Hermann Gausterer wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 hi klaus On Sunday 21 November 2004 11:57, Klaus Schmidinger wrote:After having this driver run for a whole night, my second and third DVB card (a budget WinTV NOVA-S and another FF Siemens DVB-S Rev. 1.3, respectively) apparently "died" at around 05:00 this morning. That's what I can see from the system load log here - normally these cards do EPG scans, which causes a low load, but as of 05:00 this morning that load was gone, and I wasn't able to get any more data from either of these cards (tested via Transfer mode). My primary card was still working this morning.
i am using linux-dvb.2003-11-08, with kernel 2.4.20; i get this similar behavior with a multicard-setup (3 * 1.3 dvb-s) on smp;
for me it looks like the primary card is always ok, but the additonal
cards get bad ~3-4h after a driver load, there is no data anymore
coming out of them. but after one minute (on a revording on such
a card) vdr restarts anyway with driver reload (in my setup) , so it does not "disturb" me any more; but sometimes the situation is even worse; there is still coming data out of the second/third card, but this data is corrupt, it does not render full screen, the sound is
disturbed, and the .vdr files are only 1/4 of the size for a recording the should have; but in this case, vdr does not notice this, and the recording is waste ...
did you never experienced this, klaus?
No, not that I remember.
have you tried to disable the epg scan? it was supposed that the big amount of channel swiches could disturb the firmware of the cards, but i could find that out ....
I doubt that this causes it, because it also happens on the primary card, and also very soon after a fresh start of driver and VDR.
this is not a bug on the hardware, because i switched the primary card in the vdrconfig, and the situation changed as expected, still the actual primary is ok, the rest not ...I'm getting the feeling that the dvb-kernel driver doesn't like me.my situation is even worse with dvb-kernel .. there was a report that the dvb-kernel driver for the 1.3 dvb-s is not smp-safe. and this is very true .... is this still so?? @developers (havent tried for month)Anyway, joerg@hakker.de suggested in a posting (that unfortunetaly broke the thread) to set pids_off=1 - and ever since I did this, I haven't had this problem any more :-)can you tell me, where such a switch is? i hearing that the first time :)
That's a load parameter to the dvb-ttpci.o driver file. But as I wrote, this didn't really fix it - it just happened to work for a little while... Klaus