xref: /freebsd/contrib/llvm-project/llvm/lib/CodeGen/MIRParser/MILexer.h (revision 5ffd83dbcc34f10e07f6d3e968ae6365869615f4)
10b57cec5SDimitry Andric //===- MILexer.h - Lexer for machine instructions ---------------*- C++ -*-===//
20b57cec5SDimitry Andric //
30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information.
50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
60b57cec5SDimitry Andric //
70b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
80b57cec5SDimitry Andric //
90b57cec5SDimitry Andric // This file declares the function that lexes the machine instruction source
100b57cec5SDimitry Andric // string.
110b57cec5SDimitry Andric //
120b57cec5SDimitry Andric //===----------------------------------------------------------------------===//
130b57cec5SDimitry Andric 
140b57cec5SDimitry Andric #ifndef LLVM_LIB_CODEGEN_MIRPARSER_MILEXER_H
150b57cec5SDimitry Andric #define LLVM_LIB_CODEGEN_MIRPARSER_MILEXER_H
160b57cec5SDimitry Andric 
170b57cec5SDimitry Andric #include "llvm/ADT/APSInt.h"
180b57cec5SDimitry Andric #include "llvm/ADT/StringRef.h"
190b57cec5SDimitry Andric #include <string>
200b57cec5SDimitry Andric 
210b57cec5SDimitry Andric namespace llvm {
220b57cec5SDimitry Andric 
230b57cec5SDimitry Andric class Twine;
240b57cec5SDimitry Andric 
250b57cec5SDimitry Andric /// A token produced by the machine instruction lexer.
260b57cec5SDimitry Andric struct MIToken {
270b57cec5SDimitry Andric   enum TokenKind {
280b57cec5SDimitry Andric     // Markers
290b57cec5SDimitry Andric     Eof,
300b57cec5SDimitry Andric     Error,
310b57cec5SDimitry Andric     Newline,
320b57cec5SDimitry Andric 
330b57cec5SDimitry Andric     // Tokens with no info.
340b57cec5SDimitry Andric     comma,
350b57cec5SDimitry Andric     equal,
360b57cec5SDimitry Andric     underscore,
370b57cec5SDimitry Andric     colon,
380b57cec5SDimitry Andric     coloncolon,
390b57cec5SDimitry Andric     dot,
400b57cec5SDimitry Andric     exclaim,
410b57cec5SDimitry Andric     lparen,
420b57cec5SDimitry Andric     rparen,
430b57cec5SDimitry Andric     lbrace,
440b57cec5SDimitry Andric     rbrace,
450b57cec5SDimitry Andric     plus,
460b57cec5SDimitry Andric     minus,
470b57cec5SDimitry Andric     less,
480b57cec5SDimitry Andric     greater,
490b57cec5SDimitry Andric 
500b57cec5SDimitry Andric     // Keywords
510b57cec5SDimitry Andric     kw_implicit,
520b57cec5SDimitry Andric     kw_implicit_define,
530b57cec5SDimitry Andric     kw_def,
540b57cec5SDimitry Andric     kw_dead,
550b57cec5SDimitry Andric     kw_dereferenceable,
560b57cec5SDimitry Andric     kw_killed,
570b57cec5SDimitry Andric     kw_undef,
580b57cec5SDimitry Andric     kw_internal,
590b57cec5SDimitry Andric     kw_early_clobber,
600b57cec5SDimitry Andric     kw_debug_use,
610b57cec5SDimitry Andric     kw_renamable,
620b57cec5SDimitry Andric     kw_tied_def,
630b57cec5SDimitry Andric     kw_frame_setup,
640b57cec5SDimitry Andric     kw_frame_destroy,
650b57cec5SDimitry Andric     kw_nnan,
