AppImage 2.2.1 crashes when try to restore tabs

[andr222]

Using AppImage 2.2.1 (on Linux 64bit Crash-2017-12-05T01:11:18.txt

) When try to restore tabs, Qupzilla will crash on attempting to reload certain pages on start. If enough pages are unchecked, Qupzilla will no longer crash, but it can take a large number of retries before unchecking the page(s) causing the crash. Since it is important to keep most of the pages from the previous session, it is not useful to avoid restoring. On a new profile, this no longer occurs, since there are no pages to restore. These crashes generally occur when there is much free memory available. Most recently there was more than 3 Gb disponible, as well as more than 3Gb available on swap.

error message in terminal : [andr qupzilla]$ ./QupZilla-2.2.1.AppImage QupZilla: 4 extensions loaded [4693:4723:1205/011059.940971:ERROR:nss_util.cc(808)] After loading Root Certs, loaded==false: NSS error code: -8018 QupZilla: Crashed :( Saving backtrace in /home/andr/.config/qupzilla/crashlog ... Backtrace successfully saved in /home/andr/.config/qupzilla/crashlog/Crash-2017-12-05T01:11:18.txt [1:1:0100/000000.235617:ERROR:broker_posix.cc(41)] Invalid node channel message [1:1:0100/000000.235772:ERROR:broker_posix.cc(41)] Invalid node channel message [1:1:0100/000000.235853:ERROR:broker_posix.cc(41)] Invalid node channel message Abandon (core dumped) [andr qupzilla]$

[nowrep]

Please try with new version (2.2.2) when it is available as AppImage.

[ghost]

@nowrep Just for the record: i can reproduce that not only with AppImage but also with system binaries...

[nowrep]

@cranes-bill With 2.2.2?

[ghost]

@nowrep Yes But i thing i was facing that in earlier version. Just try the following: Open as more tabs as possible, then send SIGTERM signal to Qupzilla and then start it again. If you are able to restore the tabs, so something in my system is wrong.

[nowrep]

Yes, I did just that. I had issues before 8949cb415165161ee9bdbfa31d501f7c493ea73f but now I can't reproduce it anymore.

[ghost]

Unfortunately almost all libraries are stripped, so probably i can't get good backtrace, but if have some way to find where is the reason....?

[nowrep]

In QtWebEngine as you can see from the backtrace. Just for the record I am using Qt 5.9.3 from Arch repo.

[ghost]

Seems it crashing "where it wants":

============== BACKTRACE ==============
#0: /usr/bin/qupzilla() [0x403648]
#1: /lib64/libc.so.6(+0x35270) [0x7fffee2c0270]
#2: /usr/lib64/dri/i965_dri.so(+0x8e7b10) [0x7fffd923fb10]

And a bit from GDB:

Starting program: /usr/bin/qupzilla 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7fffd9d56700 (LWP 2920)]
[New Thread 0x7fffd1517700 (LWP 2921)]
Detaching after fork from child process 2922.
[New Thread 0x7fffd0d16700 (LWP 2923)]
[New Thread 0x7fffcbfff700 (LWP 2924)]
[New Thread 0x7fffcb7fe700 (LWP 2925)]
[New Thread 0x7fffcaffd700 (LWP 2926)]
[New Thread 0x7fffca7fc700 (LWP 2927)]
[New Thread 0x7fffc9ffb700 (LWP 2928)]
[New Thread 0x7fffc97fa700 (LWP 2929)]
[New Thread 0x7fffc8ff9700 (LWP 2930)]
[New Thread 0x7fffa7fff700 (LWP 2931)]
[New Thread 0x7fffa77fe700 (LWP 2932)]
[New Thread 0x7fffa67fc700 (LWP 2934)]
[New Thread 0x7fffa6ffd700 (LWP 2933)]
[New Thread 0x7fffa5ffb700 (LWP 2935)]
[New Thread 0x7fffa57fa700 (LWP 2936)]
[New Thread 0x7fffa4ff9700 (LWP 2937)]
[New Thread 0x7fff87fff700 (LWP 2938)]
[New Thread 0x7fff86efa700 (LWP 2939)]
[New Thread 0x7fff866f9700 (LWP 2940)]
[New Thread 0x7fff85ef8700 (LWP 2941)]
[New Thread 0x7fff669ba700 (LWP 2944)]
[New Thread 0x7fff661b9700 (LWP 2953)]
[Thread 0x7fff661b9700 (LWP 2953) exited]

Program received signal SIGSEGV, Segmentation fault.
0x00007fffd923fb10 in ?? () from /usr/lib64/dri/i965_dri.so
(gdb) continue
Continuing.

Program received signal SIGSEGV, Segmentation fault.
0x00007fffd923fb10 in ?? () from /usr/lib64/dri/i965_dri.so
(gdb) continue
Continuing.

Program received signal SIGABRT, Aborted.
0x00007fffee2c01f7 in raise () from /lib64/libc.so.6
(gdb) continue
Continuing.
[Thread 0x7fff669ba700 (LWP 2944) exited]
[Thread 0x7fff866f9700 (LWP 2940) exited]
[Thread 0x7fff86efa700 (LWP 2939) exited]
[Thread 0x7fff87fff700 (LWP 2938) exited]
[Thread 0x7fffa4ff9700 (LWP 2937) exited]
[Thread 0x7fff85ef8700 (LWP 2941) exited]
[Thread 0x7fffa57fa700 (LWP 2936) exited]
[Thread 0x7fffa5ffb700 (LWP 2935) exited]
[Thread 0x7fffa67fc700 (LWP 2934) exited]
[Thread 0x7fffa6ffd700 (LWP 2933) exited]
[Thread 0x7fffa77fe700 (LWP 2932) exited]
[Thread 0x7fffa7fff700 (LWP 2931) exited]
[Thread 0x7fffc8ff9700 (LWP 2930) exited]
[Thread 0x7fffc97fa700 (LWP 2929) exited]
[Thread 0x7fffc9ffb700 (LWP 2928) exited]
[Thread 0x7fffca7fc700 (LWP 2927) exited]
[Thread 0x7fffcaffd700 (LWP 2926) exited]
[Thread 0x7fffcb7fe700 (LWP 2925) exited]
[Thread 0x7fffcbfff700 (LWP 2924) exited]
[Thread 0x7fffd0d16700 (LWP 2923) exited]
[Thread 0x7fffd1517700 (LWP 2921) exited]
[Thread 0x7ffff7e04900 (LWP 2916) exited]
[2959:2959:0100/000000.688219:ERROR:broker_posix.cc(41)] Invalid node channel message
[2957:2957:0100/000000.688219:ERROR:broker_posix.cc(41)] Invalid node channel message
[2955:2955:0100/000000.688495:ERROR:broker_posix.cc(41)] Invalid node channel message
[2961:2961:0100/000000.688721:ERROR:broker_posix.cc(41)] Invalid node channel message
[2963:2963:0100/000000.691169:ERROR:broker_posix.cc(41)] Invalid node channel message
[2965:2965:0100/000000.691886:ERROR:broker_posix.cc(41)] Invalid node channel message
[2967:2967:0100/000000.695538:ERROR:broker_posix.cc(41)] Invalid node channel message
[2969:2969:0100/000000.698430:ERROR:broker_posix.cc(41)] Invalid node channel message
[2971:2971:0100/000000.699553:ERROR:broker_posix.cc(41)] Invalid node channel message
[2975:2975:0100/000000.699597:ERROR:broker_posix.cc(41)] Invalid node channel message
[2977:2977:0100/000000.701122:ERROR:broker_posix.cc(41)] Invalid node channel message
[2973:2973:0100/000000.703727:ERROR:broker_posix.cc(41)] Invalid node channel message
[2983:2983:0100/000000.706552:ERROR:broker_posix.cc(41)] Invalid node channel message
[2979:2979:0100/000000.706647:ERROR:broker_posix.cc(41)] Invalid node channel message
[2985:2985:0100/000000.707727:ERROR:broker_posix.cc(41)] Invalid node channel message
[2981:2981:0100/000000.710027:ERROR:broker_posix.cc(41)] Invalid node channel message
[2987:2987:0100/000000.714994:ERROR:broker_posix.cc(41)] Invalid node channel message
[2989:2989:0100/000000.721539:ERROR:broker_posix.cc(41)] Invalid node channel message
[2991:2991:0100/000000.724025:ERROR:broker_posix.cc(41)] Invalid node channel message
[2995:2995:0100/000000.726910:ERROR:broker_posix.cc(41)] Invalid node channel message
[2993:2993:0100/000000.727046:ERROR:broker_posix.cc(41)] Invalid node channel message

