summaryrefslogtreecommitdiffstats
path: root/man7/sched.7
blob: 78545053fca987dd9c0b0a3a07f8fab5f52f9216 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
.\" Copyright (C) 2014 Michael Kerrisk <mtk.manpages@gmail.com>
.\" and Copyright (C) 2014 Peter Zijlstra <peterz@infradead.org>
.\" and Copyright (C) 2014 Juri Lelli <juri.lelli@gmail.com>
.\" Various pieces from the old sched_setscheduler(2) page
.\" 	Copyright (C) Tom Bjorkholm, Markus Kuhn & David A. Wheeler 1996-1999
.\" 	and Copyright (C) 2007 Carsten Emde <Carsten.Emde@osadl.org>
.\" 	and Copyright (C) 2008 Michael Kerrisk <mtk.manpages@gmail.com>
.\"
.\" SPDX-License-Identifier: GPL-2.0-or-later
.\"
.\" Worth looking at: http://rt.wiki.kernel.org/index.php
.\"
.TH sched 7 2023-02-10 "Linux man-pages 6.05.01"
.SH NAME
sched \- overview of CPU scheduling
.SH DESCRIPTION
Since Linux 2.6.23, the default scheduler is CFS,
the "Completely Fair Scheduler".
The CFS scheduler replaced the earlier "O(1)" scheduler.
.\"
.SS API summary
Linux provides the following system calls for controlling
the CPU scheduling behavior, policy, and priority of processes
(or, more precisely, threads).
.TP
.BR nice (2)
Set a new nice value for the calling thread,
and return the new nice value.
.TP
.BR getpriority (2)
Return the nice value of a thread, a process group,
or the set of threads owned by a specified user.
.TP
.BR setpriority (2)
Set the nice value of a thread, a process group,
or the set of threads owned by a specified user.
.TP
.BR sched_setscheduler (2)
Set the scheduling policy and parameters of a specified thread.
.TP
.BR sched_getscheduler (2)
Return the scheduling policy of a specified thread.
.TP
.BR sched_setparam (2)
Set the scheduling parameters of a specified thread.
.TP
.BR sched_getparam (2)
Fetch the scheduling parameters of a specified thread.
.TP
.BR sched_get_priority_max (2)
Return the maximum priority available in a specified scheduling policy.
.TP
.BR sched_get_priority_min (2)
Return the minimum priority available in a specified scheduling policy.
.TP
.BR sched_rr_get_interval (2)
Fetch the quantum used for threads that are scheduled under
the "round-robin" scheduling policy.
.TP
.BR sched_yield (2)
Cause the caller to relinquish the CPU,
so that some other thread be executed.
.TP
.BR sched_setaffinity (2)
(Linux-specific)
Set the CPU affinity of a specified thread.
.TP
.BR sched_getaffinity (2)
(Linux-specific)
Get the CPU affinity of a specified thread.
.TP
.BR sched_setattr (2)
Set the scheduling policy and parameters of a specified thread.
This (Linux-specific) system call provides a superset of the functionality of
.BR sched_setscheduler (2)
and
.BR sched_setparam (2).
.TP
.BR sched_getattr (2)
Fetch the scheduling policy and parameters of a specified thread.
This (Linux-specific) system call provides a superset of the functionality of
.BR sched_getscheduler (2)
and
.BR sched_getparam (2).
.\"
.SS Scheduling policies
The scheduler is the kernel component that decides which runnable thread
will be executed by the CPU next.
Each thread has an associated scheduling policy and a \fIstatic\fP
scheduling priority,
.IR sched_priority .
The scheduler makes its decisions based on knowledge of the scheduling
policy and static priority of all threads on the system.
.PP
For threads scheduled under one of the normal scheduling policies
(\fBSCHED_OTHER\fP, \fBSCHED_IDLE\fP, \fBSCHED_BATCH\fP),
\fIsched_priority\fP is not used in scheduling
decisions (it must be specified as 0).
.PP
Processes scheduled under one of the real-time policies
(\fBSCHED_FIFO\fP, \fBSCHED_RR\fP) have a
\fIsched_priority\fP value in the range 1 (low) to 99 (high).
(As the numbers imply, real-time threads always have higher priority
than normal threads.)
Note well: POSIX.1 requires an implementation to support only a
minimum 32 distinct priority levels for the real-time policies,
and some systems supply just this minimum.
Portable programs should use
.BR sched_get_priority_min (2)
and
.BR sched_get_priority_max (2)
to find the range of priorities supported for a particular policy.
.PP
Conceptually, the scheduler maintains a list of runnable
threads for each possible \fIsched_priority\fP value.
In order to determine which thread runs next, the scheduler looks for
the nonempty list with the highest static priority and selects the
thread at the head of this list.
.PP
A thread's scheduling policy determines
where it will be inserted into the list of threads
with equal static priority and how it will move inside this list.
.PP
All scheduling is preemptive: if a thread with a higher static
priority becomes ready to run, the currently running thread
will be preempted and
returned to the wait list for its static priority level.
The scheduling policy determines the
ordering only within the list of runnable threads with equal static
priority.
.SS SCHED_FIFO: First in-first out scheduling
\fBSCHED_FIFO\fP can be used only with static priorities higher than
0, which means that when a \fBSCHED_FIFO\fP thread becomes runnable,
it will always immediately preempt any currently running
\fBSCHED_OTHER\fP, \fBSCHED_BATCH\fP, or \fBSCHED_IDLE\fP thread.
\fBSCHED_FIFO\fP is a simple scheduling
algorithm without time slicing.
For threads scheduled under the
\fBSCHED_FIFO\fP policy, the following rules apply:
.IP \[bu] 3
A running \fBSCHED_FIFO\fP thread that has been preempted by another thread of
higher priority will stay at the head of the list for its priority and
will resume execution as soon as all threads of higher priority are
blocked again.
.IP \[bu]
When a blocked \fBSCHED_FIFO\fP thread becomes runnable, it
will be inserted at the end of the list for its priority.
.IP \[bu]
If a call to
.BR sched_setscheduler (2),
.BR sched_setparam (2),
.BR sched_setattr (2),
.BR pthread_setschedparam (3),
or
.BR pthread_setschedprio (3)
changes the priority of the running or runnable
.B SCHED_FIFO
thread identified by
.I pid
the effect on the thread's position in the list depends on
the direction of the change to threads priority:
.RS
.IP (a) 5
If the thread's priority is raised,
it is placed at the end of the list for its new priority.
As a consequence,
it may preempt a currently running thread with the same priority.
.IP (b)
If the thread's priority is unchanged,
its position in the run list is unchanged.
.IP (c)
If the thread's priority is lowered,
it is placed at the front of the list for its new priority.
.RE
.IP
According to POSIX.1-2008,
changes to a thread's priority (or policy) using any mechanism other than
.BR pthread_setschedprio (3)
should result in the thread being placed at the end of
the list for its priority.
.\" In Linux 2.2.x and Linux 2.4.x, the thread is placed at the front of the queue
.\" In Linux 2.0.x, the Right Thing happened: the thread went to the back -- MTK
.IP \[bu]
A thread calling
.BR sched_yield (2)
will be put at the end of the list.
.PP
No other events will move a thread
scheduled under the \fBSCHED_FIFO\fP policy in the wait list of
runnable threads with equal static priority.
.PP
A \fBSCHED_FIFO\fP
thread runs until either it is blocked by an I/O request, it is
preempted by a higher priority thread, or it calls
.BR sched_yield (2).
.SS SCHED_RR: Round-robin scheduling
\fBSCHED_RR\fP is a simple enhancement of \fBSCHED_FIFO\fP.
Everything
described above for \fBSCHED_FIFO\fP also applies to \fBSCHED_RR\fP,
except that each thread is allowed to run only for a maximum time
quantum.
If a \fBSCHED_RR\fP thread has been running for a time
period equal to or longer than the time quantum, it will be put at the
end of the list for its priority.
A \fBSCHED_RR\fP thread that has
been preempted by a higher priority thread and subsequently resumes
execution as a running thread will complete the unexpired portion of
its round-robin time quantum.
The length of the time quantum can be
retrieved using
.BR sched_rr_get_interval (2).
.\" On Linux 2.4, the length of the RR interval is influenced
.\" by the process nice value -- MTK
.\"
.SS SCHED_DEADLINE: Sporadic task model deadline scheduling
Since Linux 3.14, Linux provides a deadline scheduling policy
.RB ( SCHED_DEADLINE ).
This policy is currently implemented using
GEDF (Global Earliest Deadline First)
in conjunction with CBS (Constant Bandwidth Server).
To set and fetch this policy and associated attributes,
one must use the Linux-specific
.BR sched_setattr (2)
and
.BR sched_getattr (2)
system calls.
.PP
A sporadic task is one that has a sequence of jobs, where each
job is activated at most once per period.
Each job also has a
.IR "relative deadline" ,
before which it should finish execution, and a
.IR "computation time" ,
which is the CPU time necessary for executing the job.
The moment when a task wakes up
because a new job has to be executed is called the
.I arrival time
(also referred to as the request time or release time).
The
.I start time
is the time at which a task starts its execution.
The
.I absolute deadline
is thus obtained by adding the relative deadline to the arrival time.
.PP
The following diagram clarifies these terms:
.PP
.in +4n
.EX
arrival/wakeup                    absolute deadline
     |    start time                    |
     |        |                         |
     v        v                         v
-----x--------xooooooooooooooooo--------x--------x---
              |<- comp. time ->|
     |<------- relative deadline ------>|
     |<-------------- period ------------------->|
.EE
.in
.PP
When setting a
.B SCHED_DEADLINE
policy for a thread using
.BR sched_setattr (2),
one can specify three parameters:
.IR Runtime ,
.IR Deadline ,
and
.IR Period .
These parameters do not necessarily correspond to the aforementioned terms:
usual practice is to set Runtime to something bigger than the average
computation time (or worst-case execution time for hard real-time tasks),
Deadline to the relative deadline, and Period to the period of the task.
Thus, for
.B SCHED_DEADLINE
scheduling, we have:
.PP
.in +4n
.EX
arrival/wakeup                    absolute deadline
     |    start time                    |
     |        |                         |
     v        v                         v
-----x--------xooooooooooooooooo--------x--------x---
              |<-- Runtime ------->|
     |<----------- Deadline ----------->|
     |<-------------- Period ------------------->|
.EE
.in
.PP
The three deadline-scheduling parameters correspond to the
.IR sched_runtime ,
.IR sched_deadline ,
and
.I sched_period
fields of the
.I sched_attr
structure; see
.BR sched_setattr (2).
These fields express values in nanoseconds.
.\" FIXME It looks as though specifying sched_period as 0 means
.\" "make sched_period the same as sched_deadline".
.\" This needs to be documented.
If
.I sched_period
is specified as 0, then it is made the same as
.IR sched_deadline .
.PP
The kernel requires that:
.PP
.in +4n
.EX
sched_runtime <= sched_deadline <= sched_period
.EE
.in
.PP
.\" See __checkparam_dl in kernel/sched/core.c
In addition, under the current implementation,
all of the parameter values must be at least 1024
(i.e., just over one microsecond,
which is the resolution of the implementation), and less than 2\[ha]63.
If any of these checks fails,
.BR sched_setattr (2)
fails with the error
.BR EINVAL .
.PP
The CBS guarantees non-interference between tasks, by throttling
threads that attempt to over-run their specified Runtime.
.PP
To ensure deadline scheduling guarantees,
the kernel must prevent situations where the set of
.B SCHED_DEADLINE
threads is not feasible (schedulable) within the given constraints.
The kernel thus performs an admittance test when setting or changing
.B SCHED_DEADLINE
policy and attributes.
This admission test calculates whether the change is feasible;
if it is not,
.BR sched_setattr (2)
fails with the error
.BR EBUSY .
.PP
For example, it is required (but not necessarily sufficient) for
the total utilization to be less than or equal to the total number of
CPUs available, where, since each thread can maximally run for
Runtime per Period, that thread's utilization is its
Runtime divided by its Period.
.PP
In order to fulfill the guarantees that are made when
a thread is admitted to the
.B SCHED_DEADLINE
policy,
.B SCHED_DEADLINE
threads are the highest priority (user controllable) threads in the
system; if any
.B SCHED_DEADLINE
thread is runnable,
it will preempt any thread scheduled under one of the other policies.
.PP
A call to
.BR fork (2)
by a thread scheduled under the
.B SCHED_DEADLINE
policy fails with the error
.BR EAGAIN ,
unless the thread has its reset-on-fork flag set (see below).
.PP
A
.B SCHED_DEADLINE
thread that calls
.BR sched_yield (2)
will yield the current job and wait for a new period to begin.
.\"
.\" FIXME Calling sched_getparam() on a SCHED_DEADLINE thread
.\" fails with EINVAL, but sched_getscheduler() succeeds.
