Bug#941076: libgcr-base-3-1: pinentry-gnome3 causes gnome-shell crash if used with GLib >= 2.59

Pirate Praveen praveen at onenetbeyond.org
Fri Sep 27 16:05:59 BST 2019



On Fri, Sep 27, 2019 at 15:49, Simon McVittie <smcv at debian.org> wrote:
> Control: tags -1 = bullseye sid
> Control: retitle -1 libgcr-base-3-1: pinentry-gnome3 causes 
> gnome-shell crash if used with GLib >= 2.59
> Control: reassign -1 libgcr-base-3-1 3.28.0-1
> Control: affects -1 + gnome-shell pinentry-gnome3
> Control: fixed -1 3.28.0-4
> 
> On Wed, 25 Sep 2019 at 16:59:01 +0530, Pirate Praveen wrote:
>>  2019-09-25 12:13:52 upgrade libgcr-base-3-1:amd64 3.28.0-1 3.33.4-2
> 
> On Fri, 27 Sep 2019 at 19:20:24 +0530, Pirate Praveen wrote:
>>  Thread 1 (Thread 0x7fe7cc222ac0 (LWP 1660)):
>>  #0  0x00007fe7d5a23468 in prompt_stop_prompting 
>> (self=0x556f95d03770 [GcrSystemPrompter], callback=<optimized out>, 
>> send_done_message=1, wait_for_reply=0) at 
>> gcr/gcr-system-prompter.c:501
> 
> This looks like the bug fixed in 27c58b7f "system-prompter: Fix use
> of g_hash_table_lookup_extended", which was an incompatibility between
> gcr < 3.28.1 and GLib >= 2.59 - and indeed you had gcr < 3.28.1 before
> upgrading. (The fix was also backported in 3.28.0-4, but you didn't
> have that either.) I'll close this bug when the reassign operation has
> been processed.
> 
> So upgrading this library was the correct solution, and there
> is no missing dependency in gnome-shell, pinentry-gnome3 or
> gnome-session.
> 
> If anything, the correct relationship would have been for
> libglib2.0-0 (>= 2.59) to have a Breaks on libgcr-base-3-1 (<< 
> 3.28.0-4~).
> However, that version is older than buster (which was released with
> gcr 3.28.1-1), and in general Debian does not support upgrades from
> older-than-buster to bullseye (or in general from older than Debian n
> to Debian n+1).
> 
> In future please upgrade the entire system to at least the versions in
> stable, then reboot, before attempting to use any package from 
> testing,
> unstable or experimental. This would have saved you a lot of time in
> dealing with an already-fixed bug.
> 

This machine was always in sid, though I only used to update specific 
packages like gnome, firefox, libreoffice etc. I guess I should start 
doing dist-upgrades and not try to save bandwidth.

> This would also have gone considerably faster if you had started the
> bug report by locating the process that segfaulted (gnome-shell) and
> using reportbug to report it as a bug in that process (the gnome-shell
> package). This would have given us a complete list of the libraries
> loaded into that process.
> 

Thanks for tracking down the issue and sorry for the noise.

> Thanks,
>     smcv

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/pkg-gnome-maintainers/attachments/20190927/308a54bc/attachment-0003.html>


More information about the pkg-gnome-maintainers mailing list