Re: [PATCH v7 3/4] block: implement runtime pm strategy

[Aaron Lu <aaron.lu@intel.com>]

On Wed, Jan 16, 2013 at 10:30:45AM -0500, Alan Stern wrote:
> On Wed, 16 Jan 2013, Aaron Lu wrote:
> 
> > From: Lin Ming <ming.m.lin@intel.com>
> > 
> > When a request is added:
> >     If device is suspended or is suspending and the request is not a
> >     PM request, resume the device.
> > 
> > When the last request finishes:
> >     Call pm_runtime_mark_last_busy().
> > 
> > When pick a request:
> >     If device is resuming/suspending, then only PM request is allowed
> >     to go.
> > 
> > Signed-off-by: Lin Ming <ming.m.lin@intel.com>
> > Signed-off-by: Aaron Lu <aaron.lu@intel.com>
> 
> Just a couple of minor problems remaining...
> 
> > --- a/block/blk-core.c
> > +++ b/block/blk-core.c
> 
> > @@ -2051,6 +2063,28 @@ static void blk_account_io_done(struct request *req)
> >  	}
> >  }
> >  
> > +#ifdef CONFIG_PM_RUNTIME
> > +/*
> > + * Don't process normal requests when queue is suspended
> > + * or in the process of suspending/resuming
> > + */
> > +static struct request *blk_pm_peek_request(struct request_queue *q,
> > +					   struct request *rq)
> > +{
> > +	if (q->rpm_status == RPM_SUSPENDED ||
> > +	    (q->rpm_status != RPM_ACTIVE && !(rq->cmd_flags & REQ_PM)))
> > +		return NULL;
> > +	else
> > +		return rq;
> > +}
> 
> You don't check q->dev here, so the result is indefinite for devices
> that don't use runtime PM.  (Actually it will work out because
> RPM_ACTIVE is defined as 0, but that's a pretty fragile approach.)
> 
> Either do check q->dev here, or else explicitly initialize
> q->rpm_status when the queue is created.
> 
> 
> > --- a/block/elevator.c
> > +++ b/block/elevator.c
> > @@ -34,6 +34,7 @@
> >  #include <linux/blktrace_api.h>
> >  #include <linux/hash.h>
> >  #include <linux/uaccess.h>
> > +#include <linux/pm_runtime.h>
> >  
> >  #include <trace/events/block.h>
> >  
> > @@ -515,6 +516,27 @@ void elv_bio_merged(struct request_queue *q, struct request *rq,
> >  		e->type->ops.elevator_bio_merged_fn(q, rq, bio);
> >  }
> >  
> > +#ifdef CONFIG_PM_RUNTIME
> > +static void blk_pm_requeue_request(struct request *rq)
> > +{
> > +	if (!(rq->cmd_flags & REQ_PM))
> > +		rq->q->nr_pending--;
> > +}
> 
> You don't check q->dev here.  That's okay, but it means that
> q->nr_pending will be meaningless or wrong if any I/O takes place
> before blk_pm_runtime_init is called.
> 
> Therefore the kerneldoc for blk_pm_runtime_init should mention that it
> must not be called after any requests have been submitted.  Also

------------
> mention that blk_pm_runtime_init enables runtime PM for q->dev, so the 
> caller shouldn't do it.

I may misunderstandd this in last email.

We didn't enable runtime PM for the device in blk_pm_runtime_init, just
some auto suspend related settings.

Thanks,
Aaron