[llvm] r315272 - Renable r314928
Xinliang David Li via llvm-commits
llvm-commits at lists.llvm.org
Thu May 17 10:22:01 PDT 2018
I agree in principle. However there is a also advantages of keeping this in
instcombine so creating a separate pass to handle the extreme rare case
seems like an overkill to though I do object to anyone reworking the
optimization in that direction. For the time being, I have a patch ready
for review.
thanks,
David
On Thu, May 17, 2018 at 9:08 AM Michael Zolotukhin <mzolotukhin at apple.com>
wrote:
> A limit would definitely help here, but don’t we have a more fundamental
> problem here? I think the root of all evils is that in a local to an
> instruction routine we’re iterating over all instructions in its parent.
> Would it make more sense to convert it to a separate function pass (from an
> instruction visitor)? This way we can probably save some unnecessary
> traversals.
>
> Michael
>
> On May 16, 2018, at 10:06 PM, Xinliang David Li <davidxl at google.com>
> wrote:
>
> Some kind of limit is probably needed. I can take a look at it if you do
> not beat me to it.
>
> David
>
>
> On Wed, May 16, 2018 at 6:57 PM Mikhail Zolotukhin <mzolotukhin at apple.com>
> wrote:
>
>> Hi David,
>>
>> We found a huge compile time regression in an internal clang build caused
>> by this change. I looked at the change and it seems that we’re scanning all
>> phi-nodes in the basic block for every phi, which makes the algorithm
>> quadratic. Of course, it’s not always the case, but as our build showed,
>> it’s quite possible. How can we address it?
>>
>> Here is a simple script to expose the issue:
>>
>>
>> Usage:
>> > for i in `seq 4000 2000 20000`; do echo $i; python genphis.py $i | time
>> opt -instcombine -o /dev/null; done
>>
>> 4000
>> 0.20 real 0.16 user 0.00 sys
>> 6000
>> 0.37 real 0.32 user 0.00 sys
>> 8000
>> 0.98 real 0.92 user 0.01 sys
>> 10000
>> 2.65 real 2.58 user 0.01 sys
>> 12000
>> 3.79 real 3.71 user 0.02 sys
>> 14000
>> 5.08 real 4.99 user 0.02 sys
>> 16000
>> 6.80 real 6.71 user 0.03 sys
>> 18000
>> 8.41 real 8.31 user 0.03 sys
>> 20000
>> 10.43 real 10.31 user 0.04 sys
>>
>>
>> Thanks,
>> Michael
>>
>> On Oct 9, 2017, at 10:07 PM, Xinliang David Li via llvm-commits <
>> llvm-commits at lists.llvm.org> wrote:
>>
>> Author: davidxl
>> Date: Mon Oct 9 22:07:54 2017
>> New Revision: 315272
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=315272&view=rev
>> Log:
>> Renable r314928
>>
>>
>> Eliminate inttype phi with inttoptr/ptrtoint.
>>
>> This version fixed a bug in finding the matching
>> phi -- the order of the incoming blocks may be
>> different (triggered in self build on Windows).
>> A new test case is added.
>>
>> Added:
>> llvm/trunk/test/Transforms/InstCombine/intptr1.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr1.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr2.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr2.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr3.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr3.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr4.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr4.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr5.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr5.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr6.ll
>> - copied unchanged from r315107,
>> llvm/trunk/test/Transforms/InstCombine/intptr6.ll
>> llvm/trunk/test/Transforms/InstCombine/intptr7.ll
>> Modified:
>> llvm/trunk/lib/Transforms/InstCombine/InstCombineInternal.h
>> llvm/trunk/lib/Transforms/InstCombine/InstCombinePHI.cpp
>>
>> Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineInternal.h
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineInternal.h?rev=315272&r1=315271&r2=315272&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Transforms/InstCombine/InstCombineInternal.h (original)
>> +++ llvm/trunk/lib/Transforms/InstCombine/InstCombineInternal.h Mon Oct
>> 9 22:07:54 2017
>> @@ -670,6 +670,10 @@ private:
>> Instruction *FoldPHIArgGEPIntoPHI(PHINode &PN);
>> Instruction *FoldPHIArgLoadIntoPHI(PHINode &PN);
>> Instruction *FoldPHIArgZextsIntoPHI(PHINode &PN);
>> + /// If an integer typed PHI has only one use which is an IntToPtr
>> operation,
>> + /// replace the PHI with an existing pointer typed PHI if it exists.
