1
1
openmpi/ompi/mpi/cxx/intercepts.cc

507 строки
18 KiB
C++
Исходник Обычный вид История

// -*- c++ -*-
//
// Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
// University Research and Technology
// Corporation. All rights reserved.
// Copyright (c) 2004-2005 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 (c) 2006-2008 Cisco Systems, Inc. All rights reserved.
// Copyright (c) 2009 Sun Microsystems, Inc. All rights reserved.
// $COPYRIGHT$
//
// Additional copyrights may follow
//
// $HEADER$
//
#include "mpicxx.h"
#include <stdio.h>
#include "ompi_config.h"
#include "ompi/errhandler/errhandler.h"
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
#include "ompi/communicator/communicator.h"
#include "ompi/datatype/datatype.h"
extern "C"
void ompi_mpi_cxx_throw_exception(int *errcode)
{
#if OMPI_HAVE_CXX_EXCEPTION_SUPPORT
throw(MPI::Exception(*errcode));
#else
// Ick. This is really ugly, but necesary if someone uses a C compiler
// and -lmpi++ (which can legally happen in the LAM MPI implementation,
// and probably in MPICH and others who include -lmpi++ by default in their
// wrapper compilers)
fprintf(stderr, "MPI 2 C++ exception throwing is disabled, MPI::mpi_errno has the error code\n");
MPI::mpi_errno = *errcode;
#endif
}
extern "C"
void ompi_mpi_cxx_comm_throw_excptn_fctn(MPI_Comm *, int *errcode, ...)
{
/* Portland compiler raises a warning if va_start is not used in a
* variable argument function */
va_list ap;
va_start(ap, errcode);
ompi_mpi_cxx_throw_exception(errcode);
va_end(ap);
}
#if OMPI_PROVIDE_MPI_FILE_INTERFACE
extern "C"
void ompi_mpi_cxx_file_throw_excptn_fctn(MPI_File *, int *errcode, ...)
{
va_list ap;
va_start(ap, errcode);
ompi_mpi_cxx_throw_exception(errcode);
va_end(ap);
}
#endif
extern "C"
void ompi_mpi_cxx_win_throw_excptn_fctn(MPI_Win *, int *errcode, ...)
{
va_list ap;
va_start(ap, errcode);
ompi_mpi_cxx_throw_exception(errcode);
va_end(ap);
}
void
MPI::InitializeIntercepts()
{
ompi_mpi_errors_throw_exceptions.eh.eh_comm_fn =
ompi_mpi_cxx_comm_throw_excptn_fctn;
#if OMPI_PROVIDE_MPI_FILE_INTERFACE
ompi_mpi_errors_throw_exceptions.eh.eh_file_fn =
ompi_mpi_cxx_file_throw_excptn_fctn;
#endif
ompi_mpi_errors_throw_exceptions.eh.eh_win_fn =
ompi_mpi_cxx_win_throw_excptn_fctn;
}
// This function uses OMPI types, and is invoked with C linkage for
// the express purpose of having a C++ entity call back the C++
// function (so that types can be converted, etc.).
extern "C"
void ompi_mpi_cxx_comm_errhandler_invoke(ompi_errhandler_t *c_errhandler,
MPI_Comm *c_comm, int *err,
const char *message)
{
// MPI::Comm is an abstract base class; can't instantiate one of
// those. So fake it by instantiating an MPI::Intracomm and then
// casting it down to an (MPI::Comm&) when invoking the callback.
MPI::Intracomm cxx_comm(*c_comm);
MPI::Comm::Errhandler_fn *cxx_fn =
(MPI::Comm::Errhandler_fn*) c_errhandler->eh_comm_fn;
cxx_fn((MPI::Comm&) cxx_comm, err, message);
}
#if OMPI_PROVIDE_MPI_FILE_INTERFACE
// This function uses OMPI types, and is invoked with C linkage for
// the express purpose of having a C++ entity call back the C++
// function (so that types can be converted, etc.).
extern "C"
void ompi_mpi_cxx_file_errhandler_invoke(ompi_errhandler_t *c_errhandler,
MPI_File *c_file, int *err,
const char *message)
{
MPI::File cxx_file(*c_file);
MPI::File::Errhandler_fn *cxx_fn =
(MPI::File::Errhandler_fn*) c_errhandler->eh_file_fn;
cxx_fn(cxx_file, err, message);
}
#endif
// This function uses OMPI types, and is invoked with C linkage for
// the express purpose of having a C++ entity call back the C++
// function (so that types can be converted, etc.).
extern "C"
void ompi_mpi_cxx_win_errhandler_invoke(ompi_errhandler_t *c_errhandler,
MPI_Win *c_win, int *err,
const char *message)
{
MPI::Win cxx_win(*c_win);
MPI::Win::Errhandler_fn *cxx_fn =
(MPI::Win::Errhandler_fn*) c_errhandler->eh_win_fn;
cxx_fn(cxx_win, err, message);
}
// This is a bit weird; bear with me. The user-supplied function for
// MPI::Op contains a C++ object reference. So it must be called from
// a C++-compiled function. However, libmpi does not contain any C++
// code because there are portability and bootstrapping issues
// involved if someone tries to make a 100% C application link against
// a libmpi that contains C++ code. At a minimum, the user will have
// to use the C++ compiler to link. LA-MPI has shown that users don't
// want to do this (there are other problems, but this one is easy to
// cite).
//
// Hence, there are two problems when trying to invoke the user's
// callback funcion from an MPI::Op:
//
// 1. The MPI_Datatype that the C library has must be converted to an
// (MPI::Datatype)
// 2. The C++ callback function must then be called with a
// (MPI::Datatype&)
//
// Some relevant facts for the discussion:
//
// - The main engine for invoking Op callback functions is in libmpi
// (i.e., in C code).
//
// - The C++ bindings are a thin layer on top of the C bindings.
//
// - The C++ bindings are a separate library from the C bindings
// (libmpi_cxx.la).
//
// - As a direct result, the mpiCC wrapper compiler must generate a
// link order thus: "... -lmpi_cxx -lmpi ...", meaning that we cannot
// have a direct function call from the libmpi to libmpi_cxx. We can
// only do it by function pointer.
//
// So the problem remains -- how to invoke a C++ MPI::Op callback
// function (which only occurrs for user-defined datatypes, BTW) from
// within the C Op callback engine in libmpi?
//
// It is easy to cache a function pointer to the
// ompi_mpi_cxx_op_intercept() function on the MPI_Op (that is located
// in the libmpi_cxx library, and is therefore compiled with a C++
// compiler). But the normal C callback MPI_User_function type
// signature is (void*, void*, int*, MPI_Datatype*) -- so if
// ompi_mpi_cxx_op_intercept() is invoked with these arguments, it has
// no way to deduce what the user-specified callback function is that
// is associated with the MPI::Op.
//
// One can easily imagine a scenario of caching the callback pointer
// of the current MPI::Op in a global variable somewhere, and when
// ompi_mpi_cxx_op_intercept() is invoked, simply use that global
// variable. This is unfortunately not thread safe.
//
// So what we do is as follows:
//
// 1. The C++ dispatch function ompi_mpi_cxx_op_intercept() is *not*
// of type (MPI_User_function*). More specifically, it takes an
// additional argument: a function pointer. its signature is (void*,
// void*, int*, MPI_Datatype*, MPI_Op*, MPI::User_function*). This
// last argument is the function pointer of the user callback function
// to be invoked.
//
// The careful reader will notice that it is impossible for the C Op
// dispatch code in libmpi to call this function properly because the
// last argument is of a type that is not defined in libmpi (i.e.,
// it's only in libmpi_cxx). Keep reading -- this is explained below.
//
// 2. When the MPI::Op is created (in MPI::Op::Init()), we call the
// back-end C MPI_Op_create() function as normal (just like the F77
// bindings, in fact), and pass it the ompi_mpi_cxx_op_intercept()
// function (casting it to (MPI_User_function*) -- it's a function
// pointer, so its size is guaranteed to be the same, even if the
// signature of the real function is different).
//
// 3. The function pointer to ompi_mpi_cxx_op_intercept() will be
// cached in the MPI_Op in op->o_func[0].cxx_intercept_fn.
//
// Recall that MPI_Op is implemented to have an array of function
// pointers so that optimized versions of reduction operations can be
// invoked based on the corresponding datatype. But when an MPI_Op
// represents a user-defined function operation, there is only one
// function, so it is always stored in function pointer array index 0.
//
// 4. When MPI_Op_create() returns, the C++ MPI::Op::Init function
// manually sets OMPI_OP_FLAGS_CXX_FUNC flag on the resulting MPI_Op
// (again, very similar to the F77 MPI_OP_CREATE wrapper). It also
// caches the user's C++ callback function in op->o_func[1].c_fn
// (recall that the array of function pointers is actually a union of
// multiple different function pointer types -- it doesn't matter
// which type the user's callback function pointer is stored in; since
// all the types in the union are function pointers, it's guaranteed
// to be large enough to hold what we need.
//
// Note that we don't have a member of the union for the C++ callback
// function because its signature includes a (MPI::Datatype&), which
// we can't put in the C library libmpi.
//
// 5. When the user invokes an function that uses the MPI::Op (or,
// more specifically, when the Op dispatch engine in ompi/op/op.c [in
// libmpi] tries to dispatch off to it), it will see the
// OMPI_OP_FLAGS_CXX_FUNC flag and know to use the
// op->o_func[0].cxx_intercept_fn and also pass as the 4th argument,
// op->o_func[1].c_fn.
//
// 6. ompi_mpi_cxx_op_intercept() is therefore invoked and receives
// both the (MPI_Datatype*) (which is easy to convert to
// (MPI::Datatype&)) and a pointer to the user's C++ callback function
// (albiet cast as the wrong type). So it casts the callback function
// pointer to (MPI::User_function*) and invokes it.
//
// Wasn't that simple?
//
extern "C" void
ompi_mpi_cxx_op_intercept(void *invec, void *outvec, int *len,
MPI_Datatype *datatype, MPI_User_function *c_fn)
{
MPI::Datatype cxx_datatype = *datatype;
MPI::User_function *cxx_callback = (MPI::User_function*) c_fn;
cxx_callback(invec, outvec, *len, cxx_datatype);
}
//
// Attribute copy functions -- comm, type, and win
//
extern "C" int
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
ompi_mpi_cxx_comm_copy_attr_intercept(MPI_Comm comm, int keyval,
void *extra_state,
void *attribute_val_in,
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
void *attribute_val_out, int *flag,
MPI_Comm newcomm)
{
int ret = 0;
MPI::Comm::keyval_intercept_data_t *kid =
(MPI::Comm::keyval_intercept_data_t*) extra_state;
// The callback may be in C or C++. If it's in C, it's easy - just
// call it with no extra C++ machinery.
if (NULL != kid->c_copy_fn) {
return kid->c_copy_fn(comm, keyval, kid->extra_state, attribute_val_in,
attribute_val_out, flag);
}
// If the callback was C++, we have to do a little more work
MPI::Intracomm intracomm;
MPI::Intercomm intercomm;
MPI::Graphcomm graphcomm;
MPI::Cartcomm cartcomm;
bool bflag = OPAL_INT_TO_BOOL(*flag);
if (NULL != kid->cxx_copy_fn) {
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
if (OMPI_COMM_IS_GRAPH(comm)) {
graphcomm = MPI::Graphcomm(comm);
ret = kid->cxx_copy_fn(graphcomm, keyval, kid->extra_state,
attribute_val_in, attribute_val_out,
bflag);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_CART(comm)) {
cartcomm = MPI::Cartcomm(comm);
ret = kid->cxx_copy_fn(cartcomm, keyval, kid->extra_state,
attribute_val_in, attribute_val_out,
bflag);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_INTRA(comm)) {
intracomm = MPI::Intracomm(comm);
ret = kid->cxx_copy_fn(intracomm, keyval, kid->extra_state,
attribute_val_in, attribute_val_out,
bflag);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_INTER(comm)) {
intercomm = MPI::Intercomm(comm);
ret = kid->cxx_copy_fn(intercomm, keyval, kid->extra_state,
attribute_val_in, attribute_val_out,
bflag);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else {
ret = MPI::ERR_COMM;
}
} else {
ret = MPI::ERR_OTHER;
}
*flag = (int)bflag;
return ret;
}
extern "C" int
ompi_mpi_cxx_comm_delete_attr_intercept(MPI_Comm comm, int keyval,
void *attribute_val, void *extra_state)
{
int ret = 0;
MPI::Comm::keyval_intercept_data_t *kid =
(MPI::Comm::keyval_intercept_data_t*) extra_state;
// The callback may be in C or C++. If it's in C, it's easy - just
// call it with no extra C++ machinery.
