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