>> Otherwise
>> + /// insert a new pointer typed PHI and replace the original one.
>> + Instruction *FoldIntegerTypedPHI(PHINode &PN);
>>
>> /// Helper function for FoldPHIArgXIntoPHI() to set debug location for
>> the
>> /// folded operation.
>>
>> Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombinePHI.cpp
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombinePHI.cpp?rev=315272&r1=315271&r2=315272&view=diff
>>
>> ==============================================================================
>> --- llvm/trunk/lib/Transforms/InstCombine/InstCombinePHI.cpp (original)
>> +++ llvm/trunk/lib/Transforms/InstCombine/InstCombinePHI.cpp Mon Oct 9
>> 22:07:54 2017
>> @@ -40,6 +40,238 @@ void InstCombiner::PHIArgMergedDebugLoc(
>> }
>> }
>>
>> +// Replace Integer typed PHI PN if the PHI's value is used as a pointer
>> value.
>> +// If there is an existing pointer typed PHI that produces the same
>> value as PN,
>> +// replace PN and the IntToPtr operation with it. Otherwise, synthesize
>> a new
>> +// PHI node:
>> +//
>> +// Case-1:
>> +// bb1:
>> +// int_init = PtrToInt(ptr_init)
>> +// br label %bb2
>> +// bb2:
>> +// int_val = PHI([int_init, %bb1], [int_val_inc, %bb2]
>> +// ptr_val = PHI([ptr_init, %bb1], [ptr_val_inc, %bb2]
>> +// ptr_val2 = IntToPtr(int_val)
>> +// ...
>> +// use(ptr_val2)
>> +// ptr_val_inc = ...
>> +// inc_val_inc = PtrToInt(ptr_val_inc)
>> +//
>> +// ==>
>> +// bb1:
>> +// br label %bb2
>> +// bb2:
>> +// ptr_val = PHI([ptr_init, %bb1], [ptr_val_inc, %bb2]
>> +// ...
>> +// use(ptr_val)
>> +// ptr_val_inc = ...
>> +//
>> +// Case-2:
>> +// bb1:
>> +// int_ptr = BitCast(ptr_ptr)
>> +// int_init = Load(int_ptr)
>> +// br label %bb2
>> +// bb2:
>> +// int_val = PHI([int_init, %bb1], [int_val_inc, %bb2]
>> +// ptr_val2 = IntToPtr(int_val)
>> +// ...
>> +// use(ptr_val2)
>> +// ptr_val_inc = ...
>> +// inc_val_inc = PtrToInt(ptr_val_inc)
>> +// ==>
>> +// bb1:
>> +// ptr_init = Load(ptr_ptr)
>> +// br label %bb2
>> +// bb2:
>> +// ptr_val = PHI([ptr_init, %bb1], [ptr_val_inc, %bb2]
>> +// ...
>> +// use(ptr_val)
>> +// ptr_val_inc = ...
>> +// ...
