Subject: Re: wi0 timeouts and subsequent unresponsiveness
To: Brian de Alwis <bsd@cs.ubc.ca>
From: Pavel Cahyna <pavel.cahyna@st.mff.cuni.cz>
List: netbsd-help
Date: 01/12/2005 09:26:58
On Tue, 11 Jan 2005 16:00:21 -0800, Brian de Alwis wrote:

> I've been experiencing some frustrating problems with my Lucent
> Orinoco 802.11b (Classic Gold) wireless card (dmesg below).  I'm
> running on a NetBSD 2.0 kernel, and use dhclient to get leases.
> 
> The main symptom is that all wireless traffic goes dead.  The
> console shows messages like:
> 
>     wi0: device timeout
>     wi0: command timed out, cmd=0x2, arg=0x0
> 
> Thinking originally there was a minor hiccup with the DHCP leases, I
> would invoke dhclient, but would see messages like:
> 
>     send_packet: No buffer space available
> 
> The only solution I've found is to ifconfig the card down and then
> up again.  Traffic usually resumes with no problem; if I catch it
> early enough, I don't need to release/reget my DHCP lease.

I see this too, not very frwquently, with my PCI XI-626 card. Linux users
also saw some issues like this with the HostAP driver (I don't know any
detail). I plan to write a monitoring daemon that would reset the card
when such condition is encountered.

> I'm wondering if this is related to my hard drive reporting lost
> interrupts:

Isn't the wi driver spinning for a long time with interrupts disabled?

Bye	Pavel