maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Add a missing pattern for X86ISD::MOVLPD. rdar://10436044 

llvm-svn: 144566
This commit is contained in:
Evan Cheng 2011-11-14 20:35:52 +00:00
parent adfd200bcb
commit fb13d32b3f
2 changed files with 23 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
llvm/lib/Target/X86/X86InstrSSE.td
View File

@ -520,6 +520,8 @@ let Predicates = [HasSSE2] in {
// is during lowering, where it's not possible to recognize the fold cause
// it has two uses through a bitcast. One use disappears at isel time and the
// fold opportunity reappears.
def : Pat<(v2f64 (X86Movlpd VR128:$src1, VR128:$src2)),
(MOVSDrr VR128:$src1, (EXTRACT_SUBREG (v2f64 VR128:$src2),sub_sd))>;
def : Pat<(v4f32 (X86Movlps VR128:$src1, VR128:$src2)),
(MOVSDrr VR128:$src1, (EXTRACT_SUBREG (v4f32 VR128:$src2),sub_sd))>;
def : Pat<(v4i32 (X86Movlps VR128:$src1, VR128:$src2)),
@ -647,6 +649,9 @@ let Predicates = [HasAVX] in {
// is during lowering, where it's not possible to recognize the fold cause
// it has two uses through a bitcast. One use disappears at isel time and the
// fold opportunity reappears.
def : Pat<(v2f64 (X86Movlpd VR128:$src1, VR128:$src2)),
(VMOVSDrr VR128:$src1, (EXTRACT_SUBREG (v2f64 VR128:$src2),
sub_sd))>;
def : Pat<(v4f32 (X86Movlps VR128:$src1, VR128:$src2)),
(VMOVSDrr VR128:$src1, (EXTRACT_SUBREG (v4f32 VR128:$src2),
sub_sd))>;

18
llvm/test/CodeGen/X86/vec_shuffle-39.ll
View File

@ -49,3 +49,21 @@ entry:
store double %2, double* %0
ret void
}
; rdar://10436044
define <2 x double> @t3() nounwind readonly {
bb:
; CHECK: t3:
; CHECK: punpcklqdq %xmm1, %xmm0
; CHECK: movq (%rax), %xmm1
; CHECK: movsd %xmm1, %xmm0
%tmp0 = load i128* null, align 1
%tmp1 = load <2 x i32>* undef, align 8
%tmp2 = bitcast i128 %tmp0 to <16 x i8>
%tmp3 = bitcast <2 x i32> %tmp1 to i64
%tmp4 = insertelement <2 x i64> undef, i64 %tmp3, i32 0
%tmp5 = bitcast <16 x i8> %tmp2 to <2 x double>
%tmp6 = bitcast <2 x i64> %tmp4 to <2 x double>
%tmp7 = shufflevector <2 x double> %tmp5, <2 x double> %tmp6, <2 x i32> <i32 2, i32 1>
ret <2 x double> %tmp7
}