#include "tmpfile-util.h"
#include "umask-util.h"
#include "user-util.h"
+#include "vpick.h"
#define DEV_MOUNT_OPTIONS (MS_NOSUID|MS_STRICTATIME|MS_NOEXEC)
typedef enum MountMode {
/* This is ordered by priority! */
- INACCESSIBLE,
- OVERLAY_MOUNT,
- MOUNT_IMAGES,
- BIND_MOUNT,
- BIND_MOUNT_RECURSIVE,
- PRIVATE_TMP,
- PRIVATE_TMP_READONLY,
- PRIVATE_DEV,
- BIND_DEV,
- EMPTY_DIR,
- PRIVATE_SYSFS,
- BIND_SYSFS,
- PROCFS,
- READONLY,
- READWRITE,
- NOEXEC,
- EXEC,
- TMPFS,
- RUN,
- EXTENSION_DIRECTORIES, /* Bind-mounted outside the root directory, and used by subsequent mounts */
- EXTENSION_IMAGES, /* Mounted outside the root directory, and used by subsequent mounts */
- MQUEUEFS,
- READWRITE_IMPLICIT, /* Should have the lowest priority. */
+ MOUNT_INACCESSIBLE,
+ MOUNT_OVERLAY,
+ MOUNT_IMAGE,
+ MOUNT_BIND,
+ MOUNT_BIND_RECURSIVE,
+ MOUNT_PRIVATE_TMP,
+ MOUNT_PRIVATE_TMP_READ_ONLY,
+ MOUNT_PRIVATE_DEV,
+ MOUNT_BIND_DEV,
+ MOUNT_EMPTY_DIR,
+ MOUNT_PRIVATE_SYSFS,
+ MOUNT_BIND_SYSFS,
+ MOUNT_PROCFS,
+ MOUNT_READ_ONLY,
+ MOUNT_READ_WRITE,
+ MOUNT_NOEXEC,
+ MOUNT_EXEC,
+ MOUNT_TMPFS,
+ MOUNT_RUN,
+ MOUNT_EXTENSION_DIRECTORY, /* Bind-mounted outside the root directory, and used by subsequent mounts */
+ MOUNT_EXTENSION_IMAGE, /* Mounted outside the root directory, and used by subsequent mounts */
+ MOUNT_MQUEUEFS,
+ MOUNT_READ_WRITE_IMPLICIT, /* Should have the lowest priority. */
_MOUNT_MODE_MAX,
+ _MOUNT_MODE_INVALID = -EINVAL,
} MountMode;
+typedef enum MountEntryState {
+ MOUNT_PENDING,
+ MOUNT_APPLIED,
+ MOUNT_SKIPPED,
+ _MOUNT_ENTRY_STATE_MAX,
+ _MOUNT_ENTRY_STATE_INVALID = -EINVAL,
+} MountEntryState;
+
typedef struct MountEntry {
const char *path_const; /* Memory allocated on stack or static */
- MountMode mode:5;
+ MountMode mode;
bool ignore:1; /* Ignore if path does not exist? */
bool has_prefix:1; /* Already is prefixed by the root dir? */
bool read_only:1; /* Shall this mount point be read-only? */
bool nosuid:1; /* Shall set MS_NOSUID on the mount itself */
bool noexec:1; /* Shall set MS_NOEXEC on the mount itself */
bool exec:1; /* Shall clear MS_NOEXEC on the mount itself */
- bool applied:1; /* Already applied */
+ MountEntryState state; /* Whether it was already processed or skipped */
char *path_malloc; /* Use this instead of 'path_const' if we had to allocate memory */
const char *unprefixed_path_const; /* If the path was amended with a prefix, these will save the original */
char *unprefixed_path_malloc;
unsigned long flags; /* Mount flags used by EMPTY_DIR and TMPFS. Do not include MS_RDONLY here, but please use read_only. */
unsigned n_followed;
LIST_HEAD(MountOptions, image_options_const);
+ char **overlay_layers;
} MountEntry;
typedef struct MountList {
/* If MountAPIVFS= is used, let's mount /sys, /proc, /dev and /run into the it, but only as a fallback if the user hasn't mounted
* something there already. These mounts are hence overridden by any other explicitly configured mounts. */
static const MountEntry apivfs_table[] = {
- { "/proc", PROCFS, false },
- { "/dev", BIND_DEV, false },
- { "/sys", BIND_SYSFS, false },
- { "/run", RUN, false, .options_const = "mode=0755" TMPFS_LIMITS_RUN, .flags = MS_NOSUID|MS_NODEV|MS_STRICTATIME },
+ { "/proc", MOUNT_PROCFS, false },
+ { "/dev", MOUNT_BIND_DEV, false },
+ { "/sys", MOUNT_BIND_SYSFS, false },
+ { "/run", MOUNT_RUN, false, .options_const = "mode=0755" TMPFS_LIMITS_RUN, .flags = MS_NOSUID|MS_NODEV|MS_STRICTATIME },
};
/* ProtectKernelTunables= option and the related filesystem APIs */
static const MountEntry protect_kernel_tunables_proc_table[] = {
- { "/proc/acpi", READONLY, true },
- { "/proc/apm", READONLY, true }, /* Obsolete API, there's no point in permitting access to this, ever */
- { "/proc/asound", READONLY, true },
- { "/proc/bus", READONLY, true },
- { "/proc/fs", READONLY, true },
- { "/proc/irq", READONLY, true },
- { "/proc/kallsyms", INACCESSIBLE, true },
- { "/proc/kcore", INACCESSIBLE, true },
- { "/proc/latency_stats", READONLY, true },
- { "/proc/mtrr", READONLY, true },
- { "/proc/scsi", READONLY, true },
- { "/proc/sys", READONLY, true },
- { "/proc/sysrq-trigger", READONLY, true },
- { "/proc/timer_stats", READONLY, true },
+ { "/proc/acpi", MOUNT_READ_ONLY, true },
+ { "/proc/apm", MOUNT_READ_ONLY, true }, /* Obsolete API, there's no point in permitting access to this, ever */
+ { "/proc/asound", MOUNT_READ_ONLY, true },
+ { "/proc/bus", MOUNT_READ_ONLY, true },
+ { "/proc/fs", MOUNT_READ_ONLY, true },
+ { "/proc/irq", MOUNT_READ_ONLY, true },
+ { "/proc/kallsyms", MOUNT_INACCESSIBLE, true },
+ { "/proc/kcore", MOUNT_INACCESSIBLE, true },
+ { "/proc/latency_stats", MOUNT_READ_ONLY, true },
+ { "/proc/mtrr", MOUNT_READ_ONLY, true },
+ { "/proc/scsi", MOUNT_READ_ONLY, true },
+ { "/proc/sys", MOUNT_READ_ONLY, true },
+ { "/proc/sysrq-trigger", MOUNT_READ_ONLY, true },
+ { "/proc/timer_stats", MOUNT_READ_ONLY, true },
};
static const MountEntry protect_kernel_tunables_sys_table[] = {
- { "/sys", READONLY, false },
- { "/sys/fs/bpf", READONLY, true },
- { "/sys/fs/cgroup", READWRITE_IMPLICIT, false }, /* READONLY is set by ProtectControlGroups= option */
- { "/sys/fs/selinux", READWRITE_IMPLICIT, true },
- { "/sys/kernel/debug", READONLY, true },
- { "/sys/kernel/tracing", READONLY, true },
+ { "/sys", MOUNT_READ_ONLY, false },
+ { "/sys/fs/bpf", MOUNT_READ_ONLY, true },
