Resolve sycl-web test failures caused by https://github.com/llvm/llvm-project/pull/140282 #19895
Conversation
clang/test/CodeGenSPIRV/Builtins/generic_cast_to_ptr_explicit.c
Outdated
Show resolved
Hide resolved
| @@ -3,7 +3,7 @@ | |||
| // RUN: %clang_cc1 -O1 -triple spirv32 -cl-std=CL3.0 -x cl %s -emit-llvm -o - | FileCheck %s | |||
|
|
|||
| #ifdef __SYCL_DEVICE_ONLY__ | |||
There was a problem hiding this comment.
Something seems off here. Why would __SYCL_DEVICE_ONLY__ be defined for the runs that don't include -fsycl-is-device? -x cl doesn't somehow imply SYCL, does it?
There was a problem hiding this comment.
The very first line runs clang with -fsycl-is-device.
There was a problem hiding this comment.
Yes, and this makes sense for that RUN line. My question is about the other two RUN lines that pass -x cl and don't pass -fsycl-is-device. The code presumes that __SYCL_DEVICE_ONLY__ is defined for those RUN lines as well and I don't understand why that would be the case. I also find it odd that an attribute named sycl_device is expected to be used in a non-SYCL OpenCL compilation. I would expect an attribute named opencl_device or similar instead.
clang/test/CodeGenSPIRV/Builtins/generic_cast_to_ptr_explicit.c
Outdated
Show resolved
Hide resolved
clang/lib/CodeGen/CodeGenModule.cpp
Outdated
| !D->hasAttr<SYCLDeviceAttr>() && | ||
| !(D->hasAttr<SYCLDeviceAttr>() || D->hasAttr<SYCLExternalAttr>()) && | ||
| !SemaSYCL::isTypeDecoratedWithDeclAttribute<SYCLDeviceGlobalAttr>( | ||
| D->getType())) | ||
| return getLangOpts().GPURelocatableDeviceCode |
There was a problem hiding this comment.
We'll need to do something similar to this in upstream as part of fixing emit of inline functions in device code.
clang/test/CodeGenSPIRV/Builtins/generic_cast_to_ptr_explicit.c
Outdated
Show resolved
Hide resolved
|
Thank you for reviewing this, @tahonermann @Fznamznon @premanandrao ! |
This patch fixes test failures caused by the upstream patch that adds sycl_external attribute and restricts emitting device code.