build-sys: move *-setup out of shared to avoid selinux being pulled in

6 files changed, 1052 insertions, 0 deletions

diff --git a/src/core/loopback-setup.c b/src/core/loopback-setup.c
new file mode 100644
index 0000000000..46c1fc843a
--- /dev/null
+++ b/src/core/loopback-setup.c
@@ -0,0 +1,274 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <errno.h>
+#include <sys/socket.h>
+#include <net/if.h>
+#include <asm/types.h>
+#include <netinet/in.h>
+#include <string.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
+
+#include "util.h"
+#include "macro.h"
+#include "loopback-setup.h"
+#include "socket-util.h"
+
+#define NLMSG_TAIL(nmsg)                                                \
+        ((struct rtattr *) (((uint8_t*) (nmsg)) + NLMSG_ALIGN((nmsg)->nlmsg_len)))
+
+static int add_rtattr(struct nlmsghdr *n, size_t max_length, int type, const void *data, size_t data_length) {
+        size_t length;
+        struct rtattr *rta;
+
+        length = RTA_LENGTH(data_length);
+
+        if (NLMSG_ALIGN(n->nlmsg_len) + RTA_ALIGN(length) > max_length)
+                return -E2BIG;
+
+        rta = NLMSG_TAIL(n);
+        rta->rta_type = type;
+        rta->rta_len = length;
+        memcpy(RTA_DATA(rta), data, data_length);
+        n->nlmsg_len = NLMSG_ALIGN(n->nlmsg_len) + RTA_ALIGN(length);
+
+        return 0;
+}
+
+static ssize_t sendto_loop(int fd, const void *buf, size_t buf_len, int flags, const struct sockaddr *sa, socklen_t sa_len) {
+
+        for (;;) {
+                ssize_t l;
+
+                if ((l = sendto(fd, buf, buf_len, flags, sa, sa_len)) >= 0)
+                        return l;
+
+                if (errno != EINTR)
+                        return -errno;
+        }
+}
+
+static ssize_t recvfrom_loop(int fd, void *buf, size_t buf_len, int flags, struct sockaddr *sa, socklen_t *sa_len) {
+
+        for (;;) {
+                ssize_t l;
+
+                if ((l = recvfrom(fd, buf, buf_len, flags, sa, sa_len)) >= 0)
+                        return l;
+
+                if (errno != EINTR)
+                        return -errno;
+        }
+}
+
+static int add_adresses(int fd, int if_loopback, unsigned *requests) {
+        union {
+                struct sockaddr sa;
+                struct sockaddr_nl nl;
+        } sa;
+        union {
+                struct nlmsghdr header;
+                uint8_t buf[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
+                            NLMSG_ALIGN(sizeof(struct ifaddrmsg)) +
+                            RTA_LENGTH(sizeof(struct in6_addr))];
+        } request;
+
+        struct ifaddrmsg *ifaddrmsg;
+        uint32_t ipv4_address = htonl(INADDR_LOOPBACK);
+        int r;
+
+        zero(request);
+
+        request.header.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifaddrmsg));
+        request.header.nlmsg_type = RTM_NEWADDR;
+        request.header.nlmsg_flags = NLM_F_REQUEST|NLM_F_CREATE|NLM_F_ACK;
+        request.header.nlmsg_seq = *requests + 1;
+
+        ifaddrmsg = NLMSG_DATA(&request.header);
+        ifaddrmsg->ifa_family = AF_INET;
+        ifaddrmsg->ifa_prefixlen = 8;
+        ifaddrmsg->ifa_flags = IFA_F_PERMANENT;
+        ifaddrmsg->ifa_scope = RT_SCOPE_HOST;
+        ifaddrmsg->ifa_index = if_loopback;
+
+        if ((r = add_rtattr(&request.header, sizeof(request), IFA_LOCAL, &ipv4_address, sizeof(ipv4_address))) < 0)
+                return r;
+
+        zero(sa);
+        sa.nl.nl_family = AF_NETLINK;
+
+        if (sendto_loop(fd, &request, request.header.nlmsg_len, 0, &sa.sa, sizeof(sa)) < 0)
+                return -errno;
+        (*requests)++;
+
+        if (!socket_ipv6_is_supported())
+                return 0;
+
+        request.header.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifaddrmsg));
+        request.header.nlmsg_seq = *requests + 1;
+
+        ifaddrmsg->ifa_family = AF_INET6;
+        ifaddrmsg->ifa_prefixlen = 128;
+
+        if ((r = add_rtattr(&request.header, sizeof(request), IFA_LOCAL, &in6addr_loopback, sizeof(in6addr_loopback))) < 0)
+                return r;
+
+        if (sendto_loop(fd, &request, request.header.nlmsg_len, 0, &sa.sa, sizeof(sa)) < 0)
+                return -errno;
+        (*requests)++;
+
+        return 0;
+}
+
+static int start_interface(int fd, int if_loopback, unsigned *requests) {
+        union {
+                struct sockaddr sa;
+                struct sockaddr_nl nl;
