Re: wi(4) reset problem

[Martin Lexa <martin_(_at_)_martinlexa_(_dot_)_cz>]

V Po, 21. 06. 2004 v 09:12, Martin Lexa píše:
>   When this happen all network traffic completely stops.
> Even ping on ethernet card don't work. Only solution is to reboot
> the computer. And if I remember correctly even tty input is disabled
> during the resetting period.
> 
>   Scenario:
>     time  0 sec - problem happen 
>     time  0 sec - sytem unresponsible (no network traffic on all
>                   interfaces)
>     time ~8 sec - system responsible (network traffic flows)
>     time ~9 sec - system unresponsible
>     time ~17 sec - system responsible
>       and so on.
> 
>   Looks like that system is only responsible after wi card is going
> to DELAY / timeout after (un)successful reset/init (May be
> timeout_add() at the end of wi_init() function?).
> 
>   If I could say, failed resetting of the (wi2) card isn't too much
> critical for me. However, the unresponsible thing is. I'm not
> OpenBSD guru and I don't know OpenBSD's internals so may be
> I'm wrong, but couldn't it be that splimp() function blocks all others
> interrupts (tty, network devices, ...)... ? Have found this (rather
> old) post:
> 
>     http://mail-index.netbsd.org/tech-kern/2001/04/12/0012.html
> 
>   Probably next week I'll have some time to spend on this. If
> somebody experienced could point me to what should I look for,
> it would be great. Thank you for your time.

  Question:

  Does the use of splimp() function means that others
network (or tty) interrupts are disabled too?

  Answer:

  from man splimp : 

   splvm()         blocks hard interrupts from all devices that are
                   allowed to use the kernel malloc(9).  That includes
                   all disk, network, and tty device interrupts.

  Sorry, that I didn't catch that earlier. However, this makes
me not so happy...
 
Thank you all,
Martin.