man2/execve.2

   1 .\" Copyright (c) 1992 Drew Eckhardt (drew@cs.colorado.edu), March 28, 1992
   2 .\" and Copyright (c) 2006 Michael Kerrisk <mtk.manpages@gmail.com>
   3 .\"
   4 .\" %%%LICENSE_START(VERBATIM)
   5 .\" Permission is granted to make and distribute verbatim copies of this
   6 .\" manual provided the copyright notice and this permission notice are
   7 .\" preserved on all copies.
   8 .\"
   9 .\" Permission is granted to copy and distribute modified versions of this
  10 .\" manual under the conditions for verbatim copying, provided that the
  11 .\" entire resulting derived work is distributed under the terms of a
  12 .\" permission notice identical to this one.
  13 .\"
  14 .\" Since the Linux kernel and libraries are constantly changing, this
  15 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  16 .\" responsibility for errors or omissions, or for damages resulting from
  17 .\" the use of the information contained herein.  The author(s) may not
  18 .\" have taken the same level of care in the production of this manual,
  19 .\" which is licensed free of charge, as they might when working
  20 .\" professionally.
  21 .\"
  22 .\" Formatted or processed versions of this manual, if unaccompanied by
  23 .\" the source, must acknowledge the copyright and authors of this work.
  24 .\" %%%LICENSE_END
  25 .\"
  26 .\" Modified by Michael Haardt <michael@moria.de>
  27 .\" Modified 1993-07-21 by Rik Faith <faith@cs.unc.edu>
  28 .\" Modified 1994-08-21 by Michael Chastain <mec@shell.portal.com>:
  29 .\" Modified 1997-01-31 by Eric S. Raymond <esr@thyrsus.com>
  30 .\" Modified 1999-11-12 by Urs Thuermann <urs@isnogud.escape.de>
  31 .\" Modified 2004-06-23 by Michael Kerrisk <mtk.manpages@gmail.com>
  32 .\" 2006-09-04 Michael Kerrisk <mtk.manpages@gmail.com>
  33 .\"     Added list of process attributes that are not preserved on exec().
  34 .\" 2007-09-14 Ollie Wild <aaw@google.com>, mtk
  35 .\"     Add text describing limits on command-line arguments + environment
  36 .\"
  37 .TH EXECVE 2 2018-04-30 "Linux" "Linux Programmer's Manual"
  38 .SH NAME
  39 execve \- execute program
  40 .SH SYNOPSIS
  41 .B #include <unistd.h>
  42 .PP
  43 .BI "int execve(const char *" pathname ", char *const " argv "[], "
  44 .br
  45 .BI "           char *const " envp []);
  46 .SH DESCRIPTION
  47 .BR execve ()
  48 executes the program referred to by \fIpathname\fP.
  49 This causes the program that is currently being run by the calling process
  50 to be replaced with a new program, with newly initialized stack, heap,
  51 and (initialized and uninitialized) data segments.
  52 .PP
  53 \fIpathname\fP must be either a binary executable, or a script
  54 starting with a line of the form:
  55 .PP
  56 .in +4n
  57 .EX
  58 \fB#!\fP \fIinterpreter \fP[optional-arg]
  59 .EE
  60 .in
  61 .PP
  62 For details of the latter case, see "Interpreter scripts" below.
  63 .PP
  64 \fIargv\fP is an array of argument strings passed to the new program.
  65 By convention, the first of these strings (i.e.,
  66 .IR argv[0] )
  67 should contain the filename associated with the file being executed.
  68 \fIenvp\fP is an array of strings, conventionally of the form
  69 \fBkey=value\fP, which are passed as environment to the new program.
  70 The \fIargv\fP and \fIenvp\fP arrays must each include a null pointer
  71 at the end of the array.
  72 .PP
  73 The argument vector and environment can be accessed by the
  74 called program's main function, when it is defined as:
  75 .PP
  76 .in +4n
  77 .EX
  78 int main(int argc, char *argv[], char *envp[])
  79 .EE
  80 .in
  81 .PP
  82 Note, however, that the use of a third argument to the main function
  83 is not specified in POSIX.1;
  84 according to POSIX.1,
  85 the environment should be accessed via the external variable
  86 .BR environ (7).
  87 .PP
  88 .BR execve ()
  89 does not return on success, and the text, initialized data,
  90 uninitialized data (bss), and stack of the calling process are overwritten
  91 according to the contents of the newly loaded program.
  92 .PP
  93 If the current program is being ptraced, a \fBSIGTRAP\fP signal is sent to it
  94 after a successful
  95 .BR execve ().
  96 .PP
  97 If the set-user-ID bit is set on the program file referred to by
  98 \fIpathname\fP,
  99 then the effective user ID of the calling process is changed
 100 to that of the owner of the program file.
 101 Similarly, when the set-group-ID
 102 bit of the program file is set the effective group ID of the calling
 103 process is set to the group of the program file.
 104 .PP
 105 The aforementioned transformations of the effective IDs are
 106 .I not
 107 performed (i.e., the set-user-ID and set-group-ID bits are ignored)
 108 if any of the following is true:
 109 .IP * 3
 110 the
 111 .I no_new_privs
 112 attribute is set for the calling thread (see
 113 .BR prctl (2));
 114 .IP *
 115 the underlying filesystem is mounted
 116 .I nosuid
 117 (the
 118 .B MS_NOSUID
 119 flag for
 120 .BR mount (2));
 121 or
 122 .IP *
 123 the calling process is being ptraced.
 124 .PP
 125 The capabilities of the program file (see
 126 .BR capabilities (7))
 127 are also ignored if any of the above are true.
 128 .PP
 129 The effective user ID of the process is copied to the saved set-user-ID;
 130 similarly, the effective group ID is copied to the saved set-group-ID.
 131 This copying takes place after any effective ID changes that occur
 132 because of the set-user-ID and set-group-ID mode bits.
 133 .PP
 134 The process's real UID and real GID, as well its supplementary group IDs,
 135 are unchanged by a call to
 136 .BR execve ().
 137 .PP
 138 If the executable is an a.out dynamically linked
 139 binary executable containing
 140 shared-library stubs, the Linux dynamic linker
 141 .BR ld.so (8)
 142 is called at the start of execution to bring
 143 needed shared objects into memory
 144 and link the executable with them.
 145 .PP
 146 If the executable is a dynamically linked ELF executable, the
 147 interpreter named in the PT_INTERP segment is used to load the needed
 148 shared objects.
 149 This interpreter is typically
 150 .I /lib/ld-linux.so.2
 151 for binaries linked with glibc (see
 152 .BR ld-linux.so (8)).
 153 .PP
 154 All process attributes are preserved during an
 155 .BR execve (),
 156 except the following:
 157 .IP * 3
 158 The dispositions of any signals that are being caught are
 159 reset to the default
 160 .RB ( signal (7)).
 161 .IP *
 162 Any alternate signal stack is not preserved
 163 .RB ( sigaltstack (2)).
 164 .IP *
 165 Memory mappings are not preserved
 166 .RB ( mmap (2)).
 167 .IP *
 168 Attached System\ V shared memory segments are detached
 169 .RB ( shmat (2)).
 170 .IP *
 171 POSIX shared memory regions are unmapped
 172 .RB ( shm_open (3)).
 173 .IP *
 174 Open POSIX message queue descriptors are closed
 175 .RB ( mq_overview (7)).
 176 .IP *
 177 Any open POSIX named semaphores are closed
 178 .RB ( sem_overview (7)).
 179 .IP *
 180 POSIX timers are not preserved
 181 .RB ( timer_create (2)).
 182 .IP *
 183 Any open directory streams are closed
 184 .RB ( opendir (3)).
 185 .IP *
 186 Memory locks are not preserved
 187 .RB ( mlock (2),
 188 .BR mlockall (2)).
 189 .IP *
 190 Exit handlers are not preserved
 191 .RB ( atexit (3),
 192 .BR on_exit (3)).
