-/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
-
/***
This file is part of systemd.
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
-#include <sys/mount.h>
#include <errno.h>
-#include <sys/stat.h>
+#include <ftw.h>
#include <stdlib.h>
-#include <string.h>
-#include <libgen.h>
-#include <assert.h>
+#include <sys/mount.h>
#include <unistd.h>
-#include <ftw.h>
-#include "mount-setup.h"
-#include "dev-setup.h"
-#include "log.h"
-#include "macro.h"
-#include "util.h"
+#include "alloc-util.h"
+//#include "bus-util.h"
+#include "cgroup-util.h"
+//#include "dev-setup.h"
+//#include "efivars.h"
+#include "fs-util.h"
#include "label.h"
-#include "set.h"
-#include "strv.h"
+//#include "log.h"
+#include "macro.h"
+//#include "missing.h"
#include "mkdir.h"
+#include "mount-setup.h"
+#include "mount-util.h"
#include "path-util.h"
-
-#ifndef TTY_GID
-#define TTY_GID 5
-#endif
+//#include "set.h"
+//#include "smack-util.h"
+//#include "strv.h"
+#include "user-util.h"
+//#include "util.h"
+#include "virt.h"
+
+/// Additional includes needed by elogind
+#include "string-util.h"
+
+typedef enum MountMode {
+ MNT_NONE = 0,
+ MNT_FATAL = 1 << 0,
+ MNT_IN_CONTAINER = 1 << 1,
+} MountMode;
typedef struct MountPoint {
const char *what;
const char *type;
const char *options;
unsigned long flags;
- bool fatal;
+ bool (*condition_fn)(void);
+ MountMode mode;
} MountPoint;
/* The first three entries we might need before SELinux is up. The
* fourth (securityfs) is needed by IMA to load a custom policy. The
- * other ones we can delay until SELinux and IMA are loaded. */
+ * other ones we can delay until SELinux and IMA are loaded. When
+ * SMACK is enabled we need smackfs, too, so it's a fifth one. */
+#ifdef HAVE_SMACK
+#define N_EARLY_MOUNT 5
+#else
#define N_EARLY_MOUNT 4
+#endif
static const MountPoint mount_table[] = {
- { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
- { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true },
- { "devtmpfs", "/dev", "devtmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME, true },
- { "securityfs", "/sys/kernel/security", "securityfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, false },
- { "tmpfs", "/dev/shm", "tmpfs", "mode=1777", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true },
- { "devpts", "/dev/pts", "devpts", "mode=620,gid=" STRINGIFY(TTY_GID), MS_NOSUID|MS_NOEXEC, false },
- { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true },
- { "tmpfs", "/sys/fs/cgroup", "tmpfs", "mode=755", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, false },
- { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd", MS_NOSUID|MS_NOEXEC|MS_NODEV, false },
+#if 0 /// UNNEEDED by elogind
+ { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ NULL, MNT_FATAL|MNT_IN_CONTAINER },
+ { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ NULL, MNT_FATAL|MNT_IN_CONTAINER },
+ { "devtmpfs", "/dev", "devtmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME,
+ NULL, MNT_FATAL|MNT_IN_CONTAINER },
+ { "securityfs", "/sys/kernel/security", "securityfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ NULL, MNT_NONE },
+#ifdef HAVE_SMACK
+ { "smackfs", "/sys/fs/smackfs", "smackfs", "smackfsdef=*", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ mac_smack_use, MNT_FATAL },
+ { "tmpfs", "/dev/shm", "tmpfs", "mode=1777,smackfsroot=*", MS_NOSUID|MS_NODEV|MS_STRICTATIME,
+ mac_smack_use, MNT_FATAL },
+#endif
+ { "tmpfs", "/dev/shm", "tmpfs", "mode=1777", MS_NOSUID|MS_NODEV|MS_STRICTATIME,
+ NULL, MNT_FATAL|MNT_IN_CONTAINER },
+ { "devpts", "/dev/pts", "devpts", "mode=620,gid=" STRINGIFY(TTY_GID), MS_NOSUID|MS_NOEXEC,
+ NULL, MNT_IN_CONTAINER },
+#ifdef HAVE_SMACK
+ { "tmpfs", "/run", "tmpfs", "mode=755,smackfsroot=*", MS_NOSUID|MS_NODEV|MS_STRICTATIME,
+ mac_smack_use, MNT_FATAL },
+#endif
+ { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME,
+ NULL, MNT_FATAL|MNT_IN_CONTAINER },
+ { "cgroup", "/sys/fs/cgroup", "cgroup2", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_unified_wanted, MNT_FATAL|MNT_IN_CONTAINER },
+#endif // 0
+ { "tmpfs", "/sys/fs/cgroup", "tmpfs", "mode=755", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME,
+ cg_is_legacy_wanted, MNT_FATAL|MNT_IN_CONTAINER },
+#if 0 /// UNNEEDED by elogind
+ { "cgroup", "/sys/fs/cgroup/systemd", "cgroup2", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_unified_systemd_controller_wanted, MNT_IN_CONTAINER },
+ { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd,xattr", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_legacy_systemd_controller_wanted, MNT_IN_CONTAINER },
+ { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_legacy_systemd_controller_wanted, MNT_IN_CONTAINER },
+#else
+ { "cgroup", "/sys/fs/cgroup/elogind", "cgroup", "none,name=elogind,xattr", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_legacy_systemd_controller_wanted, MNT_IN_CONTAINER },
+ { "cgroup", "/sys/fs/cgroup/elogind", "cgroup", "none,name=elogind", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ cg_is_legacy_systemd_controller_wanted, MNT_IN_CONTAINER },
+#endif // 0
+#if 0 /// UNNEEDED by elogind
+ { "pstore", "/sys/fs/pstore", "pstore", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ NULL, MNT_NONE },
+#ifdef ENABLE_EFI
+ { "efivarfs", "/sys/firmware/efi/efivars", "efivarfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ is_efi_boot, MNT_NONE },
+#endif
+#endif // 0
};
+#if 0 /// UNNEEDED by elogind
/* These are API file systems that might be mounted by other software,
* we just list them here so that we know that we should ignore them */
static const char ignore_paths[] =
/* SELinux file systems */
"/sys/fs/selinux\0"
- "/selinux\0"
- /* Legacy cgroup mount points */
- "/dev/cgroup\0"
- "/cgroup\0"
- /* Legacy kernel file system */
- "/proc/bus/usb\0"
/* Container bind mounts */
"/proc/sys\0"
"/dev/console\0"
- "/proc/kmsg\0"
- "/etc/localtime\0"
- "/etc/timezone\0"
- "/etc/machine-id\0";
+ "/proc/kmsg\0";
bool mount_point_is_api(const char *path) {
unsigned i;
return false;
}
+#endif // 0
static int mount_one(const MountPoint *p, bool relabel) {
int r;
assert(p);
+ if (p->condition_fn && !p->condition_fn())
+ return 0;
+
/* Relabel first, just in case */
if (relabel)
- label_fix(p->where, true);
-
- if ((r = path_is_mount_point(p->where, true)) < 0)
- return r;
+ (void) label_fix(p->where, true, true);
+ r = path_is_mount_point(p->where, NULL, AT_SYMLINK_FOLLOW);
+ if (r < 0 && r != -ENOENT) {
+ log_full_errno((p->mode & MNT_FATAL) ? LOG_ERR : LOG_DEBUG, r, "Failed to determine whether %s is a mount point: %m", p->where);
+ return (p->mode & MNT_FATAL) ? r : 0;
+ }
if (r > 0)
return 0;
+ /* Skip securityfs in a container */
+ if (!(p->mode & MNT_IN_CONTAINER) && detect_container() > 0)
+ return 0;
+
/* The access mode here doesn't really matter too much, since
* the mounted file system will take precedence anyway. */
- mkdir_p_label(p->where, 0755);
+ if (relabel)
+ (void) mkdir_p_label(p->where, 0755);
+ else
+ (void) mkdir_p(p->where, 0755);
log_debug("Mounting %s to %s of type %s with options %s.",
p->what,
p->type,
p->flags,
p->options) < 0) {
- log_full(p->fatal ? LOG_ERR : LOG_DEBUG, "Failed to mount %s: %s", p->where, strerror(errno));
- return p->fatal ? -errno : 0;
+ log_full_errno((p->mode & MNT_FATAL) ? LOG_ERR : LOG_DEBUG, errno, "Failed to mount %s at %s: %m", p->type, p->where);
+ return (p->mode & MNT_FATAL) ? -errno : 0;
}
/* Relabel again, since we now mounted something fresh here */
if (relabel)
- label_fix(p->where, false);
+ (void) label_fix(p->where, false, false);
return 1;
}
-int mount_setup_early(void) {
+static int mount_points_setup(unsigned n, bool loaded_policy) {
unsigned i;
int r = 0;
- assert_cc(N_EARLY_MOUNT <= ELEMENTSOF(mount_table));
-
- /* Do a minimal mount of /proc and friends to enable the most
- * basic stuff, such as SELinux */
- for (i = 0; i < N_EARLY_MOUNT; i ++) {
+ for (i = 0; i < n; i ++) {
int j;
- j = mount_one(mount_table + i, false);
- if (r == 0)
+ j = mount_one(mount_table + i, loaded_policy);
+ if (j != 0 && r >= 0)
r = j;
}
return r;
}
+#if 0 /// UNNEEDED by elogind
+int mount_setup_early(void) {
+ assert_cc(N_EARLY_MOUNT <= ELEMENTSOF(mount_table));
+
+ /* Do a minimal mount of /proc and friends to enable the most
+ * basic stuff, such as SELinux */
+ return mount_points_setup(N_EARLY_MOUNT, false);
+}
+
int mount_cgroup_controllers(char ***join_controllers) {
+ _cleanup_set_free_free_ Set *controllers = NULL;
int r;
- FILE *f;
- char buf[LINE_MAX];
- Set *controllers;
- /* Mount all available cgroup controllers that are built into the kernel. */
-
- f = fopen("/proc/cgroups", "re");
- if (!f) {
- log_error("Failed to enumerate cgroup controllers: %m");
+ if (!cg_is_legacy_wanted())
return 0;
- }
-
- controllers = set_new(string_hash_func, string_compare_func);
- if (!controllers) {
- r = -ENOMEM;
- log_error("Failed to allocate controller set.");
- goto finish;
- }
- /* Ignore the header line */
- (void) fgets(buf, sizeof(buf), f);
-
- for (;;) {
- char *controller;
- int enabled = 0;
-
- if (fscanf(f, "%ms %*i %*i %i", &controller, &enabled) != 2) {
-
- if (feof(f))
- break;
-
- log_error("Failed to parse /proc/cgroups.");
- r = -EIO;
- goto finish;
- }
+ /* Mount all available cgroup controllers that are built into the kernel. */
- if (!enabled) {
- free(controller);
- continue;
- }
+ controllers = set_new(&string_hash_ops);
+ if (!controllers)
+ return log_oom();
- r = set_put(controllers, controller);
- if (r < 0) {
- log_error("Failed to add controller to set.");
- free(controller);
- goto finish;
- }
- }
+ r = cg_kernel_controllers(controllers);
+ if (r < 0)
+ return log_error_errno(r, "Failed to enumerate cgroup controllers: %m");
for (;;) {
- MountPoint p;
- char *controller, *where, *options;
+ _cleanup_free_ char *options = NULL, *controller = NULL, *where = NULL;
+ MountPoint p = {
+ .what = "cgroup",
+ .type = "cgroup",
+ .flags = MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ .mode = MNT_IN_CONTAINER,
+ };
char ***k = NULL;
controller = set_steal_first(controllers);
for (i = *k, j = *k; *i; i++) {
if (!streq(*i, controller)) {
- char *t;
+ _cleanup_free_ char *t;
t = set_remove(controllers, *i);
if (!t) {
free(*i);
continue;
}
- free(t);
}
*(j++) = *i;
*j = NULL;
options = strv_join(*k, ",");
- if (!options) {
- log_error("Failed to join options");
- free(controller);
- r = -ENOMEM;
- goto finish;
- }
-
+ if (!options)
+ return log_oom();
} else {
options = controller;
controller = NULL;
}
where = strappend("/sys/fs/cgroup/", options);
- if (!where) {
- log_error("Failed to build path");
- free(options);
- r = -ENOMEM;
- goto finish;
- }
+ if (!where)
+ return log_oom();
- zero(p);
- p.what = "cgroup";
p.where = where;
- p.type = "cgroup";
p.options = options;
- p.flags = MS_NOSUID|MS_NOEXEC|MS_NODEV;
- p.fatal = false;
r = mount_one(&p, true);
- free(controller);
- free(where);
-
- if (r < 0) {
- free(options);
- goto finish;
- }
+ if (r < 0)
+ return r;
if (r > 0 && k && *k) {
char **i;
for (i = *k; *i; i++) {
- char *t;
+ _cleanup_free_ char *t = NULL;
t = strappend("/sys/fs/cgroup/", *i);
- if (!t) {
- log_error("Failed to build path");
- r = -ENOMEM;
- free(options);
- goto finish;
- }
+ if (!t)
+ return log_oom();
r = symlink(options, t);
- free(t);
-
- if (r < 0 && errno != EEXIST) {
- log_error("Failed to create symlink: %m");
- r = -errno;
- free(options);
- goto finish;
- }
+ if (r >= 0) {
+#ifdef SMACK_RUN_LABEL
+ _cleanup_free_ char *src;
+ src = strappend("/sys/fs/cgroup/", options);
+ if (!src)
+ return log_oom();
+ r = mac_smack_copy(t, src);
+ if (r < 0 && r != -EOPNOTSUPP)
+ return log_error_errno(r, "Failed to copy smack label from %s to %s: %m", src, t);
+#endif
+ } else if (errno != EEXIST)
+ return log_error_errno(errno, "Failed to create symlink %s: %m", t);
}
}
-
- free(options);
}
- r = 0;
-
-finish:
- set_free_free(controllers);
