[PATCH 1/8] cgroup: kill cgroup_subsys->__DEPRECATED_clear_css_refs

[Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>]

2ef37d3fe4 ("memcg: Simplify mem_cgroup_force_empty_list error
handling") removed the last user of __DEPRECATED_clear_css_refs.  This
patch removes __DEPRECATED_clear_css_refs and mechanisms to support
it.

* Conditionals dependent on __DEPRECATED_clear_css_refs removed.

* ->pre_destroy() now can only fail if a new task is attached or child
  cgroup is created while ->pre_destroy()s are being called.  As the
  condition is checked again after re-acquiring cgroup_mutex
  afterwards, we don't need to take any immediate action on
  ->pre_destroy() failures.  This reduces cgroup_call_pre_destroy() to
  a simple loop surrounding ->pre_destory().  Remove
  cgroup_call_pre_destroy() and open-code the loop into
  cgroup_rmdir().

* cgroup_clear_css_refs() can no longer fail.  All that needs to be
  done are deactivating refcnts, setting CSS_REMOVED and putting the
  base reference on each css.  Remove cgroup_clear_css_refs() and the
  failure path, and open-code the loops into cgroup_rmdir().

Note that cgroup_rmdir() will see more cleanup soon.

Signed-off-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 include/linux/cgroup.h |  12 ----
 kernel/cgroup.c        | 159 ++++++++++++-------------------------------------
 2 files changed, 38 insertions(+), 133 deletions(-)

diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index c90eaa8..02e09c0 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -86,7 +86,6 @@ struct cgroup_subsys_state {
 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 */
@@ -485,17 +484,6 @@ 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;
-
 #define MAX_CGROUP_TYPE_NAMELEN 32
 	const char *name;
 
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 7981850..033bf4b 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -851,30 +851,6 @@ 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 */
@@ -3901,14 +3877,12 @@ 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);
 }
 
 /*
@@ -3978,10 +3952,9 @@ 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));
@@ -4066,71 +4039,6 @@ 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;
@@ -4138,10 +4046,9 @@ 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);
@@ -4168,11 +4075,9 @@ again:
 	 * 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;
-	}
+	for_each_subsys(cgrp->root, ss)
+		if (ss->pre_destroy)
+			WARN_ON_ONCE(ss->pre_destroy(cgrp));
 
 	mutex_lock(&cgroup_mutex);
 	parent = cgrp->parent;
@@ -4182,21 +4087,34 @@ again:
 		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;
+
+	local_irq_disable();
+
+	/* block new css_tryget() by deactivating refcnt */
+	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);
 	}
-	/* NO css_tryget() can success after here. */
+
+	/*
+	 * Set REMOVED.  All in-progress css_tryget() will be released.
+	 * 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) {
+		struct cgroup_subsys_state *css = cgrp->subsys[ss->subsys_id];
+
+		set_bit(CSS_REMOVED, &css->flags);
+		css_put(css);
+	}
+
+	local_irq_enable();
+
 	finish_wait(&cgroup_rmdir_waitq, &wait);
 	clear_bit(CGRP_WAIT_ON_RMDIR, &cgrp->flags);
 
@@ -4949,8 +4867,7 @@ void __css_put(struct cgroup_subsys_state *css)
 		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();
-- 
1.7.11.7