[U-Boot] [PATCH 1/2] mmc: sunxi: Support cd-inverted DT property
Commit 8620f384098b ("dm: sunxi: Linksprite_pcDuino3: Correct polarity of MMC card detect") claims that the Pcduino3 device tree has an incorrect polarity for the card detect pin, but the actual problem is that unlike the Linux MMC driver, the U-Boot driver isn't respecting the cd-invert property (see Documentation/devicetree/bindings/mmc/mmc.txt) which tells that the card detect signal level should be inverted.
Fix this properly by adding support for the cd-inverted property while reverting the original commit at the same time. While at it, I noticed the driver always enables pullups for the card detect line, which is not right if the card detect GPIO is active-high, so fix that as well.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com --- arch/arm/dts/sun7i-a20-pcduino3.dts | 2 +- drivers/mmc/sunxi_mmc.c | 12 ++++++++++-- 2 files changed, 11 insertions(+), 3 deletions(-)
diff --git a/arch/arm/dts/sun7i-a20-pcduino3.dts b/arch/arm/dts/sun7i-a20-pcduino3.dts index 37b1e0ee9b..1a8b39be1d 100644 --- a/arch/arm/dts/sun7i-a20-pcduino3.dts +++ b/arch/arm/dts/sun7i-a20-pcduino3.dts @@ -164,7 +164,7 @@ pinctrl-0 = <&mmc0_pins_a>, <&mmc0_cd_pin_reference_design>; vmmc-supply = <®_vcc3v3>; bus-width = <4>; - cd-gpios = <&pio 7 1 GPIO_ACTIVE_LOW>; /* PH1 */ + cd-gpios = <&pio 7 1 GPIO_ACTIVE_HIGH>; /* PH1 */ cd-inverted; status = "okay"; }; diff --git a/drivers/mmc/sunxi_mmc.c b/drivers/mmc/sunxi_mmc.c index 4edb4be46c..7cc7303570 100644 --- a/drivers/mmc/sunxi_mmc.c +++ b/drivers/mmc/sunxi_mmc.c @@ -30,6 +30,7 @@ struct sunxi_mmc_priv { uint32_t *mclkreg; unsigned fatal_err; struct gpio_desc cd_gpio; /* Change Detect GPIO */ + int cd_inverted; struct sunxi_mmc *reg; struct mmc_config cfg; }; @@ -545,7 +546,7 @@ static int sunxi_mmc_getcd(struct udevice *dev) struct sunxi_mmc_priv *priv = dev_get_priv(dev);
if (dm_gpio_is_valid(&priv->cd_gpio)) - return dm_gpio_get_value(&priv->cd_gpio); + return dm_gpio_get_value(&priv->cd_gpio) ^ priv->cd_inverted;
return 1; } @@ -606,8 +607,15 @@ static int sunxi_mmc_probe(struct udevice *dev) if (!gpio_request_by_name(dev, "cd-gpios", 0, &priv->cd_gpio, GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio); + int cd_state = priv->cd_gpio.flags & GPIOD_ACTIVE_LOW ? 0 : 1;
- sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP); + priv->cd_inverted = dev_read_bool(dev, "cd-inverted"); + cd_state ^= priv->cd_inverted; + + if (cd_state) + sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_DOWN); + else + sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP); }
upriv->mmc = &plat->mmc;
After the previous commit which adds support for the cd-inverted property to the sunxi MMC driver, driver-model MMC works fine on this board.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com --- configs/Linksprite_pcDuino3_Nano_defconfig | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/configs/Linksprite_pcDuino3_Nano_defconfig b/configs/Linksprite_pcDuino3_Nano_defconfig index 13538fafd1..44ce8b3ead 100644 --- a/configs/Linksprite_pcDuino3_Nano_defconfig +++ b/configs/Linksprite_pcDuino3_Nano_defconfig @@ -17,9 +17,11 @@ CONFIG_SPL_I2C_SUPPORT=y # CONFIG_SPL_ISO_PARTITION is not set # CONFIG_SPL_EFI_PARTITION is not set CONFIG_SCSI_AHCI=y +CONFIG_DM_MMC=y CONFIG_ETH_DESIGNWARE=y CONFIG_RGMII=y CONFIG_SUN7I_GMAC=y CONFIG_SCSI=y +CONFIG_DM_SCSI=y CONFIG_USB_EHCI_HCD=y CONFIG_SYS_USB_EVENT_POLL_VIA_INT_QUEUE=y
On Mon, Dec 18, 2017 at 10:09:42PM +0200, Tuomas Tynkkynen wrote:
After the previous commit which adds support for the cd-inverted property to the sunxi MMC driver, driver-model MMC works fine on this board.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com
configs/Linksprite_pcDuino3_Nano_defconfig | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/configs/Linksprite_pcDuino3_Nano_defconfig b/configs/Linksprite_pcDuino3_Nano_defconfig index 13538fafd1..44ce8b3ead 100644 --- a/configs/Linksprite_pcDuino3_Nano_defconfig +++ b/configs/Linksprite_pcDuino3_Nano_defconfig @@ -17,9 +17,11 @@ CONFIG_SPL_I2C_SUPPORT=y # CONFIG_SPL_ISO_PARTITION is not set # CONFIG_SPL_EFI_PARTITION is not set CONFIG_SCSI_AHCI=y +CONFIG_DM_MMC=y CONFIG_ETH_DESIGNWARE=y CONFIG_RGMII=y CONFIG_SUN7I_GMAC=y CONFIG_SCSI=y +CONFIG_DM_SCSI=y
Those should not be in a single defconfig, but enabled for all the Allwinner boards (that have SATA and MMC).
