diff options
Diffstat (limited to 'src/isa-l/crc')
28 files changed, 9098 insertions, 0 deletions
diff --git a/src/isa-l/crc/Makefile.am b/src/isa-l/crc/Makefile.am new file mode 100644 index 00000000..74b96474 --- /dev/null +++ b/src/isa-l/crc/Makefile.am @@ -0,0 +1,64 @@ +######################################################################## +# Copyright(c) 2011-2017 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. +######################################################################## + +lsrc += \ + crc/crc_base.c \ + crc/crc64_base.c + +lsrc_base_aliases += crc/crc_base_aliases.c +lsrc_x86_32 += crc/crc_base_aliases.c + +lsrc_x86_64 += \ + crc/crc16_t10dif_01.asm \ + crc/crc16_t10dif_by4.asm \ + crc/crc32_ieee_01.asm \ + crc/crc32_ieee_by4.asm \ + crc/crc32_iscsi_01.asm \ + crc/crc32_iscsi_00.asm \ + crc/crc_multibinary.asm \ + crc/crc64_multibinary.asm \ + crc/crc64_ecma_refl_by8.asm \ + crc/crc64_ecma_norm_by8.asm \ + crc/crc64_iso_refl_by8.asm \ + crc/crc64_iso_norm_by8.asm \ + crc/crc64_jones_refl_by8.asm \ + crc/crc64_jones_norm_by8.asm + +src_include += -I $(srcdir)/crc +extern_hdrs += include/crc.h include/crc64.h + +other_src += include/reg_sizes.asm include/types.h include/test.h + +check_tests += crc/crc16_t10dif_test crc/crc32_ieee_test crc/crc32_iscsi_test \ + crc/crc64_funcs_test + +perf_tests += crc/crc16_t10dif_perf crc/crc32_ieee_perf crc/crc32_iscsi_perf \ + crc/crc64_funcs_perf + +examples += crc/crc_simple_test crc/crc64_example diff --git a/src/isa-l/crc/crc16_t10dif_01.asm b/src/isa-l/crc/crc16_t10dif_01.asm new file mode 100644 index 00000000..fb9c73c8 --- /dev/null +++ b/src/isa-l/crc/crc16_t10dif_01.asm @@ -0,0 +1,665 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function API: +; UINT16 crc16_t10dif_01( +; UINT16 init_crc, //initial CRC value, 16 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; UINT64 len //buffer length in bytes (64-bit data) +; ); +; +; Authors: +; Erdinc Ozturk +; Vinodh Gopal +; James Guilford +; +; Reference paper titled "Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction" +; URL: http://www.intel.com/content/dam/www/public/us/en/documents/white-papers/fast-crc-computation-generic-polynomials-pclmulqdq-paper.pdf + +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 + + %xdefine arg1_low32 ecx +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx + + %xdefine arg1_low32 edi +%endif + +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif + +align 16 +global crc16_t10dif_01:function +crc16_t10dif_01: + + ; adjust the 16-bit initial_crc value, scale it to 32 bits + shl arg1_low32, 16 + + ; After this point, code flow is exactly same as a 32-bit CRC. + ; The only difference is before returning eax, we will shift it right 16 bits, to scale back to 16 bits. + + sub rsp, VARIABLE_OFFSET +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp+16*2],xmm6 + movdqa [rsp+16*3],xmm7 + movdqa [rsp+16*4],xmm8 + movdqa [rsp+16*5],xmm9 + movdqa [rsp+16*6],xmm10 + movdqa [rsp+16*7],xmm11 + movdqa [rsp+16*8],xmm12 + movdqa [rsp+16*9],xmm13 +%endif + + ; check if smaller than 256 + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movd xmm10, arg1_low32 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register. + ; because data will be byte-reflected and will align with initial crc at correct place. + pslldq xmm10, 12 + + movdqa xmm11, [SHUF_MASK] + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + pshufb xmm0, xmm11 + ; XOR the initial_crc value + pxor xmm0, xmm10 + pshufb xmm1, xmm11 + pshufb xmm2, xmm11 + pshufb xmm3, xmm11 + pshufb xmm4, xmm11 + pshufb xmm5, xmm11 + pshufb xmm6, xmm11 + pshufb xmm7, xmm11 + + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 ; buf += 128; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm1, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm3, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm5, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm7, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer + ; fold the 8 xmm registers to 1 xmm register with different constants + + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + + movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2 + ;imm value of pclmulqdq instruction will determine which constant to use + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pshufb xmm0, xmm11 + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm7 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm11 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm7 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm7, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10 + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0x1 + pslldq xmm0, 8 + pxor xmm7, xmm0 + + ;32b fold + movdqa xmm0, xmm7 + + pand xmm0, [mask2] + + psrldq xmm7, 12 + pclmulqdq xmm7, xmm10, 0x10 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10 + movdqa xmm0, xmm7 + pclmulqdq xmm7, xmm10, 0x01 + pslldq xmm7, 4 + pclmulqdq xmm7, xmm10, 0x11 + + pslldq xmm7, 4 + pxor xmm7, xmm0 + pextrd eax, xmm7,1 + +_cleanup: + ; scale the result back to 16 bits + shr eax, 16 +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp+16*2] + movdqa xmm7, [rsp+16*3] + movdqa xmm8, [rsp+16*4] + movdqa xmm9, [rsp+16*5] + movdqa xmm10, [rsp+16*6] + movdqa xmm11, [rsp+16*7] + movdqa xmm12, [rsp+16*8] + movdqa xmm13, [rsp+16*9] +%endif + add rsp, VARIABLE_OFFSET + ret + + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm11, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov eax, arg1_low32 + test arg3, arg3 + je _cleanup + + movdqa xmm11, [SHUF_MASK] + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + cmp arg3, 4 + jl _only_less_than_4 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm7, xmm0 + jmp _128_done + +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +_only_less_than_4: + cmp arg3, 3 + jl _only_less_than_3 + + ; load 3 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + mov al, [arg2+2] + mov [r11+2], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 5 + + jmp _barrett +_only_less_than_3: + cmp arg3, 2 + jl _only_less_than_2 + + ; load 2 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 6 + + jmp _barrett +_only_less_than_2: + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 7 + + jmp _barrett + +section .data + +; precomputed constants +; these constants are precomputed from the poly: 0x8bb70000 (0x8bb7 scaled to 32 bits) +align 16 +; Q = 0x18BB70000 +; rk1 = 2^(32*3) mod Q << 32 +; rk2 = 2^(32*5) mod Q << 32 +; rk3 = 2^(32*15) mod Q << 32 +; rk4 = 2^(32*17) mod Q << 32 +; rk5 = 2^(32*3) mod Q << 32 +; rk6 = 2^(32*2) mod Q << 32 +; rk7 = floor(2^64/Q) +; rk8 = Q +rk1: +DQ 0x2d56000000000000 +rk2: +DQ 0x06df000000000000 +rk3: +DQ 0x9d9d000000000000 +rk4: +DQ 0x7cf5000000000000 +rk5: +DQ 0x2d56000000000000 +rk6: +DQ 0x1368000000000000 +rk7: +DQ 0x00000001f65a57f8 +rk8: +DQ 0x000000018bb70000 + +rk9: +DQ 0xceae000000000000 +rk10: +DQ 0xbfd6000000000000 +rk11: +DQ 0x1e16000000000000 +rk12: +DQ 0x713c000000000000 +rk13: +DQ 0xf7f9000000000000 +rk14: +DQ 0x80a6000000000000 +rk15: +DQ 0x044c000000000000 +rk16: +DQ 0xe658000000000000 +rk17: +DQ 0xad18000000000000 +rk18: +DQ 0xa497000000000000 +rk19: +DQ 0x6ee3000000000000 +rk20: +DQ 0xe7b5000000000000 + + + + + + + + + +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + +;;; func core, ver, snum +slversion crc16_t10dif_01, 01, 06, 0010 + diff --git a/src/isa-l/crc/crc16_t10dif_by4.asm b/src/isa-l/crc/crc16_t10dif_by4.asm new file mode 100644 index 00000000..f79f4f6d --- /dev/null +++ b/src/isa-l/crc/crc16_t10dif_by4.asm @@ -0,0 +1,562 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; +; Function API: +; UINT16 crc16_t10dif_by4( +; UINT16 init_crc, //initial CRC value, 16 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; UINT64 len //buffer length in bytes (64-bit data) +; ); +; +; Authors: +; Erdinc Ozturk +; Vinodh Gopal +; James Guilford +; +; Reference paper titled "Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction" +; URL: http://download.intel.com/design/intarch/papers/323102.pdf +; + +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 + + %xdefine arg1_low32 ecx +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx + + %xdefine arg1_low32 edi +%endif + +align 16 +global crc16_t10dif_by4:function +crc16_t10dif_by4: + + ; adjust the 16-bit initial_crc value, scale it to 32 bits + shl arg1_low32, 16 + + ; After this point, code flow is exactly same as a 32-bit CRC. + ; The only difference is before returning eax, we will shift + ; it right 16 bits, to scale back to 16 bits. + + sub rsp,16*4+8 + + ; push the xmm registers into the stack to maintain + movdqa [rsp+16*2],xmm6 + movdqa [rsp+16*3],xmm7 + + ; check if smaller than 128B + cmp arg3, 128 + + ; for sizes less than 128, we can't fold 64B at a time... + jl _less_than_128 + + + ; load the initial crc value + movd xmm6, arg1_low32 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to + ; be moved to the high part of the register. + ; because data will be byte-reflected and will align with + ; initial crc at correct place. + pslldq xmm6, 12 + + movdqa xmm7, [SHUF_MASK] + ; receive the initial 64B data, xor the initial crc value + movdqu xmm0, [arg2] + movdqu xmm1, [arg2+16] + movdqu xmm2, [arg2+32] + movdqu xmm3, [arg2+48] + + pshufb xmm0, xmm7 + ; XOR the initial_crc value + pxor xmm0, xmm6 + pshufb xmm1, xmm7 + pshufb xmm2, xmm7 + pshufb xmm3, xmm7 + + movdqa xmm6, [rk3] ;xmm6 has rk3 and rk4 + ;imm value of pclmulqdq instruction + ;will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 128 instead of 64 to save one instruction from the loop + sub arg3, 128 + + ; at this section of the code, there is 64*x+y (0<=y<64) bytes of + ; buffer. The _fold_64_B_loop + ; loop will fold 64B at a time until we have 64+y Bytes of buffer + + + ; fold 64B at a time. This section of the code folds 4 xmm + ; registers in parallel +_fold_64_B_loop: + + ; update the buffer pointer + add arg2, 64 ; buf += 64; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm4, xmm0 + movdqu xmm5, xmm1 + + pclmulqdq xmm0, xmm6 , 0x11 + pclmulqdq xmm1, xmm6 , 0x11 + + pclmulqdq xmm4, xmm6, 0x0 + pclmulqdq xmm5, xmm6, 0x0 + + pxor xmm0, xmm4 + pxor xmm1, xmm5 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm4, xmm2 + movdqu xmm5, xmm3 + + pclmulqdq xmm2, xmm6, 0x11 + pclmulqdq xmm3, xmm6, 0x11 + + pclmulqdq xmm4, xmm6, 0x0 + pclmulqdq xmm5, xmm6, 0x0 + + pxor xmm2, xmm4 + pxor xmm3, xmm5 + + movdqu xmm4, [arg2] + movdqu xmm5, [arg2+16] + pshufb xmm4, xmm7 + pshufb xmm5, xmm7 + pxor xmm0, xmm4 + pxor xmm1, xmm5 + + movdqu xmm4, [arg2+32] + movdqu xmm5, [arg2+48] + pshufb xmm4, xmm7 + pshufb xmm5, xmm7 + + pxor xmm2, xmm4 + pxor xmm3, xmm5 + + sub arg3, 64 + + ; check if there is another 64B in the buffer to be able to fold + jge _fold_64_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + + add arg2, 64 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer + ; the 64B of folded data is in 4 of the xmm registers: xmm0, xmm1, xmm2, xmm3 + + + ; fold the 4 xmm registers to 1 xmm register with different constants + + movdqa xmm6, [rk1] ;xmm6 has rk1 and rk2 + ;imm value of pclmulqdq instruction will + ;determine which constant to use + + movdqa xmm4, xmm0 + pclmulqdq xmm0, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm1, xmm4 + pxor xmm1, xmm0 + + movdqa xmm4, xmm1 + pclmulqdq xmm1, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm2, xmm4 + pxor xmm2, xmm1 + + movdqa xmm4, xmm2 + pclmulqdq xmm2, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + pxor xmm3, xmm2 + + + ; instead of 64, we add 48 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 64-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes + ; is in register xmm3 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm4, xmm3 + pclmulqdq xmm3, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + movdqu xmm0, [arg2] + pshufb xmm0, xmm7 + pxor xmm3, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm3 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, + ; we can offset the input pointer before the actual point, + ; to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm3 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm7 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm3 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm3, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm4, xmm3 + pclmulqdq xmm3, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + pxor xmm3, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm6, [rk5] ; rk5 and rk6 in xmm6 + movdqa xmm0, xmm3 + + ;64b fold + pclmulqdq xmm3, xmm6, 0x1 + pslldq xmm0, 8 + pxor xmm3, xmm0 + + ;32b fold + movdqa xmm0, xmm3 + + pand xmm0, [mask2] + + psrldq xmm3, 12 + pclmulqdq xmm3, xmm6, 0x10 + pxor xmm3, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm6, [rk7] ; rk7 and rk8 in xmm6 + movdqa xmm0, xmm3 + pclmulqdq xmm3, xmm6, 0x01 + pslldq xmm3, 4 + pclmulqdq xmm3, xmm6, 0x11 + + pslldq xmm3, 4 + pxor xmm3, xmm0 + pextrd eax, xmm3,1 + +_cleanup: + ; scale the result back to 16 bits + shr eax, 16 + movdqa xmm6, [rsp+16*2] + movdqa xmm7, [rsp+16*3] + add rsp,16*4+8 + ret + + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_128: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm7, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm6, [rk1] ; rk1 and rk2 in xmm6 + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + movdqu xmm3, [arg2] ; load the plaintext + pshufb xmm3, xmm7 ; byte-reflect the plaintext + pxor xmm3, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov eax, arg1_low32 + test arg3, arg3 + je _cleanup + + movdqa xmm7, [SHUF_MASK] + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm3, [arg2] ; load the plaintext + pshufb xmm3, xmm7 ; byte-reflect the plaintext + pxor xmm3, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm6, [rk1] ; rk1 and rk2 in xmm6 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + cmp arg3, 4 + jl _only_less_than_4 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 ; xor the initial crc value + + ; shl r9, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm3, xmm0 + jmp _128_done + +align 16 +_exact_16_left: + movdqu xmm3, [arg2] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 ; xor the initial crc value + + jmp _128_done + +_only_less_than_4: + cmp arg3, 3 + jl _only_less_than_3 + + ; load 3 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + mov al, [arg2+2] + mov [r11+2], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 ; xor the initial crc value + + psrldq xmm3, 5 + + jmp _barrett +_only_less_than_3: + cmp arg3, 2 + jl _only_less_than_2 + + ; load 2 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 ; xor the initial crc value + + psrldq xmm3, 6 + + jmp _barrett +_only_less_than_2: + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 ; xor the initial crc value + + psrldq xmm3, 7 + + jmp _barrett + +section .data + +; precomputed constants +; these constants are precomputed from the poly: 0x8bb70000 (0x8bb7 scaled to 32 bits) +align 16 +; Q = 0x18BB70000 +; rk1 = 2^(32*3) mod Q << 32 +; rk2 = 2^(32*5) mod Q << 32 +; rk3 = 2^(32*15) mod Q << 32 +; rk4 = 2^(32*17) mod Q << 32 +; rk5 = 2^(32*3) mod Q << 32 +; rk6 = 2^(32*2) mod Q << 32 +; rk7 = floor(2^64/Q) +; rk8 = Q +rk1: +DQ 0x2d56000000000000 +rk2: +DQ 0x06df000000000000 +rk3: +DQ 0x044c000000000000 +rk4: +DQ 0xe658000000000000 +rk5: +DQ 0x2d56000000000000 +rk6: +DQ 0x1368000000000000 +rk7: +DQ 0x00000001f65a57f8 +rk8: +DQ 0x000000018bb70000 +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + +;;; func core, ver, snum +slversion crc16_t10dif_by4, 05, 02, 0016 diff --git a/src/isa-l/crc/crc16_t10dif_perf.c b/src/isa-l/crc/crc16_t10dif_perf.c new file mode 100644 index 00000000..34f1ddbd --- /dev/null +++ b/src/isa-l/crc/crc16_t10dif_perf.c @@ -0,0 +1,87 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <sys/time.h> +#include "crc.h" +#include "test.h" + +//#define CACHED_TEST +#ifdef CACHED_TEST +// Cached test, loop many times over small dataset +# define TEST_LEN 8*1024 +# define TEST_LOOPS 400000 +# define TEST_TYPE_STR "_warm" +#else +// Uncached test. Pull from large mem base. +# define GT_L3_CACHE 32*1024*1024 /* some number > last level cache */ +# define TEST_LEN (2 * GT_L3_CACHE) +# define TEST_LOOPS 100 +# define TEST_TYPE_STR "_cold" +#endif + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define TEST_MEM TEST_LEN + +int main(int argc, char *argv[]) +{ + int i; + void *buf; + uint16_t crc; + struct perf start, stop; + + printf("crc16_t10dif_perf:\n"); + + if (posix_memalign(&buf, 1024, TEST_LEN)) { + printf("alloc error: Fail"); + return -1; + } + + printf("Start timed tests\n"); + fflush(0); + + memset(buf, 0, TEST_LEN); + crc = crc16_t10dif(TEST_SEED, buf, TEST_LEN); + perf_start(&start); + for (i = 0; i < TEST_LOOPS; i++) { + crc = crc16_t10dif(TEST_SEED, buf, TEST_LEN); + } + perf_stop(&stop); + printf("crc16_t10dif" TEST_TYPE_STR ": "); + perf_print(stop, start, (long long)TEST_LEN * i); + + printf("finish 0x%x\n", crc); + return 0; +} diff --git a/src/isa-l/crc/crc16_t10dif_test.c b/src/isa-l/crc/crc16_t10dif_test.c new file mode 100644 index 00000000..e622f10a --- /dev/null +++ b/src/isa-l/crc/crc16_t10dif_test.c @@ -0,0 +1,167 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <string.h> +#include <stdint.h> +#include <stdlib.h> +#include "crc.h" +#include "types.h" + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define MAX_BUF 512 +#define TEST_SIZE 20 + +typedef uint32_t u32; +typedef uint16_t u16; +typedef uint8_t u8; + +void rand_buffer(unsigned char *buf, long buffer_size) +{ + long i; + for (i = 0; i < buffer_size; i++) + buf[i] = rand(); +} + +int main(int argc, char *argv[]) +{ + int fail = 0; + u32 r = 0; + int verbose = argc - 1; + int i, s; + void *buf_raw; + unsigned char *buf; + + printf("Test crc16_t10dif_test "); + if (posix_memalign(&buf_raw, MAX_BUF, MAX_BUF * TEST_SIZE)) { + printf("alloc error: Fail"); + return -1; + } + buf = (unsigned char *)buf_raw; + + srand(TEST_SEED); + + // Test of all zeros + memset(buf, 0, MAX_BUF * 10); + u16 crc = crc16_t10dif(TEST_SEED, buf, MAX_BUF); + u16 crc_ref = crc16_t10dif_base(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) { + fail++; + printf("\n opt ref\n"); + printf(" ------ ------\n"); + printf("crc zero = 0x%4x 0x%4x \n", crc, crc_ref); + } else + printf("."); + + // Another simple test pattern + memset(buf, 0x8a, MAX_BUF); + crc = crc16_t10dif(TEST_SEED, buf, MAX_BUF); + crc_ref = crc16_t10dif_base(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) { + fail++; + printf("crc all 8a = 0x%4x 0x%4x\n", crc, crc_ref); + } else + printf("."); + + // Do a few random tests + + rand_buffer(buf, MAX_BUF * TEST_SIZE); + + for (i = 0; i < TEST_SIZE; i++) { + crc = crc16_t10dif(TEST_SEED, buf, MAX_BUF); + crc_ref = crc16_t10dif_base(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%4x 0x%4x\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + + // Do a few random sizes + buf = (unsigned char *)buf_raw; //reset buf + r = rand(); + + for (i = MAX_BUF; i >= 0; i--) { + crc = crc16_t10dif(r, buf, i); + crc_ref = crc16_t10dif_base(r, buf, i); + if (crc != crc_ref) { + fail++; + printf("fail random size%i 0x%8x 0x%8x\n", i, crc, crc_ref); + } else + printf("."); + } + + // Try different seeds + for (s = 0; s < 20; s++) { + buf = (unsigned char *)buf_raw; //reset buf + + r = rand(); // just to get a new seed + rand_buffer(buf, MAX_BUF * TEST_SIZE); // new pseudo-rand data + + if (verbose) + printf("seed = 0x%x\n", r); + + for (i = 0; i < TEST_SIZE; i++) { + crc = crc16_t10dif(r, buf, MAX_BUF); + crc_ref = crc16_t10dif_base(r, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%4x 0x%4x\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + } + + // Run tests at end of buffer + buf = (unsigned char *)buf_raw; //reset buf + buf = buf + ((MAX_BUF - 1) * TEST_SIZE); //Line up TEST_SIZE from end + for (i = 0; i < TEST_SIZE; i++) { + crc = crc16_t10dif(TEST_SEED, buf + i, TEST_SIZE - i); + crc_ref = crc16_t10dif_base(TEST_SEED, buf + i, TEST_SIZE - i); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc eob rand%3d = 0x%4x 0x%4x\n", i, crc, crc_ref); + else + printf("."); + } + + printf("Test done: %s\n", fail ? "Fail" : "Pass"); + if (fail) + printf("\nFailed %d tests\n", fail); + + return fail; +} diff --git a/src/isa-l/crc/crc32_ieee_01.asm b/src/isa-l/crc/crc32_ieee_01.asm new file mode 100644 index 00000000..3c463dad --- /dev/null +++ b/src/isa-l/crc/crc32_ieee_01.asm @@ -0,0 +1,655 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function API: +; UINT32 crc32_ieee_01( +; UINT32 init_crc, //initial CRC value, 32 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; UINT64 len //buffer length in bytes (64-bit data) +; ); +; +; Authors: +; Erdinc Ozturk +; Vinodh Gopal +; James Guilford +; +; Reference paper titled "Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction" +; URL: http://www.intel.com/content/dam/www/public/us/en/documents/white-papers/fast-crc-computation-generic-polynomials-pclmulqdq-paper.pdf + +%include "reg_sizes.asm" + +%define fetch_dist 1024 +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 + + %xdefine arg1_low32 ecx +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx + + %xdefine arg1_low32 edi +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif +align 16 +global crc32_ieee_01:function +crc32_ieee_01: + + not arg1_low32 ;~init_crc + + sub rsp,VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + + ; check if smaller than 256 + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movd xmm10, arg1_low32 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register. + ; because data will be byte-reflected and will align with initial crc at correct place. + pslldq xmm10, 12 + + movdqa xmm11, [SHUF_MASK] + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + pshufb xmm0, xmm11 + ; XOR the initial_crc value + pxor xmm0, xmm10 + pshufb xmm1, xmm11 + pshufb xmm2, xmm11 + pshufb xmm3, xmm11 + pshufb xmm4, xmm11 + pshufb xmm5, xmm11 + pshufb xmm6, xmm11 + pshufb xmm7, xmm11 + + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 ; buf += 128; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm1, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm3, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm5, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm7, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer + ; the 128 of folded data is in 4 of the xmm registers: xmm0, xmm1, xmm2, xmm3 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + + movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2 + ;imm value of pclmulqdq instruction will determine which constant to use + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pshufb xmm0, xmm11 + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm7 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm11 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm7 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm7, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10 + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0x1 + pslldq xmm0, 8 + pxor xmm7, xmm0 + + ;32b fold + movdqa xmm0, xmm7 + + pand xmm0, [mask2] + + psrldq xmm7, 12 + pclmulqdq xmm7, xmm10, 0x10 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10 + movdqa xmm0, xmm7 + pclmulqdq xmm7, xmm10, 0x01 + pslldq xmm7, 4 + pclmulqdq xmm7, xmm10, 0x11 + + pslldq xmm7, 4 + pxor xmm7, xmm0 + pextrd eax, xmm7,1 + +_cleanup: + not eax +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp,VARIABLE_OFFSET + ret + + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm11, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov eax, arg1_low32 + test arg3, arg3 + je _cleanup + + movdqa xmm11, [SHUF_MASK] + + movd xmm0, arg1_low32 ; get the initial crc value + pslldq xmm0, 12 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + cmp arg3, 4 + jl _only_less_than_4 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + ; shl r9, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm7, xmm0 + jmp _128_done + +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +_only_less_than_4: + cmp arg3, 3 + jl _only_less_than_3 + + ; load 3 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + mov al, [arg2+2] + mov [r11+2], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 5 + + jmp _barrett +_only_less_than_3: + cmp arg3, 2 + jl _only_less_than_2 + + ; load 2 Bytes + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 6 + + jmp _barrett +_only_less_than_2: + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + psrldq xmm7, 7 + + jmp _barrett + +section .data + +; precomputed constants +align 16 + +rk1 : +DQ 0xf200aa6600000000 +rk2 : +DQ 0x17d3315d00000000 +rk3 : +DQ 0x022ffca500000000 +rk4 : +DQ 0x9d9ee22f00000000 +rk5 : +DQ 0xf200aa6600000000 +rk6 : +DQ 0x490d678d00000000 +rk7 : +DQ 0x0000000104d101df +rk8 : +DQ 0x0000000104c11db7 +rk9 : +DQ 0x6ac7e7d700000000 +rk10 : +DQ 0xfcd922af00000000 +rk11 : +DQ 0x34e45a6300000000 +rk12 : +DQ 0x8762c1f600000000 +rk13 : +DQ 0x5395a0ea00000000 +rk14 : +DQ 0x54f2d5c700000000 +rk15 : +DQ 0xd3504ec700000000 +rk16 : +DQ 0x57a8445500000000 +rk17 : +DQ 0xc053585d00000000 +rk18 : +DQ 0x766f1b7800000000 +rk19 : +DQ 0xcd8c54b500000000 +rk20 : +DQ 0xab40b71e00000000 + + + + + + + + + +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + +;;; func core, ver, snum +slversion crc32_ieee_01, 01, 06, 0011 + diff --git a/src/isa-l/crc/crc32_ieee_by4.asm b/src/isa-l/crc/crc32_ieee_by4.asm new file mode 100644 index 00000000..97f68680 --- /dev/null +++ b/src/isa-l/crc/crc32_ieee_by4.asm @@ -0,0 +1,565 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; +; Function API: +; UINT32 crc32_ieee_by4( +; UINT32 init_crc, //initial CRC value, 32 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; UINT64 len //buffer length in bytes (64-bit data) +; ); +; +; Authors: +; Erdinc Ozturk +; Vinodh Gopal +; James Guilford +; +; Reference paper titled "Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction" +; URL: http://download.intel.com/design/intarch/papers/323102.pdf +; + +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 + + %xdefine arg1_low32 ecx +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx + + %xdefine arg1_low32 edi +%endif + +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*4+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif + +align 16 +global crc32_ieee_by4:function +crc32_ieee_by4: + + not arg1_low32 + + sub rsp,VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0],xmm6 + movdqa [rsp + XMM_SAVE + 16*1],xmm7 +%endif + + ; check if smaller than 128B + cmp arg3, 128 + jl _less_than_128 + + + + ; load the initial crc value + movd xmm6, arg1_low32 ; initial crc + ; crc value does not need to be byte-reflected, but it needs to be + ; moved to the high part of the register. + ; because data will be byte-reflected and will align with initial + ; crc at correct place. + pslldq xmm6, 12 + + + + movdqa xmm7, [SHUF_MASK] + ; receive the initial 64B data, xor the initial crc value + movdqu xmm0, [arg2] + movdqu xmm1, [arg2+16] + movdqu xmm2, [arg2+32] + movdqu xmm3, [arg2+48] + + + + pshufb xmm0, xmm7 + ; XOR the initial_crc value + pxor xmm0, xmm6 + pshufb xmm1, xmm7 + pshufb xmm2, xmm7 + pshufb xmm3, xmm7 + + movdqa xmm6, [rk3] ; k3=2^480 mod POLY << 32 + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;we subtract 128 instead of 64 to save one instruction from the loop + sub arg3, 128 + + ; at this section of the code, there is 64*x+y (0<=y<64) bytes of + ; buffer. The _fold_64_B_loop loop will fold 64B at a time until we + ; have 64+y Bytes of buffer + + + ; fold 64B at a time. This section of the code folds 4 xmm registers in parallel +_fold_64_B_loop: + + ;update the buffer pointer + add arg2, 64 + + prefetchnta [arg2+fetch_dist+0] + movdqa xmm4, xmm0 + movdqa xmm5, xmm1 + + pclmulqdq xmm0, xmm6 , 0x11 + pclmulqdq xmm1, xmm6 , 0x11 + + pclmulqdq xmm4, xmm6, 0x0 + pclmulqdq xmm5, xmm6, 0x0 + + pxor xmm0, xmm4 + pxor xmm1, xmm5 + + prefetchnta [arg2+fetch_dist+32] + movdqa xmm4, xmm2 + movdqa xmm5, xmm3 + + pclmulqdq xmm2, xmm6, 0x11 + pclmulqdq xmm3, xmm6, 0x11 + + pclmulqdq xmm4, xmm6, 0x0 + pclmulqdq xmm5, xmm6, 0x0 + + pxor xmm2, xmm4 + pxor xmm3, xmm5 + + movdqu xmm4, [arg2] + movdqu xmm5, [arg2+16] + pshufb xmm4, xmm7 + pshufb xmm5, xmm7 + pxor xmm0, xmm4 + pxor xmm1, xmm5 + + movdqu xmm4, [arg2+32] + movdqu xmm5, [arg2+48] + pshufb xmm4, xmm7 + pshufb xmm5, xmm7 + + pxor xmm2, xmm4 + pxor xmm3, xmm5 + + sub arg3, 64 + + ; check if there is another 64B in the buffer to be able to fold + jge _fold_64_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + + add arg2, 64 + ;at this point, the arg2 is pointing at the last y Bytes of the buffer + ; the 64B of data is in 4 of the xmm registers: xmm0, xmm1, xmm2, xmm3 + + + movdqa xmm6, [rk1] ;k1 + + ; fold the 4 xmm registers to 1 xmm register with different constants + movdqa xmm4, xmm0 + pclmulqdq xmm0, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm1, xmm4 + xorps xmm1, xmm0 + + movdqa xmm4, xmm1 + pclmulqdq xmm1, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm2, xmm4 + xorps xmm2, xmm1 + + movdqa xmm4, xmm2 + pclmulqdq xmm2, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + pxor xmm3, xmm2 + + + ;instead of 64, we add 48 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 64-16 + jl _final_reduction_for_128 + +; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm3 and the rest is in memory +; we can fold 16 bytes at a time if y>=16 +; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm4, xmm3 + pclmulqdq xmm3, xmm6, 0x11 + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + movdqu xmm0, [arg2] + pshufb xmm0, xmm7 + pxor xmm3, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm3 register + + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset + ; the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm3 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm7 + + shl arg3, 4 + lea rax, [pshufb_shf_table + 15*16] + sub rax, arg3 + movdqu xmm0, [rax] + + pshufb xmm2, xmm0 + + pxor xmm0, [mask3] + + pshufb xmm3, xmm0 + + pblendvb xmm1, xmm2 ;xmm0 is implicit + + movdqa xmm2, xmm1 + + movdqa xmm4, xmm3 + pclmulqdq xmm3, xmm6, 0x11 + + pclmulqdq xmm4, xmm6, 0x0 + pxor xmm3, xmm4 + pxor xmm3, xmm2 + +_128_done: + + movdqa xmm6, [rk5] + movdqa xmm0, xmm3 + + ;64b fold + pclmulqdq xmm3, xmm6, 0x1 + pslldq xmm0, 8 + pxor xmm3, xmm0 + + ;32b fold + movdqa xmm0, xmm3 + + pand xmm0, [mask4] + + psrldq xmm3, 12 + pclmulqdq xmm3, xmm6, 0x10 + pxor xmm3, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm6, [rk7] + movdqa xmm0, xmm3 + pclmulqdq xmm3, xmm6, 0x01 + pslldq xmm3, 4 + pclmulqdq xmm3, xmm6, 0x11 + + pslldq xmm3, 4 + pxor xmm3, xmm0 + pextrd eax, xmm3,1 + +_cleanup: + not eax +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] +%endif + add rsp,VARIABLE_OFFSET + + + ret + + + + + + + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_128: + + ;check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm7, [SHUF_MASK] + + ;if there is, load the constants + movdqa xmm6, [rk1] ;k1 + + movd xmm0, arg1_low32 + pslldq xmm0, 12 + movdqu xmm3, [arg2] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + + ;update the buffer pointer + add arg2, 16 + + ;update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + + +align 16 +_less_than_32: + mov eax, arg1_low32 + test arg3, arg3 + je _cleanup + + movdqa xmm7, [SHUF_MASK] + + movd xmm0, arg1_low32 + pslldq xmm0, 12 + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + movd xmm0, arg1_low32 + pslldq xmm0, 12 + movdqu xmm3, [arg2] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + add arg2, 16 + sub arg3, 16 + movdqa xmm6, [rk1] ;k1 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + + + cmp arg3, 4 + jl _only_less_than_4 + + mov r9, arg3 + + + cmp arg3, 8 + jl _less_than_8_left + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + mov al, [arg2] + mov [r11], al + +_zero_left: + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + shl r9, 4 + lea rax, [pshufb_shf_table + 15*16] + sub rax, r9 + movdqu xmm0, [rax] + pxor xmm0, [mask3] + + pshufb xmm3, xmm0 + jmp _128_done + +align 16 +_exact_16_left: + movdqu xmm3, [arg2] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + jmp _128_done + +_only_less_than_4: + cmp arg3, 3 + jl _only_less_than_3 + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + mov al, [arg2+2] + mov [r11+2], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + psrldq xmm3, 5 + + jmp _barrett +_only_less_than_3: + cmp arg3, 2 + jl _only_less_than_2 + mov al, [arg2] + mov [r11], al + + mov al, [arg2+1] + mov [r11+1], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + psrldq xmm3, 6 + + jmp _barrett +_only_less_than_2: + mov al, [arg2] + mov [r11], al + + movdqa xmm3, [rsp] + pshufb xmm3, xmm7 + pxor xmm3, xmm0 + + psrldq xmm3, 7 + + jmp _barrett +; precomputed constants +section .data + +align 16 +rk1: +DQ 0xf200aa6600000000 +rk2: +DQ 0x17d3315d00000000 +rk3: +DQ 0xd3504ec700000000 +rk4: +DQ 0x57a8445500000000 +rk5: +DQ 0xf200aa6600000000 +rk6: +DQ 0x490d678d00000000 +rk7: +DQ 0x0000000104d101df +rk8: +DQ 0x0000000104c11db7 +mask: +dq 0xFFFFFFFFFFFFFFFF, 0x0000000000000000 +mask2: +dq 0xFFFFFFFF00000000, 0xFFFFFFFFFFFFFFFF +mask3: +dq 0x8080808080808080, 0x8080808080808080 +mask4: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF + align 32 +pshufb_shf_table: + + dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 + + dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 + + dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 + + dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 + + dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 + + dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 + + dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 + + dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 + + dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 + + dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 + + dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 + + dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 + + dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 + + dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 + + dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 + + +SHUF_MASK dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +;;; func core, ver, snum +slversion crc32_ieee_by4, 05, 02, 0017 diff --git a/src/isa-l/crc/crc32_ieee_perf.c b/src/isa-l/crc/crc32_ieee_perf.c new file mode 100644 index 00000000..c6c74950 --- /dev/null +++ b/src/isa-l/crc/crc32_ieee_perf.c @@ -0,0 +1,87 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <sys/time.h> +#include "crc.h" +#include "test.h" + +//#define CACHED_TEST +#ifdef CACHED_TEST +// Cached test, loop many times over small dataset +# define TEST_LEN 8*1024 +# define TEST_LOOPS 400000 +# define TEST_TYPE_STR "_warm" +#else +// Uncached test. Pull from large mem base. +# define GT_L3_CACHE 32*1024*1024 /* some number > last level cache */ +# define TEST_LEN (2 * GT_L3_CACHE) +# define TEST_LOOPS 100 +# define TEST_TYPE_STR "_cold" +#endif + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define TEST_MEM TEST_LEN + +int main(int argc, char *argv[]) +{ + int i; + void *buf; + uint32_t crc; + struct perf start, stop; + + printf("crc32_ieee_perf:\n"); + + if (posix_memalign(&buf, 1024, TEST_LEN)) { + printf("alloc error: Fail"); + return -1; + } + + printf("Start timed tests\n"); + fflush(0); + + memset(buf, 0, TEST_LEN); + crc = crc32_ieee(TEST_SEED, buf, TEST_LEN); + perf_start(&start); + for (i = 0; i < TEST_LOOPS; i++) { + crc = crc32_ieee(TEST_SEED, buf, TEST_LEN); + } + perf_stop(&stop); + printf("crc32_ieee" TEST_TYPE_STR ": "); + perf_print(stop, start, (long long)TEST_LEN * i); + + printf("finish 0x%x\n", crc); + return 0; +} diff --git a/src/isa-l/crc/crc32_ieee_test.c b/src/isa-l/crc/crc32_ieee_test.c new file mode 100644 index 00000000..4dfbc005 --- /dev/null +++ b/src/isa-l/crc/crc32_ieee_test.c @@ -0,0 +1,174 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include "crc.h" +#include "types.h" + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define MAX_BUF 512 +#define TEST_SIZE 20 + +typedef uint64_t u64; +typedef uint32_t u32; +typedef uint16_t u16; +typedef uint8_t u8; + +// Generates pseudo-random data + +void rand_buffer(unsigned char *buf, long buffer_size) +{ + long i; + for (i = 0; i < buffer_size; i++) + buf[i] = rand(); +} + +int main(int argc, char *argv[]) +{ + int fail = 0; + u32 r; + int verbose = argc - 1; + int i, s, ret; + void *buf_alloc; + unsigned char *buf; + + printf("Test crc32_ieee "); + + // Align to MAX_BUF boundary + ret = posix_memalign(&buf_alloc, MAX_BUF, MAX_BUF * TEST_SIZE); + if (ret) { + printf("alloc error: Fail"); + return -1; + } + buf = (unsigned char *)buf_alloc; + + srand(TEST_SEED); + + // Test of all zeros + memset(buf, 0, MAX_BUF * 10); + u32 crc = crc32_ieee(TEST_SEED, buf, MAX_BUF); + u32 crc_ref = crc32_ieee_base(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) { + fail++; + printf("\n opt ref\n"); + printf(" ------ ------\n"); + printf("crc zero = 0x%8x 0x%8x \n", crc, crc_ref); + } else + printf("."); + + // Another simple test pattern + memset(buf, 0x8a, MAX_BUF); + crc = crc32_ieee(TEST_SEED, buf, MAX_BUF); + crc_ref = crc32_ieee_base(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc all 8a = 0x%8x 0x%8x\n", crc, crc_ref); + else + printf("."); + + // Do a few random tests + r = rand(); + rand_buffer(buf, MAX_BUF * TEST_SIZE); + + for (i = 0; i < TEST_SIZE; i++) { + crc = crc32_ieee(r, buf, MAX_BUF); + crc_ref = crc32_ieee_base(r, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%8x 0x%8x\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + + // Do a few random sizes + buf = (unsigned char *)buf_alloc; //reset buf + r = rand(); + + for (i = MAX_BUF; i >= 0; i--) { + crc = crc32_ieee(r, buf, i); + crc_ref = crc32_ieee_base(r, buf, i); + if (crc != crc_ref) { + fail++; + printf("fail random size%i 0x%8x 0x%8x\n", i, crc, crc_ref); + } else + printf("."); + } + + // Try different seeds + for (s = 0; s < 20; s++) { + buf = (unsigned char *)buf_alloc; //reset buf + + r = rand(); // just to get a new seed + rand_buffer(buf, MAX_BUF * TEST_SIZE); // new pseudo-rand data + + if (verbose) + printf("seed = 0x%x\n", r); + + for (i = 0; i < TEST_SIZE; i++) { + crc = crc32_ieee(r, buf, MAX_BUF); + crc_ref = crc32_ieee_base(r, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%8x 0x%8x\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + } + + // Run tests at end of buffer + buf = (unsigned char *)buf_alloc; //reset buf + buf = buf + ((MAX_BUF - 1) * TEST_SIZE); //Line up TEST_SIZE from end + for (i = 0; i < TEST_SIZE; i++) { + crc = crc32_ieee(TEST_SEED, buf + i, TEST_SIZE - i); + crc_ref = crc32_ieee_base(TEST_SEED, buf + i, TEST_SIZE - i); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc eob rand%3d = 0x%4x 0x%4x\n", i, crc, crc_ref); + else + printf("."); + } + + printf("Test done: %s\n", fail ? "Fail" : "Pass"); + if (fail) + printf("\nFailed %d tests\n", fail); + + return fail; +} diff --git a/src/isa-l/crc/crc32_iscsi_00.asm b/src/isa-l/crc/crc32_iscsi_00.asm new file mode 100644 index 00000000..2833a8d0 --- /dev/null +++ b/src/isa-l/crc/crc32_iscsi_00.asm @@ -0,0 +1,656 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function to compute iscsi CRC32 with table-based recombination +; crc done "by 3" with block sizes 1920, 960, 480, 240 + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +%include "reg_sizes.asm" + +default rel +; crcB3 MACRO to implement crc32 on 3 %%bSize-byte blocks +%macro crcB3 3 +%define %%bSize %1 ; 1/3 of buffer size +%define %%td2 %2 ; table offset for crc0 (2/3 of buffer) +%define %%td1 %3 ; table offset for crc1 (1/3 of buffer) + +%IF %%bSize=640 + sub len, %%bSize*3 + js %%crcB3_end ;; jump to next level if 3*blockSize > len +%ELSE + cmp len, %%bSize*3 + jnae %%crcB3_end ;; jump to next level if 3*blockSize > len +%ENDIF + ;;;;;; Calculate CRC of 3 blocks of the buffer ;;;;;; +%%crcB3_loop: + ;; rax = crc0 = initial crc + xor rbx, rbx ;; rbx = crc1 = 0; + xor r10, r10 ;; r10 = crc2 = 0; + + %assign i 0 + %rep %%bSize/8 - 1 + %if i < %%bSize*3/4 + prefetchnta [bufptmp+ %%bSize*3 +i*4] + %endif + crc32 rax, qword [bufptmp+i + 0*%%bSize] ;; update crc0 + crc32 rbx, qword [bufptmp+i + 1*%%bSize] ;; update crc1 + crc32 r10, qword [bufptmp+i + 2*%%bSize] ;; update crc2 + %assign i (i+8) + %endrep + crc32 rax, qword [bufptmp+i + 0*%%bSize] ;; update crc0 + crc32 rbx, qword [bufptmp+i + 1*%%bSize] ;; update crc1 +; SKIP ;crc32 r10, [bufptmp+i + 2*%%bSize] ;; update crc2 + + ; merge in crc0 + movzx bufp_dw, al + mov r9d, [crc_init + bufp*4 + %%td2] + movzx bufp_dw, ah + shr eax, 16 + mov r11d, [crc_init + bufp*4 + %%td2] + shl r11, 8 + xor r9, r11 + + movzx bufp_dw, al + mov r11d, [crc_init + bufp*4 + %%td2] + movzx bufp_dw, ah + shl r11, 16 + xor r9, r11 + mov r11d, [crc_init + bufp*4 + %%td2] + shl r11, 24 + xor r9, r11 + + ; merge in crc1 + + movzx bufp_dw, bl + mov r11d, [crc_init + bufp*4 + %%td1] + movzx bufp_dw, bh + shr ebx, 16 + xor r9, r11 + mov r11d, [crc_init + bufp*4 + %%td1] + shl r11, 8 + xor r9, r11 + + movzx bufp_dw, bl + mov r11d, [crc_init + bufp*4 + %%td1] + movzx bufp_dw, bh + shl r11, 16 + xor r9, r11 + mov r11d, [crc_init + bufp*4 + %%td1] + shl r11, 24 + xor r9, r11 + + xor r9, [bufptmp+i + 2*%%bSize] + crc32 r10, r9 + mov rax, r10 + + add bufptmp, %%bSize*3 ;; move to next block + sub len, %%bSize*3 +%IF %%bSize=640 + jns %%crcB3_loop +%ENDIF + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +%%crcB3_end: +%IF %%bSize=640 + add len, %%bSize*3 +%ENDIF + je do_return ;; return if remaining data is zero +%endmacro + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;; ISCSI CRC 32 Implementation with crc32 Instruction + +;;; unsigned int crc32_iscsi_00(unsigned char * buffer, int len, unsigned int crc_init); +;;; +;;; *buf = rcx +;;; len = rdx +;;; crc_init = r8 +;;; + +global crc32_iscsi_00:function +crc32_iscsi_00: + +%ifidn __OUTPUT_FORMAT__, elf64 +%define bufp rdi +%define bufp_dw edi +%define bufp_w di +%define bufp_b dil +%define bufptmp rcx +%define block_0 rcx +%define block_1 r8 +%define block_2 r11 +%define len rsi +%define len_dw esi +%define len_w si +%define len_b sil +%define crc_init rdx +%define crc_init_dw edx +%else +%define bufp rcx +%define bufp_dw ecx +%define bufp_w cx +%define bufp_b cl +%define bufptmp rdi +%define block_0 rdi +%define block_1 rsi +%define block_2 r11 +%define len rdx +%define len_dw edx +%define len_w dx +%define len_b dl +%define crc_init r8 +%define crc_init_dw r8d +%endif + + + push rdi + push rbx + + mov rax, crc_init ;; rax = crc_init; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;; 1) ALIGN: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + mov bufptmp, bufp ;; rdi = *buf + neg bufp + and bufp, 7 ;; calculate the unalignment + ;; amount of the address + je proc_block ;; Skip if aligned + + cmp len, 8 + jb less_than_8 + + ;;;; Calculate CRC of unaligned bytes of the buffer (if any) ;;;; + mov rbx, [bufptmp] ;; load a quadword from the buffer + add bufptmp, bufp ;; align buffer pointer for + ;; quadword processing + sub len, bufp ;; update buffer length +align_loop: + crc32 eax, bl ;; compute crc32 of 1-byte + shr rbx, 8 ;; get next byte + dec bufp + jne align_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;; 2) BLOCK LEVEL: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +proc_block: + cmp len, 240 + jb bit8 + + lea crc_init, [mul_table_72] ;; load table base address + + crcB3 640, 0x1000, 0x0c00 ; 640*3 = 1920 (Tables 1280, 640) + crcB3 320, 0x0c00, 0x0800 ; 320*3 = 960 (Tables 640, 320) + crcB3 160, 0x0800, 0x0400 ; 160*3 = 480 (Tables 320, 160) + crcB3 80, 0x0400, 0x0000 ; 80*3 = 240 (Tables 160, 80) + + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;4) LESS THAN 256-bytes REMAIN AT THIS POINT (8-bits of rdx are full) + +bit8: + shl len_b, 1 ;; shift-out MSB (bit-7) + jnc bit7 ;; jump to bit-6 if bit-7 == 0 + %assign i 0 + %rep 16 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return ;; return if remaining data is zero + add bufptmp, 128 ;; buf +=64; (next 64 bytes) + +bit7: + shl len_b, 1 ;; shift-out MSB (bit-7) + jnc bit6 ;; jump to bit-6 if bit-7 == 0 + %assign i 0 + %rep 8 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return ;; return if remaining data is zero + add bufptmp, 64 ;; buf +=64; (next 64 bytes) +bit6: + shl len_b, 1 ;; shift-out MSB (bit-6) + jnc bit5 ;; jump to bit-5 if bit-6 == 0 + %assign i 0 + %rep 4 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return ;; return if remaining data is zero + add bufptmp, 32 ;; buf +=32; (next 32 bytes) +bit5: + shl len_b, 1 ;; shift-out MSB (bit-5) + jnc bit4 ;; jump to bit-4 if bit-5 == 0 + %assign i 0 + %rep 2 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return ;; return if remaining data is zero + add bufptmp, 16 ;; buf +=16; (next 16 bytes) +bit4: + shl len_b, 1 ;; shift-out MSB (bit-4) + jnc bit3 ;; jump to bit-3 if bit-4 == 0 + crc32 rax, qword [bufptmp] ;; compute crc32 of 8-byte data + je do_return ;; return if remaining data is zero + add bufptmp, 8 ;; buf +=8; (next 8 bytes) +bit3: + mov rbx, qword [bufptmp] ;; load a 8-bytes from the buffer: + shl len_b, 1 ;; shift-out MSB (bit-3) + jnc bit2 ;; jump to bit-2 if bit-3 == 0 + crc32 eax, ebx ;; compute crc32 of 4-byte data + je do_return ;; return if remaining data is zero + shr rbx, 32 ;; get next 3 bytes +bit2: + shl len_b, 1 ;; shift-out MSB (bit-2) + jnc bit1 ;; jump to bit-1 if bit-2 == 0 + crc32 eax, bx ;; compute crc32 of 2-byte data + je do_return ;; return if remaining data is zero + shr rbx, 16 ;; next byte +bit1: + test len_b,len_b + je do_return + crc32 eax, bl ;; compute crc32 of 1-byte data +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +do_return: + + pop rbx + pop rdi + ret + +less_than_8: + test len,4 + jz less_than_4 + crc32 eax, dword[bufptmp] + add bufptmp,4 +less_than_4: + test len,2 + jz less_than_2 + crc32 eax, word[bufptmp] + add bufptmp,2 +less_than_2: + test len,1 + jz do_return + crc32 rax, byte[bufptmp] + pop rbx + pop bufptmp + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;; global mul_table_72, mul_table_152, mul_table_312, mul_table_632, mul_table_1272 + +section .data +align 8 +mul_table_72: +DD 0x00000000,0x39d3b296,0x73a7652c,0x4a74d7ba +DD 0xe74eca58,0xde9d78ce,0x94e9af74,0xad3a1de2 +DD 0xcb71e241,0xf2a250d7,0xb8d6876d,0x810535fb +DD 0x2c3f2819,0x15ec9a8f,0x5f984d35,0x664bffa3 +DD 0x930fb273,0xaadc00e5,0xe0a8d75f,0xd97b65c9 +DD 0x7441782b,0x4d92cabd,0x07e61d07,0x3e35af91 +DD 0x587e5032,0x61ade2a4,0x2bd9351e,0x120a8788 +DD 0xbf309a6a,0x86e328fc,0xcc97ff46,0xf5444dd0 +DD 0x23f31217,0x1a20a081,0x5054773b,0x6987c5ad +DD 0xc4bdd84f,0xfd6e6ad9,0xb71abd63,0x8ec90ff5 +DD 0xe882f056,0xd15142c0,0x9b25957a,0xa2f627ec +DD 0x0fcc3a0e,0x361f8898,0x7c6b5f22,0x45b8edb4 +DD 0xb0fca064,0x892f12f2,0xc35bc548,0xfa8877de +DD 0x57b26a3c,0x6e61d8aa,0x24150f10,0x1dc6bd86 +DD 0x7b8d4225,0x425ef0b3,0x082a2709,0x31f9959f +DD 0x9cc3887d,0xa5103aeb,0xef64ed51,0xd6b75fc7 +DD 0x47e6242e,0x7e3596b8,0x34414102,0x0d92f394 +DD 0xa0a8ee76,0x997b5ce0,0xd30f8b5a,0xeadc39cc +DD 0x8c97c66f,0xb54474f9,0xff30a343,0xc6e311d5 +DD 0x6bd90c37,0x520abea1,0x187e691b,0x21addb8d +DD 0xd4e9965d,0xed3a24cb,0xa74ef371,0x9e9d41e7 +DD 0x33a75c05,0x0a74ee93,0x40003929,0x79d38bbf +DD 0x1f98741c,0x264bc68a,0x6c3f1130,0x55eca3a6 +DD 0xf8d6be44,0xc1050cd2,0x8b71db68,0xb2a269fe +DD 0x64153639,0x5dc684af,0x17b25315,0x2e61e183 +DD 0x835bfc61,0xba884ef7,0xf0fc994d,0xc92f2bdb +DD 0xaf64d478,0x96b766ee,0xdcc3b154,0xe51003c2 +DD 0x482a1e20,0x71f9acb6,0x3b8d7b0c,0x025ec99a +DD 0xf71a844a,0xcec936dc,0x84bde166,0xbd6e53f0 +DD 0x10544e12,0x2987fc84,0x63f32b3e,0x5a2099a8 +DD 0x3c6b660b,0x05b8d49d,0x4fcc0327,0x761fb1b1 +DD 0xdb25ac53,0xe2f61ec5,0xa882c97f,0x91517be9 +DD 0x8fcc485c,0xb61ffaca,0xfc6b2d70,0xc5b89fe6 +DD 0x68828204,0x51513092,0x1b25e728,0x22f655be +DD 0x44bdaa1d,0x7d6e188b,0x371acf31,0x0ec97da7 +DD 0xa3f36045,0x9a20d2d3,0xd0540569,0xe987b7ff +DD 0x1cc3fa2f,0x251048b9,0x6f649f03,0x56b72d95 +DD 0xfb8d3077,0xc25e82e1,0x882a555b,0xb1f9e7cd +DD 0xd7b2186e,0xee61aaf8,0xa4157d42,0x9dc6cfd4 +DD 0x30fcd236,0x092f60a0,0x435bb71a,0x7a88058c +DD 0xac3f5a4b,0x95ece8dd,0xdf983f67,0xe64b8df1 +DD 0x4b719013,0x72a22285,0x38d6f53f,0x010547a9 +DD 0x674eb80a,0x5e9d0a9c,0x14e9dd26,0x2d3a6fb0 +DD 0x80007252,0xb9d3c0c4,0xf3a7177e,0xca74a5e8 +DD 0x3f30e838,0x06e35aae,0x4c978d14,0x75443f82 +DD 0xd87e2260,0xe1ad90f6,0xabd9474c,0x920af5da +DD 0xf4410a79,0xcd92b8ef,0x87e66f55,0xbe35ddc3 +DD 0x130fc021,0x2adc72b7,0x60a8a50d,0x597b179b +DD 0xc82a6c72,0xf1f9dee4,0xbb8d095e,0x825ebbc8 +DD 0x2f64a62a,0x16b714bc,0x5cc3c306,0x65107190 +DD 0x035b8e33,0x3a883ca5,0x70fceb1f,0x492f5989 +DD 0xe415446b,0xddc6f6fd,0x97b22147,0xae6193d1 +DD 0x5b25de01,0x62f66c97,0x2882bb2d,0x115109bb +DD 0xbc6b1459,0x85b8a6cf,0xcfcc7175,0xf61fc3e3 +DD 0x90543c40,0xa9878ed6,0xe3f3596c,0xda20ebfa +DD 0x771af618,0x4ec9448e,0x04bd9334,0x3d6e21a2 +DD 0xebd97e65,0xd20accf3,0x987e1b49,0xa1ada9df +DD 0x0c97b43d,0x354406ab,0x7f30d111,0x46e36387 +DD 0x20a89c24,0x197b2eb2,0x530ff908,0x6adc4b9e +DD 0xc7e6567c,0xfe35e4ea,0xb4413350,0x8d9281c6 +DD 0x78d6cc16,0x41057e80,0x0b71a93a,0x32a21bac +DD 0x9f98064e,0xa64bb4d8,0xec3f6362,0xd5ecd1f4 +DD 0xb3a72e57,0x8a749cc1,0xc0004b7b,0xf9d3f9ed +DD 0x54e9e40f,0x6d3a5699,0x274e8123,0x1e9d33b5 + +mul_table_152: +DD 0x00000000,0x878a92a7,0x0af953bf,0x8d73c118 +DD 0x15f2a77e,0x927835d9,0x1f0bf4c1,0x98816666 +DD 0x2be54efc,0xac6fdc5b,0x211c1d43,0xa6968fe4 +DD 0x3e17e982,0xb99d7b25,0x34eeba3d,0xb364289a +DD 0x57ca9df8,0xd0400f5f,0x5d33ce47,0xdab95ce0 +DD 0x42383a86,0xc5b2a821,0x48c16939,0xcf4bfb9e +DD 0x7c2fd304,0xfba541a3,0x76d680bb,0xf15c121c +DD 0x69dd747a,0xee57e6dd,0x632427c5,0xe4aeb562 +DD 0xaf953bf0,0x281fa957,0xa56c684f,0x22e6fae8 +DD 0xba679c8e,0x3ded0e29,0xb09ecf31,0x37145d96 +DD 0x8470750c,0x03fae7ab,0x8e8926b3,0x0903b414 +DD 0x9182d272,0x160840d5,0x9b7b81cd,0x1cf1136a +DD 0xf85fa608,0x7fd534af,0xf2a6f5b7,0x752c6710 +DD 0xedad0176,0x6a2793d1,0xe75452c9,0x60dec06e +DD 0xd3bae8f4,0x54307a53,0xd943bb4b,0x5ec929ec +DD 0xc6484f8a,0x41c2dd2d,0xccb11c35,0x4b3b8e92 +DD 0x5ac60111,0xdd4c93b6,0x503f52ae,0xd7b5c009 +DD 0x4f34a66f,0xc8be34c8,0x45cdf5d0,0xc2476777 +DD 0x71234fed,0xf6a9dd4a,0x7bda1c52,0xfc508ef5 +DD 0x64d1e893,0xe35b7a34,0x6e28bb2c,0xe9a2298b +DD 0x0d0c9ce9,0x8a860e4e,0x07f5cf56,0x807f5df1 +DD 0x18fe3b97,0x9f74a930,0x12076828,0x958dfa8f +DD 0x26e9d215,0xa16340b2,0x2c1081aa,0xab9a130d +DD 0x331b756b,0xb491e7cc,0x39e226d4,0xbe68b473 +DD 0xf5533ae1,0x72d9a846,0xffaa695e,0x7820fbf9 +DD 0xe0a19d9f,0x672b0f38,0xea58ce20,0x6dd25c87 +DD 0xdeb6741d,0x593ce6ba,0xd44f27a2,0x53c5b505 +DD 0xcb44d363,0x4cce41c4,0xc1bd80dc,0x4637127b +DD 0xa299a719,0x251335be,0xa860f4a6,0x2fea6601 +DD 0xb76b0067,0x30e192c0,0xbd9253d8,0x3a18c17f +DD 0x897ce9e5,0x0ef67b42,0x8385ba5a,0x040f28fd +DD 0x9c8e4e9b,0x1b04dc3c,0x96771d24,0x11fd8f83 +DD 0xb58c0222,0x32069085,0xbf75519d,0x38ffc33a +DD 0xa07ea55c,0x27f437fb,0xaa87f6e3,0x2d0d6444 +DD 0x9e694cde,0x19e3de79,0x94901f61,0x131a8dc6 +DD 0x8b9beba0,0x0c117907,0x8162b81f,0x06e82ab8 +DD 0xe2469fda,0x65cc0d7d,0xe8bfcc65,0x6f355ec2 +DD 0xf7b438a4,0x703eaa03,0xfd4d6b1b,0x7ac7f9bc +DD 0xc9a3d126,0x4e294381,0xc35a8299,0x44d0103e +DD 0xdc517658,0x5bdbe4ff,0xd6a825e7,0x5122b740 +DD 0x1a1939d2,0x9d93ab75,0x10e06a6d,0x976af8ca +DD 0x0feb9eac,0x88610c0b,0x0512cd13,0x82985fb4 +DD 0x31fc772e,0xb676e589,0x3b052491,0xbc8fb636 +DD 0x240ed050,0xa38442f7,0x2ef783ef,0xa97d1148 +DD 0x4dd3a42a,0xca59368d,0x472af795,0xc0a06532 +DD 0x58210354,0xdfab91f3,0x52d850eb,0xd552c24c +DD 0x6636ead6,0xe1bc7871,0x6ccfb969,0xeb452bce +DD 0x73c44da8,0xf44edf0f,0x793d1e17,0xfeb78cb0 +DD 0xef4a0333,0x68c09194,0xe5b3508c,0x6239c22b +DD 0xfab8a44d,0x7d3236ea,0xf041f7f2,0x77cb6555 +DD 0xc4af4dcf,0x4325df68,0xce561e70,0x49dc8cd7 +DD 0xd15deab1,0x56d77816,0xdba4b90e,0x5c2e2ba9 +DD 0xb8809ecb,0x3f0a0c6c,0xb279cd74,0x35f35fd3 +DD 0xad7239b5,0x2af8ab12,0xa78b6a0a,0x2001f8ad +DD 0x9365d037,0x14ef4290,0x999c8388,0x1e16112f +DD 0x86977749,0x011de5ee,0x8c6e24f6,0x0be4b651 +DD 0x40df38c3,0xc755aa64,0x4a266b7c,0xcdacf9db +DD 0x552d9fbd,0xd2a70d1a,0x5fd4cc02,0xd85e5ea5 +DD 