Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SYCL][CI] Use -D_FORTIFY_SOURCE=2 in Nightly E2E CI #16470

Closed
wants to merge 1 commit into from

Conversation

aelovikov-intel
Copy link
Contributor

See #1117 and
#11436 for the context.

In this mode, std::memcpy cannot be used in device code, but its usage occasionally leaks into our code base. I think testing the configuration in CI is the easiest way to prevent that with at least some confidence. Another option would be to modify FE to report any usage of std::memcpy in device code only, but I don't know if that's even possible.

See #1117 and
#11436 for the context.

In this mode, `std::memcpy` cannot be used in device code, but its usage
occasionally leaks into our code base. I think testing the configuration
in CI is the easiest way to prevent that with at least some confidence.
Another option would be to modify FE to report any usage of
`std::memcpy` in device code only, but I don't know if that's even
possible.
@aelovikov-intel aelovikov-intel requested a review from a team as a code owner December 24, 2024 16:28
@aelovikov-intel
Copy link
Contributor Author

Test run in https://github.com/intel/llvm/actions/runs/12474534716/job/34817029585 shows that it works and catches existing bugs. I'll prepare another PR with fix(es) and will rebase/rerun this once that other PR is merged.

@aelovikov-intel aelovikov-intel marked this pull request as draft December 24, 2024 16:29
@aelovikov-intel
Copy link
Contributor Author

Had a discussion with @steffenlarsen about this. We decided that we'll wait until native device assert/abort regression is fixed and then we'll implement __memcpy_chk wrapper in our devicelib (merging his PR for a temporary sycl::detail::memcpy_no_adl workaround in the meantime). As such, std::memcpy will simply start working and it'll be possible to have a dedicated test for it, making this PR unnecessary.

@aelovikov-intel aelovikov-intel deleted the fortify-nightly branch January 3, 2025 15:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant