RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-01-31 12:41:49 +01:00
Code Issues Projects Releases Wiki Activity

Remove edis - the enhanced disassembler. Fixes PR14654.

Browse Source 
llvm-svn: 170578
This commit is contained in:
Roman Divacky 2012-12-19 19:55:47 +00:00
parent b0b73c2c50
commit 7a967134bc
36 changed files with 4 additions and 4006 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
Makefile.rules
View File

@ -1869,11 +1869,6 @@ $(ObjDir)/%GenDisassemblerTables.inc.tmp : %.td $(ObjDir)/.dir $(LLVM_TBLGEN)
$(Echo) "Building $(<F) disassembly tables with tblgen"
$(Verb) $(LLVMTableGen) -gen-disassembler -o $(call SYSPATH, $@) $<
$(TARGET:%=$(ObjDir)/%GenEDInfo.inc.tmp): \
$(ObjDir)/%GenEDInfo.inc.tmp : %.td $(ObjDir)/.dir $(LLVM_TBLGEN)
$(Echo) "Building $(<F) enhanced disassembly information with tblgen"
$(Verb) $(LLVMTableGen) -gen-enhanced-disassembly-info -o $(call SYSPATH, $@) $<
$(TARGET:%=$(ObjDir)/%GenFastISel.inc.tmp): \
$(ObjDir)/%GenFastISel.inc.tmp : %.td $(ObjDir)/.dir $(LLVM_TBLGEN)
$(Echo) "Building $(<F) \"fast\" instruction selector implementation with tblgen"

