man2/memfd_create.2

   1 .\" Copyright (C) 2014 Michael Kerrisk <mtk.manpages@gmail.com>
   2 .\" and Copyright (C) 2014 David Herrmann <dh.herrmann@gmail.com>
   3 .\"
   4 .\" %%%LICENSE_START(GPLv2+)
   5 .\"
   6 .\" This program is free software; you can redistribute it and/or modify
   7 .\" it under the terms of the GNU General Public License as published by
   8 .\" the Free Software Foundation; either version 2 of the License, or
   9 .\" (at your option) any later version.
  10 .\"
  11 .\" This program is distributed in the hope that it will be useful,
  12 .\" but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 .\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  14 .\" GNU General Public License for more details.
  15 .\"
  16 .\" You should have received a copy of the GNU General Public
  17 .\" License along with this manual; if not, see
  18 .\" <http://www.gnu.org/licenses/>.
  19 .\" %%%LICENSE_END
  20 .\"
  21 .TH MEMFD_CREATE 2 2016-03-15 Linux "Linux Programmer's Manual"
  22 .SH NAME
  23 memfd_create \- create an anonymous file
  24 .SH SYNOPSIS
  25 .B #include <sys/memfd.h>
  26 .sp
  27 .BI "int memfd_create(const char *" name ", unsigned int " flags ");"
  28 .SH DESCRIPTION
  29 .BR memfd_create ()
  30 creates an anonymous file and returns a file descriptor that refers to it.
  31 The file behaves like a regular file, and so can be modified,
  32 truncated, memory-mapped, and so on.
  33 However, unlike a regular file,
  34 it lives in RAM and has a volatile backing storage.
  35 Once all references to the file are dropped, it is automatically released.
  36 Anonymous memory is used for all backing pages of the file.
  37 Therefore, files created by
  38 .BR memfd_create ()
  39 have the same semantics as other anonymous
  40 .\" David Herrmann:
  41 .\"     memfd uses VM_NORESERVE so each page is accounted on first access.
  42 .\"     This means, the overcommit-limits (see __vm_enough_memory()) and the
  43 .\"     memory-cgroup limits (mem_cgroup_try_charge()) are applied. Note that
  44 .\"     those are accounted on "current" and "current->mm", that is, the
  45 .\"     process doing the first page access.
  46 memory allocations such as those allocated using
  47 .BR mmap (2)
  48 with the
  49 .BR MAP_ANONYMOUS
  50 flag.
  51
  52 The initial size of the file is set to 0.
  53 Following the call, the file size should be set using
  54 .BR ftruncate (2).
  55 (Alternatively, the file may be populated by calls to
  56 .BR write (2)
  57 or similar.)
  58
  59 The name supplied in
  60 .I name
  61 is used as a filename and will be displayed
  62 as the target of the corresponding symbolic link in the directory
  63 .IR /proc/self/fd/ .
  64 The displayed name is always prefixed with
  65 .IR memfd:
  66 and serves only for debugging purposes.
  67 Names do not affect the behavior of the file descriptor,
  68 and as such multiple files can have the same name without any side effects.
  69
  70 The following values may be bitwise ORed in
  71 .IR flags
  72 to change the behavior of
  73 .BR memfd_create ():
  74 .TP
  75 .BR MFD_CLOEXEC
  76 Set the close-on-exec
  77 .RB ( FD_CLOEXEC )
  78 flag on the new file descriptor.
  79 See the description of the
  80 .B O_CLOEXEC
  81 flag in
  82 .BR open (2)
  83 for reasons why this may be useful.
  84 .TP
  85 .BR MFD_ALLOW_SEALING
  86 Allow sealing operations on this file.
  87 See the discussion of the
  88 .B F_ADD_SEALS
  89 and
  90 .BR F_GET_SEALS
  91 operations in
  92 .BR fcntl (2),
  93 and also NOTES, below.
  94 The initial set of seals is empty.
  95 If this flag is not set, the initial set of seals will be
  96 .BR F_SEAL_SEAL ,
  97 meaning that no other seals can be set on the file.
  98 .\" FIXME Why is the MFD_ALLOW_SEALING behavior not simply the default?
  99 .\"       Is it worth adding some text explaining this?
 100 .PP
 101 Unused bits in
 102 .I flags
 103 must be 0.
 104
 105 As its return value,
 106 .BR memfd_create ()
 107 returns a new file descriptor that can be used to refer to the file.
 108 This file descriptor is opened for both reading and writing
 109 .RB ( O_RDWR )
 110 and
 111 .B O_LARGEFILE
 112 is set for the file descriptor.
 113
 114 With respect to
 115 .BR fork (2)
 116 and
 117 .BR execve (2),
 118 the usual semantics apply for the file descriptor created by
 119 .BR memfd_create ().
 120 A copy of the file descriptor is inherited by the child produced by
 121 .BR fork (2)
 122 and refers to the same file.
 123 The file descriptor is preserved across
 124 .BR execve (2),
 125 unless the close-on-exec flag has been set.
 126 .SH RETURN VALUE
 127 On success,
 128 .BR memfd_create ()
 129 returns a new file descriptor.
 130 On error, \-1 is returned and
 131 .I errno
 132 is set to indicate the error.
 133 .SH ERRORS
 134 .TP
 135 .B EFAULT
 136 The address in
 137 .IR name
 138 points to invalid memory.
 139 .TP
 140 .B EINVAL
 141 An unsupported value was specified in one of the arguments:
 142 .I flags
 143 included unknown bits, or
 144 .I name
 145 was too long.
 146 .TP
 147 .B EMFILE
 148 The per-process limit on the number of open file descriptors has been reached.
 149 .TP
 150 .B ENFILE
 151 The system-wide limit on the total number of open files has been reached.
 152 .TP
 153 .B ENOMEM
 154 There was insufficient memory to create a new anonymous file.
 155 .SH VERSIONS
 156 The
 157 .BR memfd_create ()
 158 system call first appeared in Linux 3.17.
 159 .\" FIXME . When glibc support appears, update the following sentence:
 160 Support in the GNU C library is pending.
 161 .SH CONFORMING TO
 162 The
 163 .BR memfd_create ()
 164 system call is Linux-specific.
 165 .SH NOTES
 166 .\" See also http://lwn.net/Articles/593918/
 167 .\" and http://lwn.net/Articles/594919/ and http://lwn.net/Articles/591108/
 168 The
 169 .BR memfd_create ()
 170 system call provides a simple alternative to manually mounting a
 171 .I tmpfs
 172 filesystem and creating and opening a file in that filesystem.
 173 The primary purpose of
 174 .BR memfd_create ()
 175 is to create files and associated file descriptors that are
 176 used with the file-sealing APIs provided by
 177 .BR fcntl (2).
 178
 179 The
 180 .BR memfd_create ()
 181 system call also has uses without file sealing
 182 (which is why file-sealing is disabled, unless explicitly requested with the
 183 .BR MFD_ALLOW_SEALING
 184 flag).
 185 In particular, it can be used as an alternative to creating files in
 186 .IR tmp
 187 or as an alternative to using the
 188 .BR  open (2)
 189 .B O_TMPFILE
 190 in cases where there is no intention to actually link the
 191 resulting file into the filesystem.
 192 .SS File sealing
 193 In the absence of file sealing,
 194 processes that communicate via shared memory must either trust each other,
 195 or take measures to deal with the possibility that an untrusted peer
 196 may manipulate the shared memory region in problematic ways.
 197 For example, an untrusted peer might modify the contents of the
 198 shared memory at any time, or shrink the shared memory region.
 199 The former possibility leaves the local process vulnerable to
 200 time-of-check-to-time-of-use race conditions
 201 (typically dealt with by copying data from
 202 the shared memory region before checking and using it).
 203 The latter possibility leaves the local process vulnerable to
 204 .BR SIGBUS
 205 signals when an attempt is made to access a now-nonexistent
 206 location in the shared memory region.
 207 (Dealing with this possibility necessitates the use of a handler for the
 208 .BR SIGBUS
 209 signal.)
 210
 211 Dealing with untrusted peers imposes extra complexity on
 212 code that employs shared memory.
 213 Memory sealing enables that extra complexity to be eliminated,
 214 by allowing a process to operate secure in the knowledge that
 215 its peer can't modify the shared memory in an undesired fashion.
 216
 217 An example of the usage of the sealing mechanism is as follows:
 218 .IP 1. 3
 219 The first process creates a
 220 .I tmpfs
 221 file using
 222 .BR memfd_create ().
 223 The call yields a file descriptor used in subsequent steps.
 224 .IP 2.
 225 The first process
 226 sizes the file created in the previous step using
 227 .BR ftruncate (2),
 228 maps it using
 229 .BR mmap (2),
 230 and populates the shared memory with the desired data.
 231 .IP 3.
 232 The first process uses the
 233 .BR fcntl (2)
 234 .B F_ADD_SEALS
 235 operation to place one or more seals on the file,
 236 in order to restrict further modifications on the file.
