| 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_ARM | |
| 8 | |
| 9 #include "src/code-stubs.h" | |
| 10 #include "src/cpu-profiler.h" | |
| 11 #include "src/log.h" | |
| 12 #include "src/macro-assembler.h" | |
| 13 #include "src/regexp-macro-assembler.h" | |
| 14 #include "src/regexp-stack.h" | |
| 15 #include "src/unicode.h" | |
| 16 | |
| 17 #include "src/arm/regexp-macro-assembler-arm.h" | |
| 18 | |
| 19 namespace v8 { | |
| 20 namespace internal { | |
| 21 | |
| 22 #ifndef V8_INTERPRETED_REGEXP | |
| 23 /* | |
| 24 * This assembler uses the following register assignment convention | |
| 25 * - r4 : Temporarily stores the index of capture start after a matching pass | |
| 26 * for a global regexp. | |
| 27 * - r5 : Pointer to current code object (Code*) including heap object tag. | |
| 28 * - r6 : Current position in input, as negative offset from end of string. | |
| 29 * Please notice that this is the byte offset, not the character offset! | |
| 30 * - r7 : Currently loaded character. Must be loaded using | |
| 31 * LoadCurrentCharacter before using any of the dispatch methods. | |
| 32 * - r8 : Points to tip of backtrack stack | |
| 33 * - r9 : Unused, might be used by C code and expected unchanged. | |
| 34 * - r10 : End of input (points to byte after last character in input). | |
| 35 * - r11 : Frame pointer. Used to access arguments, local variables and | |
| 36 * RegExp registers. | |
| 37 * - r12 : IP register, used by assembler. Very volatile. | |
| 38 * - r13/sp : Points to tip of C stack. | |
| 39 * | |
| 40 * The remaining registers are free for computations. | |
| 41 * Each call to a public method should retain this convention. | |
| 42 * | |
| 43 * The stack will have the following structure: | |
| 44 * - fp[56] Isolate* isolate (address of the current isolate) | |
| 45 * - fp[52] direct_call (if 1, direct call from JavaScript code, | |
| 46 * if 0, call through the runtime system). | |
| 47 * - fp[48] stack_area_base (high end of the memory area to use as | |
| 48 * backtracking stack). | |
| 49 * - fp[44] capture array size (may fit multiple sets of matches) | |
| 50 * - fp[40] int* capture_array (int[num_saved_registers_], for output). | |
| 51 * - fp[36] secondary link/return address used by native call. | |
| 52 * --- sp when called --- | |
| 53 * - fp[32] return address (lr). | |
| 54 * - fp[28] old frame pointer (r11). | |
| 55 * - fp[0..24] backup of registers r4..r10. | |
| 56 * --- frame pointer ---- | |
| 57 * - fp[-4] end of input (address of end of string). | |
| 58 * - fp[-8] start of input (address of first character in string). | |
| 59 * - fp[-12] start index (character index of start). | |
| 60 * - fp[-16] void* input_string (location of a handle containing the string). | |
| 61 * - fp[-20] success counter (only for global regexps to count matches). | |
| 62 * - fp[-24] Offset of location before start of input (effectively character | |
| 63 * position -1). Used to initialize capture registers to a | |
| 64 * non-position. | |
| 65 * - fp[-28] At start (if 1, we are starting at the start of the | |
| 66 * string, otherwise 0) | |
| 67 * - fp[-32] register 0 (Only positions must be stored in the first | |
| 68 * - register 1 num_saved_registers_ registers) | |
| 69 * - ... | |
| 70 * - register num_registers-1 | |
| 71 * --- sp --- | |
| 72 * | |
| 73 * The first num_saved_registers_ registers are initialized to point to | |
| 74 * "character -1" in the string (i.e., char_size() bytes before the first | |
| 75 * character of the string). The remaining registers start out as garbage. | |
| 76 * | |
| 77 * The data up to the return address must be placed there by the calling | |
| 78 * code and the remaining arguments are passed in registers, e.g. by calling the | |
| 79 * code entry as cast to a function with the signature: | |
| 80 * int (*match)(String* input_string, | |
| 81 * int start_index, | |
| 82 * Address start, | |
| 83 * Address end, | |
| 84 * Address secondary_return_address, // Only used by native call. | |
| 85 * int* capture_output_array, | |
| 86 * byte* stack_area_base, | |
| 87 * bool direct_call = false) | |
| 88 * The call is performed by NativeRegExpMacroAssembler::Execute() | |
| 89 * (in regexp-macro-assembler.cc) via the CALL_GENERATED_REGEXP_CODE macro | |
| 90 * in arm/simulator-arm.h. | |
| 91 * When calling as a non-direct call (i.e., from C++ code), the return address | |
| 92 * area is overwritten with the LR register by the RegExp code. When doing a | |
| 93 * direct call from generated code, the return address is placed there by | |
| 94 * the calling code, as in a normal exit frame. | |
| 95 */ | |
| 96 | |
| 97 #define __ ACCESS_MASM(masm_) | |
| 98 | |
| 99 RegExpMacroAssemblerARM::RegExpMacroAssemblerARM(Isolate* isolate, Zone* zone, | |
| 100 Mode mode, | |
| 101 int registers_to_save) | |
| 102 : NativeRegExpMacroAssembler(isolate, zone), | |
| 103 masm_(new MacroAssembler(isolate, NULL, kRegExpCodeSize)), | |
| 104 mode_(mode), | |
| 105 num_registers_(registers_to_save), | |
| 106 num_saved_registers_(registers_to_save), | |
| 107 entry_label_(), | |
| 108 start_label_(), | |
| 109 success_label_(), | |
| 110 backtrack_label_(), | |
| 111 exit_label_() { | |
| 112 DCHECK_EQ(0, registers_to_save % 2); | |
| 113 __ jmp(&entry_label_); // We'll write the entry code later. | |
| 114 __ bind(&start_label_); // And then continue from here. | |
| 115 } | |
| 116 | |
| 117 | |
| 118 RegExpMacroAssemblerARM::~RegExpMacroAssemblerARM() { | |
| 119 delete masm_; | |
| 120 // Unuse labels in case we throw away the assembler without calling GetCode. | |
| 121 entry_label_.Unuse(); | |
| 122 start_label_.Unuse(); | |
| 123 success_label_.Unuse(); | |
| 124 backtrack_label_.Unuse(); | |
| 125 exit_label_.Unuse(); | |
| 126 check_preempt_label_.Unuse(); | |
| 127 stack_overflow_label_.Unuse(); | |
| 128 } | |
| 129 | |
| 130 | |
| 131 int RegExpMacroAssemblerARM::stack_limit_slack() { | |
