diff options
author | Stephen Hines <srhines@google.com> | 2014-05-29 02:49:00 -0700 |
---|---|---|
committer | Stephen Hines <srhines@google.com> | 2014-05-29 02:49:00 -0700 |
commit | dce4a407a24b04eebc6a376f8e62b41aaa7b071f (patch) | |
tree | dcebc53f2b182f145a2e659393bf9a0472cedf23 /lib/Target/X86/Disassembler | |
parent | 220b921aed042f9e520c26cffd8282a94c66c3d5 (diff) | |
download | external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.zip external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.tar.gz external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.tar.bz2 |
Update LLVM for 3.5 rebase (r209712).
Change-Id: I149556c940fb7dc92d075273c87ff584f400941f
Diffstat (limited to 'lib/Target/X86/Disassembler')
-rw-r--r-- | lib/Target/X86/Disassembler/Android.mk | 3 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/CMakeLists.txt | 2 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/Makefile | 4 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/X86Disassembler.cpp | 51 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/X86Disassembler.h | 17 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/X86DisassemblerDecoder.cpp (renamed from lib/Target/X86/Disassembler/X86DisassemblerDecoder.c) | 181 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/X86DisassemblerDecoder.h | 362 | ||||
-rw-r--r-- | lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h | 221 |
8 files changed, 371 insertions, 470 deletions
diff --git a/lib/Target/X86/Disassembler/Android.mk b/lib/Target/X86/Disassembler/Android.mk index 3984266..0b3b8a5 100644 --- a/lib/Target/X86/Disassembler/Android.mk +++ b/lib/Target/X86/Disassembler/Android.mk @@ -8,7 +8,8 @@ x86_disassembler_TBLGEN_TABLES := \ x86_disassembler_SRC_FILES := \ X86Disassembler.cpp \ - X86DisassemblerDecoder.c + X86DisassemblerDecoder.cpp + # For the device # ===================================================== diff --git a/lib/Target/X86/Disassembler/CMakeLists.txt b/lib/Target/X86/Disassembler/CMakeLists.txt index deed115..4370282 100644 --- a/lib/Target/X86/Disassembler/CMakeLists.txt +++ b/lib/Target/X86/Disassembler/CMakeLists.txt @@ -1,4 +1,4 @@ add_llvm_library(LLVMX86Disassembler X86Disassembler.cpp - X86DisassemblerDecoder.c + X86DisassemblerDecoder.cpp ) diff --git a/lib/Target/X86/Disassembler/Makefile b/lib/Target/X86/Disassembler/Makefile index 8669fd8..51e7b82 100644 --- a/lib/Target/X86/Disassembler/Makefile +++ b/lib/Target/X86/Disassembler/Makefile @@ -10,7 +10,9 @@ LEVEL = ../../../.. LIBRARYNAME = LLVMX86Disassembler -# Hack: we need to include 'main' x86 target directory to grab private headers +# Hack: we need to include 'main' x86 target directory to grab private headers. CPP.Flags += -I$(PROJ_OBJ_DIR)/.. -I$(PROJ_SRC_DIR)/.. include $(LEVEL)/Makefile.common + +.PHONY: $(PROJ_SRC_DIR)/X86DisassemblerDecoder.c diff --git a/lib/Target/X86/Disassembler/X86Disassembler.cpp b/lib/Target/X86/Disassembler/X86Disassembler.cpp index d5759cd..c366725 100644 --- a/lib/Target/X86/Disassembler/X86Disassembler.cpp +++ b/lib/Target/X86/Disassembler/X86Disassembler.cpp @@ -27,6 +27,11 @@ #include "llvm/Support/TargetRegistry.h" #include "llvm/Support/raw_ostream.h" +using namespace llvm; +using namespace llvm::X86Disassembler; + +#define DEBUG_TYPE "x86-disassembler" + #define GET_REGINFO_ENUM #include "X86GenRegisterInfo.inc" #define GET_INSTRINFO_ENUM @@ -34,21 +39,18 @@ #define GET_SUBTARGETINFO_ENUM #include "X86GenSubtargetInfo.inc" -using namespace llvm; -using namespace llvm::X86Disassembler; - -void x86DisassemblerDebug(const char *file, - unsigned line, - const char *s) { +void llvm::X86Disassembler::Debug(const char *file, unsigned line, + const char *s) { dbgs() << file << ":" << line << ": " << s; } -const char *x86DisassemblerGetInstrName(unsigned Opcode, const void *mii) { +const char *llvm::X86Disassembler::GetInstrName(unsigned Opcode, + const void *mii) { const MCInstrInfo *MII = static_cast<const MCInstrInfo *>(mii); return MII->getName(Opcode); } -#define debug(s) DEBUG(x86DisassemblerDebug(__FILE__, __LINE__, s)); +#define debug(s) DEBUG(Debug(__FILE__, __LINE__, s)); namespace llvm { @@ -74,9 +76,11 @@ static bool translateInstruction(MCInst &target, InternalInstruction &source, const MCDisassembler *Dis); -X86GenericDisassembler::X86GenericDisassembler(const MCSubtargetInfo &STI, - const MCInstrInfo *MII) - : MCDisassembler(STI), MII(MII) { +X86GenericDisassembler::X86GenericDisassembler( + const MCSubtargetInfo &STI, + MCContext &Ctx, + std::unique_ptr<const MCInstrInfo> MII) + : MCDisassembler(STI, Ctx), MII(std::move(MII)) { switch (STI.getFeatureBits() & (X86::Mode16Bit | X86::Mode32Bit | X86::Mode64Bit)) { case X86::Mode16Bit: @@ -93,10 +97,6 @@ X86GenericDisassembler::X86GenericDisassembler(const MCSubtargetInfo &STI, } } -X86GenericDisassembler::~X86GenericDisassembler() { - delete MII; -} - /// regionReader - a callback function that wraps the readByte method from /// MemoryObject. /// @@ -140,14 +140,14 @@ X86GenericDisassembler::getInstruction(MCInst &instr, dlog_t loggerFn = logger; if (&vStream == &nulls()) - loggerFn = 0; // Disable logging completely if it's going to nulls(). + loggerFn = nullptr; // Disable logging completely if it's going to nulls(). int ret = decodeInstruction(&internalInstr, regionReader, (const void*)®ion, loggerFn, (void*)&vStream, - (const void*)MII, + (const void*)MII.get(), address, fMode); @@ -319,7 +319,7 @@ static void translateImmediate(MCInst &mcInst, uint64_t immediate, } // By default sign-extend all X86 immediates based on their encoding. else if (type == TYPE_IMM8 || type == TYPE_IMM16 || type == TYPE_IMM32 || - type == TYPE_IMM64) { + type == TYPE_IMM64 || type == TYPE_IMMv) { uint32_t Opcode = mcInst.getOpcode(); switch (operand.encoding) { default: @@ -787,13 +787,11 @@ static bool translateInstruction(MCInst &mcInst, mcInst.setOpcode(X86::XACQUIRE_PREFIX); } - int index; - insn.numImmediatesTranslated = 0; - for (index = 0; index < X86_MAX_OPERANDS; ++index) { - if (insn.operands[index].encoding != ENCODING_NONE) { - if (translateOperand(mcInst, insn.operands[index], insn, Dis)) { + for (const auto &Op : insn.operands) { + if (Op.encoding != ENCODING_NONE) { + if (translateOperand(mcInst, Op, insn, Dis)) { return