Skip to contentSkip to conversation

This series adds support for gpio-leds "default-state" property. The

PATCH Superseded linux-kernel-dev at beckhoff.com 3 participants 8 messages Apr 3, 2018 – Apr 4, 2018

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v2 cover This series adds support for gpio-leds "default-state" property. The

From: Patrick Bruenn <[email protected]>

main usecase in mind are LEDs which indicate a state like "power on". With this patchset applied, all you have to do is:
Add a gpio-led node with 'default-state = "on";' property to your device tree. And the LED will automatically light up during U-Boot startup.

Changes in v2:
- rebase to v2018.05-rc1
- add dm_test_led_default_state() to tests/dm/led.c

Patrick Bruenn (3):
dm: led: Support "default-state" property
dm: led: auto probe() LEDs with "default-state"
dm: led: add testcase for "default-state" property

 arch/sandbox/dts/test.dts | 12 ++++++++++++
 drivers/led/led_gpio.c    | 25 ++++++++++++++++++++++++-
 test/dm/led.c             | 16 ++++++++++++++++
 3 files changed, 52 insertions(+), 1 deletion(-)
Signature · 1 line
2.11.0
#1 lore raw

PATCH v2 1/3 dm: led: Support "default-state" property

From: Patrick Bruenn <[email protected]>

Add support for the device tree property "default-state". This feature might be useful for LEDs indicating "power on" or similar states.

Note: Even with this commit gpio-leds remain in reset state. That's because the led_gpio is not probed until DM_FLAG_ACTIVATED is set.

---

Changes in v2:
- rebase to v2018.05-rc1
- add dm_test_led_default_state() to tests/dm/led.c

 drivers/led/led_gpio.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
drivers/led/led_gpio.c +15 -1
index 9976635887..e68d8d3864 100644
@@ -60,11 +60,25 @@ static int led_gpio_probe(struct udevice *dev)
 {
 	struct led_uc_plat *uc_plat = dev_get_uclass_platdata(dev);
 	struct led_gpio_priv *priv = dev_get_priv(dev);
+	const char *default_state;
+	int ret;
 
 	/* Ignore the top-level LED node */
 	if (!uc_plat->label)
 		return 0;
-	return gpio_request_by_name(dev, "gpios", 0, &priv->gpio, GPIOD_IS_OUT);
+
+	ret = gpio_request_by_name(dev, "gpios", 0, &priv->gpio, GPIOD_IS_OUT);
+	if (ret)
+		return ret;
+
+	default_state = dev_read_string(dev, "default-state");
+	if (default_state) {
+		if (!strncmp(default_state, "on", 2))
+			gpio_led_set_state(dev, LEDST_ON);
+		else if (!strncmp(default_state, "off", 3))
+			gpio_led_set_state(dev, LEDST_OFF);
+	}
+	return 0;
 }
 
 static int led_gpio_remove(struct udevice *dev)
Signature · 1 line
2.11.0
#2 lore raw

PATCH v2 1/3 dm: led: Support "default-state" property

Quoted 10 lines · On 3 April 2018 at 15:31, <[email protected]> wrote:

From: Patrick Bruenn <[email protected]>

Add support for the device tree property "default-state". This feature might be useful for LEDs indicating "power on" or similar states.

Note: Even with this commit gpio-leds remain in reset state. That's because the led_gpio is not probed until DM_FLAG_ACTIVATED is set.

Quoted 8 lines

---

Changes in v2:
- rebase to v2018.05-rc1
- add dm_test_led_default_state() to tests/dm/led.c

 drivers/led/led_gpio.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)
#3 lore raw

PATCH v2 2/3 dm: led: auto probe() LEDs with "default-state"

In reply to linux-kernel-dev at beckhoff.com

From: Patrick Bruenn <[email protected]>

To avoid board specificy LED activation code, automatically activate gpio-leds with "default-state" property during bind().

---

Changes in v2: None

 drivers/led/led_gpio.c | 9 +++++++++
 1 file changed, 9 insertions(+)
drivers/led/led_gpio.c +9 -0
index e68d8d3864..d2fe3d5ad5 100644
@@ -11,6 +11,7 @@
 #include <led.h>
 #include <asm/gpio.h>
 #include <dm/lists.h>
+#include <dm/uclass-internal.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -120,6 +121,14 @@ static int led_gpio_bind(struct udevice *parent)
 			return ret;
 		uc_plat = dev_get_uclass_platdata(dev);
 		uc_plat->label = label;
+
+		if (ofnode_read_string(node, "default-state")) {
+			struct udevice *devp;
+
+			ret = uclass_get_device_tail(dev, 0, &devp);
+			if (ret)
+				return ret;
+		}
 	}
 
 	return 0;
Signature · 1 line
2.11.0
#4 lore raw

PATCH v2 2/3 dm: led: auto probe() LEDs with "default-state"

Quoted 5 lines · On 3 April 2018 at 15:31, <[email protected]> wrote:

From: Patrick Bruenn <[email protected]>

To avoid board specificy LED activation code, automatically activate gpio-leds with "default-state" property during bind().

Quoted 7 lines

---

Changes in v2: None

 drivers/led/led_gpio.c | 9 +++++++++
 1 file changed, 9 insertions(+)

But please see below.

Quoted 14 lines
drivers/led/led_gpio.c +1 -0
index e68d8d3864..d2fe3d5ad5 100644
@@ -11,6 +11,7 @@
 #include <led.h>
 #include <asm/gpio.h>
 #include <dm/lists.h>
+#include <dm/uclass-internal.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 

drivers/led/led_gpio.c static int led_gpio_bind(struct udevice *parent) · line 125

Quoted 5 lines
drivers/led/led_gpio.c +2 -0
@@ -120,6 +121,14 @@ static int led_gpio_bind(struct udevice *parent)
                        return ret;
                uc_plat = dev_get_uclass_platdata(dev);
                uc_plat->label = label;
+
+               if (ofnode_read_string(node, "default-state")) {

Here I think you mean ofnode_read_boot() ?

