Merge branch 'cgroup-rmdir-updates' into cgroup/for-3.8

Pull rmdir updates into for-3.8 so that further callback updates can
be put on top.  This pull created a trivial conflict between the
following two commits.

  8c7f6edbda ("cgroup: mark subsystems with broken hierarchy support and whine if cgroups are nested for them")
  ed95779340 ("cgroup: kill cgroup_subsys->__DEPRECATED_clear_css_refs")

The former added a field to cgroup_subsys and the latter removed one
from it.  They happen to be colocated causing the conflict.  Keeping
what's added and removing what's removed resolves the conflict.

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

5 files changed, 156 insertions, 336 deletions

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index cafcd7431189..6ce36ff98a41 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -600,7 +600,7 @@ struct cftype blkcg_files[] = {
  *
  * This is the blkcg counterpart of ioc_release_fn().
  */
-static int blkcg_pre_destroy(struct cgroup *cgroup)
+static void blkcg_pre_destroy(struct cgroup *cgroup)
 {
 	struct blkcg *blkcg = cgroup_to_blkcg(cgroup);
 
@@ -622,7 +622,6 @@ static int blkcg_pre_destroy(struct cgroup *cgroup)
 	}
 
 	spin_unlock_irq(&blkcg->lock);
-	return 0;
 }
 
 static void blkcg_destroy(struct cgroup *cgroup)
diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 4cd1d0fd2542..fe876a77031a 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -81,8 +81,6 @@ struct cgroup_subsys_state {
 /* bits in struct cgroup_subsys_state flags field */
 enum {
 	CSS_ROOT, /* This CSS is the root of the subsystem */
-	CSS_REMOVED, /* This CSS is dead */
-	CSS_CLEAR_CSS_REFS,		/* @ss->__DEPRECATED_clear_css_refs */
 };
 
 /* Caller must verify that the css is not for root cgroup */
@@ -105,11 +103,6 @@ static inline void css_get(struct cgroup_subsys_state *css)
 		__css_get(css, 1);
 }
 
-static inline bool css_is_removed(struct cgroup_subsys_state *css)
-{
-	return test_bit(CSS_REMOVED, &css->flags);
-}
-
 /*
  * Call css_tryget() to take a reference on a css if your existing
  * (known-valid) reference isn't already ref-counted. Returns false if
@@ -148,10 +141,6 @@ enum {
 	/* Control Group requires release notifications to userspace */
 	CGRP_NOTIFY_ON_RELEASE,
 	/*
-	 * A thread in rmdir() is wating for this cgroup.
-	 */
-	CGRP_WAIT_ON_RMDIR,
-	/*
 	 * Clone cgroup values when creating a new child cgroup
 	 */
 	CGRP_CLONE_CHILDREN,
@@ -421,23 +410,6 @@ int cgroup_task_count(const struct cgroup *cgrp);
 int cgroup_is_descendant(const struct cgroup *cgrp, struct task_struct *task);
 
 /*
- * When the subsys has to access css and may add permanent refcnt to css,
- * it should take care of racy conditions with rmdir(). Following set of
- * functions, is for stop/restart rmdir if necessary.
- * Because these will call css_get/put, "css" should be alive css.
- *
- *  cgroup_exclude_rmdir();
- *  ...do some jobs which may access arbitrary empty cgroup
- *  cgroup_release_and_wakeup_rmdir();
- *
- *  When someone removes a cgroup while cgroup_exclude_rmdir() holds it,
- *  it sleeps and cgroup_release_and_wakeup_rmdir() will wake him up.
- */
-
-void cgroup_exclude_rmdir(struct cgroup_subsys_state *css);
-void cgroup_release_and_wakeup_rmdir(struct cgroup_subsys_state *css);
-
-/*
  * Control Group taskset, used to pass around set of tasks to cgroup_subsys
  * methods.
  */
@@ -466,7 +438,7 @@ int cgroup_taskset_size(struct cgroup_taskset *tset);
 
 struct cgroup_subsys {
 	struct cgroup_subsys_state *(*create)(struct cgroup *cgrp);
-	int (*pre_destroy)(struct cgroup *cgrp);
+	void (*pre_destroy)(struct cgroup *cgrp);
 	void (*destroy)(struct cgroup *cgrp);
 	int (*can_attach)(struct cgroup *cgrp, struct cgroup_taskset *tset);
 	void (*cancel_attach)(struct cgroup *cgrp, struct cgroup_taskset *tset);
@@ -488,17 +460,6 @@ struct cgroup_subsys {
 	bool use_id;
 
