Re: Head parking (was: IBM HDAPS things are looking up)

[Jens Axboe <axboe@suse.de>]

On Thu, Jul 07 2005, Lenz Grimmer wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi,
> 
> Jens Axboe wrote:
> 
> > ATA7 defines a park maneuvre, I don't know how well supported it is 
> > yet though. You can test with this little app, if it says 'head 
> > parked' it works. If not, it has just idled the drive.
> 
> Great! Thanks for digging this up - it works on my T42, using a Fujitsu
> MHT2080AH drive:
> 
>   lenz@metis:~/work/ibm_hdaps> sudo ./headpark /dev/hda
>   head parked
>
> Judging from the sound the drive makes, this is the same operation that
> the windows tool performs.

Very cool, I wasn't sure if this was a 'new' feature waiting to be
implemented in drives or if ata7 just documented existing use in some
drives.

How long did the park take? Spec states it can take up to 500ms.

> However, the head does not remain parked for a very long time,
> especially if there is a lot of disk activity going on (I tested it by
> running a "find /" in parallel). The head parks, but leaves the park
> position immediately afterwards again. I guess now we need to find a way
> to "nail" the head into the parking position for some time - otherwise
> it may already be on its way back to the platter before the laptop hits
> the ground...

The head just remains parked until the next command is issued. This
needs to be combined with some ide help, to freeze the queue. Perhaps
some sysfs file so you could do

# echo park > /sys/block/hda/device/head_state

Or whatever, at least just exposing this possibility so that the drive
internally can block future io until a 'resume' command is issued.

-- 
Jens Axboe