#include <sys/prctl.h>
#include <sys/capability.h>
#include <getopt.h>
-#include <sys/poll.h>
-#include <sys/epoll.h>
#include <termios.h>
#include <sys/signalfd.h>
#include <grp.h>
#include <linux/fs.h>
#include <sys/un.h>
#include <sys/socket.h>
+#include <linux/netlink.h>
-#ifdef HAVE_XATTR
-#include <attr/xattr.h>
-#endif
-
-#include <systemd/sd-daemon.h>
-
+#include "sd-daemon.h"
+#include "sd-bus.h"
+#include "sd-id128.h"
#include "log.h"
#include "util.h"
#include "mkdir.h"
#include "strv.h"
#include "path-util.h"
#include "loopback-setup.h"
-#include "sd-id128.h"
#include "dev-setup.h"
#include "fdset.h"
#include "build.h"
#include "fileio.h"
+#include "bus-util.h"
+#include "bus-error.h"
+#include "ptyfwd.h"
+#include "bus-kernel.h"
+#include "env-util.h"
#ifndef TTY_GID
#define TTY_GID 5
static char *arg_directory = NULL;
static char *arg_user = NULL;
-static char **arg_controllers = NULL;
-static char *arg_uuid = NULL;
+static sd_id128_t arg_uuid = {};
static char *arg_machine = NULL;
+static const char *arg_slice = NULL;
static bool arg_private_network = false;
static bool arg_read_only = false;
static bool arg_boot = false;
(1ULL << CAP_AUDIT_CONTROL);
static char **arg_bind = NULL;
static char **arg_bind_ro = NULL;
+static char **arg_setenv = NULL;
static int help(void) {
" -D --directory=NAME Root directory for the container\n"
" -b --boot Boot up full system (i.e. invoke init)\n"
" -u --user=USER Run the command under specified user or uid\n"
- " -C --controllers=LIST Put the container in specified comma-separated\n"
- " cgroup hierarchies\n"
" --uuid=UUID Set a specific machine UUID for the container\n"
" -M --machine=NAME Set the machine name for the container\n"
+ " -S --slice=SLICE Place the container in the specified slice\n"
" --private-network Disable network in container\n"
" --read-only Mount the root directory read-only\n"
" --capability=CAP In addition to the default, retain specified\n"
" capability\n"
+ " --drop-capability=CAP Drop the specified capability from the default set\n"
" --link-journal=MODE Link up guest journal, one of no, auto, guest, host\n"
" -j Equivalent to --link-journal=host\n"
" --bind=PATH[:PATH] Bind mount a file or directory from the host into\n"
" the container\n"
- " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n",
+ " --bind-ro=PATH[:PATH] Similar, but creates a read-only bind mount\n"
+ " --setenv=NAME=VALUE Pass an environment variable to PID 1\n",
program_invocation_short_name);
return 0;
ARG_UUID,
ARG_READ_ONLY,
ARG_CAPABILITY,
+ ARG_DROP_CAPABILITY,
ARG_LINK_JOURNAL,
ARG_BIND,
- ARG_BIND_RO
+ ARG_BIND_RO,
+ ARG_SETENV,
};
static const struct option options[] = {
{ "version", no_argument, NULL, ARG_VERSION },
{ "directory", required_argument, NULL, 'D' },
{ "user", required_argument, NULL, 'u' },
- { "controllers", required_argument, NULL, 'C' },
{ "private-network", no_argument, NULL, ARG_PRIVATE_NETWORK },
{ "boot", no_argument, NULL, 'b' },
{ "uuid", required_argument, NULL, ARG_UUID },
{ "read-only", no_argument, NULL, ARG_READ_ONLY },
{ "capability", required_argument, NULL, ARG_CAPABILITY },
+ { "drop-capability", required_argument, NULL, ARG_DROP_CAPABILITY },
{ "link-journal", required_argument, NULL, ARG_LINK_JOURNAL },
{ "bind", required_argument, NULL, ARG_BIND },
{ "bind-ro", required_argument, NULL, ARG_BIND_RO },
