diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 18:45:59 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 18:45:59 +0000 |
commit | 19fcec84d8d7d21e796c7624e521b60d28ee21ed (patch) | |
tree | 42d26aa27d1e3f7c0b8bd3fd14e7d7082f5008dc /src/isa-l/crc/crc16_t10dif_copy_by4_02.asm | |
parent | Initial commit. (diff) | |
download | ceph-upstream.tar.xz ceph-upstream.zip |
Adding upstream version 16.2.11+ds.upstream/16.2.11+dsupstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'src/isa-l/crc/crc16_t10dif_copy_by4_02.asm')
-rw-r--r-- | src/isa-l/crc/crc16_t10dif_copy_by4_02.asm | 595 |
1 files changed, 595 insertions, 0 deletions
diff --git a/src/isa-l/crc/crc16_t10dif_copy_by4_02.asm b/src/isa-l/crc/crc16_t10dif_copy_by4_02.asm new file mode 100644 index 000000000..e12d81f4d --- /dev/null +++ b/src/isa-l/crc/crc16_t10dif_copy_by4_02.asm @@ -0,0 +1,595 @@ +;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; +; Copyright(c) 2011-2020 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_copy_by4_02( +; UINT16 init_crc, //initial CRC value, 16 bits +; unsigned char *dst, //buffer pointer destination for copy +; const unsigned char *src, //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 arg4 r9 + %xdefine tmp1 r10 + %xdefine arg1_low32 ecx +%else + %xdefine arg1 rdi + %xdefine arg2 rsi + %xdefine arg3 rdx + %xdefine arg4 rcx + %xdefine tmp1 r10 + %xdefine arg1_low32 edi +%endif + +align 16 +global crc16_t10dif_copy_by4_02:ISAL_SYM_TYPE_FUNCTION +crc16_t10dif_copy_by4_02: + + ; 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 arg4, 128 + + ; for sizes less than 128, we can't fold 64B at a time... + jl _less_than_128 + + + ; load the initial crc value + vmovd 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. + vpslldq xmm6, 12 + + vmovdqa xmm7, [SHUF_MASK] + ; receive the initial 64B data, xor the initial crc value + vmovdqu xmm0, [arg3] + vmovdqu xmm1, [arg3+16] + vmovdqu xmm2, [arg3+32] + vmovdqu xmm3, [arg3+48] + + ; copy initial data + vmovdqu [arg2], xmm0 + vmovdqu [arg2+16], xmm1 + vmovdqu [arg2+32], xmm2 + vmovdqu [arg2+48], xmm3 + + vpshufb xmm0, xmm7 + ; XOR the initial_crc value + vpxor xmm0, xmm6 + vpshufb xmm1, xmm7 + vpshufb xmm2, xmm7 + vpshufb xmm3, xmm7 + + vmovdqa 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 arg4, 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 arg3, 64 ; buf += 64; + add arg2, 64 + + prefetchnta [arg3+fetch_dist+0] + vmovdqu xmm4, xmm0 + vmovdqu xmm5, xmm1 + + vpclmulqdq xmm0, xmm6 , 0x11 + vpclmulqdq xmm1, xmm6 , 0x11 + + vpclmulqdq xmm4, xmm6, 0x0 + vpclmulqdq xmm5, xmm6, 0x0 + + vpxor xmm0, xmm4 + vpxor xmm1, xmm5 + + prefetchnta [arg3+fetch_dist+32] + vmovdqu xmm4, xmm2 + vmovdqu xmm5, xmm3 + + vpclmulqdq xmm2, xmm6, 0x11 + vpclmulqdq xmm3, xmm6, 0x11 + + vpclmulqdq xmm4, xmm6, 0x0 + vpclmulqdq xmm5, xmm6, 0x0 + + vpxor xmm2, xmm4 + vpxor xmm3, xmm5 + + vmovdqu xmm4, [arg3] + vmovdqu xmm5, [arg3+16] + vmovdqu [arg2], xmm4 + vmovdqu [arg2+16], xmm5 + vpshufb xmm4, xmm7 + vpshufb xmm5, xmm7 + vpxor xmm0, xmm4 + vpxor