Skip to contentSkip to conversation

LM3697 dedicated LED driver

PATCH In review Dan Murphy 5 participants 28 messages Sep 11, 2018 – Sep 25, 2018

Subsystems (3)
lore
Share & tools
lore

Dan Murphy

author
#0 lore raw

v7 cover LM3697 dedicated LED driver

All

The LM3697 LED driver has two opportunities to be included in the Linux kernel source.

The ti-lmu MFD driver and the leds-lm3697 driver contained in this series.

This LED driver is basic in nature and is not really suitable for the MFD inclusion. This device has no other function than to control the LEDs.

So this series will remove the lm3697 from the ti-lmu and add the driver in the LEDs directory.

Dan Murphy

Dan Murphy (6):
dt-bindings: ti-lmu: Remove LM3697
mfd: ti-lmu: Remove support for LM3697
dt-bindings: leds: Add bindings for lm3697 driver
leds: lm3697: Introduce the lm3697 driver
dt-bindings: leds: Add runtime ramp node for LM3697
leds: lm3697: Add ramp rate feature

 .../devicetree/bindings/leds/leds-lm3697.txt  |  98 ++++
 .../devicetree/bindings/mfd/ti-lmu.txt        |  26 +-
 drivers/leds/Kconfig                          |   9 +
 drivers/leds/Makefile                         |   1 +
 drivers/leds/leds-lm3697.c                    | 449 ++++++++++++++++++
 drivers/mfd/Kconfig                           |   2 +-
 drivers/mfd/ti-lmu.c                          |  17 -
 include/linux/mfd/ti-lmu-register.h           |  44 --
 include/linux/mfd/ti-lmu.h                    |   1 -
 9 files changed, 559 insertions(+), 88 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt
 create mode 100644 drivers/leds/leds-lm3697.c
Signature · 1 line
2.17.0.1855.g63749b2dea
author
#1 lore raw

PATCH v7 1/6 dt-bindings: ti-lmu: Remove LM3697

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

---

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 .../devicetree/bindings/mfd/ti-lmu.txt        | 26 +------------------
 1 file changed, 1 insertion(+), 25 deletions(-)
Documentation/devicetree/bindings/mfd/ti-lmu.txt +1 -25
index c885cf89b8ce..920f910be4e9 100644
@@ -9,7 +9,6 @@ TI LMU driver supports lighting devices below.
   LM3632       Backlight and regulator
   LM3633       Backlight, LED and fault monitor
   LM3695       Backlight
-  LM3697       Backlight and fault monitor
 
 Required properties:
   - compatible: Should be one of:
@@ -18,11 +17,10 @@ Required properties:
                 "ti,lm3632"
                 "ti,lm3633"
                 "ti,lm3695"
-                "ti,lm3697"
   - reg: I2C slave address.
          0x11 for LM3632
          0x29 for LM3631
-         0x36 for LM3633, LM3697
+         0x36 for LM3633
          0x38 for LM3532
          0x63 for LM3695
 
@@ -38,7 +36,6 @@ Optional nodes:
     Required properties:
       - compatible: Should be one of:
                     "ti,lm3633-fault-monitor"
-                    "ti,lm3697-fault-monitor"
   - leds: LED properties for LM3633. Please refer to [2].
   - regulators: Regulator properties for LM3631 and LM3632.
                 Please refer to [3].
@@ -220,24 +217,3 @@ lm3695@63 {
 		};
 	};
 };
-
-lm3697@36 {
-	compatible = "ti,lm3697";
-	reg = <0x36>;
-
-	enable-gpios = <&pioC 2 GPIO_ACTIVE_HIGH>;
-
-	backlight {
-		compatible = "ti,lm3697-backlight";
-
-		lcd {
-			led-sources = <0 1 2>;
-			ramp-up-msec = <200>;
-			ramp-down-msec = <200>;
-		};
-	};
-
-	fault-monitor {
-		compatible = "ti,lm3697-fault-monitor";
-	};
-};
Signature · 1 line
2.17.0.1855.g63749b2dea
#2 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Quoted 11 lines · Dan Murphy

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

---

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 .../devicetree/bindings/mfd/ti-lmu.txt        | 26 +------------------
 1 file changed, 1 insertion(+), 25 deletions(-)

For my own reference:

Signature · 4 lines
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
#3 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

In reply to Dan Murphy
Quoted 5 lines · Dan Murphy

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

I'd really like to see better explanation here.

We have existing binding, for lm3697 and similar devices. With this series, different binding is introduced, without documented reason.

That's bad.

Now, maybe you are right and the hardware should be handled by drivers/leds, not drivers/mfd. But we should have solution for all the similar chips, and that still does not mean we have to modify the binding. (But maybe we want to move it to different directory). Bindings are supposed to describe hardware, not mirror structure of our drivers.

Unless there's something fatally wrong with the binding... but in such case we'd like to know what is wrong.

[And yes, I recognize current situation is ... not ideal and I'm willing to help. But I'm not sure this is step in right direction.]

Thanks,
Pavel

Quoted 65 lines

---

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 .../devicetree/bindings/mfd/ti-lmu.txt        | 26 +------------------
 1 file changed, 1 insertion(+), 25 deletions(-)
Documentation/devicetree/bindings/mfd/ti-lmu.txt +1 -25
index c885cf89b8ce..920f910be4e9 100644
@@ -9,7 +9,6 @@ TI LMU driver supports lighting devices below.
   LM3632       Backlight and regulator
   LM3633       Backlight, LED and fault monitor
   LM3695       Backlight
-  LM3697       Backlight and fault monitor
 
 Required properties:
   - compatible: Should be one of:
@@ -18,11 +17,10 @@ Required properties:
                 "ti,lm3632"
                 "ti,lm3633"
                 "ti,lm3695"
-                "ti,lm3697"
   - reg: I2C slave address.
          0x11 for LM3632
          0x29 for LM3631
-         0x36 for LM3633, LM3697
+         0x36 for LM3633
          0x38 for LM3532
          0x63 for LM3695
 
@@ -38,7 +36,6 @@ Optional nodes:
     Required properties:
       - compatible: Should be one of:
                     "ti,lm3633-fault-monitor"
-                    "ti,lm3697-fault-monitor"
   - leds: LED properties for LM3633. Please refer to [2].
   - regulators: Regulator properties for LM3631 and LM3632.
                 Please refer to [3].
@@ -220,24 +217,3 @@ lm3695@63 {
 		};
 	};
 };
-
-lm3697@36 {
-	compatible = "ti,lm3697";
-	reg = <0x36>;
-
-	enable-gpios = <&pioC 2 GPIO_ACTIVE_HIGH>;
-
-	backlight {
-		compatible = "ti,lm3697-backlight";
-
-		lcd {
-			led-sources = <0 1 2>;
-			ramp-up-msec = <200>;
-			ramp-down-msec = <200>;
-		};
-	};
-
-	fault-monitor {
-		compatible = "ti,lm3697-fault-monitor";
-	};
-};
Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
author
#4 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Pavel

Quoted 9 lines · Pavel Machek
Quoted 5 lines · Dan Murphy

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

I'd really like to see better explanation here.

I will update the commit message but....

How do I politely explain that the original implementation was wrong for certain devices? How do I politely explain that this binding has information that does not exist? Isn't code and documentation supposed to be pushed in stages together?
Like document the current supported source features and then submit patches to amend new features or changes to the source. I did that in this series
where I introduced the driver and when I added a feature I added the binding information.

Unfortunately Milo left TI before he had a chance to finish upstreaming.

Some issues with the current binding:
1) The documentation referenced in the ti-lmu.txt does not exist but the binding was accepted.
[1] ../leds/backlight/ti-lmu-backlight.txt
[2] ../leds/leds-lm3633.txt
[3] ../regulator/lm363x-regulator.txt

2) ramp-up-msec and ramp-down-msec are not explained or defined in this current binding. Looks to be defined in the staged code though.

3) The ramp-up-msec and ramp-down-msec are not the right node parameters the code shows ti,ramp-down-ms and ti,ramp-up-ms. Looking at the clean up of the binding in the staged commits the binding still does not match the code.

https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

4) The ramp rates ranges are incorrect for the LM3697 it should be from 0 -> 117400 ms

5) The children compatibles are not defined in this binding but appear to be removed in the staged code.

6) address-cells and size-cells are missing from the binding

I am attempting to clean this up for all the dedicated LED drivers.
There is a lot of bloat with this driver to support a single LED driver that is a single function device.

This ti-lmu driver may be a great thing for the Droid 4 but from a customers stand point that just wants to use just 1 of the dedicated LED drivers this driver is overly complex and possibly hard to add code for differentiation.

Quoted 3 lines

We have existing binding, for lm3697 and similar devices. With this series, different binding is introduced, without documented reason.

I can update the commit message to indicate that this device is not a MFD device and only is a SFD that needs to be supported by a dedicated driver.

Quoted 1 line

That's bad.

This ti-lmu binding is more misleading in the current state.

Quoted 7 lines

Now, maybe you are right and the hardware should be handled by drivers/leds, not drivers/mfd. But we should have solution for all the similar chips, and that still does not mean we have to modify the binding. (But maybe we want to move it to different directory). Bindings are supposed to describe hardware, not mirror structure of our drivers.

But this is not clean to have this device in the MFD bindings directory when the support is in the LEDs directory.

I am working on creating driver for the other single function devices. It will take some time. I am waiting on the hardware to come in.

Quoted 6 lines

Unless there's something fatally wrong with the binding... but in such case we'd like to know what is wrong.

[And yes, I recognize current situation is ... not ideal and I'm willing to help. But I'm not sure this is step in right direction.]

It will take some time to get this all cleaned up. But I am willing to get this done.

Dan

Quoted 71 lines

Thanks,
Pavel

Quoted 65 lines

---

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 .../devicetree/bindings/mfd/ti-lmu.txt        | 26 +------------------
 1 file changed, 1 insertion(+), 25 deletions(-)
Documentation/devicetree/bindings/mfd/ti-lmu.txt +1 -25
index c885cf89b8ce..920f910be4e9 100644
@@ -9,7 +9,6 @@ TI LMU driver supports lighting devices below.
   LM3632       Backlight and regulator
   LM3633       Backlight, LED and fault monitor
   LM3695       Backlight
-  LM3697       Backlight and fault monitor
 
 Required properties:
   - compatible: Should be one of:
@@ -18,11 +17,10 @@ Required properties:
                 "ti,lm3632"
                 "ti,lm3633"
                 "ti,lm3695"
-                "ti,lm3697"
   - reg: I2C slave address.
          0x11 for LM3632
          0x29 for LM3631
-         0x36 for LM3633, LM3697
+         0x36 for LM3633
          0x38 for LM3532
          0x63 for LM3695
 
@@ -38,7 +36,6 @@ Optional nodes:
     Required properties:
       - compatible: Should be one of:
                     "ti,lm3633-fault-monitor"
-                    "ti,lm3697-fault-monitor"
   - leds: LED properties for LM3633. Please refer to [2].
   - regulators: Regulator properties for LM3631 and LM3632.
                 Please refer to [3].
@@ -220,24 +217,3 @@ lm3695@63 {
 		};
 	};
 };
-
-lm3697@36 {
-	compatible = "ti,lm3697";
-	reg = <0x36>;
-
-	enable-gpios = <&pioC 2 GPIO_ACTIVE_HIGH>;
-
-	backlight {
-		compatible = "ti,lm3697-backlight";
-
-		lcd {
-			led-sources = <0 1 2>;
-			ramp-up-msec = <200>;
-			ramp-down-msec = <200>;
-		};
-	};
-
-	fault-monitor {
-		compatible = "ti,lm3697-fault-monitor";
-	};
-};
Signature · 2 lines
------------------
Dan Murphy
#5 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Hi!

Quoted 6 lines · Pavel Machek
Quoted 5 lines · Dan Murphy
Quoted 4 lines

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

I'd really like to see better explanation here.

I will update the commit message but....

How do I politely explain that the original implementation was wrong for certain devices?

Implementation? Device tree is hardware description.

Quoted 2 lines

How do I politely explain that this binding has information that
does not exist?

I don't understand this sentence.

Quoted 2 lines

Isn't code and documentation supposed to be pushed in stages together?

Device tree is _not_ documentation. And yes, it is normally pushed together. But that did not happen here, and bindings are already in.

Bindings are an ABI between bootloader and kernel. They should not be changed lightly.

Here I believe they should be fixed; not deleted.

Quoted 1 line

Unfortunately Milo left TI before he had a chance to finish upstreaming.

Well -- I wanted to finish upstreaming. I still have those patches, and they still look reasonably well. Better than

Quoted 6 lines

Some issues with the current binding:
1) The documentation referenced in the ti-lmu.txt does not exist but the binding was accepted.
[1] ../leds/backlight/ti-lmu-backlight.txt
[2] ../leds/leds-lm3633.txt
[3] ../regulator/lm363x-regulator.txt

Quoted 9 lines

2) ramp-up-msec and ramp-down-msec are not explained or defined in this current binding. Looks to be defined in the staged code though.

3) The ramp-up-msec and ramp-down-msec are not the right node parameters the code shows ti,ramp-down-ms and ti,ramp-up-ms. Looking at the clean up of the binding in the staged commits the binding still does not match the code.

https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

Bindings are authoritative. Of course, they can be fixed if really neccessary, but normally code should be adapted to the bindings, not the other way around.

If slow ramping up/down is something that would be expected in other chips, too, ti, prefix is not good idea.

Quoted 5 lines

4) The ramp rates ranges are incorrect for the LM3697 it should be from 0 -> 117400 ms

5) The children compatibles are not defined in this binding but appear to be removed in the staged code.

6) address-cells and size-cells are missing from the binding

Yep, ok, so these should be fixed. Still the bindings should be fixed, not removed and started over.

Quoted 3 lines

I am attempting to clean this up for all the dedicated LED drivers.
There is a lot of bloat with this driver to support a single LED driver that is a single function device.

What about the multi function devices? They should have same binding.

Quoted 3 lines

This ti-lmu driver may be a great thing for the Droid 4 but from a customers stand point that just wants to use just 1 of the dedicated LED drivers this driver is overly complex and possibly hard to add code for differentiation.

The ti-lmu driver sounds like a way to go. We really want single driver, even if more complex, than 6 separate ones.

Quoted 7 lines
Quoted 3 lines

We have existing binding, for lm3697 and similar devices. With this series, different binding is introduced, without documented reason.

I can update the commit message to indicate that this device is not a MFD device and only is a SFD that needs to be supported by a dedicated driver.

Device bindings are operating systems independend. This is not an argument.

Quoted 9 lines
Quoted 6 lines

Now, maybe you are right and the hardware should be handled by drivers/leds, not drivers/mfd. But we should have solution for all the similar chips, and that still does not mean we have to modify the binding. (But maybe we want to move it to different directory). Bindings are supposed to describe hardware, not mirror structure of our drivers.

But this is not clean to have this device in the MFD bindings directory when the support is in the LEDs directory.

Perhaps location of the binding is wrong. Still not reason to drop it and introduce new one.

Quoted 11 lines
Quoted 11 lines

Documentation/devicetree/bindings/mfd/ti-lmu.txt TI LMU driver supports lighting devices below. · line 14

Quoted 11 lines

index c885cf89b8ce..920f910be4e9 100644

Documentation/devicetree/bindings/mfd/ti-lmu.txt +0 -1
@@ -9,7 +9,6 @@ TI LMU driver supports lighting devices below.
   LM3632       Backlight and regulator
   LM3633       Backlight, LED and fault monitor
   LM3695       Backlight
-  LM3697       Backlight and fault monitor
 
 Required properties:
   - compatible: Should be one of:

You see? Even if your "new" binding has no flaws, "flawed" binding still remains in tree, and we still have problems for lm3632, 3633, 3695. That's why it needs to be fixed, not replaced.

Best regards,
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
author
#6 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Pavel

Quoted 10 lines · Pavel Machek

Hi!

Quoted 8 lines · Pavel Machek
Quoted 7 lines · Dan Murphy
Quoted 5 lines

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

I'd really like to see better explanation here.

I will update the commit message but....

How do I politely explain that the original implementation was wrong for certain devices?

Implementation? Device tree is hardware description.

Yes this hardware description is incorrect. The hardware description is describing a MFD but this LED driver (and a couple others) only perform 1 function and that is to drive a LED string.

Quoted 5 lines
Quoted 3 lines

How do I politely explain that this binding has information that
does not exist?

I don't understand this sentence.

That was explained later on the issues with the binding.

Quoted 7 lines
Quoted 3 lines

Isn't code and documentation supposed to be pushed in stages together?

Device tree is _not_ documentation. And yes, it is normally pushed together. But that did not happen here, and bindings are already in.

Hmm.. Its not documentation but it is in the Documentation folder. And just because the bindings are in does not mean they cannot be changed.

So someone hurried up and pushed the bindings that were not accurate and they were taken and now we are stuck with this implementation?

That does not seem like progress.

Quoted 5 lines

Bindings are an ABI between bootloader and kernel. They should not be changed lightly.

Here I believe they should be fixed; not deleted.

I am fixing them. Removing devices that should not have been there to begin with.

Also if they are not to be changed lightly then why is there so much churn in the staged patches. Looks like the compatible is changing with this patch and appears that it was not documented appropriately.
According to the above statement this patch should be rejected because it is changing the ABI.

https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

Quoted 5 lines
Quoted 1 line

Unfortunately Milo left TI before he had a chance to finish upstreaming.

Well -- I wanted to finish upstreaming. I still have those patches, and they still look reasonably well. Better than

than?

Quoted 13 lines
Quoted 13 lines

Some issues with the current binding:
1) The documentation referenced in the ti-lmu.txt does not exist but the binding was accepted.
[1] ../leds/backlight/ti-lmu-backlight.txt
[2] ../leds/leds-lm3633.txt
[3] ../regulator/lm363x-regulator.txt

2) ramp-up-msec and ramp-down-msec are not explained or defined in this current binding. Looks to be defined in the staged code though.

3) The ramp-up-msec and ramp-down-msec are not the right node parameters the code shows ti,ramp-down-ms and ti,ramp-up-ms. Looking at the clean up of the binding in the staged commits the binding still does not match the code.

Quoted 6 lines
Quoted 2 lines

https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

Bindings are authoritative. Of course, they can be fixed if really neccessary, but normally code should be adapted to the bindings, not the other way around.

They should only be authoritative when they are correct.

Quoted 4 lines

If slow ramping up/down is something that would be expected in other chips, too, ti, prefix is not good idea.

Well if this is something expected for other chips shouldn't the LED framework be updated as opposed to creating a stub layer to do this?

Quoted 9 lines
Quoted 5 lines

4) The ramp rates ranges are incorrect for the LM3697 it should be from 0 -> 117400 ms

5) The children compatibles are not defined in this binding but appear to be removed in the staged code.

6) address-cells and size-cells are missing from the binding

Yep, ok, so these should be fixed. Still the bindings should be fixed, not removed and started over.

So once something is in the binding it can never be removed?

Quoted 5 lines
Quoted 3 lines

I am attempting to clean this up for all the dedicated LED drivers.
There is a lot of bloat with this driver to support a single LED driver that is a single function device.

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.
The only comment there is data section bloat that occurs in attempting to support only one device out of the supported devices. But that can be done via #ifconfig

Quoted 7 lines
Quoted 4 lines

This ti-lmu driver may be a great thing for the Droid 4 but from a customers stand point that just wants to use just 1 of the dedicated LED drivers this driver is overly complex and possibly hard to add code for differentiation.

The ti-lmu driver sounds like a way to go. We really want single driver, even if more complex, than 6 separate ones.

Not really in agreement here. The ti-lmu driver only makes sense for MFD devices where the device itself supports more then just one function. Otherwise we could just lump every single LED device into this driver.

Again not sure why we gloss over the fact that this driver with all of the device data is the correct way to go for every single device. If a target product only has the LM3697 why does the MFD need to be used? Why do we need to bring in all the support for all the additional devices?

This may be great for Droid 4 but not for any other device. Including IoT devices that need smaller kernel foot prints.