530
include/llvm-c/EnhancedDisassembly.h
View File

@ -1,530 +0,0 @@
/*===-- llvm-c/EnhancedDisassembly.h - Disassembler C Interface ---*- C -*-===*\
|* *|
|* The LLVM Compiler Infrastructure *|
|* *|
|* This file is distributed under the University of Illinois Open Source *|
|* License. See LICENSE.TXT for details. *|
|* *|
|*===----------------------------------------------------------------------===*|
|* *|
|* This header declares the C interface to EnhancedDisassembly.so, which *|
|* implements a disassembler with the ability to extract operand values and *|
|* individual tokens from assembly instructions. *|
|* *|
|* The header declares additional interfaces if the host compiler supports *|
|* the blocks API. *|
|* *|
\*===----------------------------------------------------------------------===*/
#ifndef LLVM_C_ENHANCEDDISASSEMBLY_H
#define LLVM_C_ENHANCEDDISASSEMBLY_H
#include "llvm/Support/DataTypes.h"
#ifdef __cplusplus
extern "C" {
#endif
/**
* @defgroup LLVMCEnhancedDisassembly Enhanced Disassembly
* @ingroup LLVMC
* @deprecated
*
* This module contains an interface to the Enhanced Disassembly (edis)
* library. The edis library is deprecated and will likely disappear in
* the near future. You should use the @ref LLVMCDisassembler interface
* instead.
*
* @{
*/
/*!
@typedef EDByteReaderCallback
Interface to memory from which instructions may be read.
@param byte A pointer whose target should be filled in with the data returned.
@param address The address of the byte to be read.
@param arg An anonymous argument for client use.
@result 0 on success; -1 otherwise.
*/
typedef int (*EDByteReaderCallback)(uint8_t *byte, uint64_t address, void *arg);
/*!
@typedef EDRegisterReaderCallback
Interface to registers from which registers may be read.
@param value A pointer whose target should be filled in with the value of the
register.
@param regID The LLVM register identifier for the register to read.
@param arg An anonymous argument for client use.
@result 0 if the register could be read; -1 otherwise.
*/
typedef int (*EDRegisterReaderCallback)(uint64_t *value, unsigned regID,
void* arg);
/*!
@typedef EDAssemblySyntax_t
An assembly syntax for use in tokenizing instructions.
*/
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
};
typedef unsigned EDAssemblySyntax_t;
/*!
@typedef EDDisassemblerRef
Encapsulates a disassembler for a single CPU architecture.
*/
typedef void *EDDisassemblerRef;
/*!
@typedef EDInstRef
Encapsulates a single disassembled instruction in one assembly syntax.
*/
typedef void *EDInstRef;
/*!
@typedef EDTokenRef
Encapsulates a token from the disassembly of an instruction.
*/
typedef void *EDTokenRef;
/*!
@typedef EDOperandRef
Encapsulates an operand of an instruction.
*/
typedef void *EDOperandRef;
/*!
@functiongroup Getting a disassembler
*/
/*!
@function EDGetDisassembler
Gets the disassembler for a given target.
@param disassembler A pointer whose target will be filled in with the
disassembler.
@param triple Identifies the target. Example: "x86_64-apple-darwin10"
@param syntax The assembly syntax to use when decoding instructions.
@result 0 on success; -1 otherwise.
*/
int EDGetDisassembler(EDDisassemblerRef *disassembler,
const char *triple,
EDAssemblySyntax_t syntax);
/*!
@functiongroup Generic architectural queries
*/
/*!
@function EDGetRegisterName
Gets the human-readable name for a given register.
@param regName A pointer whose target will be pointed at the name of the
register. The name does not need to be deallocated and will be
@param disassembler The disassembler to query for the name.
@param regID The register identifier, as returned by EDRegisterTokenValue.
@result 0 on success; -1 otherwise.
*/
int EDGetRegisterName(const char** regName,
EDDisassemblerRef disassembler,
unsigned regID);
/*!
@function EDRegisterIsStackPointer
Determines if a register is one of the platform's stack-pointer registers.
@param disassembler The disassembler to query.
@param regID The register identifier, as returned by EDRegisterTokenValue.
@result 1 if true; 0 otherwise.
*/
int EDRegisterIsStackPointer(EDDisassemblerRef disassembler,
unsigned regID);
/*!
@function EDRegisterIsProgramCounter
Determines if a register is one of the platform's stack-pointer registers.
@param disassembler The disassembler to query.
@param regID The register identifier, as returned by EDRegisterTokenValue.
@result 1 if true; 0 otherwise.
*/
int EDRegisterIsProgramCounter(EDDisassemblerRef disassembler,
unsigned regID);
/*!
@functiongroup Creating and querying instructions
*/
/*!
@function EDCreateInst
Gets a set of contiguous instructions from a disassembler.
@param insts A pointer to an array that will be filled in with the
instructions. Must have at least count entries. Entries not filled in will
be set to NULL.
@param count The maximum number of instructions to fill in.
@param disassembler The disassembler to use when decoding the instructions.
@param byteReader The function to use when reading the instruction's machine
code.
@param address The address of the first byte of the instruction.
@param arg An anonymous argument to be passed to byteReader.
@result The number of instructions read on success; 0 otherwise.
*/
unsigned int EDCreateInsts(EDInstRef *insts,
unsigned int count,
EDDisassemblerRef disassembler,
EDByteReaderCallback byteReader,
uint64_t address,
void *arg);
/*!
@function EDReleaseInst
Frees the memory for an instruction. The instruction can no longer be accessed
after this call.
@param inst The instruction to be freed.
*/
void EDReleaseInst(EDInstRef inst);
/*!
@function EDInstByteSize
@param inst The instruction to be queried.
@result The number of bytes in the instruction's machine-code representation.
*/
int EDInstByteSize(EDInstRef inst);
/*!
@function EDGetInstString
Gets the disassembled text equivalent of the instruction.
@param buf A pointer whose target will be filled in with a pointer to the
string. (The string becomes invalid when the instruction is released.)
@param inst The instruction to be queried.
@result 0 on success; -1 otherwise.
*/
int EDGetInstString(const char **buf,
EDInstRef inst);
/*!
@function EDInstID
@param instID A pointer whose target will be filled in with the LLVM identifier
for the instruction.
@param inst The instruction to be queried.
@result 0 on success; -1 otherwise.
*/
int EDInstID(unsigned *instID, EDInstRef inst);
/*!
@function EDInstIsBranch
@param inst The instruction to be queried.
@result 1 if the instruction is a branch instruction; 0 if it is some other
type of instruction; -1 if there was an error.
*/
int EDInstIsBranch(EDInstRef inst);
/*!
@function EDInstIsMove
@param inst The instruction to be queried.
@result 1 if the instruction is a move instruction; 0 if it is some other
type of instruction; -1 if there was an error.
*/
int EDInstIsMove(EDInstRef inst);
/*!
@function EDBranchTargetID
@param inst The instruction to be queried.
@result The ID of the branch target operand, suitable for use with
EDCopyOperand. -1 if no such operand exists.
*/
int EDBranchTargetID(EDInstRef inst);
/*!
@function EDMoveSourceID
@param inst The instruction to be queried.
@result The ID of the move source operand, suitable for use with
EDCopyOperand. -1 if no such operand exists.
*/
int EDMoveSourceID(EDInstRef inst);
/*!
@function EDMoveTargetID
@param inst The instruction to be queried.
@result The ID of the move source operand, suitable for use with
EDCopyOperand. -1 if no such operand exists.
*/
int EDMoveTargetID(EDInstRef inst);
/*!
@functiongroup Creating and querying tokens
*/
/*!
@function EDNumTokens
@param inst The instruction to be queried.
@result The number of tokens in the instruction, or -1 on error.
*/
int EDNumTokens(EDInstRef inst);
/*!
@function EDGetToken
Retrieves a token from an instruction. The token is valid until the
instruction is released.
@param token A pointer to be filled in with the token.
@param inst The instruction to be queried.
@param index The index of the token in the instruction.
@result 0 on success; -1 otherwise.
*/
int EDGetToken(EDTokenRef *token,
EDInstRef inst,
int index);
/*!
@function EDGetTokenString
Gets the disassembled text for a token.
@param buf A pointer whose target will be filled in with a pointer to the
string. (The string becomes invalid when the token is released.)
@param token The token to be queried.
@result 0 on success; -1 otherwise.
*/
int EDGetTokenString(const char **buf,
EDTokenRef token);
/*!
@function EDOperandIndexForToken
Returns the index of the operand to which a token belongs.
@param token The token to be queried.
@result The operand index on success; -1 otherwise
*/
int EDOperandIndexForToken(EDTokenRef token);
/*!
@function EDTokenIsWhitespace
@param token The token to be queried.
@result 1 if the token is whitespace; 0 if not; -1 on error.
*/
int EDTokenIsWhitespace(EDTokenRef token);
/*!
@function EDTokenIsPunctuation
@param token The token to be queried.
@result 1 if the token is punctuation; 0 if not; -1 on error.
*/
int EDTokenIsPunctuation(EDTokenRef token);
/*!
@function EDTokenIsOpcode
@param token The token to be queried.
@result 1 if the token is opcode; 0 if not; -1 on error.
*/
int EDTokenIsOpcode(EDTokenRef token);
/*!
@function EDTokenIsLiteral
@param token The token to be queried.
@result 1 if the token is a numeric literal; 0 if not; -1 on error.
*/
int EDTokenIsLiteral(EDTokenRef token);
/*!
@function EDTokenIsRegister
@param token The token to be queried.
@result 1 if the token identifies a register; 0 if not; -1 on error.
*/
int EDTokenIsRegister(EDTokenRef token);
/*!
@function EDTokenIsNegativeLiteral
@param token The token to be queried.
@result 1 if the token is a negative signed literal; 0 if not; -1 on error.
*/
int EDTokenIsNegativeLiteral(EDTokenRef token);
/*!
@function EDLiteralTokenAbsoluteValue
@param value A pointer whose target will be filled in with the absolute value
of the literal.
@param token The token to be queried.
@result 0 on success; -1 otherwise.
*/
int EDLiteralTokenAbsoluteValue(uint64_t *value,
EDTokenRef token);
/*!
@function EDRegisterTokenValue
@param registerID A pointer whose target will be filled in with the LLVM
register identifier for the token.
@param token The token to be queried.
@result 0 on success; -1 otherwise.
*/
int EDRegisterTokenValue(unsigned *registerID,
EDTokenRef token);
/*!
@functiongroup Creating and querying operands
*/
/*!
@function EDNumOperands
@param inst The instruction to be queried.
@result The number of operands in the instruction, or -1 on error.
*/
int EDNumOperands(EDInstRef inst);
/*!
@function EDGetOperand
Retrieves an operand from an instruction. The operand is valid until the
instruction is released.
@param operand A pointer to be filled in with the operand.
@param inst The instruction to be queried.
@param index The index of the operand in the instruction.
@result 0 on success; -1 otherwise.
*/
int EDGetOperand(EDOperandRef *operand,
EDInstRef inst,
int index);
/*!
@function EDOperandIsRegister
@param operand The operand to be queried.
@result 1 if the operand names a register; 0 if not; -1 on error.
*/
int EDOperandIsRegister(EDOperandRef operand);
/*!
@function EDOperandIsImmediate
@param operand The operand to be queried.
@result 1 if the operand specifies an immediate value; 0 if not; -1 on error.
*/
int EDOperandIsImmediate(EDOperandRef operand);
/*!
@function EDOperandIsMemory
@param operand The operand to be queried.
@result 1 if the operand specifies a location in memory; 0 if not; -1 on error.
*/
int EDOperandIsMemory(EDOperandRef operand);
/*!
@function EDRegisterOperandValue
@param value A pointer whose target will be filled in with the LLVM register ID
of the register named by the operand.
@param operand The operand to be queried.
@result 0 on success; -1 otherwise.
*/
int EDRegisterOperandValue(unsigned *value,
EDOperandRef operand);
/*!
@function EDImmediateOperandValue
@param value A pointer whose target will be filled in with the value of the
immediate.
@param operand The operand to be queried.
@result 0 on success; -1 otherwise.
*/
int EDImmediateOperandValue(uint64_t *value,
EDOperandRef operand);
/*!
@function EDEvaluateOperand
Evaluates an operand using a client-supplied register state accessor. Register
operands are evaluated by reading the value of the register; immediate operands
are evaluated by reporting the immediate value; memory operands are evaluated
by computing the target address (with only those relocations applied that were
already applied to the original bytes).
@param result A pointer whose target is to be filled with the result of
evaluating the operand.
@param operand The operand to be evaluated.
@param regReader The function to use when reading registers from the register
state.
@param arg An anonymous argument for client use.
@result 0 if the operand could be evaluated; -1 otherwise.
*/
int EDEvaluateOperand(uint64_t *result,
EDOperandRef operand,
EDRegisterReaderCallback regReader,
void *arg);
#ifdef __BLOCKS__
/*!
@typedef EDByteBlock_t
Block-based interface to memory from which instructions may be read.
@param byte A pointer whose target should be filled in with the data returned.
@param address The address of the byte to be read.
@result 0 on success; -1 otherwise.
*/
typedef int (^EDByteBlock_t)(uint8_t *byte, uint64_t address);
/*!
@typedef EDRegisterBlock_t
Block-based interface to registers from which registers may be read.
@param value A pointer whose target should be filled in with the value of the
register.
@param regID The LLVM register identifier for the register to read.
@result 0 if the register could be read; -1 otherwise.
*/
typedef int (^EDRegisterBlock_t)(uint64_t *value, unsigned regID);
/*!
@typedef EDTokenVisitor_t
Block-based handler for individual tokens.
@param token The current token being read.
@result 0 to continue; 1 to stop normally; -1 on error.
*/
typedef int (^EDTokenVisitor_t)(EDTokenRef token);
/*! @functiongroup Block-based interfaces */
/*!
@function EDBlockCreateInsts
Gets a set of contiguous instructions from a disassembler, using a block to
read memory.
@param insts A pointer to an array that will be filled in with the
instructions. Must have at least count entries. Entries not filled in will
be set to NULL.
@param count The maximum number of instructions to fill in.
@param disassembler The disassembler to use when decoding the instructions.
@param byteBlock The block to use when reading the instruction's machine
code.
@param address The address of the first byte of the instruction.
@result The number of instructions read on success; 0 otherwise.
*/
unsigned int EDBlockCreateInsts(EDInstRef *insts,
int count,
EDDisassemblerRef disassembler,
EDByteBlock_t byteBlock,
uint64_t address);
/*!
@function EDBlockEvaluateOperand
Evaluates an operand using a block to read registers.
@param result A pointer whose target is to be filled with the result of
evaluating the operand.
@param operand The operand to be evaluated.
@param regBlock The block to use when reading registers from the register
state.
@result 0 if the operand could be evaluated; -1 otherwise.
*/
int EDBlockEvaluateOperand(uint64_t *result,
EDOperandRef operand,
EDRegisterBlock_t regBlock);
/*!
@function EDBlockVisitTokens
Visits every token with a visitor.
@param inst The instruction with the tokens to be visited.
@param visitor The visitor.
@result 0 if the visit ended normally; -1 if the visitor encountered an error
or there was some other error.
*/
int EDBlockVisitTokens(EDInstRef inst,
EDTokenVisitor_t visitor);
/**
* @}
*/
#endif
#ifdef __cplusplus
}
#endif
#endif

10
include/llvm/MC/MCDisassembler.h
View File

@ -20,8 +20,6 @@ class MemoryObject;
class raw_ostream;
class MCContext;
struct EDInstInfo;
/// MCDisassembler - Superclass for all disassemblers. Consumes a memory region
/// and provides an array of assembly instructions.
class MCDisassembler {
@ -84,14 +82,6 @@ public:
raw_ostream &vStream,
raw_ostream &cStream) const = 0;
/// getEDInfo - Returns the enhanced instruction information corresponding to
/// the disassembler.
///
/// @return - An array of instruction information, with one entry for
/// each MCInst opcode this disassembler returns.
/// NULL if there is no info for this target.
virtual const EDInstInfo *getEDInfo() const { return (EDInstInfo*)0; }
private:
//
// Hooks for symbolic disassembly via the public 'C' interface.

5
lib/MC/MCDisassembler/CMakeLists.txt
View File

@ -1,8 +1,3 @@
add_llvm_library(LLVMMCDisassembler
Disassembler.cpp
EDDisassembler.cpp
EDInst.cpp
EDMain.cpp
EDOperand.cpp
EDToken.cpp
)

400
lib/MC/MCDisassembler/EDDisassembler.cpp
View File

@ -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 &registerInfo) {
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;
}

269
lib/MC/MCDisassembler/EDDisassembler.h
View File

@ -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 &registerInfo);
/// 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

84
lib/MC/MCDisassembler/EDInfo.h
View File

@ -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

211
lib/MC/MCDisassembler/EDInst.cpp
View File

@ -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

182
lib/MC/MCDisassembler/EDInst.h
View File

@ -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

276
lib/MC/MCDisassembler/EDMain.cpp
View File

@ -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

315
lib/MC/MCDisassembler/EDOperand.cpp
View File

@ -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

91
lib/MC/MCDisassembler/EDOperand.h
View File

@ -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

214
lib/MC/MCDisassembler/EDToken.cpp
View File

@ -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 &registerID) 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;
}

139
lib/MC/MCDisassembler/EDToken.h
View File

@ -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 &registerID) 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

1
lib/Target/ARM/CMakeLists.txt
View File

@ -11,7 +11,6 @@ tablegen(LLVM ARMGenDAGISel.inc -gen-dag-isel)
tablegen(LLVM ARMGenFastISel.inc -gen-fast-isel)
tablegen(LLVM ARMGenCallingConv.inc -gen-callingconv)
tablegen(LLVM ARMGenSubtargetInfo.inc -gen-subtarget)
tablegen(LLVM ARMGenEDInfo.inc -gen-enhanced-disassembly-info)
tablegen(LLVM ARMGenDisassemblerTables.inc -gen-disassembler)
add_public_tablegen_target(ARMCommonTableGen)

16
lib/Target/ARM/Disassembler/ARMDisassembler.cpp
View File

@ -13,7 +13,6 @@
#include "MCTargetDesc/ARMAddressingModes.h"
#include "MCTargetDesc/ARMBaseInfo.h"
#include "MCTargetDesc/ARMMCExpr.h"
#include "llvm/MC/EDInstInfo.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCFixedLenDisassembler.h"
@ -105,10 +104,6 @@ public:
uint64_t address,
raw_ostream &vStream,
raw_ostream &cStream) const;
/// getEDInfo - See MCDisassembler.
const EDInstInfo *getEDInfo() const;
private:
};
/// ThumbDisassembler - Thumb disassembler for all Thumb platforms.
@ -131,8 +126,6 @@ public:
raw_ostream &vStream,
raw_ostream &cStream) const;
/// getEDInfo - See MCDisassembler.
const EDInstInfo *getEDInfo() const;
private:
mutable ITStatus ITBlock;
DecodeStatus AddThumbPredicate(MCInst&) const;
@ -385,7 +378,6 @@ static DecodeStatus DecodeLDR(MCInst &Inst, unsigned Val,
static DecodeStatus DecodeMRRC2(llvm::MCInst &Inst, unsigned Val,
uint64_t Address, const void *Decoder);
#include "ARMGenDisassemblerTables.inc"
#include "ARMGenEDInfo.inc"
static MCDisassembler *createARMDisassembler(const Target &T, const MCSubtargetInfo &STI) {
return new ARMDisassembler(STI);
@ -395,14 +387,6 @@ static MCDisassembler *createThumbDisassembler(const Target &T, const MCSubtarge
return new ThumbDisassembler(STI);
}
const EDInstInfo *ARMDisassembler::getEDInfo() const {
return instInfoARM;
}
const EDInstInfo *ThumbDisassembler::getEDInfo() const {
return instInfoARM;
}
DecodeStatus ARMDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
const MemoryObject &Region,
uint64_t Address,