660b57cec5SDimitry Andric     kw_ninf,
670b57cec5SDimitry Andric     kw_nsz,
680b57cec5SDimitry Andric     kw_arcp,
690b57cec5SDimitry Andric     kw_contract,
700b57cec5SDimitry Andric     kw_afn,
710b57cec5SDimitry Andric     kw_reassoc,
720b57cec5SDimitry Andric     kw_nuw,
730b57cec5SDimitry Andric     kw_nsw,
740b57cec5SDimitry Andric     kw_exact,
75480093f4SDimitry Andric     kw_nofpexcept,
760b57cec5SDimitry Andric     kw_debug_location,
770b57cec5SDimitry Andric     kw_cfi_same_value,
780b57cec5SDimitry Andric     kw_cfi_offset,
790b57cec5SDimitry Andric     kw_cfi_rel_offset,
800b57cec5SDimitry Andric     kw_cfi_def_cfa_register,
810b57cec5SDimitry Andric     kw_cfi_def_cfa_offset,
820b57cec5SDimitry Andric     kw_cfi_adjust_cfa_offset,
830b57cec5SDimitry Andric     kw_cfi_escape,
840b57cec5SDimitry Andric     kw_cfi_def_cfa,
850b57cec5SDimitry Andric     kw_cfi_register,
860b57cec5SDimitry Andric     kw_cfi_remember_state,
870b57cec5SDimitry Andric     kw_cfi_restore,
880b57cec5SDimitry Andric     kw_cfi_restore_state,
890b57cec5SDimitry Andric     kw_cfi_undefined,
900b57cec5SDimitry Andric     kw_cfi_window_save,
910b57cec5SDimitry Andric     kw_cfi_aarch64_negate_ra_sign_state,
920b57cec5SDimitry Andric     kw_blockaddress,
930b57cec5SDimitry Andric     kw_intrinsic,
940b57cec5SDimitry Andric     kw_target_index,
950b57cec5SDimitry Andric     kw_half,
960b57cec5SDimitry Andric     kw_float,
970b57cec5SDimitry Andric     kw_double,
980b57cec5SDimitry Andric     kw_x86_fp80,
990b57cec5SDimitry Andric     kw_fp128,
1000b57cec5SDimitry Andric     kw_ppc_fp128,
1010b57cec5SDimitry Andric     kw_target_flags,
1020b57cec5SDimitry Andric     kw_volatile,
1030b57cec5SDimitry Andric     kw_non_temporal,
1040b57cec5SDimitry Andric     kw_invariant,
1050b57cec5SDimitry Andric     kw_align,
1060b57cec5SDimitry Andric     kw_addrspace,
1070b57cec5SDimitry Andric     kw_stack,
1080b57cec5SDimitry Andric     kw_got,
1090b57cec5SDimitry Andric     kw_jump_table,
1100b57cec5SDimitry Andric     kw_constant_pool,
1110b57cec5SDimitry Andric     kw_call_entry,
112480093f4SDimitry Andric     kw_custom,
1130b57cec5SDimitry Andric     kw_liveout,
1140b57cec5SDimitry Andric     kw_address_taken,
1150b57cec5SDimitry Andric     kw_landing_pad,
116*5ffd83dbSDimitry Andric     kw_ehfunclet_entry,
1170b57cec5SDimitry Andric     kw_liveins,
1180b57cec5SDimitry Andric     kw_successors,
1190b57cec5SDimitry Andric     kw_floatpred,
1200b57cec5SDimitry Andric     kw_intpred,
1218bcb0991SDimitry Andric     kw_shufflemask,
1220b57cec5SDimitry Andric     kw_pre_instr_symbol,
1230b57cec5SDimitry Andric     kw_post_instr_symbol,
124480093f4SDimitry Andric     kw_heap_alloc_marker,
125*5ffd83dbSDimitry Andric     kw_bbsections,
1260b57cec5SDimitry Andric     kw_unknown_size,
1270b57cec5SDimitry Andric 
1280b57cec5SDimitry Andric     // Named metadata keywords
1290b57cec5SDimitry Andric     md_tbaa,
1300b57cec5SDimitry Andric     md_alias_scope,
