sse2-memset16-atom.S 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722
  1. /*
  2. * Copyright (C) 2010 The Android Open Source Project
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. /*
  17. * Contributed by: Intel Corporation
  18. */
  19. #ifndef L
  20. # define L(label) .L##label
  21. #endif
  22. #ifndef ALIGN
  23. # define ALIGN(n) .p2align n
  24. #endif
  25. #ifndef cfi_startproc
  26. # define cfi_startproc .cfi_startproc
  27. #endif
  28. #ifndef cfi_endproc
  29. # define cfi_endproc .cfi_endproc
  30. #endif
  31. #ifndef cfi_rel_offset
  32. # define cfi_rel_offset(reg, off) .cfi_rel_offset reg, off
  33. #endif
  34. #ifndef cfi_restore
  35. # define cfi_restore(reg) .cfi_restore reg
  36. #endif
  37. #ifndef cfi_adjust_cfa_offset
  38. # define cfi_adjust_cfa_offset(off) .cfi_adjust_cfa_offset off
  39. #endif
  40. #ifndef ENTRY
  41. # define ENTRY(name) \
  42. .type name, @function; \
  43. .globl name; \
  44. .p2align 4; \
  45. name: \
  46. cfi_startproc
  47. #endif
  48. #ifndef END
  49. # define END(name) \
  50. cfi_endproc; \
  51. .size name, .-name
  52. #endif
  53. #define CFI_PUSH(REG) \
  54. cfi_adjust_cfa_offset (4); \
  55. cfi_rel_offset (REG, 0)
  56. #define CFI_POP(REG) \
  57. cfi_adjust_cfa_offset (-4); \
  58. cfi_restore (REG)
  59. #define PUSH(REG) pushl REG; CFI_PUSH (REG)
  60. #define POP(REG) popl REG; CFI_POP (REG)
  61. #ifdef USE_AS_BZERO16
  62. # define DEST PARMS
  63. # define LEN DEST+4
  64. #else
  65. # define DEST PARMS
  66. # define CHR DEST+4
  67. # define LEN CHR+4
  68. #endif
  69. #if 1
  70. # define SETRTNVAL
  71. #else
  72. # define SETRTNVAL movl DEST(%esp), %eax
  73. #endif
  74. #ifdef SHARED
  75. # define ENTRANCE PUSH (%ebx);
  76. # define RETURN_END POP (%ebx); ret
  77. # define RETURN RETURN_END; CFI_PUSH (%ebx)
  78. # define PARMS 8 /* Preserve EBX. */
  79. # define JMPTBL(I, B) I - B
  80. /* Load an entry in a jump table into EBX and branch to it. TABLE is a
  81. jump table with relative offsets. */
  82. # define BRANCH_TO_JMPTBL_ENTRY(TABLE) \
  83. /* We first load PC into EBX. */ \
  84. call __i686.get_pc_thunk.bx; \
  85. /* Get the address of the jump table. */ \
  86. add $(TABLE - .), %ebx; \
  87. /* Get the entry and convert the relative offset to the \
  88. absolute address. */ \
  89. add (%ebx,%ecx,4), %ebx; \
  90. /* We loaded the jump table and adjuested EDX. Go. */ \
  91. jmp *%ebx
  92. .section .gnu.linkonce.t.__i686.get_pc_thunk.bx,"ax",@progbits
  93. .globl __i686.get_pc_thunk.bx
  94. .hidden __i686.get_pc_thunk.bx
  95. ALIGN (4)
  96. .type __i686.get_pc_thunk.bx,@function
  97. __i686.get_pc_thunk.bx:
  98. movl (%esp), %ebx
