From: John L. Hammond <john.hammond@intel.com>
Date: Mon, 11 Dec 2017 19:03:20 +0000 (-0600)
Subject: LU-10302 ldlm: destroy lock if LVB init fails
X-Git-Tag: 2.10.57~70
X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=commitdiff_plain;h=c91cb6ee81e7751b719228efa58dc32fdea836e5

LU-10302 ldlm: destroy lock if LVB init fails

In ldlm_cli_enqueue_local(), destroy the newly created lock if
ldlm_lvbo_init() fails. Rename sanity test_232() to test_232a(), fix
the fail_loc setting to happen on the correct node, and check that the
OST can be unmounted after the failed write. Add sanity test_232b() to
do the same but for data version since that uses a different LDLM
path.

Signed-off-by: John L. Hammond <john.hammond@intel.com>
Change-Id: I18e9594b9a2461afbc66128f477d3185a6627bc0
Reviewed-on: https://review.whamcloud.com/30477
Reviewed-by: Quentin Bouget <quentin.bouget@cea.fr>
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
---

diff --git a/lustre/ldlm/ldlm_request.c b/lustre/ldlm/ldlm_request.c
index b901a46..c4b9e6f 100644
--- a/lustre/ldlm/ldlm_request.c
+++ b/lustre/ldlm/ldlm_request.c
@@ -466,6 +466,7 @@ int ldlm_cli_enqueue_local(struct ldlm_namespace *ns,
 	err = ldlm_lvbo_init(lock->l_resource);
 	if (err < 0) {
 		LDLM_ERROR(lock, "delayed lvb init failed (rc %d)", err);
+		ldlm_lock_destroy_nolock(lock);
 		GOTO(out, err);
 	}
 
diff --git a/lustre/tests/sanity.sh b/lustre/tests/sanity.sh
index 78db3fd..b1b61eb 100755
--- a/lustre/tests/sanity.sh
+++ b/lustre/tests/sanity.sh
@@ -14202,19 +14202,44 @@ test_231b() {
 }
 run_test 231b "must not assert on fully utilized OST request buffer"
 
-test_232() {
+test_232a() {
 	mkdir -p $DIR/$tdir
+	$LFS setstripe -c1 -i0 $DIR/$tdir/$tfile
+
 	#define OBD_FAIL_LDLM_OST_LVB		 0x31c
-	$LCTL set_param fail_loc=0x31c
+	do_facet ost1 $LCTL set_param fail_loc=0x31c
 
 	# ignore dd failure
 	dd if=/dev/zero of=$DIR/$tdir/$tfile bs=1M count=1 || true
 
-	$LCTL set_param fail_loc=0
+	do_facet ost1 $LCTL set_param fail_loc=0
+	umount_client $MOUNT || error "umount failed"
+	mount_client $MOUNT || error "mount failed"
+	stop ost1 || error "cannot stop ost1"
+	start ost1 $(ostdevname 1) $OST_MOUNT_OPTS || error "cannot start ost1"
+}
+run_test 232a "failed lock should not block umount"
+
+test_232b() {
+	mkdir -p $DIR/$tdir
+	$LFS setstripe -c1 -i0 $DIR/$tdir/$tfile
+	dd if=/dev/zero of=$DIR/$tdir/$tfile bs=1M count=1
+	sync
+	cancel_lru_locks osc
+
+	#define OBD_FAIL_LDLM_OST_LVB		 0x31c
+	do_facet ost1 $LCTL set_param fail_loc=0x31c
+
+	# ignore failure
+	$LFS data_version $DIR/$tdir/$tfile || true
+
+	do_facet ost1 $LCTL set_param fail_loc=0
 	umount_client $MOUNT || error "umount failed"
 	mount_client $MOUNT || error "mount failed"
+	stop ost1 || error "cannot stop ost1"
+	start ost1 $(ostdevname 1) $OST_MOUNT_OPTS || error "cannot start ost1"
 }
-run_test 232 "failed lock should not block umount"
+run_test 232b "failed data version lock should not block umount"
 
 test_233a() {
 	[ $(lustre_version_code $SINGLEMDS) -ge $(version_code 2.3.64) ] ||