Skip to content

Resolve a race condition for setting working directory #3221

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Apr 25, 2017
Merged

Resolve a race condition for setting working directory #3221

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
143 changes: 62 additions & 81 deletions orte/mca/odls/alps/odls_alps_module.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -144,11 +144,7 @@ static int orte_odls_alps_restart_proc(orte_proc_t *child);
static void send_error_show_help(int fd, int exit_status,
const char *file, const char *topic, ...)
__opal_attribute_noreturn__;
static int do_child(orte_proc_t *child,
char *app, char **argv,
char **environ_copy,
orte_job_t *jobdat, int write_fd,
orte_iof_base_io_conf_t opts)
static int do_child(orte_odls_spawn_caddy_t *cd, int write_fd)
__opal_attribute_noreturn__;


Expand Down Expand Up @@ -342,20 +338,15 @@ static int close_open_file_descriptors(int write_fd, orte_iof_base_io_conf_t opt
return ORTE_SUCCESS;
}

static int do_child( orte_proc_t *child,
char *app, char **argv,
char **environ_copy,
orte_job_t *jobdat, int write_fd,
orte_iof_base_io_conf_t opts)
static int do_child(orte_odls_spawn_caddy_t *cd, int write_fd)
{
int i, rc;
int i;
sigset_t sigs;
char *param, *msg;

/* Setup the pipe to be close-on-exec */
opal_fd_set_cloexec(write_fd);

if (NULL != child) {
if (NULL != cd->child) {
/* setup stdout/stderr so that any error messages that we
may print out will get displayed back at orterun.

Expand All @@ -369,20 +360,19 @@ static int do_child( orte_proc_t *child,
always outputs a nice, single message indicating what
happened
*/
if (ORTE_SUCCESS != (i = orte_iof_base_setup_child(&opts,
&environ_copy))) {
if (ORTE_SUCCESS != (i = orte_iof_base_setup_child(&cd->opts, &cd->env))) {
ORTE_ERROR_LOG(i);
send_error_show_help(write_fd, 1,
"help-orte-odls-alps.txt",
"iof setup failed",
orte_process_info.nodename, app);
orte_process_info.nodename, cd->app->app);
/* Does not return */
}

/* now set any child-level controls such as binding */
orte_rtc.set(jobdat, child, &environ_copy, write_fd);
orte_rtc.set(cd->jdata, cd->child, &cd->env, write_fd);

} else if (!ORTE_FLAG_TEST(jobdat, ORTE_JOB_FLAG_FORWARD_OUTPUT)) {
} else if (!ORTE_FLAG_TEST(cd->jdata, ORTE_JOB_FLAG_FORWARD_OUTPUT)) {
/* tie stdin/out/err/internal to /dev/null */
int fdnull;
for (i=0; i < 3; i++) {
Expand All @@ -393,24 +383,24 @@ static int do_child( orte_proc_t *child,
close(fdnull);
}
fdnull = open("/dev/null", O_RDONLY, 0);
if (fdnull > opts.p_internal[1]) {
dup2(fdnull, opts.p_internal[1]);
if (fdnull > cd->opts.p_internal[1]) {
dup2(fdnull, cd->opts.p_internal[1]);
}
close(fdnull);
}

if (ORTE_SUCCESS != close_open_file_descriptors(write_fd, opts)) {
if (ORTE_SUCCESS != close_open_file_descriptors(write_fd, cd->opts)) {
send_error_show_help(write_fd, 1, "help-orte-odls-alps.txt",
"close fds",
orte_process_info.nodename, app,
orte_process_info.nodename, cd->app->app,
__FILE__, __LINE__);
}


if (argv == NULL) {
argv = malloc(sizeof(char*)*2);
argv[0] = strdup(app);
argv[1] = NULL;
if (cd->argv == NULL) {
cd->argv = malloc(sizeof(char*)*2);
cd->argv[0] = strdup(cd->app->app);
cd->argv[1] = NULL;
}

/* Set signal handlers back to the default. Do this close to
Expand All @@ -437,37 +427,33 @@ static int do_child( orte_proc_t *child,

if (10 < opal_output_get_verbosity(orte_odls_base_framework.framework_output)) {
int jout;
opal_output(0, "%s STARTING %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), app);
for (jout=0; NULL != argv[jout]; jout++) {
opal_output(0, "%s\tARGV[%d]: %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), jout, argv[jout]);
opal_output(0, "%s STARTING %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), cd->app->app);
for (jout=0; NULL != cd->argv[jout]; jout++) {
opal_output(0, "%s\tARGV[%d]: %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), jout, cd->argv[jout]);
}
for (jout=0; NULL != environ_copy[jout]; jout++) {
opal_output(0, "%s\tENVIRON[%d]: %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), jout, environ_copy[jout]);
for (jout=0; NULL != cd->env[jout]; jout++) {
opal_output(0, "%s\tENVIRON[%d]: %s", ORTE_NAME_PRINT(ORTE_PROC_MY_NAME), jout, cd->env[jout]);
}
}

execve(app, argv, environ_copy);
execve(cd->app->app, cd->argv, cd->env);
send_error_show_help(write_fd, 1,
"help-orte-odls-alps.txt", "execve error",
orte_process_info.nodename, app, strerror(errno));
orte_process_info.nodename, cd->app->app, strerror(errno));
/* Does not return */
}


static int do_parent(orte_proc_t *child,
char *app, char **argv,
char **environ_copy,
orte_job_t *jobdat, int read_fd,
orte_iof_base_io_conf_t opts)
static int do_parent(orte_odls_spawn_caddy_t *cd, int read_fd)
{
int rc;
orte_odls_pipe_err_msg_t msg;
char file[ORTE_ODLS_MAX_FILE_LEN + 1], topic[ORTE_ODLS_MAX_TOPIC_LEN + 1], *str = NULL;

close(opts.p_stdin[0]);
close(opts.p_stdout[1]);
close(opts.p_stderr[1]);
close(opts.p_internal[1]);
close(cd->opts.p_stdin[0]);
close(cd->opts.p_stdout[1]);
close(cd->opts.p_stderr[1]);
close(cd->opts.p_internal[1]);

/* Block reading a message from the pipe */
while (1) {
Expand All @@ -483,18 +469,18 @@ static int do_parent(orte_proc_t *child,
ORTE_ERROR_LOG(rc);
close(read_fd);

if (NULL != child) {
child->state = ORTE_PROC_STATE_UNDEF;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_UNDEF;
}
return rc;
}

/* Otherwise, we got a warning or error message from the child */
if (NULL != child) {
if (NULL != cd->child) {
if (msg.fatal) {
ORTE_FLAG_UNSET(child, ORTE_PROC_FLAG_ALIVE);
ORTE_FLAG_UNSET(cd->child, ORTE_PROC_FLAG_ALIVE);
} else {
ORTE_FLAG_SET(child, ORTE_PROC_FLAG_ALIVE);
ORTE_FLAG_SET(cd->child, ORTE_PROC_FLAG_ALIVE);
}
}

Expand All @@ -504,10 +490,10 @@ static int do_parent(orte_proc_t *child,
if (OPAL_SUCCESS != rc) {
orte_show_help("help-orte-odls-alps.txt", "syscall fail",
true,
orte_process_info.nodename, app,
orte_process_info.nodename, cd->app,
"opal_fd_read", __FILE__, __LINE__);
if (NULL != child) {
child->state = ORTE_PROC_STATE_UNDEF;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_UNDEF;
}
return rc;
}
Expand All @@ -518,10 +504,10 @@ static int do_parent(orte_proc_t *child,
if (OPAL_SUCCESS != rc) {
orte_show_help("help-orte-odls-alps.txt", "syscall fail",
true,
orte_process_info.nodename, app,
orte_process_info.nodename, cd->app,
"opal_fd_read", __FILE__, __LINE__);
if (NULL != child) {
child->state = ORTE_PROC_STATE_UNDEF;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_UNDEF;
}
return rc;
}
Expand All @@ -532,10 +518,10 @@ static int do_parent(orte_proc_t *child,
if (NULL == str) {
orte_show_help("help-orte-odls-alps.txt", "syscall fail",
true,
orte_process_info.nodename, app,
orte_process_info.nodename, cd->app,
"opal_fd_read", __FILE__, __LINE__);
if (NULL != child) {
child->state = ORTE_PROC_STATE_UNDEF;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_UNDEF;
}
return rc;
}
Expand All @@ -556,9 +542,9 @@ static int do_parent(orte_proc_t *child,
closed, indicating that the child launched
successfully). */
if (msg.fatal) {
if (NULL != child) {
child->state = ORTE_PROC_STATE_FAILED_TO_START;
ORTE_FLAG_UNSET(child, ORTE_PROC_FLAG_ALIVE);
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_FAILED_TO_START;
ORTE_FLAG_UNSET(cd->child, ORTE_PROC_FLAG_ALIVE);
}
close(read_fd);
return ORTE_ERR_FAILED_TO_START;
Expand All @@ -568,9 +554,9 @@ static int do_parent(orte_proc_t *child,
/* If we got here, it means that the pipe closed without
indication of a fatal error, meaning that the child process
launched successfully. */
if (NULL != child) {
child->state = ORTE_PROC_STATE_RUNNING;
ORTE_FLAG_SET(child, ORTE_PROC_FLAG_ALIVE);
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_RUNNING;
ORTE_FLAG_SET(cd->child, ORTE_PROC_FLAG_ALIVE);
}
close(read_fd);

Expand All @@ -581,14 +567,10 @@ static int do_parent(orte_proc_t *child,
/**
* Fork/exec the specified processes
*/
static int odls_alps_fork_local_proc(orte_proc_t *child,
char *app,
char **argv,
char **environ_copy,
orte_job_t *jobdat,
orte_iof_base_io_conf_t opts)
static int odls_alps_fork_local_proc(void *cdptr)
{
int rc, p[2];
orte_odls_spawn_caddy_t *cd = (orte_odls_spawn_caddy_t*)cdptr;
int p[2];
pid_t pid;

/* A pipe is used to communicate between the parent and child to
Expand All @@ -601,24 +583,24 @@ static int odls_alps_fork_local_proc(orte_proc_t *child,
the pipe, then the child was letting us know why it failed. */
if (pipe(p) < 0) {
ORTE_ERROR_LOG(ORTE_ERR_SYS_LIMITS_PIPES);
if (NULL != child) {
child->state = ORTE_PROC_STATE_FAILED_TO_START;
child->exit_code = ORTE_ERR_SYS_LIMITS_PIPES;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_FAILED_TO_START;
cd->child->exit_code = ORTE_ERR_SYS_LIMITS_PIPES;
}
return ORTE_ERR_SYS_LIMITS_PIPES;
}

/* Fork off the child */
pid = fork();
if (NULL != child) {
child->pid = pid;
if (NULL != cd->child) {
cd->child->pid = pid;
}

if (pid < 0) {
ORTE_ERROR_LOG(ORTE_ERR_SYS_LIMITS_CHILDREN);
if (NULL != child) {
child->state = ORTE_PROC_STATE_FAILED_TO_START;
child->exit_code = ORTE_ERR_SYS_LIMITS_CHILDREN;
if (NULL != cd->child) {
cd->child->state = ORTE_PROC_STATE_FAILED_TO_START;
cd->child->exit_code = ORTE_ERR_SYS_LIMITS_CHILDREN;
}
return ORTE_ERR_SYS_LIMITS_CHILDREN;
}
Expand All @@ -628,12 +610,12 @@ static int odls_alps_fork_local_proc(orte_proc_t *child,
#if HAVE_SETPGID
setpgid(0, 0);
#endif
do_child(child, app, argv, environ_copy, jobdat, p[1], opts);
do_child(cd, p[1]);
/* Does not return */
}

close(p[1]);
return do_parent(child, app, argv, environ_copy, jobdat, p[0], opts);
return do_parent(cd, p[0]);
}


Expand All @@ -643,8 +625,8 @@ static int odls_alps_fork_local_proc(orte_proc_t *child,

int orte_odls_alps_launch_local_procs(opal_buffer_t *data)
{
int rc;
orte_jobid_t job;
int rc;

/* construct the list of children we are to launch */
if (ORTE_SUCCESS != (rc = orte_odls_base_default_construct_child_list(data, &job))) {
Expand Down Expand Up @@ -729,4 +711,3 @@ static int orte_odls_alps_restart_proc(orte_proc_t *child)
}
return rc;
}

Loading