Message-ID: <CAKgNAkgAT8vFr7hbEzOjJG5pCt8-kh_4Wh9yrzHHXrCJ3Fhh+w@mail.gmail.com>
Date:	Tue, 12 Mar 2013 08:07:30 +0100
From:	"Michael Kerrisk (man-pages)" <mtk.manpages@...il.com>
To:	"Eric W. Biederman" <ebiederm@...ssion.com>
Cc:	Rob Landley <rob@...dley.net>,
	linux-man <linux-man@...r.kernel.org>,
	Linux Containers <containers@...ts.linux-foundation.org>,
	lkml <linux-kernel@...r.kernel.org>,
	Vasily Kulikov <segoon@...nwall.com>,
	Li Zefan <lizefan@...wei.com>
Subject: For review: pid_namespaces(7) man page (draft 2)

Eric et al.,

[CCing Li because of reboot(2) changes]

I have (I think) addressed all previous comments in the current draft
of the pid_namespaces(7) page. This is a final sanity check before I
call this page complete (modulo any future kernel changes).

The attached page aims to provide a fairly complete overview of PID
namespaces. I'm looking for review comments (corrections,
improvements, additions, etc.) on this page. I've provided it in two
forms inline below, and reviewers can comment comment on whichever
form they are most comfortable with:

1) The rendered page as plain text
2) The *roff source (also attached); rendering that source will enable
readers to see proper formatting for the page.

Note that the namespaces(7) page referred to in this page is not yet
finished; I'll send it out for review at a future time.

Thanks,

Michael

==========

PID_NAMESPACES(7)      Linux Programmer's Manual     PID_NAMESPACES(7)



NAME
       pid_namespaces - overview of Linux PID namespaces

DESCRIPTION
       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  con‐
       tainer and migrating the container to a new host while the pro‐
       cesses inside the container maintain the same PIDs.

       PIDs in a new PID namespace start at 1, somewhat like a  stand‐
       alone  system, and calls to fork(2), vfork(2), or clone(2) will
       produce processes with PIDs that are unique within  the  names‐
       pace.

       Use of PID namespaces requires a kernel that is configured with
       the CONFIG_PID_NS option.

   The namespace init process
       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)).  A child process that is  orphaned
       within  the namespace will be reparented to this process rather
       than init(1) (unless one of the ancestors of the child
        in   the   same   PID   namespace   employed   the    prctl(2)
       PR_GET_CHILD_SUBREAPER  command to mark itself as the reaper of
       orphaned descendant processes).

       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 names‐
       pace.  In this case, a subsequent fork(2) into this PID  names‐
       pace  will  fail  with  the error ENOMEM; it is not possible to
       create a new processes 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  ter‐
       minated.   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) will fail
       with ENOMEM.

       Only signals for which the "init"  process  has  established  a
       signal  handler can be sent to the "init" process by other mem‐
       bers 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 associ‐
       ated with those signals (respectively, terminating and stopping
       the process).

       Starting with Linux 3.9, the reboot(2) system causes  a  signal
       to  be sent to the namespace "init" process.  See reboot(2) for
       more details.

   Nesting PID namespaces
       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.

       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, "visi‐
       ble" 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 namespace.
       More succinctly: a process can see  (e.g.,  send  signals  with
       kill(2),  set  nice values with setpriority(2), etc.) only pro‐
       cesses 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 names‐
       pace 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  ini‐
       tial  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  names‐
       pace from the caller of setns(2).  Calls to getppid(2) for such
       processes return 0.

   setns(2) and unshare(2) semantics
       Calls to setns(2) that specify a PID namespace file  descriptor
       and  calls to unshare(2) with the CLONE_NEWPID flag cause chil‐
       dren subsequently created by the caller to be placed in a  dif‐
       ferent PID namespace from the caller.  These calls do not, how‐
       ever, 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  relation‐
       ship  between PID namespaces: the parent of a process is either
       in the same namespace or resides in the  immediate  parent  PID
       namespace.

   Compatibility of CLONE_NEWPID with other CLONE_* flags
       CLONE_NEWPID can't be combined with some other CLONE_* flags:

       *  CLONE_THREAD  requires  being  in  the same PID namespace in
          order that 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) file system.

       *  CLONE_SIGHAND requires being in the same PID namespace; oth‐
          erwise  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)).  A sig‐
          nal queue shared by processes  in  multiple  PID  namespaces
          will defeat that.