{ "machine", required_argument, NULL, 'M' },
- { NULL, 0, NULL, 0 }
+ { "slice", required_argument, NULL, 'S' },
+ { "setenv", required_argument, NULL, ARG_SETENV },
+ {}
};
- int c;
+ int c, r;
assert(argc >= 0);
assert(argv);
- while ((c = getopt_long(argc, argv, "+hD:u:C:bM:j", options, NULL)) >= 0) {
+ while ((c = getopt_long(argc, argv, "+hD:u:bM:jS:", options, NULL)) >= 0) {
switch (c) {
case 'h':
- help();
- return 0;
+ return help();
case ARG_VERSION:
puts(PACKAGE_STRING);
free(arg_directory);
arg_directory = canonicalize_file_name(optarg);
if (!arg_directory) {
- log_error("Failed to canonicalize root directory.");
+ log_error("Invalid root directory: %m");
return -ENOMEM;
}
break;
- case 'C':
- strv_free(arg_controllers);
- arg_controllers = strv_split(optarg, ",");
- if (!arg_controllers)
- return log_oom();
-
- cg_shorten_controllers(arg_controllers);
- break;
-
case ARG_PRIVATE_NETWORK:
arg_private_network = true;
break;
break;
case ARG_UUID:
- if (!id128_is_valid(optarg)) {
+ r = sd_id128_from_string(optarg, &arg_uuid);
+ if (r < 0) {
log_error("Invalid UUID: %s", optarg);
- return -EINVAL;
+ return r;
}
+ break;
+
+ case 'S':
+ arg_slice = strdup(optarg);
+ if (!arg_slice)
+ return log_oom();
- arg_uuid = optarg;
break;
case 'M':
arg_read_only = true;
break;
- case ARG_CAPABILITY: {
+ case ARG_CAPABILITY:
+ case ARG_DROP_CAPABILITY: {
char *state, *word;
size_t length;
}
free(t);
- arg_retain |= 1ULL << (uint64_t) cap;
+
+ if (c == ARG_CAPABILITY)
+ arg_retain |= 1ULL << (uint64_t) cap;
+ else
+ arg_retain &= ~(1ULL << (uint64_t) cap);
}
break;
_cleanup_free_ char *a = NULL, *b = NULL;
char *e;
char ***x;
- int r;
x = c == ARG_BIND ? &arg_bind : &arg_bind_ro;
r = strv_extend(x, a);
if (r < 0)
- return r;
+ return log_oom();
r = strv_extend(x, b);
if (r < 0)
- return r;
+ return log_oom();
+
+ break;
+ }
+
+ case ARG_SETENV: {
+ char **n;
+
+ if (!env_assignment_is_valid(optarg)) {
+ log_error("Environment variable assignment '%s' is not valid.", optarg);
+ return -EINVAL;
+ }
+
+ n = strv_env_set(arg_setenv, optarg);
+ if (!n)
+ return log_oom();
+ strv_free(arg_setenv);
+ arg_setenv = n;
break;
}
return -EINVAL;
default:
- log_error("Unknown option code %c", c);
- return -EINVAL;
+ assert_not_reached("Unhandled option");
}
}
char **x, **y;
STRV_FOREACH_PAIR(x, y, l) {
- _cleanup_free_ char *where = NULL;
+ char *where;
+ struct stat source_st, dest_st;
+ int r;
- where = strjoin(dest, "/", *y, NULL);
- if (!where)
- return log_oom();
+ if (stat(*x, &source_st) < 0) {
+ log_error("failed to stat %s: %m", *x);
+ return -errno;
+ }
- mkdir_p_label(where, 0755);
+ where = strappenda(dest, *y);
+ r = stat(where, &dest_st);
+ if (r == 0) {
+ if ((source_st.st_mode & S_IFMT) != (dest_st.st_mode & S_IFMT)) {
+ log_error("The file types of %s and %s do not match. Refusing bind mount",
+ *x, where);
+ return -EINVAL;
+ }
+ } else if (errno == ENOENT) {
+ r = mkdir_parents_label(where, 0755);
+ if (r < 0) {
+ log_error("Failed to bind mount %s: %s", *x, strerror(-r));
+ return r;
+ }
+ } else {
+ log_error("Failed to bind mount %s: %s", *x, strerror(errno));
+ return -errno;
+ }
+ /* Create the mount point, but be conservative -- refuse to create block
+ * and char devices. */
+ if (S_ISDIR(source_st.st_mode))
+ mkdir_label(where, 0755);
+ else if (S_ISFIFO(source_st.st_mode))
+ mkfifo(where, 0644);
+ else if (S_ISSOCK(source_st.st_mode))
+ mknod(where, 0644 | S_IFSOCK, 0);
+ else if (S_ISREG(source_st.st_mode))
+ touch(where);
+ else {
+ log_error("Refusing to create mountpoint for file: %s", *x);
+ return -ENOTSUP;
+ }
if (mount(*x, where, "bind", MS_BIND, NULL) < 0) {
log_error("mount(%s) failed: %m", where);
static int setup_resolv_conf(const char *dest) {
char _cleanup_free_ *where = NULL;
- _cleanup_close_ int fd = -1;
assert(dest);
if (!where)
return log_oom();
- fd = open(where, O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW, 0644);
-
/* We don't really care for the results of this really. If it
* fails, it fails, but meh... */
- if (mount("/etc/resolv.conf", where, "bind", MS_BIND, NULL) < 0)
- log_warning("Failed to bind mount /etc/resolv.conf: %m");
- else
- if (mount("/etc/resolv.conf", where, "bind",
- MS_BIND|MS_REMOUNT|MS_RDONLY, NULL) < 0) {
- log_error("Failed to remount /etc/resolv.conf readonly: %m");
- return -errno;
- }
+ copy_file("/etc/resolv.conf", where, O_TRUNC|O_NOFOLLOW);
return 0;
}
}
static int setup_journal(const char *directory) {
- sd_id128_t machine_id;
+ sd_id128_t machine_id, this_id;
_cleanup_free_ char *p = NULL, *b = NULL, *q = NULL, *d = NULL;
char *id;
int r;
- if (arg_link_journal == LINK_NO)
- return 0;
-
p = strappend(directory, "/etc/machine-id");
if (!p)
return log_oom();
return r;
}
+ r = sd_id128_get_machine(&this_id);
+ if (r < 0) {
+ log_error("Failed to retrieve machine ID: %s", strerror(-r));
+ return r;
+ }
+
+ if (sd_id128_equal(machine_id, this_id)) {
+ log_full(arg_link_journal == LINK_AUTO ? LOG_WARNING : LOG_ERR,
+ "Host and machine ids are equal (%s): refusing to link journals", id);
+ if (arg_link_journal == LINK_AUTO)
+ return 0;
+ return
+ -EEXIST;
+ }
+
+ if (arg_link_journal == LINK_NO)
+ return 0;
+
free(p);
p = strappend("/var/log/journal/", id);
q = strjoin(directory, "/var/log/journal/", id, NULL);
return 0;
}
-static int setup_cgroup(const char *path) {
- char **c;
- int r;
+static int setup_kdbus(const char *dest, const char *path) {
+ const char *p;
- r = cg_create_and_attach(SYSTEMD_CGROUP_CONTROLLER, path, 1);
- if (r < 0) {
- log_error("Failed to create cgroup: %s", strerror(-r));
- return r;
+ if (!path)
+ return 0;
+
+ p = strappenda(dest, "/dev/kdbus");
+ if (mkdir(p, 0755) < 0) {
+ log_error("Failed to create kdbus path: %m");
+ return -errno;
}
- STRV_FOREACH(c, arg_controllers) {
- r = cg_create_and_attach(*c, path, 1);
- if (r < 0)
- log_warning("Failed to create cgroup in controller %s: %s", *c, strerror(-r));
+ if (mount(path, p, "bind", MS_BIND, NULL) < 0) {
+ log_error("Failed to mount kdbus namespace path: %m");
+ return -errno;
}
return 0;
}
-static int save_attributes(const char *cgroup, pid_t pid, const char *uuid, const char *directory) {
-#ifdef HAVE_XATTR
- _cleanup_free_ char *path = NULL;
- char buf[DECIMAL_STR_MAX(pid_t)];
- int r = 0, k;
-
- assert(cgroup);
- assert(pid >= 0);
- assert(arg_directory);
+static int drop_capabilities(void) {
+ return capability_bounding_set_drop(~arg_retain, false);
+}
- assert_se(snprintf(buf, sizeof(buf), "%lu", (unsigned long) pid) < (int) sizeof(buf));
+static int register_machine(void) {
+ _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL;
+ _cleanup_bus_unref_ sd_bus *bus = NULL;
+ int r;
- r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, cgroup, NULL, &path);
+ r = sd_bus_open_system(&bus);
if (r < 0) {
- log_error("Failed to get path: %s", strerror(-r));
+ log_error("Failed to open system bus: %s", strerror(-r));
return r;
}
- r = setxattr(path, "trusted.init_pid", buf, strlen(buf), XATTR_CREATE);
- if (r < 0)
- log_warning("Failed to set %s attribute on %s: %m", "trusted.init_pid", path);
-
- if (uuid) {
- k = setxattr(path, "trusted.machine_id", uuid, strlen(uuid), XATTR_CREATE);
- if (k < 0) {
- log_warning("Failed to set %s attribute on %s: %m", "trusted.machine_id", path);
- if (r == 0)
- r = k;
- }
+ r = sd_bus_call_method(
+ bus,
+ "org.freedesktop.machine1",
+ "/org/freedesktop/machine1",
+ "org.freedesktop.machine1.Manager",
+ "CreateMachine",
+ &error,
+ NULL,
+ "sayssusa(sv)",
+ arg_machine,
+ SD_BUS_MESSAGE_APPEND_ID128(arg_uuid),
+ "nspawn",
+ "container",
+ (uint32_t) 0,
+ strempty(arg_directory),
+ !isempty(arg_slice), "Slice", "s", arg_slice);
+ if (r < 0) {
+ log_error("Failed to register machine: %s", bus_error_message(&error, r));
+ return r;
}
- k = setxattr(path, "trusted.root_directory", directory, strlen(directory), XATTR_CREATE);
- if (k < 0) {
- log_warning("Failed to set %s attribute on %s: %m", "trusted.root_directory", path);
- if (r == 0)
- r = k;
- }
- return r;
-#else
return 0;
-#endif
-}
-
-static int drop_capabilities(void) {
- return capability_bounding_set_drop(~arg_retain, false);
}
-static int process_pty(int master, pid_t pid, sigset_t *mask) {
-
- 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;
- int ep = -1, signal_fd = -1, r;
- bool tried_orderly_shutdown = false;
-
- assert(master >= 0);
- assert(pid > 0);
- assert(mask);
-
- fd_nonblock(STDIN_FILENO, 1);
- fd_nonblock(STDOUT_FILENO, 1);
- fd_nonblock(master, 1);
-
- signal_fd = signalfd(-1, mask, SFD_NONBLOCK|SFD_CLOEXEC);
- if (signal_fd < 0) {
- log_error("signalfd(): %m");
- r = -errno;
- goto finish;
- }
-
- ep = epoll_create1(EPOLL_CLOEXEC);
- if (ep < 0) {
- log_error("Failed to create epoll: %m");
- r = -errno;
- goto finish;
- }
-
- /* We read from STDIN only if this is actually a TTY,
- * otherwise we assume non-interactivity. */
- if (isatty(STDIN_FILENO)) {
- zero(stdin_ev);
- stdin_ev.events = EPOLLIN|EPOLLET;
- stdin_ev.data.fd = STDIN_FILENO;
+static int terminate_machine(pid_t pid) {
+ _cleanup_bus_error_free_ sd_bus_error error = SD_BUS_ERROR_NULL;
+ _cleanup_bus_message_unref_ sd_bus_message *reply = NULL;
+ _cleanup_bus_unref_ sd_bus *bus = NULL;
+ const char *path;
+ int r;
- if (epoll_ctl(ep, EPOLL_CTL_ADD, STDIN_FILENO, &stdin_ev) < 0) {
- log_error("Failed to register STDIN in epoll: %m");
- r = -errno;
- goto finish;
- }
+ r = sd_bus_default_system(&bus);
+ if (r < 0) {
+ log_error("Failed to open system bus: %s", strerror(-r));
+ return r;
}
- zero(stdout_ev);
- stdout_ev.events = EPOLLOUT|EPOLLET;
- stdout_ev.data.fd = STDOUT_FILENO;
-
- zero(master_ev);
- master_ev.events = EPOLLIN|EPOLLOUT|EPOLLET;
- master_ev.data.fd = master;
-
- zero(signal_ev);
- signal_ev.events = EPOLLIN;
- signal_ev.data.fd = signal_fd;
-
- if (epoll_ctl(ep, EPOLL_CTL_ADD, STDOUT_FILENO, &stdout_ev) < 0) {
- if (errno != EPERM) {
- log_error("Failed to register stdout in epoll: %m");
- r = -errno;
- goto finish;
- }
- /* stdout without epoll support. Likely redirected to regular file. */
- stdout_writable = true;
+ r = sd_bus_call_method(
+ bus,
+ "org.freedesktop.machine1",
+ "/org/freedesktop/machine1",
+ "org.freedesktop.machine1.Manager",
+ "GetMachineByPID",
+ &error,
+ &reply,
+ "u",
+ (uint32_t) pid);
+ if (r < 0) {
+ /* Note that the machine might already have been
+ * cleaned up automatically, hence don't consider it a
+ * failure if we cannot get the machine object. */
+ log_debug("Failed to get machine: %s", bus_error_message(&error, r));
+ return 0;
}
- if (epoll_ctl(ep, EPOLL_CTL_ADD, master, &master_ev) < 0 ||
- epoll_ctl(ep, EPOLL_CTL_ADD, signal_fd, &signal_ev) < 0) {
- log_error("Failed to register fds in epoll: %m");
- r = -errno;
- goto finish;
+ r = sd_bus_message_read(reply, "o", &path);
+ if (r < 0)
+ return bus_log_parse_error(r);
+
+ r = sd_bus_call_method(
+ bus,
+ "org.freedesktop.machine1",
+ path,
+ "org.freedesktop.machine1.Machine",
+ "Terminate",
+ &error,
+ NULL,
+ NULL);
+ if (r < 0) {
+ log_debug("Failed to terminate machine: %s", bus_error_message(&error, r));
+ return 0;
}
- for (;;) {
- struct epoll_event ev[16];
- ssize_t k;
- int i, nfds;
-
- nfds = epoll_wait(ep, ev, ELEMENTSOF(ev), -1);
- if (nfds < 0) {
-
- if (errno == EINTR || errno == EAGAIN)
- continue;
-
- log_error("epoll_wait(): %m");
- r = -errno;
- goto finish;
- }
-
- assert(nfds >= 1);
-
- for (i = 0; i < nfds; i++) {
- if (ev[i].data.fd == STDIN_FILENO) {
-
- if (ev[i].events & (EPOLLIN|EPOLLHUP))
- stdin_readable = true;
-
- } else if (ev[i].data.fd == STDOUT_FILENO) {
-
- if (ev[i].events & (EPOLLOUT|EPOLLHUP))
- stdout_writable = true;
-
- } else if (ev[i].data.fd == master) {
-
- if (ev[i].events & (EPOLLIN|EPOLLHUP))
- master_readable = true;
-
- if (ev[i].events & (EPOLLOUT|EPOLLHUP))
- master_writable = true;
-
- } else if (ev[i].data.fd == signal_fd) {
- struct signalfd_siginfo sfsi;
- ssize_t n;
-
- n = read(signal_fd, &sfsi, sizeof(sfsi));
- if (n != sizeof(sfsi)) {
-
- if (n >= 0) {
- log_error("Failed to read from signalfd: invalid block size");
- r = -EIO;
- goto finish;
- }
-
- if (errno != EINTR && errno != EAGAIN) {
- log_error("Failed to read from signalfd: %m");
- r = -errno;
- goto finish;
- }
- } else {
-
- if (sfsi.ssi_signo == SIGWINCH) {
- struct winsize ws;
-
- /* The window size changed, let's forward that. */
- if (ioctl(STDIN_FILENO, TIOCGWINSZ, &ws) >= 0)
- ioctl(master, TIOCSWINSZ, &ws);
- } else if (sfsi.ssi_signo == SIGTERM && arg_boot && !tried_orderly_shutdown) {
-
- log_info("Trying to halt container. Send SIGTERM again to trigger immediate termination.");
-
- /* This only works for systemd... */
- tried_orderly_shutdown = true;
- kill(pid, SIGRTMIN+3);
-
- } else {
- r = 0;
- goto finish;
- }
- }
- }
- }
-
- while ((stdin_readable && in_buffer_full <= 0) ||
- (master_writable && in_buffer_full > 0) ||
- (master_readable && out_buffer_full <= 0) ||
- (stdout_writable && out_buffer_full > 0)) {
-
- if (stdin_readable && in_buffer_full < LINE_MAX) {
-
- k = read(STDIN_FILENO, in_buffer + in_buffer_full, LINE_MAX - in_buffer_full);
- if (k < 0) {
-
- if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
- stdin_readable = false;
- else {
- log_error("read(): %m");
- r = -errno;
- goto finish;
- }
- } else
- in_buffer_full += (size_t) k;
- }
-
- if (master_writable && in_buffer_full > 0) {
-
- k = write(master, in_buffer, in_buffer_full);
- if (k < 0) {
-
- if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
- master_writable = false;
- else {
- log_error("write(): %m");
- r = -errno;
- goto finish;
- }
-
- } else {
- assert(in_buffer_full >= (size_t) k);
- memmove(in_buffer, in_buffer + k, in_buffer_full - k);
- in_buffer_full -= k;
- }
- }
-
- if (master_readable && out_buffer_full < LINE_MAX) {
-
- k = read(master, out_buffer + out_buffer_full, LINE_MAX - out_buffer_full);
- if (k < 0) {
-
- if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
- master_readable = false;
- else {
- log_error("read(): %m");
- r = -errno;
- goto finish;
- }
- } else
- out_buffer_full += (size_t) k;
- }
-
- if (stdout_writable && out_buffer_full > 0) {
+ return 0;
+}
- k = write(STDOUT_FILENO, out_buffer, out_buffer_full);
- if (k < 0) {
+static bool audit_enabled(void) {
+ int fd;
- if (errno == EAGAIN || errno == EPIPE || errno == ECONNRESET || errno == EIO)
- stdout_writable = false;
- else {
- log_error("write(): %m");
- r = -errno;
- goto finish;
- }
-
- } else {
- assert(out_buffer_full >= (size_t) k);
- memmove(out_buffer, out_buffer + k, out_buffer_full - k);
- out_buffer_full -= k;
- }
- }
- }
+ fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_AUDIT);
+ if (fd >= 0) {
+ close_nointr_nofail(fd);
+ return true;
}
-
-finish:
- if (ep >= 0)
- close_nointr_nofail(ep);
-
- if (signal_fd >= 0)
- close_nointr_nofail(signal_fd);
-
- return r;
+ return false;
}
int main(int argc, char *argv[]) {
pid_t pid = 0;
int r = EXIT_FAILURE, k;
- _cleanup_free_ char *newcg = NULL;
- _cleanup_close_ int master = -1;
+ _cleanup_close_ int master = -1, kdbus_fd = -1;
int n_fd_passed;
const char *console = NULL;
- struct termios saved_attr, raw_attr;
sigset_t mask;
- bool saved_attr_valid = false;
- struct winsize ws;
- int kmsg_socket_pair[2] = { -1, -1 };
- FDSet *fds = NULL;
+ _cleanup_close_pipe_ int kmsg_socket_pair[2] = { -1, -1 };
+ _cleanup_fdset_free_ FDSet *fds = NULL;
+ _cleanup_free_ char *kdbus_namespace = NULL;
+ const char *ns;
log_parse_environment();
log_open();
path_kill_slashes(arg_directory);
if (!arg_machine) {
- arg_machine = strdup(path_get_file_name(arg_directory));
+ arg_machine = strdup(basename(arg_directory));
if (!arg_machine) {
log_oom();
goto finish;
goto finish;
}
+ if (arg_boot && audit_enabled()) {
+ log_warning("The kernel auditing subsystem is known to be incompatible with containers.\n"
+ "Please make sure to turn off auditing with 'audit=0' on the kernel command\n"
+ "line before using systemd-nspawn. Sleeping for 5s...\n");
+ sleep(5);
+ }
+
if (path_equal(arg_directory, "/")) {
log_error("Spawning container on root directory not supported.");
goto finish;
fdset_close_others(fds);
log_open();
- k = cg_get_machine_path(arg_machine, &newcg);
- if (k < 0) {
- log_error("Failed to determine machine cgroup path: %s", strerror(-k));
- goto finish;
- }
-
- k = cg_is_empty_recursive(SYSTEMD_CGROUP_CONTROLLER, newcg, true);
- if (k <= 0 && k != -ENOENT) {
- log_error("Container already running.");
-
- free(newcg);
- newcg = NULL;
-
- goto finish;
- }
-
master = posix_openpt(O_RDWR|O_NOCTTY|O_CLOEXEC|O_NDELAY);
if (master < 0) {
log_error("Failed to acquire pseudo tty: %m");
goto finish;
}
- log_info("Spawning namespace container on %s (console is %s).", arg_directory, console);
