64 for K7 (maybe for P3 too). + * Because L1 and L2 caches are aligned on those values. + * But I don't want to code such logic here! + */ + /* Why 32? + * For AVX ASM. SSE / NEON needs only 16. + * Why not larger? Because I did not see a difference in benchmarks ... + */ + /* benchmarks with P3 + * memalign(64) + 1 3071, 3051, 3032 + * memalign(64) + 2 3051, 3032, 3041 + * memalign(64) + 4 2911, 2896, 2915 + * memalign(64) + 8 2545, 2554, 2550 + * memalign(64) + 16 2543, 2572, 2563 + * memalign(64) + 32 2546, 2545, 2571 + * memalign(64) + 64 2570, 2533, 2558 + * + * BTW, malloc seems to do 8-byte alignment by default here. + */ +#else + ptr = malloc(size); +#endif + if(!ptr && !size) { + size = 1; + ptr= av_malloc(1); + } +#if CONFIG_MEMORY_POISONING + if (ptr) + memset(ptr, FF_MEMORY_POISON, size); +#endif + return ptr; +} + +void *av_realloc(void *ptr, size_t size) +{ + void *ret; + if (size > atomic_load_explicit(&max_alloc_size, memory_order_relaxed)) + return NULL; + +#if HAVE_ALIGNED_MALLOC + ret = _aligned_realloc(ptr, size + !size, ALIGN); +#else + ret = realloc(ptr, size + !size); +#endif +#if CONFIG_MEMORY_POISONING + if (ret && !ptr) + memset(ret, FF_MEMORY_POISON, size); +#endif + return ret; +} + +void *av_realloc_f(void *ptr, size_t nelem, size_t elsize) +{ + size_t size; + void *r; + + if (size_mult(elsize, nelem, &size)) { + av_free(ptr); + return NULL; + } + r = av_realloc(ptr, size); + if (!r) + av_free(ptr); + return r; +} + +int av_reallocp(void *ptr, size_t size) +{ + void *val; + + if (!size) { + av_freep(ptr); + return 0; + } + + memcpy(&val, ptr, sizeof(val)); + val = av_realloc(val, size); + + if (!val) { + av_freep(ptr); + return AVERROR(ENOMEM); + } + + memcpy(ptr, &val, sizeof(val)); + return 0; +} + +void *av_malloc_array(size_t nmemb, size_t size) +{ + size_t result; + if (size_mult(nmemb, size, &result) < 0) + return NULL; + return av_malloc(result); +} + +void *av_realloc_array(void *ptr, size_t nmemb, size_t size) +{ + size_t result; + if (size_mult(nmemb, size, &result) < 0) + return NULL; + return av_realloc(ptr, result); +} + +int av_reallocp_array(void *ptr, size_t nmemb, size_t size) +{ + void *val; + + memcpy(&val, ptr, sizeof(val)); + val = av_realloc_f(val, nmemb, size); + memcpy(ptr, &val, sizeof(val)); + if (!val && nmemb && size) + return AVERROR(ENOMEM); + + return 0; +} + +void av_free(void *ptr) +{ +#if HAVE_ALIGNED_MALLOC + _aligned_free(ptr); +#else + free(ptr); +#endif +} + +void av_freep(void *arg) +{ + void *val; + + memcpy(&val, arg, sizeof(val)); + memcpy(arg, &(void *){ NULL }, sizeof(val)); + av_free(val); +} + +void *av_mallocz(size_t size) +{ + void *ptr = av_malloc(size); + if (ptr) + memset(ptr, 0, size); + return ptr; +} + +void *av_calloc(size_t nmemb, size_t size) +{ + size_t result; + if (size_mult(nmemb, size, &result) < 0) + return NULL; + return av_mallocz(result); +} + +char *av_strdup(const char *s) +{ + char *ptr = NULL; + if (s) { + size_t len = strlen(s) + 1; + ptr = av_realloc(NULL, len); + if (ptr) + memcpy(ptr, s, len); + } + return ptr; +} + +char *av_strndup(const char *s, size_t len) +{ + char *ret = NULL, *end; + + if (!s) + return NULL; + + end = memchr(s, 0, len); + if (end) + len = end - s; + + ret = av_realloc(NULL, len + 1); + if (!ret) + return NULL; + + memcpy(ret, s, len); + ret[len] = 0; + return ret; +} + +void *av_memdup(const void *p, size_t size) +{ + void *ptr = NULL; + if (p) { + ptr = av_malloc(size); + if (ptr) + memcpy(ptr, p, size); + } + return ptr; +} + +int av_dynarray_add_nofree(void *tab_ptr, int *nb_ptr, void *elem) +{ + void **tab; + memcpy(&tab, tab_ptr, sizeof(tab)); + + FF_DYNARRAY_ADD(INT_MAX, sizeof(*tab), tab, *nb_ptr, { + tab[*nb_ptr] = elem; + memcpy(tab_ptr, &tab, sizeof(tab)); + }, { + return AVERROR(ENOMEM); + }); + return 0; +} + +void av_dynarray_add(void *tab_ptr, int *nb_ptr, void *elem) +{ + void **tab; + memcpy(&tab, tab_ptr, sizeof(tab)); + + FF_DYNARRAY_ADD(INT_MAX, sizeof(*tab), tab, *nb_ptr, { + tab[*nb_ptr] = elem; + memcpy(tab_ptr, &tab, sizeof(tab)); + }, { + *nb_ptr = 0; + av_freep(tab_ptr); + }); +} + +void *av_dynarray2_add(void **tab_ptr, int *nb_ptr, size_t elem_size, + const uint8_t *elem_data) +{ + uint8_t *tab_elem_data = NULL; + + FF_DYNARRAY_ADD(INT_MAX, elem_size, *tab_ptr, *nb_ptr, { + tab_elem_data = (uint8_t *)*tab_ptr + (*nb_ptr) * elem_size; + if (elem_data) + memcpy(tab_elem_data, elem_data, elem_size); + else if (CONFIG_MEMORY_POISONING) + memset(tab_elem_data, FF_MEMORY_POISON, elem_size); + }, { + av_freep(tab_ptr); + *nb_ptr = 0; + }); + return tab_elem_data; +} + +static void fill16(uint8_t *dst, int len) +{ + uint32_t v = AV_RN16(dst - 2); + + v |= v << 16; + + while (len >= 4) { + AV_WN32(dst, v); + dst += 4; + len -= 4; + } + + while (len--) { + *dst = dst[-2]; + dst++; + } +} + +static void fill24(uint8_t *dst, int len) +{ +#if HAVE_BIGENDIAN + uint32_t v = AV_RB24(dst - 3); + uint32_t a = v << 8 | v >> 16; + uint32_t b = v << 16 | v >> 8; + uint32_t c = v << 24 | v; +#else + uint32_t v = AV_RL24(dst - 3); + uint32_t a = v | v << 24; + uint32_t b = v >> 8 | v << 16; + uint32_t c = v >> 16 | v << 8; +#endif + + while (len >= 12) { + AV_WN32(dst, a); + AV_WN32(dst + 4, b); + AV_WN32(dst + 8, c); + dst += 12; + len -= 12; + } + + if (len >= 4) { + AV_WN32(dst, a); + dst += 4; + len -= 4; + } + + if (len >= 4) { + AV_WN32(dst, b); + dst += 4; + len -= 4; + } + + while (len--) { + *dst = dst[-3]; + dst++; + } +} + +static void fill32(uint8_t *dst, int len) +{ + uint32_t v = AV_RN32(dst - 4); + +#if HAVE_FAST_64BIT + uint64_t v2= v + ((uint64_t)v<<32); + while (len >= 32) { + AV_WN64(dst , v2); + AV_WN64(dst+ 8, v2); + AV_WN64(dst+16, v2); + AV_WN64(dst+24, v2); + dst += 32; + len -= 32; + } +#endif + + while (len >= 4) { + AV_WN32(dst, v); + dst += 4; + len -= 4; + } + + while (len--) { + *dst = dst[-4]; + dst++; + } +} + +void av_memcpy_backptr(uint8_t *dst, int back, int cnt) +{ + const uint8_t *src = &dst[-back]; + if (!back) + return; + + if (back == 1) { + memset(dst, *src, cnt); + } else if (back == 2) { + fill16(dst, cnt); + } else if (back == 3) { + fill24(dst, cnt); + } else if (back == 4) { + fill32(dst, cnt); + } else { + if (cnt >= 16) { + int blocklen = back; + while (cnt > blocklen) { + memcpy(dst, src, blocklen); + dst += blocklen; + cnt -= blocklen; + blocklen <<= 1; + } + memcpy(dst, src, cnt); + return; + } + if (cnt >= 8) { + AV_COPY32U(dst, src); + AV_COPY32U(dst + 4, src + 4); + src += 8; + dst += 8; + cnt -= 8; + } + if (cnt >= 4) { + AV_COPY32U(dst, src); + src += 4; + dst += 4; + cnt -= 4; + } + if (cnt >= 2) { + AV_COPY16U(dst, src); + src += 2; + dst += 2; + cnt -= 2; + } + if (cnt) + *dst = *src; + } +} + +void *av_fast_realloc(void *ptr, unsigned int *size, size_t min_size) +{ + size_t max_size; + + if (min_size <= *size) + return ptr; + + max_size = atomic_load_explicit(&max_alloc_size, memory_order_relaxed); + /* *size is an unsigned, so the real maximum is <= UINT_MAX. */ + max_size = FFMIN(max_size, UINT_MAX); + + if (min_size > max_size) { + *size = 0; + return NULL; + } + + min_size = FFMIN(max_size, FFMAX(min_size + min_size / 16 + 32, min_size)); + + ptr = av_realloc(ptr, min_size); + /* we could set this to the unmodified min_size but this is safer + * if the user lost the ptr and uses NULL now + */ + if (!ptr) + min_size = 0; + + *size = min_size; + + return ptr; +} + +static inline void fast_malloc(void *ptr, unsigned int *size, size_t min_size, int zero_realloc) +{ + size_t max_size; + void *val; + + memcpy(&val, ptr, sizeof(val)); + if (min_size <= *size) { + av_assert0(val || !min_size); + return; + } + + max_size = atomic_load_explicit(&max_alloc_size, memory_order_relaxed); + /* *size is an unsigned, so the real maximum is <= UINT_MAX. */ + max_size = FFMIN(max_size, UINT_MAX); + + if (min_size > max_size) { + av_freep(ptr); + *size = 0; + return; + } + min_size = FFMIN(max_size, FFMAX(min_size + min_size / 16 + 32, min_size)); + av_freep(ptr); + val = zero_realloc ? av_mallocz(min_size) : av_malloc(min_size); + memcpy(ptr, &val, sizeof(val)); + if (!val) + min_size = 0; + *size = min_size; + return; +} + +void av_fast_malloc(void *ptr, unsigned int *size, size_t min_size) +{ + fast_malloc(ptr, size, min_size, 0); +} + +void av_fast_mallocz(void *ptr, unsigned int *size, size_t min_size) +{ + fast_malloc(ptr, size, min_size, 1); +} + +int av_size_mult(size_t a, size_t b, size_t *r) +{ + return size_mult(a, b, r); +} diff --git a/media/ffvpx/libavutil/mem.h b/media/ffvpx/libavutil/mem.h new file mode 100644 index 0000000000..be697408a2 --- /dev/null +++ b/media/ffvpx/libavutil/mem.h @@ -0,0 +1,607 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_mem + * Memory handling functions + */ + +#ifndef AVUTIL_MEM_H +#define AVUTIL_MEM_H + +#include <stddef.h> +#include <stdint.h> + +#include "attributes.h" + +/** + * @addtogroup lavu_mem + * Utilities for manipulating memory. + * + * FFmpeg has several applications of memory that are not required of a typical + * program. For example, the computing-heavy components like video decoding and + * encoding can be sped up significantly through the use of aligned memory. + * + * However, for each of FFmpeg's applications of memory, there might not be a + * recognized or standardized API for that specific use. Memory alignment, for + * instance, varies wildly depending on operating systems, architectures, and + * compilers. Hence, this component of @ref libavutil is created to make + * dealing with memory consistently possible on all platforms. + * + * @{ + */ + +/** + * @defgroup lavu_mem_attrs Function Attributes + * Function attributes applicable to memory handling functions. + * + * These function attributes can help compilers emit more useful warnings, or + * generate better code. + * @{ + */ + +/** + * @def av_malloc_attrib + * Function attribute denoting a malloc-like function. + * + * @see <a href="https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-g_t_0040code_007bmalloc_007d-function-attribute-3251">Function attribute `malloc` in GCC's documentation</a> + */ + +#if AV_GCC_VERSION_AT_LEAST(3,1) + #define av_malloc_attrib __attribute__((__malloc__)) +#else + #define av_malloc_attrib +#endif + +/** + * @def av_alloc_size(...) + * Function attribute used on a function that allocates memory, whose size is + * given by the specified parameter(s). + * + * @code{.c} + * void *av_malloc(size_t size) av_alloc_size(1); + * void *av_calloc(size_t nmemb, size_t size) av_alloc_size(1, 2); + * @endcode + * + * @param ... One or two parameter indexes, separated by a comma + * + * @see <a href="https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-g_t_0040code_007balloc_005fsize_007d-function-attribute-3220">Function attribute `alloc_size` in GCC's documentation</a> + */ + +#if AV_GCC_VERSION_AT_LEAST(4,3) + #define av_alloc_size(...) __attribute__((alloc_size(__VA_ARGS__))) +#else + #define av_alloc_size(...) +#endif + +/** + * @} + */ + +/** + * @defgroup lavu_mem_funcs Heap Management + * Functions responsible for allocating, freeing, and copying memory. + * + * All memory allocation functions have a built-in upper limit of `INT_MAX` + * bytes. This may be changed with av_max_alloc(), although exercise extreme + * caution when doing so. + * + * @{ + */ + +/** + * Allocate a memory block with alignment suitable for all memory accesses + * (including vectors if available on the CPU). + * + * @param size Size in bytes for the memory block to be allocated + * @return Pointer to the allocated block, or `NULL` if the block cannot + * be allocated + * @see av_mallocz() + */ +void *av_malloc(size_t size) av_malloc_attrib av_alloc_size(1) __attribute__((visibility("default"))); + +/** + * Allocate a memory block with alignment suitable for all memory accesses + * (including vectors if available on the CPU) and zero all the bytes of the + * block. + * + * @param size Size in bytes for the memory block to be allocated + * @return Pointer to the allocated block, or `NULL` if it cannot be allocated + * @see av_malloc() + */ +void *av_mallocz(size_t size) av_malloc_attrib av_alloc_size(1); + +/** + * Allocate a memory block for an array with av_malloc(). + * + * The allocated memory will have size `size * nmemb` bytes. + * + * @param nmemb Number of element + * @param size Size of a single element + * @return Pointer to the allocated block, or `NULL` if the block cannot + * be allocated + * @see av_malloc() + */ +av_alloc_size(1, 2) void *av_malloc_array(size_t nmemb, size_t size); + +/** + * Allocate a memory block for an array with av_mallocz(). + * + * The allocated memory will have size `size * nmemb` bytes. + * + * @param nmemb Number of elements + * @param size Size of the single element + * @return Pointer to the allocated block, or `NULL` if the block cannot + * be allocated + * + * @see av_mallocz() + * @see av_malloc_array() + */ +void *av_calloc(size_t nmemb, size_t size) av_malloc_attrib av_alloc_size(1, 2); + +/** + * Allocate, reallocate, or free a block of memory. + * + * If `ptr` is `NULL` and `size` > 0, allocate a new block. Otherwise, expand or + * shrink that block of memory according to `size`. + * + * @param ptr Pointer to a memory block already allocated with + * av_realloc() or `NULL` + * @param size Size in bytes of the memory block to be allocated or + * reallocated + * + * @return Pointer to a newly-reallocated block or `NULL` if the block + * cannot be reallocated + * + * @warning Unlike av_malloc(), the returned pointer is not guaranteed to be + * correctly aligned. The returned pointer must be freed after even + * if size is zero. + * @see av_fast_realloc() + * @see av_reallocp() + */ +void *av_realloc(void *ptr, size_t size) av_alloc_size(2); + +/** + * Allocate, reallocate, or free a block of memory through a pointer to a + * pointer. + * + * If `*ptr` is `NULL` and `size` > 0, allocate a new block. If `size` is + * zero, free the memory block pointed to by `*ptr`. Otherwise, expand or + * shrink that block of memory according to `size`. + * + * @param[in,out] ptr Pointer to a pointer to a memory block already allocated + * with av_realloc(), or a pointer to `NULL`. The pointer + * is updated on success, or freed on failure. + * @param[in] size Size in bytes for the memory block to be allocated or + * reallocated + * + * @return Zero on success, an AVERROR error code on failure + * + * @warning Unlike av_malloc(), the allocated memory is not guaranteed to be + * correctly aligned. + */ +av_warn_unused_result +int av_reallocp(void *ptr, size_t size); + +/** + * Allocate, reallocate, or free a block of memory. + * + * This function does the same thing as av_realloc(), except: + * - It takes two size arguments and allocates `nelem * elsize` bytes, + * after checking the result of the multiplication for integer overflow. + * - It frees the input block in case of failure, thus avoiding the memory + * leak with the classic + * @code{.c} + * buf = realloc(buf); + * if (!buf) + * return -1; + * @endcode + * pattern. + */ +void *av_realloc_f(void *ptr, size_t nelem, size_t elsize); + +/** + * Allocate, reallocate, or free an array. + * + * If `ptr` is `NULL` and `nmemb` > 0, allocate a new block. + * + * @param ptr Pointer to a memory block already allocated with + * av_realloc() or `NULL` + * @param nmemb Number of elements in the array + * @param size Size of the single element of the array + * + * @return Pointer to a newly-reallocated block or NULL if the block + * cannot be reallocated + * + * @warning Unlike av_malloc(), the allocated memory is not guaranteed to be + * correctly aligned. The returned pointer must be freed after even if + * nmemb is zero. + * @see av_reallocp_array() + */ +av_alloc_size(2, 3) void *av_realloc_array(void *ptr, size_t nmemb, size_t size); + +/** + * Allocate, reallocate an array through a pointer to a pointer. + * + * If `*ptr` is `NULL` and `nmemb` > 0, allocate a new block. + * + * @param[in,out] ptr Pointer to a pointer to a memory block already + * allocated with av_realloc(), or a pointer to `NULL`. + * The pointer is updated on success, or freed on failure. + * @param[in] nmemb Number of elements + * @param[in] size Size of the single element + * + * @return Zero on success, an AVERROR error code on failure + * + * @warning Unlike av_malloc(), the allocated memory is not guaranteed to be + * correctly aligned. *ptr must be freed after even if nmemb is zero. + */ +int av_reallocp_array(void *ptr, size_t nmemb, size_t size); + +/** + * Reallocate the given buffer if it is not large enough, otherwise do nothing. + * + * If the given buffer is `NULL`, then a new uninitialized buffer is allocated. + * + * If the given buffer is not large enough, and reallocation fails, `NULL` is + * returned and `*size` is set to 0, but the original buffer is not changed or + * freed. + * + * A typical use pattern follows: + * + * @code{.c} + * uint8_t *buf = ...; + * uint8_t *new_buf = av_fast_realloc(buf, ¤t_size, size_needed); + * if (!new_buf) { + * // Allocation failed; clean up original buffer + * av_freep(&buf); + * return AVERROR(ENOMEM); + * } + * @endcode + * + * @param[in,out] ptr Already allocated buffer, or `NULL` + * @param[in,out] size Pointer to the size of buffer `ptr`. `*size` is + * updated to the new allocated size, in particular 0 + * in case of failure. + * @param[in] min_size Desired minimal size of buffer `ptr` + * @return `ptr` if the buffer is large enough, a pointer to newly reallocated + * buffer if the buffer was not large enough, or `NULL` in case of + * error + * @see av_realloc() + * @see av_fast_malloc() + */ +void *av_fast_realloc(void *ptr, unsigned int *size, size_t min_size); + +/** + * Allocate a buffer, reusing the given one if large enough. + * + * Contrary to av_fast_realloc(), the current buffer contents might not be + * preserved and on error the old buffer is freed, thus no special handling to + * avoid memleaks is necessary. + * + * `*ptr` is allowed to be `NULL`, in which case allocation always happens if + * `size_needed` is greater than 0. + * + * @code{.c} + * uint8_t *buf = ...; + * av_fast_malloc(&buf, ¤t_size, size_needed); + * if (!buf) { + * // Allocation failed; buf already freed + * return AVERROR(ENOMEM); + * } + * @endcode + * + * @param[in,out] ptr Pointer to pointer to an already allocated buffer. + * `*ptr` will be overwritten with pointer to new + * buffer on success or `NULL` on failure + * @param[in,out] size Pointer to the size of buffer `*ptr`. `*size` is + * updated to the new allocated size, in particular 0 + * in case of failure. + * @param[in] min_size Desired minimal size of buffer `*ptr` + * @see av_realloc() + * @see av_fast_mallocz() + */ +void av_fast_malloc(void *ptr, unsigned int *size, size_t min_size); + +/** + * Allocate and clear a buffer, reusing the given one if large enough. + * + * Like av_fast_malloc(), but all newly allocated space is initially cleared. + * Reused buffer is not cleared. + * + * `*ptr` is allowed to be `NULL`, in which case allocation always happens if + * `size_needed` is greater than 0. + * + * @param[in,out] ptr Pointer to pointer to an already allocated buffer. + * `*ptr` will be overwritten with pointer to new + * buffer on success or `NULL` on failure + * @param[in,out] size Pointer to the size of buffer `*ptr`. `*size` is + * updated to the new allocated size, in particular 0 + * in case of failure. + * @param[in] min_size Desired minimal size of buffer `*ptr` + * @see av_fast_malloc() + */ +void av_fast_mallocz(void *ptr, unsigned int *size, size_t min_size); + +/** + * Free a memory block which has been allocated with a function of av_malloc() + * or av_realloc() family. + * + * @param ptr Pointer to the memory block which should be freed. + * + * @note `ptr = NULL` is explicitly allowed. + * @note It is recommended that you use av_freep() instead, to prevent leaving + * behind dangling pointers. + * @see av_freep() + */ +void av_free(void *ptr); + +/** + * Free a memory block which has been allocated with a function of av_malloc() + * or av_realloc() family, and set the pointer pointing to it to `NULL`. + * + * @code{.c} + * uint8_t *buf = av_malloc(16); + * av_free(buf); + * // buf now contains a dangling pointer to freed memory, and accidental + * // dereference of buf will result in a use-after-free, which may be a + * // security risk. + * + * uint8_t *buf = av_malloc(16); + * av_freep(&buf); + * // buf is now NULL, and accidental dereference will only result in a + * // NULL-pointer dereference. + * @endcode + * + * @param ptr Pointer to the pointer to the memory block which should be freed + * @note `*ptr = NULL` is safe and leads to no action. + * @see av_free() + */ +void av_freep(void *ptr); + +/** + * Duplicate a string. + * + * @param s String to be duplicated + * @return Pointer to a newly-allocated string containing a + * copy of `s` or `NULL` if the string cannot be allocated + * @see av_strndup() + */ +char *av_strdup(const char *s) av_malloc_attrib; + +/** + * Duplicate a substring of a string. + * + * @param s String to be duplicated + * @param len Maximum length of the resulting string (not counting the + * terminating byte) + * @return Pointer to a newly-allocated string containing a + * substring of `s` or `NULL` if the string cannot be allocated + */ +char *av_strndup(const char *s, size_t len) av_malloc_attrib; + +/** + * Duplicate a buffer with av_malloc(). + * + * @param p Buffer to be duplicated + * @param size Size in bytes of the buffer copied + * @return Pointer to a newly allocated buffer containing a + * copy of `p` or `NULL` if the buffer cannot be allocated + */ +void *av_memdup(const void *p, size_t size); + +/** + * Overlapping memcpy() implementation. + * + * @param dst Destination buffer + * @param back Number of bytes back to start copying (i.e. the initial size of + * the overlapping window); must be > 0 + * @param cnt Number of bytes to copy; must be >= 0 + * + * @note `cnt > back` is valid, this will copy the bytes we just copied, + * thus creating a repeating pattern with a period length of `back`. + */ +void av_memcpy_backptr(uint8_t *dst, int back, int cnt); + +/** + * @} + */ + +/** + * @defgroup lavu_mem_dynarray Dynamic Array + * + * Utilities to make an array grow when needed. + * + * Sometimes, the programmer would want to have an array that can grow when + * needed. The libavutil dynamic array utilities fill that need. + * + * libavutil supports two systems of appending elements onto a dynamically + * allocated array, the first one storing the pointer to the value in the + * array, and the second storing the value directly. In both systems, the + * caller is responsible for maintaining a variable containing the length of + * the array, as well as freeing of the array after use. + * + * The first system stores pointers to values in a block of dynamically + * allocated memory. Since only pointers are stored, the function does not need + * to know the size of the type. Both av_dynarray_add() and + * av_dynarray_add_nofree() implement this system. + * + * @code + * type **array = NULL; //< an array of pointers to values + * int nb = 0; //< a variable to keep track of the length of the array + * + * type to_be_added = ...; + * type to_be_added2 = ...; + * + * av_dynarray_add(&array, &nb, &to_be_added); + * if (nb == 0) + * return AVERROR(ENOMEM); + * + * av_dynarray_add(&array, &nb, &to_be_added2); + * if (nb == 0) + * return AVERROR(ENOMEM); + * + * // Now: + * // nb == 2 + * // &to_be_added == array[0] + * // &to_be_added2 == array[1] + * + * av_freep(&array); + * @endcode + * + * The second system stores the value directly in a block of memory. As a + * result, the function has to know the size of the type. av_dynarray2_add() + * implements this mechanism. + * + * @code + * type *array = NULL; //< an array of values + * int nb = 0; //< a variable to keep track of the length of the array + * + * type to_be_added = ...; + * type to_be_added2 = ...; + * + * type *addr = av_dynarray2_add((void **)&array, &nb, sizeof(*array), NULL); + * if (!addr) + * return AVERROR(ENOMEM); + * memcpy(addr, &to_be_added, sizeof(to_be_added)); + * + * // Shortcut of the above. + * type *addr = av_dynarray2_add((void **)&array, &nb, sizeof(*array), + * (const void *)&to_be_added2); + * if (!addr) + * return AVERROR(ENOMEM); + * + * // Now: + * // nb == 2 + * // to_be_added == array[0] + * // to_be_added2 == array[1] + * + * av_freep(&array); + * @endcode + * + * @{ + */ + +/** + * Add the pointer to an element to a dynamic array. + * + * The array to grow is supposed to be an array of pointers to + * structures, and the element to add must be a pointer to an already + * allocated structure. + * + * The array is reallocated when its size reaches powers of 2. + * Therefore, the amortized cost of adding an element is constant. + * + * In case of success, the pointer to the array is updated in order to + * point to the new grown array, and the number pointed to by `nb_ptr` + * is incremented. + * In case of failure, the array is freed, `*tab_ptr` is set to `NULL` and + * `*nb_ptr` is set to 0. + * + * @param[in,out] tab_ptr Pointer to the array to grow + * @param[in,out] nb_ptr Pointer to the number of elements in the array + * @param[in] elem Element to add + * @see av_dynarray_add_nofree(), av_dynarray2_add() + */ +void av_dynarray_add(void *tab_ptr, int *nb_ptr, void *elem); + +/** + * Add an element to a dynamic array. + * + * Function has the same functionality as av_dynarray_add(), + * but it doesn't free memory on fails. It returns error code + * instead and leave current buffer untouched. + * + * @return >=0 on success, negative otherwise + * @see av_dynarray_add(), av_dynarray2_add() + */ +av_warn_unused_result +int av_dynarray_add_nofree(void *tab_ptr, int *nb_ptr, void *elem); + +/** + * Add an element of size `elem_size` to a dynamic array. + * + * The array is reallocated when its number of elements reaches powers of 2. + * Therefore, the amortized cost of adding an element is constant. + * + * In case of success, the pointer to the array is updated in order to + * point to the new grown array, and the number pointed to by `nb_ptr` + * is incremented. + * In case of failure, the array is freed, `*tab_ptr` is set to `NULL` and + * `*nb_ptr` is set to 0. + * + * @param[in,out] tab_ptr Pointer to the array to grow + * @param[in,out] nb_ptr Pointer to the number of elements in the array + * @param[in] elem_size Size in bytes of an element in the array + * @param[in] elem_data Pointer to the data of the element to add. If + * `NULL`, the space of the newly added element is + * allocated but left uninitialized. + * + * @return Pointer to the data of the element to copy in the newly allocated + * space + * @see av_dynarray_add(), av_dynarray_add_nofree() + */ +void *av_dynarray2_add(void **tab_ptr, int *nb_ptr, size_t elem_size, + const uint8_t *elem_data); + +/** + * @} + */ + +/** + * @defgroup lavu_mem_misc Miscellaneous Functions + * + * Other functions related to memory allocation. + * + * @{ + */ + +/** + * Multiply two `size_t` values checking for overflow. + * + * @param[in] a Operand of multiplication + * @param[in] b Operand of multiplication + * @param[out] r Pointer to the result of the operation + * @return 0 on success, AVERROR(EINVAL) on overflow + */ +int av_size_mult(size_t a, size_t b, size_t *r); + +/** + * Set the maximum size that may be allocated in one block. + * + * The value specified with this function is effective for all libavutil's @ref + * lavu_mem_funcs "heap management functions." + * + * By default, the max value is defined as `INT_MAX`. + * + * @param max Value to be set as the new maximum size + * + * @warning Exercise extreme caution when using this function. Don't touch + * this if you do not understand the full consequence of doing so. + */ +void av_max_alloc(size_t max); + +/** + * @} + * @} + */ + +#endif /* AVUTIL_MEM_H */ diff --git a/media/ffvpx/libavutil/mem_internal.h b/media/ffvpx/libavutil/mem_internal.h new file mode 100644 index 0000000000..2448c606f1 --- /dev/null +++ b/media/ffvpx/libavutil/mem_internal.h @@ -0,0 +1,138 @@ +/* + * Copyright (c) 2002 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_MEM_INTERNAL_H +#define AVUTIL_MEM_INTERNAL_H + +#include "config.h" + +#include <stdint.h> + +#include "attributes.h" +#include "macros.h" +#include "mem.h" +#include "version.h" + +/** + * @def DECLARE_ALIGNED(n,t,v) + * Declare a variable that is aligned in memory. + * + * @code{.c} + * DECLARE_ALIGNED(16, uint16_t, aligned_int) = 42; + * DECLARE_ALIGNED(32, uint8_t, aligned_array)[128]; + * + * // The default-alignment equivalent would be + * uint16_t aligned_int = 42; + * uint8_t aligned_array[128]; + * @endcode + * + * @param n Minimum alignment in bytes + * @param t Type of the variable (or array element) + * @param v Name of the variable + */ + +/** + * @def DECLARE_ASM_ALIGNED(n,t,v) + * Declare an aligned variable appropriate for use in inline assembly code. + * + * @code{.c} + * DECLARE_ASM_ALIGNED(16, uint64_t, pw_08) = UINT64_C(0x0008000800080008); + * @endcode + * + * @param n Minimum alignment in bytes + * @param t Type of the variable (or array element) + * @param v Name of the variable + */ + +/** + * @def DECLARE_ASM_CONST(n,t,v) + * Declare a static constant aligned variable appropriate for use in inline + * assembly code. + * + * @code{.c} + * DECLARE_ASM_CONST(16, uint64_t, pw_08) = UINT64_C(0x0008000800080008); + * @endcode + * + * @param n Minimum alignment in bytes + * @param t Type of the variable (or array element) + * @param v Name of the variable + */ + +#if defined(__INTEL_COMPILER) && __INTEL_COMPILER < 1110 || defined(__SUNPRO_C) + #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v + #define DECLARE_ASM_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v + #define DECLARE_ASM_CONST(n,t,v) const t __attribute__ ((aligned (n))) v +#elif defined(__DJGPP__) + #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (FFMIN(n, 16)))) v + #define DECLARE_ASM_ALIGNED(n,t,v) t av_used __attribute__ ((aligned (FFMIN(n, 16)))) v + #define DECLARE_ASM_CONST(n,t,v) static const t av_used __attribute__ ((aligned (FFMIN(n, 16)))) v +#elif defined(__GNUC__) || defined(__clang__) + #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v + #define DECLARE_ASM_ALIGNED(n,t,v) t av_used __attribute__ ((aligned (n))) v + #define DECLARE_ASM_CONST(n,t,v) static const t av_used __attribute__ ((aligned (n))) v +#elif defined(_MSC_VER) + #define DECLARE_ALIGNED(n,t,v) __declspec(align(n)) t v + #define DECLARE_ASM_ALIGNED(n,t,v) __declspec(align(n)) t v + #define DECLARE_ASM_CONST(n,t,v) __declspec(align(n)) static const t v +#else + #define DECLARE_ALIGNED(n,t,v) t v + #define DECLARE_ASM_ALIGNED(n,t,v) t v + #define DECLARE_ASM_CONST(n,t,v) static const t v +#endif + +// Some broken preprocessors need a second expansion +// to be forced to tokenize __VA_ARGS__ +#define E1(x) x + +#define LOCAL_ALIGNED_A(a, t, v, s, o, ...) \ + uint8_t la_##v[sizeof(t s o) + (a)]; \ + t (*v) o = (void *)FFALIGN((uintptr_t)la_##v, a) + +#define LOCAL_ALIGNED_D(a, t, v, s, o, ...) \ + DECLARE_ALIGNED(a, t, la_##v) s o; \ + t (*v) o = la_##v + +#define LOCAL_ALIGNED(a, t, v, ...) LOCAL_ALIGNED_##a(t, v, __VA_ARGS__) + +#if HAVE_LOCAL_ALIGNED +# define LOCAL_ALIGNED_4(t, v, ...) E1(LOCAL_ALIGNED_D(4, t, v, __VA_ARGS__,,)) +#else +# define LOCAL_ALIGNED_4(t, v, ...) E1(LOCAL_ALIGNED_A(4, t, v, __VA_ARGS__,,)) +#endif + +#if HAVE_LOCAL_ALIGNED +# define LOCAL_ALIGNED_8(t, v, ...) E1(LOCAL_ALIGNED_D(8, t, v, __VA_ARGS__,,)) +#else +# define LOCAL_ALIGNED_8(t, v, ...) E1(LOCAL_ALIGNED_A(8, t, v, __VA_ARGS__,,)) +#endif + +#if HAVE_LOCAL_ALIGNED +# define LOCAL_ALIGNED_16(t, v, ...) E1(LOCAL_ALIGNED_D(16, t, v, __VA_ARGS__,,)) +#else +# define LOCAL_ALIGNED_16(t, v, ...) E1(LOCAL_ALIGNED_A(16, t, v, __VA_ARGS__,,)) +#endif + +#if HAVE_LOCAL_ALIGNED +# define LOCAL_ALIGNED_32(t, v, ...) E1(LOCAL_ALIGNED_D(32, t, v, __VA_ARGS__,,)) +#else +# define LOCAL_ALIGNED_32(t, v, ...) E1(LOCAL_ALIGNED_A(32, t, v, __VA_ARGS__,,)) +#endif + +#endif /* AVUTIL_MEM_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/moz.build b/media/ffvpx/libavutil/moz.build new file mode 100644 index 0000000000..f498eb677e --- /dev/null +++ b/media/ffvpx/libavutil/moz.build @@ -0,0 +1,88 @@ +# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*- +# vim: set filetype=python: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +# Due to duplicate file names, we compile libavutil/x86 in its own +# moz.build file. +if CONFIG['FFVPX_ASFLAGS']: + if CONFIG['TARGET_CPU'] == 'x86' or CONFIG['TARGET_CPU'] == 'x86_64': + DIRS += ['x86'] + elif CONFIG['TARGET_CPU'] == 'arm': + DIRS += ['arm'] + +if CONFIG['TARGET_CPU'] == 'aarch64': + DIRS += ['aarch64'] + +SharedLibrary('mozavutil') +SOURCES += [ + 'avsscanf.c', + 'avstring.c', + 'bprint.c', + 'buffer.c', + 'channel_layout.c', + 'cpu.c', + 'crc.c', + 'dict.c', + 'error.c', + 'eval.c', + 'fifo.c', + 'fixed_dsp.c', + 'float_dsp.c', + 'frame.c', + 'hwcontext.c', + 'imgutils.c', + 'log.c', + 'log2_tab.c', + 'mathematics.c', + 'mem.c', + 'opt.c', + 'parseutils.c', + 'pixdesc.c', + 'rational.c', + 'reverse.c', + 'samplefmt.c', + 'slicethread.c', + 'time.c', + 'tx.c', + 'tx_double.c', + 'tx_float.c', + 'tx_int32.c', + 'utils.c' +] + +if not CONFIG['MOZ_FFVPX_AUDIOONLY']: + SOURCES += [ + 'adler32.c', + 'base64.c', + 'color_utils.c', + 'film_grain_params.c', + 'hdr_dynamic_metadata.c', + 'integer.c', + 'intmath.c', + 'lls.c', + 'mastering_display_metadata.c', + 'pixelutils.c', + 'threadmessage.c', + 'timecode.c', + 'video_enc_params.c' + ] + if CONFIG["MOZ_WIDGET_TOOLKIT"] == "gtk": + LOCAL_INCLUDES += ['/media/mozva'] + SOURCES += [ + 'hwcontext_vaapi.c', + ] + USE_LIBS += ['mozva'] + +SYMBOLS_FILE = 'avutil.symbols' +NoVisibilityFlags() + +OS_LIBS += CONFIG['REALTIME_LIBS'] +if CONFIG['OS_TARGET'] != 'WINNT': + OS_LIBS += ['m'] + +if CONFIG['MOZ_NEEDS_LIBATOMIC']: + OS_LIBS += ['atomic'] + +include("../ffvpxcommon.mozbuild") diff --git a/media/ffvpx/libavutil/opt.c b/media/ffvpx/libavutil/opt.c new file mode 100644 index 0000000000..0908751752 --- /dev/null +++ b/media/ffvpx/libavutil/opt.c @@ -0,0 +1,2241 @@ +/* + * AVOptions + * Copyright (c) 2005 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * AVOptions + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#include "avutil.h" +#include "avassert.h" +#include "avstring.h" +#include "channel_layout.h" +#include "common.h" +#include "dict.h" +#include "eval.h" +#include "log.h" +#include "parseutils.h" +#include "pixdesc.h" +#include "mathematics.h" +#include "opt.h" +#include "samplefmt.h" +#include "bprint.h" +#include "version.h" + +#include <float.h> + +const AVOption *av_opt_next(const void *obj, const AVOption *last) +{ + const AVClass *class; + if (!obj) + return NULL; + class = *(const AVClass**)obj; + if (!last && class && class->option && class->option[0].name) + return class->option; + if (last && last[1].name) + return ++last; + return NULL; +} + +static int read_number(const AVOption *o, const void *dst, double *num, int *den, int64_t *intnum) +{ + switch (o->type) { + case AV_OPT_TYPE_FLAGS: + *intnum = *(unsigned int*)dst; + return 0; + case AV_OPT_TYPE_PIXEL_FMT: + *intnum = *(enum AVPixelFormat *)dst; + return 0; + case AV_OPT_TYPE_SAMPLE_FMT: + *intnum = *(enum AVSampleFormat *)dst; + return 0; + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_INT: + *intnum = *(int *)dst; + return 0; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_DURATION: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + *intnum = *(int64_t *)dst; + return 0; + case AV_OPT_TYPE_FLOAT: + *num = *(float *)dst; + return 0; + case AV_OPT_TYPE_DOUBLE: + *num = *(double *)dst; + return 0; + case AV_OPT_TYPE_RATIONAL: + *intnum = ((AVRational *)dst)->num; + *den = ((AVRational *)dst)->den; + return 0; + case AV_OPT_TYPE_CONST: + *num = o->default_val.dbl; + return 0; + } + return AVERROR(EINVAL); +} + +static int write_number(void *obj, const AVOption *o, void *dst, double num, int den, int64_t intnum) +{ + if (o->type != AV_OPT_TYPE_FLAGS && + (!den || o->max * den < num * intnum || o->min * den > num * intnum)) { + num = den ? num * intnum / den : (num && intnum ? INFINITY : NAN); + av_log(obj, AV_LOG_ERROR, "Value %f for parameter '%s' out of range [%g - %g]\n", + num, o->name, o->min, o->max); + return AVERROR(ERANGE); + } + if (o->type == AV_OPT_TYPE_FLAGS) { + double d = num*intnum/den; + if (d < -1.5 || d > 0xFFFFFFFF+0.5 || (llrint(d*256) & 255)) { + av_log(obj, AV_LOG_ERROR, + "Value %f for parameter '%s' is not a valid set of 32bit integer flags\n", + num*intnum/den, o->name); + return AVERROR(ERANGE); + } + } + + switch (o->type) { + case AV_OPT_TYPE_PIXEL_FMT: + *(enum AVPixelFormat *)dst = llrint(num / den) * intnum; + break; + case AV_OPT_TYPE_SAMPLE_FMT: + *(enum AVSampleFormat *)dst = llrint(num / den) * intnum; + break; + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_FLAGS: + case AV_OPT_TYPE_INT: + *(int *)dst = llrint(num / den) * intnum; + break; + case AV_OPT_TYPE_DURATION: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_INT64:{ + double d = num / den; + if (intnum == 1 && d == (double)INT64_MAX) { + *(int64_t *)dst = INT64_MAX; + } else + *(int64_t *)dst = llrint(d) * intnum; + break;} + case AV_OPT_TYPE_UINT64:{ + double d = num / den; + // We must special case uint64_t here as llrint() does not support values + // outside the int64_t range and there is no portable function which does + // "INT64_MAX + 1ULL" is used as it is representable exactly as IEEE double + // while INT64_MAX is not + if (intnum == 1 && d == (double)UINT64_MAX) { + *(uint64_t *)dst = UINT64_MAX; + } else if (d > INT64_MAX + 1ULL) { + *(uint64_t *)dst = (llrint(d - (INT64_MAX + 1ULL)) + (INT64_MAX + 1ULL))*intnum; + } else { + *(uint64_t *)dst = llrint(d) * intnum; + } + break;} + case AV_OPT_TYPE_FLOAT: + *(float *)dst = num * intnum / den; + break; + case AV_OPT_TYPE_DOUBLE: + *(double *)dst = num * intnum / den; + break; + case AV_OPT_TYPE_RATIONAL: + case AV_OPT_TYPE_VIDEO_RATE: + if ((int) num == num) + *(AVRational *)dst = (AVRational) { num *intnum, den }; + else + *(AVRational *)dst = av_d2q(num * intnum / den, 1 << 24); + break; + default: + return AVERROR(EINVAL); + } + return 0; +} + +static int hexchar2int(char c) { + if (c >= '0' && c <= '9') + return c - '0'; + if (c >= 'a' && c <= 'f') + return c - 'a' + 10; + if (c >= 'A' && c <= 'F') + return c - 'A' + 10; + return -1; +} + +static int set_string_binary(void *obj, const AVOption *o, const char *val, uint8_t **dst) +{ + int *lendst = (int *)(dst + 1); + uint8_t *bin, *ptr; + int len; + + av_freep(dst); + *lendst = 0; + + if (!val || !(len = strlen(val))) + return 0; + + if (len & 1) + return AVERROR(EINVAL); + len /= 2; + + ptr = bin = av_malloc(len); + if (!ptr) + return AVERROR(ENOMEM); + while (*val) { + int a = hexchar2int(*val++); + int b = hexchar2int(*val++); + if (a < 0 || b < 0) { + av_free(bin); + return AVERROR(EINVAL); + } + *ptr++ = (a << 4) | b; + } + *dst = bin; + *lendst = len; + + return 0; +} + +static int set_string(void *obj, const AVOption *o, const char *val, uint8_t **dst) +{ + av_freep(dst); + *dst = av_strdup(val); + return *dst ? 0 : AVERROR(ENOMEM); +} + +#define DEFAULT_NUMVAL(opt) ((opt->type == AV_OPT_TYPE_INT64 || \ + opt->type == AV_OPT_TYPE_UINT64 || \ + opt->type == AV_OPT_TYPE_CONST || \ + opt->type == AV_OPT_TYPE_FLAGS || \ + opt->type == AV_OPT_TYPE_INT) \ + ? opt->default_val.i64 \ + : opt->default_val.dbl) + +static int set_string_number(void *obj, void *target_obj, const AVOption *o, const char *val, void *dst) +{ + int ret = 0; + + if (o->type == AV_OPT_TYPE_RATIONAL || o->type == AV_OPT_TYPE_VIDEO_RATE) { + int num, den; + char c; + if (sscanf(val, "%d%*1[:/]%d%c", &num, &den, &c) == 2) { + if ((ret = write_number(obj, o, dst, 1, den, num)) >= 0) + return ret; + ret = 0; + } + } + + for (;;) { + int i = 0; + char buf[256]; + int cmd = 0; + double d; + int64_t intnum = 1; + + if (o->type == AV_OPT_TYPE_FLAGS) { + if (*val == '+' || *val == '-') + cmd = *(val++); + for (; i < sizeof(buf) - 1 && val[i] && val[i] != '+' && val[i] != '-'; i++) + buf[i] = val[i]; + buf[i] = 0; + } + + { + int res; + int ci = 0; + double const_values[64]; + const char * const_names[64]; + int search_flags = (o->flags & AV_OPT_FLAG_CHILD_CONSTS) ? AV_OPT_SEARCH_CHILDREN : 0; + const AVOption *o_named = av_opt_find(target_obj, i ? buf : val, o->unit, 0, search_flags); + if (o_named && o_named->type == AV_OPT_TYPE_CONST) { + d = DEFAULT_NUMVAL(o_named); + if (o_named->flags & AV_OPT_FLAG_DEPRECATED) + av_log(obj, AV_LOG_WARNING, "The \"%s\" option is deprecated: %s\n", + o_named->name, o_named->help); + } else { + if (o->unit) { + for (o_named = NULL; o_named = av_opt_next(target_obj, o_named); ) { + if (o_named->type == AV_OPT_TYPE_CONST && + o_named->unit && + !strcmp(o_named->unit, o->unit)) { + if (ci + 6 >= FF_ARRAY_ELEMS(const_values)) { + av_log(obj, AV_LOG_ERROR, "const_values array too small for %s\n", o->unit); + return AVERROR_PATCHWELCOME; + } + const_names [ci ] = o_named->name; + const_values[ci++] = DEFAULT_NUMVAL(o_named); + } + } + } + const_names [ci ] = "default"; + const_values[ci++] = DEFAULT_NUMVAL(o); + const_names [ci ] = "max"; + const_values[ci++] = o->max; + const_names [ci ] = "min"; + const_values[ci++] = o->min; + const_names [ci ] = "none"; + const_values[ci++] = 0; + const_names [ci ] = "all"; + const_values[ci++] = ~0; + const_names [ci] = NULL; + const_values[ci] = 0; + + res = av_expr_parse_and_eval(&d, i ? buf : val, const_names, + const_values, NULL, NULL, NULL, NULL, NULL, 0, obj); + if (res < 0) { + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\"\n", val); + return res; + } + } + } + if (o->type == AV_OPT_TYPE_FLAGS) { + read_number(o, dst, NULL, NULL, &intnum); + if (cmd == '+') + d = intnum | (int64_t)d; + else if (cmd == '-') + d = intnum &~(int64_t)d; + } + + if ((ret = write_number(obj, o, dst, d, 1, 1)) < 0) + return ret; + val += i; + if (!i || !*val) + return 0; + } +} + +static int set_string_image_size(void *obj, const AVOption *o, const char *val, int *dst) +{ + int ret; + + if (!val || !strcmp(val, "none")) { + dst[0] = + dst[1] = 0; + return 0; + } + ret = av_parse_video_size(dst, dst + 1, val); + if (ret < 0) + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as image size\n", val); + return ret; +} + +static int set_string_video_rate(void *obj, const AVOption *o, const char *val, AVRational *dst) +{ + int ret = av_parse_video_rate(dst, val); + if (ret < 0) + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as video rate\n", val); + return ret; +} + +static int set_string_color(void *obj, const AVOption *o, const char *val, uint8_t *dst) +{ + int ret; + + if (!val) { + return 0; + } else { + ret = av_parse_color(dst, val, -1, obj); + if (ret < 0) + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as color\n", val); + return ret; + } + return 0; +} + +static const char *get_bool_name(int val) +{ + if (val < 0) + return "auto"; + return val ? "true" : "false"; +} + +static int set_string_bool(void *obj, const AVOption *o, const char *val, int *dst) +{ + int n; + + if (!val) + return 0; + + if (!strcmp(val, "auto")) { + n = -1; + } else if (av_match_name(val, "true,y,yes,enable,enabled,on")) { + n = 1; + } else if (av_match_name(val, "false,n,no,disable,disabled,off")) { + n = 0; + } else { + char *end = NULL; + n = strtol(val, &end, 10); + if (val + strlen(val) != end) + goto fail; + } + + if (n < o->min || n > o->max) + goto fail; + + *dst = n; + return 0; + +fail: + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as boolean\n", val); + return AVERROR(EINVAL); +} + +static int set_string_fmt(void *obj, const AVOption *o, const char *val, uint8_t *dst, + int fmt_nb, int ((*get_fmt)(const char *)), const char *desc) +{ + int fmt, min, max; + + if (!val || !strcmp(val, "none")) { + fmt = -1; + } else { + fmt = get_fmt(val); + if (fmt == -1) { + char *tail; + fmt = strtol(val, &tail, 0); + if (*tail || (unsigned)fmt >= fmt_nb) { + av_log(obj, AV_LOG_ERROR, + "Unable to parse option value \"%s\" as %s\n", val, desc); + return AVERROR(EINVAL); + } + } + } + + min = FFMAX(o->min, -1); + max = FFMIN(o->max, fmt_nb-1); + + // hack for compatibility with old ffmpeg + if(min == 0 && max == 0) { + min = -1; + max = fmt_nb-1; + } + + if (fmt < min || fmt > max) { + av_log(obj, AV_LOG_ERROR, + "Value %d for parameter '%s' out of %s format range [%d - %d]\n", + fmt, o->name, desc, min, max); + return AVERROR(ERANGE); + } + + *(int *)dst = fmt; + return 0; +} + +static int set_string_pixel_fmt(void *obj, const AVOption *o, const char *val, uint8_t *dst) +{ + return set_string_fmt(obj, o, val, dst, + AV_PIX_FMT_NB, av_get_pix_fmt, "pixel format"); +} + +static int set_string_sample_fmt(void *obj, const AVOption *o, const char *val, uint8_t *dst) +{ + return set_string_fmt(obj, o, val, dst, + AV_SAMPLE_FMT_NB, av_get_sample_fmt, "sample format"); +} + +static int set_string_dict(void *obj, const AVOption *o, const char *val, uint8_t **dst) +{ + AVDictionary *options = NULL; + + if (val) { + int ret = av_dict_parse_string(&options, val, "=", ":", 0); + if (ret < 0) { + av_dict_free(&options); + return ret; + } + } + + av_dict_free((AVDictionary **)dst); + *dst = (uint8_t *)options; + + return 0; +} + +static int set_string_channel_layout(void *obj, const AVOption *o, + const char *val, void *dst) +{ + AVChannelLayout *channel_layout = dst; + av_channel_layout_uninit(channel_layout); + if (!val) + return 0; + return av_channel_layout_from_string(channel_layout, val); +} + +int av_opt_set(void *obj, const char *name, const char *val, int search_flags) +{ + int ret = 0; + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; +FF_DISABLE_DEPRECATION_WARNINGS + if (!val && (o->type != AV_OPT_TYPE_STRING && + o->type != AV_OPT_TYPE_PIXEL_FMT && o->type != AV_OPT_TYPE_SAMPLE_FMT && + o->type != AV_OPT_TYPE_IMAGE_SIZE && + o->type != AV_OPT_TYPE_DURATION && o->type != AV_OPT_TYPE_COLOR && +#if FF_API_OLD_CHANNEL_LAYOUT + o->type != AV_OPT_TYPE_CHANNEL_LAYOUT && +#endif + o->type != AV_OPT_TYPE_BOOL)) + return AVERROR(EINVAL); +FF_ENABLE_DEPRECATION_WARNINGS + + if (o->flags & AV_OPT_FLAG_READONLY) + return AVERROR(EINVAL); + + if (o->flags & AV_OPT_FLAG_DEPRECATED) + av_log(obj, AV_LOG_WARNING, "The \"%s\" option is deprecated: %s\n", name, o->help); + + dst = ((uint8_t *)target_obj) + o->offset; + switch (o->type) { + case AV_OPT_TYPE_BOOL: + return set_string_bool(obj, o, val, dst); + case AV_OPT_TYPE_STRING: + return set_string(obj, o, val, dst); + case AV_OPT_TYPE_BINARY: + return set_string_binary(obj, o, val, dst); + case AV_OPT_TYPE_FLAGS: + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + case AV_OPT_TYPE_FLOAT: + case AV_OPT_TYPE_DOUBLE: + case AV_OPT_TYPE_RATIONAL: + return set_string_number(obj, target_obj, o, val, dst); + case AV_OPT_TYPE_IMAGE_SIZE: + return set_string_image_size(obj, o, val, dst); + case AV_OPT_TYPE_VIDEO_RATE: { + AVRational tmp; + ret = set_string_video_rate(obj, o, val, &tmp); + if (ret < 0) + return ret; + return write_number(obj, o, dst, 1, tmp.den, tmp.num); + } + case AV_OPT_TYPE_PIXEL_FMT: + return set_string_pixel_fmt(obj, o, val, dst); + case AV_OPT_TYPE_SAMPLE_FMT: + return set_string_sample_fmt(obj, o, val, dst); + case AV_OPT_TYPE_DURATION: + { + int64_t usecs = 0; + if (val) { + if ((ret = av_parse_time(&usecs, val, 1)) < 0) { + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as duration\n", val); + return ret; + } + } + if (usecs < o->min || usecs > o->max) { + av_log(obj, AV_LOG_ERROR, "Value %f for parameter '%s' out of range [%g - %g]\n", + usecs / 1000000.0, o->name, o->min / 1000000.0, o->max / 1000000.0); + return AVERROR(ERANGE); + } + *(int64_t *)dst = usecs; + return 0; + } + case AV_OPT_TYPE_COLOR: + return set_string_color(obj, o, val, dst); +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: + if (!val || !strcmp(val, "none")) { + *(int64_t *)dst = 0; + } else { + int64_t cl = av_get_channel_layout(val); + if (!cl) { + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as channel layout\n", val); + ret = AVERROR(EINVAL); + } + *(int64_t *)dst = cl; + return ret; + } + break; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_CHLAYOUT: + ret = set_string_channel_layout(obj, o, val, dst); + if (ret < 0) { + av_log(obj, AV_LOG_ERROR, "Unable to parse option value \"%s\" as channel layout\n", val); + ret = AVERROR(EINVAL); + } + return ret; + case AV_OPT_TYPE_DICT: + return set_string_dict(obj, o, val, dst); + } + + av_log(obj, AV_LOG_ERROR, "Invalid option type.\n"); + return AVERROR(EINVAL); +} + +#define OPT_EVAL_NUMBER(name, opttype, vartype) \ +int av_opt_eval_ ## name(void *obj, const AVOption *o, \ + const char *val, vartype *name ## _out) \ +{ \ + if (!o || o->type != opttype || o->flags & AV_OPT_FLAG_READONLY) \ + return AVERROR(EINVAL); \ + return set_string_number(obj, obj, o, val, name ## _out); \ +} + +OPT_EVAL_NUMBER(flags, AV_OPT_TYPE_FLAGS, int) +OPT_EVAL_NUMBER(int, AV_OPT_TYPE_INT, int) +OPT_EVAL_NUMBER(int64, AV_OPT_TYPE_INT64, int64_t) +OPT_EVAL_NUMBER(float, AV_OPT_TYPE_FLOAT, float) +OPT_EVAL_NUMBER(double, AV_OPT_TYPE_DOUBLE, double) +OPT_EVAL_NUMBER(q, AV_OPT_TYPE_RATIONAL, AVRational) + +static int set_number(void *obj, const char *name, double num, int den, int64_t intnum, + int search_flags) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + + if (o->flags & AV_OPT_FLAG_READONLY) + return AVERROR(EINVAL); + + dst = ((uint8_t *)target_obj) + o->offset; + return write_number(obj, o, dst, num, den, intnum); +} + +int av_opt_set_int(void *obj, const char *name, int64_t val, int search_flags) +{ + return set_number(obj, name, 1, 1, val, search_flags); +} + +int av_opt_set_double(void *obj, const char *name, double val, int search_flags) +{ + return set_number(obj, name, val, 1, 1, search_flags); +} + +int av_opt_set_q(void *obj, const char *name, AVRational val, int search_flags) +{ + return set_number(obj, name, val.num, val.den, 1, search_flags); +} + +int av_opt_set_bin(void *obj, const char *name, const uint8_t *val, int len, int search_flags) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + uint8_t *ptr; + uint8_t **dst; + int *lendst; + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + + if (o->type != AV_OPT_TYPE_BINARY || o->flags & AV_OPT_FLAG_READONLY) + return AVERROR(EINVAL); + + ptr = len ? av_malloc(len) : NULL; + if (len && !ptr) + return AVERROR(ENOMEM); + + dst = (uint8_t **)(((uint8_t *)target_obj) + o->offset); + lendst = (int *)(dst + 1); + + av_free(*dst); + *dst = ptr; + *lendst = len; + if (len) + memcpy(ptr, val, len); + + return 0; +} + +int av_opt_set_image_size(void *obj, const char *name, int w, int h, int search_flags) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_IMAGE_SIZE) { + av_log(obj, AV_LOG_ERROR, + "The value set by option '%s' is not an image size.\n", o->name); + return AVERROR(EINVAL); + } + if (w<0 || h<0) { + av_log(obj, AV_LOG_ERROR, + "Invalid negative size value %dx%d for size '%s'\n", w, h, o->name); + return AVERROR(EINVAL); + } + *(int *)(((uint8_t *)target_obj) + o->offset) = w; + *(int *)(((uint8_t *)target_obj+sizeof(int)) + o->offset) = h; + return 0; +} + +int av_opt_set_video_rate(void *obj, const char *name, AVRational val, int search_flags) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_VIDEO_RATE) { + av_log(obj, AV_LOG_ERROR, + "The value set by option '%s' is not a video rate.\n", o->name); + return AVERROR(EINVAL); + } + if (val.num <= 0 || val.den <= 0) + return AVERROR(EINVAL); + return set_number(obj, name, val.num, val.den, 1, search_flags); +} + +static int set_format(void *obj, const char *name, int fmt, int search_flags, + enum AVOptionType type, const char *desc, int nb_fmts) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, + search_flags, &target_obj); + int min, max; + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != type) { + av_log(obj, AV_LOG_ERROR, + "The value set by option '%s' is not a %s format", name, desc); + return AVERROR(EINVAL); + } + + min = FFMAX(o->min, -1); + max = FFMIN(o->max, nb_fmts-1); + + if (fmt < min || fmt > max) { + av_log(obj, AV_LOG_ERROR, + "Value %d for parameter '%s' out of %s format range [%d - %d]\n", + fmt, name, desc, min, max); + return AVERROR(ERANGE); + } + *(int *)(((uint8_t *)target_obj) + o->offset) = fmt; + return 0; +} + +int av_opt_set_pixel_fmt(void *obj, const char *name, enum AVPixelFormat fmt, int search_flags) +{ + return set_format(obj, name, fmt, search_flags, AV_OPT_TYPE_PIXEL_FMT, "pixel", AV_PIX_FMT_NB); +} + +int av_opt_set_sample_fmt(void *obj, const char *name, enum AVSampleFormat fmt, int search_flags) +{ + return set_format(obj, name, fmt, search_flags, AV_OPT_TYPE_SAMPLE_FMT, "sample", AV_SAMPLE_FMT_NB); +} + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS +int av_opt_set_channel_layout(void *obj, const char *name, int64_t cl, int search_flags) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_CHANNEL_LAYOUT) { + av_log(obj, AV_LOG_ERROR, + "The value set by option '%s' is not a channel layout.\n", o->name); + return AVERROR(EINVAL); + } + *(int64_t *)(((uint8_t *)target_obj) + o->offset) = cl; + return 0; +} +FF_ENABLE_DEPRECATION_WARNINGS +#endif + +int av_opt_set_dict_val(void *obj, const char *name, const AVDictionary *val, + int search_flags) +{ + void *target_obj; + AVDictionary **dst; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->flags & AV_OPT_FLAG_READONLY) + return AVERROR(EINVAL); + + dst = (AVDictionary **)(((uint8_t *)target_obj) + o->offset); + av_dict_free(dst); + av_dict_copy(dst, val, 0); + + return 0; +} + +int av_opt_set_chlayout(void *obj, const char *name, + const AVChannelLayout *channel_layout, + int search_flags) +{ + void *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + AVChannelLayout *dst; + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + + dst = (AVChannelLayout*)((uint8_t*)target_obj + o->offset); + + return av_channel_layout_copy(dst, channel_layout); +} + +static void format_duration(char *buf, size_t size, int64_t d) +{ + char *e; + + av_assert0(size >= 25); + if (d < 0 && d != INT64_MIN) { + *(buf++) = '-'; + size--; + d = -d; + } + if (d == INT64_MAX) + snprintf(buf, size, "INT64_MAX"); + else if (d == INT64_MIN) + snprintf(buf, size, "INT64_MIN"); + else if (d > (int64_t)3600*1000000) + snprintf(buf, size, "%"PRId64":%02d:%02d.%06d", d / 3600000000, + (int)((d / 60000000) % 60), + (int)((d / 1000000) % 60), + (int)(d % 1000000)); + else if (d > 60*1000000) + snprintf(buf, size, "%d:%02d.%06d", + (int)(d / 60000000), + (int)((d / 1000000) % 60), + (int)(d % 1000000)); + else + snprintf(buf, size, "%d.%06d", + (int)(d / 1000000), + (int)(d % 1000000)); + e = buf + strlen(buf); + while (e > buf && e[-1] == '0') + *(--e) = 0; + if (e > buf && e[-1] == '.') + *(--e) = 0; +} + +int av_opt_get(void *obj, const char *name, int search_flags, uint8_t **out_val) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + uint8_t *bin, buf[128]; + int len, i, ret; + int64_t i64; + + if (!o || !target_obj || (o->offset<=0 && o->type != AV_OPT_TYPE_CONST)) + return AVERROR_OPTION_NOT_FOUND; + + if (o->flags & AV_OPT_FLAG_DEPRECATED) + av_log(obj, AV_LOG_WARNING, "The \"%s\" option is deprecated: %s\n", name, o->help); + + dst = (uint8_t *)target_obj + o->offset; + + buf[0] = 0; + switch (o->type) { + case AV_OPT_TYPE_BOOL: + ret = snprintf(buf, sizeof(buf), "%s", (char *)av_x_if_null(get_bool_name(*(int *)dst), "invalid")); + break; + case AV_OPT_TYPE_FLAGS: + ret = snprintf(buf, sizeof(buf), "0x%08X", *(int *)dst); + break; + case AV_OPT_TYPE_INT: + ret = snprintf(buf, sizeof(buf), "%d", *(int *)dst); + break; + case AV_OPT_TYPE_INT64: + ret = snprintf(buf, sizeof(buf), "%"PRId64, *(int64_t *)dst); + break; + case AV_OPT_TYPE_UINT64: + ret = snprintf(buf, sizeof(buf), "%"PRIu64, *(uint64_t *)dst); + break; + case AV_OPT_TYPE_FLOAT: + ret = snprintf(buf, sizeof(buf), "%f", *(float *)dst); + break; + case AV_OPT_TYPE_DOUBLE: + ret = snprintf(buf, sizeof(buf), "%f", *(double *)dst); + break; + case AV_OPT_TYPE_VIDEO_RATE: + case AV_OPT_TYPE_RATIONAL: + ret = snprintf(buf, sizeof(buf), "%d/%d", ((AVRational *)dst)->num, ((AVRational *)dst)->den); + break; + case AV_OPT_TYPE_CONST: + ret = snprintf(buf, sizeof(buf), "%f", o->default_val.dbl); + break; + case AV_OPT_TYPE_STRING: + if (*(uint8_t **)dst) { + *out_val = av_strdup(*(uint8_t **)dst); + } else if (search_flags & AV_OPT_ALLOW_NULL) { + *out_val = NULL; + return 0; + } else { + *out_val = av_strdup(""); + } + return *out_val ? 0 : AVERROR(ENOMEM); + case AV_OPT_TYPE_BINARY: + if (!*(uint8_t **)dst && (search_flags & AV_OPT_ALLOW_NULL)) { + *out_val = NULL; + return 0; + } + len = *(int *)(((uint8_t *)dst) + sizeof(uint8_t *)); + if ((uint64_t)len * 2 + 1 > INT_MAX) + return AVERROR(EINVAL); + if (!(*out_val = av_malloc(len * 2 + 1))) + return AVERROR(ENOMEM); + if (!len) { + *out_val[0] = '\0'; + return 0; + } + bin = *(uint8_t **)dst; + for (i = 0; i < len; i++) + snprintf(*out_val + i * 2, 3, "%02X", bin[i]); + return 0; + case AV_OPT_TYPE_IMAGE_SIZE: + ret = snprintf(buf, sizeof(buf), "%dx%d", ((int *)dst)[0], ((int *)dst)[1]); + break; + case AV_OPT_TYPE_PIXEL_FMT: + ret = snprintf(buf, sizeof(buf), "%s", (char *)av_x_if_null(av_get_pix_fmt_name(*(enum AVPixelFormat *)dst), "none")); + break; + case AV_OPT_TYPE_SAMPLE_FMT: + ret = snprintf(buf, sizeof(buf), "%s", (char *)av_x_if_null(av_get_sample_fmt_name(*(enum AVSampleFormat *)dst), "none")); + break; + case AV_OPT_TYPE_DURATION: + i64 = *(int64_t *)dst; + format_duration(buf, sizeof(buf), i64); + ret = strlen(buf); // no overflow possible, checked by an assert + break; + case AV_OPT_TYPE_COLOR: + ret = snprintf(buf, sizeof(buf), "0x%02x%02x%02x%02x", + (int)((uint8_t *)dst)[0], (int)((uint8_t *)dst)[1], + (int)((uint8_t *)dst)[2], (int)((uint8_t *)dst)[3]); + break; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: + + i64 = *(int64_t *)dst; + ret = snprintf(buf, sizeof(buf), "0x%"PRIx64, i64); + break; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_CHLAYOUT: + ret = av_channel_layout_describe(dst, buf, sizeof(buf)); + break; + case AV_OPT_TYPE_DICT: + if (!*(AVDictionary **)dst && (search_flags & AV_OPT_ALLOW_NULL)) { + *out_val = NULL; + return 0; + } + return av_dict_get_string(*(AVDictionary **)dst, (char **)out_val, '=', ':'); + default: + return AVERROR(EINVAL); + } + + if (ret >= sizeof(buf)) + return AVERROR(EINVAL); + *out_val = av_strdup(buf); + return *out_val ? 0 : AVERROR(ENOMEM); +} + +static int get_number(void *obj, const char *name, const AVOption **o_out, double *num, int *den, int64_t *intnum, + int search_flags) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + goto error; + + dst = ((uint8_t *)target_obj) + o->offset; + + if (o_out) *o_out= o; + + return read_number(o, dst, num, den, intnum); + +error: + *den = + *intnum = 0; + return -1; +} + +int av_opt_get_int(void *obj, const char *name, int search_flags, int64_t *out_val) +{ + int64_t intnum = 1; + double num = 1; + int ret, den = 1; + + if ((ret = get_number(obj, name, NULL, &num, &den, &intnum, search_flags)) < 0) + return ret; + if (num == den) + *out_val = intnum; + else + *out_val = num * intnum / den; + return 0; +} + +int av_opt_get_double(void *obj, const char *name, int search_flags, double *out_val) +{ + int64_t intnum = 1; + double num = 1; + int ret, den = 1; + + if ((ret = get_number(obj, name, NULL, &num, &den, &intnum, search_flags)) < 0) + return ret; + *out_val = num * intnum / den; + return 0; +} + +int av_opt_get_q(void *obj, const char *name, int search_flags, AVRational *out_val) +{ + int64_t intnum = 1; + double num = 1; + int ret, den = 1; + + if ((ret = get_number(obj, name, NULL, &num, &den, &intnum, search_flags)) < 0) + return ret; + + if (num == 1.0 && (int)intnum == intnum) + *out_val = (AVRational){intnum, den}; + else + *out_val = av_d2q(num*intnum/den, 1<<24); + return 0; +} + +int av_opt_get_image_size(void *obj, const char *name, int search_flags, int *w_out, int *h_out) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_IMAGE_SIZE) { + av_log(obj, AV_LOG_ERROR, + "The value for option '%s' is not an image size.\n", name); + return AVERROR(EINVAL); + } + + dst = ((uint8_t*)target_obj) + o->offset; + if (w_out) *w_out = *(int *)dst; + if (h_out) *h_out = *((int *)dst+1); + return 0; +} + +int av_opt_get_video_rate(void *obj, const char *name, int search_flags, AVRational *out_val) +{ + int64_t intnum = 1; + double num = 1; + int ret, den = 1; + + if ((ret = get_number(obj, name, NULL, &num, &den, &intnum, search_flags)) < 0) + return ret; + + if (num == 1.0 && (int)intnum == intnum) + *out_val = (AVRational) { intnum, den }; + else + *out_val = av_d2q(num * intnum / den, 1 << 24); + return 0; +} + +static int get_format(void *obj, const char *name, int search_flags, int *out_fmt, + enum AVOptionType type, const char *desc) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != type) { + av_log(obj, AV_LOG_ERROR, + "The value for option '%s' is not a %s format.\n", desc, name); + return AVERROR(EINVAL); + } + + dst = ((uint8_t*)target_obj) + o->offset; + *out_fmt = *(int *)dst; + return 0; +} + +int av_opt_get_pixel_fmt(void *obj, const char *name, int search_flags, enum AVPixelFormat *out_fmt) +{ + return get_format(obj, name, search_flags, out_fmt, AV_OPT_TYPE_PIXEL_FMT, "pixel"); +} + +int av_opt_get_sample_fmt(void *obj, const char *name, int search_flags, enum AVSampleFormat *out_fmt) +{ + return get_format(obj, name, search_flags, out_fmt, AV_OPT_TYPE_SAMPLE_FMT, "sample"); +} + +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS +int av_opt_get_channel_layout(void *obj, const char *name, int search_flags, int64_t *cl) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_CHANNEL_LAYOUT) { + av_log(obj, AV_LOG_ERROR, + "The value for option '%s' is not a channel layout.\n", name); + return AVERROR(EINVAL); + } + + dst = ((uint8_t*)target_obj) + o->offset; + *cl = *(int64_t *)dst; + return 0; +} +FF_ENABLE_DEPRECATION_WARNINGS +#endif + +int av_opt_get_chlayout(void *obj, const char *name, int search_flags, AVChannelLayout *cl) +{ + void *dst, *target_obj; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_CHLAYOUT) { + av_log(obj, AV_LOG_ERROR, + "The value for option '%s' is not a channel layout.\n", name); + return AVERROR(EINVAL); + } + + dst = ((uint8_t*)target_obj) + o->offset; + return av_channel_layout_copy(cl, dst); +} + +int av_opt_get_dict_val(void *obj, const char *name, int search_flags, AVDictionary **out_val) +{ + void *target_obj; + AVDictionary *src; + const AVOption *o = av_opt_find2(obj, name, NULL, 0, search_flags, &target_obj); + + if (!o || !target_obj) + return AVERROR_OPTION_NOT_FOUND; + if (o->type != AV_OPT_TYPE_DICT) + return AVERROR(EINVAL); + + src = *(AVDictionary **)(((uint8_t *)target_obj) + o->offset); + av_dict_copy(out_val, src, 0); + + return 0; +} + +int av_opt_flag_is_set(void *obj, const char *field_name, const char *flag_name) +{ + const AVOption *field = av_opt_find(obj, field_name, NULL, 0, 0); + const AVOption *flag = av_opt_find(obj, flag_name, + field ? field->unit : NULL, 0, 0); + int64_t res; + + if (!field || !flag || flag->type != AV_OPT_TYPE_CONST || + av_opt_get_int(obj, field_name, 0, &res) < 0) + return 0; + return res & flag->default_val.i64; +} + +static void log_int_value(void *av_log_obj, int level, int64_t i) +{ + if (i == INT_MAX) { + av_log(av_log_obj, level, "INT_MAX"); + } else if (i == INT_MIN) { + av_log(av_log_obj, level, "INT_MIN"); + } else if (i == UINT32_MAX) { + av_log(av_log_obj, level, "UINT32_MAX"); + } else if (i == INT64_MAX) { + av_log(av_log_obj, level, "I64_MAX"); + } else if (i == INT64_MIN) { + av_log(av_log_obj, level, "I64_MIN"); + } else { + av_log(av_log_obj, level, "%"PRId64, i); + } +} + +static void log_value(void *av_log_obj, int level, double d) +{ + if (d == INT_MAX) { + av_log(av_log_obj, level, "INT_MAX"); + } else if (d == INT_MIN) { + av_log(av_log_obj, level, "INT_MIN"); + } else if (d == UINT32_MAX) { + av_log(av_log_obj, level, "UINT32_MAX"); + } else if (d == (double)INT64_MAX) { + av_log(av_log_obj, level, "I64_MAX"); + } else if (d == INT64_MIN) { + av_log(av_log_obj, level, "I64_MIN"); + } else if (d == FLT_MAX) { + av_log(av_log_obj, level, "FLT_MAX"); + } else if (d == FLT_MIN) { + av_log(av_log_obj, level, "FLT_MIN"); + } else if (d == -FLT_MAX) { + av_log(av_log_obj, level, "-FLT_MAX"); + } else if (d == -FLT_MIN) { + av_log(av_log_obj, level, "-FLT_MIN"); + } else if (d == DBL_MAX) { + av_log(av_log_obj, level, "DBL_MAX"); + } else if (d == DBL_MIN) { + av_log(av_log_obj, level, "DBL_MIN"); + } else if (d == -DBL_MAX) { + av_log(av_log_obj, level, "-DBL_MAX"); + } else if (d == -DBL_MIN) { + av_log(av_log_obj, level, "-DBL_MIN"); + } else { + av_log(av_log_obj, level, "%g", d); + } +} + +static const char *get_opt_const_name(void *obj, const char *unit, int64_t value) +{ + const AVOption *opt = NULL; + + if (!unit) + return NULL; + while ((opt = av_opt_next(obj, opt))) + if (opt->type == AV_OPT_TYPE_CONST && !strcmp(opt->unit, unit) && + opt->default_val.i64 == value) + return opt->name; + return NULL; +} + +static char *get_opt_flags_string(void *obj, const char *unit, int64_t value) +{ + const AVOption *opt = NULL; + char flags[512]; + + flags[0] = 0; + if (!unit) + return NULL; + while ((opt = av_opt_next(obj, opt))) { + if (opt->type == AV_OPT_TYPE_CONST && !strcmp(opt->unit, unit) && + opt->default_val.i64 & value) { + if (flags[0]) + av_strlcatf(flags, sizeof(flags), "+"); + av_strlcatf(flags, sizeof(flags), "%s", opt->name); + } + } + if (flags[0]) + return av_strdup(flags); + return NULL; +} + +static void opt_list(void *obj, void *av_log_obj, const char *unit, + int req_flags, int rej_flags, enum AVOptionType parent_type) +{ + const AVOption *opt = NULL; + AVOptionRanges *r; + int i; + + while ((opt = av_opt_next(obj, opt))) { + if (!(opt->flags & req_flags) || (opt->flags & rej_flags)) + continue; + + /* Don't print CONST's on level one. + * Don't print anything but CONST's on level two. + * Only print items from the requested unit. + */ + if (!unit && opt->type == AV_OPT_TYPE_CONST) + continue; + else if (unit && opt->type != AV_OPT_TYPE_CONST) + continue; + else if (unit && opt->type == AV_OPT_TYPE_CONST && strcmp(unit, opt->unit)) + continue; + else if (unit && opt->type == AV_OPT_TYPE_CONST) + av_log(av_log_obj, AV_LOG_INFO, " %-15s ", opt->name); + else + av_log(av_log_obj, AV_LOG_INFO, " %s%-17s ", + (opt->flags & AV_OPT_FLAG_FILTERING_PARAM) ? " " : "-", + opt->name); + + switch (opt->type) { + case AV_OPT_TYPE_FLAGS: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<flags>"); + break; + case AV_OPT_TYPE_INT: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<int>"); + break; + case AV_OPT_TYPE_INT64: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<int64>"); + break; + case AV_OPT_TYPE_UINT64: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<uint64>"); + break; + case AV_OPT_TYPE_DOUBLE: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<double>"); + break; + case AV_OPT_TYPE_FLOAT: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<float>"); + break; + case AV_OPT_TYPE_STRING: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<string>"); + break; + case AV_OPT_TYPE_RATIONAL: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<rational>"); + break; + case AV_OPT_TYPE_BINARY: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<binary>"); + break; + case AV_OPT_TYPE_DICT: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<dictionary>"); + break; + case AV_OPT_TYPE_IMAGE_SIZE: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<image_size>"); + break; + case AV_OPT_TYPE_VIDEO_RATE: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<video_rate>"); + break; + case AV_OPT_TYPE_PIXEL_FMT: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<pix_fmt>"); + break; + case AV_OPT_TYPE_SAMPLE_FMT: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<sample_fmt>"); + break; + case AV_OPT_TYPE_DURATION: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<duration>"); + break; + case AV_OPT_TYPE_COLOR: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<color>"); + break; + case AV_OPT_TYPE_CHLAYOUT: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<channel_layout>"); + break; + case AV_OPT_TYPE_BOOL: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", "<boolean>"); + break; + case AV_OPT_TYPE_CONST: + if (parent_type == AV_OPT_TYPE_INT) + av_log(av_log_obj, AV_LOG_INFO, "%-12"PRId64" ", opt->default_val.i64); + else + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", ""); + break; + default: + av_log(av_log_obj, AV_LOG_INFO, "%-12s ", ""); + break; + } + av_log(av_log_obj, AV_LOG_INFO, "%c%c%c%c%c%c%c%c%c%c%c", + (opt->flags & AV_OPT_FLAG_ENCODING_PARAM) ? 'E' : '.', + (opt->flags & AV_OPT_FLAG_DECODING_PARAM) ? 'D' : '.', + (opt->flags & AV_OPT_FLAG_FILTERING_PARAM) ? 'F' : '.', + (opt->flags & AV_OPT_FLAG_VIDEO_PARAM) ? 'V' : '.', + (opt->flags & AV_OPT_FLAG_AUDIO_PARAM) ? 'A' : '.', + (opt->flags & AV_OPT_FLAG_SUBTITLE_PARAM) ? 'S' : '.', + (opt->flags & AV_OPT_FLAG_EXPORT) ? 'X' : '.', + (opt->flags & AV_OPT_FLAG_READONLY) ? 'R' : '.', + (opt->flags & AV_OPT_FLAG_BSF_PARAM) ? 'B' : '.', + (opt->flags & AV_OPT_FLAG_RUNTIME_PARAM) ? 'T' : '.', + (opt->flags & AV_OPT_FLAG_DEPRECATED) ? 'P' : '.'); + + if (opt->help) + av_log(av_log_obj, AV_LOG_INFO, " %s", opt->help); + + if (av_opt_query_ranges(&r, obj, opt->name, AV_OPT_SEARCH_FAKE_OBJ) >= 0) { + switch (opt->type) { + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + case AV_OPT_TYPE_DOUBLE: + case AV_OPT_TYPE_FLOAT: + case AV_OPT_TYPE_RATIONAL: + for (i = 0; i < r->nb_ranges; i++) { + av_log(av_log_obj, AV_LOG_INFO, " (from "); + log_value(av_log_obj, AV_LOG_INFO, r->range[i]->value_min); + av_log(av_log_obj, AV_LOG_INFO, " to "); + log_value(av_log_obj, AV_LOG_INFO, r->range[i]->value_max); + av_log(av_log_obj, AV_LOG_INFO, ")"); + } + break; + } + av_opt_freep_ranges(&r); + } + + if (opt->type != AV_OPT_TYPE_CONST && + opt->type != AV_OPT_TYPE_BINARY && + !((opt->type == AV_OPT_TYPE_COLOR || + opt->type == AV_OPT_TYPE_IMAGE_SIZE || + opt->type == AV_OPT_TYPE_STRING || + opt->type == AV_OPT_TYPE_DICT || + opt->type == AV_OPT_TYPE_CHLAYOUT || + opt->type == AV_OPT_TYPE_VIDEO_RATE) && + !opt->default_val.str)) { + av_log(av_log_obj, AV_LOG_INFO, " (default "); + switch (opt->type) { + case AV_OPT_TYPE_BOOL: + av_log(av_log_obj, AV_LOG_INFO, "%s", (char *)av_x_if_null(get_bool_name(opt->default_val.i64), "invalid")); + break; + case AV_OPT_TYPE_FLAGS: { + char *def_flags = get_opt_flags_string(obj, opt->unit, opt->default_val.i64); + if (def_flags) { + av_log(av_log_obj, AV_LOG_INFO, "%s", def_flags); + av_freep(&def_flags); + } else { + av_log(av_log_obj, AV_LOG_INFO, "%"PRIX64, opt->default_val.i64); + } + break; + } + case AV_OPT_TYPE_DURATION: { + char buf[25]; + format_duration(buf, sizeof(buf), opt->default_val.i64); + av_log(av_log_obj, AV_LOG_INFO, "%s", buf); + break; + } + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_UINT64: + case AV_OPT_TYPE_INT64: { + const char *def_const = get_opt_const_name(obj, opt->unit, opt->default_val.i64); + if (def_const) + av_log(av_log_obj, AV_LOG_INFO, "%s", def_const); + else + log_int_value(av_log_obj, AV_LOG_INFO, opt->default_val.i64); + break; + } + case AV_OPT_TYPE_DOUBLE: + case AV_OPT_TYPE_FLOAT: + log_value(av_log_obj, AV_LOG_INFO, opt->default_val.dbl); + break; + case AV_OPT_TYPE_RATIONAL: { + AVRational q = av_d2q(opt->default_val.dbl, INT_MAX); + av_log(av_log_obj, AV_LOG_INFO, "%d/%d", q.num, q.den); } + break; + case AV_OPT_TYPE_PIXEL_FMT: + av_log(av_log_obj, AV_LOG_INFO, "%s", (char *)av_x_if_null(av_get_pix_fmt_name(opt->default_val.i64), "none")); + break; + case AV_OPT_TYPE_SAMPLE_FMT: + av_log(av_log_obj, AV_LOG_INFO, "%s", (char *)av_x_if_null(av_get_sample_fmt_name(opt->default_val.i64), "none")); + break; + case AV_OPT_TYPE_COLOR: + case AV_OPT_TYPE_IMAGE_SIZE: + case AV_OPT_TYPE_STRING: + case AV_OPT_TYPE_DICT: + case AV_OPT_TYPE_VIDEO_RATE: + case AV_OPT_TYPE_CHLAYOUT: + av_log(av_log_obj, AV_LOG_INFO, "\"%s\"", opt->default_val.str); + break; +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: + av_log(av_log_obj, AV_LOG_INFO, "0x%"PRIx64, opt->default_val.i64); + break; +FF_ENABLE_DEPRECATION_WARNINGS +#endif + } + av_log(av_log_obj, AV_LOG_INFO, ")"); + } + + av_log(av_log_obj, AV_LOG_INFO, "\n"); + if (opt->unit && opt->type != AV_OPT_TYPE_CONST) + opt_list(obj, av_log_obj, opt->unit, req_flags, rej_flags, opt->type); + } +} + +int av_opt_show2(void *obj, void *av_log_obj, int req_flags, int rej_flags) +{ + if (!obj) + return -1; + + av_log(av_log_obj, AV_LOG_INFO, "%s AVOptions:\n", (*(AVClass **)obj)->class_name); + + opt_list(obj, av_log_obj, NULL, req_flags, rej_flags, -1); + + return 0; +} + +void av_opt_set_defaults(void *s) +{ + av_opt_set_defaults2(s, 0, 0); +} + +void av_opt_set_defaults2(void *s, int mask, int flags) +{ + const AVOption *opt = NULL; + while ((opt = av_opt_next(s, opt))) { + void *dst = ((uint8_t*)s) + opt->offset; + + if ((opt->flags & mask) != flags) + continue; + + if (opt->flags & AV_OPT_FLAG_READONLY) + continue; + + switch (opt->type) { + case AV_OPT_TYPE_CONST: + /* Nothing to be done here */ + break; + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_FLAGS: + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + case AV_OPT_TYPE_DURATION: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_PIXEL_FMT: + case AV_OPT_TYPE_SAMPLE_FMT: + write_number(s, opt, dst, 1, 1, opt->default_val.i64); + break; + case AV_OPT_TYPE_DOUBLE: + case AV_OPT_TYPE_FLOAT: { + double val; + val = opt->default_val.dbl; + write_number(s, opt, dst, val, 1, 1); + } + break; + case AV_OPT_TYPE_RATIONAL: { + AVRational val; + val = av_d2q(opt->default_val.dbl, INT_MAX); + write_number(s, opt, dst, 1, val.den, val.num); + } + break; + case AV_OPT_TYPE_COLOR: + set_string_color(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_STRING: + set_string(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_IMAGE_SIZE: + set_string_image_size(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_VIDEO_RATE: + set_string_video_rate(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_BINARY: + set_string_binary(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_CHLAYOUT: + set_string_channel_layout(s, opt, opt->default_val.str, dst); + break; + case AV_OPT_TYPE_DICT: + set_string_dict(s, opt, opt->default_val.str, dst); + break; + default: + av_log(s, AV_LOG_DEBUG, "AVOption type %d of option %s not implemented yet\n", + opt->type, opt->name); + } + } +} + +/** + * Store the value in the field in ctx that is named like key. + * ctx must be an AVClass context, storing is done using AVOptions. + * + * @param buf the string to parse, buf will be updated to point at the + * separator just after the parsed key/value pair + * @param key_val_sep a 0-terminated list of characters used to + * separate key from value + * @param pairs_sep a 0-terminated list of characters used to separate + * two pairs from each other + * @return 0 if the key/value pair has been successfully parsed and + * set, or a negative value corresponding to an AVERROR code in case + * of error: + * AVERROR(EINVAL) if the key/value pair cannot be parsed, + * the error code issued by av_opt_set() if the key/value pair + * cannot be set + */ +static int parse_key_value_pair(void *ctx, const char **buf, + const char *key_val_sep, const char *pairs_sep) +{ + char *key = av_get_token(buf, key_val_sep); + char *val; + int ret; + + if (!key) + return AVERROR(ENOMEM); + + if (*key && strspn(*buf, key_val_sep)) { + (*buf)++; + val = av_get_token(buf, pairs_sep); + if (!val) { + av_freep(&key); + return AVERROR(ENOMEM); + } + } else { + av_log(ctx, AV_LOG_ERROR, "Missing key or no key/value separator found after key '%s'\n", key); + av_free(key); + return AVERROR(EINVAL); + } + + av_log(ctx, AV_LOG_DEBUG, "Setting entry with key '%s' to value '%s'\n", key, val); + + ret = av_opt_set(ctx, key, val, AV_OPT_SEARCH_CHILDREN); + if (ret == AVERROR_OPTION_NOT_FOUND) + av_log(ctx, AV_LOG_ERROR, "Key '%s' not found.\n", key); + + av_free(key); + av_free(val); + return ret; +} + +int av_set_options_string(void *ctx, const char *opts, + const char *key_val_sep, const char *pairs_sep) +{ + int ret, count = 0; + + if (!opts) + return 0; + + while (*opts) { + if ((ret = parse_key_value_pair(ctx, &opts, key_val_sep, pairs_sep)) < 0) + return ret; + count++; + + if (*opts) + opts++; + } + + return count; +} + +#define WHITESPACES " \n\t\r" + +static int is_key_char(char c) +{ + return (unsigned)((c | 32) - 'a') < 26 || + (unsigned)(c - '0') < 10 || + c == '-' || c == '_' || c == '/' || c == '.'; +} + +/** + * Read a key from a string. + * + * The key consists of is_key_char characters and must be terminated by a + * character from the delim string; spaces are ignored. + * + * @return 0 for success (even with ellipsis), <0 for failure + */ +static int get_key(const char **ropts, const char *delim, char **rkey) +{ + const char *opts = *ropts; + const char *key_start, *key_end; + + key_start = opts += strspn(opts, WHITESPACES); + while (is_key_char(*opts)) + opts++; + key_end = opts; + opts += strspn(opts, WHITESPACES); + if (!*opts || !strchr(delim, *opts)) + return AVERROR(EINVAL); + opts++; + if (!(*rkey = av_malloc(key_end - key_start + 1))) + return AVERROR(ENOMEM); + memcpy(*rkey, key_start, key_end - key_start); + (*rkey)[key_end - key_start] = 0; + *ropts = opts; + return 0; +} + +int av_opt_get_key_value(const char **ropts, + const char *key_val_sep, const char *pairs_sep, + unsigned flags, + char **rkey, char **rval) +{ + int ret; + char *key = NULL, *val; + const char *opts = *ropts; + + if ((ret = get_key(&opts, key_val_sep, &key)) < 0 && + !(flags & AV_OPT_FLAG_IMPLICIT_KEY)) + return AVERROR(EINVAL); + if (!(val = av_get_token(&opts, pairs_sep))) { + av_free(key); + return AVERROR(ENOMEM); + } + *ropts = opts; + *rkey = key; + *rval = val; + return 0; +} + +int av_opt_set_from_string(void *ctx, const char *opts, + const char *const *shorthand, + const char *key_val_sep, const char *pairs_sep) +{ + int ret, count = 0; + const char *dummy_shorthand = NULL; + char *av_uninit(parsed_key), *av_uninit(value); + const char *key; + + if (!opts) + return 0; + if (!shorthand) + shorthand = &dummy_shorthand; + + while (*opts) { + ret = av_opt_get_key_value(&opts, key_val_sep, pairs_sep, + *shorthand ? AV_OPT_FLAG_IMPLICIT_KEY : 0, + &parsed_key, &value); + if (ret < 0) { + if (ret == AVERROR(EINVAL)) + av_log(ctx, AV_LOG_ERROR, "No option name near '%s'\n", opts); + else + av_log(ctx, AV_LOG_ERROR, "Unable to parse '%s': %s\n", opts, + av_err2str(ret)); + return ret; + } + if (*opts) + opts++; + if (parsed_key) { + key = parsed_key; + while (*shorthand) /* discard all remaining shorthand */ + shorthand++; + } else { + key = *(shorthand++); + } + + av_log(ctx, AV_LOG_DEBUG, "Setting '%s' to value '%s'\n", key, value); + if ((ret = av_opt_set(ctx, key, value, 0)) < 0) { + if (ret == AVERROR_OPTION_NOT_FOUND) + av_log(ctx, AV_LOG_ERROR, "Option '%s' not found\n", key); + av_free(value); + av_free(parsed_key); + return ret; + } + + av_free(value); + av_free(parsed_key); + count++; + } + return count; +} + +void av_opt_free(void *obj) +{ + const AVOption *o = NULL; + while ((o = av_opt_next(obj, o))) { + switch (o->type) { + case AV_OPT_TYPE_STRING: + case AV_OPT_TYPE_BINARY: + av_freep((uint8_t *)obj + o->offset); + break; + + case AV_OPT_TYPE_DICT: + av_dict_free((AVDictionary **)(((uint8_t *)obj) + o->offset)); + break; + + case AV_OPT_TYPE_CHLAYOUT: + av_channel_layout_uninit((AVChannelLayout *)(((uint8_t *)obj) + o->offset)); + break; + + default: + break; + } + } +} + +int av_opt_set_dict2(void *obj, AVDictionary **options, int search_flags) +{ + const AVDictionaryEntry *t = NULL; + AVDictionary *tmp = NULL; + int ret; + + if (!options) + return 0; + + while ((t = av_dict_iterate(*options, t))) { + ret = av_opt_set(obj, t->key, t->value, search_flags); + if (ret == AVERROR_OPTION_NOT_FOUND) + ret = av_dict_set(&tmp, t->key, t->value, AV_DICT_MULTIKEY); + if (ret < 0) { + av_log(obj, AV_LOG_ERROR, "Error setting option %s to value %s.\n", t->key, t->value); + av_dict_free(&tmp); + return ret; + } + } + av_dict_free(options); + *options = tmp; + return 0; +} + +int av_opt_set_dict(void *obj, AVDictionary **options) +{ + return av_opt_set_dict2(obj, options, 0); +} + +const AVOption *av_opt_find(void *obj, const char *name, const char *unit, + int opt_flags, int search_flags) +{ + return av_opt_find2(obj, name, unit, opt_flags, search_flags, NULL); +} + +const AVOption *av_opt_find2(void *obj, const char *name, const char *unit, + int opt_flags, int search_flags, void **target_obj) +{ + const AVClass *c; + const AVOption *o = NULL; + + if(!obj) + return NULL; + + c= *(AVClass**)obj; + + if (!c) + return NULL; + + if (search_flags & AV_OPT_SEARCH_CHILDREN) { + if (search_flags & AV_OPT_SEARCH_FAKE_OBJ) { + void *iter = NULL; + const AVClass *child; + while (child = av_opt_child_class_iterate(c, &iter)) + if (o = av_opt_find2(&child, name, unit, opt_flags, search_flags, NULL)) + return o; + } else { + void *child = NULL; + while (child = av_opt_child_next(obj, child)) + if (o = av_opt_find2(child, name, unit, opt_flags, search_flags, target_obj)) + return o; + } + } + + while (o = av_opt_next(obj, o)) { + if (!strcmp(o->name, name) && (o->flags & opt_flags) == opt_flags && + ((!unit && o->type != AV_OPT_TYPE_CONST) || + (unit && o->type == AV_OPT_TYPE_CONST && o->unit && !strcmp(o->unit, unit)))) { + if (target_obj) { + if (!(search_flags & AV_OPT_SEARCH_FAKE_OBJ)) + *target_obj = obj; + else + *target_obj = NULL; + } + return o; + } + } + return NULL; +} + +void *av_opt_child_next(void *obj, void *prev) +{ + const AVClass *c = *(AVClass **)obj; + if (c->child_next) + return c->child_next(obj, prev); + return NULL; +} + +const AVClass *av_opt_child_class_iterate(const AVClass *parent, void **iter) +{ + if (parent->child_class_iterate) + return parent->child_class_iterate(iter); + return NULL; +} + +void *av_opt_ptr(const AVClass *class, void *obj, const char *name) +{ + const AVOption *opt= av_opt_find2(&class, name, NULL, 0, AV_OPT_SEARCH_FAKE_OBJ, NULL); + if(!opt) + return NULL; + return (uint8_t*)obj + opt->offset; +} + +static int opt_size(enum AVOptionType type) +{ + switch(type) { + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_FLAGS: + return sizeof(int); + case AV_OPT_TYPE_DURATION: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + return sizeof(int64_t); + case AV_OPT_TYPE_DOUBLE: + return sizeof(double); + case AV_OPT_TYPE_FLOAT: + return sizeof(float); + case AV_OPT_TYPE_STRING: + return sizeof(uint8_t*); + case AV_OPT_TYPE_VIDEO_RATE: + case AV_OPT_TYPE_RATIONAL: + return sizeof(AVRational); + case AV_OPT_TYPE_BINARY: + return sizeof(uint8_t*) + sizeof(int); + case AV_OPT_TYPE_IMAGE_SIZE: + return sizeof(int[2]); + case AV_OPT_TYPE_PIXEL_FMT: + return sizeof(enum AVPixelFormat); + case AV_OPT_TYPE_SAMPLE_FMT: + return sizeof(enum AVSampleFormat); + case AV_OPT_TYPE_COLOR: + return 4; + } + return AVERROR(EINVAL); +} + +int av_opt_copy(void *dst, const void *src) +{ + const AVOption *o = NULL; + const AVClass *c; + int ret = 0; + + if (!src) + return AVERROR(EINVAL); + + c = *(AVClass **)src; + if (!c || c != *(AVClass **)dst) + return AVERROR(EINVAL); + + while ((o = av_opt_next(src, o))) { + void *field_dst = (uint8_t *)dst + o->offset; + void *field_src = (uint8_t *)src + o->offset; + uint8_t **field_dst8 = (uint8_t **)field_dst; + uint8_t **field_src8 = (uint8_t **)field_src; + + if (o->type == AV_OPT_TYPE_STRING) { + if (*field_dst8 != *field_src8) + av_freep(field_dst8); + *field_dst8 = av_strdup(*field_src8); + if (*field_src8 && !