[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230906-eichenbaum-hektar-01140b56ca64@brauner>
Date: Wed, 6 Sep 2023 09:13:29 +0200
From: Christian Brauner <brauner@...nel.org>
To: Aleksa Sarai <cyphar@...har.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Damian Tometzki <dtometzki@...oraproject.org>,
Daniel Verkamp <dverkamp@...omium.org>,
Jeff Xu <jeffxu@...gle.com>, Kees Cook <keescook@...omium.org>,
Shuah Khan <shuah@...nel.org>, linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] memfd: drop warning for missing exec-related flags
On Wed, Sep 06, 2023 at 05:02:06PM +1000, Aleksa Sarai wrote:
> Commit 434ed3350f57 ("memfd: improve userspace warnings for missing
> exec-related flags") attempted to make these warnings more useful (so
> they would work as an incentive to get users to switch to specifying
> these flags -- as intended by the original MFD_NOEXEC_SEAL patchset).
> Unfortunately, it turns out that even INFO-level logging is too extreme
> to enable by default and alternative solutions to the spam issue (such
> as doing more extreme rate-limiting per-task) are either too ugly or
> overkill for something as simple as emitting a log as a developer aid.
>
> Given that the flags are new and there is no harm to not specifying them
> (after all, we maintain backwards compatibility) we can just drop the
> warnings for now until some time in the future when most programs have
> migrated and distributions start using vm.memfd_noexec=1 (where failing
> to pass the flag would result in unexpected errors for programs that use
> executable memfds).
>
> Fixes: 434ed3350f57 ("memfd: improve userspace warnings for missing exec-related flags")
> Reported-by: Damian Tometzki <dtometzki@...oraproject.org>
> Signed-off-by: Aleksa Sarai <cyphar@...har.com>
> ---
Looks good to me,
Reviewed-by: Christian Brauner <brauner@...nel.org>
Powered by blists - more mailing lists