[mythtv] DVBRec and hardware with limited filters?

Matthew McClement mythtv at macker.co.uk
Tue Jun 20 08:12:32 UTC 2006


Janne Grunau wrote:
> On Monday 19 June 2006 23:06, Matthew McClement wrote:
> 
>> [excessive use of pid filters in svn head]
> 
>> Looking over the SVN commit logs, it looks like previously the
>> AutoPID stuff filtered out the extra pids(data ones) so that you were
>> only left with the video and audio streams, but this was removed with
>> 9231.
> 
> You're on the wrong track. Most of the pids are PMT pids.

Fair enough.

>> Is the above breakage intentional
> 
> Listening on all PMT pids ought to speed up channel changing on the same 
> multiplex. PID filtering in software is a cheap operation. So this 
> behaviour is only bad with broken DVB device designs. Broken in this 
> respect means devices which aren't capable to deliver the complete 
> transport stream to the driver (mostly usb1.1 devices and 
> Siemens/Technotrend full featured cards).

While the hardware might be broken, I suspect this problem is likely to
crop up more. Most of the cheap USB sticks are based on the same
reference design with the same hardware limits, and, well, they're cheap.

Would it possibly be more elegant to catch the open error, check to see
if we managed to open any filters at all(in case the hardware is
completely broken), count them and if we don't already have the desired
pid in the list, reclear, and add from the desired pid onwards(looping
around the list if need be to fill in the extra slots)?

The above is completely off the top of my head and is done without
actually knowing what the various parts of DVBRecorder expect stream
wise after OpenFilterFd is done running. It also assumes that most cards
behave elegantly when they run out of pid filters and don't require a
driver reload to start working again.

>> or are there plans to work around this?
> 
> As dirty work around is attached to the ticket you missed.

Ta.

>> I'm surprised nobody has raised this issue before(unless I've 
>> missed something) but my searches have largely drawn a blank.
> 
> You missed it. See http://svn.mythtv.org/trac/ticket/1970.

My fault for not rechecking the tickets before sending the mail(I had
intended to send this 2 days ago but was missing the log snippet). Sorry.

Matt



More information about the mythtv-dev mailing list