.\" Is that intended? (Why?)
.\"
.SS SCHED_OTHER: Default Linux time-sharing scheduling
\fBSCHED_OTHER\fP can be used at only static priority 0
(i.e., threads under real-time policies always have priority over
.B SCHED_OTHER
processes).
\fBSCHED_OTHER\fP is the standard Linux time-sharing scheduler that is
intended for all threads that do not require the special
real-time mechanisms.
.PP
The thread to run is chosen from the static
priority 0 list based on a \fIdynamic\fP priority that is determined only
inside this list.
The dynamic priority is based on the nice value (see below)
and is increased for each time quantum the thread is ready to run,
but denied to run by the scheduler.
This ensures fair progress among all \fBSCHED_OTHER\fP threads.
.PP
In the Linux kernel source code, the
.B SCHED_OTHER
policy is actually named
.BR SCHED_NORMAL .
.\"
.SS The nice value
The nice value is an attribute
that can be used to influence the CPU scheduler to
favor or disfavor a process in scheduling decisions.
It affects the scheduling of
.B SCHED_OTHER
and
.B SCHED_BATCH
(see below) processes.
The nice value can be modified using
.BR nice (2),
.BR setpriority (2),
or
.BR sched_setattr (2).
.PP
According to POSIX.1, the nice value is a per-process attribute;
that is, the threads in a process should share a nice value.
However, on Linux, the nice value is a per-thread attribute:
different threads in the same process may have different nice values.
.PP
The range of the nice value
varies across UNIX systems.
On modern Linux, the range is \-20 (high priority) to +19 (low priority).
On some other systems, the range is \-20..20.
Very early Linux kernels (before Linux 2.0) had the range \-infinity..15.
.\" Linux before 1.3.36 had \-infinity..15.
.\" Since Linux 1.3.43, Linux has the range \-20..19.
.PP
The degree to which the nice value affects the relative scheduling of
.B SCHED_OTHER
processes likewise varies across UNIX systems and
across Linux kernel versions.
.PP
With the advent of the CFS scheduler in Linux 2.6.23,
Linux adopted an algorithm that causes
relative differences in nice values to have a much stronger effect.
In the current implementation, each unit of difference in the
nice values of two processes results in a factor of 1.25
in the degree to which the scheduler favors the higher priority process.
This causes very low nice values (+19) to truly provide little CPU
to a process whenever there is any other
higher priority load on the system,
and makes high nice values (\-20) deliver most of the CPU to applications
that require it (e.g., some audio applications).
.PP
On Linux, the
.B RLIMIT_NICE
resource limit can be used to define a limit to which
an unprivileged process's nice value can be raised; see
.BR setrlimit (2)
for details.
.PP
For further details on the nice value, see the subsections on
the autogroup feature and group scheduling, below.
.\"
.SS SCHED_BATCH: Scheduling batch processes
(Since Linux 2.6.16.)
\fBSCHED_BATCH\fP can be used only at static priority 0.
This policy is similar to \fBSCHED_OTHER\fP in that it schedules
the thread according to its dynamic priority
(based on the nice value).
The difference is that this policy
will cause the scheduler to always assume
that the thread is CPU-intensive.
Consequently, the scheduler will apply a small scheduling
penalty with respect to wakeup behavior,
so that this thread is mildly disfavored in scheduling decisions.
.PP
.\" The following paragraph is drawn largely from the text that
.\" accompanied Ingo Molnar's patch for the implementation of
.\" SCHED_BATCH.
.\" commit b0a9499c3dd50d333e2aedb7e894873c58da3785
This policy is useful for workloads that are noninteractive,
but do not want to lower their nice value,
and for workloads that want a deterministic scheduling policy without
interactivity causing extra preemptions (between the workload's tasks).
.\"
.SS SCHED_IDLE: Scheduling very low priority jobs
(Since Linux 2.6.23.)
\fBSCHED_IDLE\fP can be used only at static priority 0;
the process nice value has no influence for this policy.
.PP
This policy is intended for running jobs at extremely low
priority (lower even than a +19 nice value with the
.B SCHED_OTHER
or
.B SCHED_BATCH
policies).
.\"
.SS Resetting scheduling policy for child processes
Each thread has a reset-on-fork scheduling flag.
When this flag is set, children created by
.BR fork (2)
do not inherit privileged scheduling policies.
