ports/openmpi
ports/openmpi
1
1
Fork 0
Code Releases Activity

This is actually a much smaller commit than it appears at first glance - it just touches a lot of files. The --without-rte-support configuration option has never really been implemented completely. The option caused various objects not to be defined and conditionally compiled some base functions, but did nothing to prevent build of the component libraries. Unfortunately, since many of those components use objects covered by the option, it caused builds to break if those components were allowed to build.

Browse Source 
Brian dealt with this in the past by creating platform files and using "no-build" to block the components. This was clunky, but acceptable when only one organization was using that option. However, that number has now expanded to at least two more locations.

Accordingly, make --without-rte-support actually work by adding appropriate configury to prevent components from building when they shouldn't. While doing so, remove two frameworks (db and rmcast) that are no longer used as ORCM comes to a close (besides, they belonged in ORCM now anyway). Do some minor cleanups along the way.

This commit was SVN r25497.
This commit is contained in:
Ralph Castain 2011-11-22 21:24:35 +00:00
parent cce0949bda
commit 9b59d8de6f
206 changed files with 853 additions and 7231 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ompi/mca/bml/r2/Makefile.am
View File

@ -21,8 +21,8 @@ r2_sources = \
bml_r2.c \
bml_r2.h \
bml_r2_component.c \
bml_r2_ft.c \
bml_r2_ft.h
bml_r2_ft.c \
bml_r2_ft.h
dist_pkgdata_DATA = help-mca-bml-r2.txt

4
ompi/mca/bml/r2/bml_r2_ft.c
View File

@ -38,6 +38,7 @@
int mca_bml_r2_ft_event(int state)
{
#if !ORTE_DISABLE_FULL_SUPPORT
static bool first_continue_pass = false;
ompi_proc_t** procs = NULL;
size_t num_procs;
@ -297,6 +298,7 @@ int mca_bml_r2_ft_event(int state)
else {
;
}
#endif
return OMPI_SUCCESS;
}

21
ompi/mca/dpm/orte/configure.m4 Normal file
View File

@ -0,0 +1,21 @@
# -*- shell-script -*-
#
# Copyright (c) 2011 Los Alamos National Security, LLC.
# All rights reserved.
#
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
# MCA_dpm_orte_CONFIG([action-if-found], [action-if-not-found])
# -----------------------------------------------------------
AC_DEFUN([MCA_ompi_dpm_orte_CONFIG],[
AC_CONFIG_FILES([ompi/mca/dpm/orte/Makefile])
AS_IF([test "$orte_without_full_support" = 0],
[$1],
[$2])
])dnl

21
ompi/mca/pubsub/orte/configure.m4 Normal file
View File

@ -0,0 +1,21 @@
# -*- shell-script -*-
#
# Copyright (c) 2011 Los Alamos National Security, LLC.
# All rights reserved.
#
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
# MCA_pubsub_orte_CONFIG([action-if-found], [action-if-not-found])
# -----------------------------------------------------------
AC_DEFUN([MCA_ompi_pubsub_orte_CONFIG],[
AC_CONFIG_FILES([ompi/mca/pubsub/orte/Makefile])
AS_IF([test "$orte_without_full_support" = 0],
[$1],
[$2])
])dnl

