diff options
author | Alexander Kornienko <alexfh@google.com> | 2013-03-14 10:51:38 +0000 |
---|---|---|
committer | Alexander Kornienko <alexfh@google.com> | 2013-03-14 10:51:38 +0000 |
commit | 647735c781c5b37061ee03d6e9e6c7dda92218e2 (patch) | |
tree | 5a5e56606d41060263048b5a5586b3d2380898ba /lib/MC | |
parent | 6aed25d93d1cfcde5809a73ffa7dc1b0d6396f66 (diff) | |
parent | f635ef401786c84df32090251a8cf45981ecca33 (diff) |
Updating branches/google/stable to r176857
git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/google/stable@177040 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/MC')
42 files changed, 2722 insertions, 3771 deletions
diff --git a/lib/MC/CMakeLists.txt b/lib/MC/CMakeLists.txt index 99bff96bb9..db882c020b 100644 --- a/lib/MC/CMakeLists.txt +++ b/lib/MC/CMakeLists.txt @@ -36,7 +36,6 @@ add_llvm_library(LLVMMC MCStreamer.cpp MCSubtargetInfo.cpp MCSymbol.cpp - MCTargetAsmLexer.cpp MCValue.cpp MCWin64EH.cpp MachObjectWriter.cpp diff --git a/lib/MC/ELFObjectWriter.cpp b/lib/MC/ELFObjectWriter.cpp index bfe1709462..3d995484e7 100644 --- a/lib/MC/ELFObjectWriter.cpp +++ b/lib/MC/ELFObjectWriter.cpp @@ -1,4 +1,4 @@ -//===- lib/MC/ELFObjectWriter.cpp - ELF File Writer -------------------===// +//===- lib/MC/ELFObjectWriter.cpp - ELF File Writer -----------------------===// // // The LLVM Compiler Infrastructure // @@ -135,16 +135,14 @@ class ELFObjectWriter : public MCObjectWriter { const MCSymbol *undefinedExplicitRelSym(const MCValue &Target, const MCFixup &Fixup, bool IsPCRel) const { - return TargetObjectWriter->undefinedExplicitRelSym(Target, Fixup, IsPCRel); + return TargetObjectWriter->undefinedExplicitRelSym(Target, Fixup, + IsPCRel); } bool is64Bit() const { return TargetObjectWriter->is64Bit(); } bool hasRelocationAddend() const { return TargetObjectWriter->hasRelocationAddend(); } - unsigned getEFlags() const { - return TargetObjectWriter->getEFlags(); - } unsigned GetRelocType(const MCValue &Target, const MCFixup &Fixup, bool IsPCRel, bool IsRelocWithSymbol, int64_t Addend) const { @@ -152,13 +150,12 @@ class ELFObjectWriter : public MCObjectWriter { IsRelocWithSymbol, Addend); } - public: ELFObjectWriter(MCELFObjectTargetWriter *MOTW, raw_ostream &_OS, bool IsLittleEndian) : MCObjectWriter(_OS, IsLittleEndian), TargetObjectWriter(MOTW), - NeedsGOT(false), NeedsSymtabShndx(false){ + NeedsGOT(false), NeedsSymtabShndx(false) { } virtual ~ELFObjectWriter(); @@ -233,7 +230,8 @@ class ELFObjectWriter : public MCObjectWriter { F.getContents().append(&buf[0], &buf[8]); } - void WriteHeader(uint64_t SectionDataSize, + void WriteHeader(const MCAssembler &Asm, + uint64_t SectionDataSize, unsigned NumberOfSections); void WriteSymbolEntry(MCDataFragment *SymtabF, @@ -373,7 +371,8 @@ ELFObjectWriter::~ELFObjectWriter() {} // Emit the ELF header. -void ELFObjectWriter::WriteHeader(uint64_t SectionDataSize, +void ELFObjectWriter::WriteHeader(const MCAssembler &Asm, + uint64_t SectionDataSize, unsigned NumberOfSections) { // ELF Header // ---------- @@ -411,7 +410,7 @@ void ELFObjectWriter::WriteHeader(uint64_t SectionDataSize, sizeof(ELF::Elf32_Ehdr))); // e_shoff = sec hdr table off in bytes // e_flags = whatever the target wants - Write32(getEFlags()); + Write32(Asm.getELFHeaderEFlags()); // e_ehsize = ELF header size Write16(is64Bit() ? sizeof(ELF::Elf64_Ehdr) : sizeof(ELF::Elf32_Ehdr)); @@ -547,12 +546,17 @@ void ELFObjectWriter::WriteSymbol(MCDataFragment *SymtabF, bool IsReserved = Data.isCommon() || Data.getSymbol().isAbsolute() || Data.getSymbol().isVariable(); + // Binding and Type share the same byte as upper and lower nibbles uint8_t Binding = MCELF::GetBinding(OrigData); - uint8_t Visibility = MCELF::GetVisibility(OrigData); uint8_t Type = MCELF::GetType(Data); - uint8_t Info = (Binding << ELF_STB_Shift) | (Type << ELF_STT_Shift); - uint8_t Other = Visibility; + + // Other and Visibility share the same byte with Visability using the lower + // 2 bits + uint8_t Visibility = MCELF::GetVisibility(OrigData); + uint8_t Other = MCELF::getOther(OrigData) << + (ELF_Other_Shift - ELF_STV_Shift); + Other |= Visibility; uint64_t Value = SymbolValue(Data, Layout); uint64_t Size = 0; @@ -865,7 +869,7 @@ void ELFObjectWriter::ComputeSymbolTable(MCAssembler &Asm, // FIXME: Is this the correct place to do this? // FIXME: Why is an undefined reference to _GLOBAL_OFFSET_TABLE_ needed? if (NeedsGOT) { - llvm::StringRef Name = "_GLOBAL_OFFSET_TABLE_"; + StringRef Name = "_GLOBAL_OFFSET_TABLE_"; MCSymbol *Sym = Asm.getContext().GetOrCreateSymbol(Name); MCSymbolData &Data = Asm.getOrCreateSymbolData(*Sym); Data.setExternal(true); @@ -974,7 +978,7 @@ void ELFObjectWriter::ComputeSymbolTable(MCAssembler &Asm, for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i) UndefinedSymbolData[i].SymbolData->setIndex(Index++); - if (NumRegularSections > ELF::SHN_LORESERVE) + if (Index >= ELF::SHN_LORESERVE) NeedsSymtabShndx = true; } @@ -1319,6 +1323,8 @@ void ELFObjectWriter::WriteSection(MCAssembler &Asm, case ELF::SHT_FINI_ARRAY: case ELF::SHT_PREINIT_ARRAY: case ELF::SHT_X86_64_UNWIND: + case ELF::SHT_MIPS_REGINFO: + case ELF::SHT_MIPS_OPTIONS: // Nothing to do. break; @@ -1332,6 +1338,24 @@ void ELFObjectWriter::WriteSection(MCAssembler &Asm, break; } + if (TargetObjectWriter->getEMachine() == ELF::EM_ARM && + Section.getType() == ELF::SHT_ARM_EXIDX) { + StringRef SecName(Section.getSectionName()); + if (SecName == ".ARM.exidx") { + sh_link = SectionIndexMap.lookup( + Asm.getContext().getELFSection(".text", + ELF::SHT_PROGBITS, + ELF::SHF_EXECINSTR | ELF::SHF_ALLOC, + SectionKind::getText())); + } else if (SecName.startswith(".ARM.exidx")) { + sh_link = SectionIndexMap.lookup( + Asm.getContext().getELFSection(SecName.substr(sizeof(".ARM.exidx") - 1), + ELF::SHT_PROGBITS, + ELF::SHF_EXECINSTR | ELF::SHF_ALLOC, + SectionKind::getText())); + } + } + WriteSecHdrEntry(SectionStringTableIndex[&Section], Section.getType(), Section.getFlags(), 0, Offset, Size, sh_link, sh_info, Alignment, Section.getEntrySize()); @@ -1532,7 +1556,7 @@ void ELFObjectWriter::WriteObject(MCAssembler &Asm, } // Write out the ELF header ... - WriteHeader(SectionHeaderOffset, NumSections + 1); + WriteHeader(Asm, SectionHeaderOffset, NumSections + 1); // ... then the regular sections ... // + because of .shstrtab diff --git a/lib/MC/MCAsmInfo.cpp b/lib/MC/MCAsmInfo.cpp index 7ea0f3b85a..51bb435710 100644 --- a/lib/MC/MCAsmInfo.cpp +++ b/lib/MC/MCAsmInfo.cpp @@ -24,6 +24,8 @@ using namespace llvm; MCAsmInfo::MCAsmInfo() { PointerSize = 4; + CalleeSaveStackSlotSize = 4; + IsLittleEndian = true; StackGrowsUp = false; HasSubsectionsViaSymbols = false; @@ -37,6 +39,7 @@ MCAsmInfo::MCAsmInfo() { CommentColumn = 40; CommentString = "#"; LabelSuffix = ":"; + DebugLabelSuffix = ":"; GlobalPrefix = ""; PrivateGlobalPrefix = "."; LinkerPrivateGlobalPrefix = ""; diff --git a/lib/MC/MCAsmStreamer.cpp b/lib/MC/MCAsmStreamer.cpp index 6537e4e685..35613b411c 100644 --- a/lib/MC/MCAsmStreamer.cpp +++ b/lib/MC/MCAsmStreamer.cpp @@ -71,7 +71,7 @@ public: MCInstPrinter *printer, MCCodeEmitter *emitter, MCAsmBackend *asmbackend, bool showInst) - : MCStreamer(Context), OS(os), MAI(Context.getAsmInfo()), + : MCStreamer(SK_AsmStreamer, Context), OS(os), MAI(Context.getAsmInfo()), InstPrinter(printer), Emitter(emitter), AsmBackend(asmbackend), CommentStream(CommentToEmit), IsVerboseAsm(isVerboseAsm), ShowInst(showInst), UseLoc(useLoc), UseCFI(useCFI), @@ -127,17 +127,25 @@ public: virtual void ChangeSection(const MCSection *Section); virtual void InitSections() { + InitToTextSection(); + } + + virtual void InitToTextSection() { // FIXME, this is MachO specific, but the testsuite // expects this. - SwitchSection(getContext().getMachOSection("__TEXT", "__text", - MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, - 0, SectionKind::getText())); + SwitchSection(getContext().getMachOSection( + "__TEXT", "__text", + MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, + 0, SectionKind::getText())); } virtual void EmitLabel(MCSymbol *Symbol); + virtual void EmitDebugLabel(MCSymbol *Symbol); + virtual void EmitEHSymAttributes(const MCSymbol *Symbol, MCSymbol *EHSymbol); virtual void EmitAssemblerFlag(MCAssemblerFlag Flag); + virtual void EmitLinkerOptions(ArrayRef<std::string> Options); virtual void EmitDataRegion(MCDataRegionType Kind); virtual void EmitThumbFunc(MCSymbol *Func); @@ -207,7 +215,7 @@ public: virtual void EmitFileDirective(StringRef Filename); virtual bool EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, - StringRef Filename); + StringRef Filename, unsigned CUID = 0); virtual void EmitDwarfLocDirective(unsigned FileNo, unsigned Line, unsigned Column, unsigned Flags, unsigned Isa, unsigned Discriminator, @@ -257,6 +265,10 @@ public: virtual void EmitInstruction(const MCInst &Inst); + virtual void EmitBundleAlignMode(unsigned AlignPow2); + virtual void EmitBundleLock(bool AlignToEnd); + virtual void EmitBundleUnlock(); + /// EmitRawText - If this file is backed by an assembly streamer, this dumps /// the specified string in the output .s file. This capability is /// indicated by the hasRawTextSupport() predicate. @@ -265,6 +277,10 @@ public: virtual void FinishImpl(); /// @} + + static bool classof(const MCStreamer *S) { + return S->getKind() == SK_AsmStreamer; + } }; } // end anonymous namespace. @@ -345,6 +361,14 @@ void MCAsmStreamer::EmitLabel(MCSymbol *Symbol) { EmitEOL(); } +void MCAsmStreamer::EmitDebugLabel(MCSymbol *Symbol) { + assert(Symbol->isUndefined() && "Cannot define a symbol twice!"); + MCStreamer::EmitDebugLabel(Symbol); + + OS << *Symbol << MAI.getDebugLabelSuffix(); + EmitEOL(); +} + void MCAsmStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { switch (Flag) { case MCAF_SyntaxUnified: OS << "\t.syntax unified"; break; @@ -356,6 +380,16 @@ void MCAsmStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { EmitEOL(); } +void MCAsmStreamer::EmitLinkerOptions(ArrayRef<std::string> Options) { + assert(!Options.empty() && "At least one option is required!"); + OS << "\t.linker_option \"" << Options[0] << '"'; + for (ArrayRef<std::string>::iterator it = Options.begin() + 1, + ie = Options.end(); it != ie; ++it) { + OS << ", " << '"' << *it << '"'; + } + OS << "\n"; +} + void MCAsmStreamer::EmitDataRegion(MCDataRegionType Kind) { MCContext &Ctx = getContext(); const MCAsmInfo &MAI = Ctx.getAsmInfo(); @@ -794,14 +828,14 @@ void MCAsmStreamer::EmitFileDirective(StringRef Filename) { } bool MCAsmStreamer::EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, - StringRef Filename) { + StringRef Filename, unsigned CUID) { if (!UseDwarfDirectory && !Directory.empty()) { if (sys::path::is_absolute(Filename)) - return EmitDwarfFileDirective(FileNo, "", Filename); + return EmitDwarfFileDirective(FileNo, "", Filename, CUID); SmallString<128> FullPathName = Directory; sys::path::append(FullPathName, Filename); - return EmitDwarfFileDirective(FileNo, "", FullPathName); + return EmitDwarfFileDirective(FileNo, "", FullPathName, CUID); } if (UseLoc) { @@ -812,8 +846,11 @@ bool MCAsmStreamer::EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, } PrintQuotedString(Filename, OS); EmitEOL(); + // All .file will belong to a single CUID. + CUID = 0; } - return this->MCStreamer::EmitDwarfFileDirective(FileNo, Directory, Filename); + return this->MCStreamer::EmitDwarfFileDirective(FileNo, Directory, Filename, + CUID); } void MCAsmStreamer::EmitDwarfLocDirective(unsigned FileNo, unsigned Line, @@ -1351,6 +1388,23 @@ void MCAsmStreamer::EmitInstruction(const MCInst &Inst) { EmitEOL(); } +void MCAsmStreamer::EmitBundleAlignMode(unsigned AlignPow2) { + OS << "\t.bundle_align_mode " << AlignPow2; + EmitEOL(); +} + +void MCAsmStreamer::EmitBundleLock(bool AlignToEnd) { + OS << "\t.bundle_lock"; + if (AlignToEnd) + OS << " align_to_end"; + EmitEOL(); +} + +void MCAsmStreamer::EmitBundleUnlock() { + OS << "\t.bundle_unlock"; + EmitEOL(); +} + /// EmitRawText - If this file is backed by an assembly streamer, this dumps /// the specified string in the output .s file. This capability is /// indicated by the hasRawTextSupport() predicate. diff --git a/lib/MC/MCAssembler.cpp b/lib/MC/MCAssembler.cpp index 8f8ec151e5..1829266f96 100644 --- a/lib/MC/MCAssembler.cpp +++ b/lib/MC/MCAssembler.cpp @@ -34,10 +34,12 @@ using namespace llvm; namespace { namespace stats { STATISTIC(EmittedFragments, "Number of emitted assembler fragments - total"); -STATISTIC(EmittedInstFragments, - "Number of emitted assembler fragments - instruction"); +STATISTIC(EmittedRelaxableFragments, + "Number of emitted assembler fragments - relaxable"); STATISTIC(EmittedDataFragments, "Number of emitted assembler fragments - data"); +STATISTIC(EmittedCompactEncodedInstFragments, + "Number of emitted assembler fragments - compact encoded inst"); STATISTIC(EmittedAlignFragments, "Number of emitted assembler fragments - align"); STATISTIC(EmittedFillFragments, @@ -71,7 +73,7 @@ MCAsmLayout::MCAsmLayout(MCAssembler &Asm) SectionOrder.push_back(&*it); } -bool MCAsmLayout::isFragmentUpToDate(const MCFragment *F) const { +bool MCAsmLayout::isFragmentValid(const MCFragment *F) const { const MCSectionData &SD = *F->getParent(); const MCFragment *LastValid = LastValidFragment.lookup(&SD); if (!LastValid) @@ -80,17 +82,18 @@ bool MCAsmLayout::isFragmentUpToDate(const MCFragment *F) const { return F->getLayoutOrder() <= LastValid->getLayoutOrder(); } -void MCAsmLayout::invalidateFragmentsAfter(MCFragment *F) { - // If this fragment wasn't already up-to-date, we don't need to do anything. - if (!isFragmentUpToDate(F)) +void MCAsmLayout::invalidateFragmentsFrom(MCFragment *F) { + // If this fragment wasn't already valid, we don't need to do anything. + if (!isFragmentValid(F)) return; - // Otherwise, reset the last valid fragment to this fragment. + // Otherwise, reset the last valid fragment to the previous fragment + // (if this is the first fragment, it will be NULL). const MCSectionData &SD = *F->getParent(); - LastValidFragment[&SD] = F; + LastValidFragment[&SD] = F->getPrevNode(); } -void MCAsmLayout::EnsureValid(const MCFragment *F) const { +void MCAsmLayout::ensureValid(const MCFragment *F) const { MCSectionData &SD = *F->getParent(); MCFragment *Cur = LastValidFragment[&SD]; @@ -99,15 +102,16 @@ void MCAsmLayout::EnsureValid(const MCFragment *F) const { else Cur = Cur->getNextNode(); - // Advance the layout position until the fragment is up-to-date. - while (!isFragmentUpToDate(F)) { - const_cast<MCAsmLayout*>(this)->LayoutFragment(Cur); + // Advance the layout position until the fragment is valid. + while (!isFragmentValid(F)) { + assert(Cur && "Layout bookkeeping error"); + const_cast<MCAsmLayout*>(this)->layoutFragment(Cur); Cur = Cur->getNextNode(); } } uint64_t MCAsmLayout::getFragmentOffset(const MCFragment *F) const { - EnsureValid(F); + ensureValid(F); assert(F->Offset != ~UINT64_C(0) && "Address not set!"); return F->Offset; } @@ -159,6 +163,46 @@ uint64_t MCAsmLayout::getSectionFileSize(const MCSectionData *SD) const { return getSectionAddressSize(SD); } +uint64_t MCAsmLayout::computeBundlePadding(const MCFragment *F, + uint64_t FOffset, uint64_t FSize) { + uint64_t BundleSize = Assembler.getBundleAlignSize(); + assert(BundleSize > 0 && + "computeBundlePadding should only be called if bundling is enabled"); + uint64_t BundleMask = BundleSize - 1; + uint64_t OffsetInBundle = FOffset & BundleMask; + uint64_t EndOfFragment = OffsetInBundle + FSize; + + // There are two kinds of bundling restrictions: + // + // 1) For alignToBundleEnd(), add padding to ensure that the fragment will + // *end* on a bundle boundary. + // 2) Otherwise, check if the fragment would cross a bundle boundary. If it + // would, add padding until the end of the bundle so that the fragment + // will start in a new one. + if (F->alignToBundleEnd()) { + // Three possibilities here: + // + // A) The fragment just happens to end at a bundle boundary, so we're good. + // B) The fragment ends before the current bundle boundary: pad it just + // enough to reach the boundary. + // C) The fragment ends after the current bundle boundary: pad it until it + // reaches the end of the next bundle boundary. + // + // Note: this code could be made shorter with some modulo trickery, but it's + // intentionally kept in its more explicit form for simplicity. + if (EndOfFragment == BundleSize) + return 0; + else if (EndOfFragment < BundleSize) + return BundleSize - EndOfFragment; + else { // EndOfFragment > BundleSize + return 2 * BundleSize - EndOfFragment; + } + } else if (EndOfFragment > BundleSize) + return BundleSize - OffsetInBundle; + else + return 0; +} + /* *** */ MCFragment::MCFragment() : Kind(FragmentType(~0)) { @@ -181,12 +225,18 @@ MCEncodedFragment::~MCEncodedFragment() { /* *** */ +MCEncodedFragmentWithFixups::~MCEncodedFragmentWithFixups() { +} + +/* *** */ + MCSectionData::MCSectionData() : Section(0) {} MCSectionData::MCSectionData(const MCSection &_Section, MCAssembler *A) : Section(&_Section), Ordinal(~UINT32_C(0)), Alignment(1), + BundleLockState(NotBundleLocked), BundleGroupBeforeFirstInst(false), HasInstructions(false) { if (A) @@ -214,12 +264,32 @@ MCAssembler::MCAssembler(MCContext &Context_, MCAsmBackend &Backend_, MCCodeEmitter &Emitter_, MCObjectWriter &Writer_, raw_ostream &OS_) : Context(Context_), Backend(Backend_), Emitter(Emitter_), Writer(Writer_), - OS(OS_), RelaxAll(false), NoExecStack(false), SubsectionsViaSymbols(false) { + OS(OS_), BundleAlignSize(0), RelaxAll(false), NoExecStack(false), + SubsectionsViaSymbols(false), ELFHeaderEFlags(0) { } MCAssembler::~MCAssembler() { } +void MCAssembler::reset() { + Sections.clear(); + Symbols.clear(); + SectionMap.clear(); + SymbolMap.clear(); + IndirectSymbols.clear(); + DataRegions.clear(); + ThumbFuncs.clear(); + RelaxAll = false; + NoExecStack = false; + SubsectionsViaSymbols = false; + ELFHeaderEFlags = 0; + + // reset objects owned by us + getBackend().reset(); + getEmitter().reset(); + getWriter().reset(); +} + bool MCAssembler::isSymbolLinkerVisible(const MCSymbol &Symbol) const { // Non-temporary labels should always be visible to the linker. if (!Symbol.isTemporary()) @@ -326,11 +396,11 @@ uint64_t MCAssembler::computeFragmentSize(const MCAsmLayout &Layout, const MCFragment &F) const { switch (F.getKind()) { case MCFragment::FT_Data: - return cast<MCDataFragment>(F).getContents().size(); + case MCFragment::FT_Relaxable: + case MCFragment::FT_CompactEncodedInst: + return cast<MCEncodedFragment>(F).getContents().size(); case MCFragment::FT_Fill: return cast<MCFillFragment>(F).getSize(); - case MCFragment::FT_Inst: - return cast<MCInstFragment>(F).getInstSize(); case MCFragment::FT_LEB: return cast<MCLEBFragment>(F).getContents().size(); @@ -351,7 +421,7 @@ uint64_t MCAssembler::computeFragmentSize(const MCAsmLayout &Layout, } case MCFragment::FT_Org: { - MCOrgFragment &OF = cast<MCOrgFragment>(F); + const MCOrgFragment &OF = cast<MCOrgFragment>(F); int64_t TargetLocation; if (!OF.getOffset().EvaluateAsAbsolute(TargetLocation, Layout)) report_fatal_error("expected assembly-time absolute expression"); @@ -374,31 +444,61 @@ uint64_t MCAssembler::computeFragmentSize(const MCAsmLayout &Layout, llvm_unreachable("invalid fragment kind"); } -void MCAsmLayout::LayoutFragment(MCFragment *F) { +void MCAsmLayout::layoutFragment(MCFragment *F) { MCFragment *Prev = F->getPrevNode(); - // We should never try to recompute something which is up-to-date. - assert(!isFragmentUpToDate(F) && "Attempt to recompute up-to-date fragment!"); - // We should never try to compute the fragment layout if it's predecessor - // isn't up-to-date. - assert((!Prev || isFragmentUpToDate(Prev)) && - "Attempt to compute fragment before it's predecessor!"); + // We should never try to recompute something which is valid. + assert(!isFragmentValid(F) && "Attempt to recompute a valid fragment!"); + // We should never try to compute the fragment layout if its predecessor + // isn't valid. + assert((!Prev || isFragmentValid(Prev)) && + "Attempt to compute fragment before its predecessor!"); ++stats::FragmentLayouts; // Compute fragment offset and size. - uint64_t Offset = 0; if (Prev) - Offset += Prev->Offset + getAssembler().computeFragmentSize(*this, *Prev); - - F->Offset = Offset; + F->Offset = Prev->Offset + getAssembler().computeFragmentSize(*this, *Prev); + else + F->Offset = 0; LastValidFragment[F->getParent()] = F; + + // If bundling is enabled and this fragment has instructions in it, it has to + // obey the bundling restrictions. With padding, we'll have: + // + // + // BundlePadding + // ||| + // ------------------------------------- + // Prev |##########| F | + // ------------------------------------- + // ^ + // | + // F->Offset + // + // The fragment's offset will point to after the padding, and its computed + // size won't include the padding. + // + if (Assembler.isBundlingEnabled() && F->hasInstructions()) { + assert(isa<MCEncodedFragment>(F) && + "Only MCEncodedFragment implementations have instructions"); + uint64_t FSize = Assembler.computeFragmentSize(*this, *F); + + if (FSize > Assembler.getBundleAlignSize()) + report_fatal_error("Fragment can't be larger than a bundle size"); + + uint64_t RequiredBundlePadding = computeBundlePadding(F, F->Offset, FSize); + if (RequiredBundlePadding > UINT8_MAX) + report_fatal_error("Padding cannot exceed 255 bytes"); + F->setBundlePadding(static_cast<uint8_t>(RequiredBundlePadding)); + F->Offset += RequiredBundlePadding; + } } /// \brief Write the contents of a fragment to the given object writer. Expects /// a MCEncodedFragment. static void writeFragmentContents(const MCFragment &F, MCObjectWriter *OW) { - MCEncodedFragment &EF = cast<MCEncodedFragment>(F); + const MCEncodedFragment &EF = cast<MCEncodedFragment>(F); OW->WriteBytes(EF.getContents()); } @@ -406,17 +506,50 @@ static void writeFragmentContents(const MCFragment &F, MCObjectWriter *OW) { static void writeFragment(const MCAssembler &Asm, const MCAsmLayout &Layout, const MCFragment &F) { MCObjectWriter *OW = &Asm.getWriter(); + + // FIXME: Embed in fragments instead? + uint64_t FragmentSize = Asm.computeFragmentSize(Layout, F); + + // Should NOP padding be written out before this fragment? + unsigned BundlePadding = F.getBundlePadding(); + if (BundlePadding > 0) { + assert(Asm.isBundlingEnabled() && + "Writing bundle padding with disabled bundling"); + assert(F.hasInstructions() && + "Writing bundle padding for a fragment without instructions"); + + unsigned TotalLength = BundlePadding + static_cast<unsigned>(FragmentSize); + if (F.alignToBundleEnd() && TotalLength > Asm.getBundleAlignSize()) { + // If the padding itself crosses a bundle boundary, it must be emitted + // in 2 pieces, since even nop instructions must not cross boundaries. + // v--------------v <- BundleAlignSize + // v---------v <- BundlePadding + // ---------------------------- + // | Prev |####|####| F | + // ---------------------------- + // ^-------------------^ <- TotalLength + unsigned DistanceToBoundary = TotalLength - Asm.getBundleAlignSize(); + if (!Asm.getBackend().writeNopData(DistanceToBoundary, OW)) + report_fatal_error("unable to write NOP sequence of " + + Twine(DistanceToBoundary) + " bytes"); + BundlePadding -= DistanceToBoundary; + } + if (!Asm.getBackend().writeNopData(BundlePadding, OW)) + report_fatal_error("unable to write NOP sequence of " + + Twine(BundlePadding) + " bytes"); + } + + // This variable (and its dummy usage) is to participate in the assert at + // the end of the function. uint64_t Start = OW->getStream().tell(); (void) Start; ++stats::EmittedFragments; - // FIXME: Embed in fragments instead? - uint64_t FragmentSize = Asm.computeFragmentSize(Layout, F); switch (F.getKind()) { case MCFragment::FT_Align: { ++stats::EmittedAlignFragments; - MCAlignFragment &AF = cast<MCAlignFragment>(F); + const MCAlignFragment &AF = cast<MCAlignFragment>(F); uint64_t Count = FragmentSize / AF.getValueSize(); assert(AF.getValueSize() && "Invalid virtual align in concrete fragment!"); @@ -459,14 +592,19 @@ static void writeFragment(const MCAssembler &Asm, const MCAsmLayout &Layout, writeFragmentContents(F, OW); break; - case MCFragment::FT_Inst: - ++stats::EmittedInstFragments; + case MCFragment::FT_Relaxable: + ++stats::EmittedRelaxableFragments; + writeFragmentContents(F, OW); + break; + + case MCFragment::FT_CompactEncodedInst: + ++stats::EmittedCompactEncodedInstFragments; writeFragmentContents(F, OW); break; case MCFragment::FT_Fill: { ++stats::EmittedFillFragments; - MCFillFragment &FF = cast<MCFillFragment>(F); + const MCFillFragment &FF = cast<MCFillFragment>(F); assert(FF.getValueSize() && "Invalid virtual align in concrete fragment!"); @@ -483,14 +621,14 @@ static void writeFragment(const MCAssembler &Asm, const MCAsmLayout &Layout, } case MCFragment::FT_LEB: { - MCLEBFragment &LF = cast<MCLEBFragment>(F); + const MCLEBFragment &LF = cast<MCLEBFragment>(F); OW->WriteBytes(LF.getContents().str()); break; } case MCFragment::FT_Org: { ++stats::EmittedOrgFragments; - MCOrgFragment &OF = cast<MCOrgFragment>(F); + const MCOrgFragment &OF = cast<MCOrgFragment>(F); for (uint64_t i = 0, e = FragmentSize; i != e; ++i) OW->Write8(uint8_t(OF.getValue())); @@ -510,7 +648,8 @@ static void writeFragment(const MCAssembler &Asm, const MCAsmLayout &Layout, } } - assert(OW->getStream().tell() - Start == FragmentSize); + assert(OW->getStream().tell() - Start == FragmentSize && + "The stream should advance by fragment size"); } void MCAssembler::writeSectionData(const MCSectionData *SD, @@ -528,7 +667,7 @@ void MCAssembler::writeSectionData(const MCSectionData *SD, // Check that we aren't trying to write a non-zero contents (or fixups) // into a virtual section. This is to support clients which use standard // directives to fill the contents of virtual sections. - MCDataFragment &DF = cast<MCDataFragment>(*it); + const MCDataFragment &DF = cast<MCDataFragment>(*it); assert(DF.fixup_begin() == DF.fixup_end() && "Cannot have fixups in virtual section!"); for (unsigned i = 0, e = DF.getContents().size(); i != e; ++i) @@ -605,9 +744,9 @@ void MCAssembler::Finish() { SD->setLayoutOrder(i); unsigned FragmentIndex = 0; - for (MCSectionData::iterator it2 = SD->begin(), - ie2 = SD->end(); it2 != ie2; ++it2) - it2->setLayoutOrder(FragmentIndex++); + for (MCSectionData::iterator iFrag = SD->begin(), iFragEnd = SD->end(); + iFrag != iFragEnd; ++iFrag) + iFrag->setLayoutOrder(FragmentIndex++); } // Layout until everything fits. @@ -635,9 +774,10 @@ void MCAssembler::Finish() { for (MCAssembler::iterator it = begin(), ie = end(); it != ie; ++it) { for (MCSectionData::iterator it2 = it->begin(), ie2 = it->end(); it2 != ie2; ++it2) { - MCEncodedFragment *F = dyn_cast<MCEncodedFragment>(it2); + MCEncodedFragmentWithFixups *F = + dyn_cast<MCEncodedFragmentWithFixups>(it2); if (F) { - for (MCEncodedFragment::fixup_iterator it3 = F->fixup_begin(), + for (MCEncodedFragmentWithFixups::fixup_iterator it3 = F->fixup_begin(), ie3 = F->fixup_end(); it3 != ie3; ++it3) { MCFixup &Fixup = *it3; uint64_t FixedValue = handleFixup(Layout, *F, Fixup); @@ -655,11 +795,8 @@ void MCAssembler::Finish() { } bool MCAssembler::fixupNeedsRelaxation(const MCFixup &Fixup, - const MCInstFragment *DF, + const MCRelaxableFragment *DF, const MCAsmLayout &Layout) const { - if (getRelaxAll()) - return true; - // If we cannot resolve the fixup value, it requires relaxation. MCValue Target; uint64_t Value; @@ -669,25 +806,25 @@ bool MCAssembler::fixupNeedsRelaxation(const MCFixup &Fixup, return getBackend().fixupNeedsRelaxation(Fixup, Value, DF, Layout); } -bool MCAssembler::fragmentNeedsRelaxation(const MCInstFragment *IF, +bool MCAssembler::fragmentNeedsRelaxation(const MCRelaxableFragment *F, const MCAsmLayout &Layout) const { // If this inst doesn't ever need relaxation, ignore it. This occurs when we // are intentionally pushing out inst fragments, or because we relaxed a // previous instruction to one that doesn't need relaxation. - if (!getBackend().mayNeedRelaxation(IF->getInst())) + if (!getBackend().mayNeedRelaxation(F->getInst())) return false; - for (MCInstFragment::const_fixup_iterator it = IF->fixup_begin(), - ie = IF->fixup_end(); it != ie; ++it) - if (fixupNeedsRelaxation(*it, IF, Layout)) + for (MCRelaxableFragment::const_fixup_iterator it = F->fixup_begin(), + ie = F->fixup_end(); it != ie; ++it) + if (fixupNeedsRelaxation(*it, F, Layout)) return true; return false; } bool MCAssembler::relaxInstruction(MCAsmLayout &Layout, - MCInstFragment &IF) { - if (!fragmentNeedsRelaxation(&IF, Layout)) + MCRelaxableFragment &F) { + if (!fragmentNeedsRelaxation(&F, Layout)) return false; ++stats::RelaxedInstructions; @@ -698,7 +835,7 @@ bool MCAssembler::relaxInstruction(MCAsmLayout &Layout, // Relax the fragment. MCInst Relaxed; - getBackend().relaxInstruction(IF.getInst(), Relaxed); + getBackend().relaxInstruction(F.getInst(), Relaxed); // Encode the new instruction. // @@ -710,10 +847,10 @@ bool MCAssembler::relaxInstruction(MCAsmLayout &Layout, getEmitter().EncodeInstruction(Relaxed, VecOS, Fixups); VecOS.flush(); - // Update the instruction fragment. - IF.setInst(Relaxed); - IF.getContents() = Code; - IF.getFixups() = Fixups; + // Update the fragment. + F.setInst(Relaxed); + F.getContents() = Code; + F.getFixups() = Fixups; return true; } @@ -770,17 +907,21 @@ bool MCAssembler::relaxDwarfCallFrameFragment(MCAsmLayout &Layout, bool MCAssembler::layoutSectionOnce(MCAsmLayout &Layout, MCSectionData &SD) { // Holds the first fragment which needed relaxing during this layout. It will // remain NULL if none were relaxed. - MCFragment *FirstInvalidFragment = NULL; + // When a fragment is relaxed, all the fragments following it should get + // invalidated because their offset is going to change. + MCFragment *FirstRelaxedFragment = NULL; - // Scan for fragments that need relaxation. + // Attempt to relax all the fragments in the section. for (MCSectionData::iterator I = SD.begin(), IE = SD.end(); I != IE; ++I) { // Check if this is a fragment that needs relaxation. bool RelaxedFrag = false; switch(I->getKind()) { default: break; - case MCFragment::FT_Inst: - RelaxedFrag = relaxInstruction(Layout, *cast<MCInstFragment>(I)); + case MCFragment::FT_Relaxable: + assert(!getRelaxAll() && + "Did not expect a MCRelaxableFragment in RelaxAll mode"); + RelaxedFrag = relaxInstruction(Layout, *cast<MCRelaxableFragment>(I)); break; case MCFragment::FT_Dwarf: RelaxedFrag = relaxDwarfLineAddr(Layout, @@ -795,11 +936,11 @@ bool MCAssembler::layoutSectionOnce(MCAsmLayout &Layout, MCSectionData &SD) { RelaxedFrag = relaxLEB(Layout, *cast<MCLEBFragment>(I)); break; } - if (RelaxedFrag && !FirstInvalidFragment) - FirstInvalidFragment = I; + if (RelaxedFrag && !FirstRelaxedFragment) + FirstRelaxedFragment = I; } - if (FirstInvalidFragment) { - Layout.invalidateFragmentsAfter(FirstInvalidFragment); + if (FirstRelaxedFragment) { + Layout.invalidateFragmentsFrom(FirstRelaxedFragment); return true; } return false; @@ -846,8 +987,10 @@ void MCFragment::dump() { switch (getKind()) { case MCFragment::FT_Align: OS << "MCAlignFragment"; break; case MCFragment::FT_Data: OS << "MCDataFragment"; break; + case MCFragment::FT_CompactEncodedInst: + OS << "MCCompactEncodedInstFragment"; break; case MCFragment::FT_Fill: OS << "MCFillFragment"; break; - case MCFragment::FT_Inst: OS << "MCInstFragment"; break; + case MCFragment::FT_Relaxable: OS << "MCRelaxableFragment"; break; case MCFragment::FT_Org: OS << "MCOrgFragment"; break; case MCFragment::FT_Dwarf: OS << "MCDwarfFragment"; break; case MCFragment::FT_DwarfFrame: OS << "MCDwarfCallFrameFragment"; break; @@ -855,7 +998,9 @@ void MCFragment::dump() { } OS << "<MCFragment " << (void*) this << " LayoutOrder:" << LayoutOrder - << " Offset:" << Offset << ">"; + << " Offset:" << Offset + << " HasInstructions:" << hasInstructions() + << " BundlePadding:" << static_cast<unsigned>(getBundlePadding()) << ">"; switch (getKind()) { case MCFragment::FT_Align: { @@ -891,17 +1036,30 @@ void MCFragment::dump() { } break; } + case MCFragment::FT_CompactEncodedInst: { + const MCCompactEncodedInstFragment *CEIF = + cast<MCCompactEncodedInstFragment>(this); + OS << "\n "; + OS << " Contents:["; + const SmallVectorImpl<char> &Contents = CEIF->getContents(); + for (unsigned i = 0, e = Contents.size(); i != e; ++i) { + if (i) OS << ","; + OS << hexdigit((Contents[i] >> 4) & 0xF) << hexdigit(Contents[i] & 0xF); + } + OS << "] (" << Contents.size() << " bytes)"; + break; + } case MCFragment::FT_Fill: { const MCFillFragment *FF = cast<MCFillFragment>(this); OS << " Value:" << FF->getValue() << " ValueSize:" << FF->getValueSize() << " Size:" << FF->getSize(); break; } - case MCFragment::FT_Inst: { - const MCInstFragment *IF = cast<MCInstFragment>(this); + case MCFragment::FT_Relaxable: { + const MCRelaxableFragment *F = cast<MCRelaxableFragment>(this); OS << "\n "; OS << " Inst:"; - IF->getInst().dump_pretty(OS); + F->getInst().dump_pretty(OS); break; } case MCFragment::FT_Org: { @@ -937,7 +1095,8 @@ void MCSectionData::dump() { raw_ostream &OS = llvm::errs(); OS << "<MCSectionData"; - OS << " Alignment:" << getAlignment() << " Fragments:[\n "; + OS << " Alignment:" << getAlignment() + << " Fragments:[\n "; for (iterator it = begin(), ie = end(); it != ie; ++it) { if (it != begin()) OS << ",\n "; it->dump(); @@ -983,8 +1142,10 @@ void MCAssembler::dump() { // anchors for MC*Fragment vtables void MCEncodedFragment::anchor() { } +void MCEncodedFragmentWithFixups::anchor() { } void MCDataFragment::anchor() { } -void MCInstFragment::anchor() { } +void MCCompactEncodedInstFragment::anchor() { } +void MCRelaxableFragment::anchor() { } void MCAlignFragment::anchor() { } void MCFillFragment::anchor() { } void MCOrgFragment::anchor() { } diff --git a/lib/MC/MCContext.cpp b/lib/MC/MCContext.cpp index 23ec0bb12d..26d378e6c0 100644 --- a/lib/MC/MCContext.cpp +++ b/lib/MC/MCContext.cpp @@ -21,6 +21,7 @@ #include "llvm/MC/MCSymbol.h" #include "llvm/Support/ELF.h" #include "llvm/Support/ErrorHandling.h" +#include "llvm/Support/MemoryBuffer.h" #include "llvm/Support/Signals.h" #include "llvm/Support/SourceMgr.h" using namespace llvm; @@ -32,13 +33,15 @@ typedef StringMap<const MCSectionCOFF*> COFFUniqueMapTy; MCContext::MCContext(const MCAsmInfo &mai, const MCRegisterInfo &mri, const MCObjectFileInfo *mofi, const SourceMgr *mgr, - bool DoAutoInitializationFinalization ) : + bool DoAutoReset) : SrcMgr(mgr), MAI(mai), MRI(mri), MOFI(mofi), Allocator(), Symbols(Allocator), UsedNames(Allocator), NextUniqueID(0), - CurrentDwarfLoc(0,0,0,DWARF2_FLAG_IS_STMT,0,0), - AllowTemporaryLabels(true), - AutoInitializationFinalization(DoAutoInitializationFinalization) { + CompilationDir(llvm::sys::Path::GetCurrentDirectory().str()), + CurrentDwarfLoc(0,0,0,DWARF2_FLAG_IS_STMT,0,0), + DwarfLocSeen(false), GenDwarfForAssembly(false), GenDwarfFileNumber(0), + AllowTemporaryLabels(true), DwarfCompileUnitID(0), AutoReset(DoAutoReset) { + MachOUniquingMap = 0; ELFUniquingMap = 0; COFFUniquingMap = 0; @@ -47,14 +50,16 @@ MCContext::MCContext(const MCAsmInfo &mai, const MCRegisterInfo &mri, SecureLog = 0; SecureLogUsed = false; - if (AutoInitializationFinalization) - doInitialization(); + if (SrcMgr && SrcMgr->getNumBuffers() > 0) + MainFileName = SrcMgr->getMemoryBuffer(0)->getBufferIdentifier(); + else + MainFileName = ""; } MCContext::~MCContext() { - if (AutoInitializationFinalization) - doFinalization(); + if (AutoReset) + reset(); // NOTE: The symbols are all allocated out of a bump pointer allocator, // we don't need to free them here. @@ -67,25 +72,19 @@ MCContext::~MCContext() { // Module Lifetime Management //===----------------------------------------------------------------------===// -void MCContext::doInitialization() { - NextUniqueID = 0; - AllowTemporaryLabels = true; - DwarfLocSeen = false; - GenDwarfForAssembly = false; - GenDwarfFileNumber = 0; -} - -void MCContext::doFinalization() { +void MCContext::reset() { UsedNames.clear(); Symbols.clear(); Allocator.Reset(); Instances.clear(); - MCDwarfFiles.clear(); - MCDwarfDirs.clear(); + MCDwarfFilesCUMap.clear(); + MCDwarfDirsCUMap.clear(); MCGenDwarfLabelEntries.clear(); DwarfDebugFlags = StringRef(); MCLineSections.clear(); MCLineSectionOrder.clear(); + DwarfCompileUnitID = 0; + MCLineTableSymbols.clear(); CurrentDwarfLoc = MCDwarfLoc(0,0,0,DWARF2_FLAG_IS_STMT,0,0); // If we have the MachO uniquing map, free it. @@ -95,6 +94,12 @@ void MCContext::doFinalization() { MachOUniquingMap = 0; ELFUniquingMap = 0; COFFUniquingMap = 0; + + NextUniqueID = 0; + AllowTemporaryLabels = true; + DwarfLocSeen = false; + GenDwarfForAssembly = false; + GenDwarfFileNumber = 0; } //===----------------------------------------------------------------------===// @@ -294,11 +299,13 @@ const MCSection *MCContext::getCOFFSection(StringRef Section, /// error and zero is returned and the client reports the error, else the /// allocated file number is returned. The file numbers may be in any order. unsigned MCContext::GetDwarfFile(StringRef Directory, StringRef FileName, - unsigned FileNumber) { + unsigned FileNumber, unsigned CUID) { // TODO: a FileNumber of zero says to use the next available file number. // Note: in GenericAsmParser::ParseDirectiveFile() FileNumber was checked // to not be less than one. This needs to be change to be not less than zero. + std::vector<MCDwarfFile *>& MCDwarfFiles = MCDwarfFilesCUMap[CUID]; + std::vector<StringRef>& MCDwarfDirs = MCDwarfDirsCUMap[CUID]; // Make space for this FileNumber in the MCDwarfFiles vector if needed. if (FileNumber >= MCDwarfFiles.size()) { MCDwarfFiles.resize(FileNumber + 1); @@ -358,7 +365,8 @@ unsigned MCContext::GetDwarfFile(StringRef Directory, StringRef FileName, /// isValidDwarfFileNumber - takes a dwarf file number and returns true if it /// currently is assigned and false otherwise. -bool MCContext::isValidDwarfFileNumber(unsigned FileNumber) { +bool MCContext::isValidDwarfFileNumber(unsigned FileNumber, unsigned CUID) { + std::vector<MCDwarfFile *>& MCDwarfFiles = MCDwarfFilesCUMap[CUID]; if(FileNumber == 0 || FileNumber >= MCDwarfFiles.size()) return false; diff --git a/lib/MC/MCDisassembler/CMakeLists.txt b/lib/MC/MCDisassembler/CMakeLists.txt index 5e2cd8387d..5195b9e23d 100644 --- a/lib/MC/MCDisassembler/CMakeLists.txt +++ b/lib/MC/MCDisassembler/CMakeLists.txt @@ -1,8 +1,3 @@ add_llvm_library(LLVMMCDisassembler Disassembler.cpp - EDDisassembler.cpp - EDInst.cpp - EDMain.cpp - EDOperand.cpp - EDToken.cpp ) diff --git a/lib/MC/MCDisassembler/Disassembler.cpp b/lib/MC/MCDisassembler/Disassembler.cpp index ac583ac127..d3fa906a06 100644 --- a/lib/MC/MCDisassembler/Disassembler.cpp +++ b/lib/MC/MCDisassembler/Disassembler.cpp @@ -195,5 +195,21 @@ int LLVMSetDisasmOptions(LLVMDisasmContextRef DCR, uint64_t Options){ IP->setPrintImmHex(1); Options &= ~LLVMDisassembler_Option_PrintImmHex; } + if (Options & LLVMDisassembler_Option_AsmPrinterVariant){ + LLVMDisasmContext *DC = (LLVMDisasmContext *)DCR; + // Try to set up the new instruction printer. + const MCAsmInfo *MAI = DC->getAsmInfo(); + const MCInstrInfo *MII = DC->getInstrInfo(); + const MCRegisterInfo *MRI = DC->getRegisterInfo(); + const MCSubtargetInfo *STI = DC->getSubtargetInfo(); + int AsmPrinterVariant = MAI->getAssemblerDialect(); + AsmPrinterVariant = AsmPrinterVariant == 0 ? 1 : 0; + MCInstPrinter *IP = DC->getTarget()->createMCInstPrinter( + AsmPrinterVariant, *MAI, *MII, *MRI, *STI); + if (IP) { + DC->setIP(IP); + Options &= ~LLVMDisassembler_Option_AsmPrinterVariant; + } + } return (Options == 0); } diff --git a/lib/MC/MCDisassembler/Disassembler.h b/lib/MC/MCDisassembler/Disassembler.h index 28cf04b95d..6eb59d0c57 100644 --- a/lib/MC/MCDisassembler/Disassembler.h +++ b/lib/MC/MCDisassembler/Disassembler.h @@ -109,7 +109,11 @@ public: const Target *getTarget() const { return TheTarget; } const MCDisassembler *getDisAsm() const { return DisAsm.get(); } const MCAsmInfo *getAsmInfo() const { return MAI.get(); } + const MCInstrInfo *getInstrInfo() const { return MII.get(); } + const MCRegisterInfo *getRegisterInfo() const { return MRI.get(); } + const MCSubtargetInfo *getSubtargetInfo() const { return MSI.get(); } MCInstPrinter *getIP() { return IP.get(); } + void setIP(MCInstPrinter *NewIP) { IP.reset(NewIP); } }; } // namespace llvm diff --git a/lib/MC/MCDisassembler/EDDisassembler.cpp b/lib/MC/MCDisassembler/EDDisassembler.cpp deleted file mode 100644 index e667920791..0000000000 --- a/lib/MC/MCDisassembler/EDDisassembler.cpp +++ /dev/null @@ -1,400 +0,0 @@ -//===-EDDisassembler.cpp - LLVM Enhanced Disassembler ---------------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file implements the Enhanced Disassembly library's disassembler class. -// The disassembler is responsible for vending individual instructions according -// to a given architecture and disassembly syntax. -// -//===----------------------------------------------------------------------===// - -#include "EDDisassembler.h" -#include "EDInst.h" -#include "llvm/MC/EDInstInfo.h" -#include "llvm/MC/MCAsmInfo.h" -#include "llvm/MC/MCContext.h" -#include "llvm/MC/MCDisassembler.h" -#include "llvm/MC/MCExpr.h" -#include "llvm/MC/MCInst.h" -#include "llvm/MC/MCInstPrinter.h" -#include "llvm/MC/MCInstrInfo.h" -#include "llvm/MC/MCParser/AsmLexer.h" -#include "llvm/MC/MCParser/MCAsmParser.h" -#include "llvm/MC/MCParser/MCParsedAsmOperand.h" -#include "llvm/MC/MCRegisterInfo.h" -#include "llvm/MC/MCStreamer.h" -#include "llvm/MC/MCSubtargetInfo.h" -#include "llvm/MC/MCTargetAsmLexer.h" -#include "llvm/MC/MCTargetAsmParser.h" -#include "llvm/Support/MemoryBuffer.h" -#include "llvm/Support/MemoryObject.h" -#include "llvm/Support/SourceMgr.h" -#include "llvm/Support/TargetRegistry.h" -using namespace llvm; - -EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers; - -struct TripleMap { - Triple::ArchType Arch; - const char *String; -}; - -static const struct TripleMap triplemap[] = { - { Triple::x86, "i386-unknown-unknown" }, - { Triple::x86_64, "x86_64-unknown-unknown" }, - { Triple::arm, "arm-unknown-unknown" }, - { Triple::thumb, "thumb-unknown-unknown" } -}; - -/// infoFromArch - Returns the TripleMap corresponding to a given architecture, -/// or NULL if there is an error -/// -/// @arg arch - The Triple::ArchType for the desired architecture -static const char *tripleFromArch(Triple::ArchType arch) { - unsigned int infoIndex; - - for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) { - if (arch == triplemap[infoIndex].Arch) - return triplemap[infoIndex].String; - } - - return NULL; -} - -/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer -/// for the desired assembly syntax, suitable for passing to -/// Target::createMCInstPrinter() -/// -/// @arg arch - The target architecture -/// @arg syntax - The assembly syntax in sd form -static int getLLVMSyntaxVariant(Triple::ArchType arch, - EDDisassembler::AssemblySyntax syntax) { - switch (syntax) { - // Mappings below from X86AsmPrinter.cpp - case EDDisassembler::kEDAssemblySyntaxX86ATT: - if (arch == Triple::x86 || arch == Triple::x86_64) - return 0; - break; - case EDDisassembler::kEDAssemblySyntaxX86Intel: - if (arch == Triple::x86 || arch == Triple::x86_64) - return 1; - break; - case EDDisassembler::kEDAssemblySyntaxARMUAL: - if (arch == Triple::arm || arch == Triple::thumb) - return 0; - break; - } - - return -1; -} - -EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch, - AssemblySyntax syntax) { - const char *triple = tripleFromArch(arch); - return getDisassembler(StringRef(triple), syntax); -} - -EDDisassembler *EDDisassembler::getDisassembler(StringRef str, - AssemblySyntax syntax) { - CPUKey key; - key.Triple = str.str(); - key.Syntax = syntax; - - EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key); - - if (i != sDisassemblers.end()) { - return i->second; - } - - EDDisassembler *sdd = new EDDisassembler(key); - if (!sdd->valid()) { - delete sdd; - return NULL; - } - - sDisassemblers[key] = sdd; - - return sdd; -} - -EDDisassembler::EDDisassembler(CPUKey &key) : - Valid(false), - HasSemantics(false), - ErrorStream(nulls()), - Key(key), - TgtTriple(key.Triple.c_str()) { - - LLVMSyntaxVariant = getLLVMSyntaxVariant(TgtTriple.getArch(), key.Syntax); - - if (LLVMSyntaxVariant < 0) - return; - - std::string tripleString(key.Triple); - std::string errorString; - - Tgt = TargetRegistry::lookupTarget(key.Triple, - errorString); - - if (!Tgt) - return; - - MRI.reset(Tgt->createMCRegInfo(tripleString)); - - if (!MRI) - return; - - initMaps(*MRI); - - AsmInfo.reset(Tgt->createMCAsmInfo(tripleString)); - - if (!AsmInfo) - return; - - STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", "")); - - if (!STI) - return; - - Disassembler.reset(Tgt->createMCDisassembler(*STI)); - - if (!Disassembler) - return; - - InstInfos = Disassembler->getEDInfo(); - - MII.reset(Tgt->createMCInstrInfo()); - - if (!MII) - return; - - InstString.reset(new std::string); - InstStream.reset(new raw_string_ostream(*InstString)); - InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, - *MII, *MRI, *STI)); - - if (!InstPrinter) - return; - - GenericAsmLexer.reset(new AsmLexer(*AsmInfo)); - SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo)); - SpecificAsmLexer->InstallLexer(*GenericAsmLexer); - - initMaps(*MRI); - - Valid = true; -} - -EDDisassembler::~EDDisassembler() { - if (!valid()) - return; -} - -namespace { - /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback - /// as provided by the sd interface. See MemoryObject. - class EDMemoryObject : public llvm::MemoryObject { - private: - EDByteReaderCallback Callback; - void *Arg; - public: - EDMemoryObject(EDByteReaderCallback callback, - void *arg) : Callback(callback), Arg(arg) { } - ~EDMemoryObject() { } - uint64_t getBase() const { return 0x0; } - uint64_t getExtent() const { return (uint64_t)-1; } - int readByte(uint64_t address, uint8_t *ptr) const { - if (!Callback) - return -1; - - if (Callback(ptr, address, Arg)) - return -1; - - return 0; - } - }; -} - -EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader, - uint64_t address, - void *arg) { - EDMemoryObject memoryObject(byteReader, arg); - - MCInst* inst = new MCInst; - uint64_t byteSize; - - MCDisassembler::DecodeStatus S; - S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address, - ErrorStream, nulls()); - switch (S) { - case MCDisassembler::Fail: - case MCDisassembler::SoftFail: - // FIXME: Do something different on soft failure mode? - delete inst; - return NULL; - - case MCDisassembler::Success: { - const llvm::EDInstInfo *thisInstInfo = NULL; - - if (InstInfos) { - thisInstInfo = &InstInfos[inst->getOpcode()]; - } - - EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo); - return sdInst; - } - } - return NULL; -} - -void EDDisassembler::initMaps(const MCRegisterInfo ®isterInfo) { - unsigned numRegisters = registerInfo.getNumRegs(); - unsigned registerIndex; - - for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) { - const char* registerName = registerInfo.getName(registerIndex); - - RegVec.push_back(registerName); - RegRMap[registerName] = registerIndex; - } - - switch (TgtTriple.getArch()) { - default: - break; - case Triple::x86: - case Triple::x86_64: - stackPointers.insert(registerIDWithName("SP")); - stackPointers.insert(registerIDWithName("ESP")); - stackPointers.insert(registerIDWithName("RSP")); - - programCounters.insert(registerIDWithName("IP")); - programCounters.insert(registerIDWithName("EIP")); - programCounters.insert(registerIDWithName("RIP")); - break; - case Triple::arm: - case Triple::thumb: - stackPointers.insert(registerIDWithName("SP")); - - programCounters.insert(registerIDWithName("PC")); - break; - } -} - -const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const { - if (registerID >= RegVec.size()) - return NULL; - else - return RegVec[registerID].c_str(); -} - -unsigned EDDisassembler::registerIDWithName(const char *name) const { - regrmap_t::const_iterator iter = RegRMap.find(std::string(name)); - if (iter == RegRMap.end()) - return 0; - else - return (*iter).second; -} - -bool EDDisassembler::registerIsStackPointer(unsigned registerID) { - return (stackPointers.find(registerID) != stackPointers.end()); -} - -bool EDDisassembler::registerIsProgramCounter(unsigned registerID) { - return (programCounters.find(registerID) != programCounters.end()); -} - -int EDDisassembler::printInst(std::string &str, MCInst &inst) { - PrinterMutex.acquire(); - - InstPrinter->printInst(&inst, *InstStream, ""); - InstStream->flush(); - str = *InstString; - InstString->clear(); - - PrinterMutex.release(); - - return 0; -} - -static void diag_handler(const SMDiagnostic &diag, void *context) { - if (context) - diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream); -} - -int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands, - SmallVectorImpl<AsmToken> &tokens, - const std::string &str) { - int ret = 0; - - switch (TgtTriple.getArch()) { - default: - return -1; - case Triple::x86: - case Triple::x86_64: - case Triple::arm: - case Triple::thumb: - break; - } - - const char *cStr = str.c_str(); - MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr)); - - StringRef instName; - SMLoc instLoc; - - SourceMgr sourceMgr; - sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this)); - sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over - MCContext context(*AsmInfo, *MRI, NULL); - OwningPtr<MCStreamer> streamer(createNullStreamer(context)); - OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr, - context, *streamer, - *AsmInfo)); - - OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", "")); - OwningPtr<MCTargetAsmParser> - TargetParser(Tgt->createMCAsmParser(*STI, *genericParser)); - - AsmToken OpcodeToken = genericParser->Lex(); - AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to - - if (OpcodeToken.is(AsmToken::Identifier)) { - instName = OpcodeToken.getString(); - instLoc = OpcodeToken.getLoc(); - - ParseInstructionInfo Info; - if (NextToken.isNot(AsmToken::Eof) && - TargetParser->ParseInstruction(Info, instName, instLoc, operands)) - ret = -1; - } else { - ret = -1; - } - - ParserMutex.acquire(); - - if (!ret) { - GenericAsmLexer->setBuffer(buf); - - while (SpecificAsmLexer->Lex(), - SpecificAsmLexer->isNot(AsmToken::Eof) && - SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) { - if (SpecificAsmLexer->is(AsmToken::Error)) { - ret = -1; - break; - } - tokens.push_back(SpecificAsmLexer->getTok()); - } - } - - ParserMutex.release(); - - return ret; -} - -int EDDisassembler::llvmSyntaxVariant() const { - return LLVMSyntaxVariant; -} diff --git a/lib/MC/MCDisassembler/EDDisassembler.h b/lib/MC/MCDisassembler/EDDisassembler.h deleted file mode 100644 index 942b9067e6..0000000000 --- a/lib/MC/MCDisassembler/EDDisassembler.h +++ /dev/null @@ -1,269 +0,0 @@ -//===-- EDDisassembler.h - LLVM Enhanced Disassembler -----------*- C++ -*-===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file defines the interface for the Enhanced Disassembly library's -// disassembler class. The disassembler is responsible for vending individual -// instructions according to a given architecture and disassembly syntax. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_EDDISASSEMBLER_H -#define LLVM_EDDISASSEMBLER_H - -#include "EDInfo.h" -#include "llvm/ADT/OwningPtr.h" -#include "llvm/ADT/Triple.h" -#include "llvm/Support/Mutex.h" -#include "llvm/Support/raw_ostream.h" -#include <map> -#include <set> -#include <string> -#include <vector> - -namespace llvm { -class AsmLexer; -class AsmParser; -class AsmToken; -class MCContext; -class MCAsmInfo; -class MCAsmLexer; -class MCDisassembler; -class MCInst; -class MCInstPrinter; -class MCInstrInfo; -class MCParsedAsmOperand; -class MCRegisterInfo; -class MCStreamer; -class MCSubtargetInfo; -class MCTargetAsmLexer; -class MCTargetAsmParser; -template <typename T> class SmallVectorImpl; -class SourceMgr; -class Target; - -struct EDInstInfo; -struct EDInst; -struct EDOperand; -struct EDToken; - -typedef int (*EDByteReaderCallback)(uint8_t *byte, uint64_t address, void *arg); - -/// EDDisassembler - Encapsulates a disassembler for a single architecture and -/// disassembly syntax. Also manages the static disassembler registry. -struct EDDisassembler { - typedef enum { - /*! @constant kEDAssemblySyntaxX86Intel Intel syntax for i386 and x86_64. */ - kEDAssemblySyntaxX86Intel = 0, - /*! @constant kEDAssemblySyntaxX86ATT AT&T syntax for i386 and x86_64. */ - kEDAssemblySyntaxX86ATT = 1, - kEDAssemblySyntaxARMUAL = 2 - } AssemblySyntax; - - - //////////////////// - // Static members // - //////////////////// - - /// CPUKey - Encapsulates the descriptor of an architecture/disassembly-syntax - /// pair - struct CPUKey { - /// The architecture type - std::string Triple; - - /// The assembly syntax - AssemblySyntax Syntax; - - /// operator== - Equality operator - bool operator==(const CPUKey &key) const { - return (Triple == key.Triple && - Syntax == key.Syntax); - } - - /// operator< - Less-than operator - bool operator<(const CPUKey &key) const { - return ((Triple < key.Triple) || - ((Triple == key.Triple) && Syntax < (key.Syntax))); - } - }; - - typedef std::map<CPUKey, EDDisassembler*> DisassemblerMap_t; - - /// A map from disassembler specifications to disassemblers. Populated - /// lazily. - static DisassemblerMap_t sDisassemblers; - - /// getDisassembler - Returns the specified disassemble, or NULL on failure - /// - /// @arg arch - The desired architecture - /// @arg syntax - The desired disassembly syntax - static EDDisassembler *getDisassembler(llvm::Triple::ArchType arch, - AssemblySyntax syntax); - - /// getDisassembler - Returns the disassembler for a given combination of - /// CPU type, CPU subtype, and assembly syntax, or NULL on failure - /// - /// @arg str - The string representation of the architecture triple, e.g., - /// "x86_64-apple-darwin" - /// @arg syntax - The disassembly syntax for the required disassembler - static EDDisassembler *getDisassembler(llvm::StringRef str, - AssemblySyntax syntax); - - //////////////////////// - // Per-object members // - //////////////////////// - - /// True only if the object has been successfully initialized - bool Valid; - /// True if the disassembler can provide semantic information - bool HasSemantics; - - /// The stream to write errors to - llvm::raw_ostream &ErrorStream; - - /// The triple/syntax pair for the current architecture - CPUKey Key; - /// The Triple fur the current architecture - Triple TgtTriple; - /// The LLVM target corresponding to the disassembler - const llvm::Target *Tgt; - /// The assembly information for the target architecture - llvm::OwningPtr<const llvm::MCAsmInfo> AsmInfo; - /// The subtarget information for the target architecture - llvm::OwningPtr<const llvm::MCSubtargetInfo> STI; - // The instruction information for the target architecture. - llvm::OwningPtr<const llvm::MCInstrInfo> MII; - // The register information for the target architecture. - llvm::OwningPtr<const llvm::MCRegisterInfo> MRI; - /// The disassembler for the target architecture - llvm::OwningPtr<const llvm::MCDisassembler> Disassembler; - /// The output string for the instruction printer; must be guarded with - /// PrinterMutex - llvm::OwningPtr<std::string> InstString; - /// The output stream for the disassembler; must be guarded with - /// PrinterMutex - llvm::OwningPtr<llvm::raw_string_ostream> InstStream; - /// The instruction printer for the target architecture; must be guarded with - /// PrinterMutex when printing - llvm::OwningPtr<llvm::MCInstPrinter> InstPrinter; - /// The mutex that guards the instruction printer's printing functions, which - /// use a shared stream - llvm::sys::Mutex PrinterMutex; - /// The array of instruction information provided by the TableGen backend for - /// the target architecture - const llvm::EDInstInfo *InstInfos; - /// The target-specific lexer for use in tokenizing strings, in - /// target-independent and target-specific portions - llvm::OwningPtr<llvm::AsmLexer> GenericAsmLexer; - llvm::OwningPtr<llvm::MCTargetAsmLexer> SpecificAsmLexer; - /// The guard for the above - llvm::sys::Mutex ParserMutex; - /// The LLVM number used for the target disassembly syntax variant - int LLVMSyntaxVariant; - - typedef std::vector<std::string> regvec_t; - typedef std::map<std::string, unsigned> regrmap_t; - - /// A vector of registers for quick mapping from LLVM register IDs to names - regvec_t RegVec; - /// A map of registers for quick mapping from register names to LLVM IDs - regrmap_t RegRMap; - - /// A set of register IDs for aliases of the stack pointer for the current - /// architecture - std::set<unsigned> stackPointers; - /// A set of register IDs for aliases of the program counter for the current - /// architecture - std::set<unsigned> programCounters; - - /// Constructor - initializes a disassembler with all the necessary objects, - /// which come pre-allocated from the registry accessor function - /// - /// @arg key - the architecture and disassembly syntax for the - /// disassembler - EDDisassembler(CPUKey& key); - - /// valid - reports whether there was a failure in the constructor. - bool valid() { - return Valid; - } - - /// hasSemantics - reports whether the disassembler can provide operands and - /// tokens. - bool hasSemantics() { - return HasSemantics; - } - - ~EDDisassembler(); - - /// createInst - creates and returns an instruction given a callback and - /// memory address, or NULL on failure - /// - /// @arg byteReader - A callback function that provides machine code bytes - /// @arg address - The address of the first byte of the instruction, - /// suitable for passing to byteReader - /// @arg arg - An opaque argument for byteReader - EDInst *createInst(EDByteReaderCallback byteReader, - uint64_t address, - void *arg); - - /// initMaps - initializes regVec and regRMap using the provided register - /// info - /// - /// @arg registerInfo - the register information to use as a source - void initMaps(const llvm::MCRegisterInfo ®isterInfo); - /// nameWithRegisterID - Returns the name (owned by the EDDisassembler) of a - /// register for a given register ID, or NULL on failure - /// - /// @arg registerID - the ID of the register to be queried - const char *nameWithRegisterID(unsigned registerID) const; - /// registerIDWithName - Returns the ID of a register for a given register - /// name, or (unsigned)-1 on failure - /// - /// @arg name - The name of the register - unsigned registerIDWithName(const char *name) const; - - /// registerIsStackPointer - reports whether a register ID is an alias for the - /// stack pointer register - /// - /// @arg registerID - The LLVM register ID - bool registerIsStackPointer(unsigned registerID); - /// registerIsStackPointer - reports whether a register ID is an alias for the - /// stack pointer register - /// - /// @arg registerID - The LLVM register ID - bool registerIsProgramCounter(unsigned registerID); - - /// printInst - prints an MCInst to a string, returning 0 on success, or -1 - /// otherwise - /// - /// @arg str - A reference to a string which is filled in with the string - /// representation of the instruction - /// @arg inst - A reference to the MCInst to be printed - int printInst(std::string& str, - llvm::MCInst& inst); - - /// parseInst - extracts operands and tokens from a string for use in - /// tokenizing the string. Returns 0 on success, or -1 otherwise. - /// - /// @arg operands - A reference to a vector that will be filled in with the - /// parsed operands - /// @arg tokens - A reference to a vector that will be filled in with the - /// tokens - /// @arg str - The string representation of the instruction - int parseInst(llvm::SmallVectorImpl<llvm::MCParsedAsmOperand*> &operands, - llvm::SmallVectorImpl<llvm::AsmToken> &tokens, - const std::string &str); - - /// llvmSyntaxVariant - returns the LLVM syntax variant for this disassembler - int llvmSyntaxVariant() const; -}; - -} // end namespace llvm - -#endif diff --git a/lib/MC/MCDisassembler/EDInfo.h b/lib/MC/MCDisassembler/EDInfo.h deleted file mode 100644 index e43ad16352..0000000000 --- a/lib/MC/MCDisassembler/EDInfo.h +++ /dev/null @@ -1,84 +0,0 @@ -//===-- EDInfo.h - LLVM Enhanced Disassembler -------------------*- C++ -*-===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_EDINFO_H -#define LLVM_EDINFO_H - -enum { - EDIS_MAX_OPERANDS = 13, - EDIS_MAX_SYNTAXES = 2 -}; - -enum OperandTypes { - kOperandTypeNone, - kOperandTypeImmediate, - kOperandTypeRegister, - kOperandTypeX86Memory, - kOperandTypeX86EffectiveAddress, - kOperandTypeX86PCRelative, - kOperandTypeARMBranchTarget, - kOperandTypeARMSoReg, - kOperandTypeARMSoImm, - kOperandTypeARMRotImm, - kOperandTypeARMSoImm2Part, - kOperandTypeARMPredicate, - kOperandTypeAddrModeImm12, - kOperandTypeLdStSOReg, - kOperandTypeARMAddrMode2, - kOperandTypeARMAddrMode2Offset, - kOperandTypeARMAddrMode3, - kOperandTypeARMAddrMode3Offset, - kOperandTypeARMAddrMode4, - kOperandTypeARMAddrMode5, - kOperandTypeARMAddrMode6, - kOperandTypeARMAddrMode6Offset, - kOperandTypeARMAddrMode7, - kOperandTypeARMAddrModePC, - kOperandTypeARMRegisterList, - kOperandTypeARMDPRRegisterList, - kOperandTypeARMSPRRegisterList, - kOperandTypeARMTBAddrMode, - kOperandTypeThumbITMask, - kOperandTypeThumbAddrModeRegS1, - kOperandTypeThumbAddrModeRegS2, - kOperandTypeThumbAddrModeRegS4, - kOperandTypeThumbAddrModeImmS1, - kOperandTypeThumbAddrModeImmS2, - kOperandTypeThumbAddrModeImmS4, - kOperandTypeThumbAddrModeRR, - kOperandTypeThumbAddrModeSP, - kOperandTypeThumbAddrModePC, - kOperandTypeThumb2AddrModeReg, - kOperandTypeThumb2SoReg, - kOperandTypeThumb2SoImm, - kOperandTypeThumb2AddrModeImm8, - kOperandTypeThumb2AddrModeImm8Offset, - kOperandTypeThumb2AddrModeImm12, - kOperandTypeThumb2AddrModeSoReg, - kOperandTypeThumb2AddrModeImm8s4, - kOperandTypeThumb2AddrModeImm8s4Offset -}; - -enum OperandFlags { - kOperandFlagSource = 0x1, - kOperandFlagTarget = 0x2 -}; - -enum InstructionTypes { - kInstructionTypeNone, - kInstructionTypeMove, - kInstructionTypeBranch, - kInstructionTypePush, - kInstructionTypePop, - kInstructionTypeCall, - kInstructionTypeReturn -}; - - -#endif diff --git a/lib/MC/MCDisassembler/EDInst.cpp b/lib/MC/MCDisassembler/EDInst.cpp deleted file mode 100644 index 4c4fdd2568..0000000000 --- a/lib/MC/MCDisassembler/EDInst.cpp +++ /dev/null @@ -1,211 +0,0 @@ -//===-EDInst.cpp - LLVM Enhanced Disassembler -----------------------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file implements the Enhanced Disassembly library's instruction class. -// The instruction is responsible for vending the string representation, -// individual tokens, and operands for a single instruction. -// -//===----------------------------------------------------------------------===// - -#include "EDInst.h" -#include "EDDisassembler.h" -#include "EDOperand.h" -#include "EDToken.h" -#include "llvm/MC/EDInstInfo.h" -#include "llvm/MC/MCInst.h" - -using namespace llvm; - -EDInst::EDInst(llvm::MCInst *inst, - uint64_t byteSize, - EDDisassembler &disassembler, - const llvm::EDInstInfo *info) : - Disassembler(disassembler), - Inst(inst), - ThisInstInfo(info), - ByteSize(byteSize), - BranchTarget(-1), - MoveSource(-1), - MoveTarget(-1) { - OperandOrder = ThisInstInfo->operandOrders[Disassembler.llvmSyntaxVariant()]; -} - -EDInst::~EDInst() { - unsigned int index; - unsigned int numOperands = Operands.size(); - - for (index = 0; index < numOperands; ++index) - delete Operands[index]; - - unsigned int numTokens = Tokens.size(); - - for (index = 0; index < numTokens; ++index) - delete Tokens[index]; - - delete Inst; -} - -uint64_t EDInst::byteSize() { - return ByteSize; -} - -int EDInst::stringify() { - if (StringifyResult.valid()) - return StringifyResult.result(); - - if (Disassembler.printInst(String, *Inst)) - return StringifyResult.setResult(-1); - - String.push_back('\n'); - - return StringifyResult.setResult(0); -} - -int EDInst::getString(const char*& str) { - if (stringify()) - return -1; - - str = String.c_str(); - - return 0; -} - -unsigned EDInst::instID() { - return Inst->getOpcode(); -} - -bool EDInst::isBranch() { - if (ThisInstInfo) - return - ThisInstInfo->instructionType == kInstructionTypeBranch || - ThisInstInfo->instructionType == kInstructionTypeCall; - else - return false; -} - -bool EDInst::isMove() { - if (ThisInstInfo) - return ThisInstInfo->instructionType == kInstructionTypeMove; - else - return false; -} - -int EDInst::parseOperands() { - if (ParseResult.valid()) - return ParseResult.result(); - - if (!ThisInstInfo) - return ParseResult.setResult(-1); - - unsigned int opIndex; - unsigned int mcOpIndex = 0; - - for (opIndex = 0; opIndex < ThisInstInfo->numOperands; ++opIndex) { - if (isBranch() && - (ThisInstInfo->operandFlags[opIndex] & kOperandFlagTarget)) { - BranchTarget = opIndex; - } - else if (isMove()) { - if (ThisInstInfo->operandFlags[opIndex] & kOperandFlagSource) - MoveSource = opIndex; - else if (ThisInstInfo->operandFlags[opIndex] & kOperandFlagTarget) - MoveTarget = opIndex; - } - - EDOperand *operand = new EDOperand(Disassembler, *this, opIndex, mcOpIndex); - - Operands.push_back(operand); - } - - return ParseResult.setResult(0); -} - -int EDInst::branchTargetID() { - if (parseOperands()) - return -1; - return BranchTarget; -} - -int EDInst::moveSourceID() { - if (parseOperands()) - return -1; - return MoveSource; -} - -int EDInst::moveTargetID() { - if (parseOperands()) - return -1; - return MoveTarget; -} - -int EDInst::numOperands() { - if (parseOperands()) - return -1; - return Operands.size(); -} - -int EDInst::getOperand(EDOperand *&operand, unsigned int index) { - if (parseOperands()) - return -1; - - if (index >= Operands.size()) - return -1; - - operand = Operands[index]; - return 0; -} - -int EDInst::tokenize() { - if (TokenizeResult.valid()) - return TokenizeResult.result(); - - if (ThisInstInfo == NULL) - return TokenizeResult.setResult(-1); - - if (stringify()) - return TokenizeResult.setResult(-1); - - return TokenizeResult.setResult(EDToken::tokenize(Tokens, - String, - OperandOrder, - Disassembler)); - -} - -int EDInst::numTokens() { - if (tokenize()) - return -1; - return Tokens.size(); -} - -int EDInst::getToken(EDToken *&token, unsigned int index) { - if (tokenize()) - return -1; - token = Tokens[index]; - return 0; -} - -#ifdef __BLOCKS__ -int EDInst::visitTokens(EDTokenVisitor_t visitor) { - if (tokenize()) - return -1; - - tokvec_t::iterator iter; - - for (iter = Tokens.begin(); iter != Tokens.end(); ++iter) { - int ret = visitor(*iter); - if (ret == 1) - return 0; - if (ret != 0) - return -1; - } - - return 0; -} -#endif diff --git a/lib/MC/MCDisassembler/EDInst.h b/lib/MC/MCDisassembler/EDInst.h deleted file mode 100644 index cc0b562130..0000000000 --- a/lib/MC/MCDisassembler/EDInst.h +++ /dev/null @@ -1,182 +0,0 @@ -//===-- EDInst.h - LLVM Enhanced Disassembler -------------------*- C++ -*-===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file defines the interface for the Enhanced Disassembly library's -// instruction class. The instruction is responsible for vending the string -// representation, individual tokens and operands for a single instruction. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_EDINST_H -#define LLVM_EDINST_H - -#include "llvm/ADT/SmallVector.h" -#include "llvm/Support/DataTypes.h" -#include <string> -#include <vector> - -namespace llvm { - class MCInst; - struct EDInstInfo; - struct EDToken; - struct EDDisassembler; - struct EDOperand; - -#ifdef __BLOCKS__ - typedef int (^EDTokenVisitor_t)(EDToken *token); -#endif - -/// CachedResult - Encapsulates the result of a function along with the validity -/// of that result, so that slow functions don't need to run twice -struct CachedResult { - /// True if the result has been obtained by executing the function - bool Valid; - /// The result last obtained from the function - int Result; - - /// Constructor - Initializes an invalid result - CachedResult() : Valid(false) { } - /// valid - Returns true if the result has been obtained by executing the - /// function and false otherwise - bool valid() { return Valid; } - /// result - Returns the result of the function or an undefined value if - /// valid() is false - int result() { return Result; } - /// setResult - Sets the result of the function and declares it valid - /// returning the result (so that setResult() can be called from inside a - /// return statement) - /// @arg result - The result of the function - int setResult(int result) { Result = result; Valid = true; return result; } -}; - -/// EDInst - Encapsulates a single instruction, which can be queried for its -/// string representation, as well as its operands and tokens -struct EDInst { - /// The parent disassembler - EDDisassembler &Disassembler; - /// The containing MCInst - llvm::MCInst *Inst; - /// The instruction information provided by TableGen for this instruction - const llvm::EDInstInfo *ThisInstInfo; - /// The number of bytes for the machine code representation of the instruction - uint64_t ByteSize; - - /// The result of the stringify() function - CachedResult StringifyResult; - /// The string representation of the instruction - std::string String; - /// The order in which operands from the InstInfo's operand information appear - /// in String - const signed char* OperandOrder; - - /// The result of the parseOperands() function - CachedResult ParseResult; - typedef llvm::SmallVector<EDOperand*, 5> opvec_t; - /// The instruction's operands - opvec_t Operands; - /// The operand corresponding to the target, if the instruction is a branch - int BranchTarget; - /// The operand corresponding to the source, if the instruction is a move - int MoveSource; - /// The operand corresponding to the target, if the instruction is a move - int MoveTarget; - - /// The result of the tokenize() function - CachedResult TokenizeResult; - typedef std::vector<EDToken*> tokvec_t; - /// The instruction's tokens - tokvec_t Tokens; - - /// Constructor - initializes an instruction given the output of the LLVM - /// C++ disassembler - /// - /// @arg inst - The MCInst, which will now be owned by this object - /// @arg byteSize - The size of the consumed instruction, in bytes - /// @arg disassembler - The parent disassembler - /// @arg instInfo - The instruction information produced by the table - /// generator for this instruction - EDInst(llvm::MCInst *inst, - uint64_t byteSize, - EDDisassembler &disassembler, - const llvm::EDInstInfo *instInfo); - ~EDInst(); - - /// byteSize - returns the number of bytes consumed by the machine code - /// representation of the instruction - uint64_t byteSize(); - /// instID - returns the LLVM instruction ID of the instruction - unsigned instID(); - - /// stringify - populates the String and AsmString members of the instruction, - /// returning 0 on success or -1 otherwise - int stringify(); - /// getString - retrieves a pointer to the string representation of the - /// instructinon, returning 0 on success or -1 otherwise - /// - /// @arg str - A reference to a pointer that, on success, is set to point to - /// the string representation of the instruction; this string is still owned - /// by the instruction and will be deleted when it is - int getString(const char *&str); - - /// isBranch - Returns true if the instruction is a branch - bool isBranch(); - /// isMove - Returns true if the instruction is a move - bool isMove(); - - /// parseOperands - populates the Operands member of the instruction, - /// returning 0 on success or -1 otherwise - int parseOperands(); - /// branchTargetID - returns the ID (suitable for use with getOperand()) of - /// the target operand if the instruction is a branch, or -1 otherwise - int branchTargetID(); - /// moveSourceID - returns the ID of the source operand if the instruction - /// is a move, or -1 otherwise - int moveSourceID(); - /// moveTargetID - returns the ID of the target operand if the instruction - /// is a move, or -1 otherwise - int moveTargetID(); - - /// numOperands - returns the number of operands available to retrieve, or -1 - /// on error - int numOperands(); - /// getOperand - retrieves an operand from the instruction's operand list by - /// index, returning 0 on success or -1 on error - /// - /// @arg operand - A reference whose target is pointed at the operand on - /// success, although the operand is still owned by the EDInst - /// @arg index - The index of the operand in the instruction - int getOperand(EDOperand *&operand, unsigned int index); - - /// tokenize - populates the Tokens member of the instruction, returning 0 on - /// success or -1 otherwise - int tokenize(); - /// numTokens - returns the number of tokens in the instruction, or -1 on - /// error - int numTokens(); - /// getToken - retrieves a token from the instruction's token list by index, - /// returning 0 on success or -1 on error - /// - /// @arg token - A reference whose target is pointed at the token on success, - /// although the token is still owned by the EDInst - /// @arg index - The index of the token in the instrcutino - int getToken(EDToken *&token, unsigned int index); - -#ifdef __BLOCKS__ - /// visitTokens - Visits each token in turn and applies a block to it, - /// returning 0 if all blocks are visited and/or the block signals - /// termination by returning 1; returns -1 on error - /// - /// @arg visitor - The visitor block to apply to all tokens. - int visitTokens(EDTokenVisitor_t visitor); -#endif -}; - -} // end namespace llvm - -#endif diff --git a/lib/MC/MCDisassembler/EDMain.cpp b/lib/MC/MCDisassembler/EDMain.cpp deleted file mode 100644 index 5c065dbf0c..0000000000 --- a/lib/MC/MCDisassembler/EDMain.cpp +++ /dev/null @@ -1,276 +0,0 @@ -//===-- EDMain.cpp - LLVM Enhanced Disassembly C API ----------------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file implements the enhanced disassembler's public C API. -// -//===----------------------------------------------------------------------===// - -#include "EDDisassembler.h" -#include "EDInst.h" -#include "EDOperand.h" -#include "EDToken.h" -#include "llvm-c/EnhancedDisassembly.h" -using namespace llvm; - -int EDGetDisassembler(EDDisassemblerRef *disassembler, - const char *triple, - EDAssemblySyntax_t syntax) { - EDDisassembler::AssemblySyntax Syntax; - switch (syntax) { - default: llvm_unreachable("Unknown assembly syntax!"); - case kEDAssemblySyntaxX86Intel: - Syntax = EDDisassembler::kEDAssemblySyntaxX86Intel; - break; - case kEDAssemblySyntaxX86ATT: - Syntax = EDDisassembler::kEDAssemblySyntaxX86ATT; - break; - case kEDAssemblySyntaxARMUAL: - Syntax = EDDisassembler::kEDAssemblySyntaxARMUAL; - break; - } - - EDDisassemblerRef ret = EDDisassembler::getDisassembler(triple, Syntax); - - if (!ret) - return -1; - *disassembler = ret; - return 0; -} - -int EDGetRegisterName(const char** regName, - EDDisassemblerRef disassembler, - unsigned regID) { - const char *name = ((EDDisassembler*)disassembler)->nameWithRegisterID(regID); - if (!name) - return -1; - *regName = name; - return 0; -} - -int EDRegisterIsStackPointer(EDDisassemblerRef disassembler, - unsigned regID) { - return ((EDDisassembler*)disassembler)->registerIsStackPointer(regID) ? 1 : 0; -} - -int EDRegisterIsProgramCounter(EDDisassemblerRef disassembler, - unsigned regID) { - return ((EDDisassembler*)disassembler)->registerIsProgramCounter(regID) ? 1:0; -} - -unsigned int EDCreateInsts(EDInstRef *insts, - unsigned int count, - EDDisassemblerRef disassembler, - ::EDByteReaderCallback byteReader, - uint64_t address, - void *arg) { - unsigned int index; - - for (index = 0; index < count; ++index) { - EDInst *inst = ((EDDisassembler*)disassembler)->createInst(byteReader, - address, arg); - - if (!inst) - return index; - - insts[index] = inst; - address += inst->byteSize(); - } - - return count; -} - -void EDReleaseInst(EDInstRef inst) { - delete ((EDInst*)inst); -} - -int EDInstByteSize(EDInstRef inst) { - return ((EDInst*)inst)->byteSize(); -} - -int EDGetInstString(const char **buf, - EDInstRef inst) { - return ((EDInst*)inst)->getString(*buf); -} - -int EDInstID(unsigned *instID, EDInstRef inst) { - *instID = ((EDInst*)inst)->instID(); - return 0; -} - -int EDInstIsBranch(EDInstRef inst) { - return ((EDInst*)inst)->isBranch(); -} - -int EDInstIsMove(EDInstRef inst) { - return ((EDInst*)inst)->isMove(); -} - -int EDBranchTargetID(EDInstRef inst) { - return ((EDInst*)inst)->branchTargetID(); -} - -int EDMoveSourceID(EDInstRef inst) { - return ((EDInst*)inst)->moveSourceID(); -} - -int EDMoveTargetID(EDInstRef inst) { - return ((EDInst*)inst)->moveTargetID(); -} - -int EDNumTokens(EDInstRef inst) { - return ((EDInst*)inst)->numTokens(); -} - -int EDGetToken(EDTokenRef *token, - EDInstRef inst, - int index) { - return ((EDInst*)inst)->getToken(*(EDToken**)token, index); -} - -int EDGetTokenString(const char **buf, - EDTokenRef token) { - return ((EDToken*)token)->getString(*buf); -} - -int EDOperandIndexForToken(EDTokenRef token) { - return ((EDToken*)token)->operandID(); -} - -int EDTokenIsWhitespace(EDTokenRef token) { - return ((EDToken*)token)->type() == EDToken::kTokenWhitespace; -} - -int EDTokenIsPunctuation(EDTokenRef token) { - return ((EDToken*)token)->type() == EDToken::kTokenPunctuation; -} - -int EDTokenIsOpcode(EDTokenRef token) { - return ((EDToken*)token)->type() == EDToken::kTokenOpcode; -} - -int EDTokenIsLiteral(EDTokenRef token) { - return ((EDToken*)token)->type() == EDToken::kTokenLiteral; -} - -int EDTokenIsRegister(EDTokenRef token) { - return ((EDToken*)token)->type() == EDToken::kTokenRegister; -} - -int EDTokenIsNegativeLiteral(EDTokenRef token) { - if (((EDToken*)token)->type() != EDToken::kTokenLiteral) - return -1; - - return ((EDToken*)token)->literalSign(); -} - -int EDLiteralTokenAbsoluteValue(uint64_t *value, EDTokenRef token) { - if (((EDToken*)token)->type() != EDToken::kTokenLiteral) - return -1; - - return ((EDToken*)token)->literalAbsoluteValue(*value); -} - -int EDRegisterTokenValue(unsigned *registerID, - EDTokenRef token) { - if (((EDToken*)token)->type() != EDToken::kTokenRegister) - return -1; - - return ((EDToken*)token)->registerID(*registerID); -} - -int EDNumOperands(EDInstRef inst) { - return ((EDInst*)inst)->numOperands(); -} - -int EDGetOperand(EDOperandRef *operand, - EDInstRef inst, - int index) { - return ((EDInst*)inst)->getOperand(*(EDOperand**)operand, index); -} - -int EDOperandIsRegister(EDOperandRef operand) { - return ((EDOperand*)operand)->isRegister(); -} - -int EDOperandIsImmediate(EDOperandRef operand) { - return ((EDOperand*)operand)->isImmediate(); -} - -int EDOperandIsMemory(EDOperandRef operand) { - return ((EDOperand*)operand)->isMemory(); -} - -int EDRegisterOperandValue(unsigned *value, EDOperandRef operand) { - if (!((EDOperand*)operand)->isRegister()) - return -1; - *value = ((EDOperand*)operand)->regVal(); - return 0; -} - -int EDImmediateOperandValue(uint64_t *value, EDOperandRef operand) { - if (!