cluster/ec: Fix readdir de-itransform

Problem:
gf_deitransform returns the glbal client-id in the complete graph. So except
for the first disperse subvolume under dht, all the other disperse subvolumes
will return a client-id greater than ec->nodes, so readdir will always error
out in those subvolumes.

Fix:
Get the client subvolume whose client-id matches the client-id returned by
gf_deitransform of offset.

Change-Id: I26aa17504352d48d7ff14b390b62f49d7ab2d699
BUG: 1209113
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: http://review.gluster.org/10165
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Xavier Hernandez <xhernandez@datalab.es>

4 files changed, 93 insertions, 8 deletions

diff --git a/tests/basic/ec/ec-readdir.t b/tests/basic/ec/ec-readdir.t
new file mode 100644
index 00000000000..d57bb2869c7
--- /dev/null
+++ b/tests/basic/ec/ec-readdir.t
@@ -0,0 +1,20 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+# This test checks that readdir works fine on distributed disperse volume
+
+cleanup
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 disperse 3 redundancy 1 $H0:$B0/${V0}{0..5}
+TEST $CLI volume start $V0
+
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0;
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "3" ec_child_up_count $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "3" ec_child_up_count $V0 1
+
+TEST touch $M0/{1..100}
+EXPECT "100" echo $(ls $M0 | wc -l)
+cleanup
diff --git a/xlators/cluster/ec/src/ec-dir-read.c b/xlators/cluster/ec/src/ec-dir-read.c
index 17e1a3d124e..ffc3ed5a7cd 100644
--- a/xlators/cluster/ec/src/ec-dir-read.c
+++ b/xlators/cluster/ec/src/ec-dir-read.c
@@ -296,6 +296,30 @@ out:
     }
 }
 
+/* Returns -1 if client_id is invalid else index of child subvol in xl_list */
+int
+ec_deitransform (xlator_t *this, off_t offset)
+{
+        int  idx       = -1;
+        int  client_id = -1;
+        ec_t *ec       = this->private;
+        char id[32]    = {0};
+
+        client_id = gf_deitransform (this, offset);
+        sprintf (id, "%d", client_id);
+        if (dict_get_int32 (ec->leaf_to_subvolid, id, &idx)) {
+                idx = -1;
+                goto out;
+        }
+
+out:
+        if (idx < 0) {
+                gf_log (this->name, GF_LOG_ERROR,
+                        "Invalid index %d in readdirp request", client_id);
+        }
+        return idx;
+}
+
 /* FOP: readdir */
 
 void ec_adjust_readdir(ec_t * ec, int32_t idx, gf_dirent_t * entries)
@@ -412,17 +436,11 @@ int32_t ec_manager_readdir(ec_fop_data_t * fop, int32_t state)
             if (fop->offset != 0)
             {
                 int32_t idx = -1;
-                ec_t    *ec = fop->xl->private;
-
-                idx = gf_deitransform(fop->xl, fop->offset);
 
-                if ((idx < 0) || (idx >= ec->nodes)) {
-
-                        gf_log(fop->xl->name, GF_LOG_ERROR,
-                               "Invalid index %d in readdirp request", idx);
+                idx = ec_deitransform (fop->xl, fop->offset);
 
+                if (idx < 0) {
                         fop->error = EIO;
-
                         return EC_STATE_REPORT;
                 }
                 fop->mask &= 1ULL << idx;
diff --git a/xlators/cluster/ec/src/ec.c b/xlators/cluster/ec/src/ec.c
index fd5123d61da..5476ac6562e 100644
--- a/xlators/cluster/ec/src/ec.c
+++ b/xlators/cluster/ec/src/ec.c
@@ -115,6 +115,35 @@ int32_t ec_prepare_childs(xlator_t * this)
     return 0;
 }
 
+/* This function transforms the subvol to subvol-id*/
+static int
+_subvol_to_subvolid (dict_t *this, char *key, data_t *value, void *data)
+{
+        ec_t *ec = data;
+        xlator_t *subvol = NULL;
+        int      i = 0;
+        int     ret = -1;
+
+        subvol = data_to_ptr (value);
+        for (i = 0; i < ec->nodes; i++) {
+                if (ec->xl_list[i] == subvol) {
+                        ret = dict_set_int32 (this, key, i);
+                        /* -1 stops dict_foreach and returns -1*/
+                        if (ret < 0)
+                                ret = -1;
+                        goto out;
+                }
+        }
+out:
+        return ret;
+}
+
+int
+ec_subvol_to_subvol_id_transform (ec_t *ec, dict_t *leaf_to_subvolid)
+{
+        return dict_foreach (leaf_to_subvolid, _subvol_to_subvolid, ec);
+}
+
 void __ec_destroy_private(xlator_t * this)
 {
     ec_t * ec = this->private;
@@ -165,6 +194,8 @@ void __ec_destroy_private(xlator_t * this)
 
         LOCK_DESTROY(&ec->lock);
 
+        if (ec->leaf_to_subvolid)
+                dict_unref (ec->leaf_to_subvolid);
         GF_FREE(ec);
     }
 }
@@ -480,6 +511,21 @@ init (xlator_t *this)
             ec_selfheal_daemon_init (this);
     gf_log(this->name, GF_LOG_DEBUG, "Disperse translator initialized.");
 
+    ec->leaf_to_subvolid = dict_new ();
+    if (!ec->leaf_to_subvolid)
+            goto failed;
+    if (glusterfs_reachable_leaves (this, ec->leaf_to_subvolid)) {
+            gf_log (this->name, GF_LOG_ERROR, "Failed to build subvol "
+                    "dictionary");
+            goto failed;
+    }
+
+    if (ec_subvol_to_subvol_id_transform (ec, ec->leaf_to_subvolid) < 0) {
+            gf_log (this->name, GF_LOG_ERROR, "Failed to build subvol-id "
+                    "dictionary");
+            goto failed;
+    }
+
     return 0;
 
 failed:
diff --git a/xlators/cluster/ec/src/ec.h b/xlators/cluster/ec/src/ec.h
index d565bf8fcb4..54c13077993 100644
--- a/xlators/cluster/ec/src/ec.h
+++ b/xlators/cluster/ec/src/ec.h
@@ -47,5 +47,6 @@ struct _ec
     struct mem_pool * lock_pool;
     ec_self_heald_t   shd;
     char              vol_uuid[UUID_SIZE + 1];
+    dict_t           *leaf_to_subvolid;
 };
 #endif /* __EC_H__ */