[PATCH] D142033: [OpenCL] Always add nounwind attribute for OpenCL

[Sven van Haastregt via Phabricator via cfe-commits]

svenvh created this revision.
svenvh added reviewers: Anastasia, yaxunl, bader.
Herald added subscribers: kosarev, Naghasan, ldrumm, kerbowa, jvesely.
Herald added a project: All.
svenvh requested review of this revision.
Herald added a project: clang.
Herald added a subscriber: cfe-commits.

Neither OpenCL nor C++ for OpenCL support exceptions, so add the
`nounwind` attribute unconditionally for those languages.

Unblocks D138958 <https://reviews.llvm.org/D138958>.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D142033

Files:
  clang/lib/CodeGen/CGCall.cpp
  clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
  clang/test/CodeGenOpenCL/convergent.cl


Index: clang/test/CodeGenOpenCL/convergent.cl
===================================================================
--- clang/test/CodeGenOpenCL/convergent.cl
+++ clang/test/CodeGenOpenCL/convergent.cl
@@ -139,4 +139,5 @@
 // CHECK: attributes #3 = { {{[^}]*}}convergent noduplicate{{[^}]*}} }
 // CHECK: attributes #4 = { {{[^}]*}}convergent{{[^}]*}} }
 // CHECK: attributes #5 = { {{[^}]*}}convergent{{[^}]*}} }
-// CHECK: attributes #6 = { {{[^}]*}}convergent noduplicate{{[^}]*}} }
+// CHECK: attributes #6 = { {{[^}]*}}nounwind{{[^}]*}} }
+// CHECK: attributes #7 = { {{[^}]*}}convergent noduplicate nounwind{{[^}]*}} }
Index: clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
===================================================================
--- clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
+++ clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
@@ -298,7 +298,7 @@
 // CHECK: attributes #2 = { nocallback nofree nounwind willreturn 
memory(argmem: readwrite) }
 // CHECK: attributes #3 = { convergent noinline nounwind optnone 
"no-trapping-math"="true" "stack-protector-buffer-size"="8" 
"target-cpu"="gfx900" 
"target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64"
 }
 // CHECK: attributes #4 = { nounwind "enqueued-block" }
-// CHECK: attributes #5 = { convergent }
+// CHECK: attributes #5 = { convergent nounwind }
 //.
 // CHECK: !0 = !{i32 1, !"amdgpu_code_object_version", i32 400}
 // CHECK: !1 = !{i32 1, !"wchar_size", i32 4}
Index: clang/lib/CodeGen/CGCall.cpp
===================================================================
--- clang/lib/CodeGen/CGCall.cpp
+++ clang/lib/CodeGen/CGCall.cpp
@@ -1969,11 +1969,11 @@
     FuncAttrs.addAttribute(llvm::Attribute::Convergent);
   }
 
-  // TODO: NoUnwind attribute should be added for other GPU modes OpenCL, HIP,
+  // TODO: NoUnwind attribute should be added for other GPU modes HIP,
   // SYCL, OpenMP offload. AFAIK, none of them support exceptions in device
   // code.
-  if (getLangOpts().CUDA && getLangOpts().CUDAIsDevice) {
-    // Exceptions aren't supported in CUDA device code.
+  if ((getLangOpts().CUDA && getLangOpts().CUDAIsDevice) ||
+      getLangOpts().OpenCL) {
     FuncAttrs.addAttribute(llvm::Attribute::NoUnwind);
   }
 



Index: clang/test/CodeGenOpenCL/convergent.cl
===================================================================
--- clang/test/CodeGenOpenCL/convergent.cl
+++ clang/test/CodeGenOpenCL/convergent.cl
@@ -139,4 +139,5 @@
 // CHECK: attributes #3 = { {{[^}]*}}convergent noduplicate{{[^}]*}} }
 // CHECK: attributes #4 = { {{[^}]*}}convergent{{[^}]*}} }
 // CHECK: attributes #5 = { {{[^}]*}}convergent{{[^}]*}} }
-// CHECK: attributes #6 = { {{[^}]*}}convergent noduplicate{{[^}]*}} }
+// CHECK: attributes #6 = { {{[^}]*}}nounwind{{[^}]*}} }
+// CHECK: attributes #7 = { {{[^}]*}}convergent noduplicate nounwind{{[^}]*}} }
Index: clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
===================================================================
--- clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
+++ clang/test/CodeGenOpenCL/amdgpu-enqueue-kernel.cl
@@ -298,7 +298,7 @@
 // CHECK: attributes #2 = { nocallback nofree nounwind willreturn memory(argmem: readwrite) }
 // CHECK: attributes #3 = { convergent noinline nounwind optnone "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx900" "target-features"="+16-bit-insts,+ci-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
 // CHECK: attributes #4 = { nounwind "enqueued-block" }
-// CHECK: attributes #5 = { convergent }
+// CHECK: attributes #5 = { convergent nounwind }
 //.
 // CHECK: !0 = !{i32 1, !"amdgpu_code_object_version", i32 400}
 // CHECK: !1 = !{i32 1, !"wchar_size", i32 4}
Index: clang/lib/CodeGen/CGCall.cpp
===================================================================
--- clang/lib/CodeGen/CGCall.cpp
+++ clang/lib/CodeGen/CGCall.cpp
@@ -1969,11 +1969,11 @@
     FuncAttrs.addAttribute(llvm::Attribute::Convergent);
   }
 
-  // TODO: NoUnwind attribute should be added for other GPU modes OpenCL, HIP,
+  // TODO: NoUnwind attribute should be added for other GPU modes HIP,
   // SYCL, OpenMP offload. AFAIK, none of them support exceptions in device
   // code.
-  if (getLangOpts().CUDA && getLangOpts().CUDAIsDevice) {
-    // Exceptions aren't supported in CUDA device code.
+  if ((getLangOpts().CUDA && getLangOpts().CUDAIsDevice) ||
+      getLangOpts().OpenCL) {
     FuncAttrs.addAttribute(llvm::Attribute::NoUnwind);
   }
 

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits