mbox series

[v8,00/13] Adds support for PHY LEDs with offload triggers

Message ID 20230216013230.22978-1-ansuelsmth@gmail.com
Headers show
Series Adds support for PHY LEDs with offload triggers | expand

Message

Christian Marangi Feb. 16, 2023, 1:32 a.m. UTC
This is another attempt on adding this feature on LEDs, hoping this is
the right time and someone finally notice this.


Most of the times Switch/PHY have connected multiple LEDs that are
controlled by HW based on some rules/event. Currently we lack any
support for a generic way to control the HW part and normally we
either never implement the feature or only add control for brightness
or hw blink.

This is based on Marek idea of providing some API to cled but use a
different implementation that in theory should be more generilized.

The current idea is:
- LED driver implement 3 API (hw_control_status/start/stop).
  They are used to put the LED in hardware mode and to configure the
  various trigger.
- We have hardware triggers that are used to expose to userspace the
  supported hardware mode and set the hardware mode on trigger
  activation.
- We can also have triggers that both support hardware and software mode.
- The LED driver will declare each supported hardware blink mode and
  communicate with the trigger all the supported blink modes that will
  be available by sysfs.
- A trigger will use blink_set to configure the blink mode to active
  in hardware mode.
- On hardware trigger activation, only the hardware mode is enabled but
  the blink modes are not configured. The LED driver should reset any
  link mode active by default.

Each LED driver will have to declare explicit support for the offload
trigger (or return not supported error code) as we the trigger_data that
the LED driver will elaborate and understand what is referring to (based
on the current active trigger).

I posted a user for this new implementation that will benefit from this
and will add a big feature to it. Currently qca8k can have up to 3 LEDs
connected to each PHY port and we have some device that have only one of
them connected and the default configuration won't work for that.

The netdev trigger is expanded and it does now support hardware only
triggers.
The idea is to use hardware mode when a device_name is not defined.
An additional sysfs entry is added to give some info about the available
trigger modes supported in the current configuration.


It was reported that at least 3 other switch family would benefits by
this as they all lack support for a generic way to setup their leds and
netdev team NACK each try to add special code to support LEDs present
on switch in favor of a generic solution.

v8:
- Improve the documentation of the new feature
- Rename to a more symbolic name
- Fix some bug in netdev trigger (not using BIT())
- Add more define for qca8k-leds driver
- Add activity led mode
- Drop interval support
- Move qca8k brightness set to blocking variant (can sleep while
  setting the mode)
- More some function out of config define and provide variant if not
  selected
- Fix many bugs in the validate option in the netdev trigger
- Add phy generic schema for leds support
- Add additional required Documentation changes
v7:
- Rebase on top of net-next (for qca8k changes)
- Fix some typo in commit description
- Fix qca8k leds documentation warning
- Remove RFC tag
v6:
- Back to RFC.
- Drop additional trigger
- Rework netdev trigger to support common modes used by switch and
  hardware only triggers
- Refresh qca8k leds logic and driver
v5:
- Move out of RFC. (no comments from Andrew this is the right path?)
- Fix more spelling mistake (thx Randy)
- Fix error reported by kernel test bot
- Drop the additional HW_CONTROL flag. It does simplify CONFIG
  handling and hw control should be available anyway to support
  triggers as module.
v4:
- Rework implementation and drop hw_configure logic.
  We now expand blink_set.
- Address even more spelling mistake. (thx a lot Randy)
- Drop blink option and use blink_set delay.
- Rework phy-activity trigger to actually make the groups dynamic.
v3:
- Rework start/stop as Andrew asked.
- Introduce more logic to permit a trigger to run in hardware mode.
- Add additional patch with netdev hardware support.
- Use test_bit API to check flag passed to hw_control_configure.
- Added a new cmd to hw_control_configure to reset any active blink_mode.
- Refactor all the patches to follow this new implementation.
v2:
- Fix spelling mistake (sorry)
- Drop patch 02 "permit to declare supported offload triggers".
  Change the logic, now the LED driver declare support for them
  using the configure_offload with the cmd TRIGGER_SUPPORTED.
- Rework code to follow this new implementation.
- Update Documentation to better describe how this offload
  implementation work.

Christian Marangi (13):
  leds: add support for hardware driven LEDs
  leds: add function to configure hardware controlled LED
  leds: trigger: netdev: drop NETDEV_LED_MODE_LINKUP from mode
  leds: trigger: netdev: rename and expose NETDEV trigger enum modes
  leds: trigger: netdev: convert device attr to macro
  leds: trigger: netdev: add hardware control support
  leds: trigger: netdev: use mutex instead of spinlocks
  leds: trigger: netdev: add available mode sysfs attr
  leds: trigger: netdev: add additional hardware only triggers
  net: dsa: qca8k: add LEDs support
  dt-bindings: leds: Document netdev trigger
  dt-bindings: net: phy: Document support for leds node
  dt-bindings: net: dsa: qca8k: add LEDs definition example

 .../devicetree/bindings/leds/common.yaml      |   2 +
 .../devicetree/bindings/net/dsa/qca8k.yaml    |  24 +
 .../devicetree/bindings/net/ethernet-phy.yaml |  22 +
 Documentation/leds/leds-class.rst             |  94 ++++
 drivers/leds/Kconfig                          |  11 +
 drivers/leds/led-class.c                      |  27 ++
 drivers/leds/led-triggers.c                   |  38 ++
 drivers/leds/trigger/ledtrig-netdev.c         | 414 ++++++++++++-----
 drivers/net/dsa/qca/Kconfig                   |   9 +
 drivers/net/dsa/qca/Makefile                  |   1 +
 drivers/net/dsa/qca/qca8k-8xxx.c              |   4 +
 drivers/net/dsa/qca/qca8k-leds.c              | 419 ++++++++++++++++++
 drivers/net/dsa/qca/qca8k.h                   |  69 +++
 include/linux/leds.h                          |  95 +++-
 14 files changed, 1126 insertions(+), 103 deletions(-)
 create mode 100644 drivers/net/dsa/qca/qca8k-leds.c

Comments

Christian Marangi Feb. 17, 2023, 5:01 a.m. UTC | #1
On Fri, Feb 17, 2023 at 03:30:13PM +0100, Andrew Lunn wrote:
> On Thu, Feb 16, 2023 at 02:32:17AM +0100, Christian Marangi wrote:
> > This is another attempt on adding this feature on LEDs, hoping this is
> > the right time and someone finally notice this.
> 
> Hi Christian
> 
> Thanks for keeping working on this.
> 
> I want to review it, and maybe implement LED support in a PHY
> driver. But i'm busy with reworking EEE at the moment.
> 
> The merge window is about to open, so patches are not going to be
> accepted for the next two weeks. So i will take a look within that
> time and give you feedback.
> 

Sure take your time happy to discuss any improvement to this.
Christian Marangi Feb. 17, 2023, 5:58 a.m. UTC | #2
On Fri, Feb 17, 2023 at 05:03:46PM -0600, Rob Herring wrote:
> On Thu, Feb 16, 2023 at 02:32:28AM +0100, Christian Marangi wrote:
> > Document the netdev trigger that makes the LED blink or turn on based on
> > switch/phy events or an attached network interface.
> 
> NAK. What is netdev?

But netdev is a trigger, nothing new. Actually it was never documented.
Is the linux,default-trigger getting deprecated? 

