LKML Archive on lore.kernel.org
 help / color / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Josh Don <joshdon@google.com>
Cc: Joel Fernandes <joel@joelfernandes.org>,
	"Hyser,Chris" <chris.hyser@oracle.com>,
	Ingo Molnar <mingo@kernel.org>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	Valentin Schneider <valentin.schneider@arm.com>,
	Mel Gorman <mgorman@suse.de>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	dhiatt@digitalocean.com
Subject: Re: [PATCH 04/19] sched: Prepare for Core-wide rq->lock
Date: Wed, 28 Apr 2021 11:13:33 +0200
Message-ID: <YIknPXxwZvq0qmId@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <CABk29NvaH687GfOm_b5_hJF6HBQ6fu+1hzc0GFNEMv5mj3DrUw@mail.gmail.com>

On Tue, Apr 27, 2021 at 04:30:02PM -0700, Josh Don wrote:

> Also, did you mean to have a preempt_enable_no_resched() rather than
> prempt_enable() in raw_spin_rq_trylock?

No, trylock really needs to be preempt_enable(), because it can have
failed, at which point it will not have incremented the preemption count
and our decrement can hit 0, at which point we really should reschedule.

> I went over the rq_lockp stuff again after Don's reported lockup. Most
> uses are safe due to already holding an rq lock. However,
> double_rq_unlock() is prone to race:
> 
> double_rq_unlock(rq1, rq2):
> /* Initial state: core sched enabled, and rq1 and rq2 are smt
> siblings. So, double_rq_lock(rq1, rq2) only took a single rq lock */
> raw_spin_rq_unlock(rq1);
> /* now not holding any rq lock */
> /* sched core disabled. Now __rq_lockp(rq1) != __rq_lockp(rq2), so we
> falsely unlock rq2 */
> if (__rq_lockp(rq1) != __rq_lockp(rq2))
>         raw_spin_rq_unlock(rq2);
> else
>         __release(rq2->lock);
> 
> Instead we can cache __rq_lockp(rq1) and __rq_lockp(rq2) before
> releasing the lock, in order to prevent this. FWIW I think it is
> likely that Don is seeing a different issue.

Ah, indeed so.. rq_lockp() could do with an assertion, not sure how to
sanely do that. Anyway, double_rq_unlock() is simple enough to fix, we
can simply flip the unlock()s.

( I'm suffering a cold and am really quite slow atm )

How's this then?

---
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index f732642e3e09..3a534c0c1c46 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -290,6 +290,10 @@ static void sched_core_assert_empty(void)
 static void __sched_core_enable(void)
 {
 	static_branch_enable(&__sched_core_enabled);
+	/*
+	 * Ensure raw_spin_rq_*lock*() have completed before flipping.
+	 */
+	synchronize_sched();
 	__sched_core_flip(true);
 	sched_core_assert_empty();
 }
@@ -449,16 +453,23 @@ void raw_spin_rq_lock_nested(struct rq *rq, int subclass)
 {
 	raw_spinlock_t *lock;
 
+	/* Matches synchronize_sched() in __sched_core_enabled() */
+	preempt_disable();
 	if (sched_core_disabled()) {
 		raw_spin_lock_nested(&rq->__lock, subclass);
+		/* preempt-count *MUST* be > 1 */
+		preempt_enable_no_resched();
 		return;
 	}
 
 	for (;;) {
 		lock = __rq_lockp(rq);
 		raw_spin_lock_nested(lock, subclass);
-		if (likely(lock == __rq_lockp(rq)))
+		if (likely(lock == __rq_lockp(rq))) {
+			/* preempt-count *MUST* be > 1 */
+			preempt_enable_no_resched();
 			return;
+		}
 		raw_spin_unlock(lock);
 	}
 }
@@ -468,14 +479,21 @@ bool raw_spin_rq_trylock(struct rq *rq)
 	raw_spinlock_t *lock;
 	bool ret;
 
-	if (sched_core_disabled())
-		return raw_spin_trylock(&rq->__lock);
+	/* Matches synchronize_sched() in __sched_core_enabled() */
+	preempt_disable();
+	if (sched_core_disabled()) {
+		ret = raw_spin_trylock(&rq->__lock);
+		preempt_enable();
+		return ret;
+	}
 
 	for (;;) {
 		lock = __rq_lockp(rq);
 		ret = raw_spin_trylock(lock);
-		if (!ret || (likely(lock == __rq_lockp(rq))))
+		if (!ret || (likely(lock == __rq_lockp(rq)))) {
+			preempt_enable();
 			return ret;
+		}
 		raw_spin_unlock(lock);
 	}
 }
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 6a905fe19eef..c9a52231d58a 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2568,11 +2568,12 @@ static inline void double_rq_unlock(struct rq *rq1, struct rq *rq2)
 	__releases(rq1->lock)
 	__releases(rq2->lock)
 {
-	raw_spin_rq_unlock(rq1);
 	if (__rq_lockp(rq1) != __rq_lockp(rq2))
 		raw_spin_rq_unlock(rq2);
 	else
 		__release(rq2->lock);
+
+	raw_spin_rq_unlock(rq1);
 }
 
 extern void set_rq_online (struct rq *rq);

  reply index

Thread overview: 92+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-22 12:04 [PATCH 00/19] sched: Core Scheduling Peter Zijlstra
2021-04-22 12:05 ` [PATCH 01/19] sched/fair: Add a few assertions Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-05-13  8:56     ` Ning, Hongyu
2021-04-22 12:05 ` [PATCH 02/19] sched: Provide raw_spin_rq_*lock*() helpers Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 03/19] sched: Wrap rq::lock access Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 04/19] sched: Prepare for Core-wide rq->lock Peter Zijlstra
2021-04-24  1:22   ` Josh Don
2021-04-26  8:31     ` Peter Zijlstra
2021-04-26 22:21       ` Josh Don
2021-04-27 17:10         ` Don Hiatt
2021-04-27 23:35           ` Josh Don
2021-04-28  1:03             ` Aubrey Li
2021-04-28  6:05               ` Aubrey Li
2021-04-28 10:57                 ` Aubrey Li
2021-04-28 16:41                   ` Don Hiatt
2021-04-29 20:48                     ` Josh Don
2021-04-29 21:09                       ` Don Hiatt
2021-04-29 23:22                         ` Josh Don
2021-04-30 16:18                           ` Don Hiatt
2021-04-30  8:26                         ` Aubrey Li
2021-04-28 16:04             ` Don Hiatt
2021-04-27 23:30         ` Josh Don
2021-04-28  9:13           ` Peter Zijlstra [this message]
2021-04-28 10:35             ` Aubrey Li
2021-04-28 11:03               ` Peter Zijlstra
2021-04-28 14:18                 ` Paul E. McKenney
2021-04-29 20:11             ` Josh Don
2021-05-03 19:17               ` Peter Zijlstra
2021-04-28  7:13         ` Peter Zijlstra
2021-04-28  6:02   ` Aubrey Li
2021-04-29  8:03   ` Aubrey Li
2021-04-29 20:39     ` Josh Don
2021-04-30  8:20       ` Aubrey Li
2021-04-30  8:48         ` Josh Don
2021-04-30 14:15           ` Aubrey Li
2021-05-04  7:38       ` Peter Zijlstra
2021-05-05 16:20         ` Don Hiatt
2021-05-06 10:25           ` Peter Zijlstra
2021-05-07  9:50   ` [PATCH v2 " Peter Zijlstra
2021-05-08  8:07     ` Aubrey Li
2021-05-12  9:07       ` Peter Zijlstra
2021-04-22 12:05 ` [PATCH 05/19] sched: " Peter Zijlstra
2021-05-07  9:50   ` [PATCH v2 " Peter Zijlstra
2021-05-12 10:28     ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 06/19] sched: Optimize rq_lockp() usage Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 07/19] sched: Allow sched_core_put() from atomic context Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 08/19] sched: Introduce sched_class::pick_task() Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 09/19] sched: Basic tracking of matching tasks Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 10/19] sched: Add core wide task selection and scheduling Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 11/19] sched/fair: Fix forced idle sibling starvation corner case Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Vineeth Pillai
2021-04-22 12:05 ` [PATCH 12/19] sched: Fix priority inversion of cookied task with sibling Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Joel Fernandes (Google)
2021-04-22 12:05 ` [PATCH 13/19] sched/fair: Snapshot the min_vruntime of CPUs on force idle Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Joel Fernandes (Google)
2021-04-22 12:05 ` [PATCH 14/19] sched: Trivial forced-newidle balancer Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 15/19] sched: Migration changes for core scheduling Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Aubrey Li
2021-04-22 12:05 ` [PATCH 16/19] sched: Trivial core scheduling cookie management Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 17/19] sched: Inherit task cookie on fork() Peter Zijlstra
2021-05-10 16:06   ` Joel Fernandes
2021-05-10 16:22     ` Chris Hyser
2021-05-10 20:47       ` Joel Fernandes
2021-05-10 21:38         ` Chris Hyser
2021-05-12  9:05           ` Peter Zijlstra
2021-05-12 20:20             ` Josh Don
2021-05-12 21:07               ` Don Hiatt
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Peter Zijlstra
2021-04-22 12:05 ` [PATCH 18/19] sched: prctl() core-scheduling interface Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Chris Hyser
2021-06-14 23:36   ` [PATCH 18/19] " Josh Don
2021-06-15 11:31     ` Joel Fernandes
2021-04-22 12:05 ` [PATCH 19/19] kselftest: Add test for core sched prctl interface Peter Zijlstra
2021-05-12 10:28   ` [tip: sched/core] " tip-bot2 for Chris Hyser
2021-04-22 16:43 ` [PATCH 00/19] sched: Core Scheduling Don Hiatt
2021-04-22 17:29   ` Peter Zijlstra
2021-04-30  6:47 ` Ning, Hongyu
2021-05-06 10:29   ` Peter Zijlstra
2021-05-06 12:53     ` Ning, Hongyu
2021-05-07 18:02 ` Joel Fernandes
2021-05-10 16:16 ` Vincent Guittot
2021-05-11  7:00   ` Vincent Guittot

