fs: scale mntget/mntput

The problem that this patch aims to fix is vfsmount refcounting scalability.
We need to take a reference on the vfsmount for every successful path lookup,
which often go to the same mount point.

The fundamental difficulty is that a "simple" reference count can never be made
scalable, because any time a reference is dropped, we must check whether that
was the last reference. To do that requires communication with all other CPUs
that may have taken a reference count.

We can make refcounts more scalable in a couple of ways, involving keeping
distributed counters, and checking for the global-zero condition less
frequently.

- check the global sum once every interval (this will delay zero detection
  for some interval, so it's probably a showstopper for vfsmounts).

- keep a local count and only taking the global sum when local reaches 0 (this
  is difficult for vfsmounts, because we can't hold preempt off for the life of
  a reference, so a counter would need to be per-thread or tied strongly to a
  particular CPU which requires more locking).

- keep a local difference of increments and decrements, which allows us to sum
  the total difference and hence find the refcount when summing all CPUs. Then,
  keep a single integer "long" refcount for slow and long lasting references,
  and only take the global sum of local counters when the long refcount is 0.

This last scheme is what I implemented here. Attached mounts and process root
and working directory references are "long" references, and everything else is
a short reference.

This allows scalable vfsmount references during path walking over mounted
subtrees and unattached (lazy umounted) mounts with processes still running
in them.

This results in one fewer atomic op in the fastpath: mntget is now just a
per-CPU inc, rather than an atomic inc; and mntput just requires a spinlock
and non-atomic decrement in the common case. However code is otherwise bigger
and heavier, so single threaded performance is basically a wash.

Signed-off-by: Nick Piggin <npiggin@kernel.dk>

13 files changed, 283 insertions, 98 deletions

diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
index 5a24f40bb48e..f099b82703d8 100644
--- a/arch/ia64/kernel/perfmon.c
+++ b/arch/ia64/kernel/perfmon.c
@@ -1542,7 +1542,7 @@ pfm_exit_smpl_buffer(pfm_buffer_fmt_t *fmt)
  * any operations on the root directory. However, we need a non-trivial
  * d_name - pfm: will go nicely and kill the special-casing in procfs.
  */
-static struct vfsmount *pfmfs_mnt;
+static struct vfsmount *pfmfs_mnt __read_mostly;
 
 static int __init
 init_pfm_fs(void)
diff --git a/drivers/mtd/mtdchar.c b/drivers/mtd/mtdchar.c
index 4759d827e8c7..f511dd15fd31 100644
--- a/drivers/mtd/mtdchar.c
+++ b/drivers/mtd/mtdchar.c
@@ -1201,7 +1201,7 @@ err_unregister_chdev:
 static void __exit cleanup_mtdchar(void)
 {
         unregister_mtd_user(&mtdchar_notifier);
-        mntput(mtd_inode_mnt);
+        mntput_long(mtd_inode_mnt);
         unregister_filesystem(&mtd_inodefs_type);
         __unregister_chrdev(MTD_CHAR_MAJOR, 0, 1 << MINORBITS, "mtd");
 }
diff --git a/fs/anon_inodes.c b/fs/anon_inodes.c
index 9d92b33da8a0..5fd38112a6ca 100644
--- a/fs/anon_inodes.c
+++ b/fs/anon_inodes.c
@@ -232,7 +232,7 @@ static int __init anon_inode_init(void)
         return 0;
 
 err_mntput:
-        mntput(anon_inode_mnt);
+        mntput_long(anon_inode_mnt);
 err_unregister_filesystem:
         unregister_filesystem(&anon_inode_fs_type);
 err_exit:
diff --git a/fs/fs_struct.c b/fs/fs_struct.c
index 60b8531f41c5..68ca487bedb1 100644
--- a/fs/fs_struct.c
+++ b/fs/fs_struct.c
@@ -17,11 +17,11 @@ void set_fs_root(struct fs_struct *fs, struct path *path)
         write_seqcount_begin(&fs->seq);
         old_root = fs->root;
         fs->root = *path;
-        path_get(path);
+        path_get_long(path);
         write_seqcount_end(&fs->seq);
         spin_unlock(&fs->lock);
         if (old_root.dentry)
-                path_put(&old_root);
+                path_put_long(&old_root);
 }
 
 /*
@@ -36,12 +36,12 @@ void set_fs_pwd(struct fs_struct *fs, struct path *path)
         write_seqcount_begin(&fs->seq);
         old_pwd = fs->pwd;
         fs->pwd = *path;
-        path_get(path);
+        path_get_long(path);
         write_seqcount_end(&fs->seq);
         spin_unlock(&fs->lock);
 
         if (old_pwd.dentry)
-                path_put(&old_pwd);
+                path_put_long(&old_pwd);
 }
 
 void chroot_fs_refs(struct path *old_root, struct path *new_root)
@@ -59,13 +59,13 @@ void chroot_fs_refs(struct path *old_root, struct path *new_root)
                         write_seqcount_begin(&fs->seq);
                         if (fs->root.dentry == old_root->dentry
                             && fs->root.mnt == old_root->mnt) {
-                                path_get(new_root);
+                                path_get_long(new_root);
                                 fs->root = *new_root;
                                 count++;
                         }
                         if (fs->pwd.dentry == old_root->dentry
                             && fs->pwd.mnt == old_root->mnt) {
-                                path_get(new_root);
+                                path_get_long(new_root);
                                 fs->pwd = *new_root;
                                 count++;
                         }
@@ -76,13 +76,13 @@ void chroot_fs_refs(struct path *old_root, struct path *new_root)
         } while_each_thread(g, p);
         read_unlock(&tasklist_lock);
         while (count--)
-                path_put(old_root);
+                path_put_long(old_root);
 }
 
 void free_fs_struct(struct fs_struct *fs)
 {
-        path_put(&fs->root);
-        path_put(&fs->pwd);
+        path_put_long(&fs->root);
+        path_put_long(&fs->pwd);
         kmem_cache_free(fs_cachep, fs);
 }
 
@@ -115,7 +115,13 @@ struct fs_struct *copy_fs_struct(struct fs_struct *old)
                 spin_lock_init(&fs->lock);
                 seqcount_init(&fs->seq);
                 fs->umask = old->umask;
-                get_fs_root_and_pwd(old, &fs->root, &fs->pwd);
+
+                spin_lock(&old->lock);
+                fs->root = old->root;
+                path_get_long(&fs->root);
+                fs->pwd = old->pwd;
+                path_get_long(&fs->pwd);
+                spin_unlock(&old->lock);
         }
         return fs;
 }
diff --git a/fs/internal.h b/fs/internal.h
index e43b9a4dbf4e..9687c2ee2735 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -63,6 +63,7 @@ extern int copy_mount_string(const void __user *, char **);
 
 extern void free_vfsmnt(struct vfsmount *);
 extern struct vfsmount *alloc_vfsmnt(const char *);
+extern unsigned int mnt_get_count(struct vfsmount *mnt);
 extern struct vfsmount *__lookup_mnt(struct vfsmount *, struct dentry *, int);
 extern void mnt_set_mountpoint(struct vfsmount *, struct dentry *,
                                 struct vfsmount *);
diff --git a/fs/namei.c b/fs/namei.c
index 4e957bf744ae..19433cdba011 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -368,6 +368,18 @@ void path_get(struct path *path)
 EXPORT_SYMBOL(path_get);
 
 /**
+ * path_get_long - get a long reference to a path
+ * @path: path to get the reference to
+ *
+ * Given a path increment the reference count to the dentry and the vfsmount.
+ */
+void path_get_long(struct path *path)
+{
+        mntget_long(path->mnt);
+        dget(path->dentry);
+}
+
+/**
  * path_put - put a reference to a path
  * @path: path to put the reference to
  *
@@ -381,6 +393,18 @@ void path_put(struct path *path)
 EXPORT_SYMBOL(path_put);
 
 /**
+ * path_put_long - put a long reference to a path
+ * @path: path to put the reference to