Mercurial > hg > graal-jvmci-8
view graal/com.oracle.jvmci.asm.amd64/src/com/oracle/jvmci/asm/amd64/AMD64Assembler.java @ 21711:1c41d00e6b2f
moved com.oracle.graal.bytecode to JVMCI namespace (JBS:GRAAL-53)
author | Doug Simon <doug.simon@oracle.com> |
---|---|
date | Wed, 03 Jun 2015 21:42:01 +0200 |
parents | 6df25b1418be |
children |
line wrap: on
line source
/* * Copyright (c) 2009, 2015, Oracle and/or its affiliates. All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. */ package com.oracle.jvmci.asm.amd64; import com.oracle.jvmci.amd64.*; import com.oracle.jvmci.amd64.AMD64.*; import com.oracle.jvmci.asm.*; import com.oracle.jvmci.code.Register; import com.oracle.jvmci.code.TargetDescription; import com.oracle.jvmci.code.RegisterConfig; import static com.oracle.jvmci.amd64.AMD64.*; import static com.oracle.jvmci.asm.NumUtil.*; import static com.oracle.jvmci.asm.amd64.AMD64AsmOptions.*; import static com.oracle.jvmci.asm.amd64.AMD64Assembler.AMD64BinaryArithmetic.*; import static com.oracle.jvmci.asm.amd64.AMD64Assembler.AMD64MOp.*; import static com.oracle.jvmci.asm.amd64.AMD64Assembler.OperandSize.*; import static com.oracle.jvmci.code.MemoryBarriers.*; import com.oracle.jvmci.code.Register.RegisterCategory; /** * This class implements an assembler that can encode most X86 instructions. */ public class AMD64Assembler extends Assembler { private static final int MinEncodingNeedsRex = 8; /** * A sentinel value used as a place holder in an instruction stream for an address that will be * patched. */ private static final AMD64Address Placeholder = new AMD64Address(rip); /** * The x86 condition codes used for conditional jumps/moves. */ public enum ConditionFlag { Zero(0x4, "|zero|"), NotZero(0x5, "|nzero|"), Equal(0x4, "="), NotEqual(0x5, "!="), Less(0xc, "<"), LessEqual(0xe, "<="), Greater(0xf, ">"), GreaterEqual(0xd, ">="), Below(0x2, "|<|"), BelowEqual(0x6, "|<=|"), Above(0x7, "|>|"), AboveEqual(0x3, "|>=|"), Overflow(0x0, "|of|"), NoOverflow(0x1, "|nof|"), CarrySet(0x2, "|carry|"), CarryClear(0x3, "|ncarry|"), Negative(0x8, "|neg|"), Positive(0x9, "|pos|"), Parity(0xa, "|par|"), NoParity(0xb, "|npar|"); private final int value; private final String operator; private ConditionFlag(int value, String operator) { this.value = value; this.operator = operator; } public ConditionFlag negate() { switch (this) { case Zero: return NotZero; case NotZero: return Zero; case Equal: return NotEqual; case NotEqual: return Equal; case Less: return GreaterEqual; case LessEqual: return Greater; case Greater: return LessEqual; case GreaterEqual: return Less; case Below: return AboveEqual; case BelowEqual: return Above; case Above: return BelowEqual; case AboveEqual: return Below; case Overflow: return NoOverflow; case NoOverflow: return Overflow; case CarrySet: return CarryClear; case CarryClear: return CarrySet; case Negative: return Positive; case Positive: return Negative; case Parity: return NoParity; case NoParity: return Parity; } throw new IllegalArgumentException(); } public int getValue() { return value; } @Override public String toString() { return operator; } } /** * Constants for X86 prefix bytes. */ private static class Prefix { private static final int REX = 0x40; private static final int REXB = 0x41; private static final int REXX = 0x42; private static final int REXXB = 0x43; private static final int REXR = 0x44; private static final int REXRB = 0x45; private static final int REXRX = 0x46; private static final int REXRXB = 0x47; private static final int REXW = 0x48; private static final int REXWB = 0x49; private static final int REXWX = 0x4A; private static final int REXWXB = 0x4B; private static final int REXWR = 0x4C; private static final int REXWRB = 0x4D; private static final int REXWRX = 0x4E; private static final int REXWRXB = 0x4F; } /** * The x86 operand sizes. */ public static enum OperandSize { BYTE(1) { @Override protected void emitImmediate(AMD64Assembler asm, int imm) { assert imm == (byte) imm; asm.emitByte(imm); } }, WORD(2, 0x66) { @Override protected void emitImmediate(AMD64Assembler asm, int imm) { assert imm == (short) imm; asm.emitShort(imm); } }, DWORD(4) { @Override protected void emitImmediate(AMD64Assembler asm, int imm) { asm.emitInt(imm); } }, QWORD(8) { @Override protected void emitImmediate(AMD64Assembler asm, int imm) { asm.emitInt(imm); } }, SS(4, 0xF3, true), SD(8, 0xF2, true), PS(16, true), PD(16, 0x66, true); private final int sizePrefix; private final int bytes; private final boolean xmm; private OperandSize(int bytes) { this(bytes, 0); } private OperandSize(int bytes, int sizePrefix) { this(bytes, sizePrefix, false); } private OperandSize(int bytes, boolean xmm) { this(bytes, 0, xmm); } private OperandSize(int bytes, int sizePrefix, boolean xmm) { this.sizePrefix = sizePrefix; this.bytes = bytes; this.xmm = xmm; } public int getBytes() { return bytes; } public boolean isXmmType() { return xmm; } /** * Emit an immediate of this size. Note that immediate {@link #QWORD} operands are encoded * as sign-extended 32-bit values. * * @param asm * @param imm */ protected void emitImmediate(AMD64Assembler asm, int imm) { assert false; } } /** * Operand size and register type constraints. */ private static enum OpAssertion { ByteAssertion(CPU, CPU, BYTE), IntegerAssertion(CPU, CPU, WORD, DWORD, QWORD), No16BitAssertion(CPU, CPU, DWORD, QWORD), QwordOnlyAssertion(CPU, CPU, QWORD), FloatingAssertion(XMM, XMM, SS, SD, PS, PD), PackedFloatingAssertion(XMM, XMM, PS, PD), SingleAssertion(XMM, XMM, SS), DoubleAssertion(XMM, XMM, SD), IntToFloatingAssertion(XMM, CPU, DWORD, QWORD), FloatingToIntAssertion(CPU, XMM, DWORD, QWORD); private final RegisterCategory resultCategory; private final RegisterCategory inputCategory; private final OperandSize[] allowedSizes; private OpAssertion(RegisterCategory resultCategory, RegisterCategory inputCategory, OperandSize... allowedSizes) { this.resultCategory = resultCategory; this.inputCategory = inputCategory; this.allowedSizes = allowedSizes; } protected boolean checkOperands(AMD64Op op, OperandSize size, Register resultReg, Register inputReg) { assert resultReg == null || resultCategory.equals(resultReg.getRegisterCategory()) : "invalid result register " + resultReg + " used in " + op; assert inputReg == null || inputCategory.equals(inputReg.getRegisterCategory()) : "invalid input register " + inputReg + " used in " + op; for (OperandSize s : allowedSizes) { if (size == s) { return true; } } assert false : "invalid operand size " + size + " used in " + op; return false; } } /** * The register to which {@link Register#Frame} and {@link Register#CallerFrame} are bound. */ public final Register frameRegister; /** * Constructs an assembler for the AMD64 architecture. * * @param registerConfig the register configuration used to bind {@link Register#Frame} and * {@link Register#CallerFrame} to physical registers. This value can be null if this * assembler instance will not be used to assemble instructions using these logical * registers. */ public AMD64Assembler(TargetDescription target, RegisterConfig registerConfig) { super(target); this.frameRegister = registerConfig == null ? null : registerConfig.getFrameRegister(); } private boolean supports(CPUFeature feature) { return ((AMD64) target.arch).getFeatures().contains(feature); } private static int encode(Register r) { assert r.encoding < 16 && r.encoding >= 0 : "encoding out of range: " + r.encoding; return r.encoding & 0x7; } /** * Get RXB bits for register-register instruction. In that encoding, ModRM.rm contains a * register index. The R bit extends the ModRM.reg field and the B bit extends the ModRM.rm * field. The X bit must be 0. */ protected static int getRXB(Register reg, Register rm) { int rxb = (reg == null ? 0 : reg.encoding & 0x08) >> 1; rxb |= (rm == null ? 0 : rm.encoding & 0x08) >> 3; return rxb; } /** * Get RXB bits for register-memory instruction. The R bit extends the ModRM.reg field. There * are two cases for the memory operand:<br> * ModRM.rm contains the base register: In that case, B extends the ModRM.rm field and X = 0.<br> * There is an SIB byte: In that case, X extends SIB.index and B extends SIB.base. */ protected static int getRXB(Register reg, AMD64Address rm) { int rxb = (reg == null ? 0 : reg.encoding & 0x08) >> 1; if (!rm.getIndex().equals(Register.None)) { rxb |= (rm.getIndex().encoding & 0x08) >> 2; } if (!rm.getBase().equals(Register.None)) { rxb |= (rm.getBase().encoding & 0x08) >> 3; } return rxb; } /** * Emit the ModR/M byte for one register operand and an opcode extension in the R field. * <p> * Format: [ 11 reg r/m ] */ protected void emitModRM(int reg, Register rm) { assert (reg & 0x07) == reg; emitByte(0xC0 | (reg << 3) | (rm.encoding & 0x07)); } /** * Emit the ModR/M byte for two register operands. * <p> * Format: [ 11 reg r/m ] */ protected void emitModRM(Register reg, Register rm) { emitModRM(reg.encoding & 0x07, rm); } /** * Emits the ModR/M byte and optionally the SIB byte for one register and one memory operand. */ protected void emitOperandHelper(Register reg, AMD64Address addr) { assert !reg.equals(Register.None); emitOperandHelper(encode(reg), addr); } /** * Emits the ModR/M byte and optionally the SIB byte for one memory operand and an opcode * extension in the R field. */ protected void emitOperandHelper(int reg, AMD64Address addr) { assert (reg & 0x07) == reg; int regenc = reg << 3; Register base = addr.getBase(); Register index = addr.getIndex(); AMD64Address.Scale scale = addr.getScale(); int disp = addr.getDisplacement(); if (base.equals(Register.Frame)) { assert frameRegister != null : "cannot use register " + Register.Frame + " in assembler with null register configuration"; base = frameRegister; } if (base.equals(AMD64.rip)) { // also matches Placeholder // [00 000 101] disp32 assert index.equals(Register.None) : "cannot use RIP relative addressing with index register"; emitByte(0x05 | regenc); emitInt(disp); } else if (base.isValid()) { int baseenc = base.isValid() ? encode(base) : 0; if (index.isValid()) { int indexenc = encode(index) << 3; // [base + indexscale + disp] if (disp == 0 && !base.equals(rbp) && !base.equals(r13)) { // [base + indexscale] // [00 reg 100][ss index base] assert !index.equals(rsp) : "illegal addressing mode"; emitByte(0x04 | regenc); emitByte(scale.log2 << 6 | indexenc | baseenc); } else if (isByte(disp)) { // [base + indexscale + imm8] // [01 reg 100][ss index base] imm8 assert !index.equals(rsp) : "illegal addressing mode"; emitByte(0x44 | regenc); emitByte(scale.log2 << 6 | indexenc | baseenc); emitByte(disp & 0xFF); } else { // [base + indexscale + disp32] // [10 reg 100][ss index base] disp32 assert !index.equals(rsp) : "illegal addressing mode"; emitByte(0x84 | regenc); emitByte(scale.log2 << 6 | indexenc | baseenc); emitInt(disp); } } else if (base.equals(rsp) || base.equals(r12)) { // [rsp + disp] if (disp == 0) { // [rsp] // [00 reg 100][00 100 100] emitByte(0x04 | regenc); emitByte(0x24); } else if (isByte(disp)) { // [rsp + imm8] // [01 reg 100][00 100 100] disp8 emitByte(0x44 | regenc); emitByte(0x24); emitByte(disp & 0xFF); } else { // [rsp + imm32] // [10 reg 100][00 100 100] disp32 emitByte(0x84 | regenc); emitByte(0x24); emitInt(disp); } } else { // [base + disp] assert !base.equals(rsp) && !base.equals(r12) : "illegal addressing mode"; if (disp == 0 && !base.equals(rbp) && !base.equals(r13)) { // [base] // [00 reg base] emitByte(0x00 | regenc | baseenc); } else if (isByte(disp)) { // [base + disp8] // [01 reg base] disp8 emitByte(0x40 | regenc | baseenc); emitByte(disp & 0xFF); } else { // [base + disp32] // [10 reg base] disp32 emitByte(0x80 | regenc | baseenc); emitInt(disp); } } } else { if (index.isValid()) { int indexenc = encode(index) << 3; // [indexscale + disp] // [00 reg 100][ss index 101] disp32 assert !index.equals(rsp) : "illegal addressing mode"; emitByte(0x04 | regenc); emitByte(scale.log2 << 6 | indexenc | 0x05); emitInt(disp); } else { // [disp] ABSOLUTE // [00 reg 100][00 100 101] disp32 emitByte(0x04 | regenc); emitByte(0x25); emitInt(disp); } } } /** * Base class for AMD64 opcodes. */ public static class AMD64Op { protected static final int P_0F = 0x0F; protected static final int P_0F38 = 0x380F; protected static final int P_0F3A = 0x3A0F; private final String opcode; private final int prefix1; private final int prefix2; private final int op; private final boolean dstIsByte; private final boolean srcIsByte; private final OpAssertion assertion; private final CPUFeature feature; protected AMD64Op(String opcode, int prefix1, int prefix2, int op, OpAssertion assertion, CPUFeature feature) { this(opcode, prefix1, prefix2, op, assertion == OpAssertion.ByteAssertion, assertion == OpAssertion.ByteAssertion, assertion, feature); } protected AMD64Op(String opcode, int prefix1, int prefix2, int op, boolean dstIsByte, boolean srcIsByte, OpAssertion assertion, CPUFeature feature) { this.opcode = opcode; this.prefix1 = prefix1; this.prefix2 = prefix2; this.op = op; this.dstIsByte = dstIsByte; this.srcIsByte = srcIsByte; this.assertion = assertion; this.feature = feature; } protected final void emitOpcode(AMD64Assembler asm, OperandSize size, int rxb, int dstEnc, int srcEnc) { if (prefix1 != 0) { asm.emitByte(prefix1); } if (size.sizePrefix != 0) { asm.emitByte(size.sizePrefix); } int rexPrefix = 0x40 | rxb; if (size == QWORD) { rexPrefix |= 0x08; } if (rexPrefix != 0x40 || (dstIsByte && dstEnc >= 4) || (srcIsByte && srcEnc >= 4)) { asm.emitByte(rexPrefix); } if (prefix2 > 0xFF) { asm.emitShort(prefix2); } else if (prefix2 > 0) { asm.emitByte(prefix2); } asm.emitByte(op); } protected final boolean verify(AMD64Assembler asm, OperandSize size, Register resultReg, Register inputReg) { assert feature == null || asm.supports(feature) : String.format("unsupported feature %s required for %s", feature, opcode); assert assertion.checkOperands(this, size, resultReg, inputReg); return true; } @Override public String toString() { return opcode; } } /** * Base class for AMD64 opcodes with immediate operands. */ public static class AMD64ImmOp extends AMD64Op { private final boolean immIsByte; protected AMD64ImmOp(String opcode, boolean immIsByte, int prefix, int op, OpAssertion assertion) { super(opcode, 0, prefix, op, assertion, null); this.immIsByte = immIsByte; } protected final void emitImmediate(AMD64Assembler asm, OperandSize size, int imm) { if (immIsByte) { assert imm == (byte) imm; asm.emitByte(imm); } else { size.emitImmediate(asm, imm); } } } /** * Opcode with operand order of either RM or MR. */ public abstract static class AMD64RROp extends AMD64Op { protected AMD64RROp(String opcode, int prefix1, int prefix2, int op, OpAssertion assertion, CPUFeature feature) { super(opcode, prefix1, prefix2, op, assertion, feature); } protected AMD64RROp(String opcode, int prefix1, int prefix2, int op, boolean dstIsByte, boolean srcIsByte, OpAssertion assertion, CPUFeature feature) { super(opcode, prefix1, prefix2, op, dstIsByte, srcIsByte, assertion, feature); } public abstract void emit(AMD64Assembler asm, OperandSize size, Register dst, Register src); } /** * Opcode with operand order of RM. */ public static class AMD64RMOp extends AMD64RROp { // @formatter:off public static final AMD64RMOp IMUL = new AMD64RMOp("IMUL", P_0F, 0xAF); public static final AMD64RMOp BSF = new AMD64RMOp("BSF", P_0F, 0xBC); public static final AMD64RMOp BSR = new AMD64RMOp("BSR", P_0F, 0xBD); public static final AMD64RMOp POPCNT = new AMD64RMOp("POPCNT", 0xF3, P_0F, 0xB8, CPUFeature.POPCNT); public static final AMD64RMOp TZCNT = new AMD64RMOp("TZCNT", 0xF3, P_0F, 0xBC, CPUFeature.BMI1); public static final AMD64RMOp LZCNT = new AMD64RMOp("LZCNT", 0xF3, P_0F, 0xBD, CPUFeature.LZCNT); public static final AMD64RMOp MOVZXB = new AMD64RMOp("MOVZXB", P_0F, 0xB6, false, true, OpAssertion.IntegerAssertion); public static final AMD64RMOp MOVZX = new AMD64RMOp("MOVZX", P_0F, 0xB7, OpAssertion.No16BitAssertion); public static final AMD64RMOp MOVSXB = new AMD64RMOp("MOVSXB", P_0F, 0xBE, false, true, OpAssertion.IntegerAssertion); public static final AMD64RMOp MOVSX = new AMD64RMOp("MOVSX", P_0F, 0xBF, OpAssertion.No16BitAssertion); public static final AMD64RMOp MOVSXD = new AMD64RMOp("MOVSXD", 0x63, OpAssertion.QwordOnlyAssertion); public static final AMD64RMOp MOVB = new AMD64RMOp("MOVB", 0x8A, OpAssertion.ByteAssertion); public static final AMD64RMOp MOV = new AMD64RMOp("MOV", 0x8B); // MOVD/MOVQ and MOVSS/MOVSD are the same opcode, just with different operand size prefix public static final AMD64RMOp MOVD = new AMD64RMOp("MOVD", 0x66, P_0F, 0x6E, OpAssertion.IntToFloatingAssertion, CPUFeature.SSE2); public static final AMD64RMOp MOVQ = new AMD64RMOp("MOVQ", 0x66, P_0F, 0x6E, OpAssertion.IntToFloatingAssertion, CPUFeature.SSE2); public static final AMD64RMOp MOVSS = new AMD64RMOp("MOVSS", P_0F, 0x10, OpAssertion.FloatingAssertion, CPUFeature.SSE); public static final AMD64RMOp MOVSD = new AMD64RMOp("MOVSD", P_0F, 0x10, OpAssertion.FloatingAssertion, CPUFeature.SSE); // TEST is documented as MR operation, but it's symmetric, and using it as RM operation is more convenient. public static final AMD64RMOp TESTB = new AMD64RMOp("TEST", 0x84, OpAssertion.ByteAssertion); public static final AMD64RMOp TEST = new AMD64RMOp("TEST", 0x85); // @formatter:on protected AMD64RMOp(String opcode, int op) { this(opcode, 0, op); } protected AMD64RMOp(String opcode, int op, OpAssertion assertion) { this(opcode, 0, op, assertion); } protected AMD64RMOp(String opcode, int prefix, int op) { this(opcode, 0, prefix, op, null); } protected AMD64RMOp(String opcode, int prefix, int op, OpAssertion assertion) { this(opcode, 0, prefix, op, assertion, null); } protected AMD64RMOp(String opcode, int prefix, int op, OpAssertion assertion, CPUFeature feature) { this(opcode, 0, prefix, op, assertion, feature); } protected AMD64RMOp(String opcode, int prefix, int op, boolean dstIsByte, boolean srcIsByte, OpAssertion assertion) { super(opcode, 0, prefix, op, dstIsByte, srcIsByte, assertion, null); } protected AMD64RMOp(String opcode, int prefix1, int prefix2, int op, CPUFeature feature) { this(opcode, prefix1, prefix2, op, OpAssertion.IntegerAssertion, feature); } protected AMD64RMOp(String opcode, int prefix1, int prefix2, int op, OpAssertion assertion, CPUFeature feature) { super(opcode, prefix1, prefix2, op, assertion, feature); } @Override public final void emit(AMD64Assembler asm, OperandSize size, Register dst, Register src) { assert verify(asm, size, dst, src); emitOpcode(asm, size, getRXB(dst, src), dst.encoding, src.encoding); asm.emitModRM(dst, src); } public final void emit(AMD64Assembler asm, OperandSize size, Register dst, AMD64Address src) { assert verify(asm, size, dst, null); emitOpcode(asm, size, getRXB(dst, src), dst.encoding, 0); asm.emitOperandHelper(dst, src); } } /** * Opcode with operand order of MR. */ public static class AMD64MROp extends AMD64RROp { // @formatter:off public static final AMD64MROp MOVB = new AMD64MROp("MOVB", 0x88, OpAssertion.ByteAssertion); public static final AMD64MROp MOV = new AMD64MROp("MOV", 0x89); // MOVD and MOVQ are the same opcode, just with different operand size prefix // Note that as MR opcodes, they have reverse operand order, so the IntToFloatingAssertion must be used. public static final AMD64MROp MOVD = new AMD64MROp("MOVD", 0x66, P_0F, 0x7E, OpAssertion.IntToFloatingAssertion, CPUFeature.SSE2); public static final AMD64MROp MOVQ = new AMD64MROp("MOVQ", 0x66, P_0F, 0x7E, OpAssertion.IntToFloatingAssertion, CPUFeature.SSE2); // MOVSS and MOVSD are the same opcode, just with different operand size prefix public static final AMD64MROp MOVSS = new AMD64MROp("MOVSS", P_0F, 0x11, OpAssertion.FloatingAssertion, CPUFeature.SSE); public static final AMD64MROp MOVSD = new AMD64MROp("MOVSD", P_0F, 0x11, OpAssertion.FloatingAssertion, CPUFeature.SSE); // @formatter:on protected AMD64MROp(String opcode, int op) { this(opcode, 0, op); } protected AMD64MROp(String opcode, int op, OpAssertion assertion) { this(opcode, 0, op, assertion); } protected AMD64MROp(String opcode, int prefix, int op) { this(opcode, prefix, op, OpAssertion.IntegerAssertion); } protected AMD64MROp(String opcode, int prefix, int op, OpAssertion assertion) { this(opcode, prefix, op, assertion, null); } protected AMD64MROp(String opcode, int prefix, int op, OpAssertion assertion, CPUFeature feature) { this(opcode, 0, prefix, op, assertion, feature); } protected AMD64MROp(String opcode, int prefix1, int prefix2, int op, OpAssertion assertion, CPUFeature feature) { super(opcode, prefix1, prefix2, op, assertion, feature); } @Override public final void emit(AMD64Assembler asm, OperandSize size, Register dst, Register src) { assert verify(asm, size, src, dst); emitOpcode(asm, size, getRXB(src, dst), src.encoding, dst.encoding); asm.emitModRM(src, dst); } public final void emit(AMD64Assembler asm, OperandSize size, AMD64Address dst, Register src) { assert verify(asm, size, null, src); emitOpcode(asm, size, getRXB(src, dst), src.encoding, 0); asm.emitOperandHelper(src, dst); } } /** * Opcodes with operand order of M. */ public static class AMD64MOp extends AMD64Op { // @formatter:off public static final AMD64MOp NOT = new AMD64MOp("NOT", 0xF7, 2); public static final AMD64MOp NEG = new AMD64MOp("NEG", 0xF7, 3); public static final AMD64MOp MUL = new AMD64MOp("MUL", 0xF7, 4); public static final AMD64MOp IMUL = new AMD64MOp("IMUL", 0xF7, 5); public static final AMD64MOp DIV = new AMD64MOp("DIV", 0xF7, 6); public static final AMD64MOp IDIV = new AMD64MOp("IDIV", 0xF7, 7); public static final AMD64MOp INC = new AMD64MOp("INC", 0xFF, 0); public static final AMD64MOp DEC = new AMD64MOp("DEC", 0xFF, 1); // @formatter:on private final int ext; protected AMD64MOp(String opcode, int op, int ext) { this(opcode, 0, op, ext); } protected AMD64MOp(String opcode, int prefix, int op, int ext) { this(opcode, prefix, op, ext, OpAssertion.IntegerAssertion); } protected AMD64MOp(String opcode, int prefix, int op, int ext, OpAssertion assertion) { super(opcode, 0, prefix, op, assertion, null); this.ext = ext; } public final void emit(AMD64Assembler asm, OperandSize size, Register dst) { assert verify(asm, size, dst, null); emitOpcode(asm, size, getRXB(null, dst), 0, dst.encoding); asm.emitModRM(ext, dst); } public final void emit(AMD64Assembler asm, OperandSize size, AMD64Address dst) { assert verify(asm, size, null, null); emitOpcode(asm, size, getRXB(null, dst), 0, 0); asm.emitOperandHelper(ext, dst); } } /** * Opcodes with operand order of MI. */ public static class AMD64MIOp extends AMD64ImmOp { // @formatter:off public static final AMD64MIOp MOVB = new AMD64MIOp("MOVB", true, 0xC6, 0, OpAssertion.ByteAssertion); public static final AMD64MIOp MOV = new AMD64MIOp("MOV", false, 0xC7, 0); public static final AMD64MIOp TEST = new AMD64MIOp("TEST", false, 0xF7, 0); // @formatter:on private final int ext; protected AMD64MIOp(String opcode, boolean immIsByte, int op, int ext) { this(opcode, immIsByte, op, ext, OpAssertion.IntegerAssertion); } protected AMD64MIOp(String opcode, boolean immIsByte, int op, int ext, OpAssertion assertion) { this(opcode, immIsByte, 0, op, ext, assertion); } protected