Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 385433 - [gnome-overlay] =media-libs/clutter-gst-1.3.14: sandbox violations (remove and __xmknod)
Summary: [gnome-overlay] =media-libs/clutter-gst-1.3.14: sandbox violations (remove an...
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: [OLD] GNOME (show other bugs)
Hardware: All Linux
: Normal normal (vote)
Assignee: Gentoo Linux Gnome Desktop Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-10-02 23:04 UTC by Stefan Zwanenburg
Modified: 2011-10-15 22:59 UTC (History)
0 users

See Also:
Package list:
Runtime testing required: ---


Attachments
build.log for =media-libs/clutter-gst-1.3.14 (build.log,48.35 KB, text/plain)
2011-10-02 23:05 UTC, Stefan Zwanenburg
Details
emerge --info =media-libs/clutter-gst-1.3.14 (emerge-info,4.77 KB, text/plain)
2011-10-02 23:12 UTC, Stefan Zwanenburg
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Stefan Zwanenburg 2011-10-02 23:04:37 UTC
When emerging =media-libs/clutter-gst-1.3.14 (in my case as a result of updating gnome), several sandbox violations occur, all trying to remove or make a device node for my graphics card (for which I have installed ati-drivers).

Reproducible: Always

Steps to Reproduce:
1.Emerge =media-libs/clutter-gst-1.3.14
2.
3.
Actual Results:  
Sandbox violations (see attachment)

Expected Results:  
Successful merging of clutter-gst.
Comment 1 Stefan Zwanenburg 2011-10-02 23:05:27 UTC
Created attachment 288613 [details]
build.log for =media-libs/clutter-gst-1.3.14
Comment 2 Stefan Zwanenburg 2011-10-02 23:12:25 UTC
Created attachment 288615 [details]
emerge --info =media-libs/clutter-gst-1.3.14
Comment 3 Stefan Zwanenburg 2011-10-02 23:12:52 UTC
I also just tried it with =clutter-gst-1.3.12: same problem!
Comment 4 Alexandre Rostovtsev (RETIRED) gentoo-dev 2011-10-03 00:12:23 UTC
This is yet another variation on the theme of bugs #360219, #360073, #363917.

I believe the following ugly hack should put an end to this issue: deliberately set a known invalid DISPLAY variable.

In other words, adding DISPLAY="999invalid" to src_compile appears to resolve all sandbox violations (lots of error messages result, though), making the addpredict unnecessary.

This should be fixed in the clutter-gst-1.4.0 ebuild just added to the gnome overlay.
Comment 5 Stefan Zwanenburg 2011-10-03 11:15:13 UTC
I can confirm that =clutter-gst-1.4.0 does emerge successfully now. Thanks!
Comment 6 Nirbheek Chauhan (RETIRED) gentoo-dev 2011-10-04 10:50:41 UTC
(In reply to comment #4)
> This is yet another variation on the theme of bugs #360219, #360073, #363917.
> 
> I believe the following ugly hack should put an end to this issue: deliberately
> set a known invalid DISPLAY variable.
> 
> In other words, adding DISPLAY="999invalid" to src_compile appears to resolve
> all sandbox violations (lots of error messages result, though), making the
> addpredict unnecessary.
> 
> This should be fixed in the clutter-gst-1.4.0 ebuild just added to the gnome
> overlay.

Instead of working around the issue, why don't we ask upstream why they're trying to mknod in /dev during make install? That's a phenomenally stupid thing for a random package to do.
Comment 7 Alexandre Rostovtsev (RETIRED) gentoo-dev 2011-10-04 17:08:14 UTC
(In reply to comment #6)
> Instead of working around the issue, why don't we ask upstream why they're
> trying to mknod in /dev during make install? That's a phenomenally stupid thing
> for a random package to do.

Obviously upstream is not trying to mknod anything. The problem rather is that g-ir-scanner generates, compiles, executes, and immediately deletes some sort of temporary binary, and when this binary runs, as a side effect it tries to open an opengl context.
Comment 8 Nirbheek Chauhan (RETIRED) gentoo-dev 2011-10-07 08:37:46 UTC
(In reply to comment #7)
> Obviously upstream is not trying to mknod anything. The problem rather is that
> g-ir-scanner generates, compiles, executes, and immediately deletes some sort
> of temporary binary, and when this binary runs, as a side effect it tries to
> open an opengl context.

Yes, and in this case by 'upstream' I mean clutter-gst. Because I see no reason for gstreamer to require OpenGL access, unless they're doing video pipelining with GL, which is stupid.
Comment 9 Alexandre Rostovtsev (RETIRED) gentoo-dev 2011-10-07 08:44:17 UTC
(In reply to comment #8)
> Yes, and in this case by 'upstream' I mean clutter-gst. Because I see no reason
> for gstreamer to require OpenGL access, unless they're doing video pipelining
> with GL, which is stupid.

http://www.clutter-project.org/ : "Clutter uses OpenGL for rendering (and optionally OpenGL|ES for use on mobile and embedded platforms), but wraps an easy to use, efficient, flexible API around GL's complexity."
Comment 10 Nirbheek Chauhan (RETIRED) gentoo-dev 2011-10-07 12:39:42 UTC
(In reply to comment #9)
> (In reply to comment #8)
> > Yes, and in this case by 'upstream' I mean clutter-gst. Because I see no reason
> > for gstreamer to require OpenGL access, unless they're doing video pipelining
> > with GL, which is stupid.
> 
> http://www.clutter-project.org/ : "Clutter uses OpenGL for rendering (and
> optionally OpenGL|ES for use on mobile and embedded platforms), but wraps an
> easy to use, efficient, flexible API around GL's complexity."

Thank you for pasting the description of the package I maintain...

It's obvious that you did not understand my comment. Why the heck would clutter-gst need OpenGL for *introspection*? Why would it do GL video pipelining during /compilation/? That's utterly retarded.
Comment 11 Pacho Ramos gentoo-dev 2011-10-15 22:59:45 UTC
+*clutter-gst-1.4.2 (15 Oct 2011)
+
+  15 Oct 2011; Pacho Ramos <pacho@gentoo.org> -clutter-gst-1.3.12.ebuild,
+  -clutter-gst-1.3.14.ebuild, +clutter-gst-1.4.2.ebuild:
+  Version bump, remove development versions.
+