lib/MC/MCParser/WasmAsmParser.cpp

   1 //===- WasmAsmParser.cpp - Wasm Assembly Parser -----------------------------===//
   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 // Note, this is for wasm, the binary format (analogous to ELF), not wasm,
  10 // the instruction set (analogous to x86), for which parsing code lives in
  11 // WebAssemblyAsmParser.
  12 //
  13 // This file contains processing for generic directives implemented using
  14 // MCTargetStreamer, the ones that depend on WebAssemblyTargetStreamer are in
  15 // WebAssemblyAsmParser.
  16 //
  17 //===----------------------------------------------------------------------===//
  18
  19 #include "llvm/BinaryFormat/Wasm.h"
  20 #include "llvm/MC/MCContext.h"
  21 #include "llvm/MC/MCParser/MCAsmLexer.h"
  22 #include "llvm/MC/MCParser/MCAsmParser.h"
  23 #include "llvm/MC/MCParser/MCAsmParserExtension.h"
  24 #include "llvm/MC/MCSectionWasm.h"
  25 #include "llvm/MC/MCStreamer.h"
  26 #include "llvm/MC/MCSymbol.h"
  27 #include "llvm/MC/MCSymbolWasm.h"
  28 #include "llvm/Support/MachineValueType.h"
  29
  30 using namespace llvm;
  31
  32 namespace {
  33
  34 class WasmAsmParser : public MCAsmParserExtension {
  35   MCAsmParser *Parser = nullptr;
  36   MCAsmLexer *Lexer = nullptr;
  37
  38   template<bool (WasmAsmParser::*HandlerMethod)(StringRef, SMLoc)>
  39   void addDirectiveHandler(StringRef Directive) {
  40     MCAsmParser::ExtensionDirectiveHandler Handler = std::make_pair(
  41         this, HandleDirective<WasmAsmParser, HandlerMethod>);
  42
  43     getParser().addDirectiveHandler(Directive, Handler);
  44   }
  45
  46 public:
  47   WasmAsmParser() { BracketExpressionsSupported = true; }
  48
  49   void Initialize(MCAsmParser &P) override {
  50     Parser = &P;
  51     Lexer = &Parser->getLexer();
  52     // Call the base implementation.
  53     this->MCAsmParserExtension::Initialize(*Parser);
  54
  55     addDirectiveHandler<&WasmAsmParser::parseSectionDirectiveText>(".text");
  56     addDirectiveHandler<&WasmAsmParser::parseSectionDirective>(".section");
  57     addDirectiveHandler<&WasmAsmParser::parseDirectiveSize>(".size");
  58     addDirectiveHandler<&WasmAsmParser::parseDirectiveType>(".type");
  59     addDirectiveHandler<&WasmAsmParser::ParseDirectiveIdent>(".ident");
  60     addDirectiveHandler<
  61       &WasmAsmParser::ParseDirectiveSymbolAttribute>(".weak");
  62     addDirectiveHandler<
  63       &WasmAsmParser::ParseDirectiveSymbolAttribute>(".local");
  64     addDirectiveHandler<
  65       &WasmAsmParser::ParseDirectiveSymbolAttribute>(".internal");
  66     addDirectiveHandler<
  67       &WasmAsmParser::ParseDirectiveSymbolAttribute>(".hidden");
  68   }
  69
  70   bool error(const StringRef &Msg, const AsmToken &Tok) {
  71     return Parser->Error(Tok.getLoc(), Msg + Tok.getString());
  72   }
  73
  74   bool isNext(AsmToken::TokenKind Kind) {
  75     auto Ok = Lexer->is(Kind);
  76     if (Ok)
  77       Lex();
  78     return Ok;
  79   }
  80
  81   bool expect(AsmToken::TokenKind Kind, const char *KindName) {
  82     if (!isNext(Kind))
  83       return error(std::string("Expected ") + KindName + ", instead got: ",
  84                    Lexer->getTok());
  85     return false;
  86   }
  87
  88   bool parseSectionDirectiveText(StringRef, SMLoc) {
  89     // FIXME: .text currently no-op.
  90     return false;
  91   }
  92
  93   bool parseSectionFlags(StringRef FlagStr, bool &Passive) {
  94     SmallVector<StringRef, 2> Flags;
  95     // If there are no flags, keep Flags empty
  96     FlagStr.split(Flags, ",", -1, false);
  97     for (auto &Flag : Flags) {
  98       if (Flag == "passive")
  99         Passive = true;
 100       else
 101         return error("Expected section flags, instead got: ", Lexer->getTok());
 102     }
 103     return false;
 104   }
 105
 106   bool parseSectionDirective(StringRef, SMLoc) {
 107     StringRef Name;
 108     if (Parser->parseIdentifier(Name))
 109       return TokError("expected identifier in directive");
 110
 111     if (expect(AsmToken::Comma, ","))
 112       return true;
 113
 114     if (Lexer->isNot(AsmToken::String))
 115       return error("expected string in directive, instead got: ", Lexer->getTok());
 116
 117     auto Kind = StringSwitch<Optional<SectionKind>>(Name)
 118                     .StartsWith(".data", SectionKind::getData())
 119                     .StartsWith(".rodata", SectionKind::getReadOnly())
 120                     .StartsWith(".text", SectionKind::getText())
 121                     .StartsWith(".custom_section", SectionKind::getMetadata())
 122                     .StartsWith(".bss", SectionKind::getBSS())
 123                     // See use of .init_array in WasmObjectWriter and
 124                     // TargetLoweringObjectFileWasm
 125                     .StartsWith(".init_array", SectionKind::getData())
 126                     .Default(Optional<SectionKind>());
 127     if (!Kind.hasValue())
 128       return Parser->Error(Lexer->getLoc(), "unknown section kind: " + Name);
 129
 130     MCSectionWasm *Section = getContext().getWasmSection(Name, Kind.getValue());
 131
 132     // Update section flags if present in this .section directive
 133     bool Passive = false;
 134     if (parseSectionFlags(getTok().getStringContents(), Passive))
 135       return true;
 136
 137     if (Passive) {
 138       if (!Section->isWasmData())
 139         return Parser->Error(getTok().getLoc(),
 140                              "Only data sections can be passive");
 141       Section->setPassive();
 142     }
 143
 144     Lex();
 145
 146     if (expect(AsmToken::Comma, ",") || expect(AsmToken::At, "@") ||
 147         expect(AsmToken::EndOfStatement, "eol"))
 148       return true;
 149
 150     auto WS = getContext().getWasmSection(Name, Kind.getValue());
 151     getStreamer().SwitchSection(WS);
 152     return false;
 153   }
 154
 155   // TODO: This function is almost the same as ELFAsmParser::ParseDirectiveSize
 156   // so maybe could be shared somehow.
