The wake-affine scheduler feature is currently always trying to pull
the wakee close to the waker. In theory this should be beneficial if
the waker's CPU caches hot data for the wakee, and it's also beneficial
in the extreme ping-pong high context switch rate case.
Testing shows it can benefit hackbench up to 15%.
However, the feature is somewhat blind, from which some workloads
such as pgbench suffer. It's also time-consuming algorithmically.
Testing shows it can damage pgbench up to 50% - far more than the
benefit it brings in the best case.
So wake-affine should be smarter and it should realize when to
stop its thankless effort at trying to find a suitable CPU to wake on.
This patch introduces 'wakee_flips', which will be increased each
time the task flips (switches) its wakee target.
So a high 'wakee_flips' value means the task has more than one
wakee, and the bigger the number, the higher the wakeup frequency.
Now when making the decision on whether to pull or not, pay attention to
the wakee with a high 'wakee_flips', pulling such a task may benefit
the wakee. Also imply that the waker will face cruel competition later,
it could be very cruel or very fast depends on the story behind
'wakee_flips', waker therefore suffers.
Furthermore, if waker also has a high 'wakee_flips', that implies that
multiple tasks rely on it, then waker's higher latency will damage all
of them, so pulling wakee seems to be a bad deal.
Thus, when 'waker->wakee_flips / wakee->wakee_flips' becomes
higher and higher, the cost of pulling seems to be worse and worse.
The patch therefore helps the wake-affine feature to stop its pulling
work when:
wakee->wakee_flips > factor &&
waker->wakee_flips > (factor * wakee->wakee_flips)
The 'factor' here is the number of CPUs in the current CPU's NUMA node,
so a bigger node will lead to more pulling since the trial becomes more
severe.
After applying the patch, pgbench shows up to 40% improvements and no regressions.
Tested with 12 cpu x86 server and tip 3.10.0-rc7.
The percentages in the final column highlight the areas with the biggest wins,
all other areas improved as well:
pgbench base smart
| db_size | clients | tps | | tps |
+---------+---------+-------+ +-------+
| 22 MB | 1 | 10598 | | 10796 |
| 22 MB | 2 | 21257 | | 21336 |
| 22 MB | 4 | 41386 | | 41622 |
| 22 MB | 8 | 51253 | | 57932 |
| 22 MB | 12 | 48570 | | 54000 |
| 22 MB | 16 | 46748 | | 55982 | +19.75%
| 22 MB | 24 | 44346 | | 55847 | +25.93%
| 22 MB | 32 | 43460 | | 54614 | +25.66%
| 7484 MB | 1 | 8951 | | 9193 |
| 7484 MB | 2 | 19233 | | 19240 |
| 7484 MB | 4 | 37239 | | 37302 |
| 7484 MB | 8 | 46087 | | 50018 |
| 7484 MB | 12 | 42054 | | 48763 |
| 7484 MB | 16 | 40765 | | 51633 | +26.66%
| 7484 MB | 24 | 37651 | | 52377 | +39.11%
| 7484 MB | 32 | 37056 | | 51108 | +37.92%
| 15 GB | 1 | 8845 | | 9104 |
| 15 GB | 2 | 19094 | | 19162 |
| 15 GB | 4 | 36979 | | 36983 |
| 15 GB | 8 | 46087 | | 49977 |
| 15 GB | 12 | 41901 | | 48591 |
| 15 GB | 16 | 40147 | | 50651 | +26.16%
| 15 GB | 24 | 37250 | | 52365 | +40.58%
| 15 GB | 32 | 36470 | | 50015 | +37.14%
Signed-off-by: Michael Wang <wangyun@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/51D50057.9000809@linux.vnet.ibm.com
[ Improved the changelog. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The bad thing about update_h_load(), which computes hierarchical load
factor for task groups, is that it is called for each task group in the
system before every load balancer run, and since rebalance can be
triggered very often, this function can eat really a lot of cpu time if
there are many cpu cgroups in the system.
Although the situation was improved significantly by commit a35b646
('sched, cgroup: Reduce rq->lock hold times for large cgroup
hierarchies'), the problem still can arise under some kinds of loads,
e.g. when cpus are switching from idle to busy and back very frequently.