32
ompi/tools/ompi_info/components.c
View File

@ -10,7 +10,7 @@
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
* Copyright (c) 2006-2010 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2010 Los Alamos National Security, LLC.
* Copyright (c) 2010-2011 Los Alamos National Security, LLC.
* All rights reserved.
* Copyright (c) 2011 University of Houston. All rights reserved.
* $COPYRIGHT$
@ -103,15 +103,13 @@
#include "orte/mca/errmgr/base/base.h"
#include "orte/mca/grpcomm/grpcomm.h"
#include "orte/mca/grpcomm/base/base.h"
#include "orte/mca/db/db.h"
#include "orte/mca/db/base/base.h"
#include "orte/mca/ess/ess.h"
#include "orte/mca/ess/base/base.h"
#include "orte/mca/notifier/notifier.h"
#include "orte/mca/notifier/base/base.h"
#include "orte/util/show_help.h"
#include "orte/util/proc_info.h"
#if !ORTE_DISABLE_FULL_SUPPORT
#include "orte/mca/notifier/notifier.h"
#include "orte/mca/notifier/base/base.h"
#include "orte/mca/debugger/debugger.h"
#include "orte/mca/debugger/base/base.h"
#include "orte/mca/iof/iof.h"
@ -124,8 +122,6 @@
#include "orte/mca/ras/base/ras_private.h"
#include "orte/mca/rmaps/rmaps.h"
#include "orte/mca/rmaps/base/base.h"
#include "orte/mca/rmcast/rmcast.h"
#include "orte/mca/rmcast/base/base.h"
#include "orte/mca/rml/rml.h"
#include "orte/mca/rml/base/base.h"
#include "orte/mca/routed/routed.h"
@ -418,14 +414,6 @@ void ompi_info_open_components(void)
map->components = &orte_grpcomm_base.components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_db_base_open()) {
goto error;
}
map = OBJ_NEW(ompi_info_component_map_t);
map->type = strdup("db");
map->components = &orte_db_base_components_available;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_ess_base_open()) {
goto error;
}
@ -434,6 +422,7 @@ void ompi_info_open_components(void)
map->components = &orte_ess_base_components_available;
opal_pointer_array_add(&component_map, map);
#if !ORTE_DISABLE_FULL_SUPPORT
if (ORTE_SUCCESS != orte_notifier_base_open()) {
goto error;
}
@ -442,7 +431,6 @@ void ompi_info_open_components(void)
map->components = &orte_notifier_base_components_available;
opal_pointer_array_add(&component_map, map);
#if !ORTE_DISABLE_FULL_SUPPORT
if (ORTE_SUCCESS != orte_debugger_base_open()) {
goto error;
}
@ -491,14 +479,6 @@ void ompi_info_open_components(void)
map->components = &orte_rmaps_base.available_components;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_rmcast_base_open()) {
goto error;
}
map = OBJ_NEW(ompi_info_component_map_t);
map->type = strdup("rmcast");
map->components = &orte_rmcast_base.rmcast_opened;
opal_pointer_array_add(&component_map, map);
if (ORTE_SUCCESS != orte_rml_base_open()) {
goto error;
}
@ -799,8 +779,6 @@ void ompi_info_close_components()
(void) ompi_osc_base_close();
(void) orte_grpcomm_base_close();
(void) orte_db_base_close();
(void) orte_notifier_base_close();
(void) orte_ess_base_close();
(void) orte_show_help_finalize();
#if !ORTE_DISABLE_FULL_SUPPORT
@ -808,6 +786,7 @@ void ompi_info_close_components()
(void) orte_snapc_base_close();
(void) orte_sstore_base_close();
#endif
(void) orte_notifier_base_close();
(void) orte_filem_base_close();
(void) orte_iof_base_close();
(void) orte_plm_base_close();
@ -818,7 +797,6 @@ void ompi_info_close_components()
(void) orte_routed_base_close();
(void) mca_oob_base_close();
(void) orte_rmcast_base_close();
#endif
(void) orte_errmgr_base_close();

2
ompi/tools/ompi_info/ompi_info.c
View File

@ -239,7 +239,6 @@ int main(int argc, char *argv[])
opal_pointer_array_add(&mca_types, "odls");
opal_pointer_array_add(&mca_types, "ras");
opal_pointer_array_add(&mca_types, "rmaps");
opal_pointer_array_add(&mca_types, "rmcast");
opal_pointer_array_add(&mca_types, "rml");
opal_pointer_array_add(&mca_types, "routed");
opal_pointer_array_add(&mca_types, "plm");
@ -256,7 +255,6 @@ int main(int argc, char *argv[])
opal_pointer_array_add(&mca_types, "errmgr");
opal_pointer_array_add(&mca_types, "ess");
opal_pointer_array_add(&mca_types, "grpcomm");
opal_pointer_array_add(&mca_types, "db");
opal_pointer_array_add(&mca_types, "notifier");
/* Execute the desired action(s) */

33
orte/config/orte_configure_options.m4
View File

@ -77,25 +77,6 @@ AC_DEFINE_UNQUOTED([ORTE_WANT_ORTERUN_PREFIX_BY_DEFAULT],
[$orte_want_orterun_prefix_by_default],
[Whether we want orterun to effect "--prefix $prefix" by default])
#
# Do we want reliable multicast enabled?
#
AC_MSG_CHECKING([if want reliable multicast])
AC_ARG_ENABLE([multicast],
[AC_HELP_STRING([--enable-multicast],
[Enable reliable multicast messaging (default: disabled)])])
if test "$enable_multicast" = "yes"; then
AC_MSG_RESULT([yes])
orte_want_multicast=1
else
AC_MSG_RESULT([no])
orte_want_multicast=0
fi
AC_DEFINE_UNQUOTED([ORTE_ENABLE_MULTICAST],
[$orte_want_multicast],
[Whether we want multicast messaging enabled])
#
# Do we want sensors enabled?
@ -135,16 +116,22 @@ AC_DEFINE_UNQUOTED([ORTE_ENABLE_HEARTBEAT],
#
# Compile in resilient runtime code
#
AC_MSG_CHECKING([if want resilient runtime code enabled])
AC_ARG_ENABLE(resilient-orte,
[AC_HELP_STRING([--enable-resilient-orte], [Enable the resilient runtime code.])])
AS_IF( [test "$enable_resilient_orte" = "yes"], [result=1], [result=0] )
if test "$enable_resilient_orte" = "yes"; then
AC_MSG_RESULT([yes])
orte_enable_resilient_code=1
else
AC_MSG_RESULT([no])
orte_enable_resilient_code=0
fi
AM_CONDITIONAL(ORTE_RESIL_ORTE, [test "$enable_resilient_orte" = "yes"])
AC_DEFINE_UNQUOTED([ORTE_RESIL_ORTE], [$result],
AC_DEFINE_UNQUOTED([ORTE_RESIL_ORTE], [$orte_enable_resilient_code],
[Compile a resilient version of Open MPI])
AM_CONDITIONAL(ORTE_ENABLE_EPOCH, [test "$enable_resilient_orte" = "yes"])
AC_DEFINE_UNQUOTED([ORTE_ENABLE_EPOCH], [$result],
AC_DEFINE_UNQUOTED([ORTE_ENABLE_EPOCH], [$orte_enable_resilient_code],
[Support for epoch in the ORTE process name enabled or not])

