All of lore.kernel.org
 help / color / mirror / Atom feed
* [GIT PULL V2] Changes for 4.8
@ 2016-07-27  3:59 Juergen Gross
  2016-07-27 17:03 ` Linus Torvalds
  0 siblings, 1 reply; 4+ messages in thread
From: Juergen Gross @ 2016-07-27  3:59 UTC (permalink / raw)
  To: Linus Torvalds, Linux Kernel Mailing List

[-- Attachment #1: Type: text/plain, Size: 2189 bytes --]

Hi Linus,

please consider pulling a patch series for 4.8 from:

  https://github.com/jgross1/linux.git tags/for-linus-4-8

Support calling functions on dedicated physical cpu

Some hardware (e.g. Dell Studio laptops) require special functions to
be called on physical cpu 0 in order to avoid occasional hangs. When
running as dom0 under Xen this could be achieved only via special boot
parameters (vcpu pinning) limiting the hypervisor in it's scheduling
decisions.

This patch series is adding a generic function to be able to temporarily
pin a (virtual) cpu to a dedicated physical cpu for executing above
mentioned functions on that specific cpu. The drivers (dcdbas and i8k)
requiring this functionality are modified accordingly.

Unfortunately 2 of the 6 patches got no Acks as the maintainers didn't
react in spite of multiple pings and resends. The core modification in
the scheduler got an Ack from Peter and multiple tests showed no
regressions.

As the series is touching multiple subsystems I couldn't find anyone
willing to take the series via his tree (I tried Ingo, Thomas, Peter).

Juergen Gross (6):
      xen: sync xen header
      virt, sched: add generic vcpu pinning support
      smp: add function to execute a function synchronously on a cpu
      xen: add xen_pin_vcpu() to support calling functions on a
dedicated pcpu
      dcdbas: make use of smp_call_on_cpu()
      hwmon: use smp_call_on_cpu() for dell-smm i8k

 MAINTAINERS                       |   1 +
 arch/x86/include/asm/hypervisor.h |   4 ++
 arch/x86/kernel/cpu/hypervisor.c  |  11 +++++
 arch/x86/xen/enlighten.c          |  40 +++++++++++++++
 drivers/firmware/dcdbas.c         |  51 +++++++++----------
 drivers/hwmon/dell-smm-hwmon.c    |  36 ++++++++------
 include/linux/hypervisor.h        |  17 +++++++
 include/linux/smp.h               |   3 ++
 include/xen/interface/sched.h     | 100
+++++++++++++++++++++++++++++++-------
 kernel/smp.c                      |  51 +++++++++++++++++++
 kernel/up.c                       |  18 +++++++
 11 files changed, 273 insertions(+), 59 deletions(-)
 create mode 100644 include/linux/hypervisor.h




[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [GIT PULL V2] Changes for 4.8
  2016-07-27  3:59 [GIT PULL V2] Changes for 4.8 Juergen Gross
@ 2016-07-27 17:03 ` Linus Torvalds
  2016-07-27 18:50   ` Juergen Gross
  2016-08-11 13:23   ` Juergen Gross
  0 siblings, 2 replies; 4+ messages in thread
From: Linus Torvalds @ 2016-07-27 17:03 UTC (permalink / raw)
  To: Juergen Gross; +Cc: Linux Kernel Mailing List

On Tue, Jul 26, 2016 at 8:59 PM, Juergen Gross <jgross@suse.com> wrote:
>
> Support calling functions on dedicated physical cpu
>
> Some hardware (e.g. Dell Studio laptops) require special functions to
> be called on physical cpu 0 in order to avoid occasional hangs. When
> running as dom0 under Xen this could be achieved only via special boot
> parameters (vcpu pinning) limiting the hypervisor in it's scheduling
> decisions.
>
> This patch series is adding a generic function to be able to temporarily
> pin a (virtual) cpu to a dedicated physical cpu for executing above
> mentioned functions on that specific cpu. The drivers (dcdbas and i8k)
> requiring this functionality are modified accordingly.
>
> Unfortunately 2 of the 6 patches got no Acks as the maintainers didn't
> react in spite of multiple pings and resends. The core modification in
> the scheduler got an Ack from Peter and multiple tests showed no
> regressions.
>
> As the series is touching multiple subsystems I couldn't find anyone
> willing to take the series via his tree (I tried Ingo, Thomas, Peter).

Thanks, this explanation was exactly what I was looking for.

I'm still at two minds whether I'd prefer this kind of thing to come
in through the x86 maintainers, or possibly the virtualization people
who would actually use this. Or whether I'll just take it directly.

I will take another look later when I've gotten rid of the big initial
flood of pull requests.