 237 (If placing the seal
 238 .BR F_SEAL_WRITE ,
 239 then it will be necessary to first unmap the shared writable mapping
 240 created in the previous step.)
 241 .IP 4.
 242 A second process obtains a file descriptor for the
 243 .I tmpfs
 244 file and maps it.
 245 Among the possible ways in which this could happen are the following:
 246 .RS
 247 .IP * 3
 248 The process that called
 249 .BR memfd_create ()
 250 could transfer the resulting file descriptor to the second process
 251 via a UNIX domain socket (see
 252 .BR unix (7)
 253 and
 254 .BR cmsg (3)).
 255 The second process then maps the file using
 256 .BR mmap (2).
 257 .IP *
 258 The second process is created via
 259 .BR fork (2)
 260 and thus automatically inherits the file descriptor and mapping.
 261 (Note that in this case and the next,
 262 there is a natural trust relationship between the two processes,
 263 since they are running under the same user ID.
 264 Therefore, file sealing would not normally be necessary.)
 265 .IP *
 266 The second process opens the file
 267 .IR /proc/<pd>/fd/<fd> ,
 268 where
 269 .I <pid>
 270 is the PID of the first process (the one that called
 271 .BR memfd_create ()),
 272 and
 273 .I <fd>
 274 is the number of the file descriptor returned by the call to
 275 .BR memfd_create ()
 276 in that process.
 277 The second process then maps the file using
 278 .BR mmap (2).
 279 .RE
 280 .IP 5.
 281 The second process uses the
 282 .BR fcntl (2)
 283 .B F_GET_SEALS
 284 operation to retrieve the bit mask of seals
 285 that has been applied to the file.
 286 This bit mask can be inspected in order to determine
 287 what kinds of restrictions have been placed on file modifications.
 288 If desired, the second process can apply further seals
 289 to impose additional restrictions (so long as the
 290 .BR F_SEAL_SEAL
 291 seal has not yet been applied).
 292 .SH EXAMPLE
 293 Below are shown two example programs that demonstrate the use of
 294 .BR memfd_create ()
 295 and the file sealing API.
 296
 297 The first program,
 298 .IR t_memfd_create.c ,
 299 creates a
 300 .I tmpfs
 301 file using
 302 .BR memfd_create (),
 303 sets a size for the file, maps it into memory,
 304 and optionally places some seals on the file.
 305 The program accepts up to three command-line arguments,
 306 of which the first two are required.
 307 The first argument is the name to associate with the file,
 308 the second argument is the size to be set for the file,
 309 and the optional third argument is a string of characters that specify
 310 seals to be set on file.
 311
 312 The second program,
 313 .IR t_get_seals.c ,
 314 can be used to open an existing file that was created via
 315 .BR memfd_create ()
 316 and inspect the set of seals that have been applied to that file.
 317
 318 The following shell session demonstrates the use of these programs.
 319 First we create a
 320 .I tmpfs
 321 file and set some seals on it:
 322
 323 .in +4n
 324 .nf
 325 $ \fB./t_memfd_create my_memfd_file 4096 sw &\fP
 326 [1] 11775
 327 PID: 11775; fd: 3; /proc/11775/fd/3
 328 .fi
 329 .in
 330
 331 At this point, the
 332 .I t_memfd_create
 333 program continues to run in the background.
 334 From another program, we can obtain a file descriptor for the
 335 file created by
 336 .BR memfd_create ()
 337 by opening the
 338 .IR /proc/PID/fd
 339 file that corresponds to the file descriptor opened by
 340 .BR memfd_create ().
