1//===- MILexer.cpp - Machine instructions lexer implementation ------------===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This file implements the lexing of machine instructions.
10//
11//===----------------------------------------------------------------------===//
12
13#include "MILexer.h"
14#include "llvm/ADT/StringExtras.h"
15#include "llvm/ADT/StringSwitch.h"
16#include "llvm/ADT/Twine.h"
17#include <cassert>
18#include <cctype>
19#include <string>
20
21using namespace llvm;
22
23namespace {
24
25using ErrorCallbackType =
26 function_ref<void(StringRef::iterator Loc, const Twine &)>;
27
28/// This class provides a way to iterate and get characters from the source
29/// string.
30class Cursor {
31 const char *Ptr = nullptr;
32 const char *End = nullptr;
33
34public:
35 Cursor(std::nullopt_t) {}
36
37 explicit Cursor(StringRef Str) {
38 Ptr = Str.data();
39 End = Ptr + Str.size();
40 }
41
42 bool isEOF() const { return Ptr == End; }
43
44 char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
45
46 void advance(unsigned I = 1) { Ptr += I; }
47
48 StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
49
50 StringRef upto(Cursor C) const {
51 assert(C.Ptr >= Ptr && C.Ptr <= End);
52 return StringRef(Ptr, C.Ptr - Ptr);
53 }
54
55 StringRef::iterator location() const { return Ptr; }
56
57 operator bool() const { return Ptr != nullptr; }
58};
59
60} // end anonymous namespace
61
62MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
63 this->Kind = Kind;
64 this->Range = Range;
65 return *this;
66}
67
68MIToken &MIToken::setStringValue(StringRef StrVal) {
69 StringValue = StrVal;
70 return *this;
71}
72
73MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
74 StringValueStorage = std::move(StrVal);
75 StringValue = StringValueStorage;
76 return *this;
77}
78
79MIToken &MIToken::setIntegerValue(APSInt IntVal) {
80 this->IntVal = std::move(IntVal);
81 return *this;
82}
83
84/// Skip the leading whitespace characters and return the updated cursor.
85static Cursor skipWhitespace(Cursor C) {
86 while (isblank(C.peek()))
87 C.advance();
88 return C;
89}
90
91static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
92
93/// Skip a line comment and return the updated cursor.
94static Cursor skipComment(Cursor C) {
95 if (C.peek() != ';')
96 return C;
97 while (!isNewlineChar(C: C.peek()) && !C.isEOF())
98 C.advance();
99 return C;
100}
101
102/// Machine operands can have comments, enclosed between /* and */.
103/// This eats up all tokens, including /* and */.
104static Cursor skipMachineOperandComment(Cursor C) {
105 if (C.peek() != '/' || C.peek(I: 1) != '*')
106 return C;
107
108 while (C.peek() != '*' || C.peek(I: 1) != '/')
109 C.advance();
110
111 C.advance();
112 C.advance();
113 return C;
114}
115
116/// Return true if the given character satisfies the following regular
117/// expression: [-a-zA-Z$._0-9]
118static bool isIdentifierChar(char C) {
119 return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
120 C == '$';
121}
122
123/// Unescapes the given string value.
124///
125/// Expects the string value to be quoted.
126static std::string unescapeQuotedString(StringRef Value) {
127 assert(Value.front() == '"' && Value.back() == '"');
128 Cursor C = Cursor(Value.substr(Start: 1, N: Value.size() - 2));
129
130 std::string Str;
131 Str.reserve(res_arg: C.remaining().size());
132 while (!C.isEOF()) {
133 char Char = C.peek();
134 if (Char == '\\') {
135 if (C.peek(I: 1) == '\\') {
136 // Two '\' become one
137 Str += '\\';
138 C.advance(I: 2);
139 continue;
140 }
141 if (isxdigit(C.peek(I: 1)) && isxdigit(C.peek(I: 2))) {
142 Str += hexDigitValue(C: C.peek(I: 1)) * 16 + hexDigitValue(C: C.peek(I: 2));
143 C.advance(I: 3);
144 continue;
145 }
146 }
147 Str += Char;
148 C.advance();
149 }
150 return Str;
151}
152
153/// Lex a string constant using the following regular expression: \"[^\"]*\"
154static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
155 assert(C.peek() == '"');
156 for (C.advance(); C.peek() != '"'; C.advance()) {
157 if (C.isEOF() || isNewlineChar(C: C.peek())) {
158 ErrorCallback(
159 C.location(),
160 "end of machine instruction reached before the closing '\"'");
161 return std::nullopt;
162 }
163 }
164 C.advance();
165 return C;
166}
167
168static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
169 unsigned PrefixLength, ErrorCallbackType ErrorCallback) {
170 auto Range = C;
171 C.advance(I: PrefixLength);
172 if (C.peek() == '"') {
173 if (Cursor R = lexStringConstant(C, ErrorCallback)) {
174 StringRef String = Range.upto(C: R);
175 Token.reset(Kind: Type, Range: String)
176 .setOwnedStringValue(
177 unescapeQuotedString(Value: String.drop_front(N: PrefixLength)));
178 return R;
179 }
180 Token.reset(Kind: MIToken::Error, Range: Range.remaining());
181 return Range;
182 }
183 while (isIdentifierChar(C: C.peek()))
184 C.advance();
185 Token.reset(Kind: Type, Range: Range.upto(C))
186 .setStringValue(Range.upto(C).drop_front(N: PrefixLength));
187 return C;
188}
189
190static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
191 return StringSwitch<MIToken::TokenKind>(Identifier)
192 .Case(S: "_", Value: MIToken::underscore)
193 .Case(S: "implicit", Value: MIToken::kw_implicit)
194 .Case(S: "implicit-def", Value: MIToken::kw_implicit_define)
195 .Case(S: "def", Value: MIToken::kw_def)
196 .Case(S: "dead", Value: MIToken::kw_dead)
197 .Case(S: "killed", Value: MIToken::kw_killed)
198 .Case(S: "undef", Value: MIToken::kw_undef)
199 .Case(S: "internal", Value: MIToken::kw_internal)
200 .Case(S: "early-clobber", Value: MIToken::kw_early_clobber)
201 .Case(S: "debug-use", Value: MIToken::kw_debug_use)
202 .Case(S: "renamable", Value: MIToken::kw_renamable)
203 .Case(S: "tied-def", Value: MIToken::kw_tied_def)
204 .Case(S: "frame-setup", Value: MIToken::kw_frame_setup)
205 .Case(S: "frame-destroy", Value: MIToken::kw_frame_destroy)
206 .Case(S: "nnan", Value: MIToken::kw_nnan)
207 .Case(S: "ninf", Value: MIToken::kw_ninf)
208 .Case(S: "nsz", Value: MIToken::kw_nsz)
209 .Case(S: "arcp", Value: MIToken::kw_arcp)
210 .Case(S: "contract", Value: MIToken::kw_contract)
211 .Case(S: "afn", Value: MIToken::kw_afn)
212 .Case(S: "reassoc", Value: MIToken::kw_reassoc)
213 .Case(S: "nuw", Value: MIToken::kw_nuw)
214 .Case(S: "nsw", Value: MIToken::kw_nsw)
215 .Case(S: "nusw", Value: MIToken::kw_nusw)
216 .Case(S: "exact", Value: MIToken::kw_exact)
217 .Case(S: "nneg", Value: MIToken::kw_nneg)
218 .Case(S: "disjoint", Value: MIToken::kw_disjoint)
219 .Case(S: "samesign", Value: MIToken::kw_samesign)
