llvm-project/llvm/lib/CodeGen/MachineCopyPropagation.cpp

372 lines
12 KiB
C++
Raw Normal View History

//===- MachineCopyPropagation.cpp - Machine Copy Propagation Pass ---------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This is an extremely simple MachineInstr-level copy propagation pass.
//
//===----------------------------------------------------------------------===//
#include "llvm/CodeGen/Passes.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/Pass.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetInstrInfo.h"
#include "llvm/Target/TargetRegisterInfo.h"
#include "llvm/Target/TargetSubtargetInfo.h"
using namespace llvm;
#define DEBUG_TYPE "codegen-cp"
STATISTIC(NumDeletes, "Number of dead copies deleted");
namespace {
typedef SmallVector<unsigned, 4> RegList;
typedef DenseMap<unsigned, RegList> SourceMap;
typedef DenseMap<unsigned, MachineInstr*> Reg2MIMap;
class MachineCopyPropagation : public MachineFunctionPass {
const TargetRegisterInfo *TRI;
const TargetInstrInfo *TII;
const MachineRegisterInfo *MRI;
2012-02-09 05:22:43 +08:00
public:
static char ID; // Pass identification, replacement for typeid
MachineCopyPropagation() : MachineFunctionPass(ID) {
initializeMachineCopyPropagationPass(*PassRegistry::getPassRegistry());
}
void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.setPreservesCFG();
MachineFunctionPass::getAnalysisUsage(AU);
}
bool runOnMachineFunction(MachineFunction &MF) override;
MachineFunctionProperties getRequiredProperties() const override {
return MachineFunctionProperties().set(
MachineFunctionProperties::Property::AllVRegsAllocated);
}
private:
void ClobberRegister(unsigned Reg);
void CopyPropagateBlock(MachineBasicBlock &MBB);
bool eraseIfRedundant(MachineInstr &Copy, unsigned Src, unsigned Def);
/// Candidates for deletion.
SmallSetVector<MachineInstr*, 8> MaybeDeadCopies;
/// Def -> available copies map.
Reg2MIMap AvailCopyMap;
/// Def -> copies map.
Reg2MIMap CopyMap;
/// Src -> Def map
SourceMap SrcMap;
bool Changed;
};
}
char MachineCopyPropagation::ID = 0;
char &llvm::MachineCopyPropagationID = MachineCopyPropagation::ID;
INITIALIZE_PASS(MachineCopyPropagation, "machine-cp",
"Machine Copy Propagation Pass", false, false)
/// Remove any entry in \p Map where the register is a subregister or equal to
/// a register contained in \p Regs.
static void removeRegsFromMap(Reg2MIMap &Map, const RegList &Regs,
const TargetRegisterInfo &TRI) {
for (unsigned Reg : Regs) {
// Source of copy is no longer available for propagation.
for (MCSubRegIterator SR(Reg, &TRI, true); SR.isValid(); ++SR)
Map.erase(*SR);
}
}
/// Remove any entry in \p Map that is marked clobbered in \p RegMask.
/// The map will typically have a lot fewer entries than the regmask clobbers,
/// so this is more efficient than iterating the clobbered registers and calling
/// ClobberRegister() on them.
static void removeClobberedRegsFromMap(Reg2MIMap &Map,
const MachineOperand &RegMask) {
for (Reg2MIMap::iterator I = Map.begin(), E = Map.end(), Next; I != E;
I = Next) {
Next = std::next(I);
unsigned Reg = I->first;
if (RegMask.clobbersPhysReg(Reg))
Map.erase(I);
}
}
void MachineCopyPropagation::ClobberRegister(unsigned Reg) {
for (MCRegAliasIterator AI(Reg, TRI, true); AI.isValid(); ++AI) {
CopyMap.erase(*AI);
AvailCopyMap.erase(*AI);
SourceMap::iterator SI = SrcMap.find(*AI);
if (SI != SrcMap.end()) {
removeRegsFromMap(AvailCopyMap, SI->second, *TRI);
SrcMap.erase(SI);
}
}
}
/// Return true if \p PreviousCopy did copy register \p Src to register \p Def.
/// This fact may have been obscured by sub register usage or may not be true at
/// all even though Src and Def are subregisters of the registers used in
/// PreviousCopy. e.g.
