|
|
 | | From: | Stefan Frank | | Subject: | Segmentation faults with Opera 8.0b1 | | Date: | 7 Jan 2005 21:34:21 GMT |
|
|
 | Hello group,
I tried Opera 8 beta today but it exits with segmentation faults
after a few seconds or minutes of surfing. It seems to crash only
when a page is loading. I disabled all plugins and java without success.
The command line argumets -debugplugins and -debugjava returned nothing
suspicious.
The error messages are for example:
../opera: line 1: 4478 Segmentation fault './bin/opera'
../opera: line 1: 4551 Segmentation fault './bin/opera'
../opera: line 1: 4569 Segmentation fault './bin/opera'
../opera: line 1: 4589 Segmentation fault './bin/opera'
I tried the following versions (all with a new personal dir):
opera_8.0-20050104.2-shared-qt_en_woody_i386.deb
opera-8.0-20050104.2-shared-qt.i386-en.tar.bz2
opera-static_8.0-20050104.1-qt_en_i386.deb
In addition I tried 7.60p4 with the same symptoms:
opera-7.60-20041203.2-shared-qt.i386-en.tar.bz2
opera_7.60-20041203.2-shared-qt_en_woody_i386.deb
My system is a Debian woody with all updates.
Opera 7.54 and previous versions were rock solid and ran without any
problems.
Opera 8b works perfectly on my workstation in the company (SuSE 8.1), so
I think it's a Debian specific problem.
Can anyone confirm this behaviour? Can I provide more information to
track down the problem. Google doesn't return much useful stuff, except
disabling plugins.
Thanks in advance,
Stefan
|
|
| | From: | Robt. W. Fletcher Jr | | Subject: | Re: Segmentation faults with Opera 8.0b1 | | Date: | Fri, 07 Jan 2005 14:57:05 -0800 |
|
|
| Stefan Frank writes:
> Hello group,
>
> I tried Opera 8 beta today but it exits with segmentation faults
> after a few seconds or minutes of surfing. It seems to crash only
> when a page is loading. I disabled all plugins and java without success.
> The command line argumets -debugplugins and -debugjava returned nothing
> suspicious.
[snip]
> I tried the following versions (all with a new personal dir):
> opera_8.0-20050104.2-shared-qt_en_woody_i386.deb
> opera-8.0-20050104.2-shared-qt.i386-en.tar.bz2
> opera-static_8.0-20050104.1-qt_en_i386.deb
>
> In addition I tried 7.60p4 with the same symptoms:
> opera-7.60-20041203.2-shared-qt.i386-en.tar.bz2
> opera_7.60-20041203.2-shared-qt_en_woody_i386.deb
>
> My system is a Debian woody with all updates.
>
> Opera 7.54 and previous versions were rock solid and ran without any
> problems.
The .2 shared version??
> Can anyone confirm this behaviour? Can I provide more information to
> track down the problem. Google doesn't return much useful stuff, except
> disabling plugins.
Most strange... 7.54 (.2 I assume) works but 8.0 (.2 or .1) doesn't. Hmmmm?
I don't run Debian woody, but maybe this link might help:
http://my.opera.com/forums/showthread.php?s=9d55afb6f0f1ba89c2eb780f5d31b6a0&threadid=72104
If the link doesn't help, then do the following from an xterm
gcc --version
Stick with this the static version
tar -xjpf opera-8.0-20050104.1-static-qt.i386-en.tar.bz2
cd opera-8.0-20050104.1-static-qt.i386-en
./opera
Does opera still segfault?
Or you can ignore the about and wait for a Debian woody user to chime
in.
--
-rwf
|
|
| | From: | Robt. W. Fletcher Jr | | Subject: | Re: Segmentation faults with Opera 8.0b1 | | Date: | Fri, 07 Jan 2005 17:03:58 -0800 |
|
|
| Stefan Frank writes:
> On Fri, 07 Jan 2005 14:57:05 -0800, Robt. W. Fletcher Jr wrote:
>> Does opera still segfault?
>
> Yes, after a few seconds.
>
> steff@knecht:~/opera-8.0-20050104.1-static-qt.i386-en$ ./opera
> fcntl: Bad file descriptor
> fcntl: Bad file descriptor
> fcntl: Bad file descriptor
> fcntl: Bad file descriptor
> ./opera: line 1: 10953 Segmentation fault './bin/opera'
>
> The fcntl messages only appear with this static version, not with
> the other versions I tried before.
>
> I observed one thing. Opera seems to crash more often on some web
> pages. For example, I have never seen a crash on but
> very often on . Perhaps it depends on the content of
> the page?
