ch3u_rma_ops.c 19.2 KB
Newer Older
1 2 3 4 5 6 7 8 9
/* -*- Mode: C; c-basic-offset:4 ; -*- */
/*
 *  (C) 2001 by Argonne National Laboratory.
 *      See COPYRIGHT in top-level directory.
 */

#include "mpidi_ch3_impl.h"
#include "mpidrma.h"

10 11 12 13 14 15 16 17 18 19 20 21
static int enableShortACC=1;

MPIU_THREADSAFE_INIT_DECL(initRMAoptions);
#ifdef USE_MPIU_INSTR
MPIU_INSTR_DURATION_DECL(wincreate_allgather);
MPIU_INSTR_DURATION_DECL(winfree_rs);
MPIU_INSTR_DURATION_DECL(winfree_complete);
MPIU_INSTR_DURATION_DECL(rmaqueue_alloc);
extern void MPIDI_CH3_RMA_InitInstr(void);
#endif
extern void MPIDI_CH3_RMA_SetAccImmed( int );

22 23 24 25 26 27 28 29 30
#define MPIDI_PASSIVE_TARGET_DONE_TAG  348297
#define MPIDI_PASSIVE_TARGET_RMA_TAG 563924


#undef FUNCNAME
#define FUNCNAME MPIDI_Win_create
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Win_create(void *base, MPI_Aint size, int disp_unit, MPID_Info *info,
31
		     MPID_Comm *comm_ptr, MPID_Win **win_ptr )
32
{
33
    int mpi_errno=MPI_SUCCESS, i, k, comm_size, rank;
34
    MPI_Aint *tmp_buf;
35
    MPID_Comm *win_comm_ptr;
36
    int errflag = FALSE;
37 38 39 40 41 42 43 44 45
    MPIU_CHKPMEM_DECL(4);
    MPIU_CHKLMEM_DECL(1);
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_WIN_CREATE);
    
    MPIDI_RMA_FUNC_ENTER(MPID_STATE_MPIDI_WIN_CREATE);

    /* FIXME: There should be no unreferenced args */
    MPIU_UNREFERENCED_ARG(info);

46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
    if(initRMAoptions) {
	int rc;
	MPIU_THREADSAFE_INIT_BLOCK_BEGIN(initRMAoptions);
	/* Default is to enable the use of the immediate accumulate feature */
	if (!MPL_env2bool( "MPICH_RMA_ACC_IMMED", &rc ))
	    rc = 1;
	MPIDI_CH3_RMA_SetAccImmed(rc);
#ifdef USE_MPIU_INSTR
    /* Define all instrumentation handle used in the CH3 RMA here*/
	MPIU_INSTR_DURATION_INIT(wincreate_allgather,0,"WIN_CREATE:Allgather");
	MPIU_INSTR_DURATION_INIT(winfree_rs,0,"WIN_FREE:ReduceScatterBlock");
	MPIU_INSTR_DURATION_INIT(winfree_complete,0,"WIN_FREE:Complete");
	MPIU_INSTR_DURATION_INIT(rmaqueue_alloc,0,"Allocate RMA Queue element");
	MPIDI_CH3_RMA_InitInstr();

#endif    
	MPIU_THREADSAFE_INIT_CLEAR(initRMAoptions);
	MPIU_THREADSAFE_INIT_BLOCK_END(initRMAoptions);
    }

66 67 68 69
    comm_size = comm_ptr->local_size;
    rank = comm_ptr->rank;
    
    *win_ptr = (MPID_Win *)MPIU_Handle_obj_alloc( &MPID_Win_mem );
70 71
    MPIU_ERR_CHKANDJUMP1(!(*win_ptr),mpi_errno,MPI_ERR_OTHER,"**nomem",
			 "**nomem %s","MPID_Win_mem");
72

73 74
    MPIU_Object_set_ref(*win_ptr, 1);

