Skip to contentSkip to conversation

pwm: pxa: add device tree support to pwm driver

PATCH Superseded Mike Dunn 4 participants 9 messages Sep 8, 2013 – Sep 10, 2013

Subsystems (2)

v2 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Mike Dunn

author
#0 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

This patch adds device tree support to the pxa's pwm driver. Only an OF match table is added; nothing needs to be extracted from the device tree node. The existing platform_device id table is reused for the match table data. Support for inverted polarity is also added.

Tested on a Palm Treo 680 (both platform data and DT cases).

---
Changle log:
v2:
- of_match_table contains only the "pxa250-pwm" compatible string; require one device instance per pwm
- add Documentation/devicetree/bindings/pwm/pxa-pwm.txt
- add support for polarity flag in DT and implement set_polarity() method (the treo 680 inverts the signal between pwm out and backlight)
- return -EINVAL instead of -ENODEV if platform data or DT node not found
- output dev_info string if platform data missing
- expanded CC list of patch

 Documentation/devicetree/bindings/pwm/pxa-pwm.txt | 33 +++++++++++++
 arch/arm/boot/dts/pxa27x.dtsi                     | 24 ++++++++++
 drivers/pwm/pwm-pxa.c                             | 57 +++++++++++++++++++++++
 3 files changed, 114 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/pwm/pxa-pwm.txt
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +33 -0
new file mode 100644
index 0000000..7b09bfa
@@ -0,0 +1,33 @@
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible = "marvell,pxa250-pwm";
+- reg: physical base address and length of the registers used by the pwm channel
+  NB: One device instance must be created for each pwm that is used, so the
+  length covers only the register window for one pwm output, not that of the
+  entire pwm controller.  Currently length is 0x10 for all supported devices.
+- #pwm-cells: should be 3.
+   cell 1: the per-chip index of the PWM to use,
+   cell 2: the period in nanoseconds,
+   cell 3: flags; set bit 0 to specify inverse polarity.  The pwm controller
+   does not implement reverse polarity, but some boards may pass the pwm output
+   through an external inverter, which can cause problems if a client device
+   assumes that an increase in the duty cycle results in an increase in output
+   power.  The pwm-backlight is an example of such a client.
+
+Example pwm device node:
+
+pwm0: pwm at 40b00000 {
+	compatible = "marvell,pxa250-pwm";
+	reg = <0x40b00000 0x10>;
+	#pwm-cells = <3>;
+};
+
+Example pwm client node:
+
+backlight {
+      	compatible = "pwm-backlight";
+     	pwms = <&pwm0 0 5000000 1>;  /* pwm output is inverted */
+	...
+}
arch/arm/boot/dts/pxa27x.dtsi +24 -0
index d7c5d72..a12fe8e 100644
@@ -10,5 +10,29 @@
 			marvell,intc-priority;
 			marvell,intc-nr-irqs = <34>;
 		};
+
+		pwm0: pwm at 40b00000 {
+			compatible = "marvell,pxa250-pwm";
+			reg = <0x40b00000 0x10>;
+			#pwm-cells = <3>;
+		};
+
+		pwm1: pwm at 40b00010 {
+			compatible = "marvell,pxa250-pwm";
+			reg = <0x40b00010 0x10>;
+			#pwm-cells = <3>;
+		};
+
+		pwm2: pwm at 40c00000 {
+			compatible = "marvell,pxa250-pwm";
+			reg = <0x40c00000 0x10>;
+			#pwm-cells = <3>;
+		};
+
+		pwm3: pwm at 40c00010 {
+			compatible = "marvell,pxa250-pwm";
+			reg = <0x40c00010 0x10>;
+			#pwm-cells = <3>;
+		};
 	};
 };
drivers/pwm/pwm-pxa.c +57 -0

106 lines not on this page · raw message

Signature · 1 line
1.8.1.5
#1 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Dear Mike Dunn,

Quoted 43 lines

This patch adds device tree support to the pxa's pwm driver. Only an OF match table is added; nothing needs to be extracted from the device tree node. The existing platform_device id table is reused for the match table data. Support for inverted polarity is also added.

Tested on a Palm Treo 680 (both platform data and DT cases).

---
Changle log:
v2:
- of_match_table contains only the "pxa250-pwm" compatible string; require one device instance per pwm
- add Documentation/devicetree/bindings/pwm/pxa-pwm.txt
- add support for polarity flag in DT and implement set_polarity() method (the treo 680 inverts the signal between pwm out and backlight)
- return -EINVAL instead of -ENODEV if platform data or DT node not found
- output dev_info string if platform data missing
- expanded CC list of patch

 Documentation/devicetree/bindings/pwm/pxa-pwm.txt | 33 +++++++++++++
 arch/arm/boot/dts/pxa27x.dtsi                     | 24 ++++++++++
 drivers/pwm/pwm-pxa.c                             | 57
+++++++++++++++++++++++ 3 files changed, 114 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/pwm/pxa-pwm.txt
(quoted diff) +0 -0

b/Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file mode 100644 index 0000000..7b09bfa

Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +5 -0
@@ -0,0 +1,33 @@
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible =

"marvell,pxa250-pwm"; +- reg: physical base address and length of the registers used by the pwm channel + NB: One device instance must be created for each pwm that is used, so the + length covers only the register window for one pwm output, not that of the + entire pwm controller. Currently length is 0x10 for all supported devices. +- #pwm-cells: should be 3.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 12

Quoted 2 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
+   cell 1: the per-chip index of the PWM to use,
+   cell 2: the period in nanoseconds,

Is there no generic OF prop for this?

Quoted 1 line

+ cell 3: flags; set bit 0 to specify inverse polarity.

Do we not have some generic flag to indicate inverted PWM polarity?

[...]

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +3 -0
+#ifdef CONFIG_OF
+/*
+ * Device tree users should create one device instance for each pwm

channel. + * Hence we dispense with the HAS_SECONDARY_PWM and "tell" the original driver + * code that this is a single channel pxa25x-pwm.
+ */

Above ... pxa250-pwm , no ?

Documentation/devicetree/bindings/pwm/pxa-pwm.txt

Quoted 2 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},

Surely, data can be NULL, no ?

[...]

Documentation/devicetree/bindings/pwm/pxa-pwm.txt static int pwm_probe(struct platform_device *pdev) · line 203

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
@@ -145,6 +199,8 @@ static int pwm_probe(struct platform_device *pdev)
 	pwm->chip.ops = &pxa_pwm_ops;
 	pwm->chip.base = -1;
 	pwm->chip.npwm = (id->driver_data & HAS_SECONDARY_PWM) ? 2 : 1;
+	pwm->chip.of_xlate = of_pwm_xlate_with_flags;
+	pwm->chip.of_pwm_n_cells = 3;

Are these two settings needed ?

[...]
Best regards,
Marek Vasut

author
#2 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Quoted 41 lines · Marek Vasut

Dear Mike Dunn,

Quoted 39 lines

This patch adds device tree support to the pxa's pwm driver. Only an OF match table is added; nothing needs to be extracted from the device tree node. The existing platform_device id table is reused for the match table data. Support for inverted polarity is also added.

Tested on a Palm Treo 680 (both platform data and DT cases).

---
Changle log:
v2:
- of_match_table contains only the "pxa250-pwm" compatible string; require one device instance per pwm
- add Documentation/devicetree/bindings/pwm/pxa-pwm.txt
- add support for polarity flag in DT and implement set_polarity() method (the treo 680 inverts the signal between pwm out and backlight)
- return -EINVAL instead of -ENODEV if platform data or DT node not found
- output dev_info string if platform data missing
- expanded CC list of patch

 Documentation/devicetree/bindings/pwm/pxa-pwm.txt | 33 +++++++++++++
 arch/arm/boot/dts/pxa27x.dtsi                     | 24 ++++++++++
 drivers/pwm/pwm-pxa.c                             | 57
+++++++++++++++++++++++ 3 files changed, 114 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/pwm/pxa-pwm.txt
(quoted diff) +0 -0

b/Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file mode 100644 index 0000000..7b09bfa

Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +5 -0
@@ -0,0 +1,33 @@
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible =

"marvell,pxa250-pwm"; +- reg: physical base address and length of the registers used by the pwm channel + NB: One device instance must be

Quoted 8 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 12

Quoted 6 lines

created for each pwm that is used, so the + length covers only the register window for one pwm output, not that of the + entire pwm controller. Currently length is 0x10 for all supported devices. +- #pwm-cells: should be 3.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
+   cell 1: the per-chip index of the PWM to use,
+   cell 2: the period in nanoseconds,

Is there no generic OF prop for this?

