[PATCH] mmc: dw_mmc: No bus setup if default pinctrl is used

[PATCH] mmc: dw_mmc: No bus setup if default pinctrl is used

[Dongjin Kim]

Legacy gpio based bus setup is failed when the bus pins are listed with the
property 'pinctrl-0' and 'pinctrl-names', instead of 'gpios'. The default
pinctrls are handled by device core before probe, no need to configure the
pins again in a device driver.

[    1.190000] Synopsys Designware Multimedia Card Interface Driver
[    1.195000] dwmmc_exynos 12550000.mshc: Using PIO mode.
[    1.200000] dwmmc_exynos 12550000.mshc: DW MMC controller at irq 109, 32 bit host data width, 128 deep fifo
[    1.210000] dwmmc_exynos 12550000.mshc: invalid gpio: -2
[    1.215000] dwmmc_exynos: probe of 12550000.mshc failed with error -22

Cc: Thomas Abraham <thomas.ab@samsung.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Dongjin Kim <tobetter@gmail.com>
---
 drivers/mmc/host/dw_mmc-exynos.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/host/dw_mmc-exynos.c b/drivers/mmc/host/dw_mmc-exynos.c
index aafd485..f568c32 100644
--- a/drivers/mmc/host/dw_mmc-exynos.c
+++ b/drivers/mmc/host/dw_mmc-exynos.c
@@ -156,12 +156,19 @@ static int dw_mci_exynos_setup_bus(struct dw_mci *host,
 				struct device_node *slot_np, u8 bus_width)
 {
 	int idx, gpio, ret;
+	int nr_pins = NUM_PINS(bus_width);
 
 	if (!slot_np)
 		return -EINVAL;
 
+#if defined(CONFIG_PINCTRL)
+	/* Default pinctrl is used */
+	if (!IS_ERR(host->dev->pins->default_state))
+		nr_pins = 0;
+#endif
+
 	/* cmd + clock + bus-width pins */
-	for (idx = 0; idx < NUM_PINS(bus_width); idx++) {
+	for (idx = 0; idx < nr_pins; idx++) {
 		gpio = of_get_gpio(slot_np, idx);
 		if (!gpio_is_valid(gpio)) {
 			dev_err(host->dev, "invalid gpio: %d\n", gpio);
-- 
1.7.10.4

Re: [PATCH] mmc: dw_mmc: No bus setup if default pinctrl is used

[Linus Walleij]

On Thu, Feb 7, 2013 at 7:17 PM, Dongjin Kim <tobetter@gmail.com> wrote:

> Legacy gpio based bus setup is failed when the bus pins are listed with the
> property 'pinctrl-0' and 'pinctrl-names', instead of 'gpios'. The default
> pinctrls are handled by device core before probe, no need to configure the
> pins again in a device driver.

Aha.

>  {
>         int idx, gpio, ret;
> +       int nr_pins = NUM_PINS(bus_width);
>
>         if (!slot_np)
>                 return -EINVAL;
>
> +#if defined(CONFIG_PINCTRL)
> +       /* Default pinctrl is used */
> +       if (!IS_ERR(host->dev->pins->default_state))
> +               nr_pins = 0;
> +#endif

Use:

if (IS_ENABLED(CONFIG_PINCTRL))
   nr_pins = 0;

Without ifdefs. Grep kernel for examples.
Read <linux/kconfig.h> for explanations.

Put a comment above this to explain exactly why
you do this.

> +
>         /* cmd + clock + bus-width pins */
> -       for (idx = 0; idx < NUM_PINS(bus_width); idx++) {
> +       for (idx = 0; idx < nr_pins; idx++) {
>                 gpio = of_get_gpio(slot_np, idx);
>                 if (!gpio_is_valid(gpio)) {
>                         dev_err(host->dev, "invalid gpio: %d\n", gpio);

This looks kludgy, is it so that the driver falls back to GPIO
if it cannot use pinctrl? Can't it just depend on CONFIG_PINCTRL
in Kconfig or do you absolutely have to support both cases?

I was under the impression that the Exynos requires pinctrl.

Yours,
Linus Walleij

Re: [PATCH] mmc: dw_mmc: No bus setup if default pinctrl is used

[Dongjin Kim]

Hi Linus,

Thank you for reviewing.

Initially I also considered to use IS_ENABLED(CONFIG_PINCTRL) and
ignore for-block according to its status. But it would make a certain
board failed if gpios is used for pin configuration rather than
default-pinctrl even if CONFIG_PINCTRL is enabled. Also considered to
add a function on pinctrl.c to return if default-pinctrl is used, and
drop down the for-block according to its status. It could replace the
long and complicaed line, IS_ERR(host->dev->pins->default_state), to
simple one as well. :)

The best is obviously to enable CONFIG_PINCTRL and remove the
for-block, it makes the code more simple. I like to have a comment
from Thomas if have to support both cases, gpios and default-pinctrl.

Best regards,
Dongjin.

On Fri, Feb 8, 2013 at 4:03 AM, Linus Walleij <linus.walleij@linaro.org> wrote:
> On Thu, Feb 7, 2013 at 7:17 PM, Dongjin Kim <tobetter@gmail.com> wrote:
>
>> Legacy gpio based bus setup is failed when the bus pins are listed with the
>> property 'pinctrl-0' and 'pinctrl-names', instead of 'gpios'. The default
>> pinctrls are handled by device core before probe, no need to configure the
>> pins again in a device driver.
>
> Aha.
>
>>  {
>>         int idx, gpio, ret;
>> +       int nr_pins = NUM_PINS(bus_width);
>>
>>         if (!slot_np)
>>                 return -EINVAL;
>>
>> +#if defined(CONFIG_PINCTRL)
>> +       /* Default pinctrl is used */
>> +       if (!IS_ERR(host->dev->pins->default_state))
>> +               nr_pins = 0;
>> +#endif
>
> Use:
>
> if (IS_ENABLED(CONFIG_PINCTRL))
>    nr_pins = 0;
>
> Without ifdefs. Grep kernel for examples.
> Read <linux/kconfig.h> for explanations.
>
> Put a comment above this to explain exactly why
> you do this.
>
>> +
>>         /* cmd + clock + bus-width pins */
>> -       for (idx = 0; idx < NUM_PINS(bus_width); idx++) {
>> +       for (idx = 0; idx < nr_pins; idx++) {
>>                 gpio = of_get_gpio(slot_np, idx);
>>                 if (!gpio_is_valid(gpio)) {
>>                         dev_err(host->dev, "invalid gpio: %d\n", gpio);
>
> This looks kludgy, is it so that the driver falls back to GPIO
> if it cannot use pinctrl? Can't it just depend on CONFIG_PINCTRL
> in Kconfig or do you absolutely have to support both cases?
>
> I was under the impression that the Exynos requires pinctrl.
>
> Yours,
> Linus Walleij