1SHMGET(2) Linux Programmer's Manual SHMGET(2)
2
6 shmget - allocates a System V shared memory segment
7
9 #include <sys/ipc.h>
10 #include <sys/shm.h>
11 int shmget(key_t key, size_t size, int shmflg);
13
15 shmget() returns the identifier of the System V shared memory segment
16 associated with the value of the argument key. It may be used either
17 to obtain the identifier of a previously created shared memory segment
18 (when shmflg is zero and key does not have the value IPC_PRIVATE), or
19 to create a new set.
20 A new shared memory segment, with size equal to the value of size
22 rounded up to a multiple of PAGE_SIZE, is created if key has the value
23 IPC_PRIVATE or key isn't IPC_PRIVATE, no shared memory segment corre‐
24 sponding to key exists, and IPC_CREAT is specified in shmflg.
25 If shmflg specifies both IPC_CREAT and IPC_EXCL and a shared memory
27 segment already exists for key, then shmget() fails with errno set to
28 EEXIST. (This is analogous to the effect of the combination O_CREAT |
29 O_EXCL for open(2).)
30 The value shmflg is composed of:
32 IPC_CREAT
34 Create a new segment. If this flag is not used, then shmget()
35 will find the segment associated with key and check to see if
36 the user has permission to access the segment.
37 IPC_EXCL
39 This flag is used with IPC_CREAT to ensure that this call cre‐
40 ates the segment. If the segment already exists, the call
41 fails.
42 SHM_HUGETLB (since Linux 2.6)
44 Allocate the segment using "huge pages." See the Linux kernel
45 source file Documentation/admin-guide/mm/hugetlbpage.rst for
46 further information.
47 SHM_HUGE_2MB, SHM_HUGE_1GB (since Linux 3.8)
49 Used in conjunction with SHM_HUGETLB to select alternative
50 hugetlb page sizes (respectively, 2 MB and 1 GB) on systems that
51 support multiple hugetlb page sizes.
52 More generally, the desired huge page size can be configured by
54 encoding the base-2 logarithm of the desired page size in the
55 six bits at the offset SHM_HUGE_SHIFT. Thus, the above two con‐
56 stants are defined as:
57 #define SHM_HUGE_2MB (21 << SHM_HUGE_SHIFT)
59 #define SHM_HUGE_1GB (30 << SHM_HUGE_SHIFT)
60 For some additional details, see the discussion of the similarly
62 named constants in mmap(2).
63 SHM_NORESERVE (since Linux 2.6.15)
65 This flag serves the same purpose as the mmap(2) MAP_NORESERVE
66 flag. Do not reserve swap space for this segment. When swap
67 space is reserved, one has the guarantee that it is possible to
68 modify the segment. When swap space is not reserved one might
69 get SIGSEGV upon a write if no physical memory is available.
70 See also the discussion of the file /proc/sys/vm/overcommit_mem‐
71 ory in proc(5).
72 In addition to the above flags, the least significant 9 bits of shmflg
74 specify the permissions granted to the owner, group, and others. These
75 bits have the same format, and the same meaning, as the mode argument
76 of open(2). Presently, execute permissions are not used by the system.
77 When a new shared memory segment is created, its contents are initial‐
79 ized to zero values, and its associated data structure, shmid_ds (see
80 shmctl(2)), is initialized as follows:
81 • shm_perm.cuid and shm_perm.uid are set to the effective user ID of
83 the calling process.
84 • shm_perm.cgid and shm_perm.gid are set to the effective group ID of
86 the calling process.
87 • The least significant 9 bits of shm_perm.mode are set to the least
89 significant 9 bit of shmflg.
90 • shm_segsz is set to the value of size.
92 • shm_lpid, shm_nattch, shm_atime, and shm_dtime are set to 0.
94 • shm_ctime is set to the current time.
96 If the shared memory segment already exists, the permissions are veri‐
98 fied, and a check is made to see if it is marked for destruction.
99
101 On success, a valid shared memory identifier is returned. On error, -1
102 is returned, and errno is set to indicate the error.
103
105 On failure, errno is set to one of the following:
106 EACCES The user does not have permission to access the shared memory
108 segment, and does not have the CAP_IPC_OWNER capability in the
109 user namespace that governs its IPC namespace.
110 EEXIST IPC_CREAT and IPC_EXCL were specified in shmflg, but a shared
112 memory segment already exists for key.
113 EINVAL A new segment was to be created and size is less than SHMMIN or
115 greater than SHMMAX.
116 EINVAL A segment for the given key exists, but size is greater than the
118 size of that segment.
