2024-03-22 10:34:14

by Stephen Horvath

[permalink] [raw]
Subject: [PATCH 1/2] platform/chrome: cros_kbd_led_backlight: Automatically enable keyboard backlight control if feature present in EC

The keyboard backlight control is not enabled by default on all devices
with a Chromium EC, e.g. my Framework 13.
This patch adds a check for the feature and enables backlight control if
it is present.

I also fixed a bug that seemed like there was some confusion between
`cros_ec_dev` and `cros_ec_device`, so I'm not sure if this module even
worked until now.

Signed-off-by: Stephen Horvath <[email protected]>
---
drivers/mfd/cros_ec_dev.c | 9 +++++++
.../platform/chrome/cros_kbd_led_backlight.c | 24 +++++++------------
2 files changed, 17 insertions(+), 16 deletions(-)

diff --git a/drivers/mfd/cros_ec_dev.c b/drivers/mfd/cros_ec_dev.c
index a52d59cc2b1e..29b85ed5fece 100644
--- a/drivers/mfd/cros_ec_dev.c
+++ b/drivers/mfd/cros_ec_dev.c
@@ -99,6 +99,10 @@ static const struct mfd_cell cros_ec_wdt_cells[] = {
{ .name = "cros-ec-wdt", }
};

+static const struct mfd_cell cros_ec_pwm_kbd_cells[] = {
+ { .name = "chromeos-keyboard-leds", }
+};
+
static const struct cros_feature_to_cells cros_subdevices[] = {
{
.id = EC_FEATURE_CEC,
@@ -125,6 +129,11 @@ static const struct cros_feature_to_cells cros_subdevices[] = {
.mfd_cells = cros_ec_wdt_cells,
.num_cells = ARRAY_SIZE(cros_ec_wdt_cells),
},
+ {
+ .id = EC_FEATURE_PWM_KEYB,
+ .mfd_cells = cros_ec_pwm_kbd_cells,
+ .num_cells = ARRAY_SIZE(cros_ec_pwm_kbd_cells),
+ },
};

static const struct mfd_cell cros_ec_platform_cells[] = {
diff --git a/drivers/platform/chrome/cros_kbd_led_backlight.c b/drivers/platform/chrome/cros_kbd_led_backlight.c
index 793fd3f1015d..06e9a57536af 100644
--- a/drivers/platform/chrome/cros_kbd_led_backlight.c
+++ b/drivers/platform/chrome/cros_kbd_led_backlight.c
@@ -171,12 +171,15 @@ static int keyboard_led_init_ec_pwm(struct platform_device *pdev)
{
struct keyboard_led *keyboard_led = platform_get_drvdata(pdev);

- keyboard_led->ec = dev_get_drvdata(pdev->dev.parent);
- if (!keyboard_led->ec) {
+ struct cros_ec_dev *parent_ec = dev_get_drvdata(pdev->dev.parent);
+
+ if (!parent_ec) {
dev_err(&pdev->dev, "no parent EC device\n");
return -EINVAL;
}

+ keyboard_led->ec = parent_ec->ec_dev;
+
return 0;
}

@@ -200,8 +203,10 @@ static int keyboard_led_probe(struct platform_device *pdev)
int error;

drvdata = device_get_match_data(&pdev->dev);
- if (!drvdata)
- return -EINVAL;
+ if (!drvdata) {
+ /* Assume EC if no match data is provided */
+ drvdata = &keyboard_led_drvdata_ec_pwm;
+ }

keyboard_led = devm_kzalloc(&pdev->dev, sizeof(*keyboard_led), GFP_KERNEL);
if (!keyboard_led)
@@ -236,22 +241,10 @@ static const struct acpi_device_id keyboard_led_acpi_match[] = {
MODULE_DEVICE_TABLE(acpi, keyboard_led_acpi_match);
#endif

-#ifdef CONFIG_OF
-static const struct of_device_id keyboard_led_of_match[] = {
- {
- .compatible = "google,cros-kbd-led-backlight",
- .data = &keyboard_led_drvdata_ec_pwm,
- },
- {}
-};
-MODULE_DEVICE_TABLE(of, keyboard_led_of_match);
-#endif
-
static struct platform_driver keyboard_led_driver = {
.driver = {
.name = "chromeos-keyboard-leds",
.acpi_match_table = ACPI_PTR(keyboard_led_acpi_match),
- .of_match_table = of_match_ptr(keyboard_led_of_match),
},
.probe = keyboard_led_probe,
};
--
2.43.0



2024-03-25 12:25:16

by Tzung-Bi Shih

[permalink] [raw]
Subject: Re: [PATCH 1/2] platform/chrome: cros_kbd_led_backlight: Automatically enable keyboard backlight control if feature present in EC

On Fri, Mar 22, 2024 at 08:27:58PM +1000, Stephen Horvath wrote:
> I also fixed a bug that seemed like there was some confusion between
> `cros_ec_dev` and `cros_ec_device`, so I'm not sure if this module even
> worked until now.