Yes, all the pwm drivers with DT support have the same definitions for cell 1 and 2. Many have #pwm-cells=2, but for those with #pwm-cells=3, all define cell 3 the same way also.

Quoted 4 lines
Quoted 2 lines

+ cell 3: flags; set bit 0 to specify inverse polarity.

Do we not have some generic flag to indicate inverted PWM polarity?

Yes, enum pwm_polarity, defined in include/linux/pwm.h. The pwm drivers with polarity support all define cell 3 the same way. The wording in the bindings documentation is a little different in each case, but all essentially say the same thing. In Documentation/devicetree/bindings/pwm/ see for example atmel-tcb-pwm.txt and pwm-tiecap.txt.

I'm new to device tree, so maybe I am not understanding correctly this and your previous question... All DT properties are described as integers lacking any type (aside from their bit width), no?

Quoted 12 lines

[...]

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +3 -0
+#ifdef CONFIG_OF
+/*
+ * Device tree users should create one device instance for each pwm

channel. + * Hence we dispense with the HAS_SECONDARY_PWM and "tell" the original driver + * code that this is a single channel pxa25x-pwm.
+ */

Above ... pxa250-pwm , no ?

Quoted 4 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt

Quoted 2 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},

Surely, data can be NULL, no ?

It could, in which case pxa_pwm_get_id_dt() would explicitly return &pwm_id_table[0] instead of the .data element of the of_device_id. Not sure which way is better and why. That dumb platform_device_id table is causing all kinds of nuisance :)

Quoted 11 lines

[...]

Documentation/devicetree/bindings/pwm/pxa-pwm.txt static int pwm_probe(struct platform_device *pdev) · line 203

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +2 -0
@@ -145,6 +199,8 @@ static int pwm_probe(struct platform_device *pdev)
 	pwm->chip.ops = &pxa_pwm_ops;
 	pwm->chip.base = -1;
 	pwm->chip.npwm = (id->driver_data & HAS_SECONDARY_PWM) ? 2 : 1;
+	pwm->chip.of_xlate = of_pwm_xlate_with_flags;
+	pwm->chip.of_pwm_n_cells = 3;

Are these two settings needed ?

Yes. See drivers/pwm/core.c:of_pwmchip_add(), where they are set to default values of of_pwm_simple_xlate and 2 if left uninitialized.

Thanks again Marek,
Mike

#3 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Dear Mike Dunn,

[...]

Quoted 9 lines
Quoted 4 lines

drivers/pwm/pwm-pxa.c static void pxa_pwm_disable(struct pwm_chip *chip, struct pwm_device *pwm) · line 154

Quoted 2 lines
drivers/pwm/pwm-pxa.c +2 -0
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},

Surely, data can be NULL, no ?

It could, in which case pxa_pwm_get_id_dt() would explicitly return &pwm_id_table[0] instead of the .data element of the of_device_id. Not sure which way is better and why. That dumb platform_device_id table is causing all kinds of nuisance :)

Is the pwm_id_table needed at all anymore?

Quoted 8 lines
Quoted 8 lines

[...]

Quoted 6 lines
drivers/pwm/pwm-pxa.c +0 -0
@@ -145,6 +199,8 @@ static int pwm_probe(struct platform_device *pdev)
 
 	pwm->chip.ops = &pxa_pwm_ops;
 	pwm->chip.base = -1;
 	pwm->chip.npwm = (id->driver_data & HAS_SECONDARY_PWM) ? 2 : 1;
 
Quoted 7 lines
Quoted 4 lines

drivers/pwm/pwm-pxa.c static int pwm_probe(struct platform_device *pdev) · line 203

Quoted 2 lines
drivers/pwm/pwm-pxa.c +2 -0
+	pwm->chip.of_xlate = of_pwm_xlate_with_flags;
+	pwm->chip.of_pwm_n_cells = 3;

Are these two settings needed ?

Yes. See drivers/pwm/core.c:of_pwmchip_add(), where they are set to default values of of_pwm_simple_xlate and 2 if left uninitialized.

OK

Best regards,
Marek Vasut

author
#4 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Quoted 9 lines · Marek Vasut

Dear Mike Dunn,

[...]

Quoted 5 lines
Quoted 4 lines

drivers/pwm/pwm-pxa.c static void pxa_pwm_disable(struct pwm_chip *chip, struct pwm_device *pwm) · line 154

Quoted 2 lines
drivers/pwm/pwm-pxa.c +2 -0
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},

Surely, data can be NULL, no ?

Quoted 6 lines
Quoted 4 lines

It could, in which case pxa_pwm_get_id_dt() would explicitly return &pwm_id_table[0] instead of the .data element of the of_device_id. Not sure which way is better and why. That dumb platform_device_id table is causing all kinds of nuisance :)

Is the pwm_id_table needed at all anymore?

If the original driver is modified to switch to the one-device-instance-per-pwm approach, it can be removed. But changing the original driver will require changes to all the client device instantiations. Ac case can be made for the change, though. It purports to identify the number of pwms on the processor, but in fact fails at it.

Quoted 14 lines
Quoted 14 lines
Quoted 13 lines

[...]

drivers/pwm/pwm-pxa.c static int pwm_probe(struct platform_device *pdev) · line 205

Quoted 8 lines
drivers/pwm/pwm-pxa.c +2 -0
@@ -145,6 +199,8 @@ static int pwm_probe(struct platform_device *pdev)
 
 	pwm->chip.ops = &pxa_pwm_ops;
 	pwm->chip.base = -1;
 	pwm->chip.npwm = (id->driver_data & HAS_SECONDARY_PWM) ? 2 : 1;
 
+	pwm->chip.of_xlate = of_pwm_xlate_with_flags;
+	pwm->chip.of_pwm_n_cells = 3;

Are these two settings needed ?

Quoted 5 lines
Quoted 2 lines

Yes. See drivers/pwm/core.c:of_pwmchip_add(), where they are set to default values of of_pwm_simple_xlate and 2 if left uninitialized.

OK

Thanks,
Mike

#5 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

In reply to Mike Dunn
Quoted 1 line · Mike Dunn

This patch adds device tree support to the pxa's pwm driver. Only an OF match

"PXA" and "PWM"

Quoted 2 lines

table is added; nothing needs to be extracted from the device tree node. The existing platform_device id table is reused for the match table data. Support

"ID table"

Quoted 10 lines

Changle log:
v2:
- of_match_table contains only the "pxa250-pwm" compatible string; require one device instance per pwm
- add Documentation/devicetree/bindings/pwm/pxa-pwm.txt
- add support for polarity flag in DT and implement set_polarity() method (the treo 680 inverts the signal between pwm out and backlight)
- return -EINVAL instead of -ENODEV if platform data or DT node not found
- output dev_info string if platform data missing
- expanded CC list of patch

I think this needs to be reviewed by the device tree bindings maintainers (as listed in MAINTAINERS) as well. Would you mind resending the patch with all of them on Cc so that they have full context? Thanks.

Quoted 12 lines
 Documentation/devicetree/bindings/pwm/pxa-pwm.txt | 33 +++++++++++++
 arch/arm/boot/dts/pxa27x.dtsi                     | 24 ++++++++++
 drivers/pwm/pwm-pxa.c                             | 57 +++++++++++++++++++++++
 3 files changed, 114 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/pwm/pxa-pwm.txt
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +0 -0
new file mode 100644
index 0000000..7b09bfa

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 6

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +6 -0
@@ -0,0 +1,33 @@
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible = "marvell,pxa250-pwm";
+- reg: physical base address and length of the registers used by the pwm channel

"pwm" -> "PWM". There are a few more occurrences that I haven't explicitly pointed out.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 9

Quoted 3 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +3 -0
+  NB: One device instance must be created for each pwm that is used, so the
+  length covers only the register window for one pwm output, not that of the
+  entire pwm controller.  Currently length is 0x10 for all supported devices.