All be it the analog side of these devices may be the same but the digital side for each chip is different. The register maps are not the same, the number of supported LED strings are not
the same, there are features not supported by this driver and will be a pain to add. Adding features will be a complete night mare especially for the LM3632 which also support Torch and Flash.

Here are where a single driver will start getting messy and support difficult LM3533 has ALS and an ADC for an ALS analog sensor
LM3631 has no ALS functionality
LM3632 has strobe/torch functionality and no ramp support
LM3633 has lvled support coupled with the hvled support
LM3695 does not even appear to be available publicly
LM3697 is the only device that that this driver could be used for as is.

Jacek has already agreed that having a dedicated LED driver for SFD devices is preferred method.

Quoted 11 lines
Quoted 8 lines
Quoted 4 lines

We have existing binding, for lm3697 and similar devices. With this series, different binding is introduced, without documented reason.

I can update the commit message to indicate that this device is not a MFD device and only is a SFD that needs to be supported by a dedicated driver.

Device bindings are operating systems independend. This is not an argument.

I have no idea what this statement means.

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

Now, maybe you are right and the hardware should be handled by drivers/leds, not drivers/mfd. But we should have solution for all the similar chips, and that still does not mean we have to modify the binding. (But maybe we want to move it to different directory). Bindings are supposed to describe hardware, not mirror structure of our drivers.

Quoted 6 lines
Quoted 2 lines

But this is not clean to have this device in the MFD bindings directory when the support is in the LEDs directory.

Perhaps location of the binding is wrong. Still not reason to drop it and introduce new one.

Location and implementation is wrong for certain devices.

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

index c885cf89b8ce..920f910be4e9 100644

Documentation/devicetree/bindings/mfd/ti-lmu.txt +0 -1
@@ -9,7 +9,6 @@ TI LMU driver supports lighting devices below.
   LM3632       Backlight and regulator
   LM3633       Backlight, LED and fault monitor
   LM3695       Backlight
-  LM3697       Backlight and fault monitor
 
Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

Documentation/devicetree/bindings/mfd/ti-lmu.txt TI LMU driver supports lighting devices below. · line 14

Quoted 2 lines
Documentation/devicetree/bindings/mfd/ti-lmu.txt +0 -0
 Required properties:
   - compatible: Should be one of:

You see? Even if your "new" binding has no flaws, "flawed" binding still remains in tree, and we still have problems for lm3632, 3633, 3695. That's why it needs to be fixed, not replaced.

I am working on those as well. LM3632 is a MFD device and can be supported via the MFD framework.

This work for the other LED drivers needs to be done before we bring in any other code.

Dan

Quoted 4 lines

Best regards,
Pavel

Signature · 2 lines
------------------
Dan Murphy
#7 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Hi!

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

How do I politely explain that the original implementation was wrong for certain devices?

Implementation? Device tree is hardware description.

Yes this hardware description is incorrect. The hardware description is describing a MFD but this LED driver (and a couple others) only perform 1 function and that is to drive a LED string.

So what? Does not seem incorrect to me. Maybe the description should not be in MFD directory, but other than that...

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

Isn't code and documentation supposed to be pushed in stages together?

Device tree is _not_ documentation. And yes, it is normally pushed together. But that did not happen here, and bindings are already in.

Hmm.. Its not documentation but it is in the Documentation folder. And just because the bindings are in does not mean they cannot be changed.

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Best regards,
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#8 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Hi Pavel,

Quoted 19 lines · Pavel Machek

Hi!

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

How do I politely explain that the original implementation was wrong for certain devices?

Implementation? Device tree is hardware description.

Yes this hardware description is incorrect. The hardware description is describing a MFD but this LED driver (and a couple others) only perform 1 function and that is to drive a LED string.

So what? Does not seem incorrect to me. Maybe the description should not be in MFD directory, but other than that...

Quoted 6 lines
Quoted 5 lines
Quoted 2 lines

Isn't code and documentation supposed to be pushed in stages together?

Device tree is _not_ documentation. And yes, it is normally pushed together. But that did not happen here, and bindings are already in.

Quoted 6 lines
Quoted 3 lines

Hmm.. Its not documentation but it is in the Documentation folder. And just because the bindings are in does not mean they cannot be changed.

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Having the above it seems that there is no risk of breaking any users.

Quoted 7 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

Quoted 8 lines

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Easy to maintain will be a dedicated LED class driver.

Signature · 2 lines
Best regards,
Jacek Anaszewski
#9 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Hi!

Quoted 11 lines
Quoted 2 lines

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 7 lines
Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Having the above it seems that there is no risk of breaking any users.

DTBs and bindings are supposed to be portable between operating systems. You are right there are no _mainline_ _Linux_ users.

Quoted 12 lines
Quoted 12 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Quoted 5 lines
Quoted 3 lines

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#10 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Hi Pavel.

Quoted 18 lines · Pavel Machek

Hi!

Quoted 16 lines
Quoted 2 lines

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Quoted 5 lines
Quoted 2 lines

Having the above it seems that there is no risk of breaking any users.

DTBs and bindings are supposed to be portable between operating systems. You are right there are no _mainline_ _Linux_ users.

No mainline users means no users we should care of. Other people also don't care - see patch [0].

Quoted 16 lines
Quoted 16 lines
Quoted 15 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Quoted 5 lines
Quoted 1 line

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

[0]
https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

Signature · 2 lines
Best regards,
Jacek Anaszewski
author
#11 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Jacek

Quoted 46 lines · Jacek Anaszewski

Hi Pavel.

Quoted 18 lines · Pavel Machek

Hi!

Quoted 16 lines
Quoted 2 lines

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Quoted 5 lines
Quoted 2 lines

Having the above it seems that there is no risk of breaking any users.

DTBs and bindings are supposed to be portable between operating systems. You are right there are no _mainline_ _Linux_ users.

No mainline users means no users we should care of. Other people also don't care - see patch [0].

Quoted 16 lines
Quoted 16 lines
Quoted 15 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

I am currently working on that code now. I expect a RFC on this this week.

Dan

Quoted 4 lines

[0]
https://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git/commit/?h=droid4-pending-v4.19&id=d774c7e447ac911e73a1b3c775e6d89f0422218c

Signature · 2 lines
------------------
Dan Murphy
#12 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Dan,

Quoted 51 lines · Dan Murphy

Jacek

Quoted 48 lines · Jacek Anaszewski

Hi Pavel.

Quoted 18 lines · Pavel Machek

Hi!

Quoted 16 lines
Quoted 2 lines

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Quoted 5 lines
Quoted 2 lines

Having the above it seems that there is no risk of breaking any users.

DTBs and bindings are supposed to be portable between operating systems. You are right there are no _mainline_ _Linux_ users.

No mainline users means no users we should care of. Other people also don't care - see patch [0].

Quoted 18 lines
Quoted 17 lines
Quoted 7 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

Quoted 8 lines

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Easy to maintain will be a dedicated LED class driver.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

I am currently working on that code now. I expect a RFC on this this week.