 	/*
-	 * If %true, cgroup removal will try to clear css refs by retrying
-	 * ss->pre_destroy() until there's no css ref left.  This behavior
-	 * is strictly for backward compatibility and will be removed as
-	 * soon as the current user (memcg) is updated.
-	 *
-	 * If %false, ss->pre_destroy() can't fail and cgroup removal won't
-	 * wait for css refs to drop to zero before proceeding.
-	 */
-	bool __DEPRECATED_clear_css_refs;
-
-	/*
 	 * If %false, this subsystem is properly hierarchical -
 	 * configuration, resource accounting and restriction on a parent
 	 * cgroup cover those of its children.  If %true, hierarchy support
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index b7a0171067ea..e3045ad4267a 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -171,8 +171,8 @@ struct css_id {
 	 * The css to which this ID points. This pointer is set to valid value
 	 * after cgroup is populated. If cgroup is removed, this will be NULL.
 	 * This pointer is expected to be RCU-safe because destroy()
-	 * is called after synchronize_rcu(). But for safe use, css_is_removed()
-	 * css_tryget() should be used for avoiding race.
+	 * is called after synchronize_rcu(). But for safe use, css_tryget()
+	 * should be used for avoiding race.
 	 */
 	struct cgroup_subsys_state __rcu *css;
 	/*
@@ -854,30 +854,6 @@ static struct inode *cgroup_new_inode(umode_t mode, struct super_block *sb)
 	return inode;
 }
 
-/*
- * Call subsys's pre_destroy handler.
- * This is called before css refcnt check.
- */
-static int cgroup_call_pre_destroy(struct cgroup *cgrp)
-{
-	struct cgroup_subsys *ss;
-	int ret = 0;
-
-	for_each_subsys(cgrp->root, ss) {
-		if (!ss->pre_destroy)
-			continue;
-
-		ret = ss->pre_destroy(cgrp);
-		if (ret) {
-			/* ->pre_destroy() failure is being deprecated */
-			WARN_ON_ONCE(!ss->__DEPRECATED_clear_css_refs);
-			break;
-		}
-	}
-
-	return ret;
-}
-
 static void cgroup_diput(struct dentry *dentry, struct inode *inode)
 {
 	/* is dentry a directory ? if so, kfree() associated cgroup */
@@ -1015,33 +991,6 @@ static void cgroup_d_remove_dir(struct dentry *dentry)
 }
 
 /*
- * A queue for waiters to do rmdir() cgroup. A tasks will sleep when
- * cgroup->count == 0 && list_empty(&cgroup->children) && subsys has some
- * reference to css->refcnt. In general, this refcnt is expected to goes down
- * to zero, soon.
- *
- * CGRP_WAIT_ON_RMDIR flag is set under cgroup's inode->i_mutex;
- */
-static DECLARE_WAIT_QUEUE_HEAD(cgroup_rmdir_waitq);
-
-static void cgroup_wakeup_rmdir_waiter(struct cgroup *cgrp)
-{
-	if (unlikely(test_and_clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags)))
-		wake_up_all(&cgroup_rmdir_waitq);
-}
-
-void cgroup_exclude_rmdir(struct cgroup_subsys_state *css)
-{
-	css_get(css);
-}
-
-void cgroup_release_and_wakeup_rmdir(struct cgroup_subsys_state *css)
-{
-	cgroup_wakeup_rmdir_waiter(css->cgroup);
-	css_put(css);
-}
-
-/*
  * Call with cgroup_mutex held. Drops reference counts on modules, including
  * any duplicate ones that parse_cgroupfs_options took. If this function
  * returns an error, no reference counts are touched.
@@ -2026,12 +1975,6 @@ int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
 	}
 
 	synchronize_rcu();
-
-	/*
-	 * wake up rmdir() waiter. the rmdir should fail since the cgroup
-	 * is no longer empty.
-	 */
-	cgroup_wakeup_rmdir_waiter(cgrp);
 out:
 	if (retval) {
 		for_each_subsys(root, ss) {
@@ -2201,7 +2144,6 @@ static int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 	 * step 5: success! and cleanup
 	 */
 	synchronize_rcu();
-	cgroup_wakeup_rmdir_waiter(cgrp);
 	retval = 0;
 out_put_css_set_refs:
 	if (retval) {
@@ -4023,14 +3965,12 @@ static void init_cgroup_css(struct cgroup_subsys_state *css,
 	cgrp->subsys[ss->subsys_id] = css;
 
