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 typedef function_ref<void(StringRef::iterator Loc, const Twine &)>
25     ErrorCallbackType;
26 
27 /// This class provides a way to iterate and get characters from the source
28 /// string.
29 class Cursor {
30   const char *Ptr;
31   const char *End;
32 
33 public:
34   Cursor(NoneType) : Ptr(nullptr), End(nullptr) {}
35 
36   explicit Cursor(StringRef Str) {
37     Ptr = Str.data();
38     End = Ptr + Str.size();
39   }
40 
41   bool isEOF() const { return Ptr == End; }
42 
43   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
44 
45   void advance(unsigned I = 1) { Ptr += I; }
46 
47   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
48 
49   StringRef upto(Cursor C) const {
50     assert(C.Ptr >= Ptr && C.Ptr <= End);
51     return StringRef(Ptr, C.Ptr - Ptr);
52   }
53 
54   StringRef::iterator location() const { return Ptr; }
55 
56   operator bool() const { return Ptr != nullptr; }
57 };
58 
59 } // end anonymous namespace
60 
61 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
62   this->Kind = Kind;
63   this->Range = Range;
64   return *this;
65 }
66 
67 MIToken &MIToken::setStringValue(StringRef StrVal) {
68   StringValue = StrVal;
69   return *this;
70 }
71 
72 MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
73   StringValueStorage = std::move(StrVal);
74   StringValue = StringValueStorage;
75   return *this;
76 }
77 
78 MIToken &MIToken::setIntegerValue(APSInt IntVal) {
79   this->IntVal = std::move(IntVal);
80   return *this;
81 }
82 
83 /// Skip the leading whitespace characters and return the updated cursor.
84 static Cursor skipWhitespace(Cursor C) {
85   while (isblank(C.peek()))
86     C.advance();
87   return C;
88 }
89 
90 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
91 
92 /// Skip a line comment and return the updated cursor.
93 static Cursor skipComment(Cursor C) {
94   if (C.peek() != ';')
95     return C;
96   while (!isNewlineChar(C.peek()) && !C.isEOF())
97     C.advance();
98   return C;
99 }
100 
101 /// Return true if the given character satisfies the following regular
102 /// expression: [-a-zA-Z$._0-9]
103 static bool isIdentifierChar(char C) {
104   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
105          C == '$';
106 }
107 
108 /// Unescapes the given string value.
109 ///
110 /// Expects the string value to be quoted.
111 static std::string unescapeQuotedString(StringRef Value) {
112   assert(Value.front() == '"' && Value.back() == '"');
113   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
114 
115   std::string Str;
116   Str.reserve(C.remaining().size());
117   while (!C.isEOF()) {
118     char Char = C.peek();
119     if (Char == '\\') {
120       if (C.peek(1) == '\\') {
121         // Two '\' become one
122         Str += '\\';
123         C.advance(2);
124         continue;
125       }
126       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
127         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
128         C.advance(3);
129         continue;
130       }
131     }
132     Str += Char;
133     C.advance();
134   }
135   return Str;
136 }
137 
138 /// Lex a string constant using the following regular expression: \"[^\"]*\"
139 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
140   assert(C.peek() == '"');
141   for (C.advance(); C.peek() != '"'; C.advance()) {
142     if (C.isEOF() || isNewlineChar(C.peek())) {
143       ErrorCallback(
144           C.location(),
145           "end of machine instruction reached before the closing '\"'");
146       return None;
147     }
148   }
149   C.advance();
150   return C;
151 }
152 
153 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
154                       unsigned PrefixLength, ErrorCallbackType 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("def", MIToken::kw_def)
192       .Case("dead", MIToken::kw_dead)
193       .Case("killed", MIToken::kw_killed)
194       .Case("undef", MIToken::kw_undef)
195       .Case("internal", MIToken::kw_internal)
196       .Case("early-clobber", MIToken::kw_early_clobber)
197       .Case("debug-use", MIToken::kw_debug_use)
198       .Case("tied-def", MIToken::kw_tied_def)
199       .Case("frame-setup", MIToken::kw_frame_setup)
200       .Case("debug-location", MIToken::kw_debug_location)
201       .Case(".cfi_same_value", MIToken::kw_cfi_same_value)
202       .Case(".cfi_offset", MIToken::kw_cfi_offset)
203       .Case(".cfi_def_cfa_register", MIToken::kw_cfi_def_cfa_register)
204       .Case(".cfi_def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
205       .Case(".cfi_def_cfa", MIToken::kw_cfi_def_cfa)
206       .Case("blockaddress", MIToken::kw_blockaddress)
207       .Case("target-index", MIToken::kw_target_index)
208       .Case("half", MIToken::kw_half)
209       .Case("float", MIToken::kw_float)
210       .Case("double", MIToken::kw_double)
211       .Case("x86_fp80", MIToken::kw_x86_fp80)
212       .Case("fp128", MIToken::kw_fp128)
213       .Case("ppc_fp128", MIToken::kw_ppc_fp128)
214       .Case("target-flags", MIToken::kw_target_flags)
215       .Case("volatile", MIToken::kw_volatile)
216       .Case("non-temporal", MIToken::kw_non_temporal)
217       .Case("invariant", MIToken::kw_invariant)
218       .Case("align", MIToken::kw_align)
219       .Case("stack", MIToken::kw_stack)
220       .Case("got", MIToken::kw_got)
221       .Case("jump-table", MIToken::kw_jump_table)
222       .Case("constant-pool", MIToken::kw_constant_pool)
223       .Case("call-entry", MIToken::kw_call_entry)
224       .Case("liveout", MIToken::kw_liveout)
225       .Case("address-taken", MIToken::kw_address_taken)
226       .Case("landing-pad", MIToken::kw_landing_pad)
227       .Case("liveins", MIToken::kw_liveins)
228       .Case("successors", MIToken::kw_successors)
229       .Default(MIToken::Identifier);
230 }
231 
232 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
233   if (!isalpha(C.peek()) && C.peek() != '_' && C.peek() != '.')
