1 //===- MILexer.cpp - Machine instructions lexer implementation ----------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file implements the lexing of machine instructions.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "MILexer.h"
15 #include "llvm/ADT/StringExtras.h"
16 #include "llvm/ADT/StringSwitch.h"
17 #include "llvm/ADT/Twine.h"
18 #include <cctype>
19 
20 using namespace llvm;
21 
22 namespace {
23 
24 /// This class provides a way to iterate and get characters from the source
25 /// string.
26 class Cursor {
27   const char *Ptr;
28   const char *End;
29 
30 public:
31   Cursor(NoneType) : Ptr(nullptr), End(nullptr) {}
32 
33   explicit Cursor(StringRef Str) {
34     Ptr = Str.data();
35     End = Ptr + Str.size();
36   }
37 
38   bool isEOF() const { return Ptr == End; }
39 
40   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
41 
42   void advance(unsigned I = 1) { Ptr += I; }
43 
44   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
45 
46   StringRef upto(Cursor C) const {
47     assert(C.Ptr >= Ptr && C.Ptr <= End);
48     return StringRef(Ptr, C.Ptr - Ptr);
49   }
50 
51   StringRef::iterator location() const { return Ptr; }
52 
53   operator bool() const { return Ptr != nullptr; }
54 };
55 
56 } // end anonymous namespace
57 
58 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
59   this->Kind = Kind;
60   this->Range = Range;
61   return *this;
62 }
63 
64 MIToken &MIToken::setStringValue(StringRef StrVal) {
65   StringValue = StrVal;
66   return *this;
67 }
68 
69 MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
70   StringValueStorage = std::move(StrVal);
71   StringValue = StringValueStorage;
72   return *this;
73 }
74 
75 MIToken &MIToken::setIntegerValue(APSInt IntVal) {
76   this->IntVal = std::move(IntVal);
77   return *this;
78 }
79 
80 /// Skip the leading whitespace characters and return the updated cursor.
81 static Cursor skipWhitespace(Cursor C) {
82   while (isblank(C.peek()))
83     C.advance();
84   return C;
85 }
86 
87 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
88 
89 /// Skip a line comment and return the updated cursor.
90 static Cursor skipComment(Cursor C) {
91   if (C.peek() != ';')
92     return C;
93   while (!isNewlineChar(C.peek()) && !C.isEOF())
94     C.advance();
95   return C;
96 }
97 
98 /// Return true if the given character satisfies the following regular
99 /// expression: [-a-zA-Z$._0-9]
100 static bool isIdentifierChar(char C) {
101   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
102          C == '$';
103 }
104 
105 /// Unescapes the given string value.
106 ///
107 /// Expects the string value to be quoted.
108 static std::string unescapeQuotedString(StringRef Value) {
109   assert(Value.front() == '"' && Value.back() == '"');
110   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
111 
112   std::string Str;
113   Str.reserve(C.remaining().size());
114   while (!C.isEOF()) {
115     char Char = C.peek();
116     if (Char == '\\') {
117       if (C.peek(1) == '\\') {
118         // Two '\' become one
119         Str += '\\';
120         C.advance(2);
121         continue;
122       }
123       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
124         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
125         C.advance(3);
126         continue;
127       }
128     }
129     Str += Char;
130     C.advance();
131   }
132   return Str;
133 }
134 
135 /// Lex a string constant using the following regular expression: \"[^\"]*\"
136 static Cursor lexStringConstant(
137     Cursor C,
138     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
139   assert(C.peek() == '"');
140   for (C.advance(); C.peek() != '"'; C.advance()) {
141     if (C.isEOF() || isNewlineChar(C.peek())) {
142       ErrorCallback(
143           C.location(),
144           "end of machine instruction reached before the closing '\"'");
145       return None;
146     }
147   }
148   C.advance();
149   return C;
150 }
151 
152 static Cursor lexName(
153     Cursor C, MIToken &Token, MIToken::TokenKind Type, unsigned PrefixLength,
154     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
155   auto Range = C;
