1
1
Граф коммитов

717 Коммитов

Автор SHA1 Сообщение Дата
Aurelien Bouteiller
c16339944a Fix a coverity warning about using unsafe sprintf.
This commit was SVN r17999.
2008-03-27 21:24:27 +00:00
Aurelien Bouteiller
e11237aadb Introduction of the "progress" sender_based method to replace the slow isend-self method.
This commit was SVN r17998.
2008-03-27 21:19:45 +00:00
Aurelien Bouteiller
93db01871e This is part of the previous patch.
This commit was SVN r17997.
2008-03-27 21:06:14 +00:00
Aurelien Bouteiller
f8bf6f2c6a Code cleanup.
sender_based.h is now split in two files, to solve cyclic .h files inclusion. 
Most macros are now inline functions.
Variable names have been changed from places to places.
Various other small things... 

This commit was SVN r17996.
2008-03-27 21:05:44 +00:00
Gleb Natapov
cf40674369 Decide if sends should be throttled at the receiver and pass this to the sender
in an ACK message. The decision can't be done reliably at the sender.

This commit was SVN r17987.
2008-03-27 08:56:43 +00:00
Galen Shipman
0116041133 BTL shouldn't own the passive side's descriptor in the PML get protocol. The BTL
doesn't know when to free it on the passive side. 

This commit was SVN r17943.
2008-03-25 01:43:41 +00:00
George Bosilca
8943ae0b4e Cleanup plus some typos.
This commit was SVN r17858.
2008-03-18 03:03:33 +00:00
Josh Hursey
612ebdc2ac Cleanup some symbol visability issues.
This commit was SVN r17733.
2008-03-05 13:59:25 +00:00
Ralph Castain
d70e2e8c2b Merge the ORTE devel branch into the main trunk. Details of what this means will be circulated separately.
Remains to be tested to ensure everything came over cleanly, so please continue to withhold commits a little longer

This commit was SVN r17632.
2008-02-28 01:57:57 +00:00
Aurelien Bouteiller
76e6334a57 This change is a mistake. CONVERTOR METHOD does not work with unpatched trunk. Revert back to PACK_METHOD.
This commit was SVN r17629.
2008-02-27 20:02:25 +00:00
Aurelien Bouteiller
1d57b8b0e0 Replaced all the (long) cast by PRIsize_t. Should solve definitely compiler warnings that appeared from time to time depending on sizeof(size_t)...
This commit was SVN r17627.
2008-02-27 19:58:18 +00:00
George Bosilca
fa31ec81d0 Add the ownership flags to the PML/BTL interface. The layer
owning the descriptor is responsible for releasing it once
the descriptor is not in use anymore.

This commit was SVN r17497.
2008-02-18 17:39:30 +00:00
Shiqing Fan
653857ddbe Wrong function name was copied here.
This commit was SVN r17486.
2008-02-17 19:47:47 +00:00
Gleb Natapov
354c5bc5e1 Don't call progress() from OB1 fragment scheduling functions. They don't serve
any purpose and case recursion calls to progress engine.

This commit was SVN r17478.
2008-02-17 12:42:32 +00:00
Aurelien Bouteiller
3ffe845187 Fixed warning.
This commit was SVN r17454.
2008-02-14 15:18:19 +00:00
Gleb Natapov
0a1fa2cb56 req_match_received is set inside MCA_PML_OB1_RECV_REQUEST_MATCHE().
This commit was SVN r17442.
2008-02-13 08:34:39 +00:00
Gleb Natapov
876f49f1a7 Remove unnecessary assignment. It is done later in the same function.
This commit was SVN r17441.
2008-02-13 08:28:25 +00:00
Shiqing Fan
54c7b71cfd Use the correct way of including memchecker.h, which will work with '--with-devel-headers'.
This commit was SVN r17435.
2008-02-12 18:01:17 +00:00
Rainer Keller
7621800477 - Fix and add comments -- output full name for pd
- Protect argument in macro...

This commit was SVN r17434.
2008-02-12 16:59:59 +00:00
Jeff Squyres
6adc5015f9 This file was accidentally re-introduced in r17409.
This commit was SVN r17428.