((EDOperand*)operand)->isImmediate()) - return -1; - *value = ((EDOperand*)operand)->immediateVal(); - return 0; -} - -int EDEvaluateOperand(uint64_t *result, EDOperandRef operand, - ::EDRegisterReaderCallback regReader, void *arg) { - return ((EDOperand*)operand)->evaluate(*result, regReader, arg); -} - -#ifdef __BLOCKS__ - -struct ByteReaderWrapper { - EDByteBlock_t byteBlock; -}; - -static int readerWrapperCallback(uint8_t *byte, - uint64_t address, - void *arg) { - struct ByteReaderWrapper *wrapper = (struct ByteReaderWrapper *)arg; - return wrapper->byteBlock(byte, address); -} - -unsigned int EDBlockCreateInsts(EDInstRef *insts, - int count, - EDDisassemblerRef disassembler, - EDByteBlock_t byteBlock, - uint64_t address) { - struct ByteReaderWrapper wrapper; - wrapper.byteBlock = byteBlock; - - return EDCreateInsts(insts, count, disassembler, readerWrapperCallback, - address, (void*)&wrapper); -} - -int EDBlockEvaluateOperand(uint64_t *result, EDOperandRef operand, - EDRegisterBlock_t regBlock) { - return ((EDOperand*)operand)->evaluate(*result, regBlock); -} - -int EDBlockVisitTokens(EDInstRef inst, ::EDTokenVisitor_t visitor) { - return ((EDInst*)inst)->visitTokens((llvm::EDTokenVisitor_t)visitor); -} - -#else - -extern "C" unsigned int EDBlockCreateInsts() { - return 0; -} - -extern "C" int EDBlockEvaluateOperand() { - return -1; -} - -extern "C" int EDBlockVisitTokens() { - return -1; -} - -#endif diff --git a/lib/MC/MCDisassembler/EDOperand.cpp b/lib/MC/MCDisassembler/EDOperand.cpp deleted file mode 100644 index 48b374659d..0000000000 --- a/lib/MC/MCDisassembler/EDOperand.cpp +++ /dev/null @@ -1,315 +0,0 @@ -//===-- EDOperand.cpp - LLVM Enhanced Disassembler ------------------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file implements the Enhanced Disassembly library's operand class. The -// operand is responsible for allowing evaluation given a particular register -// context. -// -//===----------------------------------------------------------------------===// - -#include "EDOperand.h" -#include "EDDisassembler.h" -#include "EDInst.h" -#include "llvm/MC/EDInstInfo.h" -#include "llvm/MC/MCInst.h" -using namespace llvm; - -EDOperand::EDOperand(const EDDisassembler &disassembler, - const EDInst &inst, - unsigned int opIndex, - unsigned int &mcOpIndex) : - Disassembler(disassembler), - Inst(inst), - OpIndex(opIndex), - MCOpIndex(mcOpIndex) { - unsigned int numMCOperands = 0; - - Triple::ArchType arch = Disassembler.TgtTriple.getArch(); - - if (arch == Triple::x86 || - arch == Triple::x86_64) { - uint8_t operandType = inst.ThisInstInfo->operandTypes[opIndex]; - - switch (operandType) { - default: - break; - case kOperandTypeImmediate: - numMCOperands = 1; - break; - case kOperandTypeRegister: - numMCOperands = 1; - break; - case kOperandTypeX86Memory: - numMCOperands = 5; - break; - case kOperandTypeX86EffectiveAddress: - numMCOperands = 4; - break; - case kOperandTypeX86PCRelative: - numMCOperands = 1; - break; - } - } - else if (arch == Triple::arm || - arch == Triple::thumb) { - uint8_t operandType = inst.ThisInstInfo->operandTypes[opIndex]; - - switch (operandType) { - default: - case kOperandTypeARMRegisterList: - case kOperandTypeARMDPRRegisterList: - case kOperandTypeARMSPRRegisterList: - break; - case kOperandTypeImmediate: - case kOperandTypeRegister: - case kOperandTypeARMBranchTarget: - case kOperandTypeARMSoImm: - case kOperandTypeARMRotImm: - case kOperandTypeThumb2SoImm: - case kOperandTypeARMSoImm2Part: - case kOperandTypeARMPredicate: - case kOperandTypeThumbITMask: - case kOperandTypeThumb2AddrModeImm8Offset: - case kOperandTypeARMTBAddrMode: - case kOperandTypeThumb2AddrModeImm8s4Offset: - case kOperandTypeARMAddrMode7: - case kOperandTypeThumb2AddrModeReg: - numMCOperands = 1; - break; - case kOperandTypeThumb2SoReg: - case kOperandTypeAddrModeImm12: - case kOperandTypeARMAddrMode2Offset: - case kOperandTypeARMAddrMode3Offset: - case kOperandTypeARMAddrMode4: - case kOperandTypeARMAddrMode5: - case kOperandTypeARMAddrModePC: - case kOperandTypeThumb2AddrModeImm8: - case kOperandTypeThumb2AddrModeImm12: - case kOperandTypeThumb2AddrModeImm8s4: - case kOperandTypeThumbAddrModeImmS1: - case kOperandTypeThumbAddrModeImmS2: - case kOperandTypeThumbAddrModeImmS4: - case kOperandTypeThumbAddrModeRR: - case kOperandTypeThumbAddrModeSP: - case kOperandTypeThumbAddrModePC: - numMCOperands = 2; - break; - case kOperandTypeARMSoReg: - case kOperandTypeLdStSOReg: - case kOperandTypeARMAddrMode2: - case kOperandTypeARMAddrMode3: - case kOperandTypeThumb2AddrModeSoReg: - case kOperandTypeThumbAddrModeRegS1: - case kOperandTypeThumbAddrModeRegS2: - case kOperandTypeThumbAddrModeRegS4: - case kOperandTypeARMAddrMode6Offset: - numMCOperands = 3; - break; - case kOperandTypeARMAddrMode6: - numMCOperands = 4; - break; - } - } - - mcOpIndex += numMCOperands; -} - -EDOperand::~EDOperand() { -} - -int EDOperand::evaluate(uint64_t &result, - EDRegisterReaderCallback callback, - void *arg) { - uint8_t operandType = Inst.ThisInstInfo->operandTypes[OpIndex]; - - Triple::ArchType arch = Disassembler.TgtTriple.getArch(); - - switch (arch) { - default: - return -1; - case Triple::x86: - case Triple::x86_64: - switch (operandType) { - default: - return -1; - case kOperandTypeImmediate: - result = Inst.Inst->getOperand(MCOpIndex).getImm(); - return 0; - case kOperandTypeRegister: - { - unsigned reg = Inst.Inst->getOperand(MCOpIndex).getReg(); - return callback(&result, reg, arg); - } - case kOperandTypeX86PCRelative: - { - int64_t displacement = Inst.Inst->getOperand(MCOpIndex).getImm(); - - uint64_t ripVal; - - // TODO fix how we do this - - if (callback(&ripVal, Disassembler.registerIDWithName("RIP"), arg)) - return -1; - - result = ripVal + displacement; - return 0; - } - case kOperandTypeX86Memory: - case kOperandTypeX86EffectiveAddress: - { - unsigned baseReg = Inst.Inst->getOperand(MCOpIndex).getReg(); - uint64_t scaleAmount = Inst.Inst->getOperand(MCOpIndex+1).getImm(); - unsigned indexReg = Inst.Inst->getOperand(MCOpIndex+2).getReg(); - int64_t displacement = Inst.Inst->getOperand(MCOpIndex+3).getImm(); - - uint64_t addr = 0; - - unsigned segmentReg = Inst.Inst->getOperand(MCOpIndex+4).getReg(); - - if (segmentReg != 0 && arch == Triple::x86_64) { - unsigned fsID = Disassembler.registerIDWithName("FS"); - unsigned gsID = Disassembler.registerIDWithName("GS"); - - if (segmentReg == fsID || - segmentReg == gsID) { - uint64_t segmentBase; - if (!callback(&segmentBase, segmentReg, arg)) - addr += segmentBase; - } - } - - if (baseReg) { - uint64_t baseVal; - if (callback(&baseVal, baseReg, arg)) - return -1; - addr += baseVal; - } - - if (indexReg) { - uint64_t indexVal; - if (callback(&indexVal, indexReg, arg)) - return -1; - addr += (scaleAmount * indexVal); - } - - addr += displacement; - - result = addr; - return 0; - } - } // switch (operandType) - case Triple::arm: - case Triple::thumb: - switch (operandType) { - default: - return -1; - case kOperandTypeImmediate: - if (!Inst.Inst->getOperand(MCOpIndex).isImm()) - return -1; - - result = Inst.Inst->getOperand(MCOpIndex).getImm(); - return 0; - case kOperandTypeRegister: - { - if (!Inst.Inst->getOperand(MCOpIndex).isReg()) - return -1; - - unsigned reg = Inst.Inst->getOperand(MCOpIndex).getReg(); - return callback(&result, reg, arg); - } - case kOperandTypeARMBranchTarget: - { - if (!Inst.Inst->getOperand(MCOpIndex).isImm()) - return -1; - - int64_t displacement = Inst.Inst->getOperand(MCOpIndex).getImm(); - - uint64_t pcVal; - - if (callback(&pcVal, Disassembler.registerIDWithName("PC"), arg)) - return -1; - - result = pcVal + displacement; - return 0; - } - } - } -} - -int EDOperand::isRegister() { - return(Inst.ThisInstInfo->operandFlags[OpIndex] == kOperandTypeRegister); -} - -unsigned EDOperand::regVal() { - return Inst.Inst->getOperand(MCOpIndex).getReg(); -} - -int EDOperand::isImmediate() { - return(Inst.ThisInstInfo->operandFlags[OpIndex] == kOperandTypeImmediate); -} - -uint64_t EDOperand::immediateVal() { - return Inst.Inst->getOperand(MCOpIndex).getImm(); -} - -int EDOperand::isMemory() { - uint8_t operandType = Inst.ThisInstInfo->operandTypes[OpIndex]; - - switch (operandType) { - default: - return 0; - case kOperandTypeX86Memory: - case kOperandTypeX86PCRelative: - case kOperandTypeX86EffectiveAddress: - case kOperandTypeARMSoReg: - case kOperandTypeARMSoImm: - case kOperandTypeARMAddrMode2: - case kOperandTypeARMAddrMode2Offset: - case kOperandTypeARMAddrMode3: - case kOperandTypeARMAddrMode3Offset: - case kOperandTypeARMAddrMode4: - case kOperandTypeARMAddrMode5: - case kOperandTypeARMAddrMode6: - case kOperandTypeARMAddrMode7: - case kOperandTypeARMAddrModePC: - case kOperandTypeARMBranchTarget: - case kOperandTypeThumbAddrModeRegS1: - case kOperandTypeThumbAddrModeRegS2: - case kOperandTypeThumbAddrModeRegS4: - case kOperandTypeThumbAddrModeRR: - case kOperandTypeThumbAddrModeSP: - case kOperandTypeThumb2SoImm: - case kOperandTypeThumb2AddrModeImm8: - case kOperandTypeThumb2AddrModeImm8Offset: - case kOperandTypeThumb2AddrModeImm12: - case kOperandTypeThumb2AddrModeSoReg: - case kOperandTypeThumb2AddrModeImm8s4: - case kOperandTypeThumb2AddrModeReg: - return 1; - } -} - -#ifdef __BLOCKS__ -namespace { - struct RegisterReaderWrapper { - EDOperand::EDRegisterBlock_t regBlock; - }; -} - -static int readerWrapperCallback(uint64_t *value, unsigned regID, void *arg) { - RegisterReaderWrapper *wrapper = (RegisterReaderWrapper *)arg; - return wrapper->regBlock(value, regID); -} - -int EDOperand::evaluate(uint64_t &result, EDRegisterBlock_t regBlock) { - RegisterReaderWrapper wrapper; - wrapper.regBlock = regBlock; - return evaluate(result, readerWrapperCallback, (void*)&wrapper); -} -#endif diff --git a/lib/MC/MCDisassembler/EDOperand.h b/lib/MC/MCDisassembler/EDOperand.h deleted file mode 100644 index 50260ec965..0000000000 --- a/lib/MC/MCDisassembler/EDOperand.h +++ /dev/null @@ -1,91 +0,0 @@ -//===-EDOperand.h - LLVM Enhanced Disassembler ------------------*- C++ -*-===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file defines the interface for the Enhanced Disassembly library's -// operand class. The operand is responsible for allowing evaluation given a -// particular register context. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_EDOPERAND_H -#define LLVM_EDOPERAND_H - -#include "llvm/Support/DataTypes.h" - -namespace llvm { - -struct EDDisassembler; -struct EDInst; - -typedef int (*EDRegisterReaderCallback)(uint64_t *value, unsigned regID, - void* arg); - - -/// EDOperand - Encapsulates a single operand, which can be evaluated by the -/// client -struct EDOperand { - /// The parent disassembler - const EDDisassembler &Disassembler; - /// The parent instruction - const EDInst &Inst; - - /// The index of the operand in the EDInst - unsigned int OpIndex; - /// The index of the first component of the operand in the MCInst - unsigned int MCOpIndex; - - /// Constructor - Initializes an EDOperand - /// - /// @arg disassembler - The disassembler responsible for the operand - /// @arg inst - The instruction containing this operand - /// @arg opIndex - The index of the operand in inst - /// @arg mcOpIndex - The index of the operand in the original MCInst - EDOperand(const EDDisassembler &disassembler, - const EDInst &inst, - unsigned int opIndex, - unsigned int &mcOpIndex); - ~EDOperand(); - - /// evaluate - Returns the numeric value of an operand to the extent possible, - /// returning 0 on success or -1 if there was some problem (such as a - /// register not being readable) - /// - /// @arg result - A reference whose target is filled in with the value of - /// the operand (the address if it is a memory operand) - /// @arg callback - A function to call to obtain register values - /// @arg arg - An opaque argument to pass to callback - int evaluate(uint64_t &result, - EDRegisterReaderCallback callback, - void *arg); - - /// isRegister - Returns 1 if the operand is a register or 0 otherwise - int isRegister(); - /// regVal - Returns the register value. - unsigned regVal(); - - /// isImmediate - Returns 1 if the operand is an immediate or 0 otherwise - int isImmediate(); - /// immediateVal - Returns the immediate value. - uint64_t immediateVal(); - - /// isMemory - Returns 1 if the operand is a memory location or 0 otherwise - int isMemory(); - -#ifdef __BLOCKS__ - typedef int (^EDRegisterBlock_t)(uint64_t *value, unsigned regID); - - /// evaluate - Like evaluate for a callback, but uses a block instead - int evaluate(uint64_t &result, - EDRegisterBlock_t regBlock); -#endif -}; - -} // end namespace llvm - -#endif diff --git a/lib/MC/MCDisassembler/EDToken.cpp b/lib/MC/MCDisassembler/EDToken.cpp deleted file mode 100644 index a7fb1eb3c1..0000000000 --- a/lib/MC/MCDisassembler/EDToken.cpp +++ /dev/null @@ -1,214 +0,0 @@ -//===-- EDToken.cpp - LLVM Enhanced Disassembler --------------------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file implements the Enhanced Disassembler library's token class. The -// token is responsible for vending information about the token, such as its -// type and logical value. -// -//===----------------------------------------------------------------------===// - -#include "EDToken.h" -#include "EDDisassembler.h" -#include "llvm/ADT/SmallVector.h" -#include "llvm/MC/MCParser/MCAsmLexer.h" -#include "llvm/MC/MCParser/MCParsedAsmOperand.h" -using namespace llvm; - -EDToken::EDToken(StringRef str, - enum tokenType type, - uint64_t localType, - EDDisassembler &disassembler) : - Disassembler(disassembler), - Str(str), - Type(type), - LocalType(localType), - OperandID(-1) { -} - -EDToken::~EDToken() { -} - -void EDToken::makeLiteral(bool sign, uint64_t absoluteValue) { - Type = kTokenLiteral; - LiteralSign = sign; - LiteralAbsoluteValue = absoluteValue; -} - -void EDToken::makeRegister(unsigned registerID) { - Type = kTokenRegister; - RegisterID = registerID; -} - -void EDToken::setOperandID(int operandID) { - OperandID = operandID; -} - -enum EDToken::tokenType EDToken::type() const { - return Type; -} - -uint64_t EDToken::localType() const { - return LocalType; -} - -StringRef EDToken::string() const { - return Str; -} - -int EDToken::operandID() const { - return OperandID; -} - -int EDToken::literalSign() const { - if (Type != kTokenLiteral) - return -1; - return (LiteralSign ? 1 : 0); -} - -int EDToken::literalAbsoluteValue(uint64_t &value) const { - if (Type != kTokenLiteral) - return -1; - value = LiteralAbsoluteValue; - return 0; -} - -int EDToken::registerID(unsigned ®isterID) const { - if (Type != kTokenRegister) - return -1; - registerID = RegisterID; - return 0; -} - -int EDToken::tokenize(std::vector<EDToken*> &tokens, - std::string &str, - const signed char *operandOrder, - EDDisassembler &disassembler) { - SmallVector<MCParsedAsmOperand*, 5> parsedOperands; - SmallVector<AsmToken, 10> asmTokens; - - if (disassembler.parseInst(parsedOperands, asmTokens, str)) - { - for (unsigned i = 0, e = parsedOperands.size(); i != e; ++i) - delete parsedOperands[i]; - return -1; - } - - SmallVectorImpl<MCParsedAsmOperand*>::iterator operandIterator; - unsigned int operandIndex; - SmallVectorImpl<AsmToken>::iterator tokenIterator; - - operandIterator = parsedOperands.begin(); - operandIndex = 0; - - bool readOpcode = false; - - const char *wsPointer = asmTokens.begin()->getLoc().getPointer(); - - for (tokenIterator = asmTokens.begin(); - tokenIterator != asmTokens.end(); - ++tokenIterator) { - SMLoc tokenLoc = tokenIterator->getLoc(); - - const char *tokenPointer = tokenLoc.getPointer(); - - if (tokenPointer > wsPointer) { - unsigned long wsLength = tokenPointer - wsPointer; - - EDToken *whitespaceToken = new EDToken(StringRef(wsPointer, wsLength), - EDToken::kTokenWhitespace, - 0, - disassembler); - - tokens.push_back(whitespaceToken); - } - - wsPointer = tokenPointer + tokenIterator->getString().size(); - - while (operandIterator != parsedOperands.end() && - tokenLoc.getPointer() > - (*operandIterator)->getEndLoc().getPointer()) { - ++operandIterator; - ++operandIndex; - } - - EDToken *token; - - switch (tokenIterator->getKind()) { - case AsmToken::Identifier: - if (!readOpcode) { - token = new EDToken(tokenIterator->getString(), - EDToken::kTokenOpcode, - (uint64_t)tokenIterator->getKind(), - disassembler); - readOpcode = true; - break; - } - // any identifier that isn't an opcode is mere punctuation; so we fall - // through - default: - token = new EDToken(tokenIterator->getString(), - EDToken::kTokenPunctuation, - (uint64_t)tokenIterator->getKind(), - disassembler); - break; - case AsmToken::Integer: - { - token = new EDToken(tokenIterator->getString(), - EDToken::kTokenLiteral, - (uint64_t)tokenIterator->getKind(), - disassembler); - - int64_t intVal = tokenIterator->getIntVal(); - - if (intVal < 0) - token->makeLiteral(true, -intVal); - else - token->makeLiteral(false, intVal); - break; - } - case AsmToken::Register: - { - token = new EDToken(tokenIterator->getString(), - EDToken::kTokenLiteral, - (uint64_t)tokenIterator->getKind(), - disassembler); - - token->makeRegister((unsigned)tokenIterator->getRegVal()); - break; - } - } - - if (operandIterator != parsedOperands.end() && - tokenLoc.getPointer() >= - (*operandIterator)->getStartLoc().getPointer()) { - /// operandIndex == 0 means the operand is the instruction (which the - /// AsmParser treats as an operand but edis does not). We therefore skip - /// operandIndex == 0 and subtract 1 from all other operand indices. - - if (operandIndex > 0) - token->setOperandID(operandOrder[operandIndex - 1]); - } - - tokens.push_back(token); - } - - // Free any parsed operands. - for (unsigned i = 0, e = parsedOperands.size(); i != e; ++i) - delete parsedOperands[i]; - - return 0; -} - -int EDToken::getString(const char*& buf) { - if (PermStr.length() == 0) { - PermStr = Str.str(); - } - buf = PermStr.c_str(); - return 0; -} diff --git a/lib/MC/MCDisassembler/EDToken.h b/lib/MC/MCDisassembler/EDToken.h deleted file mode 100644 index 384079b72e..0000000000 --- a/lib/MC/MCDisassembler/EDToken.h +++ /dev/null @@ -1,139 +0,0 @@ -//===-EDToken.h - LLVM Enhanced Disassembler --------------------*- C++ -*-===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// -// -// This file defines the interface for the Enhanced Disassembly library's token -// class. The token is responsible for vending information about the token, -// such as its type and logical value. -// -//===----------------------------------------------------------------------===// - -#ifndef LLVM_EDTOKEN_H -#define LLVM_EDTOKEN_H - -#include "llvm/ADT/StringRef.h" -#include "llvm/Support/DataTypes.h" -#include <string> -#include <vector> - -namespace llvm { - -struct EDDisassembler; - -/// EDToken - Encapsulates a single token, which can provide a string -/// representation of itself or interpret itself in various ways, depending -/// on the token type. -struct EDToken { - enum tokenType { - kTokenWhitespace, - kTokenOpcode, - kTokenLiteral, - kTokenRegister, - kTokenPunctuation - }; - - /// The parent disassembler - EDDisassembler &Disassembler; - - /// The token's string representation - llvm::StringRef Str; - /// The token's string representation, but in a form suitable for export - std::string PermStr; - /// The type of the token, as exposed through the external API - enum tokenType Type; - /// The type of the token, as recorded by the syntax-specific tokenizer - uint64_t LocalType; - /// The operand corresponding to the token, or (unsigned int)-1 if not - /// part of an operand. - int OperandID; - - /// The sign if the token is a literal (1 if negative, 0 otherwise) - bool LiteralSign; - /// The absolute value if the token is a literal - uint64_t LiteralAbsoluteValue; - /// The LLVM register ID if the token is a register name - unsigned RegisterID; - - /// Constructor - Initializes an EDToken with the information common to all - /// tokens - /// - /// @arg str - The string corresponding to the token - /// @arg type - The token's type as exposed through the public API - /// @arg localType - The token's type as recorded by the tokenizer - /// @arg disassembler - The disassembler responsible for the token - EDToken(llvm::StringRef str, - enum tokenType type, - uint64_t localType, - EDDisassembler &disassembler); - - /// makeLiteral - Adds the information specific to a literal - /// @arg sign - The sign of the literal (1 if negative, 0 - /// otherwise) - /// - /// @arg absoluteValue - The absolute value of the literal - void makeLiteral(bool sign, uint64_t absoluteValue); - /// makeRegister - Adds the information specific to a register - /// - /// @arg registerID - The LLVM register ID - void makeRegister(unsigned registerID); - - /// setOperandID - Links the token to a numbered operand - /// - /// @arg operandID - The operand ID to link to - void setOperandID(int operandID); - - ~EDToken(); - - /// type - Returns the public type of the token - enum tokenType type() const; - /// localType - Returns the tokenizer-specific type of the token - uint64_t localType() const; - /// string - Returns the string representation of the token - llvm::StringRef string() const; - /// operandID - Returns the operand ID of the token - int operandID() const; - - /// literalSign - Returns the sign of the token - /// (1 if negative, 0 if positive or unsigned, -1 if it is not a literal) - int literalSign() const; - /// literalAbsoluteValue - Retrieves the absolute value of the token, and - /// returns -1 if the token is not a literal - /// @arg value - A reference to a value that is filled in with the absolute - /// value, if it is valid - int literalAbsoluteValue(uint64_t &value) const; - /// registerID - Retrieves the register ID of the token, and returns -1 if the - /// token is not a register - /// - /// @arg registerID - A reference to a value that is filled in with the - /// register ID, if it is valid - int registerID(unsigned ®isterID) const; - - /// tokenize - Tokenizes a string using the platform- and syntax-specific - /// tokenizer, and returns 0 on success (-1 on failure) - /// - /// @arg tokens - A vector that will be filled in with pointers to - /// allocated tokens - /// @arg str - The string, as outputted by the AsmPrinter - /// @arg operandOrder - The order of the operands from the operandFlags array - /// as they appear in str - /// @arg disassembler - The disassembler for the desired target and - // assembly syntax - static int tokenize(std::vector<EDToken*> &tokens, - std::string &str, - const signed char *operandOrder, - EDDisassembler &disassembler); - - /// getString - Directs a character pointer to the string, returning 0 on - /// success (-1 on failure) - /// @arg buf - A reference to a pointer that is set to point to the string. - /// The string is still owned by the token. - int getString(const char*& buf); -}; - -} // end namespace llvm -#endif diff --git a/lib/MC/MCDwarf.cpp b/lib/MC/MCDwarf.cpp index 597ee1d691..c81abe952a 100644 --- a/lib/MC/MCDwarf.cpp +++ b/lib/MC/MCDwarf.cpp @@ -101,7 +101,8 @@ void MCLineEntry::Make(MCStreamer *MCOS, const MCSection *Section) { } // Add the line entry to this section's entries. - LineSection->addLineEntry(LineEntry); + LineSection->addLineEntry(LineEntry, + MCOS->getContext().getDwarfCompileUnitID()); } // @@ -131,7 +132,12 @@ static inline const MCExpr *MakeStartMinusEndExpr(const MCStreamer &MCOS, // static inline void EmitDwarfLineTable(MCStreamer *MCOS, const MCSection *Section, - const MCLineSection *LineSection) { + const MCLineSection *LineSection, + unsigned CUID) { + // This LineSection does not contain any LineEntry for the given Compile Unit. + if (!LineSection->containEntriesForID(CUID)) + return; + unsigned FileNum = 1; unsigned LastLine = 1; unsigned Column = 0; @@ -141,8 +147,8 @@ static inline void EmitDwarfLineTable(MCStreamer *MCOS, // Loop through each MCLineEntry and encode the dwarf line number table. for (MCLineSection::const_iterator - it = LineSection->getMCLineEntries()->begin(), - ie = LineSection->getMCLineEntries()->end(); it != ie; ++it) { + it = LineSection->getMCLineEntries(CUID).begin(), + ie = LineSection->getMCLineEntries(CUID).end(); it != ie; ++it) { if (FileNum != it->getFileNum()) { FileNum = it->getFileNum(); @@ -215,9 +221,36 @@ const MCSymbol *MCDwarfFileTable::Emit(MCStreamer *MCOS) { // Switch to the section where the table will be emitted into. MCOS->SwitchSection(context.getObjectFileInfo()->getDwarfLineSection()); - // Create a symbol at the beginning of this section. - MCSymbol *LineStartSym = context.CreateTempSymbol(); - // Set the value of the symbol, as we are at the start of the section. + const DenseMap<unsigned, MCSymbol *> &MCLineTableSymbols = + MCOS->getContext().getMCLineTableSymbols(); + // CUID and MCLineTableSymbols are set in DwarfDebug, when DwarfDebug does + // not exist, CUID will be 0 and MCLineTableSymbols will be empty. + // Handle Compile Unit 0, the line table start symbol is the section symbol. + const MCSymbol *LineStartSym = EmitCU(MCOS, 0); + // Handle the rest of the Compile Units. + for (unsigned Is = 1, Ie = MCLineTableSymbols.size(); Is < Ie; Is++) + EmitCU(MCOS, Is); + + // Now delete the MCLineSections that were created in MCLineEntry::Make() + // and used to emit the line table. + const DenseMap<const MCSection *, MCLineSection *> &MCLineSections = + MCOS->getContext().getMCLineSections(); + for (DenseMap<const MCSection *, MCLineSection *>::const_iterator it = + MCLineSections.begin(), ie = MCLineSections.end(); it != ie; + ++it) + delete it->second; + + return LineStartSym; +} + +const MCSymbol *MCDwarfFileTable::EmitCU(MCStreamer *MCOS, unsigned CUID) { + MCContext &context = MCOS->getContext(); + + // Create a symbol at the beginning of the line table. + MCSymbol *LineStartSym = MCOS->getContext().getMCLineTableSymbol(CUID); + if (!LineStartSym) + LineStartSym = context.CreateTempSymbol(); + // Set the value of the symbol, as we are at the start of the line table. MCOS->EmitLabel(LineStartSym); // Create a symbol for the end of the section (to be set when we get there). @@ -239,8 +272,7 @@ const MCSymbol *MCDwarfFileTable::Emit(MCStreamer *MCOS) { // total length, the 2 bytes for the version, and these 4 bytes for the // length of the prologue. MCOS->EmitAbsValue(MakeStartMinusEndExpr(*MCOS, *LineStartSym, *ProEndSym, - (4 + 2 + 4)), - 4, 0); + (4 + 2 + 4)), 4, 0); // Parameters of the state machine, are next. MCOS->EmitIntValue(DWARF2_LINE_MIN_INSN_LENGTH, 1); @@ -267,19 +299,19 @@ const MCSymbol *MCDwarfFileTable::Emit(MCStreamer *MCOS) { // First the directory table. const std::vector<StringRef> &MCDwarfDirs = - context.getMCDwarfDirs(); + context.getMCDwarfDirs(CUID); for (unsigned i = 0; i < MCDwarfDirs.size(); i++) { - MCOS->EmitBytes(MCDwarfDirs[i], 0); // the DirectoryName - MCOS->EmitBytes(StringRef("\0", 1), 0); // the null term. of the string + MCOS->EmitBytes(MCDwarfDirs[i]); // the DirectoryName + MCOS->EmitBytes(StringRef("\0", 1)); // the null term. of the string } MCOS->EmitIntValue(0, 1); // Terminate the directory list // Second the file table. const std::vector<MCDwarfFile *> &MCDwarfFiles = - MCOS->getContext().getMCDwarfFiles(); + MCOS->getContext().getMCDwarfFiles(CUID); for (unsigned i = 1; i < MCDwarfFiles.size(); i++) { - MCOS->EmitBytes(MCDwarfFiles[i]->getName(), 0); // FileName - MCOS->EmitBytes(StringRef("\0", 1), 0); // the null term. of the string + MCOS->EmitBytes(MCDwarfFiles[i]->getName()); // FileName + MCOS->EmitBytes(StringRef("\0", 1)); // the null term. of the string // the Directory num MCOS->EmitULEB128IntValue(MCDwarfFiles[i]->getDirIndex()); MCOS->EmitIntValue(0, 1); // last modification timestamp (always 0) @@ -301,11 +333,7 @@ const MCSymbol *MCDwarfFileTable::Emit(MCStreamer *MCOS) { ++it) { const MCSection *Sec = *it; const MCLineSection *Line = MCLineSections.lookup(Sec); - EmitDwarfLineTable(MCOS, Sec, Line); - - // Now delete the MCLineSections that were created in MCLineEntry::Make() - // and used to emit the line table. - delete Line; + EmitDwarfLineTable(MCOS, Sec, Line, CUID); } if (MCOS->getContext().getAsmInfo().getLinkerRequiresNonEmptyDwarfLines() @@ -342,7 +370,7 @@ void MCDwarfLineAddr::Emit(MCStreamer *MCOS, int64_t LineDelta, SmallString<256> Tmp; raw_svector_ostream OS(Tmp); MCDwarfLineAddr::Encode(LineDelta, AddrDelta, OS); - MCOS->EmitBytes(OS.str(), /*AddrSpace=*/0); + MCOS->EmitBytes(OS.str()); } /// Utility function to encode a Dwarf pair of LineDelta and AddrDeltas. @@ -618,30 +646,35 @@ static void EmitGenDwarfInfo(MCStreamer *MCOS, const std::vector<StringRef> &MCDwarfDirs = context.getMCDwarfDirs(); if (MCDwarfDirs.size() > 0) { - MCOS->EmitBytes(MCDwarfDirs[0], 0); - MCOS->EmitBytes("/", 0); + MCOS->EmitBytes(MCDwarfDirs[0]); + MCOS->EmitBytes("/"); } const std::vector<MCDwarfFile *> &MCDwarfFiles = MCOS->getContext().getMCDwarfFiles(); - MCOS->EmitBytes(MCDwarfFiles[1]->getName(), 0); + MCOS->EmitBytes(MCDwarfFiles[1]->getName()); MCOS->EmitIntValue(0, 1); // NULL byte to terminate the string. // AT_comp_dir, the working directory the assembly was done in. - llvm::sys::Path CWD = llvm::sys::Path::GetCurrentDirectory(); - MCOS->EmitBytes(StringRef(CWD.c_str()), 0); + MCOS->EmitBytes(context.getCompilationDir()); MCOS->EmitIntValue(0, 1); // NULL byte to terminate the string. // AT_APPLE_flags, the command line arguments of the assembler tool. StringRef DwarfDebugFlags = context.getDwarfDebugFlags(); if (!DwarfDebugFlags.empty()){ - MCOS->EmitBytes(DwarfDebugFlags, 0); + MCOS->EmitBytes(DwarfDebugFlags); MCOS->EmitIntValue(0, 1); // NULL byte to terminate the string. } // AT_producer, the version of the assembler tool. - MCOS->EmitBytes(StringRef("llvm-mc (based on LLVM "), 0); - MCOS->EmitBytes(StringRef(PACKAGE_VERSION), 0); - MCOS->EmitBytes(StringRef(")"), 0); + StringRef DwarfDebugProducer = context.getDwarfDebugProducer(); + if (!DwarfDebugProducer.empty()){ + MCOS->EmitBytes(DwarfDebugProducer); + } + else { + MCOS->EmitBytes(StringRef("llvm-mc (based on LLVM ")); + MCOS->EmitBytes(StringRef(PACKAGE_VERSION)); + MCOS->EmitBytes(StringRef(")")); + } MCOS->EmitIntValue(0, 1); // NULL byte to terminate the string. // AT_language, a 4 byte value. We use DW_LANG_Mips_Assembler as the dwarf2 @@ -662,7 +695,7 @@ static void EmitGenDwarfInfo(MCStreamer *MCOS, MCOS->EmitULEB128IntValue(2); // AT_name, of the label without any leading underbar. - MCOS->EmitBytes(Entry->getName(), 0); + MCOS->EmitBytes(Entry->getName()); MCOS->EmitIntValue(0, 1); // NULL byte to terminate the string. // AT_decl_file, index into the file table. @@ -787,7 +820,7 @@ void MCGenDwarfLabelEntry::Make(MCSymbol *Symbol, MCStreamer *MCOS, static int getDataAlignmentFactor(MCStreamer &streamer) { MCContext &context = streamer.getContext(); const MCAsmInfo &asmInfo = context.getAsmInfo(); - int size = asmInfo.getPointerSize(); + int size = asmInfo.getCalleeSaveStackSlotSize(); if (asmInfo.isStackGrowthDirectionUp()) return size; else @@ -1072,7 +1105,7 @@ void FrameEmitterImpl::EmitCFIInstruction(MCStreamer &Streamer, } case MCCFIInstruction::OpEscape: if (VerboseAsm) Streamer.AddComment("Escape bytes"); - Streamer.EmitBytes(Instr.getValues(), 0); + Streamer.EmitBytes(Instr.getValues()); return; } llvm_unreachable("Unhandled case in switch"); @@ -1230,7 +1263,7 @@ const MCSymbol &FrameEmitterImpl::EmitCIE(MCStreamer &streamer, Augmentation += "R"; if (IsSignalFrame) Augmentation += "S"; - streamer.EmitBytes(Augmentation.str(), 0); + streamer.EmitBytes(Augmentation.str()); } streamer.EmitIntValue(0, 1); @@ -1494,7 +1527,7 @@ void MCDwarfFrameEmitter::EmitAdvanceLoc(MCStreamer &Streamer, SmallString<256> Tmp; raw_svector_ostream OS(Tmp); MCDwarfFrameEmitter::EncodeAdvanceLoc(AddrDelta, OS); - Streamer.EmitBytes(OS.str(), /*AddrSpace=*/0); + Streamer.EmitBytes(OS.str()); } void MCDwarfFrameEmitter::EncodeAdvanceLoc(uint64_t AddrDelta, diff --git a/lib/MC/MCELF.cpp b/lib/MC/MCELF.cpp index 4db2846bc3..560cdbc6ab 100644 --- a/lib/MC/MCELF.cpp +++ b/lib/MC/MCELF.cpp @@ -52,6 +52,8 @@ unsigned MCELF::GetType(const MCSymbolData &SD) { return Type; } +// Visibility is stored in the first two bits of st_other +// st_other values are stored in the second byte of get/setFlags void MCELF::SetVisibility(MCSymbolData &SD, unsigned Visibility) { assert(Visibility == ELF::STV_DEFAULT || Visibility == ELF::STV_INTERNAL || Visibility == ELF::STV_HIDDEN || Visibility == ELF::STV_PROTECTED); @@ -68,4 +70,17 @@ unsigned MCELF::GetVisibility(MCSymbolData &SD) { return Visibility; } +// Other is stored in the last six bits of st_other +// st_other values are stored in the second byte of get/setFlags +void MCELF::setOther(MCSymbolData &SD, unsigned Other) { + uint32_t OtherFlags = SD.getFlags() & ~(0x3f << ELF_Other_Shift); + SD.setFlags(OtherFlags | (Other << ELF_Other_Shift)); +} + +unsigned MCELF::getOther(MCSymbolData &SD) { + unsigned Other = + (SD.getFlags() & (0x3f << ELF_Other_Shift)) >> ELF_Other_Shift; + return Other; +} + } diff --git a/lib/MC/MCELFObjectTargetWriter.cpp b/lib/MC/MCELFObjectTargetWriter.cpp index 74cd042a0f..4cac84d666 100644 --- a/lib/MC/MCELFObjectTargetWriter.cpp +++ b/lib/MC/MCELFObjectTargetWriter.cpp @@ -24,11 +24,6 @@ MCELFObjectTargetWriter::MCELFObjectTargetWriter(bool Is64Bit_, IsN64(IsN64_){ } -/// Default e_flags = 0 -unsigned MCELFObjectTargetWriter::getEFlags() const { - return 0; -} - const MCSymbol *MCELFObjectTargetWriter::ExplicitRelSym(const MCAssembler &Asm, const MCValue &Target, const MCFragment &F, diff --git a/lib/MC/MCELFStreamer.cpp b/lib/MC/MCELFStreamer.cpp index 8b9bdb14a0..7f5f1b63e5 100644 --- a/lib/MC/MCELFStreamer.cpp +++ b/lib/MC/MCELFStreamer.cpp @@ -1,4 +1,4 @@ -//===- lib/MC/MCELFStreamer.cpp - ELF Object Output ------------===// +//===- lib/MC/MCELFStreamer.cpp - ELF Object Output -----------------------===// // // The LLVM Compiler Infrastructure // @@ -65,6 +65,10 @@ inline void MCELFStreamer::SetSectionBss() { MCELFStreamer::~MCELFStreamer() { } +void MCELFStreamer::InitToTextSection() { + SetSectionText(); +} + void MCELFStreamer::InitSections() { // This emulates the same behavior of GNU as. This makes it easier // to compare the output as the major sections are in the same order. @@ -86,6 +90,10 @@ void MCELFStreamer::EmitLabel(MCSymbol *Symbol) { MCELF::SetType(SD, ELF::STT_TLS); } +void MCELFStreamer::EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); +} + void MCELFStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { switch (Flag) { case MCAF_SyntaxUnified: return; // no-op here. @@ -101,6 +109,9 @@ void MCELFStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { } void MCELFStreamer::ChangeSection(const MCSection *Section) { + MCSectionData *CurSection = getCurrentSectionData(); + if (CurSection && CurSection->isBundleLocked()) + report_fatal_error("Unterminated .bundle_lock when changing a section"); const MCSymbol *Grp = static_cast<const MCSectionELF *>(Section)->getGroup(); if (Grp) getAssembler().getOrCreateSymbolData(*Grp); @@ -258,10 +269,22 @@ void MCELFStreamer::EmitLocalCommonSymbol(MCSymbol *Symbol, uint64_t Size, void MCELFStreamer::EmitValueImpl(const MCExpr *Value, unsigned Size, unsigned AddrSpace) { + if (getCurrentSectionData()->isBundleLocked()) + report_fatal_error("Emitting values inside a locked bundle is forbidden"); fixSymbolsInTLSFixups(Value); MCObjectStreamer::EmitValueImpl(Value, Size, AddrSpace); } +void MCELFStreamer::EmitValueToAlignment(unsigned ByteAlignment, + int64_t Value, + unsigned ValueSize, + unsigned MaxBytesToEmit) { + if (getCurrentSectionData()->isBundleLocked()) + report_fatal_error("Emitting values inside a locked bundle is forbidden"); + MCObjectStreamer::EmitValueToAlignment(ByteAlignment, Value, + ValueSize, MaxBytesToEmit); +} + // Add a symbol for the file name of this module. This is the second // entry in the module's symbol table (the first being the null symbol). @@ -277,7 +300,9 @@ void MCELFStreamer::EmitFileDirective(StringRef Filename) { void MCELFStreamer::fixSymbolsInTLSFixups(const MCExpr *expr) { switch (expr->getKind()) { - case MCExpr::Target: llvm_unreachable("Can't handle target exprs yet!"); + case MCExpr::Target: + cast<MCTargetExpr>(expr)->fixELFSymbolsInTLSFixups(getAssembler()); + break; case MCExpr::Constant: break; @@ -309,6 +334,19 @@ void MCELFStreamer::fixSymbolsInTLSFixups(const MCExpr *expr) { case MCSymbolRefExpr::VK_Mips_GOTTPREL: case MCSymbolRefExpr::VK_Mips_TPREL_HI: case MCSymbolRefExpr::VK_Mips_TPREL_LO: + case MCSymbolRefExpr::VK_PPC_TPREL16_HA: + case MCSymbolRefExpr::VK_PPC_TPREL16_LO: + case MCSymbolRefExpr::VK_PPC_DTPREL16_HA: + case MCSymbolRefExpr::VK_PPC_DTPREL16_LO: + case MCSymbolRefExpr::VK_PPC_GOT_TPREL16_HA: + case MCSymbolRefExpr::VK_PPC_GOT_TPREL16_LO: + case MCSymbolRefExpr::VK_PPC_TLS: + case MCSymbolRefExpr::VK_PPC_GOT_TLSGD16_HA: + case MCSymbolRefExpr::VK_PPC_GOT_TLSGD16_LO: + case MCSymbolRefExpr::VK_PPC_TLSGD: + case MCSymbolRefExpr::VK_PPC_GOT_TLSLD16_HA: + case MCSymbolRefExpr::VK_PPC_GOT_TLSLD16_LO: + case MCSymbolRefExpr::VK_PPC_TLSLD: break; } MCSymbolData &SD = getAssembler().getOrCreateSymbolData(symRef.getSymbol()); @@ -324,32 +362,116 @@ void MCELFStreamer::fixSymbolsInTLSFixups(const MCExpr *expr) { void MCELFStreamer::EmitInstToFragment(const MCInst &Inst) { this->MCObjectStreamer::EmitInstToFragment(Inst); - MCInstFragment &F = *cast<MCInstFragment>(getCurrentFragment()); + MCRelaxableFragment &F = *cast<MCRelaxableFragment>(getCurrentFragment()); for (unsigned i = 0, e = F.getFixups().size(); i != e; ++i) fixSymbolsInTLSFixups(F.getFixups()[i].getValue()); } void MCELFStreamer::EmitInstToData(const MCInst &Inst) { - MCDataFragment *DF = getOrCreateDataFragment(); - + MCAssembler &Assembler = getAssembler(); SmallVector<MCFixup, 4> Fixups; SmallString<256> Code; raw_svector_ostream VecOS(Code); - getAssembler().getEmitter().EncodeInstruction(Inst, VecOS, Fixups); + Assembler.getEmitter().EncodeInstruction(Inst, VecOS, Fixups); VecOS.flush(); for (unsigned i = 0, e = Fixups.size(); i != e; ++i) fixSymbolsInTLSFixups(Fixups[i].getValue()); + // There are several possibilities here: + // + // If bundling is disabled, append the encoded instruction to the current data + // fragment (or create a new such fragment if the current fragment is not a + // data fragment). + // + // If bundling is enabled: + // - If we're not in a bundle-locked group, emit the instruction into a + // fragment of its own. If there are no fixups registered for the + // instruction, emit a MCCompactEncodedInstFragment. Otherwise, emit a + // MCDataFragment. + // - If we're in a bundle-locked group, append the instruction to the current + // data fragment because we want all the instructions in a group to get into + // the same fragment. Be careful not to do that for the first instruction in + // the group, though. + MCDataFragment *DF; + + if (Assembler.isBundlingEnabled()) { + MCSectionData *SD = getCurrentSectionData(); + if (SD->isBundleLocked() && !SD->isBundleGroupBeforeFirstInst()) + // If we are bundle-locked, we re-use the current fragment. + // The bundle-locking directive ensures this is a new data fragment. + DF = cast<MCDataFragment>(getCurrentFragment()); + else if (!SD->isBundleLocked() && Fixups.size() == 0) { + // Optimize memory usage by emitting the instruction to a + // MCCompactEncodedInstFragment when not in a bundle-locked group and + // there are no fixups registered. + MCCompactEncodedInstFragment *CEIF = new MCCompactEncodedInstFragment(SD); + CEIF->getContents().append(Code.begin(), Code.end()); + return; + } else { + DF = new MCDataFragment(SD); + if (SD->getBundleLockState() == MCSectionData::BundleLockedAlignToEnd) { + // If this is a new fragment created for a bundle-locked group, and the + // group was marked as "align_to_end", set a flag in the fragment. + DF->setAlignToBundleEnd(true); + } + } + + // We're now emitting an instruction in a bundle group, so this flag has + // to be turned off. + SD->setBundleGroupBeforeFirstInst(false); + } else { + DF = getOrCreateDataFragment(); + } + // Add the fixups and data. for (unsigned i = 0, e = Fixups.size(); i != e; ++i) { Fixups[i].setOffset(Fixups[i].getOffset() + DF->getContents().size()); DF->getFixups().push_back(Fixups[i]); } + DF->setHasInstructions(true); DF->getContents().append(Code.begin(), Code.end()); } +void MCELFStreamer::EmitBundleAlignMode(unsigned AlignPow2) { + assert(AlignPow2 <= 30 && "Invalid bundle alignment"); + MCAssembler &Assembler = getAssembler(); + if (Assembler.getBundleAlignSize() == 0 && AlignPow2 > 0) + Assembler.setBundleAlignSize(1 << AlignPow2); + else + report_fatal_error(".bundle_align_mode should be only set once per file"); +} + +void MCELFStreamer::EmitBundleLock(bool AlignToEnd) { + MCSectionData *SD = getCurrentSectionData(); + + // Sanity checks + // + if (!getAssembler().isBundlingEnabled()) + report_fatal_error(".bundle_lock forbidden when bundling is disabled"); + else if (SD->isBundleLocked()) + report_fatal_error("Nesting of .bundle_lock is forbidden"); + + SD->setBundleLockState(AlignToEnd ? MCSectionData::BundleLockedAlignToEnd : + MCSectionData::BundleLocked); + SD->setBundleGroupBeforeFirstInst(true); +} + +void