1 .\" Copyright (c) 1992 Drew Eckhardt, March 28, 1992
2 .\" and Copyright (c) 2002, 2004, 2005, 2008, 2010 Michael Kerrisk
4 .\" %%%LICENSE_START(VERBATIM)
5 .\" Permission is granted to make and distribute verbatim copies of this
6 .\" manual provided the copyright notice and this permission notice are
7 .\" preserved on all copies.
9 .\" Permission is granted to copy and distribute modified versions of this
10 .\" manual under the conditions for verbatim copying, provided that the
11 .\" entire resulting derived work is distributed under the terms of a
12 .\" permission notice identical to this one.
14 .\" Since the Linux kernel and libraries are constantly changing, this
15 .\" manual page may be incorrect or out-of-date. The author(s) assume no
16 .\" responsibility for errors or omissions, or for damages resulting from
17 .\" the use of the information contained herein. The author(s) may not
18 .\" have taken the same level of care in the production of this manual,
19 .\" which is licensed free of charge, as they might when working
22 .\" Formatted or processed versions of this manual, if unaccompanied by
23 .\" the source, must acknowledge the copyright and authors of this work.
26 .\" Modified by Michael Haardt <michael@moria.de>
27 .\" Modified 1993-07-23 by Rik Faith <faith@cs.unc.edu>
28 .\" Modified 1996-01-13 by Arnt Gulbrandsen <agulbra@troll.no>
29 .\" Modified 1996-01-22 by aeb, following a remark by
30 .\" Tigran Aivazian <tigran@sco.com>
31 .\" Modified 1996-04-14 by aeb, following a remark by
32 .\" Robert Bihlmeyer <robbe@orcus.ping.at>
33 .\" Modified 1996-10-22 by Eric S. Raymond <esr@thyrsus.com>
34 .\" Modified 2001-05-04 by aeb, following a remark by
35 .\" HÃ¥vard Lygre <hklygre@online.no>
36 .\" Modified 2001-04-17 by Michael Kerrisk <mtk.manpages@gmail.com>
37 .\" Modified 2002-06-13 by Michael Kerrisk <mtk.manpages@gmail.com>
38 .\" Added note on nonstandard behavior when SIGCHLD is ignored.
39 .\" Modified 2002-07-09 by Michael Kerrisk <mtk.manpages@gmail.com>
40 .\" Enhanced descriptions of 'resource' values
41 .\" Modified 2003-11-28 by aeb, added RLIMIT_CORE
42 .\" Modified 2004-03-26 by aeb, added RLIMIT_AS
43 .\" Modified 2004-06-16 by Michael Kerrisk <mtk.manpages@gmail.com>
44 .\" Added notes on CAP_SYS_RESOURCE
46 .\" 2004-11-16 -- mtk: the getrlimit.2 page, which formally included
47 .\" coverage of getrusage(2), has been split, so that the latter
48 .\" is now covered in its own getrusage.2.
50 .\" Modified 2004-11-16, mtk: A few other minor changes
51 .\" Modified 2004-11-23, mtk
52 .\" Added notes on RLIMIT_MEMLOCK, RLIMIT_NPROC, and RLIMIT_RSS
53 .\" to "CONFORMING TO"
54 .\" Modified 2004-11-25, mtk
55 .\" Rewrote discussion on RLIMIT_MEMLOCK to incorporate kernel
57 .\" Added note on RLIMIT_CPU error in older kernels
58 .\" 2004-11-03, mtk, Added RLIMIT_SIGPENDING
59 .\" 2005-07-13, mtk, documented RLIMIT_MSGQUEUE limit.
60 .\" 2005-07-28, mtk, Added descriptions of RLIMIT_NICE and RLIMIT_RTPRIO
61 .\" 2008-05-07, mtk / Peter Zijlstra, Added description of RLIMIT_RTTIME
62 .\" 2010-11-06, mtk: Added documentation of prlimit()
64 .TH GETRLIMIT 2 2017-03-13 "Linux" "Linux Programmer's Manual"
66 getrlimit, setrlimit, prlimit \- get/set resource limits
68 .B #include <sys/time.h>
70 .B #include <sys/resource.h>
72 .BI "int getrlimit(int " resource ", struct rlimit *" rlim );
74 .BI "int setrlimit(int " resource ", const struct rlimit *" rlim );
76 .BI "int prlimit(pid_t " pid ", int " resource \
77 ", const struct rlimit *" new_limit ,
79 .BI " struct rlimit *" old_limit );
82 Feature Test Macro Requirements for glibc (see
83 .BR feature_test_macros (7)):
93 system calls get and set resource limits respectively.
