src/plugins/proctrack/linuxproc/kill_tree.c

/*****************************************************************************\
 *  kill_tree.c - Kill process tree based upon process IDs
 *	Used primarily for MPICH-GM
 *****************************************************************************
 *  Copyright (C) 2004 The Regents of the University of California.
 *  Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
 *  Written by Takao Hatazaki <takao.hatazaki@hp.com>
 *  CODE-OCEC-09-009. All rights reserved.
 *
 *  This file is part of Slurm, a resource management program.
 *  For details, see <https://slurm.schedmd.com/>.
 *  Please also read the included file: DISCLAIMER.
 *
 *  Slurm is free software; you can redistribute it and/or modify it under
 *  the terms of the GNU General Public License as published by the Free
 *  Software Foundation; either version 2 of the License, or (at your option)
 *  any later version.
 *
 *  In addition, as a special exception, the copyright holders give permission
 *  to link the code of portions of this program with the OpenSSL library under
 *  certain conditions as described in each individual source file, and
 *  distribute linked combinations including the two. You must obey the GNU
 *  General Public License in all respects for all of the code used other than
 *  OpenSSL. If you modify file(s) with this exception, you may extend this
 *  exception to your version of the file(s), but you are not obligated to do
 *  so. If you do not wish to do so, delete this exception statement from your
 *  version.  If you delete this exception statement from all source files in
 *  the program, then also delete it here.
 *
 *  Slurm is distributed in the hope that it will be useful, but WITHOUT ANY
 *  WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
 *  FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
 *  details.
 *
 *  You should have received a copy of the GNU General Public License along
 *  with Slurm; if not, write to the Free Software Foundation, Inc.,
 *  51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA.
\*****************************************************************************/

#include <dirent.h>
#include <fcntl.h>
#include <limits.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>

#include "slurm/slurm.h"
#include "slurm/slurm_errno.h"
#include "src/common/log.h"
#include "src/common/xmalloc.h"
#include "src/common/xstring.h"
#include "kill_tree.h"

typedef struct xpid_s {
	pid_t pid;
	int is_usercmd;
	char *cmd;
	struct xpid_s *next;
} xpid_t;

typedef struct xppid_s {
	pid_t ppid;
	xpid_t *list;
	struct xppid_s *next;
} xppid_t;

#define HASH_LEN 64

#define GET_HASH_IDX(ppid) ((ppid)%HASH_LEN)

static xpid_t *_alloc_pid(pid_t pid, int is_usercmd, char *cmd, xpid_t *next)
{
	xpid_t *new;

	new = xmalloc(sizeof(*new));
	new->pid = pid;
	new->is_usercmd = is_usercmd;
	new->cmd = xstrdup(cmd);
	new->next = next;
	return new;
}

static xppid_t *_alloc_ppid(pid_t ppid, pid_t pid, int is_usercmd, char *cmd,
			    xppid_t *next)
{
	xppid_t *new;

	new = xmalloc(sizeof(*new));
	new->ppid = ppid;
	new->list = _alloc_pid(pid, is_usercmd, cmd, NULL);
	new->next = next;
	return new;
}

static void _push_to_hashtbl(pid_t ppid, pid_t pid,
			     int is_usercmd, char *cmd, xppid_t **hashtbl)
{
	int idx;
	xppid_t *ppids, *newppid;
	xpid_t *newpid;

	idx = GET_HASH_IDX(ppid);
	ppids = hashtbl[idx];
	while (ppids) {
		if (ppids->ppid == ppid) {
			newpid = _alloc_pid(pid, is_usercmd, cmd, ppids->list);
			ppids->list = newpid;
			return;
		}
		ppids = ppids->next;
	}
	newppid = _alloc_ppid(ppid, pid, is_usercmd, cmd, hashtbl[idx]);
	hashtbl[idx] = newppid;
}

static int _get_myname(char *s)
{
	char path[PATH_MAX], *rbuf;
	ssize_t buf_used;
	int fd;

	snprintf(path, PATH_MAX, "/proc/%ld/stat", (long)getpid());
	if ((fd = open(path, O_RDONLY)) < 0) {
		error("Cannot open /proc/getpid()/stat");
		return -1;
	}
	rbuf = xmalloc(4096);
	buf_used = read(fd, rbuf, 4096);
	if ((buf_used <= 0) || (buf_used >= 4096)) {
		error("Cannot read /proc/getpid()/stat");
		xfree(rbuf);
		close(fd);
		return -1;
	}
	close(fd);
	if (sscanf(rbuf, "%*d %s ", s) != 1) {
		error("Cannot get the command name from /proc/getpid()/stat");
		xfree(rbuf);
		return -1;
	}
	xfree(rbuf);
	return 0;
}

static xppid_t **_build_hashtbl(void)
{
	DIR *dir;
	struct dirent *de;
	char path[PATH_MAX], *endptr, *num, *rbuf;
	ssize_t buf_used;
	char myname[1024], cmd[1024];
	char state;
	int fd;
	long pid, ppid, ret_l;
	xppid_t **hashtbl;

	if ((dir = opendir("/proc")) == NULL) {
		error("opendir(/proc): %m");
		return NULL;
	}
	if (_get_myname(myname) < 0)
		return NULL;
	debug3("Myname in build_hashtbl: %s", myname);

	hashtbl = xcalloc(HASH_LEN, sizeof(xppid_t *));

	errno = 0;
	rbuf = xmalloc(4096);
	while ((de = readdir(dir)) != NULL) {
		num = de->d_name;
		if ((num[0] < '0') || (num[0] > '9'))
			continue;
		ret_l = strtol(num, &endptr, 10);
		if ((ret_l == LONG_MIN) || (ret_l == LONG_MAX)) {
			error("couldn't do a strtol on str %s(%ld): %m",
			      num, ret_l);
			continue;
		}
		if (endptr == NULL || *endptr != 0)
			continue;
		snprintf(path, PATH_MAX, "/proc/%s/stat", num);
		if ((fd = open(path, O_RDONLY)) < 0) {
			continue;
		}
		buf_used = read(fd, rbuf, 4096);
		if ((buf_used <= 0) || (buf_used >= 4096)) {
			close(fd);
			continue;
		}
		close(fd);
		if (sscanf(rbuf, "%ld %s %c %ld", &pid, cmd, &state, &ppid)
		    != 4) {
			continue;
		}
		if (state == 'Z') {
			debug3("Defunct process skipped: command=%s state=%c "
			       "pid=%ld ppid=%ld", cmd, state, pid, ppid);
			continue;	/* Defunct, don't try to kill */
		}

		/* Record cmd for debugging purpose */
		_push_to_hashtbl((pid_t)ppid, (pid_t)pid,
				 xstrcmp(myname, cmd), cmd, hashtbl);
	}
	xfree(rbuf);
	closedir(dir);
	return hashtbl;
}

static void _destroy_list(xpid_t *list)
{
	xpid_t *tmp;

	while (list) {
		tmp = list->next;
		xfree(list->cmd);
		xfree(list);
		list = tmp;
	}
}

static void _destroy_hashtbl(xppid_t **hashtbl)
{
	int i;
	xppid_t *ppid, *tmp;

	for (i=0; i<HASH_LEN; i++) {
		ppid = hashtbl[i];
		while (ppid) {
			_destroy_list(ppid->list);
			tmp = ppid->next;
			xfree(ppid);
			ppid = tmp;
		}
	}
	xfree(hashtbl);
}


static xpid_t *_get_list(int top, xpid_t *list, xppid_t **hashtbl)
{
	xppid_t *ppid;
	xpid_t *children;

	ppid = hashtbl[GET_HASH_IDX(top)];
	while (ppid) {
		if (ppid->ppid == top) {
			children = ppid->list;
			while (children) {
				list = _alloc_pid(children->pid,
						  children->is_usercmd,
						  children->cmd,
						  list);
				children = children->next;
			}
			children = ppid->list;
			while (children) {
				list = _get_list(children->pid, list, hashtbl);
				children = children->next;
			}
			break;
		}
		ppid = ppid->next;
	}
	return list;
}

static int _kill_proclist(xpid_t *list, int sig)
{
	int rc;

