Index: lib/Target/AMDGPU/AMDGPUAlwaysInlinePass.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUAlwaysInlinePass.cpp
+++ lib/Target/AMDGPU/AMDGPUAlwaysInlinePass.cpp
@@ -35,8 +35,20 @@
 char AMDGPUAlwaysInline::ID = 0;
 
 bool AMDGPUAlwaysInline::runOnModule(Module &M) {
+  std::vector<GlobalAlias*> AliasesToRemove;
   std::vector<Function *> FuncsToClone;
 
+  for (GlobalAlias &A : M.aliases()) {
+    if (Function* F = dyn_cast<Function>(A.getAliasee())) {
+      A.replaceAllUsesWith(F);
+      AliasesToRemove.push_back(&A);
+    }
+  }
+
+  for (GlobalAlias* A : AliasesToRemove) {
+    A->eraseFromParent();
+  }
+
   for (Function &F : M) {
     if (!F.hasLocalLinkage() && !F.isDeclaration() && !F.use_empty() &&
         !F.hasFnAttribute(Attribute::NoInline))
Index: test/CodeGen/AMDGPU/inline-calls.ll
===================================================================
--- test/CodeGen/AMDGPU/inline-calls.ll
+++ test/CodeGen/AMDGPU/inline-calls.ll
@@ -10,6 +10,7 @@
 }
 
 ; CHECK: {{^}}kernel:
+; CHECK-NOT: call
 define void @kernel(i32 addrspace(1)* %out) {
 entry:
   %tmp0 = call i32 @func(i32 1)
@@ -18,8 +19,32 @@
 }
 
 ; CHECK: {{^}}kernel2:
+; CHECK-NOT: call
 define void @kernel2(i32 addrspace(1)* %out) {
 entry:
   call void @kernel(i32 addrspace(1)* %out)
   ret void
 }
+
+; CHECK-NOT: func_alias
+@func_alias = alias i32 (i32), i32 (i32)* @func
+
+; CHECK: {{^}}kernel3:
+; CHECK-NOT: call
+define void @kernel3(i32 addrspace(1)* %out) {
+entry:
+  %tmp0 = call i32 @func_alias(i32 1)
+  store i32 %tmp0, i32 addrspace(1)* %out
+  ret void
+}
+
+; CHECK-NOT: kernel_alias
+@kernel_alias = alias void (i32 addrspace(1)*), void (i32 addrspace(1)*)* @kernel
+
+; CHECK: {{^}}kernel4:
+; CHECK-NOT: call
+define void @kernel4(i32 addrspace(1)* %out) {
+entry:
+  call void @kernel_alias(i32 addrspace(1)* %out)
+  ret void
+}