> >> If we get a journalling implementation that's something it can take care
> >> of, whether it's PATA (flush the cache) or SCSI, where it can force
> >> access
> >> to the disc.
> >
> > Flushing the entire cache is unnecessary and will kill performance.
> 
> Unless things have changed recently, PATA gives you no other option.

PATA does have a version of queuing, but everything I've read about it
says that it sucks, and it is almost never implemented anyway.
PATA sucks in many ways.

Fixed in SATA.  Except someone needs to implement NCQ in *BSD drivers.