 193 .IP *
 194 The floating-point environment is reset to the default (see
 195 .BR fenv (3)).
 196 .PP
 197 The process attributes in the preceding list are all specified
 198 in POSIX.1.
 199 The following Linux-specific process attributes are also
 200 not preserved during an
 201 .BR execve ():
 202 .IP * 3
 203 The
 204 .BR prctl (2)
 205 .B PR_SET_DUMPABLE
 206 flag is set,
 207 unless a set-user-ID or set-group ID program is being executed,
 208 in which case it is cleared.
 209 .IP *
 210 The
 211 .BR prctl (2)
 212 .B PR_SET_KEEPCAPS
 213 flag is cleared.
 214 .IP *
 215 (Since Linux 2.4.36 / 2.6.23)
 216 If a set-user-ID or set-group-ID program is being executed,
 217 then the parent death signal set by
 218 .BR prctl (2)
 219 .B PR_SET_PDEATHSIG
 220 flag is cleared.
 221 .IP *
 222 The process name, as set by
 223 .BR prctl (2)
 224 .B PR_SET_NAME
 225 (and displayed by
 226 .IR "ps\ \-o comm" ),
 227 is reset to the name of the new executable file.
 228 .IP *
 229 The
 230 .B SECBIT_KEEP_CAPS
 231 .I securebits
 232 flag is cleared.
 233 See
 234 .BR capabilities (7).
 235 .IP *
 236 The termination signal is reset to
 237 .B SIGCHLD
 238 (see
 239 .BR clone (2)).
 240 .IP *
 241 The file descriptor table is unshared, undoing the effect of the
 242 .B CLONE_FILES
 243 flag of
 244 .BR clone (2).
 245 .PP
 246 Note the following further points:
 247 .IP * 3
 248 All threads other than the calling thread are destroyed during an
 249 .BR execve ().
 250 Mutexes, condition variables, and other pthreads objects are not preserved.
 251 .IP *
 252 The equivalent of \fIsetlocale(LC_ALL, "C")\fP
 253 is executed at program start-up.
 254 .IP *
 255 POSIX.1 specifies that the dispositions of any signals that
 256 are ignored or set to the default are left unchanged.
 257 POSIX.1 specifies one exception: if
 258 .B SIGCHLD
 259 is being ignored,
 260 then an implementation may leave the disposition unchanged or
 261 reset it to the default; Linux does the former.
 262 .IP *
 263 Any outstanding asynchronous I/O operations are canceled
 264 .RB ( aio_read (3),
 265 .BR aio_write (3)).
 266 .IP *
 267 For the handling of capabilities during
 268 .BR execve (),
 269 see
 270 .BR capabilities (7).
 271 .IP *
 272 By default, file descriptors remain open across an
 273 .BR execve ().
 274 File descriptors that are marked close-on-exec are closed;
 275 see the description of
 276 .B FD_CLOEXEC
 277 in
 278 .BR fcntl (2).
 279 (If a file descriptor is closed, this will cause the release
 280 of all record locks obtained on the underlying file by this process.
 281 See
 282 .BR fcntl (2)
 283 for details.)
 284 POSIX.1 says that if file descriptors 0, 1, and 2 would
 285 otherwise be closed after a successful
 286 .BR execve (),
 287 and the process would gain privilege because the set-user-ID or
 288 set-group_ID mode bit was set on the executed file,
 289 then the system may open an unspecified file for each of these
 290 file descriptors.
 291 As a general principle, no portable program, whether privileged or not,
 292 can assume that these three file descriptors will remain
 293 closed across an
 294 .BR execve ().
 295 .\" On Linux it appears that these file descriptors are
 296 .\" always open after an execve(), and it looks like
 297 .\" Solaris 8 and FreeBSD 6.1 are the same. -- mtk, 30 Apr 2007
 298 .SS Interpreter scripts
 299 An interpreter script is a text file that has execute
 300 permission enabled and whose first line is of the form:
 301 .PP
 302 .in +4n
 303 .EX
 304 \fB#!\fP \fIinterpreter \fP[optional-arg]
 305 .EE
 306 .in
 307 .PP
 308 The
 309 .I interpreter
 310 must be a valid pathname for an executable file.