+        } sa;
+        union {
+                struct nlmsghdr header;
+                uint8_t buf[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
+                            NLMSG_ALIGN(sizeof(struct ifinfomsg))];
+        } request;
+
+        struct ifinfomsg *ifinfomsg;
+
+        zero(request);
+
+        request.header.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg));
+        request.header.nlmsg_type = RTM_NEWLINK;
+        request.header.nlmsg_flags = NLM_F_REQUEST|NLM_F_ACK;
+        request.header.nlmsg_seq = *requests + 1;
+
+        ifinfomsg = NLMSG_DATA(&request.header);
+        ifinfomsg->ifi_family = AF_UNSPEC;
+        ifinfomsg->ifi_index = if_loopback;
+        ifinfomsg->ifi_flags = IFF_UP;
+        ifinfomsg->ifi_change = IFF_UP;
+
+        zero(sa);
+        sa.nl.nl_family = AF_NETLINK;
+
+        if (sendto_loop(fd, &request, request.header.nlmsg_len, 0, &sa.sa, sizeof(sa)) < 0)
+                return -errno;
+
+        (*requests)++;
+
+        return 0;
+}
+
+static int read_response(int fd, unsigned requests_max) {
+        union {
+                struct sockaddr sa;
+                struct sockaddr_nl nl;
+        } sa;
+        union {
+                struct nlmsghdr header;
+                uint8_t buf[NLMSG_ALIGN(sizeof(struct nlmsghdr)) +
+                            NLMSG_ALIGN(sizeof(struct nlmsgerr))];
+        } response;
+
+        ssize_t l;
+        socklen_t sa_len = sizeof(sa);
+        struct nlmsgerr *nlmsgerr;
+
+        if ((l = recvfrom_loop(fd, &response, sizeof(response), 0, &sa.sa, &sa_len)) < 0)
+                return -errno;
+
+        if (sa_len != sizeof(sa.nl) ||
+            sa.nl.nl_family != AF_NETLINK)
+                return -EIO;
+
+        if (sa.nl.nl_pid != 0)
+                return 0;
+
+        if ((size_t) l < sizeof(struct nlmsghdr))
+                return -EIO;
+
+        if (response.header.nlmsg_type != NLMSG_ERROR ||
+            (pid_t) response.header.nlmsg_pid != getpid() ||
+            response.header.nlmsg_seq >= requests_max)
+                return 0;
+
+        if ((size_t) l < NLMSG_LENGTH(sizeof(struct nlmsgerr)) ||
+            response.header.nlmsg_len < NLMSG_LENGTH(sizeof(struct nlmsgerr)))
+                return -EIO;
+
+        nlmsgerr = NLMSG_DATA(&response.header);
+
+        if (nlmsgerr->error < 0 && nlmsgerr->error != -EEXIST) {
+                log_warning("Netlink failure for request %i: %s", response.header.nlmsg_seq, strerror(-nlmsgerr->error));
+                return nlmsgerr->error;
+        }
+
+        return response.header.nlmsg_seq;
+}
+
+int loopback_setup(void) {
+        int r, if_loopback;
+        union {
+                struct sockaddr sa;
+                struct sockaddr_nl nl;
+                struct sockaddr_storage storage;
+        } sa;
+        unsigned requests = 0, i;
+        int fd;
+
+        errno = 0;
+        if ((if_loopback = (int) if_nametoindex("lo")) <= 0)
+                return errno ? -errno : -ENODEV;
+
+        if ((fd = socket(PF_NETLINK, SOCK_RAW, NETLINK_ROUTE)) < 0)
+                return -errno;
+
+        zero(sa);
+        sa.nl.nl_family = AF_NETLINK;
+
+        if (bind(fd, &sa.sa, sizeof(sa)) < 0) {
+                r = -errno;
+                goto finish;
+        }
+
+        if ((r = add_adresses(fd, if_loopback, &requests)) < 0)
+                goto finish;
+
+        if ((r = start_interface(fd, if_loopback, &requests)) < 0)
+                goto finish;
+
+        for (i = 0; i < requests; i++) {
+                if ((r = read_response(fd, requests)) < 0)
+                        goto finish;
+        }
+
+        r = 0;
+
+finish:
+        if (r < 0)
+                log_warning("Failed to configure loopback device: %s", strerror(-r));
+
+        if (fd >= 0)
+                close_nointr_nofail(fd);
+
+        return r;
+}
diff --git a/src/core/loopback-setup.h b/src/core/loopback-setup.h
new file mode 100644
index 0000000000..cbb969e1e4
--- /dev/null
+++ b/src/core/loopback-setup.h
@@ -0,0 +1,27 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+#ifndef fooloopbacksetuphfoo
+#define fooloopbacksetuphfoo
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+int loopback_setup(void);
+
+#endif
diff --git a/src/core/machine-id-setup.c b/src/core/machine-id-setup.c
new file mode 100644
index 0000000000..9e84ac0cb9
--- /dev/null
+++ b/src/core/machine-id-setup.c
