[MPlayer-users] Very slow play with gl default

[Reimar Döffinger]

On Sat, Feb 27, 2010 at 10:57:08AM -0700, Kevin DeKorte wrote:
> On 02/27/2010 10:45 AM, Reimar Döffinger wrote:
> > On Sat, Feb 27, 2010 at 10:42:51AM -0700, Kevin DeKorte wrote:
> >> On 02/27/2010 10:26 AM, Reimar Döffinger wrote:
> >>>> *** glibc detected *** mplayer: double free or corruption (out):
> >>>> 0x0a5c3f60 ***
> >>>> ======= Backtrace: =========
> >>>> /lib/libc.so.6[0x376751]
> >>>> /usr/lib/dri/r600_dri.so(+0x13482)[0xf692c482]
> >>>> /usr/lib/dri/r600_dri.so(+0x21370)[0xf693a370]
> >>>> /usr/lib/dri/r600_dri.so(+0xfcb3a)[0xf6a15b3a]
> >>>> mplayer(glSetupYUVConversion+0xcbc)[0x80a6368]
> >>>> [0x2]
> >>>
> >>> Looks like a bug in the fragment program compiler or something like that
> >>> to me (well, using mesa git you can hardly complain about bugs...).
> >>> I'd suggest trying -vo gl:yuv=0 - it should at least give you hardware scaling
> >>> and subtitle blending for free over x11 (though xv is probably still a better choice).
> >>
> 
> With the :yuv=0 option to gl I still get the BadMatch, the video does
> actually work and mplayer doesn't segfault. So that means a shader
> error? If so I'll open a bug on the radeon driver with a test case.

Most likely. Of course you can never be 100% sure until you actually fixed the bug :-)

> >> I think there might be a bug in mplayer as mplayer from the command line
> >> with -vo gl works fine, only fails when combined with the -wid option.
> >> So maybe that is why the BadMatch is emitted.
> > 
> > Well, it's possible. Or maybe the radeon driver can't handle rendering into
> > a window owned by a different process...
> 
> Then why does gl2 work fine? I will keep updating mesa and let you know
> if the problem gets fixed there.

gl2 does not have auto-detection for shader support.
-vo gl2:yuv=2 should show the same behaviour.