| Message ID | 20171005153508.32127-1-simon.budig@kernelconcepts.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Oct 05, 2017 at 05:35:06PM +0200, simon.budig@kernelconcepts.de wrote: > From: Simon Budig <simon.budig@kernelconcepts.de> > > For some reason the reset pin no longer gets toggeled when initializing > the touch. Fix that and restore the old behaviour. Hmm, the GPIO is requested as GPIOD_OUT_HIGH, so it should be driven low to being with and then released... I am not sure why we need to drive it low explicitly again. > > Signed-off-by: Simon Budig <simon.budig@kernelconcepts.de> > --- > drivers/input/touchscreen/edt-ft5x06.c | 6 +++++- > 1 file changed, 5 insertions(+), 1 deletion(-) > > diff --git a/drivers/input/touchscreen/edt-ft5x06.c b/drivers/input/touchscreen/edt-ft5x06.c > index 5bf63f7..867a61e 100644 > --- a/drivers/input/touchscreen/edt-ft5x06.c > +++ b/drivers/input/touchscreen/edt-ft5x06.c > @@ -936,8 +936,12 @@ static int edt_ft5x06_ts_probe(struct i2c_client *client, > > if (tsdata->reset_gpio) { > usleep_range(5000, 6000); > - gpiod_set_value_cansleep(tsdata->reset_gpio, 0); > + /* note that the reset pin needs to be registered as > + * active low in the devicetree. > + */ > + gpiod_set_value_cansleep(tsdata->reset_gpio, 1); > msleep(300); > + gpiod_set_value_cansleep(tsdata->reset_gpio, 0); > } > > input = devm_input_allocate_device(&client->dev); > -- > 2.1.4 > Thanks.
Hi Dmitry. On Mon, 2017-10-09 at 20:58 -0700, Dmitry Torokhov wrote: > On Thu, Oct 05, 2017 at 05:35:06PM +0200, simon.budig@kernelconcepts. > de wrote: > > From: Simon Budig <simon.budig@kernelconcepts.de> > > > > For some reason the reset pin no longer gets toggeled when > > initializing > > the touch. Fix that and restore the old behaviour. > > Hmm, the GPIO is requested as GPIOD_OUT_HIGH, so it should be driven > low > to being with and then released... I am not sure why we need to drive > it > low explicitly again. It is possible that I misinterpreted something here - at some point I did have the GPIO pin wrongly registered in the devicetree. However, API-wise it is not clear to me, that GPIOD_OUT_HIGH results in an actual low state of the Pin. That might be different if the constant was called GPIOD_OUT_ACTIVE or something, since this would refer to the "low_active" state of the pin. *If* that is actually the case (i.e. requesting an pin specified as GPIO_ACTIVE_LOW in the devicetree with the GPIOD_OUT_HIGH flag results in 0V on this pin) then this patch should probably be dropped, although I don't like how this code reads then. I also reread the focaltec datasheet and I've messed up the delays in this patch, so if the above does *not* happen then I'd need to rework this... Bye, Simon
On Tue, 2017-10-10 at 12:25 +0200, Simon Budig wrote: > *If* that is actually the case (i.e. requesting an pin specified as > GPIO_ACTIVE_LOW in the devicetree with the GPIOD_OUT_HIGH flag > results in 0V on this pin) then this patch should probably be > dropped, although I don't like how this code reads then. Ok, I've traced the code in the gpio-subsystem and indeed, requesting with the GPIOD_OUT_HIGH flag results in a LOW leveGPIOD_OUT_HIGH upon request when the GPIO_ACTIVE_LOW is set on that pin. That constant really should've been called GPIOD_OUT_ACTIVE or something. That reads confusing as heck. Anyway, please drop this patch. It doesn't actually do any good. Thanks, Simon
diff --git a/drivers/input/touchscreen/edt-ft5x06.c b/drivers/input/touchscreen/edt-ft5x06.c index 5bf63f7..867a61e 100644 --- a/drivers/input/touchscreen/edt-ft5x06.c +++ b/drivers/input/touchscreen/edt-ft5x06.c @@ -936,8 +936,12 @@ static int edt_ft5x06_ts_probe(struct i2c_client *client, if (tsdata->reset_gpio) { usleep_range(5000, 6000); - gpiod_set_value_cansleep(tsdata->reset_gpio, 0); + /* note that the reset pin needs to be registered as + * active low in the devicetree. + */ + gpiod_set_value_cansleep(tsdata->reset_gpio, 1); msleep(300); + gpiod_set_value_cansleep(tsdata->reset_gpio, 0); } input = devm_input_allocate_device(&client->dev);