execve.2: wfix

[thirdparty/man-pages.git] / man2 / execve.2

.\" Copyright (c) 1992 Drew Eckhardt (drew@cs.colorado.edu), March 28, 1992
.\" and Copyright (c) 2006 Michael Kerrisk <mtk.manpages@gmail.com>
.\"
.\" %%%LICENSE_START(VERBATIM)
.\" Permission is granted to make and distribute verbatim copies of this
.\" manual provided the copyright notice and this permission notice are
.\" preserved on all copies.
.\"
.\" Permission is granted to copy and distribute modified versions of this
.\" manual under the conditions for verbatim copying, provided that the
.\" entire resulting derived work is distributed under the terms of a
.\" permission notice identical to this one.
.\"
.\" Since the Linux kernel and libraries are constantly changing, this
.\" manual page may be incorrect or out-of-date.  The author(s) assume no
.\" responsibility for errors or omissions, or for damages resulting from
.\" the use of the information contained herein.  The author(s) may not
.\" have taken the same level of care in the production of this manual,
.\" which is licensed free of charge, as they might when working
.\" professionally.
.\"
.\" Formatted or processed versions of this manual, if unaccompanied by
.\" the source, must acknowledge the copyright and authors of this work.
.\" %%%LICENSE_END
.\"
.\" Modified by Michael Haardt <michael@moria.de>
.\" Modified 1993-07-21 by Rik Faith <faith@cs.unc.edu>
.\" Modified 1994-08-21 by Michael Chastain <mec@shell.portal.com>:
.\" Modified 1997-01-31 by Eric S. Raymond <esr@thyrsus.com>
.\" Modified 1999-11-12 by Urs Thuermann <urs@isnogud.escape.de>
.\" Modified 2004-06-23 by Michael Kerrisk <mtk.manpages@gmail.com>
.\" 2006-09-04 Michael Kerrisk <mtk.manpages@gmail.com>
.\"     Added list of process attributes that are not preserved on exec().
.\" 2007-09-14 Ollie Wild <aaw@google.com>, mtk
.\"     Add text describing limits on command-line arguments + environment
.\"
.TH EXECVE 2 2016-10-08 "Linux" "Linux Programmer's Manual"
.SH NAME
execve \- execute program
.SH SYNOPSIS
.B #include <unistd.h>
.sp
.BI "int execve(const char *" filename ", char *const " argv "[], "
.br
.BI "           char *const " envp []);
.SH DESCRIPTION
.BR execve ()
executes the program pointed to by \fIfilename\fP.
\fIfilename\fP must be either a binary executable, or a script
starting with a line of the form:

.in +4n
.nf
\fB#!\fP \fIinterpreter \fP[optional-arg]
.fi
.in

For details of the latter case, see "Interpreter scripts" below.

\fIargv\fP is an array of argument strings passed to the new program.
By convention, the first of these strings (i.e.,
.IR argv[0] )
should contain the filename associated with the file being executed.
\fIenvp\fP is an array of strings, conventionally of the form
\fBkey=value\fP, which are passed as environment to the new program.
The \fIargv\fP and \fIenvp\fP arrays must each include a null pointer
at the end of the array.

The argument vector and environment can be accessed by the
called program's main function, when it is defined as:

.in +4n
.nf
int main(int argc, char *argv[], char *envp[])
.fi
.in

Note, however, that the use of a third argument to the main function
is not specified in POSIX.1;
according to POSIX.1,
the environment should be accessed via the external variable
.BR environ (7).

.BR execve ()
does not return on success, and the text, initialized data,
uninitialized data (bss), and stack of the calling process are overwritten
according to the contents of the newly loaded program.

If the current program is being ptraced, a \fBSIGTRAP\fP signal is sent to it
after a successful
.BR execve ().

