cluster/stripe: Fix EINVAL errors on quota enabled volumes

        Backport of http://review.gluster.org/8145

Write operations on directories with quota enabled used to fail with
EINVAL on stripe volumes. This was due to assert failure in
stripe_lookup(), meant to ensure loc->path is not NULL. However,
in nameless lookup (in this particular case triggered by quotad, which
has stripe xlator in its graph), loc->path can be legitimately NULL.

The fix involves removing this check in stripe_lookup().

Change-Id: Ibbd4f68763fdd8a85f29da78b3937cef1ee4fd1e
BUG: 1100050
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
Reviewed-on: http://review.gluster.org/8186
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Pranith Kumar Karampuri <pkarampu@redhat.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>

1 files changed, 25 insertions, 0 deletions

diff --git a/tests/bugs/bug-1100050.t b/tests/bugs/bug-1100050.t
new file mode 100644
index 00000000000..537591dcd82
--- /dev/null
+++ b/tests/bugs/bug-1100050.t
@@ -0,0 +1,25 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+
+cleanup;
+
+TEST glusterd;
+TEST pidof glusterd;
+
+TEST gluster volume create $V0 stripe 2 $H0:$B0/{1,2} force;
+TEST gluster volume start $V0;
+TEST glusterfs --volfile-server=$H0 --volfile-id=$V0 $M0;
+
+TEST gluster volume quota $V0 enable;
+
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" quotad_up_status;
+
+TEST mkdir $M0/dir;
+
+TEST gluster volume quota $V0 limit-usage /dir 10MB;
+
+TEST mkdir $M0/dir/subdir;
+
+cleanup;