MCELFStreamer::EmitBundleUnlock() { + MCSectionData *SD = getCurrentSectionData(); + + // Sanity checks + if (!getAssembler().isBundlingEnabled()) + report_fatal_error(".bundle_unlock forbidden when bundling is disabled"); + else if (!SD->isBundleLocked()) + report_fatal_error(".bundle_unlock without matching lock"); + else if (SD->isBundleGroupBeforeFirstInst()) + report_fatal_error("Empty bundle-locked group is forbidden"); + + SD->setBundleLockState(MCSectionData::NotBundleLocked); +} + void MCELFStreamer::FinishImpl() { EmitFrames(true); @@ -377,7 +499,7 @@ void MCELFStreamer::FinishImpl() { } void MCELFStreamer::EmitTCEntry(const MCSymbol &S) { // Creates a R_PPC64_TOC relocation - MCObjectStreamer::EmitSymbolValue(&S, 8, 0); + MCObjectStreamer::EmitSymbolValue(&S, 8); } MCStreamer *llvm::createELFStreamer(MCContext &Context, MCAsmBackend &MAB, @@ -395,6 +517,10 @@ void MCELFStreamer::EmitThumbFunc(MCSymbol *Func) { llvm_unreachable("Generic ELF doesn't support this directive"); } +MCSymbolData &MCELFStreamer::getOrCreateSymbolData(MCSymbol *Symbol) { + return getAssembler().getOrCreateSymbolData(*Symbol); +} + void MCELFStreamer::EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) { llvm_unreachable("ELF doesn't support this directive"); } diff --git a/lib/MC/MCExpr.cpp b/lib/MC/MCExpr.cpp index 7a63bace6c..1a53934fef 100644 --- a/lib/MC/MCExpr.cpp +++ b/lib/MC/MCExpr.cpp @@ -54,14 +54,16 @@ void MCExpr::print(raw_ostream &OS) const { else OS << Sym; - if (SRE.getKind() == MCSymbolRefExpr::VK_ARM_PLT || + if (SRE.getKind() == MCSymbolRefExpr::VK_ARM_NONE || + SRE.getKind() == MCSymbolRefExpr::VK_ARM_PLT || SRE.getKind() == MCSymbolRefExpr::VK_ARM_TLSGD || SRE.getKind() == MCSymbolRefExpr::VK_ARM_GOT || SRE.getKind() == MCSymbolRefExpr::VK_ARM_GOTOFF || SRE.getKind() == MCSymbolRefExpr::VK_ARM_TPOFF || SRE.getKind() == MCSymbolRefExpr::VK_ARM_GOTTPOFF || SRE.getKind() == MCSymbolRefExpr::VK_ARM_TARGET1 || - SRE.getKind() == MCSymbolRefExpr::VK_ARM_TARGET2) + SRE.getKind() == MCSymbolRefExpr::VK_ARM_TARGET2 || + SRE.getKind() == MCSymbolRefExpr::VK_ARM_PREL31) OS << MCSymbolRefExpr::getVariantKindName(SRE.getKind()); else if (SRE.getKind() != MCSymbolRefExpr::VK_None && SRE.getKind() != MCSymbolRefExpr::VK_PPC_DARWIN_HA16 && @@ -193,6 +195,7 @@ StringRef MCSymbolRefExpr::getVariantKindName(VariantKind Kind) { case VK_DTPOFF: return "DTPOFF"; case VK_TLVP: return "TLVP"; case VK_SECREL: return "SECREL"; + case VK_ARM_NONE: return "(NONE)"; case VK_ARM_PLT: return "(PLT)"; case VK_ARM_GOT: return "(GOT)"; case VK_ARM_GOTOFF: return "(GOTOFF)"; @@ -201,6 +204,7 @@ StringRef MCSymbolRefExpr::getVariantKindName(VariantKind Kind) { case VK_ARM_TLSGD: return "(tlsgd)"; case VK_ARM_TARGET1: return "(target1)"; case VK_ARM_TARGET2: return "(target2)"; + case VK_ARM_PREL31: return "(prel31)"; case VK_PPC_TOC: return "tocbase"; case VK_PPC_TOC_ENTRY: return "toc"; case VK_PPC_DARWIN_HA16: return "ha16"; @@ -209,10 +213,19 @@ StringRef MCSymbolRefExpr::getVariantKindName(VariantKind Kind) { case VK_PPC_GAS_LO16: return "l"; case VK_PPC_TPREL16_HA: return "tprel@ha"; case VK_PPC_TPREL16_LO: return "tprel@l"; + case VK_PPC_DTPREL16_HA: return "dtprel@ha"; + case VK_PPC_DTPREL16_LO: return "dtprel@l"; case VK_PPC_TOC16_HA: return "toc@ha"; case VK_PPC_TOC16_LO: return "toc@l"; - case VK_PPC_GOT_TPREL16_DS: return "got@tprel"; + case VK_PPC_GOT_TPREL16_HA: return "got@tprel@ha"; + case VK_PPC_GOT_TPREL16_LO: return "got@tprel@l"; case VK_PPC_TLS: return "tls"; + case VK_PPC_GOT_TLSGD16_HA: return "got@tlsgd@ha"; + case VK_PPC_GOT_TLSGD16_LO: return "got@tlsgd@l"; + case VK_PPC_GOT_TLSLD16_HA: return "got@tlsld@ha"; + case VK_PPC_GOT_TLSLD16_LO: return "got@tlsld@l"; + case VK_PPC_TLSGD: return "tlsgd"; + case VK_PPC_TLSLD: return "tlsld"; case VK_Mips_GPREL: return "GPREL"; case VK_Mips_GOT_CALL: return "GOT_CALL"; case VK_Mips_GOT16: return "GOT16"; diff --git a/lib/MC/MCMachOStreamer.cpp b/lib/MC/MCMachOStreamer.cpp index f279e74e38..7d08d0ecd5 100644 --- a/lib/MC/MCMachOStreamer.cpp +++ b/lib/MC/MCMachOStreamer.cpp @@ -34,18 +34,21 @@ private: void EmitDataRegion(DataRegionData::KindTy Kind); void EmitDataRegionEnd(); public: - MCMachOStreamer(MCContext &Context, MCAsmBackend &MAB, - raw_ostream &OS, MCCodeEmitter *Emitter) - : MCObjectStreamer(Context, MAB, OS, Emitter) {} + MCMachOStreamer(MCContext &Context, MCAsmBackend &MAB, raw_ostream &OS, + MCCodeEmitter *Emitter) + : MCObjectStreamer(SK_MachOStreamer, Context, MAB, OS, Emitter) {} /// @name MCStreamer Interface /// @{ virtual void InitSections(); + virtual void InitToTextSection(); virtual void EmitLabel(MCSymbol *Symbol); + virtual void EmitDebugLabel(MCSymbol *Symbol); virtual void EmitEHSymAttributes(const MCSymbol *Symbol, MCSymbol *EHSymbol); virtual void EmitAssemblerFlag(MCAssemblerFlag Flag); + virtual void EmitLinkerOptions(ArrayRef<std::string> Options); virtual void EmitDataRegion(MCDataRegionType Kind); virtual void EmitThumbFunc(MCSymbol *Func); virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute); @@ -84,15 +87,23 @@ public: virtual void FinishImpl(); /// @} + + static bool classof(const MCStreamer *S) { + return S->getKind() == SK_MachOStreamer; + } }; } // end anonymous namespace. void MCMachOStreamer::InitSections() { - SwitchSection(getContext().getMachOSection("__TEXT", "__text", - MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, - 0, SectionKind::getText())); + InitToTextSection(); +} +void MCMachOStreamer::InitToTextSection() { + SwitchSection(getContext().getMachOSection( + "__TEXT", "__text", + MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, 0, + SectionKind::getText())); } void MCMachOStreamer::EmitEHSymAttributes(const MCSymbol *Symbol, @@ -130,6 +141,9 @@ void MCMachOStreamer::EmitLabel(MCSymbol *Symbol) { SD.setFlags(SD.getFlags() & ~SF_ReferenceTypeMask); } +void MCMachOStreamer::EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); +} void MCMachOStreamer::EmitDataRegion(DataRegionData::KindTy Kind) { if (!getAssembler().getBackend().hasDataInCodeSupport()) return; @@ -169,6 +183,10 @@ void MCMachOStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { } } +void MCMachOStreamer::EmitLinkerOptions(ArrayRef<std::string> Options) { + getAssembler().getLinkerOptions().push_back(Options); +} + void MCMachOStreamer::EmitDataRegion(MCDataRegionType Kind) { switch (Kind) { case MCDR_DataRegion: diff --git a/lib/MC/MCNullStreamer.cpp b/lib/MC/MCNullStreamer.cpp index f5ece10dc4..c872b2203f 100644 --- a/lib/MC/MCNullStreamer.cpp +++ b/lib/MC/MCNullStreamer.cpp @@ -19,11 +19,14 @@ namespace { class MCNullStreamer : public MCStreamer { public: - MCNullStreamer(MCContext &Context) : MCStreamer(Context) {} + MCNullStreamer(MCContext &Context) : MCStreamer(SK_NullStreamer, Context) {} /// @name MCStreamer Interface /// @{ + virtual void InitToTextSection() { + } + virtual void InitSections() { } @@ -35,7 +38,9 @@ namespace { assert(getCurrentSection() && "Cannot emit before setting section!"); Symbol->setSection(*getCurrentSection()); } - + virtual void EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); + } virtual void EmitAssemblerFlag(MCAssemblerFlag Flag) {} virtual void EmitThumbFunc(MCSymbol *Func) {} @@ -84,7 +89,7 @@ namespace { virtual void EmitFileDirective(StringRef Filename) {} virtual bool EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, - StringRef Filename) { + StringRef Filename, unsigned CUID = 0) { return false; } virtual void EmitDwarfLocDirective(unsigned FileNo, unsigned Line, @@ -93,6 +98,10 @@ namespace { StringRef FileName) {} virtual void EmitInstruction(const MCInst &Inst) {} + virtual void EmitBundleAlignMode(unsigned AlignPow2) {} + virtual void EmitBundleLock(bool AlignToEnd) {} + virtual void EmitBundleUnlock() {} + virtual void FinishImpl() {} virtual void EmitCFIEndProcImpl(MCDwarfFrameInfo &Frame) { @@ -100,6 +109,11 @@ namespace { } /// @} + + static bool classof(const MCStreamer *S) { + return S->getKind() == SK_NullStreamer; + } + }; } diff --git a/lib/MC/MCObjectFileInfo.cpp b/lib/MC/MCObjectFileInfo.cpp index d57e0c5b4a..2e1a045569 100644 --- a/lib/MC/MCObjectFileInfo.cpp +++ b/lib/MC/MCObjectFileInfo.cpp @@ -186,6 +186,10 @@ void MCObjectFileInfo::InitMachOMCObjectFileInfo(Triple T) { Ctx->getMachOSection("__DWARF", "__debug_frame", MCSectionMachO::S_ATTR_DEBUG, SectionKind::getMetadata()); + DwarfPubNamesSection = + Ctx->getMachOSection("__DWARF", "__debug_pubnames", + MCSectionMachO::S_ATTR_DEBUG, + SectionKind::getMetadata()); DwarfPubTypesSection = Ctx->getMachOSection("__DWARF", "__debug_pubtypes", MCSectionMachO::S_ATTR_DEBUG, @@ -256,6 +260,33 @@ void MCObjectFileInfo::InitELFMCObjectFileInfo(Triple T) { TTypeEncoding = (CMModel == CodeModel::Small) ? dwarf::DW_EH_PE_udata4 : dwarf::DW_EH_PE_absptr; } + } else if (T.getArch() == Triple::aarch64) { + FDECFIEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata4; + + // The small model guarantees static code/data size < 4GB, but not where it + // will be in memory. Most of these could end up >2GB away so even a signed + // pc-relative 32-bit address is insufficient, theoretically. + if (RelocM == Reloc::PIC_) { + PersonalityEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel | + dwarf::DW_EH_PE_sdata8; + LSDAEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata8; + FDEEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata4; + TTypeEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel | + dwarf::DW_EH_PE_sdata8; + } else { + PersonalityEncoding = dwarf::DW_EH_PE_absptr; + LSDAEncoding = dwarf::DW_EH_PE_absptr; + FDEEncoding = dwarf::DW_EH_PE_udata4; + TTypeEncoding = dwarf::DW_EH_PE_absptr; + } + } else if (T.getArch() == Triple::ppc64) { + PersonalityEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel | + dwarf::DW_EH_PE_udata8; + FDECFIEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_sdata4; + LSDAEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_udata8; + FDEEncoding = dwarf::DW_EH_PE_pcrel | dwarf::DW_EH_PE_udata8; + TTypeEncoding = dwarf::DW_EH_PE_indirect | dwarf::DW_EH_PE_pcrel | + dwarf::DW_EH_PE_udata8; } // Solaris requires different flags for .eh_frame to seemingly every other @@ -373,6 +404,9 @@ void MCObjectFileInfo::InitELFMCObjectFileInfo(Triple T) { DwarfFrameSection = Ctx->getELFSection(".debug_frame", ELF::SHT_PROGBITS, 0, SectionKind::getMetadata()); + DwarfPubNamesSection = + Ctx->getELFSection(".debug_pubnames", ELF::SHT_PROGBITS, 0, + SectionKind::getMetadata()); DwarfPubTypesSection = Ctx->getELFSection(".debug_pubtypes", ELF::SHT_PROGBITS, 0, SectionKind::getMetadata()); @@ -426,6 +460,12 @@ void MCObjectFileInfo::InitELFMCObjectFileInfo(Triple T) { DwarfLocDWOSection = Ctx->getELFSection(".debug_loc.dwo", ELF::SHT_PROGBITS, 0, SectionKind::getMetadata()); + DwarfStrOffDWOSection = + Ctx->getELFSection(".debug_str_offsets.dwo", ELF::SHT_PROGBITS, 0, + SectionKind::getMetadata()); + DwarfAddrSection = + Ctx->getELFSection(".debug_addr", ELF::SHT_PROGBITS, 0, + SectionKind::getMetadata()); } @@ -510,6 +550,11 @@ void MCObjectFileInfo::InitCOFFMCObjectFileInfo(Triple T) { COFF::IMAGE_SCN_MEM_DISCARDABLE | COFF::IMAGE_SCN_MEM_READ, SectionKind::getMetadata()); + DwarfPubNamesSection = + Ctx->getCOFFSection(".debug_pubnames", + COFF::IMAGE_SCN_MEM_DISCARDABLE | + COFF::IMAGE_SCN_MEM_READ, + SectionKind::getMetadata()); DwarfPubTypesSection = Ctx->getCOFFSection(".debug_pubtypes", COFF::IMAGE_SCN_MEM_DISCARDABLE | diff --git a/lib/MC/MCObjectStreamer.cpp b/lib/MC/MCObjectStreamer.cpp index c2171ffaa5..0d2ce83a8a 100644 --- a/lib/MC/MCObjectStreamer.cpp +++ b/lib/MC/MCObjectStreamer.cpp @@ -20,22 +20,19 @@ #include "llvm/Support/ErrorHandling.h" using namespace llvm; -MCObjectStreamer::MCObjectStreamer(MCContext &Context, MCAsmBackend &TAB, - raw_ostream &OS, MCCodeEmitter *Emitter_) - : MCStreamer(Context), - Assembler(new MCAssembler(Context, TAB, - *Emitter_, *TAB.createObjectWriter(OS), - OS)), - CurSectionData(0) -{ -} - -MCObjectStreamer::MCObjectStreamer(MCContext &Context, MCAsmBackend &TAB, - raw_ostream &OS, MCCodeEmitter *Emitter_, +MCObjectStreamer::MCObjectStreamer(StreamerKind Kind, MCContext &Context, + MCAsmBackend &TAB, raw_ostream &OS, + MCCodeEmitter *Emitter_) + : MCStreamer(Kind, Context), + Assembler(new MCAssembler(Context, TAB, *Emitter_, + *TAB.createObjectWriter(OS), OS)), + CurSectionData(0) {} + +MCObjectStreamer::MCObjectStreamer(StreamerKind Kind, MCContext &Context, + MCAsmBackend &TAB, raw_ostream &OS, + MCCodeEmitter *Emitter_, MCAssembler *_Assembler) - : MCStreamer(Context), Assembler(_Assembler), CurSectionData(0) -{ -} + : MCStreamer(Kind, Context), Assembler(_Assembler), CurSectionData(0) {} MCObjectStreamer::~MCObjectStreamer() { delete &Assembler->getBackend(); @@ -44,6 +41,13 @@ MCObjectStreamer::~MCObjectStreamer() { delete Assembler; } +void MCObjectStreamer::reset() { + if (Assembler) + Assembler->reset(); + CurSectionData = 0; + MCStreamer::reset(); +} + MCFragment *MCObjectStreamer::getCurrentFragment() const { assert(getCurrentSectionData() && "No current section!"); @@ -55,7 +59,9 @@ MCFragment *MCObjectStreamer::getCurrentFragment() const { MCDataFragment *MCObjectStreamer::getOrCreateDataFragment() const { MCDataFragment *F = dyn_cast_or_null<MCDataFragment>(getCurrentFragment()); - if (!F) + // When bundling is enabled, we don't want to add data to a fragment that + // already has instructions (see MCELFStreamer::EmitInstToData for details) + if (!F || (Assembler->isBundlingEnabled() && F->hasInstructions())) F = new MCDataFragment(getCurrentSectionData()); return F; } @@ -128,6 +134,10 @@ void MCObjectStreamer::EmitLabel(MCSymbol *Symbol) { SD.setOffset(F->getContents().size()); } +void MCObjectStreamer::EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); +} + void MCObjectStreamer::EmitULEB128Value(const MCExpr *Value) { int64_t IntValue; if (Value->EvaluateAsAbsolute(IntValue, getAssembler())) { @@ -170,21 +180,27 @@ void MCObjectStreamer::EmitInstruction(const MCInst &Inst) { if (Inst.getOperand(i).isExpr()) AddValueSymbols(Inst.getOperand(i).getExpr()); - getCurrentSectionData()->setHasInstructions(true); + MCSectionData *SD = getCurrentSectionData(); + SD->setHasInstructions(true); // Now that a machine instruction has been assembled into this section, make // a line entry for any .loc directive that has been seen. MCLineEntry::Make(this, getCurrentSection()); // If this instruction doesn't need relaxation, just emit it as data. - if (!getAssembler().getBackend().mayNeedRelaxation(Inst)) { + MCAssembler &Assembler = getAssembler(); + if (!Assembler.getBackend().mayNeedRelaxation(Inst)) { EmitInstToData(Inst); return; } - // Otherwise, if we are relaxing everything, relax the instruction as much as - // possible and emit it as data. - if (getAssembler().getRelaxAll()) { + // Otherwise, relax and emit it as data if either: + // - The RelaxAll flag was passed + // - Bundling is enabled and this instruction is inside a bundle-locked + // group. We want to emit all such instructions into the same data + // fragment. + if (Assembler.getRelaxAll() || + (Assembler.isBundlingEnabled() && SD->isBundleLocked())) { MCInst Relaxed; getAssembler().getBackend().relaxInstruction(Inst, Relaxed); while (getAssembler().getBackend().mayNeedRelaxation(Relaxed)) @@ -198,7 +214,10 @@ void MCObjectStreamer::EmitInstruction(const MCInst &Inst) { } void MCObjectStreamer::EmitInstToFragment(const MCInst &Inst) { - MCInstFragment *IF = new MCInstFragment(Inst, getCurrentSectionData()); + // Always create a new, separate fragment here, because its size can change + // during relaxation. + MCRelaxableFragment *IF = + new MCRelaxableFragment(Inst, getCurrentSectionData()); SmallString<128> Code; raw_svector_ostream VecOS(Code); @@ -207,6 +226,23 @@ void MCObjectStreamer::EmitInstToFragment(const MCInst &Inst) { IF->getContents().append(Code.begin(), Code.end()); } +#ifndef NDEBUG +static const char *BundlingNotImplementedMsg = + "Aligned bundling is not implemented for this object format"; +#endif + +void MCObjectStreamer::EmitBundleAlignMode(unsigned AlignPow2) { + llvm_unreachable(BundlingNotImplementedMsg); +} + +void MCObjectStreamer::EmitBundleLock(bool AlignToEnd) { + llvm_unreachable(BundlingNotImplementedMsg); +} + +void MCObjectStreamer::EmitBundleUnlock() { + llvm_unreachable(BundlingNotImplementedMsg); +} + void MCObjectStreamer::EmitDwarfAdvanceLineAddr(int64_t LineDelta, const MCSymbol *LastLabel, const MCSymbol *Label, @@ -280,7 +316,7 @@ bool MCObjectStreamer::EmitValueToOffset(const MCExpr *Offset, if (!Delta->EvaluateAsAbsolute(Res, getAssembler())) return true; - EmitFill(Res, Value, 0); + EmitFill(Res, Value); return false; } diff --git a/lib/MC/MCParser/AsmLexer.cpp b/lib/MC/MCParser/AsmLexer.cpp index d0492fd16c..c1c594a746 100644 --- a/lib/MC/MCParser/AsmLexer.cpp +++ b/lib/MC/MCParser/AsmLexer.cpp @@ -156,10 +156,36 @@ AsmToken AsmLexer::LexLineComment() { } static void SkipIgnoredIntegerSuffix(const char *&CurPtr) { - if (CurPtr[0] == 'L' && CurPtr[1] == 'L') - CurPtr += 2; - if (CurPtr[0] == 'U' && CurPtr[1] == 'L' && CurPtr[2] == 'L') - CurPtr += 3; + // Skip ULL, UL, U, L and LL suffices. + if (CurPtr[0] == 'U') + ++CurPtr; + if (CurPtr[0] == 'L') + ++CurPtr; + if (CurPtr[0] == 'L') + ++CurPtr; +} + +// Look ahead to search for first non-hex digit, if it's [hH], then we treat the +// integer as a hexadecimal, possibly with leading zeroes. +static unsigned doLookAhead(const char *&CurPtr, unsigned DefaultRadix) { + const char *FirstHex = 0; + const char *LookAhead = CurPtr; + while (1) { + if (isdigit(*LookAhead)) { + ++LookAhead; + } else if (isxdigit(*LookAhead)) { + if (!FirstHex) + FirstHex = LookAhead; + ++LookAhead; + } else { + break; + } + } + bool isHex = *LookAhead == 'h' || *LookAhead == 'H'; + CurPtr = isHex || !FirstHex ? LookAhead : FirstHex; + if (isHex) + return 16; + return DefaultRadix; } /// LexDigit: First character is [0-9]. @@ -167,16 +193,15 @@ static void SkipIgnoredIntegerSuffix(const char *&CurPtr) { /// Forward/Backward Label: [0-9][fb] /// Binary integer: 0b[01]+ /// Octal integer: 0[0-7]+ -/// Hex integer: 0x[0-9a-fA-F]+ +/// Hex integer: 0x[0-9a-fA-F]+ or [0x]?[0-9][0-9a-fA-F]*[hH] /// Decimal integer: [1-9][0-9]* AsmToken AsmLexer::LexDigit() { // Decimal integer: [1-9][0-9]* if (CurPtr[-1] != '0' || CurPtr[0] == '.') { - while (isdigit(*CurPtr)) - ++CurPtr; - + unsigned Radix = doLookAhead(CurPtr, 10); + bool isHex = Radix == 16; // Check for floating point literals. - if (*CurPtr == '.' || *CurPtr == 'e') { + if (!isHex && (*CurPtr == '.' || *CurPtr == 'e')) { ++CurPtr; return LexFloatLiteral(); } @@ -184,17 +209,22 @@ AsmToken AsmLexer::LexDigit() { StringRef Result(TokStart, CurPtr - TokStart); long long Value; - if (Result.getAsInteger(10, Value)) { + if (Result.getAsInteger(Radix, Value)) { // Allow positive values that are too large to fit into a signed 64-bit // integer, but that do fit in an unsigned one, we just convert them over. unsigned long long UValue; - if (Result.getAsInteger(10, UValue)) - return ReturnError(TokStart, "invalid decimal number"); + if (Result.getAsInteger(Radix, UValue)) + return ReturnError(TokStart, !isHex ? "invalid decimal number" : + "invalid hexdecimal number"); Value = (long long)UValue; } - // The darwin/x86 (and x86-64) assembler accepts and ignores ULL and LL - // suffixes on integer literals. + // Consume the [bB][hH]. + if (Radix == 2 || Radix == 16) + ++CurPtr; + + // The darwin/x86 (and x86-64) assembler accepts and ignores type + // suffices on integer literals. SkipIgnoredIntegerSuffix(CurPtr); return AsmToken(AsmToken::Integer, Result, Value); @@ -243,6 +273,10 @@ AsmToken AsmLexer::LexDigit() { if (StringRef(TokStart, CurPtr - TokStart).getAsInteger(0, Result)) return ReturnError(TokStart, "invalid hexadecimal number"); + // Consume the optional [hH]. + if (*CurPtr == 'h' || *CurPtr == 'H') + ++CurPtr; + // The darwin/x86 (and x86-64) assembler accepts and ignores ULL and LL // suffixes on integer literals. SkipIgnoredIntegerSuffix(CurPtr); @@ -251,14 +285,18 @@ AsmToken AsmLexer::LexDigit() { (int64_t)Result); } - // Must be an octal number, it starts with 0. - while (*CurPtr >= '0' && *CurPtr <= '9') - ++CurPtr; - - StringRef Result(TokStart, CurPtr - TokStart); + // Either octal or hexadecimal. long long Value; - if (Result.getAsInteger(8, Value)) - return ReturnError(TokStart, "invalid octal number"); + unsigned Radix = doLookAhead(CurPtr, 8); + bool isHex = Radix == 16; + StringRef Result(TokStart, CurPtr - TokStart); + if (Result.getAsInteger(Radix, Value)) + return ReturnError(TokStart, !isHex ? "invalid octal number" : + "invalid hexdecimal number"); + + // Consume the [hH]. + if (Radix == 16) + ++CurPtr; // The darwin/x86 (and x86-64) assembler accepts and ignores ULL and LL // suffixes on integer literals. diff --git a/lib/MC/MCParser/AsmParser.cpp b/lib/MC/MCParser/AsmParser.cpp index 2c06604be2..6ab49ec92c 100644 --- a/lib/MC/MCParser/AsmParser.cpp +++ b/lib/MC/MCParser/AsmParser.cpp @@ -13,6 +13,7 @@ #include "llvm/ADT/APFloat.h" #include "llvm/ADT/SmallString.h" +#include "llvm/ADT/STLExtras.h" #include "llvm/ADT/StringMap.h" #include "llvm/ADT/Twine.h" #include "llvm/MC/MCAsmInfo.h" @@ -46,31 +47,34 @@ static cl::opt<bool> FatalAssemblerWarnings("fatal-assembler-warnings", cl::desc("Consider warnings as error")); -MCAsmParserSemaCallback::~MCAsmParserSemaCallback() {} +MCAsmParserSemaCallback::~MCAsmParserSemaCallback() {} namespace { -/// \brief Helper class for tracking macro definitions. -typedef std::vector<AsmToken> MacroArgument; -typedef std::vector<MacroArgument> MacroArguments; -typedef std::pair<StringRef, MacroArgument> MacroParameter; -typedef std::vector<MacroParameter> MacroParameters; +/// \brief Helper types for tracking macro definitions. +typedef std::vector<AsmToken> MCAsmMacroArgument; +typedef std::vector<MCAsmMacroArgument> MCAsmMacroArguments; +typedef std::pair<StringRef, MCAsmMacroArgument> MCAsmMacroParameter; +typedef std::vector<MCAsmMacroParameter> MCAsmMacroParameters; -struct Macro { +struct MCAsmMacro { StringRef Name; StringRef Body; - MacroParameters Parameters; + MCAsmMacroParameters Parameters; public: - Macro(StringRef N, StringRef B, const MacroParameters &P) : + MCAsmMacro(StringRef N, StringRef B, const MCAsmMacroParameters &P) : Name(N), Body(B), Parameters(P) {} + + MCAsmMacro(const MCAsmMacro& Other) + : Name(Other.Name), Body(Other.Body), Parameters(Other.Parameters) {} }; /// \brief Helper class for storing information about an active macro /// instantiation. struct MacroInstantiation { /// The macro being instantiated. - const Macro *TheMacro; + const MCAsmMacro *TheMacro; /// The macro instantiation with substitutions. MemoryBuffer *Instantiation; @@ -85,11 +89,10 @@ struct MacroInstantiation { SMLoc ExitLoc; public: - MacroInstantiation(const Macro *M, SMLoc IL, int EB, SMLoc EL, + MacroInstantiation(const MCAsmMacro *M, SMLoc IL, int EB, SMLoc EL, MemoryBuffer *I); }; -//struct AsmRewrite; struct ParseStatementInfo { /// ParsedOperands - The parsed operands from the last parsed statement. SmallVector<MCParsedAsmOperand*, 8> ParsedOperands; @@ -97,11 +100,14 @@ struct ParseStatementInfo { /// Opcode - The opcode from the last parsed instruction. unsigned Opcode; + /// Error - Was there an error parsing the inline assembly? + bool ParseError; + SmallVectorImpl<AsmRewrite> *AsmRewrites; - ParseStatementInfo() : Opcode(~0U), AsmRewrites(0) {} + ParseStatementInfo() : Opcode(~0U), ParseError(false), AsmRewrites(0) {} ParseStatementInfo(SmallVectorImpl<AsmRewrite> *rewrites) - : Opcode(~0), AsmRewrites(rewrites) {} + : Opcode(~0), ParseError(false), AsmRewrites(rewrites) {} ~ParseStatementInfo() { // Free any parsed operands. @@ -113,8 +119,6 @@ struct ParseStatementInfo { /// \brief The concrete assembly parser instance. class AsmParser : public MCAsmParser { - friend class GenericAsmParser; - AsmParser(const AsmParser &) LLVM_DELETED_FUNCTION; void operator=(const AsmParser &) LLVM_DELETED_FUNCTION; private: @@ -125,7 +129,6 @@ private: SourceMgr &SrcMgr; SourceMgr::DiagHandlerTy SavedDiagHandler; void *SavedDiagContext; - MCAsmParserExtension *GenericParser; MCAsmParserExtension *PlatformParser; /// This is the current buffer index we're lexing from as managed by the @@ -135,20 +138,19 @@ private: AsmCond TheCondState; std::vector<AsmCond> TheCondStack; - /// DirectiveMap - This is a table handlers for directives. Each handler is - /// invoked after the directive identifier is read and is responsible for - /// parsing and validating the rest of the directive. The handler is passed - /// in the directive name and the location of the directive keyword. - StringMap<std::pair<MCAsmParserExtension*, DirectiveHandler> > DirectiveMap; + /// ExtensionDirectiveMap - maps directive names to handler methods in parser + /// extensions. Extensions register themselves in this map by calling + /// addDirectiveHandler. + StringMap<ExtensionDirectiveHandler> ExtensionDirectiveMap; /// MacroMap - Map of currently defined macros. - StringMap<Macro*> MacroMap; + StringMap<MCAsmMacro*> MacroMap; /// ActiveMacros - Stack of active macro instantiations. std::vector<MacroInstantiation*> ActiveMacros; /// Boolean tracking whether macro substitution is enabled. - unsigned MacrosEnabled : 1; + unsigned MacrosEnabledFlag : 1; /// Flag tracking whether any errors have been encountered. unsigned HadError : 1; @@ -175,10 +177,9 @@ public: virtual bool Run(bool NoInitialTextSection, bool NoFinalize = false); - virtual void AddDirectiveHandler(MCAsmParserExtension *Object, - StringRef Directive, - DirectiveHandler Handler) { - DirectiveMap[Directive] = std::make_pair(Object, Handler); + virtual void addDirectiveHandler(StringRef Directive, + ExtensionDirectiveHandler Handler) { + ExtensionDirectiveMap[Directive] = Handler; } public: @@ -189,9 +190,9 @@ public: virtual MCAsmLexer &getLexer() { return Lexer; } virtual MCContext &getContext() { return Ctx; } virtual MCStreamer &getStreamer() { return Out; } - virtual unsigned getAssemblerDialect() { + virtual unsigned getAssemblerDialect() { if (AssemblerDialect == ~0U) - return MAI.getAssemblerDialect(); + return MAI.getAssemblerDialect(); else return AssemblerDialect; } @@ -209,7 +210,7 @@ public: void setParsingInlineAsm(bool V) { ParsingInlineAsm = V; } bool isParsingInlineAsm() { return ParsingInlineAsm; } - bool ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, + bool parseMSInlineAsm(void *AsmLoc, std::string &AsmString, unsigned &NumOutputs, unsigned &NumInputs, SmallVectorImpl<std::pair<void *,bool> > &OpDecls, SmallVectorImpl<std::string> &Constraints, @@ -218,27 +219,70 @@ public: const MCInstPrinter *IP, MCAsmParserSemaCallback &SI); - bool ParseExpression(const MCExpr *&Res); - virtual bool ParseExpression(const MCExpr *&Res, SMLoc &EndLoc); - virtual bool ParseParenExpression(const MCExpr *&Res, SMLoc &EndLoc); - virtual bool ParseAbsoluteExpression(int64_t &Res); + bool parseExpression(const MCExpr *&Res); + virtual bool parseExpression(const MCExpr *&Res, SMLoc &EndLoc); + virtual bool parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc); + virtual bool parseAbsoluteExpression(int64_t &Res); + + /// parseIdentifier - Parse an identifier or string (as a quoted identifier) + /// and set \p Res to the identifier contents. + virtual bool parseIdentifier(StringRef &Res); + virtual void eatToEndOfStatement(); + virtual void checkForValidSection(); /// } private: - void CheckForValidSection(); bool ParseStatement(ParseStatementInfo &Info); void EatToEndOfLine(); bool ParseCppHashLineFilenameComment(const SMLoc &L); - bool HandleMacroEntry(StringRef Name, SMLoc NameLoc, const Macro *M); + void CheckForBadMacro(SMLoc DirectiveLoc, StringRef Name, StringRef Body, + MCAsmMacroParameters Parameters); bool expandMacro(raw_svector_ostream &OS, StringRef Body, - const MacroParameters &Parameters, - const MacroArguments &A, + const MCAsmMacroParameters &Parameters, + const MCAsmMacroArguments &A, const SMLoc &L); + + /// \brief Are macros enabled in the parser? + bool MacrosEnabled() {return MacrosEnabledFlag;} + + /// \brief Control a flag in the parser that enables or disables macros. + void SetMacrosEnabled(bool Flag) {MacrosEnabledFlag = Flag;} + + /// \brief Lookup a previously defined macro. + /// \param Name Macro name. + /// \returns Pointer to macro. NULL if no such macro was defined. + const MCAsmMacro* LookupMacro(StringRef Name); + + /// \brief Define a new macro with the given name and information. + void DefineMacro(StringRef Name, const MCAsmMacro& Macro); + + /// \brief Undefine a macro. If no such macro was defined, it's a no-op. + void UndefineMacro(StringRef Name); + + /// \brief Are we inside a macro instantiation? + bool InsideMacroInstantiation() {return !ActiveMacros.empty();} + + /// \brief Handle entry to macro instantiation. + /// + /// \param M The macro. + /// \param NameLoc Instantiation location. + bool HandleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc); + + /// \brief Handle exit from macro instantiation. void HandleMacroExit(); + /// \brief Extract AsmTokens for a macro argument. If the argument delimiter + /// is initially unknown, set it to AsmToken::Eof. It will be set to the + /// correct delimiter by the method. + bool ParseMacroArgument(MCAsmMacroArgument &MA, + AsmToken::TokenKind &ArgumentDelimiter); + + /// \brief Parse all macro arguments for a given macro. + bool ParseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A); + void PrintMacroInstantiations(); void PrintMessage(SMLoc Loc, SourceMgr::DiagKind Kind, const Twine &Msg, ArrayRef<SMRange> Ranges = ArrayRef<SMRange>()) const { @@ -260,16 +304,10 @@ private: /// location. void JumpToLoc(SMLoc Loc, int InBuffer=-1); - virtual void EatToEndOfStatement(); - - bool ParseMacroArgument(MacroArgument &MA, - AsmToken::TokenKind &ArgumentDelimiter); - bool ParseMacroArguments(const Macro *M, MacroArguments &A); - /// \brief Parse up to the end of statement and a return the contents from the /// current token until the end of the statement; the current token on exit /// will be either the EndOfStatement or EOF. - virtual StringRef ParseStringToEndOfStatement(); + virtual StringRef parseStringToEndOfStatement(); /// \brief Parse until the end of a statement or a comma is encountered, /// return the contents from the current token up to the end or comma. @@ -283,24 +321,95 @@ private: bool ParseParenExpr(const MCExpr *&Res, SMLoc &EndLoc); bool ParseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc); - /// ParseIdentifier - Parse an identifier or string (as a quoted identifier) - /// and set \p Res to the identifier contents. - virtual bool ParseIdentifier(StringRef &Res); - - // Directive Parsing. + bool ParseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); - // ".ascii", ".asciiz", ".string" + // Generic (target and platform independent) directive parsing. + enum DirectiveKind { + DK_NO_DIRECTIVE, // Placeholder + DK_SET, DK_EQU, DK_EQUIV, DK_ASCII, DK_ASCIZ, DK_STRING, DK_BYTE, DK_SHORT, + DK_VALUE, DK_2BYTE, DK_LONG, DK_INT, DK_4BYTE, DK_QUAD, DK_8BYTE, DK_SINGLE, + DK_FLOAT, DK_DOUBLE, DK_ALIGN, DK_ALIGN32, DK_BALIGN, DK_BALIGNW, + DK_BALIGNL, DK_P2ALIGN, DK_P2ALIGNW, DK_P2ALIGNL, DK_ORG, DK_FILL, DK_ENDR, + DK_BUNDLE_ALIGN_MODE, DK_BUNDLE_LOCK, DK_BUNDLE_UNLOCK, + DK_ZERO, DK_EXTERN, DK_GLOBL, DK_GLOBAL, DK_INDIRECT_SYMBOL, + DK_LAZY_REFERENCE, DK_NO_DEAD_STRIP, DK_SYMBOL_RESOLVER, DK_PRIVATE_EXTERN, + DK_REFERENCE, DK_WEAK_DEFINITION, DK_WEAK_REFERENCE, + DK_WEAK_DEF_CAN_BE_HIDDEN, DK_COMM, DK_COMMON, DK_LCOMM, DK_ABORT, + DK_INCLUDE, DK_INCBIN, DK_CODE16, DK_CODE16GCC, DK_REPT, DK_IRP, DK_IRPC, + DK_IF, DK_IFB, DK_IFNB, DK_IFC, DK_IFNC, DK_IFDEF, DK_IFNDEF, DK_IFNOTDEF, + DK_ELSEIF, DK_ELSE, DK_ENDIF, + DK_SPACE, DK_SKIP, DK_FILE, DK_LINE, DK_LOC, DK_STABS, + DK_CFI_SECTIONS, DK_CFI_STARTPROC, DK_CFI_ENDPROC, DK_CFI_DEF_CFA, + DK_CFI_DEF_CFA_OFFSET, DK_CFI_ADJUST_CFA_OFFSET, DK_CFI_DEF_CFA_REGISTER, + DK_CFI_OFFSET, DK_CFI_REL_OFFSET, DK_CFI_PERSONALITY, DK_CFI_LSDA, + DK_CFI_REMEMBER_STATE, DK_CFI_RESTORE_STATE, DK_CFI_SAME_VALUE, + DK_CFI_RESTORE, DK_CFI_ESCAPE, DK_CFI_SIGNAL_FRAME, DK_CFI_UNDEFINED, + DK_CFI_REGISTER, + DK_MACROS_ON, DK_MACROS_OFF, DK_MACRO, DK_ENDM, DK_ENDMACRO, DK_PURGEM, + DK_SLEB128, DK_ULEB128 + }; + + /// DirectiveKindMap - Maps directive name --> DirectiveKind enum, for + /// directives parsed by this class. + StringMap<DirectiveKind> DirectiveKindMap; + + // ".ascii", ".asciz", ".string" bool ParseDirectiveAscii(StringRef IDVal, bool ZeroTerminated); bool ParseDirectiveValue(unsigned Size); // ".byte", ".long", ... bool ParseDirectiveRealValue(const fltSemantics &); // ".single", ... bool ParseDirectiveFill(); // ".fill" - bool ParseDirectiveSpace(); // ".space" bool ParseDirectiveZero(); // ".zero" - bool ParseDirectiveSet(StringRef IDVal, bool allow_redef); // ".set", ".equ", ".equiv" + // ".set", ".equ", ".equiv" + bool ParseDirectiveSet(StringRef IDVal, bool allow_redef); bool ParseDirectiveOrg(); // ".org" // ".align{,32}", ".p2align{,w,l}" bool ParseDirectiveAlign(bool IsPow2, unsigned ValueSize); + // ".file", ".line", ".loc", ".stabs" + bool ParseDirectiveFile(SMLoc DirectiveLoc); + bool ParseDirectiveLine(); + bool ParseDirectiveLoc(); + bool ParseDirectiveStabs(); + + // .cfi directives + bool ParseDirectiveCFIRegister(SMLoc DirectiveLoc); + bool ParseDirectiveCFISections(); + bool ParseDirectiveCFIStartProc(); + bool ParseDirectiveCFIEndProc(); + bool ParseDirectiveCFIDefCfaOffset(); + bool ParseDirectiveCFIDefCfa(SMLoc DirectiveLoc); + bool ParseDirectiveCFIAdjustCfaOffset(); + bool ParseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc); + bool ParseDirectiveCFIOffset(SMLoc DirectiveLoc); + bool ParseDirectiveCFIRelOffset(SMLoc DirectiveLoc); + bool ParseDirectiveCFIPersonalityOrLsda(bool IsPersonality); + bool ParseDirectiveCFIRememberState(); + bool ParseDirectiveCFIRestoreState(); + bool ParseDirectiveCFISameValue(SMLoc DirectiveLoc); + bool ParseDirectiveCFIRestore(SMLoc DirectiveLoc); + bool ParseDirectiveCFIEscape(); + bool ParseDirectiveCFISignalFrame(); + bool ParseDirectiveCFIUndefined(SMLoc DirectiveLoc); + + // macro directives + bool ParseDirectivePurgeMacro(SMLoc DirectiveLoc); + bool ParseDirectiveEndMacro(StringRef Directive); + bool ParseDirectiveMacro(SMLoc DirectiveLoc); + bool ParseDirectiveMacrosOnOff(StringRef Directive); + + // ".bundle_align_mode" + bool ParseDirectiveBundleAlignMode(); + // ".bundle_lock" + bool ParseDirectiveBundleLock(); + // ".bundle_unlock" + bool ParseDirectiveBundleUnlock(); + + // ".space", ".skip" + bool ParseDirectiveSpace(StringRef IDVal); + + // .sleb128 (Signed=true) and .uleb128 (Signed=false) + bool ParseDirectiveLEB128(bool Signed); + /// ParseDirectiveSymbolAttribute - Parse a directive like ".globl" which /// accepts a single symbol (which should be a label or an external). bool ParseDirectiveSymbolAttribute(MCSymbolAttr Attr); @@ -321,139 +430,29 @@ private: bool ParseDirectiveElseIf(SMLoc DirectiveLoc); // ".elseif" bool ParseDirectiveElse(SMLoc DirectiveLoc); // ".else" bool ParseDirectiveEndIf(SMLoc DirectiveLoc); // .endif - - /// ParseEscapedString - Parse the current token as a string which may include - /// escaped characters and return the string contents. - bool ParseEscapedString(std::string &Data); + virtual bool parseEscapedString(std::string &Data); const MCExpr *ApplyModifierToExpr(const MCExpr *E, MCSymbolRefExpr::VariantKind Variant); // Macro-like directives - Macro *ParseMacroLikeBody(SMLoc DirectiveLoc); - void InstantiateMacroLikeBody(Macro *M, SMLoc DirectiveLoc, + MCAsmMacro *ParseMacroLikeBody(SMLoc DirectiveLoc); + void InstantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, raw_svector_ostream &OS); bool ParseDirectiveRept(SMLoc DirectiveLoc); // ".rept" bool ParseDirectiveIrp(SMLoc DirectiveLoc); // ".irp" bool ParseDirectiveIrpc(SMLoc DirectiveLoc); // ".irpc" bool ParseDirectiveEndr(SMLoc DirectiveLoc); // ".endr" - // "_emit" - bool ParseDirectiveEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info); -}; + // "_emit" or "__emit" + bool ParseDirectiveMSEmit(SMLoc DirectiveLoc, ParseStatementInfo &Info, + size_t Len); -/// \brief Generic implementations of directive handling, etc. which is shared -/// (or the default, at least) for all assembler parser. -class GenericAsmParser : public MCAsmParserExtension { - template<bool (GenericAsmParser::*Handler)(StringRef, SMLoc)> - void AddDirectiveHandler(StringRef Directive) { - getParser().AddDirectiveHandler(this, Directive, - HandleDirective<GenericAsmParser, Handler>); - } -public: - GenericAsmParser() {} - - AsmParser &getParser() { - return (AsmParser&) this->MCAsmParserExtension::getParser(); - } - - virtual void Initialize(MCAsmParser &Parser) { - // Call the base implementation. - this->MCAsmParserExtension::Initialize(Parser); - - // Debugging directives. - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveFile>(".file"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveLine>(".line"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveLoc>(".loc"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveStabs>(".stabs"); - - // CFI directives. - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFISections>( - ".cfi_sections"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIStartProc>( - ".cfi_startproc"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIEndProc>( - ".cfi_endproc"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIDefCfa>( - ".cfi_def_cfa"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIDefCfaOffset>( - ".cfi_def_cfa_offset"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIAdjustCfaOffset>( - ".cfi_adjust_cfa_offset"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIDefCfaRegister>( - ".cfi_def_cfa_register"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIOffset>( - ".cfi_offset"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveCFIRelOffset>( - ".cfi_rel_offset"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIPersonalityOrLsda>(".cfi_personality"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIPersonalityOrLsda>(".cfi_lsda"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIRememberState>(".cfi_remember_state"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIRestoreState>(".cfi_restore_state"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFISameValue>(".cfi_same_value"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIRestore>(".cfi_restore"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIEscape>(".cfi_escape"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFISignalFrame>(".cfi_signal_frame"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIUndefined>(".cfi_undefined"); - AddDirectiveHandler< - &GenericAsmParser::ParseDirectiveCFIRegister>(".cfi_register"); - - // Macro directives. - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveMacrosOnOff>( - ".macros_on"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveMacrosOnOff>( - ".macros_off"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveMacro>(".macro"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveEndMacro>(".endm"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveEndMacro>(".endmacro"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectivePurgeMacro>(".purgem"); - - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveLEB128>(".sleb128"); - AddDirectiveHandler<&GenericAsmParser::ParseDirectiveLEB128>(".uleb128"); - } - - bool ParseRegisterOrRegisterNumber(int64_t &Register, SMLoc DirectiveLoc); + // "align" + bool ParseDirectiveMSAlign(SMLoc DirectiveLoc, ParseStatementInfo &Info); - bool