Decode: Pull instruction decoding out of the StaticInst class into its own.
This change pulls the instruction decoding machinery (including caches) out of
the StaticInst class and puts it into its own class. This has a few intrinsic
benefits. First, the StaticInst code, which has gotten to be quite large, gets
simpler. Second, the code that handles decode caching is now separated out
into its own component and can be looked at in isolation, making it easier to
understand. I took the opportunity to restructure the code a bit which will
hopefully also help.
Beyond that, this change also lays some ground work for each ISA to have its
own, potentially stateful decode object. We'd be able to include less
contextualizing information in the ExtMachInst objects since that context
would be applied at the decoder. Also, the decoder could "know" ahead of time
that all the instructions it's going to see are going to be, for instance, 64
bit mode, and it will have one less thing to check when it decodes them.
Because the decode caching mechanism has been separated out, it's now possible
to have multiple caches which correspond to different types of decoding
context. Having one cache for each element of the cross product of different
configurations may become prohibitive, so it may be desirable to clear out the
cache when relatively static state changes and not to have one for each
setting.
Because the decode function is no longer universally accessible as a static
member of the StaticInst class, a new function was added to the ThreadContexts
that returns the applicable decode object.
2011-09-09 11:30:01 +02:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2011 Google
|
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions are
|
|
|
|
* met: redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer;
|
|
|
|
* redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution;
|
|
|
|
* neither the name of the copyright holders nor the names of its
|
|
|
|
* contributors may be used to endorse or promote products derived from
|
|
|
|
* this software without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
|
|
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
|
|
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
|
|
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
|
|
|
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
|
|
|
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
|
|
|
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
|
|
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
*
|
|
|
|
* Authors: Gabe Black
|
|
|
|
*/
|
|
|
|
|
2012-05-25 09:55:24 +02:00
|
|
|
#include "arch/x86/decoder.hh"
|
2012-05-26 22:44:46 +02:00
|
|
|
#include "arch/x86/regs/misc.hh"
|
|
|
|
#include "base/misc.hh"
|
|
|
|
#include "base/trace.hh"
|
|
|
|
#include "base/types.hh"
|
|
|
|
#include "cpu/thread_context.hh"
|
|
|
|
#include "debug/Decoder.hh"
|
Decode: Pull instruction decoding out of the StaticInst class into its own.
This change pulls the instruction decoding machinery (including caches) out of
the StaticInst class and puts it into its own class. This has a few intrinsic
benefits. First, the StaticInst code, which has gotten to be quite large, gets
simpler. Second, the code that handles decode caching is now separated out
into its own component and can be looked at in isolation, making it easier to
understand. I took the opportunity to restructure the code a bit which will
hopefully also help.
Beyond that, this change also lays some ground work for each ISA to have its
own, potentially stateful decode object. We'd be able to include less
contextualizing information in the ExtMachInst objects since that context
would be applied at the decoder. Also, the decoder could "know" ahead of time
that all the instructions it's going to see are going to be, for instance, 64
bit mode, and it will have one less thing to check when it decodes them.
Because the decode caching mechanism has been separated out, it's now possible
to have multiple caches which correspond to different types of decoding
context. Having one cache for each element of the cross product of different
configurations may become prohibitive, so it may be desirable to clear out the
cache when relatively static state changes and not to have one for each
setting.
Because the decode function is no longer universally accessible as a static
member of the StaticInst class, a new function was added to the ThreadContexts
that returns the applicable decode object.
2011-09-09 11:30:01 +02:00
|
|
|
|
2012-05-25 09:55:24 +02:00
|
|
|
namespace X86ISA
|
2012-05-25 09:53:37 +02:00
|
|
|
{
|
2012-05-26 22:44:46 +02:00
|
|
|
void Decoder::doReset()
|
|
|
|
{
|
|
|
|
origPC = basePC + offset;
|
|
|
|
DPRINTF(Decoder, "Setting origPC to %#x\n", origPC);
|
|
|
|
emi.rex = 0;
|
|
|
|
emi.legacy = 0;
|
|
|
|
emi.opcode.num = 0;
|
|
|
|
emi.opcode.op = 0;
|
|
|
|
emi.opcode.prefixA = emi.opcode.prefixB = 0;
|
|
|
|
|
|
|
|
immediateCollected = 0;
|
|
|
|
emi.immediate = 0;
|
|
|
|
emi.displacement = 0;
|
|
|
|
emi.dispSize = 0;
|
|
|
|
|
|
|
|
emi.modRM = 0;
|
|
|
|
emi.sib = 0;
|
|
|
|
m5Reg = tc->readMiscRegNoEffect(MISCREG_M5_REG);
|
|
|
|
emi.mode.mode = m5Reg.mode;
|
|
|
|
emi.mode.submode = m5Reg.submode;
|
|
|
|
}
|
|
|
|
|
|
|
|
void Decoder::process()
|
|
|
|
{
|
|
|
|
//This function drives the decoder state machine.