156   C.advance(PrefixLength);
157   if (C.peek() == '"') {
158     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
159       StringRef String = Range.upto(R);
160       Token.reset(Type, String)
161           .setOwnedStringValue(
162               unescapeQuotedString(String.drop_front(PrefixLength)));
163       return R;
164     }
165     Token.reset(MIToken::Error, Range.remaining());
166     return Range;
167   }
168   while (isIdentifierChar(C.peek()))
169     C.advance();
170   Token.reset(Type, Range.upto(C))
171       .setStringValue(Range.upto(C).drop_front(PrefixLength));
172   return C;
173 }
174 
175 static Cursor maybeLexIntegerType(Cursor C, MIToken &Token) {
176   if (C.peek() != 'i' || !isdigit(C.peek(1)))
177     return None;
178   auto Range = C;
179   C.advance(); // Skip 'i'
180   while (isdigit(C.peek()))
181     C.advance();
182   Token.reset(MIToken::IntegerType, Range.upto(C));
183   return C;
184 }
185 
186 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
187   return StringSwitch<MIToken::TokenKind>(Identifier)
188       .Case("_", MIToken::underscore)
189       .Case("implicit", MIToken::kw_implicit)
190       .Case("implicit-def", MIToken::kw_implicit_define)
191       .Case("dead", MIToken::kw_dead)
192       .Case("killed", MIToken::kw_killed)
193       .Case("undef", MIToken::kw_undef)
194       .Case("internal", MIToken::kw_internal)
195       .Case("early-clobber", MIToken::kw_early_clobber)
196       .Case("debug-use", MIToken::kw_debug_use)
197       .Case("frame-setup", MIToken::kw_frame_setup)
198       .Case("debug-location", MIToken::kw_debug_location)
199       .Case(".cfi_same_value", MIToken::kw_cfi_same_value)
200       .Case(".cfi_offset", MIToken::kw_cfi_offset)
201       .Case(".cfi_def_cfa_register", MIToken::kw_cfi_def_cfa_register)
202       .Case(".cfi_def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
203       .Case(".cfi_def_cfa", MIToken::kw_cfi_def_cfa)
204       .Case("blockaddress", MIToken::kw_blockaddress)
205       .Case("target-index", MIToken::kw_target_index)
206       .Case("half", MIToken::kw_half)
207       .Case("float", MIToken::kw_float)
208       .Case("double", MIToken::kw_double)
209       .Case("x86_fp80", MIToken::kw_x86_fp80)
210       .Case("fp128", MIToken::kw_fp128)
211       .Case("ppc_fp128", MIToken::kw_ppc_fp128)
212       .Case("target-flags", MIToken::kw_target_flags)
213       .Case("volatile", MIToken::kw_volatile)
214       .Case("non-temporal", MIToken::kw_non_temporal)
215       .Case("invariant", MIToken::kw_invariant)
216       .Case("align", MIToken::kw_align)
217       .Case("stack", MIToken::kw_stack)
218       .Case("got", MIToken::kw_got)
219       .Case("jump-table", MIToken::kw_jump_table)
220       .Case("constant-pool", MIToken::kw_constant_pool)
221       .Case("liveout", MIToken::kw_liveout)
222       .Case("address-taken", MIToken::kw_address_taken)
223       .Case("landing-pad", MIToken::kw_landing_pad)
224       .Case("liveins", MIToken::kw_liveins)
225       .Case("successors", MIToken::kw_successors)
226       .Default(MIToken::Identifier);
227 }
228 
229 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
230   if (!isalpha(C.peek()) && C.peek() != '_' && C.peek() != '.')
231     return None;
232   auto Range = C;
233   while (isIdentifierChar(C.peek()))
234     C.advance();
235   auto Identifier = Range.upto(C);
236   Token.reset(getIdentifierKind(Identifier), Identifier)
237       .setStringValue(Identifier);
238   return C;
239 }
240 
241 static Cursor maybeLexMachineBasicBlock(
242     Cursor C, MIToken &Token,
243     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
244   bool IsReference = C.remaining().startswith("%bb.");
245   if (!IsReference && !C.remaining().startswith("bb."))
246     return None;
247   auto Range = C;
248   unsigned PrefixLength = IsReference ? 4 : 3;
249   C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
250   if (!isdigit(C.peek())) {
251     Token.reset(MIToken::Error, C.remaining());
252     ErrorCallback(C.location(), "expected a number after '%bb.'");
253     return C;
254   }
255   auto NumberRange = C;
256   while (isdigit(C.peek()))