*field_dst8) + ret = AVERROR(ENOMEM); + } else if (o->type == AV_OPT_TYPE_BINARY) { + int len = *(int *)(field_src8 + 1); + if (*field_dst8 != *field_src8) + av_freep(field_dst8); + *field_dst8 = av_memdup(*field_src8, len); + if (len && !*field_dst8) { + ret = AVERROR(ENOMEM); + len = 0; + } + *(int *)(field_dst8 + 1) = len; + } else if (o->type == AV_OPT_TYPE_CONST) { + // do nothing + } else if (o->type == AV_OPT_TYPE_DICT) { + AVDictionary **sdict = (AVDictionary **) field_src; + AVDictionary **ddict = (AVDictionary **) field_dst; + int ret2; + if (*sdict != *ddict) + av_dict_free(ddict); + *ddict = NULL; + ret2 = av_dict_copy(ddict, *sdict, 0); + if (ret2 < 0) + ret = ret2; + } else if (o->type == AV_OPT_TYPE_CHLAYOUT) { + if (field_dst != field_src) + ret = av_channel_layout_copy(field_dst, field_src); + } else { + int size = opt_size(o->type); + if (size < 0) + ret = size; + else + memcpy(field_dst, field_src, size); + } + } + return ret; +} + +int av_opt_query_ranges(AVOptionRanges **ranges_arg, void *obj, const char *key, int flags) +{ + int ret; + const AVClass *c = *(AVClass**)obj; + int (*callback)(AVOptionRanges **, void *obj, const char *key, int flags) = c->query_ranges; + + if (!callback) + callback = av_opt_query_ranges_default; + + ret = callback(ranges_arg, obj, key, flags); + if (ret >= 0) { + if (!(flags & AV_OPT_MULTI_COMPONENT_RANGE)) + ret = 1; + (*ranges_arg)->nb_components = ret; + } + return ret; +} + +int av_opt_query_ranges_default(AVOptionRanges **ranges_arg, void *obj, const char *key, int flags) +{ + AVOptionRanges *ranges = av_mallocz(sizeof(*ranges)); + AVOptionRange **range_array = av_mallocz(sizeof(void*)); + AVOptionRange *range = av_mallocz(sizeof(*range)); + const AVOption *field = av_opt_find(obj, key, NULL, 0, flags); + int ret; + + *ranges_arg = NULL; + + if (!ranges || !range || !range_array || !field) { + ret = AVERROR(ENOMEM); + goto fail; + } + + ranges->range = range_array; + ranges->range[0] = range; + ranges->nb_ranges = 1; + ranges->nb_components = 1; + range->is_range = 1; + range->value_min = field->min; + range->value_max = field->max; + + switch (field->type) { + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_INT: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + case AV_OPT_TYPE_PIXEL_FMT: + case AV_OPT_TYPE_SAMPLE_FMT: + case AV_OPT_TYPE_FLOAT: + case AV_OPT_TYPE_DOUBLE: + case AV_OPT_TYPE_DURATION: + case AV_OPT_TYPE_COLOR: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + break; + case AV_OPT_TYPE_STRING: + range->component_min = 0; + range->component_max = 0x10FFFF; // max unicode value + range->value_min = -1; + range->value_max = INT_MAX; + break; + case AV_OPT_TYPE_RATIONAL: + range->component_min = INT_MIN; + range->component_max = INT_MAX; + break; + case AV_OPT_TYPE_IMAGE_SIZE: + range->component_min = 0; + range->component_max = INT_MAX/128/8; + range->value_min = 0; + range->value_max = INT_MAX/8; + break; + case AV_OPT_TYPE_VIDEO_RATE: + range->component_min = 1; + range->component_max = INT_MAX; + range->value_min = 1; + range->value_max = INT_MAX; + break; + default: + ret = AVERROR(ENOSYS); + goto fail; + } + + *ranges_arg = ranges; + return 1; +fail: + av_free(ranges); + av_free(range); + av_free(range_array); + return ret; +} + +void av_opt_freep_ranges(AVOptionRanges **rangesp) +{ + int i; + AVOptionRanges *ranges = *rangesp; + + if (!ranges) + return; + + for (i = 0; i < ranges->nb_ranges * ranges->nb_components; i++) { + AVOptionRange *range = ranges->range[i]; + if (range) { + av_freep(&range->str); + av_freep(&ranges->range[i]); + } + } + av_freep(&ranges->range); + av_freep(rangesp); +} + +int av_opt_is_set_to_default(void *obj, const AVOption *o) +{ + int64_t i64; + double d, d2; + float f; + AVRational q; + int ret, w, h; + char *str; + void *dst; + + if (!o || !obj) + return AVERROR(EINVAL); + + dst = ((uint8_t*)obj) + o->offset; + + switch (o->type) { + case AV_OPT_TYPE_CONST: + return 1; + case AV_OPT_TYPE_BOOL: + case AV_OPT_TYPE_FLAGS: + case AV_OPT_TYPE_PIXEL_FMT: + case AV_OPT_TYPE_SAMPLE_FMT: + case AV_OPT_TYPE_INT: +#if FF_API_OLD_CHANNEL_LAYOUT +FF_DISABLE_DEPRECATION_WARNINGS + case AV_OPT_TYPE_CHANNEL_LAYOUT: +FF_ENABLE_DEPRECATION_WARNINGS +#endif + case AV_OPT_TYPE_DURATION: + case AV_OPT_TYPE_INT64: + case AV_OPT_TYPE_UINT64: + read_number(o, dst, NULL, NULL, &i64); + return o->default_val.i64 == i64; + case AV_OPT_TYPE_CHLAYOUT: { + AVChannelLayout ch_layout = { 0 }; + if (o->default_val.str) { + if ((ret = av_channel_layout_from_string(&ch_layout, o->default_val.str)) < 0) + return ret; + } + return !av_channel_layout_compare((AVChannelLayout *)dst, &ch_layout); + } + case AV_OPT_TYPE_STRING: + str = *(char **)dst; + if (str == o->default_val.str) //2 NULLs + return 1; + if (!str || !o->default_val.str) //1 NULL + return 0; + return !strcmp(str, o->default_val.str); + case AV_OPT_TYPE_DOUBLE: + read_number(o, dst, &d, NULL, NULL); + return o->default_val.dbl == d; + case AV_OPT_TYPE_FLOAT: + read_number(o, dst, &d, NULL, NULL); + f = o->default_val.dbl; + d2 = f; + return d2 == d; + case AV_OPT_TYPE_RATIONAL: + q = av_d2q(o->default_val.dbl, INT_MAX); + return !av_cmp_q(*(AVRational*)dst, q); + case AV_OPT_TYPE_BINARY: { + struct { + uint8_t *data; + int size; + } tmp = {0}; + int opt_size = *(int *)((void **)dst + 1); + void *opt_ptr = *(void **)dst; + if (!opt_size && (!o->default_val.str || !strlen(o->default_val.str))) + return 1; + if (!opt_size || !o->default_val.str || !strlen(o->default_val.str )) + return 0; + if (opt_size != strlen(o->default_val.str) / 2) + return 0; + ret = set_string_binary(NULL, NULL, o->default_val.str, &tmp.data); + if (!ret) + ret = !memcmp(opt_ptr, tmp.data, tmp.size); + av_free(tmp.data); + return ret; + } + case AV_OPT_TYPE_DICT: { + AVDictionary *dict1 = NULL; + AVDictionary *dict2 = *(AVDictionary **)dst; + const AVDictionaryEntry *en1 = NULL; + const AVDictionaryEntry *en2 = NULL; + ret = av_dict_parse_string(&dict1, o->default_val.str, "=", ":", 0); + if (ret < 0) { + av_dict_free(&dict1); + return ret; + } + do { + en1 = av_dict_iterate(dict1, en1); + en2 = av_dict_iterate(dict2, en2); + } while (en1 && en2 && !strcmp(en1->key, en2->key) && !strcmp(en1->value, en2->value)); + av_dict_free(&dict1); + return (!en1 && !en2); + } + case AV_OPT_TYPE_IMAGE_SIZE: + if (!o->default_val.str || !strcmp(o->default_val.str, "none")) + w = h = 0; + else if ((ret = av_parse_video_size(&w, &h, o->default_val.str)) < 0) + return ret; + return (w == *(int *)dst) && (h == *((int *)dst+1)); + case AV_OPT_TYPE_VIDEO_RATE: + q = (AVRational){0, 0}; + if (o->default_val.str) { + if ((ret = av_parse_video_rate(&q, o->default_val.str)) < 0) + return ret; + } + return !av_cmp_q(*(AVRational*)dst, q); + case AV_OPT_TYPE_COLOR: { + uint8_t color[4] = {0, 0, 0, 0}; + if (o->default_val.str) { + if ((ret = av_parse_color(color, o->default_val.str, -1, NULL)) < 0) + return ret; + } + return !memcmp(color, dst, sizeof(color)); + } + default: + av_log(obj, AV_LOG_WARNING, "Not supported option type: %d, option name: %s\n", o->type, o->name); + break; + } + return AVERROR_PATCHWELCOME; +} + +int av_opt_is_set_to_default_by_name(void *obj, const char *name, int search_flags) +{ + const AVOption *o; + void *target; + if (!obj) + return AVERROR(EINVAL); + o = av_opt_find2(obj, name, NULL, 0, search_flags, &target); + if (!o) + return AVERROR_OPTION_NOT_FOUND; + return av_opt_is_set_to_default(target, o); +} + +int av_opt_serialize(void *obj, int opt_flags, int flags, char **buffer, + const char key_val_sep, const char pairs_sep) +{ + const AVOption *o = NULL; + uint8_t *buf; + AVBPrint bprint; + int ret, cnt = 0; + const char special_chars[] = {pairs_sep, key_val_sep, '\0'}; + + if (pairs_sep == '\0' || key_val_sep == '\0' || pairs_sep == key_val_sep || + pairs_sep == '\\' || key_val_sep == '\\') { + av_log(obj, AV_LOG_ERROR, "Invalid separator(s) found."); + return AVERROR(EINVAL); + } + + if (!obj || !buffer) + return AVERROR(EINVAL); + + *buffer = NULL; + av_bprint_init(&bprint, 64, AV_BPRINT_SIZE_UNLIMITED); + + while (o = av_opt_next(obj, o)) { + if (o->type == AV_OPT_TYPE_CONST) + continue; + if ((flags & AV_OPT_SERIALIZE_OPT_FLAGS_EXACT) && o->flags != opt_flags) + continue; + else if (((o->flags & opt_flags) != opt_flags)) + continue; + if (flags & AV_OPT_SERIALIZE_SKIP_DEFAULTS && av_opt_is_set_to_default(obj, o) > 0) + continue; + if ((ret = av_opt_get(obj, o->name, 0, &buf)) < 0) { + av_bprint_finalize(&bprint, NULL); + return ret; + } + if (buf) { + if (cnt++) + av_bprint_append_data(&bprint, &pairs_sep, 1); + av_bprint_escape(&bprint, o->name, special_chars, AV_ESCAPE_MODE_BACKSLASH, 0); + av_bprint_append_data(&bprint, &key_val_sep, 1); + av_bprint_escape(&bprint, buf, special_chars, AV_ESCAPE_MODE_BACKSLASH, 0); + av_freep(&buf); + } + } + ret = av_bprint_finalize(&bprint, buffer); + if (ret < 0) + return ret; + return 0; +} diff --git a/media/ffvpx/libavutil/opt.h b/media/ffvpx/libavutil/opt.h new file mode 100644 index 0000000000..461b5d3b6b --- /dev/null +++ b/media/ffvpx/libavutil/opt.h @@ -0,0 +1,891 @@ +/* + * AVOptions + * copyright (c) 2005 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_OPT_H +#define AVUTIL_OPT_H + +/** + * @file + * AVOptions + */ + +#include "rational.h" +#include "avutil.h" +#include "channel_layout.h" +#include "dict.h" +#include "log.h" +#include "pixfmt.h" +#include "samplefmt.h" + +/** + * @defgroup avoptions AVOptions + * @ingroup lavu_data + * @{ + * AVOptions provide a generic system to declare options on arbitrary structs + * ("objects"). An option can have a help text, a type and a range of possible + * values. Options may then be enumerated, read and written to. + * + * @section avoptions_implement Implementing AVOptions + * This section describes how to add AVOptions capabilities to a struct. + * + * All AVOptions-related information is stored in an AVClass. Therefore + * the first member of the struct should be a pointer to an AVClass describing it. + * The option field of the AVClass must be set to a NULL-terminated static array + * of AVOptions. Each AVOption must have a non-empty name, a type, a default + * value and for number-type AVOptions also a range of allowed values. It must + * also declare an offset in bytes from the start of the struct, where the field + * associated with this AVOption is located. Other fields in the AVOption struct + * should also be set when applicable, but are not required. + * + * The following example illustrates an AVOptions-enabled struct: + * @code + * typedef struct test_struct { + * const AVClass *class; + * int int_opt; + * char *str_opt; + * uint8_t *bin_opt; + * int bin_len; + * } test_struct; + * + * static const AVOption test_options[] = { + * { "test_int", "This is a test option of int type.", offsetof(test_struct, int_opt), + * AV_OPT_TYPE_INT, { .i64 = -1 }, INT_MIN, INT_MAX }, + * { "test_str", "This is a test option of string type.", offsetof(test_struct, str_opt), + * AV_OPT_TYPE_STRING }, + * { "test_bin", "This is a test option of binary type.", offsetof(test_struct, bin_opt), + * AV_OPT_TYPE_BINARY }, + * { NULL }, + * }; + * + * static const AVClass test_class = { + * .class_name = "test class", + * .item_name = av_default_item_name, + * .option = test_options, + * .version = LIBAVUTIL_VERSION_INT, + * }; + * @endcode + * + * Next, when allocating your struct, you must ensure that the AVClass pointer + * is set to the correct value. Then, av_opt_set_defaults() can be called to + * initialize defaults. After that the struct is ready to be used with the + * AVOptions API. + * + * When cleaning up, you may use the av_opt_free() function to automatically + * free all the allocated string and binary options. + * + * Continuing with the above example: + * + * @code + * test_struct *alloc_test_struct(void) + * { + * test_struct *ret = av_mallocz(sizeof(*ret)); + * ret->class = &test_class; + * av_opt_set_defaults(ret); + * return ret; + * } + * void free_test_struct(test_struct **foo) + * { + * av_opt_free(*foo); + * av_freep(foo); + * } + * @endcode + * + * @subsection avoptions_implement_nesting Nesting + * It may happen that an AVOptions-enabled struct contains another + * AVOptions-enabled struct as a member (e.g. AVCodecContext in + * libavcodec exports generic options, while its priv_data field exports + * codec-specific options). In such a case, it is possible to set up the + * parent struct to export a child's options. To do that, simply + * implement AVClass.child_next() and AVClass.child_class_iterate() in the + * parent struct's AVClass. + * Assuming that the test_struct from above now also contains a + * child_struct field: + * + * @code + * typedef struct child_struct { + * AVClass *class; + * int flags_opt; + * } child_struct; + * static const AVOption child_opts[] = { + * { "test_flags", "This is a test option of flags type.", + * offsetof(child_struct, flags_opt), AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT_MIN, INT_MAX }, + * { NULL }, + * }; + * static const AVClass child_class = { + * .class_name = "child class", + * .item_name = av_default_item_name, + * .option = child_opts, + * .version = LIBAVUTIL_VERSION_INT, + * }; + * + * void *child_next(void *obj, void *prev) + * { + * test_struct *t = obj; + * if (!prev && t->child_struct) + * return t->child_struct; + * return NULL + * } + * const AVClass child_class_iterate(void **iter) + * { + * const AVClass *c = *iter ? NULL : &child_class; + * *iter = (void*)(uintptr_t)c; + * return c; + * } + * @endcode + * Putting child_next() and child_class_iterate() as defined above into + * test_class will now make child_struct's options accessible through + * test_struct (again, proper setup as described above needs to be done on + * child_struct right after it is created). + * + * From the above example it might not be clear why both child_next() + * and child_class_iterate() are needed. The distinction is that child_next() + * iterates over actually existing objects, while child_class_iterate() + * iterates over all possible child classes. E.g. if an AVCodecContext + * was initialized to use a codec which has private options, then its + * child_next() will return AVCodecContext.priv_data and finish + * iterating. OTOH child_class_iterate() on AVCodecContext.av_class will + * iterate over all available codecs with private options. + * + * @subsection avoptions_implement_named_constants Named constants + * It is possible to create named constants for options. Simply set the unit + * field of the option the constants should apply to a string and + * create the constants themselves as options of type AV_OPT_TYPE_CONST + * with their unit field set to the same string. + * Their default_val field should contain the value of the named + * constant. + * For example, to add some named constants for the test_flags option + * above, put the following into the child_opts array: + * @code + * { "test_flags", "This is a test option of flags type.", + * offsetof(child_struct, flags_opt), AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT_MIN, INT_MAX, "test_unit" }, + * { "flag1", "This is a flag with value 16", 0, AV_OPT_TYPE_CONST, { .i64 = 16 }, 0, 0, "test_unit" }, + * @endcode + * + * @section avoptions_use Using AVOptions + * This section deals with accessing options in an AVOptions-enabled struct. + * Such structs in FFmpeg are e.g. AVCodecContext in libavcodec or + * AVFormatContext in libavformat. + * + * @subsection avoptions_use_examine Examining AVOptions + * The basic functions for examining options are av_opt_next(), which iterates + * over all options defined for one object, and av_opt_find(), which searches + * for an option with the given name. + * + * The situation is more complicated with nesting. An AVOptions-enabled struct + * may have AVOptions-enabled children. Passing the AV_OPT_SEARCH_CHILDREN flag + * to av_opt_find() will make the function search children recursively. + * + * For enumerating there are basically two cases. The first is when you want to + * get all options that may potentially exist on the struct and its children + * (e.g. when constructing documentation). In that case you should call + * av_opt_child_class_iterate() recursively on the parent struct's AVClass. The + * second case is when you have an already initialized struct with all its + * children and you want to get all options that can be actually written or read + * from it. In that case you should call av_opt_child_next() recursively (and + * av_opt_next() on each result). + * + * @subsection avoptions_use_get_set Reading and writing AVOptions + * When setting options, you often have a string read directly from the + * user. In such a case, simply passing it to av_opt_set() is enough. For + * non-string type options, av_opt_set() will parse the string according to the + * option type. + * + * Similarly av_opt_get() will read any option type and convert it to a string + * which will be returned. Do not forget that the string is allocated, so you + * have to free it with av_free(). + * + * In some cases it may be more convenient to put all options into an + * AVDictionary and call av_opt_set_dict() on it. A specific case of this + * are the format/codec open functions in lavf/lavc which take a dictionary + * filled with option as a parameter. This makes it possible to set some options + * that cannot be set otherwise, since e.g. the input file format is not known + * before the file is actually opened. + */ + +enum AVOptionType{ + AV_OPT_TYPE_FLAGS, + AV_OPT_TYPE_INT, + AV_OPT_TYPE_INT64, + AV_OPT_TYPE_DOUBLE, + AV_OPT_TYPE_FLOAT, + AV_OPT_TYPE_STRING, + AV_OPT_TYPE_RATIONAL, + AV_OPT_TYPE_BINARY, ///< offset must point to a pointer immediately followed by an int for the length + AV_OPT_TYPE_DICT, + AV_OPT_TYPE_UINT64, + AV_OPT_TYPE_CONST, + AV_OPT_TYPE_IMAGE_SIZE, ///< offset must point to two consecutive integers + AV_OPT_TYPE_PIXEL_FMT, + AV_OPT_TYPE_SAMPLE_FMT, + AV_OPT_TYPE_VIDEO_RATE, ///< offset must point to AVRational + AV_OPT_TYPE_DURATION, + AV_OPT_TYPE_COLOR, +#if FF_API_OLD_CHANNEL_LAYOUT + AV_OPT_TYPE_CHANNEL_LAYOUT, +#endif + AV_OPT_TYPE_BOOL, + AV_OPT_TYPE_CHLAYOUT, +}; + +/** + * AVOption + */ +typedef struct AVOption { + const char *name; + + /** + * short English help text + * @todo What about other languages? + */ + const char *help; + + /** + * The offset relative to the context structure where the option + * value is stored. It should be 0 for named constants. + */ + int offset; + enum AVOptionType type; + + /** + * the default value for scalar options + */ + union { + int64_t i64; + double dbl; + const char *str; + /* TODO those are unused now */ + AVRational q; + } default_val; + double min; ///< minimum valid value for the option + double max; ///< maximum valid value for the option + + int flags; +#define AV_OPT_FLAG_ENCODING_PARAM 1 ///< a generic parameter which can be set by the user for muxing or encoding +#define AV_OPT_FLAG_DECODING_PARAM 2 ///< a generic parameter which can be set by the user for demuxing or decoding +#define AV_OPT_FLAG_AUDIO_PARAM 8 +#define AV_OPT_FLAG_VIDEO_PARAM 16 +#define AV_OPT_FLAG_SUBTITLE_PARAM 32 +/** + * The option is intended for exporting values to the caller. + */ +#define AV_OPT_FLAG_EXPORT 64 +/** + * The option may not be set through the AVOptions API, only read. + * This flag only makes sense when AV_OPT_FLAG_EXPORT is also set. + */ +#define AV_OPT_FLAG_READONLY 128 +#define AV_OPT_FLAG_BSF_PARAM (1<<8) ///< a generic parameter which can be set by the user for bit stream filtering +#define AV_OPT_FLAG_RUNTIME_PARAM (1<<15) ///< a generic parameter which can be set by the user at runtime +#define AV_OPT_FLAG_FILTERING_PARAM (1<<16) ///< a generic parameter which can be set by the user for filtering +#define AV_OPT_FLAG_DEPRECATED (1<<17) ///< set if option is deprecated, users should refer to AVOption.help text for more information +#define AV_OPT_FLAG_CHILD_CONSTS (1<<18) ///< set if option constants can also reside in child objects +//FIXME think about enc-audio, ... style flags + + /** + * The logical unit to which the option belongs. Non-constant + * options and corresponding named constants share the same + * unit. May be NULL. + */ + const char *unit; +} AVOption; + +/** + * A single allowed range of values, or a single allowed value. + */ +typedef struct AVOptionRange { + const char *str; + /** + * Value range. + * For string ranges this represents the min/max length. + * For dimensions this represents the min/max pixel count or width/height in multi-component case. + */ + double value_min, value_max; + /** + * Value's component range. + * For string this represents the unicode range for chars, 0-127 limits to ASCII. + */ + double component_min, component_max; + /** + * Range flag. + * If set to 1 the struct encodes a range, if set to 0 a single value. + */ + int is_range; +} AVOptionRange; + +/** + * List of AVOptionRange structs. + */ +typedef struct AVOptionRanges { + /** + * Array of option ranges. + * + * Most of option types use just one component. + * Following describes multi-component option types: + * + * AV_OPT_TYPE_IMAGE_SIZE: + * component index 0: range of pixel count (width * height). + * component index 1: range of width. + * component index 2: range of height. + * + * @note To obtain multi-component version of this structure, user must + * provide AV_OPT_MULTI_COMPONENT_RANGE to av_opt_query_ranges or + * av_opt_query_ranges_default function. + * + * Multi-component range can be read as in following example: + * + * @code + * int range_index, component_index; + * AVOptionRanges *ranges; + * AVOptionRange *range[3]; //may require more than 3 in the future. + * av_opt_query_ranges(&ranges, obj, key, AV_OPT_MULTI_COMPONENT_RANGE); + * for (range_index = 0; range_index < ranges->nb_ranges; range_index++) { + * for (component_index = 0; component_index < ranges->nb_components; component_index++) + * range[component_index] = ranges->range[ranges->nb_ranges * component_index + range_index]; + * //do something with range here. + * } + * av_opt_freep_ranges(&ranges); + * @endcode + */ + AVOptionRange **range; + /** + * Number of ranges per component. + */ + int nb_ranges; + /** + * Number of componentes. + */ + int nb_components; +} AVOptionRanges; + +/** + * Show the obj options. + * + * @param req_flags requested flags for the options to show. Show only the + * options for which it is opt->flags & req_flags. + * @param rej_flags rejected flags for the options to show. Show only the + * options for which it is !(opt->flags & req_flags). + * @param av_log_obj log context to use for showing the options + */ +int av_opt_show2(void *obj, void *av_log_obj, int req_flags, int rej_flags); + +/** + * Set the values of all AVOption fields to their default values. + * + * @param s an AVOption-enabled struct (its first member must be a pointer to AVClass) + */ +void av_opt_set_defaults(void *s); + +/** + * Set the values of all AVOption fields to their default values. Only these + * AVOption fields for which (opt->flags & mask) == flags will have their + * default applied to s. + * + * @param s an AVOption-enabled struct (its first member must be a pointer to AVClass) + * @param mask combination of AV_OPT_FLAG_* + * @param flags combination of AV_OPT_FLAG_* + */ +void av_opt_set_defaults2(void *s, int mask, int flags); + +/** + * Parse the key/value pairs list in opts. For each key/value pair + * found, stores the value in the field in ctx that is named like the + * key. ctx must be an AVClass context, storing is done using + * AVOptions. + * + * @param opts options string to parse, may be NULL + * @param key_val_sep a 0-terminated list of characters used to + * separate key from value + * @param pairs_sep a 0-terminated list of characters used to separate + * two pairs from each other + * @return the number of successfully set key/value pairs, or a negative + * value corresponding to an AVERROR code in case of error: + * AVERROR(EINVAL) if opts cannot be parsed, + * the error code issued by av_opt_set() if a key/value pair + * cannot be set + */ +int av_set_options_string(void *ctx, const char *opts, + const char *key_val_sep, const char *pairs_sep); + +/** + * Parse the key-value pairs list in opts. For each key=value pair found, + * set the value of the corresponding option in ctx. + * + * @param ctx the AVClass object to set options on + * @param opts the options string, key-value pairs separated by a + * delimiter + * @param shorthand a NULL-terminated array of options names for shorthand + * notation: if the first field in opts has no key part, + * the key is taken from the first element of shorthand; + * then again for the second, etc., until either opts is + * finished, shorthand is finished or a named option is + * found; after that, all options must be named + * @param key_val_sep a 0-terminated list of characters used to separate + * key from value, for example '=' + * @param pairs_sep a 0-terminated list of characters used to separate + * two pairs from each other, for example ':' or ',' + * @return the number of successfully set key=value pairs, or a negative + * value corresponding to an AVERROR code in case of error: + * AVERROR(EINVAL) if opts cannot be parsed, + * the error code issued by av_set_string3() if a key/value pair + * cannot be set + * + * Options names must use only the following characters: a-z A-Z 0-9 - . / _ + * Separators must use characters distinct from option names and from each + * other. + */ +int av_opt_set_from_string(void *ctx, const char *opts, + const char *const *shorthand, + const char *key_val_sep, const char *pairs_sep); +/** + * Free all allocated objects in obj. + */ +void av_opt_free(void *obj); + +/** + * Check whether a particular flag is set in a flags field. + * + * @param field_name the name of the flag field option + * @param flag_name the name of the flag to check + * @return non-zero if the flag is set, zero if the flag isn't set, + * isn't of the right type, or the flags field doesn't exist. + */ +int av_opt_flag_is_set(void *obj, const char *field_name, const char *flag_name); + +/** + * Set all the options from a given dictionary on an object. + * + * @param obj a struct whose first element is a pointer to AVClass + * @param options options to process. This dictionary will be freed and replaced + * by a new one containing all options not found in obj. + * Of course this new dictionary needs to be freed by caller + * with av_dict_free(). + * + * @return 0 on success, a negative AVERROR if some option was found in obj, + * but could not be set. + * + * @see av_dict_copy() + */ +int av_opt_set_dict(void *obj, struct AVDictionary **options); + + +/** + * Set all the options from a given dictionary on an object. + * + * @param obj a struct whose first element is a pointer to AVClass + * @param options options to process. This dictionary will be freed and replaced + * by a new one containing all options not found in obj. + * Of course this new dictionary needs to be freed by caller + * with av_dict_free(). + * @param search_flags A combination of AV_OPT_SEARCH_*. + * + * @return 0 on success, a negative AVERROR if some option was found in obj, + * but could not be set. + * + * @see av_dict_copy() + */ +int av_opt_set_dict2(void *obj, struct AVDictionary **options, int search_flags); + +/** + * Extract a key-value pair from the beginning of a string. + * + * @param ropts pointer to the options string, will be updated to + * point to the rest of the string (one of the pairs_sep + * or the final NUL) + * @param key_val_sep a 0-terminated list of characters used to separate + * key from value, for example '=' + * @param pairs_sep a 0-terminated list of characters used to separate + * two pairs from each other, for example ':' or ',' + * @param flags flags; see the AV_OPT_FLAG_* values below + * @param rkey parsed key; must be freed using av_free() + * @param rval parsed value; must be freed using av_free() + * + * @return >=0 for success, or a negative value corresponding to an + * AVERROR code in case of error; in particular: + * AVERROR(EINVAL) if no key is present + * + */ +int av_opt_get_key_value(const char **ropts, + const char *key_val_sep, const char *pairs_sep, + unsigned flags, + char **rkey, char **rval); + +enum { + + /** + * Accept to parse a value without a key; the key will then be returned + * as NULL. + */ + AV_OPT_FLAG_IMPLICIT_KEY = 1, +}; + +/** + * @defgroup opt_eval_funcs Evaluating option strings + * @{ + * This group of functions can be used to evaluate option strings + * and get numbers out of them. They do the same thing as av_opt_set(), + * except the result is written into the caller-supplied pointer. + * + * @param obj a struct whose first element is a pointer to AVClass. + * @param o an option for which the string is to be evaluated. + * @param val string to be evaluated. + * @param *_out value of the string will be written here. + * + * @return 0 on success, a negative number on failure. + */ +int av_opt_eval_flags (void *obj, const AVOption *o, const char *val, int *flags_out); +int av_opt_eval_int (void *obj, const AVOption *o, const char *val, int *int_out); +int av_opt_eval_int64 (void *obj, const AVOption *o, const char *val, int64_t *int64_out); +int av_opt_eval_float (void *obj, const AVOption *o, const char *val, float *float_out); +int av_opt_eval_double(void *obj, const AVOption *o, const char *val, double *double_out); +int av_opt_eval_q (void *obj, const AVOption *o, const char *val, AVRational *q_out); +/** + * @} + */ + +#define AV_OPT_SEARCH_CHILDREN (1 << 0) /**< Search in possible children of the + given object first. */ +/** + * The obj passed to av_opt_find() is fake -- only a double pointer to AVClass + * instead of a required pointer to a struct containing AVClass. This is + * useful for searching for options without needing to allocate the corresponding + * object. + */ +#define AV_OPT_SEARCH_FAKE_OBJ (1 << 1) + +/** + * In av_opt_get, return NULL if the option has a pointer type and is set to NULL, + * rather than returning an empty string. + */ +#define AV_OPT_ALLOW_NULL (1 << 2) + +/** + * Allows av_opt_query_ranges and av_opt_query_ranges_default to return more than + * one component for certain option types. + * @see AVOptionRanges for details. + */ +#define AV_OPT_MULTI_COMPONENT_RANGE (1 << 12) + +/** + * Look for an option in an object. Consider only options which + * have all the specified flags set. + * + * @param[in] obj A pointer to a struct whose first element is a + * pointer to an AVClass. + * Alternatively a double pointer to an AVClass, if + * AV_OPT_SEARCH_FAKE_OBJ search flag is set. + * @param[in] name The name of the option to look for. + * @param[in] unit When searching for named constants, name of the unit + * it belongs to. + * @param opt_flags Find only options with all the specified flags set (AV_OPT_FLAG). + * @param search_flags A combination of AV_OPT_SEARCH_*. + * + * @return A pointer to the option found, or NULL if no option + * was found. + * + * @note Options found with AV_OPT_SEARCH_CHILDREN flag may not be settable + * directly with av_opt_set(). Use special calls which take an options + * AVDictionary (e.g. avformat_open_input()) to set options found with this + * flag. + */ +const AVOption *av_opt_find(void *obj, const char *name, const char *unit, + int opt_flags, int search_flags); + +/** + * Look for an option in an object. Consider only options which + * have all the specified flags set. + * + * @param[in] obj A pointer to a struct whose first element is a + * pointer to an AVClass. + * Alternatively a double pointer to an AVClass, if + * AV_OPT_SEARCH_FAKE_OBJ search flag is set. + * @param[in] name The name of the option to look for. + * @param[in] unit When searching for named constants, name of the unit + * it belongs to. + * @param opt_flags Find only options with all the specified flags set (AV_OPT_FLAG). + * @param search_flags A combination of AV_OPT_SEARCH_*. + * @param[out] target_obj if non-NULL, an object to which the option belongs will be + * written here. It may be different from obj if AV_OPT_SEARCH_CHILDREN is present + * in search_flags. This parameter is ignored if search_flags contain + * AV_OPT_SEARCH_FAKE_OBJ. + * + * @return A pointer to the option found, or NULL if no option + * was found. + */ +const AVOption *av_opt_find2(void *obj, const char *name, const char *unit, + int opt_flags, int search_flags, void **target_obj); + +/** + * Iterate over all AVOptions belonging to obj. + * + * @param obj an AVOptions-enabled struct or a double pointer to an + * AVClass describing it. + * @param prev result of the previous call to av_opt_next() on this object + * or NULL + * @return next AVOption or NULL + */ +const AVOption *av_opt_next(const void *obj, const AVOption *prev); + +/** + * Iterate over AVOptions-enabled children of obj. + * + * @param prev result of a previous call to this function or NULL + * @return next AVOptions-enabled child or NULL + */ +void *av_opt_child_next(void *obj, void *prev); + +/** + * Iterate over potential AVOptions-enabled children of parent. + * + * @param iter a pointer where iteration state is stored. + * @return AVClass corresponding to next potential child or NULL + */ +const AVClass *av_opt_child_class_iterate(const AVClass *parent, void **iter); + +/** + * @defgroup opt_set_funcs Option setting functions + * @{ + * Those functions set the field of obj with the given name to value. + * + * @param[in] obj A struct whose first element is a pointer to an AVClass. + * @param[in] name the name of the field to set + * @param[in] val The value to set. In case of av_opt_set() if the field is not + * of a string type, then the given string is parsed. + * SI postfixes and some named scalars are supported. + * If the field is of a numeric type, it has to be a numeric or named + * scalar. Behavior with more than one scalar and +- infix operators + * is undefined. + * If the field is of a flags type, it has to be a sequence of numeric + * scalars or named flags separated by '+' or '-'. Prefixing a flag + * with '+' causes it to be set without affecting the other flags; + * similarly, '-' unsets a flag. + * If the field is of a dictionary type, it has to be a ':' separated list of + * key=value parameters. Values containing ':' special characters must be + * escaped. + * @param search_flags flags passed to av_opt_find2. I.e. if AV_OPT_SEARCH_CHILDREN + * is passed here, then the option may be set on a child of obj. + * + * @return 0 if the value has been set, or an AVERROR code in case of + * error: + * AVERROR_OPTION_NOT_FOUND if no matching option exists + * AVERROR(ERANGE) if the value is out of range + * AVERROR(EINVAL) if the value is not valid + */ +int av_opt_set (void *obj, const char *name, const char *val, int search_flags); +int av_opt_set_int (void *obj, const char *name, int64_t val, int search_flags); +int av_opt_set_double (void *obj, const char *name, double val, int search_flags); +int av_opt_set_q (void *obj, const char *name, AVRational val, int search_flags); +int av_opt_set_bin (void *obj, const char *name, const uint8_t *val, int size, int search_flags); +int av_opt_set_image_size(void *obj, const char *name, int w, int h, int search_flags); +int av_opt_set_pixel_fmt (void *obj, const char *name, enum AVPixelFormat fmt, int search_flags); +int av_opt_set_sample_fmt(void *obj, const char *name, enum AVSampleFormat fmt, int search_flags); +int av_opt_set_video_rate(void *obj, const char *name, AVRational val, int search_flags); +#if FF_API_OLD_CHANNEL_LAYOUT +attribute_deprecated +int av_opt_set_channel_layout(void *obj, const char *name, int64_t ch_layout, int search_flags); +#endif +int av_opt_set_chlayout(void *obj, const char *name, const AVChannelLayout *layout, int search_flags); +/** + * @note Any old dictionary present is discarded and replaced with a copy of the new one. The + * caller still owns val is and responsible for freeing it. + */ +int av_opt_set_dict_val(void *obj, const char *name, const AVDictionary *val, int search_flags); + +/** + * Set a binary option to an integer list. + * + * @param obj AVClass object to set options on + * @param name name of the binary option + * @param val pointer to an integer list (must have the correct type with + * regard to the contents of the list) + * @param term list terminator (usually 0 or -1) + * @param flags search flags + */ +#define av_opt_set_int_list(obj, name, val, term, flags) \ + (av_int_list_length(val, term) > INT_MAX / sizeof(*(val)) ? \ + AVERROR(EINVAL) : \ + av_opt_set_bin(obj, name, (const uint8_t *)(val), \ + av_int_list_length(val, term) * sizeof(*(val)), flags)) + +/** + * @} + */ + +/** + * @defgroup opt_get_funcs Option getting functions + * @{ + * Those functions get a value of the option with the given name from an object. + * + * @param[in] obj a struct whose first element is a pointer to an AVClass. + * @param[in] name name of the option to get. + * @param[in] search_flags flags passed to av_opt_find2. I.e. if AV_OPT_SEARCH_CHILDREN + * is passed here, then the option may be found in a child of obj. + * @param[out] out_val value of the option will be written here + * @return >=0 on success, a negative error code otherwise + */ +/** + * @note the returned string will be av_malloc()ed and must be av_free()ed by the caller + * + * @note if AV_OPT_ALLOW_NULL is set in search_flags in av_opt_get, and the + * option is of type AV_OPT_TYPE_STRING, AV_OPT_TYPE_BINARY or AV_OPT_TYPE_DICT + * and is set to NULL, *out_val will be set to NULL instead of an allocated + * empty string. + */ +int av_opt_get (void *obj, const char *name, int search_flags, uint8_t **out_val); +int av_opt_get_int (void *obj, const char *name, int search_flags, int64_t *out_val); +int av_opt_get_double (void *obj, const char *name, int search_flags, double *out_val); +int av_opt_get_q (void *obj, const char *name, int search_flags, AVRational *out_val); +int av_opt_get_image_size(void *obj, const char *name, int search_flags, int *w_out, int *h_out); +int av_opt_get_pixel_fmt (void *obj, const char *name, int search_flags, enum AVPixelFormat *out_fmt); +int av_opt_get_sample_fmt(void *obj, const char *name, int search_flags, enum AVSampleFormat *out_fmt); +int av_opt_get_video_rate(void *obj, const char *name, int search_flags, AVRational *out_val); +#if FF_API_OLD_CHANNEL_LAYOUT +attribute_deprecated +int av_opt_get_channel_layout(void *obj, const char *name, int search_flags, int64_t *ch_layout); +#endif +int av_opt_get_chlayout(void *obj, const char *name, int search_flags, AVChannelLayout *layout); +/** + * @param[out] out_val The returned dictionary is a copy of the actual value and must + * be freed with av_dict_free() by the caller + */ +int av_opt_get_dict_val(void *obj, const char *name, int search_flags, AVDictionary **out_val); +/** + * @} + */ +/** + * Gets a pointer to the requested field in a struct. + * This function allows accessing a struct even when its fields are moved or + * renamed since the application making the access has been compiled, + * + * @returns a pointer to the field, it can be cast to the correct type and read + * or written to. + */ +void *av_opt_ptr(const AVClass *avclass, void *obj, const char *name); + +/** + * Free an AVOptionRanges struct and set it to NULL. + */ +void av_opt_freep_ranges(AVOptionRanges **ranges); + +/** + * Get a list of allowed ranges for the given option. + * + * The returned list may depend on other fields in obj like for example profile. + * + * @param flags is a bitmask of flags, undefined flags should not be set and should be ignored + * AV_OPT_SEARCH_FAKE_OBJ indicates that the obj is a double pointer to a AVClass instead of a full instance + * AV_OPT_MULTI_COMPONENT_RANGE indicates that function may return more than one component, @see AVOptionRanges + * + * The result must be freed with av_opt_freep_ranges. + * + * @return number of compontents returned on success, a negative errro code otherwise + */ +int av_opt_query_ranges(AVOptionRanges **, void *obj, const char *key, int flags); + +/** + * Copy options from src object into dest object. + * + * The underlying AVClass of both src and dest must coincide. The guarantee + * below does not apply if this is not fulfilled. + * + * Options that require memory allocation (e.g. string or binary) are malloc'ed in dest object. + * Original memory allocated for such options is freed unless both src and dest options points to the same memory. + * + * Even on error it is guaranteed that allocated options from src and dest + * no longer alias each other afterwards; in particular calling av_opt_free() + * on both src and dest is safe afterwards if dest has been memdup'ed from src. + * + * @param dest Object to copy from + * @param src Object to copy into + * @return 0 on success, negative on error + */ +int av_opt_copy(void *dest, const void *src); + +/** + * Get a default list of allowed ranges for the given option. + * + * This list is constructed without using the AVClass.query_ranges() callback + * and can be used as fallback from within the callback. + * + * @param flags is a bitmask of flags, undefined flags should not be set and should be ignored + * AV_OPT_SEARCH_FAKE_OBJ indicates that the obj is a double pointer to a AVClass instead of a full instance + * AV_OPT_MULTI_COMPONENT_RANGE indicates that function may return more than one component, @see AVOptionRanges + * + * The result must be freed with av_opt_free_ranges. + * + * @return number of compontents returned on success, a negative errro code otherwise + */ +int av_opt_query_ranges_default(AVOptionRanges **, void *obj, const char *key, int flags); + +/** + * Check if given option is set to its default value. + * + * Options o must belong to the obj. This function must not be called to check child's options state. + * @see av_opt_is_set_to_default_by_name(). + * + * @param obj AVClass object to check option on + * @param o option to be checked + * @return >0 when option is set to its default, + * 0 when option is not set its default, + * <0 on error + */ +int av_opt_is_set_to_default(void *obj, const AVOption *o); + +/** + * Check if given option is set to its default value. + * + * @param obj AVClass object to check option on + * @param name option name + * @param search_flags combination of AV_OPT_SEARCH_* + * @return >0 when option is set to its default, + * 0 when option is not set its default, + * <0 on error + */ +int av_opt_is_set_to_default_by_name(void *obj, const char *name, int search_flags); + + +#define AV_OPT_SERIALIZE_SKIP_DEFAULTS 0x00000001 ///< Serialize options that are not set to default values only. +#define AV_OPT_SERIALIZE_OPT_FLAGS_EXACT 0x00000002 ///< Serialize options that exactly match opt_flags only. + +/** + * Serialize object's options. + * + * Create a string containing object's serialized options. + * Such string may be passed back to av_opt_set_from_string() in order to restore option values. + * A key/value or pairs separator occurring in the serialized value or + * name string are escaped through the av_escape() function. + * + * @param[in] obj AVClass object to serialize + * @param[in] opt_flags serialize options with all the specified flags set (AV_OPT_FLAG) + * @param[in] flags combination of AV_OPT_SERIALIZE_* flags + * @param[out] buffer Pointer to buffer that will be allocated with string containg serialized options. + * Buffer must be freed by the caller when is no longer needed. + * @param[in] key_val_sep character used to separate key from value + * @param[in] pairs_sep character used to separate two pairs from each other + * @return >= 0 on success, negative on error + * @warning Separators cannot be neither '\\' nor '\0'. They also cannot be the same. + */ +int av_opt_serialize(void *obj, int opt_flags, int flags, char **buffer, + const char key_val_sep, const char pairs_sep); +/** + * @} + */ + +#endif /* AVUTIL_OPT_H */ diff --git a/media/ffvpx/libavutil/parseutils.c b/media/ffvpx/libavutil/parseutils.c new file mode 100644 index 0000000000..c2916c458b --- /dev/null +++ b/media/ffvpx/libavutil/parseutils.c @@ -0,0 +1,790 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * misc parsing utilities + */ + +#include <time.h> + +#include "avstring.h" +#include "avutil.h" +#include "common.h" +#include "eval.h" +#include "log.h" +/* #include "random_seed.h" */ +#include "time_internal.h" +#include "parseutils.h" +#include "fftime.h" + +#ifdef TEST + +#define av_get_random_seed av_get_random_seed_deterministic +static uint32_t av_get_random_seed_deterministic(void); + +#define av_gettime() 1331972053200000 + +#endif + +int av_parse_ratio(AVRational *q, const char *str, int max, + int log_offset, void *log_ctx) +{ + char c; + int ret; + + if (sscanf(str, "%d:%d%c", &q->num, &q->den, &c) != 2) { + double d; + ret = av_expr_parse_and_eval(&d, str, NULL, NULL, + NULL, NULL, NULL, NULL, + NULL, log_offset, log_ctx); + if (ret < 0) + return ret; + *q = av_d2q(d, max); + } else { + av_reduce(&q->num, &q->den, q->num, q->den, max); + } + + return 0; +} + +typedef struct VideoSizeAbbr { + const char *abbr; + int width, height; +} VideoSizeAbbr; + +typedef struct VideoRateAbbr { + const char *abbr; + AVRational rate; +} VideoRateAbbr; + +static const VideoSizeAbbr video_size_abbrs[] = { + { "ntsc", 720, 480 }, + { "pal", 720, 576 }, + { "qntsc", 352, 240 }, /* VCD compliant NTSC */ + { "qpal", 352, 288 }, /* VCD compliant PAL */ + { "sntsc", 640, 480 }, /* square pixel NTSC */ + { "spal", 768, 576 }, /* square pixel PAL */ + { "film", 352, 240 }, + { "ntsc-film", 352, 240 }, + { "sqcif", 128, 96 }, + { "qcif", 176, 144 }, + { "cif", 352, 288 }, + { "4cif", 704, 576 }, + { "16cif", 1408,1152 }, + { "qqvga", 160, 120 }, + { "qvga", 320, 240 }, + { "vga", 640, 480 }, + { "svga", 800, 600 }, + { "xga", 1024, 768 }, + { "uxga", 1600,1200 }, + { "qxga", 2048,1536 }, + { "sxga", 1280,1024 }, + { "qsxga", 2560,2048 }, + { "hsxga", 5120,4096 }, + { "wvga", 852, 480 }, + { "wxga", 1366, 768 }, + { "wsxga", 1600,1024 }, + { "wuxga", 1920,1200 }, + { "woxga", 2560,1600 }, + { "wqhd", 2560,1440 }, + { "wqsxga", 3200,2048 }, + { "wquxga", 3840,2400 }, + { "whsxga", 6400,4096 }, + { "whuxga", 7680,4800 }, + { "cga", 320, 200 }, + { "ega", 640, 350 }, + { "hd480", 852, 480 }, + { "hd720", 1280, 720 }, + { "hd1080", 1920,1080 }, + { "quadhd", 2560,1440 }, + { "2k", 2048,1080 }, /* Digital Cinema System Specification */ + { "2kdci", 2048,1080 }, + { "2kflat", 1998,1080 }, + { "2kscope", 2048, 858 }, + { "4k", 4096,2160 }, /* Digital Cinema System Specification */ + { "4kdci", 4096,2160 }, + { "4kflat", 3996,2160 }, + { "4kscope", 4096,1716 }, + { "nhd", 640,360 }, + { "hqvga", 240,160 }, + { "wqvga", 400,240 }, + { "fwqvga", 432,240 }, + { "hvga", 480,320 }, + { "qhd", 960,540 }, + { "uhd2160", 3840,2160 }, + { "uhd4320", 7680,4320 }, +}; + +static const VideoRateAbbr video_rate_abbrs[]= { + { "ntsc", { 30000, 1001 } }, + { "pal", { 25, 1 } }, + { "qntsc", { 30000, 1001 } }, /* VCD compliant NTSC */ + { "qpal", { 25, 1 } }, /* VCD compliant PAL */ + { "sntsc", { 30000, 1001 } }, /* square pixel NTSC */ + { "spal", { 25, 1 } }, /* square pixel PAL */ + { "film", { 24, 1 } }, + { "ntsc-film", { 24000, 1001 } }, +}; + +static const char *months[12] = { + "january", "february", "march", "april", "may", "june", "july", "august", + "september", "october", "november", "december" +}; + +int av_parse_video_size(int *width_ptr, int *height_ptr, const char *str) +{ + int i; + int n = FF_ARRAY_ELEMS(video_size_abbrs); + const char *p; + int width = 0, height = 0; + + for (i = 0; i < n; i++) { + if (!strcmp(video_size_abbrs[i].abbr, str)) { + width = video_size_abbrs[i].width; + height = video_size_abbrs[i].height; + break; + } + } + if (i == n) { + width = strtol(str, (void*)&p, 10); + if (*p) + p++; + height = strtol(p, (void*)&p, 10); + + /* trailing extraneous data detected, like in 123x345foobar */ + if (*p) + return AVERROR(EINVAL); + } + if (width <= 0 || height <= 0) + return AVERROR(EINVAL); + *width_ptr = width; + *height_ptr = height; + return 0; +} + +int av_parse_video_rate(AVRational *rate, const char *arg) +{ + int i, ret; + int n = FF_ARRAY_ELEMS(video_rate_abbrs); + + /* First, we check our abbreviation table */ + for (i = 0; i < n; ++i) + if (!strcmp(video_rate_abbrs[i].abbr, arg)) { + *rate = video_rate_abbrs[i].rate; + return 0; + } + + /* Then, we try to parse it as fraction */ + if ((ret = av_parse_ratio_quiet(rate, arg, 1001000)) < 0) + return ret; + if (rate->num <= 0 || rate->den <= 0) + return AVERROR(EINVAL); + return 0; +} + +typedef struct ColorEntry { + const char *name; ///< a string representing the name of the color + uint8_t rgb_color[3]; ///< RGB values for the color +} ColorEntry; + +static const ColorEntry color_table[] = { + { "AliceBlue", { 0xF0, 0xF8, 0xFF } }, + { "AntiqueWhite", { 0xFA, 0xEB, 0xD7 } }, + { "Aqua", { 0x00, 0xFF, 0xFF } }, + { "Aquamarine", { 0x7F, 0xFF, 0xD4 } }, + { "Azure", { 0xF0, 0xFF, 0xFF } }, + { "Beige", { 0xF5, 0xF5, 0xDC } }, + { "Bisque", { 0xFF, 0xE4, 0xC4 } }, + { "Black", { 0x00, 0x00, 0x00 } }, + { "BlanchedAlmond", { 0xFF, 0xEB, 0xCD } }, + { "Blue", { 0x00, 0x00, 0xFF } }, + { "BlueViolet", { 0x8A, 0x2B, 0xE2 } }, + { "Brown", { 0xA5, 0x2A, 0x2A } }, + { "BurlyWood", { 0xDE, 0xB8, 0x87 } }, + { "CadetBlue", { 0x5F, 0x9E, 0xA0 } }, + { "Chartreuse", { 0x7F, 0xFF, 0x00 } }, + { "Chocolate", { 0xD2, 0x69, 0x1E } }, + { "Coral", { 0xFF, 0x7F, 0x50 } }, + { "CornflowerBlue", { 0x64, 0x95, 0xED } }, + { "Cornsilk", { 0xFF, 0xF8, 0xDC } }, + { "Crimson", { 0xDC, 0x14, 0x3C } }, + { "Cyan", { 0x00, 0xFF, 0xFF } }, + { "DarkBlue", { 0x00, 0x00, 0x8B } }, + { "DarkCyan", { 0x00, 0x8B, 0x8B } }, + { "DarkGoldenRod", { 0xB8, 0x86, 0x0B } }, + { "DarkGray", { 0xA9, 0xA9, 0xA9 } }, + { "DarkGreen", { 0x00, 0x64, 0x00 } }, + { "DarkKhaki", { 0xBD, 0xB7, 0x6B } }, + { "DarkMagenta", { 0x8B, 0x00, 0x8B } }, + { "DarkOliveGreen", { 0x55, 0x6B, 0x2F } }, + { "Darkorange", { 0xFF, 0x8C, 0x00 } }, + { "DarkOrchid", { 0x99, 0x32, 0xCC } }, + { "DarkRed", { 0x8B, 0x00, 0x00 } }, + { "DarkSalmon", { 0xE9, 0x96, 0x7A } }, + { "DarkSeaGreen", { 0x8F, 0xBC, 0x8F } }, + { "DarkSlateBlue", { 0x48, 0x3D, 0x8B } }, + { "DarkSlateGray", { 0x2F, 0x4F, 0x4F } }, + { "DarkTurquoise", { 0x00, 0xCE, 0xD1 } }, + { "DarkViolet", { 0x94, 0x00, 0xD3 } }, + { "DeepPink", { 0xFF, 0x14, 0x93 } }, + { "DeepSkyBlue", { 0x00, 0xBF, 0xFF } }, + { "DimGray", { 0x69, 0x69, 0x69 } }, + { "DodgerBlue", { 0x1E, 0x90, 0xFF } }, + { "FireBrick", { 0xB2, 0x22, 0x22 } }, + { "FloralWhite", { 0xFF, 0xFA, 0xF0 } }, + { "ForestGreen", { 0x22, 0x8B, 0x22 } }, + { "Fuchsia", { 0xFF, 0x00, 0xFF } }, + { "Gainsboro", { 0xDC, 0xDC, 0xDC } }, + { "GhostWhite", { 0xF8, 0xF8, 0xFF } }, + { "Gold", { 0xFF, 0xD7, 0x00 } }, + { "GoldenRod", { 0xDA, 0xA5, 0x20 } }, + { "Gray", { 0x80, 0x80, 0x80 } }, + { "Green", { 0x00, 0x80, 0x00 } }, + { "GreenYellow", { 0xAD, 0xFF, 0x2F } }, + { "HoneyDew", { 0xF0, 0xFF, 0xF0 } }, + { "HotPink", { 0xFF, 0x69, 0xB4 } }, + { "IndianRed", { 0xCD, 0x5C, 0x5C } }, + { "Indigo", { 0x4B, 0x00, 0x82 } }, + { "Ivory", { 0xFF, 0xFF, 0xF0 } }, + { "Khaki", { 0xF0, 0xE6, 0x8C } }, + { "Lavender", { 0xE6, 0xE6, 0xFA } }, + { "LavenderBlush", { 0xFF, 0xF0, 0xF5 } }, + { "LawnGreen", { 0x7C, 0xFC, 0x00 } }, + { "LemonChiffon", { 0xFF, 0xFA, 0xCD } }, + { "LightBlue", { 0xAD, 0xD8, 0xE6 } }, + { "LightCoral", { 0xF0, 0x80, 0x80 } }, + { "LightCyan", { 0xE0, 0xFF, 0xFF } }, + { "LightGoldenRodYellow", { 0xFA, 0xFA, 0xD2 } }, + { "LightGreen", { 0x90, 0xEE, 0x90 } }, + { "LightGrey", { 0xD3, 0xD3, 0xD3 } }, + { "LightPink", { 0xFF, 0xB6, 0xC1 } }, + { "LightSalmon", { 0xFF, 0xA0, 0x7A } }, + { "LightSeaGreen", { 0x20, 0xB2, 0xAA } }, + { "LightSkyBlue", { 0x87, 0xCE, 0xFA } }, + { "LightSlateGray", { 0x77, 0x88, 0x99 } }, + { "LightSteelBlue", { 0xB0, 0xC4, 0xDE } }, + { "LightYellow", { 0xFF, 0xFF, 0xE0 } }, + { "Lime", { 0x00, 0xFF, 0x00 } }, + { "LimeGreen", { 0x32, 0xCD, 0x32 } }, + { "Linen", { 0xFA, 0xF0, 0xE6 } }, + { "Magenta", { 0xFF, 0x00, 0xFF } }, + { "Maroon", { 0x80, 0x00, 0x00 } }, + { "MediumAquaMarine", { 0x66, 0xCD, 0xAA } }, + { "MediumBlue", { 0x00, 0x00, 0xCD } }, + { "MediumOrchid", { 0xBA, 0x55, 0xD3 } }, + { "MediumPurple", { 0x93, 0x70, 0xD8 } }, + { "MediumSeaGreen", { 0x3C, 0xB3, 0x71 } }, + { "MediumSlateBlue", { 0x7B, 0x68, 0xEE } }, + { "MediumSpringGreen", { 0x00, 0xFA, 0x9A } }, + { "MediumTurquoise", { 0x48, 0xD1, 0xCC } }, + { "MediumVioletRed", { 0xC7, 0x15, 0x85 } }, + { "MidnightBlue", { 0x19, 0x19, 0x70 } }, + { "MintCream", { 0xF5, 0xFF, 0xFA } }, + { "MistyRose", { 0xFF, 0xE4, 0xE1 } }, + { "Moccasin", { 0xFF, 0xE4, 0xB5 } }, + { "NavajoWhite", { 0xFF, 0xDE, 0xAD } }, + { "Navy", { 0x00, 0x00, 0x80 } }, + { "OldLace", { 0xFD, 0xF5, 0xE6 } }, + { "Olive", { 0x80, 0x80, 0x00 } }, + { "OliveDrab", { 0x6B, 0x8E, 0x23 } }, + { "Orange", { 0xFF, 0xA5, 0x00 } }, + { "OrangeRed", { 0xFF, 0x45, 0x00 } }, + { "Orchid", { 0xDA, 0x70, 0xD6 } }, + { "PaleGoldenRod", { 0xEE, 0xE8, 0xAA } }, + { "PaleGreen", { 0x98, 0xFB, 0x98 } }, + { "PaleTurquoise", { 0xAF, 0xEE, 0xEE } }, + { "PaleVioletRed", { 0xD8, 0x70, 0x93 } }, + { "PapayaWhip", { 0xFF, 0xEF, 0xD5 } }, + { "PeachPuff", { 0xFF, 0xDA, 0xB9 } }, + { "Peru", { 0xCD, 0x85, 0x3F } }, + { "Pink", { 0xFF, 0xC0, 0xCB } }, + { "Plum", { 0xDD, 0xA0, 0xDD } }, + { "PowderBlue", { 0xB0, 0xE0, 0xE6 } }, + { "Purple", { 0x80, 0x00, 0x80 } }, + { "Red", { 0xFF, 0x00, 0x00 } }, + { "RosyBrown", { 0xBC, 0x8F, 0x8F } }, + { "RoyalBlue", { 0x41, 0x69, 0xE1 } }, + { "SaddleBrown", { 0x8B, 0x45, 0x13 } }, + { "Salmon", { 0xFA, 0x80, 0x72 } }, + { "SandyBrown", { 0xF4, 0xA4, 0x60 } }, + { "SeaGreen", { 0x2E, 0x8B, 0x57 } }, + { "SeaShell", { 0xFF, 0xF5, 0xEE } }, + { "Sienna", { 0xA0, 0x52, 0x2D } }, + { "Silver", { 0xC0, 0xC0, 0xC0 } }, + { "SkyBlue", { 0x87, 0xCE, 0xEB } }, + { "SlateBlue", { 0x6A, 0x5A, 0xCD } }, + { "SlateGray", { 0x70, 0x80, 0x90 } }, + { "Snow", { 0xFF, 0xFA, 0xFA } }, + { "SpringGreen", { 0x00, 0xFF, 0x7F } }, + { "SteelBlue", { 0x46, 0x82, 0xB4 } }, + { "Tan", { 0xD2, 0xB4, 0x8C } }, + { "Teal", { 0x00, 0x80, 0x80 } }, + { "Thistle", { 0xD8, 0xBF, 0xD8 } }, + { "Tomato", { 0xFF, 0x63, 0x47 } }, + { "Turquoise", { 0x40, 0xE0, 0xD0 } }, + { "Violet", { 0xEE, 0x82, 0xEE } }, + { "Wheat", { 0xF5, 0xDE, 0xB3 } }, + { "White", { 0xFF, 0xFF, 0xFF } }, + { "WhiteSmoke", { 0xF5, 0xF5, 0xF5 } }, + { "Yellow", { 0xFF, 0xFF, 0x00 } }, + { "YellowGreen", { 0x9A, 0xCD, 0x32 } }, +}; + +static int color_table_compare(const void *lhs, const void *rhs) +{ + return av_strcasecmp(lhs, ((const ColorEntry *)rhs)->name); +} + +#define ALPHA_SEP '@' + +int av_parse_color(uint8_t *rgba_color, const char *color_string, int slen, + void *log_ctx) +{ + char *tail, color_string2[128]; + const ColorEntry *entry; + int len, hex_offset = 0; + + if (color_string[0] == '#') { + hex_offset = 1; + } else if (!strncmp(color_string, "0x", 2)) + hex_offset = 2; + + if (slen < 0) + slen = strlen(color_string); + av_strlcpy(color_string2, color_string + hex_offset, + FFMIN(slen-hex_offset+1, sizeof(color_string2))); + if ((tail = strchr(color_string2, ALPHA_SEP))) + *tail++ = 0; + len = strlen(color_string2); + rgba_color[3] = 255; + + if (!av_strcasecmp(color_string2, "random") || !av_strcasecmp(color_string2, "bikeshed")) { + int rgba = 0xffffffff; /* av_get_random_seed(); */ + rgba_color[0] = rgba >> 24; + rgba_color[1] = rgba >> 16; + rgba_color[2] = rgba >> 8; + rgba_color[3] = rgba; + } else if (hex_offset || + strspn(color_string2, "0123456789ABCDEFabcdef") == len) { + char *tail; + unsigned int rgba = strtoul(color_string2, &tail, 16); + + if (*tail || (len != 6 && len != 8)) { + av_log(log_ctx, AV_LOG_ERROR, "Invalid 0xRRGGBB[AA] color string: '%s'\n", color_string2); + return AVERROR(EINVAL); + } + if (len == 8) { + rgba_color[3] = rgba; + rgba >>= 8; + } + rgba_color[0] = rgba >> 16; + rgba_color[1] = rgba >> 8; + rgba_color[2] = rgba; + } else { + entry = bsearch(color_string2, + color_table, + FF_ARRAY_ELEMS(color_table), + sizeof(ColorEntry), + color_table_compare); + if (!entry) { + av_log(log_ctx, AV_LOG_ERROR, "Cannot find color '%s'\n", color_string2); + return AVERROR(EINVAL); + } + memcpy(rgba_color, entry->rgb_color, 3); + } + + if (tail) { + double alpha; + const char *alpha_string = tail; + if (!strncmp(alpha_string, "0x", 2)) { + alpha = strtoul(alpha_string, &tail, 16); + } else { + double norm_alpha = strtod(alpha_string, &tail); + if (norm_alpha < 0.0 || norm_alpha > 1.0) + alpha = 256; + else + alpha = 255 * norm_alpha; + } + + if (tail == alpha_string || *tail || alpha > 255 || alpha < 0) { + av_log(log_ctx, AV_LOG_ERROR, "Invalid alpha value specifier '%s' in '%s'\n", + alpha_string, color_string); + return AVERROR(EINVAL); + } + rgba_color[3] = alpha; + } + + return 0; +} + +const char *av_get_known_color_name(int color_idx, const uint8_t **rgbp) +{ + const ColorEntry *color; + + if ((unsigned)color_idx >= FF_ARRAY_ELEMS(color_table)) + return NULL; + + color = &color_table[color_idx]; + if (rgbp) + *rgbp = color->rgb_color; + + return color->name; +} + +/* get a positive number between n_min and n_max, for a maximum length + of len_max. Return -1 if error. */ +static int date_get_num(const char **pp, + int n_min, int n_max, int len_max) +{ + int i, val, c; + const char *p; + + p = *pp; + val = 0; + for(i = 0; i < len_max; i++) { + c = *p; + if (!av_isdigit(c)) + break; + val = (val * 10) + c - '0'; + p++; + } + /* no number read ? */ + if (p == *pp) + return -1; + if (val < n_min || val > n_max) + return -1; + *pp = p; + return val; +} + +static int date_get_month(const char **pp) { + int i = 0; + for (; i < 12; i++) { + if (!av_strncasecmp(*pp, months[i], 3)) { + const char *mo_full = months[i] + 3; + int len = strlen(mo_full); + *pp += 3; + if (len > 0 && !av_strncasecmp(*pp, mo_full, len)) + *pp += len; + return i; + } + } + return -1; +} + +char *av_small_strptime(const char *p, const char *fmt, struct tm *dt) +{ + int c, val; + + while((c = *fmt++)) { + if (c != '%') { + if (av_isspace(c)) + for (; *p && av_isspace(*p); p++); + else if (*p != c) + return NULL; + else p++; + continue; + } + + c = *fmt++; + switch(c) { + case 'H': + case 'J': + val = date_get_num(&p, 0, c == 'H' ? 23 : INT_MAX, c == 'H' ? 2 : 4); + + if (val == -1) + return NULL; + dt->tm_hour = val; + break; + case 'M': + val = date_get_num(&p, 0, 59, 2); + if (val == -1) + return NULL; + dt->tm_min = val; + break; + case 'S': + val = date_get_num(&p, 0, 59, 2); + if (val == -1) + return NULL; + dt->tm_sec = val; + break; + case 'Y': + val = date_get_num(&p, 0, 9999, 4); + if (val == -1) + return NULL; + dt->tm_year = val - 1900; + break; + case 'm': + val = date_get_num(&p, 1, 12, 2); + if (val == -1) + return NULL; + dt->tm_mon = val - 1; + break; + case 'd': + val = date_get_num(&p, 1, 31, 2); + if (val == -1) + return NULL; + dt->tm_mday = val; + break; + case 'T': + p = av_small_strptime(p, "%H:%M:%S", dt); + if (!p) + return NULL; + break; + case 'b': + case 'B': + case 'h': + val = date_get_month(&p); + if (val == -1) + return NULL; + dt->tm_mon = val; + break; + case '%': + if (*p++ != '%') + return NULL; + break; + default: + return NULL; + } + } + + return (char*)p; +} + +time_t av_timegm(struct tm *tm) +{ + time_t t; + + int y = tm->tm_year + 1900, m = tm->tm_mon + 1, d = tm->tm_mday; + + if (m < 3) { + m += 12; + y--; + } + + t = 86400LL * + (d + (153 * m - 457) / 5 + 365 * y + y / 4 - y / 100 + y / 400 - 719469); + + t += 3600 * tm->tm_hour + 60 * tm->tm_min + tm->tm_sec; + + return t; +} + +int av_parse_time(int64_t *timeval, const char *timestr, int duration) +{ + const char *p, *q; + int64_t t, now64; + time_t now; + struct tm dt = { 0 }, tmbuf; + int today = 0, negative = 0, microseconds = 0, suffix = 1000000; + int i; + static const char * const date_fmt[] = { + "%Y - %m - %d", + "%Y%m%d", + }; + static const char * const time_fmt[] = { + "%H:%M:%S", + "%H%M%S", + }; + static const char * const tz_fmt[] = { + "%H:%M", + "%H%M", + "%H", + }; + + p = timestr; + q = NULL; + *timeval = INT64_MIN; + if (!duration) { + now64 = av_gettime(); + now = now64 / 1000000; + + if (!av_strcasecmp(timestr, "now")) { + *timeval = now64; + return 0; + } + + /* parse the year-month-day part */ + for (i = 0; i < FF_ARRAY_ELEMS(date_fmt); i++) { + q = av_small_strptime(p, date_fmt[i], &dt); + if (q) + break; + } + + /* if the year-month-day part is missing, then take the + * current year-month-day time */ + if (!q) { + today = 1; + q = p; + } + p = q; + + if (*p == 'T' || *p == 't') + p++; + else + while (av_isspace(*p)) + p++; + + /* parse the hour-minute-second part */ + for (i = 0; i < FF_ARRAY_ELEMS(time_fmt); i++) { + q = av_small_strptime(p, time_fmt[i], &dt); + if (q) + break; + } + } else { + /* parse timestr as a duration */ + if (p[0] == '-') { + negative = 1; + ++p; + } + /* parse timestr as HH:MM:SS */ + q = av_small_strptime(p, "%J:%M:%S", &dt); + if (!q) { + /* parse timestr as MM:SS */ + q = av_small_strptime(p, "%M:%S", &dt); + dt.tm_hour = 0; + } + if (!q) { + char *o; + /* parse timestr as S+ */ + errno = 0; + t = strtoll(p, &o, 10); + if (o == p) /* the parsing didn't succeed */ + return AVERROR(EINVAL); + if (errno == ERANGE) + return AVERROR(ERANGE); + q = o; + } else { + t = dt.tm_hour * 3600 + dt.tm_min * 60 + dt.tm_sec; + } + } + + /* Now we have all the fields that we can get */ + if (!q) + return AVERROR(EINVAL); + + /* parse the .m... part */ + if (*q == '.') { + int n; + q++; + for (n = 100000; n >= 1; n /= 10, q++) { + if (!av_isdigit(*q)) + break; + microseconds += n * (*q - '0'); + } + while (av_isdigit(*q)) + q++; + } + + if (duration) { + if (q[0] == 'm' && q[1] == 's') { + suffix = 1000; + microseconds /= 1000; + q += 2; + } else if (q[0] == 'u' && q[1] == 's') { + suffix = 1; + microseconds = 0; + q += 2; + } else if (*q == 's') + q++; + } else { + int is_utc = *q == 'Z' || *q == 'z'; + int tzoffset = 0; + q += is_utc; + if (!today && !is_utc && (*q == '+' || *q == '-')) { + struct tm tz = { 0 }; + int sign = (*q == '+' ? -1 : 1); + q++; + p = q; + for (i = 0; i < FF_ARRAY_ELEMS(tz_fmt); i++) { + q = av_small_strptime(p, tz_fmt[i], &tz); + if (q) + break; + } + if (!q) + return AVERROR(EINVAL); + tzoffset = sign * (tz.tm_hour * 60 + tz.tm_min) * 60; + is_utc = 1; + } + if (today) { /* fill in today's date */ + struct tm dt2 = is_utc ? *gmtime_r(&now, &tmbuf) : *localtime_r(&now, &tmbuf); + dt2.tm_hour = dt.tm_hour; + dt2.tm_min = dt.tm_min; + dt2.tm_sec = dt.tm_sec; + dt = dt2; + } + dt.tm_isdst = is_utc ? 0 : -1; + t = is_utc ? av_timegm(&dt) : mktime(&dt); + t += tzoffset; + } + + /* Check that we are at the end of the string */ + if (*q) + return AVERROR(EINVAL); + + if (INT64_MAX / suffix < t || t < INT64_MIN / suffix) + return AVERROR(ERANGE); + t *= suffix; + if (INT64_MAX - microseconds < t) + return AVERROR(ERANGE); + t += microseconds; + if (t == INT64_MIN && negative) + return AVERROR(ERANGE); + *timeval = negative ? -t : t; + return 0; +} + +int av_find_info_tag(char *arg, int arg_size, const char *tag1, const char *info) +{ + const char *p; + char tag[128], *q; + + p = info; + if (*p == '?') + p++; + for(;;) { + q = tag; + while (*p != '\0' && *p != '=' && *p != '&') { + if ((q - tag) < sizeof(tag) - 1) + *q++ = *p; + p++; + } + *q = '\0'; + q = arg; + if (*p == '=') { + p++; + while (*p != '&' && *p != '\0') { + if ((q - arg) < arg_size - 1) { + if (*p == '+') + *q++ = ' '; + else + *q++ = *p; + } + p++; + } + } + *q = '\0'; + if (!strcmp(tag, tag1)) + return 1; + if (*p != '&') + break; + p++; + } + return 0; +} diff --git a/media/ffvpx/libavutil/parseutils.h b/media/ffvpx/libavutil/parseutils.h new file mode 100644 index 0000000000..dad5c2775b --- /dev/null +++ b/media/ffvpx/libavutil/parseutils.h @@ -0,0 +1,197 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_PARSEUTILS_H +#define AVUTIL_PARSEUTILS_H + +#include <time.h> + +#include "rational.h" + +/** + * @file + * misc parsing utilities + */ + +/** + * Parse str and store the parsed ratio in q. + * + * Note that a ratio with infinite (1/0) or negative value is + * considered valid, so you should check on the returned value if you + * want to exclude those values. + * + * The undefined value can be expressed using the "0:0" string. + * + * @param[in,out] q pointer to the AVRational which will contain the ratio + * @param[in] str the string to parse: it has to be a string in the format + * num:den, a float number or an expression + * @param[in] max the maximum allowed numerator and denominator + * @param[in] log_offset log level offset which is applied to the log + * level of log_ctx + * @param[in] log_ctx parent logging context + * @return >= 0 on success, a negative error code otherwise + */ +int av_parse_ratio(AVRational *q, const char *str, int max, + int log_offset, void *log_ctx); + +#define av_parse_ratio_quiet(rate, str, max) \ + av_parse_ratio(rate, str, max, AV_LOG_MAX_OFFSET, NULL) + +/** + * Parse str and put in width_ptr and height_ptr the detected values. + * + * @param[in,out] width_ptr pointer to the variable which will contain the detected + * width value + * @param[in,out] height_ptr pointer to the variable which will contain the detected + * height value + * @param[in] str the string to parse: it has to be a string in the format + * width x height or a valid video size abbreviation. + * @return >= 0 on success, a negative error code otherwise + */ +int av_parse_video_size(int *width_ptr, int *height_ptr, const char *str); + +/** + * Parse str and store the detected values in *rate. + * + * @param[in,out] rate pointer to the AVRational which will contain the detected + * frame rate + * @param[in] str the string to parse: it has to be a string in the format + * rate_num / rate_den, a float number or a valid video rate abbreviation + * @return >= 0 on success, a negative error code otherwise + */ +int av_parse_video_rate(AVRational *rate, const char *str); + +/** + * Put the RGBA values that correspond to color_string in rgba_color. + * + * @param rgba_color 4-elements array of uint8_t values, where the respective + * red, green, blue and alpha component values are written. + * @param color_string a string specifying a color. It can be the name of + * a color (case insensitive match) or a [0x|#]RRGGBB[AA] sequence, + * possibly followed by "@" and a string representing the alpha + * component. + * The alpha component may be a string composed by "0x" followed by an + * hexadecimal number or a decimal number between 0.0 and 1.0, which + * represents the opacity value (0x00/0.0 means completely transparent, + * 0xff/1.0 completely opaque). + * If the alpha component is not specified then 0xff is assumed. + * The string "random" will result in a random color. + * @param slen length of the initial part of color_string containing the + * color. It can be set to -1 if color_string is a null terminated string + * containing nothing else than the color. + * @param log_ctx a pointer to an arbitrary struct of which the first field + * is a pointer to an AVClass struct (used for av_log()). Can be NULL. + * @return >= 0 in case of success, a negative value in case of + * failure (for example if color_string cannot be parsed). + */ +int av_parse_color(uint8_t *rgba_color, const char *color_string, int slen, + void *log_ctx); + +/** + * Get the name of a color from the internal table of hard-coded named + * colors. + * + * This function is meant to enumerate the color names recognized by + * av_parse_color(). + * + * @param color_idx index of the requested color, starting from 0 + * @param rgb if not NULL, will point to a 3-elements array with the color value in RGB + * @return the color name string or NULL if color_idx is not in the array + */ +const char *av_get_known_color_name(int color_idx, const uint8_t **rgb); + +/** + * Parse timestr and return in *time a corresponding number of + * microseconds. + * + * @param timeval puts here the number of microseconds corresponding + * to the string in timestr. If the string represents a duration, it + * is the number of microseconds contained in the time interval. If + * the string is a date, is the number of microseconds since 1st of + * January, 1970 up to the time of the parsed date. If timestr cannot + * be successfully parsed, set *time to INT64_MIN. + + * @param timestr a string representing a date or a duration. + * - If a date the syntax is: + * @code + * [{YYYY-MM-DD|YYYYMMDD}[T|t| ]]{{HH:MM:SS[.m...]]]}|{HHMMSS[.m...]]]}}[Z] + * now + * @endcode + * If the value is "now" it takes the current time. + * Time is local time unless Z is appended, in which case it is + * interpreted as UTC. + * If the year-month-day part is not specified it takes the current + * year-month-day. + * - If a duration the syntax is: + * @code + * [-][HH:]MM:SS[.m...] + * [-]S+[.m...] + * @endcode + * @param duration flag which tells how to interpret timestr, if not + * zero timestr is interpreted as a duration, otherwise as a date + * @return >= 0 in case of success, a negative value corresponding to an + * AVERROR code otherwise + */ +int av_parse_time(int64_t *timeval, const char *timestr, int duration); + +/** + * Attempt to find a specific tag in a URL. + * + * syntax: '?tag1=val1&tag2=val2...'. Little URL decoding is done. + * Return 1 if found. + */ +int av_find_info_tag(char *arg, int arg_size, const char *tag1, const char *info); + +/** + * Simplified version of strptime + * + * Parse the input string p according to the format string fmt and + * store its results in the structure dt. + * This implementation supports only a subset of the formats supported + * by the standard strptime(). + * + * The supported input field descriptors are listed below. + * - `%%H`: the hour as a decimal number, using a 24-hour clock, in the + * range '00' through '23' + * - `%%J`: hours as a decimal number, in the range '0' through INT_MAX + * - `%%M`: the minute as a decimal number, using a 24-hour clock, in the + * range '00' through '59' + * - `%%S`: the second as a decimal number, using a 24-hour clock, in the + * range '00' through '59' + * - `%%Y`: the year as a decimal number, using the Gregorian calendar + * - `%%m`: the month as a decimal number, in the range '1' through '12' + * - `%%d`: the day of the month as a decimal number, in the range '1' + * through '31' + * - `%%T`: alias for `%%H:%%M:%%S` + * - `%%`: a literal `%` + * + * @return a pointer to the first character not processed in this function + * call. In case the input string contains more characters than + * required by the format string the return value points right after + * the last consumed input character. In case the whole input string + * is consumed the return value points to the null byte at the end of + * the string. On failure NULL is returned. + */ +char *av_small_strptime(const char *p, const char *fmt, struct tm *dt); + +/** + * Convert the decomposed UTC time in tm to a time_t value. + */ +time_t av_timegm(struct tm *tm); + +#endif /* AVUTIL_PARSEUTILS_H */ diff --git a/media/ffvpx/libavutil/pixdesc.c b/media/ffvpx/libavutil/pixdesc.c new file mode 100644 index 0000000000..f6d4d01460 --- /dev/null +++ b/media/ffvpx/libavutil/pixdesc.c @@ -0,0 +1,3407 @@ +/* + * pixel format descriptor + * Copyright (c) 2009 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdio.h> +#include <string.h> + +#include "avstring.h" +#include "common.h" +#include "pixfmt.h" +#include "pixdesc.h" +#include "intreadwrite.h" + +void av_read_image_line2(void *dst, + const uint8_t *data[4], const int linesize[4], + const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, + int read_pal_component, + int dst_element_size) +{ + AVComponentDescriptor comp = desc->comp[c]; + int plane = comp.plane; + int depth = comp.depth; + unsigned mask = (1ULL << depth) - 1; + int shift = comp.shift; + int step = comp.step; + int flags = desc->flags; + uint16_t *dst16 = dst; + uint32_t *dst32 = dst; + + if (flags & AV_PIX_FMT_FLAG_BITSTREAM) { + if (depth == 10) { + // Assume all channels are packed into a 32bit value + const uint8_t *byte_p = data[plane] + y * linesize[plane]; + const uint32_t *p = (uint32_t *)byte_p; + + while (w--) { + int val = AV_RB32(p); + val = (val >> comp.offset) & mask; + if (read_pal_component) + val = data[1][4*val + c]; + if (dst_element_size == 4) *dst32++ = val; + else *dst16++ = val; + p++; + } + } else { + int skip = x * step + comp.offset; + const uint8_t *p = data[plane] + y * linesize[plane] + (skip >> 3); + int shift = 8 - depth - (skip & 7); + + while (w--) { + int val = (*p >> shift) & mask; + if (read_pal_component) + val = data[1][4*val + c]; + shift -= step; + p -= shift >> 3; + shift &= 7; + if (dst_element_size == 4) *dst32++ = val; + else *dst16++ = val; + } + } + } else { + const uint8_t *p = data[plane] + y * linesize[plane] + + x * step + comp.offset; + int is_8bit = shift + depth <= 8; + int is_16bit= shift + depth <=16; + + if (is_8bit) + p += !!(flags & AV_PIX_FMT_FLAG_BE); + + while (w--) { + unsigned val; + if (is_8bit) val = *p; + else if(is_16bit) val = flags & AV_PIX_FMT_FLAG_BE ? AV_RB16(p) : AV_RL16(p); + else val = flags & AV_PIX_FMT_FLAG_BE ? AV_RB32(p) : AV_RL32(p); + val = (val >> shift) & mask; + if (read_pal_component) + val = data[1][4 * val + c]; + p += step; + if (dst_element_size == 4) *dst32++ = val; + else *dst16++ = val; + } + } +} + +void av_read_image_line(uint16_t *dst, + const uint8_t *data[4], const int linesize[4], + const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, + int read_pal_component) +{ + av_read_image_line2(dst, data, linesize, desc,x, y, c, w, + read_pal_component, + 2); +} + +void av_write_image_line2(const void *src, + uint8_t *data[4], const int linesize[4], + const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, int src_element_size) +{ + AVComponentDescriptor comp = desc->comp[c]; + int plane = comp.plane; + int depth = comp.depth; + int step = comp.step; + int flags = desc->flags; + const uint32_t *src32 = src; + const uint16_t *src16 = src; + + if (flags & AV_PIX_FMT_FLAG_BITSTREAM) { + if (depth == 10) { + // Assume all channels are packed into a 32bit value + const uint8_t *byte_p = data[plane] + y * linesize[plane]; + uint32_t *p = (uint32_t *)byte_p; + int offset = comp.offset; + uint32_t mask = ((1ULL << depth) - 1) << offset; + + while (w--) { + uint16_t val = src_element_size == 4 ? *src32++ : *src16++; + AV_WB32(p, (AV_RB32(p) & ~mask) | (val << offset)); + p++; + } + } else { + int skip = x * step + comp.offset; + uint8_t *p = data[plane] + y * linesize[plane] + (skip >> 3); + int shift = 8 - depth - (skip & 7); + + while (w--) { + *p |= (src_element_size == 4 ? *src32++ : *src16++) << shift; + shift -= step; + p -= shift >> 3; + shift &= 7; + } + } + } else { + int shift = comp.shift; + uint8_t *p = data[plane] + y * linesize[plane] + + x * step + comp.offset; + + if (shift + depth <= 8) { + p += !!(flags & AV_PIX_FMT_FLAG_BE); + while (w--) { + *p |= ((src_element_size == 4 ? *src32++ : *src16++) << shift); + p += step; + } + } else { + while (w--) { + unsigned s = (src_element_size == 4 ? *src32++ : *src16++); + if (shift + depth <= 16) { + if (flags & AV_PIX_FMT_FLAG_BE) { + uint16_t val = AV_RB16(p) | (s << shift); + AV_WB16(p, val); + } else { + uint16_t val = AV_RL16(p) | (s << shift); + AV_WL16(p, val); + } + } else { + if (flags & AV_PIX_FMT_FLAG_BE) { + uint32_t val = AV_RB32(p) | (s << shift); + AV_WB32(p, val); + } else { + uint32_t val = AV_RL32(p) | (s << shift); + AV_WL32(p, val); + } + } + p += step; + } + } + } +} + +void av_write_image_line(const uint16_t *src, + uint8_t *data[4], const int linesize[4], + const AVPixFmtDescriptor *desc, + int x, int y, int c, int w) +{ + av_write_image_line2(src, data, linesize, desc, x, y, c, w, 2); +} + +static const AVPixFmtDescriptor av_pix_fmt_descriptors[AV_PIX_FMT_NB] = { + [AV_PIX_FMT_YUV420P] = { + .name = "yuv420p", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUYV422] = { + .name = "yuyv422", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 8 }, /* Y */ + { 0, 4, 1, 0, 8 }, /* U */ + { 0, 4, 3, 0, 8 }, /* V */ + }, + }, + [AV_PIX_FMT_YVYU422] = { + .name = "yvyu422", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 8 }, /* Y */ + { 0, 4, 3, 0, 8 }, /* U */ + { 0, 4, 1, 0, 8 }, /* V */ + }, + }, + [AV_PIX_FMT_Y210LE] = { + .name = "y210le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 6, 10 }, /* Y */ + { 0, 8, 2, 6, 10 }, /* U */ + { 0, 8, 6, 6, 10 }, /* V */ + }, + }, + [AV_PIX_FMT_Y210BE] = { + .name = "y210be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 6, 10 }, /* Y */ + { 0, 8, 2, 6, 10 }, /* U */ + { 0, 8, 6, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_RGB24] = { + .name = "rgb24", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 3, 0, 0, 8 }, /* R */ + { 0, 3, 1, 0, 8 }, /* G */ + { 0, 3, 2, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR24] = { + .name = "bgr24", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 3, 2, 0, 8 }, /* R */ + { 0, 3, 1, 0, 8 }, /* G */ + { 0, 3, 0, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_X2RGB10LE] = { + .name = "x2rgb10le", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 2, 4, 10 }, /* R */ + { 0, 4, 1, 2, 10 }, /* G */ + { 0, 4, 0, 0, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_X2RGB10BE] = { + .name = "x2rgb10be", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 0, 4, 10 }, /* R */ + { 0, 4, 1, 2, 10 }, /* G */ + { 0, 4, 2, 0, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_X2BGR10LE] = { + .name = "x2bgr10le", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 0, 0, 10 }, /* R */ + { 0, 4, 1, 2, 10 }, /* G */ + { 0, 4, 2, 4, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_X2BGR10BE] = { + .name = "x2bgr10be", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 2, 0, 10 }, /* R */ + { 0, 4, 1, 2, 10 }, /* G */ + { 0, 4, 0, 4, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_YUV422P] = { + .name = "yuv422p", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P] = { + .name = "yuv444p", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV410P] = { + .name = "yuv410p", + .nb_components = 3, + .log2_chroma_w = 2, + .log2_chroma_h = 2, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV411P] = { + .name = "yuv411p", + .nb_components = 3, + .log2_chroma_w = 2, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUVJ411P] = { + .name = "yuvj411p", + .nb_components = 3, + .log2_chroma_w = 2, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_GRAY8] = { + .name = "gray", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + }, + .alias = "gray8,y8", + }, + [AV_PIX_FMT_MONOWHITE] = { + .name = "monow", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 1 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BITSTREAM, + }, + [AV_PIX_FMT_MONOBLACK] = { + .name = "monob", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 7, 1 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BITSTREAM, + }, + [AV_PIX_FMT_PAL8] = { + .name = "pal8", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, + }, + .flags = AV_PIX_FMT_FLAG_PAL | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVJ420P] = { + .name = "yuvj420p", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUVJ422P] = { + .name = "yuvj422p", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUVJ444P] = { + .name = "yuvj444p", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, +#if FF_API_XVMC + [AV_PIX_FMT_XVMC] = { + .name = "xvmc", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, +#endif + [AV_PIX_FMT_UYVY422] = { + .name = "uyvy422", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 1, 0, 8 }, /* Y */ + { 0, 4, 0, 0, 8 }, /* U */ + { 0, 4, 2, 0, 8 }, /* V */ + }, + }, + [AV_PIX_FMT_UYYVYY411] = { + .name = "uyyvyy411", + .nb_components = 3, + .log2_chroma_w = 2, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 1, 0, 8 }, /* Y */ + { 0, 6, 0, 0, 8 }, /* U */ + { 0, 6, 3, 0, 8 }, /* V */ + }, + }, + [AV_PIX_FMT_BGR8] = { + .name = "bgr8", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 3 }, /* R */ + { 0, 1, 0, 3, 3 }, /* G */ + { 0, 1, 0, 6, 2 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR4] = { + .name = "bgr4", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 3, 0, 1 }, /* R */ + { 0, 4, 1, 0, 2 }, /* G */ + { 0, 4, 0, 0, 1 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BITSTREAM | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR4_BYTE] = { + .name = "bgr4_byte", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 1 }, /* R */ + { 0, 1, 0, 1, 2 }, /* G */ + { 0, 1, 0, 3, 1 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB8] = { + .name = "rgb8", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 5, 3 }, /* R */ + { 0, 1, 0, 2, 3 }, /* G */ + { 0, 1, 0, 0, 2 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB4] = { + .name = "rgb4", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 0, 1 }, /* R */ + { 0, 4, 1, 0, 2 }, /* G */ + { 0, 4, 3, 0, 1 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BITSTREAM | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB4_BYTE] = { + .name = "rgb4_byte", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 3, 1 }, /* R */ + { 0, 1, 0, 1, 2 }, /* G */ + { 0, 1, 0, 0, 1 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_NV12] = { + .name = "nv12", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 2, 0, 0, 8 }, /* U */ + { 1, 2, 1, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_NV21] = { + .name = "nv21", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 2, 1, 0, 8 }, /* U */ + { 1, 2, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_ARGB] = { + .name = "argb", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 1, 0, 8 }, /* R */ + { 0, 4, 2, 0, 8 }, /* G */ + { 0, 4, 3, 0, 8 }, /* B */ + { 0, 4, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_RGBA] = { + .name = "rgba", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 0, 8 }, /* R */ + { 0, 4, 1, 0, 8 }, /* G */ + { 0, 4, 2, 0, 8 }, /* B */ + { 0, 4, 3, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_ABGR] = { + .name = "abgr", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 3, 0, 8 }, /* R */ + { 0, 4, 2, 0, 8 }, /* G */ + { 0, 4, 1, 0, 8 }, /* B */ + { 0, 4, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_BGRA] = { + .name = "bgra", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 2, 0, 8 }, /* R */ + { 0, 4, 1, 0, 8 }, /* G */ + { 0, 4, 0, 0, 8 }, /* B */ + { 0, 4, 3, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_0RGB] = { + .name = "0rgb", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 1, 0, 8 }, /* R */ + { 0, 4, 2, 0, 8 }, /* G */ + { 0, 4, 3, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB0] = { + .name = "rgb0", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 0, 0, 8 }, /* R */ + { 0, 4, 1, 0, 8 }, /* G */ + { 0, 4, 2, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_0BGR] = { + .name = "0bgr", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 3, 0, 8 }, /* R */ + { 0, 4, 2, 0, 8 }, /* G */ + { 0, 4, 1, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR0] = { + .name = "bgr0", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 2, 0, 8 }, /* R */ + { 0, 4, 1, 0, 8 }, /* G */ + { 0, 4, 0, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GRAY9BE] = { + .name = "gray9be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + .alias = "y9be", + }, + [AV_PIX_FMT_GRAY9LE] = { + .name = "gray9le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + }, + .alias = "y9le", + }, + [AV_PIX_FMT_GRAY10BE] = { + .name = "gray10be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + .alias = "y10be", + }, + [AV_PIX_FMT_GRAY10LE] = { + .name = "gray10le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + }, + .alias = "y10le", + }, + [AV_PIX_FMT_GRAY12BE] = { + .name = "gray12be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + .alias = "y12be", + }, + [AV_PIX_FMT_GRAY12LE] = { + .name = "gray12le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + }, + .alias = "y12le", + }, + [AV_PIX_FMT_GRAY14BE] = { + .name = "gray14be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + .alias = "y14be", + }, + [AV_PIX_FMT_GRAY14LE] = { + .name = "gray14le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + }, + .alias = "y14le", + }, + [AV_PIX_FMT_GRAY16BE] = { + .name = "gray16be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + .alias = "y16be", + }, + [AV_PIX_FMT_GRAY16LE] = { + .name = "gray16le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + }, + .alias = "y16le", + }, + [AV_PIX_FMT_YUV440P] = { + .name = "yuv440p", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUVJ440P] = { + .name = "yuvj440p", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV440P10LE] = { + .name = "yuv440p10le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV440P10BE] = { + .name = "yuv440p10be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV440P12LE] = { + .name = "yuv440p12le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV440P12BE] = { + .name = "yuv440p12be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUVA420P] = { + .name = "yuva420p", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + { 3, 1, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P] = { + .name = "yuva422p", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + { 3, 1, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P] = { + .name = "yuva444p", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 1, 0, 0, 8 }, /* U */ + { 2, 1, 0, 0, 8 }, /* V */ + { 3, 1, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P9BE] = { + .name = "yuva420p9be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P9LE] = { + .name = "yuva420p9le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P9BE] = { + .name = "yuva422p9be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P9LE] = { + .name = "yuva422p9le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P9BE] = { + .name = "yuva444p9be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P9LE] = { + .name = "yuva444p9le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + { 3, 2, 0, 0, 9 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P10BE] = { + .name = "yuva420p10be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P10LE] = { + .name = "yuva420p10le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P10BE] = { + .name = "yuva422p10be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P10LE] = { + .name = "yuva422p10le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P10BE] = { + .name = "yuva444p10be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P10LE] = { + .name = "yuva444p10le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P16BE] = { + .name = "yuva420p16be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA420P16LE] = { + .name = "yuva420p16le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P16BE] = { + .name = "yuva422p16be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P16LE] = { + .name = "yuva422p16le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P16BE] = { + .name = "yuva444p16be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P16LE] = { + .name = "yuva444p16le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_RGB48BE] = { + .name = "rgb48be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 0, 0, 16 }, /* R */ + { 0, 6, 2, 0, 16 }, /* G */ + { 0, 6, 4, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_RGB48LE] = { + .name = "rgb48le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 0, 0, 16 }, /* R */ + { 0, 6, 2, 0, 16 }, /* G */ + { 0, 6, 4, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGBA64BE] = { + .name = "rgba64be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 0, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 4, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_RGBA64LE] = { + .name = "rgba64le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 0, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 4, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_RGB565BE] = { + .name = "rgb565be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, -1, 3, 5 }, /* R */ + { 0, 2, 0, 5, 6 }, /* G */ + { 0, 2, 0, 0, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB565LE] = { + .name = "rgb565le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 1, 3, 5 }, /* R */ + { 0, 2, 0, 5, 6 }, /* G */ + { 0, 2, 0, 0, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB555BE] = { + .name = "rgb555be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, -1, 2, 5 }, /* R */ + { 0, 2, 0, 5, 5 }, /* G */ + { 0, 2, 0, 0, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB555LE] = { + .name = "rgb555le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 1, 2, 5 }, /* R */ + { 0, 2, 0, 5, 5 }, /* G */ + { 0, 2, 0, 0, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB444BE] = { + .name = "rgb444be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, -1, 0, 4 }, /* R */ + { 0, 2, 0, 4, 4 }, /* G */ + { 0, 2, 0, 0, 4 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_RGB444LE] = { + .name = "rgb444le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 1, 0, 4 }, /* R */ + { 0, 2, 0, 4, 4 }, /* G */ + { 0, 2, 0, 0, 4 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR48BE] = { + .name = "bgr48be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 4, 0, 16 }, /* R */ + { 0, 6, 2, 0, 16 }, /* G */ + { 0, 6, 0, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR48LE] = { + .name = "bgr48le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 4, 0, 16 }, /* R */ + { 0, 6, 2, 0, 16 }, /* G */ + { 0, 6, 0, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGRA64BE] = { + .name = "bgra64be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 4, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 0, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_BGRA64LE] = { + .name = "bgra64le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 4, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 0, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_BGR565BE] = { + .name = "bgr565be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 5 }, /* R */ + { 0, 2, 0, 5, 6 }, /* G */ + { 0, 2, -1, 3, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR565LE] = { + .name = "bgr565le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 5 }, /* R */ + { 0, 2, 0, 5, 6 }, /* G */ + { 0, 2, 1, 3, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR555BE] = { + .name = "bgr555be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 5 }, /* R */ + { 0, 2, 0, 5, 5 }, /* G */ + { 0, 2, -1, 2, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR555LE] = { + .name = "bgr555le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 5 }, /* R */ + { 0, 2, 0, 5, 5 }, /* G */ + { 0, 2, 1, 2, 5 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR444BE] = { + .name = "bgr444be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 4 }, /* R */ + { 0, 2, 0, 4, 4 }, /* G */ + { 0, 2, -1, 0, 4 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_BGR444LE] = { + .name = "bgr444le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 4 }, /* R */ + { 0, 2, 0, 4, 4 }, /* G */ + { 0, 2, 1, 0, 4 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_VAAPI] = { + .name = "vaapi", + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_YUV420P9LE] = { + .name = "yuv420p9le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P9BE] = { + .name = "yuv420p9be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P10LE] = { + .name = "yuv420p10le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P10BE] = { + .name = "yuv420p10be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P12LE] = { + .name = "yuv420p12le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P12BE] = { + .name = "yuv420p12be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P14LE] = { + .name = "yuv420p14le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P14BE] = { + .name = "yuv420p14be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P16LE] = { + .name = "yuv420p16le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV420P16BE] = { + .name = "yuv420p16be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P9LE] = { + .name = "yuv422p9le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P9BE] = { + .name = "yuv422p9be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P10LE] = { + .name = "yuv422p10le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P10BE] = { + .name = "yuv422p10be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P12LE] = { + .name = "yuv422p12le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P12BE] = { + .name = "yuv422p12be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P14LE] = { + .name = "yuv422p14le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P14BE] = { + .name = "yuv422p14be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P16LE] = { + .name = "yuv422p16le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV422P16BE] = { + .name = "yuv422p16be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P16LE] = { + .name = "yuv444p16le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P16BE] = { + .name = "yuv444p16be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 2, 0, 0, 16 }, /* U */ + { 2, 2, 0, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P10LE] = { + .name = "yuv444p10le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P10BE] = { + .name = "yuv444p10be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 2, 0, 0, 10 }, /* U */ + { 2, 2, 0, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P9LE] = { + .name = "yuv444p9le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P9BE] = { + .name = "yuv444p9be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 9 }, /* Y */ + { 1, 2, 0, 0, 9 }, /* U */ + { 2, 2, 0, 0, 9 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P12LE] = { + .name = "yuv444p12le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P12BE] = { + .name = "yuv444p12be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P14LE] = { + .name = "yuv444p14le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_YUV444P14BE] = { + .name = "yuv444p14be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 14 }, /* Y */ + { 1, 2, 0, 0, 14 }, /* U */ + { 2, 2, 0, 0, 14 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_D3D11VA_VLD] = { + .name = "d3d11va_vld", + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_DXVA2_VLD] = { + .name = "dxva2_vld", + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_YA8] = { + .name = "ya8", + .nb_components = 2, + .comp = { + { 0, 2, 0, 0, 8 }, /* Y */ + { 0, 2, 1, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_ALPHA, + .alias = "gray8a", + }, + [AV_PIX_FMT_YA16LE] = { + .name = "ya16le", + .nb_components = 2, + .comp = { + { 0, 4, 0, 0, 16 }, /* Y */ + { 0, 4, 2, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YA16BE] = { + .name = "ya16be", + .nb_components = 2, + .comp = { + { 0, 4, 0, 0, 16 }, /* Y */ + { 0, 4, 2, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_VIDEOTOOLBOX] = { + .name = "videotoolbox_vld", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_GBRP] = { + .name = "gbrp", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 1, 0, 0, 8 }, /* R */ + { 0, 1, 0, 0, 8 }, /* G */ + { 1, 1, 0, 0, 8 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP9LE] = { + .name = "gbrp9le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 9 }, /* R */ + { 0, 2, 0, 0, 9 }, /* G */ + { 1, 2, 0, 0, 9 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP9BE] = { + .name = "gbrp9be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 9 }, /* R */ + { 0, 2, 0, 0, 9 }, /* G */ + { 1, 2, 0, 0, 9 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP10LE] = { + .name = "gbrp10le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 10 }, /* R */ + { 0, 2, 0, 0, 10 }, /* G */ + { 1, 2, 0, 0, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP10BE] = { + .name = "gbrp10be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 10 }, /* R */ + { 0, 2, 0, 0, 10 }, /* G */ + { 1, 2, 0, 0, 10 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP12LE] = { + .name = "gbrp12le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 12 }, /* R */ + { 0, 2, 0, 0, 12 }, /* G */ + { 1, 2, 0, 0, 12 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP12BE] = { + .name = "gbrp12be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 12 }, /* R */ + { 0, 2, 0, 0, 12 }, /* G */ + { 1, 2, 0, 0, 12 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP14LE] = { + .name = "gbrp14le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 14 }, /* R */ + { 0, 2, 0, 0, 14 }, /* G */ + { 1, 2, 0, 0, 14 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP14BE] = { + .name = "gbrp14be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 14 }, /* R */ + { 0, 2, 0, 0, 14 }, /* G */ + { 1, 2, 0, 0, 14 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP16LE] = { + .name = "gbrp16le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 16 }, /* R */ + { 0, 2, 0, 0, 16 }, /* G */ + { 1, 2, 0, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRP16BE] = { + .name = "gbrp16be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 16 }, /* R */ + { 0, 2, 0, 0, 16 }, /* G */ + { 1, 2, 0, 0, 16 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRAP] = { + .name = "gbrap", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 1, 0, 0, 8 }, /* R */ + { 0, 1, 0, 0, 8 }, /* G */ + { 1, 1, 0, 0, 8 }, /* B */ + { 3, 1, 0, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP16LE] = { + .name = "gbrap16le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 16 }, /* R */ + { 0, 2, 0, 0, 16 }, /* G */ + { 1, 2, 0, 0, 16 }, /* B */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP16BE] = { + .name = "gbrap16be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 16 }, /* R */ + { 0, 2, 0, 0, 16 }, /* G */ + { 1, 2, 0, 0, 16 }, /* B */ + { 3, 2, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_VDPAU] = { + .name = "vdpau", + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_XYZ12LE] = { + .name = "xyz12le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 0, 4, 12 }, /* X */ + { 0, 6, 2, 4, 12 }, /* Y */ + { 0, 6, 4, 4, 12 }, /* Z */ + }, + .flags = AV_PIX_FMT_FLAG_XYZ, + }, + [AV_PIX_FMT_XYZ12BE] = { + .name = "xyz12be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 6, 0, 4, 12 }, /* X */ + { 0, 6, 2, 4, 12 }, /* Y */ + { 0, 6, 4, 4, 12 }, /* Z */ + }, + .flags = AV_PIX_FMT_FLAG_XYZ | AV_PIX_FMT_FLAG_BE, + }, + +#define BAYER8_DESC_COMMON \ + .nb_components= 3, \ + .log2_chroma_w= 0, \ + .log2_chroma_h= 0, \ + .comp = { \ + { 0, 1, 0, 0, 2 }, \ + { 0, 1, 0, 0, 4 }, \ + { 0, 1, 0, 0, 2 }, \ + }, \ + +#define BAYER16_DESC_COMMON \ + .nb_components= 3, \ + .log2_chroma_w= 0, \ + .log2_chroma_h= 0, \ + .comp = { \ + { 0, 2, 0, 0, 4 }, \ + { 0, 2, 0, 0, 8 }, \ + { 0, 2, 0, 0, 4 }, \ + }, \ + + [AV_PIX_FMT_BAYER_BGGR8] = { + .name = "bayer_bggr8", + BAYER8_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_BGGR16LE] = { + .name = "bayer_bggr16le", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_BGGR16BE] = { + .name = "bayer_bggr16be", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_RGGB8] = { + .name = "bayer_rggb8", + BAYER8_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_RGGB16LE] = { + .name = "bayer_rggb16le", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_RGGB16BE] = { + .name = "bayer_rggb16be", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GBRG8] = { + .name = "bayer_gbrg8", + BAYER8_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GBRG16LE] = { + .name = "bayer_gbrg16le", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GBRG16BE] = { + .name = "bayer_gbrg16be", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GRBG8] = { + .name = "bayer_grbg8", + BAYER8_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GRBG16LE] = { + .name = "bayer_grbg16le", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_BAYER_GRBG16BE] = { + .name = "bayer_grbg16be", + BAYER16_DESC_COMMON + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_BAYER, + }, + [AV_PIX_FMT_NV16] = { + .name = "nv16", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 2, 0, 0, 8 }, /* U */ + { 1, 2, 1, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_NV20LE] = { + .name = "nv20le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 4, 0, 0, 10 }, /* U */ + { 1, 4, 2, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_NV20BE] = { + .name = "nv20be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 10 }, /* Y */ + { 1, 4, 0, 0, 10 }, /* U */ + { 1, 4, 2, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_QSV] = { + .name = "qsv", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_MEDIACODEC] = { + .name = "mediacodec", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_MMAL] = { + .name = "mmal", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_CUDA] = { + .name = "cuda", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_AYUV64LE] = { + .name = "ayuv64le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 2, 0, 16 }, /* Y */ + { 0, 8, 4, 0, 16 }, /* U */ + { 0, 8, 6, 0, 16 }, /* V */ + { 0, 8, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_AYUV64BE] = { + .name = "ayuv64be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 2, 0, 16 }, /* Y */ + { 0, 8, 4, 0, 16 }, /* U */ + { 0, 8, 6, 0, 16 }, /* V */ + { 0, 8, 0, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_P010LE] = { + .name = "p010le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P010BE] = { + .name = "p010be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P012LE] = { + .name = "p012le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P012BE] = { + .name = "p012be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P016LE] = { + .name = "p016le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P016BE] = { + .name = "p016be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 1, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_GBRAP14LE] = { + .name = "gbrap14le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 14 }, /* R */ + { 0, 2, 0, 0, 14 }, /* G */ + { 1, 2, 0, 0, 14 }, /* B */ + { 3, 2, 0, 0, 14 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP14BE] = { + .name = "gbrap14be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 14 }, /* R */ + { 0, 2, 0, 0, 14 }, /* G */ + { 1, 2, 0, 0, 14 }, /* B */ + { 3, 2, 0, 0, 14 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP12LE] = { + .name = "gbrap12le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 12 }, /* R */ + { 0, 2, 0, 0, 12 }, /* G */ + { 1, 2, 0, 0, 12 }, /* B */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP12BE] = { + .name = "gbrap12be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 12 }, /* R */ + { 0, 2, 0, 0, 12 }, /* G */ + { 1, 2, 0, 0, 12 }, /* B */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP10LE] = { + .name = "gbrap10le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 10 }, /* R */ + { 0, 2, 0, 0, 10 }, /* G */ + { 1, 2, 0, 0, 10 }, /* B */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_GBRAP10BE] = { + .name = "gbrap10be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 2, 0, 0, 10 }, /* R */ + { 0, 2, 0, 0, 10 }, /* G */ + { 1, 2, 0, 0, 10 }, /* B */ + { 3, 2, 0, 0, 10 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_D3D11] = { + .name = "d3d11", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_D3D12] = { + .name = "d3d12", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_GBRPF32BE] = { + .name = "gbrpf32be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 4, 0, 0, 32 }, /* R */ + { 0, 4, 0, 0, 32 }, /* G */ + { 1, 4, 0, 0, 32 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_GBRPF32LE] = { + .name = "gbrpf32le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 4, 0, 0, 32 }, /* R */ + { 0, 4, 0, 0, 32 }, /* G */ + { 1, 4, 0, 0, 32 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_FLOAT | AV_PIX_FMT_FLAG_RGB, + }, + [AV_PIX_FMT_GBRAPF32BE] = { + .name = "gbrapf32be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 4, 0, 0, 32 }, /* R */ + { 0, 4, 0, 0, 32 }, /* G */ + { 1, 4, 0, 0, 32 }, /* B */ + { 3, 4, 0, 0, 32 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | + AV_PIX_FMT_FLAG_ALPHA | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_GBRAPF32LE] = { + .name = "gbrapf32le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 2, 4, 0, 0, 32 }, /* R */ + { 0, 4, 0, 0, 32 }, /* G */ + { 1, 4, 0, 0, 32 }, /* B */ + { 3, 4, 0, 0, 32 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA | + AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_DRM_PRIME] = { + .name = "drm_prime", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_OPENCL] = { + .name = "opencl", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_GRAYF32BE] = { + .name = "grayf32be", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 0, 32 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_FLOAT, + .alias = "yf32be", + }, + [AV_PIX_FMT_GRAYF32LE] = { + .name = "grayf32le", + .nb_components = 1, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 0, 32 }, /* Y */ + }, + .flags = AV_PIX_FMT_FLAG_FLOAT, + .alias = "yf32le", + }, + [AV_PIX_FMT_YUVA422P12BE] = { + .name = "yuva422p12be", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA422P12LE] = { + .name = "yuva422p12le", + .nb_components = 4, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P12BE] = { + .name = "yuva444p12be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_YUVA444P12LE] = { + .name = "yuva444p12le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 12 }, /* Y */ + { 1, 2, 0, 0, 12 }, /* U */ + { 2, 2, 0, 0, 12 }, /* V */ + { 3, 2, 0, 0, 12 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_NV24] = { + .name = "nv24", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 2, 0, 0, 8 }, /* U */ + { 1, 2, 1, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_NV42] = { + .name = "nv42", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 1, 0, 0, 8 }, /* Y */ + { 1, 2, 1, 0, 8 }, /* U */ + { 1, 2, 0, 0, 8 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_VULKAN] = { + .name = "vulkan", + .flags = AV_PIX_FMT_FLAG_HWACCEL, + }, + [AV_PIX_FMT_P210BE] = { + .name = "p210be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P210LE] = { + .name = "p210le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P410BE] = { + .name = "p410be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P410LE] = { + .name = "p410le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 6, 10 }, /* Y */ + { 1, 4, 0, 6, 10 }, /* U */ + { 1, 4, 2, 6, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P216BE] = { + .name = "p216be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P216LE] = { + .name = "p216le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P416BE] = { + .name = "p416be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P416LE] = { + .name = "p416le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 0, 16 }, /* Y */ + { 1, 4, 0, 0, 16 }, /* U */ + { 1, 4, 2, 0, 16 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_VUYA] = { + .name = "vuya", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 2, 0, 8 }, /* Y */ + { 0, 4, 1, 0, 8 }, /* U */ + { 0, 4, 0, 0, 8 }, /* V */ + { 0, 4, 3, 0, 8 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_VUYX] = { + .name = "vuyx", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 2, 0, 8 }, /* Y */ + { 0, 4, 1, 0, 8 }, /* U */ + { 0, 4, 0, 0, 8 }, /* V */ + }, + }, + [AV_PIX_FMT_RGBAF16BE] = { + .name = "rgbaf16be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 0, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 4, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_ALPHA | AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_RGBAF16LE] = { + .name = "rgbaf16le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 8, 0, 0, 16 }, /* R */ + { 0, 8, 2, 0, 16 }, /* G */ + { 0, 8, 4, 0, 16 }, /* B */ + { 0, 8, 6, 0, 16 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_ALPHA | + AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_Y212LE] = { + .name = "y212le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 4, 12 }, /* Y */ + { 0, 8, 2, 4, 12 }, /* U */ + { 0, 8, 6, 4, 12 }, /* V */ + }, + }, + [AV_PIX_FMT_Y212BE] = { + .name = "y212be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 4, 0, 4, 12 }, /* Y */ + { 0, 8, 2, 4, 12 }, /* U */ + { 0, 8, 6, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_XV30LE] = { + .name = "xv30le", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 4, 1, 2, 10 }, /* Y */ + { 0, 4, 0, 0, 10 }, /* U */ + { 0, 4, 2, 4, 10 }, /* V */ + }, + }, + [AV_PIX_FMT_XV30BE] = { + .name = "xv30be", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 32, 10, 0, 10 }, /* Y */ + { 0, 32, 0, 0, 10 }, /* U */ + { 0, 32, 20, 0, 10 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_BITSTREAM, + }, + [AV_PIX_FMT_XV36LE] = { + .name = "xv36le", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 8, 2, 4, 12 }, /* Y */ + { 0, 8, 0, 4, 12 }, /* U */ + { 0, 8, 4, 4, 12 }, /* V */ + }, + }, + [AV_PIX_FMT_XV36BE] = { + .name = "xv36be", + .nb_components= 3, + .log2_chroma_w= 0, + .log2_chroma_h= 0, + .comp = { + { 0, 8, 2, 4, 12 }, /* Y */ + { 0, 8, 0, 4, 12 }, /* U */ + { 0, 8, 4, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_RGBF32BE] = { + .name = "rgbf32be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 12, 0, 0, 32 }, /* R */ + { 0, 12, 4, 0, 32 }, /* G */ + { 0, 12, 8, 0, 32 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_RGBF32LE] = { + .name = "rgbf32le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 12, 0, 0, 32 }, /* R */ + { 0, 12, 4, 0, 32 }, /* G */ + { 0, 12, 8, 0, 32 }, /* B */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_FLOAT, + }, + [AV_PIX_FMT_RGBAF32BE] = { + .name = "rgbaf32be", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 16, 0, 0, 32 }, /* R */ + { 0, 16, 4, 0, 32 }, /* G */ + { 0, 16, 8, 0, 32 }, /* B */ + { 0, 16, 12, 0, 32 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_RGB | + AV_PIX_FMT_FLAG_FLOAT | AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_RGBAF32LE] = { + .name = "rgbaf32le", + .nb_components = 4, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 16, 0, 0, 32 }, /* R */ + { 0, 16, 4, 0, 32 }, /* G */ + { 0, 16, 8, 0, 32 }, /* B */ + { 0, 16, 12, 0, 32 }, /* A */ + }, + .flags = AV_PIX_FMT_FLAG_RGB | AV_PIX_FMT_FLAG_FLOAT | + AV_PIX_FMT_FLAG_ALPHA, + }, + [AV_PIX_FMT_P212BE] = { + .name = "p212be", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P212LE] = { + .name = "p212le", + .nb_components = 3, + .log2_chroma_w = 1, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, + [AV_PIX_FMT_P412BE] = { + .name = "p412be", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR | AV_PIX_FMT_FLAG_BE, + }, + [AV_PIX_FMT_P412LE] = { + .name = "p412le", + .nb_components = 3, + .log2_chroma_w = 0, + .log2_chroma_h = 0, + .comp = { + { 0, 2, 0, 4, 12 }, /* Y */ + { 1, 4, 0, 4, 12 }, /* U */ + { 1, 4, 2, 4, 12 }, /* V */ + }, + .flags = AV_PIX_FMT_FLAG_PLANAR, + }, +}; + +static const char * const color_range_names[] = { + [AVCOL_RANGE_UNSPECIFIED] = "unknown", + [AVCOL_RANGE_MPEG] = "tv", + [AVCOL_RANGE_JPEG] = "pc", +}; + +static const char * const color_primaries_names[AVCOL_PRI_NB] = { + [AVCOL_PRI_RESERVED0] = "reserved", + [AVCOL_PRI_BT709] = "bt709", + [AVCOL_PRI_UNSPECIFIED] = "unknown", + [AVCOL_PRI_RESERVED] = "reserved", + [AVCOL_PRI_BT470M] = "bt470m", + [AVCOL_PRI_BT470BG] = "bt470bg", + [AVCOL_PRI_SMPTE170M] = "smpte170m", + [AVCOL_PRI_SMPTE240M] = "smpte240m", + [AVCOL_PRI_FILM] = "film", + [AVCOL_PRI_BT2020] = "bt2020", + [AVCOL_PRI_SMPTE428] = "smpte428", + [AVCOL_PRI_SMPTE431] = "smpte431", + [AVCOL_PRI_SMPTE432] = "smpte432", + [AVCOL_PRI_EBU3213] = "ebu3213", +}; + +static const char * const color_transfer_names[] = { + [AVCOL_TRC_RESERVED0] = "reserved", + [AVCOL_TRC_BT709] = "bt709", + [AVCOL_TRC_UNSPECIFIED] = "unknown", + [AVCOL_TRC_RESERVED] = "reserved", + [AVCOL_TRC_GAMMA22] = "bt470m", + [AVCOL_TRC_GAMMA28] = "bt470bg", + [AVCOL_TRC_SMPTE170M] = "smpte170m", + [AVCOL_TRC_SMPTE240M] = "smpte240m", + [AVCOL_TRC_LINEAR] = "linear", + [AVCOL_TRC_LOG] = "log100", + [AVCOL_TRC_LOG_SQRT] = "log316", + [AVCOL_TRC_IEC61966_2_4] = "iec61966-2-4", + [AVCOL_TRC_BT1361_ECG] = "bt1361e", + [AVCOL_TRC_IEC61966_2_1] = "iec61966-2-1", + [AVCOL_TRC_BT2020_10] = "bt2020-10", + [AVCOL_TRC_BT2020_12] = "bt2020-12", + [AVCOL_TRC_SMPTE2084] = "smpte2084", + [AVCOL_TRC_SMPTE428] = "smpte428", + [AVCOL_TRC_ARIB_STD_B67] = "arib-std-b67", +}; + +static const char * const color_space_names[] = { + [AVCOL_SPC_RGB] = "gbr", + [AVCOL_SPC_BT709] = "bt709", + [AVCOL_SPC_UNSPECIFIED] = "unknown", + [AVCOL_SPC_RESERVED] = "reserved", + [AVCOL_SPC_FCC] = "fcc", + [AVCOL_SPC_BT470BG] = "bt470bg", + [AVCOL_SPC_SMPTE170M] = "smpte170m", + [AVCOL_SPC_SMPTE240M] = "smpte240m", + [AVCOL_SPC_YCGCO] = "ycgco", + [AVCOL_SPC_BT2020_NCL] = "bt2020nc", + [AVCOL_SPC_BT2020_CL] = "bt2020c", + [AVCOL_SPC_SMPTE2085] = "smpte2085", + [AVCOL_SPC_CHROMA_DERIVED_NCL] = "chroma-derived-nc", + [AVCOL_SPC_CHROMA_DERIVED_CL] = "chroma-derived-c", + [AVCOL_SPC_ICTCP] = "ictcp", +}; + +static const char * const chroma_location_names[] = { + [AVCHROMA_LOC_UNSPECIFIED] = "unspecified", + [AVCHROMA_LOC_LEFT] = "left", + [AVCHROMA_LOC_CENTER] = "center", + [AVCHROMA_LOC_TOPLEFT] = "topleft", + [AVCHROMA_LOC_TOP] = "top", + [AVCHROMA_LOC_BOTTOMLEFT] = "bottomleft", + [AVCHROMA_LOC_BOTTOM] = "bottom", +}; + +static enum AVPixelFormat get_pix_fmt_internal(const char *name) +{ + enum AVPixelFormat pix_fmt; + + for (pix_fmt = 0; pix_fmt < AV_PIX_FMT_NB; pix_fmt++) + if (av_pix_fmt_descriptors[pix_fmt].name && + (!strcmp(av_pix_fmt_descriptors[pix_fmt].name, name) || + av_match_name(name, av_pix_fmt_descriptors[pix_fmt].alias))) + return pix_fmt; + + return AV_PIX_FMT_NONE; +} + +const char *av_get_pix_fmt_name(enum AVPixelFormat pix_fmt) +{ + return (unsigned)pix_fmt < AV_PIX_FMT_NB ? + av_pix_fmt_descriptors[pix_fmt].name : NULL; +} + +#if HAVE_BIGENDIAN +# define X_NE(be, le) be +#else +# define X_NE(be, le) le +#endif + +enum AVPixelFormat av_get_pix_fmt(const char *name) +{ + enum AVPixelFormat pix_fmt; + + if (!strcmp(name, "rgb32")) + name = X_NE("argb", "bgra"); + else if (!strcmp(name, "bgr32")) + name = X_NE("abgr", "rgba"); + + pix_fmt = get_pix_fmt_internal(name); + if (pix_fmt == AV_PIX_FMT_NONE) { + char name2[32]; + + snprintf(name2, sizeof(name2), "%s%s", name, X_NE("be", "le")); + pix_fmt = get_pix_fmt_internal(name2); + } + + return pix_fmt; +} + +int av_get_bits_per_pixel(const AVPixFmtDescriptor *pixdesc) +{ + int c, bits = 0; + int log2_pixels = pixdesc->log2_chroma_w + pixdesc->log2_chroma_h; + + for (c = 0; c < pixdesc->nb_components; c++) { + int s = c == 1 || c == 2 ? 0 : log2_pixels; + bits += pixdesc->comp[c].depth << s; + } + + return bits >> log2_pixels; +} + +int av_get_padded_bits_per_pixel(const AVPixFmtDescriptor *pixdesc) +{ + int c, bits = 0; + int log2_pixels = pixdesc->log2_chroma_w + pixdesc->log2_chroma_h; + int steps[4] = {0}; + + for (c = 0; c < pixdesc->nb_components; c++) { + const AVComponentDescriptor *comp = &pixdesc->comp[c]; + int s = c == 1 || c == 2 ? 0 : log2_pixels; + steps[comp->plane] = comp->step << s; + } + for (c = 0; c < 4; c++) + bits += steps[c]; + + if(!(pixdesc->flags & AV_PIX_FMT_FLAG_BITSTREAM)) + bits *= 8; + + return bits >> log2_pixels; +} + +char *av_get_pix_fmt_string(char *buf, int buf_size, + enum AVPixelFormat pix_fmt) +{ + /* print header */ + if (pix_fmt < 0) { + snprintf (buf, buf_size, "name" " nb_components" " nb_bits"); + } else { + const AVPixFmtDescriptor *pixdesc = &av_pix_fmt_descriptors[pix_fmt]; + snprintf(buf, buf_size, "%-11s %7d %10d", pixdesc->name, + pixdesc->nb_components, av_get_bits_per_pixel(pixdesc)); + } + + return buf; +} + +const AVPixFmtDescriptor *av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt) +{ + if (pix_fmt < 0 || pix_fmt >= AV_PIX_FMT_NB) + return NULL; + return &av_pix_fmt_descriptors[pix_fmt]; +} + +const AVPixFmtDescriptor *av_pix_fmt_desc_next(const AVPixFmtDescriptor *prev) +{ + if (!prev) + return &av_pix_fmt_descriptors[0]; + while (prev - av_pix_fmt_descriptors < FF_ARRAY_ELEMS(av_pix_fmt_descriptors) - 1) { + prev++; + if (prev->name) + return prev; + } + return NULL; +} + +enum AVPixelFormat av_pix_fmt_desc_get_id(const AVPixFmtDescriptor *desc) +{ + if (desc < av_pix_fmt_descriptors || + desc >= av_pix_fmt_descriptors + FF_ARRAY_ELEMS(av_pix_fmt_descriptors)) + return AV_PIX_FMT_NONE; + + return desc - av_pix_fmt_descriptors; +} + +int av_pix_fmt_get_chroma_sub_sample(enum AVPixelFormat pix_fmt, + int *h_shift, int *v_shift) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + if (!desc) + return AVERROR(ENOSYS); + *h_shift = desc->log2_chroma_w; + *v_shift = desc->log2_chroma_h; + + return 0; +} + +int av_pix_fmt_count_planes(enum AVPixelFormat pix_fmt) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int i, planes[4] = { 0 }, ret = 0; + + if (!desc) + return AVERROR(EINVAL); + + for (i = 0; i < desc->nb_components; i++) + planes[desc->comp[i].plane] = 1; + for (i = 0; i < FF_ARRAY_ELEMS(planes); i++) + ret += planes[i]; + return ret; +} + +enum AVPixelFormat av_pix_fmt_swap_endianness(enum AVPixelFormat pix_fmt) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + char name[16]; + int i; + + if (!desc || strlen(desc->name) < 2) + return AV_PIX_FMT_NONE; + av_strlcpy(name, desc->name, sizeof(name)); + i = strlen(name) - 2; + if (strcmp(name + i, "be") && strcmp(name + i, "le")) + return AV_PIX_FMT_NONE; + + name[i] ^= 'b' ^ 'l'; + + return get_pix_fmt_internal(name); +} + +#define FF_COLOR_NA -1 +#define FF_COLOR_RGB 0 /**< RGB color space */ +#define FF_COLOR_GRAY 1 /**< gray color space */ +#define FF_COLOR_YUV 2 /**< YUV color space. 16 <= Y <= 235, 16 <= U, V <= 240 */ +#define FF_COLOR_YUV_JPEG 3 /**< YUV color space. 0 <= Y <= 255, 0 <= U, V <= 255 */ +#define FF_COLOR_XYZ 4 + +#define pixdesc_has_alpha(pixdesc) \ + ((pixdesc)->flags & AV_PIX_FMT_FLAG_ALPHA) + + +static int get_color_type(const AVPixFmtDescriptor *desc) { + if (desc->flags & AV_PIX_FMT_FLAG_PAL) + return FF_COLOR_RGB; + + if(desc->nb_components == 1 || desc->nb_components == 2) + return FF_COLOR_GRAY; + + if (desc->name) { + if (av_strstart(desc->name, "yuvj", NULL)) + return FF_COLOR_YUV_JPEG; + } + + if(desc->flags & AV_PIX_FMT_FLAG_RGB) + return FF_COLOR_RGB; + + if(desc->flags & AV_PIX_FMT_FLAG_XYZ) + return FF_COLOR_XYZ; + + if(desc->nb_components == 0) + return FF_COLOR_NA; + + return FF_COLOR_YUV; +} + +static int get_pix_fmt_depth(int *min, int *max, enum AVPixelFormat pix_fmt) +{ + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt); + int i; + + if (!desc || !desc->nb_components) { + *min = *max = 0; + return AVERROR(EINVAL); + } + + *min = INT_MAX, *max = -INT_MAX; + for (i = 0; i < desc->nb_components; i++) { + *min = FFMIN(desc->comp[i].depth, *min); + *max = FFMAX(desc->comp[i].depth, *max); + } + return 0; +} + +static int get_pix_fmt_score(enum AVPixelFormat dst_pix_fmt, + enum AVPixelFormat src_pix_fmt, + unsigned *lossp, unsigned consider) +{ + const AVPixFmtDescriptor *src_desc = av_pix_fmt_desc_get(src_pix_fmt); + const AVPixFmtDescriptor *dst_desc = av_pix_fmt_desc_get(dst_pix_fmt); + int src_color, dst_color; + int src_min_depth, src_max_depth, dst_min_depth, dst_max_depth; + int ret, loss, i, nb_components; + int score = INT_MAX - 1; + + if (!src_desc || !dst_desc) + return -4; + + if ((src_desc->flags & AV_PIX_FMT_FLAG_HWACCEL) || + (dst_desc->flags & AV_PIX_FMT_FLAG_HWACCEL)) { + if (dst_pix_fmt == src_pix_fmt) + return -1; + else + return -2; + } + + /* compute loss */ + *lossp = loss = 0; + + if (dst_pix_fmt == src_pix_fmt) + return INT_MAX; + + if ((ret = get_pix_fmt_depth(&src_min_depth, &src_max_depth, src_pix_fmt)) < 0) + return -3; + if ((ret = get_pix_fmt_depth(&dst_min_depth, &dst_max_depth, dst_pix_fmt)) < 0) + return -3; + + src_color = get_color_type(src_desc); + dst_color = get_color_type(dst_desc); + if (dst_pix_fmt == AV_PIX_FMT_PAL8) + nb_components = FFMIN(src_desc->nb_components, 4); + else + nb_components = FFMIN(src_desc->nb_components, dst_desc->nb_components); + + for (i = 0; i < nb_components; i++) { + int depth_minus1 = (dst_pix_fmt == AV_PIX_FMT_PAL8) ? 7/nb_components : (dst_desc->comp[i].depth - 1); + int depth_delta = src_desc->comp[i].depth - 1 - depth_minus1; + if (depth_delta > 0 && (consider & FF_LOSS_DEPTH)) { + loss |= FF_LOSS_DEPTH; + score -= 65536 >> depth_minus1; + } else if (depth_delta < 0 && (consider & FF_LOSS_EXCESS_DEPTH)) { + // Favour formats where bit depth exactly matches. If all other + // scoring is equal, we'd rather use the bit depth that most closely + // matches the source. + loss |= FF_LOSS_EXCESS_DEPTH; + score += depth_delta; + } + } + + if (consider & FF_LOSS_RESOLUTION) { + if (dst_desc->log2_chroma_w > src_desc->log2_chroma_w) { + loss |= FF_LOSS_RESOLUTION; + score -= 256 << dst_desc->log2_chroma_w; + } + if (dst_desc->log2_chroma_h > src_desc->log2_chroma_h) { + loss |= FF_LOSS_RESOLUTION; + score -= 256 << dst_desc->log2_chroma_h; + } + // don't favor 422 over 420 if downsampling is needed, because 420 has much better support on the decoder side + if (dst_desc->log2_chroma_w == 1 && src_desc->log2_chroma_w == 0 && + dst_desc->log2_chroma_h == 1 && src_desc->log2_chroma_h == 0 ) { + score += 512; + } + } + + if (consider & FF_LOSS_EXCESS_RESOLUTION) { + // Favour formats where chroma subsampling exactly matches. If all other + // scoring is equal, we'd rather use the subsampling that most closely + // matches the source. + if (dst_desc->log2_chroma_w < src_desc->log2_chroma_w) { + loss |= FF_LOSS_EXCESS_RESOLUTION; + score -= 1 << (src_desc->log2_chroma_w - dst_desc->log2_chroma_w); + } + + if (dst_desc->log2_chroma_h < src_desc->log2_chroma_h) { + loss |= FF_LOSS_EXCESS_RESOLUTION; + score -= 1 << (src_desc->log2_chroma_h - dst_desc->log2_chroma_h); + } + + // don't favour 411 over 420, because 420 has much better support on the + // decoder side. + if (dst_desc->log2_chroma_w == 1 && src_desc->log2_chroma_w == 2 && + dst_desc->log2_chroma_h == 1 && src_desc->log2_chroma_h == 2) { + score += 4; + } + } + + if(consider & FF_LOSS_COLORSPACE) + switch(dst_color) { + case FF_COLOR_RGB: + if (src_color != FF_COLOR_RGB && + src_color != FF_COLOR_GRAY) + loss |= FF_LOSS_COLORSPACE; + break; + case FF_COLOR_GRAY: + if (src_color != FF_COLOR_GRAY) + loss |= FF_LOSS_COLORSPACE; + break; + case FF_COLOR_YUV: + if (src_color != FF_COLOR_YUV) + loss |= FF_LOSS_COLORSPACE; + break; + case FF_COLOR_YUV_JPEG: + if (src_color != FF_COLOR_YUV_JPEG && + src_color != FF_COLOR_YUV && + src_color != FF_COLOR_GRAY) + loss |= FF_LOSS_COLORSPACE; + break; + default: + /* fail safe test */ + if (src_color != dst_color) + loss |= FF_LOSS_COLORSPACE; + break; + } + if(loss & FF_LOSS_COLORSPACE) + score -= (nb_components * 65536) >> FFMIN(dst_desc->comp[0].depth - 1, src_desc->comp[0].depth - 1); + + if (dst_color == FF_COLOR_GRAY && + src_color != FF_COLOR_GRAY && (consider & FF_LOSS_CHROMA)) { + loss |= FF_LOSS_CHROMA; + score -= 2 * 65536; + } + if (!pixdesc_has_alpha(dst_desc) && (pixdesc_has_alpha(src_desc) && (consider & FF_LOSS_ALPHA))) { + loss |= FF_LOSS_ALPHA; + score -= 65536; + } + if (dst_pix_fmt == AV_PIX_FMT_PAL8 && (consider & FF_LOSS_COLORQUANT) && + (src_pix_fmt != AV_PIX_FMT_PAL8 && (src_color != FF_COLOR_GRAY || (pixdesc_has_alpha(src_desc) && (consider & FF_LOSS_ALPHA))))) { + loss |= FF_LOSS_COLORQUANT; + score -= 65536; + } + + *lossp = loss; + return score; +} + +int av_get_pix_fmt_loss(enum AVPixelFormat dst_pix_fmt, + enum AVPixelFormat src_pix_fmt, + int has_alpha) +{ + int loss; + int ret = get_pix_fmt_score(dst_pix_fmt, src_pix_fmt, &loss, has_alpha ? ~0 : ~FF_LOSS_ALPHA); + if (ret < 0) + return ret; + return loss; +} + +enum AVPixelFormat av_find_best_pix_fmt_of_2(enum AVPixelFormat dst_pix_fmt1, enum AVPixelFormat dst_pix_fmt2, + enum AVPixelFormat src_pix_fmt, int has_alpha, int *loss_ptr) +{ + enum AVPixelFormat dst_pix_fmt; + int loss1, loss2, loss_mask; + const AVPixFmtDescriptor *desc1 = av_pix_fmt_desc_get(dst_pix_fmt1); + const AVPixFmtDescriptor *desc2 = av_pix_fmt_desc_get(dst_pix_fmt2); + int score1, score2; + + if (!desc1) { + dst_pix_fmt = dst_pix_fmt2; + } else if (!desc2) { + dst_pix_fmt = dst_pix_fmt1; + } else { + loss_mask= loss_ptr?~*loss_ptr:~0; /* use loss mask if provided */ + if(!has_alpha) + loss_mask &= ~FF_LOSS_ALPHA; + + score1 = get_pix_fmt_score(dst_pix_fmt1, src_pix_fmt, &loss1, loss_mask); + score2 = get_pix_fmt_score(dst_pix_fmt2, src_pix_fmt, &loss2, loss_mask); + + if (score1 == score2) { + if(av_get_padded_bits_per_pixel(desc2) != av_get_padded_bits_per_pixel(desc1)) { + dst_pix_fmt = av_get_padded_bits_per_pixel(desc2) < av_get_padded_bits_per_pixel(desc1) ? dst_pix_fmt2 : dst_pix_fmt1; + } else { + dst_pix_fmt = desc2->nb_components < desc1->nb_components ? dst_pix_fmt2 : dst_pix_fmt1; + } + } else { + dst_pix_fmt = score1 < score2 ? dst_pix_fmt2 : dst_pix_fmt1; + } + } + + if (loss_ptr) + *loss_ptr = av_get_pix_fmt_loss(dst_pix_fmt, src_pix_fmt, has_alpha); + return dst_pix_fmt; +} + +const char *av_color_range_name(enum AVColorRange range) +{ + return (unsigned) range < AVCOL_RANGE_NB ? + color_range_names[range] : NULL; +} + +int av_color_range_from_name(const char *name) +{ + int i; + + for (i = 0; i < FF_ARRAY_ELEMS(color_range_names); i++) { + if (av_strstart(name, color_range_names[i], NULL)) + return i; + } + + return AVERROR(EINVAL); +} + +const char *av_color_primaries_name(enum AVColorPrimaries primaries) +{ + return (unsigned) primaries < AVCOL_PRI_NB ? + color_primaries_names[primaries] : NULL; +} + +int av_color_primaries_from_name(const char *name) +{ + int i; + + for (i = 0; i < FF_ARRAY_ELEMS(color_primaries_names); i++) { + if (!color_primaries_names[i]) + continue; + + if (av_strstart(name, color_primaries_names[i], NULL)) + return i; + } + + return AVERROR(EINVAL); +} + +const char *av_color_transfer_name(enum AVColorTransferCharacteristic transfer) +{ + return (unsigned) transfer < AVCOL_TRC_NB ? + color_transfer_names[transfer] : NULL; +} + +int av_color_transfer_from_name(const char *name) +{ + int i; + + for (i = 0; i < FF_ARRAY_ELEMS(color_transfer_names); i++) { + if (!color_transfer_names[i]) + continue; + + if (av_strstart(name, color_transfer_names[i], NULL)) + return i; + } + + return AVERROR(EINVAL); +} + +const char *av_color_space_name(enum AVColorSpace space) +{ + return (unsigned) space < AVCOL_SPC_NB ? + color_space_names[space] : NULL; +} + +int av_color_space_from_name(const char *name) +{ + int i; + + for (i = 0; i < FF_ARRAY_ELEMS(color_space_names); i++) { + if (!color_space_names[i]) + continue; + + if (av_strstart(name, color_space_names[i], NULL)) + return i; + } + + return AVERROR(EINVAL); +} + +const char *av_chroma_location_name(enum AVChromaLocation location) +{ + return (unsigned) location < AVCHROMA_LOC_NB ? + chroma_location_names[location] : NULL; +} + +int av_chroma_location_from_name(const char *name) +{ + int i; + + for (i = 0; i < FF_ARRAY_ELEMS(chroma_location_names); i++) { + if (!chroma_location_names[i]) + continue; + + if (av_strstart(name, chroma_location_names[i], NULL)) + return i; + } + + return AVERROR(EINVAL); +} + +int av_chroma_location_enum_to_pos(int *xpos, int *ypos, enum AVChromaLocation pos) +{ + if (pos <= AVCHROMA_LOC_UNSPECIFIED || pos >= AVCHROMA_LOC_NB) + return AVERROR(EINVAL); + pos--; + + *xpos = (pos&1) * 128; + *ypos = ((pos>>1)^(pos<4)) * 128; + + return 0; +} + +enum AVChromaLocation av_chroma_location_pos_to_enum(int xpos, int ypos) +{ + int pos, xout, yout; + + for (pos = AVCHROMA_LOC_UNSPECIFIED + 1; pos < AVCHROMA_LOC_NB; pos++) { + if (av_chroma_location_enum_to_pos(&xout, &yout, pos) == 0 && xout == xpos && yout == ypos) + return pos; + } + return AVCHROMA_LOC_UNSPECIFIED; +} diff --git a/media/ffvpx/libavutil/pixdesc.h b/media/ffvpx/libavutil/pixdesc.h new file mode 100644 index 0000000000..ba2f632814 --- /dev/null +++ b/media/ffvpx/libavutil/pixdesc.h @@ -0,0 +1,440 @@ +/* + * pixel format descriptor + * Copyright (c) 2009 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_PIXDESC_H +#define AVUTIL_PIXDESC_H + +#include <inttypes.h> + +#include "attributes.h" +#include "pixfmt.h" + +typedef struct AVComponentDescriptor { + /** + * Which of the 4 planes contains the component. + */ + int plane; + + /** + * Number of elements between 2 horizontally consecutive pixels. + * Elements are bits for bitstream formats, bytes otherwise. + */ + int step; + + /** + * Number of elements before the component of the first pixel. + * Elements are bits for bitstream formats, bytes otherwise. + */ + int offset; + + /** + * Number of least significant bits that must be shifted away + * to get the value. + */ + int shift; + + /** + * Number of bits in the component. + */ + int depth; +} AVComponentDescriptor; + +/** + * Descriptor that unambiguously describes how the bits of a pixel are + * stored in the up to 4 data planes of an image. It also stores the + * subsampling factors and number of components. + * + * @note This is separate of the colorspace (RGB, YCbCr, YPbPr, JPEG-style YUV + * and all the YUV variants) AVPixFmtDescriptor just stores how values + * are stored not what these values represent. + */ +typedef struct AVPixFmtDescriptor { + const char *name; + uint8_t nb_components; ///< The number of components each pixel has, (1-4) + + /** + * Amount to shift the luma width right to find the chroma width. + * For YV12 this is 1 for example. + * chroma_width = AV_CEIL_RSHIFT(luma_width, log2_chroma_w) + * The note above is needed to ensure rounding up. + * This value only refers to the chroma components. + */ + uint8_t log2_chroma_w; + + /** + * Amount to shift the luma height right to find the chroma height. + * For YV12 this is 1 for example. + * chroma_height= AV_CEIL_RSHIFT(luma_height, log2_chroma_h) + * The note above is needed to ensure rounding up. + * This value only refers to the chroma components. + */ + uint8_t log2_chroma_h; + + /** + * Combination of AV_PIX_FMT_FLAG_... flags. + */ + uint64_t flags; + + /** + * Parameters that describe how pixels are packed. + * If the format has 1 or 2 components, then luma is 0. + * If the format has 3 or 4 components: + * if the RGB flag is set then 0 is red, 1 is green and 2 is blue; + * otherwise 0 is luma, 1 is chroma-U and 2 is chroma-V. + * + * If present, the Alpha channel is always the last component. + */ + AVComponentDescriptor comp[4]; + + /** + * Alternative comma-separated names. + */ + const char *alias; +} AVPixFmtDescriptor; + +/** + * Pixel format is big-endian. + */ +#define AV_PIX_FMT_FLAG_BE (1 << 0) +/** + * Pixel format has a palette in data[1], values are indexes in this palette. + */ +#define AV_PIX_FMT_FLAG_PAL (1 << 1) +/** + * All values of a component are bit-wise packed end to end. + */ +#define AV_PIX_FMT_FLAG_BITSTREAM (1 << 2) +/** + * Pixel format is an HW accelerated format. + */ +#define AV_PIX_FMT_FLAG_HWACCEL (1 << 3) +/** + * At least one pixel component is not in the first data plane. + */ +#define AV_PIX_FMT_FLAG_PLANAR (1 << 4) +/** + * The pixel format contains RGB-like data (as opposed to YUV/grayscale). + */ +#define AV_PIX_FMT_FLAG_RGB (1 << 5) + +/** + * The pixel format has an alpha channel. This is set on all formats that + * support alpha in some way, including AV_PIX_FMT_PAL8. The alpha is always + * straight, never pre-multiplied. + * + * If a codec or a filter does not support alpha, it should set all alpha to + * opaque, or use the equivalent pixel formats without alpha component, e.g. + * AV_PIX_FMT_RGB0 (or AV_PIX_FMT_RGB24 etc.) instead of AV_PIX_FMT_RGBA. + */ +#define AV_PIX_FMT_FLAG_ALPHA (1 << 7) + +/** + * The pixel format is following a Bayer pattern + */ +#define AV_PIX_FMT_FLAG_BAYER (1 << 8) + +/** + * The pixel format contains IEEE-754 floating point values. Precision (double, + * single, or half) should be determined by the pixel size (64, 32, or 16 bits). + */ +#define AV_PIX_FMT_FLAG_FLOAT (1 << 9) + +/** + * The pixel format contains XYZ-like data (as opposed to YUV/RGB/grayscale). + */ +#define AV_PIX_FMT_FLAG_XYZ (1 << 10) + +/** + * Return the number of bits per pixel used by the pixel format + * described by pixdesc. Note that this is not the same as the number + * of bits per sample. + * + * The returned number of bits refers to the number of bits actually + * used for storing the pixel information, that is padding bits are + * not counted. + */ +int av_get_bits_per_pixel(const AVPixFmtDescriptor *pixdesc); + +/** + * Return the number of bits per pixel for the pixel format + * described by pixdesc, including any padding or unused bits. + */ +int av_get_padded_bits_per_pixel(const AVPixFmtDescriptor *pixdesc); + +/** + * @return a pixel format descriptor for provided pixel format or NULL if + * this pixel format is unknown. + */ +const AVPixFmtDescriptor *av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt); + +/** + * Iterate over all pixel format descriptors known to libavutil. + * + * @param prev previous descriptor. NULL to get the first descriptor. + * + * @return next descriptor or NULL after the last descriptor + */ +const AVPixFmtDescriptor *av_pix_fmt_desc_next(const AVPixFmtDescriptor *prev); + +/** + * @return an AVPixelFormat id described by desc, or AV_PIX_FMT_NONE if desc + * is not a valid pointer to a pixel format descriptor. + */ +enum AVPixelFormat av_pix_fmt_desc_get_id(const AVPixFmtDescriptor *desc); + +/** + * Utility function to access log2_chroma_w log2_chroma_h from + * the pixel format AVPixFmtDescriptor. + * + * @param[in] pix_fmt the pixel format + * @param[out] h_shift store log2_chroma_w (horizontal/width shift) + * @param[out] v_shift store log2_chroma_h (vertical/height shift) + * + * @return 0 on success, AVERROR(ENOSYS) on invalid or unknown pixel format + */ +int av_pix_fmt_get_chroma_sub_sample(enum AVPixelFormat pix_fmt, + int *h_shift, int *v_shift); + +/** + * @return number of planes in pix_fmt, a negative AVERROR if pix_fmt is not a + * valid pixel format. + */ +int av_pix_fmt_count_planes(enum AVPixelFormat pix_fmt); + +/** + * @return the name for provided color range or NULL if unknown. + */ +const char *av_color_range_name(enum AVColorRange range); + +/** + * @return the AVColorRange value for name or an AVError if not found. + */ +int av_color_range_from_name(const char *name); + +/** + * @return the name for provided color primaries or NULL if unknown. + */ +const char *av_color_primaries_name(enum AVColorPrimaries primaries); + +/** + * @return the AVColorPrimaries value for name or an AVError if not found. + */ +int av_color_primaries_from_name(const char *name); + +/** + * @return the name for provided color transfer or NULL if unknown. + */ +const char *av_color_transfer_name(enum AVColorTransferCharacteristic transfer); + +/** + * @return the AVColorTransferCharacteristic value for name or an AVError if not found. + */ +int av_color_transfer_from_name(const char *name); + +/** + * @return the name for provided color space or NULL if unknown. + */ +const char *av_color_space_name(enum AVColorSpace space); + +/** + * @return the AVColorSpace value for name or an AVError if not found. + */ +int av_color_space_from_name(const char *name); + +/** + * @return the name for provided chroma location or NULL if unknown. + */ +const char *av_chroma_location_name(enum AVChromaLocation location); + +/** + * @return the AVChromaLocation value for name or an AVError if not found. + */ +int av_chroma_location_from_name(const char *name); + +/** + * Converts AVChromaLocation to swscale x/y chroma position. + * + * The positions represent the chroma (0,0) position in a coordinates system + * with luma (0,0) representing the origin and luma(1,1) representing 256,256 + * + * @param xpos horizontal chroma sample position + * @param ypos vertical chroma sample position + */ +int av_chroma_location_enum_to_pos(int *xpos, int *ypos, enum AVChromaLocation pos); + +/** + * Converts swscale x/y chroma position to AVChromaLocation. + * + * The positions represent the chroma (0,0) position in a coordinates system + * with luma (0,0) representing the origin and luma(1,1) representing 256,256 + * + * @param xpos horizontal chroma sample position + * @param ypos vertical chroma sample position + */ +enum AVChromaLocation av_chroma_location_pos_to_enum(int xpos, int ypos); + +/** + * Return the pixel format corresponding to name. + * + * If there is no pixel format with name name, then looks for a + * pixel format with the name corresponding to the native endian + * format of name. + * For example in a little-endian system, first looks for "gray16", + * then for "gray16le". + * + * Finally if no pixel format has been found, returns AV_PIX_FMT_NONE. + */ +enum AVPixelFormat av_get_pix_fmt(const char *name); + +/** + * Return the short name for a pixel format, NULL in case pix_fmt is + * unknown. + * + * @see av_get_pix_fmt(), av_get_pix_fmt_string() + */ +const char *av_get_pix_fmt_name(enum AVPixelFormat pix_fmt); + +/** + * Print in buf the string corresponding to the pixel format with + * number pix_fmt, or a header if pix_fmt is negative. + * + * @param buf the buffer where to write the string + * @param buf_size the size of buf + * @param pix_fmt the number of the pixel format to print the + * corresponding info string, or a negative value to print the + * corresponding header. + */ +char *av_get_pix_fmt_string(char *buf, int buf_size, + enum AVPixelFormat pix_fmt); + +/** + * Read a line from an image, and write the values of the + * pixel format component c to dst. + * + * @param data the array containing the pointers to the planes of the image + * @param linesize the array containing the linesizes of the image + * @param desc the pixel format descriptor for the image + * @param x the horizontal coordinate of the first pixel to read + * @param y the vertical coordinate of the first pixel to read + * @param w the width of the line to read, that is the number of + * values to write to dst + * @param read_pal_component if not zero and the format is a paletted + * format writes the values corresponding to the palette + * component c in data[1] to dst, rather than the palette indexes in + * data[0]. The behavior is undefined if the format is not paletted. + * @param dst_element_size size of elements in dst array (2 or 4 byte) + */ +void av_read_image_line2(void *dst, const uint8_t *data[4], + const int linesize[4], const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, int read_pal_component, + int dst_element_size); + +void av_read_image_line(uint16_t *dst, const uint8_t *data[4], + const int linesize[4], const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, int read_pal_component); + +/** + * Write the values from src to the pixel format component c of an + * image line. + * + * @param src array containing the values to write + * @param data the array containing the pointers to the planes of the + * image to write into. It is supposed to be zeroed. + * @param linesize the array containing the linesizes of the image + * @param desc the pixel format descriptor for the image + * @param x the horizontal coordinate of the first pixel to write + * @param y the vertical coordinate of the first pixel to write + * @param w the width of the line to write, that is the number of + * values to write to the image line + * @param src_element_size size of elements in src array (2 or 4 byte) + */ +void av_write_image_line2(const void *src, uint8_t *data[4], + const int linesize[4], const AVPixFmtDescriptor *desc, + int x, int y, int c, int w, int src_element_size); + +void av_write_image_line(const uint16_t *src, uint8_t *data[4], + const int linesize[4], const AVPixFmtDescriptor *desc, + int x, int y, int c, int w); + +/** + * Utility function to swap the endianness of a pixel format. + * + * @param[in] pix_fmt the pixel format + * + * @return pixel format with swapped endianness if it exists, + * otherwise AV_PIX_FMT_NONE + */ +enum AVPixelFormat av_pix_fmt_swap_endianness(enum AVPixelFormat pix_fmt); + +#define FF_LOSS_RESOLUTION 0x0001 /**< loss due to resolution change */ +#define FF_LOSS_DEPTH 0x0002 /**< loss due to color depth change */ +#define FF_LOSS_COLORSPACE 0x0004 /**< loss due to color space conversion */ +#define FF_LOSS_ALPHA 0x0008 /**< loss of alpha bits */ +#define FF_LOSS_COLORQUANT 0x0010 /**< loss due to color quantization */ +#define FF_LOSS_CHROMA 0x0020 /**< loss of chroma (e.g. RGB to gray conversion) */ +#define FF_LOSS_EXCESS_RESOLUTION 0x0040 /**< loss due to unneeded extra resolution */ +#define FF_LOSS_EXCESS_DEPTH 0x0080 /**< loss due to unneeded extra color depth */ + + +/** + * Compute what kind of losses will occur when converting from one specific + * pixel format to another. + * When converting from one pixel format to another, information loss may occur. + * For example, when converting from RGB24 to GRAY, the color information will + * be lost. Similarly, other losses occur when converting from some formats to + * other formats. These losses can involve loss of chroma, but also loss of + * resolution, loss of color depth, loss due to the color space conversion, loss + * of the alpha bits or loss due to color quantization. + * av_get_fix_fmt_loss() informs you about the various types of losses + * which will occur when converting from one pixel format to another. + * + * @param[in] dst_pix_fmt destination pixel format + * @param[in] src_pix_fmt source pixel format + * @param[in] has_alpha Whether the source pixel format alpha channel is used. + * @return Combination of flags informing you what kind of losses will occur + * (maximum loss for an invalid dst_pix_fmt). + */ +int av_get_pix_fmt_loss(enum AVPixelFormat dst_pix_fmt, + enum AVPixelFormat src_pix_fmt, + int has_alpha); + +/** + * Compute what kind of losses will occur when converting from one specific + * pixel format to another. + * When converting from one pixel format to another, information loss may occur. + * For example, when converting from RGB24 to GRAY, the color information will + * be lost. Similarly, other losses occur when converting from some formats to + * other formats. These losses can involve loss of chroma, but also loss of + * resolution, loss of color depth, loss due to the color space conversion, loss + * of the alpha bits or loss due to color quantization. + * av_get_fix_fmt_loss() informs you about the various types of losses + * which will occur when converting from one pixel format to another. + * + * @param[in] dst_pix_fmt destination pixel format + * @param[in] src_pix_fmt source pixel format + * @param[in] has_alpha Whether the source pixel format alpha channel is used. + * @return Combination of flags informing you what kind of losses will occur + * (maximum loss for an invalid dst_pix_fmt). + */ +enum AVPixelFormat av_find_best_pix_fmt_of_2(enum AVPixelFormat dst_pix_fmt1, enum AVPixelFormat dst_pix_fmt2, + enum AVPixelFormat src_pix_fmt, int has_alpha, int *loss_ptr); + +#endif /* AVUTIL_PIXDESC_H */ diff --git a/media/ffvpx/libavutil/pixelutils.c b/media/ffvpx/libavutil/pixelutils.c new file mode 100644 index 0000000000..820889a143 --- /dev/null +++ b/media/ffvpx/libavutil/pixelutils.c @@ -0,0 +1,95 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stddef.h> + +#include "config.h" +#include "pixelutils.h" + +#if CONFIG_PIXELUTILS +#include <stdlib.h> +#include <string.h> + +#include "attributes.h" +#include "macros.h" + +#include "x86/pixelutils.h" + +static av_always_inline int sad_wxh(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2, + int w, int h) +{ + int x, y, sum = 0; + + for (y = 0; y < h; y++) { + for (x = 0; x < w; x++) + sum += abs(src1[x] - src2[x]); + src1 += stride1; + src2 += stride2; + } + return sum; +} + +#define DECLARE_BLOCK_FUNCTIONS(size) \ +static int block_sad_##size##x##size##_c(const uint8_t *src1, ptrdiff_t stride1, \ + const uint8_t *src2, ptrdiff_t stride2) \ +{ \ + return sad_wxh(src1, stride1, src2, stride2, size, size); \ +} + +DECLARE_BLOCK_FUNCTIONS(2) +DECLARE_BLOCK_FUNCTIONS(4) +DECLARE_BLOCK_FUNCTIONS(8) +DECLARE_BLOCK_FUNCTIONS(16) +DECLARE_BLOCK_FUNCTIONS(32) + +static const av_pixelutils_sad_fn sad_c[] = { + block_sad_2x2_c, + block_sad_4x4_c, + block_sad_8x8_c, + block_sad_16x16_c, + block_sad_32x32_c, +}; +#else +#include "log.h" +#endif /* CONFIG_PIXELUTILS */ + +av_pixelutils_sad_fn av_pixelutils_get_sad_fn(int w_bits, int h_bits, int aligned, void *log_ctx) +{ +#if !CONFIG_PIXELUTILS + av_log(log_ctx, AV_LOG_ERROR, "pixelutils support is required " + "but libavutil is not compiled with it\n"); + return NULL; +#else + av_pixelutils_sad_fn sad[FF_ARRAY_ELEMS(sad_c)]; + + memcpy(sad, sad_c, sizeof(sad)); + + if (w_bits < 1 || w_bits > FF_ARRAY_ELEMS(sad) || + h_bits < 1 || h_bits > FF_ARRAY_ELEMS(sad)) + return NULL; + if (w_bits != h_bits) // only squared sad for now + return NULL; + +#if ARCH_X86 + ff_pixelutils_sad_init_x86(sad, aligned); +#endif + + return sad[w_bits - 1]; +#endif +} diff --git a/media/ffvpx/libavutil/pixelutils.h b/media/ffvpx/libavutil/pixelutils.h new file mode 100644 index 0000000000..7a997cde1c --- /dev/null +++ b/media/ffvpx/libavutil/pixelutils.h @@ -0,0 +1,51 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_PIXELUTILS_H +#define AVUTIL_PIXELUTILS_H + +#include <stddef.h> +#include <stdint.h> + +/** + * Sum of abs(src1[x] - src2[x]) + */ +typedef int (*av_pixelutils_sad_fn)(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); + +/** + * Get a potentially optimized pointer to a Sum-of-absolute-differences + * function (see the av_pixelutils_sad_fn prototype). + * + * @param w_bits 1<<w_bits is the requested width of the block size + * @param h_bits 1<<h_bits is the requested height of the block size + * @param aligned If set to 2, the returned sad function will assume src1 and + * src2 addresses are aligned on the block size. + * If set to 1, the returned sad function will assume src1 is + * aligned on the block size. + * If set to 0, the returned sad function assume no particular + * alignment. + * @param log_ctx context used for logging, can be NULL + * + * @return a pointer to the SAD function or NULL in case of error (because of + * invalid parameters) + */ +av_pixelutils_sad_fn av_pixelutils_get_sad_fn(int w_bits, int h_bits, + int aligned, void *log_ctx); + +#endif /* AVUTIL_PIXELUTILS_H */ diff --git a/media/ffvpx/libavutil/pixfmt.h b/media/ffvpx/libavutil/pixfmt.h new file mode 100644 index 0000000000..9c87571f49 --- /dev/null +++ b/media/ffvpx/libavutil/pixfmt.h @@ -0,0 +1,710 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_PIXFMT_H +#define AVUTIL_PIXFMT_H + +/** + * @file + * pixel format definitions + */ + +#include "libavutil/avconfig.h" +#include "version.h" + +#define AVPALETTE_SIZE 1024 +#define AVPALETTE_COUNT 256 + +/** + * Pixel format. + * + * @note + * AV_PIX_FMT_RGB32 is handled in an endian-specific manner. An RGBA + * color is put together as: + * (A << 24) | (R << 16) | (G << 8) | B + * This is stored as BGRA on little-endian CPU architectures and ARGB on + * big-endian CPUs. + * + * @note + * If the resolution is not a multiple of the chroma subsampling factor + * then the chroma plane resolution must be rounded up. + * + * @par + * When the pixel format is palettized RGB32 (AV_PIX_FMT_PAL8), the palettized + * image data is stored in AVFrame.data[0]. The palette is transported in + * AVFrame.data[1], is 1024 bytes long (256 4-byte entries) and is + * formatted the same as in AV_PIX_FMT_RGB32 described above (i.e., it is + * also endian-specific). Note also that the individual RGB32 palette + * components stored in AVFrame.data[1] should be in the range 0..255. + * This is important as many custom PAL8 video codecs that were designed + * to run on the IBM VGA graphics adapter use 6-bit palette components. + * + * @par + * For all the 8 bits per pixel formats, an RGB32 palette is in data[1] like + * for pal8. This palette is filled in automatically by the function + * allocating the picture. + */ +enum AVPixelFormat { + AV_PIX_FMT_NONE = -1, + AV_PIX_FMT_YUV420P, ///< planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples) + AV_PIX_FMT_YUYV422, ///< packed YUV 4:2:2, 16bpp, Y0 Cb Y1 Cr + AV_PIX_FMT_RGB24, ///< packed RGB 8:8:8, 24bpp, RGBRGB... + AV_PIX_FMT_BGR24, ///< packed RGB 8:8:8, 24bpp, BGRBGR... + AV_PIX_FMT_YUV422P, ///< planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples) + AV_PIX_FMT_YUV444P, ///< planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples) + AV_PIX_FMT_YUV410P, ///< planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples) + AV_PIX_FMT_YUV411P, ///< planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples) + AV_PIX_FMT_GRAY8, ///< Y , 8bpp + AV_PIX_FMT_MONOWHITE, ///< Y , 1bpp, 0 is white, 1 is black, in each byte pixels are ordered from the msb to the lsb + AV_PIX_FMT_MONOBLACK, ///< Y , 1bpp, 0 is black, 1 is white, in each byte pixels are ordered from the msb to the lsb + AV_PIX_FMT_PAL8, ///< 8 bits with AV_PIX_FMT_RGB32 palette + AV_PIX_FMT_YUVJ420P, ///< planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV420P and setting color_range + AV_PIX_FMT_YUVJ422P, ///< planar YUV 4:2:2, 16bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV422P and setting color_range + AV_PIX_FMT_YUVJ444P, ///< planar YUV 4:4:4, 24bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV444P and setting color_range + AV_PIX_FMT_UYVY422, ///< packed YUV 4:2:2, 16bpp, Cb Y0 Cr Y1 + AV_PIX_FMT_UYYVYY411, ///< packed YUV 4:1:1, 12bpp, Cb Y0 Y1 Cr Y2 Y3 + AV_PIX_FMT_BGR8, ///< packed RGB 3:3:2, 8bpp, (msb)2B 3G 3R(lsb) + AV_PIX_FMT_BGR4, ///< packed RGB 1:2:1 bitstream, 4bpp, (msb)1B 2G 1R(lsb), a byte contains two pixels, the first pixel in the byte is the one composed by the 4 msb bits + AV_PIX_FMT_BGR4_BYTE, ///< packed RGB 1:2:1, 8bpp, (msb)1B 2G 1R(lsb) + AV_PIX_FMT_RGB8, ///< packed RGB 3:3:2, 8bpp, (msb)3R 3G 2B(lsb) + AV_PIX_FMT_RGB4, ///< packed RGB 1:2:1 bitstream, 4bpp, (msb)1R 2G 1B(lsb), a byte contains two pixels, the first pixel in the byte is the one composed by the 4 msb bits + AV_PIX_FMT_RGB4_BYTE, ///< packed RGB 1:2:1, 8bpp, (msb)1R 2G 1B(lsb) + AV_PIX_FMT_NV12, ///< planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (first byte U and the following byte V) + AV_PIX_FMT_NV21, ///< as above, but U and V bytes are swapped + + AV_PIX_FMT_ARGB, ///< packed ARGB 8:8:8:8, 32bpp, ARGBARGB... + AV_PIX_FMT_RGBA, ///< packed RGBA 8:8:8:8, 32bpp, RGBARGBA... + AV_PIX_FMT_ABGR, ///< packed ABGR 8:8:8:8, 32bpp, ABGRABGR... + AV_PIX_FMT_BGRA, ///< packed BGRA 8:8:8:8, 32bpp, BGRABGRA... + + AV_PIX_FMT_GRAY16BE, ///< Y , 16bpp, big-endian + AV_PIX_FMT_GRAY16LE, ///< Y , 16bpp, little-endian + AV_PIX_FMT_YUV440P, ///< planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples) + AV_PIX_FMT_YUVJ440P, ///< planar YUV 4:4:0 full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV440P and setting color_range + AV_PIX_FMT_YUVA420P, ///< planar YUV 4:2:0, 20bpp, (1 Cr & Cb sample per 2x2 Y & A samples) + AV_PIX_FMT_RGB48BE, ///< packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as big-endian + AV_PIX_FMT_RGB48LE, ///< packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as little-endian + + AV_PIX_FMT_RGB565BE, ///< packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), big-endian + AV_PIX_FMT_RGB565LE, ///< packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), little-endian + AV_PIX_FMT_RGB555BE, ///< packed RGB 5:5:5, 16bpp, (msb)1X 5R 5G 5B(lsb), big-endian , X=unused/undefined + AV_PIX_FMT_RGB555LE, ///< packed RGB 5:5:5, 16bpp, (msb)1X 5R 5G 5B(lsb), little-endian, X=unused/undefined + + AV_PIX_FMT_BGR565BE, ///< packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), big-endian + AV_PIX_FMT_BGR565LE, ///< packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), little-endian + AV_PIX_FMT_BGR555BE, ///< packed BGR 5:5:5, 16bpp, (msb)1X 5B 5G 5R(lsb), big-endian , X=unused/undefined + AV_PIX_FMT_BGR555LE, ///< packed BGR 5:5:5, 16bpp, (msb)1X 5B 5G 5R(lsb), little-endian, X=unused/undefined + + /** + * Hardware acceleration through VA-API, data[3] contains a + * VASurfaceID. + */ + AV_PIX_FMT_VAAPI, + + AV_PIX_FMT_YUV420P16LE, ///< planar YUV 4:2:0, 24bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian + AV_PIX_FMT_YUV420P16BE, ///< planar YUV 4:2:0, 24bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian + AV_PIX_FMT_YUV422P16LE, ///< planar YUV 4:2:2, 32bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_YUV422P16BE, ///< planar YUV 4:2:2, 32bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + AV_PIX_FMT_YUV444P16LE, ///< planar YUV 4:4:4, 48bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian + AV_PIX_FMT_YUV444P16BE, ///< planar YUV 4:4:4, 48bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian + AV_PIX_FMT_DXVA2_VLD, ///< HW decoding through DXVA2, Picture.data[3] contains a LPDIRECT3DSURFACE9 pointer + + AV_PIX_FMT_RGB444LE, ///< packed RGB 4:4:4, 16bpp, (msb)4X 4R 4G 4B(lsb), little-endian, X=unused/undefined + AV_PIX_FMT_RGB444BE, ///< packed RGB 4:4:4, 16bpp, (msb)4X 4R 4G 4B(lsb), big-endian, X=unused/undefined + AV_PIX_FMT_BGR444LE, ///< packed BGR 4:4:4, 16bpp, (msb)4X 4B 4G 4R(lsb), little-endian, X=unused/undefined + AV_PIX_FMT_BGR444BE, ///< packed BGR 4:4:4, 16bpp, (msb)4X 4B 4G 4R(lsb), big-endian, X=unused/undefined + AV_PIX_FMT_YA8, ///< 8 bits gray, 8 bits alpha + + AV_PIX_FMT_Y400A = AV_PIX_FMT_YA8, ///< alias for AV_PIX_FMT_YA8 + AV_PIX_FMT_GRAY8A= AV_PIX_FMT_YA8, ///< alias for AV_PIX_FMT_YA8 + + AV_PIX_FMT_BGR48BE, ///< packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as big-endian + AV_PIX_FMT_BGR48LE, ///< packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as little-endian + + /** + * The following 12 formats have the disadvantage of needing 1 format for each bit depth. + * Notice that each 9/10 bits sample is stored in 16 bits with extra padding. + * If you want to support multiple bit depths, then using AV_PIX_FMT_YUV420P16* with the bpp stored separately is better. + */ + AV_PIX_FMT_YUV420P9BE, ///< planar YUV 4:2:0, 13.5bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian + AV_PIX_FMT_YUV420P9LE, ///< planar YUV 4:2:0, 13.5bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian + AV_PIX_FMT_YUV420P10BE,///< planar YUV 4:2:0, 15bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian + AV_PIX_FMT_YUV420P10LE,///< planar YUV 4:2:0, 15bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian + AV_PIX_FMT_YUV422P10BE,///< planar YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + AV_PIX_FMT_YUV422P10LE,///< planar YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_YUV444P9BE, ///< planar YUV 4:4:4, 27bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian + AV_PIX_FMT_YUV444P9LE, ///< planar YUV 4:4:4, 27bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian + AV_PIX_FMT_YUV444P10BE,///< planar YUV 4:4:4, 30bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian + AV_PIX_FMT_YUV444P10LE,///< planar YUV 4:4:4, 30bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian + AV_PIX_FMT_YUV422P9BE, ///< planar YUV 4:2:2, 18bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + AV_PIX_FMT_YUV422P9LE, ///< planar YUV 4:2:2, 18bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_GBRP, ///< planar GBR 4:4:4 24bpp + AV_PIX_FMT_GBR24P = AV_PIX_FMT_GBRP, // alias for #AV_PIX_FMT_GBRP + AV_PIX_FMT_GBRP9BE, ///< planar GBR 4:4:4 27bpp, big-endian + AV_PIX_FMT_GBRP9LE, ///< planar GBR 4:4:4 27bpp, little-endian + AV_PIX_FMT_GBRP10BE, ///< planar GBR 4:4:4 30bpp, big-endian + AV_PIX_FMT_GBRP10LE, ///< planar GBR 4:4:4 30bpp, little-endian + AV_PIX_FMT_GBRP16BE, ///< planar GBR 4:4:4 48bpp, big-endian + AV_PIX_FMT_GBRP16LE, ///< planar GBR 4:4:4 48bpp, little-endian + AV_PIX_FMT_YUVA422P, ///< planar YUV 4:2:2 24bpp, (1 Cr & Cb sample per 2x1 Y & A samples) + AV_PIX_FMT_YUVA444P, ///< planar YUV 4:4:4 32bpp, (1 Cr & Cb sample per 1x1 Y & A samples) + AV_PIX_FMT_YUVA420P9BE, ///< planar YUV 4:2:0 22.5bpp, (1 Cr & Cb sample per 2x2 Y & A samples), big-endian + AV_PIX_FMT_YUVA420P9LE, ///< planar YUV 4:2:0 22.5bpp, (1 Cr & Cb sample per 2x2 Y & A samples), little-endian + AV_PIX_FMT_YUVA422P9BE, ///< planar YUV 4:2:2 27bpp, (1 Cr & Cb sample per 2x1 Y & A samples), big-endian + AV_PIX_FMT_YUVA422P9LE, ///< planar YUV 4:2:2 27bpp, (1 Cr & Cb sample per 2x1 Y & A samples), little-endian + AV_PIX_FMT_YUVA444P9BE, ///< planar YUV 4:4:4 36bpp, (1 Cr & Cb sample per 1x1 Y & A samples), big-endian + AV_PIX_FMT_YUVA444P9LE, ///< planar YUV 4:4:4 36bpp, (1 Cr & Cb sample per 1x1 Y & A samples), little-endian + AV_PIX_FMT_YUVA420P10BE, ///< planar YUV 4:2:0 25bpp, (1 Cr & Cb sample per 2x2 Y & A samples, big-endian) + AV_PIX_FMT_YUVA420P10LE, ///< planar YUV 4:2:0 25bpp, (1 Cr & Cb sample per 2x2 Y & A samples, little-endian) + AV_PIX_FMT_YUVA422P10BE, ///< planar YUV 4:2:2 30bpp, (1 Cr & Cb sample per 2x1 Y & A samples, big-endian) + AV_PIX_FMT_YUVA422P10LE, ///< planar YUV 4:2:2 30bpp, (1 Cr & Cb sample per 2x1 Y & A samples, little-endian) + AV_PIX_FMT_YUVA444P10BE, ///< planar YUV 4:4:4 40bpp, (1 Cr & Cb sample per 1x1 Y & A samples, big-endian) + AV_PIX_FMT_YUVA444P10LE, ///< planar YUV 4:4:4 40bpp, (1 Cr & Cb sample per 1x1 Y & A samples, little-endian) + AV_PIX_FMT_YUVA420P16BE, ///< planar YUV 4:2:0 40bpp, (1 Cr & Cb sample per 2x2 Y & A samples, big-endian) + AV_PIX_FMT_YUVA420P16LE, ///< planar YUV 4:2:0 40bpp, (1 Cr & Cb sample per 2x2 Y & A samples, little-endian) + AV_PIX_FMT_YUVA422P16BE, ///< planar YUV 4:2:2 48bpp, (1 Cr & Cb sample per 2x1 Y & A samples, big-endian) + AV_PIX_FMT_YUVA422P16LE, ///< planar YUV 4:2:2 48bpp, (1 Cr & Cb sample per 2x1 Y & A samples, little-endian) + AV_PIX_FMT_YUVA444P16BE, ///< planar YUV 4:4:4 64bpp, (1 Cr & Cb sample per 1x1 Y & A samples, big-endian) + AV_PIX_FMT_YUVA444P16LE, ///< planar YUV 4:4:4 64bpp, (1 Cr & Cb sample per 1x1 Y & A samples, little-endian) + + AV_PIX_FMT_VDPAU, ///< HW acceleration through VDPAU, Picture.data[3] contains a VdpVideoSurface + + AV_PIX_FMT_XYZ12LE, ///< packed XYZ 4:4:4, 36 bpp, (msb) 12X, 12Y, 12Z (lsb), the 2-byte value for each X/Y/Z is stored as little-endian, the 4 lower bits are set to 0 + AV_PIX_FMT_XYZ12BE, ///< packed XYZ 4:4:4, 36 bpp, (msb) 12X, 12Y, 12Z (lsb), the 2-byte value for each X/Y/Z is stored as big-endian, the 4 lower bits are set to 0 + AV_PIX_FMT_NV16, ///< interleaved chroma YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples) + AV_PIX_FMT_NV20LE, ///< interleaved chroma YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_NV20BE, ///< interleaved chroma YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + + AV_PIX_FMT_RGBA64BE, ///< packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is stored as big-endian + AV_PIX_FMT_RGBA64LE, ///< packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is stored as little-endian + AV_PIX_FMT_BGRA64BE, ///< packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is stored as big-endian + AV_PIX_FMT_BGRA64LE, ///< packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is stored as little-endian + + AV_PIX_FMT_YVYU422, ///< packed YUV 4:2:2, 16bpp, Y0 Cr Y1 Cb + + AV_PIX_FMT_YA16BE, ///< 16 bits gray, 16 bits alpha (big-endian) + AV_PIX_FMT_YA16LE, ///< 16 bits gray, 16 bits alpha (little-endian) + + AV_PIX_FMT_GBRAP, ///< planar GBRA 4:4:4:4 32bpp + AV_PIX_FMT_GBRAP16BE, ///< planar GBRA 4:4:4:4 64bpp, big-endian + AV_PIX_FMT_GBRAP16LE, ///< planar GBRA 4:4:4:4 64bpp, little-endian + /** + * HW acceleration through QSV, data[3] contains a pointer to the + * mfxFrameSurface1 structure. + * + * Before FFmpeg 5.0: + * mfxFrameSurface1.Data.MemId contains a pointer when importing + * the following frames as QSV frames: + * + * VAAPI: + * mfxFrameSurface1.Data.MemId contains a pointer to VASurfaceID + * + * DXVA2: + * mfxFrameSurface1.Data.MemId contains a pointer to IDirect3DSurface9 + * + * FFmpeg 5.0 and above: + * mfxFrameSurface1.Data.MemId contains a pointer to the mfxHDLPair + * structure when importing the following frames as QSV frames: + * + * VAAPI: + * mfxHDLPair.first contains a VASurfaceID pointer. + * mfxHDLPair.second is always MFX_INFINITE. + * + * DXVA2: + * mfxHDLPair.first contains IDirect3DSurface9 pointer. + * mfxHDLPair.second is always MFX_INFINITE. + * + * D3D11: + * mfxHDLPair.first contains a ID3D11Texture2D pointer. + * mfxHDLPair.second contains the texture array index of the frame if the + * ID3D11Texture2D is an array texture, or always MFX_INFINITE if it is a + * normal texture. + */ + AV_PIX_FMT_QSV, + /** + * HW acceleration though MMAL, data[3] contains a pointer to the + * MMAL_BUFFER_HEADER_T structure. + */ + AV_PIX_FMT_MMAL, + + AV_PIX_FMT_D3D11VA_VLD, ///< HW decoding through Direct3D11 via old API, Picture.data[3] contains a ID3D11VideoDecoderOutputView pointer + + /** + * HW acceleration through CUDA. data[i] contain CUdeviceptr pointers + * exactly as for system memory frames. + */ + AV_PIX_FMT_CUDA, + + AV_PIX_FMT_0RGB, ///< packed RGB 8:8:8, 32bpp, XRGBXRGB... X=unused/undefined + AV_PIX_FMT_RGB0, ///< packed RGB 8:8:8, 32bpp, RGBXRGBX... X=unused/undefined + AV_PIX_FMT_0BGR, ///< packed BGR 8:8:8, 32bpp, XBGRXBGR... X=unused/undefined + AV_PIX_FMT_BGR0, ///< packed BGR 8:8:8, 32bpp, BGRXBGRX... X=unused/undefined + + AV_PIX_FMT_YUV420P12BE, ///< planar YUV 4:2:0,18bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian + AV_PIX_FMT_YUV420P12LE, ///< planar YUV 4:2:0,18bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian + AV_PIX_FMT_YUV420P14BE, ///< planar YUV 4:2:0,21bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian + AV_PIX_FMT_YUV420P14LE, ///< planar YUV 4:2:0,21bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian + AV_PIX_FMT_YUV422P12BE, ///< planar YUV 4:2:2,24bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + AV_PIX_FMT_YUV422P12LE, ///< planar YUV 4:2:2,24bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_YUV422P14BE, ///< planar YUV 4:2:2,28bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian + AV_PIX_FMT_YUV422P14LE, ///< planar YUV 4:2:2,28bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian + AV_PIX_FMT_YUV444P12BE, ///< planar YUV 4:4:4,36bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian + AV_PIX_FMT_YUV444P12LE, ///< planar YUV 4:4:4,36bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian + AV_PIX_FMT_YUV444P14BE, ///< planar YUV 4:4:4,42bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian + AV_PIX_FMT_YUV444P14LE, ///< planar YUV 4:4:4,42bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian + AV_PIX_FMT_GBRP12BE, ///< planar GBR 4:4:4 36bpp, big-endian + AV_PIX_FMT_GBRP12LE, ///< planar GBR 4:4:4 36bpp, little-endian + AV_PIX_FMT_GBRP14BE, ///< planar GBR 4:4:4 42bpp, big-endian + AV_PIX_FMT_GBRP14LE, ///< planar GBR 4:4:4 42bpp, little-endian + AV_PIX_FMT_YUVJ411P, ///< planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples) full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV411P and setting color_range + + AV_PIX_FMT_BAYER_BGGR8, ///< bayer, BGBG..(odd line), GRGR..(even line), 8-bit samples + AV_PIX_FMT_BAYER_RGGB8, ///< bayer, RGRG..(odd line), GBGB..(even line), 8-bit samples + AV_PIX_FMT_BAYER_GBRG8, ///< bayer, GBGB..(odd line), RGRG..(even line), 8-bit samples + AV_PIX_FMT_BAYER_GRBG8, ///< bayer, GRGR..(odd line), BGBG..(even line), 8-bit samples + AV_PIX_FMT_BAYER_BGGR16LE, ///< bayer, BGBG..(odd line), GRGR..(even line), 16-bit samples, little-endian + AV_PIX_FMT_BAYER_BGGR16BE, ///< bayer, BGBG..(odd line), GRGR..(even line), 16-bit samples, big-endian + AV_PIX_FMT_BAYER_RGGB16LE, ///< bayer, RGRG..(odd line), GBGB..(even line), 16-bit samples, little-endian + AV_PIX_FMT_BAYER_RGGB16BE, ///< bayer, RGRG..(odd line), GBGB..(even line), 16-bit samples, big-endian + AV_PIX_FMT_BAYER_GBRG16LE, ///< bayer, GBGB..(odd line), RGRG..(even line), 16-bit samples, little-endian + AV_PIX_FMT_BAYER_GBRG16BE, ///< bayer, GBGB..(odd line), RGRG..(even line), 16-bit samples, big-endian + AV_PIX_FMT_BAYER_GRBG16LE, ///< bayer, GRGR..(odd line), BGBG..(even line), 16-bit samples, little-endian + AV_PIX_FMT_BAYER_GRBG16BE, ///< bayer, GRGR..(odd line), BGBG..