1
1
Граф коммитов

6595 Коммитов

Автор SHA1 Сообщение Дата
George Bosilca
85db48df0e Identification, tab vs. space.
This commit was SVN r29206.
2013-09-18 16:45:00 +00:00
Ralph Castain
f051500166 Sadly, there is no RTE-agnostic way to prune the modex entries, so we must send them all.
Refs trac:3766

This commit was SVN r29204.

The following Trac tickets were found above:
  Ticket 3766 --> https://svn.open-mpi.org/trac/ompi/ticket/3766
2013-09-18 14:09:23 +00:00
Mike Dubman
2a5c342587 Modifications that are necessary in order to meet latest libhcoll API.
cmr:v1.7.3:reviewer=jladd

This commit was SVN r29202.
2013-09-18 12:22:02 +00:00
Ralph Castain
865a7028f8 Per patch from George, with a few minor cleanups. Correctly address the complete exchange of required wireup information in Intercomm_create so all procs in the resulting communicator know how to talk to each other.
Refs trac:29166

This commit was SVN r29200.

The following Trac tickets were found above:
  Ticket 29166 --> https://svn.open-mpi.org/trac/ompi/ticket/29166
2013-09-18 02:01:30 +00:00
Ralph Castain
99611ac1d2 Revert r29166 in favor of a better solution from George
This commit was SVN r29199.

The following SVN revision numbers were found above:
  r29166 --> open-mpi/ompi@497c7e6abb
2013-09-18 01:41:26 +00:00
George Bosilca
55273f1c98 Cleanup spaces, nothing else.
This commit was SVN r29197.
2013-09-18 00:07:58 +00:00
George Bosilca
7b319a101d Fix the case where we build without Fortran support.
This commit was SVN r29194.
2013-09-17 20:45:46 +00:00
Nathan Hjelm
7929fb9dea Cleanup complex datatypes and update datatypes and operator code to use C99.
This commit changes the underlying opal complex datatypes to match the
C99 types: float _Complex, double _Complex, and long double _Complex. The
fortran and C++ types now are aliases to these basic types instead of
structure types. The operators in ompi/mca/op/base now work on only the
C99 types and the fortran types use these operators if the fortran type
matches a C complex type (this should almost always be the case.)

C99 is not is use in both the datatype and operator code and should make
the code both cleaner and much less fragile.

This commit was SVN r29193.
2013-09-17 17:49:42 +00:00
Rolf vandeVaart
440632b57f Add a function that will dump out the contents of the memory registration cache.
Useful for debugging any rcache issues.

This commit was SVN r29189.
2013-09-17 15:40:32 +00:00
Jeff Squyres
74d1278f48 btl_usnic_util.c:ompi_btl_usnic_util_abort() also passes in the strerror().
This commit was SVN r29188.
2013-09-17 12:35:51 +00:00
George Bosilca
5f686a90d0 Fix several issues regarding MPI_IN_PLACE and different flavors
of MPI_Alltoall.
 - add support for MPI_IN_PLACE in the self collective component.
 - fix the extent usage in the tuned collective component.
 - correctly use the peer counts instead of local - add support for MPI_IN_PLACE in the self collective component.
 - fix the extent usage in the tuned collective component.
 - correctly use the peer counts instead of local.
Thanks to Fujitsu for the patch.

This commit was SVN r29187.
2013-09-17 11:35:18 +00:00
Reese Faucette
8f235e6977 usnic: wrong SG entry used to compute length for small put()s
This commit was SVN r29186.
2013-09-17 08:18:02 +00:00
Reese Faucette
651d61f1a3 Clean up debugging logging a bit.
MSGDEBUG2 now means "print a one-liner for all PML calls into BTL, and
also when BTL calls PML with a recv completion (not send completions)"
MSGDEBUG1 means print more internal gory detail
MSGDEBUG is gone, replaced by MSGDEBUG1

In the process also found that PUT_DEST style fragments could
potentially be leaked in usnic_free() since send_fragment tests were
being applied to see if it was eligible to be freed.

