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;
Date:   Wed, 22 Apr 2020 09:18:41 -0400
User-Agent: K-9 Mail for Android
In-Reply-To: <20200422103536.GA16263@pc636>
References: <20200420153837.194532-1-joel@joelfernandes.org> <20200420153837.194532-3-joel@joelfernandes.org> <20200422103536.GA16263@pc636>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PATCH rcu/dev -fixes 2/4] rcu/tree: Skip entry into the page allocator for PREEMPT_RT
To:     Uladzislau Rezki <urezki@gmail.com>
CC:     linux-kernel@vger.kernel.org,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Josh Triplett <josh@joshtriplett.org>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
        "Paul E. McKenney" <paulmck@kernel.org>, rcu@vger.kernel.org,
        Steven Rostedt <rostedt@goodmis.org>
From:   joel@joelfernandes.org
Message-ID: <51A69C78-FDD4-4A15-8969-EEA17CBA5BBD@joelfernandes.org>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


On April 22, 2020 6:35:36 AM EDT, Uladzislau Rezki <urezki@gmail=2Ecom> wr=
ote:
>On Mon, Apr 20, 2020 at 11:38:35AM -0400, Joel Fernandes (Google)
>wrote:
>> To keep kfree_rcu() path working on raw non-preemptible sections,
>> prevent the optional entry into the allocator as it uses sleeping
>locks=2E
>> In fact, even if the caller of kfree_rcu() is preemptible, this path
>> still is not, as krcp->lock is a raw spinlock as done in previous
>> patches=2E With additional page pre-allocation in the works, hitting
>this
>> return is going to be much less likely soon so just prevent it for
>now
>> so that PREEMPT_RT does not break=2E Note that page allocation here is
>an
>> optimization and skipping it still makes kfree_rcu() work=2E
>>=20
>> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix=2Ede>
>> Co-developed-by: Uladzislau Rezki <urezki@gmail=2Ecom>
>> Signed-off-by: Uladzislau Rezki <urezki@gmail=2Ecom>
>> Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes=2Eorg>
>> ---
>>  kernel/rcu/tree=2Ec | 12 ++++++++++++
>>  1 file changed, 12 insertions(+)
>>=20
>> diff --git a/kernel/rcu/tree=2Ec b/kernel/rcu/tree=2Ec
>> index cf68d3d9f5b81=2E=2Ecd61649e1b001 100644
>> --- a/kernel/rcu/tree=2Ec
>> +++ b/kernel/rcu/tree=2Ec
>> @@ -3092,6 +3092,18 @@ kfree_call_rcu_add_ptr_to_bulk(struct
>kfree_rcu_cpu *krcp,
>>  		if (!bnode) {
>>  			WARN_ON_ONCE(sizeof(struct kfree_rcu_bulk_data) > PAGE_SIZE);
>> =20
>> +			/*
>> +			 * To keep this path working on raw non-preemptible
>> +			 * sections, prevent the optional entry into the
>> +			 * allocator as it uses sleeping locks=2E In fact, even
>> +			 * if the caller of kfree_rcu() is preemptible, this
>> +			 * path still is not, as krcp->lock is a raw spinlock=2E
>> +			 * With additional page pre-allocation in the works,
>> +			 * hitting this return is going to be much less likely=2E
>> +			 */
>> +			if (IS_ENABLED(CONFIG_PREEMPT_RT))
>> +				return false;
>> +
>>  			bnode =3D (struct kfree_rcu_bulk_data *)
>>  				__get_free_page(GFP_NOWAIT | __GFP_NOWARN);
>>  		}
>This will not be correctly working by just reverting everything to the
>"rcu_head path" for CONFIG_PREEMPT_RT kernel=2E We need to preallocate at
>least once=2E I can add caching on top of this series to address it=2E
>

I discussed with Vlad offline, this patch is fine for mainline as we don't=
 have headless support yet=2E So this patch is good=2E Future patches addin=
g caching will also add the headless support after additional caching, so s=
kipping the allocation here is ok=2E

Thanks=2E

- Joel


>--
>Vlad Rezki

--=20
Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E