75 76 77 78 79 80 81
    (*win_ptr)->fence_cnt = 0;
    (*win_ptr)->base = base;
    (*win_ptr)->size = size;
    (*win_ptr)->disp_unit = disp_unit;
    (*win_ptr)->start_group_ptr = NULL; 
    (*win_ptr)->start_assert = 0; 
    (*win_ptr)->attributes = NULL;
82 83
    (*win_ptr)->rma_ops_list_head = NULL;
    (*win_ptr)->rma_ops_list_tail = NULL;
84 85 86 87 88 89 90
    (*win_ptr)->lock_granted = 0;
    (*win_ptr)->current_lock_type = MPID_LOCK_NONE;
    (*win_ptr)->shared_lock_ref_cnt = 0;
    (*win_ptr)->lock_queue = NULL;
    (*win_ptr)->my_counter = 0;
    (*win_ptr)->my_pt_rma_puts_accs = 0;
    
91 92
    mpi_errno = MPIR_Comm_dup_impl(comm_ptr, &win_comm_ptr);
    if (mpi_errno) MPIU_ERR_POP(mpi_errno);
93 94 95 96
    (*win_ptr)->comm_ptr   = win_comm_ptr;
    (*win_ptr)->myrank = rank;

    MPIU_INSTR_DURATION_START(wincreate_allgather);
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
    /* allocate memory for the base addresses, disp_units, and
       completion counters of all processes */ 
    MPIU_CHKPMEM_MALLOC((*win_ptr)->base_addrs, void **,
			comm_size*sizeof(void *), 
			mpi_errno, "(*win_ptr)->base_addrs");

    MPIU_CHKPMEM_MALLOC((*win_ptr)->disp_units, int *, comm_size*sizeof(int), 
			mpi_errno, "(*win_ptr)->disp_units");

    MPIU_CHKPMEM_MALLOC((*win_ptr)->all_win_handles, MPI_Win *, 
			comm_size*sizeof(MPI_Win), 
			mpi_errno, "(*win_ptr)->all_win_handles");
    
    MPIU_CHKPMEM_MALLOC((*win_ptr)->pt_rma_puts_accs, int *, 
			comm_size*sizeof(int), 
			mpi_errno, "(*win_ptr)->pt_rma_puts_accs");
    for (i=0; i<comm_size; i++)	(*win_ptr)->pt_rma_puts_accs[i] = 0;
    
    /* get the addresses of the windows, window objects, and completion
       counters of all processes.  allocate temp. buffer for communication */
    MPIU_CHKLMEM_MALLOC(tmp_buf, MPI_Aint *, 3*comm_size*sizeof(MPI_Aint),
			mpi_errno, "tmp_buf");
    
    /* FIXME: This needs to be fixed for heterogeneous systems */
121 122
    /* FIXME: If we wanted to validate the transfer as within range at the
       origin, we'd also need the window size. */
123
    tmp_buf[3*rank]   = MPIU_PtrToAint(base);
124 125 126
    tmp_buf[3*rank+1] = (MPI_Aint) disp_unit;
    tmp_buf[3*rank+2] = (MPI_Aint) (*win_ptr)->handle;
    
127 128
    mpi_errno = MPIR_Allgather_impl(MPI_IN_PLACE, 0, MPI_DATATYPE_NULL,
                                    tmp_buf, 3 * sizeof(MPI_Aint), MPI_BYTE,
129
                                    comm_ptr, &errflag);
130
    MPIU_INSTR_DURATION_END(wincreate_allgather);
131
    if (mpi_errno) { MPIU_ERR_POP(mpi_errno); }
132 133
    MPIU_ERR_CHKANDJUMP(errflag, mpi_errno, MPI_ERR_OTHER, "**coll_fail");

134 135

    k = 0;
136 137
    for (i=0; i<comm_size; i++)
    {
138 139 140
	(*win_ptr)->base_addrs[i] = MPIU_AintToPtr(tmp_buf[k++]);
	(*win_ptr)->disp_units[i] = (int) tmp_buf[k++];
	(*win_ptr)->all_win_handles[i] = (MPI_Win) tmp_buf[k++];
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
    }
        
 fn_exit:
    MPIU_CHKLMEM_FREEALL();
    MPIDI_RMA_FUNC_EXIT(MPID_STATE_MPIDI_WIN_CREATE);
    return mpi_errno;
    /* --BEGIN ERROR HANDLING-- */
 fn_fail:
    MPIU_CHKPMEM_REAP();
    goto fn_exit;
    /* --END ERROR HANDLING-- */
}