Will it affect the shape of the driver for lm3697 as of v7 [0]? Or the patch set [0] state should be deemed "awaiting merge"?

[0] https://lkml.org/lkml/2018/9/11/760

Signature · 2 lines
Best regards,
Jacek Anaszewski
author
#13 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Jacek

Quoted 58 lines · Jacek Anaszewski

Dan,

Quoted 55 lines · Dan Murphy

Jacek

Quoted 52 lines · Jacek Anaszewski

Hi Pavel.

Quoted 18 lines · Pavel Machek

Hi!

Quoted 16 lines
Quoted 2 lines

You may want to learn more about device tree and/or talk to the device tree maintainers. This is an old article. https://lwn.net/Articles/561462/

The article title is "Device trees as ABI". A device tree is defined in the "*.dts" file that is then compiled to a dtb blob, which constitutes the ABI. And this ABI should be kept backwards compatible.

What is discussed here is a documentation of bindings, i.e. according to ePAPR: "requirements for how specific types and classes of devices are represented in the device tree".

Quoted 1 line

From the bindings documented in the

Documentation/devicetree/bindings/mfd/ti-lmu.txt only ti,lm3532-backlight is used in the mainline dts file (arch/arm/boot/dts/omap4-droid4-xt894.dts).

Quoted 5 lines
Quoted 2 lines

Having the above it seems that there is no risk of breaking any users.

DTBs and bindings are supposed to be portable between operating systems. You are right there are no _mainline_ _Linux_ users.

No mainline users means no users we should care of. Other people also don't care - see patch [0].

Quoted 21 lines
Quoted 12 lines
Quoted 12 lines

NAK on this patch. I see that this binding has problems, but introducing different binding for subset of devices is _not_ a fix.

Quoted 3 lines
Quoted 1 line

What about the multi function devices? They should have same binding.

The MFD devices defined are not in contention here only the SFD.

I'd like to see common solutions for SFD and MFD, as the hardware is similar, and that includes the code. Having code that is easier to maintain is important, and having many drivers are harder to maintain than one driver.

Quoted 5 lines
Quoted 3 lines

Milo's code looks better than yours in that regard. I disagree about Milo's code being "nightmare" to modify, and care about "easy to maintain" more than "binary size".

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

I am currently working on that code now. I expect a RFC on this this week.

Will it affect the shape of the driver for lm3697 as of v7 [0]? Or the patch set [0] state should be deemed "awaiting merge"?

It will probably affect the driver to call into the library. We can pull it in and modify or we can modify after the common code is there.

It should not affect the bindings though.

Dan

Quoted 2 lines

[0] https://lkml.org/lkml/2018/9/11/760

Signature · 2 lines
------------------
Dan Murphy
#14 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

In reply to Dan Murphy

Hi!

Quoted 10 lines
Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

I am currently working on that code now. I expect a RFC on this this week.

Looking forward to that.

But you really need acks for the bindings, and since Rob is usually quite slow acking them, it is easiest to use the existing binding... if it is wrong it needs to be fixed, anyway.

THanks,

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
author
#15 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

Pavel

Quoted 11 lines · Pavel Machek

Hi!

Quoted 9 lines
Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Easy to maintain will be a dedicated LED class driver.

You mean, 3 dedicated LED class drivers and 3 MFD drivers with LED parts? We'll need complex driver anyway, and I'd really like to have just one.

In the LED subsystem we can wrap common functionalities into a library object. MFD driver will be able to reuse it then.

Quoted 8 lines
Quoted 1 line

I am currently working on that code now. I expect a RFC on this this week.

Looking forward to that.

But you really need acks for the bindings, and since Rob is usually quite slow acking them, it is easiest to use the existing binding... if it is wrong it needs to be fixed, anyway.

OK I am in the midst of creating the code now. I have the common code done I just need to make sure that it scales to other devices in the list.

The bindings will need to be updated to follow the LED bindings binding style.

I am wondering for the non-MFD parts if we should move the bindings to the LED directory to make them easier to find.

Dan

Quoted 4 lines

THanks,

Pavel

Signature · 2 lines
------------------
Dan Murphy
#16 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

In reply to Dan Murphy

Hi!

Quoted 4 lines

All be it the analog side of these devices may be the same but the digital side for each chip is different. The register maps are not the same, the number of supported LED strings are not
the same, there are features not supported by this driver and will
be a pain to add.

Digital side looks similar enough that Milo was able to share code between them.

Yes, obviously register map will be different.

Quoted 10 lines

Adding features will be a complete night mare especially for the LM3632 which also support Torch and Flash.

Here are where a single driver will start getting messy and support difficult LM3533 has ALS and an ADC for an ALS analog sensor
LM3631 has no ALS functionality
LM3632 has strobe/torch functionality and no ramp support
LM3633 has lvled support coupled with the hvled support
LM3695 does not even appear to be available publicly
LM3697 is the only device that that this driver could be used for as is.

We already been through this once. More than one of these has ramp support, and more than one has overvoltage protection. (Plus, all of them are LED drivers).

Surely non-ugly design can be done where the code is shared?

Thanks,
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#17 lore raw

Re: [PATCH v7 1/6] dt-bindings: ti-lmu: Remove LM3697

In reply to Pavel Machek
Quoted 21 lines · Pavel Machek
Quoted 5 lines · Dan Murphy

Remove support for the LM3697 LED device
from the ti-lmu. The LM3697 will be supported
via a stand alone LED driver.

I'd really like to see better explanation here.

We have existing binding, for lm3697 and similar devices. With this series, different binding is introduced, without documented reason.

That's bad.

Now, maybe you are right and the hardware should be handled by drivers/leds, not drivers/mfd. But we should have solution for all the similar chips, and that still does not mean we have to modify the binding. (But maybe we want to move it to different directory). Bindings are supposed to describe hardware, not mirror structure of our drivers.

Quoted 2 lines

Unless there's something fatally wrong with the binding... but in such case we'd like to know what is wrong.

Dangling references ?

Quoted 4 lines

[And yes, I recognize current situation is ... not ideal and I'm willing to help. But I'm not sure this is step in right direction.]

Thanks,

Signature · 2 lines
Best regards,
Jacek Anaszewski
author
#18 lore raw

PATCH v7 2/6 mfd: ti-lmu: Remove support for LM3697

In reply to Dan Murphy

Remove support for the LM3697 from the ti-lmu
driver in favor of a dedicated LED driver.

---

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 drivers/mfd/Kconfig                 |  2 +-
 drivers/mfd/ti-lmu.c                | 17 -----------
 include/linux/mfd/ti-lmu-register.h | 44 -----------------------------
 include/linux/mfd/ti-lmu.h          |  1 -
 4 files changed, 1 insertion(+), 63 deletions(-)
drivers/mfd/Kconfig +1 -1
index 11841f4b7b2b..9b04dd527c68 100644
@@ -1293,7 +1293,7 @@ config MFD_TI_LMU
 	help
 	  Say yes here to enable support for TI LMU chips.
 
