afs: Implement VL server rotation
Track VL servers as independent entities rather than lumping all their
addresses together into one set and implement server-level rotation by:
(1) Add the concept of a VL server list, where each server has its own
separate address list. This code is similar to the FS server list.
(2) Use the DNS resolver to retrieve a set of servers and their associated
addresses, ports, preference and weight ratings.
(3) In the case of a legacy DNS resolver or an address list given directly
through /proc/net/afs/cells, create a list containing just a dummy
server record and attach all the addresses to that.
(4) Implement a simple rotation policy, for the moment ignoring the
priorities and weights assigned to the servers.
(5) Show the address list through /proc/net/afs/<cell>/vlservers. This
also displays the source and status of the data as indicated by the
upcall.
Signed-off-by: David Howells <dhowells@redhat.com>
diff --git a/fs/afs/vl_rotate.c b/fs/afs/vl_rotate.c
new file mode 100644
index 0000000..44a936a
--- /dev/null
+++ b/fs/afs/vl_rotate.c
@@ -0,0 +1,251 @@
+/* Handle vlserver selection and rotation.
+ *
+ * Copyright (C) 2018 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public Licence
+ * as published by the Free Software Foundation; either version
+ * 2 of the Licence, or (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/sched/signal.h>
+#include "internal.h"
+#include "afs_vl.h"
+
+/*
+ * Begin an operation on a volume location server.
+ */
+bool afs_begin_vlserver_operation(struct afs_vl_cursor *vc, struct afs_cell *cell,
+ struct key *key)
+{
+ memset(vc, 0, sizeof(*vc));
+ vc->cell = cell;
+ vc->key = key;
+ vc->error = -EDESTADDRREQ;
+ vc->ac.error = SHRT_MAX;
+
+ if (signal_pending(current)) {
+ vc->error = -EINTR;
+ vc->flags |= AFS_VL_CURSOR_STOP;
+ return false;
+ }
+
+ return true;
+}
+
+/*
+ * Begin iteration through a server list, starting with the last used server if
+ * possible, or the last recorded good server if not.
+ */
+static bool afs_start_vl_iteration(struct afs_vl_cursor *vc)
+{
+ struct afs_cell *cell = vc->cell;
+
+ if (wait_on_bit(&cell->flags, AFS_CELL_FL_NO_LOOKUP_YET,
+ TASK_INTERRUPTIBLE)) {
+ vc->error = -ERESTARTSYS;
+ return false;
+ }
+
+ read_lock(&cell->vl_servers_lock);
+ vc->server_list = afs_get_vlserverlist(
+ rcu_dereference_protected(cell->vl_servers,
+ lockdep_is_held(&cell->vl_servers_lock)));
+ read_unlock(&cell->vl_servers_lock);
+ if (!vc->server_list || !vc->server_list->nr_servers)
+ return false;
+
+ vc->start = READ_ONCE(vc->server_list->index);
+ vc->index = vc->start;
+ return true;
+}
+
+/*
+ * Select the vlserver to use. May be called multiple times to rotate
+ * through the vlservers.
+ */
+bool afs_select_vlserver(struct afs_vl_cursor *vc)
+{
+ struct afs_addr_list *alist;
+ struct afs_vlserver *vlserver;
+ int error = vc->ac.error;
+
+ _enter("%u/%u,%u/%u,%d,%d",
+ vc->index, vc->start,
+ vc->ac.index, vc->ac.start,
+ error, vc->ac.abort_code);
+
+ if (vc->flags & AFS_VL_CURSOR_STOP) {
+ _leave(" = f [stopped]");
+ return false;
+ }
+
+ /* Evaluate the result of the previous operation, if there was one. */
+ switch (error) {
+ case SHRT_MAX:
+ goto start;
+
+ default:
+ case 0:
+ /* Success or local failure. Stop. */
+ vc->error = error;
+ vc->flags |= AFS_VL_CURSOR_STOP;
+ _leave(" = f [okay/local %d]", vc->ac.error);
+ return false;
+
+ case -ECONNABORTED:
+ /* The far side rejected the operation on some grounds. This
+ * might involve the server being busy or the volume having been moved.