|
|
|
|
|
|
|
|
//Some sanity checks. You shouldn't try to process more bytes if
|
|
|
|
//there aren't any, and you shouldn't overwrite an already
|
|
|
|
//decoder ExtMachInst.
|
|
|
|
assert(!outOfBytes);
|
|
|
|
assert(!instDone);
|
|
|
|
|
|
|
|
//While there's still something to do...
|
|
|
|
while(!instDone && !outOfBytes)
|
|
|
|
{
|
|
|
|
uint8_t nextByte = getNextByte();
|
|
|
|
switch(state)
|
|
|
|
{
|
|
|
|
case ResetState:
|
|
|
|
doReset();
|
|
|
|
state = PrefixState;
|
|
|
|
case PrefixState:
|
|
|
|
state = doPrefixState(nextByte);
|
|
|
|
break;
|
|
|
|
case OpcodeState:
|
|
|
|
state = doOpcodeState(nextByte);
|
|
|
|
break;
|
|
|
|
case ModRMState:
|
|
|
|
state = doModRMState(nextByte);
|
|
|
|
break;
|
|
|
|
case SIBState:
|
|
|
|
state = doSIBState(nextByte);
|
|
|
|
break;
|
|
|
|
case DisplacementState:
|
|
|
|
state = doDisplacementState();
|
|
|
|
break;
|
|
|
|
case ImmediateState:
|
|
|
|
state = doImmediateState();
|
|
|
|
break;
|
|
|
|
case ErrorState:
|
|
|
|
panic("Went to the error state in the decoder.\n");
|
|
|
|
default:
|
|
|
|
panic("Unrecognized state! %d\n", state);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
//Either get a prefix and record it in the ExtMachInst, or send the
|
|
|
|
//state machine on to get the opcode(s).
|
|
|
|
Decoder::State Decoder::doPrefixState(uint8_t nextByte)
|
|
|
|
{
|
|
|
|
uint8_t prefix = Prefixes[nextByte];
|
|
|
|
State nextState = PrefixState;
|
|
|
|
// REX prefixes are only recognized in 64 bit mode.
|
|
|
|
if (prefix == RexPrefix && emi.mode.submode != SixtyFourBitMode)
|
|
|
|
prefix = 0;
|
|
|
|
if (prefix)
|
|
|
|
consumeByte();
|
|
|
|
switch(prefix)
|
|
|
|
{
|
|
|
|
//Operand size override prefixes
|
|
|
|
case OperandSizeOverride:
|
|
|
|
DPRINTF(Decoder, "Found operand size override prefix.\n");
|
|
|
|
emi.legacy.op = true;
|
|
|
|
break;
|
|
|
|
case AddressSizeOverride:
|
|
|
|
DPRINTF(Decoder, "Found address size override prefix.\n");
|
|
|
|
emi.legacy.addr = true;
|
|
|
|
break;
|
|
|
|
//Segment override prefixes
|
|
|
|
case CSOverride:
|
|
|
|
case DSOverride:
|
|
|
|
case ESOverride:
|
|
|
|
case FSOverride:
|
|
|
|
case GSOverride:
|
|
|
|
case SSOverride:
|
|
|
|
DPRINTF(Decoder, "Found segment override.\n");
|
|
|
|
emi.legacy.seg = prefix;
|
|
|
|
break;
|
|
|
|
case Lock:
|
|
|
|
DPRINTF(Decoder, "Found lock prefix.\n");
|
|
|
|
emi.legacy.lock = true;
|
|
|
|
break;
|
|
|
|
case Rep:
|
|
|
|
DPRINTF(Decoder, "Found rep prefix.\n");
|
|
|
|
emi.legacy.rep = true;
|
|
|
|
break;
|
|
|
|
case Repne:
|
|
|
|
DPRINTF(Decoder, "Found repne prefix.\n");
|
|
|
|
emi.legacy.repne = true;
|
|
|
|
break;
|
|
|
|
case RexPrefix:
|
|
|
|
DPRINTF(Decoder, "Found Rex prefix %#x.\n", nextByte);
|
|
|
|
emi.rex = nextByte;
|
|
|
|
break;
|
|
|
|
case 0:
|
|
|
|
nextState = OpcodeState;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
panic("Unrecognized prefix %#x\n", nextByte);
|
|
|
|
}
|
|
|
|
return nextState;
|
|
|
|
}
|
|
|
|
|
|
|
|
//Load all the opcodes (currently up to 2) and then figure out
|
|
|
|
//what immediate and/or ModRM is needed.