If the set-user-ID bit is set on the program file pointed to by
\fIfilename\fP,
then the effective user ID of the calling process is changed
to that of the owner of the program file.
Similarly, when the set-group-ID
bit of the program file is set the effective group ID of the calling
process is set to the group of the program file.
The aforementioned transformations of the effective IDs are
.I not
performed (i.e., the set-user-ID and set-group-ID bits are ignored)
if any of the following is true:
.IP * 3
the
.I no_new_privs
attribute is set for the calling thread (see
.BR prctl (2));
.IP *
the underlying filesystem is mounted
.I nosuid
(the
.B MS_NOSUID
flag for
.BR mount (2));
or
.IP *
the calling process is being ptraced.
.PP
The effective user ID of the process is copied to the saved set-user-ID;
similarly, the effective group ID is copied to the saved set-group-ID.
This copying takes place after any effective ID changes that occur
because of the set-user-ID and set-group-ID mode bits.

The process's read UID and real GID, as well its supplementary group IDs,
are unchanged by a call to
.BR execve (2).

If the executable is an a.out dynamically linked
binary executable containing
shared-library stubs, the Linux dynamic linker
.BR ld.so (8)
is called at the start of execution to bring
needed shared objects into memory
and link the executable with them.

If the executable is a dynamically linked ELF executable, the
interpreter named in the PT_INTERP segment is used to load the needed
shared objects.
This interpreter is typically
.I /lib/ld-linux.so.2
for binaries linked with glibc (see
.BR ld-linux.so (8)).

All process attributes are preserved during an
.BR execve (),
except the following:
.IP * 3
The dispositions of any signals that are being caught are
reset to the default
.RB ( signal (7)).
.IP *
Any alternate signal stack is not preserved
.RB ( sigaltstack (2)).
.IP *
Memory mappings are not preserved
.RB ( mmap (2)).
.IP *
Attached System\ V shared memory segments are detached
.RB ( shmat (2)).
.IP *
POSIX shared memory regions are unmapped
.RB ( shm_open (3)).
.IP *
Open POSIX message queue descriptors are closed
.RB ( mq_overview (7)).
.IP *
Any open POSIX named semaphores are closed
.RB ( sem_overview (7)).
.IP *
POSIX timers are not preserved
.RB ( timer_create (2)).
.IP *
Any open directory streams are closed
.RB ( opendir (3)).
.IP *
Memory locks are not preserved
.RB ( mlock (2),
.BR mlockall (2)).
.IP *
Exit handlers are not preserved
.RB ( atexit (3),
.BR on_exit (3)).
.IP *
The floating-point environment is reset to the default (see
.BR fenv (3)).
.PP
The process attributes in the preceding list are all specified
in POSIX.1.
The following Linux-specific process attributes are also
not preserved during an
.BR execve ():
.IP * 3
The
.BR prctl (2)
.B PR_SET_DUMPABLE
flag is set,
unless a set-user-ID or set-group ID program is being executed,
in which case it is cleared.
.IP *
The
.BR prctl (2)
.B PR_SET_KEEPCAPS
flag is cleared.
.IP *
(Since Linux 2.4.36 / 2.6.23)
If a set-user-ID or set-group-ID program is being executed,
then the parent death signal set by
.BR prctl (2)
.B PR_SET_PDEATHSIG
flag is cleared.
.IP *
The process name, as set by
.BR prctl (2)
.B PR_SET_NAME
(and displayed by
.IR "ps\ \-o comm" ),
is reset to the name of the new executable file.
.IP *
The
.B SECBIT_KEEP_CAPS
.I securebits
flag is cleared.
See
.BR capabilities (7).
.IP *
The termination signal is reset to
.B SIGCHLD
(see
.BR clone (2)).
.IP *
The file descriptor table is unshared, undoing the effect of the
.B CLONE_FILES
flag of
.BR clone (2).
.PP
Note the following further points:
.IP * 3
All threads other than the calling thread are destroyed during an
.BR execve ().
Mutexes, condition variables, and other pthreads objects are not preserved.
.IP *
The equivalent of \fIsetlocale(LC_ALL, "C")\fP
is executed at program start-up.
.IP *
POSIX.1 specifies that the dispositions of any signals that
are ignored or set to the default are left unchanged.
POSIX.1 specifies one exception: if
.B SIGCHLD
is being ignored,
then an implementation may leave the disposition unchanged or
reset it to the default; Linux does the former.
.IP *
Any outstanding asynchronous I/O operations are canceled
.RB ( aio_read (3),
.BR aio_write (3)).
.IP *
For the handling of capabilities during
.BR execve (),
see
.BR capabilities (7).
.IP *
By default, file descriptors remain open across an
.BR execve ().
File descriptors that are marked close-on-exec are closed;
see the description of
.B FD_CLOEXEC
in
.BR fcntl (2).
(If a file descriptor is closed, this will cause the release
of all record locks obtained on the underlying file by this process.
See
.BR fcntl (2)
for details.)
POSIX.1 says that if file descriptors 0, 1, and 2 would
otherwise be closed after a successful
.BR execve (),
and the process would gain privilege because the set-user-ID or
set-group_ID mode bit was set on the executed file,
then the system may open an unspecified file for each of these
file descriptors.
As a general principle, no portable program, whether privileged or not,
can assume that these three file descriptors will remain
closed across an
.BR execve ().
.\" On Linux it appears that these file descriptors are
.\" always open after an execve(), and it looks like
.\" Solaris 8 and FreeBSD 6.1 are the same. -- mtk, 30 Apr 2007
.SS Interpreter scripts
An interpreter script is a text file that has execute
permission enabled and whose first line is of the form:

