[mythtv] mythtv-vid branch needs testing

Mark Kendall mark.kendall at gmail.com
Fri Aug 17 12:17:02 UTC 2007


On 8/17/07, Markus Schulz <msc at antzsystem.de> wrote:
> Am Freitag, 3. August 2007 schrieb Markus Schulz:
> xorg crashs approx. every 5 start or stop of mythfrontend (start ==
> switch from menu to play-tv, stop==abort play-tv to menu). really
> annoying, i had to write this mail twice :(
>
> *****************
> Backtrace:
> 0: /usr/bin/X(xf86SigHandler+0x81) [0x80c8631]
> 1: [0xb7eee420]
>
> Fatal server error:
> Caught signal 11.  Server aborting
> *****************

Markus - is there any chance of getting a more detailed backtrace?

I've dusted off some old kit with a different distro and managed to
replicate the crashes Doug has been seeing (haven't found the fix
yet...) - it would be nice to know there aren't other gremlins in the
works.

> - OSD fading is choppy

I suspect your card will struggle to handle osd fade at your given
frame rate, resolutions and choice of hw deinterlacer.

> - PIP window has wrong position, size and aspect ratio (but there are
> similar problems in trunk)

I'm aware of this and the fix is straightforward enough.

> - mythtvosd is not working correct. On CRT video output stops completly
> during osd output, on DFP is really really slow.

I've not seen this (I don't use it) but suspect it's the same issue as
the choppy osd fade. I'll check tonight.

> - Kernel (2x HW) is not really as smooth as expected (compared to trunk
> with my own full framerate patchset) but smoother then 25fps video.

I'm not 100% satisfied with the quality of the hardware kerneldeint
but you may also find (again) that this is a performance issue. hw
kerneldeint is by far the most hardware intensive - at 2x a 6600 may
be struggling to keep up. How does linearblend (2x HW) look?

> Why only the HW Deints can be used with full framerate?

because the framework isn't in place to handle it yet. Funnily enough,
I had another look at your patch the other day...

> - sometimes mythfrontend crashes (segfault) but xorg survives.

As above, some backtraces here would be extremely helpful.

Thanks and regards

Mark


More information about the mythtv-dev mailing list