LoopVectorize: Use vectorized loop invariant gep index anchored in loop

Use vectorized instruction instead of original instruction anchored in the
original loop.

Fixes PR16452 and t2075.c of PR16455.

llvm-svn: 185081
This commit is contained in:
Arnold Schwaighofer 2013-06-27 15:11:55 +00:00
parent 24a3ebb78d
commit a2dd195fb3
2 changed files with 56 additions and 8 deletions

View File

@ -1205,16 +1205,28 @@ void InnerLoopVectorizer::vectorizeMemoryInstruction(Instruction *Instr,
// The last index does not have to be the induction. It can be
// consecutive and be a function of the index. For example A[I+1];
unsigned NumOperands = Gep->getNumOperands();
Value *LastGepOperand = Gep->getOperand(NumOperands - 1);
VectorParts &GEPParts = getVectorValue(LastGepOperand);
Value *LastIndex = GEPParts[0];
LastIndex = Builder.CreateExtractElement(LastIndex, Zero);
unsigned LastOperand = NumOperands - 1;
// Create the new GEP with the new induction variable.
GetElementPtrInst *Gep2 = cast<GetElementPtrInst>(Gep->clone());
Gep2->setOperand(NumOperands - 1, LastIndex);
Gep2->setName("gep.indvar.idx");
for (unsigned i = 0; i < NumOperands; ++i) {
Value *GepOperand = Gep->getOperand(i);
Instruction *GepOperandInst = dyn_cast<Instruction>(GepOperand);
// Update last index or loop invariant instruction anchored in loop.
if (i == LastOperand ||
(GepOperandInst && OrigLoop->contains(GepOperandInst))) {
assert((i == LastOperand ||
SE->isLoopInvariant(SE->getSCEV(GepOperandInst), OrigLoop)) &&
"Must be last index or loop invariant");
VectorParts &GEPParts = getVectorValue(GepOperand);
Value *Index = GEPParts[0];
Index = Builder.CreateExtractElement(Index, Zero);
Gep2->setOperand(i, Index);
Gep2->setName("gep.indvar.idx");
}
}
Ptr = Builder.Insert(Gep2);
} else {
// Use the induction element ptr.

View File

@ -0,0 +1,36 @@
; RUN: opt -S -loop-vectorize -force-vector-width=4 -force-vector-unroll=1 < %s | FileCheck %s
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
; We use to fail on this loop because we did not properly handle the loop
; invariant instruction anchored in the loop when used as a getelementptr index.
; We would use the index from the original loop resulting in a use not dominated
; by the definition.
; PR16452
; Verify that we don't miscompile this loop.
; CHECK: @t
; CHECK: <4 x i32>
define void @t() {
entry:
br label %for.body
for.body:
%indvars.iv17 = phi i64 [ %indvars.next, %for.body ], [ 128, %entry ]
; Loop invariant anchored in loop.
%idxprom21 = zext i32 undef to i64
%arrayidx23 = getelementptr inbounds [100 x [100 x i32]]* undef, i64 0, i64 %idxprom21, i64 %indvars.iv17
store i32 undef, i32* %arrayidx23, align 4
%indvars.next= add i64 %indvars.iv17, -1
%0 = trunc i64 %indvars.next to i32
%cmp15 = icmp ugt i32 %0, undef
br i1 %cmp15, label %for.body, label %loopexit
loopexit:
ret void
}