| 132 return RegExpStack::kStackLimitSlack; | |
| 133 } | |
| 134 | |
| 135 | |
| 136 void RegExpMacroAssemblerARM::AdvanceCurrentPosition(int by) { | |
| 137 if (by != 0) { | |
| 138 __ add(current_input_offset(), | |
| 139 current_input_offset(), Operand(by * char_size())); | |
| 140 } | |
| 141 } | |
| 142 | |
| 143 | |
| 144 void RegExpMacroAssemblerARM::AdvanceRegister(int reg, int by) { | |
| 145 DCHECK(reg >= 0); | |
| 146 DCHECK(reg < num_registers_); | |
| 147 if (by != 0) { | |
| 148 __ ldr(r0, register_location(reg)); | |
| 149 __ add(r0, r0, Operand(by)); | |
| 150 __ str(r0, register_location(reg)); | |
| 151 } | |
| 152 } | |
| 153 | |
| 154 | |
| 155 void RegExpMacroAssemblerARM::Backtrack() { | |
| 156 CheckPreemption(); | |
| 157 // Pop Code* offset from backtrack stack, add Code* and jump to location. | |
| 158 Pop(r0); | |
| 159 __ add(pc, r0, Operand(code_pointer())); | |
| 160 } | |
| 161 | |
| 162 | |
| 163 void RegExpMacroAssemblerARM::Bind(Label* label) { | |
| 164 __ bind(label); | |
| 165 } | |
| 166 | |
| 167 | |
| 168 void RegExpMacroAssemblerARM::CheckCharacter(uint32_t c, Label* on_equal) { | |
| 169 __ cmp(current_character(), Operand(c)); | |
| 170 BranchOrBacktrack(eq, on_equal); | |
| 171 } | |
| 172 | |
| 173 | |
| 174 void RegExpMacroAssemblerARM::CheckCharacterGT(uc16 limit, Label* on_greater) { | |
| 175 __ cmp(current_character(), Operand(limit)); | |
| 176 BranchOrBacktrack(gt, on_greater); | |
| 177 } | |
| 178 | |
| 179 | |
| 180 void RegExpMacroAssemblerARM::CheckAtStart(Label* on_at_start) { | |
| 181 Label not_at_start; | |
| 182 // Did we start the match at the start of the string at all? | |
| 183 __ ldr(r0, MemOperand(frame_pointer(), kStartIndex)); | |
| 184 __ cmp(r0, Operand::Zero()); | |
| 185 BranchOrBacktrack(ne, ¬_at_start); | |
| 186 | |
| 187 // If we did, are we still at the start of the input? | |
| 188 __ ldr(r1, MemOperand(frame_pointer(), kInputStart)); | |
| 189 __ add(r0, end_of_input_address(), Operand(current_input_offset())); | |
| 190 __ cmp(r0, r1); | |
| 191 BranchOrBacktrack(eq, on_at_start); | |
| 192 __ bind(¬_at_start); | |
| 193 } | |
| 194 | |
| 195 | |
| 196 void RegExpMacroAssemblerARM::CheckNotAtStart(Label* on_not_at_start) { | |
| 197 // Did we start the match at the start of the string at all? | |
| 198 __ ldr(r0, MemOperand(frame_pointer(), kStartIndex)); | |
| 199 __ cmp(r0, Operand::Zero()); | |
| 200 BranchOrBacktrack(ne, on_not_at_start); | |
| 201 // If we did, are we still at the start of the input? | |
| 202 __ ldr(r1, MemOperand(frame_pointer(), kInputStart)); | |
| 203 __ add(r0, end_of_input_address(), Operand(current_input_offset())); | |
| 204 __ cmp(r0, r1); | |
| 205 BranchOrBacktrack(ne, on_not_at_start); | |
| 206 } | |
| 207 | |
| 208 | |
| 209 void RegExpMacroAssemblerARM::CheckCharacterLT(uc16 limit, Label* on_less) { | |
| 210 __ cmp(current_character(), Operand(limit)); | |
| 211 BranchOrBacktrack(lt, on_less); | |
| 212 } | |
| 213 | |
| 214 | |
| 215 void RegExpMacroAssemblerARM::CheckGreedyLoop(Label* on_equal) { | |
| 216 __ ldr(r0, MemOperand(backtrack_stackpointer(), 0)); | |
| 217 __ cmp(current_input_offset(), r0); | |
| 218 __ add(backtrack_stackpointer(), | |
| 219 backtrack_stackpointer(), Operand(kPointerSize), LeaveCC, eq); | |
| 220 BranchOrBacktrack(eq, on_equal); | |
| 221 } | |
| 222 | |
| 223 | |
| 224 void RegExpMacroAssemblerARM::CheckNotBackReferenceIgnoreCase( | |
| 225 int start_reg, | |
| 226 Label* on_no_match) { | |
| 227 Label fallthrough; | |
| 228 __ ldr(r0, register_location(start_reg)); // Index of start of capture | |
| 229 __ ldr(r1, register_location(start_reg + 1)); // Index of end of capture | |
| 230 __ sub(r1, r1, r0, SetCC); // Length of capture. | |
| 231 | |
| 232 // If length is zero, either the capture is empty or it is not participating. | |
| 233 // In either case succeed immediately. | |
| 234 __ b(eq, &fallthrough); | |
| 235 | |
| 236 // Check that there are enough characters left in the input. | |
| 237 __ cmn(r1, Operand(current_input_offset())); | |
| 238 BranchOrBacktrack(gt, on_no_match); | |
| 239 | |
| 240 if (mode_ == LATIN1) { | |
| 241 Label success; | |
| 242 Label fail; | |
| 243 Label loop_check; | |
| 244 | |
| 245 // r0 - offset of start of capture | |
| 246 // r1 - length of capture | |
| 247 __ add(r0, r0, Operand(end_of_input_address())); | |
| 248 __ add(r2, end_of_input_address(), Operand(current_input_offset())); | |
| 249 __ add(r1, r0, Operand(r1)); | |
| 250 | |
| 251 // r0 - Address of start of capture. | |
| 252 // r1 - Address of end of capture | |
| 253 // r2 - Address of current input position. | |
| 254 | |
| 255 Label loop; | |
| 256 __ bind(&loop); | |
| 257 __ ldrb(r3, MemOperand(r0, char_size(), PostIndex)); | |
| 258 __ ldrb(r4, MemOperand(r2, char_size(), PostIndex)); | |
| 259 __ cmp(r4, r3); | |
| 260 __ b(eq, &loop_check); | |
| 261 | |
| 262 // Mismatch, try case-insensitive match (converting letters to lower-case). | |
| 263 __ orr(r3, r3, Operand(0x20)); // Convert capture character to lower-case. | |
| 264 __ orr(r4, r4, Operand(0x20)); // Also convert input character. | |
| 265 __ cmp(r4, r3); | |
| 266 __ b(ne, &fail); | |
| 267 __ sub(r3, r3, Operand('a')); | |
| 268 __ cmp(r3, Operand('z' - 'a')); // Is r3 a lowercase letter? | |
| 269 __ b(ls, &loop_check); // In range 'a'-'z'. | |
| 270 // Latin-1: Check for values in range [224,254] but not 247. | |
| 271 __ sub(r3, r3, Operand(224 - 'a')); | |
| 272 __ cmp(r3, Operand(254 - 224)); | |
| 273 __ b(hi, &fail); // Weren't Latin-1 letters. | |
| 274 __ cmp(r3, Operand(247 - 224)); // Check for 247. | |
| 275 __ b(eq, &fail); | |
| 276 | |
| 277 __ bind(&loop_check); | |
| 278 __ cmp(r0, r1); | |
| 279 __ b(lt, &loop); | |
| 280 __ jmp(&success); | |
| 281 | |
| 282 __ bind(&fail); | |
| 283 BranchOrBacktrack(al, on_no_match); | |
| 284 | |
| 285 __ bind(&success); | |
| 286 // Compute new value of character position after the matched part. | |