+ */
+ switch (vc->ac.abort_code) {
+ case AFSVL_IO:
+ case AFSVL_BADVOLOPER:
+ case AFSVL_NOMEM:
+ /* The server went weird. */
+ vc->error = -EREMOTEIO;
+ //write_lock(&vc->cell->vl_servers_lock);
+ //vc->server_list->weird_mask |= 1 << vc->index;
+ //write_unlock(&vc->cell->vl_servers_lock);
+ goto next_server;
+
+ default:
+ vc->error = afs_abort_to_error(vc->ac.abort_code);
+ goto failed;
+ }
+
+ case -ENETUNREACH:
+ case -EHOSTUNREACH:
+ case -ECONNREFUSED:
+ case -ETIMEDOUT:
+ case -ETIME:
+ _debug("no conn %d", error);
+ vc->error = error;
+ goto iterate_address;
+
+ case -ECONNRESET:
+ _debug("call reset");
+ vc->error = error;
+ vc->flags |= AFS_VL_CURSOR_RETRY;
+ goto next_server;
+ }
+
+restart_from_beginning:
+ _debug("restart");
+ afs_end_cursor(&vc->ac);
+ afs_put_vlserverlist(vc->cell->net, vc->server_list);
+ vc->server_list = NULL;
+ if (vc->flags & AFS_VL_CURSOR_RETRIED)
+ goto failed;
+ vc->flags |= AFS_VL_CURSOR_RETRIED;
+start:
+ _debug("start");
+
+ /* TODO: Consider checking the VL server list */
+
+ if (!afs_start_vl_iteration(vc))
+ goto failed;
+
+use_server:
+ _debug("use");
+ /* We're starting on a different vlserver from the list. We need to
+ * check it, find its address list and probe its capabilities before we
+ * use it.
+ */
+ ASSERTCMP(vc->ac.alist, ==, NULL);
+ vlserver = vc->server_list->servers[vc->index].server;
+
+ // TODO: Check the vlserver occasionally
+ //if (!afs_check_vlserver_record(vc, vlserver))
+ // goto failed;
+
+ _debug("USING VLSERVER: %s", vlserver->name);
+
+ read_lock(&vlserver->lock);
+ alist = rcu_dereference_protected(vlserver->addresses,
+ lockdep_is_held(&vlserver->lock));
+ afs_get_addrlist(alist);
+ read_unlock(&vlserver->lock);
+
+ memset(&vc->ac, 0, sizeof(vc->ac));
+
+ /* Probe the current vlserver if we haven't done so yet. */
+#if 0 // TODO
+ if (!test_bit(AFS_VLSERVER_FL_PROBED, &vlserver->flags)) {
+ vc->ac.alist = afs_get_addrlist(alist);
+
+ if (!afs_probe_vlserver(vc)) {
+ error = vc->ac.error;
+ switch (error) {
+ case -ENOMEM:
+ case -ERESTARTSYS:
+ case -EINTR:
+ goto failed_set_error;
+ default:
+ goto next_server;
+ }
+ }
+ }
+#endif
+
+ if (!vc->ac.alist)
+ vc->ac.alist = alist;
+ else
+ afs_put_addrlist(alist);
+
+ vc->ac.start = READ_ONCE(alist->index);
+ vc->ac.index = vc->ac.start;
+
+iterate_address:
+ ASSERT(vc->ac.alist);
+ _debug("iterate %d/%d", vc->ac.index, vc->ac.alist->nr_addrs);
+ /* Iterate over the current server's address list to try and find an
+ * address on which it will respond to us.
+ */
+ if (!afs_iterate_addresses(&vc->ac))
+ goto next_server;
+
+ _leave(" = t %pISpc", &vc->ac.addr->transport);
+ return true;
+
+next_server:
+ _debug("next");
+ afs_end_cursor(&vc->ac);
+ vc->index++;
+ if (vc->index >= vc->server_list->nr_servers)
+ vc->index = 0;
+ if (vc->index != vc->start)
+ goto use_server;
+
+ /* That's all the servers poked to no good effect. Try again if some
+ * of them were busy.
+ */
+ if (vc->flags & AFS_VL_CURSOR_RETRY)
+ goto restart_from_beginning;
+
+ goto failed;
+
+failed:
+ vc->flags |= AFS_VL_CURSOR_STOP;
+ afs_end_cursor(&vc->ac);
+ _leave(" = f [failed %d]", vc->error);
+ return false;
+}
+
+/*
+ * Tidy up a volume location server cursor and unlock the vnode.
+ */
+int afs_end_vlserver_operation(struct afs_vl_cursor *vc)
+{
+ struct afs_net *net = vc->cell->net;
+
+ afs_end_cursor(&vc->ac);
+ afs_put_vlserverlist(net, vc->server_list);
+
+ if (vc->error == -ECONNABORTED)
+ vc->error = afs_abort_to_error(vc->ac.abort_code);
+
+ return vc->error;
+}