-
- if (ioctl(STDIN_FILENO, TIOCGWINSZ, &ws) >= 0)
- ioctl(master, TIOCSWINSZ, &ws);
+ log_info("Spawning container %s on %s. Press ^] three times within 1s to abort execution.", arg_machine, arg_directory);
if (unlockpt(master) < 0) {
log_error("Failed to unlock tty: %m");
goto finish;
}
- if (tcgetattr(STDIN_FILENO, &saved_attr) >= 0) {
- saved_attr_valid = true;
-
- raw_attr = saved_attr;
- cfmakeraw(&raw_attr);
- raw_attr.c_lflag &= ~ECHO;
- }
+ ns = strappenda("machine-", arg_machine);
+ kdbus_fd = bus_kernel_create_namespace(ns, &kdbus_namespace);
+ if (r < 0)
+ log_debug("Failed to create kdbus namespace: %s", strerror(-r));
+ else
+ log_debug("Successfully created kdbus namespace as %s", kdbus_namespace);
if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_NONBLOCK|SOCK_CLOEXEC, 0, kmsg_socket_pair) < 0) {
log_error("Failed to create kmsg socket pair.");
for (;;) {
siginfo_t status;
- int pipefd[2], pipefd2[2];
-
- if (pipe2(pipefd, O_NONBLOCK|O_CLOEXEC) < 0) {
- log_error("pipe2(): %m");
- goto finish;
- }
-
- if (pipe2(pipefd2, O_NONBLOCK|O_CLOEXEC) < 0) {
- log_error("pipe2(): %m");
- close_pipe(pipefd);
- goto finish;
- }
pid = syscall(__NR_clone, SIGCHLD|CLONE_NEWIPC|CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWUTS|(arg_private_network ? CLONE_NEWNET : 0), NULL);
if (pid < 0) {
gid_t gid = (gid_t) -1;
unsigned n_env = 2;
const char *envp[] = {
- "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+ "PATH=" DEFAULT_PATH_SPLIT_USR,
"container=systemd-nspawn", /* LXC sets container=lxc, so follow the scheme here */
NULL, /* TERM */
NULL, /* HOME */
NULL, /* LISTEN_PID */
NULL
};
+ char **env_use;
envp[n_env] = strv_find_prefix(environ, "TERM=");
if (envp[n_env])
n_env ++;
- /* Wait for the parent process to log our PID */
- close_nointr_nofail(pipefd[1]);
- fd_wait_for_event(pipefd[0], POLLHUP, -1);
- close_nointr_nofail(pipefd[0]);
-
close_nointr_nofail(master);
master = -1;
- if (saved_attr_valid) {
- if (tcsetattr(STDIN_FILENO, TCSANOW, &raw_attr) < 0) {
- log_error("Failed to set terminal attributes: %m");
- goto child_fail;
- }
- }
-
close_nointr(STDIN_FILENO);
close_nointr(STDOUT_FILENO);
close_nointr(STDERR_FILENO);
goto child_fail;
}
- if (setup_cgroup(newcg) < 0)
- goto child_fail;
-
- close_pipe(pipefd2);
+ r = register_machine();
+ if (r < 0)
+ goto finish;
/* Mark everything as slave, so that we still
* receive mounts from the real root, but don't
if (mount_binds(arg_directory, arg_bind_ro, MS_RDONLY) < 0)
goto child_fail;
+ if (setup_kdbus(arg_directory, kdbus_namespace) < 0)
+ goto child_fail;
+
if (chdir(arg_directory) < 0) {
log_error("chdir(%s) failed: %m", arg_directory);
goto child_fail;
goto child_fail;
}
- if (arg_uuid) {
- if (asprintf((char**)(envp + n_env++), "container_uuid=%s", arg_uuid) < 0) {
+ if (!sd_id128_equal(arg_uuid, SD_ID128_NULL)) {
+ if (asprintf((char**)(envp + n_env++), "container_uuid=" SD_ID128_FORMAT_STR, SD_ID128_FORMAT_VAL(arg_uuid)) < 0) {
log_oom();
goto child_fail;
}
}
if ((asprintf((char **)(envp + n_env++), "LISTEN_FDS=%u", n_fd_passed) < 0) ||
- (asprintf((char **)(envp + n_env++), "LISTEN_PID=%lu", (unsigned long) 1) < 0)) {
+ (asprintf((char **)(envp + n_env++), "LISTEN_PID=1") < 0)) {
log_oom();
goto child_fail;
}
setup_hostname();
+ if (!strv_isempty(arg_setenv)) {
