[linux-dvb] Re: Latest CVS driver and CA recordings

[Klaus Schmidinger <Klaus.Schmidinger@cadsoft.de>]

Ralph Metzler wrote:
> 
> Klaus Schmidinger writes:
>  > I was hoping to be able to finalize the NAPI adaptation of VDR
>  > this weekend, but there are apparently still some severe problems
>  > in the latest CVS driver with encrypted channels.
>  >
>  > The biggest problems are with the ORF channels. Sometimes I don't get
>  > a picture on the video output of the DVB card (no matter whether I
>  > do a recording or just tune to that channel). When I do a recording
>  > it sometimes works fine, on other occasions the data coming from the
>  > Transport Stream contains video and audio packets (0xE0 and 0xC0),
>  > but inside these packets there are no "00 00 01" markers (there _is_
>  > a lot of data in the packets, but nothing that could be detected as
>  > "picture_header", "slice" or the like).
> 
> I could not find this error anymore on any of the encrypted channels
> I could test it with but this currently does not include ORF.
> Do you have this problem only on ORF or also still on other channels?

>From what I can see so far this particular effect is only with the two ORF
channels.

>  > Also, when switching between encrypted channels, I often get segfaults
>  > inside 'ioctl()'.
> 
> Well, a ksymoops dump would tell me a lot more than "it segfaults".
> Which ioctl() calls cause the segfault?

A backtrace in 'gdb' (applied to the resulting core dump after such a crash)
shows:

#0  0x401cd234 in __ioctl () from /lib/libc.so.6
#1  0x4 in ?? ()
#2  0x804abf3 in Switch__8cChannelP7cDvbApib (this=0x808ea88, DvbApi=0x0, Log=true) at config.c:278
#3  0x8068fc3 in main (argc=3, argv=0xbffff8ec) at vdr.c:368
#4  0x40124baf in __libc_start_main () from /lib/libc.so.6

If you could tell me how to do a 'ksymoops dump' I'd be glad to do so.

> I also don't understand how a channel being encrypted could make a
> difference inside an ioctl().
> Outside of the firmware there is no longer any difference between
> encrypted an unencrypted channels.
> If the firmware would make problems or crash in this case I could
> at least understand it.

I'm also wondering what the difference could be here. All I can say
so far is that apparently FTA channels work pretty well, while switching
between CA channels sometimes causes segfaults, and the ORF channels
sometimes don't deliver any useful data.

If you have any further suggestions what I could test, I'd greatly
appreciate them. Could this have to do with the way VDR switched channels
with the new API (maybe you could take a look into cDvbApi::SetChannel()
of VDR 0.80pre4)? However, there is no more 'CA` specific code in there,
since the new API doesn't need this, according to Dave Chapman, who implemented
this channel switching.

Klaus
-- 
_______________________________________________________________

Klaus Schmidinger                       Phone: +49-8635-6989-10
CadSoft Computer GmbH                   Fax:   +49-8635-6989-40
Hofmark 2                               Email:   kls@cadsoft.de
D-84568 Pleiskirchen, Germany           URL:     www.cadsoft.de
_______________________________________________________________


-- 
Info:
To unsubscribe send a mail to listar@linuxtv.org with "unsubscribe linux-dvb" as subject.