#undef FUNCNAME
#define FUNCNAME MPIDI_Win_free
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Win_free(MPID_Win **win_ptr)
{
163
    int mpi_errno=MPI_SUCCESS, total_pt_rma_puts_accs;
164
    int in_use;
165
    MPID_Comm *comm_ptr;
166
    int errflag = FALSE;
167 168 169 170
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_WIN_FREE);
        
    MPIDI_RMA_FUNC_ENTER(MPID_STATE_MPIDI_WIN_FREE);
        
171 172
    comm_ptr = (*win_ptr)->comm_ptr;
    MPIU_INSTR_DURATION_START(winfree_rs);
173 174
    mpi_errno = MPIR_Reduce_scatter_block_impl((*win_ptr)->pt_rma_puts_accs, 
                                               &total_pt_rma_puts_accs, 1, 
175
                                               MPI_INT, MPI_SUM, comm_ptr, &errflag);
176
    if (mpi_errno) { MPIU_ERR_POP(mpi_errno); }
177
    MPIU_ERR_CHKANDJUMP(errflag, mpi_errno, MPI_ERR_OTHER, "**coll_fail");
178
    MPIU_INSTR_DURATION_END(winfree_rs);
179 180 181 182 183 184

    if (total_pt_rma_puts_accs != (*win_ptr)->my_pt_rma_puts_accs)
    {
	MPID_Progress_state progress_state;
            
	/* poke the progress engine until the two are equal */
185
	MPIU_INSTR_DURATION_START(winfree_complete);
186 187 188 189 190 191 192 193
	MPID_Progress_start(&progress_state);
	while (total_pt_rma_puts_accs != (*win_ptr)->my_pt_rma_puts_accs)
	{
	    mpi_errno = MPID_Progress_wait(&progress_state);
	    /* --BEGIN ERROR HANDLING-- */
	    if (mpi_errno != MPI_SUCCESS)
	    {
		MPID_Progress_end(&progress_state);
194
		MPIU_ERR_SETANDJUMP(mpi_errno,MPI_ERR_OTHER,"**winnoprogress");
195 196 197 198
	    }
	    /* --END ERROR HANDLING-- */
	}
	MPID_Progress_end(&progress_state);
199
	MPIU_INSTR_DURATION_END(winfree_complete);
200 201
    }

202 203 204
    
    mpi_errno = MPIR_Comm_free_impl(comm_ptr);
    if (mpi_errno) MPIU_ERR_POP(mpi_errno);
205

206 207 208 209
    MPIU_Free((*win_ptr)->base_addrs);
    MPIU_Free((*win_ptr)->disp_units);
    MPIU_Free((*win_ptr)->all_win_handles);
    MPIU_Free((*win_ptr)->pt_rma_puts_accs);
210 211 212 213

    MPIU_Object_release_ref(*win_ptr, &in_use);
    /* MPI windows don't have reference count semantics, so this should always be true */
    MPIU_Assert(!in_use);
214
    MPIU_Handle_obj_free( &MPID_Win_mem, *win_ptr );
215

216 217 218
 fn_exit:
    MPIDI_RMA_FUNC_EXIT(MPID_STATE_MPIDI_WIN_FREE);
    return mpi_errno;
219

220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
 fn_fail:
    goto fn_exit;
}



#undef FUNCNAME
#define FUNCNAME MPIDI_Put
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Put(void *origin_addr, int origin_count, MPI_Datatype
            origin_datatype, int target_rank, MPI_Aint target_disp,
            int target_count, MPI_Datatype target_datatype, MPID_Win *win_ptr)
{
    int mpi_errno = MPI_SUCCESS;
    int dt_contig, rank, predefined;
236
    MPIDI_RMA_ops *new_ptr;
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
    MPID_Datatype *dtp;
    MPI_Aint dt_true_lb;
    MPIDI_msg_sz_t data_sz;
    MPIU_CHKPMEM_DECL(1);
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_PUT);
        
    MPIDI_RMA_FUNC_ENTER(MPID_STATE_MPIDI_PUT);

    MPIDI_Datatype_get_info(origin_count, origin_datatype,
			    dt_contig, data_sz, dtp,dt_true_lb); 
    
    if ((data_sz == 0) || (target_rank == MPI_PROC_NULL))
    {
	goto fn_exit;
    }

253
    rank = win_ptr->myrank;
254 255 256 257 258 259 260 261 262 263 264
    
    /* If the put is a local operation, do it here */
    if (target_rank == rank)
    {
	mpi_errno = MPIR_Localcopy(origin_addr, origin_count, origin_datatype,
				   (char *) win_ptr->base + win_ptr->disp_unit *
				   target_disp, target_count, target_datatype); 
    }
    else
    {
	/* queue it up */
265 266
	/* FIXME: For short operations, should we use a (per-thread) pool? */
	MPIU_INSTR_DURATION_START(rmaqueue_alloc);
267 268
	MPIU_CHKPMEM_MALLOC(new_ptr, MPIDI_RMA_ops *, sizeof(MPIDI_RMA_ops), 
			    mpi_errno, "RMA operation entry");
269 270 271 272 273 274 275 276
	MPIU_INSTR_DURATION_END(rmaqueue_alloc);
	if (win_ptr->rma_ops_list_tail) 
	    win_ptr->rma_ops_list_tail->next = new_ptr;
	else
	    win_ptr->rma_ops_list_head = new_ptr;
	win_ptr->rma_ops_list_tail = new_ptr;

	/* FIXME: For contig and very short operations, use a streamlined op */
277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
	new_ptr->next = NULL;  
	new_ptr->type = MPIDI_RMA_PUT;
	new_ptr->origin_addr = origin_addr;
	new_ptr->origin_count = origin_count;
	new_ptr->origin_datatype = origin_datatype;
	new_ptr->target_rank = target_rank;
	new_ptr->target_disp = target_disp;
	new_ptr->target_count = target_count;
	new_ptr->target_datatype = target_datatype;
	
	/* if source or target datatypes are derived, increment their
	   reference counts */ 
	MPIDI_CH3I_DATATYPE_IS_PREDEFINED(origin_datatype, predefined);
	if (!predefined)
	{
	    MPID_Datatype_get_ptr(origin_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
	MPIDI_CH3I_DATATYPE_IS_PREDEFINED(target_datatype, predefined);
	if (!predefined)
	{
	    MPID_Datatype_get_ptr(target_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
    }

  fn_exit:
    MPIDI_RMA_FUNC_EXIT(MPID_STATE_MPIDI_PUT);    
    return mpi_errno;

    /* --BEGIN ERROR HANDLING-- */
  fn_fail:
    MPIU_CHKPMEM_REAP();
    goto fn_exit;
    /* --END ERROR HANDLING-- */
}



#undef FUNCNAME
#define FUNCNAME MPIDI_Get
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Get(void *origin_addr, int origin_count, MPI_Datatype
            origin_datatype, int target_rank, MPI_Aint target_disp,
            int target_count, MPI_Datatype target_datatype, MPID_Win *win_ptr)
{
    int mpi_errno = MPI_SUCCESS;
    MPIDI_msg_sz_t data_sz;
    int dt_contig, rank, predefined;
    MPI_Aint dt_true_lb;
328
    MPIDI_RMA_ops *new_ptr;
329 330 331 332 333 334 335 336 337 338 339 340 341 342
    MPID_Datatype *dtp;
    MPIU_CHKPMEM_DECL(1);
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_GET);
        
    MPIDI_RMA_FUNC_ENTER(MPID_STATE_MPIDI_GET);

    MPIDI_Datatype_get_info(origin_count, origin_datatype,
			    dt_contig, data_sz, dtp, dt_true_lb); 

    if ((data_sz == 0) || (target_rank == MPI_PROC_NULL))
    {
	goto fn_exit;
    }

343
    rank = win_ptr->myrank;
344 345 346 347 348 349 350 351 352 353 354 355 356
    
