Too much work at interrupt

[Daniel Veillard]

   Donald,

 first thanks a lot for your prompt answer (especially at 1:30 A.M :-)

> > message (while they show the "SMP simultaneous entry of an interrupt handler"
> > occasionnaly but apparently without damage on 1.03).
> 
> Try the SMP patch from Loic Prylli <Loic.Prylli@ens-lyon.fr>
>    http://beowulf.gsfc.nasa.gov/linux/misc/smp-bug.html

 Ok, I have seens it on the SMP list, but there wasn't much feedback. Must
be good, otherwise one of the Gods would have expressed disagreement.
 Applied the patch, recompiling the kernel. I will reboot the test machine
tomorrow, but I may not restart the production servers right now ...

> >  Any chance that both are variations of the same 2.0.xx SMP kernel bug ?
> > Shoudl I upgrade to 1.05 ? Should I try to increase the 
> >   static int max_interrupt_work = 20;
> > increase as suggested in a past mail ? 
> 
> Yes, upgrade if you use multicast.

  Well, not really. BTW I wasn't able to find 1.05, it's listed on the
Web page Changelog but I found only "eepro100.c:v1.03 8/11/98" on the
web link or on the FTP server (I may not have searched at the right place
though).

> You can increase the max_interrupt_work value when loading the module:
>   echo "options eepro100 max_interrupt_work=40" >> /etc/conf.module

  This one is real nice, will do that and stop/start the network. Modules
are great :-)

> This is probably occuring because the machine is under high load.
> If the machine doesn't have real-time responsibilities, increasing the work
> permitted is safe.

  Ok, thanks a lot for the explanations/help,

 Daniel

-- 
Daniel.Veillard@w3.org | W3C  MIT/LCS  NE43-344  | Today's Bookmarks :
Tel : +1 617 253 5884  | 545 Technology Square   | Linux, WWW, rpm2html,
Fax : +1 617 258 5999  | Cambridge, MA 02139 USA | badminton, Kaffe,
http://www.w3.org/People/W3Cpeople.html#Veillard | HTTP-NG and Amaya.