+ char **n;
+
+ n = strv_env_merge(2, envp, arg_setenv);
+ if (!n) {
+ log_oom();
+ goto child_fail;
+ }
+
+ env_use = n;
+ } else
+ env_use = (char**) envp;
+
if (arg_boot) {
char **a;
size_t l;
memcpy(a + 1, argv + optind, l * sizeof(char*));
a[0] = (char*) "/usr/lib/systemd/systemd";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
a[0] = (char*) "/lib/systemd/systemd";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
a[0] = (char*) "/sbin/init";
- execve(a[0], a, (char**) envp);
+ execve(a[0], a, env_use);
} else if (argc > optind)
- execvpe(argv[optind], argv + optind, (char**) envp);
+ execvpe(argv[optind], argv + optind, env_use);
else {
chdir(home ? home : "/root");
- execle("/bin/bash", "-bash", NULL, (char**) envp);
+ execle("/bin/bash", "-bash", NULL, env_use);
}
log_error("execv() failed: %m");
_exit(EXIT_FAILURE);
}
- log_info("Init process in the container running as PID %lu.", (unsigned long) pid);
- close_nointr_nofail(pipefd[0]);
- close_nointr_nofail(pipefd[1]);
-
- /* Wait for the child process to establish cgroup hierarchy */
- close_nointr_nofail(pipefd2[1]);
- fd_wait_for_event(pipefd2[0], POLLHUP, -1);
- close_nointr_nofail(pipefd2[0]);
-
- save_attributes(newcg, pid, arg_uuid, arg_directory);
-
fdset_free(fds);
fds = NULL;
- if (process_pty(master, pid, &mask) < 0)
- goto finish;
+ k = process_pty(master, &mask, arg_boot ? pid : 0, SIGRTMIN+3);
+ if (k < 0) {
+ r = EXIT_FAILURE;
+ break;
+ }
- if (saved_attr_valid)
- tcsetattr(STDIN_FILENO, TCSANOW, &saved_attr);
+ putc('\n', stdout);
+
+ /* Kill if it is not dead yet anyway */
+ terminate_machine(pid);
+
+ /* Redundant, but better safe than sorry */
+ kill(pid, SIGKILL);
k = wait_for_terminate(pid, &status);
+ pid = 0;
+
if (k < 0) {
r = EXIT_FAILURE;
break;
if (status.si_code == CLD_EXITED) {
r = status.si_status;
if (status.si_status != 0) {
- log_error("Container failed with error code %i.", status.si_status);
+ log_error("Container %s failed with error code %i.", arg_machine, status.si_status);
break;
}
- log_debug("Container exited successfully.");
+ log_debug("Container %s exited successfully.", arg_machine);
break;
} else if (status.si_code == CLD_KILLED &&
status.si_status == SIGINT) {
- log_info("Container has been shut down.");
+ log_info("Container %s has been shut down.", arg_machine);
r = 0;
break;
} else if (status.si_code == CLD_KILLED &&
status.si_status == SIGHUP) {
- log_info("Container is being rebooted.");
+ log_info("Container %s is being rebooted.", arg_machine);
continue;
} else if (status.si_code == CLD_KILLED ||
status.si_code == CLD_DUMPED) {
- log_error("Container terminated by signal %s.", signal_to_string(status.si_status));
+ log_error("Container %s terminated by signal %s.", arg_machine, signal_to_string(status.si_status));
r = EXIT_FAILURE;
break;
} else {
- log_error("Container failed due to unknown reason.");
+ log_error("Container %s failed due to unknown reason.", arg_machine);
r = EXIT_FAILURE;
break;
}
}
finish:
- if (saved_attr_valid)
- tcsetattr(STDIN_FILENO, TCSANOW, &saved_attr);
-
- close_pipe(kmsg_socket_pair);
-
- if (newcg)
- cg_kill_recursive_and_wait(SYSTEMD_CGROUP_CONTROLLER, newcg, true);
+ if (pid > 0)
+ kill(pid, SIGKILL);
free(arg_directory);
free(arg_machine);
- strv_free(arg_controllers);
-
- fdset_free(fds);
+ free(arg_setenv);
return r;
}