Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   Joel Fernandes <joel@joelfernandes.org>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH v2] rcu: Fix the start_poll_synchronize_rcu_expedited() be invoked very early
Date:   Thu, 12 Jan 2023 09:54:26 -0500
Message-Id: <13E7FFFD-9F91-4221-B52C-E6CB1EFBC2AB@joelfernandes.org>
References: <20230112075629.1661429-1-qiang1.zhang@intel.com>
Cc:     paulmck@kernel.org, frederic@kernel.org, quic_neeraju@quicinc.com,
        rcu@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <20230112075629.1661429-1-qiang1.zhang@intel.com>
To:     Zqiang <qiang1.zhang@intel.com>
Precedence: bulk


> On Jan 12, 2023, at 2:51 AM, Zqiang <qiang1.zhang@intel.com> wrote:
>=20
> =EF=BB=BFCurrently, the start_poll_synchronize_rcu_expedited() can be invo=
ked
> very early. before rcu_init(), the rcu_data structure's->mynode is not
> initialized, if invoke start_poll_synchronize_rcu_expedited() before
> rcu_init(), will trigger a null rcu_node structure's->exp_seq_poll access.=

>=20
> This commit add boot_exp_seq_poll_rq member to rcu_state structure to
> store seq number return by invoke start_poll_synchronize_rcu_expedited()
> very early.
>=20
> Fixes: d96c52fe4907 ("rcu: Add polled expedited grace-period primitives")
> Signed-off-by: Zqiang <qiang1.zhang@intel.com>

Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>

Thanks.


> ---
> kernel/rcu/tree.c     | 3 ++-
> kernel/rcu/tree.h     | 1 +
> kernel/rcu/tree_exp.h | 6 ++++--
> 3 files changed, 7 insertions(+), 3 deletions(-)
>=20
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 63545d79da51..12f0891ce7f4 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -92,6 +92,7 @@ static struct rcu_state rcu_state =3D {
>    .exp_mutex =3D __MUTEX_INITIALIZER(rcu_state.exp_mutex),
>    .exp_wake_mutex =3D __MUTEX_INITIALIZER(rcu_state.exp_wake_mutex),
>    .ofl_lock =3D __ARCH_SPIN_LOCK_UNLOCKED,
> +    .boot_exp_seq_poll_rq =3D RCU_GET_STATE_COMPLETED,
> };
>=20
> /* Dump rcu_node combining tree at boot to verify correct setup. */
> @@ -4938,7 +4939,7 @@ void __init rcu_init(void)
>        qovld_calc =3D qovld;
>=20
>    // Kick-start any polled grace periods that started early.
> -    if (!(per_cpu_ptr(&rcu_data, cpu)->mynode->exp_seq_poll_rq & 0x1))
> +    if (!(rcu_state.boot_exp_seq_poll_rq & 0x1))
>        (void)start_poll_synchronize_rcu_expedited();
>=20
>    rcu_test_sync_prims();
> diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
> index 192536916f9a..ae50ca6853ad 100644
> --- a/kernel/rcu/tree.h
> +++ b/kernel/rcu/tree.h
> @@ -397,6 +397,7 @@ struct rcu_state {
>                        /* Synchronize offline with */
>                        /*  GP pre-initialization. */
>    int nocb_is_setup;            /* nocb is setup from boot */
> +    unsigned long boot_exp_seq_poll_rq;
> };
>=20
> /* Values for rcu_state structure's gp_flags field. */
> diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h
> index 956cd459ba7f..1b35a1e233d9 100644
> --- a/kernel/rcu/tree_exp.h
> +++ b/kernel/rcu/tree_exp.h
> @@ -1068,9 +1068,11 @@ unsigned long start_poll_synchronize_rcu_expedited(=
void)
>    if (rcu_init_invoked())
>        raw_spin_lock_irqsave(&rnp->exp_poll_lock, flags);
>    if (!poll_state_synchronize_rcu(s)) {
> -        rnp->exp_seq_poll_rq =3D s;
> -        if (rcu_init_invoked())
> +        if (rcu_init_invoked()) {
> +            rnp->exp_seq_poll_rq =3D s;
>            queue_work(rcu_gp_wq, &rnp->exp_poll_wq);
> +        } else
> +            rcu_state.boot_exp_seq_poll_rq =3D s;
>    }
>    if (rcu_init_invoked())
>        raw_spin_unlock_irqrestore(&rnp->exp_poll_lock, flags);
> --=20
> 2.25.1
>=20