1SHMOP(2) Linux Programmer's Manual SHMOP(2)
2
6 shmat, shmdt - System V shared memory operations
7
9 #include <sys/shm.h>
10 void *shmat(int shmid, const void *shmaddr, int shmflg);
12 int shmdt(const void *shmaddr);
13
15 shmat()
16 shmat() attaches the System V shared memory segment identified by shmid
17 to the address space of the calling process. The attaching address is
18 specified by shmaddr with one of the following criteria:
19 • If shmaddr is NULL, the system chooses a suitable (unused) page-
21 aligned address to attach the segment.
22 • If shmaddr isn't NULL and SHM_RND is specified in shmflg, the attach
24 occurs at the address equal to shmaddr rounded down to the nearest
25 multiple of SHMLBA.
26 • Otherwise, shmaddr must be a page-aligned address at which the attach
28 occurs.
29 In addition to SHM_RND, the following flags may be specified in the
31 shmflg bit-mask argument:
32 SHM_EXEC (Linux-specific; since Linux 2.6.9)
34 Allow the contents of the segment to be executed. The caller
35 must have execute permission on the segment.
36 SHM_RDONLY
38 Attach the segment for read-only access. The process must have
39 read permission for the segment. If this flag is not specified,
40 the segment is attached for read and write access, and the
41 process must have read and write permission for the segment.
42 There is no notion of a write-only shared memory segment.
43 SHM_REMAP (Linux-specific)
45 This flag specifies that the mapping of the segment should re‐
46 place any existing mapping in the range starting at shmaddr and
47 continuing for the size of the segment. (Normally, an EINVAL
48 error would result if a mapping already exists in this address
49 range.) In this case, shmaddr must not be NULL.
50 The brk(2) value of the calling process is not altered by the attach.
52 The segment will automatically be detached at process exit. The same
53 segment may be attached as a read and as a read-write one, and more
54 than once, in the process's address space.
55 A successful shmat() call updates the members of the shmid_ds structure
57 (see shmctl(2)) associated with the shared memory segment as follows:
58 • shm_atime is set to the current time.
60 • shm_lpid is set to the process-ID of the calling process.
62 • shm_nattch is incremented by one.
64 shmdt()
66 shmdt() detaches the shared memory segment located at the address spec‐
67 ified by shmaddr from the address space of the calling process. The
68 to-be-detached segment must be currently attached with shmaddr equal to
69 the value returned by the attaching shmat() call.
70 On a successful shmdt() call, the system updates the members of the
72 shmid_ds structure associated with the shared memory segment as fol‐
73 lows:
74 • shm_dtime is set to the current time.
76 • shm_lpid is set to the process-ID of the calling process.
78 • shm_nattch is decremented by one. If it becomes 0 and the segment is
80 marked for deletion, the segment is deleted.
81
83 On success, shmat() returns the address of the attached shared memory
84 segment; on error, (void *) -1 is returned, and errno is set to indi‐
85 cate the error.
86 On success, shmdt() returns 0; on error -1 is returned, and errno is
88 set to indicate the error.
89
91 shmat() can fail with one of the following errors:
92 EACCES The calling process does not have the required permissions for
94 the requested attach type, and does not have the CAP_IPC_OWNER
95 capability in the user namespace that governs its IPC namespace.
96 EIDRM shmid points to a removed identifier.
98 EINVAL Invalid shmid value, unaligned (i.e., not page-aligned and
100 SHM_RND was not specified) or invalid shmaddr value, or can't
101 attach segment at shmaddr, or SHM_REMAP was specified and
102 shmaddr was NULL.
103 ENOMEM Could not allocate memory for the descriptor or for the page ta‐
105 bles.
106 shmdt() errno can fail with one of the following errors:
108 EINVAL There is no shared memory segment attached at shmaddr; or,
110 shmaddr is not aligned on a page boundary.
111
113 POSIX.1-2001, POSIX.1-2008, SVr4.
114 In SVID 3 (or perhaps earlier), the type of the shmaddr argument was
116 changed from char * into const void *, and the returned type of shmat()
117 from char * into void *.
118
120 After a fork(2), the child inherits the attached shared memory seg‐
121 ments.
122 After an execve(2), all attached shared memory segments are detached
124 from the process.
125 Upon _exit(2), all attached shared memory segments are detached from
127 the process.
128 Using shmat() with shmaddr equal to NULL is the preferred, portable way
130 of attaching a shared memory segment. Be aware that the shared memory
131 segment attached in this way may be attached at different addresses in
132 different processes. Therefore, any pointers maintained within the
133 shared memory must be made relative (typically to the starting address
134 of the segment), rather than absolute.
135 On Linux, it is possible to attach a shared memory segment even if it
137 is already marked to be deleted. However, POSIX.1 does not specify
138 this behavior and many other implementations do not support it.
139 The following system parameter affects shmat():
141 SHMLBA Segment low boundary address multiple. When explicitly specify‐
143 ing an attach address in a call to shmat(), the caller should
144 ensure that the address is a multiple of this value. This is
145 necessary on some architectures, in order either to ensure good
146 CPU cache performance or to ensure that different attaches of
147 the same segment have consistent views within the CPU cache.
148 SHMLBA is normally some multiple of the system page size. (On
149 many Linux architectures, SHMLBA is the same as the system page
150 size.)
151 The implementation places no intrinsic per-process limit on the number
153 of shared memory segments (SHMSEG).
154
156 The two programs shown below exchange a string using a shared memory
157 segment. Further details about the programs are given below. First,
158 we show a shell session demonstrating their use.
