[PATCH v3 1/3] cgroup: fix mount failure in a corner case

[PATCH v3 1/3] cgroup: fix mount failure in a corner case

[Li Zefan]

  # cat test.sh
  #! /bin/bash

  mount -t cgroup -o cpu xxx /cgroup
  umount /cgroup

  mount -t cgroup -o cpu,cpuacct xxx /cgroup
  umount /cgroup
  # ./test.sh
  mount: xxx already mounted or /cgroup busy
  mount: according to mtab, xxx is already mounted on /cgroup

It's because the cgroupfs_root of the first mount was under destruction
asynchronously.

Fix this by delaying and then retrying mount for this case.

v3:
- put the refcnt immediately after getting it. (Tejun)

v2:
- use percpu_ref_tryget_live() rather that introducing
  percpu_ref_alive(). (Tejun)
- adjust comment.

Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
---
 kernel/cgroup.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 1c65f24..d3662ac 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -1648,10 +1648,12 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
 	struct dentry *dentry;
 	int ret;
+	int i;
 	bool new_sb;
 
 	/*
@@ -1677,6 +1679,25 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		goto out_unlock;
 	}
 
+	/*
+	 * Destruction of cgroup root is asynchronous, so we may fail to
+	 * mount a cgroupfs if it immediately follows a umount. Let's wait
+	 * a little bit and retry.
+	 */
+	for_each_subsys(ss, i) {
+		if (!(opts.subsys_mask & (1 << i)) ||
+		    ss->root == &cgrp_dfl_root)
+			continue;
+
+		if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
+			mutex_unlock(&cgroup_mutex);
+			msleep(10);
+			ret = restart_syscall();
+			goto out_free;
+		}
+		cgroup_put(&ss->root->cgrp);
+	}
+
 	for_each_root(root) {
 		bool name_match = false;
 
-- 
1.8.0.2

[PATCH v3 2/3] kernfs: introduce kernfs_pin_sb()

[Li Zefan]

kernfs_pin_sb() tries to get a refcnt of the superblock.

This will be used by cgroupfs.

v2:
- make kernfs_pin_sb() return the superblock.
- drop kernfs_drop_sb().

[ This is a prerequisite for a bugfix. ]
Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Acked-by: Greg Kroah-Hartman <gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r@public.gmane.org>
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
---
 fs/kernfs/mount.c      | 27 +++++++++++++++++++++++++++
 include/linux/kernfs.h |  1 +
 2 files changed, 28 insertions(+)

diff --git a/fs/kernfs/mount.c b/fs/kernfs/mount.c
index f25a7c0..616c5c4 100644
--- a/fs/kernfs/mount.c
+++ b/fs/kernfs/mount.c
@@ -210,6 +210,33 @@ void kernfs_kill_sb(struct super_block *sb)
 	kernfs_put(root_kn);
 }
 
+/**
+ * kernfs_pin_sb: try to pin the superblock associated with a kernfs_root
+ * @kernfs_root: the kernfs_root in question
+ * @ns: the namespace tag
+ *
+ * Pin the superblock so the superblock won't be destroyed in subsequent
+ * operations. Return NULL if there's no superblock associated to this
+ * kernfs_root, or -EINVAL if the superblock is being freed.
+ */
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns)
+{
+	struct kernfs_super_info *info;
+	struct super_block *sb = NULL;
+
+	mutex_lock(&kernfs_mutex);
+	list_for_each_entry(info, &root->supers, node) {
+		if (info->ns == ns) {
+			sb = info->sb;
+			if (!atomic_inc_not_zero(&info->sb->s_active))
+				sb = ERR_PTR(-EINVAL);
+			break;
+		}
+	}
+	mutex_unlock(&kernfs_mutex);
+	return sb;
+}
+
 void __init kernfs_init(void)
 {
 	kernfs_node_cache = kmem_cache_create("kernfs_node_cache",
diff --git a/include/linux/kernfs.h b/include/linux/kernfs.h
index 589318b..9096296 100644
--- a/include/linux/kernfs.h
+++ b/include/linux/kernfs.h
@@ -287,6 +287,7 @@ struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags,
 			       struct kernfs_root *root, bool *new_sb_created,
 			       const void *ns);
 void kernfs_kill_sb(struct super_block *sb);
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns);
 
 void kernfs_init(void);
 
-- 
1.8.0.2

Re: [PATCH v3 2/3] kernfs: introduce kernfs_pin_sb()

[Tejun Heo]

Applied to cgroup/for-3.16-fixes with minor updates.  Thanks.
------ 8< ------
From 4e26445faad366d67d7723622bf6a60a6f0f5993 Mon Sep 17 00:00:00 2001
From: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Date: Mon, 30 Jun 2014 11:50:28 +0800
Subject: [PATCH 2/3] kernfs: introduce kernfs_pin_sb()

kernfs_pin_sb() tries to get a refcnt of the superblock.

