I have no particular problem with the BSD approach. There are the basic engineering considerations of not disabling any interrupt for too long, but I think it is possible to meet those goals with the BSD approach.
In general, you want to push as much processing as reasonable outside the kernel where its failure will not impact parts of the system that don't depend on it. e.g. Push network protocol processing out, so you can debug it from the keyboard and display using normal debugging tools.
I have given up on trying to optimize rotational latency on SCSI disks. They are buffered, there a million different ones with different performance parameters, and the information on device timing is not easily available. Sigh.
Regards - Bill
Bill Frantz | The CDA means | Periwinkle -- Computer Consulting
(408)356-8506 | lost jobs and | 16345 Englewood Ave.
frantz@netcom.com | dead teenagers | Los Gatos, CA 95032, USA