.in +4n
.nf
\fB#!\fP \fIinterpreter \fP[optional-arg]
.fi
.in

The
.I interpreter
must be a valid pathname for an executable file.
If the
.I filename
argument of
.BR execve ()
specifies an interpreter script, then
.I interpreter
will be invoked with the following arguments:

.in +4n
.nf
\fIinterpreter\fP [optional-arg] \fIfilename\fP arg...
.fi
.in

where
.I arg...
is the series of words pointed to by the
.I argv
argument of
.BR execve (),
starting at
.IR argv [1].

For portable use,
.I optional-arg
should either be absent, or be specified as a single word (i.e., it
should not contain white space); see NOTES below.

Since Linux 2.6.28,
.\" commit bf2a9a39639b8b51377905397a5005f444e9a892
the kernel permits the interpreter of a script to itself be a script.
This permission is recursive, up to a limit of four recursions,
so that the interpreter may be a script which is interpreted by a script,
and so on.
.SS Limits on size of arguments and environment
Most UNIX implementations impose some limit on the total size
of the command-line argument
.RI ( argv )
and environment
.RI ( envp )
strings that may be passed to a new program.
POSIX.1 allows an implementation to advertise this limit using the
.B ARG_MAX
constant (either defined in
.I <limits.h>
or available at run time using the call
.IR "sysconf(_SC_ARG_MAX)" ).

On Linux prior to kernel 2.6.23, the memory used to store the
environment and argument strings was limited to 32 pages
(defined by the kernel constant
.BR MAX_ARG_PAGES ).
On architectures with a 4-kB page size,
this yields a maximum size of 128 kB.

