Merge tag 'afs-next-20171113' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs

Pull AFS updates from David Howells:
 "kAFS filesystem driver overhaul.

  The major points of the overhaul are:

   (1) Preliminary groundwork is laid for supporting network-namespacing
       of kAFS. The remainder of the namespacing work requires some way
       to pass namespace information to submounts triggered by an
       automount. This requires something like the mount overhaul that's
       in progress.

   (2) sockaddr_rxrpc is used in preference to in_addr for holding
       addresses internally and add support for talking to the YFS VL
       server. With this, kAFS can do everything over IPv6 as well as
       IPv4 if it's talking to servers that support it.

   (3) Callback handling is overhauled to be generally passive rather
       than active. 'Callbacks' are promises by the server to tell us
       about data and metadata changes. Callbacks are now checked when
       we next touch an inode rather than actively going and looking for
       it where possible.

   (4) File access permit caching is overhauled to store the caching
       information per-inode rather than per-directory, shared over
       subordinate files. Whilst older AFS servers only allow ACLs on
       directories (shared to the files in that directory), newer AFS
       servers break that restriction.

       To improve memory usage and to make it easier to do mass-key
       removal, permit combinations are cached and shared.

   (5) Cell database management is overhauled to allow lighter locks to
       be used and to make cell records autonomous state machines that
       look after getting their own DNS records and cleaning themselves
       up, in particular preventing races in acquiring and relinquishing
       the fscache token for the cell.

   (6) Volume caching is overhauled. The afs_vlocation record is got rid
       of to simplify things and the superblock is now keyed on the cell
       and the numeric volume ID only. The volume record is tied to a
       superblock and normal superblock management is used to mediate
       the lifetime of the volume fscache token.

   (7) File server record caching is overhauled to make server records
       independent of cells and volumes. A server can be in multiple
       cells (in such a case, the administrator must make sure that the
       VL services for all cells correctly reflect the volumes shared
       between those cells).

       Server records are now indexed using the UUID of the server
       rather than the address since a server can have multiple
       addresses.

   (8) File server rotation is overhauled to handle VMOVED, VBUSY (and
       similar), VOFFLINE and VNOVOL indications and to handle rotation
       both of servers and addresses of those servers. The rotation will
       also wait and retry if the server says it is busy.

   (9) Data writeback is overhauled. Each inode no longer stores a list
       of modified sections tagged with the key that authorised it in
       favour of noting the modified region of a page in page->private
       and storing a list of keys that made modifications in the inode.

       This simplifies things and allows other keys to be used to
       actually write to the server if a key that made a modification
       becomes useless.

  (10) Writable mmap() is implemented. This allows a kernel to be build
       entirely on AFS.

  Note that Pre AFS-3.4 servers are no longer supported, though this can
  be added back if necessary (AFS-3.4 was released in 1998)"

* tag 'afs-next-20171113' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs: (35 commits)
  afs: Protect call->state changes against signals
  afs: Trace page dirty/clean
  afs: Implement shared-writeable mmap
  afs: Get rid of the afs_writeback record
  afs: Introduce a file-private data record
  afs: Use a dynamic port if 7001 is in use
  afs: Fix directory read/modify race
  afs: Trace the sending of pages
  afs: Trace the initiation and completion of client calls
  afs: Fix documentation on # vs % prefix in mount source specification
  afs: Fix total-length calculation for multiple-page send
  afs: Only progress call state at end of Tx phase from rxrpc callback
  afs: Make use of the YFS service upgrade to fully support IPv6
  afs: Overhaul volume and server record caching and fileserver rotation
  afs: Move server rotation code into its own file
  afs: Add an address list concept
  afs: Overhaul cell database management
  afs: Overhaul permit caching
  afs: Overhaul the callback handling
  afs: Rename struct afs_call server member to cm_server
  ...

1 files changed, 136 insertions, 368 deletions

diff --git a/fs/afs/callback.c b/fs/afs/callback.c
index 25d404d22cae..f4291b576054 100644
--- a/fs/afs/callback.c
+++ b/fs/afs/callback.c
@@ -20,118 +20,151 @@
 #include <linux/sched.h>
 #include "internal.h"
 
