> Is anybody working on (or already did) improving com.c ?
> It works quite reliable for me on my box at work, which has fifo-chips.
> Only a few silo overflows under heavy load, nothing serious. But it doesnt
> work acceptable on this notebook - no fifo's and no chance to change
> the chips.

Stefan Grefen posted about small fix to com.c that might help.  It's
in with his PCMCIA work, and commented part of the patch should 
probably go into the tree.  I'm not on the commit list any longer, so I
don't know if it made it...