diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 18:45:59 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 18:45:59 +0000 |
commit | 19fcec84d8d7d21e796c7624e521b60d28ee21ed (patch) | |
tree | 42d26aa27d1e3f7c0b8bd3fd14e7d7082f5008dc /src/crypto/isa-l/isa-l_crypto/include/memcpy_inline.h | |
parent | Initial commit. (diff) | |
download | ceph-upstream.tar.xz ceph-upstream.zip |
Adding upstream version 16.2.11+ds.upstream/16.2.11+dsupstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
-rw-r--r-- | src/crypto/isa-l/isa-l_crypto/include/memcpy_inline.h | 363 |
1 files changed, 363 insertions, 0 deletions
diff --git a/src/crypto/isa-l/isa-l_crypto/include/memcpy_inline.h b/src/crypto/isa-l/isa-l_crypto/include/memcpy_inline.h new file mode 100644 index 000000000..b15da3cb1 --- /dev/null +++ b/src/crypto/isa-l/isa-l_crypto/include/memcpy_inline.h @@ -0,0 +1,363 @@ +/********************************************************************** + Copyright(c) 2011-2016 Intel Corporation All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * 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. + * Neither the name of Intel Corporation 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 COPYRIGHT HOLDERS AND CONTRIBUTORS + "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 COPYRIGHT + OWNER OR CONTRIBUTORS 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. +**********************************************************************/ + + +/** + * @file memcpy_inline.h + * @brief Defines intrinsic memcpy functions used by the new hashing API + * + */ + +#ifndef _MEMCPY_H_ +#define _MEMCPY_H_ + +#include "intrinreg.h" +#include <assert.h> + +#ifdef __cplusplus +extern "C" { +#endif + +#define memcpy_varlen memcpy_sse_varlen +#define memcpy_fixedlen memcpy_sse_fixedlen + +#define memclr_varlen memclr_sse_varlen +#define memclr_fixedlen memclr_sse_fixedlen + +static inline void memcpy_lte32_sse_fixedlen(void* dst, const void* src, size_t nbytes); +static inline void memcpy_gte16_sse_fixedlen(void* dst, const void* src, size_t nbytes); +static inline void memcpy_sse_fixedlen (void* dst, const void* src, size_t nbytes); + +static inline void memcpy_lte32_sse_varlen (void* dst, const void* src, size_t nbytes); +static inline void memcpy_gte16_sse_varlen (void* dst, const void* src, size_t nbytes); +static inline void memcpy_sse_varlen (void* dst, const void* src, size_t nbytes); + + +static inline void memclr_lte32_sse_fixedlen(void* dst, size_t nbytes); +static inline void memclr_gte16_sse_fixedlen(void* dst, size_t nbytes); +static inline void memclr_sse_fixedlen (void* dst, size_t nbytes); + +static inline void memclr_lte32_sse_varlen (void* dst, size_t nbytes); +static inline void memclr_gte16_sse_varlen (void* dst, size_t nbytes); +static inline void memclr_sse_varlen (void* dst, size_t nbytes); + +#define MEMCPY_BETWEEN_N_AND_2N_BYTES(N, fixedwidth, dst, src, nbytes) \ + do { \ + intrinreg##N head; \ + intrinreg##N tail; \ + assert(N <= nbytes && nbytes <= 2*N); \ + if(N == 1 || (fixedwidth && nbytes==N) ) { \ + head = load_intrinreg##N(src); \ + store_intrinreg##N(dst, head); \ + } \ + else { \ + head = load_intrinreg##N(src); \ + tail = load_intrinreg##N((const void*)((const char*)src + (nbytes - N))); \ + store_intrinreg##N(dst, head); \ + store_intrinreg##N((void*)((char*)dst + (nbytes - N)), tail); \ + } \ + } while(0) + +#define MEMCLR_BETWEEN_N_AND_2N_BYTES(N, fixedwidth, dst, nbytes) \ + do { \ + const intrinreg##N zero = {0}; \ + assert(N <= nbytes && nbytes <= 2*N); \ + if(N == 1 || (fixedwidth && nbytes==N) ) { \ + store_intrinreg##N(dst, zero); \ + } \ + else { \ + store_intrinreg##N(dst, zero); \ + store_intrinreg##N((void*)((char*)dst + (nbytes - N)), zero); \ + } \ + } while(0) + +// Define load/store functions uniformly. + +#define load_intrinreg16(src) _mm_loadu_ps((const float*) src) +#define store_intrinreg16(dst,val) _mm_storeu_ps((float*) dst, val) + +static inline intrinreg8 load_intrinreg8(const void *src) +{ + return *(intrinreg8 *) src; +} + +static inline void store_intrinreg8(void *dst, intrinreg8 val) +{ + *(intrinreg8 *) dst = val; +} + +static inline intrinreg4 load_intrinreg4(const void *src) +{ + return *(intrinreg4 *) src; +} + +static inline void store_intrinreg4(void *dst, intrinreg4 val) +{ + *(intrinreg4 *) dst = val; +} + +static inline intrinreg2 load_intrinreg2(const void *src) +{ + return *(intrinreg2 *) src; +} + +static inline void store_intrinreg2(void *dst, intrinreg2 val) +{ + *(intrinreg2 *) dst = val; +} + +static inline intrinreg1 load_intrinreg1(const void *src) +{ + return *(intrinreg1 *) src; +} + +static inline void store_intrinreg1(void *dst, intrinreg1 val) +{ + *(intrinreg1 *) dst = val; +} + +static inline void memcpy_gte16_sse_fixedlen(void *dst, const void *src, size_t nbytes) +{ + size_t i; + size_t j; + intrinreg16 pool[4]; + size_t remaining_moves; + size_t tail_offset; + int do_tail; + assert(nbytes >= 16); + + for (i = 0; i + 16 * 4 <= nbytes; i += 16 * 4) { + for (j = 0; j < 4; j++) + pool[j] = + load_intrinreg16((const void *)((const char *)src + i + 16 * j)); + for (j = 0; j < 4; j++) + store_intrinreg16((void *)((char *)dst + i + 16 * j), pool[j]); + } + + remaining_moves = (nbytes - i) / 16; + tail_offset = nbytes - 16; + do_tail = (tail_offset & (16 - 1)); + + for (j = 0; j < remaining_moves; j++) + pool[j] = load_intrinreg16((const void *)((const char *)src + i + 16 * j)); + + if (do_tail) + pool[j] = load_intrinreg16((const void *)((const char *)src + tail_offset)); + + for (j = 0; j < remaining_moves; j++) + store_intrinreg16((void *)((char *)dst + i + 16 * j), pool[j]); + + if (do_tail) + store_intrinreg16((void *)((char *)dst + tail_offset), pool[j]); +} + +static inline void memclr_gte16_sse_fixedlen(void *dst, size_t nbytes) +{ + size_t i; + size_t j; + const intrinreg16 zero = { 0 }; + size_t remaining_moves; + size_t tail_offset; + int do_tail; + assert(nbytes >= 16); + + for (i = 0; i + 16 * 4 <= nbytes; i += 16 * 4) + for (j = 0; j < 4; j++) + store_intrinreg16((void *)((char *)dst + i + 16 * j), zero); + + remaining_moves = (nbytes - i) / 16; + tail_offset = nbytes - 16; + do_tail = (tail_offset & (16 - 1)); + + for (j = 0; j < remaining_moves; j++) + store_intrinreg16((void *)((char *)dst + i + 16 * j), zero); + + if (do_tail) + store_intrinreg16((void *)((char *)dst + tail_offset), zero); +} + +static inline void memcpy_lte32_sse_fixedlen(void *dst, const void *src, size_t nbytes) +{ + assert(nbytes <= 32); + if (nbytes >= 16) + MEMCPY_BETWEEN_N_AND_2N_BYTES(16, 1, dst, src, nbytes); + else if (nbytes >= 8) + MEMCPY_BETWEEN_N_AND_2N_BYTES(8, 1, dst, src, nbytes); + else if (nbytes >= 4) + MEMCPY_BETWEEN_N_AND_2N_BYTES(4, 1, dst, src, nbytes); + else if (nbytes >= 2) + MEMCPY_BETWEEN_N_AND_2N_BYTES(2, 1, dst, src, nbytes); + else if (nbytes >= 1) + MEMCPY_BETWEEN_N_AND_2N_BYTES(1, 1, dst, src, nbytes); +} + +static inline void memclr_lte32_sse_fixedlen(void *dst, size_t nbytes) +{ + assert(nbytes <= 32); + if (nbytes >= 16) + MEMCLR_BETWEEN_N_AND_2N_BYTES(16, 1, dst, nbytes); + else if (nbytes >= 8) + MEMCLR_BETWEEN_N_AND_2N_BYTES(8, 1, dst, nbytes); + else if (nbytes >= 4) + MEMCLR_BETWEEN_N_AND_2N_BYTES(4, 1, dst, nbytes); + else if (nbytes >= 2) + MEMCLR_BETWEEN_N_AND_2N_BYTES(2, 1, dst, nbytes); + else if (nbytes >= 1) + MEMCLR_BETWEEN_N_AND_2N_BYTES(1, 1, dst, nbytes); +} + +static inline void memcpy_lte32_sse_varlen(void *dst, const void *src, size_t nbytes) +{ + assert(nbytes <= 32); + if (nbytes >= 16) + MEMCPY_BETWEEN_N_AND_2N_BYTES(16, 0, dst, src, nbytes); + else if (nbytes >= 8) + MEMCPY_BETWEEN_N_AND_2N_BYTES(8, 0, dst, src, nbytes); + else if (nbytes >= 4) + MEMCPY_BETWEEN_N_AND_2N_BYTES(4, 0, dst, src, nbytes); + else if (nbytes >= 2) + MEMCPY_BETWEEN_N_AND_2N_BYTES(2, 0, dst, src, nbytes); + else if (nbytes >= 1) + MEMCPY_BETWEEN_N_AND_2N_BYTES(1, 0, dst, src, nbytes); +} + +static inline void memclr_lte32_sse_varlen(void *dst, size_t nbytes) +{ + assert(nbytes <= 32); + if (nbytes >= 16) + MEMCLR_BETWEEN_N_AND_2N_BYTES(16, 0, dst, nbytes); + else if (nbytes >= 8) + MEMCLR_BETWEEN_N_AND_2N_BYTES(8, 0, dst, nbytes); + else if (nbytes >= 4) + MEMCLR_BETWEEN_N_AND_2N_BYTES(4, 0, dst, nbytes); + else if (nbytes >= 2) + MEMCLR_BETWEEN_N_AND_2N_BYTES(2, 0, dst, nbytes); + else if (nbytes >= 1) + MEMCLR_BETWEEN_N_AND_2N_BYTES(1, 0, dst, nbytes); +} + +static inline void memcpy_gte16_sse_varlen(void *dst, const void *src, size_t nbytes) +{ + size_t i = 0; + intrinreg16 tail; + + assert(nbytes >= 16); + + while (i + 128 <= nbytes) { + memcpy_gte16_sse_fixedlen((void *)((char *)dst + i), + (const void *)((const char *)src + i), 128); + i += 128; + } + if (i + 64 <= nbytes) { + memcpy_gte16_sse_fixedlen((void *)((char *)dst + i), + (const void *)((const char *)src + i), 64); + i += 64; + } + if (i + 32 <= nbytes) { + memcpy_gte16_sse_fixedlen((void *)((char *)dst + i), + (const void *)((const char *)src + i), 32); + i += 32; + } + if (i + 16 <= nbytes) { + memcpy_gte16_sse_fixedlen((void *)((char *)dst + i), + (const void *)((const char *)src + i), 16); + i += 16; + } + + i = nbytes - 16; + tail = load_intrinreg16((const void *)((const char *)src + i)); + store_intrinreg16((void *)((char *)dst + i), tail); +} + +static inline void memclr_gte16_sse_varlen(void *dst, size_t nbytes) +{ + size_t i = 0; + const intrinreg16 zero = { 0 }; + + assert(nbytes >= 16); + + while (i + 128 <= nbytes) { + memclr_gte16_sse_fixedlen((void *)((char *)dst + i), 128); + i += 128; + } + if (i + 64 <= nbytes) { + memclr_gte16_sse_fixedlen((void *)((char *)dst + i), 64); + i += 64; + } + if (i + 32 <= nbytes) { + memclr_gte16_sse_fixedlen((void *)((char *)dst + i), 32); + i += 32; + } + if (i + 16 <= nbytes) { + memclr_gte16_sse_fixedlen((void *)((char *)dst + i), 16); + i += 16; + } + + i = nbytes - 16; + store_intrinreg16((void *)((char *)dst + i), zero); +} + +static inline void memcpy_sse_fixedlen(void *dst, const void *src, size_t nbytes) +{ + if (nbytes >= 16) + memcpy_gte16_sse_fixedlen(dst, src, nbytes); + else + memcpy_lte32_sse_fixedlen(dst, src, nbytes); +} + +static inline void memclr_sse_fixedlen(void *dst, size_t nbytes) +{ + if (nbytes >= 16) + memclr_gte16_sse_fixedlen(dst, nbytes); + else + memclr_lte32_sse_fixedlen(dst, nbytes); +} + +static inline void memcpy_sse_varlen(void *dst, const void *src, size_t nbytes) +{ + if (nbytes >= 16) + memcpy_gte16_sse_varlen(dst, src, nbytes); + else + memcpy_lte32_sse_varlen(dst, src, nbytes); +} + +static inline void memclr_sse_varlen(void *dst, size_t nbytes) +{ + if (nbytes >= 16) + memclr_gte16_sse_varlen(dst, nbytes); + else + memclr_lte32_sse_varlen(dst, nbytes); +} + + +#ifdef __cplusplus +} +#endif + +#endif // __MEMCPY_H |