Re: [PATCH v6 1/1] input: add driver for Bosch Sensortec's BMA150 accelerometer

[Eric Andersson <eric.andersson@unixphere.com>]

On 22:15 Sun 07 Aug     , Dmitry Torokhov wrote:
> On Sun, Aug 07, 2011 at 06:23:25PM +0200, Eric Andersson wrote:
> > > On Sun, Jul 31, 2011 at 12:49:17AM +0200, Dmitry Torokhov wrote:
> > > > > > +static int bma150_open(struct bma150_data *bma150)
> > > > > > +{
> > > > > > + int ret;
> > > > > > +
> > > > > > + ret = pm_runtime_set_active(&bma150->client->dev);
> > > > > > + if (ret < 0)
> > > > > > +         return ret;
> > > > > > +
> > > > > > + pm_runtime_enable(&bma150->client->dev);
> > > > >
> > > > > I am pretty sure you want to call pm_runtime_enable() in bma150_probe()
> > > > > so that parent controller can be suspended until somebody calls
> > > > > bma150_open() and we mark the device as active (which, in turn, should
> > > > > wake up its parent).
> > > >
> > > > If I call pm_runtime_enable() I cannot use pm_runtime_set_active() later
> > > > according to the comment in __pm_runtime_set_status (runtime.c):
> > > > "If runtime PM of the device is disabled or its power.runtime_error
> > > > field is different from zero, the status may be changed either to
> > > > RPM_ACTIVE, or to RPM_SUSPENDED..."
> > > >
> > > > If the PM of the device is enabled it will return -EAGAIN. Of course, we
> > > > could enable() in probe, then disable(); set_active(); enable(); in
> > > > open, but that seems a bit confused, right?
> > > 
> > > Hmm, indeed. I do not like the idea about disable/set_active/enable so I
> > > guess we'll have to keep track of the current mode themselves and call
> > > bma150_set_mode() ourselves if, after calling pm_runtime_get/put_sync()
> > > we find that our mode is different from what we expect it to be.
> > > 
> > > I also noticed that you did not properly free IRQ on device removal and
> > > also polled devices need to be freed always (unlike regular input
> > > devices that should be only unregistered). The default_cfg can't be
> > > __initdata because it is used by __devinit functions. And my version of
> > > GCC can't figure out that ipoll_dev never used uninitialized and gives
> > > false warning.
> > > 
> > > I tried correcting these issues in the patch below, along with renaming
> > > 'ret' to 'error' which I prefer when we dealing with error handling.
> > > Could you please git it a try and if everything still works I'll fold it
> > > and commit.
> > 
> > Thanks Dmitry! I tried your patch and it worked like a charm!
> 
> Excellent, thanks for testing.
> 
> > 
> > Do you take it from here or would you like me to send an updated
> > version that includes your patch?
> 
> No, there is no need for that. I'll fold them all together and queue for
> 2.6.32.
Great thanks! Don't forget to add Alan as reviewer. I think he forgot
reply-all when he mailed me:
"I'm happy with this one.
Reviewed-by: Alan Cox <alan@linux.intel.com>"

> I take the other patch that makes polled devices do poll upon opening
> working well for you too.. Should I queue for .2 as well?
Yes, why not!

-- 
Best regards,
 Eric

http://www.unixphere.com
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html