 	/*
-	 * If !clear_css_refs, css holds an extra ref to @cgrp->dentry
-	 * which is put on the last css_put().  dput() requires process
-	 * context, which css_put() may be called without.  @css->dput_work
-	 * will be used to invoke dput() asynchronously from css_put().
+	 * css holds an extra ref to @cgrp->dentry which is put on the last
+	 * css_put().  dput() requires process context, which css_put() may
+	 * be called without.  @css->dput_work will be used to invoke
+	 * dput() asynchronously from css_put().
 	 */
 	INIT_WORK(&css->dput_work, css_dput_fn);
-	if (ss->__DEPRECATED_clear_css_refs)
-		set_bit(CSS_CLEAR_CSS_REFS, &css->flags);
 }
 
 /*
@@ -4054,6 +3994,18 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry,
 	if (!cgrp)
 		return -ENOMEM;
 
+	/*
+	 * Only live parents can have children.  Note that the liveliness
+	 * check isn't strictly necessary because cgroup_mkdir() and
+	 * cgroup_rmdir() are fully synchronized by i_mutex; however, do it
+	 * anyway so that locking is contained inside cgroup proper and we
+	 * don't get nasty surprises if we ever grow another caller.
+	 */
+	if (!cgroup_lock_live_group(parent)) {
+		err = -ENODEV;
+		goto err_free;
+	}
+
 	/* Grab a reference on the superblock so the hierarchy doesn't
 	 * get deleted on unmount if there are child cgroups.  This
 	 * can be done outside cgroup_mutex, since the sb can't
@@ -4061,8 +4013,6 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry,
 	 * fs */
 	atomic_inc(&sb->s_active);
 
-	mutex_lock(&cgroup_mutex);
-
 	init_cgroup_housekeeping(cgrp);
 
 	cgrp->parent = parent;
@@ -4110,10 +4060,9 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry,
 	if (err < 0)
 		goto err_remove;
 
-	/* If !clear_css_refs, each css holds a ref to the cgroup's dentry */
+	/* each css holds a ref to the cgroup's dentry */
 	for_each_subsys(root, ss)
-		if (!ss->__DEPRECATED_clear_css_refs)
-			dget(dentry);
+		dget(dentry);
 
 	/* The cgroup directory was pre-locked for us */
 	BUG_ON(!mutex_is_locked(&cgrp->dentry->d_inode->i_mutex));
@@ -4144,7 +4093,7 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry,
 
 	/* Release the reference count that we took on the superblock */
 	deactivate_super(sb);
-
+err_free:
 	kfree(cgrp);
 	return err;
 }
@@ -4198,71 +4147,6 @@ static int cgroup_has_css_refs(struct cgroup *cgrp)
 	return 0;
 }
 
