load_balance() fixes for affinity

We noticed in testing that affined workloads do not provide consistent performance under certain circumstances. To isolate the issue, we began testing with a representative CPU workload. In some cases the CPU workload results would vary by a large percentage from run to run. We used JTAG and scheduler trace to try and profile the issue and noticed that at least one core was left idle for the duration of the test in the low score runs.

The specific test methodology used in our investigation involved using an executable that was compiled to fork a specific number of workers. The executable was then affined to a number of cores equal to the number of workers forked, using the userspace taskset utility. The expectation was that all cores in the mask of runnable CPUs would run a process until the work was complete. In practice, for several affinity masks, one to two cores would not receive work until late in the test or not at all.

Our first observation was that it appeared initial cpu assignment of the forked threads appeared suboptimal. We observed that it was highly probable that many or all of the threads would be initially assigned to the same CPU. Our limited investigation into this indicated that the forking of the threads occurs fast enough that load estimates for CPUs have not adjusted to the work they have been assigned. The function select_task_rq_fair() does not take number of running process into consideration and thus can easily assign several tasks spawned close together to the same CPU within the allowed CPUs for a workload. While this seems suboptimal, the load_balance() algorithm run on other idle cores in the affinity mask should resolve imbalances that result from it.

As our test case was leaving cores idle for many seconds at a time, we dug into load_balance() code and specifically the group_imbalance path. We added ftrace to the path in question and noticed that two branches in load_balance() conflict in their assessments of the ability of the algorithm to migrate load. The group_imbalance path correctly sets the flag to indicate the group can not be properly balanced due to affinity, but the redo condition right after this branch incorrectly assumes that there may be other cores with work to be pulled by considering cores outside of the scheduling domain in question.

Consider the following circumstance where (T) represents tasks and (*) represents the allowed CPUs:

A B C
{ 0 1 } { 2 3 } { 4 5 }
T T T T
T
* * * * *

In this case, core 0 in scheduling group 'A' should run load balance on its local scheduling domain and try to pull work from core 1. When it fails to move any work due to affinity, it will mark scheduling group 'A' as "group_imbalance". However, right after doing so, core 0 load_balance() evaluates the redo path with core 1 removed from consideration. This should leave no other cores to consider and result in a jump to "out_all_pinned", leaving the "group_imbalance" flag to be seen by future runs of the algorithm on core 3. As the code stands currently, core 0 evaluates all cores instead of those in the scheduling domain under consideration. It incorrectly sees other cores in the system and attempts a redo. With core 1 removed from the mask, the redo identifies no load imbalance and results in a jump to "out_balanced", clearing the "group_imbalance" flag.

We propose correcting this logic with patch 1 of this series.

Making this correction revealed an issue in the calculate_imbalance() path. Patch 2 removes a branch that does not make sense with the current load_balance() algorithm because it has no scenario where it benifits the "group_imbalance" case in calculate_imbalance() and which causes problems in systems with affined workloads and many idle or lightly loaded CPUs.

[V5]
-updated comment to explain the "why" behind the redo check -fixed panic triggered from active_load_balance_cpu_stop()

[V4]
-restricted active cpus mask to the domain span -fixed dst_cpu masking logic to work for the entire local group

[V3]
-fixed botched subject lines

[V2]
-Corrected use of Signed-off-by tags
-Removed temp cpumask variable from stack

Jeffrey Hugo (2):
sched/fair: Fix load_balance() affinity redo path sched/fair: Remove group imbalance from calculate_imbalance()

 kernel/sched/fair.c | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

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.

If load_balance() fails to migrate any tasks because all tasks were affined, load_balance() removes the source cpu from consideration and attempts to redo and balance among the new subset of cpus.

There is a bug in this code path where the algorithm considers all active cpus in the system (minus the source that was just masked out). This is not valid for two reasons: some active cpus may not be in the current scheduling domain and one of the active cpus is dst_cpu. These cpus should not be considered, as we cannot pull load from them.

Instead of failing out of load_balance(), we may end up redoing the search with no valid cpus and incorrectly concluding the domain is balanced. Additionally, if the group_imbalance flag was just set, it may also be incorrectly unset, thus the flag will not be seen by other cpus in future load_balance() runs as that algorithm intends.

Fix the check by removing cpus not in the current domain and the dst_cpu from considertation, thus limiting the evaluation to valid remaining cpus from which load might be migrated.

