cluster/afr: Make local->child_up immutable

Afr transaction performs lock, pre-op, op, post-op and unlock steps in that
order. The child_up[] is overloaded with the information of where all
the first two steps succeeded. This works perfectly fine for
Transaction, but the locking/unlocking part of the code is re-used by
data self-heal. In that each loop_frame does lock, rchecksum,
read-from-source and write-to-sinks, unlock steps.

Rchecksum fop assumes that the fop needs to happen on one source + all
sinks and sets the call_count to that number. But if the lock step fails
on any of the sinks it will mark the child_up of that child to 0, which
will result in call_count mismatch and the frame will hang thinking that
some more cbks need to come. When this happens loop_frame will never go
to unlock step leading to hangs on that file.

Change-Id: I3dd0449cc6193a980bacf637d935881f4b22210a
BUG: 3597
Reviewed-on: http://review.gluster.com/474
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Amar Tumballi <amar@gluster.com>
Reviewed-by: Vijay Bellur <vijay@gluster.com>

1 files changed, 15 insertions, 1 deletions

diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 0e4e9735..2e5ca71b 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -762,6 +762,7 @@ afr_local_transaction_cleanup (afr_local_t *local, xlator_t *this)
                 GF_FREE (local->internal_lock.lower_locked_nodes);
 
 
+        GF_FREE (local->transaction.pre_op);
         GF_FREE (local->transaction.child_errno);
         GF_FREE (local->child_errno);
         GF_FREE (local->transaction.eager_lock);
@@ -936,6 +937,13 @@ afr_locked_children_count (unsigned char *children, unsigned int child_count)
         return afr_set_elem_count_get (children, child_count);
 }
 
+unsigned int
+afr_pre_op_done_children_count (unsigned char *pre_op,
+                                unsigned int child_count)
+{
+        return afr_set_elem_count_get (pre_op, child_count);
+}
+
 gf_boolean_t
 afr_is_fresh_lookup (loc_t *loc, xlator_t *this)
 {
@@ -3680,11 +3688,17 @@ afr_transaction_local_init (afr_local_t *local, xlator_t *this)
         if (local->fd) {
                 local->fd_open_on = GF_CALLOC (sizeof (*local->fd_open_on),
                                                priv->child_count,
-                                               gf_afr_mt_int32_t);
+                                               gf_afr_mt_char);
                 if (!local->fd_open_on)
                         goto out;
         }
 
+        local->transaction.pre_op = GF_CALLOC (sizeof (*local->transaction.pre_op),
+                                               priv->child_count,
+                                               gf_afr_mt_char);
+        if (!local->transaction.pre_op)
+                goto out;
+
         for (i = 0; i < priv->child_count; i++) {
                 local->pending[i] = GF_CALLOC (sizeof (*local->pending[i]),
                                                3, /* data + metadata + entry */