Hi Krzysztof,
On Mon, 2022-09-19 at 11:38 +0200, Krzysztof Kozlowski wrote:
> On 16/09/2022 14:07, B. Niedermayr wrote:
> > From: Benedikt Niedermayr <benedikt.niedermayr@xxxxxxxxxxx>
> >
> > The waitpin polarity can be configured via the WAITPIN<X>POLARITY bits
> > in the GPMC_CONFIG register. This is currently not supported by the
> > driver. This patch adds support for setting the required register bits
> > with the "gpmc,wait-pin-polarity" dt-property.
> >
> > Signed-off-by: Benedikt Niedermayr <benedikt.niedermayr@xxxxxxxxxxx>
> > ---
> > drivers/memory/omap-gpmc.c | 27 +++++++++++++++++++++++++
> > include/linux/platform_data/gpmc-omap.h | 6 ++++++
> > 2 files changed, 33 insertions(+)
> >
> > diff --git a/drivers/memory/omap-gpmc.c b/drivers/memory/omap-gpmc.c
> > index ea495e93766b..2853fc28bccc 100644
> > --- a/drivers/memory/omap-gpmc.c
> > +++ b/drivers/memory/omap-gpmc.c
> > @@ -132,6 +132,7 @@
> > #define GPMC_CONFIG_DEV_SIZE 0x00000002
> > #define GPMC_CONFIG_DEV_TYPE 0x00000003
> >
> > +#define GPMC_CONFIG_WAITPINPOLARITY(pin) (BIT(pin) << 8)
> > #define GPMC_CONFIG1_WRAPBURST_SUPP (1 << 31)
> > #define GPMC_CONFIG1_READMULTIPLE_SUPP (1 << 30)
> > #define GPMC_CONFIG1_READTYPE_ASYNC (0 << 29)
> > @@ -1882,6 +1883,17 @@ int gpmc_cs_program_settings(int cs, struct gpmc_settings *p)
> >
> > gpmc_cs_write_reg(cs, GPMC_CS_CONFIG1, config1);
> >
> > + if (p->wait_pin_polarity != WAITPINPOLARITY_DEFAULT) {
> > + config1 = gpmc_read_reg(GPMC_CONFIG);
> > +
> > + if (p->wait_pin_polarity == WAITPINPOLARITY_ACTIVE_LOW)
> > + config1 &= ~GPMC_CONFIG_WAITPINPOLARITY(p->wait_pin);
> > + else if (p->wait_pin_polarity == WAITPINPOLARITY_ACTIVE_HIGH)
> > + config1 |= GPMC_CONFIG_WAITPINPOLARITY(p->wait_pin);
> > +
> > + gpmc_write_reg(GPMC_CONFIG, config1);
>
> What happens if wait pin is shared and you have different polarities in
> both of devices?
In this case the second one wins and will overwrite the polarity of the first one.
But that would be the result of a misconfiguration in the DT.
I'm not sure how to proceed here? Does it make sense to add a check for different
waitpin polarities?
>
> > + }
> > +
> > return 0;
> > }
> >
> > @@ -1981,7 +1993,22 @@ void gpmc_read_settings_dt(struct device_node *np, struct gpmc_settings *p)
> > __func__);
> > }
> >
> > + p->wait_pin_polarity = WAITPINPOLARITY_DEFAULT;
> > +
> > if (!of_property_read_u32(np, "gpmc,wait-pin", &p->wait_pin)) {
> > + if (!of_property_read_u32(np, "gpmc,wait-pin-polarity",
> > + &p->wait_pin_polarity)) {
> > + if (p->wait_pin_polarity != WAITPINPOLARITY_ACTIVE_HIGH &&
> > + p->wait_pin_polarity != WAITPINPOLARITY_ACTIVE_LOW &&
> > + p->wait_pin_polarity != WAITPINPOLARITY_DEFAULT) {
>
> WAITPINPOLARITY_DEFAULT is not allowed in DT, so you can skip it.
This value is not assigned from the DT. It is only assigned within the GPMC and serves as a init
value (right before the if clause). This helps in case no configuration from DT is done where the
GPMC registers should stay untouched.
>
> > + pr_err("%s: Invalid wait-pin-polarity (pin: %d, pol: %d)\n",
>
> dev_err, not pr_err
Ok. I didn't want to introduce dev_* functions. Currently pr_* functions where used all over the place.
>
> > + __func__, p->wait_pin, p->wait_pin_polarity);
>
> Skip __func__
>
Ok.
> > + p->wait_pin_polarity = WAITPINPOLARITY_DEFAULT;
> > + }
> > + } else {
> > + pr_err("%s: Failed to read gpmc,wait-pin-polarity\n", __func__);
>
> Ditto.
Ok.
>
> > + }
> > +
> > p->wait_on_read = of_property_read_bool(np,
> > "gpmc,wait-on-read");
> > p->wait_on_write = of_property_read_bool(np,
> > diff --git a/include/linux/platform_data/gpmc-omap.h b/include/linux/platform_data/gpmc-omap.h
> > index c9cc4e32435d..c46c28069c31 100644
> > --- a/include/linux/platform_data/gpmc-omap.h
> > +++ b/include/linux/platform_data/gpmc-omap.h
> > @@ -136,6 +136,11 @@ struct gpmc_device_timings {
> > #define GPMC_MUX_AAD 1 /* Addr-Addr-Data multiplex */
> > #define GPMC_MUX_AD 2 /* Addr-Data multiplex */
> >
> > +/* Wait pin polarity values */
> > +#define WAITPINPOLARITY_DEFAULT -1
>
> Missing prefix. This is a global header.
Ok. Makes sense.
>
> > +#define WAITPINPOLARITY_ACTIVE_LOW 0
> > +#define WAITPINPOLARITY_ACTIVE_HIGH 1
> > +
> > struct gpmc_settings {
> > bool burst_wrap; /* enables wrap bursting */
> > bool burst_read; /* enables read page/burst mode */
> > @@ -149,6 +154,7 @@ struct gpmc_settings {
> > u32 device_width; /* device bus width (8 or 16 bit) */
> > u32 mux_add_data; /* multiplex address & data */
> > u32 wait_pin; /* wait-pin to be used */
> > + u32 wait_pin_polarity; /* wait-pin polarity */
>
> Skip the comment. You just copied the name of variable. Such comments
> are useless.
>
> We do not have KPIs in kernel to achieve some comment-ratio...
>
Ok, makes sense.
> Best regards,
> Krzysztof
