/*
* Process one line from the orte_base_user_debugger MCA param and
* look for that debugger in the path. If we find it, fill in
* new_argv.
*/
static int process(char *orig_line, char *basename, opal_cmd_line_t *cmd_line,
int argc, char **argv, char ***new_argv, int num_procs)
{
int ret = ORTE_SUCCESS;
int i, j, count;
char *line = NULL, *tmp = NULL, *full_line = strdup(orig_line);
char **orterun_argv = NULL, **executable_argv = NULL, **line_argv = NULL;
char cwd[OPAL_PATH_MAX];
bool used_num_procs = false;
bool single_app = false;
bool fail_needed_executable = false;
line = full_line;
if (NULL == line) {
ret = ORTE_ERR_OUT_OF_RESOURCE;
goto out;
}
/* Trim off whitespace at the beginning and ending of line */
for (i = 0; '\0' != line[i] && isspace(line[i]); ++line) {
continue;
}
for (i = strlen(line) - 2; i > 0 && isspace(line[i]); ++i) {
line[i] = '\0';
}
if (strlen(line) <= 0) {
ret = ORTE_ERROR;
goto out;
}
/* Get the tail of the command line (i.e., the user executable /
argv) */
opal_cmd_line_get_tail(cmd_line, &i, &executable_argv);
/* Make a new copy of the orterun command line args, without the
orterun token itself, and without the --debug, --debugger, and
-tv flags. */
orterun_argv = opal_argv_copy(argv);
count = opal_argv_count(orterun_argv);
opal_argv_delete(&count, &orterun_argv, 0, 1);
for (i = 0; NULL != orterun_argv[i]; ++i) {
count = opal_argv_count(orterun_argv);
if (0 == strcmp(orterun_argv[i], "-debug") ||
0 == strcmp(orterun_argv[i], "--debug")) {
opal_argv_delete(&count, &orterun_argv, i, 1);
} else if (0 == strcmp(orterun_argv[i], "-tv") ||
0 == strcmp(orterun_argv[i], "--tv")) {
opal_argv_delete(&count, &orterun_argv, i, 1);
} else if (0 == strcmp(orterun_argv[i], "--debugger") ||
0 == strcmp(orterun_argv[i], "-debugger")) {
opal_argv_delete(&count, &orterun_argv, i, 2);
}
}
/* Replace @@ tokens - line should never realistically be bigger
than MAX_INT, so just cast to int to remove compiler warning */
*new_argv = NULL;
line_argv = opal_argv_split(line, ' ');
if (NULL == line_argv) {
ret = ORTE_ERR_NOT_FOUND;
goto out;
}
for (i = 0; NULL != line_argv[i]; ++i) {
if (0 == strcmp(line_argv[i], "@mpirun@") ||
0 == strcmp(line_argv[i], "@orterun@")) {
opal_argv_append_nosize(new_argv, argv[0]);
} else if (0 == strcmp(line_argv[i], "@mpirun_args@") ||
0 == strcmp(line_argv[i], "@orterun_args@")) {
for (j = 0; NULL != orterun_argv && NULL != orterun_argv[j]; ++j) {
opal_argv_append_nosize(new_argv, orterun_argv[j]);
}
} else if (0 == strcmp(line_argv[i], "@np@")) {
asprintf(&tmp, "%d", num_procs);
opal_argv_append_nosize(new_argv, tmp);
free(tmp);
} else if (0 == strcmp(line_argv[i], "@single_app@")) {
/* This token is only a flag; it is not replaced with any
alternate text */
single_app = true;
} else if (0 == strcmp(line_argv[i], "@executable@")) {
/* If we found the executable, paste it in. Otherwise,
this is a possible error. */
if (NULL != executable_argv) {
opal_argv_append_nosize(new_argv, executable_argv[0]);
} else {
fail_needed_executable = true;
}
} else if (0 == strcmp(line_argv[i], "@executable_argv@")) {
/* If we found the tail, paste in the argv. Otherwise,
this is a possible error. */
if (NULL != executable_argv) {
for (j = 1; NULL != executable_argv[j]; ++j) {
opal_argv_append_nosize(new_argv, executable_argv[j]);
}
} else {
fail_needed_executable = true;
}
} else {
/* It wasn't a special token, so just copy it over */
opal_argv_append_nosize(new_argv, line_argv[i]);
}
}
/* Can we find argv[0] in the path? */
getcwd(cwd, OPAL_PATH_MAX);
tmp = opal_path_findv((*new_argv)[0], X_OK, environ, cwd);
if (NULL != tmp) {
free(tmp);
/* Ok, we found a good debugger. Check for some error
conditions. */
tmp = opal_argv_join(argv, ' ');
/* We do not support launching a debugger that requires the
-np value if the user did not specify -np on the command
line. */
if (used_num_procs && 0 == num_procs) {
free(tmp);
tmp = opal_argv_join(orterun_argv, ' ');
orte_show_help("help-orterun.txt", "debugger requires -np",
true, (*new_argv)[0], argv[0], tmp,