    /* If the get is a local operation, do it here */
    if (target_rank == rank)
    {
	mpi_errno = MPIR_Localcopy((char *) win_ptr->base +
				   win_ptr->disp_unit * target_disp,
				   target_count, target_datatype,
				   origin_addr, origin_count,
				   origin_datatype);  
    }
    else
    {
	/* queue it up */
357
	MPIU_INSTR_DURATION_START(rmaqueue_alloc);
358 359
	MPIU_CHKPMEM_MALLOC(new_ptr, MPIDI_RMA_ops *, sizeof(MPIDI_RMA_ops), 
			    mpi_errno, "RMA operation entry");
360 361 362
	MPIU_INSTR_DURATION_END(rmaqueue_alloc);
	if (win_ptr->rma_ops_list_tail) 
	    win_ptr->rma_ops_list_tail->next = new_ptr;
363
	else
364 365
	    win_ptr->rma_ops_list_head = new_ptr;
	win_ptr->rma_ops_list_tail = new_ptr;
366
            
367
	/* FIXME: For contig and very short operations, use a streamlined op */
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
	new_ptr->next = NULL;  
	new_ptr->type = MPIDI_RMA_GET;
	new_ptr->origin_addr = origin_addr;
	new_ptr->origin_count = origin_count;
	new_ptr->origin_datatype = origin_datatype;
	new_ptr->target_rank = target_rank;
	new_ptr->target_disp = target_disp;
	new_ptr->target_count = target_count;
	new_ptr->target_datatype = target_datatype;
	
	/* if source or target datatypes are derived, increment their
	   reference counts */ 
	MPIDI_CH3I_DATATYPE_IS_PREDEFINED(origin_datatype, predefined);
	if (!predefined)
	{
	    MPID_Datatype_get_ptr(origin_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
	MPIDI_CH3I_DATATYPE_IS_PREDEFINED(target_datatype, predefined);
	if (!predefined)
	{
	    MPID_Datatype_get_ptr(target_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
    }

  fn_exit:
    MPIDI_RMA_FUNC_EXIT(MPID_STATE_MPIDI_GET);
    return mpi_errno;

    /* --BEGIN ERROR HANDLING-- */
  fn_fail:
    MPIU_CHKPMEM_REAP();
    goto fn_exit;
    /* --END ERROR HANDLING-- */
}



#undef FUNCNAME
#define FUNCNAME MPIDI_Accumulate
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Accumulate(void *origin_addr, int origin_count, MPI_Datatype
                    origin_datatype, int target_rank, MPI_Aint target_disp,
                    int target_count, MPI_Datatype target_datatype, MPI_Op op,
                    MPID_Win *win_ptr)
{
    int mpi_errno=MPI_SUCCESS;
    MPIDI_msg_sz_t data_sz;
    int dt_contig, rank, origin_predefined, target_predefined;
    MPI_Aint dt_true_lb;
420
    MPIDI_RMA_ops *new_ptr;
421 422 423 424 425 426 427 428 429 430 431 432 433 434
    MPID_Datatype *dtp;
    MPIU_CHKLMEM_DECL(2);
    MPIU_CHKPMEM_DECL(1);
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_ACCUMULATE);
    
    MPIDI_RMA_FUNC_ENTER(MPID_STATE_MPIDI_ACCUMULATE);

    MPIDI_Datatype_get_info(origin_count, origin_datatype,
			    dt_contig, data_sz, dtp, dt_true_lb);  
    
    if ((data_sz == 0) || (target_rank == MPI_PROC_NULL))
    {
	goto fn_exit;
    }
435 436

    rank = win_ptr->myrank;
437 438 439 440
    
    MPIDI_CH3I_DATATYPE_IS_PREDEFINED(origin_datatype, origin_predefined);
    MPIDI_CH3I_DATATYPE_IS_PREDEFINED(target_datatype, target_predefined);

