X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fcore%2Fnamespace.c;h=fe9537787185c6755f9778593ae5aa393d760e7a;hp=ba18ddc5b0049dd005a1a7090850cd78826f0316;hb=e2d7c1a0758ce80d7cb439745deefefdffd67655;hpb=1e41be20158a6d982c34cea20e66ff271302abc5

diff --git a/src/core/namespace.c b/src/core/namespace.c
index ba18ddc5b..fe9537787 100644
--- a/src/core/namespace.c
+++ b/src/core/namespace.c
@@ -30,33 +30,52 @@
 #include <sys/syscall.h>
 #include <limits.h>
 #include <linux/fs.h>
+#include <sys/file.h>
 
 #include "strv.h"
 #include "util.h"
 #include "path-util.h"
 #include "namespace.h"
 #include "missing.h"
-
-typedef enum PathMode {
+#include "execute.h"
+#include "loopback-setup.h"
+#include "mkdir.h"
+#include "dev-setup.h"
+#include "def.h"
+#include "label.h"
+
+typedef enum MountMode {
         /* This is ordered by priority! */
         INACCESSIBLE,
         READONLY,
         PRIVATE_TMP,
         PRIVATE_VAR_TMP,
+        PRIVATE_DEV,
         READWRITE
-} PathMode;
+} MountMode;
 
-typedef struct Path {
+typedef struct BindMount {
         const char *path;
-        PathMode mode;
+        MountMode mode;
         bool done;
-} Path;
+        bool ignore;
+} BindMount;
 
-static int append_paths(Path **p, char **strv, PathMode mode) {
+static int append_mounts(BindMount **p, char **strv, MountMode mode) {
         char **i;
 
+        assert(p);
+
         STRV_FOREACH(i, strv) {
 
+                (*p)->ignore = false;
+                (*p)->done = false;
+
+                if ((mode == INACCESSIBLE || mode == READONLY || mode == READWRITE) && (*i)[0] == '-') {
+                        (*p)->ignore = true;
+                        (*i)++;
+                }
+
                 if (!path_is_absolute(*i))
                         return -EINVAL;
 
@@ -68,8 +87,8 @@ static int append_paths(Path **p, char **strv, PathMode mode) {
         return 0;
 }
 
-static int path_compare(const void *a, const void *b) {
-        const Path *p = a, *q = b;
+static int mount_path_compare(const void *a, const void *b) {
+        const BindMount *p = a, *q = b;
 
         if (path_equal(p->path, q->path)) {
 
@@ -93,54 +112,193 @@ static int path_compare(const void *a, const void *b) {
         return 0;
 }
 
-static void drop_duplicates(Path *p, unsigned *n, bool *need_inaccessible) {
-        Path *f, *t, *previous;
+static void drop_duplicates(BindMount *m, unsigned *n) {
+        BindMount *f, *t, *previous;
 
-        assert(p);
+        assert(m);
         assert(n);
-        assert(need_inaccessible);
 
-        for (f = p, t = p, previous = NULL; f < p+*n; f++) {
+        for (f = m, t = m, previous = NULL; f < m+*n; f++) {
 
                 /* The first one wins */
                 if (previous && path_equal(f->path, previous->path))
                         continue;
 
-                t->path = f->path;
-                t->mode = f->mode;
-
-                if (t->mode == INACCESSIBLE)
-                        *need_inaccessible = true;
+                *t = *f;
 
                 previous = t;
 
                 t++;
         }
 
-        *n = t - p;
+        *n = t - m;
+}
+
+static int mount_dev(BindMount *m) {
+        static const char devnodes[] =
+                "/dev/null\0"
+                "/dev/zero\0"
+                "/dev/full\0"
+                "/dev/random\0"
+                "/dev/urandom\0"
+                "/dev/tty\0";
+
+        char temporary_mount[] = "/tmp/namespace-dev-XXXXXX";
+        const char *d, *dev = NULL, *devpts = NULL, *devshm = NULL, *devkdbus = NULL, *devhugepages = NULL, *devmqueue = NULL, *devlog = NULL, *devptmx = NULL;
+        _cleanup_umask_ mode_t u;
+        int r;
+
+        assert(m);
+
+        u = umask(0000);
+
+        if (!mkdtemp(temporary_mount))
+                return -errno;
+
+        dev = strappenda(temporary_mount, "/dev");
+        mkdir(dev, 0755);
+        if (mount("tmpfs", dev, "tmpfs", MS_NOSUID|MS_STRICTATIME, "mode=755") < 0) {
+                r = -errno;
+                goto fail;
+        }
+
+        devpts = strappenda(temporary_mount, "/dev/pts");
+        mkdir(devpts, 0755);
+        if (mount("/dev/pts", devpts, NULL, MS_BIND, NULL) < 0) {
+                r = -errno;
+                goto fail;
+        }
+
+        devptmx = strappenda(temporary_mount, "/dev/ptmx");
+        symlink("pts/ptmx", devptmx);
+
+        devshm = strappenda(temporary_mount, "/dev/shm");
+        mkdir(devshm, 01777);
+        r = mount("/dev/shm", devshm, NULL, MS_BIND, NULL);
+        if (r < 0) {
+                r = -errno;
+                goto fail;
+        }
+
+        devmqueue = strappenda(temporary_mount, "/dev/mqueue");
+        mkdir(devmqueue, 0755);
+        mount("/dev/mqueue", devmqueue, NULL, MS_BIND, NULL);
+
+        devkdbus = strappenda(temporary_mount, "/dev/kdbus");
+        mkdir(devkdbus, 0755);
+        mount("/dev/kdbus", devkdbus, NULL, MS_BIND, NULL);
+
+        devhugepages = strappenda(temporary_mount, "/dev/hugepages");
+        mkdir(devhugepages, 0755);
+        mount("/dev/hugepages", devhugepages, NULL, MS_BIND, NULL);
+
+        devlog = strappenda(temporary_mount, "/dev/log");
+        symlink("/run/systemd/journal/dev-log", devlog);
+
+        NULSTR_FOREACH(d, devnodes) {
+                _cleanup_free_ char *dn = NULL;
+                struct stat st;
+
+                r = stat(d, &st);
+                if (r < 0) {
+
+                        if (errno == ENOENT)
+                                continue;
+
+                        r = -errno;
+                        goto fail;
+                }
+
+                if (!S_ISBLK(st.st_mode) &&
+                    !S_ISCHR(st.st_mode)) {
+                        r = -EINVAL;
+                        goto fail;
+                }
+
+                if (st.st_rdev == 0)
+                        continue;
+
+                dn = strappend(temporary_mount, d);
+                if (!dn) {
+                        r = -ENOMEM;
+                        goto fail;
+                }
+
+                label_context_set(d, st.st_mode);
+                r = mknod(dn, st.st_mode, st.st_rdev);
+                label_context_clear();
+
+                if (r < 0) {
+                        r = -errno;
+                        goto fail;
+                }
+        }
+
+        dev_setup(temporary_mount);
+
+        if (mount(dev, "/dev/", NULL, MS_MOVE, NULL) < 0) {
+                r = -errno;
+                goto fail;
+        }
+
+        rmdir(dev);
+        rmdir(temporary_mount);
+
+        return 0;
+
+fail:
+        if (devpts)
+                umount(devpts);
+
+        if (devshm)
+                umount(devshm);
+
+        if (devkdbus)
+                umount(devkdbus);
+
+        if (devhugepages)
+                umount(devhugepages);
+
+        if (devmqueue)
+                umount(devmqueue);
+
+        if (dev) {
+                umount(dev);
+                rmdir(dev);
+        }
+
+        rmdir(temporary_mount);
+
+        return r;
 }
 
 static int apply_mount(
-                Path *p,
+                BindMount *m,
                 const char *tmp_dir,
-                const char *var_tmp_dir,
-                const char *inaccessible_dir) {
+                const char *var_tmp_dir) {
 
         const char *what;
         int r;
 
-        assert(p);
+        assert(m);
 
-        switch (p->mode) {
+        switch (m->mode) {
 
         case INACCESSIBLE:
-                what = inaccessible_dir;
+
+                /* First, get rid of everything that is below if there
+                 * is anything... Then, overmount it with an
+                 * inaccessible directory. */
+                umount_recursive(m->path, 0);
+
+                what = "/run/systemd/inaccessible";
                 break;
 
         case READONLY:
         case READWRITE:
-                what = p->path;
-                break;
+                /* Nothing to mount here, we just later toggle the
+                 * MS_RDONLY bit for the mount point */
+                return 0;
 
         case PRIVATE_TMP:
                 what = tmp_dir;
@@ -150,181 +308,329 @@ static int apply_mount(
                 what = var_tmp_dir;
                 break;
 
+        case PRIVATE_DEV:
+                return mount_dev(m);
+
         default:
                 assert_not_reached("Unknown mode");
         }
 
         assert(what);
 
-        r = mount(what, p->path, NULL, MS_BIND|MS_REC, NULL);
+        r = mount(what, m->path, NULL, MS_BIND|MS_REC, NULL);
         if (r >= 0)
-                log_debug("Successfully mounted %s to %s", what, p->path);
+                log_debug("Successfully mounted %s to %s", what, m->path);
+        else if (m->ignore && errno == ENOENT)
+                return 0;
 
         return r;
 }
 
-static int make_read_only(Path *p) {
+static int make_read_only(BindMount *m) {
         int r;
 
-        assert(p);
+        assert(m);
+
+        if (IN_SET(m->mode, INACCESSIBLE, READONLY))
+                r = bind_remount_recursive(m->path, true);
+        else if (IN_SET(m->mode, READWRITE, PRIVATE_TMP, PRIVATE_VAR_TMP, PRIVATE_DEV))
+                r = bind_remount_recursive(m->path, false);
+        else
+                r = 0;
 
-        if (p->mode != INACCESSIBLE && p->mode != READONLY)
+        if (m->ignore && r == -ENOENT)
                 return 0;
 
-        r = mount(NULL, p->path, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY|MS_REC, NULL);
-        if (r < 0)
+        return r;
+}
+
+int setup_namespace(
+                char** read_write_dirs,
+                char** read_only_dirs,
+                char** inaccessible_dirs,
+                char* tmp_dir,
+                char* var_tmp_dir,
+                bool private_dev,
+                ProtectHome protect_home,
+                ProtectSystem protect_system,
+                unsigned mount_flags) {
+
+        BindMount *m, *mounts = NULL;
+        unsigned n;
+        int r = 0;
+
+        if (mount_flags == 0)
+                mount_flags = MS_SHARED;
+
+        if (unshare(CLONE_NEWNS) < 0)
                 return -errno;
 
