All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>,
	tj@kernel.org, mingo@redhat.com, linux-kernel@vger.kernel.org,
	der.herr@hofr.at, dave@stgolabs.net, riel@redhat.com,
	viro@ZenIV.linux.org.uk, torvalds@linux-foundation.org
Subject: Re: [RFC][PATCH 12/13] stop_machine: Remove lglock
Date: Wed, 24 Jun 2015 20:14:44 +0200	[thread overview]
Message-ID: <20150624181444.GL18673@twins.programming.kicks-ass.net> (raw)
In-Reply-To: <20150624172818.GQ3644@twins.programming.kicks-ass.net>

On Wed, Jun 24, 2015 at 07:28:18PM +0200, Peter Zijlstra wrote:
> How about something like this, it replaced mutex and start/done ticket
> thing with an MCS style lockless FIFO queue.
> 
> I further uses the gpnum/completed thing to short circuit things if
> we've waited long enough.

Prettier version

--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -3253,23 +3253,41 @@ void cond_synchronize_rcu(unsigned long
 }
 EXPORT_SYMBOL_GPL(cond_synchronize_rcu);
 
+struct expedited_task_state {
+	struct expedited_task_state *next;
+	struct task_struct *task;
+	atomic_t count;
+	int done;
+};
+
 static int synchronize_sched_expedited_cpu_stop(void *data)
 {
+	struct expedited_task_state *ets = data;
+
 	/*
 	 * There must be a full memory barrier on each affected CPU
 	 * between the time that try_stop_cpus() is called and the
 	 * time that it returns.
-	 *
-	 * In the current initial implementation of cpu_stop, the
-	 * above condition is already met when the control reaches
-	 * this point and the following smp_mb() is not strictly
-	 * necessary.  Do smp_mb() anyway for documentation and
-	 * robustness against future implementation changes.
 	 */
-	smp_mb(); /* See above comment block. */
+	if (atomic_dec_and_test(&ets->count))
+		wake_up_process(ets->task);
+
 	return 0;
 }
 
+static DEFINE_PER_CPU(struct cpu_stop_work, exp_stop_work);
+
+#define current_wait(cond)					\
+do {								\
+	for (;;) {						\
+		set_current_state(TASK_UNINTERRUPTIBLE);	\
+		if (cond)					\
+			break;					\
+		schedule();					\
+	}							\
+	__set_current_state(TASK_RUNNING);			\
+} while (0)
+
 /**
  * synchronize_sched_expedited - Brute-force RCU-sched grace period
  *
@@ -3304,138 +3322,71 @@ static int synchronize_sched_expedited_c
  */
 void synchronize_sched_expedited(void)
 {
-	cpumask_var_t cm;
-	bool cma = false;
-	int cpu;
-	long firstsnap, s, snap;
-	int trycount = 0;
 	struct rcu_state *rsp = &rcu_sched_state;
+	struct expedited_task_state *prev, *next, entry = {
+		.task = current,
+		.count = ATOMIC_INIT(1), /* avoid spurious wakeups */
+	};
+	long gpnum;
+	int cpu;
 
-	/*
-	 * If we are in danger of counter wrap, just do synchronize_sched().
-	 * By allowing sync_sched_expedited_started to advance no more than
-	 * ULONG_MAX/8 ahead of sync_sched_expedited_done, we are ensuring
-	 * that more than 3.5 billion CPUs would be required to force a
-	 * counter wrap on a 32-bit system.  Quite a few more CPUs would of
-	 * course be required on a 64-bit system.
-	 */
-	if (ULONG_CMP_GE((ulong)atomic_long_read(&rsp->expedited_start),
-			 (ulong)atomic_long_read(&rsp->expedited_done) +
-			 ULONG_MAX / 8)) {
-		wait_rcu_gp(call_rcu_sched);
-		atomic_long_inc(&rsp->expedited_wrap);
-		return;
-	}
-
-	/*
-	 * Take a ticket.  Note that atomic_inc_return() implies a
-	 * full memory barrier.
-	 */
-	snap = atomic_long_inc_return(&rsp->expedited_start);
-	firstsnap = snap;
 	if (!try_get_online_cpus()) {
 		/* CPU hotplug operation in flight, fall back to normal GP. */
 		wait_rcu_gp(call_rcu_sched);
-		atomic_long_inc(&rsp->expedited_normal);
 		return;
 	}
 	WARN_ON_ONCE(cpu_is_offline(raw_smp_processor_id()));
 