The following SVN revision numbers were found above:
  r17409 --> open-mpi/ompi@98f70d6318
2008-02-12 13:07:44 +00:00
Shiqing Fan
f5792bbda5 merging the memchecker into trunk.
This commit was SVN r17424.
2008-02-12 08:46:27 +00:00
George Bosilca
55179b833c Unexpected ... Removing unistd.h from datatype.h break the compilation
of the pml_base_bsend ... 

This commit was SVN r17412.
2008-02-10 21:49:19 +00:00
Aurelien Bouteiller
4da1258d60 Quick fix for static builds (mca_component_retain always return failure in static build mode, so just blatently ignore the failure. Though, this may crash severly sometime later if the failure occurs while in dso mode.
This commit was SVN r17328.
2008-01-30 10:41:49 +00:00
George Bosilca
4e703741b7 Move the PML tags into the legal range.
This commit was SVN r17326.
2008-01-30 00:09:45 +00:00
Aurelien Bouteiller
2fd8230025 Windows might not be the only one...
This commit was SVN r17296.
2008-01-29 07:44:33 +00:00
Aurelien Bouteiller
bd10a0231f Replaced the explicit include of inttypes.h by the opal replacement.
This commit was SVN r17295.
2008-01-29 07:35:14 +00:00
Aurelien Bouteiller
e261861f4a Major build system modification. Removed symlinks (problem with make dist), solved issues with static builds and can accept most compile options. The only unsupported compile option for now is --enable-mca-no-build=pml-v. Still investigating this...
This commit was SVN r17294.
2008-01-29 06:07:57 +00:00
George Bosilca
fad6136794 To be or not to be ! As DR require 64 bits atomics, only allow it to
build when thread support is disabled or we have 64 bits atomics support.

This commit was SVN r17293.
2008-01-29 05:24:56 +00:00
George Bosilca
c5d5fcf50a Protect the standard header file, and allow the PML V to compile
on Windows.

This commit was SVN r17250.
2008-01-26 18:43:06 +00:00
Aurelien Bouteiller
ca8eb1fb30 There should be no leftovers of configuration phase after distclean
This commit was SVN r17249.
2008-01-26 09:56:02 +00:00
Aurelien Bouteiller
b5d44261a0 Fix one warning about extremely long lines (due to macro expansion)
This commit was SVN r17247.
2008-01-26 00:38:33 +00:00
Aurelien Bouteiller
48cabdc40b Changed build system. Should be more distcheck, VPATH, static and other compilation mode friendly.
This commit was SVN r17245.
2008-01-25 23:57:01 +00:00
Rainer Keller
f7e586fc01 - allow --enable-mca-direct=pml-ob1
This commit was SVN r17227.
2008-01-25 09:56:45 +00:00
Aurelien Bouteiller
e471abb55e put back ompi ignore until long filenames and other dist issues are fixed
This commit was SVN r17219.
2008-01-25 00:28:30 +00:00
Aurelien Bouteiller
11815d9773 Fixed two warnings (especially the one that get repeted a large number of times in 64bit builds)
This commit was SVN r17197.
2008-01-24 04:59:31 +00:00
Aurelien Bouteiller
a9045402c4 remove a pedantic warning
This commit was SVN r17196.
2008-01-24 02:29:07 +00:00
Aurelien Bouteiller
76b13f91b9 fixed link:wq error in static mode
This commit was SVN r17194.
2008-01-23 23:54:02 +00:00
Aurelien Bouteiller
f29ed2ed53 fixed missing errno.h on some architectures
This commit was SVN r17186.
2008-01-23 20:24:54 +00:00
Aurelien Bouteiller
6fe17aff4a solve compatibility issue from MMAP_NOCACHE
This commit was SVN r17184.
2008-01-23 19:29:19 +00:00
Aurelien Bouteiller
69b3bae999 removed ignore, as the code is robust enough to avoid interfering with others
This commit was SVN r17182.
2008-01-23 17:27:23 +00:00
Gleb Natapov
6e4155d111 Initialize local variable before use.
This commit was SVN r17170.
2008-01-21 15:17:49 +00:00
George Bosilca
6310ce955c The first patch related to the Active Message stuff. So far, here is what we have:
- the registration array is now global instead of one by BTL.
- each framework have to declare the entries in the registration array reserved. Then
  it have to define the internal way of sharing (or not) these entries between all
  components. As an example, the PML will not share as there is only one active PML
  at any moment, while the BTLs will have to. The tag is 8 bits long, the first 3
  are reserved for the framework while the remaining 5 are use internally by each
  framework.
- The registration function is optional. If a BTL do not provide such function,
  nothing happens. However, in the case where such function is provided in the BTL
  structure, it will be called by the BML, when a tag is registered.

