cgroup: stable threadgroup during attach & subsys methods consolidation

index 453100a..efa5886 100644
@@ -63,7 +63,24 @@
 
 #include <linux/atomic.h>
 
+/*
+ * cgroup_mutex is the master lock.  Any modification to cgroup or its
+ * hierarchy must be performed while holding it.
+ *
+ * cgroup_root_mutex nests inside cgroup_mutex and should be held to modify
+ * cgroupfs_root of any cgroup hierarchy - subsys list, flags,
+ * release_agent_path and so on.  Modifying requires both cgroup_mutex and
+ * cgroup_root_mutex.  Readers can acquire either of the two.  This is to
+ * break the following locking order cycle.
+ *
+ *  A. cgroup_mutex -> cred_guard_mutex -> s_type->i_mutex_key -> namespace_sem
+ *  B. namespace_sem -> cgroup_mutex
+ *
+ * B happens only through cgroup_show_options() and using cgroup_root_mutex
+ * breaks it.
+ */
 static DEFINE_MUTEX(cgroup_mutex);
+static DEFINE_MUTEX(cgroup_root_mutex);
 
 /*
  * Generate an array of cgroup subsystem pointers. At boot time, this is

@@ -953,6 +970,7 @@ static int rebind_subsystems(struct cgroupfs_root *root,
 	int i;
 
 	BUG_ON(!mutex_is_locked(&cgroup_mutex));
+	BUG_ON(!mutex_is_locked(&cgroup_root_mutex));
 
 	removed_bits = root->actual_subsys_bits & ~final_bits;
 	added_bits = final_bits & ~root->actual_subsys_bits;

@@ -1043,7 +1061,7 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs)
 	struct cgroupfs_root *root = vfs->mnt_sb->s_fs_info;
 	struct cgroup_subsys *ss;
 
-	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 	for_each_subsys(root, ss)
 		seq_printf(seq, ",%s", ss->name);
 	if (test_bit(ROOT_NOPREFIX, &root->flags))

@@ -1054,7 +1072,7 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs)
 		seq_puts(seq, ",clone_children");
 	if (strlen(root->name))
 		seq_printf(seq, ",name=%s", root->name);
-	mutex_unlock(&cgroup_mutex);
+	mutex_unlock(&cgroup_root_mutex);
 	return 0;
 }
 

@@ -1269,6 +1287,7 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data)
 
 	mutex_lock(&cgrp->dentry->d_inode->i_mutex);
 	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 
 	/* See what subsystems are wanted */
 	ret = parse_cgroupfs_options(data, &opts);

@@ -1297,6 +1316,7 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data)
  out_unlock:
 	kfree(opts.release_agent);
 	kfree(opts.name);
+	mutex_unlock(&cgroup_root_mutex);
 	mutex_unlock(&cgroup_mutex);
 	mutex_unlock(&cgrp->dentry->d_inode->i_mutex);
 	return ret;

@@ -1481,6 +1501,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	int ret = 0;
 	struct super_block *sb;
 	struct cgroupfs_root *new_root;
+	struct inode *inode;
 
 	/* First find the desired set of subsystems */
 	mutex_lock(&cgroup_mutex);

@@ -1514,7 +1535,6 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		/* We used the new root structure, so this is a new hierarchy */
 		struct list_head tmp_cg_links;
 		struct cgroup *root_cgrp = &root->top_cgroup;
-		struct inode *inode;
 		struct cgroupfs_root *existing_root;
 		const struct cred *cred;
 		int i;

@@ -1528,18 +1548,14 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 
 		mutex_lock(&inode->i_mutex);
 		mutex_lock(&cgroup_mutex);
+		mutex_lock(&cgroup_root_mutex);
 
-		if (strlen(root->name)) {
-			/* Check for name clashes with existing mounts */
-			for_each_active_root(existing_root) {
-				if (!strcmp(existing_root->name, root->name)) {
-					ret = -EBUSY;
-					mutex_unlock(&cgroup_mutex);
-					mutex_unlock(&inode->i_mutex);
-					goto drop_new_super;
-				}
-			}
-		}
+		/* Check for name clashes with existing mounts */
+		ret = -EBUSY;
+		if (strlen(root->name))
+			for_each_active_root(existing_root)
+				if (!strcmp(existing_root->name, root->name))
+					goto unlock_drop;
 
 		/*
 		 * We're accessing css_set_count without locking

@@ -1549,18 +1565,13 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		 * have some link structures left over
 		 */
 		ret = allocate_cg_links(css_set_count, &tmp_cg_links);
-		if (ret) {
-			mutex_unlock(&cgroup_mutex);
-			mutex_unlock(&inode->i_mutex);
-			goto drop_new_super;
-		}
+		if (ret)
+			goto unlock_drop;
 
 		ret = rebind_subsystems(root, root->subsys_bits);
 		if (ret == -EBUSY) {
-			mutex_unlock(&cgroup_mutex);
-			mutex_unlock(&inode->i_mutex);
 			free_cg_links(&tmp_cg_links);
-			goto drop_new_super;
+			goto unlock_drop;
 		}
 		/*
 		 * There must be no failure case after here, since rebinding

@@ -1599,6 +1610,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		cred = override_creds(&init_cred);
 		cgroup_populate_dir(root_cgrp);
 		revert_creds(cred);
+		mutex_unlock(&cgroup_root_mutex);
 		mutex_unlock(&cgroup_mutex);
 		mutex_unlock(&inode->i_mutex);
 	} else {

@@ -1615,6 +1627,10 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	kfree(opts.name);
 	return dget(sb->s_root);
 
+ unlock_drop:
+	mutex_unlock(&cgroup_root_mutex);
+	mutex_unlock(&cgroup_mutex);
+	mutex_unlock(&inode->i_mutex);
  drop_new_super:
 	deactivate_locked_super(sb);
  drop_modules:

@@ -1639,6 +1655,7 @@ static void cgroup_kill_sb(struct super_block *sb) {
 	BUG_ON(!list_empty(&cgrp->sibling));
 
 	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 
 	/* Rebind all subsystems back to the default hierarchy */
 	ret = rebind_subsystems(root, 0);

@@ -1664,6 +1681,7 @@ static void cgroup_kill_sb(struct super_block *sb) {
 		root_count--;
 	}
 
+	mutex_unlock(&cgroup_root_mutex);
 	mutex_unlock(&cgroup_mutex);
 
 	kill_litter_super(sb);

@@ -2308,7 +2326,9 @@ static int cgroup_release_agent_write(struct cgroup *cgrp, struct cftype *cft,
 		return -EINVAL;
 	if (!cgroup_lock_live_group(cgrp))
 		return -ENODEV;
+	mutex_lock(&cgroup_root_mutex);
 	strcpy(cgrp->root->release_agent_path, buffer);
+	mutex_unlock(&cgroup_root_mutex);
 	cgroup_unlock();
 	return 0;
 }

index 453100a..efa5886 100644
@@ -63,7 +63,24 @@
 
 #include <linux/atomic.h>
 
+/*
+ * cgroup_mutex is the master lock.  Any modification to cgroup or its
+ * hierarchy must be performed while holding it.
+ *
+ * cgroup_root_mutex nests inside cgroup_mutex and should be held to modify
+ * cgroupfs_root of any cgroup hierarchy - subsys list, flags,
+ * release_agent_path and so on.  Modifying requires both cgroup_mutex and
+ * cgroup_root_mutex.  Readers can acquire either of the two.  This is to
+ * break the following locking order cycle.
+ *
+ *  A. cgroup_mutex -> cred_guard_mutex -> s_type->i_mutex_key -> namespace_sem
+ *  B. namespace_sem -> cgroup_mutex
+ *
+ * B happens only through cgroup_show_options() and using cgroup_root_mutex
+ * breaks it.
+ */
 static DEFINE_MUTEX(cgroup_mutex);
+static DEFINE_MUTEX(cgroup_root_mutex);
 
 /*
  * Generate an array of cgroup subsystem pointers. At boot time, this is

@@ -953,6 +970,7 @@ static int rebind_subsystems(struct cgroupfs_root *root,
 	int i;
 
 	BUG_ON(!mutex_is_locked(&cgroup_mutex));
+	BUG_ON(!mutex_is_locked(&cgroup_root_mutex));
 
 	removed_bits = root->actual_subsys_bits & ~final_bits;
 	added_bits = final_bits & ~root->actual_subsys_bits;

@@ -1043,7 +1061,7 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs)
 	struct cgroupfs_root *root = vfs->mnt_sb->s_fs_info;
 	struct cgroup_subsys *ss;
 
-	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 	for_each_subsys(root, ss)
 		seq_printf(seq, ",%s", ss->name);
 	if (test_bit(ROOT_NOPREFIX, &root->flags))

@@ -1054,7 +1072,7 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs)
 		seq_puts(seq, ",clone_children");
 	if (strlen(root->name))
 		seq_printf(seq, ",name=%s", root->name);
-	mutex_unlock(&cgroup_mutex);
+	mutex_unlock(&cgroup_root_mutex);
 	return 0;
 }
 

@@ -1269,6 +1287,7 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data)
 
 	mutex_lock(&cgrp->dentry->d_inode->i_mutex);
 	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 
 	/* See what subsystems are wanted */
 	ret = parse_cgroupfs_options(data, &opts);

@@ -1297,6 +1316,7 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data)
  out_unlock:
 	kfree(opts.release_agent);
 	kfree(opts.name);
+	mutex_unlock(&cgroup_root_mutex);
 	mutex_unlock(&cgroup_mutex);
 	mutex_unlock(&cgrp->dentry->d_inode->i_mutex);
 	return ret;

@@ -1481,6 +1501,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	int ret = 0;
 	struct super_block *sb;
 	struct cgroupfs_root *new_root;
+	struct inode *inode;
 
 	/* First find the desired set of subsystems */
 	mutex_lock(&cgroup_mutex);

@@ -1514,7 +1535,6 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		/* We used the new root structure, so this is a new hierarchy */
 		struct list_head tmp_cg_links;
 		struct cgroup *root_cgrp = &root->top_cgroup;
-		struct inode *inode;
 		struct cgroupfs_root *existing_root;
 		const struct cred *cred;
 		int i;

@@ -1528,18 +1548,14 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 
 		mutex_lock(&inode->i_mutex);
 		mutex_lock(&cgroup_mutex);
+		mutex_lock(&cgroup_root_mutex);
 
-		if (strlen(root->name)) {
-			/* Check for name clashes with existing mounts */
-			for_each_active_root(existing_root) {
-				if (!strcmp(existing_root->name, root->name)) {
-					ret = -EBUSY;
-					mutex_unlock(&cgroup_mutex);
-					mutex_unlock(&inode->i_mutex);
-					goto drop_new_super;
-				}
-			}
-		}
+		/* Check for name clashes with existing mounts */
+		ret = -EBUSY;
+		if (strlen(root->name))
+			for_each_active_root(existing_root)
+				if (!strcmp(existing_root->name, root->name))
+					goto unlock_drop;
 
 		/*
 		 * We're accessing css_set_count without locking

@@ -1549,18 +1565,13 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		 * have some link structures left over
 		 */
 		ret = allocate_cg_links(css_set_count, &tmp_cg_links);
-		if (ret) {
-			mutex_unlock(&cgroup_mutex);
-			mutex_unlock(&inode->i_mutex);
-			goto drop_new_super;
-		}
+		if (ret)
+			goto unlock_drop;
 
 		ret = rebind_subsystems(root, root->subsys_bits);
 		if (ret == -EBUSY) {
-			mutex_unlock(&cgroup_mutex);
-			mutex_unlock(&inode->i_mutex);
 			free_cg_links(&tmp_cg_links);
-			goto drop_new_super;
+			goto unlock_drop;
 		}
 		/*
 		 * There must be no failure case after here, since rebinding

@@ -1599,6 +1610,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		cred = override_creds(&init_cred);
 		cgroup_populate_dir(root_cgrp);
 		revert_creds(cred);
+		mutex_unlock(&cgroup_root_mutex);
 		mutex_unlock(&cgroup_mutex);
 		mutex_unlock(&inode->i_mutex);
 	} else {

@@ -1615,6 +1627,10 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	kfree(opts.name);
 	return dget(sb->s_root);
 
+ unlock_drop:
+	mutex_unlock(&cgroup_root_mutex);
+	mutex_unlock(&cgroup_mutex);
+	mutex_unlock(&inode->i_mutex);
  drop_new_super:
 	deactivate_locked_super(sb);
  drop_modules:

@@ -1639,6 +1655,7 @@ static void cgroup_kill_sb(struct super_block *sb) {
 	BUG_ON(!list_empty(&cgrp->sibling));
 
 	mutex_lock(&cgroup_mutex);
+	mutex_lock(&cgroup_root_mutex);
 
 	/* Rebind all subsystems back to the default hierarchy */
 	ret = rebind_subsystems(root, 0);

@@ -1664,6 +1681,7 @@ static void cgroup_kill_sb(struct super_block *sb) {
 		root_count--;
 	}
 
+	mutex_unlock(&cgroup_root_mutex);
 	mutex_unlock(&cgroup_mutex);
 
 	kill_litter_super(sb);

@@ -2308,7 +2326,9 @@ static int cgroup_release_agent_write(struct cgroup *cgrp, struct cftype *cft,
 		return -EINVAL;
 	if (!cgroup_lock_live_group(cgrp))
 		return -ENODEV;
+	mutex_lock(&cgroup_root_mutex);
 	strcpy(cgrp->root->release_agent_path, buffer);
+	mutex_unlock(&cgroup_root_mutex);
 	cgroup_unlock();
 	return 0;
 }

index aa47d0f..6fb546e 100644
@@ -635,11 +635,12 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting.  fork and exit paths are protected
+	 * with this rwsem using threadgroup_change_begin/end().  Users
+	 * which require threadgroup to remain stable should use
+	 * threadgroup_[un]lock() which also takes care of exec path.
+	 * Currently, cgroup is the only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2367,7 +2368,6 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2377,13 +2377,37 @@ static inline void threadgroup_change_done(struct task_struct *tsk)
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  exec is already synchronized with cread_guard_mutex
+ * which we grab here.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/* exec uses exit for de-threading, grab cred_guard_mutex first */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

index 2913b35..3565f00 100644
@@ -938,6 +938,12 @@ NORET_TYPE void do_exit(long code)
 		schedule();
 	}
 
