Merge pull request #17118 from poettering/sync-shutdown-loop

fsync() block devices before detaching them during shutdown

TODO

@@ -20,10 +20,6 @@ Janitorial Clean-ups:
 
 Features:
 
-* add wrapper for mount() that uses O_PATH on the destination and than mounts
-  to /proc/self/fd/xxx so that we basically have a mount() with O_NOFOLLOW like
-  behaviour. (in case of bind mounts do it on both source and target)
-
 * add root=tmpfs that mounts a tmpfs to /sysroot (to be used in combination
   with usr=…, for a similar effect as systemd.volatile=yes but without the
   "hide-out" effect). Also, add root=gpt-auto-late support or so, that is like

src/core/automount.c

@@ -94,7 +94,7 @@ static void unmount_autofs(Automount *a) {
                 automount_send_ready(a, a->expire_tokens, -EHOSTDOWN);
 
                 if (a->where) {
-                        r = repeat_unmount(a->where, MNT_DETACH);
+                        r = repeat_unmount(a->where, MNT_DETACH|UMOUNT_NOFOLLOW);
                         if (r < 0)
                                 log_error_errno(r, "Failed to unmount: %m");
                 }
@@ -601,10 +601,9 @@ static void automount_enter_waiting(Automount *a) {
 
         xsprintf(options, "fd=%i,pgrp="PID_FMT",minproto=5,maxproto=5,direct", p[1], getpgrp());
         xsprintf(name, "systemd-"PID_FMT, getpid_cached());
-        if (mount(name, a->where, "autofs", 0, options) < 0) {
+        r = mount_nofollow(name, a->where, "autofs", 0, options);
-                r = -errno;
+        if (r < 0)
                 goto fail;
-        }
 
         mounted = true;
 
@@ -648,7 +647,7 @@ fail:
         safe_close_pair(p);
 
         if (mounted) {
-                r = repeat_unmount(a->where, MNT_DETACH);
+                r = repeat_unmount(a->where, MNT_DETACH|UMOUNT_NOFOLLOW);
                 if (r < 0)
                         log_error_errno(r, "Failed to unmount, ignoring: %m");
         }

src/core/execute.c

@@ -67,6 +67,7 @@
 #include "memory-util.h"
 #include "missing_fs.h"
 #include "mkdir.h"
+#include "mount-util.h"
 #include "mountpoint-util.h"
 #include "namespace.h"
 #include "parse-util.h"
@@ -2652,11 +2653,13 @@ static int setup_credentials_internal(
                          * the final version to the workspace, and make it writable, so that we can make
                          * changes */
 
-                        if (mount(final, workspace, NULL, MS_BIND|MS_REC, NULL) < 0)
+                        r = mount_nofollow_verbose(LOG_DEBUG, final, workspace, NULL, MS_BIND|MS_REC, NULL);
-                                return -errno;
+                        if (r < 0)
+                                return r;
 
-                        if (mount(NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL) < 0)
+                        r = mount_nofollow_verbose(LOG_DEBUG, NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL);
-                                return -errno;
+                        if (r < 0)
+                                return r;
 
                         workspace_mounted = true;
                 }
@@ -2669,7 +2672,8 @@ static int setup_credentials_internal(
 
                         if (try == 0) {
                                 /* Try "ramfs" first, since it's not swap backed */
-                                if (mount("ramfs", workspace, "ramfs", MS_NODEV|MS_NOEXEC|MS_NOSUID, "mode=0700") >= 0) {
+                                r = mount_nofollow_verbose(LOG_DEBUG, "ramfs", workspace, "ramfs", MS_NODEV|MS_NOEXEC|MS_NOSUID, "mode=0700");
+                                if (r >= 0) {
                                         workspace_mounted = true;
                                         break;
                                 }
@@ -2681,20 +2685,22 @@ static int setup_credentials_internal(
                                         return -ENOMEM;
 
                                 /* Fall back to "tmpfs" otherwise */
-                                if (mount("tmpfs", workspace, "tmpfs", MS_NODEV|MS_NOEXEC|MS_NOSUID, opts) >= 0) {
+                                r = mount_nofollow_verbose(LOG_DEBUG, "tmpfs", workspace, "tmpfs", MS_NODEV|MS_NOEXEC|MS_NOSUID, opts);
+                                if (r >= 0) {
                                         workspace_mounted = true;
                                         break;
                                 }
 
                         } else {
                                 /* If that didn't work, try to make a bind mount from the final to the workspace, so that we can make it writable there. */
-                                if (mount(final, workspace, NULL, MS_BIND|MS_REC, NULL) < 0) {
+                                r = mount_nofollow_verbose(LOG_DEBUG, final, workspace, NULL, MS_BIND|MS_REC, NULL);
-                                        if (!ERRNO_IS_PRIVILEGE(errno)) /* Propagate anything that isn't a permission problem */
+                                if (r < 0) {
-                                                return -errno;
+                                        if (!ERRNO_IS_PRIVILEGE(r)) /* Propagate anything that isn't a permission problem */
+                                                return r;
 
                                         if (must_mount) /* If we it's not OK to use the plain directory
                                                          * fallback, propagate all errors too */
-                                                return -errno;
+                                                return r;
 
                                         /* If we lack privileges to bind mount stuff, then let's gracefully
                                          * proceed for compat with container envs, and just use the final dir
@@ -2705,8 +2711,9 @@ static int setup_credentials_internal(
                                 }
 
                                 /* Make the new bind mount writable (i.e. drop MS_RDONLY) */
-                                if (mount(NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL) < 0)
+                                r = mount_nofollow_verbose(LOG_DEBUG, NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL);
-                                        return -errno;
+                                if (r < 0)
+                                        return r;
 
                                 workspace_mounted = true;
                                 break;
@@ -2723,17 +2730,17 @@ static int setup_credentials_internal(
 
         if (workspace_mounted) {
                 /* Make workspace read-only now, so that any bind mount we make from it defaults to read-only too */
-                if (mount(NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL) < 0)
+                r = mount_nofollow_verbose(LOG_DEBUG, NULL, workspace, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL);
-                        return -errno;
+                if (r < 0)
+                        return r;
 
                 /* And mount it to the final place, read-only */
-                if (final_mounted) {
+                if (final_mounted)
-                        if (umount2(workspace, MNT_DETACH|UMOUNT_NOFOLLOW) < 0)
+                        r = umount_verbose(LOG_DEBUG, workspace, MNT_DETACH|UMOUNT_NOFOLLOW);
-                                return -errno;
+                else
-                } else {
+                        r = mount_nofollow_verbose(LOG_DEBUG, workspace, final, NULL, MS_MOVE, NULL);
-                        if (mount(workspace, final, NULL, MS_MOVE, NULL) < 0)
+                if (r < 0)
-                                return -errno;
+                        return r;
-                }
         } else {
                 _cleanup_free_ char *parent = NULL;
 
@@ -2847,7 +2854,8 @@ static int setup_credentials(
                  * given that the we do this in a privately namespaced short-lived single-threaded process
                  * that no one else sees this should be OK to do.*/
 
-                if (mount(NULL, "/dev", NULL, MS_SLAVE|MS_REC, NULL) < 0) /* Turn off propagation from our namespace to host */
+                r = mount_nofollow_verbose(LOG_DEBUG, NULL, "/dev", NULL, MS_SLAVE|MS_REC, NULL); /* Turn off propagation from our namespace to host */
+                if (r < 0)
                         goto child_fail;
 
                 r = setup_credentials_internal(

src/core/machine-id-setup.c

@@ -15,6 +15,7 @@
 #include "machine-id-setup.h"
 #include "macro.h"
 #include "mkdir.h"
+#include "mount-util.h"
 #include "mountpoint-util.h"
 #include "namespace-util.h"
 #include "path-util.h"
@@ -160,16 +161,18 @@ int machine_id_setup(const char *root, sd_id128_t machine_id, sd_id128_t *ret) {
         }
 
         /* And now, let's mount it over */
-        if (mount(run_machine_id, etc_machine_id, NULL, MS_BIND, NULL) < 0) {
+        r = mount_follow_verbose(LOG_ERR, run_machine_id, etc_machine_id, NULL, MS_BIND, NULL);
-                (void) unlink_noerrno(run_machine_id);
+        if (r < 0) {
-                return log_error_errno(errno, "Failed to mount %s: %m", etc_machine_id);
+                (void) unlink(run_machine_id);
+                return r;
         }
 
         log_info("Installed transient %s file.", etc_machine_id);
 
         /* Mark the mount read-only */
-        if (mount(NULL, etc_machine_id, NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, NULL) < 0)
+        r = mount_follow_verbose(LOG_WARNING, NULL, etc_machine_id, NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, NULL);
-                log_warning_errno(errno, "Failed to make transient %s read-only, ignoring: %m", etc_machine_id);
+        if (r < 0)
+                return r;
 
 finish:
         if (ret)
@@ -227,8 +230,9 @@ int machine_id_commit(const char *root) {
         if (r < 0)
                 return log_error_errno(r, "Failed to set up new mount namespace: %m");
 
-        if (umount(etc_machine_id) < 0)
+        r = umount_verbose(LOG_ERR, etc_machine_id, 0);
-                return log_error_errno(errno, "Failed to unmount transient %s file in our private namespace: %m", etc_machine_id);
+        if (r < 0)
+                return r;
 