For instance, when I start 1000 of processes that wake up every
millisecond on my 8 cpus host, 'top' and 'perf top' show:
Cpu(s): 17.8%us, 24.3%sy, 0.0%ni, 57.9%id, 0.0%wa, 0.0%hi, 0.0%si
Events: 243K cycles
7.57% [kernel] [k] __schedule
7.08% [kernel] [k] timerqueue_add
6.13% libc-2.12.so [.] usleep
Then if I create 10000 *idle* cpu cgroups (no processes in them), cpu
usage increases significantly although the 'wakers' are still executing
in the root cpu cgroup:
Cpu(s): 19.1%us, 48.7%sy, 0.0%ni, 31.6%id, 0.0%wa, 0.0%hi, 0.7%si
Events: 230K cycles
24.56% [kernel] [k] tg_load_down
5.76% [kernel] [k] __schedule
This happens because this particular kind of load triggers 'new idle'
rebalance very frequently, which requires calling update_h_load(),
which, in turn, calls tg_load_down() for every *idle* cpu cgroup even
though it is absolutely useless, because idle cpu cgroups have no tasks
to pull.
This patch tries to improve the situation by making h_load calculation
proceed only when h_load is really necessary. To achieve this, it
substitutes update_h_load() with update_cfs_rq_h_load(), which computes
h_load only for a given cfs_rq and all its ascendants, and makes the
load balancer call this function whenever it considers if a task should
be pulled, i.e. it moves h_load calculations directly to task_h_load().
For h_load of the same cfs_rq not to be updated multiple times (in case
several tasks in the same cgroup are considered during the same balance
run), the patch keeps the time of the last h_load update for each cfs_rq
and breaks calculation when it finds h_load to be uptodate.
The benefit of it is that h_load is computed only for those cfs_rq's,
which really need it, in particular all idle task groups are skipped.
Although this, in fact, moves h_load calculation under rq lock, it
should not affect latency much, because the amount of work done under rq
lock while trying to pull tasks is limited by sched_nr_migrate.
After the patch applied with the setup described above (1000 wakers in
the root cgroup and 10000 idle cgroups), I get:
Cpu(s): 16.9%us, 24.8%sy, 0.0%ni, 58.4%id, 0.0%wa, 0.0%hi, 0.0%si
Events: 242K cycles
7.57% [kernel] [k] __schedule
6.70% [kernel] [k] timerqueue_add
5.93% libc-2.12.so [.] usleep
Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1373896159-1278-1-git-send-email-vdavydov@parallels.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The __cpuinit type of throwaway sections might have made sense
some time ago when RAM was more constrained, but now the savings
do not offset the cost and complications. For example, the fix in
commit 5e427ec2d0 ("x86: Fix bit corruption at CPU resume time")
is a good example of the nasty type of bugs that can be created
with improper use of the various __init prefixes.
After a discussion on LKML[1] it was decided that cpuinit should go
the way of devinit and be phased out. Once all the users are gone,
we can then finally remove the macros themselves from linux/init.h.
This removes all the uses of the __cpuinit macros from C files in
the core kernel directories (kernel, init, lib, mm, and include)
that don't really have a specific maintainer.
[1] https://lkml.org/lkml/2013/5/20/589
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Similar to runnable_load_avg, blocked_load_avg variable, long type is
enough for removed_load in 64 bit or 32 bit machine.
Then we avoid the expensive atomic64 operations on 32 bit machine.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Reviewed-by: Paul Turner <pjt@google.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-12-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Since tg->load_avg is smaller than tg->load_weight, we don't need a
atomic64_t variable for load_avg in 32 bit machine.
The same reason for cfs_rq->tg_load_contrib.
The atomic_long_t/unsigned long variable type are more efficient and
convenience for them.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-11-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Since the 'u64 runnable_load_avg, blocked_load_avg' in cfs_rq struct are
smaller than 'unsigned long' cfs_rq->load.weight. We don't need u64
vaiables to describe them. unsigned long is more efficient and convenience.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Reviewed-by: Paul Turner <pjt@google.com>
Tested-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-10-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Aside from using runnable load average in background, move_tasks is
also the key function in load balance. We need consider the runnable
load average in it in order to make it an apple to apple load
comparison.
Morten had caught a div u64 bug on ARM, thanks!
Thanks-to: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Alex Shi <alex.shi@intel.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-8-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
They are the base values in load balance, update them with rq runnable
load average, then the load balance will consider runnable load avg
naturally.
