cluster/ec: Fix cthon failures observed with ec volumes

Since EC already winds one write after other there is no need to align
application fcntl locks with ec blocks. Also added this locking to be
done as a transaction to prevent partial upgrade/downgrade of locks
happening.

BUG: 1410425
Change-Id: I7ce8955c2174f62b11e5cb16140e30ff0f7c4c31
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Reviewed-on: https://review.gluster.org/16445
Smoke: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Xavier Hernandez <xhernandez@datalab.es>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>

1 files changed, 26 insertions, 11 deletions

diff --git a/xlators/cluster/ec/src/ec-locks.c b/xlators/cluster/ec/src/ec-locks.c
index bd525723ddf..70b07c318e0 100644
--- a/xlators/cluster/ec/src/ec-locks.c
+++ b/xlators/cluster/ec/src/ec-locks.c
@@ -1012,10 +1012,6 @@ int32_t ec_manager_lk(ec_fop_data_t * fop, int32_t state)
     switch (state)
     {
         case EC_STATE_INIT:
-            fop->flock.l_len += ec_adjust_offset(fop->xl->private,
-                                                 &fop->flock.l_start, 1);
-            fop->flock.l_len = ec_adjust_size(fop->xl->private,
-                                              fop->flock.l_len, 1);
             if ((fop->int32 == F_SETLKW) && (fop->flock.l_type != F_UNLCK))
             {
                 fop->uint32 = EC_LOCK_MODE_ALL;
@@ -1024,6 +1020,12 @@ int32_t ec_manager_lk(ec_fop_data_t * fop, int32_t state)
 
         /* Fall through */
 
+        case EC_STATE_LOCK:
+            ec_lock_prepare_fd(fop, fop->fd, EC_UPDATE_DATA | EC_QUERY_INFO);
+            ec_lock(fop);
+
+            return EC_STATE_DISPATCH;
+
         case EC_STATE_DISPATCH:
             ec_dispatch_all(fop);
 
@@ -1037,20 +1039,20 @@ int32_t ec_manager_lk(ec_fop_data_t * fop, int32_t state)
                 ec_fop_set_error (fop, ec_lock_check(fop, &mask));
                 if (fop->error != 0) {
                     if (mask != 0) {
-                        ec_t *ec = fop->xl->private;
-                        struct gf_flock flock;
+                        struct gf_flock flock = {0};
 
                         flock.l_type = F_UNLCK;
                         flock.l_whence = fop->flock.l_whence;
-                        flock.l_start = fop->flock.l_start * ec->fragments;
-                        flock.l_len = fop->flock.l_len * ec->fragments;
-                        flock.l_pid = 0;
-                        flock.l_owner.len = 0;
+                        flock.l_start = fop->flock.l_start;
+                        flock.l_len = fop->flock.l_len;
+                        flock.l_pid = fop->flock.l_pid;
+                        lk_owner_copy (&flock.l_owner, &fop->flock.l_owner);
 
                         ec_lk(fop->frame, fop->xl, mask, 1,
                               ec_lock_lk_unlocked, NULL, fop->fd, F_SETLK,
                               &flock, fop->xdata);
                     }
+
                     if (fop->error < 0) {
                         fop->error = 0;
 
@@ -1078,9 +1080,10 @@ int32_t ec_manager_lk(ec_fop_data_t * fop, int32_t state)
                              cbk->op_errno, &cbk->flock, cbk->xdata);
             }
 
-            return EC_STATE_END;
+            return EC_STATE_LOCK_REUSE;
 
         case -EC_STATE_INIT:
+        case -EC_STATE_LOCK:
         case -EC_STATE_DISPATCH:
         case -EC_STATE_REPORT:
             GF_ASSERT(fop->error != 0);
@@ -1091,6 +1094,18 @@ int32_t ec_manager_lk(ec_fop_data_t * fop, int32_t state)
                              NULL, NULL);
             }
 
+            return EC_STATE_LOCK_REUSE;
+
+        case -EC_STATE_LOCK_REUSE:
+        case EC_STATE_LOCK_REUSE:
+            ec_lock_reuse(fop);
+
+            return EC_STATE_UNLOCK;
+
+        case -EC_STATE_UNLOCK:
+        case EC_STATE_UNLOCK:
+            ec_unlock(fop);
+
             return EC_STATE_END;
 
         default: