[lkml]   [2017]   [Jul]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH V5 2/2] sched/fair: Remove group imbalance from calculate_imbalance()
On 7/5/2017 5:22 AM, Peter Zijlstra wrote:
> On Wed, Jun 07, 2017 at 01:18:58PM -0600, Jeffrey Hugo wrote:
>> The group_imbalance path in calculate_imbalance() made sense when it was
>> added back in 2007 with commit 908a7c1b9b80 ("sched: fix improper load
>> balance across sched domain") because busiest->load_per_task factored into
>> the amount of imbalance that was calculated. That is not the case today.
> It would be nice to have some more information on which patch(es)
> changed that.

From the history it looks like commit dd5feea14a7d ("sched: Fix
SCHED_MC regression caused by change in sched cpu_power") removed
load_per_task from imbalance calculations (was factored into max_pull).
After this change, the group imbalance modifications to this variable
appear to no longer work as originally intended.

>> The group_imbalance path can only affect the outcome of
>> calculate_imbalance() when the average load of the domain is less than the
>> original busiest->load_per_task. In this case, busiest->load_per_task is
>> overwritten with the scheduling domain load average. Thus
>> busiest->load_per_task no longer represents actual load that can be moved.
>> At the final comparison between env->imbalance and busiest->load_per_task,
>> imbalance may be larger than the new busiest->load_per_task causing the
>> check to fail under the assumption that there is a task that could be
>> migrated to satisfy the imbalance. However env->imbalance may still be
>> smaller than the original busiest->load_per_task, thus it is unlikely that
>> there is a task that can be migrated to satisfy the imbalance.
>> Calculate_imbalance() would not choose to run fix_small_imbalance() when we
>> expect it should. In the worst case, this can result in idle cpus.
>> Since the group imbalance path in calculate_imbalance() is at best a NOP
>> but otherwise harmful, remove it.
> load_per_task is horrible and should die. Ever since we did cgroup
> support the number is complete crap, but even before that the concept
> was dubious.
> Most of the logic that uses the number stems from the pre-smp-nice era.
> This also of course means that fix_small_imbalance() is probably a load
> of crap. Digging through all that has been on the todo list for a long
> while but somehow not something I've ever gotten to :/

Based on the state of the code today, our change fixes a current issue
that we are encoutering. If we add the above history to the commit text,
is our change sufficent as a short term solution between now and
whenever the load_per_task path is rearchitected?

Jeffrey Hugo
Qualcomm Datacenter Technologies as an affiliate of Qualcomm
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

 \ /
  Last update: 2017-07-07 17:27    [W:0.044 / U:0.556 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site