6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-01-01 00:33:09 +00:00
Code Issues Projects Releases Wiki Activity

LoopVectorizer: Fix a bug in the vectorization of BinaryOperators. The BinaryOperator can be folded to an Undef, and we don't want to set NSW flags to undef vals.

Browse Source 
PR14878



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@172079 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Nadav Rotem 2013-01-10 17:34:39 +00:00
parent 12da505d93
commit c560bf638b
2 changed files with 29 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
lib/Transforms/Vectorize/LoopVectorize.cpp
View File

@ -1682,13 +1682,13 @@ InnerLoopVectorizer::vectorizeBlockInLoop(LoopVectorizationLegality *Legal,
for (unsigned Part = 0; Part < UF; ++Part) { for (unsigned Part = 0; Part < UF; ++Part) {
Value *V = Builder.CreateBinOp(BinOp->getOpcode(), A[Part], B[Part]); Value *V = Builder.CreateBinOp(BinOp->getOpcode(), A[Part], B[Part]);
// Update the NSW, NUW and Exact flags. // Update the NSW, NUW and Exact flags. Notice: V can be an Undef.
BinaryOperator *VecOp = cast<BinaryOperator>(V); BinaryOperator *VecOp = dyn_cast<BinaryOperator>(V);
if (isa<OverflowingBinaryOperator>(BinOp)) { if (VecOp && isa<OverflowingBinaryOperator>(BinOp)) {
VecOp->setHasNoSignedWrap(BinOp->hasNoSignedWrap()); VecOp->setHasNoSignedWrap(BinOp->hasNoSignedWrap());
VecOp->setHasNoUnsignedWrap(BinOp->hasNoUnsignedWrap()); VecOp->setHasNoUnsignedWrap(BinOp->hasNoUnsignedWrap());
} }
if (isa<PossiblyExactOperator>(VecOp)) if (VecOp && isa<PossiblyExactOperator>(VecOp))
VecOp->setIsExact(BinOp->isExact()); VecOp->setIsExact(BinOp->isExact());
Entry[Part] = V; Entry[Part] = V;

25
test/Transforms/LoopVectorize/nsw-crash.ll Normal file
View File

@ -0,0 +1,25 @@
; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4
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"
target triple = "x86_64-apple-macosx10.7.0"
define void @test() {
entry:
br i1 undef, label %while.end, label %while.body.lr.ph
while.body.lr.ph:
br label %while.body
while.body:
%it.sroa.0.091 = phi i32* [ undef, %while.body.lr.ph ], [ %incdec.ptr.i, %while.body ]
%incdec.ptr.i = getelementptr inbounds i32* %it.sroa.0.091, i64 1
%inc32 = add i32 undef, 1 ; <------------- Make sure we don't set NSW flags to the undef.
%cmp.i11 = icmp eq i32* %incdec.ptr.i, undef
br i1 %cmp.i11, label %while.end, label %while.body
while.end:
ret void
}