This commit was SVN r29185.
2013-09-17 07:29:40 +00:00
Reese Faucette
f35d9b50e3 Cisco CSCuj22803: fixes for Bsend
changes required to support MPI_Bsend().  Introduces concept of
attaching a buffer to a large segment that the PML can scribble into and
we will send from.  The reason we don't use a pinned buffer and send
directly from that is that usnic_verbs does not (yes) support num_sge>1
for regular sends.  This means the data gets copied twice, but that is
unavoidable.

changed the logic in handle_large_send to be more sensible

Incorporated David's review comments

This commit was SVN r29184.
2013-09-17 07:27:39 +00:00
Reese Faucette
25b5c84d0f Cisco CSCuj13135: Data corruption in MPI_Bsend_ator_c
Do not assume that the "size" passed to alloc_send() will be the same as
the size of the message the resulting fragment will hold when
usnic_send() is called.  This means usnic_send()/usnic_put() can never
trust any pre-computed size values, and are only allowed to look at the
lengths and pointers of the elements in the desc SG list.

This commit was SVN r29183.
2013-09-17 07:25:05 +00:00
Reese Faucette
b9103c0f66 Cisco CSCuj12524: c_put_big segfault
- usnic_free() cannot free the fragment until ACK is received

This commit was SVN r29182.
2013-09-17 07:23:15 +00:00
Reese Faucette
89b5f0899b Cisco CSCuj12520: various problems running c_fence_put_1
- tag needs to be sent in *our* header, not the PML header
- usnic_alloc() should return smaller value if too much data requested
- be careful about callbacks vs removing items from lists
  (we need to remove from outr lists *before* the callback)
- improve send callback handling
- add some more MSGDEBUG2 logging and cleanup

This commit was SVN r29181.
2013-09-17 07:20:44 +00:00
Mike Dubman
432c10750a enable mxm2 from np>0
reviewed by yossi
cmr:v1.7.3:reviewer=jsquyres

This commit was SVN r29178.
2013-09-16 12:36:28 +00:00
Mike Dubman
44bfa95553 enable mxm2 by default on np>=0
reviewed by yossi
cmr:v1.7.3:reviewer=jsquyres

This commit was SVN r29177.
2013-09-16 12:32:29 +00:00
Ralph Castain
52caa75552 Gar - forgot to commit a few more cleanups
Refs trac:3696

This commit was SVN r29168.

The following Trac tickets were found above:
  Ticket 3696 --> https://svn.open-mpi.org/trac/ompi/ticket/3696
2013-09-15 15:32:01 +00:00
Ralph Castain
b64c8dafd8 Cleanup some errors in pubsub - must set the active flag before posting the recv in case the message has already arrived
Refs trac:3696

This commit was SVN r29167.

The following Trac tickets were found above:
  Ticket 3696 --> https://svn.open-mpi.org/trac/ompi/ticket/3696
2013-09-15 15:26:32 +00:00
Ralph Castain
497c7e6abb Fixes trac:2904
The intercomm "merge" function can create a linkage between procs that was not reflected anywhere in a modex, and so at least some of the procs in the resulting communicator don't know how to talk to some of the new communicator's peers.

For example, consider the case where:

1. parent job A comm_spawns a process (job B) - these processes exchange modex and can communicate

2. parent job A now comm_spawns another process (job C) - again, these can communicate, but the proc in C knows nothing of B

3. do an intercomm merge across the communicators created by the two comm_spawns. This puts B and C into the same communicator, but they know nothing about how to talk to each other as they were not involved in any exchange of contact info. Hence, collectives on that communicator now fail. 

