Re: WD_QUIRK_FORCE_LBA48

["Jonathan A. Kollasch" <jakllsch%kollasch.net@localhost>]

On Tue, Dec 15, 2009 at 10:24:32AM +0100, Manuel Bouyer wrote:
> On Tue, Dec 15, 2009 at 08:26:15AM +0000, Michael van Elst wrote:
> > bouyer%antioche.eu.org@localhost (Manuel Bouyer) writes:
> > 
> > >There are ATA-5 drives reporting LBA48 supports. LBA48 showed up before
> > >ATA-6 was out (like much features in ATA world, in fact).
> > 
> > I think it is bogus to rely on ATA-5 specifics (LBA28 is up to 2^28)
> > for devices that do not follow that standard by extending it.
> > 
> > Making the decision on the LBA48 capability is more likely to
> > succeed. In particular, because this can only fail for a
> > very rare condition and only for the last two sectors.
> 
> The fact that it's only for the last sector doesn't make the problem
> less critical.

And the fact that wd(4) now hardlocks Chris's hardware isn't
even more critical?

Drives that conform to ATA-6 and greater are going to continue
to become more common.  Treating them as the exception is not
a good idea.

People using drives larger than 2^28-1 sectors on LBA48-incompatible
controllers are treading on thin ice to begin with.  Very many LBA48
drives won't even let you access sector 2^28-1 using a LBA28 command.
These drives conform to the current ATA standard.

I don't like what the standard now says either.  But we can't just
ignore the current standard because we don't like what it says.

We probably aren't going to get any complaints if we require
LBA48 commands to access sector 0xfffffff.  If we do get
complaints, we would have the standard on our side.

We are probably going to see a problem like Chris's again.

        Jonathan Kollasch