[GIT PULL] scheduler changes for v6.4

[GIT PULL] scheduler changes for v6.4

[Ingo Molnar]

Linus,

Please pull the latest sched/core git tree from:

   git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git sched-core-2023-04-27

   # HEAD: f31dcb152a3d0816e2f1deab4e64572336da197d sched/clock: Fix local_clock() before sched_clock_init()

Scheduler changes for v6.4:

 - Allow unprivileged PSI poll()ing

 - Fix performance regression introduced by mm_cid

 - Improve livepatch stalls by adding livepatch task switching to cond_resched(),
   this resolves livepatching busy-loop stalls with certain CPU-bound kthreads.

 - Improve sched_move_task() performance on autogroup configs.

 - On core-scheduling CPUs, avoid selecting throttled tasks to run

 - Misc cleanups, fixes and improvements.

 Thanks,

	Ingo

------------------>
Aaron Thompson (1):
      sched/clock: Fix local_clock() before sched_clock_init()

Domenico Cerasuolo (4):
      sched/psi: Rearrange polling code in preparation
      sched/psi: Rename existing poll members in preparation
      sched/psi: Extract update_triggers side effect
      sched/psi: Allow unprivileged polling of N*2s period

Hao Jia (1):
      sched/core: Avoid selecting the task that is throttled to run when core-sched enable

Josh Poimboeuf (5):
      livepatch: Convert stack entries array to percpu
      livepatch: Skip task_call_func() for current task
      livepatch,sched: Add livepatch task switching to cond_resched()
      vhost: Fix livepatch timeouts in vhost_worker()
      sched/core: Make sched_dynamic_mutex static

Libo Chen (1):
      sched/fair: Fix inaccurate tally of ttwu_move_affine

Mathieu Desnoyers (1):
      sched: Fix performance regression introduced by mm_cid

Schspa Shi (1):
      sched/rt: Fix bad task migration for rt tasks

Shrikanth Hegde (1):
      sched: Interleave cfs bandwidth timers for improved single thread performance at low utilization

Tom Rix (1):
      sched/topology: Make sched_energy_mutex,update static

wuchi (1):
      sched/core: Reduce cost of sched_move_task when config autogroup


 Documentation/accounting/psi.rst |   4 +
 drivers/vhost/vhost.c            |   3 +-
 include/linux/livepatch.h        |   1 +
 include/linux/livepatch_sched.h  |  29 ++
 include/linux/mm_types.h         |  82 ++++-
 include/linux/psi.h              |   2 +-
 include/linux/psi_types.h        |  43 +--
 include/linux/sched.h            |  23 +-
 include/linux/sched/mm.h         |   5 +
 kernel/cgroup/cgroup.c           |   2 +-
 kernel/fork.c                    |   9 +-
 kernel/livepatch/core.c          |   1 +
 kernel/livepatch/transition.c    | 122 +++++--
 kernel/sched/clock.c             |   3 +
 kernel/sched/core.c              | 669 ++++++++++++++++++++++++++++++++++++---
 kernel/sched/deadline.c          |  11 +
 kernel/sched/fair.c              |  22 +-
 kernel/sched/psi.c               | 473 ++++++++++++++-------------
 kernel/sched/rt.c                |  23 ++
 kernel/sched/sched.h             | 243 ++++++++++++--
 kernel/sched/topology.c          |   4 +-
 21 files changed, 1424 insertions(+), 350 deletions(-)
 create mode 100644 include/linux/livepatch_sched.h

Re: [GIT PULL] scheduler changes for v6.4

[Linus Torvalds]

On Thu, Apr 27, 2023 at 1:51 PM Ingo Molnar <mingo@kernel.org> wrote:
>
>  - Fix performance regression introduced by mm_cid

This causes a conflict with commit b20b0368c614 ("mm: fix memory leak
on mm_init error handling") that came in through the MM tree.

That conflict is trivial to resolve, and I did so.

Except I also checked my resolution with what was going on in
linux-next, and it's different.

I'm pretty sure the resolution in linux-next is wrong, but I thought
I'd mention this, since clearly this wasn't caught in linux-next.

Or maybe it's me that did it wrong, but hey, that couldn't actually
happen, could it?

                  Linus

Re: [GIT PULL] scheduler changes for v6.4

[Mathieu Desnoyers]

On 2023-04-28 18:02, Linus Torvalds wrote:
> On Thu, Apr 27, 2023 at 1:51 PM Ingo Molnar <mingo@kernel.org> wrote:
>>
>>   - Fix performance regression introduced by mm_cid
> 
> This causes a conflict with commit b20b0368c614 ("mm: fix memory leak
> on mm_init error handling") that came in through the MM tree.
> 
> That conflict is trivial to resolve, and I did so.
> 
> Except I also checked my resolution with what was going on in
> linux-next, and it's different.
> 
> I'm pretty sure the resolution in linux-next is wrong, but I thought
> I'd mention this, since clearly this wasn't caught in linux-next.
> 
> Or maybe it's me that did it wrong, but hey, that couldn't actually
> happen, could it?

I've reviewed both merge commits (c79e0731da from next-20230428 and
586b222d74 from master) and I confirm that your conflict resolution
is correct. The one in next was wrong.

Thanks,

Mathieu

-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

Re: [GIT PULL] scheduler changes for v6.4

[Ingo Molnar]

* Mathieu Desnoyers <mathieu.desnoyers@efficios.com> wrote:

> On 2023-04-28 18:02, Linus Torvalds wrote:
> > On Thu, Apr 27, 2023 at 1:51 PM Ingo Molnar <mingo@kernel.org> wrote:
> > > 
> > >   - Fix performance regression introduced by mm_cid
> > 
> > This causes a conflict with commit b20b0368c614 ("mm: fix memory leak
> > on mm_init error handling") that came in through the MM tree.
> > 
> > That conflict is trivial to resolve, and I did so.
> > 
> > Except I also checked my resolution with what was going on in
> > linux-next, and it's different.
> > 
> > I'm pretty sure the resolution in linux-next is wrong, but I thought
> > I'd mention this, since clearly this wasn't caught in linux-next.
> > 
> > Or maybe it's me that did it wrong, but hey, that couldn't actually
> > happen, could it?
> 
> I've reviewed both merge commits (c79e0731da from next-20230428 and
> 586b222d74 from master) and I confirm that your conflict resolution
> is correct. The one in next was wrong.

Indeed - mm_alloc_cid() happens after the MM context has already been 
initialized, so the fail_cid label needs to deinit the context aka call 
destroy_context().

So the resolution by Linus is the correct one:

        mm_destroy_cid(mm);
 fail_cid:
        destroy_context(mm);
 fail_nocontext:
        mm_free_pgd(mm);

Thanks,

	Ingo

Re: [GIT PULL] scheduler changes for v6.4

[pr-tracker-bot]

The pull request you sent on Thu, 27 Apr 2023 22:51:05 +0200:

> git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git sched-core-2023-04-27

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/586b222d748e91c619d68e9239654ebc7fed9b0c

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html