> Erik E. Fair wrote:
> > 
> > How fast are you running the serial interface to the modem?
> 
> Ahhh...now that's a good question ;-)  How do you set that?  Is it via
> stty?  I believe it is being set somewhere, but I may be missing it...I'll
> have to pour over my scripts again.

What's the baud rate you're using to talk to the modem? It should be
fairly ovious in the ppp scripts. :-)

> > We've just been discussing fifo overruns (in the generic case) on another
> > NetBSD list...
> 
> Yeah, I noticed...the reason why I didn't post there was that it seemed
> more focused on silo overflows, which I thought had a different root
> cause, and which I thought that the mac68k port had just about eliminated
> (or am I thinking of ring buffer overruns???)

As I understand it, silo == ring. The silo is the place where things
are stored until they can be used (processed) by the input routines.
The ring buffer is a particular implimentation used by the zs drivers
(they all pretty much do the same thing).

fifo overruns happen (on mac68k) when we spend too much time at
splhigh.  ring overruns happen when we spend too much time with
interrupts blocked, as pretty much any interrupt blocking will keep us
from servicing pending tty characters.

Paul Goyette also complained of problems with the newest current giving
fifo overruns. I wonder what we changed.

Take care,

Bill