|
|
|
|
Decoder::State Decoder::doOpcodeState(uint8_t nextByte)
|
|
|
|
{
|
|
|
|
State nextState = ErrorState;
|
|
|
|
emi.opcode.num++;
|
|
|
|
//We can't handle 3+ byte opcodes right now
|
|
|
|
assert(emi.opcode.num < 4);
|
|
|
|
consumeByte();
|
|
|
|
if(emi.opcode.num == 1 && nextByte == 0x0f)
|
|
|
|
{
|
|
|
|
nextState = OpcodeState;
|
|
|
|
DPRINTF(Decoder, "Found two byte opcode.\n");
|
|
|
|
emi.opcode.prefixA = nextByte;
|
|
|
|
}
|
|
|
|
else if(emi.opcode.num == 2 && (nextByte == 0x38 || nextByte == 0x3A))
|
|
|
|
{
|
|
|
|
nextState = OpcodeState;
|
|
|
|
DPRINTF(Decoder, "Found three byte opcode.\n");
|
|
|
|
emi.opcode.prefixB = nextByte;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
DPRINTF(Decoder, "Found opcode %#x.\n", nextByte);
|
|
|
|
emi.opcode.op = nextByte;
|
|
|
|
|
|
|
|
//Figure out the effective operand size. This can be overriden to
|
|
|
|
//a fixed value at the decoder level.
|
|
|
|
int logOpSize;
|
|
|
|
if (emi.rex.w)
|
|
|
|
logOpSize = 3; // 64 bit operand size
|
|
|
|
else if (emi.legacy.op)
|
|
|
|
logOpSize = m5Reg.altOp;
|
|
|
|
else
|
|
|
|
logOpSize = m5Reg.defOp;
|
|
|
|
|
|
|
|
//Set the actual op size
|
|
|
|
emi.opSize = 1 << logOpSize;
|
|
|
|
|
|
|
|
//Figure out the effective address size. This can be overriden to
|
|
|
|
//a fixed value at the decoder level.
|
|
|
|
int logAddrSize;
|
|
|
|
if(emi.legacy.addr)
|
|
|
|
logAddrSize = m5Reg.altAddr;
|
|
|
|
else
|
|
|
|
logAddrSize = m5Reg.defAddr;
|
|
|
|
|
|
|
|
//Set the actual address size
|
|
|
|
emi.addrSize = 1 << logAddrSize;
|
|
|
|
|
|
|
|
//Figure out the effective stack width. This can be overriden to
|
|
|
|
//a fixed value at the decoder level.
|
|
|
|
emi.stackSize = 1 << m5Reg.stack;
|
|
|
|
|
|
|
|
//Figure out how big of an immediate we'll retreive based
|
|
|
|
//on the opcode.
|
|
|
|
int immType = ImmediateType[emi.opcode.num - 1][nextByte];
|
|
|
|
if (emi.opcode.num == 1 && nextByte >= 0xA0 && nextByte <= 0xA3)
|
|
|
|
immediateSize = SizeTypeToSize[logAddrSize - 1][immType];
|
|
|
|
else
|
|
|
|
immediateSize = SizeTypeToSize[logOpSize - 1][immType];
|
|
|
|
|
|
|
|
//Determine what to expect next
|
|
|
|
if (UsesModRM[emi.opcode.num - 1][nextByte]) {
|
|
|
|
nextState = ModRMState;
|
|
|
|
} else {
|
|
|
|
if(immediateSize) {
|
|
|
|
nextState = ImmediateState;
|
|
|
|
} else {
|
|
|
|
instDone = true;
|
|
|
|
nextState = ResetState;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return nextState;
|
|
|
|
}
|
|
|
|
|
|
|
|
//Get the ModRM byte and determine what displacement, if any, there is.
