[llvm] X86: Use hasOneNonDBGUse instead of kill flag for cascaded select (PR #228939)

[via llvm-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-backend-x86

Author: Matt Arsenault (arsenm)

<details>
<summary>Changes</summary>

The cascaded CMOV lowering requires the first select's result to only be
used by the second select. Avoid relying on the kill flag in preparation
for not setting it.

Co-authored-by: Claude Opus 5.5 <noreply@<!-- -->anthropic.com>

---
Full diff: https://github.com/llvm/llvm-project/pull/228939.diff


1 Files Affected:

- (modified) llvm/lib/Target/X86/X86ISelLowering.cpp (+2-1) 


``````````diff
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index c29b09ae846bb..b8edb2dc6f13d 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -37704,7 +37704,8 @@ X86TargetLowering::EmitLoweredSelect(MachineInstr &MI,
       NextMIIt->getOpcode() == MI.getOpcode() &&
       NextMIIt->getOperand(2).getReg() == MI.getOperand(2).getReg() &&
       NextMIIt->getOperand(1).getReg() == MI.getOperand(0).getReg() &&
-      NextMIIt->getOperand(1).isKill()) {
+      ThisMBB->getParent()->getRegInfo().hasOneNonDBGUse(
+          MI.getOperand(0).getReg())) {
     return EmitLoweredCascadedSelect(MI, *NextMIIt, ThisMBB);
   }
 

``````````

</details>


https://github.com/llvm/llvm-project/pull/228939