Re: [RFC PATCHv2 3/4] drivers/otp: allow an ioctl to be specified

[Jamie Iles <jamie@jamieiles.com>]

Hi Mike,

On Thu, Mar 24, 2011 at 02:35:00PM -0400, Mike Frysinger wrote:
> On Thu, Mar 24, 2011 at 11:21, Jamie Iles wrote:
> > Some drivers may need an ioctl method to provide device specific control
> > such as blackfin for providing locking.  Regions can specify an ioctl to
> > be used but this field method is optional.
> 
> shouldnt this just be squashed into patch 1 since you havent merged
> any of this yet ?

Probably yes, I was just trying to keep the patches smaller to review, 
but in this case it probably isn't a sensible split.

> 
> > +static long otp_ioctl(struct file *filp, unsigned cmd, unsigned long arg)
> > +{
> > +       struct otp_region *region = filp->private_data;
> > +       long ret = -ENOTTY;
> > +
> > +       if (down_interruptible(&otp_sem))
> > +               return -ERESTARTSYS;
> > +
> > +       if (region->ops->ioctl)
> > +               ret = region->ops->ioctl(region, cmd, arg);
> > +
> > +       up(&otp_sem);
> > +
> > +       return ret;
> > +}
> 
> the existence check can be pulled out of the locking
> 
> long ret;
> 
> if (!region->ops->ioctl)
>     return -ENOTTY;
> 
> if (down_interruptible(&otp_sem))
> 
>     return -ERESTARTSYS;
> 
> ret = region->ops->ioctl(region, cmd, arg);
> 
> up(&otp_sem);
> 
> return ret;
>
Ok, will do.

Jamie