diff options
Diffstat (limited to 'src/core/namespace.c')
-rw-r--r-- | src/core/namespace.c | 291 |
1 files changed, 164 insertions, 127 deletions
diff --git a/src/core/namespace.c b/src/core/namespace.c index e4930db15c..c2ca3e0334 100644 --- a/src/core/namespace.c +++ b/src/core/namespace.c @@ -20,6 +20,7 @@ #include "missing.h" #include "mkdir.h" #include "mount-util.h" +#include "mountpoint-util.h" #include "namespace.h" #include "path-util.h" #include "selinux-util.h" @@ -236,7 +237,8 @@ static int append_access_mounts(MountEntry **p, char **strv, MountMode mode, boo } if (!path_is_absolute(e)) - return -EINVAL; + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "Path is not absolute: %s", e); *((*p)++) = (MountEntry) { .path_const = e, @@ -263,7 +265,6 @@ static int append_empty_dir_mounts(MountEntry **p, char **strv) { .path_const = *i, .mode = EMPTY_DIR, .ignore = false, - .has_prefix = false, .read_only = true, .options_const = "mode=755", .flags = MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, @@ -302,35 +303,33 @@ static int append_tmpfs_mounts(MountEntry **p, const TemporaryFileSystem *tmpfs, for (i = 0; i < n; i++) { const TemporaryFileSystem *t = tmpfs + i; _cleanup_free_ char *o = NULL, *str = NULL; - unsigned long flags = MS_NODEV|MS_STRICTATIME; + unsigned long flags; bool ro = false; if (!path_is_absolute(t->path)) - return -EINVAL; + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "Path is not absolute: %s", + t->path); - if (!isempty(t->options)) { - str = strjoin("mode=0755,", t->options); - if (!str) - return -ENOMEM; + str = strjoin("mode=0755,", t->options); + if (!str) + return -ENOMEM; - r = mount_option_mangle(str, MS_NODEV|MS_STRICTATIME, &flags, &o); - if (r < 0) - return r; + r = mount_option_mangle(str, MS_NODEV|MS_STRICTATIME, &flags, &o); + if (r < 0) + return log_debug_errno(r, "Failed to parse mount option '%s': %m", str); - ro = flags & MS_RDONLY; - if (ro) - flags ^= MS_RDONLY; - } + ro = flags & MS_RDONLY; + if (ro) + flags ^= MS_RDONLY; *((*p)++) = (MountEntry) { .path_const = t->path, .mode = TMPFS, .read_only = ro, - .options_malloc = o, + .options_malloc = TAKE_PTR(o), .flags = flags, }; - - o = NULL; } return 0; @@ -398,32 +397,22 @@ static int append_protect_system(MountEntry **p, ProtectSystem protect_system, b } } -static int mount_path_compare(const void *a, const void *b) { - const MountEntry *p = a, *q = b; +static int mount_path_compare(const MountEntry *a, const MountEntry *b) { int d; /* If the paths are not equal, then order prefixes first */ - d = path_compare(mount_entry_path(p), mount_entry_path(q)); + d = path_compare(mount_entry_path(a), mount_entry_path(b)); if (d != 0) return d; /* If the paths are equal, check the mode */ - if (p->mode < q->mode) - return -1; - if (p->mode > q->mode) - return 1; - - return 0; + return CMP((int) a->mode, (int) b->mode); } static int prefix_where_needed(MountEntry *m, size_t n, const char *root_directory) { size_t i; - /* Prefixes all paths in the bind mount table with the root directory if it is specified and the entry needs - * that. */ - - if (!root_directory) - return 0; + /* Prefixes all paths in the bind mount table with the root directory if the entry needs that. */ for (i = 0; i < n; i++) { char *s; @@ -566,36 +555,44 @@ static void drop_outside_root(const char *root_directory, MountEntry *m, size_t *n = t - m; } -static int clone_device_node(const char *d, const char *temporary_mount, bool *make_devnode) { - const char *dn; +static int clone_device_node( + const char *d, + const char *temporary_mount, + bool *make_devnode) { + + _cleanup_free_ char *sl = NULL; + const char *dn, *bn, *t; struct stat st; int r; if (stat(d, &st) < 0) { - if (errno == ENOENT) + if (errno == ENOENT) { + log_debug_errno(errno, "Device node '%s' to clone does not exist, ignoring.", d); return -ENXIO; - return -errno; + } + + return log_debug_errno(errno, "Failed to stat() device node '%s' to clone, ignoring: %m", d); } if (!S_ISBLK(st.st_mode) && !S_ISCHR(st.st_mode)) - return -EINVAL; - - if (st.st_rdev == 0) - return -ENXIO; + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "Device node '%s' to clone is not a device node, ignoring.", + d); dn = strjoina(temporary_mount, d); + /* First, try to create device node properly */ if (*make_devnode) { mac_selinux_create_file_prepare(d, st.st_mode); r = mknod(dn, st.st_mode, st.st_rdev); mac_selinux_create_file_clear(); - - if (r == 0) - return 0; + if (r >= 0) + goto add_symlink; if (errno != EPERM) return log_debug_errno(errno, "mknod failed for %s: %m", d); + /* This didn't work, let's not try this again for the next iterations. */ *make_devnode = false; } @@ -604,9 +601,8 @@ static int clone_device_node(const char *d, const char *temporary_mount, bool *m mac_selinux_create_file_prepare(d, 0); r = mknod(dn, S_IFREG, 0); mac_selinux_create_file_clear(); - if (r < 0 && errno != EEXIST) - return log_debug_errno(errno, "mknod fallback failed for %s: %m", d); + return log_debug_errno(errno, "mknod() fallback failed for '%s': %m", d); /* Fallback to bind-mounting: * The assumption here is that all used device nodes carry standard @@ -614,7 +610,23 @@ static int clone_device_node(const char *d, const char *temporary_mount, bool *m * either be owned by root:root or root:tty (e.g. /dev/tty, /dev/ptmx) * and should not carry ACLs. */ if (mount(d, dn, NULL, MS_BIND, NULL) < 0) - return log_debug_errno(errno, "mount failed for %s: %m", d); + return log_debug_errno(errno, "Bind mounting failed for '%s': %m", d); + +add_symlink: + bn = path_startswith(d, "/dev/"); + if (!bn) + return 0; + + /* Create symlinks like /dev/char/1:9 → ../urandom */ + if (asprintf(&sl, "%s/dev/%s/%u:%u", temporary_mount, S_ISCHR(st.st_mode) ? "char" : "block", major(st.st_rdev), minor(st.st_rdev)) < 0) + return log_oom(); + + (void) mkdir_parents(sl, 0755); + + t = strjoina("../", bn); + + if (symlink(t, sl) < 0) + log_debug_errno(errno, "Failed to symlink '%s' to '%s', ignoring: %m", t, sl); return 0; } @@ -639,35 +651,34 @@ static int mount_private_dev(MountEntry *m) { u = umask(0000); if (!mkdtemp(temporary_mount)) - return -errno; + return log_debug_errno(errno, "Failed to create temporary directory '%s': %m", temporary_mount); dev = strjoina(temporary_mount, "/dev"); (void) mkdir(dev, 0755); if (mount("tmpfs", dev, "tmpfs", DEV_MOUNT_OPTIONS, "mode=755") < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to mount tmpfs on '%s': %m", dev); goto fail; } devpts = strjoina(temporary_mount, "/dev/pts"); (void) mkdir(devpts, 0755); if (mount("/dev/pts", devpts, NULL, MS_BIND, NULL) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to bind mount /dev/pts on '%s': %m", devpts); goto fail; } - /* /dev/ptmx can either be a device node or a symlink to /dev/pts/ptmx - * when /dev/ptmx a device node, /dev/pts/ptmx has 000 permissions making it inaccessible - * thus, in that case make a clone - * - * in nspawn and other containers it will be a symlink, in that case make it a symlink - */ + /* /dev/ptmx can either be a device node or a symlink to /dev/pts/ptmx. + * When /dev/ptmx a device node, /dev/pts/ptmx has 000 permissions making it inaccessible. + * Thus, in that case make a clone. + * In nspawn and other containers it will be a symlink, in that case make it a symlink. */ r = is_symlink("/dev/ptmx"); - if (r < 0) + if (r < 0) { + log_debug_errno(r, "Failed to detect whether /dev/ptmx is a symlink or not: %m"); goto fail; - if (r > 0) { + } else if (r > 0) { devptmx = strjoina(temporary_mount, "/dev/ptmx"); if (symlink("pts/ptmx", devptmx) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to create a symlink '%s' to pts/ptmx: %m", devptmx); goto fail; } } else { @@ -680,20 +691,23 @@ static int mount_private_dev(MountEntry *m) { (void) mkdir(devshm, 0755); r = mount("/dev/shm", devshm, NULL, MS_BIND, NULL); if (r < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to bind mount /dev/shm on '%s': %m", devshm); goto fail; } devmqueue = strjoina(temporary_mount, "/dev/mqueue"); (void) mkdir(devmqueue, 0755); - (void) mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL); + if (mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL) < 0) + log_debug_errno(errno, "Failed to bind mount /dev/mqueue on '%s', ignoring: %m", devmqueue); devhugepages = strjoina(temporary_mount, "/dev/hugepages"); (void) mkdir(devhugepages, 0755); - (void) mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL); + if (mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL) < 0) + log_debug_errno(errno, "Failed to bind mount /dev/hugepages on '%s', ignoring: %m", devhugepages); devlog = strjoina(temporary_mount, "/dev/log"); - (void) symlink("/run/systemd/journal/dev-log", devlog); + if (symlink("/run/systemd/journal/dev-log", devlog) < 0) + log_debug_errno(errno, "Failed to create a symlink '%s' to /run/systemd/journal/dev-log, ignoring: %m", devlog); NULSTR_FOREACH(d, devnodes) { r = clone_device_node(d, temporary_mount, &can_mknod); @@ -702,7 +716,9 @@ static int mount_private_dev(MountEntry *m) { goto fail; } - dev_setup(temporary_mount, UID_INVALID, GID_INVALID); + r = dev_setup(temporary_mount, UID_INVALID, GID_INVALID); + if (r < 0) + log_debug_errno(r, "Failed to setup basic device tree at '%s', ignoring: %m", temporary_mount); /* Create the /dev directory if missing. It is more likely to be * missing when the service is started with RootDirectory. This is @@ -711,9 +727,12 @@ static int mount_private_dev(MountEntry *m) { (void) mkdir_p_label(mount_entry_path(m), 0755); /* Unmount everything in old /dev */ - umount_recursive(mount_entry_path(m), 0); + r = umount_recursive(mount_entry_path(m), 0); + if (r < 0) + log_debug_errno(r, "Failed to unmount directories below '%s', ignoring: %m", mount_entry_path(m)); + if (mount(dev, mount_entry_path(m), NULL, MS_MOVE, NULL) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to move mount point '%s' to '%s': %m", dev, mount_entry_path(m)); goto fail; } @@ -836,10 +855,10 @@ static int follow_symlink( if (r > 0) /* Reached the end, nothing more to resolve */ return 1; - if (m->n_followed >= CHASE_SYMLINKS_MAX) { /* put a boundary on things */ - log_debug("Symlink loop on '%s'.", mount_entry_path(m)); - return -ELOOP; - } + if (m->n_followed >= CHASE_SYMLINKS_MAX) /* put a boundary on things */ + return log_debug_errno(SYNTHETIC_ERRNO(ELOOP), + "Symlink loop on '%s'.", + mount_entry_path(m)); log_debug("Followed mount entry path symlink %s → %s.", mount_entry_path(m), target); @@ -881,10 +900,9 @@ static int apply_mount( } what = mode_to_inaccessible_node(target.st_mode); - if (!what) { - log_debug("File type not supported for inaccessible mounts. Note that symlinks are not allowed"); - return -ELOOP; - } + if (!what) + return log_debug_errno(SYNTHETIC_ERRNO(ELOOP), + "File type not supported for inaccessible mounts. Note that symlinks are not allowed"); break; } @@ -999,7 +1017,17 @@ static int apply_mount( return 0; } +/* Change the per-mount readonly flag on an existing mount */ +static int remount_bind_readonly(const char *path, unsigned long orig_flags) { + int r; + + r = mount(NULL, path, NULL, MS_REMOUNT | MS_BIND | MS_RDONLY | orig_flags, NULL); + + return r < 0 ? -errno : 0; +} + static int make_read_only(const MountEntry *m, char **blacklist, FILE *proc_self_mountinfo) { + bool submounts = false; int r = 0; assert(m); @@ -1007,15 +1035,15 @@ static int make_read_only(const MountEntry *m, char **blacklist, FILE *proc_self if (mount_entry_read_only(m)) { if (IN_SET(m->mode, EMPTY_DIR, TMPFS)) { - /* Make superblock readonly */ - if (mount(NULL, mount_entry_path(m), NULL, MS_REMOUNT | MS_RDONLY | m->flags, mount_entry_options(m)) < 0) - r = -errno; - } else + r = remount_bind_readonly(mount_entry_path(m), m->flags); + } else { + submounts = true; r = bind_remount_recursive_with_mountinfo(mount_entry_path(m), true, blacklist, proc_self_mountinfo); + } } else if (m->mode == PRIVATE_DEV) { - /* Superblock can be readonly but the submounts can't */ - if (mount(NULL, mount_entry_path(m), NULL, MS_REMOUNT|DEV_MOUNT_OPTIONS|MS_RDONLY, NULL) < 0) - r = -errno; + /* Set /dev readonly, but not submounts like /dev/shm. Also, we only set the per-mount read-only flag. + * We can't set it on the superblock, if we are inside a user namespace and running Linux <= 4.17. */ + r = remount_bind_readonly(mount_entry_path(m), DEV_MOUNT_OPTIONS); } else return 0; @@ -1026,27 +1054,28 @@ static int make_read_only(const MountEntry *m, char **blacklist, FILE *proc_self if (r == -ENOENT && m->ignore) r = 0; - return r; + if (r < 0) + return log_debug_errno(r, "Failed to re-mount '%s'%s read-only: %m", mount_entry_path(m), + submounts ? " and its submounts" : ""); + + return 0; } -static bool namespace_info_mount_apivfs(const char *root_directory, const NamespaceInfo *ns_info) { +static bool namespace_info_mount_apivfs(const NamespaceInfo *ns_info) { assert(ns_info); /* * ProtectControlGroups= and ProtectKernelTunables= imply MountAPIVFS=, * since to protect the API VFS mounts, they need to be around in the - * first place... and RootDirectory= or RootImage= need to be set. + * first place... */ - /* root_directory should point to a mount point */ - return root_directory && - (ns_info->mount_apivfs || - ns_info->protect_control_groups || - ns_info->protect_kernel_tunables); + return ns_info->mount_apivfs || + ns_info->protect_control_groups || + ns_info->protect_kernel_tunables; } static size_t namespace_calculate_mounts( - const char* root_directory, const NamespaceInfo *ns_info, char** read_write_paths, char** read_only_paths, @@ -1088,14 +1117,15 @@ static size_t namespace_calculate_mounts( (ns_info->protect_control_groups ? 1 : 0) + (ns_info->protect_kernel_modules ? ELEMENTSOF(protect_kernel_modules_table) : 0) + protect_home_cnt + protect_system_cnt + - (namespace_info_mount_apivfs(root_directory, ns_info) ? ELEMENTSOF(apivfs_table) : 0); + (namespace_info_mount_apivfs(ns_info) ? ELEMENTSOF(apivfs_table) : 0); } static void normalize_mounts(const char *root_directory, MountEntry *mounts, size_t *n_mounts) { + assert(root_directory); assert(n_mounts); assert(mounts || *n_mounts == 0); - qsort_safe(mounts, *n_mounts, sizeof(MountEntry), mount_path_compare); + typesafe_qsort(mounts, *n_mounts, mount_path_compare); drop_duplicates(mounts, n_mounts); drop_outside_root(root_directory, mounts, n_mounts); @@ -1127,11 +1157,9 @@ int setup_namespace( _cleanup_(dissected_image_unrefp) DissectedImage *dissected_image = NULL; _cleanup_free_ void *root_hash = NULL; MountEntry *m, *mounts = NULL; - size_t root_hash_size = 0; - const char *root; - size_t n_mounts; - bool make_slave; + size_t n_mounts, root_hash_size = 0; bool require_prefix = false; + const char *root; int r = 0; assert(ns_info); @@ -1151,19 +1179,19 @@ int setup_namespace( dissect_image_flags & DISSECT_IMAGE_READ_ONLY ? O_RDONLY : O_RDWR, &loop_device); if (r < 0) - return r; + return log_debug_errno(r, "Failed to create loop device for root image: %m"); r = root_hash_load(root_image, &root_hash, &root_hash_size); if (r < 0) - return r; + return log_debug_errno(r, "Failed to load root hash: %m"); r = dissect_image(loop_device->fd, root_hash, root_hash_size, dissect_image_flags, &dissected_image); if (r < 0) - return r; + return log_debug_errno(r, "Failed to dissect image: %m"); r = dissected_image_decrypt(dissected_image, NULL, root_hash, root_hash_size, dissect_image_flags, &decrypted_image); if (r < 0) - return r; + return log_debug_errno(r, "Failed to decrypt dissected image: %m"); } if (root_directory) @@ -1181,7 +1209,6 @@ int setup_namespace( } n_mounts = namespace_calculate_mounts( - root, ns_info, read_write_paths, read_only_paths, @@ -1192,9 +1219,6 @@ int setup_namespace( tmp_dir, var_tmp_dir, protect_home, protect_system); - /* Set mount slave mode */ - make_slave = root || n_mounts > 0 || ns_info->private_mounts; - if (n_mounts > 0) { m = mounts = (MountEntry *) alloca0(n_mounts * sizeof(MountEntry)); r = append_access_mounts(&m, read_write_paths, READWRITE, require_prefix); @@ -1271,7 +1295,7 @@ int setup_namespace( if (r < 