jonkeane commented on issue #46950: URL: https://github.com/apache/arrow/issues/46950#issuecomment-3117599354
I also just got the email, we have until 2025-08-25 > The macOS ones seem to have since disappeared. We've got a new sanitizer issue though: [gcc-UBSAN](https://www.stats.ox.ac.uk/pub/bdr/memtests/gcc-UBSAN/arrow/arrow-Ex.Rout) Oooph, yeah for this, looks like I was wrong that `gcc-asan` includes both the address and undefined behavior sanitizers https://github.com/apache/arrow/blob/25f8f008061b978137a2c1e5bc934d07fae56e3e/dev/tasks/r/github.linux.sanitizers.yml#L39 it should be easy enough to add the ubsan job, replicate and then fix. Though it appears these are in mimalloc, which might mean patching (ugh) after submitting the fix upstream, or maybe we "just" revert to the system memory allocator for devel linux (and thus these sanitizer runs)? ``` /tmp/RtmpGxyXyv/file5ff2a769a2dd5/mimalloc_ep-prefix/src/mimalloc_ep/src/arena.c:859:49: runtime error: index 2 out of bounds for type 'mi_bitmap_field_t [1]' ``` > Can we suppress the warning on Fedora with -Wno-gnu-zero-variadic-macro-arguments or will this likely be problematic? It's worth a shot, I wonder if this is coming from CRAN using CLANG 20.1.0 and containers still using CLANG 20.0.0? We might need to do something like https://github.com/r-hub/containers/pull/89 for containers to catch this in CI. Do we want to split these two tasks up and tackle them in parallel? Which would you(all?) rather take? I'm happy to do one of them, and either is fine by me -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
