linux-kernel - [PATCH 0/5] turn signal_struct.stats_lock into seqcount_rwlock

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20230913154907.GA26210@redhat.com>
Date:   Wed, 13 Sep 2023 17:49:07 +0200
From:   Oleg Nesterov <oleg@...hat.com>
To:     Boqun Feng <boqun.feng@...il.com>, Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Rik van Riel <riel@...riel.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Waiman Long <longman@...hat.com>, Will Deacon <will@...nel.org>
Cc:     Alexey Gladkov <legion@...nel.org>,
        "Eric W. Biederman" <ebiederm@...ssion.com>,
        linux-kernel@...r.kernel.org
Subject: [PATCH 0/5] turn signal_struct.stats_lock into seqcount_rwlock_t

Hello,

RFC, not for inclusion yet. Please review, at least the intent.

In particular, can you look at the changelog from the last patch?
Am I right that currently thread_group_cputime() is not RT friendly?

During the (ongoing) s/while_each_thread/for_each_thread/ conversion
I noticed that some of these users can use signal->stats_lock instead
of lock_task_sighand(). And if we change them, we can try to avoid
stats_lock under siglock in wait_task_zombie() at least.

However, signal->stats_lock is seqlock_t but I think seqcount_rwlock_t
make more sense. So let me try to turn it into seqcount_rwlock_t first.

OTOH... I am not sure I understand the value of signal->stats_lock.
I mean, do we have any numbers which prove that seqlock_t is really
"better" than the plain rwlock_t ?

So far only compile tested, and I need to re-read these changes with
a clear head. In any case, it is not that I myself like this series
very much, and most probably I did something wrong...

Oleg.
---
 include/linux/sched/signal.h |   4 +-
 include/linux/seqlock.h      | 105 +++++++++++++++++++++++++++++++++++--------
 kernel/exit.c                |  12 +++--
 kernel/fork.c                |   3 +-
 kernel/sched/cputime.c       |  10 +++--
 5 files changed, 106 insertions(+), 28 deletions(-)