94 Each resource has an associated soft and hard limit, as defined by the
101 rlim_t rlim_cur; /* Soft limit */
102 rlim_t rlim_max; /* Hard limit (ceiling for rlim_cur) */
107 The soft limit is the value that the kernel enforces for the
108 corresponding resource.
109 The hard limit acts as a ceiling for the soft limit:
110 an unprivileged process may set only its soft limit to a value in the
111 range from 0 up to the hard limit, and (irreversibly) lower its hard limit.
112 A privileged process (under Linux: one with the
114 capability) may make arbitrary changes to either limit value.
118 denotes no limit on a resource (both in the structure returned by
120 and in the structure passed to
125 argument must be one of:
128 This is the maximum size of the process's virtual memory
129 (address space) in bytes.
130 .\" since 2.0.27 / 2.1.12
131 This limit affects calls to
136 which fail with the error
138 upon exceeding this limit.
139 Also automatic stack expansion will fail
142 that kills the process if no alternate stack
143 has been made available via
144 .BR sigaltstack (2)).
145 Since the value is a \fIlong\fP, on machines with a 32-bit \fIlong\fP
146 either this limit is at most 2\ GiB, or this resource is unlimited.
149 This is the maximum size of a
153 that the process may dump.
154 When 0 no core dump files are created.
155 When nonzero, larger dumps are truncated to this size.
158 This is a limit, in seconds,
159 on the amount of CPU time that the process can consume.
160 When the process reaches the soft limit, it is sent a
163 The default action for this signal is to terminate the process.
164 However, the signal can be caught, and the handler can return control to
166 If the process continues to consume CPU time, it will be sent
168 once per second until the hard limit is reached, at which time
171 (This latter point describes Linux behavior.
172 Implementations vary in how they treat processes which continue to
173 consume CPU time after reaching the soft limit.
174 Portable applications that need to catch this signal should
175 perform an orderly termination upon first receipt of
179 This is the maximum size of the process's data segment (initialized data,
180 uninitialized data, and heap).
181 This limit affects calls to
185 which fail with the error
187 upon encountering the soft limit of this resource.
190 This is the maximum size of files that the process may create.
191 Attempts to extend a file beyond this limit result in delivery of a
194 By default, this signal terminates a process, but a process can
195 catch this signal instead, in which case the relevant system call (e.g.,
201 .BR RLIMIT_LOCKS " (early Linux 2.4 only)"
202 .\" to be precise: Linux 2.4.0-test9; no longer in 2.4.25 / 2.5.65
203 This is a limit on the combined number of
207 leases that this process may establish.
210 This is the maximum number of bytes of memory that may be locked
212 This limit is in effect rounded down to the nearest multiple
213 of the system page size.
221 Since Linux 2.6.9, it also affects the
224 operation, where it sets a maximum on the total bytes in
225 shared memory segments (see
227 that may be locked by the real user ID of the calling process.
231 locks are accounted for separately from the per-process memory
238 a process can lock bytes up to this limit in each of these
241 In Linux kernels before 2.6.9, this limit controlled the amount of
242 memory that could be locked by a privileged process.
243 Since Linux 2.6.9, no limits are placed on the amount of memory
244 that a privileged process may lock, and this limit instead governs
245 the amount of memory that an unprivileged process may lock.
247 .BR RLIMIT_MSGQUEUE " (since Linux 2.6.8)"
248 This is a limit on the number of bytes that can be allocated
249 for POSIX message queues for the real user ID of the calling process.
