Return to BSD News archive
Xref: sserve comp.os.386bsd.development:1323 comp.os.386bsd.bugs:1609 comp.os.386bsd.questions:6003
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!uwm.edu!linac!news.ssc.gov!fnnews.fnal.gov!cdfsga.fnal.gov!dejan
From: dejan@cdfsga.fnal.gov (Dejan Vucinic)
Newsgroups: comp.os.386bsd.development,comp.os.386bsd.bugs,comp.os.386bsd.questions
Subject: Re: NetBSD 0.9 silo overflow fix
Date: 18 Oct 1993 00:36:31 GMT
Organization: Fermi National Accelerator Laboratory, Batavia IL
Lines: 36
Distribution: world
Message-ID: <29soef$3gd@fnnews.fnal.gov>
References: <1993Oct15.122522.8764@cs.chalmers.se>
NNTP-Posting-Host: cdfsga.fnal.gov
>Silo overflow fix
>=================
>Here's a fix for those of you that don't have UARTs with fifo.
>It uses Bruce Evans wonderful fast interrupts to be able to
>service the interrupts at once. It's not as elaborate
>as the sio driver but it works for me (I've transferred a
>total of over 50M with 38400 SLIP without a single overrun error).
Yes, it fixes silo overfelows, but my SLIP still hangs...
Zmodem works fine, I've transferred (many)Mb without a glitch
and with lots of other things happening on the machine, so there's
no question that characters don't get lost. But SLIP hangs. Aaargh.
It's not the server (many people using it all the time without trouble),
it's not header compression (tried without it, same), it's not flow
control (tried at 2400 no flow control, same), it's nothing I could
think of. It seems to me that it happens only when large chunk of data
arrives (ftp or cat 'largefile'), and slattach can't exit when killed:
it hangs so hard the box won't reboot. With new com it only works longer.
486DX/33, 8Mb, local bus serial card (with HDC and kitchen sink),
USR Sportster (14400), Cisco server on the other side. Any ideas?
>The patch is for NetBSD-current, but it should work for
>NetBSD 0.9 as well.
Nope. config changed a lot. Getting the whole new kernel source is
probably better than patching the patch... :)
Have a Good Day,
Dejan
PS. About THE serial problem: couple of days ago there was a posting on
comp.sources.unix, FAS serial driver for System V i?86 Un*ces. Anyone
looked at that?