diff options
Diffstat (limited to 'src/core/execute.c')
-rw-r--r-- | src/core/execute.c | 1804 |
1 files changed, 1357 insertions, 447 deletions
diff --git a/src/core/execute.c b/src/core/execute.c index 21721dc240..0c983f4953 100644 --- a/src/core/execute.c +++ b/src/core/execute.c @@ -1,5 +1,3 @@ -/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/ - /*** This file is part of systemd. @@ -21,17 +19,21 @@ #include <errno.h> #include <fcntl.h> -#include <unistd.h> -#include <string.h> -#include <signal.h> -#include <sys/socket.h> -#include <sys/un.h> -#include <sys/prctl.h> -#include <sys/stat.h> +#include <glob.h> #include <grp.h> #include <poll.h> -#include <glob.h> +#include <signal.h> +#include <string.h> +#include <sys/capability.h> +#include <sys/eventfd.h> +#include <sys/mman.h> #include <sys/personality.h> +#include <sys/prctl.h> +#include <sys/socket.h> +#include <sys/stat.h> +#include <sys/un.h> +#include <unistd.h> +#include <utmpx.h> #ifdef HAVE_PAM #include <security/pam_appl.h> @@ -50,45 +52,55 @@ #endif #include "sd-messages.h" -#include "rm-rf.h" -#include "strv.h" -#include "macro.h" -#include "capability.h" -#include "util.h" -#include "log.h" -#include "ioprio.h" -#include "securebits.h" -#include "namespace.h" -#include "exit-status.h" -#include "missing.h" -#include "utmp-wtmp.h" + +#include "af-list.h" +#include "alloc-util.h" +#ifdef HAVE_APPARMOR +#include "apparmor-util.h" +#endif +#include "async.h" +#include "barrier.h" +#include "cap-list.h" +#include "capability-util.h" #include "def.h" -#include "path-util.h" #include "env-util.h" -#include "fileio.h" -#include "unit.h" -#include "async.h" -#include "selinux-util.h" #include "errno-list.h" -#include "af-list.h" -#include "mkdir.h" -#include "smack-util.h" -#include "bus-endpoint.h" -#include "cap-list.h" +#include "execute.h" +#include "exit-status.h" +#include "fd-util.h" +#include "fileio.h" #include "formats-util.h" +#include "fs-util.h" +#include "glob-util.h" +#include "io-util.h" +#include "ioprio.h" +#include "log.h" +#include "macro.h" +#include "missing.h" +#include "mkdir.h" +#include "namespace.h" +#include "parse-util.h" +#include "path-util.h" #include "process-util.h" -#include "terminal-util.h" -#include "signal-util.h" - -#ifdef HAVE_APPARMOR -#include "apparmor-util.h" -#endif - +#include "rlimit-util.h" +#include "rm-rf.h" #ifdef HAVE_SECCOMP #include "seccomp-util.h" #endif - -#include "execute.h" +#include "securebits.h" +#include "selinux-util.h" +#include "signal-util.h" +#include "smack-util.h" +#include "special.h" +#include "string-table.h" +#include "string-util.h" +#include "strv.h" +#include "syslog-util.h" +#include "terminal-util.h" +#include "unit.h" +#include "user-util.h" +#include "util.h" +#include "utmp-wtmp.h" #define IDLE_TIMEOUT_USEC (5*USEC_PER_SEC) #define IDLE_TIMEOUT2_USEC (1*USEC_PER_SEC) @@ -121,7 +133,8 @@ static int shift_fds(int fds[], unsigned n_fds) { if (fds[i] == i+3) continue; - if ((nfd = fcntl(fds[i], F_DUPFD, i+3)) < 0) + nfd = fcntl(fds[i], F_DUPFD, i + 3); + if (nfd < 0) return -errno; safe_close(fds[i]); @@ -155,48 +168,89 @@ static int flags_fds(const int fds[], unsigned n_fds, bool nonblock) { for (i = 0; i < n_fds; i++) { - if ((r = fd_nonblock(fds[i], nonblock)) < 0) + r = fd_nonblock(fds[i], nonblock); + if (r < 0) return r; /* We unconditionally drop FD_CLOEXEC from the fds, * since after all we want to pass these fds to our * children */ - if ((r = fd_cloexec(fds[i], false)) < 0) + r = fd_cloexec(fds[i], false); + if (r < 0) return r; } return 0; } -_pure_ static const char *tty_path(const ExecContext *context) { +static const char *exec_context_tty_path(const ExecContext *context) { assert(context); + if (context->stdio_as_fds) + return NULL; + if (context->tty_path) return context->tty_path; return "/dev/console"; } -static void exec_context_tty_reset(const ExecContext *context) { +static void exec_context_tty_reset(const ExecContext *context, const ExecParameters *p) { + const char *path; + assert(context); - if (context->tty_vhangup) - terminal_vhangup(tty_path(context)); + path = exec_context_tty_path(context); + + if (context->tty_vhangup) { + if (p && p->stdin_fd >= 0) + (void) terminal_vhangup_fd(p->stdin_fd); + else if (path) + (void) terminal_vhangup(path); + } - if (context->tty_reset) - reset_terminal(tty_path(context)); + if (context->tty_reset) { + if (p && p->stdin_fd >= 0) + (void) reset_terminal_fd(p->stdin_fd, true); + else if (path) + (void) reset_terminal(path); + } - if (context->tty_vt_disallocate && context->tty_path) - vt_disallocate(context->tty_path); + if (context->tty_vt_disallocate && path) + (void) vt_disallocate(path); +} + +static bool is_terminal_input(ExecInput i) { + return IN_SET(i, + EXEC_INPUT_TTY, + EXEC_INPUT_TTY_FORCE, + EXEC_INPUT_TTY_FAIL); } static bool is_terminal_output(ExecOutput o) { - return - o == EXEC_OUTPUT_TTY || - o == EXEC_OUTPUT_SYSLOG_AND_CONSOLE || - o == EXEC_OUTPUT_KMSG_AND_CONSOLE || - o == EXEC_OUTPUT_JOURNAL_AND_CONSOLE; + return IN_SET(o, + EXEC_OUTPUT_TTY, + EXEC_OUTPUT_SYSLOG_AND_CONSOLE, + EXEC_OUTPUT_KMSG_AND_CONSOLE, + EXEC_OUTPUT_JOURNAL_AND_CONSOLE); +} + +static bool exec_context_needs_term(const ExecContext *c) { + assert(c); + + /* Return true if the execution context suggests we should set $TERM to something useful. */ + + if (is_terminal_input(c->std_input)) + return true; + + if (is_terminal_output(c->std_output)) + return true; + + if (is_terminal_output(c->std_error)) + return true; + + return !!c->tty_path; } static int open_null_as(int flags, int nfd) { @@ -244,7 +298,7 @@ static int connect_journal_socket(int fd, uid_t uid, gid_t gid) { } } - r = connect(fd, &sa.sa, offsetof(struct sockaddr_un, sun_path) + strlen(sa.un.sun_path)); + r = connect(fd, &sa.sa, SOCKADDR_UN_LEN(sa.un)); if (r < 0) r = -errno; @@ -261,7 +315,15 @@ static int connect_journal_socket(int fd, uid_t uid, gid_t gid) { return r; } -static int connect_logger_as(const ExecContext *context, ExecOutput output, const char *ident, const char *unit_id, int nfd, uid_t uid, gid_t gid) { +static int connect_logger_as( + Unit *unit, + const ExecContext *context, + ExecOutput output, + const char *ident, + int nfd, + uid_t uid, + gid_t gid) { + int fd, r; assert(context); @@ -282,7 +344,7 @@ static int connect_logger_as(const ExecContext *context, ExecOutput output, cons return -errno; } - fd_inc_sndbuf(fd, SNDBUF_SIZE); + (void) fd_inc_sndbuf(fd, SNDBUF_SIZE); dprintf(fd, "%s\n" @@ -293,18 +355,18 @@ static int connect_logger_as(const ExecContext *context, ExecOutput output, cons "%i\n" "%i\n", context->syslog_identifier ? context->syslog_identifier : ident, - unit_id, + unit->id, context->syslog_priority, !!context->syslog_level_prefix, output == EXEC_OUTPUT_SYSLOG || output == EXEC_OUTPUT_SYSLOG_AND_CONSOLE, output == EXEC_OUTPUT_KMSG || output == EXEC_OUTPUT_KMSG_AND_CONSOLE, is_terminal_output(output)); - if (fd != nfd) { - r = dup2(fd, nfd) < 0 ? -errno : nfd; - safe_close(fd); - } else - r = nfd; + if (fd == nfd) + return nfd; + + r = dup2(fd, nfd) < 0 ? -errno : nfd; + safe_close(fd); return r; } @@ -314,7 +376,8 @@ static int open_terminal_as(const char *path, mode_t mode, int nfd) { assert(path); assert(nfd >= 0); - if ((fd = open_terminal(path, mode | O_NOCTTY)) < 0) + fd = open_terminal(path, mode | O_NOCTTY); + if (fd < 0) return fd; if (fd != nfd) { @@ -326,13 +389,6 @@ static int open_terminal_as(const char *path, mode_t mode, int nfd) { return r; } -static bool is_terminal_input(ExecInput i) { - return - i == EXEC_INPUT_TTY || - i == EXEC_INPUT_TTY_FORCE || - i == EXEC_INPUT_TTY_FAIL; -} - static int fixup_input(ExecInput std_input, int socket_fd, bool apply_tty_stdin) { if (is_terminal_input(std_input) && !apply_tty_stdin) @@ -352,12 +408,28 @@ static int fixup_output(ExecOutput std_output, int socket_fd) { return std_output; } -static int setup_input(const ExecContext *context, int socket_fd, bool apply_tty_stdin) { +static int setup_input( + const ExecContext *context, + const ExecParameters *params, + int socket_fd) { + ExecInput i; assert(context); + assert(params); + + if (params->stdin_fd >= 0) { + if (dup2(params->stdin_fd, STDIN_FILENO) < 0) + return -errno; + + /* Try to make this the controlling tty, if it is a tty, and reset it */ + (void) ioctl(STDIN_FILENO, TIOCSCTTY, context->std_input == EXEC_INPUT_TTY_FORCE); + (void) reset_terminal_fd(STDIN_FILENO, true); + + return STDIN_FILENO; + } - i = fixup_input(context->std_input, socket_fd, apply_tty_stdin); + i = fixup_input(context->std_input, socket_fd, params->flags & EXEC_APPLY_TTY_STDIN); switch (i) { @@ -369,7 +441,7 @@ static int setup_input(const ExecContext *context, int socket_fd, bool apply_tty case