| 287 __ sub(current_input_offset(), r2, end_of_input_address()); | |
| 288 } else { | |
| 289 DCHECK(mode_ == UC16); | |
| 290 int argument_count = 4; | |
| 291 __ PrepareCallCFunction(argument_count, r2); | |
| 292 | |
| 293 // r0 - offset of start of capture | |
| 294 // r1 - length of capture | |
| 295 | |
| 296 // Put arguments into arguments registers. | |
| 297 // Parameters are | |
| 298 // r0: Address byte_offset1 - Address captured substring's start. | |
| 299 // r1: Address byte_offset2 - Address of current character position. | |
| 300 // r2: size_t byte_length - length of capture in bytes(!) | |
| 301 // r3: Isolate* isolate | |
| 302 | |
| 303 // Address of start of capture. | |
| 304 __ add(r0, r0, Operand(end_of_input_address())); | |
| 305 // Length of capture. | |
| 306 __ mov(r2, Operand(r1)); | |
| 307 // Save length in callee-save register for use on return. | |
| 308 __ mov(r4, Operand(r1)); | |
| 309 // Address of current input position. | |
| 310 __ add(r1, current_input_offset(), Operand(end_of_input_address())); | |
| 311 // Isolate. | |
| 312 __ mov(r3, Operand(ExternalReference::isolate_address(isolate()))); | |
| 313 | |
| 314 { | |
| 315 AllowExternalCallThatCantCauseGC scope(masm_); | |
| 316 ExternalReference function = | |
| 317 ExternalReference::re_case_insensitive_compare_uc16(isolate()); | |
| 318 __ CallCFunction(function, argument_count); | |
| 319 } | |
| 320 | |
| 321 // Check if function returned non-zero for success or zero for failure. | |
| 322 __ cmp(r0, Operand::Zero()); | |
| 323 BranchOrBacktrack(eq, on_no_match); | |
| 324 // On success, increment position by length of capture. | |
| 325 __ add(current_input_offset(), current_input_offset(), Operand(r4)); | |
| 326 } | |
| 327 | |
| 328 __ bind(&fallthrough); | |
| 329 } | |
| 330 | |
| 331 | |
| 332 void RegExpMacroAssemblerARM::CheckNotBackReference( | |
| 333 int start_reg, | |
| 334 Label* on_no_match) { | |
| 335 Label fallthrough; | |
| 336 Label success; | |
| 337 | |
| 338 // Find length of back-referenced capture. | |
| 339 __ ldr(r0, register_location(start_reg)); | |
| 340 __ ldr(r1, register_location(start_reg + 1)); | |
| 341 __ sub(r1, r1, r0, SetCC); // Length to check. | |
| 342 // Succeed on empty capture (including no capture). | |
| 343 __ b(eq, &fallthrough); | |
| 344 | |
| 345 // Check that there are enough characters left in the input. | |
| 346 __ cmn(r1, Operand(current_input_offset())); | |
| 347 BranchOrBacktrack(gt, on_no_match); | |
| 348 | |
| 349 // Compute pointers to match string and capture string | |
| 350 __ add(r0, r0, Operand(end_of_input_address())); | |
| 351 __ add(r2, end_of_input_address(), Operand(current_input_offset())); | |
| 352 __ add(r1, r1, Operand(r0)); | |
| 353 | |
| 354 Label loop; | |
| 355 __ bind(&loop); | |
| 356 if (mode_ == LATIN1) { | |
| 357 __ ldrb(r3, MemOperand(r0, char_size(), PostIndex)); | |
| 358 __ ldrb(r4, MemOperand(r2, char_size(), PostIndex)); | |
| 359 } else { | |
| 360 DCHECK(mode_ == UC16); | |
| 361 __ ldrh(r3, MemOperand(r0, char_size(), PostIndex)); | |
| 362 __ ldrh(r4, MemOperand(r2, char_size(), PostIndex)); | |
| 363 } | |
| 364 __ cmp(r3, r4); | |
| 365 BranchOrBacktrack(ne, on_no_match); | |
| 366 __ cmp(r0, r1); | |
| 367 __ b(lt, &loop); | |
| 368 | |
| 369 // Move current character position to position after match. | |
| 370 __ sub(current_input_offset(), r2, end_of_input_address()); | |
| 371 __ bind(&fallthrough); | |
| 372 } | |
| 373 | |
| 374 | |
| 375 void RegExpMacroAssemblerARM::CheckNotCharacter(unsigned c, | |
| 376 Label* on_not_equal) { | |
| 377 __ cmp(current_character(), Operand(c)); | |
| 378 BranchOrBacktrack(ne, on_not_equal); | |
| 379 } | |
| 380 | |
| 381 | |
| 382 void RegExpMacroAssemblerARM::CheckCharacterAfterAnd(uint32_t c, | |
| 383 uint32_t mask, | |
| 384 Label* on_equal) { | |
| 385 if (c == 0) { | |
| 386 __ tst(current_character(), Operand(mask)); | |
| 387 } else { | |
| 388 __ and_(r0, current_character(), Operand(mask)); | |
| 389 __ cmp(r0, Operand(c)); | |
| 390 } | |
| 391 BranchOrBacktrack(eq, on_equal); | |
| 392 } | |
| 393 | |
| 394 | |
| 395 void RegExpMacroAssemblerARM::CheckNotCharacterAfterAnd(unsigned c, | |
| 396 unsigned mask, | |
| 397 Label* on_not_equal) { | |
| 398 if (c == 0) { | |
| 399 __ tst(current_character(), Operand(mask)); | |
| 400 } else { | |
| 401 __ and_(r0, current_character(), Operand(mask)); | |
| 402 __ cmp(r0, Operand(c)); | |
| 403 } | |
| 404 BranchOrBacktrack(ne, on_not_equal); | |
| 405 } | |
| 406 | |
| 407 | |
| 408 void RegExpMacroAssemblerARM::CheckNotCharacterAfterMinusAnd( | |
| 409 uc16 c, | |
| 410 uc16 minus, | |
| 411 uc16 mask, | |
| 412 Label* on_not_equal) { | |
| 413 DCHECK(minus < String::kMaxUtf16CodeUnit); | |
| 414 __ sub(r0, current_character(), Operand(minus)); | |
| 415 __ and_(r0, r0, Operand(mask)); | |
| 416 __ cmp(r0, Operand(c)); | |
| 417 BranchOrBacktrack(ne, on_not_equal); | |
| 418 } | |
| 419 | |
| 420 | |
| 421 void RegExpMacroAssemblerARM::CheckCharacterInRange( | |
| 422 uc16 from, | |
| 423 uc16 to, | |
| 424 Label* on_in_range) { | |
| 425 __ sub(r0, current_character(), Operand(from)); | |
| 426 __ cmp(r0, Operand(to - from)); | |
| 427 BranchOrBacktrack(ls, on_in_range); // Unsigned lower-or-same condition. | |
| 428 } | |
| 429 | |
| 430 | |
| 431 void RegExpMacroAssemblerARM::CheckCharacterNotInRange( | |
| 432 uc16 from, | |
| 433 uc16 to, | |
| 434 Label* on_not_in_range) { | |
| 435 __ sub(r0, current_character(), Operand(from)); | |
| 436 __ cmp(r0, Operand(to - from)); | |
| 437 BranchOrBacktrack(hi, on_not_in_range); // Unsigned higher condition. | |
| 438 } | |
| 439 | |
| 440 | |
| 441 void RegExpMacroAssemblerARM::CheckBitInTable( | |
| 442 Handle<ByteArray> table, | |
| 443 Label* on_bit_set) { | |
| 444 __ mov(r0, Operand(table)); | |
| 445 if (mode_ != LATIN1 || kTableMask != String::kMaxOneByteCharCode) { | |