 311 If the
 312 .I pathname
 313 argument of
 314 .BR execve ()
 315 specifies an interpreter script, then
 316 .I interpreter
 317 will be invoked with the following arguments:
 318 .PP
 319 .in +4n
 320 .EX
 321 \fIinterpreter\fP [optional-arg] \fIpathname\fP arg...
 322 .EE
 323 .in
 324 .PP
 325 where
 326 .I arg...
 327 is the series of words pointed to by the
 328 .I argv
 329 argument of
 330 .BR execve (),
 331 starting at
 332 .IR argv [1].
 333 .PP
 334 For portable use,
 335 .I optional-arg
 336 should either be absent, or be specified as a single word (i.e., it
 337 should not contain white space); see NOTES below.
 338 .PP
 339 Since Linux 2.6.28,
 340 .\" commit bf2a9a39639b8b51377905397a5005f444e9a892
 341 the kernel permits the interpreter of a script to itself be a script.
 342 This permission is recursive, up to a limit of four recursions,
 343 so that the interpreter may be a script which is interpreted by a script,
 344 and so on.
 345 .SS Limits on size of arguments and environment
 346 Most UNIX implementations impose some limit on the total size
 347 of the command-line argument
 348 .RI ( argv )
 349 and environment
 350 .RI ( envp )
 351 strings that may be passed to a new program.
 352 POSIX.1 allows an implementation to advertise this limit using the
 353 .B ARG_MAX
 354 constant (either defined in
 355 .I <limits.h>
 356 or available at run time using the call
 357 .IR "sysconf(_SC_ARG_MAX)" ).
 358 .PP
 359 On Linux prior to kernel 2.6.23, the memory used to store the
 360 environment and argument strings was limited to 32 pages
 361 (defined by the kernel constant
 362 .BR MAX_ARG_PAGES ).
 363 On architectures with a 4-kB page size,
 364 this yields a maximum size of 128\ kB.
 365 .PP
 366 On kernel 2.6.23 and later, most architectures support a size limit
 367 derived from the soft
 368 .B RLIMIT_STACK
 369 resource limit (see
 370 .BR getrlimit (2))
 371 that is in force at the time of the
 372 .BR execve ()
 373 call.
 374 (Architectures with no memory management unit are excepted:
 375 they maintain the limit that was in effect before kernel 2.6.23.)
 376 This change allows programs to have a much larger
 377 argument and/or environment list.
 378 .\" For some background on the changes to ARG_MAX in kernels 2.6.23 and
 379 .\" 2.6.25, see:
 380 .\"     http://sourceware.org/bugzilla/show_bug.cgi?id=5786
 381 .\"     http://bugzilla.kernel.org/show_bug.cgi?id=10095
 382 .\"     http://thread.gmane.org/gmane.linux.kernel/646709/focus=648101,
 383 .\"     checked into 2.6.25 as commit a64e715fc74b1a7dcc5944f848acc38b2c4d4ee2.
 384 For these architectures, the total size is limited to 1/4 of the allowed
 385 stack size.
 386 (Imposing the 1/4-limit
 387 ensures that the new program always has some stack space.)
 388 .\" Ollie: That doesn't include the lists of pointers, though,
 389 .\" so the actual usage is a bit higher (1 pointer per argument).
 390 Additionally, the total size is limited to 3/4 of the value
 391 of the kernel constant
 392 .B _STK_LIM
 393 (8 Mibibytes).
 394 Since Linux 2.6.25,
 395 the kernel places a floor of 32 pages on this size limit,
 396 so that, even when
 397 .BR RLIMIT_STACK
 398 is set very low,
 399 applications are guaranteed to have at least as much argument and
 400 environment space as was provided by Linux 2.6.23 and earlier.
 401 (This guarantee was not provided in Linux 2.6.23 and 2.6.24.)
 402 Additionally, the limit per string is 32 pages (the kernel constant
 403 .BR MAX_ARG_STRLEN ),
 404 and the maximum number of strings is 0x7FFFFFFF.
 405 .SH RETURN VALUE
 406 On success,
 407 .BR execve ()
 408 does not return, on error \-1 is returned, and
 409 .I errno
 410 is set appropriately.
 411 .SH ERRORS
 412 .TP
 413 .B E2BIG
 414 The total number of bytes in the environment
 415 .RI ( envp )
 416 and argument list
 417 .RI ( argv )
 418 is too large.
 419 .TP
 420 .B EACCES
 421 Search permission is denied on a component of the path prefix of
 422 .I pathname
 423 or the name of a script interpreter.
 424 (See also
 425 .BR path_resolution (7).)
 426 .TP
 427 .B EACCES
 428 The file or a script interpreter is not a regular file.
 429 .TP
 430 .B EACCES
 431 Execute permission is denied for the file or a script or ELF interpreter.
 432 .TP
 433 .B EACCES
 434 The filesystem is mounted
 435 .IR noexec .
 436 .TP
 437 .BR EAGAIN " (since Linux 3.1)"
 438 .\" commit 72fa59970f8698023045ab0713d66f3f4f96945c
 439 Having changed its real UID using one of the
 440 .BR set*uid ()
 441 calls, the caller was\(emand is now still\(emabove its
 442 .BR RLIMIT_NPROC
 443 resource limit (see
 444 .BR setrlimit (2)).
 445 For a more detailed explanation of this error, see NOTES.
 446 .TP
 447 .B EFAULT
 448 .I pathname
 449 or one of the pointers in the vectors
 450 .I argv
 451 or
 452 .I envp
 453 points outside your accessible address space.
 454 .TP
 455 .B EINVAL
 456 An ELF executable had more than one PT_INTERP segment (i.e., tried to
 457 name more than one interpreter).
 458 .TP
 459 .B EIO
 460 An I/O error occurred.
 461 .TP
 462 .B EISDIR
 463 An ELF interpreter was a directory.
 464 .TP
 465 .B ELIBBAD
 466 An ELF interpreter was not in a recognized format.
 467 .TP
 468 .B ELOOP
 469 Too many symbolic links were encountered in resolving
 470 .I pathname
 471 or the name of a script or ELF interpreter.
 472 .TP
 473 .B ELOOP
 474 The maximum recursion limit was reached during recursive script
 475 interpretation (see "Interpreter scripts", above).
 476 Before Linux 3.8,
 477 .\" commit d740269867021faf4ce38a449353d2b986c34a67
 478 the error produced for this case was
 479 .BR ENOEXEC .
 480 .TP
 481 .B EMFILE
 482 The per-process limit on the number of open file descriptors has been reached.
 483 .TP
 484 .B ENAMETOOLONG
 485 .I pathname
 486 is too long.
 487 .TP
 488 .B ENFILE
 489 The system-wide limit on the total number of open files has been reached.
 490 .TP
 491 .B ENOENT
 492 The file
 493 .I pathname
 494 or a script or ELF interpreter does not exist, or a shared library
 495 .\" FIXME but see http://sourceware.org/bugzilla/show_bug.cgi?id=12241
 496 needed for the file or interpreter cannot be found.
 497 .TP
 498 .B ENOEXEC
 499 An executable is not in a recognized format, is for the wrong
 500 architecture, or has some other format error that means it cannot be
 501 executed.
 502 .TP
 503 .B ENOMEM
 504 Insufficient kernel memory was available.
 505 .TP
 506 .B ENOTDIR
 507 A component of the path prefix of
 508 .I pathname
 509 or a script or ELF interpreter is not a directory.
 510 .TP
 511 .B EPERM
 512 The filesystem is mounted
 513 .IR nosuid ,
 514 the user is not the superuser,
 515 and the file has the set-user-ID or set-group-ID bit set.
 516 .TP
 517 .B EPERM
 518 The process is being traced, the user is not the superuser and the
 519 file has the set-user-ID or set-group-ID bit set.
 520 .TP
 521 .B EPERM
 522 A "capability-dumb" applications would not obtain the full set of
 523 permitted capabilities granted by the executable file.
 524 See
 525 .BR capabilities (7).
 526 .TP
 527 .B ETXTBSY
 528 The specified executable was open for writing by one or more processes.
 529 .SH CONFORMING TO
 530 POSIX.1-2001, POSIX.1-2008, SVr4, 4.3BSD.
 531 POSIX does not document the #! behavior, but it exists
 532 (with some variations) on other UNIX systems.
 533 .\" SVr4 documents additional error
 534 .\" conditions EAGAIN, EINTR, ELIBACC, ENOLINK, EMULTIHOP; POSIX does not
 535 .\" document ETXTBSY, EPERM, EFAULT, ELOOP, EIO, ENFILE, EMFILE, EINVAL,
 536 .\" EISDIR or ELIBBAD error conditions.
 537 .SH NOTES
 538 One sometimes sees
 539 .BR execve ()
 540 (and the related functions described in
 541 .BR exec (3))
 542 described as "executing a
 543 .I new
 544 process" (or similar).
 545 This is a highly misleading description:
 546 there is no new process;
 547 many attributes of the calling process remain unchanged
 548 (in particular, its PID).
 549 All that
 550 .BR execve (2)
 551 does is arrange for an existing process (the calling process)
 552 to execute a new program.
 553 .PP
 554 Set-user-ID and set-group-ID processes can not be
 555 .BR ptrace (2)d.
 556 .PP
 557 The result of mounting a filesystem
 558 .I nosuid
 559 varies across Linux kernel versions:
 560 some will refuse execution of set-user-ID and set-group-ID
 561 executables when this would
 562 give the user powers she did not have already (and return
 563 .BR EPERM ),
 564 some will just ignore the set-user-ID and set-group-ID bits and
 565 .BR exec ()
 566 successfully.
 567 .PP
 568 On Linux,
 569 .I argv
 570 and
 571 .I envp
 572 can be specified as NULL.
 573 In both cases, this has the same effect as specifying the argument
 574 as a pointer to a list containing a single null pointer.
 575 .B "Do not take advantage of this nonstandard and nonportable misfeature!"
 576 On many other UNIX systems, specifying
 577 .I argv
 578 as NULL will result in an error
 579 .RB ( EFAULT ).
 580 .I Some
 581 other UNIX systems treat the
 582 .I envp==NULL
 583 case the same as Linux.
 584 .\" e.g., EFAULT on Solaris 8 and FreeBSD 6.1; but
 585 .\" HP-UX 11 is like Linux -- mtk, Apr 2007
 586 .\" Bug filed 30 Apr 2007: http://bugzilla.kernel.org/show_bug.cgi?id=8408
 587 .\" Bug rejected (because fix would constitute an ABI change).
 588 .\"
 589 .PP
 590 POSIX.1 says that values returned by
 591 .BR sysconf (3)
 592 should be invariant over the lifetime of a process.
 593 However, since Linux 2.6.23, if the
 594 .BR RLIMIT_STACK
 595 resource limit changes, then the value reported by
 596 .B _SC_ARG_MAX
 597 will also change,
 598 to reflect the fact that the limit on space for holding
 599 command-line arguments and environment variables has changed.
 600 .PP
 601 In most cases where
 602 .BR execve ()
 603 fails, control returns to the original executable image,
 604 and the caller of
 605 .BR execve ()
 606 can then handle the error.
 607 However, in (rare) cases (typically caused by resource exhaustion),
 608 failure may occur past the point of no return:
 609 the original executable image has been torn down,
 610 but the new image could not be completely built.
 611 In such cases, the kernel kills the process with a
 612 .BR SIGKILL
 613 signal.
 614 .\"
 615 .SS Interpreter scripts
 616 A maximum line length of 127 characters is allowed for the first line in
 617 an interpreter script.
 618 .PP
 619 The semantics of the
 620 .I optional-arg
 621 argument of an interpreter script vary across implementations.
 622 On Linux, the entire string following the
 623 .I interpreter
 624 name is passed as a single argument to the interpreter,
 625 and this string can include white space.
 626 However, behavior differs on some other systems.
 627 Some systems
 628 .\" e.g., Solaris 8
 629 use the first white space to terminate
 630 .IR optional-arg .
 631 On some systems,
 632 .\" e.g., FreeBSD before 6.0, but not FreeBSD 6.0 onward
 633 an interpreter script can have multiple arguments,
 634 and white spaces in
 635 .I optional-arg
 636 are used to delimit the arguments.
 637 .PP
 638 Linux ignores the set-user-ID and set-group-ID bits on scripts.
 639 .\"
 640 .\" .SH BUGS
 641 .\" Some Linux versions have failed to check permissions on ELF
 642 .\" interpreters.  This is a security hole, because it allows users to
 643 .\" open any file, such as a rewinding tape device, for reading.  Some
 644 .\" Linux versions have also had other security holes in
 645 .\" .BR execve ()
 646 .\" that could be exploited for denial of service by a suitably crafted
 647 .\" ELF binary. There are no known problems with 2.0.34 or 2.2.15.
 648 .SS execve() and EAGAIN
 649 A more detailed explanation of the
 650 .BR EAGAIN
 651 error that can occur (since Linux 3.1) when calling
 652 .BR execve ()
 653 is as follows.
 654 .PP
 655 The
 656 .BR EAGAIN
 657 error can occur when a
 658 .I preceding
 659 call to
 660 .BR setuid (2),
 661 .BR setreuid (2),
 662 or
 663 .BR setresuid (2)
 664 caused the real user ID of the process to change,
 665 and that change caused the process to exceed its
 666 .BR RLIMIT_NPROC
 667 resource limit (i.e., the number of processes belonging
 668 to the new real UID exceeds the resource limit).
 669 From Linux 2.6.0 to 3.0, this caused the
 670 .BR set*uid ()
 671 call to fail.
 672 (Prior to 2.6,
 673 .\" commit 909cc4ae86f3380152a18e2a3c44523893ee11c4
 674 the resource limit was not imposed on processes that
 675 changed their user IDs.)
 676 .PP
 677 Since Linux 3.1, the scenario just described no longer causes the
 678 .BR set*uid ()
 679 call to fail,
 680 because it too often led to security holes where buggy applications
 681 didn't check the return status and assumed
 682 that\(emif the caller had root privileges\(emthe call would always succeed.
 683 Instead, the
 684 .BR set*uid ()
 685 calls now successfully change the real UID,
 686 but the kernel sets an internal flag, named
 687 .BR PF_NPROC_EXCEEDED ,
 688 to note that the
 689 .BR RLIMIT_NPROC
 690 resource limit has been exceeded.
 691 If the
 692 .BR PF_NPROC_EXCEEDED
 693 flag is set and the resource limit is still
 694 exceeded at the time of a subsequent
 695 .BR execve ()
 696 call, that call fails with the error
 697 .BR EAGAIN .
 698 This kernel logic ensures that the
 699 .BR RLIMIT_NPROC
 700 resource limit is still enforced for the
 701 common privileged daemon workflow\(emnamely,
 702 .BR fork (2)
 703 +
 704 .BR set*uid ()
 705 +
 706 .BR execve ().
 707 .PP
 708 If the resource limit was not still exceeded at the time of the
 709 .BR execve ()
 710 call
 711 (because other processes belonging to this real UID terminated between the
 712 .BR set*uid ()
 713 call and the
 714 .BR execve ()
 715 call), then the
 716 .BR execve ()
 717 call succeeds and the kernel clears the
 718 .BR PF_NPROC_EXCEEDED
 719 process flag.
 720 The flag is also cleared if a subsequent call to
 721 .BR fork (2)
 722 by this process succeeds.
 723 .SS Historical
 724 With UNIX\ V6, the argument list of an
 725 .BR exec ()
 726 call was ended by 0,
 727 while the argument list of
 728 .I main
 729 was ended by \-1.
 730 Thus, this argument list was not directly usable in a further
 731 .BR exec ()
 732 call.
 733 Since UNIX\ V7, both are NULL.
 734 .\"
 735 .\" .SH BUGS
 736 .\" Some Linux versions have failed to check permissions on ELF
 737 .\" interpreters.  This is a security hole, because it allows users to
 738 .\" open any file, such as a rewinding tape device, for reading.  Some
 739 .\" Linux versions have also had other security holes in
 740 .\" .BR execve ()
 741 .\" that could be exploited for denial of service by a suitably crafted
 742 .\" ELF binary. There are no known problems with 2.0.34 or 2.2.15.
 743 .SH EXAMPLE
 744 The following program is designed to be execed by the second program below.
 745 It just echoes its command-line arguments, one per line.
 746 .PP
 747 .in +4n
 748 .EX
 749 /* myecho.c */
 750
 751 #include <stdio.h>
 752 #include <stdlib.h>
 753
 754 int
 755 main(int argc, char *argv[])
 756 {
 757     int j;
 758
 759     for (j = 0; j < argc; j++)
 760         printf("argv[%d]: %s\en", j, argv[j]);
 761
 762     exit(EXIT_SUCCESS);
 763 }
 764 .EE
 765 .in
 766 .PP
 767 This program can be used to exec the program named in its command-line
 768 argument:
 769 .PP
 770 .in +4n
 771 .EX
 772 /* execve.c */
 773
 774 #include <stdio.h>
 775 #include <stdlib.h>
 776 #include <unistd.h>
 777
 778 int
 779 main(int argc, char *argv[])
 780 {
 781     char *newargv[] = { NULL, "hello", "world", NULL };
 782     char *newenviron[] = { NULL };
 783
 784     if (argc != 2) {
 785         fprintf(stderr, "Usage: %s <file\-to\-exec>\en", argv[0]);
 786         exit(EXIT_FAILURE);
 787     }
 788
 789     newargv[0] = argv[1];
 790
 791     execve(argv[1], newargv, newenviron);
 792     perror("execve");   /* execve() returns only on error */
 793     exit(EXIT_FAILURE);
 794 }
 795 .EE
 796 .in
 797 .PP
 798 We can use the second program to exec the first as follows:
 799 .PP
 800 .in +4n
 801 .EX
 802 .RB "$" " cc myecho.c \-o myecho"
 803 .RB "$" " cc execve.c \-o execve"
 804 .RB "$" " ./execve ./myecho"
 805 argv[0]: ./myecho
 806 argv[1]: hello
 807 argv[2]: world
 808 .EE
 809 .in
 810 .PP
 811 We can also use these programs to demonstrate the use of a script
 812 interpreter.
 813 To do this we create a script whose "interpreter" is our
 814 .I myecho
 815 program:
 816 .PP
 817 .in +4n
 818 .EX
 819 .RB "$" " cat > script"
 820 .B #!./myecho script-arg
 821 .B ^D
 822 .RB "$" " chmod +x script"
 823 .EE
 824 .in
 825 .PP
 826 We can then use our program to exec the script:
 827 .PP
 828 .in +4n
 829 .EX
 830 .RB "$" " ./execve ./script"
 831 argv[0]: ./myecho
 832 argv[1]: script-arg
 833 argv[2]: ./script
 834 argv[3]: hello
 835 argv[4]: world
 836 .EE
 837 .in
 838 .SH SEE ALSO
 839 .BR chmod (2),
 840 .BR execveat (2),
 841 .BR fork (2),
 842 .BR get_robust_list (2),
 843 .BR ptrace (2),
 844 .BR exec (3),
 845 .BR fexecve (3),
 846 .BR getopt (3),
 847 .BR system (3),
 848 .BR credentials (7),
 849 .BR environ (7),
 850 .BR path_resolution (7),
 851 .BR ld.so (8)