Subject: Re: Support for removable MSCP devices
To: None <port-vax@NetBSD.ORG>
From: John Wilson <wilson@dbit.dbit.com>
List: port-vax
Date: 02/01/1998 09:29:19
>From: Anders Magnusson <ragge@ludd.luth.se>

>> raioerror
>>  unit available (unknown subcode) (code 4, subcode 0)

>Humm, this is probably true. My goal is to support even RA disk unit
>changes "on the fly", but there are lots of "trial-and-error" to see
>how that work. 
>
>The floppy problem is more important, though. I will look at it later
>this day. I think also that I should break out the floppy friver from
>the ra disk driver, instead of having them act the same.

Floppies are likely to be the same as hard drives for this (I haven't done
much testing on RQDX3 floppies), anyway it's pretty simple.  The controller
returns ST.AVL (004) for any I/O attempt to a drive that has gone "available",
so you just need to OP.ONL (011) it and re-issue the failed command.  If
you enabled attention messages you'll also get notified whenever the user
changes the online status of a drive (which officially changes it to AVAILABLE
whether it was previously ONLINE or OFFLINE, that's why there's only one
style of attention message for this).

Note that CMD SCSI controllers are buggy in this regard and return ST.OFL
(003) instead of ST.AVL when the SCSI device gives a UNIT ATTENTION, and it
doesn't poll the drive to check for this so you don't get the attention
message until you try to do I/O.

John Wilson
D Bit