Re: [PATCH try #2] Input/Joystick Driver: add support AD7142 joystick driver

["Dmitry Torokhov" <dmitry.torokhov@gmail.com>]

On 10/12/07, Bryan Wu <bryan.wu@analog.com> wrote:
> On Fri, 2007-10-12 at 11:50 -0400, Dmitry Torokhov wrote:
> > On 10/12/07, Bryan Wu <bryan.wu@analog.com> wrote:
> > > On Fri, 2007-10-12 at 10:49 -0400, Dmitry Torokhov wrote:
> > > > Hi Bryan,
> > > >
> > > > On 10/12/07, Bryan Wu <bryan.wu@analog.com> wrote:
> > > > > +
> > > [snip]
> > > > > +
> > > > > +static void ad7142_close(struct input_dev *dev)
> > > > > +{
> > > > > +       free_irq(CONFIG_BFIN_JOYSTICK_IRQ_PFX, ad7142_interrupt);
> > > > > +       kthread_stop(ad7142_task);
> > > >
> > > > Don't you need to write something over i2c to shut the devoce off?
> > > > What stops it from continuing to generate interrupts?
> > > >
> > > Actually, I am going to use "completion" to replace the whole
> > > wait_interrupt_xxx and intr_flag things which original from Aubrey. How
> > > do you think of that?
> > >
> >
> > I don't think it is a very good idea - for me completion is one-time
> > deal. You use it and then you are done. How about firing a work from
> > interrupt and either rely on the default workqueue (keventd) or create
> > your own to execute it?
> >
>
> completion is a wrapper of workqueue and simpler to use.
> my method:
> 1. In kthread:
> do {
> |_______|_______wait_for_completion(&ad7142_completion);
> |_______|_______ad7142_decode();
> |_______|_______enable_irq(CONFIG_BFIN_JOYSTICK_IRQ_PFX);
> |_______} while (!kthread_should_stop());
>
> 2. In irq handler will fire "complete(&ad7142_completion);"
>
> This is simpler and understand easier
>

You also need to re-initialize completion every time you done
processing in kthread otherwise you will be constantly doing
ad7142_decode(). Plus, how are you going to stop kthread (completion
may not be signalled for a long time if nobody touches the joystick).

-- 
Dmitry