[mythtv] Ticket #1887: invalid PAT in recordings causes Frontend Crash, Backtrace Attached...

[Janne Grunau]

On Thursday 01 June 2006 15:35, Daniel Kristjansson wrote:
> On Thu, 2006-06-01 at 13:29 +0200, Janne Grunau wrote:
>
> I just looked at those logs. There are *TWO* PAT's in the source
> stream.

Haven't seen the logs before.

> > But an error in the transmitted stream seems at least equally
> > unlikely. And the zero pid in the PAT is maybe an initialized
> > value. But I should look at the code before making wild guesses.
>
> Multiple PAT's is certainly not something we handle at the moment;
> I don't think we even handle multiple section PATs. But two PAT's
> with two different TSID's was definately not anticipated. I don't
> see this being banned in the MPEG standard, I don't know about DVB.

I haven't found statement in the DVB standard that explicitly forbidds 
multiple PAT with differen tranport stream ids in one transport. But 
it's unlikely that earlier versions intended this usage. The two last 
parts of the standard (DVB-H and DVB-S2) explicitly defines multiple 
streams on one physical transport. If my understanding is correct the 
hardware is supposed to select the transports so the only necessary 
change for applications is the change in the tuning data.

> Now are the MPEG program numbers always going to be unique across
> multiple PAT's or should we really be keyed to the TSID + Program
> Number, rather than just the Program Number?

I'm in favor of hoping that the program numbers are different. PATs with 
different TSIDs in one stream are probably rather uncommen.

Janne