blob: 89dd3d8b81780e579f695a65711b2eb8a2b61ab8 [file] [log] [blame]
Derek Schuffe4825972018-03-20 20:06:35 +00001//==- WebAssemblyAsmParser.cpp - Assembler for WebAssembly -*- C++ -*-==//
2//
Chandler Carruth2946cd72019-01-19 08:50:56 +00003// 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
Derek Schuffe4825972018-03-20 20:06:35 +00006//
7//===----------------------------------------------------------------------===//
8///
9/// \file
Adrian Prantl5f8f34e42018-05-01 15:54:18 +000010/// This file is part of the WebAssembly Assembler.
Derek Schuffe4825972018-03-20 20:06:35 +000011///
12/// It contains code to translate a parsed .s file into MCInsts.
13///
14//===----------------------------------------------------------------------===//
15
16#include "MCTargetDesc/WebAssemblyMCTargetDesc.h"
17#include "MCTargetDesc/WebAssemblyTargetStreamer.h"
Richard Trieuc6c42132019-05-15 01:03:00 +000018#include "TargetInfo/WebAssemblyTargetInfo.h"
Derek Schuffe4825972018-03-20 20:06:35 +000019#include "WebAssembly.h"
20#include "llvm/MC/MCContext.h"
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +000021#include "llvm/MC/MCExpr.h"
Derek Schuffe4825972018-03-20 20:06:35 +000022#include "llvm/MC/MCInst.h"
23#include "llvm/MC/MCInstrInfo.h"
Heejin Ahnf208f632018-09-05 01:27:38 +000024#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
25#include "llvm/MC/MCParser/MCTargetAsmParser.h"
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +000026#include "llvm/MC/MCSectionWasm.h"
Heejin Ahnf208f632018-09-05 01:27:38 +000027#include "llvm/MC/MCStreamer.h"
Derek Schuffe4825972018-03-20 20:06:35 +000028#include "llvm/MC/MCSubtargetInfo.h"
29#include "llvm/MC/MCSymbol.h"
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +000030#include "llvm/MC/MCSymbolWasm.h"
Derek Schuffe4825972018-03-20 20:06:35 +000031#include "llvm/Support/Endian.h"
32#include "llvm/Support/TargetRegistry.h"
33
34using namespace llvm;
35
36#define DEBUG_TYPE "wasm-asm-parser"
37
Derek Schuffe4825972018-03-20 20:06:35 +000038namespace {
39
Derek Schuffe4825972018-03-20 20:06:35 +000040/// WebAssemblyOperand - Instances of this class represent the operands in a
41/// parsed WASM machine instruction.
42struct WebAssemblyOperand : public MCParsedAsmOperand {
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +000043 enum KindTy { Token, Integer, Float, Symbol, BrList } Kind;
Derek Schuffe4825972018-03-20 20:06:35 +000044
45 SMLoc StartLoc, EndLoc;
46
47 struct TokOp {
48 StringRef Tok;
49 };
50
Derek Schuffe4825972018-03-20 20:06:35 +000051 struct IntOp {
52 int64_t Val;
53 };
54
55 struct FltOp {
56 double Val;
57 };
58
59 struct SymOp {
60 const MCExpr *Exp;
61 };
62
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +000063 struct BrLOp {
64 std::vector<unsigned> List;
65 };
66
Derek Schuffe4825972018-03-20 20:06:35 +000067 union {
68 struct TokOp Tok;
Derek Schuffe4825972018-03-20 20:06:35 +000069 struct IntOp Int;
70 struct FltOp Flt;
71 struct SymOp Sym;
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +000072 struct BrLOp BrL;
Derek Schuffe4825972018-03-20 20:06:35 +000073 };
74
75 WebAssemblyOperand(KindTy K, SMLoc Start, SMLoc End, TokOp T)
Heejin Ahnf208f632018-09-05 01:27:38 +000076 : Kind(K), StartLoc(Start), EndLoc(End), Tok(T) {}
Derek Schuffe4825972018-03-20 20:06:35 +000077 WebAssemblyOperand(KindTy K, SMLoc Start, SMLoc End, IntOp I)
Heejin Ahnf208f632018-09-05 01:27:38 +000078 : Kind(K), StartLoc(Start), EndLoc(End), Int(I) {}
Derek Schuffe4825972018-03-20 20:06:35 +000079 WebAssemblyOperand(KindTy K, SMLoc Start, SMLoc End, FltOp F)
Heejin Ahnf208f632018-09-05 01:27:38 +000080 : Kind(K), StartLoc(Start), EndLoc(End), Flt(F) {}
Derek Schuffe4825972018-03-20 20:06:35 +000081 WebAssemblyOperand(KindTy K, SMLoc Start, SMLoc End, SymOp S)
Heejin Ahnf208f632018-09-05 01:27:38 +000082 : Kind(K), StartLoc(Start), EndLoc(End), Sym(S) {}
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +000083 WebAssemblyOperand(KindTy K, SMLoc Start, SMLoc End)
84 : Kind(K), StartLoc(Start), EndLoc(End), BrL() {}
85
86 ~WebAssemblyOperand() {
87 if (isBrList())
88 BrL.~BrLOp();
89 }
Derek Schuffe4825972018-03-20 20:06:35 +000090
91 bool isToken() const override { return Kind == Token; }
Heejin Ahnf208f632018-09-05 01:27:38 +000092 bool isImm() const override {
93 return Kind == Integer || Kind == Float || Kind == Symbol;
94 }
Derek Schuffe4825972018-03-20 20:06:35 +000095 bool isMem() const override { return false; }
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +000096 bool isReg() const override { return false; }
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +000097 bool isBrList() const { return Kind == BrList; }
Derek Schuffe4825972018-03-20 20:06:35 +000098
99 unsigned getReg() const override {
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000100 llvm_unreachable("Assembly inspects a register operand");
101 return 0;
Derek Schuffe4825972018-03-20 20:06:35 +0000102 }
103
104 StringRef getToken() const {
105 assert(isToken());
106 return Tok.Tok;
107 }
108
109 SMLoc getStartLoc() const override { return StartLoc; }
110 SMLoc getEndLoc() const override { return EndLoc; }
111
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000112 void addRegOperands(MCInst &, unsigned) const {
113 // Required by the assembly matcher.
