namespaces: ipc namespaces: implement support for posix msqueues

Implement multiple mounts of the mqueue file system, and link it to usage
of CLONE_NEWIPC.

Each ipc ns has a corresponding mqueuefs superblock.  When a user does
clone(CLONE_NEWIPC) or unshare(CLONE_NEWIPC), the unshare will cause an
internal mount of a new mqueuefs sb linked to the new ipc ns.

When a user does 'mount -t mqueue mqueue /dev/mqueue', he mounts the
mqueuefs superblock.

Posix message queues can be worked with both through the mq_* system calls
(see mq_overview(7)), and through the VFS through the mqueue mount.  Any
usage of mq_open() and friends will work with the acting task's ipc
namespace.  Any actions through the VFS will work with the mqueuefs in
which the file was created.  So if a user doesn't remount mqueuefs after
unshare(CLONE_NEWIPC), mq_open("/ab") will not be reflected in "ls
/dev/mqueue".

If task a mounts mqueue for ipc_ns:1, then clones task b with a new ipcns,
ipcns:2, and then task a is the last task in ipc_ns:1 to exit, then (1)
ipc_ns:1 will be freed, (2) it's superblock will live on until task b
umounts the corresponding mqueuefs, and vfs actions will continue to
succeed, but (3) sb->s_fs_info will be NULL for the sb corresponding to
the deceased ipc_ns:1.

To make this happen, we must protect the ipc reference count when

a) a task exits and drops its ipcns->count, since it might be dropping
   it to 0 and freeing the ipcns

b) a task accesses the ipcns through its mqueuefs interface, since it
   bumps the ipcns refcount and might race with the last task in the ipcns
   exiting.

So the kref is changed to an atomic_t so we can use
atomic_dec_and_lock(&ns->count,mq_lock), and every access to the ipcns
through ns = mqueuefs_sb->s_fs_info is protected by the same lock.

Signed-off-by: Cedric Le Goater <clg@fr.ibm.com>
Signed-off-by: Serge E. Hallyn <serue@us.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 35 insertions, 6 deletions

diff --git a/ipc/namespace.c b/ipc/namespace.c
index 4b4dc6d847f..4a5e752a927 100644
--- a/ipc/namespace.c
+++ b/ipc/namespace.c
@@ -9,23 +9,31 @@
 #include <linux/rcupdate.h>
 #include <linux/nsproxy.h>
 #include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/mount.h>
 
 #include "util.h"
 
 static struct ipc_namespace *clone_ipc_ns(struct ipc_namespace *old_ns)
 {
 	struct ipc_namespace *ns;
+	int err;
 
 	ns = kmalloc(sizeof(struct ipc_namespace), GFP_KERNEL);
 	if (ns == NULL)
 		return ERR_PTR(-ENOMEM);
 
+	atomic_set(&ns->count, 1);
+	err = mq_init_ns(ns);
+	if (err) {
+		kfree(ns);
+		return ERR_PTR(err);
+	}
 	atomic_inc(&nr_ipc_ns);
 
 	sem_init_ns(ns);
 	msg_init_ns(ns);
 	shm_init_ns(ns);
-	mq_init_ns(ns);
 
 	/*
 	 * msgmni has already been computed for the new ipc ns.
@@ -35,7 +43,6 @@ static struct ipc_namespace *clone_ipc_ns(struct ipc_namespace *old_ns)
 	ipcns_notify(IPCNS_CREATED);
 	register_ipcns_notifier(ns);
 
-	kref_init(&ns->kref);
 	return ns;
 }
 
@@ -85,11 +92,34 @@ void free_ipcs(struct ipc_namespace *ns, struct ipc_ids *ids,
 	up_write(&ids->rw_mutex);
 }
 
-void free_ipc_ns(struct kref *kref)
+/*
+ * put_ipc_ns - drop a reference to an ipc namespace.
+ * @ns: the namespace to put
+ *
+ * If this is the last task in the namespace exiting, and
+ * it is dropping the refcount to 0, then it can race with
+ * a task in another ipc namespace but in a mounts namespace
+ * which has this ipcns's mqueuefs mounted, doing some action
+ * with one of the mqueuefs files.  That can raise the refcount.
+ * So dropping the refcount, and raising the refcount when
+ * accessing it through the VFS, are protected with mq_lock.
+ *
+ * (Clearly, a task raising the refcount on its own ipc_ns
+ * needn't take mq_lock since it can't race with the last task
+ * in the ipcns exiting).
+ */
+void put_ipc_ns(struct ipc_namespace *ns)
 {
-	struct ipc_namespace *ns;
+	if (atomic_dec_and_lock(&ns->count, &mq_lock)) {
+		mq_clear_sbinfo(ns);
+		spin_unlock(&mq_lock);
+		mq_put_mnt(ns);
+		free_ipc_ns(ns);
+	}
+}
 
-	ns = container_of(kref, struct ipc_namespace, kref);
+void free_ipc_ns(struct ipc_namespace *ns)
+{
 	/*
 	 * Unregistering the hotplug notifier at the beginning guarantees
 	 * that the ipc namespace won't be freed while we are inside the
@@ -102,7 +132,6 @@ void free_ipc_ns(struct kref *kref)
 	sem_exit_ns(ns);
 	msg_exit_ns(ns);
 	shm_exit_ns(ns);
-	mq_exit_ns(ns);
 	kfree(ns);
 	atomic_dec(&nr_ipc_ns);