+	/*
+	 * @tsk's threadgroup is going through changes - lock out users
+	 * which expect stable threadgroup.
+	 */
+	threadgroup_change_begin(tsk);
+
 	exit_irq_thread();
 
 	exit_signals(tsk);  /* sets PF_EXITING */

@@ -1020,10 +1026,6 @@ NORET_TYPE void do_exit(long code)
 		kfree(current->pi_state_cache);
 #endif
 	/*
-	 * Make sure we are holding no locks:
-	 */
-	debug_check_no_locks_held(tsk);
-	/*
 	 * We can do this unlocked here. The futex code uses this flag
 	 * just to verify whether the pi state cleanup has been done
 	 * or not. In the worst case it loops once more.

@@ -1040,6 +1042,13 @@ NORET_TYPE void do_exit(long code)
 
 	preempt_disable();
 	exit_rcu();
+
+	/*
+	 * Release threadgroup and make sure we are holding no locks.
+	 */
+	threadgroup_change_done(tsk);
+	debug_check_no_locks_held(tsk);
+
 	/* causes final put_task_struct in finish_task_switch(). */
 	tsk->state = TASK_DEAD;
 	schedule();

index f0e099f..83e10f9 100644
@@ -1762,7 +1762,7 @@ EXPORT_SYMBOL_GPL(cgroup_path);
  *
  * 'guarantee' is set if the caller promises that a new css_set for the task
  * will already exist. If not set, this function might sleep, and can fail with
- * -ENOMEM. Otherwise, it can only fail with -ESRCH.
+ * -ENOMEM. Must be called with cgroup_mutex and threadgroup locked.
  */
 static int cgroup_task_migrate(struct cgroup *cgrp, struct cgroup *oldcgrp,
 			       struct task_struct *tsk, bool guarantee)

@@ -1800,13 +1800,9 @@ static int cgroup_task_migrate(struct cgroup *cgrp, struct cgroup *oldcgrp,
 	}
 	put_css_set(oldcg);
 
-	/* if PF_EXITING is set, the tsk->cgroups pointer is no longer safe. */
+	/* @tsk can't exit as its threadgroup is locked */
 	task_lock(tsk);
-	if (tsk->flags & PF_EXITING) {
-		task_unlock(tsk);
-		put_css_set(newcg);
-		return -ESRCH;
-	}
+	WARN_ON_ONCE(tsk->flags & PF_EXITING);
 	rcu_assign_pointer(tsk->cgroups, newcg);
 	task_unlock(tsk);
 

@@ -1832,8 +1828,8 @@ static int cgroup_task_migrate(struct cgroup *cgrp, struct cgroup *oldcgrp,
  * @cgrp: the cgroup the task is attaching to
  * @tsk: the task to be attached
  *
- * Call holding cgroup_mutex. May take task_lock of
- * the task 'tsk' during call.
+ * Call with cgroup_mutex and threadgroup locked. May take task_lock of
+ * @tsk during call.
  */
 int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
 {

@@ -1842,6 +1838,10 @@ int cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
 	struct cgroup *oldcgrp;
 	struct cgroupfs_root *root = cgrp->root;
 
+	/* @tsk either already exited or can't exit until the end */
+	if (tsk->flags & PF_EXITING)
+		return -ESRCH;
+
 	/* Nothing to do if the task is already in that cgroup */
 	oldcgrp = task_cgroup_from_root(tsk, root);
 	if (cgrp == oldcgrp)

@@ -2062,6 +2062,10 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 	tsk = leader;
 	i = 0;
 	do {
+		/* @tsk either already exited or can't exit until the end */
+		if (tsk->flags & PF_EXITING)
+			continue;
+
 		/* as per above, nr_threads may decrease, but not increase. */
 		BUG_ON(i >= group_size);
 		get_task_struct(tsk);

@@ -2116,11 +2120,6 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 			continue;
 		/* get old css_set pointer */
 		task_lock(tsk);
-		if (tsk->flags & PF_EXITING) {
-			/* ignore this task if it's going away */
-			task_unlock(tsk);
-			continue;
-		}
 		oldcg = tsk->cgroups;
 		get_css_set(oldcg);
 		task_unlock(tsk);

@@ -2158,9 +2157,8 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 			if (ss->attach_task)
 				ss->attach_task(cgrp, tsk);
 		}
-		/* if the thread is PF_EXITING, it can just get skipped. */
 		retval = cgroup_task_migrate(cgrp, oldcgrp, tsk, true);
-		BUG_ON(retval != 0 && retval != -ESRCH);
+		BUG_ON(retval != 0);
 	}
 	/* nothing is sensitive to fork() after this point. */
 

@@ -2212,8 +2210,8 @@ out_free_group_list:
 
 /*
  * Find the task_struct of the task to attach by vpid and pass it along to the
- * function to attach either it or all tasks in its threadgroup. Will take
- * cgroup_mutex; may take task_lock of task.
+ * function to attach either it or all tasks in its threadgroup. Will lock
+ * cgroup_mutex and threadgroup; may take task_lock of task.
  */
 static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 {

@@ -2240,11 +2238,6 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 			 * detect it later.
 			 */
 			tsk = tsk->group_leader;
-		} else if (tsk->flags & PF_EXITING) {
-			/* optimization for the single-task-only case */
-			rcu_read_unlock();
-			cgroup_unlock();
-			return -ESRCH;
 		}
 		/*
 		 * even if we're attaching all tasks in the thread group, we

@@ -2268,13 +2261,15 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 		get_task_struct(tsk);
 	}
 
