ceph: set dn offset when spliced

We want to assign an offset when the dentry goes from null to linked, which
is always done by splice_dentry().  Notably, we should NOT assign an
offset when a dentry is first created and is still null.

BUG if we try to splice a non-null dentry (we shouldn't).

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

2 files changed, 44 insertions, 40 deletions

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 0b0a39d05ca6..d3bb8132a1aa 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -128,7 +128,8 @@ more:
 	dentry = list_entry(p, struct dentry, d_u.d_child);
 	di = ceph_dentry(dentry);
 	while (1) {
-		dout(" p %p/%p d_subdirs %p/%p\n", p->prev, p->next,
+		dout(" p %p/%p %s d_subdirs %p/%p\n", p->prev, p->next,
+		     d_unhashed(dentry) ? "!hashed" : "hashed",
 		     parent->d_subdirs.prev, parent->d_subdirs.next);
 		if (p == &parent->d_subdirs) {
 			fi->at_end = 1;
@@ -571,7 +572,6 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
 		    !is_root_ceph_dentry(dir, dentry) &&
 		    (ci->i_ceph_flags & CEPH_I_COMPLETE) &&
 		    (__ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1))) {
-			di->offset = ci->i_max_offset++;
 			spin_unlock(&dir->i_lock);
 			dout(" dir %p complete, -ENOENT\n", dir);
 			d_add(dentry, NULL);
@@ -984,8 +984,9 @@ static int ceph_d_revalidate(struct dentry *dentry, struct nameidata *nd)
 {
 	struct inode *dir = dentry->d_parent->d_inode;
 
-	dout("d_revalidate %p '%.*s' inode %p\n", dentry,
-	     dentry->d_name.len, dentry->d_name.name, dentry->d_inode);
+	dout("d_revalidate %p '%.*s' inode %p offset %lld\n", dentry,
+	     dentry->d_name.len, dentry->d_name.name, dentry->d_inode,
+	     ceph_dentry(dentry)->offset);
 
 	/* always trust cached snapped dentries, snapdir dentry */
 	if (ceph_snap(dir) != CEPH_NOSNAP) {
@@ -1177,8 +1178,8 @@ void ceph_dentry_lru_touch(struct dentry *dn)
 	struct ceph_dentry_info *di = ceph_dentry(dn);
 	struct ceph_mds_client *mdsc;
 
-	dout("dentry_lru_touch %p %p '%.*s'\n", di, dn,
-	     dn->d_name.len, dn->d_name.name);
+	dout("dentry_lru_touch %p %p '%.*s' (offset %lld)\n", di, dn,
+	     dn->d_name.len, dn->d_name.name, di->offset);
 	if (di) {
 		mdsc = &ceph_sb_to_client(dn->d_sb)->mdsc;
 		spin_lock(&mdsc->dentry_lru_lock);
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index aa22a0bce52f..1bcf98bd0255 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -804,6 +804,37 @@ out_unlock:
 }
 
 /*
+ * Set dentry's directory position based on the current dir's max, and
+ * order it in d_subdirs, so that dcache_readdir behaves.
+ */
+static void ceph_set_dentry_offset(struct dentry *dn)
+{
+	struct dentry *dir = dn->d_parent;
+	struct inode *inode = dn->d_parent->d_inode;
+	struct ceph_dentry_info *di;
+
+	BUG_ON(!inode);
+
+	di = ceph_dentry(dn);
+
+	spin_lock(&inode->i_lock);
+	if ((ceph_inode(inode)->i_ceph_flags & CEPH_I_COMPLETE) == 0) {
+		spin_unlock(&inode->i_lock);
+		return;
+	}
+	di->offset = ceph_inode(inode)->i_max_offset++;
+	spin_unlock(&inode->i_lock);
+
+	spin_lock(&dcache_lock);
+	spin_lock(&dn->d_lock);
+	list_move_tail(&dir->d_subdirs, &dn->d_u.d_child);
+	dout("set_dentry_offset %p %lld (%p %p)\n", dn, di->offset,
+	     dn->d_u.d_child.prev, dn->d_u.d_child.next);
+	spin_unlock(&dn->d_lock);
+	spin_unlock(&dcache_lock);
+}
+
+/*
  * splice a dentry to an inode.
  * caller must hold directory i_mutex for this to be safe.
  *
@@ -816,6 +847,8 @@ static struct dentry *splice_dentry(struct dentry *dn, struct inode *in,
 {
 	struct dentry *realdn;
 
+	BUG_ON(dn->d_inode);
+
 	/* dn must be unhashed */
 	if (!d_unhashed(dn))
 		d_drop(dn);
@@ -837,48 +870,17 @@ static struct dentry *splice_dentry(struct dentry *dn, struct inode *in,
 		dn = realdn;
 	} else {
 		BUG_ON(!ceph_dentry(dn));
-
 		dout("dn %p attached to %p ino %llx.%llx\n",
 		     dn, dn->d_inode, ceph_vinop(dn->d_inode));
 	}
 	if ((!prehash || *prehash) && d_unhashed(dn))
 		d_rehash(dn);
+	ceph_set_dentry_offset(dn);
 out:
 	return dn;
 }
 
 /*
- * Set dentry's directory position based on the current dir's max, and
- * order it in d_subdirs, so that dcache_readdir behaves.
- */
-static void ceph_set_dentry_offset(struct dentry *dn)
-{
-	struct dentry *dir = dn->d_parent;
-	struct inode *inode = dn->d_parent->d_inode;
-	struct ceph_dentry_info *di;
-
-	BUG_ON(!inode);
-
-	di = ceph_dentry(dn);
-
-	spin_lock(&inode->i_lock);
-	if ((ceph_inode(inode)->i_ceph_flags & CEPH_I_COMPLETE) == 0) {
-		spin_unlock(&inode->i_lock);
-		return;
-	}
-	di->offset = ceph_inode(inode)->i_max_offset++;
-	spin_unlock(&inode->i_lock);
-
-	spin_lock(&dcache_lock);
-	spin_lock(&dn->d_lock);
-	list_move_tail(&dir->d_subdirs, &dn->d_u.d_child);
-	dout("set_dentry_offset %p %lld (%p %p)\n", dn, di->offset,
-	     dn->d_u.d_child.prev, dn->d_u.d_child.next);
-	spin_unlock(&dn->d_lock);
-	spin_unlock(&dcache_lock);
-}
-
-/*
  * Incorporate results into the local cache.  This is either just
  * one inode, or a directory, dentry, and possibly linked-to inode (e.g.,
  * after a lookup).
@@ -1030,6 +1032,9 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
 			ceph_invalidate_dentry_lease(dn);
 
 			/* take overwritten dentry's readdir offset */
+			dout("dn %p gets %p offset %lld (old offset %lld)\n",
+			     req->r_old_dentry, dn, ceph_dentry(dn)->offset,
+			     ceph_dentry(req->r_old_dentry)->offset);
 			ceph_dentry(req->r_old_dentry)->offset =
 				ceph_dentry(dn)->offset;
 
@@ -1074,7 +1079,6 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
 				goto done;
 			}
 			req->r_dentry = dn;  /* may have spliced */
-			ceph_set_dentry_offset(dn);
 			igrab(in);
 		} else if (ceph_ino(in) == vino.ino &&
 			   ceph_snap(in) == vino.snap) {
@@ -1117,7 +1121,6 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
 			err = PTR_ERR(dn);
 			goto done;
 		}
-		ceph_set_dentry_offset(dn);
 		req->r_dentry = dn;  /* may have spliced */
 		igrab(in);
 		rinfo->head->is_dentry = 1;  /* fool notrace handlers */