  99. ret
  100. #else
  101. # define ENTRANCE
  102. # define RETURN_END ret
  103. # define RETURN RETURN_END
  104. # define PARMS 4
  105. # define JMPTBL(I, B) I
  106. /* Branch to an entry in a jump table. TABLE is a jump table with
  107. absolute offsets. */
  108. # define BRANCH_TO_JMPTBL_ENTRY(TABLE) \
  109. jmp *TABLE(,%ecx,4)
  110. #endif
  111. .section .text.sse2,"ax",@progbits
  112. ALIGN (4)
  113. ENTRY (sse2_memset16_atom)
  114. ENTRANCE
  115. movl LEN(%esp), %ecx
  116. #ifdef USE_AS_ANDROID
  117. shr $1, %ecx
  118. #endif
  119. #ifdef USE_AS_BZERO16
  120. xor %eax, %eax
  121. #else
  122. movzwl CHR(%esp), %eax
  123. mov %eax, %edx
  124. shl $16, %eax
  125. or %edx, %eax
  126. #endif
  127. movl DEST(%esp), %edx
  128. cmp $32, %ecx
  129. jae L(32wordsormore)
  130. L(write_less32words):
  131. lea (%edx, %ecx, 2), %edx
  132. BRANCH_TO_JMPTBL_ENTRY (L(table_less32words))
  133. .pushsection .rodata.sse2,"a",@progbits
  134. ALIGN (2)
  135. L(table_less32words):
  136. .int JMPTBL (L(write_0words), L(table_less32words))
  137. .int JMPTBL (L(write_1words), L(table_less32words))
  138. .int JMPTBL (L(write_2words), L(table_less32words))
  139. .int JMPTBL (L(write_3words), L(table_less32words))
  140. .int JMPTBL (L(write_4words), L(table_less32words))
  141. .int JMPTBL (L(write_5words), L(table_less32words))
  142. .int JMPTBL (L(write_6words), L(table_less32words))
  143. .int JMPTBL (L(write_7words), L(table_less32words))
  144. .int JMPTBL (L(write_8words), L(table_less32words))
  145. .int JMPTBL (L(write_9words), L(table_less32words))
  146. .int JMPTBL (L(write_10words), L(table_less32words))
  147. .int JMPTBL (L(write_11words), L(table_less32words))
  148. .int JMPTBL (L(write_12words), L(table_less32words))
  149. .int JMPTBL (L(write_13words), L(table_less32words))
  150. .int JMPTBL (L(write_14words), L(table_less32words))
  151. .int JMPTBL (L(write_15words), L(table_less32words))
  152. .int JMPTBL (L(write_16words), L(table_less32words))
  153. .int JMPTBL (L(write_17words), L(table_less32words))
  154. .int JMPTBL (L(write_18words), L(table_less32words))
  155. .int JMPTBL (L(write_19words), L(table_less32words))
  156. .int JMPTBL (L(write_20words), L(table_less32words))
  157. .int JMPTBL (L(write_21words), L(table_less32words))
  158. .int JMPTBL (L(write_22words), L(table_less32words))
  159. .int JMPTBL (L(write_23words), L(table_less32words))
  160. .int JMPTBL (L(write_24words), L(table_less32words))
  161. .int JMPTBL (L(write_25words), L(table_less32words))
  162. .int JMPTBL (L(write_26words), L(table_less32words))
  163. .int JMPTBL (L(write_27words), L(table_less32words))