| 446 __ and_(r1, current_character(), Operand(kTableSize - 1)); | |
| 447 __ add(r1, r1, Operand(ByteArray::kHeaderSize - kHeapObjectTag)); | |
| 448 } else { | |
| 449 __ add(r1, | |
| 450 current_character(), | |
| 451 Operand(ByteArray::kHeaderSize - kHeapObjectTag)); | |
| 452 } | |
| 453 __ ldrb(r0, MemOperand(r0, r1)); | |
| 454 __ cmp(r0, Operand::Zero()); | |
| 455 BranchOrBacktrack(ne, on_bit_set); | |
| 456 } | |
| 457 | |
| 458 | |
| 459 bool RegExpMacroAssemblerARM::CheckSpecialCharacterClass(uc16 type, | |
| 460 Label* on_no_match) { | |
| 461 // Range checks (c in min..max) are generally implemented by an unsigned | |
| 462 // (c - min) <= (max - min) check | |
| 463 switch (type) { | |
| 464 case 's': | |
| 465 // Match space-characters | |
| 466 if (mode_ == LATIN1) { | |
| 467 // One byte space characters are '\t'..'\r', ' ' and \u00a0. | |
| 468 Label success; | |
| 469 __ cmp(current_character(), Operand(' ')); | |
| 470 __ b(eq, &success); | |
| 471 // Check range 0x09..0x0d | |
| 472 __ sub(r0, current_character(), Operand('\t')); | |
| 473 __ cmp(r0, Operand('\r' - '\t')); | |
| 474 __ b(ls, &success); | |
| 475 // \u00a0 (NBSP). | |
| 476 __ cmp(r0, Operand(0x00a0 - '\t')); | |
| 477 BranchOrBacktrack(ne, on_no_match); | |
| 478 __ bind(&success); | |
| 479 return true; | |
| 480 } | |
| 481 return false; | |
| 482 case 'S': | |
| 483 // The emitted code for generic character classes is good enough. | |
| 484 return false; | |
| 485 case 'd': | |
| 486 // Match ASCII digits ('0'..'9') | |
| 487 __ sub(r0, current_character(), Operand('0')); | |
| 488 __ cmp(r0, Operand('9' - '0')); | |
| 489 BranchOrBacktrack(hi, on_no_match); | |
| 490 return true; | |
| 491 case 'D': | |
| 492 // Match non ASCII-digits | |
| 493 __ sub(r0, current_character(), Operand('0')); | |
| 494 __ cmp(r0, Operand('9' - '0')); | |
| 495 BranchOrBacktrack(ls, on_no_match); | |
| 496 return true; | |
| 497 case '.': { | |
| 498 // Match non-newlines (not 0x0a('\n'), 0x0d('\r'), 0x2028 and 0x2029) | |
| 499 __ eor(r0, current_character(), Operand(0x01)); | |
| 500 // See if current character is '\n'^1 or '\r'^1, i.e., 0x0b or 0x0c | |
| 501 __ sub(r0, r0, Operand(0x0b)); | |
| 502 __ cmp(r0, Operand(0x0c - 0x0b)); | |
| 503 BranchOrBacktrack(ls, on_no_match); | |
| 504 if (mode_ == UC16) { | |
| 505 // Compare original value to 0x2028 and 0x2029, using the already | |
| 506 // computed (current_char ^ 0x01 - 0x0b). I.e., check for | |
| 507 // 0x201d (0x2028 - 0x0b) or 0x201e. | |
| 508 __ sub(r0, r0, Operand(0x2028 - 0x0b)); | |
| 509 __ cmp(r0, Operand(1)); | |
| 510 BranchOrBacktrack(ls, on_no_match); | |
| 511 } | |
| 512 return true; | |
| 513 } | |
| 514 case 'n': { | |
| 515 // Match newlines (0x0a('\n'), 0x0d('\r'), 0x2028 and 0x2029) | |
| 516 __ eor(r0, current_character(), Operand(0x01)); | |
| 517 // See if current character is '\n'^1 or '\r'^1, i.e., 0x0b or 0x0c | |
| 518 __ sub(r0, r0, Operand(0x0b)); | |
| 519 __ cmp(r0, Operand(0x0c - 0x0b)); | |
| 520 if (mode_ == LATIN1) { | |
| 521 BranchOrBacktrack(hi, on_no_match); | |
| 522 } else { | |
| 523 Label done; | |
| 524 __ b(ls, &done); | |
| 525 // Compare original value to 0x2028 and 0x2029, using the already | |
| 526 // computed (current_char ^ 0x01 - 0x0b). I.e., check for | |
| 527 // 0x201d (0x2028 - 0x0b) or 0x201e. | |
| 528 __ sub(r0, r0, Operand(0x2028 - 0x0b)); | |
| 529 __ cmp(r0, Operand(1)); | |
| 530 BranchOrBacktrack(hi, on_no_match); | |
| 531 __ bind(&done); | |
| 532 } | |
| 533 return true; | |
| 534 } | |
| 535 case 'w': { | |
| 536 if (mode_ != LATIN1) { | |
| 537 // Table is 256 entries, so all Latin1 characters can be tested. | |
| 538 __ cmp(current_character(), Operand('z')); | |
| 539 BranchOrBacktrack(hi, on_no_match); | |
| 540 } | |
| 541 ExternalReference map = ExternalReference::re_word_character_map(); | |
| 542 __ mov(r0, Operand(map)); | |
| 543 __ ldrb(r0, MemOperand(r0, current_character())); | |
| 544 __ cmp(r0, Operand::Zero()); | |
| 545 BranchOrBacktrack(eq, on_no_match); | |
| 546 return true; | |
| 547 } | |
| 548 case 'W': { | |
| 549 Label done; | |
| 550 if (mode_ != LATIN1) { | |
| 551 // Table is 256 entries, so all Latin1 characters can be tested. | |
| 552 __ cmp(current_character(), Operand('z')); | |
| 553 __ b(hi, &done); | |
| 554 } | |
| 555 ExternalReference map = ExternalReference::re_word_character_map(); | |
| 556 __ mov(r0, Operand(map)); | |
| 557 __ ldrb(r0, MemOperand(r0, current_character())); | |
| 558 __ cmp(r0, Operand::Zero()); | |
| 559 BranchOrBacktrack(ne, on_no_match); | |
| 560 if (mode_ != LATIN1) { | |
| 561 __ bind(&done); | |
| 562 } | |
| 563 return true; | |
| 564 } | |
| 565 case '*': | |
| 566 // Match any character. | |
| 567 return true; | |
| 568 // No custom implementation (yet): s(UC16), S(UC16). | |
| 569 default: | |
| 570 return false; | |
| 571 } | |
| 572 } | |
| 573 | |
| 574 | |
| 575 void RegExpMacroAssemblerARM::Fail() { | |
| 576 __ mov(r0, Operand(FAILURE)); | |
| 577 __ jmp(&exit_label_); | |
| 578 } | |
| 579 | |
| 580 | |
| 581 Handle<HeapObject> RegExpMacroAssemblerARM::GetCode(Handle<String> source) { | |
| 582 Label return_r0; | |
| 583 // Finalize code - write the entry point code now we know how many | |
| 584 // registers we need. | |
| 585 | |
| 586 // Entry code: | |
| 587 __ bind(&entry_label_); | |
| 588 | |
| 589 // Tell the system that we have a stack frame. Because the type is MANUAL, no | |
| 590 // is generated. | |
| 591 FrameScope scope(masm_, StackFrame::MANUAL); | |
| 592 | |
| 593 // Actually emit code to start a new stack frame. | |
| 594 // Push arguments | |
| 595 // Save callee-save registers. | |
| 596 // Start new stack frame. | |
| 597 // Store link register in existing stack-cell. | |
| 598 // Order here should correspond to order of offset constants in header file. | |
| 599 RegList registers_to_retain = r4.bit() | r5.bit() | r6.bit() | | |
| 600 r7.bit() | r8.bit() | r9.bit() | r10.bit() | fp.bit(); | |
