On Wed, 24 Jan 96, is@Beverly.Rhein.DE wrote:

>I've seen PPP lines on NetBSD-1.1-ix86 producing silo overflows; after
>looking at the code, I realized that the damned NS16550 lookalikes don't
>have _hardware_ flow control, but only _software_ flow control (that is,
>if the interupt isn't serviced fast enough, you loose). 
>
>Not I never saw more tha 0.1% lost packets or bytes, so this is no real
>problem (as PPP recognizes broken packets). I wonder, however, why this
>shows on certain NetBSD-i386 machines and not on others. Could it be
>that the wd driver manages to block serial interupts for more than 1 ms
>or 2?

There was a go-round mid-late last year about how NetBSD simply
doesn't service interrupts as quickly as, say, FreeBSD.  I believe
FreeBSD implemented some kind of virtual-DMA thing in their interrupt
handler that's supposed to especially help their com (sio) driver.

-----------------------------------------------------------------------------
  Michael L. VanLoon                                 michaelv@HeadCandy.com
       --<  Free your mind and your machine -- NetBSD free un*x  >--
     NetBSD working ports: 386+PC, Mac 68k, Amiga, HP300, Sun3, Sun4,
                           DEC PMAX (MIPS), DEC Alpha, PC532
     NetBSD ports in progress: VAX, Atari 68k, others...
-----------------------------------------------------------------------------