On Mon, Mar 27, 2006 at 09:45:06PM +0100, David Laight wrote:
> On Mon, Mar 27, 2006 at 08:49:34PM +0200, Manuel Bouyer wrote:
> > The problem is not disk, it's 128GB LBA48-capable disk (or larger drives
> > clamped at 128GB) on non-LBA48-capable controller. You'll suddenly get a
> > unaccessible sector, at best (this could also hang the system, or worse,
> > corrupt data by reading/writing the wrong sector).
> 
> How would they be 'clamped' at 128GB ?
> If the NetBSD partition extends beyond that point then we'll have to be
> able to do LDA48 accesses anyway.
> If the controller doesn't support LDA48, then a disk > 128GB will cause
> problems - unless we know the controller won't do LBA48.
> If the drive is jumpered to report a size of 128GB then maybe we can't tell.

Yes, this is what I meant - jumpered to 128MB

> 
> Plausably we could retry the transfer as an LDA48 one if we beleive it
> is not likely to fail (as in lockup the system).  Then the kernel would
> auto-detect the bad driver firmwares.

This adds complexity in the code and I don't think it's worth it.
The quirk regexp we have now work well enough for this serie of
drives.

> Oh - are these failures on single sector transfers, or on multi-sector
> transfers that cross the 128GB boundary?

I don't know - actually I don't think I own one of these buggy drives.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--