+        n = !!tmp_dir + !!var_tmp_dir +
+                strv_length(read_write_dirs) +
+                strv_length(read_only_dirs) +
+                strv_length(inaccessible_dirs) +
+                private_dev +
+                (protect_home != PROTECT_HOME_NO ? 3 : 0) +
+                (protect_system != PROTECT_SYSTEM_NO ? 2 : 0) +
+                (protect_system == PROTECT_SYSTEM_FULL ? 1 : 0);
+
+        if (n > 0) {
+                m = mounts = (BindMount *) alloca0(n * sizeof(BindMount));
+                r = append_mounts(&m, read_write_dirs, READWRITE);
+                if (r < 0)
+                        return r;
+
+                r = append_mounts(&m, read_only_dirs, READONLY);
+                if (r < 0)
+                        return r;
+
+                r = append_mounts(&m, inaccessible_dirs, INACCESSIBLE);
+                if (r < 0)
+                        return r;
+
+                if (tmp_dir) {
+                        m->path = "/tmp";
+                        m->mode = PRIVATE_TMP;
+                        m++;
+                }
+
+                if (var_tmp_dir) {
+                        m->path = "/var/tmp";
+                        m->mode = PRIVATE_VAR_TMP;
+                        m++;
+                }
+
+                if (private_dev) {
+                        m->path = "/dev";
+                        m->mode = PRIVATE_DEV;
+                        m++;
+                }
+
+                if (protect_home != PROTECT_HOME_NO) {
+                        r = append_mounts(&m, STRV_MAKE("-/home", "-/run/user", "-/root"), protect_home == PROTECT_HOME_READ_ONLY ? READONLY : INACCESSIBLE);
+                        if (r < 0)
+                                return r;
+                }
+
+                if (protect_system != PROTECT_SYSTEM_NO) {
+                        r = append_mounts(&m, protect_system == PROTECT_SYSTEM_FULL ? STRV_MAKE("/usr", "-/boot", "/etc") : STRV_MAKE("/usr", "-/boot"), READONLY);
+                        if (r < 0)
+                                return r;
+                }
+
+                assert(mounts + n == m);
+
+                qsort(mounts, n, sizeof(BindMount), mount_path_compare);
+                drop_duplicates(mounts, &n);
+        }
+
+        if (n > 0) {
+                /* Remount / as SLAVE so that nothing now mounted in the namespace
+                   shows up in the parent */
+                if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0)
+                        return -errno;
+
+                for (m = mounts; m < mounts + n; ++m) {
+                        r = apply_mount(m, tmp_dir, var_tmp_dir);
+                        if (r < 0)
+                                goto fail;
+                }
+
+                for (m = mounts; m < mounts + n; ++m) {
+                        r = make_read_only(m);
+                        if (r < 0)
+                                goto fail;
+                }
+        }
+
+        /* Remount / as the desired mode. Not that this will not
+         * reestablish propagation from our side to the host, since
+         * what's disconnected is disconnected. */
+        if (mount(NULL, "/", NULL, mount_flags | MS_REC, NULL) < 0) {
+                r = -errno;
+                goto fail;
+        }
+
         return 0;
+
+fail:
+        if (n > 0) {
+                for (m = mounts; m < mounts + n; ++m)
+                        if (m->done)
+                                umount2(m->path, MNT_DETACH);
+        }
+
+        return r;
 }
 
-int setup_namespace(
-                char **writable,
-                char **readable,
-                char **inaccessible,
-                bool private_tmp,
-                unsigned long flags) {
-
-        char
-                tmp_dir[] = "/tmp/systemd-private-XXXXXX",
-                var_tmp_dir[] = "/var/tmp/systemd-private-XXXXXX",
-                inaccessible_dir[] = "/tmp/systemd-inaccessible-XXXXXX";
-
-        Path *paths, *p;
-        unsigned n;
-        bool need_inaccessible = false;
-        bool remove_tmp = false, remove_var_tmp = false, remove_inaccessible = false;
+static int setup_one_tmp_dir(const char *id, const char *prefix, char **path) {
+        _cleanup_free_ char *x = NULL;
+        char bid[SD_ID128_STRING_MAX];
+        sd_id128_t boot_id;
         int r;
 
-        if (!flags)
-                flags = MS_SHARED;
+        assert(id);
+        assert(prefix);
+        assert(path);
 
-        n =
-                strv_length(writable) +
-                strv_length(readable) +
-                strv_length(inaccessible) +
-                (private_tmp ? 2 : 0);
+        /* We include the boot id in the directory so that after a
+         * reboot we can easily identify obsolete directories. */
 
