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