AMDGPU/SI: Give up in promote aloca when a pointer may be captured.

[cfang]

Summary

Give up in promote alloca when a pointer may be captured.

[arsenm]

LGTM with the 2 additional tests

test/CodeGen/AMDGPU/addrspacecast-captured.ll

15–16

No unnamed values

19

Can you add another test which directly stores the addrspacecast, and another which passes the ptrtoint to an external call

[arsenm]

Typo aloca in commit message

[cfang]

Add two additional tests.

[arsenm]

LGTM

lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp

 #include "AMDGPUSubtarget.h"
 #include "Utils/AMDGPUBaseInfo.h"
 #include "llvm/ADT/APInt.h"
 #include "llvm/ADT/None.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/Triple.h"
 #include "llvm/ADT/Twine.h"
+#include "llvm/Analysis/CaptureTracking.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/Attributes.h"
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/Constant.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/Function.h"

       if (!binaryOpIsDerivedFromSameAlloca(BaseAlloca, Val, ICmp, 0, 1))
         return false;
 
       // May need to rewrite constant operands.
       WorkList.push_back(ICmp);
     }
 
     if (UseInst->getOpcode() == Instruction::AddrSpaceCast) {
+      // Give up if the pointer may be captured.
+      if (PointerMayBeCaptured(UseInst, true, true))
+        return false;
       // Don't collect the users of this.
       WorkList.push_back(User);
       continue;
     }
 
     if (!User->getType()->isPointerTy())
       continue;
 

test/CodeGen/AMDGPU/addrspacecast-captured.ll

+; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck %s
+
+; Nothing should be done if the addrspacecast is captured.
+
+declare void @consume_ptr2int(i32) #0
+
+; CHECK-LABEL: @addrspacecast_captured(
+; CHECK: %data = alloca i32, align 4
+; CHECK: %cast = addrspacecast i32* %data to i32 addrspace(4)*
+; CHECK: %ptr2int = ptrtoint i32 addrspace(4)* %cast to i32
+; CHECK: store i32 %ptr2int, i32 addrspace(1)* %out
+define void @addrspacecast_captured(i32 addrspace(1)* %out) #0 {
+entry:
+  %data = alloca i32, align 4
+  %cast = addrspacecast i32* %data to i32 addrspace(4)*
+  %ptr2int = ptrtoint i32 addrspace(4)* %cast to i32
+  store i32 %ptr2int, i32 addrspace(1)* %out
+  ret void
+}
+
+; CHECK-LABEL: @addrspacecast_captured_store(
+; CHECK: %data = alloca i32, align 4
+; CHECK: %cast = addrspacecast i32* %data to i32 addrspace(4)*
+; CHECK: store i32 addrspace(4)* %cast, i32 addrspace(4)* addrspace(1)* %out
+define void @addrspacecast_captured_store(i32 addrspace(4)* addrspace(1)* %out) #0 {
+entry:
+  %data = alloca i32, align 4
+  %cast = addrspacecast i32* %data to i32 addrspace(4)*
+  store i32 addrspace(4)* %cast, i32 addrspace(4)* addrspace(1)* %out
+  ret void
+}
+
+; CHECK-LABEL: @addrspacecast_captured_call(
+; CHECK: %data = alloca i32, align 4
+; CHECK: %cast = addrspacecast i32* %data to i32 addrspace(4)*
+; CHECK: %ptr2int = ptrtoint i32 addrspace(4)* %cast to i32
+; CHECK: call void @consume_ptr2int(i32 %ptr2int)
+define void @addrspacecast_captured_call() #0 {
+entry:
+  %data = alloca i32, align 4
+  %cast = addrspacecast i32* %data to i32 addrspace(4)*
+  %ptr2int = ptrtoint i32 addrspace(4)* %cast to i32
+  call void @consume_ptr2int(i32 %ptr2int)
+  ret void
+}
+
+attributes #0 = { nounwind }