32
orte/mca/db/Makefile.am
View File

@ -1,32 +0,0 @@
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
AM_CPPFLAGS = $(LTDLINCL)
# main library setup
noinst_LTLIBRARIES = libmca_db.la
libmca_db_la_SOURCES =
# pkgdata setup
dist_pkgdata_DATA =
# local files
headers = db.h
libmca_db_la_SOURCES += $(headers)
# Conditionally install the header files
if WANT_INSTALL_HEADERS
ortedir = $(includedir)/openmpi/$(subdir)
nobase_orte_HEADERS = $(headers)
endif
include base/Makefile.am
distclean-local:
rm -f base/static-components.h

18
orte/mca/db/base/Makefile.am
View File

@ -1,18 +0,0 @@
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
dist_pkgdata_DATA += base/help-db-base.txt
headers += \
base/base.h
libmca_db_la_SOURCES += \
base/db_base_open.c \
base/db_base_close.c \
base/db_base_select.c

54
orte/mca/db/base/base.h
View File

@ -1,54 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
/** @file:
*/
#ifndef MCA_DB_BASE_H
#define MCA_DB_BASE_H
#include "orte_config.h"
#include "orte/types.h"
#include "opal/mca/mca.h"
#include "opal/class/opal_list.h"
#include "orte/mca/db/db.h"
BEGIN_C_DECLS
/*
* Global functions for MCA overall collective open and close
*/
/**
* Open the db framework
*/
ORTE_DECLSPEC int orte_db_base_open(void);
/**
* Select a db module
*/
ORTE_DECLSPEC int orte_db_base_select(void);
/**
* Close the db framework
*/
ORTE_DECLSPEC int orte_db_base_close(void);
/*
* The verbose channel for debug output
*/
ORTE_DECLSPEC extern int orte_db_base_output;
ORTE_DECLSPEC extern opal_list_t orte_db_base_components_available;
END_C_DECLS
#endif

50
orte/mca/db/base/db_base_close.c
View File

@ -1,50 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "orte_config.h"
#include "orte/constants.h"
#include <stdio.h>
#include "opal/mca/mca.h"
#include "opal/util/output.h"
#include "opal/mca/base/base.h"
#include "opal/mca/base/mca_base_component_repository.h"
#include "orte/mca/db/base/base.h"
extern opal_list_t orte_db_base_components_available;
int
orte_db_base_close(void)
{
opal_list_item_t *item;
mca_base_component_list_item_t *cli;
if (NULL != orte_db.finalize) {
orte_db.finalize();
}
/* unload all remaining components */
while (NULL != (item = opal_list_remove_first(&orte_db_base_components_available))) {
orte_db_base_component_t* component;
cli = (mca_base_component_list_item_t *) item;
component = (orte_db_base_component_t*) cli->cli_component;
opal_output_verbose(10, 0,
"orte_db_base_close: module %s unloaded",
component->base_version.mca_component_name);
mca_base_component_repository_release((mca_base_component_t *) component);
OBJ_RELEASE(item);
}
OBJ_DESTRUCT(&orte_db_base_components_available);
opal_output_close(orte_db_base_output);
return ORTE_SUCCESS;
}

97
orte/mca/db/base/db_base_open.c
View File

@ -1,97 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "orte_config.h"
#include "orte/constants.h"
#include "opal/mca/mca.h"
#include "opal/util/output.h"
#include "opal/mca/base/base.h"
#include "opal/dss/dss_types.h"
#include "orte/mca/db/base/base.h"
/*
* The following file was created by configure. It contains extern
* dbments and the definition of an array of pointers to each
* module's public mca_base_module_t struct.
*/
#include "orte/mca/db/base/static-components.h"
opal_list_t orte_db_base_components_available;
/* provide "NULL" functions */
static int init(void);
static int finalize(void);
static int store(char *key, void *object, opal_data_type_t type);
static int set_source(orte_process_name_t *name);
static int fetch(char *key, void *object, opal_data_type_t type);
static int update(char *key, void *object, opal_data_type_t type);
static int remove_data(char *key);
orte_db_base_module_t orte_db = {
init,
finalize,
store,
set_source,
fetch,
update,
remove_data
};
int orte_db_base_output;
int
orte_db_base_open(void)
{
orte_db_base_output = opal_output_open(NULL);
OBJ_CONSTRUCT(&orte_db_base_components_available, opal_list_t);
/* Open up all available components */
if (ORTE_SUCCESS !=
mca_base_components_open("db", orte_db_base_output, mca_db_base_static_components,
&orte_db_base_components_available,
true)) {
return ORTE_ERROR;
}
return ORTE_SUCCESS;
}
static int init(void)
{
return ORTE_SUCCESS;
}
static int finalize(void)
{
return ORTE_SUCCESS;
}
static int store(char *key, void *object, opal_data_type_t type)
{
return ORTE_SUCCESS;
}
static int set_source(orte_process_name_t *name)
{
return ORTE_SUCCESS;
}
static int fetch(char *key, void *object, opal_data_type_t type)
{
return ORTE_SUCCESS;
}
static int update(char *key, void *object, opal_data_type_t type)
{
return ORTE_SUCCESS;
}
static int remove_data(char *key)
{
return ORTE_SUCCESS;
}

