pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work. If they have a custom pid_ns_for_children
before dump, they must have the same ns after restore. Otherwise,
restored task bumped into enviroment it does not expect.
This patchset solves the problem. It exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
---
Kirill Tkhai (2):
ns: Allow ns_entries to have custom symlink content
pidns: Expose task pid_ns_for_children to userspace
fs/nsfs.c | 4 +++-
fs/proc/namespaces.c | 1 +
include/linux/proc_ns.h | 2 ++
kernel/pid_namespace.c | 25 +++++++++++++++++++++++++
4 files changed, 31 insertions(+), 1 deletion(-)
--
Signed-off-by: Kirill Tkhai <redacted>
Make possible to have link content prefix yyy
different from the link name xxx:
$ readlink /proc/[pid]/ns/xxx
yyy:[4026531838]
This will be used in next patch.
Signed-off-by: Kirill Tkhai <redacted>
Reviewed-by: Cyrill Gorcunov <redacted>
Acked-by: Andrei Vagin <redacted>
---
fs/nsfs.c | 4 +++-
include/linux/proc_ns.h | 1 +
2 files changed, 4 insertions(+), 1 deletion(-)
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
Signed-off-by: Kirill Tkhai <redacted>
Reviewed-by: Cyrill Gorcunov <redacted>
Acked-by: Andrei Vagin <redacted>
---
fs/proc/namespaces.c | 1 +
include/linux/proc_ns.h | 1 +
kernel/pid_namespace.c | 25 +++++++++++++++++++++++++
3 files changed, 27 insertions(+)
From: Andrew Morton <akpm@linux-foundation.org> Date: 2017-03-30 22:05:23
On Thu, 30 Mar 2017 13:27:59 +0300 Kirill Tkhai [off-list ref] wrote:
quoted hunk
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
This interface should be documented somewhere under Documentation/.
But I can't immediately find where the /proc/pid/ns/ pseudo-files are
documented...
From: Andrei Vagin <hidden> Date: 2017-03-31 01:04:31
On Thu, Mar 30, 2017 at 03:05:20PM -0700, Andrew Morton wrote:
On Thu, 30 Mar 2017 13:27:59 +0300 Kirill Tkhai [off-list ref] wrote:
quoted
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
This interface should be documented somewhere under Documentation/.
But I can't immediately find where the /proc/pid/ns/ pseudo-files are
documented...
On Thu, Mar 30, 2017 at 03:05:20PM -0700, Andrew Morton wrote:
quoted
On Thu, 30 Mar 2017 13:27:59 +0300 Kirill Tkhai [off-list ref] wrote:
quoted
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
This interface should be documented somewhere under Documentation/.
But I can't immediately find where the /proc/pid/ns/ pseudo-files are
documented...
I suggest the below patch, but it's too early for the man description till
the feature is in mainline, because the man page requires commit id of the feature.
[PATCH] namespaces.7: Document the /proc/[pid]/ns/pid_for_children file
Signed-off-by: Kirill Tkhai <redacted>
---
man7/namespaces.7 | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
@@ -186,7 +187,14 @@ This file is a handle for the network namespace of the process. .TP .IR/proc/[pid]/ns/pid" (since Linux 3.8)".\" commit 57e8391d327609cbf12d843259c968b9e5c1838f-This file is a handle for the PID namespace of the process.+This file is a handle for the PID namespace of the process. It's+permanent during the whole process life.+.TP+.IR/proc/[pid]/ns/pid_for_children" (since Linux 4.12)"+.\" commit FIXME+This file is a handle for the PID namespace of a next born child+of the process. It's changed after unshare(2) and via setns(2),+so the file may differ from /proc/[pid]/ns/pid. .TP .IR/proc/[pid]/ns/user" (since Linux 3.8)".\" commit cde1975bc242f3e1072bde623ef378e547b73f91
On Fri, Mar 31, 2017 at 2:45 AM, Kirill Tkhai [off-list ref] wrote:
quoted hunk
On 31.03.2017 04:04, Andrei Vagin wrote:
quoted
On Thu, Mar 30, 2017 at 03:05:20PM -0700, Andrew Morton wrote:
quoted
On Thu, 30 Mar 2017 13:27:59 +0300 Kirill Tkhai [off-list ref] wrote:
quoted
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
This interface should be documented somewhere under Documentation/.
But I can't immediately find where the /proc/pid/ns/ pseudo-files are
documented...
I suggest the below patch, but it's too early for the man description till
the feature is in mainline, because the man page requires commit id of the feature.
[PATCH] namespaces.7: Document the /proc/[pid]/ns/pid_for_children file
Signed-off-by: Kirill Tkhai <redacted>
---
man7/namespaces.7 | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
@@ -186,7 +187,14 @@ This file is a handle for the network namespace of the process. .TP .IR /proc/[pid]/ns/pid " (since Linux 3.8)" .\" commit 57e8391d327609cbf12d843259c968b9e5c1838f-This file is a handle for the PID namespace of the process.+This file is a handle for the PID namespace of the process. It's+permanent during the whole process life.+.TP+.IR /proc/[pid]/ns/pid_for_children " (since Linux 4.12)"+.\" commit FIXME+This file is a handle for the PID namespace of a next born child+of the process. It's changed after unshare(2) and via setns(2),+so the file may differ from /proc/[pid]/ns/pid. .TP .IR /proc/[pid]/ns/user " (since Linux 3.8)" .\" commit cde1975bc242f3e1072bde623ef378e547b73f91
On Fri, Mar 31, 2017 at 2:45 AM, Kirill Tkhai [off-list ref] wrote:
quoted
On 31.03.2017 04:04, Andrei Vagin wrote:
quoted
On Thu, Mar 30, 2017 at 03:05:20PM -0700, Andrew Morton wrote:
quoted
On Thu, 30 Mar 2017 13:27:59 +0300 Kirill Tkhai [off-list ref] wrote:
quoted
pid_ns_for_children set by a task is known only to the task itself,
and it's impossible to identify it from outside.
It's a big problem for checkpoint/restore software like CRIU,
because it can't correctly handle tasks, that do setns(CLONE_NEWPID)
in proccess of their work.
This patch solves the problem, and it exposes pid_ns_for_children
to ns directory in standard way with the name "pid_for_children":
~# ls /proc/5531/ns -l | grep pid
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid -> pid:[4026531836]
lrwxrwxrwx 1 root root 0 Jan 14 16:38 pid_for_children -> pid:[4026532286]
This interface should be documented somewhere under Documentation/.
But I can't immediately find where the /proc/pid/ns/ pseudo-files are
documented...
I suggest the below patch, but it's too early for the man description till
the feature is in mainline, because the man page requires commit id of the feature.
[PATCH] namespaces.7: Document the /proc/[pid]/ns/pid_for_children file
Signed-off-by: Kirill Tkhai <redacted>
---
man7/namespaces.7 | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
@@ -186,7 +187,14 @@ This file is a handle for the network namespace of the process. .TP .IR/proc/[pid]/ns/pid" (since Linux 3.8)".\" commit 57e8391d327609cbf12d843259c968b9e5c1838f-This file is a handle for the PID namespace of the process.+This file is a handle for the PID namespace of the process. It's+permanent during the whole process life.+.TP+.IR/proc/[pid]/ns/pid_for_children" (since Linux 4.12)"+.\" commit FIXME+This file is a handle for the PID namespace of a next born child+of the process. It's changed after unshare(2) and via setns(2),+so the file may differ from /proc/[pid]/ns/pid. .TP .IR/proc/[pid]/ns/user" (since Linux 3.8)".\" commit cde1975bc242f3e1072bde623ef378e547b73f91