1GETRLIMIT(2) Linux Programmer's Manual GETRLIMIT(2)
2
6 getrlimit, setrlimit, prlimit - get/set resource limits
7
9 #include <sys/time.h>
10 #include <sys/resource.h>
11 int getrlimit(int resource, struct rlimit *rlim);
13 int setrlimit(int resource, const struct rlimit *rlim);
14 int prlimit(pid_t pid, int resource, const struct rlimit *new_limit,
16 struct rlimit *old_limit);
17 Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
19 prlimit(): _GNU_SOURCE && _FILE_OFFSET_BITS == 64
21
23 The getrlimit() and setrlimit() system calls get and set resource lim‐
24 its respectively. Each resource has an associated soft and hard limit,
25 as defined by the rlimit structure:
26 struct rlimit {
28 rlim_t rlim_cur; /* Soft limit */
29 rlim_t rlim_max; /* Hard limit (ceiling for rlim_cur) */
30 };
31 The soft limit is the value that the kernel enforces for the corre‐
33 sponding resource. The hard limit acts as a ceiling for the soft
34 limit: an unprivileged process may set only its soft limit to a value
35 in the range from 0 up to the hard limit, and (irreversibly) lower its
36 hard limit. A privileged process (under Linux: one with the
37 CAP_SYS_RESOURCE capability) may make arbitrary changes to either limit
38 value.
39 The value RLIM_INFINITY denotes no limit on a resource (both in the
41 structure returned by getrlimit() and in the structure passed to setr‐
42 limit()).
43 The resource argument must be one of:
45 RLIMIT_AS
47 The maximum size of the process's virtual memory (address space)
48 in bytes. This limit affects calls to brk(2), mmap(2) and
49 mremap(2), which fail with the error ENOMEM upon exceeding this
50 limit. Also automatic stack expansion will fail (and generate a
51 SIGSEGV that kills the process if no alternate stack has been
52 made available via sigaltstack(2)). Since the value is a long,
53 on machines with a 32-bit long either this limit is at most 2
54 GiB, or this resource is unlimited.
55 RLIMIT_CORE
57 Maximum size of core file. When 0 no core dump files are cre‐
58 ated. When nonzero, larger dumps are truncated to this size.
59 RLIMIT_CPU
61 CPU time limit in seconds. When the process reaches the soft
62 limit, it is sent a SIGXCPU signal. The default action for this
63 signal is to terminate the process. However, the signal can be
64 caught, and the handler can return control to the main program.
65 If the process continues to consume CPU time, it will be sent
66 SIGXCPU once per second until the hard limit is reached, at
67 which time it is sent SIGKILL. (This latter point describes
68 Linux behavior. Implementations vary in how they treat pro‐
69 cesses which continue to consume CPU time after reaching the
70 soft limit. Portable applications that need to catch this sig‐
71 nal should perform an orderly termination upon first receipt of
72 SIGXCPU.)
73 RLIMIT_DATA
75 The maximum size of the process's data segment (initialized
76 data, uninitialized data, and heap). This limit affects calls
77 to brk(2) and sbrk(2), which fail with the error ENOMEM upon
78 encountering the soft limit of this resource.
79 RLIMIT_FSIZE
81 The maximum size of files that the process may create. Attempts
82 to extend a file beyond this limit result in delivery of a
83 SIGXFSZ signal. By default, this signal terminates a process,
84 but a process can catch this signal instead, in which case the
85 relevant system call (e.g., write(2), truncate(2)) fails with
86 the error EFBIG.
87 RLIMIT_LOCKS (Early Linux 2.4 only)
89 A limit on the combined number of flock(2) locks and fcntl(2)
90 leases that this process may establish.
91 RLIMIT_MEMLOCK
93 The maximum number of bytes of memory that may be locked into
94 RAM. In effect this limit is rounded down to the nearest multi‐
95 ple of the system page size. This limit affects mlock(2) and
96 mlockall(2) and the mmap(2) MAP_LOCKED operation. Since Linux
97 2.6.9 it also affects the shmctl(2) SHM_LOCK operation, where it
98 sets a maximum on the total bytes in shared memory segments (see
99 shmget(2)) that may be locked by the real user ID of the calling
100 process. The shmctl(2) SHM_LOCK locks are accounted for sepa‐
101 rately from the per-process memory locks established by
102 mlock(2), mlockall(2), and mmap(2) MAP_LOCKED; a process can
103 lock bytes up to this limit in each of these two categories. In
104 Linux kernels before 2.6.9, this limit controlled the amount of
105 memory that could be locked by a privileged process. Since
106 Linux 2.6.9, no limits are placed on the amount of memory that a
107 privileged process may lock, and this limit instead governs the
108 amount of memory that an unprivileged process may lock.
109 RLIMIT_MSGQUEUE (Since Linux 2.6.8)
111 Specifies the limit on the number of bytes that can be allocated
112 for POSIX message queues for the real user ID of the calling
113 process. This limit is enforced for mq_open(3). Each message
114 queue that the user creates counts (until it is removed) against
115 this limit according to the formula:
116 bytes = attr.mq_maxmsg * sizeof(struct msg_msg *) +
118 attr.mq_maxmsg * attr.mq_msgsize
119 where attr is the mq_attr structure specified as the fourth
121 argument to mq_open(3).
122 The first addend in the formula, which includes sizeof(struct
124 msg_msg *) (4 bytes on Linux/i386), ensures that the user cannot
125 create an unlimited number of zero-length messages (such mes‐
126 sages nevertheless each consume some system memory for bookkeep‐
127 ing overhead).
128 RLIMIT_NICE (since Linux 2.6.12, but see BUGS below)
130 Specifies a ceiling to which the process's nice value can be
131 raised using setpriority(2) or nice(2). The actual ceiling for
132 the nice value is calculated as 20 - rlim_cur. (This strange‐
133 ness occurs because negative numbers cannot be specified as
134 resource limit values, since they typically have special mean‐
135 ings. For example, RLIM_INFINITY typically is the same as -1.)
136 RLIMIT_NOFILE
138 Specifies a value one greater than the maximum file descriptor
139 number that can be opened by this process. Attempts (open(2),
140 pipe(2), dup(2), etc.) to exceed this limit yield the error
141 EMFILE. (Historically, this limit was named RLIMIT_OFILE on
142 BSD.)
143 RLIMIT_NPROC
145 The maximum number of processes (or, more precisely on Linux,
146 threads) that can be created for the real user ID of the calling
147 process. Upon encountering this limit, fork(2) fails with the
148 error EAGAIN.
149 RLIMIT_RSS
151 Specifies the limit (in pages) of the process's resident set
152 (the number of virtual pages resident in RAM). This limit has
153 effect only in Linux 2.4.x, x < 30, and there affects only calls
154 to madvise(2) specifying MADV_WILLNEED.
155 RLIMIT_RTPRIO (Since Linux 2.6.12, but see BUGS)
157 Specifies a ceiling on the real-time priority that may be set
158 for this process using sched_setscheduler(2) and sched_set‐
159 param(2).
160 RLIMIT_RTTIME (Since Linux 2.6.25)
162 Specifies a limit (in microseconds) on the amount of CPU time
163 that a process scheduled under a real-time scheduling policy may
164 consume without making a blocking system call. For the purpose
165 of this limit, each time a process makes a blocking system call,
166 the count of its consumed CPU time is reset to zero. The CPU
167 time count is not reset if the process continues trying to use
168 the CPU but is preempted, its time slice expires, or it calls
169 sched_yield(2).
170 Upon reaching the soft limit, the process is sent a SIGXCPU sig‐
172 nal. If the process catches or ignores this signal and contin‐
173 ues consuming CPU time, then SIGXCPU will be generated once each
174 second until the hard limit is reached, at which point the
175 process is sent a SIGKILL signal.
176 The intended use of this limit is to stop a runaway real-time
178 process from locking up the system.
