Re: [PATCHv1] staging: Synaptics RMI4 touchpad driver support

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

On Wed, Nov 03, 2010 at 04:37:14PM -0500, Arce, Abraham wrote:
> Hi Naveen,
> 
> > From: linux-input-owner@vger.kernel.org [mailto:linux-input-
> > owner@vger.kernel.org] On Behalf Of Greg KH
> > Sent: Wednesday, November 03, 2010 1:20 PM
> > To: Dmitry Torokhov
> > Cc: Linus Walleij; Naveen Kumar GADDIPATI; linux-kernel@vger.kernel.org;
> > STEricsson_nomadik_linux; linux-input@vger.kernel.org
> > Subject: Re: [PATCHv1] staging: Synaptics RMI4 touchpad driver support
> > 
> > On Wed, Nov 03, 2010 at 12:07:52PM -0700, Dmitry Torokhov wrote:
> > > On Wed, Nov 03, 2010 at 11:10:58AM -0700, Greg KH wrote:
> > > > On Tue, Nov 02, 2010 at 08:40:30AM -0700, Dmitry Torokhov wrote:
> > > > > On Tue, Nov 02, 2010 at 03:14:48PM +0100, Linus Walleij wrote:
> > > > > > Greg KH wrote:
> > > > > >
> > > > > > >> +Wait for the official upstream synaptics rmi4 clearpad drivers as
> > promised over the past few months
> > > > > > >> +Merge any device support needed from this driver into it
> > > > > > >> +Delete this driver
> > > > > > >
> > > > > > > Huh?
> > > > > > >
> > > > > > > Why not just add this driver to the kernel tree instead?  When the
> > > > > > > "promised" driver then eventually shows up (who is promising it?)
> > then
> > > > > > > delete the thing.
> > > > >
> > > > > The same reason as there were several wireless drivers in staging?
> > > >
> > > > Ok, that's a good enough reason for me :)
> > > >
> > > > > > Well, Alan (on behalf of Ramesh Agarwal) sent out a very similar
> > > > > > patch (Titled "Synaptics TM1217 Touchscreen Controller driver")
> > > > > > the other day, and OTOMH that was after discussions with Synaptics
> > > > > > where they said they were working on a "real" driver (a rather
> > > > > > complex RMI4 bus driver) and we believe they will fix that sooner
> > > > > > or later.
> > > > > >
> > > > > > So, until sooner or later happens we thought we'd keep it in staging.
> > > > > >
> > > > > > If you prefer both Alan and we can probably submit our patches for
> > > > > > inclusion in the proper place.
> > > > > >
> > > > > > A side effect may be that the Synaptics RMI4 people may have trouble
> > > > > > to merge their driver into input/ since they may be requested to
> > > > > > refactor the existing drivers to use it rather than merging new stuff,
> > > > > > putting some burden on their shoulders.
> > > > > >
> > > > >
> > > > > No, they have not been requested to change any existing [in tree]
> > > > > drivers. They however been asked to convert to the driver core
> > > > > primitives instead of rolling their own infprastructure to implement
> > > > > devices and drivers binding.
> > > >
> > > > If you don't object to this driver going in, and there is a path forward
> > > > in the future for it to be able to be removed, then I will be glad to
> > > > add it.
> > > >
> > > > Can I add your Acked-by: to the patch for this?
> > > >
> > >
> > > Umm... I haven't looked in detail, but I think it is sane enough for
> > > staging... Much better than Alan's version that seemed to create a
> > > separate input device for every finger. Needs to depend on INPUT and
> > > probably regulators framework...  Probe function may call
> > > input_free_device() after calling input_register_device() which is not
> > > good.
> > 
> > Great, thanks for this.
> > 
> > > Basically staging is yours, you can add whatever you want to it ;)
> > 
> > Heh, well, I try to at least cooperate with the different subsystem
> > maintainers.  I don't need even more people pissed-off at me than I
> > normally get :)
> > 
> > I'll queue this up for .38.
> > 
> 
> I have done some initial testing with this patch, it includes some modifications to the board file, I have a gpio acting as irq plus driver code itself
> 
> I won't be able to get events (board will hang) unless adding the following code, is this due to the usage of a gpio as interrupt? If you could please clarify...
> 
> @@ -481,6 +483,9 @@ static int synaptics_rmi4_sensor_report(struct synaptics_rmi4_data *pdata)
>                                                                         rfi);
>                 }
>         }
> +
> +       enable_irq(gpio_to_irq(pdata->irq_number));
> +
>         /* return the number of touch points */
>         return touch;
>  }
> @@ -498,6 +503,8 @@ static irqreturn_t synaptics_rmi4_irq(int irq, void *data)
>         struct synaptics_rmi4_data *pdata = data;
>         int touch_count;
>         do {
> +               disable_irq_nosync(gpio_to_irq(pdata->irq_number));
> +
>                 touch_count = synaptics_rmi4_sensor_report(pdata);
>                 if (touch_count)
>                         wait_event_timeout(pdata->wait, pdata->touch_stopped,
> 

It probably wants a threaded IRQ...

-- 
Dmitry