         /* Update a persistent version of etc_machine_id */
         r = id128_write(etc_machine_id, ID128_PLAIN, id, true);

src/core/mount-setup.c

@@ -38,6 +38,7 @@ typedef enum MountMode {
         MNT_FATAL          = 1 << 0,
         MNT_IN_CONTAINER   = 1 << 1,
         MNT_CHECK_WRITABLE = 1 << 2,
+        MNT_FOLLOW_SYMLINK = 1 << 3,
 } MountMode;
 
 typedef struct MountPoint {
@@ -61,9 +62,9 @@ typedef struct MountPoint {
 #endif
 
 static const MountPoint mount_table[] = {
-        { "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|MNT_FOLLOW_SYMLINK },
+        { "sysfs",       "/sys",                      "sysfs",      NULL,                                      MS_NOSUID|MS_NOEXEC|MS_NODEV,
           NULL,          MNT_FATAL|MNT_IN_CONTAINER },
         { "devtmpfs",    "/dev",                      "devtmpfs",   "mode=755" TMPFS_LIMITS_DEV,               MS_NOSUID|MS_NOEXEC|MS_STRICTATIME,
           NULL,          MNT_FATAL|MNT_IN_CONTAINER },
@@ -184,13 +185,13 @@ static int mount_one(const MountPoint *p, bool relabel) {
                   p->type,
                   strna(p->options));
 
-        if (mount(p->what,
+        if (FLAGS_SET(p->mode, MNT_FOLLOW_SYMLINK))
-                  p->where,
+                r = mount(p->what, p->where, p->type, p->flags, p->options) < 0 ? -errno : 0;
-                  p->type,
+        else
-                  p->flags,
+                r = mount_nofollow(p->what, p->where, p->type, p->flags, p->options);
-                  p->options) < 0) {
+        if (r < 0) {
-                log_full_errno(priority, errno, "Failed to mount %s at %s: %m", p->type, p->where);
+                log_full_errno(priority, r, "Failed to mount %s at %s: %m", p->type, p->where);
-                return (p->mode & MNT_FATAL) ? -errno : 0;
+                return (p->mode & MNT_FATAL) ? r : 0;
         }
 
         /* Relabel again, since we now mounted something fresh here */
@@ -201,7 +202,7 @@ static int mount_one(const MountPoint *p, bool relabel) {
                 if (access(p->where, W_OK) < 0) {
                         r = -errno;
 
-                        (void) umount(p->where);
+                        (void) umount2(p->where, UMOUNT_NOFOLLOW);
                         (void) rmdir(p->where);
 
                         log_full_errno(priority, r, "Mount point %s not writable after mounting: %m", p->where);
@@ -355,7 +356,7 @@ int mount_cgroup_controllers(void) {
         }
 
         /* 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" TMPFS_LIMITS_SYS_FS_CGROUP);
+        (void) mount_nofollow("tmpfs", "/sys/fs/cgroup", "tmpfs", MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755" TMPFS_LIMITS_SYS_FS_CGROUP);
 
         return 0;
 }
@@ -397,13 +398,13 @@ static int relabel_cgroup_filesystems(void) {
                         return log_error_errno(errno, "Failed to determine mount flags for /sys/fs/cgroup: %m");
 
                 if (st.f_flags & ST_RDONLY)
-                        (void) mount(NULL, "/sys/fs/cgroup", NULL, MS_REMOUNT, NULL);
+                        (void) mount_nofollow(NULL, "/sys/fs/cgroup", NULL, MS_REMOUNT, NULL);
 
                 (void) label_fix("/sys/fs/cgroup", 0);
                 (void) nftw("/sys/fs/cgroup", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL);
 
                 if (st.f_flags & ST_RDONLY)
-                        (void) mount(NULL, "/sys/fs/cgroup", NULL, MS_REMOUNT|MS_RDONLY, NULL);
+                        (void) mount_nofollow(NULL, "/sys/fs/cgroup", NULL, MS_REMOUNT|MS_RDONLY, NULL);
 
         } else if (r < 0)
                 return log_error_errno(r, "Failed to determine whether we are in all unified mode: %m");

src/core/namespace.c

@@ -657,13 +657,12 @@ static int clone_device_node(
         if (r < 0 && errno != EEXIST)
                 return log_debug_errno(errno, "mknod() fallback failed for '%s': %m", d);
 
-        /* Fallback to bind-mounting:
+        /* Fallback to bind-mounting: The assumption here is that all used device nodes carry standard
-         * The assumption here is that all used device nodes carry standard
+         * properties. Specifically, the devices nodes we bind-mount should either be owned by root:root or
-         * properties. Specifically, the devices nodes we bind-mount should
+         * root:tty (e.g. /dev/tty, /dev/ptmx) and should not carry ACLs. */
-         * either be owned by root:root or root:tty (e.g. /dev/tty, /dev/ptmx)
+        r = mount_nofollow_verbose(LOG_DEBUG, d, dn, NULL, MS_BIND, NULL);
-         * and should not carry ACLs. */
+        if (r < 0)
-        if (mount(d, dn, NULL, MS_BIND, NULL) < 0)
+                return r;
-                return log_debug_errno(errno, "Bind mounting failed for '%s': %m", d);
 
 add_symlink:
         bn = path_startswith(d, "/dev/");
@@ -710,10 +709,10 @@ static int mount_private_dev(MountEntry *m) {
 
         dev = strjoina(temporary_mount, "/dev");
         (void) mkdir(dev, 0755);
-        if (mount("tmpfs", dev, "tmpfs", DEV_MOUNT_OPTIONS, "mode=755" TMPFS_LIMITS_DEV) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, "tmpfs", dev, "tmpfs", DEV_MOUNT_OPTIONS, "mode=755" TMPFS_LIMITS_DEV);
-                r = log_debug_errno(errno, "Failed to mount tmpfs on '%s': %m", dev);
+        if (r < 0)
                 goto fail;
-        }
+
         r = label_fix_container(dev, "/dev", 0);
         if (r < 0) {
                 log_debug_errno(errno, "Failed to fix label of '%s' as /dev: %m", dev);
@@ -722,10 +721,9 @@ static int mount_private_dev(MountEntry *m) {
 
         devpts = strjoina(temporary_mount, "/dev/pts");
         (void) mkdir(devpts, 0755);
-        if (mount("/dev/pts", devpts, NULL, MS_BIND, NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, "/dev/pts", devpts, NULL, MS_BIND, NULL);
-                r = log_debug_errno(errno, "Failed to bind mount /dev/pts on '%s': %m", devpts);
+        if (r < 0)
                 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.
@@ -749,21 +747,17 @@ static int mount_private_dev(MountEntry *m) {
 
         devshm = strjoina(temporary_mount, "/dev/shm");
         (void) mkdir(devshm, 0755);
-        r = mount("/dev/shm", devshm, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_DEBUG, "/dev/shm", devshm, NULL, MS_BIND, NULL);
-        if (r < 0) {
+        if (r < 0)
-                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);
-        if (mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL) < 0)
+        (void) mount_nofollow_verbose(LOG_DEBUG, "/dev/mqueue", devmqueue, NULL, MS_BIND, NULL);
-                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);
-        if (mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL) < 0)
+        (void) mount_nofollow_verbose(LOG_DEBUG, "/dev/hugepages", devhugepages, NULL, MS_BIND, NULL);
-                log_debug_errno(errno, "Failed to bind mount /dev/hugepages on '%s', ignoring: %m", devhugepages);
 
         devlog = strjoina(temporary_mount, "/dev/log");
         if (symlink("/run/systemd/journal/dev-log", devlog) < 0)
@@ -791,10 +785,9 @@ static int mount_private_dev(MountEntry *m) {
         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 = mount_nofollow_verbose(LOG_DEBUG, dev, mount_entry_path(m), NULL, MS_MOVE, NULL);
-                r = log_debug_errno(errno, "Failed to move mount point '%s' to '%s': %m", dev, mount_entry_path(m));
+        if (r < 0)
                 goto fail;
-        }
 
         (void) rmdir(dev);
         (void) rmdir(temporary_mount);
@@ -803,18 +796,18 @@ static int mount_private_dev(MountEntry *m) {
 
 fail:
         if (devpts)
-                (void) umount(devpts);
+                (void) umount_verbose(LOG_DEBUG, devpts, UMOUNT_NOFOLLOW);
 
         if (devshm)
-                (void) umount(devshm);
+                (void) umount_verbose(LOG_DEBUG, devshm, UMOUNT_NOFOLLOW);
 
         if (devhugepages)
-                (void) umount(devhugepages);
+                (void) umount_verbose(LOG_DEBUG, devhugepages, UMOUNT_NOFOLLOW);
 
         if (devmqueue)
-                (void) umount(devmqueue);
+                (void) umount_verbose(LOG_DEBUG, devmqueue, UMOUNT_NOFOLLOW);
 
-        (void) umount(dev);
+        (void) umount_verbose(LOG_DEBUG, dev, UMOUNT_NOFOLLOW);
         (void) rmdir(dev);
         (void) rmdir(temporary_mount);
 
@@ -837,8 +830,9 @@ static int mount_bind_dev(const MountEntry *m) {
         if (r > 0) /* make this a NOP if /dev is already a mount point */
                 return 0;
 
-        if (mount("/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0)
+        r = mount_nofollow_verbose(LOG_DEBUG, "/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
-                return log_debug_errno(errno, "Failed to bind mount %s: %m", mount_entry_path(m));
+        if (r < 0)
+                return r;
 
         return 1;
 }
@@ -857,14 +851,16 @@ static int mount_sysfs(const MountEntry *m) {
                 return 0;
 
