> 
> 
> To all the ppp wizards,
> 
> I've recently run into some problems with the kernal spewing errors or 
> the sort:
> 
> Dec 30 21:19:41 opus /netbsd: zstty0: ring overrun
> Dec 30 21:19:41 opus /netbsd: zstty0: ring overrun
> Dec 30 21:19:41 opus /netbsd: zstty0: fifo overrun
> Dec 30 21:19:41 opus /netbsd: zstty0: fifo overrun
> 
> Occasionally (although never while I'm watching it) the kernal panics and 
> these are the last errrors I see.  It would be really nice if the kernal 
> didn't panic twice a day, it's getting to be a real pain in the butt.

As mentioned elsewhere, these errors shouldn't be the cause of the
panic. You can reduce the ring errors by compiling your own kernel
with the ring buffer increased. You need to pick a # which is a power
of two. 2k should be fine. There's nothing you (we) can do about the
fifo errors. Basically characters have been lost BEFORE our interrupt
servicer gets called.

> I seem to remember somewhere  that this is due to not enough bandwidth 
> for the serial ports, but I can't remember where I saw that or if there 
> is a fix.  Does anyone know where I'm getting this type of error?
> 
> My system is an SE/30 with 8MB and a 150MB external SCSI drive.  A Sonic 
> Ethernet card in the PDS slot (which works great) and a 28.8 modem on the 
> modem port.

Sounds like you're overloading the machine, or that the ethernet
driver is not quick-enough. Basically to get fifo overruns, SOMETHING
has to be sitting around with interrupts off for over 1 ms (at
28800, characters arrive at 2880 bytes/sec max, and the fifo
holds 3 characters).

Good luck!

Take care,

Bill