114 llvm_unreachable("Assembly matcher creates register operands");
Derek Schuffe4825972018-03-20 20:06:35 +0000115 }
116
117 void addImmOperands(MCInst &Inst, unsigned N) const {
118 assert(N == 1 && "Invalid number of operands!");
119 if (Kind == Integer)
120 Inst.addOperand(MCOperand::createImm(Int.Val));
121 else if (Kind == Float)
122 Inst.addOperand(MCOperand::createFPImm(Flt.Val));
123 else if (Kind == Symbol)
124 Inst.addOperand(MCOperand::createExpr(Sym.Exp));
125 else
126 llvm_unreachable("Should be immediate or symbol!");
127 }
128
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +0000129 void addBrListOperands(MCInst &Inst, unsigned N) const {
130 assert(N == 1 && isBrList() && "Invalid BrList!");
131 for (auto Br : BrL.List)
132 Inst.addOperand(MCOperand::createImm(Br));
133 }
134
Derek Schuffe4825972018-03-20 20:06:35 +0000135 void print(raw_ostream &OS) const override {
136 switch (Kind) {
137 case Token:
138 OS << "Tok:" << Tok.Tok;
139 break;
Derek Schuffe4825972018-03-20 20:06:35 +0000140 case Integer:
141 OS << "Int:" << Int.Val;
142 break;
143 case Float:
144 OS << "Flt:" << Flt.Val;
145 break;
146 case Symbol:
147 OS << "Sym:" << Sym.Exp;
148 break;
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +0000149 case BrList:
150 OS << "BrList:" << BrL.List.size();
151 break;
Derek Schuffe4825972018-03-20 20:06:35 +0000152 }
153 }
154};
155
156class WebAssemblyAsmParser final : public MCTargetAsmParser {
157 MCAsmParser &Parser;
158 MCAsmLexer &Lexer;
Derek Schuffe4825972018-03-20 20:06:35 +0000159
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000160 // Much like WebAssemblyAsmPrinter in the backend, we have to own these.
161 std::vector<std::unique_ptr<wasm::WasmSignature>> Signatures;
162
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000163 // Order of labels, directives and instructions in a .s file have no
164 // syntactical enforcement. This class is a callback from the actual parser,
165 // and yet we have to be feeding data to the streamer in a very particular
166 // order to ensure a correct binary encoding that matches the regular backend
167 // (the streamer does not enforce this). This "state machine" enum helps
168 // guarantee that correct order.
169 enum ParserState {
170 FileStart,
171 Label,
172 FunctionStart,
173 FunctionLocals,
174 Instructions,
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000175 EndFunction,
Wouter van Oortmerssenf3feb6a2019-03-04 17:18:04 +0000176 DataSection,
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000177 } CurrentState = FileStart;
178
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000179 // For ensuring blocks are properly nested.
180 enum NestingType {
181 Function,
182 Block,
183 Loop,
184 Try,
185 If,
186 Else,
187 Undefined,
188 };
189 std::vector<NestingType> NestingStack;
190
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000191 // We track this to see if a .functype following a label is the same,
192 // as this is how we recognize the start of a function.
193 MCSymbol *LastLabel = nullptr;
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000194 MCSymbol *LastFunctionLabel = nullptr;
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000195
Derek Schuffe4825972018-03-20 20:06:35 +0000196public:
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000197 WebAssemblyAsmParser(const MCSubtargetInfo &STI, MCAsmParser &Parser,
198 const MCInstrInfo &MII, const MCTargetOptions &Options)
199 : MCTargetAsmParser(Options, STI, MII), Parser(Parser),
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000200 Lexer(Parser.getLexer()) {
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000201 setAvailableFeatures(ComputeAvailableFeatures(STI.getFeatureBits()));
Derek Schuffe4825972018-03-20 20:06:35 +0000202 }
203
204#define GET_ASSEMBLER_HEADER
205#include "WebAssemblyGenAsmMatcher.inc"
206
207 // TODO: This is required to be implemented, but appears unused.
Heejin Ahnf208f632018-09-05 01:27:38 +0000208 bool ParseRegister(unsigned & /*RegNo*/, SMLoc & /*StartLoc*/,
209 SMLoc & /*EndLoc*/) override {
Derek Schuffe4825972018-03-20 20:06:35 +0000210 llvm_unreachable("ParseRegister is not implemented.");
211 }
212
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000213 bool error(const Twine &Msg, const AsmToken &Tok) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000214 return Parser.Error(Tok.getLoc(), Msg + Tok.getString());
Derek Schuffe4825972018-03-20 20:06:35 +0000215 }
216
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000217 bool error(const Twine &Msg) {
218 return Parser.Error(Lexer.getTok().getLoc(), Msg);
219 }
220
221 void addSignature(std::unique_ptr<wasm::WasmSignature> &&Sig) {
222 Signatures.push_back(std::move(Sig));
223 }
224
225 std::pair<StringRef, StringRef> nestingString(NestingType NT) {
226 switch (NT) {
227 case Function:
228 return {"function", "end_function"};
229 case Block:
230 return {"block", "end_block"};
231 case Loop:
232 return {"loop", "end_loop"};
233 case Try:
234 return {"try", "end_try"};
235 case If:
236 return {"if", "end_if"};
237 case Else:
238 return {"else", "end_if"};
239 default:
240 llvm_unreachable("unknown NestingType");
241 }
242 }
243
244 void push(NestingType NT) { NestingStack.push_back(NT); }
245
246 bool pop(StringRef Ins, NestingType NT1, NestingType NT2 = Undefined) {
247 if (NestingStack.empty())
248 return error(Twine("End of block construct with no start: ") + Ins);
249 auto Top = NestingStack.back();
250 if (Top != NT1 && Top != NT2)
251 return error(Twine("Block construct type mismatch, expected: ") +
252 nestingString(Top).second + ", instead got: " + Ins);
253 NestingStack.pop_back();
254 return false;
255 }
256
257 bool ensureEmptyNestingStack() {
Heejin Ahn18c56a02019-02-04 19:13:39 +0000258 auto Err = !NestingStack.empty();
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000259 while (!NestingStack.empty()) {
260 error(Twine("Unmatched block construct(s) at function end: ") +
261 nestingString(NestingStack.back()).first);
262 NestingStack.pop_back();
263 }
Heejin Ahn18c56a02019-02-04 19:13:39 +0000264 return Err;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000265 }
266
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000267 bool isNext(AsmToken::TokenKind Kind) {
268 auto Ok = Lexer.is(Kind);
269 if (Ok)
Heejin Ahnf208f632018-09-05 01:27:38 +0000270 Parser.Lex();
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000271 return Ok;
Derek Schuffe4825972018-03-20 20:06:35 +0000272 }
273
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000274 bool expect(AsmToken::TokenKind Kind, const char *KindName) {
275 if (!isNext(Kind))
276 return error(std::string("Expected ") + KindName + ", instead got: ",
Derek Schuffe4825972018-03-20 20:06:35 +0000277 Lexer.getTok());
278 return false;
279 }
280
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000281 StringRef expectIdent() {
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000282 if (!Lexer.is(AsmToken::Identifier)) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000283 error("Expected identifier, got: ", Lexer.getTok());
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000284 return StringRef();
285 }
286 auto Name = Lexer.getTok().getString();
287 Parser.Lex();
288 return Name;
Derek Schuffe4825972018-03-20 20:06:35 +0000289 }
290
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000291 Optional<wasm::ValType> parseType(const StringRef &Type) {
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000292 // FIXME: can't use StringSwitch because wasm::ValType doesn't have a
293 // "invalid" value.