         /* Bind mount the host's version so that we get all child mounts of it, too. */
-        if (mount("/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0)
+        r = mount_nofollow_verbose(LOG_DEBUG, "/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL);
-                return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m));
+        if (r < 0)
+                return r;
 
         return 1;
 }
 
 static int mount_procfs(const MountEntry *m, const NamespaceInfo *ns_info) {
         const char *entry_path;
+        int r;
 
         assert(m);
         assert(ns_info);
@@ -896,9 +892,10 @@ static int mount_procfs(const MountEntry *m, const NamespaceInfo *ns_info) {
                 if (!opts)
                         return -ENOMEM;
 
-                if (mount("proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, opts) < 0) {
+                r = mount_nofollow_verbose(LOG_DEBUG, "proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, opts);
-                        if (errno != EINVAL)
+                if (r < 0) {
-                                return log_debug_errno(errno, "Failed to mount %s (options=%s): %m", mount_entry_path(m), opts);
+                        if (r != -EINVAL)
+                                return r;
 
                         /* 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
@@ -908,8 +905,9 @@ static int mount_procfs(const MountEntry *m, const NamespaceInfo *ns_info) {
                         return 1;
         }
 
-        if (mount("proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL) < 0)
+        r = mount_nofollow_verbose(LOG_DEBUG, "proc", entry_path, "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
-                return log_debug_errno(errno, "Failed to mount %s (no options): %m", mount_entry_path(m));
+        if (r < 0)
+                return r;
 
         return 1;
 }
@@ -928,8 +926,9 @@ static int mount_tmpfs(const MountEntry *m) {
         (void) mkdir_p_label(entry_path, 0755);
         (void) umount_recursive(entry_path, 0);
 
-        if (mount("tmpfs", entry_path, "tmpfs", m->flags, mount_entry_options(m)) < 0)
+        r = mount_nofollow_verbose(LOG_DEBUG, "tmpfs", entry_path, "tmpfs", m->flags, mount_entry_options(m));
-                return log_debug_errno(errno, "Failed to mount %s: %m", entry_path);
+        if (r < 0)
+                return r;
 
         r = label_fix_container(entry_path, inner_path, 0);
         if (r < 0)
@@ -1172,9 +1171,9 @@ static int apply_mount(
 
         assert(what);
 
-        if (mount(what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL);
+        if (r < 0) {
                 bool try_again = false;
-                r = -errno;
 
                 if (r == -ENOENT && make) {
                         struct stat st;
@@ -1202,13 +1201,8 @@ static int apply_mount(
                         }
                 }
 
-                if (try_again) {
+                if (try_again)
-                        if (mount(what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL) < 0)
+                        r = mount_nofollow_verbose(LOG_DEBUG, what, mount_entry_path(m), NULL, MS_BIND|(rbind ? MS_REC : 0), NULL);
-                                r = -errno;
-                        else
-                                r = 0;
-                }
-
                 if (r < 0)
                         return log_error_errno(r, "Failed to mount %s to %s: %m", what, mount_entry_path(m));
         }
@@ -1798,19 +1792,16 @@ int setup_namespace(
                         goto finish;
                 }
                 if (r == 0) {
-                        if (mount(root, root, NULL, MS_BIND|MS_REC, NULL) < 0) {
+                        r = mount_nofollow_verbose(LOG_DEBUG, root, root, NULL, MS_BIND|MS_REC, NULL);
-                                r = log_debug_errno(errno, "Failed to bind mount '%s': %m", root);
+                        if (r < 0)
                                 goto finish;
-                        }
                 }
 
         } 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 = mount_nofollow_verbose(LOG_DEBUG, "/", root, NULL, MS_BIND|MS_REC, NULL);
-                        r = log_debug_errno(errno, "Failed to bind mount '/' on '%s': %m", root);
+                if (r < 0)
                         goto finish;
-                }
         }
 
         /* Try to set up the new root directory before mounting anything else there. */

src/home/homework-directory.c

@@ -61,13 +61,13 @@ int home_activate_directory(
         /* Create a mount point (even if the directory is already placed correctly), as a way to indicate
          * this mount point is now "activated". Moreover, we want to set per-user
          * MS_NOSUID/MS_NOEXEC/MS_NODEV. */
-        r = mount_verbose(LOG_ERR, ip, hd, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, ip, hd, NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
-        r = mount_verbose(LOG_ERR, NULL, hd, NULL, MS_BIND|MS_REMOUNT|user_record_mount_flags(h), NULL);
+        r = mount_nofollow_verbose(LOG_ERR, NULL, hd, NULL, MS_BIND|MS_REMOUNT|user_record_mount_flags(h), NULL);
         if (r < 0) {
-                (void) umount_verbose(hd);
+                (void) umount_verbose(LOG_ERR, hd, UMOUNT_NOFOLLOW);
                 return r;
         }
 

src/home/homework-luks.c

@@ -5,6 +5,7 @@
 #include <poll.h>
 #include <sys/file.h>
 #include <sys/ioctl.h>
+#include <sys/mount.h>
 #include <sys/xattr.h>
 
 #include "blkid-util.h"
@@ -1258,7 +1259,7 @@ int home_prepare_luks(
 
 fail:
         if (mounted)
-                (void) umount_verbose("/run/systemd/user-home-mount");
+                (void) umount_verbose(LOG_ERR, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
 
         if (dm_activated)
                 (void) crypt_deactivate(cd, setup->dm_name);
@@ -2167,7 +2168,7 @@ int home_create_luks(
 
         root_fd = safe_close(root_fd);
 
-        r = umount_verbose("/run/systemd/user-home-mount");
+        r = umount_verbose(LOG_ERR, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
         if (r < 0)
                 goto fail;
 
@@ -2237,7 +2238,7 @@ fail:
         root_fd = safe_close(root_fd);
 
         if (mounted)
-                (void) umount_verbose("/run/systemd/user-home-mount");
+                (void) umount_verbose(LOG_WARNING, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
 
         if (dm_activated)
                 (void) crypt_deactivate(cd, dm_name);
@@ -2339,7 +2340,7 @@ static int ext4_offline_resize_fs(HomeSetup *setup, uint64_t new_size, bool disc
         }
 
         if (setup->undo_mount) {
-                r = umount_verbose("/run/systemd/user-home-mount");
+                r = umount_verbose(LOG_ERR, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
                 if (r < 0)
                         return r;
 

src/home/homework-mount.c

@@ -38,7 +38,7 @@ int home_mount_node(const char *node, const char *fstype, bool discard, unsigned
         } else
                 options = discard_option;
 
-        r = mount_verbose(LOG_ERR, node, "/run/systemd/user-home-mount", fstype, flags|MS_RELATIME, strempty(options));
+        r = mount_nofollow_verbose(LOG_ERR, node, "/run/systemd/user-home-mount", fstype, flags|MS_RELATIME, strempty(options));
         if (r < 0)
                 return r;
 
@@ -52,7 +52,7 @@ int home_unshare_and_mount(const char *node, const char *fstype, bool discard, u
         if (unshare(CLONE_NEWNS) < 0)
                 return log_error_errno(errno, "Couldn't unshare file system namespace: %m");
 
-        r = mount_verbose(LOG_ERR, "/run", "/run", NULL, MS_SLAVE|MS_REC, NULL); /* Mark /run as MS_SLAVE in our new namespace */
+        r = mount_nofollow_verbose(LOG_ERR, "/run", "/run", NULL, MS_SLAVE|MS_REC, NULL); /* Mark /run as MS_SLAVE in our new namespace */
         if (r < 0)
                 return r;
 
@@ -83,11 +83,11 @@ int home_move_mount(const char *user_name_and_realm, const char *target) {
 
         (void) mkdir_p(target, 0700);
 
-        r = mount_verbose(LOG_ERR, d, target, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, d, target, NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
-        r = umount_verbose("/run/systemd/user-home-mount");
+        r = umount_verbose(LOG_ERR, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
         if (r < 0)
                 return r;
 

src/home/homework.c

@@ -291,7 +291,7 @@ int home_setup_undo(HomeSetup *setup) {
         }
 
         if (setup->undo_mount) {
-                q = umount_verbose("/run/systemd/user-home-mount");
+                q = umount_verbose(LOG_DEBUG, "/run/systemd/user-home-mount", UMOUNT_NOFOLLOW);
                 if (q < 0)
                         r = q;
         }

src/login/user-runtime-dir.c

@@ -13,6 +13,7 @@
 #include "limits-util.h"
 #include "main-func.h"
 #include "mkdir.h"
+#include "mount-util.h"
 #include "mountpoint-util.h"
 #include "path-util.h"
 #include "rm-rf.h"
@@ -81,14 +82,14 @@ static int user_mkdir_runtime_path(
 
                 (void) mkdir_label(runtime_path, 0700);
 
-                r = mount("tmpfs", runtime_path, "tmpfs", MS_NODEV|MS_NOSUID, options);
+                r = mount_nofollow_verbose(LOG_DEBUG, "tmpfs", runtime_path, "tmpfs", MS_NODEV|MS_NOSUID, options);
                 if (r < 0) {
-                        if (!ERRNO_IS_PRIVILEGE(errno)) {
+                        if (!ERRNO_IS_PRIVILEGE(r)) {
-                                r = log_error_errno(errno, "Failed to mount per-user tmpfs directory %s: %m", runtime_path);
+                                log_error_errno(r, "Failed to mount per-user tmpfs directory %s: %m", runtime_path);
                                 goto fail;
                         }
 
-                        log_debug_errno(errno,
+                        log_debug_errno(r,
                                         "Failed to mount per-user tmpfs directory %s.\n"
                                         "Assuming containerized execution, ignoring: %m", runtime_path);
 

src/machine/machine-dbus.c

@@ -31,6 +31,7 @@
 #include "machine.h"
 #include "missing_capability.h"
 #include "mkdir.h"
+#include "mount-util.h"
 #include "namespace-util.h"
 #include "os-util.h"
 #include "path-util.h"
@@ -891,15 +892,17 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
 
         mount_slave_created = true;
 
