Copyright 2010 Lennart Poettering
systemd 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
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 of the License, or
(at your option) any later version.
systemd 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.
+ Lesser General Public License for more details.
- You should have received a copy of the GNU General Public License
+ You should have received a copy of the GNU Lesser General Public License
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
#include <termios.h>
#include <sys/signalfd.h>
#include <grp.h>
+#include <linux/fs.h>
+
+#include <systemd/sd-daemon.h>
#include "log.h"
#include "util.h"
+#include "mkdir.h"
+#include "audit.h"
#include "missing.h"
#include "cgroup-util.h"
-#include "sd-daemon.h"
#include "strv.h"
+#include "loopback-setup.h"
static char *arg_directory = NULL;
static char *arg_user = NULL;
+static char **arg_controllers = NULL;
+static bool arg_private_network = false;
static int help(void) {
printf("%s [OPTIONS...] [PATH] [ARGUMENTS...]\n\n"
"Spawn a minimal namespace container for debugging, testing and building.\n\n"
- " -h --help Show this help\n"
- " -D --directory=NAME Root directory for the container\n"
- " -u --user=USER Run the command under specified user or uid\n",
+ " -h --help Show this help\n"
+ " -D --directory=NAME Root directory for the container\n"
+ " -u --user=USER Run the command under specified user or uid\n"
+ " -C --controllers=LIST Put the container in specified comma-separated cgroup hierarchies\n"
+ " --private-network Disable network in container\n",
program_invocation_short_name);
return 0;
static int parse_argv(int argc, char *argv[]) {
+ enum {
+ ARG_PRIVATE_NETWORK = 0x100
+ };
+
static const struct option options[] = {
- { "help", no_argument, NULL, 'h' },
- { "directory", required_argument, NULL, 'D' },
- { "user", optional_argument, NULL, 'u' },
- { NULL, 0, NULL, 0 }
+ { "help", no_argument, NULL, 'h' },
+ { "directory", required_argument, NULL, 'D' },
+ { "user", required_argument, NULL, 'u' },
+ { "controllers", required_argument, NULL, 'C' },
+ { "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
+ { NULL, 0, NULL, 0 }
};
int c;
assert(argc >= 0);
assert(argv);
- while ((c = getopt_long(argc, argv, "+hD:u:", options, NULL)) >= 0) {
+ while ((c = getopt_long(argc, argv, "+hD:u:C:", options, NULL)) >= 0) {
switch (c) {
break;
+ case 'C':
+ strv_free(arg_controllers);
+ arg_controllers = strv_split(optarg, ",");
+ if (!arg_controllers) {
+ log_error("Failed to split controllers list.");
+ return -ENOMEM;
+ }
+ strv_uniq(arg_controllers);
+
+ break;
+
+ case ARG_PRIVATE_NETWORK:
+ arg_private_network = true;
+ break;
+
case '?':
return -EINVAL;
} MountPoint;
static const MountPoint mount_table[] = {
- { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
- { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
- { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
- { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, true },
- { "tmpfs", "/dev", "tmpfs", "mode=755", MS_NOSUID, true },
- { "/dev/pts", "/dev/pts", "bind", NULL, MS_BIND, true },
- { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV, true },
+ { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
+ { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
+ { "/proc/sys", "/proc/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
+ { "/sys", "/sys", "bind", NULL, MS_BIND, true }, /* Bind mount first */
+ { "/sys", "/sys", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, true }, /* Then, make it r/o */
+ { "tmpfs", "/dev", "tmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME, true },
+ { "/dev/pts", "/dev/pts", "bind", NULL, MS_BIND, true },
+ { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true },
#ifdef HAVE_SELINUX
- { "selinux", "/selinux", "selinuxfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, false },
+ { "/sys/fs/selinux", "/sys/fs/selinux", "bind", NULL, MS_BIND, false }, /* Bind mount first */
+ { "/sys/fs/selinux", "/sys/fs/selinux", "bind", NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, false }, /* Then, make it r/o */
#endif
};
break;
}
- if ((t = path_is_mount_point(where)) < 0) {
+ t = path_is_mount_point(where, false);
+ if (t < 0) {
log_error("Failed to detect whether %s is a mount point: %s", where, strerror(-t));
free(where);
}
/* Fix the timezone, if possible */
- if (asprintf(&where, "%s/%s", dest, "/etc/localtime") >= 0) {
- mount("/etc/localtime", where, "bind", MS_BIND, NULL);
- mount("/etc/localtime", where, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+ if (asprintf(&where, "%s/etc/localtime", dest) >= 0) {
+
+ if (mount("/etc/localtime", where, "bind", MS_BIND, NULL) >= 0)
+ mount("/etc/localtime", where, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+
+ free(where);
+ }
+
+ if (asprintf(&where, "%s/etc/timezone", dest) >= 0) {
+
+ if (mount("/etc/timezone", where, "bind", MS_BIND, NULL) >= 0)
+ mount("/etc/timezone", where, "bind", MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+
+ free(where);
