Bug 12619

Summary:	evince crashes with cairo_surface_destroy
Product:	cairo	Reporter:	Sebastien Bacher <seb128>
Component:	general	Assignee:	Carl Worth <cworth>
Status:	RESOLVED FIXED	QA Contact:	cairo-bugs mailing list <cairo-bugs>
Severity:	normal
Priority:	medium	CC:	aklitzing, burner, deshantm
Version:	1.4.10
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:

Description Sebastien Bacher 2007-09-29 10:42:54 UTC

The bug has been opened on https://bugs.launchpad.net/bugs/146860

"Binary package hint: evince

Evince crashes a few seconds after loading the following file:
  http://www.londontown.com/ALondonCalendar2007.pdf
with the following stdout:
~$ evince
/bin/sh: /usr/bin/esd: not found
evince: /build/buildd/libcairo-1.4.10/src/cairo-surface.c:394: cairo_surface_destroy: Assertion `surface->ref_count > 0' failed.
Aborted (core dumped)"

Comment 1 Michael R Head 2007-10-10 11:44:16 UTC

It also crashes around page 11 in this PDF: http://www.acm.org/acmqueue/digital/Queuevol5no6_September2007.pdf

Comment 2 Michael R Head 2007-10-10 11:47:53 UTC

BTW: I get the assertion failure on my 64-bit machine with a Xinerama setup on an nVidia card using the binary driver version 100.14.19.

On the 32-bit install on my laptop with the same nvidia driver version, with a single monitor, I don't get it.

Comment 3 Chris Wilson 2007-10-10 12:08:09 UTC

(In reply to comment #2)
> BTW: I get the assertion failure on my 64-bit machine with a Xinerama setup on
> an nVidia card using the binary driver version 100.14.19.
> 
> On the 32-bit install on my laptop with the same nvidia driver version, with a
> single monitor, I don't get it.

That might explain why I've failed to reproduce this bug with either the two pdfs. Using cairo 1.4.10 evince reports "cairo context error: out of memory" which is fixed with HEAD. I presume it was Behdad's fix to handle the return value of FT_Load_Glyph more intelligently which fixed that false error. Could you try git HEAD on your 64-bit machine? One of the numerous bug fixes which have been made since 1.4.10 is the switch to atomic reference counting, which seems the most relevant to the assertion.

Comment 4 Todd Deshane 2007-10-10 15:34:04 UTC

deshantm@gmail.com

Comment 5 André Klitzing 2007-11-08 13:46:20 UTC

Same problem with that PDF on amd64: http://www.bademeister.com/v7/pix/visions1107_opener.pdf

Comment 6 André Klitzing 2008-06-15 13:52:37 UTC

Seems to be fixed here. I have no problems anymore with that PDFs.

Cairo 1.6.4
Evince 2.22.2
Poppler 0.8.2

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.