USM/fill_any_size.cpp XPASS in OCL CPU in Nightly

[sarnex]

Describe the bug

I tried just removing the XFAIL but then it fails in precommit on OCL CPU, which is really weird

#16433

Needs investigation

 XPASS: SYCL :: USM/fill_any_size.cpp (1994 of 2242)
******************** TEST 'SYCL :: USM/fill_any_size.cpp' FAILED ********************
Exit Code: 0

Command Output (stdout):
--
# RUN: at line 1
/__w/llvm/llvm/toolchain/bin//clang++ -D_GLIBCXX_USE_CXX11_ABI=0  -fsycl -fsycl-targets=spir64  /__w/llvm/llvm/llvm/sycl/test-e2e/USM/fill_any_size.cpp -o /__w/llvm/llvm/build-e2e/USM/Output/fill_any_size.cpp.tmp1.out
# executed command: /__w/llvm/llvm/toolchain/bin//clang++ -D_GLIBCXX_USE_CXX11_ABI=0 -fsycl -fsycl-targets=spir64 /__w/llvm/llvm/llvm/sycl/test-e2e/USM/fill_any_size.cpp -o /__w/llvm/llvm/build-e2e/USM/Output/fill_any_size.cpp.tmp1.out
# note: command had no output on stdout or stderr
# RUN: at line 2
env ONEAPI_DEVICE_SELECTOR=opencl:cpu  /__w/llvm/llvm/build-e2e/USM/Output/fill_any_size.cpp.tmp1.out
# executed command: env ONEAPI_DEVICE_SELECTOR=opencl:cpu /__w/llvm/llvm/build-e2e/USM/Output/fill_any_size.cpp.tmp1.out
# .---command stdout------------
# | All 64 tests passed
# `-----------------------------

--

https://github.com/intel/llvm/actions/runs/12405518106/job/34632838094

To reproduce

No response

Environment

No response

Additional context

No response

[sarnex]

FYI @rafbiels

[rafbiels]

Thanks for flagging. This is expected, the issue happens semi-randomly as explained in the linked oneapi-src/unified-runtime#2440. This is a memory alignment issue so any change in memory layout from an unrelated change (even as simple as renaming the binary) may cause the issue to disappear or reappear.

Shall I change the XFAIL to UNSUPPORTED to avoid XPASS reports? In principle this is "supported", just the implementation has a bug.

[sarnex]

thanks for the explanation, yeah if its sporadic lets make it UNSUPPORTED, thats what we usually do for sporadic failures, thx

[rafbiels]

The fix PR #16443 got held up over my holiday and the test addition was reverted instead in #16465. I'll resubmit the test addition with UNSUPPORTED shortly.

[rafbiels]

@sarnex would you be able to advise on how to format the UNSUPPORTED-TRACKER line given that the clang-format check fails with the following suggestion:

-// UNSUPPORTED-TRACKER: https://github.com/oneapi-src/unified-runtime/issues/2440
+// UNSUPPORTED-TRACKER:
+// https://github.com/oneapi-src/unified-runtime/issues/2440

I assume there are parsers extracting this information from the comment. Do they allow a line break here? Is this documented somewhere?

[sarnex]

the grep string is here, but i would just do

// clang-format off
// UNSUPPORTED: whatever
// clang-format on