1SHMGET(2) Linux Programmer's Manual SHMGET(2)
2
6 shmget - allocates a System V shared memory segment
7
9 #include <sys/shm.h>
10 int shmget(key_t key, size_t size, int shmflg);
12
14 shmget() returns the identifier of the System V shared memory segment
15 associated with the value of the argument key. It may be used either
16 to obtain the identifier of a previously created shared memory segment
17 (when shmflg is zero and key does not have the value IPC_PRIVATE), or
18 to create a new set.
19 A new shared memory segment, with size equal to the value of size
21 rounded up to a multiple of PAGE_SIZE, is created if key has the value
22 IPC_PRIVATE or key isn't IPC_PRIVATE, no shared memory segment corre‐
23 sponding to key exists, and IPC_CREAT is specified in shmflg.
24 If shmflg specifies both IPC_CREAT and IPC_EXCL and a shared memory
26 segment already exists for key, then shmget() fails with errno set to
27 EEXIST. (This is analogous to the effect of the combination O_CREAT |
28 O_EXCL for open(2).)
29 The value shmflg is composed of:
31 IPC_CREAT
33 Create a new segment. If this flag is not used, then shmget()
34 will find the segment associated with key and check to see if
35 the user has permission to access the segment.
36 IPC_EXCL
38 This flag is used with IPC_CREAT to ensure that this call cre‐
39 ates the segment. If the segment already exists, the call
40 fails.
41 SHM_HUGETLB (since Linux 2.6)
43 Allocate the segment using "huge" pages. See the Linux kernel
44 source file Documentation/admin-guide/mm/hugetlbpage.rst for
45 further information.
46 SHM_HUGE_2MB, SHM_HUGE_1GB (since Linux 3.8)
48 Used in conjunction with SHM_HUGETLB to select alternative
49 hugetlb page sizes (respectively, 2 MB and 1 GB) on systems that
50 support multiple hugetlb page sizes.
51 More generally, the desired huge page size can be configured by
53 encoding the base-2 logarithm of the desired page size in the
54 six bits at the offset SHM_HUGE_SHIFT. Thus, the above two con‐
55 stants are defined as:
56 #define SHM_HUGE_2MB (21 << SHM_HUGE_SHIFT)
58 #define SHM_HUGE_1GB (30 << SHM_HUGE_SHIFT)
59 For some additional details, see the discussion of the similarly
61 named constants in mmap(2).
62 SHM_NORESERVE (since Linux 2.6.15)
64 This flag serves the same purpose as the mmap(2) MAP_NORESERVE
65 flag. Do not reserve swap space for this segment. When swap
66 space is reserved, one has the guarantee that it is possible to
67 modify the segment. When swap space is not reserved one might
68 get SIGSEGV upon a write if no physical memory is available.
69 See also the discussion of the file /proc/sys/vm/overcommit_mem‐
70 ory in proc(5).
71 In addition to the above flags, the least significant 9 bits of shmflg
73 specify the permissions granted to the owner, group, and others. These
74 bits have the same format, and the same meaning, as the mode argument
75 of open(2). Presently, execute permissions are not used by the system.
76 When a new shared memory segment is created, its contents are initial‐
78 ized to zero values, and its associated data structure, shmid_ds (see
79 shmctl(2)), is initialized as follows:
80 • shm_perm.cuid and shm_perm.uid are set to the effective user ID of
82 the calling process.
83 • shm_perm.cgid and shm_perm.gid are set to the effective group ID of
85 the calling process.
86 • The least significant 9 bits of shm_perm.mode are set to the least
88 significant 9 bit of shmflg.
89 • shm_segsz is set to the value of size.
91 • shm_lpid, shm_nattch, shm_atime, and shm_dtime are set to 0.
93 • shm_ctime is set to the current time.
95 If the shared memory segment already exists, the permissions are veri‐
97 fied, and a check is made to see if it is marked for destruction.
98
100 On success, a valid shared memory identifier is returned. On error, -1
101 is returned, and errno is set to indicate the error.
102
104 EACCES The user does not have permission to access the shared memory
105 segment, and does not have the CAP_IPC_OWNER capability in the
106 user namespace that governs its IPC namespace.
107 EEXIST IPC_CREAT and IPC_EXCL were specified in shmflg, but a shared
109 memory segment already exists for key.
110 EINVAL A new segment was to be created and size is less than SHMMIN or
112 greater than SHMMAX.
113 EINVAL A segment for the given key exists, but size is greater than the
115 size of that segment.