>
> After surfing I observed another thing. It seems to be
> a java error. Perhaps it's related. The log, which was written out
> by the java process, is at
> .
>
> The error message from Opera was:
> ./opera: line 1: 11485 Aborted './bin/opera'
>
> Thanks a lot for your hints,
Fat lot of good it did ya.
I'm not sure what fcntl() is telling ya, but if it were my problem I
would:
1. stick with the static version -- running from the download directory
2. if after starting opera (./opera); it stays up, I would disable
java Multimedia> and select Synchronous DNS
Network>. Then check the plugin paths Plug-ins ->
Change path> and de-select everything except for the NS4PluginProxy
from the down load directory. Re-start Opera .
3. Use gdb and attach it to the running pid of opera.
gdb ~/opera-8.0-20050104.1-static-qt.i386-en/opera/bin/opera PidNumberOfOpera
4. Surf away, when opera segfaults, gdb will get control, then enter
bt or bt full to get the backtrace.
5. Hope that somebody who knows what they're doing will pick up on this
thread. ;-)
--
-rwf
|
|
| | From: | Stefan Frank | | Subject: | Re: Segmentation faults with Opera 8.0b1 | | Date: | 8 Jan 2005 10:26:16 GMT |
|
|
| On Fri, 07 Jan 2005 17:03:58 -0800, Robt. W. Fletcher Jr wrote:
> 1. stick with the static version -- running from the download directory
>
> 2. if after starting opera (./opera); it stays up, I would disable
> java Multimedia> and select Synchronous DNS
> Network>. Then check the plugin paths Plug-ins ->
> Change path> and de-select everything except for the NS4PluginProxy
> from the down load directory. Re-start Opera .
[snip]
I tried all your suggestions. If I select Synchronous DNS, everything
runs fine without a crash, even if I reenable java and plugins.
So this fixes my problem.
Synchronous DNS is disabled in my Opera 7.54 settings.
Thanks again for your efforts,
Stefan
A backtrace with Synchronous DNS disabled is attached:
(opera-8.0-20050104.1-static-qt.i386-en)
....
(gdb) c
Continuing.
[New Thread 2049 (LWP 4483)]
[New Thread 1026 (LWP 4484)]
[New Thread 2050 (LWP 4485)]
[New Thread 3074 (LWP 4486)]
[New Thread 4098 (LWP 4487)]
[New Thread 5122 (LWP 4490)]
[New Thread 6146 (LWP 4493)]
[New Thread 7170 (LWP 4494)]
[New Thread 8194 (LWP 4497)]
[New Thread 9218 (LWP 4498)]
[New Thread 10242 (LWP 4510)]
[New Thread 11267 (LWP 4511)]
[New Thread 12290 (LWP 4512)]
[New Thread 13315 (LWP 4513)]
Program received signal SIGSEGV, Segmentation fault.
0x0849b0bb in XAllocColor ()
(gdb) info threads
15 Thread 13315 (LWP 4513) 0xb7b3dbb0 in poll () from /lib/libc.so.6
14 Thread 12290 (LWP 4512) 0xb7b3dbb0 in poll () from /lib/libc.so.6
2 Thread 2049 (LWP 4483) 0xb7b3dbb0 in poll () from /lib/libc.so.6
* 1 Thread 1024 (LWP 4470) 0x0849b0bb in XAllocColor ()
(gdb) bt
#0 0x0849b0bb in XAllocColor ()
#1 0x084938e7 in XAllocColor ()
#2 0x084935b5 in XAllocColor ()
#3 0x081f18ec in XMapRaised ()
#4 0x081f1a32 in XMapRaised ()
#5 0x081eff30 in XMapRaised ()
#6 0x0820eee7 in XMapRaised ()
#7 0x08196bc2 in XMapRaised ()
#8 0x081969eb in XMapRaised ()
#9 0x08197067 in XMapRaised ()
#10 0x08125667 in XMapRaised ()
#11 0x08126397 in XMapRaised ()
#12 0x08196cb3 in XMapRaised ()
#13 0x081969df in XMapRaised ()
#14 0x08196892 in XMapRaised ()
#15 0x08191613 in XMapRaised ()
#16 0x08193097 in XMapRaised ()
#17 0x081910c2 in XMapRaised ()
#18 0x081910c2 in XMapRaised ()
#19 0x081aaaec in XMapRaised ()
#20 0x081abcb0 in XMapRaised ()
#21 0x081236ea in XMapRaised ()
#22 0x08124a63 in XMapRaised ()
#23 0x0810f950 in XMapRaised ()
#24 0x08111d89 in XMapRaised ()
#25 0x081334a5 in XMapRaised ()
#26 0x082266de in XMapRaised ()
#27 0x08225816 in XMapRaised ()
#28 0x0842de2f in XMapRaised ()
#29 0x0842e341 in XMapRaised ()
#30 0x0842dddc in XMapRaised ()
#31 0x0842e55e in XMapRaised ()