ParseDirectiveFile(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveLine(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveStabs(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFISections(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIStartProc(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIEndProc(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIDefCfa(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIDefCfaOffset(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIAdjustCfaOffset(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIDefCfaRegister(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIOffset(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIRelOffset(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIPersonalityOrLsda(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIRememberState(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIRestoreState(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFISameValue(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIRestore(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIEscape(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFISignalFrame(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIUndefined(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveCFIRegister(StringRef, SMLoc DirectiveLoc); - - bool ParseDirectiveMacrosOnOff(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveMacro(StringRef, SMLoc DirectiveLoc); - bool ParseDirectiveEndMacro(StringRef, SMLoc DirectiveLoc); - bool ParseDirectivePurgeMacro(StringRef, SMLoc DirectiveLoc); - - bool ParseDirectiveLEB128(StringRef, SMLoc); + void initializeDirectiveKindMap(); }; - } namespace llvm { @@ -469,8 +468,8 @@ enum { DEFAULT_ADDRSPACE = 0 }; AsmParser::AsmParser(SourceMgr &_SM, MCContext &_Ctx, MCStreamer &_Out, const MCAsmInfo &_MAI) : Lexer(_MAI), Ctx(_Ctx), Out(_Out), MAI(_MAI), SrcMgr(_SM), - GenericParser(new GenericAsmParser), PlatformParser(0), - CurBuffer(0), MacrosEnabled(true), CppHashLineNumber(0), + PlatformParser(0), + CurBuffer(0), MacrosEnabledFlag(true), CppHashLineNumber(0), AssemblerDialect(~0U), IsDarwin(false), ParsingInlineAsm(false) { // Save the old handler. SavedDiagHandler = SrcMgr.getDiagHandler(); @@ -479,9 +478,6 @@ AsmParser::AsmParser(SourceMgr &_SM, MCContext &_Ctx, SrcMgr.setDiagHandler(DiagHandler, this); Lexer.setBuffer(SrcMgr.getMemoryBuffer(CurBuffer)); - // Initialize the generic parser. - GenericParser->Initialize(*this); - // Initialize the platform / file format parser. // // FIXME: This is a hack, we need to (majorly) cleanup how these objects are @@ -497,18 +493,19 @@ AsmParser::AsmParser(SourceMgr &_SM, MCContext &_Ctx, PlatformParser = createELFAsmParser(); PlatformParser->Initialize(*this); } + + initializeDirectiveKindMap(); } AsmParser::~AsmParser() { assert(ActiveMacros.empty() && "Unexpected active macro instantiation!"); // Destroy any macros. - for (StringMap<Macro*>::iterator it = MacroMap.begin(), + for (StringMap<MCAsmMacro*>::iterator it = MacroMap.begin(), ie = MacroMap.end(); it != ie; ++it) delete it->getValue(); delete PlatformParser; - delete GenericParser; } void AsmParser::PrintMacroInstantiations() { @@ -609,7 +606,8 @@ bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { getStreamer().EmitLabel(SectionStartSym); getContext().setGenDwarfSectionStartSym(SectionStartSym); getStreamer().EmitDwarfFileDirective(getContext().nextGenDwarfFileNumber(), - StringRef(), SrcMgr.getMemoryBuffer(CurBuffer)->getBufferIdentifier()); + StringRef(), + getContext().getMainFileName()); } // While we have input, parse each statement. @@ -620,7 +618,7 @@ bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { // We had an error, validate that one was emitted and recover by skipping to // the next line. assert(HadError && "Parse statement returned an error, but none emitted!"); - EatToEndOfStatement(); + eatToEndOfStatement(); } if (TheCondState.TheCond != StartingCondState.TheCond || @@ -667,18 +665,15 @@ bool AsmParser::Run(bool NoInitialTextSection, bool NoFinalize) { return HadError; } -void AsmParser::CheckForValidSection() { +void AsmParser::checkForValidSection() { if (!ParsingInlineAsm && !getStreamer().getCurrentSection()) { TokError("expected section directive before assembly directive"); - Out.SwitchSection(Ctx.getMachOSection( - "__TEXT", "__text", - MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, - 0, SectionKind::getText())); + Out.InitToTextSection(); } } -/// EatToEndOfStatement - Throw away the rest of the line for testing purposes. -void AsmParser::EatToEndOfStatement() { +/// eatToEndOfStatement - Throw away the rest of the line for testing purposes. +void AsmParser::eatToEndOfStatement() { while (Lexer.isNot(AsmToken::EndOfStatement) && Lexer.isNot(AsmToken::Eof)) Lex(); @@ -688,7 +683,7 @@ void AsmParser::EatToEndOfStatement() { Lex(); } -StringRef AsmParser::ParseStringToEndOfStatement() { +StringRef AsmParser::parseStringToEndOfStatement() { const char *Start = getTok().getLoc().getPointer(); while (Lexer.isNot(AsmToken::EndOfStatement) && @@ -717,10 +712,10 @@ StringRef AsmParser::ParseStringToComma() { /// parenexpr ::= expr) /// bool AsmParser::ParseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { - if (ParseExpression(Res)) return true; + if (parseExpression(Res)) return true; if (Lexer.isNot(AsmToken::RParen)) return TokError("expected ')' in parentheses expression"); - EndLoc = Lexer.getLoc(); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); return false; } @@ -731,10 +726,10 @@ bool AsmParser::ParseParenExpr(const MCExpr *&Res, SMLoc &EndLoc) { /// bracketexpr ::= expr] /// bool AsmParser::ParseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { - if (ParseExpression(Res)) return true; + if (parseExpression(Res)) return true; if (Lexer.isNot(AsmToken::RBrac)) return TokError("expected ']' in brackets expression"); - EndLoc = Lexer.getLoc(); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); return false; } @@ -746,7 +741,9 @@ bool AsmParser::ParseBracketExpr(const MCExpr *&Res, SMLoc &EndLoc) { /// primaryexpr ::= '.' /// primaryexpr ::= ~,+,- primaryexpr bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { - switch (Lexer.getKind()) { + SMLoc FirstTokenLoc = getLexer().getLoc(); + AsmToken::TokenKind FirstTokenKind = Lexer.getKind(); + switch (FirstTokenKind) { default: return TokError("unknown token in expression"); // If we have an error assume that we've already handled it. @@ -761,11 +758,14 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { case AsmToken::Dollar: case AsmToken::String: case AsmToken::Identifier: { - EndLoc = Lexer.getLoc(); - StringRef Identifier; - if (ParseIdentifier(Identifier)) + if (parseIdentifier(Identifier)) { + if (FirstTokenKind == AsmToken::Dollar) + return Error(FirstTokenLoc, "invalid token in expression"); return true; + } + + EndLoc = SMLoc::getFromPointer(Identifier.end()); // This is a symbol reference. std::pair<StringRef, StringRef> Split = Identifier.split('@'); @@ -799,7 +799,7 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { SMLoc Loc = getTok().getLoc(); int64_t IntVal = getTok().getIntVal(); Res = MCConstantExpr::Create(IntVal, getContext()); - EndLoc = Lexer.getLoc(); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); // Eat token. // Look for 'b' or 'f' following an Integer as a directional label if (Lexer.getKind() == AsmToken::Identifier) { @@ -811,7 +811,7 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { getContext()); if (IDVal == "b" && Sym->isUndefined()) return Error(Loc, "invalid reference to undefined symbol"); - EndLoc = Lexer.getLoc(); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); // Eat identifier. } } @@ -821,6 +821,7 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { APFloat RealVal(APFloat::IEEEdouble, getTok().getString()); uint64_t IntVal = RealVal.bitcastToAPInt().getZExtValue(); Res = MCConstantExpr::Create(IntVal, getContext()); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); // Eat token. return false; } @@ -830,7 +831,7 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { MCSymbol *Sym = Ctx.CreateTempSymbol(); Out.EmitLabel(Sym); Res = MCSymbolRefExpr::Create(Sym, MCSymbolRefExpr::VK_None, getContext()); - EndLoc = Lexer.getLoc(); + EndLoc = Lexer.getTok().getEndLoc(); Lex(); // Eat identifier. return false; } @@ -863,9 +864,9 @@ bool AsmParser::ParsePrimaryExpr(const MCExpr *&Res, SMLoc &EndLoc) { } } -bool AsmParser::ParseExpression(const MCExpr *&Res) { +bool AsmParser::parseExpression(const MCExpr *&Res) { SMLoc EndLoc; - return ParseExpression(Res, EndLoc); + return parseExpression(Res, EndLoc); } const MCExpr * @@ -916,7 +917,7 @@ AsmParser::ApplyModifierToExpr(const MCExpr *E, llvm_unreachable("Invalid expression kind!"); } -/// ParseExpression - Parse an expression and return it. +/// parseExpression - Parse an expression and return it. /// /// expr ::= expr &&,|| expr -> lowest. /// expr ::= expr |,^,&,! expr @@ -926,7 +927,7 @@ AsmParser::ApplyModifierToExpr(const MCExpr *E, /// expr ::= expr *,/,% expr -> highest. /// expr ::= primaryexpr /// -bool AsmParser::ParseExpression(const MCExpr *&Res, SMLoc &EndLoc) { +bool AsmParser::parseExpression(const MCExpr *&Res, SMLoc &EndLoc) { // Parse the expression. Res = 0; if (ParsePrimaryExpr(Res, EndLoc) || ParseBinOpRHS(1, Res, EndLoc)) @@ -964,17 +965,17 @@ bool AsmParser::ParseExpression(const MCExpr *&Res, SMLoc &EndLoc) { return false; } -bool AsmParser::ParseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { +bool AsmParser::parseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) { Res = 0; return ParseParenExpr(Res, EndLoc) || ParseBinOpRHS(1, Res, EndLoc); } -bool AsmParser::ParseAbsoluteExpression(int64_t &Res) { +bool AsmParser::parseAbsoluteExpression(int64_t &Res) { const MCExpr *Expr; SMLoc StartLoc = Lexer.getLoc(); - if (ParseExpression(Expr)) + if (parseExpression(Expr)) return true; if (!Expr->EvaluateAsAbsolute(Res)) @@ -1121,8 +1122,7 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { if (!TheCondState.Ignore) return TokError("unexpected token at start of statement"); IDVal = ""; - } - else { + } else { IDVal = getTok().getString(); Lex(); // Consume the integer token to be used as an identifier token. if (Lexer.getKind() != AsmToken::Colon) { @@ -1130,46 +1130,54 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { return TokError("unexpected token at start of statement"); } } - } else if (Lexer.is(AsmToken::Dot)) { // Treat '.' as a valid identifier in this context. Lex(); IDVal = "."; - - } else if (ParseIdentifier(IDVal)) { + } else if (parseIdentifier(IDVal)) { if (!TheCondState.Ignore) return TokError("unexpected token at start of statement"); IDVal = ""; } - // Handle conditional assembly here before checking for skipping. We // have to do this so that .endif isn't skipped in a ".if 0" block for // example. - if (IDVal == ".if") - return ParseDirectiveIf(IDLoc); - if (IDVal == ".ifb") - return ParseDirectiveIfb(IDLoc, true); - if (IDVal == ".ifnb") - return ParseDirectiveIfb(IDLoc, false); - if (IDVal == ".ifc") - return ParseDirectiveIfc(IDLoc, true); - if (IDVal == ".ifnc") - return ParseDirectiveIfc(IDLoc, false); - if (IDVal == ".ifdef") - return ParseDirectiveIfdef(IDLoc, true); - if (IDVal == ".ifndef" || IDVal == ".ifnotdef") - return ParseDirectiveIfdef(IDLoc, false); - if (IDVal == ".elseif") - return ParseDirectiveElseIf(IDLoc); - if (IDVal == ".else") - return ParseDirectiveElse(IDLoc); - if (IDVal == ".endif") - return ParseDirectiveEndIf(IDLoc); - - // If we are in a ".if 0" block, ignore this statement. + StringMap<DirectiveKind>::const_iterator DirKindIt = + DirectiveKindMap.find(IDVal); + DirectiveKind DirKind = + (DirKindIt == DirectiveKindMap.end()) ? DK_NO_DIRECTIVE : + DirKindIt->getValue(); + switch (DirKind) { + default: + break; + case DK_IF: + return ParseDirectiveIf(IDLoc); + case DK_IFB: + return ParseDirectiveIfb(IDLoc, true); + case DK_IFNB: + return ParseDirectiveIfb(IDLoc, false); + case DK_IFC: + return ParseDirectiveIfc(IDLoc, true); + case DK_IFNC: + return ParseDirectiveIfc(IDLoc, false); + case DK_IFDEF: + return ParseDirectiveIfdef(IDLoc, true); + case DK_IFNDEF: + case DK_IFNOTDEF: + return ParseDirectiveIfdef(IDLoc, false); + case DK_ELSEIF: + return ParseDirectiveElseIf(IDLoc); + case DK_ELSE: + return ParseDirectiveElse(IDLoc); + case DK_ENDIF: + return ParseDirectiveEndIf(IDLoc); + } + + // Ignore the statement if in the middle of inactive conditional + // (e.g. ".if 0"). if (TheCondState.Ignore) { - EatToEndOfStatement(); + eatToEndOfStatement(); return false; } @@ -1178,7 +1186,7 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { // See what kind of statement we have. switch (Lexer.getKind()) { case AsmToken::Colon: { - CheckForValidSection(); + checkForValidSection(); // identifier ':' -> Label. Lex(); @@ -1201,7 +1209,8 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { return Error(IDLoc, "invalid symbol redefinition"); // Emit the label. - Out.EmitLabel(Sym); + if (!ParsingInlineAsm) + Out.EmitLabel(Sym); // If we are generating dwarf for assembly source files then gather the // info to make a dwarf label entry for this label if needed. @@ -1231,160 +1240,234 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { } // If macros are enabled, check to see if this is a macro instantiation. - if (MacrosEnabled) - if (const Macro *M = MacroMap.lookup(IDVal)) - return HandleMacroEntry(IDVal, IDLoc, M); + if (MacrosEnabled()) + if (const MCAsmMacro *M = LookupMacro(IDVal)) { + return HandleMacroEntry(M, IDLoc); + } // Otherwise, we have a normal instruction or directive. + + // Directives start with "." if (IDVal[0] == '.' && IDVal != ".") { - - // Target hook for parsing target specific directives. + // There are several entities interested in parsing directives: + // + // 1. The target-specific assembly parser. Some directives are target + // specific or may potentially behave differently on certain targets. + // 2. Asm parser extensions. For example, platform-specific parsers + // (like the ELF parser) register themselves as extensions. + // 3. The generic directive parser implemented by this class. These are + // all the directives that behave in a target and platform independent + // manner, or at least have a default behavior that's shared between + // all targets and platforms. + + // First query the target-specific parser. It will return 'true' if it + // isn't interested in this directive. if (!getTargetParser().ParseDirective(ID)) return false; - // Assembler features - if (IDVal == ".set" || IDVal == ".equ") - return ParseDirectiveSet(IDVal, true); - if (IDVal == ".equiv") - return ParseDirectiveSet(IDVal, false); - - // Data directives - - if (IDVal == ".ascii") - return ParseDirectiveAscii(IDVal, false); - if (IDVal == ".asciz" || IDVal == ".string") - return ParseDirectiveAscii(IDVal, true); - - if (IDVal == ".byte") - return ParseDirectiveValue(1); - if (IDVal == ".short") - return ParseDirectiveValue(2); - if (IDVal == ".value") - return ParseDirectiveValue(2); - if (IDVal == ".2byte") - return ParseDirectiveValue(2); - if (IDVal == ".long") - return ParseDirectiveValue(4); - if (IDVal == ".int") - return ParseDirectiveValue(4); - if (IDVal == ".4byte") - return ParseDirectiveValue(4); - if (IDVal == ".quad") - return ParseDirectiveValue(8); - if (IDVal == ".8byte") - return ParseDirectiveValue(8); - if (IDVal == ".single" || IDVal == ".float") - return ParseDirectiveRealValue(APFloat::IEEEsingle); - if (IDVal == ".double") - return ParseDirectiveRealValue(APFloat::IEEEdouble); - - if (IDVal == ".align") { - bool IsPow2 = !getContext().getAsmInfo().getAlignmentIsInBytes(); - return ParseDirectiveAlign(IsPow2, /*ExprSize=*/1); - } - if (IDVal == ".align32") { - bool IsPow2 = !getContext().getAsmInfo().getAlignmentIsInBytes(); - return ParseDirectiveAlign(IsPow2, /*ExprSize=*/4); - } - if (IDVal == ".balign") - return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); - if (IDVal == ".balignw") - return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); - if (IDVal == ".balignl") - return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); - if (IDVal == ".p2align") - return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); - if (IDVal == ".p2alignw") - return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); - if (IDVal == ".p2alignl") - return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); - - if (IDVal == ".org") - return ParseDirectiveOrg(); - - if (IDVal == ".fill") - return ParseDirectiveFill(); - if (IDVal == ".space" || IDVal == ".skip") - return ParseDirectiveSpace(); - if (IDVal == ".zero") - return ParseDirectiveZero(); - - // Symbol attribute directives - - if (IDVal == ".extern") { - EatToEndOfStatement(); // .extern is the default, ignore it. - return false; - } - if (IDVal == ".globl" || IDVal == ".global") - return ParseDirectiveSymbolAttribute(MCSA_Global); - if (IDVal == ".indirect_symbol") - return ParseDirectiveSymbolAttribute(MCSA_IndirectSymbol); - if (IDVal == ".lazy_reference") - return ParseDirectiveSymbolAttribute(MCSA_LazyReference); - if (IDVal == ".no_dead_strip") - return ParseDirectiveSymbolAttribute(MCSA_NoDeadStrip); - if (IDVal == ".symbol_resolver") - return ParseDirectiveSymbolAttribute(MCSA_SymbolResolver); - if (IDVal == ".private_extern") - return ParseDirectiveSymbolAttribute(MCSA_PrivateExtern); - if (IDVal == ".reference") - return ParseDirectiveSymbolAttribute(MCSA_Reference); - if (IDVal == ".weak_definition") - return ParseDirectiveSymbolAttribute(MCSA_WeakDefinition); - if (IDVal == ".weak_reference") - return ParseDirectiveSymbolAttribute(MCSA_WeakReference); - if (IDVal == ".weak_def_can_be_hidden") - return ParseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); - - if (IDVal == ".comm" || IDVal == ".common") - return ParseDirectiveComm(/*IsLocal=*/false); - if (IDVal == ".lcomm") - return ParseDirectiveComm(/*IsLocal=*/true); - - if (IDVal == ".abort") - return ParseDirectiveAbort(); - if (IDVal == ".include") - return ParseDirectiveInclude(); - if (IDVal == ".incbin") - return ParseDirectiveIncbin(); - - if (IDVal == ".code16" || IDVal == ".code16gcc") - return TokError(Twine(IDVal) + " not supported yet"); - - // Macro-like directives - if (IDVal == ".rept") - return ParseDirectiveRept(IDLoc); - if (IDVal == ".irp") - return ParseDirectiveIrp(IDLoc); - if (IDVal == ".irpc") - return ParseDirectiveIrpc(IDLoc); - if (IDVal == ".endr") - return ParseDirectiveEndr(IDLoc); - - // Look up the handler in the handler table. + // Next, check the extention directive map to see if any extension has + // registered itself to parse this directive. std::pair<MCAsmParserExtension*, DirectiveHandler> Handler = - DirectiveMap.lookup(IDVal); + ExtensionDirectiveMap.lookup(IDVal); if (Handler.first) return (*Handler.second)(Handler.first, IDVal, IDLoc); + // Finally, if no one else is interested in this directive, it must be + // generic and familiar to this class. + switch (DirKind) { + default: + break; + case DK_SET: + case DK_EQU: + return ParseDirectiveSet(IDVal, true); + case DK_EQUIV: + return ParseDirectiveSet(IDVal, false); + case DK_ASCII: + return ParseDirectiveAscii(IDVal, false); + case DK_ASCIZ: + case DK_STRING: + return ParseDirectiveAscii(IDVal, true); + case DK_BYTE: + return ParseDirectiveValue(1); + case DK_SHORT: + case DK_VALUE: + case DK_2BYTE: + return ParseDirectiveValue(2); + case DK_LONG: + case DK_INT: + case DK_4BYTE: + return ParseDirectiveValue(4); + case DK_QUAD: + case DK_8BYTE: + return ParseDirectiveValue(8); + case DK_SINGLE: + case DK_FLOAT: + return ParseDirectiveRealValue(APFloat::IEEEsingle); + case DK_DOUBLE: + return ParseDirectiveRealValue(APFloat::IEEEdouble); + case DK_ALIGN: { + bool IsPow2 = !getContext().getAsmInfo().getAlignmentIsInBytes(); + return ParseDirectiveAlign(IsPow2, /*ExprSize=*/1); + } + case DK_ALIGN32: { + bool IsPow2 = !getContext().getAsmInfo().getAlignmentIsInBytes(); + return ParseDirectiveAlign(IsPow2, /*ExprSize=*/4); + } + case DK_BALIGN: + return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/1); + case DK_BALIGNW: + return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/2); + case DK_BALIGNL: + return ParseDirectiveAlign(/*IsPow2=*/false, /*ExprSize=*/4); + case DK_P2ALIGN: + return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/1); + case DK_P2ALIGNW: + return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/2); + case DK_P2ALIGNL: + return ParseDirectiveAlign(/*IsPow2=*/true, /*ExprSize=*/4); + case DK_ORG: + return ParseDirectiveOrg(); + case DK_FILL: + return ParseDirectiveFill(); + case DK_ZERO: + return ParseDirectiveZero(); + case DK_EXTERN: + eatToEndOfStatement(); // .extern is the default, ignore it. + return false; + case DK_GLOBL: + case DK_GLOBAL: + return ParseDirectiveSymbolAttribute(MCSA_Global); + case DK_INDIRECT_SYMBOL: + return ParseDirectiveSymbolAttribute(MCSA_IndirectSymbol); + case DK_LAZY_REFERENCE: + return ParseDirectiveSymbolAttribute(MCSA_LazyReference); + case DK_NO_DEAD_STRIP: + return ParseDirectiveSymbolAttribute(MCSA_NoDeadStrip); + case DK_SYMBOL_RESOLVER: + return ParseDirectiveSymbolAttribute(MCSA_SymbolResolver); + case DK_PRIVATE_EXTERN: + return ParseDirectiveSymbolAttribute(MCSA_PrivateExtern); + case DK_REFERENCE: + return ParseDirectiveSymbolAttribute(MCSA_Reference); + case DK_WEAK_DEFINITION: + return ParseDirectiveSymbolAttribute(MCSA_WeakDefinition); + case DK_WEAK_REFERENCE: + return ParseDirectiveSymbolAttribute(MCSA_WeakReference); + case DK_WEAK_DEF_CAN_BE_HIDDEN: + return ParseDirectiveSymbolAttribute(MCSA_WeakDefAutoPrivate); + case DK_COMM: + case DK_COMMON: + return ParseDirectiveComm(/*IsLocal=*/false); + case DK_LCOMM: + return ParseDirectiveComm(/*IsLocal=*/true); + case DK_ABORT: + return ParseDirectiveAbort(); + case DK_INCLUDE: + return ParseDirectiveInclude(); + case DK_INCBIN: + return ParseDirectiveIncbin(); + case DK_CODE16: + case DK_CODE16GCC: + return TokError(Twine(IDVal) + " not supported yet"); + case DK_REPT: + return ParseDirectiveRept(IDLoc); + case DK_IRP: + return ParseDirectiveIrp(IDLoc); + case DK_IRPC: + return ParseDirectiveIrpc(IDLoc); + case DK_ENDR: + return ParseDirectiveEndr(IDLoc); + case DK_BUNDLE_ALIGN_MODE: + return ParseDirectiveBundleAlignMode(); + case DK_BUNDLE_LOCK: + return ParseDirectiveBundleLock(); + case DK_BUNDLE_UNLOCK: + return ParseDirectiveBundleUnlock(); + case DK_SLEB128: + return ParseDirectiveLEB128(true); + case DK_ULEB128: + return ParseDirectiveLEB128(false); + case DK_SPACE: + case DK_SKIP: + return ParseDirectiveSpace(IDVal); + case DK_FILE: + return ParseDirectiveFile(IDLoc); + case DK_LINE: + return ParseDirectiveLine(); + case DK_LOC: + return ParseDirectiveLoc(); + case DK_STABS: + return ParseDirectiveStabs(); + case DK_CFI_SECTIONS: + return ParseDirectiveCFISections(); + case DK_CFI_STARTPROC: + return ParseDirectiveCFIStartProc(); + case DK_CFI_ENDPROC: + return ParseDirectiveCFIEndProc(); + case DK_CFI_DEF_CFA: + return ParseDirectiveCFIDefCfa(IDLoc); + case DK_CFI_DEF_CFA_OFFSET: + return ParseDirectiveCFIDefCfaOffset(); + case DK_CFI_ADJUST_CFA_OFFSET: + return ParseDirectiveCFIAdjustCfaOffset(); + case DK_CFI_DEF_CFA_REGISTER: + return ParseDirectiveCFIDefCfaRegister(IDLoc); + case DK_CFI_OFFSET: + return ParseDirectiveCFIOffset(IDLoc); + case DK_CFI_REL_OFFSET: + return ParseDirectiveCFIRelOffset(IDLoc); + case DK_CFI_PERSONALITY: + return ParseDirectiveCFIPersonalityOrLsda(true); + case DK_CFI_LSDA: + return ParseDirectiveCFIPersonalityOrLsda(false); + case DK_CFI_REMEMBER_STATE: + return ParseDirectiveCFIRememberState(); + case DK_CFI_RESTORE_STATE: + return ParseDirectiveCFIRestoreState(); + case DK_CFI_SAME_VALUE: + return ParseDirectiveCFISameValue(IDLoc); + case DK_CFI_RESTORE: + return ParseDirectiveCFIRestore(IDLoc); + case DK_CFI_ESCAPE: + return ParseDirectiveCFIEscape(); + case DK_CFI_SIGNAL_FRAME: + return ParseDirectiveCFISignalFrame(); + case DK_CFI_UNDEFINED: + return ParseDirectiveCFIUndefined(IDLoc); + case DK_CFI_REGISTER: + return ParseDirectiveCFIRegister(IDLoc); + case DK_MACROS_ON: + case DK_MACROS_OFF: + return ParseDirectiveMacrosOnOff(IDVal); + case DK_MACRO: + return ParseDirectiveMacro(IDLoc); + case DK_ENDM: + case DK_ENDMACRO: + return ParseDirectiveEndMacro(IDVal); + case DK_PURGEM: + return ParseDirectivePurgeMacro(IDLoc); + } return Error(IDLoc, "unknown directive"); } - // _emit - if (ParsingInlineAsm && IDVal == "_emit") - return ParseDirectiveEmit(IDLoc, Info); + // __asm _emit or __asm __emit + if (ParsingInlineAsm && (IDVal == "_emit" || IDVal == "__emit" || + IDVal == "_EMIT" || IDVal == "__EMIT")) + return ParseDirectiveMSEmit(IDLoc, Info, IDVal.size()); - CheckForValidSection(); + // __asm align + if (ParsingInlineAsm && (IDVal == "align" || IDVal == "ALIGN")) + return ParseDirectiveMSAlign(IDLoc, Info); - // Canonicalize the opcode to lower case. - SmallString<128> OpcodeStr; - for (unsigned i = 0, e = IDVal.size(); i != e; ++i) - OpcodeStr.push_back(tolower(IDVal[i])); + checkForValidSection(); + // Canonicalize the opcode to lower case. + std::string OpcodeStr = IDVal.lower(); ParseInstructionInfo IInfo(Info.AsmRewrites); - bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr.str(), - IDLoc,Info.ParsedOperands); + bool HadError = getTargetParser().ParseInstruction(IInfo, OpcodeStr, + IDLoc, Info.ParsedOperands); + Info.ParseError = HadError; // Dump the parsed representation, if requested. if (getShowParsedOperands()) { @@ -1405,24 +1488,24 @@ bool AsmParser::ParseStatement(ParseStatementInfo &Info) { // section is the initial text section then generate a .loc directive for // the instruction. if (!HadError && getContext().getGenDwarfForAssembly() && - getContext().getGenDwarfSection() == getStreamer().getCurrentSection() ) { + getContext().getGenDwarfSection() == getStreamer().getCurrentSection()) { - unsigned Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); + unsigned Line = SrcMgr.FindLineNumber(IDLoc, CurBuffer); - // If we previously parsed a cpp hash file line comment then make sure the - // current Dwarf File is for the CppHashFilename if not then emit the - // Dwarf File table for it and adjust the line number for the .loc. - const std::vector<MCDwarfFile *> &MCDwarfFiles = - getContext().getMCDwarfFiles(); - if (CppHashFilename.size() != 0) { - if(MCDwarfFiles[getContext().getGenDwarfFileNumber()]->getName() != + // If we previously parsed a cpp hash file line comment then make sure the + // current Dwarf File is for the CppHashFilename if not then emit the + // Dwarf File table for it and adjust the line number for the .loc. + const std::vector<MCDwarfFile *> &MCDwarfFiles = + getContext().getMCDwarfFiles(); + if (CppHashFilename.size() != 0) { + if (MCDwarfFiles[getContext().getGenDwarfFileNumber()]->getName() != CppHashFilename) - getStreamer().EmitDwarfFileDirective( - getContext().nextGenDwarfFileNumber(), StringRef(), CppHashFilename); + getStreamer().EmitDwarfFileDirective( + getContext().nextGenDwarfFileNumber(), StringRef(), CppHashFilename); unsigned CppHashLocLineNo = SrcMgr.FindLineNumber(CppHashLoc,CppHashBuf); Line = CppHashLineNumber - 1 + (Line - CppHashLocLineNo); - } + } getStreamer().EmitDwarfLocDirective(getContext().getGenDwarfFileNumber(), Line, 0, DWARF2_LINE_DEFAULT_IS_STMT ? @@ -1508,7 +1591,7 @@ void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { DiagSrcMgr.PrintIncludeStack(ParentIncludeLoc, OS); } - // If we have not parsed a cpp hash line filename comment or the source + // If we have not parsed a cpp hash line filename comment or the source // manager changed or buffer changed (like in a nested include) then just // print the normal diagnostic using its Filename and LineNo. if (!Parser->CppHashLineNumber || @@ -1521,7 +1604,7 @@ void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { return; } - // Use the CppHashFilename and calculate a line number based on the + // Use the CppHashFilename and calculate a line number based on the // CppHashLoc and CppHashLineNumber relative to this Diag's SMLoc for // the diagnostic. const std::string Filename = Parser->CppHashFilename; @@ -1548,12 +1631,13 @@ void AsmParser::DiagHandler(const SMDiagnostic &Diag, void *Context) { // we can't do that. AsmLexer.cpp should probably be changed to handle // '@' as a special case when needed. static bool isIdentifierChar(char c) { - return isalnum(c) || c == '_' || c == '$' || c == '.'; + return isalnum(static_cast<unsigned char>(c)) || c == '_' || c == '$' || + c == '.'; } bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, - const MacroParameters &Parameters, - const MacroArguments &A, + const MCAsmMacroParameters &Parameters, + const MCAsmMacroArguments &A, const SMLoc &L) { unsigned NParameters = Parameters.size(); if (NParameters != 0 && NParameters != A.size()) @@ -1572,7 +1656,8 @@ bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, continue; char Next = Body[Pos + 1]; - if (Next == '$' || Next == 'n' || isdigit(Next)) + if (Next == '$' || Next == 'n' || + isdigit(static_cast<unsigned char>(Next))) break; } else { // This macro has parameters, look for \foo, \bar, etc. @@ -1608,7 +1693,7 @@ bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, break; // Otherwise substitute with the token values, with spaces eliminated. - for (MacroArgument::const_iterator it = A[Index].begin(), + for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), ie = A[Index].end(); it != ie; ++it) OS << it->getString(); break; @@ -1635,7 +1720,7 @@ bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, Pos = I; } } else { - for (MacroArgument::const_iterator it = A[Index].begin(), + for (MCAsmMacroArgument::const_iterator it = A[Index].begin(), ie = A[Index].end(); it != ie; ++it) if (it->getKind() == AsmToken::String) OS << it->getStringContents(); @@ -1652,7 +1737,7 @@ bool AsmParser::expandMacro(raw_svector_ostream &OS, StringRef Body, return false; } -MacroInstantiation::MacroInstantiation(const Macro *M, SMLoc IL, +MacroInstantiation::MacroInstantiation(const MCAsmMacro *M, SMLoc IL, int EB, SMLoc EL, MemoryBuffer *I) : TheMacro(M), Instantiation(I), InstantiationLoc(IL), ExitBuffer(EB), @@ -1693,10 +1778,7 @@ static bool IsOperator(AsmToken::TokenKind kind) } } -/// ParseMacroArgument - Extract AsmTokens for a macro argument. -/// This is used for both default macro parameter values and the -/// arguments in macro invocations -bool AsmParser::ParseMacroArgument(MacroArgument &MA, +bool AsmParser::ParseMacroArgument(MCAsmMacroArgument &MA, AsmToken::TokenKind &ArgumentDelimiter) { unsigned ParenLevel = 0; unsigned AddTokens = 0; @@ -1733,7 +1815,7 @@ bool AsmParser::ParseMacroArgument(MacroArgument &MA, if (IsOperator(Lexer.getKind())) { // Check to see whether the token is used as an operator, // or part of an identifier - const char *NextChar = getTok().getEndLoc().getPointer() + 1; + const char *NextChar = getTok().getEndLoc().getPointer(); if (*NextChar == ' ') AddTokens = 2; } @@ -1772,7 +1854,7 @@ bool AsmParser::ParseMacroArgument(MacroArgument &MA, } // Parse the macro instantiation arguments. -bool AsmParser::ParseMacroArguments(const Macro *M, MacroArguments &A) { +bool AsmParser::ParseMacroArguments(const MCAsmMacro *M, MCAsmMacroArguments &A) { const unsigned NParameters = M ? M->Parameters.size() : 0; // Argument delimiter is initially unknown. It will be set by // ParseMacroArgument() @@ -1783,7 +1865,7 @@ bool AsmParser::ParseMacroArguments(const Macro *M, MacroArguments &A) { // - macros defined with parameters accept at most that many of them for (unsigned Parameter = 0; !NParameters || Parameter < NParameters; ++Parameter) { - MacroArgument MA; + MCAsmMacroArgument MA; if (ParseMacroArgument(MA, ArgumentDelimiter)) return true; @@ -1816,14 +1898,30 @@ bool AsmParser::ParseMacroArguments(const Macro *M, MacroArguments &A) { return TokError("Too many arguments"); } -bool AsmParser::HandleMacroEntry(StringRef Name, SMLoc NameLoc, - const Macro *M) { +const MCAsmMacro* AsmParser::LookupMacro(StringRef Name) { + StringMap<MCAsmMacro*>::iterator I = MacroMap.find(Name); + return (I == MacroMap.end()) ? NULL : I->getValue(); +} + +void AsmParser::DefineMacro(StringRef Name, const MCAsmMacro& Macro) { + MacroMap[Name] = new MCAsmMacro(Macro); +} + +void AsmParser::UndefineMacro(StringRef Name) { + StringMap<MCAsmMacro*>::iterator I = MacroMap.find(Name); + if (I != MacroMap.end()) { + delete I->getValue(); + MacroMap.erase(I); + } +} + +bool AsmParser::HandleMacroEntry(const MCAsmMacro *M, SMLoc NameLoc) { // Arbitrarily limit macro nesting depth, to match 'as'. We can eliminate // this, although we should protect against infinite loops. if (ActiveMacros.size() == 20) return TokError("macros cannot be nested more than 20 levels deep"); - MacroArguments A; + MCAsmMacroArguments A; if (ParseMacroArguments(M, A)) return true; @@ -1842,7 +1940,7 @@ bool AsmParser::HandleMacroEntry(StringRef Name, SMLoc NameLoc, if (expandMacro(OS, Body, M->Parameters, A, getTok().getLoc())) return true; - // We include the .endmacro in the buffer as our queue to exit the macro + // We include the .endmacro in the buffer as our cue to exit the macro // instantiation. OS << ".endmacro\n"; @@ -1904,7 +2002,7 @@ bool AsmParser::ParseAssignment(StringRef Name, bool allow_redef, SMLoc EqualLoc = Lexer.getLoc(); const MCExpr *Value; - if (ParseExpression(Value)) + if (parseExpression(Value)) return true; // Note: we don't count b as used in "a = b". This is to allow @@ -1961,10 +2059,10 @@ bool AsmParser::ParseAssignment(StringRef Name, bool allow_redef, return false; } -/// ParseIdentifier: +/// parseIdentifier: /// ::= identifier /// ::= string -bool AsmParser::ParseIdentifier(StringRef &Res) { +bool AsmParser::parseIdentifier(StringRef &Res) { // The assembler has relaxed rules for accepting identifiers, in particular we // allow things like '.globl $foo', which would normally be separate // tokens. At this level, we have already lexed so we cannot (currently) @@ -2007,7 +2105,7 @@ bool AsmParser::ParseIdentifier(StringRef &Res) { bool AsmParser::ParseDirectiveSet(StringRef IDVal, bool allow_redef) { StringRef Name; - if (ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("expected identifier after '" + Twine(IDVal) + "'"); if (getLexer().isNot(AsmToken::Comma)) @@ -2017,7 +2115,7 @@ bool AsmParser::ParseDirectiveSet(StringRef IDVal, bool allow_redef) { return ParseAssignment(Name, allow_redef, true); } -bool AsmParser::ParseEscapedString(std::string &Data) { +bool AsmParser::parseEscapedString(std::string &Data) { assert(getLexer().is(AsmToken::String) && "Unexpected current token!"); Data = ""; @@ -2079,14 +2177,14 @@ bool AsmParser::ParseEscapedString(std::string &Data) { /// ::= ( .ascii | .asciz | .string ) [ "string" ( , "string" )* ] bool AsmParser::ParseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { if (getLexer().isNot(AsmToken::EndOfStatement)) { - CheckForValidSection(); + checkForValidSection(); for (;;) { if (getLexer().isNot(AsmToken::String)) return TokError("expected string in '" + Twine(IDVal) + "' directive"); std::string Data; - if (ParseEscapedString(Data)) + if (parseEscapedString(Data)) return true; getStreamer().EmitBytes(Data, DEFAULT_ADDRSPACE); @@ -2112,12 +2210,12 @@ bool AsmParser::ParseDirectiveAscii(StringRef IDVal, bool ZeroTerminated) { /// ::= (.byte | .short | ... ) [ expression (, expression)* ] bool AsmParser::ParseDirectiveValue(unsigned Size) { if (getLexer().isNot(AsmToken::EndOfStatement)) { - CheckForValidSection(); + checkForValidSection(); for (;;) { const MCExpr *Value; SMLoc ExprLoc = getLexer().getLoc(); - if (ParseExpression(Value)) + if (parseExpression(Value)) return true; // Special case constant expressions to match code generator. @@ -2148,7 +2246,7 @@ bool AsmParser::ParseDirectiveValue(unsigned Size) { /// ::= (.single | .double) [ expression (, expression)* ] bool AsmParser::ParseDirectiveRealValue(const fltSemantics &Semantics) { if (getLexer().isNot(AsmToken::EndOfStatement)) { - CheckForValidSection(); + checkForValidSection(); for (;;) { // We don't truly support arithmetic on floating point expressions, so we @@ -2202,52 +2300,19 @@ bool AsmParser::ParseDirectiveRealValue(const fltSemantics &Semantics) { return false; } -/// ParseDirectiveSpace -/// ::= .space expression [ , expression ] -bool AsmParser::ParseDirectiveSpace() { - CheckForValidSection(); - - int64_t NumBytes; - if (ParseAbsoluteExpression(NumBytes)) - return true; - - int64_t FillExpr = 0; - if (getLexer().isNot(AsmToken::EndOfStatement)) { - if (getLexer().isNot(AsmToken::Comma)) - return TokError("unexpected token in '.space' directive"); - Lex(); - - if (ParseAbsoluteExpression(FillExpr)) - return true; - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.space' directive"); - } - - Lex(); - - if (NumBytes <= 0) - return TokError("invalid number of bytes in '.space' directive"); - - // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. - getStreamer().EmitFill(NumBytes, FillExpr, DEFAULT_ADDRSPACE); - - return false; -} - /// ParseDirectiveZero /// ::= .zero expression bool AsmParser::ParseDirectiveZero() { - CheckForValidSection(); + checkForValidSection(); int64_t NumBytes; - if (ParseAbsoluteExpression(NumBytes)) + if (parseAbsoluteExpression(NumBytes)) return true; int64_t Val = 0; if (getLexer().is(AsmToken::Comma)) { Lex(); - if (ParseAbsoluteExpression(Val)) + if (parseAbsoluteExpression(Val)) return true; } @@ -2264,10 +2329,10 @@ bool AsmParser::ParseDirectiveZero() { /// ParseDirectiveFill /// ::= .fill expression , expression , expression bool AsmParser::ParseDirectiveFill() { - CheckForValidSection(); + checkForValidSection(); int64_t NumValues; - if (ParseAbsoluteExpression(NumValues)) + if (parseAbsoluteExpression(NumValues)) return true; if (getLexer().isNot(AsmToken::Comma)) @@ -2275,7 +2340,7 @@ bool AsmParser::ParseDirectiveFill() { Lex(); int64_t FillSize; - if (ParseAbsoluteExpression(FillSize)) + if (parseAbsoluteExpression(FillSize)) return true; if (getLexer().isNot(AsmToken::Comma)) @@ -2283,7 +2348,7 @@ bool AsmParser::ParseDirectiveFill() { Lex(); int64_t FillExpr; - if (ParseAbsoluteExpression(FillExpr)) + if (parseAbsoluteExpression(FillExpr)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -2303,11 +2368,11 @@ bool AsmParser::ParseDirectiveFill() { /// ParseDirectiveOrg /// ::= .org expression [ , expression ] bool AsmParser::ParseDirectiveOrg() { - CheckForValidSection(); + checkForValidSection(); const MCExpr *Offset; SMLoc Loc = getTok().getLoc(); - if (ParseExpression(Offset)) + if (parseExpression(Offset)) return true; // Parse optional fill expression. @@ -2317,7 +2382,7 @@ bool AsmParser::ParseDirectiveOrg() { return TokError("unexpected token in '.org' directive"); Lex(); - if (ParseAbsoluteExpression(FillExpr)) + if (parseAbsoluteExpression(FillExpr)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -2338,11 +2403,11 @@ bool AsmParser::ParseDirectiveOrg() { /// ParseDirectiveAlign /// ::= {.align, ...