mpid_rsend.c

mpi并行计算的c++代码 可用vc或gcc编译通过 可以用来搭建并行计算试验环境

/* -*- Mode: C; c-basic-offset:4 ; -*- */
/*
 *  (C) 2001 by Argonne National Laboratory.
 *      See COPYRIGHT in top-level directory.
 */

#include "mpidimpl.h"

/* FIXME: HOMOGENEOUS SYSTEMS ONLY -- no data conversion is performed */

/* FIXME: How does this differ from eager send?  It should differ in 
   only a few bits (e.g., indicate that the send is ready and should
   fail if there is no matching receive) */

/*
 * MPID_Rsend()
 */
#undef FUNCNAME
#define FUNCNAME MPID_Rsend
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPID_Rsend(const void * buf, int count, MPI_Datatype datatype, int rank, int tag, MPID_Comm * comm, int context_offset,
	       MPID_Request ** request)
{
    MPIDI_CH3_Pkt_t upkt;
    MPIDI_CH3_Pkt_ready_send_t * const ready_pkt = &upkt.ready_send;
    MPIDI_msg_sz_t data_sz;
    int dt_contig;
    MPI_Aint dt_true_lb;
    MPID_Datatype * dt_ptr;
    MPID_Request * sreq = NULL;
    MPID_IOV iov[MPID_IOV_LIMIT];
    MPIDI_VC_t * vc;
#if defined(MPID_USE_SEQUENCE_NUMBERS)
    MPID_Seqnum_t seqnum;
#endif    
    int mpi_errno = MPI_SUCCESS;    
    MPIDI_STATE_DECL(MPID_STATE_MPID_RSEND);

    MPIDI_FUNC_ENTER(MPID_STATE_MPID_RSEND);

    MPIDI_DBG_PRINTF((10, FCNAME, "entering"));
    MPIDI_DBG_PRINTF((15, FCNAME, "rank=%d, tag=%d, context=%d", rank, tag, comm->context_id + context_offset));
    
    if (rank == comm->rank && comm->comm_kind != MPID_INTERCOMM)
    {
	mpi_errno = MPIDI_Isend_self(buf, count, datatype, rank, tag, comm, context_offset, MPIDI_REQUEST_TYPE_RSEND, request);
	goto fn_exit;
    }

    if (rank == MPI_PROC_NULL)
    {
	goto fn_exit;
    }

    MPIDI_Datatype_get_info(count, datatype, dt_contig, data_sz, dt_ptr, dt_true_lb);

    MPIDI_Comm_get_vc(comm, rank, &vc);
    
    MPIDI_Pkt_init(ready_pkt, MPIDI_CH3_PKT_READY_SEND);
    ready_pkt->match.rank = comm->rank;
    ready_pkt->match.tag = tag;
    ready_pkt->match.context_id = comm->context_id + context_offset;
    ready_pkt->sender_req_id = MPI_REQUEST_NULL;
    ready_pkt->data_sz = data_sz;

    if (data_sz == 0)
    {
	MPIDI_DBG_PRINTF((15, FCNAME, "sending zero length message"));
	
	MPIDI_VC_FAI_send_seqnum(vc, seqnum);
	MPIDI_Pkt_set_seqnum(ready_pkt, seqnum);
	
	mpi_errno = MPIDI_CH3_iStartMsg(vc, ready_pkt, sizeof(*ready_pkt), &sreq);
	/* --BEGIN ERROR HANDLING-- */
	if (mpi_errno != MPI_SUCCESS)
	{
	    mpi_errno = MPIR_Err_create_code(mpi_errno, MPIR_ERR_FATAL, FCNAME, __LINE__, MPI_ERR_OTHER, "**ch3|eagermsg", 0);
	    goto fn_exit;
	}
	/* --END ERROR HANDLING-- */
	if (sreq != NULL)
	{
	    MPIDI_Request_set_seqnum(sreq, seqnum);
	    MPIDI_Request_set_type(sreq, MPIDI_REQUEST_TYPE_RSEND);
	    /* sreq->comm = comm;
	       MPIR_Comm_add_ref(comm); -- not needed for blocking operations */
	}

	goto fn_exit;
    }
    
    iov[0].MPID_IOV_BUF = (MPID_IOV_BUF_CAST)ready_pkt;
    iov[0].MPID_IOV_LEN = sizeof(*ready_pkt);

    if (dt_contig)
    {
	MPIDI_DBG_PRINTF((15, FCNAME, "sending contiguous ready-mode message, data_sz=" MPIDI_MSG_SZ_FMT, data_sz));
	    
