[linux-dvb] Re: TS Continuity problem reason found for budget-cards

[Michael Hunold <hunold@linuxtv.org>]

Hi,

Am 12/01/04 11:19, Markus Schulz schrieb:
> i've build a ts-continuity check for one pid (hardcoded) in
> dvb_demux.c:dvb_dmx_swfilter_packets and find out that there are the 
> place where packets lost.
> this happens cause the ts-sync byte sometimes don't find at position 0 
> in buffer but at position 1 or (possible) other.
Interesting. Do you have more informations about the garbled TS packet?
If it starts at position 1, is it only 187 bytes long? Or is it 188 
bytes long (and valid) and the following packets are only shifted 
throughout the buffer?

> there we can see, that every supposedly lost packet is found in the 
> buffer at start position 1 instead of zero.
>
> but i don't know how this can be possible in vpeirq function 
> (budget-core.c) which reads the data from dma and calls the 
> dvb_dmx_swfilter_packets().
> are there some ts-packets larger then 188 bytes? 
No.

> or how can a single 
> byte be between two ts-pakets?
It's possible that it's a hardware bug inside the saa7146. Or some sort 
of wrap-around bug at the end of the buffer, that's not handled 
correctly by the dma. After all it's a dma engine for analog video, so 
perhaps there are some corner cases.

CU
Michael.