(*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Some debuggers do not support launching MPMD */
else if (single_app && NULL != strchr(tmp, ':')) {
orte_show_help("help-orterun.txt",
"debugger only accepts single app", true,
(*new_argv)[0], (*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Some debuggers do not use orterun/mpirun, and therefore
must have an executable to run (e.g., cannot use mpirun's
app context file feature). */
else if (fail_needed_executable) {
orte_show_help("help-orterun.txt",
"debugger requires executable", true,
(*new_argv)[0], argv[0], (*new_argv)[0], argv[0],
(*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Otherwise, we succeeded. Return happiness. */
else {
goto out;
}
free(tmp);
}
/* All done -- didn't find it */
opal_argv_free(*new_argv);
*new_argv = NULL;
ret = ORTE_ERR_NOT_FOUND;
out:
if (NULL != orterun_argv) {
opal_argv_free(orterun_argv);
}
if (NULL != executable_argv) {
opal_argv_free(executable_argv);
}
if (NULL != line_argv) {
opal_argv_free(line_argv);
}
if (NULL != tmp) {
free(tmp);
}
if (NULL != full_line) {
free(full_line);
}
return ret;
}
static int parse_args(int argc, char *argv[]) {
int i, ret, len, exit_status = ORTE_SUCCESS ;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
char * tmp_env_var = NULL;
/* Init structure */
memset(&orte_checkpoint_globals, 0, sizeof(orte_checkpoint_globals_t));
orte_checkpoint_globals.help = false;
orte_checkpoint_globals.pid = -1;
orte_checkpoint_globals.term = false;
orte_checkpoint_globals.verbose = false;
orte_checkpoint_globals.req_hnp = ORTE_JOBID_INVALID;
orte_checkpoint_globals.nowait = false;
orte_checkpoint_globals.status = false;
orte_checkpoint_globals.output = -1;
orte_checkpoint_globals.ckpt_status = ORTE_SNAPC_CKPT_STATE_NONE;
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, true, argc, argv);
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
tmp_env_var = mca_base_param_env_var("opal_cr_is_tool");
opal_setenv(tmp_env_var,
"1",
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/**
* Now start parsing our specific arguments
*/
/* get the remaining bits */
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
#if OPAL_ENABLE_FT == 0
/* Warn and exit if not configured with Checkpoint/Restart */
{
char *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
orte_show_help("help-orte-checkpoint.txt", "usage-no-cr",
true, args);
free(args);
exit_status = ORTE_ERROR;
goto cleanup;
}
#endif
if (OPAL_SUCCESS != ret ||
orte_checkpoint_globals.help ||
(0 >= argc && ORTE_JOBID_INVALID == orte_checkpoint_globals.req_hnp)) {
char *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
orte_show_help("help-orte-checkpoint.txt", "usage", true,
args);
free(args);
exit_status = ORTE_ERROR;
goto cleanup;
}
/*
* If the user did not supply an hnp jobid, then they must
* supply the PID of MPIRUN
*/
if(0 >= argc &&
ORTE_JOBID_INVALID != orte_checkpoint_globals.req_hnp) {
exit_status = ORTE_SUCCESS;
goto cleanup;
}
orte_checkpoint_globals.pid = atoi(argv[0]);
if ( 0 >= orte_checkpoint_globals.pid ) {
orte_show_help("help-orte-checkpoint.txt", "invalid_pid", true,
orte_checkpoint_globals.pid);
exit_status = ORTE_ERROR;
goto cleanup;
}
/*
* JJH: No wait is currently not implemented or tested
*/
if(orte_checkpoint_globals.nowait) {
orte_checkpoint_globals.nowait = false;
orte_show_help("help-orte-checkpoint.txt", "not_impl",
true,
"Disconnected checkpoint");
}
if(orte_checkpoint_globals.verbose) {
orte_checkpoint_globals.status = true;
}
cleanup:
return exit_status;
}
int main(int argc, char *argv[])
{
int32_t ret, i;
opal_cmd_line_t cmd_line;
char **inpt;
opal_buffer_t *buf;
int count;
char cwd[OPAL_PATH_MAX];
orcm_tool_cmd_t flag = ORCM_TOOL_STOP_CMD;
int32_t master=0;
uint16_t jfam=0;
/***************
* Initialize
***************/
/*
* Make sure to init util before parse_args
* to ensure installdirs is setup properly
* before calling mca_base_open();
*/
if( ORTE_SUCCESS != (ret = orcm_init_util()) ) {
return ret;
}
/* initialize the globals */
my_globals.help = false;
my_globals.replicas = NULL;
my_globals.sched = NULL;
my_globals.hnp_uri = NULL;
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, true, argc, argv);
/* extract the MCA/GMCA params */
mca_base_cmd_line_process_args(&cmd_line, &environ, &environ);
/**
* Now start parsing our specific arguments
*/
if (OPAL_SUCCESS != ret || my_globals.help) {
char *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
orte_show_help("help-orcm-stop.txt", "usage", true, args);
free(args);
return ORTE_ERROR;
}
if (NULL != my_globals.sched) {
if (0 == strncmp(my_globals.sched, "file", strlen("file")) ||
0 == strncmp(my_globals.sched, "FILE", strlen("FILE"))) {
char input[1024], *filename;
FILE *fp;
/* it is a file - get the filename */
filename = strchr(my_globals.sched, ':');
if (NULL == filename) {
/* filename is not correctly formatted */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-bad", true, "scheduler", my_globals.sched);
return ORTE_ERROR;
}
++filename; /* space past the : */
if (0 >= strlen(filename)) {
/* they forgot to give us the name! */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-bad", true, "scheduler", my_globals.sched);
return ORTE_ERROR;
}
/* open the file and extract the pid */
fp = fopen(filename, "r");
if (NULL == fp) { /* can't find or read file! */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-access", true, "scheduler", filename);
return ORTE_ERROR;
}
if (NULL == fgets(input, 1024, fp)) {
/* something malformed about file */
fclose(fp);
orte_show_help("help-openrcm-runtime.txt", "hnp-file-bad", "scheduler", true, filename);
return ORTE_ERROR;
}
fclose(fp);
input[strlen(input)-1] = '\0'; /* remove newline */
/* convert the pid */
master = strtoul(input, NULL, 10);
} else {
/* should just be the master itself */
master = strtoul(my_globals.sched, NULL, 10);
}
}
/* if we were given HNP contact info, parse it and
* setup the process_info struct with that info
*/
if (NULL != my_globals.hnp_uri) {
if (0 == strncmp(my_globals.hnp_uri, "file", strlen("file")) ||
0 == strncmp(my_globals.hnp_uri, "FILE", strlen("FILE"))) {
char input[1024], *filename;
FILE *fp;
/* it is a file - get the filename */
filename = strchr(my_globals.hnp_uri, ':');
if (NULL == filename) {
/* filename is not correctly formatted */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-bad", true, "uri", my_globals.hnp_uri);
goto cleanup;
}
++filename; /* space past the : */
if (0 >= strlen(filename)) {
/* they forgot to give us the name! */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-bad", true, "uri", my_globals.hnp_uri);
goto cleanup;
}
/* open the file and extract the uri */
fp = fopen(filename, "r");
if (NULL == fp) { /* can't find or read file! */
orte_show_help("help-openrcm-runtime.txt", "hnp-filename-access", true, filename);
goto cleanup;
}
if (NULL == fgets(input, 1024, fp)) {
/* something malformed about file */
fclose(fp);
orte_show_help("help-openrcm-runtime.txt", "hnp-file-bad", true, filename);
goto cleanup;
}
fclose(fp);
input[strlen(input)-1] = '\0'; /* remove newline */
/* put into the process info struct */
orte_process_info.my_hnp_uri = strdup(input);
} else {
/* should just be the uri itself */
orte_process_info.my_hnp_uri = strdup(my_globals.hnp_uri);
}
}
if (OPAL_SUCCESS != opal_getcwd(cwd, sizeof(cwd))) {
opal_output(orte_clean_output, "failed to get cwd\n");
return ORTE_ERR_NOT_FOUND;
}
/***************************
* We need all of OPAL and ORTE - this will
* automatically connect us to the CM
***************************/
if (ORTE_SUCCESS != orcm_init(ORCM_TOOL)) {
orcm_finalize();
return 1;
}
/* if we were given the hnp uri, extract the job family for the
* master id
*/
if (NULL != my_globals.hnp_uri) {
master = ORTE_JOB_FAMILY(ORTE_PROC_MY_HNP->jobid);
}
/* register to receive responses */
if (ORCM_SUCCESS != (ret = orcm_pnp.register_receive("orcm-stop", "0.1", "alpha",
ORCM_PNP_GROUP_INPUT_CHANNEL,
ORCM_PNP_TAG_TOOL,
ack_recv, NULL))) {
ORTE_ERROR_LOG(ret);
goto cleanup;
}
/* announce my existence */
if (ORCM_SUCCESS != (ret = orcm_pnp.announce("orcm-stop", "0.1", "alpha", NULL))) {
ORTE_ERROR_LOG(ret);
goto cleanup;
}
/* setup the buffer to send our cmd */
buf = OBJ_NEW(opal_buffer_t);
/* indicate the scheduler to be used */
jfam = master & 0x0000ffff;
opal_dss.pack(buf, &jfam, 1, OPAL_UINT16);
/* get the apps to stop */
inpt = NULL;
opal_cmd_line_get_tail(&cmd_line, &count, &inpt);
if (0 == count) {
/* if no apps were given, then we stop the entire
* DVM itself by telling the daemon's to terminate
*/
if (ORCM_SUCCESS != (ret = orcm_pnp.output_nb(ORCM_PNP_SYS_CHANNEL,
NULL, ORCM_PNP_TAG_TERMINATE,
NULL, 0, buf, cbfunc, NULL))) {
ORTE_ERROR_LOG(ret);
}
goto cleanup;
} else {
/* load the stop cmd */
opal_dss.pack(buf, &flag, 1, ORCM_TOOL_CMD_T);
/* for each app */
for (i=0; NULL != inpt[i]; i++) {
opal_dss.pack(buf, &inpt[i], 1, OPAL_STRING);
/* pack the replicas to be stopped */
opal_dss.pack(buf, &my_globals.replicas, 1, OPAL_STRING);
}
opal_argv_free(inpt);
if (ORCM_SUCCESS != (ret = orcm_pnp.output_nb(ORCM_PNP_SYS_CHANNEL,
NULL, ORCM_PNP_TAG_TOOL,
NULL, 0, buf, cbfunc, NULL))) {
ORTE_ERROR_LOG(ret);
}
}
/* now wait for ack */
opal_event_dispatch(opal_event_base);
/***************
* Cleanup
***************/
cleanup:
orcm_finalize();
return ret;
}
static int parse_args(int argc, char *argv[]) {
int i, ret, len, exit_status = ORTE_SUCCESS ;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
char * tmp_env_var = NULL;
char *argv0 = NULL;
/* Init structure */
memset(&orte_migrate_globals, 0, sizeof(orte_migrate_globals_t));
orte_migrate_globals.help = false;
orte_migrate_globals.pid = -1;
orte_migrate_globals.verbose = false;
orte_migrate_globals.verbose_level = 0;
orte_migrate_globals.status = false;
orte_migrate_globals.output = -1;
orte_migrate_globals.off_nodes = NULL;
orte_migrate_globals.off_procs = NULL;
orte_migrate_globals.onto_nodes = NULL;
#if OPAL_ENABLE_FT_CR == 0
/* Warn and exit if not configured with Migrate/Restart */
{
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-migrate.txt", "usage-no-cr",
true, args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
exit_status = ORTE_ERROR;
goto cleanup;
}
#endif
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, false, argc, argv);
if (OPAL_SUCCESS != ret) {
if (OPAL_ERR_SILENT != ret) {
fprintf(stderr, "%s: command line error (%s)\n", argv[0],
opal_strerror(ret));
}
exit_status = 1;
goto cleanup;
}
if (orte_migrate_globals.help) {
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-migrate.txt", "usage", true,
args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
/* If we show the help message, that should be all we do */
exit(0);
}
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
tmp_env_var = mca_base_param_env_var("opal_cr_is_tool");
opal_setenv(tmp_env_var,
"1",
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/**
* Now start parsing our specific arguments
*/
/* get the remaining bits */
argv0 = strdup(argv[0]);
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
if (NULL == orte_migrate_globals.off_nodes &&
NULL == orte_migrate_globals.off_procs) {
fprintf(stderr, "%s: Nothing to do\n", argv0);
fprintf(stderr, "Type '%s --help' for usage.\n", argv0);
exit_status = 1;
goto cleanup;
}
if(orte_migrate_globals.verbose_level < 0 ) {
orte_migrate_globals.verbose_level = 0;
}
if(orte_migrate_globals.verbose_level > 0) {
orte_migrate_globals.verbose = true;
}
/*
* If the user did not supply an hnp jobid, then they must
* supply the PID of MPIRUN
*/
if(0 >= argc ) {
fprintf(stderr, "%s: Nothing to do\n", argv[0]);
fprintf(stderr, "Type '%s --help' for usage.\n", argv[0]);
exit_status = ORTE_ERROR;
goto cleanup;
}
orte_migrate_globals.pid = atoi(argv[0]);
if ( 0 >= orte_migrate_globals.pid ) {
opal_show_help("help-orte-migrate.txt", "invalid_pid", true,
orte_migrate_globals.pid);
exit_status = ORTE_ERROR;
goto cleanup;
}
if(orte_migrate_globals.verbose) {
orte_migrate_globals.status = true;
}
if(orte_migrate_globals.verbose) {
pretty_print_migration();
}
cleanup:
if (NULL != argv0) {
free(argv0);
}
return exit_status;
}
/*
* Process one line from the orte_base_user_debugger MCA param and
* look for that debugger in the path. If we find it, fill in
* new_argv.
*/
static int process(char *orig_line, char *basename, opal_cmd_line_t *cmd_line,
int argc, char **argv, char ***new_argv)
{
int i;
char *line, *full_line = strdup(orig_line);
char *user_argv, *tmp, *tmp2, **tmp_argv, **executable;
char cwd[PATH_MAX];
bool used_num_procs = false;
bool fail_single_app = false;
bool fail_needed_executable = false;
line = full_line;
if (NULL == line) {
return ORTE_ERR_OUT_OF_RESOURCE;
}
/* Trim off whitespace at the beginning and ending of line */
for (i = 0; '\0' != line[i] && isspace(line[i]); ++line) {
continue;
}
for (i = strlen(line) - 2; i > 0 && isspace(line[i]); ++i) {
line[i] = '\0';
}
if (strlen(line) <= 0) {
return ORTE_ERROR;
}
/* Get the tail of the command line (i.e., the user executable /
argv) */
opal_cmd_line_get_tail(cmd_line, &i, &executable);
/* Remove --debug, --debugger, and -tv from the user command line
params */
if (1 == argc) {
user_argv = strdup("");
} else {
tmp_argv = opal_argv_copy(argv);
for (i = 0; NULL != tmp_argv[i]; ++i) {
if (0 == strcmp(tmp_argv[i], "-debug") ||
0 == strcmp(tmp_argv[i], "--debug")) {
free(tmp_argv[i]);
tmp_argv[i] = strdup("");
} else if (0 == strcmp(tmp_argv[i], "-tv") ||
0 == strcmp(tmp_argv[i], "--tv")) {
free(tmp_argv[i]);
tmp_argv[i] = strdup("");
} else if (0 == strcmp(tmp_argv[i], "--debugger") ||
0 == strcmp(tmp_argv[i], "-debugger")) {
free(tmp_argv[i]);
tmp_argv[i] = strdup("");
if (NULL != tmp_argv[i + 1]) {
++i;
free(tmp_argv[i]);
tmp_argv[i] = strdup("");
}
}
}
user_argv = opal_argv_join(tmp_argv + 1, ' ');
opal_argv_free(tmp_argv);
}
/* Replace @@ tokens - line should never realistically be bigger
than MAX_INT, so just cast to int to remove compiler warning */
for (i = 0; i < (int) strlen(line); ++i) {
tmp = NULL;
if (0 == strncmp(line + i, "@mpirun@", 8)) {
line[i] = '\0';
asprintf(&tmp, "%s%s%s", line, argv[0], line + i + 8);
} else if (0 == strncmp(line + i, "@orterun@", 9)) {
line[i] = '\0';
asprintf(&tmp, "%s%s%s", line, argv[0], line + i + 9);
} else if (0 == strncmp(line + i, "@mpirun_args@", 13)) {
line[i] = '\0';
asprintf(&tmp, "%s%s%s", line, user_argv, line + i + 13);
} else if (0 == strncmp(line + i, "@orterun_args@", 14)) {
line[i] = '\0';
asprintf(&tmp, "%s%s%s", line, user_argv, line + i + 14);
} else if (0 == strncmp(line + i, "@np@", 4)) {
line[i] = '\0';
asprintf(&tmp, "%s%d%s", line, orterun_globals.num_procs,
line + i + 4);
used_num_procs = true;
} else if (0 == strncmp(line + i, "@single_app@", 12)) {
line[i] = '\0';
/* This token is only a flag; it is not replaced with any
alternate text */
asprintf(&tmp, "%s%s", line, line + i + 12);
tmp2 = opal_argv_join(executable + 1, ' ');
if (NULL != strchr(tmp2, ':')) {
fail_single_app = true;
}
free(tmp2);
} else if (0 == strncmp(line + i, "@executable@", 12)) {
line[i] = '\0';
/* If we found the executable, paste it in. Otherwise,
this is a possible error. */
if (NULL != executable) {
asprintf(&tmp, "%s%s%s", line, executable[0], line + i + 12);
} else {
fail_needed_executable = true;
}
} else if (0 == strncmp(line + i, "@executable_argv@", 17)) {
line[i] = '\0';
/* If we found the tail, paste in the argv. Otherwise,
this is a possible error. */
if (NULL != executable) {
if (NULL != executable[1]) {
/* Put in the argv */
tmp2 = opal_argv_join(executable + 1, ' ');
asprintf(&tmp, "%s%s%s", line, tmp2, line + i + 17);
free(tmp2);
} else {
/* There is no argv; just paste the front and back
together, removing the @token@ */
asprintf(&tmp, "%s%s", line, line + i + 17);
}
} else {
fail_needed_executable = true;
}
}
if (NULL != tmp) {
free(full_line);
full_line = line = tmp;
--i;
}
}
/* Split up into argv */
*new_argv = opal_argv_split(line, ' ');
free(line);
/* Can we find argv[0] in the path? */
getcwd(cwd, PATH_MAX);
tmp = opal_path_findv((*new_argv)[0], X_OK, environ, cwd);
if (NULL != tmp) {
free(tmp);
/* Ok, we found a good debugger. Check for some error
conditions. */
/* We do not support launching a debugger that requires the
-np value if the user did not specify -np on the command
line. */
if (used_num_procs && 0 == orterun_globals.num_procs) {
opal_show_help("help-orterun.txt", "debugger requires -np",
true, (*new_argv)[0], argv[0], user_argv,
(*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Some debuggers do not support launching MPMD */
else if (fail_single_app) {
opal_show_help("help-orterun.txt",
"debugger only accepts single app", true,
(*new_argv)[0], (*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Some debuggers do not use orterun/mpirun, and therefore
must have an executable to run (e.g., cannot use mpirun's
app context file feature). */
else if (fail_needed_executable) {
opal_show_help("help-orterun.txt",
"debugger requires executable", true,
(*new_argv)[0], argv[0], (*new_argv)[0], argv[0],
(*new_argv)[0]);
/* Fall through to free / fail, below */
}
/* Otherwise, we succeeded. Return happiness. */
else {
return ORTE_SUCCESS;
}
}
/* All done -- didn't find it */
opal_argv_free(*new_argv);
*new_argv = NULL;
return ORTE_ERR_NOT_FOUND;
}
static int parse_args(int argc, char *argv[]) {
int i, ret, len, exit_status = ORTE_SUCCESS ;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
char * tmp_env_var = NULL;
char *argv0 = NULL;
/* Init structure */
memset(&orte_checkpoint_globals, 0, sizeof(orte_checkpoint_globals_t));
orte_checkpoint_globals.help = false;
orte_checkpoint_globals.pid = -1;
orte_checkpoint_globals.verbose = false;
orte_checkpoint_globals.verbose_level = 0;
orte_checkpoint_globals.req_hnp = ORTE_JOBID_INVALID;
orte_checkpoint_globals.nowait = false;
orte_checkpoint_globals.status = false;
orte_checkpoint_globals.output = -1;
orte_checkpoint_globals.ckpt_status = ORTE_SNAPC_CKPT_STATE_NONE;
orte_checkpoint_globals.list_only = false;
#if OPAL_ENABLE_CRDEBUG == 1
orte_checkpoint_globals.enable_crdebug = false;
#endif
orte_checkpoint_globals.options = OBJ_NEW(opal_crs_base_ckpt_options_t);
orte_checkpoint_globals.term = false;
orte_checkpoint_globals.stop = false;
#if OPAL_ENABLE_CRDEBUG == 1
orte_checkpoint_globals.attach_debugger = false;
orte_checkpoint_globals.detach_debugger = false;
#endif
#if OPAL_ENABLE_FT_CR == 0
/* Warn and exit if not configured with Checkpoint/Restart */
{
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-checkpoint.txt", "usage-no-cr",
true, args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
exit_status = ORTE_ERROR;
goto cleanup;
}
#endif
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, true, argc, argv);
if (OPAL_SUCCESS != ret) {
if (OPAL_ERR_SILENT != ret) {
fprintf(stderr, "%s: command line error (%s)\n", argv[0],
opal_strerror(ret));
}
exit_status = 1;
goto cleanup;
}
if (orte_checkpoint_globals.help) {
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-checkpoint.txt", "usage", true,
args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
/* If we show the help message, that should be all we do */
exit(0);
}
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
(void) mca_base_var_env_name("opal_cr_is_tool", &tmp_env_var);
opal_setenv(tmp_env_var,
"1",
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/**
* Now start parsing our specific arguments
*/
/* get the remaining bits */
argv0 = strdup(argv[0]);
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
if(orte_checkpoint_globals.list_only ) {
exit_status = ORTE_SUCCESS;
goto cleanup;
}
if (0 >= argc && ORTE_JOBID_INVALID == orte_checkpoint_globals.req_hnp) {
fprintf(stderr, "%s: Nothing to do\n", argv0);
fprintf(stderr, "Type '%s --help' for usage.\n", argv0);
exit_status = 1;
goto cleanup;
}
orte_checkpoint_globals.options->term = orte_checkpoint_globals.term;
orte_checkpoint_globals.options->stop = orte_checkpoint_globals.stop;
#if OPAL_ENABLE_CRDEBUG == 1
orte_checkpoint_globals.options->attach_debugger = orte_checkpoint_globals.attach_debugger;
orte_checkpoint_globals.options->detach_debugger = orte_checkpoint_globals.detach_debugger;
#endif
if(orte_checkpoint_globals.verbose_level < 0 ) {
orte_checkpoint_globals.verbose_level = 0;
}
if(orte_checkpoint_globals.verbose_level > 0) {
orte_checkpoint_globals.verbose = true;
}
/*
* If the user did not supply an hnp jobid, then they must
* supply the PID of MPIRUN
*/
if(0 >= argc &&
ORTE_JOBID_INVALID != orte_checkpoint_globals.req_hnp) {
exit_status = ORTE_SUCCESS;
goto cleanup;
}
orte_checkpoint_globals.pid = atoi(argv[0]);
if ( 0 >= orte_checkpoint_globals.pid ) {
opal_show_help("help-orte-checkpoint.txt", "invalid_pid", true,
orte_checkpoint_globals.pid);
exit_status = ORTE_ERROR;
goto cleanup;
}
/*
* JJH: No wait is currently not implemented or tested
*/
if(orte_checkpoint_globals.nowait) {
orte_checkpoint_globals.nowait = false;
opal_show_help("help-orte-checkpoint.txt", "not_impl",
true,
"Disconnected checkpoint");
}
if(orte_checkpoint_globals.verbose) {
orte_checkpoint_globals.status = true;
}
cleanup:
if (NULL != argv0) {
free(argv0);
}
return exit_status;
}
static int parse_args(int argc, char *argv[])
{
int i, ret, len;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
char * tmp_env_var = NULL;
char *argv0 = NULL;
orte_restart_globals_t tmp = { false, /* help */
NULL, /* filename */
NULL, /* appfile */
false, /* verbose */
false, /* forked */
-1, /* seq_number */
NULL, /* hostfile */
-1, /* output*/
false, /* info only */
false, /* app only */
false, /* showme */
NULL}; /* mpirun_opts */
orte_restart_globals = tmp;
#if OPAL_ENABLE_CRDEBUG == 1
orte_restart_globals.enable_crdebug = false;
#endif
#if OPAL_ENABLE_FT_CR == 0
/* Warn and exit if not configured with Checkpoint/Restart */
{
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-restart.txt", "usage-no-cr",
true, args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
return ORTE_ERROR;
}
#endif
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, true, argc, argv);
if (OPAL_SUCCESS != ret) {
if (OPAL_ERR_SILENT != ret) {
fprintf(stderr, "%s: command line error (%s)\n", argv[0],
opal_strerror(ret));
}
return 1;
}
if (orte_restart_globals.help) {
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-orte-restart.txt", "usage", true,
args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
/* If we show the help message, that should be all we do */
exit(0);
}
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
(void) mca_base_var_env_name("opal_cr_is_tool", &tmp_env_var);
opal_setenv(tmp_env_var,
"1",
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/**
* Now start parsing our specific arguments
*/
/* get the remaining bits */
argv0 = strdup(argv[0]);
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
if (0 == argc) {
fprintf(stderr, "%s: Nothing to do\n", argv0);
fprintf(stderr, "Type '%s --help' for usge.\n", argv0);
free(argv0);
return ORTE_ERROR;
}
free(argv0);
orte_restart_globals.snapshot_ref = strdup(argv[0]);
if ( NULL == orte_restart_globals.snapshot_ref ||
0 >= strlen(orte_restart_globals.snapshot_ref) ) {
opal_show_help("help-orte-restart.txt", "invalid_filename", true,
"<none provided>");
return ORTE_ERROR;
}
/* If we have arguments after the command, then assume they
* need to be grouped together.