-        p = paths = alloca(sizeof(Path) * n);
-        if ((r = append_paths(&p, writable, READWRITE)) < 0 ||
-            (r = append_paths(&p, readable, READONLY)) < 0 ||
-            (r = append_paths(&p, inaccessible, INACCESSIBLE)) < 0)
-                goto fail;
+        r = sd_id128_get_boot(&boot_id);
+        if (r < 0)
+                return r;
+
+        x = strjoin(prefix, "/systemd-private-", sd_id128_to_string(boot_id, bid), "-", id, "-XXXXXX", NULL);
+        if (!x)
+                return -ENOMEM;
+
+        RUN_WITH_UMASK(0077)
+                if (!mkdtemp(x))
+                        return -errno;
 
-        if (private_tmp) {
-                p->path = "/tmp";
-                p->mode = PRIVATE_TMP;
-                p++;
+        RUN_WITH_UMASK(0000) {
+                char *y;
 
-                p->path = "/var/tmp";
-                p->mode = PRIVATE_VAR_TMP;
-                p++;
+                y = strappenda(x, "/tmp");
+
+                if (mkdir(y, 0777 | S_ISVTX) < 0)
+                        return -errno;
         }
 
-        assert(paths + n == p);
+        *path = x;
+        x = NULL;
 
-        qsort(paths, n, sizeof(Path), path_compare);
-        drop_duplicates(paths, &n, &need_inaccessible);
+        return 0;
+}
 
-        if (need_inaccessible) {
-                mode_t u;
-                char *d;
+int setup_tmp_dirs(const char *id, char **tmp_dir, char **var_tmp_dir) {
+        char *a, *b;
+        int r;
 
-                u = umask(0777);
-                d = mkdtemp(inaccessible_dir);
-                umask(u);
+        assert(id);
+        assert(tmp_dir);
+        assert(var_tmp_dir);
 
-                if (!d) {
-                        r = -errno;
-                        goto fail;
-                }
+        r = setup_one_tmp_dir(id, "/tmp", &a);
+        if (r < 0)
+                return r;
+
+        r = setup_one_tmp_dir(id, "/var/tmp", &b);
+        if (r < 0) {
+                char *t;
 
-                remove_inaccessible = true;
+                t = strappenda(a, "/tmp");
+                rmdir(t);
+                rmdir(a);
+
+                free(a);
+                return r;
         }
 
-        if (private_tmp) {
-                mode_t u;
-                char *d;
+        *tmp_dir = a;
+        *var_tmp_dir = b;
 
-                u = umask(0000);
-                d = mkdtemp(tmp_dir);
-                umask(u);
+        return 0;
+}
 
-                if (!d) {
-                        r = -errno;
-                        goto fail;
-                }
+int setup_netns(int netns_storage_socket[2]) {
+        _cleanup_close_ int netns = -1;
+        union {
+                struct cmsghdr cmsghdr;
+                uint8_t buf[CMSG_SPACE(sizeof(int))];
+        } control = {};
+        struct msghdr mh = {
+                .msg_control = &control,
+                .msg_controllen = sizeof(control),
+        };
+        struct cmsghdr *cmsg;
+        int r;
 
-                remove_tmp = true;
+        assert(netns_storage_socket);
+        assert(netns_storage_socket[0] >= 0);
+        assert(netns_storage_socket[1] >= 0);
 
-                u = umask(0000);
-                d = mkdtemp(var_tmp_dir);
-                umask(u);
+        /* We use the passed socketpair as a storage buffer for our
+         * namespace reference fd. Whatever process runs this first
+         * shall create a new namespace, all others should just join
+         * it. To serialize that we use a file lock on the socket
+         * pair.
+         *
+         * It's a bit crazy, but hey, works great! */
 
-                if (!d) {
+        if (lockf(netns_storage_socket[0], F_LOCK, 0) < 0)
+                return -errno;
+
+        if (recvmsg(netns_storage_socket[0], &mh, MSG_DONTWAIT|MSG_CMSG_CLOEXEC) < 0) {
+                if (errno != EAGAIN) {
                         r = -errno;
                         goto fail;
                 }
 
-                remove_var_tmp = true;
+                /* Nothing stored yet, so let's create a new namespace */
 
