[winswitch] testing required: OpenGL client rendering

Antoine Martin antoine at nagafix.co.uk
Mon Apr 15 08:07:19 BST 2013


On 04/07/2013 07:54 PM, Nick Burrett wrote:
> I see the problem with any attached session, be that emacs,
> gnome-terminal, thunderbird etc.  It's 100% reliable.  I do not seem to
> have a session->info window.
The session-info window is in the system-tray applet menu. Some DEs,
like gnome3, make it very hard to use.

I have failed to reproduce the problem with any of the applications you
listed, which makes me think that something else is at play.
Which DE are you using? (a compositing window manager perhaps?)
And which distro / version / etc? (will try it in a VM)

I assume that the problem goes away if you change the encoding from the
system tray to a non video encoding (not vpx or x264)?

> nickb at nickb-pc:~/src/xpra/src$ dpkg -l | grep fglrx
> ii  fglrx                                       2:9.000-0ubuntu3        
>                                 amd64        Video driver for the AMD
> graphics accelerators
> ii  fglrx-amdcccle                              2:9.000-0ubuntu3        
>                                 amd64        Catalyst Control Center for
> the AMD graphics accelerators
$ rpm -qa | grep -i catalyst | grep x86_64
xorg-x11-drv-catalyst-libs-13.1-1.fc18.x86_64
xorg-x11-drv-catalyst-13.1-1.fc18.x86_64
akmod-catalyst-13.1-2.fc18.x86_64

> nickb at nickb-pc:~/src/xpra/src$ fglrxinfo 
> display: :0  screen: 0
> OpenGL vendor string: Advanced Micro Devices, Inc.
> OpenGL renderer string: ATI Radeon HD 5670
> OpenGL version string: 4.2.11903 Compatibility Profile Context
$ fglrxinfo
display: :0  screen: 0
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: ATI Radeon HD 5450
OpenGL version string: 4.2.12002 Compatibility Profile Context 9.002

> However I do not see the issue when I create a local xpra session and
> access that locally.  Though if I try to start a xpra local server with
> --no-mmap, I currently get this error
Do you mean that the server fails to start if you use "--no-mmap"?

> 2013-04-07 13:51:48,582 pulseaudio server started with pid 2586
> 2013-04-07 13:51:48,589 xpra server version 0.9.0
> E: [pulseaudio] pid.c: Daemon already running.
> /usr/local/lib/python2.7/dist-packages/xpra/server_base.py:297:
> GtkWarning: IA__gdk_error_trap_pop: assertion `gdk_error_traps != NULL'
> failed
>   gtk.main()
^ this is the only real error, and unfortunately it is a gdk internal
error and those often get reported against the main loop invocation
point, which is not helpful for figuring out the real cause..

This will produce *a lot* mode debugging info, which may help:
XPRA_X11_DEBUG=1 xpra ...

> 2013-04-07 13:51:48,777 xpra is ready.
> 
> (emacs:2589): Gdk-CRITICAL **: IA__gdk_error_trap_pop: assertion
> `gdk_error_traps != NULL' failed
^ same again...

(snip)