179 RLIMIT_SIGPENDING (Since Linux 2.6.8)
181 Specifies the limit on the number of signals that may be queued
182 for the real user ID of the calling process. Both standard and
183 real-time signals are counted for the purpose of checking this
184 limit. However, the limit is enforced only for sigqueue(3); it
185 is always possible to use kill(2) to queue one instance of any
186 of the signals that are not already queued to the process.
187 RLIMIT_STACK
189 The maximum size of the process stack, in bytes. Upon reaching
190 this limit, a SIGSEGV signal is generated. To handle this sig‐
191 nal, a process must employ an alternate signal stack (sigalt‐
192 stack(2)).
193 Since Linux 2.6.23, this limit also determines the amount of
195 space used for the process's command-line arguments and environ‐
196 ment variables; for details, see execve(2).
197 prlimit()
199 The Linux-specific prlimit() system call combines and extends the func‐
200 tionality of setrlimit() and getrlimit(). It can be used to both set
201 and get the resource limits of an arbitrary process.
202 The resource argument has the same meaning as for setrlimit() and getr‐
204 limit().
205 If the new_limit argument is a not NULL, then the rlimit structure to
207 which it points is used to set new values for the soft and hard limits
208 for resource. If the old_limit argument is a not NULL, then a success‐
209 ful call to prlimit() places the previous soft and hard limits for
210 resource in the rlimit structure pointed to by old_limit.
211 The pid argument specifies the ID of the process on which the call is
213 to operate. If pid is 0, then the call applies to the calling process.
214 To set or get the resources of a process other than itself, the caller
215 must have the CAP_SYS_RESOURCE capability, or the real, effective, and
216 saved set user IDs of the target process must match the real user ID of
217 the caller and the real, effective, and saved set group IDs of the tar‐
218 get process must match the real group ID of the caller.
219
221 On success, these system calls return 0. On error, -1 is returned, and
222 errno is set appropriately.
223
225 EFAULT A pointer argument points to a location outside the accessible
226 address space.
227 EINVAL The value specified in resource is not valid; or, for setr‐
229 limit() or prlimit(): rlim->rlim_cur was greater than
230 rlim->rlim_max.
231 EPERM An unprivileged process tried to raise the hard limit; the
233 CAP_SYS_RESOURCE capability is required to do this. Or, the
234 caller tried to increase the hard RLIMIT_NOFILE limit above the
235 current kernel maximum (NR_OPEN). Or, the calling process did
236 not have permission to set limits for the process specified by
237 pid.
238 ESRCH Could not find a process with the ID specified in pid.
240
242 The prlimit() system call is available since Linux 2.6.36. Library
243 support is available since glibc 2.13.
244
246 getrlimit(), setrlimit(): SVr4, 4.3BSD, POSIX.1-2001.
247 prlimit(): Linux-specific.
248 RLIMIT_MEMLOCK and RLIMIT_NPROC derive from BSD and are not specified
250 in POSIX.1-2001; they are present on the BSDs and Linux, but on few
251 other implementations. RLIMIT_RSS derives from BSD and is not speci‐
252 fied in POSIX.1-2001; it is nevertheless present on most implementa‐
253 tions. RLIMIT_MSGQUEUE, RLIMIT_NICE, RLIMIT_RTPRIO, RLIMIT_RTTIME, and
254 RLIMIT_SIGPENDING are Linux-specific.
255
257 A child process created via fork(2) inherits its parent's resource lim‐
258 its. Resource limits are preserved across execve(2).
259 One can set the resource limits of the shell using the built-in ulimit
261 command (limit in csh(1)). The shell's resource limits are inherited
262 by the processes that it creates to execute commands.
263 Since Linux 2.6.24, the resource limits of any process can be inspected
265 via /proc/[pid]/limits; see proc(5).
266 Ancient systems provided a vlimit() function with a similar purpose to
268 setrlimit(). For backward compatibility, glibc also provides vlimit().
269 All new applications should be written using setrlimit().
270
272 In older Linux kernels, the SIGXCPU and SIGKILL signals delivered when
273 a process encountered the soft and hard RLIMIT_CPU limits were deliv‐
274 ered one (CPU) second later than they should have been. This was fixed
275 in kernel 2.6.8.
276 In 2.6.x kernels before 2.6.17, a RLIMIT_CPU limit of 0 is wrongly
278 treated as "no limit" (like RLIM_INFINITY). Since Linux 2.6.17, set‐
279 ting a limit of 0 does have an effect, but is actually treated as a
280 limit of 1 second.
281 A kernel bug means that RLIMIT_RTPRIO does not work in kernel 2.6.12;
283 the problem is fixed in kernel 2.6.13.
284 In kernel 2.6.12, there was an off-by-one mismatch between the priority
286 ranges returned by getpriority(2) and RLIMIT_NICE. This had the effect
287 that the actual ceiling for the nice value was calculated as
288 19 - rlim_cur. This was fixed in kernel 2.6.13.
289 Since Linux 2.6.12, if a process reaches its soft RLIMIT_CPU limit and
291 has a handler installed for SIGXCPU, then, in addition to invoking the
292 signal handler, the kernel increases the soft limit by one second.
293 This behavior repeats if the process continues to consume CPU time,
294 until the hard limit is reached, at which point the process is killed.
295 Other implementations do not change the RLIMIT_CPU soft limit in this
296 manner, and the Linux behavior is probably not standards conformant;
297 portable applications should avoid relying on this Linux-specific
298 behavior. The Linux-specific RLIMIT_RTTIME limit exhibits the same
299 behavior when the soft limit is encountered.
300 Kernels before 2.4.22 did not diagnose the error EINVAL for setrlimit()
302 when rlim->rlim_cur was greater than rlim->rlim_max.
303
305 The program below demonstrates the use of prlimit().
306 #define _GNU_SOURCE
308 #define _FILE_OFFSET_BITS 64
309 #include <stdio.h>
310 #include <time.h>
311 #include <stdlib.h>
312 #include <unistd.h>
313 #include <sys/resource.h>
314 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
316 } while (0)
317 int
319 main(int argc, char *argv[])
320 {
321 struct rlimit old, new;
322 struct rlimit *newp;
323 pid_t pid;
324 if (!(argc == 2 || argc == 4)) {
326 fprintf(stderr, "Usage: %s <pid> [<new-soft-limit> "
327 "<new-hard-limit>]\n", argv[0]);
328 exit(EXIT_FAILURE);
329 }
330 pid = atoi(argv[1]); /* PID of target process */
332 newp = NULL;
334 if (argc == 4) {
335 new.rlim_cur = atoi(argv[2]);
336 new.rlim_max = atoi(argv[3]);
337 newp = &new;
338 }
339 /* Set CPU time limit of target process; retrieve and display
341 previous limit */
342 if (prlimit(pid, RLIMIT_CPU, newp, &old) == -1)
344 errExit("prlimit-1");
345 printf("Previous limits: soft=%lld; hard=%lld\n",
346 (long long) old.rlim_cur, (long long) old.rlim_max);
347 /* Retrieve and display new CPU time limit */
349 if (prlimit(pid, RLIMIT_CPU, NULL, &old) == -1)
351 errExit("prlimit-2");
352 printf("New limits: soft=%lld; hard=%lld\n",
353 (long long) old.rlim_cur, (long long) old.rlim_max);
354 exit(EXIT_FAILURE);
356 }
357
359 prlimit(1), dup(2), fcntl(2), fork(2), getrusage(2), mlock(2), mmap(2),
360 open(2), quotactl(2), sbrk(2), shmctl(2), malloc(3), sigqueue(3),
361 ulimit(3), core(5), capabilities(7), signal(7)
362
364 This page is part of release 3.53 of the Linux man-pages project. A
365 description of the project, and information about reporting bugs, can
366 be found at http://www.kernel.org/doc/man-pages/.
367Linux 2013-02-11 GETRLIMIT(2)