ceph: skip if no auth cap in flush_snaps

If we have a capsnap but no auth cap (e.g. because it is migrating to
another mds), bail out and do nothing for now.  Do NOT remove the capsnap
from the flush list.

Signed-off-by: Sage Weil <sage@newdream.net>

1 files changed, 12 insertions, 7 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 82ffde618af0..dbf0d6a02a77 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -339,7 +339,7 @@ static struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci, int mds)
 /*
  * Return id of any MDS with a cap, preferably FILE_WR|BUFFER|EXCL, else -1.
  */
-static int __ceph_get_cap_mds(struct ceph_inode_info *ci, u32 *mseq)
+static int __ceph_get_cap_mds(struct ceph_inode_info *ci)
 {
         struct ceph_cap *cap;
         int mds = -1;
@@ -349,8 +349,6 @@ static int __ceph_get_cap_mds(struct ceph_inode_info *ci, u32 *mseq)
         for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
                 cap = rb_entry(p, struct ceph_cap, ci_node);
                 mds = cap->mds;
-                if (mseq)
-                        *mseq = cap->mseq;
                 if (cap->issued & (CEPH_CAP_FILE_WR |
                                    CEPH_CAP_FILE_BUFFER |
                                    CEPH_CAP_FILE_EXCL))
@@ -363,7 +361,7 @@ int ceph_get_cap_mds(struct inode *inode)
 {
         int mds;
         spin_lock(&inode->i_lock);
-        mds = __ceph_get_cap_mds(ceph_inode(inode), NULL);
+        mds = __ceph_get_cap_mds(ceph_inode(inode));
         spin_unlock(&inode->i_lock);
         return mds;
 }
@@ -1231,7 +1229,13 @@ retry:
                 BUG_ON(capsnap->dirty == 0);
 
                 /* pick mds, take s_mutex */
-                mds = __ceph_get_cap_mds(ci, &mseq);
+                if (ci->i_auth_cap == NULL) {
+                        dout("no auth cap (migrating?), doing nothing\n");
+                        goto out;
+                }
+                mds = ci->i_auth_cap->session->s_mds;
+                mseq = ci->i_auth_cap->mseq;
+
                 if (session && session->s_mds != mds) {
                         dout("oops, wrong session %p mutex\n", session);
                         mutex_unlock(&session->s_mutex);
@@ -1250,8 +1254,8 @@ retry:
                         }
                         /*
                          * if session == NULL, we raced against a cap
-                         * deletion.  retry, and we'll get a better
-                         * @mds value next time.
+                         * deletion or migration.  retry, and we'll
+                         * get a better @mds value next time.
                          */
                         spin_lock(&inode->i_lock);
                         goto retry;
@@ -1289,6 +1293,7 @@ retry:
         list_del_init(&ci->i_snap_flush_item);
         spin_unlock(&mdsc->snap_flush_lock);
 
+out:
         if (psession)
                 *psession = session;
         else if (session) {