1MLOCK(2) Linux Programmer's Manual MLOCK(2)
2
6 mlock, mlock2, munlock, mlockall, munlockall - lock and unlock memory
7
9 #include <sys/mman.h>
10 int mlock(const void *addr, size_t len);
12 int mlock2(const void *addr, size_t len, int flags);
13 int munlock(const void *addr, size_t len);
14 int mlockall(int flags);
16 int munlockall(void);
17
19 mlock(), mlock2(), and mlockall() lock part or all of the calling
20 process's virtual address space into RAM, preventing that memory from
21 being paged to the swap area.
22 munlock() and munlockall() perform the converse operation, unlocking
24 part or all of the calling process's virtual address space, so that
25 pages in the specified virtual address range may once more to be
26 swapped out if required by the kernel memory manager.
27 Memory locking and unlocking are performed in units of whole pages.
29 mlock(), mlock2(), and munlock()
31 mlock() locks pages in the address range starting at addr and continu‐
32 ing for len bytes. All pages that contain a part of the specified
33 address range are guaranteed to be resident in RAM when the call
34 returns successfully; the pages are guaranteed to stay in RAM until
35 later unlocked.
36 mlock2() also locks pages in the specified range starting at addr and
38 continuing for len bytes. However, the state of the pages contained in
39 that range after the call returns successfully will depend on the value
40 in the flags argument.
41 The flags argument can be either 0 or the following constant:
43 MLOCK_ONFAULT
45 Lock pages that are currently resident and mark the entire range
46 so that the remaining nonresident pages are locked when they are
47 populated by a page fault.
48 If flags is 0, mlock2() behaves exactly the same as mlock().
50 munlock() unlocks pages in the address range starting at addr and con‐
52 tinuing for len bytes. After this call, all pages that contain a part
53 of the specified memory range can be moved to external swap space again
54 by the kernel.
55 mlockall() and munlockall()
57 mlockall() locks all pages mapped into the address space of the calling
58 process. This includes the pages of the code, data and stack segment,
59 as well as shared libraries, user space kernel data, shared memory, and
60 memory-mapped files. All mapped pages are guaranteed to be resident in
61 RAM when the call returns successfully; the pages are guaranteed to
62 stay in RAM until later unlocked.
63 The flags argument is constructed as the bitwise OR of one or more of
65 the following constants:
66 MCL_CURRENT Lock all pages which are currently mapped into the address
68 space of the process.
69 MCL_FUTURE Lock all pages which will become mapped into the address
71 space of the process in the future. These could be, for
72 instance, new pages required by a growing heap and stack as
73 well as new memory-mapped files or shared memory regions.
74 MCL_ONFAULT (since Linux 4.4)
76 Used together with MCL_CURRENT, MCL_FUTURE, or both. Mark
77 all current (with MCL_CURRENT) or future (with MCL_FUTURE)
78 mappings to lock pages when they are faulted in. When used
79 with MCL_CURRENT, all present pages are locked, but mlock‐
80 all() will not fault in non-present pages. When used with
81 MCL_FUTURE, all future mappings will be marked to lock
82 pages when they are faulted in, but they will not be popu‐
83 lated by the lock when the mapping is created. MCL_ONFAULT
84 must be used with either MCL_CURRENT or MCL_FUTURE or both.
85 If MCL_FUTURE has been specified, then a later system call (e.g.,
87 mmap(2), sbrk(2), malloc(3)), may fail if it would cause the number of
88 locked bytes to exceed the permitted maximum (see below). In the same
89 circumstances, stack growth may likewise fail: the kernel will deny
90 stack expansion and deliver a SIGSEGV signal to the process.
91 munlockall() unlocks all pages mapped into the address space of the
93 calling process.
94
96 On success, these system calls return 0. On error, -1 is returned,
97 errno is set appropriately, and no changes are made to any locks in the
98 address space of the process.
