[PATCH 1/3] platform/x86: msi-laptop: Fix old-ec check for backlight registering

[PATCH 1/3] platform/x86: msi-laptop: Fix old-ec check for backlight registering

[Hans de Goede]

Commit 2cc6c717799f ("msi-laptop: Port to new backlight interface
selection API") replaced this check:

	if (!quirks->old_ec_model || acpi_video_backlight_support())
		pr_info("Brightness ignored, ...");
	else
		do_register();

With:

	if (quirks->old_ec_model ||
	    acpi_video_get_backlight_type() == acpi_backlight_vendor)
		do_register();

But since the do_register() part was part of the else branch, the entire
condition should be inverted.  So not only the 2 statements on either
side of the || should be inverted, but the || itself should be replaced
with a &&.

In practice this has likely not been an issue because the new-ec models
(old_ec_model==false) likely all support ACPI video backlight control,
making acpi_video_get_backlight_type() return acpi_backlight_video
turning the second part of the || also false when old_ec_model == false.

Fixes: 2cc6c717799f ("msi-laptop: Port to new backlight interface selection API")
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/platform/x86/msi-laptop.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/platform/x86/msi-laptop.c b/drivers/platform/x86/msi-laptop.c
index 93ef8851b93e..54170172a666 100644
--- a/drivers/platform/x86/msi-laptop.c
+++ b/drivers/platform/x86/msi-laptop.c
@@ -1047,8 +1047,7 @@ static int __init msi_init(void)
 		return -EINVAL;
 
 	/* Register backlight stuff */
-
-	if (quirks->old_ec_model ||
+	if (quirks->old_ec_model &&
 	    acpi_video_get_backlight_type() == acpi_backlight_vendor) {
 		struct backlight_properties props;
 		memset(&props, 0, sizeof(struct backlight_properties));
-- 
2.37.2

[PATCH 2/3] platform/x86: msi-laptop: Simplify ec_delay handling

[Hans de Goede]

There is no reason to have both non-delayed and delayed work structs
for the rfkill and touchpad work.

Instead simply call schedule_delayed_work() with a delay of 0 for
the quirks->ec_delay == fase case.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/platform/x86/msi-laptop.c | 31 +++++++++++--------------------
 1 file changed, 11 insertions(+), 20 deletions(-)

diff --git a/drivers/platform/x86/msi-laptop.c b/drivers/platform/x86/msi-laptop.c
index 54170172a666..65db18c6e3e8 100644
--- a/drivers/platform/x86/msi-laptop.c
+++ b/drivers/platform/x86/msi-laptop.c
@@ -590,6 +590,14 @@ static int dmi_check_cb(const struct dmi_system_id *dmi)
 	return 1;
 }
 
+static unsigned long msi_work_delay(int msecs)
+{
+	if (quirks->ec_delay)
+		return msecs_to_jiffies(msecs);
+
+	return 0;
+}
+
 static const struct dmi_system_id msi_dmi_table[] __initconst = {
 	{
 		.ident = "MSI S270",
@@ -784,7 +792,6 @@ static void msi_update_rfkill(struct work_struct *ignored)
 		msi_rfkill_set_state(rfk_threeg, !threeg_s);
 }
 static DECLARE_DELAYED_WORK(msi_rfkill_dwork, msi_update_rfkill);
-static DECLARE_WORK(msi_rfkill_work, msi_update_rfkill);
 
 static void msi_send_touchpad_key(struct work_struct *ignored)
 {
@@ -800,7 +807,6 @@ static void msi_send_touchpad_key(struct work_struct *ignored)
 		KEY_TOUCHPAD_ON : KEY_TOUCHPAD_OFF, 1, true);
 }
 static DECLARE_DELAYED_WORK(msi_touchpad_dwork, msi_send_touchpad_key);
-static DECLARE_WORK(msi_touchpad_work, msi_send_touchpad_key);
 
 static bool msi_laptop_i8042_filter(unsigned char data, unsigned char str,
 				struct serio *port)
@@ -818,20 +824,12 @@ static bool msi_laptop_i8042_filter(unsigned char data, unsigned char str,
 		extended = false;
 		switch (data) {
 		case 0xE4:
-			if (quirks->ec_delay) {
-				schedule_delayed_work(&msi_touchpad_dwork,
-					round_jiffies_relative(0.5 * HZ));
-			} else
-				schedule_work(&msi_touchpad_work);
+			schedule_delayed_work(&msi_touchpad_dwork, msi_work_delay(500));
 			break;
 		case 0x54:
 		case 0x62:
 		case 0x76:
-			if (quirks->ec_delay) {
-				schedule_delayed_work(&msi_rfkill_dwork,
-					round_jiffies_relative(0.5 * HZ));
-			} else
-				schedule_work(&msi_rfkill_work);
+			schedule_delayed_work(&msi_rfkill_dwork, msi_work_delay(500));
 			break;
 		}
 	}
@@ -898,12 +896,7 @@ static int rfkill_init(struct platform_device *sdev)
 	}
 
