Message-ID: <0baab14d-abea-4248-e6d5-23c465f0a4a5@redhat.com>
Date:   Thu, 16 Feb 2023 16:07:25 -0500
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.6.0
Subject: Re: [PATCH 2/2] locking/rwsem: Wake up all readers for wait queue
 waker
Content-Language: en-US
To:     Hillf Danton <hdanton@sina.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>, Will Deacon <will@kernel.org>,
        Boqun Feng <boqun.feng@gmail.com>, linux-kernel@vger.kernel.org
References: <20230213194832.832256-1-longman@redhat.com>
 <20230214030901.3250-1-hdanton@sina.com>
From:   Waiman Long <longman@redhat.com>
In-Reply-To: <20230214030901.3250-1-hdanton@sina.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2/13/23 22:09, Hillf Danton wrote:
> On Mon, 13 Feb 2023 14:48:32 -0500 Waiman Long <longman@redhat.com>
>>   
>> @@ -1281,7 +1282,7 @@ static struct rw_semaphore *rwsem_downgrade_wake(struct rw_semaphore *sem)
>>   	raw_spin_lock_irqsave(&sem->wait_lock, flags);
>>   
>>   	if (!list_empty(&sem->wait_list))
>> -		rwsem_mark_wake(sem, RWSEM_WAKE_READ_OWNED, &wake_q);
>> +		rwsem_mark_wake(sem, RWSEM_WAKE_READ_OWNED, &wake_q, false);
>>   
>>   	raw_spin_unlock_irqrestore(&sem->wait_lock, flags);
>>   	wake_up_q(&wake_q);
>> -- 
>> 2.31.1
> Downgrade is conceptually the right time to let all read waiters go
> regardless write waiter.

Still, a downgraded task is still in the read critical section and we 
shouldn't introduce arbitrary latency to that. Let's focus on the easy 
one and we can discuss about other possibility later.

Thanks,
Longman