On kernel 2.6.23 and later, most architectures support a size limit
derived from the soft
.B RLIMIT_STACK
resource limit (see
.BR getrlimit (2))
that is in force at the time of the
.BR execve ()
call.
(Architectures with no memory management unit are excepted:
they maintain the limit that was in effect before kernel 2.6.23.)
This change allows programs to have a much larger
argument and/or environment list.
.\" For some background on the changes to ARG_MAX in kernels 2.6.23 and
.\" 2.6.25, see:
.\"     http://sourceware.org/bugzilla/show_bug.cgi?id=5786
.\"     http://bugzilla.kernel.org/show_bug.cgi?id=10095
.\"     http://thread.gmane.org/gmane.linux.kernel/646709/focus=648101,
.\"     checked into 2.6.25 as commit a64e715fc74b1a7dcc5944f848acc38b2c4d4ee2.
For these architectures, the total size is limited to 1/4 of the allowed
stack size.
(Imposing the 1/4-limit
ensures that the new program always has some stack space.)
.\" Ollie: That doesn't include the lists of pointers, though,
.\" so the actual usage is a bit higher (1 pointer per argument).
Since Linux 2.6.25,
the kernel places a floor of 32 pages on this size limit,
so that, even when
.BR RLIMIT_STACK
is set very low,
applications are guaranteed to have at least as much argument and
environment space as was provided by Linux 2.6.23 and earlier.
(This guarantee was not provided in Linux 2.6.23 and 2.6.24.)
Additionally, the limit per string is 32 pages (the kernel constant
.BR MAX_ARG_STRLEN ),
and the maximum number of strings is 0x7FFFFFFF.
.SH RETURN VALUE
On success,
.BR execve ()
does not return, on error \-1 is returned, and
.I errno
is set appropriately.
.SH ERRORS
.TP
.B E2BIG
The total number of bytes in the environment
.RI ( envp )
and argument list
.RI ( argv )
is too large.
.TP
.B EACCES
Search permission is denied on a component of the path prefix of
.I filename
or the name of a script interpreter.
(See also
.BR path_resolution (7).)
.TP
.B EACCES
The file or a script interpreter is not a regular file.
.TP
.B EACCES
Execute permission is denied for the file or a script or ELF interpreter.
.TP
.B EACCES
The filesystem is mounted
.IR noexec .
.TP
.BR EAGAIN " (since Linux 3.1)"
.\" commit 72fa59970f8698023045ab0713d66f3f4f96945c
Having changed its real UID using one of the
.BR set*uid ()
calls, the caller was\(emand is now still\(emabove its
.BR RLIMIT_NPROC
resource limit (see
.BR setrlimit (2)).
For a more detailed explanation of this error, see NOTES.
.TP
.B EFAULT
.I filename
or one of the pointers in the vectors
.I argv
or
.I envp
points outside your accessible address space.
.TP
.B EINVAL
An ELF executable had more than one PT_INTERP segment (i.e., tried to
name more than one interpreter).
.TP
.B EIO
An I/O error occurred.
.TP
.B EISDIR
An ELF interpreter was a directory.
.TP
.B ELIBBAD
An ELF interpreter was not in a recognized format.
.TP
.B ELOOP
Too many symbolic links were encountered in resolving
.I filename
or the name of a script or ELF interpreter.
.TP
.B ELOOP
The maximum recursion limit was reached during recursive script
interpretation (see "Interpreter scripts", above).
Before Linux 3.8,
.\" commit d740269867021faf4ce38a449353d2b986c34a67
the error produced for this case was
.BR ENOEXEC .
.TP
.B EMFILE
The per-process limit on the number of open file descriptors has been reached.
.TP
.B ENAMETOOLONG
.I filename
is too long.
.TP
.B ENFILE
The system-wide limit on the total number of open files has been reached.
.TP
.B ENOENT
The file
.I filename
or a script or ELF interpreter does not exist, or a shared library
.\" FIXME but see http://sourceware.org/bugzilla/show_bug.cgi?id=12241
needed for the file or interpreter cannot be found.
.TP
.B ENOEXEC
An executable is not in a recognized format, is for the wrong
architecture, or has some other format error that means it cannot be
executed.
.TP
.B ENOMEM
Insufficient kernel memory was available.
.TP
.B ENOTDIR
A component of the path prefix of
.I filename
or a script or ELF interpreter is not a directory.
.TP
.B EPERM
The filesystem is mounted
.IR nosuid ,
the user is not the superuser,
and the file has the set-user-ID or set-group-ID bit set.
.TP
.B EPERM
The process is being traced, the user is not the superuser and the
file has the set-user-ID or set-group-ID bit set.
.TP
.B EPERM
A "capability-dumb" applications would not obtain the full set of
permitted capabilities granted by the executable file.
See
.BR capabilities (7).
.TP
.B ETXTBSY
The specified executable was open for writing by one or more processes.
.SH CONFORMING TO
POSIX.1-2001, POSIX.1-2008, SVr4, 4.3BSD.
POSIX does not document the #! behavior, but it exists
(with some variations) on other UNIX systems.
.\" SVr4 documents additional error
.\" conditions EAGAIN, EINTR, ELIBACC, ENOLINK, EMULTIHOP; POSIX does not
.\" document ETXTBSY, EPERM, EFAULT, ELOOP, EIO, ENFILE, EMFILE, EINVAL,
.\" EISDIR or ELIBBAD error conditions.
.SH NOTES
Set-user-ID and set-group-ID processes can not be
.BR ptrace (2)d.