2
lib/Target/ARM/Makefile
View File

@ -16,7 +16,7 @@ BUILT_SOURCES = ARMGenRegisterInfo.inc ARMGenInstrInfo.inc \
ARMGenAsmWriter.inc ARMGenAsmMatcher.inc \
ARMGenDAGISel.inc ARMGenSubtargetInfo.inc \
ARMGenCodeEmitter.inc ARMGenCallingConv.inc \
ARMGenEDInfo.inc ARMGenFastISel.inc ARMGenMCCodeEmitter.inc \
ARMGenFastISel.inc ARMGenMCCodeEmitter.inc \
ARMGenMCPseudoLowering.inc ARMGenDisassemblerTables.inc
DIRS = InstPrinter AsmParser Disassembler TargetInfo MCTargetDesc

1
lib/Target/MBlaze/CMakeLists.txt
View File

@ -9,7 +9,6 @@ tablegen(LLVM MBlazeGenDAGISel.inc -gen-dag-isel)
tablegen(LLVM MBlazeGenCallingConv.inc -gen-callingconv)
tablegen(LLVM MBlazeGenSubtargetInfo.inc -gen-subtarget)
tablegen(LLVM MBlazeGenIntrinsics.inc -gen-tgt-intrinsic)
tablegen(LLVM MBlazeGenEDInfo.inc -gen-enhanced-disassembly-info)
add_public_tablegen_target(MBlazeCommonTableGen)
add_llvm_target(MBlazeCodeGen

6
lib/Target/MBlaze/Disassembler/MBlazeDisassembler.cpp
View File

@ -14,7 +14,6 @@
#include "MBlazeDisassembler.h"
#include "MBlaze.h"
#include "llvm/MC/EDInstInfo.h"
#include "llvm/MC/MCDisassembler.h"
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCInstrDesc.h"
@ -25,7 +24,6 @@
// #include "MBlazeGenDecoderTables.inc"
// #include "MBlazeGenRegisterNames.inc"
#include "MBlazeGenEDInfo.inc"
namespace llvm {
extern const MCInstrDesc MBlazeInsts[];
@ -491,10 +489,6 @@ static unsigned getOPCODE(uint32_t insn) {
}
}
const EDInstInfo *MBlazeDisassembler::getEDInfo() const {
return instInfoMBlaze;
}
//
// Public interface for the disassembler
//

5
lib/Target/MBlaze/Disassembler/MBlazeDisassembler.h
View File

@ -23,8 +23,6 @@ class MCInst;
class MemoryObject;
class raw_ostream;
struct EDInstInfo;
/// MBlazeDisassembler - Disassembler for all MBlaze platforms.
class MBlazeDisassembler : public MCDisassembler {
public:
@ -44,9 +42,6 @@ public:
uint64_t address,
raw_ostream &vStream,
raw_ostream &cStream) const;
/// getEDInfo - See MCDisassembler.
const EDInstInfo *getEDInfo() const;
};
} // namespace llvm

3
lib/Target/MBlaze/Makefile
View File

@ -15,8 +15,7 @@ BUILT_SOURCES = MBlazeGenRegisterInfo.inc MBlazeGenInstrInfo.inc \
MBlazeGenAsmWriter.inc \
MBlazeGenDAGISel.inc MBlazeGenAsmMatcher.inc \
MBlazeGenCodeEmitter.inc MBlazeGenCallingConv.inc \
MBlazeGenSubtargetInfo.inc MBlazeGenIntrinsics.inc \
MBlazeGenEDInfo.inc
MBlazeGenSubtargetInfo.inc MBlazeGenIntrinsics.inc
DIRS = InstPrinter AsmParser Disassembler TargetInfo MCTargetDesc

1
lib/Target/Mips/CMakeLists.txt
View File

@ -9,7 +9,6 @@ tablegen(LLVM MipsGenAsmWriter.inc -gen-asm-writer)
tablegen(LLVM MipsGenDAGISel.inc -gen-dag-isel)
tablegen(LLVM MipsGenCallingConv.inc -gen-callingconv)
tablegen(LLVM MipsGenSubtargetInfo.inc -gen-subtarget)
tablegen(LLVM MipsGenEDInfo.inc -gen-enhanced-disassembly-info)
tablegen(LLVM MipsGenAsmMatcher.inc -gen-asm-matcher)
tablegen(LLVM MipsGenMCPseudoLowering.inc -gen-pseudo-lowering)
add_public_tablegen_target(MipsCommonTableGen)

