From owner-freebsd-questions  Wed May  3 20:22:45 1995
Return-Path: questions-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id UAA16177
          for questions-outgoing; Wed, 3 May 1995 20:22:45 -0700
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id UAA16152
          ; Wed, 3 May 1995 20:22:15 -0700
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id NAA32266; Thu, 4 May 1995 13:20:31 +1000
Date: Thu, 4 May 1995 13:20:31 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199505040320.NAA32266@godzilla.zeta.org.au>
To: hardware@FreeBSD.org, nc@ai.net, questions@FreeBSD.org
Subject: Re: silo overflow!? w/ 16550s!
Sender: questions-owner@FreeBSD.org
Precedence: bulk

>I have a system with 4 16550 capable serial ports. All run on their own 
>IRQ and I/O address. Two of the ports are internal modems [a 14.4 and a 
>28.8] and card has another 2 ports each connected to 28.8s. All ports [as 
>above] are 16550 UART'd.

>When one connection is open full blast [a slip] there is no problem. The 
>minute I try to use kermit to dial out to a local system for giggles, or 
>even another modem on the same machine I get :

>system /kernel: sio1: 1 more silo overflow (total of 10)  for example.

The 16550 fifo trigger level is set very agressively (to 14) to maximise
efficiency.  Dynamic adjustment of the trigger level (to 8, then 4, then
2, then 1) after silo overflows is #ifdef'ed off.  A level of 8 would
probably work for all 386/16 and faster systems with only 4 ports,
provided they don't have bus-hogging DMA controller.  You currently have
to edit the sources to change the level.

Bruce