true; } } @@ -803,9 +801,10 @@ static bool translateInstruction(MCInst &mcInst, } static MCDisassembler *createX86Disassembler(const Target &T, - const MCSubtargetInfo &STI) { - return new X86Disassembler::X86GenericDisassembler(STI, - T.createMCInstrInfo()); + const MCSubtargetInfo &STI, + MCContext &Ctx) { + std::unique_ptr<const MCInstrInfo> MII(T.createMCInstrInfo()); + return new X86Disassembler::X86GenericDisassembler(STI, Ctx, std::move(MII)); } extern "C" void LLVMInitializeX86Disassembler() { diff --git a/lib/Target/X86/Disassembler/X86Disassembler.h b/lib/Target/X86/Disassembler/X86Disassembler.h index 4e6e297..4dc7c29 100644 --- a/lib/Target/X86/Disassembler/X86Disassembler.h +++ b/lib/Target/X86/Disassembler/X86Disassembler.h @@ -74,17 +74,7 @@ #ifndef X86DISASSEMBLER_H #define X86DISASSEMBLER_H -#define INSTRUCTION_SPECIFIER_FIELDS \ - uint16_t operands; - -#define INSTRUCTION_IDS \ - uint16_t instructionIDs; - #include "X86DisassemblerDecoderCommon.h" - -#undef INSTRUCTION_SPECIFIER_FIELDS -#undef INSTRUCTION_IDS - #include "llvm/MC/MCDisassembler.h" namespace llvm { @@ -101,13 +91,12 @@ namespace X86Disassembler { /// All each platform class should have to do is subclass the constructor, and /// provide a different disassemblerMode value. class X86GenericDisassembler : public MCDisassembler { - const MCInstrInfo *MII; + std::unique_ptr<const MCInstrInfo> MII; public: /// Constructor - Initializes the disassembler. /// - X86GenericDisassembler(const MCSubtargetInfo &STI, const MCInstrInfo *MII); -private: - ~X86GenericDisassembler(); + X86GenericDisassembler(const MCSubtargetInfo &STI, MCContext &Ctx, + std::unique_ptr<const MCInstrInfo> MII); public: /// getInstruction - See MCDisassembler. diff --git a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.cpp index 0801c96..804606d 100644 --- a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.c +++ b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.cpp @@ -1,17 +1,17 @@ -/*===-- X86DisassemblerDecoder.c - Disassembler decoder ------------*- C -*-===* - * - * The LLVM Compiler Infrastructure - * - * This file is distributed under the University of Illinois Open Source - * License. See LICENSE.TXT for details. - * - *===----------------------------------------------------------------------===* - * - * This file is part of the X86 Disassembler. - * It contains the implementation of the instruction decoder. - * Documentation for the disassembler can be found in X86Disassembler.h. - * - *===----------------------------------------------------------------------===*/ +//===-- X86DisassemblerDecoder.c - Disassembler decoder -------------------===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file is part of the X86 Disassembler. +// It contains the implementation of the instruction decoder. +// Documentation for the disassembler can be found in X86Disassembler.h. +// +//===----------------------------------------------------------------------===// #include <stdarg.h> /* for va_*() */ #include <stdio.h> /* for vsnprintf() */ @@ -20,13 +20,35 @@ #include "X86DisassemblerDecoder.h" -#include "X86GenDisassemblerTables.inc" +using namespace llvm::X86Disassembler; + +/// Specifies whether a ModR/M byte is needed and (if so) which +/// instruction each possible value of the ModR/M byte corresponds to. Once +/// this information is known, we have narrowed down to a single instruction. +struct ModRMDecision { + uint8_t modrm_type; + uint16_t instructionIDs; +}; + +/// Specifies which set of ModR/M->instruction tables to look at +/// given a particular opcode. +struct OpcodeDecision { + ModRMDecision modRMDecisions[256]; +}; + +/// Specifies which opcode->instruction tables to look at given +/// a particular context (set of attributes). Since there are many possible +/// contexts, the decoder first uses CONTEXTS_SYM to determine which context +/// applies given a specific set of attributes. Hence there are only IC_max +/// entries in this table, rather than 2^(ATTR_max). +struct ContextDecision { + OpcodeDecision opcodeDecisions[IC_max]; +}; -#define TRUE 1 -#define FALSE 0 +#include "X86GenDisassemblerTables.inc" #ifndef NDEBUG -#define debug(s) do { x86DisassemblerDebug(__FILE__, __LINE__, s); } while (0) +#define debug(s) do { Debug(__FILE__, __LINE__, s); } while (0) #else #define debug(s) do { } while (0) #endif @@ -41,7 +63,7 @@ * an instruction with these attributes. */ static InstructionContext contextForAttrs(uint16_t attrMask) { - return CONTEXTS_SYM[attrMask]; + return static_cast<InstructionContext>(CONTEXTS_SYM[attrMask]); } /* @@ -53,12 +75,12 @@ static InstructionContext contextForAttrs(uint16_t attrMask) { * contextForAttrs. * @param opcode - The last byte of the instruction's opcode, not counting * ModR/M extensions and escapes. - * @return - TRUE if the ModR/M byte is required, FALSE otherwise. + * @return - true if the ModR/M byte is required, false otherwise. */ static int modRMRequired(OpcodeType type, InstructionContext insnContext, uint16_t opcode) { - const struct ContextDecision* decision = 0; + const struct ContextDecision* decision = nullptr; switch (type) { case ONEBYTE: @@ -102,7 +124,7 @@ static InstrUID decode(OpcodeType type, InstructionContext insnContext, uint8_t opcode, uint8_t modRM) { - const struct ModRMDecision* dec = 0; + const struct ModRMDecision* dec = nullptr; switch (type) { case ONEBYTE: @@ -284,15 +306,15 @@ static void setPrefixPresent(struct InternalInstruction* insn, * @param location - The location to query. * @return - Whether the prefix is at that location. */ -static BOOL isPrefixAtLocation(struct InternalInstruction* insn, +static bool isPrefixAtLocation(struct InternalInstruction* insn, uint8_t prefix, uint64_t location) { if (insn->prefixPresent[prefix] == 1 && insn->prefixLocations[prefix] == location) - return TRUE; + return true; else - return FALSE; + return false; } /* @@ -305,14 +327,14 @@ static BOOL isPrefixAtLocation(struct InternalInstruction* insn, * bytes, and no prefixes conflicted; nonzero otherwise. */ static int readPrefixes(struct InternalInstruction* insn) { - BOOL isPrefix = TRUE; - BOOL prefixGroups[4] = { FALSE }; + bool isPrefix = true; + bool prefixGroups[4] = { false }; uint64_t prefixLocation; uint8_t byte = 0; uint8_t nextByte; - BOOL