mbox series

[v5,0/9] net: fix sysfs permssions when device changes network

Message ID 20200225124948.80682-1-christian.brauner@ubuntu.com
Headers show
Series net: fix sysfs permssions when device changes network | expand

Message

Christian Brauner Feb. 25, 2020, 12:49 p.m. UTC
Hey everyone,

/* v5 */
This is v5 with a small fixup requested by Rafael.

/* v4 */
This is v4 with more documentation and other fixes that Greg requested.

/* v3 */
This is v3 with explicit uid and gid parameters added to functions that
change sysfs object ownership as Greg requested.

(I've tagged this with net-next since it's triggered by a bug for
 network device files but it also touches driver core aspects so it's
 not clear-cut. I can of course split this series into separate
 patchsets.) 
We have been struggling with a bug surrounding the ownership of network
device sysfs files when moving network devices between network
namespaces owned by different user namespaces reported by multiple
users.

Currently, when moving network devices between network namespaces the
ownership of the corresponding sysfs entries is not changed. This leads
to problems when tools try to operate on the corresponding sysfs files.

I also causes a bug when creating a network device in a network
namespaces owned by a user namespace and moving that network device back
to the host network namespaces. Because when a network device is created
in a network namespaces it will be owned by the root user of the user
namespace and all its associated sysfs files will also be owned by the
root user of the corresponding user namespace.
If such a network device has to be moved back to the host network
namespace the permissions will still be set to the root user of the
owning user namespaces of the originating network namespace. This means
unprivileged users can e.g. re-trigger uevents for such incorrectly
owned devices on the host or in other network namespaces. They can also
modify the settings of the device itself through sysfs when they
wouldn't be able to do the same through netlink. Both of these things
are unwanted.

For example, quite a few workloads will create network devices in the
host network namespace. Other tools will then proceed to move such
devices between network namespaces owner by other user namespaces. While
the ownership of the device itself is updated in
net/core/net-sysfs.c:dev_change_net_namespace() the corresponding sysfs
entry for the device is not. Below you'll find that moving a network
device (here a veth device) from a network namespace into another
network namespaces owned by a different user namespace with a different
id mapping. As you can see the permissions are wrong even though it is
owned by the userns root user after it has been moved and can be
interacted with through netlink: 

drwxr-xr-x 5 nobody nobody    0 Jan 25 18:08 .
drwxr-xr-x 9 nobody nobody    0 Jan 25 18:08 ..
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 addr_assign_type
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 addr_len
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 address
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 broadcast
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 carrier
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 carrier_changes
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 carrier_down_count
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 carrier_up_count
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 dev_id
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 dev_port
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 dormant
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 duplex
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 flags
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 gro_flush_timeout
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 ifalias
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 ifindex
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 iflink
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 link_mode
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 mtu
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 name_assign_type
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 netdev_group
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 operstate
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 phys_port_id
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 phys_port_name
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 phys_switch_id
drwxr-xr-x 2 nobody nobody    0 Jan 25 18:09 power
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 proto_down
drwxr-xr-x 4 nobody nobody    0 Jan 25 18:09 queues
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 speed
drwxr-xr-x 2 nobody nobody    0 Jan 25 18:09 statistics
lrwxrwxrwx 1 nobody nobody    0 Jan 25 18:08 subsystem -> ../../../../class/net
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:09 tx_queue_len
-r--r--r-- 1 nobody nobody 4096 Jan 25 18:09 type
-rw-r--r-- 1 nobody nobody 4096 Jan 25 18:08 uevent

Constrast this with creating a device of the same type in the network
namespace directly. In this case the device's sysfs permissions will be
correctly updated.
(Please also note, that in a lot of workloads this strategy of creating
 the network device directly in the network device to workaround this
 issue can not be used. Either because the network device is dedicated
 after it has been created or because it used by a process that is
 heavily sandboxed and couldn't create network devices itself.):