This will be used by cgroupfs.

v2:
- make kernfs_pin_sb() return the superblock.
- drop kernfs_drop_sb().

tj: Updated the comment a bit.

[ This is a prerequisite for a bugfix. ]
Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Acked-by: Greg Kroah-Hartman <gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r@public.gmane.org>
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Signed-off-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 fs/kernfs/mount.c      | 30 ++++++++++++++++++++++++++++++
 include/linux/kernfs.h |  1 +
 2 files changed, 31 insertions(+)

diff --git a/fs/kernfs/mount.c b/fs/kernfs/mount.c
index d171b98..f973ae9 100644
--- a/fs/kernfs/mount.c
+++ b/fs/kernfs/mount.c
@@ -211,6 +211,36 @@ void kernfs_kill_sb(struct super_block *sb)
 	kernfs_put(root_kn);
 }
 
+/**
+ * kernfs_pin_sb: try to pin the superblock associated with a kernfs_root
+ * @kernfs_root: the kernfs_root in question
+ * @ns: the namespace tag
+ *
+ * Pin the superblock so the superblock won't be destroyed in subsequent
+ * operations.  This can be used to block ->kill_sb() which may be useful
+ * for kernfs users which dynamically manage superblocks.
+ *
+ * Returns NULL if there's no superblock associated to this kernfs_root, or
+ * -EINVAL if the superblock is being freed.
+ */
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns)
+{
+	struct kernfs_super_info *info;
+	struct super_block *sb = NULL;
+
+	mutex_lock(&kernfs_mutex);
+	list_for_each_entry(info, &root->supers, node) {
+		if (info->ns == ns) {
+			sb = info->sb;
+			if (!atomic_inc_not_zero(&info->sb->s_active))
+				sb = ERR_PTR(-EINVAL);
+			break;
+		}
+	}
+	mutex_unlock(&kernfs_mutex);
+	return sb;
+}
+
 void __init kernfs_init(void)
 {
 	kernfs_node_cache = kmem_cache_create("kernfs_node_cache",
diff --git a/include/linux/kernfs.h b/include/linux/kernfs.h
index 17aa1cc..20f4935 100644
--- a/include/linux/kernfs.h
+++ b/include/linux/kernfs.h
@@ -304,6 +304,7 @@ struct dentry *kernfs_mount_ns(struct file_system_type *fs_type, int flags,
 			       struct kernfs_root *root, unsigned long magic,
 			       bool *new_sb_created, const void *ns);
 void kernfs_kill_sb(struct super_block *sb);
+struct super_block *kernfs_pin_sb(struct kernfs_root *root, const void *ns);
 
 void kernfs_init(void);
 
-- 
1.9.3

[PATCH v3 3/3] cgroup: fix a race between cgroup_mount() and cgroup_kill_sb()

[Li Zefan]

We've converted cgroup to kernfs so cgroup won't be intertwined with
vfs objects and locking, but there are dark areas.

Run two instances of this script concurrently:

    for ((; ;))
    {
    	mount -t cgroup -o cpuacct xxx /cgroup
    	umount /cgroup
    }

After a while, I saw two mount processes were stuck at retrying, because
they were waiting for a subsystem to become free, but the root associated
with this subsystem never got freed.

This can happen, if thread A is in the process of killing superblock but
hasn't called percpu_ref_kill(), and at this time thread B is mounting
the same cgroup root and finds the root in the root list and performs
percpu_ref_try_get().

To fix this, we try to increase both the refcnt of the superblock and the
percpu refcnt of cgroup root.

v2:
- we should try to get both the superblock refcnt and cgroup_root refcnt,
  because cgroup_root may have no superblock assosiated with it.
- adjust/add comments.

Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
---
 kernel/cgroup.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index d3662ac..11e40cf 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -1655,6 +1655,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 	int ret;
 	int i;
 	bool new_sb;
+	struct super_block *sb = NULL;
 