EXEC_INPUT_TTY_FAIL: { int fd, r; - fd = acquire_terminal(tty_path(context), + fd = acquire_terminal(exec_context_tty_path(context), i == EXEC_INPUT_TTY_FAIL, i == EXEC_INPUT_TTY_FORCE, false, @@ -394,16 +466,45 @@ static int setup_input(const ExecContext *context, int socket_fd, bool apply_tty } } -static int setup_output(Unit *unit, const ExecContext *context, int fileno, int socket_fd, const char *ident, bool apply_tty_stdin, uid_t uid, gid_t gid) { +static int setup_output( + Unit *unit, + const ExecContext *context, + const ExecParameters *params, + int fileno, + int socket_fd, + const char *ident, + uid_t uid, + gid_t gid, + dev_t *journal_stream_dev, + ino_t *journal_stream_ino) { + ExecOutput o; ExecInput i; int r; assert(unit); assert(context); + assert(params); assert(ident); + assert(journal_stream_dev); + assert(journal_stream_ino); + + if (fileno == STDOUT_FILENO && params->stdout_fd >= 0) { + + if (dup2(params->stdout_fd, STDOUT_FILENO) < 0) + return -errno; + + return STDOUT_FILENO; + } + + if (fileno == STDERR_FILENO && params->stderr_fd >= 0) { + if (dup2(params->stderr_fd, STDERR_FILENO) < 0) + return -errno; + + return STDERR_FILENO; + } - i = fixup_input(context->std_input, socket_fd, apply_tty_stdin); + i = fixup_input(context->std_input, socket_fd, params->flags & EXEC_APPLY_TTY_STDIN); o = fixup_output(context->std_output, socket_fd); if (fileno == STDERR_FILENO) { @@ -430,7 +531,7 @@ static int setup_output(Unit *unit, const ExecContext *context, int fileno, int } else if (o == EXEC_OUTPUT_INHERIT) { /* If input got downgraded, inherit the original value */ if (i == EXEC_INPUT_NULL && is_terminal_input(context->std_input)) - return open_terminal_as(tty_path(context), O_WRONLY, fileno); + return open_terminal_as(exec_context_tty_path(context), O_WRONLY, fileno); /* If the input is connected to anything that's not a /dev/null, inherit that... */ if (i != EXEC_INPUT_NULL) @@ -454,7 +555,7 @@ static int setup_output(Unit *unit, const ExecContext *context, int fileno, int return dup2(STDIN_FILENO, fileno) < 0 ? -errno : fileno; /* We don't reset the terminal if this is just about output */ - return open_terminal_as(tty_path(context), O_WRONLY, fileno); + return open_terminal_as(exec_context_tty_path(context), O_WRONLY, fileno); case EXEC_OUTPUT_SYSLOG: case EXEC_OUTPUT_SYSLOG_AND_CONSOLE: @@ -462,10 +563,21 @@ static int setup_output(Unit *unit, const ExecContext *context, int fileno, int case EXEC_OUTPUT_KMSG_AND_CONSOLE: case EXEC_OUTPUT_JOURNAL: case EXEC_OUTPUT_JOURNAL_AND_CONSOLE: - r = connect_logger_as(context, o, ident, unit->id, fileno, uid, gid); + r = connect_logger_as(unit, context, o, ident, fileno, uid, gid); if (r < 0) { log_unit_error_errno(unit, r, "Failed to connect %s to the journal socket, ignoring: %m", fileno == STDOUT_FILENO ? "stdout" : "stderr"); r = open_null_as(O_WRONLY, fileno); + } else { + struct stat st; + + /* If we connected this fd to the journal via a stream, patch the device/inode into the passed + * parameters, but only then. This is useful so that we can set $JOURNAL_STREAM that permits + * services to detect whether they are connected to the journal or not. */ + + if (fstat(fileno, &st) >= 0) { + *journal_stream_dev = st.st_dev; + *journal_stream_ino = st.st_ino; + } } return r; @@ -483,6 +595,10 @@ static int chown_terminal(int fd, uid_t uid) { assert(fd >= 0); + /* Before we chown/chmod the TTY, let's ensure this is actually a tty */ + if (isatty(fd) < 1) + return 0; + /* This might fail. What matters are the results. */ (void) fchown(fd, uid, -1); (void) fchmod(fd, TTY_MODE); @@ -496,9 +612,9 @@ static int chown_terminal(int fd, uid_t uid) { return 0; } -static int setup_confirm_stdio(int *_saved_stdin, - int *_saved_stdout) { - int fd = -1, saved_stdin, saved_stdout = -1, r; +static int setup_confirm_stdio(int *_saved_stdin, int *_saved_stdout) { + _cleanup_close_ int fd = -1, saved_stdin = -1, saved_stdout = -1; + int r; assert(_saved_stdin); assert(_saved_stdout); @@ -508,10 +624,8 @@ static int setup_confirm_stdio(int *_saved_stdin, return -errno; saved_stdout = fcntl(STDOUT_FILENO, F_DUPFD, 3); - if (saved_stdout < 0) { - r = errno; - goto fail; - } + if (saved_stdout < 0) + return -errno; fd = acquire_terminal( "/dev/console", @@ -519,39 +633,33 @@ static int setup_confirm_stdio(int *_saved_stdin, false, false, DEFAULT_CONFIRM_USEC); - if (fd < 0) { - r = fd; - goto fail; - } + if (fd < 0) + return fd; r = chown_terminal(fd, getuid()); if (r < 0) - goto fail; + return r; - if (dup2(fd, STDIN_FILENO) < 0) { - r = -errno; - goto fail; - } + r = reset_terminal_fd(fd, true); + if (r < 0) + return r; - if (dup2(fd, STDOUT_FILENO) < 0) { - r = -errno; - goto fail; - } + if (dup2(fd, STDIN_FILENO) < 0) + return -errno; + + if (dup2(fd, STDOUT_FILENO) < 0) + return -errno; if (fd >= 2) safe_close(fd); + fd = -1; *_saved_stdin = saved_stdin; *_saved_stdout = saved_stdout; - return 0; - -fail: - safe_close(saved_stdout); - safe_close(saved_stdin); - safe_close(fd); + saved_stdin = saved_stdout = -1; - return r; + return 0; } _printf_(1, 2) static int write_confirm_message(const char *format, ...) { @@ -571,9 +679,7 @@ _printf_(1, 2) static int write_confirm_message(const char *format, ...) { return 0; } -static int restore_confirm_stdio(int *saved_stdin, - int *saved_stdout) { - +static int restore_confirm_stdio(int *saved_stdin, int *saved_stdout) { int r = 0; assert(saved_stdin); @@ -589,8 +695,8 @@ static int restore_confirm_stdio(int *saved_stdin, if (dup2(*saved_stdout, STDOUT_FILENO) < 0) r = -errno; - safe_close(*saved_stdin); - safe_close(*saved_stdout); + *saved_stdin = safe_close(*saved_stdin); + *saved_stdout = safe_close(*saved_stdout); return r; } @@ -624,14 +730,6 @@ static int enforce_groups(const ExecContext *context, const char *username, gid_ * we avoid NSS lookups for gid=0. */ if (context->group || username) { - - if (context->group) { - const char *g = context->group; - - if ((r = get_group_creds(&g, &gid)) < 0) - return r; - } - /* First step, initialize groups from /etc/groups */ if (username && gid != 0) { if (initgroups(username, gid) < 0) @@ -657,7 +755,8 @@ static int enforce_groups(const ExecContext *context, const char *username, gid_ return -ENOMEM; if (keep_groups) { - if ((k = getgroups(ngroups_max, gids)) < 0) { + k = getgroups(ngroups_max, gids); + if (k < 0) { free(gids); return -errno; } @@ -682,9 +781,10 @@ static int enforce_groups(const ExecContext *context, const char *username, gid_ k++; } - if (setgroups(k, gids) < 0) { + r = maybe_setgroups(k, gids); + if (r < 0) { free(gids); - return -errno; + return r; } free(gids); @@ -696,15 +796,10 @@ static int enforce_groups(const ExecContext *context, const char *username, gid_ static int enforce_user(const ExecContext *context, uid_t uid) { assert(context); - /* Sets (but doesn't lookup) the uid and make sure we keep the + /* Sets (but doesn't look up) the uid and make sure we keep the * capabilities while doing so. */ - if (context->capabilities) { - _cleanup_cap_free_ cap_t d = NULL; - static const cap_value_t bits[] = { - CAP_SETUID, /* Necessary so that we can run setresuid() below */ - CAP_SETPCAP /* Necessary so that we can set PR_SET_SECUREBITS later on */ - }; + if (context->capability_ambient_set != 0) { /* First step: If we need to keep capabilities but * drop privileges we need to make sure we keep our @@ -716,23 +811,9 @@ static int enforce_user(const ExecContext *context, uid_t uid) { if (prctl(PR_SET_SECUREBITS, sb) < 0) return -errno; } - - /* Second step: set the capabilities. This will reduce - * the capabilities to the minimum we need. */ - - d = cap_dup(context->capabilities); - if (!d) - return -errno; - - if (cap_set_flag(d, CAP_EFFECTIVE, ELEMENTSOF(bits), bits, CAP_SET) < 0 || - cap_set_flag(d, CAP_PERMITTED, ELEMENTSOF(bits), bits, CAP_SET) < 0) - return -errno; - - if (cap_set_proc(d) < 0) - return -errno; } - /* Third step: actually set the uids */ + /* Second step: actually set the uids */ if (setresuid(uid, uid, uid) < 0) return -errno; @@ -757,31 +838,36 @@ static int null_conv( return PAM_CONV_ERR; } +#endif + static int setup_pam( const char *name, const char *user, uid_t uid, + gid_t gid, const char *tty, - char ***pam_env, + char ***env, int fds[], unsigned n_fds) { +#ifdef HAVE_PAM + static const struct pam_conv conv = { .conv = null_conv, .appdata_ptr = NULL }; + _cleanup_(barrier_destroy) Barrier barrier = BARRIER_NULL; pam_handle_t *handle = NULL; sigset_t old_ss; - int pam_code = PAM_SUCCESS; - int err; - char **e = NULL; + int pam_code = PAM_SUCCESS, r; + char **nv, **e = NULL; bool close_session = false; pid_t pam_pid = 0, parent_pid; int flags = 0; assert(name); assert(user); - assert(pam_env); + assert(env); /* We