       *  CLONE_VM  requires  all of the threads to be in the same PID
          namespace, because, from the point of view of a  core  dump,
          if  two  processes  share  the  same  address space they are
          threads and will be core dumped together.  When a core  dump
          is  written, the PID of each thread is written into the core
          dump.  Writing the process IDs could not  meaningfully  suc‐
          ceed  if some of the process IDs were in a parent PID names‐
          pace.

       To summarize: there is a  technical  requirement  for  each  of
       CLONE_THREAD, CLONE_SIGHAND, and CLONE_VM to share a PID names‐
       pace.  (Note furthermore that in clone(2) requires CLONE_VM  to
       be  specified  if  CLONE_THREAD or CLONE_SIGHAND is specified.)
       Thus, call sequences such as the following will fail (with  the
       error EINVAL):

           unshare(CLONE_NEWPID);
           clone(..., CLONE_VM, ...);    /* Fails */

           setns(fd, CLONE_NEWPID);
           clone(..., CLONE_VM, ...);    /* Fails */

           clone(..., CLONE_VM, ...);
           setns(fd, CLONE_NEWPID);      /* Fails */

           clone(..., CLONE_VM, ...);
           unshare(CLONE_NEWPID);        /* Fails */

   /proc and PID namespaces
       A  /proc  file system shows (in the /proc/PID directories) only
       processes visible in the PID namespace of the process that per‐
       formed  the mount, even if the /proc file system 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.

   Miscellaneous
       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  corre‐
       sponding PID value in the receiving process's PID namespace.

CONFORMING TO
       Namespaces are a Linux-specific feature.

EXAMPLE
       See user_namespaces(7).

SEE ALSO
       clone(2),  setns(2), unshare(2), proc(5), credentials(7), capa‐
       bilities(7), user_namespaces(7), switch_root(8)



Linux                         2013-01-14             PID_NAMESPACES(7)


========== *roff source ==========

$ cat pid_namespaces.7
.\" Copyright (c) 2013 by Michael Kerrisk <mtk.manpages@...il.com>
.\" and Copyright (c) 2012 by Eric W. Biederman <ebiederm@...ssion.com>
.\"
.\" 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.
.\"
.\"
.TH PID_NAMESPACES 7 2013-01-14 "Linux" "Linux Programmer's Manual"
.SH NAME
pid_namespaces \- overview of Linux PID namespaces
.SH DESCRIPTION
For an overview of namespaces, see
.BR 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
.BR fork (2),
.BR vfork (2),
or
.BR 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
.B CONFIG_PID_NS
option.
.\"
.\" ============================================================
.\"
.SS The namespace "init" process
The first process created in a new namespace
(i.e., the process created using
.BR clone (2)
with the
.BR CLONE_NEWPID
flag, or the first child created by a process after a call to
.BR unshare (2)
using the
.BR CLONE_NEWPID
flag) has the PID 1, and is the "init" process for the namespace (see
.BR init (1)).
A child process that is orphaned within the namespace will be reparented
to this process rather than
.BR init (1)
(unless one of the ancestors of the child
 in the same PID namespace employed the
.BR prctl (2)
.B PR_GET_CHILD_SUBREAPER
command to mark itself as the reaper of orphaned descendant processes).

If the "init" process of a PID namespace terminates,
the kernel terminates all of the processes in the namespace via a
.BR 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
.BR fork (2)
into this PID namespace will fail with the error
.BR ENOMEM ;
it is not possible to create a new processes 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
.I /proc/[pid]/ns/pid
file corresponding to a process that was in a namespace to
.BR setns (2)
into that namespace after the "init" process has terminated.
Another possible scenario can occur after a call to
.BR unshare (2):
if the first child subsequently created by a
.BR fork (2)
terminates, then subsequent calls to
.BR fork (2)
will fail with
.BR 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\(emsubject to the usual permission checks described in
.BR kill (2)\(emsend
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
.I siginfo_t
.I si_pid
field described in
.BR sigaction (2)
will be zero.)
.B SIGKILL
or
.B 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 3.9, the
.BR reboot (2)
system causes a signal to be sent to the namespace "init" process.
See
.BR reboot(2)
for more details.
.\"
.\" ============================================================
.\"
.SS Nesting PID namespaces
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
.BR clone (2)
or
.BR unshare (2).
PID namespaces thus form a tree,
with all namespaces ultimately tracing their ancestry to the root namespace.

