GNOME Bugzilla – Bug 600552
crash in Document Viewer: Opening a dvi file, this...
Last modified: 2013-06-14 05:20:18 UTC
Version: 2.28.1 What were you doing when the application crashed? Opening a dvi file, this comes from https://bugzilla.gnome.org/show_bug.cgi?id=563716 , but now, evince crashes just after starting it (with 2.26, it crashed when seeing page 5 or so, like with previous) Distribution: Gentoo Base System release 1.12.13 Gnome Release: 2.26.3 2009-10-08 (Gentoo) BugBuddy Version: 2.28.0 System: Linux 2.6.31-zen4 #4 ZEN PREEMPT Wed Oct 21 23:17:17 CEST 2009 x86_64 X Vendor: The X.Org Foundation X Vendor Release: 10603901 Selinux: No Accessibility: Disabled GTK+ Theme: Shiki-Gentoo Icon Theme: gnome-gentoo GTK+ Modules: canberra-gtk-module, gnomebreakpad Memory status: size: 312180736 vsize: 312180736 resident: 43917312 share: 18853888 rss: 43917312 rss_rlim: 18446744073709551615 CPU usage: start_time: 1257257366 rtime: 34 utime: 28 stime: 6 cutime:641 cstime: 47 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/bin/evince' [Thread debugging using libthread_db enabled] [New Thread 0x7fb9a8d40790 (LWP 30953)] [New Thread 0x7fb999aa5950 (LWP 32222)] [New Thread 0x7fb99c051950 (LWP 30954)] 0x00007fb9a4d80434 in __lll_lock_wait () from /lib/libpthread.so.0
+ Trace 218771
Thread 3 (Thread 0x7fb99c051950 (LWP 30954))
----------- .xsession-errors (49 sec old) --------------------- kbuildsycoca: WARNING: '/usr/share/applications/evince.desktop' specifies undefined mimetype/servicetype 'application/x-cbr' kbuildsycoca: WARNING: '/usr/share/applications/evince.desktop' specifies undefined mimetype/servicetype 'application/x-cbz' kbuildsycoca: WARNING: '/usr/share/applications/evince.desktop' specifies undefined mimetype/servicetype 'application/x-cb7' kbuildsycoca: WARNING: 'katepart.desktop' specifies undefined mimetype/servicetype 'text/x-fortran' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/rtf' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.ms-works' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.oasis.opendocument.text-master' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.sun.xml.writer.global' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.wordperfect' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/x-extension-txt' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/x-t602' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.ms-word.document.macroEnabled.12' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.openxmlformats-officedocument.wordprocessingml.document' kbuildsycoca: WARNING: '/usr/share/applications/writer.desktop' specifies undefined mimetype/servicetype 'application/vnd.ms-word.template.macroEna ...Too much output, ignoring rest... --------------------------------------------------
*** Bug 563716 has been marked as a duplicate of this bug. ***
DVI file from https://bugzilla.gnome.org/attachment.cgi?id=124186 (bug 563716) can be used for this also
Couldn't reproduce this one with Evince 2.29.2 (git master). I believe this issue to be obsolete now.
But, isn't 2.28 still supported? (it's current stable release)
Ok, I didn't use the proper way to say it. :) But, I can't find this issue in the git master. Could you try with a newer version of Evince and confirm that it is still here ?
Created attachment 148753 [details] backtrace with 2.29.2 It still crashes with 2.29.2 :'(
(In reply to comment #5) > Ok, I didn't use the proper way to say it. :) > No problem, I was compiling 2.29.2 to test :-)
Created attachment 148756 [details] backtrace This is a more complete backtrace after compiling ghostscript with debugging CFLAGS also
Please, get use to attach the trace in the body of your message. The trace will not be registered inside the bugzilla if you don't. Thanks ! :) Distribution: Gentoo Base System release 1.12.13 Gnome Release: 2.26.3 2009-11-17 (Gentoo) BugBuddy Version: 2.28.0 System: Linux 2.6.31-gentoo-r6 #1 PREEMPT Sat Nov 14 21:21:30 CET 2009 x86_64 X Vendor: The X.Org Foundation X Vendor Release: 10603901 Selinux: No Accessibility: Disabled GTK+ Theme: Shiki-Gentoo Icon Theme: gnome-gentoo GTK+ Modules: canberra-gtk-module, gnomebreakpad Memory status: size: 303366144 vsize: 303366144 resident: 37330944 share: 19849216 rss: 37330944 rss_rlim: 18446744073709551615 CPU usage: start_time: 1259595618 rtime: 41 utime: 32 stime: 9 cutime:1344 cstime: 285 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/bin/evince' [Thread debugging using libthread_db enabled] [New Thread 0x7f2bc1d5c790 (LWP 28859)] [New Thread 0x7f2bb2386950 (LWP 31391)] [New Thread 0x7f2bb4f61950 (LWP 28862)] 0x00007f2bbba31af6 in poll () from /lib/libc.so.6
+ Trace 219389
Thread 3 (Thread 0x7f2bb4f61950 (LWP 28862))
----------- .xsession-errors (763 sec old) --------------------- (epiphany:27326): Gtk-CRITICAL **: gtk_entry_set_progress_fraction: assertion `GTK_IS_ENTRY (entry)' failed (epiphany:27326): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `GtkEntry' (epiphany:27326): Gtk-CRITICAL **: gtk_entry_set_progress_fraction: assertion `GTK_IS_ENTRY (entry)' failed (epiphany:27326): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `GtkEntry' (epiphany:27326): Gtk-CRITICAL **: gtk_entry_set_progress_fraction: assertion `GTK_IS_ENTRY (entry)' failed (epiphany:27326): GLib-GObject-WARNING **: invalid uninstantiatable type `(null)' in cast to `GtkEntry' (epiphany:27326): Gtk-CRITICAL **: gtk_entry_set_progress_fraction: assertion `GTK_IS_ENTRY (entry)' failed ...Too much output, ignoring rest... --------------------------------------------------
No, I really can't reproduce this crash on my system... :-/ I sum up what I think is relevant: - Evince 2.29.2 (git master) - Cairo 1.9.5 (git master) Maybe it's coming from your Cairo lib which is not up-to-date ??? If you don't have the same as me, maybe try to get the latest one: http://www.cairographics.org/download/ If it still crash, then I have no damn idea of what it could ! :)
Wait, I just noticed that you are using a 64bits machine... Maybe it's important !
(In reply to comment #9) > Please, get use to attach the trace in the body of your message. The trace will > not be registered inside the bugzilla if you don't. Thanks ! :) Ups, sorry, I thought it would waste a lot of space but seems that gnome bugzilla detects backtraces (In reply to comment #10) > No, I really can't reproduce this crash on my system... :-/ > What distribution do you use? For example, opensuse's ghostscript includes really a lot of patches, while gentoo's one is more close to upstream (maybe the problem is in ghostscript :-|, but I am not sure how could I confirm it) > I sum up what I think is relevant: > - Evince 2.29.2 (git master) > - Cairo 1.9.5 (git master) > > Maybe it's coming from your Cairo lib which is not up-to-date ??? > > If you don't have the same as me, maybe try to get the latest one: > http://www.cairographics.org/download/ > > If it still crash, then I have no damn idea of what it could ! :) I have tried also with cairo-1.9.4 with the same result :-( (In reply to comment #11) > Wait, I just noticed that you are using a 64bits machine... Maybe it's > important ! On the only 32bits machine I have tried, the file is never opened (well, maybe I should try to leave it opening the file more hours :-/), it runs some commands related with kpathsea but never opens the file, while xdvi shows it fine
I can't reproduce it either on 32 bits. Could you try opening the eps files instead of the dvi to see whether it's an issue in the ps backend?
I have just tried and seems that evince is able to read all of them properly (even if evince-thumbnailers seems to not work for some of them :-/)
I'm closing this issue as we cannot reproduce this bug and bug 603396. Feel free to reopen it when you are sure to have a way for us to reproduce it.
And, what I am supposed to do?! What more can I try? Could you still give me any thoughts about where could be the problem at least? I am suffering this crash with this document (even with a new created user account) since always and this way of closing bugs is really surprising This bug is different than bug 603396, as the last one only occurs with 2.29 and maybe because I am still under gnome-2.26, but this bug report occurs with all evince versions I have tried
It also fails on Fedora12 (just installed), it doesn't crash but evince starts to waste CPU resources and never opens the file, while xdvi opens it. Some kpathsea related errors are shown: $ LC_ALL=C evince paper.dvi kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+360/600 --dpi 360 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. kpathsea: Appending font creation commands to missfont.log. page: Warning: font `ptmb8r' at 360x360 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+360/600 --dpi 360 ptmr8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmr8r' at 360x360 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 1+0/600 --dpi 600 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmb8r' at 600x600 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 1+0/600 --dpi 600 ptmri8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmri8r' at 600x600 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+420/600 --dpi 420 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmb8r' at 420x420 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+540/600 --dpi 540 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmb8r' at 540x540 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+360/600 --dpi 360 ptmri8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmri8r' at 360x360 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+540/600 --dpi 540 ptmri8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmri8r' at 540x540 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+420/600 --dpi 420 ptmr8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmr8r' at 420x420 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 1+120/600 --dpi 720 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmb8r' at 720x720 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 0+540/600 --dpi 540 ptmr8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmr8r' at 540x540 not found, trying `cmr10' instead kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 1+0/600 --dpi 600 ptmr8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmr8r' at 600x600 not found, trying `cmr10' instead -> On my gentoo systems it crashes here, while on fedora it continues showing this forever (and wasting a lot of CPU)
When running it as root, seems that fonts are created but, when it ends, evince crashes. On fedora the error is different, I get: page: Warning: font `cmr10' not found, trying metric files instead page: Error: /usr/share/texmf/fonts/tfm/public/cm/cmr10.tfm: File corrupted, or not a TFM file forever, and evince never shows up.
I would say: "It's time for memtest86+ !"
I already run it without memory errors, also, take care that I am a gentoo user and I compile a lot of software. If I don't misremember, they should show random problems that I don't see. Also, ther crash is ALWAYS reproducible for me :-/ On the other hand, maybe the problem is texlive related (or, maybe, evince-texlive interaction related, since seems to not affect to xdvi)
The problem is that we don't know if we can trust your platform... so it put some doubt on most of your bug reports. It might be that a lot of GNOME developers will loose a huge amount of time trying to reproduce bugs that you are the only one to experiment... And we sincerely don't need it. In such case, I would _strongly_ advise you to stop reporting new bugs and to find the root of all Evil (aka the origin of your problem) and to fix it. Concerning the memtest, if you have some memory problem while a compilation, then it is always reproducible (as the binary will always be the same at any run).
I have just tried on a different computer (a Dell XPS1330, also x86_64) and it also crashes, in that case, I have launch evince, it seemed to show it, but, just while trying to move to next page (like in bug 563716) it crashed... I really doubt it's a hardware problem then I have just also tried on a remote i686 system with exactly the same crash
I'll try on an amd 64 box at home. But, this seems extremely strange to me...
I couldn't make it crash again on my AMD64 box.
Created attachment 149135 [details] fig12.eps Interesting, trying on my XPS in situ (I mean, not connecting remotely), evince doesn't crash, but it's unable to show any image (eps file) in the document and, while, on my first system I am able to open every eps file individually, on my laptop, I am unable to open files like attached one (also provided in original tarball) When trying to open attached eps file evince keeps "Loading" and I get: $ evince fig12.eps rangecheck -15 rangecheck -15 rangecheck -15 rangecheck -15 ** (evince:22718): WARNING **: Error rendering thumbnail in terminal
Umm, I have seen that, after updating from ghostscript-gpl-8.64 to 8.70 (like in first and "crashing" system) I get exactly the same behavior with both systems: evince is able to open all eps files but crashes when trying to open dvi file
*** Bug 606047 has been marked as a duplicate of this bug. ***
*** Bug 616366 has been marked as a duplicate of this bug. ***
Created attachment 167029 [details] [review] check "glyph->data" for emptiness Hi, the backtrace in the comment #9 is similar to the one in https://bugzilla.redhat.com/show_bug.cgi?id=555113 which can be triggered by the dvi attached in https://bugzilla.redhat.com/show_bug.cgi?id=555135. It is reproducible in Fedora 12 where evince is not compiled with --enable-t1lib and if package texlive-utils (binary mktexpk) is not installed. It segfaults because MDVI sets data pointer to 0x1 (MDVI_GLYPH_EMPTY constant) for empty glyphs and doesn't check it before use sometimes. Attached patch fixes the segfault for me. Regards Marek
Pushed to git master, thanks Marek!
Still valid after applying http://git.gnome.org/browse/evince/patch/?id=9f258bd6009e1ae92579cf159d81ad8c28280666 to 2.30.3 :-( Distribution: Gentoo Base System release 1.12.13 Gnome Release: 2.30.2 2010-07-24 (Gentoo) BugBuddy Version: 2.30.0 System: Linux 2.6.35-ck-r2 #1 PREEMPT Wed Sep 1 16:05:23 CEST 2010 x86_64 X Vendor: The X.Org Foundation X Vendor Release: 10707000 Selinux: No Accessibility: Disabled GTK+ Theme: Shiki-Gentoo Icon Theme: gnome-gentoo GTK+ Modules: canberra-gtk-module, gnomebreakpad Memory status: size: 420278272 vsize: 420278272 resident: 29724672 share: 13570048 rss: 29724672 rss_rlim: 18446744073709551615 CPU usage: start_time: 1283359369 rtime: 25 utime: 24 stime: 1 cutime:0 cstime: 0 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/bin/evince' [?1034h[Thread debugging using libthread_db enabled] [New Thread 0x7f0a069b5710 (LWP 14451)] [New Thread 0x7f0a071b6710 (LWP 14450)] 0x000000327820dc54 in __lll_lock_wait () from /lib/libpthread.so.0
+ Trace 223515
Thread 1 (Thread 0x7f0a0cb3f8c0 (LWP 14449))
Inferior 1 [process 14449] will be detached. Quit anyway? (y or n) [answered Y; input not from terminal] ----------- .xsession-errors (48 sec old) --------------------- (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed (gnome-terminal:7928): Gtk-CRITICAL **: gtk_widget_is_ancestor: assertion `ancestor != NULL' failed --------------------------------------------------
Pacho, I can't reproduce the crash either. Unfortunately, it's very difficult to evince devs to help you with the problem if they can't reproduce it. One question though. Since we are close to the release of 2.32, Could you check if you can reproduce in evince 2.31.6 or later?
The most strange thing is that this has been always difficult to reproduce depending on running distributions :-/, but I have always been unable if the crash is because evince, ghostscript... About testing 2.31... sadly I won't be able to test it in the next days, but will try it in the future for sure :-| Sorry for not being able to provide that info
Review of attachment 167029 [details] [review]: Updating patch status since Carlos already committed it.
Pacho, do you still have issues with newer versions of evince?
The problem is a mix between permissions for the fonts directory and the use of cmr10 instead. See: kpathsea: Running mktexpk --mfmode / --bdpi 600 --mag 1+120/600 --dpi 720 ptmb8r mkdir: cannot create directory `././var/cache/fonts/pk': Permission denied mktexpk: /usr/share/texmf/web2c/mktexdir /var/cache/fonts/pk/modeless/adobe/times failed. page: Warning: font `ptmb8r' at 720x720 not found, trying `cmr10' instead So, I changed the permissions of my $HOME/.textmf-var/fonts to 600 and I was able to reproduce the issue with evince 2.30.3. Here is the backtrace:
+ Trace 232053
It seems an issue either in mdvi or cairo. However, there is no crash in newer versions (3.4.0 nor master) with the permissions changed. So, I am closing this bug as obsolete. If you are able to reproduce this bug with a newer evince (>= 3.8), please feel free to reopen this bug.