} expression [ , expression [ , expression ]] bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { - CheckForValidSection(); + checkForValidSection(); SMLoc AlignmentLoc = getLexer().getLoc(); int64_t Alignment; - if (ParseAbsoluteExpression(Alignment)) + if (parseAbsoluteExpression(Alignment)) return true; SMLoc MaxBytesLoc; @@ -2359,7 +2424,7 @@ bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { // .align 3,,4 if (getLexer().isNot(AsmToken::Comma)) { HasFillExpr = true; - if (ParseAbsoluteExpression(FillExpr)) + if (parseAbsoluteExpression(FillExpr)) return true; } @@ -2369,7 +2434,7 @@ bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { Lex(); MaxBytesLoc = getLexer().getLoc(); - if (ParseAbsoluteExpression(MaxBytesToFill)) + if (parseAbsoluteExpression(MaxBytesToFill)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -2391,6 +2456,10 @@ bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { } Alignment = 1ULL << Alignment; + } else { + // Reject alignments that aren't a power of two, for gas compatibility. + if (!isPowerOf2_64(Alignment)) + Error(AlignmentLoc, "alignment must be a power of 2"); } // Diagnose non-sensical max bytes to align. @@ -2423,368 +2492,10 @@ bool AsmParser::ParseDirectiveAlign(bool IsPow2, unsigned ValueSize) { return false; } -/// ParseDirectiveSymbolAttribute -/// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] -bool AsmParser::ParseDirectiveSymbolAttribute(MCSymbolAttr Attr) { - if (getLexer().isNot(AsmToken::EndOfStatement)) { - for (;;) { - StringRef Name; - SMLoc Loc = getTok().getLoc(); - - if (ParseIdentifier(Name)) - return Error(Loc, "expected identifier in directive"); - - MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); - - // Assembler local symbols don't make any sense here. Complain loudly. - if (Sym->isTemporary()) - return Error(Loc, "non-local symbol required in directive"); - - getStreamer().EmitSymbolAttribute(Sym, Attr); - - if (getLexer().is(AsmToken::EndOfStatement)) - break; - - if (getLexer().isNot(AsmToken::Comma)) - return TokError("unexpected token in directive"); - Lex(); - } - } - - Lex(); - return false; -} - -/// ParseDirectiveComm -/// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] -bool AsmParser::ParseDirectiveComm(bool IsLocal) { - CheckForValidSection(); - - SMLoc IDLoc = getLexer().getLoc(); - StringRef Name; - if (ParseIdentifier(Name)) - return TokError("expected identifier in directive"); - - // Handle the identifier as the key symbol. - MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); - - if (getLexer().isNot(AsmToken::Comma)) - return TokError("unexpected token in directive"); - Lex(); - - int64_t Size; - SMLoc SizeLoc = getLexer().getLoc(); - if (ParseAbsoluteExpression(Size)) - return true; - - int64_t Pow2Alignment = 0; - SMLoc Pow2AlignmentLoc; - if (getLexer().is(AsmToken::Comma)) { - Lex(); - Pow2AlignmentLoc = getLexer().getLoc(); - if (ParseAbsoluteExpression(Pow2Alignment)) - return true; - - LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); - if (IsLocal && LCOMM == LCOMM::NoAlignment) - return Error(Pow2AlignmentLoc, "alignment not supported on this target"); - - // If this target takes alignments in bytes (not log) validate and convert. - if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || - (IsLocal && LCOMM == LCOMM::ByteAlignment)) { - if (!isPowerOf2_64(Pow2Alignment)) - return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); - Pow2Alignment = Log2_64(Pow2Alignment); - } - } - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.comm' or '.lcomm' directive"); - - Lex(); - - // NOTE: a size of zero for a .comm should create a undefined symbol - // but a size of .lcomm creates a bss symbol of size zero. - if (Size < 0) - return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " - "be less than zero"); - - // NOTE: The alignment in the directive is a power of 2 value, the assembler - // may internally end up wanting an alignment in bytes. - // FIXME: Diagnose overflow. - if (Pow2Alignment < 0) - return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " - "alignment, can't be less than zero"); - - if (!Sym->isUndefined()) - return Error(IDLoc, "invalid symbol redefinition"); - - // Create the Symbol as a common or local common with Size and Pow2Alignment - if (IsLocal) { - getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); - return false; - } - - getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); - return false; -} - -/// ParseDirectiveAbort -/// ::= .abort [... message ...] -bool AsmParser::ParseDirectiveAbort() { - // FIXME: Use loc from directive. - SMLoc Loc = getLexer().getLoc(); - - StringRef Str = ParseStringToEndOfStatement(); - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.abort' directive"); - - Lex(); - - if (Str.empty()) - Error(Loc, ".abort detected. Assembly stopping."); - else - Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); - // FIXME: Actually abort assembly here. - - return false; -} - -/// ParseDirectiveInclude -/// ::= .include "filename" -bool AsmParser::ParseDirectiveInclude() { - if (getLexer().isNot(AsmToken::String)) - return TokError("expected string in '.include' directive"); - - std::string Filename = getTok().getString(); - SMLoc IncludeLoc = getLexer().getLoc(); - Lex(); - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.include' directive"); - - // Strip the quotes. - Filename = Filename.substr(1, Filename.size()-2); - - // Attempt to switch the lexer to the included file before consuming the end - // of statement to avoid losing it when we switch. - if (EnterIncludeFile(Filename)) { - Error(IncludeLoc, "Could not find include file '" + Filename + "'"); - return true; - } - - return false; -} - -/// ParseDirectiveIncbin -/// ::= .incbin "filename" -bool AsmParser::ParseDirectiveIncbin() { - if (getLexer().isNot(AsmToken::String)) - return TokError("expected string in '.incbin' directive"); - - std::string Filename = getTok().getString(); - SMLoc IncbinLoc = getLexer().getLoc(); - Lex(); - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.incbin' directive"); - - // Strip the quotes. - Filename = Filename.substr(1, Filename.size()-2); - - // Attempt to process the included file. - if (ProcessIncbinFile(Filename)) { - Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); - return true; - } - - return false; -} - -/// ParseDirectiveIf -/// ::= .if expression -bool AsmParser::ParseDirectiveIf(SMLoc DirectiveLoc) { - TheCondStack.push_back(TheCondState); - TheCondState.TheCond = AsmCond::IfCond; - if (TheCondState.Ignore) { - EatToEndOfStatement(); - } else { - int64_t ExprValue; - if (ParseAbsoluteExpression(ExprValue)) - return true; - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.if' directive"); - - Lex(); - - TheCondState.CondMet = ExprValue; - TheCondState.Ignore = !TheCondState.CondMet; - } - - return false; -} - -/// ParseDirectiveIfb -/// ::= .ifb string -bool AsmParser::ParseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { - TheCondStack.push_back(TheCondState); - TheCondState.TheCond = AsmCond::IfCond; - - if (TheCondState.Ignore) { - EatToEndOfStatement(); - } else { - StringRef Str = ParseStringToEndOfStatement(); - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.ifb' directive"); - - Lex(); - - TheCondState.CondMet = ExpectBlank == Str.empty(); - TheCondState.Ignore = !TheCondState.CondMet; - } - - return false; -} - -/// ParseDirectiveIfc -/// ::= .ifc string1, string2 -bool AsmParser::ParseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { - TheCondStack.push_back(TheCondState); - TheCondState.TheCond = AsmCond::IfCond; - - if (TheCondState.Ignore) { - EatToEndOfStatement(); - } else { - StringRef Str1 = ParseStringToComma(); - - if (getLexer().isNot(AsmToken::Comma)) - return TokError("unexpected token in '.ifc' directive"); - - Lex(); - - StringRef Str2 = ParseStringToEndOfStatement(); - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.ifc' directive"); - - Lex(); - - TheCondState.CondMet = ExpectEqual == (Str1 == Str2); - TheCondState.Ignore = !TheCondState.CondMet; - } - - return false; -} - -/// ParseDirectiveIfdef -/// ::= .ifdef symbol -bool AsmParser::ParseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { - StringRef Name; - TheCondStack.push_back(TheCondState); - TheCondState.TheCond = AsmCond::IfCond; - - if (TheCondState.Ignore) { - EatToEndOfStatement(); - } else { - if (ParseIdentifier(Name)) - return TokError("expected identifier after '.ifdef'"); - - Lex(); - - MCSymbol *Sym = getContext().LookupSymbol(Name); - - if (expect_defined) - TheCondState.CondMet = (Sym != NULL && !Sym->isUndefined()); - else - TheCondState.CondMet = (Sym == NULL || Sym->isUndefined()); - TheCondState.Ignore = !TheCondState.CondMet; - } - - return false; -} - -/// ParseDirectiveElseIf -/// ::= .elseif expression -bool AsmParser::ParseDirectiveElseIf(SMLoc DirectiveLoc) { - if (TheCondState.TheCond != AsmCond::IfCond && - TheCondState.TheCond != AsmCond::ElseIfCond) - Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " - " an .elseif"); - TheCondState.TheCond = AsmCond::ElseIfCond; - - bool LastIgnoreState = false; - if (!TheCondStack.empty()) - LastIgnoreState = TheCondStack.back().Ignore; - if (LastIgnoreState || TheCondState.CondMet) { - TheCondState.Ignore = true; - EatToEndOfStatement(); - } - else { - int64_t ExprValue; - if (ParseAbsoluteExpression(ExprValue)) - return true; - - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.elseif' directive"); - - Lex(); - TheCondState.CondMet = ExprValue; - TheCondState.Ignore = !TheCondState.CondMet; - } - - return false; -} - -/// ParseDirectiveElse -/// ::= .else -bool AsmParser::ParseDirectiveElse(SMLoc DirectiveLoc) { - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.else' directive"); - - Lex(); - - if (TheCondState.TheCond != AsmCond::IfCond && - TheCondState.TheCond != AsmCond::ElseIfCond) - Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " - ".elseif"); - TheCondState.TheCond = AsmCond::ElseCond; - bool LastIgnoreState = false; - if (!TheCondStack.empty()) - LastIgnoreState = TheCondStack.back().Ignore; - if (LastIgnoreState || TheCondState.CondMet) - TheCondState.Ignore = true; - else - TheCondState.Ignore = false; - - return false; -} - -/// ParseDirectiveEndIf -/// ::= .endif -bool AsmParser::ParseDirectiveEndIf(SMLoc DirectiveLoc) { - if (getLexer().isNot(AsmToken::EndOfStatement)) - return TokError("unexpected token in '.endif' directive"); - - Lex(); - - if ((TheCondState.TheCond == AsmCond::NoCond) || - TheCondStack.empty()) - Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " - ".else"); - if (!TheCondStack.empty()) { - TheCondState = TheCondStack.back(); - TheCondStack.pop_back(); - } - - return false; -} - /// ParseDirectiveFile /// ::= .file [number] filename /// ::= .file number directory filename -bool GenericAsmParser::ParseDirectiveFile(StringRef, SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveFile(SMLoc DirectiveLoc) { // FIXME: I'm not sure what this is. int64_t FileNumber = -1; SMLoc FileNumberLoc = getLexer().getLoc(); @@ -2836,7 +2547,7 @@ bool GenericAsmParser::ParseDirectiveFile(StringRef, SMLoc DirectiveLoc) { /// ParseDirectiveLine /// ::= .line [number] -bool GenericAsmParser::ParseDirectiveLine(StringRef, SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveLine() { if (getLexer().isNot(AsmToken::EndOfStatement)) { if (getLexer().isNot(AsmToken::Integer)) return TokError("unexpected token in '.line' directive"); @@ -2854,7 +2565,6 @@ bool GenericAsmParser::ParseDirectiveLine(StringRef, SMLoc DirectiveLoc) { return false; } - /// ParseDirectiveLoc /// ::= .loc FileNumber [LineNumber] [ColumnPos] [basic_block] [prologue_end] /// [epilogue_begin] [is_stmt VALUE] [isa VALUE] @@ -2862,8 +2572,7 @@ bool GenericAsmParser::ParseDirectiveLine(StringRef, SMLoc DirectiveLoc) { /// a .file directive, the second number is the line number and optionally the /// third number is a column position (zero if not specified). The remaining /// optional items are .loc sub-directives. -bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { - +bool AsmParser::ParseDirectiveLoc() { if (getLexer().isNot(AsmToken::Integer)) return TokError("unexpected token in '.loc' directive"); int64_t FileNumber = getTok().getIntVal(); @@ -2899,7 +2608,7 @@ bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { StringRef Name; SMLoc Loc = getTok().getLoc(); - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("unexpected token in '.loc' directive"); if (Name == "basic_block") @@ -2909,9 +2618,9 @@ bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { else if (Name == "epilogue_begin") Flags |= DWARF2_FLAG_EPILOGUE_BEGIN; else if (Name == "is_stmt") { - SMLoc Loc = getTok().getLoc(); + Loc = getTok().getLoc(); const MCExpr *Value; - if (getParser().ParseExpression(Value)) + if (parseExpression(Value)) return true; // The expression must be the constant 0 or 1. if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { @@ -2928,9 +2637,9 @@ bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { } } else if (Name == "isa") { - SMLoc Loc = getTok().getLoc(); + Loc = getTok().getLoc(); const MCExpr *Value; - if (getParser().ParseExpression(Value)) + if (parseExpression(Value)) return true; // The expression must be a constant greater or equal to 0. if (const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value)) { @@ -2944,7 +2653,7 @@ bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { } } else if (Name == "discriminator") { - if (getParser().ParseAbsoluteExpression(Discriminator)) + if (parseAbsoluteExpression(Discriminator)) return true; } else { @@ -2964,20 +2673,18 @@ bool GenericAsmParser::ParseDirectiveLoc(StringRef, SMLoc DirectiveLoc) { /// ParseDirectiveStabs /// ::= .stabs string, number, number, number -bool GenericAsmParser::ParseDirectiveStabs(StringRef Directive, - SMLoc DirectiveLoc) { - return TokError("unsupported directive '" + Directive + "'"); +bool AsmParser::ParseDirectiveStabs() { + return TokError("unsupported directive '.stabs'"); } /// ParseDirectiveCFISections /// ::= .cfi_sections section [, section] -bool GenericAsmParser::ParseDirectiveCFISections(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFISections() { StringRef Name; bool EH = false; bool Debug = false; - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("Expected an identifier"); if (Name == ".eh_frame") @@ -2988,7 +2695,7 @@ bool GenericAsmParser::ParseDirectiveCFISections(StringRef, if (getLexer().is(AsmToken::Comma)) { Lex(); - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("Expected an identifier"); if (Name == ".eh_frame") @@ -2998,45 +2705,41 @@ bool GenericAsmParser::ParseDirectiveCFISections(StringRef, } getStreamer().EmitCFISections(EH, Debug); - return false; } /// ParseDirectiveCFIStartProc /// ::= .cfi_startproc -bool GenericAsmParser::ParseDirectiveCFIStartProc(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIStartProc() { getStreamer().EmitCFIStartProc(); return false; } /// ParseDirectiveCFIEndProc /// ::= .cfi_endproc -bool GenericAsmParser::ParseDirectiveCFIEndProc(StringRef, SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIEndProc() { getStreamer().EmitCFIEndProc(); return false; } /// ParseRegisterOrRegisterNumber - parse register name or number. -bool GenericAsmParser::ParseRegisterOrRegisterNumber(int64_t &Register, - SMLoc DirectiveLoc) { +bool AsmParser::ParseRegisterOrRegisterNumber(int64_t &Register, + SMLoc DirectiveLoc) { unsigned RegNo; if (getLexer().isNot(AsmToken::Integer)) { - if (getParser().getTargetParser().ParseRegister(RegNo, DirectiveLoc, - DirectiveLoc)) + if (getTargetParser().ParseRegister(RegNo, DirectiveLoc, DirectiveLoc)) return true; Register = getContext().getRegisterInfo().getDwarfRegNum(RegNo, true); } else - return getParser().ParseAbsoluteExpression(Register); + return parseAbsoluteExpression(Register); return false; } /// ParseDirectiveCFIDefCfa /// ::= .cfi_def_cfa register, offset -bool GenericAsmParser::ParseDirectiveCFIDefCfa(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIDefCfa(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) return true; @@ -3046,7 +2749,7 @@ bool GenericAsmParser::ParseDirectiveCFIDefCfa(StringRef, Lex(); int64_t Offset = 0; - if (getParser().ParseAbsoluteExpression(Offset)) + if (parseAbsoluteExpression(Offset)) return true; getStreamer().EmitCFIDefCfa(Register, Offset); @@ -3055,22 +2758,39 @@ bool GenericAsmParser::ParseDirectiveCFIDefCfa(StringRef, /// ParseDirectiveCFIDefCfaOffset /// ::= .cfi_def_cfa_offset offset -bool GenericAsmParser::ParseDirectiveCFIDefCfaOffset(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIDefCfaOffset() { int64_t Offset = 0; - if (getParser().ParseAbsoluteExpression(Offset)) + if (parseAbsoluteExpression(Offset)) return true; getStreamer().EmitCFIDefCfaOffset(Offset); return false; } +/// ParseDirectiveCFIRegister +/// ::= .cfi_register register, register +bool AsmParser::ParseDirectiveCFIRegister(SMLoc DirectiveLoc) { + int64_t Register1 = 0; + if (ParseRegisterOrRegisterNumber(Register1, DirectiveLoc)) + return true; + + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in directive"); + Lex(); + + int64_t Register2 = 0; + if (ParseRegisterOrRegisterNumber(Register2, DirectiveLoc)) + return true; + + getStreamer().EmitCFIRegister(Register1, Register2); + return false; +} + /// ParseDirectiveCFIAdjustCfaOffset /// ::= .cfi_adjust_cfa_offset adjustment -bool GenericAsmParser::ParseDirectiveCFIAdjustCfaOffset(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIAdjustCfaOffset() { int64_t Adjustment = 0; - if (getParser().ParseAbsoluteExpression(Adjustment)) + if (parseAbsoluteExpression(Adjustment)) return true; getStreamer().EmitCFIAdjustCfaOffset(Adjustment); @@ -3079,8 +2799,7 @@ bool GenericAsmParser::ParseDirectiveCFIAdjustCfaOffset(StringRef, /// ParseDirectiveCFIDefCfaRegister /// ::= .cfi_def_cfa_register register -bool GenericAsmParser::ParseDirectiveCFIDefCfaRegister(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIDefCfaRegister(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) return true; @@ -3091,7 +2810,7 @@ bool GenericAsmParser::ParseDirectiveCFIDefCfaRegister(StringRef, /// ParseDirectiveCFIOffset /// ::= .cfi_offset register, offset -bool GenericAsmParser::ParseDirectiveCFIOffset(StringRef, SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIOffset(SMLoc DirectiveLoc) { int64_t Register = 0; int64_t Offset = 0; @@ -3102,7 +2821,7 @@ bool GenericAsmParser::ParseDirectiveCFIOffset(StringRef, SMLoc DirectiveLoc) { return TokError("unexpected token in directive"); Lex(); - if (getParser().ParseAbsoluteExpression(Offset)) + if (parseAbsoluteExpression(Offset)) return true; getStreamer().EmitCFIOffset(Register, Offset); @@ -3111,8 +2830,7 @@ bool GenericAsmParser::ParseDirectiveCFIOffset(StringRef, SMLoc DirectiveLoc) { /// ParseDirectiveCFIRelOffset /// ::= .cfi_rel_offset register, offset -bool GenericAsmParser::ParseDirectiveCFIRelOffset(StringRef, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIRelOffset(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) @@ -3123,7 +2841,7 @@ bool GenericAsmParser::ParseDirectiveCFIRelOffset(StringRef, Lex(); int64_t Offset = 0; - if (getParser().ParseAbsoluteExpression(Offset)) + if (parseAbsoluteExpression(Offset)) return true; getStreamer().EmitCFIRelOffset(Register, Offset); @@ -3153,12 +2871,12 @@ static bool isValidEncoding(int64_t Encoding) { } /// ParseDirectiveCFIPersonalityOrLsda +/// IsPersonality true for cfi_personality, false for cfi_lsda /// ::= .cfi_personality encoding, [symbol_name] /// ::= .cfi_lsda encoding, [symbol_name] -bool GenericAsmParser::ParseDirectiveCFIPersonalityOrLsda(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIPersonalityOrLsda(bool IsPersonality) { int64_t Encoding = 0; - if (getParser().ParseAbsoluteExpression(Encoding)) + if (parseAbsoluteExpression(Encoding)) return true; if (Encoding == dwarf::DW_EH_PE_omit) return false; @@ -3171,70 +2889,61 @@ bool GenericAsmParser::ParseDirectiveCFIPersonalityOrLsda(StringRef IDVal, Lex(); StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("expected identifier in directive"); MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); - if (IDVal == ".cfi_personality") + if (IsPersonality) getStreamer().EmitCFIPersonality(Sym, Encoding); - else { - assert(IDVal == ".cfi_lsda"); + else getStreamer().EmitCFILsda(Sym, Encoding); - } return false; } /// ParseDirectiveCFIRememberState /// ::= .cfi_remember_state -bool GenericAsmParser::ParseDirectiveCFIRememberState(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIRememberState() { getStreamer().EmitCFIRememberState(); return false; } /// ParseDirectiveCFIRestoreState /// ::= .cfi_remember_state -bool GenericAsmParser::ParseDirectiveCFIRestoreState(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIRestoreState() { getStreamer().EmitCFIRestoreState(); return false; } /// ParseDirectiveCFISameValue /// ::= .cfi_same_value register -bool GenericAsmParser::ParseDirectiveCFISameValue(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFISameValue(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) return true; getStreamer().EmitCFISameValue(Register); - return false; } /// ParseDirectiveCFIRestore /// ::= .cfi_restore register -bool GenericAsmParser::ParseDirectiveCFIRestore(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIRestore(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) return true; getStreamer().EmitCFIRestore(Register); - return false; } /// ParseDirectiveCFIEscape /// ::= .cfi_escape expression[,...] -bool GenericAsmParser::ParseDirectiveCFIEscape(StringRef IDVal, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIEscape() { std::string Values; int64_t CurrValue; - if (getParser().ParseAbsoluteExpression(CurrValue)) + if (parseAbsoluteExpression(CurrValue)) return true; Values.push_back((uint8_t)CurrValue); @@ -3242,7 +2951,7 @@ bool GenericAsmParser::ParseDirectiveCFIEscape(StringRef IDVal, while (getLexer().is(AsmToken::Comma)) { Lex(); - if (getParser().ParseAbsoluteExpression(CurrValue)) + if (parseAbsoluteExpression(CurrValue)) return true; Values.push_back((uint8_t)CurrValue); @@ -3254,89 +2963,59 @@ bool GenericAsmParser::ParseDirectiveCFIEscape(StringRef IDVal, /// ParseDirectiveCFISignalFrame /// ::= .cfi_signal_frame -bool GenericAsmParser::ParseDirectiveCFISignalFrame(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFISignalFrame() { if (getLexer().isNot(AsmToken::EndOfStatement)) return Error(getLexer().getLoc(), - "unexpected token in '" + Directive + "' directive"); + "unexpected token in '.cfi_signal_frame'"); getStreamer().EmitCFISignalFrame(); - return false; } /// ParseDirectiveCFIUndefined /// ::= .cfi_undefined register -bool GenericAsmParser::ParseDirectiveCFIUndefined(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveCFIUndefined(SMLoc DirectiveLoc) { int64_t Register = 0; if (ParseRegisterOrRegisterNumber(Register, DirectiveLoc)) return true; getStreamer().EmitCFIUndefined(Register); - - return false; -} - -/// ParseDirectiveCFIRegister -/// ::= .cfi_register register, register -bool GenericAsmParser::ParseDirectiveCFIRegister(StringRef Directive, - SMLoc DirectiveLoc) { - int64_t Register1 = 0; - - if (ParseRegisterOrRegisterNumber(Register1, DirectiveLoc)) - return true; - - if (getLexer().isNot(AsmToken::Comma)) - return TokError("unexpected token in directive"); - Lex(); - - int64_t Register2 = 0; - - if (ParseRegisterOrRegisterNumber(Register2, DirectiveLoc)) - return true; - - getStreamer().EmitCFIRegister(Register1, Register2); - return false; } /// ParseDirectiveMacrosOnOff /// ::= .macros_on /// ::= .macros_off -bool GenericAsmParser::ParseDirectiveMacrosOnOff(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveMacrosOnOff(StringRef Directive) { if (getLexer().isNot(AsmToken::EndOfStatement)) return Error(getLexer().getLoc(), "unexpected token in '" + Directive + "' directive"); - getParser().MacrosEnabled = Directive == ".macros_on"; - + SetMacrosEnabled(Directive == ".macros_on"); return false; } /// ParseDirectiveMacro /// ::= .macro name [parameters] -bool GenericAsmParser::ParseDirectiveMacro(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveMacro(SMLoc DirectiveLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("expected identifier in '.macro' directive"); - MacroParameters Parameters; + MCAsmMacroParameters Parameters; // Argument delimiter is initially unknown. It will be set by // ParseMacroArgument() AsmToken::TokenKind ArgumentDelimiter = AsmToken::Eof; if (getLexer().isNot(AsmToken::EndOfStatement)) { for (;;) { - MacroParameter Parameter; - if (getParser().ParseIdentifier(Parameter.first)) + MCAsmMacroParameter Parameter; + if (parseIdentifier(Parameter.first)) return TokError("expected identifier in '.macro' directive"); if (getLexer().is(AsmToken::Equal)) { Lex(); - if (getParser().ParseMacroArgument(Parameter.second, ArgumentDelimiter)) + if (ParseMacroArgument(Parameter.second, ArgumentDelimiter)) return true; } @@ -3373,32 +3052,134 @@ bool GenericAsmParser::ParseDirectiveMacro(StringRef Directive, } // Otherwise, scan til the end of the statement. - getParser().EatToEndOfStatement(); + eatToEndOfStatement(); } - if (getParser().MacroMap.lookup(Name)) { + if (LookupMacro(Name)) { return Error(DirectiveLoc, "macro '" + Name + "' is already defined"); } const char *BodyStart = StartToken.getLoc().getPointer(); const char *BodyEnd = EndToken.getLoc().getPointer(); StringRef Body = StringRef(BodyStart, BodyEnd - BodyStart); - getParser().MacroMap[Name] = new Macro(Name, Body, Parameters); + CheckForBadMacro(DirectiveLoc, Name, Body, Parameters); + DefineMacro(Name, MCAsmMacro(Name, Body, Parameters)); return false; } +/// CheckForBadMacro +/// +/// With the support added for named parameters there may be code out there that +/// is transitioning from positional parameters. In versions of gas that did +/// not support named parameters they would be ignored on the macro defintion. +/// But to support both styles of parameters this is not possible so if a macro +/// defintion has named parameters but does not use them and has what appears +/// to be positional parameters, strings like $1, $2, ... and $n, then issue a +/// warning that the positional parameter found in body which have no effect. +/// Hoping the developer will either remove the named parameters from the macro +/// definiton so the positional parameters get used if that was what was +/// intended or change the macro to use the named parameters. It is possible +/// this warning will trigger when the none of the named parameters are used +/// and the strings like $1 are infact to simply to be passed trough unchanged. +void AsmParser::CheckForBadMacro(SMLoc DirectiveLoc, StringRef Name, + StringRef Body, + MCAsmMacroParameters Parameters) { + // If this macro is not defined with named parameters the warning we are + // checking for here doesn't apply. + unsigned NParameters = Parameters.size(); + if (NParameters == 0) + return; + + bool NamedParametersFound = false; + bool PositionalParametersFound = false; + + // Look at the body of the macro for use of both the named parameters and what + // are likely to be positional parameters. This is what expandMacro() is + // doing when it finds the parameters in the body. + while (!Body.empty()) { + // Scan for the next possible parameter. + std::size_t End = Body.size(), Pos = 0; + for (; Pos != End; ++Pos) { + // Check for a substitution or escape. + // This macro is defined with parameters, look for \foo, \bar, etc. + if (Body[Pos] == '\\' && Pos + 1 != End) + break; + + // This macro should have parameters, but look for $0, $1, ..., $n too. + if (Body[Pos] != '$' || Pos + 1 == End) + continue; + char Next = Body[Pos + 1]; + if (Next == '$' || Next == 'n' || + isdigit(static_cast<unsigned char>(Next))) + break; + } + + // Check if we reached the end. + if (Pos == End) + break; + + if (Body[Pos] == '$') { + switch (Body[Pos+1]) { + // $$ => $ + case '$': + break; + + // $n => number of arguments + case 'n': + PositionalParametersFound = true; + break; + + // $[0-9] => argument + default: { + PositionalParametersFound = true; + break; + } + } + Pos += 2; + } else { + unsigned I = Pos + 1; + while (isIdentifierChar(Body[I]) && I + 1 != End) + ++I; + + const char *Begin = Body.data() + Pos +1; + StringRef Argument(Begin, I - (Pos +1)); + unsigned Index = 0; + for (; Index < NParameters; ++Index) + if (Parameters[Index].first == Argument) + break; + + if (Index == NParameters) { + if (Body[Pos+1] == '(' && Body[Pos+2] == ')') + Pos += 3; + else { + Pos = I; + } + } else { + NamedParametersFound = true; + Pos += 1 + Argument.size(); + } + } + // Update the scan point. + Body = Body.substr(Pos); + } + + if (!NamedParametersFound && PositionalParametersFound) + Warning(DirectiveLoc, "macro defined with named parameters which are not " + "used in macro body, possible positional parameter " + "found in body which will have no effect"); +} + /// ParseDirectiveEndMacro /// ::= .endm /// ::= .endmacro -bool GenericAsmParser::ParseDirectiveEndMacro(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectiveEndMacro(StringRef Directive) { if (getLexer().isNot(AsmToken::EndOfStatement)) return TokError("unexpected token in '" + Directive + "' directive"); // If we are inside a macro instantiation, terminate the current // instantiation. - if (!getParser().ActiveMacros.empty()) { - getParser().HandleMacroExit(); + if (InsideMacroInstantiation()) { + HandleMacroExit(); return false; } @@ -3410,37 +3191,136 @@ bool GenericAsmParser::ParseDirectiveEndMacro(StringRef Directive, /// ParseDirectivePurgeMacro /// ::= .purgem -bool GenericAsmParser::ParseDirectivePurgeMacro(StringRef Directive, - SMLoc DirectiveLoc) { +bool AsmParser::ParseDirectivePurgeMacro(SMLoc DirectiveLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (parseIdentifier(Name)) return TokError("expected identifier in '.purgem' directive"); if (getLexer().isNot(AsmToken::EndOfStatement)) return TokError("unexpected token in '.purgem' directive"); - StringMap<Macro*>::iterator I = getParser().MacroMap.find(Name); - if (I == getParser().MacroMap.end()) + if (!LookupMacro(Name)) return Error(DirectiveLoc, "macro '" + Name + "' is not defined"); - // Undefine the macro. - delete I->getValue(); - getParser().MacroMap.erase(I); + UndefineMacro(Name); return false; } -bool GenericAsmParser::ParseDirectiveLEB128(StringRef DirName, SMLoc) { - getParser().CheckForValidSection(); +/// ParseDirectiveBundleAlignMode +/// ::= {.bundle_align_mode} expression +bool AsmParser::ParseDirectiveBundleAlignMode() { + checkForValidSection(); + + // Expect a single argument: an expression that evaluates to a constant + // in the inclusive range 0-30. + SMLoc ExprLoc = getLexer().getLoc(); + int64_t AlignSizePow2; + if (parseAbsoluteExpression(AlignSizePow2)) + return true; + else if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token after expression in" + " '.bundle_align_mode' directive"); + else if (AlignSizePow2 < 0 || AlignSizePow2 > 30) + return Error(ExprLoc, + "invalid bundle alignment size (expected between 0 and 30)"); + Lex(); + + // Because of AlignSizePow2's verified range we can safely truncate it to + // unsigned. + getStreamer().EmitBundleAlignMode(static_cast<unsigned>(AlignSizePow2)); + return false; +} + +/// ParseDirectiveBundleLock +/// ::= {.bundle_lock} [align_to_end] +bool AsmParser::ParseDirectiveBundleLock() { + checkForValidSection(); + bool AlignToEnd = false; + + if (getLexer().isNot(AsmToken::EndOfStatement)) { + StringRef Option; + SMLoc Loc = getTok().getLoc(); + const char *kInvalidOptionError = + "invalid option for '.bundle_lock' directive"; + + if (parseIdentifier(Option)) + return Error(Loc, kInvalidOptionError); + + if (Option != "align_to_end") + return Error(Loc, kInvalidOptionError); + else if (getLexer().isNot(AsmToken::EndOfStatement)) + return Error(Loc, + "unexpected token after '.bundle_lock' directive option"); + AlignToEnd = true; + } + + Lex(); + + getStreamer().EmitBundleLock(AlignToEnd); + return false; +} + +/// ParseDirectiveBundleLock +/// ::= {.bundle_lock} +bool AsmParser::ParseDirectiveBundleUnlock() { + checkForValidSection(); + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.bundle_unlock' directive"); + Lex(); + + getStreamer().EmitBundleUnlock(); + return false; +} + +/// ParseDirectiveSpace +/// ::= (.skip | .space) expression [ , expression ] +bool AsmParser::ParseDirectiveSpace(StringRef IDVal) { + checkForValidSection(); + + int64_t NumBytes; + if (parseAbsoluteExpression(NumBytes)) + return true; + + int64_t FillExpr = 0; + if (getLexer().isNot(AsmToken::EndOfStatement)) { + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); + Lex(); + + if (parseAbsoluteExpression(FillExpr)) + return true; + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); + } + + Lex(); + + if (NumBytes <= 0) + return TokError("invalid number of bytes in '" + + Twine(IDVal) + "' directive"); + + // FIXME: Sometimes the fill expr is 'nop' if it isn't supplied, instead of 0. + getStreamer().EmitFill(NumBytes, FillExpr, DEFAULT_ADDRSPACE); + + return false; +} + +/// ParseDirectiveLEB128 +/// ::= (.sleb128 | .uleb128) expression +bool AsmParser::ParseDirectiveLEB128(bool Signed) { + checkForValidSection(); const MCExpr *Value; - if (getParser().ParseExpression(Value)) + if (parseExpression(Value)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) return TokError("unexpected token in directive"); - if (DirName[1] == 's') + if (Signed) getStreamer().EmitSLEB128Value(Value); else getStreamer().EmitULEB128Value(Value); @@ -3448,7 +3328,469 @@ bool GenericAsmParser::ParseDirectiveLEB128(StringRef DirName, SMLoc) { return false; } -Macro *AsmParser::ParseMacroLikeBody(SMLoc DirectiveLoc) { +/// ParseDirectiveSymbolAttribute +/// ::= { ".globl", ".weak", ... } [ identifier ( , identifier )* ] +bool AsmParser::ParseDirectiveSymbolAttribute(MCSymbolAttr Attr) { + if (getLexer().isNot(AsmToken::EndOfStatement)) { + for (;;) { + StringRef Name; + SMLoc Loc = getTok().getLoc(); + + if (parseIdentifier(Name)) + return Error(Loc, "expected identifier in directive"); + + MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); + + // Assembler local symbols don't make any sense here. Complain loudly. + if (Sym->isTemporary()) + return Error(Loc, "non-local symbol required in directive"); + + getStreamer().EmitSymbolAttribute(Sym, Attr); + + if (getLexer().is(AsmToken::EndOfStatement)) + break; + + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in directive"); + Lex(); + } + } + + Lex(); + return false; +} + +/// ParseDirectiveComm +/// ::= ( .comm | .lcomm ) identifier , size_expression [ , align_expression ] +bool AsmParser::ParseDirectiveComm(bool IsLocal) { + checkForValidSection(); + + SMLoc IDLoc = getLexer().getLoc(); + StringRef Name; + if (parseIdentifier(Name)) + return TokError("expected identifier in directive"); + + // Handle the identifier as the key symbol. + MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); + + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in directive"); + Lex(); + + int64_t Size; + SMLoc SizeLoc = getLexer().getLoc(); + if (parseAbsoluteExpression(Size)) + return true; + + int64_t Pow2Alignment = 0; + SMLoc Pow2AlignmentLoc; + if (getLexer().is(AsmToken::Comma)) { + Lex(); + Pow2AlignmentLoc = getLexer().getLoc(); + if (parseAbsoluteExpression(Pow2Alignment)) + return true; + + LCOMM::LCOMMType LCOMM = Lexer.getMAI().getLCOMMDirectiveAlignmentType(); + if (IsLocal && LCOMM == LCOMM::NoAlignment) + return Error(Pow2AlignmentLoc, "alignment not supported on this target"); + + // If this target takes alignments in bytes (not log) validate and convert. + if ((!IsLocal && Lexer.getMAI().getCOMMDirectiveAlignmentIsInBytes()) || + (IsLocal && LCOMM == LCOMM::ByteAlignment)) { + if (!isPowerOf2_64(Pow2Alignment)) + return Error(Pow2AlignmentLoc, "alignment must be a power of 2"); + Pow2Alignment = Log2_64(Pow2Alignment); + } + } + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.comm' or '.lcomm' directive"); + + Lex(); + + // NOTE: a size of zero for a .comm should create a undefined symbol + // but a size of .lcomm creates a bss symbol of size zero. + if (Size < 0) + return Error(SizeLoc, "invalid '.comm' or '.lcomm' directive size, can't " + "be less than zero"); + + // NOTE: The alignment in the directive is a power of 2 value, the assembler + // may internally end up wanting an alignment in bytes. + // FIXME: Diagnose overflow. + if (Pow2Alignment < 0) + return Error(Pow2AlignmentLoc, "invalid '.comm' or '.lcomm' directive " + "alignment, can't be less than zero"); + + if (!Sym->isUndefined()) + return Error(IDLoc, "invalid symbol redefinition"); + + // Create the Symbol as a common or local common with Size and Pow2Alignment + if (IsLocal) { + getStreamer().EmitLocalCommonSymbol(Sym, Size, 1 << Pow2Alignment); + return false; + } + + getStreamer().EmitCommonSymbol(Sym, Size, 1 << Pow2Alignment); + return false; +} + +/// ParseDirectiveAbort +/// ::= .abort [... message ...] +bool AsmParser::ParseDirectiveAbort() { + // FIXME: Use loc from directive. + SMLoc Loc = getLexer().getLoc(); + + StringRef Str = parseStringToEndOfStatement(); + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.abort' directive"); + + Lex(); + + if (Str.empty()) + Error(Loc, ".abort detected. Assembly stopping."); + else + Error(Loc, ".abort '" + Str + "' detected. Assembly stopping."); + // FIXME: Actually abort assembly here. + + return false; +} + +/// ParseDirectiveInclude +/// ::= .include "filename" +bool AsmParser::ParseDirectiveInclude() { + if (getLexer().isNot(AsmToken::String)) + return TokError("expected string in '.include' directive"); + + std::string Filename = getTok().getString(); + SMLoc IncludeLoc = getLexer().getLoc(); + Lex(); + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.include' directive"); + + // Strip the quotes. + Filename = Filename.substr(1, Filename.size()-2); + + // Attempt to switch the lexer to the included file before consuming the end + // of statement to avoid losing it when we switch. + if (EnterIncludeFile(Filename)) { + Error(IncludeLoc, "Could not find include file '" + Filename + "'"); + return true; + } + + return false; +} + +/// ParseDirectiveIncbin +/// ::= .incbin "filename" +bool AsmParser::ParseDirectiveIncbin() { + if (getLexer().isNot(AsmToken::String)) + return TokError("expected string in '.incbin' directive"); + + std::string Filename = getTok().getString(); + SMLoc IncbinLoc = getLexer().getLoc(); + Lex(); + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.incbin' directive"); + + // Strip the quotes. + Filename = Filename.substr(1, Filename.size()-2); + + // Attempt to process the included file. + if (ProcessIncbinFile(Filename)) { + Error(IncbinLoc, "Could not find incbin file '" + Filename + "'"); + return true; + } + + return false; +} + +/// ParseDirectiveIf +/// ::= .if expression +bool AsmParser::ParseDirectiveIf(SMLoc DirectiveLoc) { + TheCondStack.push_back(TheCondState); + TheCondState.TheCond = AsmCond::IfCond; + if (TheCondState.Ignore) { + eatToEndOfStatement(); + } else { + int64_t ExprValue; + if (parseAbsoluteExpression(ExprValue)) + return true; + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.if' directive"); + + Lex(); + + TheCondState.CondMet = ExprValue; + TheCondState.Ignore = !TheCondState.CondMet; + } + + return false; +} + +/// ParseDirectiveIfb +/// ::= .ifb string +bool AsmParser::ParseDirectiveIfb(SMLoc DirectiveLoc, bool ExpectBlank) { + TheCondStack.push_back(TheCondState); + TheCondState.TheCond = AsmCond::IfCond; + + if (TheCondState.Ignore) { + eatToEndOfStatement(); + } else { + StringRef Str = parseStringToEndOfStatement(); + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.ifb' directive"); + + Lex(); + + TheCondState.CondMet = ExpectBlank == Str.empty(); + TheCondState.Ignore = !TheCondState.CondMet; + } + + return false; +} + +/// ParseDirectiveIfc +/// ::= .ifc string1, string2 +bool AsmParser::ParseDirectiveIfc(SMLoc DirectiveLoc, bool ExpectEqual) { + TheCondStack.push_back(TheCondState); + TheCondState.TheCond = AsmCond::IfCond; + + if (TheCondState.Ignore) { + eatToEndOfStatement(); + } else { + StringRef Str1 = ParseStringToComma(); + + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in '.ifc' directive"); + + Lex(); + + StringRef Str2 = parseStringToEndOfStatement(); + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.ifc' directive"); + + Lex(); + + TheCondState.CondMet = ExpectEqual == (Str1 == Str2); + TheCondState.Ignore = !TheCondState.CondMet; + } + + return false; +} + +/// ParseDirectiveIfdef +/// ::= .ifdef symbol +bool AsmParser::ParseDirectiveIfdef(SMLoc DirectiveLoc, bool expect_defined) { + StringRef Name; + TheCondStack.push_back(TheCondState); + TheCondState.TheCond = AsmCond::IfCond; + + if (TheCondState.Ignore) { + eatToEndOfStatement(); + } else { + if (parseIdentifier(Name)) + return TokError("expected identifier after '.ifdef'"); + + Lex(); + + MCSymbol *Sym = getContext().LookupSymbol(Name); + + if (expect_defined) + TheCondState.CondMet = (Sym != NULL && !Sym->isUndefined()); + else + TheCondState.CondMet = (Sym == NULL || Sym->isUndefined()); + TheCondState.Ignore = !TheCondState.CondMet; + } + + return false; +} + +/// ParseDirectiveElseIf +/// ::= .elseif