 	/* schedule to run rfkill state initial */
-	if (quirks->ec_delay) {
-		schedule_delayed_work(&msi_rfkill_init,
-			round_jiffies_relative(1 * HZ));
-	} else
-		schedule_work(&msi_rfkill_work);
-
+	schedule_delayed_work(&msi_rfkill_init, msi_work_delay(1000));
 	return 0;
 
 err_threeg:
@@ -1114,7 +1107,6 @@ static int __init msi_init(void)
 	if (quirks->load_scm_model) {
 		i8042_remove_filter(msi_laptop_i8042_filter);
 		cancel_delayed_work_sync(&msi_rfkill_dwork);
-		cancel_work_sync(&msi_rfkill_work);
 		rfkill_cleanup();
 	}
 fail_scm_model_init:
@@ -1135,7 +1127,6 @@ static void __exit msi_cleanup(void)
 		i8042_remove_filter(msi_laptop_i8042_filter);
 		input_unregister_device(msi_laptop_input_dev);
 		cancel_delayed_work_sync(&msi_rfkill_dwork);
-		cancel_work_sync(&msi_rfkill_work);
 		rfkill_cleanup();
 	}
 
-- 
2.37.2

Re: [PATCH 2/3] platform/x86: msi-laptop: Simplify ec_delay handling

[Andy Shevchenko]

On Thu, Aug 25, 2022 at 04:13:35PM +0200, Hans de Goede wrote:
> There is no reason to have both non-delayed and delayed work structs
> for the rfkill and touchpad work.
> 
> Instead simply call schedule_delayed_work() with a delay of 0 for
> the quirks->ec_delay == fase case.

fase ?!

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 2/3] platform/x86: msi-laptop: Simplify ec_delay handling

[Hans de Goede]

Hi,

On 8/25/22 17:05, Andy Shevchenko wrote:
> On Thu, Aug 25, 2022 at 04:13:35PM +0200, Hans de Goede wrote:
>> There is no reason to have both non-delayed and delayed work structs
>> for the rfkill and touchpad work.
>>
>> Instead simply call schedule_delayed_work() with a delay of 0 for
>> the quirks->ec_delay == fase case.
> 
> fase ?!

That should be "false", my bad.

Regards,

Hans

[PATCH 3/3] platform/x86: msi-laptop: Fix resource cleanup

[Hans de Goede]

Fix the input-device not getting free-ed on probe-errors and
fix the msi_touchpad_dwork not getting cancelled on neither
probe-errors nor on remove.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/platform/x86/msi-laptop.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/platform/x86/msi-laptop.c b/drivers/platform/x86/msi-laptop.c
index 65db18c6e3e8..5d4b10b8d771 100644
--- a/drivers/platform/x86/msi-laptop.c
+++ b/drivers/platform/x86/msi-laptop.c
@@ -1106,6 +1106,8 @@ static int __init msi_init(void)
 fail_create_group:
 	if (quirks->load_scm_model) {
 		i8042_remove_filter(msi_laptop_i8042_filter);
+		cancel_delayed_work_sync(&msi_touchpad_dwork);
+		input_unregister_device(msi_laptop_input_dev);
 		cancel_delayed_work_sync(&msi_rfkill_dwork);
 		rfkill_cleanup();
 	}
@@ -1125,6 +1127,7 @@ static void __exit msi_cleanup(void)
 {
 	if (quirks->load_scm_model) {
 		i8042_remove_filter(msi_laptop_i8042_filter);
+		cancel_delayed_work_sync(&msi_touchpad_dwork);
 		input_unregister_device(msi_laptop_input_dev);
 		cancel_delayed_work_sync(&msi_rfkill_dwork);
 		rfkill_cleanup();
-- 
2.37.2

Re: [PATCH 3/3] platform/x86: msi-laptop: Fix resource cleanup

[Andy Shevchenko]

On Thu, Aug 25, 2022 at 04:13:36PM +0200, Hans de Goede wrote:
> Fix the input-device not getting free-ed on probe-errors and
> fix the msi_touchpad_dwork not getting cancelled on neither
> probe-errors nor on remove.

