Subject: Re: IBM drives vs Adaptec controllers/driver [was Re: 10G IBM ide hard disk]
To: NetBSD-current Discussion List <current-users@netbsd.org>
From: Greg A. Woods <woods@most.weird.com>
List: current-users
Date: 01/23/1999 13:19:52
[ On Sat, January 23, 1999 at 08:07:48 (-0500), Bill Sommerfeld wrote: ]
> Subject: Re: IBM drives vs Adaptec controllers/driver [was Re: 10G IBM ide hard disk] 
>
> > I have yet to have my controller not puke when AHC_SCBPAGING_ENABLE is on.
> 
> ditto.  tried it once, it went kaboom.

It's been working OK for me so far (knock on wood!), but I've not done
much other than run some benchmarks and dd's.  I should probably fire up
a dozen or more bonnie runs on different directories and partitions and
see how it fares.  I know people complained about SCBPAGING in about the
same vintage of FreeBSD's driver, but I was using it there at the time
with Seagate Cheetahs with no problem (with CCD mirroring in one case no
less! ;-).

> For a system I built about a year and a half ago, it started (out of
> the box) limiting the Ultra-Wide disks to 10MHz.  I had to go into the
> SCSI BIOS menus to enable the 20MHz speed for each target.  It would
> be nice (particularly for non-x86 platforms...) if the driver would
> provide some way to do this for you..

Perhaps I need to reduce the SCSI BIOS from 40MHz to 20MHz?  In any case
why did the FreeBSD-3.0 kernel work fine with the same settings?

I've read through some of the code that does this negotiation and
reporting, and compared it to the same code in FreeBSD-3.0, and there
are massive differences, any one of which could easily account for why
FreeBSD works better.

I'd much rather help port SCAM and test Justin's new AIC driver than try
to fix our decrepit version.  ;-)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>