Re: [PATCH v2 4/4] platform/x86: topstar-laptop: add optional WLAN LED workaround

[Andy Shevchenko]

On Fri, Nov 3, 2017 at 5:07 PM, Guillaume DouÃzan-Grard
<gdouezangrard@xxxxxxxxx> wrote:
> On Fri, Nov 03, 2017 at 02:50:52PM +0200, Andy Shevchenko wrote:
>> On Sun, Oct 29, 2017 at 1:53 AM, Guillaume DouÃzan-Grard
>> <gdouezangrard@xxxxxxxxx> wrote:
>> > Topstar U931 laptops provide an LED synced with the WLAN adapter
>> > hard-blocking state. Unfortunately, some models seem to be defective,
>> > making impossible to hard-block the adapter with the WLAN switch and
>> > thus the LED is useless.
>> >
>> > An ACPI method is available to programmatically control this switch and
>> > it indirectly allows to control the LED.
>> >
>> > This commit registers the LED within the corresponding subsystem, making
>> > possible for instance to use an rfkill-based trigger to synchronize the
>> > LED with the soft-blocking state.
>> >
>> > This feature is disabled by default and can be enabled with the
>> > `led_workaround` module parameter.
>>
>> >  #include <linux/platform_device.h>
>> >  #include <linux/input.h>
>> >  #include <linux/input/sparse-keymap.h>
>> > +#include <linux/leds.h>
>>
>> Yep, exact place, esp. after moving platform_device to the right place.
>>
>> > +static bool led_workaround;
>> > +module_param_named(led_workaround, led_workaround, bool, 0444);
>> > +MODULE_PARM_DESC(led_workaround,
>> > +               "Enables software-based WLAN LED control on systems with defective hardware switch");
>>
>> So, this is most problematic piece in the series.
>>
>> We are not encouraging module parameters. Why do we need one? Can't be
>> detected automatically (perhaps based on DMI strings)?
>
> Darren told me that.

> I tried to answer this question in the cover letter:

Perhaps it makes sense to put this explanation in the commit message.

> "These are barebone laptops, sold under quite a lot of brands and
> configurations, with different firmwares and so on. I can only say for sure
> that this issue is present for all the models sold under a specific brand,
> that's why I'm reluctant to enable this by default with a DMI check."
>
> In my case for instance, the DMI info has not been filled in by the retailler
> since I only have the ODM base board information to identify a model.

I see. I would like to have a consensus on this one with Darren, the
rest (after addressing comments) looks good to me.

> For now, I will prepare a new version containing the other needed changes you
> pointed at for the other patches.

OK.

-- 
With Best Regards,
Andy Shevchenko