+ { "/sys/fs/cgroup", MOUNT_READ_WRITE_IMPLICIT, false }, /* READ_ONLY is set by ProtectControlGroups= option */
+ { "/sys/fs/selinux", MOUNT_READ_WRITE_IMPLICIT, true },
+ { "/sys/kernel/debug", MOUNT_READ_ONLY, true },
+ { "/sys/kernel/tracing", MOUNT_READ_ONLY, true },
};
/* ProtectKernelModules= option */
static const MountEntry protect_kernel_modules_table[] = {
- { "/usr/lib/modules", INACCESSIBLE, true },
+ { "/usr/lib/modules", MOUNT_INACCESSIBLE, true },
};
/* ProtectKernelLogs= option */
static const MountEntry protect_kernel_logs_proc_table[] = {
- { "/proc/kmsg", INACCESSIBLE, true },
+ { "/proc/kmsg", MOUNT_INACCESSIBLE, true },
};
static const MountEntry protect_kernel_logs_dev_table[] = {
- { "/dev/kmsg", INACCESSIBLE, true },
+ { "/dev/kmsg", MOUNT_INACCESSIBLE, true },
};
/*
* system should be protected by ProtectSystem=
*/
static const MountEntry protect_home_read_only_table[] = {
- { "/home", READONLY, true },
- { "/run/user", READONLY, true },
- { "/root", READONLY, true },
+ { "/home", MOUNT_READ_ONLY, true },
+ { "/run/user", MOUNT_READ_ONLY, true },
+ { "/root", MOUNT_READ_ONLY, true },
};
/* ProtectHome=tmpfs table */
static const MountEntry protect_home_tmpfs_table[] = {
- { "/home", TMPFS, true, .read_only = true, .options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
- { "/run/user", TMPFS, true, .read_only = true, .options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
- { "/root", TMPFS, true, .read_only = true, .options_const = "mode=0700" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
+ { "/home", MOUNT_TMPFS, true, .read_only = true, .options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
+ { "/run/user", MOUNT_TMPFS, true, .read_only = true, .options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
+ { "/root", MOUNT_TMPFS, true, .read_only = true, .options_const = "mode=0700" TMPFS_LIMITS_EMPTY_OR_ALMOST, .flags = MS_NODEV|MS_STRICTATIME },
};
/* ProtectHome=yes table */
static const MountEntry protect_home_yes_table[] = {
- { "/home", INACCESSIBLE, true },
- { "/run/user", INACCESSIBLE, true },
- { "/root", INACCESSIBLE, true },
+ { "/home", MOUNT_INACCESSIBLE, true },
+ { "/run/user", MOUNT_INACCESSIBLE, true },
+ { "/root", MOUNT_INACCESSIBLE, true },
};
/* ProtectSystem=yes table */
static const MountEntry protect_system_yes_table[] = {
- { "/usr", READONLY, false },
- { "/boot", READONLY, true },
- { "/efi", READONLY, true },
+ { "/usr", MOUNT_READ_ONLY, false },
+ { "/boot", MOUNT_READ_ONLY, true },
+ { "/efi", MOUNT_READ_ONLY, true },
};
/* ProtectSystem=full includes ProtectSystem=yes */
static const MountEntry protect_system_full_table[] = {
- { "/usr", READONLY, false },
- { "/boot", READONLY, true },
- { "/efi", READONLY, true },
- { "/etc", READONLY, false },
+ { "/usr", MOUNT_READ_ONLY, false },
+ { "/boot", MOUNT_READ_ONLY, true },
+ { "/efi", MOUNT_READ_ONLY, true },
+ { "/etc", MOUNT_READ_ONLY, false },
};
/* ProtectSystem=strict table. In this strict mode, we mount everything read-only, except for /proc, /dev,
* left writable, as ProtectHome= shall manage those, orthogonally).
*/
static const MountEntry protect_system_strict_table[] = {
- { "/", READONLY, false },
- { "/proc", READWRITE_IMPLICIT, false }, /* ProtectKernelTunables= */
- { "/sys", READWRITE_IMPLICIT, false }, /* ProtectKernelTunables= */
- { "/dev", READWRITE_IMPLICIT, false }, /* PrivateDevices= */
- { "/home", READWRITE_IMPLICIT, true }, /* ProtectHome= */
- { "/run/user", READWRITE_IMPLICIT, true }, /* ProtectHome= */
- { "/root", READWRITE_IMPLICIT, true }, /* ProtectHome= */
+ { "/", MOUNT_READ_ONLY, false },
+ { "/proc", MOUNT_READ_WRITE_IMPLICIT, false }, /* ProtectKernelTunables= */
+ { "/sys", MOUNT_READ_WRITE_IMPLICIT, false }, /* ProtectKernelTunables= */
+ { "/dev", MOUNT_READ_WRITE_IMPLICIT, false }, /* PrivateDevices= */
+ { "/home", MOUNT_READ_WRITE_IMPLICIT, true }, /* ProtectHome= */
+ { "/run/user", MOUNT_READ_WRITE_IMPLICIT, true }, /* ProtectHome= */
+ { "/root", MOUNT_READ_WRITE_IMPLICIT, true }, /* ProtectHome= */
};
/* ProtectHostname=yes able */
static const MountEntry protect_hostname_table[] = {
- { "/proc/sys/kernel/hostname", READONLY, false },
- { "/proc/sys/kernel/domainname", READONLY, false },
+ { "/proc/sys/kernel/hostname", MOUNT_READ_ONLY, false },
+ { "/proc/sys/kernel/domainname", MOUNT_READ_ONLY, false },
};
static const char * const mount_mode_table[_MOUNT_MODE_MAX] = {
- [INACCESSIBLE] = "inaccessible",
- [OVERLAY_MOUNT] = "overlay",
- [MOUNT_IMAGES] = "mount-images",
- [BIND_MOUNT] = "bind",
- [BIND_MOUNT_RECURSIVE] = "rbind",
- [PRIVATE_TMP] = "private-tmp",
- [PRIVATE_TMP_READONLY] = "private-tmp-read-only",
- [PRIVATE_DEV] = "private-dev",
- [BIND_DEV] = "bind-dev",
- [EMPTY_DIR] = "empty",
- [PRIVATE_SYSFS] = "private-sysfs",
- [BIND_SYSFS] = "bind-sysfs",
- [PROCFS] = "procfs",
- [READONLY] = "read-only",
- [READWRITE] = "read-write",
- [NOEXEC] = "noexec",
- [EXEC] = "exec",
- [TMPFS] = "tmpfs",
- [RUN] = "run",
- [EXTENSION_DIRECTORIES] = "extension-directories",
- [EXTENSION_IMAGES] = "extension-images",
- [MQUEUEFS] = "mqueuefs",
- [READWRITE_IMPLICIT] = "read-write-implicit",
+ [MOUNT_INACCESSIBLE] = "inaccessible",
+ [MOUNT_OVERLAY] = "overlay",
+ [MOUNT_IMAGE] = "image",
+ [MOUNT_BIND] = "bind",
+ [MOUNT_BIND_RECURSIVE] = "bind-recursive",
+ [MOUNT_PRIVATE_TMP] = "private-tmp",
