man7/cgroup_namespaces.7

   1 .\" Copyright (c) 2016 by Michael Kerrisk <mtk.manpages@gmail.com>
   2 .\"
   3 .\" %%%LICENSE_START(VERBATIM)
   4 .\" Permission is granted to make and distribute verbatim copies of this
   5 .\" manual provided the copyright notice and this permission notice are
   6 .\" preserved on all copies.
   7 .\"
   8 .\" Permission is granted to copy and distribute modified versions of this
   9 .\" manual under the conditions for verbatim copying, provided that the
  10 .\" entire resulting derived work is distributed under the terms of a
  11 .\" permission notice identical to this one.
  12 .\"
  13 .\" Since the Linux kernel and libraries are constantly changing, this
  14 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  15 .\" responsibility for errors or omissions, or for damages resulting from
  16 .\" the use of the information contained herein.  The author(s) may not
  17 .\" have taken the same level of care in the production of this manual,
  18 .\" which is licensed free of charge, as they might when working
  19 .\" professionally.
  20 .\"
  21 .\" Formatted or processed versions of this manual, if unaccompanied by
  22 .\" the source, must acknowledge the copyright and authors of this work.
  23 .\" %%%LICENSE_END
  24 .\"
  25 .\"
  26 .TH CGROUP_NAMESPACES 7 2017-09-15 "Linux" "Linux Programmer's Manual"
  27 .SH NAME
  28 cgroup_namespaces \- overview of Linux cgroup namespaces
  29 .SH DESCRIPTION
  30 For an overview of namespaces, see
  31 .BR namespaces (7).
  32 .PP
  33 Cgroup namespaces virtualize the view of a process's cgroups (see
  34 .BR cgroups (7))
  35 as seen via
  36 .IR /proc/[pid]/cgroup
  37 and
  38 .IR /proc/[pid]/mountinfo .
  39 .PP
  40 Each cgroup namespace has its own set of cgroup root directories.
  41 These root directories are the base points for the relative
  42 locations displayed in the corresponding records in the
  43 .IR /proc/[pid]/cgroup
  44 file.
  45 When a process creates a new cgroup namespace using
  46 .BR clone (2)
  47 or
  48 .BR unshare (2)
  49 with the
  50 .BR CLONE_NEWCGROUP
  51 flag, it enters a new cgroup namespace in which its current
  52 cgroups directories become the cgroup root directories
  53 of the new namespace.
  54 (This applies both for the cgroups version 1 hierarchies
  55 and the cgroups version 2 unified hierarchy.)
  56 .PP
  57 When viewing
  58 .IR /proc/[pid]/cgroup ,
  59 the pathname shown in the third field of each record will be
  60 relative to the reading process's root directory
  61 for the corresponding cgroup hierarchy.
  62 If the cgroup directory of the target process lies outside
  63 the root directory of the reading process's cgroup namespace,
  64 then the pathname will show
  65 .I ../
  66 entries for each ancestor level in the cgroup hierarchy.
  67 .PP
  68 The following shell session demonstrates the effect of creating
  69 a new cgroup namespace.
  70 First, (as superuser) we create a child cgroup in the
  71 .I freezer
  72 hierarchy, and put the shell into that cgroup:
  73 .PP
  74 .in +4n
  75 .EX
  76 # \fBmkdir \-p /sys/fs/cgroup/freezer/sub\fP
  77 # \fBecho $$\fP                      # Show PID of this shell
  78 30655
  79 # \fBecho 30655 > /sys/fs/cgroup/freezer/sub/cgroup.procs\fP
  80 # \fBcat /proc/self/cgroup | grep freezer\fP
  81 7:freezer:/sub
  82 .EE
  83 .in
  84 .PP
  85 Next, we use
  86 .BR unshare (1)
  87 to create a process running a new shell in new cgroup and mount namespaces:
  88 .PP
  89 .EX
  90 .in +4n
  91 # \fBunshare \-Cm bash\fP
  92 .in
  93 .EE
  94 .PP
  95 We then inspect the
  96 .IR /proc/[pid]/cgroup
  97 files of, respectively, the new shell process started by the
  98 .BR unshare (1)
  99 command, a process that is in the original cgroup namespace
 100 .RI ( init ,
 101 with PID 1), and a process in a sibling cgroup
 102 .RI ( sub2 ):
 103 .PP
 104 .EX
 105 .in +4n
 106 $ \fBcat /proc/self/cgroup | grep freezer\fP
 107 7:freezer:/
 108 $ \fBcat /proc/1/cgroup | grep freezer\fP
 109 7:freezer:/..
 110 $ \fBcat /proc/20124/cgroup | grep freezer\fP
 111 7:freezer:/../sub2
 112 .in
 113 .EE
 114 .PP
 115 From the output of the first command,
 116 we see that the freezer cgroup membership of the new shell
 117 (which is in the same cgroup as the initial shell)
 118 is shown defined relative to the freezer cgroup root directory
 119 that was established when the new cgroup namespace was created.