Now, it's time for the second step... Converting OB1 from a switch based PML to an
active message one.

This commit was SVN r17140.
2008-01-15 05:32:53 +00:00
George Bosilca
98f79f2ea0 Remove the second declaration of the PML V component.
This commit was SVN r17139.
2008-01-15 05:26:26 +00:00
Jon Mason
a0d4122606 The new cpc selection framework is now in place. The patch below allows
for dynamic selection of cpc methods based on what is available.  It
also allows for inclusion/exclusions of methods.  It even futher allows
for modifying the priorities of certain cpc methods to better determine
the optimal cpc method.

This patch also contains XRC compile time disablement (per Jeff's
patch).

At a high level, the cpc selections works by walking through each cpc
and allowing it to test to see if it is permissable to run on this
mpirun.  It returns a priority if it is permissable or a -1 if not.  All
of the cpc names and priorities are rolled into a string.  This string
is then encapsulated in a message and passed around all the ompi
processes.  Once received and unpacked, the list received is compared
to a local copy of the list.  The connection method is chosen by
comparing the lists passed around to all nodes via modex with the list
generated locally.  Any non-negative number is a potentially valid
connection method.  The method below of determining the optimal
connection method is to take the cross-section of the two lists.  The
highest single value (and the other side being non-negative) is selected
as the cpc method.

svn merge -r 16948:17128 https://svn.open-mpi.org/svn/ompi/tmp-public/openib-cpc/ .

This commit was SVN r17138.
2008-01-14 23:22:03 +00:00
George Bosilca
1bd31aa3ac Cleanup the OMPI_DECLSPEC/OMPI_MODULE_DECLSPEC in the PMLs.
This commit was SVN r17093.
2008-01-09 20:32:39 +00:00
Gleb Natapov
b37ff74a24 Make function that is used only in one file static. Remove static functions
declaration.

This commit was SVN r17080.
2008-01-09 09:54:35 +00:00
Ethan Mallove
f32dcb1636 The Sun Studio 12 compilers need to have inline specified as
`static` in cases where a function is not part of a separate
compilation unit (such as `append_recv_req_to_queue`).

This commit was SVN r17069.
2008-01-08 18:45:51 +00:00
Aurelien Bouteiller
9bf54e1604 Windows compatibility patch.
Also introduces work in progress "convertor" sender based copy algorithm. This algorithm cannot be selected without other modifications in the convertor (not currently available in trunk). The default old synchronous copy algorithm is selected by default.

This commit was SVN r17063.
2008-01-07 23:35:44 +00:00
George Bosilca
d2324050f8 Allow the PML V component to be compiled on Windows. Force all
.c files to include the ompi_config.h as the first #include.

This commit was SVN r17056.
2008-01-05 00:17:32 +00:00
George Bosilca
42414b27e9 Use BEGIN_C_DECLS and END_C_DECLS instead of the ugly #if/#endif.
This commit was SVN r17009.
2007-12-21 06:19:46 +00:00
George Bosilca
b58dae00db Allow PERUSE to compile correctly.
This commit was SVN r17008.
2007-12-21 06:18:19 +00:00
George Bosilca
906e8bf1d1 Replace the ompi_pointer_array with opal_pointer_array. The next step
(sometimes after the merge with the ORTE branch), the opal_pointer_array
will became the only pointer_array implementation (the orte_pointer_array
will be removed).

This commit was SVN r17007.
2007-12-21 06:02:00 +00:00
Gleb Natapov
35bf8c7c46 Rewrite OB1 matching logic. Get rid of macros, make the code shorter.
This commit was SVN r16993.
2007-12-19 09:16:20 +00:00
Gleb Natapov
5cd38b8b06 Better encapsulate heterogeneous arch handling in ob1.
This commit was SVN r16970.
2007-12-16 08:45:44 +00:00
Gleb Natapov
8b511b969d Introduce a new BTL parameter btl_rndv_eager_limit which determines size of a
first fragment of rendezvous protocol. Remove no longer used btl_min_send_size
parameter.

