X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fcore%2Fnamespace.c;h=ab03aebf5b5dd9209a3c32b60c32be94eee94599;hp=7e33d84156348497d574308a7557f7e330aa9b7f;hb=a004cb4cb2271e04e87d4fd6f7353307038e83ea;hpb=d5a3f0eac7cd28ad28fb5239b0dca14c7ba11eeb

diff --git a/src/core/namespace.c b/src/core/namespace.c
index 7e33d8415..ab03aebf5 100644
--- a/src/core/namespace.c
+++ b/src/core/namespace.c
@@ -30,6 +30,7 @@
 #include <sys/syscall.h>
 #include <limits.h>
 #include <linux/fs.h>
+#include <sys/file.h>
 
 #include "strv.h"
 #include "util.h"
@@ -37,6 +38,11 @@
 #include "namespace.h"
 #include "missing.h"
 #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! */
@@ -44,6 +50,8 @@ typedef enum MountMode {
         READONLY,
         PRIVATE_TMP,
         PRIVATE_VAR_TMP,
+        PRIVATE_DEV,
+        PRIVATE_BUS_ENDPOINT,
         READWRITE
 } MountMode;
 
@@ -51,13 +59,24 @@ typedef struct BindMount {
         const char *path;
         MountMode mode;
         bool done;
+        bool ignore;
 } BindMount;
 
 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;
 
@@ -106,8 +125,7 @@ static void drop_duplicates(BindMount *m, unsigned *n) {
                 if (previous && path_equal(f->path, previous->path))
                         continue;
 
-                t->path = f->path;
-                t->mode = f->mode;
+                *t = *f;
 
                 previous = t;
 
@@ -117,6 +135,216 @@ static void drop_duplicates(BindMount *m, unsigned *n) {
         *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);
+
+        umount(dev);
+        rmdir(dev);
+        rmdir(temporary_mount);
+
+        return r;
+}
+
+static int mount_kdbus(BindMount *m) {
+
+        char temporary_mount[] = "/tmp/kdbus-dev-XXXXXX";
+        _cleanup_free_ char *basepath = NULL;
+        _cleanup_umask_ mode_t u;
+        char *busnode = NULL, *root;
+        struct stat st;
+        int r;
+
+        assert(m);
+
+        u = umask(0000);
+
+        if (!mkdtemp(temporary_mount)) {
+                log_error("Failed create temp dir: %m");
+                return -errno;
+        }
+
+        root = strappenda(temporary_mount, "/kdbus");
+        mkdir(root, 0755);
+        if (mount("tmpfs", root, "tmpfs", MS_NOSUID|MS_STRICTATIME, "mode=777") < 0) {
+                r = -errno;
+                goto fail;
+        }
+
+        /* create a new /dev/null dev node copy so we have some fodder to
+         * bind-mount the custom endpoint over. */
+        if (stat("/dev/null", &st) < 0) {
+                log_error("Failed to stat /dev/null: %m");
+                r = -errno;
+                goto fail;
+        }
+
+        busnode = strappenda(root, "/bus");
+        if (mknod(busnode, (st.st_mode & ~07777) | 0600, st.st_rdev) < 0) {
+                log_error("mknod() for %s failed: %m", busnode);
+                r = -errno;
+                goto fail;
+        }
+
+        r = mount(m->path, busnode, "bind", MS_BIND, NULL);
+        if (r < 0) {
+                log_error("bind mount of %s failed: %m", m->path);
+                r = -errno;
+                goto fail;
+        }
+
+        basepath = dirname_malloc(m->path);
+        if (!basepath) {
+                r = -ENOMEM;
+                goto fail;
+        }
+
+        if (mount(root, basepath, NULL, MS_MOVE, NULL) < 0) {
+                log_error("bind mount of %s failed: %m", basepath);
+                r = -errno;
+                goto fail;
+        }
+
+        rmdir(temporary_mount);
+        return 0;
+
+fail:
+        if (busnode) {
+                umount(busnode);
+                unlink(busnode);
+        }
+
+        umount(root);
+        rmdir(root);
+        rmdir(temporary_mount);
+
+        return r;
+}
+
 static int apply_mount(
                 BindMount *m,
                 const char *tmp_dir,
@@ -130,13 +358,20 @@ static int apply_mount(
         switch (m->mode) {
 
         case INACCESSIBLE:
+
+                /* 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 = m->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;
@@ -146,6 +381,12 @@ static int apply_mount(
                 what = var_tmp_dir;
                 break;
 