|
|
|
|
//Also determine whether or not to get the SIB byte, displacement, or
|
|
|
|
//immediate next.
|
|
|
|
Decoder::State Decoder::doModRMState(uint8_t nextByte)
|
|
|
|
{
|
|
|
|
State nextState = ErrorState;
|
|
|
|
ModRM modRM;
|
|
|
|
modRM = nextByte;
|
|
|
|
DPRINTF(Decoder, "Found modrm byte %#x.\n", nextByte);
|
|
|
|
if (m5Reg.defOp == 1) {
|
|
|
|
//figure out 16 bit displacement size
|
|
|
|
if ((modRM.mod == 0 && modRM.rm == 6) || modRM.mod == 2)
|
|
|
|
displacementSize = 2;
|
|
|
|
else if (modRM.mod == 1)
|
|
|
|
displacementSize = 1;
|
|
|
|
else
|
|
|
|
displacementSize = 0;
|
|
|
|
} else {
|
|
|
|
//figure out 32/64 bit displacement size
|
|
|
|
if ((modRM.mod == 0 && modRM.rm == 5) || modRM.mod == 2)
|
|
|
|
displacementSize = 4;
|
|
|
|
else if (modRM.mod == 1)
|
|
|
|
displacementSize = 1;
|
|
|
|
else
|
|
|
|
displacementSize = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
// The "test" instruction in group 3 needs an immediate, even though
|
|
|
|
// the other instructions with the same actual opcode don't.
|
|
|
|
if (emi.opcode.num == 1 && (modRM.reg & 0x6) == 0) {
|
|
|
|
if (emi.opcode.op == 0xF6)
|
|
|
|
immediateSize = 1;
|
|
|
|
else if (emi.opcode.op == 0xF7)
|
|
|
|
immediateSize = (emi.opSize == 8) ? 4 : emi.opSize;
|
|
|
|
}
|
|
|
|
|
|
|
|
//If there's an SIB, get that next.
|
|
|
|
//There is no SIB in 16 bit mode.
|
|
|
|
if (modRM.rm == 4 && modRM.mod != 3) {
|
|
|
|
// && in 32/64 bit mode)
|
|
|
|
nextState = SIBState;
|
|
|
|
} else if(displacementSize) {
|
|
|
|
nextState = DisplacementState;
|
|
|
|
} else if(immediateSize) {
|
|
|
|
nextState = ImmediateState;
|
|
|
|
} else {
|
|
|
|
instDone = true;
|
|
|
|
nextState = ResetState;
|
|
|
|
}
|
|
|
|
//The ModRM byte is consumed no matter what
|
|
|
|
consumeByte();
|
|
|
|
emi.modRM = modRM;
|
|
|
|
return nextState;
|
|
|
|
}
|
|
|
|
|
|
|
|
//Get the SIB byte. We don't do anything with it at this point, other
|
|
|
|
//than storing it in the ExtMachInst. Determine if we need to get a
|
|
|
|
//displacement or immediate next.
|
|
|
|
Decoder::State Decoder::doSIBState(uint8_t nextByte)
|
|
|
|
{
|
|
|
|
State nextState = ErrorState;
|
|
|
|
emi.sib = nextByte;
|
|
|
|
DPRINTF(Decoder, "Found SIB byte %#x.\n", nextByte);
|
|
|
|
consumeByte();
|
|
|
|
if (emi.modRM.mod == 0 && emi.sib.base == 5)
|
|
|
|
displacementSize = 4;
|
|
|
|
if (displacementSize) {
|
|
|
|
nextState = DisplacementState;
|
|
|
|
} else if(immediateSize) {
|
|
|
|
nextState = ImmediateState;
|
|
|
|
} else {
|
|
|
|
instDone = true;
|
|
|
|
nextState = ResetState;
|
|
|
|
}
|
|
|
|
return nextState;
|
|
|
|
}
|
|
|
|
|
|
|
|
//Gather up the displacement, or at least as much of it
|
|
|
|
//as we can get.
|
|
|
|
Decoder::State Decoder::doDisplacementState()
|
|
|
|
{
|
|
|
|
State nextState = ErrorState;
|
|
|
|
|
|
|
|
getImmediate(immediateCollected,
|
|
|
|
emi.displacement,
|
|
|
|
displacementSize);
|
|
|
|
|
|
|
|
DPRINTF(Decoder, "Collecting %d byte displacement, got %d bytes.\n",
|
|
|
|
displacementSize, immediateCollected);
|
|
|
|
|
|
|
|
if(displacementSize == immediateCollected) {
|
|
|
|
//Reset this for other immediates.
