1
  2To support containers, we now allow multiple instances of devpts filesystem,
  3such that indices of ptys allocated in one instance are independent of indices
  4allocated in other instances of devpts.
  5
  6To preserve backward compatibility, this support for multiple instances is
  7enabled only if:
  8
  9	- CONFIG_DEVPTS_MULTIPLE_INSTANCES=y, and
 10	- '-o newinstance' mount option is specified while mounting devpts
 11
 12IOW, devpts now supports both single-instance and multi-instance semantics.
 13
 14If CONFIG_DEVPTS_MULTIPLE_INSTANCES=n, there is no change in behavior and
 15this referred to as the "legacy" mode. In this mode, the new mount options
 16(-o newinstance and -o ptmxmode) will be ignored with a 'bogus option' message
 17on console.
 18
 19If CONFIG_DEVPTS_MULTIPLE_INSTANCES=y and devpts is mounted without the
 20'newinstance' option (as in current start-up scripts) the new mount binds
 21to the initial kernel mount of devpts. This mode is referred to as the
 22'single-instance' mode and the current, single-instance semantics are
 23preserved, i.e PTYs are common across the system.
 24
 25The only difference between this single-instance mode and the legacy mode
 26is the presence of new, '/dev/pts/ptmx' node with permissions 0000, which
 27can safely be ignored.
 28
 29If CONFIG_DEVPTS_MULTIPLE_INSTANCES=y and 'newinstance' option is specified,
 30the mount is considered to be in the multi-instance mode and a new instance
 31of the devpts fs is created. Any ptys created in this instance are independent
 32of ptys in other instances of devpts. Like in the single-instance mode, the
 33/dev/pts/ptmx node is present. To effectively use the multi-instance mode,
 34open of /dev/ptmx must be a redirected to '/dev/pts/ptmx' using a symlink or
 35bind-mount.
 36
 37Eg: A container startup script could do the following:
 38
 39	$ chmod 0666 /dev/pts/ptmx
 40	$ rm /dev/ptmx
 41	$ ln -s pts/ptmx /dev/ptmx
 42	$ ns_exec -cm /bin/bash
 43
 44	# We are now in new container
 45
 46	$ umount /dev/pts
 47	$ mount -t devpts -o newinstance lxcpts /dev/pts
 48	$ sshd -p 1234
 49
 50where 'ns_exec -cm /bin/bash' calls clone() with CLONE_NEWNS flag and execs
 51/bin/bash in the child process.  A pty created by the sshd is not visible in
 52the original mount of /dev/pts.
 53
 54User-space changes
 55------------------
 56
 57In multi-instance mode (i.e '-o newinstance' mount option is specified at least
 58once), following user-space issues should be noted.
 59
 601. If -o newinstance mount option is never used, /dev/pts/ptmx can be ignored
 61   and no change is needed to system-startup scripts.
 62
 632. To effectively use multi-instance mode (i.e -o newinstance is specified)
 64   administrators or startup scripts should "redirect" open of /dev/ptmx to
 65   /dev/pts/ptmx using either a bind mount or symlink.
 66
 67	$ mount -t devpts -o newinstance devpts /dev/pts
 68
 69   followed by either
 70
 71	$ rm /dev/ptmx
 72	$ ln -s pts/ptmx /dev/ptmx
 73	$ chmod 666 /dev/pts/ptmx
 74   or
 75	$ mount -o bind /dev/pts/ptmx /dev/ptmx
 76
 773. The '/dev/ptmx -> pts/ptmx' symlink is the preferred method since it
 78   enables better error-reporting and treats both single-instance and
 79   multi-instance mounts similarly.
 80
 81   But this method requires that system-startup scripts set the mode of
 82   /dev/pts/ptmx correctly (default mode is 0000). The scripts can set the
 83   mode by, either
 84
 85   	- adding ptmxmode mount option to devpts entry in /etc/fstab, or
 86	- using 'chmod 0666 /dev/pts/ptmx'
 87
 884. If multi-instance mode mount is needed for containers, but the system
 89   startup scripts have not yet been updated, container-startup scripts
 90   should bind mount /dev/ptmx to /dev/pts/ptmx to avoid breaking single-
 91   instance mounts.
 92
 93   Or, in general, container-startup scripts should use:
 94
 95	mount -t devpts -o newinstance -o ptmxmode=0666 devpts /dev/pts
 96	if [ ! -L /dev/ptmx ]; then
 97		mount -o bind /dev/pts/ptmx /dev/ptmx
 98	fi
 99
100   When all devpts mounts are multi-instance, /dev/ptmx can permanently be
101   a symlink to pts/ptmx and the bind mount can be ignored.
102
1035. A multi-instance mount that is not accompanied by the /dev/ptmx to
104   /dev/pts/ptmx redirection would result in an unusable/unreachable pty.
105
106	mount -t devpts -o newinstance lxcpts /dev/pts
107
108   immediately followed by:
109
110	open("/dev/ptmx")
111
112    would create a pty, say /dev/pts/7, in the initial kernel mount.
113    But /dev/pts/7 would be invisible in the new mount.
114
1156. The permissions for /dev/pts/ptmx node should be specified when mounting
116   /dev/pts, using the '-o ptmxmode=%o' mount option (default is 0000).
117
118	mount -t devpts -o newinstance -o ptmxmode=0644 devpts /dev/pts
119
120   The permissions can be later be changed as usual with 'chmod'.
121
122	chmod 666 /dev/pts/ptmx
123
1247. A mount of devpts without the 'newinstance' option results in binding to
125   initial kernel mount.  This behavior while preserving legacy semantics,
126   does not provide strict isolation in a container environment. i.e by
127   mounting devpts without the 'newinstance' option, a container could
128   get visibility into the 'host' or root container's devpts.
129   
130   To workaround this and have strict isolation, all mounts of devpts,
131   including the mount in the root container, should use the newinstance
132   option.