220 .Case(S: "inbounds", Value: MIToken::kw_inbounds)
221 .Case(S: "nonnull", Value: MIToken::kw_nonnull)
222 .Case(S: "nofpexcept", Value: MIToken::kw_nofpexcept)
223 .Case(S: "unpredictable", Value: MIToken::kw_unpredictable)
224 .Case(S: "debug-location", Value: MIToken::kw_debug_location)
225 .Case(S: "debug-instr-number", Value: MIToken::kw_debug_instr_number)
226 .Case(S: "dbg-instr-ref", Value: MIToken::kw_dbg_instr_ref)
227 .Case(S: "same_value", Value: MIToken::kw_cfi_same_value)
228 .Case(S: "offset", Value: MIToken::kw_cfi_offset)
229 .Case(S: "rel_offset", Value: MIToken::kw_cfi_rel_offset)
230 .Case(S: "def_cfa_register", Value: MIToken::kw_cfi_def_cfa_register)
231 .Case(S: "def_cfa_offset", Value: MIToken::kw_cfi_def_cfa_offset)
232 .Case(S: "adjust_cfa_offset", Value: MIToken::kw_cfi_adjust_cfa_offset)
233 .Case(S: "escape", Value: MIToken::kw_cfi_escape)
234 .Case(S: "def_cfa", Value: MIToken::kw_cfi_def_cfa)
235 .Case(S: "llvm_def_aspace_cfa", Value: MIToken::kw_cfi_llvm_def_aspace_cfa)
236 .Case(S: "remember_state", Value: MIToken::kw_cfi_remember_state)
237 .Case(S: "restore", Value: MIToken::kw_cfi_restore)
238 .Case(S: "restore_state", Value: MIToken::kw_cfi_restore_state)
239 .Case(S: "undefined", Value: MIToken::kw_cfi_undefined)
240 .Case(S: "register", Value: MIToken::kw_cfi_register)
241 .Case(S: "window_save", Value: MIToken::kw_cfi_window_save)
242 .Case(S: "negate_ra_sign_state",
243 Value: MIToken::kw_cfi_aarch64_negate_ra_sign_state)
244 .Case(S: "negate_ra_sign_state_with_pc",
245 Value: MIToken::kw_cfi_aarch64_negate_ra_sign_state_with_pc)
246 .Case(S: "llvm_set_ra_state", Value: MIToken::kw_cfi_set_ra_state)
247 .Case(S: "llvm_register_pair", Value: MIToken::kw_cfi_llvm_register_pair)
248 .Case(S: "llvm_vector_registers", Value: MIToken::kw_cfi_llvm_vector_registers)
249 .Case(S: "llvm_vector_offset", Value: MIToken::kw_cfi_llvm_vector_offset)
250 .Case(S: "llvm_vector_register_mask",
251 Value: MIToken::kw_cfi_llvm_vector_register_mask)
252 .Case(S: "blockaddress", Value: MIToken::kw_blockaddress)
253 .Case(S: "intrinsic", Value: MIToken::kw_intrinsic)
254 .Case(S: "target-index", Value: MIToken::kw_target_index)
255 .Case(S: "half", Value: MIToken::kw_half)
256 .Case(S: "bfloat", Value: MIToken::kw_bfloat)
257 .Case(S: "float", Value: MIToken::kw_float)
258 .Case(S: "double", Value: MIToken::kw_double)
259 .Case(S: "x86_fp80", Value: MIToken::kw_x86_fp80)
260 .Case(S: "fp128", Value: MIToken::kw_fp128)
261 .Case(S: "ppc_fp128", Value: MIToken::kw_ppc_fp128)
262 .Case(S: "target-flags", Value: MIToken::kw_target_flags)
263 .Case(S: "volatile", Value: MIToken::kw_volatile)
264 .Case(S: "non-temporal", Value: MIToken::kw_non_temporal)
265 .Case(S: "dereferenceable", Value: MIToken::kw_dereferenceable)
266 .Case(S: "invariant", Value: MIToken::kw_invariant)
267 .Case(S: "align", Value: MIToken::kw_align)
268 .Case(S: "basealign", Value: MIToken::kw_basealign)
269 .Case(S: "addrspace", Value: MIToken::kw_addrspace)
270 .Case(S: "stack", Value: MIToken::kw_stack)
271 .Case(S: "got", Value: MIToken::kw_got)
272 .Case(S: "jump-table", Value: MIToken::kw_jump_table)
273 .Case(S: "constant-pool", Value: MIToken::kw_constant_pool)
274 .Case(S: "call-entry", Value: MIToken::kw_call_entry)
275 .Case(S: "custom", Value: MIToken::kw_custom)
276 .Case(S: "lanemask", Value: MIToken::kw_lanemask)
277 .Case(S: "liveout", Value: MIToken::kw_liveout)
278 .Case(S: "landing-pad", Value: MIToken::kw_landing_pad)
279 .Case(S: "inlineasm-br-indirect-target",
280 Value: MIToken::kw_inlineasm_br_indirect_target)
281 .Case(S: "ehscope-entry", Value: MIToken::kw_ehscope_entry)
282 .Case(S: "ehfunclet-entry", Value: MIToken::kw_ehfunclet_entry)
283 .Case(S: "cleanup-funclet-entry", Value: MIToken::kw_cleanup_funclet_entry)
284 .Case(S: "ehcont-target", Value: MIToken::kw_ehcont_target)
285 .Case(S: "liveins", Value: MIToken::kw_liveins)
286 .Case(S: "successors", Value: MIToken::kw_successors)
287 .Case(S: "floatpred", Value: MIToken::kw_floatpred)
288 .Case(S: "intpred", Value: MIToken::kw_intpred)
289 .Case(S: "shufflemask", Value: MIToken::kw_shufflemask)
290 .Case(S: "pre-instr-symbol", Value: MIToken::kw_pre_instr_symbol)
291 .Case(S: "post-instr-symbol", Value: MIToken::kw_post_instr_symbol)
292 .Case(S: "heap-alloc-marker", Value: MIToken::kw_heap_alloc_marker)
293 .Case(S: "pcsections", Value: MIToken::kw_pcsections)
294 .Case(S: "cfi-type", Value: MIToken::kw_cfi_type)
295 .Case(S: "deactivation-symbol", Value: MIToken::kw_deactivation_symbol)
296 .Case(S: "bbsections", Value: MIToken::kw_bbsections)
297 .Case(S: "bb_id", Value: MIToken::kw_bb_id)
298 .Case(S: "unknown-size", Value: MIToken::kw_unknown_size)
299 .Case(S: "unknown-address", Value: MIToken::kw_unknown_address)
300 .Case(S: "distinct", Value: MIToken::kw_distinct)
301 .Case(S: "ir-block-address-taken", Value: MIToken::kw_ir_block_address_taken)
302 .Case(S: "machine-block-address-taken",
303 Value: MIToken::kw_machine_block_address_taken)
304 .Case(S: "call-frame-size", Value: MIToken::kw_call_frame_size)
305 .Case(S: "max-bytes-for-alignment", Value: MIToken::kw_max_bytes_for_alignment)
306 .Case(S: "noconvergent", Value: MIToken::kw_noconvergent)
307 .Case(S: "mmra", Value: MIToken::kw_mmra)
308 .Case(S: "lr-split", Value: MIToken::kw_lr_split)
309 .Default(Value: MIToken::Identifier);
310}
311
312static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
313 if (!isalpha(C.peek()) && C.peek() != '_')
314 return std::nullopt;
315 auto Range = C;
316 while (isIdentifierChar(C: C.peek()))
317 C.advance();
318 auto Identifier = Range.upto(C);
319 Token.reset(Kind: getIdentifierKind(Identifier), Range: Identifier)
320 .setStringValue(Identifier);
321 return C;
322}
323
324static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
325 ErrorCallbackType ErrorCallback) {
326 bool IsReference = C.remaining().starts_with(Prefix: "%bb.");
327 if (!IsReference && !C.remaining().starts_with(Prefix: "bb."))
328 return std::nullopt;
329 auto Range = C;
330 unsigned PrefixLength = IsReference ? 4 : 3;
331 C.advance(I: PrefixLength); // Skip '%bb.' or 'bb.'
332 if (!isdigit(C.peek())) {
333 Token.reset(Kind: MIToken::Error, Range: C.remaining());
334 ErrorCallback(C.location(), "expected a number after '%bb.'");
335 return C;
336 }
337 auto NumberRange = C;
338 while (isdigit(C.peek()))
339 C.advance();
340 StringRef Number = NumberRange.upto(C);
341 unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
342 // TODO: The format bb.<id>.<irname> is supported only when it's not a
343 // reference. Once we deprecate the format where the irname shows up, we
344 // should only lex forward if it is a reference.
345 if (C.peek() == '.') {
346 C.advance(); // Skip '.'
347 ++StringOffset;
348 // The name is quoted if it is not a plain identifier.