Maxime
On Wed, 2017-12-20 at 13:26 +0100, Maxime Ripard wrote:
On Mon, Dec 18, 2017 at 10:09:42PM +0200, Tuomas Tynkkynen wrote:
After the previous commit which adds support for the cd-inverted property to the sunxi MMC driver, driver-model MMC works fine on this board.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com
configs/Linksprite_pcDuino3_Nano_defconfig | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/configs/Linksprite_pcDuino3_Nano_defconfig b/configs/Linksprite_pcDuino3_Nano_defconfig index 13538fafd1..44ce8b3ead 100644 --- a/configs/Linksprite_pcDuino3_Nano_defconfig +++ b/configs/Linksprite_pcDuino3_Nano_defconfig @@ -17,9 +17,11 @@ CONFIG_SPL_I2C_SUPPORT=y # CONFIG_SPL_ISO_PARTITION is not set # CONFIG_SPL_EFI_PARTITION is not set CONFIG_SCSI_AHCI=y +CONFIG_DM_MMC=y CONFIG_ETH_DESIGNWARE=y CONFIG_RGMII=y CONFIG_SUN7I_GMAC=y CONFIG_SCSI=y +CONFIG_DM_SCSI=y
Those should not be in a single defconfig, but enabled for all the Allwinner boards (that have SATA and MMC).
Maxime
Makes sense. I don't have any other boards to test with so I just imitated Simon's change for the non-Nano pcduino3.
I will try to see if this could be imply'd in Kconfig.
On Wed, Dec 20, 2017 at 03:37:13PM +0200, Tuomas Tynkkynen wrote:
On Wed, 2017-12-20 at 13:26 +0100, Maxime Ripard wrote:
On Mon, Dec 18, 2017 at 10:09:42PM +0200, Tuomas Tynkkynen wrote:
After the previous commit which adds support for the cd-inverted property to the sunxi MMC driver, driver-model MMC works fine on this board.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com
configs/Linksprite_pcDuino3_Nano_defconfig | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/configs/Linksprite_pcDuino3_Nano_defconfig b/configs/Linksprite_pcDuino3_Nano_defconfig index 13538fafd1..44ce8b3ead 100644 --- a/configs/Linksprite_pcDuino3_Nano_defconfig +++ b/configs/Linksprite_pcDuino3_Nano_defconfig @@ -17,9 +17,11 @@ CONFIG_SPL_I2C_SUPPORT=y # CONFIG_SPL_ISO_PARTITION is not set # CONFIG_SPL_EFI_PARTITION is not set CONFIG_SCSI_AHCI=y +CONFIG_DM_MMC=y CONFIG_ETH_DESIGNWARE=y CONFIG_RGMII=y CONFIG_SUN7I_GMAC=y CONFIG_SCSI=y +CONFIG_DM_SCSI=y
Those should not be in a single defconfig, but enabled for all the Allwinner boards (that have SATA and MMC).
Maxime
Makes sense. I don't have any other boards to test with so I just imitated Simon's change for the non-Nano pcduino3.
I will try to see if this could be imply'd in Kconfig.
You can also use a default y if ARCH_SUNXI. That's actually a better option, since imply still have a few glitches that generate an error when the situation is valid.
Thanks! Maxime
On Mon, Dec 18, 2017 at 10:09:41PM +0200, Tuomas Tynkkynen wrote:
Commit 8620f384098b ("dm: sunxi: Linksprite_pcDuino3: Correct polarity of MMC card detect") claims that the Pcduino3 device tree has an incorrect polarity for the card detect pin, but the actual problem is that unlike the Linux MMC driver, the U-Boot driver isn't respecting the cd-invert property (see Documentation/devicetree/bindings/mmc/mmc.txt) which tells that the card detect signal level should be inverted.
Fix this properly by adding support for the cd-inverted property while reverting the original commit at the same time. While at it, I noticed the driver always enables pullups for the card detect line, which is not right if the card detect GPIO is active-high, so fix that as well.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com
arch/arm/dts/sun7i-a20-pcduino3.dts | 2 +- drivers/mmc/sunxi_mmc.c | 12 ++++++++++-- 2 files changed, 11 insertions(+), 3 deletions(-)
diff --git a/arch/arm/dts/sun7i-a20-pcduino3.dts b/arch/arm/dts/sun7i-a20-pcduino3.dts index 37b1e0ee9b..1a8b39be1d 100644 --- a/arch/arm/dts/sun7i-a20-pcduino3.dts +++ b/arch/arm/dts/sun7i-a20-pcduino3.dts @@ -164,7 +164,7 @@ pinctrl-0 = <&mmc0_pins_a>, <&mmc0_cd_pin_reference_design>; vmmc-supply = <®_vcc3v3>; bus-width = <4>;
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_LOW>; /* PH1 */
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_HIGH>; /* PH1 */ cd-inverted; status = "okay";
}; diff --git a/drivers/mmc/sunxi_mmc.c b/drivers/mmc/sunxi_mmc.c index 4edb4be46c..7cc7303570 100644 --- a/drivers/mmc/sunxi_mmc.c +++ b/drivers/mmc/sunxi_mmc.c @@ -30,6 +30,7 @@ struct sunxi_mmc_priv { uint32_t *mclkreg; unsigned fatal_err; struct gpio_desc cd_gpio; /* Change Detect GPIO */
- int cd_inverted; struct sunxi_mmc *reg; struct mmc_config cfg;
}; @@ -545,7 +546,7 @@ static int sunxi_mmc_getcd(struct udevice *dev) struct sunxi_mmc_priv *priv = dev_get_priv(dev);
if (dm_gpio_is_valid(&priv->cd_gpio))
return dm_gpio_get_value(&priv->cd_gpio);
return dm_gpio_get_value(&priv->cd_gpio) ^ priv->cd_inverted;return 1;
} @@ -606,8 +607,15 @@ static int sunxi_mmc_probe(struct udevice *dev) if (!gpio_request_by_name(dev, "cd-gpios", 0, &priv->cd_gpio, GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio);
int cd_state = priv->cd_gpio.flags & GPIOD_ACTIVE_LOW ? 0 : 1;
Isn't that change itself enough to get what you wanted?
You really shouldn't be doing anything else.
Maxime
On Wed, 2017-12-20 at 13:26 +0100, Maxime Ripard wrote:
On Mon, Dec 18, 2017 at 10:09:41PM +0200, Tuomas Tynkkynen wrote:
Commit 8620f384098b ("dm: sunxi: Linksprite_pcDuino3: Correct polarity of MMC card detect") claims that the Pcduino3 device tree has an incorrect polarity for the card detect pin, but the actual problem is that unlike the Linux MMC driver, the U-Boot driver isn't respecting the cd-invert property (see Documentation/devicetree/bindings/mmc/mmc.txt) which tells that the card detect signal level should be inverted.
Fix this properly by adding support for the cd-inverted property while reverting the original commit at the same time. While at it, I noticed the driver always enables pullups for the card detect line, which is not right if the card detect GPIO is active-high, so fix that as well.
