Subject: Re: powerhooks and SCSI disks
To: Jason Thorpe <firstname.lastname@example.org>
From: Michael Lorenz <email@example.com>
Date: 12/16/2005 12:36:15
Content-Type: text/plain; charset=US-ASCII
> > I'm currently looking for ways to improve power management on
> > SPARCbooks, mainly adding things like some sort of standby mode, =20
> > turning off unused peripherals and so on. We can for instance turn
> > off power for the internl SCSI disk ( in case a drive wouldn't
> > honour a motor stop command ) So, I have a few questions:
> What drive would not honor a STOP command? Seems broken... anyway...
Well, different SPARCbooks use either small SCSI drives or several
different SCSI-to-IDE adaptors, mine has an adaptor and a Toshiba
drive. It spins down after a scsictl sd0 stop but I think there's a
reason why Tadpole added hardware to physically power down the drive.
> > - are there any objections against adding a power hook to sd so it =20
> > would flush the drive's cache before we're going to remove power?
> This seems like a fine thing to do.
Ok, so the initial version of the hook function would just call
sd_shutdown() when we're going into suspend or standby and do nothing on
> > - is there anything to do after turning the drive back on to resume
> > operation?
> I can't think of anything special.
> One thing you might want to make sure of is that our SCSI I/O path =20
> can handle the case where a command fails BECAUSE the drive is not =20
> spinning, and can recover by issuing a START command to the drive.
Not quite sure - scsictl sd0 stop works as expected and when there's
something to do the drive just spins up again. Not sure if it got a
START or if it just spun up on its own.
Hmm, should I issue STOP resp. START commands in the powerhook ( and
keep a status bit to avoid confusion ) or is that a Bad Idea?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (NetBSD)
-----END PGP SIGNATURE-----