*/
#define _GNU_SOURCE
-#include <sys/syscall.h>
-#include <sys/mman.h>
-#include <sys/utsname.h>
-#include <sys/types.h>
-#include <sys/syscall.h>
-#include <sys/types.h>
#include <sys/mount.h>
#include <sys/prctl.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <unistd.h>
-#include <values.h>
#include <errno.h>
-#include <stdio.h>
#include <string.h>
#include <sys/stat.h>
#include <fcntl.h>
-#include <syslog.h>
#include <libgen.h>
-#include <glob.h>
-#include <elf.h>
#include <sched.h>
+#include <linux/limits.h>
+#include <signal.h>
+#include "capabilities.h"
#include "elf.h"
+#include "fs.h"
+#include "jail.h"
+#include "log.h"
-#include <libubox/utils.h>
-#include <libubox/list.h>
#include <libubox/uloop.h>
#define STACK_SIZE (1024 * 1024)
-#define OPT_ARGS "P:S:n:r:w:psuldo"
-
-struct extra {
- struct list_head list;
-
- const char *path;
- const char *name;
- int readonly;
-};
-
-static LIST_HEAD(extras);
+#define OPT_ARGS "S:C:n:h:r:w:d:psuloc"
+
+static struct {
+ char *name;
+ char *hostname;
+ char **jail_argv;
+ char *seccomp;
+ char *capabilities;
+ int no_new_privs;
+ int namespace;
+ int procfs;
+ int ronly;
+ int sysfs;
+} opts;
extern int pivot_root(const char *new_root, const char *put_old);
return 0;
if (ret)
- ERROR("mkdir failed on %s: %s\n", dir, strerror(errno));
+ ERROR("mkdir(%s, %d) failed: %s\n", dir, mask, strerror(errno));
return ret;
}
-static int mount_bind(const char *root, const char *path, const char *name, int readonly, int error)
+int mount_bind(const char *root, const char *path, int readonly, int error)
{
- const char *p = path;
struct stat s;
- char old[256];
- char new[256];
+ char new[PATH_MAX];
int fd;
- if (strstr(p, "local"))
- p = "/lib";
-
- snprintf(old, sizeof(old), "%s/%s", path, name);
- snprintf(new, sizeof(new), "%s%s", root, p);
-
- mkdir_p(new, 0755);
-
- snprintf(new, sizeof(new), "%s%s/%s", root, p, name);
-
- if (stat(old, &s)) {
- ERROR("%s does not exist\n", old);
+ if (stat(path, &s)) {
+ ERROR("stat(%s) failed: %s\n", path, strerror(errno));
return error;
}
+ snprintf(new, sizeof(new), "%s%s", root, path);
if (S_ISDIR(s.st_mode)) {
mkdir_p(new, 0755);
} else {
+ mkdir_p(dirname(new), 0755);
+ snprintf(new, sizeof(new), "%s%s", root, path);
fd = creat(new, 0644);
if (fd == -1) {
- ERROR("failed to create %s: %s\n", new, strerror(errno));
+ ERROR("creat(%s) failed: %s\n", new, strerror(errno));
return -1;
}
close(fd);
}
- if (mount(old, new, NULL, MS_BIND, NULL)) {
- ERROR("failed to mount -B %s %s: %s\n", old, new, strerror(errno));
+ if (mount(path, new, NULL, MS_BIND, NULL)) {
+ ERROR("failed to mount -B %s %s: %s\n", path, new, strerror(errno));
return -1;
}
- if (readonly && mount(old, new, NULL, MS_BIND | MS_REMOUNT | MS_RDONLY, NULL)) {
+ if (readonly && mount(NULL, new, NULL, MS_BIND | MS_REMOUNT | MS_RDONLY, NULL)) {
ERROR("failed to remount ro %s: %s\n", new, strerror(errno));
return -1;
}
- DEBUG("mount -B %s %s\n", old, new);
+ DEBUG("mount -B %s %s (%s)\n", path, new, readonly?"ro":"rw");
return 0;
}
-static int build_jail(const char *path)
+static int build_jail_fs(void)
{
- struct library *l;
- struct extra *m;
- int ret = 0;
+ char jail_root[] = "/tmp/ujail-XXXXXX";
+ if (mkdtemp(jail_root) == NULL) {
+ ERROR("mkdtemp(%s) failed: %s\n", jail_root, strerror(errno));
+ return -1;
+ }
- mkdir(path, 0755);
+ /* oldroot can't be MS_SHARED else pivot_root() fails */
+ if (mount("none", "/", NULL, MS_REC|MS_PRIVATE, NULL)) {
+ ERROR("private mount failed %s\n", strerror(errno));
+ return -1;
+ }
- if (mount("tmpfs", path, "tmpfs", MS_NOATIME, "mode=0744")) {
+ if (mount("tmpfs", jail_root, "tmpfs", MS_NOATIME, "mode=0755")) {
ERROR("tmpfs mount failed %s\n", strerror(errno));
return -1;
}
- avl_for_each_element(&libraries, l, avl)
- if (mount_bind(path, l->path, l->name, 1, -1))
- return -1;
-
- list_for_each_entry(m, &extras, list)
- if (mount_bind(path, m->path, m->name, m->readonly, 0))
- return -1;
-
- return ret;
-}
-
-static void _umount(const char *root, const char *path)
-{
- char *buf = NULL;
-
- if (asprintf(&buf, "%s%s", root, path) < 0) {
- ERROR("failed to alloc umount buffer: %s\n", strerror(errno));
- } else {
- DEBUG("umount %s\n", buf);
- umount(buf);
- free(buf);
+ if (chdir(jail_root)) {
+ ERROR("chdir(%s) (jail_root) failed: %s\n", jail_root, strerror(errno));
+ return -1;
}
-}
-
-static int stop_jail(const char *root)
-{
- struct library *l;
- struct extra *m;
-
- avl_for_each_element(&libraries, l, avl) {
- char path[256];
- char *p = l->path;
- if (strstr(p, "local"))
- p = "/lib";
-
- snprintf(path, sizeof(path), "%s%s/%s", root, p, l->name);
- DEBUG("umount %s\n", path);
- umount(path);
+ if (mount_all(jail_root)) {
+ ERROR("mount_all() failed\n");
+ return -1;
}
- list_for_each_entry(m, &extras, list) {
- char path[256];
+ char dirbuf[sizeof(jail_root) + 4];
+ snprintf(dirbuf, sizeof(dirbuf), "%s/old", jail_root);
+ mkdir(dirbuf, 0755);
- snprintf(path, sizeof(path), "%s%s/%s", root, m->path, m->name);
- DEBUG("umount %s\n", path);
- umount(path);
+ if (pivot_root(jail_root, dirbuf) == -1) {
+ ERROR("pivot_root(%s, %s) failed: %s\n", jail_root, dirbuf, strerror(errno));
+ return -1;
+ }
+ if (chdir("/")) {
+ ERROR("chdir(/) (after pivot_root) failed: %s\n", strerror(errno));
+ return -1;
}
- _umount(root, "/proc");
- _umount(root, "/sys");
+ snprintf(dirbuf, sizeof(dirbuf), "/old%s", jail_root);
+ rmdir(dirbuf);
+ umount2("/old", MNT_DETACH);
+ rmdir("/old");
- DEBUG("umount %s\n", root);
- umount(root);
- rmdir(root);
+ if (opts.procfs) {
+ mkdir("/proc", 0755);
+ mount("proc", "/proc", "proc", MS_NOATIME | MS_NODEV | MS_NOEXEC | MS_NOSUID, 0);
+ }
+ if (opts.sysfs) {
+ mkdir("/sys", 0755);
+ mount("sysfs", "/sys", "sysfs", MS_NOATIME | MS_NODEV | MS_NOEXEC | MS_NOSUID, 0);
+ }
+ if (opts.ronly)
+ mount(NULL, "/", NULL, MS_RDONLY | MS_REMOUNT, 0);
return 0;
}
#define MAX_ENVP 8
-static char** build_envp(const char *seccomp, int debug)
+static char** build_envp(const char *seccomp)
{
static char *envp[MAX_ENVP];
- static char preload_var[64];
- static char seccomp_var[64];
+ static char preload_var[PATH_MAX];
+ static char seccomp_var[PATH_MAX];
static char debug_var[] = "LD_DEBUG=all";
- char *preload_lib = find_lib("libpreload-seccomp.so");
+ const char *preload_lib = find_lib("libpreload-seccomp.so");
int count = 0;
if (seccomp && !preload_lib) {
snprintf(preload_var, sizeof(preload_var), "LD_PRELOAD=%s", preload_lib);
envp[count++] = preload_var;
}
- if (debug)
+ if (debug > 1)
envp[count++] = debug_var;
return envp;
}
-static int spawn(const char *path, char **argv, const char *seccomp)
-{
- pid_t pid = fork();
-
- if (pid < 0) {
- ERROR("failed to spawn %s: %s\n", *argv, strerror(errno));
- return -1;
- } else if (!pid) {
- char **envp = build_envp(seccomp, 0);
-
- INFO("spawning %s\n", *argv);
- execve(*argv, argv, envp);
- ERROR("failed to spawn child %s: %s\n", *argv, strerror(errno));
- exit(-1);
- }
-
- return pid;
-}
-
-static int usage(void)
+static void usage(void)
{
- fprintf(stderr, "jail <options> -D <binary> <params ...>\n");
- fprintf(stderr, " -P <path>\tpath where the jail will be staged\n");
- fprintf(stderr, " -S <file>\tseccomp filter\n");
+ fprintf(stderr, "ujail <options> -- <binary> <params ...>\n");
+ fprintf(stderr, " -d <num>\tshow debug log (increase num to increase verbosity)\n");
+ fprintf(stderr, " -S <file>\tseccomp filter config\n");
+ fprintf(stderr, " -C <file>\tcapabilities drop config\n");
+ fprintf(stderr, " -c\t\tset PR_SET_NO_NEW_PRIVS\n");
fprintf(stderr, " -n <name>\tthe name of the jail\n");
+ fprintf(stderr, "namespace jail options:\n");
+ fprintf(stderr, " -h <hostname>\tchange the hostname of the jail\n");
fprintf(stderr, " -r <file>\treadonly files that should be staged\n");
fprintf(stderr, " -w <file>\twriteable files that should be staged\n");
- fprintf(stderr, " -p\t\tjail has /proc\t\n");
- fprintf(stderr, " -s\t\tjail has /sys\t\n");
- fprintf(stderr, " -l\t\tjail has /dev/log\t\n");
- fprintf(stderr, " -u\t\tjail has a ubus socket\t\n");
-
- return -1;
+ fprintf(stderr, " -p\t\tjail has /proc\n");
+ fprintf(stderr, " -s\t\tjail has /sys\n");
+ fprintf(stderr, " -l\t\tjail has /dev/log\n");
+ fprintf(stderr, " -u\t\tjail has a ubus socket\n");
+ fprintf(stderr, " -o\t\tremont jail root (/) read only\n");
+ fprintf(stderr, "\nWarning: by default root inside the jail is the same\n\
+and he has the same powers as root outside the jail,\n\
+thus he can escape the jail and/or break stuff.\n\
+Please use seccomp/capabilities (-S/-C) to restrict his powers\n\n\
+If you use none of the namespace jail options,\n\
+ujail will not use namespace/build a jail,\n\
+and will only drop capabilities/apply seccomp filter.\n\n");
}
-static int child_running = 1;
-