hasAdSize = FALSE; - BOOL hasOpSize = FALSE; + bool hasAdSize = false; + bool hasOpSize = false; dbgprintf(insn, "readPrefixes()"); @@ -344,7 +366,7 @@ static int readPrefixes(struct InternalInstruction* insn) { if ((byte == 0xf2 || byte == 0xf3) && ((nextByte == 0xf0) | ((nextByte & 0xfe) == 0x86 || (nextByte & 0xf8) == 0x90))) - insn->xAcquireRelease = TRUE; + insn->xAcquireRelease = true; /* * Also if the byte is 0xf3, and the following condition is met: * - it is followed by a "mov mem, reg" (opcode 0x88/0x89) or @@ -354,7 +376,7 @@ static int readPrefixes(struct InternalInstruction* insn) { if (byte == 0xf3 && (nextByte == 0x88 || nextByte == 0x89 || nextByte == 0xc6 || nextByte == 0xc7)) - insn->xAcquireRelease = TRUE; + insn->xAcquireRelease = true; if (insn->mode == MODE_64BIT && (nextByte & 0xf0) == 0x40) { if (consumeByte(insn, &nextByte)) return -1; @@ -372,7 +394,7 @@ static int readPrefixes(struct InternalInstruction* insn) { case 0xf3: /* REP or REPE/REPZ */ if (prefixGroups[0]) dbgprintf(insn, "Redundant Group 1 prefix"); - prefixGroups[0] = TRUE; + prefixGroups[0] = true; setPrefixPresent(insn, byte, prefixLocation); break; case 0x2e: /* CS segment override -OR- Branch not taken */ @@ -406,25 +428,25 @@ static int readPrefixes(struct InternalInstruction* insn) { } if (prefixGroups[1]) dbgprintf(insn, "Redundant Group 2 prefix"); - prefixGroups[1] = TRUE; + prefixGroups[1] = true; setPrefixPresent(insn, byte, prefixLocation); break; case 0x66: /* Operand-size override */ if (prefixGroups[2]) dbgprintf(insn, "Redundant Group 3 prefix"); - prefixGroups[2] = TRUE; - hasOpSize = TRUE; + prefixGroups[2] = true; + hasOpSize = true; setPrefixPresent(insn, byte, prefixLocation); break; case 0x67: /* Address-size override */ if (prefixGroups[3]) dbgprintf(insn, "Redundant Group 4 prefix"); - prefixGroups[3] = TRUE; - hasAdSize = TRUE; + prefixGroups[3] = true; + hasAdSize = true; setPrefixPresent(insn, byte, prefixLocation); break; default: /* Not a prefix byte */ - isPrefix = FALSE; + isPrefix = false; break; } @@ -549,7 +571,7 @@ static int readPrefixes(struct InternalInstruction* insn) { default: break; case VEX_PREFIX_66: - hasOpSize = TRUE; + hasOpSize = true; break; } @@ -595,7 +617,7 @@ static int readPrefixes(struct InternalInstruction* insn) { default: break; case VEX_PREFIX_66: - hasOpSize = TRUE; + hasOpSize = true; break; } @@ -790,11 +812,9 @@ static int readModRM(struct InternalInstruction* insn); static int getIDWithAttrMask(uint16_t* instructionID, struct InternalInstruction* insn, uint16_t attrMask) { - BOOL hasModRMExtension; - - uint16_t instructionClass; + bool hasModRMExtension; - instructionClass = contextForAttrs(attrMask); + InstructionContext instructionClass = contextForAttrs(attrMask); hasModRMExtension = modRMRequired(insn->opcodeType, instructionClass, @@ -825,14 +845,14 @@ static int getIDWithAttrMask(uint16_t* instructionID, * @param orig - The instruction that is not 16-bit * @param equiv - The instruction that is 16-bit */ -static BOOL is16BitEquivalent(const char* orig, const char* equiv) { +static bool is16BitEquivalent(const char* orig, const char* equiv) { off_t i; for (i = 0;; i++) { if (orig[i] == '\0' && equiv[i] == '\0') - return TRUE; + return true; if (orig[i] == '\0' || equiv[i] == '\0') - return FALSE; + return false; if (orig[i] != equiv[i]) { if ((orig[i] == 'Q' || orig[i] == 'L') && equiv[i] == 'W') continue; @@ -840,7 +860,7 @@ static BOOL is16BitEquivalent(const char* orig, const char* equiv) { continue; if ((orig[i] == '4' || orig[i] == '2') && equiv[i] == '6') continue; - return FALSE; + return false; } } } @@ -1011,9 +1031,8 @@ static int getID(struct InternalInstruction* insn, const void *miiArg) { return 0; } - specName = x86DisassemblerGetInstrName(instructionID, miiArg); - specWithOpSizeName = - x86DisassemblerGetInstrName(instructionIDWithOpsize, miiArg); + specName = GetInstrName(instructionID, miiArg); + specWithOpSizeName = GetInstrName(instructionIDWithOpsize, miiArg); if (is16BitEquivalent(specName, specWithOpSizeName) && (insn->mode == MODE_16BIT) ^ insn->prefixPresent[0x66]) { @@ -1077,8 +1096,8 @@ static int getID(struct InternalInstruction* insn, const void *miiArg) { * @return - 0 if the SIB byte was successfully read; nonzero otherwise. */ static int readSIB(struct InternalInstruction* insn) { - SIBIndex sibIndexBase = 0; - SIBBase sibBaseBase = 0; + SIBIndex sibIndexBase = SIB_INDEX_NONE; + SIBBase sibBaseBase = SIB_BASE_NONE; uint8_t index, base; dbgprintf(insn, "readSIB()"); @@ -1086,7 +1105,7 @@ static int readSIB(struct InternalInstruction* insn) { if (insn->consumedSIB) return 0; - insn->consumedSIB = TRUE; + insn->consumedSIB = true; switch (insn->addressSize) { case 2: @@ -1184,12 +1203,12 @@ static int readDisplacement(struct InternalInstruction* insn) { if (insn->consumedDisplacement) return 0; - insn->consumedDisplacement = TRUE; + insn->consumedDisplacement = true; insn->displacementOffset = insn->readerCursor - insn->startLocation; switch (insn->eaDisplacement) { case EA_DISP_NONE: - insn->consumedDisplacement = FALSE; + insn->consumedDisplacement = false; break; case EA_DISP_8: if (consumeInt8(insn, &d8)) @@ -1208,7 +1227,7 @@ static int readDisplacement(struct InternalInstruction* insn) { break; } - insn->consumedDisplacement = TRUE; + insn->consumedDisplacement = true; return 0; } @@ -1229,7 +1248,7 @@ static int readModRM(struct InternalInstruction* insn) { if (consumeByte(insn, &insn->modRM)) return -1; - insn->consumedModRM = TRUE; + insn->consumedModRM = true; mod = modFromModRM(insn->modRM); rm = rmFromModRM(insn->modRM); @@ -1599,20 +1618,22 @@ static int readImmediate(struct InternalInstruction* insn, uint8_t size) { static int readVVVV(struct InternalInstruction* insn) { dbgprintf(insn, "readVVVV()"); + int vvvv; if (insn->vectorExtensionType == TYPE_EVEX) - insn->vvvv = vvvvFromEVEX3of4(insn->vectorExtensionPrefix[2]); + vvvv = vvvvFromEVEX3of4(insn->vectorExtensionPrefix[2]); else if (insn->vectorExtensionType == TYPE_VEX_3B) - insn->vvvv = vvvvFromVEX3of3(insn->vectorExtensionPrefix[2]); + vvvv = vvvvFromVEX3of3(insn->vectorExtensionPrefix[2]); else if (insn->vectorExtensionType == TYPE_VEX_2B) - insn->vvvv = vvvvFromVEX2of2(insn->vectorExtensionPrefix[1]); + vvvv = vvvvFromVEX2of2(insn->vectorExtensionPrefix[1]); else if (insn->vectorExtensionType == TYPE_XOP) - insn->vvvv = vvvvFromXOP3of3(insn->vectorExtensionPrefix[2]); + vvvv = vvvvFromXOP3of3(insn->vectorExtensionPrefix[2]); else return -1; if (insn->mode != MODE_64BIT) - insn->vvvv &= 0x7; + vvvv &= 0x7; + insn->vvvv = static_cast<Reg>(vvvv); return 0; } @@ -1629,7 +1650,8 @@ static int readMaskRegister(struct InternalInstruction* insn) { if (insn->vectorExtensionType != TYPE_EVEX) return -1; - insn->writemask = aaaFromEVEX4of4(insn->vectorExtensionPrefix[3]); + insn->writemask = + static_cast<Reg>(aaaFromEVEX4of4(insn->vectorExtensionPrefix[3])); return 0; } @@ -1641,7 +1663,6 @@ static int readMaskRegister(struct InternalInstruction* insn) { * @return - 0 if all operands could be read; nonzero otherwise. */ static int readOperands(struct InternalInstruction* insn) { - int index; int hasVVVV, needVVVV; int sawRegImm = 0; @@ -1652,8 +1673,8 @@ static int readOperands(struct InternalInstruction* insn) { hasVVVV = !readVVVV(insn); needVVVV = hasVVVV && (insn->vvvv != 0); - for (index = 0; index < X86_MAX_OPERANDS; ++index) { - switch (x86OperandSets[insn->spec->operands][index].encoding) { + for (const auto &Op : x86OperandSets[insn->spec->operands]) { + switch (Op.encoding) { case ENCODING_NONE: case ENCODING_SI: case ENCODING_DI: @@ -1662,7 +1683,7 @@ static int readOperands(struct InternalInstruction* insn) { case ENCODING_RM: if (readModRM(insn)) return -1; - if (fixupReg(insn, &x86OperandSets[insn->spec->operands][index])) + if (fixupReg(insn, &Op)) return -1; break; case ENCODING_CB: @@ -1684,14 +1705,14 @@ static int readOperands(struct InternalInstruction* insn) { } if (readImmediate(insn, 1)) return -1; - if (x86OperandSets[insn->spec->operands][index].type == TYPE_IMM3 && + if (Op.type == TYPE_IMM3 && insn->immediates[insn->numImmediatesConsumed - 1] > 7) return -1; - if (x86OperandSets[insn->spec->operands][index].type == TYPE_IMM5 && + if (Op.type == TYPE_IMM5 && insn->immediates[insn->numImmediatesConsumed - 1] > 31) return -1; - if (x86OperandSets[insn->spec->operands][index].type == TYPE_XMM128 || - x86OperandSets[insn->spec->operands][index].type == TYPE_XMM256) + if (Op.type == TYPE_XMM128 || + Op.type == TYPE_XMM256) sawRegImm = 1; break; case ENCODING_IW: @@ -1740,7 +1761,7 @@ static int readOperands(struct InternalInstruction* insn) { needVVVV = 0; /* Mark that we have found a VVVV operand. */ if (!hasVVVV) return -1; - if (fixupReg(insn, &x86OperandSets[insn->spec->operands][index])) + if (fixupReg(insn, &Op)) return -1; break; case ENCODING_WRITEMASK: @@ -1781,14 +1802,10 @@ static int readOperands(struct InternalInstruction* insn) { * @return - 0 if the instruction's memory could be read; nonzero if * not. */ -int decodeInstruction(struct InternalInstruction* insn, - byteReader_t reader, - const void* readerArg, - dlog_t logger, - void* loggerArg, - const void* miiArg, - uint64_t startLoc, - DisassemblerMode mode) { +int llvm::X86Disassembler::decodeInstruction( + struct InternalInstruction *insn, byteReader_t reader, + const void *readerArg, dlog_t logger, void *loggerArg, const void *miiArg, + uint64_t startLoc, DisassemblerMode mode) { memset(insn, 0, sizeof(struct InternalInstruction)); insn->reader = reader; @@ -1807,7 +1824,7 @@ int decodeInstruction(struct InternalInstruction* insn, readOperands(insn)) return -1; - insn->operands = &x86OperandSets[insn->spec->operands][0]; + insn->operands = x86OperandSets[insn->spec->operands]; insn->length = insn->readerCursor - insn->startLocation; diff --git a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.h b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.h index ac3b39d..8c45402 100644 --- a/lib/Target/X86/Disassembler/X86DisassemblerDecoder.h +++ b/lib/Target/X86/Disassembler/X86DisassemblerDecoder.h @@ -1,39 +1,28 @@ -/*===-- X86DisassemblerDecoderInternal.h - Disassembler decoder ---*- C -*-===* - * - * The LLVM Compiler Infrastructure - * - * This file is distributed under the University of Illinois Open Source - * License. See LICENSE.TXT for details. - * - *===----------------------------------------------------------------------===* - * - * This file is part of the X86 Disassembler. - * It contains the public interface of the instruction decoder. - * Documentation for the disassembler can be found in X86Disassembler.h. - * - *===----------------------------------------------------------------------===*/ +//===-- X86DisassemblerDecoderInternal.h - Disassembler decoder -*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file is part of the X86 Disassembler. +// It contains the public interface of the instruction decoder. +// Documentation for the disassembler can be found in X86Disassembler.h. +// +//===----------------------------------------------------------------------===// #ifndef X86DISASSEMBLERDECODER_H #define X86DISASSEMBLERDECODER_H -#ifdef __cplusplus -extern "C" { -#endif - -#define INSTRUCTION_SPECIFIER_FIELDS \ - uint16_t operands; - -#define INSTRUCTION_IDS \ - uint16_t instructionIDs; - #include "X86DisassemblerDecoderCommon.h" +#include "llvm/ADT/ArrayRef.h" -#undef INSTRUCTION_SPECIFIER_FIELDS -#undef INSTRUCTION_IDS +namespace llvm { +namespace X86Disassembler { -/* - * Accessor functions for various fields of an Intel instruction - */ +// Accessor functions for various fields of an Intel instruction #define modFromModRM(modRM) (((modRM) & 0xc0) >> 6) #define regFromModRM(modRM) (((modRM) & 0x38) >> 3) #define rmFromModRM(modRM) ((modRM) & 0x7) @@ -83,10 +72,7 @@ extern "C" { #define lFromXOP3of3(xop) (((xop) & 0x4) >> 2) #define ppFromXOP3of3(xop) ((xop) & 0x3) -/* - * These enums represent Intel registers for use by the decoder. - */ - +// These enums represent Intel registers for use by the decoder. #define REGS_8BIT \ ENTRY(AL) \ ENTRY(CL) \ @@ -392,13 +378,11 @@ extern "C" { REGS_CONTROL \ ENTRY(RIP) -/* - * EABase - All possible values of the base field for effective-address - * computations, a.k.a. the Mod and R/M fields of the ModR/M byte. We - * distinguish between bases (EA_BASE_*) and registers that just happen to be - * referred to when Mod == 0b11 (EA_REG_*). - */ -typedef enum { +/// \brief All possible values of the base field for effective-address +/// computations, a.k.a. the Mod and R/M fields of the ModR/M byte. +/// We distinguish between bases (EA_BASE_*) and registers that just happen +/// to be referred to when Mod == 0b11 (EA_REG_*). +enum EABase { EA_BASE_NONE, #define ENTRY(x) EA_BASE_##x, ALL_EA_BASES @@ -407,15 +391,13 @@ typedef enum { ALL_REGS #undef ENTRY EA_max -} EABase; - -/* - * SIBIndex - All possible values of the SIB index field. - * Borrows entries from ALL_EA_BASES with the special case that - * sib is synonymous with NONE. - * Vector SIB: index can be XMM or YMM. - */ -typedef enum { +}; + +/// \brief All possible values of the SIB index field. +/// borrows entries from ALL_EA_BASES with the special case that +/// sib is synonymous with NONE. +/// Vector SIB: index can be XMM or YMM. +enum SIBIndex { SIB_INDEX_NONE, #define ENTRY(x) SIB_INDEX_##x, ALL_EA_BASES @@ -424,23 +406,18 @@ typedef enum { REGS_ZMM #undef ENTRY SIB_INDEX_max -} SIBIndex; +}; -/* - * SIBBase - All possible values of the SIB base field. - */ -typedef enum { +/// \brief All possible values of the SIB base field. +enum SIBBase { SIB_BASE_NONE, #define ENTRY(x) SIB_BASE_##x, ALL_SIB_BASES #undef ENTRY SIB_BASE_max -} SIBBase; +}; -/* - * EADisplacement - Possible displacement types for effective-address - * computations. - */ +/// \brief Possible displacement types for effective-address computations. typedef enum { EA_DISP_NONE, EA_DISP_8, @@ -448,20 +425,16 @@ typedef enum { EA_DISP_32 } EADisplacement; -/* - * Reg - All possible values of the reg field in the ModR/M byte. - */ -typedef enum { +/// \brief All possible values of the reg field in the ModR/M byte. +enum Reg { #define ENTRY(x) MODRM_REG_##x, ALL_REGS #undef ENTRY MODRM_REG_max -} Reg; +}; -/* - * SegmentOverride - All possible segment overrides. - */ -typedef enum { +/// \brief All possible segment overrides. +enum SegmentOverride { SEG_OVERRIDE_NONE, SEG_OVERRIDE_CS, SEG_OVERRIDE_SS, @@ -470,235 +443,220 @@ typedef enum { SEG_OVERRIDE_FS, SEG_OVERRIDE_GS, SEG_OVERRIDE_max -} SegmentOverride; - -/* - * VEXLeadingOpcodeByte - Possible values for the VEX.m-mmmm field - */ +}; -typedef enum { +/// \brief Possible values for the VEX.m-mmmm field +enum VEXLeadingOpcodeByte { VEX_LOB_0F = 0x1, VEX_LOB_0F38 = 0x2, VEX_LOB_0F3A = 0x3 -} VEXLeadingOpcodeByte; +}; -typedef enum { +enum XOPMapSelect { XOP_MAP_SELECT_8 = 0x8, XOP_MAP_SELECT_9 = 0x9, XOP_MAP_SELECT_A = 0xA -} XOPMapSelect; - -/* - * VEXPrefixCode - Possible values for the VEX.pp/EVEX.pp field - */ +}; -typedef enum { +/// \brief Possible values for the VEX.pp/EVEX.pp field +enum VEXPrefixCode { VEX_PREFIX_NONE = 0x0, VEX_PREFIX_66 = 0x1, VEX_PREFIX_F3 = 0x2, VEX_PREFIX_F2 = 0x3 -} VEXPrefixCode; +}; -typedef enum { +enum VectorExtensionType { TYPE_NO_VEX_XOP = 0x0, TYPE_VEX_2B = 0x1, TYPE_VEX_3B = 0x2, TYPE_EVEX = 0x3, TYPE_XOP = 0x4 -} VectorExtensionType; - -typedef uint8_t BOOL; - -/* - * byteReader_t - Type for the byte reader that the consumer must provide to - * the decoder. Reads a single byte from the instruction's address space. - * @param arg - A baton that the consumer can associate with any internal - * state that it needs. - * @param byte - A pointer to a single byte in memory that should be set to - * contain the value at address. - * @param address - The address in the instruction's address space that should - * be read from. - * @return - -1 if the byte cannot be read for any reason; 0 otherwise. - */ -typedef int (*byteReader_t)(const void* arg, uint8_t* byte, uint64_t address); - -/* - * dlog_t - Type for the logging function that the consumer can provide to - * get debugging output from the decoder. - * @param arg - A baton that the consumer can associate with any internal - * state that it needs. - * @param log - A string that contains the message. Will be reused after - * the logger returns. - */ -typedef void (*dlog_t)(void* arg, const char *log); - -/* - * The x86 internal instruction, which is produced by the decoder. - */ +}; + +/// \brief Type for the byte reader that the consumer must provide to +/// the decoder. Reads a single byte from the instruction's address space. +/// \param arg A baton that the consumer can associate with any internal +/// state that it needs. +/// \param byte A pointer to a single byte in memory that should be set to +/// contain the value at address. +/// \param address The address in the instruction's address space that should +/// be read from. +/// \return -1 if the byte cannot be read for any reason; 0 otherwise. +typedef int (*byteReader_t)(const void *arg, uint8_t *byte, uint64_t address); + +/// \brief Type for the logging function that the consumer can provide to +/// get debugging output from the decoder. +/// \param arg A baton that the consumer can associate with any internal +/// state that it needs. +/// \param log A string that contains the message. Will be reused after +/// the logger returns. +typedef void (*dlog_t)(void *arg, const char *log); + +/// The specification for how to extract and interpret a full instruction and +/// its operands. +struct InstructionSpecifier { + uint16_t operands; +}; + +/// The x86 internal instruction, which is produced by the decoder. struct InternalInstruction { - /* Reader interface (C) */ + // Reader interface (C) byteReader_t reader; - /* Opaque value passed to the reader */ + // Opaque value passed to the reader const void* readerArg; - /* The address of the next byte to read via the reader */ + // The address of the next byte to read via the reader uint64_t readerCursor; - /* Logger interface (C) */ + // Logger interface (C) dlog_t dlog; - /* Opaque value passed to the logger */ + // Opaque value passed to the logger void* dlogArg; - /* General instruction information */ + // General instruction information - /* The mode to disassemble for (64-bit, protected, real) */ + // The mode to disassemble for (64-bit, protected, real) DisassemblerMode mode; - /* The start of the instruction, usable with the reader */ + // The start of the instruction, usable with the reader uint64_t startLocation; - /* The length of the instruction, in bytes */ + // The length of the instruction, in bytes size_t length; - /* Prefix state */ + // Prefix state - /* 1 if the prefix byte corresponding to the entry is