-        if (mount(mount_slave, mount_slave, NULL, MS_BIND, NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, mount_slave, mount_slave, NULL, MS_BIND, NULL);
-                r = sd_bus_error_set_errnof(error, errno, "Failed to make bind mount %s: %m", mount_slave);
+        if (r < 0) {
+                sd_bus_error_set_errnof(error, r, "Failed to make bind mount %s: %m", mount_slave);
                 goto finish;
         }
 
         mount_slave_mounted = true;
 
-        if (mount(NULL, mount_slave, NULL, MS_SLAVE, NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, NULL, mount_slave, NULL, MS_SLAVE, NULL);
-                r = sd_bus_error_set_errnof(error, errno, "Failed to remount slave %s: %m", mount_slave);
+        if (r < 0) {
+                sd_bus_error_set_errnof(error, r, "Failed to remount slave %s: %m", mount_slave);
                 goto finish;
         }
 
@@ -916,19 +919,22 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
 
         mount_tmp_created = true;
 
-        if (mount(chased_src, mount_tmp, NULL, MS_BIND, NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, chased_src, mount_tmp, NULL, MS_BIND, NULL);
-                r = sd_bus_error_set_errnof(error, errno, "Failed to mount %s: %m", chased_src);
+        if (r < 0) {
+                sd_bus_error_set_errnof(error, r, "Failed to mount %s: %m", chased_src);
                 goto finish;
         }
 
         mount_tmp_mounted = true;
 
         /* Third, we remount the new bind mount read-only if requested. */
-        if (read_only)
+        if (read_only) {
-                if (mount(NULL, mount_tmp, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL) < 0) {
+                r = mount_nofollow_verbose(LOG_DEBUG, NULL, mount_tmp, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
-                        r = sd_bus_error_set_errnof(error, errno, "Failed to remount read-only %s: %m", mount_tmp);
+                if (r < 0) {
+                        sd_bus_error_set_errnof(error, r, "Failed to remount read-only %s: %m", mount_tmp);
                         goto finish;
                 }
+        }
 
         /* Fourth, we move the new bind mount into the propagation directory. This way it will appear there read-only
          * right-away. */
@@ -947,8 +953,9 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
 
         mount_outside_created = true;
 
-        if (mount(mount_tmp, mount_outside, NULL, MS_MOVE, NULL) < 0) {
+        r = mount_nofollow_verbose(LOG_DEBUG, mount_tmp, mount_outside, NULL, MS_MOVE, NULL);
-                r = sd_bus_error_set_errnof(error, errno, "Failed to move %s to %s: %m", mount_tmp, mount_outside);
+        if (r < 0) {
+                sd_bus_error_set_errnof(error, r, "Failed to move %s to %s: %m", mount_tmp, mount_outside);
                 goto finish;
         }
 
@@ -961,7 +968,7 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
                 (void) unlink(mount_tmp);
         mount_tmp_created = false;
 
-        (void) umount(mount_slave);
+        (void) umount_verbose(LOG_DEBUG, mount_slave, UMOUNT_NOFOLLOW);
         mount_slave_mounted = false;
 
         (void) rmdir(mount_slave);
@@ -1005,10 +1012,9 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
                 }
 
                 mount_inside = strjoina("/run/host/incoming/", basename(mount_outside));
-                if (mount(mount_inside, dest, NULL, MS_MOVE, NULL) < 0) {
+                r = mount_nofollow_verbose(LOG_ERR, mount_inside, dest, NULL, MS_MOVE, NULL);
-                        r = log_error_errno(errno, "Failed to mount: %m");
+                if (r < 0)
                         goto child_fail;
-                }
 
                 _exit(EXIT_SUCCESS);
 
@@ -1038,7 +1044,7 @@ int bus_machine_method_bind_mount(sd_bus_message *message, void *userdata, sd_bu
 
 finish:
         if (mount_outside_mounted)
-                (void) umount(mount_outside);
+                (void) umount_verbose(LOG_DEBUG, mount_outside, UMOUNT_NOFOLLOW);
         if (mount_outside_created) {
                 if (S_ISDIR(st.st_mode))
                         (void) rmdir(mount_outside);
@@ -1047,7 +1053,7 @@ finish:
         }
 
         if (mount_tmp_mounted)
-                (void) umount(mount_tmp);
+                (void) umount_verbose(LOG_DEBUG, mount_tmp, UMOUNT_NOFOLLOW);
         if (mount_tmp_created) {
                 if (S_ISDIR(st.st_mode))
                         (void) rmdir(mount_tmp);
@@ -1056,7 +1062,7 @@ finish:
         }
 
         if (mount_slave_mounted)
-                (void) umount(mount_slave);
+                (void) umount_verbose(LOG_DEBUG, mount_slave, UMOUNT_NOFOLLOW);
         if (mount_slave_created)
                 (void) rmdir(mount_slave);
 

src/nspawn/nspawn-cgroup.c

@@ -105,11 +105,11 @@ int sync_cgroup(pid_t pid, CGroupUnified unified_requested, uid_t uid_shift) {
                 return log_error_errno(errno, "Failed to generate temporary mount point for unified hierarchy: %m");
 
         if (unified_controller > 0)
-                r = mount_verbose(LOG_ERR, "cgroup", tree, "cgroup",
+                r = mount_nofollow_verbose(LOG_ERR, "cgroup", tree, "cgroup",
-                                  MS_NOSUID|MS_NOEXEC|MS_NODEV, "none,name=systemd,xattr");
+                                           MS_NOSUID|MS_NOEXEC|MS_NODEV, "none,name=systemd,xattr");
         else
-                r = mount_verbose(LOG_ERR, "cgroup", tree, "cgroup2",
+                r = mount_nofollow_verbose(LOG_ERR, "cgroup", tree, "cgroup2",
-                                  MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
+                                           MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
         if (r < 0)
                 goto finish;
 
@@ -136,7 +136,7 @@ int sync_cgroup(pid_t pid, CGroupUnified unified_requested, uid_t uid_shift) {
                 log_error_errno(r, "Failed to chown() cgroup %s: %m", fn);
 finish:
         if (undo_mount)
-                (void) umount_verbose(tree);
+                (void) umount_verbose(LOG_ERR, tree, UMOUNT_NOFOLLOW);
 
         (void) rmdir(tree);
         return r;
@@ -275,14 +275,14 @@ static int mount_legacy_cgroup_hierarchy(
                 opts = controller;
         }
 
-        r = mount_verbose(LOG_ERR, "cgroup", to, fstype, MS_NOSUID|MS_NOEXEC|MS_NODEV, opts);
+        r = mount_nofollow_verbose(LOG_ERR, "cgroup", to, fstype, MS_NOSUID|MS_NOEXEC|MS_NODEV, opts);
         if (r < 0)
                 return r;
 
         /* ... hence let's only make the bind mount read-only, not the superblock. */
         if (read_only) {
-                r = mount_verbose(LOG_ERR, NULL, to, NULL,
+                r = mount_nofollow_verbose(LOG_ERR, NULL, to, NULL,
-                                  MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL);
+                                           MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL);
                 if (r < 0)
                         return r;
         }
@@ -323,8 +323,8 @@ static int mount_legacy_cgns_supported(
                 if (r < 0)
                         return log_oom();
 
-                r = mount_verbose(LOG_ERR, "tmpfs", cgroup_root, "tmpfs",
+                r = mount_nofollow_verbose(LOG_ERR, "tmpfs", cgroup_root, "tmpfs",
-                                  MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, options);
+                                           MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, options);
                 if (r < 0)
                         return r;
         }
@@ -391,8 +391,8 @@ skip_controllers:
                 return r;
 
         if (!userns)
-                return mount_verbose(LOG_ERR, NULL, cgroup_root, NULL,
+                return mount_nofollow_verbose(LOG_ERR, NULL, cgroup_root, NULL,
-                                     MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755");
+                                              MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755");
 
         return 0;
 }
@@ -425,8 +425,8 @@ static int mount_legacy_cgns_unsupported(
                 if (r < 0)
                         return log_oom();
 
-                r = mount_verbose(LOG_ERR, "tmpfs", cgroup_root, "tmpfs",
+                r = mount_nofollow_verbose(LOG_ERR, "tmpfs", cgroup_root, "tmpfs",
-                                  MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, options);
+                                           MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, options);
                 if (r < 0)
                         return r;
         }
@@ -499,8 +499,8 @@ skip_controllers:
         if (r < 0)
                 return r;
 
-        return mount_verbose(LOG_ERR, NULL, cgroup_root, NULL,
+        return mount_nofollow_verbose(LOG_ERR, NULL, cgroup_root, NULL,
-                             MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755");
+                                      MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME|MS_RDONLY, "mode=755");
 }
 
 static int mount_unified_cgroups(const char *dest) {
@@ -527,7 +527,7 @@ static int mount_unified_cgroups(const char *dest) {
                                        "%s is already mounted but not a unified cgroup hierarchy. Refusing.", p);
         }
 
-        return mount_verbose(LOG_ERR, "cgroup", p, "cgroup2", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
+        return mount_nofollow_verbose(LOG_ERR, "cgroup", p, "cgroup2", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
 }
 
 int mount_cgroups(
@@ -554,13 +554,13 @@ static int mount_systemd_cgroup_writable_one(const char *root, const char *own)
         assert(own);
 