11
lib/Target/Mips/Disassembler/MipsDisassembler.cpp
View File

@ -14,7 +14,6 @@
#include "Mips.h"
#include "MipsRegisterInfo.h"
#include "MipsSubtarget.h"
#include "llvm/MC/EDInstInfo.h"
#include "llvm/MC/MCDisassembler.h"
#include "llvm/MC/MCFixedLenDisassembler.h"
#include "llvm/MC/MCInst.h"
@ -23,9 +22,6 @@
#include "llvm/Support/MemoryObject.h"
#include "llvm/Support/TargetRegistry.h"
// Not a normal header, this must come last.
#include "MipsGenEDInfo.inc"
using namespace llvm;
typedef MCDisassembler::DecodeStatus DecodeStatus;
@ -43,9 +39,6 @@ public:
virtual ~MipsDisassemblerBase() {}
/// getEDInfo - See MCDisassembler.
const EDInstInfo *getEDInfo() const;
const MCRegisterInfo *getRegInfo() const { return RegInfo; }
private:
@ -93,10 +86,6 @@ public:
} // end anonymous namespace
const EDInstInfo *MipsDisassemblerBase::getEDInfo() const {
return instInfoMips;
}
// Forward declare these because the autogenerated code will reference them.
// Definitions are further down.
static DecodeStatus DecodeCPU64RegsRegisterClass(MCInst &Inst,

2
lib/Target/Mips/Makefile
View File

@ -16,7 +16,7 @@ BUILT_SOURCES = MipsGenRegisterInfo.inc MipsGenInstrInfo.inc \
MipsGenAsmWriter.inc MipsGenCodeEmitter.inc \
MipsGenDAGISel.inc MipsGenCallingConv.inc \
MipsGenSubtargetInfo.inc MipsGenMCCodeEmitter.inc \
MipsGenEDInfo.inc MipsGenDisassemblerTables.inc \
MipsGenDisassemblerTables.inc \
MipsGenMCPseudoLowering.inc MipsGenAsmMatcher.inc
DIRS = InstPrinter Disassembler AsmParser TargetInfo MCTargetDesc

1
lib/Target/X86/CMakeLists.txt
View File

@ -10,7 +10,6 @@ tablegen(LLVM X86GenDAGISel.inc -gen-dag-isel)
tablegen(LLVM X86GenFastISel.inc -gen-fast-isel)
tablegen(LLVM X86GenCallingConv.inc -gen-callingconv)
tablegen(LLVM X86GenSubtargetInfo.inc -gen-subtarget)
tablegen(LLVM X86GenEDInfo.inc -gen-enhanced-disassembly-info)
add_public_tablegen_target(X86CommonTableGen)
set(sources

6
lib/Target/X86/Disassembler/X86Disassembler.cpp
View File

@ -16,7 +16,6 @@
#include "X86Disassembler.h"
#include "X86DisassemblerDecoder.h"
#include "llvm/MC/EDInstInfo.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCDisassembler.h"
#include "llvm/MC/MCExpr.h"
@ -32,7 +31,6 @@
#include "X86GenRegisterInfo.inc"
#define GET_INSTRINFO_ENUM
#include "X86GenInstrInfo.inc"
#include "X86GenEDInfo.inc"
using namespace llvm;
using namespace llvm::X86Disassembler;
@ -83,10 +81,6 @@ X86GenericDisassembler::~X86GenericDisassembler() {
delete MII;
}
const EDInstInfo *X86GenericDisassembler::getEDInfo() const {
return instInfoX86;
}
/// regionReader - a callback function that wraps the readByte method from
/// MemoryObject.
///

4
lib/Target/X86/Disassembler/X86Disassembler.h
View File

@ -95,8 +95,6 @@ class MCSubtargetInfo;
class MemoryObject;
class raw_ostream;
struct EDInstInfo;
namespace X86Disassembler {
/// X86GenericDisassembler - Generic disassembler for all X86 platforms.
@ -122,8 +120,6 @@ public:
raw_ostream &vStream,
raw_ostream &cStream) const;
/// getEDInfo - See MCDisassembler.
const EDInstInfo *getEDInfo() const;
private:
DisassemblerMode fMode;
};

3
lib/Target/X86/Makefile
View File

@ -16,8 +16,7 @@ BUILT_SOURCES = X86GenRegisterInfo.inc X86GenInstrInfo.inc \
X86GenAsmWriter.inc X86GenAsmMatcher.inc \
X86GenAsmWriter1.inc X86GenDAGISel.inc \
X86GenDisassemblerTables.inc X86GenFastISel.inc \
X86GenCallingConv.inc X86GenSubtargetInfo.inc \
X86GenEDInfo.inc
X86GenCallingConv.inc X86GenSubtargetInfo.inc
DIRS = InstPrinter AsmParser Disassembler TargetInfo MCTargetDesc Utils

