dart-lang/sdk
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity
Files
c9bcfe877b43391af4e470fec00ef036557a7fee
sdk/runtime/vm/compiler/assembler/assembler_x64.h
T
Ryan Macnak ebca3aba60 [vm, gc] Concurrent marking barrier. 
Code size impact on flutter_gallery after enabled:
ARM32 Instructions(CodeSize): 5892064 -> 6110592 (+3.71%)
ARM64 Instructions(CodeSize): 6307104 -> 6514528 (+3.28%)

Bug: https://github.com/dart-lang/sdk/issues/34002
Change-Id: If093f24e4dc6bf29f407cc45e95bb2274fc53dce
Reviewed-on: https://dart-review.googlesource.com/68481
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
2018-09-11 21:14:25 +00:00

1145 lines
40 KiB
C++
Raw Blame History

// Copyright (c) 2013, the Dart project authors. Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.
#ifndef RUNTIME_VM_COMPILER_ASSEMBLER_ASSEMBLER_X64_H_
#define RUNTIME_VM_COMPILER_ASSEMBLER_ASSEMBLER_X64_H_
#ifndef RUNTIME_VM_COMPILER_ASSEMBLER_ASSEMBLER_H_
#error Do not include assembler_x64.h directly; use assembler.h instead.
#endif
#include "platform/assert.h"
#include "platform/utils.h"
#include "vm/constants_x64.h"
#include "vm/constants_x86.h"
#include "vm/hash_map.h"
#include "vm/object.h"
namespace dart {
// Forward declarations.
class RuntimeEntry;
class StubEntry;
class Immediate : public ValueObject {
public:
explicit Immediate(int64_t value) : value_(value) {}
Immediate(const Immediate& other) : ValueObject(), value_(other.value_) {}
int64_t value() const { return value_; }
bool is_int8() const { return Utils::IsInt(8, value_); }
bool is_uint8() const { return Utils::IsUint(8, value_); }
bool is_int16() const { return Utils::IsInt(16, value_); }
bool is_uint16() const { return Utils::IsUint(16, value_); }
bool is_int32() const { return Utils::IsInt(32, value_); }
bool is_uint32() const { return Utils::IsUint(32, value_); }
private:
const int64_t value_;
// TODO(5411081): Add DISALLOW_COPY_AND_ASSIGN(Immediate) once the mac
// build issue is resolved.
// And remove the unnecessary copy constructor.
};
class Operand : public ValueObject {
public:
uint8_t rex() const { return rex_; }
uint8_t mod() const { return (encoding_at(0) >> 6) & 3; }
Register rm() const {
int rm_rex = (rex_ & REX_B) << 3;
return static_cast<Register>(rm_rex + (encoding_at(0) & 7));
}
ScaleFactor scale() const {
return static_cast<ScaleFactor>((encoding_at(1) >> 6) & 3);
}
Register index() const {
int index_rex = (rex_ & REX_X) << 2;
return static_cast<Register>(index_rex + ((encoding_at(1) >> 3) & 7));
}
Register base() const {
int base_rex = (rex_ & REX_B) << 3;
return static_cast<Register>(base_rex + (encoding_at(1) & 7));
}
int8_t disp8() const {
ASSERT(length_ >= 2);
return static_cast<int8_t>(encoding_[length_ - 1]);
}
int32_t disp32() const {
ASSERT(length_ >= 5);
return bit_copy<int32_t>(encoding_[length_ - 4]);
}
Operand(const Operand& other)
: ValueObject(), length_(other.length_), rex_(other.rex_) {
memmove(&encoding_[0], &other.encoding_[0], other.length_);
}
Operand& operator=(const Operand& other) {
length_ = other.length_;
rex_ = other.rex_;
memmove(&encoding_[0], &other.encoding_[0], other.length_);
return *this;
}
bool Equals(const Operand& other) const {
if (length_ != other.length_) return false;
if (rex_ != other.rex_) return false;
for (uint8_t i = 0; i < length_; i++) {
if (encoding_[i] != other.encoding_[i]) return false;
}
return true;
}
protected:
Operand() : length_(0), rex_(REX_NONE) {} // Needed by subclass Address.
void SetModRM(int mod, Register rm) {
ASSERT((mod & ~3) == 0);
if ((rm > 7) && !((rm == R12) && (mod != 3))) {
rex_ |= REX_B;
}
encoding_[0] = (mod << 6) | (rm & 7);
length_ = 1;
}
void SetSIB(ScaleFactor scale, Register index, Register base) {
ASSERT(length_ == 1);
ASSERT((scale & ~3) == 0);
if (base > 7) {
ASSERT((rex_ & REX_B) == 0); // Must not have REX.B already set.
rex_ |= REX_B;
}
if (index > 7) rex_ |= REX_X;
encoding_[1] = (scale << 6) | ((index & 7) << 3) | (base & 7);
length_ = 2;
}
void SetDisp8(int8_t disp) {
ASSERT(length_ == 1 || length_ == 2);
encoding_[length_++] = static_cast<uint8_t>(disp);
}
void SetDisp32(int32_t disp) {
ASSERT(length_ == 1 || length_ == 2);
memmove(&encoding_[length_], &disp, sizeof(disp));
length_ += sizeof(disp);
}
private:
uint8_t length_;
uint8_t rex_;
uint8_t encoding_[6];
explicit Operand(Register reg) : rex_(REX_NONE) { SetModRM(3, reg); }
// Get the operand encoding byte at the given index.
uint8_t encoding_at(intptr_t index) const {
ASSERT(index >= 0 && index < length_);
return encoding_[index];
}
// Returns whether or not this operand is really the given register in
// disguise. Used from the assembler to generate better encodings.
bool IsRegister(Register reg) const {
return ((reg > 7 ? 1 : 0) == (rex_ & REX_B)) // REX.B match.
&& ((encoding_at(0) & 0xF8) == 0xC0) // Addressing mode is register.
&& ((encoding_at(0) & 0x07) == reg); // Register codes match.
}
friend class Assembler;
};
class Address : public Operand {
public:
Address(Register base, int32_t disp) {
if ((disp == 0) && ((base & 7) != RBP)) {
SetModRM(0, base);
if ((base & 7) == RSP) {
SetSIB(TIMES_1, RSP, base);
}
} else if (Utils::IsInt(8, disp)) {
SetModRM(1, base);
if ((base & 7) == RSP) {
SetSIB(TIMES_1, RSP, base);
}
SetDisp8(disp);
} else {
SetModRM(2, base);
if ((base & 7) == RSP) {
SetSIB(TIMES_1, RSP, base);
}
SetDisp32(disp);
}
}
// This addressing mode does not exist.
Address(Register base, Register r);
Address(Register index, ScaleFactor scale, int32_t disp) {
ASSERT(index != RSP); // Illegal addressing mode.
SetModRM(0, RSP);
SetSIB(scale, index, RBP);
SetDisp32(disp);
}
// This addressing mode does not exist.
Address(Register index, ScaleFactor scale, Register r);
Address(Register base, Register index, ScaleFactor scale, int32_t disp) {
ASSERT(index != RSP); // Illegal addressing mode.
if ((disp == 0) && ((base & 7) != RBP)) {
SetModRM(0, RSP);
SetSIB(scale, index, base);
} else if (Utils::IsInt(8, disp)) {
SetModRM(1, RSP);
SetSIB(scale, index, base);
SetDisp8(disp);
} else {
SetModRM(2, RSP);
SetSIB(scale, index, base);
SetDisp32(disp);
}
}
// This addressing mode does not exist.
Address(Register base, Register index, ScaleFactor scale, Register r);
Address(const Address& other) : Operand(other) {}
Address& operator=(const Address& other) {
Operand::operator=(other);
return *this;
}
static Address AddressRIPRelative(int32_t disp) {
return Address(RIPRelativeDisp(disp));
}
static Address AddressBaseImm32(Register base, int32_t disp) {
return Address(base, disp, true);
}
// This addressing mode does not exist.
static Address AddressBaseImm32(Register base, Register r);
private:
Address(Register base, int32_t disp, bool fixed) {
ASSERT(fixed);
SetModRM(2, base);
if ((base & 7) == RSP) {
SetSIB(TIMES_1, RSP, base);
}
SetDisp32(disp);
}
struct RIPRelativeDisp {
explicit RIPRelativeDisp(int32_t disp) : disp_(disp) {}
const int32_t disp_;
};
explicit Address(const RIPRelativeDisp& disp) {
SetModRM(0, static_cast<Register>(0x5));
SetDisp32(disp.disp_);
}
};
class FieldAddress : public Address {
public:
FieldAddress(Register base, int32_t disp)
: Address(base, disp - kHeapObjectTag) {}
// This addressing mode does not exist.
FieldAddress(Register base, Register r);
FieldAddress(Register base, Register index, ScaleFactor scale, int32_t disp)
: Address(base, index, scale, disp - kHeapObjectTag) {}
// This addressing mode does not exist.
FieldAddress(Register base, Register index, ScaleFactor scale, Register r);
FieldAddress(const FieldAddress& other) : Address(other) {}
FieldAddress& operator=(const FieldAddress& other) {
Address::operator=(other);
return *this;
}
};
class Assembler : public ValueObject {
public:
explicit Assembler(ObjectPoolWrapper* object_pool_wrapper,
bool use_far_branches = false);
~Assembler() {}
static const bool kNearJump = true;
static const bool kFarJump = false;
/*
* Emit Machine Instructions.
*/
void call(Register reg) { EmitUnaryL(reg, 0xFF, 2); }
void call(const Address& address) { EmitUnaryL(address, 0xFF, 2); }
void call(Label* label);
void call(const ExternalLabel* label);
static const intptr_t kCallExternalLabelSize = 15;
void pushq(Register reg);
void pushq(const Address& address) { EmitUnaryL(address, 0xFF, 6); }
void pushq(const Immediate& imm);
void PushImmediate(const Immediate& imm);
void popq(Register reg);
void popq(const Address& address) { EmitUnaryL(address, 0x8F, 0); }
void setcc(Condition condition, ByteRegister dst);
// Register-register, register-address and address-register instructions.
#define RR(width, name, ...) \
void name(Register dst, Register src) { Emit##width(dst, src, __VA_ARGS__); }
#define RA(width, name, ...) \
void name(Register dst, const Address& src) { \
Emit##width(dst, src, __VA_ARGS__); \
}
#define AR(width, name, ...) \
void name(const Address& dst, Register src) { \
Emit##width(src, dst, __VA_ARGS__); \
}
#define REGULAR_INSTRUCTION(name, ...) \
RA(W, name##w, __VA_ARGS__) \
RA(L, name##l, __VA_ARGS__) \
RA(Q, name##q, __VA_ARGS__) \
RR(W, name##w, __VA_ARGS__) \
RR(L, name##l, __VA_ARGS__) \
RR(Q, name##q, __VA_ARGS__)
REGULAR_INSTRUCTION(test, 0x85)
REGULAR_INSTRUCTION(xchg, 0x87)
REGULAR_INSTRUCTION(imul, 0xAF, 0x0F)
REGULAR_INSTRUCTION(bsr, 0xBD, 0x0F)
#undef REGULAR_INSTRUCTION
RA(Q, movsxd, 0x63)
RR(Q, movsxd, 0x63)
AR(L, movb, 0x88)
AR(L, movl, 0x89)
AR(Q, movq, 0x89)
AR(W, movw, 0x89)
RA(L, movb, 0x8A)
RA(L, movl, 0x8B)
RA(Q, movq, 0x8B)
RR(L, movl, 0x8B)
RA(Q, leaq, 0x8D)
RA(L, leal, 0x8D)
AR(L, cmpxchgl, 0xB1, 0x0F)
AR(Q, cmpxchgq, 0xB1, 0x0F)
RA(L, cmpxchgl, 0xB1, 0x0F)
RA(Q, cmpxchgq, 0xB1, 0x0F)
RR(L, cmpxchgl, 0xB1, 0x0F)
RR(Q, cmpxchgq, 0xB1, 0x0F)
RA(Q, movzxb, 0xB6, 0x0F)
RR(Q, movzxb, 0xB6, 0x0F)
RA(Q, movzxw, 0xB7, 0x0F)
RR(Q, movzxw, 0xB7, 0x0F)
RA(Q, movsxb, 0xBE, 0x0F)
RR(Q, movsxb, 0xBE, 0x0F)
RA(Q, movsxw, 0xBF, 0x0F)
RR(Q, movsxw, 0xBF, 0x0F)
#define DECLARE_CMOV(name, code) \
RR(Q, cmov##name##q, 0x40 + code, 0x0F) \
RR(L, cmov##name##l, 0x40 + code, 0x0F) \
RA(Q, cmov##name##q, 0x40 + code, 0x0F) \
RA(L, cmov##name##l, 0x40 + code, 0x0F)
X86_CONDITIONAL_SUFFIXES(DECLARE_CMOV)
#undef DECLARE_CMOV
#undef AA
#undef RA
#undef AR
#define SIMPLE(name, ...) \
void name() { EmitSimple(__VA_ARGS__); }
SIMPLE(cpuid, 0x0F, 0xA2)
SIMPLE(fcos, 0xD9, 0xFF)
SIMPLE(fincstp, 0xD9, 0xF7)
SIMPLE(fsin, 0xD9, 0xFE)
SIMPLE(lock, 0xF0)
SIMPLE(rep_movsb, 0xF3, 0xA4)
#undef SIMPLE
// XmmRegister operations with another register or an address.
#define XX(width, name, ...) \
void name(XmmRegister dst, XmmRegister src) { \
Emit##width(dst, src, __VA_ARGS__); \
}
#define XA(width, name, ...) \
void name(XmmRegister dst, const Address& src) { \
Emit##width(dst, src, __VA_ARGS__); \
}
#define AX(width, name, ...) \
void name(const Address& dst, XmmRegister src) { \
Emit##width(src, dst, __VA_ARGS__); \
}
// We could add movupd here, but movups does the same and is shorter.
XA(L, movups, 0x10, 0x0F);
XA(L, movsd, 0x10, 0x0F, 0xF2)
XA(L, movss, 0x10, 0x0F, 0xF3)
AX(L, movups, 0x11, 0x0F);
AX(L, movsd, 0x11, 0x0F, 0xF2)
AX(L, movss, 0x11, 0x0F, 0xF3)
XX(L, movhlps, 0x12, 0x0F)
XX(L, unpcklps, 0x14, 0x0F)
XX(L, unpcklpd, 0x14, 0x0F, 0x66)
XX(L, unpckhps, 0x15, 0x0F)
XX(L, unpckhpd, 0x15, 0x0F, 0x66)
XX(L, movlhps, 0x16, 0x0F)
XX(L, movaps, 0x28, 0x0F)
XX(L, comisd, 0x2F, 0x0F, 0x66)
#define DECLARE_XMM(name, code) \
XX(L, name##ps, 0x50 + code, 0x0F) \
XA(L, name##ps, 0x50 + code, 0x0F) \
AX(L, name##ps, 0x50 + code, 0x0F) \
XX(L, name##pd, 0x50 + code, 0x0F, 0x66) \
XA(L, name##pd, 0x50 + code, 0x0F, 0x66) \
AX(L, name##pd, 0x50 + code, 0x0F, 0x66) \
XX(L, name##sd, 0x50 + code, 0x0F, 0xF2) \
XA(L, name##sd, 0x50 + code, 0x0F, 0xF2) \
AX(L, name##sd, 0x50 + code, 0x0F, 0xF2) \
XX(L, name##ss, 0x50 + code, 0x0F, 0xF3) \
XA(L, name##ss, 0x50 + code, 0x0F, 0xF3) \
AX(L, name##ss, 0x50 + code, 0x0F, 0xF3)
XMM_ALU_CODES(DECLARE_XMM)
#undef DECLARE_XMM
XX(L, cvtps2pd, 0x5A, 0x0F)
XX(L, cvtpd2ps, 0x5A, 0x0F, 0x66)
XX(L, cvtsd2ss, 0x5A, 0x0F, 0xF2)
XX(L, cvtss2sd, 0x5A, 0x0F, 0xF3)
XX(L, pxor, 0xEF, 0x0F, 0x66)
XX(L, subpl, 0xFA, 0x0F, 0x66)
XX(L, addpl, 0xFE, 0x0F, 0x66)
#undef XX
#undef AX
#undef XA
#define DECLARE_CMPPS(name, code) \
void cmpps##name(XmmRegister dst, XmmRegister src) { \
EmitL(dst, src, 0xC2, 0x0F); \
AssemblerBuffer::EnsureCapacity ensured(&buffer_); \
EmitUint8(code); \
}
XMM_CONDITIONAL_CODES(DECLARE_CMPPS)
#undef DECLARE_CMPPS
#define DECLARE_SIMPLE(name, opcode) \
void name() { EmitSimple(opcode); }
X86_ZERO_OPERAND_1_BYTE_INSTRUCTIONS(DECLARE_SIMPLE)
#undef DECLARE_SIMPLE
void movl(Register dst, const Immediate& imm);
void movl(const Address& dst, const Immediate& imm);
void movb(const Address& dst, const Immediate& imm);
void movw(Register dst, const Address& src);
void movw(const Address& dst, const Immediate& imm);
void movq(Register dst, const Immediate& imm);
void movq(const Address& dst, const Immediate& imm);
// Destination and source are reversed for some reason.
void movq(Register dst, XmmRegister src) {
EmitQ(src, dst, 0x7E, 0x0F, 0x66);
}
void movl(Register dst, XmmRegister src) {
EmitL(src, dst, 0x7E, 0x0F, 0x66);
}
void movss(XmmRegister dst, XmmRegister src) {
EmitL(src, dst, 0x11, 0x0F, 0xF3);
}
void movsd(XmmRegister dst, XmmRegister src) {
EmitL(src, dst, 0x11, 0x0F, 0xF2);
}
// Use the reversed operand order and the 0x89 bytecode instead of the
// obvious 0x88 encoding for this some, because it is expected by gdb64 older
// than 7.3.1-gg5 when disassembling a function's prologue (movq rbp, rsp)
// for proper unwinding of Dart frames (use --generate_gdb_symbols and -O0).
void movq(Register dst, Register src) { EmitQ(src, dst, 0x89); }
void movd(XmmRegister dst, Register src) {
EmitL(dst, src, 0x6E, 0x0F, 0x66);
}
void cvtsi2sdq(XmmRegister dst, Register src) {
EmitQ(dst, src, 0x2A, 0x0F, 0xF2);
}
void cvtsi2sdl(XmmRegister dst, Register src) {
EmitL(dst, src, 0x2A, 0x0F, 0xF2);
}
void cvttsd2siq(Register dst, XmmRegister src) {
EmitQ(dst, src, 0x2C, 0x0F, 0xF2);
}
void cvttsd2sil(Register dst, XmmRegister src) {
EmitL(dst, src, 0x2C, 0x0F, 0xF2);
}
void movmskpd(Register dst, XmmRegister src) {
EmitL(dst, src, 0x50, 0x0F, 0x66);
}
void movmskps(Register dst, XmmRegister src) { EmitL(dst, src, 0x50, 0x0F); }
void btl(Register dst, Register src) { EmitL(src, dst, 0xA3, 0x0F); }
void btq(Register dst, Register src) { EmitQ(src, dst, 0xA3, 0x0F); }
void notps(XmmRegister dst, XmmRegister src);
void negateps(XmmRegister dst, XmmRegister src);
void absps(XmmRegister dst, XmmRegister src);
void zerowps(XmmRegister dst, XmmRegister src);
void set1ps(XmmRegister dst, Register tmp, const Immediate& imm);
void shufps(XmmRegister dst, XmmRegister src, const Immediate& mask);
void negatepd(XmmRegister dst, XmmRegister src);
void abspd(XmmRegister dst, XmmRegister src);
void shufpd(XmmRegister dst, XmmRegister src, const Immediate& mask);
enum RoundingMode {
kRoundToNearest = 0x0,
kRoundDown = 0x1,
kRoundUp = 0x2,
kRoundToZero = 0x3
};
void roundsd(XmmRegister dst, XmmRegister src, RoundingMode mode);
void CompareImmediate(Register reg, const Immediate& imm);
void CompareImmediate(const Address& address, const Immediate& imm);
void CompareImmediate(Register reg, int32_t immediate) {
return CompareImmediate(reg, Immediate(immediate));
}
void testl(Register reg, const Immediate& imm) { testq(reg, imm); }
void testb(const Address& address, const Immediate& imm);
void testb(const Address& address, Register reg);
void testq(Register reg, const Immediate& imm);
void TestImmediate(Register dst, const Immediate& imm);
void AndImmediate(Register dst, const Immediate& imm);
void OrImmediate(Register dst, const Immediate& imm);
void XorImmediate(Register dst, const Immediate& imm);
void shldq(Register dst, Register src, Register shifter) {
ASSERT(shifter == RCX);
EmitQ(src, dst, 0xA5, 0x0F);
}
void shrdq(Register dst, Register src, Register shifter) {
ASSERT(shifter == RCX);
EmitQ(src, dst, 0xAD, 0x0F);
}
#define DECLARE_ALU(op, c) \
void op##w(Register dst, Register src) { EmitW(dst, src, c * 8 + 3); } \
void op##l(Register dst, Register src) { EmitL(dst, src, c * 8 + 3); } \
void op##q(Register dst, Register src) { EmitQ(dst, src, c * 8 + 3); } \
void op##w(Register dst, const Address& src) { EmitW(dst, src, c * 8 + 3); } \
void op##l(Register dst, const Address& src) { EmitL(dst, src, c * 8 + 3); } \
void op##q(Register dst, const Address& src) { EmitQ(dst, src, c * 8 + 3); } \
void op##w(const Address& dst, Register src) { EmitW(src, dst, c * 8 + 1); } \
void op##l(const Address& dst, Register src) { EmitL(src, dst, c * 8 + 1); } \
void op##q(const Address& dst, Register src) { EmitQ(src, dst, c * 8 + 1); } \
void op##l(Register dst, const Immediate& imm) { AluL(c, dst, imm); } \
void op##q(Register dst, const Immediate& imm) { \
AluQ(c, c * 8 + 3, dst, imm); \
} \
void op##b(const Address& dst, const Immediate& imm) { AluB(c, dst, imm); } \
void op##w(const Address& dst, const Immediate& imm) { AluW(c, dst, imm); } \
void op##l(const Address& dst, const Immediate& imm) { AluL(c, dst, imm); } \
void op##q(const Address& dst, const Immediate& imm) { \
AluQ(c, c * 8 + 3, dst, imm); \
}
X86_ALU_CODES(DECLARE_ALU)
#undef DECLARE_ALU
#undef ALU_OPS
void cqo();
#define REGULAR_UNARY(name, opcode, modrm) \
void name##q(Register reg) { EmitUnaryQ(reg, opcode, modrm); } \
void name##l(Register reg) { EmitUnaryL(reg, opcode, modrm); } \
void name##q(const Address& address) { EmitUnaryQ(address, opcode, modrm); } \
void name##l(const Address& address) { EmitUnaryL(address, opcode, modrm); }
REGULAR_UNARY(not, 0xF7, 2)
REGULAR_UNARY(neg, 0xF7, 3)
REGULAR_UNARY(mul, 0xF7, 4)
REGULAR_UNARY(div, 0xF7, 6)
REGULAR_UNARY(idiv, 0xF7, 7)
REGULAR_UNARY(inc, 0xFF, 0)
REGULAR_UNARY(dec, 0xFF, 1)
#undef REGULAR_UNARY
// We could use kWord, kDoubleWord, and kQuadWord here, but it is rather
// confusing since the same sizes mean something different on ARM.
enum OperandWidth { k32Bit, k64Bit };
void imull(Register reg, const Immediate& imm);
void imulq(Register dst, const Immediate& imm);
void MulImmediate(Register reg,
const Immediate& imm,
OperandWidth width = k64Bit);
void shll(Register reg, const Immediate& imm);
void shll(Register operand, Register shifter);
void shrl(Register reg, const Immediate& imm);
void shrl(Register operand, Register shifter);
void sarl(Register reg, const Immediate& imm);
void sarl(Register operand, Register shifter);
void shldl(Register dst, Register src, const Immediate& imm);
void shlq(Register reg, const Immediate& imm);
void shlq(Register operand, Register shifter);
void shrq(Register reg, const Immediate& imm);
void shrq(Register operand, Register shifter);
void sarq(Register reg, const Immediate& imm);
void sarq(Register operand, Register shifter);
void shldq(Register dst, Register src, const Immediate& imm);
void btq(Register base, int bit);
void enter(const Immediate& imm);
void fldl(const Address& src);
void fstpl(const Address& dst);
void ffree(intptr_t value);
// 'size' indicates size in bytes and must be in the range 1..8.
void nop(int size = 1);
static uword GetBreakInstructionFiller() { return 0xCCCCCCCCCCCCCCCC; }
void j(Condition condition, Label* label, bool near = kFarJump);
void jmp(Register reg) { EmitUnaryL(reg, 0xFF, 4); }
void jmp(const Address& address) { EmitUnaryL(address, 0xFF, 4); }
void jmp(Label* label, bool near = kFarJump);
void jmp(const ExternalLabel* label);
void jmp(const StubEntry& stub_entry);
// Issue memory to memory move through a TMP register.
// TODO(koda): Assert that these are not used for heap objects.
void MoveMemoryToMemory(const Address& dst, const Address& src) {
movq(TMP, src);
movq(dst, TMP);
}
void Exchange(Register reg, const Address& mem) {
movq(TMP, mem);
movq(mem, reg);
movq(reg, TMP);
}
void Exchange(const Address& mem1, const Address& mem2) {
movq(TMP, mem1);
xorq(TMP, mem2);
xorq(mem1, TMP);
xorq(mem2, TMP);
}
// Methods for High-level operations and implemented on all architectures.
void Ret() { ret(); }
void CompareRegisters(Register a, Register b);
void BranchIf(Condition condition, Label* label) { j(condition, label); }
// Issues a move instruction if 'to' is not the same as 'from'.
void MoveRegister(Register to, Register from);
void PushRegister(Register r);
void PopRegister(Register r);
// Methods for adding/subtracting an immediate value that may be loaded from
// the constant pool.
// TODO(koda): Assert that these are not used for heap objects.
void AddImmediate(Register reg,
const Immediate& imm,
OperandWidth width = k64Bit);
void AddImmediate(const Address& address, const Immediate& imm);
void SubImmediate(Register reg,
const Immediate& imm,
OperandWidth width = k64Bit);
void SubImmediate(const Address& address, const Immediate& imm);
void Drop(intptr_t stack_elements, Register tmp = TMP);
bool constant_pool_allowed() const { return constant_pool_allowed_; }
void set_constant_pool_allowed(bool b) { constant_pool_allowed_ = b; }
// Unlike movq this can affect the flags or use the constant pool.
void LoadImmediate(Register reg, const Immediate& imm);
void LoadIsolate(Register dst);
void LoadObject(Register dst, const Object& obj);
void LoadUniqueObject(Register dst, const Object& obj);
void LoadNativeEntry(Register dst,
const ExternalLabel* label,
ObjectPool::Patchability patchable);
void LoadFunctionFromCalleePool(Register dst,
const Function& function,
Register new_pp);
void JmpPatchable(const StubEntry& stub_entry, Register pp);
void Jmp(const StubEntry& stub_entry, Register pp = PP);
void J(Condition condition, const StubEntry& stub_entry, Register pp);
void CallPatchable(const StubEntry& stub_entry,
Code::EntryKind entry_kind = Code::EntryKind::kNormal);
void Call(const StubEntry& stub_entry);
void CallToRuntime();
void CallNullErrorShared(bool save_fpu_registers);
// Emit a call that shares its object pool entries with other calls
// that have the same equivalence marker.
void CallWithEquivalence(
const StubEntry& stub_entry,
const Object& equivalence,
Code::EntryKind entry_kind = Code::EntryKind::kNormal);
// Unaware of write barrier (use StoreInto* methods for storing to objects).
// TODO(koda): Add StackAddress/HeapAddress types to prevent misuse.
void StoreObject(const Address& dst, const Object& obj);
void PushObject(const Object& object);
void CompareObject(Register reg, const Object& object);
enum CanBeSmi {
kValueIsNotSmi,
kValueCanBeSmi,
};
// Store into a heap object and apply the generational and incremental write
// barriers. All stores into heap objects must pass through this function or,
// if the value can be proven either Smi or old-and-premarked, its NoBarrier
// variants.
// Preserves object and value registers.
void StoreIntoObject(Register object, // Object we are storing into.
const Address& dest, // Where we are storing into.
Register value, // Value we are storing.
CanBeSmi can_be_smi = kValueCanBeSmi);
void StoreIntoObjectNoBarrier(Register object,
const Address& dest,
Register value);
void StoreIntoObjectNoBarrier(Register object,
const Address& dest,
const Object& value);
// Stores a Smi value into a heap object field that always contains a Smi.
void StoreIntoSmiField(const Address& dest, Register value);
void ZeroInitSmiField(const Address& dest);
// Increments a Smi field. Leaves flags in same state as an 'addq'.
void IncrementSmiField(const Address& dest, int64_t increment);
void DoubleNegate(XmmRegister dst, XmmRegister src);
void DoubleAbs(XmmRegister dst, XmmRegister src);
void LockCmpxchgq(const Address& address, Register reg) {
lock();
cmpxchgq(address, reg);
}
void LockCmpxchgl(const Address& address, Register reg) {
lock();
cmpxchgl(address, reg);
}
void PushRegisters(intptr_t cpu_register_set, intptr_t xmm_register_set);
void PopRegisters(intptr_t cpu_register_set, intptr_t xmm_register_set);
void CheckCodePointer();
void EnterFrame(intptr_t frame_space);
void LeaveFrame();
void ReserveAlignedFrameSpace(intptr_t frame_space);
// Create a frame for calling into runtime that preserves all volatile
// registers. Frame's RSP is guaranteed to be correctly aligned and
// frame_space bytes are reserved under it.
void EnterCallRuntimeFrame(intptr_t frame_space);
void LeaveCallRuntimeFrame();
void CallRuntime(const RuntimeEntry& entry, intptr_t argument_count);
void CallRuntimeSavingRegisters(const RuntimeEntry& entry,
intptr_t argument_count);
// Call runtime function. Reserves shadow space on the stack before calling
// if platform ABI requires that. Does not restore RSP after the call itself.
void CallCFunction(Register reg);
// Loading and comparing classes of objects.
void LoadClassId(Register result, Register object);
// Overwrites class_id register (it will be tagged afterwards).
void LoadClassById(Register result, Register class_id);
void LoadClass(Register result, Register object);
void CompareClassId(Register object,
intptr_t class_id,
Register scratch = kNoRegister);
void LoadClassIdMayBeSmi(Register result, Register object);
void LoadTaggedClassIdMayBeSmi(Register result, Register object);
// CheckClassIs fused with optimistic SmiUntag.
// Value in the register object is untagged optimistically.
void SmiUntagOrCheckClass(Register object, intptr_t class_id, Label* smi);
// Misc. functionality.
void SmiTag(Register reg) { addq(reg, reg); }
void SmiUntag(Register reg) { sarq(reg, Immediate(kSmiTagSize)); }
void BranchIfNotSmi(Register reg, Label* label) {
testq(reg, Immediate(kSmiTagMask));
j(NOT_ZERO, label);
}
void BranchIfSmi(Register reg, Label* label) {
testq(reg, Immediate(kSmiTagMask));
j(ZERO, label);
}
void Align(int alignment, intptr_t offset);
void Bind(Label* label);
void Jump(Label* label) { jmp(label); }
void LoadField(Register dst, FieldAddress address) { movq(dst, address); }
void CompareWithFieldValue(Register value, FieldAddress address) {
cmpq(value, address);
}
void Comment(const char* format, ...) PRINTF_ATTRIBUTE(2, 3);
static bool EmittingComments();
const Code::Comments& GetCodeComments() const;
// Address of code at offset.