349 if (C.peek() == '"') {
350 Cursor R = lexStringConstant(C, ErrorCallback);
351 if (!R) {
352 ErrorCallback(C.location(),
353 "unable to parse quoted string from opening quote");
354 Token.reset(Kind: MIToken::Error, Range: Range.remaining());
355 return Range;
356 }
357 MIToken::TokenKind Kind = IsReference ? MIToken::MachineBasicBlock
358 : MIToken::MachineBasicBlockLabel;
359 Token.reset(Kind, Range: Range.upto(C: R))
360 .setIntegerValue(APSInt(Number))
361 .setOwnedStringValue(
362 unescapeQuotedString(Value: Range.upto(C: R).drop_front(N: StringOffset)));
363 return R;
364 }
365 while (isIdentifierChar(C: C.peek()))
366 C.advance();
367 }
368 Token.reset(Kind: IsReference ? MIToken::MachineBasicBlock
369 : MIToken::MachineBasicBlockLabel,
370 Range: Range.upto(C))
371 .setIntegerValue(APSInt(Number))
372 .setStringValue(Range.upto(C).drop_front(N: StringOffset));
373 return C;
374}
375
376static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
377 MIToken::TokenKind Kind) {
378 if (!C.remaining().starts_with(Prefix: Rule) || !isdigit(C.peek(I: Rule.size())))
379 return std::nullopt;
380 auto Range = C;
381 C.advance(I: Rule.size());
382 auto NumberRange = C;
383 while (isdigit(C.peek()))
384 C.advance();
385 Token.reset(Kind, Range: Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
386 return C;
387}
388
389static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
390 MIToken::TokenKind Kind) {
391 if (!C.remaining().starts_with(Prefix: Rule) || !isdigit(C.peek(I: Rule.size())))
392 return std::nullopt;
393 auto Range = C;
394 C.advance(I: Rule.size());
395 auto NumberRange = C;
396 while (isdigit(C.peek()))
397 C.advance();
398 StringRef Number = NumberRange.upto(C);
399 unsigned StringOffset = Rule.size() + Number.size();
400 if (C.peek() == '.') {
401 C.advance();
402 ++StringOffset;
403 while (isIdentifierChar(C: C.peek()))
404 C.advance();
405 }
406 Token.reset(Kind, Range: Range.upto(C))
407 .setIntegerValue(APSInt(Number))
408 .setStringValue(Range.upto(C).drop_front(N: StringOffset));
409 return C;
410}
411
412static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
413 return maybeLexIndex(C, Token, Rule: "%jump-table.", Kind: MIToken::JumpTableIndex);
414}
415
416static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
417 return maybeLexIndexAndName(C, Token, Rule: "%stack.", Kind: MIToken::StackObject);
418}
419
420static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
421 return maybeLexIndex(C, Token, Rule: "%fixed-stack.", Kind: MIToken::FixedStackObject);
422}
423
424static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
425 return maybeLexIndex(C, Token, Rule: "%const.", Kind: MIToken::ConstantPoolItem);
426}
427
428static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token,
429 ErrorCallbackType ErrorCallback) {
430 const StringRef Rule = "%subreg.";
431 if (!C.remaining().starts_with(Prefix: Rule))
432 return std::nullopt;
433 return lexName(C, Token, Type: MIToken::SubRegisterIndex, PrefixLength: Rule.size(),
434 ErrorCallback);
435}
436
437static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
438 ErrorCallbackType ErrorCallback) {
439 const StringRef Rule = "%ir-block.";
440 if (!C.remaining().starts_with(Prefix: Rule))
441 return std::nullopt;
442 if (isdigit(C.peek(I: Rule.size())))
443 return maybeLexIndex(C, Token, Rule, Kind: MIToken::IRBlock);
444 return lexName(C, Token, Type: MIToken::NamedIRBlock, PrefixLength: Rule.size(), ErrorCallback);
445}
446
447static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
448 ErrorCallbackType ErrorCallback) {
449 const StringRef Rule = "%ir.";
450 if (!C.remaining().starts_with(Prefix: Rule))
451 return std::nullopt;
452 if (isdigit(C.peek(I: Rule.size())))
453 return maybeLexIndex(C, Token, Rule, Kind: MIToken::IRValue);
454 return lexName(C, Token, Type: MIToken::NamedIRValue, PrefixLength: Rule.size(), ErrorCallback);
455}
456
457static Cursor maybeLexStringConstant(Cursor C, MIToken &Token,
458 ErrorCallbackType ErrorCallback) {
459 if (C.peek() != '"')
460 return std::nullopt;
461 return lexName(C, Token, Type: MIToken::StringConstant, /*PrefixLength=*/0,
462 ErrorCallback);
463}
464
465static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
466 auto Range = C;
467 C.advance(); // Skip '%'
468 auto NumberRange = C;
469 while (isdigit(C.peek()))
470 C.advance();
471 Token.reset(Kind: MIToken::VirtualRegister, Range: Range.upto(C))
472 .setIntegerValue(APSInt(NumberRange.upto(C)));
473 return C;
474}
475
476/// Returns true for a character allowed in a register name.