         /* Make our own cgroup a (writable) bind mount */
-        r = mount_verbose(LOG_ERR, own, own, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, own, own, NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
         /* And then remount the systemd cgroup root read-only */
-        return mount_verbose(LOG_ERR, NULL, root, NULL,
+        return mount_nofollow_verbose(LOG_ERR, NULL, root, NULL,
-                             MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL);
+                                      MS_BIND|MS_REMOUNT|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_RDONLY, NULL);
 }
 
 int mount_systemd_cgroup_writable(

src/nspawn/nspawn-mount.c

@@ -40,7 +40,9 @@ CustomMount* custom_mount_add(CustomMount **l, size_t *n, CustomMountType t) {
         ret = *l + *n;
         (*n)++;
 
-        *ret = (CustomMount) { .type = t };
+        *ret = (CustomMount) {
+                .type = t
+        };
 
         return ret;
 }
@@ -442,8 +444,8 @@ int mount_sysfs(const char *dest, MountSettingsMask mount_settings) {
         if (FLAGS_SET(mount_settings, MOUNT_APPLY_APIVFS_RO))
                 extra_flags |= MS_RDONLY;
 
-        r = mount_verbose(LOG_ERR, "sysfs", full, "sysfs",
+        r = mount_nofollow_verbose(LOG_ERR, "sysfs", full, "sysfs",
-                          MS_NOSUID|MS_NOEXEC|MS_NODEV|extra_flags, NULL);
+                                   MS_NOSUID|MS_NOEXEC|MS_NODEV|extra_flags, NULL);
         if (r < 0)
                 return r;
 
@@ -460,17 +462,17 @@ int mount_sysfs(const char *dest, MountSettingsMask mount_settings) {
 
                 (void) mkdir(to, 0755);
 
-                r = mount_verbose(LOG_ERR, from, to, NULL, MS_BIND, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, from, to, NULL, MS_BIND, NULL);
                 if (r < 0)
                         return r;
 
-                r = mount_verbose(LOG_ERR, NULL, to, NULL,
+                r = mount_nofollow_verbose(LOG_ERR, NULL, to, NULL,
-                                  MS_BIND|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT|extra_flags, NULL);
+                                           MS_BIND|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT|extra_flags, NULL);
                 if (r < 0)
                         return r;
         }
 
-        r = umount_verbose(full);
+        r = umount_verbose(LOG_ERR, full, UMOUNT_NOFOLLOW);
         if (r < 0)
                 return r;
 
@@ -483,8 +485,8 @@ int mount_sysfs(const char *dest, MountSettingsMask mount_settings) {
         x = prefix_roota(top, "/fs/cgroup");
         (void) mkdir_p(x, 0755);
 
-        return mount_verbose(LOG_ERR, NULL, top, NULL,
+        return mount_nofollow_verbose(LOG_ERR, NULL, top, NULL,
-                             MS_BIND|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT|extra_flags, NULL);
+                                      MS_BIND|MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_REMOUNT|extra_flags, NULL);
 }
 
 int mount_all(const char *dest,
@@ -516,7 +518,7 @@ int mount_all(const char *dest,
         static const MountPoint mount_table[] = {
                 /* First we list inner child mounts (i.e. mounts applied *after* entering user namespacing) */
                 { "proc",            "/proc",           "proc",  NULL,        MS_NOSUID|MS_NOEXEC|MS_NODEV,
-                  MOUNT_FATAL|MOUNT_IN_USERNS|MOUNT_MKDIR },
+                  MOUNT_FATAL|MOUNT_IN_USERNS|MOUNT_MKDIR|MOUNT_FOLLOW_SYMLINKS }, /* we follow symlinks here since not following them requires /proc/ already being mounted, which we don't have here. */
 
                 { "/proc/sys",       "/proc/sys",       NULL,    NULL,        MS_BIND,
                   MOUNT_FATAL|MOUNT_IN_USERNS|MOUNT_APPLY_APIVFS_RO },                          /* Bind mount first ... */
@@ -670,12 +672,14 @@ int mount_all(const char *dest,
                                 return log_error_errno(r, "Failed to resolve %s/%s: %m", dest, mount_table[k].what);
                 }
 
-                r = mount_verbose(fatal ? LOG_ERR : LOG_DEBUG,
+                r = mount_verbose_full(
-                                  prefixed ?: mount_table[k].what,
+                                fatal ? LOG_ERR : LOG_DEBUG,
-                                  where,
+                                prefixed ?: mount_table[k].what,
-                                  mount_table[k].type,
+                                where,
-                                  mount_table[k].flags,
+                                mount_table[k].type,
-                                  o);
+                                mount_table[k].flags,
+                                o,
+                                FLAGS_SET(mount_table[k].mount_settings, MOUNT_FOLLOW_SYMLINKS));
                 if (r < 0 && fatal)
                         return r;
         }
@@ -771,7 +775,7 @@ static int mount_bind(const char *dest, CustomMount *m) {
                         return log_error_errno(r, "Failed to create mount point %s: %m", where);
         }
 
-        r = mount_verbose(LOG_ERR, m->source, where, NULL, mount_flags, mount_opts);
+        r = mount_nofollow_verbose(LOG_ERR, m->source, where, NULL, mount_flags, mount_opts);
         if (r < 0)
                 return r;
 
@@ -807,7 +811,7 @@ static int mount_tmpfs(const char *dest, CustomMount *m, uid_t uid_shift, const
                 return log_oom();
         options = r > 0 ? buf : m->options;
 
-        return mount_verbose(LOG_ERR, "tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, options);
+        return mount_nofollow_verbose(LOG_ERR, "tmpfs", where, "tmpfs", MS_NODEV|MS_STRICTATIME, options);
 }
 
 static char *joined_and_escaped_lower_dirs(char **lower) {
@@ -864,7 +868,7 @@ static int mount_overlay(const char *dest, CustomMount *m) {
                 options = strjoina("lowerdir=", lower, ",upperdir=", escaped_source, ",workdir=", escaped_work_dir);
         }
 
-        return mount_verbose(LOG_ERR, "overlay", where, "overlay", m->read_only ? MS_RDONLY : 0, options);
+        return mount_nofollow_verbose(LOG_ERR, "overlay", where, "overlay", m->read_only ? MS_RDONLY : 0, options);
 }
 
 static int mount_inaccessible(const char *dest, CustomMount *m) {
@@ -885,13 +889,13 @@ static int mount_inaccessible(const char *dest, CustomMount *m) {
         if (r < 0)
                 return m->graceful ? 0 : r;
 
-        r = mount_verbose(m->graceful ? LOG_DEBUG : LOG_ERR, source, where, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(m->graceful ? LOG_DEBUG : LOG_ERR, source, where, NULL, MS_BIND, NULL);
         if (r < 0)
                 return m->graceful ? 0 : r;
 
-        r = mount_verbose(m->graceful ? LOG_DEBUG : LOG_ERR, NULL, where, NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, NULL);
+        r = mount_nofollow_verbose(m->graceful ? LOG_DEBUG : LOG_ERR, NULL, where, NULL, MS_BIND|MS_RDONLY|MS_REMOUNT, NULL);
         if (r < 0) {
-                (void) umount_verbose(where);
+                (void) umount_verbose(m->graceful ? LOG_DEBUG : LOG_ERR, where, UMOUNT_NOFOLLOW);
                 return m->graceful ? 0 : r;
         }
 
@@ -914,7 +918,7 @@ static int mount_arbitrary(const char *dest, CustomMount *m) {
                         return log_error_errno(r, "Creating mount point for mount %s failed: %m", where);
         }
 
-        return mount_verbose(LOG_ERR, m->source, where, m->type_argument, 0, m->options);
+        return mount_nofollow_verbose(LOG_ERR, m->source, where, m->type_argument, 0, m->options);
 }
 
 int mount_custom(
@@ -1013,7 +1017,7 @@ static int setup_volatile_state(const char *directory, uid_t uid_shift, const ch
         if (r > 0)
                 options = buf;
 
-        return mount_verbose(LOG_ERR, "tmpfs", p, "tmpfs", MS_STRICTATIME, options);
+        return mount_nofollow_verbose(LOG_ERR, "tmpfs", p, "tmpfs", MS_STRICTATIME, options);
 }
 
 static int setup_volatile_yes(const char *directory, uid_t uid_shift, const char *selinux_apifs_context) {
@@ -1058,7 +1062,7 @@ static int setup_volatile_yes(const char *directory, uid_t uid_shift, const char
         if (r > 0)
                 options = buf;
 
-        r = mount_verbose(LOG_ERR, "tmpfs", template, "tmpfs", MS_STRICTATIME, options);
+        r = mount_nofollow_verbose(LOG_ERR, "tmpfs", template, "tmpfs", MS_STRICTATIME, options);
         if (r < 0)
                 goto fail;
 
@@ -1073,7 +1077,7 @@ static int setup_volatile_yes(const char *directory, uid_t uid_shift, const char
                 goto fail;
         }
 
-        r = mount_verbose(LOG_ERR, f, t, NULL, MS_BIND|MS_REC, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, f, t, NULL, MS_BIND|MS_REC, NULL);
         if (r < 0)
                 goto fail;
 
@@ -1085,7 +1089,7 @@ static int setup_volatile_yes(const char *directory, uid_t uid_shift, const char
                 goto fail;
         }
 
-        r = mount_verbose(LOG_ERR, template, directory, NULL, MS_MOVE, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, template, directory, NULL, MS_MOVE, NULL);
         if (r < 0)
                 goto fail;
 
@@ -1095,10 +1099,11 @@ static int setup_volatile_yes(const char *directory, uid_t uid_shift, const char
 
 fail:
         if (bind_mounted)
-                (void) umount_verbose(t);
+                (void) umount_verbose(LOG_ERR, t, UMOUNT_NOFOLLOW);
 
         if (tmpfs_mounted)
-                (void) umount_verbose(template);
+                (void) umount_verbose(LOG_ERR, template, UMOUNT_NOFOLLOW);
+
         (void) rmdir(template);
         return r;
 }
@@ -1125,7 +1130,7 @@ static int setup_volatile_overlay(const char *directory, uid_t uid_shift, const
         if (r > 0)
                 options = buf;
 
-        r = mount_verbose(LOG_ERR, "tmpfs", template, "tmpfs", MS_STRICTATIME, options);
+        r = mount_nofollow_verbose(LOG_ERR, "tmpfs", template, "tmpfs", MS_STRICTATIME, options);
         if (r < 0)
                 goto finish;
 
@@ -1155,11 +1160,11 @@ static int setup_volatile_overlay(const char *directory, uid_t uid_shift, const
         }
 
         options = strjoina("lowerdir=", escaped_directory, ",upperdir=", escaped_upper, ",workdir=", escaped_work);
-        r = mount_verbose(LOG_ERR, "overlay", directory, "overlay", 0, options);
+        r = mount_nofollow_verbose(LOG_ERR, "overlay", directory, "overlay", 0, options);
 
 finish:
         if (tmpfs_mounted)
-                (void) umount_verbose(template);
+                (void) umount_verbose(LOG_ERR, template, UMOUNT_NOFOLLOW);
 
         (void) rmdir(template);
         return r;
@@ -1265,7 +1270,7 @@ int setup_pivot_root(const char *directory, const char *pivot_root_new, const ch
                 return log_oom();
 
         /* Remount directory_pivot_root_new to make it movable. */
-        r = mount_verbose(LOG_ERR, directory_pivot_root_new, directory_pivot_root_new, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, directory_pivot_root_new, directory_pivot_root_new, NULL, MS_BIND, NULL);
         if (r < 0)
                 goto done;
 
@@ -1282,19 +1287,19 @@ int setup_pivot_root(const char *directory, const char *pivot_root_new, const ch
                         goto done;
                 }
 
-                r = mount_verbose(LOG_ERR, directory_pivot_root_new, pivot_tmp, NULL, MS_MOVE, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, directory_pivot_root_new, pivot_tmp, NULL, MS_MOVE, NULL);
                 if (r < 0)
                         goto done;
 
-                r = mount_verbose(LOG_ERR, directory, pivot_tmp_pivot_root_old, NULL, MS_MOVE, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, directory, pivot_tmp_pivot_root_old, NULL, MS_MOVE, NULL);
                 if (r < 0)
                         goto done;
 
-                r = mount_verbose(LOG_ERR, pivot_tmp, directory, NULL, MS_MOVE, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, pivot_tmp, directory, NULL, MS_MOVE, NULL);
                 if (r < 0)
                         goto done;
         } else {
-                r = mount_verbose(LOG_ERR, directory_pivot_root_new, directory, NULL, MS_MOVE, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, directory_pivot_root_new, directory, NULL, MS_MOVE, NULL);
                 if (r < 0)
                         goto done;
         }

src/nspawn/nspawn-mount.h

@@ -19,6 +19,7 @@ typedef enum MountSettingsMask {
         MOUNT_MKDIR              = 1 << 8, /* if set, make directory to mount over first */
         MOUNT_TOUCH              = 1 << 9, /* if set, touch file to mount over first */
         MOUNT_PREFIX_ROOT        = 1 << 10,/* if set, prefix the source path with the container's root directory */
+        MOUNT_FOLLOW_SYMLINKS    = 1 << 11,/* if set, we'll follow symlinks for the mount target */
 } MountSettingsMask;
 
 typedef enum CustomMountType {

src/nspawn/nspawn.c

@@ -1918,9 +1918,9 @@ static int setup_timezone(const char *dest) {
                 if (found == 0) /* missing? */
                         (void) touch(resolved);
 
-                r = mount_verbose(LOG_WARNING, "/etc/localtime", resolved, NULL, MS_BIND, NULL);
+                r = mount_nofollow_verbose(LOG_WARNING, "/etc/localtime", resolved, NULL, MS_BIND, NULL);
                 if (r >= 0)
-                        return mount_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
+                        return mount_nofollow_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
 
                 _fallthrough_;
         }
@@ -2053,9 +2053,9 @@ static int setup_resolv_conf(const char *dest) {
                 if (found == 0) /* missing? */
                         (void) touch(resolved);
 
-                r = mount_verbose(LOG_WARNING, what, resolved, NULL, MS_BIND, NULL);
+                r = mount_nofollow_verbose(LOG_WARNING, what, resolved, NULL, MS_BIND, NULL);
                 if (r >= 0)
-                        return mount_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
+                        return mount_nofollow_verbose(LOG_ERR, NULL, resolved, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NODEV, NULL);
 
                 /* If that didn't work, let's copy the file */
         }
@@ -2107,11 +2107,11 @@ static int setup_boot_id(void) {
         from = TAKE_PTR(path);
         to = "/proc/sys/kernel/random/boot_id";
 
-        r = mount_verbose(LOG_ERR, from, to, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, from, to, NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
-        return mount_verbose(LOG_ERR, NULL, to, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
+        return mount_nofollow_verbose(LOG_ERR, NULL, to, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
 }
 
 static int copy_devnodes(const char *dest) {
@@ -2170,7 +2170,7 @@ static int copy_devnodes(const char *dest) {
                                 r = touch(to);
                                 if (r < 0)
                                         return log_error_errno(r, "touch (%s) failed: %m", to);
-                                r = mount_verbose(LOG_DEBUG, from, to, NULL, MS_BIND, NULL);
+                                r = mount_nofollow_verbose(LOG_DEBUG, from, to, NULL, MS_BIND, NULL);
                                 if (r < 0)
                                         return log_error_errno(r, "Both mknod and bind mount (%s) failed: %m", to);
                         }
@@ -2258,7 +2258,7 @@ static int setup_pts(const char *dest) {
         if (r < 0)
                 return log_error_errno(r, "Failed to create /dev/pts: %m");
 
-        r = mount_verbose(LOG_ERR, "devpts", p, "devpts", MS_NOSUID|MS_NOEXEC, options);
+        r = mount_nofollow_verbose(LOG_ERR, "devpts", p, "devpts", MS_NOSUID|MS_NOEXEC, options);
         if (r < 0)
                 return r;
         r = userns_lchown(p, 0, 0);
@@ -2360,7 +2360,7 @@ static int setup_credentials(const char *root) {
                 return log_error_errno(r, "Failed to create /run/host/credentials: %m");
 
         q = prefix_roota(root, "/run/host/credentials");
-        r = mount_verbose(LOG_ERR, NULL, q, "ramfs", MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0700");
+        r = mount_nofollow_verbose(LOG_ERR, NULL, q, "ramfs", MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0700");
         if (r < 0)
                 return r;
 
@@ -2397,11 +2397,11 @@ static int setup_credentials(const char *root) {
                 return r;
 
         /* Make both mount and superblock read-only now */
-        r = mount_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_BIND|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_BIND|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL);
         if (r < 0)
                 return r;
 
-        return mount_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0500");
+        return mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_REMOUNT|MS_RDONLY|MS_NOSUID|MS_NOEXEC|MS_NODEV, "mode=0500");
 }
 
 static int setup_kmsg(int kmsg_socket) {
@@ -2429,7 +2429,7 @@ static int setup_kmsg(int kmsg_socket) {
 
         from = TAKE_PTR(fifo);
 
-        r = mount_verbose(LOG_ERR, from, "/proc/kmsg", NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, from, "/proc/kmsg", NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
@@ -2595,7 +2595,7 @@ static int setup_journal(const char *directory) {
         if (r < 0)
                 return log_error_errno(r, "Failed to create %s: %m", q);
 
-        r = mount_verbose(LOG_DEBUG, p, q, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_DEBUG, p, q, NULL, MS_BIND, NULL);
         if (r < 0)
                 return log_error_errno(errno, "Failed to bind mount journal from host into guest: %m");
 
@@ -2700,16 +2700,16 @@ static int setup_propagate(const char *root) {
                 return log_error_errno(r, "Failed to create /run/host/incoming: %m");
 
         q = prefix_roota(root, "/run/host/incoming");
-        r = mount_verbose(LOG_ERR, p, q, NULL, MS_BIND, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, p, q, NULL, MS_BIND, NULL);
         if (r < 0)
                 return r;
 
-        r = mount_verbose(LOG_ERR, NULL, q, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
         if (r < 0)
                 return r;
 
         /* machined will MS_MOVE into that directory, and that's only supported for non-shared mounts. */
-        return mount_verbose(LOG_ERR, NULL, q, NULL, MS_SLAVE, NULL);
+        return mount_nofollow_verbose(LOG_ERR, NULL, q, NULL, MS_SLAVE, NULL);
 }
 
 static int setup_machine_id(const char *directory) {
@@ -3157,7 +3157,7 @@ static int inner_child(
                 /* Creating a new user namespace means all MS_SHARED mounts become MS_SLAVE. Let's put them
                  * back to MS_SHARED here, since that's what we want as defaults. (This will not reconnect
                  * propagation, but simply create new peer groups for all our mounts). */
-                r = mount_verbose(LOG_ERR, NULL, "/", NULL, MS_SHARED|MS_REC, NULL);
+                r = mount_follow_verbose(LOG_ERR, NULL, "/", NULL, MS_SHARED|MS_REC, NULL);
                 if (r < 0)
                         return r;
         }
@@ -3542,7 +3542,7 @@ static int outer_child(
 
