mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: K Prateek Nayak <kprateek.nayak@amd.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Doug Smythies <dsmythies@telus.net>, <mingo@kernel.org>,
	<juri.lelli@redhat.com>, <vincent.guittot@linaro.org>,
	<dietmar.eggemann@arm.com>, <rostedt@goodmis.org>,
	<bsegall@google.com>, <mgorman@suse.de>, <vschneid@redhat.com>,
	<linux-kernel@vger.kernel.org>, <wangtao554@huawei.com>,
	<quzicheng@huawei.com>, <wuyun.abel@bytedance.com>
Subject: Re: [PATCH 0/4] sched: Various reweight_entity() fixes
Date: Wed, 11 Feb 2026 14:20:31 +0530	[thread overview]
Message-ID: <d3e16de8-cf41-40b0-a976-fb2f37608b70@amd.com> (raw)
In-Reply-To: <4035fa61-cc90-498b-af84-c9ac4903cd5f@amd.com>

[-- Attachment #1: Type: text/plain, Size: 6412 bytes --]

Hello Peter,

On 2/11/2026 11:58 AM, K Prateek Nayak wrote:
> I'll go back and fix my debug patch to give a full view of series of
> events that leads to this but in the meantime I've added the full log
> below if someone wants to take a look.

Attached is archive of trace. Should expand to give a "timeline.trace"
file. Following are some series of events before the failed pick:


  Note: "cfs_rq: depth(1) ..." is the cfs_rq which fails but the first
  "dump_h_overflow_cfs_rq: cfs_rq: ..." will note depth as "0" which is
  the depth of padding and is not the actual cfs-rq depth.

  The lines after "pick_task_fair: cfs_rq of overflow:" dumps the cfs_rq
  and the queued entities that failed the pick.

  ...
  [006] ... : __enqueue_entity: Enqueue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(-7389160032126894080) sum_weight(2169503744) zero_vruntime(18446744052406571137) sum_shift(0) avg_vruntime(18446744052406571137)
  [006] ... : __dequeue_entity: Dequeue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(3512818056739946496) sum_weight(2168455168) zero_vruntime(18446744055883646744) sum_shift(0) avg_vruntime(18446744055883646744)
  [006] ... : __enqueue_entity: Enqueue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(1653261820069150720) sum_weight(2169503744) zero_vruntime(18446744056741357296) sum_shift(0) avg_vruntime(18446744056741357296)
  [006] ... : __dequeue_entity: Dequeue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(3337798565006147584) sum_weight(2168455168) zero_vruntime(18446744055964215508) sum_shift(0) avg_vruntime(18446744055964215508)
  ...
  [006] ... : __enqueue_entity: Enqueue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(-4817868778594369536) sum_weight(2169503744) zero_vruntime(18446744051221380982) sum_shift(0) avg_vruntime(18446744051221380982)
  [006] ... : __dequeue_entity: Dequeue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(8703580922935508992) sum_weight(2168455168) zero_vruntime(18446744053489887930) sum_shift(0) avg_vruntime(18446744053489887930)
  ...
  [006] ... : __enqueue_entity: Enqueue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(1049195167414747136) sum_weight(2169503744) zero_vruntime(18446744057019812035) sum_shift(0) avg_vruntime(18446744057019812035)
  [006] ... : __dequeue_entity: Dequeue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(2118331978874880000) sum_weight(2168455168) zero_vruntime(18446744056526599437) sum_shift(0) avg_vruntime(18446744056526599437)
  ...
  [006] ... : __enqueue_entity: Enqueue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(8649466989629145088) sum_weight(2169503744) zero_vruntime(18446744053516583405) sum_shift(0) avg_vruntime(18446744053516583405)
  [006] ... : __dequeue_entity: Dequeue cfs_rq: depth(1) weight(2169503744) nr_queued(2069) sum_w_vruntime(-984823152929406976) sum_weight(2168455168) zero_vruntime(18446744049450913118) sum_shift(0) avg_vruntime(18446744049450913118)

  ...

  [006] ... : pick_task_fair: cfs_rq of overflow: /* Essentially failed pick. */
  [006] ... : dump_h_overflow_cfs_rq: cfs_rq: ... weight(2169503744) nr_queued(2069) sum_w_vruntime(-59587237455593472) sum_weight(2168455168) zero_vruntime(18446744049024233365) sum_shift(0) avg_vruntime(18446744049024233365)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744057295755695) slice(2800000) deadline(18446744057298552230) curr?(1) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744057379145967) slice(2800000) deadline(18446744057381940989) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744057463106601) slice(2800000) deadline(18446744057465901974) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744057487667304) slice(2800000) deadline(18446744057490461976) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744058420769904) slice(2800000) deadline(18446744058423566108) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744058486927372) slice(2800000) deadline(18446744058489723636) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744061273836896) slice(2800000) deadline(18446744061276631077) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744061535759861) slice(2800000) deadline(18446744061538554473) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744061788668421) slice(2800000) deadline(18446744061791464976) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744062846656914) slice(2800000) deadline(18446744062849454680) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744064164417134) slice(2800000) deadline(18446744064167214230) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(18446744065239426914) slice(2800000) deadline(18446744065242224540) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(880604900) slice(2800000) deadline(883328346) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(880537321) slice(2800000) deadline(883331492) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(880537194) slice(2800000) deadline(883333328) curr?(0) task?(1) delayed?(0) se_depth(1)
  [006] ... : dump_h_overflow_entity: se: weight(1048576) vruntime(880540318) slice(2800000) deadline(883334689) curr?(0) task?(1) delayed?(0) se_depth(1)
  ...


