[mythtv-users] Long delays with remote occasionally

[Josh Mastronarde]

On Sun, Mar 29, 2009 at 9:41 AM, lj <larry at foxgulch.com> wrote:
>
> Hi Josh and list,
>
> I tried 1)  mythfrontend -v important,libav > log   and followed up by
> viewing an HD ATSC recording during which I experienced one remote ir
> button press delay.   The recording was made using one of my ATSC tuner
> cards.  Here's the output of 3)  grep "Input #" log on the HD recording:
>
> 2009-03-29 10:20:17.845 Input #0, mpeg, from
> '/var/lib/mythtv/recordings/2182_20090329070000.mpg':
> 2009-03-29 10:20:19.380 Input #0, mpeg, from
> '/var/lib/mythtv/recordings/2182_20090329070000.mpg':
> 2009-03-29 10:20:22.576 Input #0, mpegts, from
> '/var/lib/mythtv/recordings/1131_20090329070000.mpg':
> 2009-03-29 10:20:23.334 Input #0, mpegts, from
> '/var/lib/mythtv/recordings/1081_20090329070000.mpg':
> 2009-03-29 10:20:26.159 Input #0, mpegts, from
> '/var/lib/mythtv/recordings/1081_20090328220000.mpg':

<snip>

> When I watch a recording made using my s-video input to the WinTV
> HVR-1600 mpeg encoder,  3)  grep "Input #" log  shows:
> 2009-03-29 10:34:43.701 Input #0, mpeg, from
> '/var/lib/mythtv/recordings/2182_20090329070000.mpg':
> 2009-03-29 10:34:44.809 Input #0, mpeg, from
> '/var/lib/mythtv/recordings/2182_20090329070000.mpg':
> 2009-03-29 10:34:49.903 Input #0, mpeg, from
> '/var/lib/mythtv/recordings/2182_20090329070000.mpg':

Yes, this appears to be a problem -- note how it's probing the same
file, first with "mpeg", then with "mpegts", but then it keeps probing
over and over.  And later when you play an "mpeg" file, it keeps
probing that over and over.  In my system (with the mpeg probe fix), I
only see one "Input #" line for each show I watch (and it's correct
the first time for mpegts shows), no matter how long or how much I
skip around in that show.

Another thing to check -- do the same verbose logging with a fresh
mythfrontend, but this time only view the HVR-160 show (don't even let
it preview other shows; disable that if necessary).  In that case, do
you see just one "Input#0, mpeg" line?

I strongly suspect the root cause is the mpeg probing giving too high
of a score, compounded by some code later on that's trying to be
fault-tolerant (by trying again when there's a failure later -- many
of the probe procedures automatically lower their score each time
they're called, which is why mpegts sort-of succeeds the third time).
However, *that* code seems to get stuck in a loop of re-probing in the
background, which is what I think causes the other weirdnesses (like
the lirc response delays).  Insidious, because it appears to be
working when it's not quite right.

I'm going to file my ticket now with patch for 0.21-fixes, but I don't
know if it'll be accepted or how long it'd take to propagate to your
repo.  If you send me your "mythfrontend --version" output, I can try
to compile a new "libmythavformat.o" for you to manually try to drop
in, but no guarantees if that will work, and you'd have to trust my
motives :-)

Josh