/// isNopCopy("ecx = COPY eax", AX, CX) == true
/// isNopCopy("ecx = COPY eax", AH, CL) == false
static bool isNopCopy(const MachineInstr &PreviousCopy, unsigned Src,
unsigned Def, const TargetRegisterInfo *TRI) {
unsigned PreviousSrc = PreviousCopy.getOperand(1).getReg();
unsigned PreviousDef = PreviousCopy.getOperand(0).getReg();
if (Src == PreviousSrc) {
assert(Def == PreviousDef);
return true;
}
if (!TRI->isSubRegister(PreviousSrc, Src))
return false;
unsigned SubIdx = TRI->getSubRegIndex(PreviousSrc, Src);
return SubIdx == TRI->getSubRegIndex(PreviousDef, Def);
}
/// Remove instruction \p Copy if there exists a previous copy that copies the
/// register \p Src to the register \p Def; This may happen indirectly by
/// copying the super registers.
bool MachineCopyPropagation::eraseIfRedundant(MachineInstr &Copy, unsigned Src,
unsigned Def) {
// Avoid eliminating a copy from/to a reserved registers as we cannot predict
// the value (Example: The sparc zero register is writable but stays zero).
if (MRI->isReserved(Src) || MRI->isReserved(Def))
return false;
// Search for an existing copy.
Reg2MIMap::iterator CI = AvailCopyMap.find(Def);
if (CI == AvailCopyMap.end())
return false;
// Check that the existing copy uses the correct sub registers.
MachineInstr &PrevCopy = *CI->second;
if (!isNopCopy(PrevCopy, Src, Def, TRI))
return false;
DEBUG(dbgs() << "MCP: copy is a NOP, removing: "; Copy.dump());
// Copy was redundantly redefining either Src or Def. Remove earlier kill
// flags between Copy and PrevCopy because the value will be reused now.
assert(Copy.isCopy());
unsigned CopyDef = Copy.getOperand(0).getReg();
assert(CopyDef == Src || CopyDef == Def);
for (MachineInstr &MI :
make_range(PrevCopy.getIterator(), Copy.getIterator()))
MI.clearRegisterKills(CopyDef, TRI);
Copy.eraseFromParent();
Changed = true;
++NumDeletes;
return true;
}
void MachineCopyPropagation::CopyPropagateBlock(MachineBasicBlock &MBB) {
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
DEBUG(dbgs() << "MCP: CopyPropagateBlock " << MBB.getName() << "\n");
for (MachineBasicBlock::iterator I = MBB.begin(), E = MBB.end(); I != E; ) {
MachineInstr *MI = &*I;
++I;
if (MI->isCopy()) {
unsigned Def = MI->getOperand(0).getReg();
unsigned Src = MI->getOperand(1).getReg();
assert(!TargetRegisterInfo::isVirtualRegister(Def) &&
!TargetRegisterInfo::isVirtualRegister(Src) &&
"MachineCopyPropagation should be run after register allocation!");
// The two copies cancel out and the source of the first copy
// hasn't been overridden, eliminate the second one. e.g.
// %ECX<def> = COPY %EAX
// ... nothing clobbered EAX.
// %EAX<def> = COPY %ECX
// =>
// %ECX<def> = COPY %EAX
//
// or
//
// %ECX<def> = COPY %EAX
// ... nothing clobbered EAX.
// %ECX<def> = COPY %EAX
// =>
// %ECX<def> = COPY %EAX
if (eraseIfRedundant(*MI, Def, Src) || eraseIfRedundant(*MI, Src, Def))
continue;
// If Src is defined by a previous copy, the previous copy cannot be
// eliminated.
for (MCRegAliasIterator AI(Src, TRI, true); AI.isValid(); ++AI) {
Reg2MIMap::iterator CI = CopyMap.find(*AI);
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
if (CI != CopyMap.end()) {
DEBUG(dbgs() << "MCP: Copy is no longer dead: "; CI->second->dump());
MaybeDeadCopies.remove(CI->second);
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
}
}
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
DEBUG(dbgs() << "MCP: Copy is a deletion candidate: "; MI->dump());
// Copy is now a candidate for deletion.
if (!MRI->isReserved(Def))
MaybeDeadCopies.insert(MI);
// If 'Def' is previously source of another copy, then this earlier copy's
// source is no longer available. e.g.
// %xmm9<def> = copy %xmm2
// ...
// %xmm2<def> = copy %xmm0
// ...
// %xmm2<def> = copy %xmm9
ClobberRegister(Def);
// Remember Def is defined by the copy.
for (MCSubRegIterator SR(Def, TRI, /*IncludeSelf=*/true); SR.isValid();
++SR) {
CopyMap[*SR] = MI;
AvailCopyMap[*SR] = MI;
}
// Remember source that's copied to Def. Once it's clobbered, then
// it's no longer available for copy propagation.
RegList &DestList = SrcMap[Src];
if (std::find(DestList.begin(), DestList.end(), Def) == DestList.end())
DestList.push_back(Def);
continue;
}
// Not a copy.
SmallVector<unsigned, 2> Defs;
const MachineOperand *RegMask = nullptr;
for (const MachineOperand &MO : MI->operands()) {
if (MO.isRegMask())
RegMask = &MO;
if (!MO.isReg())
continue;
unsigned Reg = MO.getReg();
if (!Reg)
continue;
assert(!TargetRegisterInfo::isVirtualRegister(Reg) &&
"MachineCopyPropagation should be run after register allocation!");
if (MO.isDef()) {
Defs.push_back(Reg);
continue;
}
// If 'Reg' is defined by a copy, the copy is no longer a candidate
// for elimination.
for (MCRegAliasIterator AI(Reg, TRI, true); AI.isValid(); ++AI) {
Reg2MIMap::iterator CI = CopyMap.find(*AI);
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
if (CI != CopyMap.end()) {
DEBUG(dbgs() << "MCP: Copy is used - not dead: "; CI->second->dump());
MaybeDeadCopies.remove(CI->second);
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
}
}
// Treat undef use like defs for copy propagation but not for
// dead copy. We would need to do a liveness check to be sure the copy
// is dead for undef uses.
// The backends are allowed to do whatever they want with undef value
// and we cannot be sure this register will not be rewritten to break
// some false dependencies for the hardware for instance.
if (MO.isUndef())
Defs.push_back(Reg);
}
// The instruction has a register mask operand which means that it clobbers
// a large set of registers. Treat clobbered registers the same way as
// defined registers.
if (RegMask) {
// Erase any MaybeDeadCopies whose destination register is clobbered.
for (SmallSetVector<MachineInstr *, 8>::iterator DI =
MaybeDeadCopies.begin();
DI != MaybeDeadCopies.end();) {
MachineInstr *MaybeDead = *DI;
unsigned Reg = MaybeDead->getOperand(0).getReg();
assert(!MRI->isReserved(Reg));
if (!RegMask->clobbersPhysReg(Reg)) {
++DI;
continue;
}
MachineCopyPropagation has special logic for removing COPY instructions. It will remove plain COPYs using eraseFromParent(), but if the COPY has imp-defs/imp-uses it will convert it to a KILL, to keep the imp-def around. This actually totally breaks and causes the machine verifier to cry in several cases, one of which being: %RAX<def> = COPY %RCX<kill> %ECX<def> = COPY %EAX<kill>, %RAX<imp-use,kill> These subregister copies are together identified as noops, so are both removed. However, the second one as it has an imp-use gets converted into a kill: %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> As the original COPY has been removed, the verifier goes into tears at the use of undefined EAX and RAX. There are several hacky solutions to this hacky problem (which is all to do with imp-use/def weirdnesses), but the least hacky I've come up with is to *always* remove COPYs by converting to KILLs. KILLs are no-ops to the code generator so the generated code doesn't change (which is why they were partially used in the first place), but using them also keeps the def/use and imp-def/imp-use chains alive: %RAX<def> = KILL %RCX<kill> %ECX<def> = KILL %EAX<kill>, %RAX<imp-use,kill> The patch passes all test cases including the ones that check the removal of MOVs in this circumstance, along with an extra test I added to check subregister behaviour (which made the machine verifier fall over before my patch). The patch also adds some DEBUG() statements because the file hadn't got any. llvm-svn: 199797
2014-01-22 17:12:27 +08:00
DEBUG(dbgs() << "MCP: Removing copy due to regmask clobbering: ";
MaybeDead->dump());
// erase() will return the next valid iterator pointing to the next
// element after the erased one.
DI = MaybeDeadCopies.erase(DI);
MaybeDead->eraseFromParent();
Changed = true;
++NumDeletes;
}
removeClobberedRegsFromMap(AvailCopyMap, *RegMask);
removeClobberedRegsFromMap(CopyMap, *RegMask);
for (SourceMap::iterator I = SrcMap.begin(), E = SrcMap.end(), Next;
I != E; I = Next) {
Next = std::next(I);
if (RegMask->clobbersPhysReg(I->first)) {
removeRegsFromMap(AvailCopyMap, I->second, *TRI);
SrcMap.erase(I);
}
}
}
// Any previous copy definition or reading the Defs is no longer available.
for (unsigned Reg : Defs)
ClobberRegister(Reg);
}
// If MBB doesn't have successors, delete the copies whose defs are not used.
// If MBB does have successors, then conservative assume the defs are live-out
// since we don't want to trust live-in lists.
if (MBB.succ_empty()) {
for (MachineInstr *MaybeDead : MaybeDeadCopies) {
assert(!MRI->isReserved(MaybeDead->getOperand(0).getReg()));
MaybeDead->eraseFromParent();
Changed = true;
++NumDeletes;
}
}
MaybeDeadCopies.clear();
AvailCopyMap.clear();
CopyMap.clear();
SrcMap.clear();
}
bool MachineCopyPropagation::runOnMachineFunction(MachineFunction &MF) {
if (skipFunction(*MF.getFunction()))
return false;
Changed = false;
TRI = MF.getSubtarget().getRegisterInfo();
TII = MF.getSubtarget().getInstrInfo();
MRI = &MF.getRegInfo();
for (MachineBasicBlock &MBB : MF)
CopyPropagateBlock(MBB);
return Changed;
}