 	/*
 	 * The first time anyone tries to mount a cgroup, enable the list
@@ -1739,14 +1740,18 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 
 		/*
 		 * A root's lifetime is governed by its root cgroup.
-		 * tryget_live failure indicate that the root is being
-		 * destroyed.  Wait for destruction to complete so that the
-		 * subsystems are free.  We can use wait_queue for the wait
-		 * but this path is super cold.  Let's just sleep for a bit
-		 * and retry.
+		 * pin_sb and tryget_live failure indicate that the root is
+		 * being destroyed.  Wait for destruction to complete so that
+		 * the subsystems are free.  We can use wait_queue for the
+		 * wait but this path is super cold.  Let's just sleep for
+		 * a bit and retry.
 		 */
-		if (!percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
+		sb = kernfs_pin_sb(root->kf_root, NULL);
+		if (IS_ERR(sb) ||
+		    !percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
 			mutex_unlock(&cgroup_mutex);
+			if (!IS_ERR_OR_NULL(sb))
+				deactivate_super(sb);
 			msleep(10);
 			ret = restart_syscall();
 			goto out_free;
@@ -1790,6 +1795,17 @@ out_free:
 	dentry = kernfs_mount(fs_type, flags, root->kf_root, &new_sb);
 	if (IS_ERR(dentry) || !new_sb)
 		cgroup_put(&root->cgrp);
+
+	if (sb) {
+		/*
+		 * On success kernfs_mount() returns with sb->s_umount held,
+		 * but kernfs_mount() also increases the superblock's refcnt,
+		 * so calling deactivate_super() to drop the refcnt we got when
+		 * looking up cgroup root won't acquire sb->s_umount again.
+		 */
+		WARN_ON(new_sb);
+		deactivate_super(sb);
+	}
 	return dentry;
 }
 
-- 
1.8.0.2

Re: [PATCH v3 3/3] cgroup: fix a race between cgroup_mount() and cgroup_kill_sb()

[Tejun Heo]

Applied to cgroup/for-3.16-fixes with minor updates.  Thanks.
------ 8< ------
From 3a32bd72d77058d768dbb38183ad517f720dd1bc Mon Sep 17 00:00:00 2001
From: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Date: Mon, 30 Jun 2014 11:50:59 +0800
Subject: [PATCH 3/3] cgroup: fix a race between cgroup_mount() and
 cgroup_kill_sb()

We've converted cgroup to kernfs so cgroup won't be intertwined with
vfs objects and locking, but there are dark areas.

Run two instances of this script concurrently:

    for ((; ;))
    {
    	mount -t cgroup -o cpuacct xxx /cgroup
    	umount /cgroup
    }

After a while, I saw two mount processes were stuck at retrying, because
they were waiting for a subsystem to become free, but the root associated
with this subsystem never got freed.

This can happen, if thread A is in the process of killing superblock but
hasn't called percpu_ref_kill(), and at this time thread B is mounting
the same cgroup root and finds the root in the root list and performs
percpu_ref_try_get().

To fix this, we try to increase both the refcnt of the superblock and the
percpu refcnt of cgroup root.

v2:
- we should try to get both the superblock refcnt and cgroup_root refcnt,
  because cgroup_root may have no superblock assosiated with it.
- adjust/add comments.

tj: Updated comments.  Renamed @sb to @pinned_sb.

Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Signed-off-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 kernel/cgroup.c | 33 ++++++++++++++++++++++++++-------
 1 file changed, 26 insertions(+), 7 deletions(-)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 6406866..70776ae 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -1648,6 +1648,7 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct super_block *pinned_sb = NULL;
 	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
@@ -1740,15 +1741,23 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		}
 
 		/*
-		 * A root's lifetime is governed by its root cgroup.
-		 * tryget_live failure indicate that the root is being
-		 * destroyed.  Wait for destruction to complete so that the
-		 * subsystems are free.  We can use wait_queue for the wait
-		 * but this path is super cold.  Let's just sleep for a bit
-		 * and retry.
+		 * We want to reuse @root whose lifetime is governed by its
+		 * ->cgrp.  Let's check whether @root is alive and keep it
+		 * that way.  As cgroup_kill_sb() can happen anytime, we
+		 * want to block it by pinning the sb so that @root doesn't
+		 * get killed before mount is complete.
+		 *
+		 * With the sb pinned, tryget_live can reliably indicate
+		 * whether @root can be reused.  If it's being killed,
+		 * drain it.  We can use wait_queue for the wait but this
+		 * path is super cold.  Let's just sleep a bit and retry.
 		 */
-		if (!percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
+		pinned_sb = kernfs_pin_sb(root->kf_root, NULL);
+		if (IS_ERR(pinned_sb) ||
+		    !percpu_ref_tryget_live(&root->cgrp.self.refcnt)) {
 			mutex_unlock(&cgroup_mutex);
+			if (!IS_ERR_OR_NULL(pinned_sb))
+				deactivate_super(pinned_sb);
 			msleep(10);
 			ret = restart_syscall();
 			goto out_free;
@@ -1793,6 +1802,16 @@ out_free:
 				CGROUP_SUPER_MAGIC, &new_sb);
 	if (IS_ERR(dentry) || !new_sb)
 		cgroup_put(&root->cgrp);
+
+	/*
+	 * If @pinned_sb, we're reusing an existing root and holding an
+	 * extra ref on its sb.  Mount is complete.  Put the extra ref.
+	 */
+	if (pinned_sb) {
+		WARN_ON(new_sb);
+		deactivate_super(pinned_sb);
+	}
+
 	return dentry;
 }
 
-- 
1.9.3

Re: [PATCH v3 3/3] cgroup: fix a race between cgroup_mount() and cgroup_kill_sb()

[Tejun Heo]

Hello, Li.

I applied the three patches with some updates (mostly comments).
Dynamic root management is turning out to be pretty ugly but I think
we can clean it up a bit.

On Mon, Jun 30, 2014 at 11:50:59AM +0800, Li Zefan wrote:
> @@ -1790,6 +1795,17 @@ out_free:
>  	dentry = kernfs_mount(fs_type, flags, root->kf_root, &new_sb);
>  	if (IS_ERR(dentry) || !new_sb)
>  		cgroup_put(&root->cgrp);
> +
> +	if (sb) {
> +		/*
> +		 * On success kernfs_mount() returns with sb->s_umount held,
> +		 * but kernfs_mount() also increases the superblock's refcnt,
> +		 * so calling deactivate_super() to drop the refcnt we got when
> +		 * looking up cgroup root won't acquire sb->s_umount again.
> +		 */
> +		WARN_ON(new_sb);
> +		deactivate_super(sb);
> +	}

So, @new_sb and @sb indicate the same conditions now.  When we reuse
an existing root, we always reuse its sb too which means that we at
least no longer need the @new_sb ugliness.  I actually kinda prefer it
as this means that the ugliness is confined to one ugly function -
kernfs_pin_sb() - instead of the generic kernfs_mount().  Can you
please prepare patches to clean these up?  I'll pull in for-3.16-fixes
into for-3.17 and apply the cleanup on top.

Thanks!

-- 
tejun

Re: [PATCH v3 1/3] cgroup: fix mount failure in a corner case

[Tejun Heo]

Applied with minor updates.

Thanks.
------- 8< -------
From 970317aa48c6ef66cd023c039c2650c897bad927 Mon Sep 17 00:00:00 2001
From: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Date: Mon, 30 Jun 2014 11:49:58 +0800
Subject: [PATCH 1/3] cgroup: fix mount failure in a corner case

  # cat test.sh
  #! /bin/bash

  mount -t cgroup -o cpu xxx /cgroup
  umount /cgroup

  mount -t cgroup -o cpu,cpuacct xxx /cgroup
  umount /cgroup
  # ./test.sh
  mount: xxx already mounted or /cgroup busy
  mount: according to mtab, xxx is already mounted on /cgroup

It's because the cgroupfs_root of the first mount was under destruction
asynchronously.

Fix this by delaying and then retrying mount for this case.

v3:
- put the refcnt immediately after getting it. (Tejun)

v2:
- use percpu_ref_tryget_live() rather that introducing
  percpu_ref_alive(). (Tejun)
- adjust comment.

tj: Updated the comment a bit.

Cc: <stable-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> # 3.15
Signed-off-by: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
Signed-off-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
 kernel/cgroup.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index d9a8be9..6406866 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -1648,10 +1648,12 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 			 int flags, const char *unused_dev_name,
 			 void *data)
 {
+	struct cgroup_subsys *ss;
 	struct cgroup_root *root;
 	struct cgroup_sb_opts opts;
 	struct dentry *dentry;
 	int ret;
+	int i;
 	bool new_sb;
 
 	/*
@@ -1677,6 +1679,27 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
 		goto out_unlock;
 	}
 
+	/*
+	 * Destruction of cgroup root is asynchronous, so subsystems may
+	 * still be dying after the previous unmount.  Let's drain the
+	 * dying subsystems.  We just need to ensure that the ones
+	 * unmounted previously finish dying and don't care about new ones
+	 * starting.  Testing ref liveliness is good enough.
+	 */
+	for_each_subsys(ss, i) {
+		if (!(opts.subsys_mask & (1 << i)) ||
+		    ss->root == &cgrp_dfl_root)
+			continue;
+
+		if (!percpu_ref_tryget_live(&ss->root->cgrp.self.refcnt)) {
+			mutex_unlock(&cgroup_mutex);
+			msleep(10);
+			ret = restart_syscall();
+			goto out_free;
+		}
+		cgroup_put(&ss->root->cgrp);
+	}
+
 	for_each_root(root) {
 		bool name_match = false;
 
-- 
1.9.3