234     return None;
235   auto Range = C;
236   while (isIdentifierChar(C.peek()))
237     C.advance();
238   auto Identifier = Range.upto(C);
239   Token.reset(getIdentifierKind(Identifier), Identifier)
240       .setStringValue(Identifier);
241   return C;
242 }
243 
244 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
245                                         ErrorCallbackType ErrorCallback) {
246   bool IsReference = C.remaining().startswith("%bb.");
247   if (!IsReference && !C.remaining().startswith("bb."))
248     return None;
249   auto Range = C;
250   unsigned PrefixLength = IsReference ? 4 : 3;
251   C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
252   if (!isdigit(C.peek())) {
253     Token.reset(MIToken::Error, C.remaining());
254     ErrorCallback(C.location(), "expected a number after '%bb.'");
255     return C;
256   }
257   auto NumberRange = C;
258   while (isdigit(C.peek()))
259     C.advance();
260   StringRef Number = NumberRange.upto(C);
261   unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
262   if (C.peek() == '.') {
263     C.advance(); // Skip '.'
264     ++StringOffset;
265     while (isIdentifierChar(C.peek()))
266       C.advance();
267   }
268   Token.reset(IsReference ? MIToken::MachineBasicBlock
269                           : MIToken::MachineBasicBlockLabel,
270               Range.upto(C))
271       .setIntegerValue(APSInt(Number))
272       .setStringValue(Range.upto(C).drop_front(StringOffset));
273   return C;
274 }
275 
276 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
277                             MIToken::TokenKind Kind) {
278   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
279     return None;
280   auto Range = C;
281   C.advance(Rule.size());
282   auto NumberRange = C;
283   while (isdigit(C.peek()))
284     C.advance();
285   Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
286   return C;
287 }
288 
289 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
290                                    MIToken::TokenKind Kind) {
291   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
292     return None;
293   auto Range = C;
294   C.advance(Rule.size());
295   auto NumberRange = C;
296   while (isdigit(C.peek()))
297     C.advance();
298   StringRef Number = NumberRange.upto(C);
299   unsigned StringOffset = Rule.size() + Number.size();
300   if (C.peek() == '.') {
301     C.advance();
302     ++StringOffset;
303     while (isIdentifierChar(C.peek()))
304       C.advance();
305   }
306   Token.reset(Kind, Range.upto(C))
307       .setIntegerValue(APSInt(Number))
308       .setStringValue(Range.upto(C).drop_front(StringOffset));
309   return C;
310 }
311 
312 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
313   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
314 }
315 
316 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
317   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
318 }
319 
320 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
321   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
322 }
323 
324 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
325   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
326 }
327 
328 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
329                               ErrorCallbackType ErrorCallback) {
330   const StringRef Rule = "%ir-block.";
331   if (!C.remaining().startswith(Rule))
332     return None;
333   if (isdigit(C.peek(Rule.size())))
334     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
335   return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
336 }
337 
338 static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
339                               ErrorCallbackType ErrorCallback) {
340   const StringRef Rule = "%ir.";
341   if (!C.remaining().startswith(Rule))
342     return None;
343   if (isdigit(C.peek(Rule.size())))
344     return maybeLexIndex(C, Token, Rule, MIToken::IRValue);
345   return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
346 }
347 
348 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
349   auto Range = C;
350   C.advance(); // Skip '%'
351   auto NumberRange = C;
352   while (isdigit(C.peek()))
353     C.advance();
354   Token.reset(MIToken::VirtualRegister, Range.upto(C))
355       .setIntegerValue(APSInt(NumberRange.upto(C)));
356   return C;
357 }
358 
359 static Cursor maybeLexRegister(Cursor C, MIToken &Token) {
360   if (C.peek() != '%')
361     return None;
362   if (isdigit(C.peek(1)))
363     return lexVirtualRegister(C, Token);
364   auto Range = C;
365   C.advance(); // Skip '%'
366   while (isIdentifierChar(C.peek()))
367     C.advance();
368   Token.reset(MIToken::NamedRegister, Range.upto(C))
369       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
370   return C;
371 }
372 
373 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
374                                   ErrorCallbackType ErrorCallback) {
375   if (C.peek() != '@')
376     return None;
377   if (!isdigit(C.peek(1)))
378     return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
379                    ErrorCallback);
380   auto Range = C;
381   C.advance(1); // Skip the '@'
382   auto NumberRange = C;
383   while (isdigit(C.peek()))
384     C.advance();
385   Token.reset(MIToken::GlobalValue, Range.upto(C))
386       .setIntegerValue(APSInt(NumberRange.upto(C)));
387   return C;
388 }
389 
390 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
391                                      ErrorCallbackType 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(Cursor C, MIToken &Token,
455                              ErrorCallbackType ErrorCallback) {
456   if (C.peek() != '!')