10
test/MC/Disassembler/X86/enhanced.txt
View File

@ -1,10 +0,0 @@
# RUN: llvm-mc --edis %s -triple=x86_64-apple-darwin9 2>&1 | FileCheck %s
# CHECK: [o:jne][w: ][0-p:-][0-l:10=10] <br> 0:[RIP/{{[0-9]+}}](pc)=18446744073709551606
0x0f 0x85 0xf6 0xff 0xff 0xff
# CHECK: [o:movq][w: ][1-r:%gs=r{{[0-9]+}}][1-p::][1-l:8=8][p:,][w: ][0-r:%rcx=r{{[0-9]+}}] <mov> 0:[RCX/{{[0-9]+}}]=0 1:[GS/{{[0-9]+}}]=8
0x65 0x48 0x8b 0x0c 0x25 0x08 0x00 0x00 0x00
# CHECK: [o:xorps][w: ][2-r:%xmm1=r{{[0-9]+}}][p:,][w: ][0-r:%xmm2=r{{[0-9]+}}] 0:[XMM2/{{[0-9]+}}]=0 1:[XMM2/{{[0-9]+}}]=0 2:[XMM1/{{[0-9]+}}]=0
0x0f 0x57 0xd1
# CHECK: [o:andps][w: ][2-r:%xmm1=r{{[0-9]+}}][p:,][w: ][0-r:%xmm2=r{{[0-9]+}}] 0:[XMM2/{{[0-9]+}}]=0 1:[XMM2/{{[0-9]+}}]=0 2:[XMM1/{{[0-9]+}}]=0
0x0f 0x54 0xd1

176
tools/llvm-mc/Disassembler.cpp
View File

@ -13,10 +13,6 @@
//===----------------------------------------------------------------------===//
#include "Disassembler.h"
#include "../../lib/MC/MCDisassembler/EDDisassembler.h"
#include "../../lib/MC/MCDisassembler/EDInst.h"
#include "../../lib/MC/MCDisassembler/EDOperand.h"
#include "../../lib/MC/MCDisassembler/EDToken.h"
#include "llvm/ADT/OwningPtr.h"
#include "llvm/ADT/Triple.h"
#include "llvm/MC/MCDisassembler.h"
@ -169,175 +165,3 @@ int Disassembler::disassemble(const Target &T,
return ErrorOccurred;
}
static int byteArrayReader(uint8_t *B, uint64_t A, void *Arg) {
ByteArrayTy &ByteArray = *((ByteArrayTy*)Arg);
if (A >= ByteArray.size())
return -1;
*B = ByteArray[A].first;
return 0;
}
static int verboseEvaluator(uint64_t *V, unsigned R, void *Arg) {
EDDisassembler &disassembler = *(EDDisassembler *)((void **)Arg)[0];
raw_ostream &Out = *(raw_ostream *)((void **)Arg)[1];
if (const char *regName = disassembler.nameWithRegisterID(R))
Out << "[" << regName << "/" << R << "]";
if (disassembler.registerIsStackPointer(R))
Out << "(sp)";
if (disassembler.registerIsProgramCounter(R))
Out << "(pc)";
*V = 0;
return 0;
}
int Disassembler::disassembleEnhanced(const std::string &TS,
MemoryBuffer &Buffer,
SourceMgr &SM,
raw_ostream &Out) {
ByteArrayTy ByteArray;
StringRef Str = Buffer.getBuffer();
if (ByteArrayFromString(ByteArray, Str, SM)) {
return -1;
}
Triple T(TS);
EDDisassembler::AssemblySyntax AS;
switch (T.getArch()) {
default:
errs() << "error: no default assembly syntax for " << TS.c_str() << "\n";
return -1;
case Triple::arm:
case Triple::thumb:
AS = EDDisassembler::kEDAssemblySyntaxARMUAL;
break;
case Triple::x86:
case Triple::x86_64:
AS = EDDisassembler::kEDAssemblySyntaxX86ATT;
break;
}
OwningPtr<EDDisassembler>
disassembler(EDDisassembler::getDisassembler(TS.c_str(), AS));
if (disassembler == 0) {
errs() << "error: couldn't get disassembler for " << TS << '\n';
return -1;
}
while (ByteArray.size()) {
OwningPtr<EDInst>
inst(disassembler->createInst(byteArrayReader, 0, &ByteArray));
if (inst == 0) {
errs() << "error: Didn't get an instruction\n";
return -1;
}
ByteArray.erase (ByteArray.begin(), ByteArray.begin() + inst->byteSize());
unsigned numTokens = inst->numTokens();
if ((int)numTokens < 0) {
errs() << "error: couldn't count the instruction's tokens\n";
return -1;
}
for (unsigned tokenIndex = 0; tokenIndex != numTokens; ++tokenIndex) {
EDToken *token;
if (inst->getToken(token, tokenIndex)) {
errs() << "error: Couldn't get token\n";
return -1;
}
const char *buf;
if (token->getString(buf)) {
errs() << "error: Couldn't get string for token\n";
return -1;
}
Out << '[';
int operandIndex = token->operandID();
if (operandIndex >= 0)
Out << operandIndex << "-";
switch (token->type()) {
case EDToken::kTokenWhitespace: Out << "w"; break;
case EDToken::kTokenPunctuation: Out << "p"; break;
case EDToken::kTokenOpcode: Out << "o"; break;
case EDToken::kTokenLiteral: Out << "l"; break;
case EDToken::kTokenRegister: Out << "r"; break;
}
Out << ":" << buf;
if (token->type() == EDToken::kTokenLiteral) {
Out << "=";
if (token->literalSign())
Out << "-";
uint64_t absoluteValue;
if (token->literalAbsoluteValue(absoluteValue)) {
errs() << "error: Couldn't get the value of a literal token\n";
return -1;
}
Out << absoluteValue;
} else if (token->type() == EDToken::kTokenRegister) {
Out << "=";
unsigned regID;
if (token->registerID(regID)) {
errs() << "error: Couldn't get the ID of a register token\n";
return -1;
}
Out << "r" << regID;
}
Out << "]";
}
Out << " ";
if (inst->isBranch())
Out << "<br> ";
if (inst->isMove())
Out << "<mov> ";
unsigned numOperands = inst->numOperands();
if ((int)numOperands < 0) {
errs() << "error: Couldn't count operands\n";
return -1;
}
for (unsigned operandIndex = 0; operandIndex != numOperands;
++operandIndex) {
Out << operandIndex << ":";
EDOperand *operand;
if (inst->getOperand(operand, operandIndex)) {
errs() << "error: couldn't get operand\n";
return -1;
}
uint64_t evaluatedResult;
void *Arg[] = { disassembler.get(), &Out };
if (operand->evaluate(evaluatedResult, verboseEvaluator, Arg)) {
errs() << "error: Couldn't evaluate an operand\n";
return -1;
}
Out << "=" << evaluatedResult << " ";
}
Out << '\n';
}
return 0;
}