expression +bool AsmParser::ParseDirectiveElseIf(SMLoc DirectiveLoc) { + if (TheCondState.TheCond != AsmCond::IfCond && + TheCondState.TheCond != AsmCond::ElseIfCond) + Error(DirectiveLoc, "Encountered a .elseif that doesn't follow a .if or " + " an .elseif"); + TheCondState.TheCond = AsmCond::ElseIfCond; + + bool LastIgnoreState = false; + if (!TheCondStack.empty()) + LastIgnoreState = TheCondStack.back().Ignore; + if (LastIgnoreState || TheCondState.CondMet) { + TheCondState.Ignore = true; + eatToEndOfStatement(); + } + else { + int64_t ExprValue; + if (parseAbsoluteExpression(ExprValue)) + return true; + + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.elseif' directive"); + + Lex(); + TheCondState.CondMet = ExprValue; + TheCondState.Ignore = !TheCondState.CondMet; + } + + return false; +} + +/// ParseDirectiveElse +/// ::= .else +bool AsmParser::ParseDirectiveElse(SMLoc DirectiveLoc) { + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.else' directive"); + + Lex(); + + if (TheCondState.TheCond != AsmCond::IfCond && + TheCondState.TheCond != AsmCond::ElseIfCond) + Error(DirectiveLoc, "Encountered a .else that doesn't follow a .if or an " + ".elseif"); + TheCondState.TheCond = AsmCond::ElseCond; + bool LastIgnoreState = false; + if (!TheCondStack.empty()) + LastIgnoreState = TheCondStack.back().Ignore; + if (LastIgnoreState || TheCondState.CondMet) + TheCondState.Ignore = true; + else + TheCondState.Ignore = false; + + return false; +} + +/// ParseDirectiveEndIf +/// ::= .endif +bool AsmParser::ParseDirectiveEndIf(SMLoc DirectiveLoc) { + if (getLexer().isNot(AsmToken::EndOfStatement)) + return TokError("unexpected token in '.endif' directive"); + + Lex(); + + if ((TheCondState.TheCond == AsmCond::NoCond) || + TheCondStack.empty()) + Error(DirectiveLoc, "Encountered a .endif that doesn't follow a .if or " + ".else"); + if (!TheCondStack.empty()) { + TheCondState = TheCondStack.back(); + TheCondStack.pop_back(); + } + + return false; +} + +void AsmParser::initializeDirectiveKindMap() { + DirectiveKindMap[".set"] = DK_SET; + DirectiveKindMap[".equ"] = DK_EQU; + DirectiveKindMap[".equiv"] = DK_EQUIV; + DirectiveKindMap[".ascii"] = DK_ASCII; + DirectiveKindMap[".asciz"] = DK_ASCIZ; + DirectiveKindMap[".string"] = DK_STRING; + DirectiveKindMap[".byte"] = DK_BYTE; + DirectiveKindMap[".short"] = DK_SHORT; + DirectiveKindMap[".value"] = DK_VALUE; + DirectiveKindMap[".2byte"] = DK_2BYTE; + DirectiveKindMap[".long"] = DK_LONG; + DirectiveKindMap[".int"] = DK_INT; + DirectiveKindMap[".4byte"] = DK_4BYTE; + DirectiveKindMap[".quad"] = DK_QUAD; + DirectiveKindMap[".8byte"] = DK_8BYTE; + DirectiveKindMap[".single"] = DK_SINGLE; + DirectiveKindMap[".float"] = DK_FLOAT; + DirectiveKindMap[".double"] = DK_DOUBLE; + DirectiveKindMap[".align"] = DK_ALIGN; + DirectiveKindMap[".align32"] = DK_ALIGN32; + DirectiveKindMap[".balign"] = DK_BALIGN; + DirectiveKindMap[".balignw"] = DK_BALIGNW; + DirectiveKindMap[".balignl"] = DK_BALIGNL; + DirectiveKindMap[".p2align"] = DK_P2ALIGN; + DirectiveKindMap[".p2alignw"] = DK_P2ALIGNW; + DirectiveKindMap[".p2alignl"] = DK_P2ALIGNL; + DirectiveKindMap[".org"] = DK_ORG; + DirectiveKindMap[".fill"] = DK_FILL; + DirectiveKindMap[".zero"] = DK_ZERO; + DirectiveKindMap[".extern"] = DK_EXTERN; + DirectiveKindMap[".globl"] = DK_GLOBL; + DirectiveKindMap[".global"] = DK_GLOBAL; + DirectiveKindMap[".indirect_symbol"] = DK_INDIRECT_SYMBOL; + DirectiveKindMap[".lazy_reference"] = DK_LAZY_REFERENCE; + DirectiveKindMap[".no_dead_strip"] = DK_NO_DEAD_STRIP; + DirectiveKindMap[".symbol_resolver"] = DK_SYMBOL_RESOLVER; + DirectiveKindMap[".private_extern"] = DK_PRIVATE_EXTERN; + DirectiveKindMap[".reference"] = DK_REFERENCE; + DirectiveKindMap[".weak_definition"] = DK_WEAK_DEFINITION; + DirectiveKindMap[".weak_reference"] = DK_WEAK_REFERENCE; + DirectiveKindMap[".weak_def_can_be_hidden"] = DK_WEAK_DEF_CAN_BE_HIDDEN; + DirectiveKindMap[".comm"] = DK_COMM; + DirectiveKindMap[".common"] = DK_COMMON; + DirectiveKindMap[".lcomm"] = DK_LCOMM; + DirectiveKindMap[".abort"] = DK_ABORT; + DirectiveKindMap[".include"] = DK_INCLUDE; + DirectiveKindMap[".incbin"] = DK_INCBIN; + DirectiveKindMap[".code16"] = DK_CODE16; + DirectiveKindMap[".code16gcc"] = DK_CODE16GCC; + DirectiveKindMap[".rept"] = DK_REPT; + DirectiveKindMap[".irp"] = DK_IRP; + DirectiveKindMap[".irpc"] = DK_IRPC; + DirectiveKindMap[".endr"] = DK_ENDR; + DirectiveKindMap[".bundle_align_mode"] = DK_BUNDLE_ALIGN_MODE; + DirectiveKindMap[".bundle_lock"] = DK_BUNDLE_LOCK; + DirectiveKindMap[".bundle_unlock"] = DK_BUNDLE_UNLOCK; + DirectiveKindMap[".if"] = DK_IF; + DirectiveKindMap[".ifb"] = DK_IFB; + DirectiveKindMap[".ifnb"] = DK_IFNB; + DirectiveKindMap[".ifc"] = DK_IFC; + DirectiveKindMap[".ifnc"] = DK_IFNC; + DirectiveKindMap[".ifdef"] = DK_IFDEF; + DirectiveKindMap[".ifndef"] = DK_IFNDEF; + DirectiveKindMap[".ifnotdef"] = DK_IFNOTDEF; + DirectiveKindMap[".elseif"] = DK_ELSEIF; + DirectiveKindMap[".else"] = DK_ELSE; + DirectiveKindMap[".endif"] = DK_ENDIF; + DirectiveKindMap[".skip"] = DK_SKIP; + DirectiveKindMap[".space"] = DK_SPACE; + DirectiveKindMap[".file"] = DK_FILE; + DirectiveKindMap[".line"] = DK_LINE; + DirectiveKindMap[".loc"] = DK_LOC; + DirectiveKindMap[".stabs"] = DK_STABS; + DirectiveKindMap[".sleb128"] = DK_SLEB128; + DirectiveKindMap[".uleb128"] = DK_ULEB128; + DirectiveKindMap[".cfi_sections"] = DK_CFI_SECTIONS; + DirectiveKindMap[".cfi_startproc"] = DK_CFI_STARTPROC; + DirectiveKindMap[".cfi_endproc"] = DK_CFI_ENDPROC; + DirectiveKindMap[".cfi_def_cfa"] = DK_CFI_DEF_CFA; + DirectiveKindMap[".cfi_def_cfa_offset"] = DK_CFI_DEF_CFA_OFFSET; + DirectiveKindMap[".cfi_adjust_cfa_offset"] = DK_CFI_ADJUST_CFA_OFFSET; + DirectiveKindMap[".cfi_def_cfa_register"] = DK_CFI_DEF_CFA_REGISTER; + DirectiveKindMap[".cfi_offset"] = DK_CFI_OFFSET; + DirectiveKindMap[".cfi_rel_offset"] = DK_CFI_REL_OFFSET; + DirectiveKindMap[".cfi_personality"] = DK_CFI_PERSONALITY; + DirectiveKindMap[".cfi_lsda"] = DK_CFI_LSDA; + DirectiveKindMap[".cfi_remember_state"] = DK_CFI_REMEMBER_STATE; + DirectiveKindMap[".cfi_restore_state"] = DK_CFI_RESTORE_STATE; + DirectiveKindMap[".cfi_same_value"] = DK_CFI_SAME_VALUE; + DirectiveKindMap[".cfi_restore"] = DK_CFI_RESTORE; + DirectiveKindMap[".cfi_escape"] = DK_CFI_ESCAPE; + DirectiveKindMap[".cfi_signal_frame"] = DK_CFI_SIGNAL_FRAME; + DirectiveKindMap[".cfi_undefined"] = DK_CFI_UNDEFINED; + DirectiveKindMap[".cfi_register"] = DK_CFI_REGISTER; + DirectiveKindMap[".macros_on"] = DK_MACROS_ON; + DirectiveKindMap[".macros_off"] = DK_MACROS_OFF; + DirectiveKindMap[".macro"] = DK_MACRO; + DirectiveKindMap[".endm"] = DK_ENDM; + DirectiveKindMap[".endmacro"] = DK_ENDMACRO; + DirectiveKindMap[".purgem"] = DK_PURGEM; +} + + +MCAsmMacro *AsmParser::ParseMacroLikeBody(SMLoc DirectiveLoc) { AsmToken EndToken, StartToken = getTok(); unsigned NestLevel = 0; @@ -3480,7 +3822,7 @@ Macro *AsmParser::ParseMacroLikeBody(SMLoc DirectiveLoc) { } // Otherwise, scan till the end of the statement. - EatToEndOfStatement(); + eatToEndOfStatement(); } const char *BodyStart = StartToken.getLoc().getPointer(); @@ -3489,11 +3831,11 @@ Macro *AsmParser::ParseMacroLikeBody(SMLoc DirectiveLoc) { // We Are Anonymous. StringRef Name; - MacroParameters Parameters; - return new Macro(Name, Body, Parameters); + MCAsmMacroParameters Parameters; + return new MCAsmMacro(Name, Body, Parameters); } -void AsmParser::InstantiateMacroLikeBody(Macro *M, SMLoc DirectiveLoc, +void AsmParser::InstantiateMacroLikeBody(MCAsmMacro *M, SMLoc DirectiveLoc, raw_svector_ostream &OS) { OS << ".endr\n"; @@ -3516,7 +3858,7 @@ void AsmParser::InstantiateMacroLikeBody(Macro *M, SMLoc DirectiveLoc, bool AsmParser::ParseDirectiveRept(SMLoc DirectiveLoc) { int64_t Count; - if (ParseAbsoluteExpression(Count)) + if (parseAbsoluteExpression(Count)) return TokError("unexpected token in '.rept' directive"); if (Count < 0) @@ -3529,15 +3871,15 @@ bool AsmParser::ParseDirectiveRept(SMLoc DirectiveLoc) { Lex(); // Lex the rept definition. - Macro *M = ParseMacroLikeBody(DirectiveLoc); + MCAsmMacro *M = ParseMacroLikeBody(DirectiveLoc); if (!M) return true; // Macro instantiation is lexical, unfortunately. We construct a new buffer // to hold the macro body with substitutions. SmallString<256> Buf; - MacroParameters Parameters; - MacroArguments A; + MCAsmMacroParameters Parameters; + MCAsmMacroArguments A; raw_svector_ostream OS(Buf); while (Count--) { if (expandMacro(OS, M->Body, Parameters, A, getTok().getLoc())) @@ -3551,10 +3893,10 @@ bool AsmParser::ParseDirectiveRept(SMLoc DirectiveLoc) { /// ParseDirectiveIrp /// ::= .irp symbol,values bool AsmParser::ParseDirectiveIrp(SMLoc DirectiveLoc) { - MacroParameters Parameters; - MacroParameter Parameter; + MCAsmMacroParameters Parameters; + MCAsmMacroParameter Parameter; - if (ParseIdentifier(Parameter.first)) + if (parseIdentifier(Parameter.first)) return TokError("expected identifier in '.irp' directive"); Parameters.push_back(Parameter); @@ -3564,7 +3906,7 @@ bool AsmParser::ParseDirectiveIrp(SMLoc DirectiveLoc) { Lex(); - MacroArguments A; + MCAsmMacroArguments A; if (ParseMacroArguments(0, A)) return true; @@ -3572,7 +3914,7 @@ bool AsmParser::ParseDirectiveIrp(SMLoc DirectiveLoc) { Lex(); // Lex the irp definition. - Macro *M = ParseMacroLikeBody(DirectiveLoc); + MCAsmMacro *M = ParseMacroLikeBody(DirectiveLoc); if (!M) return true; @@ -3581,8 +3923,8 @@ bool AsmParser::ParseDirectiveIrp(SMLoc DirectiveLoc) { SmallString<256> Buf; raw_svector_ostream OS(Buf); - for (MacroArguments::iterator i = A.begin(), e = A.end(); i != e; ++i) { - MacroArguments Args; + for (MCAsmMacroArguments::iterator i = A.begin(), e = A.end(); i != e; ++i) { + MCAsmMacroArguments Args; Args.push_back(*i); if (expandMacro(OS, M->Body, Parameters, Args, getTok().getLoc())) @@ -3597,10 +3939,10 @@ bool AsmParser::ParseDirectiveIrp(SMLoc DirectiveLoc) { /// ParseDirectiveIrpc /// ::= .irpc symbol,values bool AsmParser::ParseDirectiveIrpc(SMLoc DirectiveLoc) { - MacroParameters Parameters; - MacroParameter Parameter; + MCAsmMacroParameters Parameters; + MCAsmMacroParameter Parameter; - if (ParseIdentifier(Parameter.first)) + if (parseIdentifier(Parameter.first)) return TokError("expected identifier in '.irpc' directive"); Parameters.push_back(Parameter); @@ -3610,7 +3952,7 @@ bool AsmParser::ParseDirectiveIrpc(SMLoc DirectiveLoc) { Lex(); - MacroArguments A; + MCAsmMacroArguments A; if (ParseMacroArguments(0, A)) return true; @@ -3621,7 +3963,7 @@ bool AsmParser::ParseDirectiveIrpc(SMLoc DirectiveLoc) { Lex(); // Lex the irpc definition. - Macro *M = ParseMacroLikeBody(DirectiveLoc); + MCAsmMacro *M = ParseMacroLikeBody(DirectiveLoc); if (!M) return true; @@ -3633,10 +3975,10 @@ bool AsmParser::ParseDirectiveIrpc(SMLoc DirectiveLoc) { StringRef Values = A.front().front().getString(); std::size_t I, End = Values.size(); for (I = 0; I < End; ++I) { - MacroArgument Arg; + MCAsmMacroArgument Arg; Arg.push_back(AsmToken(AsmToken::Identifier, Values.slice(I, I+1))); - MacroArguments Args; + MCAsmMacroArguments Args; Args.push_back(Arg); if (expandMacro(OS, M->Body, Parameters, Args, getTok().getLoc())) @@ -3660,10 +4002,11 @@ bool AsmParser::ParseDirectiveEndr(SMLoc DirectiveLoc) { return false; } -bool AsmParser::ParseDirectiveEmit(SMLoc IDLoc, ParseStatementInfo &Info) { +bool AsmParser::ParseDirectiveMSEmit(SMLoc IDLoc, ParseStatementInfo &Info, + size_t Len) { const MCExpr *Value; SMLoc ExprLoc = getLexer().getLoc(); - if (ParseExpression(Value)) + if (parseExpression(Value)) return true; const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); if (!MCE) @@ -3672,27 +4015,71 @@ bool AsmParser::ParseDirectiveEmit(SMLoc IDLoc, ParseStatementInfo &Info) { if (!isUIntN(8, IntValue) && !isIntN(8, IntValue)) return Error(ExprLoc, "literal value out of range for directive"); - Info.AsmRewrites->push_back(AsmRewrite(AOK_Emit, IDLoc, 5)); + Info.AsmRewrites->push_back(AsmRewrite(AOK_Emit, IDLoc, Len)); return false; } -bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, - unsigned &NumOutputs, unsigned &NumInputs, - SmallVectorImpl<std::pair<void *, bool> > &OpDecls, - SmallVectorImpl<std::string> &Constraints, - SmallVectorImpl<std::string> &Clobbers, - const MCInstrInfo *MII, - const MCInstPrinter *IP, - MCAsmParserSemaCallback &SI) { +bool AsmParser::ParseDirectiveMSAlign(SMLoc IDLoc, ParseStatementInfo &Info) { + const MCExpr *Value; + SMLoc ExprLoc = getLexer().getLoc(); + if (parseExpression(Value)) + return true; + const MCConstantExpr *MCE = dyn_cast<MCConstantExpr>(Value); + if (!MCE) + return Error(ExprLoc, "unexpected expression in align"); + uint64_t IntValue = MCE->getValue(); + if (!isPowerOf2_64(IntValue)) + return Error(ExprLoc, "literal value not a power of two greater then zero"); + + Info.AsmRewrites->push_back(AsmRewrite(AOK_Align, IDLoc, 5, + Log2_64(IntValue))); + return false; +} + +// We are comparing pointers, but the pointers are relative to a single string. +// Thus, this should always be deterministic. +static int RewritesSort(const void *A, const void *B) { + const AsmRewrite *AsmRewriteA = static_cast<const AsmRewrite *>(A); + const AsmRewrite *AsmRewriteB = static_cast<const AsmRewrite *>(B); + if (AsmRewriteA->Loc.getPointer() < AsmRewriteB->Loc.getPointer()) + return -1; + if (AsmRewriteB->Loc.getPointer() < AsmRewriteA->Loc.getPointer()) + return 1; + + // It's possible to have a SizeDirective rewrite and an Input/Output rewrite + // to the same location. Make sure the SizeDirective rewrite is performed + // first. This also ensure the sort algorithm is stable. + if (AsmRewriteA->Kind == AOK_SizeDirective) { + assert ((AsmRewriteB->Kind == AOK_Input || AsmRewriteB->Kind == AOK_Output) && + "Expected an Input/Output rewrite!"); + return -1; + } + if (AsmRewriteB->Kind == AOK_SizeDirective) { + assert ((AsmRewriteA->Kind == AOK_Input || AsmRewriteA->Kind == AOK_Output) && + "Expected an Input/Output rewrite!"); + return 1; + } + llvm_unreachable ("Unstable rewrite sort."); +} + +bool +AsmParser::parseMSInlineAsm(void *AsmLoc, std::string &AsmString, + unsigned &NumOutputs, unsigned &NumInputs, + SmallVectorImpl<std::pair<void *, bool> > &OpDecls, + SmallVectorImpl<std::string> &Constraints, + SmallVectorImpl<std::string> &Clobbers, + const MCInstrInfo *MII, + const MCInstPrinter *IP, + MCAsmParserSemaCallback &SI) { SmallVector<void *, 4> InputDecls; SmallVector<void *, 4> OutputDecls; - SmallVector<bool, 4> InputDeclsOffsetOf; - SmallVector<bool, 4> OutputDeclsOffsetOf; + SmallVector<bool, 4> InputDeclsAddressOf; + SmallVector<bool, 4> OutputDeclsAddressOf; SmallVector<std::string, 4> InputConstraints; SmallVector<std::string, 4> OutputConstraints; - std::set<std::string> ClobberRegs; + SmallVector<unsigned, 4> ClobberRegs; - SmallVector<struct AsmRewrite, 4> AsmStrRewrites; + SmallVector<AsmRewrite, 4> AsmStrRewrites; // Prime the lexer. Lex(); @@ -3705,64 +4092,63 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, if (ParseStatement(Info)) return true; - if (Info.Opcode != ~0U) { - const MCInstrDesc &Desc = MII->get(Info.Opcode); + if (Info.ParseError) + return true; - // Build the list of clobbers, outputs and inputs. - for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { - MCParsedAsmOperand *Operand = Info.ParsedOperands[i]; + if (Info.Opcode == ~0U) + continue; - // Immediate. - if (Operand->isImm()) { - if (Operand->needAsmRewrite()) - AsmStrRewrites.push_back(AsmRewrite(AOK_ImmPrefix, - Operand->getStartLoc())); - continue; - } + const MCInstrDesc &Desc = MII->get(Info.Opcode); - // Register operand. - if (Operand->isReg() && !Operand->isOffsetOf()) { - unsigned NumDefs = Desc.getNumDefs(); - // Clobber. - if (NumDefs && Operand->getMCOperandNum() < NumDefs) { - std::string Reg; - raw_string_ostream OS(Reg); - IP->printRegName(OS, Operand->getReg()); - ClobberRegs.insert(StringRef(OS.str())); - } - continue; - } + // Build the list of clobbers, outputs and inputs. + for (unsigned i = 1, e = Info.ParsedOperands.size(); i != e; ++i) { + MCParsedAsmOperand *Operand = Info.ParsedOperands[i]; - // Expr/Input or Output. - unsigned Size; - void *OpDecl = SI.LookupInlineAsmIdentifier(Operand->getName(), AsmLoc, - Size); - if (OpDecl) { - bool isOutput = (i == 1) && Desc.mayStore(); - if (!Operand->isOffsetOf() && Operand->needSizeDirective()) - AsmStrRewrites.push_back(AsmRewrite(AOK_SizeDirective, - Operand->getStartLoc(), - /*Len*/0, - Operand->getMemSize())); - if (isOutput) { - std::string Constraint = "="; - ++InputIdx; - OutputDecls.push_back(OpDecl); - OutputDeclsOffsetOf.push_back(Operand->isOffsetOf()); - Constraint += Operand->getConstraint().str(); - OutputConstraints.push_back(Constraint); - AsmStrRewrites.push_back(AsmRewrite(AOK_Output, - Operand->getStartLoc(), - Operand->getNameLen())); - } else { - InputDecls.push_back(OpDecl); - InputDeclsOffsetOf.push_back(Operand->isOffsetOf()); - InputConstraints.push_back(Operand->getConstraint().str()); - AsmStrRewrites.push_back(AsmRewrite(AOK_Input, - Operand->getStartLoc(), - Operand->getNameLen())); - } - } + // Immediate. + if (Operand->isImm()) { + if (Operand->needAsmRewrite()) + AsmStrRewrites.push_back(AsmRewrite(AOK_ImmPrefix, + Operand->getStartLoc())); + continue; + } + + // Register operand. + if (Operand->isReg() && !Operand->needAddressOf()) { + unsigned NumDefs = Desc.getNumDefs(); + // Clobber. + if (NumDefs && Operand->getMCOperandNum() < NumDefs) + ClobberRegs.push_back(Operand->getReg()); + continue; + } + + // Expr/Input or Output. + bool IsVarDecl; + unsigned Length, Size, Type; + void *OpDecl = SI.LookupInlineAsmIdentifier(Operand->getName(), AsmLoc, + Length, Size, Type, + IsVarDecl); + if (!OpDecl) + continue; + + bool isOutput = (i == 1) && Desc.mayStore(); + if (Operand->isMem() && Operand->needSizeDirective()) + AsmStrRewrites.push_back(AsmRewrite(AOK_SizeDirective, + Operand->getStartLoc(), /*Len*/0, + Operand->getMemSize())); + + if (isOutput) { + ++InputIdx; + OutputDecls.push_back(OpDecl); + OutputDeclsAddressOf.push_back(Operand->needAddressOf()); + OutputConstraints.push_back('=' + Operand->getConstraint().str()); + AsmStrRewrites.push_back(AsmRewrite(AOK_Output, Operand->getStartLoc(), + Operand->getNameLen())); + } else { + InputDecls.push_back(OpDecl); + InputDeclsAddressOf.push_back(Operand->needAddressOf()); + InputConstraints.push_back(Operand->getConstraint().str()); + AsmStrRewrites.push_back(AsmRewrite(AOK_Input, Operand->getStartLoc(), + Operand->getNameLen())); } } } @@ -3772,24 +4158,27 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, NumInputs = InputDecls.size(); // Set the unique clobbers. - for (std::set<std::string>::iterator I = ClobberRegs.begin(), - E = ClobberRegs.end(); I != E; ++I) - Clobbers.push_back(*I); + array_pod_sort(ClobberRegs.begin(), ClobberRegs.end()); + ClobberRegs.erase(std::unique(ClobberRegs.begin(), ClobberRegs.end()), + ClobberRegs.end()); + Clobbers.assign(ClobberRegs.size(), std::string()); + for (unsigned I = 0, E = ClobberRegs.size(); I != E; ++I) { + raw_string_ostream OS(Clobbers[I]); + IP->printRegName(OS, ClobberRegs[I]); + } // Merge the various outputs and inputs. Output are expected first. if (NumOutputs || NumInputs) { unsigned NumExprs = NumOutputs + NumInputs; OpDecls.resize(NumExprs); Constraints.resize(NumExprs); - // FIXME: Constraints are hard coded to 'm', but we need an 'r' - // constraint for offsetof. This needs to be cleaned up! for (unsigned i = 0; i < NumOutputs; ++i) { - OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsOffsetOf[i]); - Constraints[i] = OutputDeclsOffsetOf[i] ? "=r" : OutputConstraints[i]; + OpDecls[i] = std::make_pair(OutputDecls[i], OutputDeclsAddressOf[i]); + Constraints[i] = OutputConstraints[i]; } for (unsigned i = 0, j = NumOutputs; i < NumInputs; ++i, ++j) { - OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsOffsetOf[i]); - Constraints[j] = InputDeclsOffsetOf[i] ? "r" : InputConstraints[i]; + OpDecls[j] = std::make_pair(InputDecls[i], InputDeclsAddressOf[i]); + Constraints[j] = InputConstraints[i]; } } @@ -3798,10 +4187,14 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, AsmRewriteKind PrevKind = AOK_Imm; raw_string_ostream OS(AsmStringIR); const char *Start = SrcMgr.getMemoryBuffer(0)->getBufferStart(); - for (SmallVectorImpl<struct AsmRewrite>::iterator - I = AsmStrRewrites.begin(), E = AsmStrRewrites.end(); I != E; ++I) { + array_pod_sort(AsmStrRewrites.begin(), AsmStrRewrites.end(), RewritesSort); + for (SmallVectorImpl<AsmRewrite>::iterator I = AsmStrRewrites.begin(), + E = AsmStrRewrites.end(); + I != E; ++I) { const char *Loc = (*I).Loc.getPointer(); + assert(Loc >= Start && "Expected Loc to be after Start!"); + unsigned AdditionalSkip = 0; AsmRewriteKind Kind = (*I).Kind; // Emit everything up to the immediate/expression. If the previous rewrite @@ -3820,22 +4213,19 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, switch (Kind) { default: break; case AOK_Imm: - OS << Twine("$$"); - OS << (*I).Val; + OS << "$$" << (*I).Val; break; case AOK_ImmPrefix: - OS << Twine("$$"); + OS << "$$"; break; case AOK_Input: - OS << '$'; - OS << InputIdx++; + OS << '$' << InputIdx++; break; case AOK_Output: - OS << '$'; - OS << OutputIdx++; + OS << '$' << OutputIdx++; break; case AOK_SizeDirective: - switch((*I).Val) { + switch ((*I).Val) { default: break; case 8: OS << "byte ptr "; break; case 16: OS << "word ptr "; break; @@ -3849,6 +4239,15 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, case AOK_Emit: OS << ".byte"; break; + case AOK_Align: { + unsigned Val = (*I).Val; + OS << ".align " << Val; + + // Skip the original immediate. + assert(Val < 10 && "Expected alignment less then 2^10."); + AdditionalSkip = (Val < 4) ? 2 : Val < 7 ? 3 : 4; + break; + } case AOK_DotOperator: OS << (*I).Val; break; @@ -3856,7 +4255,7 @@ bool AsmParser::ParseMSInlineAsm(void *AsmLoc, std::string &AsmString, // Skip the original expression. if (Kind != AOK_SizeDirective) - Start = Loc + (*I).Len; + Start = Loc + (*I).Len + AdditionalSkip; } // Emit the remainder of the asm string. diff --git a/lib/MC/MCParser/COFFAsmParser.cpp b/lib/MC/MCParser/COFFAsmParser.cpp index e7c564a243..a50eab217d 100644 --- a/lib/MC/MCParser/COFFAsmParser.cpp +++ b/lib/MC/MCParser/COFFAsmParser.cpp @@ -24,10 +24,11 @@ using namespace llvm; namespace { class COFFAsmParser : public MCAsmParserExtension { - template<bool (COFFAsmParser::*Handler)(StringRef, SMLoc)> - void AddDirectiveHandler(StringRef Directive) { - getParser().AddDirectiveHandler(this, Directive, - HandleDirective<COFFAsmParser, Handler>); + template<bool (COFFAsmParser::*HandlerMethod)(StringRef, SMLoc)> + void addDirectiveHandler(StringRef Directive) { + MCAsmParser::ExtensionDirectiveHandler Handler = std::make_pair( + this, HandleDirective<COFFAsmParser, HandlerMethod>); + getParser().addDirectiveHandler(Directive, Handler); } bool ParseSectionSwitch(StringRef Section, @@ -38,43 +39,43 @@ class COFFAsmParser : public MCAsmParserExtension { // Call the base implementation. MCAsmParserExtension::Initialize(Parser); - AddDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveText>(".text"); - AddDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveData>(".data"); - AddDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveBSS>(".bss"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveDef>(".def"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveScl>(".scl"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveType>(".type"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveEndef>(".endef"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveSecRel32>(".secrel32"); + addDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveText>(".text"); + addDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveData>(".data"); + addDirectiveHandler<&COFFAsmParser::ParseSectionDirectiveBSS>(".bss"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveDef>(".def"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveScl>(".scl"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveType>(".type"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveEndef>(".endef"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveSecRel32>(".secrel32"); // Win64 EH directives. - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveStartProc>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveStartProc>( ".seh_proc"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndProc>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndProc>( ".seh_endproc"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveStartChained>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveStartChained>( ".seh_startchained"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndChained>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndChained>( ".seh_endchained"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveHandler>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveHandler>( ".seh_handler"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveHandlerData>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveHandlerData>( ".seh_handlerdata"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectivePushReg>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectivePushReg>( ".seh_pushreg"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSetFrame>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSetFrame>( ".seh_setframe"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveAllocStack>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveAllocStack>( ".seh_stackalloc"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSaveReg>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSaveReg>( ".seh_savereg"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSaveXMM>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveSaveXMM>( ".seh_savexmm"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectivePushFrame>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectivePushFrame>( ".seh_pushframe"); - AddDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndProlog>( + addDirectiveHandler<&COFFAsmParser::ParseSEHDirectiveEndProlog>( ".seh_endprologue"); - AddDirectiveHandler<&COFFAsmParser::ParseDirectiveSymbolAttribute>(".weak"); + addDirectiveHandler<&COFFAsmParser::ParseDirectiveSymbolAttribute>(".weak"); } bool ParseSectionDirectiveText(StringRef, SMLoc) { @@ -140,7 +141,7 @@ bool COFFAsmParser::ParseDirectiveSymbolAttribute(StringRef Directive, SMLoc) { for (;;) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); @@ -176,7 +177,7 @@ bool COFFAsmParser::ParseSectionSwitch(StringRef Section, bool COFFAsmParser::ParseDirectiveDef(StringRef, SMLoc) { StringRef SymbolName; - if (getParser().ParseIdentifier(SymbolName)) + if (getParser().parseIdentifier(SymbolName)) return TokError("expected identifier in directive"); MCSymbol *Sym = getContext().GetOrCreateSymbol(SymbolName); @@ -189,7 +190,7 @@ bool COFFAsmParser::ParseDirectiveDef(StringRef, SMLoc) { bool COFFAsmParser::ParseDirectiveScl(StringRef, SMLoc) { int64_t SymbolStorageClass; - if (getParser().ParseAbsoluteExpression(SymbolStorageClass)) + if (getParser().parseAbsoluteExpression(SymbolStorageClass)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -202,7 +203,7 @@ bool COFFAsmParser::ParseDirectiveScl(StringRef, SMLoc) { bool COFFAsmParser::ParseDirectiveType(StringRef, SMLoc) { int64_t Type; - if (getParser().ParseAbsoluteExpression(Type)) + if (getParser().parseAbsoluteExpression(Type)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -221,7 +222,7 @@ bool COFFAsmParser::ParseDirectiveEndef(StringRef, SMLoc) { bool COFFAsmParser::ParseDirectiveSecRel32(StringRef, SMLoc) { StringRef SymbolID; - if (getParser().ParseIdentifier(SymbolID)) + if (getParser().parseIdentifier(SymbolID)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -236,7 +237,7 @@ bool COFFAsmParser::ParseDirectiveSecRel32(StringRef, SMLoc) { bool COFFAsmParser::ParseSEHDirectiveStartProc(StringRef, SMLoc) { StringRef SymbolID; - if (getParser().ParseIdentifier(SymbolID)) + if (getParser().parseIdentifier(SymbolID)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -269,7 +270,7 @@ bool COFFAsmParser::ParseSEHDirectiveEndChained(StringRef, SMLoc) { bool COFFAsmParser::ParseSEHDirectiveHandler(StringRef, SMLoc) { StringRef SymbolID; - if (getParser().ParseIdentifier(SymbolID)) + if (getParser().parseIdentifier(SymbolID)) return true; if (getLexer().isNot(AsmToken::Comma)) @@ -322,7 +323,7 @@ bool COFFAsmParser::ParseSEHDirectiveSetFrame(StringRef, SMLoc L) { Lex(); SMLoc startLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Off)) + if (getParser().parseAbsoluteExpression(Off)) return true; if (Off & 0x0F) @@ -339,7 +340,7 @@ bool COFFAsmParser::ParseSEHDirectiveSetFrame(StringRef, SMLoc L) { bool COFFAsmParser::ParseSEHDirectiveAllocStack(StringRef, SMLoc) { int64_t Size; SMLoc startLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Size)) + if (getParser().parseAbsoluteExpression(Size)) return true; if (Size & 7) @@ -363,7 +364,7 @@ bool COFFAsmParser::ParseSEHDirectiveSaveReg(StringRef, SMLoc L) { Lex(); SMLoc startLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Off)) + if (getParser().parseAbsoluteExpression(Off)) return true; if (Off & 7) @@ -390,7 +391,7 @@ bool COFFAsmParser::ParseSEHDirectiveSaveXMM(StringRef, SMLoc L) { Lex(); SMLoc startLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Off)) + if (getParser().parseAbsoluteExpression(Off)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -411,7 +412,7 @@ bool COFFAsmParser::ParseSEHDirectivePushFrame(StringRef, SMLoc) { if (getLexer().is(AsmToken::At)) { SMLoc startLoc = getLexer().getLoc(); Lex(); - if (!getParser().ParseIdentifier(CodeID)) { + if (!getParser().parseIdentifier(CodeID)) { if (CodeID != "code") return Error(startLoc, "expected @code"); Code = true; @@ -438,7 +439,7 @@ bool COFFAsmParser::ParseAtUnwindOrAtExcept(bool &unwind, bool &except) { return TokError("a handler attribute must begin with '@'"); SMLoc startLoc = getLexer().getLoc(); Lex(); - if (getParser().ParseIdentifier(identifier)) + if (getParser().parseIdentifier(identifier)) return Error(startLoc, "expected @unwind or @except"); if (identifier == "unwind") unwind = true; @@ -479,7 +480,7 @@ bool COFFAsmParser::ParseSEHRegisterNumber(unsigned &RegNo) { } else { int64_t n; - if (getParser().ParseAbsoluteExpression(n)) + if (getParser().parseAbsoluteExpression(n)) return true; if (n > 15) return Error(startLoc, "register number is too high"); diff --git a/lib/MC/MCParser/DarwinAsmParser.cpp b/lib/MC/MCParser/DarwinAsmParser.cpp index 20c949dbda..6d6409fb69 100644 --- a/lib/MC/MCParser/DarwinAsmParser.cpp +++ b/lib/MC/MCParser/DarwinAsmParser.cpp @@ -26,10 +26,11 @@ namespace { /// \brief Implementation of directive handling which is shared across all /// Darwin targets. class DarwinAsmParser : public MCAsmParserExtension { - template<bool (DarwinAsmParser::*Handler)(StringRef, SMLoc)> - void AddDirectiveHandler(StringRef Directive) { - getParser().AddDirectiveHandler(this, Directive, - HandleDirective<DarwinAsmParser, Handler>); + template<bool (DarwinAsmParser::*HandlerMethod)(StringRef, SMLoc)> + void addDirectiveHandler(StringRef Directive) { + MCAsmParser::ExtensionDirectiveHandler Handler = std::make_pair( + this, HandleDirective<DarwinAsmParser, HandlerMethod>); + getParser().addDirectiveHandler(Directive, Handler); } bool ParseSectionSwitch(const char *Segment, const char *Section, @@ -43,77 +44,128 @@ public: // Call the base implementation. this->MCAsmParserExtension::Initialize(Parser); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveDesc>(".desc"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveLsym>(".lsym"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveSubsectionsViaSymbols>( + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveDesc>(".desc"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveLsym>(".lsym"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveSubsectionsViaSymbols>( ".subsections_via_symbols"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveDumpOrLoad>(".dump"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveDumpOrLoad>(".load"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveSection>(".section"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectivePushSection>(".pushsection"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectivePopSection>(".popsection"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectivePrevious>(".previous"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveSecureLogUnique>( + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveDumpOrLoad>(".dump"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveDumpOrLoad>(".load"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveSection>(".section"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectivePushSection>( + ".pushsection"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectivePopSection>( + ".popsection"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectivePrevious>(".previous"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveSecureLogUnique>( ".secure_log_unique"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveSecureLogReset>( + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveSecureLogReset>( ".secure_log_reset"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveTBSS>(".tbss"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveZerofill>(".zerofill"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveTBSS>(".tbss"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveZerofill>(".zerofill"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveDataRegion>(".data_region"); - AddDirectiveHandler<&DarwinAsmParser::ParseDirectiveDataRegionEnd>(".end_data_region"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveDataRegion>( + ".data_region"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveDataRegionEnd>( + ".end_data_region"); // Special section directives. - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConst>(".const"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConstData>(".const_data"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConstructor>(".constructor"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveCString>(".cstring"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveData>(".data"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveDestructor>(".destructor"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveDyld>(".dyld"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveFVMLibInit0>(".fvmlib_init0"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveFVMLibInit1>(".fvmlib_init1"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLazySymbolPointers>(".lazy_symbol_pointer"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral16>(".literal16"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral4>(".literal4"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral8>(".literal8"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveModInitFunc>(".mod_init_func"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveModTermFunc>(".mod_term_func"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveNonLazySymbolPointers>(".non_lazy_symbol_pointer"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCatClsMeth>(".objc_cat_cls_meth"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCatInstMeth>(".objc_cat_inst_meth"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCategory>(".objc_category"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClass>(".objc_class"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClassNames>(".objc_class_names"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClassVars>(".objc_class_vars"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClsMeth>(".objc_cls_meth"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClsRefs>(".objc_cls_refs"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCInstMeth>(".objc_inst_meth"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCInstanceVars>(".objc_instance_vars"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMessageRefs>(".objc_message_refs"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMetaClass>(".objc_meta_class"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMethVarNames>(".objc_meth_var_names"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMethVarTypes>(".objc_meth_var_types"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCModuleInfo>(".objc_module_info"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCProtocol>(".objc_protocol"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCSelectorStrs>(".objc_selector_strs"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCStringObject>(".objc_string_object"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCSymbols>(".objc_symbols"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectivePICSymbolStub>(".picsymbol_stub"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveStaticConst>(".static_const"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveStaticData>(".static_data"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveSymbolStub>(".symbol_stub"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveTData>(".tdata"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveText>(".text"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveThreadInitFunc>(".thread_init_func"); - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveTLV>(".tlv"); - - AddDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveIdent>(".ident"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConst>(".const"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConstData>( + ".const_data"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveConstructor>( + ".constructor"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveCString>( + ".cstring"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveData>(".data"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveDestructor>( + ".destructor"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveDyld>(".dyld"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveFVMLibInit0>( + ".fvmlib_init0"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveFVMLibInit1>( + ".fvmlib_init1"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveLazySymbolPointers>( + ".lazy_symbol_pointer"); + addDirectiveHandler<&DarwinAsmParser::ParseDirectiveLinkerOption>( + ".linker_option"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral16>( + ".literal16"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral4>( + ".literal4"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveLiteral8>( + ".literal8"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveModInitFunc>( + ".mod_init_func"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveModTermFunc>( + ".mod_term_func"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveNonLazySymbolPointers>( + ".non_lazy_symbol_pointer"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCatClsMeth>( + ".objc_cat_cls_meth"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCatInstMeth>( + ".objc_cat_inst_meth"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCCategory>( + ".objc_category"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClass>( + ".objc_class"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClassNames>( + ".objc_class_names"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClassVars>( + ".objc_class_vars"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClsMeth>( + ".objc_cls_meth"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCClsRefs>( + ".objc_cls_refs"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCInstMeth>( + ".objc_inst_meth"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveObjCInstanceVars>( + ".objc_instance_vars"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMessageRefs>( + ".objc_message_refs"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCMetaClass>( + ".objc_meta_class"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveObjCMethVarNames>( + ".objc_meth_var_names"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveObjCMethVarTypes>( + ".objc_meth_var_types"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCModuleInfo>( + ".objc_module_info"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCProtocol>( + ".objc_protocol"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveObjCSelectorStrs>( + ".objc_selector_strs"); + addDirectiveHandler< + &DarwinAsmParser::ParseSectionDirectiveObjCStringObject>( + ".objc_string_object"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveObjCSymbols>( + ".objc_symbols"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectivePICSymbolStub>( + ".picsymbol_stub"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveStaticConst>( + ".static_const"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveStaticData>( + ".static_data"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveSymbolStub>( + ".symbol_stub"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveTData>(".tdata"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveText>(".text"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveThreadInitFunc>( + ".thread_init_func"); + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveTLV>(".tlv"); + + addDirectiveHandler<&DarwinAsmParser::ParseSectionDirectiveIdent>(".ident"); } bool