1 files changed, 14 insertions, 3 deletions
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
index c865082..c1f1703 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
@@ -104,7 +104,9 @@
 #include "llvm/Transforms/Scalar/FlattenCFG.h"
 #include "llvm/Transforms/Scalar/GVN.h"
 #include "llvm/Transforms/Scalar/InferAddressSpaces.h"
+#include "llvm/Transforms/Scalar/LICM.h"
 #include "llvm/Transforms/Scalar/LoopDataPrefetch.h"
+#include "llvm/Transforms/Scalar/LoopPassManager.h"
 #include "llvm/Transforms/Scalar/NaryReassociate.h"
 #include "llvm/Transforms/Scalar/SeparateConstOffsetFromGEP.h"
 #include "llvm/Transforms/Scalar/Sink.h"
@@ -836,8 +838,10 @@ void AMDGPUTargetMachine::registerPassBuilderCallbacks(PassBuilder &PB) {
           // When we are not using -fgpu-rdc, we can run accelerator code
           // selection relatively early, but still after linking to prevent
           // eager removal of potentially reachable symbols.
-          if (EnableHipStdPar)
+          if (EnableHipStdPar) {
+            PM.addPass(HipStdParMathFixupPass());
             PM.addPass(HipStdParAcceleratorCodeSelectionPass());
+          }
           PM.addPass(AMDGPUPrintfRuntimeBindingPass());
         }
 
@@ -916,8 +920,10 @@ void AMDGPUTargetMachine::registerPassBuilderCallbacks(PassBuilder &PB) {
         // selection after linking to prevent, otherwise we end up removing
         // potentially reachable symbols that were exported as external in other
         // modules.
-        if (EnableHipStdPar)
+        if (EnableHipStdPar) {
+          PM.addPass(HipStdParMathFixupPass());
           PM.addPass(HipStdParAcceleratorCodeSelectionPass());
+        }
         // We want to support the -lto-partitions=N option as "best effort".
         // For that, we need to lower LDS earlier in the pipeline before the
         // module is partitioned for codegen.
@@ -2062,7 +2068,12 @@ void AMDGPUCodeGenPassBuilder::addIRPasses(AddIRPass &addPass) const {
     // TODO: May want to move later or split into an early and late one.
     addPass(AMDGPUCodeGenPreparePass(TM));
 
-    // TODO: LICM
+    // Try to hoist loop invariant parts of divisions AMDGPUCodeGenPrepare may
+    // have expanded.
+    if (TM.getOptLevel() > CodeGenOptLevel::Less) {
+      addPass(createFunctionToLoopPassAdaptor(LICMPass(LICMOptions()),
+                                              /*UseMemorySSA=*/true));
+    }
   }
 
   Base::addIRPasses(addPass);