         /* Mark everything as slave, so that we still receive mounts from the real root, but don't propagate
          * mounts to the real root. */
-        r = mount_verbose(LOG_ERR, NULL, "/", NULL, MS_SLAVE|MS_REC, NULL);
+        r = mount_follow_verbose(LOG_ERR, NULL, "/", NULL, MS_SLAVE|MS_REC, NULL);
         if (r < 0)
                 return r;
 
@@ -3600,7 +3600,7 @@ static int outer_child(
                  * already, and thus don't need to be afraid of colliding with anyone else's mounts).*/
                 (void) mkdir_p("/run/systemd/nspawn-root", 0755);
 
-                r = mount_verbose(LOG_ERR, "/", "/run/systemd/nspawn-root", NULL, MS_BIND|MS_REC, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, "/", "/run/systemd/nspawn-root", NULL, MS_BIND|MS_REC, NULL);
                 if (r < 0)
                         return r;
 
@@ -3634,7 +3634,7 @@ static int outer_child(
 
         /* Make sure we always have a mount that we can move to root later on. */
         if (!path_is_mount_point(directory, NULL, 0)) {
-                r = mount_verbose(LOG_ERR, directory, directory, NULL, MS_BIND|MS_REC, NULL);
+                r = mount_nofollow_verbose(LOG_ERR, directory, directory, NULL, MS_BIND|MS_REC, NULL);
                 if (r < 0)
                         return r;
         }
@@ -3677,7 +3677,7 @@ static int outer_child(
          * enable moving the root directory mount to root later on.
          * https://github.com/systemd/systemd/issues/3847#issuecomment-562735251
          */
-        r = mount_verbose(LOG_ERR, NULL, directory, NULL, MS_SHARED|MS_REC, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, NULL, directory, NULL, MS_SHARED|MS_REC, NULL);
         if (r < 0)
                 return r;
 

src/partition/repart.c

@@ -2691,7 +2691,7 @@ static int partition_copy_files(Partition *p, const char *node) {
                         _exit(EXIT_FAILURE);
                 }
 
-                if (mount_verbose(LOG_ERR, node, fs, p->format, MS_NOATIME|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL) < 0)
+                if (mount_nofollow_verbose(LOG_ERR, node, fs, p->format, MS_NOATIME|MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL) < 0)
                         _exit(EXIT_FAILURE);
 
                 if (do_copy_files(p, fs) < 0)

src/shared/dissect-image.c

@@ -1177,7 +1177,7 @@ static int mount_partition(
                         return r;
         }
 
-        r = mount_verbose(LOG_DEBUG, node, p, fstype, MS_NODEV|(rw ? 0 : MS_RDONLY), options);
+        r = mount_nofollow_verbose(LOG_DEBUG, node, p, fstype, MS_NODEV|(rw ? 0 : MS_RDONLY), options);
         if (r < 0)
                 return r;
 

src/shared/mount-util.c

@@ -19,10 +19,60 @@
 #include "parse-util.h"
 #include "path-util.h"
 #include "set.h"
+#include "stat-util.h"
 #include "stdio-util.h"
 #include "string-util.h"
 #include "strv.h"
 
+int mount_fd(const char *source,
+             int target_fd,
+             const char *filesystemtype,
+             unsigned long mountflags,
+             const void *data) {
+
+        char path[STRLEN("/proc/self/fd/") + DECIMAL_STR_MAX(int)];
+
+        xsprintf(path, "/proc/self/fd/%i", target_fd);
+        if (mount(source, path, filesystemtype, mountflags, data) < 0) {
+                if (errno != ENOENT)
+                        return -errno;
+
+                /* ENOENT can mean two things: either that the source is missing, or that /proc/ isn't
+                 * mounted. Check for the latter to generate better error messages. */
+                if (proc_mounted() == 0)
+                        return -ENOSYS;
+
+                return -ENOENT;
+        }
+
+        return 0;
+}
+
+int mount_nofollow(
+                const char *source,
+                const char *target,
+                const char *filesystemtype,
+                unsigned long mountflags,
+                const void *data) {
+
+        _cleanup_close_ int fd = -1;
+
+        /* In almost all cases we want to manipulate the mount table without following symlinks, hence
+         * mount_nofollow() is usually the way to go. The only exceptions are environments where /proc/ is
+         * not available yet, since we need /proc/self/fd/ for this logic to work. i.e. during the early
+         * initialization of namespacing/container stuff where /proc is not yet mounted (and maybe even the
+         * fs to mount) we can only use traditional mount() directly.
+         *
+         * Note that this disables following only for the final component of the target, i.e symlinks within
+         * the path of the target are honoured, as are symlinks in the source path everywhere. */
+
+        fd = open(target, O_PATH|O_CLOEXEC|O_NOFOLLOW);
+        if (fd < 0)
+                return -errno;
+
+        return mount_fd(source, fd, filesystemtype, mountflags, data);
+}
+
 int umount_recursive(const char *prefix, int flags) {
         int n = 0, r;
         bool again;
@@ -79,14 +129,18 @@ static int get_mount_flags(
                 struct libmnt_table *table,
                 const char *path,
                 unsigned long *ret) {
+
+        _cleanup_close_ int fd = -1;
         struct libmnt_fs *fs;
         struct statvfs buf;
         const char *opts;
-        int r = 0;
+        int r;
 
         /* Get the mount flags for the mountpoint at "path" from "table". We have a fallback using statvfs()
          * in place (which provides us with mostly the same info), but it's just a fallback, since using it
-         * means triggering autofs or NFS mounts, which we'd rather avoid needlessly. */
+         * means triggering autofs or NFS mounts, which we'd rather avoid needlessly.
+         *
+         * This generally doesn't follow symlinks. */
 
         fs = mnt_table_find_target(table, path, MNT_ITER_FORWARD);
         if (!fs) {
@@ -111,7 +165,11 @@ static int get_mount_flags(
         return 0;
 
 fallback:
-        if (statvfs(path, &buf) < 0)
+        fd = open(path, O_PATH|O_CLOEXEC|O_NOFOLLOW);
+        if (fd < 0)
+                return -errno;
+
+        if (fstatvfs(fd, &buf) < 0)
                 return -errno;
 
         /* The statvfs() flags and the mount flags mostly have the same values, but for some cases do
@@ -254,14 +312,16 @@ int bind_remount_recursive_with_mountinfo(
                 if (!set_contains(done, simplified) &&
                     !set_contains(todo, simplified)) {
                         /* The prefix directory itself is not yet a mount, make it one. */
-                        if (mount(simplified, simplified, NULL, MS_BIND|MS_REC, NULL) < 0)
+                        r = mount_nofollow(simplified, simplified, NULL, MS_BIND|MS_REC, NULL);
-                                return -errno;
+                        if (r < 0)
+                                return r;
 
                         orig_flags = 0;
                         (void) get_mount_flags(table, simplified, &orig_flags);
 
-                        if (mount(NULL, simplified, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL) < 0)
+                        r = mount_nofollow(NULL, simplified, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL);
-                                return -errno;
+                        if (r < 0)
+                                return r;
 
                         log_debug("Made top-level directory %s a mount point.", prefix);
 
@@ -302,8 +362,9 @@ int bind_remount_recursive_with_mountinfo(
                         orig_flags = 0;
                         (void) get_mount_flags(table, x, &orig_flags);
 
-                        if (mount(NULL, x, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL) < 0)
+                        r = mount_nofollow(NULL, x, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL);
-                                return -errno;
+                        if (r < 0)
+                                return r;
 
                         log_debug("Remounted %s read-only.", x);
                 }
@@ -352,8 +413,9 @@ int bind_remount_one_with_mountinfo(
         /* Try to reuse the original flag set */
         (void) get_mount_flags(table, path, &orig_flags);
 
-        if (mount(NULL, path, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL) < 0)
+        r = mount_nofollow(NULL, path, NULL, (orig_flags & ~flags_mask)|MS_BIND|MS_REMOUNT|new_flags, NULL);
-                return -errno;
+        if (r < 0)
+                return r;
 
         return 0;
 }
@@ -548,13 +610,14 @@ static char* mount_flags_to_string(long unsigned flags) {
         return x;
 }
 
-int mount_verbose(
+int mount_verbose_full(
                 int error_log_level,
                 const char *what,
                 const char *where,
                 const char *type,
                 unsigned long flags,
-                const char *options) {
+                const char *options,
+                bool follow_symlink) {
 
         _cleanup_free_ char *fl = NULL, *o = NULL;
         unsigned long f;
@@ -583,17 +646,31 @@ int mount_verbose(
         else
                 log_debug("Mounting %s on %s (%s \"%s\")...",
                           strna(type), where, strnull(fl), strempty(o));
-        if (mount(what, where, type, f, o) < 0)
+
-                return log_full_errno(error_log_level, errno,
+        if (follow_symlink)
+                r = mount(what, where, type, f, o) < 0 ? -errno : 0;
+        else
+                r = mount_nofollow(what, where, type, f, o);
+        if (r < 0)
+                return log_full_errno(error_log_level, r,
                                       "Failed to mount %s (type %s) on %s (%s \"%s\"): %m",
                                       strna(what), strna(type), where, strnull(fl), strempty(o));
         return 0;
 }
 
-int umount_verbose(const char *what) {
+int umount_verbose(
+                int error_log_level,
+                const char *what,
+                int flags) {
+
+        assert(what);
+
         log_debug("Umounting %s...", what);
-        if (umount(what) < 0)
+
-                return log_error_errno(errno, "Failed to unmount %s: %m", what);
+        if (umount2(what, flags) < 0)
+                return log_full_errno(error_log_level, errno,
+                                      "Failed to unmount %s: %m", what);
+
         return 0;
 }
 

src/shared/mount-util.h

@@ -32,6 +32,9 @@
 #define TMPFS_LIMITS_ROOTFS          TMPFS_LIMITS_VAR
 #define TMPFS_LIMITS_VOLATILE_STATE  TMPFS_LIMITS_VAR
 