ParseDirectiveDesc(StringRef, SMLoc); bool ParseDirectiveDumpOrLoad(StringRef, SMLoc); bool ParseDirectiveLsym(StringRef, SMLoc); + bool ParseDirectiveLinkerOption(StringRef, SMLoc); bool ParseDirectiveSection(StringRef, SMLoc); bool ParseDirectivePushSection(StringRef, SMLoc); bool ParseDirectivePopSection(StringRef, SMLoc); @@ -293,7 +345,7 @@ public: } bool ParseSectionDirectiveIdent(StringRef, SMLoc) { // Darwin silently ignores the .ident directive. - getParser().EatToEndOfStatement(); + getParser().eatToEndOfStatement(); return false; } bool ParseSectionDirectiveThreadInitFunc(StringRef, SMLoc) { @@ -314,7 +366,7 @@ bool DarwinAsmParser::ParseSectionSwitch(const char *Segment, Lex(); // FIXME: Arch specific. - bool isText = StringRef(Segment) == "__TEXT"; // FIXME: Hack. + bool isText = TAA & MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS; getStreamer().SwitchSection(getContext().getMachOSection( Segment, Section, TAA, StubSize, isText ? SectionKind::getText() @@ -338,7 +390,7 @@ bool DarwinAsmParser::ParseSectionSwitch(const char *Segment, /// ::= .desc identifier , expression bool DarwinAsmParser::ParseDirectiveDesc(StringRef, SMLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); // Handle the identifier as the key symbol. @@ -349,7 +401,7 @@ bool DarwinAsmParser::ParseDirectiveDesc(StringRef, SMLoc) { Lex(); int64_t DescValue; - if (getParser().ParseAbsoluteExpression(DescValue)) + if (getParser().parseAbsoluteExpression(DescValue)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -386,11 +438,38 @@ bool DarwinAsmParser::ParseDirectiveDumpOrLoad(StringRef Directive, return Warning(IDLoc, "ignoring directive .load for now"); } +/// ParseDirectiveLinkerOption +/// ::= .linker_option "string" ( , "string" )* +bool DarwinAsmParser::ParseDirectiveLinkerOption(StringRef IDVal, SMLoc) { + SmallVector<std::string, 4> Args; + for (;;) { + if (getLexer().isNot(AsmToken::String)) + return TokError("expected string in '" + Twine(IDVal) + "' directive"); + + std::string Data; + if (getParser().parseEscapedString(Data)) + return true; + + Args.push_back(Data); + + Lex(); + if (getLexer().is(AsmToken::EndOfStatement)) + break; + + if (getLexer().isNot(AsmToken::Comma)) + return TokError("unexpected token in '" + Twine(IDVal) + "' directive"); + Lex(); + } + + getStreamer().EmitLinkerOptions(Args); + return false; +} + /// ParseDirectiveLsym /// ::= .lsym identifier , expression bool DarwinAsmParser::ParseDirectiveLsym(StringRef, SMLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); // Handle the identifier as the key symbol. @@ -401,7 +480,7 @@ bool DarwinAsmParser::ParseDirectiveLsym(StringRef, SMLoc) { Lex(); const MCExpr *Value; - if (getParser().ParseExpression(Value)) + if (getParser().parseExpression(Value)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -422,7 +501,7 @@ bool DarwinAsmParser::ParseDirectiveSection(StringRef, SMLoc) { SMLoc Loc = getLexer().getLoc(); StringRef SectionName; - if (getParser().ParseIdentifier(SectionName)) + if (getParser().parseIdentifier(SectionName)) return Error(Loc, "expected identifier after '.section' directive"); // Verify there is a following comma. @@ -497,7 +576,7 @@ bool DarwinAsmParser::ParseDirectivePrevious(StringRef DirName, SMLoc) { /// ParseDirectiveSecureLogUnique /// ::= .secure_log_unique ... message ... bool DarwinAsmParser::ParseDirectiveSecureLogUnique(StringRef, SMLoc IDLoc) { - StringRef LogMessage = getParser().ParseStringToEndOfStatement(); + StringRef LogMessage = getParser().parseStringToEndOfStatement(); if (getLexer().isNot(AsmToken::EndOfStatement)) return TokError("unexpected token in '.secure_log_unique' directive"); @@ -565,7 +644,7 @@ bool DarwinAsmParser::ParseDirectiveSubsectionsViaSymbols(StringRef, SMLoc) { bool DarwinAsmParser::ParseDirectiveTBSS(StringRef, SMLoc) { SMLoc IDLoc = getLexer().getLoc(); StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); // Handle the identifier as the key symbol. @@ -577,7 +656,7 @@ bool DarwinAsmParser::ParseDirectiveTBSS(StringRef, SMLoc) { int64_t Size; SMLoc SizeLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Size)) + if (getParser().parseAbsoluteExpression(Size)) return true; int64_t Pow2Alignment = 0; @@ -585,7 +664,7 @@ bool DarwinAsmParser::ParseDirectiveTBSS(StringRef, SMLoc) { if (getLexer().is(AsmToken::Comma)) { Lex(); Pow2AlignmentLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Pow2Alignment)) + if (getParser().parseAbsoluteExpression(Pow2Alignment)) return true; } @@ -620,7 +699,7 @@ bool DarwinAsmParser::ParseDirectiveTBSS(StringRef, SMLoc) { /// , align_expression ]] bool DarwinAsmParser::ParseDirectiveZerofill(StringRef, SMLoc) { StringRef Segment; - if (getParser().ParseIdentifier(Segment)) + if (getParser().parseIdentifier(Segment)) return TokError("expected segment name after '.zerofill' directive"); if (getLexer().isNot(AsmToken::Comma)) @@ -628,7 +707,7 @@ bool DarwinAsmParser::ParseDirectiveZerofill(StringRef, SMLoc) { Lex(); StringRef Section; - if (getParser().ParseIdentifier(Section)) + if (getParser().parseIdentifier(Section)) return TokError("expected section name after comma in '.zerofill' " "directive"); @@ -648,7 +727,7 @@ bool DarwinAsmParser::ParseDirectiveZerofill(StringRef, SMLoc) { SMLoc IDLoc = getLexer().getLoc(); StringRef IDStr; - if (getParser().ParseIdentifier(IDStr)) + if (getParser().parseIdentifier(IDStr)) return TokError("expected identifier in directive"); // handle the identifier as the key symbol. @@ -660,7 +739,7 @@ bool DarwinAsmParser::ParseDirectiveZerofill(StringRef, SMLoc) { int64_t Size; SMLoc SizeLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Size)) + if (getParser().parseAbsoluteExpression(Size)) return true; int64_t Pow2Alignment = 0; @@ -668,7 +747,7 @@ bool DarwinAsmParser::ParseDirectiveZerofill(StringRef, SMLoc) { if (getLexer().is(AsmToken::Comma)) { Lex(); Pow2AlignmentLoc = getLexer().getLoc(); - if (getParser().ParseAbsoluteExpression(Pow2Alignment)) + if (getParser().parseAbsoluteExpression(Pow2Alignment)) return true; } @@ -712,7 +791,7 @@ bool DarwinAsmParser::ParseDirectiveDataRegion(StringRef, SMLoc) { } StringRef RegionType; SMLoc Loc = getParser().getTok().getLoc(); - if (getParser().ParseIdentifier(RegionType)) + if (getParser().parseIdentifier(RegionType)) return TokError("expected region type after '.data_region' directive"); int Kind = StringSwitch<int>(RegionType) .Case("jt8", MCDR_DataRegionJT8) diff --git a/lib/MC/MCParser/ELFAsmParser.cpp b/lib/MC/MCParser/ELFAsmParser.cpp index d55de1f3fb..4c45e08744 100644 --- a/lib/MC/MCParser/ELFAsmParser.cpp +++ b/lib/MC/MCParser/ELFAsmParser.cpp @@ -22,10 +22,12 @@ using namespace llvm; namespace { class ELFAsmParser : public MCAsmParserExtension { - template<bool (ELFAsmParser::*Handler)(StringRef, SMLoc)> - void AddDirectiveHandler(StringRef Directive) { - getParser().AddDirectiveHandler(this, Directive, - HandleDirective<ELFAsmParser, Handler>); + template<bool (ELFAsmParser::*HandlerMethod)(StringRef, SMLoc)> + void addDirectiveHandler(StringRef Directive) { + MCAsmParser::ExtensionDirectiveHandler Handler = std::make_pair( + this, HandleDirective<ELFAsmParser, HandlerMethod>); + + getParser().addDirectiveHandler(Directive, Handler); } bool ParseSectionSwitch(StringRef Section, unsigned Type, @@ -41,38 +43,38 @@ public: // Call the base implementation. this->MCAsmParserExtension::Initialize(Parser); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveData>(".data"); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveText>(".text"); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveBSS>(".bss"); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveRoData>(".rodata"); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTData>(".tdata"); - AddDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTBSS>(".tbss"); - AddDirectiveHandler< + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveData>(".data"); + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveText>(".text"); + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveBSS>(".bss"); + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveRoData>(".rodata"); + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTData>(".tdata"); + addDirectiveHandler<&ELFAsmParser::ParseSectionDirectiveTBSS>(".tbss"); + addDirectiveHandler< &ELFAsmParser::ParseSectionDirectiveDataRel>(".data.rel"); - AddDirectiveHandler< + addDirectiveHandler< &ELFAsmParser::ParseSectionDirectiveDataRelRo>(".data.rel.ro"); - AddDirectiveHandler< + addDirectiveHandler< &ELFAsmParser::ParseSectionDirectiveDataRelRoLocal>(".data.rel.ro.local"); - AddDirectiveHandler< + addDirectiveHandler< &ELFAsmParser::ParseSectionDirectiveEhFrame>(".eh_frame"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSection>(".section"); - AddDirectiveHandler< + addDirectiveHandler<&ELFAsmParser::ParseDirectiveSection>(".section"); + addDirectiveHandler< &ELFAsmParser::ParseDirectivePushSection>(".pushsection"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectivePopSection>(".popsection"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSize>(".size"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectivePrevious>(".previous"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveType>(".type"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveIdent>(".ident"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymver>(".symver"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveVersion>(".version"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveWeakref>(".weakref"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".weak"); - AddDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".local"); - AddDirectiveHandler< + addDirectiveHandler<&ELFAsmParser::ParseDirectivePopSection>(".popsection"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveSize>(".size"); + addDirectiveHandler<&ELFAsmParser::ParseDirectivePrevious>(".previous"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveType>(".type"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveIdent>(".ident"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveSymver>(".symver"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveVersion>(".version"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveWeakref>(".weakref"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".weak"); + addDirectiveHandler<&ELFAsmParser::ParseDirectiveSymbolAttribute>(".local"); + addDirectiveHandler< &ELFAsmParser::ParseDirectiveSymbolAttribute>(".protected"); - AddDirectiveHandler< + addDirectiveHandler< &ELFAsmParser::ParseDirectiveSymbolAttribute>(".internal"); - AddDirectiveHandler< + addDirectiveHandler< &ELFAsmParser::ParseDirectiveSymbolAttribute>(".hidden"); } @@ -167,7 +169,7 @@ bool ELFAsmParser::ParseDirectiveSymbolAttribute(StringRef Directive, SMLoc) { for (;;) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); @@ -201,7 +203,7 @@ bool ELFAsmParser::ParseSectionSwitch(StringRef Section, unsigned Type, bool ELFAsmParser::ParseDirectiveSize(StringRef, SMLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); MCSymbol *Sym = getContext().GetOrCreateSymbol(Name); @@ -210,7 +212,7 @@ bool ELFAsmParser::ParseDirectiveSize(StringRef, SMLoc) { Lex(); const MCExpr *Expr; - if (getParser().ParseExpression(Expr)) + if (getParser().parseExpression(Expr)) return true; if (getLexer().isNot(AsmToken::EndOfStatement)) @@ -222,7 +224,7 @@ bool ELFAsmParser::ParseDirectiveSize(StringRef, SMLoc) { bool ELFAsmParser::ParseSectionName(StringRef &SectionName) { // A section name can contain -, so we cannot just use - // ParseIdentifier. + // parseIdentifier. SMLoc FirstLoc = getLexer().getLoc(); unsigned Size = 0; @@ -375,14 +377,14 @@ bool ELFAsmParser::ParseDirectiveSection(StringRef, SMLoc) { return TokError("expected '@' or '%' before type"); Lex(); - if (getParser().ParseIdentifier(TypeName)) + if (getParser().parseIdentifier(TypeName)) return TokError("expected identifier in directive"); if (Mergeable) { if (getLexer().isNot(AsmToken::Comma)) return TokError("expected the entry size"); Lex(); - if (getParser().ParseAbsoluteExpression(Size)) + if (getParser().parseAbsoluteExpression(Size)) return true; if (Size <= 0) return TokError("entry size must be positive"); @@ -392,12 +394,12 @@ bool ELFAsmParser::ParseDirectiveSection(StringRef, SMLoc) { if (getLexer().isNot(AsmToken::Comma)) return TokError("expected group name"); Lex(); - if (getParser().ParseIdentifier(GroupName)) + if (getParser().parseIdentifier(GroupName)) return true; if (getLexer().is(AsmToken::Comma)) { Lex(); StringRef Linkage; - if (getParser().ParseIdentifier(Linkage)) + if (getParser().parseIdentifier(Linkage)) return true; if (Linkage != "comdat") return TokError("Linkage must be 'comdat'"); @@ -411,7 +413,16 @@ bool ELFAsmParser::ParseDirectiveSection(StringRef, SMLoc) { unsigned Type = ELF::SHT_PROGBITS; - if (!TypeName.empty()) { + if (TypeName.empty()) { + if (SectionName.startswith(".note")) + Type = ELF::SHT_NOTE; + else if (SectionName == ".init_array") + Type = ELF::SHT_INIT_ARRAY; + else if (SectionName == ".fini_array") + Type = ELF::SHT_FINI_ARRAY; + else if (SectionName == ".preinit_array") + Type = ELF::SHT_PREINIT_ARRAY; + } else { if (TypeName == "init_array") Type = ELF::SHT_INIT_ARRAY; else if (TypeName == "fini_array") @@ -450,7 +461,7 @@ bool ELFAsmParser::ParseDirectivePrevious(StringRef DirName, SMLoc) { /// ::= .type identifier , @attribute bool ELFAsmParser::ParseDirectiveType(StringRef, SMLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); // Handle the identifier as the key symbol. @@ -468,7 +479,7 @@ bool ELFAsmParser::ParseDirectiveType(StringRef, SMLoc) { SMLoc TypeLoc; TypeLoc = getLexer().getLoc(); - if (getParser().ParseIdentifier(Type)) + if (getParser().parseIdentifier(Type)) return TokError("expected symbol type in directive"); MCSymbolAttr Attr = StringSwitch<MCSymbolAttr>(Type) @@ -517,7 +528,7 @@ bool ELFAsmParser::ParseDirectiveIdent(StringRef, SMLoc) { getStreamer().EmitIntValue(0, 1); SeenIdent = true; } - getStreamer().EmitBytes(Data, 0); + getStreamer().EmitBytes(Data); getStreamer().EmitIntValue(0, 1); getStreamer().PopSection(); return false; @@ -527,7 +538,7 @@ bool ELFAsmParser::ParseDirectiveIdent(StringRef, SMLoc) { /// ::= .symver foo, bar2@zed bool ELFAsmParser::ParseDirectiveSymver(StringRef, SMLoc) { StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); if (getLexer().isNot(AsmToken::Comma)) @@ -536,7 +547,7 @@ bool ELFAsmParser::ParseDirectiveSymver(StringRef, SMLoc) { Lex(); StringRef AliasName; - if (getParser().ParseIdentifier(AliasName)) + if (getParser().parseIdentifier(AliasName)) return TokError("expected identifier in directive"); if (AliasName.find('@') == StringRef::npos) @@ -569,7 +580,7 @@ bool ELFAsmParser::ParseDirectiveVersion(StringRef, SMLoc) { getStreamer().EmitIntValue(Data.size()+1, 4); // namesz. getStreamer().EmitIntValue(0, 4); // descsz = 0 (no description). getStreamer().EmitIntValue(1, 4); // type = NT_VERSION. - getStreamer().EmitBytes(Data, 0); // name. + getStreamer().EmitBytes(Data); // name. getStreamer().EmitIntValue(0, 1); // terminate the string. getStreamer().EmitValueToAlignment(4); // ensure 4 byte alignment. getStreamer().PopSection(); @@ -582,7 +593,7 @@ bool ELFAsmParser::ParseDirectiveWeakref(StringRef, SMLoc) { // FIXME: Share code with the other alias building directives. StringRef AliasName; - if (getParser().ParseIdentifier(AliasName)) + if (getParser().parseIdentifier(AliasName)) return TokError("expected identifier in directive"); if (getLexer().isNot(AsmToken::Comma)) @@ -591,7 +602,7 @@ bool ELFAsmParser::ParseDirectiveWeakref(StringRef, SMLoc) { Lex(); StringRef Name; - if (getParser().ParseIdentifier(Name)) + if (getParser().parseIdentifier(Name)) return TokError("expected identifier in directive"); MCSymbol *Alias = getContext().GetOrCreateSymbol(AliasName); diff --git a/lib/MC/MCParser/MCAsmLexer.cpp b/lib/MC/MCParser/MCAsmLexer.cpp index 384b341bc7..3867691107 100644 --- a/lib/MC/MCParser/MCAsmLexer.cpp +++ b/lib/MC/MCParser/MCAsmLexer.cpp @@ -28,5 +28,5 @@ SMLoc AsmToken::getLoc() const { } SMLoc AsmToken::getEndLoc() const { - return SMLoc::getFromPointer(Str.data() + Str.size() - 1); + return SMLoc::getFromPointer(Str.data() + Str.size()); } diff --git a/lib/MC/MCParser/MCAsmParser.cpp b/lib/MC/MCParser/MCAsmParser.cpp index a8b00cd84a..6e1ebad36c 100644 --- a/lib/MC/MCParser/MCAsmParser.cpp +++ b/lib/MC/MCParser/MCAsmParser.cpp @@ -38,9 +38,9 @@ bool MCAsmParser::TokError(const Twine &Msg, ArrayRef<SMRange> Ranges) { return true; } -bool MCAsmParser::ParseExpression(const MCExpr *&Res) { +bool MCAsmParser::parseExpression(const MCExpr *&Res) { SMLoc L; - return ParseExpression(Res, L); + return parseExpression(Res, L); } void MCParsedAsmOperand::dump() const { diff --git a/lib/MC/MCPureStreamer.cpp b/lib/MC/MCPureStreamer.cpp index 1563bdd107..0e04c5537a 100644 --- a/lib/MC/MCPureStreamer.cpp +++ b/lib/MC/MCPureStreamer.cpp @@ -28,15 +28,17 @@ private: virtual void EmitInstToData(const MCInst &Inst); public: - MCPureStreamer(MCContext &Context, MCAsmBackend &TAB, - raw_ostream &OS, MCCodeEmitter *Emitter) - : MCObjectStreamer(Context, TAB, OS, Emitter) {} + MCPureStreamer(MCContext &Context, MCAsmBackend &TAB, raw_ostream &OS, + MCCodeEmitter *Emitter) + : MCObjectStreamer(SK_PureStreamer, Context, TAB, OS, Emitter) {} /// @name MCStreamer Interface /// @{ virtual void InitSections(); + virtual void InitToTextSection(); virtual void EmitLabel(MCSymbol *Symbol); + virtual void EmitDebugLabel(MCSymbol *Symbol); virtual void EmitZerofill(const MCSection *Section, MCSymbol *Symbol = 0, uint64_t Size = 0, unsigned ByteAlignment = 0); virtual void EmitBytes(StringRef Data, unsigned AddrSpace); @@ -93,21 +95,28 @@ public: report_fatal_error("unsupported directive in pure streamer"); } virtual bool EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, - StringRef Filename) { + StringRef Filename, unsigned CUID = 0) { report_fatal_error("unsupported directive in pure streamer"); } /// @} + + static bool classof(const MCStreamer *S) { + return S->getKind() == SK_PureStreamer; + } }; } // end anonymous namespace. void MCPureStreamer::InitSections() { + InitToTextSection(); +} + +void MCPureStreamer::InitToTextSection() { // FIMXE: To what!? SwitchSection(getContext().getMachOSection("__TEXT", "__text", MCSectionMachO::S_ATTR_PURE_INSTRUCTIONS, 0, SectionKind::getText())); - } void MCPureStreamer::EmitLabel(MCSymbol *Symbol) { @@ -134,6 +143,11 @@ void MCPureStreamer::EmitLabel(MCSymbol *Symbol) { SD.setOffset(F->getContents().size()); } + +void MCPureStreamer::EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); +} + void MCPureStreamer::EmitZerofill(const MCSection *Section, MCSymbol *Symbol, uint64_t Size, unsigned ByteAlignment) { report_fatal_error("not yet implemented in pure streamer"); @@ -182,7 +196,8 @@ bool MCPureStreamer::EmitValueToOffset(const MCExpr *Offset, } void MCPureStreamer::EmitInstToFragment(const MCInst &Inst) { - MCInstFragment *IF = new MCInstFragment(Inst, getCurrentSectionData()); + MCRelaxableFragment *IF = + new MCRelaxableFragment(Inst, getCurrentSectionData()); // Add the fixups and data. // diff --git a/lib/MC/MCSectionMachO.cpp b/lib/MC/MCSectionMachO.cpp index e771556262..fc323155be 100644 --- a/lib/MC/MCSectionMachO.cpp +++ b/lib/MC/MCSectionMachO.cpp @@ -165,9 +165,9 @@ bool MCSectionMachO::isVirtualSection() const { /// StripSpaces - This removes leading and trailing spaces from the StringRef. static void StripSpaces(StringRef &Str) { - while (!Str.empty() && isspace(Str[0])) + while (!Str.empty() && isspace(static_cast<unsigned char>(Str[0]))) Str = Str.substr(1); - while (!Str.empty() && isspace(Str.back())) + while (!Str.empty() && isspace(static_cast<unsigned char>(Str.back()))) Str = Str.substr(0, Str.size()-1); } diff --git a/lib/MC/MCStreamer.cpp b/lib/MC/MCStreamer.cpp index 96d6d691d2..51ef415542 100644 --- a/lib/MC/MCStreamer.cpp +++ b/lib/MC/MCStreamer.cpp @@ -21,10 +21,9 @@ #include <cstdlib> using namespace llvm; -MCStreamer::MCStreamer(MCContext &Ctx) - : Context(Ctx), EmitEHFrame(true), EmitDebugFrame(false), - CurrentW64UnwindInfo(0), LastSymbol(0), - AutoInitSections(false) { +MCStreamer::MCStreamer(StreamerKind Kind, MCContext &Ctx) + : Kind(Kind), Context(Ctx), EmitEHFrame(true), EmitDebugFrame(false), + CurrentW64UnwindInfo(0), LastSymbol(0), AutoInitSections(false) { const MCSection *section = NULL; SectionStack.push_back(std::make_pair(section, section)); } @@ -34,6 +33,18 @@ MCStreamer::~MCStreamer() { delete W64UnwindInfos[i]; } +void MCStreamer::reset() { + for (unsigned i = 0; i < getNumW64UnwindInfos(); ++i) + delete W64UnwindInfos[i]; + EmitEHFrame = true; + EmitDebugFrame = false; + CurrentW64UnwindInfo = 0; + LastSymbol = 0; + const MCSection *section = NULL; + SectionStack.clear(); + SectionStack.push_back(std::make_pair(section, section)); +} + const MCExpr *MCStreamer::BuildSymbolDiff(MCContext &Context, const MCSymbol *A, const MCSymbol *B) { @@ -92,8 +103,8 @@ void MCStreamer::EmitIntValue(uint64_t Value, unsigned Size, /// EmitULEB128Value - Special case of EmitULEB128Value that avoids the /// client having to pass in a MCExpr for constant integers. -void MCStreamer::EmitULEB128IntValue(uint64_t Value, unsigned AddrSpace, - unsigned Padding) { +void MCStreamer::EmitULEB128IntValue(uint64_t Value, unsigned Padding, + unsigned AddrSpace) { SmallString<128> Tmp; raw_svector_ostream OSE(Tmp); encodeULEB128(Value, OSE, Padding); @@ -146,8 +157,8 @@ void MCStreamer::EmitFill(uint64_t NumBytes, uint8_t FillValue, bool MCStreamer::EmitDwarfFileDirective(unsigned FileNo, StringRef Directory, - StringRef Filename) { - return getContext().GetDwarfFile(Directory, Filename, FileNo) == 0; + StringRef Filename, unsigned CUID) { + return getContext().GetDwarfFile(Directory, Filename, FileNo, CUID) == 0; } void MCStreamer::EmitDwarfLocDirective(unsigned FileNo, unsigned Line, @@ -182,6 +193,13 @@ void MCStreamer::EmitLabel(MCSymbol *Symbol) { LastSymbol = Symbol; } +void MCStreamer::EmitDebugLabel(MCSymbol *Symbol) { + assert(!Symbol->isVariable() && "Cannot emit a variable symbol!"); + assert(getCurrentSection() && "Cannot emit before setting section!"); + Symbol->setSection(*getCurrentSection()); + LastSymbol = Symbol; +} + void MCStreamer::EmitCompactUnwindEncoding(uint32_t CompactUnwindEncoding) { EnsureValidFrame(); MCDwarfFrameInfo *CurFrame = getCurrentFrameInfo(); @@ -602,3 +620,8 @@ void MCStreamer::Finish() { FinishImpl(); } + +MCSymbolData &MCStreamer::getOrCreateSymbolData(MCSymbol *Symbol) { + report_fatal_error("Not supported!"); + return *(static_cast<MCSymbolData*> (NULL)); +} diff --git a/lib/MC/MCTargetAsmLexer.cpp b/lib/MC/MCTargetAsmLexer.cpp deleted file mode 100644 index c01c914cec..0000000000 --- a/lib/MC/MCTargetAsmLexer.cpp +++ /dev/null @@ -1,16 +0,0 @@ -//===-- llvm/MC/MCTargetAsmLexer.cpp - Target Assembly Lexer --------------===// -// -// The LLVM Compiler Infrastructure -// -// This file is distributed under the University of Illinois Open Source -// License. See LICENSE.TXT for details. -// -//===----------------------------------------------------------------------===// - -#include "llvm/MC/MCTargetAsmLexer.h" -using namespace llvm; - -MCTargetAsmLexer::MCTargetAsmLexer(const Target &T) - : TheTarget(T), Lexer(NULL) { -} -MCTargetAsmLexer::~MCTargetAsmLexer() {} diff --git a/lib/MC/MachObjectWriter.cpp b/lib/MC/MachObjectWriter.cpp index cc8d2fb477..a5ba3c3653 100644 --- a/lib/MC/MachObjectWriter.cpp +++ b/lib/MC/MachObjectWriter.cpp @@ -27,6 +27,16 @@ using namespace llvm; using namespace llvm::object; +void MachObjectWriter::reset() { + Relocations.clear(); + IndirectSymBase.clear(); + StringTable.clear(); + LocalSymbolData.clear(); + ExternalSymbolData.clear(); + UndefinedSymbolData.clear(); + MCObjectWriter::reset(); +} + bool MachObjectWriter:: doesSymbolRequireExternRelocation(const MCSymbolData *SD) { // Undefined symbols are always extern. @@ -366,6 +376,39 @@ void MachObjectWriter::WriteLinkeditLoadCommand(uint32_t Type, assert(OS.tell() - Start == macho::LinkeditLoadCommandSize); } +static unsigned ComputeLinkerOptionsLoadCommandSize( + const std::vector<std::string> &Options, bool is64Bit) +{ + unsigned Size = sizeof(macho::LinkerOptionsLoadCommand); + for (unsigned i = 0, e = Options.size(); i != e; ++i) + Size += Options[i].size() + 1; + return RoundUpToAlignment(Size, is64Bit ? 8 : 4); +} + +void MachObjectWriter::WriteLinkerOptionsLoadCommand( + const std::vector<std::string> &Options) +{ + unsigned Size = ComputeLinkerOptionsLoadCommandSize(Options, is64Bit()); + uint64_t Start = OS.tell(); + (void) Start; + + Write32(macho::LCT_LinkerOptions); + Write32(Size); + Write32(Options.size()); + uint64_t BytesWritten = sizeof(macho::LinkerOptionsLoadCommand); + for (unsigned i = 0, e = Options.size(); i != e; ++i) { + // Write each string, including the null byte. + const std::string &Option = Options[i]; + WriteBytes(Option.c_str(), Option.size() + 1); + BytesWritten += Option.size() + 1; + } + + // Pad to a multiple of the pointer size. + WriteBytes("", OffsetToAlignment(BytesWritten, is64Bit() ? 8 : 4)); + + assert(OS.tell() - Start == Size); +} + void MachObjectWriter::RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout, @@ -683,6 +726,13 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, macho::SegmentLoadCommand64Size + NumSections * macho::Section64Size : macho::SegmentLoadCommand32Size + NumSections * macho::Section32Size; + // Add the data-in-code load command size, if used. + unsigned NumDataRegions = Asm.getDataRegions().size(); + if (NumDataRegions) { + ++NumLoadCommands; + LoadCommandsSize += macho::LinkeditLoadCommandSize; + } + // Add the symbol table load command sizes, if used. unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() + UndefinedSymbolData.size(); @@ -692,13 +742,15 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, macho::DysymtabLoadCommandSize); } - // Add the data-in-code load command size, if used. - unsigned NumDataRegions = Asm.getDataRegions().size(); - if (NumDataRegions) { + // Add the linker option load commands sizes. + const std::vector<std::vector<std::string> > &LinkerOptions = + Asm.getLinkerOptions(); + for (unsigned i = 0, e = LinkerOptions.size(); i != e; ++i) { ++NumLoadCommands; - LoadCommandsSize += macho::LinkeditLoadCommandSize; + LoadCommandsSize += ComputeLinkerOptionsLoadCommandSize(LinkerOptions[i], + is64Bit()); } - + // Compute the total size of the section data, as well as its file size and vm // size. uint64_t SectionDataStart = (is64Bit() ? macho::Header64Size : @@ -789,6 +841,11 @@ void MachObjectWriter::WriteObject(MCAssembler &Asm, IndirectSymbolOffset, NumIndirectSymbols); } + // Write the linker options load commands. + for (unsigned i = 0, e = LinkerOptions.size(); i != e; ++i) { + WriteLinkerOptionsLoadCommand(LinkerOptions[i]); + } + // Write the actual section data. for (MCAssembler::const_iterator it = Asm.begin(), ie = Asm.end(); it != ie; ++it) { diff --git a/lib/MC/WinCOFFObjectWriter.cpp b/lib/MC/WinCOFFObjectWriter.cpp index 01860c5d7f..6dffed73df 100644 --- a/lib/MC/WinCOFFObjectWriter.cpp +++ b/lib/MC/WinCOFFObjectWriter.cpp @@ -36,7 +36,7 @@ using namespace llvm; namespace { -typedef llvm::SmallString<COFF::NameSize> name; +typedef SmallString<COFF::NameSize> name; enum AuxiliaryType { ATFunctionDefinition, @@ -58,7 +58,7 @@ class COFFSymbol { public: COFF::symbol Data; - typedef llvm::SmallVector<AuxSymbol, 1> AuxiliarySymbols; + typedef SmallVector<AuxSymbol, 1> AuxiliarySymbols; name Name; int Index; @@ -69,7 +69,7 @@ public: MCSymbolData const *MCData; - COFFSymbol(llvm::StringRef name); + COFFSymbol(StringRef name); size_t size() const; void set_name_offset(uint32_t Offset); @@ -97,13 +97,13 @@ public: COFFSymbol *Symbol; relocations Relocations; - COFFSection(llvm::StringRef name); + COFFSection(StringRef name); static size_t size(); }; // This class holds the COFF string table. class StringTable { - typedef llvm::StringMap<size_t> map; + typedef StringMap<size_t> map; map Map; void update_length(); @@ -112,7 +112,7 @@ public: StringTable(); size_t size() const; - size_t insert(llvm::StringRef String); + size_t insert(StringRef String); }; class WinCOFFObjectWriter : public MCObjectWriter { @@ -144,10 +144,12 @@ public: COFFSection *createSection(StringRef Name); template <typename object_t, typename list_t> - object_t *createCOFFEntity(llvm::StringRef Name, list_t &List); + object_t *createCOFFEntity(StringRef Name, list_t &List); void DefineSection(MCSectionData const &SectionData); - void DefineSymbol(MCSymbolData const &SymbolData, MCAssembler &Assembler); + void DefineSymbol(MCSymbol const &Symbol, + MCSymbolData const &SymbolData, + MCAssembler &Assembler); void MakeSymbolReal(COFFSymbol &S, size_t Index); void MakeSectionReal(COFFSection &S, size_t Number); @@ -202,7 +204,7 @@ static inline void write_uint8_le(void *Data, uint8_t const &Value) { //------------------------------------------------------------------------------ // Symbol class implementation -COFFSymbol::COFFSymbol(llvm::StringRef name) +COFFSymbol::COFFSymbol(StringRef name) : Name(name.begin(), name.end()) , Other(NULL) , Section(NULL) @@ -254,7 +256,7 @@ bool COFFSymbol::should_keep() const { //------------------------------------------------------------------------------ // Section class implementation -COFFSection::COFFSection(llvm::StringRef name) +COFFSection::COFFSection(StringRef name) : Name(name) , MCData(NULL) , Symbol(NULL) { @@ -287,7 +289,7 @@ size_t StringTable::size() const { /// Add String to the table iff it is not already there. /// @returns the index into the string table where the string is now located. -size_t StringTable::insert(llvm::StringRef String) { +size_t StringTable::insert(StringRef String) { map::iterator i = Map.find(String); if (i != Map.end()) @@ -341,14 +343,14 @@ COFFSymbol *WinCOFFObjectWriter::GetOrCreateCOFFSymbol(const MCSymbol * Symbol){ return RetSymbol; } -COFFSection *WinCOFFObjectWriter::createSection(llvm::StringRef Name) { +COFFSection *WinCOFFObjectWriter::createSection(StringRef Name) { return createCOFFEntity<COFFSection>(Name, Sections); } /// A template used to lookup or create a symbol/section, and initialize it if /// needed. template <typename object_t, typename list_t> -object_t *WinCOFFObjectWriter::createCOFFEntity(llvm::StringRef Name, +object_t *WinCOFFObjectWriter::createCOFFEntity(StringRef Name, list_t &List) { object_t *Object = new object_t(Name); @@ -408,9 +410,10 @@ void WinCOFFObjectWriter::DefineSection(MCSectionData const &SectionData) { /// This function takes a section data object from the assembler /// and creates the associated COFF symbol staging object. -void WinCOFFObjectWriter::DefineSymbol(MCSymbolData const &SymbolData, +void WinCOFFObjectWriter::DefineSymbol(MCSymbol const &Symbol, + MCSymbolData const &SymbolData, MCAssembler &Assembler) { - COFFSymbol *coff_symbol = GetOrCreateCOFFSymbol(&SymbolData.getSymbol()); + COFFSymbol *coff_symbol = GetOrCreateCOFFSymbol(&Symbol); coff_symbol->Data.Type = (SymbolData.getFlags() & 0x0000FFFF) >> 0; coff_symbol->Data.StorageClass = (SymbolData.getFlags() & 0x00FF0000) >> 16; @@ -418,20 +421,17 @@ void WinCOFFObjectWriter::DefineSymbol(MCSymbolData const &SymbolData, if (SymbolData.getFlags() & COFF::SF_WeakExternal) { coff_symbol->Data.StorageClass = COFF::IMAGE_SYM_CLASS_WEAK_EXTERNAL; - if (SymbolData.getSymbol().isVariable()) { + if (Symbol.isVariable()) { coff_symbol->Data.StorageClass = COFF::IMAGE_SYM_CLASS_WEAK_EXTERNAL; - const MCExpr *Value = SymbolData.getSymbol().getVariableValue(); // FIXME: This assert message isn't very good. - assert(Value->getKind() == MCExpr::SymbolRef && + assert(Symbol.getVariableValue()->getKind() == MCExpr::SymbolRef && "Value must be a SymbolRef!"); - const MCSymbolRefExpr *SymbolRef = - static_cast<const MCSymbolRefExpr *>(Value); - coff_symbol->Other = GetOrCreateCOFFSymbol(&SymbolRef->getSymbol()); + coff_symbol->Other = GetOrCreateCOFFSymbol(&Symbol); } else { std::string WeakName = std::string(".weak.") - + SymbolData.getSymbol().getName().str() + + Symbol.getName().str() + ".default"; COFFSymbol *WeakDefault = createSymbol(WeakName); WeakDefault->Data.SectionNumber = COFF::IMAGE_SYM_ABSOLUTE; @@ -464,7 +464,7 @@ void WinCOFFObjectWriter::DefineSymbol(MCSymbolData const &SymbolData, // Bind internal COFF symbol to MC symbol. coff_symbol->MCData = &SymbolData; - SymbolMap[&SymbolData.getSymbol()] = coff_symbol; + SymbolMap[&Symbol] = coff_symbol; } /// making a section real involves assigned it a number and putting @@ -619,8 +619,11 @@ void WinCOFFObjectWriter::ExecutePostLayoutBinding(MCAssembler &Asm, for (MCAssembler::const_symbol_iterator i = Asm.symbol_begin(), e = Asm.symbol_end(); i != e; i++) { - if (ExportSymbol(*i, Asm)) - DefineSymbol(*i, Asm); + if (ExportSymbol(*i, Asm)) { + const MCSymbol &Alias = i->getSymbol(); + const MCSymbol &Symbol = Alias.AliasedSymbol(); + DefineSymbol(Alias, Asm.getSymbolData(Symbol), Asm); + } } } diff --git a/lib/MC/WinCOFFStreamer.cpp b/lib/MC/WinCOFFStreamer.cpp index 359b388618..75f343c421 100644 --- a/lib/MC/WinCOFFStreamer.cpp +++ b/lib/MC/WinCOFFStreamer.cpp @@ -50,10 +50,11 @@ public: // MCStreamer interface virtual void InitSections(); + virtual void InitToTextSection(); virtual void EmitLabel(MCSymbol *Symbol); + virtual void EmitDebugLabel(MCSymbol *Symbol); virtual void EmitAssemblerFlag(MCAssemblerFlag Flag); virtual void EmitThumbFunc(MCSymbol *Func); - virtual void EmitAssignment(MCSymbol *Symbol, const MCExpr *Value); virtual void EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute); virtual void EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue); virtual void BeginCOFFSymbolDef(MCSymbol const *Symbol); @@ -71,16 +72,29 @@ public: virtual void EmitTBSSSymbol(const MCSection *Section, MCSymbol *Symbol, uint64_t Size, unsigned ByteAlignment); virtual void EmitFileDirective(StringRef Filename); - virtual void EmitInstruction(const MCInst &Instruction); virtual void EmitWin64EHHandlerData(); virtual void FinishImpl(); -private: - virtual void EmitInstToFragment(const MCInst &Inst) { - llvm_unreachable("Not used by WinCOFF."); + static bool classof(const MCStreamer *S) { + return S->getKind() == SK_WinCOFFStreamer; } + +private: virtual void EmitInstToData(const MCInst &Inst) { - llvm_unreachable("Not used by WinCOFF."); + MCDataFragment *DF = getOrCreateDataFragment(); + + SmallVector<MCFixup, 4> Fixups; + SmallString<256> Code; + raw_svector_ostream VecOS(Code); + getAssembler().getEmitter().EncodeInstruction(Inst, VecOS, Fixups); + VecOS.flush(); + + // Add the fixups and data. + for (unsigned i = 0, e = Fixups.size(); i != e; ++i) { + Fixups[i].setOffset(Fixups[i].getOffset() + DF->getContents().size()); + DF->getFixups().push_back(Fixups[i]); + } + DF->getContents().append(Code.begin(), Code.end()); } void SetSection(StringRef Section, @@ -115,17 +129,13 @@ private: SectionKind::getBSS()); EmitCodeAlignment(4, 0); } - }; } // end anonymous namespace. -WinCOFFStreamer::WinCOFFStreamer(MCContext &Context, - MCAsmBackend &MAB, - MCCodeEmitter &CE, - raw_ostream &OS) - : MCObjectStreamer(Context, MAB, OS, &CE) - , CurSymbol(NULL) { -} +WinCOFFStreamer::WinCOFFStreamer(MCContext &Context, MCAsmBackend &MAB, + MCCodeEmitter &CE, raw_ostream &OS) + : MCObjectStreamer(SK_WinCOFFStreamer, Context, MAB, OS, &CE), + CurSymbol(NULL) {} void WinCOFFStreamer::AddCommonSymbol(MCSymbol *Symbol, uint64_t Size, unsigned ByteAlignment, bool External) { @@ -164,6 +174,10 @@ void WinCOFFStreamer::AddCommonSymbol(MCSymbol *Symbol, uint64_t Size, // MCStreamer interface +void WinCOFFStreamer::InitToTextSection() { + SetSectionText(); +} + void WinCOFFStreamer::InitSections() { SetSectionText(); SetSectionData(); @@ -176,6 +190,9 @@ void WinCOFFStreamer::EmitLabel(MCSymbol *Symbol) { MCObjectStreamer::EmitLabel(Symbol); } +void WinCOFFStreamer::EmitDebugLabel(MCSymbol *Symbol) { + EmitLabel(Symbol); +} void WinCOFFStreamer::EmitAssemblerFlag(MCAssemblerFlag Flag) { llvm_unreachable("not implemented"); } @@ -184,44 +201,6 @@ void WinCOFFStreamer::EmitThumbFunc(MCSymbol *Func) { llvm_unreachable("not implemented"); } -void WinCOFFStreamer::EmitAssignment(MCSymbol *Symbol, const MCExpr *Value) { - assert((Symbol->isInSection() - ? Symbol->getSection().getVariant() == MCSection::SV_COFF - : true) && "Got non COFF section in the COFF backend!"); - // FIXME: This is all very ugly and depressing. What needs to happen here - // depends on quite a few things that are all part of relaxation, which we - // don't really even do. - - if (Value->getKind() != MCExpr::SymbolRef) { - MCObjectStreamer::EmitAssignment(Symbol, Value); - } else { - // FIXME: This is a horrible way to do this :(. This should really be - // handled after we are done with the MC* objects and immediately before - // writing out the object file when we know exactly what the symbol should - // look like in the coff symbol table. I'm not doing that now because the - // COFF object writer doesn't have a clearly defined separation between MC - // data structures, the object writers data structures, and the raw, POD, - // data structures that get written to disk. - - // Copy over the aliased data. - MCSymbolData &SD = getAssembler().getOrCreateSymbolData(*Symbol); - const MCSymbolData &RealSD = getAssembler().getOrCreateSymbolData( - dyn_cast<const MCSymbolRefExpr>(Value)->getSymbol()); - - // FIXME: This is particularly nasty because it breaks as soon as any data - // members of MCSymbolData change. - SD.CommonAlign = RealSD.CommonAlign; - SD.CommonSize = RealSD.CommonSize; - SD.Flags = RealSD.Flags; - SD.Fragment = RealSD.Fragment; - SD.Index = RealSD.Index; - SD.IsExternal = RealSD.IsExternal; - SD.IsPrivateExtern = RealSD.IsPrivateExtern; - SD.Offset = RealSD.Offset; - SD.SymbolSize = RealSD.SymbolSize; - } -} - void WinCOFFStreamer::EmitSymbolAttribute(MCSymbol *Symbol, MCSymbolAttr Attribute) { assert(Symbol && "Symbol must be non-null!"); @@ -330,22 +309,6 @@ void WinCOFFStreamer::EmitFileDirective(StringRef Filename) { // info will be a much large effort. } -void WinCOFFStreamer::EmitInstruction(const MCInst &Instruction) { - for (unsigned i = 0, e = Instruction.getNumOperands(); i != e; ++i) - if (Instruction.getOperand(i).isExpr()) - AddValueSymbols(Instruction.getOperand(i).getExpr()); - - getCurrentSectionData()->setHasInstructions(true); - - MCInstFragment *Fragment = - new MCInstFragment(Instruction, getCurrentSectionData()); - - raw_svector_ostream VecOS(Fragment->getContents()); - - getAssembler().getEmitter().EncodeInstruction(Instruction, VecOS, - Fragment->getFixups()); -} - void WinCOFFStreamer::EmitWin64EHHandlerData() { MCStreamer::EmitWin64EHHandlerData(); |