119 ENFILE The system-wide limit on the total number of open files has been
121 reached.
122 ENOENT No segment exists for the given key, and IPC_CREAT was not spec‐
124 ified.
125 ENOMEM No memory could be allocated for segment overhead.
127 ENOSPC All possible shared memory IDs have been taken (SHMMNI), or al‐
129 locating a segment of the requested size would cause the system
130 to exceed the system-wide limit on shared memory (SHMALL).
131 EPERM The SHM_HUGETLB flag was specified, but the caller was not priv‐
133 ileged (did not have the CAP_IPC_LOCK capability).
134
136 POSIX.1-2001, POSIX.1-2008, SVr4.
137 SHM_HUGETLB and SHM_NORESERVE are Linux extensions.
139
141 The inclusion of <sys/types.h> and <sys/ipc.h> isn't required on Linux
142 or by any version of POSIX. However, some old implementations required
143 the inclusion of these header files, and the SVID also documented their
144 inclusion. Applications intended to be portable to such old systems
145 may need to include these header files.
146 IPC_PRIVATE isn't a flag field but a key_t type. If this special value
148 is used for key, the system call ignores all but the least significant
149 9 bits of shmflg and creates a new shared memory segment.
150 Shared memory limits
152 The following limits on shared memory segment resources affect the
153 shmget() call:
154 SHMALL System-wide limit on the total amount of shared memory, measured
156 in units of the system page size.
157 On Linux, this limit can be read and modified via /proc/sys/ker‐
159 nel/shmall. Since Linux 3.16, the default value for this limit
160 is:
161 ULONG_MAX - 2^24
163 The effect of this value (which is suitable for both 32-bit and
165 64-bit systems) is to impose no limitation on allocations. This
166 value, rather than ULONG_MAX, was chosen as the default to pre‐
167 vent some cases where historical applications simply raised the
168 existing limit without first checking its current value. Such
169 applications would cause the value to overflow if the limit was
170 set at ULONG_MAX.
171 From Linux 2.4 up to Linux 3.15, the default value for this
173 limit was:
174 SHMMAX / PAGE_SIZE * (SHMMNI / 16)
176 If SHMMAX and SHMMNI were not modified, then multiplying the re‐
178 sult of this formula by the page size (to get a value in bytes)
179 yielded a value of 8 GB as the limit on the total memory used by
180 all shared memory segments.
181 SHMMAX Maximum size in bytes for a shared memory segment.
183 On Linux, this limit can be read and modified via /proc/sys/ker‐
185 nel/shmmax. Since Linux 3.16, the default value for this limit
186 is:
187 ULONG_MAX - 2^24
189 The effect of this value (which is suitable for both 32-bit and
191 64-bit systems) is to impose no limitation on allocations. See
192 the description of SHMALL for a discussion of why this default
193 value (rather than ULONG_MAX) is used.
194 From Linux 2.2 up to Linux 3.15, the default value of this limit
196 was 0x2000000 (32 MB).
197 Because it is not possible to map just part of a shared memory
199 segment, the amount of virtual memory places another limit on
200 the maximum size of a usable segment: for example, on i386 the
201 largest segments that can be mapped have a size of around
202 2.8 GB, and on x86-64 the limit is around 127 TB.
203 SHMMIN Minimum size in bytes for a shared memory segment: implementa‐
205 tion dependent (currently 1 byte, though PAGE_SIZE is the effec‐
206 tive minimum size).
207 SHMMNI System-wide limit on the number of shared memory segments. In
209 Linux 2.2, the default value for this limit was 128; since Linux
210 2.4, the default value is 4096.
211 On Linux, this limit can be read and modified via /proc/sys/ker‐
213 nel/shmmni.
214 The implementation has no specific limits for the per-process maximum
216 number of shared memory segments (SHMSEG).
217 Linux notes
219 Until version 2.3.30, Linux would return EIDRM for a shmget() on a
220 shared memory segment scheduled for deletion.
221
223 The name choice IPC_PRIVATE was perhaps unfortunate, IPC_NEW would more
224 clearly show its function.
225
227 See shmop(2).
228
230 memfd_create(2), shmat(2), shmctl(2), shmdt(2), ftok(3), capabili‐
231 ties(7), shm_overview(7), sysvipc(7)
232
234 This page is part of release 5.10 of the Linux man-pages project. A
235 description of the project, information about reporting bugs, and the
236 latest version of this page, can be found at
237 https://www.kernel.org/doc/man-pages/.
238Linux 2020-04-11 SHMGET(2)