On Tue, 2022-02-01 at 16:23 +0000, Thorsten Glaser wrote: > retitle 925358 qemu-user-static: mis-emulates something to do with > process/signal handling (m68k, s390x, …) > affects 925358 klibc-dev > thanks > > This still happens. (And retitling because I almost filed a bug > against klibc again… oops…) > > Look for “mtest-external” (second occurrence) in: > https://buildd.debian.org/status/fetch.php?pkg=mksh&arch=m68k&ver=59c-16&stamp=1643675884&raw=0 > > Incidentally, the last buildd build on which this worked was an ARAnyM > one which I ran myself, back then. > > For some reason, this does work with glibc and musl, so perhaps there > *is* a bug or… at least a chance to work around maybe? in klibc…
klibc's signal handling is probably a bit different from the others, and it certainly has been buggy on some architectures in the past. It seems to be solid now on real hardware. I regularly build and test klibc across most supported architectures, using QEMU, and have had to use a locally patched version due to multiple regressions: <https://git.kernel.org/pub/scm/linux/kernel/git/bwh/klibc-maint.git/plain/status.md>. Ben. -- Ben Hutchings Time is nature's way of making sure that everything doesn't happen at once.
signature.asc
Description: This is a digitally signed message part
