Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1571509 > unrolled thread

Re: [PATCH 0/8] asus-wmi: Get rid of quirk_no_rfkill{,_wapf4}

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH 0/8] asus-wmi: Get rid of quirk_no_rfkill{,_wapf4} João Paulo Rechi Vita <jprvita@gmail.com> - 2017-02-01 13:30 +0100
    Re: [PATCH 0/8] asus-wmi: Get rid of quirk_no_rfkill{,_wapf4} Andy Shevchenko <andy.shevchenko@gmail.com> - 2017-02-04 16:10 +0100

#1571509 — Re: [PATCH 0/8] asus-wmi: Get rid of quirk_no_rfkill{,_wapf4}

On 27 January 2017 at 10:37, Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
> On Thu, Jan 26, 2017 at 5:30 PM, João Paulo Rechi Vita
> <jprvita@gmail.com> wrote:
>> This quirk avoids the conflicting usage of ASUS_WMI_DEVID_WLAN_LED (0x00010002)
>> by both asus-wireless and asus-wmi on machines where the BIOS can't set and
>> record the wlan status (wlan_ctrl_by_user in asus-wmi.c). At the moment we have
>> six models using that quirk upstream, but there are another fifteen for which
>> we have downstream patches, and will send upstream unless we find a more
>> dynamic solution. We also expect this list to continue growing.
>>
>> This series makes use of ASUS_WMI_DSTS_USER_BIT plus the status of the ASHS
>> device to skip asus_wmi_rfkill_init(). Since asus-wmi now needs to know the ASHS
>> _HIDs, I have moved device_ids[] to a new header file, asus-wireless.h. I'm not
>> sure if there are any best practices in the kernel community against having one
>> module including headers from a different module, but I wanted to avoid
>> duplicating this list or creating an ordering dependency between the modules.
>> Any advice on alternatives here are welcome.
>>
>> PS: I don't think asus-wireless.h needs to also be listed in MAINTAINERS,
>> please correct me if I'm wrong.
>>
>> João Paulo Rechi Vita (8):
>
>>   Revert "asus-wmi: Add quirk_no_rfkill_wapf4 for the Asus X456UA"
>>   Revert "asus-wmi: Add quirk_no_rfkill_wapf4 for the Asus X456UF"
>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus Z550MA"
>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus U303LB"
>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus N552VW"
>>   Revert "asus-wmi: Create quirk for airplane_mode LED"
>
> Don't do this.
> Each of them will break the working case -> means regression on user experience.
>

They are being substituted by a generic fix implemented by the
following two  commits, so the user experience will not change at all
after the whole series get merged.

>>   asus-wireless: Export ids list
>>   asus-wmi: Don't register rfkill if ASHS and user bit are present
>>

Or your point is that these two should come before the reverts? I
thought patchsets were looked at as an unit from the user experience /
functionally perspective. Can you please advise on what is the correct
approach here?

Thanks again for your review on this series.

--
João Paulo Rechi Vita
http://about.me/jprvita

[toc] | [next] | [standalone]

#1573679

On Wed, Feb 1, 2017 at 2:21 PM, João Paulo Rechi Vita <jprvita@gmail.com> wrote:
> On 27 January 2017 at 10:37, Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
>> On Thu, Jan 26, 2017 at 5:30 PM, João Paulo Rechi Vita
>> <jprvita@gmail.com> wrote:

>>>   Revert "asus-wmi: Add quirk_no_rfkill_wapf4 for the Asus X456UA"
>>>   Revert "asus-wmi: Add quirk_no_rfkill_wapf4 for the Asus X456UF"
>>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus Z550MA"
>>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus U303LB"
>>>   Revert "asus-wmi: Add quirk_no_rfkill for the Asus N552VW"
>>>   Revert "asus-wmi: Create quirk for airplane_mode LED"
>>
>> Don't do this.
>> Each of them will break the working case -> means regression on user experience.
>>
>
> They are being substituted by a generic fix implemented by the
> following two  commits, so the user experience will not change at all
> after the whole series get merged.
>
>>>   asus-wireless: Export ids list
>>>   asus-wmi: Don't register rfkill if ASHS and user bit are present
>>>
>
> Or your point is that these two should come before the reverts? I
> thought patchsets were looked at as an unit from the user experience /
> functionally perspective. Can you please advise on what is the correct
> approach here?

You prepare entire new infrastructure and then you have two options:
1. Remove old stuff and apply new use in one patch
2. Replace old calls by new ones in one patch, and remove not used
anymore old stuff in the other patch.

Choose whatever suits better.

-- 
With Best Regards,
Andy Shevchenko

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web