This commit was SVN r16969.
2007-12-16 08:35:17 +00:00
Jeff Squyres
213b5d5c6e Per long threads on the mailing list and much confusion discussion
about linkers, have all OPAL, ORTE, and OMPI components '''not'' link
against the OPAL, ORTE, or OMPI libraries.

See ttp://www.open-mpi.org/community/lists/users/2007/10/4220.php for
details (or https://svn.open-mpi.org/trac/ompi/wiki/Linkers for a
better-formatted version of the same info).

This commit was SVN r16968.
2007-12-15 13:32:02 +00:00
Aurelien Bouteiller
93f39fa190 Fixes various issues with --enable-visibility, C++ and exotic C compilers.
Aurelien 

This commit was SVN r16949.
2007-12-12 19:13:23 +00:00
Gleb Natapov
e0dc53e516 Use mca_bml_base_send_status() in OB1.
This commit was SVN r16905.
2007-12-09 14:13:24 +00:00
Gleb Natapov
e2e211f23b Add flags parameter to btl_alloc() and btl_prepare_src() functions. If BTL
knows at the time of allocation priority of a descriptor it may do some
optimizations.

This commit was SVN r16901.
2007-12-09 14:08:01 +00:00
Gleb Natapov
2d784752dd Remove descriptor caching form BML. With descriptor caching some optimizations
are impossible.

This commit was SVN r16897.
2007-12-09 13:58:17 +00:00
Aurelien Bouteiller
6190c97ee9 PML V and vprotocol framework management of customizable wait/test. This is still a fast and dirty implementation (cleanup of the customized request functions is not totally correct if several component modify them out of order).
This commit was SVN r16890.
2007-12-07 08:21:25 +00:00
Aurelien Bouteiller
859169214c Vprotocol pessimist benefits from customizable requests. Waitany, waitsome, test, testany, testall, testsome can now be hooked and are therefore logged correctly.
This commit was SVN r16885.
2007-12-07 08:17:30 +00:00
Aurelien Bouteiller
15ffe6c89c Accomoding the new interface for free_lists.
This commit was SVN r16727.
2007-11-16 00:00:38 +00:00
Rich Graham
27a748e7eb change all instances of ompi_free_list_init to ompi_free_list_init_new. Header
and payload data are specified separately at this stage.

This commit was SVN r16633.
2007-11-01 23:38:50 +00:00
Rich Graham
67f4b69848 propogate fix for out of buffered send memory space to dr and ob1 - thanks
George.

This commit was SVN r16593.
2007-10-27 00:17:53 +00:00
Rich Graham
9c0483088a if unable to get buffered space, try and progress communications to
free up resources.

This commit was SVN r16591.
2007-10-26 23:16:31 +00:00
Gleb Natapov
52c6160252 MCA_PML_BASE_REQUEST_MPI_COMPLETE() macro does nothing except call to
ompi_request_complete(). Remove the macro and call the function directly.

This commit was SVN r16498.
2007-10-18 14:20:24 +00:00
George Bosilca
aa20a94b6f Remove warning about an unused variable.
This commit was SVN r16497.
2007-10-18 13:48:56 +00:00
Gleb Natapov
4f865e22e8 We have two different version of ompi_request_complete. One as a function
another as a macro. Make it one inline function.

This commit was SVN r16495.
2007-10-18 13:02:27 +00:00
Gleb Natapov
e0a3a7e53e Move duplicated code all over the code to a single function ompi_request_wait_completion().
This commit was SVN r16494.
2007-10-18 12:33:21 +00:00
Gleb Natapov
807f49ed7f If there are more then one BTL present we may divide payload between them in
such a way that converter will not be able to pack some of it. This commit adds
handling of such cases. If converter can't pack any data for a BTL the data is
sent over another BTL that has data to send.

This commit was SVN r16493.
2007-10-18 12:07:37 +00:00
Gleb Natapov
1330974e5e eager_limit is no longer needed in OB1 PML. Remove it.
This commit was SVN r16442.
2007-10-15 09:26:42 +00:00
George Bosilca
e3105a85be Don't require a progress function from the PML. If there is one then the
PML base will take care of the registration with the event library.
Otherwise, (and this apply for the CM case) the MTL are in charge of
registering their own progress function.

