Wolfgang Thiel wrote:
Many thanks to all of you trying to track down my problems with the SS2:-)Looking at the code I'd say that the driver supports 6 hardware PID filters. If you open more than 6 at a time, it should dump the whole stream, IMO the 7th PID should not be 8192 but another legal value. 8192 is caught nowhere in the code, I cannot say what effect it could have -- maybe it's just taken as another 0 PID.
My configuration:
- Debian Gnu/Linux running a 2.4.22 kernel from ftp.kernel.org.
- gcc 2.95.4 (Debian)
- one FF card (frontend: ves1x93: Alps-BSRV)
- one SS2 card rev. 2.6B (frontend: stv0229)
- drivers: dvb-kernel CVS 031116
- dvbstream from CVS Dvbtools 031114, patched to *not* close the
frontend if not requested from the command line (the current dvbstream
seems to have the same default now as well.) - scan and szap from from linuxtv-dvb-1.0.1.tar.gz
For my tests, I only loaded the modules needed for the SS2, *not* the
ones needed for the FF.
I didn't have much time yet, but so far I can tell:
- I found no way to get the full transport stream, no matter how many (valid)
pids dvbstream was supposed to stream, with the last being 8192
[I used 30w, 11811v, with its (radio) pids 256-...]
- With szap -r and then cat /dev/dvb/adapter0/dvr0 |mplayer - everything is fine. - But: scan -c is not able to set all relevant filters (the one on the PMT seems to work fine, but not the filters for the different PATs). At the same time, cat /dev/dvb/adapter0/dvr0 stops to deliver any data: I have to kill mplayer and szap; restarting szap and mplayer usually works, but not always; if it doesn't work, reloading the driver, some- times, doesn't work: dvb-core is used by some ghost and cannot be unloaded. Sometimes, this then even crashes the kernel: I only get some flashing CapsLock/ScrollLock leds on the keyboard, and that's it :-( [And no oops in the logs.] - Using pid 0x2000 for streaming seems to return the data of filters that other running programs have set already, but not all the streams, and not very reliably (first comes - first gets the data?).
See above: 0x2000 seems to be simply ignored.
I'll make more tests when I find the time. Unfortunately, I cannot reproduce the dvb-core unload problems/kernel crashes reliably.
Have seen that rarely, too. (Also 2.6B)
To solve filtering issue, maybe you could try to add _more_ than 6 different PIDs? This would indicate if (at least) the OpenWholeBandwidth subroutine works as expected.My conclusion so far is that there must be something wrong - at least - with the filter handling in the skystar2 driver. I don't know whether this is a problem of the frontend. So: Q: Is anybody with a SS2 having no problems? What frontend? I'll postpone my other problem for later [driver needs reloading after the tuner lost lock.]