Subject: Re: kern/37508: After jmcneill-pm, disks are switched to standby
To: None <joerg@NetBSD.org, gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,>
From: Joerg Sonnenberger <joerg@britannica.bec.de>
List: netbsd-bugs
Date: 12/15/2007 10:40:02
The following reply was made to PR kern/37508; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg@britannica.bec.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/37508: After jmcneill-pm, disks are switched to standby
	before rebooting
Date: Sat, 15 Dec 2007 10:25:47 +0100

 On Mon, Dec 10, 2007 at 04:50:02PM +0000, Manuel Bouyer wrote:
 >  It may be OK for a power down (there is even a PR for this: kern/32766:
 >  some macppc systems rely on the system doing the wd_standby() before powering
 >  down, the firmware doesn't do it). But it's definitively not OK
 >  for a reboot, or a halt without power down.
 
 I would even reference the older PR kern/21531.
 
 >  From the discussion about the PR mentionned above, it came that we should
 >  not spin down a drive on reboot or halt without -p, and an extra argument to
 >  the PM callbacks is needed. 
 
 I would like to avoid having to carry an extra argument that is used
 only by a few drivers. What do you think of extending boothowto to
 actually have a RB_REBOOT flag and use that to decide if we are
 currently going to suspend the system/drive or reboot/halt? I have a
 patch for that.
 
 Joerg