(even line), 16-bit samples, big-endian + +#if FF_API_XVMC + AV_PIX_FMT_XVMC,///< XVideo Motion Acceleration via common packet passing +#endif + + AV_PIX_FMT_YUV440P10LE, ///< planar YUV 4:4:0,20bpp, (1 Cr & Cb sample per 1x2 Y samples), little-endian + AV_PIX_FMT_YUV440P10BE, ///< planar YUV 4:4:0,20bpp, (1 Cr & Cb sample per 1x2 Y samples), big-endian + AV_PIX_FMT_YUV440P12LE, ///< planar YUV 4:4:0,24bpp, (1 Cr & Cb sample per 1x2 Y samples), little-endian + AV_PIX_FMT_YUV440P12BE, ///< planar YUV 4:4:0,24bpp, (1 Cr & Cb sample per 1x2 Y samples), big-endian + AV_PIX_FMT_AYUV64LE, ///< packed AYUV 4:4:4,64bpp (1 Cr & Cb sample per 1x1 Y & A samples), little-endian + AV_PIX_FMT_AYUV64BE, ///< packed AYUV 4:4:4,64bpp (1 Cr & Cb sample per 1x1 Y & A samples), big-endian + + AV_PIX_FMT_VIDEOTOOLBOX, ///< hardware decoding through Videotoolbox + + AV_PIX_FMT_P010LE, ///< like NV12, with 10bpp per component, data in the high bits, zeros in the low bits, little-endian + AV_PIX_FMT_P010BE, ///< like NV12, with 10bpp per component, data in the high bits, zeros in the low bits, big-endian + + AV_PIX_FMT_GBRAP12BE, ///< planar GBR 4:4:4:4 48bpp, big-endian + AV_PIX_FMT_GBRAP12LE, ///< planar GBR 4:4:4:4 48bpp, little-endian + + AV_PIX_FMT_GBRAP10BE, ///< planar GBR 4:4:4:4 40bpp, big-endian + AV_PIX_FMT_GBRAP10LE, ///< planar GBR 4:4:4:4 40bpp, little-endian + + AV_PIX_FMT_MEDIACODEC, ///< hardware decoding through MediaCodec + + AV_PIX_FMT_GRAY12BE, ///< Y , 12bpp, big-endian + AV_PIX_FMT_GRAY12LE, ///< Y , 12bpp, little-endian + AV_PIX_FMT_GRAY10BE, ///< Y , 10bpp, big-endian + AV_PIX_FMT_GRAY10LE, ///< Y , 10bpp, little-endian + + AV_PIX_FMT_P016LE, ///< like NV12, with 16bpp per component, little-endian + AV_PIX_FMT_P016BE, ///< like NV12, with 16bpp per component, big-endian + + /** + * Hardware surfaces for Direct3D11. + * + * This is preferred over the legacy AV_PIX_FMT_D3D11VA_VLD. The new D3D11 + * hwaccel API and filtering support AV_PIX_FMT_D3D11 only. + * + * data[0] contains a ID3D11Texture2D pointer, and data[1] contains the + * texture array index of the frame as intptr_t if the ID3D11Texture2D is + * an array texture (or always 0 if it's a normal texture). + */ + AV_PIX_FMT_D3D11, + + AV_PIX_FMT_GRAY9BE, ///< Y , 9bpp, big-endian + AV_PIX_FMT_GRAY9LE, ///< Y , 9bpp, little-endian + + AV_PIX_FMT_GBRPF32BE, ///< IEEE-754 single precision planar GBR 4:4:4, 96bpp, big-endian + AV_PIX_FMT_GBRPF32LE, ///< IEEE-754 single precision planar GBR 4:4:4, 96bpp, little-endian + AV_PIX_FMT_GBRAPF32BE, ///< IEEE-754 single precision planar GBRA 4:4:4:4, 128bpp, big-endian + AV_PIX_FMT_GBRAPF32LE, ///< IEEE-754 single precision planar GBRA 4:4:4:4, 128bpp, little-endian + + /** + * DRM-managed buffers exposed through PRIME buffer sharing. + * + * data[0] points to an AVDRMFrameDescriptor. + */ + AV_PIX_FMT_DRM_PRIME, + /** + * Hardware surfaces for OpenCL. + * + * data[i] contain 2D image objects (typed in C as cl_mem, used + * in OpenCL as image2d_t) for each plane of the surface. + */ + AV_PIX_FMT_OPENCL, + + AV_PIX_FMT_GRAY14BE, ///< Y , 14bpp, big-endian + AV_PIX_FMT_GRAY14LE, ///< Y , 14bpp, little-endian + + AV_PIX_FMT_GRAYF32BE, ///< IEEE-754 single precision Y, 32bpp, big-endian + AV_PIX_FMT_GRAYF32LE, ///< IEEE-754 single precision Y, 32bpp, little-endian + + AV_PIX_FMT_YUVA422P12BE, ///< planar YUV 4:2:2,24bpp, (1 Cr & Cb sample per 2x1 Y samples), 12b alpha, big-endian + AV_PIX_FMT_YUVA422P12LE, ///< planar YUV 4:2:2,24bpp, (1 Cr & Cb sample per 2x1 Y samples), 12b alpha, little-endian + AV_PIX_FMT_YUVA444P12BE, ///< planar YUV 4:4:4,36bpp, (1 Cr & Cb sample per 1x1 Y samples), 12b alpha, big-endian + AV_PIX_FMT_YUVA444P12LE, ///< planar YUV 4:4:4,36bpp, (1 Cr & Cb sample per 1x1 Y samples), 12b alpha, little-endian + + AV_PIX_FMT_NV24, ///< planar YUV 4:4:4, 24bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (first byte U and the following byte V) + AV_PIX_FMT_NV42, ///< as above, but U and V bytes are swapped + + /** + * Vulkan hardware images. + * + * data[0] points to an AVVkFrame + */ + AV_PIX_FMT_VULKAN, + + AV_PIX_FMT_Y210BE, ///< packed YUV 4:2:2 like YUYV422, 20bpp, data in the high bits, big-endian + AV_PIX_FMT_Y210LE, ///< packed YUV 4:2:2 like YUYV422, 20bpp, data in the high bits, little-endian + + AV_PIX_FMT_X2RGB10LE, ///< packed RGB 10:10:10, 30bpp, (msb)2X 10R 10G 10B(lsb), little-endian, X=unused/undefined + AV_PIX_FMT_X2RGB10BE, ///< packed RGB 10:10:10, 30bpp, (msb)2X 10R 10G 10B(lsb), big-endian, X=unused/undefined + AV_PIX_FMT_X2BGR10LE, ///< packed BGR 10:10:10, 30bpp, (msb)2X 10B 10G 10R(lsb), little-endian, X=unused/undefined + AV_PIX_FMT_X2BGR10BE, ///< packed BGR 10:10:10, 30bpp, (msb)2X 10B 10G 10R(lsb), big-endian, X=unused/undefined + + AV_PIX_FMT_P210BE, ///< interleaved chroma YUV 4:2:2, 20bpp, data in the high bits, big-endian + AV_PIX_FMT_P210LE, ///< interleaved chroma YUV 4:2:2, 20bpp, data in the high bits, little-endian + + AV_PIX_FMT_P410BE, ///< interleaved chroma YUV 4:4:4, 30bpp, data in the high bits, big-endian + AV_PIX_FMT_P410LE, ///< interleaved chroma YUV 4:4:4, 30bpp, data in the high bits, little-endian + + AV_PIX_FMT_P216BE, ///< interleaved chroma YUV 4:2:2, 32bpp, big-endian + AV_PIX_FMT_P216LE, ///< interleaved chroma YUV 4:2:2, 32bpp, little-endian + + AV_PIX_FMT_P416BE, ///< interleaved chroma YUV 4:4:4, 48bpp, big-endian + AV_PIX_FMT_P416LE, ///< interleaved chroma YUV 4:4:4, 48bpp, little-endian + + AV_PIX_FMT_VUYA, ///< packed VUYA 4:4:4, 32bpp, VUYAVUYA... + + AV_PIX_FMT_RGBAF16BE, ///< IEEE-754 half precision packed RGBA 16:16:16:16, 64bpp, RGBARGBA..., big-endian + AV_PIX_FMT_RGBAF16LE, ///< IEEE-754 half precision packed RGBA 16:16:16:16, 64bpp, RGBARGBA..., little-endian + + AV_PIX_FMT_VUYX, ///< packed VUYX 4:4:4, 32bpp, Variant of VUYA where alpha channel is left undefined + + AV_PIX_FMT_P012LE, ///< like NV12, with 12bpp per component, data in the high bits, zeros in the low bits, little-endian + AV_PIX_FMT_P012BE, ///< like NV12, with 12bpp per component, data in the high bits, zeros in the low bits, big-endian + + AV_PIX_FMT_Y212BE, ///< packed YUV 4:2:2 like YUYV422, 24bpp, data in the high bits, zeros in the low bits, big-endian + AV_PIX_FMT_Y212LE, ///< packed YUV 4:2:2 like YUYV422, 24bpp, data in the high bits, zeros in the low bits, little-endian + + AV_PIX_FMT_XV30BE, ///< packed XVYU 4:4:4, 32bpp, (msb)2X 10V 10Y 10U(lsb), big-endian, variant of Y410 where alpha channel is left undefined + AV_PIX_FMT_XV30LE, ///< packed XVYU 4:4:4, 32bpp, (msb)2X 10V 10Y 10U(lsb), little-endian, variant of Y410 where alpha channel is left undefined + + AV_PIX_FMT_XV36BE, ///< packed XVYU 4:4:4, 48bpp, data in the high bits, zeros in the low bits, big-endian, variant of Y412 where alpha channel is left undefined + AV_PIX_FMT_XV36LE, ///< packed XVYU 4:4:4, 48bpp, data in the high bits, zeros in the low bits, little-endian, variant of Y412 where alpha channel is left undefined + + AV_PIX_FMT_RGBF32BE, ///< IEEE-754 single precision packed RGB 32:32:32, 96bpp, RGBRGB..., big-endian + AV_PIX_FMT_RGBF32LE, ///< IEEE-754 single precision packed RGB 32:32:32, 96bpp, RGBRGB..., little-endian + + AV_PIX_FMT_RGBAF32BE, ///< IEEE-754 single precision packed RGBA 32:32:32:32, 128bpp, RGBARGBA..., big-endian + AV_PIX_FMT_RGBAF32LE, ///< IEEE-754 single precision packed RGBA 32:32:32:32, 128bpp, RGBARGBA..., little-endian + + AV_PIX_FMT_P212BE, ///< interleaved chroma YUV 4:2:2, 24bpp, data in the high bits, big-endian + AV_PIX_FMT_P212LE, ///< interleaved chroma YUV 4:2:2, 24bpp, data in the high bits, little-endian + + AV_PIX_FMT_P412BE, ///< interleaved chroma YUV 4:4:4, 36bpp, data in the high bits, big-endian + AV_PIX_FMT_P412LE, ///< interleaved chroma YUV 4:4:4, 36bpp, data in the high bits, little-endian + + AV_PIX_FMT_GBRAP14BE, ///< planar GBR 4:4:4:4 56bpp, big-endian + AV_PIX_FMT_GBRAP14LE, ///< planar GBR 4:4:4:4 56bpp, little-endian + + /** + * Hardware surfaces for Direct3D 12. + * + * data[0] points to an AVD3D12VAFrame + */ + AV_PIX_FMT_D3D12, + + AV_PIX_FMT_NB ///< number of pixel formats, DO NOT USE THIS if you want to link with shared libav* because the number of formats might differ between versions +}; + +#if AV_HAVE_BIGENDIAN +# define AV_PIX_FMT_NE(be, le) AV_PIX_FMT_##be +#else +# define AV_PIX_FMT_NE(be, le) AV_PIX_FMT_##le +#endif + +#define AV_PIX_FMT_RGB32 AV_PIX_FMT_NE(ARGB, BGRA) +#define AV_PIX_FMT_RGB32_1 AV_PIX_FMT_NE(RGBA, ABGR) +#define AV_PIX_FMT_BGR32 AV_PIX_FMT_NE(ABGR, RGBA) +#define AV_PIX_FMT_BGR32_1 AV_PIX_FMT_NE(BGRA, ARGB) +#define AV_PIX_FMT_0RGB32 AV_PIX_FMT_NE(0RGB, BGR0) +#define AV_PIX_FMT_0BGR32 AV_PIX_FMT_NE(0BGR, RGB0) + +#define AV_PIX_FMT_GRAY9 AV_PIX_FMT_NE(GRAY9BE, GRAY9LE) +#define AV_PIX_FMT_GRAY10 AV_PIX_FMT_NE(GRAY10BE, GRAY10LE) +#define AV_PIX_FMT_GRAY12 AV_PIX_FMT_NE(GRAY12BE, GRAY12LE) +#define AV_PIX_FMT_GRAY14 AV_PIX_FMT_NE(GRAY14BE, GRAY14LE) +#define AV_PIX_FMT_GRAY16 AV_PIX_FMT_NE(GRAY16BE, GRAY16LE) +#define AV_PIX_FMT_YA16 AV_PIX_FMT_NE(YA16BE, YA16LE) +#define AV_PIX_FMT_RGB48 AV_PIX_FMT_NE(RGB48BE, RGB48LE) +#define AV_PIX_FMT_RGB565 AV_PIX_FMT_NE(RGB565BE, RGB565LE) +#define AV_PIX_FMT_RGB555 AV_PIX_FMT_NE(RGB555BE, RGB555LE) +#define AV_PIX_FMT_RGB444 AV_PIX_FMT_NE(RGB444BE, RGB444LE) +#define AV_PIX_FMT_RGBA64 AV_PIX_FMT_NE(RGBA64BE, RGBA64LE) +#define AV_PIX_FMT_BGR48 AV_PIX_FMT_NE(BGR48BE, BGR48LE) +#define AV_PIX_FMT_BGR565 AV_PIX_FMT_NE(BGR565BE, BGR565LE) +#define AV_PIX_FMT_BGR555 AV_PIX_FMT_NE(BGR555BE, BGR555LE) +#define AV_PIX_FMT_BGR444 AV_PIX_FMT_NE(BGR444BE, BGR444LE) +#define AV_PIX_FMT_BGRA64 AV_PIX_FMT_NE(BGRA64BE, BGRA64LE) + +#define AV_PIX_FMT_YUV420P9 AV_PIX_FMT_NE(YUV420P9BE , YUV420P9LE) +#define AV_PIX_FMT_YUV422P9 AV_PIX_FMT_NE(YUV422P9BE , YUV422P9LE) +#define AV_PIX_FMT_YUV444P9 AV_PIX_FMT_NE(YUV444P9BE , YUV444P9LE) +#define AV_PIX_FMT_YUV420P10 AV_PIX_FMT_NE(YUV420P10BE, YUV420P10LE) +#define AV_PIX_FMT_YUV422P10 AV_PIX_FMT_NE(YUV422P10BE, YUV422P10LE) +#define AV_PIX_FMT_YUV440P10 AV_PIX_FMT_NE(YUV440P10BE, YUV440P10LE) +#define AV_PIX_FMT_YUV444P10 AV_PIX_FMT_NE(YUV444P10BE, YUV444P10LE) +#define AV_PIX_FMT_YUV420P12 AV_PIX_FMT_NE(YUV420P12BE, YUV420P12LE) +#define AV_PIX_FMT_YUV422P12 AV_PIX_FMT_NE(YUV422P12BE, YUV422P12LE) +#define AV_PIX_FMT_YUV440P12 AV_PIX_FMT_NE(YUV440P12BE, YUV440P12LE) +#define AV_PIX_FMT_YUV444P12 AV_PIX_FMT_NE(YUV444P12BE, YUV444P12LE) +#define AV_PIX_FMT_YUV420P14 AV_PIX_FMT_NE(YUV420P14BE, YUV420P14LE) +#define AV_PIX_FMT_YUV422P14 AV_PIX_FMT_NE(YUV422P14BE, YUV422P14LE) +#define AV_PIX_FMT_YUV444P14 AV_PIX_FMT_NE(YUV444P14BE, YUV444P14LE) +#define AV_PIX_FMT_YUV420P16 AV_PIX_FMT_NE(YUV420P16BE, YUV420P16LE) +#define AV_PIX_FMT_YUV422P16 AV_PIX_FMT_NE(YUV422P16BE, YUV422P16LE) +#define AV_PIX_FMT_YUV444P16 AV_PIX_FMT_NE(YUV444P16BE, YUV444P16LE) + +#define AV_PIX_FMT_GBRP9 AV_PIX_FMT_NE(GBRP9BE , GBRP9LE) +#define AV_PIX_FMT_GBRP10 AV_PIX_FMT_NE(GBRP10BE, GBRP10LE) +#define AV_PIX_FMT_GBRP12 AV_PIX_FMT_NE(GBRP12BE, GBRP12LE) +#define AV_PIX_FMT_GBRP14 AV_PIX_FMT_NE(GBRP14BE, GBRP14LE) +#define AV_PIX_FMT_GBRP16 AV_PIX_FMT_NE(GBRP16BE, GBRP16LE) +#define AV_PIX_FMT_GBRAP10 AV_PIX_FMT_NE(GBRAP10BE, GBRAP10LE) +#define AV_PIX_FMT_GBRAP12 AV_PIX_FMT_NE(GBRAP12BE, GBRAP12LE) +#define AV_PIX_FMT_GBRAP14 AV_PIX_FMT_NE(GBRAP14BE, GBRAP14LE) +#define AV_PIX_FMT_GBRAP16 AV_PIX_FMT_NE(GBRAP16BE, GBRAP16LE) + +#define AV_PIX_FMT_BAYER_BGGR16 AV_PIX_FMT_NE(BAYER_BGGR16BE, BAYER_BGGR16LE) +#define AV_PIX_FMT_BAYER_RGGB16 AV_PIX_FMT_NE(BAYER_RGGB16BE, BAYER_RGGB16LE) +#define AV_PIX_FMT_BAYER_GBRG16 AV_PIX_FMT_NE(BAYER_GBRG16BE, BAYER_GBRG16LE) +#define AV_PIX_FMT_BAYER_GRBG16 AV_PIX_FMT_NE(BAYER_GRBG16BE, BAYER_GRBG16LE) + +#define AV_PIX_FMT_GBRPF32 AV_PIX_FMT_NE(GBRPF32BE, GBRPF32LE) +#define AV_PIX_FMT_GBRAPF32 AV_PIX_FMT_NE(GBRAPF32BE, GBRAPF32LE) + +#define AV_PIX_FMT_GRAYF32 AV_PIX_FMT_NE(GRAYF32BE, GRAYF32LE) + +#define AV_PIX_FMT_YUVA420P9 AV_PIX_FMT_NE(YUVA420P9BE , YUVA420P9LE) +#define AV_PIX_FMT_YUVA422P9 AV_PIX_FMT_NE(YUVA422P9BE , YUVA422P9LE) +#define AV_PIX_FMT_YUVA444P9 AV_PIX_FMT_NE(YUVA444P9BE , YUVA444P9LE) +#define AV_PIX_FMT_YUVA420P10 AV_PIX_FMT_NE(YUVA420P10BE, YUVA420P10LE) +#define AV_PIX_FMT_YUVA422P10 AV_PIX_FMT_NE(YUVA422P10BE, YUVA422P10LE) +#define AV_PIX_FMT_YUVA444P10 AV_PIX_FMT_NE(YUVA444P10BE, YUVA444P10LE) +#define AV_PIX_FMT_YUVA422P12 AV_PIX_FMT_NE(YUVA422P12BE, YUVA422P12LE) +#define AV_PIX_FMT_YUVA444P12 AV_PIX_FMT_NE(YUVA444P12BE, YUVA444P12LE) +#define AV_PIX_FMT_YUVA420P16 AV_PIX_FMT_NE(YUVA420P16BE, YUVA420P16LE) +#define AV_PIX_FMT_YUVA422P16 AV_PIX_FMT_NE(YUVA422P16BE, YUVA422P16LE) +#define AV_PIX_FMT_YUVA444P16 AV_PIX_FMT_NE(YUVA444P16BE, YUVA444P16LE) + +#define AV_PIX_FMT_XYZ12 AV_PIX_FMT_NE(XYZ12BE, XYZ12LE) +#define AV_PIX_FMT_NV20 AV_PIX_FMT_NE(NV20BE, NV20LE) +#define AV_PIX_FMT_AYUV64 AV_PIX_FMT_NE(AYUV64BE, AYUV64LE) +#define AV_PIX_FMT_P010 AV_PIX_FMT_NE(P010BE, P010LE) +#define AV_PIX_FMT_P012 AV_PIX_FMT_NE(P012BE, P012LE) +#define AV_PIX_FMT_P016 AV_PIX_FMT_NE(P016BE, P016LE) + +#define AV_PIX_FMT_Y210 AV_PIX_FMT_NE(Y210BE, Y210LE) +#define AV_PIX_FMT_Y212 AV_PIX_FMT_NE(Y212BE, Y212LE) +#define AV_PIX_FMT_XV30 AV_PIX_FMT_NE(XV30BE, XV30LE) +#define AV_PIX_FMT_XV36 AV_PIX_FMT_NE(XV36BE, XV36LE) +#define AV_PIX_FMT_X2RGB10 AV_PIX_FMT_NE(X2RGB10BE, X2RGB10LE) +#define AV_PIX_FMT_X2BGR10 AV_PIX_FMT_NE(X2BGR10BE, X2BGR10LE) + +#define AV_PIX_FMT_P210 AV_PIX_FMT_NE(P210BE, P210LE) +#define AV_PIX_FMT_P410 AV_PIX_FMT_NE(P410BE, P410LE) +#define AV_PIX_FMT_P212 AV_PIX_FMT_NE(P212BE, P212LE) +#define AV_PIX_FMT_P412 AV_PIX_FMT_NE(P412BE, P412LE) +#define AV_PIX_FMT_P216 AV_PIX_FMT_NE(P216BE, P216LE) +#define AV_PIX_FMT_P416 AV_PIX_FMT_NE(P416BE, P416LE) + +#define AV_PIX_FMT_RGBAF16 AV_PIX_FMT_NE(RGBAF16BE, RGBAF16LE) + +#define AV_PIX_FMT_RGBF32 AV_PIX_FMT_NE(RGBF32BE, RGBF32LE) +#define AV_PIX_FMT_RGBAF32 AV_PIX_FMT_NE(RGBAF32BE, RGBAF32LE) + +/** + * Chromaticity coordinates of the source primaries. + * These values match the ones defined by ISO/IEC 23091-2_2019 subclause 8.1 and ITU-T H.273. + */ +enum AVColorPrimaries { + AVCOL_PRI_RESERVED0 = 0, + AVCOL_PRI_BT709 = 1, ///< also ITU-R BT1361 / IEC 61966-2-4 / SMPTE RP 177 Annex B + AVCOL_PRI_UNSPECIFIED = 2, + AVCOL_PRI_RESERVED = 3, + AVCOL_PRI_BT470M = 4, ///< also FCC Title 47 Code of Federal Regulations 73.682 (a)(20) + + AVCOL_PRI_BT470BG = 5, ///< also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM + AVCOL_PRI_SMPTE170M = 6, ///< also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC + AVCOL_PRI_SMPTE240M = 7, ///< identical to above, also called "SMPTE C" even though it uses D65 + AVCOL_PRI_FILM = 8, ///< colour filters using Illuminant C + AVCOL_PRI_BT2020 = 9, ///< ITU-R BT2020 + AVCOL_PRI_SMPTE428 = 10, ///< SMPTE ST 428-1 (CIE 1931 XYZ) + AVCOL_PRI_SMPTEST428_1 = AVCOL_PRI_SMPTE428, + AVCOL_PRI_SMPTE431 = 11, ///< SMPTE ST 431-2 (2011) / DCI P3 + AVCOL_PRI_SMPTE432 = 12, ///< SMPTE ST 432-1 (2010) / P3 D65 / Display P3 + AVCOL_PRI_EBU3213 = 22, ///< EBU Tech. 3213-E (nothing there) / one of JEDEC P22 group phosphors + AVCOL_PRI_JEDEC_P22 = AVCOL_PRI_EBU3213, + AVCOL_PRI_NB ///< Not part of ABI +}; + +/** + * Color Transfer Characteristic. + * These values match the ones defined by ISO/IEC 23091-2_2019 subclause 8.2. + */ +enum AVColorTransferCharacteristic { + AVCOL_TRC_RESERVED0 = 0, + AVCOL_TRC_BT709 = 1, ///< also ITU-R BT1361 + AVCOL_TRC_UNSPECIFIED = 2, + AVCOL_TRC_RESERVED = 3, + AVCOL_TRC_GAMMA22 = 4, ///< also ITU-R BT470M / ITU-R BT1700 625 PAL & SECAM + AVCOL_TRC_GAMMA28 = 5, ///< also ITU-R BT470BG + AVCOL_TRC_SMPTE170M = 6, ///< also ITU-R BT601-6 525 or 625 / ITU-R BT1358 525 or 625 / ITU-R BT1700 NTSC + AVCOL_TRC_SMPTE240M = 7, + AVCOL_TRC_LINEAR = 8, ///< "Linear transfer characteristics" + AVCOL_TRC_LOG = 9, ///< "Logarithmic transfer characteristic (100:1 range)" + AVCOL_TRC_LOG_SQRT = 10, ///< "Logarithmic transfer characteristic (100 * Sqrt(10) : 1 range)" + AVCOL_TRC_IEC61966_2_4 = 11, ///< IEC 61966-2-4 + AVCOL_TRC_BT1361_ECG = 12, ///< ITU-R BT1361 Extended Colour Gamut + AVCOL_TRC_IEC61966_2_1 = 13, ///< IEC 61966-2-1 (sRGB or sYCC) + AVCOL_TRC_BT2020_10 = 14, ///< ITU-R BT2020 for 10-bit system + AVCOL_TRC_BT2020_12 = 15, ///< ITU-R BT2020 for 12-bit system + AVCOL_TRC_SMPTE2084 = 16, ///< SMPTE ST 2084 for 10-, 12-, 14- and 16-bit systems + AVCOL_TRC_SMPTEST2084 = AVCOL_TRC_SMPTE2084, + AVCOL_TRC_SMPTE428 = 17, ///< SMPTE ST 428-1 + AVCOL_TRC_SMPTEST428_1 = AVCOL_TRC_SMPTE428, + AVCOL_TRC_ARIB_STD_B67 = 18, ///< ARIB STD-B67, known as "Hybrid log-gamma" + AVCOL_TRC_NB ///< Not part of ABI +}; + +/** + * YUV colorspace type. + * These values match the ones defined by ISO/IEC 23091-2_2019 subclause 8.3. + */ +enum AVColorSpace { + AVCOL_SPC_RGB = 0, ///< order of coefficients is actually GBR, also IEC 61966-2-1 (sRGB), YZX and ST 428-1 + AVCOL_SPC_BT709 = 1, ///< also ITU-R BT1361 / IEC 61966-2-4 xvYCC709 / derived in SMPTE RP 177 Annex B + AVCOL_SPC_UNSPECIFIED = 2, + AVCOL_SPC_RESERVED = 3, ///< reserved for future use by ITU-T and ISO/IEC just like 15-255 are + AVCOL_SPC_FCC = 4, ///< FCC Title 47 Code of Federal Regulations 73.682 (a)(20) + AVCOL_SPC_BT470BG = 5, ///< also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM / IEC 61966-2-4 xvYCC601 + AVCOL_SPC_SMPTE170M = 6, ///< also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC / functionally identical to above + AVCOL_SPC_SMPTE240M = 7, ///< derived from 170M primaries and D65 white point, 170M is derived from BT470 System M's primaries + AVCOL_SPC_YCGCO = 8, ///< used by Dirac / VC-2 and H.264 FRext, see ITU-T SG16 + AVCOL_SPC_YCOCG = AVCOL_SPC_YCGCO, + AVCOL_SPC_BT2020_NCL = 9, ///< ITU-R BT2020 non-constant luminance system + AVCOL_SPC_BT2020_CL = 10, ///< ITU-R BT2020 constant luminance system + AVCOL_SPC_SMPTE2085 = 11, ///< SMPTE 2085, Y'D'zD'x + AVCOL_SPC_CHROMA_DERIVED_NCL = 12, ///< Chromaticity-derived non-constant luminance system + AVCOL_SPC_CHROMA_DERIVED_CL = 13, ///< Chromaticity-derived constant luminance system + AVCOL_SPC_ICTCP = 14, ///< ITU-R BT.2100-0, ICtCp + AVCOL_SPC_NB ///< Not part of ABI +}; + +/** + * Visual content value range. + * + * These values are based on definitions that can be found in multiple + * specifications, such as ITU-T BT.709 (3.4 - Quantization of RGB, luminance + * and colour-difference signals), ITU-T BT.2020 (Table 5 - Digital + * Representation) as well as ITU-T BT.2100 (Table 9 - Digital 10- and 12-bit + * integer representation). At the time of writing, the BT.2100 one is + * recommended, as it also defines the full range representation. + * + * Common definitions: + * - For RGB and luma planes such as Y in YCbCr and I in ICtCp, + * 'E' is the original value in range of 0.0 to 1.0. + * - For chroma planes such as Cb,Cr and Ct,Cp, 'E' is the original + * value in range of -0.5 to 0.5. + * - 'n' is the output bit depth. + * - For additional definitions such as rounding and clipping to valid n + * bit unsigned integer range, please refer to BT.2100 (Table 9). + */ +enum AVColorRange { + AVCOL_RANGE_UNSPECIFIED = 0, + + /** + * Narrow or limited range content. + * + * - For luma planes: + * + * (219 * E + 16) * 2^(n-8) + * + * F.ex. the range of 16-235 for 8 bits + * + * - For chroma planes: + * + * (224 * E + 128) * 2^(n-8) + * + * F.ex. the range of 16-240 for 8 bits + */ + AVCOL_RANGE_MPEG = 1, + + /** + * Full range content. + * + * - For RGB and luma planes: + * + * (2^n - 1) * E + * + * F.ex. the range of 0-255 for 8 bits + * + * - For chroma planes: + * + * (2^n - 1) * E + 2^(n - 1) + * + * F.ex. the range of 1-255 for 8 bits + */ + AVCOL_RANGE_JPEG = 2, + AVCOL_RANGE_NB ///< Not part of ABI +}; + +/** + * Location of chroma samples. + * + * Illustration showing the location of the first (top left) chroma sample of the + * image, the left shows only luma, the right + * shows the location of the chroma sample, the 2 could be imagined to overlay + * each other but are drawn separately due to limitations of ASCII + * + * 1st 2nd 1st 2nd horizontal luma sample positions + * v v v v + * ______ ______ + *1st luma line > |X X ... |3 4 X ... X are luma samples, + * | |1 2 1-6 are possible chroma positions + *2nd luma line > |X X ... |5 6 X ... 0 is undefined/unknown position + */ +enum AVChromaLocation { + AVCHROMA_LOC_UNSPECIFIED = 0, + AVCHROMA_LOC_LEFT = 1, ///< MPEG-2/4 4:2:0, H.264 default for 4:2:0 + AVCHROMA_LOC_CENTER = 2, ///< MPEG-1 4:2:0, JPEG 4:2:0, H.263 4:2:0 + AVCHROMA_LOC_TOPLEFT = 3, ///< ITU-R 601, SMPTE 274M 296M S314M(DV 4:1:1), mpeg2 4:2:2 + AVCHROMA_LOC_TOP = 4, + AVCHROMA_LOC_BOTTOMLEFT = 5, + AVCHROMA_LOC_BOTTOM = 6, + AVCHROMA_LOC_NB ///< Not part of ABI +}; + +#endif /* AVUTIL_PIXFMT_H */ diff --git a/media/ffvpx/libavutil/qsort.h b/media/ffvpx/libavutil/qsort.h new file mode 100644 index 0000000000..6014f88be3 --- /dev/null +++ b/media/ffvpx/libavutil/qsort.h @@ -0,0 +1,122 @@ +/* + * copyright (c) 2012 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_QSORT_H +#define AVUTIL_QSORT_H + +#include "macros.h" + + +/** + * Quicksort + * This sort is fast, and fully inplace but not stable and it is possible + * to construct input that requires O(n^2) time but this is very unlikely to + * happen with non constructed input. + */ +#define AV_QSORT(p, num, type, cmp) do {\ + void *stack[64][2];\ + int sp= 1;\ + stack[0][0] = p;\ + stack[0][1] = (p)+(num)-1;\ + while(sp){\ + type *start= stack[--sp][0];\ + type *end = stack[ sp][1];\ + while(start < end){\ + if(start < end-1) {\ + int checksort=0;\ + type *right = end-2;\ + type *left = start+1;\ + type *mid = start + ((end-start)>>1);\ + if(cmp(start, end) > 0) {\ + if(cmp( end, mid) > 0) FFSWAP(type, *start, *mid);\ + else FFSWAP(type, *start, *end);\ + }else{\ + if(cmp(start, mid) > 0) FFSWAP(type, *start, *mid);\ + else checksort= 1;\ + }\ + if(cmp(mid, end) > 0){ \ + FFSWAP(type, *mid, *end);\ + checksort=0;\ + }\ + if(start == end-2) break;\ + FFSWAP(type, end[-1], *mid);\ + while(left <= right){\ + while(left<=right && cmp(left, end-1) < 0)\ + left++;\ + while(left<=right && cmp(right, end-1) > 0)\ + right--;\ + if(left <= right){\ + FFSWAP(type, *left, *right);\ + left++;\ + right--;\ + }\ + }\ + FFSWAP(type, end[-1], *left);\ + if(checksort && (mid == left-1 || mid == left)){\ + mid= start;\ + while(mid<end && cmp(mid, mid+1) <= 0)\ + mid++;\ + if(mid==end)\ + break;\ + }\ + if(end-left < left-start){\ + stack[sp ][0]= start;\ + stack[sp++][1]= right;\ + start = left+1;\ + }else{\ + stack[sp ][0]= left+1;\ + stack[sp++][1]= end;\ + end = right;\ + }\ + }else{\ + if(cmp(start, end) > 0)\ + FFSWAP(type, *start, *end);\ + break;\ + }\ + }\ + }\ +} while (0) + +/** + * Merge sort, this sort requires a temporary buffer and is stable, its worst + * case time is O(n log n) + * @param p must be a lvalue pointer, this function may exchange it with tmp + * @param tmp must be a lvalue pointer, this function may exchange it with p + */ +#define AV_MSORT(p, tmp, num, type, cmp) do {\ + unsigned i, j, step;\ + for(step=1; step<(num); step+=step){\ + for(i=0; i<(num); i+=2*step){\ + unsigned a[2] = {i, i+step};\ + unsigned end = FFMIN(i+2*step, (num));\ + for(j=i; a[0]<i+step && a[1]<end; j++){\ + int idx= cmp(p+a[0], p+a[1]) > 0;\ + tmp[j] = p[ a[idx]++ ];\ + }\ + if(a[0]>=i+step) a[0] = a[1];\ + for(; j<end; j++){\ + tmp[j] = p[ a[0]++ ];\ + }\ + }\ + FFSWAP(type*, p, tmp);\ + }\ +} while (0) + +#endif /* AVUTIL_QSORT_H */ diff --git a/media/ffvpx/libavutil/rational.c b/media/ffvpx/libavutil/rational.c new file mode 100644 index 0000000000..eb148ddb12 --- /dev/null +++ b/media/ffvpx/libavutil/rational.c @@ -0,0 +1,193 @@ +/* + * rational numbers + * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * rational numbers + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#include "avassert.h" +#include <limits.h> + +#include "common.h" +#include "mathematics.h" +#include "rational.h" + +int av_reduce(int *dst_num, int *dst_den, + int64_t num, int64_t den, int64_t max) +{ + AVRational a0 = { 0, 1 }, a1 = { 1, 0 }; + int sign = (num < 0) ^ (den < 0); + int64_t gcd = av_gcd(FFABS(num), FFABS(den)); + + if (gcd) { + num = FFABS(num) / gcd; + den = FFABS(den) / gcd; + } + if (num <= max && den <= max) { + a1 = (AVRational) { num, den }; + den = 0; + } + + while (den) { + uint64_t x = num / den; + int64_t next_den = num - den * x; + int64_t a2n = x * a1.num + a0.num; + int64_t a2d = x * a1.den + a0.den; + + if (a2n > max || a2d > max) { + if (a1.num) x = (max - a0.num) / a1.num; + if (a1.den) x = FFMIN(x, (max - a0.den) / a1.den); + + if (den * (2 * x * a1.den + a0.den) > num * a1.den) + a1 = (AVRational) { x * a1.num + a0.num, x * a1.den + a0.den }; + break; + } + + a0 = a1; + a1 = (AVRational) { a2n, a2d }; + num = den; + den = next_den; + } + av_assert2(av_gcd(a1.num, a1.den) <= 1U); + av_assert2(a1.num <= max && a1.den <= max); + + *dst_num = sign ? -a1.num : a1.num; + *dst_den = a1.den; + + return den == 0; +} + +AVRational av_mul_q(AVRational b, AVRational c) +{ + av_reduce(&b.num, &b.den, + b.num * (int64_t) c.num, + b.den * (int64_t) c.den, INT_MAX); + return b; +} + +AVRational av_div_q(AVRational b, AVRational c) +{ + return av_mul_q(b, (AVRational) { c.den, c.num }); +} + +AVRational av_add_q(AVRational b, AVRational c) { + av_reduce(&b.num, &b.den, + b.num * (int64_t) c.den + + c.num * (int64_t) b.den, + b.den * (int64_t) c.den, INT_MAX); + return b; +} + +AVRational av_sub_q(AVRational b, AVRational c) +{ + return av_add_q(b, (AVRational) { -c.num, c.den }); +} + +AVRational av_d2q(double d, int max) +{ + AVRational a; + int exponent; + int64_t den; + if (isnan(d)) + return (AVRational) { 0,0 }; + if (fabs(d) > INT_MAX + 3LL) + return (AVRational) { d < 0 ? -1 : 1, 0 }; + frexp(d, &exponent); + exponent = FFMAX(exponent-1, 0); + den = 1LL << (61 - exponent); + // (int64_t)rint() and llrint() do not work with gcc on ia64 and sparc64, + // see Ticket2713 for affected gcc/glibc versions + av_reduce(&a.num, &a.den, floor(d * den + 0.5), den, max); + if ((!a.num || !a.den) && d && max>0 && max<INT_MAX) + av_reduce(&a.num, &a.den, floor(d * den + 0.5), den, INT_MAX); + + return a; +} + +int av_nearer_q(AVRational q, AVRational q1, AVRational q2) +{ + /* n/d is q, a/b is the median between q1 and q2 */ + int64_t a = q1.num * (int64_t)q2.den + q2.num * (int64_t)q1.den; + int64_t b = 2 * (int64_t)q1.den * q2.den; + + /* rnd_up(a*d/b) > n => a*d/b > n */ + int64_t x_up = av_rescale_rnd(a, q.den, b, AV_ROUND_UP); + + /* rnd_down(a*d/b) < n => a*d/b < n */ + int64_t x_down = av_rescale_rnd(a, q.den, b, AV_ROUND_DOWN); + + return ((x_up > q.num) - (x_down < q.num)) * av_cmp_q(q2, q1); +} + +int av_find_nearest_q_idx(AVRational q, const AVRational* q_list) +{ + int i, nearest_q_idx = 0; + for (i = 0; q_list[i].den; i++) + if (av_nearer_q(q, q_list[i], q_list[nearest_q_idx]) > 0) + nearest_q_idx = i; + + return nearest_q_idx; +} + +uint32_t av_q2intfloat(AVRational q) { + int64_t n; + int shift; + int sign = 0; + + if (q.den < 0) { + q.den *= -1; + q.num *= -1; + } + if (q.num < 0) { + q.num *= -1; + sign = 1; + } + + if (!q.num && !q.den) return 0xFFC00000; + if (!q.num) return 0; + if (!q.den) return 0x7F800000 | (q.num & 0x80000000); + + shift = 23 + av_log2(q.den) - av_log2(q.num); + if (shift >= 0) n = av_rescale(q.num, 1LL<<shift, q.den); + else n = av_rescale(q.num, 1, ((int64_t)q.den) << -shift); + + shift -= n >= (1<<24); + shift += n < (1<<23); + + if (shift >= 0) n = av_rescale(q.num, 1LL<<shift, q.den); + else n = av_rescale(q.num, 1, ((int64_t)q.den) << -shift); + + av_assert1(n < (1<<24)); + av_assert1(n >= (1<<23)); + + return sign<<31 | (150-shift)<<23 | (n - (1<<23)); +} + +AVRational av_gcd_q(AVRational a, AVRational b, int max_den, AVRational def) +{ + int64_t gcd, lcm; + + gcd = av_gcd(a.den, b.den); + lcm = (a.den / gcd) * b.den; + return lcm < max_den ? av_make_q(av_gcd(a.num, b.num), lcm) : def; +} diff --git a/media/ffvpx/libavutil/rational.h b/media/ffvpx/libavutil/rational.h new file mode 100644 index 0000000000..8cbfc8e066 --- /dev/null +++ b/media/ffvpx/libavutil/rational.h @@ -0,0 +1,221 @@ +/* + * rational numbers + * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu_math_rational + * Utilties for rational number calculation. + * @author Michael Niedermayer <michaelni@gmx.at> + */ + +#ifndef AVUTIL_RATIONAL_H +#define AVUTIL_RATIONAL_H + +#include <stdint.h> +#include <limits.h> +#include "attributes.h" + +/** + * @defgroup lavu_math_rational AVRational + * @ingroup lavu_math + * Rational number calculation. + * + * While rational numbers can be expressed as floating-point numbers, the + * conversion process is a lossy one, so are floating-point operations. On the + * other hand, the nature of FFmpeg demands highly accurate calculation of + * timestamps. This set of rational number utilities serves as a generic + * interface for manipulating rational numbers as pairs of numerators and + * denominators. + * + * Many of the functions that operate on AVRational's have the suffix `_q`, in + * reference to the mathematical symbol "ℚ" (Q) which denotes the set of all + * rational numbers. + * + * @{ + */ + +/** + * Rational number (pair of numerator and denominator). + */ +typedef struct AVRational{ + int num; ///< Numerator + int den; ///< Denominator +} AVRational; + +/** + * Create an AVRational. + * + * Useful for compilers that do not support compound literals. + * + * @note The return value is not reduced. + * @see av_reduce() + */ +static inline AVRational av_make_q(int num, int den) +{ + AVRational r = { num, den }; + return r; +} + +/** + * Compare two rationals. + * + * @param a First rational + * @param b Second rational + * + * @return One of the following values: + * - 0 if `a == b` + * - 1 if `a > b` + * - -1 if `a < b` + * - `INT_MIN` if one of the values is of the form `0 / 0` + */ +static inline int av_cmp_q(AVRational a, AVRational b){ + const int64_t tmp= a.num * (int64_t)b.den - b.num * (int64_t)a.den; + + if(tmp) return (int)((tmp ^ a.den ^ b.den)>>63)|1; + else if(b.den && a.den) return 0; + else if(a.num && b.num) return (a.num>>31) - (b.num>>31); + else return INT_MIN; +} + +/** + * Convert an AVRational to a `double`. + * @param a AVRational to convert + * @return `a` in floating-point form + * @see av_d2q() + */ +static inline double av_q2d(AVRational a){ + return a.num / (double) a.den; +} + +/** + * Reduce a fraction. + * + * This is useful for framerate calculations. + * + * @param[out] dst_num Destination numerator + * @param[out] dst_den Destination denominator + * @param[in] num Source numerator + * @param[in] den Source denominator + * @param[in] max Maximum allowed values for `dst_num` & `dst_den` + * @return 1 if the operation is exact, 0 otherwise + */ +int av_reduce(int *dst_num, int *dst_den, int64_t num, int64_t den, int64_t max); + +/** + * Multiply two rationals. + * @param b First rational + * @param c Second rational + * @return b*c + */ +AVRational av_mul_q(AVRational b, AVRational c) av_const; + +/** + * Divide one rational by another. + * @param b First rational + * @param c Second rational + * @return b/c + */ +AVRational av_div_q(AVRational b, AVRational c) av_const; + +/** + * Add two rationals. + * @param b First rational + * @param c Second rational + * @return b+c + */ +AVRational av_add_q(AVRational b, AVRational c) av_const; + +/** + * Subtract one rational from another. + * @param b First rational + * @param c Second rational + * @return b-c + */ +AVRational av_sub_q(AVRational b, AVRational c) av_const; + +/** + * Invert a rational. + * @param q value + * @return 1 / q + */ +static av_always_inline AVRational av_inv_q(AVRational q) +{ + AVRational r = { q.den, q.num }; + return r; +} + +/** + * Convert a double precision floating point number to a rational. + * + * In case of infinity, the returned value is expressed as `{1, 0}` or + * `{-1, 0}` depending on the sign. + * + * @param d `double` to convert + * @param max Maximum allowed numerator and denominator + * @return `d` in AVRational form + * @see av_q2d() + */ +AVRational av_d2q(double d, int max) av_const; + +/** + * Find which of the two rationals is closer to another rational. + * + * @param q Rational to be compared against + * @param q1 Rational to be tested + * @param q2 Rational to be tested + * @return One of the following values: + * - 1 if `q1` is nearer to `q` than `q2` + * - -1 if `q2` is nearer to `q` than `q1` + * - 0 if they have the same distance + */ +int av_nearer_q(AVRational q, AVRational q1, AVRational q2); + +/** + * Find the value in a list of rationals nearest a given reference rational. + * + * @param q Reference rational + * @param q_list Array of rationals terminated by `{0, 0}` + * @return Index of the nearest value found in the array + */ +int av_find_nearest_q_idx(AVRational q, const AVRational* q_list); + +/** + * Convert an AVRational to a IEEE 32-bit `float` expressed in fixed-point + * format. + * + * @param q Rational to be converted + * @return Equivalent floating-point value, expressed as an unsigned 32-bit + * integer. + * @note The returned value is platform-indepedant. + */ +uint32_t av_q2intfloat(AVRational q); + +/** + * Return the best rational so that a and b are multiple of it. + * If the resulting denominator is larger than max_den, return def. + */ +AVRational av_gcd_q(AVRational a, AVRational b, int max_den, AVRational def); + +/** + * @} + */ + +#endif /* AVUTIL_RATIONAL_H */ diff --git a/media/ffvpx/libavutil/replaygain.h b/media/ffvpx/libavutil/replaygain.h new file mode 100644 index 0000000000..b49bf1a3d9 --- /dev/null +++ b/media/ffvpx/libavutil/replaygain.h @@ -0,0 +1,50 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_REPLAYGAIN_H +#define AVUTIL_REPLAYGAIN_H + +#include <stdint.h> + +/** + * ReplayGain information (see + * http://wiki.hydrogenaudio.org/index.php?title=ReplayGain_1.0_specification). + * The size of this struct is a part of the public ABI. + */ +typedef struct AVReplayGain { + /** + * Track replay gain in microbels (divide by 100000 to get the value in dB). + * Should be set to INT32_MIN when unknown. + */ + int32_t track_gain; + /** + * Peak track amplitude, with 100000 representing full scale (but values + * may overflow). 0 when unknown. + */ + uint32_t track_peak; + /** + * Same as track_gain, but for the whole album. + */ + int32_t album_gain; + /** + * Same as track_peak, but for the whole album, + */ + uint32_t album_peak; +} AVReplayGain; + +#endif /* AVUTIL_REPLAYGAIN_H */ diff --git a/media/ffvpx/libavutil/reverse.c b/media/ffvpx/libavutil/reverse.c new file mode 100644 index 0000000000..105eb03dda --- /dev/null +++ b/media/ffvpx/libavutil/reverse.c @@ -0,0 +1,40 @@ +/* + * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdint.h> + +const uint8_t ff_reverse[256] = { +0x00,0x80,0x40,0xC0,0x20,0xA0,0x60,0xE0,0x10,0x90,0x50,0xD0,0x30,0xB0,0x70,0xF0, +0x08,0x88,0x48,0xC8,0x28,0xA8,0x68,0xE8,0x18,0x98,0x58,0xD8,0x38,0xB8,0x78,0xF8, +0x04,0x84,0x44,0xC4,0x24,0xA4,0x64,0xE4,0x14,0x94,0x54,0xD4,0x34,0xB4,0x74,0xF4, +0x0C,0x8C,0x4C,0xCC,0x2C,0xAC,0x6C,0xEC,0x1C,0x9C,0x5C,0xDC,0x3C,0xBC,0x7C,0xFC, +0x02,0x82,0x42,0xC2,0x22,0xA2,0x62,0xE2,0x12,0x92,0x52,0xD2,0x32,0xB2,0x72,0xF2, +0x0A,0x8A,0x4A,0xCA,0x2A,0xAA,0x6A,0xEA,0x1A,0x9A,0x5A,0xDA,0x3A,0xBA,0x7A,0xFA, +0x06,0x86,0x46,0xC6,0x26,0xA6,0x66,0xE6,0x16,0x96,0x56,0xD6,0x36,0xB6,0x76,0xF6, +0x0E,0x8E,0x4E,0xCE,0x2E,0xAE,0x6E,0xEE,0x1E,0x9E,0x5E,0xDE,0x3E,0xBE,0x7E,0xFE, +0x01,0x81,0x41,0xC1,0x21,0xA1,0x61,0xE1,0x11,0x91,0x51,0xD1,0x31,0xB1,0x71,0xF1, +0x09,0x89,0x49,0xC9,0x29,0xA9,0x69,0xE9,0x19,0x99,0x59,0xD9,0x39,0xB9,0x79,0xF9, +0x05,0x85,0x45,0xC5,0x25,0xA5,0x65,0xE5,0x15,0x95,0x55,0xD5,0x35,0xB5,0x75,0xF5, +0x0D,0x8D,0x4D,0xCD,0x2D,0xAD,0x6D,0xED,0x1D,0x9D,0x5D,0xDD,0x3D,0xBD,0x7D,0xFD, +0x03,0x83,0x43,0xC3,0x23,0xA3,0x63,0xE3,0x13,0x93,0x53,0xD3,0x33,0xB3,0x73,0xF3, +0x0B,0x8B,0x4B,0xCB,0x2B,0xAB,0x6B,0xEB,0x1B,0x9B,0x5B,0xDB,0x3B,0xBB,0x7B,0xFB, +0x07,0x87,0x47,0xC7,0x27,0xA7,0x67,0xE7,0x17,0x97,0x57,0xD7,0x37,0xB7,0x77,0xF7, +0x0F,0x8F,0x4F,0xCF,0x2F,0xAF,0x6F,0xEF,0x1F,0x9F,0x5F,0xDF,0x3F,0xBF,0x7F,0xFF, +}; diff --git a/media/ffvpx/libavutil/reverse.h b/media/ffvpx/libavutil/reverse.h new file mode 100644 index 0000000000..4eb6123932 --- /dev/null +++ b/media/ffvpx/libavutil/reverse.h @@ -0,0 +1,28 @@ +/* + * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_REVERSE_H +#define AVUTIL_REVERSE_H + +#include <stdint.h> + +extern const uint8_t ff_reverse[256]; + +#endif /* AVUTIL_REVERSE_H */ diff --git a/media/ffvpx/libavutil/samplefmt.c b/media/ffvpx/libavutil/samplefmt.c new file mode 100644 index 0000000000..e1be5f0547 --- /dev/null +++ b/media/ffvpx/libavutil/samplefmt.c @@ -0,0 +1,263 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "error.h" +#include "macros.h" +#include "mem.h" +#include "samplefmt.h" + +#include <limits.h> +#include <stdio.h> +#include <string.h> + +typedef struct SampleFmtInfo { + char name[8]; + int bits; + int planar; + enum AVSampleFormat altform; ///< planar<->packed alternative form +} SampleFmtInfo; + +/** this table gives more information about formats */ +static const SampleFmtInfo sample_fmt_info[AV_SAMPLE_FMT_NB] = { + [AV_SAMPLE_FMT_U8] = { .name = "u8", .bits = 8, .planar = 0, .altform = AV_SAMPLE_FMT_U8P }, + [AV_SAMPLE_FMT_S16] = { .name = "s16", .bits = 16, .planar = 0, .altform = AV_SAMPLE_FMT_S16P }, + [AV_SAMPLE_FMT_S32] = { .name = "s32", .bits = 32, .planar = 0, .altform = AV_SAMPLE_FMT_S32P }, + [AV_SAMPLE_FMT_S64] = { .name = "s64", .bits = 64, .planar = 0, .altform = AV_SAMPLE_FMT_S64P }, + [AV_SAMPLE_FMT_FLT] = { .name = "flt", .bits = 32, .planar = 0, .altform = AV_SAMPLE_FMT_FLTP }, + [AV_SAMPLE_FMT_DBL] = { .name = "dbl", .bits = 64, .planar = 0, .altform = AV_SAMPLE_FMT_DBLP }, + [AV_SAMPLE_FMT_U8P] = { .name = "u8p", .bits = 8, .planar = 1, .altform = AV_SAMPLE_FMT_U8 }, + [AV_SAMPLE_FMT_S16P] = { .name = "s16p", .bits = 16, .planar = 1, .altform = AV_SAMPLE_FMT_S16 }, + [AV_SAMPLE_FMT_S32P] = { .name = "s32p", .bits = 32, .planar = 1, .altform = AV_SAMPLE_FMT_S32 }, + [AV_SAMPLE_FMT_S64P] = { .name = "s64p", .bits = 64, .planar = 1, .altform = AV_SAMPLE_FMT_S64 }, + [AV_SAMPLE_FMT_FLTP] = { .name = "fltp", .bits = 32, .planar = 1, .altform = AV_SAMPLE_FMT_FLT }, + [AV_SAMPLE_FMT_DBLP] = { .name = "dblp", .bits = 64, .planar = 1, .altform = AV_SAMPLE_FMT_DBL }, +}; + +const char *av_get_sample_fmt_name(enum AVSampleFormat sample_fmt) +{ + if (sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB) + return NULL; + return sample_fmt_info[sample_fmt].name; +} + +enum AVSampleFormat av_get_sample_fmt(const char *name) +{ + int i; + + for (i = 0; i < AV_SAMPLE_FMT_NB; i++) + if (!strcmp(sample_fmt_info[i].name, name)) + return i; + return AV_SAMPLE_FMT_NONE; +} + +enum AVSampleFormat av_get_alt_sample_fmt(enum AVSampleFormat sample_fmt, int planar) +{ + if (sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB) + return AV_SAMPLE_FMT_NONE; + if (sample_fmt_info[sample_fmt].planar == planar) + return sample_fmt; + return sample_fmt_info[sample_fmt].altform; +} + +enum AVSampleFormat av_get_packed_sample_fmt(enum AVSampleFormat sample_fmt) +{ + if (sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB) + return AV_SAMPLE_FMT_NONE; + if (sample_fmt_info[sample_fmt].planar) + return sample_fmt_info[sample_fmt].altform; + return sample_fmt; +} + +enum AVSampleFormat av_get_planar_sample_fmt(enum AVSampleFormat sample_fmt) +{ + if (sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB) + return AV_SAMPLE_FMT_NONE; + if (sample_fmt_info[sample_fmt].planar) + return sample_fmt; + return sample_fmt_info[sample_fmt].altform; +} + +char *av_get_sample_fmt_string (char *buf, int buf_size, enum AVSampleFormat sample_fmt) +{ + /* print header */ + if (sample_fmt < 0) + snprintf(buf, buf_size, "name " " depth"); + else if (sample_fmt < AV_SAMPLE_FMT_NB) { + SampleFmtInfo info = sample_fmt_info[sample_fmt]; + snprintf (buf, buf_size, "%-6s" " %2d ", info.name, info.bits); + } + + return buf; +} + +int av_get_bytes_per_sample(enum AVSampleFormat sample_fmt) +{ + return sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB ? + 0 : sample_fmt_info[sample_fmt].bits >> 3; +} + +int av_sample_fmt_is_planar(enum AVSampleFormat sample_fmt) +{ + if (sample_fmt < 0 || sample_fmt >= AV_SAMPLE_FMT_NB) + return 0; + return sample_fmt_info[sample_fmt].planar; +} + +int av_samples_get_buffer_size(int *linesize, int nb_channels, int nb_samples, + enum AVSampleFormat sample_fmt, int align) +{ + int line_size; + int sample_size = av_get_bytes_per_sample(sample_fmt); + int planar = av_sample_fmt_is_planar(sample_fmt); + + /* validate parameter ranges */ + if (!sample_size || nb_samples <= 0 || nb_channels <= 0) + return AVERROR(EINVAL); + + /* auto-select alignment if not specified */ + if (!align) { + if (nb_samples > INT_MAX - 31) + return AVERROR(EINVAL); + align = 1; + nb_samples = FFALIGN(nb_samples, 32); + } + + /* check for integer overflow */ + if (nb_channels > INT_MAX / align || + (int64_t)nb_channels * nb_samples > (INT_MAX - (align * nb_channels)) / sample_size) + return AVERROR(EINVAL); + + line_size = planar ? FFALIGN(nb_samples * sample_size, align) : + FFALIGN(nb_samples * sample_size * nb_channels, align); + if (linesize) + *linesize = line_size; + + return planar ? line_size * nb_channels : line_size; +} + +int av_samples_fill_arrays(uint8_t **audio_data, int *linesize, + const uint8_t *buf, int nb_channels, int nb_samples, + enum AVSampleFormat sample_fmt, int align) +{ + int ch, planar, buf_size, line_size; + + planar = av_sample_fmt_is_planar(sample_fmt); + buf_size = av_samples_get_buffer_size(&line_size, nb_channels, nb_samples, + sample_fmt, align); + if (buf_size < 0) + return buf_size; + + if (linesize) + *linesize = line_size; + + memset(audio_data, 0, planar + ? sizeof(*audio_data) * nb_channels + : sizeof(*audio_data)); + + if (!buf) + return buf_size; + + audio_data[0] = (uint8_t *)buf; + for (ch = 1; planar && ch < nb_channels; ch++) + audio_data[ch] = audio_data[ch-1] + line_size; + + return buf_size; +} + +int av_samples_alloc(uint8_t **audio_data, int *linesize, int nb_channels, + int nb_samples, enum AVSampleFormat sample_fmt, int align) +{ + uint8_t *buf; + int size = av_samples_get_buffer_size(NULL, nb_channels, nb_samples, + sample_fmt, align); + if (size < 0) + return size; + + buf = av_malloc(size); + if (!buf) + return AVERROR(ENOMEM); + + size = av_samples_fill_arrays(audio_data, linesize, buf, nb_channels, + nb_samples, sample_fmt, align); + if (size < 0) { + av_free(buf); + return size; + } + + av_samples_set_silence(audio_data, 0, nb_samples, nb_channels, sample_fmt); + + return size; +} + +int av_samples_alloc_array_and_samples(uint8_t ***audio_data, int *linesize, int nb_channels, + int nb_samples, enum AVSampleFormat sample_fmt, int align) +{ + int ret, nb_planes = av_sample_fmt_is_planar(sample_fmt) ? nb_channels : 1; + + *audio_data = av_calloc(nb_planes, sizeof(**audio_data)); + if (!*audio_data) + return AVERROR(ENOMEM); + ret = av_samples_alloc(*audio_data, linesize, nb_channels, + nb_samples, sample_fmt, align); + if (ret < 0) + av_freep(audio_data); + return ret; +} + +int av_samples_copy(uint8_t * const *dst, uint8_t * const *src, int dst_offset, + int src_offset, int nb_samples, int nb_channels, + enum AVSampleFormat sample_fmt) +{ + int planar = av_sample_fmt_is_planar(sample_fmt); + int planes = planar ? nb_channels : 1; + int block_align = av_get_bytes_per_sample(sample_fmt) * (planar ? 1 : nb_channels); + int data_size = nb_samples * block_align; + int i; + + dst_offset *= block_align; + src_offset *= block_align; + + if((dst[0] < src[0] ? src[0] - dst[0] : dst[0] - src[0]) >= data_size) { + for (i = 0; i < planes; i++) + memcpy(dst[i] + dst_offset, src[i] + src_offset, data_size); + } else { + for (i = 0; i < planes; i++) + memmove(dst[i] + dst_offset, src[i] + src_offset, data_size); + } + + return 0; +} + +int av_samples_set_silence(uint8_t * const *audio_data, int offset, int nb_samples, + int nb_channels, enum AVSampleFormat sample_fmt) +{ + int planar = av_sample_fmt_is_planar(sample_fmt); + int planes = planar ? nb_channels : 1; + int block_align = av_get_bytes_per_sample(sample_fmt) * (planar ? 1 : nb_channels); + int data_size = nb_samples * block_align; + int fill_char = (sample_fmt == AV_SAMPLE_FMT_U8 || + sample_fmt == AV_SAMPLE_FMT_U8P) ? 0x80 : 0x00; + int i; + + offset *= block_align; + + for (i = 0; i < planes; i++) + memset(audio_data[i] + offset, fill_char, data_size); + + return 0; +} diff --git a/media/ffvpx/libavutil/samplefmt.h b/media/ffvpx/libavutil/samplefmt.h new file mode 100644 index 0000000000..43a57a422c --- /dev/null +++ b/media/ffvpx/libavutil/samplefmt.h @@ -0,0 +1,269 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_SAMPLEFMT_H +#define AVUTIL_SAMPLEFMT_H + +#include <stdint.h> + +/** + * @addtogroup lavu_audio + * @{ + * + * @defgroup lavu_sampfmts Audio sample formats + * + * Audio sample format enumeration and related convenience functions. + * @{ + */ + +/** + * Audio sample formats + * + * - The data described by the sample format is always in native-endian order. + * Sample values can be expressed by native C types, hence the lack of a signed + * 24-bit sample format even though it is a common raw audio data format. + * + * - The floating-point formats are based on full volume being in the range + * [-1.0, 1.0]. Any values outside this range are beyond full volume level. + * + * - The data layout as used in av_samples_fill_arrays() and elsewhere in FFmpeg + * (such as AVFrame in libavcodec) is as follows: + * + * @par + * For planar sample formats, each audio channel is in a separate data plane, + * and linesize is the buffer size, in bytes, for a single plane. All data + * planes must be the same size. For packed sample formats, only the first data + * plane is used, and samples for each channel are interleaved. In this case, + * linesize is the buffer size, in bytes, for the 1 plane. + * + */ +enum AVSampleFormat { + AV_SAMPLE_FMT_NONE = -1, + AV_SAMPLE_FMT_U8, ///< unsigned 8 bits + AV_SAMPLE_FMT_S16, ///< signed 16 bits + AV_SAMPLE_FMT_S32, ///< signed 32 bits + AV_SAMPLE_FMT_FLT, ///< float + AV_SAMPLE_FMT_DBL, ///< double + + AV_SAMPLE_FMT_U8P, ///< unsigned 8 bits, planar + AV_SAMPLE_FMT_S16P, ///< signed 16 bits, planar + AV_SAMPLE_FMT_S32P, ///< signed 32 bits, planar + AV_SAMPLE_FMT_FLTP, ///< float, planar + AV_SAMPLE_FMT_DBLP, ///< double, planar + AV_SAMPLE_FMT_S64, ///< signed 64 bits + AV_SAMPLE_FMT_S64P, ///< signed 64 bits, planar + + AV_SAMPLE_FMT_NB ///< Number of sample formats. DO NOT USE if linking dynamically +}; + +/** + * Return the name of sample_fmt, or NULL if sample_fmt is not + * recognized. + */ +const char *av_get_sample_fmt_name(enum AVSampleFormat sample_fmt); + +/** + * Return a sample format corresponding to name, or AV_SAMPLE_FMT_NONE + * on error. + */ +enum AVSampleFormat av_get_sample_fmt(const char *name); + +/** + * Return the planar<->packed alternative form of the given sample format, or + * AV_SAMPLE_FMT_NONE on error. If the passed sample_fmt is already in the + * requested planar/packed format, the format returned is the same as the + * input. + */ +enum AVSampleFormat av_get_alt_sample_fmt(enum AVSampleFormat sample_fmt, int planar); + +/** + * Get the packed alternative form of the given sample format. + * + * If the passed sample_fmt is already in packed format, the format returned is + * the same as the input. + * + * @return the packed alternative form of the given sample format or + AV_SAMPLE_FMT_NONE on error. + */ +enum AVSampleFormat av_get_packed_sample_fmt(enum AVSampleFormat sample_fmt); + +/** + * Get the planar alternative form of the given sample format. + * + * If the passed sample_fmt is already in planar format, the format returned is + * the same as the input. + * + * @return the planar alternative form of the given sample format or + AV_SAMPLE_FMT_NONE on error. + */ +enum AVSampleFormat av_get_planar_sample_fmt(enum AVSampleFormat sample_fmt); + +/** + * Generate a string corresponding to the sample format with + * sample_fmt, or a header if sample_fmt is negative. + * + * @param buf the buffer where to write the string + * @param buf_size the size of buf + * @param sample_fmt the number of the sample format to print the + * corresponding info string, or a negative value to print the + * corresponding header. + * @return the pointer to the filled buffer or NULL if sample_fmt is + * unknown or in case of other errors + */ +char *av_get_sample_fmt_string(char *buf, int buf_size, enum AVSampleFormat sample_fmt); + +/** + * Return number of bytes per sample. + * + * @param sample_fmt the sample format + * @return number of bytes per sample or zero if unknown for the given + * sample format + */ +int av_get_bytes_per_sample(enum AVSampleFormat sample_fmt); + +/** + * Check if the sample format is planar. + * + * @param sample_fmt the sample format to inspect + * @return 1 if the sample format is planar, 0 if it is interleaved + */ +int av_sample_fmt_is_planar(enum AVSampleFormat sample_fmt); + +/** + * Get the required buffer size for the given audio parameters. + * + * @param[out] linesize calculated linesize, may be NULL + * @param nb_channels the number of channels + * @param nb_samples the number of samples in a single channel + * @param sample_fmt the sample format + * @param align buffer size alignment (0 = default, 1 = no alignment) + * @return required buffer size, or negative error code on failure + */ +int av_samples_get_buffer_size(int *linesize, int nb_channels, int nb_samples, + enum AVSampleFormat sample_fmt, int align); + +/** + * @} + * + * @defgroup lavu_sampmanip Samples manipulation + * + * Functions that manipulate audio samples + * @{ + */ + +/** + * Fill plane data pointers and linesize for samples with sample + * format sample_fmt. + * + * The audio_data array is filled with the pointers to the samples data planes: + * for planar, set the start point of each channel's data within the buffer, + * for packed, set the start point of the entire buffer only. + * + * The value pointed to by linesize is set to the aligned size of each + * channel's data buffer for planar layout, or to the aligned size of the + * buffer for all channels for packed layout. + * + * The buffer in buf must be big enough to contain all the samples + * (use av_samples_get_buffer_size() to compute its minimum size), + * otherwise the audio_data pointers will point to invalid data. + * + * @see enum AVSampleFormat + * The documentation for AVSampleFormat describes the data layout. + * + * @param[out] audio_data array to be filled with the pointer for each channel + * @param[out] linesize calculated linesize, may be NULL + * @param buf the pointer to a buffer containing the samples + * @param nb_channels the number of channels + * @param nb_samples the number of samples in a single channel + * @param sample_fmt the sample format + * @param align buffer size alignment (0 = default, 1 = no alignment) + * @return minimum size in bytes required for the buffer on success, + * or a negative error code on failure + */ +int av_samples_fill_arrays(uint8_t **audio_data, int *linesize, + const uint8_t *buf, + int nb_channels, int nb_samples, + enum AVSampleFormat sample_fmt, int align); + +/** + * Allocate a samples buffer for nb_samples samples, and fill data pointers and + * linesize accordingly. + * The allocated samples buffer can be freed by using av_freep(&audio_data[0]) + * Allocated data will be initialized to silence. + * + * @see enum AVSampleFormat + * The documentation for AVSampleFormat describes the data layout. + * + * @param[out] audio_data array to be filled with the pointer for each channel + * @param[out] linesize aligned size for audio buffer(s), may be NULL + * @param nb_channels number of audio channels + * @param nb_samples number of samples per channel + * @param sample_fmt the sample format + * @param align buffer size alignment (0 = default, 1 = no alignment) + * @return >=0 on success or a negative error code on failure + * @todo return the size of the allocated buffer in case of success at the next bump + * @see av_samples_fill_arrays() + * @see av_samples_alloc_array_and_samples() + */ +int av_samples_alloc(uint8_t **audio_data, int *linesize, int nb_channels, + int nb_samples, enum AVSampleFormat sample_fmt, int align); + +/** + * Allocate a data pointers array, samples buffer for nb_samples + * samples, and fill data pointers and linesize accordingly. + * + * This is the same as av_samples_alloc(), but also allocates the data + * pointers array. + * + * @see av_samples_alloc() + */ +int av_samples_alloc_array_and_samples(uint8_t ***audio_data, int *linesize, int nb_channels, + int nb_samples, enum AVSampleFormat sample_fmt, int align); + +/** + * Copy samples from src to dst. + * + * @param dst destination array of pointers to data planes + * @param src source array of pointers to data planes + * @param dst_offset offset in samples at which the data will be written to dst + * @param src_offset offset in samples at which the data will be read from src + * @param nb_samples number of samples to be copied + * @param nb_channels number of audio channels + * @param sample_fmt audio sample format + */ +int av_samples_copy(uint8_t * const *dst, uint8_t * const *src, int dst_offset, + int src_offset, int nb_samples, int nb_channels, + enum AVSampleFormat sample_fmt); + +/** + * Fill an audio buffer with silence. + * + * @param audio_data array of pointers to data planes + * @param offset offset in samples at which to start filling + * @param nb_samples number of samples to fill + * @param nb_channels number of audio channels + * @param sample_fmt audio sample format + */ +int av_samples_set_silence(uint8_t * const *audio_data, int offset, int nb_samples, + int nb_channels, enum AVSampleFormat sample_fmt); + +/** + * @} + * @} + */ +#endif /* AVUTIL_SAMPLEFMT_H */ diff --git a/media/ffvpx/libavutil/sfc64.h b/media/ffvpx/libavutil/sfc64.h new file mode 100644 index 0000000000..a1593a0a7b --- /dev/null +++ b/media/ffvpx/libavutil/sfc64.h @@ -0,0 +1,84 @@ +/* + * Copyright (c) 2024 Michael Niedermayer <michael-ffmpeg@niedermayer.cc> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + * + */ + +/** + * @file + * simple Pseudo Random Number Generator + * + * This is a implementation of SFC64, a 64-bit PRNG by Chris Doty-Humphrey. + * + * This Generator is much faster (0m1.872s) than 64bit KISS (0m3.823s) and PCG-XSH-RR-64/32 (0m2.700s) + * And passes testu01 and practrand test suits. + */ + +#ifndef AVUTIL_SFC64_H +#define AVUTIL_SFC64_H + +#include <inttypes.h> + +typedef struct FFSFC64 { + uint64_t a,b,c,counter; +} FFSFC64; + +static inline uint64_t ff_sfc64_get(FFSFC64 *s) { + uint64_t tmp = s->a + s->b + s->counter++; + s->a = s->b ^ (s->b >> 11); + s->b = s->c + (s->c << 3); // This is a multiply by 9 + s->c = (s->c << 24 | s->c >> 40) + tmp; + return tmp; +} + +/** + * Return the previous random value, and step the generator backward. + * + * It is safe to take values before the first, but such values can be highly + * correlated to the seeds. + */ +static inline uint64_t ff_sfc64_reverse_get(FFSFC64 *s) { + uint64_t prev_c = s->b * 0x8E38E38E38E38E39; + uint64_t tmp = s->c - (prev_c << 24 | prev_c >> 40); + s->b = s->a ^ (s->a >> 11); + s->b ^= s->b >> 22; + s->b ^= s->b >> 44; + + s->a = tmp - s->b - --s->counter; + s->c = prev_c; + + return tmp; +} + +/** + * Initialize sfc64 with up to 3 seeds. + * + * @param rounds number of rounds mixing up state during init. Generally 8-18, larger numbers will help with bad quality seeds. + * 12 is a good choice if all 3 seeds are equal + * + */ +static inline void ff_sfc64_init(FFSFC64 *s, uint64_t seeda, uint64_t seedb, uint64_t seedc, int rounds) { + s->a = seeda; + s->b = seedb; + s->c = seedc; + s->counter = 1; + while (rounds--) + ff_sfc64_get(s); +} + +#endif // AVUTIL_SFC64_H diff --git a/media/ffvpx/libavutil/slicethread.c b/media/ffvpx/libavutil/slicethread.c new file mode 100644 index 0000000000..115b099736 --- /dev/null +++ b/media/ffvpx/libavutil/slicethread.c @@ -0,0 +1,259 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdatomic.h> +#include "cpu.h" +#include "internal.h" +#include "slicethread.h" +#include "mem.h" +#include "thread.h" +#include "avassert.h" + +#define MAX_AUTO_THREADS 16 + +#if HAVE_PTHREADS || HAVE_W32THREADS || HAVE_OS2THREADS + +typedef struct WorkerContext { + AVSliceThread *ctx; + pthread_mutex_t mutex; + pthread_cond_t cond; + pthread_t thread; + int done; +} WorkerContext; + +struct AVSliceThread { + WorkerContext *workers; + int nb_threads; + int nb_active_threads; + int nb_jobs; + + atomic_uint first_job; + atomic_uint current_job; + pthread_mutex_t done_mutex; + pthread_cond_t done_cond; + int done; + int finished; + + void *priv; + void (*worker_func)(void *priv, int jobnr, int threadnr, int nb_jobs, int nb_threads); + void (*main_func)(void *priv); +}; + +static int run_jobs(AVSliceThread *ctx) +{ + unsigned nb_jobs = ctx->nb_jobs; + unsigned nb_active_threads = ctx->nb_active_threads; + unsigned first_job = atomic_fetch_add_explicit(&ctx->first_job, 1, memory_order_acq_rel); + unsigned current_job = first_job; + + do { + ctx->worker_func(ctx->priv, current_job, first_job, nb_jobs, nb_active_threads); + } while ((current_job = atomic_fetch_add_explicit(&ctx->current_job, 1, memory_order_acq_rel)) < nb_jobs); + + return current_job == nb_jobs + nb_active_threads - 1; +} + +static void *attribute_align_arg thread_worker(void *v) +{ + WorkerContext *w = v; + AVSliceThread *ctx = w->ctx; + + pthread_mutex_lock(&w->mutex); + pthread_cond_signal(&w->cond); + + while (1) { + w->done = 1; + while (w->done) + pthread_cond_wait(&w->cond, &w->mutex); + + if (ctx->finished) { + pthread_mutex_unlock(&w->mutex); + return NULL; + } + + if (run_jobs(ctx)) { + pthread_mutex_lock(&ctx->done_mutex); + ctx->done = 1; + pthread_cond_signal(&ctx->done_cond); + pthread_mutex_unlock(&ctx->done_mutex); + } + } +} + +int avpriv_slicethread_create(AVSliceThread **pctx, void *priv, + void (*worker_func)(void *priv, int jobnr, int threadnr, int nb_jobs, int nb_threads), + void (*main_func)(void *priv), + int nb_threads) +{ + AVSliceThread *ctx; + int nb_workers, i; + + av_assert0(nb_threads >= 0); + if (!nb_threads) { + int nb_cpus = av_cpu_count(); + if (nb_cpus > 1) + nb_threads = FFMIN(nb_cpus + 1, MAX_AUTO_THREADS); + else + nb_threads = 1; + } + + nb_workers = nb_threads; + if (!main_func) + nb_workers--; + + *pctx = ctx = av_mallocz(sizeof(*ctx)); + if (!ctx) + return AVERROR(ENOMEM); + + if (nb_workers && !(ctx->workers = av_calloc(nb_workers, sizeof(*ctx->workers)))) { + av_freep(pctx); + return AVERROR(ENOMEM); + } + + ctx->priv = priv; + ctx->worker_func = worker_func; + ctx->main_func = main_func; + ctx->nb_threads = nb_threads; + ctx->nb_active_threads = 0; + ctx->nb_jobs = 0; + ctx->finished = 0; + + atomic_init(&ctx->first_job, 0); + atomic_init(&ctx->current_job, 0); + pthread_mutex_init(&ctx->done_mutex, NULL); + pthread_cond_init(&ctx->done_cond, NULL); + ctx->done = 0; + + for (i = 0; i < nb_workers; i++) { + WorkerContext *w = &ctx->workers[i]; + int ret; + w->ctx = ctx; + pthread_mutex_init(&w->mutex, NULL); + pthread_cond_init(&w->cond, NULL); + pthread_mutex_lock(&w->mutex); + w->done = 0; + + if (ret = pthread_create(&w->thread, NULL, thread_worker, w)) { + ctx->nb_threads = main_func ? i : i + 1; + pthread_mutex_unlock(&w->mutex); + pthread_cond_destroy(&w->cond); + pthread_mutex_destroy(&w->mutex); + avpriv_slicethread_free(pctx); + return AVERROR(ret); + } + + while (!w->done) + pthread_cond_wait(&w->cond, &w->mutex); + pthread_mutex_unlock(&w->mutex); + } + + return nb_threads; +} + +void avpriv_slicethread_execute(AVSliceThread *ctx, int nb_jobs, int execute_main) +{ + int nb_workers, i, is_last = 0; + + av_assert0(nb_jobs > 0); + ctx->nb_jobs = nb_jobs; + ctx->nb_active_threads = FFMIN(nb_jobs, ctx->nb_threads); + atomic_store_explicit(&ctx->first_job, 0, memory_order_relaxed); + atomic_store_explicit(&ctx->current_job, ctx->nb_active_threads, memory_order_relaxed); + nb_workers = ctx->nb_active_threads; + if (!ctx->main_func || !execute_main) + nb_workers--; + + for (i = 0; i < nb_workers; i++) { + WorkerContext *w = &ctx->workers[i]; + pthread_mutex_lock(&w->mutex); + w->done = 0; + pthread_cond_signal(&w->cond); + pthread_mutex_unlock(&w->mutex); + } + + if (ctx->main_func && execute_main) + ctx->main_func(ctx->priv); + else + is_last = run_jobs(ctx); + + if (!is_last) { + pthread_mutex_lock(&ctx->done_mutex); + while (!ctx->done) + pthread_cond_wait(&ctx->done_cond, &ctx->done_mutex); + ctx->done = 0; + pthread_mutex_unlock(&ctx->done_mutex); + } +} + +void avpriv_slicethread_free(AVSliceThread **pctx) +{ + AVSliceThread *ctx; + int nb_workers, i; + + if (!pctx || !*pctx) + return; + + ctx = *pctx; + nb_workers = ctx->nb_threads; + if (!ctx->main_func) + nb_workers--; + + ctx->finished = 1; + for (i = 0; i < nb_workers; i++) { + WorkerContext *w = &ctx->workers[i]; + pthread_mutex_lock(&w->mutex); + w->done = 0; + pthread_cond_signal(&w->cond); + pthread_mutex_unlock(&w->mutex); + } + + for (i = 0; i < nb_workers; i++) { + WorkerContext *w = &ctx->workers[i]; + pthread_join(w->thread, NULL); + pthread_cond_destroy(&w->cond); + pthread_mutex_destroy(&w->mutex); + } + + pthread_cond_destroy(&ctx->done_cond); + pthread_mutex_destroy(&ctx->done_mutex); + av_freep(&ctx->workers); + av_freep(pctx); +} + +#else /* HAVE_PTHREADS || HAVE_W32THREADS || HAVE_OS32THREADS */ + +int avpriv_slicethread_create(AVSliceThread **pctx, void *priv, + void (*worker_func)(void *priv, int jobnr, int threadnr, int nb_jobs, int nb_threads), + void (*main_func)(void *priv), + int nb_threads) +{ + *pctx = NULL; + return AVERROR(ENOSYS); +} + +void avpriv_slicethread_execute(AVSliceThread *ctx, int nb_jobs, int execute_main) +{ + av_assert0(0); +} + +void avpriv_slicethread_free(AVSliceThread **pctx) +{ + av_assert0(!pctx || !*pctx); +} + +#endif /* HAVE_PTHREADS || HAVE_W32THREADS || HAVE_OS32THREADS */ diff --git a/media/ffvpx/libavutil/slicethread.h b/media/ffvpx/libavutil/slicethread.h new file mode 100644 index 0000000000..f6f6f302c4 --- /dev/null +++ b/media/ffvpx/libavutil/slicethread.h @@ -0,0 +1,52 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_SLICETHREAD_H +#define AVUTIL_SLICETHREAD_H + +typedef struct AVSliceThread AVSliceThread; + +/** + * Create slice threading context. + * @param pctx slice threading context returned here + * @param priv private pointer to be passed to callback function + * @param worker_func callback function to be executed + * @param main_func special callback function, called from main thread, may be NULL + * @param nb_threads number of threads, 0 for automatic, must be >= 0 + * @return return number of threads or negative AVERROR on failure + */ +int avpriv_slicethread_create(AVSliceThread **pctx, void *priv, + void (*worker_func)(void *priv, int jobnr, int threadnr, int nb_jobs, int nb_threads), + void (*main_func)(void *priv), + int nb_threads); + +/** + * Execute slice threading. + * @param ctx slice threading context + * @param nb_jobs number of jobs, must be > 0 + * @param execute_main also execute main_func + */ +void avpriv_slicethread_execute(AVSliceThread *ctx, int nb_jobs, int execute_main); + +/** + * Destroy slice threading context. + * @param pctx pointer to context + */ +void avpriv_slicethread_free(AVSliceThread **pctx); + +#endif diff --git a/media/ffvpx/libavutil/thread.h b/media/ffvpx/libavutil/thread.h new file mode 100644 index 0000000000..2ded498c89 --- /dev/null +++ b/media/ffvpx/libavutil/thread.h @@ -0,0 +1,223 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +// This header should only be used to simplify code where +// threading is optional, not as a generic threading abstraction. + +#ifndef AVUTIL_THREAD_H +#define AVUTIL_THREAD_H + +#include "config.h" + +#if HAVE_PRCTL +#include <sys/prctl.h> +#endif + +#include "error.h" + +#if HAVE_PTHREADS || HAVE_W32THREADS || HAVE_OS2THREADS + +#if HAVE_PTHREADS +#include <pthread.h> + +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1 + +#include <stdlib.h> + +#include "log.h" +#include "macros.h" + +#define ASSERT_PTHREAD_ABORT(func, ret) do { \ + char errbuf[AV_ERROR_MAX_STRING_SIZE] = ""; \ + av_log(NULL, AV_LOG_FATAL, AV_STRINGIFY(func) \ + " failed with error: %s\n", \ + av_make_error_string(errbuf, AV_ERROR_MAX_STRING_SIZE, \ + AVERROR(ret))); \ + abort(); \ +} while (0) + +#define ASSERT_PTHREAD_NORET(func, ...) do { \ + int ret = func(__VA_ARGS__); \ + if (ret) \ + ASSERT_PTHREAD_ABORT(func, ret); \ +} while (0) + +#define ASSERT_PTHREAD(func, ...) do { \ + ASSERT_PTHREAD_NORET(func, __VA_ARGS__); \ + return 0; \ +} while (0) + +static inline int strict_pthread_join(pthread_t thread, void **value_ptr) +{ + ASSERT_PTHREAD(pthread_join, thread, value_ptr); +} + +static inline int strict_pthread_mutex_init(pthread_mutex_t *mutex, const pthread_mutexattr_t *attr) +{ + if (attr) { + ASSERT_PTHREAD_NORET(pthread_mutex_init, mutex, attr); + } else { + pthread_mutexattr_t local_attr; + ASSERT_PTHREAD_NORET(pthread_mutexattr_init, &local_attr); + ASSERT_PTHREAD_NORET(pthread_mutexattr_settype, &local_attr, PTHREAD_MUTEX_ERRORCHECK); + ASSERT_PTHREAD_NORET(pthread_mutex_init, mutex, &local_attr); + ASSERT_PTHREAD_NORET(pthread_mutexattr_destroy, &local_attr); + } + return 0; +} + +static inline int strict_pthread_mutex_destroy(pthread_mutex_t *mutex) +{ + ASSERT_PTHREAD(pthread_mutex_destroy, mutex); +} + +static inline int strict_pthread_mutex_lock(pthread_mutex_t *mutex) +{ + ASSERT_PTHREAD(pthread_mutex_lock, mutex); +} + +static inline int strict_pthread_mutex_unlock(pthread_mutex_t *mutex) +{ + ASSERT_PTHREAD(pthread_mutex_unlock, mutex); +} + +static inline int strict_pthread_cond_init(pthread_cond_t *cond, const pthread_condattr_t *attr) +{ + ASSERT_PTHREAD(pthread_cond_init, cond, attr); +} + +static inline int strict_pthread_cond_destroy(pthread_cond_t *cond) +{ + ASSERT_PTHREAD(pthread_cond_destroy, cond); +} + +static inline int strict_pthread_cond_signal(pthread_cond_t *cond) +{ + ASSERT_PTHREAD(pthread_cond_signal, cond); +} + +static inline int strict_pthread_cond_broadcast(pthread_cond_t *cond) +{ + ASSERT_PTHREAD(pthread_cond_broadcast, cond); +} + +static inline int strict_pthread_cond_wait(pthread_cond_t *cond, pthread_mutex_t *mutex) +{ + ASSERT_PTHREAD(pthread_cond_wait, cond, mutex); +} + +static inline int strict_pthread_cond_timedwait(pthread_cond_t *cond, pthread_mutex_t *mutex, + const struct timespec *abstime) +{ + int ret = pthread_cond_timedwait(cond, mutex, abstime); + if (ret && ret != ETIMEDOUT) + ASSERT_PTHREAD_ABORT(pthread_cond_timedwait, ret); + return ret; +} + +static inline int strict_pthread_once(pthread_once_t *once_control, void (*init_routine)(void)) +{ + ASSERT_PTHREAD(pthread_once, once_control, init_routine); +} + +#define pthread_join strict_pthread_join +#define pthread_mutex_init strict_pthread_mutex_init +#define pthread_mutex_destroy strict_pthread_mutex_destroy +#define pthread_mutex_lock strict_pthread_mutex_lock +#define pthread_mutex_unlock strict_pthread_mutex_unlock +#define pthread_cond_init strict_pthread_cond_init +#define pthread_cond_destroy strict_pthread_cond_destroy +#define pthread_cond_signal strict_pthread_cond_signal +#define pthread_cond_broadcast strict_pthread_cond_broadcast +#define pthread_cond_wait strict_pthread_cond_wait +#define pthread_cond_timedwait strict_pthread_cond_timedwait +#define pthread_once strict_pthread_once +#endif + +#elif HAVE_OS2THREADS +#include "compat/os2threads.h" +#else +#include "compat/w32pthreads.h" +#endif + +#define AVMutex pthread_mutex_t +#define AV_MUTEX_INITIALIZER PTHREAD_MUTEX_INITIALIZER + +#define ff_mutex_init pthread_mutex_init +#define ff_mutex_lock pthread_mutex_lock +#define ff_mutex_unlock pthread_mutex_unlock +#define ff_mutex_destroy pthread_mutex_destroy + +#define AVCond pthread_cond_t + +#define ff_cond_init pthread_cond_init +#define ff_cond_destroy pthread_cond_destroy +#define ff_cond_signal pthread_cond_signal +#define ff_cond_broadcast pthread_cond_broadcast +#define ff_cond_wait pthread_cond_wait +#define ff_cond_timedwait pthread_cond_timedwait + +#define AVOnce pthread_once_t +#define AV_ONCE_INIT PTHREAD_ONCE_INIT + +#define ff_thread_once(control, routine) pthread_once(control, routine) + +#else + +#define AVMutex char +#define AV_MUTEX_INITIALIZER 0 + +static inline int ff_mutex_init(AVMutex *mutex, const void *attr){ return 0; } +static inline int ff_mutex_lock(AVMutex *mutex){ return 0; } +static inline int ff_mutex_unlock(AVMutex *mutex){ return 0; } +static inline int ff_mutex_destroy(AVMutex *mutex){ return 0; } + +#define AVCond char + +static inline int ff_cond_init(AVCond *cond, const void *attr){ return 0; } +static inline int ff_cond_destroy(AVCond *cond){ return 0; } +static inline int ff_cond_signal(AVCond *cond){ return 0; } +static inline int ff_cond_broadcast(AVCond *cond){ return 0; } +static inline int ff_cond_wait(AVCond *cond, AVMutex *mutex){ return 0; } +static inline int ff_cond_timedwait(AVCond *cond, AVMutex *mutex, + const void *abstime){ return 0; } + +#define AVOnce char +#define AV_ONCE_INIT 0 + +static inline int ff_thread_once(char *control, void (*routine)(void)) +{ + if (!*control) { + routine(); + *control = 1; + } + return 0; +} + +#endif + +static inline int ff_thread_setname(const char *name) +{ +#if HAVE_PRCTL + return AVERROR(prctl(PR_SET_NAME, name)); +#endif + + return AVERROR(ENOSYS); +} + +#endif /* AVUTIL_THREAD_H */ diff --git a/media/ffvpx/libavutil/threadmessage.c b/media/ffvpx/libavutil/threadmessage.c new file mode 100644 index 0000000000..20b3882033 --- /dev/null +++ b/media/ffvpx/libavutil/threadmessage.c @@ -0,0 +1,240 @@ +/* + * Copyright (c) 2014 Nicolas George + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public License + * as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public License + * along with FFmpeg; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <limits.h> +#include <stddef.h> + +#include "error.h" +#include "fifo.h" +#include "mem.h" +#include "threadmessage.h" +#include "thread.h" + +struct AVThreadMessageQueue { +#if HAVE_THREADS + AVFifo *fifo; + pthread_mutex_t lock; + pthread_cond_t cond_recv; + pthread_cond_t cond_send; + int err_send; + int err_recv; + unsigned elsize; + void (*free_func)(void *msg); +#else + int dummy; +#endif +}; + +int av_thread_message_queue_alloc(AVThreadMessageQueue **mq, + unsigned nelem, + unsigned elsize) +{ +#if HAVE_THREADS + AVThreadMessageQueue *rmq; + int ret = 0; + + if (nelem > INT_MAX / elsize) + return AVERROR(EINVAL); + if (!(rmq = av_mallocz(sizeof(*rmq)))) + return AVERROR(ENOMEM); + if ((ret = pthread_mutex_init(&rmq->lock, NULL))) { + av_free(rmq); + return AVERROR(ret); + } + if ((ret = pthread_cond_init(&rmq->cond_recv, NULL))) { + pthread_mutex_destroy(&rmq->lock); + av_free(rmq); + return AVERROR(ret); + } + if ((ret = pthread_cond_init(&rmq->cond_send, NULL))) { + pthread_cond_destroy(&rmq->cond_recv); + pthread_mutex_destroy(&rmq->lock); + av_free(rmq); + return AVERROR(ret); + } + if (!(rmq->fifo = av_fifo_alloc2(nelem, elsize, 0))) { + pthread_cond_destroy(&rmq->cond_send); + pthread_cond_destroy(&rmq->cond_recv); + pthread_mutex_destroy(&rmq->lock); + av_free(rmq); + return AVERROR(ENOMEM); + } + rmq->elsize = elsize; + *mq = rmq; + return 0; +#else + *mq = NULL; + return AVERROR(ENOSYS); +#endif /* HAVE_THREADS */ +} + +void av_thread_message_queue_set_free_func(AVThreadMessageQueue *mq, + void (*free_func)(void *msg)) +{ +#if HAVE_THREADS + mq->free_func = free_func; +#endif +} + +void av_thread_message_queue_free(AVThreadMessageQueue **mq) +{ +#if HAVE_THREADS + if (*mq) { + av_thread_message_flush(*mq); + av_fifo_freep2(&(*mq)->fifo); + pthread_cond_destroy(&(*mq)->cond_send); + pthread_cond_destroy(&(*mq)->cond_recv); + pthread_mutex_destroy(&(*mq)->lock); + av_freep(mq); + } +#endif +} + +int av_thread_message_queue_nb_elems(AVThreadMessageQueue *mq) +{ +#if HAVE_THREADS + int ret; + pthread_mutex_lock(&mq->lock); + ret = av_fifo_can_read(mq->fifo); + pthread_mutex_unlock(&mq->lock); + return ret; +#else + return AVERROR(ENOSYS); +#endif +} + +#if HAVE_THREADS + +static int av_thread_message_queue_send_locked(AVThreadMessageQueue *mq, + void *msg, + unsigned flags) +{ + while (!mq->err_send && !av_fifo_can_write(mq->fifo)) { + if ((flags & AV_THREAD_MESSAGE_NONBLOCK)) + return AVERROR(EAGAIN); + pthread_cond_wait(&mq->cond_send, &mq->lock); + } + if (mq->err_send) + return mq->err_send; + av_fifo_write(mq->fifo, msg, 1); + /* one message is sent, signal one receiver */ + pthread_cond_signal(&mq->cond_recv); + return 0; +} + +static int av_thread_message_queue_recv_locked(AVThreadMessageQueue *mq, + void *msg, + unsigned flags) +{ + while (!mq->err_recv && !av_fifo_can_read(mq->fifo)) { + if ((flags & AV_THREAD_MESSAGE_NONBLOCK)) + return AVERROR(EAGAIN); + pthread_cond_wait(&mq->cond_recv, &mq->lock); + } + if (!av_fifo_can_read(mq->fifo)) + return mq->err_recv; + av_fifo_read(mq->fifo, msg, 1); + /* one message space appeared, signal one sender */ + pthread_cond_signal(&mq->cond_send); + return 0; +} + +#endif /* HAVE_THREADS */ + +int av_thread_message_queue_send(AVThreadMessageQueue *mq, + void *msg, + unsigned flags) +{ +#if HAVE_THREADS + int ret; + + pthread_mutex_lock(&mq->lock); + ret = av_thread_message_queue_send_locked(mq, msg, flags); + pthread_mutex_unlock(&mq->lock); + return ret; +#else + return AVERROR(ENOSYS); +#endif /* HAVE_THREADS */ +} + +int av_thread_message_queue_recv(AVThreadMessageQueue *mq, + void *msg, + unsigned flags) +{ +#if HAVE_THREADS + int ret; + + pthread_mutex_lock(&mq->lock); + ret = av_thread_message_queue_recv_locked(mq, msg, flags); + pthread_mutex_unlock(&mq->lock); + return ret; +#else + return AVERROR(ENOSYS); +#endif /* HAVE_THREADS */ +} + +void av_thread_message_queue_set_err_send(AVThreadMessageQueue *mq, + int err) +{ +#if HAVE_THREADS + pthread_mutex_lock(&mq->lock); + mq->err_send = err; + pthread_cond_broadcast(&mq->cond_send); + pthread_mutex_unlock(&mq->lock); +#endif /* HAVE_THREADS */ +} + +void av_thread_message_queue_set_err_recv(AVThreadMessageQueue *mq, + int err) +{ +#if HAVE_THREADS + pthread_mutex_lock(&mq->lock); + mq->err_recv = err; + pthread_cond_broadcast(&mq->cond_recv); + pthread_mutex_unlock(&mq->lock); +#endif /* HAVE_THREADS */ +} + +#if HAVE_THREADS +static int free_func_wrap(void *arg, void *buf, size_t *nb_elems) +{ + AVThreadMessageQueue *mq = arg; + uint8_t *msg = buf; + for (size_t i = 0; i < *nb_elems; i++) + mq->free_func(msg + i * mq->elsize); + return 0; +} +#endif + +void av_thread_message_flush(AVThreadMessageQueue *mq) +{ +#if HAVE_THREADS + size_t used; + + pthread_mutex_lock(&mq->lock); + used = av_fifo_can_read(mq->fifo); + if (mq->free_func) + av_fifo_read_to_cb(mq->fifo, free_func_wrap, mq, &used); + /* only the senders need to be notified since the queue is empty and there + * is nothing to read */ + pthread_cond_broadcast(&mq->cond_send); + pthread_mutex_unlock(&mq->lock); +#endif /* HAVE_THREADS */ +} diff --git a/media/ffvpx/libavutil/threadmessage.h b/media/ffvpx/libavutil/threadmessage.h new file mode 100644 index 0000000000..42ce655f36 --- /dev/null +++ b/media/ffvpx/libavutil/threadmessage.h @@ -0,0 +1,115 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public License + * as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public License + * along with FFmpeg; if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_THREADMESSAGE_H +#define AVUTIL_THREADMESSAGE_H + +typedef struct AVThreadMessageQueue AVThreadMessageQueue; + +typedef enum AVThreadMessageFlags { + + /** + * Perform non-blocking operation. + * If this flag is set, send and recv operations are non-blocking and + * return AVERROR(EAGAIN) immediately if they can not proceed. + */ + AV_THREAD_MESSAGE_NONBLOCK = 1, + +} AVThreadMessageFlags; + +/** + * Allocate a new message queue. + * + * @param mq pointer to the message queue + * @param nelem maximum number of elements in the queue + * @param elsize size of each element in the queue + * @return >=0 for success; <0 for error, in particular AVERROR(ENOSYS) if + * lavu was built without thread support + */ +int av_thread_message_queue_alloc(AVThreadMessageQueue **mq, + unsigned nelem, + unsigned elsize); + +/** + * Free a message queue. + * + * The message queue must no longer be in use by another thread. + */ +void av_thread_message_queue_free(AVThreadMessageQueue **mq); + +/** + * Send a message on the queue. + */ +int av_thread_message_queue_send(AVThreadMessageQueue *mq, + void *msg, + unsigned flags); + +/** + * Receive a message from the queue. + */ +int av_thread_message_queue_recv(AVThreadMessageQueue *mq, + void *msg, + unsigned flags); + +/** + * Set the sending error code. + * + * If the error code is set to non-zero, av_thread_message_queue_send() will + * return it immediately. Conventional values, such as AVERROR_EOF or + * AVERROR(EAGAIN), can be used to cause the sending thread to stop or + * suspend its operation. + */ +void av_thread_message_queue_set_err_send(AVThreadMessageQueue *mq, + int err); + +/** + * Set the receiving error code. + * + * If the error code is set to non-zero, av_thread_message_queue_recv() will + * return it immediately when there are no longer available messages. + * Conventional values, such as AVERROR_EOF or AVERROR(EAGAIN), can be used + * to cause the receiving thread to stop or suspend its operation. + */ +void av_thread_message_queue_set_err_recv(AVThreadMessageQueue *mq, + int err); + +/** + * Set the optional free message callback function which will be called if an + * operation is removing messages from the queue. + */ +void av_thread_message_queue_set_free_func(AVThreadMessageQueue *mq, + void (*free_func)(void *msg)); + +/** + * Return the current number of messages in the queue. + * + * @return the current number of messages or AVERROR(ENOSYS) if lavu was built + * without thread support + */ +int av_thread_message_queue_nb_elems(AVThreadMessageQueue *mq); + +/** + * Flush the message queue + * + * This function is mostly equivalent to reading and free-ing every message + * except that it will be done in a single operation (no lock/unlock between + * reads). + */ +void av_thread_message_flush(AVThreadMessageQueue *mq); + +#endif /* AVUTIL_THREADMESSAGE_H */ diff --git a/media/ffvpx/libavutil/time.c b/media/ffvpx/libavutil/time.c new file mode 100644 index 0000000000..83743da0b2 --- /dev/null +++ b/media/ffvpx/libavutil/time.c @@ -0,0 +1,98 @@ +/* + * Copyright (c) 2000-2003 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include <stddef.h> +#include <stdint.h> +#include <time.h> +#if HAVE_GETTIMEOFDAY +#include <sys/time.h> +#endif +#if HAVE_UNISTD_H +#include <unistd.h> +#endif +#if HAVE_WINDOWS_H +#include <windows.h> +#endif + +#include "fftime.h" +#include "error.h" + +int64_t av_gettime(void) +{ +#if HAVE_GETTIMEOFDAY + struct timeval tv; + gettimeofday(&tv, NULL); + return (int64_t)tv.tv_sec * 1000000 + tv.tv_usec; +#elif HAVE_GETSYSTEMTIMEASFILETIME + FILETIME ft; + int64_t t; + GetSystemTimeAsFileTime(&ft); + t = (int64_t)ft.dwHighDateTime << 32 | ft.dwLowDateTime; + return t / 10 - 11644473600000000; /* Jan 1, 1601 */ +#else + return -1; +#endif +} + +int64_t av_gettime_relative(void) +{ +#if HAVE_CLOCK_GETTIME && defined(CLOCK_MONOTONIC) +#ifdef __APPLE__ + if (&clock_gettime) +#endif + { + struct timespec ts; + clock_gettime(CLOCK_MONOTONIC, &ts); + return (int64_t)ts.tv_sec * 1000000 + ts.tv_nsec / 1000; + } +#endif + return av_gettime() + 42 * 60 * 60 * INT64_C(1000000); +} + +int av_gettime_relative_is_monotonic(void) +{ +#if HAVE_CLOCK_GETTIME && defined(CLOCK_MONOTONIC) +#ifdef __APPLE__ + if (!&clock_gettime) + return 0; +#endif + return 1; +#else + return 0; +#endif +} + +int av_usleep(unsigned usec) +{ +#if HAVE_NANOSLEEP + struct timespec ts = { usec / 1000000, usec % 1000000 * 1000 }; + while (nanosleep(&ts, &ts) < 0 && errno == EINTR); + return 0; +#elif HAVE_USLEEP + return usleep(usec); +#elif HAVE_SLEEP + Sleep(usec / 1000); + return 0; +#else + return AVERROR(ENOSYS); +#endif +} diff --git a/media/ffvpx/libavutil/time_internal.h b/media/ffvpx/libavutil/time_internal.h new file mode 100644 index 0000000000..d0f007ab1c --- /dev/null +++ b/media/ffvpx/libavutil/time_internal.h @@ -0,0 +1,49 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_TIME_INTERNAL_H +#define AVUTIL_TIME_INTERNAL_H + +#include <time.h> +#include "config.h" + +#if !HAVE_GMTIME_R && !defined(gmtime_r) +static inline struct tm *ff_gmtime_r(const time_t* clock, struct tm *result) +{ + struct tm *ptr = gmtime(clock); + if (!ptr) + return NULL; + *result = *ptr; + return result; +} +#define gmtime_r ff_gmtime_r +#endif + +#if !HAVE_LOCALTIME_R && !defined(localtime_r) +static inline struct tm *ff_localtime_r(const time_t* clock, struct tm *result) +{ + struct tm *ptr = localtime(clock); + if (!ptr) + return NULL; + *result = *ptr; + return result; +} +#define localtime_r ff_localtime_r +#endif + +#endif /* AVUTIL_TIME_INTERNAL_H */ diff --git a/media/ffvpx/libavutil/timecode.c b/media/ffvpx/libavutil/timecode.c new file mode 100644 index 0000000000..b93f05b4b8 --- /dev/null +++ b/media/ffvpx/libavutil/timecode.c @@ -0,0 +1,265 @@ +/* + * Copyright (c) 2006 Smartjog S.A.S, Baptiste Coudurier <baptiste.coudurier@gmail.com> + * Copyright (c) 2011-2012 Smartjog S.A.S, Clément Bœsch <clement.boesch@smartjog.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Timecode helpers + * @see https://en.wikipedia.org/wiki/SMPTE_time_code + * @see http://www.dropframetimecode.org + */ + +#include <stdio.h> +#include "common.h" +#include "timecode.h" +#include "log.h" +#include "error.h" + +int av_timecode_adjust_ntsc_framenum2(int framenum, int fps) +{ + /* only works for multiples of NTSC 29.97 */ + int drop_frames = 0; + int d, m, frames_per_10mins; + + if (fps && fps % 30 == 0) { + drop_frames = fps / 30 * 2; + frames_per_10mins = fps / 30 * 17982; + } else + return framenum; + + d = framenum / frames_per_10mins; + m = framenum % frames_per_10mins; + + return framenum + 9U * drop_frames * d + drop_frames * ((m - drop_frames) / (frames_per_10mins / 10)); +} + +uint32_t av_timecode_get_smpte_from_framenum(const AVTimecode *tc, int framenum) +{ + unsigned fps = tc->fps; + int drop = !!(tc->flags & AV_TIMECODE_FLAG_DROPFRAME); + int hh, mm, ss, ff; + + framenum += tc->start; + if (drop) + framenum = av_timecode_adjust_ntsc_framenum2(framenum, tc->fps); + ff = framenum % fps; + ss = framenum / fps % 60; + mm = framenum / (fps*60) % 60; + hh = framenum / (fps*3600) % 24; + return av_timecode_get_smpte(tc->rate, drop, hh, mm, ss, ff); +} + +uint32_t av_timecode_get_smpte(AVRational rate, int drop, int hh, int mm, int ss, int ff) +{ + uint32_t tc = 0; + + /* For SMPTE 12-M timecodes, frame count is a special case if > 30 FPS. + See SMPTE ST 12-1:2014 Sec 12.1 for more info. */ + if (av_cmp_q(rate, (AVRational) {30, 1}) == 1) { + if (ff % 2 == 1) { + if (av_cmp_q(rate, (AVRational) {50, 1}) == 0) + tc |= (1 << 7); + else + tc |= (1 << 23); + } + ff /= 2; + } + + hh = hh % 24; + mm = av_clip(mm, 0, 59); + ss = av_clip(ss, 0, 59); + ff = ff % 40; + + tc |= drop << 30; + tc |= (ff / 10) << 28; + tc |= (ff % 10) << 24; + tc |= (ss / 10) << 20; + tc |= (ss % 10) << 16; + tc |= (mm / 10) << 12; + tc |= (mm % 10) << 8; + tc |= (hh / 10) << 4; + tc |= (hh % 10); + + return tc; +} + +char *av_timecode_make_string(const AVTimecode *tc, char *buf, int framenum) +{ + int fps = tc->fps; + int drop = tc->flags & AV_TIMECODE_FLAG_DROPFRAME; + int hh, mm, ss, ff, ff_len, neg = 0; + + framenum += tc->start; + if (drop) + framenum = av_timecode_adjust_ntsc_framenum2(framenum, fps); + if (framenum < 0) { + framenum = -framenum; + neg = tc->flags & AV_TIMECODE_FLAG_ALLOWNEGATIVE; + } + ff = framenum % fps; + ss = framenum / fps % 60; + mm = framenum / (fps*60LL) % 60; + hh = framenum / (fps*3600LL); + if (tc->flags & AV_TIMECODE_FLAG_24HOURSMAX) + hh = hh % 24; + ff_len = fps > 10000 ? 5 : fps > 1000 ? 4 : fps > 100 ? 3 : fps > 10 ? 2 : 1; + snprintf(buf, AV_TIMECODE_STR_SIZE, "%s%02d:%02d:%02d%c%0*d", + neg ? "-" : "", + hh, mm, ss, drop ? ';' : ':', ff_len, ff); + return buf; +} + +static unsigned bcd2uint(uint8_t bcd) +{ + unsigned low = bcd & 0xf; + unsigned high = bcd >> 4; + if (low > 9 || high > 9) + return 0; + return low + 10*high; +} + +char *av_timecode_make_smpte_tc_string2(char *buf, AVRational rate, uint32_t tcsmpte, int prevent_df, int skip_field) +{ + unsigned hh = bcd2uint(tcsmpte & 0x3f); // 6-bit hours + unsigned mm = bcd2uint(tcsmpte>>8 & 0x7f); // 7-bit minutes + unsigned ss = bcd2uint(tcsmpte>>16 & 0x7f); // 7-bit seconds + unsigned ff = bcd2uint(tcsmpte>>24 & 0x3f); // 6-bit frames + unsigned drop = tcsmpte & 1<<30 && !prevent_df; // 1-bit drop if not arbitrary bit + + if (av_cmp_q(rate, (AVRational) {30, 1}) == 1) { + ff <<= 1; + if (!skip_field) { + if (av_cmp_q(rate, (AVRational) {50, 1}) == 0) + ff += !!(tcsmpte & 1 << 7); + else + ff += !!