[Mplayer-dvb] Re: Reproducible MPlayer Crash with dvb-t

[Nico Sabbi]

David Maslen wrote:

>Haven't managed to subscribe to this list, so apologies for the cut and
>paste reply.
>
>Nico Sabbi nsabbi at tiscali.it 
>Tue May 11 16:28:30 CEST 2004 
>
>  
>
>>the error you see below is due to the fact that mplayer didn't read() 
>>fast enough from the
>>DVB drivers, which overflew.
>>This is logical if  your system isn't fast enough to handle the load.
>>    
>>
>
>The system does seem fast enough to play dvd's and other video files,
>so I think the processor and disk speed are adequate. I have seen error
>about system being to slow, but only while accessing a DVB stream, not
>when reading from disk.
>
>  
>
>>>The Mplayer session immediately ends, and won't start again until I
>>>reload the kernel module dvb-bt8xx. (rmmod then modprobe).
>>>      
>>>
>
>  
>
>>this sounds like a bug in the dvb drivers (either bttv or dvb-bt8xx).
>>Does tzap -r + mplayer /dev/dvb/adapter0/dvr0  work correctly when you
>>    
>>
>
>  
>
>>have this problem?
>>    
>>
>
>Not had a chance to try this yet. Will do soon. That would be something
>to report to the standard kernel bugs process?
>  
>

no, to the linux-dvb at linuxtv.org ML

>  
>
>>Maybe using your network card triggers something that raises the
>>system load?
>>Is your disk very slow? Something in 2.6.x  seems to have a similar 
>>effect on my system:
>>every massive I/O on the disk seems to almost paralize my PC; I
>>suspect anticipatory scheduling, so I'll try to disable it.
>>Are you sharing interrupts?
>>    
>>
>
>Last night I tried disabling the parallel and serial ports, which
>change the interupts. The BT878 driver is now sharing with some USB
>drivers, but they aren't being used con-currently.
>
>I also changed the PCI latency in the bios from 33 to 128. 
>This seems to have created some instability in my system, with
>Windowmaker reporting a SIG11 bug. However I switched to open box and
>kept testing.
>
>Basically, it does appear to be continual ethernet card access on eth1,
>which is not sharing a interupt, that causes the process to end
>prematurely.
>
>Once this has happened I need to reload the bt878 module, before I can
>run mplayer again.
>
>I also added echo 1024 > /proc/sys/dev/rtc/max-user-freq to my start up
>scripts while I was at it.
>
>I am using all 5 pci slots on the motherboard. I was wondering if there
>were any point in moving the cards physically?
>
yes, it's the usual way to change interrupts for pci cards, but some 
board such as your Asus
(once I had one) should permit to assign irqs by hand in the bios. Some 
of them
will still be hardwired (usb mainly) and shared with the highest 
numbered PCI slots.
I'm not sure that assigning irqs in a different manner will solve or 
reduce your problem,
but trying doesn't harm

> I also saw an older post
>which suggested in a motherboard with an agp video card, the first pci
>slot should be left free. I could do this if I removed the ocasionally
>used USB card. Would the first slot be the closest slot to the agp
>card? Is there any point in trying this?
>
>Finally on a related issue, the motherboard is a Asus P2B-F.
>It's been a good board, but I'd consider replacing it with a new board
>that was compatible with the CPU (Celeron 1400), Memory SDRAM and hard
>disks (IDE). Anyone got a suggestion for a suitable replacement that
>might have a faster bus and therefore help in my DVB reception?
>
>
>
>=====
>--
>David..
>
>
>  
>

you should ask to someone with a BT878 card at that ML