1
1
openmpi/orte/mca/oob/tcp/oob_tcp.c

1165 строки
40 KiB
C
Исходник Обычный вид История

/*
* Copyright (c) 2004-2005 The Trustees of Indiana University.
* All rights reserved.
* Copyright (c) 2004-2005 The Trustees of the University of Tennessee.
* 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.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*/
#include "ompi_config.h"
#include "include/orte_types.h"
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#include <fcntl.h>
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif
#ifdef HAVE_ARPA_INET_H
#include <arpa/inet.h>
#endif
#include "include/ompi_socket_errno.h"
#include "opal/util/output.h"
#include "opal/util/if.h"
#include "orte/class/orte_proc_table.h"
#include "mca/oob/tcp/oob_tcp.h"
#include "mca/errmgr/errmgr.h"
#include "mca/ns/ns.h"
#include "mca/gpr/gpr.h"
#define IMPORTANT_WINDOWS_COMMENT() \
/* In windows, many of the socket functions return an EWOULDBLOCK instead of \
things like EAGAIN, EINPROGRESS, etc. It has been verified that this will \
not conflict with other error codes that are returned by these functions \
under UNIX/Linux environments */
/*
* Data structure for accepting connections.
*/
struct mca_oob_tcp_event_t {
opal_list_item_t item;
opal_event_t event;
};
typedef struct mca_oob_tcp_event_t mca_oob_tcp_event_t;
static void mca_oob_tcp_event_construct(mca_oob_tcp_event_t* event)
{
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
opal_list_append(&mca_oob_tcp_component.tcp_events, &event->item);
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
}
static void mca_oob_tcp_event_destruct(mca_oob_tcp_event_t* event)
{
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
opal_list_remove_item(&mca_oob_tcp_component.tcp_events, &event->item);
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
}
OBJ_CLASS_INSTANCE(
mca_oob_tcp_event_t,
opal_list_item_t,
mca_oob_tcp_event_construct,
mca_oob_tcp_event_destruct);
/*
* Local utility functions
*/
static int mca_oob_tcp_create_listen(void);
static void mca_oob_tcp_recv_handler(int sd, short flags, void* user);
static void mca_oob_tcp_accept(void);
struct mca_oob_tcp_subscription_t {
opal_list_item_t item;
orte_jobid_t jobid;
orte_gpr_subscription_id_t subid;
};
typedef struct mca_oob_tcp_subscription_t mca_oob_tcp_subscription_t;
OBJ_CLASS_INSTANCE(
mca_oob_tcp_subscription_t,
opal_list_item_t,
NULL,
NULL);
/*
* Struct of function pointers and all that to let us be initialized
*/
- massive change for module<-->component name fixes throughout the code base. - many (most) mca type names have "component" or "module" in them, as relevant, just to further distinguish the difference between component data/actions and module data/actions. All developers are encouraged to perpetuate this convention when you create types that are specific to a framework, component, or module - did very little to entire framework (just the basics to make it compile) because it's just about to be almost entirely replaced - ditto for io / romio - did not work on elan or ib components; have to commit and then convert those on a different machine with the right libraries and headers - renamed a bunch of *_module.c files to *_component.c and *module*c to *component*c (a few still remain, e.g., ptl/ib, ptl/elan, etc.) - modified autogen/configure/build process to match new filenames (e.g., output static-components.h instead of static-modules.h) - removed DOS-style cr/lf stuff in ns/ns.h - added newline to end of file src/util/numtostr.h - removed some redundant error checking in the top-level topo functions - added a few {} here and there where people "forgot" to put them in for 1 line blocks ;-) - removed a bunch of MPI_* types from mca header files (replaced with corresponding ompi_* types) - all the ptl components had version numbers in their structs; removed - converted a few more elements in the MCA base to use the OBJ interface -- removed some old manual reference counting kruft This commit was SVN r1830.
2004-08-02 04:24:22 +04:00
mca_oob_tcp_component_t mca_oob_tcp_component = {
{
{
MCA_OOB_BASE_VERSION_1_0_0,
"tcp", /* MCA module name */
1, /* MCA component major version */
0, /* MCA component minor version */
0, /* MCA component release version */
mca_oob_tcp_component_open, /* component open */
mca_oob_tcp_component_close /* component close */
},
{
false /* checkpoint / restart */
},
mca_oob_tcp_component_init
- massive change for module<-->component name fixes throughout the code base. - many (most) mca type names have "component" or "module" in them, as relevant, just to further distinguish the difference between component data/actions and module data/actions. All developers are encouraged to perpetuate this convention when you create types that are specific to a framework, component, or module - did very little to entire framework (just the basics to make it compile) because it's just about to be almost entirely replaced - ditto for io / romio - did not work on elan or ib components; have to commit and then convert those on a different machine with the right libraries and headers - renamed a bunch of *_module.c files to *_component.c and *module*c to *component*c (a few still remain, e.g., ptl/ib, ptl/elan, etc.) - modified autogen/configure/build process to match new filenames (e.g., output static-components.h instead of static-modules.h) - removed DOS-style cr/lf stuff in ns/ns.h - added newline to end of file src/util/numtostr.h - removed some redundant error checking in the top-level topo functions - added a few {} here and there where people "forgot" to put them in for 1 line blocks ;-) - removed a bunch of MPI_* types from mca header files (replaced with corresponding ompi_* types) - all the ptl components had version numbers in their structs; removed - converted a few more elements in the MCA base to use the OBJ interface -- removed some old manual reference counting kruft This commit was SVN r1830.
2004-08-02 04:24:22 +04:00
}
};
static mca_oob_t mca_oob_tcp = {
mca_oob_tcp_get_addr,
mca_oob_tcp_set_addr,
mca_oob_tcp_ping,
mca_oob_tcp_send,
mca_oob_tcp_recv,
mca_oob_tcp_send_nb,
mca_oob_tcp_recv_nb,
mca_oob_tcp_recv_cancel,
mca_oob_tcp_init,
mca_oob_tcp_fini,
mca_oob_xcast
};
/*
* Utility function to register/lookup module parameters.
*/
static inline int mca_oob_tcp_param_register_int(
const char* param_name,
int default_value)
{
int id = mca_base_param_register_int("oob","tcp",param_name,NULL,default_value);
int param_value = default_value;
mca_base_param_lookup_int(id,&param_value);
return param_value;
}
static inline char* mca_oob_tcp_param_register_str(
const char* param_name,
const char* default_value)
{
int id = mca_base_param_register_string("oob","tcp",param_name,NULL,default_value);
char* param_value = NULL;
mca_base_param_lookup_string(id,&param_value);
return param_value;
}
/*
* Initialize global variables used w/in this module.
*/
int mca_oob_tcp_component_open(void)
{
#ifdef WIN32
WSADATA win_sock_data;
if (WSAStartup(MAKEWORD(2,2), &win_sock_data) != 0) {
opal_output (0, "mca_oob_tcp_component_init: failed to initialise windows sockets: error %d\n", WSAGetLastError());
return OMPI_ERROR;
}
#endif
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_subscriptions, opal_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_peer_list, opal_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_peers, opal_hash_table_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_peer_names, opal_hash_table_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_peer_free, opal_free_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_msgs, opal_free_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_lock, opal_mutex_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_events, opal_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_msg_post, opal_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_msg_recv, opal_list_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_match_lock, opal_mutex_t);
OBJ_CONSTRUCT(&mca_oob_tcp_component.tcp_match_cond, opal_condition_t);
/* register oob module parameters */
mca_oob_tcp_component.tcp_peer_limit =
mca_oob_tcp_param_register_int("peer_limit", -1);
mca_oob_tcp_component.tcp_peer_retries =
mca_oob_tcp_param_register_int("peer_retries", 60);
mca_oob_tcp_component.tcp_debug =
mca_oob_tcp_param_register_int("debug", 0);
mca_oob_tcp_component.tcp_include =
mca_oob_tcp_param_register_str("include", NULL);
mca_oob_tcp_component.tcp_exclude =
mca_oob_tcp_param_register_str("exclude", NULL);
/* initialize state */
mca_oob_tcp_component.tcp_listen_sd = -1;
mca_oob_tcp_component.tcp_match_count = 0;
return OMPI_SUCCESS;
}
/*
* Cleanup of global variables used by this module.
*/
int mca_oob_tcp_component_close(void)
{
#ifdef WIN32
WSACleanup();
#endif
/* cleanup resources */
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_peer_list);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_peers);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_peer_free);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_events);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_subscriptions);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_msgs);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_lock);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_msg_post);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_msg_recv);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_match_lock);
OBJ_DESTRUCT(&mca_oob_tcp_component.tcp_match_cond);
return OMPI_SUCCESS;
}
/*
* Called by mca_oob_tcp_recv_handler() when the TCP listen
* socket has pending connection requests. Accept incoming
* requests and queue for completion of the connection handshake.
*/
static void mca_oob_tcp_accept(void)
{
while(true) {
ompi_socklen_t addrlen = sizeof(struct sockaddr_in);
struct sockaddr_in addr;
mca_oob_tcp_event_t* event;
int sd;
sd = accept(mca_oob_tcp_component.tcp_listen_sd, (struct sockaddr*)&addr, &addrlen);
if(sd < 0) {
IMPORTANT_WINDOWS_COMMENT();
if(ompi_socket_errno == EINTR)
continue;
if(ompi_socket_errno != EAGAIN || ompi_socket_errno != EWOULDBLOCK)
opal_output(0, "mca_oob_tcp_accept: accept() failed with errno %d.", ompi_socket_errno);
return;
}
/* log the accept */
if(mca_oob_tcp_component.tcp_debug) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_accept: %s:%d\n",
ORTE_NAME_ARGS(orte_process_info.my_name),
inet_ntoa(addr.sin_addr),
addr.sin_port);
}
/* wait for receipt of peers process identifier to complete this connection */
event = OBJ_NEW(mca_oob_tcp_event_t);
opal_event_set(&event->event, sd, OPAL_EV_READ, mca_oob_tcp_recv_handler, event);
opal_event_add(&event->event, 0);
}
}
/*
* Create a listen socket and bind to all interfaces
*/
static int mca_oob_tcp_create_listen(void)
{
int flags;
struct sockaddr_in inaddr;
ompi_socklen_t addrlen;
/* create a listen socket for incoming connections */
mca_oob_tcp_component.tcp_listen_sd = socket(AF_INET, SOCK_STREAM, 0);
if(mca_oob_tcp_component.tcp_listen_sd < 0) {
opal_output(0,"mca_oob_tcp_component_init: socket() failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
}
memset(&inaddr, 0, sizeof(inaddr));
inaddr.sin_family = AF_INET;
inaddr.sin_addr.s_addr = INADDR_ANY;
inaddr.sin_port = 0;
if(bind(mca_oob_tcp_component.tcp_listen_sd, (struct sockaddr*)&inaddr, sizeof(inaddr)) < 0) {
opal_output(0,"mca_oob_tcp_create_listen: bind() failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
}
/* resolve system assigned port */
addrlen = sizeof(struct sockaddr_in);
if(getsockname(mca_oob_tcp_component.tcp_listen_sd, (struct sockaddr*)&inaddr, &addrlen) < 0) {
opal_output(0, "mca_oob_tcp_create_listen: getsockname() failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
}
mca_oob_tcp_component.tcp_listen_port = inaddr.sin_port;
/* setup listen backlog to maximum allowed by kernel */
if(listen(mca_oob_tcp_component.tcp_listen_sd, SOMAXCONN) < 0) {
opal_output(0, "mca_oob_tcp_component_init: listen() failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
}
/* set socket up to be non-blocking, otherwise accept could block */
if((flags = fcntl(mca_oob_tcp_component.tcp_listen_sd, F_GETFL, 0)) < 0) {
opal_output(0, "mca_oob_tcp_component_init: fcntl(F_GETFL) failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
} else {
flags |= O_NONBLOCK;
if(fcntl(mca_oob_tcp_component.tcp_listen_sd, F_SETFL, flags) < 0) {
opal_output(0, "mca_oob_tcp_component_init: fcntl(F_SETFL) failed with errno=%d", ompi_socket_errno);
return OMPI_ERROR;
}
}
/* register listen port */
opal_event_set(
&mca_oob_tcp_component.tcp_recv_event,
mca_oob_tcp_component.tcp_listen_sd,
OPAL_EV_READ|OPAL_EV_PERSIST,
mca_oob_tcp_recv_handler,
0);
opal_event_add(&mca_oob_tcp_component.tcp_recv_event, 0);
return OMPI_SUCCESS;
}
/*
* Handle probe
*/
static void mca_oob_tcp_recv_probe(int sd, mca_oob_tcp_hdr_t* hdr)
{
unsigned char* ptr = (unsigned char*)hdr;
size_t cnt = 0;
hdr->msg_type = MCA_OOB_TCP_PROBE;
hdr->msg_dst = hdr->msg_src;
hdr->msg_src = *orte_process_info.my_name;
MCA_OOB_TCP_HDR_HTON(hdr);
while(cnt < sizeof(mca_oob_tcp_hdr_t)) {
int retval = send(sd, (char *)ptr+cnt, sizeof(mca_oob_tcp_hdr_t)-cnt, 0);
if(retval < 0) {
IMPORTANT_WINDOWS_COMMENT();
if(ompi_socket_errno != EINTR && ompi_socket_errno != EAGAIN && ompi_socket_errno != EWOULDBLOCK) {
opal_output(0, "[%lu,%lu,%lu]-[%lu,%lu,%lu] mca_oob_tcp_peer_recv_probe: send() failed with errno=%d\n",
ORTE_NAME_ARGS(orte_process_info.my_name),
ORTE_NAME_ARGS(&(hdr->msg_src)),
ompi_socket_errno);
close(sd);
return;
}
continue;
}
cnt += retval;
}
close(sd);
}
/*
* Handle connection request
*/
static void mca_oob_tcp_recv_connect(int sd, mca_oob_tcp_hdr_t* hdr)
{
mca_oob_tcp_peer_t* peer;
int flags;
int cmpval;
/* now set socket up to be non-blocking */
if((flags = fcntl(sd, F_GETFL, 0)) < 0) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: fcntl(F_GETFL) failed with errno=%d",
ORTE_NAME_ARGS(orte_process_info.my_name), ompi_socket_errno);
} else {
flags |= O_NONBLOCK;
if(fcntl(sd, F_SETFL, flags) < 0) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: fcntl(F_SETFL) failed with errno=%d",
ORTE_NAME_ARGS(orte_process_info.my_name), ompi_socket_errno);
}
}
/* check for wildcard name - if this is true - we allocate a name from the name server
* and return to the peer
*/
cmpval = orte_ns.compare(ORTE_NS_CMP_ALL, &hdr->msg_src, MCA_OOB_NAME_ANY);
if (cmpval == 0) {
if (ORTE_SUCCESS != orte_ns.create_jobid(&hdr->msg_src.jobid)) {
return;
}
if (ORTE_SUCCESS != orte_ns.reserve_range(hdr->msg_src.jobid, 1, &hdr->msg_src.vpid)) {
return;
}
if (ORTE_SUCCESS != orte_ns.assign_cellid_to_process(&hdr->msg_src)) {
return;
}
}
/* lookup the corresponding process */
peer = mca_oob_tcp_peer_lookup(&hdr->msg_src);
if(NULL == peer) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: unable to locate peer",
ORTE_NAME_ARGS(orte_process_info.my_name));
close(sd);
return;
}
/* is the peer instance willing to accept this connection */
if(mca_oob_tcp_peer_accept(peer, sd) == false) {
if(mca_oob_tcp_component.tcp_debug > 0) {
opal_output(0, "[%lu,%lu,%lu]-[%lu,%lu,%lu] mca_oob_tcp_recv_handler: "
"rejected connection from [%lu,%lu,%lu] connection state %d",
ORTE_NAME_ARGS(orte_process_info.my_name),
ORTE_NAME_ARGS(&(peer->peer_name)),
ORTE_NAME_ARGS(&(hdr->msg_src)),
peer->peer_state);
}
close(sd);
return;
}
}
/*
* Event callback when there is data available on the registered
* socket to recv.
*/
static void mca_oob_tcp_recv_handler(int sd, short flags, void* user)
{
mca_oob_tcp_hdr_t hdr;
mca_oob_tcp_event_t* event = (mca_oob_tcp_event_t *)user;
int rc;
/* accept new connections on the listen socket */
if(mca_oob_tcp_component.tcp_listen_sd == sd) {
mca_oob_tcp_accept();
return;
}
OBJ_RELEASE(event);
/* recv the process identifier */
while((rc = recv(sd, (char *)&hdr, sizeof(hdr), 0)) != sizeof(hdr)) {
if(rc >= 0) {
if(mca_oob_tcp_component.tcp_debug > 1) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: peer closed connection",
ORTE_NAME_ARGS(orte_process_info.my_name));
}
close(sd);
return;
}
if(ompi_socket_errno != EINTR) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: recv() failed with errno=%d\n",
ORTE_NAME_ARGS(orte_process_info.my_name), ompi_socket_errno);
close(sd);
return;
}
}
MCA_OOB_TCP_HDR_NTOH(&hdr);
/* dispatch based on message type */
switch(hdr.msg_type) {
case MCA_OOB_TCP_PROBE:
mca_oob_tcp_recv_probe(sd, &hdr);
break;
case MCA_OOB_TCP_CONNECT:
mca_oob_tcp_recv_connect(sd, &hdr);
break;
default:
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_recv_handler: invalid message type: %d\n",
ORTE_NAME_ARGS(orte_process_info.my_name), hdr.msg_type);
close(sd);
break;
}
}
/*
* Component initialization - create a module.
* (1) initialize static resources
* (2) create listen socket
*/
mca_oob_t* mca_oob_tcp_component_init(int* priority)
{
*priority = 1;
/* are there any interfaces? */
if(opal_ifcount() == 0)
return NULL;
/* initialize data structures */
opal_hash_table_init(&mca_oob_tcp_component.tcp_peers, 128);
opal_hash_table_init(&mca_oob_tcp_component.tcp_peer_names, 128);
opal_free_list_init(&mca_oob_tcp_component.tcp_peer_free,
sizeof(mca_oob_tcp_peer_t),
OBJ_CLASS(mca_oob_tcp_peer_t),
8, /* initial number */
mca_oob_tcp_component.tcp_peer_limit, /* maximum number */
8); /* increment to grow by */
opal_free_list_init(&mca_oob_tcp_component.tcp_msgs,
sizeof(mca_oob_tcp_msg_t),
OBJ_CLASS(mca_oob_tcp_msg_t),
8, /* initial number */
-1, /* maximum number */
8); /* increment to grow by */
/* intialize event library */
memset(&mca_oob_tcp_component.tcp_recv_event, 0, sizeof(opal_event_t));
memset(&mca_oob_tcp_component.tcp_send_event, 0, sizeof(opal_event_t));
/* create a listen socket */
if(mca_oob_tcp_create_listen() != OMPI_SUCCESS) {
opal_output(0, "mca_oob_tcp_init: unable to create listen socket\n");
return NULL;
}
return &mca_oob_tcp;
}
/*
* Callback from registry on change to subscribed segments.
*/
void mca_oob_tcp_registry_callback(
orte_gpr_notify_data_t* data,
void* cbdata)
{
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
size_t i, j, k;
orte_gpr_value_t **values, *value;
orte_gpr_keyval_t *keyval;
orte_buffer_t buffer;
mca_oob_tcp_addr_t* addr, *existing;
mca_oob_tcp_peer_t* peer;
if(mca_oob_tcp_component.tcp_debug > 1) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_registry_callback\n",
ORTE_NAME_ARGS(orte_process_info.my_name));
}
/* process the callback */
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
values = (orte_gpr_value_t**)(data->values)->addr;
for(i = 0, k=0; k < data->cnt &&
i < (data->values)->size; i++) {
if (NULL != values[i]) {
k++;
value = values[i];
for(j = 0; j < value->cnt; j++) {
/* check to make sure this is the requested key */
keyval = value->keyvals[j];
if(strcmp(keyval->key,"oob-tcp") != 0)
continue;
/* transfer ownership of registry object to buffer and unpack */
OBJ_CONSTRUCT(&buffer, orte_buffer_t);
if(orte_dps.load(&buffer,
keyval->value.byteobject.bytes,
keyval->value.byteobject.size) != ORTE_SUCCESS) {
/* TSW - throw ERROR */
continue;
}
keyval->type = ORTE_NULL;
keyval->value.byteobject.bytes = NULL;
keyval->value.byteobject.size = 0;
addr = mca_oob_tcp_addr_unpack(&buffer);
OBJ_DESTRUCT(&buffer);
if(NULL == addr) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_registry_callback: unable to unpack peer address\n",
ORTE_NAME_ARGS(orte_process_info.my_name));
continue;
}
if(mca_oob_tcp_component.tcp_debug > 1) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_registry_callback: received peer [%lu,%lu,%lu]\n",
ORTE_NAME_ARGS(orte_process_info.my_name),
ORTE_NAME_ARGS(&(addr->addr_name)));
}
/* check for existing cache entry */
existing = (mca_oob_tcp_addr_t *)orte_hash_table_get_proc(
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
&mca_oob_tcp_component.tcp_peer_names, &addr->addr_name);
if(NULL != existing) {
/* TSW - need to update existing entry */
OBJ_RELEASE(addr);
continue;
}
/* insert into cache and notify peer */
orte_hash_table_set_proc(&mca_oob_tcp_component.tcp_peer_names, &addr->addr_name, addr);
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
peer = (mca_oob_tcp_peer_t *)opal_hash_table_get_proc(
&mca_oob_tcp_component.tcp_peers, &addr->addr_name);
if(NULL != peer)
mca_oob_tcp_peer_resolved(peer, addr);
}
}
}
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
}
/*
* Attempt to resolve peer name.