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000294 if (Type == "i32")
295 return wasm::ValType::I32;
296 if (Type == "i64")
297 return wasm::ValType::I64;
298 if (Type == "f32")
299 return wasm::ValType::F32;
300 if (Type == "f64")
301 return wasm::ValType::F64;
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000302 if (Type == "v128" || Type == "i8x16" || Type == "i16x8" ||
303 Type == "i32x4" || Type == "i64x2" || Type == "f32x4" ||
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000304 Type == "f64x2")
305 return wasm::ValType::V128;
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000306 if (Type == "except_ref")
307 return wasm::ValType::EXCEPT_REF;
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000308 return Optional<wasm::ValType>();
309 }
310
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000311 WebAssembly::ExprType parseBlockType(StringRef ID) {
312 return StringSwitch<WebAssembly::ExprType>(ID)
313 .Case("i32", WebAssembly::ExprType::I32)
314 .Case("i64", WebAssembly::ExprType::I64)
315 .Case("f32", WebAssembly::ExprType::F32)
316 .Case("f64", WebAssembly::ExprType::F64)
317 .Case("v128", WebAssembly::ExprType::V128)
318 .Case("except_ref", WebAssembly::ExprType::ExceptRef)
319 .Case("void", WebAssembly::ExprType::Void)
320 .Default(WebAssembly::ExprType::Invalid);
321 }
322
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000323 bool parseRegTypeList(SmallVectorImpl<wasm::ValType> &Types) {
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000324 while (Lexer.is(AsmToken::Identifier)) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000325 auto Type = parseType(Lexer.getTok().getString());
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000326 if (!Type)
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000327 return error("unknown type: ", Lexer.getTok());
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000328 Types.push_back(Type.getValue());
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000329 Parser.Lex();
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000330 if (!isNext(AsmToken::Comma))
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000331 break;
332 }
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000333 return false;
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000334 }
335
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000336 void parseSingleInteger(bool IsNegative, OperandVector &Operands) {
Derek Schuffe4825972018-03-20 20:06:35 +0000337 auto &Int = Lexer.getTok();
338 int64_t Val = Int.getIntVal();
Heejin Ahnf208f632018-09-05 01:27:38 +0000339 if (IsNegative)
340 Val = -Val;
Derek Schuffe4825972018-03-20 20:06:35 +0000341 Operands.push_back(make_unique<WebAssemblyOperand>(
Heejin Ahnf208f632018-09-05 01:27:38 +0000342 WebAssemblyOperand::Integer, Int.getLoc(), Int.getEndLoc(),
343 WebAssemblyOperand::IntOp{Val}));
Derek Schuffe4825972018-03-20 20:06:35 +0000344 Parser.Lex();
345 }
346
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000347 bool parseOperandStartingWithInteger(bool IsNegative, OperandVector &Operands,
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000348 StringRef InstName) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000349 parseSingleInteger(IsNegative, Operands);
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000350 // FIXME: there is probably a cleaner way to do this.
351 auto IsLoadStore = InstName.startswith("load") ||
352 InstName.startswith("store") ||
Heejin Ahn3477bd12019-02-20 01:14:36 +0000353 InstName.startswith("atomic");
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000354 if (IsLoadStore) {
355 // Parse load/store operands of the form: offset align
356 auto &Offset = Lexer.getTok();
357 if (Offset.is(AsmToken::Integer)) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000358 parseSingleInteger(false, Operands);
Derek Schuffe4825972018-03-20 20:06:35 +0000359 } else {
360 // Alignment not specified.
361 // FIXME: correctly derive a default from the instruction.
Wouter van Oortmerssen8a9cb242018-08-27 15:45:51 +0000362 // We can't just call WebAssembly::GetDefaultP2Align since we don't have
363 // an opcode until after the assembly matcher.
Derek Schuffe4825972018-03-20 20:06:35 +0000364 Operands.push_back(make_unique<WebAssemblyOperand>(
Heejin Ahnf208f632018-09-05 01:27:38 +0000365 WebAssemblyOperand::Integer, Offset.getLoc(), Offset.getEndLoc(),
366 WebAssemblyOperand::IntOp{0}));
Derek Schuffe4825972018-03-20 20:06:35 +0000367 }
368 }
369 return false;
370 }
371
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000372 void addBlockTypeOperand(OperandVector &Operands, SMLoc NameLoc,
373 WebAssembly::ExprType BT) {
374 Operands.push_back(make_unique<WebAssemblyOperand>(
375 WebAssemblyOperand::Integer, NameLoc, NameLoc,
376 WebAssemblyOperand::IntOp{static_cast<int64_t>(BT)}));
377 }
378
Heejin Ahnf208f632018-09-05 01:27:38 +0000379 bool ParseInstruction(ParseInstructionInfo & /*Info*/, StringRef Name,
Derek Schuffe4825972018-03-20 20:06:35 +0000380 SMLoc NameLoc, OperandVector &Operands) override {
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000381 // Note: Name does NOT point into the sourcecode, but to a local, so
382 // use NameLoc instead.
383 Name = StringRef(NameLoc.getPointer(), Name.size());
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000384
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000385 // WebAssembly has instructions with / in them, which AsmLexer parses
386 // as seperate tokens, so if we find such tokens immediately adjacent (no
387 // whitespace), expand the name to include them:
388 for (;;) {
389 auto &Sep = Lexer.getTok();
390 if (Sep.getLoc().getPointer() != Name.end() ||
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000391 Sep.getKind() != AsmToken::Slash)
392 break;
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000393 // Extend name with /
394 Name = StringRef(Name.begin(), Name.size() + Sep.getString().size());
395 Parser.Lex();
396 // We must now find another identifier, or error.