48
orte/mca/db/base/db_base_select.c
View File

@ -1,48 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "orte_config.h"
#include "orte/constants.h"
#include "opal/class/opal_list.h"
#include "opal/mca/mca.h"
#include "opal/mca/base/base.h"
#include "opal/mca/base/mca_base_component_repository.h"
#include "orte/mca/db/base/base.h"
extern opal_list_t orte_db_base_components_available;
int
orte_db_base_select(void)
{
orte_db_base_component_t *best_component = NULL;
orte_db_base_module_t *best_module = NULL;
/*
* Select the best component
*/
if( OPAL_SUCCESS != mca_base_select("db", orte_db_base_output,
&orte_db_base_components_available,
(mca_base_module_t **) &best_module,
(mca_base_component_t **) &best_component) ) {
/* It is okay to not select a component - default
* to using the base NULL component
*/
return ORTE_SUCCESS;
}
/* Save and init the winner */
orte_db = *best_module;
if (NULL != orte_db.init) {
orte_db.init();
}
return ORTE_SUCCESS;
}

19
orte/mca/db/base/help-db-base.txt
View File

@ -1,19 +0,0 @@
-*- text -*-
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
#
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
# This is the US/English general help file for ORTE Errmgr HNP module.
#
[errmgr-hnp:unknown-job-error]
An error has occurred in an unknown job. This generally should not happen
except due to an internal ORTE error.
Job state: %s
This information should probably be reported to the OMPI developers.

38
orte/mca/db/daemon/Makefile.am
View File

@ -1,38 +0,0 @@
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
sources = \
db_daemon.h \
db_daemon_component.c \
db_daemon.c
# Make the output library in this directory, and name it either
# mca_<type>_<name>.la (for DSO builds) or libmca_<type>_<name>.la
# (for static builds).
if MCA_BUILD_orte_db_daemon_DSO
component_noinst =
component_install = mca_db_daemon.la
else
component_noinst = libmca_db_daemon.la
component_install =
endif
mcacomponentdir = $(pkglibdir)
mcacomponent_LTLIBRARIES = $(component_install)
mca_db_daemon_la_CPPFLAGS = $(db_daemon_CPPFLAGS)
mca_db_daemon_la_SOURCES = $(sources)
mca_db_daemon_la_LDFLAGS = -module -avoid-version $(db_daemon_LDFLAGS)
mca_db_daemon_la_LIBADD = $(db_daemon_LIBS)
noinst_LTLIBRARIES = $(component_noinst)
libmca_db_daemon_la_CPPFLAGS = $(db_daemon_CPPFLAGS)
libmca_db_daemon_la_SOURCES =$(sources)
libmca_db_daemon_la_LDFLAGS = -module -avoid-version $(db_daemon_LDFLAGS)
libmca_db_daemon_la_LIBADD = $(db_daemon_LIBS)

19
orte/mca/db/daemon/configure.m4
View File

@ -1,19 +0,0 @@
# -*- shell-script -*-
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
# MCA_db_daemon_CONFIG([action-if-found], [action-if-not-found])
# -----------------------------------------------------------
AC_DEFUN([MCA_orte_db_daemon_CONFIG], [
AC_CONFIG_FILES([orte/mca/db/daemon/Makefile])
# cant run this component without multicast
AS_IF([test "$orte_want_multicast" = "1"],
[$1], [$2])
])dnl

