> One of Thor's motivations was the fact that, to cache metadata 
> effectively in some configurations (cvs.netbsd.org, in particular), you 
> need a LOT of *really small* buffers.  Your patch will help achieve 

I haven't addressed that (yet); buffer memory allocation is still done
with page-size granularity. But it should be relatively simple to
mix in a "small buffer pool" consisting of say 1k or 2k sized memory
buffers.

> Also something to consider there... Since you're using kernel_map, you 
> can't actually free pages back from an interrupt context.  Can any of 
> the functions that free pages back actually be called from interrupt 
> context?  I can't really remember, but perhaps we need some assertions 
> in there to catch stuff like this.

Currently, all of that is done in allocbuf() which is never called
from an interrupt context. I'll make sure it stays that way...

-pk