The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Mathieu Desnoyers" <mathieu.desnoyers@efficios.com>
Cc: linux-rt-devel@lists.linux.dev, tj@kernel.org,
	hannes@cmpxchg.org, mkoutny@suse.com,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v7 27/40] sched/deadline: Move a task's bandwidth to the root domain it runs in
Date: Tue, 17 Nov 2026 15:29:57 +0000	[thread overview]
Message-ID: <sashiko-outbox-219077@kernel.org> (raw)
In-Reply-To: <3292905cac4b511f515b4d353f02a41d747b8015.1794922146.git.mathieu.desnoyers@efficios.com>

Thank you for your contribution! Sashiko AI review found 3 potential
issues to consider:
- [Medium] sched/deadline: dl_bw_rebuilding() drops bw moves during
  rebuild task walk
- [Medium] sched/deadline: migrate_task_rq_dl() drops bw move for
  mutex-blocked tasks
- [Medium] sched/deadline: dl_task_move_bw() moves departed bw of
  PI-boosted tasks
--

commit 00a2592b3e4a7ff7a78ebc503a680dacf59e846b
Author: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

sched/deadline: Move a task's bandwidth to the root domain it runs in

This commit moves a deadline task's bandwidth to the root domain of its
new CPU in migrate_task_rq_dl() and task_woken_dl() when it differs from
the root domain that currently counts it, while suppressing moves during
sched domain rebuilds and proxy execution migrations.

> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 9ff79141c6306..6c6a4d9cd0548 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c

[ ... ]

> @@ -2930,11 +2968,86 @@ select_task_rq_dl(struct task_struct *p, int cpu, int flags)
>  	return cpu;
>  }
>  
> -static void migrate_task_rq_dl(struct task_struct *p, int new_cpu __maybe_unused)
> +/*
> + * Return true if the root domain of @cpu may not count the bandwidth of
> + * @p. Either another root domain counts it, or a rebuild left @p out.
> + * Test without dl_departed_lock, so that a move inside a root domain
> + * takes no lock. dl_task_move_bw() tests again under it.
> + *
> + * Return false if @cpu is not active. A migration-disabled task can wake
> + * up on a CPU that goes offline, after it joined the default root domain.
> + * Keep the bandwidth where it is counted. balance_push() moves the task
> + * off that CPU once it enables migration, or its next wakeup does once
> + * the CPU is offline. That migration moves the bandwidth if needed.
> + *
> + * Return false while the sched domains are rebuilt. The CPUs then move
> + * to their new root domains one by one. That alone can make the two root
> + * domains differ. The rebuild of the accounting
> + * that follows counts @p.
> + */
> +static bool dl_task_bw_elsewhere(struct task_struct *p, int cpu)
> +{
> +	int bw_cpu = READ_ONCE(p->dl.dl_bw_cpu);
> +
> +	if (bw_cpu < 0 || !p->dl.dl_bw || !cpu_active(cpu) || dl_bw_rebuilding())
> +		return false;
> +	return READ_ONCE(p->dl.dl_bw_gen) != READ_ONCE(dl_bw_gen) ||
> +	       cpu_rq(bw_cpu)->rd != cpu_rq(cpu)->rd;
> +}
> +
> +/*
> + * Move the bandwidth of @p to the root domain of @new_cpu. Count it there
> + * also if a rebuild left @p out. Add it even without room, since @p runs
> + * there. The caller holds the rq lock of @p, or its pi_lock for a wakeup.
> + */
> +static void dl_task_move_bw(struct task_struct *p, int new_cpu)
> +{
> +	struct dl_bw *dl_b;
> +	int cpu;
> +
> +	if (dl_entity_is_special(&p->dl) || !p->dl.dl_bw)
> +		return;
> +
> +	guard(raw_spinlock)(&dl_departed_lock);
> +	cpu = p->dl.dl_bw_cpu;
> +	if (cpu < 0 || dl_bw_rebuilding())
> +		return;
> +	if (p->dl.dl_bw_gen == dl_bw_gen) {
> +		if (cpu_rq(cpu)->rd == cpu_rq(new_cpu)->rd)
> +			return;
> +		dl_b = dl_bw_of(cpu);
> +		raw_spin_lock(&dl_b->lock);
> +		__dl_sub(dl_b, p->dl.dl_bw, dl_bw_cpus(cpu));
> +		raw_spin_unlock(&dl_b->lock);
> +	}
> +	dl_b = dl_bw_of(new_cpu);
> +	raw_spin_lock(&dl_b->lock);
> +	__dl_add(dl_b, p->dl.dl_bw, dl_bw_cpus(new_cpu));
> +	raw_spin_unlock(&dl_b->lock);
> +	p->dl.dl_bw_cpu = new_cpu;
> +	p->dl.dl_bw_gen = dl_bw_gen;
> +}

[Severity: Medium]
Can dl_task_move_bw() move the departing bandwidth of a non-deadline task
that was PI-boosted back into the deadline class?

When a task leaves SCHED_DEADLINE before its 0-lag time, switched_from_dl()
arms inactive_timer and leaves p->dl.dl_bw and p->dl.dl_bw_cpu set in the
original root domain.  As a non-deadline task (!task_has_dl_policy(p)), it
can move to another root domain without admission control, and if it is
then PI-boosted by a deadline waiter before its 0-lag time,
switched_to_dl() cancels the inactive timer and takes it off
dl_departed_list while keeping the original dl_bw_cpu recorded.

Because neither dl_task_bw_elsewhere() nor dl_task_move_bw() checks
task_has_dl_policy(p), waking or migrating the boosted task in the new root
domain subtracts p->dl.dl_bw from the original root domain before its
0-lag time and adds that unreserved bandwidth to the new root domain.

> +
> +static void migrate_task_rq_dl(struct task_struct *p, int new_cpu)
>  {
>  	struct rq_flags rf;
>  	struct rq *rq;
>  
> +	/*
> +	 * The new CPU can be in another root domain than the one that counts
> +	 * the bandwidth. A mask can hold CPUs of two root domains, and
> +	 * select_fallback_rq() can widen it. Move the bandwidth there, as
> +	 * dl_task_offline_migration() does for a throttled task.
> +	 *
> +	 * Not for a task that is blocked on a mutex and only queued for
> +	 * proxy execution. It does not run on the new CPU, so its bandwidth
> +	 * stays in its root domain. If its wake-up then picks a CPU of another
> +	 * root domain, the bandwidth moves at that time.
> +	 */
> +	if (unlikely(dl_task_bw_elsewhere(p, new_cpu)) &&
> +	    !(p->is_blocked && task_on_rq_migrating(p)))
> +		dl_task_move_bw(p, new_cpu);

[Severity: Medium]
Does skipping dl_task_move_bw() when p->is_blocked && task_on_rq_migrating(p)
also drop the bandwidth move for normal queued migrations of mutex-blocked
tasks under CONFIG_SCHED_PROXY_EXEC?

When a mutex-blocked task (p->is_blocked == 1) is migrated across root
domains by move_queued_task() or move_queued_task_locked() (for example
from __balance_push_cpu_stop() during CPU hotplug or push_dl_task()),
deactivate_task() sets p->on_rq = TASK_ON_RQ_MIGRATING and set_task_cpu()
updates both p->cpu and p->wake_cpu to new_cpu:

kernel/sched/core.c:move_queued_task() {
	...
	deactivate_task(rq, p, DEQUEUE_NOCLOCK);
	set_task_cpu(p, new_cpu);
	...
	activate_task(rq, p, 0);
}

When the mutex is unlocked, try_to_wake_up() sees p->on_rq == TASK_ON_RQ_QUEUED
and takes ttwu_runnable():

kernel/sched/core.c:ttwu_runnable() {
	...
	if (p->is_blocked) {
		if (p->se.sched_delayed)
			enqueue_task(rq, p, ENQUEUE_NOCLOCK | ENQUEUE_DELAYED);
		if (proxy_needs_return(rq, p))
			return 0;
	}
	...
	ttwu_do_wakeup(p);
	return 1;
}

Because task_cpu(p) == p->wake_cpu, proxy_needs_return() returns false and
ttwu_runnable() finishes the wakeup in place without calling
migrate_task_rq_dl() or task_woken_dl(), leaving p's bandwidth counted in
the old root domain.