set up PAM in the parent process, then fork. The child * will then stay around until killed via PR_GET_PDEATHSIG or @@ -790,6 +876,10 @@ static int setup_pam( * daemon. We do things this way to ensure that the main PID * of the daemon is the one we initially fork()ed. */ + r = barrier_create(&barrier); + if (r < 0) + goto fail; + if (log_get_max_level() < LOG_DEBUG) flags |= PAM_SILENT; @@ -805,6 +895,12 @@ static int setup_pam( goto fail; } + STRV_FOREACH(nv, *env) { + pam_code = pam_putenv(handle, *nv); + if (pam_code != PAM_SUCCESS) + goto fail; + } + pam_code = pam_acct_mgmt(handle, flags); if (pam_code != PAM_SUCCESS) goto fail; @@ -829,15 +925,17 @@ static int setup_pam( parent_pid = getpid(); pam_pid = fork(); - if (pam_pid < 0) + if (pam_pid < 0) { + r = -errno; goto fail; + } if (pam_pid == 0) { - int sig; - int r = EXIT_PAM; + int sig, ret = EXIT_PAM; /* The child's job is to reset the PAM session on * termination */ + barrier_set_role(&barrier, BARRIER_CHILD); /* This string must fit in 10 chars (i.e. the length * of "/sbin/init"), to look pretty in /bin/ps */ @@ -852,8 +950,14 @@ static int setup_pam( * and this will make PR_SET_PDEATHSIG work in most cases. * If this fails, ignore the error - but expect sd-pam threads * to fail to exit normally */ + + r = maybe_setgroups(0, NULL); + if (r < 0) + log_warning_errno(r, "Failed to setgroups() in sd-pam: %m"); + if (setresgid(gid, gid, gid) < 0) + log_warning_errno(errno, "Failed to setresgid() in sd-pam: %m"); if (setresuid(uid, uid, uid) < 0) - log_error_errno(r, "Error: Failed to setresuid() in sd-pam: %m"); + log_warning_errno(errno, "Failed to setresuid() in sd-pam: %m"); (void) ignore_signals(SIGPIPE, -1); @@ -865,6 +969,11 @@ static int setup_pam( if (prctl(PR_SET_PDEATHSIG, SIGTERM) < 0) goto child_finish; + /* Tell the parent that our setup is done. This is especially + * important regarding dropping privileges. Otherwise, unit + * setup might race against our setresuid(2) call. */ + barrier_place(&barrier); + /* Check if our parent process might already have * died? */ if (getppid() == parent_pid) { @@ -893,13 +1002,15 @@ static int setup_pam( goto child_finish; } - r = 0; + ret = 0; child_finish: pam_end(handle, pam_code | flags); - _exit(r); + _exit(ret); } + barrier_set_role(&barrier, BARRIER_PARENT); + /* If the child was forked off successfully it will do all the * cleanups, so forget about the handle here. */ handle = NULL; @@ -911,19 +1022,22 @@ static int setup_pam( * might have opened it, but we don't want this fd around. */ closelog(); - *pam_env = e; - e = NULL; + /* Synchronously wait for the child to initialize. We don't care for + * errors as we cannot recover. However, warn loudly if it happens. */ + if (!barrier_place_and_sync(&barrier)) + log_error("PAM initialization failed"); + + strv_free(*env); + *env = e; return 0; fail: if (pam_code != PAM_SUCCESS) { log_error("PAM failed: %s", pam_strerror(handle, pam_code)); - err = -EPERM; /* PAM errors do not map to errno */ - } else { - log_error_errno(errno, "PAM failed: %m"); - err = -errno; - } + r = -EPERM; /* PAM errors do not map to errno */ + } else + log_error_errno(r, "PAM failed: %m"); if (handle) { if (close_session) @@ -933,17 +1047,13 @@ fail: } strv_free(e); - closelog(); - if (pam_pid > 1) { - kill(pam_pid, SIGTERM); - kill(pam_pid, SIGCONT); - } - - return err; -} + return r; +#else + return 0; #endif +} static void rename_process_from_path(const char *path) { char process_name[11]; @@ -978,7 +1088,17 @@ static void rename_process_from_path(const char *path) { #ifdef HAVE_SECCOMP -static int apply_seccomp(const ExecContext *c) { +static bool skip_seccomp_unavailable(const Unit* u, const char* msg) { + if (!is_seccomp_available()) { + log_open(); + log_unit_debug(u, "SECCOMP features not detected in the kernel, skipping %s", msg); + log_close(); + return true; + } + return false; +} + +static int apply_seccomp(const Unit* u, const ExecContext *c) { uint32_t negative_action, action; scmp_filter_ctx *seccomp; Iterator i; @@ -987,6 +1107,9 @@ static int apply_seccomp(const ExecContext *c) { assert(c); + if (skip_seccomp_unavailable(u, "syscall filtering")) + return 0; + negative_action = c->syscall_errno == 0 ? SCMP_ACT_KILL : SCMP_ACT_ERRNO(c->syscall_errno); seccomp = seccomp_init(c->syscall_whitelist ? negative_action : SCMP_ACT_ALLOW); @@ -1027,13 +1150,16 @@ finish: return r; } -static int apply_address_families(const ExecContext *c) { +static int apply_address_families(const Unit* u, const ExecContext *c) { scmp_filter_ctx *seccomp; Iterator i; int r; assert(c); + if (skip_seccomp_unavailable(u, "RestrictAddressFamilies=")) + return 0; + seccomp = seccomp_init(SCMP_ACT_ALLOW); if (!seccomp) return -ENOMEM; @@ -1148,14 +1274,246 @@ finish: return r; } +static int apply_memory_deny_write_execute(const Unit* u, const ExecContext *c) { + scmp_filter_ctx *seccomp; + int r; + + assert(c); + + if (skip_seccomp_unavailable(u, "MemoryDenyWriteExecute=")) + return 0; + + seccomp = seccomp_init(SCMP_ACT_ALLOW); + if (!seccomp) + return -ENOMEM; + + r = seccomp_add_secondary_archs(seccomp); + if (r < 0) + goto finish; + + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(mmap), + 1, + SCMP_A2(SCMP_CMP_MASKED_EQ, PROT_EXEC|PROT_WRITE, PROT_EXEC|PROT_WRITE)); + if (r < 0) + goto finish; + + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(mprotect), + 1, + SCMP_A2(SCMP_CMP_MASKED_EQ, PROT_EXEC, PROT_EXEC)); + if (r < 0) + goto finish; + + r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0); + if (r < 0) + goto finish; + + r = seccomp_load(seccomp); + +finish: + seccomp_release(seccomp); + return r; +} + +static int apply_restrict_realtime(const Unit* u, const ExecContext *c) { + static const int permitted_policies[] = { + SCHED_OTHER, + SCHED_BATCH, + SCHED_IDLE, + }; + + scmp_filter_ctx *seccomp; + unsigned i; + int r, p, max_policy = 0; + + assert(c); + + if (skip_seccomp_unavailable(u, "RestrictRealtime=")) + return 0; + + seccomp = seccomp_init(SCMP_ACT_ALLOW); + if (!seccomp) + return -ENOMEM; + + r = seccomp_add_secondary_archs(seccomp); + if (r < 0) + goto finish; + + /* Determine the highest policy constant we want to allow */ + for (i = 0; i < ELEMENTSOF(permitted_policies); i++) + if (permitted_policies[i] > max_policy) + max_policy = permitted_policies[i]; + + /* Go through all policies with lower values than that, and block them -- unless they appear in the + * whitelist. */ + for (p = 0; p < max_policy; p++) { + bool good = false; + + /* Check if this is in the whitelist. */ + for (i = 0; i < ELEMENTSOF(permitted_policies); i++) + if (permitted_policies[i] == p) { + good = true; + break; + } + + if (good) + continue; + + /* Deny this policy */ + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(sched_setscheduler), + 1, + SCMP_A1(SCMP_CMP_EQ, p)); + if (r < 0) + goto finish; + } + + /* Blacklist all other policies, i.e. the ones with higher values. Note that all comparisons are unsigned here, + * hence no need no check for < 0 values. */ + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(sched_setscheduler), + 1, + SCMP_A1(SCMP_CMP_GT, max_policy)); + if (r < 0) + goto finish; + + r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0); + if (r < 0) + goto finish; + + r = seccomp_load(seccomp); + +finish: + seccomp_release(seccomp); + return r; +} + +static int apply_protect_sysctl(Unit *u, const ExecContext *c) { + scmp_filter_ctx *seccomp; + int r; + + assert(c); + + /* Turn off the legacy sysctl() system call. Many distributions turn this off while building the kernel, but + * let's protect even those systems where this is left on in the kernel. */ + + if (skip_seccomp_unavailable(u, "ProtectKernelTunables=")) + return 0; + + seccomp = seccomp_init(SCMP_ACT_ALLOW); + if (!seccomp) + return -ENOMEM; + + r = seccomp_add_secondary_archs(seccomp); + if (r < 0) + goto finish; + + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + SCMP_SYS(_sysctl), + 0); + if (r < 0) + goto finish; + + r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0); + if (r < 0) + goto finish; + + r = seccomp_load(seccomp); + +finish: + seccomp_release(seccomp); + return r; +} + +static int apply_private_devices(Unit *u, const ExecContext *c) { + const SystemCallFilterSet *set; + scmp_filter_ctx *seccomp; + const char *sys; + bool syscalls_found = false; + int r; + + assert(c); + + /* If PrivateDevices= is set, also turn off iopl and all @raw-io syscalls. */ + + if (skip_seccomp_unavailable(u, "PrivateDevices=")) + return 0; + + seccomp = seccomp_init(SCMP_ACT_ALLOW); + if (!seccomp) + return -ENOMEM; + + r = seccomp_add_secondary_archs(seccomp); + if (r < 0) + goto finish; + + for (set = syscall_filter_sets; set->set_name; set++) + if (streq(set->set_name, "@raw-io")) { + syscalls_found = true; + break; + } + + /* We should never fail here */ + if (!syscalls_found) { + r = -EOPNOTSUPP; + goto finish; + } + + NULSTR_FOREACH(sys, set->value) { + int id; + bool add = true; + +#ifndef __NR_s390_pci_mmio_read + if (streq(sys, "s390_pci_mmio_read")) + add = false; +#endif +#ifndef __NR_s390_pci_mmio_write + if (streq(sys, "s390_pci_mmio_write")) + add = false; +#endif + + if (!add) + continue; + + id = seccomp_syscall_resolve_name(sys); + + r = seccomp_rule_add( + seccomp, + SCMP_ACT_ERRNO(EPERM), + id, 0); + if (r < 0) + goto finish; + } + + r = seccomp_attr_set(seccomp, SCMP_FLTATR_CTL_NNP, 0); + if (r < 0) + goto finish; + + r = seccomp_load(seccomp); + +finish: + seccomp_release(seccomp); + return r; +} + #endif static void do_idle_pipe_dance(int idle_pipe[4]) { assert(idle_pipe); - - safe_close(idle_pipe[1]); - safe_close(idle_pipe[2]); + idle_pipe[1] = safe_close(idle_pipe[1]); + idle_pipe[2] = safe_close(idle_pipe[2]); if (idle_pipe[0] >= 0) { int r; @@ -1163,41 +1521,49 @@ static void do_idle_pipe_dance(int idle_pipe[4]) { r = fd_wait_for_event(idle_pipe[0], POLLHUP, IDLE_TIMEOUT_USEC); if (idle_pipe[3] >= 0 && r == 0 /* timeout */) { + ssize_t n; + /* Signal systemd that we are bored and want to continue. */ - r = write(idle_pipe[3], "x", 1); - if (r > 0) + n = write(idle_pipe[3], "x", 1); + if (n > 0) /* Wait for systemd to react to the signal above. */ fd_wait_for_event(idle_pipe[0], POLLHUP, IDLE_TIMEOUT2_USEC); } - safe_close(idle_pipe[0]); + idle_pipe[0] = safe_close(idle_pipe[0]); } - safe_close(idle_pipe[3]); + idle_pipe[3] = safe_close(idle_pipe[3]); } static int build_environment( + Unit *u, const ExecContext *c, + const ExecParameters *p, unsigned n_fds, - usec_t watchdog_usec, const char *home, const char *username, const char *shell, + dev_t journal_stream_dev, + ino_t journal_stream_ino, char ***ret) { _cleanup_strv_free_ char **our_env = NULL; unsigned n_env = 0; char *x; + assert(u); assert(c); assert(ret); - our_env = new0(char*, 10); + our_env = new0(char*, 14); if (!our_env) return -ENOMEM; if (n_fds > 0) { + _cleanup_free_ char *joined = NULL; + if (asprintf(&x, "LISTEN_PID="PID_FMT, getpid()) < 0) return -ENOMEM; our_env[n_env++] = x; @@ -1205,14 +1571,33 @@ static int build_environment( if (asprintf(&x, "LISTEN_FDS=%u", n_fds) < 0) return -ENOMEM; our_env[n_env++] = x; + + joined = strv_join(p->fd_names, ":"); + if (!joined) + return -ENOMEM; + + x = strjoin("LISTEN_FDNAMES=", joined, NULL); + if (!x) + return -ENOMEM; + our_env[n_env++] = x; } - if (watchdog_usec > 0) { + if ((p->flags & EXEC_SET_WATCHDOG) && p->watchdog_usec > 0) { if (asprintf(&x, "WATCHDOG_PID="PID_FMT, getpid()) < 0) return -ENOMEM; our_env[n_env++] = x; - if (asprintf(&x, "WATCHDOG_USEC="USEC_FMT, watchdog_usec) < 0) + if (asprintf(&x, "WATCHDOG_USEC="USEC_FMT, p->watchdog_usec) < 0) + return -ENOMEM; + our_env[n_env++] = x; + } + + /* If this is D-Bus, tell the nss-systemd module, since it relies on being able to use D-Bus look up dynamic + * users via PID 1, possibly dead-locking the dbus daemon. This way it will not use D-Bus to resolve names, but + * check the database directly. */ + if (unit_has_name(u, SPECIAL_DBUS_SERVICE)) { + x = strdup("SYSTEMD_NSS_BYPASS_BUS=1"); + if (!x) return -ENOMEM; our_env[n_env++] = x; } @@ -1243,19 +1628,42 @@ static int build_environment( our_env[n_env++] = x; } - if (is_terminal_input(c->std_input) || - c->std_output == EXEC_OUTPUT_TTY || - c->std_error == EXEC_OUTPUT_TTY || - c->tty_path) { + if (!sd_id128_is_null(u->invocation_id)) { + if (asprintf(&x, "INVOCATION_ID=" SD_ID128_FORMAT_STR, SD_ID128_FORMAT_VAL(u->invocation_id)) < 0) + return -ENOMEM; + + our_env[n_env++] = x; + } + + if (exec_context_needs_term(c)) { + const char *tty_path, *term = NULL; + + tty_path = exec_context_tty_path(c); - x = strdup(default_term_for_tty(tty_path(c))); + /* If we are forked off PID 1 and we are supposed to operate on /dev/console, then let's try to inherit + * the $TERM set for PID 1. This is useful for containers so that the $TERM the container manager + * passes to PID 1 ends up all the way in the console login shown. */ + + if (path_equal(tty_path, "/dev/console") && getppid() == 1) + term = getenv("TERM"); + if (!term) + term = default_term_for_tty(tty_path); + + x = strappend("TERM=", term); if (!x) return -ENOMEM; our_env[n_env++] = x; } + if (journal_stream_dev != 0 && journal_stream_ino != 0) { + if (asprintf(&x, "JOURNAL_STREAM=" DEV_FMT ":" INO_FMT, journal_stream_dev, journal_stream_ino) < 0) + return -ENOMEM; + + our_env[n_env++] = x; + } + our_env[n_env++] = NULL; - assert(n_env <= 10); + assert(n_env <= 12); *ret = our_env; our_env = NULL; @@ -1263,6 +1671,34 @@ static int build_environment( return 0; } +static int build_pass_environment(const ExecContext *c, char ***ret) { + _cleanup_strv_free_ char **pass_env = NULL; + size_t n_env = 0, n_bufsize = 0; + char **i; + + STRV_FOREACH(i, c->pass_environment) { + _cleanup_free_ char *x = NULL; + char *v; + + v = getenv(*i); + if (!v) + continue; + x = strjoin(*i, "=", v, NULL); + if (!x) + return -ENOMEM; + if (!GREEDY_REALLOC(pass_env, n_bufsize, n_env + 2)) + return -ENOMEM; + pass_env[n_env++] = x; + pass_env[n_env] = NULL; + x = NULL; + } + + *ret = pass_env; + pass_env = NULL; + + return 0; +} + static bool exec_needs_mount_namespace( const ExecContext *context, const ExecParameters *params, @@ -1271,9 +1707,9 @@ static bool exec_needs_mount_namespace( assert(context); assert(params); - if (!strv_isempty(context->read_write_dirs) || - !strv_isempty(context->read_only_dirs) || - !strv_isempty(context->inaccessible_dirs)) + if (!strv_isempty(context->read_write_paths) || + !strv_isempty(context->read_only_paths) || + !strv_isempty(context->inaccessible_paths)) return true; if (context->mount_flags != 0) @@ -1282,38 +1718,412 @@ static bool exec_needs_mount_namespace( if (context->private_tmp && runtime && (runtime->tmp_dir || runtime->var_tmp_dir)) return true; - if (params->bus_endpoint_path) - return true; - if (context->private_devices || context->protect_system != PROTECT_SYSTEM_NO || - context->protect_home != PROTECT_HOME_NO) + context->protect_home != PROTECT_HOME_NO || + context->protect_kernel_tunables || + context->protect_control_groups) return true; return false; } +static int setup_private_users(uid_t uid, gid_t gid) { + _cleanup_free_ char *uid_map = NULL, *gid_map = NULL; + _cleanup_close_pair_ int errno_pipe[2] = { -1, -1 }; + _cleanup_close_ int unshare_ready_fd = -1; + _cleanup_(sigkill_waitp) pid_t pid = 0; + uint64_t c = 1; + siginfo_t si; + ssize_t n; + int r; + + /* Set up a user namespace and map root to root, the selected UID/GID to itself, and everything else to + * nobody. In order to be able to write this mapping we need CAP_SETUID in the original user namespace, which + * we however lack after opening the user namespace. To work around this we fork() a temporary child process, + * which waits for the parent to create the new user namespace while staying in the original namespace. The + * child then writes the UID mapping, under full privileges. The parent waits for the child to finish and + * continues execution normally. */ + + if (uid != 0 && uid_is_valid(uid)) + asprintf(&uid_map, + "0 0 1\n" /* Map root → root */ + UID_FMT " " UID_FMT " 1\n", /* Map $UID → $UID */ + uid, uid); /* The case where the above is the same */ + else + uid_map = strdup("0 0 1\n"); + if (!uid_map) + return -ENOMEM; + + if (gid != 0 && gid_is_valid(gid)) + asprintf(&gid_map, + "0 0 1\n" /* Map root → root */ + GID_FMT " " GID_FMT " 1\n", /* Map $GID → $GID */ + gid, gid); + else + gid_map = strdup("0 0 1\n"); /* The case where the above is the same */ + if (!gid_map) + return -ENOMEM; + + /* Create a communication channel so that the parent can tell the child when it finished creating the user + * namespace. */ + unshare_ready_fd = eventfd(0, EFD_CLOEXEC); + if (unshare_ready_fd < 0) + return -errno; + + /* Create a communication channel so that the child can tell the parent a proper error code in case it + * failed. */ + if (pipe2(errno_pipe, O_CLOEXEC) < 0) + return -errno; + + pid = fork(); + if (pid < 0) + return -errno; + + if (pid == 0) { + _cleanup_close_ int fd = -1; + const char *a; + pid_t ppid; + + /* Child process, running in the original user namespace. Let's update the parent's UID/GID map from + * here, after the parent opened its own user namespace. */ + + ppid = getppid(); + errno_pipe[0] = safe_close(errno_pipe[0]); + + /* Wait until the parent unshared the user namespace */ + if (read(unshare_ready_fd, &c, sizeof(c)) < 0) { + r = -errno; + goto child_fail; + } + + /* Disable