---
 kernel/sched/fair.c | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)

index d711093..2ba4407 100644
@@ -6737,10 +6737,10 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env)
 		 * our sched_group. We may want to revisit it if we couldn't
 		 * meet load balance goals by pulling other tasks on src_cpu.
 		 *
-		 * Also avoid computing new_dst_cpu if we have already computed
-		 * one in current iteration.
+		 * Avoid computing new_dst_cpu for NEWLY_IDLE or if we have
+		 * already computed one in current iteration.
 		 */
-		if (!env->dst_grpmask || (env->flags & LBF_DST_PINNED))
+		if (env->idle == CPU_NEWLY_IDLE || (env->flags & LBF_DST_PINNED))
 			return 0;
 
 		/* Prevent to re-select dst_cpu via env's cpus */

@@ -8091,14 +8091,7 @@ static int load_balance(int this_cpu, struct rq *this_rq,
 		.tasks		= LIST_HEAD_INIT(env.tasks),
 	};
 
-	/*
-	 * For NEWLY_IDLE load_balancing, we don't need to consider
-	 * other cpus in our group
-	 */
-	if (idle == CPU_NEWLY_IDLE)
-		env.dst_grpmask = NULL;
-
-	cpumask_copy(cpus, cpu_active_mask);
+	cpumask_and(cpus, sched_domain_span(sd), cpu_active_mask);
 
 	schedstat_inc(sd->lb_count[idle]);
 

@@ -8220,7 +8213,15 @@ static int load_balance(int this_cpu, struct rq *this_rq,
 		/* All tasks on this runqueue were pinned by CPU affinity */
 		if (unlikely(env.flags & LBF_ALL_PINNED)) {
 			cpumask_clear_cpu(cpu_of(busiest), cpus);
-			if (!cpumask_empty(cpus)) {
+			/*
+			 * Attempting to continue load balancing at the current
+			 * sched_domain level only makes sense if there are
+			 * active cpus remaining as possible busiest cpus to
+			 * pull load from which are not contained within the
+			 * destination group that is receiving any migrated
+			 * load.
+			 */
+			if (!cpumask_subset(cpus, env.dst_grpmask)) {
 				env.loop = 0;
 				env.loop_break = sched_nr_migrate_break;
 				goto redo;

@@ -8516,6 +8517,13 @@ static int active_load_balance_cpu_stop(void *data)
 			.src_cpu	= busiest_rq->cpu,
 			.src_rq		= busiest_rq,
 			.idle		= CPU_IDLE,
+			/*
+			 * can_migrate_task() doesn't need to compute new_dst_cpu
+			 * for active balancing. Since we have CPU_IDLE, but no
+			 * @dst_grpmask we need to make that test go away with lying
+			 * about DST_PINNED.
+			 */
+			.flags		= LBF_DST_PINNED,
 		};
 
 		schedstat_inc(sd->alb_count);

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.

Yes, this looks good. Thanks!

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.

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.

---
 kernel/sched/fair.c | 9 ---------
 1 file changed, 9 deletions(-)

index 84255ab..3600713 100644
@@ -7760,15 +7760,6 @@ static inline void calculate_imbalance(struct lb_env *env, struct sd_lb_stats *s
 	local = &sds->local_stat;
 	busiest = &sds->busiest_stat;
 
-	if (busiest->group_type == group_imbalanced) {
-		/*
-		 * In the group_imb case we cannot rely on group-wide averages
-		 * to ensure cpu-load equilibrium, look at wider averages. XXX
-		 */
-		busiest->load_per_task =
-			min(busiest->load_per_task, sds->avg_load);
-	}
-
 	/*
 	 * Avg load of busiest sg can be less and avg load of local sg can
 	 * be greater than avg load across all sgs of sd because avg load

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.

It would be nice to have some more information on which patch(es) changed that.

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 :/

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.

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.

Peter (and/or Ingo?), is this series likely to make 4.13? V5 has been quiet, so I'm not sure if folks are happy with things, or just haven't had time to review.

Thanks

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.

Ping?

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.

Ping again.

Since there has been no comments for the past 3 weeks, I'm assuming everyone is happy with this series. With 4.12-rc7 out, and Linus saying it looks good, it seems likely that the merge window is going to open next week. We really want this in 4.13.

Would you please confirm that the series will be pulled into 4.13?

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.

Sorry, I had to deal with becoming a parent for the 3rd time, let me go stare at them ;-)