Merge pull request #828 from hjelmn/openib_thread_fix

openib thread fixes

opal/mca/btl/openib/btl_openib_component.c

@@ -2534,12 +2534,6 @@ btl_openib_component_init(int *num_btl_modules,
         malloc_hook_set = true;
     }
 #endif
-    /* Currently refuse to run if MPI_THREAD_MULTIPLE is enabled */
-    if (enable_mpi_threads && !mca_btl_base_thread_multiple_override) {
-        opal_output_verbose(5, opal_btl_base_framework.framework_output,
-                            "btl:openib: MPI_THREAD_MULTIPLE not suppported; skipping this component");
-        goto no_btls;
-    }
 
     /* Per https://svn.open-mpi.org/trac/ompi/ticket/1305, check to
        see if $sysfsdir/class/infiniband exists.  If it does not,

opal/mca/btl/openib/btl_openib_eager_rdma.h

@@ -1,5 +1,8 @@
+/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
 /*
  * Copyright (c) 2006-2007 Voltaire All rights reserved.
+ * Copyright (c) 2015      Los Alamos National Security, LLC. All rights
+ *                         reserved.
  * $COPYRIGHT$
  *
  * Additional copyrights may follow
@@ -81,17 +84,33 @@ typedef struct mca_btl_openib_eager_rdma_remote_t mca_btl_openib_eager_rdma_remo
                                    mca_btl_openib_component.eager_rdma_num) \
                                 (I) = 0;                                    \
                         } while (0)
-#define MCA_BTL_OPENIB_RDMA_MOVE_INDEX(HEAD, OLD_HEAD)                      \
+
-    do {                                                                    \
+
-        int32_t new_head;                                                   \
+#if OPAL_ENABLE_DEBUG
-        do {                                                                \
+
-            OLD_HEAD = HEAD;                                                \
+/**
-            new_head = OLD_HEAD + 1;                                        \
+ * @brief read and increment the remote head index and generate a sequence
-            if(new_head == mca_btl_openib_component.eager_rdma_num)         \
+ *        number
-                new_head = 0;                                               \
+ */
-        } while(!OPAL_ATOMIC_CMPSET_32(&HEAD, OLD_HEAD, new_head));         \
+
+#define MCA_BTL_OPENIB_RDMA_MOVE_INDEX(HEAD, OLD_HEAD, SEQ)             \
+    do {                                                                \
+        (SEQ) = OPAL_THREAD_ADD32(&(HEAD), 1) - 1;                      \
+        (OLD_HEAD) = (SEQ) % mca_btl_openib_component.eager_rdma_num;   \
     } while(0)
 
+#else
+
+/**
+ * @brief read and increment the remote head index
+ */
+
+#define MCA_BTL_OPENIB_RDMA_MOVE_INDEX(HEAD, OLD_HEAD)                  \
+    do {                                                                \
+        (OLD_HEAD) = (OPAL_THREAD_ADD32(&(HEAD), 1) - 1) % mca_btl_openib_component.eager_rdma_num; \
+    } while(0)
+
+#endif
 
 END_C_DECLS
 #endif

opal/mca/btl/openib/btl_openib_endpoint.h

@@ -569,17 +569,18 @@ static inline int post_send(mca_btl_openib_endpoint_t *ep,
         MCA_BTL_OPENIB_RDMA_FRAG_SET_SIZE(ftr, sg->length);
         MCA_BTL_OPENIB_RDMA_MAKE_LOCAL(ftr);
 #if OPAL_ENABLE_DEBUG
-        do {
+        /* NTH: generate the sequence from the remote head index to ensure that the
-          ftr->seq = ep->eager_rdma_remote.seq;
+         * wrong sequence isn't set. The way this code used to look the sequence number
-        } while (!OPAL_ATOMIC_CMPSET_32((int32_t*) &ep->eager_rdma_remote.seq,
+         * and head were updated independently and it led to false positives for incorrect
-                                        (int32_t) ftr->seq,
+         * sequence numbers. */
-                                        (int32_t) (ftr->seq+1)));
+        MCA_BTL_OPENIB_RDMA_MOVE_INDEX(ep->eager_rdma_remote.head, head, ftr->seq);
+#else
+        MCA_BTL_OPENIB_RDMA_MOVE_INDEX(ep->eager_rdma_remote.head, head);
 #endif
         if(ep->nbo)
             BTL_OPENIB_FOOTER_HTON(*ftr);
 
         sr_desc->wr.rdma.rkey = ep->eager_rdma_remote.rkey;
-        MCA_BTL_OPENIB_RDMA_MOVE_INDEX(ep->eager_rdma_remote.head, head);
 #if BTL_OPENIB_FAILOVER_ENABLED
         /* frag->ftr is unused on the sending fragment, so use it
          * to indicate it is an eager fragment.  A non-zero value