[PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Hans de Goede]

acpi_bus_find_device_by_name() call returns a pointer to the device object
on the ACPI bus, aka the ACPI companion device.

gpio_secondary_fwnode_init() then continues with setting the secondary
fwnode on this device. But this is not the actual physical device for
the GPIO controller (e.g. the GPIO controller platform bus device).

This mismatch is causing GPIO lookups by secondary fwnode to not work.

Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
of the first physical device associated with the ACPI companion device.

This fixes the GPIO lookups not working.

Fixes: 1448c2d2ca5c ("platform/x86: x86-android-tablets: enable fwnode matching of GPIO chips")
Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
---
 drivers/platform/x86/x86-android-tablets/core.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/platform/x86/x86-android-tablets/core.c b/drivers/platform/x86/x86-android-tablets/core.c
index b028af1c9942..1706287fd136 100644
--- a/drivers/platform/x86/x86-android-tablets/core.c
+++ b/drivers/platform/x86/x86-android-tablets/core.c
@@ -417,9 +417,15 @@ static int gpio_secondary_fwnode_init(struct device *parent,
 		if (WARN_ON(!fwnode))
 			return -ENOENT;
 
-		set_secondary_fwnode(dev, fwnode);
+		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
+		if (!phys_dev)
+			return dev_err_probe(parent,
+					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
+					     (*swnode)->name);
 
-		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(dev));
+		set_secondary_fwnode(phys_dev, fwnode);
+
+		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(phys_dev));
 		if (ret)
 			return ret;
 	}
-- 
2.55.0

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Dmitry Torokhov]

On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
> acpi_bus_find_device_by_name() call returns a pointer to the device object
> on the ACPI bus, aka the ACPI companion device.
> 
> gpio_secondary_fwnode_init() then continues with setting the secondary
> fwnode on this device. But this is not the actual physical device for
> the GPIO controller (e.g. the GPIO controller platform bus device).
> 
> This mismatch is causing GPIO lookups by secondary fwnode to not work.
> 
> Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
> of the first physical device associated with the ACPI companion device.
> 
> This fixes the GPIO lookups not working.

Nice find, thank you!

> 
> Fixes: 1448c2d2ca5c ("platform/x86: x86-android-tablets: enable fwnode matching of GPIO chips")
> Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
> ---
>  drivers/platform/x86/x86-android-tablets/core.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/platform/x86/x86-android-tablets/core.c b/drivers/platform/x86/x86-android-tablets/core.c
> index b028af1c9942..1706287fd136 100644
> --- a/drivers/platform/x86/x86-android-tablets/core.c
> +++ b/drivers/platform/x86/x86-android-tablets/core.c
> @@ -417,9 +417,15 @@ static int gpio_secondary_fwnode_init(struct device *parent,
>  		if (WARN_ON(!fwnode))
>  			return -ENOENT;
>  
> -		set_secondary_fwnode(dev, fwnode);
> +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
> +		if (!phys_dev)
> +			return dev_err_probe(parent,
> +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
> +					     (*swnode)->name);
>  
> -		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(dev));
> +		set_secondary_fwnode(phys_dev, fwnode);
> +
> +		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(phys_dev));

I am surprised that acpi_get_first_physical_node() does not bump up
refcount of the device...

>  		if (ret)
>  			return ret;

This leaks refcount on error though...

>  	}

Thanks.

-- 
Dmitry

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Hans de Goede]

Hi,

On 31-Aug-26 22:22, Dmitry Torokhov wrote:
> On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
>> acpi_bus_find_device_by_name() call returns a pointer to the device object
>> on the ACPI bus, aka the ACPI companion device.
>>
>> gpio_secondary_fwnode_init() then continues with setting the secondary
>> fwnode on this device. But this is not the actual physical device for
>> the GPIO controller (e.g. the GPIO controller platform bus device).
>>
>> This mismatch is causing GPIO lookups by secondary fwnode to not work.
>>
>> Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
>> of the first physical device associated with the ACPI companion device.
>>
>> This fixes the GPIO lookups not working.
> 
> Nice find, thank you!
> 
>>
>> Fixes: 1448c2d2ca5c ("platform/x86: x86-android-tablets: enable fwnode matching of GPIO chips")
>> Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
>> ---
>>  drivers/platform/x86/x86-android-tablets/core.c | 10 ++++++++--
>>  1 file changed, 8 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/platform/x86/x86-android-tablets/core.c b/drivers/platform/x86/x86-android-tablets/core.c
>> index b028af1c9942..1706287fd136 100644
>> --- a/drivers/platform/x86/x86-android-tablets/core.c
>> +++ b/drivers/platform/x86/x86-android-tablets/core.c
>> @@ -417,9 +417,15 @@ static int gpio_secondary_fwnode_init(struct device *parent,
>>  		if (WARN_ON(!fwnode))
>>  			return -ENOENT;
>>  
>> -		set_secondary_fwnode(dev, fwnode);
>> +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
>> +		if (!phys_dev)
>> +			return dev_err_probe(parent,
>> +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
>> +					     (*swnode)->name);
>>  
>> -		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(dev));
>> +		set_secondary_fwnode(phys_dev, fwnode);
>> +
>> +		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(phys_dev));
> 
> I am surprised that acpi_get_first_physical_node() does not bump up
> refcount of the device...
> >>  		if (ret)
>>  			return ret;
> 
> This leaks refcount on error though...

