1 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2 ; Copyright(c) 2011-2016 Intel Corporation All rights reserved.
4 ; Redistribution and use in source and binary forms, with or without
5 ; modification, are permitted provided that the following conditions
7 ; * Redistributions of source code must retain the above copyright
8 ; notice, this list of conditions and the following disclaimer.
9 ; * Redistributions in binary form must reproduce the above copyright
10 ; notice, this list of conditions and the following disclaimer in
11 ; the documentation and/or other materials provided with the
13 ; * Neither the name of Intel Corporation nor the names of its
14 ; contributors may be used to endorse or promote products derived
15 ; from this software without specific prior written permission.
17 ; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
18 ; "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
19 ; LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
20 ; A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
21 ; OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
22 ; SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
23 ; LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
24 ; DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
25 ; THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
26 ; (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
27 ; OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
31 ; uint64_t crc64_jones_norm_by8(
32 ; uint64_t init_crc, //initial CRC value, 64 bits
33 ; const unsigned char *buf, //buffer pointer to calculate CRC on
34 ; uint64_t len //buffer length in bytes (64-bit data)
37 %include "reg_sizes.asm"
39 %define fetch_dist 1024
46 %ifidn __OUTPUT_FORMAT__, win64
57 %ifidn __OUTPUT_FORMAT__, win64
59 %define VARIABLE_OFFSET 16*10+8
61 %define VARIABLE_OFFSET 16*2+8
64 global crc64_jones_norm_by8:function
69 sub rsp,VARIABLE_OFFSET
71 %ifidn __OUTPUT_FORMAT__, win64
72 ; push the xmm registers into the stack to maintain
73 movdqa [rsp + XMM_SAVE + 16*0], xmm6
74 movdqa [rsp + XMM_SAVE + 16*1], xmm7
75 movdqa [rsp + XMM_SAVE + 16*2], xmm8
76 movdqa [rsp + XMM_SAVE + 16*3], xmm9
77 movdqa [rsp + XMM_SAVE + 16*4], xmm10
78 movdqa [rsp + XMM_SAVE + 16*5], xmm11
79 movdqa [rsp + XMM_SAVE + 16*6], xmm12
80 movdqa [rsp + XMM_SAVE + 16*7], xmm13
84 ; check if smaller than 256
87 ; for sizes less than 256, we can't fold 128B at a time...
91 ; load the initial crc value
92 movq xmm10, arg1 ; initial crc
94 ; crc value does not need to be byte-reflected, but it needs to be moved to the high part of the register.
95 ; because data will be byte-reflected and will align with initial crc at correct place.
98 movdqa xmm11, [SHUF_MASK]
99 ; receive the initial 128B data, xor the initial crc value
100 movdqu xmm0, [arg2+16*0]
101 movdqu xmm1, [arg2+16*1]
102 movdqu xmm2, [arg2+16*2]
103 movdqu xmm3, [arg2+16*3]
104 movdqu xmm4, [arg2+16*4]
105 movdqu xmm5, [arg2+16*5]
106 movdqu xmm6, [arg2+16*6]
107 movdqu xmm7, [arg2+16*7]
110 ; XOR the initial_crc value
120 movdqa xmm10, [rk3] ;xmm10 has rk3 and rk4
121 ;imm value of pclmulqdq instruction will determine which constant to use
122 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
123 ; we subtract 256 instead of 128 to save one instruction from the loop
126 ; at this section of the code, there is 128*x+y (0<=y<128) bytes of buffer. The _fold_128_B_loop
127 ; loop will fold 128B at a time until we have 128+y Bytes of buffer
130 ; fold 128B at a time. This section of the code folds 8 xmm registers in parallel
133 ; update the buffer pointer
134 add arg2, 128 ; buf += 128;
136 prefetchnta [arg2+fetch_dist+0]
137 movdqu xmm9, [arg2+16*0]
138 movdqu xmm12, [arg2+16*1]
143 pclmulqdq xmm0, xmm10, 0x0
144 pclmulqdq xmm8, xmm10 , 0x11
145 pclmulqdq xmm1, xmm10, 0x0
146 pclmulqdq xmm13, xmm10 , 0x11
152 prefetchnta [arg2+fetch_dist+32]
153 movdqu xmm9, [arg2+16*2]
154 movdqu xmm12, [arg2+16*3]
159 pclmulqdq xmm2, xmm10, 0x0
160 pclmulqdq xmm8, xmm10 , 0x11
161 pclmulqdq xmm3, xmm10, 0x0
162 pclmulqdq xmm13, xmm10 , 0x11
168 prefetchnta [arg2+fetch_dist+64]
169 movdqu xmm9, [arg2+16*4]
170 movdqu xmm12, [arg2+16*5]