 157   bool parseDirectiveSize(StringRef, SMLoc) {
 158     StringRef Name;
 159     if (Parser->parseIdentifier(Name))
 160       return TokError("expected identifier in directive");
 161     auto Sym = getContext().getOrCreateSymbol(Name);
 162     if (expect(AsmToken::Comma, ","))
 163       return true;
 164     const MCExpr *Expr;
 165     if (Parser->parseExpression(Expr))
 166       return true;
 167     if (expect(AsmToken::EndOfStatement, "eol"))
 168       return true;
 169     // This is done automatically by the assembler for functions currently,
 170     // so this is only currently needed for data sections:
 171     getStreamer().emitELFSize(Sym, Expr);
 172     return false;
 173   }
 174
 175   bool parseDirectiveType(StringRef, SMLoc) {
 176     // This could be the start of a function, check if followed by
 177     // "label,@function"
 178     if (!Lexer->is(AsmToken::Identifier))
 179       return error("Expected label after .type directive, got: ",
 180                    Lexer->getTok());
 181     auto WasmSym = cast<MCSymbolWasm>(
 182                      getStreamer().getContext().getOrCreateSymbol(
 183                        Lexer->getTok().getString()));
 184     Lex();
 185     if (!(isNext(AsmToken::Comma) && isNext(AsmToken::At) &&
 186           Lexer->is(AsmToken::Identifier)))
 187       return error("Expected label,@type declaration, got: ", Lexer->getTok());
 188     auto TypeName = Lexer->getTok().getString();
 189     if (TypeName == "function")
 190       WasmSym->setType(wasm::WASM_SYMBOL_TYPE_FUNCTION);
 191     else if (TypeName == "global")
 192       WasmSym->setType(wasm::WASM_SYMBOL_TYPE_GLOBAL);
 193     else if (TypeName == "object")
 194       WasmSym->setType(wasm::WASM_SYMBOL_TYPE_DATA);
 195     else
 196       return error("Unknown WASM symbol type: ", Lexer->getTok());
 197     Lex();
 198     return expect(AsmToken::EndOfStatement, "EOL");
 199   }
 200
 201   // FIXME: Shared with ELF.
 202   /// ParseDirectiveIdent
 203   ///  ::= .ident string
 204   bool ParseDirectiveIdent(StringRef, SMLoc) {
 205     if (getLexer().isNot(AsmToken::String))
 206       return TokError("unexpected token in '.ident' directive");
 207     StringRef Data = getTok().getIdentifier();
 208     Lex();
 209     if (getLexer().isNot(AsmToken::EndOfStatement))
 210       return TokError("unexpected token in '.ident' directive");
 211     Lex();
 212     getStreamer().EmitIdent(Data);
 213     return false;
 214   }
 215
 216   // FIXME: Shared with ELF.
 217   /// ParseDirectiveSymbolAttribute
 218   ///  ::= { ".local", ".weak", ... } [ identifier ( , identifier )* ]
 219   bool ParseDirectiveSymbolAttribute(StringRef Directive, SMLoc) {
 220     MCSymbolAttr Attr = StringSwitch<MCSymbolAttr>(Directive)
 221       .Case(".weak", MCSA_Weak)
 222       .Case(".local", MCSA_Local)
 223       .Case(".hidden", MCSA_Hidden)
 224       .Case(".internal", MCSA_Internal)
 225       .Case(".protected", MCSA_Protected)
 226       .Default(MCSA_Invalid);
 227     assert(Attr != MCSA_Invalid && "unexpected symbol attribute directive!");
 228     if (getLexer().isNot(AsmToken::EndOfStatement)) {
 229       while (true) {
 230         StringRef Name;
 231         if (getParser().parseIdentifier(Name))
 232           return TokError("expected identifier in directive");
 233         MCSymbol *Sym = getContext().getOrCreateSymbol(Name);
 234         getStreamer().EmitSymbolAttribute(Sym, Attr);
 235         if (getLexer().is(AsmToken::EndOfStatement))
 236           break;
 237         if (getLexer().isNot(AsmToken::Comma))
 238           return TokError("unexpected token in directive");
 239         Lex();
 240       }
 241     }
 242     Lex();
 243     return false;
 244   }
 245 };
 246
 247 } // end anonymous namespace
 248
 249 namespace llvm {
 250
 251 MCAsmParserExtension *createWasmAsmParser() {
 252   return new WasmAsmParser;
 253 }
 254
 255 } // end namespace llvm