Signed-off-by: Tuomas Tynkkynen tuomas@tuxera.com
arch/arm/dts/sun7i-a20-pcduino3.dts | 2 +- drivers/mmc/sunxi_mmc.c | 12 ++++++++++-- 2 files changed, 11 insertions(+), 3 deletions(-)
diff --git a/arch/arm/dts/sun7i-a20-pcduino3.dts b/arch/arm/dts/sun7i-a20-pcduino3.dts index 37b1e0ee9b..1a8b39be1d 100644 --- a/arch/arm/dts/sun7i-a20-pcduino3.dts +++ b/arch/arm/dts/sun7i-a20-pcduino3.dts @@ -164,7 +164,7 @@ pinctrl-0 = <&mmc0_pins_a>, <&mmc0_cd_pin_reference_design>; vmmc-supply = <®_vcc3v3>; bus-width = <4>;
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_LOW>; /* PH1 */
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_HIGH>; /* PH1 */ cd-inverted; status = "okay";
}; diff --git a/drivers/mmc/sunxi_mmc.c b/drivers/mmc/sunxi_mmc.c index 4edb4be46c..7cc7303570 100644 --- a/drivers/mmc/sunxi_mmc.c +++ b/drivers/mmc/sunxi_mmc.c @@ -30,6 +30,7 @@ struct sunxi_mmc_priv { uint32_t *mclkreg; unsigned fatal_err; struct gpio_desc cd_gpio; /* Change Detect GPIO */
- int cd_inverted; struct sunxi_mmc *reg; struct mmc_config cfg;
}; @@ -545,7 +546,7 @@ static int sunxi_mmc_getcd(struct udevice *dev) struct sunxi_mmc_priv *priv = dev_get_priv(dev);
if (dm_gpio_is_valid(&priv->cd_gpio))
return dm_gpio_get_value(&priv->cd_gpio);
return dm_gpio_get_value(&priv->cd_gpio) ^ priv-cd_inverted;
return 1; } @@ -606,8 +607,15 @@ static int sunxi_mmc_probe(struct udevice *dev) if (!gpio_request_by_name(dev, "cd-gpios", 0, &priv-
cd_gpio,
GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio);
int cd_state = priv->cd_gpio.flags &GPIOD_ACTIVE_LOW ? 0 : 1;
Isn't that change itself enough to get what you wanted?
You really shouldn't be doing anything else.
Was that the correct part you meant to quote? The cd_state was only for the somewhat pull-up change... Sorry about that, here's the minimal (probably whitespace-damaged) change:
diff --git a/arch/arm/dts/sun7i-a20-pcduino3.dts b/arch/arm/dts/sun7i-a20-pcduino3.dts index 37b1e0ee9b..1a8b39be1d 100644 --- a/arch/arm/dts/sun7i-a20-pcduino3.dts +++ b/arch/arm/dts/sun7i-a20-pcduino3.dts @@ -164,7 +164,7 @@ pinctrl-0 = <&mmc0_pins_a>, <&mmc0_cd_pin_reference_design>; vmmc-supply = <®_vcc3v3>; bus-width = <4>; - cd-gpios = <&pio 7 1 GPIO_ACTIVE_LOW>; /* PH1 */ + cd-gpios = <&pio 7 1 GPIO_ACTIVE_HIGH>; /* PH1 */ cd-inverted; status = "okay"; }; diff --git a/drivers/mmc/sunxi_mmc.c b/drivers/mmc/sunxi_mmc.c index 4edb4be46c..a974543148 100644 --- a/drivers/mmc/sunxi_mmc.c +++ b/drivers/mmc/sunxi_mmc.c @@ -30,6 +30,7 @@ struct sunxi_mmc_priv { uint32_t *mclkreg; unsigned fatal_err; struct gpio_desc cd_gpio; /* Change Detect GPIO */ + int cd_inverted; struct sunxi_mmc *reg; struct mmc_config cfg; }; @@ -545,7 +546,7 @@ static int sunxi_mmc_getcd(struct udevice *dev) struct sunxi_mmc_priv *priv = dev_get_priv(dev);
if (dm_gpio_is_valid(&priv->cd_gpio)) - return dm_gpio_get_value(&priv->cd_gpio); + return dm_gpio_get_value(&priv->cd_gpio) ^ priv->cd_inverted;
return 1; } @@ -607,6 +608,8 @@ static int sunxi_mmc_probe(struct udevice *dev) GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio);
+ priv->cd_inverted = dev_read_bool(dev, "cd-inverted"); + sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP); }
So: if the DT specifies GPIO_ACTIVE_LOW, dm_gpio_get_value() does its own inversion of the GPIO level. Then, on top of that we do another inversion if the "cd-inverted" property was specified. This matches what the Linux MMC driver does.
Does that make sense?
Thanks.
Hi Tuomas,
On Wed, Dec 20, 2017 at 03:34:47PM +0200, Tuomas Tynkkynen wrote:
@@ -606,8 +607,15 @@ static int sunxi_mmc_probe(struct udevice *dev) if (!gpio_request_by_name(dev, "cd-gpios", 0, &priv-
cd_gpio,
GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio);
int cd_state = priv->cd_gpio.flags &GPIOD_ACTIVE_LOW ? 0 : 1;
Isn't that change itself enough to get what you wanted?
You really shouldn't be doing anything else.
Was that the correct part you meant to quote? The cd_state was only for the somewhat pull-up change... Sorry about that, here's the minimal (probably whitespace-damaged) change:
Sorry if it wasn't really clear, I actually wanted to say that cd-inverted is deprecated, and you should just teach about the gpio flags like you did.