441
    /* Do =! rank first (most likely branch?) */
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459
    if (target_rank == rank)
    {
	MPI_User_function *uop;
	
	if (op == MPI_REPLACE)
	{
	    mpi_errno = MPIR_Localcopy(origin_addr, origin_count, 
				origin_datatype,
				(char *) win_ptr->base + win_ptr->disp_unit *
				target_disp, target_count, target_datatype); 
	    goto fn_exit;
	}
	
	MPIU_ERR_CHKANDJUMP1((HANDLE_GET_KIND(op) != HANDLE_KIND_BUILTIN), 
			     mpi_errno, MPI_ERR_OP, "**opnotpredefined",
			     "**opnotpredefined %d", op );
	
	/* get the function by indexing into the op table */
460
	uop = MPIR_Op_table[((op)&0xf) - 1];
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
	
	if (origin_predefined && target_predefined)
	{    
	    (*uop)(origin_addr, (char *) win_ptr->base + win_ptr->disp_unit *
		   target_disp, &target_count, &target_datatype);
	}
	else
	{
	    /* derived datatype */
	    
	    MPID_Segment *segp;
	    DLOOP_VECTOR *dloop_vec;
	    MPI_Aint first, last;
	    int vec_len, i, type_size, count;
	    MPI_Datatype type;
	    MPI_Aint true_lb, true_extent, extent;
	    void *tmp_buf=NULL, *source_buf, *target_buf;
	    
	    if (origin_datatype != target_datatype)
	    {
		/* first copy the data into a temporary buffer with
		   the same datatype as the target. Then do the
		   accumulate operation. */
		
485
		MPIR_Type_get_true_extent_impl(target_datatype, &true_lb, &true_extent);
486 487 488 489 490 491 492 493 494 495 496 497 498
		MPID_Datatype_get_extent_macro(target_datatype, extent); 
		
		MPIU_CHKLMEM_MALLOC(tmp_buf, void *, 
			target_count * (MPIR_MAX(extent,true_extent)), 
			mpi_errno, "temporary buffer");
		/* adjust for potential negative lower bound in datatype */
		tmp_buf = (void *)((char*)tmp_buf - true_lb);
		
		mpi_errno = MPIR_Localcopy(origin_addr, origin_count,
					   origin_datatype, tmp_buf,
					   target_count, target_datatype);  
		if (mpi_errno) { MPIU_ERR_POP(mpi_errno); }
	    }
499 500 501 502 503 504

	    if (target_predefined) { 
		/* target predefined type, origin derived datatype */

		(*uop)(tmp_buf, (char *) win_ptr->base + win_ptr->disp_unit *
		   target_disp, &target_count, &target_datatype);
505
	    }
506
	    else {
507
	    
508
		segp = MPID_Segment_alloc();
509 510
		MPIU_ERR_CHKANDJUMP1((!segp), mpi_errno, MPI_ERR_OTHER, 
				    "**nomem","**nomem %s","MPID_Segment_alloc"); 
511 512 513 514 515
		MPID_Segment_init(NULL, target_count, target_datatype, segp, 0);
		first = 0;
		last  = SEGMENT_IGNORE_LAST;
		
		MPID_Datatype_get_ptr(target_datatype, dtp);
516
		vec_len = dtp->max_contig_blocks * target_count + 1; 
517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
		/* +1 needed because Rob says so */
		MPIU_CHKLMEM_MALLOC(dloop_vec, DLOOP_VECTOR *, 
				    vec_len * sizeof(DLOOP_VECTOR), 
				    mpi_errno, "dloop vector");
		
		MPID_Segment_pack_vector(segp, first, &last, dloop_vec, &vec_len);
		
		source_buf = (tmp_buf != NULL) ? tmp_buf : origin_addr;
		target_buf = (char *) win_ptr->base + 
		    win_ptr->disp_unit * target_disp;
		type = dtp->eltype;
		type_size = MPID_Datatype_get_basic_size(type);
		for (i=0; i<vec_len; i++)
		{
		    count = (dloop_vec[i].DLOOP_VECTOR_LEN)/type_size;
		    (*uop)((char *)source_buf + MPIU_PtrToAint(dloop_vec[i].DLOOP_VECTOR_BUF),
			   (char *)target_buf + MPIU_PtrToAint(dloop_vec[i].DLOOP_VECTOR_BUF),
			   &count, &type);
		}
		