drwxr-xr-x 5 root   root      0 Jan 25 18:12 .
drwxr-xr-x 9 nobody nobody    0 Jan 25 18:08 ..
-r--r--r-- 1 root   root   4096 Jan 25 18:12 addr_assign_type
-r--r--r-- 1 root   root   4096 Jan 25 18:12 addr_len
-r--r--r-- 1 root   root   4096 Jan 25 18:12 address
-r--r--r-- 1 root   root   4096 Jan 25 18:12 broadcast
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 carrier
-r--r--r-- 1 root   root   4096 Jan 25 18:12 carrier_changes
-r--r--r-- 1 root   root   4096 Jan 25 18:12 carrier_down_count
-r--r--r-- 1 root   root   4096 Jan 25 18:12 carrier_up_count
-r--r--r-- 1 root   root   4096 Jan 25 18:12 dev_id
-r--r--r-- 1 root   root   4096 Jan 25 18:12 dev_port
-r--r--r-- 1 root   root   4096 Jan 25 18:12 dormant
-r--r--r-- 1 root   root   4096 Jan 25 18:12 duplex
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 flags
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 gro_flush_timeout
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 ifalias
-r--r--r-- 1 root   root   4096 Jan 25 18:12 ifindex
-r--r--r-- 1 root   root   4096 Jan 25 18:12 iflink
-r--r--r-- 1 root   root   4096 Jan 25 18:12 link_mode
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 mtu
-r--r--r-- 1 root   root   4096 Jan 25 18:12 name_assign_type
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 netdev_group
-r--r--r-- 1 root   root   4096 Jan 25 18:12 operstate
-r--r--r-- 1 root   root   4096 Jan 25 18:12 phys_port_id
-r--r--r-- 1 root   root   4096 Jan 25 18:12 phys_port_name
-r--r--r-- 1 root   root   4096 Jan 25 18:12 phys_switch_id
drwxr-xr-x 2 root   root      0 Jan 25 18:12 power
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 proto_down
drwxr-xr-x 4 root   root      0 Jan 25 18:12 queues
-r--r--r-- 1 root   root   4096 Jan 25 18:12 speed
drwxr-xr-x 2 root   root      0 Jan 25 18:12 statistics
lrwxrwxrwx 1 nobody nobody    0 Jan 25 18:12 subsystem -> ../../../../class/net
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 tx_queue_len
-r--r--r-- 1 root   root   4096 Jan 25 18:12 type
-rw-r--r-- 1 root   root   4096 Jan 25 18:12 uevent

Now, when creating a network device in a network namespace owned by a
user namespace and moving it to the host the permissions will be set to
the id that the user namespace root user has been mapped to on the host
leading to all sorts of permission issues mentioned above:

458752
drwxr-xr-x 5 458752 458752      0 Jan 25 18:12 .
drwxr-xr-x 9 root   root        0 Jan 25 18:08 ..
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 addr_assign_type
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 addr_len
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 address
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 broadcast
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 carrier
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 carrier_changes
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 carrier_down_count
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 carrier_up_count
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 dev_id
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 dev_port
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 dormant
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 duplex
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 flags
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 gro_flush_timeout
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 ifalias
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 ifindex
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 iflink
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 link_mode
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 mtu
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 name_assign_type
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 netdev_group
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 operstate
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 phys_port_id
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 phys_port_name
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 phys_switch_id
drwxr-xr-x 2 458752 458752      0 Jan 25 18:12 power
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 proto_down
drwxr-xr-x 4 458752 458752      0 Jan 25 18:12 queues
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 speed
drwxr-xr-x 2 458752 458752      0 Jan 25 18:12 statistics
lrwxrwxrwx 1 root   root        0 Jan 25 18:12 subsystem -> ../../../../class/net
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 tx_queue_len
-r--r--r-- 1 458752 458752   4096 Jan 25 18:12 type
-rw-r--r-- 1 458752 458752   4096 Jan 25 18:12 uevent

Fix this by changing the basic sysfs files associated with network
devices when moving them between network namespaces. To this end we add
some infrastructure to sysfs.

The patchset takes care to only do this when the owning user namespaces
changes and the kids differ. So there's only a performance overhead,
when the owning user namespace of the network namespace is different
__and__ the kid mappings for the root user are different for the two
user namespaces:
Assume we have a netdev eth0 which we create in netns1 owned by userns1.
userns1 has an id mapping of 0 100000 100000. Now we move eth0 into
netns2 which is owned by userns2 which also defines an id mapping of 0
100000 100000. In this case sysfs doesn't need updating. The patch will
handle this case and not do any needless work. Now assume eth0 is moved
into netns3 which is owned by userns3 which defines an id mapping of 0
123456 65536. In this case the root user in each namespace corresponds
to different kid and sysfs needs updating.

Thanks!
Christian

Christian Brauner (9):
  sysfs: add sysfs_file_change_owner_by_name()
  sysfs: add sysfs_link_change_owner()
  sysfs: add sysfs_group{s}_change_owner()
  sysfs: add sysfs_change_owner()
  device: add device_change_owner()
  drivers/base/power: add dpm_sysfs_change_owner()
  net-sysfs: add netdev_change_owner()
  net-sysfs: add queue_change_owner()
  net: fix sysfs permssions when device changes network namespace

 drivers/base/core.c        | 120 ++++++++++++++++++++++++++++++
 drivers/base/power/power.h |   3 +
 drivers/base/power/sysfs.c |  55 +++++++++++++-
 fs/sysfs/file.c            | 148 +++++++++++++++++++++++++++++++++++++
 fs/sysfs/group.c           | 117 +++++++++++++++++++++++++++++
 include/linux/device.h     |   1 +
 include/linux/sysfs.h      |  46 ++++++++++++
 net/core/dev.c             |   9 ++-
 net/core/net-sysfs.c       | 133 +++++++++++++++++++++++++++++++++
 net/core/net-sysfs.h       |   2 +
 10 files changed, 632 insertions(+), 2 deletions(-)


base-commit: bb6d3fb354c5ee8d6bde2d576eb7220ea09862b9