The result of mounting a filesystem
.I nosuid
varies across Linux kernel versions:
some will refuse execution of set-user-ID and set-group-ID
executables when this would
give the user powers she did not have already (and return
.BR EPERM ),
some will just ignore the set-user-ID and set-group-ID bits and
.BR exec ()
successfully.

On Linux,
.I argv
and
.I envp
can be specified as NULL.
In both cases, this has the same effect as specifying the argument
as a pointer to a list containing a single null pointer.
.B "Do not take advantage of this nonstandard and nonportable misfeature!"
On many other UNIX systems, specifying
.I argv
as NULL will result in an error
.RB ( EFAULT ).
.I Some
other UNIX systems treat the
.I envp==NULL
case the same as Linux.
.\" e.g., EFAULT on Solaris 8 and FreeBSD 6.1; but
.\" HP-UX 11 is like Linux -- mtk, Apr 2007
.\" Bug filed 30 Apr 2007: http://bugzilla.kernel.org/show_bug.cgi?id=8408
.\" Bug rejected (because fix would constitute an ABI change).
.\"

POSIX.1 says that values returned by
.BR sysconf (3)
should be invariant over the lifetime of a process.
However, since Linux 2.6.23, if the
.BR RLIMIT_STACK
resource limit changes, then the value reported by
.B _SC_ARG_MAX
will also change,
to reflect the fact that the limit on space for holding
command-line arguments and environment variables has changed.

In most cases where
.BR execve ()
fails, control returns to the original executable image,
and the caller of
.BR execve ()
can then handle the error.
However, in (rare) cases (typically caused by resource exhaustion),
failure may occur past the point of no return:
the original executable image has been torn down,
but the new image could not be completely built.
In such cases, the kernel kills the process with a
.BR SIGKILL
signal.
.\"
.SS Interpreter scripts
A maximum line length of 127 characters is allowed for the first line in
an interpreter scripts.

The semantics of the
.I optional-arg
argument of an interpreter script vary across implementations.
On Linux, the entire string following the
.I interpreter
name is passed as a single argument to the interpreter,
and this string can include white space.
However, behavior differs on some other systems.
Some systems
.\" e.g., Solaris 8
use the first white space to terminate
.IR optional-arg .
On some systems,
.\" e.g., FreeBSD before 6.0, but not FreeBSD 6.0 onward
an interpreter script can have multiple arguments,
and white spaces in
.I optional-arg
are used to delimit the arguments.

Linux ignores the set-user-ID and set-group-ID bits on scripts.
.\"
.\" .SH BUGS
.\" Some Linux versions have failed to check permissions on ELF
.\" interpreters.  This is a security hole, because it allows users to
.\" open any file, such as a rewinding tape device, for reading.  Some
.\" Linux versions have also had other security holes in
.\" .BR execve ()
.\" that could be exploited for denial of service by a suitably crafted
.\" ELF binary. There are no known problems with 2.0.34 or 2.2.15.
.SS execve() and EAGAIN
A more detailed explanation of the
.BR EAGAIN
error that can occur (since Linux 3.1) when calling
.BR execve ()
is as follows.

The
.BR EAGAIN
error can occur when a
.I preceding
call to
.BR setuid (2),
.BR setreuid (2),
or
.BR setresuid (2)
caused the real user ID of the process to change,
and that change caused the process to exceed its
.BR RLIMIT_NPROC
resource limit (i.e., the number of processes belonging
to the new real UID exceeds the resource limit).
From Linux 2.6.0 to 3.0, this caused the
.BR set*uid ()
call to fail.
(Prior to 2.6,
.\" commit 909cc4ae86f3380152a18e2a3c44523893ee11c4
the resource limit was not imposed on processes that
changed their user IDs.)

