[1/1] mmc: core: don't override the CD GPIO level when "cd-inverted" is set
diff mbox series

Message ID 20190101194414.24766-2-martin.blumenstingl@googlemail.com
State New
Headers show
Series
  • mmc: fix boards with CD GPIO and "cd-inverted"
Related show

Commit Message

Martin Blumenstingl Jan. 1, 2019, 7:44 p.m. UTC
Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
if "cd-inverted" is also set. This results in the "cd-inverted" property
being evaluated twice, which effectively makes it a no-op:
- first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
  setting up the CD GPIO
- then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
  reading the CD GPIO value at runtime

On boards which are using device-tree with the "cd-inverted" property
being set any inserted card are not detected anymore. This is due to the
MMC core treating the CD GPIO with the wrong polarity.

Disable "override_cd_active_level" for the card detection GPIO which is
parsed using mmc_of_parse. This fixes SD card detection on the boards
which are currently using the "cd-inverted" device-tree property (tested
on Meson8b Odroid-C1 and Meson8b EC-100).

This does not remove the CD GPIO inversion logic from the MMC core
because there's at least one driver (sdhci-pci-core for Intel BayTrail
based boards) which still passes "override_cd_active_level = true" to
mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
untouched.
In the future the GPIO inversion logic for both, card and read-only
detection can be removed once no driver is using it anymore.

Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
---
 drivers/mmc/core/host.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Linus Walleij Jan. 3, 2019, 10:05 p.m. UTC | #1
On Tue, Jan 1, 2019 at 8:44 PM Martin Blumenstingl
<martin.blumenstingl@googlemail.com> wrote:

> Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
> tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
> if "cd-inverted" is also set. This results in the "cd-inverted" property
> being evaluated twice, which effectively makes it a no-op:
> - first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
>   setting up the CD GPIO
> - then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
>   reading the CD GPIO value at runtime
>
> On boards which are using device-tree with the "cd-inverted" property
> being set any inserted card are not detected anymore. This is due to the
> MMC core treating the CD GPIO with the wrong polarity.
>
> Disable "override_cd_active_level" for the card detection GPIO which is
> parsed using mmc_of_parse. This fixes SD card detection on the boards
> which are currently using the "cd-inverted" device-tree property (tested
> on Meson8b Odroid-C1 and Meson8b EC-100).
>
> This does not remove the CD GPIO inversion logic from the MMC core
> because there's at least one driver (sdhci-pci-core for Intel BayTrail
> based boards) which still passes "override_cd_active_level = true" to
> mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
> untouched.
> In the future the GPIO inversion logic for both, card and read-only
> detection can be removed once no driver is using it anymore.
>
> Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>

OK two steps forward, one step back, that's a good fix for now, I
can fix it properly for v4.22.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>

Yours,
Linus Walleij
Martin Blumenstingl Jan. 8, 2019, 10:20 p.m. UTC | #2
Hi Linus,

On Thu, Jan 3, 2019 at 11:06 PM Linus Walleij <linus.walleij@linaro.org> wrote:
>
> On Tue, Jan 1, 2019 at 8:44 PM Martin Blumenstingl
> <martin.blumenstingl@googlemail.com> wrote:
>
> > Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
> > tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
> > if "cd-inverted" is also set. This results in the "cd-inverted" property
> > being evaluated twice, which effectively makes it a no-op:
> > - first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
> >   setting up the CD GPIO
> > - then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
> >   reading the CD GPIO value at runtime
> >
> > On boards which are using device-tree with the "cd-inverted" property
> > being set any inserted card are not detected anymore. This is due to the
> > MMC core treating the CD GPIO with the wrong polarity.
> >
> > Disable "override_cd_active_level" for the card detection GPIO which is
> > parsed using mmc_of_parse. This fixes SD card detection on the boards
> > which are currently using the "cd-inverted" device-tree property (tested
> > on Meson8b Odroid-C1 and Meson8b EC-100).
> >
> > This does not remove the CD GPIO inversion logic from the MMC core
> > because there's at least one driver (sdhci-pci-core for Intel BayTrail
> > based boards) which still passes "override_cd_active_level = true" to
> > mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
> > untouched.
> > In the future the GPIO inversion logic for both, card and read-only
> > detection can be removed once no driver is using it anymore.
> >
> > Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
> > Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
>
> OK two steps forward, one step back, that's a good fix for now, I
> can fix it properly for v4.22.
> Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
great, thanks for reviewing this!