 120 (In absolute terms,
 121 the new shell is in the
 122 .I /sub
 123 freezer cgroup,
 124 and the root directory of the freezer cgroup hierarchy
 125 in the new cgroup namespace is also
 126 .IR /sub .
 127 Thus, the new shell's cgroup membership is displayed as \(aq/\(aq.)
 128 .PP
 129 However, when we look in
 130 .IR /proc/self/mountinfo
 131 we see the following anomaly:
 132 .PP
 133 .EX
 134 .in +4n
 135 # \fBcat /proc/self/mountinfo | grep freezer\fP
 136 155 145 0:32 /.. /sys/fs/cgroup/freezer ...
 137 .in
 138 .EE
 139 .PP
 140 The fourth field of this line
 141 .RI ( /.. )
 142 should show the
 143 directory in the cgroup filesystem which forms the root of this mount.
 144 Since by the definition of cgroup namespaces, the process's current
 145 freezer cgroup directory became its root freezer cgroup directory,
 146 we should see \(aq/\(aq in this field.
 147 The problem here is that we are seeing a mount entry for the cgroup
 148 filesystem corresponding to our initial shell process's cgroup namespace
 149 (whose cgroup filesystem is indeed rooted in the parent directory of
 150 .IR sub ).
 151 We need to remount the freezer cgroup filesystem
 152 inside this cgroup namespace, after which we see the expected results:
 153 .PP
 154 .EX
 155 .in +4n
 156 # \fBmount \-\-make\-rslave /\fP     # Don't propagate mount events
 157                             # to other namespaces
 158 # \fBumount /sys/fs/cgroup/freezer\fP
 159 # \fBmount \-t cgroup \-o freezer freezer /sys/fs/cgroup/freezer\fP
 160 # \fBcat /proc/self/mountinfo | grep freezer\fP
 161 155 145 0:32 / /sys/fs/cgroup/freezer rw,relatime ...
 162 .in
 163 .EE
 164 .PP
 165 Use of cgroup namespaces requires a kernel that is configured with the
 166 .B CONFIG_CGROUPS
 167 option.
 168 .\"
 169 .SH CONFORMING TO
 170 Namespaces are a Linux-specific feature.
 171 .SH NOTES
 172 The virtualization provided by cgroup namespaces serves a number of purposes:
 173 .IP * 2
 174 It prevents information leaks whereby cgroup directory paths outside of
 175 a container would otherwise be visible to processes in the container.
 176 Such leakages could, for example,
 177 reveal information about the container framework
 178 to containerized applications.
 179 .IP *
 180 It eases tasks such as container migration.
 181 The virtualization provided by cgroup namespaces
 182 allows containers to be isolated from knowledge of
 183 the pathnames of ancestor cgroups.
 184 Without such isolation, the full cgroup pathnames (displayed in
 185 .IR /proc/self/cgroups )
 186 would need to be replicated on the target system when migrating a container;
 187 those pathnames would also need to be unique,
 188 so that they don't conflict with other pathnames on the target system.
 189 .IP *
 190 It allows better confinement of containerized processes,
 191 because it is possible to mount the container's cgroup filesystems such that
 192 the container processes can't gain access to ancestor cgroup directories.
 193 Consider, for example, the following scenario:
 194 .RS 4
 195 .IP \(bu 2
 196 We have a cgroup directory,
 197 .IR /cg/1 ,
 198 that is owned by user ID 9000.
 199 .IP \(bu
 200 We have a process,
 201 .IR X ,
 202 also owned by user ID 9000,
 203 that is namespaced under the cgroup
 204 .IR /cg/1/2
 205 (i.e.,
 206 .I X
 207 was placed in a new cgroup namespace via
 208 .BR clone (2)
 209 or
 210 .BR unshare (2)
 211 with the
 212 .BR CLONE_NEWCGROUP
 213 flag).
 214 .RE
 215 .IP
 216 In the absence of cgroup namespacing, because the cgroup directory
 217 .IR /cg/1
 218 is owned (and writable) by UID 9000 and process
 219 .I X
 220 is also owned by user ID 9000, then process
 221 .I X
 222 would be able to modify the contents of cgroups files
 223 (i.e., change cgroup settings) not only in
 224 .IR /cg/1/2
 225 but also in the ancestor cgroup directory
 226 .IR /cg/1 .
 227 Namespacing process
 228 .IR X
 229 under the cgroup directory
 230 .IR /cg/1/2 ,
 231 in combination with suitable mount operations
 232 for the cgroup filesystem (as shown above),
 233 prevents it modifying files in
 234 .IR /cg/1 ,
 235 since it cannot even see the contents of that directory
 236 (or of further removed cgroup ancestor directories).
 237 Combined with correct enforcement of hierarchical limits,
 238 this prevents process
 239 .I X
 240 from escaping the limits imposed by ancestor cgroups.
 241 .SH SEE ALSO
 242 .BR unshare (1),
 243 .BR clone (2),
 244 .BR setns (2),
 245 .BR unshare (2),
 246 .BR proc (5),
 247 .BR cgroups (7),
 248 .BR credentials (7),
 249 .BR namespaces (7),
 250 .BR user_namespaces (7)