-	/* Offline CPUs, idle CPUs, and any CPU we run on are quiescent. */
-	cma = zalloc_cpumask_var(&cm, GFP_KERNEL);
-	if (cma) {
-		cpumask_copy(cm, cpu_online_mask);
-		cpumask_clear_cpu(raw_smp_processor_id(), cm);
-		for_each_cpu(cpu, cm) {
-			struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);
-
-			if (!(atomic_add_return(0, &rdtp->dynticks) & 0x1))
-				cpumask_clear_cpu(cpu, cm);
-		}
-		if (cpumask_weight(cm) == 0)
-			goto all_cpus_idle;
+	smp_mb();
+	gpnum = smp_load_acquire(&rsp->gpnum);
+
+	/* MCS style queue 'lock' */
+	prev = xchg(&rsp->expedited_queue, &entry);
+	if (prev) {
+		WRITE_ONCE(prev->next, &entry);
+		current_wait(smp_load_acquire(&entry.done));
 	}
 
 	/*
-	 * Each pass through the following loop attempts to force a
-	 * context switch on each CPU.
+	 * Check to see if someone else did our work for us, while we were
+	 * waiting on the queue.
 	 */
-	while (try_stop_cpus(cma ? cm : cpu_online_mask,
-			     synchronize_sched_expedited_cpu_stop,
-			     NULL) == -EAGAIN) {
-		put_online_cpus();
-		atomic_long_inc(&rsp->expedited_tryfail);
-
-		/* Check to see if someone else did our work for us. */
-		s = atomic_long_read(&rsp->expedited_done);
-		if (ULONG_CMP_GE((ulong)s, (ulong)firstsnap)) {
-			/* ensure test happens before caller kfree */
-			smp_mb__before_atomic(); /* ^^^ */
-			atomic_long_inc(&rsp->expedited_workdone1);
-			free_cpumask_var(cm);
-			return;
-		}
-
-		/* No joy, try again later.  Or just synchronize_sched(). */
-		if (trycount++ < 10) {
-			udelay(trycount * num_online_cpus());
-		} else {
-			wait_rcu_gp(call_rcu_sched);
-			atomic_long_inc(&rsp->expedited_normal);
-			free_cpumask_var(cm);
-			return;
-		}
-
-		/* Recheck to see if someone else did our work for us. */
-		s = atomic_long_read(&rsp->expedited_done);
-		if (ULONG_CMP_GE((ulong)s, (ulong)firstsnap)) {
-			/* ensure test happens before caller kfree */
-			smp_mb__before_atomic(); /* ^^^ */
-			atomic_long_inc(&rsp->expedited_workdone2);
-			free_cpumask_var(cm);
-			return;
-		}
-
-		/*
-		 * Refetching sync_sched_expedited_started allows later
-		 * callers to piggyback on our grace period.  We retry
-		 * after they started, so our grace period works for them,
-		 * and they started after our first try, so their grace
-		 * period works for us.
-		 */
-		if (!try_get_online_cpus()) {
-			/* CPU hotplug operation in flight, use normal GP. */
-			wait_rcu_gp(call_rcu_sched);
-			atomic_long_inc(&rsp->expedited_normal);
-			free_cpumask_var(cm);
-			return;
-		}
-		snap = atomic_long_read(&rsp->expedited_start);
-		smp_mb(); /* ensure read is before try_stop_cpus(). */
+	if (ULONG_CMP_LT(gpnum, smp_load_acquire(&rsp->completed)))
+		goto unlock;
+
+	/* Stop each CPU that is online, non-idle, and not us. */
+	for_each_online_cpu(cpu) {
+		struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);
+
+		/* Skip our CPU, */
+		if (raw_smp_processor_id() == cpu)
+			continue;
+
+		/* and any idle CPUs. */
+		if (!(atomic_add_return(0, &rdtp->dynticks) & 0x1))
+			continue;
+
+		atomic_inc(&entry.count);
+		stop_one_cpu_nowait(cpu, synchronize_sched_expedited_cpu_stop,
+				&entry, &per_cpu(exp_stop_work, cpu));
 	}
-	atomic_long_inc(&rsp->expedited_stoppedcpus);
 
-all_cpus_idle:
-	free_cpumask_var(cm);
+	atomic_dec(&entry.count); /* let the wakeups in */
+	current_wait(!atomic_read(&entry.count));
 
-	/*
-	 * Everyone up to our most recent fetch is covered by our grace
-	 * period.  Update the counter, but only if our work is still
-	 * relevant -- which it won't be if someone who started later
-	 * than we did already did their update.
-	 */
-	do {
-		atomic_long_inc(&rsp->expedited_done_tries);
-		s = atomic_long_read(&rsp->expedited_done);
-		if (ULONG_CMP_GE((ulong)s, (ulong)snap)) {
-			/* ensure test happens before caller kfree */
-			smp_mb__before_atomic(); /* ^^^ */
-			atomic_long_inc(&rsp->expedited_done_lost);
-			break;
-		}
-	} while (atomic_long_cmpxchg(&rsp->expedited_done, s, snap) != s);
-	atomic_long_inc(&rsp->expedited_done_exit);
+unlock:
+	/* MCS style queue 'unlock' */
+	next = READ_ONCE(entry.next);
+	if (!next) {
+		if (cmpxchg(&rsp->expedited_queue, &entry, NULL) == &entry)
+			goto done;
+		while (!(next = READ_ONCE(entry.next)))
+			cpu_relax();
+	}
+	smp_store_release(&next->done, 1);
+	wake_up_process(next->task);
 