[ ... ]

> @@ -4068,8 +4192,10 @@ void dl_rebuild_rd_accounting(void)
>  	 * a resume run one for each CPU. A task gets a bandwidth under the
>  	 * locks this rebuild holds, so the count cannot leave 0 meanwhile.
>  	 */
> -	if (!atomic_read(&dl_nr_bw_tasks))
> +	if (!atomic_read(&dl_nr_bw_tasks)) {
> +		dl_bw_rebuild_end();
>  		return;
> +	}
>  
>  	read_lock(&tasklist_lock);
>  	for_each_process_thread(g, p) {
> @@ -4083,6 +4209,7 @@ void dl_rebuild_rd_accounting(void)
>  			dl_add_task_root_domain(p);
>  	}
>  	read_unlock(&tasklist_lock);
> +	dl_bw_rebuild_end();
>  }

[Severity: Medium]
Can keeping dl_bw_rebuilding() true until the end of the tasklist walk
drop bandwidth moves for tasks that dl_add_task_root_domain() has already
visited?

During CPU deactivation, tasks are woken via select_fallback_rq() or pushed
by __balance_push_cpu_stop() while dl_rebuild_rd_accounting() walks
for_each_process_thread() under read_lock(&tasklist_lock):

CPU0                                    CPU1
partition_sched_domains_locked()
  dl_bw_rebuild_begin()
  ...
  dl_rebuild_rd_accounting()
    dl_clear_root_domains()
    read_lock(&tasklist_lock)
    for_each_process_thread(g, p)
      dl_add_task_root_domain(p)
        /* counts p in RD_A (or skips
         * p if cpu >= nr_cpu_ids) */
                                        try_to_wake_up(p)
                                          select_fallback_rq() -> cpu in RD_B
                                          set_task_cpu(p, new_cpu)
                                            migrate_task_rq_dl(p, new_cpu)
                                              dl_task_bw_elsewhere()
                                                dl_bw_rebuilding() == true
                                                --> returns false (bw not moved)
      /* still visiting later tasks */
    read_unlock(&tasklist_lock)
    dl_bw_rebuild_end()

Once dl_add_task_root_domain(p) releases p->pi_lock, p can wake up or
migrate to an active CPU in another root domain before the walk finishes.
Because dl_bw_rebuilding() is still true, migrate_task_rq_dl() and
task_woken_dl() ignore the move, leaving p's bandwidth counted in RD_A (or
uncounted if dl_task_rebuild_cpu(p) returned nr_cpu_ids).

In addition, when dl_task_bw_elsewhere() runs under rq->__lock without
p->pi_lock, it reads bw_cpu = READ_ONCE(p->dl.dl_bw_cpu) before checking
dl_bw_rebuilding() and READ_ONCE(dl_bw_gen), whereas
dl_add_task_root_domain() updates p->dl.dl_bw_cpu and p->dl.dl_bw_gen under
p->pi_lock and dl_departed_lock, so dl_task_bw_elsewhere() can compare
cpu_rq(bw_cpu)->rd using a stale bw_cpu read from before the rebuild ended.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1794922146.git.mathieu.desnoyers@efficios.com?part=27

  reply	other threads:[~2026-11-17 15:29 UTC|newest]

