ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
openmpi/orte/runtime/orte_info_support.c

457 строки
14 KiB
C
Исходник Обычный вид История

Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
/*
* Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana
* University Research and Technology
* Corporation. All rights reserved.
* Copyright (c) 2004-2011 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* Copyright (c) 2004-2005 The Regents of the University of California.
* All rights reserved.
* Copyright (c) 2006-2012 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2010-2012 Los Alamos National Security, LLC.
* All rights reserved.
* Copyright (c) 2011-2012 University of Houston. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "orte_config.h"
#include "orte/constants.h"
#include "orte/types.h"
#include "opal/class/opal_pointer_array.h"
#include "opal/runtime/opal_info_support.h"
#include "orte/mca/db/base/base.h"
Add the dfs to ompi_info This commit was SVN r27613.
2012-11-15 15:54:07 +00:00
#include "orte/mca/dfs/base/base.h"
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
#include "orte/mca/errmgr/base/base.h"
#include "orte/mca/ess/base/base.h"
#include "orte/mca/grpcomm/base/base.h"
#include "orte/mca/iof/base/base.h"
#include "orte/mca/odls/base/base.h"
#include "orte/mca/oob/base/base.h"
#include "orte/mca/plm/base/base.h"
#include "orte/mca/ras/base/ras_private.h"
#include "orte/mca/rmaps/base/base.h"
#include "orte/mca/rml/base/base.h"
#include "orte/mca/routed/base/base.h"
#if OPAL_ENABLE_FT_CR == 1
#include "orte/mca/snapc/base/base.h"
#include "orte/mca/sstore/base/base.h"
#endif
#if ORTE_ENABLE_SENSORS
#include "orte/mca/sensor/base/base.h"
#endif
#include "orte/mca/filem/base/base.h"
#include "orte/mca/state/base/base.h"
#include "orte/util/proc_info.h"
#include "orte/runtime/runtime.h"
#include "orte/runtime/orte_info_support.h"
const char *orte_info_type_orte = "orte";
void orte_info_register_types(opal_pointer_array_t *mca_types)
{
/* frameworks */
opal_pointer_array_add(mca_types, "db");
Add the dfs to ompi_info This commit was SVN r27613.
2012-11-15 15:54:07 +00:00
opal_pointer_array_add(mca_types, "dfs");
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
opal_pointer_array_add(mca_types, "errmgr");
opal_pointer_array_add(mca_types, "ess");
opal_pointer_array_add(mca_types, "filem");
opal_pointer_array_add(mca_types, "grpcomm");
opal_pointer_array_add(mca_types, "iof");
opal_pointer_array_add(mca_types, "odls");
opal_pointer_array_add(mca_types, "oob");
opal_pointer_array_add(mca_types, "orte");
opal_pointer_array_add(mca_types, "plm");
opal_pointer_array_add(mca_types, "ras");
opal_pointer_array_add(mca_types, "rmaps");
opal_pointer_array_add(mca_types, "rml");
opal_pointer_array_add(mca_types, "routed");
#if ORTE_ENABLE_SENSORS
opal_pointer_array_add(mca_types, "sensor");
#endif
#if OPAL_ENABLE_FT_CR == 1
opal_pointer_array_add(mca_types, "snapc");
opal_pointer_array_add(mca_types, "sstore");
#endif
opal_pointer_array_add(mca_types, "state");
}
int orte_info_register_components(opal_pointer_array_t *mca_types,
opal_pointer_array_t *component_map)
{
opal_info_component_map_t *map;
char *env, *str;
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
int i, rc;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
char *target, *save, *type;
char **env_save=NULL;
/* Clear out the environment. Use strdup() to orphan the resulting
* strings because items are placed in the environment by reference,
* not by value.
*/
for (i = 0; i < mca_types->size; ++i) {
if (NULL == (type = (char*)opal_pointer_array_get_item(mca_types, i))) {
continue;
}
asprintf(&env, "OMPI_MCA_%s", type);
if (NULL != (save = getenv(env))) {
/* save this param so it can later be restored */
asprintf(&str, "%s=%s", env, save);
opal_argv_append_nosize(&env_save, str);
free(str);
/* can't manipulate it directly, so make a copy first */
asprintf(&target, "%s=", env);
putenv(target);
}
free(env);
}
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
/* Set orte_process_info.proc_type to HNP to force all frameworks to
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
* open components
*/
orte_process_info.proc_type = ORTE_PROC_HNP;
Set the orte_event_base to the opal_event_base in ompi_info - we aren't doing anything with progress threads anyway This commit was SVN r27488.
2012-10-25 22:36:08 +00:00
/* set the event base to be the opal event base as we
* aren't attempting to do anything with progress threads here
*/
orte_event_base = opal_event_base;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
/* Register the ORTE layer's MCA parameters */
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_register_params()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "orte_register_params";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_db_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "db_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("db");
map->components = &orte_db_base.available_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "db";
goto breakout;
}
Add the dfs to ompi_info This commit was SVN r27613.
2012-11-15 15:54:07 +00:00
if (ORTE_SUCCESS != (rc = orte_dfs_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "dfs_base_open";
goto error;
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("dfs");
map->components = &orte_dfs_base.components_available;
opal_pointer_array_add(component_map, map);
if (ORTE_ERR_BAD_PARAM == rc) {
str = "dfs";
goto breakout;
}
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_errmgr_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "errmgr_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("errmgr");
map->components = &orte_errmgr_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "errmgr";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_ess_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "ess_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("ess");
map->components = &orte_ess_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "ess";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_filem_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "filem_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("filem");
map->components = &orte_filem_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "filem";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_grpcomm_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "grpcomm_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("grpcomm");
map->components = &orte_grpcomm_base.components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "grpcomm";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_iof_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "iof_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("iof");
map->components = &orte_iof_base.iof_components_opened;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "iof";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_odls_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "odls_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("odls");
map->components = &orte_odls_base.available_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "odls";
goto breakout;
}
if (ORTE_SUCCESS != (rc = mca_oob_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "oob_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("oob");
map->components = &mca_oob_base_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "oob";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_plm_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "plm_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("plm");
map->components = &orte_plm_base.available_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "plm";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_ras_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "ras_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("ras");
map->components = &orte_ras_base.ras_opened;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "ras";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_rmaps_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "rmaps_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("rmaps");
map->components = &orte_rmaps_base.available_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "rmaps";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_routed_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "routed_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("routed");
map->components = &orte_routed_base_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "routed";
goto breakout;
}
if (ORTE_SUCCESS != (rc = orte_rml_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "rml_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("rml");
map->components = &orte_rml_base_components;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "rml";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
#if ORTE_ENABLE_SENSORS
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_sensor_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "sensor_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("sensor");
map->components = &mca_sensor_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "sensor";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
#endif
#if OPAL_ENABLE_FT_CR == 1
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_snapc_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "snapc_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("snapc");
map->components = &orte_snapc_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "snapc";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_sstore_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "sstore_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("sstore");
map->components = &orte_sstore_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "sstore";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
#endif
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_SUCCESS != (rc = orte_state_base_open()) &&
ORTE_ERR_BAD_PARAM != rc) {
str = "state_base_open";
goto error;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
map = OBJ_NEW(opal_info_component_map_t);
map->type = strdup("state");
map->components = &orte_state_base_components_available;
opal_pointer_array_add(component_map, map);
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
str = "state";
goto breakout;
}
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
breakout:
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
/* Restore the environment to what it was before we started so that
* if users setenv OMPI_MCA_<framework name> to some value, they'll
* see that value when it is shown via --param output.