+int mount_fd(const char *source, int target_fd, const char *filesystemtype, unsigned long mountflags, const void *data);
+int mount_nofollow(const char *source, const char *target, const char *filesystemtype, unsigned long mountflags, const void *data);
+
 int repeat_unmount(const char *path, int flags);
 int umount_recursive(const char *target, int flags);
 int bind_remount_recursive(const char *prefix, unsigned long new_flags, unsigned long flags_mask, char **deny_list);
@@ -43,14 +46,39 @@ int mount_move_root(const char *path);
 DEFINE_TRIVIAL_CLEANUP_FUNC(FILE*, endmntent);
 #define _cleanup_endmntent_ _cleanup_(endmntentp)
 
-int mount_verbose(
+int mount_verbose_full(
                 int error_log_level,
                 const char *what,
                 const char *where,
                 const char *type,
                 unsigned long flags,
-                const char *options);
+                const char *options,
-int umount_verbose(const char *where);
+                bool follow_symlink);
+
+static inline int mount_follow_verbose(
+                int error_log_level,
+                const char *what,
+                const char *where,
+                const char *type,
+                unsigned long flags,
+                const char *options) {
+        return mount_verbose_full(error_log_level, what, where, type, flags, options, true);
+}
+
+static inline int mount_nofollow_verbose(
+                int error_log_level,
+                const char *what,
+                const char *where,
+                const char *type,
+                unsigned long flags,
+                const char *options) {
+        return mount_verbose_full(error_log_level, what, where, type, flags, options, false);
+}
+
+int umount_verbose(
+                int error_log_level,
+                const char *where,
+                int flags);
 
 int mount_option_mangle(
                 const char *options,

src/shutdown/umount.c

@@ -23,6 +23,7 @@
 #include "device-util.h"
 #include "escape.h"
 #include "fd-util.h"
+#include "fs-util.h"
 #include "fstab-util.h"
 #include "libmount-util.h"
 #include "mount-setup.h"
@@ -387,8 +388,15 @@ static int delete_loopback(const char *device) {
         assert(device);
 
         fd = open(device, O_RDONLY|O_CLOEXEC);
-        if (fd < 0)
+        if (fd < 0) {
+                log_debug_errno(errno, "Failed to open loopback device %s: %m", device);
                 return errno == ENOENT ? 0 : -errno;
+        }
+
+        /* Loopback block devices don't sync in-flight blocks when we clear the fd, hence sync explicitly
+         * first */
+        if (fsync(fd) < 0)
+                log_debug_errno(errno, "Failed to sync loop block device %s, ignoring: %m", device);
 
         if (ioctl(fd, LOOP_CLR_FD, 0) < 0) {
                 if (errno == ENXIO) /* Nothing bound, didn't do anything */
@@ -440,43 +448,50 @@ static int delete_loopback(const char *device) {
         return -EBUSY; /* Nothing changed, the device is still attached, hence it apparently is still busy */
 }
 
-static int delete_dm(dev_t devnum) {
+static int delete_dm(MountPoint *m) {
-
-        struct dm_ioctl dm = {
-                .version = {
-                        DM_VERSION_MAJOR,
-                        DM_VERSION_MINOR,
-                        DM_VERSION_PATCHLEVEL
-                },
-                .data_size = sizeof(dm),
-                .dev = devnum,
-        };
-
         _cleanup_close_ int fd = -1;
+        int r;
 
-        assert(major(devnum) != 0);
+        assert(m);
+        assert(major(m->devnum) != 0);
+        assert(m->path);
 
         fd = open("/dev/mapper/control", O_RDWR|O_CLOEXEC);
         if (fd < 0)
                 return -errno;
 
-        if (ioctl(fd, DM_DEV_REMOVE, &dm) < 0)
+        r = fsync_path_at(AT_FDCWD, m->path);
+        if (r < 0)
+                log_debug_errno(r, "Failed to sync DM block device %s, ignoring: %m", m->path);
+
+        if (ioctl(fd, DM_DEV_REMOVE, &(struct dm_ioctl) {
+                .version = {
+                        DM_VERSION_MAJOR,
+                        DM_VERSION_MINOR,
+                        DM_VERSION_PATCHLEVEL
+                },
+                .data_size = sizeof(struct dm_ioctl),
+                .dev = m->devnum,
+        }) < 0)
                 return -errno;
 
         return 0;
 }
 
 static int delete_md(MountPoint *m) {
-
         _cleanup_close_ int fd = -1;
 
+        assert(m);
         assert(major(m->devnum) != 0);
-        assert(m->path != 0);
+        assert(m->path);
 
         fd = open(m->path, O_RDONLY|O_CLOEXEC|O_EXCL);
         if (fd < 0)
                 return -errno;
 
+        if (fsync(fd) < 0)
+                log_debug_errno(errno, "Failed to sync MD block device %s, ignoring: %m", m->path);
+
         if (ioctl(fd, STOP_ARRAY, NULL) < 0)
                 return -errno;
 
@@ -692,7 +707,7 @@ static int dm_points_list_detach(MountPoint **head, bool *changed, int umount_lo
                 }
 
                 log_info("Detaching DM %s (%u:%u).", m->path, major(m->devnum), minor(m->devnum));
-                r = delete_dm(m->devnum);
+                r = delete_dm(m);
                 if (r < 0) {
                         log_full_errno(umount_log_level, r, "Could not detach DM %s: %m", m->path);
                         n_failed++;

src/test/test-udev.c

@@ -45,8 +45,8 @@ static int fake_filesystems(void) {
                 return log_error_errno(r, "Failed to detach mount namespace: %m");
 
         for (size_t i = 0; i < ELEMENTSOF(fakefss); i++) {
-                r = mount_verbose(fakefss[i].ignore_mount_error ? LOG_NOTICE : LOG_ERR,
+                r = mount_nofollow_verbose(fakefss[i].ignore_mount_error ? LOG_NOTICE : LOG_ERR,
-                                  fakefss[i].src, fakefss[i].target, NULL, MS_BIND, NULL);
+                                           fakefss[i].src, fakefss[i].target, NULL, MS_BIND, NULL);
                 if (r < 0 && !fakefss[i].ignore_mount_error)
                         return r;
         }

src/volatile-root/volatile-root.c

@@ -29,7 +29,7 @@ static int make_volatile(const char *path) {
         if (r < 0)
                 return log_error_errno(r, "Couldn't generate volatile sysroot directory: %m");
 
-        r = mount_verbose(LOG_ERR, "tmpfs", "/run/systemd/volatile-sysroot", "tmpfs", MS_STRICTATIME, "mode=755" TMPFS_LIMITS_ROOTFS);
+        r = mount_nofollow_verbose(LOG_ERR, "tmpfs", "/run/systemd/volatile-sysroot", "tmpfs", MS_STRICTATIME, "mode=755" TMPFS_LIMITS_ROOTFS);
         if (r < 0)
                 goto finish_rmdir;
 
@@ -38,7 +38,7 @@ static int make_volatile(const char *path) {
                 goto finish_umount;
         }
 
-        r = mount_verbose(LOG_ERR, old_usr, "/run/systemd/volatile-sysroot/usr", NULL, MS_BIND|MS_REC, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, old_usr, "/run/systemd/volatile-sysroot/usr", NULL, MS_BIND|MS_REC, NULL);
         if (r < 0)
                 goto finish_umount;
 
@@ -57,7 +57,7 @@ static int make_volatile(const char *path) {
         if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0)
                 log_warning_errno(errno, "Failed to remount %s MS_SLAVE|MS_REC, ignoring: %m", path);
 
-        r = mount_verbose(LOG_ERR, "/run/systemd/volatile-sysroot", path, NULL, MS_MOVE, NULL);
+        r = mount_nofollow_verbose(LOG_ERR, "/run/systemd/volatile-sysroot", path, NULL, MS_MOVE, NULL);
 
 finish_umount:
         (void) umount_recursive("/run/systemd/volatile-sysroot", 0);
@@ -80,7 +80,7 @@ static int make_overlay(const char *path) {
         if (r < 0)
                 return log_error_errno(r, "Couldn't create overlay sysroot directory: %m");
 
-        r = mount_verbose(LOG_ERR, "tmpfs", "/run/systemd/overlay-sysroot", "tmpfs", MS_STRICTATIME, "mode=755" TMPFS_LIMITS_ROOTFS);
+        r = mount_nofollow_verbose(LOG_ERR, "tmpfs", "/run/systemd/overlay-sysroot", "tmpfs", MS_STRICTATIME, "mode=755" TMPFS_LIMITS_ROOTFS);
         if (r < 0)
                 goto finish;
 
@@ -103,11 +103,11 @@ static int make_overlay(const char *path) {
         }
 
         options = strjoina("lowerdir=", escaped_path, ",upperdir=/run/systemd/overlay-sysroot/upper,workdir=/run/systemd/overlay-sysroot/work");
-        r = mount_verbose(LOG_ERR, "overlay", path, "overlay", 0, options);
+        r = mount_nofollow_verbose(LOG_ERR, "overlay", path, "overlay", 0, options);
 
 finish:
         if (tmpfs_mounted)
-                (void) umount_verbose("/run/systemd/overlay-sysroot");
+                (void) umount_verbose(LOG_ERR, "/run/systemd/overlay-sysroot", UMOUNT_NOFOLLOW);
 
         (void) rmdir("/run/systemd/overlay-sysroot");
         return r;