> 
> Don't add new linux,default-trigger entries either. We have better ways 
> to define trigger sources, namely 'trigger-sources'.
> 
> > Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
> > ---
> >  Documentation/devicetree/bindings/leds/common.yaml | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/Documentation/devicetree/bindings/leds/common.yaml b/Documentation/devicetree/bindings/leds/common.yaml
> > index d34bb58c0037..6e016415a4d8 100644
> > --- a/Documentation/devicetree/bindings/leds/common.yaml
> > +++ b/Documentation/devicetree/bindings/leds/common.yaml
> > @@ -98,6 +98,8 @@ properties:
> >              # LED alters the brightness for the specified duration with one software
> >              # timer (requires "led-pattern" property)
> >            - pattern
> > +            # LED blink and turns on based on netdev events
> > +          - netdev
> >        - pattern: "^cpu[0-9]*$"
> >        - pattern: "^hci[0-9]+-power$"
> >          # LED is triggered by Bluetooth activity
> > -- 
> > 2.38.1
> >
Andrew Lunn Feb. 17, 2023, 2:30 p.m. UTC | #3
On Thu, Feb 16, 2023 at 02:32:17AM +0100, Christian Marangi wrote:
> This is another attempt on adding this feature on LEDs, hoping this is
> the right time and someone finally notice this.

Hi Christian

Thanks for keeping working on this.

I want to review it, and maybe implement LED support in a PHY
driver. But i'm busy with reworking EEE at the moment.

The merge window is about to open, so patches are not going to be
accepted for the next two weeks. So i will take a look within that
time and give you feedback.

     Andrew
Rob Herring (Arm) Feb. 17, 2023, 11:03 p.m. UTC | #4
On Thu, Feb 16, 2023 at 02:32:28AM +0100, Christian Marangi wrote:
> Document the netdev trigger that makes the LED blink or turn on based on
> switch/phy events or an attached network interface.

NAK. What is netdev?

Don't add new linux,default-trigger entries either. We have better ways 
to define trigger sources, namely 'trigger-sources'.

> Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
> ---
>  Documentation/devicetree/bindings/leds/common.yaml | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/leds/common.yaml b/Documentation/devicetree/bindings/leds/common.yaml
> index d34bb58c0037..6e016415a4d8 100644
> --- a/Documentation/devicetree/bindings/leds/common.yaml
> +++ b/Documentation/devicetree/bindings/leds/common.yaml
> @@ -98,6 +98,8 @@ properties:
>              # LED alters the brightness for the specified duration with one software
>              # timer (requires "led-pattern" property)
>            - pattern
> +            # LED blink and turns on based on netdev events
> +          - netdev
>        - pattern: "^cpu[0-9]*$"
>        - pattern: "^hci[0-9]+-power$"
>          # LED is triggered by Bluetooth activity
> -- 
> 2.38.1
>
Rob Herring (Arm) Feb. 21, 2023, 1:44 a.m. UTC | #5
On Fri, Feb 17, 2023 at 06:58:39AM +0100, Christian Marangi wrote:
> On Fri, Feb 17, 2023 at 05:03:46PM -0600, Rob Herring wrote:
> > On Thu, Feb 16, 2023 at 02:32:28AM +0100, Christian Marangi wrote:
> > > Document the netdev trigger that makes the LED blink or turn on based on
> > > switch/phy events or an attached network interface.
> > 
> > NAK. What is netdev?
> 
> But netdev is a trigger, nothing new. Actually it was never documented.

Okay, please state that in the commit message.

> Is the linux,default-trigger getting deprecated? 

Not quite, but it shouldn't be used for anything tied to a device IMO. 

Rob
Lee Jones Feb. 22, 2023, 3:02 p.m. UTC | #6
On Fri, 17 Feb 2023, Andrew Lunn wrote:

> On Thu, Feb 16, 2023 at 02:32:17AM +0100, Christian Marangi wrote:
> > This is another attempt on adding this feature on LEDs, hoping this is
> > the right time and someone finally notice this.
> 
> Hi Christian
> 
> Thanks for keeping working on this.
> 
> I want to review it, and maybe implement LED support in a PHY
> driver. But i'm busy with reworking EEE at the moment.
> 
> The merge window is about to open, so patches are not going to be
> accepted for the next two weeks. So i will take a look within that
> time and give you feedback.