Thread overview: 55+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 14:10 [PATCH v7 00/40] sched/deadline: bandwidth accounting fixes and selftests Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 01/40] sched: Refuse SCHED_FLAG_KEEP_PARAMS that would change the policy or the class Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 02/40] sched/deadline: Do not admit bandwidth for parameters that are kept Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 03/40] sched: Let a task boosted by a deadline waiter change its policy Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 04/40] sched/deadline: Do not divide by zero active CPUs of a root domain Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 05/40] sched/deadline: Refuse dl_server debugfs writes while the rq is offline Mathieu Desnoyers
2026-11-17 14:38   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 06/40] sched: Take cpuset_mutex for a deadline policy change only with pi Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 07/40] sched/deadline: Do not lose a throttled task switched out and back Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 08/40] sched/deadline: Do not replenish a boost from cleared parameters Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 09/40] rtmutex: Continue the chain walk when a waiter's deadline donor changes Mathieu Desnoyers
2026-11-17 14:50   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 10/40] sched/deadline: Reset the deadline and the runtime with the parameters Mathieu Desnoyers
2026-11-17 14:55   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 11/40] sched/deadline: Reset the donor's runtime at a deboost in the class Mathieu Desnoyers
2026-11-17 14:55   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 12/40] sched/deadline: Replenish a throttled task boosted while asleep Mathieu Desnoyers
2026-11-17 14:55   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 13/40] sched/deadline: Release the bandwidth of a task that leaves the class asleep Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 14/40] sched/deadline: Reject SCHED_DEADLINE on exited tasks Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 15/40] sched/deadline: Keep departing bandwidth across a root domain rebuild Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 16/40] sched/deadline: Fix this_bw when a task is changed during a remote wakeup Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 17/40] sched/deadline: Rebuild the root domains' accounting from the task list Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 18/40] sched/deadline: Record the root domain holding a task's bandwidth Mathieu Desnoyers
2026-11-17 15:14   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 19/40] sched/deadline: Leave a dying task out of a rebuild of the accounting Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 20/40] sched/deadline: Check SCHED_DEADLINE affinity under the rq lock Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 21/40] sched/deadline: Move a task's bandwidth only where room was made Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 22/40] cgroup/cpuset: Wait for an attach before rebuilding the sched domains Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 23/40] sched: Make deadline policy and affinity changes wait for a cpuset attach Mathieu Desnoyers
2026-11-17 15:18   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 24/40] cgroup/cpuset: Give back the bandwidth an attach did not move Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 25/40] cgroup/cpuset: Reserve the bandwidth in the root domain a task moves to Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 26/40] sched/deadline: Make a rebuild count a task in the root domain it runs in Mathieu Desnoyers
2026-11-17 15:31   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 27/40] sched/deadline: Move a task's bandwidth to " Mathieu Desnoyers
2026-11-17 15:29   ` sashiko-bot [this message]
2026-11-17 14:10 ` [PATCH v7 28/40] sched/deadline: Keep a task in the root domain that counts its bandwidth Mathieu Desnoyers
2026-11-17 15:32   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 29/40] sched/deadline: Keep the last CPU a task has in its root domain Mathieu Desnoyers
2026-11-17 15:41   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 30/40] cgroup/cpuset: Refuse to take a deadline task out of " Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 31/40] sched/deadline: Cancel the replenishment timer with the parameters Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 32/40] sched/deadline: Count the bandwidth a boosted task is given Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 33/40] sched/deadline: Start a period for a task that enters the class blocked Mathieu Desnoyers
2026-11-17 15:46   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 34/40] sched/deadline: Cap the runtime left of a constrained task at dl_runtime Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 35/40] sched/deadline: Check the runtime of a task that comes back runnable Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 36/40] sched/deadline: Check the runtime of a task whose parameters change Mathieu Desnoyers
2026-11-17 15:45   ` sashiko-bot
2026-11-17 14:10 ` [PATCH v7 37/40] sched/debug: Show the deadline bandwidth of each runqueue Mathieu Desnoyers
2026-11-17 14:10 ` [PATCH v7 38/40] selftests/sched: Add a test of SCHED_DEADLINE policy round trips Mathieu Desnoyers
2026-11-17 15:23   ` sashiko-bot
2026-11-17 14:11 ` [PATCH v7 39/40] selftests/sched: Test SCHED_DEADLINE bandwidth across a root domain rebuild Mathieu Desnoyers
2026-11-17 14:11 ` [PATCH v7 40/40] selftests/sched: Add a test of SCHED_DEADLINE bandwidth accounting Mathieu Desnoyers

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=sashiko-outbox-219077@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=hannes@cmpxchg.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rt-devel@lists.linux.dev \
    --cc=mathieu.desnoyers@efficios.com \
    --cc=mkoutny@suse.com \
    --cc=sashiko-reviews@lists.linux.dev \
    --cc=tj@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