397 auto &Id = Lexer.getTok();
398 if (Id.getKind() != AsmToken::Identifier ||
399 Id.getLoc().getPointer() != Name.end())
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000400 return error("Incomplete instruction name: ", Id);
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000401 Name = StringRef(Name.begin(), Name.size() + Id.getString().size());
402 Parser.Lex();
403 }
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000404
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000405 // Now construct the name as first operand.
Heejin Ahnf208f632018-09-05 01:27:38 +0000406 Operands.push_back(make_unique<WebAssemblyOperand>(
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000407 WebAssemblyOperand::Token, NameLoc, SMLoc::getFromPointer(Name.end()),
408 WebAssemblyOperand::TokOp{Name}));
Derek Schuffe4825972018-03-20 20:06:35 +0000409 auto NamePair = Name.split('.');
410 // If no '.', there is no type prefix.
Wouter van Oortmerssen0c83c3f2018-10-01 17:20:31 +0000411 auto BaseName = NamePair.second.empty() ? NamePair.first : NamePair.second;
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000412
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000413 // If this instruction is part of a control flow structure, ensure
414 // proper nesting.
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000415 bool ExpectBlockType = false;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000416 if (BaseName == "block") {
417 push(Block);
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000418 ExpectBlockType = true;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000419 } else if (BaseName == "loop") {
420 push(Loop);
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000421 ExpectBlockType = true;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000422 } else if (BaseName == "try") {
423 push(Try);
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000424 ExpectBlockType = true;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000425 } else if (BaseName == "if") {
426 push(If);
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000427 ExpectBlockType = true;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000428 } else if (BaseName == "else") {
429 if (pop(BaseName, If))
430 return true;
431 push(Else);
432 } else if (BaseName == "catch") {
433 if (pop(BaseName, Try))
434 return true;
435 push(Try);
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000436 } else if (BaseName == "end_if") {
437 if (pop(BaseName, If, Else))
438 return true;
439 } else if (BaseName == "end_try") {
440 if (pop(BaseName, Try))
441 return true;
442 } else if (BaseName == "end_loop") {
443 if (pop(BaseName, Loop))
444 return true;
445 } else if (BaseName == "end_block") {
446 if (pop(BaseName, Block))
447 return true;
448 } else if (BaseName == "end_function") {
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000449 CurrentState = EndFunction;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000450 if (pop(BaseName, Function) || ensureEmptyNestingStack())
451 return true;
452 }
453
Derek Schuffe4825972018-03-20 20:06:35 +0000454 while (Lexer.isNot(AsmToken::EndOfStatement)) {
455 auto &Tok = Lexer.getTok();
456 switch (Tok.getKind()) {
Derek Schuffe4825972018-03-20 20:06:35 +0000457 case AsmToken::Identifier: {
458 auto &Id = Lexer.getTok();
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000459 if (ExpectBlockType) {
460 // Assume this identifier is a block_type.
461 auto BT = parseBlockType(Id.getString());
462 if (BT == WebAssembly::ExprType::Invalid)
463 return error("Unknown block type: ", Id);
464 addBlockTypeOperand(Operands, NameLoc, BT);
465 Parser.Lex();
466 } else {
467 // Assume this identifier is a label.
468 const MCExpr *Val;
469 SMLoc End;
470 if (Parser.parsePrimaryExpr(Val, End))
471 return error("Cannot parse symbol: ", Lexer.getTok());
472 Operands.push_back(make_unique<WebAssemblyOperand>(
473 WebAssemblyOperand::Symbol, Id.getLoc(), Id.getEndLoc(),
474 WebAssemblyOperand::SymOp{Val}));
475 }
Derek Schuffe4825972018-03-20 20:06:35 +0000476 break;
477 }
478 case AsmToken::Minus:
479 Parser.Lex();
480 if (Lexer.isNot(AsmToken::Integer))
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000481 return error("Expected integer instead got: ", Lexer.getTok());
482 if (parseOperandStartingWithInteger(true, Operands, BaseName))
Derek Schuffe4825972018-03-20 20:06:35 +0000483 return true;
484 break;
485 case AsmToken::Integer:
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000486 if (parseOperandStartingWithInteger(false, Operands, BaseName))
Derek Schuffe4825972018-03-20 20:06:35 +0000487 return true;
488 break;
489 case AsmToken::Real: {
490 double Val;
491 if (Tok.getString().getAsDouble(Val, false))
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000492 return error("Cannot parse real: ", Tok);
Derek Schuffe4825972018-03-20 20:06:35 +0000493 Operands.push_back(make_unique<WebAssemblyOperand>(
Heejin Ahnf208f632018-09-05 01:27:38 +0000494 WebAssemblyOperand::Float, Tok.getLoc(), Tok.getEndLoc(),
495 WebAssemblyOperand::FltOp{Val}));
Derek Schuffe4825972018-03-20 20:06:35 +0000496 Parser.Lex();
497 break;
498 }
Wouter van Oortmerssend3c544a2018-12-17 22:04:44 +0000499 case AsmToken::LCurly: {
500 Parser.Lex();
501 auto Op = make_unique<WebAssemblyOperand>(
502 WebAssemblyOperand::BrList, Tok.getLoc(), Tok.getEndLoc());
503 if (!Lexer.is(AsmToken::RCurly))
504 for (;;) {
505 Op->BrL.List.push_back(Lexer.getTok().getIntVal());
506 expect(AsmToken::Integer, "integer");
507 if (!isNext(AsmToken::Comma))
508 break;
509 }
510 expect(AsmToken::RCurly, "}");
511 Operands.push_back(std::move(Op));
512 break;
513 }
Derek Schuffe4825972018-03-20 20:06:35 +0000514 default:
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000515 return error("Unexpected token in operand: ", Tok);
Derek Schuffe4825972018-03-20 20:06:35 +0000516 }
517 if (Lexer.isNot(AsmToken::EndOfStatement)) {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000518 if (expect(AsmToken::Comma, ","))
Heejin Ahnf208f632018-09-05 01:27:38 +0000519 return true;
Derek Schuffe4825972018-03-20 20:06:35 +0000520 }
521 }
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000522 if (ExpectBlockType && Operands.size() == 1) {
523 // Support blocks with no operands as default to void.
524 addBlockTypeOperand(Operands, NameLoc, WebAssembly::ExprType::Void);
Derek Schuffe4825972018-03-20 20:06:35 +0000525 }
Wouter van Oortmerssenad72f682019-01-02 23:23:51 +0000526 Parser.Lex();
Derek Schuffe4825972018-03-20 20:06:35 +0000527 return false;
528 }
529
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000530 void onLabelParsed(MCSymbol *Symbol) override {
531 LastLabel = Symbol;
532 CurrentState = Label;
533 }
534
Heejin Ahnbe5e5872018-12-11 01:11:04 +0000535 bool parseSignature(wasm::WasmSignature *Signature) {
536 if (expect(AsmToken::LParen, "("))
537 return true;
538 if (parseRegTypeList(Signature->Params))
539 return true;
540 if (expect(AsmToken::RParen, ")"))
541 return true;
542 if (expect(AsmToken::MinusGreater, "->"))
543 return true;
544 if (expect(AsmToken::LParen, "("))
545 return true;
546 if (parseRegTypeList(Signature->Returns))
547 return true;
548 if (expect(AsmToken::RParen, ")"))
549 return true;
550 return false;
551 }
552
Wouter van Oortmerssenf3feb6a2019-03-04 17:18:04 +0000553 bool CheckDataSection() {
554 if (CurrentState != DataSection) {
555 auto WS = cast<MCSectionWasm>(getStreamer().getCurrentSection().first);
556 if (WS && WS->getKind().isText())
557 return error("data directive must occur in a data segment: ",
558 Lexer.getTok());
559 }
560 CurrentState = DataSection;
561 return false;
562 }
563
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000564 // This function processes wasm-specific directives streamed to
565 // WebAssemblyTargetStreamer, all others go to the generic parser
566 // (see WasmAsmParser).
Derek Schuffe4825972018-03-20 20:06:35 +0000567 bool ParseDirective(AsmToken DirectiveID) override {
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000568 // This function has a really weird return value behavior that is different
569 // from all the other parsing functions:
570 // - return true && no tokens consumed -> don't know this directive / let
571 // the generic parser handle it.
572 // - return true && tokens consumed -> a parsing error occurred.
573 // - return false -> processed this directive successfully.
Derek Schuffe4825972018-03-20 20:06:35 +0000574 assert(DirectiveID.getKind() == AsmToken::Identifier);
575 auto &Out = getStreamer();
Heejin Ahnf208f632018-09-05 01:27:38 +0000576 auto &TOut =
577 reinterpret_cast<WebAssemblyTargetStreamer &>(*Out.getTargetStreamer());
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000578 auto &Ctx = Out.getContext();
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000579
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000580 // TODO: any time we return an error, at least one token must have been
581 // consumed, otherwise this will not signal an error to the caller.
Wouter van Oortmerssencc75e772018-11-12 20:15:01 +0000582 if (DirectiveID.getString() == ".globaltype") {
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000583 auto SymName = expectIdent();
584 if (SymName.empty())
585 return true;
586 if (expect(AsmToken::Comma, ","))
587 return true;
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000588 auto TypeTok = Lexer.getTok();
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000589 auto TypeName = expectIdent();
590 if (TypeName.empty())
591 return true;
592 auto Type = parseType(TypeName);
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000593 if (!Type)
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000594 return error("Unknown type in .globaltype directive: ", TypeTok);
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000595 // Now set this symbol with the correct type.
