1getrusage(2) System Calls Manual getrusage(2)
2
6 getrusage - get resource usage
7
9 Standard C library (libc, -lc)
10
12 #include <sys/resource.h>
13 int getrusage(int who, struct rusage *usage);
15
17 getrusage() returns resource usage measures for who, which can be one
18 of the following:
19 RUSAGE_SELF
21 Return resource usage statistics for the calling process, which
22 is the sum of resources used by all threads in the process.
23 RUSAGE_CHILDREN
25 Return resource usage statistics for all children of the calling
26 process that have terminated and been waited for. These statis‐
27 tics will include the resources used by grandchildren, and fur‐
28 ther removed descendants, if all of the intervening descendants
29 waited on their terminated children.
30 RUSAGE_THREAD (since Linux 2.6.26)
32 Return resource usage statistics for the calling thread. The
33 _GNU_SOURCE feature test macro must be defined (before including
34 any header file) in order to obtain the definition of this con‐
35 stant from <sys/resource.h>.
36 The resource usages are returned in the structure pointed to by usage,
38 which has the following form:
39 struct rusage {
41 struct timeval ru_utime; /* user CPU time used */
42 struct timeval ru_stime; /* system CPU time used */
43 long ru_maxrss; /* maximum resident set size */
44 long ru_ixrss; /* integral shared memory size */
45 long ru_idrss; /* integral unshared data size */
46 long ru_isrss; /* integral unshared stack size */
47 long ru_minflt; /* page reclaims (soft page faults) */
48 long ru_majflt; /* page faults (hard page faults) */
49 long ru_nswap; /* swaps */
50 long ru_inblock; /* block input operations */
51 long ru_oublock; /* block output operations */
52 long ru_msgsnd; /* IPC messages sent */
53 long ru_msgrcv; /* IPC messages received */
54 long ru_nsignals; /* signals received */
55 long ru_nvcsw; /* voluntary context switches */
56 long ru_nivcsw; /* involuntary context switches */
57 };
58 Not all fields are completed; unmaintained fields are set to zero by
60 the kernel. (The unmaintained fields are provided for compatibility
61 with other systems, and because they may one day be supported on
62 Linux.) The fields are interpreted as follows:
63 ru_utime
65 This is the total amount of time spent executing in user mode,
66 expressed in a timeval structure (seconds plus microseconds).
67 ru_stime
69 This is the total amount of time spent executing in kernel mode,
70 expressed in a timeval structure (seconds plus microseconds).
71 ru_maxrss (since Linux 2.6.32)
73 This is the maximum resident set size used (in kilobytes). For
74 RUSAGE_CHILDREN, this is the resident set size of the largest
75 child, not the maximum resident set size of the process tree.
76 ru_ixrss (unmaintained)
78 This field is currently unused on Linux.
79 ru_idrss (unmaintained)
81 This field is currently unused on Linux.
82 ru_isrss (unmaintained)
84 This field is currently unused on Linux.
85 ru_minflt
87 The number of page faults serviced without any I/O activity;
88 here I/O activity is avoided by “reclaiming” a page frame from
89 the list of pages awaiting reallocation.
90 ru_majflt
92 The number of page faults serviced that required I/O activity.
93 ru_nswap (unmaintained)
95 This field is currently unused on Linux.
96 ru_inblock (since Linux 2.6.22)
98 The number of times the filesystem had to perform input.
99 ru_oublock (since Linux 2.6.22)
101 The number of times the filesystem had to perform output.
102 ru_msgsnd (unmaintained)
104 This field is currently unused on Linux.
105 ru_msgrcv (unmaintained)
107 This field is currently unused on Linux.
108 ru_nsignals (unmaintained)
110 This field is currently unused on Linux.
111 ru_nvcsw (since Linux 2.6)
113 The number of times a context switch resulted due to a process
114 voluntarily giving up the processor before its time slice was
115 completed (usually to await availability of a resource).
116 ru_nivcsw (since Linux 2.6)
118 The number of times a context switch resulted due to a higher
119 priority process becoming runnable or because the current
120 process exceeded its time slice.
121
123 On success, zero is returned. On error, -1 is returned, and errno is
124 set to indicate the error.
125
127 EFAULT usage points outside the accessible address space.
128 EINVAL who is invalid.
130
132 For an explanation of the terms used in this section, see at‐
133 tributes(7).
134 ┌────────────────────────────────────────────┬───────────────┬─────────┐
136 │Interface │ Attribute │ Value │
137 ├────────────────────────────────────────────┼───────────────┼─────────┤
138 │getrusage() │ Thread safety │ MT-Safe │
139 └────────────────────────────────────────────┴───────────────┴─────────┘
140
142 POSIX.1-2008.
143 POSIX.1 specifies getrusage(), but specifies only the fields ru_utime
145 and ru_stime.
146 RUSAGE_THREAD is Linux-specific.
148
150 POSIX.1-2001, SVr4, 4.3BSD.
151 Before Linux 2.6.9, if the disposition of SIGCHLD is set to SIG_IGN
153 then the resource usages of child processes are automatically included
154 in the value returned by RUSAGE_CHILDREN, although POSIX.1-2001
155 explicitly prohibits this. This nonconformance is rectified in Linux
156 2.6.9 and later.
157 The structure definition shown at the start of this page was taken from
159 4.3BSD Reno.
160 Ancient systems provided a vtimes() function with a similar purpose to
162 getrusage(). For backward compatibility, glibc (up until Linux 2.32)
163 also provides vtimes(). All new applications should be written using
164 getrusage(). (Since Linux 2.33, glibc no longer provides an vtimes()
165 implementation.)
166
168 Resource usage metrics are preserved across an execve(2).
169 See also the description of /proc/pid/stat in proc(5).
171
173 clock_gettime(2), getrlimit(2), times(2), wait(2), wait4(2), clock(3)
174Linux man-pages 6.05 2023-07-20 getrusage(2)