aboutsummaryrefslogtreecommitdiffstats
path: root/fs/ocfs2/dlm
diff options
context:
space:
mode:
authorSunil Mushran <sunil.mushran@oracle.com>2008-12-16 18:49:19 -0500
committerMark Fasheh <mfasheh@suse.com>2009-01-05 11:40:35 -0500
commit2b83256407687613e906bee93d98a25339128a4d (patch)
tree5886c51adcb20b3d735ccf38e1f7e70671aefb43 /fs/ocfs2/dlm
parent58896c4d0e5868360ea0693c607d5bf74f79da6b (diff)
ocfs2/dlm: Fix a race between migrate request and exit domain
Patch address a racing migrate request message and an exit domain message. Instead of blocking exit domains for the duration of the migrate, we ignore failure to deliver that message. This is because an exiting domain should not have any active locks and thus has no role to play in the migration. Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Diffstat (limited to 'fs/ocfs2/dlm')
-rw-r--r--fs/ocfs2/dlm/dlmmaster.c23
1 files changed, 19 insertions, 4 deletions
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 44f87caf3683..92fd1d7d6126 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2949,7 +2949,7 @@ static int dlm_do_migrate_request(struct dlm_ctxt *dlm,
2949 struct dlm_node_iter *iter) 2949 struct dlm_node_iter *iter)
2950{ 2950{
2951 struct dlm_migrate_request migrate; 2951 struct dlm_migrate_request migrate;
2952 int ret, status = 0; 2952 int ret, skip, status = 0;
2953 int nodenum; 2953 int nodenum;
2954 2954
2955 memset(&migrate, 0, sizeof(migrate)); 2955 memset(&migrate, 0, sizeof(migrate));
@@ -2966,12 +2966,27 @@ static int dlm_do_migrate_request(struct dlm_ctxt *dlm,
2966 nodenum == new_master) 2966 nodenum == new_master)
2967 continue; 2967 continue;
2968 2968
2969 /* We could race exit domain. If exited, skip. */
2970 spin_lock(&dlm->spinlock);
2971 skip = (!test_bit(nodenum, dlm->domain_map));
2972 spin_unlock(&dlm->spinlock);
2973 if (skip) {
2974 clear_bit(nodenum, iter->node_map);
2975 continue;
2976 }
2977
2969 ret = o2net_send_message(DLM_MIGRATE_REQUEST_MSG, dlm->key, 2978 ret = o2net_send_message(DLM_MIGRATE_REQUEST_MSG, dlm->key,
2970 &migrate, sizeof(migrate), nodenum, 2979 &migrate, sizeof(migrate), nodenum,
2971 &status); 2980 &status);
2972 if (ret < 0) 2981 if (ret < 0) {
2973 mlog_errno(ret); 2982 mlog(0, "migrate_request returned %d!\n", ret);
2974 else if (status < 0) { 2983 if (!dlm_is_host_down(ret)) {
2984 mlog(ML_ERROR, "unhandled error=%d!\n", ret);
2985 BUG();
2986 }
2987 clear_bit(nodenum, iter->node_map);
2988 ret = 0;
2989 } else if (status < 0) {
2975 mlog(0, "migrate request (node %u) returned %d!\n", 2990 mlog(0, "migrate request (node %u) returned %d!\n",
2976 nodenum, status); 2991 nodenum, status);
2977 ret = status; 2992 ret = status;