#32 0x08624c86 in QObject::activate_signal ()
#33 0x087957da in QSignal::signal ()
#34 0x0873d6dc in QSignal::activate ()
#35 0x08632d4e in QSingleShotTimer::event ()
#36 0x085fd520 in QApplication::internalNotify ()
#37 0x085fd3e1 in QApplication::notify ()
#38 0x085c0082 in qt_activate_timers ()
#39 0x085be132 in QApplication::processNextEvent ()
#40 0x085fe67f in QApplication::enter_loop ()
#41 0x085bdd92 in QApplication::exec ()
#42 0x0842e07c in XMapRaised ()
#43 0x080fe7f0 in XMapRaised ()
#44 0x080fafab in XMapRaised ()
#45 0xb7a8e14f in __libc_start_main () from /lib/libc.so.6
(gdb)
|
|
| | From: | Stefan Frank | | Subject: | Re: Segmentation faults with Opera 8.0b1 | | Date: | 8 Jan 2005 00:18:41 GMT |
|
|
| On Fri, 07 Jan 2005 14:57:05 -0800, Robt. W. Fletcher Jr wrote:
>> My system is a Debian woody with all updates.
>>
>> Opera 7.54 and previous versions were rock solid and ran without any
>> problems.
>
> The .2 shared version??
Yes.
>> Can anyone confirm this behaviour? Can I provide more information to
>> track down the problem. Google doesn't return much useful stuff, except
>> disabling plugins.
>
> http://my.opera.com/forums/showthread.php?s=9d55afb6f0f1ba89c2eb780f5d31b6a0&threadid=72104
>
Debian Woody uses gcc 2.95.4, so the .2 package is the right one.
> Stick with this the static version
>
> tar -xjpf opera-8.0-20050104.1-static-qt.i386-en.tar.bz2
> cd opera-8.0-20050104.1-static-qt.i386-en
> ./opera
>
> Does opera still segfault?
Yes, after a few seconds.
steff@knecht:~/opera-8.0-20050104.1-static-qt.i386-en$ ./opera
fcntl: Bad file descriptor
fcntl: Bad file descriptor
fcntl: Bad file descriptor
fcntl: Bad file descriptor
../opera: line 1: 10953 Segmentation fault './bin/opera'
The fcntl messages only appear with this static version, not with the
other versions I tried before.
I observed one thing. Opera seems to crash more often on some web pages.
For example, I have never seen a crash on but very often
on . Perhaps it depends on the content of the page?
After surfing I observed another thing. It seems to be a
java error. Perhaps it's related. The log, which was written out by the
java process, is at .
The error message from Opera was:
../opera: line 1: 11485 Aborted './bin/opera'
Thanks a lot for your hints,
Stefan
|
|
| | From: | Robt. W. Fletcher Jr | | Subject: | Re: Segmentation faults with Opera 8.0b1 | | Date: | Sat, 08 Jan 2005 09:53:53 -0800 |
|
|
| Stefan Frank writes:
> On Fri, 07 Jan 2005 17:03:58 -0800, Robt. W. Fletcher Jr wrote:
>>
>> 1. stick with the static version -- running from the download
>> directory
>>
>> 2. if after starting opera (./opera); it stays up, I would disable
>> java Multimedia> and select Synchronous DNS
>> Network>. Then check the plugin paths Plug-ins ->
>> Change path> and de-select everything except for the
>> NS4PluginProxy from the down load directory. Re-start Opera
>> .
> [snip]
>
> I tried all your suggestions. If I select Synchronous DNS, everything
> runs fine without a crash, even if I reenable java and plugins.
> So this fixes my problem.
>
> Synchronous DNS is disabled in my Opera 7.54 settings.
>
Very strange! I run Gentoo kernel 2.6.9 which has a lot of
improvements for threading, but...
If I were you, I'd bug report it. Debian woody 'should' run with the
static version (you didn't mention if the .2 version of 8.0b1 ran okay
with Synchronous DNS selected). When you report it make use you tell
them you ran from the download directory w/ all factory defaults
except for selecting Synchronous DNS. Too ensure is, rm -rf the
usersettings directory and start opera i.e.:
cd ~/opera-8.0-20050104.1-static-qt.i386-en
rm -rf usersettings
./opera
Set Synchronous DNS
If they have a debian woody system, they 'might' be able to re-create
the crash.
--
-rwf
|
|
|