*/
int mca_oob_tcp_resolve(mca_oob_tcp_peer_t* peer)
{
mca_oob_tcp_addr_t* addr;
mca_oob_tcp_subscription_t* subscription;
orte_gpr_trigger_t trig, *trigs;
orte_gpr_subscription_t sub, *subs;
opal_list_item_t* item;
int rc;
/* if the address is already cached - simply return it */
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
addr = (mca_oob_tcp_addr_t *)opal_hash_table_get_proc(&mca_oob_tcp_component.tcp_peer_names,
&peer->peer_name);
if(NULL != addr) {
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
mca_oob_tcp_peer_resolved(peer, addr);
return OMPI_SUCCESS;
}
/* check to see if we have subscribed to this registry segment */
for( item = opal_list_get_first(&mca_oob_tcp_component.tcp_subscriptions);
item != opal_list_get_end(&mca_oob_tcp_component.tcp_subscriptions);
item = opal_list_get_next(item)) {
subscription = (mca_oob_tcp_subscription_t*)item;
if(subscription->jobid == peer->peer_name.jobid) {
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
return OMPI_SUCCESS;
}
}
OBJ_CONSTRUCT(&sub, orte_gpr_subscription_t);
/* indicate that this is a standard subscription. This indicates that the
* subscription will be common to all processes. Thus, the resulting data
* can be consolidated into a process-independent message and broadcast
* to all processes
*/
if (ORTE_SUCCESS != (rc = orte_schema.get_std_subscription_name(&(sub.name),
OMPI_OOB_SUBSCRIPTION, peer->peer_name.jobid))) {
ORTE_ERROR_LOG(rc);
return rc;
}
/* send data when trigger fires, continue to monitor. The default
* action for any subscription that includes a trigger condition is
* to send the specified data when the trigger fires. This set of flags
* indicates that - AFTER the trigger fires - the subscription should
* continue to send data any time an entry is added or changed.
*/
sub.action = ORTE_GPR_NOTIFY_ADD_ENTRY |
ORTE_GPR_NOTIFY_VALUE_CHG |
ORTE_GPR_NOTIFY_STARTS_AFTER_TRIG;
/* setup the value structures that describe the data to
* be monitored and returned by this subscription
*/
sub.cnt = 1;
sub.values = (orte_gpr_value_t**)malloc(sizeof(orte_gpr_value_t*));
if (NULL == sub.values) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0] = OBJ_NEW(orte_gpr_value_t);
if (NULL == sub.values[0]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.cnt = 1;
/* define the segment */
if (ORTE_SUCCESS != (rc = orte_schema.get_job_segment_name(
&(sub.values[0]->segment),
peer->peer_name.jobid))) {
ORTE_ERROR_LOG(rc);
OBJ_DESTRUCT(&sub);
return rc;
}
sub.values[0]->addr_mode = ORTE_GPR_KEYS_OR | ORTE_GPR_TOKENS_OR;
/* look at all containers on this segment */
sub.values[0]->tokens = NULL;
sub.values[0]->num_tokens = 0;
/* look for any keyval with "modex" key */
sub.values[0]->cnt = 1;
sub.values[0]->keyvals = (orte_gpr_keyval_t**)malloc(sizeof(orte_gpr_keyval_t*));
if (NULL == sub.values[0]->keyvals) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0]->keyvals[0] = OBJ_NEW(orte_gpr_keyval_t);
if (NULL == sub.values[0]->keyvals[0]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0]->keyvals[0]->key = strdup("oob-tcp");
if (NULL == sub.values[0]->keyvals[0]->key) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
/* define the callback function */
sub.cbfunc = mca_oob_tcp_registry_callback;
sub.user_tag = NULL;
/* setup the trigger value */
OBJ_CONSTRUCT(&trig, orte_gpr_trigger_t);
if (ORTE_SUCCESS != (rc = orte_schema.get_std_trigger_name(&(trig.name),
ORTE_STG1_TRIGGER, peer->peer_name.jobid))) {
ORTE_ERROR_LOG(rc);
return rc;
}
/* this is an ORTE-standard trigger that is defined by the ORTE resource manager
* when the job was launched - therefore, we don't need to provide any additional
* info
*/
trigs = &trig;
subs = &sub;
subscription = OBJ_NEW(mca_oob_tcp_subscription_t);
subscription->jobid = peer->peer_name.jobid;
rc = orte_gpr.subscribe(1, &subs, 1, &trigs);
if(rc != OMPI_SUCCESS) {
ORTE_ERROR_LOG(rc);
OBJ_DESTRUCT(&sub);
OBJ_DESTRUCT(&trig);
return rc;
}
/* the id of each subscription is stored by the system in the corresponding
* subscription object we passed into orte_gpr.subscribe. We record it
* here so we can (if desired) cancel that subscription later
*/
subscription->subid = sub.id;
/* done with these, so release any memory */
OBJ_DESTRUCT(&sub);
OBJ_DESTRUCT(&trig);
opal_list_append(&mca_oob_tcp_component.tcp_subscriptions, &subscription->item);
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
return rc;
}
/*
* Setup contact information in the registry.
*/
int mca_oob_tcp_init(void)
{
orte_jobid_t jobid;
orte_buffer_t *buffer;
orte_gpr_trigger_t trig, *trigs;
orte_gpr_value_t *value;
mca_oob_tcp_subscription_t *subscription;
orte_gpr_subscription_t sub, *subs;
int rc;
opal_list_item_t* item;
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
char *tmp, *tmp2, *tmp3;
/* random delay to stagger connections back to seed */
#if defined(WIN32)
sleep((orte_process_info.my_name->vpid % orte_process_info.num_procs % 1000) * 1000);
#else
usleep((orte_process_info.my_name->vpid % orte_process_info.num_procs % 1000) * 1000);
#endif
/* get my jobid */
if (ORTE_SUCCESS != (rc = orte_ns.get_jobid(&jobid,
orte_process_info.my_name))) {
ORTE_ERROR_LOG(rc);
return rc;
}
/* iterate through the open connections and send an ident message to all peers -
* note that we initially come up w/out knowing our process name - and are assigned
* a temporary name by our peer. once we have determined our real name - we send it
* to the peer.
*/
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
for(item = opal_list_get_first(&mca_oob_tcp_component.tcp_peer_list);
item != opal_list_get_end(&mca_oob_tcp_component.tcp_peer_list);
item = opal_list_get_next(item)) {
mca_oob_tcp_peer_t* peer = (mca_oob_tcp_peer_t*)item;
mca_oob_tcp_peer_send_ident(peer);
}
/* register subscribe callback to receive notification when all processes have registered */
subscription = OBJ_NEW(mca_oob_tcp_subscription_t);
subscription->jobid = jobid;
opal_list_append(&mca_oob_tcp_component.tcp_subscriptions, &subscription->item);
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
if(mca_oob_tcp_component.tcp_debug > 2) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_init: calling orte_gpr.subscribe\n",
ORTE_NAME_ARGS(orte_process_info.my_name));
}
/* setup the subscription description value */
OBJ_CONSTRUCT(&sub, orte_gpr_subscription_t);
/* indicate that this is a standard subscription. This indicates that the
* subscription will be common to all processes. Thus, the resulting data
* can be consolidated into a process-independent message and broadcast
* to all processes
*/
if (ORTE_SUCCESS != (rc = orte_schema.get_std_subscription_name(&(sub.name),
OMPI_OOB_SUBSCRIPTION, jobid))) {
ORTE_ERROR_LOG(rc);
return rc;
}
/* send data when trigger fires, continue to monitor. The default
* action for any subscription that includes a trigger condition is
* to send the specified data when the trigger fires. This set of flags
* indicates that - AFTER the trigger fires - the subscription should
* continue to send data any time an entry is added or changed.
*/
sub.action = ORTE_GPR_NOTIFY_ADD_ENTRY |
ORTE_GPR_NOTIFY_VALUE_CHG |
ORTE_GPR_NOTIFY_STARTS_AFTER_TRIG;
/* setup the value structures that describe the data to
* be monitored and returned by this subscription
*/
sub.cnt = 1;
sub.values = (orte_gpr_value_t**)malloc(sizeof(orte_gpr_value_t*));
if (NULL == sub.values) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0] = OBJ_NEW(orte_gpr_value_t);
if (NULL == sub.values[0]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
/* define the segment */
if (ORTE_SUCCESS != (rc = orte_schema.get_job_segment_name(
&(sub.values[0]->segment),
jobid))) {
ORTE_ERROR_LOG(rc);
OBJ_DESTRUCT(&sub);
return rc;
}
sub.values[0]->addr_mode = ORTE_GPR_KEYS_OR | ORTE_GPR_TOKENS_OR;
/* look at all containers on this segment */
sub.values[0]->tokens = NULL;
sub.values[0]->num_tokens = 0;
/* look for any keyval with "modex" key */
sub.values[0]->cnt = 1;
sub.values[0]->keyvals = (orte_gpr_keyval_t**)malloc(sizeof(orte_gpr_keyval_t*));
if (NULL == sub.values[0]->keyvals) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0]->keyvals[0] = OBJ_NEW(orte_gpr_keyval_t);
if (NULL == sub.values[0]->keyvals[0]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
sub.values[0]->keyvals[0]->key = strdup("oob-tcp");
if (NULL == sub.values[0]->keyvals[0]->key) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
OBJ_DESTRUCT(&sub);
return ORTE_ERR_OUT_OF_RESOURCE;
}
/* define the callback function */
sub.cbfunc = mca_oob_tcp_registry_callback;
sub.user_tag = NULL;
/* setup the trigger value */
OBJ_CONSTRUCT(&trig, orte_gpr_trigger_t);
if (ORTE_SUCCESS != (rc = orte_schema.get_std_trigger_name(&(trig.name),
ORTE_STG1_TRIGGER, jobid))) {
ORTE_ERROR_LOG(rc);
return rc;
}
/* this is an ORTE-standard trigger that is defined by the ORTE resource manager
* when the job was launched - therefore, we don't need to provide any additional
* info
*/
trigs = &trig;
subs = &sub;
subscription = OBJ_NEW(mca_oob_tcp_subscription_t);
subscription->jobid = jobid;
rc = orte_gpr.subscribe(1, &subs, 1, &trigs);
if(rc != OMPI_SUCCESS) {
ORTE_ERROR_LOG(rc);
OBJ_DESTRUCT(&sub);
OBJ_DESTRUCT(&trig);
return rc;
}
/* the id of each subscription is stored by the system in the corresponding
* subscription object we passed into orte_gpr.subscribe. We record it
* here so we can (if desired) cancel that subscription later
*/
subscription->subid = sub.id;
/* done with these, so release any memory */
OBJ_DESTRUCT(&sub);
OBJ_DESTRUCT(&trig);
buffer = OBJ_NEW(orte_buffer_t);
if(buffer == NULL) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return OMPI_ERR_OUT_OF_RESOURCE;
}
rc = mca_oob_tcp_addr_pack(buffer);
if(rc != OMPI_SUCCESS) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(buffer);
return rc;
}
/* put our contact info in registry */
value = OBJ_NEW(orte_gpr_value_t);
if (NULL == value) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
value->addr_mode = ORTE_GPR_OVERWRITE | ORTE_GPR_TOKENS_XAND;
if (ORTE_SUCCESS != (rc = orte_schema.get_job_segment_name(&(value->segment), jobid))) {
ORTE_ERROR_LOG(rc);
return rc;
}
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
value->cnt = 2;
value->keyvals = (orte_gpr_keyval_t**)malloc(value->cnt * sizeof(orte_gpr_keyval_t*));
if(NULL == value->keyvals) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
value->keyvals[0] = OBJ_NEW(orte_gpr_keyval_t);
if (NULL == value->keyvals[0]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
value->keyvals[1] = OBJ_NEW(orte_gpr_keyval_t);
if (NULL == value->keyvals[1]) {
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
return ORTE_ERR_OUT_OF_RESOURCE;
}
if (ORTE_SUCCESS != (rc = orte_schema.get_proc_tokens(&(value->tokens),
&(value->num_tokens), orte_process_info.my_name))) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(value);
return rc;
}
(value->keyvals[0])->type = ORTE_BYTE_OBJECT;
(value->keyvals[0])->key = strdup("oob-tcp");
rc = orte_dps.unload(buffer, (void**)&(value->keyvals[0])->value.byteobject.bytes,
&(value->keyvals[0])->value.byteobject.size);
if(rc != ORTE_SUCCESS) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(value);
OBJ_RELEASE(buffer);
return rc;
}
First phase of the scalable RTE changes: 1. Modify the registry to eliminate redundant data copying for startup messages. 2. Revise the subscription/trigger system to avoid redundant storage of triggers and subscriptions. This dramatically reduces the search time when a registry action occurs - to illustrate the point, there are now only a handful of triggers on the system for each job. Before, there were a handful of triggers for each PROCESS in the job, all of which had to be checked every time something happened on the registry. This is much, much faster now. 3. Update all subscriptions to the new format. There are now "named" subscriptions - this allows you to "name" a subscription that all the processes will be using. The first one to hit the registry actually defines the subscription. From then on, any subsequent "subscribes" to the same name just cause that process to "attach" to the existing subscription. This keeps the number of subscriptions being tracked by the registry to a minimum, while ensuring that each process still gets notified. 4. Do the same for triggers. Also fixed a duplicate subscription problem that was causing people to receive data equal to the number of processes times the data they should have received from a trigger/subscription. Sorry about that... :-( ...but it's all better now! Uncovered a situation where the modex data seems to be getting entered on the registry a second time - the latter time coming after the compound command has been "fired", thereby causing all the subscriptions to fire. Asked Tim and Jeff to look into this. Second phase of the changes will involve modifying the xcast system so that the same message gets sent to all processes. This will further reduce the message traffic, and - once we have a true "broadcast" version of xcast - really speed things up and improve scalability. This commit was SVN r6542.
2005-07-18 22:49:00 +04:00
(value->keyvals[1])->type = ORTE_STRING;
(value->keyvals[1])->key = strdup(ORTE_PROC_RML_IP_ADDRESS_KEY);
tmp = mca_oob.oob_get_addr();
tmp2 = strrchr(tmp, '/');
tmp3 = strrchr(tmp, ':');
*tmp3 = '\0';
(value->keyvals[1])->value.strptr = strdup(tmp2);
free(tmp);
if(mca_oob_tcp_component.tcp_debug > 2) {
opal_output(0, "[%lu,%lu,%lu] mca_oob_tcp_init: calling orte_gpr.put(%s)\n",
ORTE_NAME_ARGS(orte_process_info.my_name),
value->segment);
}
rc = orte_gpr.put(1, &value);
if(rc != OMPI_SUCCESS) {
ORTE_ERROR_LOG(rc);
OBJ_RELEASE(value);
OBJ_RELEASE(buffer);
return rc;
}
OBJ_RELEASE(buffer);
OBJ_RELEASE(value);
if(rc != ORTE_SUCCESS) {
ORTE_ERROR_LOG(rc);
return rc;
}
return OMPI_SUCCESS;
}
/*
* Module cleanup.