440
orte/mca/db/daemon/db_daemon.c
View File

@ -1,440 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2004-2011 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*
*/
#include "orte_config.h"
#include "orte/constants.h"
#include <time.h>
#include "opal/class/opal_pointer_array.h"
#include "opal/dss/dss_types.h"
#include "opal/util/output.h"
#include "orte/util/show_help.h"
#include "orte/util/name_fns.h"
#include "orte/mca/errmgr/errmgr.h"
#include "orte/mca/rmcast/rmcast.h"
#include "orte/runtime/orte_globals.h"
#include "orte/runtime/orte_wait.h"
#include "orte/mca/db/base/base.h"
#include "db_daemon.h"
static int init(void);
static int finalize(void);
static int store(char *key, void *object, opal_data_type_t type);
static int set_source(orte_process_name_t *name);
static int fetch(char *key, void *object, opal_data_type_t type);
static int update(char *key, void *object, opal_data_type_t type);
static int remove_data(char *key);
orte_db_base_module_t orte_db_daemon_module = {
init,
finalize,
store,
set_source,
fetch,
update,
remove_data
};
/* local types */
typedef struct {
opal_object_t super;
orte_process_name_t name;
char *key;
int32_t size;
uint8_t *bytes;
} orte_db_data_t;
static void dtconstructor(orte_db_data_t *dt)
{
dt->key = NULL;
dt->bytes = NULL;
dt->size = 0;
}
static void dtdestructor(orte_db_data_t *dt)
{
if (NULL != dt->key) {
free(dt->key);
}
if (NULL != dt->bytes) {
free(dt->bytes);
}
}
OBJ_CLASS_INSTANCE(orte_db_data_t,
opal_object_t,
dtconstructor,
dtdestructor);
/* local variables */
static orte_vpid_t num_recvd;
static bool ack_reqd;
static opal_pointer_array_t datastore;
static orte_rmcast_channel_t my_group_channel;
/* local functions */
static void callback_fn(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata);
static void recv_cmd(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata);
static void recv_ack(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata);
#include MCA_timer_IMPLEMENTATION_HEADER
static inline double gettime(void) __opal_attribute_always_inline__;
static inline double gettime(void)
{
double wtime;
#if OPAL_TIMER_USEC_NATIVE
wtime = ((double) opal_timer_base_get_usec()) / 1000000.0;
#else
struct timeval tv;
gettimeofday(&tv, NULL);
wtime = tv.tv_sec;
wtime += (double)tv.tv_usec / 1000000.0;
#endif
return wtime;
}
#define TIMER_START(x) (x) = gettime();
#define TIMER_STOP(y,x) (y) = (gettime() - (x));
static int init(void)
{
int rc;
if (ORTE_PROC_IS_DAEMON || ORTE_PROC_IS_HNP) {
/* daemons recv data server cmds */
if (ORTE_SUCCESS != (rc = orte_rmcast.recv_buffer_nb(ORTE_RMCAST_DATA_SERVER_CHANNEL,
ORTE_RMCAST_TAG_DATA,
ORTE_RMCAST_PERSISTENT,
recv_cmd, NULL))) {
ORTE_ERROR_LOG(rc);
}
OBJ_CONSTRUCT(&datastore, opal_pointer_array_t);
opal_pointer_array_init(&datastore, 16, INT_MAX, 16);
} else if (ORTE_PROC_IS_APP) {
/* get my multicast output group */
orte_rmcast.query_channel(&my_group_channel, NULL);
/* recv responses */
if (ORTE_SUCCESS != (rc = orte_rmcast.recv_buffer_nb(my_group_channel,
ORTE_RMCAST_TAG_CMD_ACK,
ORTE_RMCAST_PERSISTENT,
recv_ack, NULL))) {
ORTE_ERROR_LOG(rc);
}
}
return ORTE_SUCCESS;
}
static int finalize(void)
{
int i;
orte_db_data_t *dat;
/* cancel the callbacks */
if (ORTE_PROC_IS_DAEMON || ORTE_PROC_IS_HNP) {
orte_rmcast.cancel_recv(ORTE_RMCAST_DATA_SERVER_CHANNEL, ORTE_RMCAST_TAG_DATA);
for (i=0; i < datastore.size; i++) {
if (NULL != (dat = (orte_db_data_t*)opal_pointer_array_get_item(&datastore, i))) {
OBJ_RELEASE(dat);
}
}
OBJ_DESTRUCT(&datastore);
} else if (ORTE_PROC_IS_APP) {
orte_rmcast.cancel_recv(my_group_channel, ORTE_RMCAST_TAG_WILDCARD);
}
return ORTE_SUCCESS;
}
static int send_data(orte_db_cmd_t cmd, char *key, void *object, opal_data_type_t type)
{
opal_buffer_t *buf, dat;
orte_job_t *jdata;
orte_proc_t *proc;
orte_job_state_t *job_state;
orte_proc_state_t *proc_state;
int rc;
bool got_response;
opal_byte_object_t bo;
/* construct the buffer we will use for packing the data */
buf = OBJ_NEW(opal_buffer_t);
opal_dss.pack(buf, &cmd, 1, ORTE_DB_CMD_T); /* add cmd */
opal_dss.pack(buf, &my_group_channel, 1, ORTE_RMCAST_CHANNEL_T); /* tell the server my channel */
opal_dss.pack(buf, &key, 1, OPAL_STRING); /* pack the key */
if (NULL != object) {
OBJ_CONSTRUCT(&dat, opal_buffer_t);
/* pack the data */
switch (type) {
case ORTE_JOB:
jdata = (orte_job_t*)object;
opal_dss.pack(&dat, &jdata, 1, ORTE_JOB);
break;
case ORTE_JOB_STATE:
job_state = (orte_job_state_t*)object;
opal_dss.pack(&dat, job_state, 1, ORTE_JOB_STATE);
break;
case ORTE_PROC:
proc = (orte_proc_t*)object;
opal_dss.pack(&dat, &proc, 1, ORTE_PROC);
break;
case ORTE_PROC_STATE:
proc_state = (orte_proc_state_t*)object;
opal_dss.pack(&dat, proc_state, 1, ORTE_PROC_STATE);
break;
default:
orte_show_help("help-db-base.txt", "unrecognized-type", true, type);
rc = ORTE_ERR_BAD_PARAM;
goto cleanup;
break;
}
opal_dss.unload(&dat, (void**)&bo.bytes, &bo.size);