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 namespace.
More succinctly: a process can see (e.g., send signals with
.BR kill(2),
set nice values with
.BR 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
.BR 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
.BR init (1)
process with PID 1) is necessarily in another namespace.
Likewise, the direct children of a process that uses
.BR setns (2)
to cause its children to join a PID namespace are in a different
PID namespace from the caller of
.BR setns (2).
Calls to
.BR getppid (2)
for such processes return 0.
.\"
.\" ============================================================
.\"
.SS setns(2) and unshare(2) semantics
Calls to
.BR setns (2)
that specify a PID namespace file descriptor
and calls to
.BR unshare (2)
with the
.BR CLONE_NEWPID
flag cause children subsequently created
by the caller to be placed in a different PID namespace from the caller.
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
.BR 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.
.SS Compatibility of CLONE_NEWPID with other CLONE_* flags
.BR CLONE_NEWPID
can't be combined with some other
.BR CLONE_*
flags:
.IP * 3
.B CLONE_THREAD
requires being in the same PID namespace in order that 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
.BR proc (5)
file system.
.IP *
.BR CLONE_SIGHAND
requires being in the same PID namespace;
otherwise the process ID of the process sending a signal
could not be meaningfully encoded when a signal is sent
(see the description of the
.I siginfo_t
type in
.BR sigaction (2)).
A signal queue shared by processes in multiple PID namespaces
will defeat that.
.IP *
.BR CLONE_VM
requires all of the threads to be in the same PID namespace,
because, from the point of view of a core dump,
if two processes share the same address space they are threads and will
be core dumped together.
When a core dump is written, the PID of each
thread is written into the core dump.
Writing the process IDs could not meaningfully succeed
if some of the process IDs were in a parent PID namespace.
.PP
To summarize: there is a technical requirement for each of
.BR CLONE_THREAD ,
.BR CLONE_SIGHAND ,
and
.BR CLONE_VM
to share a PID namespace.
(Note furthermore that in
.BR clone (2)
requires
.BR CLONE_VM
to be specified if
.BR CLONE_THREAD
or
.BR CLONE_SIGHAND
is specified.)
Thus, call sequences such as the following will fail (with the error
.BR EINVAL ):

.nf
    unshare(CLONE_NEWPID);
    clone(..., CLONE_VM, ...);    /* Fails */

    setns(fd, CLONE_NEWPID);
    clone(..., CLONE_VM, ...);    /* Fails */

    clone(..., CLONE_VM, ...);
    setns(fd, CLONE_NEWPID);      /* Fails */

    clone(..., CLONE_VM, ...);
    unshare(CLONE_NEWPID);        /* Fails */
.fi
.\"
.\" ============================================================
.\"
.SS /proc and PID namespaces
A
.I /proc
file system shows (in the
.I /proc/PID
directories) only processes visible in the PID namespace
of the process that performed the mount, even if the
.I /proc
file system 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
.I /proc
so that tools such as
.BR ps (1)
work correctly.
If a new mount namespace is simultaneously created by including
.BR CLONE_NEWNS
in the
.IR flags
argument of
.BR clone (2)
or
.BR unshare (2),
then it isn't necessary to change the root directory:
a new procfs instance can be mounted directly over
.IR /proc .

>From a shell, the command to mount
.I /proc
is:

    $ mount -t proc proc /proc

Calling
.BR readlink (2)
on the path
.I /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.
.\"
.\" ============================================================
.\"
.SS Miscellaneous
When a process ID is passed over a UNIX domain socket to a
process in a different PID namespace (see the description of
.B SCM_CREDENTIALS
in
.BR unix (7)),
it is translated into the corresponding PID value in
the receiving process's PID namespace.
.SH CONFORMING TO
Namespaces are a Linux-specific feature.
.SH EXAMPLE
See
.BR user_namespaces (7).
.SH SEE ALSO
.BR clone (2),
.BR setns (2),
.BR unshare (2),
.BR proc (5),
.BR credentials (7),
.BR capabilities (7),
.BR user_namespaces (7),
.BR switch_root (8)

Download attachment "pid_namespaces.7" of type "application/octet-stream" (11307 bytes)

Powered by blists - more mailing lists