aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorGeorge Spelvin <linux@horizon.com>2014-06-06 23:08:58 -0400
committerHerbert Xu <herbert@gondor.apana.org.au>2014-06-20 09:27:57 -0400
commit473946e674eb797940c0a1aea8a999bc8fa9fcff (patch)
treed3d8cf86e95ebb7aaea477b5551aaebcf0f7a162
parentcea4001ae1f80270a30031c6de139313e4dda213 (diff)
crypto: crc32c-pclmul - Shrink K_table to 32-bit words
There's no need for the K_table to be made of 64-bit words. For some reason, the original authors didn't fully reduce the values modulo the CRC32C polynomial, and so had some 33-bit values in there. They can all be reduced to 32 bits. Doing that cuts the table size in half. Since the code depends on both pclmulq and crc32, SSE 4.1 is obviously present, so we can use pmovzxdq to fetch it in the correct format. This adds (measured on Ivy Bridge) 1 cycle per main loop iteration (CRC of up to 3K bytes), less than 0.2%. The hope is that the reduced D-cache footprint will make up the loss in other code. Two other related fixes: * K_table is read-only, so belongs in .rodata, and * There's no need for more than 8-byte alignment Acked-by: Tim Chen <tim.c.chen@linux.intel.com> Signed-off-by: George Spelvin <linux@horizon.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
-rw-r--r--arch/x86/crypto/crc32c-pcl-intel-asm_64.S281
1 files changed, 139 insertions, 142 deletions
diff --git a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
index dbc4339b5417..26d49ebae040 100644
--- a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
+++ b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
@@ -72,6 +72,7 @@
72 72
73# unsigned int crc_pcl(u8 *buffer, int len, unsigned int crc_init); 73# unsigned int crc_pcl(u8 *buffer, int len, unsigned int crc_init);
74 74
75.text
75ENTRY(crc_pcl) 76ENTRY(crc_pcl)
76#define bufp %rdi 77#define bufp %rdi
77#define bufp_dw %edi 78#define bufp_dw %edi
@@ -216,15 +217,11 @@ LABEL crc_ %i
216 ## 4) Combine three results: 217 ## 4) Combine three results:
217 ################################################################ 218 ################################################################
218 219
219 lea (K_table-16)(%rip), bufp # first entry is for idx 1 220 lea (K_table-8)(%rip), bufp # first entry is for idx 1
220 shlq $3, %rax # rax *= 8 221 shlq $3, %rax # rax *= 8
221 subq %rax, tmp # tmp -= rax*8 222 pmovzxdq (bufp,%rax), %xmm0 # 2 consts: K1:K2
222 shlq $1, %rax 223 leal (%eax,%eax,2), %eax # rax *= 3 (total *24)