The rest seems superfluous, especially since I'd expect the combination of our GPIO flags and cd-inverted in our DTs to be buggy in most cases.
diff --git a/arch/arm/dts/sun7i-a20-pcduino3.dts b/arch/arm/dts/sun7i-a20-pcduino3.dts index 37b1e0ee9b..1a8b39be1d 100644 --- a/arch/arm/dts/sun7i-a20-pcduino3.dts +++ b/arch/arm/dts/sun7i-a20-pcduino3.dts @@ -164,7 +164,7 @@ pinctrl-0 = <&mmc0_pins_a>, <&mmc0_cd_pin_reference_design>; vmmc-supply = <®_vcc3v3>; bus-width = <4>;
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_LOW>; /* PH1 */
- cd-gpios = <&pio 7 1 GPIO_ACTIVE_HIGH>; /* PH1 */ cd-inverted; status = "okay";
}; diff --git a/drivers/mmc/sunxi_mmc.c b/drivers/mmc/sunxi_mmc.c index 4edb4be46c..a974543148 100644 --- a/drivers/mmc/sunxi_mmc.c +++ b/drivers/mmc/sunxi_mmc.c @@ -30,6 +30,7 @@ struct sunxi_mmc_priv { uint32_t *mclkreg; unsigned fatal_err; struct gpio_desc cd_gpio; /* Change Detect GPIO */
- int cd_inverted; struct sunxi_mmc *reg; struct mmc_config cfg;
}; @@ -545,7 +546,7 @@ static int sunxi_mmc_getcd(struct udevice *dev) struct sunxi_mmc_priv *priv = dev_get_priv(dev);
if (dm_gpio_is_valid(&priv->cd_gpio))
return dm_gpio_get_value(&priv->cd_gpio);
return dm_gpio_get_value(&priv->cd_gpio) ^ priv->cd_inverted;return 1;
} @@ -607,6 +608,8 @@ static int sunxi_mmc_probe(struct udevice *dev) GPIOD_IS_IN)) { int cd_pin = gpio_get_number(&priv->cd_gpio);
priv->cd_inverted = dev_read_bool(dev, "cd-inverted");- sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP); }
So: if the DT specifies GPIO_ACTIVE_LOW, dm_gpio_get_value() does its own inversion of the GPIO level. Then, on top of that we do another inversion if the "cd-inverted" property was specified. This matches what the Linux MMC driver does.
Hmmm, right.
I guess in your particular case then, you'd be better off droping the cd-inverted property. Like I said, it's really legacy these days. That of course doesn't prevent the rest of the patch to go in.
Maxime
On Thu, 2017-12-21 at 14:09 +0100, Maxime Ripard wrote:
sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP);}
So: if the DT specifies GPIO_ACTIVE_LOW, dm_gpio_get_value() does its own inversion of the GPIO level. Then, on top of that we do another inversion if the "cd-inverted" property was specified. This matches what the Linux MMC driver does.
Hmmm, right.
I guess in your particular case then, you'd be better off droping the cd-inverted property. Like I said, it's really legacy these days. That of course doesn't prevent the rest of the patch to go in.
Yes, getting rid of the cd-inverted properties from the DTs should make card detection work without any code changes. So the plan of action will be:
- Drop cd-inverted from U-Boot's sun7i-a20-pcduino3.dts - Flip the GPIO_ACTIVE_* around + drop cd-inverted from Linux DTs - Re-sync DTs from Linux to U-Boot - Finally switch all sunxi boards to DM_MMC in U-Boot
On Thu, Dec 21, 2017 at 04:09:46PM +0200, Tuomas Tynkkynen wrote:
On Thu, 2017-12-21 at 14:09 +0100, Maxime Ripard wrote:
sunxi_gpio_set_pull(cd_pin, SUNXI_GPIO_PULL_UP);}
So: if the DT specifies GPIO_ACTIVE_LOW, dm_gpio_get_value() does its own inversion of the GPIO level. Then, on top of that we do another inversion if the "cd-inverted" property was specified. This matches what the Linux MMC driver does.
Hmmm, right.
I guess in your particular case then, you'd be better off droping the cd-inverted property. Like I said, it's really legacy these days. That of course doesn't prevent the rest of the patch to go in.
Yes, getting rid of the cd-inverted properties from the DTs should make card detection work without any code changes. So the plan of action will be:
- Drop cd-inverted from U-Boot's sun7i-a20-pcduino3.dts
- Flip the GPIO_ACTIVE_* around + drop cd-inverted from Linux DTs
- Re-sync DTs from Linux to U-Boot
- Finally switch all sunxi boards to DM_MMC in U-Boot
Looks good to me :)
Maxime
participants (2)
-
Maxime Ripard -
Tuomas Tynkkynen