  164. .int JMPTBL (L(write_28words), L(table_less32words))
  165. .int JMPTBL (L(write_29words), L(table_less32words))
  166. .int JMPTBL (L(write_30words), L(table_less32words))
  167. .int JMPTBL (L(write_31words), L(table_less32words))
  168. .popsection
  169. ALIGN (4)
  170. L(write_28words):
  171. movl %eax, -56(%edx)
  172. movl %eax, -52(%edx)
  173. L(write_24words):
  174. movl %eax, -48(%edx)
  175. movl %eax, -44(%edx)
  176. L(write_20words):
  177. movl %eax, -40(%edx)
  178. movl %eax, -36(%edx)
  179. L(write_16words):
  180. movl %eax, -32(%edx)
  181. movl %eax, -28(%edx)
  182. L(write_12words):
  183. movl %eax, -24(%edx)
  184. movl %eax, -20(%edx)
  185. L(write_8words):
  186. movl %eax, -16(%edx)
  187. movl %eax, -12(%edx)
  188. L(write_4words):
  189. movl %eax, -8(%edx)
  190. movl %eax, -4(%edx)
  191. L(write_0words):
  192. SETRTNVAL
  193. RETURN
  194. ALIGN (4)
  195. L(write_29words):
  196. movl %eax, -58(%edx)
  197. movl %eax, -54(%edx)
  198. L(write_25words):
  199. movl %eax, -50(%edx)
  200. movl %eax, -46(%edx)
  201. L(write_21words):
  202. movl %eax, -42(%edx)
  203. movl %eax, -38(%edx)
  204. L(write_17words):
  205. movl %eax, -34(%edx)
  206. movl %eax, -30(%edx)
  207. L(write_13words):
  208. movl %eax, -26(%edx)
  209. movl %eax, -22(%edx)
  210. L(write_9words):
  211. movl %eax, -18(%edx)
  212. movl %eax, -14(%edx)
  213. L(write_5words):
  214. movl %eax, -10(%edx)
  215. movl %eax, -6(%edx)
  216. L(write_1words):
  217. mov %ax, -2(%edx)
  218. SETRTNVAL
  219. RETURN
  220. ALIGN (4)
  221. L(write_30words):
  222. movl %eax, -60(%edx)
  223. movl %eax, -56(%edx)
  224. L(write_26words):
  225. movl %eax, -52(%edx)
  226. movl %eax, -48(%edx)
  227. L(write_22words):
  228. movl %eax, -44(%edx)
  229. movl %eax, -40(%edx)
  230. L(write_18words):
  231. movl %eax, -36(%edx)
  232. movl %eax, -32(%edx)
  233. L(write_14words):
  234. movl %eax, -28(%edx)
  235. movl %eax, -24(%edx)
  236. L(write_10words):
  237. movl %eax, -20(%edx)
  238. movl %eax, -16(%edx)
  239. L(write_6words):
  240. movl %eax, -12(%edx)
  241. movl %eax, -8(%edx)
  242. L(write_2words):
  243. movl %eax, -4(%edx)
  244. SETRTNVAL
  245. RETURN
  246. ALIGN (4)
  247. L(write_31words):
  248. movl %eax, -62(%edx)
  249. movl %eax, -58(%edx)
  250. L(write_27words):
  251. movl %eax, -54(%edx)
  252. movl %eax, -50(%edx)
  253. L(write_23words):
  254. movl %eax, -46(%edx)
  255. movl %eax, -42(%edx)
  256. L(write_19words):
  257. movl %eax, -38(%edx)
  258. movl %eax, -34(%edx)
  259. L(write_15words):
  260. movl %eax, -30(%edx)
  261. movl %eax, -26(%edx)
  262. L(write_11words):
  263. movl %eax, -22(%edx)
  264. movl %eax, -18(%edx)
  265. L(write_7words):
