All of lore.kernel.org
 help / color / mirror / Atom feed
From: Valentin Schneider <valentin.schneider@arm.com>
To: linux-kernel@vger.kernel.org
Cc: Will Deacon <will@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>,
	Morten Rasmussen <morten.rasmussen@arm.com>,
	Qais Yousef <qais.yousef@arm.com>,
	Dietmar Eggemann <dietmar.eggemann@arm.com>,
	Quentin Perret <qperret@google.com>,
	Juri Lelli <juri.lelli@redhat.com>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	kernel-team@android.com
Subject: [PATCH 2/2] sched: Plug race between SCA, hotplug and migration_cpu_stop()
Date: Wed, 26 May 2021 21:57:51 +0100	[thread overview]
Message-ID: <20210526205751.842360-3-valentin.schneider@arm.com> (raw)
In-Reply-To: <20210526205751.842360-1-valentin.schneider@arm.com>

migration_cpu_stop() is handed a CPU that was a valid pick at the time
set_cpus_allowed_ptr() was invoked. However, hotplug may have happened
between then and the stopper work being executed, meaning
__move_queued_task() could fail when passed arg.dest_cpu. This could mean
leaving a task on its current CPU (despite it being outside the task's
affinity) up until its next wakeup, which is a big no-no.

Verify the validity of the pick in the stopper callback, and invoke
select_fallback_rq() there if need be.

Fixes: 6d337eab041d ("sched: Fix migrate_disable() vs set_cpus_allowed_ptr()")
Reported-by: Will Deacon <will@kernel.org>
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
---
 kernel/sched/core.c | 36 ++++++++++++++++++++++++------------
 1 file changed, 24 insertions(+), 12 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index d40511c55e80..0679efb6c22d 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2263,6 +2263,8 @@ static struct rq *__migrate_task(struct rq *rq, struct rq_flags *rf,
 	return rq;
 }
 
+static int select_fallback_rq(int cpu, struct task_struct *p);
+
 /*
  * migration_cpu_stop - this will be executed by a highprio stopper thread
  * and performs thread migration by bumping thread off CPU then
@@ -2276,6 +2278,7 @@ static int migration_cpu_stop(void *data)
 	struct rq *rq = this_rq();
 	bool complete = false;
 	struct rq_flags rf;
+	int dest_cpu;
 
 	/*
 	 * The original target CPU might have gone down and we might
@@ -2315,18 +2318,27 @@ static int migration_cpu_stop(void *data)
 				goto out;
 		}
 
-		if (task_on_rq_queued(p))
-			rq = __migrate_task(rq, &rf, p, arg->dest_cpu);
-		else
-			p->wake_cpu = arg->dest_cpu;
-
-		/*
-		 * XXX __migrate_task() can fail, at which point we might end
-		 * up running on a dodgy CPU, AFAICT this can only happen
-		 * during CPU hotplug, at which point we'll get pushed out
-		 * anyway, so it's probably not a big deal.
-		 */
-
+		dest_cpu = arg->dest_cpu;
+		if (task_on_rq_queued(p)) {
+			/*
+			 * A hotplug operation could have happened between
+			 * set_cpus_allowed_ptr() and here, making dest_cpu no
+			 * longer allowed.
+			 */
+			if (!is_cpu_allowed(p, dest_cpu))
+				dest_cpu = select_fallback_rq(cpu_of(rq), p);
+			/*
+			 * dest_cpu can be victim of hotplug between is_cpu_allowed()
+			 * and here. However, per the synchronize_rcu() in
+			 * sched_cpu_deactivate(), it can't have gone lower than
+			 * CPUHP_AP_ACTIVE, so it's safe to punt it over and let
+			 * balance_push() route it elsewhere.
+			 */
+			update_rq_clock(rq);
+			rq = move_queued_task(rq, &rf, p, dest_cpu);
+		} else {
+			p->wake_cpu = dest_cpu;
+		}
 	} else if (pending) {
 		/*
 		 * This happens when we get migrated between migrate_enable()'s
-- 
2.25.1


  parent reply	other threads:[~2021-05-26 20:58 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-26 20:57 [PATCH 0/2] sched: SCA vs hotplug vs stopper races fixes Valentin Schneider
2021-05-26 20:57 ` [PATCH 1/2] sched: Don't defer CPU pick to migration_cpu_stop() Valentin Schneider
2021-06-01 14:04   ` [tip: sched/core] " tip-bot2 for Valentin Schneider
2021-05-26 20:57 ` Valentin Schneider [this message]
2021-05-27  9:12   ` [PATCH 2/2] sched: Plug race between SCA, hotplug and migration_cpu_stop() Peter Zijlstra
2021-06-01 16:59   ` Valentin Schneider
2021-06-02 15:26     ` Will Deacon
2021-06-02 19:43       ` Valentin Schneider
2021-06-22 13:51         ` Will Deacon
2021-06-24  9:33           ` Valentin Schneider

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=20210526205751.842360-3-valentin.schneider@arm.com \
    --to=valentin.schneider@arm.com \
    --cc=dietmar.eggemann@arm.com \
    --cc=juri.lelli@redhat.com \
    --cc=kernel-team@android.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=morten.rasmussen@arm.com \
    --cc=peterz@infradead.org \
    --cc=qais.yousef@arm.com \
    --cc=qperret@google.com \
    --cc=vincent.guittot@linaro.org \
    --cc=will@kernel.org \
    --subject='Re: [PATCH 2/2] sched: Plug race between SCA, hotplug and migration_cpu_stop()' \
    /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

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.