| OLD | NEW |
|---|
| (Empty) |
| 1 // Copyright 2012 the V8 project authors. All rights reserved. | |
| 2 // Use of this source code is governed by a BSD-style license that can be | |
| 3 // found in the LICENSE file. | |
| 4 | |
| 5 #include "src/v8.h" | |
| 6 | |
| 7 #if V8_TARGET_ARCH_IA32 | |
| 8 | |
| 9 #include "src/cpu-profiler.h" | |
| 10 #include "src/log.h" | |
| 11 #include "src/macro-assembler.h" | |
| 12 #include "src/regexp-macro-assembler.h" | |
| 13 #include "src/regexp-stack.h" | |
| 14 #include "src/unicode.h" | |
| 15 | |
| 16 #include "src/ia32/regexp-macro-assembler-ia32.h" | |
| 17 | |
| 18 namespace v8 { | |
| 19 namespace internal { | |
| 20 | |
| 21 #ifndef V8_INTERPRETED_REGEXP | |
| 22 /* | |
| 23 * This assembler uses the following register assignment convention | |
| 24 * - edx : Current character. Must be loaded using LoadCurrentCharacter | |
| 25 * before using any of the dispatch methods. Temporarily stores the | |
| 26 * index of capture start after a matching pass for a global regexp. | |
| 27 * - edi : Current position in input, as negative offset from end of string. | |
| 28 * Please notice that this is the byte offset, not the character offset! | |
| 29 * - esi : end of input (points to byte after last character in input). | |
| 30 * - ebp : Frame pointer. Used to access arguments, local variables and | |
| 31 * RegExp registers. | |
| 32 * - esp : Points to tip of C stack. | |
| 33 * - ecx : Points to tip of backtrack stack | |
| 34 * | |
| 35 * The registers eax and ebx are free to use for computations. | |
| 36 * | |
| 37 * Each call to a public method should retain this convention. | |
| 38 * The stack will have the following structure: | |
| 39 * - Isolate* isolate (address of the current isolate) | |
| 40 * - direct_call (if 1, direct call from JavaScript code, if 0 | |
| 41 * call through the runtime system) | |
| 42 * - stack_area_base (high end of the memory area to use as | |
| 43 * backtracking stack) | |
| 44 * - capture array size (may fit multiple sets of matches) | |
| 45 * - int* capture_array (int[num_saved_registers_], for output). | |
| 46 * - end of input (address of end of string) | |
| 47 * - start of input (address of first character in string) | |
| 48 * - start index (character index of start) | |
| 49 * - String* input_string (location of a handle containing the string) | |
| 50 * --- frame alignment (if applicable) --- | |
| 51 * - return address | |
| 52 * ebp-> - old ebp | |
| 53 * - backup of caller esi | |
| 54 * - backup of caller edi | |
| 55 * - backup of caller ebx | |
| 56 * - success counter (only for global regexps to count matches). | |
| 57 * - Offset of location before start of input (effectively character | |
| 58 * position -1). Used to initialize capture registers to a non-position. | |
| 59 * - register 0 ebp[-4] (only positions must be stored in the first | |
| 60 * - register 1 ebp[-8] num_saved_registers_ registers) | |
| 61 * - ... | |
| 62 * | |
| 63 * The first num_saved_registers_ registers are initialized to point to | |
| 64 * "character -1" in the string (i.e., char_size() bytes before the first | |
| 65 * character of the string). The remaining registers starts out as garbage. | |
| 66 * | |
| 67 * The data up to the return address must be placed there by the calling | |
| 68 * code, by calling the code entry as cast to a function with the signature: | |
| 69 * int (*match)(String* input_string, | |
| 70 * int start_index, | |
| 71 * Address start, | |
| 72 * Address end, | |
| 73 * int* capture_output_array, | |
| 74 * bool at_start, | |
| 75 * byte* stack_area_base, | |
| 76 * bool direct_call) | |
| 77 */ | |
| 78 | |
| 79 #define __ ACCESS_MASM(masm_) | |
| 80 | |
| 81 RegExpMacroAssemblerIA32::RegExpMacroAssemblerIA32(Isolate* isolate, Zone* zone, | |
| 82 Mode mode, | |
| 83 int registers_to_save) | |
| 84 : NativeRegExpMacroAssembler(isolate, zone), | |
| 85 masm_(new MacroAssembler(isolate, NULL, kRegExpCodeSize)), | |
| 86 mode_(mode), | |
| 87 num_registers_(registers_to_save), | |
| 88 num_saved_registers_(registers_to_save), | |
| 89 entry_label_(), | |
| 90 start_label_(), | |
| 91 success_label_(), | |
| 92 backtrack_label_(), | |
| 93 exit_label_() { | |
| 94 DCHECK_EQ(0, registers_to_save % 2); | |
| 95 __ jmp(&entry_label_); // We'll write the entry code later. | |
| 96 __ bind(&start_label_); // And then continue from here. | |
| 97 } | |
| 98 | |
| 99 | |
| 100 RegExpMacroAssemblerIA32::~RegExpMacroAssemblerIA32() { | |
| 101 delete masm_; | |
| 102 // Unuse labels in case we throw away the assembler without calling GetCode. | |
| 103 entry_label_.Unuse(); | |
| 104 start_label_.Unuse(); | |
| 105 success_label_.Unuse(); | |
| 106 backtrack_label_.Unuse(); | |
| 107 exit_label_.Unuse(); | |
| 108 check_preempt_label_.Unuse(); | |
| 109 stack_overflow_label_.Unuse(); | |
| 110 } | |
| 111 | |
| 112 | |
| 113 int RegExpMacroAssemblerIA32::stack_limit_slack() { | |
| 114 return RegExpStack::kStackLimitSlack; | |
| 115 } | |
| 116 | |
| 117 | |
| 118 void RegExpMacroAssemblerIA32::AdvanceCurrentPosition(int by) { | |
| 119 if (by != 0) { | |
| 120 __ add(edi, Immediate(by * char_size())); | |
| 121 } | |
| 122 } | |
| 123 | |
| 124 | |
| 125 void RegExpMacroAssemblerIA32::AdvanceRegister(int reg, int by) { | |
| 126 DCHECK(reg >= 0); | |
| 127 DCHECK(reg < num_registers_); | |
| 128 if (by != 0) { | |
| 129 __ add(register_location(reg), Immediate(by)); | |
| 130 } | |
| 131 } | |
| 132 | |
| 133 | |
| 134 void RegExpMacroAssemblerIA32::Backtrack() { | |
| 135 CheckPreemption(); | |
| 136 // Pop Code* offset from backtrack stack, add Code* and jump to location. | |
| 137 Pop(ebx); | |
| 138 __ add(ebx, Immediate(masm_->CodeObject())); | |
| 139 __ jmp(ebx); | |
| 140 } | |
| 141 | |
| 142 | |
| 143 void RegExpMacroAssemblerIA32::Bind(Label* label) { | |
| 144 __ bind(label); | |
| 145 } | |
| 146 | |
| 147 | |
| 148 void RegExpMacroAssemblerIA32::CheckCharacter(uint32_t c, Label* on_equal) { | |
| 149 __ cmp(current_character(), c); | |
| 150 BranchOrBacktrack(equal, on_equal); | |
| 151 } | |
| 152 | |
| 153 | |
| 154 void RegExpMacroAssemblerIA32::CheckCharacterGT(uc16 limit, Label* on_greater) { | |
| 155 __ cmp(current_character(), limit); | |
| 156 BranchOrBacktrack(greater, on_greater); | |
| 157 } | |
| 158 | |
| 159 | |
| 160 void RegExpMacroAssemblerIA32::CheckAtStart(Label* on_at_start) { | |
| 161 Label not_at_start; | |
| 162 // Did we start the match at the start of the string at all? | |
| 163 __ cmp(Operand(ebp, kStartIndex), Immediate(0)); | |
| 164 BranchOrBacktrack(not_equal, ¬_at_start); | |
| 165 // If we did, are we still at the start of the input? | |
| 166 __ lea(eax, Operand(esi, edi, times_1, 0)); | |
| 167 __ cmp(eax, Operand(ebp, kInputStart)); | |
| 168 BranchOrBacktrack(equal, on_at_start); | |
| 169 __ bind(¬_at_start); | |
| 170 } | |
| 171 | |
| 172 | |
| 173 void RegExpMacroAssemblerIA32::CheckNotAtStart(Label* on_not_at_start) { | |
| 174 // Did we start the match at the start of the string at all? | |
| 175 __ cmp(Operand(ebp, kStartIndex), Immediate(0)); | |
| 176 BranchOrBacktrack(not_equal, on_not_at_start); | |
| 177 // If we did, are we still at the start of the input? | |
| 178 __ lea(eax, Operand(esi, edi, times_1, 0)); | |
| 179 __ cmp(eax, Operand(ebp, kInputStart)); | |
| 180 BranchOrBacktrack(not_equal, on_not_at_start); | |
| 181 } | |
| 182 | |
| 183 | |
| 184 void RegExpMacroAssemblerIA32::CheckCharacterLT(uc16 limit, Label* on_less) { | |
| 185 __ cmp(current_character(), limit); | |
| 186 BranchOrBacktrack(less, on_less); | |
| 187 } | |
| 188 | |
| 189 | |
| 190 void RegExpMacroAssemblerIA32::CheckGreedyLoop(Label* on_equal) { | |
| 191 Label fallthrough; | |
| 192 __ cmp(edi, Operand(backtrack_stackpointer(), 0)); | |
| 193 __ j(not_equal, &fallthrough); | |
| 194 __ add(backtrack_stackpointer(), Immediate(kPointerSize)); // Pop. | |
| 195 BranchOrBacktrack(no_condition, on_equal); | |
| 196 __ bind(&fallthrough); | |
| 197 } | |
| 198 | |
| 199 | |
| 200 void RegExpMacroAssemblerIA32::CheckNotBackReferenceIgnoreCase( | |
| 201 int start_reg, | |
| 202 Label* on_no_match) { | |
| 203 Label fallthrough; | |
| 204 __ mov(edx, register_location(start_reg)); // Index of start of capture | |
| 205 __ mov(ebx, register_location(start_reg + 1)); // Index of end of capture | |
| 206 __ sub(ebx, edx); // Length of capture. | |
| 207 | |
| 208 // The length of a capture should not be negative. This can only happen | |
| 209 // if the end of the capture is unrecorded, or at a point earlier than | |
| 210 // the start of the capture. | |
| 211 BranchOrBacktrack(less, on_no_match); | |
| 212 | |
| 213 // If length is zero, either the capture is empty or it is completely | |
| 214 // uncaptured. In either case succeed immediately. | |
| 215 __ j(equal, &fallthrough); | |
| 216 | |
| 217 // Check that there are sufficient characters left in the input. | |
| 218 __ mov(eax, edi); | |
| 219 __ add(eax, ebx); | |
| 220 BranchOrBacktrack(greater, on_no_match); | |
| 221 | |
| 222 if (mode_ == LATIN1) { | |
| 223 Label success; | |
| 224 Label fail; | |
| 225 Label loop_increment; | |
| 226 // Save register contents to make the registers available below. | |
| 227 __ push(edi); | |
| 228 __ push(backtrack_stackpointer()); | |
| 229 // After this, the eax, ecx, and edi registers are available. | |
| 230 | |
| 231 __ add(edx, esi); // Start of capture | |
| 232 __ add(edi, esi); // Start of text to match against capture. | |
| 233 __ add(ebx, edi); // End of text to match against capture. | |
| 234 | |
| 235 Label loop; | |
| 236 __ bind(&loop); | |
| 237 __ movzx_b(eax, Operand(edi, 0)); | |
| 238 __ cmpb_al(Operand(edx, 0)); | |
| 239 __ j(equal, &loop_increment); | |
| 240 | |
| 241 // Mismatch, try case-insensitive match (converting letters to lower-case). | |
| 242 __ or_(eax, 0x20); // Convert match character to lower-case. | |
| 243 __ lea(ecx, Operand(eax, -'a')); | |
| 244 __ cmp(ecx, static_cast<int32_t>('z' - 'a')); // Is eax a lowercase letter? | |
| 245 Label convert_capture; | |
| 246 __ j(below_equal, &convert_capture); // In range 'a'-'z'. | |
| 247 // Latin-1: Check for values in range [224,254] but not 247. | |
| 248 __ sub(ecx, Immediate(224 - 'a')); | |
| 249 __ cmp(ecx, Immediate(254 - 224)); | |
| 250 __ j(above, &fail); // Weren't Latin-1 letters. | |
| 251 __ cmp(ecx, Immediate(247 - 224)); // Check for 247. | |
| 252 __ j(equal, &fail); | |
| 253 __ bind(&convert_capture); | |
| 254 // Also convert capture character. | |
| 255 __ movzx_b(ecx, Operand(edx, 0)); | |
| 256 __ or_(ecx, 0x20); | |
| 257 | |
| 258 __ cmp(eax, ecx); | |
| 259 __ j(not_equal, &fail); | |
| 260 | |
| 261 __ bind(&loop_increment); | |
| 262 // Increment pointers into match and capture strings. | |
| 263 __ add(edx, Immediate(1)); | |
| 264 __ add(edi, Immediate(1)); | |
| 265 // Compare to end of match, and loop if not done. | |
| 266 __ cmp(edi, ebx); | |
| 267 __ j(below, &loop); | |
| 268 __ jmp(&success); | |
| 269 | |
| 270 __ bind(&fail); | |
| 271 // Restore original values before failing. | |
| 272 __ pop(backtrack_stackpointer()); | |
| 273 __ pop(edi); | |
| 274 BranchOrBacktrack(no_condition, on_no_match); | |
| 275 | |
| 276 __ bind(&success); | |
| 277 // Restore original value before continuing. | |
| 278 __ pop(backtrack_stackpointer()); | |
| 279 // Drop original value of character position. | |
| 280 __ add(esp, Immediate(kPointerSize)); | |
| 281 // Compute new value of character position after the matched part. | |
| 282 __ sub(edi, esi); | |
| 283 } else { | |
| 284 DCHECK(mode_ == UC16); | |
| 285 // Save registers before calling C function. | |
| 286 __ push(esi); | |
| 287 __ push(edi); | |
| 288 __ push(backtrack_stackpointer()); | |
| 289 __ push(ebx); | |
| 290 | |
| 291 static const int argument_count = 4; | |
| 292 __ PrepareCallCFunction(argument_count, ecx); | |