116 ENFILE The system-wide limit on the total number of open files has been
118 reached.
119 ENOENT No segment exists for the given key, and IPC_CREAT was not spec‐
121 ified.
122 ENOMEM No memory could be allocated for segment overhead.
124 ENOSPC All possible shared memory IDs have been taken (SHMMNI), or al‐
126 locating a segment of the requested size would cause the system
127 to exceed the system-wide limit on shared memory (SHMALL).
128 EPERM The SHM_HUGETLB flag was specified, but the caller was not priv‐
130 ileged (did not have the CAP_IPC_LOCK capability) and is not a
131 member of the sysctl_hugetlb_shm_group group; see the descrip‐
132 tion of /proc/sys/vm/sysctl_hugetlb_shm_group in proc(5).
133
135 POSIX.1-2001, POSIX.1-2008, SVr4.
136 SHM_HUGETLB and SHM_NORESERVE are Linux extensions.
138
140 IPC_PRIVATE isn't a flag field but a key_t type. If this special value
141 is used for key, the system call ignores all but the least significant
142 9 bits of shmflg and creates a new shared memory segment.
143 Shared memory limits
145 The following limits on shared memory segment resources affect the
146 shmget() call:
147 SHMALL System-wide limit on the total amount of shared memory, measured
149 in units of the system page size.
150 On Linux, this limit can be read and modified via /proc/sys/ker‐
152 nel/shmall. Since Linux 3.16, the default value for this limit
153 is:
154 ULONG_MAX - 2^24
156 The effect of this value (which is suitable for both 32-bit and
158 64-bit systems) is to impose no limitation on allocations. This
159 value, rather than ULONG_MAX, was chosen as the default to pre‐
160 vent some cases where historical applications simply raised the
161 existing limit without first checking its current value. Such
162 applications would cause the value to overflow if the limit was
163 set at ULONG_MAX.
164 From Linux 2.4 up to Linux 3.15, the default value for this
166 limit was:
167 SHMMAX / PAGE_SIZE * (SHMMNI / 16)
169 If SHMMAX and SHMMNI were not modified, then multiplying the re‐
171 sult of this formula by the page size (to get a value in bytes)
172 yielded a value of 8 GB as the limit on the total memory used by
173 all shared memory segments.
174 SHMMAX Maximum size in bytes for a shared memory segment.
176 On Linux, this limit can be read and modified via /proc/sys/ker‐
178 nel/shmmax. Since Linux 3.16, the default value for this limit
179 is:
180 ULONG_MAX - 2^24
182 The effect of this value (which is suitable for both 32-bit and
184 64-bit systems) is to impose no limitation on allocations. See
185 the description of SHMALL for a discussion of why this default
186 value (rather than ULONG_MAX) is used.
187 From Linux 2.2 up to Linux 3.15, the default value of this limit
189 was 0x2000000 (32 MB).
190 Because it is not possible to map just part of a shared memory
192 segment, the amount of virtual memory places another limit on
193 the maximum size of a usable segment: for example, on i386 the
194 largest segments that can be mapped have a size of around
195 2.8 GB, and on x86-64 the limit is around 127 TB.
196 SHMMIN Minimum size in bytes for a shared memory segment: implementa‐
198 tion dependent (currently 1 byte, though PAGE_SIZE is the effec‐
199 tive minimum size).
200 SHMMNI System-wide limit on the number of shared memory segments. In
202 Linux 2.2, the default value for this limit was 128; since Linux
203 2.4, the default value is 4096.
204 On Linux, this limit can be read and modified via /proc/sys/ker‐
206 nel/shmmni.
207 The implementation has no specific limits for the per-process maximum
209 number of shared memory segments (SHMSEG).
210 Linux notes
212 Until version 2.3.30, Linux would return EIDRM for a shmget() on a
213 shared memory segment scheduled for deletion.
214
216 The name choice IPC_PRIVATE was perhaps unfortunate, IPC_NEW would more
217 clearly show its function.
218
220 See shmop(2).
221
223 memfd_create(2), shmat(2), shmctl(2), shmdt(2), ftok(3), capabili‐
224 ties(7), shm_overview(7), sysvipc(7)
225
227 This page is part of release 5.13 of the Linux man-pages project. A
228 description of the project, information about reporting bugs, and the
229 latest version of this page, can be found at
230 https://www.kernel.org/doc/man-pages/.
231Linux 2021-03-22 SHMGET(2)