afs: Overhaul the callback handling

Overhaul the AFS callback handling by the following means:

 (1) Don't give up callback promises on vnodes that we are no longer using,
     rather let them just expire on the server or let the server break
     them.  This is actually more efficient for the server as the callback
     lookup is expensive if there are lots of extant callbacks.

 (2) Only give up the callback promises we have from a server when the
     server record is destroyed.  Then we can just give up *all* the
     callback promises on it in one go.

 (3) Servers can end up being shared between cells if cells are aliased, so
     don't add all the vnodes being backed by a particular server into a
     big FID-indexed tree on that server as there may be duplicates.

     Instead have each volume instance (~= superblock) register an interest
     in a server as it starts to make use of it and use this to allow the
     processor for callbacks from the server to find the superblock and
     thence the inode corresponding to the FID being broken by means of
     ilookup_nowait().

 (4) Rather than iterating over the entire callback list when a mass-break
     comes in from the server, maintain a counter of mass-breaks in
     afs_server (cb_seq) and make afs_validate() check it against the copy
     in afs_vnode.

     It would be nice not to have to take a read_lock whilst doing this,
     but that's tricky without using RCU.

 (5) Save a ref on the fileserver we're using for a call in the afs_call
     struct so that we can access its cb_s_break during call decoding.

 (6) Write-lock around callback and status storage in a vnode and read-lock
     around getattr so that we don't see the status mid-update.

This has the following consequences:

 (1) Data invalidation isn't seen until someone calls afs_validate() on a
     vnode.  Unfortunately, we need to use a key to query the server, but
     getting one from a background thread is tricky without caching loads
     of keys all over the place.

 (2) Mass invalidation isn't seen until someone calls afs_validate().

 (3) Callback breaking is going to hit the inode_hash_lock quite a bit.
     Could this be replaced with rcu_read_lock() since inodes are destroyed
     under RCU conditions.

Signed-off-by: David Howells <dhowells@redhat.com>

1 files changed, 15 insertions, 188 deletions

diff --git a/fs/afs/vnode.c b/fs/afs/vnode.c
index b79d05374878..c1bf2124cef5 100644
--- a/fs/afs/vnode.c
+++ b/fs/afs/vnode.c
@@ -16,189 +16,20 @@
 #include <linux/sched.h>
 #include "internal.h"
 
-#if 0
-static noinline bool dump_tree_aux(struct rb_node *node, struct rb_node *parent,
-				   int depth, char lr)
-{
-	struct afs_vnode *vnode;
-	bool bad = false;
-
-	if (!node)
-		return false;
-
-	if (node->rb_left)
-		bad = dump_tree_aux(node->rb_left, node, depth + 2, '/');
-
-	vnode = rb_entry(node, struct afs_vnode, cb_promise);
-	_debug("%c %*.*s%c%p {%d}",
-	       rb_is_red(node) ? 'R' : 'B',
-	       depth, depth, "", lr,
-	       vnode, vnode->cb_expires_at);
-	if (rb_parent(node) != parent) {
-		printk("BAD: %p != %p\n", rb_parent(node), parent);
-		bad = true;
-	}
-
-	if (node->rb_right)
-		bad |= dump_tree_aux(node->rb_right, node, depth + 2, '\\');
-
-	return bad;
-}
-
-static noinline void dump_tree(const char *name, struct afs_server *server)
-{
-	_enter("%s", name);
-	if (dump_tree_aux(server->cb_promises.rb_node, NULL, 0, '-'))
-		BUG();
-}
-#endif
-
 /*
- * insert a vnode into the backing server's vnode tree
- */
-static void afs_install_vnode(struct afs_vnode *vnode,
-			      struct afs_server *server)
-{
-	struct afs_server *old_server = vnode->server;
-	struct afs_vnode *xvnode;
-	struct rb_node *parent, **p;
-
-	_enter("%p,%p", vnode, server);
-
-	if (old_server) {
-		spin_lock(&old_server->fs_lock);
-		rb_erase(&vnode->server_rb, &old_server->fs_vnodes);
-		spin_unlock(&old_server->fs_lock);
-	}
-
-	afs_get_server(server);
-	vnode->server = server;
-	afs_put_server(afs_v2net(vnode), old_server);
-
-	/* insert into the server's vnode tree in FID order */
-	spin_lock(&server->fs_lock);
-
-	parent = NULL;
-	p = &server->fs_vnodes.rb_node;
-	while (*p) {
-		parent = *p;
-		xvnode = rb_entry(parent, struct afs_vnode, server_rb);
-		if (vnode->fid.vid < xvnode->fid.vid)
-			p = &(*p)->rb_left;
-		else if (vnode->fid.vid > xvnode->fid.vid)
-			p = &(*p)->rb_right;
-		else if (vnode->fid.vnode < xvnode->fid.vnode)
-			p = &(*p)->rb_left;
-		else if (vnode->fid.vnode > xvnode->fid.vnode)
-			p = &(*p)->rb_right;
-		else if (vnode->fid.unique < xvnode->fid.unique)
-			p = &(*p)->rb_left;
-		else if (vnode->fid.unique > xvnode->fid.unique)
-			p = &(*p)->rb_right;
-		else
-			BUG(); /* can't happen unless afs_iget() malfunctions */
-	}
-
-	rb_link_node(&vnode->server_rb, parent, p);
-	rb_insert_color(&vnode->server_rb, &server->fs_vnodes);
-
-	spin_unlock(&server->fs_lock);
-	_leave("");
-}
-
-/*
- * insert a vnode into the promising server's update/expiration tree
- * - caller must hold vnode->lock
- */
-static void afs_vnode_note_promise(struct afs_vnode *vnode,
-				   struct afs_server *server)
-{
-	struct afs_server *old_server;
-	struct afs_vnode *xvnode;
-	struct rb_node *parent, **p;
-
-	_enter("%p,%p", vnode, server);
-
-	ASSERT(server != NULL);
-
-	old_server = vnode->server;
-	if (vnode->cb_promised) {
-		if (server == old_server &&
-		    vnode->cb_expires == vnode->cb_expires_at) {
-			_leave(" [no change]");
-			return;
-		}
-
-		spin_lock(&old_server->cb_lock);
-		if (vnode->cb_promised) {
-			_debug("delete");
-			rb_erase(&vnode->cb_promise, &old_server->cb_promises);
-			vnode->cb_promised = false;
-		}
-		spin_unlock(&old_server->cb_lock);
-	}
-
-	if (vnode->server != server)
-		afs_install_vnode(vnode, server);
-
-	vnode->cb_expires_at = vnode->cb_expires;
-	_debug("PROMISE on %p {%lu}",
-	       vnode, (unsigned long) vnode->cb_expires_at);
-
-	/* abuse an RB-tree to hold the expiration order (we may have multiple
-	 * items with the same expiration time) */
-	spin_lock(&server->cb_lock);
-
-	parent = NULL;
-	p = &server->cb_promises.rb_node;
-	while (*p) {
-		parent = *p;
-		xvnode = rb_entry(parent, struct afs_vnode, cb_promise);
-		if (vnode->cb_expires_at < xvnode->cb_expires_at)
-			p = &(*p)->rb_left;
-		else
-			p = &(*p)->rb_right;
-	}
-
-	rb_link_node(&vnode->cb_promise, parent, p);
-	rb_insert_color(&vnode->cb_promise, &server->cb_promises);
-	vnode->cb_promised = true;
-
-	spin_unlock(&server->cb_lock);
-	_leave("");
-}
-
-/*
- * handle remote file deletion by discarding the callback promise
+ * Handle remote file deletion.
  */
 static void afs_vnode_deleted_remotely(struct afs_vnode *vnode)
 {
-	struct afs_server *server;
+	struct afs_cb_interest *cbi = vnode->cb_interest;
 
-	_enter("{%p}", vnode->server);
+	_enter("{%p}", cbi);
 
 	set_bit(AFS_VNODE_DELETED, &vnode->flags);
 
