Subject: Re: problems assessing disk size with ahcisata
To: Arto Huusko <arto.huusko@pp2.inet.fi>
From: Steven M. Bellovin <smb@cs.columbia.edu>
List: current-users
Date: 11/05/2007 12:36:25
On Mon, 05 Nov 2007 08:34:48 +0200
Arto Huusko <arto.huusko@pp2.inet.fi> wrote:

> Steven M. Bellovin wrote:
> > I just ran into a serious problem with a new Thinkpad T61: anything
> > looking at the MBR on the 160G disk running with ahcisata got very
> > confused.  I'll skip all the gory details (and the state it left my
> > disk in); the net, though, is the fdisk and sysinst think the drive
> > has only 67MB.  When I changed it in the BIOS to compatibility
> > mode, it picked up the size correctly.
> 
> I've seen this behaviour on T61 too, but the way the disk is probed
> seems to depend on what order devices are probed and attached:
> tweaking unrelated things in kernel config make or break the disk
> attachment.
> 
> I guess there is some delay or interrupt issue with the disk probed:
> when the disk is probed wrong, there is a noticeable delay in kernel
> messages, and often some other device attaches in the middle of
> disk attachment.
> 
> Anyway, for what ever reason, attachment of pckbc fixes or breaks
> the disk probe for me: with pckbc at isa disk is fine, with pckbc
> at acpi it is not.
> 
Interesting.  It is indeed at acpi, since I'm running the power
management branch.


		--Steve Bellovin, http://www.cs.columbia.edu/~smb