[linux-dvb] Re: The origin of the pts ?

[Peter Maersk-Moller <peter@maersk-moller.net>]

Hi Robert

Robert Schlabbach wrote:
> From: "Peter Maersk-Moller" <peter@maersk-moller.net>
> > ....
> TT *budget* cards deliver the stream unaltered. Only TT "premium" cards
> need to re-create parts of the TS.
Johannes wrote that the PTS part is relayed unchanged as-is as far
as I understood it.

> > The reason I'm asking is that every now and then I seem to get
> > a PTS of a frame belonging in the past. See the log below, where I
> > receive data with a TT Budget (NOVA-s), broadcast with dvbstream,
> > receive and decode with libmpeg2 (in mp4live).
> Since you have a budget card, there must be a bug somewhere in the chain -
> either the broadcaster is already delivering erroneous PTS, or maybe
> libmpeg2 is buggy.
To tell you the truth, since I have a mixture of premium, budget
and budget-ci, I'm no longer sure which I was using. I have to
try to test it again.

> > Assuming libmpeg2 is decoding correct (and that I use libmpeg2
> > correct, could the *PTS fro mthe past* be caused by loosing data
> > every now and then on tha satellite down-link ?
> Packet loss should lead to continuity count errors.
Not really.

> What are you using as
> the demultiplexer? Make sure it reports continuity count errors...
I'm using the TS-demuxer example code from mpeg2dec.c (libmpeg2)
and that seems to work fine.

Apparently I didn't detect a RTP (MPEG-2 TS) packet loss so I
don't think there was any packet loss between dvbstream and libmpeg2
decoding. Libmpeg2 seems to quite nicely handle lost packets, but I
don't think there was lost any packets. But you got a point.

I detect when I see a PTS that in this case every now and then (minutes
in-between) is 1-4 frames before the PTS of the previous decoded
frame. I my case I drop the frame, but apparently, I may fail
to recognise how many of the following frames have a PTS that
is before the last *not dropped frame*.

However I do detect if PTS is larger than previous PTS plus
one frame period, but that is not a problem here.

So far, for me it seems that every now and then, the PTS
(in display order) of a decoded frame is smaller than PTS
of previous frame. Not really sure where to go from here
except try to figure out if any event leads up the the
PTS being lower than the previous.

Kind regards

--PMM