1SHMOP(2) System Calls Manual SHMOP(2)
2
6 shmat, shmdt - System V shared memory operations
7
9 Standard C library (libc, -lc)
10
12 #include <sys/shm.h>
13 void *shmat(int shmid, const void *_Nullable shmaddr, int shmflg);
15 int shmdt(const void *shmaddr);
16
18 shmat()
19 shmat() attaches the System V shared memory segment identified by shmid
20 to the address space of the calling process. The attaching address is
21 specified by shmaddr with one of the following criteria:
22 • If shmaddr is NULL, the system chooses a suitable (unused) page-
24 aligned address to attach the segment.
25 • If shmaddr isn't NULL and SHM_RND is specified in shmflg, the attach
27 occurs at the address equal to shmaddr rounded down to the nearest
28 multiple of SHMLBA.
29 • Otherwise, shmaddr must be a page-aligned address at which the at‐
31 tach occurs.
32 In addition to SHM_RND, the following flags may be specified in the
34 shmflg bit-mask argument:
35 SHM_EXEC (Linux-specific; since Linux 2.6.9)
37 Allow the contents of the segment to be executed. The caller
38 must have execute permission on the segment.
39 SHM_RDONLY
41 Attach the segment for read-only access. The process must have
42 read permission for the segment. If this flag is not specified,
43 the segment is attached for read and write access, and the
44 process must have read and write permission for the segment.
45 There is no notion of a write-only shared memory segment.
46 SHM_REMAP (Linux-specific)
48 This flag specifies that the mapping of the segment should re‐
49 place any existing mapping in the range starting at shmaddr and
50 continuing for the size of the segment. (Normally, an EINVAL
51 error would result if a mapping already exists in this address
52 range.) In this case, shmaddr must not be NULL.
53 The brk(2) value of the calling process is not altered by the attach.
55 The segment will automatically be detached at process exit. The same
56 segment may be attached as a read and as a read-write one, and more
57 than once, in the process's address space.
58 A successful shmat() call updates the members of the shmid_ds structure
60 (see shmctl(2)) associated with the shared memory segment as follows:
61 • shm_atime is set to the current time.
63 • shm_lpid is set to the process-ID of the calling process.
65 • shm_nattch is incremented by one.
67 shmdt()
69 shmdt() detaches the shared memory segment located at the address spec‐
70 ified by shmaddr from the address space of the calling process. The
71 to-be-detached segment must be currently attached with shmaddr equal to
72 the value returned by the attaching shmat() call.
73 On a successful shmdt() call, the system updates the members of the
75 shmid_ds structure associated with the shared memory segment as fol‐
76 lows:
77 • shm_dtime is set to the current time.
79 • shm_lpid is set to the process-ID of the calling process.
81 • shm_nattch is decremented by one. If it becomes 0 and the segment
83 is marked for deletion, the segment is deleted.
84
86 On success, shmat() returns the address of the attached shared memory
87 segment; on error, (void *) -1 is returned, and errno is set to indi‐
88 cate the error.
89 On success, shmdt() returns 0; on error -1 is returned, and errno is
91 set to indicate the error.
92
94 shmat() can fail with one of the following errors:
95 EACCES The calling process does not have the required permissions for
97 the requested attach type, and does not have the CAP_IPC_OWNER
98 capability in the user namespace that governs its IPC namespace.
99 EIDRM shmid points to a removed identifier.
101 EINVAL Invalid shmid value, unaligned (i.e., not page-aligned and
103 SHM_RND was not specified) or invalid shmaddr value, or can't
104 attach segment at shmaddr, or SHM_REMAP was specified and
105 shmaddr was NULL.
106 ENOMEM Could not allocate memory for the descriptor or for the page ta‐
108 bles.
109 shmdt() can fail with one of the following errors:
111 EINVAL There is no shared memory segment attached at shmaddr; or,
113 shmaddr is not aligned on a page boundary.
