[llvm] [AMDGPU] Remember to clear a DenseMap between runs of SIInsertWaitcnts (PR #110650)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-amdgpu

Author: Jay Foad (jayfoad)

<details>
<summary>Changes</summary>

This caused nondeterministic codegen in some cases.


---
Full diff: https://github.com/llvm/llvm-project/pull/110650.diff


1 Files Affected:

- (modified) llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp (+1) 


``````````diff
diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
index 80a7529002ac90..b2cfc3f3172564 100644
--- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
@@ -2598,6 +2598,7 @@ bool SIInsertWaitcnts::runOnMachineFunction(MachineFunction &MF) {
     Modified = true;
   }
   ReleaseVGPRInsts.clear();
+  PreheadersToFlush.clear();
   SLoadAddresses.clear();
 
   return Modified;

``````````

</details>


https://github.com/llvm/llvm-project/pull/110650