| 293 // Put arguments into allocated stack area, last argument highest on stack. | |
| 294 // Parameters are | |
| 295 // Address byte_offset1 - Address captured substring's start. | |
| 296 // Address byte_offset2 - Address of current character position. | |
| 297 // size_t byte_length - length of capture in bytes(!) | |
| 298 // Isolate* isolate | |
| 299 | |
| 300 // Set isolate. | |
| 301 __ mov(Operand(esp, 3 * kPointerSize), | |
| 302 Immediate(ExternalReference::isolate_address(isolate()))); | |
| 303 // Set byte_length. | |
| 304 __ mov(Operand(esp, 2 * kPointerSize), ebx); | |
| 305 // Set byte_offset2. | |
| 306 // Found by adding negative string-end offset of current position (edi) | |
| 307 // to end of string. | |
| 308 __ add(edi, esi); | |
| 309 __ mov(Operand(esp, 1 * kPointerSize), edi); | |
| 310 // Set byte_offset1. | |
| 311 // Start of capture, where edx already holds string-end negative offset. | |
| 312 __ add(edx, esi); | |
| 313 __ mov(Operand(esp, 0 * kPointerSize), edx); | |
| 314 | |
| 315 { | |
| 316 AllowExternalCallThatCantCauseGC scope(masm_); | |
| 317 ExternalReference compare = | |
| 318 ExternalReference::re_case_insensitive_compare_uc16(isolate()); | |
| 319 __ CallCFunction(compare, argument_count); | |
| 320 } | |
| 321 // Pop original values before reacting on result value. | |
| 322 __ pop(ebx); | |
| 323 __ pop(backtrack_stackpointer()); | |
| 324 __ pop(edi); | |
| 325 __ pop(esi); | |
| 326 | |
| 327 // Check if function returned non-zero for success or zero for failure. | |
| 328 __ or_(eax, eax); | |
| 329 BranchOrBacktrack(zero, on_no_match); | |
| 330 // On success, increment position by length of capture. | |
| 331 __ add(edi, ebx); | |
| 332 } | |
| 333 __ bind(&fallthrough); | |
| 334 } | |
| 335 | |
| 336 | |
| 337 void RegExpMacroAssemblerIA32::CheckNotBackReference( | |
| 338 int start_reg, | |
| 339 Label* on_no_match) { | |
| 340 Label fallthrough; | |
| 341 Label success; | |
| 342 Label fail; | |
| 343 | |
| 344 // Find length of back-referenced capture. | |
| 345 __ mov(edx, register_location(start_reg)); | |
| 346 __ mov(eax, register_location(start_reg + 1)); | |
| 347 __ sub(eax, edx); // Length to check. | |
| 348 // Fail on partial or illegal capture (start of capture after end of capture). | |
| 349 BranchOrBacktrack(less, on_no_match); | |
| 350 // Succeed on empty capture (including no capture) | |
| 351 __ j(equal, &fallthrough); | |
| 352 | |
| 353 // Check that there are sufficient characters left in the input. | |
| 354 __ mov(ebx, edi); | |
| 355 __ add(ebx, eax); | |
| 356 BranchOrBacktrack(greater, on_no_match); | |
| 357 | |
| 358 // Save register to make it available below. | |
| 359 __ push(backtrack_stackpointer()); | |
| 360 | |
| 361 // Compute pointers to match string and capture string | |
| 362 __ lea(ebx, Operand(esi, edi, times_1, 0)); // Start of match. | |
| 363 __ add(edx, esi); // Start of capture. | |
| 364 __ lea(ecx, Operand(eax, ebx, times_1, 0)); // End of match | |
| 365 | |
| 366 Label loop; | |
| 367 __ bind(&loop); | |
| 368 if (mode_ == LATIN1) { | |
| 369 __ movzx_b(eax, Operand(edx, 0)); | |
| 370 __ cmpb_al(Operand(ebx, 0)); | |
| 371 } else { | |
| 372 DCHECK(mode_ == UC16); | |
| 373 __ movzx_w(eax, Operand(edx, 0)); | |
| 374 __ cmpw_ax(Operand(ebx, 0)); | |
| 375 } | |
| 376 __ j(not_equal, &fail); | |
| 377 // Increment pointers into capture and match string. | |
| 378 __ add(edx, Immediate(char_size())); | |
| 379 __ add(ebx, Immediate(char_size())); | |
| 380 // Check if we have reached end of match area. | |
| 381 __ cmp(ebx, ecx); | |
| 382 __ j(below, &loop); | |
| 383 __ jmp(&success); | |
| 384 | |
| 385 __ bind(&fail); | |
| 386 // Restore backtrack stackpointer. | |
| 387 __ pop(backtrack_stackpointer()); | |
| 388 BranchOrBacktrack(no_condition, on_no_match); | |
| 389 | |
| 390 __ bind(&success); | |
| 391 // Move current character position to position after match. | |
| 392 __ mov(edi, ecx); | |
| 393 __ sub(edi, esi); | |
| 394 // Restore backtrack stackpointer. | |
| 395 __ pop(backtrack_stackpointer()); | |
| 396 | |
| 397 __ bind(&fallthrough); | |
| 398 } | |
| 399 | |
| 400 | |
| 401 void RegExpMacroAssemblerIA32::CheckNotCharacter(uint32_t c, | |
| 402 Label* on_not_equal) { | |
| 403 __ cmp(current_character(), c); | |
| 404 BranchOrBacktrack(not_equal, on_not_equal); | |
| 405 } | |
| 406 | |
| 407 | |
| 408 void RegExpMacroAssemblerIA32::CheckCharacterAfterAnd(uint32_t c, | |
| 409 uint32_t mask, | |
| 410 Label* on_equal) { | |
| 411 if (c == 0) { | |
| 412 __ test(current_character(), Immediate(mask)); | |
| 413 } else { | |
| 414 __ mov(eax, mask); | |
| 415 __ and_(eax, current_character()); | |
| 416 __ cmp(eax, c); | |
| 417 } | |
| 418 BranchOrBacktrack(equal, on_equal); | |
| 419 } | |
| 420 | |
| 421 | |
| 422 void RegExpMacroAssemblerIA32::CheckNotCharacterAfterAnd(uint32_t c, | |
| 423 uint32_t mask, | |
| 424 Label* on_not_equal) { | |
| 425 if (c == 0) { | |
| 426 __ test(current_character(), Immediate(mask)); | |
| 427 } else { | |
| 428 __ mov(eax, mask); | |
| 429 __ and_(eax, current_character()); | |
| 430 __ cmp(eax, c); | |
| 431 } | |
| 432 BranchOrBacktrack(not_equal, on_not_equal); | |
| 433 } | |
| 434 | |
| 435 | |
| 436 void RegExpMacroAssemblerIA32::CheckNotCharacterAfterMinusAnd( | |
| 437 uc16 c, | |
| 438 uc16 minus, | |
| 439 uc16 mask, | |
| 440 Label* on_not_equal) { | |
| 441 DCHECK(minus < String::kMaxUtf16CodeUnit); | |
| 442 __ lea(eax, Operand(current_character(), -minus)); | |
| 443 if (c == 0) { | |
| 444 __ test(eax, Immediate(mask)); | |
| 445 } else { | |
| 446 __ and_(eax, mask); | |
| 447 __ cmp(eax, c); | |
| 448 } | |
| 449 BranchOrBacktrack(not_equal, on_not_equal); | |
| 450 } | |
| 451 | |
| 452 | |
| 453 void RegExpMacroAssemblerIA32::CheckCharacterInRange( | |
| 454 uc16 from, | |
| 455 uc16 to, | |
| 456 Label* on_in_range) { | |
| 457 __ lea(eax, Operand(current_character(), -from)); | |
| 458 __ cmp(eax, to - from); | |
| 459 BranchOrBacktrack(below_equal, on_in_range); | |
| 460 } | |
| 461 | |
| 462 | |
| 463 void RegExpMacroAssemblerIA32::CheckCharacterNotInRange( | |
| 464 uc16 from, | |
| 465 uc16 to, | |
| 466 Label* on_not_in_range) { | |
| 467 __ lea(eax, Operand(current_character(), -from)); | |
| 468 __ cmp(eax, to - from); | |
| 469 BranchOrBacktrack(above, on_not_in_range); | |
| 470 } | |
| 471 | |
| 472 | |
| 473 void RegExpMacroAssemblerIA32::CheckBitInTable( | |
| 474 Handle<ByteArray> table, | |
| 475 Label* on_bit_set) { | |
| 476 __ mov(eax, Immediate(table)); | |
| 477 Register index = current_character(); | |
| 478 if (mode_ != LATIN1 || kTableMask != String::kMaxOneByteCharCode) { | |
| 479 __ mov(ebx, kTableSize - 1); | |
| 480 __ and_(ebx, current_character()); | |
| 481 index = ebx; | |
| 482 } | |
| 483 __ cmpb(FieldOperand(eax, index, times_1, ByteArray::kHeaderSize), 0); | |
| 484 BranchOrBacktrack(not_equal, on_bit_set); | |
| 485 } | |
| 486 | |
| 487 | |
| 488 bool RegExpMacroAssemblerIA32::CheckSpecialCharacterClass(uc16 type, | |
| 489 Label* on_no_match) { | |
| 490 // Range checks (c in min..max) are generally implemented by an unsigned | |
| 491 // (c - min) <= (max - min) check | |
| 492 switch (type) { | |
| 493 case 's': | |
| 494 // Match space-characters | |
| 495 if (mode_ == LATIN1) { | |
| 496 // One byte space characters are '\t'..'\r', ' ' and \u00a0. | |
| 497 Label success; | |
| 498 __ cmp(current_character(), ' '); | |
| 499 __ j(equal, &success, Label::kNear); | |
| 500 // Check range 0x09..0x0d | |
| 501 __ lea(eax, Operand(current_character(), -'\t')); | |
| 502 __ cmp(eax, '\r' - '\t'); | |
| 503 __ j(below_equal, &success, Label::kNear); | |
| 504 // \u00a0 (NBSP). | |
| 505 __ cmp(eax, 0x00a0 - '\t'); | |
| 506 BranchOrBacktrack(not_equal, on_no_match); | |
| 507 __ bind(&success); | |
| 508 return true; | |
| 509 } | |
| 510 return false; | |
| 511 case 'S': | |
| 512 // The emitted code for generic character classes is good enough. | |
| 513 return false; | |
| 514 case 'd': | |
| 515 // Match ASCII digits ('0'..'9') | |
| 516 __ lea(eax, Operand(current_character(), -'0')); | |
| 517 __ cmp(eax, '9' - '0'); | |
| 518 BranchOrBacktrack(above, on_no_match); | |
| 519 return true; | |
| 520 case 'D': | |
| 521 // Match non ASCII-digits | |
| 522 __ lea(eax, Operand(current_character(), -'0')); | |
| 523 __ cmp(eax, '9' - '0'); | |
| 524 BranchOrBacktrack(below_equal, on_no_match); | |
| 525 return true; | |
| 526 case '.': { | |
| 527 // Match non-newlines (not 0x0a('\n'), 0x0d('\r'), 0x2028 and 0x2029) | |
| 528 __ mov(eax, current_character()); | |
| 529 __ xor_(eax, Immediate(0x01)); | |
| 530 // See if current character is '\n'^1 or '\r'^1, i.e., 0x0b or 0x0c | |
| 531 __ sub(eax, Immediate(0x0b)); | |
| 532 __ cmp(eax, 0x0c - 0x0b); | |
| 533 BranchOrBacktrack(below_equal, on_no_match); | |
| 534 if (mode_ == UC16) { | |
| 535 // Compare original value to 0x2028 and 0x2029, using the already | |
| 536 // computed (current_char ^ 0x01 - 0x0b). I.e., check for | |
| 537 // 0x201d (0x2028 - 0x0b) or 0x201e. | |
| 538 __ sub(eax, Immediate(0x2028 - 0x0b)); | |
| 539 __ cmp(eax, 0x2029 - 0x2028); | |
| 540 BranchOrBacktrack(below_equal, on_no_match); | |
| 541 } | |
| 542 return true; | |
| 543 } | |
| 544 case 'w': { | |
| 545 if (mode_ != LATIN1) { | |
| 546 // Table is 256 entries, so all Latin1 characters can be tested. | |
| 547 __ cmp(current_character(), Immediate('z')); | |
| 548 BranchOrBacktrack(above, on_no_match); | |
| 549 } | |
| 550 DCHECK_EQ(0, word_character_map[0]); // Character '\0' is not a word char. | |
| 551 ExternalReference word_map = ExternalReference::re_word_character_map(); | |
| 552 __ test_b(current_character(), | |
| 553 Operand::StaticArray(current_character(), times_1, word_map)); | |
| 554 BranchOrBacktrack(zero, on_no_match); | |
| 555 return true; | |
| 556 } | |
| 557 case 'W': { | |
| 558 Label done; | |
| 559 if (mode_ != LATIN1) { | |
| 560 // Table is 256 entries, so all Latin1 characters can be tested. | |
| 561 __ cmp(current_character(), Immediate('z')); | |
| 562 __ j(above, &done); | |
| 563 } | |
| 564 DCHECK_EQ(0, word_character_map[0]); // Character '\0' is not a word char. | |
| 565 ExternalReference word_map = ExternalReference::re_word_character_map(); | |
| 566 __ test_b(current_character(), | |
| 567 Operand::StaticArray(current_character(), times_1, word_map)); | |
| 568 BranchOrBacktrack(not_zero, on_no_match); | |
| 569 if (mode_ != LATIN1) { | |
| 570 __ bind(&done); | |
| 571 } | |
| 572 return true; | |
| 573 } | |
| 574 // Non-standard classes (with no syntactic shorthand) used internally. | |
| 575 case '*': | |
| 576 // Match any character. | |
| 577 return true; | |
| 578 case 'n': { | |
| 579 // Match newlines (0x0a('\n'), 0x0d('\r'), 0x2028 or 0x2029). | |
| 580 // The opposite of '.'. | |
| 581 __ mov(eax, current_character()); | |
| 582 __ xor_(eax, Immediate(0x01)); | |
| 583 // See if current character is '\n'^1 or '\r'^1, i.e., 0x0b or 0x0c | |
| 584 __ sub(eax, Immediate(0x0b)); | |
| 585 __ cmp(eax, 0x0c - 0x0b); | |
| 586 if (mode_ == LATIN1) { | |
| 587 BranchOrBacktrack(above, on_no_match); | |
| 588 } else { | |
| 589 Label done; | |
| 590 BranchOrBacktrack(below_equal, &done); | |
| 591 DCHECK_EQ(UC16, mode_); | |
| 592 // Compare original value to 0x2028 and 0x2029, using the already | |
| 593 // computed (current_char ^ 0x01 - 0x0b). I.e., check for | |
| 594 // 0x201d (0x2028 - 0x0b) or 0x201e. | |
| 595 __ sub(eax, Immediate(0x2028 - 0x0b)); | |
| 596 __ cmp(eax, 1); | |
| 597 BranchOrBacktrack(above, on_no_match); | |
| 598 __ bind(&done); | |
| 599 } | |
| 600 return true; | |
| 601 } | |
| 602 // No custom implementation (yet): s(UC16), S(UC16). | |
| 603 default: | |
| 604 return false; | |
| 605 } | |
| 606 } | |
| 607 | |
| 608 | |
| 609 void RegExpMacroAssemblerIA32::Fail() { | |