0) goto finish; - if (namespace_info_mount_apivfs(root, ns_info)) { + if (namespace_info_mount_apivfs(ns_info)) { r = append_static_mounts(&m, apivfs_table, ELEMENTSOF(apivfs_table), ns_info->ignore_protect_paths); if (r < 0) goto finish; @@ -1284,33 +1308,44 @@ int setup_namespace( if (r < 0) goto finish; - normalize_mounts(root_directory, mounts, &n_mounts); + normalize_mounts(root, mounts, &n_mounts); } + /* All above is just preparation, figuring out what to do. Let's now actually start doing something. */ + if (unshare(CLONE_NEWNS) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to unshare the mount namespace: %m"); + if (IN_SET(r, -EACCES, -EPERM, -EOPNOTSUPP, -ENOSYS)) + /* If the kernel doesn't support namespaces, or when there's a MAC or seccomp filter in place + * that doesn't allow us to create namespaces (or a missing cap), then propagate a recognizable + * error back, which the caller can use to detect this case (and only this) and optionally + * continue without namespacing applied. */ + r = -ENOANO; + goto finish; } - if (make_slave) { - /* Remount / as SLAVE so that nothing now mounted in the namespace - shows up in the parent */ - if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) { - r = -errno; - goto finish; - } + /* Remount / as SLAVE so that nothing now mounted in the namespace + * shows up in the parent */ + if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) { + r = log_debug_errno(errno, "Failed to remount '/' as SLAVE: %m"); + goto finish; } if (root_image) { /* A root image is specified, mount it to the right place */ r = dissected_image_mount(dissected_image, root, UID_INVALID, dissect_image_flags); - if (r < 0) + if (r < 0) { + log_debug_errno(r, "Failed to mount root image: %m"); goto finish; + } if (decrypted_image) { r = decrypted_image_relinquish(decrypted_image); - if (r < 0) + if (r < 0) { + log_debug_errno(r, "Failed to relinquish decrypted image: %m"); goto finish; + } } loop_device_relinquish(loop_device); @@ -1319,20 +1354,22 @@ int setup_namespace( /* A root directory is specified. Turn its directory into bind mount, if it isn't one yet. */ r = path_is_mount_point(root, NULL, AT_SYMLINK_FOLLOW); - if (r < 0) + if (r < 0) { + log_debug_errno(r, "Failed to detect that %s is a mount point or not: %m", root); goto finish; + } if (r == 0) { if (mount(root, root, NULL, MS_BIND|MS_REC, NULL) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to bind mount '%s': %m", root); goto finish; } } - } else if (root) { + } else { /* Let's mount the main root directory to the root directory to use */ if (mount("/", root, NULL, MS_BIND|MS_REC, NULL) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to bind mount '/' on '%s': %m", root); goto finish; } } @@ -1350,7 +1387,7 @@ int setup_namespace( * For example, this is the case with the option: 'InaccessiblePaths=/proc' */ proc_self_mountinfo = fopen("/proc/self/mountinfo", "re"); if (!proc_self_mountinfo) { - r = -errno; + r = log_debug_errno(errno, "Failed to open /proc/self/mountinfo: %m"); goto finish; } @@ -1385,7 +1422,7 @@ int setup_namespace( if (!again) break; - normalize_mounts(root_directory, mounts, &n_mounts); + normalize_mounts(root, mounts, &n_mounts); } /* Create a blacklist we can pass to bind_mount_recursive() */ @@ -1402,18 +1439,18 @@ int setup_namespace( } } - if (root) { - /* MS_MOVE does not work on MS_SHARED so the remount MS_SHARED will be done later */ - r = mount_move_root(root); - if (r < 0) - goto finish; + /* MS_MOVE does not work on MS_SHARED so the remount MS_SHARED will be done later */ + r = mount_move_root(root); + if (r < 0) { + log_debug_errno(r, "Failed to mount root with MS_MOVE: %m"); + goto finish; } /* Remount / as the desired mode. Note that this will not * reestablish propagation from our side to the host, since * what's disconnected is disconnected. */ if (mount(NULL, "/", NULL, mount_flags | MS_REC, NULL) < 0) { - r = -errno; + r = log_debug_errno(errno, "Failed to remount '/' with desired mount flags: %m"); goto finish; } |