1 Real-Time group scheduling
2 --------------------------
12 2.1 System-wide settings
14 2.3 Basis for grouping tasks
21 Fiddling with these settings can result in an unstable system, the knobs are
22 root only and assumes root knows what he is doing.
26 * very small values in sched_rt_period_us can result in an unstable
27 system when the period is smaller than either the available hrtimer
28 resolution, or the time it takes to handle the budget refresh itself.
30 * very small values in sched_rt_runtime_us can result in an unstable
31 system when the runtime is so small the system has difficulty making
32 forward progress (NOTE: the migration thread and kstopmachine both
33 are real-time processes).
42 Realtime scheduling is all about determinism, a group has to be able to rely on
43 the amount of bandwidth (eg. CPU time) being constant. In order to schedule
44 multiple groups of realtime tasks, each group must be assigned a fixed portion
45 of the CPU time available. Without a minimum guarantee a realtime group can
46 obviously fall short. A fuzzy upper limit is of no use since it cannot be
47 relied upon. Which leaves us with just the single fixed portion.
52 CPU time is divided by means of specifying how much time can be spent running
53 in a given period. We allocate this "run time" for each realtime group which
54 the other realtime groups will not be permitted to use.
56 Any time not allocated to a realtime group will be used to run normal priority
57 tasks (SCHED_OTHER). Any allocated run time not used will also be picked up by
60 Let's consider an example: a frame fixed realtime renderer must deliver 25
61 frames a second, which yields a period of 0.04s per frame. Now say it will also
62 have to play some music and respond to input, leaving it with around 80% CPU
63 time dedicated for the graphics. We can then give this group a run time of 0.8
66 This way the graphics group will have a 0.04s period with a 0.032s run time
67 limit. Now if the audio thread needs to refill the DMA buffer every 0.005s, but
68 needs only about 3% CPU time to do so, it can do with a 0.03 * 0.005s =
69 0.00015s. So this group can be scheduled with a period of 0.005s and a run time
72 The remaining CPU time will be used for user input and other tasks. Because
73 realtime tasks have explicitly allocated the CPU time they need to perform
74 their tasks, buffer underruns in the graphics or audio can be eliminated.
76 NOTE: the above example is not fully implemented as of yet (2.6.25). We still
77 lack an EDF scheduler to make non-uniform periods usable.
84 2.1 System wide settings
85 ------------------------
87 The system wide settings are configured under the /proc virtual file system:
89 /proc/sys/kernel/sched_rt_period_us:
90 The scheduling period that is equivalent to 100% CPU bandwidth
92 /proc/sys/kernel/sched_rt_runtime_us:
93 A global limit on how much time realtime scheduling may use. Even without
94 CONFIG_RT_GROUP_SCHED enabled, this will limit time reserved to realtime
95 processes. With CONFIG_RT_GROUP_SCHED it signifies the total bandwidth
96 available to all realtime groups.
98 * Time is specified in us because the interface is s32. This gives an
99 operating range from 1us to about 35 minutes.
100 * sched_rt_period_us takes values from 1 to INT_MAX.
101 * sched_rt_runtime_us takes values from -1 to (INT_MAX - 1).
102 * A run time of -1 specifies runtime == period, ie. no limit.
105 2.2 Default behaviour
106 ---------------------
108 The default values for sched_rt_period_us (1000000 or 1s) and
109 sched_rt_runtime_us (950000 or 0.95s). This gives 0.05s to be used by
110 SCHED_OTHER (non-RT tasks). These defaults were chosen so that a run-away
111 realtime tasks will not lock up the machine but leave a little time to recover
112 it. By setting runtime to -1 you'd get the old behaviour back.
114 By default all bandwidth is assigned to the root group and new groups get the
115 period from /proc/sys/kernel/sched_rt_period_us and a run time of 0. If you
116 want to assign bandwidth to another group, reduce the root group's bandwidth
117 and assign some or all of the difference to another group.
119 Realtime group scheduling means you have to assign a portion of total CPU
120 bandwidth to the group before it will accept realtime tasks. Therefore you will
121 not be able to run realtime tasks as any user other than root until you have
122 done that, even if the user has the rights to run processes with realtime
126 2.3 Basis for grouping tasks
127 ----------------------------
129 There are two compile-time settings for allocating CPU bandwidth. These are
130 configured using the "Basis for grouping tasks" multiple choice menu under
131 General setup > Group CPU Scheduler:
133 a. CONFIG_USER_SCHED (aka "Basis for grouping tasks" = "user id")
135 This lets you use the virtual files under
136 "/sys/kernel/uids/<uid>/cpu_rt_runtime_us" to control he CPU time reserved for
141 .o CONFIG_CGROUP_SCHED (aka "Basis for grouping tasks" = "Control groups")
143 This uses the /cgroup virtual file system and "/cgroup/<cgroup>/cpu.rt_runtime_us"
144 to control the CPU time reserved for each control group instead.
146 For more information on working with control groups, you should read
147 Documentation/cgroups/cgroups.txt as well.
149 Group settings are checked against the following limits in order to keep the configuration
152 \Sum_{i} runtime_{i} / global_period <= global_runtime / global_period
154 For now, this can be simplified to just the following (but see Future plans):
156 \Sum_{i} runtime_{i} <= global_runtime
162 There is work in progress to make the scheduling period for each group
163 ("/sys/kernel/uids/<uid>/cpu_rt_period_us" or
164 "/cgroup/<cgroup>/cpu.rt_period_us" respectively) configurable as well.
166 The constraint on the period is that a subgroup must have a smaller or
167 equal period to its parent. But realistically its not very useful _yet_
168 as its prone to starvation without deadline scheduling.
170 Consider two sibling groups A and B; both have 50% bandwidth, but A's
171 period is twice the length of B's.
173 * group A: period=100000us, runtime=10000us
174 - this runs for 0.01s once every 0.1s
176 * group B: period= 50000us, runtime=10000us
177 - this runs for 0.01s twice every 0.1s (or once every 0.05 sec).
179 This means that currently a while (1) loop in A will run for the full period of
180 B and can starve B's tasks (assuming they are of lower priority) for a whole
183 The next project will be SCHED_EDF (Earliest Deadline First scheduling) to bring
184 full deadline scheduling to the linux kernel. Deadline scheduling the above
185 groups and treating end of the period as a deadline will ensure that they both
186 get their allocated time.
188 Implementing SCHED_EDF might take a while to complete. Priority Inheritance is
189 the biggest challenge as the current linux PI infrastructure is geared towards
190 the limited static priority levels 0-99. With deadline scheduling you need to
191 do deadline inheritance (since priority is inversely proportional to the
192 deadline delta (deadline - now).
194 This means the whole PI machinery will have to be reworked - and that is one of
195 the most complex pieces of code we have.