Thanks Andrew.  If Pavel is still unavailable to conduct reviews, I'm
going to need all the help I can get with complex submissions such as
these.
Christian Marangi March 6, 2023, 6:43 p.m. UTC | #7
On Wed, Feb 22, 2023 at 03:02:04PM +0000, Lee Jones wrote:
> On Fri, 17 Feb 2023, Andrew Lunn wrote:
> 
> > On Thu, Feb 16, 2023 at 02:32:17AM +0100, Christian Marangi wrote:
> > > This is another attempt on adding this feature on LEDs, hoping this is
> > > the right time and someone finally notice this.
> > 
> > Hi Christian
> > 
> > Thanks for keeping working on this.
> > 
> > I want to review it, and maybe implement LED support in a PHY
> > driver. But i'm busy with reworking EEE at the moment.
> > 
> > The merge window is about to open, so patches are not going to be
> > accepted for the next two weeks. So i will take a look within that
> > time and give you feedback.
> 
> Thanks Andrew.  If Pavel is still unavailable to conduct reviews, I'm
> going to need all the help I can get with complex submissions such as
> these.
> 

Hi Lee,
thanks for stepping in. Just wanted to tell you I got some message with
Andrew to make this thing less problematic and to dry/make it more
review friendly.

We decided on pushing this in 3 step:
1. Propose most basic things for some switch and some PHY. (brightness
and blink_set support only, already supported by LED core)
2. A small series that should be just a cleanup for the netdev trigger
3. Support for hw_control in the most possible clean and way with small
patch to they are not hard to track and understand the concept of this
feature.

I'm starting with the step 1 and sending some of my patch and Andrew
patch to add basic support and I will add you and LED mailing list in
Cc.

Again thanks for starting checking this and feel free to ask any
question about this to me also privately, I'm very open to any help.
Lee Jones March 8, 2023, 2:06 p.m. UTC | #8
On Mon, 06 Mar 2023, Christian Marangi wrote:

> On Wed, Feb 22, 2023 at 03:02:04PM +0000, Lee Jones wrote:
> > On Fri, 17 Feb 2023, Andrew Lunn wrote:
> >
> > > On Thu, Feb 16, 2023 at 02:32:17AM +0100, Christian Marangi wrote:
> > > > This is another attempt on adding this feature on LEDs, hoping this is
> > > > the right time and someone finally notice this.
> > >
> > > Hi Christian
> > >
> > > Thanks for keeping working on this.
> > >
> > > I want to review it, and maybe implement LED support in a PHY
> > > driver. But i'm busy with reworking EEE at the moment.
> > >
> > > The merge window is about to open, so patches are not going to be
> > > accepted for the next two weeks. So i will take a look within that
> > > time and give you feedback.
> >
> > Thanks Andrew.  If Pavel is still unavailable to conduct reviews, I'm
> > going to need all the help I can get with complex submissions such as
> > these.
> >
>
> Hi Lee,
> thanks for stepping in. Just wanted to tell you I got some message with
> Andrew to make this thing less problematic and to dry/make it more
> review friendly.
>
> We decided on pushing this in 3 step:
> 1. Propose most basic things for some switch and some PHY. (brightness
> and blink_set support only, already supported by LED core)
> 2. A small series that should be just a cleanup for the netdev trigger
> 3. Support for hw_control in the most possible clean and way with small
> patch to they are not hard to track and understand the concept of this
> feature.
>
> I'm starting with the step 1 and sending some of my patch and Andrew
> patch to add basic support and I will add you and LED mailing list in
> Cc.

Sounds like a plan.  Thank you both.

> Again thanks for starting checking this and feel free to ask any
> question about this to me also privately, I'm very open to any help.

--
Lee Jones [李琼斯]
Linus Walleij March 9, 2023, 9:09 a.m. UTC | #9
Hi Christian,

thanks for your patch!

On Thu, Feb 16, 2023 at 2:36 AM Christian Marangi <ansuelsmth@gmail.com> wrote:

> The current idea is:
> - LED driver implement 3 API (hw_control_status/start/stop).
>   They are used to put the LED in hardware mode and to configure the
>   various trigger.
> - We have hardware triggers that are used to expose to userspace the
>   supported hardware mode and set the hardware mode on trigger
>   activation.
> - We can also have triggers that both support hardware and software mode.
> - The LED driver will declare each supported hardware blink mode and
>   communicate with the trigger all the supported blink modes that will
>   be available by sysfs.
> - A trigger will use blink_set to configure the blink mode to active
>   in hardware mode.
> - On hardware trigger activation, only the hardware mode is enabled but
>   the blink modes are not configured. The LED driver should reset any
>   link mode active by default.

The series looks good as a start.
There are some drivers and HW definitions etc for switch-controlled
LEDs, which is great.

I am a bit reluctant on the ambition to rely on configuration from sysfs
for the triggers, and I am also puzzled to how a certain trigger on a
certain LED is going to associate itself with, say, a certain port.

I want to draw your attention to this recently merged patch series
from Hans de Goede:
https://lore.kernel.org/linux-leds/20230120114524.408368-1-hdegoede@redhat.com/

This adds the devm_led_get() API which works similar to getting
regulators, clocks, GPIOs or any other resources.

It is not yet (I think) hooked into the device tree framework, but it
supports software nodes so adding DT handling should be sort of
trivial.

I think the ambition should be something like this (conjured example)
for a DSA switch:

    platform {
            switch {
                    compatible = "foo";

                    leds {
                            #address-cells = <1>;
                            #size-cells = <0>;
                            led0: led@0 {
                                    reg = <0>;
                                    color =...
                                    function = ...
                                    function-enumerator = ...
                                    default-state = ...
                            };
                            led1: led@1 {
                                    reg = <1>;
                                    color =...
                                    function = ...
                                    function-enumerator = ...
                                    default-state = ...
                            };
                    };

                    ports {
                            #address-cells = <1>;
                            #size-cells = <0>;
                            port@0 {
                                    reg = <0>;
                                    label = "lan0";
                                    phy-handle = <&phy0>;
                                    leds = <&led0>;
                            };
                            port@1 {
                                    reg = <1>;
                                    label = "lan1";
                                    phy-handle = <&phy1>;
                                    leds = <&led0>;
                            };
                    };

                    mdio {
                            compatible = "foo-mdio";
                            #address-cells = <1>;
                            #size-cells = <0>;

                            phy0: ethernet-phy@0 {
                                    reg = <0>;
                            };
                            phy1: ethernet-phy@1 {
                                    reg = <1>;
                            };
                    };
            };
    };

I am not the man to tell whether the leds = <&led0>; phandle should be on
the port or actually on the phy, it may even vary. You guys know the answer
to this.

But certainly something like this resource phandle will be necessary to
assign the right LED to the right port or phy, I hope you were not going
to rely on strings and naming conventions?

Yours,
Linus Walleij
Hans de Goede March 9, 2023, 9:32 a.m. UTC | #10
Hi,

On 3/9/23 10:09, Linus Walleij wrote:
> Hi Christian,
> 
> thanks for your patch!
> 
> On Thu, Feb 16, 2023 at 2:36 AM Christian Marangi <ansuelsmth@gmail.com> wrote:
> 
>> The current idea is:
>> - LED driver implement 3 API (hw_control_status/start/stop).
>>   They are used to put the LED in hardware mode and to configure the
>>   various trigger.
>> - We have hardware triggers that are used to expose to userspace the
>>   supported hardware mode and set the hardware mode on trigger
>>   activation.
>> - We can also have triggers that both support hardware and software mode.
>> - The LED driver will declare each supported hardware blink mode and
>>   communicate with the trigger all the supported blink modes that will
>>   be available by sysfs.
>> - A trigger will use blink_set to configure the blink mode to active
>>   in hardware mode.
>> - On hardware trigger activation, only the hardware mode is enabled but
>>   the blink modes are not configured. The LED driver should reset any
>>   link mode active by default.
> 
> The series looks good as a start.
> There are some drivers and HW definitions etc for switch-controlled
> LEDs, which is great.
> 
> I am a bit reluctant on the ambition to rely on configuration from sysfs
> for the triggers, and I am also puzzled to how a certain trigger on a
> certain LED is going to associate itself with, say, a certain port.
> 
> I want to draw your attention to this recently merged patch series
> from Hans de Goede:
> https://lore.kernel.org/linux-leds/20230120114524.408368-1-hdegoede@redhat.com/
> 
> This adds the devm_led_get() API which works similar to getting
> regulators, clocks, GPIOs or any other resources.
> 
> It is not yet (I think) hooked into the device tree framework, but it
> supports software nodes so adding DT handling should be sort of
> trivial.