The or_reset part of devm_add_action_or_reset() will call gpio_secondary_unset
on failure which puts the ref we take here on phys_dev.

And the original reference on the ACPI bus device returned by 
acpi_bus_find_device_by_name() us auto-free-ed since dev is declared as:

	struct device *dev __free(put_device) = ...

or am I missing something here ?

(also note the refcounting is not changed, this just exchanges dev with
phys_dev)

Regards,

Hans

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Dmitry Torokhov]

On Mon, Aug 31, 2026 at 10:28:33PM +0200, Hans de Goede wrote:
> Hi,
> 
> On 31-Aug-26 22:22, Dmitry Torokhov wrote:
> > On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
> >> acpi_bus_find_device_by_name() call returns a pointer to the device object
> >> on the ACPI bus, aka the ACPI companion device.
> >>
> >> gpio_secondary_fwnode_init() then continues with setting the secondary
> >> fwnode on this device. But this is not the actual physical device for
> >> the GPIO controller (e.g. the GPIO controller platform bus device).
> >>
> >> This mismatch is causing GPIO lookups by secondary fwnode to not work.
> >>
> >> Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
> >> of the first physical device associated with the ACPI companion device.
> >>
> >> This fixes the GPIO lookups not working.
> > 
> > Nice find, thank you!
> > 
> >>
> >> Fixes: 1448c2d2ca5c ("platform/x86: x86-android-tablets: enable fwnode matching of GPIO chips")
> >> Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
> >> ---
> >>  drivers/platform/x86/x86-android-tablets/core.c | 10 ++++++++--
> >>  1 file changed, 8 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/drivers/platform/x86/x86-android-tablets/core.c b/drivers/platform/x86/x86-android-tablets/core.c
> >> index b028af1c9942..1706287fd136 100644
> >> --- a/drivers/platform/x86/x86-android-tablets/core.c
> >> +++ b/drivers/platform/x86/x86-android-tablets/core.c
> >> @@ -417,9 +417,15 @@ static int gpio_secondary_fwnode_init(struct device *parent,
> >>  		if (WARN_ON(!fwnode))
> >>  			return -ENOENT;
> >>  
> >> -		set_secondary_fwnode(dev, fwnode);
> >> +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
> >> +		if (!phys_dev)
> >> +			return dev_err_probe(parent,
> >> +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
> >> +					     (*swnode)->name);
> >>  
> >> -		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(dev));
> >> +		set_secondary_fwnode(phys_dev, fwnode);
> >> +
> >> +		ret = devm_add_action_or_reset(parent, gpio_secondary_unset, get_device(phys_dev));
> > 
> > I am surprised that acpi_get_first_physical_node() does not bump up
> > refcount of the device...
> > >>  		if (ret)
> >>  			return ret;
> > 
> > This leaks refcount on error though...
> 
> The or_reset part of devm_add_action_or_reset() will call gpio_secondary_unset
> on failure which puts the ref we take here on phys_dev.

Ah, yes, you are right. Sorry for the noise.

Reviewed-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Thanks.

-- 
Dmitry

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Ilpo Järvinen]

On Tue, 1 Sep 2026, Andy Shevchenko wrote:

> On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
> > acpi_bus_find_device_by_name() call returns a pointer to the device object
> > on the ACPI bus, aka the ACPI companion device.
> > 
> > gpio_secondary_fwnode_init() then continues with setting the secondary
> > fwnode on this device. But this is not the actual physical device for
> > the GPIO controller (e.g. the GPIO controller platform bus device).
> > 
> > This mismatch is causing GPIO lookups by secondary fwnode to not work.
> > 
> > Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
> > of the first physical device associated with the ACPI companion device.
> > 
> > This fixes the GPIO lookups not working.
> 
> ...
> 
> > static int gpio_secondary_fwnode_init(struct device *parent,
> 
> >  		if (WARN_ON(!fwnode))
> >  			return -ENOENT;
> >  
> > -		set_secondary_fwnode(dev, fwnode);
> > +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
> 
> It doesn't look like an auto cleaning pointer, so let's make a declaration
> outside of the code?
> 
> > +		if (!phys_dev)
> > +			return dev_err_probe(parent,
> > +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
> > +					     (*swnode)->name);
> 
> Hmm... Why not %pfwP?

Hi Hans,

I'm waiting for v2.

-- 
 i.

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Hans de Goede]

Hi Ilpo,

On 3-Sep-26 13:54, Ilpo Järvinen wrote:
> On Tue, 1 Sep 2026, Andy Shevchenko wrote:
> 
>> On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
>>> acpi_bus_find_device_by_name() call returns a pointer to the device object
>>> on the ACPI bus, aka the ACPI companion device.
>>>
>>> gpio_secondary_fwnode_init() then continues with setting the secondary
>>> fwnode on this device. But this is not the actual physical device for
>>> the GPIO controller (e.g. the GPIO controller platform bus device).
>>>
>>> This mismatch is causing GPIO lookups by secondary fwnode to not work.
>>>
>>> Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
>>> of the first physical device associated with the ACPI companion device.
>>>
>>> This fixes the GPIO lookups not working.
>>
>> ...
>>
>>> static int gpio_secondary_fwnode_init(struct device *parent,
>>
>>>  		if (WARN_ON(!fwnode))
>>>  			return -ENOENT;
>>>  
>>> -		set_secondary_fwnode(dev, fwnode);
>>> +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
>>
>> It doesn't look like an auto cleaning pointer, so let's make a declaration
>> outside of the code?
>>
>>> +		if (!phys_dev)
>>> +			return dev_err_probe(parent,
>>> +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
>>> +					     (*swnode)->name);
>>
>> Hmm... Why not %pfwP?
> 
> Hi Hans,
> 
> I'm waiting for v2.

I see, thank you for quoting Andy's email since it never made the list (1),
nor did it reach me.

I'll prepare a v2 addressing Andy's remarks.

Regards,


Hans


1) If you go to:

https://lore.kernel.org/platform-driver-x86/20260831201157.36397-1-johannes.goede@oss.qualcomm.com/#r

it shows a "not found" <apcRAvoHkh9blDwK@black.igk.intel.com> entry, which it
only knows about because of this reply to it by you (Ilpo).

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Ilpo Järvinen]

[-- Attachment #1: Type: text/plain, Size: 2250 bytes --]

On Tue, 8 Sep 2026, Hans de Goede wrote:

> Hi Ilpo,
> 
> On 3-Sep-26 13:54, Ilpo Järvinen wrote:
> > On Tue, 1 Sep 2026, Andy Shevchenko wrote:
> > 
> >> On Mon, Aug 31, 2026 at 10:11:57PM +0200, Hans de Goede wrote:
> >>> acpi_bus_find_device_by_name() call returns a pointer to the device object
> >>> on the ACPI bus, aka the ACPI companion device.
> >>>
> >>> gpio_secondary_fwnode_init() then continues with setting the secondary
> >>> fwnode on this device. But this is not the actual physical device for
> >>> the GPIO controller (e.g. the GPIO controller platform bus device).
> >>>
> >>> This mismatch is causing GPIO lookups by secondary fwnode to not work.
> >>>
> >>> Modify gpio_secondary_fwnode_init() to instead set the secondary fwnode
> >>> of the first physical device associated with the ACPI companion device.
> >>>
> >>> This fixes the GPIO lookups not working.
> >>
> >> ...
> >>
> >>> static int gpio_secondary_fwnode_init(struct device *parent,
> >>
> >>>  		if (WARN_ON(!fwnode))
> >>>  			return -ENOENT;
> >>>  
> >>> -		set_secondary_fwnode(dev, fwnode);
> >>> +		struct device *phys_dev = acpi_get_first_physical_node(to_acpi_device(dev));
> >>
> >> It doesn't look like an auto cleaning pointer, so let's make a declaration
> >> outside of the code?
> >>
> >>> +		if (!phys_dev)
> >>> +			return dev_err_probe(parent,
> >>> +					     -ENODEV, "No physical device for ACPI GPIO dev: %s\n",
> >>> +					     (*swnode)->name);
> >>
> >> Hmm... Why not %pfwP?
> > 
> > Hi Hans,
> > 
> > I'm waiting for v2.
> 
> I see, thank you for quoting Andy's email since it never made the list (1),
> nor did it reach me.
> 
> I'll prepare a v2 addressing Andy's remarks.
> 
> Regards,
> 
> 
> Hans
> 
> 
> 1) If you go to:
> 
> https://lore.kernel.org/platform-driver-x86/20260831201157.36397-1-johannes.goede@oss.qualcomm.com/#r
> 
> it shows a "not found" <apcRAvoHkh9blDwK@black.igk.intel.com> entry, which it
> only knows about because of this reply to it by you (Ilpo).

Yeah, I noticed it didn't end up into patchwork either but figured you'd 
see the content from my reply if the original had failed to reach you.

I've applied v2 now, thanks.

-- 
 i.

Re: [PATCH] platform/x86: x86-android-tablets: fix gpio_secondary_fwnode_init() not working

[Andy Shevchenko]

On Tue, Sep 08, 2026 at 07:51:55PM +0200, Hans de Goede wrote:
> On 3-Sep-26 13:54, Ilpo Järvinen wrote:

...

> I see, thank you for quoting Andy's email since it never made the list (1),
> nor did it reach me.

Oh, it was not intended to be not public. Thanks, Ilpo, for sharing the content.

> I'll prepare a v2 addressing Andy's remarks.

Thank you!

-- 
With Best Regards,
Andy Shevchenko