| 601 RegList argument_registers = r0.bit() | r1.bit() | r2.bit() | r3.bit(); | |
| 602 __ stm(db_w, sp, argument_registers | registers_to_retain | lr.bit()); | |
| 603 // Set frame pointer in space for it if this is not a direct call | |
| 604 // from generated code. | |
| 605 __ add(frame_pointer(), sp, Operand(4 * kPointerSize)); | |
| 606 __ mov(r0, Operand::Zero()); | |
| 607 __ push(r0); // Make room for success counter and initialize it to 0. | |
| 608 __ push(r0); // Make room for "position - 1" constant (value is irrelevant). | |
| 609 // Check if we have space on the stack for registers. | |
| 610 Label stack_limit_hit; | |
| 611 Label stack_ok; | |
| 612 | |
| 613 ExternalReference stack_limit = | |
| 614 ExternalReference::address_of_stack_limit(isolate()); | |
| 615 __ mov(r0, Operand(stack_limit)); | |
| 616 __ ldr(r0, MemOperand(r0)); | |
| 617 __ sub(r0, sp, r0, SetCC); | |
| 618 // Handle it if the stack pointer is already below the stack limit. | |
| 619 __ b(ls, &stack_limit_hit); | |
| 620 // Check if there is room for the variable number of registers above | |
| 621 // the stack limit. | |
| 622 __ cmp(r0, Operand(num_registers_ * kPointerSize)); | |
| 623 __ b(hs, &stack_ok); | |
| 624 // Exit with OutOfMemory exception. There is not enough space on the stack | |
| 625 // for our working registers. | |
| 626 __ mov(r0, Operand(EXCEPTION)); | |
| 627 __ jmp(&return_r0); | |
| 628 | |
| 629 __ bind(&stack_limit_hit); | |
| 630 CallCheckStackGuardState(r0); | |
| 631 __ cmp(r0, Operand::Zero()); | |
| 632 // If returned value is non-zero, we exit with the returned value as result. | |
| 633 __ b(ne, &return_r0); | |
| 634 | |
| 635 __ bind(&stack_ok); | |
| 636 | |
| 637 // Allocate space on stack for registers. | |
| 638 __ sub(sp, sp, Operand(num_registers_ * kPointerSize)); | |
| 639 // Load string end. | |
| 640 __ ldr(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd)); | |
| 641 // Load input start. | |
| 642 __ ldr(r0, MemOperand(frame_pointer(), kInputStart)); | |
| 643 // Find negative length (offset of start relative to end). | |
| 644 __ sub(current_input_offset(), r0, end_of_input_address()); | |
| 645 // Set r0 to address of char before start of the input string | |
| 646 // (effectively string position -1). | |
| 647 __ ldr(r1, MemOperand(frame_pointer(), kStartIndex)); | |
| 648 __ sub(r0, current_input_offset(), Operand(char_size())); | |
| 649 __ sub(r0, r0, Operand(r1, LSL, (mode_ == UC16) ? 1 : 0)); | |
| 650 // Store this value in a local variable, for use when clearing | |
| 651 // position registers. | |
| 652 __ str(r0, MemOperand(frame_pointer(), kInputStartMinusOne)); | |
| 653 | |
| 654 // Initialize code pointer register | |
| 655 __ mov(code_pointer(), Operand(masm_->CodeObject())); | |
| 656 | |
| 657 Label load_char_start_regexp, start_regexp; | |
| 658 // Load newline if index is at start, previous character otherwise. | |
| 659 __ cmp(r1, Operand::Zero()); | |
| 660 __ b(ne, &load_char_start_regexp); | |
| 661 __ mov(current_character(), Operand('\n'), LeaveCC, eq); | |
| 662 __ jmp(&start_regexp); | |
| 663 | |
| 664 // Global regexp restarts matching here. | |
| 665 __ bind(&load_char_start_regexp); | |
| 666 // Load previous char as initial value of current character register. | |
| 667 LoadCurrentCharacterUnchecked(-1, 1); | |
| 668 __ bind(&start_regexp); | |
| 669 | |
| 670 // Initialize on-stack registers. | |
| 671 if (num_saved_registers_ > 0) { // Always is, if generated from a regexp. | |
| 672 // Fill saved registers with initial value = start offset - 1 | |
| 673 if (num_saved_registers_ > 8) { | |
| 674 // Address of register 0. | |
| 675 __ add(r1, frame_pointer(), Operand(kRegisterZero)); | |
| 676 __ mov(r2, Operand(num_saved_registers_)); | |
| 677 Label init_loop; | |
| 678 __ bind(&init_loop); | |
| 679 __ str(r0, MemOperand(r1, kPointerSize, NegPostIndex)); | |
| 680 __ sub(r2, r2, Operand(1), SetCC); | |
| 681 __ b(ne, &init_loop); | |
| 682 } else { | |
| 683 for (int i = 0; i < num_saved_registers_; i++) { | |
| 684 __ str(r0, register_location(i)); | |
| 685 } | |
| 686 } | |
| 687 } | |
| 688 | |
| 689 // Initialize backtrack stack pointer. | |
| 690 __ ldr(backtrack_stackpointer(), MemOperand(frame_pointer(), kStackHighEnd)); | |
| 691 | |
| 692 __ jmp(&start_label_); | |
| 693 | |
| 694 // Exit code: | |
| 695 if (success_label_.is_linked()) { | |
| 696 // Save captures when successful. | |
| 697 __ bind(&success_label_); | |
| 698 if (num_saved_registers_ > 0) { | |
| 699 // copy captures to output | |
| 700 __ ldr(r1, MemOperand(frame_pointer(), kInputStart)); | |
| 701 __ ldr(r0, MemOperand(frame_pointer(), kRegisterOutput)); | |
| 702 __ ldr(r2, MemOperand(frame_pointer(), kStartIndex)); | |
| 703 __ sub(r1, end_of_input_address(), r1); | |
| 704 // r1 is length of input in bytes. | |
| 705 if (mode_ == UC16) { | |
| 706 __ mov(r1, Operand(r1, LSR, 1)); | |
| 707 } | |
| 708 // r1 is length of input in characters. | |
| 709 __ add(r1, r1, Operand(r2)); | |
| 710 // r1 is length of string in characters. | |
| 711 | |
| 712 DCHECK_EQ(0, num_saved_registers_ % 2); | |
| 713 // Always an even number of capture registers. This allows us to | |
| 714 // unroll the loop once to add an operation between a load of a register | |
| 715 // and the following use of that register. | |
| 716 for (int i = 0; i < num_saved_registers_; i += 2) { | |
| 717 __ ldr(r2, register_location(i)); | |
| 718 __ ldr(r3, register_location(i + 1)); | |
| 719 if (i == 0 && global_with_zero_length_check()) { | |
| 720 // Keep capture start in r4 for the zero-length check later. | |
| 721 __ mov(r4, r2); | |
| 722 } | |
| 723 if (mode_ == UC16) { | |
| 724 __ add(r2, r1, Operand(r2, ASR, 1)); | |
| 725 __ add(r3, r1, Operand(r3, ASR, 1)); | |
| 726 } else { | |
| 727 __ add(r2, r1, Operand(r2)); | |
| 728 __ add(r3, r1, Operand(r3)); | |
| 729 } | |
| 730 __ str(r2, MemOperand(r0, kPointerSize, PostIndex)); | |
| 731 __ str(r3, MemOperand(r0, kPointerSize, PostIndex)); | |
| 732 } | |
| 733 } | |
| 734 | |
| 735 if (global()) { | |
| 736 // Restart matching if the regular expression is flagged as global. | |
| 737 __ ldr(r0, MemOperand(frame_pointer(), kSuccessfulCaptures)); | |
| 738 __ ldr(r1, MemOperand(frame_pointer(), kNumOutputRegisters)); | |
| 739 __ ldr(r2, MemOperand(frame_pointer(), kRegisterOutput)); | |
| 740 // Increment success counter. | |
| 741 __ add(r0, r0, Operand(1)); | |
| 742 __ str(r0, MemOperand(frame_pointer(), kSuccessfulCaptures)); | |
| 743 // Capture results have been stored, so the number of remaining global | |
| 744 // output registers is reduced by the number of stored captures. | |
| 745 __ sub(r1, r1, Operand(num_saved_registers_)); | |
| 746 // Check whether we have enough room for another set of capture results. | |
| 747 __ cmp(r1, Operand(num_saved_registers_)); | |
| 748 __ b(lt, &return_r0); | |
| 749 | |
| 750 __ str(r1, MemOperand(frame_pointer(), kNumOutputRegisters)); | |
| 751 // Advance the location for output. | |
| 752 __ add(r2, r2, Operand(num_saved_registers_ * kPointerSize)); | |
| 753 __ str(r2, MemOperand(frame_pointer(), kRegisterOutput)); | |
| 754 | |
| 755 // Prepare r0 to initialize registers with its value in the next run. | |
| 756 __ ldr(r0, MemOperand(frame_pointer(), kInputStartMinusOne)); | |
| 757 | |
| 758 if (global_with_zero_length_check()) { | |
| 759 // Special case for zero-length matches. | |
| 760 // r4: capture start index | |
| 761 __ cmp(current_input_offset(), r4); | |
| 762 // Not a zero-length match, restart. | |
| 763 __ b(ne, &load_char_start_regexp); | |
| 764 // Offset from the end is zero if we already reached the end. | |
| 765 __ cmp(current_input_offset(), Operand::Zero()); | |
| 766 __ b(eq, &exit_label_); | |
| 767 // Advance current position after a zero-length match. | |
| 768 __ add(current_input_offset(), | |
| 769 current_input_offset(), | |
| 770 Operand((mode_ == UC16) ? 2 : 1)); | |
| 771 } | |
| 772 | |
| 773 __ b(&load_char_start_regexp); | |
| 774 } else { | |
| 775 __ mov(r0, Operand(SUCCESS)); | |
| 776 } | |
| 777 } | |
| 778 | |
| 779 // Exit and return r0 | |
| 780 __ bind(&exit_label_); | |
| 781 if (global()) { | |
| 782 __ ldr(r0, MemOperand(frame_pointer(), kSuccessfulCaptures)); | |
| 783 } | |
| 784 | |
| 785 __ bind(&return_r0); | |
| 786 // Skip sp past regexp registers and local variables.. | |
| 787 __ mov(sp, frame_pointer()); | |
| 788 // Restore registers r4..r11 and return (restoring lr to pc). | |
| 789 __ ldm(ia_w, sp, registers_to_retain | pc.bit()); | |
| 790 | |
| 791 // Backtrack code (branch target for conditional backtracks). | |
| 792 if (backtrack_label_.is_linked()) { | |
| 793 __ bind(&backtrack_label_); | |
| 794 Backtrack(); | |
| 795 } | |
| 796 | |
| 797 Label exit_with_exception; | |
| 798 | |
| 799 // Preempt-code | |
| 800 if (check_preempt_label_.is_linked()) { | |
| 801 SafeCallTarget(&check_preempt_label_); | |
| 802 | |
| 803 CallCheckStackGuardState(r0); | |
| 804 __ cmp(r0, Operand::Zero()); | |
| 805 // If returning non-zero, we should end execution with the given | |
| 806 // result as return value. | |
| 807 __ b(ne, &return_r0); | |
| 808 | |
| 809 // String might have moved: Reload end of string from frame. | |
| 810 __ ldr(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd)); | |
| 811 SafeReturn(); | |
| 812 } | |
| 813 | |
| 814 // Backtrack stack overflow code. | |
| 815 if (stack_overflow_label_.is_linked()) { | |
| 816 SafeCallTarget(&stack_overflow_label_); | |
| 817 // Reached if the backtrack-stack limit has been hit. | |
| 818 Label grow_failed; | |
| 819 | |
| 820 // Call GrowStack(backtrack_stackpointer(), &stack_base) | |
| 821 static const int num_arguments = 3; | |
| 822 __ PrepareCallCFunction(num_arguments, r0); | |
| 823 __ mov(r0, backtrack_stackpointer()); | |
| 824 __ add(r1, frame_pointer(), Operand(kStackHighEnd)); | |
| 825 __ mov(r2, Operand(ExternalReference::isolate_address(isolate()))); | |
| 826 ExternalReference grow_stack = | |
| 827 ExternalReference::re_grow_stack(isolate()); | |
| 828 __ CallCFunction(grow_stack, num_arguments); | |
| 829 // If return NULL, we have failed to grow the stack, and | |
| 830 // must exit with a stack-overflow exception. | |
| 831 __ cmp(r0, Operand::Zero()); | |
| 832 __ b(eq, &exit_with_exception); | |
| 833 // Otherwise use return value as new stack pointer. | |
| 834 __ mov(backtrack_stackpointer(), r0); | |
| 835 // Restore saved registers and continue. | |
| 836 SafeReturn(); | |
| 837 } | |
| 838 | |
| 839 if (exit_with_exception.is_linked()) { | |
| 840 // If any of the code above needed to exit with an exception. | |
| 841 __ bind(&exit_with_exception); | |
| 842 // Exit with Result EXCEPTION(-1) to signal thrown exception. | |
| 843 __ mov(r0, Operand(EXCEPTION)); | |
| 844 __ jmp(&return_r0); | |
| 845 } | |
| 846 | |
| 847 CodeDesc code_desc; | |
| 848 masm_->GetCode(&code_desc); | |
| 849 Handle<Code> code = isolate()->factory()->NewCode( | |
| 850 code_desc, Code::ComputeFlags(Code::REGEXP), masm_->CodeObject()); | |
| 851 PROFILE(masm_->isolate(), RegExpCodeCreateEvent(*code, *source)); | |
| 852 return Handle<HeapObject>::cast(code); | |
| 853 } | |
| 854 | |
| 855 | |
| 856 void RegExpMacroAssemblerARM::GoTo(Label* to) { | |
| 857 BranchOrBacktrack(al, to); | |
| 858 } | |
| 859 | |
| 860 | |
| 861 void RegExpMacroAssemblerARM::IfRegisterGE(int reg, | |
| 862 int comparand, | |
| 863 Label* if_ge) { | |
| 864 __ ldr(r0, register_location(reg)); | |
| 865 __ cmp(r0, Operand(comparand)); | |
| 866 BranchOrBacktrack(ge, if_ge); | |
| 867 } | |
| 868 | |
| 869 | |
| 870 void RegExpMacroAssemblerARM::IfRegisterLT(int reg, | |
| 871 int comparand, | |
| 872 Label* if_lt) { | |
| 873 __ ldr(r0, register_location(reg)); | |
| 874 __ cmp(r0, Operand(comparand)); | |
| 875 BranchOrBacktrack(lt, if_lt); | |
| 876 } | |
| 877 | |
| 878 | |