	/* FIXME: handle case where data_sz is greater than what can be stored in iov.MPID_IOV_LEN.  hand off to segment code? */
	iov[1].MPID_IOV_BUF = (MPID_IOV_BUF_CAST)((char *) buf + dt_true_lb);
	iov[1].MPID_IOV_LEN = data_sz;
	    
	MPIDI_VC_FAI_send_seqnum(vc, seqnum);
	MPIDI_Pkt_set_seqnum(ready_pkt, seqnum);
	    
	mpi_errno = MPIDI_CH3_iStartMsgv(vc, iov, 2, &sreq);
	/* --BEGIN ERROR HANDLING-- */
	if (mpi_errno != MPI_SUCCESS)
	{
	    mpi_errno = MPIR_Err_create_code(mpi_errno, MPIR_ERR_FATAL, FCNAME, __LINE__, MPI_ERR_OTHER, "**ch3|eagermsg", 0);
	    goto fn_exit;
	}
	/* --END ERROR HANDLING-- */
	if (sreq != NULL)
	{
	    MPIDI_Request_set_seqnum(sreq, seqnum);
	    MPIDI_Request_set_type(sreq, MPIDI_REQUEST_TYPE_RSEND);
	    /* sreq->comm = comm;
	       MPIR_Comm_add_ref(comm); -- not needed for blocking operations */
	}
    }
    else
    {
	int iov_n;
	    
	MPIDI_DBG_PRINTF((15, FCNAME, "sending non-contiguous ready-mode message, data_sz=" MPIDI_MSG_SZ_FMT, data_sz));
	    
	MPIDI_Request_create_sreq(sreq, mpi_errno, goto fn_exit);
	MPIDI_Request_set_type(sreq, MPIDI_REQUEST_TYPE_RSEND);
	    
	MPID_Segment_init(buf, count, datatype, &sreq->dev.segment, 0);
	sreq->dev.segment_first = 0;
	sreq->dev.segment_size = data_sz;
	    
	iov_n = MPID_IOV_LIMIT - 1;
	mpi_errno = MPIDI_CH3U_Request_load_send_iov(sreq, &iov[1], &iov_n);
	if (mpi_errno == MPI_SUCCESS)
	{
	    iov_n += 1;
		
	    MPIDI_VC_FAI_send_seqnum(vc, seqnum);
	    MPIDI_Pkt_set_seqnum(ready_pkt, seqnum);
	    MPIDI_Request_set_seqnum(sreq, seqnum);
	    
	    if (sreq->dev.ca != MPIDI_CH3_CA_COMPLETE)
	    {
		/* sreq->dev.datatype_ptr = dt_ptr;
		   MPID_Datatype_add_ref(dt_ptr); -- not needed for blocking operations */
	    }
	    
	    mpi_errno = MPIDI_CH3_iSendv(vc, sreq, iov, iov_n);
	    /* --BEGIN ERROR HANDLING-- */
	    if (mpi_errno != MPI_SUCCESS)
	    {
		MPIU_Object_set_ref(sreq, 0);
		MPIDI_CH3_Request_destroy(sreq);
		sreq = NULL;
		mpi_errno = MPIR_Err_create_code(mpi_errno, MPIR_ERR_FATAL, FCNAME, __LINE__, MPI_ERR_OTHER, "**ch3|eagermsg", 0);
		goto fn_exit;
	    }
	    /* --END ERROR HANDLING-- */
	}
	else
	{
	    /* --BEGIN ERROR HANDLING-- */
	    MPIU_Object_set_ref(sreq, 0);
	    MPIDI_CH3_Request_destroy(sreq);
	    sreq = NULL;
	    mpi_errno = MPIR_Err_create_code(mpi_errno, MPIR_ERR_RECOVERABLE, FCNAME, __LINE__, MPI_ERR_OTHER, "**ch3|loadsendiov", 0);
	    goto fn_exit;
	    /* --END ERROR HANDLING-- */
	}
    }

  fn_exit:
    *request = sreq;
    
#   if defined(MPICH_DBG_OUTPUT)
    {
	if (mpi_errno == MPI_SUCCESS)
	{
	    if (sreq)
	    {
		MPIDI_DBG_PRINTF((15, FCNAME, "request allocated, handle=0x%08x", sreq->handle));
	    }
	    else
	    {
		MPIDI_DBG_PRINTF((15, FCNAME, "operation complete, no requests allocated"));
	    }
	}
    }
#   endif
    
    MPIDI_DBG_PRINTF((10, FCNAME, "exiting"));
    MPIDI_FUNC_EXIT(MPID_STATE_MPID_RSEND);
    return mpi_errno;
}