(tcsmpte & 1 << 23); + } + } + + snprintf(buf, AV_TIMECODE_STR_SIZE, "%02u:%02u:%02u%c%02u", + hh, mm, ss, drop ? ';' : ':', ff); + return buf; + +} + +char *av_timecode_make_smpte_tc_string(char *buf, uint32_t tcsmpte, int prevent_df) +{ + return av_timecode_make_smpte_tc_string2(buf, (AVRational){30, 1}, tcsmpte, prevent_df, 1); +} + +char *av_timecode_make_mpeg_tc_string(char *buf, uint32_t tc25bit) +{ + snprintf(buf, AV_TIMECODE_STR_SIZE, + "%02"PRIu32":%02"PRIu32":%02"PRIu32"%c%02"PRIu32, + tc25bit>>19 & 0x1f, // 5-bit hours + tc25bit>>13 & 0x3f, // 6-bit minutes + tc25bit>>6 & 0x3f, // 6-bit seconds + tc25bit & 1<<24 ? ';' : ':', // 1-bit drop flag + tc25bit & 0x3f); // 6-bit frames + return buf; +} + +static int check_fps(int fps) +{ + int i; + static const int supported_fps[] = { + 24, 25, 30, 48, 50, 60, 100, 120, 150, + }; + + for (i = 0; i < FF_ARRAY_ELEMS(supported_fps); i++) + if (fps == supported_fps[i]) + return 0; + return -1; +} + +static int check_timecode(void *log_ctx, AVTimecode *tc) +{ + if ((int)tc->fps <= 0) { + av_log(log_ctx, AV_LOG_ERROR, "Valid timecode frame rate must be specified. Minimum value is 1\n"); + return AVERROR(EINVAL); + } + if ((tc->flags & AV_TIMECODE_FLAG_DROPFRAME) && tc->fps % 30 != 0) { + av_log(log_ctx, AV_LOG_ERROR, "Drop frame is only allowed with multiples of 30000/1001 FPS\n"); + return AVERROR(EINVAL); + } + if (check_fps(tc->fps) < 0) { + av_log(log_ctx, AV_LOG_WARNING, "Using non-standard frame rate %d/%d\n", + tc->rate.num, tc->rate.den); + } + return 0; +} + +static int fps_from_frame_rate(AVRational rate) +{ + if (!rate.den || !rate.num) + return -1; + return (rate.num + rate.den/2) / rate.den; +} + +int av_timecode_check_frame_rate(AVRational rate) +{ + return check_fps(fps_from_frame_rate(rate)); +} + +int av_timecode_init(AVTimecode *tc, AVRational rate, int flags, int frame_start, void *log_ctx) +{ + memset(tc, 0, sizeof(*tc)); + tc->start = frame_start; + tc->flags = flags; + tc->rate = rate; + tc->fps = fps_from_frame_rate(rate); + return check_timecode(log_ctx, tc); +} + +int av_timecode_init_from_components(AVTimecode *tc, AVRational rate, int flags, int hh, int mm, int ss, int ff, void *log_ctx) +{ + int ret; + + memset(tc, 0, sizeof(*tc)); + tc->flags = flags; + tc->rate = rate; + tc->fps = fps_from_frame_rate(rate); + + ret = check_timecode(log_ctx, tc); + if (ret < 0) + return ret; + + tc->start = (hh*3600 + mm*60 + ss) * tc->fps + ff; + if (tc->flags & AV_TIMECODE_FLAG_DROPFRAME) { /* adjust frame number */ + int tmins = 60*hh + mm; + tc->start -= (tc->fps / 30 * 2) * (tmins - tmins/10); + } + return 0; +} + +int av_timecode_init_from_string(AVTimecode *tc, AVRational rate, const char *str, void *log_ctx) +{ + char c; + int hh, mm, ss, ff, flags; + + if (sscanf(str, "%d:%d:%d%c%d", &hh, &mm, &ss, &c, &ff) != 5) { + av_log(log_ctx, AV_LOG_ERROR, "Unable to parse timecode, " + "syntax: hh:mm:ss[:;.]ff\n"); + return AVERROR_INVALIDDATA; + } + flags = c != ':' ? AV_TIMECODE_FLAG_DROPFRAME : 0; // drop if ';', '.', ... + + return av_timecode_init_from_components(tc, rate, flags, hh, mm, ss, ff, log_ctx); +} diff --git a/media/ffvpx/libavutil/timecode.h b/media/ffvpx/libavutil/timecode.h new file mode 100644 index 0000000000..fe0fc83576 --- /dev/null +++ b/media/ffvpx/libavutil/timecode.h @@ -0,0 +1,199 @@ +/* + * Copyright (c) 2006 Smartjog S.A.S, Baptiste Coudurier <baptiste.coudurier@gmail.com> + * Copyright (c) 2011-2012 Smartjog S.A.S, Clément Bœsch <clement.boesch@smartjog.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * Timecode helpers header + */ + +#ifndef AVUTIL_TIMECODE_H +#define AVUTIL_TIMECODE_H + +#include <stdint.h> +#include "rational.h" + +#define AV_TIMECODE_STR_SIZE 23 + +enum AVTimecodeFlag { + AV_TIMECODE_FLAG_DROPFRAME = 1<<0, ///< timecode is drop frame + AV_TIMECODE_FLAG_24HOURSMAX = 1<<1, ///< timecode wraps after 24 hours + AV_TIMECODE_FLAG_ALLOWNEGATIVE = 1<<2, ///< negative time values are allowed +}; + +typedef struct { + int start; ///< timecode frame start (first base frame number) + uint32_t flags; ///< flags such as drop frame, +24 hours support, ... + AVRational rate; ///< frame rate in rational form + unsigned fps; ///< frame per second; must be consistent with the rate field +} AVTimecode; + +/** + * Adjust frame number for NTSC drop frame time code. + * + * @param framenum frame number to adjust + * @param fps frame per second, multiples of 30 + * @return adjusted frame number + * @warning adjustment is only valid for multiples of NTSC 29.97 + */ +int av_timecode_adjust_ntsc_framenum2(int framenum, int fps); + +/** + * Convert frame number to SMPTE 12M binary representation. + * + * @param tc timecode data correctly initialized + * @param framenum frame number + * @return the SMPTE binary representation + * + * See SMPTE ST 314M-2005 Sec 4.4.2.2.1 "Time code pack (TC)" + * the format description as follows: + * bits 0-5: hours, in BCD(6bits) + * bits 6: BGF1 + * bits 7: BGF2 (NTSC) or FIELD (PAL) + * bits 8-14: minutes, in BCD(7bits) + * bits 15: BGF0 (NTSC) or BGF2 (PAL) + * bits 16-22: seconds, in BCD(7bits) + * bits 23: FIELD (NTSC) or BGF0 (PAL) + * bits 24-29: frames, in BCD(6bits) + * bits 30: drop frame flag (0: non drop, 1: drop) + * bits 31: color frame flag (0: unsync mode, 1: sync mode) + * @note BCD numbers (6 or 7 bits): 4 or 5 lower bits for units, 2 higher bits for tens. + * @note Frame number adjustment is automatically done in case of drop timecode, + * you do NOT have to call av_timecode_adjust_ntsc_framenum2(). + * @note The frame number is relative to tc->start. + * @note Color frame (CF) and binary group flags (BGF) bits are set to zero. + */ +uint32_t av_timecode_get_smpte_from_framenum(const AVTimecode *tc, int framenum); + +/** + * Convert sei info to SMPTE 12M binary representation. + * + * @param rate frame rate in rational form + * @param drop drop flag + * @param hh hour + * @param mm minute + * @param ss second + * @param ff frame number + * @return the SMPTE binary representation + */ +uint32_t av_timecode_get_smpte(AVRational rate, int drop, int hh, int mm, int ss, int ff); + +/** + * Load timecode string in buf. + * + * @param tc timecode data correctly initialized + * @param buf destination buffer, must be at least AV_TIMECODE_STR_SIZE long + * @param framenum frame number + * @return the buf parameter + * + * @note Timecode representation can be a negative timecode and have more than + * 24 hours, but will only be honored if the flags are correctly set. + * @note The frame number is relative to tc->start. + */ +char *av_timecode_make_string(const AVTimecode *tc, char *buf, int framenum); + +/** + * Get the timecode string from the SMPTE timecode format. + * + * In contrast to av_timecode_make_smpte_tc_string this function supports 50/60 + * fps timecodes by using the field bit. + * + * @param buf destination buffer, must be at least AV_TIMECODE_STR_SIZE long + * @param rate frame rate of the timecode + * @param tcsmpte the 32-bit SMPTE timecode + * @param prevent_df prevent the use of a drop flag when it is known the DF bit + * is arbitrary + * @param skip_field prevent the use of a field flag when it is known the field + * bit is arbitrary (e.g. because it is used as PC flag) + * @return the buf parameter + */ +char *av_timecode_make_smpte_tc_string2(char *buf, AVRational rate, uint32_t tcsmpte, int prevent_df, int skip_field); + +/** + * Get the timecode string from the SMPTE timecode format. + * + * @param buf destination buffer, must be at least AV_TIMECODE_STR_SIZE long + * @param tcsmpte the 32-bit SMPTE timecode + * @param prevent_df prevent the use of a drop flag when it is known the DF bit + * is arbitrary + * @return the buf parameter + */ +char *av_timecode_make_smpte_tc_string(char *buf, uint32_t tcsmpte, int prevent_df); + +/** + * Get the timecode string from the 25-bit timecode format (MPEG GOP format). + * + * @param buf destination buffer, must be at least AV_TIMECODE_STR_SIZE long + * @param tc25bit the 25-bits timecode + * @return the buf parameter + */ +char *av_timecode_make_mpeg_tc_string(char *buf, uint32_t tc25bit); + +/** + * Init a timecode struct with the passed parameters. + * + * @param tc pointer to an allocated AVTimecode + * @param rate frame rate in rational form + * @param flags miscellaneous flags such as drop frame, +24 hours, ... + * (see AVTimecodeFlag) + * @param frame_start the first frame number + * @param log_ctx a pointer to an arbitrary struct of which the first field + * is a pointer to an AVClass struct (used for av_log) + * @return 0 on success, AVERROR otherwise + */ +int av_timecode_init(AVTimecode *tc, AVRational rate, int flags, int frame_start, void *log_ctx); + +/** + * Init a timecode struct from the passed timecode components. + * + * @param tc pointer to an allocated AVTimecode + * @param rate frame rate in rational form + * @param flags miscellaneous flags such as drop frame, +24 hours, ... + * (see AVTimecodeFlag) + * @param hh hours + * @param mm minutes + * @param ss seconds + * @param ff frames + * @param log_ctx a pointer to an arbitrary struct of which the first field + * is a pointer to an AVClass struct (used for av_log) + * @return 0 on success, AVERROR otherwise + */ +int av_timecode_init_from_components(AVTimecode *tc, AVRational rate, int flags, int hh, int mm, int ss, int ff, void *log_ctx); + +/** + * Parse timecode representation (hh:mm:ss[:;.]ff). + * + * @param tc pointer to an allocated AVTimecode + * @param rate frame rate in rational form + * @param str timecode string which will determine the frame start + * @param log_ctx a pointer to an arbitrary struct of which the first field is a + * pointer to an AVClass struct (used for av_log). + * @return 0 on success, AVERROR otherwise + */ +int av_timecode_init_from_string(AVTimecode *tc, AVRational rate, const char *str, void *log_ctx); + +/** + * Check if the timecode feature is available for the given frame rate + * + * @return 0 if supported, <0 otherwise + */ +int av_timecode_check_frame_rate(AVRational rate); + +#endif /* AVUTIL_TIMECODE_H */ diff --git a/media/ffvpx/libavutil/timer.h b/media/ffvpx/libavutil/timer.h new file mode 100644 index 0000000000..2cd299eca3 --- /dev/null +++ b/media/ffvpx/libavutil/timer.h @@ -0,0 +1,158 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * high precision timer, useful to profile code + */ + +#ifndef AVUTIL_TIMER_H +#define AVUTIL_TIMER_H + +#include "config.h" + +#if CONFIG_LINUX_PERF +# ifndef _GNU_SOURCE +# define _GNU_SOURCE +# endif +# include <unistd.h> // read(3) +# include <sys/ioctl.h> +# include <asm/unistd.h> +# include <linux/perf_event.h> +#endif + +#include <stdlib.h> +#include <stdint.h> +#include <inttypes.h> + +#if CONFIG_MACOS_KPERF +#include "macos_kperf.h" +#elif HAVE_MACH_ABSOLUTE_TIME +#include <mach/mach_time.h> +#endif + +#include "common.h" +#include "log.h" + +#if ARCH_AARCH64 +# include "aarch64/timer.h" +#elif ARCH_ARM +# include "arm/timer.h" +#elif ARCH_PPC +# include "ppc/timer.h" +#elif ARCH_RISCV +# include "riscv/timer.h" +#elif ARCH_X86 +# include "x86/timer.h" +#elif ARCH_LOONGARCH +# include "loongarch/timer.h" +#endif + +#if !defined(AV_READ_TIME) +# if HAVE_GETHRTIME +# define AV_READ_TIME gethrtime +# elif HAVE_MACH_ABSOLUTE_TIME +# define AV_READ_TIME mach_absolute_time +# endif +#endif + +#ifndef FF_TIMER_UNITS +# define FF_TIMER_UNITS "UNITS" +#endif + +#define TIMER_REPORT(id, tdiff) \ + { \ + static uint64_t tsum = 0; \ + static int tcount = 0; \ + static int tskip_count = 0; \ + static int thistogram[32] = {0}; \ + thistogram[av_log2(tdiff)]++; \ + if (tcount < 2 || \ + (tdiff) < 8 * tsum / tcount || \ + (tdiff) < 2000) { \ + tsum += (tdiff); \ + tcount++; \ + } else \ + tskip_count++; \ + if (((tcount + tskip_count) & (tcount + tskip_count - 1)) == 0) { \ + int i; \ + av_log(NULL, AV_LOG_ERROR, \ + "%7" PRIu64 " " FF_TIMER_UNITS " in %s,%8d runs,%7d skips",\ + tsum * 10 / tcount, id, tcount, tskip_count); \ + for (i = 0; i < 32; i++) \ + av_log(NULL, AV_LOG_VERBOSE, " %2d", av_log2(2*thistogram[i]));\ + av_log(NULL, AV_LOG_ERROR, "\n"); \ + } \ + } + +#if CONFIG_LINUX_PERF + +#define START_TIMER \ + static int linux_perf_fd = -1; \ + uint64_t tperf; \ + if (linux_perf_fd == -1) { \ + struct perf_event_attr attr = { \ + .type = PERF_TYPE_HARDWARE, \ + .size = sizeof(struct perf_event_attr), \ + .config = PERF_COUNT_HW_CPU_CYCLES, \ + .disabled = 1, \ + .exclude_kernel = 1, \ + .exclude_hv = 1, \ + }; \ + linux_perf_fd = syscall(__NR_perf_event_open, &attr, \ + 0, -1, -1, 0); \ + } \ + if (linux_perf_fd == -1) { \ + av_log(NULL, AV_LOG_ERROR, "perf_event_open failed: %s\n", \ + av_err2str(AVERROR(errno))); \ + } else { \ + ioctl(linux_perf_fd, PERF_EVENT_IOC_RESET, 0); \ + ioctl(linux_perf_fd, PERF_EVENT_IOC_ENABLE, 0); \ + } + +#define STOP_TIMER(id) \ + ioctl(linux_perf_fd, PERF_EVENT_IOC_DISABLE, 0); \ + read(linux_perf_fd, &tperf, sizeof(tperf)); \ + TIMER_REPORT(id, tperf) + +#elif CONFIG_MACOS_KPERF + +#define START_TIMER \ + uint64_t tperf; \ + ff_kperf_init(); \ + tperf = ff_kperf_cycles(); + +#define STOP_TIMER(id) \ + TIMER_REPORT(id, ff_kperf_cycles() - tperf); + +#elif defined(AV_READ_TIME) +#define START_TIMER \ + uint64_t tend; \ + uint64_t tstart = AV_READ_TIME(); \ + +#define STOP_TIMER(id) \ + tend = AV_READ_TIME(); \ + TIMER_REPORT(id, tend - tstart) +#else +#define START_TIMER +#define STOP_TIMER(id) { } +#endif + +#endif /* AVUTIL_TIMER_H */ diff --git a/media/ffvpx/libavutil/timestamp.h b/media/ffvpx/libavutil/timestamp.h new file mode 100644 index 0000000000..9ae64da8a1 --- /dev/null +++ b/media/ffvpx/libavutil/timestamp.h @@ -0,0 +1,78 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * timestamp utils, mostly useful for debugging/logging purposes + */ + +#ifndef AVUTIL_TIMESTAMP_H +#define AVUTIL_TIMESTAMP_H + +#include "avutil.h" + +#if defined(__cplusplus) && !defined(__STDC_FORMAT_MACROS) && !defined(PRId64) +#error missing -D__STDC_FORMAT_MACROS / #define __STDC_FORMAT_MACROS +#endif + +#define AV_TS_MAX_STRING_SIZE 32 + +/** + * Fill the provided buffer with a string containing a timestamp + * representation. + * + * @param buf a buffer with size in bytes of at least AV_TS_MAX_STRING_SIZE + * @param ts the timestamp to represent + * @return the buffer in input + */ +static inline char *av_ts_make_string(char *buf, int64_t ts) +{ + if (ts == AV_NOPTS_VALUE) snprintf(buf, AV_TS_MAX_STRING_SIZE, "NOPTS"); + else snprintf(buf, AV_TS_MAX_STRING_SIZE, "%" PRId64, ts); + return buf; +} + +/** + * Convenience macro, the return value should be used only directly in + * function arguments but never stand-alone. + */ +#define av_ts2str(ts) av_ts_make_string((char[AV_TS_MAX_STRING_SIZE]){0}, ts) + +/** + * Fill the provided buffer with a string containing a timestamp time + * representation. + * + * @param buf a buffer with size in bytes of at least AV_TS_MAX_STRING_SIZE + * @param ts the timestamp to represent + * @param tb the timebase of the timestamp + * @return the buffer in input + */ +static inline char *av_ts_make_time_string(char *buf, int64_t ts, AVRational *tb) +{ + if (ts == AV_NOPTS_VALUE) snprintf(buf, AV_TS_MAX_STRING_SIZE, "NOPTS"); + else snprintf(buf, AV_TS_MAX_STRING_SIZE, "%.6g", av_q2d(*tb) * ts); + return buf; +} + +/** + * Convenience macro, the return value should be used only directly in + * function arguments but never stand-alone. + */ +#define av_ts2timestr(ts, tb) av_ts_make_time_string((char[AV_TS_MAX_STRING_SIZE]){0}, ts, tb) + +#endif /* AVUTIL_TIMESTAMP_H */ diff --git a/media/ffvpx/libavutil/tx.c b/media/ffvpx/libavutil/tx.c new file mode 100644 index 0000000000..a1f767039b --- /dev/null +++ b/media/ffvpx/libavutil/tx.c @@ -0,0 +1,935 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "avassert.h" +#include "intmath.h" +#include "cpu.h" +#include "qsort.h" +#include "bprint.h" + +#include "tx_priv.h" + +#define TYPE_IS(type, x) \ + (((x) == AV_TX_FLOAT_ ## type) || \ + ((x) == AV_TX_DOUBLE_ ## type) || \ + ((x) == AV_TX_INT32_ ## type)) + +/* Calculates the modular multiplicative inverse */ +static av_always_inline int mulinv(int n, int m) +{ + n = n % m; + for (int x = 1; x < m; x++) + if (((n * x) % m) == 1) + return x; + av_assert0(0); /* Never reached */ + return 0; +} + +int ff_tx_gen_pfa_input_map(AVTXContext *s, FFTXCodeletOptions *opts, + int d1, int d2) +{ + const int sl = d1*d2; + + s->map = av_malloc(s->len*sizeof(*s->map)); + if (!s->map) + return AVERROR(ENOMEM); + + for (int k = 0; k < s->len; k += sl) { + if (s->inv || (opts && opts->map_dir == FF_TX_MAP_SCATTER)) { + for (int m = 0; m < d2; m++) + for (int n = 0; n < d1; n++) + s->map[k + ((m*d1 + n*d2) % (sl))] = m*d1 + n; + } else { + for (int m = 0; m < d2; m++) + for (int n = 0; n < d1; n++) + s->map[k + m*d1 + n] = (m*d1 + n*d2) % (sl); + } + + if (s->inv) + for (int w = 1; w <= ((sl) >> 1); w++) + FFSWAP(int, s->map[k + w], s->map[k + sl - w]); + } + + s->map_dir = opts ? opts->map_dir : FF_TX_MAP_GATHER; + + return 0; +} + +/* Guaranteed to work for any n, m where gcd(n, m) == 1 */ +int ff_tx_gen_compound_mapping(AVTXContext *s, FFTXCodeletOptions *opts, + int inv, int n, int m) +{ + int *in_map, *out_map; + const int len = n*m; /* Will not be equal to s->len for MDCTs */ + int m_inv, n_inv; + + /* Make sure the numbers are coprime */ + if (av_gcd(n, m) != 1) + return AVERROR(EINVAL); + + m_inv = mulinv(m, n); + n_inv = mulinv(n, m); + + if (!(s->map = av_malloc(2*len*sizeof(*s->map)))) + return AVERROR(ENOMEM); + + in_map = s->map; + out_map = s->map + len; + + /* Ruritanian map for input, CRT map for output, can be swapped */ + if (opts && opts->map_dir == FF_TX_MAP_SCATTER) { + for (int j = 0; j < m; j++) { + for (int i = 0; i < n; i++) { + in_map[(i*m + j*n) % len] = j*n + i; + out_map[(i*m*m_inv + j*n*n_inv) % len] = i*m + j; + } + } + } else { + for (int j = 0; j < m; j++) { + for (int i = 0; i < n; i++) { + in_map[j*n + i] = (i*m + j*n) % len; + out_map[(i*m*m_inv + j*n*n_inv) % len] = i*m + j; + } + } + } + + if (inv) { + for (int i = 0; i < m; i++) { + int *in = &in_map[i*n + 1]; /* Skip the DC */ + for (int j = 0; j < ((n - 1) >> 1); j++) + FFSWAP(int, in[j], in[n - j - 2]); + } + } + + s->map_dir = opts ? opts->map_dir : FF_TX_MAP_GATHER; + + return 0; +} + +static inline int split_radix_permutation(int i, int len, int inv) +{ + len >>= 1; + if (len <= 1) + return i & 1; + if (!(i & len)) + return split_radix_permutation(i, len, inv) * 2; + len >>= 1; + return split_radix_permutation(i, len, inv) * 4 + 1 - 2*(!(i & len) ^ inv); +} + +int ff_tx_gen_ptwo_revtab(AVTXContext *s, FFTXCodeletOptions *opts) +{ + int len = s->len; + + if (!(s->map = av_malloc(len*sizeof(*s->map)))) + return AVERROR(ENOMEM); + + if (opts && opts->map_dir == FF_TX_MAP_SCATTER) { + for (int i = 0; i < s->len; i++) + s->map[-split_radix_permutation(i, len, s->inv) & (len - 1)] = i; + } else { + for (int i = 0; i < s->len; i++) + s->map[i] = -split_radix_permutation(i, len, s->inv) & (len - 1); + } + + s->map_dir = opts ? opts->map_dir : FF_TX_MAP_GATHER; + + return 0; +} + +int ff_tx_gen_inplace_map(AVTXContext *s, int len) +{ + int *src_map, out_map_idx = 0; + + if (!s->sub || !s->sub->map) + return AVERROR(EINVAL); + + if (!(s->map = av_mallocz(len*sizeof(*s->map)))) + return AVERROR(ENOMEM); + + src_map = s->sub->map; + + /* The first coefficient is always already in-place */ + for (int src = 1; src < s->len; src++) { + int dst = src_map[src]; + int found = 0; + + if (dst <= src) + continue; + + /* This just checks if a closed loop has been encountered before, + * and if so, skips it, since to fully permute a loop we must only + * enter it once. */ + do { + for (int j = 0; j < out_map_idx; j++) { + if (dst == s->map[j]) { + found = 1; + break; + } + } + dst = src_map[dst]; + } while (dst != src && !found); + + if (!found) + s->map[out_map_idx++] = src; + } + + s->map[out_map_idx++] = 0; + + return 0; +} + +static void parity_revtab_generator(int *revtab, int n, int inv, int offset, + int is_dual, int dual_high, int len, + int basis, int dual_stride, int inv_lookup) +{ + len >>= 1; + + if (len <= basis) { + int k1, k2, stride, even_idx, odd_idx; + + is_dual = is_dual && dual_stride; + dual_high = is_dual & dual_high; + stride = is_dual ? FFMIN(dual_stride, len) : 0; + + even_idx = offset + dual_high*(stride - 2*len); + odd_idx = even_idx + len + (is_dual && !dual_high)*len + dual_high*len; + + for (int i = 0; i < len; i++) { + k1 = -split_radix_permutation(offset + i*2 + 0, n, inv) & (n - 1); + k2 = -split_radix_permutation(offset + i*2 + 1, n, inv) & (n - 1); + if (inv_lookup) { + revtab[even_idx++] = k1; + revtab[odd_idx++] = k2; + } else { + revtab[k1] = even_idx++; + revtab[k2] = odd_idx++; + } + if (stride && !((i + 1) % stride)) { + even_idx += stride; + odd_idx += stride; + } + } + + return; + } + + parity_revtab_generator(revtab, n, inv, offset, + 0, 0, len >> 0, basis, dual_stride, inv_lookup); + parity_revtab_generator(revtab, n, inv, offset + (len >> 0), + 1, 0, len >> 1, basis, dual_stride, inv_lookup); + parity_revtab_generator(revtab, n, inv, offset + (len >> 0) + (len >> 1), + 1, 1, len >> 1, basis, dual_stride, inv_lookup); +} + +int ff_tx_gen_split_radix_parity_revtab(AVTXContext *s, int len, int inv, + FFTXCodeletOptions *opts, + int basis, int dual_stride) +{ + basis >>= 1; + if (len < basis) + return AVERROR(EINVAL); + + if (!(s->map = av_mallocz(len*sizeof(*s->map)))) + return AVERROR(ENOMEM); + + av_assert0(!dual_stride || !(dual_stride & (dual_stride - 1))); + av_assert0(dual_stride <= basis); + + parity_revtab_generator(s->map, len, inv, 0, 0, 0, len, + basis, dual_stride, + opts ? opts->map_dir == FF_TX_MAP_GATHER : FF_TX_MAP_GATHER); + + s->map_dir = opts ? opts->map_dir : FF_TX_MAP_GATHER; + + return 0; +} + +static void reset_ctx(AVTXContext *s, int free_sub) +{ + if (!s) + return; + + if (s->sub) + for (int i = 0; i < TX_MAX_SUB; i++) + reset_ctx(&s->sub[i], free_sub + 1); + + if (s->cd_self && s->cd_self->uninit) + s->cd_self->uninit(s); + + if (free_sub) + av_freep(&s->sub); + + av_freep(&s->map); + av_freep(&s->exp); + av_freep(&s->tmp); + + /* Nothing else needs to be reset, it gets overwritten if another + * ff_tx_init_subtx() call is made. */ + s->nb_sub = 0; + s->opaque = NULL; + memset(s->fn, 0, sizeof(*s->fn)); +} + +void ff_tx_clear_ctx(AVTXContext *s) +{ + reset_ctx(s, 0); +} + +av_cold void av_tx_uninit(AVTXContext **ctx) +{ + if (!(*ctx)) + return; + + reset_ctx(*ctx, 1); + av_freep(ctx); +} + +static av_cold int ff_tx_null_init(AVTXContext *s, const FFTXCodelet *cd, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale) +{ + /* Can only handle one sample+type to one sample+type transforms */ + if (TYPE_IS(MDCT, s->type) || TYPE_IS(RDFT, s->type)) + return AVERROR(EINVAL); + return 0; +} + +/* Null transform when the length is 1 */ +static void ff_tx_null(AVTXContext *s, void *_out, void *_in, ptrdiff_t stride) +{ + memcpy(_out, _in, stride); +} + +static const FFTXCodelet ff_tx_null_def = { + .name = NULL_IF_CONFIG_SMALL("null"), + .function = ff_tx_null, + .type = TX_TYPE_ANY, + .flags = AV_TX_UNALIGNED | FF_TX_ALIGNED | + FF_TX_OUT_OF_PLACE | AV_TX_INPLACE, + .factors[0] = TX_FACTOR_ANY, + .min_len = 1, + .max_len = 1, + .init = ff_tx_null_init, + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_MAX, +}; + +static const FFTXCodelet * const ff_tx_null_list[] = { + &ff_tx_null_def, + NULL, +}; + +/* Array of all compiled codelet lists. Order is irrelevant. */ +static const FFTXCodelet * const * const codelet_list[] = { + ff_tx_codelet_list_float_c, + ff_tx_codelet_list_double_c, + ff_tx_codelet_list_int32_c, + ff_tx_null_list, +#if HAVE_X86ASM + ff_tx_codelet_list_float_x86, +#endif +#if ARCH_AARCH64 + ff_tx_codelet_list_float_aarch64, +#endif +}; +static const int codelet_list_num = FF_ARRAY_ELEMS(codelet_list); + +static const int cpu_slow_mask = AV_CPU_FLAG_SSE2SLOW | AV_CPU_FLAG_SSE3SLOW | + AV_CPU_FLAG_ATOM | AV_CPU_FLAG_SSSE3SLOW | + AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER; + +static const int cpu_slow_penalties[][2] = { + { AV_CPU_FLAG_SSE2SLOW, 1 + 64 }, + { AV_CPU_FLAG_SSE3SLOW, 1 + 64 }, + { AV_CPU_FLAG_SSSE3SLOW, 1 + 64 }, + { AV_CPU_FLAG_ATOM, 1 + 128 }, + { AV_CPU_FLAG_AVXSLOW, 1 + 128 }, + { AV_CPU_FLAG_SLOW_GATHER, 1 + 32 }, +}; + +static int get_codelet_prio(const FFTXCodelet *cd, int cpu_flags, int len) +{ + int prio = cd->prio; + int max_factor = 0; + + /* If the CPU has a SLOW flag, and the instruction is also flagged + * as being slow for such, reduce its priority */ + for (int i = 0; i < FF_ARRAY_ELEMS(cpu_slow_penalties); i++) { + if ((cpu_flags & cd->cpu_flags) & cpu_slow_penalties[i][0]) + prio -= cpu_slow_penalties[i][1]; + } + + /* Prioritize aligned-only codelets */ + if ((cd->flags & FF_TX_ALIGNED) && !(cd->flags & AV_TX_UNALIGNED)) + prio += 64; + + /* Codelets for specific lengths are generally faster */ + if ((len == cd->min_len) && (len == cd->max_len)) + prio += 64; + + /* Forward-only or inverse-only transforms are generally better */ + if ((cd->flags & (FF_TX_FORWARD_ONLY | FF_TX_INVERSE_ONLY))) + prio += 64; + + /* Larger factors are generally better */ + for (int i = 0; i < TX_MAX_SUB; i++) + max_factor = FFMAX(cd->factors[i], max_factor); + if (max_factor) + prio += 16*max_factor; + + return prio; +} + +typedef struct FFTXLenDecomp { + int len; + int len2; + int prio; + const FFTXCodelet *cd; +} FFTXLenDecomp; + +static int cmp_decomp(FFTXLenDecomp *a, FFTXLenDecomp *b) +{ + return FFDIFFSIGN(b->prio, a->prio); +} + +int ff_tx_decompose_length(int dst[TX_MAX_DECOMPOSITIONS], enum AVTXType type, + int len, int inv) +{ + int nb_decomp = 0; + FFTXLenDecomp ld[TX_MAX_DECOMPOSITIONS]; + int codelet_list_idx = codelet_list_num; + + const int cpu_flags = av_get_cpu_flags(); + + /* Loop through all codelets in all codelet lists to find matches + * to the requirements */ + while (codelet_list_idx--) { + const FFTXCodelet * const * list = codelet_list[codelet_list_idx]; + const FFTXCodelet *cd = NULL; + + while ((cd = *list++)) { + int fl = len; + int skip = 0, prio; + int factors_product = 1, factors_mod = 0; + + if (nb_decomp >= TX_MAX_DECOMPOSITIONS) + goto sort; + + /* Check if the type matches */ + if (cd->type != TX_TYPE_ANY && type != cd->type) + continue; + + /* Check direction for non-orthogonal codelets */ + if (((cd->flags & FF_TX_FORWARD_ONLY) && inv) || + ((cd->flags & (FF_TX_INVERSE_ONLY | AV_TX_FULL_IMDCT)) && !inv) || + ((cd->flags & (FF_TX_FORWARD_ONLY | AV_TX_REAL_TO_REAL)) && inv) || + ((cd->flags & (FF_TX_FORWARD_ONLY | AV_TX_REAL_TO_IMAGINARY)) && inv)) + continue; + + /* Check if the CPU supports the required ISA */ + if (cd->cpu_flags != FF_TX_CPU_FLAGS_ALL && + !(cpu_flags & (cd->cpu_flags & ~cpu_slow_mask))) + continue; + + for (int i = 0; i < TX_MAX_FACTORS; i++) { + if (!cd->factors[i] || (fl == 1)) + break; + + if (cd->factors[i] == TX_FACTOR_ANY) { + factors_mod++; + factors_product *= fl; + } else if (!(fl % cd->factors[i])) { + factors_mod++; + if (cd->factors[i] == 2) { + int b = ff_ctz(fl); + fl >>= b; + factors_product <<= b; + } else { + do { + fl /= cd->factors[i]; + factors_product *= cd->factors[i]; + } while (!(fl % cd->factors[i])); + } + } + } + + /* Disqualify if factor requirements are not satisfied or if trivial */ + if ((factors_mod < cd->nb_factors) || (len == factors_product)) + continue; + + if (av_gcd(factors_product, fl) != 1) + continue; + + /* Check if length is supported and factorization was successful */ + if ((factors_product < cd->min_len) || + (cd->max_len != TX_LEN_UNLIMITED && (factors_product > cd->max_len))) + continue; + + prio = get_codelet_prio(cd, cpu_flags, factors_product) * factors_product; + + /* Check for duplicates */ + for (int i = 0; i < nb_decomp; i++) { + if (factors_product == ld[i].len) { + /* Update priority if new one is higher */ + if (prio > ld[i].prio) + ld[i].prio = prio; + skip = 1; + break; + } + } + + /* Add decomposition if unique */ + if (!skip) { + ld[nb_decomp].cd = cd; + ld[nb_decomp].len = factors_product; + ld[nb_decomp].len2 = fl; + ld[nb_decomp].prio = prio; + nb_decomp++; + } + } + } + + if (!nb_decomp) + return AVERROR(EINVAL); + +sort: + AV_QSORT(ld, nb_decomp, FFTXLenDecomp, cmp_decomp); + + for (int i = 0; i < nb_decomp; i++) { + if (ld[i].cd->nb_factors > 1) + dst[i] = ld[i].len2; + else + dst[i] = ld[i].len; + } + + return nb_decomp; +} + +int ff_tx_gen_default_map(AVTXContext *s, FFTXCodeletOptions *opts) +{ + s->map = av_malloc(s->len*sizeof(*s->map)); + if (!s->map) + return AVERROR(ENOMEM); + + s->map[0] = 0; /* DC is always at the start */ + if (s->inv) /* Reversing the ACs flips the transform direction */ + for (int i = 1; i < s->len; i++) + s->map[i] = s->len - i; + else + for (int i = 1; i < s->len; i++) + s->map[i] = i; + + s->map_dir = FF_TX_MAP_GATHER; + + return 0; +} + +#if !CONFIG_SMALL +static void print_flags(AVBPrint *bp, uint64_t f) +{ + int prev = 0; + const char *sep = ", "; + av_bprintf(bp, "flags: ["); + if ((f & FF_TX_ALIGNED) && ++prev) + av_bprintf(bp, "aligned"); + if ((f & AV_TX_UNALIGNED) && ++prev) + av_bprintf(bp, "%sunaligned", prev > 1 ? sep : ""); + if ((f & AV_TX_INPLACE) && ++prev) + av_bprintf(bp, "%sinplace", prev > 1 ? sep : ""); + if ((f & FF_TX_OUT_OF_PLACE) && ++prev) + av_bprintf(bp, "%sout_of_place", prev > 1 ? sep : ""); + if ((f & FF_TX_FORWARD_ONLY) && ++prev) + av_bprintf(bp, "%sfwd_only", prev > 1 ? sep : ""); + if ((f & FF_TX_INVERSE_ONLY) && ++prev) + av_bprintf(bp, "%sinv_only", prev > 1 ? sep : ""); + if ((f & FF_TX_PRESHUFFLE) && ++prev) + av_bprintf(bp, "%spreshuf", prev > 1 ? sep : ""); + if ((f & AV_TX_FULL_IMDCT) && ++prev) + av_bprintf(bp, "%simdct_full", prev > 1 ? sep : ""); + if ((f & AV_TX_REAL_TO_REAL) && ++prev) + av_bprintf(bp, "%sreal_to_real", prev > 1 ? sep : ""); + if ((f & AV_TX_REAL_TO_IMAGINARY) && ++prev) + av_bprintf(bp, "%sreal_to_imaginary", prev > 1 ? sep : ""); + if ((f & FF_TX_ASM_CALL) && ++prev) + av_bprintf(bp, "%sasm_call", prev > 1 ? sep : ""); + av_bprintf(bp, "]"); +} + +static void print_type(AVBPrint *bp, enum AVTXType type) +{ + av_bprintf(bp, "%s", + type == TX_TYPE_ANY ? "any" : + type == AV_TX_FLOAT_FFT ? "fft_float" : + type == AV_TX_FLOAT_MDCT ? "mdct_float" : + type == AV_TX_FLOAT_RDFT ? "rdft_float" : + type == AV_TX_FLOAT_DCT_I ? "dctI_float" : + type == AV_TX_FLOAT_DST_I ? "dstI_float" : + type == AV_TX_DOUBLE_FFT ? "fft_double" : + type == AV_TX_DOUBLE_MDCT ? "mdct_double" : + type == AV_TX_DOUBLE_RDFT ? "rdft_double" : + type == AV_TX_DOUBLE_DCT_I ? "dctI_double" : + type == AV_TX_DOUBLE_DST_I ? "dstI_double" : + type == AV_TX_INT32_FFT ? "fft_int32" : + type == AV_TX_INT32_MDCT ? "mdct_int32" : + type == AV_TX_INT32_RDFT ? "rdft_int32" : + type == AV_TX_INT32_DCT_I ? "dctI_int32" : + type == AV_TX_INT32_DST_I ? "dstI_int32" : + "unknown"); +} + +static void print_cd_info(const FFTXCodelet *cd, int prio, int len, int print_prio) +{ + AVBPrint bp; + av_bprint_init(&bp, 0, AV_BPRINT_SIZE_AUTOMATIC); + + av_bprintf(&bp, "%s - type: ", cd->name); + + print_type(&bp, cd->type); + + av_bprintf(&bp, ", len: "); + if (!len) { + if (cd->min_len != cd->max_len) + av_bprintf(&bp, "[%i, ", cd->min_len); + + if (cd->max_len == TX_LEN_UNLIMITED) + av_bprintf(&bp, "unlimited"); + else + av_bprintf(&bp, "%i", cd->max_len); + } else { + av_bprintf(&bp, "%i", len); + } + + if (cd->factors[1]) { + av_bprintf(&bp, "%s, factors", !len && cd->min_len != cd->max_len ? "]" : ""); + if (!cd->nb_factors) + av_bprintf(&bp, ": ["); + else + av_bprintf(&bp, "[%i]: [", cd->nb_factors); + + for (int i = 0; i < TX_MAX_FACTORS; i++) { + if (i && cd->factors[i]) + av_bprintf(&bp, ", "); + if (cd->factors[i] == TX_FACTOR_ANY) + av_bprintf(&bp, "any"); + else if (cd->factors[i]) + av_bprintf(&bp, "%i", cd->factors[i]); + else + break; + } + + av_bprintf(&bp, "], "); + } else { + av_bprintf(&bp, "%s, factor: %i, ", + !len && cd->min_len != cd->max_len ? "]" : "", cd->factors[0]); + } + print_flags(&bp, cd->flags); + + if (print_prio) + av_bprintf(&bp, ", prio: %i", prio); + + av_log(NULL, AV_LOG_DEBUG, "%s\n", bp.str); +} + +static void print_tx_structure(AVTXContext *s, int depth) +{ + const FFTXCodelet *cd = s->cd_self; + + for (int i = 0; i <= depth; i++) + av_log(NULL, AV_LOG_DEBUG, " "); + + print_cd_info(cd, cd->prio, s->len, 0); + + for (int i = 0; i < s->nb_sub; i++) + print_tx_structure(&s->sub[i], depth + 1); +} +#endif /* CONFIG_SMALL */ + +typedef struct TXCodeletMatch { + const FFTXCodelet *cd; + int prio; +} TXCodeletMatch; + +static int cmp_matches(TXCodeletMatch *a, TXCodeletMatch *b) +{ + return FFDIFFSIGN(b->prio, a->prio); +} + +/* We want all factors to completely cover the length */ +static inline int check_cd_factors(const FFTXCodelet *cd, int len) +{ + int matches = 0, any_flag = 0; + + for (int i = 0; i < TX_MAX_FACTORS; i++) { + int factor = cd->factors[i]; + + if (factor == TX_FACTOR_ANY) { + any_flag = 1; + matches++; + continue; + } else if (len <= 1 || !factor) { + break; + } else if (factor == 2) { /* Fast path */ + int bits_2 = ff_ctz(len); + if (!bits_2) + continue; /* Factor not supported */ + + len >>= bits_2; + matches++; + } else { + int res = len % factor; + if (res) + continue; /* Factor not supported */ + + while (!res) { + len /= factor; + res = len % factor; + } + matches++; + } + } + + return (cd->nb_factors <= matches) && (any_flag || len == 1); +} + +av_cold int ff_tx_init_subtx(AVTXContext *s, enum AVTXType type, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale) +{ + int ret = 0; + AVTXContext *sub = NULL; + TXCodeletMatch *cd_tmp, *cd_matches = NULL; + unsigned int cd_matches_size = 0; + int codelet_list_idx = codelet_list_num; + int nb_cd_matches = 0; +#if !CONFIG_SMALL + AVBPrint bp; +#endif + + /* We still accept functions marked with SLOW, even if the CPU is + * marked with the same flag, but we give them lower priority. */ + const int cpu_flags = av_get_cpu_flags(); + + /* Flags the transform wants */ + uint64_t req_flags = flags; + + /* Flags the codelet may require to be present */ + uint64_t inv_req_mask = AV_TX_FULL_IMDCT | + AV_TX_REAL_TO_REAL | + AV_TX_REAL_TO_IMAGINARY | + FF_TX_PRESHUFFLE | + FF_TX_ASM_CALL; + + /* Unaligned codelets are compatible with the aligned flag */ + if (req_flags & FF_TX_ALIGNED) + req_flags |= AV_TX_UNALIGNED; + + /* If either flag is set, both are okay, so don't check for an exact match */ + if ((req_flags & AV_TX_INPLACE) && (req_flags & FF_TX_OUT_OF_PLACE)) + req_flags &= ~(AV_TX_INPLACE | FF_TX_OUT_OF_PLACE); + if ((req_flags & FF_TX_ALIGNED) && (req_flags & AV_TX_UNALIGNED)) + req_flags &= ~(FF_TX_ALIGNED | AV_TX_UNALIGNED); + + /* Loop through all codelets in all codelet lists to find matches + * to the requirements */ + while (codelet_list_idx--) { + const FFTXCodelet * const * list = codelet_list[codelet_list_idx]; + const FFTXCodelet *cd = NULL; + + while ((cd = *list++)) { + /* Check if the type matches */ + if (cd->type != TX_TYPE_ANY && type != cd->type) + continue; + + /* Check direction for non-orthogonal codelets */ + if (((cd->flags & FF_TX_FORWARD_ONLY) && inv) || + ((cd->flags & (FF_TX_INVERSE_ONLY | AV_TX_FULL_IMDCT)) && !inv) || + ((cd->flags & (FF_TX_FORWARD_ONLY | AV_TX_REAL_TO_REAL)) && inv) || + ((cd->flags & (FF_TX_FORWARD_ONLY | AV_TX_REAL_TO_IMAGINARY)) && inv)) + continue; + + /* Check if the requested flags match from both sides */ + if (((req_flags & cd->flags) != (req_flags)) || + ((inv_req_mask & cd->flags) != (req_flags & inv_req_mask))) + continue; + + /* Check if length is supported */ + if ((len < cd->min_len) || (cd->max_len != -1 && (len > cd->max_len))) + continue; + + /* Check if the CPU supports the required ISA */ + if (cd->cpu_flags != FF_TX_CPU_FLAGS_ALL && + !(cpu_flags & (cd->cpu_flags & ~cpu_slow_mask))) + continue; + + /* Check for factors */ + if (!check_cd_factors(cd, len)) + continue; + + /* Realloc array and append */ + cd_tmp = av_fast_realloc(cd_matches, &cd_matches_size, + sizeof(*cd_tmp) * (nb_cd_matches + 1)); + if (!cd_tmp) { + av_free(cd_matches); + return AVERROR(ENOMEM); + } + + cd_matches = cd_tmp; + cd_matches[nb_cd_matches].cd = cd; + cd_matches[nb_cd_matches].prio = get_codelet_prio(cd, cpu_flags, len); + nb_cd_matches++; + } + } + +#if !CONFIG_SMALL + /* Print debugging info */ + av_bprint_init(&bp, 0, AV_BPRINT_SIZE_AUTOMATIC); + av_bprintf(&bp, "For transform of length %i, %s, ", len, + inv ? "inverse" : "forward"); + print_type(&bp, type); + av_bprintf(&bp, ", "); + print_flags(&bp, flags); + av_bprintf(&bp, ", found %i matches%s", nb_cd_matches, + nb_cd_matches ? ":" : "."); +#endif + + /* No matches found */ + if (!nb_cd_matches) + return AVERROR(ENOSYS); + + /* Sort the list */ + AV_QSORT(cd_matches, nb_cd_matches, TXCodeletMatch, cmp_matches); + +#if !CONFIG_SMALL + av_log(NULL, AV_LOG_DEBUG, "%s\n", bp.str); + + for (int i = 0; i < nb_cd_matches; i++) { + av_log(NULL, AV_LOG_DEBUG, " %i: ", i + 1); + print_cd_info(cd_matches[i].cd, cd_matches[i].prio, 0, 1); + } +#endif + + if (!s->sub) { + s->sub = sub = av_mallocz(TX_MAX_SUB*sizeof(*sub)); + if (!sub) { + ret = AVERROR(ENOMEM); + goto end; + } + } + + /* Attempt to initialize each */ + for (int i = 0; i < nb_cd_matches; i++) { + const FFTXCodelet *cd = cd_matches[i].cd; + AVTXContext *sctx = &s->sub[s->nb_sub]; + + sctx->len = len; + sctx->inv = inv; + sctx->type = type; + sctx->flags = cd->flags | flags; + sctx->cd_self = cd; + + s->fn[s->nb_sub] = cd->function; + s->cd[s->nb_sub] = cd; + + ret = 0; + if (cd->init) + ret = cd->init(sctx, cd, flags, opts, len, inv, scale); + + if (ret >= 0) { + if (opts && opts->map_dir != FF_TX_MAP_NONE && + sctx->map_dir == FF_TX_MAP_NONE) { + /* If a specific map direction was requested, and it doesn't + * exist, create one.*/ + sctx->map = av_malloc(len*sizeof(*sctx->map)); + if (!sctx->map) { + ret = AVERROR(ENOMEM); + goto end; + } + + for (int i = 0; i < len; i++) + sctx->map[i] = i; + } else if (opts && (opts->map_dir != sctx->map_dir)) { + int *tmp = av_malloc(len*sizeof(*sctx->map)); + if (!tmp) { + ret = AVERROR(ENOMEM); + goto end; + } + + memcpy(tmp, sctx->map, len*sizeof(*sctx->map)); + + for (int i = 0; i < len; i++) + sctx->map[tmp[i]] = i; + + av_free(tmp); + } + + s->nb_sub++; + goto end; + } + + s->fn[s->nb_sub] = NULL; + s->cd[s->nb_sub] = NULL; + + reset_ctx(sctx, 0); + if (ret == AVERROR(ENOMEM)) + break; + } + + if (!s->nb_sub) + av_freep(&s->sub); + +end: + av_free(cd_matches); + return ret; +} + +av_cold int av_tx_init(AVTXContext **ctx, av_tx_fn *tx, enum AVTXType type, + int inv, int len, const void *scale, uint64_t flags) +{ + int ret; + AVTXContext tmp = { 0 }; + const double default_scale_d = 1.0; + const float default_scale_f = 1.0f; + + if (!len || type >= AV_TX_NB || !ctx || !tx) + return AVERROR(EINVAL); + + if (!(flags & AV_TX_UNALIGNED)) + flags |= FF_TX_ALIGNED; + if (!(flags & AV_TX_INPLACE)) + flags |= FF_TX_OUT_OF_PLACE; + + if (!scale && ((type == AV_TX_FLOAT_MDCT) || (type == AV_TX_INT32_MDCT))) + scale = &default_scale_f; + else if (!scale && (type == AV_TX_DOUBLE_MDCT)) + scale = &default_scale_d; + + ret = ff_tx_init_subtx(&tmp, type, flags, NULL, len, inv, scale); + if (ret < 0) + return ret; + + *ctx = &tmp.sub[0]; + *tx = tmp.fn[0]; + +#if !CONFIG_SMALL + av_log(NULL, AV_LOG_DEBUG, "Transform tree:\n"); + print_tx_structure(*ctx, 0); +#endif + + return ret; +} diff --git a/media/ffvpx/libavutil/tx.h b/media/ffvpx/libavutil/tx.h new file mode 100644 index 0000000000..4696988cae --- /dev/null +++ b/media/ffvpx/libavutil/tx.h @@ -0,0 +1,210 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_TX_H +#define AVUTIL_TX_H + +#include <stdint.h> +#include <stddef.h> + +typedef struct AVTXContext AVTXContext; + +typedef struct AVComplexFloat { + float re, im; +} AVComplexFloat; + +typedef struct AVComplexDouble { + double re, im; +} AVComplexDouble; + +typedef struct AVComplexInt32 { + int32_t re, im; +} AVComplexInt32; + +enum AVTXType { + /** + * Standard complex to complex FFT with sample data type of AVComplexFloat, + * AVComplexDouble or AVComplexInt32, for each respective variant. + * + * Output is not 1/len normalized. Scaling currently unsupported. + * The stride parameter must be set to the size of a single sample in bytes. + */ + AV_TX_FLOAT_FFT = 0, + AV_TX_DOUBLE_FFT = 2, + AV_TX_INT32_FFT = 4, + + /** + * Standard MDCT with a sample data type of float, double or int32_t, + * respecively. For the float and int32 variants, the scale type is + * 'float', while for the double variant, it's 'double'. + * If scale is NULL, 1.0 will be used as a default. + * + * Length is the frame size, not the window size (which is 2x frame). + * For forward transforms, the stride specifies the spacing between each + * sample in the output array in bytes. The input must be a flat array. + * + * For inverse transforms, the stride specifies the spacing between each + * sample in the input array in bytes. The output must be a flat array. + * + * NOTE: the inverse transform is half-length, meaning the output will not + * contain redundant data. This is what most codecs work with. To do a full + * inverse transform, set the AV_TX_FULL_IMDCT flag on init. + */ + AV_TX_FLOAT_MDCT = 1, + AV_TX_DOUBLE_MDCT = 3, + AV_TX_INT32_MDCT = 5, + + /** + * Real to complex and complex to real DFTs. + * For the float and int32 variants, the scale type is 'float', while for + * the double variant, it's a 'double'. If scale is NULL, 1.0 will be used + * as a default. + * + * For forward transforms (R2C), stride must be the spacing between two + * samples in bytes. For inverse transforms, the stride must be set + * to the spacing between two complex values in bytes. + * + * The forward transform performs a real-to-complex DFT of N samples to + * N/2+1 complex values. + * + * The inverse transform performs a complex-to-real DFT of N/2+1 complex + * values to N real samples. The output is not normalized, but can be + * made so by setting the scale value to 1.0/len. + * NOTE: the inverse transform always overwrites the input. + */ + AV_TX_FLOAT_RDFT = 6, + AV_TX_DOUBLE_RDFT = 7, + AV_TX_INT32_RDFT = 8, + + /** + * Real to real (DCT) transforms. + * + * The forward transform is a DCT-II. + * The inverse transform is a DCT-III. + * + * The input array is always overwritten. DCT-III requires that the + * input be padded with 2 extra samples. Stride must be set to the + * spacing between two samples in bytes. + */ + AV_TX_FLOAT_DCT = 9, + AV_TX_DOUBLE_DCT = 10, + AV_TX_INT32_DCT = 11, + + /** + * Discrete Cosine Transform I + * + * The forward transform is a DCT-I. + * The inverse transform is a DCT-I multiplied by 2/(N + 1). + * + * The input array is always overwritten. + */ + AV_TX_FLOAT_DCT_I = 12, + AV_TX_DOUBLE_DCT_I = 13, + AV_TX_INT32_DCT_I = 14, + + /** + * Discrete Sine Transform I + * + * The forward transform is a DST-I. + * The inverse transform is a DST-I multiplied by 2/(N + 1). + * + * The input array is always overwritten. + */ + AV_TX_FLOAT_DST_I = 15, + AV_TX_DOUBLE_DST_I = 16, + AV_TX_INT32_DST_I = 17, + + /* Not part of the API, do not use */ + AV_TX_NB, +}; + +/** + * Function pointer to a function to perform the transform. + * + * @note Using a different context than the one allocated during av_tx_init() + * is not allowed. + * + * @param s the transform context + * @param out the output array + * @param in the input array + * @param stride the input or output stride in bytes + * + * The out and in arrays must be aligned to the maximum required by the CPU + * architecture unless the AV_TX_UNALIGNED flag was set in av_tx_init(). + * The stride must follow the constraints the transform type has specified. + */ +typedef void (*av_tx_fn)(AVTXContext *s, void *out, void *in, ptrdiff_t stride); + +/** + * Flags for av_tx_init() + */ +enum AVTXFlags { + /** + * Allows for in-place transformations, where input == output. + * May be unsupported or slower for some transform types. + */ + AV_TX_INPLACE = 1ULL << 0, + + /** + * Relaxes alignment requirement for the in and out arrays of av_tx_fn(). + * May be slower with certain transform types. + */ + AV_TX_UNALIGNED = 1ULL << 1, + + /** + * Performs a full inverse MDCT rather than leaving out samples that can be + * derived through symmetry. Requires an output array of 'len' floats, + * rather than the usual 'len/2' floats. + * Ignored for all transforms but inverse MDCTs. + */ + AV_TX_FULL_IMDCT = 1ULL << 2, + + /** + * Perform a real to half-complex RDFT. + * Only the real, or imaginary coefficients will + * be output, depending on the flag used. Only available for forward RDFTs. + * Output array must have enough space to hold N complex values + * (regular size for a real to complex transform). + */ + AV_TX_REAL_TO_REAL = 1ULL << 3, + AV_TX_REAL_TO_IMAGINARY = 1ULL << 4, +}; + +/** + * Initialize a transform context with the given configuration + * (i)MDCTs with an odd length are currently not supported. + * + * @param ctx the context to allocate, will be NULL on error + * @param tx pointer to the transform function pointer to set + * @param type type the type of transform + * @param inv whether to do an inverse or a forward transform + * @param len the size of the transform in samples + * @param scale pointer to the value to scale the output if supported by type + * @param flags a bitmask of AVTXFlags or 0 + * + * @return 0 on success, negative error code on failure + */ +int av_tx_init(AVTXContext **ctx, av_tx_fn *tx, enum AVTXType type, + int inv, int len, const void *scale, uint64_t flags); + +/** + * Frees a context and sets *ctx to NULL, does nothing when *ctx == NULL. + */ +void av_tx_uninit(AVTXContext **ctx); + +#endif /* AVUTIL_TX_H */ diff --git a/media/ffvpx/libavutil/tx_double.c b/media/ffvpx/libavutil/tx_double.c new file mode 100644 index 0000000000..7ea4283c94 --- /dev/null +++ b/media/ffvpx/libavutil/tx_double.c @@ -0,0 +1,21 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TX_DOUBLE +#include "tx_priv.h" +#include "tx_template.c" diff --git a/media/ffvpx/libavutil/tx_float.c b/media/ffvpx/libavutil/tx_float.c new file mode 100644 index 0000000000..018f2a21ac --- /dev/null +++ b/media/ffvpx/libavutil/tx_float.c @@ -0,0 +1,21 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TX_FLOAT +#include "tx_priv.h" +#include "tx_template.c" diff --git a/media/ffvpx/libavutil/tx_int32.c b/media/ffvpx/libavutil/tx_int32.c new file mode 100644 index 0000000000..9261013bf6 --- /dev/null +++ b/media/ffvpx/libavutil/tx_int32.c @@ -0,0 +1,21 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TX_INT32 +#include "tx_priv.h" +#include "tx_template.c" diff --git a/media/ffvpx/libavutil/tx_priv.h b/media/ffvpx/libavutil/tx_priv.h new file mode 100644 index 0000000000..d3fcdbf563 --- /dev/null +++ b/media/ffvpx/libavutil/tx_priv.h @@ -0,0 +1,392 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_TX_PRIV_H +#define AVUTIL_TX_PRIV_H + +#include "tx.h" +#include "thread.h" +#include "mem_internal.h" +#include "common.h" +#include "attributes.h" + +#ifdef TX_FLOAT +#define TX_TAB(x) x ## _float +#define TX_NAME(x) x ## _float_c +#define TX_NAME_STR(x) NULL_IF_CONFIG_SMALL(x "_float_c") +#define TX_TYPE(x) AV_TX_FLOAT_ ## x +#define TX_FN_NAME(fn, suffix) ff_tx_ ## fn ## _float_ ## suffix +#define TX_FN_NAME_STR(fn, suffix) NULL_IF_CONFIG_SMALL(#fn "_float_" #suffix) +#define MULT(x, m) ((x) * (m)) +#define SCALE_TYPE float +typedef float TXSample; +typedef float TXUSample; +typedef AVComplexFloat TXComplex; +#elif defined(TX_DOUBLE) +#define TX_TAB(x) x ## _double +#define TX_NAME(x) x ## _double_c +#define TX_NAME_STR(x) NULL_IF_CONFIG_SMALL(x "_double_c") +#define TX_TYPE(x) AV_TX_DOUBLE_ ## x +#define TX_FN_NAME(fn, suffix) ff_tx_ ## fn ## _double_ ## suffix +#define TX_FN_NAME_STR(fn, suffix) NULL_IF_CONFIG_SMALL(#fn "_double_" #suffix) +#define MULT(x, m) ((x) * (m)) +#define SCALE_TYPE double +typedef double TXSample; +typedef double TXUSample; +typedef AVComplexDouble TXComplex; +#elif defined(TX_INT32) +#define TX_TAB(x) x ## _int32 +#define TX_NAME(x) x ## _int32_c +#define TX_NAME_STR(x) NULL_IF_CONFIG_SMALL(x "_int32_c") +#define TX_TYPE(x) AV_TX_INT32_ ## x +#define TX_FN_NAME(fn, suffix) ff_tx_ ## fn ## _int32_ ## suffix +#define TX_FN_NAME_STR(fn, suffix) NULL_IF_CONFIG_SMALL(#fn "_int32_" #suffix) +#define MULT(x, m) (((((int64_t)(x)) * (int64_t)(m)) + 0x40000000) >> 31) +#define SCALE_TYPE float +typedef int32_t TXSample; +typedef uint32_t TXUSample; +typedef AVComplexInt32 TXComplex; +#else +typedef void TXComplex; +#endif + +#define TX_DECL_FN(fn, suffix) \ + void TX_FN_NAME(fn, suffix)(AVTXContext *s, void *o, void *i, ptrdiff_t st); + +#define TX_DEF(fn, tx_type, len_min, len_max, f1, f2, \ + p, init_fn, suffix, cf, cd_flags, cf2) \ + &(const FFTXCodelet){ \ + .name = TX_FN_NAME_STR(fn, suffix), \ + .function = TX_FN_NAME(fn, suffix), \ + .type = TX_TYPE(tx_type), \ + .flags = FF_TX_ALIGNED | FF_TX_OUT_OF_PLACE | cd_flags, \ + .factors = { (f1), (f2) }, \ + .nb_factors = !!(f1) + !!(f2), \ + .min_len = len_min, \ + .max_len = len_max, \ + .init = init_fn, \ + .cpu_flags = cf2 | AV_CPU_FLAG_ ## cf, \ + .prio = p, \ + } + +#if defined(TX_FLOAT) || defined(TX_DOUBLE) + +#define CMUL(dre, dim, are, aim, bre, bim) \ + do { \ + (dre) = (are) * (bre) - (aim) * (bim); \ + (dim) = (are) * (bim) + (aim) * (bre); \ + } while (0) + +#define SMUL(dre, dim, are, aim, bre, bim) \ + do { \ + (dre) = (are) * (bre) - (aim) * (bim); \ + (dim) = (are) * (bim) - (aim) * (bre); \ + } while (0) + +#define UNSCALE(x) (x) +#define RESCALE(x) (x) + +#define FOLD(a, b) ((a) + (b)) + +#define BF(x, y, a, b) \ + do { \ + x = (a) - (b); \ + y = (a) + (b); \ + } while (0) + +#elif defined(TX_INT32) + +/* Properly rounds the result */ +#define CMUL(dre, dim, are, aim, bre, bim) \ + do { \ + int64_t accu; \ + (accu) = (int64_t)(bre) * (are); \ + (accu) -= (int64_t)(bim) * (aim); \ + (dre) = (int)(((accu) + 0x40000000) >> 31); \ + (accu) = (int64_t)(bim) * (are); \ + (accu) += (int64_t)(bre) * (aim); \ + (dim) = (int)(((accu) + 0x40000000) >> 31); \ + } while (0) + +#define SMUL(dre, dim, are, aim, bre, bim) \ + do { \ + int64_t accu; \ + (accu) = (int64_t)(bre) * (are); \ + (accu) -= (int64_t)(bim) * (aim); \ + (dre) = (int)(((accu) + 0x40000000) >> 31); \ + (accu) = (int64_t)(bim) * (are); \ + (accu) -= (int64_t)(bre) * (aim); \ + (dim) = (int)(((accu) + 0x40000000) >> 31); \ + } while (0) + +#define UNSCALE(x) ((double)(x)/2147483648.0) +#define RESCALE(x) (av_clip64(llrintf((x) * 2147483648.0), INT32_MIN, INT32_MAX)) + +#define FOLD(x, y) ((int32_t)((x) + (unsigned)(y) + 32) >> 6) + +#define BF(x, y, a, b) \ + do { \ + x = (a) - (unsigned)(b); \ + y = (a) + (unsigned)(b); \ + } while (0) + +#endif /* TX_INT32 */ + +#define CMUL3(c, a, b) CMUL((c).re, (c).im, (a).re, (a).im, (b).re, (b).im) + +/* Codelet flags, used to pick codelets. Must be a superset of enum AVTXFlags, + * but if it runs out of bits, it can be made separate. */ +#define FF_TX_OUT_OF_PLACE (1ULL << 63) /* Can be OR'd with AV_TX_INPLACE */ +#define FF_TX_ALIGNED (1ULL << 62) /* Cannot be OR'd with AV_TX_UNALIGNED */ +#define FF_TX_PRESHUFFLE (1ULL << 61) /* Codelet expects permuted coeffs */ +#define FF_TX_INVERSE_ONLY (1ULL << 60) /* For non-orthogonal inverse-only transforms */ +#define FF_TX_FORWARD_ONLY (1ULL << 59) /* For non-orthogonal forward-only transforms */ +#define FF_TX_ASM_CALL (1ULL << 58) /* For asm->asm functions only */ + +typedef enum FFTXCodeletPriority { + FF_TX_PRIO_BASE = 0, /* Baseline priority */ + + /* For SIMD, set base prio to the register size in bits and increment in + * steps of 64 depending on faster/slower features, like FMA. */ + + FF_TX_PRIO_MIN = -131072, /* For naive implementations */ + FF_TX_PRIO_MAX = 32768, /* For custom implementations/ASICs */ +} FFTXCodeletPriority; + +typedef enum FFTXMapDirection { + /* No map. Make a map up. */ + FF_TX_MAP_NONE = 0, + + /* Lookup table must be applied via dst[i] = src[lut[i]]; */ + FF_TX_MAP_GATHER, + + /* Lookup table must be applied via dst[lut[i]] = src[i]; */ + FF_TX_MAP_SCATTER, +} FFTXMapDirection; + +/* Codelet options */ +typedef struct FFTXCodeletOptions { + /* Request a specific lookup table direction. Codelets MUST put the + * direction in AVTXContext. If the codelet does not respect this, a + * conversion will be performed. */ + FFTXMapDirection map_dir; +} FFTXCodeletOptions; + +/* Maximum number of factors a codelet may have. Arbitrary. */ +#define TX_MAX_FACTORS 16 + +/* Maximum amount of subtransform functions, subtransforms and factors. Arbitrary. */ +#define TX_MAX_SUB 4 + +/* Maximum number of returned results for ff_tx_decompose_length. Arbitrary. */ +#define TX_MAX_DECOMPOSITIONS 512 + +typedef struct FFTXCodelet { + const char *name; /* Codelet name, for debugging */ + av_tx_fn function; /* Codelet function, != NULL */ + enum AVTXType type; /* Type of codelet transform */ +#define TX_TYPE_ANY INT32_MAX /* Special type to allow all types */ + + uint64_t flags; /* A combination of AVTXFlags and codelet + * flags that describe its properties. */ + + int factors[TX_MAX_FACTORS]; /* Length factors. MUST be coprime. */ +#define TX_FACTOR_ANY -1 /* When used alone, signals that the codelet + * supports all factors. Otherwise, if other + * factors are present, it signals that whatever + * remains will be supported, as long as the + * other factors are a component of the length */ + + int nb_factors; /* Minimum number of factors that have to + * be a modulo of the length. Must not be 0. */ + + int min_len; /* Minimum length of transform, must be >= 1 */ + int max_len; /* Maximum length of transform */ +#define TX_LEN_UNLIMITED -1 /* Special length value to permit all lengths */ + + int (*init)(AVTXContext *s, /* Optional callback for current context initialization. */ + const struct FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale); + + int (*uninit)(AVTXContext *s); /* Optional callback for uninitialization. */ + + int cpu_flags; /* CPU flags. If any negative flags like + * SLOW are present, will avoid picking. + * 0x0 to signal it's a C codelet */ +#define FF_TX_CPU_FLAGS_ALL 0x0 /* Special CPU flag for C */ + + int prio; /* < 0 = least, 0 = no pref, > 0 = prefer */ +} FFTXCodelet; + +struct AVTXContext { + /* Fields the root transform and subtransforms use or may use. + * NOTE: This section is used by assembly, do not reorder or change */ + int len; /* Length of the transform */ + int inv; /* If transform is inverse */ + int *map; /* Lookup table(s) */ + TXComplex *exp; /* Any non-pre-baked multiplication factors, + * or extra temporary buffer */ + TXComplex *tmp; /* Temporary buffer, if needed */ + + AVTXContext *sub; /* Subtransform context(s), if needed */ + av_tx_fn fn[TX_MAX_SUB]; /* Function(s) for the subtransforms */ + int nb_sub; /* Number of subtransforms. + * The reason all of these are set here + * rather than in each separate context + * is to eliminate extra pointer + * dereferences. */ + + /* Fields mainly useul/applicable for the root transform or initialization. + * Fields below are not used by assembly code. */ + const FFTXCodelet *cd[TX_MAX_SUB]; /* Subtransform codelets */ + const FFTXCodelet *cd_self; /* Codelet for the current context */ + enum AVTXType type; /* Type of transform */ + uint64_t flags; /* A combination of AVTXFlags and + * codelet flags used when creating */ + FFTXMapDirection map_dir; /* Direction of AVTXContext->map */ + float scale_f; + double scale_d; + void *opaque; /* Free to use by implementations */ +}; + +/* This function embeds a Ruritanian PFA input map into an existing lookup table + * to avoid double permutation. This allows for compound factors to be + * synthesized as fast PFA FFTs and embedded into either other or standalone + * transforms. + * The output CRT map must still be pre-baked into the transform. */ +#define TX_EMBED_INPUT_PFA_MAP(map, tot_len, d1, d2) \ + do { \ + int mtmp[(d1)*(d2)]; \ + for (int k = 0; k < tot_len; k += (d1)*(d2)) { \ + memcpy(mtmp, &map[k], (d1)*(d2)*sizeof(*mtmp)); \ + for (int m = 0; m < (d2); m++) \ + for (int n = 0; n < (d1); n++) \ + map[k + m*(d1) + n] = mtmp[(m*(d1) + n*(d2)) % ((d1)*(d2))]; \ + } \ + } while (0) + +/* This function generates a Ruritanian PFA input map into s->map. */ +int ff_tx_gen_pfa_input_map(AVTXContext *s, FFTXCodeletOptions *opts, + int d1, int d2); + +/* Create a subtransform in the current context with the given parameters. + * The flags parameter from FFTXCodelet.init() should be preserved as much + * as that's possible. + * MUST be called during the sub() callback of each codelet. */ +int ff_tx_init_subtx(AVTXContext *s, enum AVTXType type, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale); + +/* Clear the context by freeing all tables, maps and subtransforms. */ +void ff_tx_clear_ctx(AVTXContext *s); + +/* Attempt to factorize a length into 2 integers such that + * len / dst1 == dst2, where dst1 and dst2 are coprime. */ +int ff_tx_decompose_length(int dst[TX_MAX_DECOMPOSITIONS], enum AVTXType type, + int len, int inv); + +/* Generate a default map (0->len or 0, (len-1)->1 for inverse transforms) + * for a context. */ +int ff_tx_gen_default_map(AVTXContext *s, FFTXCodeletOptions *opts); + +/* + * Generates the PFA permutation table into AVTXContext->pfatab. The end table + * is appended to the start table. + * The `inv` flag should only be enabled if the lookup tables of subtransforms + * won't get flattened. + */ +int ff_tx_gen_compound_mapping(AVTXContext *s, FFTXCodeletOptions *opts, + int inv, int n, int m); + +/* + * Generates a standard-ish (slightly modified) Split-Radix revtab into + * AVTXContext->map. Invert lookup changes how the mapping needs to be applied. + * If it's set to 0, it has to be applied like out[map[i]] = in[i], otherwise + * if it's set to 1, has to be applied as out[i] = in[map[i]] + */ +int ff_tx_gen_ptwo_revtab(AVTXContext *s, FFTXCodeletOptions *opts); + +/* + * Generates an index into AVTXContext->inplace_idx that if followed in the + * specific order, allows the revtab to be done in-place. The sub-transform + * and its map should already be initialized. + */ +int ff_tx_gen_inplace_map(AVTXContext *s, int len); + +/* + * This generates a parity-based revtab of length len and direction inv. + * + * Parity means even and odd complex numbers will be split, e.g. the even + * coefficients will come first, after which the odd coefficients will be + * placed. For example, a 4-point transform's coefficients after reordering: + * z[0].re, z[0].im, z[2].re, z[2].im, z[1].re, z[1].im, z[3].re, z[3].im + * + * The basis argument is the length of the largest non-composite transform + * supported, and also implies that the basis/2 transform is supported as well, + * as the split-radix algorithm requires it to be. + * + * The dual_stride argument indicates that both the basis, as well as the + * basis/2 transforms support doing two transforms at once, and the coefficients + * will be interleaved between each pair in a split-radix like so (stride == 2): + * tx1[0], tx1[2], tx2[0], tx2[2], tx1[1], tx1[3], tx2[1], tx2[3] + * A non-zero number switches this on, with the value indicating the stride + * (how many values of 1 transform to put first before switching to the other). + * Must be a power of two or 0. Must be less than the basis. + * Value will be clipped to the transform size, so for a basis of 16 and a + * dual_stride of 8, dual 8-point transforms will be laid out as if dual_stride + * was set to 4. + * Usually you'll set this to half the complex numbers that fit in a single + * register or 0. This allows to reuse SSE functions as dual-transform + * functions in AVX mode. + * + * If length is smaller than basis/2 this function will not do anything. + * + * If inv_lookup is set to 1, it will flip the lookup from out[map[i]] = src[i] + * to out[i] = src[map[i]]. + */ +int ff_tx_gen_split_radix_parity_revtab(AVTXContext *s, int len, int inv, + FFTXCodeletOptions *opts, + int basis, int dual_stride); + +/* Typed init function to initialize shared tables. Will initialize all tables + * for all factors of a length. */ +void ff_tx_init_tabs_float (int len); +void ff_tx_init_tabs_double(int len); +void ff_tx_init_tabs_int32 (int len); + +/* Typed init function to initialize an MDCT exptab in a context. + * If pre_tab is set, duplicates the entire table, with the first + * copy being shuffled according to pre_tab, and the second copy + * being the original. */ +int ff_tx_mdct_gen_exp_float (AVTXContext *s, int *pre_tab); +int ff_tx_mdct_gen_exp_double(AVTXContext *s, int *pre_tab); +int ff_tx_mdct_gen_exp_int32 (AVTXContext *s, int *pre_tab); + +/* Lists of codelets */ +extern const FFTXCodelet * const ff_tx_codelet_list_float_c []; +extern const FFTXCodelet * const ff_tx_codelet_list_float_x86 []; +extern const FFTXCodelet * const ff_tx_codelet_list_float_aarch64 []; + +extern const FFTXCodelet * const ff_tx_codelet_list_double_c []; + +extern const FFTXCodelet * const ff_tx_codelet_list_int32_c []; + +#endif /* AVUTIL_TX_PRIV_H */ diff --git a/media/ffvpx/libavutil/tx_template.c b/media/ffvpx/libavutil/tx_template.c new file mode 100644 index 0000000000..a2c27465cb --- /dev/null +++ b/media/ffvpx/libavutil/tx_template.c @@ -0,0 +1,2214 @@ +/* + * Copyright (c) Lynne + * + * Power of two FFT: + * Copyright (c) Lynne + * Copyright (c) 2008 Loren Merritt + * Copyright (c) 2002 Fabrice Bellard + * Partly based on libdjbfft by D. J. Bernstein + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TABLE_DEF(name, size) \ + DECLARE_ALIGNED(32, TXSample, TX_TAB(ff_tx_tab_ ##name))[size] + +#define SR_POW2_TABLES \ + SR_TABLE(8) \ + SR_TABLE(16) \ + SR_TABLE(32) \ + SR_TABLE(64) \ + SR_TABLE(128) \ + SR_TABLE(256) \ + SR_TABLE(512) \ + SR_TABLE(1024) \ + SR_TABLE(2048) \ + SR_TABLE(4096) \ + SR_TABLE(8192) \ + SR_TABLE(16384) \ + SR_TABLE(32768) \ + SR_TABLE(65536) \ + SR_TABLE(131072) \ + SR_TABLE(262144) \ + SR_TABLE(524288) \ + SR_TABLE(1048576) \ + SR_TABLE(2097152) \ + +#define SR_TABLE(len) \ + TABLE_DEF(len, len/4 + 1); +/* Power of two tables */ +SR_POW2_TABLES +#undef SR_TABLE + +/* Other factors' tables */ +TABLE_DEF(53, 12); +TABLE_DEF( 7, 6); +TABLE_DEF( 9, 8); + +typedef struct FFTabInitData { + void (*func)(void); + int factors[TX_MAX_SUB]; /* Must be sorted high -> low */ +} FFTabInitData; + +#define SR_TABLE(len) \ +static av_cold void TX_TAB(ff_tx_init_tab_ ##len)(void) \ +{ \ + double freq = 2*M_PI/len; \ + TXSample *tab = TX_TAB(ff_tx_tab_ ##len); \ + \ + for (int i = 0; i < len/4; i++) \ + *tab++ = RESCALE(cos(i*freq)); \ + \ + *tab = 0; \ +} +SR_POW2_TABLES +#undef SR_TABLE + +static void (*const sr_tabs_init_funcs[])(void) = { +#define SR_TABLE(len) TX_TAB(ff_tx_init_tab_ ##len), + SR_POW2_TABLES +#undef SR_TABLE +}; + +static AVOnce sr_tabs_init_once[] = { +#define SR_TABLE(len) AV_ONCE_INIT, + SR_POW2_TABLES +#undef SR_TABLE +}; + +static av_cold void TX_TAB(ff_tx_init_tab_53)(void) +{ + /* 5pt, doubled to eliminate AVX lane shuffles */ + TX_TAB(ff_tx_tab_53)[0] = RESCALE(cos(2 * M_PI / 5)); + TX_TAB(ff_tx_tab_53)[1] = RESCALE(cos(2 * M_PI / 5)); + TX_TAB(ff_tx_tab_53)[2] = RESCALE(cos(2 * M_PI / 10)); + TX_TAB(ff_tx_tab_53)[3] = RESCALE(cos(2 * M_PI / 10)); + TX_TAB(ff_tx_tab_53)[4] = RESCALE(sin(2 * M_PI / 5)); + TX_TAB(ff_tx_tab_53)[5] = RESCALE(sin(2 * M_PI / 5)); + TX_TAB(ff_tx_tab_53)[6] = RESCALE(sin(2 * M_PI / 10)); + TX_TAB(ff_tx_tab_53)[7] = RESCALE(sin(2 * M_PI / 10)); + + /* 3pt */ + TX_TAB(ff_tx_tab_53)[ 8] = RESCALE(cos(2 * M_PI / 12)); + TX_TAB(ff_tx_tab_53)[ 9] = RESCALE(cos(2 * M_PI / 12)); + TX_TAB(ff_tx_tab_53)[10] = RESCALE(cos(2 * M_PI / 6)); + TX_TAB(ff_tx_tab_53)[11] = RESCALE(cos(8 * M_PI / 6)); +} + +static av_cold void TX_TAB(ff_tx_init_tab_7)(void) +{ + TX_TAB(ff_tx_tab_7)[0] = RESCALE(cos(2 * M_PI / 7)); + TX_TAB(ff_tx_tab_7)[1] = RESCALE(sin(2 * M_PI / 7)); + TX_TAB(ff_tx_tab_7)[2] = RESCALE(sin(2 * M_PI / 28)); + TX_TAB(ff_tx_tab_7)[3] = RESCALE(cos(2 * M_PI / 28)); + TX_TAB(ff_tx_tab_7)[4] = RESCALE(cos(2 * M_PI / 14)); + TX_TAB(ff_tx_tab_7)[5] = RESCALE(sin(2 * M_PI / 14)); +} + +static av_cold void TX_TAB(ff_tx_init_tab_9)(void) +{ + TX_TAB(ff_tx_tab_9)[0] = RESCALE(cos(2 * M_PI / 3)); + TX_TAB(ff_tx_tab_9)[1] = RESCALE(sin(2 * M_PI / 3)); + TX_TAB(ff_tx_tab_9)[2] = RESCALE(cos(2 * M_PI / 9)); + TX_TAB(ff_tx_tab_9)[3] = RESCALE(sin(2 * M_PI / 9)); + TX_TAB(ff_tx_tab_9)[4] = RESCALE(cos(2 * M_PI / 36)); + TX_TAB(ff_tx_tab_9)[5] = RESCALE(sin(2 * M_PI / 36)); + TX_TAB(ff_tx_tab_9)[6] = TX_TAB(ff_tx_tab_9)[2] + TX_TAB(ff_tx_tab_9)[5]; + TX_TAB(ff_tx_tab_9)[7] = TX_TAB(ff_tx_tab_9)[3] - TX_TAB(ff_tx_tab_9)[4]; +} + +static const FFTabInitData nptwo_tabs_init_data[] = { + { TX_TAB(ff_tx_init_tab_53), { 15, 5, 3 } }, + { TX_TAB(ff_tx_init_tab_9), { 9 } }, + { TX_TAB(ff_tx_init_tab_7), { 7 } }, +}; + +static AVOnce nptwo_tabs_init_once[] = { + AV_ONCE_INIT, + AV_ONCE_INIT, + AV_ONCE_INIT, +}; + +av_cold void TX_TAB(ff_tx_init_tabs)(int len) +{ + int factor_2 = ff_ctz(len); + if (factor_2) { + int idx = factor_2 - 3; + for (int i = 0; i <= idx; i++) + ff_thread_once(&sr_tabs_init_once[i], + sr_tabs_init_funcs[i]); + len >>= factor_2; + } + + for (int i = 0; i < FF_ARRAY_ELEMS(nptwo_tabs_init_data); i++) { + int f, f_idx = 0; + + if (len <= 1) + return; + + while ((f = nptwo_tabs_init_data[i].factors[f_idx++])) { + if (f % len) + continue; + + ff_thread_once(&nptwo_tabs_init_once[i], + nptwo_tabs_init_data[i].func); + len /= f; + break; + } + } +} + +static av_always_inline void fft3(TXComplex *out, TXComplex *in, + ptrdiff_t stride) +{ + TXComplex tmp[3]; + const TXSample *tab = TX_TAB(ff_tx_tab_53); +#ifdef TX_INT32 + int64_t mtmp[4]; +#endif + + tmp[0] = in[0]; + BF(tmp[1].re, tmp[2].im, in[1].im, in[2].im); + BF(tmp[1].im, tmp[2].re, in[1].re, in[2].re); + +#ifdef TX_INT32 + out[0*stride].re = (int64_t)tmp[0].re + tmp[2].re; + out[0*stride].im = (int64_t)tmp[0].im + tmp[2].im; + mtmp[0] = (int64_t)tab[ 8] * tmp[1].re; + mtmp[1] = (int64_t)tab[ 9] * tmp[1].im; + mtmp[2] = (int64_t)tab[10] * tmp[2].re; + mtmp[3] = (int64_t)tab[10] * tmp[2].im; + out[1*stride].re = tmp[0].re - (mtmp[2] + mtmp[0] + 0x40000000 >> 31); + out[1*stride].im = tmp[0].im - (mtmp[3] - mtmp[1] + 0x40000000 >> 31); + out[2*stride].re = tmp[0].re - (mtmp[2] - mtmp[0] + 0x40000000 >> 31); + out[2*stride].im = tmp[0].im - (mtmp[3] + mtmp[1] + 0x40000000 >> 31); +#else + out[0*stride].re = tmp[0].re + tmp[2].re; + out[0*stride].im = tmp[0].im + tmp[2].im; + tmp[1].re = tab[ 8] * tmp[1].re; + tmp[1].im = tab[ 9] * tmp[1].im; + tmp[2].re = tab[10] * tmp[2].re; + tmp[2].im = tab[10] * tmp[2].im; + out[1*stride].re = tmp[0].re - tmp[2].re + tmp[1].re; + out[1*stride].im = tmp[0].im - tmp[2].im - tmp[1].im; + out[2*stride].re = tmp[0].re - tmp[2].re - tmp[1].re; + out[2*stride].im = tmp[0].im - tmp[2].im + tmp[1].im; +#endif +} + +#define DECL_FFT5(NAME, D0, D1, D2, D3, D4) \ +static av_always_inline void NAME(TXComplex *out, TXComplex *in, \ + ptrdiff_t stride) \ +{ \ + TXComplex dc, z0[4], t[6]; \ + const TXSample *tab = TX_TAB(ff_tx_tab_53); \ + \ + dc = in[0]; \ + BF(t[1].im, t[0].re, in[1].re, in[4].re); \ + BF(t[1].re, t[0].im, in[1].im, in[4].im); \ + BF(t[3].im, t[2].re, in[2].re, in[3].re); \ + BF(t[3].re, t[2].im, in[2].im, in[3].im); \ + \ + out[D0*stride].re = dc.re + (TXUSample)t[0].re + t[2].re; \ + out[D0*stride].im = dc.im + (TXUSample)t[0].im + t[2].im; \ + \ + SMUL(t[4].re, t[0].re, tab[0], tab[2], t[2].re, t[0].re); \ + SMUL(t[4].im, t[0].im, tab[0], tab[2], t[2].im, t[0].im); \ + CMUL(t[5].re, t[1].re, tab[4], tab[6], t[3].re, t[1].re); \ + CMUL(t[5].im, t[1].im, tab[4], tab[6], t[3].im, t[1].im); \ + \ + BF(z0[0].re, z0[3].re, t[0].re, t[1].re); \ + BF(z0[0].im, z0[3].im, t[0].im, t[1].im); \ + BF(z0[2].re, z0[1].re, t[4].re, t[5].re); \ + BF(z0[2].im, z0[1].im, t[4].im, t[5].im); \ + \ + out[D1*stride].re = dc.re + (TXUSample)z0[3].re; \ + out[D1*stride].im = dc.im + (TXUSample)z0[0].im; \ + out[D2*stride].re = dc.re + (TXUSample)z0[2].re; \ + out[D2*stride].im = dc.im + (TXUSample)z0[1].im; \ + out[D3*stride].re = dc.re + (TXUSample)z0[1].re; \ + out[D3*stride].im = dc.im + (TXUSample)z0[2].im; \ + out[D4*stride].re = dc.re + (TXUSample)z0[0].re; \ + out[D4*stride].im = dc.im + (TXUSample)z0[3].im; \ +} + +DECL_FFT5(fft5, 0, 1, 2, 3, 4) +DECL_FFT5(fft5_m1, 0, 6, 12, 3, 9) +DECL_FFT5(fft5_m2, 10, 1, 7, 13, 4) +DECL_FFT5(fft5_m3, 5, 11, 2, 8, 14) + +static av_always_inline void fft7(TXComplex *out, TXComplex *in, + ptrdiff_t stride) +{ + TXComplex dc, t[6], z[3]; + const TXComplex *tab = (const TXComplex *)TX_TAB(ff_tx_tab_7); +#ifdef TX_INT32 + int64_t mtmp[12]; +#endif + + dc = in[0]; + BF(t[1].re, t[0].re, in[1].re, in[6].re); + BF(t[1].im, t[0].im, in[1].im, in[6].im); + BF(t[3].re, t[2].re, in[2].re, in[5].re); + BF(t[3].im, t[2].im, in[2].im, in[5].im); + BF(t[5].re, t[4].re, in[3].re, in[4].re); + BF(t[5].im, t[4].im, in[3].im, in[4].im); + + out[0*stride].re = dc.re + t[0].re + t[2].re + t[4].re; + out[0*stride].im = dc.im + t[0].im + t[2].im + t[4].im; + +#ifdef TX_INT32 /* NOTE: it's possible to do this with 16 mults but 72 adds */ + mtmp[ 0] = ((int64_t)tab[0].re)*t[0].re - ((int64_t)tab[2].re)*t[4].re; + mtmp[ 1] = ((int64_t)tab[0].re)*t[4].re - ((int64_t)tab[1].re)*t[0].re; + mtmp[ 2] = ((int64_t)tab[0].re)*t[2].re - ((int64_t)tab[2].re)*t[0].re; + mtmp[ 3] = ((int64_t)tab[0].re)*t[0].im - ((int64_t)tab[1].re)*t[2].im; + mtmp[ 4] = ((int64_t)tab[0].re)*t[4].im - ((int64_t)tab[1].re)*t[0].im; + mtmp[ 5] = ((int64_t)tab[0].re)*t[2].im - ((int64_t)tab[2].re)*t[0].im; + + mtmp[ 6] = ((int64_t)tab[2].im)*t[1].im + ((int64_t)tab[1].im)*t[5].im; + mtmp[ 7] = ((int64_t)tab[0].im)*t[5].im + ((int64_t)tab[2].im)*t[3].im; + mtmp[ 8] = ((int64_t)tab[2].im)*t[5].im + ((int64_t)tab[1].im)*t[3].im; + mtmp[ 9] = ((int64_t)tab[0].im)*t[1].re + ((int64_t)tab[1].im)*t[3].re; + mtmp[10] = ((int64_t)tab[2].im)*t[3].re + ((int64_t)tab[0].im)*t[5].re; + mtmp[11] = ((int64_t)tab[2].im)*t[1].re + ((int64_t)tab[1].im)*t[5].re; + + z[0].re = (int32_t)(mtmp[ 0] - ((int64_t)tab[1].re)*t[2].re + 0x40000000 >> 31); + z[1].re = (int32_t)(mtmp[ 1] - ((int64_t)tab[2].re)*t[2].re + 0x40000000 >> 31); + z[2].re = (int32_t)(mtmp[ 2] - ((int64_t)tab[1].re)*t[4].re + 0x40000000 >> 31); + z[0].im = (int32_t)(mtmp[ 3] - ((int64_t)tab[2].re)*t[4].im + 0x40000000 >> 31); + z[1].im = (int32_t)(mtmp[ 4] - ((int64_t)tab[2].re)*t[2].im + 0x40000000 >> 31); + z[2].im = (int32_t)(mtmp[ 5] - ((int64_t)tab[1].re)*t[4].im + 0x40000000 >> 31); + + t[0].re = (int32_t)(mtmp[ 6] - ((int64_t)tab[0].im)*t[3].im + 0x40000000 >> 31); + t[2].re = (int32_t)(mtmp[ 7] - ((int64_t)tab[1].im)*t[1].im + 0x40000000 >> 31); + t[4].re = (int32_t)(mtmp[ 8] + ((int64_t)tab[0].im)*t[1].im + 0x40000000 >> 31); + t[0].im = (int32_t)(mtmp[ 9] + ((int64_t)tab[2].im)*t[5].re + 0x40000000 >> 31); + t[2].im = (int32_t)(mtmp[10] - ((int64_t)tab[1].im)*t[1].re + 0x40000000 >> 31); + t[4].im = (int32_t)(mtmp[11] - ((int64_t)tab[0].im)*t[3].re + 0x40000000 >> 31); +#else + z[0].re = tab[0].re*t[0].re - tab[2].re*t[4].re - tab[1].re*t[2].re; + z[1].re = tab[0].re*t[4].re - tab[1].re*t[0].re - tab[2].re*t[2].re; + z[2].re = tab[0].re*t[2].re - tab[2].re*t[0].re - tab[1].re*t[4].re; + z[0].im = tab[0].re*t[0].im - tab[1].re*t[2].im - tab[2].re*t[4].im; + z[1].im = tab[0].re*t[4].im - tab[1].re*t[0].im - tab[2].re*t[2].im; + z[2].im = tab[0].re*t[2].im - tab[2].re*t[0].im - tab[1].re*t[4].im; + + /* It's possible to do t[4].re and t[0].im with 2 multiplies only by + * multiplying the sum of all with the average of the twiddles */ + + t[0].re = tab[2].im*t[1].im + tab[1].im*t[5].im - tab[0].im*t[3].im; + t[2].re = tab[0].im*t[5].im + tab[2].im*t[3].im - tab[1].im*t[1].im; + t[4].re = tab[2].im*t[5].im + tab[1].im*t[3].im + tab[0].im*t[1].im; + t[0].im = tab[0].im*t[1].re + tab[1].im*t[3].re + tab[2].im*t[5].re; + t[2].im = tab[2].im*t[3].re + tab[0].im*t[5].re - tab[1].im*t[1].re; + t[4].im = tab[2].im*t[1].re + tab[1].im*t[5].re - tab[0].im*t[3].re; +#endif + + BF(t[1].re, z[0].re, z[0].re, t[4].re); + BF(t[3].re, z[1].re, z[1].re, t[2].re); + BF(t[5].re, z[2].re, z[2].re, t[0].re); + BF(t[1].im, z[0].im, z[0].im, t[0].im); + BF(t[3].im, z[1].im, z[1].im, t[2].im); + BF(t[5].im, z[2].im, z[2].im, t[4].im); + + out[1*stride].re = dc.re + z[0].re; + out[1*stride].im = dc.im + t[1].im; + out[2*stride].re = dc.re + t[3].re; + out[2*stride].im = dc.im + z[1].im; + out[3*stride].re = dc.re + z[2].re; + out[3*stride].im = dc.im + t[5].im; + out[4*stride].re = dc.re + t[5].re; + out[4*stride].im = dc.im + z[2].im; + out[5*stride].re = dc.re + z[1].re; + out[5*stride].im = dc.im + t[3].im; + out[6*stride].re = dc.re + t[1].re; + out[6*stride].im = dc.im + z[0].im; +} + +static av_always_inline void fft9(TXComplex *out, TXComplex *in, + ptrdiff_t stride) +{ + const TXComplex *tab = (const TXComplex *)TX_TAB(ff_tx_tab_9); + TXComplex dc, t[16], w[4], x[5], y[5], z[2]; +#ifdef TX_INT32 + int64_t mtmp[12]; +#endif + + dc = in[0]; + BF(t[1].re, t[0].re, in[1].re, in[8].re); + BF(t[1].im, t[0].im, in[1].im, in[8].im); + BF(t[3].re, t[2].re, in[2].re, in[7].re); + BF(t[3].im, t[2].im, in[2].im, in[7].im); + BF(t[5].re, t[4].re, in[3].re, in[6].re); + BF(t[5].im, t[4].im, in[3].im, in[6].im); + BF(t[7].re, t[6].re, in[4].re, in[5].re); + BF(t[7].im, t[6].im, in[4].im, in[5].im); + + w[0].re = t[0].re - t[6].re; + w[0].im = t[0].im - t[6].im; + w[1].re = t[2].re - t[6].re; + w[1].im = t[2].im - t[6].im; + w[2].re = t[1].re - t[7].re; + w[2].im = t[1].im - t[7].im; + w[3].re = t[3].re + t[7].re; + w[3].im = t[3].im + t[7].im; + + z[0].re = dc.re + t[4].re; + z[0].im = dc.im + t[4].im; + + z[1].re = t[0].re + t[2].re + t[6].re; + z[1].im = t[0].im + t[2].im + t[6].im; + + out[0*stride].re = z[0].re + z[1].re; + out[0*stride].im = z[0].im + z[1].im; + +#ifdef TX_INT32 + mtmp[0] = t[1].re - t[3].re + t[7].re; + mtmp[1] = t[1].im - t[3].im + t[7].im; + + y[3].re = (int32_t)(((int64_t)tab[0].im)*mtmp[0] + 0x40000000 >> 31); + y[3].im = (int32_t)(((int64_t)tab[0].im)*mtmp[1] + 0x40000000 >> 31); + + mtmp[0] = (int32_t)(((int64_t)tab[0].re)*z[1].re + 0x40000000 >> 31); + mtmp[1] = (int32_t)(((int64_t)tab[0].re)*z[1].im + 0x40000000 >> 31); + mtmp[2] = (int32_t)(((int64_t)tab[0].re)*t[4].re + 0x40000000 >> 31); + mtmp[3] = (int32_t)(((int64_t)tab[0].re)*t[4].im + 0x40000000 >> 31); + + x[3].re = z[0].re + (int32_t)mtmp[0]; + x[3].im = z[0].im + (int32_t)mtmp[1]; + z[0].re = in[0].re + (int32_t)mtmp[2]; + z[0].im = in[0].im + (int32_t)mtmp[3]; + + mtmp[0] = ((int64_t)tab[1].re)*w[0].re; + mtmp[1] = ((int64_t)tab[1].re)*w[0].im; + mtmp[2] = ((int64_t)tab[2].im)*w[0].re; + mtmp[3] = ((int64_t)tab[2].im)*w[0].im; + mtmp[4] = ((int64_t)tab[1].im)*w[2].re; + mtmp[5] = ((int64_t)tab[1].im)*w[2].im; + mtmp[6] = ((int64_t)tab[2].re)*w[2].re; + mtmp[7] = ((int64_t)tab[2].re)*w[2].im; + + x[1].re = (int32_t)(mtmp[0] + ((int64_t)tab[2].im)*w[1].re + 0x40000000 >> 31); + x[1].im = (int32_t)(mtmp[1] + ((int64_t)tab[2].im)*w[1].im + 0x40000000 >> 31); + x[2].re = (int32_t)(mtmp[2] - ((int64_t)tab[3].re)*w[1].re + 0x40000000 >> 31); + x[2].im = (int32_t)(mtmp[3] - ((int64_t)tab[3].re)*w[1].im + 0x40000000 >> 31); + y[1].re = (int32_t)(mtmp[4] + ((int64_t)tab[2].re)*w[3].re + 0x40000000 >> 31); + y[1].im = (int32_t)(mtmp[5] + ((int64_t)tab[2].re)*w[3].im + 0x40000000 >> 31); + y[2].re = (int32_t)(mtmp[6] - ((int64_t)tab[3].im)*w[3].re + 0x40000000 >> 31); + y[2].im = (int32_t)(mtmp[7] - ((int64_t)tab[3].im)*w[3].im + 0x40000000 >> 31); + + y[0].re = (int32_t)(((int64_t)tab[0].im)*t[5].re + 0x40000000 >> 31); + y[0].im = (int32_t)(((int64_t)tab[0].im)*t[5].im + 0x40000000 >> 31); + +#else + y[3].re = tab[0].im*(t[1].re - t[3].re + t[7].re); + y[3].im = tab[0].im*(t[1].im - t[3].im + t[7].im); + + x[3].re = z[0].re + tab[0].re*z[1].re; + x[3].im = z[0].im + tab[0].re*z[1].im; + z[0].re = dc.re + tab[0].re*t[4].re; + z[0].im = dc.im + tab[0].re*t[4].im; + + x[1].re = tab[1].re*w[0].re + tab[2].im*w[1].re; + x[1].im = tab[1].re*w[0].im + tab[2].im*w[1].im; + x[2].re = tab[2].im*w[0].re - tab[3].re*w[1].re; + x[2].im = tab[2].im*w[0].im - tab[3].re*w[1].im; + y[1].re = tab[1].im*w[2].re + tab[2].re*w[3].re; + y[1].im = tab[1].im*w[2].im + tab[2].re*w[3].im; + y[2].re = tab[2].re*w[2].re - tab[3].im*w[3].re; + y[2].im = tab[2].re*w[2].im - tab[3].im*w[3].im; + + y[0].re = tab[0].im*t[5].re; + y[0].im = tab[0].im*t[5].im; +#endif + + x[4].re = x[1].re + x[2].re; + x[4].im = x[1].im + x[2].im; + + y[4].re = y[1].re - y[2].re; + y[4].im = y[1].im - y[2].im; + x[1].re = z[0].re + x[1].re; + x[1].im = z[0].im + x[1].im; + y[1].re = y[0].re + y[1].re; + y[1].im = y[0].im + y[1].im; + x[2].re = z[0].re + x[2].re; + x[2].im = z[0].im + x[2].im; + y[2].re = y[2].re - y[0].re; + y[2].im = y[2].im - y[0].im; + x[4].re = z[0].re - x[4].re; + x[4].im = z[0].im - x[4].im; + y[4].re = y[0].re - y[4].re; + y[4].im = y[0].im - y[4].im; + + out[1*stride] = (TXComplex){ x[1].re + y[1].im, x[1].im - y[1].re }; + out[2*stride] = (TXComplex){ x[2].re + y[2].im, x[2].im - y[2].re }; + out[3*stride] = (TXComplex){ x[3].re + y[3].im, x[3].im - y[3].re }; + out[4*stride] = (TXComplex){ x[4].re + y[4].im, x[4].im - y[4].re }; + out[5*stride] = (TXComplex){ x[4].re - y[4].im, x[4].im + y[4].re }; + out[6*stride] = (TXComplex){ x[3].re - y[3].im, x[3].im + y[3].re }; + out[7*stride] = (TXComplex){ x[2].re - y[2].im, x[2].im + y[2].re }; + out[8*stride] = (TXComplex){ x[1].re - y[1].im, x[1].im + y[1].re }; +} + +static av_always_inline void fft15(TXComplex *out, TXComplex *in, + ptrdiff_t stride) +{ + TXComplex tmp[15]; + + for (int i = 0; i < 5; i++) + fft3(tmp + i, in + i*3, 5); + + fft5_m1(out, tmp + 0, stride); + fft5_m2(out, tmp + 5, stride); + fft5_m3(out, tmp + 10, stride); +} + +static av_cold int TX_NAME(ff_tx_fft_factor_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret = 0; + TX_TAB(ff_tx_init_tabs)(len); + + if (len == 15) + ret = ff_tx_gen_pfa_input_map(s, opts, 3, 5); + else if (flags & FF_TX_PRESHUFFLE) + ret = ff_tx_gen_default_map(s, opts); + + return ret; +} + +#define DECL_FACTOR_S(n) \ +static void TX_NAME(ff_tx_fft##n)(AVTXContext *s, void *dst, \ + void *src, ptrdiff_t stride) \ +{ \ + fft##n((TXComplex *)dst, (TXComplex *)src, stride / sizeof(TXComplex)); \ +} \ +static const FFTXCodelet TX_NAME(ff_tx_fft##n##_ns_def) = { \ + .name = TX_NAME_STR("fft" #n "_ns"), \ + .function = TX_NAME(ff_tx_fft##n), \ + .type = TX_TYPE(FFT), \ + .flags = AV_TX_INPLACE | FF_TX_OUT_OF_PLACE | \ + AV_TX_UNALIGNED | FF_TX_PRESHUFFLE, \ + .factors[0] = n, \ + .nb_factors = 1, \ + .min_len = n, \ + .max_len = n, \ + .init = TX_NAME(ff_tx_fft_factor_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +#define DECL_FACTOR_F(n) \ +DECL_FACTOR_S(n) \ +static const FFTXCodelet TX_NAME(ff_tx_fft##n##_fwd_def) = { \ + .name = TX_NAME_STR("fft" #n "_fwd"), \ + .function = TX_NAME(ff_tx_fft##n), \ + .type = TX_TYPE(FFT), \ + .flags = AV_TX_INPLACE | FF_TX_OUT_OF_PLACE | \ + AV_TX_UNALIGNED | FF_TX_FORWARD_ONLY, \ + .factors[0] = n, \ + .nb_factors = 1, \ + .min_len = n, \ + .max_len = n, \ + .init = TX_NAME(ff_tx_fft_factor_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +DECL_FACTOR_F(3) +DECL_FACTOR_F(5) +DECL_FACTOR_F(7) +DECL_FACTOR_F(9) +DECL_FACTOR_S(15) + +#define BUTTERFLIES(a0, a1, a2, a3) \ + do { \ + r0=a0.re; \ + i0=a0.im; \ + r1=a1.re; \ + i1=a1.im; \ + BF(t3, t5, t5, t1); \ + BF(a2.re, a0.re, r0, t5); \ + BF(a3.im, a1.im, i1, t3); \ + BF(t4, t6, t2, t6); \ + BF(a3.re, a1.re, r1, t4); \ + BF(a2.im, a0.im, i0, t6); \ + } while (0) + +#define TRANSFORM(a0, a1, a2, a3, wre, wim) \ + do { \ + CMUL(t1, t2, a2.re, a2.im, wre, -wim); \ + CMUL(t5, t6, a3.re, a3.im, wre, wim); \ + BUTTERFLIES(a0, a1, a2, a3); \ + } while (0) + +/* z[0...8n-1], w[1...2n-1] */ +static inline void TX_NAME(ff_tx_fft_sr_combine)(TXComplex *z, + const TXSample *cos, int len) +{ + int o1 = 2*len; + int o2 = 4*len; + int o3 = 6*len; + const TXSample *wim = cos + o1 - 7; + TXUSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1; + + for (int i = 0; i < len; i += 4) { + TRANSFORM(z[0], z[o1 + 0], z[o2 + 0], z[o3 + 0], cos[0], wim[7]); + TRANSFORM(z[2], z[o1 + 2], z[o2 + 2], z[o3 + 2], cos[2], wim[5]); + TRANSFORM(z[4], z[o1 + 4], z[o2 + 4], z[o3 + 4], cos[4], wim[3]); + TRANSFORM(z[6], z[o1 + 6], z[o2 + 6], z[o3 + 6], cos[6], wim[1]); + + TRANSFORM(z[1], z[o1 + 1], z[o2 + 1], z[o3 + 1], cos[1], wim[6]); + TRANSFORM(z[3], z[o1 + 3], z[o2 + 3], z[o3 + 3], cos[3], wim[4]); + TRANSFORM(z[5], z[o1 + 5], z[o2 + 5], z[o3 + 5], cos[5], wim[2]); + TRANSFORM(z[7], z[o1 + 7], z[o2 + 7], z[o3 + 7], cos[7], wim[0]); + + z += 2*4; + cos += 2*4; + wim -= 2*4; + } +} + +static av_cold int TX_NAME(ff_tx_fft_sr_codelet_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + TX_TAB(ff_tx_init_tabs)(len); + return ff_tx_gen_ptwo_revtab(s, opts); +} + +#define DECL_SR_CODELET_DEF(n) \ +static const FFTXCodelet TX_NAME(ff_tx_fft##n##_ns_def) = { \ + .name = TX_NAME_STR("fft" #n "_ns"), \ + .function = TX_NAME(ff_tx_fft##n##_ns), \ + .type = TX_TYPE(FFT), \ + .flags = FF_TX_OUT_OF_PLACE | AV_TX_INPLACE | \ + AV_TX_UNALIGNED | FF_TX_PRESHUFFLE, \ + .factors[0] = 2, \ + .nb_factors = 1, \ + .min_len = n, \ + .max_len = n, \ + .init = TX_NAME(ff_tx_fft_sr_codelet_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +#define DECL_SR_CODELET(n, n2, n4) \ +static void TX_NAME(ff_tx_fft##n##_ns)(AVTXContext *s, void *_dst, \ + void *_src, ptrdiff_t stride) \ +{ \ + TXComplex *src = _src; \ + TXComplex *dst = _dst; \ + const TXSample *cos = TX_TAB(ff_tx_tab_##n); \ + \ + TX_NAME(ff_tx_fft##n2##_ns)(s, dst, src, stride); \ + TX_NAME(ff_tx_fft##n4##_ns)(s, dst + n4*2, src + n4*2, stride); \ + TX_NAME(ff_tx_fft##n4##_ns)(s, dst + n4*3, src + n4*3, stride); \ + TX_NAME(ff_tx_fft_sr_combine)(dst, cos, n4 >> 1); \ +} \ + \ +DECL_SR_CODELET_DEF(n) + +static void TX_NAME(ff_tx_fft2_ns)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + TXComplex tmp; + + BF(tmp.re, dst[0].re, src[0].re, src[1].re); + BF(tmp.im, dst[0].im, src[0].im, src[1].im); + dst[1] = tmp; +} + +static void TX_NAME(ff_tx_fft4_ns)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + TXSample t1, t2, t3, t4, t5, t6, t7, t8; + + BF(t3, t1, src[0].re, src[1].re); + BF(t8, t6, src[3].re, src[2].re); + BF(dst[2].re, dst[0].re, t1, t6); + BF(t4, t2, src[0].im, src[1].im); + BF(t7, t5, src[2].im, src[3].im); + BF(dst[3].im, dst[1].im, t4, t8); + BF(dst[3].re, dst[1].re, t3, t7); + BF(dst[2].im, dst[0].im, t2, t5); +} + +static void TX_NAME(ff_tx_fft8_ns)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + TXUSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1; + const TXSample cos = TX_TAB(ff_tx_tab_8)[1]; + + TX_NAME(ff_tx_fft4_ns)(s, dst, src, stride); + + BF(t1, dst[5].re, src[4].re, -src[5].re); + BF(t2, dst[5].im, src[4].im, -src[5].im); + BF(t5, dst[7].re, src[6].re, -src[7].re); + BF(t6, dst[7].im, src[6].im, -src[7].im); + + BUTTERFLIES(dst[0], dst[2], dst[4], dst[6]); + TRANSFORM(dst[1], dst[3], dst[5], dst[7], cos, cos); +} + +static void TX_NAME(ff_tx_fft16_ns)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + const TXSample *cos = TX_TAB(ff_tx_tab_16); + + TXUSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1; + TXSample cos_16_1 = cos[1]; + TXSample cos_16_2 = cos[2]; + TXSample cos_16_3 = cos[3]; + + TX_NAME(ff_tx_fft8_ns)(s, dst + 0, src + 0, stride); + TX_NAME(ff_tx_fft4_ns)(s, dst + 8, src + 8, stride); + TX_NAME(ff_tx_fft4_ns)(s, dst + 12, src + 12, stride); + + t1 = dst[ 8].re; + t2 = dst[ 8].im; + t5 = dst[12].re; + t6 = dst[12].im; + BUTTERFLIES(dst[0], dst[4], dst[8], dst[12]); + + TRANSFORM(dst[ 2], dst[ 6], dst[10], dst[14], cos_16_2, cos_16_2); + TRANSFORM(dst[ 1], dst[ 5], dst[ 9], dst[13], cos_16_1, cos_16_3); + TRANSFORM(dst[ 3], dst[ 7], dst[11], dst[15], cos_16_3, cos_16_1); +} + +DECL_SR_CODELET_DEF(2) +DECL_SR_CODELET_DEF(4) +DECL_SR_CODELET_DEF(8) +DECL_SR_CODELET_DEF(16) +DECL_SR_CODELET(32,16,8) +DECL_SR_CODELET(64,32,16) +DECL_SR_CODELET(128,64,32) +DECL_SR_CODELET(256,128,64) +DECL_SR_CODELET(512,256,128) +DECL_SR_CODELET(1024,512,256) +DECL_SR_CODELET(2048,1024,512) +DECL_SR_CODELET(4096,2048,1024) +DECL_SR_CODELET(8192,4096,2048) +DECL_SR_CODELET(16384,8192,4096) +DECL_SR_CODELET(32768,16384,8192) +DECL_SR_CODELET(65536,32768,16384) +DECL_SR_CODELET(131072,65536,32768) +DECL_SR_CODELET(262144,131072,65536) +DECL_SR_CODELET(524288,262144,131072) +DECL_SR_CODELET(1048576,524288,262144) +DECL_SR_CODELET(2097152,1048576,524288) + +static av_cold int TX_NAME(ff_tx_fft_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + int is_inplace = !!(flags & AV_TX_INPLACE); + FFTXCodeletOptions sub_opts = { + .map_dir = is_inplace ? FF_TX_MAP_SCATTER : FF_TX_MAP_GATHER, + }; + + flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */ + flags |= AV_TX_INPLACE; /* in-place */ + flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */ + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len, inv, scale))) + return ret; + + if (is_inplace && (ret = ff_tx_gen_inplace_map(s, len))) + return ret; + + return 0; +} + +static av_cold int TX_NAME(ff_tx_fft_inplace_small_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + if (!