-/*
- * Atomically mark all (or else none) of the cgroup's CSS objects as
- * CSS_REMOVED. Return true on success, or false if the cgroup has
- * busy subsystems. Call with cgroup_mutex held
- *
- * Depending on whether a subsys has __DEPRECATED_clear_css_refs set or
- * not, cgroup removal behaves differently.
- *
- * If clear is set, css refcnt for the subsystem should be zero before
- * cgroup removal can be committed.  This is implemented by
- * CGRP_WAIT_ON_RMDIR and retry logic around ->pre_destroy(), which may be
- * called multiple times until all css refcnts reach zero and is allowed to
- * veto removal on any invocation.  This behavior is deprecated and will be
- * removed as soon as the existing user (memcg) is updated.
- *
- * If clear is not set, each css holds an extra reference to the cgroup's
- * dentry and cgroup removal proceeds regardless of css refs.
- * ->pre_destroy() will be called at least once and is not allowed to fail.
- * On the last put of each css, whenever that may be, the extra dentry ref
- * is put so that dentry destruction happens only after all css's are
- * released.
- */
-static int cgroup_clear_css_refs(struct cgroup *cgrp)
-{
-	struct cgroup_subsys *ss;
-	unsigned long flags;
-	bool failed = false;
-
-	local_irq_save(flags);
-
-	/*
-	 * Block new css_tryget() by deactivating refcnt.  If all refcnts
-	 * for subsystems w/ clear_css_refs set were 1 at the moment of
-	 * deactivation, we succeeded.
-	 */
-	for_each_subsys(cgrp->root, ss) {
-		struct cgroup_subsys_state *css = cgrp->subsys[ss->subsys_id];
-
-		WARN_ON(atomic_read(&css->refcnt) < 0);
-		atomic_add(CSS_DEACT_BIAS, &css->refcnt);
-
-		if (ss->__DEPRECATED_clear_css_refs)
-			failed |= css_refcnt(css) != 1;
-	}
-
-	/*
-	 * If succeeded, set REMOVED and put all the base refs; otherwise,
-	 * restore refcnts to positive values.  Either way, all in-progress
-	 * css_tryget() will be released.
-	 */
-	for_each_subsys(cgrp->root, ss) {
-		struct cgroup_subsys_state *css = cgrp->subsys[ss->subsys_id];
-
-		if (!failed) {
-			set_bit(CSS_REMOVED, &css->flags);
-			css_put(css);
-		} else {
-			atomic_sub(CSS_DEACT_BIAS, &css->refcnt);
-		}
-	}
-
-	local_irq_restore(flags);
-	return !failed;
-}
-
 static int cgroup_rmdir(struct inode *unused_dir, struct dentry *dentry)
 {
 	struct cgroup *cgrp = dentry->d_fsdata;
@@ -4270,70 +4154,52 @@ static int cgroup_rmdir(struct inode *unused_dir, struct dentry *dentry)
 	struct cgroup *parent;
 	DEFINE_WAIT(wait);
 	struct cgroup_event *event, *tmp;
-	int ret;
+	struct cgroup_subsys *ss;
 
 	/* the vfs holds both inode->i_mutex already */
-again:
 	mutex_lock(&cgroup_mutex);
-	if (atomic_read(&cgrp->count) != 0) {
-		mutex_unlock(&cgroup_mutex);
-		return -EBUSY;
-	}
-	if (!list_empty(&cgrp->children)) {
+	parent = cgrp->parent;
+	if (atomic_read(&cgrp->count) || !list_empty(&cgrp->children)) {
 		mutex_unlock(&cgroup_mutex);
 		return -EBUSY;
 	}
-	mutex_unlock(&cgroup_mutex);
 
 	/*
-	 * In general, subsystem has no css->refcnt after pre_destroy(). But
-	 * in racy cases, subsystem may have to get css->refcnt after
-	 * pre_destroy() and it makes rmdir return with -EBUSY. This sometimes
-	 * make rmdir return -EBUSY too often. To avoid that, we use waitqueue
-	 * for cgroup's rmdir. CGRP_WAIT_ON_RMDIR is for synchronizing rmdir
-	 * and subsystem's reference count handling. Please see css_get/put
-	 * and css_tryget() and cgroup_wakeup_rmdir_waiter() implementation.
+	 * Block new css_tryget() by deactivating refcnt and mark @cgrp
+	 * removed.  This makes future css_tryget() and child creation
+	 * attempts fail thus maintaining the removal conditions verified
+	 * above.
 	 */
-	set_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
+	for_each_subsys(cgrp->root, ss) {
+		struct cgroup_subsys_state *css = cgrp->subsys[ss->subsys_id];
 
-	/*
-	 * Call pre_destroy handlers of subsys. Notify subsystems
-	 * that rmdir() request comes.
-	 */
-	ret = cgroup_call_pre_destroy(cgrp);
-	if (ret) {
-		clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
-		return ret;
+		WARN_ON(atomic_read(&css->refcnt) < 0);
+		atomic_add(CSS_DEACT_BIAS, &css->refcnt);
 	}
+	set_bit(CGRP_REMOVED, &cgrp->flags);
 