Sounds like a fix, shouldn't have a Fixes tag?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 3/3] platform/x86: msi-laptop: Fix resource cleanup

[Hans de Goede]

Hi,

On 8/25/22 17:05, Andy Shevchenko wrote:
> On Thu, Aug 25, 2022 at 04:13:36PM +0200, Hans de Goede wrote:
>> Fix the input-device not getting free-ed on probe-errors and
>> fix the msi_touchpad_dwork not getting cancelled on neither
>> probe-errors nor on remove.
> 
> Sounds like a fix, shouldn't have a Fixes tag?

You right, I'll add a:

Fixes: 143a4c0284dc ("msi-laptop: send out touchpad on/off key")

tag before merging.

Regards,

Hans

Re: [PATCH 1/3] platform/x86: msi-laptop: Fix old-ec check for backlight registering

[Andy Shevchenko]

On Thu, Aug 25, 2022 at 04:13:34PM +0200, Hans de Goede wrote:
> Commit 2cc6c717799f ("msi-laptop: Port to new backlight interface
> selection API") replaced this check:
> 
> 	if (!quirks->old_ec_model || acpi_video_backlight_support())
> 		pr_info("Brightness ignored, ...");
> 	else
> 		do_register();
> 
> With:
> 
> 	if (quirks->old_ec_model ||
> 	    acpi_video_get_backlight_type() == acpi_backlight_vendor)
> 		do_register();
> 
> But since the do_register() part was part of the else branch, the entire
> condition should be inverted.  So not only the 2 statements on either
> side of the || should be inverted, but the || itself should be replaced
> with a &&.
> 
> In practice this has likely not been an issue because the new-ec models
> (old_ec_model==false) likely all support ACPI video backlight control,
> making acpi_video_get_backlight_type() return acpi_backlight_video
> turning the second part of the || also false when old_ec_model == false.

...

>  	/* Register backlight stuff */
> -
> -	if (quirks->old_ec_model ||
> +	if (quirks->old_ec_model &&
>  	    acpi_video_get_backlight_type() == acpi_backlight_vendor) {
>  		struct backlight_properties props;

checkpatch might complain for absence of blank line here, btw. Perhaps
you may just move the above one here at the same patch.

>  		memset(&props, 0, sizeof(struct backlight_properties));

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 1/3] platform/x86: msi-laptop: Fix old-ec check for backlight registering

[Hans de Goede]

Hi,

On 8/25/22 16:13, Hans de Goede wrote:
> Commit 2cc6c717799f ("msi-laptop: Port to new backlight interface
> selection API") replaced this check:
> 
> 	if (!quirks->old_ec_model || acpi_video_backlight_support())
> 		pr_info("Brightness ignored, ...");
> 	else
> 		do_register();
> 
> With:
> 
> 	if (quirks->old_ec_model ||
> 	    acpi_video_get_backlight_type() == acpi_backlight_vendor)
> 		do_register();
> 
> But since the do_register() part was part of the else branch, the entire
> condition should be inverted.  So not only the 2 statements on either
> side of the || should be inverted, but the || itself should be replaced
> with a &&.
> 
> In practice this has likely not been an issue because the new-ec models
> (old_ec_model==false) likely all support ACPI video backlight control,
> making acpi_video_get_backlight_type() return acpi_backlight_video
> turning the second part of the || also false when old_ec_model == false.
> 
> Fixes: 2cc6c717799f ("msi-laptop: Port to new backlight interface selection API")
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>

I've added this series to my review-hans (soon to be for-next) branch now.

Regards,

Hans

> ---
>  drivers/platform/x86/msi-laptop.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/platform/x86/msi-laptop.c b/drivers/platform/x86/msi-laptop.c
> index 93ef8851b93e..54170172a666 100644
> --- a/drivers/platform/x86/msi-laptop.c
> +++ b/drivers/platform/x86/msi-laptop.c
> @@ -1047,8 +1047,7 @@ static int __init msi_init(void)
>  		return -EINVAL;
>  
>  	/* Register backlight stuff */
> -
> -	if (quirks->old_ec_model ||
> +	if (quirks->old_ec_model &&
>  	    acpi_video_get_backlight_type() == acpi_backlight_vendor) {
>  		struct backlight_properties props;
>  		memset(&props, 0, sizeof(struct backlight_properties));