175 pclmulqdq xmm4, xmm10, 0x0
176 pclmulqdq xmm8, xmm10 , 0x11
177 pclmulqdq xmm5, xmm10, 0x0
178 pclmulqdq xmm13, xmm10 , 0x11
184 prefetchnta [arg2+fetch_dist+96]
185 movdqu xmm9, [arg2+16*6]
186 movdqu xmm12, [arg2+16*7]
191 pclmulqdq xmm6, xmm10, 0x0
192 pclmulqdq xmm8, xmm10 , 0x11
193 pclmulqdq xmm7, xmm10, 0x0
194 pclmulqdq xmm13, xmm10 , 0x11
202 ; check if there is another 128B in the buffer to be able to fold
204 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
207 ; at this point, the buffer pointer is pointing at the last y Bytes of the buffer, where 0 <= y < 128
208 ; the 128B of folded data is in 8 of the xmm registers: xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7
211 ; fold the 8 xmm registers to 1 xmm register with different constants
215 pclmulqdq xmm0, xmm10, 0x11
216 pclmulqdq xmm8, xmm10, 0x0
222 pclmulqdq xmm1, xmm10, 0x11
223 pclmulqdq xmm8, xmm10, 0x0
229 pclmulqdq xmm2, xmm10, 0x11
230 pclmulqdq xmm8, xmm10, 0x0
236 pclmulqdq xmm3, xmm10, 0x11
237 pclmulqdq xmm8, xmm10, 0x0
243 pclmulqdq xmm4, xmm10, 0x11
244 pclmulqdq xmm8, xmm10, 0x0
250 pclmulqdq xmm5, xmm10, 0x11
251 pclmulqdq xmm8, xmm10, 0x0
255 movdqa xmm10, [rk1] ;xmm10 has rk1 and rk2
258 pclmulqdq xmm6, xmm10, 0x11
259 pclmulqdq xmm8, xmm10, 0x0
264 ; instead of 128, we add 112 to the loop counter to save 1 instruction from the loop
265 ; instead of a cmp instruction, we use the negative flag with the jl instruction
267 jl _final_reduction_for_128
269 ; now we have 16+y bytes left to reduce. 16 Bytes is in register xmm7 and the rest is in memory
270 ; we can fold 16 bytes at a time if y>=16
271 ; continue folding 16B at a time
275 pclmulqdq xmm7, xmm10, 0x11
276 pclmulqdq xmm8, xmm10, 0x0
283 ; instead of a cmp instruction, we utilize the flags with the jge instruction
284 ; equivalent of: cmp arg3, 16-16
285 ; check if there is any more 16B in the buffer to be able to fold
286 jge _16B_reduction_loop
288 ;now we have 16+z bytes left to reduce, where 0<= z < 16.
289 ;first, we reduce the data in the xmm7 register
292 _final_reduction_for_128:
293 ; check if any more data to fold. If not, compute the CRC of the final 128 bits
297 ; here we are getting data that is less than 16 bytes.
298 ; 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.
299 ; after that the registers need to be adjusted.
303 movdqu xmm1, [arg2 - 16 + arg3]
306 ; get rid of the extra data that was loaded before
307 ; load the shift constant
308 lea rax, [pshufb_shf_table + 16]
312 ; shift xmm2 to the left by arg3 bytes
315 ; shift xmm7 to the right by 16-arg3 bytes
318 pblendvb xmm1, xmm2 ;xmm0 is implicit
323 pclmulqdq xmm7, xmm10, 0x11
324 pclmulqdq xmm8, xmm10, 0x0
329 ; compute crc of a 128-bit value
330 movdqa xmm10, [rk5] ; rk5 and rk6 in xmm10
334 pclmulqdq xmm7, xmm10, 0x01 ; H*L
340 movdqa xmm10, [rk7] ; rk7 and rk8 in xmm10
345 pclmulqdq xmm7, xmm10, 0x01
348 pclmulqdq xmm7, xmm10, 0x11
354 %ifidn __OUTPUT_FORMAT__, win64
355 movdqa xmm6, [rsp + XMM_SAVE + 16*0]
356 movdqa xmm7, [rsp + XMM_SAVE + 16*1]
357 movdqa xmm8, [rsp + XMM_SAVE + 16*2]
358 movdqa xmm9, [rsp + XMM_SAVE + 16*3]
359 movdqa xmm10, [rsp + XMM_SAVE + 16*4]
360 movdqa xmm11, [rsp + XMM_SAVE + 16*5]
361 movdqa xmm12, [rsp + XMM_SAVE + 16*6]
362 movdqa xmm13, [rsp + XMM_SAVE + 16*7]
364 add rsp, VARIABLE_OFFSET
367 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
368 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
369 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
370 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
375 ; check if there is enough buffer to be able to fold 16B at a time
378 movdqa xmm11, [SHUF_MASK]
380 ; if there is, load the constants
381 movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10
383 movq xmm0, arg1 ; get the initial crc value
384 pslldq xmm0, 8 ; align it to its correct place
385 movdqu xmm7, [arg2] ; load the plaintext
386 pshufb xmm7, xmm11 ; byte-reflect the plaintext
390 ; update the buffer pointer
393 ; update the counter. subtract 32 instead of 16 to save one instruction from the loop
396 jmp _16B_reduction_loop
399 ; mov initial crc to the return value. this is necessary for zero-length buffers.