Quoted 10 lines
drivers/led/led_gpio.c +6 -1
+                       struct udevice *devp;
+
+                       ret = uclass_get_device_tail(dev, 0, &devp);
+                       if (ret)
+                               return ret;
+               }
        }
 
        return 0;
--
Quoted 3 lines

2.11.0

Regards,
Simon

#5 lore raw

PATCH v2 2/3 dm: led: auto probe() LEDs with "default-state"

From: [email protected] [mailto:sjg at google.com] On Behalf Of Simon Glass

Quoted 9 lines

Sent: Dienstag, 3. April 2018 19:53
To: linux-kernel-dev <[email protected]>
Cc: U-Boot Mailing List <[email protected]>; Patrick Brünn <[email protected]>; Ziping Chen <[email protected]>
Subject: Re: [PATCH v2 2/3] dm: led: auto probe() LEDs with "default-state"

Quoted 2 lines · On 3 April 2018 at 15:31, <[email protected]> wrote:

From: Patrick Bruenn <[email protected]>

...

Quoted 13 lines
Quoted 13 lines
drivers/led/led_gpio.c +1 -0
index e68d8d3864..d2fe3d5ad5 100644
@@ -11,6 +11,7 @@
 #include <led.h>
 #include <asm/gpio.h>
 #include <dm/lists.h>
+#include <dm/uclass-internal.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
Quoted 9 lines

drivers/led/led_gpio.c static int led_gpio_bind(struct udevice *parent) · line 125

Quoted 6 lines
drivers/led/led_gpio.c +2 -0
@@ -120,6 +121,14 @@ static int led_gpio_bind(struct udevice *parent)
                        return ret;
                uc_plat = dev_get_uclass_platdata(dev);
                uc_plat->label = label;
+
+               if (ofnode_read_string(node, "default-state")) {

Here I think you mean ofnode_read_boot() ?

Yes, you are right I will resend a v3 with ofnode_read_bool().

It took me quite a while to understand that "read_bool" means "has_property".

Thanks for your help!
Patrick
Beckhoff Automation GmbH & Co. KG | Managing Director: Dipl. Phys. Hans Beckhoff Registered office: Verl, Germany | Register court: Guetersloh HRA 7075

#6 lore raw

PATCH v2 3/3 dm: led: add testcase for "default-state" property

In reply to linux-kernel-dev at beckhoff.com

From: Patrick Bruenn <[email protected]>

Add two more gpio-leds to sandbox test device tree with default-state property set to "on"/"off".
Add dm_test_led_default_state() to check that these new LED's are set to LEDST_ON and LEDST_OFF.

---
patman complains about: test/dm/led.c:45: check: Please use a blank line after function/struct/union/enum declarations.
I compared with other DM_TEST() usage and decided to ignore this check. Should we fix the macro, patman or keep ignoring this?

Changes in v2: None

 arch/sandbox/dts/test.dts | 12 ++++++++++++
 test/dm/led.c             | 16 ++++++++++++++++
 2 files changed, 28 insertions(+)
arch/sandbox/dts/test.dts +12 -0
index 06d0e8ce85..07f9bad258 100644
@@ -254,6 +254,18 @@
 			gpios = <&gpio_a 2 0>;
 			label = "sandbox:green";
 		};
+
+		default_on {
+			gpios = <&gpio_a 3 0>;
+			label = "sandbox:default_on";
+			default-state = "on";
+		};
+
+		default_off {
+			gpios = <&gpio_a 4 0>;
+			label = "sandbox:default_off";
+			default-state = "off";
+		};
 	};
 
 	mbox: mbox {
test/dm/led.c +16 -0
index fde700be38..c560abc4fb 100644
@@ -28,6 +28,22 @@ static int dm_test_led_base(struct unit_test_state *uts)
 }
 DM_TEST(dm_test_led_base, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
 
+/* Test of the LED 'default-state' device tree property */
+static int dm_test_led_default_state(struct unit_test_state *uts)
+{
+	struct udevice *dev;
+
+	/* Check that we handle the default-state property correctly. */
+	ut_assertok(led_get_by_label("sandbox:default_on", &dev));
+	ut_asserteq(LEDST_ON, led_get_state(dev));
+
+	ut_assertok(led_get_by_label("sandbox:default_off", &dev));
+	ut_asserteq(LEDST_OFF, led_get_state(dev));
+
+	return 0;
+}
+DM_TEST(dm_test_led_default_state, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
 /* Test of the led uclass using the led_gpio driver */
 static int dm_test_led_gpio(struct unit_test_state *uts)
 {
Signature · 1 line
2.11.0
#7 lore raw

PATCH v2 3/3 dm: led: add testcase for "default-state" property

Quoted 14 lines · On 3 April 2018 at 15:31, <[email protected]> wrote:

From: Patrick Bruenn <[email protected]>

Add two more gpio-leds to sandbox test device tree with default-state property set to "on"/"off".
Add dm_test_led_default_state() to check that these new LED's are set to LEDST_ON and LEDST_OFF.

---
patman complains about: test/dm/led.c:45: check: Please use a blank line after function/struct/union/enum declarations.
I compared with other DM_TEST() usage and decided to ignore this check. Should we fix the macro, patman or keep ignoring this?

Quoted 5 lines

Changes in v2: None

 arch/sandbox/dts/test.dts | 12 ++++++++++++
 test/dm/led.c             | 16 ++++++++++++++++
 2 files changed, 28 insertions(+)