We also try to include the blocked_load_avg as cpu load in balancing,
but that cause kbuild performance drop 6% on every Intel machine, and
aim7/oltp drop on some of 4 CPU sockets machines.
Or only add blocked_load_avg into get_rq_runable_load, hackbench still
drop a little on NHM EX.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Reviewed-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-7-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The woken migrated task will __synchronize_entity_decay(se); in
migrate_task_rq_fair, then it needs to set
`se->avg.last_runnable_update -= (-se->avg.decay_count) << 20' before
update_entity_load_avg, in order to avoid sleep time is updated twice
for se.avg.load_avg_contrib in both __syncchronize and
update_entity_load_avg.
However if the sleeping task is woken up from the same cpu, it miss
the last_runnable_update before update_entity_load_avg(se, 0, 1), then
the sleep time was used twice in both functions. So we need to remove
the double sleep time accounting.
Paul also contributed some code comments in this commit.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Reviewed-by: Paul Turner <pjt@google.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-5-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
We need to initialize the se.avg.{decay_count, load_avg_contrib} for a
new forked task. Otherwise random values of above variables cause a
mess when a new task is enqueued:
enqueue_task_fair
enqueue_entity
enqueue_entity_load_avg
and make fork balancing imbalance due to incorrect load_avg_contrib.
Further more, Morten Rasmussen notice some tasks were not launched at
once after created. So Paul and Peter suggest giving a start value for
new task runnable avg time same as sched_slice().
PeterZ said:
> So the 'problem' is that our running avg is a 'floating' average; ie. it
> decays with time. Now we have to guess about the future of our newly
> spawned task -- something that is nigh impossible seeing these CPU
> vendors keep refusing to implement the crystal ball instruction.
>
> So there's two asymptotic cases we want to deal well with; 1) the case
> where the newly spawned program will be 'nearly' idle for its lifetime;
> and 2) the case where its cpu-bound.
>
> Since we have to guess, we'll go for worst case and assume its
> cpu-bound; now we don't want to make the avg so heavy adjusting to the
> near-idle case takes forever. We want to be able to quickly adjust and
> lower our running avg.
>
> Now we also don't want to make our avg too light, such that it gets
> decremented just for the new task not having had a chance to run yet --
> even if when it would run, it would be more cpu-bound than not.
>
> So what we do is we make the initial avg of the same duration as that we
> guess it takes to run each task on the system at least once -- aka
> sched_slice().
>
> Of course we can defeat this with wakeup/fork bombs, but in the 'normal'
> case it should be good enough.
Paul also contributed most of the code comments in this commit.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Reviewed-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Reviewed-by: Paul Turner <pjt@google.com>
[peterz; added explanation of sched_slice() usage]
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1371694737-29336-4-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Remove CONFIG_FAIR_GROUP_SCHED that covers the runnable info, then
we can use runnable load variables.
Also remove 2 CONFIG_FAIR_GROUP_SCHED setting which is not in reverted
patch(introduced in 9ee474f), but also need to revert.
Signed-off-by: Alex Shi <alex.shi@intel.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/51CA76A3.3050207@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
default_cfs_period(), do_sched_cfs_period_timer(), do_sched_cfs_slack_timer()
already defined previously, no need to declare again.
Signed-off-by: Michael Wang <wangyun@linux.vnet.ibm.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/51AD8808.7020608@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Directly use rq to save some code.
Signed-off-by: Michael Wang <wangyun@linux.vnet.ibm.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/51AD87EB.1070605@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Commit 78becc2709 ("sched: Use an accessor to read the rq clock")
introduces rq_clock(), which obsoletes the use of the "rq" variable
in expire_cfs_rq_runtime() and triggers this build warning:
kernel/sched/fair.c: In function 'expire_cfs_rq_runtime':
kernel/sched/fair.c:2159:13: warning: unused variable 'rq' [-Wunused-variable]
Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Turner <pjt@google.com>
Cc: peterz@infradead.org
Link: http://lkml.kernel.org/r/1369904660-14169-1-git-send-email-kamalesh@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Read the runqueue clock through an accessor. This
prepares for adding a debugging infrastructure to
detect missing or redundant calls to update_rq_clock()
between a scheduler's entry and exit point.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Paul Turner <pjt@google.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1365724262-20142-6-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
In this function we are making use of rq->clock right before the
update of the rq clock, let's just call update_rq_clock() just
before that to avoid using a stale rq clock value.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Paul Turner <pjt@google.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1365724262-20142-5-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Because we may update the execution time in
sched_group_set_shares()->update_cfs_shares()->reweight_entity()->update_curr()
before reweighting the entity while setting the group shares and this requires
an uptodate version of the runqueue clock.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Paul Turner <pjt@google.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1365724262-20142-3-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
It is a few instructions more efficent to and slightly more
readable to use this_rq()-> instead of cpu_rq(smp_processor_id())-> .