| 610 STATIC_ASSERT(FAILURE == 0); // Return value for failure is zero. | |
| 611 if (!global()) { | |
| 612 __ Move(eax, Immediate(FAILURE)); | |
| 613 } | |
| 614 __ jmp(&exit_label_); | |
| 615 } | |
| 616 | |
| 617 | |
| 618 Handle<HeapObject> RegExpMacroAssemblerIA32::GetCode(Handle<String> source) { | |
| 619 Label return_eax; | |
| 620 // Finalize code - write the entry point code now we know how many | |
| 621 // registers we need. | |
| 622 | |
| 623 // Entry code: | |
| 624 __ bind(&entry_label_); | |
| 625 | |
| 626 // Tell the system that we have a stack frame. Because the type is MANUAL, no | |
| 627 // code is generated. | |
| 628 FrameScope scope(masm_, StackFrame::MANUAL); | |
| 629 | |
| 630 // Actually emit code to start a new stack frame. | |
| 631 __ push(ebp); | |
| 632 __ mov(ebp, esp); | |
| 633 // Save callee-save registers. Order here should correspond to order of | |
| 634 // kBackup_ebx etc. | |
| 635 __ push(esi); | |
| 636 __ push(edi); | |
| 637 __ push(ebx); // Callee-save on MacOS. | |
| 638 __ push(Immediate(0)); // Number of successful matches in a global regexp. | |
| 639 __ push(Immediate(0)); // Make room for "input start - 1" constant. | |
| 640 | |
| 641 // Check if we have space on the stack for registers. | |
| 642 Label stack_limit_hit; | |
| 643 Label stack_ok; | |
| 644 | |
| 645 ExternalReference stack_limit = | |
| 646 ExternalReference::address_of_stack_limit(isolate()); | |
| 647 __ mov(ecx, esp); | |
| 648 __ sub(ecx, Operand::StaticVariable(stack_limit)); | |
| 649 // Handle it if the stack pointer is already below the stack limit. | |
| 650 __ j(below_equal, &stack_limit_hit); | |
| 651 // Check if there is room for the variable number of registers above | |
| 652 // the stack limit. | |
| 653 __ cmp(ecx, num_registers_ * kPointerSize); | |
| 654 __ j(above_equal, &stack_ok); | |
| 655 // Exit with OutOfMemory exception. There is not enough space on the stack | |
| 656 // for our working registers. | |
| 657 __ mov(eax, EXCEPTION); | |
| 658 __ jmp(&return_eax); | |
| 659 | |
| 660 __ bind(&stack_limit_hit); | |
| 661 CallCheckStackGuardState(ebx); | |
| 662 __ or_(eax, eax); | |
| 663 // If returned value is non-zero, we exit with the returned value as result. | |
| 664 __ j(not_zero, &return_eax); | |
| 665 | |
| 666 __ bind(&stack_ok); | |
| 667 // Load start index for later use. | |
| 668 __ mov(ebx, Operand(ebp, kStartIndex)); | |
| 669 | |
| 670 // Allocate space on stack for registers. | |
| 671 __ sub(esp, Immediate(num_registers_ * kPointerSize)); | |
| 672 // Load string length. | |
| 673 __ mov(esi, Operand(ebp, kInputEnd)); | |
| 674 // Load input position. | |
| 675 __ mov(edi, Operand(ebp, kInputStart)); | |
| 676 // Set up edi to be negative offset from string end. | |
| 677 __ sub(edi, esi); | |
| 678 | |
| 679 // Set eax to address of char before start of the string. | |
| 680 // (effectively string position -1). | |
| 681 __ neg(ebx); | |
| 682 if (mode_ == UC16) { | |
| 683 __ lea(eax, Operand(edi, ebx, times_2, -char_size())); | |
| 684 } else { | |
| 685 __ lea(eax, Operand(edi, ebx, times_1, -char_size())); | |
| 686 } | |
| 687 // Store this value in a local variable, for use when clearing | |
| 688 // position registers. | |
| 689 __ mov(Operand(ebp, kInputStartMinusOne), eax); | |
| 690 | |
| 691 #if V8_OS_WIN | |
| 692 // Ensure that we write to each stack page, in order. Skipping a page | |
| 693 // on Windows can cause segmentation faults. Assuming page size is 4k. | |
| 694 const int kPageSize = 4096; | |
| 695 const int kRegistersPerPage = kPageSize / kPointerSize; | |
| 696 for (int i = num_saved_registers_ + kRegistersPerPage - 1; | |
| 697 i < num_registers_; | |
| 698 i += kRegistersPerPage) { | |
| 699 __ mov(register_location(i), eax); // One write every page. | |
| 700 } | |
| 701 #endif // V8_OS_WIN | |
| 702 | |
| 703 Label load_char_start_regexp, start_regexp; | |
| 704 // Load newline if index is at start, previous character otherwise. | |
| 705 __ cmp(Operand(ebp, kStartIndex), Immediate(0)); | |
| 706 __ j(not_equal, &load_char_start_regexp, Label::kNear); | |
| 707 __ mov(current_character(), '\n'); | |
| 708 __ jmp(&start_regexp, Label::kNear); | |
| 709 | |
| 710 // Global regexp restarts matching here. | |
| 711 __ bind(&load_char_start_regexp); | |
| 712 // Load previous char as initial value of current character register. | |
| 713 LoadCurrentCharacterUnchecked(-1, 1); | |
| 714 __ bind(&start_regexp); | |
| 715 | |
| 716 // Initialize on-stack registers. | |
| 717 if (num_saved_registers_ > 0) { // Always is, if generated from a regexp. | |
| 718 // Fill saved registers with initial value = start offset - 1 | |
| 719 // Fill in stack push order, to avoid accessing across an unwritten | |
| 720 // page (a problem on Windows). | |
| 721 if (num_saved_registers_ > 8) { | |
| 722 __ mov(ecx, kRegisterZero); | |
| 723 Label init_loop; | |
| 724 __ bind(&init_loop); | |
| 725 __ mov(Operand(ebp, ecx, times_1, 0), eax); | |
| 726 __ sub(ecx, Immediate(kPointerSize)); | |
| 727 __ cmp(ecx, kRegisterZero - num_saved_registers_ * kPointerSize); | |
| 728 __ j(greater, &init_loop); | |
| 729 } else { // Unroll the loop. | |
| 730 for (int i = 0; i < num_saved_registers_; i++) { | |
| 731 __ mov(register_location(i), eax); | |
| 732 } | |
| 733 } | |
| 734 } | |
| 735 | |
| 736 // Initialize backtrack stack pointer. | |
| 737 __ mov(backtrack_stackpointer(), Operand(ebp, kStackHighEnd)); | |
| 738 | |
| 739 __ jmp(&start_label_); | |
| 740 | |
| 741 // Exit code: | |
| 742 if (success_label_.is_linked()) { | |
| 743 // Save captures when successful. | |
| 744 __ bind(&success_label_); | |
| 745 if (num_saved_registers_ > 0) { | |
| 746 // copy captures to output | |
| 747 __ mov(ebx, Operand(ebp, kRegisterOutput)); | |
| 748 __ mov(ecx, Operand(ebp, kInputEnd)); | |
| 749 __ mov(edx, Operand(ebp, kStartIndex)); | |
| 750 __ sub(ecx, Operand(ebp, kInputStart)); | |