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=YIknPXxwZvq0qmId@hirez.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=chris.hyser@oracle.com \
    --cc=dhiatt@digitalocean.com \
    --cc=joel@joelfernandes.org \
    --cc=joshdon@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mgorman@suse.de \
    --cc=mingo@kernel.org \
    --cc=tglx@linutronix.de \
    --cc=valentin.schneider@arm.com \
    --cc=vincent.guittot@linaro.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

LKML Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/lkml/0 lkml/git/0.git
	git clone --mirror https://lore.kernel.org/lkml/1 lkml/git/1.git
	git clone --mirror https://lore.kernel.org/lkml/2 lkml/git/2.git
	git clone --mirror https://lore.kernel.org/lkml/3 lkml/git/3.git
	git clone --mirror https://lore.kernel.org/lkml/4 lkml/git/4.git
	git clone --mirror https://lore.kernel.org/lkml/5 lkml/git/5.git
	git clone --mirror https://lore.kernel.org/lkml/6 lkml/git/6.git
	git clone --mirror https://lore.kernel.org/lkml/7 lkml/git/7.git
	git clone --mirror https://lore.kernel.org/lkml/8 lkml/git/8.git
	git clone --mirror https://lore.kernel.org/lkml/9 lkml/git/9.git
	git clone --mirror https://lore.kernel.org/lkml/10 lkml/git/10.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 lkml lkml/ https://lore.kernel.org/lkml \
		linux-kernel@vger.kernel.org
	public-inbox-index lkml

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-kernel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git