+ [MOUNT_PRIVATE_TMP_READ_ONLY] = "private-tmp-read-only",
+ [MOUNT_PRIVATE_DEV] = "private-dev",
+ [MOUNT_BIND_DEV] = "bind-dev",
+ [MOUNT_EMPTY_DIR] = "empty-dir",
+ [MOUNT_PRIVATE_SYSFS] = "private-sysfs",
+ [MOUNT_BIND_SYSFS] = "bind-sysfs",
+ [MOUNT_PROCFS] = "procfs",
+ [MOUNT_READ_ONLY] = "read-only",
+ [MOUNT_READ_WRITE] = "read-write",
+ [MOUNT_NOEXEC] = "noexec",
+ [MOUNT_EXEC] = "exec",
+ [MOUNT_TMPFS] = "tmpfs",
+ [MOUNT_RUN] = "run",
+ [MOUNT_EXTENSION_DIRECTORY] = "extension-directory",
+ [MOUNT_EXTENSION_IMAGE] = "extension-image",
+ [MOUNT_MQUEUEFS] = "mqueuefs",
+ [MOUNT_READ_WRITE_IMPLICIT] = "read-write-implicit",
};
/* Helper struct for naming simplicity and reusability */
static bool mount_entry_read_only(const MountEntry *p) {
assert(p);
- return p->read_only || IN_SET(p->mode, READONLY, INACCESSIBLE, PRIVATE_TMP_READONLY);
+ return p->read_only || IN_SET(p->mode, MOUNT_READ_ONLY, MOUNT_INACCESSIBLE, MOUNT_PRIVATE_TMP_READ_ONLY);
}
static bool mount_entry_noexec(const MountEntry *p) {
assert(p);
- return p->noexec || IN_SET(p->mode, NOEXEC, INACCESSIBLE, PRIVATE_SYSFS, BIND_SYSFS, PROCFS);
+ return p->noexec || IN_SET(p->mode, MOUNT_NOEXEC, MOUNT_INACCESSIBLE, MOUNT_PRIVATE_SYSFS, MOUNT_BIND_SYSFS, MOUNT_PROCFS);
}
static bool mount_entry_exec(const MountEntry *p) {
assert(p);
- return p->exec || p->mode == EXEC;
+ return p->exec || p->mode == MOUNT_EXEC;
}
static const char *mount_entry_source(const MountEntry *p) {
p->unprefixed_path_malloc = mfree(p->unprefixed_path_malloc);
p->source_malloc = mfree(p->source_malloc);
p->options_malloc = mfree(p->options_malloc);
+ p->overlay_layers = strv_free(p->overlay_layers);
}
static void mount_list_done(MountList *ml) {
static int append_access_mounts(MountList *ml, char **strv, MountMode mode, bool forcibly_require_prefix) {
assert(ml);
- /* Adds a list of user-supplied READWRITE/READWRITE_IMPLICIT/READONLY/INACCESSIBLE entries */
+ /* Adds a list of user-supplied READ_WRITE/READ_WRITE_IMPLICIT/READ_ONLY/INACCESSIBLE entries */
STRV_FOREACH(i, strv) {
bool ignore = false, needs_prefix = false;
*me = (MountEntry) {
.path_const = *i,
- .mode = EMPTY_DIR,
+ .mode = MOUNT_EMPTY_DIR,
.ignore = false,
.read_only = true,
.options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST,
*me = (MountEntry) {
.path_const = b->destination,
- .mode = b->recursive ? BIND_MOUNT_RECURSIVE : BIND_MOUNT,
+ .mode = b->recursive ? MOUNT_BIND_RECURSIVE : MOUNT_BIND,
.read_only = b->read_only,
.nosuid = b->nosuid,
.source_const = b->source,
*me = (MountEntry) {
.path_const = m->destination,
- .mode = MOUNT_IMAGES,
+ .mode = MOUNT_IMAGE,
.source_const = m->source,
.image_options_const = m->mount_options,
.ignore = m->ignore_enoent,
size_t n,
char **extension_directories) {
- _cleanup_strv_free_ char **overlays = NULL;
+ char ***overlays = NULL;
+ size_t n_overlays = 0;
int r;
assert(ml);
assert(extension_dir);
- /* Prepare a list of overlays, that will have as each element a string suitable for being
- * passed as a lowerdir= parameter, so start with the hierarchy on the root.
+ n_overlays = strv_length(hierarchies);
+ if (n_overlays == 0)
+ return 0;
+
+ /* Prepare a list of overlays, that will have as each element a strv containing all the layers that
+ * will later be concatenated as a lowerdir= parameter for the mount operation.
* The overlays vector will have the same number of elements and will correspond to the
* hierarchies vector, so they can be iterated upon together. */
- STRV_FOREACH(hierarchy, hierarchies) {
- _cleanup_free_ char *prefixed_hierarchy = NULL;
-
- prefixed_hierarchy = path_join(root, *hierarchy);
- if (!prefixed_hierarchy)
- return -ENOMEM;
+ overlays = new0(char**, n_overlays);
+ if (!overlays)
+ return -ENOMEM;
- r = strv_consume(&overlays, TAKE_PTR(prefixed_hierarchy));
- if (r < 0)
- return r;
- }
+ CLEANUP_ARRAY(overlays, n_overlays, strv_free_many);
/* First, prepare a mount for each image, but these won't be visible to the unit, instead
* they will be mounted in our propagate directory, and used as a source for the overlay. */
for (size_t i = 0; i < n; i++) {
+ _cleanup_(pick_result_done) PickResult result = PICK_RESULT_NULL;
_cleanup_free_ char *mount_point = NULL;
const MountImage *m = mount_images + i;
+ r = path_pick(/* toplevel_path= */ NULL,
+ /* toplevel_fd= */ AT_FDCWD,
+ m->source,
+ &pick_filter_image_raw,
+ PICK_ARCHITECTURE|PICK_TRIES,
+ &result);
+ if (r < 0)
+ return r;
+ if (!result.path)
+ return log_debug_errno(
+ SYNTHETIC_ERRNO(ENOENT),
+ "No matching entry in .v/ directory %s found.",
+ m->source);
+
if (asprintf(&mount_point, "%s/%zu", extension_dir, i) < 0)
return -ENOMEM;
for (size_t j = 0; hierarchies && hierarchies[j]; ++j) {
- _cleanup_free_ char *prefixed_hierarchy = NULL, *escaped = NULL, *lowerdir = NULL;
-
- prefixed_hierarchy = path_join(mount_point, hierarchies[j]);
+ char *prefixed_hierarchy = path_join(mount_point, hierarchies[j]);
if (!prefixed_hierarchy)
return -ENOMEM;
- escaped = shell_escape(prefixed_hierarchy, ",:");
- if (!escaped)
- return -ENOMEM;
-
- /* Note that lowerdir= parameters are in 'reverse' order, so the
- * top-most directory in the overlay comes first in the list. */
- lowerdir = strjoin(escaped, ":", overlays[j]);
- if (!lowerdir)
- return -ENOMEM;
-
- free_and_replace(overlays[j], lowerdir);
+ r = strv_consume(&overlays[j], TAKE_PTR(prefixed_hierarchy));