223 subq %rax, tmp # tmp -= rax*16 224 subq %rax, tmp # tmp -= rax*24
224 # (total tmp -= rax*24)
225 addq %rax, bufp
226
227 movdqa (bufp), %xmm0 # 2 consts: K1:K2
228 225
229 movq crc_init, %xmm1 # CRC for block 1 226 movq crc_init, %xmm1 # CRC for block 1
230 PCLMULQDQ 0x00,%xmm0,%xmm1 # Multiply by K2 227 PCLMULQDQ 0x00,%xmm0,%xmm1 # Multiply by K2
@@ -238,9 +235,9 @@ LABEL crc_ %i
238 mov crc2, crc_init 235 mov crc2, crc_init
239 crc32 %rax, crc_init 236 crc32 %rax, crc_init
240 237
241################################################################ 238 ################################################################
242## 5) Check for end: 239 ## 5) Check for end:
243################################################################ 240 ################################################################
244 241
245LABEL crc_ 0 242LABEL crc_ 0
246 mov tmp, len 243 mov tmp, len
@@ -331,136 +328,136 @@ ENDPROC(crc_pcl)
331 328
332 ################################################################ 329 ################################################################
333 ## PCLMULQDQ tables 330 ## PCLMULQDQ tables
334 ## Table is 128 entries x 2 quad words each 331 ## Table is 128 entries x 2 words (8 bytes) each
335 ################################################################ 332 ################################################################
336.data 333.section .rotata, "a", %progbits
337.align 64 334.align 8
338K_table: 335K_table:
339 .quad 0x14cd00bd6,0x105ec76f0 336 .long 0x493c7d27, 0x00000001
340 .quad 0x0ba4fc28e,0x14cd00bd6 337 .long 0xba4fc28e, 0x493c7d27
341 .quad 0x1d82c63da,0x0f20c0dfe 338 .long 0xddc0152b, 0xf20c0dfe
342 .quad 0x09e4addf8,0x0ba4fc28e 339 .long 0x9e4addf8, 0xba4fc28e
343 .quad 0x039d3b296,0x1384aa63a 340 .long 0x39d3b296, 0x3da6d0cb
344 .quad 0x102f9b8a2,0x1d82c63da 341 .long 0x0715ce53, 0xddc0152b
345 .quad 0x14237f5e6,0x01c291d04 342 .long 0x47db8317, 0x1c291d04
346 .quad 0x00d3b6092,0x09e4addf8 343 .long 0x0d3b6092, 0x9e4addf8
347 .quad 0x0c96cfdc0,0x0740eef02 344 .long 0xc96cfdc0, 0x740eef02
348 .quad 0x18266e456,0x039d3b296 345 .long 0x878a92a7, 0x39d3b296
349 .quad 0x0daece73e,0x0083a6eec 346 .long 0xdaece73e, 0x083a6eec
350 .quad 0x0ab7aff2a,0x102f9b8a2 347 .long 0xab7aff2a, 0x0715ce53
351 .quad 0x1248ea574,0x1c1733996 348 .long 0x2162d385, 0xc49f4f67
352 .quad 0x083348832,0x14237f5e6 349 .long 0x83348832, 0x47db8317
353 .quad 0x12c743124,0x02ad91c30 350 .long 0x299847d5, 0x2ad91c30
354 .quad 0x0b9e02b86,0x00d3b6092 351 .long 0xb9e02b86, 0x0d3b6092