+ }
+
+ if (asprintf(&where, "%s/proc/kmsg", dest) >= 0) {
+ mount("/dev/null", where, "bind", MS_BIND, NULL);
free(where);
}
}
finish:
-
umask(u);
return r;
unsigned long l;
- for (l = 0; l <= MAX(63LU, (unsigned long) CAP_LAST_CAP); l++) {
+ for (l = 0; l <= cap_last_cap(); l++) {
unsigned i;
for (i = 0; i < ELEMENTSOF(retain); i++)
continue;
if (prctl(PR_CAPBSET_DROP, l) < 0) {
-
- /* If this capability is not known, EINVAL
- * will be returned, let's ignore this. */
- if (errno == EINVAL)
- break;
-
log_error("PR_CAPBSET_DROP failed: %m");
return -errno;
}
return r < 0 ? 0 : 1;
}
-#define BUFFER_SIZE 1024
-
static int process_pty(int master, sigset_t *mask) {
- char in_buffer[BUFFER_SIZE], out_buffer[BUFFER_SIZE];
+ char in_buffer[LINE_MAX], out_buffer[LINE_MAX];
size_t in_buffer_full = 0, out_buffer_full = 0;
struct epoll_event stdin_ev, stdout_ev, master_ev, signal_ev;
bool stdin_readable = false, stdout_writable = false, master_readable = false, master_writable = false;
(master_readable && out_buffer_full <= 0) ||
(stdout_writable && out_buffer_full > 0)) {
- if (stdin_readable && in_buffer_full < BUFFER_SIZE) {
+ if (stdin_readable && in_buffer_full < LINE_MAX) {
- if ((k = read(STDIN_FILENO, in_buffer + in_buffer_full, BUFFER_SIZE - in_buffer_full)) < 0) {
+ if ((k = read(STDIN_FILENO, in_buffer + in_buffer_full, LINE_MAX - in_buffer_full)) < 0) {
if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
stdin_readable = false;
}
}
- if (master_readable && out_buffer_full < BUFFER_SIZE) {
+ if (master_readable && out_buffer_full < LINE_MAX) {
- if ((k = read(master, out_buffer + out_buffer_full, BUFFER_SIZE - out_buffer_full)) < 0) {
+ if ((k = read(master, out_buffer + out_buffer_full, LINE_MAX - out_buffer_full)) < 0) {
if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
master_readable = false;
pid_t pid = 0;
int r = EXIT_FAILURE, k;
char *oldcg = NULL, *newcg = NULL;
+ char **controller = NULL;
int master = -1;
const char *console = NULL;
struct termios saved_attr, raw_attr;
goto finish;
}
- if ((k = cg_create_and_attach(SYSTEMD_CGROUP_CONTROLLER, newcg, 0)) < 0) {
+ k = cg_create_and_attach(SYSTEMD_CGROUP_CONTROLLER, newcg, 0);
+ if (k < 0) {
log_error("Failed to create cgroup: %s", strerror(-k));
goto finish;
}
+ STRV_FOREACH(controller,arg_controllers) {
+ k = cg_create_and_attach(*controller, newcg, 0);
+ if (k < 0)
+ log_warning("Failed to create cgroup in controller %s: %s", *controller, strerror(-k));
+ }
+
if ((master = posix_openpt(O_RDWR|O_NOCTTY|O_CLOEXEC|O_NDELAY)) < 0) {
log_error("Failed to acquire pseudo tty: %m");
goto finish;
sigset_add_many(&mask, SIGCHLD, SIGWINCH, SIGTERM, SIGINT, -1);
assert_se(sigprocmask(SIG_BLOCK, &mask, NULL) == 0);
- if ((pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS, NULL)) < 0) {
- log_error("clone() failed: %m");
+ pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL);
+ if (pid < 0) {
+ if (errno == EINVAL)
+ log_error("clone() failed, do you have namespace support enabled in your kernel? (You need UTS, IPC, PID and NET namespacing built in): %m");
+ else
+ log_error("clone() failed: %m");
+
goto finish;
}
gid_t gid = (gid_t) -1;
const char *envp[] = {
"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+ "container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */
NULL, /* TERM */
NULL, /* HOME */
NULL, /* USER */
NULL
};
- envp[1] = strv_find_prefix(environ, "TERM=");
+ envp[2] = strv_find_prefix(environ, "TERM=");
close_nointr_nofail(master);
if (prctl(PR_SET_PDEATHSIG, SIGKILL) < 0)
goto child_fail;
+ /* Mark / as private, in case somebody marked it shared */
+ if (mount(NULL, "/", NULL, MS_PRIVATE|MS_REC, NULL) < 0)
+ goto child_fail;
+
if (mount_all(arg_directory) < 0)
goto child_fail;
goto child_fail;
}
- umask(0002);
+ umask(0022);
+
+ loopback_setup();
if (drop_capabilities() < 0)
goto child_fail;
goto child_fail;
}
- if (setregid(gid, gid) < 0) {
+ if (setresgid(gid, gid, gid) < 0) {
log_error("setregid() failed: %m");
goto child_fail;
}
- if (setreuid(uid, uid) < 0) {
+ if (setresuid(uid, uid, uid) < 0) {
log_error("setreuid() failed: %m");
goto child_fail;
}
}
- if ((asprintf((char**)(envp + 2), "HOME=%s", home? home: "/root") < 0) ||
- (asprintf((char**)(envp + 3), "USER=%s", arg_user? arg_user : "root") < 0) ||
- (asprintf((char**)(envp + 4), "LOGNAME=%s", arg_user? arg_user : "root") < 0)) {
- log_error("environment setup failed: %m");
+ if ((asprintf((char**)(envp + 3), "HOME=%s", home? home: "/root") < 0) ||
+ (asprintf((char**)(envp + 4), "USER=%s", arg_user? arg_user : "root") < 0) ||
+ (asprintf((char**)(envp + 5), "LOGNAME=%s", arg_user? arg_user : "root") < 0)) {
+ log_error("Out of memory");
goto child_fail;
}
if (argc > optind)
execvpe(argv[optind], argv + optind, (char**) envp);
else {
- chdir(home? home : "/root");
+ chdir(home ? home : "/root");
execle("/bin/bash", "-bash", NULL, (char**) envp);
}
cg_kill_recursive_and_wait(SYSTEMD_CGROUP_CONTROLLER, newcg, true);
free(arg_directory);
+ strv_free(arg_controllers);
free(oldcg);
free(newcg);
~ianmdlvl / elogind.git / blobdiff