257     C.advance();
258   StringRef Number = NumberRange.upto(C);
259   unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
260   if (C.peek() == '.') {
261     C.advance(); // Skip '.'
262     ++StringOffset;
263     while (isIdentifierChar(C.peek()))
264       C.advance();
265   }
266   Token.reset(IsReference ? MIToken::MachineBasicBlock
267                           : MIToken::MachineBasicBlockLabel,
268               Range.upto(C))
269       .setIntegerValue(APSInt(Number))
270       .setStringValue(Range.upto(C).drop_front(StringOffset));
271   return C;
272 }
273 
274 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
275                             MIToken::TokenKind Kind) {
276   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
277     return None;
278   auto Range = C;
279   C.advance(Rule.size());
280   auto NumberRange = C;
281   while (isdigit(C.peek()))
282     C.advance();
283   Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
284   return C;
285 }
286 
287 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
288                                    MIToken::TokenKind Kind) {
289   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
290     return None;
291   auto Range = C;
292   C.advance(Rule.size());
293   auto NumberRange = C;
294   while (isdigit(C.peek()))
295     C.advance();
296   StringRef Number = NumberRange.upto(C);
297   unsigned StringOffset = Rule.size() + Number.size();
298   if (C.peek() == '.') {
299     C.advance();
300     ++StringOffset;
301     while (isIdentifierChar(C.peek()))
302       C.advance();
303   }
304   Token.reset(Kind, Range.upto(C))
305       .setIntegerValue(APSInt(Number))
306       .setStringValue(Range.upto(C).drop_front(StringOffset));
307   return C;
308 }
309 
310 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
311   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
312 }
313 
314 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
315   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
316 }
317 
318 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
319   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
320 }
321 
322 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
323   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
324 }
325 
326 static Cursor maybeLexIRBlock(
327     Cursor C, MIToken &Token,
328     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
329   const StringRef Rule = "%ir-block.";
330   if (!C.remaining().startswith(Rule))
331     return None;
332   if (isdigit(C.peek(Rule.size())))
333     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
334   return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
335 }
336 
337 static Cursor maybeLexIRValue(
338     Cursor C, MIToken &Token,
339     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
340   const StringRef Rule = "%ir.";
341   if (!C.remaining().startswith(Rule))
342     return None;
343   return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
344 }
345 
346 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
347   auto Range = C;
348   C.advance(); // Skip '%'
349   auto NumberRange = C;
350   while (isdigit(C.peek()))
351     C.advance();
352   Token.reset(MIToken::VirtualRegister, Range.upto(C))
353       .setIntegerValue(APSInt(NumberRange.upto(C)));
354   return C;
355 }
356 
357 static Cursor maybeLexRegister(Cursor C, MIToken &Token) {
358   if (C.peek() != '%')
359     return None;
360   if (isdigit(C.peek(1)))
361     return lexVirtualRegister(C, Token);
362   auto Range = C;
363   C.advance(); // Skip '%'
364   while (isIdentifierChar(C.peek()))
365     C.advance();
366   Token.reset(MIToken::NamedRegister, Range.upto(C))
367       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
368   return C;
369 }
370 
371 static Cursor maybeLexGlobalValue(
372     Cursor C, MIToken &Token,
373     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
374   if (C.peek() != '@')
375     return None;
376   if (!isdigit(C.peek(1)))
377     return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
378                    ErrorCallback);
379   auto Range = C;
380   C.advance(1); // Skip the '@'
381   auto NumberRange = C;
382   while (isdigit(C.peek()))
383     C.advance();
384   Token.reset(MIToken::GlobalValue, Range.upto(C))
385       .setIntegerValue(APSInt(NumberRange.upto(C)));
386   return C;
387 }
388 
389 static Cursor maybeLexExternalSymbol(
390     Cursor C, MIToken &Token,
391     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
392   if (C.peek() != '$')
393     return None;
394   return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1,
395                  ErrorCallback);
396 }
397 
398 static bool isValidHexFloatingPointPrefix(char C) {
399   return C == 'H' || C == 'K' || C == 'L' || C == 'M';
400 }
401 
402 static Cursor maybeLexHexFloatingPointLiteral(Cursor C, MIToken &Token) {
403   if (C.peek() != '0' || C.peek(1) != 'x')
404     return None;
405   Cursor Range = C;
406   C.advance(2); // Skip '0x'
407   if (isValidHexFloatingPointPrefix(C.peek()))
408     C.advance();
409   while (isxdigit(C.peek()))
410     C.advance();
411   Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
412   return C;
413 }
414 
415 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
416   C.advance();
417   // Skip over [0-9]*([eE][-+]?[0-9]+)?
418   while (isdigit(C.peek()))
419     C.advance();
420   if ((C.peek() == 'e' || C.peek() == 'E') &&
421       (isdigit(C.peek(1)) ||
422        ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) {
423     C.advance(2);
424     while (isdigit(C.peek()))
425       C.advance();
426   }
427   Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
428   return C;
429 }
430 
431 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
432   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
433     return None;
434   auto Range = C;
435   C.advance();
436   while (isdigit(C.peek()))
437     C.advance();
438   if (C.peek() == '.')