+	/*
+	 * Tell subsystems to initate destruction.  pre_destroy() should be
+	 * called with cgroup_mutex unlocked.  See 3fa59dfbc3 ("cgroup: fix
+	 * potential deadlock in pre_destroy") for details.
+	 */
+	mutex_unlock(&cgroup_mutex);
+	for_each_subsys(cgrp->root, ss)
+		if (ss->pre_destroy)
+			ss->pre_destroy(cgrp);
 	mutex_lock(&cgroup_mutex);
-	parent = cgrp->parent;
-	if (atomic_read(&cgrp->count) || !list_empty(&cgrp->children)) {
-		clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
-		mutex_unlock(&cgroup_mutex);
-		return -EBUSY;
-	}
-	prepare_to_wait(&cgroup_rmdir_waitq, &wait, TASK_INTERRUPTIBLE);
-	if (!cgroup_clear_css_refs(cgrp)) {
-		mutex_unlock(&cgroup_mutex);
-		/*
-		 * Because someone may call cgroup_wakeup_rmdir_waiter() before
-		 * prepare_to_wait(), we need to check this flag.
-		 */
-		if (test_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags))
-			schedule();
-		finish_wait(&cgroup_rmdir_waitq, &wait);
-		clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
-		if (signal_pending(current))
-			return -EINTR;
-		goto again;
-	}
-	/* NO css_tryget() can success after here. */
-	finish_wait(&cgroup_rmdir_waitq, &wait);
-	clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
+
+	/*
+	 * Put all the base refs.  Each css holds an extra reference to the
+	 * cgroup's dentry and cgroup removal proceeds regardless of css
+	 * refs.  On the last put of each css, whenever that may be, the
+	 * extra dentry ref is put so that dentry destruction happens only
+	 * after all css's are released.
+	 */
+	for_each_subsys(cgrp->root, ss)
+		css_put(cgrp->subsys[ss->subsys_id]);
 
 	raw_spin_lock(&release_list_lock);
-	set_bit(CGRP_REMOVED, &cgrp->flags);
 	if (!list_empty(&cgrp->release_list))
 		list_del_init(&cgrp->release_list);
 	raw_spin_unlock(&release_list_lock);
@@ -5041,15 +4907,17 @@ static void check_for_release(struct cgroup *cgrp)
 /* Caller must verify that the css is not for root cgroup */
 bool __css_tryget(struct cgroup_subsys_state *css)
 {
-	do {
-		int v = css_refcnt(css);
+	while (true) {
+		int t, v;
 
-		if (atomic_cmpxchg(&css->refcnt, v, v + 1) == v)
+		v = css_refcnt(css);
+		t = atomic_cmpxchg(&css->refcnt, v, v + 1);
+		if (likely(t == v))
 			return true;
+		else if (t < 0)
+			return false;
 		cpu_relax();
-	} while (!test_bit(CSS_REMOVED, &css->flags));
-
-	return false;
+	}
 }
 EXPORT_SYMBOL_GPL(__css_tryget);
 
@@ -5068,11 +4936,9 @@ void __css_put(struct cgroup_subsys_state *css)
 			set_bit(CGRP_RELEASABLE, &cgrp->flags);
 			check_for_release(cgrp);
 		}
-		cgroup_wakeup_rmdir_waiter(cgrp);
 		break;
 	case 0:
-		if (!test_bit(CSS_CLEAR_CSS_REFS, &css->flags))
-			schedule_work(&css->dput_work);
+		schedule_work(&css->dput_work);
 		break;
 	}
 	rcu_read_unlock();
diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c
index a3f358fb8a0c..0d3a1a317731 100644
--- a/mm/hugetlb_cgroup.c
+++ b/mm/hugetlb_cgroup.c
@@ -155,18 +155,13 @@ out:
  * Force the hugetlb cgroup to empty the hugetlb resources by moving them to
  * the parent cgroup.
  */
-static int hugetlb_cgroup_pre_destroy(struct cgroup *cgroup)
+static void hugetlb_cgroup_pre_destroy(struct cgroup *cgroup)
 {
 	struct hstate *h;
 	struct page *page;
-	int ret = 0, idx = 0;
+	int idx = 0;
 
