On Sat, 3 Jul 2010 10:59:56 +0200 Daniel Kobras wrote: [...] > On Wed, Jun 30, 2010 at 10:13:36PM +0200, Francesco Poli wrote: > > It seems that the new PATA libata-based drivers (see > > http://lists.debian.org/debian-kernel/2009/12/msg00587.html) > > turned /dev/hda to /dev/sda and make noflushd think it's dealing with a > > SCSI device. > > The comment in /etc/default/noflushd states: > > > > # [...] Noflushd currently only works with ide hard disks; > > # scsi disks require a kernel patch. [...] > > > > If my guess is correct, I assume that all (or almost all) noflushd > > users will be unable to go on using the package. > > You are correct.
Good, thanks for confirming! :-) > Therefore I'm raising the severity to grave. I agree that this is the most appropriate severity, given the status of this issue. > I need to > implement a new methode spindown_ata using the SG_IO interface instead of the > old IDE ioctl()s. I've just checked the hdparm source code, and the simple > tasks noflushd needs to perform require a bit more code than I was hoping, but > it's certainly doable without problems. This is good news, thanks for investigating and addressing this bug! > The second thing that needs to be done > is way to determine when to use the SG_IO interface. I certainly want to avoid > spinning down a SCSI disk that won't spin up by itself afterwards. Still needs > to investigate here. This sounds reasonable: I hope there's a simply way to tell actual SCSI disks from IDE disks driven by libata... > > Regards, Bye and many thanks for your time! :) -- http://www.inventati.org/frx/progs/scripts/pdebuild-hooks.html Need some pdebuild hook scripts? ..................................................... Francesco Poli . GnuPG key fpr == C979 F34B 27CE 5CD8 DC12 31B5 78F4 279B DD6D FCF4
pgpVaSsQaVAjG.pgp
Description: PGP signature