159 In one terminal window, we run the "reader" program, which creates a
161 System V shared memory segment and a System V semaphore set. The pro‐
162 gram prints out the IDs of the created objects, and then waits for the
163 semaphore to change value.
164 $ ./svshm_string_read
166 shmid = 1114194; semid = 15
167 In another terminal window, we run the "writer" program. The "writer"
169 program takes three command-line arguments: the IDs of the shared mem‐
170 ory segment and semaphore set created by the "reader", and a string.
171 It attaches the existing shared memory segment, copies the string to
172 the shared memory, and modifies the semaphore value.
173 $ ./svshm_string_write 1114194 15 'Hello, world'
175 Returning to the terminal where the "reader" is running, we see that
177 the program has ceased waiting on the semaphore and has printed the
178 string that was copied into the shared memory segment by the writer:
179 Hello, world
181 Program source: svshm_string.h
183 The following header file is included by the "reader" and "writer" pro‐
184 grams.
185 #include <sys/types.h>
187 #include <sys/ipc.h>
188 #include <sys/shm.h>
189 #include <sys/sem.h>
190 #include <stdio.h>
191 #include <stdlib.h>
192 #include <string.h>
193 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
195 } while (0)
196 union semun { /* Used in calls to semctl() */
198 int val;
199 struct semid_ds * buf;
200 unsigned short * array;
201 #if defined(__linux__)
202 struct seminfo * __buf;
203 #endif
204 };
205 #define MEM_SIZE 4096
207 Program source: svshm_string_read.c
209 The "reader" program creates a shared memory segment and a semaphore
210 set containing one semaphore. It then attaches the shared memory ob‐
211 ject into its address space and initializes the semaphore value to 1.
212 Finally, the program waits for the semaphore value to become 0, and af‐
213 terwards prints the string that has been copied into the shared memory
214 segment by the "writer".
215 /* svshm_string_read.c
217 Licensed under GNU General Public License v2 or later.
219 */
220 #include "svshm_string.h"
221 int
223 main(int argc, char *argv[])
224 {
225 int semid, shmid;
226 union semun arg, dummy;
227 struct sembuf sop;
228 char *addr;
229 /* Create shared memory and semaphore set containing one
231 semaphore. */
232 shmid = shmget(IPC_PRIVATE, MEM_SIZE, IPC_CREAT | 0600);
234 if (shmid == -1)
235 errExit("shmget");
236 semid = semget(IPC_PRIVATE, 1, IPC_CREAT | 0600);
238 if (semid == -1)
239 errExit("semget");
240 /* Attach shared memory into our address space. */
242 addr = shmat(shmid, NULL, SHM_RDONLY);
244 if (addr == (void *) -1)
245 errExit("shmat");
246 /* Initialize semaphore 0 in set with value 1. */
248 arg.val = 1;
250 if (semctl(semid, 0, SETVAL, arg) == -1)
251 errExit("semctl");
252 printf("shmid = %d; semid = %d\n", shmid, semid);
254 /* Wait for semaphore value to become 0. */
256 sop.sem_num = 0;
258 sop.sem_op = 0;
259 sop.sem_flg = 0;
260 if (semop(semid, &sop, 1) == -1)
262 errExit("semop");
263 /* Print the string from shared memory. */
265 printf("%s\n", addr);
267 /* Remove shared memory and semaphore set. */
269 if (shmctl(shmid, IPC_RMID, NULL) == -1)
271 errExit("shmctl");
272 if (semctl(semid, 0, IPC_RMID, dummy) == -1)
273 errExit("semctl");
274 exit(EXIT_SUCCESS);
276 }
277 Program source: svshm_string_write.c
279 The writer program takes three command-line arguments: the IDs of the
280 shared memory segment and semaphore set that have already been created
281 by the "reader", and a string. It attaches the shared memory segment
282 into its address space, and then decrements the semaphore value to 0 in
283 order to inform the "reader" that it can now examine the contents of
284 the shared memory.
285 /* svshm_string_write.c
287 Licensed under GNU General Public License v2 or later.
289 */
290 #include "svshm_string.h"
291 int
293 main(int argc, char *argv[])
294 {
295 int semid, shmid;
296 struct sembuf sop;
297 char *addr;
298 size_t len;
299 if (argc != 4) {
301 fprintf(stderr, "Usage: %s shmid semid string\n", argv[0]);
302 exit(EXIT_FAILURE);
303 }
304 len = strlen(argv[3]) + 1; /* +1 to include trailing '\0' */
306 if (len > MEM_SIZE) {
307 fprintf(stderr, "String is too big!\n");
308 exit(EXIT_FAILURE);
309 }
310 /* Get object IDs from command-line. */
312 shmid = atoi(argv[1]);
314 semid = atoi(argv[2]);
315 /* Attach shared memory into our address space and copy string
317 (including trailing null byte) into memory. */
318 addr = shmat(shmid, NULL, 0);
320 if (addr == (void *) -1)
321 errExit("shmat");
322 memcpy(addr, argv[3], len);
324 /* Decrement semaphore to 0. */
326 sop.sem_num = 0;
328 sop.sem_op = -1;
329 sop.sem_flg = 0;
330 if (semop(semid, &sop, 1) == -1)
332 errExit("semop");
333 exit(EXIT_SUCCESS);
335 }
336
338 brk(2), mmap(2), shmctl(2), shmget(2), capabilities(7), shm_over‐
339 view(7), sysvipc(7)
340
342 This page is part of release 5.12 of the Linux man-pages project. A
343 description of the project, information about reporting bugs, and the
344 latest version of this page, can be found at
345 https://www.kernel.org/doc/man-pages/.
346Linux 2021-03-22 SHMOP(2)