vdr 1.2.6 - cyclic restarting of vdr when recording

[Tobias Grimm]

Stefan-W. Hahn wrote:

>after upgrading from c't VDR to debian main, I have the problem that after
>starting a recording the vdr is doing cyclic restarts after some seconds of
>recording. There are no messages in the log.
>I run a full featured HAUPPAGE-Nexus S, and a second low budget HAUPPAUGE on a
>Debian woody/sarge mix with a Kernel 2.4.22.
>  
>
Sounds like the well known "Video Data Stream Broken" bug. But you 
should find a log entry complaining about a broken video stream in this 
case (in /var/log/syslog only - not in messages!).

There is a workaorund for it - just szap on both cards before starting 
vdr. I've built this piece of code into the runvdr of the c't vdr:

if [ "$VDSB_WORKAROUND" = "yes" ] && [ -x /usr/bin/szap ] ; then
    szap -a 0 -n 1 -c /var/lib/vdr/channels.conf >/dev/null &
    szap -a 1 -n 1 -c /var/lib/vdr/channels.conf >/dev/null &
    szap -a 3 -n 1 -c /var/lib/vdr/channels.conf >/dev/null &
    szap -a 3 -n 1 -c /var/lib/vdr/channels.conf >/dev/null &
    sleep 5
    killall szap
fi


You may want to try this out.

>After diff'ing the two source trees I got all the differences in scripts,
>descriptions and the things for the elchi patch.
>Left over two interresting things:
>
>-#define MAXFRAMESIZE  KILOBYTE(512)
>+#define MAXFRAMESIZE  KILOBYTE(192)
>-#define MMAX_PLENGTH (64*MAX_PLENGTH) // some stations send PES packets that are extremely large, e.g. DVB-T in Finland
>+#define MMAX_PLENGTH (8*MAX_PLENGTH) // some stations send PES packets that are extremely large, e.g. DVB-T in Finland
>  
>
These changes are from the elchi patch too.

Tobias