Hi Jacek, Thanks for reverting. The setup seems fine because, when we express these as PCA955X_TYPE_GPIO and then define the LEDs under “compatibility=gpio-leds” section, then the LEDs do retain the states post BMC reboot with Andrew Jeffrey’s patch: https://github.com/torvalds/linux/commit/f5808ac158f2b16b686a3d3c0879c5d6048aba14 Problem is only when we express LEDs as PCA955X_TYPE_LED. With Andrew Jeffrey’s help now, I got some better understanding of load/unload. I then created this hack, but that does not do the trick as well. --- a/drivers/leds/led-class.c +++ b/drivers/leds/led-class.c @@ -442,9 +442,9 @@ void led_classdev_unregister(struct led_classdev *led_cdev) led_cdev->flags |= LED_UNREGISTERING; /* Stop blinking */ - led_stop_software_blink(led_cdev); + /* led_stop_software_blink(led_cdev); */ - led_set_brightness(led_cdev, LED_OFF); + /*led_set_brightness(led_cdev, LED_OFF); */ flush_work(&led_cdev->set_brightness_work); diff --git a/drivers/leds/led-core.c b/drivers/leds/led-core.c index f1f718d..833a16b 100644 --- a/drivers/leds/led-core.c +++ b/drivers/leds/led-core.c @@ -121,8 +121,8 @@ static void set_brightness_delayed(struct work_struct *ws) int ret = 0; if (test_and_clear_bit(LED_BLINK_DISABLE, &led_cdev->work_flags)) { - led_cdev->delayed_set_value = LED_OFF; - led_stop_software_blink(led_cdev); + /* led_cdev->delayed_set_value = LED_OFF; */ + /* led_stop_software_blink(led_cdev); */ } ret = __led_set_brightness(led_cdev, led_cdev->delayed_set_value); — Thank you, !! Vishwa !! > On 28-Apr-2021, at 2:52 AM, Jacek Anaszewski <jacek.anaszewski@xxxxxxxxx> wrote: > > Hi Vishwanatha, > > On 4/26/21 7:59 AM, vishwanatha subbanna wrote: >> Joel, >> With the experiments that I have done, I can not express LEDs with PCA955X_TYPE_LED predominantly because LEDs won’t >> retain states after the BMC reboot. I cooked a patch and tried but it does not work. I did an experiment where >> I put the patch and then did a reboot and saw that the LEDs were [OFF] in the very early stage of probe itself. >>> From a9fe9e956c624c15a455b88cc05262358519a541 Mon Sep 17 00:00:00 2001 >> From: Vishwanatha Subbanna <vishwa@xxxxxxxxxxxxxxxxxx> >> Date: Fri, 23 Apr 2021 06:57:56 -0500 >> Subject: [PATCH 1/2] leds: pca955x: Add support for default-state >> Signed-off-by: Vishwanatha Subbanna <vishwa@xxxxxxxxxxxxxxxxxx> >> --- >> drivers/leds/leds-pca955x.c | 12 ++++++++++-- >> 1 file changed, 10 insertions(+), 2 deletions(-) >> diff --git a/drivers/leds/leds-pca955x.c b/drivers/leds/leds-pca955x.c >> index bf7ead4..987415b 100644 >> --- a/drivers/leds/leds-pca955x.c >> +++ b/drivers/leds/leds-pca955x.c >> @@ -130,6 +130,7 @@ struct pca955x_led { >> char name[32]; >> u32 type; >> const char *default_trigger; >> + const char *default_state; >> }; >> struct pca955x_platform_data { >> @@ -408,6 +409,8 @@ static int pca955x_gpio_direction_output(struct gpio_chip *gc, >> fwnode_property_read_u32(child, "type", &pdata->leds[reg].type); >> fwnode_property_read_string(child, "linux,default-trigger", >> &pdata->leds[reg].default_trigger); >> + fwnode_property_read_string(child, "default-state", >> + &pdata->leds[reg].default_state); >> } >> pdata->num_leds = chip->bits; >> @@ -520,8 +523,13 @@ static int pca955x_probe(struct i2c_client *client, >> if (err) >> return err; >> - /* Turn off LED */ >> - err = pca955x_led_set(&pca955x_led->led_cdev, LED_OFF); >> + /* If the default-state is "keep", don't change states */ >> + if (strcmp(pdata->leds[i].default_state, "keep")) { >> + if (!strcmp(pdata->leds[i].default_state, "on")) >> + err = pca955x_led_set(&pca955x_led->led_cdev, LED_ON); >> + else >> + err = pca955x_led_set(&pca955x_led->led_cdev, LED_OFF); >> + } >> if (err) >> return err; >> } >> — >> 1.8.3.1 >> For `leds-gpio`, Andrew had put a patch, but I don’t see how that can be mapped to PCA955X. https://github.com/torvalds/linux/commit/f5808ac158f2b16b686a3d3c0879c5d6048aba14 >> Jacek, >> Please could you help me here ?.. I need to express LEDs as PCA955X_TYPE_LED and also retain states post BMC reboot. > > If in your setup the LED controller loses power on reboot then there > is nothing you can do to retain the state. > > -- > Best regards, > Jacek Anaszewski