Plugin hangs browser

[Nathan]

When I install the lightpsark-mozilla-plugin firefox hangs on loading a lot of sites. Even ones without a flash applet. (planet.gnome.org is one exampe)

Note: Firefox doesnt crash, it just doesnt response to anything anymore. When I kill the browser and retries it happens consistently again. Uninstalling the plugin makes firefox usable again...

I'm using Firefox 3.6.3 from normal ubuntu lucid release

[Edit: This is only with flashblock installed]

[Alessandro Pignotti (a-pignotti)]

Hi,

are you running a 32 bit or 64 bit distribution?

[Nathan (nathansamson)]

64

[Nathan (nathansamson)]

I ran firefox in debug mode, and then loaded planet.gnome.org

It hangs (as expected), and then I ran a bt

[New Thread 0x7fffc69f9710 (LWP 3602)]
INFO: Creating input thread
^C
Program received signal SIGINT, Interrupt.
0x00007ffff7bba03d in pthread_join (threadid=140736525735696,
    thread_return=0x0) at pthread_join.c:89
89 pthread_join.c: Bestand of map bestaat niet.
 in pthread_join.c
(gdb) bt
#0 0x00007ffff7bba03d in pthread_join (threadid=140736525735696,
    thread_return=0x0) at pthread_join.c:89
#1 0x00007fffcfac0e14 in lightspark::RenderThread::wait() ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#2 0x00007fffcfbaff9d in nsPluginInstance::~nsPluginInstance() ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#3 0x00007fffcfbaf076 in NPP_Destroy ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#4 0x00007ffff7157975 in ?? () from /usr/lib/firefox-3.6.3/libxul.so
... Much of the same in libxul

(89 pthread_join.c: Bestand of map bestaat niet.) It says: "File or directory doesn't exists."

When I continued, and ran a bt once again a few seconds later it gave the same result.

It seems to be a race-condition (that happens always on my computer?)

Still I don't get why lightspark plugin would want to render something on planet.gnome since it doens't contain any flash?

[Alessandro Pignotti (a-pignotti)]

lightspark, as any plugin is loaded by the browser when the corresponding mimetype is found. And although planet gnome does not use any flash content, aggregated blogs may. I've just now found a youtube video on planet gnome main page. This may be a race condition triggered by the plugin instance being created and destroyed too fast. Could you please attach the whole log using last version? (0.3.5)

[Nathan (nathansamson)]

I've found the cause of the problem: I've flashblock installed. When I disable it, firefox doesn't hang anymore, it just plain old crashes. (I'll post another bug report for that one)

Complete log (i think?):

