2011-11-02 19:07:57 +04:00
|
|
|
/* -*- Mode: C; c-basic-offset:4 ; indent-tabs-mode:nil -*- */
|
2011-10-21 01:39:44 +04:00
|
|
|
/*
|
2014-10-18 03:06:33 +04:00
|
|
|
* Copyright (c) 2010-2014 Los Alamos National Security, LLC. All rights
|
|
|
|
* reserved.
|
2011-10-21 01:39:44 +04:00
|
|
|
* $COPYRIGHT$
|
|
|
|
*
|
|
|
|
* Additional copyrights may follow
|
|
|
|
*
|
|
|
|
* $HEADER$
|
|
|
|
*/
|
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
#include "opal_config.h"
|
2011-10-21 01:39:44 +04:00
|
|
|
|
|
|
|
#include "btl_vader.h"
|
|
|
|
#include "btl_vader_frag.h"
|
|
|
|
#include "btl_vader_endpoint.h"
|
2013-07-12 00:54:12 +04:00
|
|
|
#include "btl_vader_xpmem.h"
|
2011-10-21 01:39:44 +04:00
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
#if OPAL_BTL_VADER_HAVE_CMA
|
2014-01-06 23:51:44 +04:00
|
|
|
#include <sys/uio.h>
|
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
#if OPAL_CMA_NEED_SYSCALL_DEFS
|
2014-01-06 23:51:44 +04:00
|
|
|
#include "opal/sys/cma.h"
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
#endif /* OPAL_CMA_NEED_SYSCALL_DEFS */
|
2014-01-06 23:51:44 +04:00
|
|
|
|
|
|
|
#endif
|
|
|
|
|
2011-10-21 01:39:44 +04:00
|
|
|
/**
|
|
|
|
* Initiate an synchronous get.
|
|
|
|
*
|
|
|
|
* @param btl (IN) BTL module
|
|
|
|
* @param endpoint (IN) BTL addressing information
|
|
|
|
* @param descriptor (IN) Description of the data to be transferred
|
|
|
|
*/
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
#if OPAL_BTL_VADER_HAVE_XPMEM
|
2015-01-06 02:02:41 +03:00
|
|
|
int mca_btl_vader_get_xpmem (mca_btl_base_module_t *btl, mca_btl_base_endpoint_t *endpoint, void *local_address,
|
|
|
|
uint64_t remote_address, mca_btl_base_registration_handle_t *local_handle,
|
|
|
|
mca_btl_base_registration_handle_t *remote_handle, size_t size, int flags,
|
|
|
|
int order, mca_btl_base_rdma_completion_fn_t cbfunc, void *cbcontext, void *cbdata)
|
2011-10-21 01:39:44 +04:00
|
|
|
{
|
2015-11-02 22:07:08 +03:00
|
|
|
mca_rcache_base_registration_t *reg;
|
2011-10-21 01:39:44 +04:00
|
|
|
void *rem_ptr;
|
|
|
|
|
2015-01-06 02:02:41 +03:00
|
|
|
/* silence warning about unused arguments */
|
|
|
|
(void) local_handle;
|
|
|
|
(void) remote_handle;
|
|
|
|
|
|
|
|
reg = vader_get_registation (endpoint, (void *)(intptr_t) remote_address, size, 0, &rem_ptr);
|
2013-07-12 00:54:12 +04:00
|
|
|
if (OPAL_UNLIKELY(NULL == rem_ptr)) {
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
return OPAL_ERROR;
|
2011-10-21 01:39:44 +04:00
|
|
|
}
|
|
|
|
|
2015-01-06 02:02:41 +03:00
|
|
|
vader_memmove (local_address, rem_ptr, size);
|
2011-10-21 01:39:44 +04:00
|
|
|
|
2013-03-28 02:10:02 +04:00
|
|
|
vader_return_registration (reg, endpoint);
|
2011-10-21 01:39:44 +04:00
|
|
|
|
2014-09-23 22:11:22 +04:00
|
|
|
/* always call the callback function */
|
2015-01-06 02:02:41 +03:00
|
|
|
cbfunc (btl, endpoint, local_address, local_handle, cbcontext, cbdata, OPAL_SUCCESS);
|
2011-10-21 01:39:44 +04:00
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
return OPAL_SUCCESS;
|
2011-10-21 01:39:44 +04:00
|
|
|
}
|
2014-10-18 03:06:33 +04:00
|
|
|
#endif
|
|
|
|
|
|
|
|
#if OPAL_BTL_VADER_HAVE_CMA
|
2015-01-06 02:02:41 +03:00
|
|
|
int mca_btl_vader_get_cma (mca_btl_base_module_t *btl, mca_btl_base_endpoint_t *endpoint, void *local_address,
|
|
|
|
uint64_t remote_address, mca_btl_base_registration_handle_t *local_handle,
|
|
|
|
mca_btl_base_registration_handle_t *remote_handle, size_t size, int flags,
|
|
|
|
int order, mca_btl_base_rdma_completion_fn_t cbfunc, void *cbcontext, void *cbdata)
|
2014-01-06 23:51:44 +04:00
|
|
|
{
|
2015-01-06 02:02:41 +03:00
|
|
|
struct iovec src_iov = {.iov_base = (void *)(intptr_t) remote_address, .iov_len = size};
|
|
|
|
struct iovec dst_iov = {.iov_base = local_address, .iov_len = size};
|
2014-01-06 23:51:44 +04:00
|
|
|
ssize_t ret;
|
|
|
|
|
2014-10-18 03:06:33 +04:00
|
|
|
ret = process_vm_readv (endpoint->segment_data.other.seg_ds->seg_cpid, &dst_iov, 1, &src_iov, 1, 0);
|
2014-08-11 17:34:44 +04:00
|
|
|
if (ret != (ssize_t)size) {
|
|
|
|
opal_output(0, "Read %ld, expected %lu, errno = %d\n", (long)ret, (unsigned long)size, errno);
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
return OPAL_ERROR;
|
2014-01-06 23:51:44 +04:00
|
|
|
}
|
|
|
|
|
2014-09-23 22:11:22 +04:00
|
|
|
/* always call the callback function */
|
2015-01-06 02:02:41 +03:00
|
|
|
cbfunc (btl, endpoint, local_address, local_handle, cbcontext, cbdata, OPAL_SUCCESS);
|
2014-01-06 23:51:44 +04:00
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
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.
