[patch 3/3] OCFS2 Configurable timeouts - Protocol changes
commit828ae6afbef03bfe107a4a8cc38798419d6a2765
authorAndrew Beekhof <abeekhof@suse.de>
Mon, 4 Dec 2006 13:04:55 +0000 (4 14:04 +0100)
committerMark Fasheh <mark.fasheh@oracle.com>
Mon, 11 Dec 2006 22:26:44 +0000 (11 14:26 -0800)
treeaa9cd680db9af2070f124cfd66aad88da279a5b9
parentb5dd80304da482d77b2320e1a01a189e656b9770
[patch 3/3] OCFS2 Configurable timeouts - Protocol changes

Modify the OCFS2 handshake to ensure essential timeouts are configured
identically on all nodes.

Only allow changes when there are no connected peers

Improves the logic in o2net_advance_rx() which broke now that
sizeof(struct o2net_handshake) is greater than sizeof(struct o2net_msg)

Included is the field for userspace-heartbeat timeout to avoid the need for
further protocol changes.

Uses a global spinlock to ensure the decisions to update configfs entries
are made on the correct value.  The region covered by the spinlock when
incrementing the counter is much larger as this is the more critical case.

Small cleanup contributed by Adrian Bunk <bunk@stusta.de>

Signed-off-by: Andrew Beekhof <abeekhof@suse.de>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
fs/ocfs2/cluster/nodemanager.c
fs/ocfs2/cluster/tcp.c
fs/ocfs2/cluster/tcp.h
fs/ocfs2/cluster/tcp_internal.h