cgroup_attach_proc() behaves differently from cgroup_attach_task() in
the following aspects.

* All hooks are invoked even if no task is actually being moved.

* ->can_attach_task() is called for all tasks in the group whether the
  new cgrp is different from the current cgrp or not; however,
  ->attach_task() is skipped if new equals new.  This makes the calls
  asymmetric.

This patch improves old cgroup handling in cgroup_attach_proc() by
looking up the current cgroup at the head, recording it in the flex
array along with the task itself, and using it to remove the above two
differences.  This will also ease further changes.

-v2: nr_todo renamed to nr_migrating_tasks as per Paul Menage's
     suggestion.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Menage <paul@paulmenage.org>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>

Update cgroup to take advantage of the fack that threadgroup_lock()
guarantees stable threadgroup.

* Lock threadgroup even if the target is a single task.  This
  guarantees that when the target tasks stay stable during migration
  regardless of the target type.

* Remove PF_EXITING early exit optimization from attach_task_by_pid()
  and check it in cgroup_task_migrate() instead.  The optimization was
  for rather cold path to begin with and PF_EXITING state can be
  trusted throughout migration by checking it after locking
  threadgroup.

* Don't add PF_EXITING tasks to target task array in
  cgroup_attach_proc().  This ensures that task migration is performed
  only for live tasks.

* Remove -ESRCH failure path from cgroup_task_migrate().  With the
  above changes, it's guaranteed to be called only for live tasks.

After the changes, only live tasks are migrated and they're guaranteed
to stay alive until migration is complete.  This removes problems
caused by exec and exit racing against cgroup migration including
symmetry among cgroup attach methods and different cgroup methods
racing each other.

v2: Oleg pointed out that one more PF_EXITING check can be removed
    from cgroup_attach_proc().  Removed.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Paul Menage <paul@paulmenage.org>

threadgroup_lock() protected only protected against new addition to
the threadgroup, which was inherently somewhat incomplete and
problematic for its only user cgroup.  On-going migration could race
against exec and exit leading to interesting problems - the symmetry
between various attach methods, task exiting during method execution,
->exit() racing against attach methods, migrating task switching basic
properties during exec and so on.

This patch extends threadgroup_lock() such that it protects against
all three threadgroup altering operations - fork, exit and exec.  For
exit, threadgroup_change_begin/end() calls are added to exit_signals
around assertion of PF_EXITING.  For exec, threadgroup_[un]lock() are
updated to also grab and release cred_guard_mutex.

With this change, threadgroup_lock() guarantees that the target
threadgroup will remain stable - no new task will be added, no new
PF_EXITING will be set and exec won't happen.

The next patch will update cgroup so that it can take full advantage
of this change.

-v2: beefed up comment as suggested by Frederic.

-v3: narrowed scope of protection in exit path as suggested by
     Frederic.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Paul Menage <paul@paulmenage.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>

Make the following renames to prepare for extension of threadgroup
locking.

* s/signal->threadgroup_fork_lock/signal->group_rwsem/
* s/threadgroup_fork_read_lock()/threadgroup_change_begin()/
* s/threadgroup_fork_read_unlock()/threadgroup_change_end()/
* s/threadgroup_fork_write_lock()/threadgroup_lock()/
* s/threadgroup_fork_write_unlock()/threadgroup_unlock()/

This patch doesn't cause any behavior change.

-v2: Rename threadgroup_change_done() to threadgroup_change_end() per
     KAMEZAWA's suggestion.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Paul Menage <paul@paulmenage.org>

cgroup wants to make threadgroup stable while modifying cgroup
hierarchies which will introduce locking dependency on
cred_guard_mutex from cgroup_mutex.  This unfortunately completes
circular dependency.

 A. cgroup_mutex -> cred_guard_mutex -> s_type->i_mutex_key -> namespace_sem
 B. namespace_sem -> cgroup_mutex

B is from cgroup_show_options() and this patch breaks it by
introducing another mutex cgroup_root_mutex which nests inside
cgroup_mutex and protects cgroupfs_root.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>

The hibernation core code forgets to release memory preallocated
for hibernation if there's an error in its early stages or if test
modes causing hibernation_snapshot() to return early are used.  This
causes the system to be hardly usable, because the amount of
preallocated memory is usually huge.  Fix this problem.

