linux-toradex.git/kernel/cgroup.c, branch v4.6-rc2

Merge branch 'for-4.6-ns' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

2016-03-21T17:05:13+00:00

Pull cgroup namespace support from Tejun Heo:
 "These are changes to implement namespace support for cgroup which has
  been pending for quite some time now.  It is very straight-forward and
  only affects what part of cgroup hierarchies are visible.

  After unsharing, mounting a cgroup fs will be scoped to the cgroups
  the task belonged to at the time of unsharing and the cgroup paths
  exposed to userland would be adjusted accordingly"

* 'for-4.6-ns' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: fix and restructure error handling in copy_cgroup_ns()
  cgroup: fix alloc_cgroup_ns() error handling in copy_cgroup_ns()
  Add FS_USERNS_FLAG to cgroup fs
  cgroup: Add documentation for cgroup namespaces
  cgroup: mount cgroupns-root when inside non-init cgroupns
  kernfs: define kernfs_node_dentry
  cgroup: cgroup namespace setns support
  cgroup: introduce cgroup namespaces
  sched: new clone flag CLONE_NEWCGROUP for cgroup namespace
  kernfs: Add API to generate relative kernfs path

cgroup: avoid false positive gcc-6 warning

2016-03-16T20:32:23+00:00

When all subsystems are disabled, gcc notices that cgroup_subsys_enabled_key
is a zero-length array and that any access to it must be out of bounds:

In file included from ../include/linux/cgroup.h:19:0,
                 from ../kernel/cgroup.c:31:
../kernel/cgroup.c: In function 'cgroup_add_cftypes':
../kernel/cgroup.c:261:53: error: array subscript is above array bounds [-Werror=array-bounds]
  return static_key_enabled(cgroup_subsys_enabled_key[ssid]);
                            ~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
../include/linux/jump_label.h:271:40: note: in definition of macro 'static_key_enabled'
  static_key_count((struct static_key *)x) > 0;    \
                                        ^

We should never call the function in this particular case, so this is
not a bug. In order to silence the warning, this adds an explicit check
for the CGROUP_SUBSYS_COUNT==0 case.

Signed-off-by: Arnd Bergmann 
Signed-off-by: Tejun Heo

cgroup: ignore css_sets associated with dead cgroups during migration

2016-03-16T20:31:46+00:00

Before 2e91fa7f6d45 ("cgroup: keep zombies associated with their
original cgroups"), all dead tasks were associated with init_css_set.
If a zombie task is requested for migration, while migration prep
operations would still be performed on init_css_set, the actual
migration would ignore zombie tasks.  As init_css_set is always valid,
this worked fine.

However, after 2e91fa7f6d45, zombie tasks stay with the css_set it was
associated with at the time of death.  Let's say a task T associated
with cgroup A on hierarchy H-1 and cgroup B on hiearchy H-2.  After T
becomes a zombie, it would still remain associated with A and B.  If A
only contains zombie tasks, it can be removed.  On removal, A gets
marked offline but stays pinned until all zombies are drained.  At
this point, if migration is initiated on T to a cgroup C on hierarchy
H-2, migration path would try to prepare T's css_set for migration and
trigger the following.

 WARNING: CPU: 0 PID: 1576 at kernel/cgroup.c:474 cgroup_get+0x121/0x160()
 CPU: 0 PID: 1576 Comm: bash Not tainted 4.4.0-work+ #289
 ...
 Call Trace:
  [] dump_stack+0x4e/0x82
  [] warn_slowpath_common+0x78/0xb0
  [] warn_slowpath_null+0x15/0x20
  [] cgroup_get+0x121/0x160
  [] link_css_set+0x7b/0x90
  [] find_css_set+0x3bc/0x5e0
  [] cgroup_migrate_prepare_dst+0x89/0x1f0
  [] cgroup_attach_task+0x157/0x230
  [] __cgroup_procs_write+0x2b7/0x470
  [] cgroup_tasks_write+0xc/0x10
  [] cgroup_file_write+0x30/0x1b0
  [] kernfs_fop_write+0x13c/0x180
  [] __vfs_write+0x23/0xe0
  [] vfs_write+0xa4/0x1a0
  [] SyS_write+0x44/0xa0
  [] entry_SYSCALL_64_fastpath+0x12/0x6f

It doesn't make sense to prepare migration for css_sets pointing to
dead cgroups as they are guaranteed to contain only zombies which are
ignored later during migration.  This patch makes cgroup destruction
path mark all affected css_sets as dead and updates the migration path
to ignore them during preparation.

Signed-off-by: Tejun Heo 
Fixes: 2e91fa7f6d45 ("cgroup: keep zombies associated with their original cgroups")
Cc: stable@vger.kernel.org # v4.4+

cgroup: implement cgroup_subsys->implicit_on_dfl

2016-03-08T16:51:26+00:00

Some controllers, perf_event for now and possibly freezer in the
future, don't really make sense to control explicitly through
"cgroup.subtree_control".  For example, the primary role of perf_event
is identifying the cgroups of tasks; however, because the controller
also keeps a small amount of state per cgroup, it can't be replaced
with simple cgroup membership tests.