 	do {
-		if (cgroup_task_count(cgroup) ||
-		    !list_empty(&cgroup->children)) {
-			ret = -EBUSY;
-			goto out;
-		}
 		for_each_hstate(h) {
 			spin_lock(&hugetlb_lock);
 			list_for_each_entry(page, &h->hugepage_activelist, lru)
@@ -177,8 +172,6 @@ static int hugetlb_cgroup_pre_destroy(struct cgroup *cgroup)
 		}
 		cond_resched();
 	} while (hugetlb_cgroup_have_usage(cgroup));
-out:
-	return ret;
 }
 
 int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages,
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 7acf43bf04a2..08adaaae6fcc 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2337,7 +2337,6 @@ static int __mem_cgroup_try_charge(struct mm_struct *mm,
 again:
 	if (*ptr) { /* css should be a valid one */
 		memcg = *ptr;
-		VM_BUG_ON(css_is_removed(&memcg->css));
 		if (mem_cgroup_is_root(memcg))
 			goto done;
 		if (nr_pages == 1 && consume_stock(memcg))
@@ -2477,9 +2476,9 @@ static void __mem_cgroup_cancel_local_charge(struct mem_cgroup *memcg,
 
 /*
  * A helper function to get mem_cgroup from ID. must be called under
- * rcu_read_lock(). The caller must check css_is_removed() or some if
- * it's concern. (dropping refcnt from swap can be called against removed
- * memcg.)
+ * rcu_read_lock().  The caller is responsible for calling css_tryget if
+ * the mem_cgroup is used for charging. (dropping refcnt from swap can be
+ * called against removed memcg.)
  */
 static struct mem_cgroup *mem_cgroup_lookup(unsigned short id)
 {
@@ -2676,13 +2675,6 @@ static int mem_cgroup_move_account(struct page *page,
 	/* caller should have done css_get */
 	pc->mem_cgroup = to;
 	mem_cgroup_charge_statistics(to, anon, nr_pages);
-	/*
-	 * We charges against "to" which may not have any tasks. Then, "to"
-	 * can be under rmdir(). But in current implementation, caller of
-	 * this function is just force_empty() and move charge, so it's
-	 * guaranteed that "to" is never removed. So, we don't check rmdir
-	 * status here.
-	 */
 	move_unlock_mem_cgroup(from, &flags);
 	ret = 0;
 unlock:
@@ -2696,10 +2688,27 @@ out:
 	return ret;
 }
 
-/*
- * move charges to its parent.
+/**
+ * mem_cgroup_move_parent - moves page to the parent group
+ * @page: the page to move
+ * @pc: page_cgroup of the page
+ * @child: page's cgroup
+ *
+ * move charges to its parent or the root cgroup if the group has no
+ * parent (aka use_hierarchy==0).
+ * Although this might fail (get_page_unless_zero, isolate_lru_page or
+ * mem_cgroup_move_account fails) the failure is always temporary and
+ * it signals a race with a page removal/uncharge or migration. In the
+ * first case the page is on the way out and it will vanish from the LRU
+ * on the next attempt and the call should be retried later.
+ * Isolation from the LRU fails only if page has been isolated from
+ * the LRU since we looked at it and that usually means either global
+ * reclaim or migration going on. The page will either get back to the
+ * LRU or vanish.
+ * Finaly mem_cgroup_move_account fails only if the page got uncharged
+ * (!PageCgroupUsed) or moved to a different group. The page will
+ * disappear in the next attempt.
  */
-
 static int mem_cgroup_move_parent(struct page *page,
 				  struct page_cgroup *pc,
 				  struct mem_cgroup *child)
@@ -2709,9 +2718,7 @@ static int mem_cgroup_move_parent(struct page *page,
 	unsigned long uninitialized_var(flags);
 	int ret;
 
-	/* Is ROOT ? */
-	if (mem_cgroup_is_root(child))
-		return -EINVAL;
+	VM_BUG_ON(mem_cgroup_is_root(child));
 
 	ret = -EBUSY;
 	if (!get_page_unless_zero(page))
@@ -2728,8 +2735,10 @@ static int mem_cgroup_move_parent(struct page *page,
 	if (!parent)
 		parent = root_mem_cgroup;
 
