> Now that you mention it, I remember noticing that Reno and NetBSD do
> this differently: with Reno, the port indicator lamp on an RA81 drive
> will stay continuously lit during the bootstrap process, but once the
> system is up, and file systems are mounted, it will only light up when
> actual disk access is going on.  With NetBSD, the lamp is always lit
> while the system is running.
> 
> Is fixing this this a quick hack or a lot of code to do complex port
> access negotiation with the drive?

It's not a lot of code, but I'd be wary of messing with the MSCP
driver to get what is (in this case) only a cosmetic effect.  In
particular, you'll probably find (after making this modification)
that some third-party MSCP-emulating controllers don't work under
NetBSD anymore.  (Many third-party MSCP controllers - like the
SCSI controllers made by CMD and Anromeda - aren't entirely faithful
to what a real UDA50/KDA50/RQDX3 report back for drives that are in
anything but a perfectly "on-line, ready, and accessible" state.)

The NetBSD 'ra.c' driver is still short and sleek, but you can
easily bloat a MSCP driver into incredible size by deciding to add OS
support for dual-porting/multi-hosting/assisted-bad-block-replacement/etc.
While each concept may be digestible, the way that the concepts
can interfere with each other can be extremely messy.

Still, it'd be easier than adding RLV11 support !!!

Tim. (shoppa@triumf.ca)