355 .quad 0x018b33a4e,0x06992cea2 352 .long 0x18b33a4e, 0x6992cea2
356 .quad 0x1b331e26a,0x0c96cfdc0 353 .long 0xb6dd949b, 0xc96cfdc0
357 .quad 0x17d35ba46,0x07e908048 354 .long 0x78d9ccb7, 0x7e908048
358 .quad 0x1bf2e8b8a,0x18266e456 355 .long 0xbac2fd7b, 0x878a92a7
359 .quad 0x1a3e0968a,0x11ed1f9d8 356 .long 0xa60ce07b, 0x1b3d8f29
360 .quad 0x0ce7f39f4,0x0daece73e 357 .long 0xce7f39f4, 0xdaece73e
361 .quad 0x061d82e56,0x0f1d0f55e 358 .long 0x61d82e56, 0xf1d0f55e
362 .quad 0x0d270f1a2,0x0ab7aff2a 359 .long 0xd270f1a2, 0xab7aff2a
363 .quad 0x1c3f5f66c,0x0a87ab8a8 360 .long 0xc619809d, 0xa87ab8a8
364 .quad 0x12ed0daac,0x1248ea574 361 .long 0x2b3cac5d, 0x2162d385
365 .quad 0x065863b64,0x08462d800 362 .long 0x65863b64, 0x8462d800
366 .quad 0x11eef4f8e,0x083348832 363 .long 0x1b03397f, 0x83348832
367 .quad 0x1ee54f54c,0x071d111a8 364 .long 0xebb883bd, 0x71d111a8
368 .quad 0x0b3e32c28,0x12c743124 365 .long 0xb3e32c28, 0x299847d5
369 .quad 0x0064f7f26,0x0ffd852c6 366 .long 0x064f7f26, 0xffd852c6
370 .quad 0x0dd7e3b0c,0x0b9e02b86 367 .long 0xdd7e3b0c, 0xb9e02b86
371 .quad 0x0f285651c,0x0dcb17aa4 368 .long 0xf285651c, 0xdcb17aa4
372 .quad 0x010746f3c,0x018b33a4e 369 .long 0x10746f3c, 0x18b33a4e
373 .quad 0x1c24afea4,0x0f37c5aee 370 .long 0xc7a68855, 0xf37c5aee
374 .quad 0x0271d9844,0x1b331e26a 371 .long 0x271d9844, 0xb6dd949b
375 .quad 0x08e766a0c,0x06051d5a2 372 .long 0x8e766a0c, 0x6051d5a2
376 .quad 0x093a5f730,0x17d35ba46 373 .long 0x93a5f730, 0x78d9ccb7
377 .quad 0x06cb08e5c,0x11d5ca20e 374 .long 0x6cb08e5c, 0x18b0d4ff
378 .quad 0x06b749fb2,0x1bf2e8b8a 375 .long 0x6b749fb2, 0xbac2fd7b
379 .quad 0x1167f94f2,0x021f3d99c 376 .long 0x1393e203, 0x21f3d99c
380 .quad 0x0cec3662e,0x1a3e0968a 377 .long 0xcec3662e, 0xa60ce07b
381 .quad 0x19329634a,0x08f158014 378 .long 0x96c515bb, 0x8f158014
382 .quad 0x0e6fc4e6a,0x0ce7f39f4 379 .long 0xe6fc4e6a, 0xce7f39f4
383 .quad 0x08227bb8a,0x1a5e82106 380 .long 0x8227bb8a, 0xa00457f7
384 .quad 0x0b0cd4768,0x061d82e56 381 .long 0xb0cd4768, 0x61d82e56
385 .quad 0x13c2b89c4,0x188815ab2 382 .long 0x39c7ff35, 0x8d6d2c43
386 .quad 0x0d7a4825c,0x0d270f1a2 383 .long 0xd7a4825c, 0xd270f1a2
387 .quad 0x10f5ff2ba,0x105405f3e 384 .long 0x0ab3844b, 0x00ac29cf
388 .quad 0x00167d312,0x1c3f5f66c 385 .long 0x0167d312, 0xc619809d
389 .quad 0x0f6076544,0x0e9adf796 386 .long 0xf6076544, 0xe9adf796
390 .quad 0x026f6a60a,0x12ed0daac 387 .long 0x26f6a60a, 0x2b3cac5d
391 .quad 0x1a2adb74e,0x096638b34 388 .long 0xa741c1bf, 0x96638b34