-	if (nr_pages > 1)
+	if (nr_pages > 1) {
+		VM_BUG_ON(!PageTransHuge(page));
 		flags = compound_lock_irqsave(page);
+	}
 
 	ret = mem_cgroup_move_account(page, nr_pages,
 				pc, child, parent);
@@ -2871,7 +2880,6 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *memcg,
 		return;
 	if (!memcg)
 		return;
-	cgroup_exclude_rmdir(&memcg->css);
 
 	__mem_cgroup_commit_charge(memcg, page, 1, ctype, true);
 	/*
@@ -2885,12 +2893,6 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *memcg,
 		swp_entry_t ent = {.val = page_private(page)};
 		mem_cgroup_uncharge_swap(ent);
 	}
-	/*
-	 * At swapin, we may charge account against cgroup which has no tasks.
-	 * So, rmdir()->pre_destroy() can be called while we do this charge.
-	 * In that case, we need to call pre_destroy() again. check it here.
-	 */
-	cgroup_release_and_wakeup_rmdir(&memcg->css);
 }
 
 void mem_cgroup_commit_charge_swapin(struct page *page,
@@ -3338,8 +3340,7 @@ void mem_cgroup_end_migration(struct mem_cgroup *memcg,
 
 	if (!memcg)
 		return;
-	/* blocks rmdir() */
-	cgroup_exclude_rmdir(&memcg->css);
+
 	if (!migration_ok) {
 		used = oldpage;
 		unused = newpage;
@@ -3373,13 +3374,6 @@ void mem_cgroup_end_migration(struct mem_cgroup *memcg,
 	 */
 	if (anon)
 		mem_cgroup_uncharge_page(used);
-	/*
-	 * At migration, we may charge account against cgroup which has no
-	 * tasks.
-	 * So, rmdir()->pre_destroy() can be called while we do this charge.
-	 * In that case, we need to call pre_destroy() again. check it here.
-	 */
-	cgroup_release_and_wakeup_rmdir(&memcg->css);
 }
 
 /*
@@ -3679,17 +3673,22 @@ unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order,
 	return nr_reclaimed;
 }
 
-/*
+/**
+ * mem_cgroup_force_empty_list - clears LRU of a group
+ * @memcg: group to clear
+ * @node: NUMA node
+ * @zid: zone id
+ * @lru: lru to to clear
+ *
  * Traverse a specified page_cgroup list and try to drop them all.  This doesn't
- * reclaim the pages page themselves - it just removes the page_cgroups.
- * Returns true if some page_cgroups were not freed, indicating that the caller
- * must retry this operation.
+ * reclaim the pages page themselves - pages are moved to the parent (or root)
+ * group.
  */
-static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg,
+static void mem_cgroup_force_empty_list(struct mem_cgroup *memcg,
 				int node, int zid, enum lru_list lru)
 {
 	struct mem_cgroup_per_zone *mz;
-	unsigned long flags, loop;
+	unsigned long flags;
 	struct list_head *list;
 	struct page *busy;
 	struct zone *zone;
@@ -3698,11 +3697,8 @@ static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg,
 	mz = mem_cgroup_zoneinfo(memcg, node, zid);
 	list = &mz->lruvec.lists[lru];
 
-	loop = mz->lru_size[lru];
-	/* give some margin against EBUSY etc...*/
-	loop += 256;
 	busy = NULL;
-	while (loop--) {
+	do {
 		struct page_cgroup *pc;
 		struct page *page;
 
@@ -3728,76 +3724,72 @@ static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg,
 			cond_resched();
 		} else
 			busy = NULL;
-	}
-	return !list_empty(list);
+	} while (!list_empty(list));
 }
 
 /*
- * make mem_cgroup's charge to be 0 if there is no task.
+ * make mem_cgroup's charge to be 0 if there is no task by moving
+ * all the charges and pages to the parent.
  * This enables deleting this mem_cgroup.
+ *
+ * Caller is responsible for holding css reference on the memcg.
  */
-static int mem_cgroup_force_empty(struct mem_cgroup *memcg, bool free_all)
+static void mem_cgroup_reparent_charges(struct mem_cgroup *memcg)
 {
-	int ret;
-	int node, zid, shrink;
-	int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
-	struct cgroup *cgrp = memcg->css.cgroup;
-
-	css_get(&memcg->css);
+	int node, zid;
 
