================
@@ -0,0 +1,27 @@
+// Verify the ELF packaging of OpenMP SPIR-V device images.
+// REQUIRES: system-linux
+// REQUIRES: spirv-tools
+// REQUIRES: spirv-registered-target
+// REQUIRES: x86-registered-target
+
+// RUN: %clang -cc1 %s -triple x86_64-unknown-linux-gnu -emit-llvm-bc -o %t.bc
+// RUN: %clang -cc1 %s -triple spirv64-intel -fopenmp-is-target-device -o 
%t.device.o
+// RUN: llvm-offload-binary -o %t.bundle 
--image=file=%t.device.o,triple=spirv64-intel,arch=generic,kind=openmp
+// RUN: %clang -cc1 -triple x86_64-unknown-linux-gnu -emit-obj 
-fembed-offload-object=%t.bundle -x ir %t.bc -o %t.host.o
----------------
bader wrote:

> This is an existing test I wrote like a year ago, Yury just moved/renamed it 
> from `clang/test/Tooling/clang-linker-wrapper-spirv.cpp`.

`clang/test/Tooling/clang-linker-wrapper-spirv.cpp` has been added by @adurang 
in https://github.com/llvm/llvm-project/pull/185425 two weeks ago.
It's not a simple "move/rename" - Yury changed the object file generation 
commands!

> I'm not sure we need to make any content changes to it as part of this PR, 
> it's been working fine,

However, the fact that it works today doesn’t mean it’s the best approach, and 
it’s worth considering whether we can simplify or make the test more robust.

So my question is: do we really need to invoke clang to produce the input files 
for testing the llvm-objdump functionality introduced in 
https://github.com/llvm/llvm-project/pull/185425, or are there simpler/more 
direct ways to construct the required test inputs?

https://github.com/llvm/llvm-project/pull/198434
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to