Reported-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>

There's no in-kernel user of set_freezable_with_signal() left.  Mixing
TIF_SIGPENDING with kernel threads can lead to nasty corner cases as
kernel threads never travel signal delivery path on their own.

e.g. the current implementation is buggy in the cancelation path of
__thaw_task().  It calls recalc_sigpending_and_wake() in an attempt to
clear TIF_SIGPENDING but the function never clears it regardless of
sigpending state.  This means that signallable freezable kthreads may
continue executing with !freezing() && stuck TIF_SIGPENDING, which can
be troublesome.

This patch removes set_freezable_with_signal() along with
PF_FREEZER_NOSIG and recalc_sigpending*() calls in freezer.  User
tasks get TIF_SIGPENDING, kernel tasks get woken up and the spurious
sigpending is dealt with in the usual signal delivery path.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Oleg Nesterov <oleg@redhat.com>

After "freezer: make freezing() test freeze conditions in effect
instead of TIF_FREEZE", freezing() returns authoritative answer on
whether the current task should freeze or not and freeze_task()
doesn't need or use @sig_only.  Remove it.

While at it, rewrite function comment for freeze_task() and rename
@sig_only to @user_only in try_to_freeze_tasks().

This patch doesn't cause any functional change.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Oleg Nesterov <oleg@redhat.com>

cgroup_freezer calls freeze_task() without holding tasklist_lock and,
if the task is exiting, its ->sighand may be gone by the time
fake_signal_wake_up() is called.  Use lock_task_sighand() instead of
accessing ->sighand directly.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Paul Menage <paul@paulmenage.org>

If another freeze happens before all tasks leave FROZEN state after
being thawed, the freezer can see the existing FROZEN and consider the
tasks to be frozen but they can clear FROZEN without checking the new
freezing().

Oleg suggested restructuring __refrigerator() such that there's single
condition check section inside freezer_lock and sigpending is cleared
afterwards, which fixes the problem and simplifies the code.
Restructure accordingly.

-v2: Frozen loop exited without releasing freezer_lock.  Fixed.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>

A kthread doing set_freezable*() may race with on-going PM freeze and
the freezer might think all tasks are frozen while the new freezable
kthread is merrily proceeding to execute code paths which aren't
supposed to be executing during PM freeze.

Reimplement set_freezable[_with_signal]() using __set_freezable() such
that freezable PF flags are modified under freezer_lock and
try_to_freeze() is called afterwards.  This eliminates race condition
against freezing.

Note: Separated out from larger patch to resolve fix order dependency
      Oleg pointed out.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

should_send_signal() is only used in freezer.c.  Exporting them only
increases chance of abuse.  Open code the two users and remove it.

Update frozen() to return bool.

Signed-off-by: Tejun Heo <tj@kernel.org>

Using TIF_FREEZE for freezing worked when there was only single
freezing condition (the PM one); however, now there is also the
cgroup_freezer and single bit flag is getting clumsy.
thaw_processes() is already testing whether cgroup freezing in in
effect to avoid thawing tasks which were frozen by both PM and cgroup
freezers.

This is racy (nothing prevents race against cgroup freezing) and
fragile.  A much simpler way is to test actual freeze conditions from
freezing() - ie. directly test whether PM or cgroup freezing is in
effect.

This patch adds variables to indicate whether and what type of
freezing conditions are in effect and reimplements freezing() such
that it directly tests whether any of the two freezing conditions is
active and the task should freeze.  On fast path, freezing() is still
very cheap - it only tests system_freezing_cnt.

This makes the clumsy dancing aroung TIF_FREEZE unnecessary and
freeze/thaw operations more usual - updating state variables for the
new state and nudging target tasks so that they notice the new state
and comply.  As long as the nudging happens after state update, it's
race-free.

* This allows use of freezing() in freeze_task().  Replace the open
  coded tests with freezing().

* p != current test is added to warning printing conditions in
  try_to_freeze_tasks() failure path.  This is necessary as freezing()
  is now true for the task which initiated freezing too.