-#if 0
-unsigned afs_vnode_update_timeout = 10;
-#endif  /*  0  */
-
-#define afs_breakring_space(server) \
-	CIRC_SPACE((server)->cb_break_head, (server)->cb_break_tail,	\
-		   ARRAY_SIZE((server)->cb_break))
-
-//static void afs_callback_updater(struct work_struct *);
-
-static struct workqueue_struct *afs_callback_update_worker;
-
 /*
- * allow the fileserver to request callback state (re-)initialisation
+ * Set up an interest-in-callbacks record for a volume on a server and
+ * register it with the server.
+ * - Called with volume->server_sem held.
  */
-void afs_init_callback_state(struct afs_server *server)
+int afs_register_server_cb_interest(struct afs_vnode *vnode,
+				    struct afs_server_entry *entry)
 {
-	struct afs_vnode *vnode;
-
-	_enter("{%p}", server);
+	struct afs_cb_interest *cbi = entry->cb_interest, *vcbi, *new, *x;
+	struct afs_server *server = entry->server;
+
+again:
+	vcbi = vnode->cb_interest;
+	if (vcbi) {
+		if (vcbi == cbi)
+			return 0;
+
+		if (cbi && vcbi->server == cbi->server) {
+			write_seqlock(&vnode->cb_lock);
+			vnode->cb_interest = afs_get_cb_interest(cbi);
+			write_sequnlock(&vnode->cb_lock);
+			afs_put_cb_interest(afs_v2net(vnode), cbi);
+			return 0;
+		}
 
-	spin_lock(&server->cb_lock);
+		if (!cbi && vcbi->server == server) {
+			afs_get_cb_interest(vcbi);
+			x = cmpxchg(&entry->cb_interest, cbi, vcbi);
+			if (x != cbi) {
+				cbi = x;
+				afs_put_cb_interest(afs_v2net(vnode), vcbi);
+				goto again;
+			}
+			return 0;
+		}
+	}
 
-	/* kill all the promises on record from this server */
-	while (!RB_EMPTY_ROOT(&server->cb_promises)) {
-		vnode = rb_entry(server->cb_promises.rb_node,
-				 struct afs_vnode, cb_promise);
-		_debug("UNPROMISE { vid=%x:%u uq=%u}",
-		       vnode->fid.vid, vnode->fid.vnode, vnode->fid.unique);
-		rb_erase(&vnode->cb_promise, &server->cb_promises);
-		vnode->cb_promised = false;
+	if (!cbi) {
+		new = kzalloc(sizeof(struct afs_cb_interest), GFP_KERNEL);
+		if (!new)
+			return -ENOMEM;
+
+		refcount_set(&new->usage, 1);
+		new->sb = vnode->vfs_inode.i_sb;
+		new->vid = vnode->volume->vid;
+		new->server = afs_get_server(server);
+		INIT_LIST_HEAD(&new->cb_link);
+
+		write_lock(&server->cb_break_lock);
+		list_add_tail(&new->cb_link, &server->cb_interests);
+		write_unlock(&server->cb_break_lock);
+
+		x = cmpxchg(&entry->cb_interest, cbi, new);
+		if (x == cbi) {
+			cbi = new;
+		} else {
+			cbi = x;
+			afs_put_cb_interest(afs_v2net(vnode), new);
+		}
 	}
 
-	spin_unlock(&server->cb_lock);
-	_leave("");
+	ASSERT(cbi);
+
+	/* Change the server the vnode is using.  This entails scrubbing any
+	 * interest the vnode had in the previous server it was using.
+	 */
+	write_seqlock(&vnode->cb_lock);
+
+	vnode->cb_interest = afs_get_cb_interest(cbi);
+	vnode->cb_s_break = cbi->server->cb_s_break;
+	clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
+
+	write_sequnlock(&vnode->cb_lock);
+	return 0;
 }
 
 /*
- * handle the data invalidation side of a callback being broken
+ * Set a vnode's interest on a server.
  */
-void afs_broken_callback_work(struct work_struct *work)
+void afs_set_cb_interest(struct afs_vnode *vnode, struct afs_cb_interest *cbi)
 {
-	struct afs_vnode *vnode =
-		container_of(work, struct afs_vnode, cb_broken_work);
+	struct afs_cb_interest *old_cbi = NULL;
 
-	_enter("");
-
-	if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
+	if (vnode->cb_interest == cbi)
 		return;
 
-	/* we're only interested in dealing with a broken callback on *this*
-	 * vnode and only if no-one else has dealt with it yet */
-	if (!mutex_trylock(&vnode->validate_lock))
-		return; /* someone else is dealing with it */
-
-	if (test_bit(AFS_VNODE_CB_BROKEN, &vnode->flags)) {
-		if (S_ISDIR(vnode->vfs_inode.i_mode))
-			afs_clear_permits(vnode);
-
-		if (afs_vnode_fetch_status(vnode, NULL, NULL) < 0)
-			goto out;
-
-		if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
-			goto out;
-
-		/* if the vnode's data version number changed then its contents
-		 * are different */
-		if (test_and_clear_bit(AFS_VNODE_ZAP_DATA, &vnode->flags))
-			afs_zap_data(vnode);
+	write_seqlock(&vnode->cb_lock);
+	if (vnode->cb_interest != cbi) {
+		afs_get_cb_interest(cbi);
+		old_cbi = vnode->cb_interest;
+		vnode->cb_interest = cbi;
 	}
+	write_sequnlock(&vnode->cb_lock);
+	afs_put_cb_interest(afs_v2net(vnode), cbi);
+}
 
-out:
-	mutex_unlock(&vnode->validate_lock);
-
-	/* avoid the potential race whereby the mutex_trylock() in this
-	 * function happens again between the clear_bit() and the
-	 * mutex_unlock() */
-	if (test_bit(AFS_VNODE_CB_BROKEN, &vnode->flags)) {
-		_debug("requeue");
-		queue_work(afs_callback_update_worker, &vnode->cb_broken_work);
+/*
+ * Remove an interest on a server.
+ */
+void afs_put_cb_interest(struct afs_net *net, struct afs_cb_interest *cbi)
+{
+	if (cbi && refcount_dec_and_test(&cbi->usage)) {
+		if (!list_empty(&cbi->cb_link)) {
+			write_lock(&cbi->server->cb_break_lock);
+			list_del_init(&cbi->cb_link);
+			write_unlock(&cbi->server->cb_break_lock);
+			afs_put_server(net, cbi->server);
+		}
+		kfree(cbi);
 	}
-	_leave("");
+}
+
+/*
+ * allow the fileserver to request callback state (re-)initialisation
+ */
+void afs_init_callback_state(struct afs_server *server)
+{
+	if (!test_and_clear_bit(AFS_SERVER_FL_NEW, &server->flags))
+		server->cb_s_break++;
 }
 
 /*
  * actually break a callback
  */
-static void afs_break_callback(struct afs_server *server,
-			       struct afs_vnode *vnode)
+void afs_break_callback(struct afs_vnode *vnode)
 {
 	_enter("");
 
-	set_bit(AFS_VNODE_CB_BROKEN, &vnode->flags);
+	write_seqlock(&vnode->cb_lock);
+
+	if (test_and_clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
+		vnode->cb_break++;
+		afs_clear_permits(vnode);
 
-	if (vnode->cb_promised) {
 		spin_lock(&vnode->lock);
 
 		_debug("break callback");
 
-		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);
-
-		queue_work(afs_callback_update_worker, &vnode->cb_broken_work);
 		if (list_empty(&vnode->granted_locks) &&
 		    !list_empty(&vnode->pending_locks))
 			afs_lock_may_be_available(vnode);
 		spin_unlock(&vnode->lock);
 	}
+
+	write_sequnlock(&vnode->cb_lock);
 }
 
 /*
@@ -143,49 +176,31 @@ static void afs_break_callback(struct afs_server *server,
 static void afs_break_one_callback(struct afs_server *server,
 				   struct afs_fid *fid)
 {
+	struct afs_cb_interest *cbi;
+	struct afs_iget_data data;
 	struct afs_vnode *vnode;
-	struct rb_node *p;
-
-	_debug("find");
-	spin_lock(&server->fs_lock);
-	p = server->fs_vnodes.rb_node;
-	while (p) {
-		vnode = rb_entry(p, struct afs_vnode, server_rb);
-		if (fid->vid < vnode->fid.vid)
-			p = p->rb_left;
-		else if (fid->vid > vnode->fid.vid)
-			p = p->rb_right;
-		else if (fid->vnode < vnode->fid.vnode)
-			p = p->rb_left;
-		else if (fid->vnode > vnode->fid.vnode)
-			p = p->rb_right;
-		else if (fid->unique < vnode->fid.unique)
-			p = p->rb_left;
-		else if (fid->unique > vnode->fid.unique)
-			p = p->rb_right;
-		else
-			goto found;
-	}
-
-	/* not found so we just ignore it (it may have moved to another
-	 * server) */
-not_available:
-	_debug("not avail");
-	spin_unlock(&server->fs_lock);
-	_leave("");
-	return;
+	struct inode *inode;
 
-found:
-	_debug("found");
-	ASSERTCMP(server, ==, vnode->server);
+	read_lock(&server->cb_break_lock);
 
-	if (!igrab(AFS_VNODE_TO_I(vnode)))
-		goto not_available;
-	spin_unlock(&server->fs_lock);
+	/* Step through all interested superblocks.  There may be more than one
+	 * because of cell aliasing.
+	 */
+	list_for_each_entry(cbi, &server->cb_interests, cb_link) {
+		if (cbi->vid != fid->vid)
+			continue;
+
+		data.volume = NULL;
+		data.fid = *fid;
+		inode = ilookup5_nowait(cbi->sb, fid->vnode, afs_iget5_test, &data);
+		if (inode) {
+			vnode = AFS_FS_I(inode);
+			afs_break_callback(vnode);
+			iput(inode);
+		}
+	}
 
-	afs_break_callback(server, vnode);
-	iput(&vnode->vfs_inode);
-	_leave("");
+	read_unlock(&server->cb_break_lock);
 }
 
 /*
@@ -216,261 +231,14 @@ void afs_break_callbacks(struct afs_server *server, size_t count,
 }
 
 /*
- * record the callback for breaking
- * - the caller must hold server->cb_lock
+ * Clear the callback interests in a server list.
  */
-static void afs_do_give_up_callback(struct afs_server *server,
-				    struct afs_vnode *vnode)
+void afs_clear_callback_interests(struct afs_net *net, struct afs_server_list *slist)
 {
-	struct afs_callback *cb;
-
-	_enter("%p,%p", server, vnode);
-
-	cb = &server->cb_break[server->cb_break_head];
-	cb->fid		= vnode->fid;
-	cb->version	= vnode->cb_version;
-	cb->expiry	= vnode->cb_expiry;
-	cb->type	= vnode->cb_type;
-	smp_wmb();
-	server->cb_break_head =
-		(server->cb_break_head + 1) &
-		(ARRAY_SIZE(server->cb_break) - 1);
-
-	/* defer the breaking of callbacks to try and collect as many as
-	 * possible to ship in one operation */
-	switch (atomic_inc_return(&server->cb_break_n)) {
-	case 1 ... AFSCBMAX - 1:
-		queue_delayed_work(afs_callback_update_worker,
-				   &server->cb_break_work, HZ * 2);
-		break;
-	case AFSCBMAX:
-		afs_flush_callback_breaks(server);
-		break;
-	default:
-		break;
-	}
-
-	ASSERT(server->cb_promises.rb_node != NULL);
-	rb_erase(&vnode->cb_promise, &server->cb_promises);
-	vnode->cb_promised = false;
-	_leave("");
-}
-
-/*
- * discard the callback on a deleted item
- */
-void afs_discard_callback_on_delete(struct afs_vnode *vnode)
-{
-	struct afs_server *server = vnode->server;
+	int i;
 
-	_enter("%d", vnode->cb_promised);
-
-	if (!vnode->cb_promised) {
-		_leave(" [not promised]");
-		return;
-	}
-
-	ASSERT(server != NULL);
-
-	spin_lock(&server->cb_lock);
-	if (vnode->cb_promised) {
-		ASSERT(server->cb_promises.rb_node != NULL);
-		rb_erase(&vnode->cb_promise, &server->cb_promises);
-		vnode->cb_promised = false;
+	for (i = 0; i < slist->nr_servers; i++) {
+		afs_put_cb_interest(net, slist->servers[i].cb_interest);
+		slist->servers[i].cb_interest = NULL;
 	}
-	spin_unlock(&server->cb_lock);
-	_leave("");
-}
-
-/*
- * give up the callback registered for a vnode on the file server when the
- * inode is being cleared
- */
-void afs_give_up_callback(struct afs_vnode *vnode)
-{
-	struct afs_server *server = vnode->server;
-
-	DECLARE_WAITQUEUE(myself, current);
-
-	_enter("%d", vnode->cb_promised);
-
-	_debug("GIVE UP INODE %p", &vnode->vfs_inode);
-
-	if (!vnode->cb_promised) {
-		_leave(" [not promised]");
-		return;
-	}
-
-	ASSERT(server != NULL);
-
-	spin_lock(&server->cb_lock);
-	if (vnode->cb_promised && afs_breakring_space(server) == 0) {
-		add_wait_queue(&server->cb_break_waitq, &myself);
-		for (;;) {
-			set_current_state(TASK_UNINTERRUPTIBLE);
-			if (!vnode->cb_promised ||
-			    afs_breakring_space(server) != 0)
-				break;
-			spin_unlock(&server->cb_lock);
-			schedule();
-			spin_lock(&server->cb_lock);
-		}
-		remove_wait_queue(&server->cb_break_waitq, &myself);
-		__set_current_state(TASK_RUNNING);
-	}
-
-	/* of course, it's always possible for the server to break this vnode's
-	 * callback first... */
-	if (vnode->cb_promised)
-		afs_do_give_up_callback(server, vnode);
-
-	spin_unlock(&server->cb_lock);
-	_leave("");
-}
-
-/*
- * dispatch a deferred give up callbacks operation
- */
-void afs_dispatch_give_up_callbacks(struct work_struct *work)
-{
-	struct afs_server *server =
-		container_of(work, struct afs_server, cb_break_work.work);
-
-	_enter("");
-
-	/* tell the fileserver to discard the callback promises it has
-	 * - in the event of ENOMEM or some other error, we just forget that we
-	 *   had callbacks entirely, and the server will call us later to break
-	 *   them
-	 */
-	afs_fs_give_up_callbacks(server, true);
-}
-
-/*
- * flush the outstanding callback breaks on a server
- */
-void afs_flush_callback_breaks(struct afs_server *server)
-{
-	mod_delayed_work(afs_callback_update_worker, &server->cb_break_work, 0);
-}
-
-#if 0
-/*
- * update a bunch of callbacks
- */
-static void afs_callback_updater(struct work_struct *work)
-{
-	struct afs_server *server;
-	struct afs_vnode *vnode, *xvnode;
-	time64_t now;
-	long timeout;
-	int ret;
-
-	server = container_of(work, struct afs_server, updater);
-
-	_enter("");
-
-	now = ktime_get_real_seconds();
-
-	/* find the first vnode to update */
-	spin_lock(&server->cb_lock);
-	for (;;) {
-		if (RB_EMPTY_ROOT(&server->cb_promises)) {
-			spin_unlock(&server->cb_lock);
-			_leave(" [nothing]");
-			return;
-		}
-
-		vnode = rb_entry(rb_first(&server->cb_promises),
-				 struct afs_vnode, cb_promise);
-		if (atomic_read(&vnode->usage) > 0)
-			break;
-		rb_erase(&vnode->cb_promise, &server->cb_promises);
-		vnode->cb_promised = false;
-	}
-
-	timeout = vnode->update_at - now;
-	if (timeout > 0) {
-		queue_delayed_work(afs_vnode_update_worker,
-				   &afs_vnode_update, timeout * HZ);
-		spin_unlock(&server->cb_lock);
-		_leave(" [nothing]");
-		return;
-	}
-
-	list_del_init(&vnode->update);
-	atomic_inc(&vnode->usage);
-	spin_unlock(&server->cb_lock);
-
-	/* we can now perform the update */
-	_debug("update %s", vnode->vldb.name);
-	vnode->state = AFS_VL_UPDATING;
-	vnode->upd_rej_cnt = 0;
-	vnode->upd_busy_cnt = 0;
-
-	ret = afs_vnode_update_record(vl, &vldb);
-	switch (ret) {
-	case 0:
-		afs_vnode_apply_update(vl, &vldb);
-		vnode->state = AFS_VL_UPDATING;
-		break;
-	case -ENOMEDIUM:
-		vnode->state = AFS_VL_VOLUME_DELETED;
-		break;
-	default:
-		vnode->state = AFS_VL_UNCERTAIN;
-		break;
-	}
-
-	/* and then reschedule */
-	_debug("reschedule");
-	vnode->update_at = ktime_get_real_seconds() +
-			afs_vnode_update_timeout;
-
-	spin_lock(&server->cb_lock);
-
-	if (!list_empty(&server->cb_promises)) {
-		/* next update in 10 minutes, but wait at least 1 second more
-		 * than the newest record already queued so that we don't spam
-		 * the VL server suddenly with lots of requests
-		 */
-		xvnode = list_entry(server->cb_promises.prev,
-				    struct afs_vnode, update);
-		if (vnode->update_at <= xvnode->update_at)
-			vnode->update_at = xvnode->update_at + 1;
-		xvnode = list_entry(server->cb_promises.next,
-				    struct afs_vnode, update);
-		timeout = xvnode->update_at - now;
-		if (timeout < 0)
-			timeout = 0;
-	} else {
-		timeout = afs_vnode_update_timeout;
-	}
-
-	list_add_tail(&vnode->update, &server->cb_promises);
-
-	_debug("timeout %ld", timeout);
-	queue_delayed_work(afs_vnode_update_worker,
-			   &afs_vnode_update, timeout * HZ);
-	spin_unlock(&server->cb_lock);
-	afs_put_vnode(vl);
-}
-#endif
-
-/*
- * initialise the callback update process
- */
-int __init afs_callback_update_init(void)
-{
-	afs_callback_update_worker = alloc_ordered_workqueue("kafs_callbackd",
-							     WQ_MEM_RECLAIM);
-	return afs_callback_update_worker ? 0 : -ENOMEM;
-}
-
-/*
- * shut down the callback update process
- */
-void afs_callback_update_kill(void)
-{
-	destroy_workqueue(afs_callback_update_worker);
 }