Program terminated with signal SIGABRT, Aborted.

[nowrep]

Alright, you should type bt after it crashes to get backtrace, but even without that it can be seen that it crashes inside your GPU driver.

[ghost]

Alright, you should type bt after it crashes to get backtrace

It says No stack. (just forgot to mention)

but even without that it can be seen that it crashes inside your GPU driver.

Yeah, so i'm guessing you don't need it.

[ghost]

Always has libc in the report, so probably there is the problem....? Anyway, i think the issue can be closed.

[nowrep]

Always has libc in the report, so probably there is the problem....?

No, the issue most likely isn't in libc.

Anyway, i think the issue can be closed.

Not unless @andr222 reports back.

[ghost]

I also i got from libQt5WebEngineCore and libQt5Core, but not from QupZilla. However, if you need backtrace i can post it.

Anyway, i think the issue can be closed. Not unless @andr222 reports back.

That was addressed most likely to him.

[andr222]

OK, I'll try AppImage 2.2.2 when it is available.

I was finally able to restart Qupzilla successfully, after a clean reboot, and unclicking many pages before restore. Strangely, many (but not all) of the pages I unselected were reloaded.

I say clean reboot since usually the computer is (explicitly or automatically after timeout) put in sleep mode, instead of totally shutting down. This allows for faster reboot, but leaves something in the swap partition. A clean reboot clears the swap. The swap is used even when lots of main memory is available. If I put the computer explicitly in sleep mode, I usually first close Qupzilla & other large applications.

Note that I tend to have a lot of pages loaded. Most of the time, Qupzilla starts without problem, although it does crash (at least once) almost every day. Normally it restarts without problem afterwards.

Hopefully all this info helps find the cause ...

[nowrep]

@andr222 It is available right now, download on www.qupzilla.com/download

[andr222]

I'm now running on AppImage 2.2.2, and like 2.2.1 at the moment, it starts ok. I'll post if the problem starts occuring again. (should I open a new issue if it does ?) Remember that it was intermittant before. I see that the translation to french has improved.

There are still a lot of warnings when I strart in the console. See below :

[32582:32609:1205/173409.051773:ERROR:nss_util.cc(808)] After loading Root Certs, loaded==false: NSS error code: -8018 Xlib: sequence lost (0x106fc > 0x6fe) in reply type 0x0! Xlib: sequence lost (0x109c6 > 0x9c8) in reply type 0x0! [32582:32602:1205/173411.846925:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [32582:32602:1205/173411.846960:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [32582:32602:1205/173411.846990:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com [32582:32602:1205/173411.848763:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [32582:32602:1205/173411.848781:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [32582:32602:1205/173411.848802:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com

Off topic : Qupzilla doesn't start to the same place & size it had when I shut down. I tried changing a line (I forget exactly which) in settings.ini , but it didn't work. It always loads in the same position. Which line in settings.ini changes the load place & size ? Or is it in another file ? Or should I erase some temporary file (like for Mozilla) or something similar ?