439     return lexFloatingPointLiteral(Range, C, Token);
440   StringRef StrVal = Range.upto(C);
441   Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal));
442   return C;
443 }
444 
445 static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) {
446   return StringSwitch<MIToken::TokenKind>(Identifier)
447       .Case("!tbaa", MIToken::md_tbaa)
448       .Case("!alias.scope", MIToken::md_alias_scope)
449       .Case("!noalias", MIToken::md_noalias)
450       .Case("!range", MIToken::md_range)
451       .Default(MIToken::Error);
452 }
453 
454 static Cursor maybeLexExlaim(
455     Cursor C, MIToken &Token,
456     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
457   if (C.peek() != '!')
458     return None;
459   auto Range = C;
460   C.advance(1);
461   if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
462     Token.reset(MIToken::exclaim, Range.upto(C));
463     return C;
464   }
465   while (isIdentifierChar(C.peek()))
466     C.advance();
467   StringRef StrVal = Range.upto(C);
468   Token.reset(getMetadataKeywordKind(StrVal), StrVal);
469   if (Token.isError())
470     ErrorCallback(Token.location(),
471                   "use of unknown metadata keyword '" + StrVal + "'");
472   return C;
473 }
474 
475 static MIToken::TokenKind symbolToken(char C) {
476   switch (C) {
477   case ',':
478     return MIToken::comma;
479   case '=':
480     return MIToken::equal;
481   case ':':
482     return MIToken::colon;
483   case '(':
484     return MIToken::lparen;
485   case ')':
486     return MIToken::rparen;
487   case '{':
488     return MIToken::lbrace;
489   case '}':
490     return MIToken::rbrace;
491   case '+':
492     return MIToken::plus;
493   case '-':
494     return MIToken::minus;
495   default:
496     return MIToken::Error;
497   }
498 }
499 
500 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
501   MIToken::TokenKind Kind;
502   unsigned Length = 1;
503   if (C.peek() == ':' && C.peek(1) == ':') {
504     Kind = MIToken::coloncolon;
505     Length = 2;
506   } else
507     Kind = symbolToken(C.peek());
508   if (Kind == MIToken::Error)
509     return None;
510   auto Range = C;
511   C.advance(Length);
512   Token.reset(Kind, Range.upto(C));
513   return C;
514 }
515 
516 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
517   if (!isNewlineChar(C.peek()))
518     return None;
519   auto Range = C;
520   C.advance();
521   Token.reset(MIToken::Newline, Range.upto(C));
522   return C;
523 }
524 
525 StringRef llvm::lexMIToken(
526     StringRef Source, MIToken &Token,
527     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
528   auto C = skipComment(skipWhitespace(Cursor(Source)));
529   if (C.isEOF()) {
530     Token.reset(MIToken::Eof, C.remaining());
531     return C.remaining();
532   }
533 
534   if (Cursor R = maybeLexIntegerType(C, Token))
535     return R.remaining();
536   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
537     return R.remaining();
538   if (Cursor R = maybeLexIdentifier(C, Token))
539     return R.remaining();
540   if (Cursor R = maybeLexJumpTableIndex(C, Token))
541     return R.remaining();
542   if (Cursor R = maybeLexStackObject(C, Token))
543     return R.remaining();
544   if (Cursor R = maybeLexFixedStackObject(C, Token))
545     return R.remaining();
546   if (Cursor R = maybeLexConstantPoolItem(C, Token))
547     return R.remaining();
548   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
549     return R.remaining();
550   if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
551     return R.remaining();
552   if (Cursor R = maybeLexRegister(C, Token))
553     return R.remaining();
554   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
555     return R.remaining();
556   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
557     return R.remaining();
558   if (Cursor R = maybeLexHexFloatingPointLiteral(C, Token))
559     return R.remaining();
560   if (Cursor R = maybeLexNumericalLiteral(C, Token))
561     return R.remaining();
562   if (Cursor R = maybeLexExlaim(C, Token, ErrorCallback))
563     return R.remaining();
564   if (Cursor R = maybeLexSymbol(C, Token))
565     return R.remaining();
566   if (Cursor R = maybeLexNewline(C, Token))
567     return R.remaining();
568 
569   Token.reset(MIToken::Error, C.remaining());
570   ErrorCallback(C.location(),
571                 Twine("unexpected character '") + Twine(C.peek()) + "'");
572   return C.remaining();
573 }
574