Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [RFC PATCH 2/2] pipe: use pipe busy wait
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     linux-kernel@vger.kernel.org, dhaval.giani@oracle.com,
        steven.sistare@oracle.com
References: <20180830202458.32579-1-subhra.mazumdar@oracle.com>
 <20180830202458.32579-3-subhra.mazumdar@oracle.com>
 <20180907122557.GY24106@hirez.programming.kicks-ass.net>
 <304ffc06-1ab9-19d5-d5ca-e63e2abe2900@oracle.com>
 <20180917224344.GB3284@worktop.programming.kicks-ass.net>
From:   Subhra Mazumdar <subhra.mazumdar@oracle.com>
Message-ID: <00c2fe72-0672-ea97-fe58-c7cd767c0837@oracle.com>
Date:   Mon, 17 Sep 2018 18:07:41 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.9.1
MIME-Version: 1.0
In-Reply-To: <20180917224344.GB3284@worktop.programming.kicks-ass.net>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


On 09/17/2018 03:43 PM, Peter Zijlstra wrote:
> On Mon, Sep 17, 2018 at 02:05:40PM -0700, Subhra Mazumdar wrote:
>> On 09/07/2018 05:25 AM, Peter Zijlstra wrote:
>>> Why not just busy wait on current->state ? A little something like:
>>>
>>> diff --git a/fs/pipe.c b/fs/pipe.c
>>> index bdc5d3c0977d..8d9f1c95ff99 100644
>>> --- a/fs/pipe.c
>>> +++ b/fs/pipe.c
>>> @@ -106,6 +106,7 @@ void pipe_double_lock(struct pipe_inode_info *pipe1,
>>>   void pipe_wait(struct pipe_inode_info *pipe)
>>>   {
>>>   	DEFINE_WAIT(wait);
>>> +	u64 start;
>>>   	/*
>>>   	 * Pipes are system-local resources, so sleeping on them
>>> @@ -113,7 +114,15 @@ void pipe_wait(struct pipe_inode_info *pipe)
>>>   	 */
>>>   	prepare_to_wait(&pipe->wait, &wait, TASK_INTERRUPTIBLE);
>>>   	pipe_unlock(pipe);
>>> -	schedule();
>>> +
>>> +	preempt_disable();
>>> +	start = local_clock();
>>> +	while (!need_resched() && current->state != TASK_RUNNING &&
>>> +			(local_clock() - start) < pipe->poll_usec)
>>> +		cpu_relax();
>>> +	schedule_preempt_disabled();
>>> +	preempt_enable();
>>> +
>>>   	finish_wait(&pipe->wait, &wait);
>>>   	pipe_lock(pipe);
>>>   }
>> This will make the current thread always spin and block as it itself does
>> the state change to TASK_RUNNING in finish_wait.
> Nah, the actual wakeup will also do that state change. The one in
> finish_wait() is for the case where the wait condition became true
> without wakeup, such that we don't 'leak' the INTERRUPTIBLE state.
Ok, it works. I see similar improvements with hackbench as the original
patch.