+done:
 	put_online_cpus();
 }
 EXPORT_SYMBOL_GPL(synchronize_sched_expedited);
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -483,17 +483,7 @@ struct rcu_state {
 						/*  _rcu_barrier(). */
 	/* End of fields guarded by barrier_mutex. */
 
-	atomic_long_t expedited_start;		/* Starting ticket. */
-	atomic_long_t expedited_done;		/* Done ticket. */
-	atomic_long_t expedited_wrap;		/* # near-wrap incidents. */
-	atomic_long_t expedited_tryfail;	/* # acquisition failures. */
-	atomic_long_t expedited_workdone1;	/* # done by others #1. */
-	atomic_long_t expedited_workdone2;	/* # done by others #2. */
-	atomic_long_t expedited_normal;		/* # fallbacks to normal. */
-	atomic_long_t expedited_stoppedcpus;	/* # successful stop_cpus. */
-	atomic_long_t expedited_done_tries;	/* # tries to update _done. */
-	atomic_long_t expedited_done_lost;	/* # times beaten to _done. */
-	atomic_long_t expedited_done_exit;	/* # times exited _done loop. */
+	void *expedited_queue;
 
 	unsigned long jiffies_force_qs;		/* Time at which to invoke */
 						/*  force_quiescent_state(). */

  parent reply	other threads:[~2015-06-24 18:14 UTC|newest]

