Control: severity -1 serious Control: tag -1 + patch Hi,
On Fri, 30 Oct 2020 01:58:31 +0100 Johannes 'josch' Schauer <jo...@debian.org> wrote: > For armhf I was able to bisect Debian using snapshot.debian.net and find out > that the first snapshot timestamp that produces the timeout is 2020-09-15 > 06:00:00+01:00. sorry, this was actually wrong. What my bisection script saw on that date was a M-A:same version skew and not the problem I raised in this bug report. Further investigation revealed that the problem indeed started somewhere between 2020-09-23 and 2020-10-05. Since there was a fakeroot upload during that time I tried bisecting fakeroot itself and built the shared sysv library for my native amd64 architecture as well as for armhf. The offending commit is: > commit 2a53909e732b19ccbaf6d1534e932dff74fa757e (refs/bisect/bad) > Author: Vasyl Gello <vasek.ge...@gmail.com> > Date: Fri Oct 2 03:48:45 2020 +0000 > > Use fixed-width members in fake_msg > > * Also usevthe same padding on all architectures > * Fixes cross-architecture SysV IPC connection failures > (i.e using system-wide fakeroot in chroot-ed jails) If I revert that commit on the current master branch, then everything works fine again, thus I'm tagging this bug with "patch". Please either fix this issue or revert the commit until you have a solution that doesn't break this use-case. It currently makes the mmdebstrap autopkgtest fail and thus will block it from migration to testing (thus raising severity). Many thanks to Jessica Clarke and Guillem Jover for their input that helped me track down this issue! cheers, josch
signature.asc
Description: signature