uword CodeAddress(intptr_t offset) { return buffer_.Address(offset); }
intptr_t CodeSize() const { return buffer_.Size(); }
intptr_t prologue_offset() const { return prologue_offset_; }
bool has_single_entry_point() const { return has_single_entry_point_; }
// Count the fixups that produce a pointer offset, without processing
// the fixups.
intptr_t CountPointerOffsets() const { return buffer_.CountPointerOffsets(); }
const ZoneGrowableArray<intptr_t>& GetPointerOffsets() const {
return buffer_.pointer_offsets();
}
ObjectPoolWrapper& object_pool_wrapper() { return *object_pool_wrapper_; }
RawObjectPool* MakeObjectPool() {
return object_pool_wrapper_->MakeObjectPool();
}
void FinalizeInstructions(const MemoryRegion& region) {
buffer_.FinalizeInstructions(region);
}
void RestoreCodePointer();
void LoadPoolPointer(Register pp = PP);
// Set up a Dart frame on entry with a frame pointer and PC information to
// enable easy access to the RawInstruction object of code corresponding
// to this frame.
// The dart frame layout is as follows:
// ....
// locals space <=== RSP
// saved PP
// pc (used to derive the RawInstruction Object of the dart code)
// saved RBP <=== RBP
// ret PC
// .....
// This code sets this up with the sequence:
// pushq rbp
// movq rbp, rsp
// call L
// L: <code to adjust saved pc if there is any intrinsification code>
// ...
// pushq r15
// .....
void EnterDartFrame(intptr_t frame_size, Register new_pp);
void LeaveDartFrame(RestorePP restore_pp = kRestoreCallerPP);
// Set up a Dart frame for a function compiled for on-stack replacement.
// The frame layout is a normal Dart frame, but the frame is partially set
// up on entry (it is the frame of the unoptimized code).
void EnterOsrFrame(intptr_t extra_size);
// Set up a stub frame so that the stack traversal code can easily identify
// a stub frame.
// The stub frame layout is as follows:
// .... <=== RSP
// pc (used to derive the RawInstruction Object of the stub)
// saved RBP <=== RBP
// ret PC
// .....
// This code sets this up with the sequence:
// pushq rbp
// movq rbp, rsp
// pushq immediate(0)
// .....
void EnterStubFrame();
void LeaveStubFrame();
void MonomorphicCheckedEntry();
void UpdateAllocationStats(intptr_t cid, Heap::Space space);
void UpdateAllocationStatsWithSize(intptr_t cid,
Register size_reg,
Heap::Space space);
void UpdateAllocationStatsWithSize(intptr_t cid,
intptr_t instance_size,
Heap::Space space);
// If allocation tracing for |cid| is enabled, will jump to |trace| label,
// which will allocate in the runtime where tracing occurs.
void MaybeTraceAllocation(intptr_t cid, Label* trace, bool near_jump);
// Inlined allocation of an instance of class 'cls', code has no runtime
// calls. Jump to 'failure' if the instance cannot be allocated here.
// Allocated instance is returned in 'instance_reg'.
// Only the tags field of the object is initialized.
void TryAllocate(const Class& cls,
Label* failure,
bool near_jump,
Register instance_reg,
Register temp);
void TryAllocateArray(intptr_t cid,
intptr_t instance_size,
Label* failure,
bool near_jump,
Register instance,
Register end_address,
Register temp);
// Debugging and bringup support.
void Breakpoint() { int3(); }
void Stop(const char* message, bool fixed_length_encoding = false);
void Unimplemented(const char* message);
void Untested(const char* message);
void Unreachable(const char* message);
static void InitializeMemoryWithBreakpoints(uword data, intptr_t length);
static const char* RegisterName(Register reg);
static const char* FpuRegisterName(FpuRegister reg);
static Address ElementAddressForIntIndex(bool is_external,
intptr_t cid,
intptr_t index_scale,
Register array,
intptr_t index);
static Address ElementAddressForRegIndex(bool is_external,
intptr_t cid,
intptr_t index_scale,
Register array,
Register index);
static Address VMTagAddress() {
return Address(THR, Thread::vm_tag_offset());
}
// On some other platforms, we draw a distinction between safe and unsafe
// smis.
static bool IsSafe(const Object& object) { return true; }
static bool IsSafeSmi(const Object& object) { return object.IsSmi(); }
private:
AssemblerBuffer buffer_;
ObjectPoolWrapper* object_pool_wrapper_;
intptr_t prologue_offset_;
bool has_single_entry_point_;
class CodeComment : public ZoneAllocated {
public:
CodeComment(intptr_t pc_offset, const String& comment)
: pc_offset_(pc_offset), comment_(comment) {}
intptr_t pc_offset() const { return pc_offset_; }
const String& comment() const { return comment_; }
private:
intptr_t pc_offset_;
const String& comment_;
DISALLOW_COPY_AND_ASSIGN(CodeComment);
};
GrowableArray<CodeComment*> comments_;
bool constant_pool_allowed_;
intptr_t FindImmediate(int64_t imm);
bool CanLoadFromObjectPool(const Object& object) const;
void LoadObjectHelper(Register dst, const Object& obj, bool is_unique);
void LoadWordFromPoolOffset(Register dst, int32_t offset);
void AluL(uint8_t modrm_opcode, Register dst, const Immediate& imm);
void AluB(uint8_t modrm_opcode, const Address& dst, const Immediate& imm);
void AluW(uint8_t modrm_opcode, const Address& dst, const Immediate& imm);
void AluL(uint8_t modrm_opcode, const Address& dst, const Immediate& imm);
void AluQ(uint8_t modrm_opcode,
uint8_t opcode,
Register dst,
const Immediate& imm);
void AluQ(uint8_t modrm_opcode,
uint8_t opcode,
const Address& dst,
const Immediate& imm);
void EmitSimple(int opcode, int opcode2 = -1);
void EmitUnaryQ(Register reg, int opcode, int modrm_code);
void EmitUnaryL(Register reg, int opcode, int modrm_code);
void EmitUnaryQ(const Address& address, int opcode, int modrm_code);
void EmitUnaryL(const Address& address, int opcode, int modrm_code);
// The prefixes are in reverse order due to the rules of default arguments in
// C++.