114
116 POSIX.1-2008.
117
119 POSIX.1-2001, SVr4.
120 In SVID 3 (or perhaps earlier), the type of the shmaddr argument was
122 changed from char * into const void *, and the returned type of shmat()
123 from char * into void *.
124
126 After a fork(2), the child inherits the attached shared memory seg‐
127 ments.
128 After an execve(2), all attached shared memory segments are detached
130 from the process.
131 Upon _exit(2), all attached shared memory segments are detached from
133 the process.
134 Using shmat() with shmaddr equal to NULL is the preferred, portable way
136 of attaching a shared memory segment. Be aware that the shared memory
137 segment attached in this way may be attached at different addresses in
138 different processes. Therefore, any pointers maintained within the
139 shared memory must be made relative (typically to the starting address
140 of the segment), rather than absolute.
141 On Linux, it is possible to attach a shared memory segment even if it
143 is already marked to be deleted. However, POSIX.1 does not specify
144 this behavior and many other implementations do not support it.
145 The following system parameter affects shmat():
147 SHMLBA Segment low boundary address multiple. When explicitly specify‐
149 ing an attach address in a call to shmat(), the caller should
150 ensure that the address is a multiple of this value. This is
151 necessary on some architectures, in order either to ensure good
152 CPU cache performance or to ensure that different attaches of
153 the same segment have consistent views within the CPU cache.
154 SHMLBA is normally some multiple of the system page size. (On
155 many Linux architectures, SHMLBA is the same as the system page
156 size.)
157 The implementation places no intrinsic per-process limit on the number
159 of shared memory segments (SHMSEG).
160
162 The two programs shown below exchange a string using a shared memory
163 segment. Further details about the programs are given below. First,
164 we show a shell session demonstrating their use.
165 In one terminal window, we run the "reader" program, which creates a
167 System V shared memory segment and a System V semaphore set. The pro‐
168 gram prints out the IDs of the created objects, and then waits for the
169 semaphore to change value.
170 $ ./svshm_string_read
172 shmid = 1114194; semid = 15
173 In another terminal window, we run the "writer" program. The "writer"
175 program takes three command-line arguments: the IDs of the shared mem‐
176 ory segment and semaphore set created by the "reader", and a string.
177 It attaches the existing shared memory segment, copies the string to
178 the shared memory, and modifies the semaphore value.
179 $ ./svshm_string_write 1114194 15 'Hello, world'
181 Returning to the terminal where the "reader" is running, we see that
183 the program has ceased waiting on the semaphore and has printed the
184 string that was copied into the shared memory segment by the writer:
185 Hello, world
187 Program source: svshm_string.h
189 The following header file is included by the "reader" and "writer" pro‐
190 grams:
191 /* svshm_string.h
193 Licensed under GNU General Public License v2 or later.
195 */
196 #include <sys/types.h>
197 #include <sys/ipc.h>
198 #include <sys/shm.h>
199 #include <sys/sem.h>
200 #include <stdio.h>
201 #include <stdlib.h>
202 #include <string.h>
203 #define errExit(msg) do { perror(msg); exit(EXIT_FAILURE); \
205 } while (0)
206 union semun { /* Used in calls to semctl() */
208 int val;
209 struct semid_ds * buf;
210 unsigned short * array;
211 #if defined(__linux__)
212 struct seminfo * __buf;
213 #endif
214 };
215 #define MEM_SIZE 4096
217 Program source: svshm_string_read.c
219 The "reader" program creates a shared memory segment and a semaphore
220 set containing one semaphore. It then attaches the shared memory ob‐
221 ject into its address space and initializes the semaphore value to 1.
222 Finally, the program waits for the semaphore value to become 0, and af‐
223 terwards prints the string that has been copied into the shared memory
224 segment by the "writer".