xmm1, xmm5 + + vmovdqu xmm4, [arg3+32] + vmovdqu xmm5, [arg3+48] + vmovdqu [arg2+32], xmm4 + vmovdqu [arg2+48], xmm5 + vpshufb xmm4, xmm7 + vpshufb xmm5, xmm7 + + vpxor xmm2, xmm4 + vpxor xmm3, xmm5 + + sub arg4, 64 + + ; check if there is another 64B in the buffer to be able to fold + jge _fold_64_B_loop + ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; + + + add arg3, 64 + 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 + + vmovdqa xmm6, [rk1] ;xmm6 has rk1 and rk2 + ;imm value of pclmulqdq instruction will + ;determine which constant to use + + vmovdqa xmm4, xmm0 + vpclmulqdq xmm0, xmm6, 0x11 + vpclmulqdq xmm4, xmm6, 0x0 + vpxor xmm1, xmm4 + vpxor xmm1, xmm0 + + vmovdqa xmm4, xmm1 + vpclmulqdq xmm1, xmm6, 0x11 + vpclmulqdq xmm4, xmm6, 0x0 + vpxor xmm2, xmm4 + vpxor xmm2, xmm1 + + vmovdqa xmm4, xmm2 + vpclmulqdq xmm2, xmm6, 0x11 + vpclmulqdq xmm4, xmm6, 0x0 + vpxor xmm3, xmm4 + vpxor 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 arg4, 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: + vmovdqa xmm4, xmm3 + vpclmulqdq xmm3, xmm6, 0x11 + vpclmulqdq xmm4, xmm6, 0x0 + vpxor xmm3, xmm4 + vmovdqu xmm0, [arg3] + vmovdqu [arg2], xmm0 + vpshufb xmm0, xmm7 + vpxor xmm3, xmm0 + add arg3, 16 + add arg2, 16 + sub arg4, 16 + ; instead of a cmp instruction, we utilize the flags with the jge instruction + ; equivalent of: cmp arg4, 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 arg4, 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: + vmovdqa xmm2, xmm3 + + vmovdqu xmm1, [arg3 - 16 + arg4] + vmovdqu [arg2 - 16 + arg4], xmm1 + vpshufb xmm1, xmm7 + + ; get rid of the extra data that was loaded before + ; load the shift constant + lea rax, [pshufb_shf_table + 16] + sub rax, arg4 + vmovdqu xmm0, [rax] + + ; shift xmm2 to the left by arg4 bytes + vpshufb xmm2, xmm0 + + ; shift xmm3 to the right by 16-arg4 bytes + vpxor xmm0, [mask1] + vpshufb xmm3, xmm0 + vpblendvb xmm1, xmm1, xmm2, xmm0 + + ; fold 16 Bytes + vmovdqa xmm2, xmm1 + vmovdqa xmm4, xmm3 + vpclmulqdq xmm3, xmm6, 0x11 + vpclmulqdq xmm4, xmm6, 0x0 + vpxor xmm3, xmm4 + vpxor xmm3, xmm2 + +_128_done: + ; compute crc of a 128-bit value + vmovdqa xmm6, [rk5] ; rk5 and rk6 in xmm6 + vmovdqa xmm0, xmm3 + + ;64b fold + vpclmulqdq xmm3, xmm6, 0x1 + vpslldq xmm0, 8 + vpxor xmm3, xmm0 + + ;32b fold + vmovdqa xmm0, xmm3 + + vpand xmm0, [mask2] + + vpsrldq xmm3, 12 + vpclmulqdq xmm3, xmm6, 0x10 + vpxor xmm3, xmm0 + + ;barrett reduction +_barrett: + vmovdqa xmm6, [rk7] ; rk7 and rk8 in xmm6 + vmovdqa xmm0, xmm3 + vpclmulqdq xmm3, xmm6, 0x01 + vpslldq xmm3, 4 + vpclmulqdq xmm3, xmm6, 0x11 + + vpslldq xmm3, 4 + vpxor xmm3, xmm0 + vpextrd eax, xmm3,1 + +_cleanup: + ; scale the result back to 16 bits + shr eax, 16 + vmovdqa xmm6, [rsp+16*2] + vmovdqa 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 arg4, 32 + jl _less_than_32 + vmovdqa xmm7, [SHUF_MASK] + + ; if there is, load the constants + vmovdqa xmm6, [rk1] ; rk1 and rk2 in xmm6 + + vmovd xmm0, arg1_low32 ; get the initial crc value + vpslldq xmm0, 12 ; align it to