This patch implements cgroup_subsys->implicit_on_dfl flag.  When set,
the controller is implicitly enabled on all cgroups on the v2
hierarchy so that utility type controllers such as perf_event can be
enabled and function transparently.

An implicit controller doesn't show up in "cgroup.controllers" or
"cgroup.subtree_control", is exempt from no internal process rule and
can be stolen from the default hierarchy even if there are non-root
csses.

v2: Reimplemented on top of the recent updates to css handling and
    subsystem rebinding.  Rebinding implicit subsystems is now a
    simple matter of exempting it from the busy subsystem check.

Signed-off-by: Tejun Heo

cgroup: use css_set->mg_dst_cgrp for the migration target cgroup

2016-03-08T16:51:26+00:00

Migration can be multi-target on the default hierarchy when a
controller is enabled - processes belonging to each child cgroup have
to be moved to the child cgroup itself to refresh css association.

This isn't a problem for cgroup_migrate_add_src() as each source
css_set still maps to single source and target cgroups; however,
cgroup_migrate_prepare_dst() is called once after all source css_sets
are added and thus might not have a single destination cgroup.  This
is currently worked around by specifying NULL for @dst_cgrp and using
the source's default cgroup as destination as the only multi-target
migration in use is self-targetting.  While this works, it's subtle
and clunky.

As all taget cgroups are already specified while preparing the source
css_sets, this clunkiness can easily be removed by recording the
target cgroup in each source css_set.  This patch adds
css_set->mg_dst_cgrp which is recorded on cgroup_migrate_src() and
used by cgroup_migrate_prepare_dst().  This also makes migration code
ready for arbitrary multi-target migration.

Signed-off-by: Tejun Heo

cgroup: make cgroup[_taskset]_migrate() take cgroup_root instead of cgroup

2016-03-08T16:51:26+00:00

On the default hierarchy, a migration can be multi-source and/or
multi-destination.  cgroup_taskest_migrate() used to incorrectly
assume single destination cgroup but the bug has been fixed by
1f7dd3e5a6e4 ("cgroup: fix handling of multi-destination migration
from subtree_control enabling").

Since the commit, @dst_cgrp to cgroup[_taskset]_migrate() is only used
to determine which subsystems are affected or which cgroup_root the
migration is taking place in.  As such, @dst_cgrp is misleading.  This
patch replaces @dst_cgrp with @root.

Signed-off-by: Tejun Heo

cgroup: move migration destination verification out of cgroup_migrate_prepare_dst()

2016-03-08T16:51:25+00:00

cgroup_migrate_prepare_dst() verifies whether the destination cgroup
is allowable; however, the test doesn't really belong there.  It's too
deep and common in the stack and as a result the test itself is gated
by another test.

Separate the test out into cgroup_may_migrate_to() and update
cgroup_attach_task() and cgroup_transfer_tasks() to perform the test
directly.  This doesn't cause any behavior differences.

Signed-off-by: Tejun Heo

cgroup: fix incorrect destination cgroup in cgroup_update_dfl_csses()

2016-03-08T16:51:25+00:00

cgroup_update_dfl_csses() should move each task in the subtree to
self; however, it was incorrectly calling cgroup_migrate_add_src()
with the root of the subtree as @dst_cgrp.  Fortunately,
cgroup_migrate_add_src() currently uses @dst_cgrp only to determine
the hierarchy and the bug doesn't cause any actual breakages.  Fix it.

Signed-off-by: Tejun Heo

cgroup: update css iteration in cgroup_update_dfl_csses()

2016-03-03T14:58:01+00:00

The existing sequences of operations ensure that the offlining csses
are drained before cgroup_update_dfl_csses(), so even though
cgroup_update_dfl_csses() uses css_for_each_descendant_pre() to walk
the target cgroups, it doesn't end up operating on dead cgroups.
Also, the function explicitly excludes the subtree root from
operation.

This is fragile and inconsistent with the rest of css update
operations.  This patch updates cgroup_update_dfl_csses() to use
cgroup_for_each_live_descendant_pre() instead and include the subtree
root.

Signed-off-by: Tejun Heo 
Acked-by: Zefan Li

cgroup: allocate 2x cgrp_cset_links when setting up a new root

2016-03-03T14:58:01+00:00

During prep, cgroup_setup_root() allocates cgrp_cset_links matching
the number of existing css_sets to later link the new root.  This is
fine for now as the only operation which can happen inbetween is
rebind_subsystems() and rebinding of empty subsystems doesn't create
new css_sets.

However, while not yet allowed, with the recent reimplementation,
rebind_subsystems() can rebind subsystems with descendant csses and
thus can create new css_sets.  This patch makes cgroup_setup_root()
allocate 2x of the existing css_sets so that later use of live
subsystem rebinding doesn't blow up.

Signed-off-by: Tejun Heo 
Acked-by: Zefan Li