X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fcore%2Fmount-setup.c;h=343ff066a97d987622266ec72d9a4849e5964324;hp=d902211ead787d179435f3e903e32ee427edfddd;hb=5bd4b173605142c7be493aa4d958ebaef21f421d;hpb=71f737d2de635d4ac6183face7e9c8b7981631ab diff --git a/src/core/mount-setup.c b/src/core/mount-setup.c index d902211ea..343ff066a 100644 --- a/src/core/mount-setup.c +++ b/src/core/mount-setup.c @@ -21,11 +21,7 @@ #include #include -#include #include -#include -#include -#include #include #include @@ -38,10 +34,18 @@ #include "set.h" #include "strv.h" #include "mkdir.h" - -#ifndef TTY_GID -#define TTY_GID 5 -#endif +#include "path-util.h" +#include "missing.h" +#include "virt.h" +#include "efivars.h" +#include "smack-util.h" +#include "cgroup-util.h" + +typedef enum MountMode { + MNT_NONE = 0, + MNT_FATAL = 1 << 0, + MNT_IN_CONTAINER = 1 << 1, +} MountMode; typedef struct MountPoint { const char *what; @@ -49,34 +53,73 @@ typedef struct MountPoint { const char *type; const char *options; unsigned long flags; - bool fatal; + bool (*condition_fn)(void); + MountMode mode; } MountPoint; /* The first three entries we might need before SELinux is up. The * fourth (securityfs) is needed by IMA to load a custom policy. The - * other ones we can delay until SELinux and IMA are loaded. */ + * other ones we can delay until SELinux and IMA are loaded. When + * SMACK is enabled we need smackfs, too, so it's a fifth one. */ +#ifdef HAVE_SMACK +#define N_EARLY_MOUNT 5 +#else #define N_EARLY_MOUNT 4 +#endif static const MountPoint mount_table[] = { - { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true }, - { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, true }, - { "devtmpfs", "/dev", "devtmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME, true }, - { "securityfs", "/sys/kernel/security", "securityfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, false }, - { "tmpfs", "/dev/shm", "tmpfs", "mode=1777", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true }, - { "devpts", "/dev/pts", "devpts", "mode=620,gid=" STRINGIFY(TTY_GID), MS_NOSUID|MS_NOEXEC, false }, - { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME, true }, - { "tmpfs", "/sys/fs/cgroup", "tmpfs", "mode=755", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, false }, - { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd", MS_NOSUID|MS_NOEXEC|MS_NODEV, false }, + { "sysfs", "/sys", "sysfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "proc", "/proc", "proc", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "devtmpfs", "/dev", "devtmpfs", "mode=755", MS_NOSUID|MS_STRICTATIME, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "securityfs", "/sys/kernel/security", "securityfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_NONE }, +#ifdef HAVE_SMACK + { "smackfs", "/sys/fs/smackfs", "smackfs", "smackfsdef=*", MS_NOSUID|MS_NOEXEC|MS_NODEV, + mac_smack_use, MNT_FATAL }, + { "tmpfs", "/dev/shm", "tmpfs", "mode=1777,smackfsroot=*", MS_NOSUID|MS_NODEV|MS_STRICTATIME, + mac_smack_use, MNT_FATAL }, +#endif + { "tmpfs", "/dev/shm", "tmpfs", "mode=1777", MS_NOSUID|MS_NODEV|MS_STRICTATIME, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "devpts", "/dev/pts", "devpts", "mode=620,gid=" STRINGIFY(TTY_GID), MS_NOSUID|MS_NOEXEC, + NULL, MNT_IN_CONTAINER }, +#ifdef HAVE_SMACK + { "tmpfs", "/run", "tmpfs", "mode=755,smackfsroot=*", MS_NOSUID|MS_NODEV|MS_STRICTATIME, + mac_smack_use, MNT_FATAL }, +#endif + { "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "tmpfs", "/sys/fs/cgroup", "tmpfs", "mode=755", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd,xattr", MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_IN_CONTAINER }, + { "cgroup", "/sys/fs/cgroup/systemd", "cgroup", "none,name=systemd", MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_FATAL|MNT_IN_CONTAINER }, + { "pstore", "/sys/fs/pstore", "pstore", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_NONE }, +#ifdef ENABLE_EFI + { "efivarfs", "/sys/firmware/efi/efivars", "efivarfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + is_efi_boot, MNT_NONE }, +#endif +#ifdef ENABLE_KDBUS + { "kdbusfs", "/sys/fs/kdbus", "kdbusfs", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV, + NULL, MNT_IN_CONTAINER }, +#endif }; /* These are API file systems that might be mounted by other software, * we just list them here so that we know that we should ignore them */ -static const char * const ignore_paths[] = { - "/sys/fs/selinux", - "/selinux", - "/proc/bus/usb" -}; +static const char ignore_paths[] = + /* SELinux file systems */ + "/sys/fs/selinux\0" + /* Container bind mounts */ + "/proc/sys\0" + "/dev/console\0" + "/proc/kmsg\0"; bool mount_point_is_api(const char *path) { unsigned i; @@ -92,10 +135,10 @@ bool mount_point_is_api(const char *path) { } bool mount_point_ignore(const char *path) { - unsigned i; + const char *i; - for (i = 0; i < ELEMENTSOF(ignore_paths); i++) - if (path_equal(path, ignore_paths[i])) + NULSTR_FOREACH(i, ignore_paths) + if (path_equal(path, i)) return true; return false; @@ -106,19 +149,30 @@ static int mount_one(const MountPoint *p, bool relabel) { assert(p); + if (p->condition_fn && !p->condition_fn()) + return 0; + /* Relabel first, just in case */ if (relabel) - label_fix(p->where, true); + label_fix(p->where, true, true); - if ((r = path_is_mount_point(p->where, true)) < 0) + r = path_is_mount_point(p->where, true); + if (r < 0) return r; if (r > 0) return 0; + /* Skip securityfs in a container */ + if (!(p->mode & MNT_IN_CONTAINER) && detect_container(NULL) > 0) + return 0; + /* The access mode here doesn't really matter too much, since * the mounted file system will take precedence anyway. */ - mkdir_p(p->where, 0755); + if (relabel) + mkdir_p_label(p->where, 0755); + else + mkdir_p(p->where, 0755); log_debug("Mounting %s to %s of type %s with options %s.", p->what, @@ -131,13 +185,13 @@ static int mount_one(const MountPoint *p, bool relabel) { p->type, p->flags, p->options) < 0) { - log_full(p->fatal ? LOG_ERR : LOG_DEBUG, "Failed to mount %s: %s", p->where, strerror(errno)); - return p->fatal ? -errno : 0; + log_full((p->mode & MNT_FATAL) ? LOG_ERR : LOG_DEBUG, "Failed to mount %s at %s: %m", p->type, p->where); + return (p->mode & MNT_FATAL) ? -errno : 0; } /* Relabel again, since we now mounted something fresh here */ if (relabel) - label_fix(p->where, false); + label_fix(p->where, false, false); return 1; } @@ -162,59 +216,27 @@ int mount_setup_early(void) { } int mount_cgroup_controllers(char ***join_controllers) { + _cleanup_set_free_free_ Set *controllers = NULL; int r; - FILE *f; - char buf[LINE_MAX]; - Set *controllers; /* Mount all available cgroup controllers that are built into the kernel. */ - f = fopen("/proc/cgroups", "re"); - if (!f) { - log_error("Failed to enumerate cgroup controllers: %m"); - return 0; - } - - controllers = set_new(string_hash_func, string_compare_func); - if (!controllers) { - r = -ENOMEM; - log_error("Failed to allocate controller set."); - goto finish; - } + controllers = set_new(&string_hash_ops); + if (!controllers) + return log_oom(); - /* Ignore the header line */ - (void) fgets(buf, sizeof(buf), f); + r = cg_kernel_controllers(controllers); + if (r < 0) + return log_error_errno(r, "Failed to enumerate cgroup controllers: %m"); for (;;) { - char *controller; - int enabled = 0; - - if (fscanf(f, "%ms %*i %*i %i", &controller, &enabled) != 2) { - - if (feof(f)) - break; - - log_error("Failed to parse /proc/cgroups."); - r = -EIO; - goto finish; - } - - if (!enabled) { - free(controller); - continue; - } - - r = set_put(controllers, controller); - if (r < 0) { - log_error("Failed to add controller to set."); - free(controller); - goto finish; - } - } - - for (;;) { - MountPoint p; - char *controller, *where, *options; + _cleanup_free_ char *options = NULL, *controller = NULL, *where = NULL; + MountPoint p = { + .what = "cgroup", + .type = "cgroup", + .flags = MS_NOSUID|MS_NOEXEC|MS_NODEV, + .mode = MNT_IN_CONTAINER, + }; char ***k = NULL; controller = set_steal_first(controllers); @@ -232,14 +254,13 @@ int mount_cgroup_controllers(char ***join_controllers) { for (i = *k, j = *k; *i; i++) { if (!streq(*i, controller)) { - char *t; + _cleanup_free_ char *t; t = set_remove(controllers, *i); if (!t) { free(*i); continue; } - free(t); } *(j++) = *i; @@ -248,82 +269,49 @@ int mount_cgroup_controllers(char ***join_controllers) { *j = NULL; options = strv_join(*k, ","); - if (!options) { - log_error("Failed