Size comparison of kernel/sched/fair.o:
text data bss dec hex filename
27972 122 26 28120 6dd8 fair.o.before
27956 122 26 28104 6dc8 fair.o.after
Signed-off-by: Nathan Zimmer <nzimmer@sgi.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1368116643-87971-1-git-send-email-nzimmer@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
These inlines are only used by kernel/sched/fair.c so they do
not need to be present in the main kernel/sched/sched.h file.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/1366398650-31599-3-git-send-email-paul.gortmaker@windriver.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The full dynticks tree needs the latest RCU and sched
upstream updates in order to fix some dependencies.
Merge a common upstream merge point that has these
updates.
Conflicts:
include/linux/perf_event.h
kernel/rcutree.h
kernel/rcutree_plugin.h
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
On my SMP platform which is made of 5 cores in 2 clusters, I
have the nr_busy_cpu field of sched_group_power struct that is
not null when the platform is fully idle - which makes the
scheduler unhappy.
The root cause is:
During the boot sequence, some CPUs reach the idle loop and set
their NOHZ_IDLE flag while waiting for others CPUs to boot. But
the nr_busy_cpus field is initialized later with the assumption
that all CPUs are in the busy state whereas some CPUs have
already set their NOHZ_IDLE flag.
More generally, the NOHZ_IDLE flag must be initialized when new
sched_domains are created in order to ensure that NOHZ_IDLE and
nr_busy_cpus are aligned.
This condition can be ensured by adding a synchronize_rcu()
between the destruction of old sched_domains and the creation of
new ones so the NOHZ_IDLE flag will not be updated with old
sched_domain once it has been initialized. But this solution
introduces a additionnal latency in the rebuild sequence that is
called during cpu hotplug.
As suggested by Frederic Weisbecker, another solution is to have
the same rcu lifecycle for both NOHZ_IDLE and sched_domain
struct. A new nohz_idle field is added to sched_domain so both
status and sched_domain will share the same RCU lifecycle and
will be always synchronized. In addition, there is no more need
to protect nohz_idle against concurrent access as it is only
modified by 2 exclusive functions called by local cpu.
This solution has been prefered to the creation of a new struct
with an extra pointer indirection for sched_domain.
The synchronization is done at the cost of :
- An additional indirection and a rcu_dereference for accessing nohz_idle.
- We use only the nohz_idle field of the top sched_domain.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: linaro-kernel@lists.linaro.org
Cc: peterz@infradead.org
Cc: fweisbec@gmail.com
Cc: pjt@google.com
Cc: rostedt@goodmis.org
Cc: efault@gmx.de
Link: http://lkml.kernel.org/r/1366729142-14662-1-git-send-email-vincent.guittot@linaro.org
[ Fixed !NO_HZ build bug. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Commit 88b8dac0 makes load_balance() consider other cpus in its
group. But, in that, there is no code for preventing to
re-select dst-cpu. So, same dst-cpu can be selected over and
over.
This patch add functionality to load_balance() in order to
exclude cpu which is selected once. We prevent to re-select
dst_cpu via env's cpus, so now, env's cpus is a candidate not
only for src_cpus, but also dst_cpus.
With this patch, we can remove lb_iterations and
max_lb_iterations, because we decide whether we can go ahead or
not via env's cpus.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: Jason Low <jason.low2@hp.com>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-7-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This name doesn't represent specific meaning.
So rename it to imply it's purpose.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: Jason Low <jason.low2@hp.com>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-6-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Currently, LBF_ALL_PINNED is cleared after affinity check is
passed. So, if task migration is skipped by small load value or
small imbalance value in move_tasks(), we don't clear
LBF_ALL_PINNED. At last, we trigger 'redo' in load_balance().
Imbalance value is often so small that any tasks cannot be moved
to other cpus and, of course, this situation may be continued
after we change the target cpu. So this patch move up affinity
check code and clear LBF_ALL_PINNED before evaluating load value
in order to mitigate useless redoing overhead.