present; 0 if not */ + // 1 if the prefix byte corresponding to the entry is present; 0 if not uint8_t prefixPresent[0x100]; - /* contains the location (for use with the reader) of the prefix byte */ + // contains the location (for use with the reader) of the prefix byte uint64_t prefixLocations[0x100]; - /* The value of the vector extension prefix(EVEX/VEX/XOP), if present */ + // The value of the vector extension prefix(EVEX/VEX/XOP), if present uint8_t vectorExtensionPrefix[4]; - /* The type of the vector extension prefix */ + // The type of the vector extension prefix VectorExtensionType vectorExtensionType; - /* The value of the REX prefix, if present */ + // The value of the REX prefix, if present uint8_t rexPrefix; - /* The location where a mandatory prefix would have to be (i.e., right before - the opcode, or right before the REX prefix if one is present) */ + // The location where a mandatory prefix would have to be (i.e., right before + // the opcode, or right before the REX prefix if one is present). uint64_t necessaryPrefixLocation; - /* The segment override type */ + // The segment override type SegmentOverride segmentOverride; - /* 1 if the prefix byte, 0xf2 or 0xf3 is xacquire or xrelease */ - BOOL xAcquireRelease; + // 1 if the prefix byte, 0xf2 or 0xf3 is xacquire or xrelease + bool xAcquireRelease; - /* Sizes of various critical pieces of data, in bytes */ + // Sizes of various critical pieces of data, in bytes uint8_t registerSize; uint8_t addressSize; uint8_t displacementSize; uint8_t immediateSize; - /* Offsets from the start of the instruction to the pieces of data, which is - needed to find relocation entries for adding symbolic operands */ + // Offsets from the start of the instruction to the pieces of data, which is + // needed to find relocation entries for adding symbolic operands. uint8_t displacementOffset; uint8_t immediateOffset; - /* opcode state */ + // opcode state - /* The last byte of the opcode, not counting any ModR/M extension */ + // The last byte of the opcode, not counting any ModR/M extension uint8_t opcode; - /* The ModR/M byte of the instruction, if it is an opcode extension */ + // The ModR/M byte of the instruction, if it is an opcode extension uint8_t modRMExtension; - /* decode state */ + // decode state - /* The type of opcode, used for indexing into the array of decode tables */ + // The type of opcode, used for indexing into the array of decode tables OpcodeType opcodeType; - /* The instruction ID, extracted from the decode table */ + // The instruction ID, extracted from the decode table uint16_t instructionID; - /* The specifier for the instruction, from the instruction info table */ - const struct InstructionSpecifier *spec; + // The specifier for the instruction, from the instruction info table + const InstructionSpecifier *spec; - /* state for additional bytes, consumed during operand decode. Pattern: - consumed___ indicates that the byte was already consumed and does not - need to be consumed again */ + // state for additional bytes, consumed during operand decode. Pattern: + // consumed___ indicates that the byte was already consumed and does not + // need to be consumed again. - /* The VEX.vvvv field, which contains a third register operand for some AVX - instructions */ + // The VEX.vvvv field, which contains a third register operand for some AVX + // instructions. Reg vvvv; - /* The writemask for AVX-512 instructions which is contained in EVEX.aaa */ + // The writemask for AVX-512 instructions which is contained in EVEX.aaa Reg writemask; - /* The ModR/M byte, which contains most register operands and some portion of - all memory operands */ - BOOL consumedModRM; + // The ModR/M byte, which contains most register operands and some portion of + // all memory operands. + bool consumedModRM; uint8_t modRM; - /* The SIB byte, used for more complex 32- or 64-bit memory operands */ - BOOL consumedSIB; + // The SIB byte, used for more complex 32- or 64-bit memory operands + bool consumedSIB; uint8_t sib; - /* The displacement, used for memory operands */ - BOOL consumedDisplacement; + // The displacement, used for memory operands + bool consumedDisplacement; int32_t displacement; - /* Immediates. There can be two in some cases */ + // Immediates. There can be two in some cases uint8_t numImmediatesConsumed; uint8_t numImmediatesTranslated; uint64_t immediates[2]; - /* A register or immediate operand encoded into the opcode */ + // A register or immediate operand encoded into the opcode Reg opcodeRegister; - /* Portions of the ModR/M byte */ + // Portions of the ModR/M byte - /* These fields determine the allowable values for the ModR/M fields, which - depend on operand and address widths */ + // These fields determine the allowable values for the ModR/M fields, which + // depend on operand and address widths. EABase eaBaseBase; EABase eaRegBase; Reg regBase; - /* The Mod and R/M fields can encode a base for an effective address, or a - register. These are separated into two fields here */ + // The Mod and R/M fields can encode a base for an effective address, or a + // register. These are separated into two fields here. EABase eaBase; EADisplacement eaDisplacement; - /* The reg field always encodes a register */ + // The reg field always encodes a register Reg reg; - /* SIB state */ + // SIB state SIBIndex sibIndex; uint8_t sibScale; SIBBase sibBase; - const struct OperandSpecifier *operands; + ArrayRef<OperandSpecifier> operands; }; -/* decodeInstruction - Decode one instruction and store the decoding results in - * a buffer provided by the consumer. - * @param insn - The buffer to store the instruction in. Allocated by the - * consumer. - * @param reader - The byteReader_t for the bytes to be read. - * @param readerArg - An argument to pass to the reader for storing context - * specific to the consumer. May be NULL. - * @param logger - The dlog_t to be used in printing status messages from the - * disassembler. May be NULL. - * @param loggerArg - An argument to pass to the logger for storing context - * specific to the logger. May be NULL. - * @param startLoc - The address (in the reader's address space) of the first - * byte in the instruction. - * @param mode - The mode (16-bit, 32-bit, 64-bit) to decode