amy-kwan wrote: > @amy-kwan Sorry, my mistake, I didn't connect this to #223248. Do the zeroed > attributes cause link failures on z/OS? If so we can take it as a fix; it'd > also need @jh7370's re-review and a squash.
Thanks a lot, @tru! My apologies once again about forgetting to explicitly link this PR in the other comment. As for the zeroed attributes flags for the archive, they have not caused any immediate link failures that we have observed. However, the zeroed attributes are considered incorrect from the z/OS binder perspective, and they would likely cause the binder to make incorrect assumptions about the linked code. In my testing, I haven't observed any link failures or incorrect runtime behaviour (yet), but it would be more correct for llvm-ar to match the behaviour of z/OS ar a bit more closely when it comes to GOFF archives. This patch just fills in the bits that were always supposed to be there but were hardcoded to zero. https://github.com/llvm/llvm-project/pull/226705 _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