99
101 ENOMEM (Linux 2.6.9 and later) the caller had a nonzero RLIMIT_MEMLOCK
102 soft resource limit, but tried to lock more memory than the
103 limit permitted. This limit is not enforced if the process is
104 privileged (CAP_IPC_LOCK).
105 ENOMEM (Linux 2.4 and earlier) the calling process tried to lock more
107 than half of RAM.
108 EPERM The caller is not privileged, but needs privilege (CAP_IPC_LOCK)
110 to perform the requested operation.
111 For mlock(), mlock2(), and munlock():
113 EAGAIN Some or all of the specified address range could not be locked.
115 EINVAL The result of the addition addr+len was less than addr (e.g.,
117 the addition may have resulted in an overflow).
118 EINVAL (Not on Linux) addr was not a multiple of the page size.
120 ENOMEM Some of the specified address range does not correspond to
122 mapped pages in the address space of the process.
123 ENOMEM Locking or unlocking a region would result in the total number
125 of mappings with distinct attributes (e.g., locked versus
126 unlocked) exceeding the allowed maximum. (For example, unlock‐
127 ing a range in the middle of a currently locked mapping would
128 result in three mappings: two locked mappings at each end and an
129 unlocked mapping in the middle.)
130 For mlock2():
132 EINVAL Unknown flags were specified.
134 For mlockall():
136 EINVAL Unknown flags were specified or MCL_ONFAULT was specified with‐
138 out either MCL_FUTURE or MCL_CURRENT.
139 For munlockall():
141 EPERM (Linux 2.6.8 and earlier) The caller was not privileged
143 (CAP_IPC_LOCK).
144
146 mlock2() is available since Linux 4.4; glibc support was added in ver‐
147 sion 2.27.
148
150 POSIX.1-2001, POSIX.1-2008, SVr4.
151 mlock2 () is Linux specific.
153
155 On POSIX systems on which mlock() and munlock() are available,
156 _POSIX_MEMLOCK_RANGE is defined in <unistd.h> and the number of bytes
157 in a page can be determined from the constant PAGESIZE (if defined) in
158 <limits.h> or by calling sysconf(_SC_PAGESIZE).
159 On POSIX systems on which mlockall() and munlockall() are available,
161 _POSIX_MEMLOCK is defined in <unistd.h> to a value greater than 0.
162 (See also sysconf(3).)
163
165 Memory locking has two main applications: real-time algorithms and
166 high-security data processing. Real-time applications require deter‐
167 ministic timing, and, like scheduling, paging is one major cause of
168 unexpected program execution delays. Real-time applications will usu‐
169 ally also switch to a real-time scheduler with sched_setscheduler(2).
170 Cryptographic security software often handles critical bytes like pass‐
171 words or secret keys as data structures. As a result of paging, these
172 secrets could be transferred onto a persistent swap store medium, where
173 they might be accessible to the enemy long after the security software
174 has erased the secrets in RAM and terminated. (But be aware that the
175 suspend mode on laptops and some desktop computers will save a copy of
176 the system's RAM to disk, regardless of memory locks.)
177 Real-time processes that are using mlockall() to prevent delays on page
179 faults should reserve enough locked stack pages before entering the
180 time-critical section, so that no page fault can be caused by function
181 calls. This can be achieved by calling a function that allocates a
182 sufficiently large automatic variable (an array) and writes to the mem‐
183 ory occupied by this array in order to touch these stack pages. This
184 way, enough pages will be mapped for the stack and can be locked into
185 RAM. The dummy writes ensure that not even copy-on-write page faults
186 can occur in the critical section.
187 Memory locks are not inherited by a child created via fork(2) and are
189 automatically removed (unlocked) during an execve(2) or when the
190 process terminates. The mlockall() MCL_FUTURE and MCL_FUTURE |
191 MCL_ONFAULT settings are not inherited by a child created via fork(2)
192 and are cleared during an execve(2).
193 Note that fork(2) will prepare the address space for a copy-on-write
195 operation. The consequence is that any write access that follows will
196 cause a page fault that in turn may cause high latencies for a real-
197 time process. Therefore, it is crucial not to invoke fork(2) after an
198 mlockall() or mlock() operation—not even from a thread which runs at a
199 low priority within a process which also has a thread running at ele‐
200 vated priority.
201 The memory lock on an address range is automatically removed if the
203 address range is unmapped via munmap(2).
204 Memory locks do not stack, that is, pages which have been locked sev‐
206 eral times by calls to mlock(), mlock2(), or mlockall() will be
207 unlocked by a single call to munlock() for the corresponding range or
208 by munlockall(). Pages which are mapped to several locations or by
209 several processes stay locked into RAM as long as they are locked at
210 least at one location or by at least one process.
211 If a call to mlockall() which uses the MCL_FUTURE flag is followed by
213 another call that does not specify this flag, the changes made by the
214 MCL_FUTURE call will be lost.
215 The mlock2() MLOCK_ONFAULT flag and the mlockall() MCL_ONFAULT flag
217 allow efficient memory locking for applications that deal with large
218 mappings where only a (small) portion of pages in the mapping are
219 touched. In such cases, locking all of the pages in a mapping would
220 incur a significant penalty for memory locking.
221 Linux notes
223 Under Linux, mlock(), mlock2(), and munlock() automatically round addr
224 down to the nearest page boundary. However, the POSIX.1 specification
225 of mlock() and munlock() allows an implementation to require that addr
226 is page aligned, so portable applications should ensure this.
227 The VmLck field of the Linux-specific /proc/[pid]/status file shows how
229 many kilobytes of memory the process with ID PID has locked using
230 mlock(), mlock2(), mlockall(), and mmap(2) MAP_LOCKED.
231 Limits and permissions
233 In Linux 2.6.8 and earlier, a process must be privileged (CAP_IPC_LOCK)
234 in order to lock memory and the RLIMIT_MEMLOCK soft resource limit
235 defines a limit on how much memory the process may lock.
236 Since Linux 2.6.9, no limits are placed on the amount of memory that a
238 privileged process can lock and the RLIMIT_MEMLOCK soft resource limit
239 instead defines a limit on how much memory an unprivileged process may
240 lock.
241
243 In Linux 4.8 and earlier, a bug in the kernel's accounting of locked
244 memory for unprivileged processes (i.e., without CAP_IPC_LOCK) meant
245 that if the region specified by addr and len overlapped an existing
246 lock, then the already locked bytes in the overlapping region were
247 counted twice when checking against the limit. Such double accounting
248 could incorrectly calculate a "total locked memory" value for the
249 process that exceeded the RLIMIT_MEMLOCK limit, with the result that
250 mlock() and mlock2() would fail on requests that should have succeeded.
251 This bug was fixed in Linux 4.9
252 In the 2.4 series Linux kernels up to and including 2.4.17, a bug
254 caused the mlockall() MCL_FUTURE flag to be inherited across a fork(2).
255 This was rectified in kernel 2.4.18.
256 Since kernel 2.6.9, if a privileged process calls mlockall(MCL_FUTURE)
258 and later drops privileges (loses the CAP_IPC_LOCK capability by, for
259 example, setting its effective UID to a nonzero value), then subsequent
260 memory allocations (e.g., mmap(2), brk(2)) will fail if the RLIMIT_MEM‐
261 LOCK resource limit is encountered.
262
264 mincore(2), mmap(2), setrlimit(2), shmctl(2), sysconf(3), proc(5),
265 capabilities(7)
266
268 This page is part of release 5.04 of the Linux man-pages project. A
269 description of the project, information about reporting bugs, and the
270 latest version of this page, can be found at
271 https://www.kernel.org/doc/man-pages/.
272Linux 2018-02-02 MLOCK(2)