1GETRUSAGE(2) Linux Programmer's Manual GETRUSAGE(2)
2
6 getrusage - get resource usage
7
9 #include <sys/time.h>
10 #include <sys/resource.h>
11 int getrusage(int who, struct rusage *usage);
13
15 getrusage() returns resource usage measures for who, which can be one
16 of the following:
17 RUSAGE_SELF
19 Return resource usage statistics for the calling process, which
20 is the sum of resources used by all threads in the process.
21 RUSAGE_CHILDREN
23 Return resource usage statistics for all children of the calling
24 process that have terminated and been waited for. These statis‐
25 tics will include the resources used by grandchildren, and fur‐
26 ther removed descendants, if all of the intervening descendants
27 waited on their terminated children.
28 RUSAGE_THREAD (since Linux 2.6.26)
30 Return resource usage statistics for the calling thread. The
31 _GNU_SOURCE feature test macro must be defined (before including
32 any header file) in order to obtain the definition of this con‐
33 stant from <sys/resource.h>.
34 The resource usages are returned in the structure pointed to by usage,
36 which has the following form:
37 struct rusage {
39 struct timeval ru_utime; /* user CPU time used */
40 struct timeval ru_stime; /* system CPU time used */
41 long ru_maxrss; /* maximum resident set size */
42 long ru_ixrss; /* integral shared memory size */
43 long ru_idrss; /* integral unshared data size */
44 long ru_isrss; /* integral unshared stack size */
45 long ru_minflt; /* page reclaims (soft page faults) */
46 long ru_majflt; /* page faults (hard page faults) */
47 long ru_nswap; /* swaps */
48 long ru_inblock; /* block input operations */
49 long ru_oublock; /* block output operations */
50 long ru_msgsnd; /* IPC messages sent */
51 long ru_msgrcv; /* IPC messages received */
52 long ru_nsignals; /* signals received */
53 long ru_nvcsw; /* voluntary context switches */
54 long ru_nivcsw; /* involuntary context switches */
55 };
56 Not all fields are completed; unmaintained fields are set to zero by
58 the kernel. (The unmaintained fields are provided for compatibility
59 with other systems, and because they may one day be supported on
60 Linux.) The fields are interpreted as follows:
61 ru_utime
63 This is the total amount of time spent executing in user mode,
64 expressed in a timeval structure (seconds plus microseconds).
65 ru_stime
67 This is the total amount of time spent executing in kernel mode,
68 expressed in a timeval structure (seconds plus microseconds).
69 ru_maxrss (since Linux 2.6.32)
71 This is the maximum resident set size used (in kilobytes). For
72 RUSAGE_CHILDREN, this is the resident set size of the largest
73 child, not the maximum resident set size of the process tree.
74 ru_ixrss (unmaintained)
76 This field is currently unused on Linux.
77 ru_idrss (unmaintained)
79 This field is currently unused on Linux.
80 ru_isrss (unmaintained)
82 This field is currently unused on Linux.
83 ru_minflt
85 The number of page faults serviced without any I/O activity;
86 here I/O activity is avoided by “reclaiming” a page frame from
87 the list of pages awaiting reallocation.
88 ru_majflt
90 The number of page faults serviced that required I/O activity.
91 ru_nswap (unmaintained)
93 This field is currently unused on Linux.
94 ru_inblock (since Linux 2.6.22)
96 The number of times the filesystem had to perform input.
97 ru_oublock (since Linux 2.6.22)
99 The number of times the filesystem had to perform output.
100 ru_msgsnd (unmaintained)
102 This field is currently unused on Linux.
103 ru_msgrcv (unmaintained)
105 This field is currently unused on Linux.
106 ru_nsignals (unmaintained)
108 This field is currently unused on Linux.
109 ru_nvcsw (since Linux 2.6)
111 The number of times a context switch resulted due to a process
112 voluntarily giving up the processor before its time slice was
113 completed (usually to await availability of a resource).
114 ru_nivcsw (since Linux 2.6)
116 The number of times a context switch resulted due to a higher
117 priority process becoming runnable or because the current
118 process exceeded its time slice.
119
121 On success, zero is returned. On error, -1 is returned, and errno is
122 set appropriately.
123
125 EFAULT usage points outside the accessible address space.
126 EINVAL who is invalid.
128
130 For an explanation of the terms used in this section, see
131 attributes(7).
132 ┌────────────┬───────────────┬─────────┐
134 │Interface │ Attribute │ Value │
135 ├────────────┼───────────────┼─────────┤
136 │getrusage() │ Thread safety │ MT-Safe │
137 └────────────┴───────────────┴─────────┘
138
140 POSIX.1-2001, POSIX.1-2008, SVr4, 4.3BSD. POSIX.1 specifies
141 getrusage(), but specifies only the fields ru_utime and ru_stime.
142 RUSAGE_THREAD is Linux-specific.
144
146 Resource usage metrics are preserved across an execve(2).
147 Including <sys/time.h> is not required these days, but increases porta‐
149 bility. (Indeed, struct timeval is defined in <sys/time.h>.)
150 In Linux kernel versions before 2.6.9, if the disposition of SIGCHLD is
152 set to SIG_IGN then the resource usages of child processes are automat‐
153 ically included in the value returned by RUSAGE_CHILDREN, although
154 POSIX.1-2001 explicitly prohibits this. This nonconformance is recti‐
155 fied in Linux 2.6.9 and later.
156 The structure definition shown at the start of this page was taken from
158 4.3BSD Reno.
159 Ancient systems provided a vtimes() function with a similar purpose to
161 getrusage(). For backward compatibility, glibc also provides vtimes().
162 All new applications should be written using getrusage().
163 See also the description of /proc/[pid]/stat in proc(5).
165
167 clock_gettime(2), getrlimit(2), times(2), wait(2), wait4(2), clock(3)
168
170 This page is part of release 5.02 of the Linux man-pages project. A
171 description of the project, information about reporting bugs, and the
172 latest version of this page, can be found at
173 https://www.kernel.org/doc/man-pages/.
174Linux 2017-09-15 GETRUSAGE(2)