| 879 void RegExpMacroAssemblerARM::IfRegisterEqPos(int reg, | |
| 880 Label* if_eq) { | |
| 881 __ ldr(r0, register_location(reg)); | |
| 882 __ cmp(r0, Operand(current_input_offset())); | |
| 883 BranchOrBacktrack(eq, if_eq); | |
| 884 } | |
| 885 | |
| 886 | |
| 887 RegExpMacroAssembler::IrregexpImplementation | |
| 888 RegExpMacroAssemblerARM::Implementation() { | |
| 889 return kARMImplementation; | |
| 890 } | |
| 891 | |
| 892 | |
| 893 void RegExpMacroAssemblerARM::LoadCurrentCharacter(int cp_offset, | |
| 894 Label* on_end_of_input, | |
| 895 bool check_bounds, | |
| 896 int characters) { | |
| 897 DCHECK(cp_offset >= -1); // ^ and \b can look behind one character. | |
| 898 DCHECK(cp_offset < (1<<30)); // Be sane! (And ensure negation works) | |
| 899 if (check_bounds) { | |
| 900 CheckPosition(cp_offset + characters - 1, on_end_of_input); | |
| 901 } | |
| 902 LoadCurrentCharacterUnchecked(cp_offset, characters); | |
| 903 } | |
| 904 | |
| 905 | |
| 906 void RegExpMacroAssemblerARM::PopCurrentPosition() { | |
| 907 Pop(current_input_offset()); | |
| 908 } | |
| 909 | |
| 910 | |
| 911 void RegExpMacroAssemblerARM::PopRegister(int register_index) { | |
| 912 Pop(r0); | |
| 913 __ str(r0, register_location(register_index)); | |
| 914 } | |
| 915 | |
| 916 | |
| 917 void RegExpMacroAssemblerARM::PushBacktrack(Label* label) { | |
| 918 __ mov_label_offset(r0, label); | |
| 919 Push(r0); | |
| 920 CheckStackLimit(); | |
| 921 } | |
| 922 | |
| 923 | |
| 924 void RegExpMacroAssemblerARM::PushCurrentPosition() { | |
| 925 Push(current_input_offset()); | |
| 926 } | |
| 927 | |
| 928 | |
| 929 void RegExpMacroAssemblerARM::PushRegister(int register_index, | |
| 930 StackCheckFlag check_stack_limit) { | |
| 931 __ ldr(r0, register_location(register_index)); | |
| 932 Push(r0); | |
| 933 if (check_stack_limit) CheckStackLimit(); | |
| 934 } | |
| 935 | |
| 936 | |
| 937 void RegExpMacroAssemblerARM::ReadCurrentPositionFromRegister(int reg) { | |
| 938 __ ldr(current_input_offset(), register_location(reg)); | |
| 939 } | |
| 940 | |
| 941 | |
| 942 void RegExpMacroAssemblerARM::ReadStackPointerFromRegister(int reg) { | |
| 943 __ ldr(backtrack_stackpointer(), register_location(reg)); | |
| 944 __ ldr(r0, MemOperand(frame_pointer(), kStackHighEnd)); | |
| 945 __ add(backtrack_stackpointer(), backtrack_stackpointer(), Operand(r0)); | |
| 946 } | |
| 947 | |
| 948 | |
| 949 void RegExpMacroAssemblerARM::SetCurrentPositionFromEnd(int by) { | |
| 950 Label after_position; | |
| 951 __ cmp(current_input_offset(), Operand(-by * char_size())); | |
| 952 __ b(ge, &after_position); | |
| 953 __ mov(current_input_offset(), Operand(-by * char_size())); | |
| 954 // On RegExp code entry (where this operation is used), the character before | |
| 955 // the current position is expected to be already loaded. | |
| 956 // We have advanced the position, so it's safe to read backwards. | |
| 957 LoadCurrentCharacterUnchecked(-1, 1); | |
| 958 __ bind(&after_position); | |
| 959 } | |
| 960 | |
| 961 | |
| 962 void RegExpMacroAssemblerARM::SetRegister(int register_index, int to) { | |
| 963 DCHECK(register_index >= num_saved_registers_); // Reserved for positions! | |
| 964 __ mov(r0, Operand(to)); | |
| 965 __ str(r0, register_location(register_index)); | |
| 966 } | |
| 967 | |
| 968 | |
| 969 bool RegExpMacroAssemblerARM::Succeed() { | |
| 970 __ jmp(&success_label_); | |
| 971 return global(); | |
| 972 } | |
| 973 | |
| 974 | |
| 975 void RegExpMacroAssemblerARM::WriteCurrentPositionToRegister(int reg, | |
| 976 int cp_offset) { | |
| 977 if (cp_offset == 0) { | |
| 978 __ str(current_input_offset(), register_location(reg)); | |
| 979 } else { | |
| 980 __ add(r0, current_input_offset(), Operand(cp_offset * char_size())); | |
| 981 __ str(r0, register_location(reg)); | |
| 982 } | |
| 983 } | |
| 984 | |
| 985 | |
| 986 void RegExpMacroAssemblerARM::ClearRegisters(int reg_from, int reg_to) { | |
| 987 DCHECK(reg_from <= reg_to); | |
| 988 __ ldr(r0, MemOperand(frame_pointer(), kInputStartMinusOne)); | |
| 989 for (int reg = reg_from; reg <= reg_to; reg++) { | |
| 990 __ str(r0, register_location(reg)); | |
| 991 } | |
| 992 } | |
| 993 | |
| 994 | |
| 995 void RegExpMacroAssemblerARM::WriteStackPointerToRegister(int reg) { | |
| 996 __ ldr(r1, MemOperand(frame_pointer(), kStackHighEnd)); | |
| 997 __ sub(r0, backtrack_stackpointer(), r1); | |
| 998 __ str(r0, register_location(reg)); | |
| 999 } | |
| 1000 | |
| 1001 | |
| 1002 // Private methods: | |
| 1003 | |
| 1004 void RegExpMacroAssemblerARM::CallCheckStackGuardState(Register scratch) { | |
| 1005 __ PrepareCallCFunction(3, scratch); | |
| 1006 | |
| 1007 // RegExp code frame pointer. | |
| 1008 __ mov(r2, frame_pointer()); | |
| 1009 // Code* of self. | |
| 1010 __ mov(r1, Operand(masm_->CodeObject())); | |
| 1011 | |
| 1012 // We need to make room for the return address on the stack. | |
| 1013 int stack_alignment = base::OS::ActivationFrameAlignment(); | |
| 1014 DCHECK(IsAligned(stack_alignment, kPointerSize)); | |
| 1015 __ sub(sp, sp, Operand(stack_alignment)); | |
| 1016 | |
| 1017 // r0 will point to the return address, placed by DirectCEntry. | |
| 1018 __ mov(r0, sp); | |
| 1019 | |
| 1020 ExternalReference stack_guard_check = | |
| 1021 ExternalReference::re_check_stack_guard_state(isolate()); | |
| 1022 __ mov(ip, Operand(stack_guard_check)); | |
| 1023 DirectCEntryStub stub(isolate()); | |
| 1024 stub.GenerateCall(masm_, ip); | |
| 1025 | |
| 1026 // Drop the return address from the stack. | |
| 1027 __ add(sp, sp, Operand(stack_alignment)); | |
| 1028 | |
| 1029 DCHECK(stack_alignment != 0); | |
| 1030 __ ldr(sp, MemOperand(sp, 0)); | |
| 1031 | |
| 1032 __ mov(code_pointer(), Operand(masm_->CodeObject())); | |
| 1033 } | |
| 1034 | |
| 1035 | |
| 1036 // Helper function for reading a value out of a stack frame. | |
| 1037 template <typename T> | |
| 1038 static T& frame_entry(Address re_frame, int frame_offset) { | |