The avg_vruntime(18446744049024233365) = -24685318251 (s64) is smaller
than that of all the vruntime of queued entities.

Peter, Could it be the "left bias" in avg_vruntime() coming into play
with larger "cfs_rq->sum_weight"?

-- 
Thanks and Regards,
Prateek

[-- Attachment #2: timeline.tar.gz --]
[-- Type: application/x-gzip, Size: 1200020 bytes --]

  reply	other threads:[~2026-02-11  8:50 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-30  9:34 Peter Zijlstra
2026-01-30  9:34 ` [PATCH 1/4] sched/fair: Only set slice protection at pick time Peter Zijlstra
2026-01-30 15:52   ` Vincent Guittot
2026-01-30  9:34 ` [PATCH 2/4] sched/eevdf: Update se->vprot in reweight_entity() Peter Zijlstra
2026-01-30 16:20   ` Vincent Guittot
2026-01-30  9:34 ` [PATCH 3/4] sched/fair: Increase weight bits for avg_vruntime Peter Zijlstra
2026-01-30  9:34 ` [PATCH 4/4] sched/fair: Revert 6d71a9c61604 ("sched/fair: Fix EEVDF entity placement bug causing scheduling lag") Peter Zijlstra
2026-01-31  1:47   ` Zhang Qiao
2026-01-31 15:21     ` Peter Zijlstra
2026-02-02  9:12       ` Peter Zijlstra
2026-02-02  9:24         ` Peter Zijlstra
2026-02-02 11:23         ` Zhang Qiao
2026-02-01 17:13 ` [PATCH 0/4] sched: Various reweight_entity() fixes Doug Smythies
2026-02-03  6:45 ` K Prateek Nayak
2026-02-03 11:11   ` Peter Zijlstra
2026-02-03 12:19     ` K Prateek Nayak
2026-02-03 16:36       ` Doug Smythies
2026-02-10 18:13         ` Peter Zijlstra
2026-02-11  5:51           ` Doug Smythies
2026-02-04 10:15       ` K Prateek Nayak
2026-02-09 15:47         ` Peter Zijlstra
2026-02-09 16:52           ` K Prateek Nayak
2026-02-10  5:16           ` K Prateek Nayak
2026-02-10 10:29             ` Peter Zijlstra
2026-02-10 15:41           ` Doug Smythies
2026-02-10 18:09             ` K Prateek Nayak
2026-02-10 18:35               ` Peter Zijlstra
2026-02-10 20:04                 ` K Prateek Nayak
2026-02-11  6:28                   ` K Prateek Nayak
2026-02-11  8:50                     ` K Prateek Nayak [this message]
2026-02-11 23:09               ` Doug Smythies
2026-02-10 18:52             ` Peter Zijlstra
2026-02-10 20:52           ` Vincent Guittot
2026-02-11  5:21             ` Doug Smythies
2026-02-11  8:49               ` Vincent Guittot
2026-02-11  9:01                 ` Peter Zijlstra
2026-02-11 10:48                   ` Peter Zijlstra
2026-02-11 10:49                     ` Peter Zijlstra
2026-02-11 11:15                     ` Vincent Guittot
2026-02-11 16:28                       ` Peter Zijlstra
2026-02-12  7:43                         ` K Prateek Nayak
2026-02-12 11:59                           ` Peter Zijlstra
2026-02-12 17:16                             ` Peter Zijlstra
2026-02-12 17:24                               ` Vincent Guittot
2026-02-12 19:31                                 ` Peter Zijlstra
2026-02-13  5:22                                   ` K Prateek Nayak
2026-02-13  6:44                                     ` Peter Zijlstra
2026-02-13 10:50                                       ` Peter Zijlstra
2026-02-13 14:29                                         ` K Prateek Nayak
2026-02-14  6:31                                         ` Doug Smythies
2026-02-21 22:51                                           ` Doug Smythies
2026-02-12 19:29                               ` Peter Zijlstra
2026-02-12 19:37                                 ` Doug Smythies
2026-02-13  6:04                                 ` K Prateek Nayak
2026-02-11 16:21                     ` Peter Zijlstra
2026-02-12  5:54                       ` Doug Smythies
2026-02-12  7:51                         ` Peter Zijlstra
2026-02-12 15:47                           ` Doug Smythies
2026-02-12  7:46                       ` Peter Zijlstra
2026-02-11 23:25                     ` Doug Smythies
2026-02-11  8:48             ` Peter Zijlstra
2026-02-04 10:44       ` Peter Zijlstra
2026-02-14  7:20 ` Shubhang Kaushik
2026-02-16  3:14   ` K Prateek Nayak
2026-02-16 10:59     ` Dietmar Eggemann
2026-02-17 14:37       ` Dietmar Eggemann
2026-02-17 22:02     ` Shubhang Kaushik
2026-02-17  4:20 ` K Prateek Nayak
2026-02-18 18:37 ` Shubhang Kaushik
2026-02-19  7:53   ` 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=d3e16de8-cf41-40b0-a976-fb2f37608b70@amd.com \
    --to=kprateek.nayak@amd.com \
    --cc=bsegall@google.com \
    --cc=dietmar.eggemann@arm.com \
    --cc=dsmythies@telus.net \
    --cc=juri.lelli@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mgorman@suse.de \
    --cc=mingo@kernel.org \
    --cc=peterz@infradead.org \
    --cc=quzicheng@huawei.com \
    --cc=rostedt@goodmis.org \
    --cc=vincent.guittot@linaro.org \
    --cc=vschneid@redhat.com \
    --cc=wangtao554@huawei.com \
    --cc=wuyun.abel@bytedance.com \
    /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

Powered by JetHome