From owner-freebsd-questions Fri Dec 30 09:03:47 1994 Return-Path: questions-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id JAA21919 for questions-outgoing; Fri, 30 Dec 1994 09:03:47 -0800 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.9/8.6.6) with SMTP id JAA21909 for ; Fri, 30 Dec 1994 09:03:45 -0800 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA12067; Fri, 30 Dec 94 09:57:41 MST From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9412301657.AA12067@cs.weber.edu> Subject: Re: Bernoulli drives and FreeBSD? To: dufault@hda.com (Peter Dufault) Date: Fri, 30 Dec 94 9:57:40 MST Cc: MULHOLLENJ@cpva.saic.com, freebsd-questions@freefall.cdrom.com In-Reply-To: <199412300159.UAA23786@hda.com> from "Peter Dufault" at Dec 29, 94 08:59:44 pm X-Mailer: ELM [version 2.4dev PL52] Sender: questions-owner@freebsd.org Precedence: bulk > > (2) If we did, the Bernoulli drives *lie* and say they are done > > formatting immediately. A correct implementation would > > have them ack the format *after* they were in a state to > > accept further SCSI commands. I've seen problems with a > > Kawai synthesizer that uses removable SCSI media for patches > > and so on, whereas it worked fine with a Syquest. > > There must be something you can ask it to see if it is done > and then write a special Bernoulli format. How does it respond > to a Test Unit Ready while it is formatting? Actually, it says "busy". This *is* the trick. I would have preferred they use a vendor private SCSI command for the format that keeps on going, though... I supposed one might handle it by setting a bit in the driver when a format command is issued, causing a TUR to be issued prior to any further commands. I don't know if it takes both LUNs offline in a two drive situation (almost all dual Bernoulli's are dual LUNs at the same target), so this may be a problem (or it may not). The test will have to be limited to the particular target or the particular target plus LUN, depending. That's actually a lot of state, but I don't think it would be worth traversing a queue of formatting devices when a command is registered, since it would up the expense for everyone else. Better to keep flag bits for everything. Anyway, it's an issue that needs to be dealt with at the generic SCSI level, but like the scanner code, may end up being dependent on some potentially non-existant features being present in the underlying card/card-driver. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.