Problems with dropped callouts

[John Baldwin]

On Tuesday 22 August 2006 11:01, Jim Skon wrote:
> Hello,
> 
> I have been developing a modification to the 802.11 driver to support a
> version of PCF (polling mode).  The AP part of this solution depends on
> several active callouts to time the operation of the protocol.  The
> problem I am having is that after several hours of stress testing
> (pushing the system at maximum bandwidth), the system appears to start
> losing set callouts.  The callouts are still active, and past due, but
> are never called.
> 
> I have already done similar implementations in the netbsd and linux
> environments, and have never had such a problem.
> 
> Note that I am setting callouts (only 3) at a rate in the 100's per
> second, so a lot is going on.
> 
> Has anyone ever had such a problem?

I'm not aware of anyone else having this problem (yet).  If you generate a 
core dump, can you see the callouts on the calloutwheels?

-- 
John Baldwin