5
tools/llvm-mc/Disassembler.h
View File

@ -35,11 +35,6 @@ public:
MemoryBuffer &Buffer,
SourceMgr &SM,
raw_ostream &Out);
static int disassembleEnhanced(const std::string &tripleString,
MemoryBuffer &buffer,
SourceMgr &SM,
raw_ostream &Out);
};
} // namespace llvm

6
tools/llvm-mc/llvm-mc.cpp
View File

@ -169,7 +169,6 @@ enum ActionType {
AC_AsLex,
AC_Assemble,
AC_Disassemble,
AC_EDisassemble,
AC_MDisassemble,
AC_HDisassemble
};
@ -183,8 +182,6 @@ Action(cl::desc("Action to perform:"),
"Assemble a .s file (default)"),
clEnumValN(AC_Disassemble, "disassemble",
"Disassemble strings of hex bytes"),
clEnumValN(AC_EDisassemble, "edis",
"Enhanced disassembly of strings of hex bytes"),
clEnumValN(AC_MDisassemble, "mdis",
"Marked up disassembly of strings of hex bytes"),
clEnumValN(AC_HDisassemble, "hdis",
@ -472,9 +469,6 @@ int main(int argc, char **argv) {
case AC_Disassemble:
disassemble = true;
break;
case AC_EDisassemble:
Res = Disassembler::disassembleEnhanced(TripleName, *Buffer, SrcMgr, Out->os());
break;
}
if (disassemble)
Res = Disassembler::disassemble(*TheTarget, TripleName, *STI, *Str,

1
utils/TableGen/CMakeLists.txt
View File

@ -19,7 +19,6 @@ add_tablegen(llvm-tblgen LLVM
DAGISelMatcher.cpp
DFAPacketizerEmitter.cpp
DisassemblerEmitter.cpp
EDEmitter.cpp
FastISelEmitter.cpp
FixedLenDecoderEmitter.cpp
InstrInfoEmitter.cpp

1012
utils/TableGen/EDEmitter.cpp
View File

File diff suppressed because it is too large Load Diff

6
utils/TableGen/TableGen.cpp
View File

@ -38,7 +38,6 @@ enum ActionType {
GenSubtarget,
GenIntrinsic,
GenTgtIntrinsic,
GenEDInfo,
PrintEnums,
PrintSets,
GenOptParserDefs
@ -77,8 +76,6 @@ namespace {
"Generate intrinsic information"),
clEnumValN(GenTgtIntrinsic, "gen-tgt-intrinsic",
"Generate target intrinsic information"),
clEnumValN(GenEDInfo, "gen-enhanced-disassembly-info",
"Generate enhanced disassembly info"),
clEnumValN(PrintEnums, "print-enums",
"Print enum values for a class"),
clEnumValN(PrintSets, "print-sets",
@ -138,9 +135,6 @@ bool LLVMTableGenMain(raw_ostream &OS, RecordKeeper &Records) {
case GenTgtIntrinsic:
EmitIntrinsics(Records, OS, true);
break;
case GenEDInfo:
EmitEnhancedDisassemblerInfo(Records, OS);
break;
case GenOptParserDefs:
EmitOptParser(Records, OS);
break;

1
utils/TableGen/TableGenBackends.h
View File

@ -68,7 +68,6 @@ void EmitCodeEmitter(RecordKeeper &RK, raw_ostream &OS);
void EmitDAGISel(RecordKeeper &RK, raw_ostream &OS);
void EmitDFAPacketizer(RecordKeeper &RK, raw_ostream &OS);
void EmitDisassembler(RecordKeeper &RK, raw_ostream &OS);
void EmitEnhancedDisassemblerInfo(RecordKeeper &RK, raw_ostream &OS);
void EmitFastISel(RecordKeeper &RK, raw_ostream &OS);
void EmitInstrInfo(RecordKeeper &RK, raw_ostream &OS);
void EmitPseudoLowering(RecordKeeper &RK, raw_ostream &OS);