This commit was SVN r16415.
2007-10-09 23:28:53 +00:00
Galen Shipman
62ade993ca Seperate finalize and close for the PML, this gives the PML a chance to complete any outstanding operations prior to close. Before this change we just called pml_finalize in pml_close which causes problems if there are outstanding events that a BTL/MTL needs to progress during finalize. The problem is that MPI_COMM_WORLD and others were destroyed prior to closing the PML, pml_close would call pml_finalize, events would progress in the BTL, and these events expected MPI_COMM_WORLD to still be around..
This commit was SVN r16405.
2007-10-09 15:28:56 +00:00
Josh Hursey
7437f37e96 This commit contains the following:
* Fix some missing includes in a few places.
 * Add the cr_request() functionality to the BLCR CRS component.
   We are now dependent upon the 0.6.* series of BLCR.
 * Made the CR notification mechanism a registered function.
   This way we can have an OPAL-only version and it can be replaced at
   runtime with the ORTE version.
 * Add a 'opal_cr_allow_opal_only' parameter that will enable OPAL-only
   CR functionality when the user wants it. Default: Disabled.
 * Fix the placement of a checkpoint request check in MPI_Init
 * Pull the OPAL notification mechanism into the SnapC framework.
   * We no longer fork/exec the 'opal-checkpoint' command for local
   checkpointing, the Local coordinator in the orted does this directly.
   * The Local and Application coordinator talk together bypassing the OPAL
   notifiation mechanism.
   * Optimized the Local <-> App Coordinator communication.
   * Improved the structure used to track vpid_snapshots in the local coord.
 * Fix a race condition in which an application under heavy communication load
   may produce an inconsistent global checkpoint.

This commit was SVN r16389.
2007-10-08 20:53:02 +00:00
Ralph Castain
54b2cf747e These changes were mostly captured in a prior RFC (except for #2 below) and are aimed specifically at improving startup performance and setting up the remaining modifications described in that RFC.
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component.

This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done:

As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in.

In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in.

The incoming changes revamp these procedures in three ways:

1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step.

The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic.

Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure.


2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed.

The size of this data has been reduced in three ways:

(a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes.

To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose.

(b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction.

(c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using.

While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly.


3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup.

It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging.

Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future.


There are a few minor additional changes in the commit that I'll just note in passing:

* propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details.

* requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details.

* cleanup of some stale header files

This commit was SVN r16364.
2007-10-05 19:48:23 +00:00
Aurelien Bouteiller
670956e172 Another cast mistake.
This commit was SVN r16247.
2007-09-26 21:14:35 +00:00
Aurelien Bouteiller
f7d7d58fb6 Various cast type errors on 64bit architectures
This commit was SVN r16246.
2007-09-26 20:54:18 +00:00
Aurelien Bouteiller
0df0087f17 Investigating improvement of cache line management on shared memory
This commit was SVN r16183.
2007-09-21 20:02:56 +00:00
Josh Hursey
3e51d7bb25 Implement the MPI_Iprobe and MPI_Probe wrappers.
Remove some old, unused code.

This commit was SVN r16178.
2007-09-21 16:28:46 +00:00
Aurelien Bouteiller
d3b376a340 This patch adds actual non-blocking sender-based message logging. This improves bandwidth. Still need to work on malloc/mmap storage to reach optimal bandwidth.
This commit was SVN r16172.
2007-09-21 03:24:08 +00:00
Aurelien Bouteiller
bc318b35e2 There is room in convertor to copy the packed data. It works just need to add the correct memcopy. It does not manage the short messages but I alreqdy think of a workaround for this (and it might even be better regarding latency).
This commit was SVN r16169.
2007-09-20 21:57:21 +00:00
Aurelien Bouteiller
bbac6e650a New improved version of sender-based. Under dev but a new framework for expressing various methods have been added.
This commit was SVN r16159.
2007-09-19 03:42:56 +00:00
Gleb Natapov
097b17d30e Prevent a receive request from been freed while other thread holds a reference
to it or there is an outstanding completion for the request.