the setgroups() system call in the child user namespace, for good. */ + a = procfs_file_alloca(ppid, "setgroups"); + fd = open(a, O_WRONLY|O_CLOEXEC); + if (fd < 0) { + if (errno != ENOENT) { + r = -errno; + goto child_fail; + } + + /* If the file is missing the kernel is too old, let's continue anyway. */ + } else { + if (write(fd, "deny\n", 5) < 0) { + r = -errno; + goto child_fail; + } + + fd = safe_close(fd); + } + + /* First write the GID map */ + a = procfs_file_alloca(ppid, "gid_map"); + fd = open(a, O_WRONLY|O_CLOEXEC); + if (fd < 0) { + r = -errno; + goto child_fail; + } + if (write(fd, gid_map, strlen(gid_map)) < 0) { + r = -errno; + goto child_fail; + } + fd = safe_close(fd); + + /* The write the UID map */ + a = procfs_file_alloca(ppid, "uid_map"); + fd = open(a, O_WRONLY|O_CLOEXEC); + if (fd < 0) { + r = -errno; + goto child_fail; + } + if (write(fd, uid_map, strlen(uid_map)) < 0) { + r = -errno; + goto child_fail; + } + + _exit(EXIT_SUCCESS); + + child_fail: + (void) write(errno_pipe[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + errno_pipe[1] = safe_close(errno_pipe[1]); + + if (unshare(CLONE_NEWUSER) < 0) + return -errno; + + /* Let the child know that the namespace is ready now */ + if (write(unshare_ready_fd, &c, sizeof(c)) < 0) + return -errno; + + /* Try to read an error code from the child */ + n = read(errno_pipe[0], &r, sizeof(r)); + if (n < 0) + return -errno; + if (n == sizeof(r)) { /* an error code was sent to us */ + if (r < 0) + return r; + return -EIO; + } + if (n != 0) /* on success we should have read 0 bytes */ + return -EIO; + + r = wait_for_terminate(pid, &si); + if (r < 0) + return r; + pid = 0; + + /* If something strange happened with the child, let's consider this fatal, too */ + if (si.si_code != CLD_EXITED || si.si_status != 0) + return -EIO; + + return 0; +} + +static int setup_runtime_directory( + const ExecContext *context, + const ExecParameters *params, + uid_t uid, + gid_t gid) { + + char **rt; + int r; + + assert(context); + assert(params); + + STRV_FOREACH(rt, context->runtime_directory) { + _cleanup_free_ char *p; + + p = strjoin(params->runtime_prefix, "/", *rt, NULL); + if (!p) + return -ENOMEM; + + r = mkdir_p_label(p, context->runtime_directory_mode); + if (r < 0) + return r; + + r = chmod_and_chown(p, context->runtime_directory_mode, uid, gid); + if (r < 0) + return r; + } + + return 0; +} + +static int setup_smack( + const ExecContext *context, + const ExecCommand *command) { + +#ifdef HAVE_SMACK + int r; + + assert(context); + assert(command); + + if (!mac_smack_use()) + return 0; + + if (context->smack_process_label) { + r = mac_smack_apply_pid(0, context->smack_process_label); + if (r < 0) + return r; + } +#ifdef SMACK_DEFAULT_PROCESS_LABEL + else { + _cleanup_free_ char *exec_label = NULL; + + r = mac_smack_read(command->path, SMACK_ATTR_EXEC, &exec_label); + if (r < 0 && r != -ENODATA && r != -EOPNOTSUPP) + return r; + + r = mac_smack_apply_pid(0, exec_label ? : SMACK_DEFAULT_PROCESS_LABEL); + if (r < 0) + return r; + } +#endif +#endif + + return 0; +} + +static int compile_read_write_paths( + const ExecContext *context, + const ExecParameters *params, + char ***ret) { + + _cleanup_strv_free_ char **l = NULL; + char **rt; + + /* Compile the list of writable paths. This is the combination of the explicitly configured paths, plus all + * runtime directories. */ + + if (strv_isempty(context->read_write_paths) && + strv_isempty(context->runtime_directory)) { + *ret = NULL; /* NOP if neither is set */ + return 0; + } + + l = strv_copy(context->read_write_paths); + if (!l) + return -ENOMEM; + + STRV_FOREACH(rt, context->runtime_directory) { + char *s; + + s = strjoin(params->runtime_prefix, "/", *rt, NULL); + if (!s) + return -ENOMEM; + + if (strv_consume(&l, s) < 0) + return -ENOMEM; + } + + *ret = l; + l = NULL; + + return 0; +} + +static void append_socket_pair(int *array, unsigned *n, int pair[2]) { + assert(array); + assert(n); + + if (!pair) + return; + + if (pair[0] >= 0) + array[(*n)++] = pair[0]; + if (pair[1] >= 0) + array[(*n)++] = pair[1]; +} + +static int close_remaining_fds( + const ExecParameters *params, + ExecRuntime *runtime, + DynamicCreds *dcreds, + int user_lookup_fd, + int socket_fd, + int *fds, unsigned n_fds) { + + unsigned n_dont_close = 0; + int dont_close[n_fds + 12]; + + assert(params); + + if (params->stdin_fd >= 0) + dont_close[n_dont_close++] = params->stdin_fd; + if (params->stdout_fd >= 0) + dont_close[n_dont_close++] = params->stdout_fd; + if (params->stderr_fd >= 0) + dont_close[n_dont_close++] = params->stderr_fd; + + if (socket_fd >= 0) + dont_close[n_dont_close++] = socket_fd; + if (n_fds > 0) { + memcpy(dont_close + n_dont_close, fds, sizeof(int) * n_fds); + n_dont_close += n_fds; + } + + if (runtime) + append_socket_pair(dont_close, &n_dont_close, runtime->netns_storage_socket); + + if (dcreds) { + if (dcreds->user) + append_socket_pair(dont_close, &n_dont_close, dcreds->user->storage_socket); + if (dcreds->group) + append_socket_pair(dont_close, &n_dont_close, dcreds->group->storage_socket); + } + + if (user_lookup_fd >= 0) + dont_close[n_dont_close++] = user_lookup_fd; + + return close_all_fds(dont_close, n_dont_close); +} + +static bool context_has_address_families(const ExecContext *c) { + assert(c); + + return c->address_families_whitelist || + !set_isempty(c->address_families); +} + +static bool context_has_syscall_filters(const ExecContext *c) { + assert(c); + + return c->syscall_whitelist || + !set_isempty(c->syscall_filter) || + !set_isempty(c->syscall_archs); +} + +static bool context_has_no_new_privileges(const ExecContext *c) { + assert(c); + + if (c->no_new_privileges) + return true; + + if (have_effective_cap(CAP_SYS_ADMIN)) /* if we are privileged, we don't need NNP */ + return false; + + return context_has_address_families(c) || /* we need NNP if we have any form of seccomp and are unprivileged */ + c->memory_deny_write_execute || + c->restrict_realtime || + c->protect_kernel_tunables || + context_has_syscall_filters(c); +} + +static int send_user_lookup( + Unit *unit, + int user_lookup_fd, + uid_t uid, + gid_t gid) { + + assert(unit); + + /* Send the resolved UID/GID to PID 1 after we learnt it. We send a single datagram, containing the UID/GID + * data as well as the unit name. Note that we suppress sending this if no user/group to resolve was + * specified. */ + + if (user_lookup_fd < 0) + return 0; + + if (!uid_is_valid(uid) && !gid_is_valid(gid)) + return 0; + + if (writev(user_lookup_fd, + (struct iovec[]) { + { .iov_base = &uid, .iov_len = sizeof(uid) }, + { .iov_base = &gid, .iov_len = sizeof(gid) }, + { .iov_base = unit->id, .iov_len = strlen(unit->id) }}, 3) < 0) + return -errno; + + return 0; +} + static int exec_child( Unit *unit, ExecCommand *command, const ExecContext *context, const ExecParameters *params, ExecRuntime *runtime, + DynamicCreds *dcreds, char **argv, int socket_fd, int *fds, unsigned n_fds, char **files_env, + int user_lookup_fd, int *exit_status) { - _cleanup_strv_free_ char **our_env = NULL, **pam_env = NULL, **final_env = NULL, **final_argv = NULL; + _cleanup_strv_free_ char **our_env = NULL, **pass_env = NULL, **accum_env = NULL, **final_argv = NULL; _cleanup_free_ char *mac_selinux_context_net = NULL; - const char *username = NULL, *home = NULL, *shell = NULL; - unsigned n_dont_close = 0; - int dont_close[n_fds + 4]; + const char *username = NULL, *home = NULL, *shell = NULL, *wd; + dev_t journal_stream_dev = 0; + ino_t journal_stream_ino = 0; + bool needs_mount_namespace; uid_t uid = UID_INVALID; gid_t gid = GID_INVALID; int i, r; - bool needs_mount_namespace; assert(unit); assert(command); @@ -1349,22 +2159,7 @@ static int exec_child( log_forget_fds(); - if (socket_fd >= 0) - dont_close[n_dont_close++] = socket_fd; - if (n_fds > 0) { - memcpy(dont_close + n_dont_close, fds, sizeof(int) * n_fds); - n_dont_close += n_fds; - } - if (params->bus_endpoint_fd >= 0) - dont_close[n_dont_close++] = params->bus_endpoint_fd; - if (runtime) { - if (runtime->netns_storage_socket[0] >= 0) - dont_close[n_dont_close++] = runtime->netns_storage_socket[0]; - if (runtime->netns_storage_socket[1] >= 0) - dont_close[n_dont_close++] = runtime->netns_storage_socket[1]; - } - - r = close_all_fds(dont_close, n_dont_close); + r = close_remaining_fds(params, runtime, dcreds, user_lookup_fd, socket_fd, fds, n_fds); if (r < 0) { *exit_status = EXIT_FDS; return r; @@ -1376,9 +2171,9 @@ static int exec_child( return -errno; } - exec_context_tty_reset(context); + exec_context_tty_reset(context, params); - if (params->confirm_spawn) { + if (params->flags & EXEC_CONFIRM_SPAWN) { char response; r = ask_for_confirmation(&response, argv); @@ -1397,33 +2192,78 @@ static int exec_child( } } - if (context->user) { - username = context->user; - r = get_user_creds(&username, &uid, &gid, &home, &shell); + if (context->dynamic_user && dcreds) { + + /* Make sure we bypass our own NSS module for any NSS checks */ + if (putenv((char*) "SYSTEMD_NSS_DYNAMIC_BYPASS=1") != 0) { + *exit_status = EXIT_USER; + return -errno; + } + + r = dynamic_creds_realize(dcreds, &uid, &gid); if (r < 0) { *exit_status = EXIT_USER; return r; } + + if (!uid_is_valid(uid) || !gid_is_valid(gid)) { + *exit_status = EXIT_USER; + return -ESRCH; + } + + if (dcreds->user) + username = dcreds->user->name; + + } else { + if (context->user) { + username = context->user; + r = get_user_creds_clean(&username, &uid, &gid, &home, &shell); + if (r < 0) { + *exit_status = EXIT_USER; + return r; + } + + /* Note that we don't set $HOME or $SHELL if they are not particularly enlightening anyway + * (i.e. are "/" or "/bin/nologin"). */ + } + + if (context->group) { + const char *g = context->group; + + r = get_group_creds(&g, &gid); + if (r < 0) { + *exit_status = EXIT_GROUP; + return r; + } + } + } + + r = send_user_lookup(unit, user_lookup_fd, uid, gid); + if (r < 0) { + *exit_status = EXIT_USER; + return r; } + user_lookup_fd = safe_close(user_lookup_fd); + /* If a socket is connected to STDIN/STDOUT/STDERR, we * must sure to drop O_NONBLOCK */ if (socket_fd >= 0) - fd_nonblock(socket_fd, false); + (void) fd_nonblock(socket_fd, false); - r = setup_input(context, socket_fd, params->apply_tty_stdin); + r = setup_input(context, params, socket_fd); if (r < 0) { *exit_status = EXIT_STDIN; return r; } - r = setup_output(unit, context, STDOUT_FILENO, socket_fd, basename(command->path), params->apply_tty_stdin, uid, gid); + r = setup_output(unit, context, params, STDOUT_FILENO, socket_fd, basename(command->path), uid, gid, &journal_stream_dev, &journal_stream_ino); if (r < 0) { *exit_status = EXIT_STDOUT; return r; } - r = setup_output(unit, context, STDERR_FILENO, socket_fd, basename(command->path), params->apply_tty_stdin, uid, gid); + r = setup_output(unit, context, params, STDERR_FILENO, socket_fd, basename(command->path), uid, gid, &journal_stream_dev, &journal_stream_ino); if (r < 0) { *exit_status = EXIT_STDERR; return r; @@ -1504,9 +2344,13 @@ static int exec_child( } if (context->utmp_id) - utmp_put_init_process(context->utmp_id, getpid(), getsid(0), context->tty_path); + utmp_put_init_process(context->utmp_id, getpid(), getsid(0), context->tty_path, + context->utmp_mode == EXEC_UTMP_INIT ? INIT_PROCESS : + context->utmp_mode == EXEC_UTMP_LOGIN ? LOGIN_PROCESS : + USER_PROCESS, + username ? "root" : context->user); - if (context->user && is_terminal_input(context->std_input)) { + if (context->user) { r = chown_terminal(STDIN_FILENO, uid); if (r < 0) { *exit_status = EXIT_STDIN; @@ -1514,16 +2358,6 @@ static int exec_child( } } - if (params->bus_endpoint_fd >= 0 && context->bus_endpoint) { - uid_t ep_uid = (uid == UID_INVALID) ? 0 : uid; - - r = bus_kernel_set_endpoint_policy(params->bus_endpoint_fd, ep_uid, context->bus_endpoint); - if (r < 0) { - *exit_status = EXIT_BUS_ENDPOINT; - return r; - } - } - /* If delegation is enabled we'll pass ownership of the cgroup * (but only in systemd's own controller hierarchy!) to the * user of the new process. */ @@ -1543,44 +2377,65 @@ static int exec_child( } if (!strv_isempty(context->runtime_directory) && params->runtime_prefix) { - char **rt; - - STRV_FOREACH(rt, context->runtime_directory) { - _cleanup_free_ char *p; + r = setup_runtime_directory(context, params, uid, gid); + if (r < 0) { + *exit_status = EXIT_RUNTIME_DIRECTORY; + return r; + } + } - p = strjoin(params->runtime_prefix, "/", *rt, NULL); - if (!p) { - *exit_status = EXIT_RUNTIME_DIRECTORY; - return -ENOMEM; - } + r = build_environment( + unit, + context, + params, + n_fds, + home, + username, + shell, + journal_stream_dev, + journal_stream_ino, + &our_env); + if (r < 0) { + *exit_status = EXIT_MEMORY; + return r; + } - r = mkdir_safe_label(p, context->runtime_directory_mode, uid, gid); - if (r < 0) { - *exit_status = EXIT_RUNTIME_DIRECTORY; - return r; - } - } + r = build_pass_environment(context, &pass_env); + if (r < 0) { + *exit_status = EXIT_MEMORY; + return r; } - if (params->apply_permissions) { - r = enforce_groups(context, username, gid); - if (r < 0) { - *exit_status = EXIT_GROUP; - return r; - } + accum_env = strv_env_merge(5, + params->environment, + our_env, + pass_env, + context->environment, + files_env, + NULL); + if (!accum_env) { + *exit_status = EXIT_MEMORY; + return -ENOMEM; } + accum_env = strv_env_clean(accum_env); - umask(context->umask); + (void) umask(context->umask); -#ifdef HAVE_PAM - if (params->apply_permissions && context->pam_name && username) { - r = setup_pam(context->pam_name, username, uid, context->tty_path, &pam_env, fds, n_fds); + if ((params->flags & EXEC_APPLY_PERMISSIONS) && !command->privileged) { + r = setup_smack(context, command); if (r < 0) { - *exit_status = EXIT_PAM; + *exit_status = EXIT_SMACK_PROCESS_LABEL; return r; } + + if (context->pam_name && username) { + r = setup_pam(context->pam_name, username, uid, gid, context->tty_path, &accum_env, fds, n_fds); + if (r < 0) { + *exit_status = EXIT_PAM; + return r; + } + } } -#endif if (context->private_network && runtime && runtime->netns_storage_socket[0] >= 0) { r = setup_netns(runtime->netns_storage_socket); @@ -1591,8 +2446,8 @@ static int exec_child( } needs_mount_namespace = exec_needs_mount_namespace(context, params, runtime); - if (needs_mount_namespace) { + _cleanup_free_ char **rw = NULL; char *tmp = NULL, *var = NULL; /* The runtime struct only contains the parent @@ -1608,15 +2463,22 @@ static int exec_child( var = strjoina(runtime->var_tmp_dir, "/tmp"); } + r = compile_read_write_paths(context, params, &rw); + if (r < 0) { + *exit_status = EXIT_NAMESPACE; + return r; + } + r = setup_namespace( - params->apply_chroot ? context->root_directory : NULL, - context->read_write_dirs, - context->read_only_dirs, - context->inaccessible_dirs, + (params->flags & EXEC_APPLY_CHROOT) ? context->root_directory : NULL, + rw, + context->read_only_paths, + context->inaccessible_paths, tmp, var, - params->bus_endpoint_path, context->private_devices, + context->protect_kernel_tunables, + context->protect_control_groups, context->protect_home, context->protect_system, context->mount_flags); @@ -1634,28 +2496,37 @@ static int exec_child( } } - if (params->apply_chroot) { + if ((params->flags & EXEC_APPLY_PERMISSIONS) && !command->privileged) { + r = enforce_groups(context, username, gid); + if (r < 0) { + *exit_status = EXIT_GROUP; + return r; + } + } + + if (context->working_directory_home) + wd = home; + else if (context->working_directory) + wd = context->working_directory; + else + wd = "/"; + + if (params->flags & EXEC_APPLY_CHROOT) { if (!needs_mount_namespace && context->root_directory) if (chroot(context->root_directory) < 0) { *exit_status = EXIT_CHROOT; return -errno; } - if (chdir(context->working_directory ?: "/") < 0 && + if (chdir(wd) < 0 && !context->working_directory_missing_ok) { *exit_status = EXIT_CHDIR; return -errno; } } else { - _cleanup_free_ char *d = NULL; - - if (asprintf(&d, "%s/%s", - context->root_directory ?: "", - context->working_directory ?: "") < 0) { - *exit_status = EXIT_MEMORY; - return -ENOMEM; - } + const char *d; + d = strjoina(strempty(context->root_directory), "/", strempty(wd)); if (chdir(d) < 0 && !context->working_directory_missing_ok) { *exit_status = EXIT_CHDIR; @@ -1664,7 +2535,12 @@ static int exec_child( } #ifdef HAVE_SELINUX - if (params->apply_permissions && mac_selinux_use() && params->selinux_context_net && socket_fd >= 0) { + if ((params->flags & EXEC_APPLY_PERMISSIONS) && + mac_selinux_use() && + params->selinux_context_net && + socket_fd >= 0 && + !command->privileged) { + r = mac_selinux_get_child_mls_label(socket_fd, command->path, context->selinux_context, &mac_selinux_context_net); if (r < 0) { *exit_status = EXIT_SELINUX_CONTEXT; @@ -1673,6 +2549,14 @@ static int exec_child( } #endif + if ((params->flags & EXEC_APPLY_PERMISSIONS) && context->private_users) { + r = setup_private_users(uid, gid); + if (r < 0) { + *exit_status = EXIT_USER; + return r; + } + } + /* We repeat the fd closing here, to make sure that * nothing is leaked from the PAM modules. Note that * we are more aggressive this time since socket_fd @@ -1689,44 +2573,47 @@ static int exec_child( return r; } - if (params->apply_permissions) { + if ((params->flags & EXEC_APPLY_PERMISSIONS) && !command->privileged) { + + int secure_bits = context->secure_bits; for (i = 0; i < _RLIMIT_MAX; i++) { + if (!context->rlimit[i]) continue; - if (setrlimit_closest(i, context->rlimit[i]) < 