-	if (threadgroup) {
-		threadgroup_lock(tsk);
+	threadgroup_lock(tsk);
+
+	if (threadgroup)
 		ret = cgroup_attach_proc(cgrp, tsk);
-		threadgroup_unlock(tsk);
-	} else {
+	else
 		ret = cgroup_attach_task(cgrp, tsk);
-	}
+
+	threadgroup_unlock(tsk);
+
 	put_task_struct(tsk);
 	cgroup_unlock();
 	return ret;

index 83e10f9..3679fb6 100644
@@ -2152,13 +2152,15 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 		oldcgrp = task_cgroup_from_root(tsk, root);
 		if (cgrp == oldcgrp)
 			continue;
+
+		retval = cgroup_task_migrate(cgrp, oldcgrp, tsk, true);
+		BUG_ON(retval != 0);
+
 		/* attach each task to each subsystem */
 		for_each_subsys(root, ss) {
 			if (ss->attach_task)
 				ss->attach_task(cgrp, tsk);
 		}
-		retval = cgroup_task_migrate(cgrp, oldcgrp, tsk, true);
-		BUG_ON(retval != 0);
 	}
 	/* nothing is sensitive to fork() after this point. */
 

index b596e54..b5e3fef 100644
@@ -30,8 +30,10 @@ EXPORT_SYMBOL_GPL(blkio_root_cgroup);
 
 static struct cgroup_subsys_state *blkiocg_create(struct cgroup_subsys *,
 						  struct cgroup *);
-static int blkiocg_can_attach_task(struct cgroup *, struct task_struct *);
-static void blkiocg_attach_task(struct cgroup *, struct task_struct *);
+static int blkiocg_can_attach(struct cgroup_subsys *, struct cgroup *,
+			      struct cgroup_taskset *);
+static void blkiocg_attach(struct cgroup_subsys *, struct cgroup *,
+			   struct cgroup_taskset *);
 static void blkiocg_destroy(struct cgroup_subsys *, struct cgroup *);
 static int blkiocg_populate(struct cgroup_subsys *, struct cgroup *);
 

@@ -44,8 +46,8 @@ static int blkiocg_populate(struct cgroup_subsys *, struct cgroup *);
 struct cgroup_subsys blkio_subsys = {
 	.name = "blkio",
 	.create = blkiocg_create,
-	.can_attach_task = blkiocg_can_attach_task,
-	.attach_task = blkiocg_attach_task,
+	.can_attach = blkiocg_can_attach,
+	.attach = blkiocg_attach,
 	.destroy = blkiocg_destroy,
 	.populate = blkiocg_populate,
 #ifdef CONFIG_BLK_CGROUP

@@ -1609,30 +1611,39 @@ done:
  * of the main cic data structures.  For now we allow a task to change
  * its cgroup only if it's the only owner of its ioc.
  */
-static int blkiocg_can_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
+static int blkiocg_can_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
+			      struct cgroup_taskset *tset)
 {
+	struct task_struct *task;
 	struct io_context *ioc;
 	int ret = 0;
 
 	/* task_lock() is needed to avoid races with exit_io_context() */
-	task_lock(tsk);
-	ioc = tsk->io_context;
-	if (ioc && atomic_read(&ioc->nr_tasks) > 1)
-		ret = -EINVAL;
-	task_unlock(tsk);
-
+	cgroup_taskset_for_each(task, cgrp, tset) {
+		task_lock(task);
+		ioc = task->io_context;
+		if (ioc && atomic_read(&ioc->nr_tasks) > 1)
+			ret = -EINVAL;
+		task_unlock(task);
+		if (ret)
+			break;
+	}
 	return ret;
 }
 
-static void blkiocg_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
+static void blkiocg_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
+			   struct cgroup_taskset *tset)
 {
+	struct task_struct *task;
 	struct io_context *ioc;
 
-	task_lock(tsk);
-	ioc = tsk->io_context;
-	if (ioc)
-		ioc->cgroup_changed = 1;
-	task_unlock(tsk);
+	cgroup_taskset_for_each(task, cgrp, tset) {
+		task_lock(task);
+		ioc = task->io_context;
+		if (ioc)
+			ioc->cgroup_changed = 1;
+		task_unlock(task);
+	}
 }
 
 void blkio_policy_register(struct blkio_policy_type *blkiop)

index a72d5fa..68ef861 100644
@@ -163,10 +163,19 @@ static int freezer_can_attach(struct cgroup_subsys *ss,
 			      struct cgroup_taskset *tset)
 {
 	struct freezer *freezer;
+	struct task_struct *task;
 
 	/*
 	 * Anything frozen can't move or be moved to/from.
 	 */
+	rcu_read_lock();
+	cgroup_taskset_for_each(task, new_cgroup, tset) {
+		if (__cgroup_freezing_or_frozen(task)) {
+			rcu_read_unlock();
+			return -EBUSY;
+		}
+	}
+	rcu_read_unlock();
 
 	freezer = cgroup_freezer(new_cgroup);
 	if (freezer->state != CGROUP_THAWED)

@@ -175,17 +184,6 @@ static int freezer_can_attach(struct cgroup_subsys *ss,
 	return 0;
 }
 
-static int freezer_can_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
-{
-	rcu_read_lock();
-	if (__cgroup_freezing_or_frozen(tsk)) {
-		rcu_read_unlock();
-		return -EBUSY;
-	}
-	rcu_read_unlock();
-	return 0;
-}
-
 static void freezer_fork(struct cgroup_subsys *ss, struct task_struct *task)
 {
 	struct freezer *freezer;

@@ -381,10 +379,5 @@ struct cgroup_subsys freezer_subsys = {
 	.populate	= freezer_populate,
 	.subsys_id	= freezer_subsys_id,
 	.can_attach	= freezer_can_attach,
-	.can_attach_task = freezer_can_attach_task,
-	.pre_attach	= NULL,
-	.attach_task	= NULL,
-	.attach		= NULL,
 	.fork		= freezer_fork,
-	.exit		= NULL,
 };

113 lines not on this page · raw message

index d1a1bee..ed35211 100644
@@ -7136,10 +7136,13 @@ static int __perf_cgroup_move(void *info)
 	return 0;
 }
 
-static void
-perf_cgroup_attach_task(struct cgroup *cgrp, struct task_struct *task)
+static void perf_cgroup_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
+			       struct cgroup_taskset *tset)
 {
-	task_function_call(task, __perf_cgroup_move, task);
+	struct task_struct *task;
+
+	cgroup_taskset_for_each(task, cgrp, tset)
+		task_function_call(task, __perf_cgroup_move, task);
 }
 
 static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp,

@@ -7153,7 +7156,7 @@ static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp,
 	if (!(task->flags & PF_EXITING))
 		return;
 
-	perf_cgroup_attach_task(cgrp, task);
+	task_function_call(task, __perf_cgroup_move, task);
 }
 
 struct cgroup_subsys perf_subsys = {

@@ -7162,6 +7165,6 @@ struct cgroup_subsys perf_subsys = {
 	.create		= perf_cgroup_create,
 	.destroy	= perf_cgroup_destroy,
 	.exit		= perf_cgroup_exit,
-	.attach_task	= perf_cgroup_attach_task,
+	.attach		= perf_cgroup_attach,
 };
 #endif /* CONFIG_CGROUP_PERF */

index d87c6e5..49e3605 100644
@@ -9129,24 +9129,31 @@ cpu_cgroup_destroy(struct cgroup_subsys *ss, struct cgroup *cgrp)
 	sched_destroy_group(tg);
 }
 
-static int
-cpu_cgroup_can_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
+static int cpu_cgroup_can_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
+				 struct cgroup_taskset *tset)
 {
+	struct task_struct *task;
+
+	cgroup_taskset_for_each(task, cgrp, tset) {
 #ifdef CONFIG_RT_GROUP_SCHED
-	if (!sched_rt_can_attach(cgroup_tg(cgrp), tsk))
-		return -EINVAL;
+		if (!sched_rt_can_attach(cgroup_tg(cgrp), task))
+			return -EINVAL;
 #else
-	/* We don't support RT-tasks being in separate groups */
-	if (tsk->sched_class != &fair_sched_class)
-		return -EINVAL;
+		/* We don't support RT-tasks being in separate groups */
+		if (task->sched_class != &fair_sched_class)
+			return -EINVAL;
 #endif
+	}
 	return 0;
 }
 
-static void
-cpu_cgroup_attach_task(struct cgroup *cgrp, struct task_struct *tsk)
+static void cpu_cgroup_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
+			      struct cgroup_taskset *tset)
 {
-	sched_move_task(tsk);
+	struct task_struct *task;
+
+	cgroup_taskset_for_each(task, cgrp, tset)
+		sched_move_task(task);
 }
 
 static void

@@ -9482,8 +9489,8 @@ struct cgroup_subsys cpu_cgroup_subsys = {
 	.name		= "cpu",
 	.create		= cpu_cgroup_create,
 	.destroy	= cpu_cgroup_destroy,
-	.can_attach_task = cpu_cgroup_can_attach_task,
-	.attach_task	= cpu_cgroup_attach_task,
+	.can_attach	= cpu_cgroup_can_attach,
+	.attach		= cpu_cgroup_attach,
 	.exit		= cpu_cgroup_exit,
 	.populate	= cpu_cgroup_populate,
 	.subsys_id	= cpu_cgroup_subsys_id,

index 08ffab0..ef20cbe 100644
@@ -23,11 +23,10 @@ extern struct files_struct init_files;
 extern struct fs_struct init_fs;
 
 #ifdef CONFIG_CGROUPS
-#define INIT_THREADGROUP_FORK_LOCK(sig)					\
-	.threadgroup_fork_lock =					\
-		__RWSEM_INITIALIZER(sig.threadgroup_fork_lock),
+#define INIT_GROUP_RWSEM(sig)						\
+	.group_rwsem = __RWSEM_INITIALIZER(sig.group_rwsem),
 #else
-#define INIT_THREADGROUP_FORK_LOCK(sig)
+#define INIT_GROUP_RWSEM(sig)
 #endif
 
 #define INIT_SIGNALS(sig) {						\

@@ -46,7 +45,7 @@ extern struct fs_struct init_fs;
 	},								\
 	.cred_guard_mutex =						\
 		 __MUTEX_INITIALIZER(sig.cred_guard_mutex),		\
-	INIT_THREADGROUP_FORK_LOCK(sig)					\
+	INIT_GROUP_RWSEM(sig)						\
 }
 
 extern struct nsproxy init_nsproxy;

index e8acce7..aa47d0f 100644
@@ -635,13 +635,13 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The threadgroup_fork_lock prevents threads from forking with
+	 * The group_rwsem prevents threads from forking with
 	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
 	 * threadgroup-wide operations. It's taken for reading in fork.c in
 	 * copy_process().
 	 * Currently only needed write-side by cgroups.
 	 */
-	struct rw_semaphore threadgroup_fork_lock;
+	struct rw_semaphore group_rwsem;
 #endif
 
 	int oom_adj;		/* OOM kill score adjustment (bit shift) */

@@ -2367,29 +2367,29 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of threadgroup_fork_lock in signal_struct. */
+/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
-static inline void threadgroup_fork_read_lock(struct task_struct *tsk)
+static inline void threadgroup_change_begin(struct task_struct *tsk)
 {
-	down_read(&tsk->signal->threadgroup_fork_lock);
+	down_read(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_read_unlock(struct task_struct *tsk)
+static inline void threadgroup_change_done(struct task_struct *tsk)
 {
-	up_read(&tsk->signal->threadgroup_fork_lock);
+	up_read(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_write_lock(struct task_struct *tsk)
+static inline void threadgroup_lock(struct task_struct *tsk)
 {
-	down_write(&tsk->signal->threadgroup_fork_lock);
+	down_write(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_write_unlock(struct task_struct *tsk)
+static inline void threadgroup_unlock(struct task_struct *tsk)
 {
-	up_write(&tsk->signal->threadgroup_fork_lock);
+	up_write(&tsk->signal->group_rwsem);
 }
 #else
-static inline void threadgroup_fork_read_lock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_read_unlock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_write_lock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_write_unlock(struct task_struct *tsk) {}
+static inline void threadgroup_change_begin(struct task_struct *tsk) {}
+static inline void threadgroup_change_done(struct task_struct *tsk) {}
+static inline void threadgroup_lock(struct task_struct *tsk) {}
+static inline void threadgroup_unlock(struct task_struct *tsk) {}
 #endif
 
 #ifndef __HAVE_THREAD_FUNCTIONS

index efa5886..f0e099f 100644
@@ -2003,8 +2003,8 @@ static int css_set_prefetch(struct cgroup *cgrp, struct css_set *cg,
  * @cgrp: the cgroup to attach to
  * @leader: the threadgroup leader task_struct of the group to be attached
  *
- * Call holding cgroup_mutex and the threadgroup_fork_lock of the leader. Will
- * take task_lock of each thread in leader's threadgroup individually in turn.
+ * Call holding cgroup_mutex and the group_rwsem of the leader. Will take
+ * task_lock of each thread in leader's threadgroup individually in turn.
  */
 int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 {

@@ -2030,8 +2030,8 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 	 * step 0: in order to do expensive, possibly blocking operations for
 	 * every thread, we cannot iterate the thread group list, since it needs
 	 * rcu or tasklist locked. instead, build an array of all threads in the
-	 * group - threadgroup_fork_lock prevents new threads from appearing,
-	 * and if threads exit, this will just be an over-estimate.
+	 * group - group_rwsem prevents new threads from appearing, and if
+	 * threads exit, this will just be an over-estimate.
 	 */
 	group_size = get_nr_threads(leader);
 	/* flex_array supports very large thread-groups better than kmalloc. */

@@ -2246,7 +2246,6 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 			cgroup_unlock();
 			return -ESRCH;
 		}
-
 		/*
 		 * even if we're attaching all tasks in the thread group, we
 		 * only need to check permissions on one of them.