		MPID_Segment_free(segp);
	    }
539 540 541 542 543
	}
    }
    else
    {
	/* queue it up */
544
	MPIU_INSTR_DURATION_START(rmaqueue_alloc);
545 546
	MPIU_CHKPMEM_MALLOC(new_ptr, MPIDI_RMA_ops *, sizeof(MPIDI_RMA_ops), 
			    mpi_errno, "RMA operation entry");
547 548 549
	MPIU_INSTR_DURATION_END(rmaqueue_alloc);
	if (win_ptr->rma_ops_list_tail) 
	    win_ptr->rma_ops_list_tail->next = new_ptr;
550
	else
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
	    win_ptr->rma_ops_list_head = new_ptr;
	win_ptr->rma_ops_list_tail = new_ptr;

	/* If predefined and contiguous, use a simplified element */
	if (origin_predefined && target_predefined && enableShortACC) {
	    new_ptr->next = NULL;
	    new_ptr->type = MPIDI_RMA_ACC_CONTIG;
	    /* Only the information needed for the contig/predefined acc */
	    new_ptr->origin_addr = origin_addr;
	    new_ptr->origin_count = origin_count;
	    new_ptr->origin_datatype = origin_datatype;
	    new_ptr->target_rank = target_rank;
	    new_ptr->target_disp = target_disp;
	    new_ptr->target_count = target_count;
	    new_ptr->target_datatype = target_datatype;
	    new_ptr->op = op;
	    goto fn_exit;
568
	}
569

570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641
	new_ptr->next = NULL;  
	new_ptr->type = MPIDI_RMA_ACCUMULATE;
	new_ptr->origin_addr = origin_addr;
	new_ptr->origin_count = origin_count;
	new_ptr->origin_datatype = origin_datatype;
	new_ptr->target_rank = target_rank;
	new_ptr->target_disp = target_disp;
	new_ptr->target_count = target_count;
	new_ptr->target_datatype = target_datatype;
	new_ptr->op = op;
	
	/* if source or target datatypes are derived, increment their
	   reference counts */ 
	if (!origin_predefined)
	{
	    MPID_Datatype_get_ptr(origin_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
	if (!target_predefined)
	{
	    MPID_Datatype_get_ptr(target_datatype, dtp);
	    MPID_Datatype_add_ref(dtp);
	}
    }

 fn_exit:
    MPIU_CHKLMEM_FREEALL();
    MPIDI_RMA_FUNC_EXIT(MPID_STATE_MPIDI_ACCUMULATE);
    return mpi_errno;

    /* --BEGIN ERROR HANDLING-- */
  fn_fail:
    MPIU_CHKPMEM_REAP();
    goto fn_exit;
    /* --END ERROR HANDLING-- */
}


#undef FUNCNAME
#define FUNCNAME MPIDI_Alloc_mem
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
void *MPIDI_Alloc_mem( size_t size, MPID_Info *info_ptr )
{
    void *ap;
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_ALLOC_MEM);

    MPIDI_FUNC_ENTER(MPID_STATE_MPIDI_ALLOC_MEM);

    ap = MPIU_Malloc(size);
    
    MPIDI_FUNC_EXIT(MPID_STATE_MPIDI_ALLOC_MEM);
    return ap;
}


#undef FUNCNAME
#define FUNCNAME MPIDI_Free_mem
#undef FCNAME
#define FCNAME MPIDI_QUOTE(FUNCNAME)
int MPIDI_Free_mem( void *ptr )
{
    int mpi_errno = MPI_SUCCESS;
    MPIDI_STATE_DECL(MPID_STATE_MPIDI_FREE_MEM);

    MPIDI_FUNC_ENTER(MPID_STATE_MPIDI_FREE_MEM);

    MPIU_Free(ptr);
    
    MPIDI_FUNC_EXIT(MPID_STATE_MPIDI_FREE_MEM);
    return mpi_errno;
}