 341 Using that pathname, we inspect the content of the
 342 .IR /proc/PID/fd
 343 symbolic link, and use our
 344 .I t_get_seals
 345 program to view the seals that have been placed on the file:
 346
 347 .in +4n
 348 .nf
 349 $ \fBreadlink /proc/11775/fd/3\fP
 350 /memfd:my_memfd_file (deleted)
 351 $ \fB./t_get_seals /proc/11775/fd/3\fP
 352 Existing seals: WRITE SHRINK
 353 .fi
 354 .in
 355 .SS Program source: t_memfd_create.c
 356 \&
 357 .nf
 358 #include <sys/memfd.h>
 359 #include <fcntl.h>
 360 #include <stdlib.h>
 361 #include <unistd.h>
 362 #include <string.h>
 363 #include <stdio.h>
 364
 365 #define errExit(msg)    do { perror(msg); exit(EXIT_FAILURE); \\
 366                         } while (0)
 367
 368 int
 369 main(int argc, char *argv[])
 370 {
 371     int fd;
 372     unsigned int seals;
 373     char *addr;
 374     char *name, *seals_arg;
 375     ssize_t len;
 376
 377     if (argc < 3) {
 378         fprintf(stderr, "%s name size [seals]\\n", argv[0]);
 379         fprintf(stderr, "\\t\(aqseals\(aq can contain any of the "
 380                 "following characters:\\n");
 381         fprintf(stderr, "\\t\\tg \- F_SEAL_GROW\\n");
 382         fprintf(stderr, "\\t\\ts \- F_SEAL_SHRINK\\n");
 383         fprintf(stderr, "\\t\\tw \- F_SEAL_WRITE\\n");
 384         fprintf(stderr, "\\t\\tS \- F_SEAL_SEAL\\n");
 385         exit(EXIT_FAILURE);
 386     }
 387
 388     name = argv[1];
 389     len = atoi(argv[2]);
 390     seals_arg = argv[3];
 391
 392     /* Create an anonymous file in tmpfs; allow seals to be
 393        placed on the file */
 394
 395     fd = memfd_create(name, MFD_ALLOW_SEALING);
 396     if (fd == \-1)
 397         errExit("memfd_create");
 398
 399     /* Size the file as specified on the command line */
 400
 401     if (ftruncate(fd, len) == \-1)
 402         errExit("truncate");
 403
 404     printf("PID: %ld; fd: %d; /proc/%ld/fd/%d\\n",
 405             (long) getpid(), fd, (long) getpid(), fd);
 406
 407     /* Code to map the file and populate the mapping with data
 408        omitted */
 409
 410     /* If a \(aqseals\(aq command\-line argument was supplied, set some
 411        seals on the file */
 412
 413     if (seals_arg != NULL) {
 414         seals = 0;
 415
 416         if (strchr(seals_arg, \(aqg\(aq) != NULL)
 417             seals |= F_SEAL_GROW;
 418         if (strchr(seals_arg, \(aqs\(aq) != NULL)
 419             seals |= F_SEAL_SHRINK;
 420         if (strchr(seals_arg, \(aqw\(aq) != NULL)
 421             seals |= F_SEAL_WRITE;
 422         if (strchr(seals_arg, \(aqS\(aq) != NULL)
 423             seals |= F_SEAL_SEAL;
 424
 425         if (fcntl(fd, F_ADD_SEALS, seals) == \-1)
 426             errExit("fcntl");
 427     }
 428
 429     /* Keep running, so that the file created by memfd_create()
 430        continues to exist */
 431
 432     pause();
 433
 434     exit(EXIT_SUCCESS);
 435 }
 436 .fi
 437 .SS Program source: t_get_seals.c
 438 \&
 439 .nf
 440 #include <sys/memfd.h>
 441 #include <fcntl.h>
 442 #include <unistd.h>
 443 #include <stdlib.h>
 444 #include <string.h>
 445 #include <stdio.h>
 446
 447 #define errExit(msg)    do { perror(msg); exit(EXIT_FAILURE); \\
 448                         } while (0)
 449
 450 int
 451 main(int argc, char *argv[])
 452 {
 453     int fd;
 454     unsigned int seals;
 455
 456     if (argc != 2) {
 457         fprintf(stderr, "%s /proc/PID/fd/FD\\n", argv[0]);
 458         exit(EXIT_FAILURE);
 459     }
 460
 461     fd = open(argv[1], O_RDWR);
 462     if (fd == \-1)
 463         errExit("open");
 464
 465     seals = fcntl(fd, F_GET_SEALS);
 466     if (seals == \-1)
 467         errExit("fcntl");
 468
 469     printf("Existing seals:");
 470     if (seals & F_SEAL_SEAL)
 471         printf(" SEAL");
 472     if (seals & F_SEAL_GROW)
 473         printf(" GROW");
 474     if (seals & F_SEAL_WRITE)
 475         printf(" WRITE");
 476     if (seals & F_SEAL_SHRINK)
 477         printf(" SHRINK");
 478     printf("\\n");
 479
 480     /* Code to map the file and access the contents of the
 481        resulting mapping omitted */
 482
 483     exit(EXIT_SUCCESS);
 484 }
 485 .fi
 486 .SH SEE ALSO
 487 .BR fcntl (2),
 488 .BR ftruncate (2),
 489 .BR mmap (2),
 490 .BR shmget (2),
 491 .BR shm_open (3)