This fix adds an API to the ompi/dpm framework that (a) exchanges the modex info across the procs in the merge to ensure all procs know how to communicate, and (b) calls add_procs to give the btl's a chance to select transports to any new procs.

cmr:v1.7.3:reviewer=jsquyres

This commit was SVN r29166.

The following Trac tickets were found above:
  Ticket 2904 --> https://svn.open-mpi.org/trac/ompi/ticket/2904
2013-09-15 15:00:40 +00:00
Rolf vandeVaart
096b8c022e Also add flag to debug output.
This commit was SVN r29163.
2013-09-13 19:47:05 +00:00
Rolf vandeVaart
c15b2a26b8 Fix some formatting. Move some CUDA-aware mca parameter initialization earlier.
This commit was SVN r29162.
2013-09-13 17:43:41 +00:00
Rolf vandeVaart
d247c26b84 In the case that HAVE_IBV_FORK_INIT is not defined, we will need this variable so we can give the user an error if they ask for it. Also fixes compile error when HAVE_IBV_FORK_INIT is not defined.
This commit was SVN r29160.
2013-09-13 14:38:49 +00:00
Rolf vandeVaart
ba9ec1b8bc For debug builds, add the ability to view memory registrations and deregistrations in the openib BTL.
This commit was SVN r29159.
2013-09-13 14:28:26 +00:00
Joshua Ladd
b3f88c4a1d Per the RFC schedule, this commit adds Mellanox OpenSHMEM to the trunk. It does not yet run on OSX or with CM PML for an MTL other than MXM. Mellanox is aware of these issues and is in the process of resolving them. This should be added to \ncmr=v1.7.4:subject=Move OSHMEM to 1.7.4:reviewer=rhc
This commit was SVN r29153.
2013-09-10 15:34:09 +00:00
Jeff Squyres
c9f05a2664 Delineate OMPI_FREE_LIST_*_MT separately.
The FREE_LIST_*_MT stuff was introduced on the SVN trunk in r28722
(2013-07-04), but so far, has not been merged into the v1.7 branch yet
(2013-09-06).  So put it in its own #ifdef, rather than defining it
based on OMPI_MAJOR_VERSION/OMPI_MINOR_VERSION.

This commit was SVN r29148.

The following SVN revision numbers were found above:
  r28722 --> open-mpi/ompi@c9e5ab9ed1
2013-09-06 19:22:56 +00:00
Jeff Squyres
e02cc0a7ec No need for this header file.
This commit was SVN r29147.
2013-09-06 19:22:28 +00:00
Jeff Squyres
c53b0890cf Ensure that btl_usnic_compat.h is in the tarball.
This commit was SVN r29140.
2013-09-06 15:53:56 +00:00
Dave Goodell
75fa28c303 usnic: v1.6<->trunk unification, trunk side
The Cisco-maintained v1.6 port of the usnic BTL has diverged from the
upstream trunk and v1.7 branches.  This commit adjusts the trunk to more
closely match the v1.6 branch to simplify future merging and
cherry-picking.

The usnic MCA parameters also need work on this side.

Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29138.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:21:34 +00:00
Dave Goodell
a669bd01e6 usnic: revamp convertor handling.
The fix for the HPL SEGV was incorrect because it assumed the
prepare_src() routine was always allowed to return "bytes processed"
less than the requested "bytes to send".  It turns out this is only true
if the convertor is what limits the size, we are not allowed to limit
the data sent for our own reasons, else we break login in the upper
layers.

This means we need to learn the number of bytes out of the size
requested the convertor will give us, no matter how big the size is.
Unfortunately, this is a destructive test, and (currently) the only way to
learn that number is to actually have the convertor copy the data out into
buffers.

This change implements this, copying the entire data out into a chain of
send segments which are attached to the large send fragment.  Now we can
always return the proper size value to the PML.