opal_dss.pack(buf, &bo.size, 1, OPAL_INT32);
opal_dss.pack(buf, &bo.bytes, bo.size, OPAL_UINT8);
OBJ_DESTRUCT(&dat);
free(bo.bytes);
}
got_response = false;
num_recvd = 0;
ack_reqd = true;
/* send the data to all the daemons */
if (ORTE_SUCCESS != (rc = orte_rmcast.send_buffer_nb(ORTE_RMCAST_DATA_SERVER_CHANNEL,
ORTE_RMCAST_TAG_DATA, buf,
callback_fn, NULL))) {
ORTE_ERROR_LOG(rc);
}
/* wait for all daemons to ack the request */
ORTE_PROGRESSED_WAIT(got_response, num_recvd, orte_process_info.num_daemons);
ack_reqd = false;
cleanup:
return rc;
}
static int store(char *key, void *object, opal_data_type_t type)
{
int rc;
double start;
double cpu_time_used;
TIMER_START(start);
if (ORTE_SUCCESS != (rc = send_data(ORTE_DB_STORE_CMD, key, object, type))) {
ORTE_ERROR_LOG(rc);
}
TIMER_STOP(cpu_time_used, start);
opal_output(0, "%s TOOK %g usecs TO STORE",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
cpu_time_used * 1000000.0);
return rc;
}
static int set_source(orte_process_name_t *name)
{
return ORTE_SUCCESS;
}
static int fetch(char *key, void *object, opal_data_type_t type)
{
int rc;
double cpu_time_used, start;
TIMER_START(start);
if (ORTE_SUCCESS != (rc = send_data(ORTE_DB_FETCH_CMD, key, NULL, OPAL_INT32))) {
ORTE_ERROR_LOG(rc);
}
TIMER_STOP(cpu_time_used, start);
opal_output(0, "%s TOOK %g usecs TO FETCH",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
cpu_time_used * 1000000.0);
return ORTE_SUCCESS;
}
static int update(char *key, void *object, opal_data_type_t type)
{
int rc;
double start;
double cpu_time_used;
TIMER_START(start);
if (ORTE_SUCCESS != (rc = send_data(ORTE_DB_UPDATE_CMD, key, object, type))) {
ORTE_ERROR_LOG(rc);
}
TIMER_STOP(cpu_time_used, start);
opal_output(0, "%s TOOK %g usecs TO UPDATE",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
cpu_time_used * 1000000.0);
return rc;
}
static int remove_data(char *key)
{
int rc;
double start;
double cpu_time_used;
TIMER_START(start);
if (ORTE_SUCCESS != (rc = send_data(ORTE_DB_REMOVE_CMD, key, NULL, OPAL_INT32))) {
ORTE_ERROR_LOG(rc);
}
TIMER_STOP(cpu_time_used, start);
opal_output(0, "%s TOOK %g usecs TO REMOVE",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
cpu_time_used * 1000000.0);
return rc;
}
static void callback_fn(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata)
{
OBJ_RELEASE(buf);
}
static void recv_ack(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata)
{
if (ack_reqd) {
num_recvd++;
}
}
static void recv_cmd(int status,
orte_rmcast_channel_t channel,
orte_rmcast_seq_t seq_num,
orte_rmcast_tag_t tag,
orte_process_name_t *sender,
opal_buffer_t *buf, void* cbdata)
{
orte_db_cmd_t cmd;
opal_buffer_t *ans;
int count, i;
int32_t rc;
char *key;
orte_db_data_t *dat;
orte_rmcast_channel_t ch;
char *ch_name;
OPAL_OUTPUT_VERBOSE((2, orte_db_base_output,
"%s db:daemon: cmd recvd from %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender)));
count=1;
opal_dss.unpack(buf, &cmd, &count, ORTE_DB_CMD_T);
count=1;
opal_dss.unpack(buf, &ch, &count, ORTE_RMCAST_CHANNEL_T);
count=1;
opal_dss.unpack(buf, &ch_name, &count, OPAL_STRING);
count=1;
opal_dss.unpack(buf, &key, &count, OPAL_STRING);
ans = OBJ_NEW(opal_buffer_t);
opal_dss.pack(ans, &cmd, 1, ORTE_DB_CMD_T);
switch (cmd) {
case ORTE_DB_STORE_CMD:
dat = OBJ_NEW(orte_db_data_t);
dat->name.jobid = sender->jobid;
dat->name.vpid = sender->vpid;
ORTE_EPOCH_SET(dat->name.epoch,sender->epoch);
dat->key = key;
count=1;
opal_dss.unpack(buf, &dat->size, &count, OPAL_INT32);
dat->bytes = (uint8_t*)malloc(dat->size);
opal_dss.unpack(buf, dat->bytes, &dat->size, OPAL_UINT8);
opal_pointer_array_add(&datastore, dat);
OPAL_OUTPUT_VERBOSE((2, orte_db_base_output,
"%s db:daemon: data from %s stored: key %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender), key));
rc = ORTE_SUCCESS;
opal_dss.pack(ans, &rc, 1, OPAL_INT32);
break;
case ORTE_DB_FETCH_CMD:
/* find the key */
for (i=0; i < datastore.size; i++) {
if (NULL == (dat = (orte_db_data_t*)opal_pointer_array_get_item(&datastore, i))) {
continue;
}
if (0 != strcmp(key, dat->key)) {
continue;
}
/* found the data - return it */
rc = ORTE_SUCCESS;
opal_dss.pack(ans, &rc, 1, OPAL_INT32);
opal_dss.pack(ans, &dat->size, 1, OPAL_INT32);
opal_dss.pack(ans, dat->bytes, dat->size, OPAL_UINT8);
OPAL_OUTPUT_VERBOSE((2, orte_db_base_output,
"%s db:daemon: data fetched for %s: key %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender), key));
break;
}
rc = ORTE_ERR_NOT_FOUND;
opal_dss.pack(ans, &rc, 1, OPAL_INT32);
OPAL_OUTPUT_VERBOSE((2, orte_db_base_output,
"%s db:daemon: data fetch request from %s not found: key %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_NAME_PRINT(sender), key));
break;
default:
rc = ORTE_ERR_NOT_FOUND;
break;
}
/* ensure the return channel is open */
orte_rmcast.open_channel(ch, ch_name, NULL, -1, NULL, ORTE_RMCAST_XMIT);
orte_rmcast.send_buffer_nb(ch, ORTE_RMCAST_TAG_CMD_ACK, ans, callback_fn, NULL);
}

