From mboxrd@z Thu Jan  1 00:00:00 1970
From: xuwei5@hisilicon.com (Wei Xu)
Date: Fri, 18 May 2018 11:48:48 +0100
Subject: [GIT PULL] arm64: defconfig: hisilicon config updates for v4.18
In-Reply-To: <20180515230014.GS29062@mai>
References: <5AF5A94A.4080308@hisilicon.com>
 <20180514201445.6sglddhalvclu36v@localhost> <20180515082811.GL29062@mai>
 <20180515203755.esjnphfep5ep7wbe@localhost> <20180515230014.GS29062@mai>
Message-ID: <5AFEAF90.2030405@hisilicon.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Daniel,

On 2018/5/16 0:00, Daniel Lezcano wrote:
> On Tue, May 15, 2018 at 01:37:55PM -0700, Olof Johansson wrote:
>> On Tue, May 15, 2018 at 10:28:11AM +0200, Daniel Lezcano wrote:
>>> On Mon, May 14, 2018 at 01:14:45PM -0700, Olof Johansson wrote:
>>>> Hi Wei,
>>>>
>>>> On Fri, May 11, 2018 at 03:31:38PM +0100, Wei Xu wrote:
>>>>> Hi Arnd, Hi Olof,
>>>>>
>>>>> Please help to pull the following changes.
>>>>>
>>>>> About the CLOCK_STUB and the MAILBOX consolidate patch,
>>>>> Jassi and Stephen have acked it.
>>>>> Could you let me know how to handle this kind case
>>>>> if it is not OK to be in this pull?
>>>>
>>>> I don't think there's any need to group the Kconfig changes with the defconfig
>>>> updates here, is there?
>>>
>>> I don't have the patches history, but likely this patch should come together with:
>>>
>>> https://patchwork.kernel.org/patch/10399799/
>>> https://patchwork.kernel.org/patch/10399801/
>>>
>>> Otherwise the compilation options won't be consistent with what is enabled in
>>> the DT.
>>
>> As long as neither side regresses due to the changes, there should be no
>> problem. Just because a DT node is added in the tree there's no need to
>> configure the driver. Or am I missing some aspect of it here?
> 
> Actually, the DT node being added do not introduce regressions.
> 
> However it is expected by adding the clock stub and the mailbox to have the
> cpufreq working which is not necessarily the case because the config may be
> inconsistent, so the cpufreq may be working on some config if the user had the
> options for the clock and the mailbox enabled but these options can disappear
> and not come back because of this Kconfig inconsistencies.
> 
>>>From my point of view, by adding those DT nodes, it makes sense to give a
> consolidated Kconfig coming together and ensuring the drivers are enabled when
> the node is parsed.
> 
> On the other side, the patch is simple enough to be split and submitted in
> separated trees, hoping the maintainer Wu Xei and the branch users keep in mind
> if the board does not boot or has inconsistent behavior they will have to
> double check the options are enabled for the clock stub and the mailbox.
> 
> I don't have a strong opinion on this actually, whatever the decision is, I
> will be fine with resubmitting the patch to the different trees, or keep it as
> is and merge it through the hisi tree.

Thanks to make it clear!
In this case, I will drop this patch in the pull firstly.

Best Regards,
Wei

> 
>   -- Daniel
>