| 1039 return reinterpret_cast<T&>(Memory::int32_at(re_frame + frame_offset)); | |
| 1040 } | |
| 1041 | |
| 1042 | |
| 1043 template <typename T> | |
| 1044 static T* frame_entry_address(Address re_frame, int frame_offset) { | |
| 1045 return reinterpret_cast<T*>(re_frame + frame_offset); | |
| 1046 } | |
| 1047 | |
| 1048 | |
| 1049 int RegExpMacroAssemblerARM::CheckStackGuardState(Address* return_address, | |
| 1050 Code* re_code, | |
| 1051 Address re_frame) { | |
| 1052 return NativeRegExpMacroAssembler::CheckStackGuardState( | |
| 1053 frame_entry<Isolate*>(re_frame, kIsolate), | |
| 1054 frame_entry<int>(re_frame, kStartIndex), | |
| 1055 frame_entry<int>(re_frame, kDirectCall) == 1, return_address, re_code, | |
| 1056 frame_entry_address<String*>(re_frame, kInputString), | |
| 1057 frame_entry_address<const byte*>(re_frame, kInputStart), | |
| 1058 frame_entry_address<const byte*>(re_frame, kInputEnd)); | |
| 1059 } | |
| 1060 | |
| 1061 | |
| 1062 MemOperand RegExpMacroAssemblerARM::register_location(int register_index) { | |
| 1063 DCHECK(register_index < (1<<30)); | |
| 1064 if (num_registers_ <= register_index) { | |
| 1065 num_registers_ = register_index + 1; | |
| 1066 } | |
| 1067 return MemOperand(frame_pointer(), | |
| 1068 kRegisterZero - register_index * kPointerSize); | |
| 1069 } | |
| 1070 | |
| 1071 | |
| 1072 void RegExpMacroAssemblerARM::CheckPosition(int cp_offset, | |
| 1073 Label* on_outside_input) { | |
| 1074 __ cmp(current_input_offset(), Operand(-cp_offset * char_size())); | |
| 1075 BranchOrBacktrack(ge, on_outside_input); | |
| 1076 } | |
| 1077 | |
| 1078 | |
| 1079 void RegExpMacroAssemblerARM::BranchOrBacktrack(Condition condition, | |
| 1080 Label* to) { | |
| 1081 if (condition == al) { // Unconditional. | |
| 1082 if (to == NULL) { | |
| 1083 Backtrack(); | |
| 1084 return; | |
| 1085 } | |
| 1086 __ jmp(to); | |
| 1087 return; | |
| 1088 } | |
| 1089 if (to == NULL) { | |
| 1090 __ b(condition, &backtrack_label_); | |
| 1091 return; | |
| 1092 } | |
| 1093 __ b(condition, to); | |
| 1094 } | |
| 1095 | |
| 1096 | |
| 1097 void RegExpMacroAssemblerARM::SafeCall(Label* to, Condition cond) { | |
| 1098 __ bl(to, cond); | |
| 1099 } | |
| 1100 | |
| 1101 | |
| 1102 void RegExpMacroAssemblerARM::SafeReturn() { | |
| 1103 __ pop(lr); | |
| 1104 __ add(pc, lr, Operand(masm_->CodeObject())); | |
| 1105 } | |
| 1106 | |
| 1107 | |
| 1108 void RegExpMacroAssemblerARM::SafeCallTarget(Label* name) { | |
| 1109 __ bind(name); | |
| 1110 __ sub(lr, lr, Operand(masm_->CodeObject())); | |
| 1111 __ push(lr); | |
| 1112 } | |
| 1113 | |
| 1114 | |
| 1115 void RegExpMacroAssemblerARM::Push(Register source) { | |
| 1116 DCHECK(!source.is(backtrack_stackpointer())); | |
| 1117 __ str(source, | |
| 1118 MemOperand(backtrack_stackpointer(), kPointerSize, NegPreIndex)); | |
| 1119 } | |
| 1120 | |
| 1121 | |
| 1122 void RegExpMacroAssemblerARM::Pop(Register target) { | |
| 1123 DCHECK(!target.is(backtrack_stackpointer())); | |
| 1124 __ ldr(target, | |
| 1125 MemOperand(backtrack_stackpointer(), kPointerSize, PostIndex)); | |
| 1126 } | |
| 1127 | |
| 1128 | |
| 1129 void RegExpMacroAssemblerARM::CheckPreemption() { | |
| 1130 // Check for preemption. | |
| 1131 ExternalReference stack_limit = | |
| 1132 ExternalReference::address_of_stack_limit(isolate()); | |
| 1133 __ mov(r0, Operand(stack_limit)); | |
| 1134 __ ldr(r0, MemOperand(r0)); | |
| 1135 __ cmp(sp, r0); | |
| 1136 SafeCall(&check_preempt_label_, ls); | |
| 1137 } | |
| 1138 | |
| 1139 | |
| 1140 void RegExpMacroAssemblerARM::CheckStackLimit() { | |
| 1141 ExternalReference stack_limit = | |
| 1142 ExternalReference::address_of_regexp_stack_limit(isolate()); | |
| 1143 __ mov(r0, Operand(stack_limit)); | |
| 1144 __ ldr(r0, MemOperand(r0)); | |
| 1145 __ cmp(backtrack_stackpointer(), Operand(r0)); | |
| 1146 SafeCall(&stack_overflow_label_, ls); | |
| 1147 } | |
| 1148 | |
| 1149 | |
| 1150 bool RegExpMacroAssemblerARM::CanReadUnaligned() { | |
| 1151 return CpuFeatures::IsSupported(UNALIGNED_ACCESSES) && !slow_safe(); | |
| 1152 } | |
| 1153 | |
| 1154 | |
| 1155 void RegExpMacroAssemblerARM::LoadCurrentCharacterUnchecked(int cp_offset, | |
| 1156 int characters) { | |
| 1157 Register offset = current_input_offset(); | |
| 1158 if (cp_offset != 0) { | |
| 1159 // r4 is not being used to store the capture start index at this point. | |
| 1160 __ add(r4, current_input_offset(), Operand(cp_offset * char_size())); | |
| 1161 offset = r4; | |
| 1162 } | |
| 1163 // The ldr, str, ldrh, strh instructions can do unaligned accesses, if the CPU | |
| 1164 // and the operating system running on the target allow it. | |
| 1165 // If unaligned load/stores are not supported then this function must only | |
| 1166 // be used to load a single character at a time. | |
| 1167 if (!CanReadUnaligned()) { | |
| 1168 DCHECK(characters == 1); | |
| 1169 } | |
| 1170 | |
| 1171 if (mode_ == LATIN1) { | |
| 1172 if (characters == 4) { | |
| 1173 __ ldr(current_character(), MemOperand(end_of_input_address(), offset)); | |
| 1174 } else if (characters == 2) { | |
| 1175 __ ldrh(current_character(), MemOperand(end_of_input_address(), offset)); | |
| 1176 } else { | |
| 1177 DCHECK(characters == 1); | |
| 1178 __ ldrb(current_character(), MemOperand(end_of_input_address(), offset)); | |
| 1179 } | |
| 1180 } else { | |
| 1181 DCHECK(mode_ == UC16); | |
| 1182 if (characters == 2) { | |
| 1183 __ ldr(current_character(), MemOperand(end_of_input_address(), offset)); | |
| 1184 } else { | |
| 1185 DCHECK(characters == 1); | |
| 1186 __ ldrh(current_character(), MemOperand(end_of_input_address(), offset)); | |
| 1187 } | |
| 1188 } | |
| 1189 } | |
| 1190 | |
| 1191 | |
| 1192 #undef __ | |
| 1193 | |
| 1194 #endif // V8_INTERPRETED_REGEXP | |
| 1195 | |
| 1196 } // namespace internal | |
| 1197 } // namespace v8 | |
| 1198 | |
| 1199 #endif // V8_TARGET_ARCH_ARM | |
| 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