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000596 auto WasmSym = cast<MCSymbolWasm>(Ctx.getOrCreateSymbol(SymName));
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000597 WasmSym->setType(wasm::WASM_SYMBOL_TYPE_GLOBAL);
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000598 WasmSym->setGlobalType(
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000599 wasm::WasmGlobalType{uint8_t(Type.getValue()), true});
Wouter van Oortmerssende28b5d2018-11-02 22:04:33 +0000600 // And emit the directive again.
601 TOut.emitGlobalType(WasmSym);
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000602 return expect(AsmToken::EndOfStatement, "EOL");
603 }
604
605 if (DirectiveID.getString() == ".functype") {
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000606 // This code has to send things to the streamer similar to
607 // WebAssemblyAsmPrinter::EmitFunctionBodyStart.
608 // TODO: would be good to factor this into a common function, but the
609 // assembler and backend really don't share any common code, and this code
610 // parses the locals seperately.
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000611 auto SymName = expectIdent();
612 if (SymName.empty())
613 return true;
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000614 auto WasmSym = cast<MCSymbolWasm>(Ctx.getOrCreateSymbol(SymName));
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000615 if (CurrentState == Label && WasmSym == LastLabel) {
616 // This .functype indicates a start of a function.
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000617 if (ensureEmptyNestingStack())
618 return true;
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000619 CurrentState = FunctionStart;
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000620 LastFunctionLabel = LastLabel;
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000621 push(Function);
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000622 }
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000623 auto Signature = make_unique<wasm::WasmSignature>();
Heejin Ahnbe5e5872018-12-11 01:11:04 +0000624 if (parseSignature(Signature.get()))
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000625 return true;
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000626 WasmSym->setSignature(Signature.get());
627 addSignature(std::move(Signature));
628 WasmSym->setType(wasm::WASM_SYMBOL_TYPE_FUNCTION);
629 TOut.emitFunctionType(WasmSym);
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000630 // TODO: backend also calls TOut.emitIndIdx, but that is not implemented.
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000631 return expect(AsmToken::EndOfStatement, "EOL");
632 }
633
Heejin Ahnbe5e5872018-12-11 01:11:04 +0000634 if (DirectiveID.getString() == ".eventtype") {
635 auto SymName = expectIdent();
636 if (SymName.empty())
637 return true;
Wouter van Oortmerssenf3b762a2019-01-17 18:14:09 +0000638 auto WasmSym = cast<MCSymbolWasm>(Ctx.getOrCreateSymbol(SymName));
Heejin Ahnbe5e5872018-12-11 01:11:04 +0000639 auto Signature = make_unique<wasm::WasmSignature>();
640 if (parseRegTypeList(Signature->Params))
641 return true;
642 WasmSym->setSignature(Signature.get());
643 addSignature(std::move(Signature));
644 WasmSym->setType(wasm::WASM_SYMBOL_TYPE_EVENT);
645 TOut.emitEventType(WasmSym);
646 // TODO: backend also calls TOut.emitIndIdx, but that is not implemented.
647 return expect(AsmToken::EndOfStatement, "EOL");
648 }
649
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000650 if (DirectiveID.getString() == ".local") {
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000651 if (CurrentState != FunctionStart)
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000652 return error(".local directive should follow the start of a function",
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000653 Lexer.getTok());
Wouter van Oortmerssen49482f82018-11-19 17:10:36 +0000654 SmallVector<wasm::ValType, 4> Locals;
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000655 if (parseRegTypeList(Locals))
656 return true;
Derek Schuffe4825972018-03-20 20:06:35 +0000657 TOut.emitLocal(Locals);
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000658 CurrentState = FunctionLocals;
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000659 return expect(AsmToken::EndOfStatement, "EOL");
Derek Schuffe4825972018-03-20 20:06:35 +0000660 }
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000661
Wouter van Oortmerssenf3feb6a2019-03-04 17:18:04 +0000662 if (DirectiveID.getString() == ".int8") {
663 if (CheckDataSection()) return true;
664 int64_t V;
665 if (Parser.parseAbsoluteExpression(V))
666 return error("Cannot parse int8 constant: ", Lexer.getTok());
667 // TODO: error if value doesn't fit?
668 Out.EmitIntValue(static_cast<uint64_t>(V), 1);
669 return expect(AsmToken::EndOfStatement, "EOL");
670 }
671
672 if (DirectiveID.getString() == ".asciz") {
673 if (CheckDataSection()) return true;
674 std::string S;
675 if (Parser.parseEscapedString(S))
676 return error("Cannot parse string constant: ", Lexer.getTok());
677 Out.EmitBytes(StringRef(S.c_str(), S.length() + 1));
678 return expect(AsmToken::EndOfStatement, "EOL");
679 }
680
Heejin Ahn7ce5edf2018-12-07 21:35:37 +0000681 return true; // We didn't process this directive.