-v2: Oleg pointed out that re-freezing FROZEN cgroup could increment
     system_freezing_cnt.  Fixed.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Paul Menage <paul@paulmenage.org>  (for the cgroup portions)

TIF_FREEZE will be removed soon and freezing() will directly test
whether any freezing condition is in effect.  Make the following
changes in preparation.

* Rename cgroup_freezing_or_frozen() to cgroup_freezing() and make it
  return bool.

* Make cgroup_freezing() access task_freezer() under rcu read lock
  instead of task_lock().  This makes the state dereferencing racy
  against task moving to another cgroup; however, it was already racy
  without this change as ->state dereference wasn't synchronized.
  This will be later dealt with using attach hooks.

* freezer->state is now set before trying to push tasks into the
  target state.

-v2: Oleg pointed out that freeze_change_state() was setting
     freeze->state incorrectly to CGROUP_FROZEN instead of
     CGROUP_FREEZING.  Fixed.

-v3: Matt pointed out that setting CGROUP_FROZEN used to always invoke
     try_to_freeze_cgroup() regardless of the current state.  Patch
     updated such that the actual freeze/thaw operations are always
     performed on invocation.  This shouldn't make any difference
     unless something is broken.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Paul Menage <paul@paulmenage.org>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Oleg Nesterov <oleg@redhat.com>

freeze_processes() failure path is rather messy.  Freezing is canceled
for workqueues and tasks which aren't frozen yet but frozen tasks are
left alone and should be thawed by the caller and of course some
callers (xen and kexec) didn't do it.

This patch updates __thaw_task() to handle cancelation correctly and
makes freeze_processes() and freeze_kernel_threads() call
thaw_processes() on failure instead so that the system is fully thawed
on failure.  Unnecessary [suspend_]thaw_processes() calls are removed
from kernel/power/hibernate.c, suspend.c and user.c.

While at it, restructure error checking if clause in suspend_prepare()
to be less weird.

-v2: Srivatsa spotted missing removal of suspend_thaw_processes() in
     suspend_prepare() and error in commit message.  Updated.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>

With the previous changes, there's no meaningful difference between
PF_FREEZING and PF_FROZEN.  Remove PF_FREEZING and use PF_FROZEN
instead in task_contributes_to_load().

Signed-off-by: Tejun Heo <tj@kernel.org>

try_to_freeze_tasks() and thaw_processes() use freezable() and
frozen() as preliminary tests before initiating operations on a task.
These are done without any synchronization and hinder with
synchronization cleanup without any real performance benefits.

In try_to_freeze_tasks(), open code self test and move PF_NOFREEZE and
frozen() tests inside freezer_lock in freeze_task().

thaw_processes() can simply drop freezable() test as frozen() test in
__thaw_task() is enough.

Note: This used to be a part of larger patch to fix set_freezable()
      race.  Separated out to satisfy ordering among dependent fixes.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

Currently freezing (TIF_FREEZE) and frozen (PF_FROZEN) states are
interlocked - freezing is set to request freeze and when the task
actually freezes, it clears freezing and sets frozen.

This interlocking makes things more complex than necessary - freezing
doesn't mean there's freezing condition in effect and frozen doesn't
match the task actually entering and leaving frozen state (it's
cleared by the thawing task).

This patch makes freezing indicate that freeze condition is in effect.
A task enters and stays frozen if freezing.  This makes PF_FROZEN
manipulation done only by the task itself and prevents wakeup from
__thaw_task() leaking outside of refrigerator.

The only place which needs to tell freezing && !frozen is
try_to_freeze_task() to whine about tasks which don't enter frozen.
It's updated to test the condition explicitly.

With the change, frozen() state my linger after __thaw_task() until
the task wakes up and exits fridge.  This can trigger BUG_ON() in
update_if_frozen().  Work it around by testing freezing() && frozen()
instead of frozen().

-v2: Oleg pointed out missing re-check of freezing() when trying to
     clear FROZEN and possible spurious BUG_ON() trigger in
     update_if_frozen().  Both fixed.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Paul Menage <paul@paulmenage.org>

Freezer synchronization is needlessly complicated - it's by no means a
hot path and the priority is staying unintrusive and safe.  This patch
makes it simply use a dedicated lock instead of piggy-backing on
task_lock() and playing with memory barriers.