| 751 if (mode_ == UC16) { | |
| 752 __ lea(ecx, Operand(ecx, edx, times_2, 0)); | |
| 753 } else { | |
| 754 __ add(ecx, edx); | |
| 755 } | |
| 756 for (int i = 0; i < num_saved_registers_; i++) { | |
| 757 __ mov(eax, register_location(i)); | |
| 758 if (i == 0 && global_with_zero_length_check()) { | |
| 759 // Keep capture start in edx for the zero-length check later. | |
| 760 __ mov(edx, eax); | |
| 761 } | |
| 762 // Convert to index from start of string, not end. | |
| 763 __ add(eax, ecx); | |
| 764 if (mode_ == UC16) { | |
| 765 __ sar(eax, 1); // Convert byte index to character index. | |
| 766 } | |
| 767 __ mov(Operand(ebx, i * kPointerSize), eax); | |
| 768 } | |
| 769 } | |
| 770 | |
| 771 if (global()) { | |
| 772 // Restart matching if the regular expression is flagged as global. | |
| 773 // Increment success counter. | |
| 774 __ inc(Operand(ebp, kSuccessfulCaptures)); | |
| 775 // Capture results have been stored, so the number of remaining global | |
| 776 // output registers is reduced by the number of stored captures. | |
| 777 __ mov(ecx, Operand(ebp, kNumOutputRegisters)); | |
| 778 __ sub(ecx, Immediate(num_saved_registers_)); | |
| 779 // Check whether we have enough room for another set of capture results. | |
| 780 __ cmp(ecx, Immediate(num_saved_registers_)); | |
| 781 __ j(less, &exit_label_); | |
| 782 | |
| 783 __ mov(Operand(ebp, kNumOutputRegisters), ecx); | |
| 784 // Advance the location for output. | |
| 785 __ add(Operand(ebp, kRegisterOutput), | |
| 786 Immediate(num_saved_registers_ * kPointerSize)); | |
| 787 | |
| 788 // Prepare eax to initialize registers with its value in the next run. | |
| 789 __ mov(eax, Operand(ebp, kInputStartMinusOne)); | |
| 790 | |
| 791 if (global_with_zero_length_check()) { | |
| 792 // Special case for zero-length matches. | |
| 793 // edx: capture start index | |
| 794 __ cmp(edi, edx); | |
| 795 // Not a zero-length match, restart. | |
| 796 __ j(not_equal, &load_char_start_regexp); | |
| 797 // edi (offset from the end) is zero if we already reached the end. | |
| 798 __ test(edi, edi); | |
| 799 __ j(zero, &exit_label_, Label::kNear); | |
| 800 // Advance current position after a zero-length match. | |
| 801 if (mode_ == UC16) { | |
| 802 __ add(edi, Immediate(2)); | |
| 803 } else { | |
| 804 __ inc(edi); | |
| 805 } | |
| 806 } | |
| 807 | |
| 808 __ jmp(&load_char_start_regexp); | |
| 809 } else { | |
| 810 __ mov(eax, Immediate(SUCCESS)); | |
| 811 } | |
| 812 } | |
| 813 | |
| 814 __ bind(&exit_label_); | |
| 815 if (global()) { | |
| 816 // Return the number of successful captures. | |
| 817 __ mov(eax, Operand(ebp, kSuccessfulCaptures)); | |
| 818 } | |
| 819 | |
| 820 __ bind(&return_eax); | |
| 821 // Skip esp past regexp registers. | |
| 822 __ lea(esp, Operand(ebp, kBackup_ebx)); | |
| 823 // Restore callee-save registers. | |
| 824 __ pop(ebx); | |
| 825 __ pop(edi); | |
| 826 __ pop(esi); | |
| 827 // Exit function frame, restore previous one. | |
| 828 __ pop(ebp); | |
| 829 __ ret(0); | |
| 830 | |
| 831 // Backtrack code (branch target for conditional backtracks). | |
| 832 if (backtrack_label_.is_linked()) { | |
| 833 __ bind(&backtrack_label_); | |
| 834 Backtrack(); | |
| 835 } | |
| 836 | |
| 837 Label exit_with_exception; | |
| 838 | |
| 839 // Preempt-code | |
| 840 if (check_preempt_label_.is_linked()) { | |
| 841 SafeCallTarget(&check_preempt_label_); | |
| 842 | |
| 843 __ push(backtrack_stackpointer()); | |
| 844 __ push(edi); | |
| 845 | |
| 846 CallCheckStackGuardState(ebx); | |
| 847 __ or_(eax, eax); | |
| 848 // If returning non-zero, we should end execution with the given | |
| 849 // result as return value. | |
| 850 __ j(not_zero, &return_eax); | |
| 851 | |
| 852 __ pop(edi); | |
| 853 __ pop(backtrack_stackpointer()); | |
| 854 // String might have moved: Reload esi from frame. | |
| 855 __ mov(esi, Operand(ebp, kInputEnd)); | |
| 856 SafeReturn(); | |
| 857 } | |
| 858 | |
| 859 // Backtrack stack overflow code. | |
| 860 if (stack_overflow_label_.is_linked()) { | |
| 861 SafeCallTarget(&stack_overflow_label_); | |
| 862 // Reached if the backtrack-stack limit has been hit. | |
| 863 | |
| 864 Label grow_failed; | |
| 865 // Save registers before calling C function | |
| 866 __ push(esi); | |
| 867 __ push(edi); | |
| 868 | |
| 869 // Call GrowStack(backtrack_stackpointer()) | |
| 870 static const int num_arguments = 3; | |
| 871 __ PrepareCallCFunction(num_arguments, ebx); | |
| 872 __ mov(Operand(esp, 2 * kPointerSize), | |
| 873 Immediate(ExternalReference::isolate_address(isolate()))); | |
| 874 __ lea(eax, Operand(ebp, kStackHighEnd)); | |
| 875 __ mov(Operand(esp, 1 * kPointerSize), eax); | |
| 876 __ mov(Operand(esp, 0 * kPointerSize), backtrack_stackpointer()); | |
| 877 ExternalReference grow_stack = | |
| 878 ExternalReference::re_grow_stack(isolate()); | |
| 879 __ CallCFunction(grow_stack, num_arguments); | |
| 880 // If return NULL, we have failed to grow the stack, and | |
| 881 // must exit with a stack-overflow exception. | |
| 882 __ or_(eax, eax); | |
| 883 __ j(equal, &exit_with_exception); | |
| 884 // Otherwise use return value as new stack pointer. | |
| 885 __ mov(backtrack_stackpointer(), eax); | |
| 886 // Restore saved registers and continue. | |
| 887 __ pop(edi); | |
| 888 __ pop(esi); | |
| 889 SafeReturn(); | |
| 890 } | |
| 891 | |
| 892 if (exit_with_exception.is_linked()) { | |
| 893 // If any of the code above needed to exit with an exception. | |
| 894 __ bind(&exit_with_exception); | |
| 895 // Exit with Result EXCEPTION(-1) to signal thrown exception. | |
| 896 __ mov(eax, EXCEPTION); | |
| 897 __ jmp(&return_eax); | |
| 898 } | |
| 899 | |
| 900 CodeDesc code_desc; | |
| 901 masm_->GetCode(&code_desc); | |
| 902 Handle<Code> code = | |
| 903 isolate()->factory()->NewCode(code_desc, | |
| 904 Code::ComputeFlags(Code::REGEXP), | |
| 905 masm_->CodeObject()); | |
| 906 PROFILE(isolate(), RegExpCodeCreateEvent(*code, *source)); | |