AMD64MIOp(String opcode, boolean immIsByte, int prefix, int op, int ext, OpAssertion assertion) { super(opcode, immIsByte, prefix, op, assertion); this.ext = ext; } public final void emit(AMD64Assembler asm, OperandSize size, Register dst, int imm) { assert verify(asm, size, dst, null); emitOpcode(asm, size, getRXB(null, dst), 0, dst.encoding); asm.emitModRM(ext, dst); emitImmediate(asm, size, imm); } public final void emit(AMD64Assembler asm, OperandSize size, AMD64Address dst, int imm) { assert verify(asm, size, null, null); emitOpcode(asm, size, getRXB(null, dst), 0, 0); asm.emitOperandHelper(ext, dst); emitImmediate(asm, size, imm); } } /** * Opcodes with operand order of RMI. */ public static class AMD64RMIOp extends AMD64ImmOp { // @formatter:off public static final AMD64RMIOp IMUL = new AMD64RMIOp("IMUL", false, 0x69); public static final AMD64RMIOp IMUL_SX = new AMD64RMIOp("IMUL", true, 0x6B); // @formatter:on protected AMD64RMIOp(String opcode, boolean immIsByte, int op) { this(opcode, immIsByte, 0, op, OpAssertion.IntegerAssertion); } protected AMD64RMIOp(String opcode, boolean immIsByte, int prefix, int op, OpAssertion assertion) { super(opcode, immIsByte, prefix, op, assertion); } public final void emit(AMD64Assembler asm, OperandSize size, Register dst, Register src, int imm) { assert verify(asm, size, dst, src); emitOpcode(asm, size, getRXB(dst, src), dst.encoding, src.encoding); asm.emitModRM(dst, src); emitImmediate(asm, size, imm); } public final void emit(AMD64Assembler asm, OperandSize size, Register dst, AMD64Address src, int imm) { assert verify(asm, size, dst, null); emitOpcode(asm, size, getRXB(dst, src), dst.encoding, 0); asm.emitOperandHelper(dst, src); emitImmediate(asm, size, imm); } } public static class SSEOp extends AMD64RMOp { // @formatter:off public static final SSEOp CVTSI2SS = new SSEOp("CVTSI2SS", 0xF3, P_0F, 0x2A, OpAssertion.IntToFloatingAssertion); public static final SSEOp CVTSI2SD = new SSEOp("CVTSI2SS", 0xF2, P_0F, 0x2A, OpAssertion.IntToFloatingAssertion); public static final SSEOp CVTTSS2SI = new SSEOp("CVTTSS2SI", 0xF3, P_0F, 0x2C, OpAssertion.FloatingToIntAssertion); public static final SSEOp CVTTSD2SI = new SSEOp("CVTTSD2SI", 0xF2, P_0F, 0x2C, OpAssertion.FloatingToIntAssertion); public static final SSEOp UCOMIS = new SSEOp("UCOMIS", P_0F, 0x2E, OpAssertion.PackedFloatingAssertion); public static final SSEOp SQRT = new SSEOp("SQRT", P_0F, 0x51); public static final SSEOp AND = new SSEOp("AND", P_0F, 0x54, OpAssertion.PackedFloatingAssertion); public static final SSEOp ANDN = new SSEOp("ANDN", P_0F, 0x55, OpAssertion.PackedFloatingAssertion); public static final SSEOp OR = new SSEOp("OR", P_0F, 0x56, OpAssertion.PackedFloatingAssertion); public static final SSEOp XOR = new SSEOp("XOR", P_0F, 0x57, OpAssertion.PackedFloatingAssertion); public static final SSEOp ADD = new SSEOp("ADD", P_0F, 0x58); public static final SSEOp MUL = new SSEOp("MUL", P_0F, 0x59); public static final SSEOp CVTSS2SD = new SSEOp("CVTSS2SD", P_0F, 0x5A, OpAssertion.SingleAssertion); public static final SSEOp CVTSD2SS = new SSEOp("CVTSD2SS", P_0F, 0x5A, OpAssertion.DoubleAssertion); public static final SSEOp SUB = new SSEOp("SUB", P_0F, 0x5C); public static final SSEOp MIN = new SSEOp("MIN", P_0F, 0x5D); public static final SSEOp DIV = new SSEOp("DIV", P_0F, 0x5E); public static final SSEOp MAX = new SSEOp("MAX", P_0F, 0x5F); // @formatter:on protected SSEOp(String opcode, int prefix, int op) { this(opcode, prefix, op, OpAssertion.FloatingAssertion); } protected SSEOp(String opcode, int prefix, int op, OpAssertion assertion) { this(opcode, 0, prefix, op, assertion); } protected SSEOp(String opcode, int mandatoryPrefix, int prefix, int op, OpAssertion assertion) { super(opcode, mandatoryPrefix, prefix, op, assertion, CPUFeature.SSE2); } } /** * Arithmetic operation with operand order of RM, MR or MI. */ public static final class AMD64BinaryArithmetic { // @formatter:off public static final AMD64BinaryArithmetic ADD = new AMD64BinaryArithmetic("ADD", 0); public static final AMD64BinaryArithmetic OR = new AMD64BinaryArithmetic("OR", 1); public static final AMD64BinaryArithmetic ADC = new AMD64BinaryArithmetic("ADC", 2); public static final AMD64BinaryArithmetic SBB = new AMD64BinaryArithmetic("SBB", 3); public static final AMD64BinaryArithmetic AND = new AMD64BinaryArithmetic("AND", 4); public static final AMD64BinaryArithmetic SUB = new AMD64BinaryArithmetic("SUB", 5); public static final AMD64BinaryArithmetic XOR = new AMD64BinaryArithmetic("XOR", 6); public static final AMD64BinaryArithmetic CMP = new AMD64BinaryArithmetic("CMP", 7); // @formatter:on private final AMD64MIOp byteImmOp; private final AMD64MROp byteMrOp; private final AMD64RMOp byteRmOp; private final AMD64MIOp immOp; private final AMD64MIOp immSxOp; private final AMD64MROp mrOp; private final AMD64RMOp rmOp; private AMD64BinaryArithmetic(String opcode, int code) { int baseOp = code << 3; byteImmOp = new AMD64MIOp(opcode, true, 0, 0x80, code, OpAssertion.ByteAssertion); byteMrOp = new AMD64MROp(opcode, 0, baseOp, OpAssertion.ByteAssertion); byteRmOp = new AMD64RMOp(opcode, 0, baseOp | 0x02, OpAssertion.ByteAssertion); immOp = new AMD64MIOp(opcode, false, 0, 0x81, code, OpAssertion.IntegerAssertion); immSxOp = new AMD64MIOp(opcode, true, 0, 0x83, code, OpAssertion.IntegerAssertion); mrOp = new AMD64MROp(opcode, 0, baseOp | 0x01, OpAssertion.IntegerAssertion); rmOp = new AMD64RMOp(opcode, 0, baseOp | 0x03, OpAssertion.IntegerAssertion); } public AMD64MIOp getMIOpcode(OperandSize size, boolean sx) { if (size == BYTE) { return byteImmOp; } else if (sx) { return immSxOp; } else { return immOp; } } public AMD64MROp getMROpcode(OperandSize size) { if (size == BYTE) { return byteMrOp; } else { return mrOp; } } public AMD64RMOp getRMOpcode(OperandSize size) { if (size == BYTE) { return byteRmOp; } else { return rmOp; } } } /** * Shift operation with operand order of M1, MC or MI. */ public static final class AMD64Shift { // @formatter:off public static final AMD64Shift ROL = new AMD64Shift("ROL", 0); public static final AMD64Shift ROR = new AMD64Shift("ROR", 1); public static final AMD64Shift RCL = new AMD64Shift("RCL", 2); public static final AMD64Shift RCR = new AMD64Shift("RCR", 3); public static final AMD64Shift SHL = new AMD64Shift("SHL", 4); public static final AMD64Shift SHR = new AMD64Shift("SHR", 5); public static final AMD64Shift SAR = new AMD64Shift("SAR", 7); // @formatter:on public final AMD64MOp m1Op; public final AMD64MOp mcOp; public final AMD64MIOp miOp; private AMD64Shift(String opcode, int code) { m1Op = new AMD64MOp(opcode, 0, 0xD1, code, OpAssertion.IntegerAssertion); mcOp = new AMD64MOp(opcode, 0, 0xD3, code, OpAssertion.IntegerAssertion); miOp = new AMD64MIOp(opcode, true, 0, 0xC1, code, OpAssertion.IntegerAssertion); } } public final void addl(AMD64Address dst, int imm32) { ADD.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } public final void addl(Register dst, int imm32) { ADD.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } private void addrNop4() { // 4 bytes: NOP DWORD PTR [EAX+0] emitByte(0x0F); emitByte(0x1F); emitByte(0x40); // emitRm(cbuf, 0x1, EAXEnc, EAXEnc); emitByte(0); // 8-bits offset (1 byte) } private void addrNop5() { // 5 bytes: NOP DWORD PTR [EAX+EAX*0+0] 8-bits offset emitByte(0x0F); emitByte(0x1F); emitByte(0x44); // emitRm(cbuf, 0x1, EAXEnc, 0x4); emitByte(0x00); // emitRm(cbuf, 0x0, EAXEnc, EAXEnc); emitByte(0); // 8-bits offset (1 byte) } private void addrNop7() { // 7 bytes: NOP DWORD PTR [EAX+0] 32-bits offset emitByte(0x0F); emitByte(0x1F); emitByte(0x80); // emitRm(cbuf, 0x2, EAXEnc, EAXEnc); emitInt(0); // 32-bits offset (4 bytes) } private void addrNop8() { // 8 bytes: NOP DWORD PTR [EAX+EAX*0+0] 32-bits offset emitByte(0x0F); emitByte(0x1F); emitByte(0x84); // emitRm(cbuf, 0x2, EAXEnc, 0x4); emitByte(0x00); // emitRm(cbuf, 0x0, EAXEnc, EAXEnc); emitInt(0); // 32-bits offset (4 bytes) } public final void andl(Register dst, int imm32) { AND.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } public final void bswapl(Register reg) { int encode = prefixAndEncode(reg.encoding); emitByte(0x0F); emitByte(0xC8 | encode); } public final void cdql() { emitByte(0x99); } public final void cmovl(ConditionFlag cc, Register dst, Register src) { int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x40 | cc.getValue()); emitByte(0xC0 | encode); } public final void cmovl(ConditionFlag cc, Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x0F); emitByte(0x40 | cc.getValue()); emitOperandHelper(dst, src); } public final void cmpl(Register dst, int imm32) { CMP.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } public final void cmpl(Register dst, Register src) { CMP.rmOp.emit(this, DWORD, dst, src); } public final void cmpl(Register dst, AMD64Address src) { CMP.rmOp.emit(this, DWORD, dst, src); } public final void cmpl(AMD64Address dst, int imm32) { CMP.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } // The 32-bit cmpxchg compares the value at adr with the contents of X86.rax, // and stores reg into adr if so; otherwise, the value at adr is loaded into X86.rax,. // The ZF is set if the compared values were equal, and cleared otherwise. public final void cmpxchgl(Register reg, AMD64Address adr) { // cmpxchg prefix(adr, reg); emitByte(0x0F); emitByte(0xB1); emitOperandHelper(reg, adr); } protected final void decl(AMD64Address dst) { prefix(dst); emitByte(0xFF); emitOperandHelper(1, dst); } public final void hlt() { emitByte(0xF4); } public final void imull(Register dst, Register src, int value) { if (isByte(value)) { AMD64RMIOp.IMUL_SX.emit(this, DWORD, dst, src, value); } else { AMD64RMIOp.IMUL.emit(this, DWORD, dst, src, value); } } protected final void incl(AMD64Address dst) { prefix(dst); emitByte(0xFF); emitOperandHelper(0, dst); } public void jcc(ConditionFlag cc, int jumpTarget, boolean forceDisp32) { int shortSize = 2; int longSize = 6; long disp = jumpTarget - position(); if (!forceDisp32 && isByte(disp - shortSize)) { // 0111 tttn #8-bit disp emitByte(0x70 | cc.getValue()); emitByte((int) ((disp - shortSize) & 0xFF)); } else { // 0000 1111 1000 tttn #32-bit disp assert isInt(disp - longSize) : "must be 32bit offset (call4)"; emitByte(0x0F); emitByte(0x80 | cc.getValue()); emitInt((int) (disp - longSize)); } } public final void jcc(ConditionFlag cc, Label l) { assert (0 <= cc.getValue()) && (cc.getValue() < 16) : "illegal cc"; if (l.isBound()) { jcc(cc, l.position(), false); } else { // Note: could eliminate cond. jumps to this jump if condition // is the same however, seems to be rather unlikely case. // Note: use jccb() if label to be bound is very close to get // an 8-bit displacement l.addPatchAt(position()); emitByte(0x0F); emitByte(0x80 | cc.getValue()); emitInt(0); } } public final void jccb(ConditionFlag cc, Label l) { if (l.isBound()) { int shortSize = 2; int entry = l.position(); assert isByte(entry - (position() + shortSize)) : "Dispacement too large for a short jmp"; long disp = entry - position(); // 0111 tttn #8-bit disp emitByte(0x70 | cc.getValue()); emitByte((int) ((disp - shortSize) & 0xFF)); } else { l.addPatchAt(position()); emitByte(0x70 | cc.getValue()); emitByte(0); } } public final void jmp(int jumpTarget, boolean forceDisp32) { int shortSize = 2; int longSize = 5; long disp = jumpTarget - position(); if (!forceDisp32 && isByte(disp - shortSize)) { emitByte(0xEB); emitByte((int) ((disp - shortSize) & 0xFF)); } else { emitByte(0xE9); emitInt((int) (disp - longSize)); } } @Override public final void jmp(Label l) { if (l.isBound()) { jmp(l.position(), false); } else { // By default, forward jumps are always 32-bit displacements, since // we can't yet know where the label will be bound. If you're sure that // the forward jump will not run beyond 256 bytes, use jmpb to // force an 8-bit displacement. l.addPatchAt(position()); emitByte(0xE9); emitInt(0); } } public final void jmp(Register entry) { int encode = prefixAndEncode(entry.encoding); emitByte(0xFF); emitByte(0xE0 | encode); } public final void jmpb(Label l) { if (l.isBound()) { int shortSize = 2; int entry = l.position(); assert isByte((entry - position()) + shortSize) : "Dispacement too large for a short jmp"; long offs = entry - position(); emitByte(0xEB); emitByte((int) ((offs - shortSize) & 0xFF)); } else { l.addPatchAt(position()); emitByte(0xEB); emitByte(0); } } public final void leaq(Register dst, AMD64Address src) { prefixq(src, dst); emitByte(0x8D); emitOperandHelper(dst, src); } public final void leave() { emitByte(0xC9); } public final void lock() { emitByte(0xF0); } public final void movapd(Register dst, Register src) { assert dst.getRegisterCategory().equals(AMD64.XMM); assert src.getRegisterCategory().equals(AMD64.XMM); int dstenc = dst.encoding; int srcenc = src.encoding; emitByte(0x66); if (dstenc < 8) { if (srcenc >= 8) { emitByte(Prefix.REXB); srcenc -= 8; } } else { if (srcenc < 8) { emitByte(Prefix.REXR); } else { emitByte(Prefix.REXRB); srcenc -= 8; } dstenc -= 8; } emitByte(0x0F); emitByte(0x28); emitByte(0xC0 | dstenc << 3 | srcenc); } public final void movaps(Register dst, Register src) { assert dst.getRegisterCategory().equals(AMD64.XMM); assert src.getRegisterCategory().equals(AMD64.XMM); int dstenc = dst.encoding; int srcenc = src.encoding; if (dstenc < 8) { if (srcenc >= 8) { emitByte(Prefix.REXB); srcenc -= 8; } } else { if (srcenc < 8) { emitByte(Prefix.REXR); } else { emitByte(Prefix.REXRB); srcenc -= 8; } dstenc -= 8; } emitByte(0x0F); emitByte(0x28); emitByte(0xC0 | dstenc << 3 | srcenc); } public final void movb(AMD64Address dst, int imm8) { prefix(dst); emitByte(0xC6); emitOperandHelper(0, dst); emitByte(imm8); } public final void movb(AMD64Address dst, Register src) { assert src.getRegisterCategory().equals(AMD64.CPU) : "must have byte register"; prefix(dst, src, true); emitByte(0x88); emitOperandHelper(src, dst); } public final void movl(Register dst, int imm32) { int encode = prefixAndEncode(dst.encoding); emitByte(0xB8 | encode); emitInt(imm32); } public final void movl(Register dst, Register src) { int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x8B); emitByte(0xC0 | encode); } public final void movl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x8B); emitOperandHelper(dst, src); } public final void movl(AMD64Address dst, int imm32) { prefix(dst); emitByte(0xC7); emitOperandHelper(0, dst); emitInt(imm32); } public final void movl(AMD64Address dst, Register src) { prefix(dst, src); emitByte(0x89); emitOperandHelper(src, dst); } /** * New CPUs require use of movsd and movss to avoid partial register stall when loading from * memory. But for old Opteron use movlpd instead of movsd. The selection is done in * {@link AMD64MacroAssembler#movdbl(Register, AMD64Address)} and * {@link AMD64MacroAssembler#movflt(Register, Register)}. */ public final void movlpd(Register dst, AMD64Address src) { assert dst.getRegisterCategory().equals(AMD64.XMM); emitByte(0x66); prefix(src, dst); emitByte(0x0F); emitByte(0x12); emitOperandHelper(dst, src); } public final void movq(Register dst, AMD64Address src) { if (dst.getRegisterCategory().equals(AMD64.XMM)) { emitByte(0xF3); prefixq(src, dst); emitByte(0x0F); emitByte(0x7E); emitOperandHelper(dst, src); } else { prefixq(src, dst); emitByte(0x8B); emitOperandHelper(dst, src); } } public final void movq(Register dst, Register src) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x8B); emitByte(0xC0 | encode); } public final void movq(AMD64Address dst, Register src) { if (src.getRegisterCategory().equals(AMD64.XMM)) { emitByte(0x66); prefixq(dst, src); emitByte(0x0F); emitByte(0xD6); emitOperandHelper(src, dst); } else { prefixq(dst, src); emitByte(0x89); emitOperandHelper(src, dst); } } public final void movsbl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x0F); emitByte(0xBE); emitOperandHelper(dst, src); } public final void movsbl(Register dst, Register src) { int encode = prefixAndEncode(dst.encoding, false, src.encoding, true); emitByte(0x0F); emitByte(0xBE); emitByte(0xC0 | encode); } public final void movsbq(Register dst, AMD64Address src) { prefixq(src, dst); emitByte(0x0F); emitByte(0xBE); emitOperandHelper(dst, src); } public final void movsbq(Register dst, Register src) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0xBE); emitByte(0xC0 | encode); } public final void movsd(Register dst, Register src) { assert dst.getRegisterCategory().equals(AMD64.XMM); assert src.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF2); int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x10); emitByte(0xC0 | encode); } public final void movsd(Register dst, AMD64Address src) { assert dst.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF2); prefix(src, dst); emitByte(0x0F); emitByte(0x10); emitOperandHelper(dst, src); } public final void movsd(AMD64Address dst, Register src) { assert src.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF2); prefix(dst, src); emitByte(0x0F); emitByte(0x11); emitOperandHelper(src, dst); } public final void movss(Register dst, Register src) { assert dst.getRegisterCategory().equals(AMD64.XMM); assert src.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF3); int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x10); emitByte(0xC0 | encode); } public final void movss(Register dst, AMD64Address src) { assert dst.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF3); prefix(src, dst); emitByte(0x0F); emitByte(0x10); emitOperandHelper(dst, src); } public final void movss(AMD64Address dst, Register src) { assert src.getRegisterCategory().equals(AMD64.XMM); emitByte(0xF3); prefix(dst, src); emitByte(0x0F); emitByte(0x11); emitOperandHelper(src, dst); } public final void movswl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x0F); emitByte(0xBF); emitOperandHelper(dst, src); } public final void movw(AMD64Address dst, int imm16) { emitByte(0x66); // switch to 16-bit mode prefix(dst); emitByte(0xC7); emitOperandHelper(0, dst); emitShort(imm16); } public final void movw(AMD64Address dst, Register src) { emitByte(0x66); prefix(dst, src); emitByte(0x89); emitOperandHelper(src, dst); } public final void movzbl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x0F); emitByte(0xB6); emitOperandHelper(dst, src); } public final void movzwl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x0F); emitByte(0xB7); emitOperandHelper(dst, src); } @Override public final void ensureUniquePC() { nop(); } public final void nop() { nop(1); } public void nop(int count) { int i = count; if (UseNormalNop) { assert i > 0 : " "; // The fancy nops aren't currently recognized by debuggers making it a // pain to disassemble code while debugging. If assert are on clearly // speed is not an issue so simply use the single byte traditional nop // to do alignment. for (; i > 0; i--) { emitByte(0x90); } return; } if (UseAddressNop) { // // Using multi-bytes nops "0x0F 0x1F [Address]" for AMD. // 1: 0x90 // 2: 0x66 0x90 // 3: 0x66 0x66 0x90 (don't use "0x0F 0x1F 0x00" - need patching safe padding) // 4: 0x0F 0x1F 0x40 0x00 // 5: 0x0F 0x1F 0x44 0x00 0x00 // 6: 0x66 0x0F 0x1F 0x44 0x00 0x00 // 7: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 // 8: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 // 9: 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 // 10: 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 // 11: 0x66 0x66 0x66 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 // The rest coding is AMD specific - use consecutive Address nops // 12: 0x66 0x0F 0x1F 0x44 0x00 0x00 0x66 0x0F 0x1F 0x44 0x00 0x00 // 13: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 0x66 0x0F 0x1F 0x44 0x00 0x00 // 14: 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 // 15: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x0F 0x1F 0x80 0x00 0x00 0x00 0x00 // 16: 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 0x0F 0x1F 0x84 0x00 0x00 0x00 0x00 0x00 // Size prefixes (0x66) are added for larger sizes while (i >= 22) { i -= 11; emitByte(0x66); // size prefix emitByte(0x66); // size prefix emitByte(0x66); // size prefix addrNop8(); } // Generate first nop for size between 21-12 switch (i) { case 21: i -= 1; emitByte(0x66); // size prefix // fall through case 20: // fall through case 19: i -= 1; emitByte(0x66); // size prefix // fall through case 18: // fall through case 17: i -= 1; emitByte(0x66); // size prefix // fall through case 16: // fall through case 15: i -= 8; addrNop8(); break; case 14: case 13: i -= 7; addrNop7(); break; case 12: i -= 6; emitByte(0x66); // size prefix addrNop5(); break; default: assert i < 12; } // Generate second nop for size between 11-1 switch (i) { case 11: emitByte(0x66); // size prefix emitByte(0x66); // size prefix emitByte(0x66); // size prefix addrNop8(); break; case 10: emitByte(0x66); // size prefix emitByte(0x66); // size prefix addrNop8(); break; case 9: emitByte(0x66); // size prefix addrNop8(); break; case 8: addrNop8(); break; case 7: addrNop7(); break; case 6: emitByte(0x66); // size prefix addrNop5(); break; case 5: addrNop5(); break; case 4: addrNop4(); break; case 3: // Don't use "0x0F 0x1F 0x00" - need patching safe padding emitByte(0x66); // size prefix emitByte(0x66); // size prefix emitByte(0x90); // nop break; case 2: emitByte(0x66); // size prefix emitByte(0x90); // nop break; case 1: emitByte(0x90); // nop break; default: assert i == 0; } return; } // Using nops with size prefixes "0x66 0x90". // From AMD Optimization Guide: // 1: 0x90 // 2: 0x66 0x90 // 3: 0x66 0x66 0x90 // 4: 0x66 0x66 0x66 0x90 // 5: 0x66 0x66 0x90 0x66 0x90 // 6: 0x66 0x66 0x90 0x66 0x66 0x90 // 7: 0x66 0x66 0x66 0x90 0x66 0x66 0x90 // 8: 0x66 0x66 0x66 0x90 0x66 0x66 0x66 0x90 // 9: 0x66 0x66 0x90 0x66 0x66 0x90 0x66 0x66 0x90 // 10: 0x66 0x66 0x66 0x90 0x66 0x66 0x90 0x66 0x66 0x90 // while (i > 12) { i -= 4; emitByte(0x66); // size prefix emitByte(0x66); emitByte(0x66); emitByte(0x90); // nop } // 1 - 12 nops if (i > 8) { if (i > 9) { i -= 1; emitByte(0x66); } i -= 3; emitByte(0x66); emitByte(0x66); emitByte(0x90); } // 1 - 8 nops if (i > 4) { if (i > 6) { i -= 1; emitByte(0x66); } i -= 3; emitByte(0x66); emitByte(0x66); emitByte(0x90); } switch (i) { case 4: emitByte(0x66); emitByte(0x66); emitByte(0x66); emitByte(0x90); break; case 3: emitByte(0x66); emitByte(0x66); emitByte(0x90); break; case 2: emitByte(0x66); emitByte(0x90); break; case 1: emitByte(0x90); break; default: assert i == 0; } } public final void pop(Register dst) { int encode = prefixAndEncode(dst.encoding); emitByte(0x58 | encode); } public void popfq() { emitByte(0x9D); } public final void ptest(Register dst, Register src) { assert supports(CPUFeature.SSE4_1); emitByte(0x66); int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x38); emitByte(0x17); emitByte(0xC0 | encode); } public final void push(Register src) { int encode = prefixAndEncode(src.encoding); emitByte(0x50 | encode); } public void pushfq() { emitByte(0x9c); } public final void pxor(Register dst, Register src) { emitByte(0x66); int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0xEF); emitByte(0xC0 | encode); } public final void ret(int imm16) { if (imm16 == 0) { emitByte(0xC3); } else { emitByte(0xC2); emitShort(imm16); } } public final void subl(AMD64Address dst, int imm32) { SUB.