-static void child_process_handler(struct uloop_process *c, int ret)
+static int exec_jail(void *_notused)
{
- INFO("child (%d) exited: %d\n", c->pid, ret);
- uloop_end();
- child_running = 0;
-}
-
-struct uloop_process child_process = {
- .cb = child_process_handler,
-};
+ if (opts.capabilities && drop_capabilities(opts.capabilities))
+ exit(EXIT_FAILURE);
-static int spawn_child(void *arg)
-{
- char *path = get_current_dir_name();
- int procfs = 0, sysfs = 0;
- char *seccomp = NULL;
- char **argv = arg;
- int argc = 0, ch;
- char *mpoint;
- int ronly = 0;
-
- while (argv[argc])
- argc++;
-
- optind = 0;
- while ((ch = getopt(argc, argv, OPT_ARGS)) != -1) {
- switch (ch) {
- case 'd':
- debug = 1;
- break;
- case 'S':
- seccomp = optarg;
- break;
- case 'p':
- procfs = 1;
- break;
- case 'o':
- ronly = 1;
- break;
- case 's':
- sysfs = 1;
- break;
- case 'n':
- sethostname(optarg, strlen(optarg));
- break;
- }
+ if (opts.no_new_privs && prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
+ ERROR("prctl(PR_SET_NO_NEW_PRIVS) failed: %s\n", strerror(errno));
+ exit(EXIT_FAILURE);
}
- asprintf(&mpoint, "%s/old", path);
- mkdir_p(mpoint, 0755);
- if (pivot_root(path, mpoint) == -1) {
- ERROR("pivot_root failed:%s\n", strerror(errno));
- return -1;
- }
- free(mpoint);
- umount2("/old", MNT_DETACH);
- rmdir("/old");
- if (procfs) {
- mkdir("/proc", 0755);
- mount("proc", "/proc", "proc", MS_NOATIME, 0);
- }
- if (sysfs) {
- mkdir("/sys", 0755);
- mount("sysfs", "/sys", "sysfs", MS_NOATIME, 0);
+ if (opts.namespace && opts.hostname && strlen(opts.hostname) > 0
+ && sethostname(opts.hostname, strlen(opts.hostname))) {
+ ERROR("sethostname(%s) failed: %s\n", opts.hostname, strerror(errno));
+ exit(EXIT_FAILURE);
}
- if (ronly)
- mount(NULL, "/", NULL, MS_RDONLY | MS_REMOUNT, 0);
- uloop_init();
-
- child_process.pid = spawn(path, &argv[optind], seccomp);
- uloop_process_add(&child_process);
- uloop_run();
- uloop_done();
- if (child_running) {
- kill(child_process.pid, SIGTERM);
- waitpid(child_process.pid, NULL, 0);
+ if (opts.namespace && build_jail_fs()) {
+ ERROR("failed to build jail fs\n");
+ exit(EXIT_FAILURE);
}
- return 0;
+ char **envp = build_envp(opts.seccomp);
+ if (!envp)
+ exit(EXIT_FAILURE);
+
+ INFO("exec-ing %s\n", *opts.jail_argv);
+ execve(*opts.jail_argv, opts.jail_argv, envp);
+ /* we get there only if execve fails */
+ ERROR("failed to execve %s: %s\n", *opts.jail_argv, strerror(errno));
+ exit(EXIT_FAILURE);
}
-static int namespace_running = 1;
+static int jail_running = 1;
+static int jail_return_code = 0;
-static void namespace_process_handler(struct uloop_process *c, int ret)
+static void jail_process_timeout_cb(struct uloop_timeout *t);
+static struct uloop_timeout jail_process_timeout = {
+ .cb = jail_process_timeout_cb,
+};
+
+static void jail_process_handler(struct uloop_process *c, int ret)
{
- INFO("namespace (%d) exited: %d\n", c->pid, ret);
+ uloop_timeout_cancel(&jail_process_timeout);
+ if (WIFEXITED(ret)) {
+ jail_return_code = WEXITSTATUS(ret);
+ INFO("jail (%d) exited with exit: %d\n", c->pid, jail_return_code);
+ } else {
+ jail_return_code = WTERMSIG(ret);
+ INFO("jail (%d) exited with signal: %d\n", c->pid, jail_return_code);
+ }
+ jail_running = 0;
uloop_end();
- namespace_running = 0;
}
-struct uloop_process namespace_process = {
- .cb = namespace_process_handler,
+static struct uloop_process jail_process = {
+ .cb = jail_process_handler,
};
-static void spawn_namespace(const char *path, int argc, char **argv)
+static void jail_process_timeout_cb(struct uloop_timeout *t)
{
- char *dir = get_current_dir_name();
-
- uloop_init();
- chdir(path);
- namespace_process.pid = clone(spawn_child,
- child_stack + STACK_SIZE,
- CLONE_NEWUTS | CLONE_NEWPID | CLONE_NEWNS | SIGCHLD, argv);
-
- if (namespace_process.pid != -1) {
- chdir(dir);
- free(dir);
- uloop_process_add(&namespace_process);
- uloop_run();
- uloop_done();
- if (namespace_running) {
- kill(namespace_process.pid, SIGTERM);
- waitpid(namespace_process.pid, NULL, 0);
- }
- } else {
- ERROR("failed to spawn namespace: %s\n", strerror(errno));
- }
+ DEBUG("jail process failed to stop, sending SIGKILL\n");
+ kill(jail_process.pid, SIGKILL);
}
-static void add_extra(char *name, int readonly)
+static void jail_handle_signal(int signo)
{
- struct extra *f;
-
- if (*name != '/') {
- ERROR("%s is not an absolute path\n", name);
- return;
- }
-
- f = calloc(1, sizeof(struct extra));
-
- f->name = basename(name);
- f->path = dirname(strdup(name));
- f->readonly = readonly;
-
- list_add_tail(&f->list, &extras);
+ DEBUG("forwarding signal %d to the jailed process\n", signo);
+ kill(jail_process.pid, signo);
}
int main(int argc, char **argv)
{
+ sigset_t sigmask;
uid_t uid = getuid();
- const char *name = NULL;
- char *path = NULL;
- struct stat s;
- int ch, ret;
char log[] = "/dev/log";
char ubus[] = "/var/run/ubus.sock";
+ int ch, i;
if (uid) {
ERROR("not root, aborting: %s\n", strerror(errno));
- return -1;
+ return EXIT_FAILURE;
}
umask(022);
+ mount_list_init();
+ init_library_search();
while ((ch = getopt(argc, argv, OPT_ARGS)) != -1) {
switch (ch) {
case 'd':
- debug = 1;
+ debug = atoi(optarg);
break;
- case 'P':
- path = optarg;
+ case 'p':
+ opts.namespace = 1;
+ opts.procfs = 1;
break;
- case 'n':
- name = optarg;
+ case 'o':
+ opts.namespace = 1;
+ opts.ronly = 1;
+ break;
+ case 's':
+ opts.namespace = 1;
+ opts.sysfs = 1;
break;
case 'S':
+ opts.seccomp = optarg;
+ add_mount(optarg, 1, -1);
+ break;
+ case 'C':
+ opts.capabilities = optarg;
+ break;
+ case 'c':
+ opts.no_new_privs = 1;
+ break;
+ case 'n':
+ opts.name = optarg;
+ break;
+ case 'h':
+ opts.hostname = optarg;
+ break;
case 'r':
- add_extra(optarg, 1);
+ opts.namespace = 1;
+ add_path_and_deps(optarg, 1, 0, 0);
break;
case 'w':
- add_extra(optarg, 0);
+ opts.namespace = 1;
+ add_path_and_deps(optarg, 0, 0, 0);
break;
case 'u':
- add_extra(ubus, 0);
+ opts.namespace = 1;
+ add_mount(ubus, 0, -1);
break;
case 'l':
- add_extra(log, 0);
+ opts.namespace = 1;
+ add_mount(log, 0, -1);
break;
}
}
- if (argc - optind < 1)
- return usage();
-
- if (!path && asprintf(&path, "/tmp/%s", basename(argv[optind])) == -1) {
- ERROR("failed to set root path\n: %s", strerror(errno));
- return -1;
+ /* no <binary> param found */
+ if (argc - optind < 1) {
+ usage();
+ return EXIT_FAILURE;
}
-
- if (!stat(path, &s)) {
- ERROR("%s already exists: %s\n", path, strerror(errno));
- return -1;
+ if (!(opts.namespace||opts.capabilities||opts.seccomp)) {
+ ERROR("Not using namespaces, capabilities or seccomp !!!\n\n");
+ usage();
+ return EXIT_FAILURE;
}
+ DEBUG("Using namespaces(%d), capabilities(%d), seccomp(%d)\n",
+ opts.namespace,
+ opts.capabilities != 0,
+ opts.seccomp != 0);
- if (name)
- prctl(PR_SET_NAME, name, NULL, NULL, NULL);
-
- avl_init(&libraries, avl_strcmp, false, NULL);
- alloc_library_path("/lib64");
- alloc_library_path("/lib");
- alloc_library_path("/usr/lib");
- load_ldso_conf("/etc/ld.so.conf");
+ opts.jail_argv = &argv[optind];
- if (elf_load_deps(argv[optind])) {
+ if (opts.namespace && add_path_and_deps(*opts.jail_argv, 1, -1, 0)) {
ERROR("failed to load dependencies\n");
return -1;
}
- if (elf_load_deps("libpreload-seccomp.so")) {
+ if (opts.namespace && opts.seccomp && add_path_and_deps("libpreload-seccomp.so", 1, -1, 1)) {
ERROR("failed to load libpreload-seccomp.so\n");
return -1;
}
- ret = build_jail(path);
+ if (opts.name)
+ prctl(PR_SET_NAME, opts.name, NULL, NULL, NULL);
- if (!ret)
- spawn_namespace(path, argc, argv);
- else
- ERROR("failed to build jail\n");
+ uloop_init();
- stop_jail(path);
+ sigfillset(&sigmask);
+ for (i = 0; i < _NSIG; i++) {
+ struct sigaction s = { 0 };
- return ret;
+ if (!sigismember(&sigmask, i))
+ continue;
+ if ((i == SIGCHLD) || (i == SIGPIPE))
+ continue;
+
+ s.sa_handler = jail_handle_signal;
+ sigaction(i, &s, NULL);
+ }
+
+ if (opts.namespace) {
+ add_mount("/dev/full", 0, -1);
+ add_mount("/dev/null", 0, -1);
+ add_mount("/dev/urandom", 0, -1);
+ add_mount("/dev/zero", 0, -1);
+
+ int flags = CLONE_NEWPID | CLONE_NEWNS | CLONE_NEWIPC | SIGCHLD;
+ if (opts.hostname)
+ flags |= CLONE_NEWUTS;
+ jail_process.pid = clone(exec_jail, child_stack + STACK_SIZE, flags, NULL);
+ } else {
+ jail_process.pid = fork();
+ }
+
+ if (jail_process.pid > 0) {
+ /* parent process */
+ uloop_process_add(&jail_process);
+ uloop_run();
+ if (jail_running) {
+ DEBUG("uloop interrupted, killing jail process\n");
+ kill(jail_process.pid, SIGTERM);
+ uloop_timeout_set(&jail_process_timeout, 1000);
+ uloop_run();
+ }
+ uloop_done();
+ return jail_return_code;
+ } else if (jail_process.pid == 0) {
+ /* fork child process */
+ return exec_jail(NULL);
+ } else {
+ ERROR("failed to clone/fork: %s\n", strerror(errno));
+ return EXIT_FAILURE;
+ }
}
projects / project / procd.git / blobdiff