| 907 return Handle<HeapObject>::cast(code); | |
| 908 } | |
| 909 | |
| 910 | |
| 911 void RegExpMacroAssemblerIA32::GoTo(Label* to) { | |
| 912 BranchOrBacktrack(no_condition, to); | |
| 913 } | |
| 914 | |
| 915 | |
| 916 void RegExpMacroAssemblerIA32::IfRegisterGE(int reg, | |
| 917 int comparand, | |
| 918 Label* if_ge) { | |
| 919 __ cmp(register_location(reg), Immediate(comparand)); | |
| 920 BranchOrBacktrack(greater_equal, if_ge); | |
| 921 } | |
| 922 | |
| 923 | |
| 924 void RegExpMacroAssemblerIA32::IfRegisterLT(int reg, | |
| 925 int comparand, | |
| 926 Label* if_lt) { | |
| 927 __ cmp(register_location(reg), Immediate(comparand)); | |
| 928 BranchOrBacktrack(less, if_lt); | |
| 929 } | |
| 930 | |
| 931 | |
| 932 void RegExpMacroAssemblerIA32::IfRegisterEqPos(int reg, | |
| 933 Label* if_eq) { | |
| 934 __ cmp(edi, register_location(reg)); | |
| 935 BranchOrBacktrack(equal, if_eq); | |
| 936 } | |
| 937 | |
| 938 | |
| 939 RegExpMacroAssembler::IrregexpImplementation | |
| 940 RegExpMacroAssemblerIA32::Implementation() { | |
| 941 return kIA32Implementation; | |
| 942 } | |
| 943 | |
| 944 | |
| 945 void RegExpMacroAssemblerIA32::LoadCurrentCharacter(int cp_offset, | |
| 946 Label* on_end_of_input, | |
| 947 bool check_bounds, | |
| 948 int characters) { | |
| 949 DCHECK(cp_offset >= -1); // ^ and \b can look behind one character. | |
| 950 DCHECK(cp_offset < (1<<30)); // Be sane! (And ensure negation works) | |
| 951 if (check_bounds) { | |
| 952 CheckPosition(cp_offset + characters - 1, on_end_of_input); | |
| 953 } | |
| 954 LoadCurrentCharacterUnchecked(cp_offset, characters); | |
| 955 } | |
| 956 | |
| 957 | |
| 958 void RegExpMacroAssemblerIA32::PopCurrentPosition() { | |
| 959 Pop(edi); | |
| 960 } | |
| 961 | |
| 962 | |
| 963 void RegExpMacroAssemblerIA32::PopRegister(int register_index) { | |
| 964 Pop(eax); | |
| 965 __ mov(register_location(register_index), eax); | |
| 966 } | |
| 967 | |
| 968 | |
| 969 void RegExpMacroAssemblerIA32::PushBacktrack(Label* label) { | |
| 970 Push(Immediate::CodeRelativeOffset(label)); | |
| 971 CheckStackLimit(); | |
| 972 } | |
| 973 | |
| 974 | |
| 975 void RegExpMacroAssemblerIA32::PushCurrentPosition() { | |
| 976 Push(edi); | |
| 977 } | |
| 978 | |
| 979 | |
| 980 void RegExpMacroAssemblerIA32::PushRegister(int register_index, | |
| 981 StackCheckFlag check_stack_limit) { | |
| 982 __ mov(eax, register_location(register_index)); | |
| 983 Push(eax); | |
| 984 if (check_stack_limit) CheckStackLimit(); | |
| 985 } | |
| 986 | |
| 987 | |
| 988 void RegExpMacroAssemblerIA32::ReadCurrentPositionFromRegister(int reg) { | |
| 989 __ mov(edi, register_location(reg)); | |
| 990 } | |
| 991 | |
| 992 | |
| 993 void RegExpMacroAssemblerIA32::ReadStackPointerFromRegister(int reg) { | |
| 994 __ mov(backtrack_stackpointer(), register_location(reg)); | |
| 995 __ add(backtrack_stackpointer(), Operand(ebp, kStackHighEnd)); | |
| 996 } | |
| 997 | |
| 998 void RegExpMacroAssemblerIA32::SetCurrentPositionFromEnd(int by) { | |
| 999 Label after_position; | |
| 1000 __ cmp(edi, -by * char_size()); | |
| 1001 __ j(greater_equal, &after_position, Label::kNear); | |
| 1002 __ mov(edi, -by * char_size()); | |
| 1003 // On RegExp code entry (where this operation is used), the character before | |
| 1004 // the current position is expected to be already loaded. | |
| 1005 // We have advanced the position, so it's safe to read backwards. | |
| 1006 LoadCurrentCharacterUnchecked(-1, 1); | |
| 1007 __ bind(&after_position); | |
| 1008 } | |
| 1009 | |
| 1010 | |
| 1011 void RegExpMacroAssemblerIA32::SetRegister(int register_index, int to) { | |
| 1012 DCHECK(register_index >= num_saved_registers_); // Reserved for positions! | |
| 1013 __ mov(register_location(register_index), Immediate(to)); | |
| 1014 } | |
| 1015 | |
| 1016 | |
| 1017 bool RegExpMacroAssemblerIA32::Succeed() { | |
| 1018 __ jmp(&success_label_); | |
| 1019 return global(); | |
| 1020 } | |
| 1021 | |
| 1022 | |
| 1023 void RegExpMacroAssemblerIA32::WriteCurrentPositionToRegister(int reg, | |
| 1024 int cp_offset) { | |
| 1025 if (cp_offset == 0) { | |
| 1026 __ mov(register_location(reg), edi); | |
| 1027 } else { | |
| 1028 __ lea(eax, Operand(edi, cp_offset * char_size())); | |
| 1029 __ mov(register_location(reg), eax); | |
| 1030 } | |
| 1031 } | |
| 1032 | |
| 1033 | |
| 1034 void RegExpMacroAssemblerIA32::ClearRegisters(int reg_from, int reg_to) { | |
| 1035 DCHECK(reg_from <= reg_to); | |
| 1036 __ mov(eax, Operand(ebp, kInputStartMinusOne)); | |
| 1037 for (int reg = reg_from; reg <= reg_to; reg++) { | |
| 1038 __ mov(register_location(reg), eax); | |
| 1039 } | |
| 1040 } | |
| 1041 | |
| 1042 | |
| 1043 void RegExpMacroAssemblerIA32::WriteStackPointerToRegister(int reg) { | |
| 1044 __ mov(eax, backtrack_stackpointer()); | |
| 1045 __ sub(eax, Operand(ebp, kStackHighEnd)); | |
| 1046 __ mov(register_location(reg), eax); | |
| 1047 } | |
| 1048 | |
| 1049 | |
| 1050 // Private methods: | |
| 1051 | |
| 1052 void RegExpMacroAssemblerIA32::CallCheckStackGuardState(Register scratch) { | |
| 1053 static const int num_arguments = 3; | |
| 1054 __ PrepareCallCFunction(num_arguments, scratch); | |
| 1055 // RegExp code frame pointer. | |
| 1056 __ mov(Operand(esp, 2 * kPointerSize), ebp); | |
| 1057 // Code* of self. | |
| 1058 __ mov(Operand(esp, 1 * kPointerSize), Immediate(masm_->CodeObject())); | |
| 1059 // Next address on the stack (will be address of return address). | |
| 1060 __ lea(eax, Operand(esp, -kPointerSize)); | |
| 1061 __ mov(Operand(esp, 0 * kPointerSize), eax); | |
| 1062 ExternalReference check_stack_guard = | |
| 1063 ExternalReference::re_check_stack_guard_state(isolate()); | |
| 1064 __ CallCFunction(check_stack_guard, num_arguments); | |
| 1065 } | |
| 1066 | |
| 1067 | |
| 1068 // Helper function for reading a value out of a stack frame. | |
| 1069 template <typename T> | |