This commit was SVN r16153.
2007-09-18 16:18:47 +00:00
Aurelien Bouteiller
a1f5312afb Fixed two little warnings
This commit was SVN r16116.
2007-09-12 21:07:11 +00:00
Aurelien Bouteiller
ccb3f75e8f Make sure that the pml v parasite never get loaded when user did not requested FT. This does not break the ability to switch protocol on the fly.
This commit was SVN r16114.
2007-09-12 20:47:17 +00:00
Aurelien Bouteiller
828af95be8 Major modification of the vprotocol framework build system. With a better integration in autogen.sh, it allows for generating static-components.h the usual way.
NOTE: This build system does not work with the current autogen.sh. Modified one is under heavy testing to make sure it does not have side effects 

This commit was SVN r16110.
2007-09-12 18:46:37 +00:00
George Bosilca
05ae27c68b Don't segfault if we receive a fragment for a non existing communicator.
Instead, drop it by now.

This commit was SVN r16105.
2007-09-12 17:52:02 +00:00
Shiqing Fan
a0660f4deb - Just some type casts.
This commit was SVN r16100.
2007-09-12 15:29:58 +00:00
Gleb Natapov
07c8fddeef Fix scheduling of pending send request. It should be scheduled req_lock times.
This commit was SVN r16096.
2007-09-12 07:08:38 +00:00
George Bosilca
d8fed2cfa1 Set a default value so that some compilers stop complaining about
uninitialized values.

This commit was SVN r16094.
2007-09-11 18:00:53 +00:00
Gleb Natapov
79011279e5 Remove debug output.
This commit was SVN r16016.
2007-08-30 13:29:41 +00:00
Gleb Natapov
690fb95bda Cleanup send scheduling code.
This commit was SVN r16014.
2007-08-30 12:10:04 +00:00
Gleb Natapov
0b0f9d14aa Mark send request complete on PML level only when absolutely sure there is
no more work associated with this request. No more outstanding completions or
packets and send scheduling isn't running in another thread.

This commit was SVN r16013.
2007-08-30 12:08:33 +00:00
Gleb Natapov
eac2674f66 The inner voice tells me this is a typo.
This commit was SVN r16004.
2007-08-29 13:28:47 +00:00
Brian Barrett
59b22533f2 Enable RDMA for heterogeneous situations. Currently done by overloading
the ompi_convertor_need_buffers function to only return 0 if the convertor
is homogeneous (which it never does on the trunk, but does to on v1.2, but
that's a different issue).  Only enable the heterogeneous rdma code for
a btl if it supports it (via a flag), as some btls need some work for this
to work properly.  Currently only TCP and OpenIB extensively tested

This commit was SVN r15990.
2007-08-28 21:23:44 +00:00
Gleb Natapov
fa69c5cc10 If a memory on a sender's size is not registered don't register it on a receive
side too. Otherwise a content of the recvreq->req_rdma array is replaced later
without freeing previous content and refcount on registration in mpool become
wrong.

This commit was SVN r15978.
2007-08-28 07:43:06 +00:00
Gleb Natapov
e1a1d9d90e Receive request converter can be accessed in parallel by a thread that receives
data and a thread that run RDMA schedule function. Protect access to the
converter by a lock.

This commit was SVN r15967.
2007-08-27 11:41:42 +00:00
Gleb Natapov
065d04dfde Do not free recvreq while schedule function is running in another thread.
This commit was SVN r15964.
2007-08-27 11:31:40 +00:00
Rainer Keller
1b5fa48a29 - Add missing PERUSE_COMM_REQ_REMOVE_FROM_POSTED_Q when matching
from the posted generic_recv-queue.
 - Move the PERUSE_COMM_MSG_MATCH_POSTED_REQ from
   MCA_PML_OB1_RECV_REQUEST_MATCHED to
   mca_pml_ob1_recv_frag_match() as suggested by Terry Dontje
   Only post, if this is not a probe/iprobe request.
 - Do not post PERUSE_COMM_REQ_MATCH_UNEX for probes / iprobes and
   do in correct order before PERUSE_COMM_MSG_REMOVE_FROM_UNEX_Q

This commit was SVN r15947.
2007-08-23 07:09:43 +00:00