[llvm] [AMDGPU] Set success flag for weak cmpxchg in LowerBufferFatPointers (PR #203033)

[via llvm-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-amdgpu

Author: Arseniy Obolenskiy (aobolensk)

<details>
<summary>Changes</summary>

The weak cmpxchg lowering never set the success bool leaving it poison and breaking code that reads it

The buffer atomic cmpswap never fails by chance, so the strong path works for weak too

---
Full diff: https://github.com/llvm/llvm-project/pull/203033.diff


2 Files Affected:

- (modified) llvm/lib/Target/AMDGPU/AMDGPULowerBufferFatPointers.cpp (+2-4) 
- (modified) llvm/test/CodeGen/AMDGPU/lower-buffer-fat-pointers-memops.ll (+3-1) 


``````````diff
diff --git a/llvm/lib/Target/AMDGPU/AMDGPULowerBufferFatPointers.cpp b/llvm/lib/Target/AMDGPU/AMDGPULowerBufferFatPointers.cpp
index 679f7d5d8329f..300c2f8782ae1 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPULowerBufferFatPointers.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPULowerBufferFatPointers.cpp
@@ -1880,10 +1880,8 @@ PtrParts SplitPtrStructs::visitAtomicCmpXchgInst(AtomicCmpXchgInst &AI) {
 
   Value *Res = PoisonValue::get(AI.getType());
   Res = IRB.CreateInsertValue(Res, Call, 0);
-  if (!AI.isWeak()) {
-    Value *Succeeded = IRB.CreateICmpEQ(Call, AI.getCompareOperand());
-    Res = IRB.CreateInsertValue(Res, Succeeded, 1);
-  }
+  Value *Succeeded = IRB.CreateICmpEQ(Call, AI.getCompareOperand());
+  Res = IRB.CreateInsertValue(Res, Succeeded, 1);
   SplitUsers.insert(&AI);
   AI.replaceAllUsesWith(Res);
   return {nullptr, nullptr};
diff --git a/llvm/test/CodeGen/AMDGPU/lower-buffer-fat-pointers-memops.ll b/llvm/test/CodeGen/AMDGPU/lower-buffer-fat-pointers-memops.ll
index ca51a08920aba..756a6af4ccbcd 100644
--- a/llvm/test/CodeGen/AMDGPU/lower-buffer-fat-pointers-memops.ll
+++ b/llvm/test/CodeGen/AMDGPU/lower-buffer-fat-pointers-memops.ll
@@ -175,7 +175,9 @@ define {i32, i1} @cmpxchg_weak(ptr addrspace(8) %buf, i32 %wanted, i32 %new) {
 ; CHECK-NEXT:    [[RET:%.*]] = call i32 @llvm.amdgcn.raw.ptr.buffer.atomic.cmpswap.i32(i32 [[NEW]], i32 [[WANTED]], ptr addrspace(8) align 4 [[BUF]], i32 16, i32 0, i32 0)
 ; CHECK-NEXT:    fence syncscope("wavefront") acquire
 ; CHECK-NEXT:    [[TMP1:%.*]] = insertvalue { i32, i1 } poison, i32 [[RET]], 0
-; CHECK-NEXT:    ret { i32, i1 } [[TMP1]]
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[RET]], [[WANTED]]
+; CHECK-NEXT:    [[TMP3:%.*]] = insertvalue { i32, i1 } [[TMP1]], i1 [[TMP2]], 1
+; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
 ;
   %base = addrspacecast ptr addrspace(8) %buf to ptr addrspace(7)
   %p = getelementptr i32, ptr addrspace(7) %base, i32 4

``````````

</details>


https://github.com/llvm/llvm-project/pull/203033