diff options
| author | Craig Topper <craig.topper@intel.com> | 2019-06-03 19:29:14 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@intel.com> | 2019-06-03 19:29:14 +0000 |
| commit | dcf865f0ca557e60b1e902f004012bc54e418878 (patch) | |
| tree | a775eabfa0c49df80696df7601a7e6a39ac998d6 /llvm/lib/Target | |
| parent | 0b7f98da6569bbc49aed2675fdda95f89198d356 (diff) | |
| download | bcm5719-llvm-dcf865f0ca557e60b1e902f004012bc54e418878.tar.gz bcm5719-llvm-dcf865f0ca557e60b1e902f004012bc54e418878.zip | |
[X86] Fix the pattern for merge masked vcvtps2pd.
r362199 fixed it for zero masking, but not zero masking. The load
folding in the peephole pass hid the bug. This patch turns off
the peephole pass on the relevant test to ensure coverage.
llvm-svn: 362440
Diffstat (limited to 'llvm/lib/Target')
| -rw-r--r-- | llvm/lib/Target/X86/X86InstrAVX512.td | 5 |
1 files changed, 1 insertions, 4 deletions
diff --git a/llvm/lib/Target/X86/X86InstrAVX512.td b/llvm/lib/Target/X86/X86InstrAVX512.td index 753f1b71b07..eebb6401db0 100644 --- a/llvm/lib/Target/X86/X86InstrAVX512.td +++ b/llvm/lib/Target/X86/X86InstrAVX512.td @@ -7629,10 +7629,7 @@ multiclass avx512_vcvt_fp<bits<8> opc, string OpcodeStr, X86VectorVTInfo _, (ins MaskRC:$mask, MemOp:$src), OpcodeStr#Alias, "$src", "$src", LdDAG, - (vselect MaskRC:$mask, - (_.VT (OpNode (_Src.VT - (_Src.LdFrag addr:$src)))), - _.RC:$src0), + (vselect MaskRC:$mask, LdDAG, _.RC:$src0), vselect, "$src0 = $dst">, EVEX, Sched<[sched.Folded]>; |
