[mythtv] Experimental pvr 350 output patch
john.timon at labatt.com
Tue Dec 30 10:23:55 EST 2003
Where does this stand?
is it committed to CVS or is it being held apart pending approval? Just
curious, I think I want to play Guniea Pig and try it out :)
Unix like TeePee no windows, no gates, Apache inside.
From: David Engel [mailto:dlengel at attbi.com]
Sent: Monday, December 15, 2003 5:13 PM
To: Development of mythtv
Subject: Re: [mythtv] Experimental pvr 350 output patch
FYI, if you have not tried this patch yet, now might be a good time. All
known issues have been addressed and the improved functionality is, IMHO,
well worth the effort.
On Tue, Dec 09, 2003 at 10:45:45PM -0600, David Engel wrote:
> Another day, another version. The major changes in this version are
> as follows.
After many trials and tribulations with the ivtv driver and trying to
understand the idiosyncracies of the itvc15 decoder with very little
documentation, here is the latest version of my pvr350 patch. The major
changes from the previous version (as best I remember) are as follows.
Moved some funtionality from the video thread to the decoder thread to avoid
race conditions while pausing and unpausing. The video thread now
exclusively controls the OSD aspects of the pvr350 and the decoder thread
exclusively controls the decoding aspects.
Finished seeking to exact frames with video update. It's kind of klugy
because I have to single step the decoder to the desired frame. As a result,
exact seeks other than one frame forward will look a little funny, but at
least it works. Since I still have concerns about the flakiness of the
decoder, debug output is still enabled for this code. As far as I can tell,
the only way to improve on this way of doing exact seeks is to do a more
complete parsing of the mpeg2 stream and only feeding the decoder exactly
what is needed.
Some driver issues could not be worked around from user space, so the pvr350
patch now requires an ivtv driver patch. The ivtv patch is also attached.
It fixes output polling and adds support for non-blocking output flushing
queued data. Once I get some feedback from others on how well this works, I
will address the issues with the ivtv developers.
Again, if you have a pvr350 and haven't tried this yet, please consider
doing so. It seems to work pretty well for me.
dlengel at attbi.com
More information about the mythtv-dev