1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-26 12:43:36 +01:00
llvm-mirror/tools/llvm-mc/MC-X86Specific.cpp
Daniel Dunbar 0afb18f239 llvm-mc/x86: Fix various nit-picky bugs in displacement parsing.
- Test case to follow.

llvm-svn: 74687
2009-07-02 02:26:39 +00:00

267 lines
7.9 KiB
C++

//===- MC-X86Specific.cpp - X86-Specific code for MC ----------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements X86-specific parsing, encoding and decoding stuff for
// MC.
//
//===----------------------------------------------------------------------===//
#include "AsmParser.h"
#include "llvm/MC/MCInst.h"
#include "llvm/Support/SourceMgr.h"
using namespace llvm;
/// X86Operand - Instances of this class represent one X86 machine instruction.
struct AsmParser::X86Operand {
enum {
Register,
Immediate,
Memory
} Kind;
union {
struct {
unsigned RegNo;
} Reg;
struct {
MCValue Val;
} Imm;
struct {
unsigned SegReg;
MCValue Disp;
unsigned BaseReg;
unsigned IndexReg;
unsigned Scale;
} Mem;
};
unsigned getReg() const {
assert(Kind == Register && "Invalid access!");
return Reg.RegNo;
}
static X86Operand CreateReg(unsigned RegNo) {
X86Operand Res;
Res.Kind = Register;
Res.Reg.RegNo = RegNo;
return Res;
}
static X86Operand CreateImm(MCValue Val) {
X86Operand Res;
Res.Kind = Immediate;
Res.Imm.Val = Val;
return Res;
}
static X86Operand CreateMem(unsigned SegReg, MCValue Disp, unsigned BaseReg,
unsigned IndexReg, unsigned Scale) {
// If there is no index register, we should never have a scale, and we
// should always have a scale (in {1,2,4,8}) if we do.
assert(((Scale == 0 && !IndexReg) ||
(IndexReg && (Scale == 1 || Scale == 2 ||
Scale == 4 || Scale == 8))) &&
"Invalid scale!");
X86Operand Res;
Res.Kind = Memory;
Res.Mem.SegReg = SegReg;
Res.Mem.Disp = Disp;
Res.Mem.BaseReg = BaseReg;
Res.Mem.IndexReg = IndexReg;
Res.Mem.Scale = Scale;
return Res;
}
};
bool AsmParser::ParseX86Register(X86Operand &Op) {
assert(Lexer.getKind() == asmtok::Register && "Invalid token kind!");
// FIXME: Decode register number.
Op = X86Operand::CreateReg(123);
Lexer.Lex(); // Eat register token.
return false;
}
bool AsmParser::ParseX86Operand(X86Operand &Op) {
switch (Lexer.getKind()) {
default:
return ParseX86MemOperand(Op);
case asmtok::Register:
// FIXME: if a segment register, this could either be just the seg reg, or
// the start of a memory operand.
return ParseX86Register(Op);
case asmtok::Dollar: {
// $42 -> immediate.
Lexer.Lex();
MCValue Val;
if (ParseRelocatableExpression(Val))
return true;
Op = X86Operand::CreateImm(Val);
return false;
}
case asmtok::Star: {
Lexer.Lex(); // Eat the star.
if (Lexer.is(asmtok::Register)) {
if (ParseX86Register(Op))
return true;
} else if (ParseX86MemOperand(Op))
return true;
// FIXME: Note the '*' in the operand for use by the matcher.
return false;
}
}
}
/// ParseX86MemOperand: segment: disp(basereg, indexreg, scale)
bool AsmParser::ParseX86MemOperand(X86Operand &Op) {
// FIXME: If SegReg ':' (e.g. %gs:), eat and remember.
unsigned SegReg = 0;
// We have to disambiguate a parenthesized expression "(4+5)" from the start
// of a memory operand with a missing displacement "(%ebx)" or "(,%eax)". The
// only way to do this without lookahead is to eat the ( and see what is after
// it.
MCValue Disp = MCValue::get(0, 0, 0);
if (Lexer.isNot(asmtok::LParen)) {
if (ParseRelocatableExpression(Disp)) return true;
// After parsing the base expression we could either have a parenthesized
// memory address or not. If not, return now. If so, eat the (.
