[llvm] r262457 - AMDGPU: Fix bug 26659.

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Tue Mar 1 20:12:39 PST 2016


Author: arsenm
Date: Tue Mar  1 22:12:39 2016
New Revision: 262457

URL: http://llvm.org/viewvc/llvm-project?rev=262457&view=rev
Log:
AMDGPU: Fix bug 26659.

Fix checking the same instruction twice instead of the
second branch that uses vccz. I don't think this matters
currently because s_branch_vccnz is always used currently.

Modified:
    llvm/trunk/lib/Target/AMDGPU/SIInsertWaits.cpp

Modified: llvm/trunk/lib/Target/AMDGPU/SIInsertWaits.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/AMDGPU/SIInsertWaits.cpp?rev=262457&r1=262456&r2=262457&view=diff
==============================================================================
--- llvm/trunk/lib/Target/AMDGPU/SIInsertWaits.cpp (original)
+++ llvm/trunk/lib/Target/AMDGPU/SIInsertWaits.cpp Tue Mar  1 22:12:39 2016
@@ -164,7 +164,7 @@ const Counters SIInsertWaits::WaitCounts
 const Counters SIInsertWaits::ZeroCounts = { { 0, 0, 0 } };
 
 static bool readsVCCZ(unsigned Opcode) {
-  return Opcode == AMDGPU::S_CBRANCH_VCCNZ || Opcode == AMDGPU::S_CBRANCH_VCCNZ;
+  return Opcode == AMDGPU::S_CBRANCH_VCCNZ || Opcode == AMDGPU::S_CBRANCH_VCCZ;
 }
 
 bool SIInsertWaits::hasOutstandingLGKM() const {




More information about the llvm-commits mailing list