@@ -2270,9 +2269,9 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 	}
 
 	if (threadgroup) {
-		threadgroup_fork_write_lock(tsk);
+		threadgroup_lock(tsk);
 		ret = cgroup_attach_proc(cgrp, tsk);
-		threadgroup_fork_write_unlock(tsk);
+		threadgroup_unlock(tsk);
 	} else {
 		ret = cgroup_attach_task(cgrp, tsk);
 	}

index 8e6b6f4..c2af839 100644
@@ -980,7 +980,7 @@ static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
 	sched_autogroup_fork(sig);
 
 #ifdef CONFIG_CGROUPS
-	init_rwsem(&sig->threadgroup_fork_lock);
+	init_rwsem(&sig->group_rwsem);
 #endif
 
 	sig->oom_adj = current->signal->oom_adj;

@@ -1166,7 +1166,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	p->io_context = NULL;
 	p->audit_context = NULL;
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_lock(current);
+		threadgroup_change_begin(current);
 	cgroup_fork(p);
 #ifdef CONFIG_NUMA
 	p->mempolicy = mpol_dup(p->mempolicy);

@@ -1378,7 +1378,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	proc_fork_connector(p);
 	cgroup_post_fork(p);
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_unlock(current);
+		threadgroup_change_done(current);
 	perf_event_fork(p);
 	return p;
 

@@ -1418,7 +1418,7 @@ bad_fork_cleanup_policy:
 bad_fork_cleanup_cgroup:
 #endif
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_unlock(current);
+		threadgroup_change_done(current);
 	cgroup_exit(p, cgroup_callbacks_done);
 	delayacct_tsk_free(p);
 	module_put(task_thread_info(p)->exec_domain->module);

index 08ffab0..ef20cbe 100644
@@ -23,11 +23,10 @@ extern struct files_struct init_files;
 extern struct fs_struct init_fs;
 
 #ifdef CONFIG_CGROUPS
-#define INIT_THREADGROUP_FORK_LOCK(sig)					\
-	.threadgroup_fork_lock =					\
-		__RWSEM_INITIALIZER(sig.threadgroup_fork_lock),
+#define INIT_GROUP_RWSEM(sig)						\
+	.group_rwsem = __RWSEM_INITIALIZER(sig.group_rwsem),
 #else
-#define INIT_THREADGROUP_FORK_LOCK(sig)
+#define INIT_GROUP_RWSEM(sig)
 #endif
 
 #define INIT_SIGNALS(sig) {						\

@@ -46,7 +45,7 @@ extern struct fs_struct init_fs;
 	},								\
 	.cred_guard_mutex =						\
 		 __MUTEX_INITIALIZER(sig.cred_guard_mutex),		\
-	INIT_THREADGROUP_FORK_LOCK(sig)					\
+	INIT_GROUP_RWSEM(sig)						\
 }
 
 extern struct nsproxy init_nsproxy;

index e8acce7..aa47d0f 100644
@@ -635,13 +635,13 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The threadgroup_fork_lock prevents threads from forking with
+	 * The group_rwsem prevents threads from forking with
 	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
 	 * threadgroup-wide operations. It's taken for reading in fork.c in
 	 * copy_process().
 	 * Currently only needed write-side by cgroups.
 	 */
-	struct rw_semaphore threadgroup_fork_lock;
+	struct rw_semaphore group_rwsem;
 #endif
 
 	int oom_adj;		/* OOM kill score adjustment (bit shift) */

@@ -2367,29 +2367,29 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of threadgroup_fork_lock in signal_struct. */
+/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
-static inline void threadgroup_fork_read_lock(struct task_struct *tsk)
+static inline void threadgroup_change_begin(struct task_struct *tsk)
 {
-	down_read(&tsk->signal->threadgroup_fork_lock);
+	down_read(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_read_unlock(struct task_struct *tsk)
+static inline void threadgroup_change_done(struct task_struct *tsk)
 {
-	up_read(&tsk->signal->threadgroup_fork_lock);
+	up_read(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_write_lock(struct task_struct *tsk)
+static inline void threadgroup_lock(struct task_struct *tsk)
 {
-	down_write(&tsk->signal->threadgroup_fork_lock);
+	down_write(&tsk->signal->group_rwsem);
 }
-static inline void threadgroup_fork_write_unlock(struct task_struct *tsk)
+static inline void threadgroup_unlock(struct task_struct *tsk)
 {
-	up_write(&tsk->signal->threadgroup_fork_lock);
+	up_write(&tsk->signal->group_rwsem);
 }
 #else
-static inline void threadgroup_fork_read_lock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_read_unlock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_write_lock(struct task_struct *tsk) {}
-static inline void threadgroup_fork_write_unlock(struct task_struct *tsk) {}
+static inline void threadgroup_change_begin(struct task_struct *tsk) {}
+static inline void threadgroup_change_done(struct task_struct *tsk) {}
+static inline void threadgroup_lock(struct task_struct *tsk) {}
+static inline void threadgroup_unlock(struct task_struct *tsk) {}
 #endif
 
 #ifndef __HAVE_THREAD_FUNCTIONS

index efa5886..f0e099f 100644
@@ -2003,8 +2003,8 @@ static int css_set_prefetch(struct cgroup *cgrp, struct css_set *cg,
  * @cgrp: the cgroup to attach to
  * @leader: the threadgroup leader task_struct of the group to be attached
  *
- * Call holding cgroup_mutex and the threadgroup_fork_lock of the leader. Will
- * take task_lock of each thread in leader's threadgroup individually in turn.
+ * Call holding cgroup_mutex and the group_rwsem of the leader. Will take
+ * task_lock of each thread in leader's threadgroup individually in turn.
  */
 int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 {

@@ -2030,8 +2030,8 @@ int cgroup_attach_proc(struct cgroup *cgrp, struct task_struct *leader)
 	 * step 0: in order to do expensive, possibly blocking operations for
 	 * every thread, we cannot iterate the thread group list, since it needs
 	 * rcu or tasklist locked. instead, build an array of all threads in the
-	 * group - threadgroup_fork_lock prevents new threads from appearing,
-	 * and if threads exit, this will just be an over-estimate.
+	 * group - group_rwsem prevents new threads from appearing, and if
+	 * threads exit, this will just be an over-estimate.
 	 */
 	group_size = get_nr_threads(leader);
 	/* flex_array supports very large thread-groups better than kmalloc. */

@@ -2246,7 +2246,6 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 			cgroup_unlock();
 			return -ESRCH;
 		}
-
 		/*
 		 * even if we're attaching all tasks in the thread group, we
 		 * only need to check permissions on one of them.

@@ -2270,9 +2269,9 @@ static int attach_task_by_pid(struct cgroup *cgrp, u64 pid, bool threadgroup)
 	}
 
 	if (threadgroup) {
-		threadgroup_fork_write_lock(tsk);
+		threadgroup_lock(tsk);
 		ret = cgroup_attach_proc(cgrp, tsk);
-		threadgroup_fork_write_unlock(tsk);
+		threadgroup_unlock(tsk);
 	} else {
 		ret = cgroup_attach_task(cgrp, tsk);
 	}

index 8e6b6f4..c2af839 100644
@@ -980,7 +980,7 @@ static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
 	sched_autogroup_fork(sig);
 
 #ifdef CONFIG_CGROUPS
-	init_rwsem(&sig->threadgroup_fork_lock);
+	init_rwsem(&sig->group_rwsem);
 #endif
 
 	sig->oom_adj = current->signal->oom_adj;

@@ -1166,7 +1166,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	p->io_context = NULL;
 	p->audit_context = NULL;
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_lock(current);
+		threadgroup_change_begin(current);
 	cgroup_fork(p);
 #ifdef CONFIG_NUMA
 	p->mempolicy = mpol_dup(p->mempolicy);

@@ -1378,7 +1378,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	proc_fork_connector(p);
 	cgroup_post_fork(p);
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_unlock(current);
+		threadgroup_change_done(current);
 	perf_event_fork(p);
 	return p;
 

@@ -1418,7 +1418,7 @@ bad_fork_cleanup_policy:
 bad_fork_cleanup_cgroup:
 #endif
 	if (clone_flags & CLONE_THREAD)
-		threadgroup_fork_read_unlock(current);
+		threadgroup_change_done(current);
 	cgroup_exit(p, cgroup_callbacks_done);
 	delayacct_tsk_free(p);
 	module_put(task_thread_info(p)->exec_domain->module);

index aa47d0f..6fb546e 100644
@@ -635,11 +635,12 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting.  fork and exit paths are protected
+	 * with this rwsem using threadgroup_change_begin/end().  Users
+	 * which require threadgroup to remain stable should use
+	 * threadgroup_[un]lock() which also takes care of exec path.
+	 * Currently, cgroup is the only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2367,7 +2368,6 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2377,13 +2377,37 @@ static inline void threadgroup_change_done(struct task_struct *tsk)
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  exec is already synchronized with cread_guard_mutex
+ * which we grab here.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/* exec uses exit for de-threading, grab cred_guard_mutex first */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

index 2913b35..3565f00 100644
@@ -938,6 +938,12 @@ NORET_TYPE void do_exit(long code)
 		schedule();
 	}
 
+	/*
+	 * @tsk's threadgroup is going through changes - lock out users
+	 * which expect stable threadgroup.
+	 */
+	threadgroup_change_begin(tsk);
+
 	exit_irq_thread();
 
 	exit_signals(tsk);  /* sets PF_EXITING */

@@ -1020,10 +1026,6 @@ NORET_TYPE void do_exit(long code)
 		kfree(current->pi_state_cache);
 #endif
 	/*
-	 * Make sure we are holding no locks:
-	 */
-	debug_check_no_locks_held(tsk);
-	/*
 	 * We can do this unlocked here. The futex code uses this flag
 	 * just to verify whether the pi state cleanup has been done
 	 * or not. In the worst case it loops once more.

@@ -1040,6 +1042,13 @@ NORET_TYPE void do_exit(long code)
 
 	preempt_disable();
 	exit_rcu();
+
+	/*
+	 * Release threadgroup and make sure we are holding no locks.
+	 */
+	threadgroup_change_done(tsk);
+	debug_check_no_locks_held(tsk);
+
 	/* causes final put_task_struct in finish_task_switch(). */
 	tsk->state = TASK_DEAD;
 	schedule();

index aa47d0f..6fb546e 100644
@@ -635,11 +635,12 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting.  fork and exit paths are protected
+	 * with this rwsem using threadgroup_change_begin/end().  Users
+	 * which require threadgroup to remain stable should use
+	 * threadgroup_[un]lock() which also takes care of exec path.
+	 * Currently, cgroup is the only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2367,7 +2368,6 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2377,13 +2377,37 @@ static inline void threadgroup_change_done(struct task_struct *tsk)
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  exec is already synchronized with cread_guard_mutex
+ * which we grab here.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/* exec uses exit for de-threading, grab cred_guard_mutex first */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

index 2913b35..3565f00 100644
@@ -938,6 +938,12 @@ NORET_TYPE void do_exit(long code)
 		schedule();
 	}
 
+	/*
+	 * @tsk's threadgroup is going through changes - lock out users
+	 * which expect stable threadgroup.
+	 */
+	threadgroup_change_begin(tsk);
+
 	exit_irq_thread();
 
 	exit_signals(tsk);  /* sets PF_EXITING */

@@ -1020,10 +1026,6 @@ NORET_TYPE void do_exit(long code)
 		kfree(current->pi_state_cache);
 #endif
 	/*
-	 * Make sure we are holding no locks:
-	 */
-	debug_check_no_locks_held(tsk);
-	/*
 	 * We can do this unlocked here. The futex code uses this flag
 	 * just to verify whether the pi state cleanup has been done
 	 * or not. In the worst case it loops once more.

@@ -1040,6 +1042,13 @@ NORET_TYPE void do_exit(long code)
 
 	preempt_disable();
 	exit_rcu();
+
+	/*
+	 * Release threadgroup and make sure we are holding no locks.
+	 */
+	threadgroup_change_done(tsk);
+	debug_check_no_locks_held(tsk);
+
 	/* causes final put_task_struct in finish_task_switch(). */
 	tsk->state = TASK_DEAD;
 	schedule();

index c5acbce..481c5ed 100644
@@ -635,11 +635,12 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting.  fork and exit paths are protected
+	 * with this rwsem using threadgroup_change_begin/end().  Users
+	 * which require threadgroup to remain stable should use
+	 * threadgroup_[un]lock() which also takes care of exec path.
+	 * Currently, cgroup is the only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2374,7 +2375,6 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2384,13 +2384,48 @@ static inline void threadgroup_change_end(struct task_struct *tsk)
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  This excludes most of do_exit() to ensure that, while
+ * locked, tasks belonging to a locked group are not in the process of
+ * deconstruction - they're either alive or dead.
+ *
+ * During exec, a task goes and puts its thread group through unusual
+ * changes.  After de-threading, exclusive access is assumed to resources
+ * which are usually shared by tasks in the same group - e.g. sighand may
+ * be replaced with a new one.  Also, the exec'ing task takes over group
+ * leader role including its pid.  Exclude these changes while locked by
+ * grabbing cred_guard_mutex which is used to synchronize exec path.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/*
+	 * exec uses exit for de-threading nesting group_rwsem inside
+	 * cred_guard_mutex. Grab cred_guard_mutex first.
+	 */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

index d0b7d98..b2cb562 100644
@@ -936,6 +936,14 @@ NORET_TYPE void do_exit(long code)
 		schedule();
 	}
 
+	/*
+	 * @tsk's threadgroup is going through changes - lock out users
+	 * which expect stable threadgroup.  Do this before actually
+	 * starting tearing down @tsk so that locked threadgroup has either
+	 * alive or dead tasks, not something inbetween.
+	 */
+	threadgroup_change_begin(tsk);
+
 	exit_irq_thread();
 
 	exit_signals(tsk);  /* sets PF_EXITING */

@@ -1018,10 +1026,6 @@ NORET_TYPE void do_exit(long code)
 		kfree(current->pi_state_cache);
 #endif
 	/*
-	 * Make sure we are holding no locks:
-	 */
-	debug_check_no_locks_held(tsk);
-	/*
 	 * We can do this unlocked here. The futex code uses this flag
 	 * just to verify whether the pi state cleanup has been done
 	 * or not. In the worst case it loops once more.

