Re: [PATCH v4 04/11] leds: aw200xx: calculate dts property display_rows in the driver

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, 24 Nov 2023, Dmitry Rokosov wrote:

> On Thu, Nov 23, 2023 at 04:32:52PM +0000, Lee Jones wrote:
> > On Tue, 21 Nov 2023, Dmitry Rokosov wrote:
> > 
> > > From: George Stark <gnstark@xxxxxxxxxxxxxxxxx>
> > > 
> > > Get rid of device tree property "awinic,display-rows". The property
> > > value actually means number of current switches and depends on how leds
> > 
> > Nit: LEDs
> > 
> > > are connected to the device. It should be calculated manually by max
> > > used led number. In the same way it is computed automatically now.
> > 
> > As above - I won't mention this again.
> > 
> > > Max used led is taken from led definition subnodes.
> > > 
> > > Signed-off-by: George Stark <gnstark@xxxxxxxxxxxxxxxxx>
> > > Signed-off-by: Dmitry Rokosov <ddrokosov@xxxxxxxxxxxxxxxxx>
> > > ---
> > >  drivers/leds/leds-aw200xx.c | 39 +++++++++++++++++++++++++------------
> > >  1 file changed, 27 insertions(+), 12 deletions(-)
> > > 
> > > diff --git a/drivers/leds/leds-aw200xx.c b/drivers/leds/leds-aw200xx.c
> > > index 7762b3a132ac..4bce5e7381c0 100644
> > > --- a/drivers/leds/leds-aw200xx.c
> > > +++ b/drivers/leds/leds-aw200xx.c
> > > @@ -379,6 +379,30 @@ static void aw200xx_disable(const struct aw200xx *const chip)
> > >  	return gpiod_set_value_cansleep(chip->hwen, 0);
> > >  }
> > >  
> > > +static bool aw200xx_probe_get_display_rows(struct device *dev, struct aw200xx *chip)
> > > +{
> > > +	struct fwnode_handle *child;
> > > +	u32 max_source = 0;
> > > +
> > > +	device_for_each_child_node(dev, child) {
> > > +		u32 source;
> > > +		int ret;
> > > +
> > > +		ret = fwnode_property_read_u32(child, "reg", &source);
> > > +		if (ret || source >= chip->cdef->channels)
> > 
> > Shouldn't the second clause fail instantly?
> > 
> 
> We already have such logic in the aw200xx_probe_fw() function, which
> skips the LED node with the wrong reg value too. Furthermore, we have
> strict reg constraints in the dt-bindings parts (in the current patch
> series), so we assume that the DT developer will not create an LED with
> the wrong reg value.

Why is it being checked again then?

-- 
Lee Jones [李琼斯]




[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux