/*-*- Mode: C; c-basic-offset: 8 -*-*/
+/***
+ This file is part of systemd.
+
+ Copyright 2010 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2 of the License, or
+ (at your option) any later version.
+
+ systemd is distributed in the hope that it will be useful, but
+ WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
#include <assert.h>
#include <dirent.h>
#include <errno.h>
#include <signal.h>
#include <sys/socket.h>
#include <sys/un.h>
+#include <sys/prctl.h>
+#include <linux/sched.h>
+#include <sys/types.h>
+#include <sys/stat.h>
#include "execute.h"
#include "strv.h"
#include "util.h"
#include "log.h"
#include "ioprio.h"
+#include "securebits.h"
static int close_fds(int except[], unsigned n_except) {
DIR *d;
return 0;
}
-static int flags_fds(int fds[], unsigned n_fds) {
+static int flags_fds(int fds[], unsigned n_fds, bool nonblock) {
unsigned i;
if (n_fds <= 0)
assert(fds);
- /* Drops O_NONBLOCK and FD_CLOEXEC from the file flags */
+ /* Drops/Sets O_NONBLOCK and FD_CLOEXEC from the file flags */
for (i = 0; i < n_fds; i++) {
int flags;
if ((flags = fcntl(fds[i], F_GETFL, 0)) < 0)
return -errno;
- /* Since we are at it, let's make sure that nobody
- * forgot setting O_NONBLOCK for all our fds */
+ if (nonblock)
+ flags |= O_NONBLOCK;
+ else
+ flags &= ~O_NONBLOCK;
- if (fcntl(fds[i], F_SETFL, flags &~O_NONBLOCK) < 0)
+ if (fcntl(fds[i], F_SETFL, flags) < 0)
return -errno;
+ /* We unconditionally drop FD_CLOEXEC from the fds,
+ * since after all we want to pass these fds to our
+ * children */
if ((flags = fcntl(fds[i], F_GETFD, 0)) < 0)
return -errno;
- /* Also make sure nobody forgot O_CLOEXEC for all our
- * fds */
if (fcntl(fds[i], F_SETFD, flags &~FD_CLOEXEC) < 0)
return -errno;
}
switch (context->output) {
- case EXEC_CONSOLE:
+ case EXEC_OUTPUT_CONSOLE:
return 0;
- case EXEC_NULL:
+ case EXEC_OUTPUT_NULL:
- if ((r = replace_null_fd(STDIN_FILENO, O_RDONLY)) < 0 ||
- (r = replace_null_fd(STDOUT_FILENO, O_WRONLY)) < 0 ||
+ if ((r = replace_null_fd(STDOUT_FILENO, O_WRONLY)) < 0 ||
(r = replace_null_fd(STDERR_FILENO, O_WRONLY)) < 0)
return r;
return 0;
- case EXEC_KERNEL:
- case EXEC_SYSLOG: {
+ case EXEC_OUTPUT_KERNEL:
+ case EXEC_OUTPUT_SYSLOG: {
int fd;
union {
struct sockaddr_un un;
} sa;
- if ((r = replace_null_fd(STDIN_FILENO, O_RDONLY)) < 0)
- return r;
-
close_nointr(STDOUT_FILENO);
close_nointr(STDERR_FILENO);
"%s\n"
"%i\n"
"%s\n",
- context->output == EXEC_KERNEL ? "kmsg" : "syslog",
+ context->output == EXEC_OUTPUT_KERNEL ? "kmsg" : "syslog",
context->syslog_priority,
context->syslog_identifier ? context->syslog_identifier : ident);
return 0;
}
+
+ default:
+ assert_not_reached("Unknown output type");
}
+}
+
+static int setup_input(const ExecContext *context) {
+ int r;
+
+ assert(context);
+
+ switch (context->input) {
+
+ case EXEC_INPUT_CONSOLE:
+ return 0;
+
+ case EXEC_INPUT_NULL:
+ if ((r = replace_null_fd(STDIN_FILENO, O_RDONLY)) < 0)
+ return r;
+
+ return 0;
- assert_not_reached("Unknown logging type");
+ default:
+ assert_not_reached("Unknown input type");
+ }
}
int exec_spawn(const ExecCommand *command, const ExecContext *context, int *fds, unsigned n_fds, pid_t *ret) {
umask(context->umask);
+ if (setup_input(context) < 0) {
+ r = EXIT_INPUT;
+ goto fail;
+ }
+
if (setup_output(context, file_name_from_path(command->path)) < 0) {
r = EXIT_OUTPUT;
goto fail;
goto fail;
}
+ if (context->cpu_sched_set) {
+ struct sched_param param;
+
+ zero(param);
+ param.sched_priority = context->cpu_sched_priority;
+
+ if (sched_setscheduler(0, context->cpu_sched_policy |
+ (context->cpu_sched_reset_on_fork ? SCHED_RESET_ON_FORK : 0), ¶m) < 0) {
+ r = EXIT_SETSCHEDULER;
+ goto fail;
+ }
+ }
+
+ if (context->cpu_affinity_set)
+ if (sched_setaffinity(0, sizeof(context->cpu_affinity), &context->cpu_affinity) < 0) {
+ r = EXIT_CPUAFFINITY;
+ goto fail;
+ }
+
if (context->ioprio_set)
if (ioprio_set(IOPRIO_WHO_PROCESS, 0, context->ioprio) < 0) {
r = EXIT_IOPRIO;
goto fail;
}
+ if (context->timer_slack_ns_set)
+ if (prctl(PR_SET_TIMERSLACK, context->timer_slack_ns_set) < 0) {
+ r = EXIT_TIMERSLACK;
+ goto fail;
+ }
+
if (close_fds(fds, n_fds) < 0 ||
shift_fds(fds, n_fds) < 0 ||
- flags_fds(fds, n_fds) < 0) {
+ flags_fds(fds, n_fds, context->non_blocking) < 0) {
r = EXIT_FDS;
goto fail;
}
}
}
+ if (context->secure_bits) {
+ if (prctl(PR_SET_SECUREBITS, context->secure_bits) < 0) {
+ r = EXIT_SECUREBITS;
+ goto fail;
+ }
+ }
+
if (n_fds > 0) {
char a[64], b[64];
char *listen_env[3] = {
assert(c);
c->umask = 0002;
- cap_clear(c->capabilities);
- c->capabilities_set = false;
c->oom_adjust = 0;
c->oom_adjust_set = false;
c->nice = 0;
c->nice_set = false;
c->ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 0);
c->ioprio_set = false;
+ c->cpu_sched_policy = SCHED_OTHER;
+ c->cpu_sched_priority = 0;
+ c->cpu_sched_set = false;
+ CPU_ZERO(&c->cpu_affinity);
+ c->cpu_affinity_set = false;
+ c->input = 0;
c->output = 0;
c->syslog_priority = LOG_DAEMON|LOG_INFO;
+
+ c->secure_bits = 0;
+ c->capability_bounding_set_drop = 0;
}
void exec_context_done(ExecContext *c) {
strv_free(c->supplementary_groups);
c->supplementary_groups = NULL;
+
+ if (c->capabilities) {
+ cap_free(c->capabilities);
+ c->capabilities = NULL;
+ }
}
void exec_command_free_list(ExecCommand *c) {
}
void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) {
-
- static const char * const table[] = {
- [IOPRIO_CLASS_NONE] = "none",
- [IOPRIO_CLASS_RT] = "realtime",
- [IOPRIO_CLASS_BE] = "best-effort",
- [IOPRIO_CLASS_IDLE] = "idle"
- };
+ char ** e;
+ unsigned i;
assert(c);
assert(f);
prefix = "";
fprintf(f,
- "%sUmask: %04o\n"
- "%sWorking Directory: %s\n"
- "%sRoot Directory: %s\n",
+ "%sUMask: %04o\n"
+ "%sWorkingDirectory: %s\n"
+ "%sRootDirectory: %s\n"
+ "%sNonBlocking: %s\n",
prefix, c->umask,
prefix, c->working_directory ? c->working_directory : "/",
- prefix, c->root_directory ? c->root_directory : "/");
+ prefix, c->root_directory ? c->root_directory : "/",
+ prefix, yes_no(c->non_blocking));
+
+ if (c->environment)
+ for (e = c->environment; *e; e++)
+ fprintf(f, "%sEnvironment: %s\n", prefix, *e);
if (c->nice_set)
fprintf(f,
"%sOOMAdjust: %i\n",
prefix, c->oom_adjust);
+ for (i = 0; i < RLIM_NLIMITS; i++)
+ if (c->rlimit[i])
+ fprintf(f, "%s%s: %llu\n", prefix, rlimit_to_string(i), (unsigned long long) c->rlimit[i]->rlim_max);
+
if (c->ioprio_set)
fprintf(f,
"%sIOSchedulingClass: %s\n"
"%sIOPriority: %i\n",
- prefix, table[IOPRIO_PRIO_CLASS(c->ioprio)],
