On Tue, 2014-05-13 at 10:08 -0400, Rik van Riel wrote:
> OK, after doing some other NUMA stuff, and then looking at the scheduler
> again with a fresh mind, I have drawn some more conclusions about what
> the scheduler does, and how it breaks NUMA locality :)
>
> 1) If the node_distance between
On 05/09/2014 11:54 PM, Mike Galbraith wrote:
> On Fri, 2014-05-09 at 14:16 -0400, Rik van Riel wrote:
>
>> That leaves the big question: do we want to fall back to
>> prev_cpu if it is not idle, and it has an idle sibling,
>> or would it be better to find an idle sibling of prev_cpu
>> when we
On 05/09/2014 11:54 PM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 14:16 -0400, Rik van Riel wrote:
That leaves the big question: do we want to fall back to
prev_cpu if it is not idle, and it has an idle sibling,
or would it be better to find an idle sibling of prev_cpu
when we wake up a
On Tue, 2014-05-13 at 10:08 -0400, Rik van Riel wrote:
OK, after doing some other NUMA stuff, and then looking at the scheduler
again with a fresh mind, I have drawn some more conclusions about what
the scheduler does, and how it breaks NUMA locality :)
1) If the node_distance between nodes
On Fri, 2014-05-09 at 14:16 -0400, Rik van Riel wrote:
> That leaves the big question: do we want to fall back to
> prev_cpu if it is not idle, and it has an idle sibling,
> or would it be better to find an idle sibling of prev_cpu
> when we wake up a task?
Yes. If there was A correct answer,
On 05/09/2014 01:55 PM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 11:24 -0400, Rik van Riel wrote:
On 05/09/2014 11:24 AM, Mike Galbraith wrote:
If no ->flags & SD_BALANCE_WAKE is encountered during traversal, sd
remains NULL, we fall through to return prev_cpu.
We do
On Fri, 2014-05-09 at 11:24 -0400, Rik van Riel wrote:
> On 05/09/2014 11:24 AM, Mike Galbraith wrote:
> > If no ->flags & SD_BALANCE_WAKE is encountered during traversal, sd
> > remains NULL, we fall through to return prev_cpu.
> We do fall through, but into this loop:
>
On 05/09/2014 11:24 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 10:22 -0400, Rik van Riel wrote:
On 05/09/2014 03:34 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel wrote:
Looks like SD_BALANCE_WAKE is not
On Fri, 2014-05-09 at 10:22 -0400, Rik van Riel wrote:
> On 05/09/2014 03:34 AM, Mike Galbraith wrote:
> > On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
> >> On Thu, 08 May 2014 22:20:25 -0400
> >> Rik van Riel wrote:
> >>
> >>> Looks like SD_BALANCE_WAKE is not gotten from the sd
On 05/09/2014 03:34 AM, Mike Galbraith wrote:
> On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
>> On Thu, 08 May 2014 22:20:25 -0400
>> Rik van Riel wrote:
>>
>>> Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
>>> all, but passed into select_task_rq by try_to_wake_up, as a
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
> On Thu, 08 May 2014 22:20:25 -0400
> Rik van Riel wrote:
>
> > Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
> > all, but passed into select_task_rq by try_to_wake_up, as a
> > hard coded sd_flags argument.
>
> > Should
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel r...@redhat.com wrote:
Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
all, but passed into select_task_rq by try_to_wake_up, as a
hard coded sd_flags argument.
On 05/09/2014 03:34 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel r...@redhat.com wrote:
Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
all, but passed into select_task_rq by try_to_wake_up, as
On Fri, 2014-05-09 at 10:22 -0400, Rik van Riel wrote:
On 05/09/2014 03:34 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel r...@redhat.com wrote:
Looks like SD_BALANCE_WAKE is not gotten from the sd flags
On 05/09/2014 11:24 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 10:22 -0400, Rik van Riel wrote:
On 05/09/2014 03:34 AM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 01:27 -0400, Rik van Riel wrote:
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel r...@redhat.com wrote:
Looks like
On Fri, 2014-05-09 at 11:24 -0400, Rik van Riel wrote:
On 05/09/2014 11:24 AM, Mike Galbraith wrote:
If no -flags SD_BALANCE_WAKE is encountered during traversal, sd
remains NULL, we fall through to return prev_cpu.
We do fall through, but into this loop:
On 05/09/2014 01:55 PM, Mike Galbraith wrote:
On Fri, 2014-05-09 at 11:24 -0400, Rik van Riel wrote:
On 05/09/2014 11:24 AM, Mike Galbraith wrote:
If no -flags SD_BALANCE_WAKE is encountered during traversal, sd
remains NULL, we fall through to return prev_cpu.
We do
On Fri, 2014-05-09 at 14:16 -0400, Rik van Riel wrote:
That leaves the big question: do we want to fall back to
prev_cpu if it is not idle, and it has an idle sibling,
or would it be better to find an idle sibling of prev_cpu
when we wake up a task?
Yes. If there was A correct answer, this
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel wrote:
> Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
> all, but passed into select_task_rq by try_to_wake_up, as a
> hard coded sd_flags argument.
> Should we do that, if SD_WAKE_BALANCE is not set for any sched domain?
I
On Thu, 08 May 2014 22:20:25 -0400
Rik van Riel r...@redhat.com wrote:
Looks like SD_BALANCE_WAKE is not gotten from the sd flags at
all, but passed into select_task_rq by try_to_wake_up, as a
hard coded sd_flags argument.
Should we do that, if SD_WAKE_BALANCE is not set for any sched
20 matches
Mail list logo