afs: Try to avoid taking RCU read lock when checking vnode validity

Try to avoid taking the RCU read lock when checking the validity of a
vnode's callback state.  The only thing it's needed for is to pin the
parent volume's server list whilst we search it to find the record of the
server we're currently using to see if it has been reinitialised (ie. it
sent us a CB.InitCallBackState* RPC).

Do this by the following means:

 (1) Keep an additional per-cell counter (fs_s_break) that's incremented
     each time any of the fileservers in the cell reinitialises.

     Since the new counter can be accessed without RCU from the vnode, we
     can check that first - and only if it differs, get the RCU read lock
     and check the volume's server list.

 (2) Replace afs_get_s_break_rcu() with afs_check_server_good() which now
     indicates whether the callback promise is still expected to be present
     on the server.  This does the checks as described in (1).

 (3) Restructure afs_check_validity() to take account of the change in (2).

     We can also get rid of the valid variable and just use the need_clear
     variable with the addition of the afs_cb_break_no_promise reason.

 (4) afs_check_validity() probably shouldn't be altering vnode->cb_v_break
     and vnode->cb_s_break when it doesn't have cb_lock exclusively locked.

     Move the change to vnode->cb_v_break to __afs_break_callback().

     Delegate the change to vnode->cb_s_break to afs_select_fileserver()
     and set vnode->cb_fs_s_break there also.

 (5) afs_validate() no longer needs to get the RCU read lock around its
     call to afs_check_validity() - and can skip the call entirely if we
     don't have a promise.

Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Markus Suvanto <markus.suvanto@gmail.com>
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/163111669583.283156.1397603105683094563.stgit@warthog.procyon.org.uk/

1 files changed, 43 insertions, 45 deletions

diff --git a/fs/afs/inode.c b/fs/afs/inode.c
index 80b6c8d967d5..126daf9969db 100644
--- a/fs/afs/inode.c
+++ b/fs/afs/inode.c
@@ -587,22 +587,32 @@ static void afs_zap_data(struct afs_vnode *vnode)
 }
 
 /*
- * Get the server reinit counter for a vnode's current server.
+ * Check to see if we have a server currently serving this volume and that it
+ * hasn't been reinitialised or dropped from the list.
  */
-static bool afs_get_s_break_rcu(struct afs_vnode *vnode, unsigned int *_s_break)
+static bool afs_check_server_good(struct afs_vnode *vnode)
 {
-	struct afs_server_list *slist = rcu_dereference(vnode->volume->servers);
+	struct afs_server_list *slist;
 	struct afs_server *server;
+	bool good;
 	int i;
 
+	if (vnode->cb_fs_s_break == atomic_read(&vnode->volume->cell->fs_s_break))
+		return true;
+
+	rcu_read_lock();
+
+	slist = rcu_dereference(vnode->volume->servers);
 	for (i = 0; i < slist->nr_servers; i++) {
 		server = slist->servers[i].server;
 		if (server == vnode->cb_server) {
-			*_s_break = READ_ONCE(server->cb_s_break);
-			return true;
+			good = (vnode->cb_s_break == server->cb_s_break);
+			rcu_read_unlock();
+			return good;
 		}
 	}
 
+	rcu_read_unlock();
 	return false;
 }
 
@@ -611,57 +621,46 @@ static bool afs_get_s_break_rcu(struct afs_vnode *vnode, unsigned int *_s_break)
  */
 bool afs_check_validity(struct afs_vnode *vnode)
 {
-	struct afs_volume *volume = vnode->volume;
 	enum afs_cb_break_reason need_clear = afs_cb_break_no_break;
 	time64_t now = ktime_get_real_seconds();
-	bool valid;
-	unsigned int cb_break, cb_s_break, cb_v_break;
+	unsigned int cb_break;
 	int seq = 0;
 
 	do {
 		read_seqbegin_or_lock(&vnode->cb_lock, &seq);
-		cb_v_break = READ_ONCE(volume->cb_v_break);
 		cb_break = vnode->cb_break;
 
-		if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags) &&
-		    afs_get_s_break_rcu(vnode, &cb_s_break)) {
-			if (vnode->cb_s_break != cb_s_break ||
-			    vnode->cb_v_break != cb_v_break) {
-				vnode->cb_s_break = cb_s_break;
-				vnode->cb_v_break = cb_v_break;
-				need_clear = afs_cb_break_for_vsbreak;
-				valid = false;
-			} else if (test_bit(AFS_VNODE_ZAP_DATA, &vnode->flags)) {
+		if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
+			if (vnode->cb_v_break != vnode->volume->cb_v_break)
+				need_clear = afs_cb_break_for_v_break;
+			else if (!afs_check_server_good(vnode))
+				need_clear = afs_cb_break_for_s_reinit;
+			else if (test_bit(AFS_VNODE_ZAP_DATA, &vnode->flags))
 				need_clear = afs_cb_break_for_zap;
-				valid = false;
-			} else if (vnode->cb_expires_at - 10 <= now) {
+			else if (vnode->cb_expires_at - 10 <= now)
 				need_clear = afs_cb_break_for_lapsed;
-				valid = false;
-			} else {
-				valid = true;
-			}
 		} else if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
-			valid = true;
+			;
 		} else {
-			vnode->cb_v_break = cb_v_break;
-			valid = false;
+			need_clear = afs_cb_break_no_promise;
 		}
 
 	} while (need_seqretry(&vnode->cb_lock, seq));
 
 	done_seqretry(&vnode->cb_lock, seq);
 
-	if (need_clear != afs_cb_break_no_break) {
-		write_seqlock(&vnode->cb_lock);
-		if (cb_break == vnode->cb_break)
-			__afs_break_callback(vnode, need_clear);
-		else
-			trace_afs_cb_miss(&vnode->fid, need_clear);
-		write_sequnlock(&vnode->cb_lock);
-		valid = false;
-	}
+	if (need_clear == afs_cb_break_no_break)
+		return true;
 
-	return valid;
+	write_seqlock(&vnode->cb_lock);
+	if (need_clear == afs_cb_break_no_promise)
+		vnode->cb_v_break = vnode->volume->cb_v_break;
+	else if (cb_break == vnode->cb_break)
+		__afs_break_callback(vnode, need_clear);
+	else
+		trace_afs_cb_miss(&vnode->fid, need_clear);
+	write_sequnlock(&vnode->cb_lock);
+	return false;
 }
 
 /*
@@ -675,21 +674,20 @@ bool afs_check_validity(struct afs_vnode *vnode)
  */
 int afs_validate(struct afs_vnode *vnode, struct key *key)
 {
-	bool valid;
 	int ret;
 
 	_enter("{v={%llx:%llu} fl=%lx},%x",
 	       vnode->fid.vid, vnode->fid.vnode, vnode->flags,
 	       key_serial(key));
 
-	rcu_read_lock();
-	valid = afs_check_validity(vnode);
-	rcu_read_unlock();
-
-	if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
-		clear_nlink(&vnode->vfs_inode);
+	if (unlikely(test_bit(AFS_VNODE_DELETED, &vnode->flags))) {
+		if (vnode->vfs_inode.i_nlink)
+			clear_nlink(&vnode->vfs_inode);
+		goto valid;
+	}
 
-	if (valid)
+	if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags) &&
+	    afs_check_validity(vnode))
 		goto valid;
 
 	down_write(&vnode->validate_lock);