man2/shmop.2

   1 .\" Copyright 1993 Giorgio Ciucci (giorgio@crcc.it)
   2 .\"
   3 .\" Permission is granted to make and distribute verbatim copies of this
   4 .\" manual provided the copyright notice and this permission notice are
   5 .\" preserved on all copies.
   6 .\"
   7 .\" Permission is granted to copy and distribute modified versions of this
   8 .\" manual under the conditions for verbatim copying, provided that the
   9 .\" entire resulting derived work is distributed under the terms of a
  10 .\" permission notice identical to this one.
  11 .\"
  12 .\" Since the Linux kernel and libraries are constantly changing, this
  13 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  14 .\" responsibility for errors or omissions, or for damages resulting from
  15 .\" the use of the information contained herein.  The author(s) may not
  16 .\" have taken the same level of care in the production of this manual,
  17 .\" which is licensed free of charge, as they might when working
  18 .\" professionally.
  19 .\"
  20 .\" Formatted or processed versions of this manual, if unaccompanied by
  21 .\" the source, must acknowledge the copyright and authors of this work.
  22 .\"
  23 .\" Modified Sun Nov 28 17:06:19 1993, Rik Faith (faith@cs.unc.edu)
  24 .\"          with material from Luigi P. Bai (lpb@softint.com)
  25 .\" Portions Copyright 1993 Luigi P. Bai
  26 .\" Modified Tue Oct 22 22:04:23 1996 by Eric S. Raymond <esr@thyrsus.com>
  27 .\" Modified, 5 Jan 2002, Michael Kerrisk <mtk-manpages@gmx.net>
  28 .\" Modified, 19 Sep 2002, Michael Kerrisk <mtk-manpages@gmx.net>
  29 .\"     Added SHM_REMAP flag description
  30 .\" Modified, 27 May 2004, Michael Kerrisk <mtk-manpages@gmx.net>
  31 .\"     Added notes on capability requirements
  32 .\" Modified, 11 Nov 2004, Michael Kerrisk <mtk-manpages@gmx.net>
  33 .\"     Language and formatting clean-ups
  34 .\"     Changed wording and placement of sentence regarding attachment
  35 .\"             of segments marked for destruction
  36 .\"
  37 .TH SHMOP 2 2004-11-10 "Linux 2.6.9" "Linux Programmer's Manual"
  38 .SH NAME
  39 shmop, shmat, shmdt \- shared memory operations
  40 .SH SYNOPSIS
  41 .nf
  42 .B
  43 #include <sys/types.h>
  44 .B
  45 #include <sys/shm.h>
  46 .fi
  47 .sp
  48 .BI "void *shmat(int " shmid ,
  49 .BI "const void *" shmaddr ,
  50 .BI "int " shmflg );
  51 .sp
  52 .BI "int shmdt(const void *" shmaddr );
  53 .SH DESCRIPTION
  54 .BR shmat ()
  55 attaches the shared memory segment identified by
  56 .I shmid
  57 to the address space of the calling process.
  58 The attaching address is specified by
  59 .I shmaddr
  60 with one of the following criteria:
  61 .LP
  62 If
  63 .I shmaddr
  64 is NULL,
  65 the system chooses a suitable (unused) address at which to attach
  66 the segment.
  67 .LP
  68 If
  69 .I shmaddr
  70 isn't NULL
  71 and
  72 .B SHM_RND
  73 is specified in
  74 .IR shmflg ,
  75 the attach occurs at the address equal to
  76 .I shmaddr
  77 rounded down to the nearest multiple of
  78 .BR SHMLBA .
  79 Otherwise
  80 .I shmaddr
  81 must be a page-aligned address at which the attach occurs.
  82 .PP
  83 If
  84 .B SHM_RDONLY
  85 is specified in
  86 .IR shmflg ,
  87 the segment is attached for reading and the process must have
  88 read permission for the segment.
  89 Otherwise the segment is attached for read and write
  90 and the process must have read and write permission for the segment.
  91 There is no notion of a write-only shared memory segment.
  92 .PP
  93 The (Linux-specific)
  94 .B SHM_REMAP
  95 flag may be specified in
  96 .I shmflg
  97 to indicate that the mapping of the segment should replace
  98 any existing mapping in the range starting at
  99 .I shmaddr
 100 and continuing for the size of the segment.
 101 (Normally an
 102 .B EINVAL
 103 error would result if a mapping already exists in this address range.)
 104 In this case,
 105 .I shmaddr
 106 must not be NULL.
 107 .PP
 108 The
 109 .BR brk (2)
 110 value of the calling process is not altered by the attach.
 111 The segment will automatically be detached at process exit.
 112 The same segment may be attached as a read and as a read-write
 113 one, and more than once, in the process's address space.
 114 .PP
 115 A successful
 116 .BR shmat ()
 117 call updates the members of the
 118 .B shmid_ds
 119 structure (see
 120 .BR shmctl (2))
 121 associated with the shared memory segment as follows:
 122 .IP
 123 .I shm_atime
 124 is set to the current time.
 125 .IP
 126 .I shm_lpid
 127 is set to the process-ID of the calling process.
 128 .IP
 129 .I shm_nattch
 130 is incremented by one.
 131 .PP
 132 .BR shmdt ()
 133 detaches the shared memory segment located at the address specified by
 134 .I shmaddr
 135 from the address space of the calling process.