*/
if(argc > 1) {
orte_restart_globals.snapshot_ref = strdup(opal_argv_join(argv, ' '));
}
return ORTE_SUCCESS;
}
static int parse_args(int argc, char *argv[]) {
int i, ret, len;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
char * tmp_env_var = NULL;
char *argv0 = NULL;
memset(&opal_checkpoint_globals, 0, sizeof(opal_checkpoint_globals_t));
opal_checkpoint_globals.snapshot_name = NULL;
opal_checkpoint_globals.snapshot_loc = NULL;
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, true, false, argc, argv);
if (OPAL_SUCCESS != ret) {
if (OPAL_ERR_SILENT != ret) {
fprintf(stderr, "%s: command line error (%s)\n", argv[0],
opal_strerror(ret));
}
return 1;
}
if (opal_checkpoint_globals.help) {
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-opal-checkpoint.txt", "usage", true,
args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
/* If we show the help message, that should be all we do */
exit(0);
}
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
(void) mca_base_var_env_name("opal_cr_is_tool", &tmp_env_var);
opal_setenv(tmp_env_var,
"1",
true, &environ);
free(tmp_env_var);
tmp_env_var = NULL;
/**
* Now start parsing our specific arguments
*/
if( NULL == opal_checkpoint_globals.snapshot_name )
opal_checkpoint_globals.snapshot_name = strdup("");
if( NULL == opal_checkpoint_globals.snapshot_loc ) {
opal_checkpoint_globals.snapshot_loc = strdup("");
}
/* get the remaining bits */
argv0 = strdup(argv[0]);
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
if (0 == argc) {
fprintf(stderr, "%s: Nothing to do\n", argv0);
fprintf(stderr, "Type '%s --help' for usage.\n", argv0);
free(argv0);
return OPAL_ERROR;
}
free(argv0);
opal_checkpoint_globals.pid = atoi(argv[0]);
if ( 0 >= opal_checkpoint_globals.pid ) {
opal_show_help("help-opal-checkpoint.txt", "invalid_pid", true,
opal_checkpoint_globals.pid);
return OPAL_ERROR;
}
return OPAL_SUCCESS;
}
static int parse_args(int argc, char *argv[])
{
int i, ret, len;
opal_cmd_line_t cmd_line;
char **app_env = NULL, **global_env = NULL;
opal_restart_globals.help = false;
opal_restart_globals.verbose = false;
opal_restart_globals.snapshot_ref = NULL;
opal_restart_globals.snapshot_loc = NULL;
opal_restart_globals.snapshot_metadata = NULL;
opal_restart_globals.snapshot_cache = NULL;
opal_restart_globals.snapshot_compress = NULL;
opal_restart_globals.snapshot_compress_postfix = NULL;
opal_restart_globals.output = 0;
/* Parse the command line options */
opal_cmd_line_create(&cmd_line, cmd_line_opts);
mca_base_open();
mca_base_cmd_line_setup(&cmd_line);
ret = opal_cmd_line_parse(&cmd_line, false, false, argc, argv);
if (OPAL_SUCCESS != ret) {
if (OPAL_ERR_SILENT != ret) {
fprintf(stderr, "%s: command line error (%s)\n", argv[0],
opal_strerror(ret));
}
return 1;
}
if (opal_restart_globals.help ) {
char *str, *args = NULL;
args = opal_cmd_line_get_usage_msg(&cmd_line);
str = opal_show_help_string("help-opal-restart.txt", "usage", true,
args);
if (NULL != str) {
printf("%s", str);
free(str);
}
free(args);
/* If we show the help message, that should be all we do */
exit(0);
}
/**
* Put all of the MCA arguments in the environment
*/
mca_base_cmd_line_process_args(&cmd_line, &app_env, &global_env);
len = opal_argv_count(app_env);
for(i = 0; i < len; ++i) {
putenv(app_env[i]);
}
len = opal_argv_count(global_env);
for(i = 0; i < len; ++i) {
putenv(global_env[i]);
}
/**
* Now start parsing our specific arguments
*/
/* get the remaining bits */
opal_cmd_line_get_tail(&cmd_line, &argc, &argv);
if ( NULL == opal_restart_globals.snapshot_ref ||
0 >= strlen(opal_restart_globals.snapshot_ref) ) {
opal_show_help("help-opal-restart.txt", "invalid_filename", true,
"<none provided>");
return OPAL_ERROR;
}
/* If we have arguments after the command, then assume they
* need to be grouped together.
* Useful in the 'mca crs self' instance.
*/
if(argc > 0) {
opal_restart_globals.snapshot_ref = strdup(opal_argv_join(argv, ' '));
}
return OPAL_SUCCESS;
}