After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 691961 - Race condition prevents EOG from displaying images with dimensions matching a specific range / the problem is HW dependent
Race condition prevents EOG from displaying images with dimensions matching a...
Status: RESOLVED DUPLICATE of bug 656224
Product: eog
Classification: Core
Component: general
3.4.x
Other Linux
: Normal normal
: ---
Assigned To: EOG Maintainers
EOG Maintainers
Depends on:
Blocks:
 
 
Reported: 2013-01-17 17:15 UTC by Jaromír Cápík
Modified: 2013-02-24 18:15 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Jaromír Cápík 2013-01-17 17:15:24 UTC
Hello.

I'm experiencing a problem with Eog (tested with 3.4.x
and 3.6.2).
It fails to display images with specific dimension range
on my workstation, whilst they're displayed correctly
on the same OS release running in a virtual machine
or different hardware. It seems to be a HW dependent
race condition ...
I tested 5 different image viewers and they display all three
images without problems.

Please, check the following link:

http://jcapik.fedorapeople.org/files/eog/Bug_827005/


All three test images were created by resizing
the same original image. Only the 5480x5480 sized
image can't be displayed correctly on my workstation.
The background is black during the image decoding,
but when Eog tries to finally display the image,
only a chessboard is shown. AFAIK the chessboard
is shown in case of transparent pixels. The strange
thing is, that the miniature shown when I try to drag
the chessboard with mouse is shown correctly.
The attached OGV videos demonstrate the issue.

As the problem is HW dependent, I'm willing to provide
you with any debug data you need in order to resolve
this issue. 

Please, let me know.

Thank you.

Best regards,
Jaromir.
Comment 1 Felix Riemann 2013-01-23 23:34:48 UTC
Hmm, this could be related to bug 656224

What gfx card and driver do you use?
I cannot reproduce it on my system.
Comment 2 Jaromír Cápík 2013-01-24 12:39:45 UTC
Hello Felix.

'lspci -nn' says:
-----------------
0f:00.0 VGA compatible controller [0300]: nVidia Corporation G96 [Quadro FX 380] [10de:0658] (rev a1)

kernel / module:
----------------
kernel-3.6.11-1.fc17.x86_64 / nouveau

Xorg driver:
------------
xorg-x11-drv-nouveau-0.0.16-37.20120306gitf5d1cd2.fc17.x86_64


xrandr says :
-------------
Screen 0: minimum 320 x 200, current 3840 x 1080, maximum 8192 x 8192
DVI-I-1 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 477mm x 268mm
DVI-I-2 connected 1920x1080+1920+0 (normal left inverted right x axis y axis) 477mm x 268mm



I've checked the Bug 656224 and see that Jasper Krijgsman experiences this issue only with dual screen. I'm gonna do some test here and let you know.
Comment 3 Jaromír Cápík 2013-01-24 12:46:08 UTC
I can confirm, that switching ANY of the two displays off helped.
Comment 4 Felix Riemann 2013-02-11 17:42:30 UTC
That's really strange!

Would it be possible for you to check if this exists with nvidia's binary driver as well?

If it doesn't nouveau developers should probably be involved.
Comment 5 Jaromír Cápík 2013-02-19 13:36:43 UTC
Hello Felix.

Sorry for the delay, I was sick.
Sure, I'm gonna test that.
The biggest question is, why this happens with eog only. Other viewers work well.

Regards,
Jaromir.
Comment 6 Jaromír Cápík 2013-02-19 15:28:24 UTC
The problem can't be reproduced with the proprietary driver. I tested twinview, separate X screens (with and without xinerama).
Comment 7 Felix Riemann 2013-02-24 18:15:53 UTC
Hi Jaromír

if I really knew what's the problem I would try to fix it. :)

As it works by exchanging drivers, my assumption is that we might possibly hit a texture size bug in the open driver that breaks compositing. Other viewers probably still do tiling when drawing an image to the screen, which we disabled around 3.0. Then again I tend to think that cairo should handle that case for eog.

But I guess only a driver developer can really tell what we are doing wrong.

Marking this as a duplicate now.

*** This bug has been marked as a duplicate of bug 656224 ***