1SCHED_SETAFFINITY(2) Linux Programmer's Manual SCHED_SETAFFINITY(2)
2
6 sched_setaffinity, sched_getaffinity - set and get a thread's CPU
7 affinity mask
8
10 #define _GNU_SOURCE /* See feature_test_macros(7) */
11 #include <sched.h>
12 int sched_setaffinity(pid_t pid, size_t cpusetsize,
14 const cpu_set_t *mask);
15 int sched_getaffinity(pid_t pid, size_t cpusetsize,
17 cpu_set_t *mask);
18
20 A thread's CPU affinity mask determines the set of CPUs on which it is
21 eligible to run. On a multiprocessor system, setting the CPU affinity
22 mask can be used to obtain performance benefits. For example, by dedi‐
23 cating one CPU to a particular thread (i.e., setting the affinity mask
24 of that thread to specify a single CPU, and setting the affinity mask
25 of all other threads to exclude that CPU), it is possible to ensure
26 maximum execution speed for that thread. Restricting a thread to run
27 on a single CPU also avoids the performance cost caused by the cache
28 invalidation that occurs when a thread ceases to execute on one CPU and
29 then recommences execution on a different CPU.
30 A CPU affinity mask is represented by the cpu_set_t structure, a "CPU
32 set", pointed to by mask. A set of macros for manipulating CPU sets is
33 described in CPU_SET(3).
34 sched_setaffinity() sets the CPU affinity mask of the thread whose ID
36 is pid to the value specified by mask. If pid is zero, then the call‐
37 ing thread is used. The argument cpusetsize is the length (in bytes)
38 of the data pointed to by mask. Normally this argument would be speci‐
39 fied as sizeof(cpu_set_t).
40 If the thread specified by pid is not currently running on one of the
42 CPUs specified in mask, then that thread is migrated to one of the CPUs
43 specified in mask.
44 sched_getaffinity() writes the affinity mask of the thread whose ID is
46 pid into the cpu_set_t structure pointed to by mask. The cpusetsize
47 argument specifies the size (in bytes) of mask. If pid is zero, then
48 the mask of the calling thread is returned.
49
51 On success, sched_setaffinity() and sched_getaffinity() return 0. On
52 error, -1 is returned, and errno is set appropriately.
53
55 EFAULT A supplied memory address was invalid.
56 EINVAL The affinity bit mask mask contains no processors that are cur‐
58 rently physically on the system and permitted to the thread
59 according to any restrictions that may be imposed by cpuset
60 cgroups or the "cpuset" mechanism described in cpuset(7).
61 EINVAL (sched_getaffinity() and, in kernels before 2.6.9,
63 sched_setaffinity()) cpusetsize is smaller than the size of the
64 affinity mask used by the kernel.
65 EPERM (sched_setaffinity()) The calling thread does not have appropri‐
67 ate privileges. The caller needs an effective user ID equal to
68 the real user ID or effective user ID of the thread identified
69 by pid, or it must possess the CAP_SYS_NICE capability in the
70 user namespace of the thread pid.
71 ESRCH The thread whose ID is pid could not be found.
73
75 The CPU affinity system calls were introduced in Linux kernel 2.5.8.
76 The system call wrappers were introduced in glibc 2.3. Initially, the
77 glibc interfaces included a cpusetsize argument, typed as unsigned int.
78 In glibc 2.3.3, the cpusetsize argument was removed, but was then
79 restored in glibc 2.3.4, with type size_t.
80
82 These system calls are Linux-specific.
83
85 After a call to sched_setaffinity(), the set of CPUs on which the
86 thread will actually run is the intersection of the set specified in
87 the mask argument and the set of CPUs actually present on the system.
88 The system may further restrict the set of CPUs on which the thread
89 runs if the "cpuset" mechanism described in cpuset(7) is being used.
90 These restrictions on the actual set of CPUs on which the thread will
91 run are silently imposed by the kernel.
92 There are various ways of determining the number of CPUs available on
94 the system, including: inspecting the contents of /proc/cpuinfo; using
95 sysconf(3) to obtain the values of the _SC_NPROCESSORS_CONF and
96 _SC_NPROCESSORS_ONLN parameters; and inspecting the list of CPU direc‐
97 tories under /sys/devices/system/cpu/.
98 sched(7) has a description of the Linux scheduling scheme.
100 The affinity mask is a per-thread attribute that can be adjusted inde‐
102 pendently for each of the threads in a thread group. The value
103 returned from a call to gettid(2) can be passed in the argument pid.
104 Specifying pid as 0 will set the attribute for the calling thread, and
105 passing the value returned from a call to getpid(2) will set the
106 attribute for the main thread of the thread group. (If you are using
107 the POSIX threads API, then use pthread_setaffinity_np(3) instead of
108 sched_setaffinity().)
109 The isolcpus boot option can be used to isolate one or more CPUs at
111 boot time, so that no processes are scheduled onto those CPUs. Follow‐
112 ing the use of this boot option, the only way to schedule processes
113 onto the isolated CPUs is via sched_setaffinity() or the cpuset(7)
114 mechanism. For further information, see the kernel source file Docu‐
115 mentation/admin-guide/kernel-parameters.txt. As noted in that file,
116 isolcpus is the preferred mechanism of isolating CPUs (versus the
117 alternative of manually setting the CPU affinity of all processes on
118 the system).
119 A child created via fork(2) inherits its parent's CPU affinity mask.
121 The affinity mask is preserved across an execve(2).
122 C library/kernel differences
124 This manual page describes the glibc interface for the CPU affinity
125 calls. The actual system call interface is slightly different, with
126 the mask being typed as unsigned long *, reflecting the fact that the
127 underlying implementation of CPU sets is a simple bit mask. On suc‐
128 cess, the raw sched_getaffinity() system call returns the size (in
129 bytes) of the cpumask_t data type that is used internally by the kernel
130 to represent the CPU set bit mask.
131 Handling systems with large CPU affinity masks
133 The underlying system calls (which represent CPU masks as bit masks of
134 type unsigned long *) impose no restriction on the size of the CPU
135 mask. However, the cpu_set_t data type used by glibc has a fixed size
136 of 128 bytes, meaning that the maximum CPU number that can be repre‐
137 sented is 1023. If the kernel CPU affinity mask is larger than 1024,
138 then calls of the form:
139 sched_getaffinity(pid, sizeof(cpu_set_t), &mask);
141 fail with the error EINVAL, the error produced by the underlying system
143 call for the case where the mask size specified in cpusetsize is
144 smaller than the size of the affinity mask used by the kernel.
145 (Depending on the system CPU topology, the kernel affinity mask can be
146 substantially larger than the number of active CPUs in the system.)
147 When working on systems with large kernel CPU affinity masks, one must
149 dynamically allocate the mask argument (see CPU_ALLOC(3)). Currently,
150 the only way to do this is by probing for the size of the required mask
151 using sched_getaffinity() calls with increasing mask sizes (until the
152 call does not fail with the error EINVAL).
153 Be aware that CPU_ALLOC(3) may allocate a slightly larger CPU set than
155 requested (because CPU sets are implemented as bit masks allocated in
156 units of sizeof(long)). Consequently, sched_getaffinity() can set bits
157 beyond the requested allocation size, because the kernel sees a few
158 additional bits. Therefore, the caller should iterate over the bits in
159 the returned set, counting those which are set, and stop upon reaching
160 the value returned by CPU_COUNT(3) (rather than iterating over the num‐
161 ber of bits requested to be allocated).
162
164 The program below creates a child process. The parent and child then
165 each assign themselves to a specified CPU and execute identical loops
166 that consume some CPU time. Before terminating, the parent waits for
167 the child to complete. The program takes three command-line arguments:
168 the CPU number for the parent, the CPU number for the child, and the
169 number of loop iterations that both processes should perform.
170 As the sample runs below demonstrate, the amount of real and CPU time
172 consumed when running the program will depend on intra-core caching
173 effects and whether the processes are using the same CPU.
174 We first employ lscpu(1) to determine that this (x86) system has two
176 cores, each with two CPUs:
177 $ lscpu | grep -i 'core.*:|socket'
179 Thread(s) per core: 2
180 Core(s) per socket: 2
181 Socket(s): 1
182 We then time the operation of the example program for three cases: both
184 processes running on the same CPU; both processes running on different
185 CPUs on the same core; and both processes running on different CPUs on
186 different cores.
187 $ time -p ./a.out 0 0 100000000
189 real 14.75
190 user 3.02
191 sys 11.73
192 $ time -p ./a.out 0 1 100000000
193 real 11.52
194 user 3.98
195 sys 19.06
196 $ time -p ./a.out 0 3 100000000
197 real 7.89
198 user 3.29
199 sys 12.07
200 Program source
202 #define _GNU_SOURCE
204 #include <sched.h>
205 #include <stdio.h>
206 #include <stdlib.h>
207 #include <unistd.h>
208 #include <sys/wait.h>
209 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
211 } while (0)
212 int
214 main(int argc, char *argv[])
215 {
216 cpu_set_t set;
217 int parentCPU, childCPU;
218 int nloops, j;
219 if (argc != 4) {
221 fprintf(stderr, "Usage: %s parent-cpu child-cpu num-loops\n",
222 argv[0]);
223 exit(EXIT_FAILURE);
224 }
225 parentCPU = atoi(argv[1]);
227 childCPU = atoi(argv[2]);
228 nloops = atoi(argv[3]);
229 CPU_ZERO(&set);
231 switch (fork()) {
233 case -1: /* Error */
234 errExit("fork");
235 case 0: /* Child */
237 CPU_SET(childCPU, &set);
238 if (sched_setaffinity(getpid(), sizeof(set), &set) == -1)
240 errExit("sched_setaffinity");
241 for (j = 0; j < nloops; j++)
243 getppid();
244 exit(EXIT_SUCCESS);
246 default: /* Parent */
248 CPU_SET(parentCPU, &set);
249 if (sched_setaffinity(getpid(), sizeof(set), &set) == -1)
251 errExit("sched_setaffinity");
252 for (j = 0; j < nloops; j++)
254 getppid();
255 wait(NULL); /* Wait for child to terminate */
257 exit(EXIT_SUCCESS);
258 }
259 }
260
262 lscpu(1), nproc(1), taskset(1), clone(2), getcpu(2), getpriority(2),
263 gettid(2), nice(2), sched_get_priority_max(2),
264 sched_get_priority_min(2), sched_getscheduler(2),
265 sched_setscheduler(2), setpriority(2), CPU_SET(3), get_nprocs(3),
266 pthread_setaffinity_np(3), sched_getcpu(3), capabilities(7), cpuset(7),
267 sched(7), numactl(8)
268
270 This page is part of release 4.16 of the Linux man-pages project. A
271 description of the project, information about reporting bugs, and the
272 latest version of this page, can be found at
273 https://www.kernel.org/doc/man-pages/.
274Linux 2017-09-15 SCHED_SETAFFINITY(2)