Thread overview: 106+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-06-22 12:16 [RFC][PATCH 00/13] percpu rwsem -v2 Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 01/13] rcu: Create rcu_sync infrastructure Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 02/13] rcusync: Introduce struct rcu_sync_ops Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 03/13] rcusync: Add the CONFIG_PROVE_RCU checks Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 04/13] rcusync: Introduce rcu_sync_dtor() Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 05/13] percpu-rwsem: Optimize readers and reduce global impact Peter Zijlstra
2015-06-22 23:02   ` Oleg Nesterov
2015-06-23  7:28   ` Nicholas Mc Guire
2015-06-25 19:08     ` Peter Zijlstra
2015-06-25 19:17       ` Tejun Heo
2015-06-29  9:32         ` Peter Zijlstra
2015-06-29 15:12           ` Tejun Heo
2015-06-29 15:14             ` Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 06/13] percpu-rwsem: Provide percpu_down_read_trylock() Peter Zijlstra
2015-06-22 23:08   ` Oleg Nesterov
2015-06-22 12:16 ` [RFC][PATCH 07/13] sched: Reorder task_struct Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 08/13] percpu-rwsem: DEFINE_STATIC_PERCPU_RWSEM Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 09/13] hotplug: Replace hotplug lock with percpu-rwsem Peter Zijlstra
2015-06-22 22:57   ` Oleg Nesterov
2015-06-23  7:16     ` Peter Zijlstra
2015-06-23 17:01       ` Oleg Nesterov
2015-06-23 17:53         ` Peter Zijlstra
2015-06-24 13:50           ` Oleg Nesterov
2015-06-24 14:13             ` Peter Zijlstra
2015-06-24 15:12               ` Oleg Nesterov
2015-06-24 16:15                 ` Peter Zijlstra
2015-06-28 23:56             ` [PATCH 0/3] percpu-rwsem: introduce percpu_rw_semaphore->recursive mode Oleg Nesterov
2015-06-28 23:56               ` [PATCH 1/3] rcusync: introduce rcu_sync_struct->exclusive mode Oleg Nesterov
2015-06-28 23:56               ` [PATCH 2/3] percpu-rwsem: don't use percpu_rw_semaphore->rw_sem to exclude writers Oleg Nesterov
2015-06-28 23:56               ` [PATCH 3/3] percpu-rwsem: introduce percpu_rw_semaphore->recursive mode Oleg Nesterov
2015-06-22 12:16 ` [RFC][PATCH 10/13] fs/locks: Replace lg_global with a percpu-rwsem Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 11/13] fs/locks: Replace lg_local with a per-cpu spinlock Peter Zijlstra
2015-06-23  0:19   ` Oleg Nesterov
2015-06-22 12:16 ` [RFC][PATCH 12/13] stop_machine: Remove lglock Peter Zijlstra
2015-06-22 22:21   ` Oleg Nesterov
2015-06-23 10:09     ` Peter Zijlstra
2015-06-23 10:55       ` Peter Zijlstra
2015-06-23 11:20         ` Peter Zijlstra
2015-06-23 13:08           ` Peter Zijlstra
2015-06-23 16:36             ` Oleg Nesterov
2015-06-23 17:30             ` Paul E. McKenney
2015-06-23 18:04               ` Peter Zijlstra
2015-06-23 18:26                 ` Paul E. McKenney
2015-06-23 19:05                   ` Paul E. McKenney
2015-06-24  2:23                     ` Paul E. McKenney
2015-06-24  8:32                       ` Peter Zijlstra
2015-06-24  9:31                         ` Peter Zijlstra
2015-06-24 13:48                           ` Paul E. McKenney
2015-06-24 15:01                         ` Paul E. McKenney
2015-06-24 15:34                           ` Peter Zijlstra
2015-06-24  7:35                   ` Peter Zijlstra
2015-06-24  8:42                     ` Ingo Molnar
2015-06-24 13:39                       ` Paul E. McKenney
2015-06-24 13:43                         ` Ingo Molnar
2015-06-24 14:03                           ` Paul E. McKenney
2015-06-24 14:50                     ` Paul E. McKenney
2015-06-24 15:01                       ` Peter Zijlstra
2015-06-24 15:27                         ` Paul E. McKenney
2015-06-24 15:40                           ` Peter Zijlstra
2015-06-24 16:09                             ` Paul E. McKenney
2015-06-24 16:42                               ` Peter Zijlstra
2015-06-24 17:10                                 ` Paul E. McKenney
2015-06-24 17:20                                   ` Paul E. McKenney
2015-06-24 17:29                                     ` Peter Zijlstra
2015-06-24 17:28                                   ` Peter Zijlstra
2015-06-24 17:32                                     ` Peter Zijlstra
2015-06-24 18:14                                     ` Peter Zijlstra [this message]
2015-06-24 17:58                                   ` Peter Zijlstra
2015-06-25  3:23                                     ` Paul E. McKenney
2015-06-25 11:07                                       ` Peter Zijlstra
2015-06-25 13:47                                         ` Paul E. McKenney
2015-06-25 14:20                                           ` Peter Zijlstra
2015-06-25 14:51                                             ` Paul E. McKenney
2015-06-26 12:32                                               ` Peter Zijlstra
2015-06-26 16:14                                                 ` Paul E. McKenney
2015-06-29  7:56                                                   ` Peter Zijlstra
2015-06-30 21:32                                                     ` Paul E. McKenney
2015-07-01 11:56                                                       ` Peter Zijlstra
2015-07-01 15:56                                                         ` Paul E. McKenney
2015-07-01 16:16                                                           ` Peter Zijlstra
2015-07-01 18:45                                                             ` Paul E. McKenney
2015-06-23 14:39         ` Paul E. McKenney
2015-06-23 16:20       ` Oleg Nesterov
2015-06-23 17:24         ` Oleg Nesterov
2015-06-25 19:18           ` Peter Zijlstra
2015-06-22 12:16 ` [RFC][PATCH 13/13] locking: " Peter Zijlstra
2015-06-22 12:36 ` [RFC][PATCH 00/13] percpu rwsem -v2 Peter Zijlstra
2015-06-22 18:11 ` Daniel Wagner
2015-06-22 19:05   ` Peter Zijlstra
2015-06-23  9:35     ` Daniel Wagner
2015-06-23 10:00       ` Ingo Molnar
2015-06-23 14:34       ` Peter Zijlstra
2015-06-23 14:56         ` Daniel Wagner
2015-06-23 17:50           ` Peter Zijlstra
2015-06-23 19:36             ` Peter Zijlstra
2015-06-24  8:46               ` Ingo Molnar
2015-06-24  9:01                 ` Peter Zijlstra
2015-06-24  9:18                 ` Daniel Wagner
2015-07-01  5:57                   ` Daniel Wagner
2015-07-01 21:54                     ` Linus Torvalds
2015-07-02  9:41                       ` Peter Zijlstra
2015-07-20  5:53                         ` Daniel Wagner
2015-07-20 18:44                           ` Linus Torvalds
2015-06-22 20:06 ` Linus Torvalds
2015-06-23 16:10 ` Davidlohr Bueso
2015-06-23 16:21   ` Peter Zijlstra

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=20150624181444.GL18673@twins.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=dave@stgolabs.net \
    --cc=der.herr@hofr.at \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=oleg@redhat.com \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=riel@redhat.com \
    --cc=tj@kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@ZenIV.linux.org.uk \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.