unity8-dash hangs in scopes backend

[Michał Sawicz]

This is a forced crash, it was unity8-dash hanging. I got it twice on startup, once during normal operation.

Most suspect are threads 12 to 22.

A `restart unity8-dash` it works again, so this is a race of some kind.

ProblemType: Crash
DistroRelease: Ubuntu RTM 14.09
Package: unity8 8.00+14.10.20140825.3-0ubuntu1
Uname: Linux 3.4.67 armv7l
ApportVersion: 2.14.6-0ubuntu2
Architecture: armhf
CurrentDesktop: Unity
Date: Thu Aug 28 13:19:11 2014
ExecutablePath: /usr/bin/unity8-dash
ExecutableTimestamp: 1408985173
InstallationDate: Installed on 2014-08-27 (0 days ago)
InstallationMedia: Ubuntu Utopic Unicorn (development branch) - armhf (20140827-202410)
ProcCmdline: unity8-dash --desktop_file_hint=/usr/share/applications/unity8-dash.desktop
ProcCwd: /home/phablet
SegvAnalysis: Skipped: missing required field "Disassembly"
Signal: 11
SourcePackage: unity8
Stacktrace:
 #0 0xb6221022 in poll () at ../sysdeps/unix/syscall-template.S:81
 No locals.
 #1 0xb5d460c0 in ?? () from /lib/arm-linux-gnueabihf/libglib-2.0.so.0
 No symbol table info available.
 Backtrace stopped: previous frame identical to this frame (corrupt stack?)
StacktraceTop:
 poll () at ../sysdeps/unix/syscall-template.S:81
 ?? () from /lib/arm-linux-gnueabihf/libglib-2.0.so.0
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dialout dip plugdev sudo tty video

[Apport retracing service (apport)]

StacktraceTop:
 poll () at ../sysdeps/unix/syscall-template.S:81
 g_main_context_poll (priority=2147483647, n_fds=5, fds=0x2799690, timeout=-1, context=0x203b508) at /build/buildd/glib2.0-2.41.3/./glib/gmain.c:4029
 g_main_context_iterate (context=context@entry=0x203b508, block=block@entry=1, dispatch=dispatch@entry=1, self=<optimized out>) at /build/buildd/glib2.0-2.41.3/./glib/gmain.c:3729
 g_main_context_iteration (context=0x203b508, may_block=1) at /build/buildd/glib2.0-2.41.3/./glib/gmain.c:3795
 QEventDispatcherGlib::processEvents (this=0x203b170, flags=...) at kernel/qeventdispatcher_glib.cpp:426

[Michał Sawicz (saviq)]

And I can't reproduce this any more...

[Michał Sawicz (saviq)]

Got this again now during normal operation. I don't think unity8 can do anything here, it looks like the deadlock is somewhere between scopes and zmq.

[Michi Henning (michihenning)]

Hmmm... The stack traces all look normal for an idle process. It could be related to the thread starvation issue on the adapter side (which I haven't gotten back to yet), but in that case, I would have expected a timeout exception.

How long was the dash hung before you took the stack trace? More than 40 seconds or so or definitely less?

[Michał Sawicz (saviq)]

W dniu 02.09.2014 o 03:41, Michi Henning pisze:
> Hmmm... The stack traces all look normal for an idle process. It could
> be related to the thread starvation issue on the adapter side (which I
> haven't gotten back to yet), but in that case, I would have expected a
> timeout exception.
>
> How long was the dash hung before you took the stack trace? More than 40
> seconds or so or definitely less?

Oh much more than that. It was more than 10 minutes actually. I had to
download and install some dbg packages to try and find why it was stuck.

[Jean-Baptiste Lallement (jibel)]

Setting importance to high, it happens during normal operations of the device and requires a reboot.

[taiebot65 (dedreuil)]

Is it similar to the bug i have reported with weather scope? i can reproduce it every time.
https://bugs.launchpad.net/bugs/1362769

[Michał Sawicz (saviq)]

W dniu 08.09.2014 o 21:00, taiebot65 pisze:
> Is it similar to the bug i have reported with weather scope? i can reproduce it every time.
> https://bugs.launchpad.net/bugs/1362769

I doubt it, the one you reported seems to be related to the trust
prompt, nothing like that in this here.

I'll try and reproduce your steps again.

[Michi Henning (michihenning)]

The stack trace isn't all that complete. Quite a few frames are cut off. But the ones that look right are as I would expect them to be for an idle run time. There are no outgoing requests in flight, and everything is basically idle.

Saviq, any pointers as to how I might reproduce and with which image?

[Michał Sawicz (saviq)]

So... I've just tried reproducing this on two devices with both devel-proposed@233 and rtm-proposed@23, flashing, rebooting... I always got a fully working dash on boot...

[Michał Sawicz (saviq)]

I went back to rtm-proposed@17, still nothing.

[Michi Henning (michihenning)]

I've rebooted at least 15 times with proposed@233 and didn't see the problem :-(

[Michał Sawicz (saviq)]

I'm marking incomplete, jibel didn't see it either. It's not great that we don't know what happened, let's revisit this if it comes back.

[Pat McGowan (pat-mcgowan)]

I saw this on yesterdays images (cant recall if it was rtm-proposed or devel-proposed or both)
Twice saw the Unity8 starting splash when open the welcome screen and it never starts

[Jean-Baptiste Lallement (jibel)]

I saw it with krillin rtm-proposed #25.

[Andrea Cimitan (cimi)]

I have this on krillin, today image

[Andrea Cimitan (cimi)]

and image 33 too

[Launchpad Janitor (janitor)]

[Expired for unity8 (Ubuntu) because there has been no activity for 60 days.]

[Michi Henning (michihenning)]

No-one has seen this for months, so I'm marking as Invalid.