That series contains this (unmerged) patch to hookup DT handling:

https://lore.kernel.org/linux-leds/20230120114524.408368-6-hdegoede@redhat.com/

this was not merged because there are no current users, but adding
support is as easy as picking up that patch :)

Note there also already is a devicetree *only*:

struct led_classdev *of_led_get(struct device_node *np, int index);

Since I was working on a x86/ACPI platform I needed something more
generic though and ideally new code would use the generic approach.

Regards,

Hans





> 
> I think the ambition should be something like this (conjured example)
> for a DSA switch:
> 
>     platform {
>             switch {
>                     compatible = "foo";
> 
>                     leds {
>                             #address-cells = <1>;
>                             #size-cells = <0>;
>                             led0: led@0 {
>                                     reg = <0>;
>                                     color =...
>                                     function = ...
>                                     function-enumerator = ...
>                                     default-state = ...
>                             };
>                             led1: led@1 {
>                                     reg = <1>;
>                                     color =...
>                                     function = ...
>                                     function-enumerator = ...
>                                     default-state = ...
>                             };
>                     };
> 
>                     ports {
>                             #address-cells = <1>;
>                             #size-cells = <0>;
>                             port@0 {
>                                     reg = <0>;
>                                     label = "lan0";
>                                     phy-handle = <&phy0>;
>                                     leds = <&led0>;
>                             };
>                             port@1 {
>                                     reg = <1>;
>                                     label = "lan1";
>                                     phy-handle = <&phy1>;
>                                     leds = <&led0>;
>                             };
>                     };
> 
>                     mdio {
>                             compatible = "foo-mdio";
>                             #address-cells = <1>;
>                             #size-cells = <0>;
> 
>                             phy0: ethernet-phy@0 {
>                                     reg = <0>;
>                             };
>                             phy1: ethernet-phy@1 {
>                                     reg = <1>;
>                             };
>                     };
>             };
>     };
> 
> I am not the man to tell whether the leds = <&led0>; phandle should be on
> the port or actually on the phy, it may even vary. You guys know the answer
> to this.
> 
> But certainly something like this resource phandle will be necessary to
> assign the right LED to the right port or phy, I hope you were not going
> to rely on strings and naming conventions?
> 
> Yours,
> Linus Walleij
>
Andrew Lunn March 9, 2023, 3:22 p.m. UTC | #11
> I am a bit reluctant on the ambition to rely on configuration from sysfs
> for the triggers, and I am also puzzled to how a certain trigger on a
> certain LED is going to associate itself with, say, a certain port.

Hi Linus

There will need to be a device tree binding for the default
trigger. That is what nearly all the rejected hacks so far have been
about, write registers in the PHYs LEDs registers to put it into a
specific mode. I don't see that part of the overall problem too
problematic, apart from the old issue, is it describing configuration,
not hardware.

As to 'how a certain trigger on a certain LED is going to associate
itself with, say, a certain port' is clearly a property of the
hardware, when offloading is supported. I've not seen a switch you can
arbitrarily assign LEDs to ports. The Marvell switches have the LED
registers within the port registers, for example, two LEDs per port.

> 
> I want to draw your attention to this recently merged patch series
> from Hans de Goede:
> https://lore.kernel.org/linux-leds/20230120114524.408368-1-hdegoede@redhat.com/
> 
> This adds the devm_led_get() API which works similar to getting
> regulators, clocks, GPIOs or any other resources.

Interesting. Thanks for pointing this out. But i don't think it is of
interest in our use case, which is hardware offload. For a purely
software controlled LED, where the LED could be anywhere,
devm_led_get() makes sense. But in our case, the LED is in a well
defined place, either the MAC or the PHY, where it has access to the
RX and TX packets, link status etc. So we don't have the problem of
finding it in an arbitrary location.