477static bool isRegisterChar(char C) {
478 return isIdentifierChar(C) && C != '.';
479}
480
481static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) {
482 Cursor Range = C;
483 C.advance(); // Skip '%'
484 while (isRegisterChar(C: C.peek()))
485 C.advance();
486 Token.reset(Kind: MIToken::NamedVirtualRegister, Range: Range.upto(C))
487 .setStringValue(Range.upto(C).drop_front(N: 1)); // Drop the '%'
488 return C;
489}
490
491static Cursor maybeLexRegister(Cursor C, MIToken &Token,
492 ErrorCallbackType ErrorCallback) {
493 if (C.peek() != '%' && C.peek() != '$')
494 return std::nullopt;
495
496 if (C.peek() == '%') {
497 if (isdigit(C.peek(I: 1)))
498 return lexVirtualRegister(C, Token);
499
500 if (isRegisterChar(C: C.peek(I: 1)))
501 return lexNamedVirtualRegister(C, Token);
502
503 return std::nullopt;
504 }
505
506 assert(C.peek() == '$');
507 auto Range = C;
508 C.advance(); // Skip '$'
509 while (isRegisterChar(C: C.peek()))
510 C.advance();
511 Token.reset(Kind: MIToken::NamedRegister, Range: Range.upto(C))
512 .setStringValue(Range.upto(C).drop_front(N: 1)); // Drop the '$'
513 return C;
514}
515
516static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
517 ErrorCallbackType ErrorCallback) {
518 if (C.peek() != '@')
519 return std::nullopt;
520 if (!isdigit(C.peek(I: 1)))
521 return lexName(C, Token, Type: MIToken::NamedGlobalValue, /*PrefixLength=*/1,
522 ErrorCallback);
523 auto Range = C;
524 C.advance(I: 1); // Skip the '@'
525 auto NumberRange = C;
526 while (isdigit(C.peek()))
527 C.advance();
528 Token.reset(Kind: MIToken::GlobalValue, Range: Range.upto(C))
529 .setIntegerValue(APSInt(NumberRange.upto(C)));
530 return C;
531}
532
533static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
534 ErrorCallbackType ErrorCallback) {
535 if (C.peek() != '&')
536 return std::nullopt;
537 return lexName(C, Token, Type: MIToken::ExternalSymbol, /*PrefixLength=*/1,
538 ErrorCallback);
539}
540
541static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token,
542 ErrorCallbackType ErrorCallback) {
543 const StringRef Rule = "<mcsymbol ";
544 if (!C.remaining().starts_with(Prefix: Rule))
545 return std::nullopt;
546 auto Start = C;
547 C.advance(I: Rule.size());
548
549 // Try a simple unquoted name.
550 if (C.peek() != '"') {
551 while (isIdentifierChar(C: C.peek()))
552 C.advance();
553 StringRef String = Start.upto(C).drop_front(N: Rule.size());
554 if (C.peek() != '>') {
555 ErrorCallback(C.location(),
556 "expected the '<mcsymbol ...' to be closed by a '>'");
557 Token.reset(Kind: MIToken::Error, Range: Start.remaining());
558 return Start;
559 }
560 C.advance();
561
562 Token.reset(Kind: MIToken::MCSymbol, Range: Start.upto(C)).setStringValue(String);
563 return C;
564 }
565
566 // Otherwise lex out a quoted name.
567 Cursor R = lexStringConstant(C, ErrorCallback);
568 if (!R) {
569 ErrorCallback(C.location(),
570 "unable to parse quoted string from opening quote");
571 Token.reset(Kind: MIToken::Error, Range: Start.remaining());
572 return Start;
573 }
574 StringRef String = Start.upto(C: R).drop_front(N: Rule.size());
575 if (R.peek() != '>') {
576 ErrorCallback(R.location(),
577 "expected the '<mcsymbol ...' to be closed by a '>'");
578 Token.reset(Kind: MIToken::Error, Range: Start.remaining());
579 return Start;
580 }
581 R.advance();
582
583 Token.reset(Kind: MIToken::MCSymbol, Range: Start.upto(C: R))
584 .setOwnedStringValue(unescapeQuotedString(Value: String));
585 return R;
586}
587
588static bool isValidHexFloatingPointPrefix(char C) {
589 return C == 'H' || C == 'K' || C == 'L' || C == 'M' || C == 'R';
590}
591
592static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
593 C.advance();
594 // Skip over [0-9]*([eE][-+]?[0-9]+)?
595 while (isdigit(C.peek()))
596 C.advance();
597 if ((C.peek() == 'e' || C.peek() == 'E') &&
598 (isdigit(C.peek(I: 1)) ||
599 ((C.peek(I: 1) == '-' || C.peek(I: 1) == '+') && isdigit(C.peek(I: 2))))) {
600 C.advance(I: 2);
601 while (isdigit(C.peek()))
602 C.advance();
603 }
604 Token.reset(Kind: MIToken::FloatingPointLiteral, Range: Range.upto(C));
605 return C;
606}
607
608static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) {
609 if (C.peek() != '0' || (C.peek(I: 1) != 'x' && C.peek(I: 1) != 'X'))
610 return std::nullopt;
611 Cursor Range = C;
612 C.advance(I: 2);
613 unsigned PrefLen = 2;
614 if (isValidHexFloatingPointPrefix(C: C.peek())) {
615 C.advance();
616 PrefLen++;
617 }
618 while (isxdigit(C.peek()))
619 C.advance();
620 StringRef StrVal = Range.upto(C);
621 if (StrVal.size() <= PrefLen)
622 return std::nullopt;
623 if (PrefLen == 2)
624 Token.reset(Kind: MIToken::HexLiteral, Range: Range.upto(C));
625 else // It must be 3, which means that there was a floating-point prefix.
626 Token.reset(Kind: MIToken::FloatingPointLiteral, Range: Range.upto(C));
627 return C;
628}
629
630static Cursor maybeLexFloatHexBits(Cursor C, MIToken &Token) {
631 if (C.peek() != 'f')
632 return std::nullopt;
633 if (C.peek(I: 1) != '0' || (C.peek(I: 2) != 'x' && C.peek(I: 2) != 'X'))
634 return std::nullopt;
635 Cursor Range = C;
636 C.advance(I: 3);
637 while (isxdigit(C.peek()))
638 C.advance();
639 StringRef StrVal = Range.upto(C);
640 if (StrVal.size() <= 3)
641 return std::nullopt;
642 Token.reset(Kind: MIToken::FloatingPointLiteral, Range: Range.upto(C));
643 return C;
644}
645
646static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
647 if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(I: 1))))
648 return std::nullopt;
649 auto Range = C;
650 C.advance();
651 while (isdigit(C.peek()))
652 C.advance();
653 if (C.peek() == '.')
654 return lexFloatingPointLiteral(Range, C, Token);
655 StringRef StrVal = Range.upto(C);
656 Token.reset(Kind: MIToken::IntegerLiteral, Range: StrVal).setIntegerValue(APSInt(StrVal));
657 return C;
658}
659
660static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) {
661 return StringSwitch<MIToken::TokenKind>(Identifier)
662 .Case(S: "!tbaa", Value: MIToken::md_tbaa)
663 .Case(S: "!alias.scope", Value: MIToken::md_alias_scope)
664 .Case(S: "!noalias", Value: MIToken::md_noalias)
665 .Case(S: "!range", Value: MIToken::md_range)
666 .Case(S: "!mem.cache_hint", Value: MIToken::md_mem_cache_hint)
667 .Case(S: "!DIExpression", Value: MIToken::md_diexpr)
668 .Case(S: "!DILocation", Value: MIToken::md_dilocation)
669 .Case(S: "!noalias.addrspace", Value: MIToken::md_noalias_addrspace)
670 .Default(Value: MIToken::Error);
671}
672
673static Cursor maybeLexExclaim(Cursor C, MIToken &Token,
674 ErrorCallbackType ErrorCallback) {
675 if (C.peek() != '!')
676 return std::nullopt;
677 auto Range = C;
678 C.advance(I: 1);
679 if (isdigit(C.peek()) || !isIdentifierChar(C: C.peek())) {
680 Token.reset(Kind: MIToken::exclaim, Range: Range.upto(C));
681 return C;
682 }
683 while (isIdentifierChar(C: C.peek()))
684 C.advance();
685 StringRef StrVal = Range.upto(C);
686 Token.reset(Kind: getMetadataKeywordKind(Identifier: StrVal), Range: StrVal);
687 if (Token.isError())
688 ErrorCallback(Token.location(),
689 "use of unknown metadata keyword '" + StrVal + "'");
690 return C;
691}
692
693static MIToken::TokenKind symbolToken(char C) {
694 switch (C) {
695 case ',':
696 return MIToken::comma;
697 case '.':
698 return MIToken::dot;
699 case '=':
700 return MIToken::equal;
701 case ':':
702 return MIToken::colon;
703 case '(':
704 return MIToken::lparen;
705 case ')':
706 return MIToken::rparen;
707 case '{':
708 return MIToken::lbrace;
709 case '}':
710 return MIToken::rbrace;
711 case '+':
712 return MIToken::plus;
713 case '-':
714 return MIToken::minus;
715 case '<':
716 return MIToken::less;
717 case '>':
718 return MIToken::greater;
719 default:
720 return MIToken::Error;
721 }
722}
723
724static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
725 MIToken::TokenKind Kind;
726 unsigned Length = 1;
727 if (C.peek() == ':' && C.peek(I: 1) == ':') {
728 Kind = MIToken::coloncolon;
729 Length = 2;
730 } else
731 Kind = symbolToken(C: C.peek());
732 if (Kind == MIToken::Error)
733 return std::nullopt;
734 auto Range = C;
735 C.advance(I: Length);
736 Token.reset(Kind, Range: Range.upto(C));
737 return C;
738}
739
740static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
741 if (!isNewlineChar(C: C.peek()))
742 return std::nullopt;
743 auto Range = C;
744 C.advance();
745 Token.reset(Kind: MIToken::Newline, Range: Range.upto(C));
746 return C;
747}
748
749static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
750 ErrorCallbackType ErrorCallback) {
751 if (C.peek() != '`')
752 return std::nullopt;
753 auto Range = C;
754 C.advance();
755 auto StrRange = C;
756 while (C.peek() != '`') {
757 if (C.isEOF() || isNewlineChar(C: C.peek())) {
758 ErrorCallback(
759 C.location(),
760 "end of machine instruction reached before the closing '`'");
761 Token.reset(Kind: MIToken::Error, Range: Range.remaining());
762 return C;
763 }
764 C.advance();
765 }
766 StringRef Value = StrRange.upto(C);
767 C.advance();
768 Token.reset(Kind: MIToken::QuotedIRValue, Range: Range.upto(C)).setStringValue(Value);
769 return C;
770}
771
772StringRef llvm::lexMIToken(StringRef Source, MIToken &Token,
773 ErrorCallbackType ErrorCallback) {
774 auto C = skipComment(C: skipWhitespace(C: Cursor(Source)));
775 if (C.isEOF()) {
776 Token.reset(Kind: MIToken::Eof, Range: C.remaining());
777 return C.remaining();
778 }
779
780 C = skipWhitespace(C: skipMachineOperandComment(C));
781
782 if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
783 return R.remaining();
784 if (Cursor R = maybeLexFloatHexBits(C, Token))
785 return R.remaining();
786 if (Cursor R = maybeLexIdentifier(C, Token))
787 return R.remaining();
788 if (Cursor R = maybeLexJumpTableIndex(C, Token))
789 return R.remaining();
790 if (Cursor R = maybeLexStackObject(C, Token))
791 return R.remaining();
792 if (Cursor R = maybeLexFixedStackObject(C, Token))
793 return R.remaining();
794 if (Cursor R = maybeLexConstantPoolItem(C, Token))
795 return R.remaining();
796 if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback))
797 return R.remaining();
798 if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
799 return R.remaining();
800 if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
801 return R.remaining();
802 if (Cursor R = maybeLexRegister(C, Token, ErrorCallback))
803 return R.remaining();
804 if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
805 return R.remaining();
806 if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
807 return R.remaining();
808 if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback))
809 return R.remaining();
810 if (Cursor R = maybeLexHexadecimalLiteral(C, Token))
811 return R.remaining();
812 if (Cursor R = maybeLexNumericalLiteral(C, Token))
813 return R.remaining();
814 if (Cursor R = maybeLexExclaim(C, Token, ErrorCallback))
815 return R.remaining();
816 if (Cursor R = maybeLexSymbol(C, Token))
817 return R.remaining();
818 if (Cursor R = maybeLexNewline(C, Token))
819 return R.remaining();
820 if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
821 return R.remaining();
822 if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback))
823 return R.remaining();
824
825 Token.reset(Kind: MIToken::Error, Range: C.remaining());
826 ErrorCallback(C.location(),
827 Twine("unexpected character '") + Twine(C.peek()) + "'");
828 return C.remaining();
829}
830