-	shrink = 0;
-	/* should free all ? */
-	if (free_all)
-		goto try_to_free;
-move_account:
 	do {
-		ret = -EBUSY;
-		if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children))
-			goto out;
 		/* This is for making all *used* pages to be on LRU. */
 		lru_add_drain_all();
 		drain_all_stock_sync(memcg);
-		ret = 0;
 		mem_cgroup_start_move(memcg);
 		for_each_node_state(node, N_HIGH_MEMORY) {
-			for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) {
+			for (zid = 0; zid < MAX_NR_ZONES; zid++) {
 				enum lru_list lru;
 				for_each_lru(lru) {
-					ret = mem_cgroup_force_empty_list(memcg,
+					mem_cgroup_force_empty_list(memcg,
 							node, zid, lru);
-					if (ret)
-						break;
 				}
 			}
-			if (ret)
-				break;
 		}
 		mem_cgroup_end_move(memcg);
 		memcg_oom_recover(memcg);
 		cond_resched();
-	/* "ret" should also be checked to ensure all lists are empty. */
-	} while (res_counter_read_u64(&memcg->res, RES_USAGE) > 0 || ret);
-out:
-	css_put(&memcg->css);
-	return ret;
 
-try_to_free:
+		/*
+		 * This is a safety check because mem_cgroup_force_empty_list
+		 * could have raced with mem_cgroup_replace_page_cache callers
+		 * so the lru seemed empty but the page could have been added
+		 * right after the check. RES_USAGE should be safe as we always
+		 * charge before adding to the LRU.
+		 */
+	} while (res_counter_read_u64(&memcg->res, RES_USAGE) > 0);
+}
+
+/*
+ * Reclaims as many pages from the given memcg as possible and moves
+ * the rest to the parent.
+ *
+ * Caller is responsible for holding css reference for memcg.
+ */
+static int mem_cgroup_force_empty(struct mem_cgroup *memcg)
+{
+	int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+	struct cgroup *cgrp = memcg->css.cgroup;
+
 	/* returns EBUSY if there is a task or if we come here twice. */
-	if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children) || shrink) {
-		ret = -EBUSY;
-		goto out;
-	}
+	if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children))
+		return -EBUSY;
+
 	/* we call try-to-free pages for make this cgroup empty */
 	lru_add_drain_all();
 	/* try to free all pages in this cgroup */
-	shrink = 1;
 	while (nr_retries && res_counter_read_u64(&memcg->res, RES_USAGE) > 0) {
 		int progress;
 
-		if (signal_pending(current)) {
-			ret = -EINTR;
-			goto out;
-		}
+		if (signal_pending(current))
+			return -EINTR;
+
 		progress = try_to_free_mem_cgroup_pages(memcg, GFP_KERNEL,
 						false);
 		if (!progress) {
@@ -3808,13 +3800,23 @@ try_to_free:
 
 	}
 	lru_add_drain();
-	/* try move_account...there may be some *locked* pages. */
-	goto move_account;
+	mem_cgroup_reparent_charges(memcg);
+
+	return 0;
 }
 
 static int mem_cgroup_force_empty_write(struct cgroup *cont, unsigned int event)
 {
-	return mem_cgroup_force_empty(mem_cgroup_from_cont(cont), true);
+	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
+	int ret;
+
+	if (mem_cgroup_is_root(memcg))
+		return -EINVAL;
+	css_get(&memcg->css);
+	ret = mem_cgroup_force_empty(memcg);
+	css_put(&memcg->css);
+
+	return ret;
 }
 
 
@@ -5001,11 +5003,11 @@ free_out:
 	return ERR_PTR(error);
 }
 
-static int mem_cgroup_pre_destroy(struct cgroup *cont)
+static void mem_cgroup_pre_destroy(struct cgroup *cont)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
 
-	return mem_cgroup_force_empty(memcg, false);
+	mem_cgroup_reparent_charges(memcg);
 }
 
 static void mem_cgroup_destroy(struct cgroup *cont)
@@ -5607,7 +5609,6 @@ struct cgroup_subsys mem_cgroup_subsys = {
 	.base_cftypes = mem_cgroup_files,
 	.early_init = 0,
 	.use_id = 1,
-	.__DEPRECATED_clear_css_refs = true,
 };
 
 #ifdef CONFIG_MEMCG_SWAP