Since Linux 3.1, the scenario just described no longer causes the
.BR set*uid ()
call to fail,
because it too often led to security holes where buggy applications
didn't check the return status and assumed
that\(emif the caller had root privileges\(emthe call would always succeed.
Instead, the
.BR set*uid ()
calls now successfully change the real UID,
but the kernel sets an internal flag, named
.BR PF_NPROC_EXCEEDED ,
to note that the
.BR RLIMIT_NPROC
resource limit has been exceeded.
If the
.BR PF_NPROC_EXCEEDED
flag is set and the resource limit is still
exceeded at the time of a subsequent
.BR execve ()
call, that call fails with the error
.BR EAGAIN .
This kernel logic ensures that the
.BR RLIMIT_NPROC
resource limit is still enforced for the
common privileged daemon workflow\(emnamely,
.BR fork (2)
+
.BR set*uid ()
+
.BR execve ().

If the resource limit was not still exceeded at the time of the
.BR execve ()
call
(because other processes belonging to this real UID terminated between the
.BR set*uid ()
call and the
.BR execve ()
call), then the
.BR execve ()
call succeeds and the kernel clears the
.BR PF_NPROC_EXCEEDED
process flag.
The flag is also cleared if a subsequent call to
.BR fork (2)
by this process succeeds.
.SS Historical
With UNIX\ V6, the argument list of an
.BR exec ()
call was ended by 0,
while the argument list of
.I main
was ended by \-1.
Thus, this argument list was not directly usable in a further
.BR exec ()
call.
Since UNIX\ V7, both are NULL.
.\"
.\" .SH BUGS
.\" Some Linux versions have failed to check permissions on ELF
.\" interpreters.  This is a security hole, because it allows users to
.\" open any file, such as a rewinding tape device, for reading.  Some
.\" Linux versions have also had other security holes in
.\" .BR execve ()
.\" that could be exploited for denial of service by a suitably crafted
.\" ELF binary. There are no known problems with 2.0.34 or 2.2.15.
.SH EXAMPLE
The following program is designed to be execed by the second program below.
It just echoes its command-line arguments, one per line.

.in +4n
.nf
/* myecho.c */

#include <stdio.h>
#include <stdlib.h>

int
main(int argc, char *argv[])
{
    int j;

    for (j = 0; j < argc; j++)
        printf("argv[%d]: %s\\n", j, argv[j]);

    exit(EXIT_SUCCESS);
}
.fi
.in

This program can be used to exec the program named in its command-line
argument:
.in +4n
.nf

/* execve.c */

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>

int
main(int argc, char *argv[])
{
    char *newargv[] = { NULL, "hello", "world", NULL };
    char *newenviron[] = { NULL };

    if (argc != 2) {
        fprintf(stderr, "Usage: %s <file\-to\-exec>\\n", argv[0]);
        exit(EXIT_FAILURE);
    }

    newargv[0] = argv[1];

    execve(argv[1], newargv, newenviron);
    perror("execve");   /* execve() returns only on error */
    exit(EXIT_FAILURE);
}
.fi
.in

We can use the second program to exec the first as follows:

.in +4n
.nf
.RB "$" " cc myecho.c \-o myecho"
.RB "$" " cc execve.c \-o execve"
.RB "$" " ./execve ./myecho"
argv[0]: ./myecho
argv[1]: hello
argv[2]: world
.fi
.in

We can also use these programs to demonstrate the use of a script
interpreter.
To do this we create a script whose "interpreter" is our
.I myecho
program:

.in +4n
.nf
.RB "$" " cat > script"
.B #!./myecho script-arg
.B ^D
.RB "$" " chmod +x script"
.fi
.in

We can then use our program to exec the script:

.in +4n
.nf
.RB "$" " ./execve ./script"
argv[0]: ./myecho
argv[1]: script-arg
argv[2]: ./script
argv[3]: hello
argv[4]: world
.fi
.in
.SH SEE ALSO
.BR chmod (2),
.BR execveat (2),
.BR fork (2),
.BR ptrace (2),
.BR execl (3),
.BR fexecve (3),
.BR getopt (3),
.BR system (3),
.BR credentials (7),
.BR environ (7),
.BR path_resolution (7),
.BR ld.so (8)