-	  TI LMU MFD supports LM3532, LM3631, LM3632, LM3633, LM3695 and LM3697.
+	  TI LMU MFD supports LM3532, LM3631, LM3632, LM3633, and LM3695.
 	  It consists of backlight, LED and regulator driver.
 	  It provides consistent device controls for lighting functions.
 
drivers/mfd/ti-lmu.c +0 -17
index cfb411cde51c..b6bfa99a29dd 100644
@@ -128,20 +128,6 @@ static struct mfd_cell lm3695_devices[] = {
 	},
 };
 
-static struct mfd_cell lm3697_devices[] = {
-	{
-		.name          = "ti-lmu-backlight",
-		.id            = LM3697,
-		.of_compatible = "ti,lm3697-backlight",
-	},
-	/* Monitoring driver for open/short circuit detection */
-	{
-		.name          = "ti-lmu-fault-monitor",
-		.id            = LM3697,
-		.of_compatible = "ti,lm3697-fault-monitor",
-	},
-};
-
 #define TI_LMU_DATA(chip, max_reg)		\
 static const struct ti_lmu_data chip##_data =	\
 {						\
@@ -155,7 +141,6 @@ TI_LMU_DATA(lm3631, LM3631_MAX_REG);
 TI_LMU_DATA(lm3632, LM3632_MAX_REG);
 TI_LMU_DATA(lm3633, LM3633_MAX_REG);
 TI_LMU_DATA(lm3695, LM3695_MAX_REG);
-TI_LMU_DATA(lm3697, LM3697_MAX_REG);
 
 static const struct of_device_id ti_lmu_of_match[] = {
 	{ .compatible = "ti,lm3532", .data = &lm3532_data },
@@ -163,7 +148,6 @@ static const struct of_device_id ti_lmu_of_match[] = {
 	{ .compatible = "ti,lm3632", .data = &lm3632_data },
 	{ .compatible = "ti,lm3633", .data = &lm3633_data },
 	{ .compatible = "ti,lm3695", .data = &lm3695_data },
-	{ .compatible = "ti,lm3697", .data = &lm3697_data },
 	{ }
 };
 MODULE_DEVICE_TABLE(of, ti_lmu_of_match);
@@ -237,7 +221,6 @@ static const struct i2c_device_id ti_lmu_ids[] = {
 	{ "lm3632", LM3632 },
 	{ "lm3633", LM3633 },
 	{ "lm3695", LM3695 },
-	{ "lm3697", LM3697 },
 	{ }
 };
 MODULE_DEVICE_TABLE(i2c, ti_lmu_ids);
include/linux/mfd/ti-lmu-register.h +0 -44

48 lines not on this page · raw message

include/linux/mfd/ti-lmu.h +0 -1
index 09d5f30384e5..bc9272f08f47 100644
@@ -26,7 +26,6 @@ enum ti_lmu_id {
 	LM3632,
 	LM3633,
 	LM3695,
-	LM3697,
 	LMU_MAX_ID,
 };
 
Signature · 1 line
2.17.0.1855.g63749b2dea
#19 lore raw

Re: [PATCH v7 2/6] mfd: ti-lmu: Remove support for LM3697

Quoted 6 lines · Dan Murphy

Remove support for the LM3697 from the ti-lmu
driver in favor of a dedicated LED driver.

---

Quoted 7 lines

v7 - New change for the series based on the comments in https://lore.kernel.org/patchwork/patch/982550/

 drivers/mfd/Kconfig                 |  2 +-
 drivers/mfd/ti-lmu.c                | 17 -----------
 include/linux/mfd/ti-lmu-register.h | 44 -----------------------------
 include/linux/mfd/ti-lmu.h          |  1 -
 4 files changed, 1 insertion(+), 63 deletions(-)

For my own reference:

Signature · 4 lines
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
author
#20 lore raw

PATCH v7 3/6 dt-bindings: leds: Add bindings for lm3697 driver

In reply to Dan Murphy

Add the device tree bindings for the lm3697
LED driver for backlighting and display.

---

v7 - Fix compatible from 3967 to 3697, and other comments. - https://lore.kernel.org/patchwork/patch/982550/

v6 - Fix minor issues - https://lore.kernel.org/patchwork/patch/975387/
v5 - Fix the comment for the example - https://lore.kernel.org/patchwork/patch/975060/
v4 - Removed HVLED definition in favor of HVLED place definition - https://lore.kernel.org/patchwork/patch/974812/
v3 - Updated subject with prefered title - https://lore.kernel.org/patchwork/patch/972337/
v2 - Fixed subject and patch commit message - https://lore.kernel.org/patchwork/patch/971326/

 .../devicetree/bindings/leds/leds-lm3697.txt  | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt
Documentation/devicetree/bindings/leds/leds-lm3697.txt new file +86 -0
new file mode 100644
index 000000000000..85ae075f6677
@@ -0,0 +1,86 @@
+* Texas Instruments - LM3697 Highly Efficient White LED Driver
+
+The LM3697 11-bit LED driver provides high-
+performance backlight dimming for 1, 2, or 3 series
+LED strings while delivering up to 90% efficiency.
+
+This device is suitable for display and keypad Lighting
+
+Required properties:
+	- compatible:
+		"ti,lm3697"
+	- reg :  I2C slave address
+	- #address-cells : 1
+	- #size-cells : 0
+
+Optional properties:
+	- enable-gpios : GPIO pin to enable/disable the device
+	- vled-supply : LED supply
+
+Required child properties:
+	- reg : 0 - LED is Controlled by bank A
+		1 - LED is Controlled by bank B
+	- led-sources : Indicates which HVLED string is associated to which
+			control bank.  Each element in the array is associated
+			with a specific HVLED string.  Element 0 is HVLED1,
+			element 1 is HVLED2 and element 2 HVLED3.
+			Additional information is contained
+			in Documentation/devicetree/bindings/leds/common.txt
+			0 - HVLED is not active in this control bank
+			1 - HVLED string is controlled by this control bank
+
+Optional child properties:
+	- label : see Documentation/devicetree/bindings/leds/common.txt
+	- linux,default-trigger :
+	   see Documentation/devicetree/bindings/leds/common.txt
+
+Example:
+
+HVLED string 1 and 3 are controlled by control bank A and HVLED 2 string is
+controlled by control bank B.
+
+led-controller@36 {
+	compatible = "ti,lm3697";
+	reg = <0x36>;
+	#address-cells = <1>;
+	#size-cells = <0>;
+
+	enable-gpios = <&gpio1 28 GPIO_ACTIVE_HIGH>;
+	vled-supply = <&vbatt>;
+
+	led@0 {
+		reg = <0>;
+		led-sources = <1 0 1>;
+		label = "white:first_backlight_cluster";
+		linux,default-trigger = "backlight";
+	};
+
+	led@1 {
+		reg = <1>;
+		led-sources = <0 1 0>;
+		label = "white:second_backlight_cluster";
+		linux,default-trigger = "backlight";
+	};
+}
+
+All HVLED strings controlled by control bank A
+
+led-controller@36 {
+	compatible = "ti,lm3697";
+	reg = <0x36>;
+	#address-cells = <1>;
+	#size-cells = <0>;
+
+	enable-gpios = <&gpio1 28 GPIO_ACTIVE_HIGH>;
+	vled-supply = <&vbatt>;
+
+	led@0 {
+		reg = <0>;
+		led-sources = <1 1 1>;
+		label = "white:backlight_cluster";
+		linux,default-trigger = "backlight";
+	};
+}
+
+For more product information please see the link below:
+http://www.ti.com/lit/ds/symlink/lm3697.pdf
Signature · 1 line
2.17.0.1855.g63749b2dea
maintainer Reviewed-by on v7 3/6
#21 lore raw