[andr222]

I got a crash again, this time with AppImage 2.2.2 (I'm using only AppImage now, to facilitate bug reports) When I restarted, it crashed the first time. The second try started, but the lateral tab bar was lost. (This has happened a few times with previous versions.) I was able to restore the lateral tab bar by reactivating horizontal tabs, exiting tab manager extension, then re-entering to activate lateral tabs, exiting again, then re-entering to deactivate horizontal tabs. The same procedure I have used before. I then quit Qupzilla & restarted, and everything works fine. Now it even restores the same position & size of the Qupzilla window, which it didn't do before. (A lot of nice display fixes in 2.2.2.) After the crash on restarting, I ran from terminal to show output. (Twice, separated by ****)

---

[andr qupzilla]$ ./QupZilla-2.2.2.AppImage init called QupZilla: 6 extensions loaded [25109:25140:1206/004248.524374:ERROR:nss_util.cc(808)] After loading Root Certs, loaded==false: NSS error code: -8018 [25109:25634:1206/004252.192887:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [25109:25634:1206/004252.193029:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [25109:25634:1206/004252.193097:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com [25109:25634:1206/004252.195249:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [25109:25634:1206/004252.195387:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [25109:25634:1206/004252.195447:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com Xlib: sequence lost (0x11150 > 0x1152) in reply type 0x0! Xlib: sequence lost (0x11258 > 0x125a) in reply type 0x0! Xlib: sequence lost (0x113c6 > 0x13ca) in reply type 0x0! Xlib: sequence lost (0x11417 > 0x141a) in reply type 0x0! js: Mixed Content: The page at 'https://meteo.gc.ca/forecast/hourly/qc-96_metric_f.html' was loaded over a secure connection, but contains a form which targets an insecure endpoint 'http://recherche-search.gc.ca/rGs/s_r?#wb-land'. This endpoint should be made available over a secure connection. js: Uncaught TypeError: Cannot read property 'appendChild' of undefined js: Uncaught TypeError: Cannot read property 'appendChild' of undefined js: Mixed Content: The page at 'https://meteo.gc.ca/forecast/hourly/qc-96_metric_f.html' was loaded over a secure connection, but contains a form which targets an insecure endpoint 'http://recherche-search.gc.ca/rGs/s_r?#wb-land'. This endpoint should be made available over a secure connection. libpng warning: iCCP: known incorrect sRGB profile QXcbConnection: XCB error: 3 (BadWindow), sequence: 8325, resource id: 26531019, major code: 40 (TranslateCoords), minor code: 0 QXcbConnection: XCB error: 3 (BadWindow), sequence: 12328, resource id: 25987264, major code: 40 (TranslateCoords), minor code: 0 unload called

---

[andr qupzilla]$ ./QupZilla-2.2.2.AppImage init called QupZilla: 6 extensions loaded [26694:26723:1206/004702.723850:ERROR:nss_util.cc(808)] After loading Root Certs, loaded==false: NSS error code: -8018 [26694:26716:1206/004705.328175:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [26694:26716:1206/004705.328210:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [26694:26716:1206/004705.328241:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com [26694:26716:1206/004705.330687:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [26694:26716:1206/004705.330716:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: ocsp.digicert.com [26694:26716:1206/004705.330740:ERROR:nss_ocsp.cc(591)] No URLRequestContext for NSS HTTP handler. host: crl4.digicert.com js: Mixed Content: The page at 'https://meteo.gc.ca/forecast/hourly/qc-96_metric_f.html' was loaded over a secure connection, but contains a form which targets an insecure endpoint 'http://recherche-search.gc.ca/rGs/s_r?#wb-land'. This endpoint should be made available over a secure connection. js: Uncaught TypeError: Cannot read property 'appendChild' of undefined js: Uncaught TypeError: Cannot read property 'appendChild' of undefined js: Mixed Content: The page at 'https://meteo.gc.ca/forecast/hourly/qc-96_metric_f.html' was loaded over a secure connection, but contains a form which targets an insecure endpoint 'http://recherche-search.gc.ca/rGs/s_r?#wb-land'. This endpoint should be made available over a secure connection.

---

[nowrep]

The second try started, but the lateral tab bar was lost. (This has happened a few times with previous versions.) I was able to restore the lateral tab bar by reactivating horizontal tabs, exiting tab manager extension, then re-entering to activate lateral tabs, exiting again, then re-entering to deactivate horizontal tabs.

Can you please report this in a separate issue?

Also for the rest, please post backtrace, terminal output doesn't say anything useful.

[andr222]

(Sorry for overloading this issue, and forgetting the backtraces.) Repeated crashes on starting occured again. After a clean reboot of system, which cleared the swap, Qupzilla started without problem. Again, some (but not most) of the tabs deselected were loaded. (Swap was 44k or 44m used before reboot, I forget which, of 4G disponible. Main memory approx 3G disponible.) I suspect that swap is related to crashes, since it was a clean reboot that preceded successful starting last time. However please note that Mageia (my Linux distro) clears the /tmp folder on boot, so that may be related. All 5 backtraces are below : (The messages about link to unsecure page is govt weather site link to terminology definitions.)

Crash-2017-12-06T11:32:21.txt Crash-2017-12-06T11:32:57.txt Crash-2017-12-06T11:37:22.txt Crash-2017-12-06T11:39:21.txt Crash-2017-12-06T11:43:17.txt

[nowrep]

@cranes-bill Why does the AppImage use bundled libcrypto but system libssl (first backtrace)? That should be fixed.

[ghost]

It can be fixed tomorrow?

[andr222]

Crash again with AppImage 2.2.2, after starting system from sleep mode. Qupzilla (and other large apps) had been closed before putting in sleep. After initial crash, got restore tabs screen, as usual. Removing many (non-essential) tabs didn't work. (Subsequent tries in console) Adding zeroing swap didn't work. Adding removing /tmp/qtsingleapp-QupZ* entries didn't work. Clean reboot, and QupZilla started automatically without problem. (No restore tab screen.) (Except (I had thought that I had closed QupZilla before reboot, but maybe not.)

[andr222]

(I accidently pushed return before finishing comment) I just had to reactivate extensions, window by window, which I had also deactivated at some point in testing. In sum, it looks like a clean reboot after a crash will recover. Note that usually after awaking from sleep mode, Qupzilla restarts (or re-activates) normally.

Crash logs below : Crash-2017-12-07T10:44:56.txt Crash-2017-12-07T10:45:35.txt Crash-2017-12-07T10:48:49.txt Crash-2017-12-07T10:52:50.txt Crash-2017-12-07T10:55:43.txt Crash-2017-12-07T11:07:34.txt

[andr222]

A further observation about this issue : It only seems to occur if I quit Qupzilla before putting the computer in sleep mode. I have tested leaving Qupzilla open before putting computer in sleep mode, and no more crashes. Before, I would quit Qupzilla before putting the computer in sleep mode, and crashes most days. Also the computer goes automatically into sleep mode after one hour, which happens typically several times a day. So when I quit Qupzilla, it must be leaving some state that causes it to crash on next execution. I suspect that it is qupzilla files in /tmp, since a clean reboot doesn't result in a crash, even if I have quit Qupzilla before the reboot. Note that Mageia (Linux) clears /tmp on boot.

[andr222]

A further note : If I quit Qupzilla before a clean reboot, when I start Qupzilla after the reboot, it simply restores the previous windows & tabs. If I leave Qupzilla running, it is started automatically with an extra empty window, and I am queried to restore tabs. (It would be better without the empty window.) Both acceptable behavior.

[nowrep]

This should be "fixed" in 2.2.4, can you please test?

[the-Arioch]

To me it stil crashes often even 2.2.6....