afs: Probe multiple fileservers simultaneously

Send probes to all the unprobed fileservers in a fileserver list on all
addresses simultaneously in an attempt to find out the fastest route whilst
not getting stuck for 20s on any server or address that we don't get a
reply from.

This alleviates the problem whereby attempting to access a new server can
take a long time because the rotation algorithm ends up rotating through
all servers and addresses until it finds one that responds.

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

1 files changed, 7 insertions, 102 deletions

diff --git a/fs/afs/server.c b/fs/afs/server.c
index 7c1be8b4dc9a..642afa2e9783 100644
--- a/fs/afs/server.c
+++ b/fs/afs/server.c
@@ -231,6 +231,8 @@ static struct afs_server *afs_alloc_server(struct afs_net *net,
 	rwlock_init(&server->fs_lock);
 	INIT_HLIST_HEAD(&server->cb_volumes);
 	rwlock_init(&server->cb_break_lock);
+	init_waitqueue_head(&server->probe_wq);
+	spin_lock_init(&server->probe_lock);
 
 	afs_inc_servers_outstanding(net);
 	_leave(" = %p", server);
@@ -254,7 +256,7 @@ static struct afs_addr_list *afs_vl_lookup_addrs(struct afs_cell *cell,
 	ret = -ERESTARTSYS;
 	if (afs_begin_vlserver_operation(&vc, cell, key)) {
 		while (afs_select_vlserver(&vc)) {
-			if (test_bit(vc.ac.index, &vc.ac.alist->yfs))
+			if (test_bit(AFS_VLSERVER_FL_IS_YFS, &vc.server->flags))
 				alist = afs_yfsvl_get_endpoints(&vc, uuid);
 			else
 				alist = afs_vl_get_addrs_u(&vc, uuid);
@@ -365,8 +367,7 @@ static void afs_destroy_server(struct afs_net *net, struct afs_server *server)
 	struct afs_addr_list *alist = rcu_access_pointer(server->addresses);
 	struct afs_addr_cursor ac = {
 		.alist	= alist,
-		.start	= alist->index,
-		.index	= 0,
+		.index	= alist->preferred,
 		.error	= 0,
 	};
 	_enter("%p", server);
@@ -374,6 +375,9 @@ static void afs_destroy_server(struct afs_net *net, struct afs_server *server)
 	if (test_bit(AFS_SERVER_FL_MAY_HAVE_CB, &server->flags))
 		afs_fs_give_up_all_callbacks(net, server, &ac, NULL);
 
+	wait_var_event(&server->probe_outstanding,
+		       atomic_read(&server->probe_outstanding) == 0);
+
 	call_rcu(&server->rcu, afs_server_rcu);
 	afs_dec_servers_outstanding(net);
 }
@@ -507,105 +511,6 @@ void afs_purge_servers(struct afs_net *net)
 }
 
 /*
- * Probe a fileserver to find its capabilities.
- *
- * TODO: Try service upgrade.
- */
-static bool afs_do_probe_fileserver(struct afs_fs_cursor *fc)
-{
-	int i;
-
-	_enter("");
-
-	fc->ac.start = READ_ONCE(fc->ac.alist->index);
-	fc->ac.index = fc->ac.start;
-	fc->ac.error = 0;
-	fc->ac.begun = false;
-
-	while (afs_iterate_addresses(&fc->ac)) {
-		afs_fs_get_capabilities(afs_v2net(fc->vnode), fc->cbi->server,
-					&fc->ac, fc->key);
-		switch (fc->ac.error) {
-		case 0:
-			if (test_bit(AFS_SERVER_FL_IS_YFS, &fc->cbi->server->flags)) {
-				for (i = 0; i < fc->ac.alist->nr_addrs; i++)
-					fc->ac.alist->addrs[i].srx_service =
-						YFS_FS_SERVICE;
-			}
-			afs_end_cursor(&fc->ac);
-			set_bit(AFS_SERVER_FL_PROBED, &fc->cbi->server->flags);
-			return true;
-		case -ECONNABORTED:
-			fc->ac.error = afs_abort_to_error(fc->ac.abort_code);
-			goto error;
-		case -ENOMEM:
-		case -ENONET:
-			goto error;
-		case -ENETUNREACH:
-		case -EHOSTUNREACH:
-		case -ECONNREFUSED:
-		case -ETIMEDOUT:
-		case -ETIME:
-			break;
-		default:
-			fc->ac.error = afs_io_error(NULL, afs_io_error_fs_probe_fail);
-			goto error;
-		}
-	}
-
-error:
-	afs_end_cursor(&fc->ac);
-	return false;
-}
-
-/*
- * If we haven't already, try probing the fileserver to get its capabilities.
- * We try not to instigate parallel probes, but it's possible that the parallel
- * probes will fail due to authentication failure when ours would succeed.
- *
- * TODO: Try sending an anonymous probe if an authenticated probe fails.
- */
-bool afs_probe_fileserver(struct afs_fs_cursor *fc)
-{
-	bool success;
-	int ret, retries = 0;
-
-	_enter("");
-
-retry:
-	if (test_bit(AFS_SERVER_FL_PROBED, &fc->cbi->server->flags)) {
-		_leave(" = t");
-		return true;
-	}
-
-	if (!test_and_set_bit_lock(AFS_SERVER_FL_PROBING, &fc->cbi->server->flags)) {
-		success = afs_do_probe_fileserver(fc);
-		clear_bit_unlock(AFS_SERVER_FL_PROBING, &fc->cbi->server->flags);
-		wake_up_bit(&fc->cbi->server->flags, AFS_SERVER_FL_PROBING);
-		_leave(" = t");
-		return success;
-	}
-
-	_debug("wait");
-	ret = wait_on_bit(&fc->cbi->server->flags, AFS_SERVER_FL_PROBING,
-			  TASK_INTERRUPTIBLE);
-	if (ret == -ERESTARTSYS) {
-		fc->ac.error = ret;
-		_leave(" = f [%d]", ret);
-		return false;
-	}
-
-	retries++;
-	if (retries == 4) {
-		fc->ac.error = -ESTALE;
-		_leave(" = f [stale]");
-		return false;
-	}
-	_debug("retry");
-	goto retry;
-}
-
-/*
  * Get an update for a server's address list.
  */
 static noinline bool afs_update_server_record(struct afs_fs_cursor *fc, struct afs_server *server)