[patch 3/7] vfs: mountinfo: add mount ID
authorMiklos Szeredi <mszeredi@suse.cz>
Wed, 26 Mar 2008 21:11:34 +0000 (22:11 +0100)
committerAl Viro <viro@zeniv.linux.org.uk>
Wed, 23 Apr 2008 04:04:45 +0000 (00:04 -0400)
Add a unique ID to each vfsmount using the IDR infrastructure.  The
identifiers are reused after the vfsmount is freed.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
fs/namespace.c
include/linux/mount.h

index 1bf302d0478b9b52d5a302278febdf40d474eda5..8ca6317cb401b303fde6f3d6fda73b33ecc617d8 100644 (file)
@@ -27,6 +27,7 @@
 #include <linux/mount.h>
 #include <linux/ramfs.h>
 #include <linux/log2.h>
+#include <linux/idr.h>
 #include <asm/uaccess.h>
 #include <asm/unistd.h>
 #include "pnode.h"
@@ -39,6 +40,7 @@
 __cacheline_aligned_in_smp DEFINE_SPINLOCK(vfsmount_lock);
 
 static int event;
+static DEFINE_IDA(mnt_id_ida);
 
 static struct list_head *mount_hashtable __read_mostly;
 static struct kmem_cache *mnt_cache __read_mostly;
@@ -58,10 +60,41 @@ static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
 
 #define MNT_WRITER_UNDERFLOW_LIMIT -(1<<16)
 
+/* allocation is serialized by namespace_sem */
+static int mnt_alloc_id(struct vfsmount *mnt)
+{
+       int res;
+
+retry:
+       ida_pre_get(&mnt_id_ida, GFP_KERNEL);
+       spin_lock(&vfsmount_lock);
+       res = ida_get_new(&mnt_id_ida, &mnt->mnt_id);
+       spin_unlock(&vfsmount_lock);
+       if (res == -EAGAIN)
+               goto retry;
+
+       return res;
+}
+
+static void mnt_free_id(struct vfsmount *mnt)
+{
+       spin_lock(&vfsmount_lock);
+       ida_remove(&mnt_id_ida, mnt->mnt_id);
+       spin_unlock(&vfsmount_lock);
+}
+
 struct vfsmount *alloc_vfsmnt(const char *name)
 {
        struct vfsmount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL);
        if (mnt) {
+               int err;
+
+               err = mnt_alloc_id(mnt);
+               if (err) {
+                       kmem_cache_free(mnt_cache, mnt);
+                       return NULL;
+               }
+
                atomic_set(&mnt->mnt_count, 1);
                INIT_LIST_HEAD(&mnt->mnt_hash);
                INIT_LIST_HEAD(&mnt->mnt_child);
@@ -353,6 +386,7 @@ EXPORT_SYMBOL(simple_set_mnt);
 void free_vfsmnt(struct vfsmount *mnt)
 {
        kfree(mnt->mnt_devname);
+       mnt_free_id(mnt);
        kmem_cache_free(mnt_cache, mnt);
 }
 
index 87b24cea18635d110317ddd2400f0ca07c0200a2..f0dfb17ffccce344c67b5e0569bc8a0bc132272c 100644 (file)
@@ -56,6 +56,7 @@ struct vfsmount {
        struct list_head mnt_slave;     /* slave list entry */
        struct vfsmount *mnt_master;    /* slave is on master->mnt_slave_list */
        struct mnt_namespace *mnt_ns;   /* containing namespace */
+       int mnt_id;                     /* mount identifier */
        /*
         * We put mnt_count & mnt_expiry_mark at the end of struct vfsmount
         * to let these frequently modified fields in a separate cache line