The reset-on-fork flag can be set by either:
.IP \[bu] 3
ORing the
.B SCHED_RESET_ON_FORK
flag into the
.I policy
argument when calling
.BR sched_setscheduler (2)
(since Linux 2.6.32);
or
.IP \[bu]
specifying the
.B SCHED_FLAG_RESET_ON_FORK
flag in
.I attr.sched_flags
when calling
.BR sched_setattr (2).
.PP
Note that the constants used with these two APIs have different names.
The state of the reset-on-fork flag can analogously be retrieved using
.BR sched_getscheduler (2)
and
.BR sched_getattr (2).
.PP
The reset-on-fork feature is intended for media-playback applications,
and can be used to prevent applications evading the
.B RLIMIT_RTTIME
resource limit (see
.BR getrlimit (2))
by creating multiple child processes.
.PP
More precisely, if the reset-on-fork flag is set,
the following rules apply for subsequently created children:
.IP \[bu] 3
If the calling thread has a scheduling policy of
.B SCHED_FIFO
or
.BR SCHED_RR ,
the policy is reset to
.B SCHED_OTHER
in child processes.
.IP \[bu]
If the calling process has a negative nice value,
the nice value is reset to zero in child processes.
.PP
After the reset-on-fork flag has been enabled,
it can be reset only if the thread has the
.B CAP_SYS_NICE
capability.
This flag is disabled in child processes created by
.BR fork (2).
.\"
.SS Privileges and resource limits
Before Linux 2.6.12, only privileged
.RB ( CAP_SYS_NICE )
threads can set a nonzero static priority (i.e., set a real-time
scheduling policy).
The only change that an unprivileged thread can make is to set the
.B SCHED_OTHER
policy, and this can be done only if the effective user ID of the caller
matches the real or effective user ID of the target thread
(i.e., the thread specified by
.IR pid )
whose policy is being changed.
.PP
A thread must be privileged
.RB ( CAP_SYS_NICE )
in order to set or modify a
.B SCHED_DEADLINE
policy.
.PP
Since Linux 2.6.12, the
.B RLIMIT_RTPRIO
resource limit defines a ceiling on an unprivileged thread's
static priority for the
.B SCHED_RR
and
.B SCHED_FIFO
policies.
The rules for changing scheduling policy and priority are as follows:
.IP \[bu] 3
If an unprivileged thread has a nonzero
.B RLIMIT_RTPRIO
soft limit, then it can change its scheduling policy and priority,
subject to the restriction that the priority cannot be set to a
value higher than the maximum of its current priority and its
.B RLIMIT_RTPRIO
soft limit.
.IP \[bu]
If the
.B RLIMIT_RTPRIO
soft limit is 0, then the only permitted changes are to lower the priority,
or to switch to a non-real-time policy.
.IP \[bu]
Subject to the same rules,
another unprivileged thread can also make these changes,
as long as the effective user ID of the thread making the change
matches the real or effective user ID of the target thread.
.IP \[bu]
Special rules apply for the
.B SCHED_IDLE
policy.
Before Linux 2.6.39,
an unprivileged thread operating under this policy cannot
change its policy, regardless of the value of its
.B RLIMIT_RTPRIO
resource limit.
Since Linux 2.6.39,
.\" commit c02aa73b1d18e43cfd79c2f193b225e84ca497c8
an unprivileged thread can switch to either the
.B SCHED_BATCH
or the
.B SCHED_OTHER
policy so long as its nice value falls within the range permitted by its
.B RLIMIT_NICE
resource limit (see
.BR getrlimit (2)).
.PP
Privileged
.RB ( CAP_SYS_NICE )
threads ignore the
.B RLIMIT_RTPRIO
limit; as with older kernels,
they can make arbitrary changes to scheduling policy and priority.
See
.BR getrlimit (2)
for further information on
.BR RLIMIT_RTPRIO .
.SS Limiting the CPU usage of real-time and deadline processes
A nonblocking infinite loop in a thread scheduled under the
.BR SCHED_FIFO ,
.BR SCHED_RR ,
or
.B SCHED_DEADLINE
policy can potentially block all other threads from accessing
the CPU forever.
Before Linux 2.6.25, the only way of preventing a runaway real-time
process from freezing the system was to run (at the console)
a shell scheduled under a higher static priority than the tested application.
This allows an emergency kill of tested
real-time applications that do not block or terminate as expected.
.PP
Since Linux 2.6.25, there are other techniques for dealing with runaway
real-time and deadline processes.
One of these is to use the
.B RLIMIT_RTTIME
resource limit to set a ceiling on the CPU time that
a real-time process may consume.
See
.BR getrlimit (2)
for details.
.PP
Since Linux 2.6.25, Linux also provides two
.I /proc
files that can be used to reserve a certain amount of CPU time
to be used by non-real-time processes.
Reserving CPU time in this fashion allows some CPU time to be
allocated to (say) a root shell that can be used to kill a runaway process.
Both of these files specify time values in microseconds:
.TP
.I /proc/sys/kernel/sched_rt_period_us
This file specifies a scheduling period that is equivalent to
100% CPU bandwidth.
The value in this file can range from 1 to
.BR INT_MAX ,
giving an operating range of 1 microsecond to around 35 minutes.
The default value in this file is 1,000,000 (1 second).
.TP
.I /proc/sys/kernel/sched_rt_runtime_us
The value in this file specifies how much of the "period" time
can be used by all real-time and deadline scheduled processes
on the system.
The value in this file can range from \-1 to
.BR INT_MAX \-1.
Specifying \-1 makes the run time the same as the period;
that is, no CPU time is set aside for non-real-time processes
(which was the behavior before Linux 2.6.25).
The default value in this file is 950,000 (0.95 seconds),
meaning that 5% of the CPU time is reserved for processes that
don't run under a real-time or deadline scheduling policy.
.SS Response time
A blocked high priority thread waiting for I/O has a certain
response time before it is scheduled again.
The device driver writer
can greatly reduce this response time by using a "slow interrupt"
interrupt handler.
.\" as described in
.\" .BR request_irq (9).
.SS Miscellaneous
Child processes inherit the scheduling policy and parameters across a
.BR fork (2).
The scheduling policy and parameters are preserved across
.BR execve (2).
.PP
Memory locking is usually needed for real-time processes to avoid
paging delays; this can be done with
.BR mlock (2)
or
.BR mlockall (2).
.\"
.SS The autogroup feature
.\" commit 5091faa449ee0b7d73bc296a93bca9540fc51d0a
Since Linux 2.6.38,
the kernel provides a feature known as autogrouping to improve interactive
desktop performance in the face of multiprocess, CPU-intensive
workloads such as building the Linux kernel with large numbers of
parallel build processes (i.e., the
.BR make (1)
.B \-j
flag).
.PP
This feature operates in conjunction with the
CFS scheduler and requires a kernel that is configured with
.BR CONFIG_SCHED_AUTOGROUP .
On a running system, this feature is enabled or disabled via the file
.IR /proc/sys/kernel/sched_autogroup_enabled ;
a value of 0 disables the feature, while a value of 1 enables it.
The default value in this file is 1, unless the kernel was booted with the
.I noautogroup
parameter.
.PP
A new autogroup is created when a new session is created via
.BR setsid (2);
this happens, for example, when a new terminal window is started.
A new process created by
.BR fork (2)
inherits its parent's autogroup membership.
Thus, all of the processes in a session are members of the same autogroup.
An autogroup is automatically destroyed when the last process
in the group terminates.
.PP
When autogrouping is enabled, all of the members of an autogroup
are placed in the same kernel scheduler "task group".
The CFS scheduler employs an algorithm that equalizes the
distribution of CPU cycles across task groups.
The benefits of this for interactive desktop performance
can be described via the following example.
.PP
Suppose that there are two autogroups competing for the same CPU
(i.e., presume either a single CPU system or the use of
.BR taskset (1)
to confine all the processes to the same CPU on an SMP system).
The first group contains ten CPU-bound processes from
a kernel build started with
.IR "make\~\-j10" .
The other contains a single CPU-bound process: a video player.
The effect of autogrouping is that the two groups will
each receive half of the CPU cycles.
That is, the video player will receive 50% of the CPU cycles,
rather than just 9% of the cycles,
which would likely lead to degraded video playback.
