pid_namespaces - overview of Linux PID namespaces
For an overview of namespaces, see
namespaces(7).
PID namespaces isolate the process ID number space, meaning that processes in
different PID namespaces can have the same PID. PID namespaces allow
containers to provide functionality such as suspending/resuming the set of
processes in the container and migrating the container to a new host while the
processes inside the container maintain the same PIDs.
PIDs in a new PID namespace start at 1, somewhat like a standalone system, and
calls to
fork(2),
vfork(2), or
clone(2) will produce
processes with PIDs that are unique within the namespace.
Use of PID namespaces requires a kernel that is configured with the
CONFIG_PID_NS option.
The first process created in a new namespace (i.e., the process created using
clone(2) with the
CLONE_NEWPID flag, or the first child created
by a process after a call to
unshare(2) using the
CLONE_NEWPID
flag) has the PID 1, and is the "init" process for the namespace
(see
init(1)). This process becomes the parent of any child processes
that are orphaned because a process that resides in this PID namespace
terminated (see below for further details).
If the "init" process of a PID namespace terminates, the kernel
terminates all of the processes in the namespace via a
SIGKILL signal.
This behavior reflects the fact that the "init" process is essential
for the correct operation of a PID namespace. In this case, a subsequent
fork(2) into this PID namespace fail with the error
ENOMEM; it
is not possible to create a new process in a PID namespace whose
"init" process has terminated. Such scenarios can occur when, for
example, a process uses an open file descriptor for a
/proc/[pid]/ns/pid file corresponding to a process that was in a
namespace to
setns(2) into that namespace after the "init"
process has terminated. Another possible scenario can occur after a call to
unshare(2): if the first child subsequently created by a
fork(2)
terminates, then subsequent calls to
fork(2) fail with
ENOMEM.
Only signals for which the "init" process has established a signal
handler can be sent to the "init" process by other members of the
PID namespace. This restriction applies even to privileged processes, and
prevents other members of the PID namespace from accidentally killing the
"init" process.
Likewise, a process in an ancestor namespace can—subject to the usual
permission checks described in
kill(2)—send signals to the
"init" process of a child PID namespace only if the "init"
process has established a handler for that signal. (Within the handler, the
siginfo_t si_pid field described in
sigaction(2) will be
zero.)
SIGKILL or
SIGSTOP are treated exceptionally: these
signals are forcibly delivered when sent from an ancestor PID namespace.
Neither of these signals can be caught by the "init" process, and so
will result in the usual actions associated with those signals (respectively,
terminating and stopping the process).
Starting with Linux 3.4, the
reboot(2) system call causes a signal to be
sent to the namespace "init" process. See
reboot(2) for more
details.
PID namespaces can be nested: each PID namespace has a parent, except for the
initial ("root") PID namespace. The parent of a PID namespace is the
PID namespace of the process that created the namespace using
clone(2)
or
unshare(2). PID namespaces thus form a tree, with all namespaces
ultimately tracing their ancestry to the root namespace. Since Linux 3.7, the
kernel limits the maximum nesting depth for PID namespaces to 32.
A process is visible to other processes in its PID namespace, and to the
processes in each direct ancestor PID namespace going back to the root PID
namespace. In this context, "visible" means that one process can be
the target of operations by another process using system calls that specify a
process ID. Conversely, the processes in a child PID namespace can't see
processes in the parent and further removed ancestor namespaces. More
succinctly: a process can see (e.g., send signals with
kill(2), set
nice values with
setpriority(2), etc.) only processes contained in its
own PID namespace and in descendants of that namespace.
A process has one process ID in each of the layers of the PID namespace
hierarchy in which is visible, and walking back though each direct ancestor
namespace through to the root PID namespace. System calls that operate on
process IDs always operate using the process ID that is visible in the PID
namespace of the caller. A call to
getpid(2) always returns the PID
associated with the namespace in which the process was created.
Some processes in a PID namespace may have parents that are outside of the
namespace. For example, the parent of the initial process in the namespace
(i.e., the
init(1) process with PID 1) is necessarily in another
namespace. Likewise, the direct children of a process that uses
setns(2) to cause its children to join a PID namespace are in a
different PID namespace from the caller of
setns(2). Calls to
getppid(2) for such processes return 0.
While processes may freely descend into child PID namespaces (e.g., using
setns(2) with a PID namespace file descriptor), they may not move in
the other direction. That is to say, processes may not enter any ancestor
namespaces (parent, grandparent, etc.). Changing PID namespaces is a one-way
operation.
The
NS_GET_PARENT ioctl(2) operation can be used to discover the
parental relationship between PID namespaces; see
ioctl_ns(2).
Calls to
setns(2) that specify a PID namespace file descriptor and calls
to
unshare(2) with the
CLONE_NEWPID flag cause children
subsequently created by the caller to be placed in a different PID namespace
from the caller. (Since Linux 4.12, that PID namespace is shown via the
/proc/[pid]/ns/pid_for_children file, as described in
namespaces(7).) These calls do not, however, change the PID namespace
of the calling process, because doing so would change the caller's idea of its
own PID (as reported by
getpid()), which would break many applications
and libraries.
To put things another way: a process's PID namespace membership is determined
when the process is created and cannot be changed thereafter. Among other
things, this means that the parental relationship between processes mirrors
the parental relationship between PID namespaces: the parent of a process is
either in the same namespace or resides in the immediate parent PID namespace.
A process may call
unshare(2) with the
CLONE_NEWPID flag only
once. After it has performed this operation, its
/proc/PID/ns/pid_for_children symbolic link will be empty until the
first child is created in the namespace.
When a child process becomes orphaned, it is reparented to the "init"
process in the PID namespace of its parent (unless one of the nearer ancestors
of the parent employed the
prctl(2)
PR_SET_CHILD_SUBREAPER
command to mark itself as the reaper of orphaned descendant processes). Note
that because of the
setns(2) and
unshare(2) semantics described
above, this may be the "init" process in the PID namespace that is
the
parent of the child's PID namespace, rather than the
"init" process in the child's own PID namespace.
In current versions of Linux,
CLONE_NEWPID can't be combined with
CLONE_THREAD. Threads are required to be in the same PID namespace such
that the threads in a process can send signals to each other. Similarly, it
must be possible to see all of the threads of a processes in the
proc(5) filesystem. Additionally, if two threads were in different PID
namespaces, the process ID of the process sending a signal could not be
meaningfully encoded when a signal is sent (see the description of the
siginfo_t type in
sigaction(2)). Since this is computed when a
signal is enqueued, a signal queue shared by processes in multiple PID
namespaces would defeat that.
In earlier versions of Linux,
CLONE_NEWPID was additionally disallowed
(failing with the error
EINVAL) in combination with
CLONE_SIGHAND (before Linux 4.3) as well as
CLONE_VM (before
Linux 3.12). The changes that lifted these restrictions have also been ported
to earlier stable kernels.
A
/proc filesystem shows (in the
/proc/[pid] directories) only
processes visible in the PID namespace of the process that performed the
mount, even if the
/proc filesystem is viewed from processes in other
namespaces.
After creating a new PID namespace, it is useful for the child to change its
root directory and mount a new procfs instance at
/proc so that tools
such as
ps(1) work correctly. If a new mount namespace is
simultaneously created by including
CLONE_NEWNS in the
flags
argument of
clone(2) or
unshare(2), then it isn't necessary to
change the root directory: a new procfs instance can be mounted directly over
/proc.
From a shell, the command to mount
/proc is:
$ mount -t proc proc /proc
Calling
readlink(2) on the path
/proc/self yields the process ID
of the caller in the PID namespace of the procfs mount (i.e., the PID
namespace of the process that mounted the procfs). This can be useful for
introspection purposes, when a process wants to discover its PID in other
namespaces.
- /proc/sys/kernel/ns_last_pid (since Linux 3.3)
- This file displays the last PID that was allocated in this PID namespace.
When the next PID is allocated, the kernel will search for the lowest
unallocated PID that is greater than this value, and when this file is
subsequently read it will show that PID.
- This file is writable by a process that has the CAP_SYS_ADMIN
capability inside its user namespace. This makes it possible to determine
the PID that is allocated to the next process that is created inside this
PID namespace.
When a process ID is passed over a UNIX domain socket to a process in a
different PID namespace (see the description of
SCM_CREDENTIALS in
unix(7)), it is translated into the corresponding PID value in the
receiving process's PID namespace.
Namespaces are a Linux-specific feature.
See
user_namespaces(7).
clone(2),
reboot(2),
setns(2),
unshare(2),
proc(5),
capabilities(7),
credentials(7),
mount_namespaces(7),
namespaces(7),
user_namespaces(7),
switch_root(8)