if (NULL != kid->c_delete_fn) {
return kid->c_delete_fn(comm, keyval, attribute_val, kid->extra_state);
}
// If the callback was C++, we have to do a little more work
MPI::Intracomm intracomm;
MPI::Intercomm intercomm;
MPI::Graphcomm graphcomm;
MPI::Cartcomm cartcomm;
if (NULL != kid->cxx_delete_fn) {
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
if (OMPI_COMM_IS_GRAPH(comm)) {
graphcomm = MPI::Graphcomm(comm);
ret = kid->cxx_delete_fn(graphcomm, keyval, attribute_val,
kid->extra_state);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_CART(comm)) {
cartcomm = MPI::Cartcomm(comm);
ret = kid->cxx_delete_fn(cartcomm, keyval, attribute_val,
kid->extra_state);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_INTRA(comm)) {
intracomm = MPI::Intracomm(comm);
ret = kid->cxx_delete_fn(intracomm, keyval, attribute_val,
kid->extra_state);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else if (OMPI_COMM_IS_INTER(comm)) {
intercomm = MPI::Intercomm(comm);
ret = kid->cxx_delete_fn(intercomm, keyval, attribute_val,
kid->extra_state);
Fixes trac:817 The C++ bindings were not tracking keyvals properly -- they were freeing some internal meta data when Free_keyval() was called, not when the keyval was actually destroyed (keyvals are refcounted in the C layer, just like all other MPI objects, because they can live for long after their corresponding Free call is invoked). This commit fixes this problem and several other things: * Add infrastructure on the ompi_attribute_keyval_t for an "extra" destructor pointer that will be invoked during the "real" constructor (i.e., when OBJ_RELEASE puts the refcount to 0). This allows calling back into the C++ layer to release meta data associated with the keyval. * Adjust all cases where keyvals are created to pass in relevant destructors (NULL or the C++ destructor). * Do essentially the same for MPI::Comm, MPI::Win, and MPI:Datatype: * Move several functions out of the .cc file into the _inln.h file since they no longer require locks * Make the 4 Create_keyval() functions call a common back-end keyval creation function that does the Right Thing depending on whether C or C++ function pointers were used for the keyval functions. The back-end function does not call the corresponding C MPI_*_create_keyval function, but rather does the work itself so that it can associate a "destructor" callback for the C++ bindings for when the keyval is actually destroyed. * Change a few type names to be more indicative of what they are (mostly dealing with keyvals [not "keys"]). * Add the 3 missing bindings for MPI::Comm::Create_keyval(). * Remove MPI::Comm::comm_map (and associated types) because it's no longer necessary in the intercepts -- it was a by-product of being a portable C++ bindings layer. Now we can just query the C layer directly to figure out what type a communicator is. This solves some logistics / callback issues, too. * Rename several types, variables, and fix many comments in the back-end C attribute implementation to make the names really reflect what they are (keyvals vs. attributes). The previous names heavily overloaded the name "key" and were ''extremely'' confusing. This commit was SVN r13565. The following Trac tickets were found above: Ticket 817 --> https://svn.open-mpi.org/trac/ompi/ticket/817
2007-02-09 02:50:04 +03:00
} else {
ret = MPI::ERR_COMM;
}
} else {
ret = MPI::ERR_OTHER;
}
return ret;
}
extern "C" int
ompi_mpi_cxx_type_copy_attr_intercept(MPI_Datatype oldtype, int keyval,
void *extra_state, void *attribute_val_in,
void *attribute_val_out, int *flag)
{
int ret = 0;
MPI::Datatype::keyval_intercept_data_t *kid =
(MPI::Datatype::keyval_intercept_data_t*) extra_state;
if (NULL != kid->c_copy_fn) {
// The callback may be in C or C++. If it's in C, it's easy - just
// call it with no extra C++ machinery.