1310b57cec5SDimitry Andric     md_noalias,
1320b57cec5SDimitry Andric     md_range,
1330b57cec5SDimitry Andric     md_diexpr,
1340b57cec5SDimitry Andric     md_dilocation,
1350b57cec5SDimitry Andric 
1360b57cec5SDimitry Andric     // Identifier tokens
1370b57cec5SDimitry Andric     Identifier,
1380b57cec5SDimitry Andric     NamedRegister,
1390b57cec5SDimitry Andric     NamedVirtualRegister,
1400b57cec5SDimitry Andric     MachineBasicBlockLabel,
1410b57cec5SDimitry Andric     MachineBasicBlock,
1420b57cec5SDimitry Andric     StackObject,
1430b57cec5SDimitry Andric     FixedStackObject,
1440b57cec5SDimitry Andric     NamedGlobalValue,
1450b57cec5SDimitry Andric     GlobalValue,
1460b57cec5SDimitry Andric     ExternalSymbol,
1470b57cec5SDimitry Andric     MCSymbol,
1480b57cec5SDimitry Andric 
1490b57cec5SDimitry Andric     // Other tokens
1500b57cec5SDimitry Andric     IntegerLiteral,
1510b57cec5SDimitry Andric     FloatingPointLiteral,
1520b57cec5SDimitry Andric     HexLiteral,
1538bcb0991SDimitry Andric     VectorLiteral,
1540b57cec5SDimitry Andric     VirtualRegister,
1550b57cec5SDimitry Andric     ConstantPoolItem,
1560b57cec5SDimitry Andric     JumpTableIndex,
1570b57cec5SDimitry Andric     NamedIRBlock,
1580b57cec5SDimitry Andric     IRBlock,
1590b57cec5SDimitry Andric     NamedIRValue,
1600b57cec5SDimitry Andric     IRValue,
1610b57cec5SDimitry Andric     QuotedIRValue, // `<constant value>`
1620b57cec5SDimitry Andric     SubRegisterIndex,
1630b57cec5SDimitry Andric     StringConstant
1640b57cec5SDimitry Andric   };
1650b57cec5SDimitry Andric 
1660b57cec5SDimitry Andric private:
1670b57cec5SDimitry Andric   TokenKind Kind = Error;
1680b57cec5SDimitry Andric   StringRef Range;
1690b57cec5SDimitry Andric   StringRef StringValue;
1700b57cec5SDimitry Andric   std::string StringValueStorage;
1710b57cec5SDimitry Andric   APSInt IntVal;
1720b57cec5SDimitry Andric 
1730b57cec5SDimitry Andric public:
1740b57cec5SDimitry Andric   MIToken() = default;
1750b57cec5SDimitry Andric 
1760b57cec5SDimitry Andric   MIToken &reset(TokenKind Kind, StringRef Range);
1770b57cec5SDimitry Andric 
1780b57cec5SDimitry Andric   MIToken &setStringValue(StringRef StrVal);
1790b57cec5SDimitry Andric   MIToken &setOwnedStringValue(std::string StrVal);
1800b57cec5SDimitry Andric   MIToken &setIntegerValue(APSInt IntVal);
1810b57cec5SDimitry Andric 
1820b57cec5SDimitry Andric   TokenKind kind() const { return Kind; }
1830b57cec5SDimitry Andric 
1840b57cec5SDimitry Andric   bool isError() const { return Kind == Error; }
1850b57cec5SDimitry Andric 
1860b57cec5SDimitry Andric   bool isNewlineOrEOF() const { return Kind == Newline || Kind == Eof; }
1870b57cec5SDimitry Andric 
1880b57cec5SDimitry Andric   bool isErrorOrEOF() const { return Kind == Error || Kind == Eof; }
1890b57cec5SDimitry Andric 
1900b57cec5SDimitry Andric   bool isRegister() const {
1910b57cec5SDimitry Andric     return Kind == NamedRegister || Kind == underscore ||