In addition, re-order some comments correctly.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: Jason Low <jason.low2@hp.com>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-5-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Commit 88b8dac0 makes load_balance() consider other cpus in its
group, regardless of idle type. When we do NEWLY_IDLE balancing,
we should not consider it, because a motivation of NEWLY_IDLE
balancing is to turn this cpu to non idle state if needed. This
is not the case of other cpus. So, change code not to consider
other cpus for NEWLY_IDLE balancing.
With this patch, assign 'if (pulled_task) this_rq->idle_stamp =
0' in idle_balance() is corrected, because NEWLY_IDLE balancing
doesn't consider other cpus. Assigning to 'this_rq->idle_stamp'
is now valid.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Tested-by: Jason Low <jason.low2@hp.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-4-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
After commit 88b8dac0, dst-cpu can be changed in load_balance(),
then we can't know cpu_idle_type of dst-cpu when load_balance()
return positive. So, add explicit cpu_idle_type checking.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Tested-by: Jason Low <jason.low2@hp.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-3-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
cur_ld_moved is reset if env.flags hit LBF_NEED_BREAK.
So, there is possibility that we miss doing resched_cpu().
Correct it as changing position of resched_cpu()
before checking LBF_NEED_BREAK.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Tested-by: Jason Low <jason.low2@hp.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366705662-3587-2-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The current update of the rq's load can be erroneous when RT
tasks are involved.
The update of the load of a rq that becomes idle, is done only
if the avg_idle is less than sysctl_sched_migration_cost. If RT
tasks and short idle duration alternate, the runnable_avg will
not be updated correctly and the time will be accounted as idle
time when a CFS task wakes up.
A new idle_enter function is called when the next task is the
idle function so the elapsed time will be accounted as run time
in the load of the rq, whatever the average idle time is. The
function update_rq_runnable_avg is removed from idle_balance.
When a RT task is scheduled on an idle CPU, the update of the
rq's load is not done when the rq exit idle state because CFS's
functions are not called. Then, the idle_balance, which is
called just before entering the idle function, updates the rq's
load and makes the assumption that the elapsed time since the
last update, was only running time.
As a consequence, the rq's load of a CPU that only runs a
periodic RT task, is close to LOAD_AVG_MAX whatever the running
duration of the RT task is.
A new idle_exit function is called when the prev task is the
idle function so the elapsed time will be accounted as idle time
in the rq's load.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: linaro-kernel@lists.linaro.org
Cc: peterz@infradead.org
Cc: pjt@google.com
Cc: fweisbec@gmail.com
Cc: efault@gmx.de
Link: http://lkml.kernel.org/r/1366302867-5055-1-git-send-email-vincent.guittot@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
A comment in function rebalance_domains() mentions
arch_init_sched_domains(), but that function does not exist
anymore. The proper function is init_sched_domains().
Signed-off-by: Libin <huawei.libin@huawei.com>
Cc: <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1364814841-49156-1-git-send-email-huawei.libin@huawei.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
At this point tsk_cache_hot is always true, so no need to check it.
Signed-off-by: Zhang Hang <bob.zhanghang@huawei.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/51650107.9040606@huawei.com
[ Also remove unnecessary schedstat #ifdefs. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
We are planning to convert the dynticks Kconfig options layout
into a choice menu. The user must be able to easily pick
any of the following implementations: constant periodic tick,
idle dynticks, full dynticks.
As this implies a mutual exclusion, the two dynticks implementions
need to converge on the selection of a common Kconfig option in order
to ease the sharing of a common infrastructure.
It would thus seem pretty natural to reuse CONFIG_NO_HZ to
that end. It already implements all the idle dynticks code
and the full dynticks depends on all that code for now.
So ideally the choice menu would propose CONFIG_NO_HZ_IDLE and
CONFIG_NO_HZ_EXTENDED then both would select CONFIG_NO_HZ.
On the other hand we want to stay backward compatible: if
CONFIG_NO_HZ is set in an older config file, we want to
enable CONFIG_NO_HZ_IDLE by default.
But we can't afford both at the same time or we run into
a circular dependency:
1) CONFIG_NO_HZ_IDLE and CONFIG_NO_HZ_EXTENDED both select
CONFIG_NO_HZ
2) If CONFIG_NO_HZ is set, we default to CONFIG_NO_HZ_IDLE
We might be able to support that from Kconfig/Kbuild but it
may not be wise to introduce such a confusing behaviour.
