ec/fini: Fix race between xlator cleanup and on going async fop

Problem:
While we process a cleanup, there is a chance for a race between
async operations, for example ec_launch_replace_heal. So this can
lead to invalid mem access.

Solution:
Just like we track on going heal fops, we can also track fops like
ec_launch_replace_heal, so that we can decide when to send a
PARENT_DOWN request.

Change-Id: I055391c5c6c34d58aef7336847f3b570cb831298
fixes: bz#1703948
Signed-off-by: Mohammed Rafi KC <rkavunga@redhat.com>

1 files changed, 15 insertions, 2 deletions

diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 8844c292f8..237fea2235 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -2814,8 +2814,20 @@ int
 ec_replace_heal_done(int ret, call_frame_t *heal, void *opaque)
 {
     ec_t *ec = opaque;
+    gf_boolean_t last_fop = _gf_false;
 
+    if (GF_ATOMIC_DEC(ec->async_fop_count) == 0) {
+        LOCK(&ec->lock);
+        {
+            last_fop = __ec_is_last_fop(ec);
+        }
+        UNLOCK(&ec->lock);
+    }
     gf_msg_debug(ec->xl->name, 0, "getxattr on bricks is done ret %d", ret);
+
+    if (last_fop)
+        ec_pending_fops_completed(ec);
+
     return 0;
 }
 
@@ -2869,14 +2881,15 @@ ec_launch_replace_heal(ec_t *ec)
 {
     int ret = -1;
 
-    if (!ec)
-        return ret;
     ret = synctask_new(ec->xl->ctx->env, ec_replace_brick_heal_wrap,
                        ec_replace_heal_done, NULL, ec);
+
     if (ret < 0) {
         gf_msg_debug(ec->xl->name, 0, "Heal failed for replace brick ret = %d",
                      ret);
+        ec_replace_heal_done(-1, NULL, ec);
     }
+
     return ret;
 }