From: Matthieu Baerts <matttbe@xxxxxxxxxx>
Date: Thu, 15 Feb 2024 10:58:58 +0100
> Hi Kuniyuki,
>
> On 14/02/2024 20:54, Kuniyuki Iwashima wrote:
> > Recently, commit 8e5443d2b866 ("net: remove SOCK_DEBUG leftovers")
> > removed the last users of SOCK_DEBUG(), and commit b1dffcf0da22 ("net:
> > remove SOCK_DEBUG macro") removed the macro.
> >
> > Now is the time to deprecate the oldest socket option.
> >
> > Note that setsockopt(SO_DEBUG) is moved not to acquire lock_sock().
> >
> > Reviewed-by: Gerd Bayer <gbayer@xxxxxxxxxxxxx>
> > Signed-off-by: Kuniyuki Iwashima <kuniyu@xxxxxxxxxx>
> > ---
> > v2:
> > * Move setsockopt(SO_DEBUG) code not to acquire lock_sock().
>
> Thank you for the v2!
>
> Good idea to have modified that in net/core/sock.c too!
>
> Reviewed-by: Matthieu Baerts (NGI0) <matttbe@xxxxxxxxxx>
>
>
> I don't think we need to do anything else, but just to be sure: do we
> need to tell the userspace this socket option has been deprecated?
> SO_DEBUG is a bit particular, so I guess it is fine not to do anything
> else, but except by looking at the kernel version, I don't think the
> userspace can know it has no more effect.
>
> I didn't find many examples of other deprecated socket options, apart
> from SO_BSDCOMPAT. For years, there was a warning, removed a few years
> ago: f4ecc748533d ("net: Stop warning about SO_BSDCOMPAT usage"). I
> guess we don't want that for SO_DEBUG, right?
I was also wondering if I should add pr_info_once(), but it seems not
worth adding.
With a rough grep,
$ git log -S SOCK_DEBUG net
except for the legacy protocols (appletalk, dccp, x25) that 8e5443d2b866
touched recently, TCP was the last real user of SOCK_DEBUG() and it's
2019-02-25 (5 years ago!) that 9946b3410b61 removed the use.
So, I think we don't need such warning.
Thanks!
