On some newer laptops it appears that an AMD driver can register a
platform_profile handler. If this happens then the asus_wmi driver would
error with -EEXIST when trying to register its own handler leaving the
user with a possibly unusable system - this is especially true for
laptops with an MCU that emit a stream of HID packets, some of which can
be misinterpreted as shutdown signals.

We can safely continue loading the driver instead of bombing out.

Signed-off-by: Luke D. Jones <luke@ljones.dev>
---
 drivers/platform/x86/asus-wmi.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/platform/x86/asus-wmi.c b/drivers/platform/x86/asus-wmi.c
index cc5dc296fb2d..2fdfa84f7efb 100644
--- a/drivers/platform/x86/asus-wmi.c
+++ b/drivers/platform/x86/asus-wmi.c
@@ -3897,8 +3897,13 @@ static int platform_profile_setup(struct asus_wmi *asus)
 		asus->platform_profile_handler.choices);
 
 	err = platform_profile_register(&asus->platform_profile_handler);
-	if (err)
+	if (err == -EEXIST) {
+		pr_warn("A platform_profile handler is already registered, this may be a driver conflict.\n");
+		return 0;
+	} else if (err) {
+		pr_err("Failed to register platform_profile: %d\n", err);
 		return err;
+	}
 
 	asus->platform_profile_support = true;
 	return 0;
@@ -4773,7 +4778,7 @@ static int asus_wmi_add(struct platform_device *pdev)
 		goto fail_fan_boost_mode;
 
 	err = platform_profile_setup(asus);
-	if (err)
+	if (err && err != -EEXIST)
 		goto fail_platform_profile_setup;
 
 	err = asus_wmi_sysfs_init(asus->platform_device);
-- 
2.45.2

Hi Luke,

On 8/6/24 1:58 AM, Luke D. Jones wrote:
> On some newer laptops it appears that an AMD driver can register a
> platform_profile handler. If this happens then the asus_wmi driver would
> error with -EEXIST when trying to register its own handler leaving the
> user with a possibly unusable system - this is especially true for
> laptops with an MCU that emit a stream of HID packets, some of which can
> be misinterpreted as shutdown signals.
> 
> We can safely continue loading the driver instead of bombing out.
> 
> Signed-off-by: Luke D. Jones <luke@ljones.dev>
> ---
>  drivers/platform/x86/asus-wmi.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/platform/x86/asus-wmi.c b/drivers/platform/x86/asus-wmi.c
> index cc5dc296fb2d..2fdfa84f7efb 100644
> --- a/drivers/platform/x86/asus-wmi.c
> +++ b/drivers/platform/x86/asus-wmi.c
> @@ -3897,8 +3897,13 @@ static int platform_profile_setup(struct asus_wmi *asus)
>  		asus->platform_profile_handler.choices);
>  
>  	err = platform_profile_register(&asus->platform_profile_handler);
> -	if (err)
> +	if (err == -EEXIST) {
> +		pr_warn("A platform_profile handler is already registered, this may be a driver conflict.\n");
> +		return 0;
> +	} else if (err) {
> +		pr_err("Failed to register platform_profile: %d\n", err);
>  		return err;
> +	}
>  
>  	asus->platform_profile_support = true;
>  	return 0;
> @@ -4773,7 +4778,7 @@ static int asus_wmi_add(struct platform_device *pdev)
>  		goto fail_fan_boost_mode;
>  
>  	err = platform_profile_setup(asus);
> -	if (err)
> +	if (err && err != -EEXIST)
>  		goto fail_platform_profile_setup;

Since you already do "return 0" on EEXIST in platform_profile_setup() there
is no need for this part of the patch.

Regards,

Hans



>  
>  	err = asus_wmi_sysfs_init(asus->platform_device);