	rc = 0;
	while (list) {
		if (list->pid > 1) {
			if (! list->is_usercmd) {
				debug2("%ld %s is not a user command.  "
				       "Skipped sending signal %d",
				       (long)list->pid, list->cmd, sig);
			} else {
				verbose("Sending signal %d to pid %d %s",
					sig, list->pid, list->cmd);
				if (kill(list->pid, sig))
					rc = errno; /* save the last error */
			}
		}
		list = list->next;
	}

	return rc;
}


/*
 * Some of processes may not be in the same process group
 * (e.g. GMPI processes).  So, find out the process tree,
 * then kill all that subtree.
 */
extern int kill_proc_tree(pid_t top, int sig)
{
	xpid_t *list;
	int rc = -1;
	xppid_t **hashtbl;

	if ((hashtbl = _build_hashtbl()) == NULL)
		return -1;

	list = _get_list(top, NULL, hashtbl);
	rc = _kill_proclist(list, sig);
	_destroy_hashtbl(hashtbl);
	_destroy_list(list);
	return rc;
}


/*
 * Return the pid of the process named "process_name"
 * which is the ancestor of "process".
 */
extern pid_t find_ancestor(pid_t process, char *process_name)
{
	char path[PATH_MAX], *rbuf;
	ssize_t buf_used;
	int fd, len;
	long pid, ppid;

	len = strlen(process_name);
	rbuf = xmalloc_nz(4097);
	pid = ppid = (long)process;
	while (1) {
		if (ppid <= 1) {
			pid = 0;
			break;
		}

		snprintf(path, PATH_MAX, "/proc/%ld/stat", ppid);
		if ((fd = open(path, O_RDONLY)) < 0) {
			pid = 0;
			break;
		}
		buf_used = read(fd, rbuf, 4096);
		if (buf_used >= 0)
			rbuf[buf_used] = '\0';
		else
			rbuf[0] = '\0';
		if ((buf_used <= 0) || (buf_used >= 4096)) {
			close(fd);
			pid = 0;
			break;
		}
		close(fd);
		if (sscanf(rbuf, "%ld %*s %*s %ld", &pid, &ppid) != 2) {
			pid = 0;
			break;
		}

		snprintf(path, PATH_MAX, "/proc/%ld/cmdline", pid);
		if ((fd = open(path, O_RDONLY)) < 0) {
			continue;
		}
		buf_used = read(fd, rbuf, 4096);
		if (buf_used >= 0)
			rbuf[buf_used] = '\0';
		else
			rbuf[0] = '\0';
		if ((buf_used <= 0) || (buf_used >= 4096)) {
			close(fd);
			continue;
		}
		close(fd);
		if (strncmp(rbuf, process_name, len) == 0)
			break;
	}
	xfree(rbuf);

	return pid;
}

/* The returned "pids" array does NOT include the slurmstepd */
extern int proctrack_linuxproc_get_pids(pid_t top, pid_t **pids, int *npids)
{
	xppid_t **hashtbl;
	xpid_t *list, *ptr;
	pid_t *p;
	int i, len = 32, rc;

	if ((hashtbl = _build_hashtbl()) == NULL)
		return SLURM_ERROR;

	list = _get_list(top, NULL, hashtbl);
	if (list == NULL) {
		*pids = NULL;
		*npids = 0;
		_destroy_hashtbl(hashtbl);
		return SLURM_ERROR;
	}

	p = xcalloc(len, sizeof(pid_t));
	ptr = list;
	i = 0;
	while (ptr != NULL) {
		if (ptr->is_usercmd) { /* don't include the slurmstepd */
			if (i >= len - 1) {
				len *= 2;
				xrealloc(p, (sizeof(pid_t) * len));
			}
			p[i] = ptr->pid;
			i++;
		}
		ptr = ptr->next;
	}

	if (i == 0) {
		xfree(p);
		*pids = NULL;
		*npids = 0;
		rc = SLURM_ERROR;
	} else {
		*pids = p;
		*npids = i;
		rc = SLURM_SUCCESS;
	}
	_destroy_hashtbl(hashtbl);
	_destroy_list(list);
	return rc;
}