-	server = vnode->server;
-	if (server) {
-		if (vnode->cb_promised) {
-			spin_lock(&server->cb_lock);
-			if (vnode->cb_promised) {
-				rb_erase(&vnode->cb_promise,
-					 &server->cb_promises);
-				vnode->cb_promised = false;
-			}
-			spin_unlock(&server->cb_lock);
-		}
-
-		spin_lock(&server->fs_lock);
-		rb_erase(&vnode->server_rb, &server->fs_vnodes);
-		spin_unlock(&server->fs_lock);
-
-		vnode->server = NULL;
-		afs_put_server(afs_v2net(vnode), server);
-	} else {
-		ASSERT(!vnode->cb_promised);
+	if (cbi) {
+		vnode->cb_interest = NULL;
+		afs_put_cb_interest(afs_v2net(vnode), cbi);
 	}
 
 	_leave("");
@@ -218,8 +49,6 @@ void afs_vnode_finalise_status_update(struct afs_vnode *vnode,
 	_enter("%p,%p", vnode, server);
 
 	spin_lock(&vnode->lock);
-	clear_bit(AFS_VNODE_CB_BROKEN, &vnode->flags);
-	afs_vnode_note_promise(vnode, server);
 	vnode->update_cnt--;
 	ASSERTCMP(vnode->update_cnt, >=, 0);
 	spin_unlock(&vnode->lock);
@@ -238,8 +67,6 @@ static void afs_vnode_status_update_failed(struct afs_vnode *vnode, int ret)
 
 	spin_lock(&vnode->lock);
 
-	clear_bit(AFS_VNODE_CB_BROKEN, &vnode->flags);
-
 	if (ret == -ENOENT) {
 		/* the file was deleted on the server */
 		_debug("got NOENT from server - marking file deleted");
@@ -261,8 +88,8 @@ static void afs_vnode_status_update_failed(struct afs_vnode *vnode, int ret)
  *   - there are any outstanding ops that will fetch the status
  * - TODO implement local caching
  */
-int afs_vnode_fetch_status(struct afs_vnode *vnode,
-			   struct afs_vnode *auth_vnode, struct key *key)
+int afs_vnode_fetch_status(struct afs_vnode *vnode, struct afs_vnode *auth_vnode,
+			   struct key *key, bool force)
 {
 	struct afs_server *server;
 	unsigned long acl_order;
@@ -270,12 +97,13 @@ int afs_vnode_fetch_status(struct afs_vnode *vnode,
 
 	DECLARE_WAITQUEUE(myself, current);
 
-	_enter("%s,{%x:%u.%u}",
+	_enter("%s,{%x:%u.%u,S=%lx},%u",
 	       vnode->volume->vlocation->vldb.name,
-	       vnode->fid.vid, vnode->fid.vnode, vnode->fid.unique);
+	       vnode->fid.vid, vnode->fid.vnode, vnode->fid.unique,
+	       vnode->flags,
+	       force);
 
-	if (!test_bit(AFS_VNODE_CB_BROKEN, &vnode->flags) &&
-	    vnode->cb_promised) {
+	if (!force && test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
 		_leave(" [unchanged]");
 		return 0;
 	}
@@ -291,8 +119,7 @@ int afs_vnode_fetch_status(struct afs_vnode *vnode,
 
 	spin_lock(&vnode->lock);
 
-	if (!test_bit(AFS_VNODE_CB_BROKEN, &vnode->flags) &&
-	    vnode->cb_promised) {
+	if (!force && test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
 		spin_unlock(&vnode->lock);
 		_leave(" [unchanged]");
 		return 0;
@@ -310,7 +137,7 @@ int afs_vnode_fetch_status(struct afs_vnode *vnode,
 
 		/* wait for the status to be updated */
 		for (;;) {
-			if (!test_bit(AFS_VNODE_CB_BROKEN, &vnode->flags))
+			if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags))
 				break;
 			if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
 				break;