Derek Schuffe4825972018-03-20 20:06:35 +0000682 }
683
Heejin Ahnf208f632018-09-05 01:27:38 +0000684 bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned & /*Opcode*/,
685 OperandVector &Operands, MCStreamer &Out,
686 uint64_t &ErrorInfo,
Derek Schuffe4825972018-03-20 20:06:35 +0000687 bool MatchingInlineAsm) override {
688 MCInst Inst;
689 unsigned MatchResult =
690 MatchInstructionImpl(Operands, Inst, ErrorInfo, MatchingInlineAsm);
691 switch (MatchResult) {
692 case Match_Success: {
Wouter van Oortmerssenc7b89f02018-12-03 20:30:28 +0000693 if (CurrentState == FunctionStart) {
694 // This is the first instruction in a function, but we haven't seen
695 // a .local directive yet. The streamer requires locals to be encoded
696 // as a prelude to the instructions, so emit an empty list of locals
697 // here.
698 auto &TOut = reinterpret_cast<WebAssemblyTargetStreamer &>(
699 *Out.getTargetStreamer());
700 TOut.emitLocal(SmallVector<wasm::ValType, 0>());
701 }
Derek Schuffe4825972018-03-20 20:06:35 +0000702 Out.EmitInstruction(Inst, getSTI());
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000703 if (CurrentState == EndFunction) {
704 onEndOfFunction();
705 } else {
706 CurrentState = Instructions;
707 }
Derek Schuffe4825972018-03-20 20:06:35 +0000708 return false;
709 }
710 case Match_MissingFeature:
Heejin Ahnf208f632018-09-05 01:27:38 +0000711 return Parser.Error(
712 IDLoc, "instruction requires a WASM feature not currently enabled");
Derek Schuffe4825972018-03-20 20:06:35 +0000713 case Match_MnemonicFail:
714 return Parser.Error(IDLoc, "invalid instruction");
715 case Match_NearMisses:
716 return Parser.Error(IDLoc, "ambiguous instruction");
717 case Match_InvalidTiedOperand:
718 case Match_InvalidOperand: {
719 SMLoc ErrorLoc = IDLoc;
720 if (ErrorInfo != ~0ULL) {
721 if (ErrorInfo >= Operands.size())
722 return Parser.Error(IDLoc, "too few operands for instruction");
723 ErrorLoc = Operands[ErrorInfo]->getStartLoc();
724 if (ErrorLoc == SMLoc())
725 ErrorLoc = IDLoc;
726 }
727 return Parser.Error(ErrorLoc, "invalid operand for instruction");
728 }
729 }
730 llvm_unreachable("Implement any new match types added!");
731 }
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000732
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000733 void doBeforeLabelEmit(MCSymbol *Symbol) override {
734 // Start a new section for the next function automatically, since our
735 // object writer expects each function to have its own section. This way
736 // The user can't forget this "convention".
737 auto SymName = Symbol->getName();
738 if (SymName.startswith(".L"))
739 return; // Local Symbol.
740 auto SecName = ".text." + SymName;
741 auto WS = getContext().getWasmSection(SecName, SectionKind::getText());
742 getStreamer().SwitchSection(WS);
743 }
744
745 void onEndOfFunction() {
746 // Automatically output a .size directive, so it becomes optional for the
747 // user.
Wouter van Oortmerssenf3feb6a2019-03-04 17:18:04 +0000748 if (!LastFunctionLabel) return;
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000749 auto TempSym = getContext().createLinkerPrivateTempSymbol();
750 getStreamer().EmitLabel(TempSym);
Wouter van Oortmerssenf3feb6a2019-03-04 17:18:04 +0000751 auto Start = MCSymbolRefExpr::create(LastFunctionLabel, getContext());
Wouter van Oortmerssen0b3cf242019-02-04 18:03:11 +0000752 auto End = MCSymbolRefExpr::create(TempSym, getContext());
753 auto Expr =
754 MCBinaryExpr::create(MCBinaryExpr::Sub, End, Start, getContext());
755 getStreamer().emitELFSize(LastFunctionLabel, Expr);
756 }
757
Wouter van Oortmerssen29c6ce52018-12-26 22:46:18 +0000758 void onEndOfFile() override { ensureEmptyNestingStack(); }
Derek Schuffe4825972018-03-20 20:06:35 +0000759};
760} // end anonymous namespace
761
762// Force static initialization.
Tom Stellard4b0b2612019-06-11 03:21:13 +0000763extern "C" void LLVMInitializeWebAssemblyAsmParser() {
Derek Schuffe4825972018-03-20 20:06:35 +0000764 RegisterMCAsmParser<WebAssemblyAsmParser> X(getTheWebAssemblyTarget32());
765 RegisterMCAsmParser<WebAssemblyAsmParser> Y(getTheWebAssemblyTarget64());
766}
767
768#define GET_REGISTER_MATCHER
769#define GET_MATCHER_IMPLEMENTATION
770#include "WebAssemblyGenAsmMatcher.inc"