|
|
|
|
immediateCollected = 0;
|
|
|
|
//Sign extend the displacement
|
|
|
|
switch(displacementSize)
|
|
|
|
{
|
|
|
|
case 1:
|
|
|
|
emi.displacement = sext<8>(emi.displacement);
|
|
|
|
break;
|
|
|
|
case 2:
|
|
|
|
emi.displacement = sext<16>(emi.displacement);
|
|
|
|
break;
|
|
|
|
case 4:
|
|
|
|
emi.displacement = sext<32>(emi.displacement);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
panic("Undefined displacement size!\n");
|
|
|
|
}
|
|
|
|
DPRINTF(Decoder, "Collected displacement %#x.\n",
|
|
|
|
emi.displacement);
|
|
|
|
if(immediateSize) {
|
|
|
|
nextState = ImmediateState;
|
|
|
|
} else {
|
|
|
|
instDone = true;
|
|
|
|
nextState = ResetState;
|
|
|
|
}
|
|
|
|
|
|
|
|
emi.dispSize = displacementSize;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
nextState = DisplacementState;
|
|
|
|
return nextState;
|
|
|
|
}
|
|
|
|
|
|
|
|
//Gather up the immediate, or at least as much of it
|
|
|
|
//as we can get
|
|
|
|
Decoder::State Decoder::doImmediateState()
|
|
|
|
{
|
|
|
|
State nextState = ErrorState;
|
|
|
|
|
|
|
|
getImmediate(immediateCollected,
|
|
|
|
emi.immediate,
|
|
|
|
immediateSize);
|
|
|
|
|
|
|
|
DPRINTF(Decoder, "Collecting %d byte immediate, got %d bytes.\n",
|
|
|
|
immediateSize, immediateCollected);
|
|
|
|
|
|
|
|
if(immediateSize == immediateCollected)
|
|
|
|
{
|
|
|
|
//Reset this for other immediates.
|
|
|
|
immediateCollected = 0;
|
|
|
|
|
|
|
|
//XXX Warning! The following is an observed pattern and might
|
|
|
|
//not always be true!
|
|
|
|
|
|
|
|
//Instructions which use 64 bit operands but 32 bit immediates
|
|
|
|
//need to have the immediate sign extended to 64 bits.
|
|
|
|
//Instructions which use true 64 bit immediates won't be
|
|
|
|
//affected, and instructions that use true 32 bit immediates
|
|
|
|
//won't notice.
|
|
|
|
switch(immediateSize)
|
|
|
|
{
|
|
|
|
case 4:
|
|
|
|
emi.immediate = sext<32>(emi.immediate);
|
|
|
|
break;
|
|
|
|
case 1:
|
|
|
|
emi.immediate = sext<8>(emi.immediate);
|
|
|
|
}
|
|
|
|
|
|
|
|
DPRINTF(Decoder, "Collected immediate %#x.\n",
|
|
|
|
emi.immediate);
|
|
|
|
instDone = true;
|
|
|
|
nextState = ResetState;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
nextState = ImmediateState;
|
|
|
|
return nextState;
|
|
|
|
}
|
2012-05-25 09:53:37 +02:00
|
|
|
|
2012-05-26 22:45:12 +02:00
|
|
|
DecodeCache::InstMap Decoder::instMap;
|
|
|
|
DecodeCache::AddrMap<StaticInstPtr> Decoder::decodePages;
|
|
|
|
|
|
|
|
StaticInstPtr
|
|
|
|
Decoder::decode(ExtMachInst mach_inst, Addr addr)
|
|
|
|
{
|
|
|
|
StaticInstPtr &si = decodePages.lookup(addr);
|
|
|
|
if (si && (si->machInst == mach_inst))
|
|
|
|
return si;
|
|
|
|
|
|
|
|
DecodeCache::InstMap::iterator iter = instMap.find(mach_inst);
|
|
|
|
if (iter != instMap.end()) {
|
|
|
|
si = iter->second;
|
|
|
|
return si;
|
|
|
|
}
|
|
|
|
|
|
|
|
si = decodeInst(mach_inst);
|
|
|
|
instMap[mach_inst] = si;
|
|
|
|
return si;
|
|
|
|
}
|
2012-05-25 09:53:37 +02:00
|
|
|
|
|
|
|
}
|