mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Luca Abeni <luca.abeni@unitn.it>
To: linux-kernel@vger.kernel.org
Cc: Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>, Juri Lelli <juri.lelli@arm.com>,
	Luca Abeni <luca.abeni@unitn.it>
Subject: [RFC v2 0/7] CPU reclaiming for SCHED_DEADLINE
Date: Fri,  1 Apr 2016 17:12:26 +0200	[thread overview]
Message-ID: <1459523553-29089-1-git-send-email-luca.abeni@unitn.it> (raw)

Hi all,

this patchset implements CPU reclaiming (using the GRUB algorithm[1])
for SCHED_DEADLINE: basically, this feature allows SCHED_DEADLINE tasks
to consume more than their reserved runtime, up to a maximum fraction
of the CPU time (so that other tasks are left some spare CPU time to
execute), if this does not break the guarantees of other SCHED_DEADLINE
tasks.
The patchset applies on top of tip/master.

Respect to the first version of the RFC:
- I tried to address all the comments I received
- I removed some patches that were not really used in the patchset
- I rebased on tip/master
- I removed all the checkpatch warnings
- I added a new patch (patch 0004) to update the total -deadline
  utilization (dl_b->total_bw) at the correct time, addressing the
  large comment in __setparam_dl() (see both  dl_overflow() and
  __setparam_dl()). 


The implemented CPU reclaiming algorithm is based on tracking the
utilization U_act of active tasks (first 3 patches), and modifying the
runtime accounting rule (see patch 0005). The original GRUB algorithm is
modified as described in [2] to support multiple CPUs (the original
algorithm only considered one single CPU, this one tracks U_act per
runqueue) and to leave an "unreclaimable" fraction of CPU time to non
SCHED_DEADLINE tasks (the original algorithm can consume 100% of the CPU
time, starving all the other tasks).

I tried to split the patches so that the whole patchset can be better
understood; if they should be organized in a different way, let me know.
The first 3 patches (tracking of per-runqueue active utilization) can
be useful for frequency scaling too.
Patches 0005-0007 implement the reclaiming algorithm. and patch 0004
uses the newly introduced "inactive timer" (introduced in patch 0003)
to fix dl_overflow() and __setparam_dl().


Luca Abeni (7):
  Track the active utilisation
  Correctly track the active utilisation for migrating tasks
  Improve the tracking of active utilisation
  Fix the update of the total -deadline utilization
  GRUB accounting
  Make GRUB a task's flag
  Do not reclaim the whole CPU bandwidth

 include/linux/sched.h      |   1 +
 include/uapi/linux/sched.h |   1 +
 kernel/sched/core.c        |  44 ++++-----
 kernel/sched/deadline.c    | 225 +++++++++++++++++++++++++++++++++++++++++----
 kernel/sched/sched.h       |  13 +++
 5 files changed, 239 insertions(+), 45 deletions(-)

-- 
2.5.0

             reply	other threads:[~2016-04-01 15:18 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-04-01 15:12 Luca Abeni [this message]
2016-04-01 15:12 ` [RFC v2 1/7] Track the active utilisation Luca Abeni
2016-04-05 12:23   ` Peter Zijlstra
2016-04-05 16:47     ` luca abeni
2016-04-01 15:12 ` [RFC v2 2/7] Correctly track the active utilisation for migrating tasks Luca Abeni
2016-04-05 12:24   ` Peter Zijlstra
2016-04-05 16:50     ` luca abeni
2016-04-01 15:12 ` [RFC v2 3/7] Improve the tracking of active utilisation Luca Abeni
2016-04-05 12:42   ` Peter Zijlstra
2016-04-05 17:00     ` luca abeni
2016-04-05 12:42   ` Peter Zijlstra
2016-04-05 17:05     ` luca abeni
2016-04-05 14:48   ` Peter Zijlstra
2016-04-05 17:17     ` luca abeni
2016-04-05 15:00   ` Peter Zijlstra
2016-04-05 17:56     ` luca abeni
2016-04-05 18:00       ` Peter Zijlstra
2016-04-05 19:35         ` luca abeni
2016-04-05 18:02       ` Peter Zijlstra
2016-04-05 19:24         ` luca abeni
2016-04-05 19:31           ` Peter Zijlstra
2016-04-05 19:32           ` luca abeni
2016-04-05 18:11       ` Peter Zijlstra
2016-04-01 15:12 ` [RFC v2 4/7] Fix the update of the total -deadline utilization Luca Abeni
2016-04-05 12:58   ` Peter Zijlstra
2016-04-05 17:16     ` luca abeni
2016-04-01 15:12 ` [RFC v2 5/7] GRUB accounting Luca Abeni
2016-04-01 15:12 ` [RFC v2 6/7] Make GRUB a task's flag Luca Abeni
2016-04-01 15:12 ` [RFC v2 7/7] Do not reclaim the whole CPU bandwidth Luca Abeni

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=1459523553-29089-1-git-send-email-luca.abeni@unitn.it \
    --to=luca.abeni@unitn.it \
    --cc=juri.lelli@arm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=peterz@infradead.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®