man2/unshare.2

   1 .\" Copyright (C) 2006, Janak Desai <janak@us.ibm.com>
   2 .\" and Copyright (C) 2006, Michael Kerrisk <mtk.manpages@gmail.com>
   3 .\" Licensed under the GPL
   4 .\"
   5 .\" Patch Justification:
   6 .\" unshare system call is needed to implement, using PAM,
   7 .\" per-security_context and/or per-user namespace to provide
   8 .\" polyinstantiated directories. Using unshare and bind mounts, a
   9 .\" PAM module can create private namespace with appropriate
  10 .\" directories(based on user's security context) bind mounted on
  11 .\" public directories such as /tmp, thus providing an instance of
  12 .\" /tmp that is based on user's security context. Without the
  13 .\" unshare system call, namespace separation can only be achieved
  14 .\" by clone, which would require porting and maintaining all commands
  15 .\" such as login, and su, that establish a user session.
  16 .\"
  17 .TH UNSHARE 2 2010-10-24 "Linux" "Linux Programmer's Manual"
  18 .SH NAME
  19 unshare \- disassociate parts of the process execution context
  20 .SH SYNOPSIS
  21 .nf
  22 .BR "#define _GNU_SOURCE" "             /* See feature_test_macros(7) */"
  23 .\" Actually _BSD_SOURCE || _SVID_SOURCE
  24 .\" See http://sources.redhat.com/bugzilla/show_bug.cgi?id=4749
  25 .B #include <sched.h>
  26 .sp
  27 .BI "int unshare(int " flags );
  28 .fi
  29 .SH DESCRIPTION
  30 .BR unshare ()
  31 allows a process to disassociate parts of its execution
  32 context that are currently being shared with other processes.
  33 Part of the execution context, such as the mount namespace, is shared
  34 implicitly when a new process is created using
  35 .BR fork (2)
  36 or
  37 .BR vfork (2),
  38 while other parts, such as virtual memory, may be
  39 shared by explicit request when creating a process using
  40 .BR clone (2).
  41
  42 The main use of
  43 .BR unshare ()
  44 is to allow a process to control its
  45 shared execution context without creating a new process.
  46
  47 The
  48 .I flags
  49 argument is a bit mask that specifies which parts of
  50 the execution context should be unshared.
  51 This argument is specified by ORing together zero or more
  52 of the following constants:
  53 .TP
  54 .B CLONE_FILES
  55 Reverse the effect of the
  56 .BR clone (2)
  57 .B CLONE_FILES
  58 flag.
  59 Unshare the file descriptor table, so that the calling process
  60 no longer shares its file descriptors with any other process.
  61 .TP
  62 .B CLONE_FS
  63 Reverse the effect of the
  64 .BR clone (2)
  65 .B CLONE_FS
  66 flag.
  67 Unshare file system attributes, so that the calling process
  68 no longer shares its root directory, current directory,
  69 or umask attributes with any other process.
  70 .BR chroot (2),
  71 .BR chdir (2),
  72 or
  73 .BR umask (2)
  74 .TP
  75 .BR CLONE_NEWIPC " (since Linux 2.6.19)
  76 This flag has the same effect as the
  77 .BR clone (2)
  78 .B CLONE_NEWIPC
  79 flag.
  80 Unshare the System V IPC namespace,
  81 so that the calling process has a private copy of the
  82 System V IPC namespace which is not shared with any other process.
  83 Specifying this flag automatically implies
  84 .BR CLONE_SYSVSEM
  85 as well.
  86 Use of
  87 .BR CLONE_NEWIPC
  88 requires the
  89 .BR CAP_SYS_ADMIN
  90 capability.
  91 .TP
  92 .B CLONE_NEWNS
  93 .\" These flag name are inconsistent:
  94 .\" CLONE_NEWNS does the same thing in clone(), but CLONE_VM,
  95 .\" CLONE_FS, and CLONE_FILES reverse the action of the clone()
  96 .\" flags of the same name.
  97 This flag has the same effect as the
  98 .BR clone (2)
  99 .B CLONE_NEWNS
 100 flag.
 101 Unshare the mount namespace,
 102 so that the calling process has a private copy of
 103 its namespace which is not shared with any other process.
 104 Specifying this flag automatically implies
 105 .B CLONE_FS
 106 as well.
 107 Use of
 108 .BR CLONE_NEWNS
 109 requires the
 110 .BR CAP_SYS_ADMIN
 111 capability.