2014-07-26 04:47:28 +04:00
|
|
|
return OPAL_SUCCESS;
|
2014-01-06 23:51:44 +04:00
|
|
|
}
|
|
|
|
#endif
|
2014-10-18 03:06:33 +04:00
|
|
|
|
|
|
|
#if OPAL_BTL_VADER_HAVE_KNEM
|
2015-01-06 02:02:41 +03:00
|
|
|
int mca_btl_vader_get_knem (mca_btl_base_module_t *btl, mca_btl_base_endpoint_t *endpoint, void *local_address,
|
|
|
|
uint64_t remote_address, mca_btl_base_registration_handle_t *local_handle,
|
|
|
|
mca_btl_base_registration_handle_t *remote_handle, size_t size, int flags,
|
|
|
|
int order, mca_btl_base_rdma_completion_fn_t cbfunc, void *cbcontext, void *cbdata)
|
2014-10-18 03:06:33 +04:00
|
|
|
{
|
|
|
|
struct knem_cmd_param_iovec recv_iovec;
|
|
|
|
struct knem_cmd_inline_copy icopy;
|
|
|
|
|
|
|
|
/* Fill in the ioctl data fields. There's no async completion, so
|
|
|
|
we don't need to worry about getting a slot, etc. */
|
2015-01-06 02:02:41 +03:00
|
|
|
recv_iovec.base = (uintptr_t) local_address;
|
2014-10-18 03:06:33 +04:00
|
|
|
recv_iovec.len = size;
|
|
|
|
icopy.local_iovec_array = (uintptr_t) &recv_iovec;
|
|
|
|
icopy.local_iovec_nr = 1;
|
2015-01-06 02:02:41 +03:00
|
|
|
icopy.remote_cookie = remote_handle->cookie;
|
|
|
|
icopy.remote_offset = remote_address - remote_handle->base_addr;
|
2014-10-18 03:06:33 +04:00
|
|
|
icopy.write = 0;
|
|
|
|
icopy.flags = 0;
|
|
|
|
|
|
|
|
/* Use the DMA flag if knem supports it *and* the segment length
|
|
|
|
* is greater than the cutoff. Not that if DMA is not supported
|
|
|
|
* or the user specified 0 for knem_dma_min the knem_dma_min was
|
|
|
|
* set to UINT_MAX in mca_btl_vader_knem_init. */
|
2015-01-06 02:02:41 +03:00
|
|
|
if (mca_btl_vader_component.knem_dma_min <= size) {
|
2014-10-18 03:06:33 +04:00
|
|
|
icopy.flags = KNEM_FLAG_DMA;
|
|
|
|
}
|
|
|
|
/* synchronous flags only, no need to specify icopy.async_status_index */
|
|
|
|
|
|
|
|
/* When the ioctl returns, the transfer is done and we can invoke
|
|
|
|
the btl callback and return the frag */
|
|
|
|
if (OPAL_UNLIKELY(0 != ioctl (mca_btl_vader.knem_fd, KNEM_CMD_INLINE_COPY, &icopy))) {
|
|
|
|
return OPAL_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (KNEM_STATUS_FAILED == icopy.current_status) {
|
|
|
|
return OPAL_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* always call the callback function */
|
2015-01-06 02:02:41 +03:00
|
|
|
cbfunc (btl, endpoint, local_address, local_handle, cbcontext, cbdata, OPAL_SUCCESS);
|
2014-10-18 03:06:33 +04:00
|
|
|
|
|
|
|
return OPAL_SUCCESS;
|
|
|
|
}
|
|
|
|
#endif
|