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } public final void subl(Register dst, int imm32) { SUB.getMIOpcode(DWORD, isByte(imm32)).emit(this, DWORD, dst, imm32); } public final void testl(Register dst, int imm32) { // not using emitArith because test // doesn't support sign-extension of // 8bit operands int encode = dst.encoding; if (encode == 0) { emitByte(0xA9); } else { encode = prefixAndEncode(encode); emitByte(0xF7); emitByte(0xC0 | encode); } emitInt(imm32); } public final void testl(Register dst, Register src) { int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x85); emitByte(0xC0 | encode); } public final void testl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x85); emitOperandHelper(dst, src); } public final void xorl(Register dst, Register src) { XOR.rmOp.emit(this, DWORD, dst, src); } public final void xorpd(Register dst, Register src) { emitByte(0x66); xorps(dst, src); } public final void xorps(Register dst, Register src) { assert dst.getRegisterCategory().equals(AMD64.XMM) && src.getRegisterCategory().equals(AMD64.XMM); int encode = prefixAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x57); emitByte(0xC0 | encode); } protected final void decl(Register dst) { // Use two-byte form (one-byte form is a REX prefix in 64-bit mode) int encode = prefixAndEncode(dst.encoding); emitByte(0xFF); emitByte(0xC8 | encode); } protected final void incl(Register dst) { // Use two-byte form (one-byte from is a REX prefix in 64-bit mode) int encode = prefixAndEncode(dst.encoding); emitByte(0xFF); emitByte(0xC0 | encode); } private int prefixAndEncode(int regEnc) { return prefixAndEncode(regEnc, false); } private int prefixAndEncode(int regEnc, boolean byteinst) { if (regEnc >= 8) { emitByte(Prefix.REXB); return regEnc - 8; } else if (byteinst && regEnc >= 4) { emitByte(Prefix.REX); } return regEnc; } private int prefixqAndEncode(int regEnc) { if (regEnc < 8) { emitByte(Prefix.REXW); return regEnc; } else { emitByte(Prefix.REXWB); return regEnc - 8; } } private int prefixAndEncode(int dstEnc, int srcEnc) { return prefixAndEncode(dstEnc, false, srcEnc, false); } private int prefixAndEncode(int dstEncoding, boolean dstIsByte, int srcEncoding, boolean srcIsByte) { int srcEnc = srcEncoding; int dstEnc = dstEncoding; if (dstEnc < 8) { if (srcEnc >= 8) { emitByte(Prefix.REXB); srcEnc -= 8; } else if ((srcIsByte && srcEnc >= 4) || (dstIsByte && dstEnc >= 4)) { emitByte(Prefix.REX); } } else { if (srcEnc < 8) { emitByte(Prefix.REXR); } else { emitByte(Prefix.REXRB); srcEnc -= 8; } dstEnc -= 8; } return dstEnc << 3 | srcEnc; } /** * Creates prefix and the encoding of the lower 6 bits of the ModRM-Byte. It emits an operand * prefix. If the given operands exceed 3 bits, the 4th bit is encoded in the prefix. * * @param regEncoding the encoding of the register part of the ModRM-Byte * @param rmEncoding the encoding of the r/m part of the ModRM-Byte * @return the lower 6 bits of the ModRM-Byte that should be emitted */ private int prefixqAndEncode(int regEncoding, int rmEncoding) { int rmEnc = rmEncoding; int regEnc = regEncoding; if (regEnc < 8) { if (rmEnc < 8) { emitByte(Prefix.REXW); } else { emitByte(Prefix.REXWB); rmEnc -= 8; } } else { if (rmEnc < 8) { emitByte(Prefix.REXWR); } else { emitByte(Prefix.REXWRB); rmEnc -= 8; } regEnc -= 8; } return regEnc << 3 | rmEnc; } private static boolean needsRex(Register reg) { return reg.encoding >= MinEncodingNeedsRex; } private void prefix(AMD64Address adr) { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXXB); } else { emitByte(Prefix.REXB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXX); } } } private void prefixq(AMD64Address adr) { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWXB); } else { emitByte(Prefix.REXWB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWX); } else { emitByte(Prefix.REXW); } } } private void prefix(AMD64Address adr, Register reg) { prefix(adr, reg, false); } private void prefix(AMD64Address adr, Register reg, boolean byteinst) { if (reg.encoding < 8) { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXXB); } else { emitByte(Prefix.REXB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXX); } else if (byteinst && reg.encoding >= 4) { emitByte(Prefix.REX); } } } else { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXRXB); } else { emitByte(Prefix.REXRB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXRX); } else { emitByte(Prefix.REXR); } } } } private void prefixq(AMD64Address adr, Register src) { if (src.encoding < 8) { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWXB); } else { emitByte(Prefix.REXWB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWX); } else { emitByte(Prefix.REXW); } } } else { if (needsRex(adr.getBase())) { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWRXB); } else { emitByte(Prefix.REXWRB); } } else { if (needsRex(adr.getIndex())) { emitByte(Prefix.REXWRX); } else { emitByte(Prefix.REXWR); } } } } public final void addq(Register dst, int imm32) { ADD.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void addq(AMD64Address dst, int imm32) { ADD.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void addq(Register dst, Register src) { ADD.rmOp.emit(this, QWORD, dst, src); } public final void addq(AMD64Address dst, Register src) { ADD.mrOp.emit(this, QWORD, dst, src); } public final void andq(Register dst, int imm32) { AND.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void bswapq(Register reg) { int encode = prefixqAndEncode(reg.encoding); emitByte(0x0F); emitByte(0xC8 | encode); } public final void cdqq() { emitByte(Prefix.REXW); emitByte(0x99); } public final void cmovq(ConditionFlag cc, Register dst, Register src) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x40 | cc.getValue()); emitByte(0xC0 | encode); } public final void cmovq(ConditionFlag cc, Register dst, AMD64Address src) { prefixq(src, dst); emitByte(0x0F); emitByte(0x40 | cc.getValue()); emitOperandHelper(dst, src); } public final void cmpq(Register dst, int imm32) { CMP.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void cmpq(Register dst, Register src) { CMP.rmOp.emit(this, QWORD, dst, src); } public final void cmpq(Register dst, AMD64Address src) { CMP.rmOp.emit(this, QWORD, dst, src); } public final void cmpxchgq(Register reg, AMD64Address adr) { prefixq(adr, reg); emitByte(0x0F); emitByte(0xB1); emitOperandHelper(reg, adr); } protected final void decq(Register dst) { // Use two-byte form (one-byte from is a REX prefix in 64-bit mode) int encode = prefixqAndEncode(dst.encoding); emitByte(0xFF); emitByte(0xC8 | encode); } public final void decq(AMD64Address dst) { DEC.emit(this, QWORD, dst); } public final void incq(Register dst) { // Don't use it directly. Use Macroincrementq() instead. // Use two-byte form (one-byte from is a REX prefix in 64-bit mode) int encode = prefixqAndEncode(dst.encoding); emitByte(0xFF); emitByte(0xC0 | encode); } public final void incq(AMD64Address dst) { INC.emit(this, QWORD, dst); } public final void movq(Register dst, long imm64) { int encode = prefixqAndEncode(dst.encoding); emitByte(0xB8 | encode); emitLong(imm64); } public final void movslq(Register dst, int imm32) { int encode = prefixqAndEncode(dst.encoding); emitByte(0xC7); emitByte(0xC0 | encode); emitInt(imm32); } public final void movdq(Register dst, Register src) { // table D-1 says MMX/SSE2 emitByte(0x66); if (dst.getRegisterCategory().equals(AMD64.XMM)) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x0F); emitByte(0x6E); emitByte(0xC0 | encode); } else if (src.getRegisterCategory().equals(AMD64.XMM)) { // swap src/dst to get correct prefix int encode = prefixqAndEncode(src.encoding, dst.encoding); emitByte(0x0F); emitByte(0x7E); emitByte(0xC0 | encode); } else { throw new InternalError("should not reach here"); } } public final void movdqu(Register dst, AMD64Address src) { emitByte(0xF3); prefix(src, dst); emitByte(0x0F); emitByte(0x6F); emitOperandHelper(dst, src); } public final void movslq(AMD64Address dst, int imm32) { prefixq(dst); emitByte(0xC7); emitOperandHelper(0, dst); emitInt(imm32); } public final void movslq(Register dst, AMD64Address src) { prefixq(src, dst); emitByte(0x63); emitOperandHelper(dst, src); } public final void movslq(Register dst, Register src) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x63); emitByte(0xC0 | encode); } public final void negq(Register dst) { int encode = prefixqAndEncode(dst.encoding); emitByte(0xF7); emitByte(0xD8 | encode); } public final void shlq(Register dst, int imm8) { assert isShiftCount(imm8 >> 1) : "illegal shift count"; int encode = prefixqAndEncode(dst.encoding); if (imm8 == 1) { emitByte(0xD1); emitByte(0xE0 | encode); } else { emitByte(0xC1); emitByte(0xE0 | encode); emitByte(imm8); } } public final void shrq(Register dst, int imm8) { assert isShiftCount(imm8 >> 1) : "illegal shift count"; int encode = prefixqAndEncode(dst.encoding); if (imm8 == 1) { emitByte(0xD1); emitByte(0xE8 | encode); } else { emitByte(0xC1); emitByte(0xE8 | encode); emitByte(imm8); } } public final void subq(Register dst, int imm32) { SUB.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void subq(AMD64Address dst, int imm32) { SUB.getMIOpcode(QWORD, isByte(imm32)).emit(this, QWORD, dst, imm32); } public final void subqWide(Register dst, int imm32) { // don't use the sign-extending version, forcing a 32-bit immediate SUB.getMIOpcode(QWORD, false).emit(this, QWORD, dst, imm32); } public final void subq(Register dst, Register src) { SUB.rmOp.emit(this, QWORD, dst, src); } public final void testq(Register dst, Register src) { int encode = prefixqAndEncode(dst.encoding, src.encoding); emitByte(0x85); emitByte(0xC0 | encode); } public final void xaddl(AMD64Address dst, Register src) { prefix(dst, src); emitByte(0x0F); emitByte(0xC1); emitOperandHelper(src, dst); } public final void xaddq(AMD64Address dst, Register src) { prefixq(dst, src); emitByte(0x0F); emitByte(0xC1); emitOperandHelper(src, dst); } public final void xchgl(Register dst, AMD64Address src) { prefix(src, dst); emitByte(0x87); emitOperandHelper(dst, src); } public final void xchgq(Register dst, AMD64Address src) { prefixq(src, dst); emitByte(0x87); emitOperandHelper(dst, src); } public final void membar(int barriers) { if (target.isMP) { // We only have to handle StoreLoad if ((barriers & STORE_LOAD) != 0) { // All usable chips support "locked" instructions which suffice // as barriers, and are much faster than the alternative of // using cpuid instruction. We use here a locked add [rsp],0. // This is conveniently otherwise a no-op except for blowing // flags. // Any change to this code may need to revisit other places in // the code where this idiom is used, in particular the // orderAccess code. lock(); addl(new AMD64Address(rsp, 0), 0); // Assert the lock# signal here } } } @Override protected final void patchJumpTarget(int branch, int branchTarget) { int op = getByte(branch); assert op == 0xE8 // call || op == 0x00 // jump table entry || op == 0xE9 // jmp || op == 0xEB // short jmp || (op & 0xF0) == 0x70 // short jcc || op == 0x0F && (getByte(branch + 1) & 0xF0) == 0x80 // jcc : "Invalid opcode at patch point branch=" + branch + ", branchTarget=" + branchTarget + ", op=" + op; if (op == 0x00) { int offsetToJumpTableBase = getShort(branch + 1); int jumpTableBase = branch - offsetToJumpTableBase; int imm32 = branchTarget - jumpTableBase; emitInt(imm32, branch); } else if (op == 0xEB || (op & 0xF0) == 0x70) { // short offset operators (jmp and jcc) final int imm8 = branchTarget - (branch + 2); /* * Since a wrongly patched short branch can potentially lead to working but really bad * behaving code we should always fail with an exception instead of having an assert. */ if (!NumUtil.isByte(imm8)) { throw new InternalError("branch displacement out of range: " + imm8); } emitByte(imm8, branch + 1); } else { int off = 1; if (op == 0x0F) { off = 2; } int imm32 = branchTarget - (branch + 4 + off); emitInt(imm32, branch + off); } } public void nullCheck(AMD64Address address) { testl(AMD64.rax, address); } @Override public void align(int modulus) { if (position() % modulus != 0) { nop(modulus - (position() % modulus)); } } /** * Emits a direct call instruction. Note that the actual call target is not specified, because * all calls need patching anyway. Therefore, 0 is emitted as the call target, and the user is * responsible to add the call address to the appropriate patching tables. */ public final void call() { emitByte(0xE8); emitInt(0); } public final void call(Register src) { int encode = prefixAndEncode(src.encoding); emitByte(0xFF); emitByte(0xD0 | encode); } public final void int3() { emitByte(0xCC); } private void emitx87(int b1, int b2, int i) { assert 0 <= i && i < 8 : "illegal stack offset"; emitByte(b1); emitByte(b2 + i); } public final void fldd(AMD64Address src) { emitByte(0xDD); emitOperandHelper(0, src); } public final void flds(AMD64Address src) { emitByte(0xD9); emitOperandHelper(0, src); } public final void fldln2() { emitByte(0xD9); emitByte(0xED); } public final void fldlg2() { emitByte(0xD9); emitByte(0xEC); } public final void fyl2x() { emitByte(0xD9); emitByte(0xF1); } public final void fstps(AMD64Address src) { emitByte(0xD9); emitOperandHelper(3, src); } public final void fstpd(AMD64Address src) { emitByte(0xDD); emitOperandHelper(3, src); } private void emitFPUArith(int b1, int b2, int i) { assert 0 <= i && i < 8 : "illegal FPU register: " + i; emitByte(b1); emitByte(b2 + i); } public void ffree(int i) { emitFPUArith(0xDD, 0xC0, i); } public void fincstp() { emitByte(0xD9); emitByte(0xF7); } public void fxch(int i) { emitFPUArith(0xD9, 0xC8, i); } public void fnstswAX() { emitByte(0xDF); emitByte(0xE0); } public void fwait() { emitByte(0x9B); } public void fprem() { emitByte(0xD9); emitByte(0xF8); } public final void fsin() { emitByte(0xD9); emitByte(0xFE); } public final void fcos() { emitByte(0xD9); emitByte(0xFF); } public final void fptan() { emitByte(0xD9); emitByte(0xF2); } public final void fstp(int i) { emitx87(0xDD, 0xD8, i); } @Override public AMD64Address makeAddress(Register base, int displacement) { return new AMD64Address(base, displacement); } @Override public AMD64Address getPlaceholder() { return Placeholder; } private void prefetchPrefix(AMD64Address src) { prefix(src); emitByte(0x0F); } public void prefetchnta(AMD64Address src) { prefetchPrefix(src); emitByte(0x18); emitOperandHelper(0, src); } void prefetchr(AMD64Address src) { assert supports(CPUFeature.AMD_3DNOW_PREFETCH); prefetchPrefix(src); emitByte(0x0D); emitOperandHelper(0, src); } public void prefetcht0(AMD64Address src) { assert supports(CPUFeature.SSE); prefetchPrefix(src); emitByte(0x18); emitOperandHelper(1, src); } public void prefetcht1(AMD64Address src) { assert supports(CPUFeature.SSE); prefetchPrefix(src); emitByte(0x18); emitOperandHelper(2, src); } public void prefetcht2(AMD64Address src) { assert supports(CPUFeature.SSE); prefix(src); emitByte(0x0f); emitByte(0x18); emitOperandHelper(3, src); } public void prefetchw(AMD64Address src) { assert supports(CPUFeature.AMD_3DNOW_PREFETCH); prefix(src); emitByte(0x0f); emitByte(0x0D); emitOperandHelper(1, src); } /** * Emits an instruction which is considered to be illegal. This is used if we deliberately want * to crash the program (debugging etc.). */ public void illegal() { emitByte(0x0f); emitByte(0x0b); } }