Merge pull request #7871 from devreal/osc-ucx-rget-rput-fetch-alignment

OSC UCX: make sure no-op fetch in rget/rput is properly aligned

Author: devreal

ompi/mca/osc/ucx/osc_ucx_comm.c

@@ -1086,9 +1086,10 @@ int ompi_osc_ucx_rput(const void *origin_addr, int origin_count,
     }
 
     mca_osc_ucx_component.num_incomplete_req_ops++;
+    /* TODO: investigate whether ucp_worker_flush_nb is a better choice here */
     ret = opal_common_ucx_wpmem_fetch_nb(module->mem, UCP_ATOMIC_FETCH_OP_FADD,
                                          0, target, &(module->req_result),
-                                         sizeof(uint64_t), remote_addr,
+                                         sizeof(uint64_t), remote_addr & (~0x7),
                                          req_completion, ucx_req);
     if (ret != OMPI_SUCCESS) {
         OMPI_OSC_UCX_REQUEST_RETURN(ucx_req);
@@ -1138,9 +1139,10 @@ int ompi_osc_ucx_rget(void *origin_addr, int origin_count,
     }
 
     mca_osc_ucx_component.num_incomplete_req_ops++;
+    /* TODO: investigate whether ucp_worker_flush_nb is a better choice here */
     ret = opal_common_ucx_wpmem_fetch_nb(module->mem, UCP_ATOMIC_FETCH_OP_FADD,
                                          0, target, &(module->req_result),
-                                         sizeof(uint64_t), remote_addr,
+                                         sizeof(uint64_t), remote_addr & (~0x7),
                                          req_completion, ucx_req);
     if (ret != OMPI_SUCCESS) {
         OMPI_OSC_UCX_REQUEST_RETURN(ucx_req);