Michael Hitch writes:

>   If you can build a new kernel, here's a fix I've been using.  The first
> part of the patch is a workaround for the stray interrupt.  The rest of
> the patch keeps the interface packet and bytes counts from getting counted
> twice.

Thank you, is there anything else I need (a definition for S_INTR_LATCH) ? 

I'll apply this to 1.5.3_alpha (march 2nd I believe).  Based on what you've said about
the code in -current not changing this should be a clean patch (which looks to be
the case).