[HLSL] Fix call convention mismatch for ctor/dtor #118651
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Before this patch, there was a calling-convention mismatch between the constructors and the actual call emitted for the entrypoint wrapper. Such mismatch causes the InstCombine pass to replace this call with an `unreachable`, breaking the whole function. Signed-off-by: Nathan Gauër <[email protected]>
llvmbot
added
clang
|
@llvm/pr-subscribers-clang-codegen @llvm/pr-subscribers-hlsl Author: Nathan Gauër (Keenuts) ChangesBefore this patch, there was a calling-convention mismatch between the constructors and the actual call emitted for the entrypoint wrapper. Such mismatch causes the InstCombine pass to replace this call with an Full diff: https://github.com/llvm/llvm-project/pull/118651.diff 2 Files Affected:
diff --git a/clang/lib/CodeGen/CGHLSLRuntime.cpp b/clang/lib/CodeGen/CGHLSLRuntime.cpp
index 2c293523fca8ca..9b6db8f0c5d660 100644
--- a/clang/lib/CodeGen/CGHLSLRuntime.cpp
+++ b/clang/lib/CodeGen/CGHLSLRuntime.cpp
@@ -507,13 +507,17 @@ void CGHLSLRuntime::generateGlobalCtorDtorCalls() {
IP = Token->getNextNode();
}
IRBuilder<> B(IP);
- for (auto *Fn : CtorFns)
- B.CreateCall(FunctionCallee(Fn), {}, OB);
+ for (auto *Fn : CtorFns) {
+ auto CI = B.CreateCall(FunctionCallee(Fn), {}, OB);
+ CI->setCallingConv(Fn->getCallingConv());
+ }
// Insert global dtors before the terminator of the last instruction
B.SetInsertPoint(F.back().getTerminator());
- for (auto *Fn : DtorFns)
- B.CreateCall(FunctionCallee(Fn), {}, OB);
+ for (auto *Fn : DtorFns) {
+ auto CI = B.CreateCall(FunctionCallee(Fn), {}, OB);
+ CI->setCallingConv(Fn->getCallingConv());
+ }
}
// No need to keep global ctors/dtors for non-lib profile after call to
diff --git a/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor-opt.hlsl b/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor-opt.hlsl
new file mode 100644
index 00000000000000..d834a22917c1f6
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/RWBuffer-constructor-opt.hlsl
@@ -0,0 +1,21 @@
+// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-DXIL
+// RUN: %clang_cc1 -triple spirv-vulkan-compute -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-SPIRV
+
+// CHECK-SPIRV: %"class.hlsl::RWBuffer" = type { target("spirv.Image", float, 5, 2, 0, 0, 2, 0) }
+// CHECK-DXIL: %"class.hlsl::RWBuffer" = type { target("dx.TypedBuffer", float, 1, 0, 0) }
+RWBuffer<float> Buf : register(u5, space3);
+
+[shader("compute")]
+[numthreads(1, 1, 1)]
+void main() {
+// CHECK: define void @main()
+// CHECK-NEXT: entry:
+
+// CHECK-SPIRV-NEXT: %Buf_h.i = tail call target("spirv.Image", float, 5, 2, 0, 0, 2, 0) @llvm.spv.handle.fromBinding.tspirv.Image_f32_5_2_0_0_2_0t(i32 3, i32 5, i32 1, i32 0, i1 false)
+// CHECK-SPIRV-NEXT: store target("spirv.Image", float, 5, 2, 0, 0, 2, 0) %Buf_h.i, ptr @Buf, align 8
+
+// CHECK-DXIL-NEXT: %Buf_h.i = tail call target("dx.TypedBuffer", float, 1, 0, 0) @llvm.dx.handle.fromBinding.tdx.TypedBuffer_f32_1_0_0t(i32 3, i32 5, i32 1, i32 0, i1 false)
+// CHECK-DXIL-NEXT: store target("dx.TypedBuffer", float, 1, 0, 0) %Buf_h.i, ptr @Buf, align 4
+
+// CHECK-NEXT: ret void
+}
|
llvm-beanz
reviewed
Dec 5, 2024
| @@ -0,0 +1,21 @@ | |||
| // RUN: %clang_cc1 -triple dxil-pc-shadermodel6.3-library -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-DXIL | |||
| // RUN: %clang_cc1 -triple spirv-vulkan-compute -x hlsl -emit-llvm -O3 -o - %s | FileCheck %s --check-prefixes=CHECK,CHECK-SPIRV | |||
There was a problem hiding this comment.
Can you add a run line that verifies the IR coming out of Clang before optimizations?
There was a problem hiding this comment.
There is the RWBuffer-constructor.hlsl test that tests almost the same code, but without any optimization passes. Or do you mean another kind of test?
bogner
approved these changes
Dec 16, 2024
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/134/builds/10507 Here is the relevant piece of the build log for the reference |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Before this patch, there was a calling-convention mismatch between the constructors and the actual call emitted for the entrypoint wrapper.
Such mismatch causes the InstCombine pass to replace this call with an
unreachable, breaking the whole function.