>> +//
>> +Instruction *InstCombiner::FoldIntegerTypedPHI(PHINode &PN) {
>> + if (!PN.getType()->isIntegerTy())
>> + return nullptr;
>> + if (!PN.hasOneUse())
>> + return nullptr;
>> +
>> + auto *IntToPtr = dyn_cast<IntToPtrInst>(PN.user_back());
>> + if (!IntToPtr)
>> + return nullptr;
>> +
>> + // Check if the pointer is actually used as pointer:
>> + auto HasPointerUse = [](Instruction *IIP) {
>> + for (User *U : IIP->users()) {
>> + Value *Ptr = nullptr;
>> + if (LoadInst *LoadI = dyn_cast<LoadInst>(U)) {
>> + Ptr = LoadI->getPointerOperand();
>> + } else if (StoreInst *SI = dyn_cast<StoreInst>(U)) {
>> + Ptr = SI->getPointerOperand();
>> + } else if (GetElementPtrInst *GI = dyn_cast<GetElementPtrInst>(U))
>> {
>> + Ptr = GI->getPointerOperand();
>> + }
>> +
>> + if (Ptr && Ptr == IIP)
>> + return true;
>> + }
>> + return false;
>> + };
>> +
>> + if (!HasPointerUse(IntToPtr))
>> + return nullptr;
>> +
>> + if (DL.getPointerSizeInBits(IntToPtr->getAddressSpace()) !=
>> + DL.getTypeSizeInBits(IntToPtr->getOperand(0)->getType()))
>> + return nullptr;
>> +
>> + SmallVector<Value *, 4> AvailablePtrVals;
>> + for (unsigned i = 0; i != PN.getNumIncomingValues(); ++i) {
>> + Value *Arg = PN.getIncomingValue(i);
>> +
>> + // First look backward:
>> + if (auto *PI = dyn_cast<PtrToIntInst>(Arg)) {
>> + AvailablePtrVals.emplace_back(PI->getOperand(0));
>> + continue;
>> + }
>> +
>> + // Next look forward:
>> + Value *ArgIntToPtr = nullptr;
>> + for (User *U : Arg->users()) {
>> + if (isa<IntToPtrInst>(U) && U->getType() == IntToPtr->getType() &&
>> + (DT.dominates(cast<Instruction>(U), PN.getIncomingBlock(i)) ||
>> + cast<Instruction>(U)->getParent() == PN.getIncomingBlock(i)))
>> {
>> + ArgIntToPtr = U;
>> + break;
>> + }
>> + }
>> +
>> + if (ArgIntToPtr) {
>> + AvailablePtrVals.emplace_back(ArgIntToPtr);
>> + continue;
>> + }
>> +
>> + // If Arg is defined by a PHI, allow it. This will also create
>> + // more opportunities iteratively.
>> + if (isa<PHINode>(Arg)) {
>> + AvailablePtrVals.emplace_back(Arg);
>> + continue;
>> + }
>> +
>> + // For a single use integer load:
>> + auto *LoadI = dyn_cast<LoadInst>(Arg);
>> + if (!LoadI)
>> + return nullptr;
>> +
>> + if (!LoadI->hasOneUse())
>> + return nullptr;
>> +
>> + // Push the integer typed Load instruction into the available
>> + // value set, and fix it up later when the pointer typed PHI
>> + // is synthesized.
>> + AvailablePtrVals.emplace_back(LoadI);
>> + }
>> +
>> + // Now search for a matching PHI
>> + auto *BB = PN.getParent();
>> + assert(AvailablePtrVals.size() == PN.getNumIncomingValues() &&
>> + "Not enough available ptr typed incoming values");
>> + PHINode *MatchingPtrPHI = nullptr;
>> + for (auto II = BB->begin(), EI =
>> BasicBlock::iterator(BB->getFirstNonPHI());
>> + II != EI; II++) {
>> + PHINode *PtrPHI = dyn_cast<PHINode>(II);
>> + if (!PtrPHI || PtrPHI == &PN || PtrPHI->getType() !=
>> IntToPtr->getType())
>> + continue;
>> + MatchingPtrPHI = PtrPHI;
>> + for (unsigned i = 0; i != PtrPHI->getNumIncomingValues(); ++i) {
>> + if (AvailablePtrVals[i] !=
>> + PtrPHI->getIncomingValueForBlock(PN.getIncomingBlock(i))) {
>> + MatchingPtrPHI = nullptr;
>> + break;
>> + }
>> + }
>> +
>> + if (MatchingPtrPHI)
>> + break;
>> + }
>> +
>> + if (MatchingPtrPHI) {
>> + assert(MatchingPtrPHI->getType() == IntToPtr->getType() &&
>> + "Phi's Type does not match with IntToPtr");
>> + // The PtrToCast + IntToPtr will be simplified later
>> + return CastInst::CreateBitOrPointerCast(MatchingPtrPHI,
>> +
>> IntToPtr->getOperand(0)->getType());
>> + }
>> +
>> + // If it requires a conversion for every PHI operand, do not do it.
>> + if (std::all_of(AvailablePtrVals.begin(), AvailablePtrVals.end(),
>> + [&](Value *V) {
>> + return (V->getType() != IntToPtr->getType()) ||
>> + isa<IntToPtrInst>(V);
>> + }))
>> + return nullptr;
>> +
>> + // If any of the operand that requires casting is a terminator
>> + // instruction, do not do it.
>> + if (std::any_of(AvailablePtrVals.begin(), AvailablePtrVals.end(),
>> + [&](Value *V) {
>> + return (V->getType() != IntToPtr->getType()) &&
>> + isa<TerminatorInst>(V);
>> + }))
>> + return nullptr;
>> +
>> + PHINode *NewPtrPHI = PHINode::Create(
>> + IntToPtr->getType(), PN.getNumIncomingValues(), PN.getName() +
>> ".ptr");
>> +
>> + InsertNewInstBefore(NewPtrPHI, PN);
>> + SmallDenseMap<Value *, Instruction *> Casts;
>> + for (unsigned i = 0; i != PN.getNumIncomingValues(); ++i) {
>> + auto *IncomingBB = PN.getIncomingBlock(i);
>> + auto *IncomingVal = AvailablePtrVals[i];
>> +
>> + if (IncomingVal->getType() == IntToPtr->getType()) {
>> + NewPtrPHI->addIncoming(IncomingVal, IncomingBB);
>> + continue;
>> + }
>> +
>> +#ifndef NDEBUG
>> + LoadInst *LoadI = dyn_cast<LoadInst>(IncomingVal);
>> + assert((isa<PHINode>(IncomingVal) ||
>> + IncomingVal->getType()->isPointerTy() ||
>> + (LoadI && LoadI->hasOneUse())) &&
>> + "Can not replace LoadInst with multiple uses");
>> +#endif
>> + // Need to insert a BitCast.
>> + // For an integer Load instruction with a single use, the load +
>> IntToPtr
>> + // cast will be simplified into a pointer load:
>> + // %v = load i64, i64* %a.ip, align 8
>> + // %v.cast = inttoptr i64 %v to float **
>> + // ==>
>> + // %v.ptrp = bitcast i64 * %a.ip to float **
>> + // %v.cast = load float *, float ** %v.ptrp, align 8
>> + Instruction *&CI = Casts[IncomingVal];
>> + if (!CI) {
>> + CI = CastInst::CreateBitOrPointerCast(IncomingVal,
>> IntToPtr->getType(),
>> + IncomingVal->getName() +
>> ".ptr");
>> + if (auto *IncomingI = dyn_cast<Instruction>(IncomingVal)) {
>> + BasicBlock::iterator InsertPos(IncomingI);
>> + InsertPos++;
>> + if (isa<PHINode>(IncomingI))
>> + InsertPos = IncomingI->getParent()->getFirstInsertionPt();
>> + InsertNewInstBefore(CI, *InsertPos);
>> + } else {
>> + auto *InsertBB = &IncomingBB->getParent()->getEntryBlock();
>> + InsertNewInstBefore(CI, *InsertBB->getFirstInsertionPt());
>> + }
>> + }
>> + NewPtrPHI->addIncoming(CI, IncomingBB);
>> + }
>> +
>> + // The PtrToCast + IntToPtr will be simplified later
>> + return CastInst::CreateBitOrPointerCast(NewPtrPHI,
>> +
>> IntToPtr->getOperand(0)->getType());
>> +}
>> +
>> /// If we have something like phi [add (a,b), add(a,c)] and if a/b/c and
>> the
>> /// adds all have a single use, turn this into a phi and a single binop.