Re: [PATCH v7 3/6] dt-bindings: leds: Add bindings for lm3697 driver

Quoted 14 lines · Dan Murphy

Add the device tree bindings for the lm3697
LED driver for backlighting and display.

---

v7 - Fix compatible from 3967 to 3697, and other comments. - https://lore.kernel.org/patchwork/patch/982550/

v6 - Fix minor issues - https://lore.kernel.org/patchwork/patch/975387/
v5 - Fix the comment for the example - https://lore.kernel.org/patchwork/patch/975060/
v4 - Removed HVLED definition in favor of HVLED place definition - https://lore.kernel.org/patchwork/patch/974812/
v3 - Updated subject with prefered title - https://lore.kernel.org/patchwork/patch/972337/
v2 - Fixed subject and patch commit message - https://lore.kernel.org/patchwork/patch/971326/

Quoted 3 lines
 .../devicetree/bindings/leds/leds-lm3697.txt  | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt
#22 lore raw

Re: [PATCH v7 3/6] dt-bindings: leds: Add bindings for lm3697 driver

Hi!

Quoted 10 lines
Quoted 8 lines

Add the device tree bindings for the lm3697
LED driver for backlighting and display.

 .../devicetree/bindings/leds/leds-lm3697.txt  | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt

We already have binding for lm3697 in Documentation/devicetree/bindings/mfd/ti-lmu.txt . Is it good idea to have second one?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
maintainer
#23 lore raw

Re: [PATCH v7 3/6] dt-bindings: leds: Add bindings for lm3697 driver

Quoted 12 lines · Pavel Machek

Hi!

Quoted 9 lines
Quoted 8 lines

Add the device tree bindings for the lm3697
LED driver for backlighting and display.

 .../devicetree/bindings/leds/leds-lm3697.txt  | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt
Quoted 5 lines
Quoted 1 line

We already have binding for lm3697 in Documentation/devicetree/bindings/mfd/ti-lmu.txt . Is it good idea to have second one?

Of course not. Now that you mention it, I do remember seeing some discussion on this.

Rob

author
#24 lore raw

Re: [PATCH v7 3/6] dt-bindings: leds: Add bindings for lm3697 driver

Rob/Pavel/Jacek

Quoted 15 lines · Rob Herring
Quoted 14 lines · Pavel Machek

Hi!

Quoted 10 lines
Quoted 8 lines

Add the device tree bindings for the lm3697
LED driver for backlighting and display.

 .../devicetree/bindings/leds/leds-lm3697.txt  | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/leds/leds-lm3697.txt
Quoted 9 lines
Quoted 3 lines

We already have binding for lm3697 in Documentation/devicetree/bindings/mfd/ti-lmu.txt . Is it good idea to have second one?

Of course not. Now that you mention it, I do remember seeing some discussion on this.

Rob

I will be posting a RFC tomorrow with all the changes that shows 3 devices attached to the common code along with extending out the features on the LED drivers themselves.

Its not complete or tested code but it should give a basis for discussion. This way I can do some other code while this is being discussed.

Dan

Signature · 2 lines
------------------
Dan Murphy
author
#25 lore raw

PATCH v7 4/6 leds: lm3697: Introduce the lm3697 driver

In reply to Dan Murphy

Introduce the lm3697 LED driver for
backlighting and display.

Datasheet location:
http://www.ti.com/lit/ds/symlink/lm3697.pdf

---

v7 - No change - https://lore.kernel.org/patchwork/patch/982551/

v6 - Fix nitpicks - https://lore.kernel.org/patchwork/patch/975388/
v5 - Fix nitpick issues with code - https://lore.kernel.org/patchwork/patch/975061/
v4 - Made modifications to the control bank routine to accomodate the DT changes
https://lore.kernel.org/patchwork/patch/974811/
v3 - Add code to support led-sources and control bank configuration also fix other miscellaneous issues reported- https://lore.kernel.org/patchwork/patch/972336/
v2 - Removed unneed 'of' calls in dt_parse, fixed comment, fixed checkpatch error, and change led registration - https://lore.kernel.org/patchwork/patch/971327/

 drivers/leds/Kconfig       |   9 +
 drivers/leds/Makefile      |   1 +
 drivers/leds/leds-lm3697.c | 377 +++++++++++++++++++++++++++++++++++++
 3 files changed, 387 insertions(+)
 create mode 100644 drivers/leds/leds-lm3697.c
drivers/leds/Kconfig +9 -0
index 44097a3e0fcc..784cbe375724 100644
@@ -151,6 +151,15 @@ config LEDS_LM3642
 	  converter plus 1.5A constant current driver for a high-current
 	  white LED.
 
+config LEDS_LM3697
+	tristate "LED support for LM3697 Chip"
+	depends on LEDS_CLASS && I2C
+	select REGMAP_I2C
+	help
+	  This option enables support for LEDs connected to LM3697.
+	  The LM3697 is an 11 bit high performance backlight driver for
+	  keypad and display lighting.
+
 config LEDS_LM3692X
 	tristate "LED support for LM3692x Chips"
 	depends on LEDS_CLASS && I2C && OF
drivers/leds/Makefile +1 -0
index 420b5d2cfa62..2813f089f3db 100644
@@ -19,6 +19,7 @@ obj-$(CONFIG_LEDS_LOCOMO)		+= leds-locomo.o
 obj-$(CONFIG_LEDS_LM3530)		+= leds-lm3530.o
 obj-$(CONFIG_LEDS_LM3533)		+= leds-lm3533.o
 obj-$(CONFIG_LEDS_LM3642)		+= leds-lm3642.o
+obj-$(CONFIG_LEDS_LM3697)		+= leds-lm3697.o
 obj-$(CONFIG_LEDS_MIKROTIK_RB532)	+= leds-rb532.o
 obj-$(CONFIG_LEDS_S3C24XX)		+= leds-s3c24xx.o
 obj-$(CONFIG_LEDS_NET48XX)		+= leds-net48xx.o
drivers/leds/leds-lm3697.c new file +377 -0