On the failure path of try_to_freeze_tasks(), locking is moved from it
to cancel_freezing().  This makes the frozen() test racy but the race
here is a non-issue as the warning is printed for tasks which failed
to enter frozen for 20 seconds and race on PF_FROZEN at the last
moment doesn't change anything.

This simplifies freezer implementation and eases further changes
including some race fixes.

Signed-off-by: Tejun Heo <tj@kernel.org>

There's no point in thawing nosig tasks before others.  There's no
ordering requirement between the two groups on thaw, which the staged
thawing can't guarantee anyway.  Simplify thaw_processes() by removing
the distinction and collapsing thaw_tasks() into thaw_processes().
This will help further updates to freezer.

Signed-off-by: Tejun Heo <tj@kernel.org>

clear_freeze_flag() in exit_mm() is racy.  Freezing can start
afterwards.  Remove it.  Skipping freezer for exiting task will be
properly implemented later.

Also, freezable() was testing exit_state directly to make system
freezer ignore dead tasks.  Let the exiting task set PF_NOFREEZE after
entering TASK_DEAD instead.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>

thaw_process() now has only internal users - system and cgroup
freezers.  Remove the unnecessary return value, rename, unexport and
collapse __thaw_process() into it.  This will help further updates to
the freezer code.

-v3: oom_kill grew a use of thaw_process() while this patch was
     pending.  Convert it to use __thaw_task() for now.  In the longer
     term, this should be handled by allowing tasks to die if killed
     even if it's frozen.

-v2: minor style update as suggested by Matt.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Paul Menage <menage@google.com>
Cc: Matt Helsley <matthltc@us.ibm.com>

Writeback and thinkpad_acpi have been using thaw_process() to prevent
deadlock between the freezer and kthread_stop(); unfortunately, this
is inherently racy - nothing prevents freezing from happening between
thaw_process() and kthread_stop().

This patch implements kthread_freezable_should_stop() which enters
refrigerator if necessary but is guaranteed to return if
kthread_stop() is invoked.  Both thaw_process() users are converted to
use the new function.

Note that this deadlock condition exists for many of freezable
kthreads.  They need to be converted to use the new should_stop or
freezable workqueue.

Tested with synthetic test case.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Henrique de Moraes Holschuh <ibm-acpi@hmh.eng.br>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Oleg Nesterov <oleg@redhat.com>

There is no reason to export two functions for entering the
refrigerator.  Calling refrigerator() instead of try_to_freeze()
doesn't save anything noticeable or removes any race condition.

* Rename refrigerator() to __refrigerator() and make it return bool
  indicating whether it scheduled out for freezing.

* Update try_to_freeze() to return bool and relay the return value of
  __refrigerator() if freezing().

* Convert all refrigerator() users to try_to_freeze().

* Update documentation accordingly.

* While at it, add might_sleep() to try_to_freeze().

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Samuel Ortiz <samuel@sortiz.org>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jan Kara <jack@suse.cz>
Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp>
Cc: Christoph Hellwig <hch@infradead.org>

refrigerator() saves current->state before entering frozen state and
restores it before returning using __set_current_state(); however,
this is racy, for example, please consider the following sequence.

	set_current_state(TASK_INTERRUPTIBLE);
	try_to_freeze();
	if (kthread_should_stop())
		break;
	schedule();

If kthread_stop() races with ->state restoration, the restoration can
restore ->state to TASK_INTERRUPTIBLE after kthread_stop() sets it to
TASK_RUNNING but kthread_should_stop() may still see zero
->should_stop because there's no memory barrier between restoring
TASK_INTERRUPTIBLE and kthread_should_stop() test.

This isn't restricted to kthread_should_stop().  current->state is
often used in memory barrier based synchronization and silently
restoring it w/o mb breaks them.

Use set_current_state() instead.