31
orte/mca/db/daemon/db_daemon.h
View File

@ -1,31 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#ifndef ORTE_DB_DAEMON_H
#define ORTE_DB_DAEMON_H
#include "orte/mca/db/db.h"
BEGIN_C_DECLS
/*
* Module open / close
*/
int orte_db_daemon_component_open(void);
int orte_db_daemon_component_close(void);
int orte_db_daemon_component_query(mca_base_module_t **module, int *priority);
ORTE_MODULE_DECLSPEC extern orte_db_base_component_t mca_db_daemon_component;
ORTE_DECLSPEC extern orte_db_base_module_t orte_db_daemon_module;
extern char *orte_db_daemon_directory;
END_C_DECLS
#endif /* ORTE_DB_DAEMON_H */

91
orte/mca/db/daemon/db_daemon_component.c
View File

@ -1,91 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*
* These symbols are in a file by themselves to provide nice linker
* semantics. Since linkers generally pull in symbols by object
* files, keeping these symbols as the only symbols in this file
* prevents utility programs such as "ompi_info" from having to import
* entire components just to query their version and parameters.
*/
#include "orte_config.h"
#include "orte/constants.h"
#ifdef HAVE_STRINGS_H
#include <strings.h>
#endif
#include "opal/mca/base/base.h"
#include "opal/mca/base/mca_base_param.h"
#include "orte/util/proc_info.h"
#include "orte/mca/db/db.h"
#include "orte/mca/db/base/base.h"
#include "db_daemon.h"
extern orte_db_base_module_t orte_db_daemon_module;
char *orte_db_daemon_directory;
/*
* Instantiate the public struct with all of our public information
* and pointers to our public functions in it
*/
orte_db_base_component_t mca_db_daemon_component = {
{
ORTE_DB_BASE_VERSION_1_0_0,
/* Component name and version */
"daemon",
ORTE_MAJOR_VERSION,
ORTE_MINOR_VERSION,
ORTE_RELEASE_VERSION,
/* Component open and close functions */
orte_db_daemon_component_open,
orte_db_daemon_component_close,
orte_db_daemon_component_query
},
{
/* The component is checkpoint ready */
MCA_BASE_METADATA_PARAM_CHECKPOINT
}
};
int
orte_db_daemon_component_open(void)
{
return ORTE_SUCCESS;
}
/* only select us if specifically requested */
int orte_db_daemon_component_query(mca_base_module_t **module, int *priority)
{
char *req;
if (NULL != (req = getenv("OMPI_MCA_db"))) {
if (0 == strcasecmp(req, "daemon")) {
*priority = 100;
*module = (mca_base_module_t*)&orte_db_daemon_module;
return ORTE_SUCCESS;
}
}
*priority = 0;
*module = NULL;
return ORTE_ERROR;
}
int
orte_db_daemon_component_close(void)
{
return ORTE_SUCCESS;
}

