From: "Dr. David Alan Gilbert" <linux@xxxxxxxxxxx>
>* MyungJoo Ham (myungjoo.ham@xxxxxxxxxxx) wrote:
>> >* linux@xxxxxxxxxxx (linux@xxxxxxxxxxx) wrote:
>> >> From: "Dr. David Alan Gilbert" <linux@xxxxxxxxxxx>
>> >>
>> >> devm_devfreq_register_notifier() and devm_devfreq_unregister_notifier()
>> >> have been unused since 2019's
>> >> commit 0ef7c7cce43f ("PM / devfreq: passive: Use non-devm notifiers")
>> >>
>> >> Remove them, and the helpers they used.
>> >>
>> >> Note, devm_devfreq_register_notifier() is still used as an example
>> >> in Documentation/doc-guide/contributing.rst but that's just
>> >> an example of an old doc bug rather than anything about the function
>> >> itself.
>> >>
>> >> Signed-off-by: Dr. David Alan Gilbert <linux@xxxxxxxxxxx>
>> >
>> >Ping.
>> >
>> >Thanks,
>> >
>> >Dave
>>
>> When I search github, it appears that vendors are using this API.
>
>Hmm OK.
>Of course there's a lot of random junk on github, so it can be tricky
>to know what's current/real/relevant.
>
>> NVIDIA:
>> Samsung:
>> Realtek:
>>
>> Please don't remove ABIs used by vendors even if
>> they didn't upstream their drivers.
>
>Hmm OK.
>Do you think they should be using this ABI or do they have the same bug as
>is fixed in 0ef7c7cce43f ?
>I guess they don't care.
>
>Dave
They will suffer from the same bugs mentioned in 0ef7x7cce43f.
Anyway, they probably don't care if they build these
code as built-in for their product binries.
Vendors of embedded devices, including mobile phones, usually
do not care upstreaming their device drivers, and we have too many
Linux embedded device vendors.
Even in my affiliation, we have too many different instances of
Linux kernel source repositories and binaries built
simultaneously, and I do not know how they are using the
given devfreq APIs.
Cheers,
MyungJoo