Thanks,

               Linus

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [GIT PULL V2] Changes for 4.8
  2016-07-27 17:03 ` Linus Torvalds
@ 2016-07-27 18:50   ` Juergen Gross
  2016-08-11 13:23   ` Juergen Gross
  1 sibling, 0 replies; 4+ messages in thread
From: Juergen Gross @ 2016-07-27 18:50 UTC (permalink / raw)
  To: Linus Torvalds; +Cc: Linux Kernel Mailing List

On 27/07/16 19:03, Linus Torvalds wrote:
> On Tue, Jul 26, 2016 at 8:59 PM, Juergen Gross <jgross@suse.com> wrote:
>>
>> Support calling functions on dedicated physical cpu
>>
>> Some hardware (e.g. Dell Studio laptops) require special functions to
>> be called on physical cpu 0 in order to avoid occasional hangs. When
>> running as dom0 under Xen this could be achieved only via special boot
>> parameters (vcpu pinning) limiting the hypervisor in it's scheduling
>> decisions.
>>
>> This patch series is adding a generic function to be able to temporarily
>> pin a (virtual) cpu to a dedicated physical cpu for executing above
>> mentioned functions on that specific cpu. The drivers (dcdbas and i8k)
>> requiring this functionality are modified accordingly.
>>
>> Unfortunately 2 of the 6 patches got no Acks as the maintainers didn't
>> react in spite of multiple pings and resends. The core modification in
>> the scheduler got an Ack from Peter and multiple tests showed no
>> regressions.
>>
>> As the series is touching multiple subsystems I couldn't find anyone
>> willing to take the series via his tree (I tried Ingo, Thomas, Peter).
> 
> Thanks, this explanation was exactly what I was looking for.
> 
> I'm still at two minds whether I'd prefer this kind of thing to come
> in through the x86 maintainers, or possibly the virtualization people
> who would actually use this. Or whether I'll just take it directly.

Okay.

Ingo already signaled he would be willing to take the series into one
of the -tip trees in case you don't want to take it.

I already asked David Vrabel whether he would take it via the Xen tree
and he was a little bit hesitant as some non-Xen patches got no Ack.
OTOH I'm one of the Xen maintainers, so taking the patches from me would
be fine regarding your statement you'd take the patches from "the
virtualization people". :-)

> I will take another look later when I've gotten rid of the big initial
> flood of pull requests.

Thanks,

Juergen

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: Re: [GIT PULL V2] Changes for 4.8
  2016-07-27 17:03 ` Linus Torvalds
  2016-07-27 18:50   ` Juergen Gross
@ 2016-08-11 13:23   ` Juergen Gross
  1 sibling, 0 replies; 4+ messages in thread
From: Juergen Gross @ 2016-08-11 13:23 UTC (permalink / raw)
  To: Linus Torvalds; +Cc: Linux Kernel Mailing List

On 27/07/16 19:03, Linus Torvalds wrote:
> On Tue, Jul 26, 2016 at 8:59 PM, Juergen Gross <jgross@suse.com> wrote:
>>
>> Support calling functions on dedicated physical cpu
>>
>> Some hardware (e.g. Dell Studio laptops) require special functions to
>> be called on physical cpu 0 in order to avoid occasional hangs. When
>> running as dom0 under Xen this could be achieved only via special boot
>> parameters (vcpu pinning) limiting the hypervisor in it's scheduling
>> decisions.
>>
>> This patch series is adding a generic function to be able to temporarily
>> pin a (virtual) cpu to a dedicated physical cpu for executing above
>> mentioned functions on that specific cpu. The drivers (dcdbas and i8k)
>> requiring this functionality are modified accordingly.
>>
>> Unfortunately 2 of the 6 patches got no Acks as the maintainers didn't
>> react in spite of multiple pings and resends. The core modification in
>> the scheduler got an Ack from Peter and multiple tests showed no
>> regressions.
>>
>> As the series is touching multiple subsystems I couldn't find anyone
>> willing to take the series via his tree (I tried Ingo, Thomas, Peter).
> 
> Thanks, this explanation was exactly what I was looking for.
> 
> I'm still at two minds whether I'd prefer this kind of thing to come
> in through the x86 maintainers, or possibly the virtualization people
> who would actually use this. Or whether I'll just take it directly.
> 
> I will take another look later when I've gotten rid of the big initial
> flood of pull requests.

Did you already decide what you want to do?

Thanks,

Juergen

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2016-08-11 13:23 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-07-27  3:59 [GIT PULL V2] Changes for 4.8 Juergen Gross
2016-07-27 17:03 ` Linus Torvalds
2016-07-27 18:50   ` Juergen Gross
2016-08-11 13:23   ` Juergen Gross

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.