On Thu, May 16, 2024, at 14:09, Doug Anderson wrote:
> On Thu, May 16, 2024 at 6:43 AM Doug Anderson <dianders@xxxxxxxxxxxx> wrote:
>> On Wed, May 15, 2024 at 11:55 PM <neil.armstrong@xxxxxxxxxx> wrote:
>> > On 16/05/2024 08:43, cong yang wrote:
>> >
>> > Yeah we usually don't mess with arch specific defconfig from drm tree
>>
>> In general I agree that makes sense. In this case, though, the new
>> config symbol was introduced in the previous patch and split off an
>> existing symbol. Updating "all" of the configs (AKA just arm64) that
>> had the old symbol to also have the new symbol seems like the nice
>> thing to do and it feels like it makes sense to land in the same tree
>> that did the "split" just to cause the least confusion to anyone
>> affected.
>>
>> In any case, if it's going to land in some other tree then I guess the
>> question is whether it needs to wait a few revisions to land there or
>> if it should land right away. Nobody would get a compile error if it
>> landed in a different tree right away since unknown config symbols are
>> silently ignored, but it feels a little weird to me.
>>
>> ...of course, I'm also OK just dropping the config patch. I personally
>> don't use the upstream "defconfig". It just seemed courteous to update
>> it for those who do.
>
> Hmmm, probably should have put Arnd on this thread. Added now in case
> he has any opinions. I also did manage to find when this last came up
> where I was involved. At that time Will Deacon (who get_maintainer.pl
> reports is the official maintainer of this file) said [1]:
>
>> But yes, although there are a few things I really care about
>> in defconfig (e.g. things like page size!), generally speaking we don't
>> need to Ack everything that changes in there.
>
My preferred way of getting arm/arm64 defconfig updates is to have
them picked up by the platform maintainer, the same way we handle
updates to dts files. The platform maintainers are familiar with the
process and will send the patches on to me for integration through
the soc tree.
If a change is not specific to any particular platform, I recommend
to send it to:soc@xxxxxxxxxx, cc:lakml. This makes it show up in
my patchwork, so I will eventually get around to picking it up.
When you do this, it's helpful to me if you include an explanation
(after the --- line) why this patch does not get picked up by
a platform maintainer, and it also helps me to include whether
I should include it in the current (6.10) fixes or queue it for
the next merge window.
Arnd
