From: David Teigland <teigland@redhat.com>
Date: Tue, 1 May 2012 20:50:48 +0000 (-0500)
Subject: gfs2: fix recovery during unmount
X-Git-Url: https://git.stricted.de/?a=commitdiff_plain;h=1a058f5288a74a20d5567a85ab1a04a9de69a212;p=GitHub%2Fexynos8895%2Fandroid_kernel_samsung_universal8895.git

gfs2: fix recovery during unmount

Journal recovery from lock_dlm should not be ignored
if there is an unmount in progress.  Ignoring it will
causes the recovery to get stuck.  The recovery
process will correctly handle an in-progess unmount.

Signed-off-by: David Teigland <teigland@redhat.com>
---

diff --git a/fs/gfs2/sys.c b/fs/gfs2/sys.c
index d33172c291ba..9c2592b1d5ff 100644
--- a/fs/gfs2/sys.c
+++ b/fs/gfs2/sys.c
@@ -368,10 +368,7 @@ int gfs2_recover_set(struct gfs2_sbd *sdp, unsigned jid)
 	struct gfs2_jdesc *jd;
 	int rv;
 
-	rv = -ESHUTDOWN;
 	spin_lock(&sdp->sd_jindex_spin);
-	if (test_bit(SDF_NORECOVERY, &sdp->sd_flags))
-		goto out;
 	rv = -EBUSY;
 	if (sdp->sd_jdesc->jd_jid == jid)
 		goto out;
@@ -396,8 +393,13 @@ static ssize_t recover_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
 	if (rv != 1)
 		return -EINVAL;
 
-	rv = gfs2_recover_set(sdp, jid);
+	if (test_bit(SDF_NORECOVERY, &sdp->sd_flags)) {
+		rv = -ESHUTDOWN;
+		goto out;
+	}
 
+	rv = gfs2_recover_set(sdp, jid);
+out:
 	return rv ? rv : len;
 }