to join options"); - free(controller); - r = -ENOMEM; - goto finish; - } - + if (!options) + return log_oom(); } else { options = controller; controller = NULL; } where = strappend("/sys/fs/cgroup/", options); - if (!where) { - log_error("Failed to build path"); - free(options); - r = -ENOMEM; - goto finish; - } + if (!where) + return log_oom(); - zero(p); - p.what = "cgroup"; p.where = where; - p.type = "cgroup"; p.options = options; - p.flags = MS_NOSUID|MS_NOEXEC|MS_NODEV; - p.fatal = false; r = mount_one(&p, true); - free(controller); - free(where); - - if (r < 0) { - free(options); - goto finish; - } + if (r < 0) + return r; if (r > 0 && k && *k) { char **i; for (i = *k; *i; i++) { - char *t; + _cleanup_free_ char *t = NULL; t = strappend("/sys/fs/cgroup/", *i); - if (!t) { - log_error("Failed to build path"); - r = -ENOMEM; - free(options); - goto finish; - } + if (!t) + return log_oom(); r = symlink(options, t); - free(t); - - if (r < 0 && errno != EEXIST) { - log_error("Failed to create symlink: %m"); - r = -errno; - free(options); - goto finish; - } + if (r < 0 && errno != EEXIST) + return log_error_errno(errno, "Failed to create symlink %s: %m", t); } } - - free(options); } - r = 0; - -finish: - set_free_free(controllers); - - fclose(f); + /* 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"); - return r; + return 0; } +#if defined(HAVE_SELINUX) || defined(HAVE_SMACK) static int nftw_cb( const char *fpath, const struct stat *sb, @@ -334,7 +322,7 @@ static int nftw_cb( if (_unlikely_(ftwbuf->level == 0)) return FTW_CONTINUE; - label_fix(fpath, true); + label_fix(fpath, false, false); /* /run/initramfs is static data and big, no need to * dynamically relabel its contents at boot... */ @@ -345,24 +333,24 @@ static int nftw_cb( return FTW_CONTINUE; }; +#endif int mount_setup(bool loaded_policy) { - - static const char relabel[] = - "/run/initramfs/root-fsck\0" - "/run/initramfs/shutdown\0"; - - int r; unsigned i; - const char *j; + int r = 0; for (i = 0; i < ELEMENTSOF(mount_table); i ++) { - r = mount_one(mount_table + i, true); + int j; - if (r < 0) - return r; + j = mount_one(mount_table + i, loaded_policy); + if (r == 0) + r = j; } + if (r < 0) + return r; + +#if defined(HAVE_SELINUX) || defined(HAVE_SMACK) /* Nodes in devtmpfs and /run need to be manually updated for * the appropriate labels, after mounting. The other virtual * API file systems like /sys and /proc do not need that, they @@ -376,24 +364,36 @@ int mount_setup(bool loaded_policy) { nftw("/dev", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL); nftw("/run", nftw_cb, 64, FTW_MOUNT|FTW_PHYS|FTW_ACTIONRETVAL); - /* Explicitly relabel these */ - NULSTR_FOREACH(j, relabel) - label_fix(j, true); - after_relabel = now(CLOCK_MONOTONIC); log_info("Relabelled /dev and /run in %s.", - format_timespan(timespan, sizeof(timespan), after_relabel - before_relabel)); + format_timespan(timespan, sizeof(timespan), after_relabel - before_relabel, 0)); } +#endif /* Create a few default symlinks, which are normally created * by udevd, but some scripts might need them before we start * udevd. */ - dev_setup(); - - /* Create a few directories we always want around */ - label_mkdir("/run/systemd", 0755); - label_mkdir("/run/systemd/system", 0755); + dev_setup(NULL); + + /* Mark the root directory as shared in regards to mount + * propagation. The kernel defaults to "private", but we think + * it makes more sense to have a default of "shared" so that + * nspawn and the container tools work out of the box. If + * specific setups need other settings they can reset the + * propagation mode to private if needed. */ + if (detect_container(NULL) <= 0) + if (mount(NULL, "/", NULL, MS_REC|MS_SHARED, NULL) < 0) + log_warning_errno(errno, "Failed to set up the root directory for shared mount propagation: %m"); + + /* Create a few directories we always want around, Note that + * sd_booted() checks for /run/systemd/system, so this mkdir + * really needs to stay for good, otherwise software that + * copied sd-daemon.c into their sources will misdetect + * systemd. */ + mkdir_label("/run/systemd", 0755); + mkdir_label("/run/systemd/system", 0755); + mkdir_label("/run/systemd/inaccessible", 0000); return 0; }