From: Yang Sheng <yang.sheng@intel.com>
Date: Mon, 25 Aug 2014 03:58:57 +0000 (+0800)
Subject: LU-4416 osd: using set_nlink to calm down WARN_ON message
X-Git-Tag: 2.6.52~6
X-Git-Url: https://git.whamcloud.com/?a=commitdiff_plain;h=refs%2Fchanges%2F71%2F11571%2F2;p=fs%2Flustre-release.git

LU-4416 osd: using set_nlink to calm down WARN_ON message

Invoke inc_nlink with nlink=0 will trigger a WARN_ON
message. Flag I_LINKABLE can avoid this problem in
latest upstream kernel. But there exist a case that
WARN_ON was invoked and I_LINKABLE hasn't defined.
RHEL7 does so. Then we just using set_nlink in the
nlink=0 case.

Signed-off-by: Yang Sheng <yang.sheng@intel.com>
Change-Id: I1e110872b359ebd7fa883506c5592f072f26c1dc
Reviewed-on: http://review.whamcloud.com/11571
Tested-by: Jenkins
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: James Simmons <uja.ornl@gmail.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
---

diff --git a/lustre/osd-ldiskfs/osd_handler.c b/lustre/osd-ldiskfs/osd_handler.c
index 0c5a07d..b656a79 100644
--- a/lustre/osd-ldiskfs/osd_handler.c
+++ b/lustre/osd-ldiskfs/osd_handler.c
@@ -2809,17 +2809,15 @@ static int osd_object_ref_add(const struct lu_env *env,
 	 * This also has to properly handle the case of inodes with nlink == 0
 	 * in case they are being linked into the PENDING directory
 	 */
-#ifdef I_LINKABLE
-	/* This is necessary to increment from i_nlink == 0 */
-	spin_lock(&inode->i_lock);
-	inode->i_state |= I_LINKABLE;
-	spin_unlock(&inode->i_lock);
-#endif
-
 	spin_lock(&obj->oo_guard);
-	ldiskfs_inc_count(oh->ot_handle, inode);
-	if (!S_ISDIR(inode->i_mode))
-		LASSERT(inode->i_nlink <= LDISKFS_LINK_MAX);
+	if (unlikely(inode->i_nlink == 0))
+		/* inc_nlink from 0 may cause WARN_ON */
+		set_nlink(inode, 1);
+	else {
+		ldiskfs_inc_count(oh->ot_handle, inode);
+		if (!S_ISDIR(inode->i_mode))
+			LASSERT(inode->i_nlink <= LDISKFS_LINK_MAX);
+	}
 	spin_unlock(&obj->oo_guard);
 
 	ll_dirty_inode(inode, I_DIRTY_DATASYNC);