+ prefix, ioprio_class_to_string(IOPRIO_PRIO_CLASS(c->ioprio)),
prefix, (int) IOPRIO_PRIO_DATA(c->ioprio));
+
+ if (c->cpu_sched_set)
+ fprintf(f,
+ "%sCPUSchedulingPolicy: %s\n"
+ "%sCPUSchedulingPriority: %i\n"
+ "%sCPUSchedulingResetOnFork: %s\n",
+ prefix, sched_policy_to_string(c->cpu_sched_policy),
+ prefix, c->cpu_sched_priority,
+ prefix, yes_no(c->cpu_sched_reset_on_fork));
+
+ if (c->cpu_affinity_set) {
+ fprintf(f, "%sCPUAffinity:", prefix);
+ for (i = 0; i < CPU_SETSIZE; i++)
+ if (CPU_ISSET(i, &c->cpu_affinity))
+ fprintf(f, " %i", i);
+ fputs("\n", f);
+ }
+
+ if (c->timer_slack_ns_set)
+ fprintf(f, "%sTimerSlackNS: %lu\n", prefix, c->timer_slack_ns);
+
+ fprintf(f,
+ "%sInput: %s\n"
+ "%sOutput: %s\n",
+ prefix, exec_input_to_string(c->input),
+ prefix, exec_output_to_string(c->output));
+
+ if (c->output == EXEC_OUTPUT_SYSLOG || c->output == EXEC_OUTPUT_KERNEL)
+ fprintf(f,
+ "%sSyslogFacility: %s\n"
+ "%sSyslogLevel: %s\n",
+ prefix, log_facility_to_string(LOG_FAC(c->syslog_priority)),
+ prefix, log_level_to_string(LOG_PRI(c->syslog_priority)));
+
+ if (c->capabilities) {
+ char *t;
+ if ((t = cap_to_text(c->capabilities, NULL))) {
+ fprintf(f, "%sCapabilities: %s\n",
+ prefix, t);
+ cap_free(t);
+ }
+ }
+
+ if (c->secure_bits)
+ fprintf(f, "%sSecure Bits:%s%s%s%s%s%s\n",
+ prefix,
+ (c->secure_bits & SECURE_KEEP_CAPS) ? " keep-caps" : "",
+ (c->secure_bits & SECURE_KEEP_CAPS_LOCKED) ? " keep-caps-locked" : "",
+ (c->secure_bits & SECURE_NO_SETUID_FIXUP) ? " no-setuid-fixup" : "",
+ (c->secure_bits & SECURE_NO_SETUID_FIXUP_LOCKED) ? " no-setuid-fixup-locked" : "",
+ (c->secure_bits & SECURE_NOROOT) ? " noroot" : "",
+ (c->secure_bits & SECURE_NOROOT_LOCKED) ? "noroot-locked" : "");
+
+ if (c->capability_bounding_set_drop) {
+ fprintf(f, "%sCapabilityBoundingSetDrop:", prefix);
+
+ for (i = 0; i <= CAP_LAST_CAP; i++)
+ if (c->capability_bounding_set_drop & (1 << i)) {
+ char *t;
+
+ if ((t = cap_to_name(i))) {
+ fprintf(f, " %s", t);
+ free(t);
+ }
+ }
+
+ fputs("\n", f);
+ }
+
+ if (c->user)
+ fprintf(f, "%sUser: %s", prefix, c->user);
+ if (c->group)
+ fprintf(f, "%sGroup: %s", prefix, c->group);
+
+ if (c->supplementary_groups) {
+ char **g;
+
+ fprintf(f, "%sSupplementaryGroups:", prefix);
+
+ STRV_FOREACH(g, c->supplementary_groups)
+ fprintf(f, " %s", *g);
+
+ fputs("\n", f);
+ }
}
void exec_status_fill(ExecStatus *s, pid_t pid, int code, int status) {
LIST_FOREACH(command, c, c)
exec_command_dump(c, f, prefix);
}
+
+void exec_command_append_list(ExecCommand **l, ExecCommand *e) {
+ ExecCommand *end;
+
+ assert(l);
+ assert(e);
+
+ if (*l) {
+ /* It's kinda important that we keep the order here */
+ LIST_FIND_TAIL(ExecCommand, command, *l, end);
+ LIST_INSERT_AFTER(ExecCommand, command, *l, end, e);
+ } else
+ *l = e;
+}
+
+static const char* const exec_output_table[_EXEC_OUTPUT_MAX] = {
+ [EXEC_OUTPUT_CONSOLE] = "console",
+ [EXEC_OUTPUT_NULL] = "null",
+ [EXEC_OUTPUT_SYSLOG] = "syslog",
+ [EXEC_OUTPUT_KERNEL] = "kernel"
+};
+
+DEFINE_STRING_TABLE_LOOKUP(exec_output, ExecOutput);
+
+static const char* const exec_input_table[_EXEC_INPUT_MAX] = {
+ [EXEC_INPUT_NULL] = "null",
+ [EXEC_INPUT_CONSOLE] = "console"
+};
+
+DEFINE_STRING_TABLE_LOOKUP(exec_input, ExecInput);