Off the cuff, this doesn't look bad... tho it does a lot.
The reordering of locking in device_reset and bus reset definately looks
good, especially for the corner-cases (yanked device, etc.).
I'm uncertain if your handling of the io_request_lock is right but
getting information on how to hand
Guys,
this works dandy for us in RHL and RHEL, *especially* on bigger SMP boxes.
It does not fix all the bugs in that code, but I consider it a good start,
and unless someone speaks up, it goes to Marcelo after 2.4.26 is out.
Or maybe before, depending how long it takes him to release 2.4.26.
Plea