+        case PRIVATE_DEV:
+                return mount_dev(m);
+
+        case PRIVATE_BUS_ENDPOINT:
+                return mount_kdbus(m);
+
         default:
                 assert_not_reached("Unknown mode");
         }
@@ -155,6 +396,8 @@ static int apply_mount(
         r = mount(what, m->path, NULL, MS_BIND|MS_REC, NULL);
         if (r >= 0)
                 log_debug("Successfully mounted %s to %s", what, m->path);
+        else if (m->ignore && errno == ENOENT)
+                return 0;
 
         return r;
 }
@@ -164,114 +407,313 @@ static int make_read_only(BindMount *m) {
 
         assert(m);
 
-        if (m->mode != INACCESSIBLE && m->mode != READONLY)
+        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 (m->ignore && r == -ENOENT)
                 return 0;
 
-        r = mount(NULL, m->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,
+                const char* tmp_dir,
+                const char* var_tmp_dir,
+                const char* bus_endpoint_path,
+                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 + !!bus_endpoint_path +
+                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 (bus_endpoint_path) {
+                        m->path = bus_endpoint_path;
+                        m->mode = PRIVATE_BUS_ENDPOINT;
+                        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_tmpdirs(char **tmp_dir,
-                  char **var_tmp_dir) {
-        int r = 0;
-        char tmp_dir_template[] = "/tmp/systemd-private-XXXXXX",
-             var_tmp_dir_template[] = "/var/tmp/systemd-private-XXXXXX";
+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;
+
+        assert(id);
+        assert(prefix);
+        assert(path);
+
+        /* We include the boot id in the directory so that after a
+         * reboot we can easily identify obsolete directories. */
 
+        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;
+
+        RUN_WITH_UMASK(0000) {
+                char *y;
+
+                y = strappenda(x, "/tmp");
+
+                if (mkdir(y, 0777 | S_ISVTX) < 0)
+                        return -errno;
+        }
+
+        *path = x;
+        x = NULL;
+
+        return 0;
+}
+
+int setup_tmp_dirs(const char *id, char **tmp_dir, char **var_tmp_dir) {
+        char *a, *b;
+        int r;
+
+        assert(id);
         assert(tmp_dir);
         assert(var_tmp_dir);
 
-        r = create_tmp_dir(tmp_dir_template, tmp_dir);
+        r = setup_one_tmp_dir(id, "/tmp", &a);
         if (r < 0)
                 return r;
 
-        r = create_tmp_dir(var_tmp_dir_template, var_tmp_dir);
-        if (r == 0)
-                return 0;
+        r = setup_one_tmp_dir(id, "/var/tmp", &b);
+        if (r < 0) {
+                char *t;
 
-        /* failure */
-        rmdir(*tmp_dir);
-        rmdir(tmp_dir_template);
-        free(*tmp_dir);
-        *tmp_dir = NULL;
+                t = strappenda(a, "/tmp");
+                rmdir(t);
+                rmdir(a);
 
-        return r;
+                free(a);
+                return r;
+        }
+
+        *tmp_dir = a;
+        *var_tmp_dir = b;
+
+        return 0;
 }
 
-int setup_namespace(char** read_write_dirs,
-                    char** read_only_dirs,
-                    char** inaccessible_dirs,
-                    char* tmp_dir,
-                    char* var_tmp_dir,
-                    bool private_tmp,
-                    unsigned mount_flags) {
-
-        unsigned n = strv_length(read_write_dirs) +
-                     strv_length(read_only_dirs) +
-                     strv_length(inaccessible_dirs) +
-                     (private_tmp ? 2 : 0);
-        BindMount *m, *mounts;
-        int r = 0;
+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;
 
-        if (!mount_flags)
-                mount_flags = MS_SHARED;
+        assert(netns_storage_socket);
+        assert(netns_storage_socket[0] >= 0);
+        assert(netns_storage_socket[1] >= 0);
 
-        if (unshare(CLONE_NEWNS) < 0)
+        /* 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 (lockf(netns_storage_socket[0], F_LOCK, 0) < 0)
                 return -errno;
 
-        m = mounts = (BindMount *) alloca(n * sizeof(BindMount));
-        if ((r = append_mounts(&m, read_write_dirs, READWRITE)) < 0 ||
-                (r = append_mounts(&m, read_only_dirs, READONLY)) < 0 ||
-                (r = append_mounts(&m, inaccessible_dirs, INACCESSIBLE)) < 0)
-                return r;
+        if (recvmsg(netns_storage_socket[0], &mh, MSG_DONTWAIT|MSG_CMSG_CLOEXEC) < 0) {
+                if (errno != EAGAIN) {
+                        r = -errno;
+                        goto fail;
+                }
 
-        if (private_tmp) {
-                m->path = "/tmp";
-                m->mode = PRIVATE_TMP;
-                m++;
+                /* Nothing stored yet, so let's create a new namespace */
 
-                m->path = "/var/tmp";
-                m->mode = PRIVATE_VAR_TMP;
-                m++;
-        }
+                if (unshare(CLONE_NEWNET) < 0) {
+                        r = -errno;
+                        goto fail;
+                }
 
-        assert(mounts + n == m);
+                loopback_setup();
 
-        qsort(mounts, n, sizeof(BindMount), mount_path_compare);
-        drop_duplicates(mounts, &n);
+                netns = open("/proc/self/ns/net", O_RDONLY|O_CLOEXEC|O_NOCTTY);
+                if (netns < 0) {
+                        r = -errno;
+                        goto fail;
+                }
 
-        /* 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;
+                r = 1;
+        } else {
+                /* Yay, found something, so let's join the namespace */
 
-        for (m = mounts; m < mounts + n; ++m) {
-                r = apply_mount(m, tmp_dir, var_tmp_dir);
-                if (r < 0)
-                        goto undo_mounts;
-        }
+                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 (m = mounts; m < mounts + n; ++m) {
-                r = make_read_only(m);
-                if (r < 0)
-                        goto undo_mounts;
-        }
+                if (setns(netns, CLONE_NEWNET) < 0) {
+                        r = -errno;
+                        goto fail;
+                }
 
-        /* Remount / as the desired mode */
-        if (mount(NULL, "/", NULL, mount_flags | MS_REC, NULL) < 0) {
-                r = -errno;
-                goto undo_mounts;
+                r = 0;
         }
 
-        return 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;
 
-undo_mounts:
-        for (m = mounts; m < mounts + n; ++m) {
-                if (m->done)
-                        umount2(m->path, MNT_DETACH);
+        if (sendmsg(netns_storage_socket[1], &mh, MSG_DONTWAIT|MSG_NOSIGNAL) < 0) {
+                r = -errno;
+                goto fail;
         }
 
+fail:
+        lockf(netns_storage_socket[0], F_ULOCK, 0);
+
         return r;
 }
+
+static const char *const protect_home_table[_PROTECT_HOME_MAX] = {
+        [PROTECT_HOME_NO] = "no",
+        [PROTECT_HOME_YES] = "yes",
+        [PROTECT_HOME_READ_ONLY] = "read-only",
+};
+
+DEFINE_STRING_TABLE_LOOKUP(protect_home, ProtectHome);
+
+static const char *const protect_system_table[_PROTECT_SYSTEM_MAX] = {
+        [PROTECT_SYSTEM_NO] = "no",
+        [PROTECT_SYSTEM_YES] = "yes",
+        [PROTECT_SYSTEM_FULL] = "full",
+};
+
+DEFINE_STRING_TABLE_LOOKUP(protect_system, ProtectSystem);