ret = kid->c_copy_fn(oldtype, keyval, kid->extra_state, attribute_val_in,
attribute_val_out, flag);
} else if (NULL != kid->cxx_copy_fn) {
// If the callback was C++, we have to do a little more work
bool bflag = OPAL_INT_TO_BOOL(*flag);
MPI::Datatype cxx_datatype(oldtype);
ret = kid->cxx_copy_fn(cxx_datatype, keyval, kid->extra_state,
attribute_val_in, attribute_val_out, bflag);
*flag = (int)bflag;
} else {
ret = MPI::ERR_TYPE;
}
return ret;
}
extern "C" int
ompi_mpi_cxx_type_delete_attr_intercept(MPI_Datatype type, int keyval,
void *attribute_val, void *extra_state)
{
int ret = 0;
MPI::Datatype::keyval_intercept_data_t *kid =
(MPI::Datatype::keyval_intercept_data_t*) extra_state;
if (NULL != kid->c_delete_fn) {
return kid->c_delete_fn(type, keyval, attribute_val, kid->extra_state);
} else if (NULL != kid->cxx_delete_fn) {
MPI::Datatype cxx_datatype(type);
return kid->cxx_delete_fn(cxx_datatype, keyval, attribute_val,
kid->extra_state);
} else {
ret = MPI::ERR_TYPE;
}
return ret;
}
extern "C" int
ompi_mpi_cxx_win_copy_attr_intercept(MPI_Win oldwin, int keyval,
void *extra_state, void *attribute_val_in,
void *attribute_val_out, int *flag)
{
int ret = 0;
MPI::Win::keyval_intercept_data_t *kid =
(MPI::Win::keyval_intercept_data_t*) extra_state;
if (NULL != kid->c_copy_fn) {
// The callback may be in C or C++. If it's in C, it's easy - just
// call it with no extra C++ machinery.
ret = kid->c_copy_fn(oldwin, keyval, kid->extra_state, attribute_val_in,
attribute_val_out, flag);
} else if (NULL != kid->cxx_copy_fn) {
// If the callback was C++, we have to do a little more work
bool bflag = OPAL_INT_TO_BOOL(*flag);
MPI::Win cxx_win(oldwin);
ret = kid->cxx_copy_fn(cxx_win, keyval, kid->extra_state,
attribute_val_in, attribute_val_out, bflag);
*flag = (int)bflag;
} else {
ret = MPI::ERR_WIN;
}
return ret;
}
extern "C" int
ompi_mpi_cxx_win_delete_attr_intercept(MPI_Win win, int keyval,
void *attribute_val, void *extra_state)
{
int ret = 0;
MPI::Win::keyval_intercept_data_t *kid =
(MPI::Win::keyval_intercept_data_t*) extra_state;
if (NULL != kid->c_delete_fn) {
return kid->c_delete_fn(win, keyval, attribute_val, kid->extra_state);
} else if (NULL != kid->cxx_delete_fn) {
MPI::Win cxx_win(win);
return kid->cxx_delete_fn(cxx_win, keyval, attribute_val,
kid->extra_state);
} else {
ret = MPI::ERR_WIN;
}
return ret;
}
// For similar reasons as above, we need to intercept calls for the 3
// generalized request callbacks (convert arguments to C++ types and
// invoke the C++ callback signature).
extern "C" int
ompi_mpi_cxx_grequest_query_fn_intercept(void *state, MPI_Status *status)
{
MPI::Grequest::Intercept_data_t *data =
(MPI::Grequest::Intercept_data_t *) state;
MPI::Status s(*status);
int ret = data->id_cxx_query_fn(data->id_extra, s);
*status = s;
return ret;
}
extern "C" int
ompi_mpi_cxx_grequest_free_fn_intercept(void *state)
{
MPI::Grequest::Intercept_data_t *data =
(MPI::Grequest::Intercept_data_t *) state;
int ret = data->id_cxx_free_fn(data->id_extra);
// Delete the struct that was "new"ed in MPI::Grequest::Start()
delete data;
return ret;
}
extern "C" int
ompi_mpi_cxx_grequest_cancel_fn_intercept(void *state, int cancelled)
{
MPI::Grequest::Intercept_data_t *data =
(MPI::Grequest::Intercept_data_t *) state;
return data->id_cxx_cancel_fn(data->id_extra,
(0 != cancelled ? true : false));
}