250 This limit is enforced for
252 Each message queue that the user creates counts (until it is removed)
253 against this limit according to the formula:
258 bytes = attr.mq_maxmsg * sizeof(struct msg_msg) +
259 min(attr.mq_maxmsg, MQ_PRIO_MAX) *
260 sizeof(struct posix_msg_tree_node)+
262 attr.mq_maxmsg * attr.mq_msgsize;
263 /* For message data */
265 Linux 3.4 and earlier:
267 bytes = attr.mq_maxmsg * sizeof(struct msg_msg *) +
269 attr.mq_maxmsg * attr.mq_msgsize;
270 /* For message data */
277 structure specified as the fourth argument to
282 .I posix_msg_tree_node
283 structures are kernel-internal structures.
285 The "overhead" addend in the formula accounts for overhead
286 bytes required by the implementation
287 and ensures that the user cannot
288 create an unlimited number of zero-length messages (such messages
289 nevertheless each consume some system memory for bookkeeping overhead).
291 .BR RLIMIT_NICE " (since Linux 2.6.12, but see BUGS below)"
292 This specifies a ceiling to which the process's nice value can be raised using
296 The actual ceiling for the nice value is calculated as
297 .IR "20\ \-\ rlim_cur" .
298 The useful range for this limit is thus from 1
299 (corresponding to a nice value of 19) to 40
300 (corresponding to a nice value of -20).
301 This unusual choice of range was necessary
302 because negative numbers cannot be specified
303 as resource limit values, since they typically have special meanings.
306 typically is the same as \-1.
307 For more detail on the nice value, see
311 This specifies a value one greater than the maximum file descriptor number
312 that can be opened by this process.
318 to exceed this limit yield the error
320 (Historically, this limit was named
325 this limit also defines the maximum number of file descriptors that
326 an unprivileged process (one without the
328 capability) may have "in flight" to other processes,
329 by being passed across UNIX domain sockets.
330 This limit applies to the
333 For further details, see
337 This is the maximum number of processes
338 (or, more precisely on Linux, threads)
339 that can be created for the real user ID of the calling process.
340 Upon encountering this limit,
344 This limit is not enforced for processes that have either the
351 This is a limit (in bytes) on the process's resident set
352 (the number of virtual pages resident in RAM).
353 This limit has effect only in Linux 2.4.x, x < 30, and there
354 affects only calls to
358 .\" As at kernel 2.6.12, this limit still does nothing in 2.6 though
359 .\" talk of making it do something has surfaced from time to time in LKML
362 .BR RLIMIT_RTPRIO " (since Linux 2.6.12, but see BUGS)"
363 This specifies a ceiling on the real-time priority that may be set for
365 .BR sched_setscheduler (2)
367 .BR sched_setparam (2).
369 For further details on real-time scheduling policies, see
372 .BR RLIMIT_RTTIME " (since Linux 2.6.25)"
373 This is a limit (in microseconds)
374 on the amount of CPU time that a process scheduled
375 under a real-time scheduling policy may consume without making a blocking
377 For the purpose of this limit,
378 each time a process makes a blocking system call,
379 the count of its consumed CPU time is reset to zero.
380 The CPU time count is not reset if the process continues trying to
381 use the CPU but is preempted, its time slice expires, or it calls
384 Upon reaching the soft limit, the process is sent a
387 If the process catches or ignores this signal and
388 continues consuming CPU time, then
390 will be generated once each second until the hard limit is reached,
391 at which point the process is sent a
395 The intended use of this limit is to stop a runaway
396 real-time process from locking up the system.
398 For further details on real-time scheduling policies, see
401 .BR RLIMIT_SIGPENDING " (since Linux 2.6.8)"
402 This is a limit on the number of signals
403 that may be queued for the real user ID of the calling process.
404 Both standard and real-time signals are counted for the purpose of
406 However, the limit is enforced only for
408 it is always possible to use
410 to queue one instance of any of the signals that are not already
411 queued to the process.
412 .\" This replaces the /proc/sys/kernel/rtsig-max system-wide limit
413 .\" that was present in kernels <= 2.6.7. MTK Dec 04
416 This is the maximum size of the process stack, in bytes.
417 Upon reaching this limit, a
420 To handle this signal, a process must employ an alternate signal stack
421 .RB ( sigaltstack (2)).