404 movdqa xmm11, [SHUF_MASK]
406 movq xmm0, arg1 ; get the initial crc value
407 pslldq xmm0, 8 ; align it to its correct place
411 jl _less_than_16_left
413 movdqu xmm7, [arg2] ; load the plaintext
414 pshufb xmm7, xmm11 ; byte-reflect the plaintext
415 pxor xmm7, xmm0 ; xor the initial crc value
418 movdqa xmm10, [rk1] ; rk1 and rk2 in xmm10
419 jmp _get_last_two_xmms
422 ; use stack space to load data less than 16 bytes, zero-out the 16B in memory first.
427 ; backup the counter value
470 pxor xmm7, xmm0 ; xor the initial crc value
473 lea rax, [pshufb_shf_table + 16]
487 ; Right shift (8-length) bytes in XMM
497 pxor xmm7, xmm0 ; xor the initial crc value
503 ; precomputed constants
507 DQ 0x4445ed2750017038
509 DQ 0x698b74157cfbd736
511 DQ 0x0cfcfb5101c4b775
513 DQ 0x65403fd47cbec866
515 DQ 0x4445ed2750017038
517 DQ 0x0000000000000000
519 DQ 0xddf3eeb298be6cf8
521 DQ 0xad93d23594c935a9
523 DQ 0xd8dc208e2ba527b4
525 DQ 0xf032cfec76bb2bc5
527 DQ 0xb536044f357f4238
529 DQ 0xfdbf104d938ba67a
531 DQ 0xeeddad9297a843e7
533 DQ 0x3550bce629466473
535 DQ 0x4e501e58ca43d25e
537 DQ 0x13c961588f27f643
539 DQ 0x3b60d00dcb1099bc
541 DQ 0x44bf1f468c53b9a3
543 DQ 0x96f2236e317179ee
545 DQ 0xf00839aa0dd64bac
548 dq 0x8080808080808080, 0x8080808080808080
550 dq 0xFFFFFFFFFFFFFFFF, 0x00000000FFFFFFFF
552 dq 0x0000000000000000, 0xFFFFFFFFFFFFFFFF
555 dq 0x08090A0B0C0D0E0F, 0x0001020304050607
558 ; use these values for shift constants for the pshufb instruction
559 ; different alignments result in values as shown:
560 ; dq 0x8887868584838281, 0x008f8e8d8c8b8a89 ; shl 15 (16-1) / shr1
561 ; dq 0x8988878685848382, 0x01008f8e8d8c8b8a ; shl 14 (16-3) / shr2
562 ; dq 0x8a89888786858483, 0x0201008f8e8d8c8b ; shl 13 (16-4) / shr3
563 ; dq 0x8b8a898887868584, 0x030201008f8e8d8c ; shl 12 (16-4) / shr4
564 ; dq 0x8c8b8a8988878685, 0x04030201008f8e8d ; shl 11 (16-5) / shr5
565 ; dq 0x8d8c8b8a89888786, 0x0504030201008f8e ; shl 10 (16-6) / shr6
566 ; dq 0x8e8d8c8b8a898887, 0x060504030201008f ; shl 9 (16-7) / shr7
567 ; dq 0x8f8e8d8c8b8a8988, 0x0706050403020100 ; shl 8 (16-8) / shr8
568 ; dq 0x008f8e8d8c8b8a89, 0x0807060504030201 ; shl 7 (16-9) / shr9
569 ; dq 0x01008f8e8d8c8b8a, 0x0908070605040302 ; shl 6 (16-10) / shr10
570 ; dq 0x0201008f8e8d8c8b, 0x0a09080706050403 ; shl 5 (16-11) / shr11
571 ; dq 0x030201008f8e8d8c, 0x0b0a090807060504 ; shl 4 (16-12) / shr12
572 ; dq 0x04030201008f8e8d, 0x0c0b0a0908070605 ; shl 3 (16-13) / shr13
573 ; dq 0x0504030201008f8e, 0x0d0c0b0a09080706 ; shl 2 (16-14) / shr14
574 ; dq 0x060504030201008f, 0x0e0d0c0b0a090807 ; shl 1 (16-15) / shr15
575 dq 0x8786858483828100, 0x8f8e8d8c8b8a8988
576 dq 0x0706050403020100, 0x0f0e0d0c0b0a0908
577 dq 0x8080808080808080, 0x0f0e0d0c0b0a0908
578 dq 0x8080808080808080, 0x8080808080808080
580 ;;; func core, ver, snum
581 slversion crc64_jones_norm_by8, 01, 00, 0026