[PATCH] D151153: [AMDGPU] Add attribute to AMDGPU ctor / dtor to indicate single threadedness

[Joseph Huber via Phabricator via llvm-commits]

jhuber6 created this revision.
jhuber6 added reviewers: JonChesterfield, yaxunl, arsenm, rampitec.
Herald added subscribers: foad, kerbowa, hiraditya, tpr, dstuttard, jvesely, kzhuravl.
Herald added a project: All.
jhuber6 requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

We only expect these ctor / dtor functions to be called with a single
thread. Add the appropriate attributes to indicate this to the backend.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D151153

Files:
  llvm/lib/Target/AMDGPU/AMDGPUCtorDtorLowering.cpp
  llvm/test/CodeGen/AMDGPU/lower-ctor-dtor-constexpr-alias.ll
  llvm/test/CodeGen/AMDGPU/lower-ctor-dtor.ll
  llvm/test/CodeGen/AMDGPU/lower-multiple-ctor-dtor.ll


Index: llvm/test/CodeGen/AMDGPU/lower-multiple-ctor-dtor.ll
===================================================================
--- llvm/test/CodeGen/AMDGPU/lower-multiple-ctor-dtor.ll
+++ llvm/test/CodeGen/AMDGPU/lower-multiple-ctor-dtor.ll
@@ -57,5 +57,5 @@
   ret void
 }
 
-; CHECK: attributes #0 = { "device-init" }
-; CHECK: attributes #1 = { "device-fini" }
+; CHECK: attributes #0 = { "amdgpu-flat-work-group-size"="1,1" "device-init" }
+; CHECK: attributes #1 = { "amdgpu-flat-work-group-size"="1,1" "device-fini" }
Index: llvm/test/CodeGen/AMDGPU/lower-ctor-dtor.ll
===================================================================
--- llvm/test/CodeGen/AMDGPU/lower-ctor-dtor.ll
+++ llvm/test/CodeGen/AMDGPU/lower-ctor-dtor.ll
@@ -58,5 +58,5 @@
   ret void
 }
 
-; CHECK: attributes #0 = { "device-init" }
-; CHECK: attributes #1 = { "device-fini" }
+; CHECK: attributes #0 = { "amdgpu-flat-work-group-size"="1,1" "device-init" }
+; CHECK: attributes #1 = { "amdgpu-flat-work-group-size"="1,1" "device-fini" }
Index: llvm/test/CodeGen/AMDGPU/lower-ctor-dtor-constexpr-alias.ll
===================================================================
--- llvm/test/CodeGen/AMDGPU/lower-ctor-dtor-constexpr-alias.ll
+++ llvm/test/CodeGen/AMDGPU/lower-ctor-dtor-constexpr-alias.ll
@@ -64,5 +64,5 @@
 ; CHECK:       while.end:
 ; CHECK-NEXT:    ret void
 
-; CHECK: attributes #[[ATTR0:[0-9]+]] = { "device-init" }
-; CHECK: attributes #[[ATTR1:[0-9]+]] = { "device-fini" }
+; CHECK: attributes #[[ATTR0:[0-9]+]] = { "amdgpu-flat-work-group-size"="1,1" "device-init" }
+; CHECK: attributes #[[ATTR1:[0-9]+]] = { "amdgpu-flat-work-group-size"="1,1" "device-fini" }
Index: llvm/lib/Target/AMDGPU/AMDGPUCtorDtorLowering.cpp
===================================================================
--- llvm/lib/Target/AMDGPU/AMDGPUCtorDtorLowering.cpp
+++ llvm/lib/Target/AMDGPU/AMDGPUCtorDtorLowering.cpp
@@ -38,6 +38,7 @@
       FunctionType::get(Type::getVoidTy(M.getContext()), false),
       GlobalValue::WeakODRLinkage, 0, InitOrFiniKernelName, &M);
   InitOrFiniKernel->setCallingConv(CallingConv::AMDGPU_KERNEL);
+  InitOrFiniKernel->addFnAttr("amdgpu-flat-work-group-size", "1,1");
   if (IsCtor)
     InitOrFiniKernel->addFnAttr("device-init");
   else
@@ -58,6 +59,7 @@
 // void call_init_array_callbacks() {
 //   for (auto start = __init_array_start; start != __init_array_end; ++start)
 //     reinterpret_cast<InitCallback *>(*start)();
+// }
 static void createInitOrFiniCalls(Function &F, bool IsCtor) {
   Module &M = *F.getParent();
   LLVMContext &C = M.getContext();


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D151153.524473.patch
Type: text/x-patch
Size: 2583 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230522/5e8b5e51/attachment.bin>