replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Fix a regression from r125393; 

It caused a crash in MultiSource/Benchmarks/Bullet.
Opt hit an assertion with "opt -std-compile-opts" because
Constant::getAllOnesValue doesn't know how to handle floats.

This patch added a test to reproduce the problem and a check that the
destination vector is of integer type.

Thank you Benjamin!

llvm-svn: 125459
This commit is contained in:
Nadav Rotem 2011-02-13 15:45:34 +00:00
parent d5f0b1148a
commit 27b848afb0
2 changed files with 19 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
llvm/lib/VMCore/ConstantFold.cpp
View File

@ -43,7 +43,8 @@ using namespace llvm;
static Constant *BitCastConstantVector(ConstantVector *CV,
const VectorType *DstTy) {
if (CV->isAllOnesValue()) return Constant::getAllOnesValue(DstTy);
if (CV->isAllOnesValue() && DstTy->getElementType()->isIntegerTy())
return Constant::getAllOnesValue(DstTy);
if (CV->isNullValue()) return Constant::getNullValue(DstTy);
// If this cast changes element count then we can't handle it here:

18
llvm/test/Transforms/InstCombine/bitcast-vec-uniform.ll
View File

@ -1,14 +1,30 @@
; RUN: opt < %s -instcombine -S | not grep bitcast
; RUN: opt < %s -instcombine -S | FileCheck %s
; CHECK: @a
; CHECK-NOT: bitcast
; CHECK: ret
define <4 x i32> @a(<1 x i64> %y) {
%c = bitcast <2 x i64> <i64 0, i64 0> to <4 x i32>
ret <4 x i32> %c
}
; CHECK: @b
; CHECK: bitcast
; CHECK: ret
define <4 x i32> @b(<1 x i64> %y) {
%c = bitcast <2 x i64> <i64 -1, i64 -1> to <4 x i32>
ret <4 x i32> %c
}
; CHECK: @foo
; CHECK: bitcast
; from MultiSource/Benchmarks/Bullet
define <2 x float> @foo() {
%cast = bitcast i64 -1 to <2 x float>
ret <2 x float> %cast
}