+ if (r < 0)
+ return r;
}
MountEntry *me = mount_list_extend(ml);
if (!me)
- return log_oom_debug();
+ return -ENOMEM;
*me = (MountEntry) {
.path_malloc = TAKE_PTR(mount_point),
.image_options_const = m->mount_options,
.ignore = m->ignore_enoent,
- .source_const = m->source,
- .mode = EXTENSION_IMAGES,
+ .source_malloc = TAKE_PTR(result.path),
+ .mode = MOUNT_EXTENSION_IMAGE,
.has_prefix = true,
};
}
* Bind mount them in the same location as the ExtensionImages, so that we
* can check that they are valid trees (extension-release.d). */
STRV_FOREACH(extension_directory, extension_directories) {
- _cleanup_free_ char *mount_point = NULL, *source = NULL;
+ _cleanup_(pick_result_done) PickResult result = PICK_RESULT_NULL;
+ _cleanup_free_ char *mount_point = NULL;
const char *e = *extension_directory;
bool ignore_enoent = false;
if (startswith(e, "+"))
e++;
- source = strdup(e);
- if (!source)
- return -ENOMEM;
+ r = path_pick(/* toplevel_path= */ NULL,
+ /* toplevel_fd= */ AT_FDCWD,
+ e,
+ &pick_filter_image_dir,
+ PICK_ARCHITECTURE|PICK_TRIES,
+ &result);
+ if (r < 0)
+ return r;
+ if (!result.path)
+ return log_debug_errno(
+ SYNTHETIC_ERRNO(ENOENT),
+ "No matching entry in .v/ directory %s found.",
+ e);
for (size_t j = 0; hierarchies && hierarchies[j]; ++j) {
- _cleanup_free_ char *prefixed_hierarchy = NULL, *escaped = NULL, *lowerdir = NULL;
-
- prefixed_hierarchy = path_join(mount_point, hierarchies[j]);
+ char *prefixed_hierarchy = path_join(mount_point, hierarchies[j]);
if (!prefixed_hierarchy)
return -ENOMEM;
- escaped = shell_escape(prefixed_hierarchy, ",:");
- if (!escaped)
- return -ENOMEM;
-
- /* Note that lowerdir= parameters are in 'reverse' order, so the
- * top-most directory in the overlay comes first in the list. */
- lowerdir = strjoin(escaped, ":", overlays[j]);
- if (!lowerdir)
- return -ENOMEM;
-
- free_and_replace(overlays[j], lowerdir);
+ r = strv_consume(&overlays[j], TAKE_PTR(prefixed_hierarchy));
+ if (r < 0)
+ return r;
}
MountEntry *me = mount_list_extend(ml);
if (!me)
- return log_oom_debug();
+ return -ENOMEM;
*me = (MountEntry) {
.path_malloc = TAKE_PTR(mount_point),
- .source_malloc = TAKE_PTR(source),
- .mode = EXTENSION_DIRECTORIES,
+ .source_malloc = TAKE_PTR(result.path),
+ .mode = MOUNT_EXTENSION_DIRECTORY,
.ignore = ignore_enoent,
.has_prefix = true,
.read_only = true,
MountEntry *me = mount_list_extend(ml);
if (!me)
- return log_oom_debug();
+ return -ENOMEM;
*me = (MountEntry) {
.path_malloc = TAKE_PTR(prefixed_hierarchy),
- .options_malloc = TAKE_PTR(overlays[i]),
- .mode = OVERLAY_MOUNT,
+ .overlay_layers = TAKE_PTR(overlays[i]),
+ .mode = MOUNT_OVERLAY,
.has_prefix = true,
.ignore = true, /* If the source image doesn't set the ignore bit it will fail earlier. */
};
return log_debug_errno(r, "Failed to parse mount option '%s': %m", str);
ro = flags & MS_RDONLY;
- if (ro)
- flags ^= MS_RDONLY;
+ flags &= ~MS_RDONLY;
MountEntry *me = mount_list_extend(ml);
if (!me)
*me = (MountEntry) {
.path_const = t->path,
- .mode = TMPFS,
+ .mode = MOUNT_TMPFS,
.read_only = ro,
.options_malloc = TAKE_PTR(o),
.flags = flags,
/* ExtensionImages/Directories will be used by other mounts as a base, so sort them first
* regardless of the prefix - they are set up in the propagate directory anyway */
- d = -CMP(a->mode == EXTENSION_IMAGES, b->mode == EXTENSION_IMAGES);
+ d = -CMP(a->mode == MOUNT_EXTENSION_IMAGE, b->mode == MOUNT_EXTENSION_IMAGE);
if (d != 0)
return d;
- d = -CMP(a->mode == EXTENSION_DIRECTORIES, b->mode == EXTENSION_DIRECTORIES);
+ d = -CMP(a->mode == MOUNT_EXTENSION_DIRECTORY, b->mode == MOUNT_EXTENSION_DIRECTORY);
if (d != 0)
return d;
* above. Note that we only drop duplicates that haven't been mounted yet. */
if (previous &&
path_equal(mount_entry_path(f), mount_entry_path(previous)) &&
- !f->applied && !previous->applied) {
+ f->state == MOUNT_PENDING && previous->state == MOUNT_PENDING) {
log_debug("%s (%s) is duplicate.", mount_entry_path(f), mount_mode_to_string(f->mode));
/* Propagate the flags to the remaining entry */
previous->read_only = previous->read_only || mount_entry_read_only(f);
continue;
}
- clear = f->mode == INACCESSIBLE ? mount_entry_path(f) : NULL;
+ clear = f->mode == MOUNT_INACCESSIBLE ? mount_entry_path(f) : NULL;
*t = *f;
t++;
for (f = ml->mounts, t = ml->mounts; f < ml->mounts + ml->n_mounts; f++) {
- /* Only suppress such subtrees for READONLY, READWRITE and READWRITE_IMPLICIT entries */
- if (IN_SET(f->mode, READONLY, READWRITE, READWRITE_IMPLICIT)) {
+ /* Only suppress such subtrees for READ_ONLY, READ_WRITE and READ_WRITE_IMPLICIT entries */
+ if (IN_SET(f->mode, MOUNT_READ_ONLY, MOUNT_READ_WRITE, MOUNT_READ_WRITE_IMPLICIT)) {
MountEntry *found = NULL;
/* Now let's find the first parent of the entry we are looking at. */
for (f = ml->mounts, t = ml->mounts; f < ml->mounts + ml->n_mounts; f++) {
/* ExtensionImages/Directories bases are opened in /run/systemd/unit-extensions on the host */
- if (!IN_SET(f->mode, EXTENSION_IMAGES, EXTENSION_DIRECTORIES) && !path_startswith(mount_entry_path(f), root_directory)) {
+ if (!IN_SET(f->mode, MOUNT_EXTENSION_IMAGE, MOUNT_EXTENSION_DIRECTORY) && !path_startswith(mount_entry_path(f), root_directory)) {
log_debug("%s is outside of root directory.", mount_entry_path(f));
mount_entry_done(f);
continue;
return runtime_dir;
}