(s->tmp = av_malloc(len*sizeof(*s->tmp)))) + return AVERROR(ENOMEM); + flags &= ~AV_TX_INPLACE; + return TX_NAME(ff_tx_fft_init)(s, cd, flags, opts, len, inv, scale); +} + +static void TX_NAME(ff_tx_fft)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst1 = s->flags & AV_TX_INPLACE ? s->tmp : _dst; + TXComplex *dst2 = _dst; + int *map = s->sub[0].map; + int len = s->len; + + /* Compilers can't vectorize this anyway without assuming AVX2, which they + * generally don't, at least without -march=native -mtune=native */ + for (int i = 0; i < len; i++) + dst1[i] = src[map[i]]; + + s->fn[0](&s->sub[0], dst2, dst1, stride); +} + +static void TX_NAME(ff_tx_fft_inplace)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + TXComplex tmp; + const int *map = s->sub->map; + const int *inplace_idx = s->map; + int src_idx, dst_idx; + + src_idx = *inplace_idx++; + do { + tmp = src[src_idx]; + dst_idx = map[src_idx]; + do { + FFSWAP(TXComplex, tmp, src[dst_idx]); + dst_idx = map[dst_idx]; + } while (dst_idx != src_idx); /* Can be > as well, but was less predictable */ + src[dst_idx] = tmp; + } while ((src_idx = *inplace_idx++)); + + s->fn[0](&s->sub[0], dst, src, stride); +} + +static const FFTXCodelet TX_NAME(ff_tx_fft_def) = { + .name = TX_NAME_STR("fft"), + .function = TX_NAME(ff_tx_fft), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE, + .factors[0] = TX_FACTOR_ANY, + .nb_factors = 1, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_fft_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static const FFTXCodelet TX_NAME(ff_tx_fft_inplace_small_def) = { + .name = TX_NAME_STR("fft_inplace_small"), + .function = TX_NAME(ff_tx_fft), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | AV_TX_INPLACE, + .factors[0] = TX_FACTOR_ANY, + .nb_factors = 1, + .min_len = 2, + .max_len = 65536, + .init = TX_NAME(ff_tx_fft_inplace_small_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE - 256, +}; + +static const FFTXCodelet TX_NAME(ff_tx_fft_inplace_def) = { + .name = TX_NAME_STR("fft_inplace"), + .function = TX_NAME(ff_tx_fft_inplace), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | AV_TX_INPLACE, + .factors[0] = TX_FACTOR_ANY, + .nb_factors = 1, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_fft_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE - 512, +}; + +static av_cold int TX_NAME(ff_tx_fft_init_naive_small)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + const double phase = s->inv ? 2.0*M_PI/len : -2.0*M_PI/len; + + if (!(s->exp = av_malloc(len*len*sizeof(*s->exp)))) + return AVERROR(ENOMEM); + + for (int i = 0; i < len; i++) { + for (int j = 0; j < len; j++) { + const double factor = phase*i*j; + s->exp[i*j] = (TXComplex){ + RESCALE(cos(factor)), + RESCALE(sin(factor)), + }; + } + } + + return 0; +} + +static void TX_NAME(ff_tx_fft_naive)(AVTXContext *s, void *_dst, void *_src, + ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + const int n = s->len; + double phase = s->inv ? 2.0*M_PI/n : -2.0*M_PI/n; + + stride /= sizeof(*dst); + + for (int i = 0; i < n; i++) { + TXComplex tmp = { 0 }; + for (int j = 0; j < n; j++) { + const double factor = phase*i*j; + const TXComplex mult = { + RESCALE(cos(factor)), + RESCALE(sin(factor)), + }; + TXComplex res; + CMUL3(res, src[j], mult); + tmp.re += res.re; + tmp.im += res.im; + } + dst[i*stride] = tmp; + } +} + +static void TX_NAME(ff_tx_fft_naive_small)(AVTXContext *s, void *_dst, void *_src, + ptrdiff_t stride) +{ + TXComplex *src = _src; + TXComplex *dst = _dst; + const int n = s->len; + + stride /= sizeof(*dst); + + for (int i = 0; i < n; i++) { + TXComplex tmp = { 0 }; + for (int j = 0; j < n; j++) { + TXComplex res; + const TXComplex mult = s->exp[i*j]; + CMUL3(res, src[j], mult); + tmp.re += res.re; + tmp.im += res.im; + } + dst[i*stride] = tmp; + } +} + +static const FFTXCodelet TX_NAME(ff_tx_fft_naive_small_def) = { + .name = TX_NAME_STR("fft_naive_small"), + .function = TX_NAME(ff_tx_fft_naive_small), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE, + .factors[0] = TX_FACTOR_ANY, + .nb_factors = 1, + .min_len = 2, + .max_len = 1024, + .init = TX_NAME(ff_tx_fft_init_naive_small), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_MIN/2, +}; + +static const FFTXCodelet TX_NAME(ff_tx_fft_naive_def) = { + .name = TX_NAME_STR("fft_naive"), + .function = TX_NAME(ff_tx_fft_naive), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE, + .factors[0] = TX_FACTOR_ANY, + .nb_factors = 1, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = NULL, + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_MIN, +}; + +static av_cold int TX_NAME(ff_tx_fft_pfa_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret, *tmp, ps = flags & FF_TX_PRESHUFFLE; + FFTXCodeletOptions sub_opts = { .map_dir = FF_TX_MAP_GATHER }; + size_t extra_tmp_len = 0; + int len_list[TX_MAX_DECOMPOSITIONS]; + + if ((ret = ff_tx_decompose_length(len_list, TX_TYPE(FFT), len, inv)) < 0) + return ret; + + /* Two iterations to test both orderings. */ + for (int i = 0; i < ret; i++) { + int len1 = len_list[i]; + int len2 = len / len1; + + /* Our ptwo transforms don't support striding the output. */ + if (len2 & (len2 - 1)) + FFSWAP(int, len1, len2); + + ff_tx_clear_ctx(s); + + /* First transform */ + sub_opts.map_dir = FF_TX_MAP_GATHER; + flags &= ~AV_TX_INPLACE; + flags |= FF_TX_OUT_OF_PLACE; + flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */ + ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + len1, inv, scale); + + if (ret == AVERROR(ENOMEM)) { + return ret; + } else if (ret < 0) { /* Try again without a preshuffle flag */ + flags &= ~FF_TX_PRESHUFFLE; + ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + len1, inv, scale); + if (ret == AVERROR(ENOMEM)) + return ret; + else if (ret < 0) + continue; + } + + /* Second transform. */ + sub_opts.map_dir = FF_TX_MAP_SCATTER; + flags |= FF_TX_PRESHUFFLE; +retry: + flags &= ~FF_TX_OUT_OF_PLACE; + flags |= AV_TX_INPLACE; + ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + len2, inv, scale); + + if (ret == AVERROR(ENOMEM)) { + return ret; + } else if (ret < 0) { /* Try again with an out-of-place transform */ + flags |= FF_TX_OUT_OF_PLACE; + flags &= ~AV_TX_INPLACE; + ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + len2, inv, scale); + if (ret == AVERROR(ENOMEM)) { + return ret; + } else if (ret < 0) { + if (flags & FF_TX_PRESHUFFLE) { /* Retry again without a preshuf flag */ + flags &= ~FF_TX_PRESHUFFLE; + goto retry; + } else { + continue; + } + } + } + + /* Success */ + break; + } + + /* If nothing was sucessful, error out */ + if (ret < 0) + return ret; + + /* Generate PFA map */ + if ((ret = ff_tx_gen_compound_mapping(s, opts, 0, + s->sub[0].len, s->sub[1].len))) + return ret; + + if (!(s->tmp = av_malloc(len*sizeof(*s->tmp)))) + return AVERROR(ENOMEM); + + /* Flatten input map */ + tmp = (int *)s->tmp; + for (int k = 0; k < len; k += s->sub[0].len) { + memcpy(tmp, &s->map[k], s->sub[0].len*sizeof(*tmp)); + for (int i = 0; i < s->sub[0].len; i++) + s->map[k + i] = tmp[s->sub[0].map[i]]; + } + + /* Only allocate extra temporary memory if we need it */ + if (!(s->sub[1].flags & AV_TX_INPLACE)) + extra_tmp_len = len; + else if (!ps) + extra_tmp_len = s->sub[0].len; + + if (extra_tmp_len && !(s->exp = av_malloc(extra_tmp_len*sizeof(*s->exp)))) + return AVERROR(ENOMEM); + + return 0; +} + +static void TX_NAME(ff_tx_fft_pfa)(AVTXContext *s, void *_out, + void *_in, ptrdiff_t stride) +{ + const int n = s->sub[0].len, m = s->sub[1].len, l = s->len; + const int *in_map = s->map, *out_map = in_map + l; + const int *sub_map = s->sub[1].map; + TXComplex *tmp1 = s->sub[1].flags & AV_TX_INPLACE ? s->tmp : s->exp; + TXComplex *in = _in, *out = _out; + + stride /= sizeof(*out); + + for (int i = 0; i < m; i++) { + for (int j = 0; j < n; j++) + s->exp[j] = in[in_map[i*n + j]]; + s->fn[0](&s->sub[0], &s->tmp[sub_map[i]], s->exp, m*sizeof(TXComplex)); + } + + for (int i = 0; i < n; i++) + s->fn[1](&s->sub[1], &tmp1[m*i], &s->tmp[m*i], sizeof(TXComplex)); + + for (int i = 0; i < l; i++) + out[i*stride] = tmp1[out_map[i]]; +} + +static void TX_NAME(ff_tx_fft_pfa_ns)(AVTXContext *s, void *_out, + void *_in, ptrdiff_t stride) +{ + const int n = s->sub[0].len, m = s->sub[1].len, l = s->len; + const int *in_map = s->map, *out_map = in_map + l; + const int *sub_map = s->sub[1].map; + TXComplex *tmp1 = s->sub[1].flags & AV_TX_INPLACE ? s->tmp : s->exp; + TXComplex *in = _in, *out = _out; + + stride /= sizeof(*out); + + for (int i = 0; i < m; i++) + s->fn[0](&s->sub[0], &s->tmp[sub_map[i]], &in[i*n], m*sizeof(TXComplex)); + + for (int i = 0; i < n; i++) + s->fn[1](&s->sub[1], &tmp1[m*i], &s->tmp[m*i], sizeof(TXComplex)); + + for (int i = 0; i < l; i++) + out[i*stride] = tmp1[out_map[i]]; +} + +static const FFTXCodelet TX_NAME(ff_tx_fft_pfa_def) = { + .name = TX_NAME_STR("fft_pfa"), + .function = TX_NAME(ff_tx_fft_pfa), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE, + .factors = { 7, 5, 3, 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2*3, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_fft_pfa_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static const FFTXCodelet TX_NAME(ff_tx_fft_pfa_ns_def) = { + .name = TX_NAME_STR("fft_pfa_ns"), + .function = TX_NAME(ff_tx_fft_pfa_ns), + .type = TX_TYPE(FFT), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE | + FF_TX_PRESHUFFLE, + .factors = { 7, 5, 3, 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2*3, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_fft_pfa_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static av_cold int TX_NAME(ff_tx_mdct_naive_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + return 0; +} + +static void TX_NAME(ff_tx_mdct_naive_fwd)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *src = _src; + TXSample *dst = _dst; + double scale = s->scale_d; + int len = s->len; + const double phase = M_PI/(4.0*len); + + stride /= sizeof(*dst); + + for (int i = 0; i < len; i++) { + double sum = 0.0; + for (int j = 0; j < len*2; j++) { + int a = (2*j + 1 + len) * (2*i + 1); + sum += UNSCALE(src[j]) * cos(a * phase); + } + dst[i*stride] = RESCALE(sum*scale); + } +} + +static void TX_NAME(ff_tx_mdct_naive_inv)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *src = _src; + TXSample *dst = _dst; + double scale = s->scale_d; + int len = s->len >> 1; + int len2 = len*2; + const double phase = M_PI/(4.0*len2); + + stride /= sizeof(*src); + + for (int i = 0; i < len; i++) { + double sum_d = 0.0; + double sum_u = 0.0; + double i_d = phase * (4*len - 2*i - 1); + double i_u = phase * (3*len2 + 2*i + 1); + for (int j = 0; j < len2; j++) { + double a = (2 * j + 1); + double a_d = cos(a * i_d); + double a_u = cos(a * i_u); + double val = UNSCALE(src[j*stride]); + sum_d += a_d * val; + sum_u += a_u * val; + } + dst[i + 0] = RESCALE( sum_d*scale); + dst[i + len] = RESCALE(-sum_u*scale); + } +} + +static const FFTXCodelet TX_NAME(ff_tx_mdct_naive_fwd_def) = { + .name = TX_NAME_STR("mdct_naive_fwd"), + .function = TX_NAME(ff_tx_mdct_naive_fwd), + .type = TX_TYPE(MDCT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, + .factors = { 2, TX_FACTOR_ANY }, /* MDCTs need an even length */ + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_mdct_naive_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_MIN, +}; + +static const FFTXCodelet TX_NAME(ff_tx_mdct_naive_inv_def) = { + .name = TX_NAME_STR("mdct_naive_inv"), + .function = TX_NAME(ff_tx_mdct_naive_inv), + .type = TX_TYPE(MDCT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_mdct_naive_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_MIN, +}; + +static av_cold int TX_NAME(ff_tx_mdct_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + FFTXCodeletOptions sub_opts = { + .map_dir = !inv ? FF_TX_MAP_SCATTER : FF_TX_MAP_GATHER, + }; + + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + + flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */ + flags |= AV_TX_INPLACE; /* in-place */ + flags |= FF_TX_PRESHUFFLE; /* First try with an in-place transform */ + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len >> 1, + inv, scale))) { + flags &= ~FF_TX_PRESHUFFLE; /* Now try with a generic FFT */ + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len >> 1, + inv, scale))) + return ret; + } + + s->map = av_malloc((len >> 1)*sizeof(*s->map)); + if (!s->map) + return AVERROR(ENOMEM); + + /* If we need to preshuffle copy the map from the subcontext */ + if (s->sub[0].flags & FF_TX_PRESHUFFLE) { + memcpy(s->map, s->sub->map, (len >> 1)*sizeof(*s->map)); + } else { + for (int i = 0; i < len >> 1; i++) + s->map[i] = i; + } + + if ((ret = TX_TAB(ff_tx_mdct_gen_exp)(s, inv ? s->map : NULL))) + return ret; + + /* Saves a multiply in a hot path. */ + if (inv) + for (int i = 0; i < (s->len >> 1); i++) + s->map[i] <<= 1; + + return 0; +} + +static void TX_NAME(ff_tx_mdct_fwd)(AVTXContext *s, void *_dst, void *_src, + ptrdiff_t stride) +{ + TXSample *src = _src, *dst = _dst; + TXComplex *exp = s->exp, tmp, *z = _dst; + const int len2 = s->len >> 1; + const int len4 = s->len >> 2; + const int len3 = len2 * 3; + const int *sub_map = s->map; + + stride /= sizeof(*dst); + + for (int i = 0; i < len2; i++) { /* Folding and pre-reindexing */ + const int k = 2*i; + const int idx = sub_map[i]; + if (k < len2) { + tmp.re = FOLD(-src[ len2 + k], src[1*len2 - 1 - k]); + tmp.im = FOLD(-src[ len3 + k], -src[1*len3 - 1 - k]); + } else { + tmp.re = FOLD(-src[ len2 + k], -src[5*len2 - 1 - k]); + tmp.im = FOLD( src[-len2 + k], -src[1*len3 - 1 - k]); + } + CMUL(z[idx].im, z[idx].re, tmp.re, tmp.im, exp[i].re, exp[i].im); + } + + s->fn[0](&s->sub[0], z, z, sizeof(TXComplex)); + + for (int i = 0; i < len4; i++) { + const int i0 = len4 + i, i1 = len4 - i - 1; + TXComplex src1 = { z[i1].re, z[i1].im }; + TXComplex src0 = { z[i0].re, z[i0].im }; + + CMUL(dst[2*i1*stride + stride], dst[2*i0*stride], src0.re, src0.im, + exp[i0].im, exp[i0].re); + CMUL(dst[2*i0*stride + stride], dst[2*i1*stride], src1.re, src1.im, + exp[i1].im, exp[i1].re); + } +} + +static void TX_NAME(ff_tx_mdct_inv)(AVTXContext *s, void *_dst, void *_src, + ptrdiff_t stride) +{ + TXComplex *z = _dst, *exp = s->exp; + const TXSample *src = _src, *in1, *in2; + const int len2 = s->len >> 1; + const int len4 = s->len >> 2; + const int *sub_map = s->map; + + stride /= sizeof(*src); + in1 = src; + in2 = src + ((len2*2) - 1) * stride; + + for (int i = 0; i < len2; i++) { + int k = sub_map[i]; + TXComplex tmp = { in2[-k*stride], in1[k*stride] }; + CMUL3(z[i], tmp, exp[i]); + } + + s->fn[0](&s->sub[0], z, z, sizeof(TXComplex)); + + exp += len2; + for (int i = 0; i < len4; i++) { + const int i0 = len4 + i, i1 = len4 - i - 1; + TXComplex src1 = { z[i1].im, z[i1].re }; + TXComplex src0 = { z[i0].im, z[i0].re }; + + CMUL(z[i1].re, z[i0].im, src1.re, src1.im, exp[i1].im, exp[i1].re); + CMUL(z[i0].re, z[i1].im, src0.re, src0.im, exp[i0].im, exp[i0].re); + } +} + +static const FFTXCodelet TX_NAME(ff_tx_mdct_fwd_def) = { + .name = TX_NAME_STR("mdct_fwd"), + .function = TX_NAME(ff_tx_mdct_fwd), + .type = TX_TYPE(MDCT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_mdct_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static const FFTXCodelet TX_NAME(ff_tx_mdct_inv_def) = { + .name = TX_NAME_STR("mdct_inv"), + .function = TX_NAME(ff_tx_mdct_inv), + .type = TX_TYPE(MDCT), + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_mdct_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static av_cold int TX_NAME(ff_tx_mdct_inv_full_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + + flags &= ~AV_TX_FULL_IMDCT; + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(MDCT), flags, NULL, len, 1, scale))) + return ret; + + return 0; +} + +static void TX_NAME(ff_tx_mdct_inv_full)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + int len = s->len << 1; + int len2 = len >> 1; + int len4 = len >> 2; + TXSample *dst = _dst; + + s->fn[0](&s->sub[0], dst + len4, _src, stride); + + stride /= sizeof(*dst); + + for (int i = 0; i < len4; i++) { + dst[ i*stride] = -dst[(len2 - i - 1)*stride]; + dst[(len - i - 1)*stride] = dst[(len2 + i + 0)*stride]; + } +} + +static const FFTXCodelet TX_NAME(ff_tx_mdct_inv_full_def) = { + .name = TX_NAME_STR("mdct_inv_full"), + .function = TX_NAME(ff_tx_mdct_inv_full), + .type = TX_TYPE(MDCT), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | + FF_TX_OUT_OF_PLACE | AV_TX_FULL_IMDCT, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_mdct_inv_full_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static av_cold int TX_NAME(ff_tx_mdct_pfa_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret, sub_len; + FFTXCodeletOptions sub_opts = { .map_dir = FF_TX_MAP_SCATTER }; + + len >>= 1; + sub_len = len / cd->factors[0]; + + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + + flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */ + flags |= AV_TX_INPLACE; /* in-place */ + flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */ + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + sub_len, inv, scale))) + return ret; + + if ((ret = ff_tx_gen_compound_mapping(s, opts, s->inv, cd->factors[0], sub_len))) + return ret; + + /* Our 15-point transform is also a compound one, so embed its input map */ + if (cd->factors[0] == 15) + TX_EMBED_INPUT_PFA_MAP(s->map, len, 3, 5); + + if ((ret = TX_TAB(ff_tx_mdct_gen_exp)(s, inv ? s->map : NULL))) + return ret; + + /* Saves multiplies in loops. */ + for (int i = 0; i < len; i++) + s->map[i] <<= 1; + + if (!(s->tmp = av_malloc(len*sizeof(*s->tmp)))) + return AVERROR(ENOMEM); + + TX_TAB(ff_tx_init_tabs)(len / sub_len); + + return 0; +} + +#define DECL_COMP_IMDCT(N) \ +static void TX_NAME(ff_tx_mdct_pfa_##N##xM_inv)(AVTXContext *s, void *_dst, \ + void *_src, ptrdiff_t stride) \ +{ \ + TXComplex fft##N##in[N]; \ + TXComplex *z = _dst, *exp = s->exp; \ + const TXSample *src = _src, *in1, *in2; \ + const int len4 = s->len >> 2; \ + const int len2 = s->len >> 1; \ + const int m = s->sub->len; \ + const int *in_map = s->map, *out_map = in_map + N*m; \ + const int *sub_map = s->sub->map; \ + \ + stride /= sizeof(*src); /* To convert it from bytes */ \ + in1 = src; \ + in2 = src + ((N*m*2) - 1) * stride; \ + \ + for (int i = 0; i < len2; i += N) { \ + for (int j = 0; j < N; j++) { \ + const int k = in_map[j]; \ + TXComplex tmp = { in2[-k*stride], in1[k*stride] }; \ + CMUL3(fft##N##in[j], tmp, exp[j]); \ + } \ + fft##N(s->tmp + *(sub_map++), fft##N##in, m); \ + exp += N; \ + in_map += N; \ + } \ + \ + for (int i = 0; i < N; i++) \ + s->fn[0](&s->sub[0], s->tmp + m*i, s->tmp + m*i, sizeof(TXComplex)); \ + \ + for (int i = 0; i < len4; i++) { \ + const int i0 = len4 + i, i1 = len4 - i - 1; \ + const int s0 = out_map[i0], s1 = out_map[i1]; \ + TXComplex src1 = { s->tmp[s1].im, s->tmp[s1].re }; \ + TXComplex src0 = { s->tmp[s0].im, s->tmp[s0].re }; \ + \ + CMUL(z[i1].re, z[i0].im, src1.re, src1.im, exp[i1].im, exp[i1].re); \ + CMUL(z[i0].re, z[i1].im, src0.re, src0.im, exp[i0].im, exp[i0].re); \ + } \ +} \ + \ +static const FFTXCodelet TX_NAME(ff_tx_mdct_pfa_##N##xM_inv_def) = { \ + .name = TX_NAME_STR("mdct_pfa_" #N "xM_inv"), \ + .function = TX_NAME(ff_tx_mdct_pfa_##N##xM_inv), \ + .type = TX_TYPE(MDCT), \ + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY, \ + .factors = { N, TX_FACTOR_ANY }, \ + .nb_factors = 2, \ + .min_len = N*2, \ + .max_len = TX_LEN_UNLIMITED, \ + .init = TX_NAME(ff_tx_mdct_pfa_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +DECL_COMP_IMDCT(3) +DECL_COMP_IMDCT(5) +DECL_COMP_IMDCT(7) +DECL_COMP_IMDCT(9) +DECL_COMP_IMDCT(15) + +#define DECL_COMP_MDCT(N) \ +static void TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd)(AVTXContext *s, void *_dst, \ + void *_src, ptrdiff_t stride) \ +{ \ + TXComplex fft##N##in[N]; \ + TXSample *src = _src, *dst = _dst; \ + TXComplex *exp = s->exp, tmp; \ + const int m = s->sub->len; \ + const int len4 = N*m; \ + const int len3 = len4 * 3; \ + const int len8 = s->len >> 2; \ + const int *in_map = s->map, *out_map = in_map + N*m; \ + const int *sub_map = s->sub->map; \ + \ + stride /= sizeof(*dst); \ + \ + for (int i = 0; i < m; i++) { /* Folding and pre-reindexing */ \ + for (int j = 0; j < N; j++) { \ + const int k = in_map[i*N + j]; \ + if (k < len4) { \ + tmp.re = FOLD(-src[ len4 + k], src[1*len4 - 1 - k]); \ + tmp.im = FOLD(-src[ len3 + k], -src[1*len3 - 1 - k]); \ + } else { \ + tmp.re = FOLD(-src[ len4 + k], -src[5*len4 - 1 - k]); \ + tmp.im = FOLD( src[-len4 + k], -src[1*len3 - 1 - k]); \ + } \ + CMUL(fft##N##in[j].im, fft##N##in[j].re, tmp.re, tmp.im, \ + exp[k >> 1].re, exp[k >> 1].im); \ + } \ + fft##N(s->tmp + sub_map[i], fft##N##in, m); \ + } \ + \ + for (int i = 0; i < N; i++) \ + s->fn[0](&s->sub[0], s->tmp + m*i, s->tmp + m*i, sizeof(TXComplex)); \ + \ + for (int i = 0; i < len8; i++) { \ + const int i0 = len8 + i, i1 = len8 - i - 1; \ + const int s0 = out_map[i0], s1 = out_map[i1]; \ + TXComplex src1 = { s->tmp[s1].re, s->tmp[s1].im }; \ + TXComplex src0 = { s->tmp[s0].re, s->tmp[s0].im }; \ + \ + CMUL(dst[2*i1*stride + stride], dst[2*i0*stride], src0.re, src0.im, \ + exp[i0].im, exp[i0].re); \ + CMUL(dst[2*i0*stride + stride], dst[2*i1*stride], src1.re, src1.im, \ + exp[i1].im, exp[i1].re); \ + } \ +} \ + \ +static const FFTXCodelet TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd_def) = { \ + .name = TX_NAME_STR("mdct_pfa_" #N "xM_fwd"), \ + .function = TX_NAME(ff_tx_mdct_pfa_##N##xM_fwd), \ + .type = TX_TYPE(MDCT), \ + .flags = AV_TX_UNALIGNED | FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, \ + .factors = { N, TX_FACTOR_ANY }, \ + .nb_factors = 2, \ + .min_len = N*2, \ + .max_len = TX_LEN_UNLIMITED, \ + .init = TX_NAME(ff_tx_mdct_pfa_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +DECL_COMP_MDCT(3) +DECL_COMP_MDCT(5) +DECL_COMP_MDCT(7) +DECL_COMP_MDCT(9) +DECL_COMP_MDCT(15) + +static av_cold int TX_NAME(ff_tx_rdft_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + double f, m; + TXSample *tab; + uint64_t r2r = flags & AV_TX_REAL_TO_REAL; + int len4 = FFALIGN(len, 4) / 4; + + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + + flags &= ~(AV_TX_REAL_TO_REAL | AV_TX_REAL_TO_IMAGINARY); + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, NULL, len >> 1, inv, scale))) + return ret; + + if (!(s->exp = av_mallocz((8 + 2*len4)*sizeof(*s->exp)))) + return AVERROR(ENOMEM); + + tab = (TXSample *)s->exp; + + f = 2*M_PI/len; + + m = (inv ? 2*s->scale_d : s->scale_d); + + *tab++ = RESCALE((inv ? 0.5 : 1.0) * m); + *tab++ = RESCALE(inv ? 0.5*m : 1.0*m); + *tab++ = RESCALE( m); + *tab++ = RESCALE(-m); + + *tab++ = RESCALE( (0.5 - 0.0) * m); + if (r2r) + *tab++ = 1 / s->scale_f; + else + *tab++ = RESCALE( (0.0 - 0.5) * m); + *tab++ = RESCALE( (0.5 - inv) * m); + *tab++ = RESCALE(-(0.5 - inv) * m); + + for (int i = 0; i < len4; i++) + *tab++ = RESCALE(cos(i*f)); + + tab = ((TXSample *)s->exp) + len4 + 8; + + for (int i = 0; i < len4; i++) + *tab++ = RESCALE(cos(((len - i*4)/4.0)*f)) * (inv ? 1 : -1); + + return 0; +} + +#define DECL_RDFT(n, inv) \ +static void TX_NAME(ff_tx_rdft_ ##n)(AVTXContext *s, void *_dst, \ + void *_src, ptrdiff_t stride) \ +{ \ + const int len2 = s->len >> 1; \ + const int len4 = s->len >> 2; \ + const TXSample *fact = (void *)s->exp; \ + const TXSample *tcos = fact + 8; \ + const TXSample *tsin = tcos + len4; \ + TXComplex *data = inv ? _src : _dst; \ + TXComplex t[3]; \ + \ + if (!inv) \ + s->fn[0](&s->sub[0], data, _src, sizeof(TXComplex)); \ + else \ + data[0].im = data[len2].re; \ + \ + /* The DC value's both components are real, but we need to change them \ + * into complex values. Also, the middle of the array is special-cased. \ + * These operations can be done before or after the loop. */ \ + t[0].re = data[0].re; \ + data[0].re = t[0].re + data[0].im; \ + data[0].im = t[0].re - data[0].im; \ + data[ 0].re = MULT(fact[0], data[ 0].re); \ + data[ 0].im = MULT(fact[1], data[ 0].im); \ + data[len4].re = MULT(fact[2], data[len4].re); \ + data[len4].im = MULT(fact[3], data[len4].im); \ + \ + for (int i = 1; i < len4; i++) { \ + /* Separate even and odd FFTs */ \ + t[0].re = MULT(fact[4], (data[i].re + data[len2 - i].re)); \ + t[0].im = MULT(fact[5], (data[i].im - data[len2 - i].im)); \ + t[1].re = MULT(fact[6], (data[i].im + data[len2 - i].im)); \ + t[1].im = MULT(fact[7], (data[i].re - data[len2 - i].re)); \ + \ + /* Apply twiddle factors to the odd FFT and add to the even FFT */ \ + CMUL(t[2].re, t[2].im, t[1].re, t[1].im, tcos[i], tsin[i]); \ + \ + data[ i].re = t[0].re + t[2].re; \ + data[ i].im = t[2].im - t[0].im; \ + data[len2 - i].re = t[0].re - t[2].re; \ + data[len2 - i].im = t[2].im + t[0].im; \ + } \ + \ + if (inv) { \ + s->fn[0](&s->sub[0], _dst, data, sizeof(TXComplex)); \ + } else { \ + /* Move [0].im to the last position, as convention requires */ \ + data[len2].re = data[0].im; \ + data[ 0].im = data[len2].im = 0; \ + } \ +} \ + \ +static const FFTXCodelet TX_NAME(ff_tx_rdft_ ##n## _def) = { \ + .name = TX_NAME_STR("rdft_" #n), \ + .function = TX_NAME(ff_tx_rdft_ ##n), \ + .type = TX_TYPE(RDFT), \ + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE | \ + (inv ? FF_TX_INVERSE_ONLY : FF_TX_FORWARD_ONLY), \ + .factors = { 4, TX_FACTOR_ANY }, \ + .nb_factors = 2, \ + .min_len = 4, \ + .max_len = TX_LEN_UNLIMITED, \ + .init = TX_NAME(ff_tx_rdft_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +DECL_RDFT(r2c, 0) +DECL_RDFT(c2r, 1) + +#define DECL_RDFT_HALF(n, mode, mod2) \ +static void TX_NAME(ff_tx_rdft_ ##n)(AVTXContext *s, void *_dst, \ + void *_src, ptrdiff_t stride) \ +{ \ + const int len = s->len; \ + const int len2 = len >> 1; \ + const int len4 = len >> 2; \ + const int aligned_len4 = FFALIGN(len, 4)/4; \ + const TXSample *fact = (void *)s->exp; \ + const TXSample *tcos = fact + 8; \ + const TXSample *tsin = tcos + aligned_len4; \ + TXComplex *data = _dst; \ + TXSample *out = _dst; /* Half-complex is forward-only */ \ + TXSample tmp_dc; \ + av_unused TXSample tmp_mid; \ + TXSample tmp[4]; \ + TXComplex sf, sl; \ + \ + s->fn[0](&s->sub[0], _dst, _src, sizeof(TXComplex)); \ + \ + tmp_dc = data[0].re; \ + data[ 0].re = tmp_dc + data[0].im; \ + tmp_dc = tmp_dc - data[0].im; \ + \ + data[ 0].re = MULT(fact[0], data[ 0].re); \ + tmp_dc = MULT(fact[1], tmp_dc); \ + data[len4].re = MULT(fact[2], data[len4].re); \ + \ + if (!mod2) { \ + data[len4].im = MULT(fact[3], data[len4].im); \ + } else { \ + sf = data[len4]; \ + sl = data[len4 + 1]; \ + if (mode == AV_TX_REAL_TO_REAL) \ + tmp[0] = MULT(fact[4], (sf.re + sl.re)); \ + else \ + tmp[0] = MULT(fact[5], (sf.im - sl.im)); \ + tmp[1] = MULT(fact[6], (sf.im + sl.im)); \ + tmp[2] = MULT(fact[7], (sf.re - sl.re)); \ + \ + if (mode == AV_TX_REAL_TO_REAL) { \ + tmp[3] = tmp[1]*tcos[len4] - tmp[2]*tsin[len4]; \ + tmp_mid = (tmp[0] - tmp[3]); \ + } else { \ + tmp[3] = tmp[1]*tsin[len4] + tmp[2]*tcos[len4]; \ + tmp_mid = (tmp[0] + tmp[3]); \ + } \ + } \ + \ + /* NOTE: unrolling this breaks non-mod8 lengths */ \ + for (int i = 1; i <= len4; i++) { \ + TXSample tmp[4]; \ + TXComplex sf = data[i]; \ + TXComplex sl = data[len2 - i]; \ + \ + if (mode == AV_TX_REAL_TO_REAL) \ + tmp[0] = MULT(fact[4], (sf.re + sl.re)); \ + else \ + tmp[0] = MULT(fact[5], (sf.im - sl.im)); \ + \ + tmp[1] = MULT(fact[6], (sf.im + sl.im)); \ + tmp[2] = MULT(fact[7], (sf.re - sl.re)); \ + \ + if (mode == AV_TX_REAL_TO_REAL) { \ + tmp[3] = tmp[1]*tcos[i] - tmp[2]*tsin[i]; \ + out[i] = (tmp[0] + tmp[3]); \ + out[len - i] = (tmp[0] - tmp[3]); \ + } else { \ + tmp[3] = tmp[1]*tsin[i] + tmp[2]*tcos[i]; \ + out[i - 1] = (tmp[3] - tmp[0]); \ + out[len - i - 1] = (tmp[0] + tmp[3]); \ + } \ + } \ + \ + for (int i = 1; i < (len4 + (mode == AV_TX_REAL_TO_IMAGINARY)); i++) \ + out[len2 - i] = out[len - i]; \ + \ + if (mode == AV_TX_REAL_TO_REAL) { \ + out[len2] = tmp_dc; \ + if (mod2) \ + out[len4 + 1] = tmp_mid * fact[5]; \ + } else if (mod2) { \ + out[len4] = tmp_mid; \ + } \ +} \ + \ +static const FFTXCodelet TX_NAME(ff_tx_rdft_ ##n## _def) = { \ + .name = TX_NAME_STR("rdft_" #n), \ + .function = TX_NAME(ff_tx_rdft_ ##n), \ + .type = TX_TYPE(RDFT), \ + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | mode | \ + FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, \ + .factors = { 2 + 2*(!mod2), TX_FACTOR_ANY }, \ + .nb_factors = 2, \ + .min_len = 2 + 2*(!mod2), \ + .max_len = TX_LEN_UNLIMITED, \ + .init = TX_NAME(ff_tx_rdft_init), \ + .cpu_flags = FF_TX_CPU_FLAGS_ALL, \ + .prio = FF_TX_PRIO_BASE, \ +}; + +DECL_RDFT_HALF(r2r, AV_TX_REAL_TO_REAL, 0) +DECL_RDFT_HALF(r2r_mod2, AV_TX_REAL_TO_REAL, 1) +DECL_RDFT_HALF(r2i, AV_TX_REAL_TO_IMAGINARY, 0) +DECL_RDFT_HALF(r2i_mod2, AV_TX_REAL_TO_IMAGINARY, 1) + +static av_cold int TX_NAME(ff_tx_dct_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + double freq; + TXSample *tab; + SCALE_TYPE rsc = *((SCALE_TYPE *)scale); + + if (inv) { + len *= 2; + s->len *= 2; + rsc *= 0.5; + } + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(RDFT), flags, NULL, len, inv, &rsc))) + return ret; + + s->exp = av_malloc((len/2)*3*sizeof(TXSample)); + if (!s->exp) + return AVERROR(ENOMEM); + + tab = (TXSample *)s->exp; + + freq = M_PI/(len*2); + + for (int i = 0; i < len; i++) + tab[i] = RESCALE(cos(i*freq)*(!inv + 1)); + + if (inv) { + for (int i = 0; i < len/2; i++) + tab[len + i] = RESCALE(0.5 / sin((2*i + 1)*freq)); + } else { + for (int i = 0; i < len/2; i++) + tab[len + i] = RESCALE(cos((len - 2*i - 1)*freq)); + } + + return 0; +} + +static void TX_NAME(ff_tx_dctII)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *dst = _dst; + TXSample *src = _src; + const int len = s->len; + const int len2 = len >> 1; + const TXSample *exp = (void *)s->exp; + TXSample next; +#ifdef TX_INT32 + int64_t tmp1, tmp2; +#else + TXSample tmp1, tmp2; +#endif + + for (int i = 0; i < len2; i++) { + TXSample in1 = src[i]; + TXSample in2 = src[len - i - 1]; + TXSample s = exp[len + i]; + +#ifdef TX_INT32 + tmp1 = in1 + in2; + tmp2 = in1 - in2; + + tmp1 >>= 1; + tmp2 *= s; + + tmp2 = (tmp2 + 0x40000000) >> 31; +#else + tmp1 = (in1 + in2)*0.5; + tmp2 = (in1 - in2)*s; +#endif + + src[i] = tmp1 + tmp2; + src[len - i - 1] = tmp1 - tmp2; + } + + s->fn[0](&s->sub[0], dst, src, sizeof(TXComplex)); + + next = dst[len]; + + for (int i = len - 2; i > 0; i -= 2) { + TXSample tmp; + + CMUL(tmp, dst[i], exp[len - i], exp[i], dst[i + 0], dst[i + 1]); + + dst[i + 1] = next; + + next += tmp; + } + +#ifdef TX_INT32 + tmp1 = ((int64_t)exp[0]) * ((int64_t)dst[0]); + dst[0] = (tmp1 + 0x40000000) >> 31; +#else + dst[0] = exp[0] * dst[0]; +#endif + dst[1] = next; +} + +static void TX_NAME(ff_tx_dctIII)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *dst = _dst; + TXSample *src = _src; + const int len = s->len; + const int len2 = len >> 1; + const TXSample *exp = (void *)s->exp; +#ifdef TX_INT32 + int64_t tmp1, tmp2 = src[len - 1]; + tmp2 = (2*tmp2 + 0x40000000) >> 31; +#else + TXSample tmp1, tmp2 = 2*src[len - 1]; +#endif + + src[len] = tmp2; + + for (int i = len - 2; i >= 2; i -= 2) { + TXSample val1 = src[i - 0]; + TXSample val2 = src[i - 1] - src[i + 1]; + + CMUL(src[i + 1], src[i], exp[len - i], exp[i], val1, val2); + } + + s->fn[0](&s->sub[0], dst, src, sizeof(float)); + + for (int i = 0; i < len2; i++) { + TXSample in1 = dst[i]; + TXSample in2 = dst[len - i - 1]; + TXSample c = exp[len + i]; + + tmp1 = in1 + in2; + tmp2 = in1 - in2; + tmp2 *= c; +#ifdef TX_INT32 + tmp2 = (tmp2 + 0x40000000) >> 31; +#endif + + dst[i] = tmp1 + tmp2; + dst[len - i - 1] = tmp1 - tmp2; + } +} + +static const FFTXCodelet TX_NAME(ff_tx_dctII_def) = { + .name = TX_NAME_STR("dctII"), + .function = TX_NAME(ff_tx_dctII), + .type = TX_TYPE(DCT), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | + FF_TX_OUT_OF_PLACE | FF_TX_FORWARD_ONLY, + .factors = { 2, TX_FACTOR_ANY }, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_dct_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static const FFTXCodelet TX_NAME(ff_tx_dctIII_def) = { + .name = TX_NAME_STR("dctIII"), + .function = TX_NAME(ff_tx_dctIII), + .type = TX_TYPE(DCT), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | + FF_TX_OUT_OF_PLACE | FF_TX_INVERSE_ONLY, + .factors = { 2, TX_FACTOR_ANY }, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_dct_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static av_cold int TX_NAME(ff_tx_dcstI_init)(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + SCALE_TYPE rsc = *((SCALE_TYPE *)scale); + + if (inv) { + len *= 2; + s->len *= 2; + rsc *= 0.5; + } + + /* We want a half-complex RDFT */ + flags |= cd->type == TX_TYPE(DCT_I) ? AV_TX_REAL_TO_REAL : + AV_TX_REAL_TO_IMAGINARY; + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(RDFT), flags, NULL, + (len - 1 + 2*(cd->type == TX_TYPE(DST_I)))*2, + 0, &rsc))) + return ret; + + s->tmp = av_mallocz((len + 1)*2*sizeof(TXSample)); + if (!s->tmp) + return AVERROR(ENOMEM); + + return 0; +} + +static void TX_NAME(ff_tx_dctI)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *dst = _dst; + TXSample *src = _src; + const int len = s->len - 1; + TXSample *tmp = (TXSample *)s->tmp; + + stride /= sizeof(TXSample); + + for (int i = 0; i < len; i++) + tmp[i] = tmp[2*len - i] = src[i * stride]; + + tmp[len] = src[len * stride]; /* Middle */ + + s->fn[0](&s->sub[0], dst, tmp, sizeof(TXSample)); +} + +static void TX_NAME(ff_tx_dstI)(AVTXContext *s, void *_dst, + void *_src, ptrdiff_t stride) +{ + TXSample *dst = _dst; + TXSample *src = _src; + const int len = s->len + 1; + TXSample *tmp = (void *)s->tmp; + + stride /= sizeof(TXSample); + + tmp[0] = 0; + + for (int i = 1; i < len; i++) { + TXSample a = src[(i - 1) * stride]; + tmp[i] = -a; + tmp[2*len - i] = a; + } + + tmp[len] = 0; /* i == n, Nyquist */ + + s->fn[0](&s->sub[0], dst, tmp, sizeof(float)); +} + +static const FFTXCodelet TX_NAME(ff_tx_dctI_def) = { + .name = TX_NAME_STR("dctI"), + .function = TX_NAME(ff_tx_dctI), + .type = TX_TYPE(DCT_I), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_dcstI_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +static const FFTXCodelet TX_NAME(ff_tx_dstI_def) = { + .name = TX_NAME_STR("dstI"), + .function = TX_NAME(ff_tx_dstI), + .type = TX_TYPE(DST_I), + .flags = AV_TX_UNALIGNED | AV_TX_INPLACE | FF_TX_OUT_OF_PLACE, + .factors = { 2, TX_FACTOR_ANY }, + .nb_factors = 2, + .min_len = 2, + .max_len = TX_LEN_UNLIMITED, + .init = TX_NAME(ff_tx_dcstI_init), + .cpu_flags = FF_TX_CPU_FLAGS_ALL, + .prio = FF_TX_PRIO_BASE, +}; + +int TX_TAB(ff_tx_mdct_gen_exp)(AVTXContext *s, int *pre_tab) +{ + int off = 0; + int len4 = s->len >> 1; + double scale = s->scale_d; + const double theta = (scale < 0 ? len4 : 0) + 1.0/8.0; + size_t alloc = pre_tab ? 2*len4 : len4; + + if (!(s->exp = av_malloc_array(alloc, sizeof(*s->exp)))) + return AVERROR(ENOMEM); + + scale = sqrt(fabs(scale)); + + if (pre_tab) + off = len4; + + for (int i = 0; i < len4; i++) { + const double alpha = M_PI_2 * (i + theta) / len4; + s->exp[off + i] = (TXComplex){ RESCALE(cos(alpha) * scale), + RESCALE(sin(alpha) * scale) }; + } + + if (pre_tab) + for (int i = 0; i < len4; i++) + s->exp[i] = s->exp[len4 + pre_tab[i]]; + + return 0; +} + +const FFTXCodelet * const TX_NAME(ff_tx_codelet_list)[] = { + /* Split-Radix codelets */ + &TX_NAME(ff_tx_fft2_ns_def), + &TX_NAME(ff_tx_fft4_ns_def), + &TX_NAME(ff_tx_fft8_ns_def), + &TX_NAME(ff_tx_fft16_ns_def), + &TX_NAME(ff_tx_fft32_ns_def), + &TX_NAME(ff_tx_fft64_ns_def), + &TX_NAME(ff_tx_fft128_ns_def), + &TX_NAME(ff_tx_fft256_ns_def), + &TX_NAME(ff_tx_fft512_ns_def), + &TX_NAME(ff_tx_fft1024_ns_def), + &TX_NAME(ff_tx_fft2048_ns_def), + &TX_NAME(ff_tx_fft4096_ns_def), + &TX_NAME(ff_tx_fft8192_ns_def), + &TX_NAME(ff_tx_fft16384_ns_def), + &TX_NAME(ff_tx_fft32768_ns_def), + &TX_NAME(ff_tx_fft65536_ns_def), + &TX_NAME(ff_tx_fft131072_ns_def), + &TX_NAME(ff_tx_fft262144_ns_def), + &TX_NAME(ff_tx_fft524288_ns_def), + &TX_NAME(ff_tx_fft1048576_ns_def), + &TX_NAME(ff_tx_fft2097152_ns_def), + + /* Prime factor codelets */ + &TX_NAME(ff_tx_fft3_ns_def), + &TX_NAME(ff_tx_fft5_ns_def), + &TX_NAME(ff_tx_fft7_ns_def), + &TX_NAME(ff_tx_fft9_ns_def), + &TX_NAME(ff_tx_fft15_ns_def), + + /* We get these for free */ + &TX_NAME(ff_tx_fft3_fwd_def), + &TX_NAME(ff_tx_fft5_fwd_def), + &TX_NAME(ff_tx_fft7_fwd_def), + &TX_NAME(ff_tx_fft9_fwd_def), + + /* Standalone transforms */ + &TX_NAME(ff_tx_fft_def), + &TX_NAME(ff_tx_fft_inplace_def), + &TX_NAME(ff_tx_fft_inplace_small_def), + &TX_NAME(ff_tx_fft_pfa_def), + &TX_NAME(ff_tx_fft_pfa_ns_def), + &TX_NAME(ff_tx_fft_naive_def), + &TX_NAME(ff_tx_fft_naive_small_def), + &TX_NAME(ff_tx_mdct_fwd_def), + &TX_NAME(ff_tx_mdct_inv_def), + &TX_NAME(ff_tx_mdct_pfa_3xM_fwd_def), + &TX_NAME(ff_tx_mdct_pfa_5xM_fwd_def), + &TX_NAME(ff_tx_mdct_pfa_7xM_fwd_def), + &TX_NAME(ff_tx_mdct_pfa_9xM_fwd_def), + &TX_NAME(ff_tx_mdct_pfa_15xM_fwd_def), + &TX_NAME(ff_tx_mdct_pfa_3xM_inv_def), + &TX_NAME(ff_tx_mdct_pfa_5xM_inv_def), + &TX_NAME(ff_tx_mdct_pfa_7xM_inv_def), + &TX_NAME(ff_tx_mdct_pfa_9xM_inv_def), + &TX_NAME(ff_tx_mdct_pfa_15xM_inv_def), + &TX_NAME(ff_tx_mdct_naive_fwd_def), + &TX_NAME(ff_tx_mdct_naive_inv_def), + &TX_NAME(ff_tx_mdct_inv_full_def), + &TX_NAME(ff_tx_rdft_r2c_def), + &TX_NAME(ff_tx_rdft_r2r_def), + &TX_NAME(ff_tx_rdft_r2r_mod2_def), + &TX_NAME(ff_tx_rdft_r2i_def), + &TX_NAME(ff_tx_rdft_r2i_mod2_def), + &TX_NAME(ff_tx_rdft_c2r_def), + &TX_NAME(ff_tx_dctII_def), + &TX_NAME(ff_tx_dctIII_def), + &TX_NAME(ff_tx_dctI_def), + &TX_NAME(ff_tx_dstI_def), + + NULL, +}; diff --git a/media/ffvpx/libavutil/utils.c b/media/ffvpx/libavutil/utils.c new file mode 100644 index 0000000000..94d247bbee --- /dev/null +++ b/media/ffvpx/libavutil/utils.c @@ -0,0 +1,112 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" +#include "avutil.h" +#include "avassert.h" + +/** + * @file + * various utility functions + */ + +const char *av_get_media_type_string(enum AVMediaType media_type) +{ + switch (media_type) { + case AVMEDIA_TYPE_VIDEO: return "video"; + case AVMEDIA_TYPE_AUDIO: return "audio"; + case AVMEDIA_TYPE_DATA: return "data"; + case AVMEDIA_TYPE_SUBTITLE: return "subtitle"; + case AVMEDIA_TYPE_ATTACHMENT: return "attachment"; + default: return NULL; + } +} + +char av_get_picture_type_char(enum AVPictureType pict_type) +{ + switch (pict_type) { + case AV_PICTURE_TYPE_I: return 'I'; + case AV_PICTURE_TYPE_P: return 'P'; + case AV_PICTURE_TYPE_B: return 'B'; + case AV_PICTURE_TYPE_S: return 'S'; + case AV_PICTURE_TYPE_SI: return 'i'; + case AV_PICTURE_TYPE_SP: return 'p'; + case AV_PICTURE_TYPE_BI: return 'b'; + default: return '?'; + } +} + +unsigned av_int_list_length_for_size(unsigned elsize, + const void *list, uint64_t term) +{ + unsigned i; + + if (!list) + return 0; +#define LIST_LENGTH(type) \ + { type t = term, *l = (type *)list; for (i = 0; l[i] != t; i++); } + switch (elsize) { + case 1: LIST_LENGTH(uint8_t); break; + case 2: LIST_LENGTH(uint16_t); break; + case 4: LIST_LENGTH(uint32_t); break; + case 8: LIST_LENGTH(uint64_t); break; + default: av_assert0(!"valid element size"); + } + return i; +} + +char *av_fourcc_make_string(char *buf, uint32_t fourcc) +{ + int i; + char *orig_buf = buf; + size_t buf_size = AV_FOURCC_MAX_STRING_SIZE; + + for (i = 0; i < 4; i++) { + const int c = fourcc & 0xff; + const int print_chr = (c >= '0' && c <= '9') || + (c >= 'a' && c <= 'z') || + (c >= 'A' && c <= 'Z') || + (c && strchr(". -_", c)); + const int len = snprintf(buf, buf_size, print_chr ? "%c" : "[%d]", c); + if (len < 0) + break; + buf += len; + buf_size = buf_size > len ? buf_size - len : 0; + fourcc >>= 8; + } + + return orig_buf; +} + +AVRational av_get_time_base_q(void) +{ + return (AVRational){1, AV_TIME_BASE}; +} + +void av_assert0_fpu(void) { +#if HAVE_MMX_INLINE + uint16_t state[14]; + __asm__ volatile ( + "fstenv %0 \n\t" + : "+m" (state) + : + : "memory" + ); + av_assert0((state[4] & 3) == 3); +#endif +} diff --git a/media/ffvpx/libavutil/version.h b/media/ffvpx/libavutil/version.h new file mode 100644 index 0000000000..772c4e209c --- /dev/null +++ b/media/ffvpx/libavutil/version.h @@ -0,0 +1,128 @@ +/* + * copyright (c) 2003 Fabrice Bellard + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * @ingroup lavu + * Libavutil version macros + */ + +#ifndef AVUTIL_VERSION_H +#define AVUTIL_VERSION_H + +#include "macros.h" + +/** + * @addtogroup version_utils + * + * Useful to check and match library version in order to maintain + * backward compatibility. + * + * The FFmpeg libraries follow a versioning sheme very similar to + * Semantic Versioning (http://semver.org/) + * The difference is that the component called PATCH is called MICRO in FFmpeg + * and its value is reset to 100 instead of 0 to keep it above or equal to 100. + * Also we do not increase MICRO for every bugfix or change in git master. + * + * Prior to FFmpeg 3.2 point releases did not change any lib version number to + * avoid aliassing different git master checkouts. + * Starting with FFmpeg 3.2, the released library versions will occupy + * a separate MAJOR.MINOR that is not used on the master development branch. + * That is if we branch a release of master 55.10.123 we will bump to 55.11.100 + * for the release and master will continue at 55.12.100 after it. Each new + * point release will then bump the MICRO improving the usefulness of the lib + * versions. + * + * @{ + */ + +#define AV_VERSION_INT(a, b, c) ((a)<<16 | (b)<<8 | (c)) +#define AV_VERSION_DOT(a, b, c) a ##.## b ##.## c +#define AV_VERSION(a, b, c) AV_VERSION_DOT(a, b, c) + +/** + * Extract version components from the full ::AV_VERSION_INT int as returned + * by functions like ::avformat_version() and ::avcodec_version() + */ +#define AV_VERSION_MAJOR(a) ((a) >> 16) +#define AV_VERSION_MINOR(a) (((a) & 0x00FF00) >> 8) +#define AV_VERSION_MICRO(a) ((a) & 0xFF) + +/** + * @} + */ + +/** + * @defgroup lavu_ver Version and Build diagnostics + * + * Macros and function useful to check at compiletime and at runtime + * which version of libavutil is in use. + * + * @{ + */ + +#define LIBAVUTIL_VERSION_MAJOR 58 +#define LIBAVUTIL_VERSION_MINOR 36 +#define LIBAVUTIL_VERSION_MICRO 101 + +#define LIBAVUTIL_VERSION_INT AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \ + LIBAVUTIL_VERSION_MINOR, \ + LIBAVUTIL_VERSION_MICRO) +#define LIBAVUTIL_VERSION AV_VERSION(LIBAVUTIL_VERSION_MAJOR, \ + LIBAVUTIL_VERSION_MINOR, \ + LIBAVUTIL_VERSION_MICRO) +#define LIBAVUTIL_BUILD LIBAVUTIL_VERSION_INT + +#define LIBAVUTIL_IDENT "Lavu" AV_STRINGIFY(LIBAVUTIL_VERSION) + +/** + * @defgroup lavu_depr_guards Deprecation Guards + * FF_API_* defines may be placed below to indicate public API that will be + * dropped at a future version bump. The defines themselves are not part of + * the public API and may change, break or disappear at any time. + * + * @note, when bumping the major version it is recommended to manually + * disable each FF_API_* in its own commit instead of disabling them all + * at once through the bump. This improves the git bisect-ability of the change. + * + * @{ + */ + +#define FF_API_FIFO_PEEK2 (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_FIFO_OLD_API (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_XVMC (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_OLD_CHANNEL_LAYOUT (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_AV_FOPEN_UTF8 (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_PKT_DURATION (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_REORDERED_OPAQUE (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_FRAME_PICTURE_NUMBER (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_HDR_VIVID_THREE_SPLINE (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_FRAME_PKT (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_INTERLACED_FRAME (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_FRAME_KEY (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_PALETTE_HAS_CHANGED (LIBAVUTIL_VERSION_MAJOR < 59) +#define FF_API_VULKAN_CONTIGUOUS_MEMORY (LIBAVUTIL_VERSION_MAJOR < 59) + +/** + * @} + * @} + */ + +#endif /* AVUTIL_VERSION_H */ diff --git a/media/ffvpx/libavutil/video_enc_params.c b/media/ffvpx/libavutil/video_enc_params.c new file mode 100644 index 0000000000..33592dc128 --- /dev/null +++ b/media/ffvpx/libavutil/video_enc_params.c @@ -0,0 +1,80 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stddef.h> +#include <stdint.h> + +#include "buffer.h" +#include "frame.h" +#include "mem.h" +#include "video_enc_params.h" + +AVVideoEncParams *av_video_enc_params_alloc(enum AVVideoEncParamsType type, + unsigned int nb_blocks, size_t *out_size) +{ + struct TestStruct { + AVVideoEncParams p; + AVVideoBlockParams b; + }; + const size_t blocks_offset = offsetof(struct TestStruct, b); + size_t size = blocks_offset; + AVVideoEncParams *par; + + if (nb_blocks > (SIZE_MAX - size) / sizeof(AVVideoBlockParams)) + return NULL; + size += sizeof(AVVideoBlockParams) * nb_blocks; + + par = av_mallocz(size); + if (!par) + return NULL; + + par->type = type; + par->nb_blocks = nb_blocks; + par->block_size = sizeof(AVVideoBlockParams); + par->blocks_offset = blocks_offset; + + if (out_size) + *out_size = size; + + return par; +} + +AVVideoEncParams* +av_video_enc_params_create_side_data(AVFrame *frame, enum AVVideoEncParamsType type, + unsigned int nb_blocks) +{ + AVBufferRef *buf; + AVVideoEncParams *par; + size_t size; + + par = av_video_enc_params_alloc(type, nb_blocks, &size); + if (!par) + return NULL; + buf = av_buffer_create((uint8_t *)par, size, NULL, NULL, 0); + if (!buf) { + av_freep(&par); + return NULL; + } + + if (!av_frame_new_side_data_from_buf(frame, AV_FRAME_DATA_VIDEO_ENC_PARAMS, buf)) { + av_buffer_unref(&buf); + return NULL; + } + + return par; +} diff --git a/media/ffvpx/libavutil/video_enc_params.h b/media/ffvpx/libavutil/video_enc_params.h new file mode 100644 index 0000000000..62265a5c06 --- /dev/null +++ b/media/ffvpx/libavutil/video_enc_params.h @@ -0,0 +1,171 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_VIDEO_ENC_PARAMS_H +#define AVUTIL_VIDEO_ENC_PARAMS_H + +#include <stddef.h> +#include <stdint.h> + +#include "libavutil/avassert.h" +#include "libavutil/frame.h" + +enum AVVideoEncParamsType { + AV_VIDEO_ENC_PARAMS_NONE = -1, + /** + * VP9 stores: + * - per-frame base (luma AC) quantizer index, exported as AVVideoEncParams.qp + * - deltas for luma DC, chroma AC and chroma DC, exported in the + * corresponding entries in AVVideoEncParams.delta_qp + * - per-segment delta, exported as for each block as AVVideoBlockParams.delta_qp + * + * To compute the resulting quantizer index for a block: + * - for luma AC, add the base qp and the per-block delta_qp, saturating to + * unsigned 8-bit. + * - for luma DC and chroma AC/DC, add the corresponding + * AVVideoBlockParams.delta_qp to the luma AC index, again saturating to + * unsigned 8-bit. + */ + AV_VIDEO_ENC_PARAMS_VP9, + + /** + * H.264 stores: + * - in PPS (per-picture): + * * initial QP_Y (luma) value, exported as AVVideoEncParams.qp + * * delta(s) for chroma QP values (same for both, or each separately), + * exported as in the corresponding entries in AVVideoEncParams.delta_qp + * - per-slice QP delta, not exported directly, added to the per-MB value + * - per-MB delta; not exported directly; the final per-MB quantizer + * parameter - QP_Y - minus the value in AVVideoEncParams.qp is exported + * as AVVideoBlockParams.qp_delta. + */ + AV_VIDEO_ENC_PARAMS_H264, + + /* + * MPEG-2-compatible quantizer. + * + * Summing the frame-level qp with the per-block delta_qp gives the + * resulting quantizer for the block. + */ + AV_VIDEO_ENC_PARAMS_MPEG2, +}; + +/** + * Video encoding parameters for a given frame. This struct is allocated along + * with an optional array of per-block AVVideoBlockParams descriptors. + * Must be allocated with av_video_enc_params_alloc(). + */ +typedef struct AVVideoEncParams { + /** + * Number of blocks in the array. + * + * May be 0, in which case no per-block information is present. In this case + * the values of blocks_offset / block_size are unspecified and should not + * be accessed. + */ + unsigned int nb_blocks; + /** + * Offset in bytes from the beginning of this structure at which the array + * of blocks starts. + */ + size_t blocks_offset; + /* + * Size of each block in bytes. May not match sizeof(AVVideoBlockParams). + */ + size_t block_size; + + /** + * Type of the parameters (the codec they are used with). + */ + enum AVVideoEncParamsType type; + + /** + * Base quantisation parameter for the frame. The final quantiser for a + * given block in a given plane is obtained from this value, possibly + * combined with {@code delta_qp} and the per-block delta in a manner + * documented for each type. + */ + int32_t qp; + + /** + * Quantisation parameter offset from the base (per-frame) qp for a given + * plane (first index) and AC/DC coefficients (second index). + */ + int32_t delta_qp[4][2]; +} AVVideoEncParams; + +/** + * Data structure for storing block-level encoding information. + * It is allocated as a part of AVVideoEncParams and should be retrieved with + * av_video_enc_params_block(). + * + * sizeof(AVVideoBlockParams) is not a part of the ABI and new fields may be + * added to it. + */ +typedef struct AVVideoBlockParams { + /** + * Distance in luma pixels from the top-left corner of the visible frame + * to the top-left corner of the block. + * Can be negative if top/right padding is present on the coded frame. + */ + int src_x, src_y; + /** + * Width and height of the block in luma pixels. + */ + int w, h; + + /** + * Difference between this block's final quantization parameter and the + * corresponding per-frame value. + */ + int32_t delta_qp; +} AVVideoBlockParams; + +/** + * Get the block at the specified {@code idx}. Must be between 0 and nb_blocks - 1. + */ +static av_always_inline AVVideoBlockParams* +av_video_enc_params_block(AVVideoEncParams *par, unsigned int idx) +{ + av_assert0(idx < par->nb_blocks); + return (AVVideoBlockParams *)((uint8_t *)par + par->blocks_offset + + idx * par->block_size); +} + +/** + * Allocates memory for AVVideoEncParams of the given type, plus an array of + * {@code nb_blocks} AVVideoBlockParams and initializes the variables. Can be + * freed with a normal av_free() call. + * + * @param out_size if non-NULL, the size in bytes of the resulting data array is + * written here. + */ +AVVideoEncParams *av_video_enc_params_alloc(enum AVVideoEncParamsType type, + unsigned int nb_blocks, size_t *out_size); + +/** + * Allocates memory for AVEncodeInfoFrame plus an array of + * {@code nb_blocks} AVEncodeInfoBlock in the given AVFrame {@code frame} + * as AVFrameSideData of type AV_FRAME_DATA_VIDEO_ENC_PARAMS + * and initializes the variables. + */ +AVVideoEncParams* +av_video_enc_params_create_side_data(AVFrame *frame, enum AVVideoEncParamsType type, + unsigned int nb_blocks); + +#endif /* AVUTIL_VIDEO_ENC_PARAMS_H */ diff --git a/media/ffvpx/libavutil/x86/asm.h b/media/ffvpx/libavutil/x86/asm.h new file mode 100644 index 0000000000..9bff42d628 --- /dev/null +++ b/media/ffvpx/libavutil/x86/asm.h @@ -0,0 +1,154 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_ASM_H +#define AVUTIL_X86_ASM_H + +#include <stdint.h> +#include "config.h" + +typedef struct xmm_reg { uint64_t a, b; } xmm_reg; +typedef struct ymm_reg { uint64_t a, b, c, d; } ymm_reg; + +#if ARCH_X86_64 +# define FF_OPSIZE "q" +# define FF_REG_a "rax" +# define FF_REG_b "rbx" +# define FF_REG_c "rcx" +# define FF_REG_d "rdx" +# define FF_REG_D "rdi" +# define FF_REG_S "rsi" +# define FF_PTR_SIZE "8" +typedef int64_t x86_reg; + +/* FF_REG_SP is defined in Solaris sys headers, so use FF_REG_sp */ +# define FF_REG_sp "rsp" +# define FF_REG_BP "rbp" +# define FF_REGBP rbp +# define FF_REGa rax +# define FF_REGb rbx +# define FF_REGc rcx +# define FF_REGd rdx +# define FF_REGSP rsp + +#elif ARCH_X86_32 + +# define FF_OPSIZE "l" +# define FF_REG_a "eax" +# define FF_REG_b "ebx" +# define FF_REG_c "ecx" +# define FF_REG_d "edx" +# define FF_REG_D "edi" +# define FF_REG_S "esi" +# define FF_PTR_SIZE "4" +typedef int32_t x86_reg; + +# define FF_REG_sp "esp" +# define FF_REG_BP "ebp" +# define FF_REGBP ebp +# define FF_REGa eax +# define FF_REGb ebx +# define FF_REGc ecx +# define FF_REGd edx +# define FF_REGSP esp +#else +typedef int x86_reg; +#endif + +#define HAVE_7REGS (ARCH_X86_64 || (HAVE_EBX_AVAILABLE && HAVE_EBP_AVAILABLE)) +#define HAVE_6REGS (ARCH_X86_64 || (HAVE_EBX_AVAILABLE || HAVE_EBP_AVAILABLE)) + +#if ARCH_X86_64 && defined(PIC) +# define BROKEN_RELOCATIONS 1 +#endif + +/* + * If gcc is not set to support sse (-msse) it will not accept xmm registers + * in the clobber list for inline asm. XMM_CLOBBERS takes a list of xmm + * registers to be marked as clobbered and evaluates to nothing if they are + * not supported, or to the list itself if they are supported. Since a clobber + * list may not be empty, XMM_CLOBBERS_ONLY should be used if the xmm + * registers are the only in the clobber list. + * For example a list with "eax" and "xmm0" as clobbers should become: + * : XMM_CLOBBERS("xmm0",) "eax" + * and a list with only "xmm0" should become: + * XMM_CLOBBERS_ONLY("xmm0") + */ +#if HAVE_XMM_CLOBBERS +# define XMM_CLOBBERS(...) __VA_ARGS__ +# define XMM_CLOBBERS_ONLY(...) : __VA_ARGS__ +#else +# define XMM_CLOBBERS(...) +# define XMM_CLOBBERS_ONLY(...) +#endif + +/* Use to export labels from asm. */ +#define LABEL_MANGLE(a) EXTERN_PREFIX #a + +// Use rip-relative addressing if compiling PIC code on x86-64. +#if ARCH_X86_64 && defined(PIC) +# define LOCAL_MANGLE(a) #a "(%%rip)" +#else +# define LOCAL_MANGLE(a) #a +#endif + +#if HAVE_INLINE_ASM_DIRECT_SYMBOL_REFS +# define MANGLE(a) EXTERN_PREFIX LOCAL_MANGLE(a) +# define NAMED_CONSTRAINTS_ADD(...) +# define NAMED_CONSTRAINTS(...) +# define NAMED_CONSTRAINTS_ARRAY_ADD(...) +# define NAMED_CONSTRAINTS_ARRAY(...) +#else + /* When direct symbol references are used in code passed to a compiler that does not support them + * then these references need to be converted to named asm constraints instead. + * Instead of returning a direct symbol MANGLE now returns a named constraint for that specific symbol. + * In order for this to work there must also be a corresponding entry in the asm-interface. To add this + * entry use the macro NAMED_CONSTRAINTS() and pass in a list of each symbol reference used in the + * corresponding block of code. (e.g. NAMED_CONSTRAINTS(var1,var2,var3) where var1 is the first symbol etc. ). + * If there are already existing constraints then use NAMED_CONSTRAINTS_ADD to add to the existing constraint list. + */ +# define MANGLE(a) "%["#a"]" + // Intel/MSVC does not correctly expand va-args so we need a rather ugly hack in order to get it to work +# define FE_0(P,X) P(X) +# define FE_1(P,X,X1) P(X), FE_0(P,X1) +# define FE_2(P,X,X1,X2) P(X), FE_1(P,X1,X2) +# define FE_3(P,X,X1,X2,X3) P(X), FE_2(P,X1,X2,X3) +# define FE_4(P,X,X1,X2,X3,X4) P(X), FE_3(P,X1,X2,X3,X4) +# define FE_5(P,X,X1,X2,X3,X4,X5) P(X), FE_4(P,X1,X2,X3,X4,X5) +# define FE_6(P,X,X1,X2,X3,X4,X5,X6) P(X), FE_5(P,X1,X2,X3,X4,X5,X6) +# define FE_7(P,X,X1,X2,X3,X4,X5,X6,X7) P(X), FE_6(P,X1,X2,X3,X4,X5,X6,X7) +# define FE_8(P,X,X1,X2,X3,X4,X5,X6,X7,X8) P(X), FE_7(P,X1,X2,X3,X4,X5,X6,X7,X8) +# define FE_9(P,X,X1,X2,X3,X4,X5,X6,X7,X8,X9) P(X), FE_8(P,X1,X2,X3,X4,X5,X6,X7,X8,X9) +# define GET_FE_IMPL(_0,_1,_2,_3,_4,_5,_6,_7,_8,_9,NAME,...) NAME +# define GET_FE(A) GET_FE_IMPL A +# define GET_FE_GLUE(x, y) x y +# define FOR_EACH_VA(P,...) GET_FE_GLUE(GET_FE((__VA_ARGS__,FE_9,FE_8,FE_7,FE_6,FE_5,FE_4,FE_3,FE_2,FE_1,FE_0)), (P,__VA_ARGS__)) +# define NAME_CONSTRAINT(x) [x] "m"(x) + // Parameters are a list of each symbol reference required +# define NAMED_CONSTRAINTS_ADD(...) , FOR_EACH_VA(NAME_CONSTRAINT,__VA_ARGS__) + // Same but without comma for when there are no previously defined constraints +# define NAMED_CONSTRAINTS(...) FOR_EACH_VA(NAME_CONSTRAINT,__VA_ARGS__) + // Same as above NAMED_CONSTRAINTS except used for passing arrays/pointers instead of normal variables +# define NAME_CONSTRAINT_ARRAY(x) [x] "m"(*x) +# define NAMED_CONSTRAINTS_ARRAY_ADD(...) , FOR_EACH_VA(NAME_CONSTRAINT_ARRAY,__VA_ARGS__) +# define NAMED_CONSTRAINTS_ARRAY(...) FOR_EACH_VA(NAME_CONSTRAINT_ARRAY,__VA_ARGS__) +#endif + +#endif /* AVUTIL_X86_ASM_H */ diff --git a/media/ffvpx/libavutil/x86/bswap.h b/media/ffvpx/libavutil/x86/bswap.h new file mode 100644 index 0000000000..b2f18b6c93 --- /dev/null +++ b/media/ffvpx/libavutil/x86/bswap.h @@ -0,0 +1,88 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +/** + * @file + * byte swapping routines + */ + +#ifndef AVUTIL_X86_BSWAP_H +#define AVUTIL_X86_BSWAP_H + +#include <stdint.h> +#if defined(_MSC_VER) +#include <stdlib.h> +#include <intrin.h> +#endif +#include "config.h" +#include "libavutil/attributes.h" + +#if defined(_MSC_VER) + +#define av_bswap16 av_bswap16 +static av_always_inline av_const uint16_t av_bswap16(uint16_t x) +{ + return _rotr16(x, 8); +} + +#define av_bswap32 av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + return _byteswap_ulong(x); +} + +#if ARCH_X86_64 +#define av_bswap64 av_bswap64 +static inline uint64_t av_const av_bswap64(uint64_t x) +{ + return _byteswap_uint64(x); +} +#endif + + +#elif HAVE_INLINE_ASM + +#if AV_GCC_VERSION_AT_MOST(4,0) +#define av_bswap16 av_bswap16 +static av_always_inline av_const unsigned av_bswap16(unsigned x) +{ + __asm__("rorw $8, %w0" : "+r"(x)); + return x; +} +#endif /* AV_GCC_VERSION_AT_MOST(4,0) */ + +#if AV_GCC_VERSION_AT_MOST(4,4) || defined(__INTEL_COMPILER) +#define av_bswap32 av_bswap32 +static av_always_inline av_const uint32_t av_bswap32(uint32_t x) +{ + __asm__("bswap %0" : "+r" (x)); + return x; +} + +#if ARCH_X86_64 +#define av_bswap64 av_bswap64 +static inline uint64_t av_const av_bswap64(uint64_t x) +{ + __asm__("bswap %0": "=r" (x) : "0" (x)); + return x; +} +#endif +#endif /* AV_GCC_VERSION_AT_MOST(4,4) */ + +#endif /* HAVE_INLINE_ASM */ +#endif /* AVUTIL_X86_BSWAP_H */ diff --git a/media/ffvpx/libavutil/x86/cpu.c b/media/ffvpx/libavutil/x86/cpu.c new file mode 100644 index 0000000000..d6cd4fab9c --- /dev/null +++ b/media/ffvpx/libavutil/x86/cpu.c @@ -0,0 +1,284 @@ +/* + * CPU detection code, extracted from mmx.h + * (c)1997-99 by H. Dietz and R. Fisher + * Converted to C and improved by Fabrice Bellard. + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stdlib.h> +#include <string.h> + +#include "libavutil/x86/asm.h" +#include "libavutil/x86/cpu.h" +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" + +#if HAVE_X86ASM + +#define cpuid(index, eax, ebx, ecx, edx) \ + ff_cpu_cpuid(index, &eax, &ebx, &ecx, &edx) + +#define xgetbv(index, eax, edx) \ + ff_cpu_xgetbv(index, &eax, &edx) + +#elif HAVE_INLINE_ASM + +/* ebx saving is necessary for PIC. gcc seems unable to see it alone */ +#define cpuid(index, eax, ebx, ecx, edx) \ + __asm__ volatile ( \ + "mov %%"FF_REG_b", %%"FF_REG_S" \n\t" \ + "cpuid \n\t" \ + "xchg %%"FF_REG_b", %%"FF_REG_S \ + : "=a" (eax), "=S" (ebx), "=c" (ecx), "=d" (edx) \ + : "0" (index), "2"(0)) + +#define xgetbv(index, eax, edx) \ + __asm__ (".byte 0x0f, 0x01, 0xd0" : "=a"(eax), "=d"(edx) : "c" (index)) + +#define get_eflags(x) \ + __asm__ volatile ("pushfl \n" \ + "pop %0 \n" \ + : "=r"(x)) + +#define set_eflags(x) \ + __asm__ volatile ("push %0 \n" \ + "popfl \n" \ + :: "r"(x)) + +#endif /* HAVE_INLINE_ASM */ + +#if ARCH_X86_64 + +#define cpuid_test() 1 + +#elif HAVE_X86ASM + +#define cpuid_test ff_cpu_cpuid_test + +#elif HAVE_INLINE_ASM + +static int cpuid_test(void) +{ + x86_reg a, c; + + /* Check if CPUID is supported by attempting to toggle the ID bit in + * the EFLAGS register. */ + get_eflags(a); + set_eflags(a ^ 0x200000); + get_eflags(c); + + return a != c; +} +#endif + +/* Function to test if multimedia instructions are supported... */ +int ff_get_cpu_flags_x86(void) +{ + int rval = 0; + +#ifdef cpuid + + int eax, ebx, ecx, edx; + int max_std_level, max_ext_level, std_caps = 0, ext_caps = 0; + int family = 0, model = 0; + union { int i[3]; char c[12]; } vendor; + int xcr0_lo = 0, xcr0_hi = 0; + + if (!cpuid_test()) + return 0; /* CPUID not supported */ + + cpuid(0, max_std_level, vendor.i[0], vendor.i[2], vendor.i[1]); + + if (max_std_level >= 1) { + cpuid(1, eax, ebx, ecx, std_caps); + family = ((eax >> 8) & 0xf) + ((eax >> 20) & 0xff); + model = ((eax >> 4) & 0xf) + ((eax >> 12) & 0xf0); + if (std_caps & (1 << 15)) + rval |= AV_CPU_FLAG_CMOV; + if (std_caps & (1 << 23)) + rval |= AV_CPU_FLAG_MMX; + if (std_caps & (1 << 25)) + rval |= AV_CPU_FLAG_MMXEXT; +#if HAVE_SSE + if (std_caps & (1 << 25)) + rval |= AV_CPU_FLAG_SSE; + if (std_caps & (1 << 26)) + rval |= AV_CPU_FLAG_SSE2; + if (ecx & 1) + rval |= AV_CPU_FLAG_SSE3; + if (ecx & 0x00000200 ) + rval |= AV_CPU_FLAG_SSSE3; + if (ecx & 0x00080000 ) + rval |= AV_CPU_FLAG_SSE4; + if (ecx & 0x00100000 ) + rval |= AV_CPU_FLAG_SSE42; + if (ecx & 0x02000000 ) + rval |= AV_CPU_FLAG_AESNI; +#if HAVE_AVX + /* Check OXSAVE and AVX bits */ + if ((ecx & 0x18000000) == 0x18000000) { + /* Check for OS support */ + xgetbv(0, xcr0_lo, xcr0_hi); + if ((xcr0_lo & 0x6) == 0x6) { + rval |= AV_CPU_FLAG_AVX; + if (ecx & 0x00001000) + rval |= AV_CPU_FLAG_FMA3; + } + } +#endif /* HAVE_AVX */ +#endif /* HAVE_SSE */ + } + if (max_std_level >= 7) { + cpuid(7, eax, ebx, ecx, edx); +#if HAVE_AVX2 + if ((rval & AV_CPU_FLAG_AVX) && (ebx & 0x00000020)) + rval |= AV_CPU_FLAG_AVX2; +#if HAVE_AVX512 /* F, CD, BW, DQ, VL */ + if ((xcr0_lo & 0xe0) == 0xe0) { /* OPMASK/ZMM state */ + if ((rval & AV_CPU_FLAG_AVX2) && (ebx & 0xd0030000) == 0xd0030000) { + rval |= AV_CPU_FLAG_AVX512; +#if HAVE_AVX512ICL + if ((ebx & 0xd0200000) == 0xd0200000 && (ecx & 0x5f42) == 0x5f42) + rval |= AV_CPU_FLAG_AVX512ICL; +#endif /* HAVE_AVX512ICL */ + } + } +#endif /* HAVE_AVX512 */ +#endif /* HAVE_AVX2 */ + /* BMI1/2 don't need OS support */ + if (ebx & 0x00000008) { + rval |= AV_CPU_FLAG_BMI1; + if (ebx & 0x00000100) + rval |= AV_CPU_FLAG_BMI2; + } + } + + cpuid(0x80000000, max_ext_level, ebx, ecx, edx); + + if (max_ext_level >= 0x80000001) { + cpuid(0x80000001, eax, ebx, ecx, ext_caps); + if (ext_caps & (1U << 31)) + rval |= AV_CPU_FLAG_3DNOW; + if (ext_caps & (1 << 30)) + rval |= AV_CPU_FLAG_3DNOWEXT; + if (ext_caps & (1 << 23)) + rval |= AV_CPU_FLAG_MMX; + if (ext_caps & (1 << 22)) + rval |= AV_CPU_FLAG_MMXEXT; + + if (!strncmp(vendor.c, "AuthenticAMD", 12)) { + /* Allow for selectively disabling SSE2 functions on AMD processors + with SSE2 support but not SSE4a. This includes Athlon64, some + Opteron, and some Sempron processors. MMX, SSE, or 3DNow! are faster + than SSE2 often enough to utilize this special-case flag. + AV_CPU_FLAG_SSE2 and AV_CPU_FLAG_SSE2SLOW are both set in this case + so that SSE2 is used unless explicitly disabled by checking + AV_CPU_FLAG_SSE2SLOW. */ + if (rval & AV_CPU_FLAG_SSE2 && !(ecx & 0x00000040)) + rval |= AV_CPU_FLAG_SSE2SLOW; + + /* Similar to the above but for AVX functions on AMD processors. + This is necessary only for functions using YMM registers on Bulldozer + and Jaguar based CPUs as they lack 256-bit execution units. SSE/AVX + functions using XMM registers are always faster on them. + AV_CPU_FLAG_AVX and AV_CPU_FLAG_AVXSLOW are both set so that AVX is + used unless explicitly disabled by checking AV_CPU_FLAG_AVXSLOW. */ + if ((family == 0x15 || family == 0x16) && (rval & AV_CPU_FLAG_AVX)) + rval |= AV_CPU_FLAG_AVXSLOW; + + /* Zen 3 and earlier have slow gather */ + if ((family <= 0x19) && (rval & AV_CPU_FLAG_AVX2)) + rval |= AV_CPU_FLAG_SLOW_GATHER; + } + + /* XOP and FMA4 use the AVX instruction coding scheme, so they can't be + * used unless the OS has AVX support. */ + if (rval & AV_CPU_FLAG_AVX) { + if (ecx & 0x00000800) + rval |= AV_CPU_FLAG_XOP; + if (ecx & 0x00010000) + rval |= AV_CPU_FLAG_FMA4; + } + } + + if (!strncmp(vendor.c, "GenuineIntel", 12)) { + if (family == 6 && (model == 9 || model == 13 || model == 14)) { + /* 6/9 (pentium-m "banias"), 6/13 (pentium-m "dothan"), and + * 6/14 (core1 "yonah") theoretically support sse2, but it's + * usually slower than mmx, so let's just pretend they don't. + * AV_CPU_FLAG_SSE2 is disabled and AV_CPU_FLAG_SSE2SLOW is + * enabled so that SSE2 is not used unless explicitly enabled + * by checking AV_CPU_FLAG_SSE2SLOW. The same situation + * applies for AV_CPU_FLAG_SSE3 and AV_CPU_FLAG_SSE3SLOW. */ + if (rval & AV_CPU_FLAG_SSE2) + rval ^= AV_CPU_FLAG_SSE2SLOW | AV_CPU_FLAG_SSE2; + if (rval & AV_CPU_FLAG_SSE3) + rval ^= AV_CPU_FLAG_SSE3SLOW | AV_CPU_FLAG_SSE3; + } + /* The Atom processor has SSSE3 support, which is useful in many cases, + * but sometimes the SSSE3 version is slower than the SSE2 equivalent + * on the Atom, but is generally faster on other processors supporting + * SSSE3. This flag allows for selectively disabling certain SSSE3 + * functions on the Atom. */ + if (family == 6 && model == 28) + rval |= AV_CPU_FLAG_ATOM; + + /* Conroe has a slow shuffle unit. Check the model number to ensure not + * to include crippled low-end Penryns and Nehalems that lack SSE4. */ + if ((rval & AV_CPU_FLAG_SSSE3) && !(rval & AV_CPU_FLAG_SSE4) && + family == 6 && model < 23) + rval |= AV_CPU_FLAG_SSSE3SLOW; + + /* Haswell has slow gather */ + if ((rval & AV_CPU_FLAG_AVX2) && family == 6 && model < 70) + rval |= AV_CPU_FLAG_SLOW_GATHER; + } + +#endif /* cpuid */ + + return rval; +} + +size_t ff_get_cpu_max_align_x86(void) +{ + int flags = av_get_cpu_flags(); + + if (flags & AV_CPU_FLAG_AVX512) + return 64; + if (flags & (AV_CPU_FLAG_AVX2 | + AV_CPU_FLAG_AVX | + AV_CPU_FLAG_XOP | + AV_CPU_FLAG_FMA4 | + AV_CPU_FLAG_FMA3 | + AV_CPU_FLAG_AVXSLOW)) + return 32; + if (flags & (AV_CPU_FLAG_AESNI | + AV_CPU_FLAG_SSE42 | + AV_CPU_FLAG_SSE4 | + AV_CPU_FLAG_SSSE3 | + AV_CPU_FLAG_SSE3 | + AV_CPU_FLAG_SSE2 | + AV_CPU_FLAG_SSE | + AV_CPU_FLAG_ATOM | + AV_CPU_FLAG_SSSE3SLOW | + AV_CPU_FLAG_SSE3SLOW | + AV_CPU_FLAG_SSE2SLOW)) + return 16; + + return 8; +} diff --git a/media/ffvpx/libavutil/x86/cpu.h b/media/ffvpx/libavutil/x86/cpu.h new file mode 100644 index 0000000000..40a1eef0ab --- /dev/null +++ b/media/ffvpx/libavutil/x86/cpu.h @@ -0,0 +1,114 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_CPU_H +#define AVUTIL_X86_CPU_H + +#include "libavutil/cpu.h" +#include "libavutil/cpu_internal.h" + +#define AV_CPU_FLAG_AMD3DNOW AV_CPU_FLAG_3DNOW +#define AV_CPU_FLAG_AMD3DNOWEXT AV_CPU_FLAG_3DNOWEXT + +#define X86_AMD3DNOW(flags) CPUEXT(flags, AMD3DNOW) +#define X86_AMD3DNOWEXT(flags) CPUEXT(flags, AMD3DNOWEXT) +#define X86_MMX(flags) CPUEXT(flags, MMX) +#define X86_MMXEXT(flags) CPUEXT(flags, MMXEXT) +#define X86_SSE(flags) CPUEXT(flags, SSE) +#define X86_SSE2(flags) CPUEXT(flags, SSE2) +#define X86_SSE2_FAST(flags) CPUEXT_FAST(flags, SSE2) +#define X86_SSE2_SLOW(flags) CPUEXT_SLOW(flags, SSE2) +#define X86_SSE3(flags) CPUEXT(flags, SSE3) +#define X86_SSE3_FAST(flags) CPUEXT_FAST(flags, SSE3) +#define X86_SSE3_SLOW(flags) CPUEXT_SLOW(flags, SSE3) +#define X86_SSSE3(flags) CPUEXT(flags, SSSE3) +#define X86_SSSE3_FAST(flags) CPUEXT_FAST(flags, SSSE3) +#define X86_SSSE3_SLOW(flags) CPUEXT_SLOW(flags, SSSE3) +#define X86_SSE4(flags) CPUEXT(flags, SSE4) +#define X86_SSE42(flags) CPUEXT(flags, SSE42) +#define X86_AVX(flags) CPUEXT(flags, AVX) +#define X86_AVX_FAST(flags) CPUEXT_FAST(flags, AVX) +#define X86_AVX_SLOW(flags) CPUEXT_SLOW(flags, AVX) +#define X86_XOP(flags) CPUEXT(flags, XOP) +#define X86_FMA3(flags) CPUEXT(flags, FMA3) +#define X86_FMA4(flags) CPUEXT(flags, FMA4) +#define X86_AVX2(flags) CPUEXT(flags, AVX2) +#define X86_AESNI(flags) CPUEXT(flags, AESNI) +#define X86_AVX512(flags) CPUEXT(flags, AVX512) + +#define EXTERNAL_AMD3DNOW(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AMD3DNOW) +#define EXTERNAL_AMD3DNOWEXT(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AMD3DNOWEXT) +#define EXTERNAL_MMX(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, MMX) +#define EXTERNAL_MMXEXT(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, MMXEXT) +#define EXTERNAL_SSE(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSE) +#define EXTERNAL_SSE2(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSE2) +#define EXTERNAL_SSE2_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _EXTERNAL, SSE2) +#define EXTERNAL_SSE2_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _EXTERNAL, SSE2) +#define EXTERNAL_SSE3(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSE3) +#define EXTERNAL_SSE3_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _EXTERNAL, SSE3) +#define EXTERNAL_SSE3_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _EXTERNAL, SSE3) +#define EXTERNAL_SSSE3(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSSE3) +#define EXTERNAL_SSSE3_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _EXTERNAL, SSSE3) +#define EXTERNAL_SSSE3_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _EXTERNAL, SSSE3) +#define EXTERNAL_SSE4(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSE4) +#define EXTERNAL_SSE42(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, SSE42) +#define EXTERNAL_AVX(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AVX) +#define EXTERNAL_AVX_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _EXTERNAL, AVX) +#define EXTERNAL_AVX_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _EXTERNAL, AVX) +#define EXTERNAL_XOP(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, XOP) +#define EXTERNAL_FMA3(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, FMA3) +#define EXTERNAL_FMA3_FAST(flags) CPUEXT_SUFFIX_FAST2(flags, _EXTERNAL, FMA3, AVX) +#define EXTERNAL_FMA3_SLOW(flags) CPUEXT_SUFFIX_SLOW2(flags, _EXTERNAL, FMA3, AVX) +#define EXTERNAL_FMA4(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, FMA4) +#define EXTERNAL_AVX2(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AVX2) +#define EXTERNAL_AVX2_FAST(flags) CPUEXT_SUFFIX_FAST2(flags, _EXTERNAL, AVX2, AVX) +#define EXTERNAL_AVX2_SLOW(flags) CPUEXT_SUFFIX_SLOW2(flags, _EXTERNAL, AVX2, AVX) +#define EXTERNAL_AESNI(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AESNI) +#define EXTERNAL_AVX512(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AVX512) +#define EXTERNAL_AVX512ICL(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AVX512ICL) + +#define INLINE_AMD3DNOW(flags) CPUEXT_SUFFIX(flags, _INLINE, AMD3DNOW) +#define INLINE_AMD3DNOWEXT(flags) CPUEXT_SUFFIX(flags, _INLINE, AMD3DNOWEXT) +#define INLINE_MMX(flags) CPUEXT_SUFFIX(flags, _INLINE, MMX) +#define INLINE_MMXEXT(flags) CPUEXT_SUFFIX(flags, _INLINE, MMXEXT) +#define INLINE_SSE(flags) CPUEXT_SUFFIX(flags, _INLINE, SSE) +#define INLINE_SSE2(flags) CPUEXT_SUFFIX(flags, _INLINE, SSE2) +#define INLINE_SSE2_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _INLINE, SSE2) +#define INLINE_SSE2_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _INLINE, SSE2) +#define INLINE_SSE3(flags) CPUEXT_SUFFIX(flags, _INLINE, SSE3) +#define INLINE_SSE3_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _INLINE, SSE3) +#define INLINE_SSE3_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _INLINE, SSE3) +#define INLINE_SSSE3(flags) CPUEXT_SUFFIX(flags, _INLINE, SSSE3) +#define INLINE_SSSE3_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _INLINE, SSSE3) +#define INLINE_SSSE3_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _INLINE, SSSE3) +#define INLINE_SSE4(flags) CPUEXT_SUFFIX(flags, _INLINE, SSE4) +#define INLINE_SSE42(flags) CPUEXT_SUFFIX(flags, _INLINE, SSE42) +#define INLINE_AVX(flags) CPUEXT_SUFFIX(flags, _INLINE, AVX) +#define INLINE_AVX_FAST(flags) CPUEXT_SUFFIX_FAST(flags, _INLINE, AVX) +#define INLINE_AVX_SLOW(flags) CPUEXT_SUFFIX_SLOW(flags, _INLINE, AVX) +#define INLINE_XOP(flags) CPUEXT_SUFFIX(flags, _INLINE, XOP) +#define INLINE_FMA3(flags) CPUEXT_SUFFIX(flags, _INLINE, FMA3) +#define INLINE_FMA4(flags) CPUEXT_SUFFIX(flags, _INLINE, FMA4) +#define INLINE_AVX2(flags) CPUEXT_SUFFIX(flags, _INLINE, AVX2) +#define INLINE_AESNI(flags) CPUEXT_SUFFIX(flags, _INLINE, AESNI) + +void ff_cpu_cpuid(int index, int *eax, int *ebx, int *ecx, int *edx); +void ff_cpu_xgetbv(int op, int *eax, int *edx); +int ff_cpu_cpuid_test(void); + +#endif /* AVUTIL_X86_CPU_H */ diff --git a/media/ffvpx/libavutil/x86/cpuid.asm b/media/ffvpx/libavutil/x86/cpuid.asm new file mode 100644 index 0000000000..766f77fcdf --- /dev/null +++ b/media/ffvpx/libavutil/x86/cpuid.asm @@ -0,0 +1,91 @@ +;***************************************************************************** +;* Copyright (C) 2005-2010 x264 project +;* +;* Authors: Loren Merritt <lorenm@u.washington.edu> +;* Fiona Glaser <fiona@x264.com> +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +;----------------------------------------------------------------------------- +; void ff_cpu_cpuid(int index, int *eax, int *ebx, int *ecx, int *edx) +;----------------------------------------------------------------------------- +cglobal cpu_cpuid, 5,7 + push rbx + push r4 + push r3 + push r2 + push r1 + mov eax, r0d + xor ecx, ecx + cpuid + pop r4 + mov [r4], eax + pop r4 + mov [r4], ebx + pop r4 + mov [r4], ecx + pop r4 + mov [r4], edx + pop rbx + RET + +;----------------------------------------------------------------------------- +; void ff_cpu_xgetbv(int op, int *eax, int *edx) +;----------------------------------------------------------------------------- +cglobal cpu_xgetbv, 3,7 + push r2 + push r1 + mov ecx, r0d + xgetbv + pop r4 + mov [r4], eax + pop r4 + mov [r4], edx + RET + +%if ARCH_X86_64 == 0 +;----------------------------------------------------------------------------- +; int ff_cpu_cpuid_test(void) +; return 0 if unsupported +;----------------------------------------------------------------------------- +cglobal cpu_cpuid_test + pushfd + push ebx + push ebp + push esi + push edi + pushfd + pop eax + mov ebx, eax + xor eax, 0x200000 + push eax + popfd + pushfd + pop eax + xor eax, ebx + pop edi + pop esi + pop ebp + pop ebx + popfd + ret +%endif diff --git a/media/ffvpx/libavutil/x86/emms.asm b/media/ffvpx/libavutil/x86/emms.asm new file mode 100644 index 0000000000..df84f2221b --- /dev/null +++ b/media/ffvpx/libavutil/x86/emms.asm @@ -0,0 +1,30 @@ +;***************************************************************************** +;* Copyright (C) 2013 Martin Storsjo +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +;----------------------------------------------------------------------------- +; void avpriv_emms_asm(void) +;----------------------------------------------------------------------------- +cvisible emms_asm, 0, 0 + emms + RET diff --git a/media/ffvpx/libavutil/x86/emms.h b/media/ffvpx/libavutil/x86/emms.h new file mode 100644 index 0000000000..8ceec110cf --- /dev/null +++ b/media/ffvpx/libavutil/x86/emms.h @@ -0,0 +1,58 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_EMMS_H +#define AVUTIL_X86_EMMS_H + +#include "config.h" +#include "libavutil/attributes.h" + +void avpriv_emms_asm(void); + +#if HAVE_MMX_INLINE +#ifndef __MMX__ +#include "libavutil/cpu.h" +#endif + +# define emms_c emms_c +/** + * Empty mmx state. + * this must be called between any dsp function and float/double code. + * for example sin(); dsp->idct_put(); emms_c(); cos() + * Note, *alloc() and *free() also use float code in some libc implementations + * thus this also applies to them or any function using them. + */ +static av_always_inline void emms_c(void) +{ +/* Some inlined functions may also use mmx instructions regardless of + * runtime cpuflags. With that in mind, we unconditionally empty the + * mmx state if the target cpu chosen at configure time supports it. + */ +#if !defined(__MMX__) + if(av_get_cpu_flags() & AV_CPU_FLAG_MMX) +#endif + __asm__ volatile ("emms" ::: "memory"); +} +#elif HAVE_MMX && HAVE_MM_EMPTY +# include <mmintrin.h> +# define emms_c _mm_empty +#elif HAVE_MMX_EXTERNAL +# define emms_c avpriv_emms_asm +#endif /* HAVE_MMX_INLINE */ + +#endif /* AVUTIL_X86_EMMS_H */ diff --git a/media/ffvpx/libavutil/x86/fixed_dsp.asm b/media/ffvpx/libavutil/x86/fixed_dsp.asm new file mode 100644 index 0000000000..2f411850f4 --- /dev/null +++ b/media/ffvpx/libavutil/x86/fixed_dsp.asm @@ -0,0 +1,48 @@ +;***************************************************************************** +;* x86-optimized Float DSP functions +;* +;* Copyright 2016 James Almer +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +;----------------------------------------------------------------------------- +; void ff_butterflies_fixed(float *src0, float *src1, int len); +;----------------------------------------------------------------------------- +INIT_XMM sse2 +cglobal butterflies_fixed, 3,3,3, src0, src1, len + shl lend, 2 + add src0q, lenq + add src1q, lenq + neg lenq + +align 16 +.loop: + mova m0, [src0q + lenq] + mova m1, [src1q + lenq] + mova m2, m0 + paddd m0, m1 + psubd m2, m1 + mova [src0q + lenq], m0 + mova [src1q + lenq], m2 + add lenq, mmsize + jl .loop + RET diff --git a/media/ffvpx/libavutil/x86/fixed_dsp_init.c b/media/ffvpx/libavutil/x86/fixed_dsp_init.c new file mode 100644 index 0000000000..d3f4b2e325 --- /dev/null +++ b/media/ffvpx/libavutil/x86/fixed_dsp_init.c @@ -0,0 +1,35 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include "libavutil/attributes.h" +#include "libavutil/cpu.h" +#include "libavutil/fixed_dsp.h" +#include "cpu.h" + +void ff_butterflies_fixed_sse2(int *av_restrict src0, int *av_restrict src1, int len); + +av_cold void ff_fixed_dsp_init_x86(AVFixedDSPContext *fdsp) +{ + int cpu_flags = av_get_cpu_flags(); + + if (EXTERNAL_SSE2(cpu_flags)) { + fdsp->butterflies_fixed = ff_butterflies_fixed_sse2; + } +} diff --git a/media/ffvpx/libavutil/x86/float_dsp.asm b/media/ffvpx/libavutil/x86/float_dsp.asm new file mode 100644 index 0000000000..e84ba52566 --- /dev/null +++ b/media/ffvpx/libavutil/x86/float_dsp.asm @@ -0,0 +1,588 @@ +;***************************************************************************** +;* x86-optimized Float DSP functions +;* +;* Copyright 2006 Loren Merritt +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION_RODATA 32 +pd_reverse: dd 7, 6, 5, 4, 3, 2, 1, 0 + +SECTION .text + +;----------------------------------------------------------------------------- +; void vector_fmul(float *dst, const float *src0, const float *src1, int len) +;----------------------------------------------------------------------------- +%macro VECTOR_FMUL 0 +cglobal vector_fmul, 4,4,2, dst, src0, src1, len + lea lenq, [lend*4 - 64] +ALIGN 16 +.loop: +%assign a 0 +%rep 32/mmsize + mova m0, [src0q + lenq + (a+0)*mmsize] + mova m1, [src0q + lenq + (a+1)*mmsize] + mulps m0, m0, [src1q + lenq + (a+0)*mmsize] + mulps m1, m1, [src1q + lenq + (a+1)*mmsize] + mova [dstq + lenq + (a+0)*mmsize], m0 + mova [dstq + lenq + (a+1)*mmsize], m1 +%assign a a+2 +%endrep + + sub lenq, 64 + jge .loop + RET +%endmacro + +INIT_XMM sse +VECTOR_FMUL +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_FMUL +%endif + +;----------------------------------------------------------------------------- +; void vector_dmul(double *dst, const double *src0, const double *src1, int len) +;----------------------------------------------------------------------------- +%macro VECTOR_DMUL 0 +cglobal vector_dmul, 4,4,4, dst, src0, src1, len + lea lend, [lenq*8 - mmsize*4] +ALIGN 16 +.loop: + movaps m0, [src0q + lenq + 0*mmsize] + movaps m1, [src0q + lenq + 1*mmsize] + movaps m2, [src0q + lenq + 2*mmsize] + movaps m3, [src0q + lenq + 3*mmsize] + mulpd m0, m0, [src1q + lenq + 0*mmsize] + mulpd m1, m1, [src1q + lenq + 1*mmsize] + mulpd m2, m2, [src1q + lenq + 2*mmsize] + mulpd m3, m3, [src1q + lenq + 3*mmsize] + movaps [dstq + lenq + 0*mmsize], m0 + movaps [dstq + lenq + 1*mmsize], m1 + movaps [dstq + lenq + 2*mmsize], m2 + movaps [dstq + lenq + 3*mmsize], m3 + + sub lenq, mmsize*4 + jge .loop + RET +%endmacro + +INIT_XMM sse2 +VECTOR_DMUL +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_DMUL +%endif + +;------------------------------------------------------------------------------ +; void ff_vector_fmac_scalar(float *dst, const float *src, float mul, int len) +;------------------------------------------------------------------------------ + +%macro VECTOR_FMAC_SCALAR 0 +%if UNIX64 +cglobal vector_fmac_scalar, 3,3,5, dst, src, len +%else +cglobal vector_fmac_scalar, 4,4,5, dst, src, mul, len +%endif +%if ARCH_X86_32 + VBROADCASTSS m0, mulm +%else +%if WIN64 + SWAP 0, 2 +%endif + shufps xm0, xm0, 0 +%if cpuflag(avx) + vinsertf128 m0, m0, xm0, 1 +%endif +%endif + lea lenq, [lend*4-64] +.loop: +%if cpuflag(fma3) + mova m1, [dstq+lenq] + mova m2, [dstq+lenq+1*mmsize] + fmaddps m1, m0, [srcq+lenq], m1 + fmaddps m2, m0, [srcq+lenq+1*mmsize], m2 +%else ; cpuflag + mulps m1, m0, [srcq+lenq] + mulps m2, m0, [srcq+lenq+1*mmsize] +%if mmsize < 32 + mulps m3, m0, [srcq+lenq+2*mmsize] + mulps m4, m0, [srcq+lenq+3*mmsize] +%endif ; mmsize + addps m1, m1, [dstq+lenq] + addps m2, m2, [dstq+lenq+1*mmsize] +%if mmsize < 32 + addps m3, m3, [dstq+lenq+2*mmsize] + addps m4, m4, [dstq+lenq+3*mmsize] +%endif ; mmsize +%endif ; cpuflag + mova [dstq+lenq], m1 + mova [dstq+lenq+1*mmsize], m2 +%if mmsize < 32 + mova [dstq+lenq+2*mmsize], m3 + mova [dstq+lenq+3*mmsize], m4 +%endif ; mmsize + sub lenq, 64 + jge .loop + RET +%endmacro + +INIT_XMM sse +VECTOR_FMAC_SCALAR +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_FMAC_SCALAR +%endif +%if HAVE_FMA3_EXTERNAL +INIT_YMM fma3 +VECTOR_FMAC_SCALAR +%endif + +;------------------------------------------------------------------------------ +; void ff_vector_fmul_scalar(float *dst, const float *src, float mul, int len) +;------------------------------------------------------------------------------ + +%macro VECTOR_FMUL_SCALAR 0 +%if UNIX64 +cglobal vector_fmul_scalar, 3,3,2, dst, src, len +%else +cglobal vector_fmul_scalar, 4,4,3, dst, src, mul, len +%endif +%if ARCH_X86_32 + movss m0, mulm +%elif WIN64 + SWAP 0, 2 +%endif + shufps m0, m0, 0 + lea lenq, [lend*4-mmsize] +.loop: + mova m1, [srcq+lenq] + mulps m1, m0 + mova [dstq+lenq], m1 + sub lenq, mmsize + jge .loop + RET +%endmacro + +INIT_XMM sse +VECTOR_FMUL_SCALAR + +;------------------------------------------------------------------------------ +; void ff_vector_dmac_scalar(double *dst, const double *src, double mul, +; int len) +;------------------------------------------------------------------------------ + +%macro VECTOR_DMAC_SCALAR 0 +%if ARCH_X86_32 +cglobal vector_dmac_scalar, 2,4,5, dst, src, mul, len, lenaddr + mov lenq, lenaddrm + VBROADCASTSD m0, mulm +%else +%if UNIX64 +cglobal vector_dmac_scalar, 3,3,5, dst, src, len +%else +cglobal vector_dmac_scalar, 4,4,5, dst, src, mul, len + SWAP 0, 2 +%endif + movlhps xm0, xm0 +%if cpuflag(avx) + vinsertf128 m0, m0, xm0, 1 +%endif +%endif + lea lenq, [lend*8-mmsize*4] +.loop: +%if cpuflag(fma3) + movaps m1, [dstq+lenq] + movaps m2, [dstq+lenq+1*mmsize] + movaps m3, [dstq+lenq+2*mmsize] + movaps m4, [dstq+lenq+3*mmsize] + fmaddpd m1, m0, [srcq+lenq], m1 + fmaddpd m2, m0, [srcq+lenq+1*mmsize], m2 + fmaddpd m3, m0, [srcq+lenq+2*mmsize], m3 + fmaddpd m4, m0, [srcq+lenq+3*mmsize], m4 +%else ; cpuflag + mulpd m1, m0, [srcq+lenq] + mulpd m2, m0, [srcq+lenq+1*mmsize] + mulpd m3, m0, [srcq+lenq+2*mmsize] + mulpd m4, m0, [srcq+lenq+3*mmsize] + addpd m1, m1, [dstq+lenq] + addpd m2, m2, [dstq+lenq+1*mmsize] + addpd m3, m3, [dstq+lenq+2*mmsize] + addpd m4, m4, [dstq+lenq+3*mmsize] +%endif ; cpuflag + movaps [dstq+lenq], m1 + movaps [dstq+lenq+1*mmsize], m2 + movaps [dstq+lenq+2*mmsize], m3 + movaps [dstq+lenq+3*mmsize], m4 + sub lenq, mmsize*4 + jge .loop + RET +%endmacro + +INIT_XMM sse2 +VECTOR_DMAC_SCALAR +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_DMAC_SCALAR +%endif +%if HAVE_FMA3_EXTERNAL +INIT_YMM fma3 +VECTOR_DMAC_SCALAR +%endif + +;------------------------------------------------------------------------------ +; void ff_vector_dmul_scalar(double *dst, const double *src, double mul, +; int len) +;------------------------------------------------------------------------------ + +%macro VECTOR_DMUL_SCALAR 0 +%if ARCH_X86_32 +cglobal vector_dmul_scalar, 3,4,3, dst, src, mul, len, lenaddr + mov lenq, lenaddrm +%elif UNIX64 +cglobal vector_dmul_scalar, 3,3,3, dst, src, len +%else +cglobal vector_dmul_scalar, 4,4,3, dst, src, mul, len +%endif +%if ARCH_X86_32 + VBROADCASTSD m0, mulm +%else +%if WIN64 + SWAP 0, 2 +%endif + movlhps xm0, xm0 +%if cpuflag(avx) + vinsertf128 ym0, ym0, xm0, 1 +%endif +%endif + lea lenq, [lend*8-2*mmsize] +.loop: + mulpd m1, m0, [srcq+lenq ] + mulpd m2, m0, [srcq+lenq+mmsize] + movaps [dstq+lenq ], m1 + movaps [dstq+lenq+mmsize], m2 + sub lenq, 2*mmsize + jge .loop + RET +%endmacro + +INIT_XMM sse2 +VECTOR_DMUL_SCALAR +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_DMUL_SCALAR +%endif + +;----------------------------------------------------------------------------- +; vector_fmul_window(float *dst, const float *src0, +; const float *src1, const float *win, int len); +;----------------------------------------------------------------------------- +INIT_XMM sse +cglobal vector_fmul_window, 5, 6, 6, dst, src0, src1, win, len, len1 + shl lend, 2 + lea len1q, [lenq - mmsize] + add src0q, lenq + add dstq, lenq + add winq, lenq + neg lenq +.loop: + mova m0, [winq + lenq] + mova m4, [src0q + lenq] + mova m1, [winq + len1q] + mova m5, [src1q + len1q] + shufps m1, m1, 0x1b + shufps m5, m5, 0x1b + mova m2, m0 + mova m3, m1 + mulps m2, m4 + mulps m3, m5 + mulps m1, m4 + mulps m0, m5 + addps m2, m3 + subps m1, m0 + shufps m2, m2, 0x1b + mova [dstq + lenq], m1 + mova [dstq + len1q], m2 + sub len1q, mmsize + add lenq, mmsize + jl .loop + RET + +;----------------------------------------------------------------------------- +; vector_fmul_add(float *dst, const float *src0, const float *src1, +; const float *src2, int len) +;----------------------------------------------------------------------------- +%macro VECTOR_FMUL_ADD 0 +cglobal vector_fmul_add, 5,5,4, dst, src0, src1, src2, len + lea lenq, [lend*4 - 2*mmsize] +ALIGN 16 +.loop: + mova m0, [src0q + lenq] + mova m1, [src0q + lenq + mmsize] +%if cpuflag(fma3) + mova m2, [src2q + lenq] + mova m3, [src2q + lenq + mmsize] + fmaddps m0, m0, [src1q + lenq], m2 + fmaddps m1, m1, [src1q + lenq + mmsize], m3 +%else + mulps m0, m0, [src1q + lenq] + mulps m1, m1, [src1q + lenq + mmsize] + addps m0, m0, [src2q + lenq] + addps m1, m1, [src2q + lenq + mmsize] +%endif + mova [dstq + lenq], m0 + mova [dstq + lenq + mmsize], m1 + + sub lenq, 2*mmsize + jge .loop + RET +%endmacro + +INIT_XMM sse +VECTOR_FMUL_ADD +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_FMUL_ADD +%endif +%if HAVE_FMA3_EXTERNAL +INIT_YMM fma3 +VECTOR_FMUL_ADD +%endif + +;----------------------------------------------------------------------------- +; void vector_fmul_reverse(float *dst, const float *src0, const float *src1, +; int len) +;----------------------------------------------------------------------------- +%macro VECTOR_FMUL_REVERSE 0 +cglobal vector_fmul_reverse, 4,4,2, dst, src0, src1, len +%if cpuflag(avx2) + movaps m2, [pd_reverse] +%endif + lea lenq, [lend*4 - 2*mmsize] +ALIGN 16 +.loop: +%if cpuflag(avx2) + vpermps m0, m2, [src1q] + vpermps m1, m2, [src1q+mmsize] +%elif cpuflag(avx) + vmovaps xmm0, [src1q + 16] + vinsertf128 m0, m0, [src1q], 1 + vshufps m0, m0, m0, q0123 + vmovaps xmm1, [src1q + mmsize + 16] + vinsertf128 m1, m1, [src1q + mmsize], 1 + vshufps m1, m1, m1, q0123 +%else + mova m0, [src1q] + mova m1, [src1q + mmsize] + shufps m0, m0, q0123 + shufps m1, m1, q0123 +%endif + mulps m0, m0, [src0q + lenq + mmsize] + mulps m1, m1, [src0q + lenq] + movaps [dstq + lenq + mmsize], m0 + movaps [dstq + lenq], m1 + add src1q, 2*mmsize + sub lenq, 2*mmsize + jge .loop + RET +%endmacro + +INIT_XMM sse +VECTOR_FMUL_REVERSE +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_FMUL_REVERSE +%endif +%if HAVE_AVX2_EXTERNAL +INIT_YMM avx2 +VECTOR_FMUL_REVERSE +%endif + +; float scalarproduct_float_sse(const float *v1, const float *v2, int len) +INIT_XMM sse +cglobal scalarproduct_float, 3,3,2, v1, v2, offset + shl offsetd, 2 + add v1q, offsetq + add v2q, offsetq + neg offsetq + xorps xmm0, xmm0 +.loop: + movaps xmm1, [v1q+offsetq] + mulps xmm1, [v2q+offsetq] + addps xmm0, xmm1 + add offsetq, 16 + js .loop + movhlps xmm1, xmm0 + addps xmm0, xmm1 + movss xmm1, xmm0 + shufps xmm0, xmm0, 1 + addss xmm0, xmm1 +%if ARCH_X86_64 == 0 + movss r0m, xmm0 + fld dword r0m +%endif + RET + +INIT_YMM fma3 +cglobal scalarproduct_float, 3,5,8, v1, v2, size, len, offset + xor offsetq, offsetq + xorps m0, m0, m0 + shl sized, 2 + mov lenq, sizeq + cmp lenq, 32 + jl .l16 + cmp lenq, 64 + jl .l32 + xorps m1, m1, m1 + cmp lenq, 128 + jl .l64 + and lenq, ~127 + xorps m2, m2, m2 + xorps m3, m3, m3 +.loop128: + movups m4, [v1q+offsetq] + movups m5, [v1q+offsetq + 32] + movups m6, [v1q+offsetq + 64] + movups m7, [v1q+offsetq + 96] + fmaddps m0, m4, [v2q+offsetq ], m0 + fmaddps m1, m5, [v2q+offsetq + 32], m1 + fmaddps m2, m6, [v2q+offsetq + 64], m2 + fmaddps m3, m7, [v2q+offsetq + 96], m3 + add offsetq, 128 + cmp offsetq, lenq + jl .loop128 + addps m0, m0, m2 + addps m1, m1, m3 + mov lenq, sizeq + and lenq, 127 + cmp lenq, 64 + jge .l64 + addps m0, m0, m1 + cmp lenq, 32 + jge .l32 + vextractf128 xmm2, m0, 1 + addps xmm0, xmm2 + cmp lenq, 16 + jge .l16 + movhlps xmm1, xmm0 + addps xmm0, xmm1 + movss xmm1, xmm0 + shufps xmm0, xmm0, 1 + addss xmm0, xmm1 +%if ARCH_X86_64 == 0 + movss r0m, xm0 + fld dword r0m +%endif + RET +.l64: + and lenq, ~63 + add lenq, offsetq +.loop64: + movups m4, [v1q+offsetq] + movups m5, [v1q+offsetq + 32] + fmaddps m0, m4, [v2q+offsetq], m0 + fmaddps m1, m5, [v2q+offsetq + 32], m1 + add offsetq, 64 + cmp offsetq, lenq + jl .loop64 + addps m0, m0, m1 + mov lenq, sizeq + and lenq, 63 + cmp lenq, 32 + jge .l32 + vextractf128 xmm2, m0, 1 + addps xmm0, xmm2 + cmp lenq, 16 + jge .l16 + movhlps xmm1, xmm0 + addps xmm0, xmm1 + movss xmm1, xmm0 + shufps xmm0, xmm0, 1 + addss xmm0, xmm1 +%if ARCH_X86_64 == 0 + movss r0m, xm0 + fld dword r0m +%endif + RET +.l32: + and lenq, ~31 + add lenq, offsetq +.loop32: + movups m4, [v1q+offsetq] + fmaddps m0, m4, [v2q+offsetq], m0 + add offsetq, 32 + cmp offsetq, lenq + jl .loop32 + vextractf128 xmm2, m0, 1 + addps xmm0, xmm2 + mov lenq, sizeq + and lenq, 31 + cmp lenq, 16 + jge .l16 + movhlps xmm1, xmm0 + addps xmm0, xmm1 + movss xmm1, xmm0 + shufps xmm0, xmm0, 1 + addss xmm0, xmm1 +%if ARCH_X86_64 == 0 + movss r0m, xm0 + fld dword r0m +%endif + RET +.l16: + and lenq, ~15 + add lenq, offsetq +.loop16: + movaps xmm1, [v1q+offsetq] + mulps xmm1, [v2q+offsetq] + addps xmm0, xmm1 + add offsetq, 16 + cmp offsetq, lenq + jl .loop16 + movhlps xmm1, xmm0 + addps xmm0, xmm1 + movss xmm1, xmm0 + shufps xmm0, xmm0, 1 + addss xmm0, xmm1 +%if ARCH_X86_64 == 0 + movss r0m, xm0 + fld dword r0m +%endif + RET + +;----------------------------------------------------------------------------- +; void ff_butterflies_float(float *src0, float *src1, int len); +;----------------------------------------------------------------------------- +INIT_XMM sse +cglobal butterflies_float, 3,3,3, src0, src1, len + shl lend, 2 + add src0q, lenq + add src1q, lenq + neg lenq +.loop: + mova m0, [src0q + lenq] + mova m1, [src1q + lenq] + subps m2, m0, m1 + addps m0, m0, m1 + mova [src1q + lenq], m2 + mova [src0q + lenq], m0 + add lenq, mmsize + jl .loop + RET diff --git a/media/ffvpx/libavutil/x86/float_dsp_init.c b/media/ffvpx/libavutil/x86/float_dsp_init.c new file mode 100644 index 0000000000..ad6b506259 --- /dev/null +++ b/media/ffvpx/libavutil/x86/float_dsp_init.c @@ -0,0 +1,118 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include "libavutil/attributes.h" +#include "libavutil/cpu.h" +#include "libavutil/float_dsp.h" +#include "cpu.h" +#include "asm.h" + +void ff_vector_fmul_sse(float *dst, const float *src0, const float *src1, + int len); +void ff_vector_fmul_avx(float *dst, const float *src0, const float *src1, + int len); + +void ff_vector_dmul_sse2(double *dst, const double *src0, const double *src1, + int len); +void ff_vector_dmul_avx(double *dst, const double *src0, const double *src1, + int len); + +void ff_vector_fmac_scalar_sse(float *dst, const float *src, float mul, + int len); +void ff_vector_fmac_scalar_avx(float *dst, const float *src, float mul, + int len); +void ff_vector_fmac_scalar_fma3(float *dst, const float *src, float mul, + int len); + +void ff_vector_fmul_scalar_sse(float *dst, const float *src, float mul, + int len); + +void ff_vector_dmac_scalar_sse2(double *dst, const double *src, double mul, + int len); +void ff_vector_dmac_scalar_avx(double *dst, const double *src, double mul, + int len); +void ff_vector_dmac_scalar_fma3(double *dst, const double *src, double mul, + int len); + +void ff_vector_dmul_scalar_sse2(double *dst, const double *src, + double mul, int len); +void ff_vector_dmul_scalar_avx(double *dst, const double *src, + double mul, int len); + +void ff_vector_fmul_window_sse(float *dst, const float *src0, + const float *src1, const float *win, int len); + +void ff_vector_fmul_add_sse(float *dst, const float *src0, const float *src1, + const float *src2, int len); +void ff_vector_fmul_add_avx(float *dst, const float *src0, const float *src1, + const float *src2, int len); +void ff_vector_fmul_add_fma3(float *dst, const float *src0, const float *src1, + const float *src2, int len); + +void ff_vector_fmul_reverse_sse(float *dst, const float *src0, + const float *src1, int len); +void ff_vector_fmul_reverse_avx(float *dst, const float *src0, + const float *src1, int len); +void ff_vector_fmul_reverse_avx2(float *dst, const float *src0, + const float *src1, int len); + +float ff_scalarproduct_float_sse(const float *v1, const float *v2, int order); +float ff_scalarproduct_float_fma3(const float *v1, const float *v2, int order); + +void ff_butterflies_float_sse(float *av_restrict src0, float *av_restrict src1, int len); + +av_cold void ff_float_dsp_init_x86(AVFloatDSPContext *fdsp) +{ + int cpu_flags = av_get_cpu_flags(); + + if (EXTERNAL_SSE(cpu_flags)) { + fdsp->vector_fmul = ff_vector_fmul_sse; + fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_sse; + fdsp->vector_fmul_scalar = ff_vector_fmul_scalar_sse; + fdsp->vector_fmul_window = ff_vector_fmul_window_sse; + fdsp->vector_fmul_add = ff_vector_fmul_add_sse; + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_sse; + fdsp->scalarproduct_float = ff_scalarproduct_float_sse; + fdsp->butterflies_float = ff_butterflies_float_sse; + } + if (EXTERNAL_SSE2(cpu_flags)) { + fdsp->vector_dmul = ff_vector_dmul_sse2; + fdsp->vector_dmac_scalar = ff_vector_dmac_scalar_sse2; + fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_sse2; + } + if (EXTERNAL_AVX_FAST(cpu_flags)) { + fdsp->vector_fmul = ff_vector_fmul_avx; + fdsp->vector_dmul = ff_vector_dmul_avx; + fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_avx; + fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_avx; + fdsp->vector_dmac_scalar = ff_vector_dmac_scalar_avx; + fdsp->vector_fmul_add = ff_vector_fmul_add_avx; + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_avx; + } + if (EXTERNAL_AVX2_FAST(cpu_flags)) { + fdsp->vector_fmul_reverse = ff_vector_fmul_reverse_avx2; + } + if (EXTERNAL_FMA3_FAST(cpu_flags)) { + fdsp->vector_fmac_scalar = ff_vector_fmac_scalar_fma3; + fdsp->vector_fmul_add = ff_vector_fmul_add_fma3; + fdsp->vector_dmac_scalar = ff_vector_dmac_scalar_fma3; + fdsp->scalarproduct_float = ff_scalarproduct_float_fma3; + } +} diff --git a/media/ffvpx/libavutil/x86/imgutils.asm b/media/ffvpx/libavutil/x86/imgutils.asm new file mode 100644 index 0000000000..3cca56cdca --- /dev/null +++ b/media/ffvpx/libavutil/x86/imgutils.asm @@ -0,0 +1,53 @@ +;***************************************************************************** +;* Copyright 2016 Anton Khirnov +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +INIT_XMM sse4 +cglobal image_copy_plane_uc_from, 6, 7, 4, dst, dst_linesize, src, src_linesize, bw, height, rowpos + add dstq, bwq + add srcq, bwq + neg bwq + +.row_start: + mov rowposq, bwq + +.loop: + movntdqa m0, [srcq + rowposq + 0 * mmsize] + movntdqa m1, [srcq + rowposq + 1 * mmsize] + movntdqa m2, [srcq + rowposq + 2 * mmsize] + movntdqa m3, [srcq + rowposq + 3 * mmsize] + + mova [dstq + rowposq + 0 * mmsize], m0 + mova [dstq + rowposq + 1 * mmsize], m1 + mova [dstq + rowposq + 2 * mmsize], m2 + mova [dstq + rowposq + 3 * mmsize], m3 + + add rowposq, 4 * mmsize + jnz .loop + + add srcq, src_linesizeq + add dstq, dst_linesizeq + dec heightd + jnz .row_start + + RET diff --git a/media/ffvpx/libavutil/x86/imgutils_init.c b/media/ffvpx/libavutil/x86/imgutils_init.c new file mode 100644 index 0000000000..91a16cf594 --- /dev/null +++ b/media/ffvpx/libavutil/x86/imgutils_init.c @@ -0,0 +1,48 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <stddef.h> +#include <stdint.h> + +#include "libavutil/cpu.h" +#include "libavutil/error.h" +#include "libavutil/imgutils_internal.h" +#include "libavutil/macros.h" + +#include "cpu.h" + +void ff_image_copy_plane_uc_from_sse4(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height); + +int ff_image_copy_plane_uc_from_x86(uint8_t *dst, ptrdiff_t dst_linesize, + const uint8_t *src, ptrdiff_t src_linesize, + ptrdiff_t bytewidth, int height) +{ + int cpu_flags = av_get_cpu_flags(); + ptrdiff_t bw_aligned = FFALIGN(bytewidth, 64); + + if (EXTERNAL_SSE4(cpu_flags) && + bw_aligned <= dst_linesize && bw_aligned <= src_linesize) + ff_image_copy_plane_uc_from_sse4(dst, dst_linesize, src, src_linesize, + bw_aligned, height); + else + return AVERROR(ENOSYS); + + return 0; +} diff --git a/media/ffvpx/libavutil/x86/intmath.h b/media/ffvpx/libavutil/x86/intmath.h new file mode 100644 index 0000000000..8a6b5ae261 --- /dev/null +++ b/media/ffvpx/libavutil/x86/intmath.h @@ -0,0 +1,169 @@ +/* + * Copyright (c) 2015 James Almer + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_INTMATH_H +#define AVUTIL_X86_INTMATH_H + +#include <stdint.h> +#include <stdlib.h> +#if HAVE_FAST_CLZ +#if defined(_MSC_VER) +#include <intrin.h> +#elif defined(__INTEL_COMPILER) +#include <immintrin.h> +#endif +#endif +#include "config.h" + +#if HAVE_FAST_CLZ +#if (defined(__INTEL_COMPILER) && (__INTEL_COMPILER>=1216)) || defined(_MSC_VER) +# if defined(__INTEL_COMPILER) +# define ff_log2(x) (_bit_scan_reverse((x)|1)) +# else +# define ff_log2 ff_log2_x86 +static av_always_inline av_const int ff_log2_x86(unsigned int v) +{ + unsigned long n; + _BitScanReverse(&n, v|1); + return n; +} +# endif +# define ff_log2_16bit av_log2 + +#if defined(__INTEL_COMPILER) || (defined(_MSC_VER) && (_MSC_VER >= 1700) && \ + (defined(__BMI__) || !defined(__clang__))) +# define ff_ctz(v) _tzcnt_u32(v) + +# if ARCH_X86_64 +# define ff_ctzll(v) _tzcnt_u64(v) +# else +# define ff_ctzll ff_ctzll_x86 +static av_always_inline av_const int ff_ctzll_x86(long long v) +{ + return ((uint32_t)v == 0) ? _tzcnt_u32((uint32_t)(v >> 32)) + 32 : _tzcnt_u32((uint32_t)v); +} +# endif +#endif /* _MSC_VER */ + +#endif /* __INTEL_COMPILER */ + +#endif /* HAVE_FAST_CLZ */ + +#if defined(__GNUC__) + +/* Our generic version of av_popcount is faster than GCC's built-in on + * CPUs that don't support the popcnt instruction. + */ +#if defined(__POPCNT__) + #define av_popcount __builtin_popcount +#if ARCH_X86_64 + #define av_popcount64 __builtin_popcountll +#endif + +#endif /* __POPCNT__ */ + +#if defined(__BMI2__) + +#if AV_GCC_VERSION_AT_LEAST(5,1) +#define av_mod_uintp2 __builtin_ia32_bzhi_si +#elif HAVE_INLINE_ASM +/* GCC releases before 5.1.0 have a broken bzhi builtin, so for those we + * implement it using inline assembly + */ +#define av_mod_uintp2 av_mod_uintp2_bmi2 +static av_always_inline av_const unsigned av_mod_uintp2_bmi2(unsigned a, unsigned p) +{ + if (av_builtin_constant_p(p)) + return a & ((1 << p) - 1); + else { + unsigned x; + __asm__ ("bzhi %2, %1, %0 \n\t" : "=r"(x) : "rm"(a), "r"(p)); + return x; + } +} +#endif /* AV_GCC_VERSION_AT_LEAST */ + +#endif /* __BMI2__ */ + +#if defined(__SSE2__) && !defined(__INTEL_COMPILER) + +#define av_clipd av_clipd_sse2 +static av_always_inline av_const double av_clipd_sse2(double a, double amin, double amax) +{ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + __asm__ ("maxsd %1, %0 \n\t" + "minsd %2, %0 \n\t" + : "+&x"(a) : "xm"(amin), "xm"(amax)); + return a; +} + +#endif /* __SSE2__ */ + +#if defined(__SSE__) && !defined(__INTEL_COMPILER) + +#define av_clipf av_clipf_sse +static av_always_inline av_const float av_clipf_sse(float a, float amin, float amax) +{ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + __asm__ ("maxss %1, %0 \n\t" + "minss %2, %0 \n\t" + : "+&x"(a) : "xm"(amin), "xm"(amax)); + return a; +} + +#endif /* __SSE__ */ + +#if defined(__AVX__) && !defined(__INTEL_COMPILER) + +#undef av_clipd +#define av_clipd av_clipd_avx +static av_always_inline av_const double av_clipd_avx(double a, double amin, double amax) +{ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + __asm__ ("vmaxsd %1, %0, %0 \n\t" + "vminsd %2, %0, %0 \n\t" + : "+&x"(a) : "xm"(amin), "xm"(amax)); + return a; +} + +#undef av_clipf +#define av_clipf av_clipf_avx +static av_always_inline av_const float av_clipf_avx(float a, float amin, float amax) +{ +#if defined(ASSERT_LEVEL) && ASSERT_LEVEL >= 2 + if (amin > amax) abort(); +#endif + __asm__ ("vmaxss %1, %0, %0 \n\t" + "vminss %2, %0, %0 \n\t" + : "+&x"(a) : "xm"(amin), "xm"(amax)); + return a; +} + +#endif /* __AVX__ */ + +#endif /* __GNUC__ */ + +#endif /* AVUTIL_X86_INTMATH_H */ diff --git a/media/ffvpx/libavutil/x86/intreadwrite.h b/media/ffvpx/libavutil/x86/intreadwrite.h new file mode 100644 index 0000000000..40f375b013 --- /dev/null +++ b/media/ffvpx/libavutil/x86/intreadwrite.h @@ -0,0 +1,99 @@ +/* + * Copyright (c) 2010 Alexander Strange <astrange@ithinksw.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_INTREADWRITE_H +#define AVUTIL_X86_INTREADWRITE_H + +#include <stdint.h> +#include "config.h" +#include "libavutil/attributes.h" + +#if HAVE_MMX + +#if !HAVE_FAST_64BIT && defined(__MMX__) + +#define FF_COPY_SWAP_ZERO_USES_MMX + +#define AV_COPY64 AV_COPY64 +static av_always_inline void AV_COPY64(void *d, const void *s) +{ + __asm__("movq %1, %%mm0 \n\t" + "movq %%mm0, %0 \n\t" + : "=m"(*(uint64_t*)d) + : "m" (*(const uint64_t*)s) + : "mm0"); +} + +#define AV_SWAP64 AV_SWAP64 +static av_always_inline void AV_SWAP64(void *a, void *b) +{ + __asm__("movq %1, %%mm0 \n\t" + "movq %0, %%mm1 \n\t" + "movq %%mm0, %0 \n\t" + "movq %%mm1, %1 \n\t" + : "+m"(*(uint64_t*)a), "+m"(*(uint64_t*)b) + ::"mm0", "mm1"); +} + +#define AV_ZERO64 AV_ZERO64 +static av_always_inline void AV_ZERO64(void *d) +{ + __asm__("pxor %%mm0, %%mm0 \n\t" + "movq %%mm0, %0 \n\t" + : "=m"(*(uint64_t*)d) + :: "mm0"); +} + +#endif /* !HAVE_FAST_64BIT && defined(__MMX__) */ + +#ifdef __SSE__ + +#define AV_COPY128 AV_COPY128 +static av_always_inline void AV_COPY128(void *d, const void *s) +{ + struct v {uint64_t v[2];}; + + __asm__("movaps %1, %%xmm0 \n\t" + "movaps %%xmm0, %0 \n\t" + : "=m"(*(struct v*)d) + : "m" (*(const struct v*)s) + : "xmm0"); +} + +#endif /* __SSE__ */ + +#ifdef __SSE2__ + +#define AV_ZERO128 AV_ZERO128 +static av_always_inline void AV_ZERO128(void *d) +{ + struct v {uint64_t v[2];}; + + __asm__("pxor %%xmm0, %%xmm0 \n\t" + "movdqa %%xmm0, %0 \n\t" + : "=m"(*(struct v*)d) + :: "xmm0"); +} + +#endif /* __SSE2__ */ + +#endif /* HAVE_MMX */ + +#endif /* AVUTIL_X86_INTREADWRITE_H */ diff --git a/media/ffvpx/libavutil/x86/lls.asm b/media/ffvpx/libavutil/x86/lls.asm new file mode 100644 index 0000000000..e8141e6c4f --- /dev/null +++ b/media/ffvpx/libavutil/x86/lls.asm @@ -0,0 +1,290 @@ +;****************************************************************************** +;* linear least squares model +;* +;* Copyright (c) 2013 Loren Merritt +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +%define MAX_VARS 32 +%define MAX_VARS_ALIGN (MAX_VARS+4) +%define COVAR_STRIDE MAX_VARS_ALIGN*8 +%define COVAR(x,y) [covarq + (x)*8 + (y)*COVAR_STRIDE] + +struc LLSModel + .covariance: resq MAX_VARS_ALIGN*MAX_VARS_ALIGN + .coeff: resq MAX_VARS*MAX_VARS + .variance: resq MAX_VARS + .indep_count: resd 1 +endstruc + +%macro ADDPD_MEM 2 +%if cpuflag(avx) + vaddpd %2, %2, %1 +%else + addpd %2, %1 +%endif + mova %1, %2 +%endmacro + +INIT_XMM sse2 +%define movdqa movaps +cglobal update_lls, 2,5,8, ctx, var, i, j, covar2 + %define covarq ctxq + mov id, [ctxq + LLSModel.indep_count] + lea varq, [varq + iq*8] + neg iq + mov covar2q, covarq +.loopi: + ; Compute all 3 pairwise products of a 2x2 block that lies on the diagonal + mova m1, [varq + iq*8] + mova m3, [varq + iq*8 + 16] + pshufd m4, m1, q1010 + pshufd m5, m1, q3232 + pshufd m6, m3, q1010 + pshufd m7, m3, q3232 + mulpd m0, m1, m4 + mulpd m1, m1, m5 + lea covarq, [covar2q + 16] + ADDPD_MEM COVAR(-2,0), m0 + ADDPD_MEM COVAR(-2,1), m1 + lea jq, [iq + 2] + cmp jd, -2 + jg .skip4x4 +.loop4x4: + ; Compute all 16 pairwise products of a 4x4 block + mulpd m0, m4, m3 + mulpd m1, m5, m3 + mulpd m2, m6, m3 + mulpd m3, m3, m7 + ADDPD_MEM COVAR(0,0), m0 + ADDPD_MEM COVAR(0,1), m1 + ADDPD_MEM COVAR(0,2), m2 + ADDPD_MEM COVAR(0,3), m3 + mova m3, [varq + jq*8 + 16] + mulpd m0, m4, m3 + mulpd m1, m5, m3 + mulpd m2, m6, m3 + mulpd m3, m3, m7 + ADDPD_MEM COVAR(2,0), m0 + ADDPD_MEM COVAR(2,1), m1 + ADDPD_MEM COVAR(2,2), m2 + ADDPD_MEM COVAR(2,3), m3 + mova m3, [varq + jq*8 + 32] + add covarq, 32 + add jq, 4 + cmp jd, -2 + jle .loop4x4 +.skip4x4: + test jd, jd + jg .skip2x4 + mulpd m4, m3 + mulpd m5, m3 + mulpd m6, m3 + mulpd m7, m3 + ADDPD_MEM COVAR(0,0), m4 + ADDPD_MEM COVAR(0,1), m5 + ADDPD_MEM COVAR(0,2), m6 + ADDPD_MEM COVAR(0,3), m7 +.skip2x4: + add iq, 4 + add covar2q, 4*COVAR_STRIDE+32 + cmp id, -2 + jle .loopi + test id, id + jg .ret + mov jq, iq + %define covarq covar2q +.loop2x1: + movsd m0, [varq + iq*8] + movlhps m0, m0 + mulpd m0, [varq + jq*8] + ADDPD_MEM COVAR(0,0), m0 + inc iq + add covarq, COVAR_STRIDE + test id, id + jle .loop2x1 +.ret: + RET + +%macro UPDATE_LLS 0 +cglobal update_lls, 3,6,8, ctx, var, count, i, j, count2 + %define covarq ctxq + mov countd, [ctxq + LLSModel.indep_count] + lea count2d, [countq-2] + xor id, id +.loopi: + ; Compute all 10 pairwise products of a 4x4 block that lies on the diagonal + mova ymm1, [varq + iq*8] + vbroadcastsd ymm4, [varq + iq*8] + vbroadcastsd ymm5, [varq + iq*8 + 8] + vbroadcastsd ymm6, [varq + iq*8 + 16] + vbroadcastsd ymm7, [varq + iq*8 + 24] + vextractf128 xmm3, ymm1, 1 +%if cpuflag(fma3) + mova ymm0, COVAR(iq ,0) + mova xmm2, COVAR(iq+2,2) + fmaddpd ymm0, ymm1, ymm4, ymm0 + fmaddpd xmm2, xmm3, xmm6, xmm2 + fmaddpd ymm1, ymm5, ymm1, COVAR(iq ,1) + fmaddpd xmm3, xmm7, xmm3, COVAR(iq+2,3) + mova COVAR(iq ,0), ymm0 + mova COVAR(iq ,1), ymm1 + mova COVAR(iq+2,2), xmm2 + mova COVAR(iq+2,3), xmm3 +%else + vmulpd ymm0, ymm1, ymm4 + vmulpd ymm1, ymm1, ymm5 + vmulpd xmm2, xmm3, xmm6 + vmulpd xmm3, xmm3, xmm7 + ADDPD_MEM COVAR(iq ,0), ymm0 + ADDPD_MEM COVAR(iq ,1), ymm1 + ADDPD_MEM COVAR(iq+2,2), xmm2 + ADDPD_MEM COVAR(iq+2,3), xmm3 +%endif ; cpuflag(fma3) + lea jd, [iq + 4] + cmp jd, count2d + jg .skip4x4 +.loop4x4: + ; Compute all 16 pairwise products of a 4x4 block + mova ymm3, [varq + jq*8] +%if cpuflag(fma3) + mova ymm0, COVAR(jq, 0) + mova ymm1, COVAR(jq, 1) + mova ymm2, COVAR(jq, 2) + fmaddpd ymm0, ymm3, ymm4, ymm0 + fmaddpd ymm1, ymm3, ymm5, ymm1 + fmaddpd ymm2, ymm3, ymm6, ymm2 + fmaddpd ymm3, ymm7, ymm3, COVAR(jq,3) + mova COVAR(jq, 0), ymm0 + mova COVAR(jq, 1), ymm1 + mova COVAR(jq, 2), ymm2 + mova COVAR(jq, 3), ymm3 +%else + vmulpd ymm0, ymm3, ymm4 + vmulpd ymm1, ymm3, ymm5 + vmulpd ymm2, ymm3, ymm6 + vmulpd ymm3, ymm3, ymm7 + ADDPD_MEM COVAR(jq,0), ymm0 + ADDPD_MEM COVAR(jq,1), ymm1 + ADDPD_MEM COVAR(jq,2), ymm2 + ADDPD_MEM COVAR(jq,3), ymm3 +%endif ; cpuflag(fma3) + add jd, 4 + cmp jd, count2d + jle .loop4x4 +.skip4x4: + cmp jd, countd + jg .skip2x4 + mova xmm3, [varq + jq*8] +%if cpuflag(fma3) + mova xmm0, COVAR(jq, 0) + mova xmm1, COVAR(jq, 1) + mova xmm2, COVAR(jq, 2) + fmaddpd xmm0, xmm3, xmm4, xmm0 + fmaddpd xmm1, xmm3, xmm5, xmm1 + fmaddpd xmm2, xmm3, xmm6, xmm2 + fmaddpd xmm3, xmm7, xmm3, COVAR(jq,3) + mova COVAR(jq, 0), xmm0 + mova COVAR(jq, 1), xmm1 + mova COVAR(jq, 2), xmm2 + mova COVAR(jq, 3), xmm3 +%else + vmulpd xmm0, xmm3, xmm4 + vmulpd xmm1, xmm3, xmm5 + vmulpd xmm2, xmm3, xmm6 + vmulpd xmm3, xmm3, xmm7 + ADDPD_MEM COVAR(jq,0), xmm0 + ADDPD_MEM COVAR(jq,1), xmm1 + ADDPD_MEM COVAR(jq,2), xmm2 + ADDPD_MEM COVAR(jq,3), xmm3 +%endif ; cpuflag(fma3) +.skip2x4: + add id, 4 + add covarq, 4*COVAR_STRIDE + cmp id, count2d + jle .loopi + cmp id, countd + jg .ret + mov jd, id +.loop2x1: + vmovddup xmm0, [varq + iq*8] +%if cpuflag(fma3) + mova xmm1, [varq + jq*8] + fmaddpd xmm0, xmm1, xmm0, COVAR(jq,0) + mova COVAR(jq,0), xmm0 +%else + vmulpd xmm0, [varq + jq*8] + ADDPD_MEM COVAR(jq,0), xmm0 +%endif ; cpuflag(fma3) + inc id + add covarq, COVAR_STRIDE + cmp id, countd + jle .loop2x1 +.ret: + RET +%endmacro ; UPDATE_LLS + +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +UPDATE_LLS +%endif +%if HAVE_FMA3_EXTERNAL +INIT_YMM fma3 +UPDATE_LLS +%endif + +INIT_XMM sse2 +cglobal evaluate_lls, 3,4,2, ctx, var, order, i + ; This function is often called on the same buffer as update_lls, but with + ; an offset. They can't both be aligned. + ; Load halves rather than movu to avoid store-forwarding stalls, since the + ; input was initialized immediately prior to this function using scalar math. + %define coefsq ctxq + mov id, orderd + imul orderd, MAX_VARS + lea coefsq, [ctxq + LLSModel.coeff + orderq*8] + movsd m0, [varq] + movhpd m0, [varq + 8] + mulpd m0, [coefsq] + lea coefsq, [coefsq + iq*8] + lea varq, [varq + iq*8] + neg iq + add iq, 2 +.loop: + movsd m1, [varq + iq*8] + movhpd m1, [varq + iq*8 + 8] + mulpd m1, [coefsq + iq*8] + addpd m0, m1 + add iq, 2 + jl .loop + jg .skip1 + movsd m1, [varq + iq*8] + mulsd m1, [coefsq + iq*8] + addpd m0, m1 +.skip1: + movhlps m1, m0 + addsd m0, m1 +%if ARCH_X86_32 + movsd r0m, m0 + fld qword r0m +%endif + RET diff --git a/media/ffvpx/libavutil/x86/lls_init.c b/media/ffvpx/libavutil/x86/lls_init.c new file mode 100644 index 0000000000..c786376915 --- /dev/null +++ b/media/ffvpx/libavutil/x86/lls_init.c @@ -0,0 +1,46 @@ +/* + * linear least squares model + * + * Copyright (c) 2013 Loren Merritt + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/attributes.h" +#include "libavutil/lls.h" +#include "libavutil/x86/cpu.h" + +void ff_update_lls_sse2(LLSModel *m, const double *var); +void ff_update_lls_avx(LLSModel *m, const double *var); +void ff_update_lls_fma3(LLSModel *m, const double *var); +double ff_evaluate_lls_sse2(LLSModel *m, const double *var, int order); + +av_cold void ff_init_lls_x86(LLSModel *m) +{ + int cpu_flags = av_get_cpu_flags(); + if (EXTERNAL_SSE2(cpu_flags)) { + m->update_lls = ff_update_lls_sse2; + if (m->indep_count >= 4) + m->evaluate_lls = ff_evaluate_lls_sse2; + } + if (EXTERNAL_AVX_FAST(cpu_flags)) { + m->update_lls = ff_update_lls_avx; + } + if (EXTERNAL_FMA3_FAST(cpu_flags)) { + m->update_lls = ff_update_lls_fma3; + } +} diff --git a/media/ffvpx/libavutil/x86/moz.build b/media/ffvpx/libavutil/x86/moz.build new file mode 100644 index 0000000000..df33768f66 --- /dev/null +++ b/media/ffvpx/libavutil/x86/moz.build @@ -0,0 +1,25 @@ +# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*- +# vim: set filetype=python: +# This Source Code Form is subject to the terms of the Mozilla Public +# License, v. 2.0. If a copy of the MPL was not distributed with this +# file, You can obtain one at http://mozilla.org/MPL/2.0/. + +SOURCES += [ + 'cpu.c', + 'cpuid.asm', + 'emms.asm', + 'fixed_dsp.asm', + 'fixed_dsp_init.c', + 'float_dsp.asm', + 'float_dsp_init.c', + 'imgutils.asm', + 'imgutils_init.c', + 'lls.asm', + 'lls_init.c', + 'tx_float.asm', + 'tx_float_init.c', +] + +FINAL_LIBRARY = 'mozavutil' + +include('/media/ffvpx/ffvpxcommon.mozbuild') diff --git a/media/ffvpx/libavutil/x86/pixelutils.asm b/media/ffvpx/libavutil/x86/pixelutils.asm new file mode 100644 index 0000000000..0bcccb51f5 --- /dev/null +++ b/media/ffvpx/libavutil/x86/pixelutils.asm @@ -0,0 +1,329 @@ +;****************************************************************************** +;* Pixel utilities SIMD +;* +;* Copyright (C) 2002-2004 Michael Niedermayer <michaelni@gmx.at> +;* Copyright (C) 2014 Clément Bœsch <u pkh me> +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%include "libavutil/x86/x86util.asm" + +SECTION .text + +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_8x8_mmxext(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +INIT_MMX mmxext +cglobal pixelutils_sad_8x8, 4,4,0, src1, stride1, src2, stride2 + pxor m2, m2 +%rep 4 + mova m0, [src1q] + mova m1, [src1q + stride1q] + psadbw m0, [src2q] + psadbw m1, [src2q + stride2q] + paddw m2, m0 + paddw m2, m1 + lea src1q, [src1q + 2*stride1q] + lea src2q, [src2q + 2*stride2q] +%endrep + movd eax, m2 + emms + RET + +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_16x16_sse2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +INIT_XMM sse2 +cglobal pixelutils_sad_16x16, 4,4,5, src1, stride1, src2, stride2 + movu m4, [src1q] + movu m2, [src2q] + movu m1, [src1q + stride1q] + movu m3, [src2q + stride2q] + psadbw m4, m2 + psadbw m1, m3 + paddw m4, m1 +%rep 7 + lea src1q, [src1q + 2*stride1q] + lea src2q, [src2q + 2*stride2q] + movu m0, [src1q] + movu m2, [src2q] + movu m1, [src1q + stride1q] + movu m3, [src2q + stride2q] + psadbw m0, m2 + psadbw m1, m3 + paddw m4, m0 + paddw m4, m1 +%endrep + movhlps m0, m4 + paddw m4, m0 + movd eax, m4 + RET + +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_[au]_16x16_sse2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +%macro SAD_XMM_16x16 1 +INIT_XMM sse2 +cglobal pixelutils_sad_%1_16x16, 4,4,3, src1, stride1, src2, stride2 + mov%1 m2, [src2q] + psadbw m2, [src1q] + mov%1 m1, [src2q + stride2q] + psadbw m1, [src1q + stride1q] + paddw m2, m1 +%rep 7 + lea src1q, [src1q + 2*stride1q] + lea src2q, [src2q + 2*stride2q] + mov%1 m0, [src2q] + psadbw m0, [src1q] + mov%1 m1, [src2q + stride2q] + psadbw m1, [src1q + stride1q] + paddw m2, m0 + paddw m2, m1 +%endrep + movhlps m0, m2 + paddw m2, m0 + movd eax, m2 + RET +%endmacro + +SAD_XMM_16x16 a +SAD_XMM_16x16 u + + +%macro PROCESS_SAD_32x4_U 0 + movu m1, [r2] + movu m2, [r2 + 16] + movu m3, [r0] + movu m4, [r0 + 16] + psadbw m1, m3 + psadbw m2, m4 + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + movu m1, [r2] + movu m2, [r2 + 16] + movu m3, [r0] + movu m4, [r0 + 16] + psadbw m1, m3 + psadbw m2, m4 + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + movu m1, [r2] + movu m2, [r2 + 16] + movu m3, [r0] + movu m4, [r0 + 16] + psadbw m1, m3 + psadbw m2, m4 + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + movu m1, [r2] + movu m2, [r2 + 16] + movu m3, [r0] + movu m4, [r0 + 16] + psadbw m1, m3 + psadbw m2, m4 + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] +%endmacro + +%macro PROCESS_SAD_32x4 1 + mov%1 m1, [r2] + mov%1 m2, [r2 + 16] + psadbw m1, [r0] + psadbw m2, [r0 + 16] + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + mov%1 m1, [r2] + mov%1 m2, [r2 + 16] + psadbw m1, [r0] + psadbw m2, [r0 + 16] + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + mov%1 m1, [r2] + mov%1 m2, [r2 + 16] + psadbw m1, [r0] + psadbw m2, [r0 + 16] + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] + + mov%1 m1, [r2] + mov%1 m2, [r2 + 16] + psadbw m1, [r0] + psadbw m2, [r0 + 16] + paddd m1, m2 + paddd m0, m1 + lea r2, [r2 + r3] + lea r0, [r0 + r1] +%endmacro + +;----------------------------------------------------------------------------- +; int ff_pixelutils_sad_32x32_sse2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;----------------------------------------------------------------------------- +INIT_XMM sse2 +cglobal pixelutils_sad_32x32, 4,5,5, src1, stride1, src2, stride2 + pxor m0, m0 + mov r4d, 4 +.loop: + PROCESS_SAD_32x4_U + PROCESS_SAD_32x4_U + dec r4d + jnz .loop + + movhlps m1, m0 + paddd m0, m1 + movd eax, m0 + RET + +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_[au]_32x32_sse2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +%macro SAD_XMM_32x32 1 +INIT_XMM sse2 +cglobal pixelutils_sad_%1_32x32, 4,5,3, src1, stride1, src2, stride2 + pxor m0, m0 + mov r4d, 4 +.loop: + PROCESS_SAD_32x4 %1 + PROCESS_SAD_32x4 %1 + dec r4d + jnz .loop + + movhlps m1, m0 + paddd m0, m1 + movd eax, m0 + RET +%endmacro + +SAD_XMM_32x32 a +SAD_XMM_32x32 u + +%if HAVE_AVX2_EXTERNAL +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_32x32_avx2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +INIT_YMM avx2 +cglobal pixelutils_sad_32x32, 4,7,5, src1, stride1, src2, stride2 + pxor m0, m0 + mov r4d, 32/4 + lea r5, [stride1q * 3] + lea r6, [stride2q * 3] + +.loop: + movu m1, [src1q] ; row 0 of pix0 + movu m2, [src2q] ; row 0 of pix1 + movu m3, [src1q + stride1q] ; row 1 of pix0 + movu m4, [src2q + stride2q] ; row 1 of pix1 + + psadbw m1, m2 + psadbw m3, m4 + paddd m0, m1 + paddd m0, m3 + + movu m1, [src1q + 2 * stride1q] ; row 2 of pix0 + movu m2, [src2q + 2 * stride2q] ; row 2 of pix1 + movu m3, [src1q + r5] ; row 3 of pix0 + movu m4, [src2q + r6] ; row 3 of pix1 + + psadbw m1, m2 + psadbw m3, m4 + paddd m0, m1 + paddd m0, m3 + + lea src2q, [src2q + 4 * stride2q] + lea src1q, [src1q + 4 * stride1q] + + dec r4d + jnz .loop + + vextracti128 xm1, m0, 1 + paddd xm0, xm1 + pshufd xm1, xm0, 2 + paddd xm0, xm1 + movd eax, xm0 + RET + +;------------------------------------------------------------------------------- +; int ff_pixelutils_sad_[au]_32x32_avx2(const uint8_t *src1, ptrdiff_t stride1, +; const uint8_t *src2, ptrdiff_t stride2); +;------------------------------------------------------------------------------- +%macro SAD_AVX2_32x32 1 +INIT_YMM avx2 +cglobal pixelutils_sad_%1_32x32, 4,7,3, src1, stride1, src2, stride2 + pxor m0, m0 + mov r4d, 32/4 + lea r5, [stride1q * 3] + lea r6, [stride2q * 3] + +.loop: + mov%1 m1, [src2q] ; row 0 of pix1 + psadbw m1, [src1q] + mov%1 m2, [src2q + stride2q] ; row 1 of pix1 + psadbw m2, [src1q + stride1q] + + paddd m0, m1 + paddd m0, m2 + + mov%1 m1, [src2q + 2 * stride2q] ; row 2 of pix1 + psadbw m1, [src1q + 2 * stride1q] + mov%1 m2, [src2q + r6] ; row 3 of pix1 + psadbw m2, [src1q + r5] + + paddd m0, m1 + paddd m0, m2 + + lea src2q, [src2q + 4 * stride2q] + lea src1q, [src1q + 4 * stride1q] + + dec r4d + jnz .loop + + vextracti128 xm1, m0, 1 + paddd xm0, xm1 + pshufd xm1, xm0, 2 + paddd xm0, xm1 + movd eax, xm0 + RET +%endmacro + +SAD_AVX2_32x32 a +SAD_AVX2_32x32 u +%endif diff --git a/media/ffvpx/libavutil/x86/pixelutils.h b/media/ffvpx/libavutil/x86/pixelutils.h new file mode 100644 index 0000000000..876cf46053 --- /dev/null +++ b/media/ffvpx/libavutil/x86/pixelutils.h @@ -0,0 +1,26 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_PIXELUTILS_H +#define AVUTIL_X86_PIXELUTILS_H + +#include "libavutil/pixelutils.h" + +void ff_pixelutils_sad_init_x86(av_pixelutils_sad_fn *sad, int aligned); + +#endif /* AVUTIL_X86_PIXELUTILS_H */ diff --git a/media/ffvpx/libavutil/x86/pixelutils_init.c b/media/ffvpx/libavutil/x86/pixelutils_init.c new file mode 100644 index 0000000000..c3c0662414 --- /dev/null +++ b/media/ffvpx/libavutil/x86/pixelutils_init.c @@ -0,0 +1,85 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "config.h" + +#include "pixelutils.h" +#include "cpu.h" + +int ff_pixelutils_sad_8x8_mmxext(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); + +int ff_pixelutils_sad_16x16_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_a_16x16_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_u_16x16_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); + +int ff_pixelutils_sad_32x32_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_a_32x32_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_u_32x32_sse2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); + +int ff_pixelutils_sad_32x32_avx2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_a_32x32_avx2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); +int ff_pixelutils_sad_u_32x32_avx2(const uint8_t *src1, ptrdiff_t stride1, + const uint8_t *src2, ptrdiff_t stride2); + +void ff_pixelutils_sad_init_x86(av_pixelutils_sad_fn *sad, int aligned) +{ + int cpu_flags = av_get_cpu_flags(); + + // The best way to use SSE2 would be to do 2 SADs in parallel, + // but we'd have to modify the pixelutils API to return SIMD functions. + + // It's probably not faster to shuffle data around + // to get two lines of 8 pixels into a single 16byte register, + // so just use the MMX 8x8 version even when SSE2 is available. + if (EXTERNAL_MMXEXT(cpu_flags)) { + sad[2] = ff_pixelutils_sad_8x8_mmxext; + } + + if (EXTERNAL_SSE2(cpu_flags)) { + switch (aligned) { + case 0: sad[3] = ff_pixelutils_sad_16x16_sse2; break; // src1 unaligned, src2 unaligned + case 1: sad[3] = ff_pixelutils_sad_u_16x16_sse2; break; // src1 aligned, src2 unaligned + case 2: sad[3] = ff_pixelutils_sad_a_16x16_sse2; break; // src1 aligned, src2 aligned + } + } + + if (EXTERNAL_SSE2(cpu_flags)) { + switch (aligned) { + case 0: sad[4] = ff_pixelutils_sad_32x32_sse2; break; // src1 unaligned, src2 unaligned + case 1: sad[4] = ff_pixelutils_sad_u_32x32_sse2; break; // src1 aligned, src2 unaligned + case 2: sad[4] = ff_pixelutils_sad_a_32x32_sse2; break; // src1 aligned, src2 aligned + } + } + + if (EXTERNAL_AVX2_FAST(cpu_flags)) { + switch (aligned) { + case 0: sad[4] = ff_pixelutils_sad_32x32_avx2; break; // src1 unaligned, src2 unaligned + case 1: sad[4] = ff_pixelutils_sad_u_32x32_avx2; break; // src1 aligned, src2 unaligned + case 2: sad[4] = ff_pixelutils_sad_a_32x32_avx2; break; // src1 aligned, src2 aligned + } + } +} diff --git a/media/ffvpx/libavutil/x86/timer.h b/media/ffvpx/libavutil/x86/timer.h new file mode 100644 index 0000000000..4d1e88def0 --- /dev/null +++ b/media/ffvpx/libavutil/x86/timer.h @@ -0,0 +1,50 @@ +/* + * copyright (c) 2006 Michael Niedermayer <michaelni@gmx.at> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#ifndef AVUTIL_X86_TIMER_H +#define AVUTIL_X86_TIMER_H + +#include <stdint.h> + +#if HAVE_INLINE_ASM + +#define FF_TIMER_UNITS "decicycles" +#define AV_READ_TIME read_time + +static inline uint64_t read_time(void) +{ + uint32_t a, d; + __asm__ volatile( +#if ARCH_X86_64 || defined(__SSE2__) + "lfence \n\t" +#endif + "rdtsc \n\t" + : "=a" (a), "=d" (d)); + return ((uint64_t)d << 32) + a; +} + +#elif HAVE_RDTSC + +#include <intrin.h> +#define AV_READ_TIME __rdtsc + +#endif /* HAVE_INLINE_ASM */ + +#endif /* AVUTIL_X86_TIMER_H */ diff --git a/media/ffvpx/libavutil/x86/tx_float.asm b/media/ffvpx/libavutil/x86/tx_float.asm new file mode 100644 index 0000000000..e1533a8595 --- /dev/null +++ b/media/ffvpx/libavutil/x86/tx_float.asm @@ -0,0 +1,1936 @@ +;****************************************************************************** +;* Copyright (c) Lynne +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +; Open `doc/transforms.md` to see the code upon which the transforms here were +; based upon and compare. + +; Intra-asm call convention: +; 320 bytes of stack available +; 14 GPRs available (last 4 must not be clobbered) +; Additionally, don't clobber ctx, in, out, stride, len, lut +; All vector regs available + +; TODO: +; carry over registers from smaller transforms to save on ~8 loads/stores +; check if vinsertf could be faster than verpm2f128 for duplication +; even faster FFT8 (current one is very #instructions optimized) +; replace some xors with blends + addsubs? +; replace some shuffles with vblends? +; avx512 split-radix + +%include "libavutil/x86/x86util.asm" + +%define private_prefix ff_tx + +%if ARCH_X86_64 +%define ptr resq +%else +%define ptr resd +%endif + +%assign i 16 +%rep 14 +cextern tab_ %+ i %+ _float ; ff_tab_i_float... +%assign i (i << 1) +%endrep + +cextern tab_53_float + +struc AVTXContext + .len: resd 1 ; Length + .inv resd 1 ; Inverse flag + .map: ptr 1 ; Lookup table(s) + .exp: ptr 1 ; Exponentiation factors + .tmp: ptr 1 ; Temporary data + + .sub: ptr 1 ; Subcontexts + .fn: ptr 4 ; Subcontext functions + .nb_sub: resd 1 ; Subcontext count + + ; Everything else is inaccessible +endstruc + +SECTION_RODATA 32 + +%define POS 0x00000000 +%define NEG 0x80000000 + +%define M_SQRT1_2 0.707106781186547524401 +%define COS16_1 0.92387950420379638671875 +%define COS16_3 0.3826834261417388916015625 + +d8_mult_odd: dd M_SQRT1_2, -M_SQRT1_2, -M_SQRT1_2, M_SQRT1_2, \ + M_SQRT1_2, -M_SQRT1_2, -M_SQRT1_2, M_SQRT1_2 + +s8_mult_odd: dd 1.0, 1.0, -1.0, 1.0, -M_SQRT1_2, -M_SQRT1_2, M_SQRT1_2, M_SQRT1_2 +s8_perm_even: dd 1, 3, 0, 2, 1, 3, 2, 0 +s8_perm_odd1: dd 3, 3, 1, 1, 1, 1, 3, 3 +s8_perm_odd2: dd 1, 2, 0, 3, 1, 0, 0, 1 + +s16_mult_even: dd 1.0, 1.0, M_SQRT1_2, M_SQRT1_2, 1.0, -1.0, M_SQRT1_2, -M_SQRT1_2 +s16_mult_odd1: dd COS16_1, COS16_1, COS16_3, COS16_3, COS16_1, -COS16_1, COS16_3, -COS16_3 +s16_mult_odd2: dd COS16_3, -COS16_3, COS16_1, -COS16_1, -COS16_3, -COS16_3, -COS16_1, -COS16_1 +s16_perm: dd 0, 1, 2, 3, 1, 0, 3, 2 + +s15_perm: dd 0, 6, 5, 3, 2, 4, 7, 1 + +mask_mmppmmmm: dd NEG, NEG, POS, POS, NEG, NEG, NEG, NEG +mask_mmmmpppm: dd NEG, NEG, NEG, NEG, POS, POS, POS, NEG +mask_ppmpmmpm: dd POS, POS, NEG, POS, NEG, NEG, POS, NEG +mask_mppmmpmp: dd NEG, POS, POS, NEG, NEG, POS, NEG, POS +mask_mpmppmpm: dd NEG, POS, NEG, POS, POS, NEG, POS, NEG +mask_pmmppmmp: dd POS, NEG, NEG, POS, POS, NEG, NEG, POS +mask_pmpmpmpm: times 4 dd POS, NEG + +SECTION .text + +; Load complex values (64 bits) via a lookup table +; %1 - output register +; %2 - GRP of base input memory address +; %3 - GPR of LUT (int32_t indices) address +; %4 - LUT offset +; %5 - temporary GPR (only used if vgather is not used) +; %6 - temporary register (for avx only) +; %7 - temporary register (for avx only, enables vgatherdpd (AVX2) if FMA3 is set) +%macro LOAD64_LUT 5-7 +%if %0 > 6 && cpuflag(avx2) + pcmpeqd %7, %7 ; pcmpeqq has a 0.5 throughput on Zen 3, this has 0.25 + movupd xmm%6, [%3 + %4] ; float mov since vgatherdpd is a float instruction + vgatherdpd %1, [%2 + xmm%6*8], %7 ; must use separate registers for args +%else + mov %5d, [%3 + %4 + 0] + movsd xmm%1, [%2 + %5q*8] +%if sizeof%1 > 16 && %0 > 5 + mov %5d, [%3 + %4 + 8] + movsd xmm%6, [%2 + %5q*8] +%endif + mov %5d, [%3 + %4 + 4] + movhps xmm%1, [%2 + %5q*8] +%if sizeof%1 > 16 && %0 > 5 + mov %5d, [%3 + %4 + 12] + movhps xmm%6, [%2 + %5q*8] + vinsertf128 %1, %1, xmm%6, 1 +%endif +%endif +%endmacro + +; Single 2-point in-place complex FFT (will do 2 transforms at once in AVX mode) +; %1 - coefficients (r0.reim, r1.reim) +; %2 - temporary +%macro FFT2 2 + shufps %2, %1, %1, q3322 + shufps %1, %1, %1, q1100 + + addsubps %1, %1, %2 + + shufps %1, %1, %1, q2031 +%endmacro + +; Single 4-point in-place complex FFT (will do 2 transforms at once in [AVX] mode) +; %1 - even coefficients (r0.reim, r2.reim, r4.reim, r6.reim) +; %2 - odd coefficients (r1.reim, r3.reim, r5.reim, r7.reim) +; %3 - temporary +%macro FFT4 3 + subps %3, %1, %2 ; r1234, [r5678] + addps %1, %1, %2 ; t1234, [t5678] + + shufps %2, %1, %3, q1010 ; t12, r12 + shufps %1, %1, %3, q2332 ; t34, r43 + + subps %3, %2, %1 ; a34, b32 + addps %2, %2, %1 ; a12, b14 + + shufps %1, %2, %3, q1010 ; a1234 even + + shufps %2, %2, %3, q2332 ; b1423 + shufps %2, %2, %2, q1320 ; b1234 odd +%endmacro + +; Single/Dual 8-point in-place complex FFT (will do 2 transforms in [AVX] mode) +; %1 - even coefficients (a0.reim, a2.reim, [b0.reim, b2.reim]) +; %2 - even coefficients (a4.reim, a6.reim, [b4.reim, b6.reim]) +; %3 - odd coefficients (a1.reim, a3.reim, [b1.reim, b3.reim]) +; %4 - odd coefficients (a5.reim, a7.reim, [b5.reim, b7.reim]) +; %5 - temporary +; %6 - temporary +%macro FFT8 6 + addps %5, %1, %3 ; q1-8 + addps %6, %2, %4 ; k1-8 + + subps %1, %1, %3 ; r1-8 + subps %2, %2, %4 ; j1-8 + + shufps %4, %1, %1, q2323 ; r4343 + shufps %3, %5, %6, q3032 ; q34, k14 + + shufps %1, %1, %1, q1010 ; r1212 + shufps %5, %5, %6, q1210 ; q12, k32 + + xorps %4, %4, [mask_pmmppmmp] ; r4343 * pmmp + addps %6, %5, %3 ; s12, g12 + + mulps %2, %2, [d8_mult_odd] ; r8 * d8_mult_odd + subps %5, %5, %3 ; s34, g43 + + addps %3, %1, %4 ; z1234 + unpcklpd %1, %6, %5 ; s1234 + + shufps %4, %2, %2, q2301 ; j2143 + shufps %6, %6, %5, q2332 ; g1234 + + addsubps %2, %2, %4 ; l2143 + shufps %5, %2, %2, q0123 ; l3412 + addsubps %5, %5, %2 ; t1234 + + subps %2, %1, %6 ; h1234 even + subps %4, %3, %5 ; u1234 odd + + addps %1, %1, %6 ; w1234 even + addps %3, %3, %5 ; o1234 odd +%endmacro + +; Single 8-point in-place complex FFT in 20 instructions +; %1 - even coefficients (r0.reim, r2.reim, r4.reim, r6.reim) +; %2 - odd coefficients (r1.reim, r3.reim, r5.reim, r7.reim) +; %3 - temporary +; %4 - temporary +%macro FFT8_AVX 4 + subps %3, %1, %2 ; r1234, r5678 + addps %1, %1, %2 ; q1234, q5678 + + vpermilps %2, %3, [s8_perm_odd1] ; r4422, r6688 + shufps %4, %1, %1, q3322 ; q1122, q5566 + + movsldup %3, %3 ; r1133, r5577 + shufps %1, %1, %1, q1100 ; q3344, q7788 + + addsubps %3, %3, %2 ; z1234, z5678 + addsubps %1, %1, %4 ; s3142, s7586 + + mulps %3, %3, [s8_mult_odd] ; z * s8_mult_odd + vpermilps %1, %1, [s8_perm_even] ; s1234, s5687 ! + + shufps %2, %3, %3, q2332 ; junk, z7887 + xorps %4, %1, [mask_mmmmpppm] ; e1234, e5687 ! + + vpermilps %3, %3, [s8_perm_odd2] ; z2314, z6556 + vperm2f128 %1, %1, %4, 0x03 ; e5687, s1234 + + addsubps %2, %2, %3 ; junk, t5678 + subps %1, %1, %4 ; w1234, w5678 even + + vperm2f128 %2, %2, %2, 0x11 ; t5678, t5678 + vperm2f128 %3, %3, %3, 0x00 ; z2314, z2314 + + xorps %2, %2, [mask_ppmpmmpm] ; t * ppmpmmpm + addps %2, %3, %2 ; u1234, u5678 odd +%endmacro + +; Single 16-point in-place complex FFT +; %1 - even coefficients (r0.reim, r2.reim, r4.reim, r6.reim) +; %2 - even coefficients (r8.reim, r10.reim, r12.reim, r14.reim) +; %3 - odd coefficients (r1.reim, r3.reim, r5.reim, r7.reim) +; %4 - odd coefficients (r9.reim, r11.reim, r13.reim, r15.reim) +; %5, %6 - temporary +; %7, %8 - temporary (optional) +%macro FFT16 6-8 + FFT4 %3, %4, %5 +%if %0 > 7 + FFT8_AVX %1, %2, %6, %7 + movaps %8, [mask_mpmppmpm] + movaps %7, [s16_perm] +%define mask %8 +%define perm %7 +%elif %0 > 6 + FFT8_AVX %1, %2, %6, %7 + movaps %7, [s16_perm] +%define mask [mask_mpmppmpm] +%define perm %7 +%else + FFT8_AVX %1, %2, %6, %5 +%define mask [mask_mpmppmpm] +%define perm [s16_perm] +%endif + xorps %5, %5, %5 ; 0 + + shufps %6, %4, %4, q2301 ; z12.imre, z13.imre... + shufps %5, %5, %3, q2301 ; 0, 0, z8.imre... + + mulps %4, %4, [s16_mult_odd1] ; z.reim * costab + xorps %5, %5, [mask_mppmmpmp] +%if cpuflag(fma3) + fmaddps %6, %6, [s16_mult_odd2], %4 ; s[8..15] + addps %5, %3, %5 ; s[0...7] +%else + mulps %6, %6, [s16_mult_odd2] ; z.imre * costab + + addps %5, %3, %5 ; s[0...7] + addps %6, %4, %6 ; s[8..15] +%endif + mulps %5, %5, [s16_mult_even] ; s[0...7]*costab + + xorps %4, %6, mask ; s[8..15]*mpmppmpm + xorps %3, %5, mask ; s[0...7]*mpmppmpm + + vperm2f128 %4, %4, %4, 0x01 ; s[12..15, 8..11] + vperm2f128 %3, %3, %3, 0x01 ; s[4..7, 0..3] + + addps %6, %6, %4 ; y56, u56, y34, u34 + addps %5, %5, %3 ; w56, x56, w34, x34 + + vpermilps %6, %6, perm ; y56, u56, y43, u43 + vpermilps %5, %5, perm ; w56, x56, w43, x43 + + subps %4, %2, %6 ; odd part 2 + addps %3, %2, %6 ; odd part 1 + + subps %2, %1, %5 ; even part 2 + addps %1, %1, %5 ; even part 1 +%undef mask +%undef perm +%endmacro + +; Single 15-point complex FFT +; Input: +; xm0 must contain in[0,1].reim +; m2 - in[3-6].reim +; m3 - in[7-11].reim +; m4 - in[12-15].reim +; xm5 must contain in[2].reimreim +; +; Output: +; m0, m1, m2 - ACs +; xm14 - out[0] +; xm15 - out[10, 5] +%macro FFT15 0 + shufps xm1, xm0, xm0, q3223 ; in[1].imrereim + shufps xm0, xm0, xm0, q1001 ; in[0].imrereim + + xorps xm1, xm11 + addps xm1, xm0 ; pc[0,1].imre + + shufps xm0, xm1, xm1, q3232 ; pc[1].reimreim + addps xm0, xm5 ; dc[0].reimreim + + mulps xm1, xm9 ; tab[0123]*pc[01] + + shufpd xm6, xm1, xm1, 01b ; pc[1,0].reim + xorps xm1, xm11 + addps xm1, xm1, xm6 + addsubps xm1, xm5, xm1 ; dc[1,2].reim + + subps m7, m2, m3 ; q[0-3].imre + addps m6, m2, m3 ; q[4-7] + shufps m7, m7, m7, q2301 ; q[0-3].reim + + addps m5, m4, m6 ; y[0-3] + + vperm2f128 m14, m9, m9, 0x11 ; tab[23232323] + vbroadcastsd m15, xm9 ; tab[01010101] + + mulps m6, m14 + mulps m7, m15 + + subps m2, m6, m7 ; k[0-3] + addps m3, m6, m7 ; k[4-7] + + shufps m12, m11, m11, q3232 ; ppppmmmm + + addsubps m6, m4, m2 ; k[0-3] + addsubps m7, m4, m3 ; k[4-7] + + ; 15pt from here on + vpermpd m2, m5, q0123 ; y[3-0] + vpermpd m3, m6, q0123 ; k[3-0] + vpermpd m4, m7, q0123 ; k[7-4] + + xorps m5, m12 + xorps m6, m12 + xorps m7, m12 + + addps m2, m5 ; t[0-3] + addps m3, m6 ; t[4-7] + addps m4, m7 ; t[8-11] + + movlhps xm14, xm2 ; out[0] + unpcklpd xm15, xm3, xm4 ; out[10,5] + unpckhpd xm5, xm3, xm4 ; out[10,5] + + addps xm14, xm2 ; out[0] + addps xm15, xm5 ; out[10,5] + addps xm14, xm0 ; out[0] + addps xm15, xm1 ; out[10,5] + + shufps m12, m10, m10, q3232 ; tab5 4 5 4 5 8 9 8 9 + shufps m13, m10, m10, q1010 ; tab5 6 7 6 7 10 11 10 11 + + mulps m5, m2, m12 ; t[0-3] + mulps m6, m3, m12 ; t[4-7] + mulps m7, m4, m12 ; t[8-11] + + mulps m2, m13 ; r[0-3] + mulps m3, m13 ; r[4-7] + mulps m4, m13 ; r[8-11] + + shufps m5, m5, m5, q1032 ; t[1,0,3,2].reim + shufps m6, m6, m6, q1032 ; t[5,4,7,6].reim + shufps m7, m7, m7, q1032 ; t[9,8,11,10].reim + + vperm2f128 m13, m11, m11, 0x01 ; mmmmmmpp + shufps m12, m11, m11, q3232 ; ppppmmmm + + xorps m5, m13 + xorps m6, m13 + xorps m7, m13 + + addps m2, m5 ; r[0,1,2,3] + addps m3, m6 ; r[4,5,6,7] + addps m4, m7 ; r[8,9,10,11] + + shufps m5, m2, m2, q2301 + shufps m6, m3, m3, q2301 + shufps m7, m4, m4, q2301 + + xorps m2, m12 + xorps m3, m12 + xorps m4, m12 + + vpermpd m5, m5, q0123 + vpermpd m6, m6, q0123 + vpermpd m7, m7, q0123 + + addps m5, m2 + addps m6, m3 + addps m7, m4 + + vpermps m5, m8, m5 + vpermps m6, m8, m6 + vpermps m7, m8, m7 + + vbroadcastsd m0, xm0 ; dc[0] + vpermpd m2, m1, q1111 ; dc[2] + vbroadcastsd m1, xm1 ; dc[1] + + addps m0, m5 + addps m1, m6 + addps m2, m7 +%endmacro + +; Cobmines m0...m8 (tx1[even, even, odd, odd], tx2,3[even], tx2,3[odd]) coeffs +; Uses all 16 of registers. +; Output is slightly permuted such that tx2,3's coefficients are interleaved +; on a 2-point basis (look at `doc/transforms.md`) +%macro SPLIT_RADIX_COMBINE 17 +%if %1 && mmsize == 32 + vperm2f128 %14, %6, %7, 0x20 ; m2[0], m2[1], m3[0], m3[1] even + vperm2f128 %16, %9, %8, 0x20 ; m2[0], m2[1], m3[0], m3[1] odd + vperm2f128 %15, %6, %7, 0x31 ; m2[2], m2[3], m3[2], m3[3] even + vperm2f128 %17, %9, %8, 0x31 ; m2[2], m2[3], m3[2], m3[3] odd +%endif + + shufps %12, %10, %10, q2200 ; cos00224466 + shufps %13, %11, %11, q1133 ; wim77553311 + movshdup %10, %10 ; cos11335577 + shufps %11, %11, %11, q0022 ; wim66442200 + +%if %1 && mmsize == 32 + shufps %6, %14, %14, q2301 ; m2[0].imre, m2[1].imre, m2[2].imre, m2[3].imre even + shufps %8, %16, %16, q2301 ; m2[0].imre, m2[1].imre, m2[2].imre, m2[3].imre odd + shufps %7, %15, %15, q2301 ; m3[0].imre, m3[1].imre, m3[2].imre, m3[3].imre even + shufps %9, %17, %17, q2301 ; m3[0].imre, m3[1].imre, m3[2].imre, m3[3].imre odd + + mulps %14, %14, %13 ; m2[0123]reim * wim7531 even + mulps %16, %16, %11 ; m2[0123]reim * wim7531 odd + mulps %15, %15, %13 ; m3[0123]reim * wim7531 even + mulps %17, %17, %11 ; m3[0123]reim * wim7531 odd +%else + mulps %14, %6, %13 ; m2,3[01]reim * wim7531 even + mulps %16, %8, %11 ; m2,3[01]reim * wim7531 odd + mulps %15, %7, %13 ; m2,3[23]reim * wim7531 even + mulps %17, %9, %11 ; m2,3[23]reim * wim7531 odd + ; reorder the multiplies to save movs reg, reg in the %if above + shufps %6, %6, %6, q2301 ; m2[0].imre, m2[1].imre, m3[0].imre, m3[1].imre even + shufps %8, %8, %8, q2301 ; m2[0].imre, m2[1].imre, m3[0].imre, m3[1].imre odd + shufps %7, %7, %7, q2301 ; m2[2].imre, m2[3].imre, m3[2].imre, m3[3].imre even + shufps %9, %9, %9, q2301 ; m2[2].imre, m2[3].imre, m3[2].imre, m3[3].imre odd +%endif + +%if cpuflag(fma3) ; 11 - 5 = 6 instructions saved through FMA! + fmaddsubps %6, %6, %12, %14 ; w[0..8] even + fmaddsubps %8, %8, %10, %16 ; w[0..8] odd + fmsubaddps %7, %7, %12, %15 ; j[0..8] even + fmsubaddps %9, %9, %10, %17 ; j[0..8] odd + movaps %13, [mask_pmpmpmpm] ; "subaddps? pfft, who needs that!" +%else + mulps %6, %6, %12 ; m2,3[01]imre * cos0246 + mulps %8, %8, %10 ; m2,3[01]imre * cos0246 + movaps %13, [mask_pmpmpmpm] ; "subaddps? pfft, who needs that!" + mulps %7, %7, %12 ; m2,3[23]reim * cos0246 + mulps %9, %9, %10 ; m2,3[23]reim * cos0246 + addsubps %6, %6, %14 ; w[0..8] + addsubps %8, %8, %16 ; w[0..8] + xorps %15, %15, %13 ; +-m2,3[23]imre * wim7531 + xorps %17, %17, %13 ; +-m2,3[23]imre * wim7531 + addps %7, %7, %15 ; j[0..8] + addps %9, %9, %17 ; j[0..8] +%endif + + addps %14, %6, %7 ; t10235476 even + addps %16, %8, %9 ; t10235476 odd + subps %15, %6, %7 ; +-r[0..7] even + subps %17, %8, %9 ; +-r[0..7] odd + + shufps %14, %14, %14, q2301 ; t[0..7] even + shufps %16, %16, %16, q2301 ; t[0..7] odd + xorps %15, %15, %13 ; r[0..7] even + xorps %17, %17, %13 ; r[0..7] odd + + subps %6, %2, %14 ; m2,3[01] even + subps %8, %4, %16 ; m2,3[01] odd + subps %7, %3, %15 ; m2,3[23] even + subps %9, %5, %17 ; m2,3[23] odd + + addps %2, %2, %14 ; m0 even + addps %4, %4, %16 ; m0 odd + addps %3, %3, %15 ; m1 even + addps %5, %5, %17 ; m1 odd +%endmacro + +; Same as above, only does one parity at a time, takes 3 temporary registers, +; however, if the twiddles aren't needed after this, the registers they use +; can be used as any of the temporary registers. +%macro SPLIT_RADIX_COMBINE_HALF 10 +%if %1 + shufps %8, %6, %6, q2200 ; cos00224466 + shufps %9, %7, %7, q1133 ; wim77553311 +%else + shufps %8, %6, %6, q3311 ; cos11335577 + shufps %9, %7, %7, q0022 ; wim66442200 +%endif + + mulps %10, %4, %9 ; m2,3[01]reim * wim7531 even + mulps %9, %9, %5 ; m2,3[23]reim * wim7531 even + + shufps %4, %4, %4, q2301 ; m2[0].imre, m2[1].imre, m3[0].imre, m3[1].imre even + shufps %5, %5, %5, q2301 ; m2[2].imre, m2[3].imre, m3[2].imre, m3[3].imre even + +%if cpuflag(fma3) + fmaddsubps %4, %4, %8, %10 ; w[0..8] even + fmsubaddps %5, %5, %8, %9 ; j[0..8] even + movaps %10, [mask_pmpmpmpm] +%else + mulps %4, %4, %8 ; m2,3[01]imre * cos0246 + mulps %5, %5, %8 ; m2,3[23]reim * cos0246 + addsubps %4, %4, %10 ; w[0..8] + movaps %10, [mask_pmpmpmpm] + xorps %9, %9, %10 ; +-m2,3[23]imre * wim7531 + addps %5, %5, %9 ; j[0..8] +%endif + + addps %8, %4, %5 ; t10235476 + subps %9, %4, %5 ; +-r[0..7] + + shufps %8, %8, %8, q2301 ; t[0..7] + xorps %9, %9, %10 ; r[0..7] + + subps %4, %2, %8 ; %3,3[01] + subps %5, %3, %9 ; %3,3[23] + + addps %2, %2, %8 ; m0 + addps %3, %3, %9 ; m1 +%endmacro + +; Same as above, tries REALLY hard to use 2 temporary registers. +%macro SPLIT_RADIX_COMBINE_LITE 9 +%if %1 + shufps %8, %6, %6, q2200 ; cos00224466 + shufps %9, %7, %7, q1133 ; wim77553311 +%else + shufps %8, %6, %6, q3311 ; cos11335577 + shufps %9, %7, %7, q0022 ; wim66442200 +%endif + + mulps %9, %9, %4 ; m2,3[01]reim * wim7531 even + shufps %4, %4, %4, q2301 ; m2[0].imre, m2[1].imre, m3[0].imre, m3[1].imre even + +%if cpuflag(fma3) + fmaddsubps %4, %4, %8, %9 ; w[0..8] even +%else + mulps %4, %4, %8 ; m2,3[01]imre * cos0246 + addsubps %4, %4, %9 ; w[0..8] +%endif + +%if %1 + shufps %9, %7, %7, q1133 ; wim77553311 +%else + shufps %9, %7, %7, q0022 ; wim66442200 +%endif + + mulps %9, %9, %5 ; m2,3[23]reim * wim7531 even + shufps %5, %5, %5, q2301 ; m2[2].imre, m2[3].imre, m3[2].imre, m3[3].imre even +%if cpuflag (fma3) + fmsubaddps %5, %5, %8, %9 ; j[0..8] even +%else + mulps %5, %5, %8 ; m2,3[23]reim * cos0246 + xorps %9, %9, [mask_pmpmpmpm] ; +-m2,3[23]imre * wim7531 + addps %5, %5, %9 ; j[0..8] +%endif + + addps %8, %4, %5 ; t10235476 + subps %9, %4, %5 ; +-r[0..7] + + shufps %8, %8, %8, q2301 ; t[0..7] + xorps %9, %9, [mask_pmpmpmpm] ; r[0..7] + + subps %4, %2, %8 ; %3,3[01] + subps %5, %3, %9 ; %3,3[23] + + addps %2, %2, %8 ; m0 + addps %3, %3, %9 ; m1 +%endmacro + +%macro SPLIT_RADIX_COMBINE_64 0 + SPLIT_RADIX_COMBINE_LITE 1, m0, m1, tx1_e0, tx2_e0, tw_e, tw_o, tmp1, tmp2 + + movaps [outq + 0*mmsize], m0 + movaps [outq + 4*mmsize], m1 + movaps [outq + 8*mmsize], tx1_e0 + movaps [outq + 12*mmsize], tx2_e0 + + SPLIT_RADIX_COMBINE_HALF 0, m2, m3, tx1_o0, tx2_o0, tw_e, tw_o, tmp1, tmp2, m0 + + movaps [outq + 2*mmsize], m2 + movaps [outq + 6*mmsize], m3 + movaps [outq + 10*mmsize], tx1_o0 + movaps [outq + 14*mmsize], tx2_o0 + + movaps tw_e, [tab_64_float + mmsize] + vperm2f128 tw_o, tw_o, [tab_64_float + 64 - 4*7 - mmsize], 0x23 + + movaps m0, [outq + 1*mmsize] + movaps m1, [outq + 3*mmsize] + movaps m2, [outq + 5*mmsize] + movaps m3, [outq + 7*mmsize] + + SPLIT_RADIX_COMBINE 0, m0, m2, m1, m3, tx1_e1, tx2_e1, tx1_o1, tx2_o1, tw_e, tw_o, \ + tmp1, tmp2, tx2_o0, tx1_o0, tx2_e0, tx1_e0 ; temporary registers + + movaps [outq + 1*mmsize], m0 + movaps [outq + 3*mmsize], m1 + movaps [outq + 5*mmsize], m2 + movaps [outq + 7*mmsize], m3 + + movaps [outq + 9*mmsize], tx1_e1 + movaps [outq + 11*mmsize], tx1_o1 + movaps [outq + 13*mmsize], tx2_e1 + movaps [outq + 15*mmsize], tx2_o1 +%endmacro + +; Perform a single even/odd split radix combination with loads and stores +; The _4 indicates this is a quarter of the iterations required to complete a full +; combine loop +; %1 must contain len*2, %2 must contain len*4, %3 must contain len*6 +%macro SPLIT_RADIX_LOAD_COMBINE_4 8 + movaps m8, [rtabq + (%5)*mmsize + %7] + vperm2f128 m9, m9, [itabq - (%5)*mmsize + %8], 0x23 + + movaps m0, [outq + (0 + %4)*mmsize + %6] + movaps m2, [outq + (2 + %4)*mmsize + %6] + movaps m1, [outq + %1 + (0 + %4)*mmsize + %6] + movaps m3, [outq + %1 + (2 + %4)*mmsize + %6] + + movaps m4, [outq + %2 + (0 + %4)*mmsize + %6] + movaps m6, [outq + %2 + (2 + %4)*mmsize + %6] + movaps m5, [outq + %3 + (0 + %4)*mmsize + %6] + movaps m7, [outq + %3 + (2 + %4)*mmsize + %6] + + SPLIT_RADIX_COMBINE 0, m0, m1, m2, m3, \ + m4, m5, m6, m7, \ + m8, m9, \ + m10, m11, m12, m13, m14, m15 + + movaps [outq + (0 + %4)*mmsize + %6], m0 + movaps [outq + (2 + %4)*mmsize + %6], m2 + movaps [outq + %1 + (0 + %4)*mmsize + %6], m1 + movaps [outq + %1 + (2 + %4)*mmsize + %6], m3 + + movaps [outq + %2 + (0 + %4)*mmsize + %6], m4 + movaps [outq + %2 + (2 + %4)*mmsize + %6], m6 + movaps [outq + %3 + (0 + %4)*mmsize + %6], m5 + movaps [outq + %3 + (2 + %4)*mmsize + %6], m7 +%endmacro + +%macro SPLIT_RADIX_LOAD_COMBINE_FULL 2-5 +%if %0 > 2 +%define offset_c %3 +%else +%define offset_c 0 +%endif +%if %0 > 3 +%define offset_r %4 +%else +%define offset_r 0 +%endif +%if %0 > 4 +%define offset_i %5 +%else +%define offset_i 0 +%endif + + SPLIT_RADIX_LOAD_COMBINE_4 %1, 2*%1, %2, 0, 0, offset_c, offset_r, offset_i + SPLIT_RADIX_LOAD_COMBINE_4 %1, 2*%1, %2, 1, 1, offset_c, offset_r, offset_i + SPLIT_RADIX_LOAD_COMBINE_4 %1, 2*%1, %2, 4, 2, offset_c, offset_r, offset_i + SPLIT_RADIX_LOAD_COMBINE_4 %1, 2*%1, %2, 5, 3, offset_c, offset_r, offset_i +%endmacro + +; Perform a single even/odd split radix combination with loads, deinterleaves and +; stores. The _2 indicates this is a half of the iterations required to complete +; a full combine+deinterleave loop +; %3 must contain len*2, %4 must contain len*4, %5 must contain len*6 +%macro SPLIT_RADIX_COMBINE_DEINTERLEAVE_2 6 + movaps m8, [rtabq + (0 + %2)*mmsize] + vperm2f128 m9, m9, [itabq - (0 + %2)*mmsize], 0x23 + + movaps m0, [outq + (0 + 0 + %1)*mmsize + %6] + movaps m2, [outq + (2 + 0 + %1)*mmsize + %6] + movaps m1, [outq + %3 + (0 + 0 + %1)*mmsize + %6] + movaps m3, [outq + %3 + (2 + 0 + %1)*mmsize + %6] + + movaps m4, [outq + %4 + (0 + 0 + %1)*mmsize + %6] + movaps m6, [outq + %4 + (2 + 0 + %1)*mmsize + %6] + movaps m5, [outq + %5 + (0 + 0 + %1)*mmsize + %6] + movaps m7, [outq + %5 + (2 + 0 + %1)*mmsize + %6] + + SPLIT_RADIX_COMBINE 0, m0, m1, m2, m3, \ + m4, m5, m6, m7, \ + m8, m9, \ + m10, m11, m12, m13, m14, m15 + + unpckhpd m10, m0, m2 + unpckhpd m11, m1, m3 + unpckhpd m12, m4, m6 + unpckhpd m13, m5, m7 + unpcklpd m0, m0, m2 + unpcklpd m1, m1, m3 + unpcklpd m4, m4, m6 + unpcklpd m5, m5, m7 + + vextractf128 [outq + (0 + 0 + %1)*mmsize + %6 + 0], m0, 0 + vextractf128 [outq + (0 + 0 + %1)*mmsize + %6 + 16], m10, 0 + vextractf128 [outq + %3 + (0 + 0 + %1)*mmsize + %6 + 0], m1, 0 + vextractf128 [outq + %3 + (0 + 0 + %1)*mmsize + %6 + 16], m11, 0 + + vextractf128 [outq + %4 + (0 + 0 + %1)*mmsize + %6 + 0], m4, 0 + vextractf128 [outq + %4 + (0 + 0 + %1)*mmsize + %6 + 16], m12, 0 + vextractf128 [outq + %5 + (0 + 0 + %1)*mmsize + %6 + 0], m5, 0 + vextractf128 [outq + %5 + (0 + 0 + %1)*mmsize + %6 + 16], m13, 0 + + vperm2f128 m10, m10, m0, 0x13 + vperm2f128 m11, m11, m1, 0x13 + vperm2f128 m12, m12, m4, 0x13 + vperm2f128 m13, m13, m5, 0x13 + + movaps m8, [rtabq + (1 + %2)*mmsize] + vperm2f128 m9, m9, [itabq - (1 + %2)*mmsize], 0x23 + + movaps m0, [outq + (0 + 1 + %1)*mmsize + %6] + movaps m2, [outq + (2 + 1 + %1)*mmsize + %6] + movaps m1, [outq + %3 + (0 + 1 + %1)*mmsize + %6] + movaps m3, [outq + %3 + (2 + 1 + %1)*mmsize + %6] + + movaps [outq + (0 + 1 + %1)*mmsize + %6], m10 ; m0 conflict + movaps [outq + %3 + (0 + 1 + %1)*mmsize + %6], m11 ; m1 conflict + + movaps m4, [outq + %4 + (0 + 1 + %1)*mmsize + %6] + movaps m6, [outq + %4 + (2 + 1 + %1)*mmsize + %6] + movaps m5, [outq + %5 + (0 + 1 + %1)*mmsize + %6] + movaps m7, [outq + %5 + (2 + 1 + %1)*mmsize + %6] + + movaps [outq + %4 + (0 + 1 + %1)*mmsize + %6], m12 ; m4 conflict + movaps [outq + %5 + (0 + 1 + %1)*mmsize + %6], m13 ; m5 conflict + + SPLIT_RADIX_COMBINE 0, m0, m1, m2, m3, \ + m4, m5, m6, m7, \ + m8, m9, \ + m10, m11, m12, m13, m14, m15 ; temporary registers + + unpcklpd m8, m0, m2 + unpcklpd m9, m1, m3 + unpcklpd m10, m4, m6 + unpcklpd m11, m5, m7 + unpckhpd m0, m0, m2 + unpckhpd m1, m1, m3 + unpckhpd m4, m4, m6 + unpckhpd m5, m5, m7 + + vextractf128 [outq + (2 + 0 + %1)*mmsize + %6 + 0], m8, 0 + vextractf128 [outq + (2 + 0 + %1)*mmsize + %6 + 16], m0, 0 + vextractf128 [outq + (2 + 1 + %1)*mmsize + %6 + 0], m8, 1 + vextractf128 [outq + (2 + 1 + %1)*mmsize + %6 + 16], m0, 1 + + vextractf128 [outq + %3 + (2 + 0 + %1)*mmsize + %6 + 0], m9, 0 + vextractf128 [outq + %3 + (2 + 0 + %1)*mmsize + %6 + 16], m1, 0 + vextractf128 [outq + %3 + (2 + 1 + %1)*mmsize + %6 + 0], m9, 1 + vextractf128 [outq + %3 + (2 + 1 + %1)*mmsize + %6 + 16], m1, 1 + + vextractf128 [outq + %4 + (2 + 0 + %1)*mmsize + %6 + 0], m10, 0 + vextractf128 [outq + %4 + (2 + 0 + %1)*mmsize + %6 + 16], m4, 0 + vextractf128 [outq + %4 + (2 + 1 + %1)*mmsize + %6 + 0], m10, 1 + vextractf128 [outq + %4 + (2 + 1 + %1)*mmsize + %6 + 16], m4, 1 + + vextractf128 [outq + %5 + (2 + 0 + %1)*mmsize + %6 + 0], m11, 0 + vextractf128 [outq + %5 + (2 + 0 + %1)*mmsize + %6 + 16], m5, 0 + vextractf128 [outq + %5 + (2 + 1 + %1)*mmsize + %6 + 0], m11, 1 + vextractf128 [outq + %5 + (2 + 1 + %1)*mmsize + %6 + 16], m5, 1 +%endmacro + +%macro SPLIT_RADIX_COMBINE_DEINTERLEAVE_FULL 2-3 +%if %0 > 2 +%define offset %3 +%else +%define offset 0 +%endif + SPLIT_RADIX_COMBINE_DEINTERLEAVE_2 0, 0, %1, %1*2, %2, offset + SPLIT_RADIX_COMBINE_DEINTERLEAVE_2 4, 2, %1, %1*2, %2, offset +%endmacro + +INIT_XMM sse3 +cglobal fft2_asm_float, 0, 0, 0, ctx, out, in, stride + movaps m0, [inq] + FFT2 m0, m1 + movaps [outq], m0 + ret + +cglobal fft2_float, 4, 4, 2, ctx, out, in, stride + movaps m0, [inq] + FFT2 m0, m1 + movaps [outq], m0 + RET + +%macro FFT4_FN 3 +INIT_XMM sse2 +%if %3 +cglobal fft4_ %+ %1 %+ _asm_float, 0, 0, 0, ctx, out, in, stride +%else +cglobal fft4_ %+ %1 %+ _float, 4, 4, 3, ctx, out, in, stride +%endif + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] + +%if %2 + shufps m2, m1, m0, q3210 + shufps m0, m0, m1, q3210 + movaps m1, m2 +%endif + + FFT4 m0, m1, m2 + + unpcklpd m2, m0, m1 + unpckhpd m0, m0, m1 + + movaps [outq + 0*mmsize], m2 + movaps [outq + 1*mmsize], m0 + +%if %3 + ret +%else + RET +%endif +%endmacro + +FFT4_FN fwd, 0, 0 +FFT4_FN fwd, 0, 1 +FFT4_FN inv, 1, 0 +FFT4_FN inv, 1, 1 + +%macro FFT8_SSE_FN 1 +INIT_XMM sse3 +%if %1 +cglobal fft8_asm_float, 0, 0, 0, ctx, out, in, stride, tmp + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] + movaps m2, [inq + 2*mmsize] + movaps m3, [inq + 3*mmsize] +%else +cglobal fft8_float, 4, 4, 6, ctx, out, in, tmp + mov ctxq, [ctxq + AVTXContext.map] + LOAD64_LUT m0, inq, ctxq, (mmsize/2)*0, tmpq + LOAD64_LUT m1, inq, ctxq, (mmsize/2)*1, tmpq + LOAD64_LUT m2, inq, ctxq, (mmsize/2)*2, tmpq + LOAD64_LUT m3, inq, ctxq, (mmsize/2)*3, tmpq +%endif + + FFT8 m0, m1, m2, m3, m4, m5 + + unpcklpd m4, m0, m3 + unpcklpd m5, m1, m2 + unpckhpd m0, m0, m3 + unpckhpd m1, m1, m2 + + movups [outq + 0*mmsize], m4 + movups [outq + 1*mmsize], m0 + movups [outq + 2*mmsize], m5 + movups [outq + 3*mmsize], m1 + +%if %1 + ret +%else + RET +%endif + +%if %1 +cglobal fft8_ns_float, 4, 5, 6, ctx, out, in, stride, tmp + call mangle(ff_tx_fft8_asm_float_sse3) + RET +%endif +%endmacro + +FFT8_SSE_FN 0 +FFT8_SSE_FN 1 + +%macro FFT8_AVX_FN 1 +INIT_YMM avx +%if %1 +cglobal fft8_asm_float, 0, 0, 0, ctx, out, in, stride, tmp + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] +%else +cglobal fft8_float, 4, 4, 4, ctx, out, in, tmp + mov ctxq, [ctxq + AVTXContext.map] + LOAD64_LUT m0, inq, ctxq, (mmsize/2)*0, tmpq, m2 + LOAD64_LUT m1, inq, ctxq, (mmsize/2)*1, tmpq, m3 +%endif + + FFT8_AVX m0, m1, m2, m3 + + unpcklpd m2, m0, m1 + unpckhpd m0, m0, m1 + + ; Around 2% faster than 2x vperm2f128 + 2x movapd + vextractf128 [outq + 16*0], m2, 0 + vextractf128 [outq + 16*1], m0, 0 + vextractf128 [outq + 16*2], m2, 1 + vextractf128 [outq + 16*3], m0, 1 + +%if %1 + ret +%else + RET +%endif + +%if %1 +cglobal fft8_ns_float, 4, 5, 4, ctx, out, in, stride, tmp + call mangle(ff_tx_fft8_asm_float_avx) + RET +%endif +%endmacro + +FFT8_AVX_FN 0 +FFT8_AVX_FN 1 + +%macro FFT16_FN 2 +INIT_YMM %1 +%if %2 +cglobal fft16_asm_float, 0, 0, 0, ctx, out, in, stride, tmp + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] + movaps m2, [inq + 2*mmsize] + movaps m3, [inq + 3*mmsize] +%else +cglobal fft16_float, 4, 4, 8, ctx, out, in, tmp + mov ctxq, [ctxq + AVTXContext.map] + LOAD64_LUT m0, inq, ctxq, (mmsize/2)*0, tmpq, m4 + LOAD64_LUT m1, inq, ctxq, (mmsize/2)*1, tmpq, m5 + LOAD64_LUT m2, inq, ctxq, (mmsize/2)*2, tmpq, m6 + LOAD64_LUT m3, inq, ctxq, (mmsize/2)*3, tmpq, m7 +%endif + + FFT16 m0, m1, m2, m3, m4, m5, m6, m7 + + unpcklpd m5, m1, m3 + unpcklpd m4, m0, m2 + unpckhpd m1, m1, m3 + unpckhpd m0, m0, m2 + + vextractf128 [outq + 16*0], m4, 0 + vextractf128 [outq + 16*1], m0, 0 + vextractf128 [outq + 16*2], m4, 1 + vextractf128 [outq + 16*3], m0, 1 + vextractf128 [outq + 16*4], m5, 0 + vextractf128 [outq + 16*5], m1, 0 + vextractf128 [outq + 16*6], m5, 1 + vextractf128 [outq + 16*7], m1, 1 + +%if %2 + ret +%else + RET +%endif + +%if %2 +cglobal fft16_ns_float, 4, 5, 8, ctx, out, in, stride, tmp + call mangle(ff_tx_fft16_asm_float_ %+ %1) + RET +%endif +%endmacro + +FFT16_FN avx, 0 +FFT16_FN avx, 1 +FFT16_FN fma3, 0 +FFT16_FN fma3, 1 + +%macro FFT32_FN 2 +INIT_YMM %1 +%if %2 +cglobal fft32_asm_float, 0, 0, 0, ctx, out, in, stride, tmp + movaps m4, [inq + 4*mmsize] + movaps m5, [inq + 5*mmsize] + movaps m6, [inq + 6*mmsize] + movaps m7, [inq + 7*mmsize] +%else +cglobal fft32_float, 4, 4, 16, ctx, out, in, tmp + mov ctxq, [ctxq + AVTXContext.map] + LOAD64_LUT m4, inq, ctxq, (mmsize/2)*4, tmpq, m8, m12 + LOAD64_LUT m5, inq, ctxq, (mmsize/2)*5, tmpq, m9, m13 + LOAD64_LUT m6, inq, ctxq, (mmsize/2)*6, tmpq, m10, m14 + LOAD64_LUT m7, inq, ctxq, (mmsize/2)*7, tmpq, m11, m15 +%endif + + FFT8 m4, m5, m6, m7, m8, m9 + +%if %2 + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] + movaps m2, [inq + 2*mmsize] + movaps m3, [inq + 3*mmsize] +%else + LOAD64_LUT m0, inq, ctxq, (mmsize/2)*0, tmpq, m8, m12 + LOAD64_LUT m1, inq, ctxq, (mmsize/2)*1, tmpq, m9, m13 + LOAD64_LUT m2, inq, ctxq, (mmsize/2)*2, tmpq, m10, m14 + LOAD64_LUT m3, inq, ctxq, (mmsize/2)*3, tmpq, m11, m15 +%endif + + movaps m8, [tab_32_float] + vperm2f128 m9, m9, [tab_32_float + 4*8 - 4*7], 0x23 + + FFT16 m0, m1, m2, m3, m10, m11, m12, m13 + + SPLIT_RADIX_COMBINE 1, m0, m1, m2, m3, m4, m5, m6, m7, m8, m9, \ + m10, m11, m12, m13, m14, m15 ; temporary registers + + unpcklpd m9, m1, m3 + unpcklpd m10, m5, m7 + unpcklpd m8, m0, m2 + unpcklpd m11, m4, m6 + unpckhpd m1, m1, m3 + unpckhpd m5, m5, m7 + unpckhpd m0, m0, m2 + unpckhpd m4, m4, m6 + + vextractf128 [outq + 16* 0], m8, 0 + vextractf128 [outq + 16* 1], m0, 0 + vextractf128 [outq + 16* 2], m8, 1 + vextractf128 [outq + 16* 3], m0, 1 + vextractf128 [outq + 16* 4], m9, 0 + vextractf128 [outq + 16* 5], m1, 0 + vextractf128 [outq + 16* 6], m9, 1 + vextractf128 [outq + 16* 7], m1, 1 + + vextractf128 [outq + 16* 8], m11, 0 + vextractf128 [outq + 16* 9], m4, 0 + vextractf128 [outq + 16*10], m11, 1 + vextractf128 [outq + 16*11], m4, 1 + vextractf128 [outq + 16*12], m10, 0 + vextractf128 [outq + 16*13], m5, 0 + vextractf128 [outq + 16*14], m10, 1 + vextractf128 [outq + 16*15], m5, 1 + +%if %2 + ret +%else + RET +%endif + +%if %2 +cglobal fft32_ns_float, 4, 5, 16, ctx, out, in, stride, tmp + call mangle(ff_tx_fft32_asm_float_ %+ %1) + RET +%endif +%endmacro + +%if ARCH_X86_64 +FFT32_FN avx, 0 +FFT32_FN avx, 1 +FFT32_FN fma3, 0 +FFT32_FN fma3, 1 +%endif + +%macro FFT_SPLIT_RADIX_DEF 1-2 +ALIGN 16 +.%1 %+ pt: + PUSH lenq + mov lenq, (%1/4) + + add outq, (%1*4) - (%1/1) + call .32pt + + add outq, (%1*2) - (%1/2) ; the synth loops also increment outq + call .32pt + + POP lenq + sub outq, (%1*4) + (%1*2) + (%1/2) + + lea rtabq, [tab_ %+ %1 %+ _float] + lea itabq, [tab_ %+ %1 %+ _float + %1 - 4*7] + +%if %0 > 1 + cmp tgtq, %1 + je .deinterleave + + mov tmpq, %1 + +.synth_ %+ %1: + SPLIT_RADIX_LOAD_COMBINE_FULL 2*%1, 6*%1, 0, 0, 0 + add outq, 8*mmsize + add rtabq, 4*mmsize + sub itabq, 4*mmsize + sub tmpq, 4*mmsize + jg .synth_ %+ %1 + + cmp lenq, %1 + jg %2 ; can't do math here, nasm doesn't get it + ret +%endif +%endmacro + +%macro FFT_SPLIT_RADIX_FN 2 +INIT_YMM %1 +%if %2 +cglobal fft_sr_asm_float, 0, 0, 0, ctx, out, in, stride, len, lut, itab, rtab, tgt, tmp +%else +cglobal fft_sr_float, 4, 10, 16, 272, ctx, out, in, stride, len, lut, itab, rtab, tgt, tmp + movsxd lenq, dword [ctxq + AVTXContext.len] + mov lutq, [ctxq + AVTXContext.map] +%endif + mov tgtq, lenq + +; Bottom-most/32-point transform =============================================== +ALIGN 16 +.32pt: +%if %2 + movaps m4, [inq + 4*mmsize] + movaps m5, [inq + 5*mmsize] + movaps m6, [inq + 6*mmsize] + movaps m7, [inq + 7*mmsize] +%else + LOAD64_LUT m4, inq, lutq, (mmsize/2)*4, tmpq, m8, m12 + LOAD64_LUT m5, inq, lutq, (mmsize/2)*5, tmpq, m9, m13 + LOAD64_LUT m6, inq, lutq, (mmsize/2)*6, tmpq, m10, m14 + LOAD64_LUT m7, inq, lutq, (mmsize/2)*7, tmpq, m11, m15 +%endif + + FFT8 m4, m5, m6, m7, m8, m9 + +%if %2 + movaps m0, [inq + 0*mmsize] + movaps m1, [inq + 1*mmsize] + movaps m2, [inq + 2*mmsize] + movaps m3, [inq + 3*mmsize] +%else + LOAD64_LUT m0, inq, lutq, (mmsize/2)*0, tmpq, m8, m12 + LOAD64_LUT m1, inq, lutq, (mmsize/2)*1, tmpq, m9, m13 + LOAD64_LUT m2, inq, lutq, (mmsize/2)*2, tmpq, m10, m14 + LOAD64_LUT m3, inq, lutq, (mmsize/2)*3, tmpq, m11, m15 +%endif + + movaps m8, [tab_32_float] + vperm2f128 m9, m9, [tab_32_float + 32 - 4*7], 0x23 + + FFT16 m0, m1, m2, m3, m10, m11, m12, m13 + + SPLIT_RADIX_COMBINE 1, m0, m1, m2, m3, m4, m5, m6, m7, m8, m9, \ + m10, m11, m12, m13, m14, m15 ; temporary registers + + movaps [outq + 1*mmsize], m1 + movaps [outq + 3*mmsize], m3 + movaps [outq + 5*mmsize], m5 + movaps [outq + 7*mmsize], m7 + +%if %2 + add inq, 8*mmsize +%else + add lutq, (mmsize/2)*8 +%endif + cmp lenq, 32 + jg .64pt + + movaps [outq + 0*mmsize], m0 + movaps [outq + 2*mmsize], m2 + movaps [outq + 4*mmsize], m4 + movaps [outq + 6*mmsize], m6 + + ret + +; 64-point transform =========================================================== +ALIGN 16 +.64pt: +; Helper defines, these make it easier to track what's happening +%define tx1_e0 m4 +%define tx1_e1 m5 +%define tx1_o0 m6 +%define tx1_o1 m7 +%define tx2_e0 m8 +%define tx2_e1 m9 +%define tx2_o0 m10 +%define tx2_o1 m11 +%define tw_e m12 +%define tw_o m13 +%define tmp1 m14 +%define tmp2 m15 + + SWAP m4, m1 + SWAP m6, m3 + +%if %2 + movaps tx1_e0, [inq + 0*mmsize] + movaps tx1_e1, [inq + 1*mmsize] + movaps tx1_o0, [inq + 2*mmsize] + movaps tx1_o1, [inq + 3*mmsize] +%else + LOAD64_LUT tx1_e0, inq, lutq, (mmsize/2)*0, tmpq, tw_e, tmp1 + LOAD64_LUT tx1_e1, inq, lutq, (mmsize/2)*1, tmpq, tw_o, tmp2 + LOAD64_LUT tx1_o0, inq, lutq, (mmsize/2)*2, tmpq, tw_e, tmp1 + LOAD64_LUT tx1_o1, inq, lutq, (mmsize/2)*3, tmpq, tw_o, tmp2 +%endif + + FFT16 tx1_e0, tx1_e1, tx1_o0, tx1_o1, tw_e, tw_o, tx2_o0, tx2_o1 + +%if %2 + movaps tx2_e0, [inq + 4*mmsize] + movaps tx2_e1, [inq + 5*mmsize] + movaps tx2_o0, [inq + 6*mmsize] + movaps tx2_o1, [inq + 7*mmsize] +%else + LOAD64_LUT tx2_e0, inq, lutq, (mmsize/2)*4, tmpq, tw_e, tmp1 + LOAD64_LUT tx2_e1, inq, lutq, (mmsize/2)*5, tmpq, tw_o, tmp2 + LOAD64_LUT tx2_o0, inq, lutq, (mmsize/2)*6, tmpq, tw_e, tmp1 + LOAD64_LUT tx2_o1, inq, lutq, (mmsize/2)*7, tmpq, tw_o, tmp2 +%endif + + FFT16 tx2_e0, tx2_e1, tx2_o0, tx2_o1, tmp1, tmp2, tw_e, tw_o + + movaps tw_e, [tab_64_float] + vperm2f128 tw_o, tw_o, [tab_64_float + 64 - 4*7], 0x23 + +%if %2 + add inq, 8*mmsize +%else + add lutq, (mmsize/2)*8 +%endif + cmp tgtq, 64 + je .64pt_deint + + SPLIT_RADIX_COMBINE_64 + + cmp lenq, 64 + jg .128pt + ret + +; 128-point transform ========================================================== +ALIGN 16 +.128pt: + PUSH lenq + mov lenq, 32 + + add outq, 16*mmsize + call .32pt + + add outq, 8*mmsize + call .32pt + + POP lenq + sub outq, 24*mmsize + + lea rtabq, [tab_128_float] + lea itabq, [tab_128_float + 128 - 4*7] + + cmp tgtq, 128 + je .deinterleave + + SPLIT_RADIX_LOAD_COMBINE_FULL 2*128, 6*128 + + cmp lenq, 128 + jg .256pt + ret + +; 256-point transform ========================================================== +ALIGN 16 +.256pt: + PUSH lenq + mov lenq, 64 + + add outq, 32*mmsize + call .32pt + + add outq, 16*mmsize + call .32pt + + POP lenq + sub outq, 48*mmsize + + lea rtabq, [tab_256_float] + lea itabq, [tab_256_float + 256 - 4*7] + + cmp tgtq, 256 + je .deinterleave + + SPLIT_RADIX_LOAD_COMBINE_FULL 2*256, 6*256 + SPLIT_RADIX_LOAD_COMBINE_FULL 2*256, 6*256, 8*mmsize, 4*mmsize, -4*mmsize + + cmp lenq, 256 + jg .512pt + ret + +; 512-point transform ========================================================== +ALIGN 16 +.512pt: + PUSH lenq + mov lenq, 128 + + add outq, 64*mmsize + call .32pt + + add outq, 32*mmsize + call .32pt + + POP lenq + sub outq, 96*mmsize + + lea rtabq, [tab_512_float] + lea itabq, [tab_512_float + 512 - 4*7] + + cmp tgtq, 512 + je .deinterleave + + mov tmpq, 4 + +.synth_512: + SPLIT_RADIX_LOAD_COMBINE_FULL 2*512, 6*512 + add outq, 8*mmsize + add rtabq, 4*mmsize + sub itabq, 4*mmsize + sub tmpq, 1 + jg .synth_512 + + cmp lenq, 512 + jg .1024pt + ret + +; 1024-point transform ========================================================== +ALIGN 16 +.1024pt: + PUSH lenq + mov lenq, 256 + + add outq, 96*mmsize + call .32pt + + add outq, 64*mmsize + call .32pt + + POP lenq + sub outq, 192*mmsize + + lea rtabq, [tab_1024_float] + lea itabq, [tab_1024_float + 1024 - 4*7] + + cmp tgtq, 1024 + je .deinterleave + + mov tmpq, 8 + +.synth_1024: + SPLIT_RADIX_LOAD_COMBINE_FULL 2*1024, 6*1024 + add outq, 8*mmsize + add rtabq, 4*mmsize + sub itabq, 4*mmsize + sub tmpq, 1 + jg .synth_1024 + + cmp lenq, 1024 + jg .2048pt + ret + +; 2048 to 131072-point transforms ============================================== +FFT_SPLIT_RADIX_DEF 2048, .4096pt +FFT_SPLIT_RADIX_DEF 4096, .8192pt +FFT_SPLIT_RADIX_DEF 8192, .16384pt +FFT_SPLIT_RADIX_DEF 16384, .32768pt +FFT_SPLIT_RADIX_DEF 32768, .65536pt +FFT_SPLIT_RADIX_DEF 65536, .131072pt +FFT_SPLIT_RADIX_DEF 131072 + +;=============================================================================== +; Final synthesis + deinterleaving code +;=============================================================================== +.deinterleave: +%if %2 + PUSH strideq +%endif + mov tgtq, lenq + imul tmpq, lenq, 2 + lea strideq, [4*lenq + tmpq] + +.synth_deinterleave: + SPLIT_RADIX_COMBINE_DEINTERLEAVE_FULL tmpq, strideq + add outq, 8*mmsize + add rtabq, 4*mmsize + sub itabq, 4*mmsize + sub tgtq, 4*mmsize + jg .synth_deinterleave + +%if %2 + POP strideq + sub outq, tmpq + neg tmpq + lea inq, [inq + tmpq*4] + ret +%else + RET +%endif + +; 64-point deinterleave which only has to load 4 registers ===================== +.64pt_deint: + SPLIT_RADIX_COMBINE_LITE 1, m0, m1, tx1_e0, tx2_e0, tw_e, tw_o, tmp1, tmp2 + SPLIT_RADIX_COMBINE_HALF 0, m2, m3, tx1_o0, tx2_o0, tw_e, tw_o, tmp1, tmp2, tw_e + + unpcklpd tmp1, m0, m2 + unpcklpd tmp2, m1, m3 + unpcklpd tw_o, tx1_e0, tx1_o0 + unpcklpd tw_e, tx2_e0, tx2_o0 + unpckhpd m0, m0, m2 + unpckhpd m1, m1, m3 + unpckhpd tx1_e0, tx1_e0, tx1_o0 + unpckhpd tx2_e0, tx2_e0, tx2_o0 + + vextractf128 [outq + 0*mmsize + 0], tmp1, 0 + vextractf128 [outq + 0*mmsize + 16], m0, 0 + vextractf128 [outq + 4*mmsize + 0], tmp2, 0 + vextractf128 [outq + 4*mmsize + 16], m1, 0 + + vextractf128 [outq + 8*mmsize + 0], tw_o, 0 + vextractf128 [outq + 8*mmsize + 16], tx1_e0, 0 + vextractf128 [outq + 9*mmsize + 0], tw_o, 1 + vextractf128 [outq + 9*mmsize + 16], tx1_e0, 1 + + vperm2f128 tmp1, tmp1, m0, 0x31 + vperm2f128 tmp2, tmp2, m1, 0x31 + + vextractf128 [outq + 12*mmsize + 0], tw_e, 0 + vextractf128 [outq + 12*mmsize + 16], tx2_e0, 0 + vextractf128 [outq + 13*mmsize + 0], tw_e, 1 + vextractf128 [outq + 13*mmsize + 16], tx2_e0, 1 + + movaps tw_e, [tab_64_float + mmsize] + vperm2f128 tw_o, tw_o, [tab_64_float + 64 - 4*7 - mmsize], 0x23 + + movaps m0, [outq + 1*mmsize] + movaps m1, [outq + 3*mmsize] + movaps m2, [outq + 5*mmsize] + movaps m3, [outq + 7*mmsize] + + movaps [outq + 1*mmsize], tmp1 + movaps [outq + 5*mmsize], tmp2 + + SPLIT_RADIX_COMBINE 0, m0, m2, m1, m3, tx1_e1, tx2_e1, tx1_o1, tx2_o1, tw_e, tw_o, \ + tmp1, tmp2, tx2_o0, tx1_o0, tx2_e0, tx1_e0 ; temporary registers + + unpcklpd tmp1, m0, m1 + unpcklpd tmp2, m2, m3 + unpcklpd tw_e, tx1_e1, tx1_o1 + unpcklpd tw_o, tx2_e1, tx2_o1 + unpckhpd m0, m0, m1 + unpckhpd m2, m2, m3 + unpckhpd tx1_e1, tx1_e1, tx1_o1 + unpckhpd tx2_e1, tx2_e1, tx2_o1 + + vextractf128 [outq + 2*mmsize + 0], tmp1, 0 + vextractf128 [outq + 2*mmsize + 16], m0, 0 + vextractf128 [outq + 3*mmsize + 0], tmp1, 1 + vextractf128 [outq + 3*mmsize + 16], m0, 1 + + vextractf128 [outq + 6*mmsize + 0], tmp2, 0 + vextractf128 [outq + 6*mmsize + 16], m2, 0 + vextractf128 [outq + 7*mmsize + 0], tmp2, 1 + vextractf128 [outq + 7*mmsize + 16], m2, 1 + + vextractf128 [outq + 10*mmsize + 0], tw_e, 0 + vextractf128 [outq + 10*mmsize + 16], tx1_e1, 0 + vextractf128 [outq + 11*mmsize + 0], tw_e, 1 + vextractf128 [outq + 11*mmsize + 16], tx1_e1, 1 + + vextractf128 [outq + 14*mmsize + 0], tw_o, 0 + vextractf128 [outq + 14*mmsize + 16], tx2_e1, 0 + vextractf128 [outq + 15*mmsize + 0], tw_o, 1 + vextractf128 [outq + 15*mmsize + 16], tx2_e1, 1 + +%if %2 + sub inq, 16*mmsize + ret +%else + RET +%endif + +%if %2 +cglobal fft_sr_ns_float, 4, 10, 16, 272, ctx, out, in, tmp, len, lut, itab, rtab, tgt, off + movsxd lenq, dword [ctxq + AVTXContext.len] + mov lutq, [ctxq + AVTXContext.map] + + call mangle(ff_tx_fft_sr_asm_float_ %+ %1) + RET +%endif +%endmacro + +%if ARCH_X86_64 +FFT_SPLIT_RADIX_FN avx, 0 +FFT_SPLIT_RADIX_FN avx, 1 +FFT_SPLIT_RADIX_FN fma3, 0 +FFT_SPLIT_RADIX_FN fma3, 1 +%if HAVE_AVX2_EXTERNAL +FFT_SPLIT_RADIX_FN avx2, 0 +FFT_SPLIT_RADIX_FN avx2, 1 +%endif +%endif + +%macro FFT15_FN 2 +INIT_YMM avx2 +cglobal fft15_ %+ %2, 4, 10, 16, ctx, out, in, stride, len, lut, tmp, tgt5, stride3, stride5 + mov lutq, [ctxq + AVTXContext.map] + + imul stride3q, strideq, 3 + imul stride5q, strideq, 5 + + movaps m11, [mask_mmppmmmm] ; mmppmmmm + movaps m10, [tab_53_float] ; tab5 + movaps xm9, [tab_53_float + 32] ; tab3 + vpermpd m9, m9, q1110 ; tab[23232323] + movaps m8, [s15_perm] + +%if %1 + movups xm0, [inq] + movddup xm5, [inq + 16] + movups m2, [inq + mmsize*0 + 24] + movups m3, [inq + mmsize*1 + 24] + movups m4, [inq + mmsize*2 + 24] +%else + LOAD64_LUT xm0, inq, lutq, 0, tmpq, m14, xm15 + LOAD64_LUT m2, inq, lutq, (mmsize/2)*0 + 12, tmpq, m6, m7 + LOAD64_LUT m3, inq, lutq, (mmsize/2)*1 + 12, tmpq, m14, m15 + LOAD64_LUT m4, inq, lutq, (mmsize/2)*2 + 12, tmpq, m6, m7 + mov tmpd, [lutq + 8] + movddup xm5, [inq + tmpq*8] +%endif + + FFT15 + + lea tgt5q, [outq + stride5q] + lea tmpq, [outq + stride5q*2] + + movhps [outq], xm14 ; out[0] + movhps [outq + stride5q*1], xm15 ; out[5] + movlps [outq + stride5q*2], xm15 ; out[10] + + vextractf128 xm3, m0, 1 + vextractf128 xm4, m1, 1 + vextractf128 xm5, m2, 1 + + movlps [outq + strideq*1], xm1 + movhps [outq + strideq*2], xm2 + movlps [outq + stride3q*1], xm3 + movhps [outq + strideq*4], xm4 + movlps [outq + stride3q*2], xm0 + movlps [outq + strideq*8], xm5 + movhps [outq + stride3q*4], xm0 + movhps [tgt5q + strideq*2], xm1 + movhps [tgt5q + strideq*4], xm3 + movlps [tmpq + strideq*1], xm2 + movlps [tmpq + stride3q*1], xm4 + movhps [tmpq + strideq*4], xm5 + + RET +%endmacro + +%if ARCH_X86_64 && HAVE_AVX2_EXTERNAL +FFT15_FN 0, float +FFT15_FN 1, ns_float +%endif + +%macro IMDCT_FN 1 +INIT_YMM %1 +cglobal mdct_inv_float, 4, 14, 16, 320, ctx, out, in, stride, len, lut, exp, t1, t2, t3, \ + t4, t5, btmp + movsxd lenq, dword [ctxq + AVTXContext.len] + mov expq, [ctxq + AVTXContext.exp] + + lea t1d, [lend - 1] + imul t1d, strided + + mov btmpq, ctxq ; backup original context + mov lutq, [ctxq + AVTXContext.map] ; load map + + cmp strideq, 4 + je .stride4 + + shl strideq, 1 + movd xm4, strided + vpbroadcastd m4, xm4 ; stride splatted + movd xm5, t1d + vpbroadcastd m5, xm5 ; offset splatted + + mov t2q, outq ; don't modify the original output + pcmpeqd m15, m15 ; set all bits to 1 + +.stridex_pre: + pmulld m2, m4, [lutq] ; multiply by stride + movaps m0, m15 + psubd m3, m5, m2 ; subtract from offset + movaps m1, m15 + vgatherdps m6, [inq + m2], m0 ; im + vgatherdps m7, [inq + m3], m1 ; re + + movaps m8, [expq + 0*mmsize] ; tab 1 + movaps m9, [expq + 1*mmsize] ; tab 2 + + unpcklps m0, m7, m6 ; re, im, re, im + unpckhps m1, m7, m6 ; re, im, re, im + + vperm2f128 m2, m1, m0, 0x02 ; output order + vperm2f128 m3, m1, m0, 0x13 ; output order + + movshdup m10, m8 ; tab 1 imim + movshdup m11, m9 ; tab 2 imim + movsldup m12, m8 ; tab 1 rere + movsldup m13, m9 ; tab 2 rere + + mulps m10, m2 ; 1 reim * imim + mulps m11, m3 ; 2 reim * imim + + shufps m10, m10, m10, q2301 + shufps m11, m11, m11, q2301 + + fmaddsubps m10, m12, m2, m10 + fmaddsubps m11, m13, m3, m11 + + movups [t2q + 0*mmsize], m10 + movups [t2q + 1*mmsize], m11 + + add expq, mmsize*2 + add lutq, mmsize + add t2q, mmsize*2 + sub lenq, mmsize/2 + jg .stridex_pre + jmp .transform + +.stride4: + lea expq, [expq + lenq*4] + lea lutq, [lutq + lenq*2] + lea t1q, [inq + t1q] + lea t1q, [t1q + strideq - mmsize] + lea t2q, [lenq*2 - mmsize/2] + +.stride4_pre: + movups m4, [inq] + movups m3, [t1q] + + movsldup m1, m4 ; im im, im im + movshdup m0, m3 ; re re, re re + movshdup m4, m4 ; re re, re re (2) + movsldup m3, m3 ; im im, im im (2) + + movups m2, [expq] ; tab + movups m5, [expq + 2*t2q] ; tab (2) + + vpermpd m0, m0, q0123 ; flip + shufps m7, m2, m2, q2301 + vpermpd m4, m4, q0123 ; flip (2) + shufps m8, m5, m5, q2301 + + mulps m1, m7 ; im im * tab.reim + mulps m3, m8 ; im im * tab.reim (2) + + fmaddsubps m0, m0, m2, m1 + fmaddsubps m4, m4, m5, m3 + + vextractf128 xm3, m0, 1 + vextractf128 xm6, m4, 1 + + ; scatter + movsxd strideq, dword [lutq + 0*4] + movsxd lenq, dword [lutq + 1*4] + movsxd t3q, dword [lutq + 2*4] + movsxd t4q, dword [lutq + 3*4] + + movlps [outq + strideq*8], xm0 + movhps [outq + lenq*8], xm0 + movlps [outq + t3q*8], xm3 + movhps [outq + t4q*8], xm3 + + movsxd strideq, dword [lutq + 0*4 + t2q] + movsxd lenq, dword [lutq + 1*4 + t2q] + movsxd t3q, dword [lutq + 2*4 + t2q] + movsxd t4q, dword [lutq + 3*4 + t2q] + + movlps [outq + strideq*8], xm4 + movhps [outq + lenq*8], xm4 + movlps [outq + t3q*8], xm6 + movhps [outq + t4q*8], xm6 + + add lutq, mmsize/2 + add expq, mmsize + add inq, mmsize + sub t1q, mmsize + sub t2q, mmsize + jge .stride4_pre + +.transform: + mov strideq, 2*4 + mov t4q, ctxq ; backup original context + mov t5q, [ctxq + AVTXContext.fn] ; subtransform's jump point + mov ctxq, [ctxq + AVTXContext.sub] + mov lutq, [ctxq + AVTXContext.map] + movsxd lenq, dword [ctxq + AVTXContext.len] + + mov inq, outq ; in-place transform + call t5q ; call the FFT + + mov ctxq, t4q ; restore original context + movsxd lenq, dword [ctxq + AVTXContext.len] + mov expq, [ctxq + AVTXContext.exp] + lea expq, [expq + lenq*4] + + xor t1q, t1q ; low + lea t2q, [lenq*4 - mmsize] ; high + +.post: + movaps m2, [expq + t2q] ; tab h + movaps m3, [expq + t1q] ; tab l + movups m0, [outq + t2q] ; in h + movups m1, [outq + t1q] ; in l + + movshdup m4, m2 ; tab h imim + movshdup m5, m3 ; tab l imim + movsldup m6, m2 ; tab h rere + movsldup m7, m3 ; tab l rere + + shufps m2, m0, m0, q2301 ; in h imre + shufps m3, m1, m1, q2301 ; in l imre + + mulps m6, m0 + mulps m7, m1 + + fmaddsubps m4, m4, m2, m6 + fmaddsubps m5, m5, m3, m7 + + vpermpd m3, m5, q0123 ; flip + vpermpd m2, m4, q0123 ; flip + + blendps m1, m2, m5, 01010101b + blendps m0, m3, m4, 01010101b + + movups [outq + t2q], m0 + movups [outq + t1q], m1 + + add t1q, mmsize + sub t2q, mmsize + sub lenq, mmsize/2 + jg .post + + RET +%endmacro + +%if ARCH_X86_64 && HAVE_AVX2_EXTERNAL +IMDCT_FN avx2 +%endif + +%macro PFA_15_FN 2 +INIT_YMM %1 +%if %2 +cglobal fft_pfa_15xM_asm_float, 0, 0, 0, ctx, out, in, stride, len, lut, buf, map, tgt, tmp, \ + tgt5, stride3, stride5, btmp +%else +cglobal fft_pfa_15xM_float, 4, 14, 16, 320, ctx, out, in, stride, len, lut, buf, map, tgt, tmp, \ + tgt5, stride3, stride5, btmp +%endif + +%if %2 + PUSH inq + PUSH tgt5q + PUSH stride3q + PUSH stride5q + PUSH btmpq +%endif + + PUSH strideq + + mov btmpq, outq + + mov outq, [ctxq + AVTXContext.tmp] +%if %2 == 0 + movsxd lenq, dword [ctxq + AVTXContext.len] + mov lutq, [ctxq + AVTXContext.map] +%endif + + ; Load stride (second transform's length) and second transform's LUT + mov tmpq, [ctxq + AVTXContext.sub] + movsxd strideq, dword [tmpq + AVTXContext.len] + mov mapq, [tmpq + AVTXContext.map] + + shl strideq, 3 + imul stride3q, strideq, 3 + imul stride5q, strideq, 5 + + movaps m11, [mask_mmppmmmm] ; mmppmmmm + movaps m10, [tab_53_float] ; tab5 + movaps xm9, [tab_53_float + 32] ; tab3 + vpermpd m9, m9, q1110 ; tab[23232323] + movaps m8, [s15_perm] + +.dim1: + mov tmpd, [mapq] + lea tgtq, [outq + tmpq*8] + +%if %2 + movups xm0, [inq] ; in[0,1].reim + movddup xm5, [inq + 16] ; in[2].reimreim + movups m2, [inq + mmsize*0 + 24] ; in[3-6].reim + movups m3, [inq + mmsize*1 + 24] ; in[7-11].reim + movups m4, [inq + mmsize*2 + 24] ; in[12-15].reim +%else + LOAD64_LUT xm0, inq, lutq, 0, tmpq, m14, xm15 ; in[0,1].reim + LOAD64_LUT m2, inq, lutq, (mmsize/2)*0 + 12, tmpq, m6, m7 + LOAD64_LUT m3, inq, lutq, (mmsize/2)*1 + 12, tmpq, m14, m15 + LOAD64_LUT m4, inq, lutq, (mmsize/2)*2 + 12, tmpq, m6, m7 + mov tmpd, [lutq + 8] + movddup xm5, [inq + tmpq*8] ; in[2].reimreim +%endif + + FFT15 + + lea tgt5q, [tgtq + stride5q] + lea tmpq, [tgtq + stride5q*2] + + movhps [tgtq], xm14 ; out[0] + movhps [tgtq + stride5q*1], xm15 ; out[5] + movlps [tgtq + stride5q*2], xm15 ; out[10] + + vextractf128 xm3, m0, 1 + vextractf128 xm4, m1, 1 + vextractf128 xm5, m2, 1 + + movlps [tgtq + strideq*1], xm1 + movhps [tgtq + strideq*2], xm2 + movlps [tgtq + stride3q*1], xm3 + movhps [tgtq + strideq*4], xm4 + movlps [tgtq + stride3q*2], xm0 + movlps [tgtq + strideq*8], xm5 + movhps [tgtq + stride3q*4], xm0 + movhps [tgt5q + strideq*2], xm1 + movhps [tgt5q + strideq*4], xm3 + movlps [tmpq + strideq*1], xm2 + movlps [tmpq + stride3q*1], xm4 + movhps [tmpq + strideq*4], xm5 + +%if %2 + add inq, mmsize*3 + 24 +%else + add lutq, (mmsize/2)*3 + 12 +%endif + add mapq, 4 + sub lenq, 15 + jg .dim1 + + ; Second transform setup + mov stride5q, ctxq ; backup original context + movsxd stride3q, dword [ctxq + AVTXContext.len] ; full length + mov tgt5q, [ctxq + AVTXContext.fn] ; subtransform's jump point + + mov inq, outq ; in-place transform + mov ctxq, [ctxq + AVTXContext.sub] ; load subtransform's context + mov lutq, [ctxq + AVTXContext.map] ; load subtransform's map + movsxd lenq, dword [ctxq + AVTXContext.len] ; load subtransform's length + +.dim2: + call tgt5q ; call the FFT + lea inq, [inq + lenq*8] + lea outq, [outq + lenq*8] + sub stride3q, lenq + jg .dim2 + + mov ctxq, stride5q ; restore original context + mov lutq, [ctxq + AVTXContext.map] + mov inq, [ctxq + AVTXContext.tmp] + movsxd lenq, dword [ctxq + AVTXContext.len] ; full length + + lea stride3q, [lutq + lenq*4] ; second part of the LUT + mov stride5q, lenq + mov tgt5q, btmpq + POP strideq + lea tmpq, [strideq + 2*strideq] + +.post: + LOAD64_LUT m0, inq, stride3q, 0, tmpq, m8, m9 + vextractf128 xm1, m0, 1 + movlps [tgt5q], xm0 + movhps [tgt5q + strideq], xm0 + movlps [tgt5q + strideq*2], xm1 + movhps [tgt5q + tmpq], xm1 + + lea tgt5q, [tgt5q + 4*strideq] + add stride3q, mmsize/2 + sub stride5q, mmsize/8 + jg .post + +%if %2 + mov outq, btmpq + POP btmpq + POP stride5q + POP stride3q + POP tgt5q + POP inq + ret +%else + RET +%endif + +%if %2 +cglobal fft_pfa_15xM_ns_float, 4, 14, 16, 320, ctx, out, in, stride, len, lut, buf, map, tgt, tmp, \ + tgt5, stride3, stride5, btmp + movsxd lenq, dword [ctxq + AVTXContext.len] + mov lutq, [ctxq + AVTXContext.map] + + call mangle(ff_tx_fft_pfa_15xM_asm_float) + RET +%endif +%endmacro + +%if ARCH_X86_64 && HAVE_AVX2_EXTERNAL +PFA_15_FN avx2, 0 +PFA_15_FN avx2, 1 +%endif diff --git a/media/ffvpx/libavutil/x86/tx_float_init.c b/media/ffvpx/libavutil/x86/tx_float_init.c new file mode 100644 index 0000000000..d3c0beb50f --- /dev/null +++ b/media/ffvpx/libavutil/x86/tx_float_init.c @@ -0,0 +1,310 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#define TX_FLOAT +#include "libavutil/tx_priv.h" +#include "libavutil/attributes.h" +#include "libavutil/x86/cpu.h" + +#include "config.h" + +TX_DECL_FN(fft2, sse3) +TX_DECL_FN(fft4_fwd, sse2) +TX_DECL_FN(fft4_inv, sse2) +TX_DECL_FN(fft8, sse3) +TX_DECL_FN(fft8_ns, sse3) +TX_DECL_FN(fft8, avx) +TX_DECL_FN(fft8_ns, avx) +TX_DECL_FN(fft15, avx2) +TX_DECL_FN(fft15_ns, avx2) +TX_DECL_FN(fft16, avx) +TX_DECL_FN(fft16_ns, avx) +TX_DECL_FN(fft16, fma3) +TX_DECL_FN(fft16_ns, fma3) +TX_DECL_FN(fft32, avx) +TX_DECL_FN(fft32_ns, avx) +TX_DECL_FN(fft32, fma3) +TX_DECL_FN(fft32_ns, fma3) +TX_DECL_FN(fft_sr, avx) +TX_DECL_FN(fft_sr_ns, avx) +TX_DECL_FN(fft_sr, fma3) +TX_DECL_FN(fft_sr_ns, fma3) +TX_DECL_FN(fft_sr, avx2) +TX_DECL_FN(fft_sr_ns, avx2) + +TX_DECL_FN(fft_pfa_15xM, avx2) +TX_DECL_FN(fft_pfa_15xM_ns, avx2) + +TX_DECL_FN(mdct_inv, avx2) + +TX_DECL_FN(fft2_asm, sse3) +TX_DECL_FN(fft4_fwd_asm, sse2) +TX_DECL_FN(fft4_inv_asm, sse2) +TX_DECL_FN(fft8_asm, sse3) +TX_DECL_FN(fft8_asm, avx) +TX_DECL_FN(fft16_asm, avx) +TX_DECL_FN(fft16_asm, fma3) +TX_DECL_FN(fft32_asm, avx) +TX_DECL_FN(fft32_asm, fma3) +TX_DECL_FN(fft_sr_asm, avx) +TX_DECL_FN(fft_sr_asm, fma3) +TX_DECL_FN(fft_sr_asm, avx2) + +TX_DECL_FN(fft_pfa_15xM_asm, avx2) + +#define DECL_INIT_FN(basis, interleave) \ +static av_cold int b ##basis## _i ##interleave(AVTXContext *s, \ + const FFTXCodelet *cd, \ + uint64_t flags, \ + FFTXCodeletOptions *opts, \ + int len, int inv, \ + const void *scale) \ +{ \ + ff_tx_init_tabs_float(len); \ + if (cd->max_len == 2) \ + return ff_tx_gen_ptwo_revtab(s, opts); \ + else \ + return ff_tx_gen_split_radix_parity_revtab(s, len, inv, opts, \ + basis, interleave); \ +} + +DECL_INIT_FN(8, 0) +DECL_INIT_FN(8, 2) + +static av_cold int factor_init(AVTXContext *s, const FFTXCodelet *cd, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale) +{ + int ret; + + /* The transformations below are performed in the gather domain, + * so override the option and let the infrastructure convert the map + * to SCATTER if needed. */ + FFTXCodeletOptions sub_opts = { .map_dir = FF_TX_MAP_GATHER }; + + TX_TAB(ff_tx_init_tabs)(len); + + if (len == 15) + ret = ff_tx_gen_pfa_input_map(s, &sub_opts, 3, 5); + else + ret = ff_tx_gen_default_map(s, &sub_opts); + + if (ret < 0) + return ret; + + if (len == 15) { + int cnt = 0, tmp[15]; + + /* Special permutation to simplify loads in the pre-permuted version */ + memcpy(tmp, s->map, 15*sizeof(*tmp)); + for (int i = 1; i < 15; i += 3) { + s->map[cnt] = tmp[i]; + cnt++; + } + for (int i = 2; i < 15; i += 3) { + s->map[cnt] = tmp[i]; + cnt++; + } + for (int i = 0; i < 15; i += 3) { + s->map[cnt] = tmp[i]; + cnt++; + } + memmove(&s->map[7], &s->map[6], 4*sizeof(int)); + memmove(&s->map[3], &s->map[1], 4*sizeof(int)); + s->map[1] = tmp[2]; + s->map[2] = tmp[0]; + } + + return 0; +} + +static av_cold int m_inv_init(AVTXContext *s, const FFTXCodelet *cd, + uint64_t flags, FFTXCodeletOptions *opts, + int len, int inv, const void *scale) +{ + int ret; + FFTXCodeletOptions sub_opts = { .map_dir = FF_TX_MAP_GATHER }; + + s->scale_d = *((SCALE_TYPE *)scale); + s->scale_f = s->scale_d; + + flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */ + flags |= AV_TX_INPLACE; /* in-place */ + flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */ + flags |= FF_TX_ASM_CALL; /* We want an assembly function, not C */ + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, len >> 1, + inv, scale))) + return ret; + + s->map = av_malloc(len*sizeof(*s->map)); + if (!s->map) + return AVERROR(ENOMEM); + + memcpy(s->map, s->sub->map, (len >> 1)*sizeof(*s->map)); + /* Invert lookup table for unstrided path */ + for (int i = 0; i < (len >> 1); i++) + s->map[(len >> 1) + s->map[i]] = i; + + if ((ret = ff_tx_mdct_gen_exp_float(s, s->map))) + return ret; + + return 0; +} + +static av_cold int fft_pfa_init(AVTXContext *s, + const FFTXCodelet *cd, + uint64_t flags, + FFTXCodeletOptions *opts, + int len, int inv, + const void *scale) +{ + int ret; + int sub_len = len / cd->factors[0]; + FFTXCodeletOptions sub_opts = { .map_dir = FF_TX_MAP_SCATTER }; + + flags &= ~FF_TX_OUT_OF_PLACE; /* We want the subtransform to be */ + flags |= AV_TX_INPLACE; /* in-place */ + flags |= FF_TX_PRESHUFFLE; /* This function handles the permute step */ + flags |= FF_TX_ASM_CALL; /* We want an assembly function, not C */ + + if ((ret = ff_tx_init_subtx(s, TX_TYPE(FFT), flags, &sub_opts, + sub_len, inv, scale))) + return ret; + + if ((ret = ff_tx_gen_compound_mapping(s, opts, s->inv, cd->factors[0], sub_len))) + return ret; + + if (cd->factors[0] == 15) { + int tmp[15]; + + /* Our 15-point transform is also a compound one, so embed its input map */ + TX_EMBED_INPUT_PFA_MAP(s->map, len, 3, 5); + + /* Special permutation to simplify loads in the pre-permuted version */ + for (int k = 0; k < s->sub[0].len; k++) { + int cnt = 0; + memcpy(tmp, &s->map[k*15], 15*sizeof(*tmp)); + for (int i = 1; i < 15; i += 3) { + s->map[k*15 + cnt] = tmp[i]; + cnt++; + } + for (int i = 2; i < 15; i += 3) { + s->map[k*15 + cnt] = tmp[i]; + cnt++; + } + for (int i = 0; i < 15; i += 3) { + s->map[k*15 + cnt] = tmp[i]; + cnt++; + } + memmove(&s->map[k*15 + 7], &s->map[k*15 + 6], 4*sizeof(int)); + memmove(&s->map[k*15 + 3], &s->map[k*15 + 1], 4*sizeof(int)); + s->map[k*15 + 1] = tmp[2]; + s->map[k*15 + 2] = tmp[0]; + } + } + + if (!(s->tmp = av_malloc(len*sizeof(*s->tmp)))) + return AVERROR(ENOMEM); + + TX_TAB(ff_tx_init_tabs)(len / sub_len); + + return 0; +} + +const FFTXCodelet * const ff_tx_codelet_list_float_x86[] = { + TX_DEF(fft2, FFT, 2, 2, 2, 0, 128, NULL, sse3, SSE3, AV_TX_INPLACE, 0), + TX_DEF(fft2_asm, FFT, 2, 2, 2, 0, 192, b8_i0, sse3, SSE3, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, 0), + TX_DEF(fft2, FFT, 2, 2, 2, 0, 192, b8_i0, sse3, SSE3, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft4_fwd, FFT, 4, 4, 2, 0, 128, NULL, sse2, SSE2, AV_TX_INPLACE | FF_TX_FORWARD_ONLY, 0), + TX_DEF(fft4_fwd_asm, FFT, 4, 4, 2, 0, 192, b8_i0, sse2, SSE2, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, 0), + TX_DEF(fft4_inv_asm, FFT, 4, 4, 2, 0, 128, NULL, sse2, SSE2, + AV_TX_INPLACE | FF_TX_INVERSE_ONLY | FF_TX_ASM_CALL, 0), + TX_DEF(fft4_fwd, FFT, 4, 4, 2, 0, 192, b8_i0, sse2, SSE2, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft4_inv, FFT, 4, 4, 2, 0, 128, NULL, sse2, SSE2, AV_TX_INPLACE | FF_TX_INVERSE_ONLY, 0), + TX_DEF(fft8, FFT, 8, 8, 2, 0, 128, b8_i0, sse3, SSE3, AV_TX_INPLACE, 0), + TX_DEF(fft8_asm, FFT, 8, 8, 2, 0, 192, b8_i0, sse3, SSE3, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, 0), + TX_DEF(fft8_ns, FFT, 8, 8, 2, 0, 192, b8_i0, sse3, SSE3, AV_TX_INPLACE | FF_TX_PRESHUFFLE, 0), + TX_DEF(fft8, FFT, 8, 8, 2, 0, 256, b8_i0, avx, AVX, AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft8_asm, FFT, 8, 8, 2, 0, 320, b8_i0, avx, AVX, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft8_ns, FFT, 8, 8, 2, 0, 320, b8_i0, avx, AVX, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16, FFT, 16, 16, 2, 0, 256, b8_i2, avx, AVX, AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16_asm, FFT, 16, 16, 2, 0, 320, b8_i2, avx, AVX, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16_ns, FFT, 16, 16, 2, 0, 320, b8_i2, avx, AVX, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16, FFT, 16, 16, 2, 0, 288, b8_i2, fma3, FMA3, AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16_asm, FFT, 16, 16, 2, 0, 352, b8_i2, fma3, FMA3, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft16_ns, FFT, 16, 16, 2, 0, 352, b8_i2, fma3, FMA3, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + +#if ARCH_X86_64 + TX_DEF(fft32, FFT, 32, 32, 2, 0, 256, b8_i2, avx, AVX, AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft32_asm, FFT, 32, 32, 2, 0, 320, b8_i2, avx, AVX, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft32_ns, FFT, 32, 32, 2, 0, 320, b8_i2, avx, AVX, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft32, FFT, 32, 32, 2, 0, 288, b8_i2, fma3, FMA3, AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft32_asm, FFT, 32, 32, 2, 0, 352, b8_i2, fma3, FMA3, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft32_ns, FFT, 32, 32, 2, 0, 352, b8_i2, fma3, FMA3, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr, FFT, 64, 131072, 2, 0, 256, b8_i2, avx, AVX, 0, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr_asm, FFT, 64, 131072, 2, 0, 320, b8_i2, avx, AVX, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr_ns, FFT, 64, 131072, 2, 0, 320, b8_i2, avx, AVX, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr, FFT, 64, 131072, 2, 0, 288, b8_i2, fma3, FMA3, 0, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr_asm, FFT, 64, 131072, 2, 0, 352, b8_i2, fma3, FMA3, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft_sr_ns, FFT, 64, 131072, 2, 0, 352, b8_i2, fma3, FMA3, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW), + +#if HAVE_AVX2_EXTERNAL + TX_DEF(fft15, FFT, 15, 15, 15, 0, 320, factor_init, avx2, AVX2, + AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW), + TX_DEF(fft15_ns, FFT, 15, 15, 15, 0, 384, factor_init, avx2, AVX2, + AV_TX_INPLACE | FF_TX_PRESHUFFLE, AV_CPU_FLAG_AVXSLOW), + + TX_DEF(fft_sr, FFT, 64, 131072, 2, 0, 320, b8_i2, avx2, AVX2, 0, + AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + TX_DEF(fft_sr_asm, FFT, 64, 131072, 2, 0, 384, b8_i2, avx2, AVX2, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + TX_DEF(fft_sr_ns, FFT, 64, 131072, 2, 0, 384, b8_i2, avx2, AVX2, AV_TX_INPLACE | FF_TX_PRESHUFFLE, + AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + + TX_DEF(fft_pfa_15xM, FFT, 60, TX_LEN_UNLIMITED, 15, 2, 320, fft_pfa_init, avx2, AVX2, + AV_TX_INPLACE, AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + TX_DEF(fft_pfa_15xM_asm, FFT, 60, TX_LEN_UNLIMITED, 15, 2, 384, fft_pfa_init, avx2, AVX2, + AV_TX_INPLACE | FF_TX_PRESHUFFLE | FF_TX_ASM_CALL, AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + TX_DEF(fft_pfa_15xM_ns, FFT, 60, TX_LEN_UNLIMITED, 15, 2, 384, fft_pfa_init, avx2, AVX2, + AV_TX_INPLACE | FF_TX_PRESHUFFLE, AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), + + TX_DEF(mdct_inv, MDCT, 16, TX_LEN_UNLIMITED, 2, TX_FACTOR_ANY, 384, m_inv_init, avx2, AVX2, + FF_TX_INVERSE_ONLY, AV_CPU_FLAG_AVXSLOW | AV_CPU_FLAG_SLOW_GATHER), +#endif +#endif + + NULL, +}; diff --git a/media/ffvpx/libavutil/x86/x86inc.asm b/media/ffvpx/libavutil/x86/x86inc.asm new file mode 100644 index 0000000000..e099ee4b10 --- /dev/null +++ b/media/ffvpx/libavutil/x86/x86inc.asm @@ -0,0 +1,1726 @@ +;***************************************************************************** +;* x86inc.asm: x264asm abstraction layer +;***************************************************************************** +;* Copyright (C) 2005-2018 x264 project +;* +;* Authors: Loren Merritt <lorenm@u.washington.edu> +;* Henrik Gramner <henrik@gramner.com> +;* Anton Mitrofanov <BugMaster@narod.ru> +;* Fiona Glaser <fiona@x264.com> +;* +;* Permission to use, copy, modify, and/or distribute this software for any +;* purpose with or without fee is hereby granted, provided that the above +;* copyright notice and this permission notice appear in all copies. +;* +;* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES +;* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF +;* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR +;* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES +;* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN +;* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF +;* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. +;***************************************************************************** + +; This is a header file for the x264ASM assembly language, which uses +; NASM/YASM syntax combined with a large number of macros to provide easy +; abstraction between different calling conventions (x86_32, win64, linux64). +; It also has various other useful features to simplify writing the kind of +; DSP functions that are most often used in x264. + +; Unlike the rest of x264, this file is available under an ISC license, as it +; has significant usefulness outside of x264 and we want it to be available +; to the largest audience possible. Of course, if you modify it for your own +; purposes to add a new feature, we strongly encourage contributing a patch +; as this feature might be useful for others as well. Send patches or ideas +; to x264-devel@videolan.org . + +%ifndef private_prefix + %define private_prefix x264 +%endif + +%ifndef public_prefix + %define public_prefix private_prefix +%endif + +%if HAVE_ALIGNED_STACK + %define STACK_ALIGNMENT 16 +%endif +%ifndef STACK_ALIGNMENT + %if ARCH_X86_64 + %define STACK_ALIGNMENT 16 + %else + %define STACK_ALIGNMENT 4 + %endif +%endif + +%define WIN64 0 +%define UNIX64 0 +%if ARCH_X86_64 + %ifidn __OUTPUT_FORMAT__,win32 + %define WIN64 1 + %elifidn __OUTPUT_FORMAT__,win64 + %define WIN64 1 + %elifidn __OUTPUT_FORMAT__,x64 + %define WIN64 1 + %else + %define UNIX64 1 + %endif +%endif + +%define FORMAT_ELF 0 +%ifidn __OUTPUT_FORMAT__,elf + %define FORMAT_ELF 1 +%elifidn __OUTPUT_FORMAT__,elf32 + %define FORMAT_ELF 1 +%elifidn __OUTPUT_FORMAT__,elf64 + %define FORMAT_ELF 1 +%endif + +%ifdef PREFIX + %define mangle(x) _ %+ x +%else + %define mangle(x) x +%endif + +; aout does not support align= +; NOTE: This section is out of sync with x264, in order to +; keep supporting OS/2. +%macro SECTION_RODATA 0-1 16 + %ifidn __OUTPUT_FORMAT__,aout + SECTION .text + %elifidn __OUTPUT_FORMAT__,coff + SECTION .text + %elifidn __OUTPUT_FORMAT__,win32 + SECTION .rdata align=%1 + %elif WIN64 + SECTION .rdata align=%1 + %else + SECTION .rodata align=%1 + %endif +%endmacro + +%if WIN64 + %define PIC +%elif ARCH_X86_64 == 0 +; x86_32 doesn't require PIC. +; Some distros prefer shared objects to be PIC, but nothing breaks if +; the code contains a few textrels, so we'll skip that complexity. + %undef PIC +%endif +%ifdef PIC + default rel +%endif + +%macro CPUNOP 1 + %if HAVE_CPUNOP + CPU %1 + %endif +%endmacro + +; Macros to eliminate most code duplication between x86_32 and x86_64: +; Currently this works only for leaf functions which load all their arguments +; into registers at the start, and make no other use of the stack. Luckily that +; covers most of x264's asm. + +; PROLOGUE: +; %1 = number of arguments. loads them from stack if needed. +; %2 = number of registers used. pushes callee-saved regs if needed. +; %3 = number of xmm registers used. pushes callee-saved xmm regs if needed. +; %4 = (optional) stack size to be allocated. The stack will be aligned before +; allocating the specified stack size. If the required stack alignment is +; larger than the known stack alignment the stack will be manually aligned +; and an extra register will be allocated to hold the original stack +; pointer (to not invalidate r0m etc.). To prevent the use of an extra +; register as stack pointer, request a negative stack size. +; %4+/%5+ = list of names to define to registers +; PROLOGUE can also be invoked by adding the same options to cglobal + +; e.g. +; cglobal foo, 2,3,7,0x40, dst, src, tmp +; declares a function (foo) that automatically loads two arguments (dst and +; src) into registers, uses one additional register (tmp) plus 7 vector +; registers (m0-m6) and allocates 0x40 bytes of stack space. + +; TODO Some functions can use some args directly from the stack. If they're the +; last args then you can just not declare them, but if they're in the middle +; we need more flexible macro. + +; RET: +; Pops anything that was pushed by PROLOGUE, and returns. + +; REP_RET: +; Use this instead of RET if it's a branch target. + +; registers: +; rN and rNq are the native-size register holding function argument N +; rNd, rNw, rNb are dword, word, and byte size +; rNh is the high 8 bits of the word size +; rNm is the original location of arg N (a register or on the stack), dword +; rNmp is native size + +%macro DECLARE_REG 2-3 + %define r%1q %2 + %define r%1d %2d + %define r%1w %2w + %define r%1b %2b + %define r%1h %2h + %define %2q %2 + %if %0 == 2 + %define r%1m %2d + %define r%1mp %2 + %elif ARCH_X86_64 ; memory + %define r%1m [rstk + stack_offset + %3] + %define r%1mp qword r %+ %1 %+ m + %else + %define r%1m [rstk + stack_offset + %3] + %define r%1mp dword r %+ %1 %+ m + %endif + %define r%1 %2 +%endmacro + +%macro DECLARE_REG_SIZE 3 + %define r%1q r%1 + %define e%1q r%1 + %define r%1d e%1 + %define e%1d e%1 + %define r%1w %1 + %define e%1w %1 + %define r%1h %3 + %define e%1h %3 + %define r%1b %2 + %define e%1b %2 + %if ARCH_X86_64 == 0 + %define r%1 e%1 + %endif +%endmacro + +DECLARE_REG_SIZE ax, al, ah +DECLARE_REG_SIZE bx, bl, bh +DECLARE_REG_SIZE cx, cl, ch +DECLARE_REG_SIZE dx, dl, dh +DECLARE_REG_SIZE si, sil, null +DECLARE_REG_SIZE di, dil, null +DECLARE_REG_SIZE bp, bpl, null + +; t# defines for when per-arch register allocation is more complex than just function arguments + +%macro DECLARE_REG_TMP 1-* + %assign %%i 0 + %rep %0 + CAT_XDEFINE t, %%i, r%1 + %assign %%i %%i+1 + %rotate 1 + %endrep +%endmacro + +%macro DECLARE_REG_TMP_SIZE 0-* + %rep %0 + %define t%1q t%1 %+ q + %define t%1d t%1 %+ d + %define t%1w t%1 %+ w + %define t%1h t%1 %+ h + %define t%1b t%1 %+ b + %rotate 1 + %endrep +%endmacro + +DECLARE_REG_TMP_SIZE 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14 + +%if ARCH_X86_64 + %define gprsize 8 +%else + %define gprsize 4 +%endif + +; Repeats an instruction/operation for multiple arguments. +; Example usage: "REPX {psrlw x, 8}, m0, m1, m2, m3" +%macro REPX 2-* ; operation, args + %xdefine %%f(x) %1 + %rep %0 - 1 + %rotate 1 + %%f(%1) + %endrep +%endmacro + +%macro PUSH 1 + push %1 + %ifidn rstk, rsp + %assign stack_offset stack_offset+gprsize + %endif +%endmacro + +%macro POP 1 + pop %1 + %ifidn rstk, rsp + %assign stack_offset stack_offset-gprsize + %endif +%endmacro + +%macro PUSH_IF_USED 1-* + %rep %0 + %if %1 < regs_used + PUSH r%1 + %endif + %rotate 1 + %endrep +%endmacro + +%macro POP_IF_USED 1-* + %rep %0 + %if %1 < regs_used + pop r%1 + %endif + %rotate 1 + %endrep +%endmacro + +%macro LOAD_IF_USED 1-* + %rep %0 + %if %1 < num_args + mov r%1, r %+ %1 %+ mp + %endif + %rotate 1 + %endrep +%endmacro + +%macro SUB 2 + sub %1, %2 + %ifidn %1, rstk + %assign stack_offset stack_offset+(%2) + %endif +%endmacro + +%macro ADD 2 + add %1, %2 + %ifidn %1, rstk + %assign stack_offset stack_offset-(%2) + %endif +%endmacro + +%macro movifnidn 2 + %ifnidn %1, %2 + mov %1, %2 + %endif +%endmacro + +%macro movsxdifnidn 2 + %ifnidn %1, %2 + movsxd %1, %2 + %endif +%endmacro + +%macro ASSERT 1 + %if (%1) == 0 + %error assertion ``%1'' failed + %endif +%endmacro + +%macro DEFINE_ARGS 0-* + %ifdef n_arg_names + %assign %%i 0 + %rep n_arg_names + CAT_UNDEF arg_name %+ %%i, q + CAT_UNDEF arg_name %+ %%i, d + CAT_UNDEF arg_name %+ %%i, w + CAT_UNDEF arg_name %+ %%i, h + CAT_UNDEF arg_name %+ %%i, b + CAT_UNDEF arg_name %+ %%i, m + CAT_UNDEF arg_name %+ %%i, mp + CAT_UNDEF arg_name, %%i + %assign %%i %%i+1 + %endrep + %endif + + %xdefine %%stack_offset stack_offset + %undef stack_offset ; so that the current value of stack_offset doesn't get baked in by xdefine + %assign %%i 0 + %rep %0 + %xdefine %1q r %+ %%i %+ q + %xdefine %1d r %+ %%i %+ d + %xdefine %1w r %+ %%i %+ w + %xdefine %1h r %+ %%i %+ h + %xdefine %1b r %+ %%i %+ b + %xdefine %1m r %+ %%i %+ m + %xdefine %1mp r %+ %%i %+ mp + CAT_XDEFINE arg_name, %%i, %1 + %assign %%i %%i+1 + %rotate 1 + %endrep + %xdefine stack_offset %%stack_offset + %assign n_arg_names %0 +%endmacro + +%define required_stack_alignment ((mmsize + 15) & ~15) +%define vzeroupper_required (mmsize > 16 && (ARCH_X86_64 == 0 || xmm_regs_used > 16 || notcpuflag(avx512))) +%define high_mm_regs (16*cpuflag(avx512)) + +%macro ALLOC_STACK 1-2 0 ; stack_size, n_xmm_regs (for win64 only) + %ifnum %1 + %if %1 != 0 + %assign %%pad 0 + %assign stack_size %1 + %if stack_size < 0 + %assign stack_size -stack_size + %endif + %if WIN64 + %assign %%pad %%pad + 32 ; shadow space + %if mmsize != 8 + %assign xmm_regs_used %2 + %if xmm_regs_used > 8 + %assign %%pad %%pad + (xmm_regs_used-8)*16 ; callee-saved xmm registers + %endif + %endif + %endif + %if required_stack_alignment <= STACK_ALIGNMENT + ; maintain the current stack alignment + %assign stack_size_padded stack_size + %%pad + ((-%%pad-stack_offset-gprsize) & (STACK_ALIGNMENT-1)) + SUB rsp, stack_size_padded + %else + %assign %%reg_num (regs_used - 1) + %xdefine rstk r %+ %%reg_num + ; align stack, and save original stack location directly above + ; it, i.e. in [rsp+stack_size_padded], so we can restore the + ; stack in a single instruction (i.e. mov rsp, rstk or mov + ; rsp, [rsp+stack_size_padded]) + %if %1 < 0 ; need to store rsp on stack + %xdefine rstkm [rsp + stack_size + %%pad] + %assign %%pad %%pad + gprsize + %else ; can keep rsp in rstk during whole function + %xdefine rstkm rstk + %endif + %assign stack_size_padded stack_size + ((%%pad + required_stack_alignment-1) & ~(required_stack_alignment-1)) + mov rstk, rsp + and rsp, ~(required_stack_alignment-1) + sub rsp, stack_size_padded + movifnidn rstkm, rstk + %endif + WIN64_PUSH_XMM + %endif + %endif +%endmacro + +%macro SETUP_STACK_POINTER 1 + %ifnum %1 + %if %1 != 0 && required_stack_alignment > STACK_ALIGNMENT + %if %1 > 0 + ; Reserve an additional register for storing the original stack pointer, but avoid using + ; eax/rax for this purpose since it can potentially get overwritten as a return value. + %assign regs_used (regs_used + 1) + %if ARCH_X86_64 && regs_used == 7 + %assign regs_used 8 + %elif ARCH_X86_64 == 0 && regs_used == 1 + %assign regs_used 2 + %endif + %endif + %if ARCH_X86_64 && regs_used < 5 + UNIX64 * 3 + ; Ensure that we don't clobber any registers containing arguments. For UNIX64 we also preserve r6 (rax) + ; since it's used as a hidden argument in vararg functions to specify the number of vector registers used. + %assign regs_used 5 + UNIX64 * 3 + %endif + %endif + %endif +%endmacro + +%if WIN64 ; Windows x64 ;================================================= + +DECLARE_REG 0, rcx +DECLARE_REG 1, rdx +DECLARE_REG 2, R8 +DECLARE_REG 3, R9 +DECLARE_REG 4, R10, 40 +DECLARE_REG 5, R11, 48 +DECLARE_REG 6, rax, 56 +DECLARE_REG 7, rdi, 64 +DECLARE_REG 8, rsi, 72 +DECLARE_REG 9, rbx, 80 +DECLARE_REG 10, rbp, 88 +DECLARE_REG 11, R14, 96 +DECLARE_REG 12, R15, 104 +DECLARE_REG 13, R12, 112 +DECLARE_REG 14, R13, 120 + +%macro PROLOGUE 2-5+ 0, 0 ; #args, #regs, #xmm_regs, [stack_size,] arg_names... + %assign num_args %1 + %assign regs_used %2 + ASSERT regs_used >= num_args + SETUP_STACK_POINTER %4 + ASSERT regs_used <= 15 + PUSH_IF_USED 7, 8, 9, 10, 11, 12, 13, 14 + ALLOC_STACK %4, %3 + %if mmsize != 8 && stack_size == 0 + WIN64_SPILL_XMM %3 + %endif + LOAD_IF_USED 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 + %if %0 > 4 + %ifnum %4 + DEFINE_ARGS %5 + %else + DEFINE_ARGS %4, %5 + %endif + %elifnnum %4 + DEFINE_ARGS %4 + %endif +%endmacro + +%macro WIN64_PUSH_XMM 0 + ; Use the shadow space to store XMM6 and XMM7, the rest needs stack space allocated. + %if xmm_regs_used > 6 + high_mm_regs + movaps [rstk + stack_offset + 8], xmm6 + %endif + %if xmm_regs_used > 7 + high_mm_regs + movaps [rstk + stack_offset + 24], xmm7 + %endif + %assign %%xmm_regs_on_stack xmm_regs_used - high_mm_regs - 8 + %if %%xmm_regs_on_stack > 0 + %assign %%i 8 + %rep %%xmm_regs_on_stack + movaps [rsp + (%%i-8)*16 + stack_size + 32], xmm %+ %%i + %assign %%i %%i+1 + %endrep + %endif +%endmacro + +%macro WIN64_SPILL_XMM 1 + %assign xmm_regs_used %1 + ASSERT xmm_regs_used <= 16 + high_mm_regs + %assign %%xmm_regs_on_stack xmm_regs_used - high_mm_regs - 8 + %if %%xmm_regs_on_stack > 0 + ; Allocate stack space for callee-saved xmm registers plus shadow space and align the stack. + %assign %%pad %%xmm_regs_on_stack*16 + 32 + %assign stack_size_padded %%pad + ((-%%pad-stack_offset-gprsize) & (STACK_ALIGNMENT-1)) + SUB rsp, stack_size_padded + %endif + WIN64_PUSH_XMM +%endmacro + +%macro WIN64_RESTORE_XMM_INTERNAL 0 + %assign %%pad_size 0 + %assign %%xmm_regs_on_stack xmm_regs_used - high_mm_regs - 8 + %if %%xmm_regs_on_stack > 0 + %assign %%i xmm_regs_used - high_mm_regs + %rep %%xmm_regs_on_stack + %assign %%i %%i-1 + movaps xmm %+ %%i, [rsp + (%%i-8)*16 + stack_size + 32] + %endrep + %endif + %if stack_size_padded > 0 + %if stack_size > 0 && required_stack_alignment > STACK_ALIGNMENT + mov rsp, rstkm + %else + add rsp, stack_size_padded + %assign %%pad_size stack_size_padded + %endif + %endif + %if xmm_regs_used > 7 + high_mm_regs + movaps xmm7, [rsp + stack_offset - %%pad_size + 24] + %endif + %if xmm_regs_used > 6 + high_mm_regs + movaps xmm6, [rsp + stack_offset - %%pad_size + 8] + %endif +%endmacro + +%macro WIN64_RESTORE_XMM 0 + WIN64_RESTORE_XMM_INTERNAL + %assign stack_offset (stack_offset-stack_size_padded) + %assign stack_size_padded 0 + %assign xmm_regs_used 0 +%endmacro + +%define has_epilogue regs_used > 7 || stack_size > 0 || vzeroupper_required || xmm_regs_used > 6+high_mm_regs + +%macro RET 0 + WIN64_RESTORE_XMM_INTERNAL + POP_IF_USED 14, 13, 12, 11, 10, 9, 8, 7 + %if vzeroupper_required + vzeroupper + %endif + AUTO_REP_RET +%endmacro + +%elif ARCH_X86_64 ; *nix x64 ;============================================= + +DECLARE_REG 0, rdi +DECLARE_REG 1, rsi +DECLARE_REG 2, rdx +DECLARE_REG 3, rcx +DECLARE_REG 4, R8 +DECLARE_REG 5, R9 +DECLARE_REG 6, rax, 8 +DECLARE_REG 7, R10, 16 +DECLARE_REG 8, R11, 24 +DECLARE_REG 9, rbx, 32 +DECLARE_REG 10, rbp, 40 +DECLARE_REG 11, R14, 48 +DECLARE_REG 12, R15, 56 +DECLARE_REG 13, R12, 64 +DECLARE_REG 14, R13, 72 + +%macro PROLOGUE 2-5+ 0, 0 ; #args, #regs, #xmm_regs, [stack_size,] arg_names... + %assign num_args %1 + %assign regs_used %2 + %assign xmm_regs_used %3 + ASSERT regs_used >= num_args + SETUP_STACK_POINTER %4 + ASSERT regs_used <= 15 + PUSH_IF_USED 9, 10, 11, 12, 13, 14 + ALLOC_STACK %4 + LOAD_IF_USED 6, 7, 8, 9, 10, 11, 12, 13, 14 + %if %0 > 4 + %ifnum %4 + DEFINE_ARGS %5 + %else + DEFINE_ARGS %4, %5 + %endif + %elifnnum %4 + DEFINE_ARGS %4 + %endif +%endmacro + +%define has_epilogue regs_used > 9 || stack_size > 0 || vzeroupper_required + +%macro RET 0 + %if stack_size_padded > 0 + %if required_stack_alignment > STACK_ALIGNMENT + mov rsp, rstkm + %else + add rsp, stack_size_padded + %endif + %endif + POP_IF_USED 14, 13, 12, 11, 10, 9 + %if vzeroupper_required + vzeroupper + %endif + AUTO_REP_RET +%endmacro + +%else ; X86_32 ;============================================================== + +DECLARE_REG 0, eax, 4 +DECLARE_REG 1, ecx, 8 +DECLARE_REG 2, edx, 12 +DECLARE_REG 3, ebx, 16 +DECLARE_REG 4, esi, 20 +DECLARE_REG 5, edi, 24 +DECLARE_REG 6, ebp, 28 +%define rsp esp + +%macro DECLARE_ARG 1-* + %rep %0 + %define r%1m [rstk + stack_offset + 4*%1 + 4] + %define r%1mp dword r%1m + %rotate 1 + %endrep +%endmacro + +DECLARE_ARG 7, 8, 9, 10, 11, 12, 13, 14 + +%macro PROLOGUE 2-5+ 0, 0 ; #args, #regs, #xmm_regs, [stack_size,] arg_names... + %assign num_args %1 + %assign regs_used %2 + ASSERT regs_used >= num_args + %if num_args > 7 + %assign num_args 7 + %endif + %if regs_used > 7 + %assign regs_used 7 + %endif + SETUP_STACK_POINTER %4 + ASSERT regs_used <= 7 + PUSH_IF_USED 3, 4, 5, 6 + ALLOC_STACK %4 + LOAD_IF_USED 0, 1, 2, 3, 4, 5, 6 + %if %0 > 4 + %ifnum %4 + DEFINE_ARGS %5 + %else + DEFINE_ARGS %4, %5 + %endif + %elifnnum %4 + DEFINE_ARGS %4 + %endif +%endmacro + +%define has_epilogue regs_used > 3 || stack_size > 0 || vzeroupper_required + +%macro RET 0 + %if stack_size_padded > 0 + %if required_stack_alignment > STACK_ALIGNMENT + mov rsp, rstkm + %else + add rsp, stack_size_padded + %endif + %endif + POP_IF_USED 6, 5, 4, 3 + %if vzeroupper_required + vzeroupper + %endif + AUTO_REP_RET +%endmacro + +%endif ;====================================================================== + +%if WIN64 == 0 + %macro WIN64_SPILL_XMM 1 + %endmacro + %macro WIN64_RESTORE_XMM 0 + %endmacro + %macro WIN64_PUSH_XMM 0 + %endmacro +%endif + +; On AMD cpus <=K10, an ordinary ret is slow if it immediately follows either +; a branch or a branch target. So switch to a 2-byte form of ret in that case. +; We can automatically detect "follows a branch", but not a branch target. +; (SSSE3 is a sufficient condition to know that your cpu doesn't have this problem.) +%macro REP_RET 0 + %if has_epilogue || cpuflag(ssse3) + RET + %else + rep ret + %endif + annotate_function_size +%endmacro + +%define last_branch_adr $$ +%macro AUTO_REP_RET 0 + %if notcpuflag(ssse3) + times ((last_branch_adr-$)>>31)+1 rep ; times 1 iff $ == last_branch_adr. + %endif + ret + annotate_function_size +%endmacro + +%macro BRANCH_INSTR 0-* + %rep %0 + %macro %1 1-2 %1 + %2 %1 + %if notcpuflag(ssse3) + %%branch_instr equ $ + %xdefine last_branch_adr %%branch_instr + %endif + %endmacro + %rotate 1 + %endrep +%endmacro + +BRANCH_INSTR jz, je, jnz, jne, jl, jle, jnl, jnle, jg, jge, jng, jnge, ja, jae, jna, jnae, jb, jbe, jnb, jnbe, jc, jnc, js, jns, jo, jno, jp, jnp + +%macro TAIL_CALL 2 ; callee, is_nonadjacent + %if has_epilogue + call %1 + RET + %elif %2 + jmp %1 + %endif + annotate_function_size +%endmacro + +;============================================================================= +; arch-independent part +;============================================================================= + +%assign function_align 16 + +; Begin a function. +; Applies any symbol mangling needed for C linkage, and sets up a define such that +; subsequent uses of the function name automatically refer to the mangled version. +; Appends cpuflags to the function name if cpuflags has been specified. +; The "" empty default parameter is a workaround for nasm, which fails if SUFFIX +; is empty and we call cglobal_internal with just %1 %+ SUFFIX (without %2). +%macro cglobal 1-2+ "" ; name, [PROLOGUE args] + cglobal_internal 1, %1 %+ SUFFIX, %2 +%endmacro +%macro cvisible 1-2+ "" ; name, [PROLOGUE args] + cglobal_internal 0, %1 %+ SUFFIX, %2 +%endmacro +%macro cglobal_internal 2-3+ + annotate_function_size + %if %1 + %xdefine %%FUNCTION_PREFIX private_prefix + %xdefine %%VISIBILITY hidden + %else + %xdefine %%FUNCTION_PREFIX public_prefix + %xdefine %%VISIBILITY + %endif + %ifndef cglobaled_%2 + %xdefine %2 mangle(%%FUNCTION_PREFIX %+ _ %+ %2) + %xdefine %2.skip_prologue %2 %+ .skip_prologue + CAT_XDEFINE cglobaled_, %2, 1 + %endif + %xdefine current_function %2 + %xdefine current_function_section __SECT__ + %if FORMAT_ELF + global %2:function %%VISIBILITY + %else + global %2 + %endif + align function_align + %2: + RESET_MM_PERMUTATION ; needed for x86-64, also makes disassembly somewhat nicer + %xdefine rstk rsp ; copy of the original stack pointer, used when greater alignment than the known stack alignment is required + %assign stack_offset 0 ; stack pointer offset relative to the return address + %assign stack_size 0 ; amount of stack space that can be freely used inside a function + %assign stack_size_padded 0 ; total amount of allocated stack space, including space for callee-saved xmm registers on WIN64 and alignment padding + %assign xmm_regs_used 0 ; number of XMM registers requested, used for dealing with callee-saved registers on WIN64 and vzeroupper + %ifnidn %3, "" + PROLOGUE %3 + %endif +%endmacro + +; Create a global symbol from a local label with the correct name mangling and type +%macro cglobal_label 1 + %if FORMAT_ELF + global current_function %+ %1:function hidden + %else + global current_function %+ %1 + %endif + %1: +%endmacro + +%macro cextern 1 + %xdefine %1 mangle(private_prefix %+ _ %+ %1) + CAT_XDEFINE cglobaled_, %1, 1 + extern %1 +%endmacro + +; like cextern, but without the prefix +%macro cextern_naked 1 + %ifdef PREFIX + %xdefine %1 mangle(%1) + %endif + CAT_XDEFINE cglobaled_, %1, 1 + extern %1 +%endmacro + +%macro const 1-2+ + %xdefine %1 mangle(private_prefix %+ _ %+ %1) + %if FORMAT_ELF + global %1:data hidden + %else + global %1 + %endif + %1: %2 +%endmacro + +; This is needed for ELF, otherwise the GNU linker assumes the stack is executable by default. +%if FORMAT_ELF + [SECTION .note.GNU-stack noalloc noexec nowrite progbits] +%endif + +; Tell debuggers how large the function was. +; This may be invoked multiple times per function; we rely on later instances overriding earlier ones. +; This is invoked by RET and similar macros, and also cglobal does it for the previous function, +; but if the last function in a source file doesn't use any of the standard macros for its epilogue, +; then its size might be unspecified. +%macro annotate_function_size 0 + %ifdef __YASM_VER__ + %ifdef current_function + %if FORMAT_ELF + current_function_section + %%ecf equ $ + size current_function %%ecf - current_function + __SECT__ + %endif + %endif + %endif +%endmacro + +; cpuflags + +%assign cpuflags_mmx (1<<0) +%assign cpuflags_mmx2 (1<<1) | cpuflags_mmx +%assign cpuflags_3dnow (1<<2) | cpuflags_mmx +%assign cpuflags_3dnowext (1<<3) | cpuflags_3dnow +%assign cpuflags_sse (1<<4) | cpuflags_mmx2 +%assign cpuflags_sse2 (1<<5) | cpuflags_sse +%assign cpuflags_sse2slow (1<<6) | cpuflags_sse2 +%assign cpuflags_lzcnt (1<<7) | cpuflags_sse2 +%assign cpuflags_sse3 (1<<8) | cpuflags_sse2 +%assign cpuflags_ssse3 (1<<9) | cpuflags_sse3 +%assign cpuflags_sse4 (1<<10)| cpuflags_ssse3 +%assign cpuflags_sse42 (1<<11)| cpuflags_sse4 +%assign cpuflags_aesni (1<<12)| cpuflags_sse42 +%assign cpuflags_avx (1<<13)| cpuflags_sse42 +%assign cpuflags_xop (1<<14)| cpuflags_avx +%assign cpuflags_fma4 (1<<15)| cpuflags_avx +%assign cpuflags_fma3 (1<<16)| cpuflags_avx +%assign cpuflags_bmi1 (1<<17)| cpuflags_avx|cpuflags_lzcnt +%assign cpuflags_bmi2 (1<<18)| cpuflags_bmi1 +%assign cpuflags_avx2 (1<<19)| cpuflags_fma3|cpuflags_bmi2 +%assign cpuflags_avx512 (1<<20)| cpuflags_avx2 ; F, CD, BW, DQ, VL +%assign cpuflags_avx512icl (1<<25)| cpuflags_avx512 + +%assign cpuflags_cache32 (1<<21) +%assign cpuflags_cache64 (1<<22) +%assign cpuflags_aligned (1<<23) ; not a cpu feature, but a function variant +%assign cpuflags_atom (1<<24) + +; Returns a boolean value expressing whether or not the specified cpuflag is enabled. +%define cpuflag(x) (((((cpuflags & (cpuflags_ %+ x)) ^ (cpuflags_ %+ x)) - 1) >> 31) & 1) +%define notcpuflag(x) (cpuflag(x) ^ 1) + +; Takes an arbitrary number of cpuflags from the above list. +; All subsequent functions (up to the next INIT_CPUFLAGS) is built for the specified cpu. +; You shouldn't need to invoke this macro directly, it's a subroutine for INIT_MMX &co. +%macro INIT_CPUFLAGS 0-* + %xdefine SUFFIX + %undef cpuname + %assign cpuflags 0 + + %if %0 >= 1 + %rep %0 + %ifdef cpuname + %xdefine cpuname cpuname %+ _%1 + %else + %xdefine cpuname %1 + %endif + %assign cpuflags cpuflags | cpuflags_%1 + %rotate 1 + %endrep + %xdefine SUFFIX _ %+ cpuname + + %if cpuflag(avx) + %assign avx_enabled 1 + %endif + %if (mmsize == 16 && notcpuflag(sse2)) || (mmsize == 32 && notcpuflag(avx2)) + %define mova movaps + %define movu movups + %define movnta movntps + %endif + %if cpuflag(aligned) + %define movu mova + %elif cpuflag(sse3) && notcpuflag(ssse3) + %define movu lddqu + %endif + %endif + + %if ARCH_X86_64 || cpuflag(sse2) + CPUNOP amdnop + %else + CPUNOP basicnop + %endif +%endmacro + +; Merge mmx, sse*, and avx* +; m# is a simd register of the currently selected size +; xm# is the corresponding xmm register if mmsize >= 16, otherwise the same as m# +; ym# is the corresponding ymm register if mmsize >= 32, otherwise the same as m# +; zm# is the corresponding zmm register if mmsize >= 64, otherwise the same as m# +; (All 4 remain in sync through SWAP.) + +%macro CAT_XDEFINE 3 + %xdefine %1%2 %3 +%endmacro + +%macro CAT_UNDEF 2 + %undef %1%2 +%endmacro + +%macro DEFINE_MMREGS 1 ; mmtype + %assign %%prev_mmregs 0 + %ifdef num_mmregs + %assign %%prev_mmregs num_mmregs + %endif + + %assign num_mmregs 8 + %if ARCH_X86_64 && mmsize >= 16 + %assign num_mmregs 16 + %if cpuflag(avx512) || mmsize == 64 + %assign num_mmregs 32 + %endif + %endif + + %assign %%i 0 + %rep num_mmregs + CAT_XDEFINE m, %%i, %1 %+ %%i + CAT_XDEFINE nn%1, %%i, %%i + %assign %%i %%i+1 + %endrep + %if %%prev_mmregs > num_mmregs + %rep %%prev_mmregs - num_mmregs + CAT_UNDEF m, %%i + CAT_UNDEF nn %+ mmtype, %%i + %assign %%i %%i+1 + %endrep + %endif + %xdefine mmtype %1 +%endmacro + +; Prefer registers 16-31 over 0-15 to avoid having to use vzeroupper +%macro AVX512_MM_PERMUTATION 0-1 0 ; start_reg + %if ARCH_X86_64 && cpuflag(avx512) + %assign %%i %1 + %rep 16-%1 + %assign %%i_high %%i+16 + SWAP %%i, %%i_high + %assign %%i %%i+1 + %endrep + %endif +%endmacro + +%macro INIT_MMX 0-1+ + %assign avx_enabled 0 + %define RESET_MM_PERMUTATION INIT_MMX %1 + %define mmsize 8 + %define mova movq + %define movu movq + %define movh movd + %define movnta movntq + INIT_CPUFLAGS %1 + DEFINE_MMREGS mm +%endmacro + +%macro INIT_XMM 0-1+ + %assign avx_enabled 0 + %define RESET_MM_PERMUTATION INIT_XMM %1 + %define mmsize 16 + %define mova movdqa + %define movu movdqu + %define movh movq + %define movnta movntdq + INIT_CPUFLAGS %1 + DEFINE_MMREGS xmm + %if WIN64 + AVX512_MM_PERMUTATION 6 ; Swap callee-saved registers with volatile registers + %endif +%endmacro + +%macro INIT_YMM 0-1+ + %assign avx_enabled 1 + %define RESET_MM_PERMUTATION INIT_YMM %1 + %define mmsize 32 + %define mova movdqa + %define movu movdqu + %undef movh + %define movnta movntdq + INIT_CPUFLAGS %1 + DEFINE_MMREGS ymm + AVX512_MM_PERMUTATION +%endmacro + +%macro INIT_ZMM 0-1+ + %assign avx_enabled 1 + %define RESET_MM_PERMUTATION INIT_ZMM %1 + %define mmsize 64 + %define mova movdqa + %define movu movdqu + %undef movh + %define movnta movntdq + INIT_CPUFLAGS %1 + DEFINE_MMREGS zmm + AVX512_MM_PERMUTATION +%endmacro + +INIT_XMM + +%macro DECLARE_MMCAST 1 + %define mmmm%1 mm%1 + %define mmxmm%1 mm%1 + %define mmymm%1 mm%1 + %define mmzmm%1 mm%1 + %define xmmmm%1 mm%1 + %define xmmxmm%1 xmm%1 + %define xmmymm%1 xmm%1 + %define xmmzmm%1 xmm%1 + %define ymmmm%1 mm%1 + %define ymmxmm%1 xmm%1 + %define ymmymm%1 ymm%1 + %define ymmzmm%1 ymm%1 + %define zmmmm%1 mm%1 + %define zmmxmm%1 xmm%1 + %define zmmymm%1 ymm%1 + %define zmmzmm%1 zmm%1 + %define xm%1 xmm %+ m%1 + %define ym%1 ymm %+ m%1 + %define zm%1 zmm %+ m%1 +%endmacro + +%assign i 0 +%rep 32 + DECLARE_MMCAST i + %assign i i+1 +%endrep + +; I often want to use macros that permute their arguments. e.g. there's no +; efficient way to implement butterfly or transpose or dct without swapping some +; arguments. +; +; I would like to not have to manually keep track of the permutations: +; If I insert a permutation in the middle of a function, it should automatically +; change everything that follows. For more complex macros I may also have multiple +; implementations, e.g. the SSE2 and SSSE3 versions may have different permutations. +; +; Hence these macros. Insert a PERMUTE or some SWAPs at the end of a macro that +; permutes its arguments. It's equivalent to exchanging the contents of the +; registers, except that this way you exchange the register names instead, so it +; doesn't cost any cycles. + +%macro PERMUTE 2-* ; takes a list of pairs to swap + %rep %0/2 + %xdefine %%tmp%2 m%2 + %rotate 2 + %endrep + %rep %0/2 + %xdefine m%1 %%tmp%2 + CAT_XDEFINE nn, m%1, %1 + %rotate 2 + %endrep +%endmacro + +%macro SWAP 2+ ; swaps a single chain (sometimes more concise than pairs) + %ifnum %1 ; SWAP 0, 1, ... + SWAP_INTERNAL_NUM %1, %2 + %else ; SWAP m0, m1, ... + SWAP_INTERNAL_NAME %1, %2 + %endif +%endmacro + +%macro SWAP_INTERNAL_NUM 2-* + %rep %0-1 + %xdefine %%tmp m%1 + %xdefine m%1 m%2 + %xdefine m%2 %%tmp + CAT_XDEFINE nn, m%1, %1 + CAT_XDEFINE nn, m%2, %2 + %rotate 1 + %endrep +%endmacro + +%macro SWAP_INTERNAL_NAME 2-* + %xdefine %%args nn %+ %1 + %rep %0-1 + %xdefine %%args %%args, nn %+ %2 + %rotate 1 + %endrep + SWAP_INTERNAL_NUM %%args +%endmacro + +; If SAVE_MM_PERMUTATION is placed at the end of a function, then any later +; calls to that function will automatically load the permutation, so values can +; be returned in mmregs. +%macro SAVE_MM_PERMUTATION 0-1 + %if %0 + %xdefine %%f %1_m + %else + %xdefine %%f current_function %+ _m + %endif + %assign %%i 0 + %rep num_mmregs + CAT_XDEFINE %%f, %%i, m %+ %%i + %assign %%i %%i+1 + %endrep +%endmacro + +%macro LOAD_MM_PERMUTATION 1 ; name to load from + %ifdef %1_m0 + %assign %%i 0 + %rep num_mmregs + CAT_XDEFINE m, %%i, %1_m %+ %%i + CAT_XDEFINE nn, m %+ %%i, %%i + %assign %%i %%i+1 + %endrep + %endif +%endmacro + +; Append cpuflags to the callee's name iff the appended name is known and the plain name isn't +%macro call 1 + %ifid %1 + call_internal %1 %+ SUFFIX, %1 + %else + call %1 + %endif +%endmacro +%macro call_internal 2 + %xdefine %%i %2 + %ifndef cglobaled_%2 + %ifdef cglobaled_%1 + %xdefine %%i %1 + %endif + %endif + call %%i + LOAD_MM_PERMUTATION %%i +%endmacro + +; Substitutions that reduce instruction size but are functionally equivalent +%macro add 2 + %ifnum %2 + %if %2==128 + sub %1, -128 + %else + add %1, %2 + %endif + %else + add %1, %2 + %endif +%endmacro + +%macro sub 2 + %ifnum %2 + %if %2==128 + add %1, -128 + %else + sub %1, %2 + %endif + %else + sub %1, %2 + %endif +%endmacro + +;============================================================================= +; AVX abstraction layer +;============================================================================= + +%assign i 0 +%rep 32 + %if i < 8 + CAT_XDEFINE sizeofmm, i, 8 + CAT_XDEFINE regnumofmm, i, i + %endif + CAT_XDEFINE sizeofxmm, i, 16 + CAT_XDEFINE sizeofymm, i, 32 + CAT_XDEFINE sizeofzmm, i, 64 + CAT_XDEFINE regnumofxmm, i, i + CAT_XDEFINE regnumofymm, i, i + CAT_XDEFINE regnumofzmm, i, i + %assign i i+1 +%endrep +%undef i + +%macro CHECK_AVX_INSTR_EMU 3-* + %xdefine %%opcode %1 + %xdefine %%dst %2 + %rep %0-2 + %ifidn %%dst, %3 + %error non-avx emulation of ``%%opcode'' is not supported + %endif + %rotate 1 + %endrep +%endmacro + +;%1 == instruction +;%2 == minimal instruction set +;%3 == 1 if float, 0 if int +;%4 == 1 if 4-operand emulation, 0 if 3-operand emulation, 255 otherwise (no emulation) +;%5 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not +;%6+: operands +%macro RUN_AVX_INSTR 6-9+ + %ifnum sizeof%7 + %assign __sizeofreg sizeof%7 + %elifnum sizeof%6 + %assign __sizeofreg sizeof%6 + %else + %assign __sizeofreg mmsize + %endif + %assign __emulate_avx 0 + %if avx_enabled && __sizeofreg >= 16 + %xdefine __instr v%1 + %else + %xdefine __instr %1 + %if %0 >= 8+%4 + %assign __emulate_avx 1 + %endif + %endif + %ifnidn %2, fnord + %ifdef cpuname + %if notcpuflag(%2) + %error use of ``%1'' %2 instruction in cpuname function: current_function + %elif cpuflags_%2 < cpuflags_sse && notcpuflag(sse2) && __sizeofreg > 8 + %error use of ``%1'' sse2 instruction in cpuname function: current_function + %endif + %endif + %endif + + %if __emulate_avx + %xdefine __src1 %7 + %xdefine __src2 %8 + %if %5 && %4 == 0 + %ifnidn %6, %7 + %ifidn %6, %8 + %xdefine __src1 %8 + %xdefine __src2 %7 + %elifnnum sizeof%8 + ; 3-operand AVX instructions with a memory arg can only have it in src2, + ; whereas SSE emulation prefers to have it in src1 (i.e. the mov). + ; So, if the instruction is commutative with a memory arg, swap them. + %xdefine __src1 %8 + %xdefine __src2 %7 + %endif + %endif + %endif + %ifnidn %6, __src1 + %if %0 >= 9 + CHECK_AVX_INSTR_EMU {%1 %6, %7, %8, %9}, %6, __src2, %9 + %else + CHECK_AVX_INSTR_EMU {%1 %6, %7, %8}, %6, __src2 + %endif + %if __sizeofreg == 8 + MOVQ %6, __src1 + %elif %3 + MOVAPS %6, __src1 + %else + MOVDQA %6, __src1 + %endif + %endif + %if %0 >= 9 + %1 %6, __src2, %9 + %else + %1 %6, __src2 + %endif + %elif %0 >= 9 + __instr %6, %7, %8, %9 + %elif %0 == 8 + __instr %6, %7, %8 + %elif %0 == 7 + __instr %6, %7 + %else + __instr %6 + %endif +%endmacro + +;%1 == instruction +;%2 == minimal instruction set +;%3 == 1 if float, 0 if int +;%4 == 1 if 4-operand emulation, 0 if 3-operand emulation, 255 otherwise (no emulation) +;%5 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not +%macro AVX_INSTR 1-5 fnord, 0, 255, 0 + %macro %1 1-10 fnord, fnord, fnord, fnord, %1, %2, %3, %4, %5 + %ifidn %2, fnord + RUN_AVX_INSTR %6, %7, %8, %9, %10, %1 + %elifidn %3, fnord + RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2 + %elifidn %4, fnord + RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3 + %elifidn %5, fnord + RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3, %4 + %else + RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3, %4, %5 + %endif + %endmacro +%endmacro + +; Instructions with both VEX/EVEX and legacy encodings +; Non-destructive instructions are written without parameters +AVX_INSTR addpd, sse2, 1, 0, 1 +AVX_INSTR addps, sse, 1, 0, 1 +AVX_INSTR addsd, sse2, 1, 0, 0 +AVX_INSTR addss, sse, 1, 0, 0 +AVX_INSTR addsubpd, sse3, 1, 0, 0 +AVX_INSTR addsubps, sse3, 1, 0, 0 +AVX_INSTR aesdec, aesni, 0, 0, 0 +AVX_INSTR aesdeclast, aesni, 0, 0, 0 +AVX_INSTR aesenc, aesni, 0, 0, 0 +AVX_INSTR aesenclast, aesni, 0, 0, 0 +AVX_INSTR aesimc, aesni +AVX_INSTR aeskeygenassist, aesni +AVX_INSTR andnpd, sse2, 1, 0, 0 +AVX_INSTR andnps, sse, 1, 0, 0 +AVX_INSTR andpd, sse2, 1, 0, 1 +AVX_INSTR andps, sse, 1, 0, 1 +AVX_INSTR blendpd, sse4, 1, 1, 0 +AVX_INSTR blendps, sse4, 1, 1, 0 +AVX_INSTR blendvpd, sse4 ; can't be emulated +AVX_INSTR blendvps, sse4 ; can't be emulated +AVX_INSTR cmpeqpd, sse2, 1, 0, 1 +AVX_INSTR cmpeqps, sse, 1, 0, 1 +AVX_INSTR cmpeqsd, sse2, 1, 0, 0 +AVX_INSTR cmpeqss, sse, 1, 0, 0 +AVX_INSTR cmplepd, sse2, 1, 0, 0 +AVX_INSTR cmpleps, sse, 1, 0, 0 +AVX_INSTR cmplesd, sse2, 1, 0, 0 +AVX_INSTR cmpless, sse, 1, 0, 0 +AVX_INSTR cmpltpd, sse2, 1, 0, 0 +AVX_INSTR cmpltps, sse, 1, 0, 0 +AVX_INSTR cmpltsd, sse2, 1, 0, 0 +AVX_INSTR cmpltss, sse, 1, 0, 0 +AVX_INSTR cmpneqpd, sse2, 1, 0, 1 +AVX_INSTR cmpneqps, sse, 1, 0, 1 +AVX_INSTR cmpneqsd, sse2, 1, 0, 0 +AVX_INSTR cmpneqss, sse, 1, 0, 0 +AVX_INSTR cmpnlepd, sse2, 1, 0, 0 +AVX_INSTR cmpnleps, sse, 1, 0, 0 +AVX_INSTR cmpnlesd, sse2, 1, 0, 0 +AVX_INSTR cmpnless, sse, 1, 0, 0 +AVX_INSTR cmpnltpd, sse2, 1, 0, 0 +AVX_INSTR cmpnltps, sse, 1, 0, 0 +AVX_INSTR cmpnltsd, sse2, 1, 0, 0 +AVX_INSTR cmpnltss, sse, 1, 0, 0 +AVX_INSTR cmpordpd, sse2 1, 0, 1 +AVX_INSTR cmpordps, sse 1, 0, 1 +AVX_INSTR cmpordsd, sse2 1, 0, 0 +AVX_INSTR cmpordss, sse 1, 0, 0 +AVX_INSTR cmppd, sse2, 1, 1, 0 +AVX_INSTR cmpps, sse, 1, 1, 0 +AVX_INSTR cmpsd, sse2, 1, 1, 0 +AVX_INSTR cmpss, sse, 1, 1, 0 +AVX_INSTR cmpunordpd, sse2, 1, 0, 1 +AVX_INSTR cmpunordps, sse, 1, 0, 1 +AVX_INSTR cmpunordsd, sse2, 1, 0, 0 +AVX_INSTR cmpunordss, sse, 1, 0, 0 +AVX_INSTR comisd, sse2 +AVX_INSTR comiss, sse +AVX_INSTR cvtdq2pd, sse2 +AVX_INSTR cvtdq2ps, sse2 +AVX_INSTR cvtpd2dq, sse2 +AVX_INSTR cvtpd2ps, sse2 +AVX_INSTR cvtps2dq, sse2 +AVX_INSTR cvtps2pd, sse2 +AVX_INSTR cvtsd2si, sse2 +AVX_INSTR cvtsd2ss, sse2, 1, 0, 0 +AVX_INSTR cvtsi2sd, sse2, 1, 0, 0 +AVX_INSTR cvtsi2ss, sse, 1, 0, 0 +AVX_INSTR cvtss2sd, sse2, 1, 0, 0 +AVX_INSTR cvtss2si, sse +AVX_INSTR cvttpd2dq, sse2 +AVX_INSTR cvttps2dq, sse2 +AVX_INSTR cvttsd2si, sse2 +AVX_INSTR cvttss2si, sse +AVX_INSTR divpd, sse2, 1, 0, 0 +AVX_INSTR divps, sse, 1, 0, 0 +AVX_INSTR divsd, sse2, 1, 0, 0 +AVX_INSTR divss, sse, 1, 0, 0 +AVX_INSTR dppd, sse4, 1, 1, 0 +AVX_INSTR dpps, sse4, 1, 1, 0 +AVX_INSTR extractps, sse4 +AVX_INSTR haddpd, sse3, 1, 0, 0 +AVX_INSTR haddps, sse3, 1, 0, 0 +AVX_INSTR hsubpd, sse3, 1, 0, 0 +AVX_INSTR hsubps, sse3, 1, 0, 0 +AVX_INSTR insertps, sse4, 1, 1, 0 +AVX_INSTR lddqu, sse3 +AVX_INSTR ldmxcsr, sse +AVX_INSTR maskmovdqu, sse2 +AVX_INSTR maxpd, sse2, 1, 0, 1 +AVX_INSTR maxps, sse, 1, 0, 1 +AVX_INSTR maxsd, sse2, 1, 0, 0 +AVX_INSTR maxss, sse, 1, 0, 0 +AVX_INSTR minpd, sse2, 1, 0, 1 +AVX_INSTR minps, sse, 1, 0, 1 +AVX_INSTR minsd, sse2, 1, 0, 0 +AVX_INSTR minss, sse, 1, 0, 0 +AVX_INSTR movapd, sse2 +AVX_INSTR movaps, sse +AVX_INSTR movd, mmx +AVX_INSTR movddup, sse3 +AVX_INSTR movdqa, sse2 +AVX_INSTR movdqu, sse2 +AVX_INSTR movhlps, sse, 1, 0, 0 +AVX_INSTR movhpd, sse2, 1, 0, 0 +AVX_INSTR movhps, sse, 1, 0, 0 +AVX_INSTR movlhps, sse, 1, 0, 0 +AVX_INSTR movlpd, sse2, 1, 0, 0 +AVX_INSTR movlps, sse, 1, 0, 0 +AVX_INSTR movmskpd, sse2 +AVX_INSTR movmskps, sse +AVX_INSTR movntdq, sse2 +AVX_INSTR movntdqa, sse4 +AVX_INSTR movntpd, sse2 +AVX_INSTR movntps, sse +AVX_INSTR movq, mmx +AVX_INSTR movsd, sse2, 1, 0, 0 +AVX_INSTR movshdup, sse3 +AVX_INSTR movsldup, sse3 +AVX_INSTR movss, sse, 1, 0, 0 +AVX_INSTR movupd, sse2 +AVX_INSTR movups, sse +AVX_INSTR mpsadbw, sse4, 0, 1, 0 +AVX_INSTR mulpd, sse2, 1, 0, 1 +AVX_INSTR mulps, sse, 1, 0, 1 +AVX_INSTR mulsd, sse2, 1, 0, 0 +AVX_INSTR mulss, sse, 1, 0, 0 +AVX_INSTR orpd, sse2, 1, 0, 1 +AVX_INSTR orps, sse, 1, 0, 1 +AVX_INSTR pabsb, ssse3 +AVX_INSTR pabsd, ssse3 +AVX_INSTR pabsw, ssse3 +AVX_INSTR packsswb, mmx, 0, 0, 0 +AVX_INSTR packssdw, mmx, 0, 0, 0 +AVX_INSTR packuswb, mmx, 0, 0, 0 +AVX_INSTR packusdw, sse4, 0, 0, 0 +AVX_INSTR paddb, mmx, 0, 0, 1 +AVX_INSTR paddw, mmx, 0, 0, 1 +AVX_INSTR paddd, mmx, 0, 0, 1 +AVX_INSTR paddq, sse2, 0, 0, 1 +AVX_INSTR paddsb, mmx, 0, 0, 1 +AVX_INSTR paddsw, mmx, 0, 0, 1 +AVX_INSTR paddusb, mmx, 0, 0, 1 +AVX_INSTR paddusw, mmx, 0, 0, 1 +AVX_INSTR palignr, ssse3, 0, 1, 0 +AVX_INSTR pand, mmx, 0, 0, 1 +AVX_INSTR pandn, mmx, 0, 0, 0 +AVX_INSTR pavgb, mmx2, 0, 0, 1 +AVX_INSTR pavgw, mmx2, 0, 0, 1 +AVX_INSTR pblendvb, sse4 ; can't be emulated +AVX_INSTR pblendw, sse4, 0, 1, 0 +AVX_INSTR pclmulqdq, fnord, 0, 1, 0 +AVX_INSTR pclmulhqhqdq, fnord, 0, 0, 0 +AVX_INSTR pclmulhqlqdq, fnord, 0, 0, 0 +AVX_INSTR pclmullqhqdq, fnord, 0, 0, 0 +AVX_INSTR pclmullqlqdq, fnord, 0, 0, 0 +AVX_INSTR pcmpestri, sse42 +AVX_INSTR pcmpestrm, sse42 +AVX_INSTR pcmpistri, sse42 +AVX_INSTR pcmpistrm, sse42 +AVX_INSTR pcmpeqb, mmx, 0, 0, 1 +AVX_INSTR pcmpeqw, mmx, 0, 0, 1 +AVX_INSTR pcmpeqd, mmx, 0, 0, 1 +AVX_INSTR pcmpeqq, sse4, 0, 0, 1 +AVX_INSTR pcmpgtb, mmx, 0, 0, 0 +AVX_INSTR pcmpgtw, mmx, 0, 0, 0 +AVX_INSTR pcmpgtd, mmx, 0, 0, 0 +AVX_INSTR pcmpgtq, sse42, 0, 0, 0 +AVX_INSTR pextrb, sse4 +AVX_INSTR pextrd, sse4 +AVX_INSTR pextrq, sse4 +AVX_INSTR pextrw, mmx2 +AVX_INSTR phaddw, ssse3, 0, 0, 0 +AVX_INSTR phaddd, ssse3, 0, 0, 0 +AVX_INSTR phaddsw, ssse3, 0, 0, 0 +AVX_INSTR phminposuw, sse4 +AVX_INSTR phsubw, ssse3, 0, 0, 0 +AVX_INSTR phsubd, ssse3, 0, 0, 0 +AVX_INSTR phsubsw, ssse3, 0, 0, 0 +AVX_INSTR pinsrb, sse4, 0, 1, 0 +AVX_INSTR pinsrd, sse4, 0, 1, 0 +AVX_INSTR pinsrq, sse4, 0, 1, 0 +AVX_INSTR pinsrw, mmx2, 0, 1, 0 +AVX_INSTR pmaddwd, mmx, 0, 0, 1 +AVX_INSTR pmaddubsw, ssse3, 0, 0, 0 +AVX_INSTR pmaxsb, sse4, 0, 0, 1 +AVX_INSTR pmaxsw, mmx2, 0, 0, 1 +AVX_INSTR pmaxsd, sse4, 0, 0, 1 +AVX_INSTR pmaxub, mmx2, 0, 0, 1 +AVX_INSTR pmaxuw, sse4, 0, 0, 1 +AVX_INSTR pmaxud, sse4, 0, 0, 1 +AVX_INSTR pminsb, sse4, 0, 0, 1 +AVX_INSTR pminsw, mmx2, 0, 0, 1 +AVX_INSTR pminsd, sse4, 0, 0, 1 +AVX_INSTR pminub, mmx2, 0, 0, 1 +AVX_INSTR pminuw, sse4, 0, 0, 1 +AVX_INSTR pminud, sse4, 0, 0, 1 +AVX_INSTR pmovmskb, mmx2 +AVX_INSTR pmovsxbw, sse4 +AVX_INSTR pmovsxbd, sse4 +AVX_INSTR pmovsxbq, sse4 +AVX_INSTR pmovsxwd, sse4 +AVX_INSTR pmovsxwq, sse4 +AVX_INSTR pmovsxdq, sse4 +AVX_INSTR pmovzxbw, sse4 +AVX_INSTR pmovzxbd, sse4 +AVX_INSTR pmovzxbq, sse4 +AVX_INSTR pmovzxwd, sse4 +AVX_INSTR pmovzxwq, sse4 +AVX_INSTR pmovzxdq, sse4 +AVX_INSTR pmuldq, sse4, 0, 0, 1 +AVX_INSTR pmulhrsw, ssse3, 0, 0, 1 +AVX_INSTR pmulhuw, mmx2, 0, 0, 1 +AVX_INSTR pmulhw, mmx, 0, 0, 1 +AVX_INSTR pmullw, mmx, 0, 0, 1 +AVX_INSTR pmulld, sse4, 0, 0, 1 +AVX_INSTR pmuludq, sse2, 0, 0, 1 +AVX_INSTR por, mmx, 0, 0, 1 +AVX_INSTR psadbw, mmx2, 0, 0, 1 +AVX_INSTR pshufb, ssse3, 0, 0, 0 +AVX_INSTR pshufd, sse2 +AVX_INSTR pshufhw, sse2 +AVX_INSTR pshuflw, sse2 +AVX_INSTR psignb, ssse3, 0, 0, 0 +AVX_INSTR psignw, ssse3, 0, 0, 0 +AVX_INSTR psignd, ssse3, 0, 0, 0 +AVX_INSTR psllw, mmx, 0, 0, 0 +AVX_INSTR pslld, mmx, 0, 0, 0 +AVX_INSTR psllq, mmx, 0, 0, 0 +AVX_INSTR pslldq, sse2, 0, 0, 0 +AVX_INSTR psraw, mmx, 0, 0, 0 +AVX_INSTR psrad, mmx, 0, 0, 0 +AVX_INSTR psrlw, mmx, 0, 0, 0 +AVX_INSTR psrld, mmx, 0, 0, 0 +AVX_INSTR psrlq, mmx, 0, 0, 0 +AVX_INSTR psrldq, sse2, 0, 0, 0 +AVX_INSTR psubb, mmx, 0, 0, 0 +AVX_INSTR psubw, mmx, 0, 0, 0 +AVX_INSTR psubd, mmx, 0, 0, 0 +AVX_INSTR psubq, sse2, 0, 0, 0 +AVX_INSTR psubsb, mmx, 0, 0, 0 +AVX_INSTR psubsw, mmx, 0, 0, 0 +AVX_INSTR psubusb, mmx, 0, 0, 0 +AVX_INSTR psubusw, mmx, 0, 0, 0 +AVX_INSTR ptest, sse4 +AVX_INSTR punpckhbw, mmx, 0, 0, 0 +AVX_INSTR punpckhwd, mmx, 0, 0, 0 +AVX_INSTR punpckhdq, mmx, 0, 0, 0 +AVX_INSTR punpckhqdq, sse2, 0, 0, 0 +AVX_INSTR punpcklbw, mmx, 0, 0, 0 +AVX_INSTR punpcklwd, mmx, 0, 0, 0 +AVX_INSTR punpckldq, mmx, 0, 0, 0 +AVX_INSTR punpcklqdq, sse2, 0, 0, 0 +AVX_INSTR pxor, mmx, 0, 0, 1 +AVX_INSTR rcpps, sse +AVX_INSTR rcpss, sse, 1, 0, 0 +AVX_INSTR roundpd, sse4 +AVX_INSTR roundps, sse4 +AVX_INSTR roundsd, sse4, 1, 1, 0 +AVX_INSTR roundss, sse4, 1, 1, 0 +AVX_INSTR rsqrtps, sse +AVX_INSTR rsqrtss, sse, 1, 0, 0 +AVX_INSTR shufpd, sse2, 1, 1, 0 +AVX_INSTR shufps, sse, 1, 1, 0 +AVX_INSTR sqrtpd, sse2 +AVX_INSTR sqrtps, sse +AVX_INSTR sqrtsd, sse2, 1, 0, 0 +AVX_INSTR sqrtss, sse, 1, 0, 0 +AVX_INSTR stmxcsr, sse +AVX_INSTR subpd, sse2, 1, 0, 0 +AVX_INSTR subps, sse, 1, 0, 0 +AVX_INSTR subsd, sse2, 1, 0, 0 +AVX_INSTR subss, sse, 1, 0, 0 +AVX_INSTR ucomisd, sse2 +AVX_INSTR ucomiss, sse +AVX_INSTR unpckhpd, sse2, 1, 0, 0 +AVX_INSTR unpckhps, sse, 1, 0, 0 +AVX_INSTR unpcklpd, sse2, 1, 0, 0 +AVX_INSTR unpcklps, sse, 1, 0, 0 +AVX_INSTR xorpd, sse2, 1, 0, 1 +AVX_INSTR xorps, sse, 1, 0, 1 + +; 3DNow instructions, for sharing code between AVX, SSE and 3DN +AVX_INSTR pfadd, 3dnow, 1, 0, 1 +AVX_INSTR pfsub, 3dnow, 1, 0, 0 +AVX_INSTR pfmul, 3dnow, 1, 0, 1 + +; base-4 constants for shuffles +%assign i 0 +%rep 256 + %assign j ((i>>6)&3)*1000 + ((i>>4)&3)*100 + ((i>>2)&3)*10 + (i&3) + %if j < 10 + CAT_XDEFINE q000, j, i + %elif j < 100 + CAT_XDEFINE q00, j, i + %elif j < 1000 + CAT_XDEFINE q0, j, i + %else + CAT_XDEFINE q, j, i + %endif + %assign i i+1 +%endrep +%undef i +%undef j + +%macro FMA_INSTR 3 + %macro %1 4-7 %1, %2, %3 + %if cpuflag(xop) + v%5 %1, %2, %3, %4 + %elifnidn %1, %4 + %6 %1, %2, %3 + %7 %1, %4 + %else + %error non-xop emulation of ``%5 %1, %2, %3, %4'' is not supported + %endif + %endmacro +%endmacro + +FMA_INSTR pmacsww, pmullw, paddw +FMA_INSTR pmacsdd, pmulld, paddd ; sse4 emulation +FMA_INSTR pmacsdql, pmuldq, paddq ; sse4 emulation +FMA_INSTR pmadcswd, pmaddwd, paddd + +; tzcnt is equivalent to "rep bsf" and is backwards-compatible with bsf. +; This lets us use tzcnt without bumping the yasm version requirement yet. +%define tzcnt rep bsf + +; Macros for consolidating FMA3 and FMA4 using 4-operand (dst, src1, src2, src3) syntax. +; FMA3 is only possible if dst is the same as one of the src registers. +; Either src2 or src3 can be a memory operand. +%macro FMA4_INSTR 2-* + %push fma4_instr + %xdefine %$prefix %1 + %rep %0 - 1 + %macro %$prefix%2 4-6 %$prefix, %2 + %if notcpuflag(fma3) && notcpuflag(fma4) + %error use of ``%5%6'' fma instruction in cpuname function: current_function + %elif cpuflag(fma4) + v%5%6 %1, %2, %3, %4 + %elifidn %1, %2 + ; If %3 or %4 is a memory operand it needs to be encoded as the last operand. + %ifnum sizeof%3 + v%{5}213%6 %2, %3, %4 + %else + v%{5}132%6 %2, %4, %3 + %endif + %elifidn %1, %3 + v%{5}213%6 %3, %2, %4 + %elifidn %1, %4 + v%{5}231%6 %4, %2, %3 + %else + %error fma3 emulation of ``%5%6 %1, %2, %3, %4'' is not supported + %endif + %endmacro + %rotate 1 + %endrep + %pop +%endmacro + +FMA4_INSTR fmadd, pd, ps, sd, ss +FMA4_INSTR fmaddsub, pd, ps +FMA4_INSTR fmsub, pd, ps, sd, ss +FMA4_INSTR fmsubadd, pd, ps +FMA4_INSTR fnmadd, pd, ps, sd, ss +FMA4_INSTR fnmsub, pd, ps, sd, ss + +; Macros for converting VEX instructions to equivalent EVEX ones. +%macro EVEX_INSTR 2-3 0 ; vex, evex, prefer_evex + %macro %1 2-7 fnord, fnord, %1, %2, %3 + %ifidn %3, fnord + %define %%args %1, %2 + %elifidn %4, fnord + %define %%args %1, %2, %3 + %else + %define %%args %1, %2, %3, %4 + %endif + %assign %%evex_required cpuflag(avx512) & %7 + %ifnum regnumof%1 + %if regnumof%1 >= 16 || sizeof%1 > 32 + %assign %%evex_required 1 + %endif + %endif + %ifnum regnumof%2 + %if regnumof%2 >= 16 || sizeof%2 > 32 + %assign %%evex_required 1 + %endif + %endif + %if %%evex_required + %6 %%args + %else + %5 %%args ; Prefer VEX over EVEX due to shorter instruction length + %endif + %endmacro +%endmacro + +EVEX_INSTR vbroadcastf128, vbroadcastf32x4 +EVEX_INSTR vbroadcasti128, vbroadcasti32x4 +EVEX_INSTR vextractf128, vextractf32x4 +EVEX_INSTR vextracti128, vextracti32x4 +EVEX_INSTR vinsertf128, vinsertf32x4 +EVEX_INSTR vinserti128, vinserti32x4 +EVEX_INSTR vmovdqa, vmovdqa32 +EVEX_INSTR vmovdqu, vmovdqu32 +EVEX_INSTR vpand, vpandd +EVEX_INSTR vpandn, vpandnd +EVEX_INSTR vpor, vpord +EVEX_INSTR vpxor, vpxord +EVEX_INSTR vrcpps, vrcp14ps, 1 ; EVEX versions have higher precision +EVEX_INSTR vrcpss, vrcp14ss, 1 +EVEX_INSTR vrsqrtps, vrsqrt14ps, 1 +EVEX_INSTR vrsqrtss, vrsqrt14ss, 1 + +; workaround: vpbroadcastq is broken in x86_32 due to a yasm bug (fixed in 1.3.0) +%ifdef __YASM_VER__ + %if __YASM_VERSION_ID__ < 0x01030000 && ARCH_X86_64 == 0 + %macro vpbroadcastq 2 + %if sizeof%1 == 16 + movddup %1, %2 + %else + vbroadcastsd %1, %2 + %endif + %endmacro + %endif +%endif diff --git a/media/ffvpx/libavutil/x86/x86util.asm b/media/ffvpx/libavutil/x86/x86util.asm new file mode 100644 index 0000000000..d7cd996842 --- /dev/null +++ b/media/ffvpx/libavutil/x86/x86util.asm @@ -0,0 +1,1028 @@ +;***************************************************************************** +;* x86util.asm +;***************************************************************************** +;* Copyright (C) 2008-2010 x264 project +;* +;* Authors: Loren Merritt <lorenm@u.washington.edu> +;* Holger Lubitz <holger@lubitz.org> +;* +;* This file is part of FFmpeg. +;* +;* FFmpeg is free software; you can redistribute it and/or +;* modify it under the terms of the GNU Lesser General Public +;* License as published by the Free Software Foundation; either +;* version 2.1 of the License, or (at your option) any later version. +;* +;* FFmpeg is distributed in the hope that it will be useful, +;* but WITHOUT ANY WARRANTY; without even the implied warranty of +;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;* Lesser General Public License for more details. +;* +;* You should have received a copy of the GNU Lesser General Public +;* License along with FFmpeg; if not, write to the Free Software +;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA +;****************************************************************************** + +%define private_prefix ff +%define public_prefix avpriv +%define cpuflags_mmxext cpuflags_mmx2 + +%include "libavutil/x86/x86inc.asm" + +; expands to [base],...,[base+7*stride] +%define PASS8ROWS(base, base3, stride, stride3) \ + [base], [base + stride], [base + 2*stride], [base3], \ + [base3 + stride], [base3 + 2*stride], [base3 + stride3], [base3 + stride*4] + +; Interleave low src0 with low src1 and store in src0, +; interleave high src0 with high src1 and store in src1. +; %1 - types +; %2 - index of the register with src0 +; %3 - index of the register with src1 +; %4 - index of the register for intermediate results +; example for %1 - wd: input: src0: x0 x1 x2 x3 z0 z1 z2 z3 +; src1: y0 y1 y2 y3 q0 q1 q2 q3 +; output: src0: x0 y0 x1 y1 x2 y2 x3 y3 +; src1: z0 q0 z1 q1 z2 q2 z3 q3 +%macro SBUTTERFLY 4 +%ifidn %1, dqqq + vperm2i128 m%4, m%2, m%3, q0301 + vinserti128 m%2, m%2, xm%3, 1 +%elif avx_enabled == 0 + mova m%4, m%2 + punpckl%1 m%2, m%3 + punpckh%1 m%4, m%3 +%else + punpckh%1 m%4, m%2, m%3 + punpckl%1 m%2, m%3 +%endif + SWAP %3, %4 +%endmacro + +%macro SBUTTERFLY2 4 + punpckl%1 m%4, m%2, m%3 + punpckh%1 m%2, m%2, m%3 + SWAP %2, %4, %3 +%endmacro + +%macro SBUTTERFLYPS 3 + unpcklps m%3, m%1, m%2 + unpckhps m%1, m%1, m%2 + SWAP %1, %3, %2 +%endmacro + +%macro SBUTTERFLYPD 3 + movlhps m%3, m%1, m%2 + movhlps m%2, m%2, m%1 + SWAP %1, %3 +%endmacro + +%macro TRANSPOSE4x4B 5 + SBUTTERFLY bw, %1, %2, %5 + SBUTTERFLY bw, %3, %4, %5 + SBUTTERFLY wd, %1, %3, %5 + SBUTTERFLY wd, %2, %4, %5 + SWAP %2, %3 +%endmacro + +%macro TRANSPOSE4x4W 5 + SBUTTERFLY wd, %1, %2, %5 + SBUTTERFLY wd, %3, %4, %5 + SBUTTERFLY dq, %1, %3, %5 + SBUTTERFLY dq, %2, %4, %5 + SWAP %2, %3 +%endmacro + +%macro TRANSPOSE2x4x4B 5 + SBUTTERFLY bw, %1, %2, %5 + SBUTTERFLY bw, %3, %4, %5 + SBUTTERFLY wd, %1, %3, %5 + SBUTTERFLY wd, %2, %4, %5 + SBUTTERFLY dq, %1, %2, %5 + SBUTTERFLY dq, %3, %4, %5 +%endmacro + +%macro TRANSPOSE2x4x4W 5 + SBUTTERFLY wd, %1, %2, %5 + SBUTTERFLY wd, %3, %4, %5 + SBUTTERFLY dq, %1, %3, %5 + SBUTTERFLY dq, %2, %4, %5 + SBUTTERFLY qdq, %1, %2, %5 + SBUTTERFLY qdq, %3, %4, %5 +%endmacro + +%macro TRANSPOSE4x4D 5 + SBUTTERFLY dq, %1, %2, %5 + SBUTTERFLY dq, %3, %4, %5 + SBUTTERFLY qdq, %1, %3, %5 + SBUTTERFLY qdq, %2, %4, %5 + SWAP %2, %3 +%endmacro + +; identical behavior to TRANSPOSE4x4D, but using SSE1 float ops +%macro TRANSPOSE4x4PS 5 + SBUTTERFLYPS %1, %2, %5 + SBUTTERFLYPS %3, %4, %5 + SBUTTERFLYPD %1, %3, %5 + SBUTTERFLYPD %2, %4, %5 + SWAP %2, %3 +%endmacro + +%macro TRANSPOSE8x4D 9-11 +%if ARCH_X86_64 + SBUTTERFLY dq, %1, %2, %9 + SBUTTERFLY dq, %3, %4, %9 + SBUTTERFLY dq, %5, %6, %9 + SBUTTERFLY dq, %7, %8, %9 + SBUTTERFLY qdq, %1, %3, %9 + SBUTTERFLY qdq, %2, %4, %9 + SBUTTERFLY qdq, %5, %7, %9 + SBUTTERFLY qdq, %6, %8, %9 + SWAP %2, %5 + SWAP %4, %7 +%else +; in: m0..m7 +; out: m0..m7, unless %11 in which case m2 is in %9 +; spills into %9 and %10 + movdqa %9, m%7 + SBUTTERFLY dq, %1, %2, %7 + movdqa %10, m%2 + movdqa m%7, %9 + SBUTTERFLY dq, %3, %4, %2 + SBUTTERFLY dq, %5, %6, %2 + SBUTTERFLY dq, %7, %8, %2 + SBUTTERFLY qdq, %1, %3, %2 + movdqa %9, m%3 + movdqa m%2, %10 + SBUTTERFLY qdq, %2, %4, %3 + SBUTTERFLY qdq, %5, %7, %3 + SBUTTERFLY qdq, %6, %8, %3 + SWAP %2, %5 + SWAP %4, %7 +%if %0<11 + movdqa m%3, %9 +%endif +%endif +%endmacro + +%macro TRANSPOSE8x8W 9-11 +%if ARCH_X86_64 + SBUTTERFLY wd, %1, %2, %9 + SBUTTERFLY wd, %3, %4, %9 + SBUTTERFLY wd, %5, %6, %9 + SBUTTERFLY wd, %7, %8, %9 + SBUTTERFLY dq, %1, %3, %9 + SBUTTERFLY dq, %2, %4, %9 + SBUTTERFLY dq, %5, %7, %9 + SBUTTERFLY dq, %6, %8, %9 + SBUTTERFLY qdq, %1, %5, %9 + SBUTTERFLY qdq, %2, %6, %9 + SBUTTERFLY qdq, %3, %7, %9 + SBUTTERFLY qdq, %4, %8, %9 + SWAP %2, %5 + SWAP %4, %7 +%else +; in: m0..m7, unless %11 in which case m6 is in %9 +; out: m0..m7, unless %11 in which case m4 is in %10 +; spills into %9 and %10 +%if %0<11 + movdqa %9, m%7 +%endif + SBUTTERFLY wd, %1, %2, %7 + movdqa %10, m%2 + movdqa m%7, %9 + SBUTTERFLY wd, %3, %4, %2 + SBUTTERFLY wd, %5, %6, %2 + SBUTTERFLY wd, %7, %8, %2 + SBUTTERFLY dq, %1, %3, %2 + movdqa %9, m%3 + movdqa m%2, %10 + SBUTTERFLY dq, %2, %4, %3 + SBUTTERFLY dq, %5, %7, %3 + SBUTTERFLY dq, %6, %8, %3 + SBUTTERFLY qdq, %1, %5, %3 + SBUTTERFLY qdq, %2, %6, %3 + movdqa %10, m%2 + movdqa m%3, %9 + SBUTTERFLY qdq, %3, %7, %2 + SBUTTERFLY qdq, %4, %8, %2 + SWAP %2, %5 + SWAP %4, %7 +%if %0<11 + movdqa m%5, %10 +%endif +%endif +%endmacro + +%macro TRANSPOSE16x16W 18-19 +; in: m0..m15, unless %19 in which case m6 is in %17 +; out: m0..m15, unless %19 in which case m4 is in %18 +; spills into %17 and %18 +%if %0 < 19 + mova %17, m%7 +%endif + + SBUTTERFLY dqqq, %1, %9, %7 + SBUTTERFLY dqqq, %2, %10, %7 + SBUTTERFLY dqqq, %3, %11, %7 + SBUTTERFLY dqqq, %4, %12, %7 + SBUTTERFLY dqqq, %5, %13, %7 + SBUTTERFLY dqqq, %6, %14, %7 + mova %18, m%14 + mova m%7, %17 + SBUTTERFLY dqqq, %7, %15, %14 + SBUTTERFLY dqqq, %8, %16, %14 + + SBUTTERFLY wd, %1, %2, %14 + SBUTTERFLY wd, %3, %4, %14 + SBUTTERFLY wd, %5, %6, %14 + SBUTTERFLY wd, %7, %8, %14 + SBUTTERFLY wd, %9, %10, %14 + SBUTTERFLY wd, %11, %12, %14 + mova %17, m%12 + mova m%14, %18 + SBUTTERFLY wd, %13, %14, %12 + SBUTTERFLY wd, %15, %16, %12 + + SBUTTERFLY dq, %1, %3, %12 + SBUTTERFLY dq, %2, %4, %12 + SBUTTERFLY dq, %5, %7, %12 + SBUTTERFLY dq, %6, %8, %12 + SBUTTERFLY dq, %9, %11, %12 + mova %18, m%11 + mova m%12, %17 + SBUTTERFLY dq, %10, %12, %11 + SBUTTERFLY dq, %13, %15, %11 + SBUTTERFLY dq, %14, %16, %11 + + SBUTTERFLY qdq, %1, %5, %11 + SBUTTERFLY qdq, %2, %6, %11 + SBUTTERFLY qdq, %3, %7, %11 + SBUTTERFLY qdq, %4, %8, %11 + + SWAP %2, %5 + SWAP %4, %7 + + SBUTTERFLY qdq, %9, %13, %11 + SBUTTERFLY qdq, %10, %14, %11 + mova m%11, %18 + mova %18, m%5 + SBUTTERFLY qdq, %11, %15, %5 + SBUTTERFLY qdq, %12, %16, %5 + +%if %0 < 19 + mova m%5, %18 +%endif + + SWAP %10, %13 + SWAP %12, %15 +%endmacro + +%macro TRANSPOSE_8X8B 8 + %if mmsize == 8 + %error "This macro does not support mmsize == 8" + %endif + punpcklbw m%1, m%2 + punpcklbw m%3, m%4 + punpcklbw m%5, m%6 + punpcklbw m%7, m%8 + TRANSPOSE4x4W %1, %3, %5, %7, %2 + MOVHL m%2, m%1 + MOVHL m%4, m%3 + MOVHL m%6, m%5 + MOVHL m%8, m%7 +%endmacro + +; PABSW macro assumes %1 != %2, while ABS1/2 macros work in-place +%macro PABSW 2 +%if cpuflag(ssse3) + pabsw %1, %2 +%elif cpuflag(mmxext) + pxor %1, %1 + psubw %1, %2 + pmaxsw %1, %2 +%else + pxor %1, %1 + pcmpgtw %1, %2 + pxor %2, %1 + psubw %2, %1 + SWAP %1, %2 +%endif +%endmacro + +%macro PSIGNW 2 +%if cpuflag(ssse3) + psignw %1, %2 +%else + pxor %1, %2 + psubw %1, %2 +%endif +%endmacro + +%macro ABS1 2 +%if cpuflag(ssse3) + pabsw %1, %1 +%elif cpuflag(mmxext) ; a, tmp + pxor %2, %2 + psubw %2, %1 + pmaxsw %1, %2 +%else ; a, tmp + pxor %2, %2 + pcmpgtw %2, %1 + pxor %1, %2 + psubw %1, %2 +%endif +%endmacro + +%macro ABS2 4 +%if cpuflag(ssse3) + pabsw %1, %1 + pabsw %2, %2 +%elif cpuflag(mmxext) ; a, b, tmp0, tmp1 + pxor %3, %3 + pxor %4, %4 + psubw %3, %1 + psubw %4, %2 + pmaxsw %1, %3 + pmaxsw %2, %4 +%else ; a, b, tmp0, tmp1 + pxor %3, %3 + pxor %4, %4 + pcmpgtw %3, %1 + pcmpgtw %4, %2 + pxor %1, %3 + pxor %2, %4 + psubw %1, %3 + psubw %2, %4 +%endif +%endmacro + +%macro ABSB 2 ; source mmreg, temp mmreg (unused for SSSE3) +%if cpuflag(ssse3) + pabsb %1, %1 +%else + pxor %2, %2 + psubb %2, %1 + pminub %1, %2 +%endif +%endmacro + +%macro ABSB2 4 ; src1, src2, tmp1, tmp2 (tmp1/2 unused for SSSE3) +%if cpuflag(ssse3) + pabsb %1, %1 + pabsb %2, %2 +%else + pxor %3, %3 + pxor %4, %4 + psubb %3, %1 + psubb %4, %2 + pminub %1, %3 + pminub %2, %4 +%endif +%endmacro + +%macro ABSD2 4 + pxor %3, %3 + pxor %4, %4 + pcmpgtd %3, %1 + pcmpgtd %4, %2 + pxor %1, %3 + pxor %2, %4 + psubd %1, %3 + psubd %2, %4 +%endmacro + +%macro ABS4 6 + ABS2 %1, %2, %5, %6 + ABS2 %3, %4, %5, %6 +%endmacro + +%macro SPLATB_LOAD 3 +%if cpuflag(ssse3) + movd %1, [%2-3] + pshufb %1, %3 +%else + movd %1, [%2-3] ;to avoid crossing a cacheline + punpcklbw %1, %1 + SPLATW %1, %1, 3 +%endif +%endmacro + +%macro SPLATB_REG 3 +%if cpuflag(ssse3) + movd %1, %2d + pshufb %1, %3 +%else + movd %1, %2d + punpcklbw %1, %1 + SPLATW %1, %1, 0 +%endif +%endmacro + +%macro HADDD 2 ; sum junk +%if sizeof%1 == 32 +%define %2 xmm%2 + vextracti128 %2, %1, 1 +%define %1 xmm%1 + paddd %1, %2 +%endif +%if mmsize >= 16 +%if cpuflag(xop) && sizeof%1 == 16 + vphadddq %1, %1 +%endif + movhlps %2, %1 + paddd %1, %2 +%endif +%if notcpuflag(xop) || sizeof%1 != 16 +%if cpuflag(mmxext) + PSHUFLW %2, %1, q0032 +%else ; mmx + mova %2, %1 + psrlq %2, 32 +%endif + paddd %1, %2 +%endif +%undef %1 +%undef %2 +%endmacro + +%macro HADDW 2 ; reg, tmp +%if cpuflag(xop) && sizeof%1 == 16 + vphaddwq %1, %1 + movhlps %2, %1 + paddd %1, %2 +%else + pmaddwd %1, [pw_1] + HADDD %1, %2 +%endif +%endmacro + +%macro HADDPS 3 ; dst, src, tmp +%if cpuflag(sse3) + haddps %1, %1, %2 +%else + movaps %3, %1 + shufps %1, %2, q2020 + shufps %3, %2, q3131 + addps %1, %3 +%endif +%endmacro + +%macro PALIGNR 4-5 +%if cpuflag(ssse3) +%if %0==5 + palignr %1, %2, %3, %4 +%else + palignr %1, %2, %3 +%endif +%else ; [dst,] src1, src2, imm, tmp + %define %%dst %1 +%if %0==5 +%ifnidn %1, %2 + mova %%dst, %2 +%endif + %rotate 1 +%endif +%ifnidn %4, %2 + mova %4, %2 +%endif +%if mmsize==8 + psllq %%dst, (8-%3)*8 + psrlq %4, %3*8 +%else + pslldq %%dst, 16-%3 + psrldq %4, %3 +%endif + por %%dst, %4 +%endif +%endmacro + +%macro PAVGB 2-4 +%if cpuflag(mmxext) + pavgb %1, %2 +%elif cpuflag(3dnow) + pavgusb %1, %2 +%elif cpuflag(mmx) + movu %3, %2 + por %3, %1 + pxor %1, %2 + pand %1, %4 + psrlq %1, 1 + psubb %3, %1 + SWAP %1, %3 +%endif +%endmacro + +%macro PSHUFLW 1+ + %if mmsize == 8 + pshufw %1 + %else + pshuflw %1 + %endif +%endmacro + +%macro PSWAPD 2 +%if cpuflag(mmxext) + pshufw %1, %2, q1032 +%elif cpuflag(3dnowext) + pswapd %1, %2 +%elif cpuflag(3dnow) + movq %1, %2 + psrlq %1, 32 + punpckldq %1, %2 +%endif +%endmacro + +%macro DEINTB 5 ; mask, reg1, mask, reg2, optional src to fill masks from +%ifnum %5 + pand m%3, m%5, m%4 ; src .. y6 .. y4 + pand m%1, m%5, m%2 ; dst .. y6 .. y4 +%else + mova m%1, %5 + pand m%3, m%1, m%4 ; src .. y6 .. y4 + pand m%1, m%1, m%2 ; dst .. y6 .. y4 +%endif + psrlw m%2, 8 ; dst .. y7 .. y5 + psrlw m%4, 8 ; src .. y7 .. y5 +%endmacro + +%macro SUMSUB_BA 3-4 +%if %0==3 + padd%1 m%2, m%3 + padd%1 m%3, m%3 + psub%1 m%3, m%2 +%else +%if avx_enabled == 0 + mova m%4, m%2 + padd%1 m%2, m%3 + psub%1 m%3, m%4 +%else + padd%1 m%4, m%2, m%3 + psub%1 m%3, m%2 + SWAP %2, %4 +%endif +%endif +%endmacro + +%macro SUMSUB_BADC 5-6 +%if %0==6 + SUMSUB_BA %1, %2, %3, %6 + SUMSUB_BA %1, %4, %5, %6 +%else + padd%1 m%2, m%3 + padd%1 m%4, m%5 + padd%1 m%3, m%3 + padd%1 m%5, m%5 + psub%1 m%3, m%2 + psub%1 m%5, m%4 +%endif +%endmacro + +%macro SUMSUB2_AB 4 +%ifnum %3 + psub%1 m%4, m%2, m%3 + psub%1 m%4, m%3 + padd%1 m%2, m%2 + padd%1 m%2, m%3 +%else + mova m%4, m%2 + padd%1 m%2, m%2 + padd%1 m%2, %3 + psub%1 m%4, %3 + psub%1 m%4, %3 +%endif +%endmacro + +%macro SUMSUB2_BA 4 +%if avx_enabled == 0 + mova m%4, m%2 + padd%1 m%2, m%3 + padd%1 m%2, m%3 + psub%1 m%3, m%4 + psub%1 m%3, m%4 +%else + padd%1 m%4, m%2, m%3 + padd%1 m%4, m%3 + psub%1 m%3, m%2 + psub%1 m%3, m%2 + SWAP %2, %4 +%endif +%endmacro + +%macro SUMSUBD2_AB 5 +%ifnum %4 + psra%1 m%5, m%2, 1 ; %3: %3>>1 + psra%1 m%4, m%3, 1 ; %2: %2>>1 + padd%1 m%4, m%2 ; %3: %3>>1+%2 + psub%1 m%5, m%3 ; %2: %2>>1-%3 + SWAP %2, %5 + SWAP %3, %4 +%else + mova %5, m%2 + mova %4, m%3 + psra%1 m%3, 1 ; %3: %3>>1 + psra%1 m%2, 1 ; %2: %2>>1 + padd%1 m%3, %5 ; %3: %3>>1+%2 + psub%1 m%2, %4 ; %2: %2>>1-%3 +%endif +%endmacro + +%macro DCT4_1D 5 +%ifnum %5 + SUMSUB_BADC w, %4, %1, %3, %2, %5 + SUMSUB_BA w, %3, %4, %5 + SUMSUB2_AB w, %1, %2, %5 + SWAP %1, %3, %4, %5, %2 +%else + SUMSUB_BADC w, %4, %1, %3, %2 + SUMSUB_BA w, %3, %4 + mova [%5], m%2 + SUMSUB2_AB w, %1, [%5], %2 + SWAP %1, %3, %4, %2 +%endif +%endmacro + +%macro IDCT4_1D 6-7 +%ifnum %6 + SUMSUBD2_AB %1, %3, %5, %7, %6 + ; %3: %3>>1-%5 %5: %3+%5>>1 + SUMSUB_BA %1, %4, %2, %7 + ; %4: %2+%4 %2: %2-%4 + SUMSUB_BADC %1, %5, %4, %3, %2, %7 + ; %5: %2+%4 + (%3+%5>>1) + ; %4: %2+%4 - (%3+%5>>1) + ; %3: %2-%4 + (%3>>1-%5) + ; %2: %2-%4 - (%3>>1-%5) +%else +%ifidn %1, w + SUMSUBD2_AB %1, %3, %5, [%6], [%6+16] +%else + SUMSUBD2_AB %1, %3, %5, [%6], [%6+32] +%endif + SUMSUB_BA %1, %4, %2 + SUMSUB_BADC %1, %5, %4, %3, %2 +%endif + SWAP %2, %5, %4 + ; %2: %2+%4 + (%3+%5>>1) row0 + ; %3: %2-%4 + (%3>>1-%5) row1 + ; %4: %2-%4 - (%3>>1-%5) row2 + ; %5: %2+%4 - (%3+%5>>1) row3 +%endmacro + + +%macro LOAD_DIFF 5 +%ifidn %3, none + movh %1, %4 + movh %2, %5 + punpcklbw %1, %2 + punpcklbw %2, %2 + psubw %1, %2 +%else + movh %1, %4 + punpcklbw %1, %3 + movh %2, %5 + punpcklbw %2, %3 + psubw %1, %2 +%endif +%endmacro + +%macro STORE_DCT 6 + movq [%5+%6+ 0], m%1 + movq [%5+%6+ 8], m%2 + movq [%5+%6+16], m%3 + movq [%5+%6+24], m%4 + movhps [%5+%6+32], m%1 + movhps [%5+%6+40], m%2 + movhps [%5+%6+48], m%3 + movhps [%5+%6+56], m%4 +%endmacro + +%macro LOAD_DIFF_8x4P 7-10 r0,r2,0 ; 4x dest, 2x temp, 2x pointer, increment? + LOAD_DIFF m%1, m%5, m%7, [%8], [%9] + LOAD_DIFF m%2, m%6, m%7, [%8+r1], [%9+r3] + LOAD_DIFF m%3, m%5, m%7, [%8+2*r1], [%9+2*r3] + LOAD_DIFF m%4, m%6, m%7, [%8+r4], [%9+r5] +%if %10 + lea %8, [%8+4*r1] + lea %9, [%9+4*r3] +%endif +%endmacro + +%macro DIFFx2 6-7 + movh %3, %5 + punpcklbw %3, %4 + psraw %1, 6 + paddsw %1, %3 + movh %3, %6 + punpcklbw %3, %4 + psraw %2, 6 + paddsw %2, %3 + packuswb %2, %1 +%endmacro + +%macro STORE_DIFF 4 + movh %2, %4 + punpcklbw %2, %3 + psraw %1, 6 + paddsw %1, %2 + packuswb %1, %1 + movh %4, %1 +%endmacro + +%macro STORE_DIFFx2 8 ; add1, add2, reg1, reg2, zero, shift, source, stride + movh %3, [%7] + movh %4, [%7+%8] + psraw %1, %6 + psraw %2, %6 + punpcklbw %3, %5 + punpcklbw %4, %5 + paddw %3, %1 + paddw %4, %2 + packuswb %3, %5 + packuswb %4, %5 + movh [%7], %3 + movh [%7+%8], %4 +%endmacro + +%macro PMINUB 3 ; dst, src, ignored +%if cpuflag(mmxext) + pminub %1, %2 +%else ; dst, src, tmp + mova %3, %1 + psubusb %3, %2 + psubb %1, %3 +%endif +%endmacro + +%macro SPLATW 2-3 0 +%if cpuflag(avx2) && %3 == 0 + vpbroadcastw %1, %2 +%elif mmsize == 16 + pshuflw %1, %2, (%3)*0x55 + punpcklqdq %1, %1 +%elif cpuflag(mmxext) + pshufw %1, %2, (%3)*0x55 +%else + %ifnidn %1, %2 + mova %1, %2 + %endif + %if %3 & 2 + punpckhwd %1, %1 + %else + punpcklwd %1, %1 + %endif + %if %3 & 1 + punpckhwd %1, %1 + %else + punpcklwd %1, %1 + %endif +%endif +%endmacro + +%macro SPLATD 1 +%if mmsize == 8 + punpckldq %1, %1 +%elif cpuflag(sse2) + pshufd %1, %1, 0 +%elif cpuflag(sse) + shufps %1, %1, 0 +%endif +%endmacro + +%macro CLIPUB 3 ;(dst, min, max) + pmaxub %1, %2 + pminub %1, %3 +%endmacro + +%macro CLIPW 3 ;(dst, min, max) + pmaxsw %1, %2 + pminsw %1, %3 +%endmacro + +%macro PMINSD 3 ; dst, src, tmp/unused +%if cpuflag(sse4) + pminsd %1, %2 +%elif cpuflag(sse2) + cvtdq2ps %1, %1 + minps %1, %2 + cvtps2dq %1, %1 +%else + mova %3, %2 + pcmpgtd %3, %1 + pxor %1, %2 + pand %1, %3 + pxor %1, %2 +%endif +%endmacro + +%macro PMAXSD 3 ; dst, src, tmp/unused +%if cpuflag(sse4) + pmaxsd %1, %2 +%else + mova %3, %1 + pcmpgtd %3, %2 + pand %1, %3 + pandn %3, %2 + por %1, %3 +%endif +%endmacro + +%macro CLIPD 3-4 +%if cpuflag(sse4); src/dst, min, max, unused + pminsd %1, %3 + pmaxsd %1, %2 +%elif cpuflag(sse2) ; src/dst, min (float), max (float), unused + cvtdq2ps %1, %1 + minps %1, %3 + maxps %1, %2 + cvtps2dq %1, %1 +%else ; src/dst, min, max, tmp + PMINSD %1, %3, %4 + PMAXSD %1, %2, %4 +%endif +%endmacro + +%macro VBROADCASTSS 2 ; dst xmm/ymm, src m32/xmm +%if cpuflag(avx2) + vbroadcastss %1, %2 +%elif cpuflag(avx) + %ifnum sizeof%2 ; avx1 register + shufps xmm%1, xmm%2, xmm%2, q0000 + %if sizeof%1 >= 32 ; mmsize>=32 + vinsertf128 %1, %1, xmm%1, 1 + %endif + %else ; avx1 memory + vbroadcastss %1, %2 + %endif +%else + %ifnum sizeof%2 ; sse register + shufps %1, %2, %2, q0000 + %else ; sse memory + movss %1, %2 + shufps %1, %1, 0 + %endif +%endif +%endmacro + +%macro VBROADCASTSD 2 ; dst xmm/ymm, src m64 +%if cpuflag(avx) && mmsize == 32 + vbroadcastsd %1, %2 +%elif cpuflag(sse3) + movddup %1, %2 +%else ; sse2 + movsd %1, %2 + movlhps %1, %1 +%endif +%endmacro + +%macro VPBROADCASTD 2 ; dst xmm/ymm, src m32/xmm +%if cpuflag(avx2) + vpbroadcastd %1, %2 +%elif cpuflag(avx) && sizeof%1 >= 32 + %error vpbroadcastd not possible with ymm on avx1. try vbroadcastss +%else + %ifnum sizeof%2 ; sse2 register + pshufd %1, %2, q0000 + %else ; sse memory + movd %1, %2 + pshufd %1, %1, 0 + %endif +%endif +%endmacro + +%macro VBROADCASTI128 2 ; dst xmm/ymm, src : 128bits val +%if mmsize > 16 + vbroadcasti128 %1, %2 +%else + mova %1, %2 +%endif +%endmacro + +%macro SHUFFLE_MASK_W 8 + %rep 8 + %if %1>=0x80 + db %1, %1 + %else + db %1*2 + db %1*2+1 + %endif + %rotate 1 + %endrep +%endmacro + +%macro PMOVSXWD 2; dst, src +%if cpuflag(sse4) + pmovsxwd %1, %2 +%else + %ifnidn %1, %2 + mova %1, %2 + %endif + punpcklwd %1, %1 + psrad %1, 16 +%endif +%endmacro + +; Wrapper for non-FMA version of fmaddps +%macro FMULADD_PS 5 + %if cpuflag(fma3) || cpuflag(fma4) + fmaddps %1, %2, %3, %4 + %elifidn %1, %4 + mulps %5, %2, %3 + addps %1, %4, %5 + %else + mulps %1, %2, %3 + addps %1, %4 + %endif +%endmacro + +%macro LSHIFT 2 +%if mmsize > 8 + pslldq %1, %2 +%else + psllq %1, 8*(%2) +%endif +%endmacro + +%macro RSHIFT 2 +%if mmsize > 8 + psrldq %1, %2 +%else + psrlq %1, 8*(%2) +%endif +%endmacro + +%macro MOVHL 2 ; dst, src +%ifidn %1, %2 + punpckhqdq %1, %2 +%elif cpuflag(avx) + punpckhqdq %1, %2, %2 +%elif cpuflag(sse4) + pshufd %1, %2, q3232 ; pshufd is slow on some older CPUs, so only use it on more modern ones +%else + movhlps %1, %2 ; may cause an int/float domain transition and has a dependency on dst +%endif +%endmacro + +; Horizontal Sum of Packed Single precision floats +; The resulting sum is in all elements. +%macro HSUMPS 2 ; dst/src, tmp +%if cpuflag(avx) + %if sizeof%1>=32 ; avx + vperm2f128 %2, %1, %1, (0)*16+(1) + addps %1, %2 + %endif + shufps %2, %1, %1, q1032 + addps %1, %2 + shufps %2, %1, %1, q0321 + addps %1, %2 +%else ; this form is a bit faster than the short avx-like emulation. + movaps %2, %1 + shufps %1, %1, q1032 + addps %1, %2 + movaps %2, %1 + shufps %1, %1, q0321 + addps %1, %2 + ; all %1 members should be equal for as long as float a+b==b+a +%endif +%endmacro + +; Emulate blendvps if not available +; +; src_b is destroyed when using emulation with logical operands +; SSE41 blendv instruction is hard coded to use xmm0 as mask +%macro BLENDVPS 3 ; dst/src_a, src_b, mask +%if cpuflag(avx) + blendvps %1, %1, %2, %3 +%elif cpuflag(sse4) + %ifnidn %3,xmm0 + %error sse41 blendvps uses xmm0 as default 3d operand, you used %3 + %endif + blendvps %1, %2, %3 +%else + xorps %2, %1 + andps %2, %3 + xorps %1, %2 +%endif +%endmacro + +; Emulate pblendvb if not available +; +; src_b is destroyed when using emulation with logical operands +; SSE41 blendv instruction is hard coded to use xmm0 as mask +%macro PBLENDVB 3 ; dst/src_a, src_b, mask +%if cpuflag(avx) + %if cpuflag(avx) && notcpuflag(avx2) && sizeof%1 >= 32 + %error pblendb not possible with ymm on avx1, try blendvps. + %endif + pblendvb %1, %1, %2, %3 +%elif cpuflag(sse4) + %ifnidn %3,xmm0 + %error sse41 pblendvd uses xmm0 as default 3d operand, you used %3 + %endif + pblendvb %1, %2, %3 +%else + pxor %2, %1 + pand %2, %3 + pxor %1, %2 +%endif +%endmacro |