After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 695169 - touchpad tap on screen shield crashes GDM/Gnome-shell
touchpad tap on screen shield crashes GDM/Gnome-shell
Status: RESOLVED INCOMPLETE
Product: gnome-shell
Classification: Core
Component: lock-screen
3.6.x
Other Linux
: Normal normal
: ---
Assigned To: gnome-shell-maint
gnome-shell-maint
Depends on:
Blocks:
 
 
Reported: 2013-03-04 23:54 UTC by bwatkins
Modified: 2013-06-17 11:16 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
Xorg.0.log (25.46 KB, application/octet-stream)
2013-03-05 14:10 UTC, bwatkins
Details
dmesg (59.35 KB, application/octet-stream)
2013-03-06 15:49 UTC, bwatkins
Details
journalctl (2.07 KB, text/plain)
2013-03-06 15:54 UTC, bwatkins
Details
session.log (18.96 KB, text/x-log)
2013-03-06 16:10 UTC, bwatkins
Details

Description bwatkins 2013-03-04 23:54:25 UTC
100% reproducible, using the latest gnome-shell 3.6 on fedora 18. If I try to unlock the screenshield by tapping a few times on the touchpad GDM immediately crashes, logging me out.

Its a system76 lemur ultra laptop with intel hd4000 graphics, and a standard synaptics touchpad.
Comment 1 André Klapper 2013-03-05 09:25:19 UTC
https://live.gnome.org/GnomeShell/Debugging
Comment 2 Giovanni Campagna 2013-03-05 13:51:25 UTC
Actually, I would place my bets on an Xorg crash (if gdm crashes alone, X doesn't go down, and you're not logged out, although the system needs a reboot anyway). Can you check /var/log/Xorg.0.log? Also, did you get any abrt report after the fact?
Comment 3 bwatkins 2013-03-05 14:10:12 UTC
Created attachment 238130 [details]
Xorg.0.log
Comment 4 bwatkins 2013-03-05 14:10:54 UTC
And it didn't seem to produce an abrt crash for some reason.
Comment 5 Giovanni Campagna 2013-03-06 14:06:39 UTC
Ok, if this is a gnome crash (gnome-session? gnome-shell?) we really need a trace then.
You can look in dmesg to find what process actually crashed, and then gdb attach it (eg. "gdb -p `pidof gnome-shell`") from a different VT before you make it crash again.
Comment 6 bwatkins 2013-03-06 15:49:01 UTC
Created attachment 238207 [details]
dmesg

strange. I just made it happen twice, and then looked in dmesg, and it doesn't look like it even added anything to dmesg, nothing to do with gnome in there and I don't see anything alarming in there at all:
Comment 7 bwatkins 2013-03-06 15:50:27 UTC
its a very odd "crash", I have to tap my touchpad a bunch of times on the lock screen (sometimes only a few times, other times a bunch of times) and then it just logs me out and brings me right to the gdm login screen.
Comment 8 bwatkins 2013-03-06 15:51:33 UTC
And I recently switched to arch with gnome 3.6 (thats where this dmesg is from) and I can fully reproduce the "crash" here too.
Comment 9 bwatkins 2013-03-06 15:54:56 UTC
Created attachment 238208 [details]
journalctl

Here's journalctl output, some potentially relevent errors here:
Comment 10 Giovanni Campagna 2013-03-06 16:04:56 UTC
Uhm, yes,
Mar 06 10:39:46 brandon-lemu4 gnome-session[5761]: Gdk-WARNING: gnome-session: Fatal IO error 11 (Resource temporarily unavailable) on X server :1.
Looks relevant.
So... something is taking X down. That must be either an X crash (but it would be logged), or gdm deciding to close the session for some reason. The timestamps are not clear if this was a volountary logout or what.

On the other hand, we have:
Mar 06 10:44:52 brandon-lemu4 gnome-session[6143]: WARNING: Detected that screensaver has left the bus
Which is an indication that gnome-shell went down. ~/.cache/gdm/session.log might shed some light on this.
Comment 11 bwatkins 2013-03-06 16:10:49 UTC
Created attachment 238211 [details]
session.log

some interesting things in that log:
Comment 12 Giovanni Campagna 2013-03-06 16:18:40 UTC
This is really weird: gnome-shell fails to create the dialog, but doesn't log the reason. Which means it's some weird race condition, like cancelling before the dialog is loaded.
And then it goes down because X goes down.

Now, one possible interpretation is that your touchpad is emitting escapes (for whatever reason), so if you keep pressing we keep opening and cancelling dialogs in rapid succession, which eventually confuses gdm and tears X down (without a crash!).
Comment 13 bwatkins 2013-03-06 16:23:46 UTC
So, is tapping the touchpad even supposed to be able to slide up the lock screen (if you have tap-to-click enabled)? that's how I coincidentally discovered the issue, I resumed my laptop, tapped the touchpad since I have tap-to-click enabled and expected the shield to slide up, it didn't so I tapped a few more times and it logged me out, then I tried it again out of curiosity sake and found I could reproduce it :)
Comment 14 Giovanni Campagna 2013-03-06 16:26:01 UTC
No, tapping is not supposed to slide up the lock. But it does activate the drag to unlock, which explains why we open and then immediately cancel the dialog.
Comment 15 Jasper St. Pierre (not reading bugmail) 2013-03-13 07:17:24 UTC
Has anybody else been able to reproduce it?
Comment 16 leebickmtu 2013-03-17 09:47:25 UTC
I tried to reproduce this on my HP dv4-1430us laptop. No matter how many times I tap the touchpad I never get any crash. I tried this both with tap-to-click enabled and disabled. No crashes either way.
Comment 17 Matthias Clasen 2013-03-19 02:37:49 UTC
I can't reproduce either. Moving this off the blocker list
Comment 18 Florian Müllner 2013-06-17 11:16:26 UTC
Closing this bug report as no further information has been provided. Please feel free to reopen this bug if you can provide the information asked for.
Thanks!