  266. movl %eax, -14(%edx)
  267. movl %eax, -10(%edx)
  268. L(write_3words):
  269. movl %eax, -6(%edx)
  270. movw %ax, -2(%edx)
  271. SETRTNVAL
  272. RETURN
  273. ALIGN (4)
  274. L(32wordsormore):
  275. shl $1, %ecx
  276. test $0x01, %edx
  277. jz L(aligned2bytes)
  278. mov %eax, (%edx)
  279. mov %eax, -4(%edx, %ecx)
  280. sub $2, %ecx
  281. add $1, %edx
  282. rol $8, %eax
  283. L(aligned2bytes):
  284. #ifdef USE_AS_BZERO16
  285. pxor %xmm0, %xmm0
  286. #else
  287. movd %eax, %xmm0
  288. pshufd $0, %xmm0, %xmm0
  289. #endif
  290. testl $0xf, %edx
  291. jz L(aligned_16)
  292. /* ECX > 32 and EDX is not 16 byte aligned. */
  293. L(not_aligned_16):
  294. movdqu %xmm0, (%edx)
  295. movl %edx, %eax
  296. and $-16, %edx
  297. add $16, %edx
  298. sub %edx, %eax
  299. add %eax, %ecx
  300. movd %xmm0, %eax
  301. ALIGN (4)
  302. L(aligned_16):
  303. cmp $128, %ecx
  304. jae L(128bytesormore)
  305. L(aligned_16_less128bytes):
  306. add %ecx, %edx
  307. shr $1, %ecx
  308. BRANCH_TO_JMPTBL_ENTRY (L(table_16_128bytes))
  309. ALIGN (4)
  310. L(128bytesormore):
  311. #ifdef SHARED_CACHE_SIZE
  312. PUSH (%ebx)
  313. mov $SHARED_CACHE_SIZE, %ebx
  314. #else
  315. # ifdef SHARED
  316. call __i686.get_pc_thunk.bx
  317. add $_GLOBAL_OFFSET_TABLE_, %ebx
  318. mov __x86_shared_cache_size@GOTOFF(%ebx), %ebx
  319. # else
  320. PUSH (%ebx)
  321. mov __x86_shared_cache_size, %ebx
  322. # endif
  323. #endif
  324. cmp %ebx, %ecx
  325. jae L(128bytesormore_nt_start)
  326. #ifdef DATA_CACHE_SIZE
  327. POP (%ebx)
  328. # define RESTORE_EBX_STATE CFI_PUSH (%ebx)
  329. cmp $DATA_CACHE_SIZE, %ecx
  330. #else
  331. # ifdef SHARED
  332. # define RESTORE_EBX_STATE
  333. call __i686.get_pc_thunk.bx
  334. add $_GLOBAL_OFFSET_TABLE_, %ebx
  335. cmp __x86_data_cache_size@GOTOFF(%ebx), %ecx
  336. # else
  337. POP (%ebx)
  338. # define RESTORE_EBX_STATE CFI_PUSH (%ebx)
  339. cmp __x86_data_cache_size, %ecx
  340. # endif
  341. #endif
  342. jae L(128bytes_L2_normal)
  343. subl $128, %ecx
  344. L(128bytesormore_normal):
  345. sub $128, %ecx
  346. movdqa %xmm0, (%edx)
  347. movdqa %xmm0, 0x10(%edx)
  348. movdqa %xmm0, 0x20(%edx)
  349. movdqa %xmm0, 0x30(%edx)
  350. movdqa %xmm0, 0x40(%edx)
  351. movdqa %xmm0, 0x50(%edx)
  352. movdqa %xmm0, 0x60(%edx)
  353. movdqa %xmm0, 0x70(%edx)
  354. lea 128(%edx), %edx
  355. jb L(128bytesless_normal)
  356. sub $128, %ecx
  357. movdqa %xmm0, (%edx)
  358. movdqa %xmm0, 0x10(%edx)
  359. movdqa %xmm0, 0x20(%edx)
  360. movdqa %xmm0, 0x30(%edx)
  361. movdqa %xmm0, 0x40(%edx)
  362. movdqa %xmm0, 0x50(%edx)
  363. movdqa %xmm0, 0x60(%edx)
  364. movdqa %xmm0, 0x70(%edx)