0x6b3a763f,0xecb0e498,0x61c32580,0xe649b727 +DD 0x7ec8d141,0xf94243e6,0x743182fe,0xf3bb1059 +DD 0x1715a53b,0x909f379c,0x1decf684,0x9a666423 +DD 0x02e70245,0x856d90e2,0x081e51fa,0x8f94c35d +DD 0x3cf0ebc7,0xbb7a7960,0x3609b878,0xb1832adf +DD 0x29024cb9,0xae88de1e,0x23fb1f06,0xa4718da1 + +mul_table_312: +DD 0x00000000,0xbac2fd7b,0x70698c07,0xcaab717c +DD 0xe0d3180e,0x5a11e575,0x90ba9409,0x2a786972 +DD 0xc44a46ed,0x7e88bb96,0xb423caea,0x0ee13791 +DD 0x24995ee3,0x9e5ba398,0x54f0d2e4,0xee322f9f +DD 0x8d78fb2b,0x37ba0650,0xfd11772c,0x47d38a57 +DD 0x6dabe325,0xd7691e5e,0x1dc26f22,0xa7009259 +DD 0x4932bdc6,0xf3f040bd,0x395b31c1,0x8399ccba +DD 0xa9e1a5c8,0x132358b3,0xd98829cf,0x634ad4b4 +DD 0x1f1d80a7,0xa5df7ddc,0x6f740ca0,0xd5b6f1db +DD 0xffce98a9,0x450c65d2,0x8fa714ae,0x3565e9d5 +DD 0xdb57c64a,0x61953b31,0xab3e4a4d,0x11fcb736 +DD 0x3b84de44,0x8146233f,0x4bed5243,0xf12faf38 +DD 0x92657b8c,0x28a786f7,0xe20cf78b,0x58ce0af0 +DD 0x72b66382,0xc8749ef9,0x02dfef85,0xb81d12fe +DD 0x562f3d61,0xecedc01a,0x2646b166,0x9c844c1d +DD 0xb6fc256f,0x0c3ed814,0xc695a968,0x7c575413 +DD 0x3e3b014e,0x84f9fc35,0x4e528d49,0xf4907032 +DD 0xdee81940,0x642ae43b,0xae819547,0x1443683c +DD 0xfa7147a3,0x40b3bad8,0x8a18cba4,0x30da36df +DD 0x1aa25fad,0xa060a2d6,0x6acbd3aa,0xd0092ed1 +DD 0xb343fa65,0x0981071e,0xc32a7662,0x79e88b19 +DD 0x5390e26b,0xe9521f10,0x23f96e6c,0x993b9317 +DD 0x7709bc88,0xcdcb41f3,0x0760308f,0xbda2cdf4 +DD 0x97daa486,0x2d1859fd,0xe7b32881,0x5d71d5fa +DD 0x212681e9,0x9be47c92,0x514f0dee,0xeb8df095 +DD 0xc1f599e7,0x7b37649c,0xb19c15e0,0x0b5ee89b +DD 0xe56cc704,0x5fae3a7f,0x95054b03,0x2fc7b678 +DD 0x05bfdf0a,0xbf7d2271,0x75d6530d,0xcf14ae76 +DD 0xac5e7ac2,0x169c87b9,0xdc37f6c5,0x66f50bbe +DD 0x4c8d62cc,0xf64f9fb7,0x3ce4eecb,0x862613b0 +DD 0x68143c2f,0xd2d6c154,0x187db028,0xa2bf4d53 +DD 0x88c72421,0x3205d95a,0xf8aea826,0x426c555d +DD 0x7c76029c,0xc6b4ffe7,0x0c1f8e9b,0xb6dd73e0 +DD 0x9ca51a92,0x2667e7e9,0xeccc9695,0x560e6bee +DD 0xb83c4471,0x02feb90a,0xc855c876,0x7297350d +DD 0x58ef5c7f,0xe22da104,0x2886d078,0x92442d03 +DD 0xf10ef9b7,0x4bcc04cc,0x816775b0,0x3ba588cb +DD 0x11dde1b9,0xab1f1cc2,0x61b46dbe,0xdb7690c5 +DD 0x3544bf5a,0x8f864221,0x452d335d,0xffefce26 +DD 0xd597a754,0x6f555a2f,0xa5fe2b53,0x1f3cd628 +DD 0x636b823b,0xd9a97f40,0x13020e3c,0xa9c0f347 +DD 0x83b89a35,0x397a674e,0xf3d11632,0x4913eb49 +DD 0xa721c4d6,0x1de339ad,0xd74848d1,0x6d8ab5aa +DD 0x47f2dcd8,0xfd3021a3,0x379b50df,0x8d59ada4 +DD 0xee137910,0x54d1846b,0x9e7af517,0x24b8086c +DD 0x0ec0611e,0xb4029c65,0x7ea9ed19,0xc46b1062 +DD 0x2a593ffd,0x909bc286,0x5a30b3fa,0xe0f24e81 +DD 0xca8a27f3,0x7048da88,0xbae3abf4,0x0021568f +DD 0x424d03d2,0xf88ffea9,0x32248fd5,0x88e672ae +DD 0xa29e1bdc,0x185ce6a7,0xd2f797db,0x68356aa0 +DD 0x8607453f,0x3cc5b844,0xf66ec938,0x4cac3443 +DD 0x66d45d31,0xdc16a04a,0x16bdd136,0xac7f2c4d +DD 0xcf35f8f9,0x75f70582,0xbf5c74fe,0x059e8985 +DD 0x2fe6e0f7,0x95241d8c,0x5f8f6cf0,0xe54d918b +DD 0x0b7fbe14,0xb1bd436f,0x7b163213,0xc1d4cf68 +DD 0xebaca61a,0x516e5b61,0x9bc52a1d,0x2107d766 +DD 0x5d508375,0xe7927e0e,0x2d390f72,0x97fbf209 +DD 0xbd839b7b,0x07416600,0xcdea177c,0x7728ea07 +DD 0x991ac598,0x23d838e3,0xe973499f,0x53b1b4e4 +DD 0x79c9dd96,0xc30b20ed,0x09a05191,0xb362acea +DD 0xd028785e,0x6aea8525,0xa041f459,0x1a830922 +DD 0x30fb6050,0x8a399d2b,0x4092ec57,0xfa50112c +DD 0x14623eb3,0xaea0c3c8,0x640bb2b4,0xdec94fcf +DD 0xf4b126bd,0x4e73dbc6,0x84d8aaba,0x3e1a57c1 + +mul_table_632: +DD 0x00000000,0x6b749fb2,0xd6e93f64,0xbd9da0d6 +DD 0xa83e0839,0xc34a978b,0x7ed7375d,0x15a3a8ef +DD 0x55906683,0x3ee4f931,0x837959e7,0xe80dc655 +DD 0xfdae6eba,0x96daf108,0x2b4751de,0x4033ce6c +DD 0xab20cd06,0xc05452b4,0x7dc9f262,0x16bd6dd0 +DD 0x031ec53f,0x686a5a8d,0xd5f7fa5b,0xbe8365e9 +DD 0xfeb0ab85,0x95c43437,0x285994e1,0x432d0b53 +DD 0x568ea3bc,0x3dfa3c0e,0x80679cd8,0xeb13036a +DD 0x53adecfd,0x38d9734f,0x8544d399,0xee304c2b +DD 0xfb93e4c4,0x90e77b76,0x2d7adba0,0x460e4412 +DD 0x063d8a7e,0x6d4915cc,0xd0d4b51a,0xbba02aa8 +DD 0xae038247,0xc5771df5,0x78eabd23,0x139e2291 +DD 0xf88d21fb,0x93f9be49,0x2e641e9f,0x4510812d +DD 0x50b329c2,0x3bc7b670,0x865a16a6,0xed2e8914 +DD 0xad1d4778,0xc669d8ca,0x7bf4781c,0x1080e7ae +DD 0x05234f41,0x6e57d0f3,0xd3ca7025,0xb8beef97 +DD 0xa75bd9fa,0xcc2f4648,0x71b2e69e,0x1ac6792c +DD 0x0f65d1c3,0x64114e71,0xd98ceea7,0xb2f87115 +DD 0xf2cbbf79,0x99bf20cb,0x2422801d,0x4f561faf +DD 0x5af5b740,0x318128f2,0x8c1c8824,0xe7681796 +DD 0x0c7b14fc,0x670f8b4e,0xda922b98,0xb1e6b42a +DD 0xa4451cc5,0xcf318377,0x72ac23a1,0x19d8bc13 +DD 0x59eb727f,0x329fedcd,0x8f024d1b,0xe476d2a9 +DD 0xf1d57a46,0x9aa1e5f4,0x273c4522,0x4c48da90 +DD 0xf4f63507,0x9f82aab5,0x221f0a63,0x496b95d1 +DD 0x5cc83d3e,0x37bca28c,0x8a21025a,0xe1559de8 +DD 0xa1665384,0xca12cc36,0x778f6ce0,0x1cfbf352 +DD 0x09585bbd,0x622cc40f,0xdfb164d9,0xb4c5fb6b +DD 0x5fd6f801,0x34a267b3,0x893fc765,0xe24b58d7 +DD 0xf7e8f038,0x9c9c6f8a,0x2101cf5c,0x4a7550ee +DD 0x0a469e82,0x61320130,0xdcafa1e6,0xb7db3e54 +DD 0xa27896bb,0xc90c0909,0x7491a9df,0x1fe5366d +DD 0x4b5bc505,0x202f5ab7,0x9db2fa61,0xf6c665d3 +DD 0xe365cd3c,0x8811528e,0x358cf258,0x5ef86dea +DD 0x1ecba386,0x75bf3c34,0xc8229ce2,0xa3560350 +DD 0xb6f5abbf,0xdd81340d,0x601c94db,0x0b680b69 +DD 0xe07b0803,0x8b0f97b1,0x36923767,0x5de6a8d5 +DD 0x4845003a,0x23319f88,0x9eac3f5e,0xf5d8a0ec +DD 0xb5eb6e80,0xde9ff132,0x630251e4,0x0876ce56 +DD 0x1dd566b9,0x76a1f90b,0xcb3c59dd,0xa048c66f +DD 0x18f629f8,0x7382b64a,0xce1f169c,0xa56b892e +DD 0xb0c821c1,0xdbbcbe73,0x66211ea5,0x0d558117 +DD 0x4d664f7b,0x2612d0c9,0x9b8f701f,0xf0fbefad +DD 0xe5584742,0x8e2cd8f0,0x33b17826,0x58c5e794 +DD 0xb3d6e4fe,0xd8a27b4c,0x653fdb9a,0x0e4b4428 +DD 0x1be8ecc7,0x709c7375,0xcd01d3a3,0xa6754c11 +DD 0xe646827d,0x8d321dcf,0x30afbd19,0x5bdb22ab +DD 0x4e788a44,0x250c15f6,0x9891b520,0xf3e52a92 +DD 0xec001cff,0x8774834d,0x3ae9239b,0x519dbc29 +DD 0x443e14c6,0x2f4a8b74,0x92d72ba2,0xf9a3b410 +DD 0xb9907a7c,0xd2e4e5ce,0x6f794518,0x040ddaaa +DD 0x11ae7245,0x7adaedf7,0xc7474d21,0xac33d293 +DD 0x4720d1f9,0x2c544e4b,0x91c9ee9d,0xfabd712f +DD 0xef1ed9c0,0x846a4672,0x39f7e6a4,0x52837916 +DD 0x12b0b77a,0x79c428c8,0xc459881e,0xaf2d17ac +DD 0xba8ebf43,0xd1fa20f1,0x6c678027,0x07131f95 +DD 0xbfadf002,0xd4d96fb0,0x6944cf66,0x023050d4 +DD 0x1793f83b,0x7ce76789,0xc17ac75f,0xaa0e58ed +DD 0xea3d9681,0x81490933,0x3cd4a9e5,0x57a03657 +DD 0x42039eb8,0x2977010a,0x94eaa1dc,0xff9e3e6e +DD 0x148d3d04,0x7ff9a2b6,0xc2640260,0xa9109dd2 +DD 0xbcb3353d,0xd7c7aa8f,0x6a5a0a59,0x012e95eb +DD 0x411d5b87,0x2a69c435,0x97f464e3,0xfc80fb51 +DD 0xe92353be,0x8257cc0c,0x3fca6cda,0x54bef368 + +mul_table_1272: +DD 0x00000000,0xdd66cbbb,0xbf21e187,0x62472a3c +DD 0x7bafb5ff,0xa6c97e44,0xc48e5478,0x19e89fc3 +DD 0xf75f6bfe,0x2a39a045,0x487e8a79,0x951841c2 +DD 0x8cf0de01,0x519615ba,0x33d13f86,0xeeb7f43d +DD 0xeb52a10d,0x36346ab6,0x5473408a,0x89158b31 +DD 0x90fd14f2,0x4d9bdf49,0x2fdcf575,0xf2ba3ece +DD 0x1c0dcaf3,0xc16b0148,0xa32c2b74,0x7e4ae0cf +DD 0x67a27f0c,0xbac4b4b7,0xd8839e8b,0x05e55530 +DD 0xd34934eb,0x0e2fff50,0x6c68d56c,0xb10e1ed7 +DD 0xa8e68114,0x75804aaf,0x17c76093,0xcaa1ab28 +DD 0x24165f15,0xf97094ae,0x9b37be92,0x46517529 +DD 0x5fb9eaea,0x82df2151,0xe0980b6d,0x3dfec0d6 +DD 0x381b95e6,0xe57d5e5d,0x873a7461,0x5a5cbfda +DD 0x43b42019,0x9ed2eba2,0xfc95c19e,0x21f30a25 +DD 0xcf44fe18,0x122235a3,0x70651f9f,0xad03d424 +DD 0xb4eb4be7,0x698d805c,0x0bcaaa60,0xd6ac61db +DD 0xa37e1f27,0x7e18d49c,0x1c5ffea0,0xc139351b +DD 0xd8d1aad8,0x05b76163,0x67f04b5f,0xba9680e4 +DD 0x542174d9,0x8947bf62,0xeb00955e,0x36665ee5 +DD 0x2f8ec126,0xf2e80a9d,0x90af20a1,0x4dc9eb1a +DD 0x482cbe2a,0x954a7591,0xf70d5fad,0x2a6b9416 +DD 0x33830bd5,0xeee5c06e,0x8ca2ea52,0x51c421e9 +DD 0xbf73d5d4,0x62151e6f,0x00523453,0xdd34ffe8 +DD 0xc4dc602b,0x19baab90,0x7bfd81ac,0xa69b4a17 +DD 0x70372bcc,0xad51e077,0xcf16ca4b,0x127001f0 +DD 0x0b989e33,0xd6fe5588,0xb4b97fb4,0x69dfb40f +DD 0x87684032,0x5a0e8b89,0x3849a1b5,0xe52f6a0e +DD 0xfcc7f5cd,0x21a13e76,0x43e6144a,0x9e80dff1 +DD 0x9b658ac1,0x4603417a,0x24446b46,0xf922a0fd +DD 0xe0ca3f3e,0x3dacf485,0x5febdeb9,0x828d1502 +DD 0x6c3ae13f,0xb15c2a84,0xd31b00b8,0x0e7dcb03 +DD 0x179554c0,0xcaf39f7b,0xa8b4b547,0x75d27efc +DD 0x431048bf,0x9e768304,0xfc31a938,0x21576283 +DD 0x38bffd40,0xe5d936fb,0x879e1cc7,0x5af8d77c +DD 0xb44f2341,0x6929e8fa,0x0b6ec2c6,0xd608097d +DD 0xcfe096be,0x12865d05,0x70c17739,0xada7bc82 +DD 0xa842e9b2,0x75242209,0x17630835,0xca05c38e +DD 0xd3ed5c4d,0x0e8b97f6,0x6cccbdca,0xb1aa7671 +DD 0x5f1d824c,0x827b49f7,0xe03c63cb,0x3d5aa870 +DD 0x24b237b3,0xf9d4fc08,0x9b93d634,0x46f51d8f +DD 0x90597c54,0x4d3fb7ef,0x2f789dd3,0xf21e5668 +DD 0xebf6c9ab,0x36900210,0x54d7282c,0x89b1e397 +DD 0x670617aa,0xba60dc11,0xd827f62d,0x05413d96 +DD 0x1ca9a255,0xc1cf69ee,0xa38843d2,0x7eee8869 +DD 0x7b0bdd59,0xa66d16e2,0xc42a3cde,0x194cf765 +DD 0x00a468a6,0xddc2a31d,0xbf858921,0x62e3429a +DD 0x8c54b6a7,0x51327d1c,0x33755720,0xee139c9b +DD 0xf7fb0358,0x2a9dc8e3,0x48dae2df,0x95bc2964 +DD 0xe06e5798,0x3d089c23,0x5f4fb61f,0x82297da4 +DD 0x9bc1e267,0x46a729dc,0x24e003e0,0xf986c85b +DD 0x17313c66,0xca57f7dd,0xa810dde1,0x7576165a +DD 0x6c9e8999,0xb1f84222,0xd3bf681e,0x0ed9a3a5 +DD 0x0b3cf695,0xd65a3d2e,0xb41d1712,0x697bdca9 +DD 0x7093436a,0xadf588d1,0xcfb2a2ed,0x12d46956 +DD 0xfc639d6b,0x210556d0,0x43427cec,0x9e24b757 +DD 0x87cc2894,0x5aaae32f,0x38edc913,0xe58b02a8 +DD 0x33276373,0xee41a8c8,0x8c0682f4,0x5160494f +DD 0x4888d68c,0x95ee1d37,0xf7a9370b,0x2acffcb0 +DD 0xc478088d,0x191ec336,0x7b59e90a,0xa63f22b1 +DD 0xbfd7bd72,0x62b176c9,0x00f65cf5,0xdd90974e +DD 0xd875c27e,0x051309c5,0x675423f9,0xba32e842 +DD 0xa3da7781,0x7ebcbc3a,0x1cfb9606,0xc19d5dbd +DD 0x2f2aa980,0xf24c623b,0x900b4807,0x4d6d83bc +DD 0x54851c7f,0x89e3d7c4,0xeba4fdf8,0x36c23643 + +;;; func core, ver, snum +slversion crc32_iscsi_00, 00, 03, 0014 + diff --git a/src/isa-l/crc/crc32_iscsi_01.asm b/src/isa-l/crc/crc32_iscsi_01.asm new file mode 100644 index 00000000..5b730f63 --- /dev/null +++ b/src/isa-l/crc/crc32_iscsi_01.asm @@ -0,0 +1,572 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;; ISCSI CRC 32 Implementation with crc32 and pclmulqdq Instruction + +%include "reg_sizes.asm" + +default rel +%define CONCAT(a,b,c) a %+ b %+ c + +; Define threshold where buffers are considered "small" and routed to more +; efficient "by-1" code. This "by-1" code only handles up to 255 bytes, so +; SMALL_SIZE can be no larger than 256. +%define SMALL_SIZE 200 + +%if (SMALL_SIZE > 256) +%error SMALL_ SIZE must be <= 256 +% error ; needed because '%error' actually generates only a warning +%endif + +;;; unsigned int crc32_iscsi_01(unsigned char * buffer, int len, unsigned int crc_init); +;;; +;;; *buf = rcx +;;; len = rdx +;;; crc_init = r8 + +global crc32_iscsi_01:function +crc32_iscsi_01: + +%ifidn __OUTPUT_FORMAT__, elf64 +%define bufp rdi +%define bufp_dw edi +%define bufp_w di +%define bufp_b dil +%define bufptmp rcx +%define block_0 rcx +%define block_1 rdx +%define block_2 r11 +%define len rsi +%define len_dw esi +%define len_w si +%define len_b sil +%define crc_init_arg rdx +%else +%define bufp rcx +%define bufp_dw ecx +%define bufp_w cx +%define bufp_b cl +%define bufptmp rdi +%define block_0 rdi +%define block_1 rsi +%define block_2 r11 +%define len rdx +%define len_dw edx +%define len_w dx +%define len_b dl +%endif + +%define tmp rbx +%define crc_init r8 +%define crc_init_dw r8d +%define crc1 r9 +%define crc2 r10 + + push rbx + push rdi + push rsi + + ;; Move crc_init for Linux to a different reg +%ifidn __OUTPUT_FORMAT__, elf64 + mov crc_init, crc_init_arg +%endif + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 1) ALIGN: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + mov bufptmp, bufp ;; rdi = *buf + neg bufp + and bufp, 7 ;; calculate the unalignment amount of + ;; the address + je proc_block ;; Skip if aligned + + ;; If len is less than 8 and we're unaligned, we need to jump + ;; to special code to avoid reading beyond the end of the buffer + cmp len, 8 + jb less_than_8 + + ;;;; Calculate CRC of unaligned bytes of the buffer (if any) ;;; + mov tmp, [bufptmp] ;; load a quadword from the buffer + add bufptmp, bufp ;; align buffer pointer for quadword + ;; processing + sub len, bufp ;; update buffer length +align_loop: + crc32 crc_init_dw, bl ;; compute crc32 of 1-byte + shr tmp, 8 ;; get next byte + dec bufp + jne align_loop + +proc_block: + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 2) PROCESS BLOCKS: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + ;; compute num of bytes to be processed + mov tmp, len ;; save num bytes in tmp + + cmp len, 128*24 + jae full_block + +continue_block: + cmp len, SMALL_SIZE + jb small + + ;; len < 128*24 + mov rax, 2731 ;; 2731 = ceil(2^16 / 24) + mul len_dw + shr rax, 16 + + ;; eax contains floor(bytes / 24) = num 24-byte chunks to do + + ;; process rax 24-byte chunks (128 >= rax >= 0) + + ;; compute end address of each block + ;; rdi -> block 0 (base addr + RAX * 8) + ;; rsi -> block 1 (base addr + RAX * 16) + ;; r11 -> block 2 (base addr + RAX * 24) + lea block_0, [bufptmp + rax * 8] + lea block_1, [block_0 + rax * 8] + lea block_2, [block_1 + rax * 8] + + xor crc1,crc1 + xor crc2,crc2 + + ;; branch into array + lea bufp, [jump_table] + movzx len, word [bufp + rax * 2] ;; len is offset from crc_array + lea bufp, [bufp + len + crc_array - jump_table] + jmp bufp + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 2a) PROCESS FULL BLOCKS: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +full_block: + mov rax, 128 + lea block_1, [block_0 + 128*8*2] + lea block_2, [block_0 + 128*8*3] + add block_0, 128*8*1 + + xor crc1,crc1 + xor crc2,crc2 + +; ;; branch into array +; jmp CONCAT(crc_,128,) + ; Fall thruogh into top of crc array (crc_128) + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 3) CRC Array: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +crc_array: +%assign i 128 +%rep 128-1 +CONCAT(crc_,i,:) + crc32 crc_init, qword [block_0 - i*8] + crc32 crc1, qword [block_1 - i*8] + crc32 crc2, qword [block_2 - i*8] + + %if i > 128*8 / 32 ; prefetch next 3KB data + prefetchnta [block_2 + 128*32 - i*32] + %endif +%assign i (i-1) +%endrep + +CONCAT(crc_,i,:) + crc32 crc_init, qword [block_0 - i*8] + crc32 crc1, qword [block_1 - i*8] +; SKIP ;crc32 crc2, [block_2 - i*8] ; Don't do this one yet + + mov block_0, block_2 + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 4) Combine three results: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + lea bufp, [K_table - 16] ; first entry is for idx 1 + shl rax, 3 ; rax *= 8 + sub tmp, rax ; tmp -= rax*8 + shl rax, 1 + sub tmp, rax ; tmp -= rax*16 (total tmp -= rax*24) + add bufp, rax + + movdqa xmm0, [bufp] ; 2 consts: K1:K2 + + movq xmm1, crc_init ; CRC for block 1 + pclmulqdq xmm1, xmm0, 0x00 ; Multiply by K2 + + movq xmm2, crc1 ; CRC for block 2 + pclmulqdq xmm2, xmm0, 0x10 ; Multiply by K1 + + pxor xmm1, xmm2 + movq rax, xmm1 + xor rax, [block_2 - i*8] + mov crc_init, crc2 + crc32 crc_init, rax + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; 5) Check for end: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +CONCAT(crc_,0,:) + mov len, tmp + cmp tmp, 128*24 + jae full_block + cmp tmp, 24 + jae continue_block + +fewer_than_24: + ;; now fewer than 24 bytes remain + cmp tmp, 16 + jae do_16 + cmp tmp, 8 + jae do_8 + + ;; 0 <= tmp <= 7 + shl ebx, 29 ; size now in bits 31:29 + jz do_return +check_4: + mov bufp, [bufptmp] + shl ebx, 1 ; shift out into carry MSB (orig size & 4) + jnc check_2 + crc32 crc_init_dw, bufp_dw + jz do_return + shr bufp, 32 ; shift data down by 4 bytes +check_2: + shl ebx, 1 ; shift out into carry MSB (orig size & 2) + jnc check_1 + crc32 crc_init_dw, bufp_w + jz do_return + shr bufp, 16 ; shift data down by 2 bytes +check_1: + crc32 crc_init_dw, bufp_b + +do_return: + mov rax, crc_init + pop rsi + pop rdi + pop rbx + ret + +do_8: + crc32 crc_init, qword [bufptmp] + add bufptmp, 8 + shl ebx, 29 ; size (0...7) in bits 31:29 + jnz check_4 + mov rax, crc_init + pop rsi + pop rdi + pop rbx + ret + +do_16: + crc32 crc_init, qword [bufptmp] + crc32 crc_init, qword [bufptmp+8] + add bufptmp, 16 + shl ebx, 29 ; size (0...7) in bits 31:29 + jnz check_4 + mov rax, crc_init + pop rsi + pop rdi + pop rbx + ret + + + + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; Handle the case of fewer than 8 bytes, unaligned. In this case + ;; we can't read 8 bytes, as this might go beyond the end of the buffer + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +less_than_8: + test len,4 + jz less_than_4 + crc32 crc_init_dw, dword[bufptmp] + add bufptmp,4 +less_than_4: + test len,2 + jz less_than_2 + crc32 crc_init_dw, word[bufptmp] + add bufptmp,2 +less_than_2: + test len,1 + jz do_return + crc32 crc_init_dw, byte[bufptmp] + mov rax, crc_init + pop rsi + pop rdi + pop rbx + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;4) LESS THAN 256-bytes REMAIN AT THIS POINT (8-bits of len are full) + +small: + mov rax, crc_init + +bit8: + shl len_b, 1 ;; shift-out MSB (bit-7) + jnc bit7 ;; jump to bit-6 if bit-7 == 0 + %assign i 0 + %rep 16 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return2 ;; return if remaining data is zero + add bufptmp, 128 ;; buf +=64; (next 64 bytes) + +bit7: + shl len_b, 1 ;; shift-out MSB (bit-7) + jnc bit6 ;; jump to bit-6 if bit-7 == 0 + %assign i 0 + %rep 8 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return2 ;; return if remaining data is zero + add bufptmp, 64 ;; buf +=64; (next 64 bytes) +bit6: + shl len_b, 1 ;; shift-out MSB (bit-6) + jnc bit5 ;; jump to bit-5 if bit-6 == 0 + %assign i 0 + %rep 4 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return2 ;; return if remaining data is zero + add bufptmp, 32 ;; buf +=32; (next 32 bytes) +bit5: + shl len_b, 1 ;; shift-out MSB (bit-5) + jnc bit4 ;; jump to bit-4 if bit-5 == 0 + %assign i 0 + %rep 2 + crc32 rax, qword [bufptmp+i] ;; compute crc32 of 8-byte data + %assign i (i+8) + %endrep + je do_return2 ;; return if remaining data is zero + add bufptmp, 16 ;; buf +=16; (next 16 bytes) +bit4: + shl len_b, 1 ;; shift-out MSB (bit-4) + jnc bit3 ;; jump to bit-3 if bit-4 == 0 + crc32 rax, qword [bufptmp] ;; compute crc32 of 8-byte data + je do_return2 ;; return if remaining data is zero + add bufptmp, 8 ;; buf +=8; (next 8 bytes) +bit3: + mov rbx, qword [bufptmp] ;; load a 8-bytes from the buffer: + shl len_b, 1 ;; shift-out MSB (bit-3) + jnc bit2 ;; jump to bit-2 if bit-3 == 0 + crc32 eax, ebx ;; compute crc32 of 4-byte data + je do_return2 ;; return if remaining data is zero + shr rbx, 32 ;; get next 3 bytes +bit2: + shl len_b, 1 ;; shift-out MSB (bit-2) + jnc bit1 ;; jump to bit-1 if bit-2 == 0 + crc32 eax, bx ;; compute crc32 of 2-byte data + je do_return2 ;; return if remaining data is zero + shr rbx, 16 ;; next byte +bit1: + test len_b,len_b + je do_return2 + crc32 eax, bl ;; compute crc32 of 1-byte data +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +do_return2: + pop rsi + pop rdi + pop rbx + ret + + + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; jump table ;; Table is 129 entries x 2 bytes each + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +align 4 +jump_table: +%assign i 0 +%rep 129 + dw CONCAT(crc_,i,) - crc_array +%assign i (i+1) +%endrep + + + + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ;; PCLMULQDQ tables + ;; Table is 128 entries x 2 quad words each + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +section .data +align 64 +K_table: + dq 0x14cd00bd6, 0x105ec76f0 + dq 0x0ba4fc28e, 0x14cd00bd6 + dq 0x1d82c63da, 0x0f20c0dfe + dq 0x09e4addf8, 0x0ba4fc28e + dq 0x039d3b296, 0x1384aa63a + dq 0x102f9b8a2, 0x1d82c63da + dq 0x14237f5e6, 0x01c291d04 + dq 0x00d3b6092, 0x09e4addf8 + dq 0x0c96cfdc0, 0x0740eef02 + dq 0x18266e456, 0x039d3b296 + dq 0x0daece73e, 0x0083a6eec + dq 0x0ab7aff2a, 0x102f9b8a2 + dq 0x1248ea574, 0x1c1733996 + dq 0x083348832, 0x14237f5e6 + dq 0x12c743124, 0x02ad91c30 + dq 0x0b9e02b86, 0x00d3b6092 + dq 0x018b33a4e, 0x06992cea2 + dq 0x1b331e26a, 0x0c96cfdc0 + dq 0x17d35ba46, 0x07e908048 + dq 0x1bf2e8b8a, 0x18266e456 + dq 0x1a3e0968a, 0x11ed1f9d8 + dq 0x0ce7f39f4, 0x0daece73e + dq 0x061d82e56, 0x0f1d0f55e + dq 0x0d270f1a2, 0x0ab7aff2a + dq 0x1c3f5f66c, 0x0a87ab8a8 + dq 0x12ed0daac, 0x1248ea574 + dq 0x065863b64, 0x08462d800 + dq 0x11eef4f8e, 0x083348832 + dq 0x1ee54f54c, 0x071d111a8 + dq 0x0b3e32c28, 0x12c743124 + dq 0x0064f7f26, 0x0ffd852c6 + dq 0x0dd7e3b0c, 0x0b9e02b86 + dq 0x0f285651c, 0x0dcb17aa4 + dq 0x010746f3c, 0x018b33a4e + dq 0x1c24afea4, 0x0f37c5aee + dq 0x0271d9844, 0x1b331e26a + dq 0x08e766a0c, 0x06051d5a2 + dq 0x093a5f730, 0x17d35ba46 + dq 0x06cb08e5c, 0x11d5ca20e + dq 0x06b749fb2, 0x1bf2e8b8a + dq 0x1167f94f2, 0x021f3d99c + dq 0x0cec3662e, 0x1a3e0968a + dq 0x19329634a, 0x08f158014 + dq 0x0e6fc4e6a, 0x0ce7f39f4 + dq 0x08227bb8a, 0x1a5e82106 + dq 0x0b0cd4768, 0x061d82e56 + dq 0x13c2b89c4, 0x188815ab2 + dq 0x0d7a4825c, 0x0d270f1a2 + dq 0x10f5ff2ba, 0x105405f3e + dq 0x00167d312, 0x1c3f5f66c + dq 0x0f6076544, 0x0e9adf796 + dq 0x026f6a60a, 0x12ed0daac + dq 0x1a2adb74e, 0x096638b34 + dq 0x19d34af3a, 0x065863b64 + dq 0x049c3cc9c, 0x1e50585a0 + dq 0x068bce87a, 0x11eef4f8e + dq 0x1524fa6c6, 0x19f1c69dc + dq 0x16cba8aca, 0x1ee54f54c + dq 0x042d98888, 0x12913343e + dq 0x1329d9f7e, 0x0b3e32c28 + dq 0x1b1c69528, 0x088f25a3a + dq 0x02178513a, 0x0064f7f26 + dq 0x0e0ac139e, 0x04e36f0b0 + dq 0x0170076fa, 0x0dd7e3b0c + dq 0x141a1a2e2, 0x0bd6f81f8 + dq 0x16ad828b4, 0x0f285651c + dq 0x041d17b64, 0x19425cbba + dq 0x1fae1cc66, 0x010746f3c + dq 0x1a75b4b00, 0x18db37e8a + dq 0x0f872e54c, 0x1c24afea4 + dq 0x01e41e9fc, 0x04c144932 + dq 0x086d8e4d2, 0x0271d9844 + dq 0x160f7af7a, 0x052148f02 + dq 0x05bb8f1bc, 0x08e766a0c + dq 0x0a90fd27a, 0x0a3c6f37a + dq 0x0b3af077a, 0x093a5f730 + dq 0x04984d782, 0x1d22c238e + dq 0x0ca6ef3ac, 0x06cb08e5c + dq 0x0234e0b26, 0x063ded06a + dq 0x1d88abd4a, 0x06b749fb2 + dq 0x04597456a, 0x04d56973c + dq 0x0e9e28eb4, 0x1167f94f2 + dq 0x07b3ff57a, 0x19385bf2e + dq 0x0c9c8b782, 0x0cec3662e + dq 0x13a9cba9e, 0x0e417f38a + dq 0x093e106a4, 0x19329634a + dq 0x167001a9c, 0x14e727980 + dq 0x1ddffc5d4, 0x0e6fc4e6a + dq 0x00df04680, 0x0d104b8fc + dq 0x02342001e, 0x08227bb8a + dq 0x00a2a8d7e, 0x05b397730 + dq 0x168763fa6, 0x0b0cd4768 + dq 0x1ed5a407a, 0x0e78eb416 + dq 0x0d2c3ed1a, 0x13c2b89c4 + dq 0x0995a5724, 0x1641378f0 + dq 0x19b1afbc4, 0x0d7a4825c + dq 0x109ffedc0, 0x08d96551c + dq 0x0f2271e60, 0x10f5ff2ba + dq 0x00b0bf8ca, 0x00bf80dd2 + dq 0x123888b7a, 0x00167d312 + dq 0x1e888f7dc, 0x18dcddd1c + dq 0x002ee03b2, 0x0f6076544 + dq 0x183e8d8fe, 0x06a45d2b2 + dq 0x133d7a042, 0x026f6a60a + dq 0x116b0f50c, 0x1dd3e10e8 + dq 0x05fabe670, 0x1a2adb74e + dq 0x130004488, 0x0de87806c + dq 0x000bcf5f6, 0x19d34af3a + dq 0x18f0c7078, 0x014338754 + dq 0x017f27698, 0x049c3cc9c + dq 0x058ca5f00, 0x15e3e77ee + dq 0x1af900c24, 0x068bce87a + dq 0x0b5cfca28, 0x0dd07448e + dq 0x0ded288f8, 0x1524fa6c6 + dq 0x059f229bc, 0x1d8048348 + dq 0x06d390dec, 0x16cba8aca + dq 0x037170390, 0x0a3e3e02c + dq 0x06353c1cc, 0x042d98888 + dq 0x0c4584f5c, 0x0d73c7bea + dq 0x1f16a3418, 0x1329d9f7e + dq 0x0531377e2, 0x185137662 + dq 0x1d8d9ca7c, 0x1b1c69528 + dq 0x0b25b29f2, 0x18a08b5bc + dq 0x19fb2a8b0, 0x02178513a + dq 0x1a08fe6ac, 0x1da758ae0 + dq 0x045cddf4e, 0x0e0ac139e + dq 0x1a91647f2, 0x169cf9eb0 + dq 0x1a0f717c4, 0x0170076fa + +;;; func core, ver, snum +slversion crc32_iscsi_01, 01, 03, 0015 + diff --git a/src/isa-l/crc/crc32_iscsi_perf.c b/src/isa-l/crc/crc32_iscsi_perf.c new file mode 100644 index 00000000..2770099a --- /dev/null +++ b/src/isa-l/crc/crc32_iscsi_perf.c @@ -0,0 +1,87 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <sys/time.h> +#include "crc.h" +#include "test.h" + +//#define CACHED_TEST +#ifdef CACHED_TEST +// Cached test, loop many times over small dataset +# define TEST_LEN 8*1024 +# define TEST_LOOPS 1000000 +# define TEST_TYPE_STR "_warm" +#else +// Uncached test. Pull from large mem base. +# define GT_L3_CACHE 32*1024*1024 /* some number > last level cache */ +# define TEST_LEN (2 * GT_L3_CACHE) +# define TEST_LOOPS 500 +# define TEST_TYPE_STR "_cold" +#endif + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define TEST_MEM TEST_LEN + +int main(int argc, char *argv[]) +{ + int i; + void *buf; + uint32_t crc; + struct perf start, stop; + + printf("crc32_iscsi_perf:\n"); + + if (posix_memalign(&buf, 1024, TEST_LEN)) { + printf("alloc error: Fail"); + return -1; + } + + printf("Start timed tests\n"); + fflush(0); + + memset(buf, 0, TEST_LEN); + crc = crc32_iscsi(buf, TEST_LEN, TEST_SEED); + perf_start(&start); + for (i = 0; i < TEST_LOOPS; i++) { + crc = crc32_iscsi(buf, TEST_LEN, TEST_SEED); + } + perf_stop(&stop); + printf("crc32_iscsi" TEST_TYPE_STR ": "); + perf_print(stop, start, (long long)TEST_LEN * i); + + printf("finish 0x%x\n", crc); + return 0; +} diff --git a/src/isa-l/crc/crc32_iscsi_test.c b/src/isa-l/crc/crc32_iscsi_test.c new file mode 100644 index 00000000..e37f23f1 --- /dev/null +++ b/src/isa-l/crc/crc32_iscsi_test.c @@ -0,0 +1,171 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include "crc.h" +#include "types.h" + +unsigned long crc32_table_iscsi[256] = { + 0x00000000, 0xF26B8303, 0xE13B70F7, 0x1350F3F4, + 0xC79A971F, 0x35F1141C, 0x26A1E7E8, 0xD4CA64EB, + 0x8AD958CF, 0x78B2DBCC, 0x6BE22838, 0x9989AB3B, + 0x4D43CFD0, 0xBF284CD3, 0xAC78BF27, 0x5E133C24, + 0x105EC76F, 0xE235446C, 0xF165B798, 0x030E349B, + 0xD7C45070, 0x25AFD373, 0x36FF2087, 0xC494A384, + 0x9A879FA0, 0x68EC1CA3, 0x7BBCEF57, 0x89D76C54, + 0x5D1D08BF, 0xAF768BBC, 0xBC267848, 0x4E4DFB4B, + 0x20BD8EDE, 0xD2D60DDD, 0xC186FE29, 0x33ED7D2A, + 0xE72719C1, 0x154C9AC2, 0x061C6936, 0xF477EA35, + 0xAA64D611, 0x580F5512, 0x4B5FA6E6, 0xB93425E5, + 0x6DFE410E, 0x9F95C20D, 0x8CC531F9, 0x7EAEB2FA, + 0x30E349B1, 0xC288CAB2, 0xD1D83946, 0x23B3BA45, + 0xF779DEAE, 0x05125DAD, 0x1642AE59, 0xE4292D5A, + 0xBA3A117E, 0x4851927D, 0x5B016189, 0xA96AE28A, + 0x7DA08661, 0x8FCB0562, 0x9C9BF696, 0x6EF07595, + 0x417B1DBC, 0xB3109EBF, 0xA0406D4B, 0x522BEE48, + 0x86E18AA3, 0x748A09A0, 0x67DAFA54, 0x95B17957, + 0xCBA24573, 0x39C9C670, 0x2A993584, 0xD8F2B687, + 0x0C38D26C, 0xFE53516F, 0xED03A29B, 0x1F682198, + 0x5125DAD3, 0xA34E59D0, 0xB01EAA24, 0x42752927, + 0x96BF4DCC, 0x64D4CECF, 0x77843D3B, 0x85EFBE38, + 0xDBFC821C, 0x2997011F, 0x3AC7F2EB, 0xC8AC71E8, + 0x1C661503, 0xEE0D9600, 0xFD5D65F4, 0x0F36E6F7, + 0x61C69362, 0x93AD1061, 0x80FDE395, 0x72966096, + 0xA65C047D, 0x5437877E, 0x4767748A, 0xB50CF789, + 0xEB1FCBAD, 0x197448AE, 0x0A24BB5A, 0xF84F3859, + 0x2C855CB2, 0xDEEEDFB1, 0xCDBE2C45, 0x3FD5AF46, + 0x7198540D, 0x83F3D70E, 0x90A324FA, 0x62C8A7F9, + 0xB602C312, 0x44694011, 0x5739B3E5, 0xA55230E6, + 0xFB410CC2, 0x092A8FC1, 0x1A7A7C35, 0xE811FF36, + 0x3CDB9BDD, 0xCEB018DE, 0xDDE0EB2A, 0x2F8B6829, + 0x82F63B78, 0x709DB87B, 0x63CD4B8F, 0x91A6C88C, + 0x456CAC67, 0xB7072F64, 0xA457DC90, 0x563C5F93, + 0x082F63B7, 0xFA44E0B4, 0xE9141340, 0x1B7F9043, + 0xCFB5F4A8, 0x3DDE77AB, 0x2E8E845F, 0xDCE5075C, + 0x92A8FC17, 0x60C37F14, 0x73938CE0, 0x81F80FE3, + 0x55326B08, 0xA759E80B, 0xB4091BFF, 0x466298FC, + 0x1871A4D8, 0xEA1A27DB, 0xF94AD42F, 0x0B21572C, + 0xDFEB33C7, 0x2D80B0C4, 0x3ED04330, 0xCCBBC033, + 0xA24BB5A6, 0x502036A5, 0x4370C551, 0xB11B4652, + 0x65D122B9, 0x97BAA1BA, 0x84EA524E, 0x7681D14D, + 0x2892ED69, 0xDAF96E6A, 0xC9A99D9E, 0x3BC21E9D, + 0xEF087A76, 0x1D63F975, 0x0E330A81, 0xFC588982, + 0xB21572C9, 0x407EF1CA, 0x532E023E, 0xA145813D, + 0x758FE5D6, 0x87E466D5, 0x94B49521, 0x66DF1622, + 0x38CC2A06, 0xCAA7A905, 0xD9F75AF1, 0x2B9CD9F2, + 0xFF56BD19, 0x0D3D3E1A, 0x1E6DCDEE, 0xEC064EED, + 0xC38D26C4, 0x31E6A5C7, 0x22B65633, 0xD0DDD530, + 0x0417B1DB, 0xF67C32D8, 0xE52CC12C, 0x1747422F, + 0x49547E0B, 0xBB3FFD08, 0xA86F0EFC, 0x5A048DFF, + 0x8ECEE914, 0x7CA56A17, 0x6FF599E3, 0x9D9E1AE0, + 0xD3D3E1AB, 0x21B862A8, 0x32E8915C, 0xC083125F, + 0x144976B4, 0xE622F5B7, 0xF5720643, 0x07198540, + 0x590AB964, 0xAB613A67, 0xB831C993, 0x4A5A4A90, + 0x9E902E7B, 0x6CFBAD78, 0x7FAB5E8C, 0x8DC0DD8F, + 0xE330A81A, 0x115B2B19, 0x020BD8ED, 0xF0605BEE, + 0x24AA3F05, 0xD6C1BC06, 0xC5914FF2, 0x37FACCF1, + 0x69E9F0D5, 0x9B8273D6, 0x88D28022, 0x7AB90321, + 0xAE7367CA, 0x5C18E4C9, 0x4F48173D, 0xBD23943E, + 0xF36E6F75, 0x0105EC76, 0x12551F82, 0xE03E9C81, + 0x34F4F86A, 0xC69F7B69, 0xD5CF889D, 0x27A40B9E, + 0x79B737BA, 0x8BDCB4B9, 0x988C474D, 0x6AE7C44E, + 0xBE2DA0A5, 0x4C4623A6, 0x5F16D052, 0xAD7D5351, +}; + +#define PAGESIZE 10240 + +int main(void) +{ + unsigned int i, j, good, test, init_crc = 1; + + printf("crc32_iscsi_test: "); + + unsigned char *q_buf = malloc(PAGESIZE); + if (q_buf == NULL) { + printf("alloc of q_buf failed\n"); + return -1; + } + // fill q_buf with semi-random data + for (i = 0; i < PAGESIZE; i++) + q_buf[i] = (unsigned char)(i ^ (13 + (i >> 8)) ^ ((i >> 16) - 13)); + + // Test case 1: Compare against base/simple crc32 implementation and + // try all offsets/alignments of buffer. + + for (j = 0; j < 128; j++) { + for (i = 0; i < PAGESIZE - j; i++) { + good = crc32_iscsi_base(q_buf + j, i, -1); + test = crc32_iscsi(q_buf + j, i, -1); + if (good != test) { + printf("Error for size %d offset %d, %08X should be %08X\n", + i, j, test, good); + return -1; + } + } // end for i + putchar('.'); + fflush(0); + } // end for j + + // Test case 2: Also vary initial CRC + + for (j = 0; j < 128; j++) { // do all offsets + for (i = 0; i < PAGESIZE - j; i++) { + good = crc32_iscsi_base(q_buf + j, i, init_crc); + test = crc32_iscsi(q_buf + j, i, init_crc); + if (good != test) { + printf("Error for size %d offset %d, %08X should be %08X\n", + i, j, test, good); + return -1; + } + // modify init_crc semi-randomly + init_crc ^= 1 << ((i * 3 + j * 5) & 31); + } // end for i + putchar('.'); + fflush(0); + } // end for j + + // Test case 3: do end of buffer + + for (i = 0; i < PAGESIZE; i++) { + good = crc32_iscsi_base(q_buf + i, PAGESIZE - i, -1); + test = crc32_iscsi(q_buf + i, PAGESIZE - i, -1); + if (good != test) { + printf("Error for size %d at eob, %08X should be %08X\n", + i, test, good); + return -1; + } + } // end for i + putchar('.'); + fflush(0); + + printf("Pass\n"); + + return 0; +} diff --git a/src/isa-l/crc/crc64_base.c b/src/isa-l/crc/crc64_base.c new file mode 100644 index 00000000..29166f90 --- /dev/null +++ b/src/isa-l/crc/crc64_base.c @@ -0,0 +1,159 @@ +/********************************************************************** + 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. +**********************************************************************/ + +#include <stdlib.h> +#include "crc64.h" + +#define MAX_ITER 8 + +// crc64_ecma baseline function +// Slow crc64 from the definition. Can be sped up with a lookup table. +uint64_t crc64_ecma_refl_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0xC96C5795D7870F42ULL; // ECMA-182 standard reflected + + for (i = 0; i < len; i++) { + rem = rem ^ (uint64_t) buf[i]; + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x1ULL ? poly : 0) ^ (rem >> 1); + } + } + return ~rem; +} + +uint64_t crc64_ecma_norm_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0x42F0E1EBA9EA3693ULL; // ECMA-182 standard + + for (i = 0; i < len; i++) { + rem = rem ^ ((uint64_t) buf[i] << 56); + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x8000000000000000ULL ? poly : 0) ^ (rem << 1); + } + } + return ~rem; +} + +// crc64_iso baseline function +// Slow crc64 from the definition. Can be sped up with a lookup table. +uint64_t crc64_iso_refl_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0xD800000000000000ULL; // ISO standard reflected + + for (i = 0; i < len; i++) { + rem = rem ^ (uint64_t) buf[i]; + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x1ULL ? poly : 0) ^ (rem >> 1); + } + } + return ~rem; +} + +uint64_t crc64_iso_norm_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0x000000000000001BULL; // ISO standard + + for (i = 0; i < len; i++) { + rem = rem ^ ((uint64_t) buf[i] << 56); + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x8000000000000000ULL ? poly : 0) ^ (rem << 1); + } + } + return ~rem; +} + +// crc64_jones baseline function +// Slow crc64 from the definition. Can be sped up with a lookup table. +uint64_t crc64_jones_refl_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0x95ac9329ac4bc9b5ULL; // Jones coefficients reflected + + for (i = 0; i < len; i++) { + rem = rem ^ (uint64_t) buf[i]; + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x1ULL ? poly : 0) ^ (rem >> 1); + } + } + return ~rem; +} + +uint64_t crc64_jones_norm_base(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint64_t poly = 0xad93d23594c935a9ULL; // Jones coefficients + + for (i = 0; i < len; i++) { + rem = rem ^ ((uint64_t) buf[i] << 56); + for (j = 0; j < MAX_ITER; j++) { + rem = (rem & 0x8000000000000000ULL ? poly : 0) ^ (rem << 1); + } + } + return ~rem; +} + +struct slver { + unsigned short snum; + unsigned char ver; + unsigned char core; +}; + +struct slver crc64_ecma_refl_base_slver_0000001c; +struct slver crc64_ecma_refl_base_slver = { 0x001c, 0x00, 0x00 }; + +struct slver crc64_ecma_norm_base_slver_00000019; +struct slver crc64_ecma_norm_base_slver = { 0x0019, 0x00, 0x00 }; + +struct slver crc64_iso_refl_base_slver_00000022; +struct slver crc64_iso_refl_base_slver = { 0x0022, 0x00, 0x00 }; + +struct slver crc64_iso_norm_base_slver_0000001f; +struct slver crc64_iso_norm_base_slver = { 0x001f, 0x00, 0x00 }; + +struct slver crc64_jones_refl_base_slver_00000028; +struct slver crc64_jones_refl_base_slver = { 0x0028, 0x00, 0x00 }; + +struct slver crc64_jones_norm_base_slver_00000025; +struct slver crc64_jones_norm_base_slver = { 0x0025, 0x00, 0x00 }; diff --git a/src/isa-l/crc/crc64_ecma_norm_by8.asm b/src/isa-l/crc/crc64_ecma_norm_by8.asm new file mode 100644 index 00000000..cff01e1c --- /dev/null +++ b/src/isa-l/crc/crc64_ecma_norm_by8.asm @@ -0,0 +1,583 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function API: +; uint64_t crc64_ecma_norm_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +; yasm -f x64 -f elf64 -X gnu -g dwarf2 crc64_ecma_norm_by8 +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif +align 16 +global crc64_ecma_norm_by8:function +crc64_ecma_norm_by8: + + not arg1 ;~init_crc + + sub rsp,VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + + ; check if smaller than 256 + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register. + ; because data will be byte-reflected and will align with initial crc at correct place. + pslldq xmm10, 8 + + movdqa xmm11, [SHUF_MASK] + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + pshufb xmm0, xmm11 + ; XOR the initial_crc value + pxor xmm0, xmm10 + pshufb xmm1, xmm11 + pshufb xmm2, xmm11 + pshufb xmm3, xmm11 + pshufb xmm4, xmm11 + pshufb xmm5, xmm11 + pshufb xmm6, xmm11 + pshufb xmm7, xmm11 + + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 ; buf += 128; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm1, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm3, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm5, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm7, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + + movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2 + + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pshufb xmm0, xmm11 + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm7 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm11 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm7 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm7, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10 + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0x01 ; H*L + pslldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10 + movdqa xmm0, xmm7 + + movdqa xmm1, xmm7 + pand xmm1, [mask3] + pclmulqdq xmm7, xmm10, 0x01 + pxor xmm7, xmm1 + + pclmulqdq xmm7, xmm10, 0x11 + pxor xmm7, xmm0 + pextrq rax, xmm7, 0 + +_cleanup: + not rax +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm11, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movdqa xmm11, [SHUF_MASK] + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + ; shl r9, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm7, xmm0 + jmp _128_done + +_end_1to7: + ; Right shift (8-length) bytes in XMM + add rax, 8 + movdqu xmm0, [rax] + pshufb xmm7,xmm0 + + jmp _barrett +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 + +rk1 : +DQ 0x5f5c3c7eb52fab6 +rk2 : +DQ 0x4eb938a7d257740e +rk3 : +DQ 0x5cf79dea9ac37d6 +rk4 : +DQ 0x001067e571d7d5c2 +rk5 : +DQ 0x5f5c3c7eb52fab6 +rk6 : +DQ 0x0000000000000000 +rk7 : +DQ 0x578d29d06cc4f872 +rk8 : +DQ 0x42f0e1eba9ea3693 +rk9 : +DQ 0xe464f4df5fb60ac1 +rk10 : +DQ 0xb649c5b35a759cf2 +rk11 : +DQ 0x9af04e1eff82d0dd +rk12 : +DQ 0x6e82e609297f8fe8 +rk13 : +DQ 0x97c516e98bd2e73 +rk14 : +DQ 0xb76477b31e22e7b +rk15 : +DQ 0x5f6843ca540df020 +rk16 : +DQ 0xddf4b6981205b83f +rk17 : +DQ 0x54819d8713758b2c +rk18 : +DQ 0x4a6b90073eb0af5a +rk19 : +DQ 0x571bee0a227ef92b +rk20 : +DQ 0x44bef2a201b5200c + + +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF +mask3: +dq 0x0000000000000000, 0xFFFFFFFFFFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_ecma_norm_by8, 01, 00, 001a diff --git a/src/isa-l/crc/crc64_ecma_refl_by8.asm b/src/isa-l/crc/crc64_ecma_refl_by8.asm new file mode 100644 index 00000000..9d3847e9 --- /dev/null +++ b/src/isa-l/crc/crc64_ecma_refl_by8.asm @@ -0,0 +1,548 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Function API: +; uint64_t crc64_ecma_refl_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +; Reference paper titled "Fast CRC Computation for Generic Polynomials Using PCLMULQDQ Instruction" +; sample yasm command line: +; yasm -f x64 -f elf64 -X gnu -g dwarf2 crc64_ecma_refl_by8 +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif + + +align 16 +global crc64_ecma_refl_by8:function +crc64_ecma_refl_by8: + ; uint64_t c = crc ^ 0xffffffff,ffffffffL; + not arg1 + sub rsp, VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + ; check if smaller than 256B + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + ; XOR the initial_crc value + pxor xmm0, xmm10 + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm1, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm3, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm5, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm7, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + ; xmm0 to xmm7 + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + ;xmm1 to xmm7 + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + ; xmm6 to xmm7 + movdqa xmm10, [rk1] + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 128-16 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + add arg3, 16 + je _128_done + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + + + movdqa xmm2, xmm7 + movdqu xmm1, [arg2 - 16 + arg3] + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table] + add rax, arg3 + movdqu xmm0, [rax] + + + pshufb xmm7, xmm0 + pxor xmm0, [mask3] + pshufb xmm2, xmm0 + + pblendvb xmm2, xmm1 ;xmm0 is implicit + ;;;;;;;;;; + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0 + psrldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm1, xmm7 + movdqa xmm10, [rk7] + + pclmulqdq xmm7, xmm10, 0 + movdqa xmm2, xmm7 + pclmulqdq xmm7, xmm10, 0x10 + pslldq xmm2, 8 + pxor xmm7, xmm2 + pxor xmm7, xmm1 + pextrq rax, xmm7, 1 + +_cleanup: + ; return c ^ 0xffffffff, ffffffffL; + not rax + + +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movq xmm0, arg1 ; get the initial crc value + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + +_zero_left: + movdqa xmm7, [rsp] + pxor xmm7, xmm0 ; xor the initial crc value + + lea rax,[pshufb_shf_table] + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax + r9] + pshufb xmm7,xmm0 + jmp _128_done + +_end_1to7: + ; Left shift (8-length) bytes in XMM + movdqu xmm0, [rax + r9 + 8] + pshufb xmm7,xmm0 + + jmp _barrett + +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 +; rk7 = floor(2^128/Q) +; rk8 = Q +rk1 : +DQ 0xdabe95afc7875f40 +rk2 : +DQ 0xe05dd497ca393ae4 +rk3 : +DQ 0xd7d86b2af73de740 +rk4 : +DQ 0x8757d71d4fcc1000 +rk5 : +DQ 0xdabe95afc7875f40 +rk6 : +DQ 0x0000000000000000 +rk7 : +DQ 0x9c3e466c172963d5 +rk8 : +DQ 0x92d8af2baf0e1e84 +rk9 : +DQ 0x947874de595052cb +rk10 : +DQ 0x9e735cb59b4724da +rk11 : +DQ 0xe4ce2cd55fea0037 +rk12 : +DQ 0x2fe3fd2920ce82ec +rk13 : +DQ 0xe31d519421a63a5 +rk14 : +DQ 0x2e30203212cac325 +rk15 : +DQ 0x81f6054a7842df4 +rk16 : +DQ 0x6ae3efbb9dd441f3 +rk17 : +DQ 0x69a35d91c3730254 +rk18 : +DQ 0xb5ea1af9c013aca4 +rk19 : +DQ 0x3be653a30fe1af51 +rk20 : +DQ 0x60095b008a9efa44 + + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + + +mask: +dq 0xFFFFFFFFFFFFFFFF, 0x0000000000000000 +mask2: +dq 0xFFFFFFFF00000000, 0xFFFFFFFFFFFFFFFF +mask3: +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_ecma_refl_by8, 01, 00, 001d diff --git a/src/isa-l/crc/crc64_example.c b/src/isa-l/crc/crc64_example.c new file mode 100644 index 00000000..64763a1b --- /dev/null +++ b/src/isa-l/crc/crc64_example.c @@ -0,0 +1,68 @@ +/********************************************************************** + 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. +**********************************************************************/ +#include <stdio.h> +#include <stdlib.h> +#include <assert.h> +#include "crc64.h" + +#define BUF_SIZE 8192 +#define INIT_SEED 0x12345678 + +int main(int argc, char *argv[]) +{ + uint8_t inbuf[BUF_SIZE]; + uint64_t avail_in, total_in = 0; + uint64_t crc64_checksum; + FILE *in; + + if (argc != 2) { + fprintf(stderr, "Usage: crc64_example infile\n"); + exit(0); + } + in = fopen(argv[1], "rb"); + if (!in) { + fprintf(stderr, "Can't open %s for reading\n", argv[1]); + exit(0); + } + + printf("crc64_example -- crc64_ecma_refl:\n"); + fflush(0); + + crc64_checksum = INIT_SEED; + while ((avail_in = fread(inbuf, 1, BUF_SIZE, in))) { + // crc update mode + crc64_checksum = crc64_ecma_refl(crc64_checksum, inbuf, avail_in); + total_in += avail_in; + } + + fclose(in); + printf("total length is %ld, checksum is 0x%lx\n", total_in, crc64_checksum); + + return 0; +} diff --git a/src/isa-l/crc/crc64_funcs_perf.c b/src/isa-l/crc/crc64_funcs_perf.c new file mode 100644 index 00000000..04135bff --- /dev/null +++ b/src/isa-l/crc/crc64_funcs_perf.c @@ -0,0 +1,109 @@ +/********************************************************************** + 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include <sys/time.h> +#include "crc64.h" +#include "test.h" + +//#define CACHED_TEST +#ifdef CACHED_TEST +// Cached test, loop many times over small dataset +# define TEST_LEN 8*1024 +# define TEST_LOOPS 400000 +# define TEST_TYPE_STR "_warm" +#else +// Uncached test. Pull from large mem base. +# define GT_L3_CACHE 32*1024*1024 /* some number > last level cache */ +# define TEST_LEN (2 * GT_L3_CACHE) +# define TEST_LOOPS 100 +# define TEST_TYPE_STR "_cold" +#endif + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define TEST_MEM TEST_LEN + +typedef uint64_t(*crc64_func_t) (uint64_t, const uint8_t *, uint64_t); + +typedef struct func_case { + char *note; + crc64_func_t crc64_func_call; + crc64_func_t crc64_ref_call; +} func_case_t; + +func_case_t test_funcs[] = { + {"crc64_ecma_norm", crc64_ecma_norm, crc64_ecma_norm_base}, + {"crc64_ecma_refl", crc64_ecma_refl, crc64_ecma_refl_base}, + {"crc64_iso_norm", crc64_iso_norm, crc64_iso_norm_base}, + {"crc64_iso_refl", crc64_iso_refl, crc64_iso_refl_base}, + {"crc64_jones_norm", crc64_jones_norm, crc64_jones_norm_base}, + {"crc64_jones_refl", crc64_jones_refl, crc64_jones_refl_base} +}; + +int main(int argc, char *argv[]) +{ + int i, j; + void *buf; + uint64_t crc; + struct perf start, stop; + func_case_t *test_func; + + if (posix_memalign(&buf, 1024, TEST_LEN)) { + printf("alloc error: Fail"); + return -1; + } + memset(buf, (char)TEST_SEED, TEST_LEN); + + for (j = 0; j < sizeof(test_funcs) / sizeof(test_funcs[0]); j++) { + test_func = &test_funcs[j]; + printf("%s_perf:\n", test_func->note); + + printf("Start timed tests\n"); + fflush(0); + + crc = test_func->crc64_func_call(TEST_SEED, buf, TEST_LEN); + perf_start(&start); + for (i = 0; i < TEST_LOOPS; i++) { + crc = test_func->crc64_func_call(TEST_SEED, buf, TEST_LEN); + } + perf_stop(&stop); + printf("%s" TEST_TYPE_STR ": ", test_func->note); + perf_print(stop, start, (long long)TEST_LEN * i); + + printf("finish 0x%lx\n", crc); + } + + return 0; +} diff --git a/src/isa-l/crc/crc64_funcs_test.c b/src/isa-l/crc/crc64_funcs_test.c new file mode 100644 index 00000000..f638f0f9 --- /dev/null +++ b/src/isa-l/crc/crc64_funcs_test.c @@ -0,0 +1,290 @@ +/********************************************************************** + 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. +**********************************************************************/ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdint.h> +#include "crc64.h" +#include "types.h" + +#ifndef TEST_SEED +# define TEST_SEED 0x1234 +#endif + +#define MAX_BUF 512 +#define TEST_SIZE 20 + +typedef uint64_t u64; +typedef uint32_t u32; +typedef uint16_t u16; +typedef uint8_t u8; + +typedef uint64_t(*crc64_func_t) (uint64_t, const uint8_t *, uint64_t); + +typedef struct func_case { + char *note; + crc64_func_t crc64_func_call; + crc64_func_t crc64_ref_call; +} func_case_t; + +func_case_t test_funcs[] = { + {"crc64_ecma_norm", crc64_ecma_norm, crc64_ecma_norm_base}, + {"crc64_ecma_refl", crc64_ecma_refl, crc64_ecma_refl_base}, + {"crc64_iso_norm", crc64_iso_norm, crc64_iso_norm_base}, + {"crc64_iso_refl", crc64_iso_refl, crc64_iso_refl_base}, + {"crc64_jones_norm", crc64_jones_norm, crc64_jones_norm_base}, + {"crc64_jones_refl", crc64_jones_refl, crc64_jones_refl_base} +}; + +// Generates pseudo-random data + +void rand_buffer(unsigned char *buf, long buffer_size) +{ + long i; + for (i = 0; i < buffer_size; i++) + buf[i] = rand(); +} + +// Test cases +int zeros_test(func_case_t * test_func); + +int simple_pattern_test(func_case_t * test_func); + +int seeds_sizes_test(func_case_t * test_func); + +int eob_test(func_case_t * test_func); + +int update_test(func_case_t * test_func); + +int verbose = 0; +void *buf_alloc = NULL; + +int main(int argc, char *argv[]) +{ + int fail = 0, fail_case; + int i, ret; + func_case_t *test_func; + + verbose = argc - 1; + + // Align to MAX_BUF boundary + ret = posix_memalign(&buf_alloc, MAX_BUF, MAX_BUF * TEST_SIZE); + if (ret) { + printf("alloc error: Fail"); + return -1; + } + srand(TEST_SEED); + printf("CRC64 Tests\n"); + + for (i = 0; i < sizeof(test_funcs) / sizeof(test_funcs[0]); i++) { + fail_case = 0; + test_func = &test_funcs[i]; + + printf("Test %s ", test_func->note); + fail_case += zeros_test(test_func); + fail_case += simple_pattern_test(test_func); + fail_case += seeds_sizes_test(test_func); + fail_case += eob_test(test_func); + fail_case += update_test(test_func); + printf("Test %s done: %s\n", test_func->note, fail_case ? "Fail" : "Pass"); + + if (fail_case) { + printf("\n%s Failed %d tests\n", test_func->note, fail_case); + fail++; + } + } + + printf("CRC64 Tests all done: %s\n", fail ? "Fail" : "Pass"); + + return fail; +} + +// Test of all zeros +int zeros_test(func_case_t * test_func) +{ + uint64_t crc, crc_ref; + int fail = 0; + unsigned char *buf = NULL; + + buf = (unsigned char *)buf_alloc; + memset(buf, 0, MAX_BUF * 10); + crc = test_func->crc64_func_call(TEST_SEED, buf, MAX_BUF * 10); + crc_ref = test_func->crc64_ref_call(TEST_SEED, buf, MAX_BUF * 10); + + if (crc != crc_ref) { + fail++; + printf("\n opt ref\n"); + printf(" ------ ------\n"); + printf("crc zero = 0x%16lx 0x%16lx \n", crc, crc_ref); + } else + printf("."); + + return fail; +} + +// Another simple test pattern +int simple_pattern_test(func_case_t * test_func) +{ + uint64_t crc, crc_ref; + int fail = 0; + unsigned char *buf = NULL; + + buf = (unsigned char *)buf_alloc; + memset(buf, 0x8a, MAX_BUF); + crc = test_func->crc64_func_call(TEST_SEED, buf, MAX_BUF); + crc_ref = test_func->crc64_ref_call(TEST_SEED, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc all 8a = 0x%16lx 0x%16lx\n", crc, crc_ref); + else + printf("."); + + return fail; +} + +int seeds_sizes_test(func_case_t * test_func) +{ + uint64_t crc, crc_ref; + int fail = 0; + int i; + uint64_t r, s; + unsigned char *buf = NULL; + + // Do a few random tests + buf = (unsigned char *)buf_alloc; //reset buf + r = rand(); + rand_buffer(buf, MAX_BUF * TEST_SIZE); + + for (i = 0; i < TEST_SIZE; i++) { + crc = test_func->crc64_func_call(r, buf, MAX_BUF); + crc_ref = test_func->crc64_ref_call(r, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%16lx 0x%16lx\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + + // Do a few random sizes + buf = (unsigned char *)buf_alloc; //reset buf + r = rand(); + + for (i = MAX_BUF; i >= 0; i--) { + crc = test_func->crc64_func_call(r, buf, i); + crc_ref = test_func->crc64_ref_call(r, buf, i); + if (crc != crc_ref) { + fail++; + printf("fail random size%i 0x%16lx 0x%16lx\n", i, crc, crc_ref); + } else + printf("."); + } + + // Try different seeds + for (s = 0; s < 20; s++) { + buf = (unsigned char *)buf_alloc; //reset buf + + r = rand(); // just to get a new seed + rand_buffer(buf, MAX_BUF * TEST_SIZE); // new pseudo-rand data + + if (verbose) + printf("seed = 0x%lx\n", r); + + for (i = 0; i < TEST_SIZE; i++) { + crc = test_func->crc64_func_call(r, buf, MAX_BUF); + crc_ref = test_func->crc64_ref_call(r, buf, MAX_BUF); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%16lx 0x%16lx\n", i, crc, crc_ref); + else + printf("."); + buf += MAX_BUF; + } + } + + return fail; +} + +// Run tests at end of buffer +int eob_test(func_case_t * test_func) +{ + uint64_t crc, crc_ref; + int fail = 0; + int i; + unsigned char *buf = NULL; + + buf = (unsigned char *)buf_alloc; //reset buf + buf = buf + ((MAX_BUF - 1) * TEST_SIZE); //Line up TEST_SIZE from end + for (i = 0; i < TEST_SIZE; i++) { + crc = test_func->crc64_func_call(TEST_SEED, buf + i, TEST_SIZE - i); + crc_ref = test_func->crc64_ref_call(TEST_SEED, buf + i, TEST_SIZE - i); + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc eob rand%3d = 0x%16lx 0x%16lx\n", i, crc, crc_ref); + else + printf("."); + } + + return fail; +} + +int update_test(func_case_t * test_func) +{ + uint64_t crc, crc_ref; + int fail = 0; + int i; + uint64_t r; + unsigned char *buf = NULL; + + buf = (unsigned char *)buf_alloc; //reset buf + r = rand(); + // Process the whole buf with reference func single call. + crc_ref = test_func->crc64_ref_call(r, buf, MAX_BUF * TEST_SIZE); + // Process buf with update method. + for (i = 0; i < TEST_SIZE; i++) { + crc = test_func->crc64_func_call(r, buf, MAX_BUF); + // Update crc seeds and buf pointer. + r = crc; + buf += MAX_BUF; + } + + if (crc != crc_ref) + fail++; + if (verbose) + printf("crc rand%3d = 0x%16lx 0x%16lx\n", i, crc, crc_ref); + else + printf("."); + + return fail; +} diff --git a/src/isa-l/crc/crc64_iso_norm_by8.asm b/src/isa-l/crc/crc64_iso_norm_by8.asm new file mode 100644 index 00000000..1a4195d6 --- /dev/null +++ b/src/isa-l/crc/crc64_iso_norm_by8.asm @@ -0,0 +1,581 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function API: +; uint64_t crc64_iso_norm_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif +align 16 +global crc64_iso_norm_by8:function +crc64_iso_norm_by8: + + not arg1 ;~init_crc + + sub rsp,VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + + ; check if smaller than 256 + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register. + ; because data will be byte-reflected and will align with initial crc at correct place. + pslldq xmm10, 8 + + movdqa xmm11, [SHUF_MASK] + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + pshufb xmm0, xmm11 + ; XOR the initial_crc value + pxor xmm0, xmm10 + pshufb xmm1, xmm11 + pshufb xmm2, xmm11 + pshufb xmm3, xmm11 + pshufb xmm4, xmm11 + pshufb xmm5, xmm11 + pshufb xmm6, xmm11 + pshufb xmm7, xmm11 + + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 ; buf += 128; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm1, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm3, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm5, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm7, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + + movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2 + + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pshufb xmm0, xmm11 + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm7 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm11 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm7 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm7, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10 + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0x01 ; H*L + pslldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10 + movdqa xmm0, xmm7 + + movdqa xmm1, xmm7 + pand xmm1, [mask3] + pclmulqdq xmm7, xmm10, 0x01 + pxor xmm7, xmm1 + + pclmulqdq xmm7, xmm10, 0x11 + pxor xmm7, xmm0 + pextrq rax, xmm7, 0 + +_cleanup: + not rax +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm11, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movdqa xmm11, [SHUF_MASK] + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + ; shl r9, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm7, xmm0 + jmp _128_done + +_end_1to7: + ; Right shift (8-length) bytes in XMM + add rax, 8 + movdqu xmm0, [rax] + pshufb xmm7,xmm0 + + jmp _barrett +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 + +rk1: +DQ 0x0000000000000145 +rk2: +DQ 0x0000000000001db7 +rk3: +DQ 0x000100000001001a +rk4: +DQ 0x001b0000001b015e +rk5: +DQ 0x0000000000000145 +rk6: +DQ 0x0000000000000000 +rk7: +DQ 0x000000000000001b +rk8: +DQ 0x000000000000001b +rk9: +DQ 0x0150145145145015 +rk10: +DQ 0x1c71db6db6db71c7 +rk11: +DQ 0x0001110110110111 +rk12: +DQ 0x001aab1ab1ab1aab +rk13: +DQ 0x0000014445014445 +rk14: +DQ 0x00001daab71daab7 +rk15: +DQ 0x0000000101000101 +rk16: +DQ 0x0000001b1b001b1b +rk17: +DQ 0x0000000001514515 +rk18: +DQ 0x000000001c6db6c7 +rk19: +DQ 0x0000000000011011 +rk20: +DQ 0x00000000001ab1ab + +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF +mask3: +dq 0x0000000000000000, 0xFFFFFFFFFFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_iso_norm_by8, 01, 00, 0020 diff --git a/src/isa-l/crc/crc64_iso_refl_by8.asm b/src/isa-l/crc/crc64_iso_refl_by8.asm new file mode 100644 index 00000000..d7ed8ae3 --- /dev/null +++ b/src/isa-l/crc/crc64_iso_refl_by8.asm @@ -0,0 +1,544 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Function API: +; uint64_t crc64_iso_refl_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif + + +align 16 +global crc64_iso_refl_by8:function +crc64_iso_refl_by8: + ; uint64_t c = crc ^ 0xffffffff,ffffffffL; + not arg1 + sub rsp, VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + ; check if smaller than 256B + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + ; XOR the initial_crc value + pxor xmm0, xmm10 + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm1, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm3, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm5, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm7, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + ; xmm0 to xmm7 + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + ;xmm1 to xmm7 + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + ; xmm6 to xmm7 + movdqa xmm10, [rk1] + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 128-16 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + add arg3, 16 + je _128_done + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + + + movdqa xmm2, xmm7 + movdqu xmm1, [arg2 - 16 + arg3] + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table] + add rax, arg3 + movdqu xmm0, [rax] + + + pshufb xmm7, xmm0 + pxor xmm0, [mask3] + pshufb xmm2, xmm0 + + pblendvb xmm2, xmm1 ;xmm0 is implicit + ;;;;;;;;;; + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0 + psrldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm1, xmm7 + movdqa xmm10, [rk7] + + pclmulqdq xmm7, xmm10, 0 + movdqa xmm2, xmm7 + pclmulqdq xmm7, xmm10, 0x10 + pslldq xmm2, 8 + pxor xmm7, xmm2 + pxor xmm7, xmm1 + pextrq rax, xmm7, 1 + +_cleanup: + ; return c ^ 0xffffffff, ffffffffL; + not rax + + +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movq xmm0, arg1 ; get the initial crc value + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + +_zero_left: + movdqa xmm7, [rsp] + pxor xmm7, xmm0 ; xor the initial crc value + + lea rax,[pshufb_shf_table] + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax + r9] + pshufb xmm7,xmm0 + jmp _128_done + +_end_1to7: + ; Left shift (8-length) bytes in XMM + movdqu xmm0, [rax + r9 + 8] + pshufb xmm7,xmm0 + + jmp _barrett + +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 +; rk7 = floor(2^128/Q) +; rk8 = Q +rk1: +DQ 0xf500000000000001 +rk2: +DQ 0x6b70000000000001 +rk3: +DQ 0xb001000000010000 +rk4: +DQ 0xf501b0000001b000 +rk5: +DQ 0xf500000000000001 +rk6: +DQ 0x0000000000000000 +rk7: +DQ 0xb000000000000001 +rk8: +DQ 0xb000000000000000 +rk9: +DQ 0xe014514514501501 +rk10: +DQ 0x771db6db6db71c71 +rk11: +DQ 0xa101101101110001 +rk12: +DQ 0x1ab1ab1ab1aab001 +rk13: +DQ 0xf445014445000001 +rk14: +DQ 0x6aab71daab700001 +rk15: +DQ 0xb100010100000001 +rk16: +DQ 0x01b001b1b0000001 +rk17: +DQ 0xe145150000000001 +rk18: +DQ 0x76db6c7000000001 +rk19: +DQ 0xa011000000000001 +rk20: +DQ 0x1b1ab00000000001 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + + +mask: +dq 0xFFFFFFFFFFFFFFFF, 0x0000000000000000 +mask2: +dq 0xFFFFFFFF00000000, 0xFFFFFFFFFFFFFFFF +mask3: +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_iso_refl_by8, 01, 00, 0023 diff --git a/src/isa-l/crc/crc64_jones_norm_by8.asm b/src/isa-l/crc/crc64_jones_norm_by8.asm new file mode 100644 index 00000000..0e5e75a2 --- /dev/null +++ b/src/isa-l/crc/crc64_jones_norm_by8.asm @@ -0,0 +1,581 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +; Function API: +; uint64_t crc64_jones_norm_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif +align 16 +global crc64_jones_norm_by8:function +crc64_jones_norm_by8: + + not arg1 ;~init_crc + + sub rsp,VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + + ; check if smaller than 256 + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + + ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register. + ; because data will be byte-reflected and will align with initial crc at correct place. + pslldq xmm10, 8 + + movdqa xmm11, [SHUF_MASK] + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + pshufb xmm0, xmm11 + ; XOR the initial_crc value + pxor xmm0, xmm10 + pshufb xmm1, xmm11 + pshufb xmm2, xmm11 + pshufb xmm3, xmm11 + pshufb xmm4, xmm11 + pshufb xmm5, xmm11 + pshufb xmm6, xmm11 + pshufb xmm7, xmm11 + + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 ; buf += 128; + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm1, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm3, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm5, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + pshufb xmm9, xmm11 + pshufb xmm12, xmm11 + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x0 + pclmulqdq xmm8, xmm10 , 0x11 + pclmulqdq xmm7, xmm10, 0x0 + pclmulqdq xmm13, xmm10 , 0x11 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + + movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2 + + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pshufb xmm0, xmm11 + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + ; check if any more data to fold. If not, compute the CRC of the final 128 bits + add arg3, 16 + je _128_done + + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + movdqa xmm2, xmm7 + + movdqu xmm1, [arg2 - 16 + arg3] + pshufb xmm1, xmm11 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg3 + movdqu xmm0, [rax] + + ; shift xmm2 to the left by arg3 bytes + pshufb xmm2, xmm0 + + ; shift xmm7 to the right by 16-arg3 bytes + pxor xmm0, [mask1] + pshufb xmm7, xmm0 + pblendvb xmm1, xmm2 ;xmm0 is implicit + + ; fold 16 Bytes + movdqa xmm2, xmm1 + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x11 + pclmulqdq xmm8, xmm10, 0x0 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10 + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0x01 ; H*L + pslldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10 + movdqa xmm0, xmm7 + + movdqa xmm1, xmm7 + pand xmm1, [mask3] + pclmulqdq xmm7, xmm10, 0x01 + pxor xmm7, xmm1 + + pclmulqdq xmm7, xmm10, 0x11 + pxor xmm7, xmm0 + pextrq rax, xmm7, 0 + +_cleanup: + not rax +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + movdqa xmm11, [SHUF_MASK] + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 + + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movdqa xmm11, [SHUF_MASK] + + movq xmm0, arg1 ; get the initial crc value + pslldq xmm0, 8 ; align it to its correct place + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pshufb xmm7, xmm11 ; byte-reflect the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al +_zero_left: + movdqa xmm7, [rsp] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + ; shl r9, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, r9 + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax] + pxor xmm0, [mask1] + + pshufb xmm7, xmm0 + jmp _128_done + +_end_1to7: + ; Right shift (8-length) bytes in XMM + add rax, 8 + movdqu xmm0, [rax] + pshufb xmm7,xmm0 + + jmp _barrett +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pshufb xmm7, xmm11 + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 + +rk1: +DQ 0x4445ed2750017038 +rk2: +DQ 0x698b74157cfbd736 +rk3: +DQ 0x0cfcfb5101c4b775 +rk4: +DQ 0x65403fd47cbec866 +rk5: +DQ 0x4445ed2750017038 +rk6: +DQ 0x0000000000000000 +rk7: +DQ 0xddf3eeb298be6cf8 +rk8: +DQ 0xad93d23594c935a9 +rk9: +DQ 0xd8dc208e2ba527b4 +rk10: +DQ 0xf032cfec76bb2bc5 +rk11: +DQ 0xb536044f357f4238 +rk12: +DQ 0xfdbf104d938ba67a +rk13: +DQ 0xeeddad9297a843e7 +rk14: +DQ 0x3550bce629466473 +rk15: +DQ 0x4e501e58ca43d25e +rk16: +DQ 0x13c961588f27f643 +rk17: +DQ 0x3b60d00dcb1099bc +rk18: +DQ 0x44bf1f468c53b9a3 +rk19: +DQ 0x96f2236e317179ee +rk20: +DQ 0xf00839aa0dd64bac + +mask1: +dq 0x8080808080808080, 0x8080808080808080 +mask2: +dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF +mask3: +dq 0x0000000000000000, 0xFFFFFFFFFFFFFFFF + +SHUF_MASK: +dq 0x08090A0B0C0D0E0F, 0x0001020304050607 + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x0f0e0d0c0b0a0908 +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_jones_norm_by8, 01, 00, 0026 diff --git a/src/isa-l/crc/crc64_jones_refl_by8.asm b/src/isa-l/crc/crc64_jones_refl_by8.asm new file mode 100644 index 00000000..39da6b82 --- /dev/null +++ b/src/isa-l/crc/crc64_jones_refl_by8.asm @@ -0,0 +1,544 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Function API: +; uint64_t crc64_jones_refl_by8( +; uint64_t init_crc, //initial CRC value, 64 bits +; const unsigned char *buf, //buffer pointer to calculate CRC on +; uint64_t len //buffer length in bytes (64-bit data) +; ); +; +%include "reg_sizes.asm" + +%define fetch_dist 1024 + +[bits 64] +default rel + +section .text + + +%ifidn __OUTPUT_FORMAT__, win64 + %xdefine arg1 rcx + %xdefine arg2 rdx + %xdefine arg3 r8 +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx +%endif + +%define TMP 16*0 +%ifidn __OUTPUT_FORMAT__, win64 + %define XMM_SAVE 16*2 + %define VARIABLE_OFFSET 16*10+8 +%else + %define VARIABLE_OFFSET 16*2+8 +%endif + + +align 16 +global crc64_jones_refl_by8:function +crc64_jones_refl_by8: + ; uint64_t c = crc ^ 0xffffffff,ffffffffL; + not arg1 + sub rsp, VARIABLE_OFFSET + +%ifidn __OUTPUT_FORMAT__, win64 + ; push the xmm registers into the stack to maintain + movdqa [rsp + XMM_SAVE + 16*0], xmm6 + movdqa [rsp + XMM_SAVE + 16*1], xmm7 + movdqa [rsp + XMM_SAVE + 16*2], xmm8 + movdqa [rsp + XMM_SAVE + 16*3], xmm9 + movdqa [rsp + XMM_SAVE + 16*4], xmm10 + movdqa [rsp + XMM_SAVE + 16*5], xmm11 + movdqa [rsp + XMM_SAVE + 16*6], xmm12 + movdqa [rsp + XMM_SAVE + 16*7], xmm13 +%endif + + ; check if smaller than 256B + cmp arg3, 256 + + ; for sizes less than 256, we can't fold 128B at a time... + jl _less_than_256 + + + ; load the initial crc value + movq xmm10, arg1 ; initial crc + ; receive the initial 128B data, xor the initial crc value + movdqu xmm0, [arg2+16*0] + movdqu xmm1, [arg2+16*1] + movdqu xmm2, [arg2+16*2] + movdqu xmm3, [arg2+16*3] + movdqu xmm4, [arg2+16*4] + movdqu xmm5, [arg2+16*5] + movdqu xmm6, [arg2+16*6] + movdqu xmm7, [arg2+16*7] + + ; XOR the initial_crc value + pxor xmm0, xmm10 + movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4 + ;imm value of pclmulqdq instruction will determine which constant to use + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + ; we subtract 256 instead of 128 to save one instruction from the loop + sub arg3, 256 + + ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop + ; loop will fold 128B at a time until we have 128+y Bytes of buffer + + + ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel +_fold_128_B_loop: + + ; update the buffer pointer + add arg2, 128 + + prefetchnta [arg2+fetch_dist+0] + movdqu xmm9, [arg2+16*0] + movdqu xmm12, [arg2+16*1] + movdqa xmm8, xmm0 + movdqa xmm13, xmm1 + pclmulqdq xmm0, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm1, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm0, xmm9 + xorps xmm0, xmm8 + pxor xmm1, xmm12 + xorps xmm1, xmm13 + + prefetchnta [arg2+fetch_dist+32] + movdqu xmm9, [arg2+16*2] + movdqu xmm12, [arg2+16*3] + movdqa xmm8, xmm2 + movdqa xmm13, xmm3 + pclmulqdq xmm2, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm3, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm2, xmm9 + xorps xmm2, xmm8 + pxor xmm3, xmm12 + xorps xmm3, xmm13 + + prefetchnta [arg2+fetch_dist+64] + movdqu xmm9, [arg2+16*4] + movdqu xmm12, [arg2+16*5] + movdqa xmm8, xmm4 + movdqa xmm13, xmm5 + pclmulqdq xmm4, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm5, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm4, xmm9 + xorps xmm4, xmm8 + pxor xmm5, xmm12 + xorps xmm5, xmm13 + + prefetchnta [arg2+fetch_dist+96] + movdqu xmm9, [arg2+16*6] + movdqu xmm12, [arg2+16*7] + movdqa xmm8, xmm6 + movdqa xmm13, xmm7 + pclmulqdq xmm6, xmm10, 0x10 + pclmulqdq xmm8, xmm10 , 0x1 + pclmulqdq xmm7, xmm10, 0x10 + pclmulqdq xmm13, xmm10 , 0x1 + pxor xmm6, xmm9 + xorps xmm6, xmm8 + pxor xmm7, xmm12 + xorps xmm7, xmm13 + + sub arg3, 128 + + ; check if there is another 128B in the buffer to be able to fold + jge _fold_128_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + add arg2, 128 + ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128 + ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7 + + + ; fold the 8 xmm registers to 1 xmm register with different constants + ; xmm0 to xmm7 + movdqa xmm10, [rk9] + movdqa xmm8, xmm0 + pclmulqdq xmm0, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm0 + ;xmm1 to xmm7 + movdqa xmm10, [rk11] + movdqa xmm8, xmm1 + pclmulqdq xmm1, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm1 + + movdqa xmm10, [rk13] + movdqa xmm8, xmm2 + pclmulqdq xmm2, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + + movdqa xmm10, [rk15] + movdqa xmm8, xmm3 + pclmulqdq xmm3, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm3 + + movdqa xmm10, [rk17] + movdqa xmm8, xmm4 + pclmulqdq xmm4, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm4 + + movdqa xmm10, [rk19] + movdqa xmm8, xmm5 + pclmulqdq xmm5, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + xorps xmm7, xmm5 + ; xmm6 to xmm7 + movdqa xmm10, [rk1] + movdqa xmm8, xmm6 + pclmulqdq xmm6, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm6 + + + ; instead of 128, we add 128-16 to the loop counter to save 1 instruction from the loop + ; instead of a cmp instruction, we use the negative flag with the jl instruction + add arg3, 128-16 + jl _final_reduction_for_128 + + ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory + ; we can fold 16 bytes at a time if y>=16 + ; continue folding 16B at a time + +_16B_reduction_loop: + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + movdqu xmm0, [arg2] + pxor xmm7, xmm0 + add arg2, 16 + sub arg3, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg3, 16-16 + ; check if there is any more 16B in the buffer to be able to fold + jge _16B_reduction_loop + + ;now we have 16+z bytes left to reduce, where 0<= z < 16. + ;first, we reduce the data in the xmm7 register + + +_final_reduction_for_128: + add arg3, 16 + je _128_done + ; here we are getting data that is less than 16 bytes. + ; since we know that there was data before the pointer, we can offset the input pointer before the actual point, to receive exactly 16 bytes. + ; after that the registers need to be adjusted. +_get_last_two_xmms: + + + movdqa xmm2, xmm7 + movdqu xmm1, [arg2 - 16 + arg3] + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table] + add rax, arg3 + movdqu xmm0, [rax] + + + pshufb xmm7, xmm0 + pxor xmm0, [mask3] + pshufb xmm2, xmm0 + + pblendvb xmm2, xmm1 ;xmm0 is implicit + ;;;;;;;;;; + movdqa xmm8, xmm7 + pclmulqdq xmm7, xmm10, 0x1 + + pclmulqdq xmm8, xmm10, 0x10 + pxor xmm7, xmm8 + pxor xmm7, xmm2 + +_128_done: + ; compute crc of a 128-bit value + movdqa xmm10, [rk5] + movdqa xmm0, xmm7 + + ;64b fold + pclmulqdq xmm7, xmm10, 0 + psrldq xmm0, 8 + pxor xmm7, xmm0 + + ;barrett reduction +_barrett: + movdqa xmm1, xmm7 + movdqa xmm10, [rk7] + + pclmulqdq xmm7, xmm10, 0 + movdqa xmm2, xmm7 + pclmulqdq xmm7, xmm10, 0x10 + pslldq xmm2, 8 + pxor xmm7, xmm2 + pxor xmm7, xmm1 + pextrq rax, xmm7, 1 + +_cleanup: + ; return c ^ 0xffffffff, ffffffffL; + not rax + + +%ifidn __OUTPUT_FORMAT__, win64 + movdqa xmm6, [rsp + XMM_SAVE + 16*0] + movdqa xmm7, [rsp + XMM_SAVE + 16*1] + movdqa xmm8, [rsp + XMM_SAVE + 16*2] + movdqa xmm9, [rsp + XMM_SAVE + 16*3] + movdqa xmm10, [rsp + XMM_SAVE + 16*4] + movdqa xmm11, [rsp + XMM_SAVE + 16*5] + movdqa xmm12, [rsp + XMM_SAVE + 16*6] + movdqa xmm13, [rsp + XMM_SAVE + 16*7] +%endif + add rsp, VARIABLE_OFFSET + ret + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +align 16 +_less_than_256: + + ; check if there is enough buffer to be able to fold 16B at a time + cmp arg3, 32 + jl _less_than_32 + + ; if there is, load the constants + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + + movq xmm0, arg1 ; get the initial crc value + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 + + ; update the buffer pointer + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg3, 32 + + jmp _16B_reduction_loop + +align 16 +_less_than_32: + ; mov initial crc to the return value. this is necessary for zero-length buffers. + mov rax, arg1 + test arg3, arg3 + je _cleanup + + movq xmm0, arg1 ; get the initial crc value + + cmp arg3, 16 + je _exact_16_left + jl _less_than_16_left + + movdqu xmm7, [arg2] ; load the plaintext + pxor xmm7, xmm0 ; xor the initial crc value + add arg2, 16 + sub arg3, 16 + movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10 + jmp _get_last_two_xmms + + +align 16 +_less_than_16_left: + ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first. + + pxor xmm1, xmm1 + mov r11, rsp + movdqa [r11], xmm1 + + ; backup the counter value + mov r9, arg3 + cmp arg3, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg2] + mov [r11], rax + add r11, 8 + sub arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg3, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg2] + mov [r11], eax + add r11, 4 + sub arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg3, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg2] + mov [r11], ax + add r11, 2 + sub arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg3, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg2] + mov [r11], al + +_zero_left: + movdqa xmm7, [rsp] + pxor xmm7, xmm0 ; xor the initial crc value + + lea rax,[pshufb_shf_table] + + cmp r9, 8 + jl _end_1to7 + +_end_8to15: + movdqu xmm0, [rax + r9] + pshufb xmm7,xmm0 + jmp _128_done + +_end_1to7: + ; Left shift (8-length) bytes in XMM + movdqu xmm0, [rax + r9 + 8] + pshufb xmm7,xmm0 + + jmp _barrett + +align 16 +_exact_16_left: + movdqu xmm7, [arg2] + pxor xmm7, xmm0 ; xor the initial crc value + + jmp _128_done + +section .data + +; precomputed constants +align 16 +; rk7 = floor(2^128/Q) +; rk8 = Q +rk1: +DQ 0x381d0015c96f4444 +rk2: +DQ 0xd9d7be7d505da32c +rk3: +DQ 0x768361524d29ed0b +rk4: +DQ 0xcc26fa7c57f8054c +rk5: +DQ 0x381d0015c96f4444 +rk6: +DQ 0x0000000000000000 +rk7: +DQ 0x3e6cfa329aef9f77 +rk8: +DQ 0x2b5926535897936a +rk9: +DQ 0x5bc94ba8e2087636 +rk10: +DQ 0x6cf09c8f37710b75 +rk11: +DQ 0x3885fd59e440d95a +rk12: +DQ 0xbccba3936411fb7e +rk13: +DQ 0xe4dd0d81cbfce585 +rk14: +DQ 0xb715e37b96ed8633 +rk15: +DQ 0xf49784a634f014e4 +rk16: +DQ 0xaf86efb16d9ab4fb +rk17: +DQ 0x7b3211a760160db8 +rk18: +DQ 0xa062b2319d66692f +rk19: +DQ 0xef3d1d18ed889ed2 +rk20: +DQ 0x6ba4d760ab38201e + +pshufb_shf_table: +; use these values for shift constants for the pshufb instruction +; different alignments result in values as shown: +; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1 +; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2 +; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3 +; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4 +; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5 +; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6 +; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7 +; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8 +; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9 +; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10 +; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11 +; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12 +; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13 +; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14 +; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15 +dq 0x8786858483828100, 0x8f8e8d8c8b8a8988 +dq 0x0706050403020100, 0x000e0d0c0b0a0908 + + +mask: +dq 0xFFFFFFFFFFFFFFFF, 0x0000000000000000 +mask2: +dq 0xFFFFFFFF00000000, 0xFFFFFFFFFFFFFFFF +mask3: +dq 0x8080808080808080, 0x8080808080808080 + +;;; func core, ver, snum +slversion crc64_jones_refl_by8, 01, 00, 0029 diff --git a/src/isa-l/crc/crc64_multibinary.asm b/src/isa-l/crc/crc64_multibinary.asm new file mode 100644 index 00000000..a20c8a79 --- /dev/null +++ b/src/isa-l/crc/crc64_multibinary.asm @@ -0,0 +1,89 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +;;; +;;; uint64_t crc64_func(uint64_t init_crc, const unsigned char *buf, uint64_t len); +;;; + +default rel +[bits 64] + +%ifidn __OUTPUT_FORMAT__, elf64 +%define WRT_OPT wrt ..plt +%else +%define WRT_OPT +%endif + +%include "reg_sizes.asm" + +extern crc64_ecma_refl_by8 +extern crc64_ecma_refl_base + +extern crc64_ecma_norm_by8 +extern crc64_ecma_norm_base + +extern crc64_iso_refl_by8 +extern crc64_iso_refl_base + +extern crc64_iso_norm_by8 +extern crc64_iso_norm_base + +extern crc64_jones_refl_by8 +extern crc64_jones_refl_base + +extern crc64_jones_norm_by8 +extern crc64_jones_norm_base + +section .text + +%include "multibinary.asm" + +mbin_interface crc64_ecma_refl +mbin_dispatch_init_clmul crc64_ecma_refl, crc64_ecma_refl_base, crc64_ecma_refl_by8 +mbin_interface crc64_ecma_norm +mbin_dispatch_init_clmul crc64_ecma_norm, crc64_ecma_norm_base, crc64_ecma_norm_by8 + +mbin_interface crc64_iso_refl +mbin_dispatch_init_clmul crc64_iso_refl, crc64_iso_refl_base, crc64_iso_refl_by8 +mbin_interface crc64_iso_norm +mbin_dispatch_init_clmul crc64_iso_norm, crc64_iso_norm_base, crc64_iso_norm_by8 + +mbin_interface crc64_jones_refl +mbin_dispatch_init_clmul crc64_jones_refl, crc64_jones_refl_base, crc64_jones_refl_by8 +mbin_interface crc64_jones_norm +mbin_dispatch_init_clmul crc64_jones_norm, crc64_jones_norm_base, crc64_jones_norm_by8 + +;;; func core, ver, snum +slversion crc64_ecma_refl, 00, 00, 001b +slversion crc64_ecma_norm, 00, 00, 0018 +slversion crc64_iso_refl, 00, 00, 0021 +slversion crc64_iso_norm, 00, 00, 001e +slversion crc64_jones_refl, 00, 00, 0027 +slversion crc64_jones_norm, 00, 00, 0024 diff --git a/src/isa-l/crc/crc_base.c b/src/isa-l/crc/crc_base.c new file mode 100644 index 00000000..ee14f059 --- /dev/null +++ b/src/isa-l/crc/crc_base.c @@ -0,0 +1,170 @@ +/********************************************************************** + Copyright(c) 2011-2015 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. +**********************************************************************/ + +#include <stdlib.h> +#include "crc.h" + +#define MAX_ITER 8 + +uint32_t crc32_table_iscsi_base[256] = { + 0x00000000, 0xF26B8303, 0xE13B70F7, 0x1350F3F4, + 0xC79A971F, 0x35F1141C, 0x26A1E7E8, 0xD4CA64EB, + 0x8AD958CF, 0x78B2DBCC, 0x6BE22838, 0x9989AB3B, + 0x4D43CFD0, 0xBF284CD3, 0xAC78BF27, 0x5E133C24, + 0x105EC76F, 0xE235446C, 0xF165B798, 0x030E349B, + 0xD7C45070, 0x25AFD373, 0x36FF2087, 0xC494A384, + 0x9A879FA0, 0x68EC1CA3, 0x7BBCEF57, 0x89D76C54, + 0x5D1D08BF, 0xAF768BBC, 0xBC267848, 0x4E4DFB4B, + 0x20BD8EDE, 0xD2D60DDD, 0xC186FE29, 0x33ED7D2A, + 0xE72719C1, 0x154C9AC2, 0x061C6936, 0xF477EA35, + 0xAA64D611, 0x580F5512, 0x4B5FA6E6, 0xB93425E5, + 0x6DFE410E, 0x9F95C20D, 0x8CC531F9, 0x7EAEB2FA, + 0x30E349B1, 0xC288CAB2, 0xD1D83946, 0x23B3BA45, + 0xF779DEAE, 0x05125DAD, 0x1642AE59, 0xE4292D5A, + 0xBA3A117E, 0x4851927D, 0x5B016189, 0xA96AE28A, + 0x7DA08661, 0x8FCB0562, 0x9C9BF696, 0x6EF07595, + 0x417B1DBC, 0xB3109EBF, 0xA0406D4B, 0x522BEE48, + 0x86E18AA3, 0x748A09A0, 0x67DAFA54, 0x95B17957, + 0xCBA24573, 0x39C9C670, 0x2A993584, 0xD8F2B687, + 0x0C38D26C, 0xFE53516F, 0xED03A29B, 0x1F682198, + 0x5125DAD3, 0xA34E59D0, 0xB01EAA24, 0x42752927, + 0x96BF4DCC, 0x64D4CECF, 0x77843D3B, 0x85EFBE38, + 0xDBFC821C, 0x2997011F, 0x3AC7F2EB, 0xC8AC71E8, + 0x1C661503, 0xEE0D9600, 0xFD5D65F4, 0x0F36E6F7, + 0x61C69362, 0x93AD1061, 0x80FDE395, 0x72966096, + 0xA65C047D, 0x5437877E, 0x4767748A, 0xB50CF789, + 0xEB1FCBAD, 0x197448AE, 0x0A24BB5A, 0xF84F3859, + 0x2C855CB2, 0xDEEEDFB1, 0xCDBE2C45, 0x3FD5AF46, + 0x7198540D, 0x83F3D70E, 0x90A324FA, 0x62C8A7F9, + 0xB602C312, 0x44694011, 0x5739B3E5, 0xA55230E6, + 0xFB410CC2, 0x092A8FC1, 0x1A7A7C35, 0xE811FF36, + 0x3CDB9BDD, 0xCEB018DE, 0xDDE0EB2A, 0x2F8B6829, + 0x82F63B78, 0x709DB87B, 0x63CD4B8F, 0x91A6C88C, + 0x456CAC67, 0xB7072F64, 0xA457DC90, 0x563C5F93, + 0x082F63B7, 0xFA44E0B4, 0xE9141340, 0x1B7F9043, + 0xCFB5F4A8, 0x3DDE77AB, 0x2E8E845F, 0xDCE5075C, + 0x92A8FC17, 0x60C37F14, 0x73938CE0, 0x81F80FE3, + 0x55326B08, 0xA759E80B, 0xB4091BFF, 0x466298FC, + 0x1871A4D8, 0xEA1A27DB, 0xF94AD42F, 0x0B21572C, + 0xDFEB33C7, 0x2D80B0C4, 0x3ED04330, 0xCCBBC033, + 0xA24BB5A6, 0x502036A5, 0x4370C551, 0xB11B4652, + 0x65D122B9, 0x97BAA1BA, 0x84EA524E, 0x7681D14D, + 0x2892ED69, 0xDAF96E6A, 0xC9A99D9E, 0x3BC21E9D, + 0xEF087A76, 0x1D63F975, 0x0E330A81, 0xFC588982, + 0xB21572C9, 0x407EF1CA, 0x532E023E, 0xA145813D, + 0x758FE5D6, 0x87E466D5, 0x94B49521, 0x66DF1622, + 0x38CC2A06, 0xCAA7A905, 0xD9F75AF1, 0x2B9CD9F2, + 0xFF56BD19, 0x0D3D3E1A, 0x1E6DCDEE, 0xEC064EED, + 0xC38D26C4, 0x31E6A5C7, 0x22B65633, 0xD0DDD530, + 0x0417B1DB, 0xF67C32D8, 0xE52CC12C, 0x1747422F, + 0x49547E0B, 0xBB3FFD08, 0xA86F0EFC, 0x5A048DFF, + 0x8ECEE914, 0x7CA56A17, 0x6FF599E3, 0x9D9E1AE0, + 0xD3D3E1AB, 0x21B862A8, 0x32E8915C, 0xC083125F, + 0x144976B4, 0xE622F5B7, 0xF5720643, 0x07198540, + 0x590AB964, 0xAB613A67, 0xB831C993, 0x4A5A4A90, + 0x9E902E7B, 0x6CFBAD78, 0x7FAB5E8C, 0x8DC0DD8F, + 0xE330A81A, 0x115B2B19, 0x020BD8ED, 0xF0605BEE, + 0x24AA3F05, 0xD6C1BC06, 0xC5914FF2, 0x37FACCF1, + 0x69E9F0D5, 0x9B8273D6, 0x88D28022, 0x7AB90321, + 0xAE7367CA, 0x5C18E4C9, 0x4F48173D, 0xBD23943E, + 0xF36E6F75, 0x0105EC76, 0x12551F82, 0xE03E9C81, + 0x34F4F86A, 0xC69F7B69, 0xD5CF889D, 0x27A40B9E, + 0x79B737BA, 0x8BDCB4B9, 0x988C474D, 0x6AE7C44E, + 0xBE2DA0A5, 0x4C4623A6, 0x5F16D052, 0xAD7D5351, +}; + +// iSCSI CRC baseline function +unsigned int crc32_iscsi_base(unsigned char *buffer, int len, unsigned int crc_init) +{ + unsigned int crc; + unsigned char *p_buf; + + p_buf = (unsigned char *)buffer; + unsigned char *p_end = buffer + len; + + crc = crc_init; + + while (p_buf < (unsigned char *)p_end) { + crc = (crc >> 8) ^ crc32_table_iscsi_base[(crc & 0x000000FF) ^ *p_buf++]; + } + return crc; +} + +// crc16_t10dif baseline function +// Slow crc16 from the definition. Can be sped up with a lookup table. +uint16_t crc16_t10dif_base(uint16_t seed, uint8_t * buf, uint64_t len) +{ + size_t rem = seed; + unsigned int i, j; + + uint16_t poly = 0x8bb7; // t10dif standard + + for (i = 0; i < len; i++) { + rem = rem ^ (buf[i] << 8); + for (j = 0; j < MAX_ITER; j++) { + rem = rem << 1; + rem = (rem & 0x10000) ? rem ^ poly : rem; + } + } + return rem; +} + +// crc32_ieee baseline function +// Slow crc32 from the definition. Can be sped up with a lookup table. +uint32_t crc32_ieee_base(uint32_t seed, uint8_t * buf, uint64_t len) +{ + uint64_t rem = ~seed; + unsigned int i, j; + + uint32_t poly = 0x04C11DB7; // IEEE standard + + for (i = 0; i < len; i++) { + rem = rem ^ (buf[i] << 24); + for (j = 0; j < MAX_ITER; j++) { + rem = rem << 1; + rem = (rem & 0x100000000ULL) ? rem ^ poly : rem; + } + } + return ~rem; +} + +struct slver { + unsigned short snum; + unsigned char ver; + unsigned char core; +}; + +struct slver crc32_iscsi_base_slver_0001011d; +struct slver crc32_iscsi_base_slver = { 0x011d, 0x02, 0x00 }; + +struct slver crc16_t10dif_base_slver_0001011e; +struct slver crc16_t10dif_base_slver = { 0x011e, 0x02, 0x00 }; + +struct slver crc32_ieee_base_slver_0001011f; +struct slver crc32_ieee_base_slver = { 0x011f, 0x02, 0x00 }; diff --git a/src/isa-l/crc/crc_base_aliases.c b/src/isa-l/crc/crc_base_aliases.c new file mode 100644 index 00000000..63dfb306 --- /dev/null +++ b/src/isa-l/crc/crc_base_aliases.c @@ -0,0 +1,77 @@ +/********************************************************************** + Copyright(c) 2011-2017 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. +**********************************************************************/ + +#include "crc.h" +#include "crc64.h" +#include <stdint.h> + +unsigned int crc32_iscsi(unsigned char *buffer, int len, unsigned int crc_init) +{ + return crc32_iscsi_base(buffer, len, crc_init); +} + +uint16_t crc16_t10dif(uint16_t seed, const unsigned char *buf, uint64_t len) +{ + return crc16_t10dif_base(seed, (uint8_t *) buf, len); +} + +uint32_t crc32_ieee(uint32_t seed, const unsigned char *buf, uint64_t len) +{ + return crc32_ieee_base(seed, (uint8_t *) buf, len); +} + +uint64_t crc64_ecma_refl(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_ecma_refl_base(seed, buf, len); +} + +uint64_t crc64_ecma_norm(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_ecma_norm_base(seed, buf, len); +} + +uint64_t crc64_iso_refl(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_iso_refl_base(seed, buf, len); +} + +uint64_t crc64_iso_norm(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_iso_norm_base(seed, buf, len); +} + +uint64_t crc64_jones_refl(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_jones_refl_base(seed, buf, len); +} + +uint64_t crc64_jones_norm(uint64_t seed, const uint8_t * buf, uint64_t len) +{ + return crc64_jones_norm_base(seed, buf, len); +} diff --git a/src/isa-l/crc/crc_multibinary.asm b/src/isa-l/crc/crc_multibinary.asm new file mode 100644 index 00000000..ae0c0498 --- /dev/null +++ b/src/isa-l/crc/crc_multibinary.asm @@ -0,0 +1,180 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2015 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. +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + +default rel +[bits 64] + +%ifidn __OUTPUT_FORMAT__, elf64 +%define WRT_OPT wrt ..plt +%else +%define WRT_OPT +%endif + +%include "reg_sizes.asm" + +extern crc32_iscsi_00 +extern crc32_iscsi_01 +extern crc32_iscsi_base + +extern crc32_ieee_01 +extern crc32_ieee_by4 ;; Optimized for SLM +extern crc32_ieee_base + +extern crc16_t10dif_01 +extern crc16_t10dif_by4 ;; Optimized for SLM +extern crc16_t10dif_base + +section .data +;;; *_mbinit are initial values for *_dispatched; is updated on first call. +;;; Therefore, *_dispatch_init is only executed on first call. + +crc32_iscsi_dispatched: + dq crc32_iscsi_mbinit + +crc32_ieee_dispatched: + dq crc32_ieee_mbinit + +crc16_t10dif_dispatched: + dq crc16_t10dif_mbinit + +section .text +;;;; +; crc32_iscsi multibinary function +;;;; +global crc32_iscsi:function +crc32_iscsi_mbinit: + call crc32_iscsi_dispatch_init +crc32_iscsi: + jmp qword [crc32_iscsi_dispatched] + +crc32_iscsi_dispatch_init: + push rax + push rbx + push rcx + push rdx + push rsi + lea rsi, [crc32_iscsi_base WRT_OPT] ; Default + + mov eax, 1 + cpuid + lea rbx, [crc32_iscsi_00 WRT_OPT] + lea rax, [crc32_iscsi_01 WRT_OPT] + + test ecx, FLAG_CPUID1_ECX_SSE4_2 + cmovne rsi, rbx + test ecx, FLAG_CPUID1_ECX_CLMUL + cmovne rsi, rax + mov [crc32_iscsi_dispatched], rsi + pop rsi + pop rdx + pop rcx + pop rbx + pop rax + ret + +;;;; +; crc32_ieee multibinary function +;;;; +global crc32_ieee:function +crc32_ieee_mbinit: + call crc32_ieee_dispatch_init +crc32_ieee: + jmp qword [crc32_ieee_dispatched] + +crc32_ieee_dispatch_init: + push rax + push rbx + push rcx + push rdx + push rsi + lea rsi, [crc32_ieee_base WRT_OPT] ; Default + + mov eax, 1 + cpuid + lea rbx, [crc32_ieee_01 WRT_OPT] + lea rdx, [crc32_ieee_by4 WRT_OPT] + + test ecx, FLAG_CPUID1_ECX_SSE3 + jz use_ieee_base + test ecx, FLAG_CPUID1_ECX_CLMUL + cmovne rsi, rbx + and eax, FLAG_CPUID1_EAX_STEP_MASK + cmp eax, FLAG_CPUID1_EAX_AVOTON + cmove rsi, rdx +use_ieee_base: + mov [crc32_ieee_dispatched], rsi + pop rsi + pop rdx + pop rcx + pop rbx + pop rax + ret + +;;;; +; crc16_t10dif multibinary function +;;;; +global crc16_t10dif:function +crc16_t10dif_mbinit: + call crc16_t10dif_dispatch_init +crc16_t10dif: + jmp qword [crc16_t10dif_dispatched] + +crc16_t10dif_dispatch_init: + push rax + push rbx + push rcx + push rdx + push rsi + lea rsi, [crc16_t10dif_base WRT_OPT] ; Default + + mov eax, 1 + cpuid + lea rbx, [crc16_t10dif_01 WRT_OPT] + lea rdx, [crc16_t10dif_by4 WRT_OPT] + + test ecx, FLAG_CPUID1_ECX_SSE3 + jz use_t10dif_base + test ecx, FLAG_CPUID1_ECX_CLMUL + cmovne rsi, rbx + and eax, FLAG_CPUID1_EAX_STEP_MASK + cmp eax, FLAG_CPUID1_EAX_AVOTON + cmove rsi, rdx +use_t10dif_base: + mov [crc16_t10dif_dispatched], rsi + pop rsi + pop rdx + pop rcx + pop rbx + pop rax + ret + +;;; func core, ver, snum +slversion crc16_t10dif, 00, 03, 011a +slversion crc32_ieee, 00, 03, 011b +slversion crc32_iscsi, 00, 03, 011c diff --git a/src/isa-l/crc/crc_simple_test.c b/src/isa-l/crc/crc_simple_test.c new file mode 100644 index 00000000..cac18f52 --- /dev/null +++ b/src/isa-l/crc/crc_simple_test.c @@ -0,0 +1,63 @@ +/********************************************************************** + Copyright(c) 2011-2013 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. +**********************************************************************/ +#include <stdio.h> +#include <stdint.h> +#include "crc.h" + +const uint16_t init_crc_16 = 0x1234; +const uint16_t t10_dif_expected = 0x60b3; +const uint32_t init_crc_32 = 0x12345678; +const uint32_t ieee_expected = 0x2ceadbe3; + +int main(void) +{ + unsigned char p_buf[48]; + uint16_t t10_dif_computed; + uint32_t ieee_computed; + int i; + + for (i = 0; i < 48; i++) + p_buf[i] = i; + + t10_dif_computed = crc16_t10dif(init_crc_16, p_buf, 48); + + if (t10_dif_computed != t10_dif_expected) + printf("WRONG CRC-16(T10 DIF) value\n"); + else + printf("CORRECT CRC-16(T10 DIF) value\n"); + + ieee_computed = crc32_ieee(init_crc_32, p_buf, 48); + + if (ieee_computed != ieee_expected) + printf("WRONG CRC-32(IEEE) value\n"); + else + printf("CORRECT CRC-32(IEEE) value\n"); + + return 0; +} |