vdr-bounces@linuxtv.org wrote:

I got something I desperately need help with: I have got one FF card plus 3 Nova CI Budget Cards, usually using the latest driver, firmware and VDR versions. After having upgraded to Kernel 2.6.13, I get the dreaded "Video Data Stream Broken" error as soon as a recording starts. Vanilla VDR 1.3.32 (no plugins besides remote) with "WAIT_FOR_TUNER_LOCK" reports "no dvb card available for recording" instead:

Sep 20 17:40:44 [vdr] timer 5 (1 1740-2040 '@TITLE EPISODE') start Sep 20 17:40:44 [vdr] record /video0/@Brisant_Boulevardmagazin/2005-09-20.17.40.10.01.rec Sep 20 17:40:49 [vdr] ERROR: device 2 has no lock, can't attach receiver! Sep 20 17:40:49 [vdr] ERROR: Keine freie DVB-Karte zum Aufnehmen!

The log reports 4 video devices found. When switching back to 2.6.12 everything works fine again (same driver, same firmware)...no problem recording from 4 different transponders... Of course I also tried the latest (yesterday's) CVS driver and the latest Firmware (fb2620).

Same to me (but 2xFF 2.1 + 1x Nova-s). I ended with putting out the budget card and run the system with 2xFF cards only. My observation was that if the PC was just booted the budget card worked (NVRAM-Wakeup) but when there are several recordings going on and the budget was used later (e.g. uptime 1h) vdr reported no lock. I tried several kernels and different cvs versions of the dvb-drivers and different firmwares but no go.

...

From looking at dmesg I can tell the cards' IRQ assignments have changed,

but still each card uses its own one. Also, they still share their IRQs with the very same other PCI devices. szap seems to be able to tune, though, as it reports FE_HAS_LOCK on all 4 cards when tuning to "Das Erste".

Any hints on how to debug this further?

Might be a powerconsumption problem, are you sure the power supply is strong enough to serve all your cards ?

Thanks, Stephan

/hgm.bg

Hi Oliver,

Oliver Endriss wrote:

There are several versions of Nova-type cards. Please provide the sub-system ids (-> "lspci -vn").

Kernel 2.6.12 reports:

0000:02:0a.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 22 Memory at fbfff800 (32-bit, non-prefetchable) [size=512]

0000:02:0b.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:0000 Flags: bus master, medium devsel, latency 64, IRQ 23 Memory at fbfff400 (32-bit, non-prefetchable) [size=512]

0000:02:0c.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 20 Memory at fbfff000 (32-bit, non-prefetchable) [size=512]

0000:02:0d.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 21 Memory at fbffec00 (32-bit, non-prefetchable) [size=512]

Kernel 2.6.13:

0000:02:0a.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 16 Memory at fbfff800 (32-bit, non-prefetchable) [size=512]

0000:02:0b.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:0000 Flags: bus master, medium devsel, latency 64, IRQ 18 Memory at fbfff400 (32-bit, non-prefetchable) [size=512]

0000:02:0c.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 23 Memory at fbfff000 (32-bit, non-prefetchable) [size=512]

0000:02:0d.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f Flags: bus master, medium devsel, latency 64, IRQ 22 Memory at fbffec00 (32-bit, non-prefetchable) [size=512]

...

Of course I also tried the latest (yesterday's) CVS driver and the latest Firmware (fb2620).

This firmware is not used by budget cards.

Yes I know that, but still some people had reported similar problems, allegedly due to different driver/firmware combinations...

Try to find out which change in the driver caused this problem...

There is no change in the driver whatsoever, unfortunately ;-) The interesting thing is that the driver version does not seem to matter at all... It simply works with Kernel 2.6.12, but doesn't with 2.6.13. I tried the built-in drivers as well as CVS from 28.08.05 and 19.09.05 respectively. Always same result. 2.6.12 -> no problem, 2.6.13 -> VDSB :-/

It might be better to report driver bugs on the linux-dvb mailing list.

Yes, sorry, it's been quite a while since I've been subscribed to either list.... So far, VDR Portal had all the answers I needed... Next time I'll try to address the correct list ;-)

Oliver

Thanks, Stephan

Stephan Schreiber wrote:

...

There are several versions of Nova-type cards. Please provide the sub-system ids (-> "lspci -vn").

... 0000:02:0a.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f

0000:02:0c.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f

0000:02:0d.0 Class 0480: 1131:7146 (rev 01) Subsystem: 13c2:100f

Hm - sub-system 13c2:100f.

There is a thread on the driver mailing list about tuning problems with exactly this type of card.

Just a wild guess, but it might help: Try to disable 'enhanced_tuning' in budget-ci.c (line 711).

Replace .enhanced_tuning = 1, by .enhanced_tuning = 0,

Oliver

Stephan Schreiber wrote: ...

I have got one FF card plus 3 Nova CI Budget Cards, usually using the latest driver, firmware and VDR versions. After having upgraded to Kernel 2.6.13, I get the dreaded "Video Data Stream Broken" error as soon as a recording starts.

...

When switching back to 2.6.12 everything works fine again (same driver, same firmware)...no problem recording from 4 different transponders...

Has this problem been solved in the meantime?

I have just received SuSE 10.0 and I am wondering if I should install it on my VDR system, which also has one FF card and two budget cards.

Carsten.