So to solve this, create a new CONFIG_NO_HZ_COMMON option
which gathers the common code between idle and full dynticks
(that common code for now is simply the idle dynticks code)
and select it from their referring Kconfig.
Then we'll later create CONFIG_NO_HZ_IDLE and map CONFIG_NO_HZ
to it for backward compatibility.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Chris Metcalf <cmetcalf@tilera.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Geoff Levand <geoff@infradead.org>
Cc: Gilad Ben Yossef <gilad@benyossef.com>
Cc: Hakan Akkan <hakanakkan@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Kevin Hilman <khilman@linaro.org>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung.kim@lge.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
As default_scale_{freq,smt}_power() and update_rt_power() are
used in kernel/sched/fair.c only, annotate them as static
functions.
Signed-off-by: Li Zefan <lizefan@huawei.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/5135A7AF.8010900@huawei.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull scheduler changes from Ingo Molnar:
"Main changes:
- scheduler side full-dynticks (user-space execution is undisturbed
and receives no timer IRQs) preparation changes that convert the
cputime accounting code to be full-dynticks ready, from Frederic
Weisbecker.
- Initial sched.h split-up changes, by Clark Williams
- select_idle_sibling() performance improvement by Mike Galbraith:
" 1 tbench pair (worst case) in a 10 core + SMT package:
pre 15.22 MB/sec 1 procs
post 252.01 MB/sec 1 procs "
- sched_rr_get_interval() ABI fix/change. We think this detail is not
used by apps (so it's not an ABI in practice), but lets keep it
under observation.
- misc RT scheduling cleanups, optimizations"
* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (24 commits)
sched/rt: Add <linux/sched/rt.h> header to <linux/init_task.h>
cputime: Remove irqsave from seqlock readers
sched, powerpc: Fix sched.h split-up build failure
cputime: Restore CPU_ACCOUNTING config defaults for PPC64
sched/rt: Move rt specific bits into new header file
sched/rt: Add a tuning knob to allow changing SCHED_RR timeslice
sched: Move sched.h sysctl bits into separate header
sched: Fix signedness bug in yield_to()
sched: Fix select_idle_sibling() bouncing cow syndrome
sched/rt: Further simplify pick_rt_task()
sched/rt: Do not account zero delta_exec in update_curr_rt()
cputime: Safely read cputime of full dynticks CPUs
kvm: Prepare to add generic guest entry/exit callbacks
cputime: Use accessors to read task cputime stats
cputime: Allow dynamic switch between tick/virtual based cputime accounting
cputime: Generic on-demand virtual cputime accounting
cputime: Move default nsecs_to_cputime() to jiffies based cputime file
cputime: Librarize per nsecs resolution cputime definitions
cputime: Avoid multiplication overflow on utime scaling
context_tracking: Export context state for generic vtime
...
Fix up conflict in kernel/context_tracking.c due to comment additions.
If the previous CPU is cache affine and idle, select it.
The current implementation simply traverses the sd_llc domain,
taking the first idle CPU encountered, which walks buddy pairs
hand in hand over the package, inflicting excruciating pain.
1 tbench pair (worst case) in a 10 core + SMT package:
pre 15.22 MB/sec 1 procs
post 252.01 MB/sec 1 procs
Signed-off-by: Mike Galbraith <bitbucket@online.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1359371965.5783.127.camel@marge.simpson.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
a4c96ae319 "sched: Unthrottle rt runqueues in
__disable_runtime()" turned the unthrottle_offline_cfs_rqs
function into a static symbol, which now triggers a warning
about it being potentially unused:
kernel/sched/fair.c:2055:13: warning: 'unthrottle_offline_cfs_rqs' defined but not used [-Wunused-function]
Marking it __maybe_unused shuts up the gcc warning and lets the
compiler safely drop the function body when it's not being used.
To reproduce, build the ARM bcm2835_defconfig.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Peter Boonstoppel <pboonstoppel@nvidia.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Turner <pjt@google.com>
Cc: linux-arm-kernel@list.infradead.org
Link: http://lkml.kernel.org/r/1359123276-15833-6-git-send-email-arnd@arndb.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The caller of sched_sliced() should pass se.cfs_rq and se as the
arguments, however in sched_rr_get_interval() we gave it
rq.cfs_rq and se, which made the following computation obviously
wrong.