Cheers
Antoine


> 
> 
> On 7 April 2013 12:56, Antoine Martin <antoine at nagafix.co.uk
> <mailto:antoine at nagafix.co.uk>> wrote:
> 
>     Hi Nick,
> 
>     Can you tell us what graphics driver you are using (as shown by "session
>     info" window preferably)?
>     And what application triggers the problem most reliably.
> 
>     Thanks
>     Antoine
> 
>     On 07/04/13 12:52, Nick Burrett wrote:
>     > With changeset 3053, the visual artefacts remain.  When your suggested
>     > change to gl_check.py, I appear to be viewing the penultimate
>     window update
>     > e.g. if I click on "File", then the background of "File" changes
>     colour,
>     > but the actual File menu does not appear until I move the mouse to the
>     > position of the first menu entry, which forces a background colour
>     change
>     > for that menu entry and the whole menu gets drawn.
>     >
>     > Regards,
>     >
>     > Nick.
>     >
>     > On 7 April 2013 12:33, Antoine Martin <antoine at nagafix.co.uk
>     <mailto:antoine at nagafix.co.uk>> wrote:
>     >
>     >> Hi Nick,
>     >>
>     >> Many thanks for the feedback.
>     >> The error with menu items should be fixed by this change:
>     >> https://www.xpra.org/trac/changeset/3053/xpra
>     >>
>     >> If you still get visual artefacts, you may want to try this patch:
>     >>
>     >> --- src/xpra/gl/gl_check.py    (revision 3046)
>     >> +++ src/xpra/gl/gl_check.py    (working copy)
>     >> @@ -22,7 +22,7 @@
>     >>  def get_DISPLAY_MODE():
>     >>      import gtk.gdkgl
>     >>      #return  gtk.gdkgl.MODE_RGB | gtk.gdkgl.MODE_DEPTH |
>     >> gtk.gdkgl.MODE_DOUBLE
>     >> -    return  gtk.gdkgl.MODE_RGB | gtk.gdkgl.MODE_DOUBLE
>     >> +    return  gtk.gdkgl.MODE_RGB
>     >>
>     >>  #by default, we raise an ImportError as soon as we find
>     something missing:
>     >>  def raise_error(msg):
>     >>
>     >> Cheers
>     >> Antoine
>     >>
>     >>
>     >> On 07/04/13 09:31, Nick Burrett wrote:
>     >>> Sorry, no, doing that just broken OpenGL-accelerate.  So it
>     appears that
>     >>> with OpenGL enabled, the double-buffering is often the previous
>     buffer
>     >> when
>     >>> moving through menus or doing any sort of window updates.
>     >>>
>     >>> On 7 April 2013 09:26, Nick Burrett <nick at sqrt.co.uk
>     <mailto:nick at sqrt.co.uk>> wrote:
>     >>>
>     >>>> I fixed this by upgrading the Python OpenGL-accelerate module:
>     >>>>
>     >>>> easy_install OpenGL-accelerate
>     >>>>
>     >>>>
>     >>>> On 7 April 2013 09:07, Nick Burrett <nick at sqrt.co.uk
>     <mailto:nick at sqrt.co.uk>> wrote:
>     >>>>
>     >>>>> I'm trying this out with the xfce4 terminal - /usr/bin/Terminal on
>     >> Ubuntu
>     >>>>> quantal (12.10)
>     >>>>>
>     >>>>> Rendering is certainly faster, however just moving the mouse
>     across the
>     >>>>> various menu items at the top of the window will flick between the
>     >> current
>     >>>>> image and a previous window image and will regularly produce this:
>     >>>>>
>     >>>>> 2013-04-07 09:03:51,306 OpenGL Version: 4.2.11903
>     Compatibility Profile
>     >>>>> Context
>     >>>>> 2013-04-07 09:03:51,333 GL Extension GL_ARB_shader_objects
>     available
>     >>>>> 2013-04-07 09:03:51,353 GL Extension GL_ARB_fragment_program
>     available
>     >>>>> 2013-04-07 09:03:51,353 GL Extension GL_ARB_texture_rectangle
>     available
>     >>>>> 2013-04-07 09:03:51,354 GL Extension GL_ARB_vertex_program
>     available
>     >>>>> 2013-04-07 09:04:00,139 Attached to ssh:devbox:100 (press
>     Control-C to
>     >>>>> detach)
>     >>>>>
>     >>>>> 2013-04-07 09:05:00,204 do_paint_rgb24 error
>     >>>>> Traceback (most recent call last):
>     >>>>>   File
>     "/usr/local/lib/python2.7/dist-packages/xpra/window_backing.py",
>     >>>>> line 371, in do_paint_rgb24
>     >>>>>     self._do_paint_rgb24(img_data, x, y, width, height, rowstride,
>     >>>>> options, callbacks)
>     >>>>>   File
>     >>>>>
>     "/usr/local/lib/python2.7/dist-packages/xpra/gl/gl_window_backing.py",
>     >> line
>     >>>>> 181, in _do_paint_rgb24
>     >>>>>     self.gl_end(drawable)
>     >>>>>   File
>     >>>>>
>     "/usr/local/lib/python2.7/dist-packages/xpra/gl/gl_window_backing.py",
>     >> line
>     >>>>> 127, in gl_end
>     >>>>>     if drawable.is_double_buffered():
>     >>>>> AttributeError: 'NoneType' object has no attribute
>     'is_double_buffered'
>     >>>>>
>     >>>>>
>     >>>>> On 16 March 2013 14:35, Antoine Martin <antoine at nagafix.co.uk
>     <mailto:antoine at nagafix.co.uk>> wrote:
>     >>>>>
>     >>>>>> Hi,
>     >>>>>>
>     >>>>>> The OpenGL client rendering has been enabled again, and this
>     time it
>     >>>>>> looks like it should work in almost all cases (the "nouveau"
>     driver on
>     >>>>>> Linux is blacklisted however - and OSX is unlikely to be
>     supported).
>     >>>>>> Please give it a whirl and report any display visual
>     artifacts / bugs
>     >> so
>     >>>>>> that the 0.9.0 release can include it enabled by default,
>     more details
>     >>>>>> here:
>     >>>>>> http://xpra.org/trac/wiki/ClientRendering#OpenGLNotes
>     >>>>>> Updated beta packages here:
>     >>>>>> http://xpra.org/beta/
>     >>>>>>
>     >>>>>> Thanks
>     >>>>>> Antoine
>     >>>>>> _______________________________________________
>     >>>>>> shifter-users mailing list
>     >>>>>> shifter-users at lists.devloop.org.uk
>     <mailto:shifter-users at lists.devloop.org.uk>
>     >>>>>> http://lists.devloop.org.uk/mailman/listinfo/shifter-users
>     >>>>>>
>     >>
> 
> 




More information about the shifter-users mailing list