[SYCL][CUDA] error: no matching function for call to 'make_context'

[jinz2014]

Building the file "interop-cuda-experimental.cpp" shows the error:

interop-cuda-experimental.cpp:91:19: error: no matching function for call to 'make_context'
  auto sycl_ctx = sycl::make_context<sycl::backend::ext_oneapi_cuda>(cu_ctx);

note: candidate template ignored: requirement 'detail::InteropFeatureSupportMap<sycl::backend::ext_oneapi_cuda>::MakeContext == true' was not satisfied [with Backend = sycl::backend::ext_oneapi_cuda]
make_context(

command:
clang++ -std=c++17 -Wall -fsycl -fsycl-targets=nvptx64-nvidia-cuda -Xsycl-target-backend --cuda-gpu-arch=sm_70 -c interop-cuda-experimental.cpp -o interop-cuda-experimental.o

clang version 17.0.0 (https://github.com/intel/llvm.git 20c1857)

[npmiller]

Great catch!

make_context was disabled in the last CUDA context re-work here: #8197

Not sure why the CI hasn't caught this, I'll update the test

[jinz2014]

Ok. Thanks.

[jinz2014]

Did you update the test ?

[npmiller]

Not yet, looking at it again I uncovered two bugs related to it, I'll open a PR with the test update when they are fixed:

• Using get_backend on an event from a host task hits an assert #11479
• [CUDA] Fix queue creation with native handle oneapi-src/unified-runtime#945

[jinz2014]

Thanks for the links.