| 1070 static T& frame_entry(Address re_frame, int frame_offset) { | |
| 1071 return reinterpret_cast<T&>(Memory::int32_at(re_frame + frame_offset)); | |
| 1072 } | |
| 1073 | |
| 1074 | |
| 1075 template <typename T> | |
| 1076 static T* frame_entry_address(Address re_frame, int frame_offset) { | |
| 1077 return reinterpret_cast<T*>(re_frame + frame_offset); | |
| 1078 } | |
| 1079 | |
| 1080 | |
| 1081 int RegExpMacroAssemblerIA32::CheckStackGuardState(Address* return_address, | |
| 1082 Code* re_code, | |
| 1083 Address re_frame) { | |
| 1084 return NativeRegExpMacroAssembler::CheckStackGuardState( | |
| 1085 frame_entry<Isolate*>(re_frame, kIsolate), | |
| 1086 frame_entry<int>(re_frame, kStartIndex), | |
| 1087 frame_entry<int>(re_frame, kDirectCall) == 1, return_address, re_code, | |
| 1088 frame_entry_address<String*>(re_frame, kInputString), | |
| 1089 frame_entry_address<const byte*>(re_frame, kInputStart), | |
| 1090 frame_entry_address<const byte*>(re_frame, kInputEnd)); | |
| 1091 } | |
| 1092 | |
| 1093 | |
| 1094 Operand RegExpMacroAssemblerIA32::register_location(int register_index) { | |
| 1095 DCHECK(register_index < (1<<30)); | |
| 1096 if (num_registers_ <= register_index) { | |
| 1097 num_registers_ = register_index + 1; | |
| 1098 } | |
| 1099 return Operand(ebp, kRegisterZero - register_index * kPointerSize); | |
| 1100 } | |
| 1101 | |
| 1102 | |
| 1103 void RegExpMacroAssemblerIA32::CheckPosition(int cp_offset, | |
| 1104 Label* on_outside_input) { | |
| 1105 __ cmp(edi, -cp_offset * char_size()); | |
| 1106 BranchOrBacktrack(greater_equal, on_outside_input); | |
| 1107 } | |
| 1108 | |
| 1109 | |
| 1110 void RegExpMacroAssemblerIA32::BranchOrBacktrack(Condition condition, | |
| 1111 Label* to) { | |
| 1112 if (condition < 0) { // No condition | |
| 1113 if (to == NULL) { | |
| 1114 Backtrack(); | |
| 1115 return; | |
| 1116 } | |
| 1117 __ jmp(to); | |
| 1118 return; | |
| 1119 } | |
| 1120 if (to == NULL) { | |
| 1121 __ j(condition, &backtrack_label_); | |
| 1122 return; | |
| 1123 } | |
| 1124 __ j(condition, to); | |
| 1125 } | |
| 1126 | |
| 1127 | |
| 1128 void RegExpMacroAssemblerIA32::SafeCall(Label* to) { | |
| 1129 Label return_to; | |
| 1130 __ push(Immediate::CodeRelativeOffset(&return_to)); | |
| 1131 __ jmp(to); | |
| 1132 __ bind(&return_to); | |
| 1133 } | |
| 1134 | |
| 1135 | |
| 1136 void RegExpMacroAssemblerIA32::SafeReturn() { | |
| 1137 __ pop(ebx); | |
| 1138 __ add(ebx, Immediate(masm_->CodeObject())); | |
| 1139 __ jmp(ebx); | |
| 1140 } | |
| 1141 | |
| 1142 | |
| 1143 void RegExpMacroAssemblerIA32::SafeCallTarget(Label* name) { | |
| 1144 __ bind(name); | |
| 1145 } | |
| 1146 | |
| 1147 | |
| 1148 void RegExpMacroAssemblerIA32::Push(Register source) { | |
| 1149 DCHECK(!source.is(backtrack_stackpointer())); | |
| 1150 // Notice: This updates flags, unlike normal Push. | |
| 1151 __ sub(backtrack_stackpointer(), Immediate(kPointerSize)); | |
| 1152 __ mov(Operand(backtrack_stackpointer(), 0), source); | |
| 1153 } | |
| 1154 | |
| 1155 | |
| 1156 void RegExpMacroAssemblerIA32::Push(Immediate value) { | |
| 1157 // Notice: This updates flags, unlike normal Push. | |
| 1158 __ sub(backtrack_stackpointer(), Immediate(kPointerSize)); | |
| 1159 __ mov(Operand(backtrack_stackpointer(), 0), value); | |
| 1160 } | |
| 1161 | |
| 1162 | |
| 1163 void RegExpMacroAssemblerIA32::Pop(Register target) { | |
| 1164 DCHECK(!target.is(backtrack_stackpointer())); | |
| 1165 __ mov(target, Operand(backtrack_stackpointer(), 0)); | |
| 1166 // Notice: This updates flags, unlike normal Pop. | |
| 1167 __ add(backtrack_stackpointer(), Immediate(kPointerSize)); | |
| 1168 } | |
| 1169 | |
| 1170 | |
| 1171 void RegExpMacroAssemblerIA32::CheckPreemption() { | |
| 1172 // Check for preemption. | |
| 1173 Label no_preempt; | |
| 1174 ExternalReference stack_limit = | |
| 1175 ExternalReference::address_of_stack_limit(isolate()); | |
| 1176 __ cmp(esp, Operand::StaticVariable(stack_limit)); | |
| 1177 __ j(above, &no_preempt); | |
| 1178 | |
| 1179 SafeCall(&check_preempt_label_); | |
| 1180 | |
| 1181 __ bind(&no_preempt); | |
| 1182 } | |
| 1183 | |
| 1184 | |
| 1185 void RegExpMacroAssemblerIA32::CheckStackLimit() { | |
| 1186 Label no_stack_overflow; | |
| 1187 ExternalReference stack_limit = | |
| 1188 ExternalReference::address_of_regexp_stack_limit(isolate()); | |
| 1189 __ cmp(backtrack_stackpointer(), Operand::StaticVariable(stack_limit)); | |
| 1190 __ j(above, &no_stack_overflow); | |
| 1191 | |
| 1192 SafeCall(&stack_overflow_label_); | |
| 1193 | |
| 1194 __ bind(&no_stack_overflow); | |
| 1195 } | |
| 1196 | |
| 1197 | |
| 1198 void RegExpMacroAssemblerIA32::LoadCurrentCharacterUnchecked(int cp_offset, | |
| 1199 int characters) { | |
| 1200 if (mode_ == LATIN1) { | |
| 1201 if (characters == 4) { | |
| 1202 __ mov(current_character(), Operand(esi, edi, times_1, cp_offset)); | |
| 1203 } else if (characters == 2) { | |
| 1204 __ movzx_w(current_character(), Operand(esi, edi, times_1, cp_offset)); | |
| 1205 } else { | |
| 1206 DCHECK(characters == 1); | |
| 1207 __ movzx_b(current_character(), Operand(esi, edi, times_1, cp_offset)); | |
| 1208 } | |
| 1209 } else { | |
| 1210 DCHECK(mode_ == UC16); | |
| 1211 if (characters == 2) { | |
| 1212 __ mov(current_character(), | |
| 1213 Operand(esi, edi, times_1, cp_offset * sizeof(uc16))); | |
| 1214 } else { | |
| 1215 DCHECK(characters == 1); | |
| 1216 __ movzx_w(current_character(), | |
| 1217 Operand(esi, edi, times_1, cp_offset * sizeof(uc16))); | |
| 1218 } | |
| 1219 } | |
| 1220 } | |
| 1221 | |
| 1222 | |
| 1223 #undef __ | |
| 1224 | |
| 1225 #endif // V8_INTERPRETED_REGEXP | |
| 1226 | |
| 1227 } // namespace internal | |
| 1228 } // namespace v8 | |
| 1229 | |
| 1230 #endif // V8_TARGET_ARCH_IA32 | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 1285163003:
Move regexp implementation into its own folder. (Closed)
Base URL: https://chromium.googlesource.com/v8/v8.git@master