diff --git a/llvm/lib/Target/AMDGPU/AMDGPU.h b/llvm/lib/Target/AMDGPU/AMDGPU.h
--- a/llvm/lib/Target/AMDGPU/AMDGPU.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPU.h
@@ -123,9 +123,27 @@
 void initializeAMDGPUPropagateAttributesEarlyPass(PassRegistry &);
 extern char &AMDGPUPropagateAttributesEarlyID;
 
+struct AMDGPUPropagateAttributesEarlyPass
+    : PassInfoMixin<AMDGPUPropagateAttributesEarlyPass> {
+  AMDGPUPropagateAttributesEarlyPass(TargetMachine &TM) : TM(TM) {}
+  PreservedAnalyses run(Function &F, FunctionAnalysisManager &AM);
+
+private:
+  TargetMachine &TM;
+};
+
 void initializeAMDGPUPropagateAttributesLatePass(PassRegistry &);
 extern char &AMDGPUPropagateAttributesLateID;
 
+struct AMDGPUPropagateAttributesLatePass
+    : PassInfoMixin<AMDGPUPropagateAttributesLatePass> {
+  AMDGPUPropagateAttributesLatePass(TargetMachine &TM) : TM(TM) {}
+  PreservedAnalyses run(Module &M, ModuleAnalysisManager &AM);
+
+private:
+  TargetMachine &TM;
+};
+
 void initializeAMDGPURewriteOutArgumentsPass(PassRegistry &);
 extern char &AMDGPURewriteOutArgumentsID;
 
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUPropagateAttributes.cpp b/llvm/lib/Target/AMDGPU/AMDGPUPropagateAttributes.cpp
--- a/llvm/lib/Target/AMDGPU/AMDGPUPropagateAttributes.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUPropagateAttributes.cpp
@@ -35,6 +35,7 @@
 #include "llvm/CodeGen/TargetPassConfig.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/Module.h"
+#include "llvm/IR/PassManager.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Transforms/Utils/Cloning.h"
 #include <string>
@@ -409,3 +410,21 @@
 *llvm::createAMDGPUPropagateAttributesLatePass(const TargetMachine *TM) {
   return new AMDGPUPropagateAttributesLate(TM);
 }
+
+PreservedAnalyses
+AMDGPUPropagateAttributesEarlyPass::run(Function &F,
+                                        FunctionAnalysisManager &AM) {
+  if (!AMDGPU::isEntryFunctionCC(F.getCallingConv()))
+    return PreservedAnalyses::all();
+
+  return AMDGPUPropagateAttributes(&TM, false).process(F)
+             ? PreservedAnalyses::none()
+             : PreservedAnalyses::all();
+}
+
+PreservedAnalyses
+AMDGPUPropagateAttributesLatePass::run(Module &M, ModuleAnalysisManager &AM) {
+  return AMDGPUPropagateAttributes(&TM, true).process(M)
+             ? PreservedAnalyses::none()
+             : PreservedAnalyses::all();
+}
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
@@ -491,6 +491,15 @@
 
 void AMDGPUTargetMachine::registerPassBuilderCallbacks(PassBuilder &PB,
                                                        bool DebugPassManager) {
+  PB.registerPipelineParsingCallback(
+      [this](StringRef PassName, ModulePassManager &PM,
+             ArrayRef<PassBuilder::PipelineElement>) {
+        if (PassName == "amdgpu-propagate-attributes-late") {
+          PM.addPass(AMDGPUPropagateAttributesLatePass(*this));
+          return true;
+        }
+        return false;
+      });
   PB.registerPipelineParsingCallback(
       [this](StringRef PassName, FunctionPassManager &PM,
              ArrayRef<PassBuilder::PipelineElement>) {
@@ -514,13 +523,19 @@
           PM.addPass(AMDGPULowerKernelAttributesPass());
           return true;
         }
+        if (PassName == "amdgpu-propagate-attributes-early") {
+          PM.addPass(AMDGPUPropagateAttributesEarlyPass(*this));
+          return true;
+        }
+
         return false;
       });
 
-  PB.registerPipelineStartEPCallback([DebugPassManager](
+  PB.registerPipelineStartEPCallback([this, DebugPassManager](
                                          ModulePassManager &PM,
                                          PassBuilder::OptimizationLevel Level) {
     FunctionPassManager FPM(DebugPassManager);
+    FPM.addPass(AMDGPUPropagateAttributesEarlyPass(*this));
     FPM.addPass(AMDGPUUseNativeCallsPass());
     if (EnableLibCallSimplify && Level != PassBuilder::OptimizationLevel::O0)
       FPM.addPass(AMDGPUSimplifyLibCallsPass());
@@ -528,12 +543,15 @@
   });
 
   PB.registerPipelineEarlySimplificationEPCallback(
-      [](ModulePassManager &PM, PassBuilder::OptimizationLevel Level) {
+      [this](ModulePassManager &PM, PassBuilder::OptimizationLevel Level) {
         if (Level == PassBuilder::OptimizationLevel::O0)
           return;
 
         if (InternalizeSymbols) {
           PM.addPass(InternalizePass(mustPreserveGV));
+        }
+        PM.addPass(AMDGPUPropagateAttributesLatePass(*this));
+        if (InternalizeSymbols) {
           PM.addPass(GlobalDCEPass());
         }
       });
diff --git a/llvm/test/CodeGen/AMDGPU/propagate-attributes-clone.ll b/llvm/test/CodeGen/AMDGPU/propagate-attributes-clone.ll
--- a/llvm/test/CodeGen/AMDGPU/propagate-attributes-clone.ll
+++ b/llvm/test/CodeGen/AMDGPU/propagate-attributes-clone.ll
@@ -1,5 +1,7 @@
 ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -O1 < %s | FileCheck -check-prefixes=OPT,OPT-EXT %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes='default<O1>' < %s | FileCheck -check-prefixes=OPT,OPT-EXT %s
 ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -O1 --amdgpu-internalize-symbols < %s | FileCheck -check-prefixes=OPT,OPT-INT %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes='default<O1>' --amdgpu-internalize-symbols < %s | FileCheck -check-prefixes=OPT,OPT-INT %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefix=LLC %s
 
 ; OPT: declare void @foo4() local_unnamed_addr #0
diff --git a/llvm/test/CodeGen/AMDGPU/propagate-attributes-flat-work-group-size.ll b/llvm/test/CodeGen/AMDGPU/propagate-attributes-flat-work-group-size.ll
--- a/llvm/test/CodeGen/AMDGPU/propagate-attributes-flat-work-group-size.ll
+++ b/llvm/test/CodeGen/AMDGPU/propagate-attributes-flat-work-group-size.ll
@@ -1,4 +1,5 @@
 ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -amdgpu-propagate-attributes-late %s | FileCheck %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-propagate-attributes-late %s | FileCheck %s
 
 ; CHECK: define internal void @max_flat_1_1024() #0 {
 define internal void @max_flat_1_1024() #0 {
diff --git a/llvm/test/CodeGen/AMDGPU/propagate-attributes-single-set.ll b/llvm/test/CodeGen/AMDGPU/propagate-attributes-single-set.ll
--- a/llvm/test/CodeGen/AMDGPU/propagate-attributes-single-set.ll
+++ b/llvm/test/CodeGen/AMDGPU/propagate-attributes-single-set.ll
@@ -1,4 +1,5 @@
 ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -O1 < %s | FileCheck -check-prefix=OPT %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes='default<O1>' < %s | FileCheck -check-prefix=OPT %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefix=LLC %s
 
 ; OPT: declare void @foo4() local_unnamed_addr #0
diff --git a/llvm/tools/opt/opt.cpp b/llvm/tools/opt/opt.cpp
--- a/llvm/tools/opt/opt.cpp
+++ b/llvm/tools/opt/opt.cpp
@@ -463,8 +463,13 @@
 // it exists.
 static bool shouldPinPassToLegacyPM(StringRef Pass) {
   std::vector<StringRef> PassNameExactToIgnore = {
-      "amdgpu-simplifylib", "amdgpu-usenative", "amdgpu-promote-alloca",
-      "amdgpu-promote-alloca-to-vector", "amdgpu-lower-kernel-attributes"};
+      "amdgpu-simplifylib",
+      "amdgpu-usenative",
+      "amdgpu-promote-alloca",
+      "amdgpu-promote-alloca-to-vector",
+      "amdgpu-lower-kernel-attributes",
+      "amdgpu-propagate-attributes-early",
+      "amdgpu-propagate-attributes-late"};
   for (const auto &P : PassNameExactToIgnore)
     if (Pass == P)
       return false;