feel free to keep me CC'ed on the "cleanup patches". I can review them
if you don't mind that I usually need a few days to do that


Regards
Martin
Anand Moon Jan. 9, 2019, 6:54 p.m. UTC | #3
Hi Martin,

On Wed, 2 Jan 2019 at 01:14, Martin Blumenstingl
<martin.blumenstingl@googlemail.com> wrote:
>
> Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
> tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
> if "cd-inverted" is also set. This results in the "cd-inverted" property
> being evaluated twice, which effectively makes it a no-op:
> - first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
>   setting up the CD GPIO
> - then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
>   reading the CD GPIO value at runtime
>
> On boards which are using device-tree with the "cd-inverted" property
> being set any inserted card are not detected anymore. This is due to the
> MMC core treating the CD GPIO with the wrong polarity.
>
> Disable "override_cd_active_level" for the card detection GPIO which is
> parsed using mmc_of_parse. This fixes SD card detection on the boards
> which are currently using the "cd-inverted" device-tree property (tested
> on Meson8b Odroid-C1 and Meson8b EC-100).
>
> This does not remove the CD GPIO inversion logic from the MMC core
> because there's at least one driver (sdhci-pci-core for Intel BayTrail
> based boards) which still passes "override_cd_active_level = true" to
> mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
> untouched.
> In the future the GPIO inversion logic for both, card and read-only
> detection can be removed once no driver is using it anymore.
>
> Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
> ---
>  drivers/mmc/core/host.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
> index f57f5de54206..cf58ccaf22d5 100644
> --- a/drivers/mmc/core/host.c
> +++ b/drivers/mmc/core/host.c
> @@ -234,7 +234,7 @@ int mmc_of_parse(struct mmc_host *host)
>                 if (device_property_read_bool(dev, "broken-cd"))
>                         host->caps |= MMC_CAP_NEEDS_POLL;
>
> -               ret = mmc_gpiod_request_cd(host, "cd", 0, true,
> +               ret = mmc_gpiod_request_cd(host, "cd", 0, false,
>                                            cd_debounce_delay_ms * 1000,
>                                            &cd_gpio_invert);
>                 if (!ret)
> --
> 2.20.1
>

Please add my.
Tested on Odroidc1+ and Odroidc2 both booted up fine.

Tested-by: Anand Moon <linux.amoon@gmail.com>

>
> _______________________________________________
> linux-amlogic mailing list
> linux-amlogic@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-amlogic
Loys Ollivier Jan. 10, 2019, 11:33 a.m. UTC | #4
Hi Martin,

On Tue 01 Jan 2019 at 19:44, Martin Blumenstingl wrote:

> Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
> tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
> if "cd-inverted" is also set. This results in the "cd-inverted" property
> being evaluated twice, which effectively makes it a no-op:
> - first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
>   setting up the CD GPIO
> - then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
>   reading the CD GPIO value at runtime
>
> On boards which are using device-tree with the "cd-inverted" property
> being set any inserted card are not detected anymore. This is due to the
> MMC core treating the CD GPIO with the wrong polarity.
>
> Disable "override_cd_active_level" for the card detection GPIO which is
> parsed using mmc_of_parse. This fixes SD card detection on the boards
> which are currently using the "cd-inverted" device-tree property (tested
> on Meson8b Odroid-C1 and Meson8b EC-100).
>
> This does not remove the CD GPIO inversion logic from the MMC core
> because there's at least one driver (sdhci-pci-core for Intel BayTrail
> based boards) which still passes "override_cd_active_level = true" to
> mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
> untouched.
> In the future the GPIO inversion logic for both, card and read-only
> detection can be removed once no driver is using it anymore.
>
> Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
> ---
>  drivers/mmc/core/host.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
> index f57f5de54206..cf58ccaf22d5 100644
> --- a/drivers/mmc/core/host.c
> +++ b/drivers/mmc/core/host.c
> @@ -234,7 +234,7 @@ int mmc_of_parse(struct mmc_host *host)
>  		if (device_property_read_bool(dev, "broken-cd"))
>  			host->caps |= MMC_CAP_NEEDS_POLL;
>  
> -		ret = mmc_gpiod_request_cd(host, "cd", 0, true,
> +		ret = mmc_gpiod_request_cd(host, "cd", 0, false,
>  					   cd_debounce_delay_ms * 1000,
>  					   &cd_gpio_invert);
>  		if (!ret)

Test on libretech-cc, sd is fixed with that patch.

Tested-by: Loys Ollivier <loys.ollivier@gmail.com>
Linus Walleij Jan. 11, 2019, 2:28 p.m. UTC | #5
On Tue, Jan 1, 2019 at 8:44 PM Martin Blumenstingl
<martin.blumenstingl@googlemail.com> wrote:

> Since commit 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device
> tree") gpiolib-of parses the "cd-gpios" property and flips the polarity
> if "cd-inverted" is also set. This results in the "cd-inverted" property
> being evaluated twice, which effectively makes it a no-op:
> - first in drivers/gpio/gpiolib-of.c (of_xlate_and_get_gpiod_flags) when
>   setting up the CD GPIO
> - then again in drivers/mmc/core/slot-gpio.c (mmc_gpio_get_cd) when
>   reading the CD GPIO value at runtime
>
> On boards which are using device-tree with the "cd-inverted" property
> being set any inserted card are not detected anymore. This is due to the
> MMC core treating the CD GPIO with the wrong polarity.
>
> Disable "override_cd_active_level" for the card detection GPIO which is
> parsed using mmc_of_parse. This fixes SD card detection on the boards
> which are currently using the "cd-inverted" device-tree property (tested
> on Meson8b Odroid-C1 and Meson8b EC-100).
>
> This does not remove the CD GPIO inversion logic from the MMC core
> because there's at least one driver (sdhci-pci-core for Intel BayTrail
> based boards) which still passes "override_cd_active_level = true" to
> mmc_gpiod_request_cd(). Due to lack of hardware for testing this is left
> untouched.
> In the future the GPIO inversion logic for both, card and read-only
> detection can be removed once no driver is using it anymore.
>
> Fixes: 89a5e15bcba87d ("gpio/mmc/of: Respect polarity in the device tree")
> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>

I applied this to the GPIO fixes with the ACKs.

Ulf is on vacation but I managed to get his ACK over chat,
so I will try to get this upstream ASAP.

Yours,
Linus Walleij

Patch
diff mbox series

diff --git a/drivers/mmc/core/host.c b/drivers/mmc/core/host.c
index f57f5de54206..cf58ccaf22d5 100644
--- a/drivers/mmc/core/host.c
+++ b/drivers/mmc/core/host.c
@@ -234,7 +234,7 @@  int mmc_of_parse(struct mmc_host *host)
 		if (device_property_read_bool(dev, "broken-cd"))
 			host->caps |= MMC_CAP_NEEDS_POLL;
 
-		ret = mmc_gpiod_request_cd(host, "cd", 0, true,
+		ret = mmc_gpiod_request_cd(host, "cd", 0, false,
 					   cd_debounce_delay_ms * 1000,
 					   &cd_gpio_invert);
 		if (!ret)