Fixes Cisco bug CSCuj08024

Authored-by: Reese Faucette <rfaucett@cisco.com>

Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29137.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:21:21 +00:00
Dave Goodell
0ef8336502 new bookkeeping code should return value indicating whether packet is good or not.
Authored-by: Reese Faucette <rfaucett@cisco.com>

Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29136.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:19:32 +00:00
Dave Goodell
122890c2fd usnic: "bookeeping" --> "bookkeeping"
Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29135.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:19:20 +00:00
Dave Goodell
0df6ed4acc usnic: squash warnings from perf improvements
Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29134.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:19:08 +00:00
Dave Goodell
6dc54d372d usnic: Basket of performance changes including:
- round segment buffer allocation to cache-line
    - split some routines into an inline fast section and a called
      slower section
    - introduce receive fastpath in component_progress that:
        o returns immediately if there is a packet available on priority
          queue and fastpath is enabled
        o disables fastpath for 1 time after use to provide fairness to
          other processing
        o defers receive buffer posting
        o defers bookeeping for receive until next call
          to usnic_component_progress

Authored-by: Reese Faucette <rfaucett@cisco.com>

Should be included in usnic v1.7.3 roll-up CMR (refs trac:3760)

This commit was SVN r29133.

The following Trac tickets were found above:
  Ticket 3760 --> https://svn.open-mpi.org/trac/ompi/ticket/3760
2013-09-06 03:18:57 +00:00
Dave Goodell
9cab9777d9 usnic: properly destroy embedded small send frag
Without this, an `--enable-debug` build would hit an assertion in the
list code when run under valgrind with `--malloc-fill=0xff` or any other
case where malloc returned non-zeroed buffers.

Also allow the normal OBJ_ machinery to handle the constructor
invocation ordering for us instead of doing it by hand (which could have
led to future bugs).

Reviewed-by: jsquyres@cisco.com

cmr=v1.7.4

Depends on trunk functionality in r29095 and r29096.  Refs trac:3740,#3741.

This commit was SVN r29127.

The following SVN revision numbers were found above:
  r29095 --> open-mpi/ompi@d1b5940e97
  r29096 --> open-mpi/ompi@a552921171

The following Trac tickets were found above:
  Ticket 3740 --> https://svn.open-mpi.org/trac/ompi/ticket/3740
2013-09-04 20:59:12 +00:00
George Bosilca
2aed876be6 Fix the bug where MPI_is_thread_main returns true for all
threads when not in MPI_THREAD_MULTIPLE mode.
Thanks to Lisandro Dalcin for pointing it out.

This commit was SVN r29113.
2013-09-04 11:10:51 +00:00
Jeff Squyres
f6619f8e9e Fix compile error in the heterogeneous case.
We've been forcing C99 compiler compliance for a while now, so use C99
syntax to keep the #if code tidy.

This commit was SVN r29101.
2013-08-31 12:56:08 +00:00
Brian Barrett
16a1166884 Remove the proc_pml and proc_bml fields from ompi_proc_t and replace with a
configure-time dynamic allocation of flags.  The net result for platforms
which only support BTL-based communication is a reduction of 8*nprocs bytes
per process.  Platforms which support both MTLs and BTLs will not see
a space reduction, but will now be able to safely run both the MTL and BTL
side-by-side, which will prove useful.

This commit was SVN r29100.
2013-08-30 16:54:55 +00:00
Rolf vandeVaart
18962d296b This has bothered me for a while. Change MCA_BTL_TAG_BTL to MCA_BTL_TAG_IB. They are the same
value so this does not change anything.  (MCA_BTL_TAG_IB = MCA_BTL_TAG_BTL + 0).  This just makes it more correct.

This commit was SVN r29099.
2013-08-30 14:53:59 +00:00
Dave Goodell
c5a7e8a079 usnic: stomp format specifier warnings
The usnic BTL now builds cleanly under `--enable-picky` when `MSGDEBUG1`
is set.