The change was introduced by commit:
77034937dc sched: fix crash in sys_sched_rr_get_interval()
... 5 years ago, while it had been the correct 'cfs_rq_of' before
the commit. The change seems to be irrelevant to the commit
msg, which was to return a 0 timeslice for tasks that are on an
idle runqueue. So I believe that was just a plain typo.
Signed-off-by: Zhu Yanhai <gaoyang.zyh@taobao.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Turner <pjt@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1357621012-15039-1-git-send-email-gaoyang.zyh@taobao.com
[ Since this is an ABI and an old bug, we'll test this via a
slow upstream route, to hopefully discover any app breakage. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
task_numa_placement() oopsed on NULL p->mm when task_numa_fault() got
called in the handling of break_ksm() for ksmd. That might be a
peculiar case, which perhaps KSM could takes steps to avoid? but it's
more robust if task_numa_placement() allows for such a possibility.
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Michal Hocko reported that the following build error occurs if
CONFIG_NUMA_BALANCING is set without THP support
kernel/sched/fair.c: In function ‘task_numa_work’:
kernel/sched/fair.c:932:55: error: call to ‘__build_bug_failed’ declared with attribute error: BUILD_BUG failed
The problem is that HPAGE_PMD_SHIFT triggers a BUILD_BUG() on
!CONFIG_TRANSPARENT_HUGEPAGE. This patch addresses the problem.
Reported-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (GNU/Linux)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=ivQa
-----END PGP SIGNATURE-----
Merge tag 'balancenuma-v11' of git://git.kernel.org/pub/scm/linux/kernel/git/mel/linux-balancenuma
Pull Automatic NUMA Balancing bare-bones from Mel Gorman:
"There are three implementations for NUMA balancing, this tree
(balancenuma), numacore which has been developed in tip/master and
autonuma which is in aa.git.
In almost all respects balancenuma is the dumbest of the three because
its main impact is on the VM side with no attempt to be smart about
scheduling. In the interest of getting the ball rolling, it would be
desirable to see this much merged for 3.8 with the view to building
scheduler smarts on top and adapting the VM where required for 3.9.
The most recent set of comparisons available from different people are
mel: https://lkml.org/lkml/2012/12/9/108
mingo: https://lkml.org/lkml/2012/12/7/331
tglx: https://lkml.org/lkml/2012/12/10/437
srikar: https://lkml.org/lkml/2012/12/10/397
The results are a mixed bag. In my own tests, balancenuma does
reasonably well. It's dumb as rocks and does not regress against
mainline. On the other hand, Ingo's tests shows that balancenuma is
incapable of converging for this workloads driven by perf which is bad
but is potentially explained by the lack of scheduler smarts. Thomas'
results show balancenuma improves on mainline but falls far short of
numacore or autonuma. Srikar's results indicate we all suffer on a
large machine with imbalanced node sizes.
My own testing showed that recent numacore results have improved
dramatically, particularly in the last week but not universally.
We've butted heads heavily on system CPU usage and high levels of
migration even when it shows that overall performance is better.
There are also cases where it regresses. Of interest is that for
specjbb in some configurations it will regress for lower numbers of
warehouses and show gains for higher numbers which is not reported by
the tool by default and sometimes missed in treports. Recently I
reported for numacore that the JVM was crashing with
NullPointerExceptions but currently it's unclear what the source of
this problem is. Initially I thought it was in how numacore batch
handles PTEs but I'm no longer think this is the case. It's possible
numacore is just able to trigger it due to higher rates of migration.
These reports were quite late in the cycle so I/we would like to start
with this tree as it contains much of the code we can agree on and has
not changed significantly over the last 2-3 weeks."
* tag 'balancenuma-v11' of git://git.kernel.org/pub/scm/linux/kernel/git/mel/linux-balancenuma: (50 commits)
mm/rmap, migration: Make rmap_walk_anon() and try_to_unmap_anon() more scalable
mm/rmap: Convert the struct anon_vma::mutex to an rwsem
mm: migrate: Account a transhuge page properly when rate limiting
mm: numa: Account for failed allocations and isolations as migration failures
mm: numa: Add THP migration for the NUMA working set scanning fault case build fix
mm: numa: Add THP migration for the NUMA working set scanning fault case.
