Two sched patchs missing from 4.4-stable: inclusion possible?

Two sched patchs missing from 4.4-stable: inclusion possible?

[Ben Pineau]

Hello,

The following two patches series, merged in mainline 4.7, was Cc'ed to stable@,
but somehow didn't made its way to the 4.4.x stable tree:
  754bd598be9bbc953bc709a9e8ed7f3188bfb9d7
  094f469172e00d6ab0a3130b0e01c83b3cf3a98d

Links to the discussions:
  https://marc.info/?l=linux-kernel&m=146608185431676
  https://marc.info/?l=linux-kernel&m=146608184031673

This patch series address a real life issue, ie. we're seeing an
oopsing machine about every day on a kubernetes cluster, and
other users reported the same problem, for instance :
  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1687512
  https://github.com/kubernetes/kops/issues/874

Those two patches are also on their way for inclusion in the current
(4.4 based) LTS Ubuntu release :
https://code.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/+ref/master-next

Could this series be considered for 4.4-stable inclusion by any chance ?

For ref, the stacktrace I'm getting on an unpatched 4.4.67 kernel:

  [168797.975683] BUG: unable to handle kernel NULL pointer
dereference at 0000000000000050
  [168797.976656] Call Trace:
  [168797.976656]  [<ffffffff810bbef5>] pick_next_task_fair+0x75/0x4b0
  [168797.976656]  [<ffffffff816f38a1>] __schedule+0x6c1/0x7f0
  [168797.976656]  [<ffffffff816f3a05>] schedule+0x35/0x80
  [168797.976656]  [<ffffffff816f3cee>] schedule_preempt_disabled+0xe/0x10
  [168797.976656]  [<ffffffff810c299a>] cpu_startup_entry+0x19a/0x360
  [168797.976656]  [<ffffffff810512f0>] start_secondary+0x160/0x1a0
  [168797.976656] RIP  [<ffffffff810b562f>] pick_next_entity+0x7f/0x160

Thanks

Re: Two sched patchs missing from 4.4-stable: inclusion possible?

[Greg KH]

On Thu, May 18, 2017 at 11:11:23AM +0200, Ben Pineau wrote:
> Hello,
> 
> The following two patches series, merged in mainline 4.7, was Cc'ed to stable@,
> but somehow didn't made its way to the 4.4.x stable tree:
>   754bd598be9bbc953bc709a9e8ed7f3188bfb9d7
>   094f469172e00d6ab0a3130b0e01c83b3cf3a98d
> 
> Links to the discussions:
>   https://marc.info/?l=linux-kernel&m=146608185431676
>   https://marc.info/?l=linux-kernel&m=146608184031673
> 
> This patch series address a real life issue, ie. we're seeing an
> oopsing machine about every day on a kubernetes cluster, and
> other users reported the same problem, for instance :
>   https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1687512
>   https://github.com/kubernetes/kops/issues/874
> 
> Those two patches are also on their way for inclusion in the current
> (4.4 based) LTS Ubuntu release :
> https://code.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/xenial/+ref/master-next
> 
> Could this series be considered for 4.4-stable inclusion by any chance ?
> 
> For ref, the stacktrace I'm getting on an unpatched 4.4.67 kernel:
> 
>   [168797.975683] BUG: unable to handle kernel NULL pointer
> dereference at 0000000000000050
>   [168797.976656] Call Trace:
>   [168797.976656]  [<ffffffff810bbef5>] pick_next_task_fair+0x75/0x4b0
>   [168797.976656]  [<ffffffff816f38a1>] __schedule+0x6c1/0x7f0
>   [168797.976656]  [<ffffffff816f3a05>] schedule+0x35/0x80
>   [168797.976656]  [<ffffffff816f3cee>] schedule_preempt_disabled+0xe/0x10
>   [168797.976656]  [<ffffffff810c299a>] cpu_startup_entry+0x19a/0x360
>   [168797.976656]  [<ffffffff810512f0>] start_secondary+0x160/0x1a0
>   [168797.976656] RIP  [<ffffffff810b562f>] pick_next_entity+0x7f/0x160

Sure, I'll be glad to take them if I can get an ack from the patch
authors, and reviewers (added to: and cc:)...

Ingo and Peter, any objections?

thanks,

greg k-h

Re: Two sched patchs missing from 4.4-stable: inclusion possible?

[Peter Zijlstra]

On Thu, May 18, 2017 at 11:35:00AM +0200, Greg KH wrote:
> On Thu, May 18, 2017 at 11:11:23AM +0200, Ben Pineau wrote:
> > Hello,
> > 
> > The following two patches series, merged in mainline 4.7, was Cc'ed to stable@,
> > but somehow didn't made its way to the 4.4.x stable tree:
> >   754bd598be9bbc953bc709a9e8ed7f3188bfb9d7
> >   094f469172e00d6ab0a3130b0e01c83b3cf3a98d

It would be so much easier if that were normal quotes like:

754bd598be9b ("sched/fair: Do not announce throttled next buddy in dequeue_task_fair()")
094f469172e0 ("sched/fair: Initialize throttle_count for new task-groups lazily")

Then I'd not have had to look them up. And I'm most certainly not going
to click through various interweb sites to provide answers.

> Sure, I'll be glad to take them if I can get an ack from the patch
> authors, and reviewers (added to: and cc:)...
> 
> Ingo and Peter, any objections?

No, those patches should be fine I think.

Re: Two sched patchs missing from 4.4-stable: inclusion possible?

[Greg KH]

On Thu, May 18, 2017 at 12:51:03PM +0200, Peter Zijlstra wrote:
> On Thu, May 18, 2017 at 11:35:00AM +0200, Greg KH wrote:
> > On Thu, May 18, 2017 at 11:11:23AM +0200, Ben Pineau wrote:
> > > Hello,
> > > 
> > > The following two patches series, merged in mainline 4.7, was Cc'ed to stable@,
> > > but somehow didn't made its way to the 4.4.x stable tree:
> > >   754bd598be9bbc953bc709a9e8ed7f3188bfb9d7
> > >   094f469172e00d6ab0a3130b0e01c83b3cf3a98d
> 
> It would be so much easier if that were normal quotes like:
> 
> 754bd598be9b ("sched/fair: Do not announce throttled next buddy in dequeue_task_fair()")
> 094f469172e0 ("sched/fair: Initialize throttle_count for new task-groups lazily")
> 
> Then I'd not have had to look them up. And I'm most certainly not going
> to click through various interweb sites to provide answers.
> 
> > Sure, I'll be glad to take them if I can get an ack from the patch
> > authors, and reviewers (added to: and cc:)...
> > 
> > Ingo and Peter, any objections?
> 
> No, those patches should be fine I think.

Thanks, now queued up.

greg k-h