if (Lexer.isNot(asmtok::LParen)) {
Op = X86Operand::CreateMem(SegReg, Disp, 0, 0, 0);
return false;
}
// Eat the '('.
Lexer.Lex();
} else {
// Okay, we have a '('. We don't know if this is an expression or not, but
// so we have to eat the ( to see beyond it.
Lexer.Lex(); // Eat the '('.
if (Lexer.is(asmtok::Register) || Lexer.is(asmtok::Comma)) {
// Nothing to do here, fall into the code below with the '(' part of the
// memory operand consumed.
} else {
// It must be an parenthesized expression, parse it now.
if (ParseParenRelocatableExpression(Disp))
return true;
// After parsing the base expression we could either have a parenthesized
// memory address or not. If not, return now. If so, eat the (.
if (Lexer.isNot(asmtok::LParen)) {
Op = X86Operand::CreateMem(SegReg, Disp, 0, 0, 0);
return false;
}
// Eat the '('.
Lexer.Lex();
}
}
// If we reached here, then we just ate the ( of the memory operand. Process
// the rest of the memory operand.
unsigned BaseReg = 0, IndexReg = 0, Scale = 0;
if (Lexer.is(asmtok::Register)) {
if (ParseX86Register(Op))
return true;
BaseReg = Op.getReg();
}
if (Lexer.is(asmtok::Comma)) {
Lexer.Lex(); // Eat the comma.
// Following the comma we should have either an index register, or a scale
// value. We don't support the later form, but we want to parse it
// correctly.
//
// Not that even though it would be completely consistent to support syntax
// like "1(%eax,,1)", the assembler doesn't.
if (Lexer.is(asmtok::Register)) {
if (ParseX86Register(Op))
return true;
IndexReg = Op.getReg();
Scale = 1; // If not specified, the scale defaults to 1.
if (Lexer.isNot(asmtok::RParen)) {
// Parse the scale amount:
// ::= ',' [scale-expression]
if (Lexer.isNot(asmtok::Comma))
return true;
Lexer.Lex(); // Eat the comma.
if (Lexer.isNot(asmtok::RParen)) {
int64_t ScaleVal;
if (ParseAbsoluteExpression(ScaleVal))
return true;
// Validate the scale amount.
if (ScaleVal != 1 && ScaleVal != 2 && ScaleVal != 4 && ScaleVal != 8)
return TokError("scale factor in address must be 1, 2, 4 or 8");
Scale = (unsigned)ScaleVal;
}
}
} else if (Lexer.isNot(asmtok::RParen)) {
// Otherwise we have the unsupported form of a scale amount without an
// index.
SMLoc Loc = Lexer.getLoc();
int64_t Value;
if (ParseAbsoluteExpression(Value))
return true;
return Error(Loc, "cannot have scale factor without index register");
}
}
// Ok, we've eaten the memory operand, verify we have a ')' and eat it too.
if (Lexer.isNot(asmtok::RParen))
return TokError("unexpected token in memory operand");
Lexer.Lex(); // Eat the ')'.
Op = X86Operand::CreateMem(SegReg, Disp, BaseReg, IndexReg, Scale);
return false;
}
/// MatchX86Inst - Convert a parsed instruction name and operand list into a
/// concrete instruction.
static bool MatchX86Inst(const char *Name,
llvm::SmallVector<AsmParser::X86Operand, 3> &Operands,
MCInst &Inst) {
return false;
}
/// ParseX86InstOperands - Parse the operands of an X86 instruction and return
/// them as the operands of an MCInst.
bool AsmParser::ParseX86InstOperands(const char *InstName, MCInst &Inst) {
llvm::SmallVector<X86Operand, 3> Operands;
if (Lexer.isNot(asmtok::EndOfStatement)) {
// Read the first operand.
Operands.push_back(X86Operand());
if (ParseX86Operand(Operands.back()))
return true;
while (Lexer.is(asmtok::Comma)) {
Lexer.Lex(); // Eat the comma.
// Parse and remember the operand.
Operands.push_back(X86Operand());
if (ParseX86Operand(Operands.back()))
return true;
}
}
return MatchX86Inst(InstName, Operands, Inst);
}