[PATCH] D81710: [amdgpu] Skip OR combining on 64-bit integer before legalizing ops.

[Michael Liao via Phabricator via llvm-commits]

hliao created this revision.
hliao added reviewers: arsenm, rampitec.
Herald added subscribers: llvm-commits, kerbowa, hiraditya, t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, jvesely, kzhuravl.
Herald added a project: LLVM.

Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D81710

Files:
  llvm/lib/Target/AMDGPU/SIISelLowering.cpp
  llvm/test/CodeGen/AMDGPU/fshr.ll


Index: llvm/test/CodeGen/AMDGPU/fshr.ll
===================================================================
--- llvm/test/CodeGen/AMDGPU/fshr.ll
+++ llvm/test/CodeGen/AMDGPU/fshr.ll
@@ -763,7 +763,7 @@
 ; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
 ; SI-NEXT:    v_cndmask_b32_e32 v1, v2, v5, vcc
 ; SI-NEXT:    v_and_b32_e32 v2, v9, v1
-; SI-NEXT:    v_alignbit_b32 v1, v2, v0, 16
+; SI-NEXT:    v_alignbit_b32 v1, v1, v0, 16
 ; SI-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; VI-LABEL: v_fshr_v3i16:
Index: llvm/lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -8906,7 +8906,7 @@
     }
   }
 
-  if (VT != MVT::i64)
+  if (VT != MVT::i64 || DCI.isBeforeLegalizeOps())
     return SDValue();
 
   // TODO: This could be a generic combine with a predicate for extracting the


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D81710.270291.patch
Type: text/x-patch
Size: 915 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200612/b33816a4/attachment.bin>