Bug#964803: dbus-daemon: unaligned trap on alpha

2021-11-01 Thread Антон Кочков
I updated the dbus from "experimental" (from 1.12 to 1.13), and the
problem indeed went for me as well.

I retried with a clean installation from
- 
https://cdimage.debian.org/cdimage/ports/current/debian-11.0.0-alpha-NETINST-1.iso
- 
https://cdimage.debian.org/cdimage/ports/debian-installer/2021-10-20/alpha/debian-installer-images_20211020_alpha.tar.gz

Unaligned trap with the standard dbus (1.12) and nothing after
upgrading it from experimental to 1.13 and rebooting.

Best regards,
Anton Kochkov.



Bug#964803: dbus-daemon: unaligned trap on alpha

2021-05-15 Thread Rich
I can't find a setting to do that in Linux sysctl or
kernel-parameters; or searching around, do you have a pointer?

I tried just telling /etc/systemd/system.conf
DefaultLimitCORE=infinity, setting a non-default path in
kernel.core_pattern in sysctl.conf, restarting, restarting dbus after
boot, and running find after systemctl status dbus reported dying with
SEGV, but I don't see any cores, in the path I set with core_pattern
or otherwise. :(

- Rich

On Sat, May 15, 2021 at 11:31 AM Simon McVittie  wrote:
>
> On Sat, 15 May 2021 at 10:42:55 -0400, Rich wrote:
> > I came here to report this very problem, but "reportbug" suggested I
> > try the dbus from experimental(!), so I did, and the problem (in my
> > very limited experimenting) seems to have gone away...
>
> Are you able to configure the system to dump core on misalignment, and
> get a backtrace from the situation(s) where it crashes?
>
> If 1.12.x crashes and 1.13.x does not, this might mean that commit 26d5d97d
> "Don't cast user-supplied pointers to DBusBasicValue *" is what solved it
> for you.
>
> smcv



Bug#964803: dbus-daemon: unaligned trap on alpha

2021-05-15 Thread Simon McVittie
On Sat, 15 May 2021 at 10:42:55 -0400, Rich wrote:
> I came here to report this very problem, but "reportbug" suggested I
> try the dbus from experimental(!), so I did, and the problem (in my
> very limited experimenting) seems to have gone away...

Are you able to configure the system to dump core on misalignment, and
get a backtrace from the situation(s) where it crashes?

If 1.12.x crashes and 1.13.x does not, this might mean that commit 26d5d97d
"Don't cast user-supplied pointers to DBusBasicValue *" is what solved it
for you.

smcv