+ /* Now that we mounted everything, let's make the tmpfs the
+ * cgroup file systems are mounted into read-only. */
+ (void) mount("tmpfs", "/sys/fs/cgroup", "tmpfs", MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755");
- fclose(f);
-
- return r;
+ return 0;
}
+#if defined(HAVE_SELINUX) || defined(HAVE_SMACK)
static int nftw_cb(
const char *fpath,
const struct stat *sb,
if (_unlikely_(ftwbuf->level == 0))
return FTW_CONTINUE;
- label_fix(fpath, true);
+ label_fix(fpath, false, false);
/* /run/initramfs is static data and big, no need to
* dynamically relabel its contents at boot... */
return FTW_CONTINUE;
};
+#endif
+#endif // 0
int mount_setup(bool loaded_policy) {
+ int r = 0;
- static const char relabel[] =
- "/run/initramfs/root-fsck\0"
- "/run/initramfs/shutdown\0";
-
- int r;
- unsigned i;
- const char *j;
-
- for (i = 0; i < ELEMENTSOF(mount_table); i ++) {
- r = mount_one(mount_table + i, true);
-
- if (r < 0)
- return r;
- }
+ r = mount_points_setup(ELEMENTSOF(mount_table), loaded_policy);
+ if (r < 0)
+ return r;
+#if 0 /// elogind does not control /, /dev, /run and /run/systemd/* are setup elsewhere.
+#if defined(HAVE_SELINUX) || defined(HAVE_SMACK)
/* Nodes in devtmpfs and /run need to be manually updated for
* the appropriate labels, after mounting. The other virtual
* API file systems like /sys and /proc do not need that, they
before_relabel = now(CLOCK_MONOTONIC);
nftw("/dev", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL);
+ nftw("/dev/shm", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL);
nftw("/run", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL);
- /* Explicitly relabel these */
- NULSTR_FOREACH(j, relabel)
- label_fix(j, true);
-
after_relabel = now(CLOCK_MONOTONIC);
log_info("Relabelled /dev and /run in %s.",
- format_timespan(timespan, sizeof(timespan), after_relabel - before_relabel));
+ format_timespan(timespan, sizeof(timespan), after_relabel - before_relabel, 0));
}
+#endif
/* Create a few default symlinks, which are normally created
* by udevd, but some scripts might need them before we start
* udevd. */
- dev_setup();
-
- /* Create a few directories we always want around */
- mkdir_label("/run/systemd", 0755);
- mkdir_label("/run/systemd/system", 0755);
+ dev_setup(NULL, UID_INVALID, GID_INVALID);
+
+ /* Mark the root directory as shared in regards to mount propagation. The kernel defaults to "private", but we
+ * think it makes more sense to have a default of "shared" so that nspawn and the container tools work out of
+ * the box. If specific setups need other settings they can reset the propagation mode to private if
+ * needed. Note that we set this only when we are invoked directly by the kernel. If we are invoked by a
+ * container manager we assume the container manager knows what it is doing (for example, because it set up
+ * some directories with different propagation modes). */
+ if (detect_container() <= 0)
+ if (mount(NULL, "/", NULL, MS_REC|MS_SHARED, NULL) < 0)
+ log_warning_errno(errno, "Failed to set up the root directory for shared mount propagation: %m");
+
+#endif // 0
+ /* Create a few directories we always want around, Note that sd_booted() checks for /run/systemd/system, so
+ * this mkdir really needs to stay for good, otherwise software that copied sd-daemon.c into their sources will
+ * misdetect systemd. */
+ (void) mkdir_label("/run/systemd", 0755);
+ (void) mkdir_label("/run/systemd/system", 0755);
+
+ /* Set up inaccessible items */
+ (void) mkdir_label("/run/systemd/inaccessible", 0000);
+ (void) mknod("/run/systemd/inaccessible/reg", S_IFREG | 0000, 0);
+ (void) mkdir_label("/run/systemd/inaccessible/dir", 0000);
+ (void) mknod("/run/systemd/inaccessible/chr", S_IFCHR | 0000, makedev(0, 0));
+ (void) mknod("/run/systemd/inaccessible/blk", S_IFBLK | 0000, makedev(0, 0));
+ (void) mkfifo("/run/systemd/inaccessible/fifo", 0000);
+ (void) mknod("/run/systemd/inaccessible/sock", S_IFSOCK | 0000, 0);
return 0;
}
~ianmdlvl / elogind.git / blobdiff