On Tue, Dec 16, 2003 at 09:19:54AM -0800, Jason Thorpe wrote:
> It would be really cool if you could resurrect this stuff.
> 
> As far as the "larger than 64K reads" thing goes, yah, it is kind of 
> annoying about any drive that uses the ATA command set has this 64K 
> limitation (I think it's hard-wired into the spec, right?)

If I remember well, the real limit is 128k, by setting the sector count to 0.
I guess we'll find broken drives that won't handle this properly :(

> 
> ...but I would still like to fix the problem so that we can e.g. issue 
> large I/Os to logical disks (e.g. ccds) that allow us to issue 64K I/Os 
> to the disks themselves.

Or larger. At last with (e)siop and probably other drivers, handling (much)
larger I/O would be almost trivial (at the expense of more RAM used for
the I/O structures, of course).

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--