in. - * @return - Nonzero if there was an error during decode, 0 otherwise. - */ -int decodeInstruction(struct InternalInstruction* insn, +/// \brief Decode one instruction and store the decoding results in +/// a buffer provided by the consumer. +/// \param insn The buffer to store the instruction in. Allocated by the +/// consumer. +/// \param reader The byteReader_t for the bytes to be read. +/// \param readerArg An argument to pass to the reader for storing context +/// specific to the consumer. May be NULL. +/// \param logger The dlog_t to be used in printing status messages from the +/// disassembler. May be NULL. +/// \param loggerArg An argument to pass to the logger for storing context +/// specific to the logger. May be NULL. +/// \param startLoc The address (in the reader's address space) of the first +/// byte in the instruction. +/// \param mode The mode (16-bit, 32-bit, 64-bit) to decode in. +/// \return Nonzero if there was an error during decode, 0 otherwise. +int decodeInstruction(InternalInstruction *insn, byteReader_t reader, - const void* readerArg, + const void *readerArg, dlog_t logger, - void* loggerArg, - const void* miiArg, + void *loggerArg, + const void *miiArg, uint64_t startLoc, DisassemblerMode mode); -/* x86DisassemblerDebug - C-accessible function for printing a message to - * debugs() - * @param file - The name of the file printing the debug message. - * @param line - The line number that printed the debug message. - * @param s - The message to print. - */ +/// \brief Print a message to debugs() +/// \param file The name of the file printing the debug message. +/// \param line The line number that printed the debug message. +/// \param s The message to print. +void Debug(const char *file, unsigned line, const char *s); -void x86DisassemblerDebug(const char *file, - unsigned line, - const char *s); +const char *GetInstrName(unsigned Opcode, const void *mii); -const char *x86DisassemblerGetInstrName(unsigned Opcode, const void *mii); - -#ifdef __cplusplus -} -#endif +} // namespace X86Disassembler +} // namespace llvm #endif diff --git a/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h b/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h index 523ae99..f59e0b6 100644 --- a/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h +++ b/lib/Target/X86/Disassembler/X86DisassemblerDecoderCommon.h @@ -1,29 +1,27 @@ -/*===-- X86DisassemblerDecoderCommon.h - Disassembler decoder -----*- C -*-===* - * - * The LLVM Compiler Infrastructure - * - * This file is distributed under the University of Illinois Open Source - * License. See LICENSE.TXT for details. - * - *===----------------------------------------------------------------------===* - * - * This file is part of the X86 Disassembler. - * It contains common definitions used by both the disassembler and the table - * generator. - * Documentation for the disassembler can be found in X86Disassembler.h. - * - *===----------------------------------------------------------------------===*/ - -/* - * This header file provides those definitions that need to be shared between - * the decoder and the table generator in a C-friendly manner. - */ +//===-- X86DisassemblerDecoderCommon.h - Disassembler decoder ---*- C++ -*-===// +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// +// +// This file is part of the X86 Disassembler. +// It contains common definitions used by both the disassembler and the table +// generator. +// Documentation for the disassembler can be found in X86Disassembler.h. +// +//===----------------------------------------------------------------------===// #ifndef X86DISASSEMBLERDECODERCOMMON_H #define X86DISASSEMBLERDECODERCOMMON_H #include "llvm/Support/DataTypes.h" +namespace llvm { +namespace X86Disassembler { + #define INSTRUCTIONS_SYM x86DisassemblerInstrSpecifiers #define CONTEXTS_SYM x86DisassemblerContexts #define ONEBYTE_SYM x86DisassemblerOneByteOpcodes @@ -44,11 +42,9 @@ #define XOP9_MAP_STR "x86DisassemblerXOP9Opcodes" #define XOPA_MAP_STR "x86DisassemblerXOPAOpcodes" -/* - * Attributes of an instruction that must be known before the opcode can be - * processed correctly. Most of these indicate the presence of particular - * prefixes, but ATTR_64BIT is simply an attribute of the decoding context. - */ +// Attributes of an instruction that must be known before the opcode can be +// processed correctly. Most of these indicate the presence of particular +// prefixes, but ATTR_64BIT is simply an attribute of the decoding context. #define ATTRIBUTE_BITS \ ENUM_ENTRY(ATTR_NONE, 0x00) \ ENUM_ENTRY(ATTR_64BIT, (0x1 << 0)) \ @@ -73,13 +69,11 @@ enum attributeBits { }; #undef ENUM_ENTRY -/* - * Combinations of the above attributes that are relevant to instruction - * decode. Although other combinations are possible, they can be reduced to - * these without affecting the ultimately decoded instruction. - */ +// Combinations of the above attributes that are relevant to instruction +// decode. Although other combinations are possible, they can be reduced to +// these without affecting the ultimately decoded instruction. -/* Class name Rank Rationale for rank assignment */ +// Class name Rank Rationale for rank assignment #define INSTRUCTION_CONTEXTS \ ENUM_ENTRY(IC, 0, "says nothing about the instruction") \ ENUM_ENTRY(IC_64BIT, 1, "says the instruction applies in " \ @@ -274,17 +268,15 @@ enum attributeBits { ENUM_ENTRY(IC_EVEX_L2_W_OPSIZE_KZ, 4, "requires EVEX_KZ, L2, W and OpSize") #define ENUM_ENTRY(n, r, d) n, -typedef enum { +enum InstructionContext { INSTRUCTION_CONTEXTS IC_max -} InstructionContext; +}; #undef ENUM_ENTRY -/* - * Opcode types, which determine which decode table to use, both in the Intel - * manual and also for the decoder. - */ -typedef enum { +// Opcode types, which determine which decode table to use, both in the Intel +// manual and also for the decoder. +enum OpcodeType { ONEBYTE = 0, TWOBYTE = 1, THREEBYTE_38 = 2, @@ -292,39 +284,33 @@ typedef enum { XOP8_MAP = 4, XOP9_MAP = 5, XOPA_MAP = 6 -} OpcodeType; - -/* - * The following structs are used for the hierarchical decode table. After - * determining the instruction's class (i.e., which IC_* constant applies to - * it), the decoder reads the opcode. Some instructions require specific - * values