itojun@iijlab.net writes:

> > > 	I'm not sure if it is documented somewhere, but how should we
> > > 	code foo_detach?  ep_pcmcia_detach() has call to bpfdetach()
> > > 	with the above change, while ep_pcmcia_attach() doesn't.
> > > 	bpfattach() is in elink3.c:epconfig().  i fear to see mismatch
> > > 	between dev/ic code and dev/pcmcia code so i'd prefer bpfdetach()
> > > 	to be done somewhere in elink3.c.
> > > 	could I see an example?
> >I even committed an example (the `tlp' driver), but it was apparently
> >ignored :-/
> 
> 	Thanks, I looked tlp and now ne can be detached without trouble.
> 
> 	Is it okay to assume that the user bring interface down (!IFF_UP)
> 	before removal?  ne2000 will lock up solid in interrupt handler.

This is probably a bug in the driver or the pcmcia code.  I've been
working on DTRT with interrupts on enable and disable.  E.g., ray would
hang on card insert previously but I fixed this in pcic.

Pcic also appears to DTWT on disable.. I'll fix this too.

Chris.