On Thu Mar 9 12:59:31 2023 +0000, Jinoh Kang wrote:
For example, if the signal occured during a malloc() call or I/O housekeeping, this could lead to an infinite signal trigger loop.
No, it's not (although it's unlikely to matter in practice, and in fact we already do potentially use fprintf in signal handlers... although there are only a couple of places this is possible and they probably should be fixed too.)
Anyway, it's easy to just use write(2, ...) instead, so I'll change it accordingly, thanks.