Patrick Boettcher wrote: > On Tue, 21 Jun 2005, Felix Domke wrote: > >Patrick Boettcher wrote: > >>>So the section should IMHO never appeared into userspace. Sure it's > >>>right to be prepared for invalid sections (and drop them), but in this > >>>case, I suspect a driver/hardware bug. > >>>Does the hardware use hardware section filtering? > >>No, this section was received using a budget Card (AirStar2). Parsing > >>was done by the linux dvb demuxer. > >>I received a lot of valid (parsable) sections before. This faulty > >>section (and others) I got, when the signal went through a normal > >>loss/regain process. > >Of course there's always the 1:16 possibility to skip a multiple of 16 > >packets on that pid, so the CC will give a failse match. > > I see. > > >Does it happen all the time after a signal regain or only very few times? > > No. It happens rarely, but I can force it, by "playing around" with the > antenna. > > What would be your suggestion? A harder syntax check in the program > (correct table_id for this PID, section_syntax_indicator == 1) or > anything else? IMHO the software demux should throw away the complete section when it sees a TS packet with the error indicator set, and then wait for a good TS packet with PUSI set. Johannes