225 /* svshm_string_read.c
227 Licensed under GNU General Public License v2 or later.
229 */
230 #include <stdio.h>
231 #include <stdlib.h>
232 #include <sys/ipc.h>
233 #include <sys/sem.h>
234 #include <sys/shm.h>
235 #include "svshm_string.h"
237 int
239 main(void)
240 {
241 int semid, shmid;
242 char *addr;
243 union semun arg, dummy;
244 struct sembuf sop;
245 /* Create shared memory and semaphore set containing one
247 semaphore. */
248 shmid = shmget(IPC_PRIVATE, MEM_SIZE, IPC_CREAT | 0600);
250 if (shmid == -1)
251 errExit("shmget");
252 semid = semget(IPC_PRIVATE, 1, IPC_CREAT | 0600);
254 if (semid == -1)
255 errExit("semget");
256 /* Attach shared memory into our address space. */
258 addr = shmat(shmid, NULL, SHM_RDONLY);
260 if (addr == (void *) -1)
261 errExit("shmat");
262 /* Initialize semaphore 0 in set with value 1. */
264 arg.val = 1;
266 if (semctl(semid, 0, SETVAL, arg) == -1)
267 errExit("semctl");
268 printf("shmid = %d; semid = %d\n", shmid, semid);
270 /* Wait for semaphore value to become 0. */
272 sop.sem_num = 0;
274 sop.sem_op = 0;
275 sop.sem_flg = 0;
276 if (semop(semid, &sop, 1) == -1)
278 errExit("semop");
279 /* Print the string from shared memory. */
281 printf("%s\n", addr);
283 /* Remove shared memory and semaphore set. */
285 if (shmctl(shmid, IPC_RMID, NULL) == -1)
287 errExit("shmctl");
288 if (semctl(semid, 0, IPC_RMID, dummy) == -1)
289 errExit("semctl");
290 exit(EXIT_SUCCESS);
292 }
293 Program source: svshm_string_write.c
295 The writer program takes three command-line arguments: the IDs of the
296 shared memory segment and semaphore set that have already been created
297 by the "reader", and a string. It attaches the shared memory segment
298 into its address space, and then decrements the semaphore value to 0 in
299 order to inform the "reader" that it can now examine the contents of
300 the shared memory.
301 /* svshm_string_write.c
303 Licensed under GNU General Public License v2 or later.
305 */
306 #include <stdio.h>
307 #include <stdlib.h>
308 #include <string.h>
309 #include <sys/sem.h>
310 #include <sys/shm.h>
311 #include "svshm_string.h"
313 int
315 main(int argc, char *argv[])
316 {
317 int semid, shmid;
318 char *addr;
319 size_t len;
320 struct sembuf sop;
321 if (argc != 4) {
323 fprintf(stderr, "Usage: %s shmid semid string\n", argv[0]);
324 exit(EXIT_FAILURE);
325 }
326 len = strlen(argv[3]) + 1; /* +1 to include trailing '\0' */
328 if (len > MEM_SIZE) {
329 fprintf(stderr, "String is too big!\n");
330 exit(EXIT_FAILURE);
331 }
332 /* Get object IDs from command-line. */
334 shmid = atoi(argv[1]);
336 semid = atoi(argv[2]);
337 /* Attach shared memory into our address space and copy string
339 (including trailing null byte) into memory. */
340 addr = shmat(shmid, NULL, 0);
342 if (addr == (void *) -1)
343 errExit("shmat");
344 memcpy(addr, argv[3], len);
346 /* Decrement semaphore to 0. */
348 sop.sem_num = 0;
350 sop.sem_op = -1;
351 sop.sem_flg = 0;
352 if (semop(semid, &sop, 1) == -1)
354 errExit("semop");
355 exit(EXIT_SUCCESS);
357 }
358
360 brk(2), mmap(2), shmctl(2), shmget(2), capabilities(7), shm_over‐
361 view(7), sysvipc(7)
362Linux man-pages 6.05 2023-05-03 SHMOP(2)