I'm not sure I like that very much. It seems a wrong representation of the hardware for the sake of modifying a smaller number of lines in the driver.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 13

Quoted 4 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +4 -0
+- #pwm-cells: should be 3.
+   cell 1: the per-chip index of the PWM to use,
+   cell 2: the period in nanoseconds,
+   cell 3: flags; set bit 0 to specify inverse polarity.  The pwm controller

This is the standard binding, so you can just refer to the standard binding documentation instead, like so:

- #pwm-cells: Should be 2. See pwm.txt in this directory for a description of the cells format.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 17

Quoted 4 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +4 -0
+   does not implement reverse polarity, but some boards may pass the pwm output
+   through an external inverter, which can cause problems if a client device
+   assumes that an increase in the duty cycle results in an increase in output
+   power.  The pwm-backlight is an example of such a client.

Please don't do that. Reverse polarity support should not be emulated. Either the hardware supports it or it doesn't. In the above case you should be able to achieve the same effect by adding the correct values in the pwm-backlight device node's brightness-levels property.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 32

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +6 -0
+Example pwm client node:
+
+backlight {
+      	compatible = "pwm-backlight";
+     	pwms = <&pwm0 0 5000000 1>;  /* pwm output is inverted */
+	...

The indentation here is funky. You should be using only tabs to indent.

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +6 -0
+#ifdef CONFIG_OF
+/*
+ * Device tree users should create one device instance for each pwm channel.
+ * Hence we dispense with the HAS_SECONDARY_PWM and "tell" the original driver
+ * code that this is a single channel pxa25x-pwm.
+ */

Documentation/devicetree/bindings/pwm/pxa-pwm.txt

Quoted 10 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +10 -0
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},
+	{ }
+};
+MODULE_DEVICE_TABLE(of, pwm_of_match);
+
+static const struct platform_device_id *pxa_pwm_get_id_dt(struct device *dev)
+{
+	const struct of_device_id *pwm_pxa_devid =
+		of_match_device(pwm_of_match, dev);

If you name this variable something much shorter, like "id", this will fit on a single line, and the code below will be slightly more readable as well.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt

Quoted 5 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +5 -0
+	if (pwm_pxa_devid)
+		return (const struct platform_device_id *)pwm_pxa_devid->data;
+	else
+		return NULL;
+}

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20130909/11a5bcaf/attachment.sig>

author
#6 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Quoted 4 lines · Thierry Reding
Quoted 1 line · Mike Dunn

This patch adds device tree support to the pxa's pwm driver. Only an OF match

"PXA" and "PWM"

OK. And thank you for the review!

Quoted 5 lines
Quoted 3 lines

table is added; nothing needs to be extracted from the device tree node. The existing platform_device id table is reused for the match table data. Support

"ID table"

OK

Quoted 12 lines
Quoted 11 lines

Changle log:
v2:
- of_match_table contains only the "pxa250-pwm" compatible string; require one device instance per pwm
- add Documentation/devicetree/bindings/pwm/pxa-pwm.txt
- add support for polarity flag in DT and implement set_polarity() method (the treo 680 inverts the signal between pwm out and backlight)
- return -EINVAL instead of -ENODEV if platform data or DT node not found
- output dev_info string if platform data missing
- expanded CC list of patch

Quoted 3 lines

I think this needs to be reviewed by the device tree bindings maintainers (as listed in MAINTAINERS) as well. Would you mind resending the patch with all of them on Cc so that they have full context? Thanks.

You bet. Sorry... should have checked that. Adding them to this reply, and will resend original patch.

Quoted 23 lines
Quoted 13 lines
 Documentation/devicetree/bindings/pwm/pxa-pwm.txt | 33 +++++++++++++
 arch/arm/boot/dts/pxa27x.dtsi                     | 24 ++++++++++
 drivers/pwm/pwm-pxa.c                             | 57 +++++++++++++++++++++++
 3 files changed, 114 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/pwm/pxa-pwm.txt
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +0 -0
new file mode 100644
index 0000000..7b09bfa

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 6

Quoted 6 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +6 -0
@@ -0,0 +1,33 @@
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible = "marvell,pxa250-pwm";
+- reg: physical base address and length of the registers used by the pwm channel

"pwm" -> "PWM". There are a few more occurrences that I haven't explicitly pointed out.

