From owner-freebsd-current  Fri Apr  5 10:31:21 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA00811
          for current-outgoing; Fri, 5 Apr 1996 10:31:21 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA00806
          for <current@FreeBSD.ORG>; Fri, 5 Apr 1996 10:31:18 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.12/8.6.9) id EAA18166; Sat, 6 Apr 1996 04:26:33 +1000
Date: Sat, 6 Apr 1996 04:26:33 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199604051826.EAA18166@godzilla.zeta.org.au>
To: bde@zeta.org.au, jgreco@brasil.moneng.mei.com
Subject: Re: tty-level buffer overflows - what to do?
Cc: current@FreeBSD.ORG, root@deadline.snafu.de
Sender: owner-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

>> >What kind of drives?  IDE drives bite, particularly if they are being used
>> >at the same time as the serial port... etc, etc.
>> 
>> IDE drives have no affect on the operation of the serial unless they
>> are so slow that the system spends too much of its time in the kernel.
>> Bus-hogging SCSI controllers bite.

>It has been my observation that IDE drives DO tend to affect the operation
>of serial I/O, at least during heavy I/O periods.  Small-memory systems tend
>to spend much more time doing "heavy I/O" (swapping), in my experience, this
>is just one reason I put 8MB in even my smallest machines these days.

It's an indirect effect.  Small-memory systems are more likely to have
slow IDE drives that make the swapping slower.  You probably can't
afford to swap the applications doing serial i/o at all if you don't use
flow control - the kernel buffers are only large enough for 100 msec of
input at 115200 bps.

Bruce