On Mon, Sep 28, 2020 at 07:26:15PM -0400, Rich Felker wrote: > In investigating the MT-fork deadlock stuff and working on a patch > that makes it attempt to trap rather than deadlocking, I found a > problem with interaction of fork and aio. A fix for that, along with > the patch to trap, and a minimal testcase for the aio bug, are > attached. > > There's also a problematic interaction of abort with fork that was > just found in glibc that also exists in musl, that I still need to > fix. I'll follow up later with a proposed solution for that. And here's the proposed abort fix.