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;
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Davidlohr Bueso <dave@stgolabs.net>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@kernel.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        Will Deacon <will@kernel.org>,
        Waiman Long <longman@redhat.com>,
        Boqun Feng <boqun.feng@gmail.com>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Mike Galbraith <efault@gmx.de>
Subject: Re: [patch V3 56/64] futex: Correct the number of requeued waiters
 for PI
In-Reply-To: <20210808170535.kotqd5t677tijh4o@offworld>
References: <20210805151300.330412127@linutronix.de>
 <20210805153956.051961951@linutronix.de>
 <20210808170535.kotqd5t677tijh4o@offworld>
Date:   Mon, 09 Aug 2021 10:18:22 +0200
Message-ID: <87o8a7t4j5.ffs@tglx>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On Sun, Aug 08 2021 at 10:05, Davidlohr Bueso wrote:
> On Thu, 05 Aug 2021, Thomas Gleixner wrote:
>
>>From: Thomas Gleixner <tglx@linutronix.de>
>>
>>The accounting is wrong when either the PI sanity check or the
>>requeue PI operation fails. Adjust it in the failure path.
>
> Ok fortunately these accounting errors are benign considering they
> are in error paths. This also made me wonder about the requeue PI
> top-waiter wakeup from futex_proxy_trylock_atomic(), which is always
> required with nr_wakers == 1. We account for it on the successful
> case we acquired the lock on it's behalf (and thus requeue_pi_wake_futex
> was called), but if the corresponding lookup_pi_state fails, we'll retry.
> So, shouldn't the task_count++ only be considered when we know the
> requeueing is next (a successful top_waiter acquiring the lock+pi state)?
>
> @@ -2260,7 +2260,6 @@ static int futex_requeue(u32 __user *uaddr1, unsigned int flags,
> 		 */
> 		if (ret > 0) {
> 			WARN_ON(pi_state);
> -                       task_count++;
> 			/*
> 			 * If we acquired the lock, then the user space value
> 			 * of uaddr2 should be vpid. It cannot be changed by
> @@ -2275,6 +2274,8 @@ static int futex_requeue(u32 __user *uaddr1, unsigned int flags,
> 			 */
> 			ret = lookup_pi_state(uaddr2, ret, hb2, &key2,
> 					      &pi_state, &exiting);
> +                       if (!ret)
> +                               task_count++;
> 		}

Yes, but if futex_proxy_trylock_atomic() succeeds and lookup_pi_state()
fails then the user space futex value is already the VPID of the top
waiter and a retry will then fail the futex_proxy_trylock_atomic().

> 		switch (ret) {
>
> Also, looking at the code, I think we can use the goto retry_private
> optimization for private futexes upon futex_proxy_trylock_atomic
> lookup_pi_state errors:
>
> @@ -2290,8 +2290,11 @@ static int futex_requeue(u32 __user *uaddr1, unsigned int flags,
> 			double_unlock_hb(hb1, hb2);
> 			hb_waiters_dec(hb2);
> 			ret = fault_in_user_writeable(uaddr2);
> -                       if (!ret)
> +                       if (!ret) {
> +                               if (!(flags & FLAGS_SHARED))
> +                                       goto retry_private;
> 				goto retry;
> +                       }

Yes, we can, but let me stare at that code some more.

Thanks,

        tglx