424 this limit also determines the amount of space used for the process's
425 command-line arguments and environment variables; for details, see
428 .\" commit c022a0acad534fd5f5d5f17280f6d4d135e74e81
429 .\" Author: Jiri Slaby <jslaby@suse.cz>
430 .\" Date: Tue May 4 18:03:50 2010 +0200
432 .\" rlimits: implement prlimit64 syscall
434 .\" commit 6a1d5e2c85d06da35cdfd93f1a27675bfdc3ad8c
435 .\" Author: Jiri Slaby <jslaby@suse.cz>
436 .\" Date: Wed Mar 24 17:06:58 2010 +0100
438 .\" rlimits: add rlimit64 structure
442 system call combines and extends the functionality of
446 It can be used to both set and get the resource limits of an arbitrary process.
450 argument has the same meaning as for
457 argument is a not NULL, then the
459 structure to which it points is used to set new values for
460 the soft and hard limits for
464 argument is a not NULL, then a successful call to
466 places the previous soft and hard limits for
470 structure pointed to by
475 argument specifies the ID of the process on which the call is to operate.
478 is 0, then the call applies to the calling process.
479 To set or get the resources of a process other than itself,
480 the caller must have the
482 capability in the user namespace of the process
483 whose resource limits are being changed, or the
484 real, effective, and saved set user IDs of the target process
485 must match the real user ID of the caller
487 the real, effective, and saved set group IDs of the target process
488 must match the real group ID of the caller.
489 .\" FIXME . this permission check is strange
490 .\" Asked about this on LKML, 7 Nov 2010
491 .\" "Inconsistent credential checking in prlimit() syscall"
493 On success, these system calls return 0.
494 On error, \-1 is returned, and
496 is set appropriately.
500 A pointer argument points to a location
501 outside the accessible address space.
504 The value specified in
513 .IR rlim\->rlim_max .
516 An unprivileged process tried to raise the hard limit; the
518 capability is required to do this.
521 The caller tried to increase the hard
523 limit above the maximum defined by
524 .IR /proc/sys/fs/nr_open
530 The calling process did not have permission to set limits
531 for the process specified by
535 Could not find a process with the ID specified in
540 system call is available since Linux 2.6.36.
541 Library support is available since glibc 2.13.
543 For an explanation of the terms used in this section, see
549 Interface Attribute Value
554 T} Thread safety MT-Safe
560 POSIX.1-2001, POSIX.1-2008, SVr4, 4.3BSD.
568 derive from BSD and are not specified in POSIX.1;
569 they are present on the BSDs and Linux, but on few other implementations.
571 derives from BSD and is not specified in POSIX.1;
572 it is nevertheless present on most implementations.
573 .BR RLIMIT_MSGQUEUE ,
581 A child process created via
583 inherits its parent's resource limits.
584 Resource limits are preserved across
587 Lowering the soft limit for a resource below the process's
588 current consumption of that resource will succeed
589 (but will prevent the process from further increasing
590 its consumption of the resource).
592 One can set the resource limits of the shell using the built-in
598 The shell's resource limits are inherited by the processes that
599 it creates to execute commands.
601 Since Linux 2.6.24, the resource limits of any process can be inspected via
602 .IR /proc/[pid]/limits ;
606 Ancient systems provided a
608 function with a similar purpose to
610 For backward compatibility, glibc also provides
612 All new applications should be written using
614 .SS C library/ kernel ABI differences
615 Since version 2.13, the glibc
619 wrapper functions no longer invoke the corresponding system calls,
622 for the reasons described in BUGS.
624 The name of the glibc wrapper function is
626 the underlying system call is
629 In older Linux kernels, the
633 signals delivered when a process encountered the soft and hard
635 limits were delivered one (CPU) second later than they should have been.
636 This was fixed in kernel 2.6.8.
638 In 2.6.x kernels before 2.6.17, a
640 limit of 0 is wrongly treated as "no limit" (like
642 Since Linux 2.6.17, setting a limit of 0 does have an effect,
643 but is actually treated as a limit of 1 second.
644 .\" see http://marc.theaimsgroup.com/?l=linux-kernel&m=114008066530167&w=2
646 A kernel bug means that
647 .\" See https://lwn.net/Articles/145008/
649 does not work in kernel 2.6.12; the problem is fixed in kernel 2.6.13.
651 In kernel 2.6.12, there was an off-by-one mismatch
652 between the priority ranges returned by
656 This had the effect that the actual ceiling for the nice value
658 .IR "19\ \-\ rlim_cur" .