110
orte/mca/db/db.h
View File

@ -1,110 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
/** @file:
*
* The OpenRTE State Save/Recovery Service
*
*/
#ifndef ORTE_DB_H
#define ORTE_DB_H
#include "orte_config.h"
#include "orte/types.h"
#include "opal/mca/mca.h"
#include "opal/dss/dss_types.h"
BEGIN_C_DECLS
/*
* API functions
*/
typedef uint8_t orte_db_cmd_t;
#define ORTE_DB_CMD_T OPAL_UINT8
#define ORTE_DB_STORE_CMD 0x01
#define ORTE_DB_FETCH_CMD 0X02
#define ORTE_DB_UPDATE_CMD 0x03
#define ORTE_DB_REMOVE_CMD 0x04
/*
* Initialize the module
*/
typedef int (*orte_db_base_module_init_fn_t)(void);
/*
* Finalize the module
*/
typedef int (*orte_db_base_module_finalize_fn_t)(void);
/*
* Save the db of the provided object
*/
typedef int (*orte_db_base_module_store_fn_t)(char *key, void *object, opal_data_type_t type);
/*
* Set the source for recovering db info
*/
typedef int (*orte_db_base_module_set_source_fn_t)(orte_process_name_t *name);
/*
* Retrieve data
*/
typedef int (*orte_db_base_module_fetch_fn_t)(char *key, void *object, opal_data_type_t type);
/*
* Update data
*/
typedef int (*orte_db_base_module_update_fn_t)(char *key, void *object, opal_data_type_t type);
/*
* Delete data
*/
typedef int (*orte_db_base_module_remove_fn_t)(char *key);
/*
* the standard module data structure
*/
struct orte_db_base_module_1_0_0_t {
orte_db_base_module_init_fn_t init;
orte_db_base_module_finalize_fn_t finalize;
orte_db_base_module_store_fn_t store;
orte_db_base_module_set_source_fn_t set_source;
orte_db_base_module_fetch_fn_t fetch;
orte_db_base_module_update_fn_t update;
orte_db_base_module_remove_fn_t remove;
};
typedef struct orte_db_base_module_1_0_0_t orte_db_base_module_1_0_0_t;
typedef struct orte_db_base_module_1_0_0_t orte_db_base_module_t;
/*
* the standard component data structure
*/
struct orte_db_base_component_1_0_0_t {
mca_base_component_t base_version;
mca_base_component_data_t base_data;
};
typedef struct orte_db_base_component_1_0_0_t orte_db_base_component_1_0_0_t;
typedef struct orte_db_base_component_1_0_0_t orte_db_base_component_t;
/*
* Macro for use in components that are of type db
*/
#define ORTE_DB_BASE_VERSION_1_0_0 \
MCA_BASE_VERSION_2_0_0, \
"db", 1, 0, 0
/* Global structure for accessing DB functions */
ORTE_DECLSPEC extern orte_db_base_module_t orte_db; /* holds selected module's function pointers */
END_C_DECLS
#endif

38
orte/mca/db/dbase/Makefile.am
View File

@ -1,38 +0,0 @@
#
# Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
# $COPYRIGHT$
#
# Additional copyrights may follow
#
# $HEADER$
#
sources = \
db_dbase.h \
db_dbase_component.c \
db_dbase.c
# Make the output library in this directory, and name it either
# mca_<type>_<name>.la (for DSO builds) or libmca_<type>_<name>.la
# (for static builds).
if MCA_BUILD_orte_db_dbase_DSO
component_noinst =
component_install = mca_db_dbase.la
else
component_noinst = libmca_db_dbase.la
component_install =
endif
mcacomponentdir = $(pkglibdir)
mcacomponent_LTLIBRARIES = $(component_install)
mca_db_dbase_la_CPPFLAGS = $(db_dbase_CPPFLAGS)
mca_db_dbase_la_SOURCES = $(sources)
mca_db_dbase_la_LDFLAGS = -module -avoid-version $(db_dbase_LDFLAGS)
mca_db_dbase_la_LIBADD = $(db_dbase_LIBS)
noinst_LTLIBRARIES = $(component_noinst)
libmca_db_dbase_la_CPPFLAGS = $(db_dbase_CPPFLAGS)
libmca_db_dbase_la_SOURCES =$(sources)
libmca_db_dbase_la_LDFLAGS = -module -avoid-version $(db_dbase_LDFLAGS)
libmca_db_dbase_la_LIBADD = $(db_dbase_LIBS)

30
orte/mca/db/dbase/configure.m4
View File

@ -1,30 +0,0 @@
dnl -*- shell-script -*-
dnl
dnl Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
dnl $COPYRIGHT$
dnl
dnl Additional copyrights may follow
dnl
dnl $HEADER$
dnl
# MCA_db_dbase_CONFIG([action-if-found], [action-if-not-found])
# -----------------------------------------------------------
AC_DEFUN([MCA_orte_db_dbase_CONFIG], [
AC_CONFIG_FILES([orte/mca/db/dbase/Makefile])
# only build if db.h and its corresponding library are found
OMPI_CHECK_PACKAGE([db_db],
[db.h],
[db],
[dbopen],
[],
[],
[],
[$1],
[$2])])
AC_SUBST(db_dbase_CPPFLAGS)
AC_SUBST(db_dbase_LDFLAGS)
AC_SUBST(db_dbase_LIBS)
])dnl

224
orte/mca/db/dbase/db_dbase.c
View File

@ -1,224 +0,0 @@
/*
* Copyright (c) 2010 Cisco Systems, Inc. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*
*/
#include "orte_config.h"
#include "orte/constants.h"
#include <string.h>
#include <sys/types.h>
#ifdef HAVE_LIMITS_H
#include <limits.h>
#endif
#include <stdio.h>
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#include <db.h>
#include "opal/dss/dss_types.h"
#include "opal/util/os_dirpath.h"
#include "opal/util/os_path.h"
#include "opal/util/output.h"
#include "opal/util/malloc.h"
#include "opal/util/basename.h"
#include "opal/mca/pstat/base/base.h"
#include "opal/mca/paffinity/base/base.h"
#include "opal/mca/sysinfo/base/base.h"
#include "orte/util/show_help.h"
#include "orte/mca/errmgr/base/base.h"
#include "orte/runtime/orte_globals.h"
#include "db_dbase.h"
static int init(void);