linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Gleixner <tglx@linutronix.de>
To: Peter Zijlstra <peterz@infradead.org>
Cc: LKML <linux-kernel@vger.kernel.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>,
	Davidlohr Bueso <dave@stgolabs.net>
Subject: Re: [patch 62/63] locking/rtmutex: Add adaptive spinwait mechanism
Date: Wed, 04 Aug 2021 19:49:56 +0200	[thread overview]
Message-ID: <87y29h14mz.ffs@tglx> (raw)
In-Reply-To: <20210804123030.GD8057@worktop.programming.kicks-ass.net>

On Wed, Aug 04 2021 at 14:30, Peter Zijlstra wrote:
> On Fri, Jul 30, 2021 at 03:51:09PM +0200, Thomas Gleixner wrote:
>> From: Steven Rostedt <rostedt@goodmis.org>
>> 
>> Going to sleep when a spinlock or rwlock is contended can be quite
>> inefficient when the contention time is short and the lock owner is running
>> on a different CPU. The MCS mechanism is not applicable to rtmutex based
>> locks, so provide a simple adaptive spinwait mechanism for the RT specific
>> spin/rwlock implementations.
>
> A better Changelog would explain *why* OSQ does not apply. I'm thinking
> this ie because the (spin) waiters can be of different priorities and we
> need to ensure the highest prio waiter gets to win?
>
> AFAICT that isn't true even without OSQ, you just get a thundering herd
> and the higher prio waiter has a better chance of winning the race but
> all bets are off either way around.

Will do.

>> +#ifdef CONFIG_SMP
>
> Existing convention would make that:
>
> #ifdef CONFIG_RTMUTEX_SPIN_ON_OWNER
>
> But I suppose that's indeed not required if we don't use OSQ.

Right.

>> +/*
>> + * Note that owner is a speculative pointer and dereferencing relies
>> + * on rcu_read_lock() and the check against the lock owner.
>> + */
>> +static bool rtlock_adaptive_spinwait(struct rt_mutex_base *lock,
>> +				     struct task_struct *owner)
>
> similarly, this would be:
>
>   rt_mutex_spin_on_owner()

Duh.
>
> Esp. when this will be on rtmutex unconditionally, you want to mirror
> the full set of conditions we also have on mutex_spin_on_owner():
>
> 	|| need_resched() || vcpu_is_preempted(task_cpu(owner))

Sure.

>> +			res = false;
>> +			break;
>> +		}
>> +		cpu_relax();
>> +	}
>> +	rcu_read_unlock();
>> +	return res;
>> +}
>
> Additionally, we could consider adding something that would compare the
> current prio to the top_waiter prio and terminate the loop if we're
> found to be of lower prio, but lifetime issues are probably going to
> make that 'interesting'.

It's only the top priority waiter which can spin. If all of them start
spinning then everything goes down the drain.

Thanks,

        tglx

  reply	other threads:[~2021-08-04 17:50 UTC|newest]

Thread overview: 92+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-30 13:50 [patch 00/63] locking, sched: The PREEMPT-RT locking infrastructure Thomas Gleixner
2021-07-30 13:50 ` [patch 01/63] sched: Split out the wakeup state check Thomas Gleixner
2021-07-30 13:50 ` [patch 02/63] sched: Introduce TASK_RTLOCK_WAIT Thomas Gleixner
2021-07-30 13:50 ` [patch 03/63] sched: Prepare for RT sleeping spin/rwlocks Thomas Gleixner
2021-08-01 15:30   ` Mike Galbraith
2021-08-03  9:48     ` Peter Zijlstra
2021-08-03 14:04       ` Thomas Gleixner
2021-08-03 14:51         ` Peter Zijlstra
2021-08-03 20:11           ` Thomas Gleixner
2021-07-30 13:50 ` [patch 04/63] sched: Rework the __schedule() preempt argument Thomas Gleixner
2021-07-30 13:50 ` [patch 05/63] sched: Provide schedule point for RT locks Thomas Gleixner
2021-07-30 13:50 ` [patch 06/63] sched/wake_q: Provide WAKE_Q_HEAD_INITIALIZER Thomas Gleixner
2021-07-30 13:50 ` [patch 07/63] media/atomisp: Use lockdep instead of *mutex_is_locked() Thomas Gleixner
2021-07-30 13:50 ` [patch 08/63] rtmutex: Remove rt_mutex_is_locked() Thomas Gleixner
2021-07-30 13:50 ` [patch 09/63] rtmutex: Convert macros to inlines Thomas Gleixner
2021-07-30 13:50 ` [patch 10/63] rtmutex: Switch to try_cmpxchg() Thomas Gleixner
2021-07-30 13:50 ` [patch 11/63] rtmutex: Split API and implementation Thomas Gleixner
2021-07-30 13:50 ` [patch 12/63] rtmutex: Split out the inner parts of struct rtmutex Thomas Gleixner
2021-07-30 13:50 ` [patch 13/63] locking/rtmutex: Provide rt_mutex_slowlock_locked() Thomas Gleixner
2021-07-30 13:50 ` [patch 14/63] rtmutex: Provide rt_mutex_base_is_locked() Thomas Gleixner
2021-07-30 13:50 ` [patch 15/63] locking: Add base code for RT rw_semaphore and rwlock Thomas Gleixner
2021-08-04 19:37   ` Waiman Long
2021-08-05  9:04     ` Thomas Gleixner
2021-08-05 14:59       ` Waiman Long
2021-07-30 13:50 ` [patch 16/63] locking/rwsem: Add rtmutex based R/W semaphore implementation Thomas Gleixner
2021-07-30 13:50 ` [patch 17/63] locking/rtmutex: Add wake_state to rt_mutex_waiter Thomas Gleixner
2021-07-30 13:50 ` [patch 18/63] locking/rtmutex: Provide rt_wake_q and helpers Thomas Gleixner
2021-07-30 13:50 ` [patch 19/63] locking/rtmutex: Use rt_mutex_wake_q_head Thomas Gleixner
2021-07-30 13:50 ` [patch 20/63] locking/rtmutex: Prepare RT rt_mutex_wake_q for RT locks Thomas Gleixner
2021-07-30 13:50 ` [patch 21/63] locking/rtmutex: Guard regular sleeping locks specific functions Thomas Gleixner
2021-07-30 13:50 ` [patch 22/63] locking/spinlock: Split the lock types header Thomas Gleixner
2021-08-04 21:17   ` Waiman Long
2021-08-05  8:54     ` Thomas Gleixner
2021-07-30 13:50 ` [patch 23/63] locking/rtmutex: Prevent future include recursion hell Thomas Gleixner
2021-07-30 13:50 ` [patch 24/63] locking/lockdep: Reduce includes in debug_locks.h Thomas Gleixner
2021-07-30 13:50 ` [patch 25/63] rbtree: Split out the rbtree type definitions Thomas Gleixner
2021-07-30 13:50 ` [patch 26/63] locking/rtmutex: Include only rbtree types Thomas Gleixner
2021-07-30 13:50 ` [patch 27/63] locking/spinlock: Provide RT specific spinlock type Thomas Gleixner
2021-07-30 13:50 ` [patch 28/63] locking/spinlock: Provide RT variant header Thomas Gleixner
2021-07-30 13:50 ` [patch 29/63] locking/rtmutex: Provide the spin/rwlock core lock function Thomas Gleixner
2021-07-30 13:50 ` [patch 30/63] locking/spinlock: Provide RT variant Thomas Gleixner
2021-08-04 23:34   ` Waiman Long
2021-08-05  8:54     ` Thomas Gleixner
2021-07-30 13:50 ` [patch 31/63] locking/rwlock: " Thomas Gleixner
2021-07-30 13:50 ` [patch 32/63] locking/mutex: Consolidate core headers Thomas Gleixner
2021-07-30 13:50 ` [patch 33/63] locking/mutex: Move waiter to core header Thomas Gleixner
2021-07-30 13:50 ` [patch 34/63] locking/ww_mutex: Move ww_mutex declarations into ww_mutex.h Thomas Gleixner
2021-07-30 13:50 ` [patch 35/63] locking/mutex: Make mutex::wait_lock raw Thomas Gleixner
2021-07-30 13:50 ` [patch 36/63] locking/ww_mutex: Simplify lockdep annotation Thomas Gleixner
2021-07-30 13:50 ` [patch 37/63] locking/ww_mutex: Gather mutex_waiter initialization Thomas Gleixner
2021-07-30 13:50 ` [patch 38/63] locking/ww_mutex: Split up ww_mutex_unlock() Thomas Gleixner
2021-07-30 13:50 ` [patch 39/63] locking/ww_mutex: Split W/W implementation logic Thomas Gleixner
2021-07-30 13:50 ` [patch 40/63] locking/ww_mutex: Remove __sched annotation Thomas Gleixner
2021-07-30 13:50 ` [patch 41/63] locking/ww_mutex: Abstract waiter iteration Thomas Gleixner
2021-07-30 13:50 ` [patch 42/63] locking/ww_mutex: Abstract waiter enqueueing Thomas Gleixner
2021-07-30 13:50 ` [patch 43/63] locking/ww_mutex: Abstract mutex accessors Thomas Gleixner
2021-07-30 13:50 ` [patch 44/63] locking/ww_mutex: Abstract mutex types Thomas Gleixner
2021-07-30 13:50 ` [patch 45/63] locking/ww_mutex: Abstract internal lock access Thomas Gleixner
2021-07-30 13:50 ` [patch 46/63] locking/ww_mutex: Implement rt_mutex accessors Thomas Gleixner
2021-07-30 13:50 ` [patch 47/63] locking/ww_mutex: Add RT priority to W/W order Thomas Gleixner
2021-07-30 13:50 ` [patch 48/63] locking/ww_mutex: Add rt_mutex based lock type and accessors Thomas Gleixner
2021-07-30 13:50 ` [patch 49/63] locking/rtmutex: Extend the rtmutex core to support ww_mutex Thomas Gleixner
2021-07-30 13:50 ` [patch 50/63] locking/ww_mutex: Implement rtmutex based ww_mutex API functions Thomas Gleixner
2021-07-31 13:26   ` Mike Galbraith
2021-08-01 21:18     ` Thomas Gleixner
2021-07-30 13:50 ` [patch 51/63] locking/rtmutex: Add mutex variant for RT Thomas Gleixner
2021-07-30 13:50 ` [patch 52/63] lib/test_lockup: Adapt to changed variables Thomas Gleixner
2021-07-30 13:51 ` [patch 53/63] futex: Validate waiter correctly in futex_proxy_trylock_atomic() Thomas Gleixner
2021-07-30 13:51 ` [patch 54/63] futex: Cleanup stale comments Thomas Gleixner
2021-07-30 13:51 ` [patch 55/63] futex: Correct the number of requeued waiters for PI Thomas Gleixner
2021-07-30 13:51 ` [patch 56/63] futex: Restructure futex_requeue() Thomas Gleixner
2021-07-30 13:51 ` [patch 57/63] futex: Clarify comment in futex_requeue() Thomas Gleixner
2021-07-30 13:51 ` [patch 58/63] futex: Prevent requeue_pi() lock nesting issue on RT Thomas Gleixner
2021-08-02 12:56   ` Peter Zijlstra
2021-08-02 13:10     ` Peter Zijlstra
2021-08-02 14:35       ` Thomas Gleixner
2021-08-02 14:34     ` Thomas Gleixner
2021-08-03 10:28     ` Peter Zijlstra
2021-08-03 21:10       ` Thomas Gleixner
2021-08-03 10:07   ` Peter Zijlstra
2021-08-03 21:10     ` Thomas Gleixner
2021-08-03 11:20   ` Peter Zijlstra
2021-08-03 21:22     ` Thomas Gleixner
2021-07-30 13:51 ` [patch 59/63] rtmutex: Prevent lockdep false positive with PI futexes Thomas Gleixner
2021-07-30 13:51 ` [patch 60/63] preempt: Adjust PREEMPT_LOCK_OFFSET for RT Thomas Gleixner
2021-07-30 13:51 ` [patch 61/63] locking/rtmutex: Implement equal priority lock stealing Thomas Gleixner
2021-07-30 13:51 ` [patch 62/63] locking/rtmutex: Add adaptive spinwait mechanism Thomas Gleixner
2021-08-04 12:30   ` Peter Zijlstra
2021-08-04 17:49     ` Thomas Gleixner [this message]
2021-07-30 13:51 ` [patch 63/63] locking/rtmutex: Use adaptive spinwait for all rtmutex based locks Thomas Gleixner
     [not found] ` <20210803063217.2325-1-hdanton@sina.com>
2021-08-03  9:10   ` [patch 30/63] locking/spinlock: Provide RT variant Thomas Gleixner
2021-08-03 12:37 ` [patch 00/63] locking, sched: The PREEMPT-RT locking infrastructure Daniel Bristot de Oliveira

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87y29h14mz.ffs@tglx \
    --to=tglx@linutronix.de \
    --cc=bigeasy@linutronix.de \
    --cc=boqun.feng@gmail.com \
    --cc=bristot@redhat.com \
    --cc=dave@stgolabs.net \
    --cc=juri.lelli@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=longman@redhat.com \
    --cc=mingo@kernel.org \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=will@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).