[564] in linux-scsi channel archive
Re: IN-2000 and Linux, some progress
daemon@ATHENA.MIT.EDU (Harik A'ttar)
Wed Aug 30 15:49:15 1995
Date: Wed, 30 Aug 1995 05:35:01 -0400 (EDT)
From: "Harik A'ttar" <harik@chaos.sub.ucf.edu>
Reply-To: drm89121@pegasus.cc.ucf.edu
To: Stephen Walton <swalton@galileo.csun.edu>
cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <Pine.HPP.3.91.950829090614.8089A-100000@galileo.csun.edu>
On Tue, 29 Aug 1995, Stephen Walton wrote:
> Summary: Everex 486/33, 8 MB RAM, TVGA card. IN-2000 SCSI host adapter
> with latest BIOS (version 887-1.3). HP C2247 1 GB disk and Exabyte
> 8200 tape drive, terminators on latter. NE-2000 Ethernet card; system
> on LAN. Slackware 2.3.0 distribution except for kernel (see below).
>
> Status up until now: system is somewhat better with a newly recompiled
> custom version 1.2.13 kernel, which includes no SCSI or Ethernet
> drivers except the ones for the cards I have. I also changed the
> IN-2000 I/O space from its factory default setting of 110h-11Fh to
> 200H-20F. The company tells me that the former setting conflicts with
> some (undocumented) addresses used by 3COM Ethernet cards. Finally, I
> moved the IRQ on the NE-2000 to 5 from 2. With the latter setting,
> the Linux kernel was claiming it was at IRQ 9 (?).
Ah, irq 9 and 2 are bad to use. :) Did you check the /var/log/messages
(sometimes /var/adm) To see if it all was detected right?
>
> However, while more stable, the system is still acting as if the
> virtual memory subsystem doesn't work. When recompiling the 1.2.13
> kernel under 1.2.8, I couldn't get it to work very well unless I
> turned off the -pipe option. Even now, weird things happen, like
> e-mail from cron to root complaining about not being able to find
> "/usr/lib,atrun" (sic) or, just now, an attempt to build the 1.3.21
> kernel attempting to give gcc the "-fomit-frale-pointer" switch. So
> far, I've not been able to access the Exabyte drive from the Linux
> side at all, but it works fine with the ASPI MS/DOS tar which I have.
Actually, it sounds like you have a corrupt swapspace. I.E. stuff
goes there and dosn't come back right (it should crash, tho. not
just text strings) Try unswapping and mkswap(8) with the checking
option. (is there a raw disk scanner/checker for linux?)
>
> Any more ideas? I'm going to give the 1.3.21 kernel a spin just to
> see what happens.
Possibility: One or more of the moved cards may be interacting
with your disk system. Try making a table of IRQ/DMA/ADDR space
and see if you get overlap anywhere. That saved my butt once,
undocumented address usage is harder to find. (ack phfft!)
Since you have the source, read drivers/net and drivers/scsi
to see what address space they use (found by main-addr + offset)
while mucking with the source, try forcing each driver to scan
the actual address first (A lot have a table of addresses to scan)
that would stop any other cards from going berserk.
I remember the fun I had in DOG with no source, trying to get
drivers to ONLY scan the correct space. ACK! Sound cards were
the worst, they all started scanning at 0x200, and that crashed
my machine INSTANTLY, nothing but BRS. Of course, that _WOULD_
be the main reason I switched to linux in the first place... :)
> Stephen Walton, California State University, Northridge
> "Be careful what you wish for; you might get it." swalton@csun.edu
chaos@dynamic.ip.don't.reply Guess what? I really _DO_ speak for my
Dan Merillat / Harik A'ttar system. And if you share my opinions,
drm89121@pegasus.cc.ucf.edu you should seek professional help.
Finger for my PGP Public Key PGP-mail prefered, thank you.
