You are not logged in.

#1 2019-09-24 16:18:33

barronmo
Member
Registered: 2019-09-23
Posts: 4

mate-panel segfault

After recent update my Mate desktop stopped working (appeared to load, mouse unresponsive, icon bar coming and going).  I have attached a part of the journalctl -b output that was repeating over and over.  Thanks for any help.


Sep 22 22:42:37 exam-desktop audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-coredump@2-839-0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Sep 22 22:42:37 exam-desktop audit[865]: ANOM_ABEND auid=1000 uid=1000 gid=100 ses=1 pid=865 comm="mate-panel" exe="/usr/bin/mate-panel" sig=11 res=1
Sep 22 22:42:37 exam-desktop audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-coredump@4-875-0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Sep 22 22:42:35 exam-desktop clock-applet[823]: Negative content width -7 (allocation 1, extents 4x4) while allocating gadget (node button, owner GtkToggleButton)
Sep 22 22:42:34 exam-desktop dbus-daemon[633]: [session uid=1000 pid=633] Successfully activated service 'org.mate.panel.applet.WnckletFactory'
Sep 22 22:42:37 exam-desktop systemd-coredump[841]: Process 746 (mate-panel) of user 1000 dumped core.
                                                   
                                                    Stack trace of thread 746:
                                                    #0  0x00000000b289b9d9 n/a (librsvg-2.so.2)
                                                    #1  0x00000000b289b782 n/a (librsvg-2.so.2)
                                                    #2  0x00000000b28a7ccf n/a (librsvg-2.so.2)
                                                    #3  0x00000000b28a7608 n/a (librsvg-2.so.2)
                                                    #4  0x00000000b28a4963 n/a (librsvg-2.so.2)
                                                    #5  0x00000000b28a2817 n/a (librsvg-2.so.2)
                                                    #6  0x00000000b28b7481 n/a (librsvg-2.so.2)
                                                    #7  0x00000000b28972b1 n/a (librsvg-2.so.2)
                                                    #8  0x00000000b28913d5 n/a (librsvg-2.so.2)

Offline

#2 2019-09-24 16:46:24

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 833
Website

Re: mate-panel segfault

Sounds like a SSE2 problem. Are you on a CPU supporting SSE2 (cat /proc/cpuinfo), then I strongly suggest to switch to the pentium4 branch.

Offline

#3 2019-09-24 17:50:45

levi
Moderator
From: Yorkshire, UK
Registered: 2018-06-16
Posts: 1,197

Re: mate-panel segfault

That should be mostly solved by pacman these days though, surely?  Most packages have had an update by now I reckon, although not all, but I'd expect something like mate and all of its plugins to have been updated by now, so even if you didn't follow the instructions to reinstall everything it should be okay.


Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.

Offline

#4 2019-09-24 18:02:30

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 833
Website

Re: mate-panel segfault

on i686 I know for sure as librsvg2 has been rebuilt with rust-bin (see AUR), this is the repackaged binary blob from upstream, and that one produces
SSE2 opcodes when compiling rust code. For pentium4, sure, there should not be any difference.

Offline

#5 2019-09-24 21:39:53

barronmo
Member
Registered: 2019-09-23
Posts: 4

Re: mate-panel segfault

Thanks for the quick replies.  One of the flags shown from cpuinfo is sse but no sse2.  Is this the same?  I assumed since my processor is AMD the Pentium4 branch did not apply to me.

Mike

Offline

#6 2019-09-25 06:06:44

deep42thought
Administrator
From: Jena, Germany
Registered: 2017-06-17
Posts: 617

Re: mate-panel segfault

barronmo wrote:

Thanks for the quick replies.  One of the flags shown from cpuinfo is sse but no sse2.  Is this the same?  I assumed since my processor is AMD the Pentium4 branch did not apply to me.

Mike

sse != sse2 - you are right to avoid pentium4 (it requires sse2) - see https://archlinux32.org/architecture/ - otoh this has nothing to do with intel vs. amd.

regards,
deep42thought

Offline

#7 2019-09-25 06:21:31

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 833
Website

Re: mate-panel segfault

Then yes, librsvg2 needs a rebuild with a rust which doesn't produce SSE2 opcodes, the problem is currently rust refuses to rebuild
on i686, see https://bugs.archlinux32.org/index.php? … task_id=71

I'm currently testing on my Pentium3 without SSE2 and it shows problems in various places:
- firefox
- thunderbird
- librsvg2
- libbabl/libgegl (used by gimp)

Though the latest one could also pe micro-optimized in the package itself or SSE2 is produced
by the gcc/binutils toolchain.

Also Gnome has some many broken packages currently, that it's a major effort to even do a
rebuild on pentium4 or x86_64.

Offline

#8 2019-09-25 16:39:55

barronmo
Member
Registered: 2019-09-23
Posts: 4

Re: mate-panel segfault

I managed to get LXDE working which is an adequate work-around for me.

Offline

#9 2019-09-25 18:21:54

andreas_baumann
Administrator
From: Zurich, Switzerland
Registered: 2017-08-10
Posts: 833
Website

Re: mate-panel segfault

Yes, sorry about Gnome, but it will re-emerge. Just needs some work.. :-)

Offline

Board footer

Powered by FluxBB