(gdb) run
Starting program: /usr/lib/firefox-3.6.3/firefox-bin
[Thread debugging using libthread_db enabled]
[New Thread 0x7fffe6edc710 (LWP 4771)]
[New Thread 0x7fffe66db710 (LWP 4772)]
[New Thread 0x7fffe5cff710 (LWP 4773)]
[New Thread 0x7fffe52f4710 (LWP 4774)]
[New Thread 0x7fffe31ff710 (LWP 4775)]
[Thread 0x7fffe31ff710 (LWP 4775) exited]
[New Thread 0x7fffe31ff710 (LWP 4776)]
[New Thread 0x7fffe0ed4710 (LWP 4777)]
[New Thread 0x7fffe02ff710 (LWP 4786)]
[Thread 0x7fffe31ff710 (LWP 4776) exited]
[New Thread 0x7fffe31ff710 (LWP 4787)]
[New Thread 0x7fffdbebd710 (LWP 4788)]
[New Thread 0x7fffdafff710 (LWP 4789)]
[New Thread 0x7fffda1ff710 (LWP 4790)]
[New Thread 0x7fffd99fe710 (LWP 4791)]
[Thread 0x7fffda1ff710 (LWP 4790) exited]
[Thread 0x7fffd99fe710 (LWP 4791) exited]
5/20/2010 12:34:35 DEBUG: checking init
5/20/2010 12:34:35 DEBUG: starting inbox checking
5/20/2010 12:34:35 DEBUG: trying to login
5/20/2010 12:34:35 DEBUG: set timeout 30000(ms) id 9 for https://www.google.com/accounts/ClientLogin
5/20/2010 12:34:36 DEBUG: parsing login
5/20/2010 12:34:36 DEBUG: cleared timeout id 9
5/20/2010 12:34:36 DEBUG: successfully logged in
5/20/2010 12:34:36 DEBUG: retrieving inbox
5/20/2010 12:34:36 DEBUG: set timeout 30000(ms) id 20 for https://wave.google.com/wave/
[New Thread 0x7fffd99fe710 (LWP 4793)]
5/20/2010 12:34:38 DEBUG: parsing inbox
5/20/2010 12:34:38 DEBUG: cleared timeout id 20
5/20/2010 12:34:38 DEBUG: setting tooltip
5/20/2010 12:34:38 DEBUG: finished checking
[New Thread 0x7fffda1ff710 (LWP 4794)]
[New Thread 0x7fffd52bb710 (LWP 4799)]
[New Thread 0x7fffd4aba710 (LWP 4800)]
[New Thread 0x7fffd42b9710 (LWP 4801)]
[New Thread 0x7fffcc8ff710 (LWP 4802)]
[New Thread 0x7fffcc0fe710 (LWP 4803)]
[New Thread 0x7fffcb8fd710 (LWP 4804)]
[New Thread 0x7fffcb0fc710 (LWP 4805)]
[New Thread 0x7fffca8fb710 (LWP 4806)]
[New Thread 0x7fffca0fa710 (LWP 4807)]
INFO: RenderThread this=0x7fffd35a8d20
[New Thread 0x7fffc98f9710 (LWP 4808)]
INFO: Creating input thread
^C
Program received signal SIGINT, Interrupt.
0x00007ffff7bba03d in pthread_join (threadid=140736575018768,
    thread_return=0x0) at pthread_join.c:89
89 pthread_join.c: Bestand of map bestaat niet.
 in pthread_join.c
(gdb) bt
#0 0x00007ffff7bba03d in pthread_join (threadid=140736575018768,
    thread_return=0x0) at pthread_join.c:89
#1 0x00007fffd29c0e14 in lightspark::RenderThread::wait() ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#2 0x00007fffd2aaff9d in nsPluginInstance::~nsPluginInstance() ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#3 0x00007fffd2aaf076 in NPP_Destroy ()
   from /usr/lib/mozilla/plugins/liblightsparkplugin.so
#4 0x00007ffff7157975 in ?? () from /usr/lib/firefox-3.6.3/libxul.so
#5 0x00007ffff6c3d35d in ?? () from /usr/lib/firefox-3.6.3/libxul.so
#6 0x00007ffff6c3d487 in ?? () from /usr/lib/firefox-3.6.3/libxul.so
#7 0x00007ffff6c3d4fc in ?? () from /usr/lib/firefox-3.6.3/libxul.so
#8 0x00007ffff6c3e10a in ?? () from...

[Alessandro Pignotti (a-pignotti)]

I have not managed to reproduce this on my lucid box. To understand what is going on I have to ask you the backtrace for each thread (use the 'info threads' command to see threads and the 'thread' command to switch thread ). Only those blocked on sem_wait or pthread_join are interesting. Thanks a lot.

[Alessandro Pignotti (a-pignotti)]

Ok, the log contains enough information to debug the issue. Please try version 0.3.7 (to be released later today). It contains some added logging to confirm my hypothesis.

[Nathan (nathansamson)]

I tried the 0.3.7. It has the same issue (but that seems to be expected), but I dont see any logging happening (I dont see the log entry you added i n the source about the deadlock).

But maybe I have to do something special to see this logging?

[Alessandro Pignotti (a-pignotti)]

Please post the whole log, my hypothesis is that there is a race condition when the browser terminate the main stream too early. The logging I've added is in the plugin specific code.

