Jens Axboe wrote:
> That should work as well. WRT ata_sg_is_last(), if we go ahead with my
> recent sg chaining updates, we can keep the test as it would be a single
> conditional and not require any looping.
> 
> Let me know when you have tested this!

The patch I attached to the last email got both sata_mv test boxes 
working reliably (so far).

I worked up a patch that kills ata_sg_is_last() (plus the 
max_phys_segments sata_mv fix), see attached.  I'm thinking this is what 
I like to see in upstream.

Of course, this doesn't explain why ata_sg_is_last() was broken, but 
since it's working _and_ slightly more efficient, I don't really care :)

	Jeff