SYCL :: EnqueueNativeCommand/custom-command-multiple-dev-cuda.cpp failing in nightly on CUDA

[sarnex]

Describe the bug

Failed Tests (1):
  SYCL :: EnqueueNativeCommand/custom-command-multiple-dev-cuda.cpp

Command Output (stdout):
--
# RUN: at line 2
/__w/llvm/llvm/toolchain/bin//clang++   -fsycl -fsycl-targets=nvptx64-nvidia-cuda  /__w/llvm/llvm/llvm/sycl/test-e2e/EnqueueNativeCommand/custom-command-multiple-dev-cuda.cpp -o /__w/llvm/llvm/build-e2e/EnqueueNativeCommand/Output/custom-command-multiple-dev-cuda.cpp.tmp.out -lcuda
# executed command: /__w/llvm/llvm/toolchain/bin//clang++ -fsycl -fsycl-targets=nvptx64-nvidia-cuda /__w/llvm/llvm/llvm/sycl/test-e2e/EnqueueNativeCommand/custom-command-multiple-dev-cuda.cpp -o /__w/llvm/llvm/build-e2e/EnqueueNativeCommand/Output/custom-command-multiple-dev-cuda.cpp.tmp.out -lcuda
# note: command had no output on stdout or stderr
# RUN: at line 3
env SYCL_PI_CUDA_ENABLE_IMAGE_SUPPORT=1 ONEAPI_DEVICE_SELECTOR=cuda:gpu  /__w/llvm/llvm/build-e2e/EnqueueNativeCommand/Output/custom-command-multiple-dev-cuda.cpp.tmp.out
# executed command: env SYCL_PI_CUDA_ENABLE_IMAGE_SUPPORT=1 ONEAPI_DEVICE_SELECTOR=cuda:gpu /__w/llvm/llvm/build-e2e/EnqueueNativeCommand/Output/custom-command-multiple-dev-cuda.cpp.tmp.out
# .---command stdout------------
# | 1 devices found
# | Using on device 0: NVIDIA GeForce RTX 3090
# | Checking result on host: FAILED
# | 0 ?= 1
# `-----------------------------
# error: command failed with exit status: 1

https://github.com/intel/llvm/actions/runs/10051976562/job/27782653780

To reproduce

No response

Environment

No response

Additional context

No response

[sarnex]

@hdelan Looks like this test was added in #14136. Can you please take a look at this?

[hdelan]

Have made a patch here to not run for single device context #14717 . Although this should still work. I'll try to repro and see what is happening. Potentially there is a race condition somewhere.