Signed-off-by: Tejun Heo <tj@kernel.org>

After commit 2a77c46d
(PM / Suspend: Add statistics debugfs file for suspend to RAM)
a missing pair of braces inside the state_store() function causes even
invalid arguments to suspend to be wrongly treated as failed suspend
attempts. Fix this.

[rjw: Put the hash/subject of the buggy commit into the changelog.]

Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Commit 2aede851
(PM / Hibernate: Freeze kernel threads after preallocating memory)
postponed the freezing of kernel threads to after preallocating memory
for hibernation. But while doing that, the hibernation test TEST_FREEZER
and the test mode HIBERNATION_TESTPROC were not moved accordingly.

As a result, when using these test modes, it only goes upto the freezing of
userspace and exits, when in fact it should go till the complete end of task
freezing stage, namely the freezing of kernel threads as well.

So, move these points of exit to appropriate places so that freezing of
kernel threads is also tested while using these test harnesses.

Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

They are not used any more.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>

Since commit 4a31a334, the name of this misc device is not initialized,
which leads to a funny device named /dev/(null) being created and
/proc/misc containing an entry with just a number but no name. The latter
leads to complaints by cryptsetup, which caused me to investigate this
matter.

Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Use of the GPL or a compatible licence doesn't necessarily make the code
any good.  We already consider staging modules to be suspect, and this
should also be true for out-of-tree modules which may receive very
little review.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Reviewed-by: Dave Jones <davej@redhat.com>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (patched oops-tracing.txt)

Dynamic debugging is currently disabled for tainted modules, except
for TAINT_CRAP.  This prevents use of dynamic debugging for
out-of-tree modules once the next patch is applied.

This condition was apparently intended to avoid a crash if a force-
loaded module has an incompatible definition of dynamic debug
structures.  However, a administrator that forces us to load a module
is claiming that it *is* compatible even though it fails our version
checks.  If they are mistaken, there are any number of ways the module
could crash the system.

As a side-effect, proprietary and other tainted modules can now use
dynamic_debug.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

PM / Freezer: Revert 27920651 "PM / Freezer: Make fake_signal_wake_up() wake TASK_KILLABLE tasks too"

Commit 27920651 "PM / Freezer: Make fake_signal_wake_up() wake
TASK_KILLABLE tasks too" updated fake_signal_wake_up() used by freezer
to wake up KILLABLE tasks.  Sending unsolicited wakeups to tasks in
killable sleep is dangerous as there are code paths which depend on
tasks not waking up spuriously from KILLABLE sleep.

For example. sys_read() or page can sleep in TASK_KILLABLE assuming
that wait/down/whatever _killable can only fail if we can not return
to the usermode.  TASK_TRACED is another obvious example.

The previous patch updated wait_event_freezekillable() such that it
doesn't depend on the spurious wakeup.  This patch reverts the
offending commit.

Note that the spurious KILLABLE wakeup had other implicit effects in
KILLABLE sleeps in nfs and cifs and those will need further updates to
regain freezekillable behavior.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

Remove an "if" check, that repeats an equivalent one 6 lines above.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

The CPU hotplug notifications sent out by the _cpu_up() and _cpu_down()
functions depend on the value of the 'tasks_frozen' argument passed to them
(which indicates whether tasks have been frozen or not).
(Examples for such CPU hotplug notifications: CPU_ONLINE, CPU_ONLINE_FROZEN,
CPU_DEAD, CPU_DEAD_FROZEN).

Thus, it is essential that while the callbacks for those notifications are
running, the state of the system with respect to the tasks being frozen or
not remains unchanged, *throughout that duration*. Hence there is a need for
synchronizing the CPU hotplug code with the freezer subsystem.

Since the freezer is involved only in the Suspend/Hibernate call paths, this
patch hooks the CPU hotplug code to the suspend/hibernate notifiers
PM_[SUSPEND|HIBERNATE]_PREPARE and PM_POST_[SUSPEND|HIBERNATE] to prevent
the race between CPU hotplug and freezer, thus ensuring that CPU hotplug
notifications will always be run with the state of the system really being
what the notifications indicate, _throughout_ their execution time.

Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>

This reverts commit 144060fe.

It causes a resume regression for Andi on his Acer Aspire 1830T post
3.1.  The screen just stays black after wakeup.

