From 5915d33874fd0278f72f2c1a8cf2047bdf0ffd31 Mon Sep 17 00:00:00 2001
From: Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com>
Date: Tue, 6 Jul 2021 15:10:58 -0700
Subject: [PATCH] [AMDGPU] Do not run IR optimizations at -O0

Differential Revision: https://reviews.llvm.org/D105515
---
 llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp | 12 +++++++-----
 llvm/test/CodeGen/AMDGPU/llc-pipeline.ll       | 11 ++---------
 2 files changed, 9 insertions(+), 14 deletions(-)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
index a34449c..395672f 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
@@ -940,11 +940,11 @@ void AMDGPUPassConfig::addIRPasses() {
           AAR.addAAResult(WrapperPass->getResult());
         }));
     }
-  }
 
-  if (TM.getTargetTriple().getArch() == Triple::amdgcn) {
-    // TODO: May want to move later or split into an early and late one.
-    addPass(createAMDGPUCodeGenPreparePass());
+    if (TM.getTargetTriple().getArch() == Triple::amdgcn) {
+      // TODO: May want to move later or split into an early and late one.
+      addPass(createAMDGPUCodeGenPreparePass());
+    }
   }
 
   TargetPassConfig::addIRPasses();
@@ -1062,7 +1062,9 @@ ScheduleDAGInstrs *GCNPassConfig::createMachineScheduler(
 bool GCNPassConfig::addPreISel() {
   AMDGPUPassConfig::addPreISel();
 
-  addPass(createAMDGPULateCodeGenPreparePass());
+  if (TM->getOptLevel() > CodeGenOpt::None)
+    addPass(createAMDGPULateCodeGenPreparePass());
+
   if (EnableAtomicOptimizations) {
     addPass(createAMDGPUAtomicOptimizerPass());
   }
diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
index 1f1609c..b62d690 100644
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
@@ -44,11 +44,6 @@
 ; GCN-O0-NEXT:     Lower OpenCL enqueued blocks
 ; GCN-O0-NEXT:     Lower uses of LDS variables from non-kernel functions
 ; GCN-O0-NEXT:     FunctionPass Manager
-; GCN-O0-NEXT:       Dominator Tree Construction
-; GCN-O0-NEXT:       Post-Dominator Tree Construction
-; GCN-O0-NEXT:       Natural Loop Information
-; GCN-O0-NEXT:       Legacy Divergence Analysis
-; GCN-O0-NEXT:       AMDGPU IR optimizations
 ; GCN-O0-NEXT:       Lower Garbage Collection Instructions
 ; GCN-O0-NEXT:       Shadow Stack GC Lowering
 ; GCN-O0-NEXT:       Lower constant intrinsics
@@ -72,13 +67,11 @@
 ; GCN-O0-NEXT:       Function Alias Analysis Results
 ; GCN-O0-NEXT:       Flatten the CFG
 ; GCN-O0-NEXT:       Dominator Tree Construction
-; GCN-O0-NEXT:       Post-Dominator Tree Construction
-; GCN-O0-NEXT:       Natural Loop Information
-; GCN-O0-NEXT:       Legacy Divergence Analysis
-; GCN-O0-NEXT:       AMDGPU IR late optimizations
 ; GCN-O0-NEXT:       Basic Alias Analysis (stateless AA impl)
 ; GCN-O0-NEXT:       Function Alias Analysis Results
+; GCN-O0-NEXT:       Natural Loop Information
 ; GCN-O0-NEXT:       Code sinking
+; GCN-O0-NEXT:       Post-Dominator Tree Construction
 ; GCN-O0-NEXT:       Legacy Divergence Analysis
 ; GCN-O0-NEXT:       Unify divergent function exit nodes
 ; GCN-O0-NEXT:       Lazy Value Information Analysis
-- 
2.7.4