 136 The to\-be\-detached segment must be currently
 137 attached with
 138 .I shmaddr
 139 equal to the value returned by the attaching
 140 .BR shmat ()
 141 call.
 142 .PP
 143 On a successful
 144 .BR shmdt ()
 145 call the system updates the members of the
 146 .I shmid_ds
 147 structure associated with the shared memory segment as follows:
 148 .IP
 149 .I shm_dtime
 150 is set to the current time.
 151 .IP
 152 .I shm_lpid
 153 is set to the process-ID of the calling process.
 154 .IP
 155 .I shm_nattch
 156 is decremented by one.
 157 If it becomes 0 and the segment is marked for deletion,
 158 the segment is deleted.
 159 .SH "SYSTEM CALLS"
 160 .TP 11
 161 .BR fork (2)
 162 After a
 163 .BR fork (2)
 164 the child inherits the attached shared memory segments.
 165 .TP
 166 .BR execve (2)
 167 After an
 168 .BR execve (2)
 169 all attached shared memory segments are detached from the process.
 170 .TP
 171 .BR _exit (2)
 172 Upon
 173 .BR _exit (2)
 174 all attached shared memory segments are detached from the process.
 175 .SH "RETURN VALUE"
 176 On success
 177 .BR shmat ()
 178 returns the address of the attached shared memory segment; on error
 179 .I (void *) \-1
 180 is returned, and
 181 .I errno
 182 is set to indicate the cause of the error.
 183
 184 On success
 185 .BR shmdt ()
 186 returns 0; on error \-1 is returned, and
 187 .I errno
 188 is set to indicate the cause of the error.
 189 .SH ERRORS
 190 When
 191 .BR shmat ()
 192 fails,
 193 .I errno
 194 is set to one of the following:
 195 .TP
 196 .B EACCES
 197 The calling process does not have the required permissions for
 198 the requested attach type, and does not have the
 199 .B CAP_IPC_OWNER
 200 capability.
 201 .TP
 202 .B EINVAL
 203 Invalid
 204 .I shmid
 205 value, unaligned (i.e., not page-aligned and \fBSHM_RND\fP was not
 206 specified) or invalid
 207 .I shmaddr
 208 value, or failing attach at
 209 .BR brk (2),
 210 .\" FIXME What does "failing attach at brk" mean?  (Is this phrase
 211 .\" just junk?)
 212 or
 213 .B SHM_REMAP
 214 was specified and
 215 .I shmaddr
 216 was NULL.
 217 .TP
 218 .B ENOMEM
 219 Could not allocate memory for the descriptor or for the page tables.
 220 .PP
 221 When
 222 .BR shmdt ()
 223 fails,
 224 .I errno
 225 is set as follows:
 226 .TP
 227 .B EINVAL
 228 There is no shared memory segment attached at
 229 .IR shmaddr ;
 230 or,
 231 .\" The following since 2.6.17-rc1:
 232 .I shmaddr
 233 is not aligned on a page boundary.
 234 .SH NOTES
 235 Using
 236 .BR shmat ()
 237 with
 238 .I shmaddr
 239 equal to NULL
 240 is the preferred, portable way of attaching a shared memory segment.
 241 Be aware that the shared memory segment attached in this way
 242 may be attached at different addresses in different processes.
 243 Therefore, any pointers maintained within the shared memory must be
 244 made relative (typically to the starting address of the segment),
 245 rather than absolute.
 246 .PP
 247 On Linux, it is possible to attach a shared memory segment even if it
 248 is already marked to be deleted.
 249 However, POSIX.1-2001 does not specify this behaviour and
 250 many other implementations do not support it.
 251 .LP
 252 The following system parameter affects
 253 .BR shmat ():
 254 .TP 11
 255 .\" FIXME A good explanation of the rationale for the existence
 256 .\" of SHMLBA would be useful here
 257 .B SHMLBA
 258 Segment low boundary address multiple.
 259 Must be page aligned.
 260 For the current implementation the
 261 .B SHMLBA
 262 value is
 263 .BR PAGE_SIZE .
 264 .\" FIXME That last sentence isn't true for all Linux
 265 .\" architectures (i.e., SHMLBA != PAGE_SIZE for some architectures)
 266 .\" -- MTK, Nov 04
 267 .PP
 268 The implementation places no intrinsic limit on the per\-process maximum
 269 number of shared memory segments
 270 .RB ( SHMSEG ).
 271 .SH "CONFORMING TO"
 272 SVr4, POSIX.1-2001.
 273 .\" SVr4 documents an additional error condition EMFILE.
 274
 275 In SVID 3 (or perhaps earlier)
 276 the type of the \fIshmaddr\fP argument was changed from
 277 .I "char *"
 278 into
 279 .IR "const void *" ,
 280 and the returned type of
 281 .BR shmat ()
 282 from
 283 .I "char *"
 284 into
 285 .IR "void *" .
 286 (Linux libc4 and libc5 have the
 287 .I "char *"
 288 prototypes; glibc2 has
 289 .IR "void *" .)
 290 .SH "SEE ALSO"
 291 .BR brk (2),
 292 .BR mmap (2),
 293 .BR shmctl (2),
 294 .BR shmget (2),
 295 .BR capabilities (7),
 296 .BR svipc (7)