Re: [PATCH] acer-wmi: check wireless capability on AMW0_GUID2 machines

[Ike Panhc <ike.pan@canonical.com>]

Hi Joey,

On 01/20/2012 10:44 AM, joeyli wrote:
> Hi Ike
> 
> 於 四，2012-01-12 於 00:54 +0800，Ike Panhc 提到：
>> On 01/11/2012 12:26 PM, joeyli wrote:
>>> 於 一，2012-01-09 於 22:15 +0000，Carlos Corbacho 提到：
>>>> On Monday 09 Jan 2012 18:00:24 joeyli wrote:
>>>>> I just checked and found I also have no AMW0_GUID2 machines on my hand,
>>>>> now.
>>>>> I have some questions about AMW0_GUID2:
>>>>>
>>>>>   + Why we enabled wireless capability in acer-wim if a non-acer machine has
>>>>> AMW0_GUID2 ?
>>>>
>>>> IIRC, there were some bug reports on non-Acer hardware (I don't think this 
>>>> involved Acer hardware, but this was some time ago...) where the wireless 
>>>> detection wasn't working with this, but if we forced it on, it worked fine.
>>>>
>>>
>>> Yes, in acer-wmi supported some non-acer machine by quirk and read EC
>>> register. That means we need find out EC register reflect to wifi RF
>>> state if we are lucky.
>>>
>>> Unfortunately, direct enable wireless capability causes some non-acer
>>> machines' wireless rfkill always disabled. e.g. Lenovo E520
>>>
>>> https://bugzilla.redhat.com/show_bug.cgi?id=674353 
>>>
>>> If we don't check the wireless capability, then they need manually add
>>> acer-wmi to backlist. On some lenovo machines, ideapad-laptop driver
>>> should support by lenovo's ACPI method.
>>>
>>
>> I've got two reports which says after poking the ec register, the phy0 rfkill reports hard blocked by mistake. And few hours ago I get the third report says the similar issue. I will summary the report.
>>
> 
> Do you mean the above 2 reports is use this patch? or actually it's
> other problem?
> Could you please kindly test this patch on issue machines?
> 

Yes, I ask them to blacklist patched acer-wmi to confirm the problem.

I am afraid that after we change the value of ec register, ec firmware might do something unexpectedly. Since acer-wmi is loaded on lots of machine with different EC, it might cause unexpected and different regression on each machine.

>> I am thinking about alternative way
>>
>> Since we have thinkpad_acpi and ideapad_laptop, both of them have rfkill capability. If there is VPC2004/IBM0068/LEN0068 in DSDT, then it will be fine if we do not register rfkill in acer-wmi.
>>
>> Any thought?
>>
> 
> It's also another idea.
> 
> 
> Thanks a lot!
> Joey Lee
> 
>