659 This was fixed in kernel 2.6.13.
660 .\" see http://marc.theaimsgroup.com/?l=linux-kernel&m=112256338703880&w=2
663 .\" The relevant patch, sent to LKML, seems to be
664 .\" http://thread.gmane.org/gmane.linux.kernel/273462
665 .\" From: Roland McGrath <roland <at> redhat.com>
666 .\" Subject: [PATCH 7/7] make RLIMIT_CPU/SIGXCPU per-process
667 .\" Date: 2005-01-23 23:27:46 GMT
668 if a process reaches its soft
670 limit and has a handler installed for
672 then, in addition to invoking the signal handler,
673 the kernel increases the soft limit by one second.
674 This behavior repeats if the process continues to consume CPU time,
675 until the hard limit is reached,
676 at which point the process is killed.
677 Other implementations
678 .\" Tested Solaris 10, FreeBSD 9, OpenBSD 5.0
681 soft limit in this manner,
682 and the Linux behavior is probably not standards conformant;
683 portable applications should avoid relying on this Linux-specific behavior.
684 .\" FIXME . https://bugzilla.kernel.org/show_bug.cgi?id=50951
687 limit exhibits the same behavior when the soft limit is encountered.
689 Kernels before 2.4.22 did not diagnose the error
696 .IR rlim\->rlim_max .
698 .SS Representation of """large""" resource limit values on 32-bit platforms
703 wrapper functions use a 64-bit
705 data type, even on 32-bit platforms.
708 data type used in the
712 system calls is a (32-bit)
713 .IR "unsigned long" .
714 Furthermore, in Linux versions before 2.6.36,
715 the kernel represents resource limits on 32-bit platforms as
716 .IR "unsigned long" .
717 However, a 32-bit data type is not wide enough.
718 .\" https://bugzilla.kernel.org/show_bug.cgi?id=5042
719 .\" http://sources.redhat.com/bugzilla/show_bug.cgi?id=12201
720 The most pertinent limit here is
722 which specifies the maximum size to which a file can grow:
723 to be useful, this limit must be represented using a type
724 that is as wide as the type used to
725 represent file offsets\(emthat is, as wide as a 64-bit
727 (assuming a program compiled with
728 .IR _FILE_OFFSET_BITS=64 ).
730 To work around this kernel limitation,
731 if a program tried to set a resource limit to a value larger than
732 can be represented in a 32-bit
733 .IR "unsigned long" ,
736 wrapper function silently converted the limit value to
738 In other words, the requested resource limit setting was silently ignored.
740 This problem was addressed in Linux 2.6.36 with two principal changes:
742 the addition of a new kernel representation of resource limits that
743 uses 64 bits, even on 32-bit platforms;
747 system call, which employs 64-bit values for its resource limit arguments.
750 .\" https://www.sourceware.org/bugzilla/show_bug.cgi?id=12201
751 glibc works around the limitations of the
755 system calls by implementing
759 as wrapper functions that call
762 The program below demonstrates the use of
767 #define _FILE_OFFSET_BITS 64
772 #include <sys/resource.h>
774 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \\
778 main(int argc, char *argv[])
780 struct rlimit old, new;
784 if (!(argc == 2 || argc == 4)) {
785 fprintf(stderr, "Usage: %s <pid> [<new\-soft\-limit> "
786 "<new\-hard\-limit>]\\n", argv[0]);
790 pid = atoi(argv[1]); /* PID of target process */
794 new.rlim_cur = atoi(argv[2]);
795 new.rlim_max = atoi(argv[3]);
799 /* Set CPU time limit of target process; retrieve and display
802 if (prlimit(pid, RLIMIT_CPU, newp, &old) == \-1)
803 errExit("prlimit\-1");
804 printf("Previous limits: soft=%lld; hard=%lld\\n",
805 (long long) old.rlim_cur, (long long) old.rlim_max);
807 /* Retrieve and display new CPU time limit */
809 if (prlimit(pid, RLIMIT_CPU, NULL, &old) == \-1)
810 errExit("prlimit\-2");
811 printf("New limits: soft=%lld; hard=%lld\\n",
812 (long long) old.rlim_cur, (long long) old.rlim_max);
833 .BR capabilities (7),