[Nathan (nathansamson)]

(gdb) r
Starting program: /usr/lib/firefox-3.6.3/firefox-bin -P
[Thread debugging using libthread_db enabled]
[New Thread 0x7fffe76ff710 (LWP 16515)]
[New Thread 0x7fffe6efe710 (LWP 16516)]
[New Thread 0x7fffe66fd710 (LWP 16517)]
[New Thread 0x7fffe42df710 (LWP 16518)]
[New Thread 0x7fffe36ff710 (LWP 16519)]
[New Thread 0x7fffe2efe710 (LWP 16520)]
[Thread 0x7fffe36ff710 (LWP 16519) exited]
[Thread 0x7fffe66fd710 (LWP 16517) exited]
[Thread 0x7fffe2efe710 (LWP 16520) exited]
[Thread 0x7fffe42df710 (LWP 16518) exited]
[Thread 0x7fffe6efe710 (LWP 16516) exited]
[Thread 0x7fffe76ff710 (LWP 16515) exited]
process 16514 is executing new program: /usr/lib/firefox-3.6.3/firefox-bin
[Thread debugging using libthread_db enabled]
[New Thread 0x7fffe75f1710 (LWP 16521)]
[New Thread 0x7fffe6be6710 (LWP 16522)]
[New Thread 0x7fffe63e5710 (LWP 16523)]
[New Thread 0x7fffe54ff710 (LWP 16524)]
[New Thread 0x7fffe44ca710 (LWP 16525)]
[Thread 0x7fffe44ca710 (LWP 16525) exited]
[New Thread 0x7fffe44ca710 (LWP 16526)]
[New Thread 0x7fffe34ff710 (LWP 16527)]
[Thread 0x7fffe44ca710 (LWP 16526) exited]
[New Thread 0x7fffe44ca710 (LWP 16528)]
[Thread 0x7fffe44ca710 (LWP 16528) exited]
[New Thread 0x7fffe44ca710 (LWP 16530)]
[New Thread 0x7fffde2bd710 (LWP 16531)]
[New Thread 0x7fffddabc710 (LWP 16532)]
[New Thread 0x7fffdd0ff710 (LWP 16533)]
[New Thread 0x7fffdc6ff710 (LWP 16534)]
[New Thread 0x7fffdbaf5710 (LWP 16535)]
[Thread 0x7fffdbaf5710 (LWP 16535) exited]
[New Thread 0x7fffdbaf5710 (LWP 16536)]
[Thread 0x7fffdbaf5710 (LWP 16536) exited]
[New Thread 0x7fffdbaf5710 (LWP 16537)]
[New Thread 0x7fffd9bff710 (LWP 16538)]
[New Thread 0x7fffd93fe710 (LWP 16539)]
[New Thread 0x7fffd8bfd710 (LWP 16540)]
[Thread 0x7fffe34ff710 (LWP 16527) exited]
[New Thread 0x7fffe34ff710 (LWP 16541)]
[Thread 0x7fffe34ff710 (LWP 16541) exited]
[New Thread 0x7fffe34ff710 (LWP 16542)]
[New Thread 0x7fffd18ff710 (LWP 16543)]
[New Thread 0x7fffd10fe710 (LWP 16544)]
[New Thread 0x7fffd08fd710 (LWP 16545)]
[New Thread 0x7fffd00fc710 (LWP 16546)]
[New Thread 0x7fffcf8fb710 (LWP 16547)]
INFO: RenderThread this=0x7fffd9dacae0
[New Thread 0x7fffcf0fa710 (LWP 16548)]
INFO: Creating input thread
^C
Program received signal SIGINT, Interrupt.
0x00007ffff7bba03d in pthread_join (threadid=140736667297552,
    thread_return=0x0) at pthread_join.c:89
89 pthread_join.c: Bestand of map bestaat niet.
 in pthread_join.c
(gdb) info thread
  34 Thread 0x7fffcf0fa710 (LWP 16548) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  33 Thread 0x7fffcf8fb710 (LWP 16547) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  32 Thread 0x7fffd00fc710 (LWP 16546) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  31 Thread 0x7fffd08fd710 (LWP 16545) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  30 Thread 0x7fffd10fe710 (LWP 16544) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  29 Thread 0x7fffd18ff710 (LWP 16543) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
  28 Thread 0x7fffe34ff710 (LWP 16542) sem_wait ()
    at ../nptl/sysdeps/unix/sysv/linux/x86_64/sem_wait.S:86
 ...

[Alessandro Pignotti (a-pignotti)]

Try release 0.3.8, this issue should be fixed

[Nathan (nathansamson)]

Confirmed, its fixed...

[Jacopo Corbetta (jacopo-arc)]

I'm afraid that still happens.

Go to http://www.corriere.it and try opening any article. Full GDB log attached.