>> Instruction *InstCombiner::FoldPHIArgBinOpIntoPHI(PHINode &PN) {
>> @@ -903,6 +1135,9 @@ Instruction *InstCombiner::visitPHINode(
>> // this PHI only has a single use (a PHI), and if that PHI only has one
>> use (a
>> // PHI)... break the cycle.
>> if (PN.hasOneUse()) {
>> + if (Instruction *Result = FoldIntegerTypedPHI(PN))
>> + return Result;
>> +
>> Instruction *PHIUser = cast<Instruction>(PN.user_back());
>> if (PHINode *PU = dyn_cast<PHINode>(PHIUser)) {
>> SmallPtrSet<PHINode*, 16> PotentiallyDeadPHIs;
>>
>> Added: llvm/trunk/test/Transforms/InstCombine/intptr7.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/intptr7.ll?rev=315272&view=auto
>>
>> ==============================================================================
>> --- llvm/trunk/test/Transforms/InstCombine/intptr7.ll (added)
>> +++ llvm/trunk/test/Transforms/InstCombine/intptr7.ll Mon Oct 9 22:07:54
>> 2017
>> @@ -0,0 +1,58 @@
>> +; RUN: opt < %s -instcombine -S | FileCheck %s
>> +
>> +define void @matching_phi(i64 %a, float* %b, i1 %cond) {
>> +; CHECK-LABEL: @matching_phi
>> +entry:
>> + %cmp1 = icmp eq i1 %cond, 0
>> + %add.int = add i64 %a, 1
>> + %add = inttoptr i64 %add.int to float *
>> +
>> + %addb = getelementptr inbounds float, float* %b, i64 2
>> + %addb.int = ptrtoint float* %addb to i64
>> + br i1 %cmp1, label %A, label %B
>> +A:
>> + br label %C
>> +B:
>> + store float 1.0e+01, float* %add, align 4
>> + br label %C
>> +
>> +C:
>> + %a.addr.03 = phi float* [ %addb, %A ], [ %add, %B ]
>> + %b.addr.02 = phi i64 [ %addb.int, %A ], [ %add.int, %B ]
>> + %tmp = inttoptr i64 %b.addr.02 to float*
>> +; CHECK: %a.addr.03 = phi
>> +; CHECK-NEXT: = load
>> + %tmp1 = load float, float* %tmp, align 4
>> + %mul.i = fmul float %tmp1, 4.200000e+01
>> + store float %mul.i, float* %a.addr.03, align 4
>> + ret void
>> +}
>> +
>> +define void @no_matching_phi(i64 %a, float* %b, i1 %cond) {
>> +; CHECK-LABEL: @no_matching_phi
>> +entry:
>> + %cmp1 = icmp eq i1 %cond, 0
>> + %add.int = add i64 %a, 1
>> + %add = inttoptr i64 %add.int to float *
>> +
>> + %addb = getelementptr inbounds float, float* %b, i64 2
>> + %addb.int = ptrtoint float* %addb to i64
>> + br i1 %cmp1, label %A, label %B
>> +A:
>> + br label %C
>> +B:
>> + store float 1.0e+01, float* %add, align 4
>> + br label %C
>> +
>> +C:
>> + %a.addr.03 = phi float* [ %addb, %A ], [ %add, %B ]
>> + %b.addr.02 = phi i64 [ %addb.int, %B ], [ %add.int, %A ]
>> + %tmp = inttoptr i64 %b.addr.02 to float*
>> + %tmp1 = load float, float* %tmp, align 4
>> +; CHECK: %a.addr.03 = phi
>> +; CHECK-NEXT: %b.addr.02.ptr = phi
>> +; CHECK-NEXT: = load
>> + %mul.i = fmul float %tmp1, 4.200000e+01
>> + store float %mul.i, float* %a.addr.03, align 4
>> + ret void
>> +}
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180517/2ee2e320/attachment.html>
More information about the llvm-commits
mailing list