of the ModR/M byte, so the ModR/M byte indexes into the final table. - * - * If a ModR/M byte is not required, "required" is left unset, and the values - * for each instructionID are identical. - */ +}; +// The following structs are used for the hierarchical decode table. After +// determining the instruction's class (i.e., which IC_* constant applies to +// it), the decoder reads the opcode. Some instructions require specific +// values of the ModR/M byte, so the ModR/M byte indexes into the final table. +// +// If a ModR/M byte is not required, "required" is left unset, and the values +// for each instructionID are identical. typedef uint16_t InstrUID; -/* - * ModRMDecisionType - describes the type of ModR/M decision, allowing the - * consumer to determine the number of entries in it. - * - * MODRM_ONEENTRY - No matter what the value of the ModR/M byte is, the decoded - * instruction is the same. - * MODRM_SPLITRM - If the ModR/M byte is between 0x00 and 0xbf, the opcode - * corresponds to one instruction; otherwise, it corresponds to - * a different instruction. - * MODRM_SPLITMISC- If the ModR/M byte is between 0x00 and 0xbf, ModR/M byte - * divided by 8 is used to select instruction; otherwise, each - * value of the ModR/M byte could correspond to a different - * instruction. - * MODRM_SPLITREG - ModR/M byte divided by 8 is used to select instruction. This - corresponds to instructions that use reg field as opcode - * MODRM_FULL - Potentially, each value of the ModR/M byte could correspond - * to a different instruction. - */ - +// ModRMDecisionType - describes the type of ModR/M decision, allowing the +// consumer to determine the number of entries in it. +// +// MODRM_ONEENTRY - No matter what the value of the ModR/M byte is, the decoded +// instruction is the same. +// MODRM_SPLITRM - If the ModR/M byte is between 0x00 and 0xbf, the opcode +// corresponds to one instruction; otherwise, it corresponds to +// a different instruction. +// MODRM_SPLITMISC- If the ModR/M byte is between 0x00 and 0xbf, ModR/M byte +// divided by 8 is used to select instruction; otherwise, each +// value of the ModR/M byte could correspond to a different +// instruction. +// MODRM_SPLITREG - ModR/M byte divided by 8 is used to select instruction. This +// corresponds to instructions that use reg field as opcode +// MODRM_FULL - Potentially, each value of the ModR/M byte could correspond +// to a different instruction. #define MODRMTYPES \ ENUM_ENTRY(MODRM_ONEENTRY) \ ENUM_ENTRY(MODRM_SPLITRM) \ @@ -333,47 +319,13 @@ typedef uint16_t InstrUID; ENUM_ENTRY(MODRM_FULL) #define ENUM_ENTRY(n) n, -typedef enum { +enum ModRMDecisionType { MODRMTYPES MODRM_max -} ModRMDecisionType; -#undef ENUM_ENTRY - -/* - * ModRMDecision - Specifies whether a ModR/M byte is needed and (if so) which - * instruction each possible value of the ModR/M byte corresponds to. Once - * this information is known, we have narrowed down to a single instruction. - */ -struct ModRMDecision { - uint8_t modrm_type; - - /* The macro below must be defined wherever this file is included. */ - INSTRUCTION_IDS -}; - -/* - * OpcodeDecision - Specifies which set of ModR/M->instruction tables to look at - * given a particular opcode. - */ -struct OpcodeDecision { - struct ModRMDecision modRMDecisions[256]; -}; - -/* - * ContextDecision - Specifies which opcode->instruction tables to look at given - * a particular context (set of attributes). Since there are many possible - * contexts, the decoder first uses CONTEXTS_SYM to determine which context - * applies given a specific set of attributes. Hence there are only IC_max - * entries in this table, rather than 2^(ATTR_max). - */ -struct ContextDecision { - struct OpcodeDecision opcodeDecisions[IC_max]; }; +#undef ENUM_ENTRY -/* - * Physical encodings of instruction operands. - */ - +// Physical encodings of instruction operands. #define ENCODINGS \ ENUM_ENTRY(ENCODING_NONE, "") \ ENUM_ENTRY(ENCODING_REG, "Register operand in ModR/M byte.") \ @@ -408,16 +360,13 @@ struct ContextDecision { ENUM_ENTRY(ENCODING_DI, "Destination index; encoded in prefixes") #define ENUM_ENTRY(n, d) n, - typedef enum { - ENCODINGS - ENCODING_max - } OperandEncoding; +enum OperandEncoding { + ENCODINGS + ENCODING_max +}; #undef ENUM_ENTRY -/* - * Semantic interpretations of instruction operands. - */ - +// Semantic interpretations of instruction operands. #define TYPES \ ENUM_ENTRY(TYPE_NONE, "") \ ENUM_ENTRY(TYPE_REL8, "1-byte immediate address") \ @@ -508,56 +457,42 @@ struct ContextDecision { ENUM_ENTRY(TYPE_M512, "512-bit FPU/MMX/XMM/MXCSR state") #define ENUM_ENTRY(n, d) n, -typedef enum { +enum OperandType { TYPES TYPE_max -} OperandType; +}; #undef ENUM_ENTRY -/* - * OperandSpecifier - The specification for how to extract and interpret one - * operand. - */ +/// \brief The specification for how to extract and interpret one operand. struct OperandSpecifier { uint8_t encoding; uint8_t type; }; -/* - * Indicates where the opcode modifier (if any) is to be found. Extended - * opcodes with AddRegFrm have the opcode modifier in the ModR/M byte. - */ - +// Indicates where the opcode modifier (if any) is to be found. Extended +// opcodes with AddRegFrm have the opcode modifier in the ModR/M byte. #define MODIFIER_TYPES \ ENUM_ENTRY(MODIFIER_NONE) #define ENUM_ENTRY(n) n, -typedef enum { +enum ModifierType { MODIFIER_TYPES MODIFIER_max -} ModifierType; +}; #undef ENUM_ENTRY -#define X86_MAX_OPERANDS 5 - -/* - * The specification for how to extract and interpret a full instruction and - * its operands. - */ -struct InstructionSpecifier { - /* The macro below must be defined wherever this file is included. */ - INSTRUCTION_SPECIFIER_FIELDS -}; +static const unsigned X86_MAX_OPERANDS = 5; -/* - * Decoding mode for the Intel disassembler. 16-bit, 32-bit, and 64-bit mode - * are supported, and represent real mode, IA-32e, and IA-32e in 64-bit mode, - * respectively. - */ -typedef enum { +/// Decoding mode for the Intel disassembler. 16-bit, 32-bit, and 64-bit mode +/// are supported, and represent real mode, IA-32e, and IA-32e in 64-bit mode, +/// respectively. +enum DisassemblerMode { MODE_16BIT, MODE_32BIT, MODE_64BIT -} DisassemblerMode; +}; + +} // namespace X86Disassembler +} // namespace llvm #endif |