+static int create_temporary_mount_point(RuntimeScope scope, char **ret) {
+ _cleanup_free_ char *runtime_dir = NULL, *temporary_mount = NULL;
+
+ assert(ret);
+
+ runtime_dir = settle_runtime_dir(scope);
+ if (!runtime_dir)
+ return log_oom_debug();
+
+ temporary_mount = path_join(runtime_dir, "systemd/namespace-XXXXXX");
+ if (!temporary_mount)
+ return log_oom_debug();
+
+ if (!mkdtemp(temporary_mount))
+ return log_debug_errno(errno, "Failed to create temporary directory '%s': %m", temporary_mount);
+
+ *ret = TAKE_PTR(temporary_mount);
+ return 0;
+}
+
static int mount_private_dev(MountEntry *m, RuntimeScope scope) {
static const char devnodes[] =
"/dev/null\0"
"/dev/urandom\0"
"/dev/tty\0";
- _cleanup_free_ char *runtime_dir = NULL, *temporary_mount = NULL;
+ _cleanup_free_ char *temporary_mount = NULL;
const char *dev = NULL, *devpts = NULL, *devshm = NULL, *devhugepages = NULL, *devmqueue = NULL, *devlog = NULL, *devptmx = NULL;
bool can_mknod = true;
int r;
assert(m);
- runtime_dir = settle_runtime_dir(scope);
- if (!runtime_dir)
- return log_oom_debug();
-
- temporary_mount = path_join(runtime_dir, "systemd/namespace-dev-XXXXXX");
- if (!temporary_mount)
- return log_oom_debug();
-
- if (!mkdtemp(temporary_mount))
- return log_debug_errno(errno, "Failed to create temporary directory '%s': %m", temporary_mount);
+ r = create_temporary_mount_point(scope, &temporary_mount);
+ if (r < 0)
+ return r;
dev = strjoina(temporary_mount, "/dev");
(void) mkdir(dev, 0755);
if (r < 0)
log_debug_errno(r, "Failed to set up basic device tree at '%s', ignoring: %m", temporary_mount);
- /* Make the bind mount read-only. */
- r = mount_nofollow_verbose(LOG_DEBUG, NULL, dev, NULL, MS_REMOUNT|MS_BIND|MS_RDONLY, NULL);
- if (r < 0)
- return r;
-
/* Create the /dev directory if missing. It is more likely to be missing when the service is started
* with RootDirectory. This is consistent with mount units creating the mount points when missing. */
(void) mkdir_p_label(mount_entry_path(m), 0755);
(void) rmdir(dev);
(void) rmdir(temporary_mount);
- return 0;
+ return 1;
fail:
if (devpts)
(void) mkdir_p_label(mount_entry_path(m), 0755);
- r = path_is_mount_point(mount_entry_path(m), NULL, 0);
+ r = path_is_mount_point(mount_entry_path(m));
if (r < 0)
return log_debug_errno(r, "Unable to determine whether /dev is already mounted: %m");
if (r > 0) /* make this a NOP if /dev is already a mount point */
return 0;
- return mount_nofollow_verbose(LOG_DEBUG, "/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
+ r = mount_nofollow_verbose(LOG_DEBUG, "/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
+ if (r < 0)
+ return r;
+
+ return 1;
}
static int mount_bind_sysfs(const MountEntry *m) {
(void) mkdir_p_label(mount_entry_path(m), 0755);
- r = path_is_mount_point(mount_entry_path(m), NULL, 0);
+ r = path_is_mount_point(mount_entry_path(m));
if (r < 0)
return log_debug_errno(r, "Unable to determine whether /sys is already mounted: %m");
if (r > 0) /* make this a NOP if /sys is already a mount point */
return 0;
/* Bind mount the host's version so that we get all child mounts of it, too. */
- return mount_nofollow_verbose(LOG_DEBUG, "/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
+ r = mount_nofollow_verbose(LOG_DEBUG, "/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
+ if (r < 0)
+ return r;
+
+ return 1;
}
-static int mount_private_sysfs(const MountEntry *m) {
- const char *entry_path = mount_entry_path(ASSERT_PTR(m));
- int r, n;
+static int mount_private_apivfs(
+ const char *fstype,
+ const char *entry_path,
+ const char *bind_source,
+ const char *opts,
+ RuntimeScope scope) {
+
+ _cleanup_(rmdir_and_freep) char *temporary_mount = NULL;
+ int r;
+
+ assert(fstype);
+ assert(entry_path);
+ assert(bind_source);
(void) mkdir_p_label(entry_path, 0755);
- n = umount_recursive(entry_path, 0);
+ /* First, check if we have enough privileges to mount a new instance. Note, a new sysfs instance
+ * cannot be mounted on an already existing mount. Let's use a temporary place. */
+ r = create_temporary_mount_point(scope, &temporary_mount);
+ if (r < 0)
+ return r;
- r = mount_nofollow_verbose(LOG_DEBUG, "sysfs", entry_path, "sysfs", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
+ r = mount_nofollow_verbose(LOG_DEBUG, fstype, temporary_mount, fstype, MS_NOSUID|MS_NOEXEC|MS_NODEV, opts);
+ if (r == -EINVAL && opts)
+ /* If this failed with EINVAL then this likely means the textual hidepid= stuff for procfs is
+ * not supported by the kernel, and thus the per-instance hidepid= neither, which means we
+ * really don't want to use it, since it would affect our host's /proc mount. Hence let's
+ * gracefully fallback to a classic, unrestricted version. */
+ r = mount_nofollow_verbose(LOG_DEBUG, fstype, temporary_mount, fstype, MS_NOSUID|MS_NOEXEC|MS_NODEV, /* opts = */ NULL);
if (ERRNO_IS_NEG_PRIVILEGE(r)) {
- /* When we do not have enough privileges to mount sysfs, fall back to use existing /sys. */
+ /* When we do not have enough privileges to mount a new instance, fall back to use an
+ * existing mount. */
+
+ r = path_is_mount_point(entry_path);
+ if (r < 0)
+ return log_debug_errno(r, "Unable to determine whether '%s' is already mounted: %m", entry_path);
+ if (r > 0)
+ return 0; /* Use the current mount as is. */
- if (n > 0)
- /* /sys or some of sub-mounts are umounted in the above. Refuse incomplete tree.
- * Propagate the original error code returned by mount() in the above. */
+ /* We lack permissions to mount a new instance, and it is not already mounted. But we can
+ * access the host's, so as a final fallback bind-mount it to the destination, as most likely
+ * we are inside a user manager in an unprivileged user namespace. */
+ r = mount_nofollow_verbose(LOG_DEBUG, bind_source, entry_path, /* fstype = */ NULL, MS_BIND|MS_REC, /* opts = */ NULL);
+ if (r < 0)
return r;
- return mount_bind_sysfs(m);
+ return 1;
} else if (r < 0)
return r;
- /* We mounted a new instance now. Let's bind mount the children over now. */
- (void) bind_mount_submounts("/sys", entry_path);
- return 0;
+ /* OK. We have a new mount instance. Let's clear an existing mount and its submounts. */
+ r = umount_recursive(entry_path, /* flags = */ 0);
+ if (r < 0)
+ log_debug_errno(r, "Failed to unmount directories below '%s', ignoring: %m", entry_path);
+
+ /* Then, move the new mount instance. */
+ r = mount_nofollow_verbose(LOG_DEBUG, temporary_mount, entry_path, /* fstype = */ NULL, MS_MOVE, /* opts = */ NULL);
+ if (r < 0)
+ return r;
+
+ /* We mounted a new instance now. Let's bind mount the children over now. This matters for nspawn
+ * where a bunch of files are overmounted, in particular the boot id. */
+ (void) bind_mount_submounts(bind_source, entry_path);
+ return 1;
+}
+
+static int mount_private_sysfs(const MountEntry *m, const NamespaceParameters *p) {
+ assert(m);
+ assert(p);
+ return mount_private_apivfs("sysfs", mount_entry_path(m), "/sys", /* opts = */ NULL, p->runtime_scope);
}
static int mount_procfs(const MountEntry *m, const NamespaceParameters *p) {
_cleanup_free_ char *opts = NULL;
- const char *entry_path;
- int r, n;
assert(m);
assert(p);
return -ENOMEM;
}
- entry_path = mount_entry_path(m);
- (void) mkdir_p_label(entry_path, 0755);
-
/* Mount a new instance, so that we get the one that matches our user namespace, if we are running in
* one. i.e we don't reuse existing mounts here under any condition, we want a new instance owned by
* our user namespace and with our hidepid= settings applied. Hence, let's get rid of everything
* mounted on /proc/ first. */
-
- n = umount_recursive(entry_path, 0);
-
- r = mount_nofollow_verbose(LOG_DEBUG, "proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, opts);
- if (r == -EINVAL && opts)
- /* If this failed with EINVAL then this likely means the textual hidepid= stuff is
- * not supported by the kernel, and thus the per-instance hidepid= neither, which
- * means we really don't want to use it, since it would affect our host's /proc
- * mount. Hence let's gracefully fallback to a classic, unrestricted version. */
- r = mount_nofollow_verbose(LOG_DEBUG, "proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
- if (ERRNO_IS_NEG_PRIVILEGE(r)) {
- /* When we do not have enough privileges to mount /proc, fall back to use existing /proc. */
-
- if (n > 0)
- /* /proc or some of sub-mounts are umounted in the above. Refuse incomplete tree.
- * Propagate the original error code returned by mount() in the above. */
- return r;
-
- r = path_is_mount_point(entry_path, NULL, 0);
- if (r < 0)
- return log_debug_errno(r, "Unable to determine whether /proc is already mounted: %m");
- if (r > 0)
- return 0;
-
- /* We lack permissions to mount a new instance of /proc, and it is not already mounted. But
- * we can access the host's, so as a final fallback bind-mount it to the destination, as most
- * likely we are inside a user manager in an unprivileged user namespace. */
- return mount_nofollow_verbose(LOG_DEBUG, "/proc", entry_path, NULL, MS_BIND|MS_REC, NULL);
-
- } else if (r < 0)
- return r;
-
- /* We mounted a new instance now. Let's bind mount the children over now. This matters for nspawn
- * where a bunch of files are overmounted, in particular the boot id */
- (void) bind_mount_submounts("/proc", entry_path);
- return 0;
+ return mount_private_apivfs("proc", mount_entry_path(m), "/proc", opts, p->runtime_scope);
}
static int mount_tmpfs(const MountEntry *m) {
if (r < 0)
return log_debug_errno(r, "Failed to fix label of '%s' as '%s': %m", entry_path, inner_path);
- return 0;
+ return 1;
}
static int mount_run(const MountEntry *m) {
assert(m);
- r = path_is_mount_point(mount_entry_path(m), NULL, 0);
+ r = path_is_mount_point(mount_entry_path(m));
if (r < 0 && r != -ENOENT)
return log_debug_errno(r, "Unable to determine whether /run is already mounted: %m");
if (r > 0) /* make this a NOP if /run is already a mount point */
if (r < 0)
return r;
- return 0;
+ return 1;
}
static int mount_image(
if (r < 0)
return log_debug_errno(r, "Failed to extract extension name from %s: %m", mount_entry_source(m));
- if (m->mode == EXTENSION_IMAGES) {
+ if (m->mode == MOUNT_EXTENSION_IMAGE) {
r = parse_os_release(
empty_to_root(root_directory),
"ID", &host_os_release_id,
if (r < 0)
return log_debug_errno(r, "Failed to mount image %s on %s: %m", mount_entry_source(m), mount_entry_path(m));
- return 0;
+ return 1;
}
static int mount_overlay(const MountEntry *m) {
- const char *options;
+ _cleanup_free_ char *options = NULL, *layers = NULL;
int r;
assert(m);
- options = strjoina("lowerdir=", mount_entry_options(m));
+ /* Extension hierarchies are optional (e.g.: confext might not have /opt) so check if they actually
+ * exist in an image before attempting to create an overlay with them, otherwise the mount will
+ * fail. We can't check before this, as the images will not be mounted until now. */
+
+ /* Note that lowerdir= parameters are in 'reverse' order, so the top-most directory in the overlay
+ * comes first in the list. */
+ STRV_FOREACH_BACKWARDS(o, m->overlay_layers) {
+ _cleanup_free_ char *escaped = NULL;
+
+ r = is_dir(*o, /* follow= */ false);
