[vdr] Re: vdr[3330]: ERROR: can't record MPEG1! - possible cause

[Andreas Schultz <andreas@aschultz.net>]

Rainer Zocholl wrote:

> andreas@aschultz.net(Andreas Schultz)  21.09.01 13:19
> 
> Once upon a time Andreas Schultz shaped the electrons to say...
> 
> 
>>Hi all,
>>
> 
>>The MPEG1 problem caught my intressted when it became apparent that
>>there could be a race condition involved that is only been triggered
>>on fast systems.
>>
> 
> I added a mean Software delay loop.
> 
> volatile int iTime=0;
> void cTS2PES::instant_repack(const uint8_t *Buf, int Count)
> {
>   int c = 0;
> 
>   while(iTime< 40000)
>   {iTime++;};
>   iTime=0;
> 
> 
>  (the 40000 was "hand optimized", 100000 leads
> "buffer 100%" messages...)
> That seems to be better, but no solution.
> (It's of cause not the right place.
> Too the power consumption rises from below 50W to over 75W!)


the loop is to tight, you have to insert an usleep here. But busy loops, 
and especialy tight one like yours, are always a bad idea. Avoid them, 
whenever possible.

>>And indeed there is a race that could be related to the problem.
>>
> 
> Maybe of interrest:
> 
> The machine hangs often while recording the garbage.
> (Not Network, Not CNTL-ALT-DEL, No Console switching, (Sysreq i did not try)


I noted similar problems with DVD playback. Currently my best guess is 
that invalid data or data that overflows the DVB internal buffers 
crashes some part of the driver or the card itself, leading to some kind 
of kernel lookup.


> But "lirc_serial" still generates messages complaining about 
> "invalid interrupts " on the console (are of no relevance)
> If i press a key on the RC, lirc_serial (Inmterrupthandler)
> writes "buffer overflow" on the concole. So the CPU is still living!


I don't have a monitor attached to my system, but network was definetly 
dead, as well as APM.

> 
> 
> 
> 
>>thread.c, arround line 70:
>>
> 
> 
>>An untested patch for the problem.is attached.
>>
> 
> I'l try.
> 
> Thanks!
> 
> 
> 

Have fun

    Andreas