377 lines not on this page · raw message

Signature · 1 line
2.17.0.1855.g63749b2dea
author
#26 lore raw

PATCH v7 5/6 dt-bindings: leds: Add runtime ramp node for LM3697

In reply to Dan Murphy

Add runtime ramp rates for the LM3697 LED driver. This LED driver supports separate runtime ramp rates for going from a higher or lower level of brightnesses

---

v7 - New change for the series to support feature in ti-lmu

 .../devicetree/bindings/leds/leds-lm3697.txt         | 12 ++++++++++++
 1 file changed, 12 insertions(+)
Documentation/devicetree/bindings/leds/leds-lm3697.txt +12 -0
index 85ae075f6677..4bb2ed51025b 100644
@@ -30,6 +30,12 @@ Required child properties:
 			1 - HVLED string is controlled by this control bank
 
 Optional child properties:
+	- runtime-ramp-up-msec: Current ramping from one brightness level to
+				the a higher brightness level.
+				Range from 2048 us - 117.44 s
+	- runtime-ramp-down-msec: Current ramping from one brightness level to
+				  the a lower brightness level.
+				  Range from 2048 us - 117.44 s
 	- label : see Documentation/devicetree/bindings/leds/common.txt
 	- linux,default-trigger :
 	   see Documentation/devicetree/bindings/leds/common.txt
@@ -51,6 +57,8 @@ led-controller@36 {
 	led@0 {
 		reg = <0>;
 		led-sources = <1 0 1>;
+		runtime-ramp-up-msec = <5000>;
+		runtime-ramp-down-msec = <1000>;
 		label = "white:first_backlight_cluster";
 		linux,default-trigger = "backlight";
 	};
@@ -58,6 +66,8 @@ led-controller@36 {
 	led@1 {
 		reg = <1>;
 		led-sources = <0 1 0>;
+		runtime-ramp-up-msec = <500>;
+		runtime-ramp-down-msec = <1000>;
 		label = "white:second_backlight_cluster";
 		linux,default-trigger = "backlight";
 	};
@@ -77,6 +87,8 @@ led-controller@36 {
 	led@0 {
 		reg = <0>;
 		led-sources = <1 1 1>;
+		runtime-ramp-up-msec = <500>;
+		runtime-ramp-down-msec = <1000>;
 		label = "white:backlight_cluster";
 		linux,default-trigger = "backlight";
 	};
Signature · 1 line
2.17.0.1855.g63749b2dea
author
#27 lore raw

PATCH v7 6/6 leds: lm3697: Add ramp rate feature

In reply to Dan Murphy

Add the runtime ramp up and down of the LEDs in the specific control bank.

Each control bank can have separate ramp up and ramp down values for the lighting zones the control banks manage.

---

v7 - New change for the series to support feature in ti-lmu

 drivers/leds/leds-lm3697.c | 72 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 72 insertions(+)
drivers/leds/leds-lm3697.c +72 -0
index 7416c545ec59..997c270a46b5 100644
@@ -59,6 +59,8 @@ struct lm3697_led {
 	struct led_classdev led_dev;
 	struct lm3697 *priv;
 	int control_bank;
+	int ramp_up_rate;
+	int ramp_down_rate;
 };
 
 /**
@@ -178,6 +180,62 @@ static int lm3697_set_control_bank(struct lm3697 *priv)
 	return ret;
 }
 
+static int lm3697_find_ramp_reg_val(int rate)
+{
+	const static int lookup[16] = { 2, 262, 524, 1049, 2090, 4194, 8389,
+					16780, 33550, 41940, 50330, 58720,
+					67110, 83880, 100660, 117440};
+	int i;
+
+	for (i = 1; i < ARRAY_SIZE(lookup); i++) {
+		if (rate == lookup[i])
+			return i;
+
+		if (rate > lookup[i - 1] && rate < lookup[i]) {
+			if (rate - lookup[i - 1] < lookup[i] - rate)
+				return i - 1;
+			else
+				return i;
+		}
+	}
+
+	return -EINVAL;
+}
+
+static int lm3697_set_ramp_rates(struct lm3697 *priv)
+{
+	u8 ramp, ramp_up, ramp_down;
+	struct lm3697_led *led;
+	u8 ramp_reg;
+	int i, ret = 0;
+
+	for (i = 0; i < 2; i++) {
+		led = &priv->leds[i];
+		if (led->ramp_up_rate == 0 && led->ramp_down_rate == 0)
+			continue;
+
+		if (led->control_bank == LM3697_CONTROL_A)
+			ramp_reg = LM3697_CTRL_A_RAMP;
+		else
+			ramp_reg = LM3697_CTRL_B_RAMP;
+
+		ramp_up = lm3697_find_ramp_reg_val(led->ramp_up_rate);
+		ramp_down = lm3697_find_ramp_reg_val(led->ramp_down_rate);
+
+		if (ramp_up < 0 || ramp_down < 0) {
+			dev_err(&priv->client->dev, "Cannot find ramp rate\n");
+			continue;
+		}
+
+		ramp = (ramp_up << 4) | ramp_down;
+		ret = regmap_write(priv->regmap, ramp_reg, ramp);
+		if (ret)
+			dev_err(&priv->client->dev, "Cannot write ramp config\n");
+	}
+
+	return ret;
+}
+
 static int lm3697_init(struct lm3697 *priv)
 {
 	int ret;
@@ -202,6 +260,9 @@ static int lm3697_init(struct lm3697 *priv)
 	if (ret)
 		dev_err(&priv->client->dev, "Setting the CRTL bank failed\n");
 
+	ret = lm3697_set_ramp_rates(priv);
+	if (ret)
+		dev_err(&priv->client->dev, "Setting the ramp rate failed\n");
 out:
 	return ret;
 }
@@ -254,6 +315,17 @@ static int lm3697_probe_dt(struct lm3697 *priv)
 			goto child_out;
 		}
 
+		ret = fwnode_property_read_u32(child, "runtime-ramp-up-msec",
+					       &led->ramp_up_rate);
+		if (ret)
+			dev_warn(&priv->client->dev, "runtime-ramp-up-msec property missing\n");
+
+		ret = fwnode_property_read_u32(child, "runtime-ramp-down-msec",
+					       &led->ramp_down_rate);
+		if (ret)
+			dev_warn(&priv->client->dev, "runtime-ramp-down-msec property missing\n");
+
+
 		fwnode_property_read_string(child, "linux,default-trigger",
 					    &led->led_dev.default_trigger);
 
Signature · 1 line
2.17.0.1855.g63749b2dea