  365. lea 128(%edx), %edx
  366. jae L(128bytesormore_normal)
  367. L(128bytesless_normal):
  368. lea 128(%ecx), %ecx
  369. add %ecx, %edx
  370. shr $1, %ecx
  371. BRANCH_TO_JMPTBL_ENTRY (L(table_16_128bytes))
  372. ALIGN (4)
  373. L(128bytes_L2_normal):
  374. prefetcht0 0x380(%edx)
  375. prefetcht0 0x3c0(%edx)
  376. sub $128, %ecx
  377. movdqa %xmm0, (%edx)
  378. movaps %xmm0, 0x10(%edx)
  379. movaps %xmm0, 0x20(%edx)
  380. movaps %xmm0, 0x30(%edx)
  381. movaps %xmm0, 0x40(%edx)
  382. movaps %xmm0, 0x50(%edx)
  383. movaps %xmm0, 0x60(%edx)
  384. movaps %xmm0, 0x70(%edx)
  385. add $128, %edx
  386. cmp $128, %ecx
  387. jae L(128bytes_L2_normal)
  388. L(128bytesless_L2_normal):
  389. add %ecx, %edx
  390. shr $1, %ecx
  391. BRANCH_TO_JMPTBL_ENTRY (L(table_16_128bytes))
  392. RESTORE_EBX_STATE
  393. L(128bytesormore_nt_start):
  394. sub %ebx, %ecx
  395. mov %ebx, %eax
  396. and $0x7f, %eax
  397. add %eax, %ecx
  398. movd %xmm0, %eax
  399. ALIGN (4)
  400. L(128bytesormore_shared_cache_loop):
  401. prefetcht0 0x3c0(%edx)
  402. prefetcht0 0x380(%edx)
  403. sub $0x80, %ebx
  404. movdqa %xmm0, (%edx)
  405. movdqa %xmm0, 0x10(%edx)
  406. movdqa %xmm0, 0x20(%edx)
  407. movdqa %xmm0, 0x30(%edx)
  408. movdqa %xmm0, 0x40(%edx)
  409. movdqa %xmm0, 0x50(%edx)
  410. movdqa %xmm0, 0x60(%edx)
  411. movdqa %xmm0, 0x70(%edx)
  412. add $0x80, %edx
  413. cmp $0x80, %ebx
  414. jae L(128bytesormore_shared_cache_loop)
  415. cmp $0x80, %ecx
  416. jb L(shared_cache_loop_end)
  417. ALIGN (4)
  418. L(128bytesormore_nt):
  419. sub $0x80, %ecx
  420. movntdq %xmm0, (%edx)
  421. movntdq %xmm0, 0x10(%edx)
  422. movntdq %xmm0, 0x20(%edx)
  423. movntdq %xmm0, 0x30(%edx)
  424. movntdq %xmm0, 0x40(%edx)
  425. movntdq %xmm0, 0x50(%edx)
  426. movntdq %xmm0, 0x60(%edx)
  427. movntdq %xmm0, 0x70(%edx)
  428. add $0x80, %edx
  429. cmp $0x80, %ecx
  430. jae L(128bytesormore_nt)
  431. sfence
  432. L(shared_cache_loop_end):
  433. #if defined DATA_CACHE_SIZE || !defined SHARED
  434. POP (%ebx)
  435. #endif
  436. add %ecx, %edx
  437. shr $1, %ecx
  438. BRANCH_TO_JMPTBL_ENTRY (L(table_16_128bytes))
  439. .pushsection .rodata.sse2,"a",@progbits
  440. ALIGN (2)
  441. L(table_16_128bytes):
  442. .int JMPTBL (L(aligned_16_0bytes), L(table_16_128bytes))
  443. .int JMPTBL (L(aligned_16_2bytes), L(table_16_128bytes))
  444. .int JMPTBL (L(aligned_16_4bytes), L(table_16_128bytes))
  445. .int JMPTBL (L(aligned_16_6bytes), L(table_16_128bytes))
  446. .int JMPTBL (L(aligned_16_8bytes), L(table_16_128bytes))
  447. .int JMPTBL (L(aligned_16_10bytes), L(table_16_128bytes))
  448. .int JMPTBL (L(aligned_16_12bytes), L(table_16_128bytes))
  449. .int JMPTBL (L(aligned_16_14bytes), L(table_16_128bytes))
  450. .int JMPTBL (L(aligned_16_16bytes), L(table_16_128bytes))
  451. .int JMPTBL (L(aligned_16_18bytes), L(table_16_128bytes))
  452. .int JMPTBL (L(aligned_16_20bytes), L(table_16_128bytes))
  453. .int JMPTBL (L(aligned_16_22bytes), L(table_16_128bytes))
  454. .int JMPTBL (L(aligned_16_24bytes), L(table_16_128bytes))
  455. .int JMPTBL (L(aligned_16_26bytes), L(table_16_128bytes))
  456. .int JMPTBL (L(aligned_16_28bytes), L(table_16_128bytes))
  457. .int JMPTBL (L(aligned_16_30bytes), L(table_16_128bytes))
  458. .int JMPTBL (L(aligned_16_32bytes), L(table_16_128bytes))
  459. .int JMPTBL (L(aligned_16_34bytes), L(table_16_128bytes))
  460. .int JMPTBL (L(aligned_16_36bytes), L(table_16_128bytes))
  461. .int JMPTBL (L(aligned_16_38bytes), L(table_16_128bytes))
  462. .int JMPTBL (L(aligned_16_40bytes), L(table_16_128bytes))
  463. .int JMPTBL (L(aligned_16_42bytes), L(table_16_128bytes))
  464. .int JMPTBL (L(aligned_16_44bytes), L(table_16_128bytes))
  465. .int JMPTBL (L(aligned_16_46bytes), L(table_16_128bytes))
  466. .int JMPTBL (L(aligned_16_48bytes), L(table_16_128bytes))
  467. .int JMPTBL (L(aligned_16_50bytes), L(table_16_128bytes))
  468. .int JMPTBL (L(aligned_16_52bytes), L(table_16_128bytes))
  469. .int JMPTBL (L(aligned_16_54bytes), L(table_16_128bytes))
  470. .int JMPTBL (L(aligned_16_56bytes), L(table_16_128bytes))
  471. .int JMPTBL (L(aligned_16_58bytes), L(table_16_128bytes))
  472. .int JMPTBL (L(aligned_16_60bytes), L(table_16_128bytes))
  473. .int JMPTBL (L(aligned_16_62bytes), L(table_16_128bytes))
  474. .int JMPTBL (L(aligned_16_64bytes), L(table_16_128bytes))
  475. .int JMPTBL (L(aligned_16_66bytes), L(table_16_128bytes))
  476. .int JMPTBL (L(aligned_16_68bytes), L(table_16_128bytes))
  477. .int JMPTBL (L(aligned_16_70bytes), L(table_16_128bytes))
  478. .int JMPTBL (L(aligned_16_72bytes), L(table_16_128bytes))
  479. .int JMPTBL (L(aligned_16_74bytes), L(table_16_128bytes))
  480. .int JMPTBL (L(aligned_16_76bytes), L(table_16_128bytes))
  481. .int JMPTBL (L(aligned_16_78bytes), L(table_16_128bytes))
  482. .int JMPTBL (L(aligned_16_80bytes), L(table_16_128bytes))
  483. .int JMPTBL (L(aligned_16_82bytes), L(table_16_128bytes))
  484. .int JMPTBL (L(aligned_16_84bytes), L(table_16_128bytes))
  485. .int JMPTBL (L(aligned_16_86bytes), L(table_16_128bytes))
  486. .int JMPTBL (L(aligned_16_88bytes), L(table_16_128bytes))
  487. .int JMPTBL (L(aligned_16_90bytes), L(table_16_128bytes))
  488. .int JMPTBL (L(aligned_16_92bytes), L(table_16_128bytes))
  489. .int JMPTBL (L(aligned_16_94bytes), L(table_16_128bytes))
  490. .int JMPTBL (L(aligned_16_96bytes), L(table_16_128bytes))