If they aren't part of commit message, please put them after "---". Please
separate to an independent patch if they fix different things.

> diff --git a/drivers/platform/chrome/cros_kbd_led_backlight.c b/drivers/platform/chrome/cros_kbd_led_backlight.c
> index 793fd3f1015d..06e9a57536af 100644
> --- a/drivers/platform/chrome/cros_kbd_led_backlight.c
> +++ b/drivers/platform/chrome/cros_kbd_led_backlight.c
> @@ -171,12 +171,15 @@ static int keyboard_led_init_ec_pwm(struct platform_device *pdev)
> {
> struct keyboard_led *keyboard_led = platform_get_drvdata(pdev);
>
> - keyboard_led->ec = dev_get_drvdata(pdev->dev.parent);
> - if (!keyboard_led->ec) {
> + struct cros_ec_dev *parent_ec = dev_get_drvdata(pdev->dev.parent);

I'm not sure if it really fixes anything. See [1] or [2].

[1]: https://elixir.bootlin.com/linux/v6.8/source/drivers/platform/chrome/cros_ec_lpc.c#L417
[2]: https://elixir.bootlin.com/linux/v6.8/source/drivers/platform/chrome/cros_ec_spi.c#L759

> @@ -200,8 +203,10 @@ static int keyboard_led_probe(struct platform_device *pdev)
> int error;
>
> drvdata = device_get_match_data(&pdev->dev);
> - if (!drvdata)
> - return -EINVAL;
> + if (!drvdata) {
> + /* Assume EC if no match data is provided */
> + drvdata = &keyboard_led_drvdata_ec_pwm;
> + }

By current design, the `drvdata` should be provided by either ACPI or OF match.

> @@ -236,22 +241,10 @@ static const struct acpi_device_id keyboard_led_acpi_match[] = {
> MODULE_DEVICE_TABLE(acpi, keyboard_led_acpi_match);
> #endif
>
> -#ifdef CONFIG_OF
> -static const struct of_device_id keyboard_led_of_match[] = {
> - {
> - .compatible = "google,cros-kbd-led-backlight",
> - .data = &keyboard_led_drvdata_ec_pwm,
> - },
> - {}
> -};
> -MODULE_DEVICE_TABLE(of, keyboard_led_of_match);
> -#endif
> -
> static struct platform_driver keyboard_led_driver = {
> .driver = {
> .name = "chromeos-keyboard-leds",
> .acpi_match_table = ACPI_PTR(keyboard_led_acpi_match),
> - .of_match_table = of_match_ptr(keyboard_led_of_match),

The patch is ruining the use cases in OF world (e.g. [3]).

[3]: https://elixir.bootlin.com/linux/v6.8/source/arch/arm64/boot/dts/mediatek/mt8195-cherry.dtsi#L1154

2024-03-25 16:33:24

by Stephen Horvath

[permalink] [raw]
Subject: Re: [PATCH 1/2] platform/chrome: cros_kbd_led_backlight: Automatically enable keyboard backlight control if feature present in EC

On 25/3/24 13:24, Tzung-Bi Shih wrote:
> If they aren't part of commit message, please put them after "---". Please
> separate to an independent patch if they fix different things.

Alright thanks, the separation thing did kinda occur to me after I'd
send it, but I will do it next time.

> I'm not sure if it really fixes anything. See [1] or [2].

Ahh okay, I guess it was probably a side effect of me breaking something
else then, since the module only seemed to crash without that change.
But I was also mildly confused how it seemed to work fine like that
elsewhere though.

> By current design, the `drvdata` should be provided by either ACPI or OF match.

Yeah sorry, I didn't really know of a simpler way to do it, since it
gets started from `mfd_add_hotplug_devices` in cros_ec_dev.c in this
patch. I'll try and find a better way if I decide to make a v2.

> The patch is ruining the use cases in OF world (e.g. [3]).

I was kinda worried that the module would somehow launch twice, but I
have a feeling that wouldn't be possible anyway.

Thanks for your feedback!
Steve