Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 18 Jun 95 14:51:20 MDT
From:      terry@cs.weber.edu (Terry Lambert)
To:        guido@gvr.win.tue.nl (Guido van Rooij)
Cc:        gibbs@freefall.cdrom.com, freebsd-bugs@freebsd.org
Subject:   Re: 2.05R: failure to get fd parameters correctly
Message-ID:  <9506182051.AA18733@cs.weber.edu>
In-Reply-To: <199506181850.UAA03786@gvr.win.tue.nl> from "Guido van Rooij" at Jun 18, 95 08:50:15 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> Justin T. Gibbs wrote:
> > >I tried to install 2.05R on a CSSI only system with a 2GB Quantum
> > >EMPIRE_2100S (Btw: dont try this at hoime..that disk sucks)
> > >Anyway: after installing all of it, it failed to boot from the
> > >fixed disk with a missing operating system.
> > >So I tried to reinstall it but now I chang4ed the geomtery to 
> > >what 2.0R reported when booting. This just failed. Every key I touched
> > >I got this popu menu shouting that I entered the wrong geometry :-(
> > >I `solved' the problem by adding a dos partition of 1 meg, but this
> > >is bad. I dont want DOS. And I dont understand why I caanot use the geometry
> > >reported by 2.0R when booting for the 2.05R installs...
> > >
> > >-Guido
> > 
> > I installed to an Empire 2100 last nigt using the standard Adaptec
> > geometry since it was hanging of a 2742T.  What kind of disk controller
> > are you using?
> 
> OOps...sorry:
> An ncr 53c810 controller.
> 
> This is waht 2.0R thought of it:
> 
> sd2: 2006MB (4108600 total sec), 3053 cyl, 12 head, 112 sec, bytes/sec 512

The problem with the quantum and the NCR controller is the sequencer
code having a window between checking for other bus users and asserting
bus ownership because it is a fast drive.

This hole was closed on the Adaptec sequencer code, and the procedural
fix reported to the hackers list.

I am unsure if the fix has been incorporated into updated NCR sequencer
code -- in fact, I doubt that it has.

Your current alternatives are:

1)	Use a non-NCR controller

2)	Use a different disk that does not exercise the hole.  I believe
	the hole is exercised by drives that are called "AV drvies".
	These drives achieve their faster throughput by not doing thermal
	recalibration.  This is in fact a terrifically bad hack to get
	the desired results, and will definitely increase the failure
	rates you'll see when using the drives.

3)	Go to the Hackers list archives, and using that information
	and information obtained from NCR, fix the sequencer code.
	Hopefully if you choose this option, you will donate the code
	back, both for the rest of the NCR controller owners benefit
	and for your own benefit in not having to reintegrate the
	changes you make each time the driver is revised.


					Terry Lambert
					terry@cs.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9506182051.AA18733>