457     return None;
458   auto Range = C;
459   C.advance(1);
460   if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
461     Token.reset(MIToken::exclaim, Range.upto(C));
462     return C;
463   }
464   while (isIdentifierChar(C.peek()))
465     C.advance();
466   StringRef StrVal = Range.upto(C);
467   Token.reset(getMetadataKeywordKind(StrVal), StrVal);
468   if (Token.isError())
469     ErrorCallback(Token.location(),
470                   "use of unknown metadata keyword '" + StrVal + "'");
471   return C;
472 }
473 
474 static MIToken::TokenKind symbolToken(char C) {
475   switch (C) {
476   case ',':
477     return MIToken::comma;
478   case '=':
479     return MIToken::equal;
480   case ':':
481     return MIToken::colon;
482   case '(':
483     return MIToken::lparen;
484   case ')':
485     return MIToken::rparen;
486   case '{':
487     return MIToken::lbrace;
488   case '}':
489     return MIToken::rbrace;
490   case '+':
491     return MIToken::plus;
492   case '-':
493     return MIToken::minus;
494   case '<':
495     return MIToken::less;
496   case '>':
497     return MIToken::greater;
498   default:
499     return MIToken::Error;
500   }
501 }
502 
503 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
504   MIToken::TokenKind Kind;
505   unsigned Length = 1;
506   if (C.peek() == ':' && C.peek(1) == ':') {
507     Kind = MIToken::coloncolon;
508     Length = 2;
509   } else
510     Kind = symbolToken(C.peek());
511   if (Kind == MIToken::Error)
512     return None;
513   auto Range = C;
514   C.advance(Length);
515   Token.reset(Kind, Range.upto(C));
516   return C;
517 }
518 
519 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
520   if (!isNewlineChar(C.peek()))
521     return None;
522   auto Range = C;
523   C.advance();
524   Token.reset(MIToken::Newline, Range.upto(C));
525   return C;
526 }
527 
528 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
529                                      ErrorCallbackType ErrorCallback) {
530   if (C.peek() != '`')
531     return None;
532   auto Range = C;
533   C.advance();
534   auto StrRange = C;
535   while (C.peek() != '`') {
536     if (C.isEOF() || isNewlineChar(C.peek())) {
537       ErrorCallback(
538           C.location(),
539           "end of machine instruction reached before the closing '`'");
540       Token.reset(MIToken::Error, Range.remaining());
541       return C;
542     }
543     C.advance();
544   }
545   StringRef Value = StrRange.upto(C);
546   C.advance();
547   Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value);
548   return C;
549 }
550 
551 StringRef llvm::lexMIToken(StringRef Source, MIToken &Token,
552                            ErrorCallbackType ErrorCallback) {
553   auto C = skipComment(skipWhitespace(Cursor(Source)));
554   if (C.isEOF()) {
555     Token.reset(MIToken::Eof, C.remaining());
556     return C.remaining();
557   }
558 
559   if (Cursor R = maybeLexIntegerType(C, Token))
560     return R.remaining();
561   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
562     return R.remaining();
563   if (Cursor R = maybeLexIdentifier(C, Token))
564     return R.remaining();
565   if (Cursor R = maybeLexJumpTableIndex(C, Token))
566     return R.remaining();
567   if (Cursor R = maybeLexStackObject(C, Token))
568     return R.remaining();
569   if (Cursor R = maybeLexFixedStackObject(C, Token))
570     return R.remaining();
571   if (Cursor R = maybeLexConstantPoolItem(C, Token))
572     return R.remaining();
573   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
574     return R.remaining();
575   if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
576     return R.remaining();
577   if (Cursor R = maybeLexRegister(C, Token))
578     return R.remaining();
579   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
580     return R.remaining();
581   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
582     return R.remaining();
583   if (Cursor R = maybeLexHexFloatingPointLiteral(C, Token))
584     return R.remaining();
585   if (Cursor R = maybeLexNumericalLiteral(C, Token))
586     return R.remaining();
587   if (Cursor R = maybeLexExlaim(C, Token, ErrorCallback))
588     return R.remaining();
589   if (Cursor R = maybeLexSymbol(C, Token))
590     return R.remaining();
591   if (Cursor R = maybeLexNewline(C, Token))
592     return R.remaining();
593   if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
594     return R.remaining();
595 
596   Token.reset(MIToken::Error, C.remaining());
597   ErrorCallback(C.location(),
598                 Twine("unexpected character '") + Twine(C.peek()) + "'");
599   return C.remaining();
600 }
601