@@ -0,0 +1,265 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <unistd.h>
+#include <stdio.h>
+#include <errno.h>
+#include <string.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <sys/mount.h>
+
+#include <systemd/sd-id128.h>
+
+#include "machine-id-setup.h"
+#include "macro.h"
+#include "util.h"
+#include "mkdir.h"
+#include "log.h"
+#include "virt.h"
+
+static int shorten_uuid(char destination[36], const char *source) {
+        unsigned i, j;
+
+        for (i = 0, j = 0; i < 36 && j < 32; i++) {
+                int t;
+
+                t = unhexchar(source[i]);
+                if (t < 0)
+                        continue;
+
+                destination[j++] = hexchar(t);
+        }
+
+        if (i == 36 && j == 32) {
+                destination[32] = '\n';
+                destination[33] = 0;
+                return 0;
+        }
+
+        return -EINVAL;
+}
+
+static int generate(char id[34]) {
+        int fd, r;
+        unsigned char *p;
+        sd_id128_t buf;
+        char *q;
+        ssize_t k;
+        const char *vm_id;
+
+        assert(id);
+
+        /* First, try reading the D-Bus machine id, unless it is a symlink */
+        fd = open("/var/lib/dbus/machine-id", O_RDONLY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW);
+        if (fd >= 0) {
+
+                k = loop_read(fd, id, 32, false);
+                close_nointr_nofail(fd);
+
+                if (k >= 32) {
+                        id[32] = '\n';
+                        id[33] = 0;
+
+                        log_info("Initializing machine ID from D-Bus machine ID.");
+                        return 0;
+                }
+        }
+
+        /* If that didn't work, see if we are running in qemu/kvm and a
+         * machine ID was passed in via -uuid on the qemu/kvm command
+         * line */
+
+        r = detect_vm(&vm_id);
+        if (r > 0 && streq(vm_id, "kvm")) {
+                char uuid[37];
+
+                fd = open("/sys/class/dmi/id/product_uuid", O_RDONLY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW);
+                if (fd >= 0) {
+                        k = loop_read(fd, uuid, 36, false);
+                        close_nointr_nofail(fd);
+
+                        if (k >= 36) {
+                                r = shorten_uuid(id, uuid);
+                                if (r >= 0) {
+                                        log_info("Initializing machine ID from KVM UUID");
+                                        return 0;
+                                }
+                        }
+                }
+        }
+
+        /* If that didn't work either, see if we are running in a
+         * container, and a machine ID was passed in via
+         * $container_uuid the way libvirt/LXC does it */
+
+        r = detect_container(NULL);
+        if (r > 0) {
+                FILE *f;
+
+                f = fopen("/proc/1/environ", "re");
+                if (f) {
+                        bool done = false;
+
+                        do {
+                                char line[LINE_MAX];
+                                unsigned i;
+
+                                for (i = 0; i < sizeof(line)-1; i++) {
+                                        int c;
+
+                                        c = getc(f);
+                                        if (_unlikely_(c == EOF)) {
+                                                done = true;
+                                                break;
+                                        } else if (c == 0)
+                                                break;
+
+                                        line[i] = c;
+                                }
+                                line[i] = 0;
+
+                                if (startswith(line, "container_uuid=") &&
+                                    strlen(line + 15) >= 36) {
+                                        r = shorten_uuid(id, line + 15);
+                                        if (r >= 0) {
+                                                log_info("Initializing machine ID from container UUID");
+                                                return 0;
+                                        }
+                                }
+
+                        } while (!done);
+
+                        fclose(f);
+                }
+        }
+
+        /* If that didn't work, generate a random machine id */
+        r = sd_id128_randomize(&buf);
+        if (r < 0) {
+                log_error("Failed to open /dev/urandom: %s", strerror(-r));
+                return r;
+        }
+
+        for (p = buf.bytes, q = id; p < buf.bytes + sizeof(buf); p++, q += 2) {
+                q[0] = hexchar(*p >> 4);
+                q[1] = hexchar(*p & 15);
+        }
+
+        id[32] = '\n';
+        id[33] = 0;
+
+        log_info("Initializing machine ID from random generator.");
+
+        return 0;
+}
+
+int machine_id_setup(void) {
+        int fd, r;
+        bool writable;
+        struct stat st;
+        char id[34]; /* 32 + \n + \0 */
+        mode_t m;
+
+        m = umask(0000);
+
+        /* We create this 0444, to indicate that this isn't really
+         * something you should ever modify. Of course, since the file
+         * will be owned by root it doesn't matter much, but maybe
+         * people look. */
+
+        fd = open("/etc/machine-id", O_RDWR|O_CREAT|O_CLOEXEC|O_NOCTTY, 0444);
+        if (fd >= 0)
+                writable = true;
+        else {
+                fd = open("/etc/machine-id", O_RDONLY|O_CLOEXEC|O_NOCTTY);
+                if (fd < 0) {
+                        umask(m);
+                        log_error("Cannot open /etc/machine-id: %m");
+                        return -errno;
+                }
+
+                writable = false;
+        }
+
+        umask(m);
+
+        if (fstat(fd, &st) < 0) {
+                log_error("fstat() failed: %m");
+                r = -errno;
+                goto finish;
+        }
+
+        if (S_ISREG(st.st_mode)) {
+                if (loop_read(fd, id, 32, false) >= 32) {
+                        r = 0;
+                        goto finish;
+                }
+        }
+
+        /* Hmm, so, the id currently stored is not useful, then let's
+         * generate one */
+
+        r = generate(id);
+        if (r < 0)
+                goto finish;
+
+        if (S_ISREG(st.st_mode) && writable) {
+                lseek(fd, 0, SEEK_SET);
+
+                if (loop_write(fd, id, 33, false) == 33) {
+                        r = 0;
+                        goto finish;
+                }
+        }
+
+        close_nointr_nofail(fd);
+        fd = -1;
+
+        /* Hmm, we couldn't write it? So let's write it to
+         * /run/machine-id as a replacement */
+
+        m = umask(0022);
+        r = write_one_line_file("/run/machine-id", id);
+        umask(m);
+
+        if (r < 0) {
+                log_error("Cannot write /run/machine-id: %s", strerror(-r));
+
+                unlink("/run/machine-id");
+                goto finish;
+        }
+
+        /* And now, let's mount it over */
+        r = mount("/run/machine-id", "/etc/machine-id", "bind", MS_BIND|MS_RDONLY, NULL) < 0 ? -errno : 0;
+        if (r < 0) {
+                unlink("/run/machine-id");
+                log_error("Failed to mount /etc/machine-id: %s", strerror(-r));
+        } else
+                log_info("Installed transient /etc/machine-id file.");
+
+finish:
+
+        if (fd >= 0)
+                close_nointr_nofail(fd);
+
+        return r;
+}
diff --git a/src/core/machine-id-setup.h b/src/core/machine-id-setup.h
new file mode 100644
index 0000000000..16f45d86d3
--- /dev/null
+++ b/src/core/machine-id-setup.h
@@ -0,0 +1,27 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+#ifndef foomachineidsetuphfoo
+#define foomachineidsetuphfoo
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+int machine_id_setup(void);
+
+#endif
diff --git a/src/core/mount-setup.c b/src/core/mount-setup.c
new file mode 100644
index 0000000000..52fe523674
--- /dev/null
+++ b/src/core/mount-setup.c
@@ -0,0 +1,423 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <sys/mount.h>
+#include <errno.h>
+#include <sys/stat.h>
+#include <stdlib.h>
+#include <string.h>
+#include <libgen.h>
+#include <assert.h>
+#include <unistd.h>
+#include <ftw.h>
+
+#include "mount-setup.h"
+#include "log.h"
+#include "macro.h"
+#include "util.h"
+#include "label.h"
+#include "set.h"
+#include "strv.h"
+#include "mkdir.h"
+
+#ifndef TTY_GID
+#define TTY_GID 5
+#endif
+
+typedef struct MountPoint {
+        const char *what;
+        const char *where;
+        const char *type;
+        const char *options;
+        unsigned long flags;
+        bool fatal;
+} 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. */
