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:   Tue, 9 Mar 2021 18:15:39 +0100
From:   Oleg Nesterov <oleg@redhat.com>
To:     Jim Newsome <jnewsome@torproject.org>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        "Eric W . Biederman" <ebiederm@xmission.com>,
        Christian Brauner <christian@brauner.io>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] do_wait: make PIDTYPE_PID case O(1) instead of O(n)
Message-ID: <20210309171539.GA32475@redhat.com>
References: <20210309161548.18786-1-jnewsome@torproject.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20210309161548.18786-1-jnewsome@torproject.org>
User-Agent: Mutt/1.5.24 (2015-08-30)
Precedence: bulk

Jim,

Thanks, the patch looks good to me. Yet I think you need to send V3 even
if I personally do not care ;) Please consider ./scripts/checkpatch.pl,
it reports all the coding-style problems I was going to mention.

I too have a couple of cosmetic nits, but feel free to ignore, this is
subjective.

On 03/09, Jim Newsome wrote:
>
> do_wait is an internal function used to implement waitpid, waitid,
> wait4, etc. To handle the general case, it does an O(n) linear scan of
> the thread group's children and tracees.
> 
> This patch adds a special-case when waiting on a pid to skip these scans
> and instead do an O(1) lookup. This improves performance when waiting on
> a pid from a thread group with many children and/or tracees.
> 
> Signed-off-by: James Newsome
> ---
>  kernel/exit.c | 54 ++++++++++++++++++++++++++++++++++++++++++---------
>  1 file changed, 45 insertions(+), 9 deletions(-)
> 
> diff --git a/kernel/exit.c b/kernel/exit.c
> index 04029e35e69a..312c4dfc9555 100644
> --- a/kernel/exit.c
> +++ b/kernel/exit.c
> @@ -1439,6 +1439,33 @@ void __wake_up_parent(struct task_struct *p, struct task_struct *parent)
>  			   TASK_INTERRUPTIBLE, p);
>  }
>  
> +// Optimization for waiting on PIDTYPE_PID. No need to iterate through child
> +// and tracee lists to find the target task.
> +static int do_wait_pid(struct wait_opts *wo, struct task_struct *tsk)
> +{
> +	struct task_struct *target = pid_task(wo->wo_pid, PIDTYPE_PID);
> +	if (!target) {
> +		return 0;
> +	}
> +	if (tsk == target->real_parent ||
> +	    (!(wo->wo_flags & __WNOTHREAD) &&
> +	     same_thread_group(tsk, target->real_parent))) {
> +		int retval = wait_consider_task(wo, /* ptrace= */ 0, target);
> +		if (retval) {
> +			return retval;
> +		}
> +	}
> +	if (target->ptrace && (tsk == target->parent ||
> +			       (!(wo->wo_flags & __WNOTHREAD) &&
> +				same_thread_group(tsk, target->parent)))) {
> +		int retval = wait_consider_task(wo, /* ptrace= */ 1, target);
> +		if (retval) {
> +			return retval;
> +		}
> +	}

Both if's use "int retval", to me it would be better to declare this variable
at the start of do_wait_pid(). But again, I won't insist this is up to you.

I am wondering if something like

	static inline bool is_parent(struct task_struct *tsk, struct task_struct *p, int flags)
	{
		return	tsk == p || !(flags & __WNOTHREAD)) && same_thread_group(tsk, p);
	}

makes any sense to make do_wait_pid() more clear... probably not.

Oleg.