+ if (r <= 0) {
+ if (r != -ENOENT)
+ log_debug_errno(r,
+ "Failed to check whether overlay layer source path '%s' exists, ignoring: %m",
+ *o);
+ continue;
+ }
+
+ escaped = shell_escape(*o, ",:");
+ if (!escaped)
+ return log_oom_debug();
+
+ if (!strextend_with_separator(&layers, ":", escaped))
+ return log_oom_debug();
+ }
+
+ if (!layers) {
+ log_debug("None of the overlays specified in '%s' exist at the source, skipping.",
+ mount_entry_options(m));
+ return 0; /* Only the root is set? Then there's nothing to overlay */
+ }
+
+ options = strjoin("lowerdir=", layers, ":", mount_entry_path(m)); /* The root goes in last */
+ if (!options)
+ return log_oom_debug();
(void) mkdir_p_label(mount_entry_path(m), 0755);
r = mount_nofollow_verbose(LOG_DEBUG, "overlay", mount_entry_path(m), "overlay", MS_RDONLY, options);
if (r == -ENOENT && m->ignore)
return 0;
+ if (r < 0)
+ return r;
- return r;
+ return 1;
}
static int follow_symlink(
mount_entry_consume_prefix(m, TAKE_PTR(target));
- m->n_followed ++;
+ m->n_followed++;
return 0;
}
const char *what;
int r;
+ /* Return 1 when the mount should be post-processed (remounted r/o, etc.), 0 otherwise. In most
+ * cases post-processing is the right thing, the typical exception is when the mount is gracefully
+ * skipped. */
+
assert(m);
assert(p);
switch (m->mode) {
- case INACCESSIBLE: {
+ case MOUNT_INACCESSIBLE: {
_cleanup_free_ char *runtime_dir = NULL;
struct stat target;
break;
}
- case READONLY:
- case READWRITE:
- case READWRITE_IMPLICIT:
- case EXEC:
- case NOEXEC:
- r = path_is_mount_point(mount_entry_path(m), root_directory, 0);
+ case MOUNT_READ_ONLY:
+ case MOUNT_READ_WRITE:
+ case MOUNT_READ_WRITE_IMPLICIT:
+ case MOUNT_EXEC:
+ case MOUNT_NOEXEC:
+ r = path_is_mount_point_full(mount_entry_path(m), root_directory, /* flags = */ 0);
if (r == -ENOENT && m->ignore)
return 0;
if (r < 0)
mount_entry_path(m));
if (r > 0) /* Nothing to do here, it is already a mount. We just later toggle the MS_RDONLY
* and MS_NOEXEC bits for the mount point if needed. */
- return 0;
+ return 1;
/* This isn't a mount point yet, let's make it one. */
what = mount_entry_path(m);
break;
- case EXTENSION_DIRECTORIES: {
+ case MOUNT_EXTENSION_DIRECTORY: {
_cleanup_free_ char *host_os_release_id = NULL, *host_os_release_version_id = NULL,
*host_os_release_level = NULL, *extension_name = NULL;
_cleanup_strv_free_ char **extension_release = NULL;
_fallthrough_;
}
- case BIND_MOUNT:
+ case MOUNT_BIND:
rbind = false;
_fallthrough_;
- case BIND_MOUNT_RECURSIVE: {
+ case MOUNT_BIND_RECURSIVE: {
_cleanup_free_ char *chased = NULL;
/* Since mount() will always follow symlinks we chase the symlinks on our own first. Note
break;
}
- case EMPTY_DIR:
- case TMPFS:
+ case MOUNT_EMPTY_DIR:
+ case MOUNT_TMPFS:
return mount_tmpfs(m);
- case PRIVATE_TMP:
- case PRIVATE_TMP_READONLY:
+ case MOUNT_PRIVATE_TMP:
+ case MOUNT_PRIVATE_TMP_READ_ONLY:
what = mount_entry_source(m);
make = true;
break;
- case PRIVATE_DEV:
+ case MOUNT_PRIVATE_DEV:
return mount_private_dev(m, p->runtime_scope);
- case BIND_DEV:
+ case MOUNT_BIND_DEV:
return mount_bind_dev(m);
- case PRIVATE_SYSFS:
- return mount_private_sysfs(m);
+ case MOUNT_PRIVATE_SYSFS:
+ return mount_private_sysfs(m, p);
- case BIND_SYSFS:
+ case MOUNT_BIND_SYSFS:
return mount_bind_sysfs(m);
- case PROCFS:
+ case MOUNT_PROCFS:
return mount_procfs(m, p);
- case RUN:
+ case MOUNT_RUN:
return mount_run(m);
- case MQUEUEFS:
+ case MOUNT_MQUEUEFS:
return mount_mqueuefs(m);
- case MOUNT_IMAGES:
+ case MOUNT_IMAGE:
return mount_image(m, NULL, p->mount_image_policy);
- case EXTENSION_IMAGES:
+ case MOUNT_EXTENSION_IMAGE:
return mount_image(m, root_directory, p->extension_image_policy);
- case OVERLAY_MOUNT:
+ case MOUNT_OVERLAY:
return mount_overlay(m);
default:
}
log_debug("Successfully mounted %s to %s", what, mount_entry_path(m));
- return 0;
+ return 1;
}
static int make_read_only(const MountEntry *m, char **deny_list, FILE *proc_self_mountinfo) {
assert(m);
assert(proc_self_mountinfo);
- if (mount_entry_read_only(m) || m->mode == PRIVATE_DEV) {
+ if (m->state != MOUNT_APPLIED)
+ return 0;
+
+ if (mount_entry_read_only(m) || m->mode == MOUNT_PRIVATE_DEV) {
new_flags |= MS_RDONLY;
flags_mask |= MS_RDONLY;
}
* and running Linux <= 4.17. */
submounts =
mount_entry_read_only(m) &&
- !IN_SET(m->mode, EMPTY_DIR, TMPFS);
+ !IN_SET(m->mode, MOUNT_EMPTY_DIR, MOUNT_TMPFS);
if (submounts)
r = bind_remount_recursive_with_mountinfo(mount_entry_path(m), new_flags, flags_mask, deny_list, proc_self_mountinfo);
else
assert(m);
assert(proc_self_mountinfo);
+ if (m->state != MOUNT_APPLIED)
+ return 0;
+
if (mount_entry_noexec(m)) {
new_flags |= MS_NOEXEC;
flags_mask |= MS_NOEXEC;
if (flags_mask == 0) /* No Change? */
return 0;
- submounts = !IN_SET(m->mode, EMPTY_DIR, TMPFS);
+ submounts = !IN_SET(m->mode, MOUNT_EMPTY_DIR, MOUNT_TMPFS);
if (submounts)
r = bind_remount_recursive_with_mountinfo(mount_entry_path(m), new_flags, flags_mask, deny_list, proc_self_mountinfo);
assert(m);
assert(proc_self_mountinfo);
- submounts = !IN_SET(m->mode, EMPTY_DIR, TMPFS);
+ if (m->state != MOUNT_APPLIED)
+ return 0;
+ submounts = !IN_SET(m->mode, MOUNT_EMPTY_DIR, MOUNT_TMPFS);
if (submounts)
r = bind_remount_recursive_with_mountinfo(mount_entry_path(m), MS_NOSUID, MS_NOSUID, NULL, proc_self_mountinfo);
else
FOREACH_ARRAY(m, ml->mounts, ml->n_mounts) {
- if (m->applied)
+ if (m->state != MOUNT_PENDING)
continue;
/* ExtensionImages/Directories are first opened in the propagate directory, not in the root_directory */
- r = follow_symlink(!IN_SET(m->mode, EXTENSION_IMAGES, EXTENSION_DIRECTORIES) ? root : NULL, m);
+ r = follow_symlink(!IN_SET(m->mode, MOUNT_EXTENSION_IMAGE, MOUNT_EXTENSION_DIRECTORY) ? root : NULL, m);
if (r < 0) {
mount_entry_path_debug_string(root, m, error_path);
