Re: [PATCH 2/5] platform/x86: wmi: Check if event data is not NULL

[Armin Wolf]

Am 15.02.24 um 13:31 schrieb Ilpo Järvinen:

> On Wed, 14 Feb 2024, Armin Wolf wrote:
>
> > WMI event drivers which do not have no_notify_data set expect
> > that each WMI event contains valid data. Evaluating _WED however
> > might return no data, which can cause issues with such drivers.
> >
> > Fix this by validating that evaluating _WED did return data.
> >
> > Signed-off-by: Armin Wolf <W_Armin@xxxxxx>
> > ---
> >   drivers/platform/x86/wmi.c | 11 +++++++++--
> >   1 file changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/platform/x86/wmi.c b/drivers/platform/x86/wmi.c
> > index 34d8f55afaad..8a916887c546 100644
> > --- a/drivers/platform/x86/wmi.c
> > +++ b/drivers/platform/x86/wmi.c
> > @@ -1211,6 +1211,7 @@ static void wmi_notify_driver(struct wmi_block *wblock)
> >   {
> >   	struct wmi_driver *driver = drv_to_wdrv(wblock->dev.dev.driver);
> >   	struct acpi_buffer data = { ACPI_ALLOCATE_BUFFER, NULL };
> > +	union acpi_object *obj = NULL;
> >   	acpi_status status;
> >
> >   	if (!driver->no_notify_data) {
> > @@ -1219,12 +1220,18 @@ static void wmi_notify_driver(struct wmi_block *wblock)
> >   			dev_warn(&wblock->dev.dev, "Failed to get event data\n");
> >   			return;
> >   		}
> > +
> > +		obj = data.pointer;
> > +		if (!obj) {
> > +			dev_warn(&wblock->dev.dev, "Event contains not event data\n");
> > +			return;
> > +		}
> >   	}
> >
> >   	if (driver->notify)
> > -		driver->notify(&wblock->dev, data.pointer);
> > +		driver->notify(&wblock->dev, obj);
> >
> > -	kfree(data.pointer);
> > +	kfree(obj);
> Hi Armin,
>
> While looking into this patch, I failed to connect the mention of
> no_notify_data in the commit message with the code change that does
> nothing differently based no_notify_data being set or not, AFAICT.
>
> It could be just that you need to explain things better in the commit
> message, I'm not sure.

Here the _WED ACPI control method is only evaluated if driver->no_notify_data is not set.
So the returned ACPI object should only be validated in this case, as we pass NULL otherwise.

Armin Wolf