0) { + r = setrlimit_closest(i, context->rlimit[i]); + if (r < 0) { *exit_status = EXIT_LIMITS; - return -errno; + return r; } } - if (context->capability_bounding_set_drop) { - r = capability_bounding_set_drop(context->capability_bounding_set_drop, false); - if (r < 0) { - *exit_status = EXIT_CAPABILITIES; - return r; + /* Set the RTPRIO resource limit to 0, but only if nothing else was explicitly requested. */ + if (context->restrict_realtime && !context->rlimit[RLIMIT_RTPRIO]) { + if (setrlimit(RLIMIT_RTPRIO, &RLIMIT_MAKE_CONST(0)) < 0) { + *exit_status = EXIT_LIMITS; + return -errno; } } -#ifdef HAVE_SMACK - if (context->smack_process_label) { - r = mac_smack_apply_pid(0, context->smack_process_label); + if (!cap_test_all(context->capability_bounding_set)) { + r = capability_bounding_set_drop(context->capability_bounding_set, false); if (r < 0) { - *exit_status = EXIT_SMACK_PROCESS_LABEL; + *exit_status = EXIT_CAPABILITIES; return r; } } -#ifdef SMACK_DEFAULT_PROCESS_LABEL - else { - r = mac_smack_apply_pid(0, SMACK_DEFAULT_PROCESS_LABEL); + + /* This is done before enforce_user, but ambient set + * does not survive over setresuid() if keep_caps is not set. */ + if (context->capability_ambient_set != 0) { + r = capability_ambient_set_apply(context->capability_ambient_set, true); if (r < 0) { - *exit_status = EXIT_SMACK_PROCESS_LABEL; + *exit_status = EXIT_CAPABILITIES; return r; } } -#endif -#endif if (context->user) { r = enforce_user(context, uid); @@ -1734,44 +2621,85 @@ static int exec_child( *exit_status = EXIT_USER; return r; } + if (context->capability_ambient_set != 0) { + + /* Fix the ambient capabilities after user change. */ + r = capability_ambient_set_apply(context->capability_ambient_set, false); + if (r < 0) { + *exit_status = EXIT_CAPABILITIES; + return r; + } + + /* If we were asked to change user and ambient capabilities + * were requested, we had to add keep-caps to the securebits + * so that we would maintain the inherited capability set + * through the setresuid(). Make sure that the bit is added + * also to the context secure_bits so that we don't try to + * drop the bit away next. */ + + secure_bits |= 1<<SECURE_KEEP_CAPS; + } } /* PR_GET_SECUREBITS is not privileged, while * PR_SET_SECUREBITS is. So to suppress * potential EPERMs we'll try not to call * PR_SET_SECUREBITS unless necessary. */ - if (prctl(PR_GET_SECUREBITS) != context->secure_bits) - if (prctl(PR_SET_SECUREBITS, context->secure_bits) < 0) { + if (prctl(PR_GET_SECUREBITS) != secure_bits) + if (prctl(PR_SET_SECUREBITS, secure_bits) < 0) { *exit_status = EXIT_SECUREBITS; return -errno; } - if (context->capabilities) - if (cap_set_proc(context->capabilities) < 0) { - *exit_status = EXIT_CAPABILITIES; - return -errno; - } - - if (context->no_new_privileges) + if (context_has_no_new_privileges(context)) if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0) { *exit_status = EXIT_NO_NEW_PRIVILEGES; return -errno; } #ifdef HAVE_SECCOMP - if (context->address_families_whitelist || - !set_isempty(context->address_families)) { - r = apply_address_families(context); + if (context_has_address_families(context)) { + r = apply_address_families(unit, context); if (r < 0) { *exit_status = EXIT_ADDRESS_FAMILIES; return r; } } - if (context->syscall_whitelist || - !set_isempty(context->syscall_filter) || - !set_isempty(context->syscall_archs)) { - r = apply_seccomp(context); + if (context->memory_deny_write_execute) { + r = apply_memory_deny_write_execute(unit, context); + if (r < 0) { + *exit_status = EXIT_SECCOMP; + return r; + } + } + + if (context->restrict_realtime) { + r = apply_restrict_realtime(unit, context); + if (r < 0) { + *exit_status = EXIT_SECCOMP; + return r; + } + } + + if (context->protect_kernel_tunables) { + r = apply_protect_sysctl(unit, context); + if (r < 0) { + *exit_status = EXIT_SECCOMP; + return r; + } + } + + if (context->private_devices) { + r = apply_private_devices(unit, context); + if (r < 0) { + *exit_status = EXIT_SECCOMP; + return r; + } + } + + if (context_has_syscall_filters(context)) { + r = apply_seccomp(unit, context); if (r < 0) { *exit_status = EXIT_SECCOMP; return r; @@ -1804,32 +2732,12 @@ static int exec_child( #endif } - r = build_environment(context, n_fds, params->watchdog_usec, home, username, shell, &our_env); - if (r < 0) { - *exit_status = EXIT_MEMORY; - return r; - } - - final_env = strv_env_merge(5, - params->environment, - our_env, - context->environment, - files_env, - pam_env, - NULL); - if (!final_env) { - *exit_status = EXIT_MEMORY; - return -ENOMEM; - } - - final_argv = replace_env_argv(argv, final_env); + final_argv = replace_env_argv(argv, accum_env); if (!final_argv) { *exit_status = EXIT_MEMORY; return -ENOMEM; } - final_env = strv_env_clean(final_env); - if (_unlikely_(log_get_max_level() >= LOG_DEBUG)) { _cleanup_free_ char *line; @@ -1845,7 +2753,7 @@ static int exec_child( } } - execve(command->path, final_argv, final_env); + execve(command->path, final_argv, accum_env); *exit_status = EXIT_EXEC; return -errno; } @@ -1855,6 +2763,7 @@ int exec_spawn(Unit *unit, const ExecContext *context, const ExecParameters *params, ExecRuntime *runtime, + DynamicCreds *dcreds, pid_t *ret) { _cleanup_strv_free_ char **files_env = NULL; @@ -1903,7 +2812,7 @@ int exec_spawn(Unit *unit, NULL); pid = fork(); if (pid < 0) - return log_unit_error_errno(unit, r, "Failed to fork: %m"); + return log_unit_error_errno(unit, errno, "Failed to fork: %m"); if (pid == 0) { int exit_status; @@ -1913,10 +2822,12 @@ int exec_spawn(Unit *unit, context, params, runtime, + dcreds, argv, socket_fd, fds, n_fds, files_env, + unit->manager->user_lookup_fds[1], &exit_status); if (r < 0) { log_open(); @@ -1961,6 +2872,7 @@ void exec_context_init(ExecContext *c) { c->timer_slack_nsec = NSEC_INFINITY; c->personality = PERSONALITY_INVALID; c->runtime_directory_mode = 0755; + c->capability_bounding_set = CAP_ALL; } void exec_context_done(ExecContext *c) { @@ -1968,80 +2880,40 @@ void exec_context_done(ExecContext *c) { assert(c); - strv_free(c->environment); - c->environment = NULL; - - strv_free(c->environment_files); - c->environment_files = NULL; - - for (l = 0; l < ELEMENTSOF(c->rlimit); l++) { - free(c->rlimit[l]); - c->rlimit[l] = NULL; - } - - free(c->working_directory); - c->working_directory = NULL; - free(c->root_directory); - c->root_directory = NULL; - - free(c->tty_path); - c->tty_path = NULL; + c->environment = strv_free(c->environment); + c->environment_files = strv_free(c->environment_files); + c->pass_environment = strv_free(c->pass_environment); - free(c->syslog_identifier); - c->syslog_identifier = NULL; + for (l = 0; l < ELEMENTSOF(c->rlimit); l++) + c->rlimit[l] = mfree(c->rlimit[l]); - free(c->user); - c->user = NULL; + c->working_directory = mfree(c->working_directory); + c->root_directory = mfree(c->root_directory); + c->tty_path = mfree(c->tty_path); + c->syslog_identifier = mfree(c->syslog_identifier); + c->user = mfree(c->user); + c->group = mfree(c->group); - free(c->group); - c->group = NULL; + c->supplementary_groups = strv_free(c->supplementary_groups); - strv_free(c->supplementary_groups); - c->supplementary_groups = NULL; + c->pam_name = mfree(c->pam_name); - free(c->pam_name); - c->pam_name = NULL; - - if (c->capabilities) { - cap_free(c->capabilities); - c->capabilities = NULL; - } - - strv_free(c->read_only_dirs); - c->read_only_dirs = NULL; - - strv_free(c->read_write_dirs); - c->read_write_dirs = NULL; - - strv_free(c->inaccessible_dirs); - c->inaccessible_dirs = NULL; + c->read_only_paths = strv_free(c->read_only_paths); + c->read_write_paths = strv_free(c->read_write_paths); + c->inaccessible_paths = strv_free(c->inaccessible_paths); if (c->cpuset) CPU_FREE(c->cpuset); - free(c->utmp_id); - c->utmp_id = NULL; - - free(c->selinux_context); - c->selinux_context = NULL; - - free(c->apparmor_profile); - c->apparmor_profile = NULL; + c->utmp_id = mfree(c->utmp_id); + c->selinux_context = mfree(c->selinux_context); + c->apparmor_profile = mfree(c->apparmor_profile); - set_free(c->syscall_filter); - c->syscall_filter = NULL; + c->syscall_filter = set_free(c->syscall_filter); + c->syscall_archs = set_free(c->syscall_archs); + c->address_families = set_free(c->address_families); - set_free(c->syscall_archs); - c->syscall_archs = NULL; - - set_free(c->address_families); - c->address_families = NULL; - - strv_free(c->runtime_directory); - c->runtime_directory = NULL; - - bus_endpoint_free(c->bus_endpoint); - c->bus_endpoint = NULL; + c->runtime_directory = strv_free(c->runtime_directory); } int exec_context_destroy_runtime_directory(ExecContext *c, const char *runtime_prefix) { @@ -2071,11 +2943,9 @@ int exec_context_destroy_runtime_directory(ExecContext *c, const char *runtime_p void exec_command_done(ExecCommand *c) { assert(c); - free(c->path); - c->path = NULL; + c->path = mfree(c->path); - strv_free(c->argv); - c->argv = NULL; + c->argv = strv_free(c->argv); } void exec_command_done_array(ExecCommand *c, unsigned n) { @@ -2133,7 +3003,7 @@ int exec_context_load_environment(Unit *unit, const ExecContext *c, char ***l) { if (fn[0] == '-') { ignore = true; - fn ++; + fn++; } if (!path_is_absolute(fn)) { @@ -2151,7 +3021,7 @@ int exec_context_load_environment(Unit *unit, const ExecContext *c, char ***l) { continue; strv_free(r); - return errno ? -errno : -EINVAL; + return errno > 0 ? -errno : -EINVAL; } count = pglob.gl_pathc; if (count == 0) { @@ -2203,7 +3073,10 @@ int exec_context_load_environment(Unit *unit, const ExecContext *c, char ***l) { static bool tty_may_match_dev_console(const char *tty) { _cleanup_free_ char *active = NULL; - char *console; + char *console; + + if (!tty) + return true; if (startswith(tty, "/dev/")) tty += 5; @@ -2222,11 +3095,14 @@ static bool tty_may_match_dev_console(const char *tty) { } bool exec_context_may_touch_console(ExecContext *ec) { - return (ec->tty_reset || ec->tty_vhangup || ec->tty_vt_disallocate || + + return (ec->tty_reset || + ec->tty_vhangup || + ec->tty_vt_disallocate || is_terminal_input(ec->std_input) || is_terminal_output(ec->std_output) || is_terminal_output(ec->std_error)) && - tty_may_match_dev_console(tty_path(ec)); + tty_may_match_dev_console(exec_context_tty_path(ec)); } static void strv_fprintf(FILE *f, char **l) { @@ -2239,7 +3115,7 @@ static void strv_fprintf(FILE *f, char **l) { } void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { - char **e; + char **e, **d; unsigned i; assert(c); @@ -2253,21 +3129,31 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { "%sRootDirectory: %s\n" "%sNonBlocking: %s\n" "%sPrivateTmp: %s\n" - "%sPrivateNetwork: %s\n" "%sPrivateDevices: %s\n" + "%sProtectKernelTunables: %s\n" + "%sProtectControlGroups: %s\n" + "%sPrivateNetwork: %s\n" + "%sPrivateUsers: %s\n" "%sProtectHome: %s\n" "%sProtectSystem: %s\n" - "%sIgnoreSIGPIPE: %s\n", + "%sIgnoreSIGPIPE: %s\n" + "%sMemoryDenyWriteExecute: %s\n" + "%sRestrictRealtime: %s\n", prefix, c->umask, prefix, c->working_directory ? c->working_directory : "/", prefix, c->root_directory ? c->root_directory : "/", prefix, yes_no(c->non_blocking), prefix, yes_no(c->private_tmp), - prefix, yes_no(c->private_network), prefix, yes_no(c->private_devices), + prefix, yes_no(c->protect_kernel_tunables), + prefix, yes_no(c->protect_control_groups), + prefix, yes_no(c->private_network), + prefix, yes_no(c->private_users), prefix, protect_home_to_string(c->protect_home), prefix, protect_system_to_string(c->protect_system), - prefix, yes_no(c->ignore_sigpipe)); + prefix, yes_no(c->ignore_sigpipe), + prefix, yes_no(c->memory_deny_write_execute), + prefix, yes_no(c->restrict_realtime)); STRV_FOREACH(e, c->environment) fprintf(f, "%sEnvironment: %s\n", prefix, *e); @@ -2275,6 +3161,14 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { STRV_FOREACH(e, c->environment_files) fprintf(f, "%sEnvironmentFile: %s\n", prefix, *e); + STRV_FOREACH(e, c->pass_environment) + fprintf(f, "%sPassEnvironment: %s\n", prefix, *e); + + fprintf(f, "%sRuntimeDirectoryMode: %04o\n", prefix, c->runtime_directory_mode); + + STRV_FOREACH(d, c->runtime_directory) + fprintf(f, "%sRuntimeDirectory: %s\n", prefix, *d); + if (c->nice_set) fprintf(f, "%sNice: %i\n", @@ -2286,9 +3180,12 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { prefix, c->oom_score_adjust); for (i = 0; i < RLIM_NLIMITS; i++) - if (c->rlimit[i]) - fprintf(f, "%s%s: "RLIM_FMT"\n", + if (c->rlimit[i]) { + fprintf(f, "%s%s: " RLIM_FMT "\n", prefix, rlimit_to_string(i), c->rlimit[i]->rlim_max); + fprintf(f, "%s%sSoft: " RLIM_FMT "\n", + prefix, rlimit_to_string(i), c->rlimit[i]->rlim_cur); + } if (c->ioprio_set) { _cleanup_free_ char *class_str = NULL; @@ -2369,14 +3266,6 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { prefix, strna(lvl_str)); } - if (c->capabilities) { - _cleanup_cap_free_charp_ char *t; - - t = cap_to_text(c->capabilities, NULL); - if (t) - fprintf(f, "%sCapabilities: %s\n", prefix, t); - } - if (c->secure_bits) fprintf(f, "%sSecure Bits:%s%s%s%s%s%s\n", prefix, @@ -2387,12 +3276,23 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { (c->secure_bits & 1<<SECURE_NOROOT) ? " noroot" : "", (c->secure_bits & 1<<SECURE_NOROOT_LOCKED) ? "noroot-locked" : ""); - if (c->capability_bounding_set_drop) { + if (c->capability_bounding_set != CAP_ALL) { unsigned long l; fprintf(f, "%sCapabilityBoundingSet:", prefix); for (l = 0; l <= cap_last_cap(); l++) - if (!(c->capability_bounding_set_drop & ((uint64_t) 1ULL << (uint64_t) l))) + if (c->capability_bounding_set & (UINT64_C(1) << l)) + fprintf(f, " %s", strna(capability_to_name(l))); + + fputs("\n", f); + } + + if (c->capability_ambient_set != 0) { + unsigned long l; + fprintf(f, "%sAmbientCapabilities:", prefix); + + for (l = 0; l <= cap_last_cap(); l++) + if (c->capability_ambient_set & (UINT64_C(1) << l)) fprintf(f, " %s", strna(capability_to_name(l))); fputs("\n", f); @@ -2403,6 +3303,8 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { if (c->group) fprintf(f, "%sGroup: %s\n", prefix, c->group); + fprintf(f, "%sDynamicUser: %s\n", prefix, yes_no(c->dynamic_user)); + if (strv_length(c->supplementary_groups) > 0) { fprintf(f, "%sSupplementaryGroups:", prefix); strv_fprintf(f, c->supplementary_groups); @@ -2412,21 +3314,21 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { if (c->pam_name) fprintf(f, "%sPAMName: %s\n", prefix, c->pam_name); - if (strv_length(c->read_write_dirs) > 0) { - fprintf(f, "%sReadWriteDirs:", prefix); - strv_fprintf(f, c->read_write_dirs); + if (strv_length(c->read_write_paths) > 0) { + fprintf(f, "%sReadWritePaths:", prefix); + strv_fprintf(f, c->read_write_paths); fputs("\n", f); } - if (strv_length(c->read_only_dirs) > 0) { - fprintf(f, "%sReadOnlyDirs:", prefix); - strv_fprintf(f, c->read_only_dirs); + if (strv_length(c->read_only_paths) > 0) { + fprintf(f, "%sReadOnlyPaths:", prefix); + strv_fprintf(f, c->read_only_paths); fputs("\n", f); } - if (strv_length(c->inaccessible_dirs) > 0) { - fprintf(f, "%sInaccessibleDirs:", prefix); - strv_fprintf(f, c->inaccessible_dirs); + if (strv_length(c->inaccessible_paths) > 0) { + fprintf(f, "%sInaccessiblePaths:", prefix); + strv_fprintf(f, c->inaccessible_paths); fputs("\n", f); } @@ -2493,7 +3395,7 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { fputc('\n', f); } - if (c->syscall_errno != 0) + if (c->syscall_errno > 0) fprintf(f, "%sSystemCallErrorNumber: %s\n", prefix, strna(errno_to_name(c->syscall_errno))); @@ -2507,7 +3409,7 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { bool exec_context_maintains_privileges(ExecContext *c) { assert(c); - /* Returns true if the process forked off would run run under + /* Returns true if the process forked off would run under * an unchanged UID or as root. */ if (!c->user) @@ -2543,7 +3445,7 @@ void exec_status_exit(ExecStatus *s, ExecContext *context, pid_t pid, int code, if (context->utmp_id) utmp_put_dead_process(context->utmp_id, pid, code, status); - exec_context_tty_reset(context); + exec_context_tty_reset(context, NULL); } } @@ -2562,12 +3464,12 @@ void exec_status_dump(ExecStatus *s, FILE *f, const char *prefix) { "%sPID: "PID_FMT"\n", prefix, s->pid); - if (s->start_timestamp.realtime > 0) + if (dual_timestamp_is_set(&s->start_timestamp)) fprintf(f, "%sStart Timestamp: %s\n", prefix, format_timestamp(buf, sizeof(buf), s->start_timestamp.realtime)); - if (s->exit_timestamp.realtime > 0) + if (dual_timestamp_is_set(&s->exit_timestamp)) fprintf(f, "%sExit Timestamp: %s\n" "%sExit Code: %s\n" @@ -2701,7 +3603,7 @@ int exec_command_append(ExecCommand *c, const char *path, ...) { if (!l) return -ENOMEM; - r = strv_extend_strv(&c->argv, l); + r = strv_extend_strv(&c->argv, l, false); if (r < 0) return r; @@ -2742,7 +3644,7 @@ int exec_runtime_make(ExecRuntime **rt, ExecContext *c, const char *id) { return r; if (c->private_network && (*rt)->netns_storage_socket[0] < 0) { - if (socketpair(AF_UNIX, SOCK_DGRAM, 0, (*rt)->netns_storage_socket) < 0) + if (socketpair(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC, 0, (*rt)->netns_storage_socket) < 0) return -errno; } @@ -2954,3 +3856,11 @@ static const char* const exec_output_table[_EXEC_OUTPUT_MAX] = { }; DEFINE_STRING_TABLE_LOOKUP(exec_output, ExecOutput); + +static const char* const exec_utmp_mode_table[_EXEC_UTMP_MODE_MAX] = { + [EXEC_UTMP_INIT] = "init", + [EXEC_UTMP_LOGIN] = "login", + [EXEC_UTMP_USER] = "user", +}; + +DEFINE_STRING_TABLE_LOOKUP(exec_utmp_mode, ExecUtmpMode); |