Reviewed-by: jsquyres

cmr=v1.7.4:reviewer=jsquyres

This commit was SVN r29097.
2013-08-29 23:24:14 +00:00
Dave Goodell
a552921171 call element destructors in ompi_free_list_destruct
The free list code called the constructor for each object it
slab allocated in ompi_free_list_grow.  This permits free list-managed
elements to safely allocate/deallocate resources in their
constructors/destructors without leaking.

It's probably best to let this soak on the trunk a little while before
moving it over to v1.7.

Reviewed-by: bosilca

cmr=v1.7.4:reviewer=bosilca

This commit was SVN r29096.
2013-08-29 22:56:31 +00:00
Dave Goodell
d1b5940e97 don't call OMPI_REQUEST_INIT in req constructor
Calling OMPI_REQUEST_INIT puts the request into an _INACTIVE state
instead of an _INVALID state, which we don't want if it's been
simply been constructed, e.g., in a free_list.  Without this change a
future change to call destructors at free list destruction time will
result in request dtor state assertion failures.

Reviewed-by: bosilca

cmr=v1.7.4:reviewer=bosilca

This commit was SVN r29095.
2013-08-29 22:56:22 +00:00
Ralph Castain
5d1fa4fa0e Silence warnings:
osc_pt2pt_data_move.c: In function 'ompi_osc_pt2pt_sendreq_recv_accum_long_cb':
osc_pt2pt_data_move.c:643:9: warning: variable 'ret' set but not used [-Wunused-but-set-variable]
osc_rdma_data_move.c: In function 'ompi_osc_rdma_control_send_cb':
osc_rdma_data_move.c:1312:37: warning: variable 'header' set but not used [-Wunused-but-set-variable]

This commit was SVN r29092.
2013-08-29 20:56:36 +00:00
Ralph Castain
537e7380b1 As per the discussion on the devel telecon, do not compute ompi_comm_world_thread_level_mult if thread multiple is disabled. We aren't using the value anyway, but we will leave the current code in-place until we understand if it is needed or not.
This commit was SVN r29080.
2013-08-28 17:44:04 +00:00
George Bosilca
305fa88d4b Remove two warnings from the SM BTL. The return code can be safely ignored
as the internals of the SM BTL will repost the fragment until the send operation
succesfully complete.

This commit was SVN r29077.
2013-08-28 06:36:01 +00:00
George Bosilca
badd011ac3 Minor cleanup.
This commit was SVN r29076.
2013-08-28 05:48:58 +00:00
Dave Goodell
dd82bd3c19 usnic: fix invalid rfstart initialization
endpoint_rfstart was being initialized from a value which was not yet
set.  Also ensure that rfstart is a valid index in the range
0..WINDOW_SIZE-1, since it is used as the index into endpoint_rcvd_segs,
which has WINDOW_SIZE elements.

Without this change there is significant risk of memory corruption or
segfaults, resulting in hangs or crashes, if malloc ever returns us a
value >=WINDOW_SIZE (4096).  Right now we seem to be getting lucky that
the malloc is returning zero-pages to us when we are allocating endpoint
structures (possibly because the freelist performs a single large
allocation for all endpoints).

Fixes Cisco bug CSCui88781.

Reviewed-by: rfaucett@cisco.com
Reviewed-by: jsquyres@cisco.com

cmr=v1.7.3:reviewer=jsquyres

This commit was SVN r29075.
2013-08-27 22:43:20 +00:00
Nathan Hjelm
c699ee7812 Update the ompi_info man page with information about variable levels
and improve the behavior of ompi_info.

This commit changes the default behavior of ompi_info --all when a
level is not specified. Instead of assuming level 1 in this case we
now assume level 9. This change is due to feedback from the community
after the introduction of the --level option.

I also added a new option: --selected-only. This option will limit the
displayed variables to components that can be selected (ie. if there
is a selection parameter set-- btl self,sm)

cmr=v1.7.3:reviewer=jsquyres

This commit was SVN r29070.
2013-08-27 19:11:37 +00:00