mm: sched: numa: Delay PTE scanning until a task is scheduled on a new node
mm: sched: numa: Control enabling and disabling of NUMA balancing if !SCHED_DEBUG
mm: sched: numa: Control enabling and disabling of NUMA balancing
mm: sched: Adapt the scanning rate if a NUMA hinting fault does not migrate
mm: numa: Use a two-stage filter to restrict pages being migrated for unlikely task<->node relationships
mm: numa: migrate: Set last_nid on newly allocated page
mm: numa: split_huge_page: Transfer last_nid on tail page
mm: numa: Introduce last_nid to the page frame
sched: numa: Slowly increase the scanning period as NUMA faults are handled
mm: numa: Rate limit setting of pte_numa if node is saturated
mm: numa: Rate limit the amount of memory that is migrated between nodes
mm: numa: Structures for Migrate On Fault per NUMA migration rate limiting
mm: numa: Migrate pages handled during a pmd_numa hinting fault
mm: numa: Migrate on reference policy
...
This reverts commit f269ae0469.
It turns out it causes a very noticeable interactivity regression with
CONFIG_SCHED_AUTOGROUP (test-case: "make -j32" of the kernel in a
terminal window, while scrolling in a browser - the autogrouping means
that the two end up in separate cgroups, and the browser should be
smooth as silk despite the high load).
Says Paul Turner:
"It seems that the update-throttling on the wake-side is reducing the
interactive tasks' ability to preempt. While I suspect the right
longer term answer here is force these updates only in the
cross-cgroup case; this is less trivial. For this release I believe
the right answer is either going to be a revert or restore the updates
on the enqueue-side."
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Bisected-by: Mike Galbraith <efault@gmx.de>
Acked-by: Paul Turner <pjt@google.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Due to the fact that migrations are driven by the CPU a task is running
on there is no point tracking NUMA faults until one task runs on a new
node. This patch tracks the first node used by an address space. Until
it changes, PTE scanning is disabled and no NUMA hinting faults are
trapped. This should help workloads that are short-lived, do not care
about NUMA placement or have bound themselves to a single node.
This takes advantage of the logic in "mm: sched: numa: Implement slow
start for working set sampling" to delay when the checks are made. This
will take advantage of processes that set their CPU and node bindings
early in their lifetime. It will also potentially allow any initial load
balancing to take place.
Signed-off-by: Mel Gorman <mgorman@suse.de>
This patch adds Kconfig options and kernel parameters to allow the
enabling and disabling of automatic NUMA balancing. The existance
of such a switch was and is very important when debugging problems
related to transparent hugepages and we should have the same for
automatic NUMA placement.
Signed-off-by: Mel Gorman <mgorman@suse.de>
The PTE scanning rate and fault rates are two of the biggest sources of
system CPU overhead with automatic NUMA placement. Ideally a proper policy
would detect if a workload was properly placed, schedule and adjust the
PTE scanning rate accordingly. We do not track the necessary information
to do that but we at least know if we migrated or not.
This patch scans slower if a page was not migrated as the result of a
NUMA hinting fault up to sysctl_numa_balancing_scan_period_max which is
now higher than the previous default. Once every minute it will reset
the scanner in case of phase changes.
This is hilariously crude and the numbers are arbitrary. Workloads will
converge quite slowly in comparison to what a proper policy should be able
to do. On the plus side, we will chew up less CPU for workloads that have
no need for automatic balancing.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Currently the rate of scanning for an address space is controlled
by the individual tasks. The next scan is simply determined by
2*p->numa_scan_period.
The 2*p->numa_scan_period is arbitrary and never changes. At this point
there is still no proper policy that decides if a task or process is
properly placed. It just scans and assumes the next NUMA fault will
place it properly. As it is assumed that pages will get properly placed
over time, increase the scan window each time a fault is incurred. This
is a big assumption as noted in the comments.
It should be noted that changing to p->numa_scan_period will increase
system CPU usage because now the scanning rate has effectively doubled.
If that is a problem then the min_rate should be made 200ms instead of
restoring the 2* logic.
Signed-off-by: Mel Gorman <mgorman@suse.de>
If there are a large number of NUMA hinting faults and all of them
are resulting in migrations it may indicate that memory is just
bouncing uselessly around. NUMA balancing cost is likely exceeding
any benefit from locality. Rate limit the PTE updates if the node
is migration rate-limited. As noted in the comments, this distorts
the NUMA faulting statistics.
Signed-off-by: Mel Gorman <mgorman@suse.de>