Quoted 7 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 9

Quoted 3 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +3 -0
+  NB: One device instance must be created for each pwm that is used, so the
+  length covers only the register window for one pwm output, not that of the
+  entire pwm controller.  Currently length is 0x10 for all supported devices.

I'm not sure I like that very much. It seems a wrong representation of the hardware for the sake of modifying a smaller number of lines in the driver.

I don't like it either, but this is an existing driver defect that will need to be corrected. To do so, I will need to to survey all the supported processors to determine how many pwm outputs is posessed by each. And there may be some confusion in that regard; the driver says "pxa25x" has one, but my pxa255 developer's manual makes no mention of a pwm. The pxa270 I am testing on has four, but the driver says "pxa27x" has two, so currently (using platform_data with existing driver) two devices are instantiated, each with two pwms. It seems that there are many variations within a single generation of the processor family, so to correctly identify the number of pwms, processor identification will have to be made on a finer granularity than the current "pxa25x", for example. I'm guessing that the hardware designers had this one-device-instance-per-pwm approach in mind when they made the decision to segregate the register sets of each pwm. I really hope that this sin can be forgiven :)

Quoted 17 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 13

Quoted 5 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +4 -0
+- #pwm-cells: should be 3.
+   cell 1: the per-chip index of the PWM to use,
+   cell 2: the period in nanoseconds,
+   cell 3: flags; set bit 0 to specify inverse polarity.  The pwm controller

This is the standard binding, so you can just refer to the standard binding documentation instead, like so:

- #pwm-cells: Should be 2. See pwm.txt in this directory for a description of the cells format.

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 17

Quoted 4 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +4 -0
+   does not implement reverse polarity, but some boards may pass the pwm output
+   through an external inverter, which can cause problems if a client device
+   assumes that an increase in the duty cycle results in an increase in output
+   power.  The pwm-backlight is an example of such a client.
Quoted 4 lines

Please don't do that. Reverse polarity support should not be emulated. Either the hardware supports it or it doesn't. In the above case you should be able to achieve the same effect by adding the correct values in the pwm-backlight device node's brightness-levels property.

That was my initial thought too. But sadly, that is not the case. The pwm-backlight driver assumes that the brightness-levels increase when parsed from left to right (more specifically, it assumes the last one is max). If emulated polarity is unacceptable, the pwm-backlight driver will need some rework to be usable in my case. The change looks minor and straightforward at first glance, though.

Quoted 9 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 32

Quoted 7 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +6 -0
+Example pwm client node:
+
+backlight {
+      	compatible = "pwm-backlight";
+     	pwms = <&pwm0 0 5000000 1>;  /* pwm output is inverted */
+	...

The indentation here is funky. You should be using only tabs to indent.

OK

Quoted 13 lines
Quoted 13 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +12 -0
+#ifdef CONFIG_OF
+/*
+ * Device tree users should create one device instance for each pwm channel.
+ * Hence we dispense with the HAS_SECONDARY_PWM and "tell" the original driver
+ * code that this is a single channel pxa25x-pwm.
+ */
+static struct of_device_id pwm_of_match[] = {
+	{ .compatible = "marvell,pxa250-pwm", .data = &pwm_id_table[0]},
+	{ }
+};
+MODULE_DEVICE_TABLE(of, pwm_of_match);
+
Quoted 8 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt

Quoted 4 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt new file +4 -0
+static const struct platform_device_id *pxa_pwm_get_id_dt(struct device *dev)
+{
+	const struct of_device_id *pwm_pxa_devid =
+		of_match_device(pwm_of_match, dev);

If you name this variable something much shorter, like "id", this will fit on a single line, and the code below will be slightly more readable as well.

OK. I struggle not to be too verbose :)

Thanks again,
Mike

#7 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Quoted 24 lines · Mike Dunn
Quoted 15 lines · Thierry Reding

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 6

Quoted 10 lines · Mike Dunn

This patch adds device tree support to the pxa's pwm driver. Only an OF match

Documentation/devicetree/bindings/pwm/pxa-pwm.txt +6 -0
 
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible = "marvell,pxa250-pwm";
+- reg: physical base address and length of the registers used by the pwm channel

"pwm" -> "PWM". There are a few more occurrences that I haven't explicitly pointed out.

Quoted 7 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 9

Quoted 3 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt +3 -0
+  NB: One device instance must be created for each pwm that is used, so the
+  length covers only the register window for one pwm output, not that of the
+  entire pwm controller.  Currently length is 0x10 for all supported devices.

I'm not sure I like that very much. It seems a wrong representation of the hardware for the sake of modifying a smaller number of lines in the driver.

Quoted 14 lines

I don't like it either, but this is an existing driver defect that will need to be corrected. To do so, I will need to to survey all the supported processors to determine how many pwm outputs is posessed by each. And there may be some confusion in that regard; the driver says "pxa25x" has one, but my pxa255 developer's manual makes no mention of a pwm. The pxa270 I am testing on has four, but the driver says "pxa27x" has two, so currently (using platform_data with existing driver) two devices are instantiated, each with two pwms. It seems that there are many variations within a single generation of the processor family, so to correctly identify the number of pwms, processor identification will have to be made on a finer granularity than the current "pxa25x", for example. I'm guessing that the hardware designers had this one-device-instance-per-pwm approach in mind when they made the decision to segregate the register sets of each pwm. I really hope that this sin can be forgiven :)

The DT binding is an ABI, so it needs to correctly represent the HW. As such, I'd say that if the binding is wrong, we need to fix it even if it means a lot of work.

That said, if each PWM truly is a *completely* independent block, with non-overlapping register spaces, there's certainly an argument that it's perfectly acceptable to represent each PWM as a separate DT node...

author
#8 lore raw

PATCH v2 1/1 pwm: pxa: add device tree support to pwm driver

Quoted 40 lines · Stephen Warren
Quoted 24 lines · Mike Dunn
Quoted 15 lines · Thierry Reding

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 6

Quoted 10 lines · Mike Dunn

This patch adds device tree support to the pxa's pwm driver. Only an OF match

Documentation/devicetree/bindings/pwm/pxa-pwm.txt +6 -0
 
+Marvell pwm controller
+
+Required properties:
+- compatible:
+  for pxa25x, pxa27x, pxa168, pxa910: must be compatible = "marvell,pxa250-pwm";
+- reg: physical base address and length of the registers used by the pwm channel

"pwm" -> "PWM". There are a few more occurrences that I haven't explicitly pointed out.

Quoted 7 lines

Documentation/devicetree/bindings/pwm/pxa-pwm.txt line 9

Quoted 3 lines
Documentation/devicetree/bindings/pwm/pxa-pwm.txt +3 -0
+  NB: One device instance must be created for each pwm that is used, so the
+  length covers only the register window for one pwm output, not that of the
+  entire pwm controller.  Currently length is 0x10 for all supported devices.

I'm not sure I like that very much. It seems a wrong representation of the hardware for the sake of modifying a smaller number of lines in the driver.

Quoted 14 lines

I don't like it either, but this is an existing driver defect that will need to be corrected. To do so, I will need to to survey all the supported processors to determine how many pwm outputs is posessed by each. And there may be some confusion in that regard; the driver says "pxa25x" has one, but my pxa255 developer's manual makes no mention of a pwm. The pxa270 I am testing on has four, but the driver says "pxa27x" has two, so currently (using platform_data with existing driver) two devices are instantiated, each with two pwms. It seems that there are many variations within a single generation of the processor family, so to correctly identify the number of pwms, processor identification will have to be made on a finer granularity than the current "pxa25x", for example. I'm guessing that the hardware designers had this one-device-instance-per-pwm approach in mind when they made the decision to segregate the register sets of each pwm. I really hope that this sin can be forgiven :)

Quoted 7 lines

The DT binding is an ABI, so it needs to correctly represent the HW. As such, I'd say that if the binding is wrong, we need to fix it even if it means a lot of work.

That said, if each PWM truly is a *completely* independent block, with non-overlapping register spaces, there's certainly an argument that it's perfectly acceptable to represent each PWM as a separate DT node...

Thanks Stephen. Yes, each pwm output has its own registers with no common control registers. The only commonality is that they all share a clock in the clock unit. But this is handled in the clock manager.

Thanks,
Mike