552c9ca5a0
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic. This commit was SVN r32317.
79 строки
2.9 KiB
C
79 строки
2.9 KiB
C
/*
|
|
* Copyright (c) 2004-2007 The Trustees of Indiana University and Indiana
|
|
* University Research and Technology
|
|
* Corporation. All rights reserved.
|
|
* Copyright (c) 2004-2006 The University of Tennessee and The University
|
|
* of Tennessee Research Foundation. All rights
|
|
* reserved.
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
* University of Stuttgart. All rights reserved.
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
* All rights reserved.
|
|
* $COPYRIGHT$
|
|
*
|
|
* Additional copyrights may follow
|
|
*
|
|
* $HEADER$
|
|
*/
|
|
/**
|
|
* @file
|
|
*
|
|
* This file will hopefully not last long in the tree, but it's
|
|
* unfortunately necessary for now.
|
|
*
|
|
* There are multiple places in the code base where we need to safely
|
|
* convert from a size_t to an int. However, on some platforms,
|
|
* sizeof(size_t) is larger than sizeof(int), so casting from size_t
|
|
* -> int will result in a compiler warning and potentially data
|
|
* truncation.
|
|
*
|
|
* But, unfortunately, we still need to do it. But we definitely do
|
|
* not want compiler warnings. So when sizeof(size_t)>sizeof(int),
|
|
* the solution is the treat the size_t value like an array and
|
|
* dereference the appropriate nibble and cast that to an int (which
|
|
* accounts for both big and little endian machines).
|
|
*
|
|
* Most places in the code where this casting must occur are because
|
|
* collision of APIs (e.g., one API requires a size_t and another API
|
|
* requires an int. And in most places, we're not going to overflow
|
|
* the int when casting down into it (e.g., it's the result of a
|
|
* strlen -- if that buffer is larger than MAX_INT, we've got other problems!).
|
|
*
|
|
* BUT -- the whole premise of casting down to an int is dangerous.
|
|
* So we provide extra protection here to detect overflow situations
|
|
* and print out appropriate warnings. So if this situation ever
|
|
* occurs, we'll still overflow, but we'll have a good indication that
|
|
* it's happening, and where.
|
|
*/
|
|
|
|
#ifndef OPAL_CONVERT_H
|
|
#define OPAL_CONVERT_H
|
|
|
|
#include "opal_config.h"
|
|
|
|
BEGIN_C_DECLS
|
|
|
|
/**
|
|
* Convert a size_t to an int.
|
|
*
|
|
* @param in The size_t value to be converted
|
|
* @param out The output int value.
|
|
* @param want_check Whether to check for truncation or not
|
|
*
|
|
* @returns OPAL_SUCESS If all went well
|
|
* @returns OPAL_NOT_SUPPORTED if the size_t value was truncated
|
|
*
|
|
* The conversion will always occur. However, if the size_t value was
|
|
* truncated (i.e., sizeof(size_t) > sizeof(int), and the cast down to
|
|
* the int actually changed the value), OPAL_NOT_SUPPORTED will be
|
|
* returned.
|
|
*
|
|
* On platforms where sizeof(size_t) <= sizeof(int), this function
|
|
* will aways return OPAL_SUCCESS.
|
|
*/
|
|
OPAL_DECLSPEC int opal_size2int(size_t in, int *out, bool want_check) __opal_attribute_nonnull__(2);
|
|
|
|
END_C_DECLS
|
|
|
|
#endif
|