void EmitQ(int reg,
const Address& address,
int opcode,
int prefix2 = -1,
int prefix1 = -1);
void EmitL(int reg,
const Address& address,
int opcode,
int prefix2 = -1,
int prefix1 = -1);
void EmitW(Register reg,
const Address& address,
int opcode,
int prefix2 = -1,
int prefix1 = -1);
void EmitQ(int dst, int src, int opcode, int prefix2 = -1, int prefix1 = -1);
void EmitL(int dst, int src, int opcode, int prefix2 = -1, int prefix1 = -1);
void EmitW(Register dst,
Register src,
int opcode,
int prefix2 = -1,
int prefix1 = -1);
void CmpPS(XmmRegister dst, XmmRegister src, int condition);
inline void EmitUint8(uint8_t value);
inline void EmitInt32(int32_t value);
inline void EmitUInt32(uint32_t value);
inline void EmitInt64(int64_t value);
inline void EmitRegisterREX(Register reg,
uint8_t rex,
bool force_emit = false);
inline void EmitOperandREX(int rm, const Operand& operand, uint8_t rex);
inline void EmitRegisterOperand(int rm, int reg);
inline void EmitFixup(AssemblerFixup* fixup);
inline void EmitOperandSizeOverride();
inline void EmitRegRegRex(int reg, int base, uint8_t rex = REX_NONE);
void EmitOperand(int rm, const Operand& operand);
void EmitImmediate(const Immediate& imm);
void EmitComplex(int rm, const Operand& operand, const Immediate& immediate);
void EmitSignExtendedInt8(int rm,
const Operand& operand,
const Immediate& immediate);
void EmitLabel(Label* label, intptr_t instruction_size);
void EmitLabelLink(Label* label);
void EmitNearLabelLink(Label* label);
void EmitGenericShift(bool wide, int rm, Register reg, const Immediate& imm);
void EmitGenericShift(bool wide, int rm, Register operand, Register shifter);
enum BarrierFilterMode {
// Filter falls through into the barrier update code. Target label
// is a "after-store" label.
kJumpToNoUpdate,
// Filter falls through to the "after-store" code. Target label
// is barrier update code label.
kJumpToBarrier,
};
void StoreIntoObjectFilter(Register object,
Register value,
Label* label,
CanBeSmi can_be_smi,
BarrierFilterMode barrier_filter_mode);
// Unaware of write barrier (use StoreInto* methods for storing to objects).
void MoveImmediate(const Address& dst, const Immediate& imm);
void ComputeCounterAddressesForCid(intptr_t cid,
Heap::Space space,
Address* count_address,
Address* size_address);
DISALLOW_ALLOCATION();
DISALLOW_COPY_AND_ASSIGN(Assembler);
};
inline void Assembler::EmitUint8(uint8_t value) {
buffer_.Emit<uint8_t>(value);
}
inline void Assembler::EmitInt32(int32_t value) {
buffer_.Emit<int32_t>(value);
}
inline void Assembler::EmitUInt32(uint32_t value) {
buffer_.Emit<uint32_t>(value);
}
inline void Assembler::EmitInt64(int64_t value) {
buffer_.Emit<int64_t>(value);
}
inline void Assembler::EmitRegisterREX(Register reg, uint8_t rex, bool force) {
ASSERT(reg != kNoRegister && reg <= R15);
ASSERT(rex == REX_NONE || rex == REX_W);
rex |= (reg > 7 ? REX_B : REX_NONE);
if (rex != REX_NONE || force) EmitUint8(REX_PREFIX | rex);
}
inline void Assembler::EmitOperandREX(int rm,
const Operand& operand,
uint8_t rex) {
rex |= (rm > 7 ? REX_R : REX_NONE) | operand.rex();
if (rex != REX_NONE) EmitUint8(REX_PREFIX | rex);
}
inline void Assembler::EmitRegRegRex(int reg, int base, uint8_t rex) {
ASSERT(reg != kNoRegister && reg <= R15);
ASSERT(base != kNoRegister && base <= R15);
ASSERT(rex == REX_NONE || rex == REX_W);
if (reg > 7) rex |= REX_R;
if (base > 7) rex |= REX_B;
if (rex != REX_NONE) EmitUint8(REX_PREFIX | rex);
}
inline void Assembler::EmitFixup(AssemblerFixup* fixup) {
buffer_.EmitFixup(fixup);
}
inline void Assembler::EmitOperandSizeOverride() {
EmitUint8(0x66);
}
} // namespace dart
#endif // RUNTIME_VM_COMPILER_ASSEMBLER_ASSEMBLER_X64_H_
Reference in New Issue View Git Blame Copy Permalink