From: Andrea Righi <arighi@nvidia.com>
To: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: John Stultz <jstultz@google.com>,
Suleiman Souhlal <suleiman@google.com>,
Ingo Molnar <mingo@redhat.com>,
Peter Zijlstra <peterz@infradead.org>,
Juri Lelli <juri.lelli@redhat.com>,
Vincent Guittot <vincent.guittot@linaro.org>,
Will Deacon <will@kernel.org>, Boqun Feng <boqun@kernel.org>,
linux-kernel@vger.kernel.org,
Dietmar Eggemann <dietmar.eggemann@arm.com>,
Steven Rostedt <rostedt@goodmis.org>,
Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
Valentin Schneider <vschneid@redhat.com>,
Waiman Long <longman@redhat.com>
Subject: Re: [RFC PATCH 06/16] sched/core: Queue blocked donor onto sleeping owner for chain-wakeup
Date: Wed, 26 Aug 2026 18:20:10 +0200 [thread overview]
Message-ID: <ao8SOkBYH1uLh6Sa@gpd4> (raw)
In-Reply-To: <20260826062901.2137-7-kprateek.nayak@amd.com>
On Wed, Aug 26, 2026 at 06:28:50AM +0000, K Prateek Nayak wrote:
> Add a blocked_donor to sleeping owner's "blocked_head" when
> find_proxy_task() resolves to a blocked task. Once added to the list,
> double check if the owner has woken up by checking "owner->on_rq".
>
> If the owner has woken up, remove the task from "blocked_head" and
> continue try find_proxy_task() agiain to re-evaluate the state of owner
> and take the correct steps.
nit: s/agiain/again/
>
> Since find_proxy_task() is called with wait_lock held, which the owner
> needs during unlock, it is guaranteed that owner cannot disappear under
> us in the process.
>
> The added data memebers in task_struct serve the following pusrpose:
nit:
s/memebers/members/
s/pusrpose/purpose/
>
> - blocked_head: Contains the blocked donors queued on us
> - blocked_node: The list head used to queue onto blocked_head of a
> sleeping owner
> - sleeping_owner: Sleeping owner on which the task is queued.
>
> Co-developed-by: John Stultz <jstultz@google.com>
> Signed-off-by: John Stultz <jstultz@google.com>
> Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
> ---
> include/linux/sched.h | 5 ++++
> init/init_task.c | 5 ++++
> kernel/fork.c | 5 ++++
> kernel/sched/core.c | 68 +++++++++++++++++++++++++++++++++++++++++--
> 4 files changed, 81 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index 373bcc0598d1..bf0f4b6be7c4 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1251,6 +1251,11 @@ struct task_struct {
>
> struct mutex *blocked_on; /* lock we're blocked on */
> raw_spinlock_t blocked_lock;
> +#ifdef CONFIG_SCHED_PROXY_EXEC
> + struct list_head blocked_head; /* tasks blocked on this task */
> + struct list_head blocked_node; /* our entry on someone elses blocked_head */
> + struct task_struct *sleeping_owner; /* task our blocked_node is enqueued on */
> +#endif
>
> /*
> * The task that is boosting this task; a back link for the current
> diff --git a/init/init_task.c b/init/init_task.c
> index b67ef6040a65..a097c0def4c4 100644
> --- a/init/init_task.c
> +++ b/init/init_task.c
> @@ -211,6 +211,11 @@ struct task_struct init_task __aligned(L1_CACHE_BYTES) = {
> &init_task.alloc_lock),
> #endif
> .blocked_donor = NULL,
> +#ifdef CONFIG_SCHED_PROXY_EXEC
> + .blocked_head = LIST_HEAD_INIT(init_task.blocked_head),
> + .blocked_node = LIST_HEAD_INIT(init_task.blocked_node),
> + .sleeping_owner = NULL,
> +#endif
> #ifdef CONFIG_RT_MUTEXES
> .pi_waiters = RB_ROOT_CACHED,
> .pi_top_task = NULL,
> diff --git a/kernel/fork.c b/kernel/fork.c
> index f0e2e131a9a5..88f2b6e08c46 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -2247,6 +2247,11 @@ __latent_entropy struct task_struct *copy_process(
>
> p->blocked_on = NULL; /* not blocked yet */
> p->blocked_donor = NULL; /* nobody is boosting p yet */
> +#ifdef CONFIG_SCHED_PROXY_EXEC
> + INIT_LIST_HEAD(&p->blocked_head);
> + INIT_LIST_HEAD(&p->blocked_node);
> + p->sleeping_owner = NULL;
> +#endif
>
> #ifdef CONFIG_BCACHE
> p->sequential_io = 0;
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index b4a2771eb290..e53967a126a9 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -2229,6 +2229,16 @@ void activate_task(struct rq *rq, struct task_struct *p, int en_flags)
> __activate_task(rq, p, en_flags | ENQUEUE_MIGRATING);
> }
>
> +#ifdef CONFIG_SCHED_PROXY_EXEC
> +
> +static void __proxy_dequeue_from_owner(struct task_struct *p)
> +{
> + list_del_init(&p->blocked_node);
> + WRITE_ONCE(p->sleeping_owner, NULL);
> +}
> +
> +#endif /* CONFIG_SCHED_PROXY_EXEC */
> +
> static void activate_blocked_task(struct rq *rq, struct task_struct *p, int en_flags)
> {
> __activate_task(rq, p, en_flags);
> @@ -6872,6 +6882,43 @@ static void proxy_migrate_task(struct rq *rq, struct rq_flags *rf,
> proxy_reacquire_rq_lock(rq, rf);
> }
>
> +static void proxy_enqueue_on_owner(struct rq *rq, struct task_struct *owner,
> + struct task_struct *p)
> +{
> + lockdep_assert_rq_held(rq);
> + lockdep_assert_held(&owner->blocked_lock);
> +
> + WARN_ON(!p->on_rq);
> + WARN_ON(p->sleeping_owner);
> +
> + WRITE_ONCE(p->sleeping_owner, owner);
> + list_add(&p->blocked_node, &owner->blocked_head);
> + proxy_resched_idle(rq);
> +
> + /*
> + * Order against __activate_blocked_task_slowpath() checking
> + * owner->blocked_list after setting owner->on_rq.