return r;
break;
}
+ /* Returns 1 if the mount should be post-processed, 0 otherwise */
r = apply_one_mount(root, m, p);
if (r < 0) {
mount_entry_path_debug_string(root, m, error_path);
return r;
}
-
- m->applied = true;
+ m->state = r == 0 ? MOUNT_SKIPPED : MOUNT_APPLIED;
}
if (!again)
/* Third round, flip the noexec bits with a simplified deny list. */
for (size_t j = 0; j < ml->n_mounts; j++)
- if (IN_SET((ml->mounts+j)->mode, EXEC, NOEXEC))
+ if (IN_SET((ml->mounts+j)->mode, MOUNT_EXEC, MOUNT_NOEXEC))
deny_list[j] = (char*) mount_entry_path(ml->mounts+j);
deny_list[ml->n_mounts] = NULL;
DISSECT_IMAGE_USR_NO_ROOT |
DISSECT_IMAGE_GROWFS |
DISSECT_IMAGE_ADD_PARTITION_DEVICES |
- DISSECT_IMAGE_PIN_PARTITION_DEVICES;
+ DISSECT_IMAGE_PIN_PARTITION_DEVICES |
+ DISSECT_IMAGE_ALLOW_USERSPACE_VERITY;
int r;
assert(p);
return r;
}
- r = append_access_mounts(&ml, p->read_write_paths, READWRITE, require_prefix);
+ r = append_access_mounts(&ml, p->read_write_paths, MOUNT_READ_WRITE, require_prefix);
if (r < 0)
return r;
- r = append_access_mounts(&ml, p->read_only_paths, READONLY, require_prefix);
+ r = append_access_mounts(&ml, p->read_only_paths, MOUNT_READ_ONLY, require_prefix);
if (r < 0)
return r;
- r = append_access_mounts(&ml, p->inaccessible_paths, INACCESSIBLE, require_prefix);
+ r = append_access_mounts(&ml, p->inaccessible_paths, MOUNT_INACCESSIBLE, require_prefix);
if (r < 0)
return r;
- r = append_access_mounts(&ml, p->exec_paths, EXEC, require_prefix);
+ r = append_access_mounts(&ml, p->exec_paths, MOUNT_EXEC, require_prefix);
if (r < 0)
return r;
- r = append_access_mounts(&ml, p->no_exec_paths, NOEXEC, require_prefix);
+ r = append_access_mounts(&ml, p->no_exec_paths, MOUNT_NOEXEC, require_prefix);
if (r < 0)
return r;
*me = (MountEntry) {
.path_const = "/tmp",
- .mode = ro ? PRIVATE_TMP_READONLY : PRIVATE_TMP,
+ .mode = ro ? MOUNT_PRIVATE_TMP_READ_ONLY : MOUNT_PRIVATE_TMP,
.source_const = p->tmp_dir,
};
}
*me = (MountEntry) {
.path_const = "/var/tmp",
- .mode = ro ? PRIVATE_TMP_READONLY : PRIVATE_TMP,
+ .mode = ro ? MOUNT_PRIVATE_TMP_READ_ONLY : MOUNT_PRIVATE_TMP,
.source_const = p->var_tmp_dir,
};
}
*me = (MountEntry) {
.path_const = "/dev",
- .mode = PRIVATE_DEV,
+ .mode = MOUNT_PRIVATE_DEV,
.flags = DEV_MOUNT_OPTIONS,
};
}
*me = (MountEntry) {
.path_const = "/sys/fs/cgroup",
- .mode = READONLY,
+ .mode = MOUNT_READ_ONLY,
};
}
*me = (MountEntry) {
.path_const = "/sys",
- .mode = PRIVATE_SYSFS,
+ .mode = MOUNT_PRIVATE_SYSFS,
};
}
*me = (MountEntry) {
.path_const = "/dev/mqueue",
- .mode = MQUEUEFS,
+ .mode = MOUNT_MQUEUEFS,
.flags = MS_NOSUID | MS_NODEV | MS_NOEXEC | MS_RELATIME,
};
}
*me = (MountEntry) {
.path_const = "/run/credentials",
- .mode = TMPFS,
+ .mode = MOUNT_TMPFS,
.read_only = true,
.options_const = "mode=0755" TMPFS_LIMITS_EMPTY_OR_ALMOST,
.flags = MS_NODEV|MS_STRICTATIME|MS_NOSUID|MS_NOEXEC,
*me = (MountEntry) {
.path_const = p->creds_path,
- .mode = BIND_MOUNT,
+ .mode = MOUNT_BIND,
.read_only = true,
.source_const = p->creds_path,
.ignore = true,
*me = (MountEntry) {
.path_const = "/run/credentials",
- .mode = INACCESSIBLE,
+ .mode = MOUNT_INACCESSIBLE,
.ignore = true,
};
}
*me = (MountEntry) {
.path_const = "/run/systemd/journal",
- .mode = BIND_MOUNT_RECURSIVE,
+ .mode = MOUNT_BIND_RECURSIVE,
.read_only = true,
.source_malloc = TAKE_PTR(q),
};
*me = (MountEntry) {
.source_const = p->propagate_dir,
.path_const = p->incoming_dir,
- .mode = BIND_MOUNT,
+ .mode = MOUNT_BIND,
.read_only = true,
};
}
*me = (MountEntry) {
.path_const = p->notify_socket,
.source_const = p->notify_socket,
- .mode = BIND_MOUNT,
+ .mode = MOUNT_BIND,
.read_only = true,
};
}
*me = (MountEntry) {
.path_const = "/run/host/.os-release-stage/",
.source_const = p->host_os_release_stage,
- .mode = BIND_MOUNT,
+ .mode = MOUNT_BIND,
.read_only = true,
.ignore = true, /* Live copy, don't hard-fail if it goes missing */
};
} else if (p->root_directory) {
/* 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);
+ r = path_is_mount_point_full(root, /* root = */ NULL, AT_SYMLINK_FOLLOW);
if (r < 0)
return log_debug_errno(r, "Failed to detect that %s is a mount point or not: %m", root);
if (r == 0) {
*b = c;
- c[(*n) ++] = (BindMount) {
+ c[(*n)++] = (BindMount) {
.source = TAKE_PTR(s),
.destination = TAKE_PTR(d),
.read_only = item->read_only,
*m = c;
- c[(*n) ++] = (MountImage) {
+ c[(*n)++] = (MountImage) {
.source = TAKE_PTR(s),
.destination = TAKE_PTR(d),
.mount_options = TAKE_PTR(options),
*t = c;
- c[(*n) ++] = (TemporaryFileSystem) {
+ c[(*n)++] = (TemporaryFileSystem) {
.path = TAKE_PTR(p),
.options = TAKE_PTR(o),
};
int setup_shareable_ns(int ns_storage_socket[static 2], unsigned long nsflag) {
_cleanup_close_ int ns = -EBADF;
- int r;
const char *ns_name, *ns_path;
+ int r;
assert(ns_storage_socket);
assert(ns_storage_socket[0] >= 0);
assert(ns_storage_socket[1] >= 0);
- ns_name = namespace_single_flag_to_string(nsflag);
- assert(ns_name);
+ ns_name = ASSERT_PTR(namespace_single_flag_to_string(nsflag));
- /* We use the passed socketpair as a storage buffer for our
- * namespace reference fd. Whatever process runs this first
- * shall create a new namespace, all others should just join
- * it. To serialize that we use a file lock on the socket
- * pair.
+ /* We use the passed socketpair as a storage buffer for our namespace reference fd. Whatever process
+ * runs this first shall create a new namespace, all others should just join it. To serialize that we
+ * use a file lock on the socket pair.
*
* It's a bit crazy, but hey, works great! */
if (unshare(nsflag) < 0)
return -errno;
- (void) loopback_setup();
+ if (nsflag == CLONE_NEWNET)
+ (void) loopback_setup();
ns_path = strjoina("/proc/self/ns/", ns_name);
ns = open(ns_path, O_RDONLY|O_CLOEXEC|O_NOCTTY);