Also, it really looks like the wrong way to suspend and resume perf
events: I think they should be done as part of the CPU suspend and
resume, rather than as a notifier that does smp_call_function().

Reported-by: Andi Kleen <andi@firstfloor.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

commit d05c65ff ("genirq: spurious: Run only one poller at a time")
introduced a regression, leaving the boot options 'irqfixup' and
'irqpoll' non-functional. The patch placed tests in each function, to
exit if the function is already running. The test in 'misrouted_irq'
exited when it should have proceeded, effectively disabling
'misrouted_irq' and 'poll_spurious_irqs'.

The check for an already running poller needs to be "!= 1" not "== 1"
as "1" is the value when the first poller starts running.

Signed-off-by: Edward Donovan <edward.donovan@numble.net>
Cc: maciej.rutecki@gmail.com
Link: http://lkml.kernel.org/r/1320175784-6745-1-git-send-email-edward.donovan@numble.net


Cc: stable@vger.kernel.org # >= 2.6.39
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

While back-porting Johannes Weiner's patch "mm: memcg-aware global
reclaim" for an internal effort, we noticed a significant performance
regression during page-reclaim heavy workloads due to high contention of
the ss->id_lock.  This lock protects idr map, and serializes calls to
idr_get_next() in css_get_next() (which is used during the memcg hierarchy
walk).

Since idr_get_next() is just doing a look up, we need only serialize it
with respect to idr_remove()/idr_get_new().  By making the ss->id_lock a
rwlock, contention is greatly reduced and performance improves.

Tested: cat a 256m file from a ramdisk in a 128m container 50 times on
each core (one file + container per core) in parallel on a NUMA machine.
Result is the time for the test to complete in 1 of the containers.
Both kernels included Johannes' memcg-aware global reclaim patches.

Before rwlock patch: 1710.778s
After rwlock patch: 152.227s

Signed-off-by: Andrew Bresticker <abrestic@google.com>
Cc: Paul Menage <menage@gmail.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Adding support for poll() in sysctl fs allows userspace to receive
notifications of changes in sysctl entries.  This adds a infrastructure to
allow files in sysctl fs to be pollable and implements it for hostname and
domainname.

[akpm@linux-foundation.org: s/declare/define/ for definitions]
Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
Cc: Greg KH <gregkh@suse.de>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

{get,put}_mems_allowed() exist so that general kernel code may locklessly
access a task's set of allowable nodes without having the chance that a
concurrent write will cause the nodemask to be empty on configurations
where MAX_NUMNODES > BITS_PER_LONG.

This could incur a significant delay, however, especially in low memory
conditions because the page allocator is blocking and reclaim requires
get_mems_allowed() itself.  It is not atypical to see writes to
cpuset.mems take over 2 seconds to complete, for example.  In low memory
conditions, this is problematic because it's one of the most imporant
times to change cpuset.mems in the first place!

The only way a task's set of allowable nodes may change is through cpusets
by writing to cpuset.mems and when attaching a task to a generic code is
not reading the nodemask with get_mems_allowed() at the same time, and
then clearing all the old nodes.  This prevents the possibility that a
reader will see an empty nodemask at the same time the writer is storing a
new nodemask.

If at least one node remains unchanged, though, it's possible to simply
set all new nodes and then clear all the old nodes.  Changing a task's
nodemask is protected by cgroup_mutex so it's guaranteed that two threads
are not changing the same task's nodemask at the same time, so the
nodemask is guaranteed to be stored before another thread changes it and
determines whether a node remains set or not.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Paul Menage <paul@paulmenage.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

If a task has exited to the point it has called cgroup_exit() already,
then we can't migrate it to another cgroup anymore.

This can happen when we are attaching a task to a new cgroup between the
call to ->can_attach_task() on subsystems and the migration that is
eventually tried in cgroup_task_migrate().

In this case cgroup_task_migrate() returns -ESRCH and we don't want to
attach the task to the subsystems because the attachment to the new cgroup
itself failed.

Fix this by only calling ->attach_task() on the subsystems if the cgroup
migration succeeded.

Reported-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Acked-by: Paul Menage <paul@paulmenage.org>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>