The situation on an SMP system is more complex,
.\" Mike Galbraith, 25 Nov 2016:
.\"     I'd say something more wishy-washy here, like cycles are
.\"     distributed fairly across groups and leave it at that, as your
.\"     detailed example is incorrect due to SMP fairness (which I don't
.\"     like much because [very unlikely] worst case scenario
.\"     renders a box sized group incapable of utilizing more that
.\"     a single CPU total).  For example, if a group of NR_CPUS
.\"     size competes with a singleton, load balancing will try to give
.\"     the singleton a full CPU of its very own.  If groups intersect for
.\"     whatever reason on say my quad lappy, distribution is 80/20 in
.\"     favor of the singleton.
but the general effect is the same:
the scheduler distributes CPU cycles across task groups such that
an autogroup that contains a large number of CPU-bound processes
does not end up hogging CPU cycles at the expense of the other
jobs on the system.
.PP
A process's autogroup (task group) membership can be viewed via the file
.IR /proc/ pid /autogroup :
.PP
.in +4n
.EX
$ \fBcat /proc/1/autogroup\fP
/autogroup\-1 nice 0
.EE
.in
.PP
This file can also be used to modify the CPU bandwidth allocated
to an autogroup.
This is done by writing a number in the "nice" range to the file
to set the autogroup's nice value.
The allowed range is from +19 (low priority) to \-20 (high priority).
(Writing values outside of this range causes
.BR write (2)
to fail with the error
.BR EINVAL .)
.\" FIXME .
.\" Because of a bug introduced in Linux 4.7
.\" (commit 2159197d66770ec01f75c93fb11dc66df81fd45b made changes
.\" that exposed the fact that autogroup didn't call scale_load()),
.\" it happened that *all* values in this range caused a task group
.\" to be further disfavored by the scheduler, with \-20 resulting
.\" in the scheduler mildly disfavoring the task group and +19 greatly
.\" disfavoring it.
.\"
.\" A patch was posted on 23 Nov 2016
.\" ("sched/autogroup: Fix 64bit kernel nice adjustment";
.\" check later to see in which kernel version it lands.
.PP
The autogroup nice setting has the same meaning as the process nice value,
but applies to distribution of CPU cycles to the autogroup as a whole,
based on the relative nice values of other autogroups.
For a process inside an autogroup, the CPU cycles that it receives
will be a product of the autogroup's nice value
(compared to other autogroups)
and the process's nice value
(compared to other processes in the same autogroup.
.PP
The use of the
.BR cgroups (7)
CPU controller to place processes in cgroups other than the
root CPU cgroup overrides the effect of autogrouping.
.PP
The autogroup feature groups only processes scheduled under
non-real-time policies
.RB ( SCHED_OTHER ,
.BR SCHED_BATCH ,
and
.BR SCHED_IDLE ).
It does not group processes scheduled under real-time and
deadline policies.
Those processes are scheduled according to the rules described earlier.
.\"
.SS The nice value and group scheduling
When scheduling non-real-time processes (i.e., those scheduled under the
.BR SCHED_OTHER ,
.BR SCHED_BATCH ,
and
.B SCHED_IDLE
policies), the CFS scheduler employs a technique known as "group scheduling",
if the kernel was configured with the
.B CONFIG_FAIR_GROUP_SCHED
option (which is typical).
.PP
Under group scheduling, threads are scheduled in "task groups".
Task groups have a hierarchical relationship,
rooted under the initial task group on the system,
known as the "root task group".
Task groups are formed in the following circumstances:
.IP \[bu] 3
All of the threads in a CPU cgroup form a task group.
The parent of this task group is the task group of the
corresponding parent cgroup.
.IP \[bu]
If autogrouping is enabled,
then all of the threads that are (implicitly) placed in an autogroup
(i.e., the same session, as created by
.BR setsid (2))
form a task group.
Each new autogroup is thus a separate task group.
The root task group is the parent of all such autogroups.
.IP \[bu]
If autogrouping is enabled, then the root task group consists of
all processes in the root CPU cgroup that were not
otherwise implicitly placed into a new autogroup.
.IP \[bu]
If autogrouping is disabled, then the root task group consists of
all processes in the root CPU cgroup.
.IP \[bu]
If group scheduling was disabled (i.e., the kernel was configured without
.BR CONFIG_FAIR_GROUP_SCHED ),
then all of the processes on the system are notionally placed
in a single task group.
.PP
Under group scheduling,
a thread's nice value has an effect for scheduling decisions
.IR "only relative to other threads in the same task group" .
This has some surprising consequences in terms of the traditional semantics
of the nice value on UNIX systems.
In particular, if autogrouping
is enabled (which is the default in various distributions), then employing
.BR setpriority (2)
or
.BR nice (1)
on a process has an effect only for scheduling relative
to other processes executed in the same session
(typically: the same terminal window).
.PP
Conversely, for two processes that are (for example)
the sole CPU-bound processes in different sessions
(e.g., different terminal windows,
each of whose jobs are tied to different autogroups),
.I modifying the nice value of the process in one of the sessions
.I has no effect
in terms of the scheduler's decisions relative to the
process in the other session.
.\" More succinctly: the nice(1) command is in many cases a no-op since
.\" Linux 2.6.38.
.\"
A possibly useful workaround here is to use a command such as
the following to modify the autogroup nice value for
.I all
of the processes in a terminal session:
.PP
.in +4n
.EX
$ \fBecho 10 > /proc/self/autogroup\fP
.EE
.in
.SS Real-time features in the mainline Linux kernel
.\" FIXME . Probably this text will need some minor tweaking
.\" ask Carsten Emde about this.
Since Linux 2.6.18, Linux is gradually
becoming equipped with real-time capabilities,
most of which are derived from the former
.I realtime\-preempt
patch set.
Until the patches have been completely merged into the
mainline kernel,
they must be installed to achieve the best real-time performance.
These patches are named:
.PP
.in +4n
.EX
patch\-\fIkernelversion\fP\-rt\fIpatchversion\fP
.EE
.in
.PP
and can be downloaded from
.UR http://www.kernel.org\:/pub\:/linux\:/kernel\:/projects\:/rt/
.UE .
.PP
Without the patches and prior to their full inclusion into the mainline
kernel, the kernel configuration offers only the three preemption classes
.BR CONFIG_PREEMPT_NONE ,
.BR CONFIG_PREEMPT_VOLUNTARY ,
and
.B CONFIG_PREEMPT_DESKTOP
which respectively provide no, some, and considerable
reduction of the worst-case scheduling latency.
.PP
With the patches applied or after their full inclusion into the mainline
kernel, the additional configuration item
.B CONFIG_PREEMPT_RT
becomes available.
If this is selected, Linux is transformed into a regular
real-time operating system.
The FIFO and RR scheduling policies are then used to run a thread
with true real-time priority and a minimum worst-case scheduling latency.
.SH NOTES
The
.BR cgroups (7)
CPU controller can be used to limit the CPU consumption of
groups of processes.
.PP
Originally, Standard Linux was intended as a general-purpose operating
system being able to handle background processes, interactive
applications, and less demanding real-time applications (applications that
need to usually meet timing deadlines).
Although the Linux 2.6
allowed for kernel preemption and the newly introduced O(1) scheduler
ensures that the time needed to schedule is fixed and deterministic
irrespective of the number of active tasks, true real-time computing
was not possible up to Linux 2.6.17.
.SH SEE ALSO
.ad l
.nh
.BR chcpu (1),
.BR chrt (1),
.BR lscpu (1),
.BR ps (1),
.BR taskset (1),
.BR top (1),
.BR getpriority (2),
.BR mlock (2),
.BR mlockall (2),
.BR munlock (2),
.BR munlockall (2),
.BR nice (2),
.BR sched_get_priority_max (2),
.BR sched_get_priority_min (2),
.BR sched_getaffinity (2),
.BR sched_getparam (2),
.BR sched_getscheduler (2),
.BR sched_rr_get_interval (2),
.BR sched_setaffinity (2),
.BR sched_setparam (2),
.BR sched_setscheduler (2),
.BR sched_yield (2),
.BR setpriority (2),
.BR pthread_getaffinity_np (3),
.BR pthread_getschedparam (3),
.BR pthread_setaffinity_np (3),
.BR sched_getcpu (3),
.BR capabilities (7),
.BR cpuset (7)
.ad
.PP
.I Programming for the real world \- POSIX.4
by Bill O.\& Gallmeister, O'Reilly & Associates, Inc., ISBN 1-56592-074-0.
.PP
The Linux kernel source files
.IR \%Documentation/\:scheduler/\:sched\-deadline\:.txt ,
.IR \%Documentation/\:scheduler/\:sched\-rt\-group\:.txt ,
.IR \%Documentation/\:scheduler/\:sched\-design\-CFS\:.txt ,
and
.I \%Documentation/\:scheduler/\:sched\-nice\-design\:.txt