392 .quad 0x19d34af3a,0x065863b64 389 .long 0x98d8d9cb, 0x65863b64
393 .quad 0x049c3cc9c,0x1e50585a0 390 .long 0x49c3cc9c, 0xe0e9f351
394 .quad 0x068bce87a,0x11eef4f8e 391 .long 0x68bce87a, 0x1b03397f
395 .quad 0x1524fa6c6,0x19f1c69dc 392 .long 0x57a3d037, 0x9af01f2d
396 .quad 0x16cba8aca,0x1ee54f54c 393 .long 0x6956fc3b, 0xebb883bd
397 .quad 0x042d98888,0x12913343e 394 .long 0x42d98888, 0x2cff42cf
398 .quad 0x1329d9f7e,0x0b3e32c28 395 .long 0x3771e98f, 0xb3e32c28
399 .quad 0x1b1c69528,0x088f25a3a 396 .long 0xb42ae3d9, 0x88f25a3a
400 .quad 0x02178513a,0x0064f7f26 397 .long 0x2178513a, 0x064f7f26
401 .quad 0x0e0ac139e,0x04e36f0b0 398 .long 0xe0ac139e, 0x4e36f0b0
402 .quad 0x0170076fa,0x0dd7e3b0c 399 .long 0x170076fa, 0xdd7e3b0c
403 .quad 0x141a1a2e2,0x0bd6f81f8 400 .long 0x444dd413, 0xbd6f81f8
404 .quad 0x16ad828b4,0x0f285651c 401 .long 0x6f345e45, 0xf285651c
405 .quad 0x041d17b64,0x19425cbba 402 .long 0x41d17b64, 0x91c9bd4b
406 .quad 0x1fae1cc66,0x010746f3c 403 .long 0xff0dba97, 0x10746f3c
407 .quad 0x1a75b4b00,0x18db37e8a 404 .long 0xa2b73df1, 0x885f087b
408 .quad 0x0f872e54c,0x1c24afea4 405 .long 0xf872e54c, 0xc7a68855
409 .quad 0x01e41e9fc,0x04c144932 406 .long 0x1e41e9fc, 0x4c144932
410 .quad 0x086d8e4d2,0x0271d9844 407 .long 0x86d8e4d2, 0x271d9844
411 .quad 0x160f7af7a,0x052148f02 408 .long 0x651bd98b, 0x52148f02
412 .quad 0x05bb8f1bc,0x08e766a0c 409 .long 0x5bb8f1bc, 0x8e766a0c
413 .quad 0x0a90fd27a,0x0a3c6f37a 410 .long 0xa90fd27a, 0xa3c6f37a
414 .quad 0x0b3af077a,0x093a5f730 411 .long 0xb3af077a, 0x93a5f730
415 .quad 0x04984d782,0x1d22c238e 412 .long 0x4984d782, 0xd7c0557f
416 .quad 0x0ca6ef3ac,0x06cb08e5c 413 .long 0xca6ef3ac, 0x6cb08e5c
417 .quad 0x0234e0b26,0x063ded06a 414 .long 0x234e0b26, 0x63ded06a
418 .quad 0x1d88abd4a,0x06b749fb2 415 .long 0xdd66cbbb, 0x6b749fb2
419 .quad 0x04597456a,0x04d56973c 416 .long 0x4597456a, 0x4d56973c
420 .quad 0x0e9e28eb4,0x1167f94f2 417 .long 0xe9e28eb4, 0x1393e203
421 .quad 0x07b3ff57a,0x19385bf2e 418 .long 0x7b3ff57a, 0x9669c9df
422 .quad 0x0c9c8b782,0x0cec3662e 419 .long 0xc9c8b782, 0xcec3662e
423 .quad 0x13a9cba9e,0x0e417f38a 420 .long 0x3f70cc6f, 0xe417f38a
424 .quad 0x093e106a4,0x19329634a 421 .long 0x93e106a4, 0x96c515bb
425 .quad 0x167001a9c,0x14e727980 422 .long 0x62ec6c6d, 0x4b9e0f71
426 .quad 0x1ddffc5d4,0x0e6fc4e6a 423 .long 0xd813b325, 0xe6fc4e6a
427 .quad 0x00df04680,0x0d104b8fc 424 .long 0x0df04680, 0xd104b8fc
428 .quad 0x02342001e,0x08227bb8a 425 .long 0x2342001e, 0x8227bb8a
429 .quad 0x00a2a8d7e,0x05b397730 426 .long 0x0a2a8d7e, 0x5b397730