*/
if (NULL != env_save) {
for (i = 0; i < opal_argv_count(env_save); ++i) {
putenv(env_save[i]);
}
}
Refs trac:3275. We ran into a case where the OMPI SVN trunk grew a new acceptable MCA parameter value, but this new value was not accepted on the v1.6 branch (hwloc_base_mem_bind_failure_action -- on the trunk it accepts the value "silent", but on the older v1.6 branch, it doesn't). If you set "hwloc_base_mem_bind_failure_action=silent" in the default MCA params file and then accidentally ran with the v1.6 branch, every OMPI executable (including ompi_info) just failed because hwloc_base_open() would say "hey, 'silent' is not a valid value for hwloc_base_mem_bind_failure_action!". Kaboom. The only problem is that it didn't give you any indication of where this value was being set. Quite maddening, from a user perspective. So we changed the ompi_info handles this case. If any framework open function return OMPI_ERR_BAD_PARAM (either because its base MCA params got a bad value or because one of its component register/open functions return OMPI_ERR_BAD_PARAM), ompi_info will stop, print out a warning that it received and error, and then dump out the parameters that it has received so far in the framework that had a problem. At a minimum, this will show the user the MCA param that had an error (it's usually the last one), and ''where it was set from'' (so that they can go fix it). We updated ompi_info to check for O???_ERR_BAD_PARAM from each from the framework opens. Also updated the doxygen docs in mca.h for this O???_BAD_PARAM behavior. And we noticed that mca.h had MCA_SUCCESS and MCA_ERR_??? codes. Why? I think we used them in exactly one place in the code base (mca_base_components_open.c). So we deleted those and just used the normal OPAL_* codes instead. While we were doing this, we also cleaned up a little memory management during ompi_info/orte-info/opal-info finalization. Valgrind still reports a truckload of memory still in use at ompi_info termination, but they mostly look to be components not freeing memory/resources properly (and outside the scope of this fix). This commit was SVN r27306. The following Trac tickets were found above: Ticket 3275 --> https://svn.open-mpi.org/trac/ompi/ticket/3275
2012-09-11 20:47:24 +00:00
if (ORTE_ERR_BAD_PARAM == rc) {
fprintf(stderr, "\nA \"bad parameter\" error was encountered when opening the ORTE %s framework\n", str);
fprintf(stderr, "The output received from that framework includes the following parameters:\n\n");
}
return rc;
error:
fprintf(stderr, "orte_info_register: %s failed\n", str);
return ORTE_ERROR;
Refactor ompi_info, splitting it into parts according to the layer involved. Thus, we call down to the opal layer to get those frameworks and components, and down to the orte layer to get those. Still some abstraction breaks, but they mostly involve renaming of OMPI_foo labels that have been around since before we split the build system by layer. This commit was SVN r26695.
2012-06-28 18:23:34 +00:00
}
void orte_info_close_components(void)
{
(void) orte_db_base_close();
(void) orte_errmgr_base_close();
(void) orte_ess_base_close();
(void) orte_filem_base_close();
(void) orte_grpcomm_base_close();
(void) orte_iof_base_close();
(void) orte_odls_base_close();
(void) mca_oob_base_close();
(void) orte_plm_base_close();
(void) orte_ras_base_close();
(void) orte_rmaps_base_close();
(void) orte_rml_base_close();
(void) orte_routed_base_close();
#if OPAL_ENABLE_FT_CR == 1
(void) orte_snapc_base_close();
(void) orte_sstore_base_close();
#endif
(void) orte_state_base_close();
}
void orte_info_show_orte_version(const char *scope)
{
char *tmp, *tmp2;
asprintf(&tmp, "%s:version:full", orte_info_type_orte);
tmp2 = opal_info_make_version_str(scope,
ORTE_MAJOR_VERSION, ORTE_MINOR_VERSION,
ORTE_RELEASE_VERSION,
ORTE_GREEK_VERSION,
ORTE_WANT_REPO_REV, ORTE_REPO_REV);
opal_info_out("Open RTE", tmp, tmp2);
free(tmp);
free(tmp2);
asprintf(&tmp, "%s:version:repo", orte_info_type_orte);
opal_info_out("Open RTE repo revision", tmp, ORTE_REPO_REV);
free(tmp);
asprintf(&tmp, "%s:version:release_date", orte_info_type_orte);
opal_info_out("Open RTE release date", tmp, ORTE_RELEASE_DATE);
free(tmp);
}
Копировать постоянную ссылку