[llvm] [AMDGPU] Non convergent readfirstlane does not depend on EXEC. NFCI. (PR #179821)

[Stanislav Mekhanoshin via llvm-commits]

https://github.com/rampitec updated https://github.com/llvm/llvm-project/pull/179821

>From 4f81463d3590102d326fa05b64f84aa7508196f7 Mon Sep 17 00:00:00 2001
From: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
Date: Wed, 4 Feb 2026 16:14:24 -0800
Subject: [PATCH] [AMDGPU] Non convergent instruction does not depend on EXEC.
 NFCI.

---
 llvm/lib/Target/AMDGPU/SIInstrInfo.cpp | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 7ad086a869bdf..996c92a8b5078 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -180,6 +180,10 @@ bool SIInstrInfo::resultDependsOnExec(const MachineInstr &MI) const {
     return false;
   }
 
+  // If it is not convergent it does not depend on EXEC.
+  if (!MI.isConvergent())
+    return false;
+
   switch (MI.getOpcode()) {
   default:
     break;