-                if (chmod(tmp_dir, 0777 + S_ISVTX) < 0) {
+                if (unshare(CLONE_NEWNET) < 0) {
                         r = -errno;
                         goto fail;
                 }
 
-                if (chmod(var_tmp_dir, 0777 + S_ISVTX) < 0) {
+                loopback_setup();
+
+                netns = open("/proc/self/ns/net", O_RDONLY|O_CLOEXEC|O_NOCTTY);
+                if (netns < 0) {
                         r = -errno;
                         goto fail;
                 }
-        }
 
-        if (unshare(CLONE_NEWNS) < 0) {
-                r = -errno;
-                goto fail;
-        }
+                r = 1;
+        } else {
+                /* Yay, found something, so let's join the namespace */
 
-        /* Remount / as SLAVE so that nothing now mounted in the namespace
-           shows up in the parent */
-        if (mount(NULL, "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) {
-                r = -errno;
-                goto fail;
-        }
+                for (cmsg = CMSG_FIRSTHDR(&mh); cmsg; cmsg = CMSG_NXTHDR(&mh, cmsg)) {
+                        if (cmsg->cmsg_level == SOL_SOCKET && cmsg->cmsg_type == SCM_RIGHTS) {
+                                assert(cmsg->cmsg_len == CMSG_LEN(sizeof(int)));
+                                netns = *(int*) CMSG_DATA(cmsg);
+                        }
+                }
 
-        for (p = paths; p < paths + n; p++) {
-                r = apply_mount(p, tmp_dir, var_tmp_dir, inaccessible_dir);
-                if (r < 0)
-                        goto undo_mounts;
-        }
+                if (setns(netns, CLONE_NEWNET) < 0) {
+                        r = -errno;
+                        goto fail;
+                }
 
-        for (p = paths; p < paths + n; p++) {
-                r = make_read_only(p);
-                if (r < 0)
-                        goto undo_mounts;
+                r = 0;
         }
 
-        /* Remount / as the desired mode */
-        if (mount(NULL, "/", NULL, flags|MS_REC, NULL) < 0) {
+        cmsg = CMSG_FIRSTHDR(&mh);
+        cmsg->cmsg_level = SOL_SOCKET;
+        cmsg->cmsg_type = SCM_RIGHTS;
+        cmsg->cmsg_len = CMSG_LEN(sizeof(int));
+        memcpy(CMSG_DATA(cmsg), &netns, sizeof(int));
+        mh.msg_controllen = cmsg->cmsg_len;
+
+        if (sendmsg(netns_storage_socket[1], &mh, MSG_DONTWAIT|MSG_NOSIGNAL) < 0) {
                 r = -errno;
-                goto undo_mounts;
+                goto fail;
         }
 
-        return 0;
+fail:
+        lockf(netns_storage_socket[0], F_ULOCK, 0);
 
-undo_mounts:
-        for (p = paths; p < paths + n; p++)
-                if (p->done)
-                        umount2(p->path, MNT_DETACH);
+        return r;
+}
 
-fail:
-        if (remove_inaccessible)
-                rmdir(inaccessible_dir);
+static const char *const protect_home_table[_PROTECT_HOME_MAX] = {
+        [PROTECT_HOME_NO] = "no",
+        [PROTECT_HOME_YES] = "yes",
+        [PROTECT_HOME_READ_ONLY] = "read-only",
+};
 
-        if (remove_tmp)
-                rmdir(tmp_dir);
+DEFINE_STRING_TABLE_LOOKUP(protect_home, ProtectHome);
 
-        if (remove_var_tmp)
-                rmdir(var_tmp_dir);
+static const char *const protect_system_table[_PROTECT_SYSTEM_MAX] = {
+        [PROTECT_SYSTEM_NO] = "no",
+        [PROTECT_SYSTEM_YES] = "yes",
+        [PROTECT_SYSTEM_FULL] = "full",
+};
 
-        return r;
-}
+DEFINE_STRING_TABLE_LOOKUP(protect_system, ProtectSystem);