its correct place + vmovdqu xmm3, [arg3] ; load the plaintext + vmovdqu [arg2], xmm3 ; store copy + vpshufb xmm3, xmm7 ; byte-reflect the plaintext + vpxor xmm3, xmm0 + + + ; update the buffer pointer + add arg3, 16 + add arg2, 16 + + ; update the counter. subtract 32 instead of 16 to save one instruction from the loop + sub arg4, 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 arg4, arg4 + je _cleanup + + vmovdqa xmm7, [SHUF_MASK] + + vmovd xmm0, arg1_low32 ; get the initial crc value + vpslldq xmm0, 12 ; align it to its correct place + + cmp arg4, 16 + je _exact_16_left + jl _less_than_16_left + + vmovdqu xmm3, [arg3] ; load the plaintext + vmovdqu [arg2], xmm3 ; store the copy + vpshufb xmm3, xmm7 ; byte-reflect the plaintext + vpxor xmm3, xmm0 ; xor the initial crc value + add arg3, 16 + add arg2, 16 + sub arg4, 16 + vmovdqa 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. + + vpxor xmm1, xmm1 + mov r11, rsp + vmovdqa [r11], xmm1 + + cmp arg4, 4 + jl _only_less_than_4 + + ; backup the counter value + mov tmp1, arg4 + cmp arg4, 8 + jl _less_than_8_left + + ; load 8 Bytes + mov rax, [arg3] + mov [arg2], rax + mov [r11], rax + add r11, 8 + sub arg4, 8 + add arg3, 8 + add arg2, 8 +_less_than_8_left: + + cmp arg4, 4 + jl _less_than_4_left + + ; load 4 Bytes + mov eax, [arg3] + mov [arg2], eax + mov [r11], eax + add r11, 4 + sub arg4, 4 + add arg3, 4 + add arg2, 4 +_less_than_4_left: + + cmp arg4, 2 + jl _less_than_2_left + + ; load 2 Bytes + mov ax, [arg3] + mov [arg2], ax + mov [r11], ax + add r11, 2 + sub arg4, 2 + add arg3, 2 + add arg2, 2 +_less_than_2_left: + cmp arg4, 1 + jl _zero_left + + ; load 1 Byte + mov al, [arg3] + mov [arg2], al + mov [r11], al +_zero_left: + vmovdqa xmm3, [rsp] + vpshufb xmm3, xmm7 + vpxor xmm3, xmm0 ; xor the initial crc value + + ; shl tmp1, 4 + lea rax, [pshufb_shf_table + 16] + sub rax, tmp1 + vmovdqu xmm0, [rax] + vpxor xmm0, [mask1] + + vpshufb xmm3, xmm0 + jmp _128_done + +align 16 +_exact_16_left: + vmovdqu xmm3, [arg3] + vmovdqu [arg2], xmm3 + vpshufb xmm3, xmm7 + vpxor xmm3, xmm0 ; xor the initial crc value + + jmp _128_done + +_only_less_than_4: + cmp arg4, 3 + jl _only_less_than_3 + + ; load 3 Bytes + mov al, [arg3] + mov [arg2], al + mov [r11], al + + mov al, [arg3+1] + mov [arg2+1], al + mov [r11+1], al + + mov al, [arg3+2] + mov [arg2+2], al + mov [r11+2], al + + vmovdqa xmm3, [rsp] + vpshufb xmm3, xmm7 + vpxor xmm3, xmm0 ; xor the initial crc value + + vpsrldq xmm3, 5 + + jmp _barrett +_only_less_than_3: + cmp arg4, 2 + jl _only_less_than_2 + + ; load 2 Bytes + mov al, [arg3] + mov [arg2], al + mov [r11], al + + mov al, [arg3+1] + mov [arg2+1], al + mov [r11+1], al + + vmovdqa xmm3, [rsp] + vpshufb xmm3, xmm7 + vpxor xmm3, xmm0 ; xor the initial crc value + + vpsrldq xmm3, 6 + + jmp _barrett +_only_less_than_2: + + ; load 1 Byte + mov al, [arg3] + mov [arg2],al + mov [r11], al + + vmovdqa xmm3, [rsp] + vpshufb xmm3, xmm7 + vpxor xmm3, xmm0 ; xor the initial crc value + + vpsrldq 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 |