1920b57cec5SDimitry Andric            Kind == NamedVirtualRegister || Kind == VirtualRegister;
1930b57cec5SDimitry Andric   }
1940b57cec5SDimitry Andric 
1950b57cec5SDimitry Andric   bool isRegisterFlag() const {
1960b57cec5SDimitry Andric     return Kind == kw_implicit || Kind == kw_implicit_define ||
1970b57cec5SDimitry Andric            Kind == kw_def || Kind == kw_dead || Kind == kw_killed ||
1980b57cec5SDimitry Andric            Kind == kw_undef || Kind == kw_internal ||
1990b57cec5SDimitry Andric            Kind == kw_early_clobber || Kind == kw_debug_use ||
2000b57cec5SDimitry Andric            Kind == kw_renamable;
2010b57cec5SDimitry Andric   }
2020b57cec5SDimitry Andric 
2030b57cec5SDimitry Andric   bool isMemoryOperandFlag() const {
2040b57cec5SDimitry Andric     return Kind == kw_volatile || Kind == kw_non_temporal ||
2050b57cec5SDimitry Andric            Kind == kw_dereferenceable || Kind == kw_invariant ||
2060b57cec5SDimitry Andric            Kind == StringConstant;
2070b57cec5SDimitry Andric   }
2080b57cec5SDimitry Andric 
2090b57cec5SDimitry Andric   bool is(TokenKind K) const { return Kind == K; }
2100b57cec5SDimitry Andric 
2110b57cec5SDimitry Andric   bool isNot(TokenKind K) const { return Kind != K; }
2120b57cec5SDimitry Andric 
2130b57cec5SDimitry Andric   StringRef::iterator location() const { return Range.begin(); }
2140b57cec5SDimitry Andric 
2150b57cec5SDimitry Andric   StringRef range() const { return Range; }
2160b57cec5SDimitry Andric 
2170b57cec5SDimitry Andric   /// Return the token's string value.
2180b57cec5SDimitry Andric   StringRef stringValue() const { return StringValue; }
2190b57cec5SDimitry Andric 
2200b57cec5SDimitry Andric   const APSInt &integerValue() const { return IntVal; }
2210b57cec5SDimitry Andric 
2220b57cec5SDimitry Andric   bool hasIntegerValue() const {
2230b57cec5SDimitry Andric     return Kind == IntegerLiteral || Kind == MachineBasicBlock ||
2240b57cec5SDimitry Andric            Kind == MachineBasicBlockLabel || Kind == StackObject ||
2250b57cec5SDimitry Andric            Kind == FixedStackObject || Kind == GlobalValue ||
2260b57cec5SDimitry Andric            Kind == VirtualRegister || Kind == ConstantPoolItem ||
2270b57cec5SDimitry Andric            Kind == JumpTableIndex || Kind == IRBlock || Kind == IRValue;
2280b57cec5SDimitry Andric   }
2290b57cec5SDimitry Andric };
2300b57cec5SDimitry Andric 
2310b57cec5SDimitry Andric /// Consume a single machine instruction token in the given source and return
2320b57cec5SDimitry Andric /// the remaining source string.
2330b57cec5SDimitry Andric StringRef lexMIToken(
2340b57cec5SDimitry Andric     StringRef Source, MIToken &Token,
2350b57cec5SDimitry Andric     function_ref<void(StringRef::iterator, const Twine &)> ErrorCallback);
2360b57cec5SDimitry Andric 
2370b57cec5SDimitry Andric } // end namespace llvm
2380b57cec5SDimitry Andric 
2390b57cec5SDimitry Andric #endif // LLVM_LIB_CODEGEN_MIRPARSER_MILEXER_H
240