Re: Re: HDAPS, Need to park the head for real

[Jens Axboe <axboe-l3A5Bk7waGM@public.gmane.org>]

On Thu, Aug 25 2005, Jon Escombe wrote:
> Alan Cox wrote:
> >@@ -1661,6 +1671,9 @@
> >                where = ELEVATOR_INSERT_FRONT;
> >                rq->flags |= REQ_PREEMPT;
> >        }
> >+       if (action == ide_next)
> >+               where = ELEVATOR_INSERT_FRONT;
> >+
> >        __elv_add_request(drive->queue, rq, where, 0);
> >        ide_do_request(hwgroup, IDE_NO_IRQ);
> >        spin_unlock_irqrestore(&ide_lock, flags);
> >
> >Also puzzles me- why is this needed ?
> 
> I wanted the park command to get in at the head of the queue (behind the 
> currently executing request).
> 
> Contrary to the comments for ide_do_drive_cmd(), ide_next didn't appear 
> to do anything to achieve this? At least from my initial testing before 
> I made this change - it could take a second or so for the park command 
> to be issued if the disk was busy....

That part seems to have been lost, apparently. The above patch is
correct.

-- 
Jens Axboe



-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf