mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-30 23:42:52 +01:00
7e501cf4c3
This update was done with the following bash script:
find test/CodeGen -name "*.ll" | \
while read NAME; do
echo "$NAME"
if ! grep -q "^; *RUN: *llc.*debug" $NAME; then
TEMP=`mktemp -t temp`
cp $NAME $TEMP
sed -n "s/^define [^@]*@\([A-Za-z0-9_]*\)(.*$/\1/p" < $NAME | \
while read FUNC; do
sed -i '' "s/;\(.*\)\([A-Za-z0-9_-]*\):\( *\)$FUNC: *\$/;\1\2-LABEL:\3$FUNC:/g" $TEMP
done
sed -i '' "s/;\(.*\)-LABEL-LABEL:/;\1-LABEL:/" $TEMP
sed -i '' "s/;\(.*\)-NEXT-LABEL:/;\1-NEXT:/" $TEMP
sed -i '' "s/;\(.*\)-NOT-LABEL:/;\1-NOT:/" $TEMP
sed -i '' "s/;\(.*\)-DAG-LABEL:/;\1-DAG:/" $TEMP
mv $TEMP $NAME
fi
done
llvm-svn: 186280
37 lines
1.5 KiB
LLVM
37 lines
1.5 KiB
LLVM
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mattr=+sse2 -mcpu=nehalem | FileCheck %s
|
|
|
|
; rdar: 12558838
|
|
; PR14221
|
|
; There is a mismatch between the intrinsic and the actual instruction.
|
|
; The actual instruction has a partial update of dest, while the intrinsic
|
|
; passes through the upper FP values. Here, we make sure the source and
|
|
; destination of rsqrtss are the same.
|
|
define void @t1(<4 x float> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: t1:
|
|
; CHECK: rsqrtss %xmm0, %xmm0
|
|
%0 = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %a) nounwind
|
|
%a.addr.0.extract = extractelement <4 x float> %0, i32 0
|
|
%conv = fpext float %a.addr.0.extract to double
|
|
%a.addr.4.extract = extractelement <4 x float> %0, i32 1
|
|
%conv3 = fpext float %a.addr.4.extract to double
|
|
tail call void @callee(double %conv, double %conv3) nounwind
|
|
ret void
|
|
}
|
|
declare void @callee(double, double)
|
|
declare <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float>) nounwind readnone
|
|
|
|
define void @t2(<4 x float> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: t2:
|
|
; CHECK: rcpss %xmm0, %xmm0
|
|
%0 = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %a) nounwind
|
|
%a.addr.0.extract = extractelement <4 x float> %0, i32 0
|
|
%conv = fpext float %a.addr.0.extract to double
|
|
%a.addr.4.extract = extractelement <4 x float> %0, i32 1
|
|
%conv3 = fpext float %a.addr.4.extract to double
|
|
tail call void @callee(double %conv, double %conv3) nounwind
|
|
ret void
|
|
}
|
|
declare <4 x float> @llvm.x86.sse.rcp.ss(<4 x float>) nounwind readnone
|