Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Davidlohr Bueso <dave@stgolabs.net>
To:     tglx@linutronix.de
Cc:     peterz@infradead.org, mingo@kernel.org, rostedt@goodmis.org,
        longman@redhat.com, bigeasy@linutronix.de, boqun.feng@gmail.com,
        dave@stgolabs.net, linux-kernel@vger.kernel.org
Subject: [PATCH 0/2] locking/rwbase_rt: Use wake_q for lockless reader waker
Date:   Wed,  1 Sep 2021 15:28:23 -0700
Message-Id: <20210901222825.6313-1-dave@stgolabs.net>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

Patch 1 does some necessary code reordering.

Patch 2 introduces wake_up_q_state() and converts rwbase_read_unlock()
slowpath to use wake_q instead of holding the wait_lock across the
entire wakeup.

I have tested equivalent changes in the preempt_rt kernel (v5.14.y-rt)
stressing mmap_sem, without anything falling out.

Thanks!

Davidlohr Bueso (2):
  sched: Move wake_q code below try_to_wake_up()
  locking/rwbase_rt: Lockless reader waking up a writer

 kernel/locking/rwbase_rt.c |   4 +-
 kernel/sched/core.c        | 182 +++++++++++++++++++------------------
 2 files changed, 99 insertions(+), 87 deletions(-)

--
2.26.2