qla_target: don't delete changed nacls

The code is currently riddled with "drop the hardware_lock to avoid a
deadlock" bugs that expose races.  One of those races seems to expose a
valid warning in tcm_qla2xxx_clear_nacl_from_fcport_map.  Add some
bandaid to it.

Signed-off-by: Joern Engel <joern@logfs.org>
Cc: <stable@vger.kernel.org> # v3.5+
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>

diff --git a/drivers/scsi/qla2xxx/tcm_qla2xxx.c b/drivers/scsi/qla2xxx/tcm_qla2xxx.c
index e2beab962096cd10d18cf1df63920c79fad216b3..4747d2c660247cc80312c1f65a0417e9fbe658b3 100644 (file)
--- a/drivers/scsi/qla2xxx/tcm_qla2xxx.c
+++ b/drivers/scsi/qla2xxx/tcm_qla2xxx.c
@@ -757,7 +757,16 @@ static void tcm_qla2xxx_clear_nacl_from_fcport_map(struct qla_tgt_sess *sess)
        pr_debug("fc_rport domain: port_id 0x%06x\n", nacl->nport_id);
 
        node = btree_remove32(&lport->lport_fcport_map, nacl->nport_id);
-       WARN_ON(node && (node != se_nacl));
+       if (WARN_ON(node && (node != se_nacl))) {
+               /*
+                * The nacl no longer matches what we think it should be.
+                * Most likely a new dynamic acl has been added while
+                * someone dropped the hardware lock.  It clearly is a
+                * bug elsewhere, but this bit can't make things worse.
+                */
+               btree_insert32(&lport->lport_fcport_map, nacl->nport_id,
+                              node, GFP_ATOMIC);
+       }
 
        pr_debug("Removed from fcport_map: %p for WWNN: 0x%016LX, port_id: 0x%06x\n",
            se_nacl, nacl->nport_wwnn, nacl->nport_id);