Re: [PATCH -mm v2 1/1] sysctl: remove CONFIG_PROC_SYSCTL, it just mirrors CONFIG_SYSCTL

[Oleg Nesterov]

On 07/27, Nicolas Schier wrote:
>
> On Mon, Jul 27, 2026 at 04:47:22PM +0200, Oleg Nesterov wrote:
> >
> > -config PROC_SYSCTL
> > +config SYSCTL
> >  	bool "Sysctl support (/proc/sys)" if EXPERT
> >  	depends on PROC_FS
> > -	select SYSCTL
> >  	default y
> >  	help
> >  	  The sysctl interface provides a means of dynamically changing
>
> Please be aware that renaming of Kconfig symbols may not be welcomed due
> to unwanted effects for users and for bisecting, cp. last paragraph of

You mean "Renaming config parameters is *bad*.". Yes, this was my concern
too, please see below.

> Personally, I do understand that it makes sense to clean this up, here.
> As the default was and stays 'y' this should affect only "few" users,
> but still I am feeling ambigious about it.
>
> Sorry for missing to point to that at the RFC.

That is why the previous [RFC PATCH 1/1] killed the hidden SYSCTL, not
PROC_SYSCTL.

But that patch touched 170 files and the output from get_maintainer.pl
was huge.

>From [RFC PATCH 0/1]:

	Alternatively, we could kill CONFIG_PROC_SYSCTL in favour of CONFIG_SYSCTL.
	This would result in a much smaller patch, but I am not sure if it is "safe"
	because CONFIG_PROC_SYSCTL is user-visible in that it has a prompt in
	fs/proc/Kconfig. But perhaps this is fine?

and after discussion with Eric we came to conclusion that "probably this is fine"
in thic case.

I do think it makes sense to cleanup the usage of PROC_SYSCTL/SYSCTL (kill one of
them). Just look at kernel/pid.c. It defines the "static int proc_do_cad_pid()" if
CONFIG_SYSCTL, but this function is only used if CONFIG_PROC_SYSCTL. This looks
obviously wrong until you grep Kconfig's and realize that both are the same thing.

Oleg.