 112 .TP
 113 .BR CLONE_SYSVSEM " (since Linux 2.6.26)
 114 This flag reverses the effect of the
 115 .BR clone (2)
 116 .B CLONE_SYSVSEM
 117 flag.
 118 Unshare System V semaphore undo values,
 119 so that the calling process has a private copy
 120 which is not shared with any other process.
 121 Use of
 122 .BR CLONE_SYSVSEM
 123 requires the
 124 .BR CAP_SYS_ADMIN
 125 capability.
 126 .\" As at 2.6.16, the following forced implications also apply,
 127 .\" although the relevant flags are not yet implemented.
 128 .\" If CLONE_THREAD is set force CLONE_VM.
 129 .\" If CLONE_VM is set, force CLONE_SIGHAND.
 130 .\" CLONE_NEWNSIf CLONE_SIGHAND is set and signals are also being shared
 131 .\" (i.e., current->signal->count > 1), force CLONE_THREAD.
 132 .\"
 133 .\" FIXME . CLONE_VM is not (yet, as at 2.6.16) implemented.
 134 .\" .TP
 135 .\" .B CLONE_VM
 136 .\" Reverse the effect of the
 137 .\" .BR clone (2)
 138 .\" .B CLONE_VM
 139 .\" flag.
 140 .\" .RB ( CLONE_VM
 141 .\" is also implicitly set by
 142 .\" .BR vfork (2),
 143 .\" and can be reversed using this
 144 .\" .BR unshare ()
 145 .\" flag.)
 146 .\" Unshare virtual memory, so that the calling process no
 147 .\" longer shares its virtual address space with any other process.
 148 .PP
 149 If
 150 .I flags
 151 is specified as zero, then
 152 .BR unshare ()
 153 is a no-op;
 154 no changes are made to the calling process's execution context.
 155 .SH RETURN VALUE
 156 On success, zero returned.
 157 On failure, \-1 is returned and
 158 .I errno
 159 is set to indicate the error.
 160 .SH ERRORS
 161 .TP
 162 .B EINVAL
 163 An invalid bit was specified in
 164 .IR flags .
 165 .TP
 166 .B ENOMEM
 167 Cannot allocate sufficient memory to copy parts of caller's
 168 context that need to be unshared.
 169 .TP
 170 .B EPERM
 171 The calling process did not have the required privileges for this operation.
 172 .SH VERSIONS
 173 The
 174 .BR unshare ()
 175 system call was added to Linux in kernel 2.6.16.
 176 .SH CONFORMING TO
 177 The
 178 .BR unshare ()
 179 system call is Linux-specific.
 180 .SH NOTES
 181 Not all of the process attributes that can be shared when
 182 a new process is created using
 183 .BR clone (2)
 184 can be unshared using
 185 .BR unshare ().
 186 In particular, as at kernel 2.6.16,
 187 .BR unshare ()
 188 does not implement flags that reverse the effects of
 189 .BR CLONE_SIGHAND ,
 190 .\" However, we can do unshare(CLONE_SIGHAND) if CLONE_SIGHAND
 191 .\" was not specified when doing clone(); i.e., unsharing
 192 .\" signal handlers is permitted if we are not actually
 193 .\" sharing signal handlers.   mtk
 194 .BR CLONE_SYSVSEM ,
 195 .BR CLONE_THREAD ,
 196 or
 197 .\" FIXME . check future kernel versions (e.g., 2.6.17)
 198 .\" to see if CLONE_VM gets implemented.
 199 .BR CLONE_VM .
 200 .\" However, as at 2.6.16, we can do unshare(CLONE_VM) if CLONE_VM
 201 .\" was not specified when doing clone(); i.e., unsharing
 202 .\" virtual memory is permitted if we are not actually
 203 .\" sharing virtual memory.   mtk
 204 Such functionality may be added in the future, if required.
 205 .\"
 206 .\"9) Future Work
 207 .\"--------------
 208 .\"The current implementation of unshare does not allow unsharing of
 209 .\"signals and signal handlers. Signals are complex to begin with and
 210 .\"to unshare signals and/or signal handlers of a currently running
 211 .\"process is even more complex. If in the future there is a specific
 212 .\"need to allow unsharing of signals and/or signal handlers, it can
 213 .\"be incrementally added to unshare without affecting legacy
 214 .\"applications using unshare.
 215 .\"
 216 .SH SEE ALSO
 217 .BR clone (2),
 218 .BR fork (2),
 219 .BR vfork (2),
 220 Documentation/unshare.txt