There is also one additional problem we have, both for MAC and PHY
LEDs. The trigger is ledtrig-netdev. All trigger state revolves around
a netdev. A DSA port is not a netdev. A PHY is not a netdev. Each of
these three things have there own life cycle. Often, a PHY does not
know what netdev it is associated to until the interface is opened,
ie. ip link set eth0 up. But once it is associated, phylib knows this
information, so can return it, without any additional configuration
data in DT. A DSA switch port can be created before the netdev
associated to it is created. But again, the DSA core does know the
mapping between a netdev and a port. Using devm_led_get() does not
gain us anything.

	Andrew
Linus Walleij March 10, 2023, 9:37 a.m. UTC | #12
On Thu, Mar 9, 2023 at 4:22 PM Andrew Lunn <andrew@lunn.ch> wrote:

> As to 'how a certain trigger on a certain LED is going to associate
> itself with, say, a certain port' is clearly a property of the
> hardware, when offloading is supported. I've not seen a switch you can
> arbitrarily assign LEDs to ports. The Marvell switches have the LED
> registers within the port registers, for example, two LEDs per port.

Aha so there is an implicit HW dependency between the port and the
LED, that we just cannot see in the device tree. Okay, it makes sense.

I think there will be a day when a switch without LED controller appears,
but the system has a few LEDs for the ports connected to an
arbitrary GPIO controller, and then we will need this. But we have
not seen that yet :)

Yours,
Linus Walleij
Andrew Lunn March 10, 2023, 3:15 p.m. UTC | #13
On Fri, Mar 10, 2023 at 10:37:25AM +0100, Linus Walleij wrote:
> On Thu, Mar 9, 2023 at 4:22 PM Andrew Lunn <andrew@lunn.ch> wrote:
> 
> > As to 'how a certain trigger on a certain LED is going to associate
> > itself with, say, a certain port' is clearly a property of the
> > hardware, when offloading is supported. I've not seen a switch you can
> > arbitrarily assign LEDs to ports. The Marvell switches have the LED
> > registers within the port registers, for example, two LEDs per port.
> 
> Aha so there is an implicit HW dependency between the port and the
> LED, that we just cannot see in the device tree. Okay, it makes sense.

Well, i would say the dependency is in the device tree, in that the
LEDs are described in the ports, not as a block of their own at a
higher level within the switch. And in some switches, they might
actually be a block of registers in there own space, rather than in
the port registers. But i still expect there is a fixed mapping
between LED and port.

> I think there will be a day when a switch without LED controller appears,
> but the system has a few LEDs for the ports connected to an
> arbitrary GPIO controller, and then we will need this. But we have
> not seen that yet :)

The microchip sparx5 might be going in that direction. It has what
looks like a reasonably generic sgpio controller:
drivers/pinctrl/pinctrl-microchip-sgpio.c

But this not just about switches. It is also plain NICs. And using
ledtrig-netdev, you could make your keyboard LED blink based on
network traffic etc. So yes, using a phandle to an LED could very well
be useful in the future.

   Andrew
Michael Walle March 13, 2023, 9:28 a.m. UTC | #14
> > I think there will be a day when a switch without LED controller appears,
> > but the system has a few LEDs for the ports connected to an
> > arbitrary GPIO controller, and then we will need this. But we have
> > not seen that yet :)
> 
> The microchip sparx5 might be going in that direction. It has what
> looks like a reasonably generic sgpio controller:
> drivers/pinctrl/pinctrl-microchip-sgpio.c

That gpio controller supports both, some kind of hardware controlled
and pure software controlled mode. AFAIK the driver only supports
software controlled mode (yet?). In any case, our board
(arch/arm/boot/dts/lan966x-kontron-kswitch-d10-mmt.dtsi) is broken in
a way that we are forced to use the software controlled mode anyway.
Therefore, there is already such a board ;)

-michael