afs: Use the serverUnique field in the UVLDB record to reduce rpc ops

The U-version VLDB volume record retrieved by the VL.GetEntryByNameU rpc op
carries a change counter (the serverUnique field) for each fileserver
listed in the record as backing that volume.  This is incremented whenever
the registration details for a fileserver change (such as its address
list).  Note that the same value will be seen in all UVLDB records that
refer to that fileserver.

This should be checked before calling the VL server to re-query the address
list for a fileserver.  If it's the same, there's no point doing the query.

Reported-by: Jeffrey Altman <jaltman@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>

1 files changed, 14 insertions, 12 deletions

diff --git a/fs/afs/server.c b/fs/afs/server.c
index 11b90ac7ea30..9e50ccde5d37 100644
--- a/fs/afs/server.c
+++ b/fs/afs/server.c
@@ -12,7 +12,6 @@
 #include "protocol_yfs.h"
 
 static unsigned afs_server_gc_delay = 10;	/* Server record timeout in seconds */
-static unsigned afs_server_update_delay = 30;	/* Time till VLDB recheck in secs */
 static atomic_t afs_server_debug_id;
 
 static void afs_inc_servers_outstanding(struct afs_net *net)
@@ -218,7 +217,6 @@ static struct afs_server *afs_alloc_server(struct afs_net *net,
 	RCU_INIT_POINTER(server->addresses, alist);
 	server->addr_version = alist->version;
 	server->uuid = *uuid;
-	server->update_at = ktime_get_real_seconds() + afs_server_update_delay;
 	rwlock_init(&server->fs_lock);
 	INIT_HLIST_HEAD(&server->cb_volumes);
 	rwlock_init(&server->cb_break_lock);
@@ -264,7 +262,7 @@ static struct afs_addr_list *afs_vl_lookup_addrs(struct afs_cell *cell,
  * Get or create a fileserver record.
  */
 struct afs_server *afs_lookup_server(struct afs_cell *cell, struct key *key,
-				     const uuid_t *uuid)
+				     const uuid_t *uuid, u32 addr_version)
 {
 	struct afs_addr_list *alist;
 	struct afs_server *server, *candidate;
@@ -272,8 +270,11 @@ struct afs_server *afs_lookup_server(struct afs_cell *cell, struct key *key,
 	_enter("%p,%pU", cell->net, uuid);
 
 	server = afs_find_server_by_uuid(cell->net, uuid);
-	if (server)
+	if (server) {
+		if (server->addr_version != addr_version)
+			set_bit(AFS_SERVER_FL_NEEDS_UPDATE, &server->flags);
 		return server;
+	}
 
 	alist = afs_vl_lookup_addrs(cell, key, uuid);
 	if (IS_ERR(alist))
@@ -558,7 +559,6 @@ static noinline bool afs_update_server_record(struct afs_fs_cursor *fc, struct a
 		write_unlock(&server->fs_lock);
 	}
 
-	server->update_at = ktime_get_real_seconds() + afs_server_update_delay;
 	afs_put_addrlist(discard);
 	_leave(" = t");
 	return true;
@@ -569,8 +569,6 @@ static noinline bool afs_update_server_record(struct afs_fs_cursor *fc, struct a
  */
 bool afs_check_server_record(struct afs_fs_cursor *fc, struct afs_server *server)
 {
-	time64_t now = ktime_get_real_seconds();
-	long diff;
 	bool success;
 	int ret, retries = 0;
 
@@ -579,13 +577,16 @@ bool afs_check_server_record(struct afs_fs_cursor *fc, struct afs_server *server
 	ASSERT(server);
 
 retry:
-	diff = READ_ONCE(server->update_at) - now;
-	if (diff > 0) {
-		_leave(" = t [not now %ld]", diff);
-		return true;
-	}
+	if (test_bit(AFS_SERVER_FL_UPDATING, &server->flags))
+		goto wait;
+	if (test_bit(AFS_SERVER_FL_NEEDS_UPDATE, &server->flags))
+		goto update;
+	_leave(" = t [good]");
+	return true;
 
+update:
 	if (!test_and_set_bit_lock(AFS_SERVER_FL_UPDATING, &server->flags)) {
+		clear_bit(AFS_SERVER_FL_NEEDS_UPDATE, &server->flags);
 		success = afs_update_server_record(fc, server);
 		clear_bit_unlock(AFS_SERVER_FL_UPDATING, &server->flags);
 		wake_up_bit(&server->flags, AFS_SERVER_FL_UPDATING);
@@ -593,6 +594,7 @@ retry:
 		return success;
 	}
 
+wait:
 	ret = wait_on_bit(&server->flags, AFS_SERVER_FL_UPDATING,
 			  (fc->flags & AFS_FS_CURSOR_INTR) ?
 			  TASK_INTERRUPTIBLE : TASK_UNINTERRUPTIBLE);