430 .quad 0x168763fa6,0x0b0cd4768 427 .long 0x6d9a4957, 0xb0cd4768
431 .quad 0x1ed5a407a,0x0e78eb416 428 .long 0xe8b6368b, 0xe78eb416
432 .quad 0x0d2c3ed1a,0x13c2b89c4 429 .long 0xd2c3ed1a, 0x39c7ff35
433 .quad 0x0995a5724,0x1641378f0 430 .long 0x995a5724, 0x61ff0e01
434 .quad 0x19b1afbc4,0x0d7a4825c 431 .long 0x9ef68d35, 0xd7a4825c
435 .quad 0x109ffedc0,0x08d96551c 432 .long 0x0c139b31, 0x8d96551c
436 .quad 0x0f2271e60,0x10f5ff2ba 433 .long 0xf2271e60, 0x0ab3844b
437 .quad 0x00b0bf8ca,0x00bf80dd2 434 .long 0x0b0bf8ca, 0x0bf80dd2
438 .quad 0x123888b7a,0x00167d312 435 .long 0x2664fd8b, 0x0167d312
439 .quad 0x1e888f7dc,0x18dcddd1c 436 .long 0xed64812d, 0x8821abed
440 .quad 0x002ee03b2,0x0f6076544 437 .long 0x02ee03b2, 0xf6076544
441 .quad 0x183e8d8fe,0x06a45d2b2 438 .long 0x8604ae0f, 0x6a45d2b2
442 .quad 0x133d7a042,0x026f6a60a 439 .long 0x363bd6b3, 0x26f6a60a
443 .quad 0x116b0f50c,0x1dd3e10e8 440 .long 0x135c83fd, 0xd8d26619
444 .quad 0x05fabe670,0x1a2adb74e 441 .long 0x5fabe670, 0xa741c1bf
445 .quad 0x130004488,0x0de87806c 442 .long 0x35ec3279, 0xde87806c
446 .quad 0x000bcf5f6,0x19d34af3a 443 .long 0x00bcf5f6, 0x98d8d9cb
447 .quad 0x18f0c7078,0x014338754 444 .long 0x8ae00689, 0x14338754
448 .quad 0x017f27698,0x049c3cc9c 445 .long 0x17f27698, 0x49c3cc9c
449 .quad 0x058ca5f00,0x15e3e77ee 446 .long 0x58ca5f00, 0x5bd2011f
450 .quad 0x1af900c24,0x068bce87a 447 .long 0xaa7c7ad5, 0x68bce87a
451 .quad 0x0b5cfca28,0x0dd07448e 448 .long 0xb5cfca28, 0xdd07448e
452 .quad 0x0ded288f8,0x1524fa6c6 449 .long 0xded288f8, 0x57a3d037
453 .quad 0x059f229bc,0x1d8048348 450 .long 0x59f229bc, 0xdde8f5b9
454 .quad 0x06d390dec,0x16cba8aca 451 .long 0x6d390dec, 0x6956fc3b
455 .quad 0x037170390,0x0a3e3e02c 452 .long 0x37170390, 0xa3e3e02c
456 .quad 0x06353c1cc,0x042d98888 453 .long 0x6353c1cc, 0x42d98888
457 .quad 0x0c4584f5c,0x0d73c7bea 454 .long 0xc4584f5c, 0xd73c7bea
458 .quad 0x1f16a3418,0x1329d9f7e 455 .long 0xf48642e9, 0x3771e98f
459 .quad 0x0531377e2,0x185137662 456 .long 0x531377e2, 0x80ff0093
460 .quad 0x1d8d9ca7c,0x1b1c69528 457 .long 0xdd35bc8d, 0xb42ae3d9
461 .quad 0x0b25b29f2,0x18a08b5bc 458 .long 0xb25b29f2, 0x8fe4c34d
462 .quad 0x19fb2a8b0,0x02178513a 459 .long 0x9a5ede41, 0x2178513a
463 .quad 0x1a08fe6ac,0x1da758ae0 460 .long 0xa563905d, 0xdf99fc11
464 .quad 0x045cddf4e,0x0e0ac139e 461 .long 0x45cddf4e, 0xe0ac139e
465 .quad 0x1a91647f2,0x169cf9eb0 462 .long 0xacfa3103, 0x6c23e841
466 .quad 0x1a0f717c4,0x0170076fa 463 .long 0xa51b6135, 0x170076fa