@@ -1038,6 +1042,13 @@ NORET_TYPE void do_exit(long code)
 
 	preempt_disable();
 	exit_rcu();
+
+	/*
+	 * Release threadgroup and make sure we are holding no locks.
+	 */
+	threadgroup_change_end(tsk);
+	debug_check_no_locks_held(tsk);
+
 	/* causes final put_task_struct in finish_task_switch(). */
 	tsk->state = TASK_DEAD;
 	schedule();

index c5acbce..481c5ed 100644
@@ -635,11 +635,12 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting.  fork and exit paths are protected
+	 * with this rwsem using threadgroup_change_begin/end().  Users
+	 * which require threadgroup to remain stable should use
+	 * threadgroup_[un]lock() which also takes care of exec path.
+	 * Currently, cgroup is the only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2374,7 +2375,6 @@ static inline void unlock_task_sighand(struct task_struct *tsk,
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2384,13 +2384,48 @@ static inline void threadgroup_change_end(struct task_struct *tsk)
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  This excludes most of do_exit() to ensure that, while
+ * locked, tasks belonging to a locked group are not in the process of
+ * deconstruction - they're either alive or dead.
+ *
+ * During exec, a task goes and puts its thread group through unusual
+ * changes.  After de-threading, exclusive access is assumed to resources
+ * which are usually shared by tasks in the same group - e.g. sighand may
+ * be replaced with a new one.  Also, the exec'ing task takes over group
+ * leader role including its pid.  Exclude these changes while locked by
+ * grabbing cred_guard_mutex which is used to synchronize exec path.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/*
+	 * exec uses exit for de-threading nesting group_rwsem inside
+	 * cred_guard_mutex. Grab cred_guard_mutex first.
+	 */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

index d0b7d98..b2cb562 100644
@@ -936,6 +936,14 @@ NORET_TYPE void do_exit(long code)
 		schedule();
 	}
 
+	/*
+	 * @tsk's threadgroup is going through changes - lock out users
+	 * which expect stable threadgroup.  Do this before actually
+	 * starting tearing down @tsk so that locked threadgroup has either
+	 * alive or dead tasks, not something inbetween.
+	 */
+	threadgroup_change_begin(tsk);
+
 	exit_irq_thread();
 
 	exit_signals(tsk);  /* sets PF_EXITING */

@@ -1018,10 +1026,6 @@ NORET_TYPE void do_exit(long code)
 		kfree(current->pi_state_cache);
 #endif
 	/*
-	 * Make sure we are holding no locks:
-	 */
-	debug_check_no_locks_held(tsk);
-	/*
 	 * We can do this unlocked here. The futex code uses this flag
 	 * just to verify whether the pi state cleanup has been done
 	 * or not. In the worst case it loops once more.

@@ -1038,6 +1042,13 @@ NORET_TYPE void do_exit(long code)
 
 	preempt_disable();
 	exit_rcu();
+
+	/*
+	 * Release threadgroup and make sure we are holding no locks.
+	 */
+	threadgroup_change_end(tsk);
+	debug_check_no_locks_held(tsk);
+
 	/* causes final put_task_struct in finish_task_switch(). */
 	tsk->state = TASK_DEAD;
 	schedule();

Index: work/include/linux/sched.h
===================================================================
@@ -635,11 +635,13 @@ struct signal_struct {
 #endif
 #ifdef CONFIG_CGROUPS
 	/*
-	 * The group_rwsem prevents threads from forking with
-	 * CLONE_THREAD while held for writing. Use this for fork-sensitive
-	 * threadgroup-wide operations. It's taken for reading in fork.c in
-	 * copy_process().
-	 * Currently only needed write-side by cgroups.
+	 * group_rwsem prevents new tasks from entering the threadgroup and
+	 * member tasks from exiting,a more specifically, setting of
+	 * PF_EXITING.  fork and exit paths are protected with this rwsem
+	 * using threadgroup_change_begin/end().  Users which require
+	 * threadgroup to remain stable should use threadgroup_[un]lock()
+	 * which also takes care of exec path.  Currently, cgroup is the
+	 * only user.
 	 */
 	struct rw_semaphore group_rwsem;
 #endif

@@ -2374,7 +2376,6 @@ static inline void unlock_task_sighand(s
 	spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
 }
 
-/* See the declaration of group_rwsem in signal_struct. */
 #ifdef CONFIG_CGROUPS
 static inline void threadgroup_change_begin(struct task_struct *tsk)
 {

@@ -2384,13 +2385,47 @@ static inline void threadgroup_change_en
 {
 	up_read(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_lock - lock threadgroup
+ * @tsk: member task of the threadgroup to lock
+ *
+ * Lock the threadgroup @tsk belongs to.  No new task is allowed to enter
+ * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
+ * perform exec.  This is useful for cases where the threadgroup needs to
+ * stay stable across blockable operations.
+ *
+ * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
+ * synchronization.  While held, no new task will be added to threadgroup
+ * and no existing live task will have its PF_EXITING set.
+ *
+ * During exec, a task goes and puts its thread group through unusual
+ * changes.  After de-threading, exclusive access is assumed to resources
+ * which are usually shared by tasks in the same group - e.g. sighand may
+ * be replaced with a new one.  Also, the exec'ing task takes over group
+ * leader role including its pid.  Exclude these changes while locked by
+ * grabbing cred_guard_mutex which is used to synchronize exec path.
+ */
 static inline void threadgroup_lock(struct task_struct *tsk)
 {
+	/*
+	 * exec uses exit for de-threading nesting group_rwsem inside
+	 * cred_guard_mutex. Grab cred_guard_mutex first.
+	 */
+	mutex_lock(&tsk->signal->cred_guard_mutex);
 	down_write(&tsk->signal->group_rwsem);
 }
+
+/**
+ * threadgroup_unlock - unlock threadgroup
+ * @tsk: member task of the threadgroup to unlock
+ *
+ * Reverse threadgroup_lock().
+ */
 static inline void threadgroup_unlock(struct task_struct *tsk)
 {
 	up_write(&tsk->signal->group_rwsem);
+	mutex_unlock(&tsk->signal->cred_guard_mutex);
 }
 #else
 static inline void threadgroup_change_begin(struct task_struct *tsk) {}

Index: work/kernel/signal.c
===================================================================
@@ -2359,8 +2359,15 @@ void exit_signals(struct task_struct *ts
 	int group_stop = 0;
 	sigset_t unblocked;
 
+	/*
+	 * @tsk is about to have PF_EXITING set - lock out users which
+	 * expect stable threadgroup.
+	 */
+	threadgroup_change_begin(tsk);
+
 	if (thread_group_empty(tsk) || signal_group_exit(tsk->signal)) {
 		tsk->flags |= PF_EXITING;
+		threadgroup_change_end(tsk);
 		return;
 	}
 

@@ -2370,6 +2377,9 @@ void exit_signals(struct task_struct *ts
 	 * see wants_signal(), do_signal_stop().
 	 */
 	tsk->flags |= PF_EXITING;
+
+	threadgroup_change_end(tsk);
+
 	if (!signal_pending(tsk))
 		goto out;