*/
int mca_oob_tcp_fini(void)
{
opal_list_item_t *item;
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
opal_event_disable(); /* disable event processing */
/* close listen socket */
if (mca_oob_tcp_component.tcp_listen_sd >= 0) {
opal_event_del(&mca_oob_tcp_component.tcp_recv_event);
close(mca_oob_tcp_component.tcp_listen_sd);
mca_oob_tcp_component.tcp_listen_sd = -1;
}
/* cleanup all peers */
for(item = opal_list_remove_first(&mca_oob_tcp_component.tcp_peer_list);
item != NULL;
item = opal_list_remove_first(&mca_oob_tcp_component.tcp_peer_list)) {
mca_oob_tcp_peer_t* peer = (mca_oob_tcp_peer_t*)item;
MCA_OOB_TCP_PEER_RETURN(peer);
}
/* delete any pending events */
for(item = opal_list_remove_first(&mca_oob_tcp_component.tcp_events);
item != NULL;
item = opal_list_remove_first(&mca_oob_tcp_component.tcp_events)) {
mca_oob_tcp_event_t* event = (mca_oob_tcp_event_t*)item;
opal_event_del(&event->event);
OBJ_RELEASE(event);
}
opal_event_enable();
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
return OMPI_SUCCESS;
}
/*
* Compare two process names for equality.
*
* @param n1 Process name 1.
* @param n2 Process name 2.
* @return (-1 for n1<n2 0 for equality, 1 for n1>n2)
*
* Note that the definition of < or > is somewhat arbitrary -
* just needs to be consistently applied to maintain an ordering
* when process names are used as indices.
*/
int mca_oob_tcp_process_name_compare(const orte_process_name_t* n1, const orte_process_name_t* n2)
{
return orte_ns.compare(ORTE_NS_CMP_ALL, n1, n2);
}
/*
* Return local process address as a URI string.
*/
char* mca_oob_tcp_get_addr(void)
{
int i;
char *contact_info = (char *)malloc((opal_ifcount()+1) * 32);
char *ptr = contact_info;
*ptr = 0;
for(i=opal_ifbegin(); i>0; i=opal_ifnext(i)) {
struct sockaddr_in addr;
char name[32];
opal_ifindextoname(i, name, sizeof(name));
if (mca_oob_tcp_component.tcp_include != NULL &&
strstr(mca_oob_tcp_component.tcp_include,name) == NULL)
continue;
if (mca_oob_tcp_component.tcp_exclude != NULL &&
strstr(mca_oob_tcp_component.tcp_exclude,name) != NULL)
continue;
opal_ifindextoaddr(i, (struct sockaddr*)&addr, sizeof(addr));
if(opal_ifcount() > 1 && addr.sin_addr.s_addr == inet_addr("127.0.0.1"))
continue;
if(ptr != contact_info) {
ptr += sprintf(ptr, ";");
}
ptr += sprintf(ptr, "tcp://%s:%d", inet_ntoa(addr.sin_addr), ntohs(mca_oob_tcp_component.tcp_listen_port));
}
return contact_info;
}
/*
* Parse a URI string into an IP address and port number.
*/
int mca_oob_tcp_parse_uri(const char* uri, struct sockaddr_in* inaddr)
{
char* tmp = strdup(uri);
char* ptr = tmp + 6;
char* addr = ptr;
char* port;
if(strncmp(tmp, "tcp://", 6) != 0) {
free(tmp);
return OMPI_ERR_BAD_PARAM;
}
ptr = strchr(addr, ':');
if(NULL == ptr) {
free(tmp);
return OMPI_ERR_BAD_PARAM;
}
*ptr = '\0';
ptr++;
port = ptr;
memset(inaddr, 0, sizeof(inaddr));
inaddr->sin_family = AF_INET;
inaddr->sin_addr.s_addr = inet_addr(addr);
if(inaddr->sin_addr.s_addr == INADDR_ANY) {
free(tmp);
return OMPI_ERR_BAD_PARAM;
}
inaddr->sin_port = htons(atoi(port));
free(tmp);
return OMPI_SUCCESS;
}
/*
* Setup address in the cache. Note that this could be called multiple
* times if a given destination exports multiple addresses.
*/
int mca_oob_tcp_set_addr(const orte_process_name_t* name, const char* uri)
{
struct sockaddr_in inaddr;
mca_oob_tcp_addr_t* addr;
mca_oob_tcp_peer_t* peer;
int rc;
if((rc = mca_oob_tcp_parse_uri(uri,&inaddr)) != OMPI_SUCCESS)
return rc;
OPAL_THREAD_LOCK(&mca_oob_tcp_component.tcp_lock);
addr = (mca_oob_tcp_addr_t*)opal_hash_table_get_proc(&mca_oob_tcp_component.tcp_peer_names, name);
if(NULL == addr) {
addr = OBJ_NEW(mca_oob_tcp_addr_t);
addr->addr_name = *name;
opal_hash_table_set_proc(&mca_oob_tcp_component.tcp_peer_names, &addr->addr_name, addr);
}
rc = mca_oob_tcp_addr_insert(addr, &inaddr);
peer = (mca_oob_tcp_peer_t *)opal_hash_table_get_proc(
&mca_oob_tcp_component.tcp_peers, &addr->addr_name);
if(NULL != peer) {
mca_oob_tcp_peer_resolved(peer, addr);
}
OPAL_THREAD_UNLOCK(&mca_oob_tcp_component.tcp_lock);
return rc;
}