  491. .int JMPTBL (L(aligned_16_98bytes), L(table_16_128bytes))
  492. .int JMPTBL (L(aligned_16_100bytes), L(table_16_128bytes))
  493. .int JMPTBL (L(aligned_16_102bytes), L(table_16_128bytes))
  494. .int JMPTBL (L(aligned_16_104bytes), L(table_16_128bytes))
  495. .int JMPTBL (L(aligned_16_106bytes), L(table_16_128bytes))
  496. .int JMPTBL (L(aligned_16_108bytes), L(table_16_128bytes))
  497. .int JMPTBL (L(aligned_16_110bytes), L(table_16_128bytes))
  498. .int JMPTBL (L(aligned_16_112bytes), L(table_16_128bytes))
  499. .int JMPTBL (L(aligned_16_114bytes), L(table_16_128bytes))
  500. .int JMPTBL (L(aligned_16_116bytes), L(table_16_128bytes))
  501. .int JMPTBL (L(aligned_16_118bytes), L(table_16_128bytes))
  502. .int JMPTBL (L(aligned_16_120bytes), L(table_16_128bytes))
  503. .int JMPTBL (L(aligned_16_122bytes), L(table_16_128bytes))
  504. .int JMPTBL (L(aligned_16_124bytes), L(table_16_128bytes))
  505. .int JMPTBL (L(aligned_16_126bytes), L(table_16_128bytes))
  506. .popsection
  507. ALIGN (4)
  508. L(aligned_16_112bytes):
  509. movdqa %xmm0, -112(%edx)
  510. L(aligned_16_96bytes):
  511. movdqa %xmm0, -96(%edx)
  512. L(aligned_16_80bytes):
  513. movdqa %xmm0, -80(%edx)
  514. L(aligned_16_64bytes):
  515. movdqa %xmm0, -64(%edx)
  516. L(aligned_16_48bytes):
  517. movdqa %xmm0, -48(%edx)
  518. L(aligned_16_32bytes):
  519. movdqa %xmm0, -32(%edx)
  520. L(aligned_16_16bytes):
  521. movdqa %xmm0, -16(%edx)
  522. L(aligned_16_0bytes):
  523. SETRTNVAL
  524. RETURN
  525. ALIGN (4)
  526. L(aligned_16_114bytes):
  527. movdqa %xmm0, -114(%edx)
  528. L(aligned_16_98bytes):
  529. movdqa %xmm0, -98(%edx)
  530. L(aligned_16_82bytes):
  531. movdqa %xmm0, -82(%edx)
  532. L(aligned_16_66bytes):
  533. movdqa %xmm0, -66(%edx)
  534. L(aligned_16_50bytes):
  535. movdqa %xmm0, -50(%edx)
  536. L(aligned_16_34bytes):
  537. movdqa %xmm0, -34(%edx)
  538. L(aligned_16_18bytes):
  539. movdqa %xmm0, -18(%edx)
  540. L(aligned_16_2bytes):
  541. movw %ax, -2(%edx)
  542. SETRTNVAL
  543. RETURN
  544. ALIGN (4)
  545. L(aligned_16_116bytes):
  546. movdqa %xmm0, -116(%edx)
  547. L(aligned_16_100bytes):
  548. movdqa %xmm0, -100(%edx)
  549. L(aligned_16_84bytes):
  550. movdqa %xmm0, -84(%edx)
  551. L(aligned_16_68bytes):
  552. movdqa %xmm0, -68(%edx)
  553. L(aligned_16_52bytes):
  554. movdqa %xmm0, -52(%edx)
  555. L(aligned_16_36bytes):
  556. movdqa %xmm0, -36(%edx)
  557. L(aligned_16_20bytes):
  558. movdqa %xmm0, -20(%edx)
  559. L(aligned_16_4bytes):
  560. movl %eax, -4(%edx)
  561. SETRTNVAL
  562. RETURN
  563. ALIGN (4)
  564. L(aligned_16_118bytes):
  565. movdqa %xmm0, -118(%edx)
  566. L(aligned_16_102bytes):
  567. movdqa %xmm0, -102(%edx)
  568. L(aligned_16_86bytes):
  569. movdqa %xmm0, -86(%edx)
  570. L(aligned_16_70bytes):
  571. movdqa %xmm0, -70(%edx)
  572. L(aligned_16_54bytes):
  573. movdqa %xmm0, -54(%edx)
  574. L(aligned_16_38bytes):
  575. movdqa %xmm0, -38(%edx)
  576. L(aligned_16_22bytes):
  577. movdqa %xmm0, -22(%edx)
  578. L(aligned_16_6bytes):
  579. movl %eax, -6(%edx)
  580. movw %ax, -2(%edx)
  581. SETRTNVAL
  582. RETURN
  583. ALIGN (4)
  584. L(aligned_16_120bytes):
  585. movdqa %xmm0, -120(%edx)
  586. L(aligned_16_104bytes):
  587. movdqa %xmm0, -104(%edx)
  588. L(aligned_16_88bytes):
  589. movdqa %xmm0, -88(%edx)
  590. L(aligned_16_72bytes):
  591. movdqa %xmm0, -72(%edx)
  592. L(aligned_16_56bytes):
  593. movdqa %xmm0, -56(%edx)
  594. L(aligned_16_40bytes):
  595. movdqa %xmm0, -40(%edx)
  596. L(aligned_16_24bytes):
  597. movdqa %xmm0, -24(%edx)
  598. L(aligned_16_8bytes):
  599. movq %xmm0, -8(%edx)
  600. SETRTNVAL
  601. RETURN
  602. ALIGN (4)
  603. L(aligned_16_122bytes):
  604. movdqa %xmm0, -122(%edx)
  605. L(aligned_16_106bytes):
  606. movdqa %xmm0, -106(%edx)
  607. L(aligned_16_90bytes):
  608. movdqa %xmm0, -90(%edx)
  609. L(aligned_16_74bytes):
  610. movdqa %xmm0, -74(%edx)
  611. L(aligned_16_58bytes):
  612. movdqa %xmm0, -58(%edx)
  613. L(aligned_16_42bytes):
  614. movdqa %xmm0, -42(%edx)
  615. L(aligned_16_26bytes):
  616. movdqa %xmm0, -26(%edx)
  617. L(aligned_16_10bytes):
  618. movq %xmm0, -10(%edx)
  619. movw %ax, -2(%edx)
  620. SETRTNVAL
  621. RETURN
  622. ALIGN (4)
  623. L(aligned_16_124bytes):
  624. movdqa %xmm0, -124(%edx)
  625. L(aligned_16_108bytes):
  626. movdqa %xmm0, -108(%edx)
  627. L(aligned_16_92bytes):
  628. movdqa %xmm0, -92(%edx)
  629. L(aligned_16_76bytes):
  630. movdqa %xmm0, -76(%edx)
  631. L(aligned_16_60bytes):
  632. movdqa %xmm0, -60(%edx)
  633. L(aligned_16_44bytes):
  634. movdqa %xmm0, -44(%edx)
  635. L(aligned_16_28bytes):
  636. movdqa %xmm0, -28(%edx)
  637. L(aligned_16_12bytes):
  638. movq %xmm0, -12(%edx)
  639. movl %eax, -4(%edx)
  640. SETRTNVAL
  641. RETURN
  642. ALIGN (4)
  643. L(aligned_16_126bytes):
  644. movdqa %xmm0, -126(%edx)
  645. L(aligned_16_110bytes):
  646. movdqa %xmm0, -110(%edx)
  647. L(aligned_16_94bytes):
  648. movdqa %xmm0, -94(%edx)
  649. L(aligned_16_78bytes):
  650. movdqa %xmm0, -78(%edx)
  651. L(aligned_16_62bytes):
  652. movdqa %xmm0, -62(%edx)
  653. L(aligned_16_46bytes):
  654. movdqa %xmm0, -46(%edx)
  655. L(aligned_16_30bytes):
  656. movdqa %xmm0, -30(%edx)
  657. L(aligned_16_14bytes):
  658. movq %xmm0, -14(%edx)
  659. movl %eax, -6(%edx)
  660. movw %ax, -2(%edx)
  661. SETRTNVAL
  662. RETURN
  663. END (sse2_memset16_atom)