This should be owner->blocked_head not owner->blocked_list, right?
> + */
> + smp_mb();
> +
> + if (READ_ONCE(owner->on_rq)) {
> + /*
> + * owner has woken up and may miss activating us.
> + * Remove ourself from "owner->blocked_head" and try
> + * find_proxy_task() again considering the owner's
> + * new state.
> + */
> + __proxy_dequeue_from_owner(p);
> + return;
> + }
> +
> + /*
> + * Owner is fully blocked. __activate_blocked_task_slowpath()
> + * will see us on the list during wakeup and DTRT.
> + */
> + block_task(rq, p, READ_ONCE(p->__state));
> +}
> +
> /*
> * Find runnable lock owner to proxy for mutex blocked donor
> *
> @@ -6962,8 +7009,25 @@ find_proxy_task(struct rq *rq, struct task_struct *donor, struct rq_flags *rf)
> /* XXX Don't handle blocked owners yet */
> if (curr_in_chain)
> return proxy_resched_idle(rq);
> - __clear_task_blocked_on(p, NULL);
> - goto deactivate;
> + /*
> + * If !@owner->on_rq, holding @rq->lock will not pin the task,
> + * so we cannot drop @mutex->wait_lock until we're sure its a blocked
> + * task on this rq.
> + *
> + * We use @owner->blocked_lock to serialize against ttwu_activate().
> + * Either we see its new owner->on_rq or it will see our list_add().
> + */
> + WARN_ON(owner == p);
> +
> + raw_spin_unlock(&p->blocked_lock);
> + raw_spin_lock(&owner->blocked_lock);
> +
> + proxy_enqueue_on_owner(rq, owner, p);
> +
> + raw_spin_unlock(&owner->blocked_lock);
> + raw_spin_lock(&p->blocked_lock);
> +
> + return NULL; /* retry task selection */
> }
>
> owner_cpu = task_cpu(owner);
> --
> 2.34.1
>
Thanks,
-Andrea
next prev parent reply other threads:[~2026-08-26 16:20 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-26 6:28 [RFC PATCH 00/16][PoC] sched/core: Alternate approach to sleeping-owner handling in PROXY_EXEC K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 01/16] sched/core: Break activation of blocked task into a separate helper K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 02/16] sched/core: Use enqueue/dequeue flags instead of task_on_rq_migrating() K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 03/16] sched/fair: Use enqueue flags for DO_ATTACH in update_load_avg() K Prateek Nayak
2026-08-26 15:36 ` Andrea Righi
2026-08-26 6:28 ` [RFC PATCH 04/16] sched/core: Activate blocked donor when no owner is found K Prateek Nayak
2026-08-26 15:56 ` Andrea Righi
2026-08-26 17:20 ` K Prateek Nayak
2026-08-28 6:04 ` K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 05/16] sched/core: Do not queue blocked donor on a delayed owner K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 06/16] sched/core: Queue blocked donor onto sleeping owner for chain-wakeup K Prateek Nayak
2026-08-26 16:20 ` Andrea Righi [this message]
2026-08-27 3:51 ` K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 07/16] sched/core: Avoid delaying blocked donors queued on sleeping owner K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 08/16] sched/deadline: Prepare for blocking and proxy activation with MIGRATING flag K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 09/16] sched/core: Track CPU where the task was blocked on K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 10/16] sched/core: Introduce p->is_linked to track if task is queued on sleeping owner K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 11/16] sched/core: Prepare to inspect ->is_linked alongside ->on_rq during wakeup K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 12/16] sched:core: Add MIGRATING flags when blocking and activating linked donors K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 13/16] sched/core: Use p->is_linked state to unlink from sleeping owner early K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 14/16] sched/core: Introduce chain-wakeup to activate blocked donors K Prateek Nayak
2026-08-26 6:28 ` [RFC PATCH 15/16] locking/mutex: Track locks owned by a task in a per-task counter K Prateek Nayak
2026-08-26 6:29 ` [RFC PATCH 16/16] sched/core: Set activation of non lock-holders to fast-path K Prateek Nayak
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=ao8SOkBYH1uLh6Sa@gpd4 \
--to=arighi@nvidia.com \
--cc=boqun@kernel.org \
--cc=bsegall@google.com \
--cc=dietmar.eggemann@arm.com \
--cc=jstultz@google.com \
--cc=juri.lelli@redhat.com \
--cc=kprateek.nayak@amd.com \
--cc=linux-kernel@vger.kernel.org \
--cc=longman@redhat.com \
--cc=mgorman@suse.de \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=suleiman@google.com \
--cc=vincent.guittot@linaro.org \
--cc=vschneid@redhat.com \
--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
all inboxes | Powered by JetHome®