+#define N_EARLY_MOUNT 4
+
+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 },
+};
+
+/* 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"
+};
+
+bool mount_point_is_api(const char *path) {
+        unsigned i;
+
+        /* Checks if this mount point is considered "API", and hence
+         * should be ignored */
+
+        for (i = 0; i < ELEMENTSOF(mount_table); i ++)
+                if (path_equal(path, mount_table[i].where))
+                        return true;
+
+        return path_startswith(path, "/sys/fs/cgroup/");
+}
+
+bool mount_point_ignore(const char *path) {
+        unsigned i;
+
+        for (i = 0; i < ELEMENTSOF(ignore_paths); i++)
+                if (path_equal(path, ignore_paths[i]))
+                        return true;
+
+        return false;
+}
+
+static int mount_one(const MountPoint *p, bool relabel) {
+        int r;
+
+        assert(p);
+
+        /* Relabel first, just in case */
+        if (relabel)
+                label_fix(p->where, true);
+
+        if ((r = path_is_mount_point(p->where, true)) < 0)
+                return r;
+
+        if (r > 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);
+
+        log_debug("Mounting %s to %s of type %s with options %s.",
+                  p->what,
+                  p->where,
+                  p->type,
+                  strna(p->options));
+
+        if (mount(p->what,
+                  p->where,
+                  p->type,
+                  p->flags,
+                  p->options) < 0) {
+                log_error("Failed to mount %s: %s", p->where, strerror(errno));
+                return p->fatal ? -errno : 0;
+        }
+
+        /* Relabel again, since we now mounted something fresh here */
+        if (relabel)
+                label_fix(p->where, false);
+
+        return 1;
+}
+
+int mount_setup_early(void) {
+        unsigned i;
+        int r = 0;
+
+        assert_cc(N_EARLY_MOUNT <= ELEMENTSOF(mount_table));
+
+        /* Do a minimal mount of /proc and friends to enable the most
+         * basic stuff, such as SELinux */
+        for (i = 0; i < N_EARLY_MOUNT; i ++)  {
+                int j;
+
+                j = mount_one(mount_table + i, false);
+                if (r == 0)
+                        r = j;
+        }
+
+        return r;
+}
+
+int mount_cgroup_controllers(char ***join_controllers) {
+        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;
+        }
+
+        /* Ignore the header line */
+        (void) fgets(buf, sizeof(buf), f);
+
+        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;
+                char ***k = NULL;
+
+                controller = set_steal_first(controllers);
+                if (!controller)
+                        break;
+
+                if (join_controllers)
+                        for (k = join_controllers; *k; k++)
+                                if (strv_find(*k, controller))
+                                        break;
+
+                if (k && *k) {
+                        char **i, **j;
+
+                        for (i = *k, j = *k; *i; i++) {
+
+                                if (!streq(*i, controller)) {
+                                        char *t;
+
+                                        t = set_remove(controllers, *i);
+                                        if (!t) {
+                                                free(*i);
+                                                continue;
+                                        }
+                                        free(t);
+                                }
+
+                                *(j++) = *i;
+                        }
+
+                        *j = NULL;
+
+                        options = strv_join(*k, ",");
+                        if (!options) {
+                                log_error("Failed to join options");
+                                free(controller);
+                                r = -ENOMEM;
+                                goto finish;
+                        }
+
+                } 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;
+                }
+
+                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 && k && *k) {
+                        char **i;
+
+                        for (i = *k; *i; i++) {
+                                char *t;
+
+                                t = strappend("/sys/fs/cgroup/", *i);
+                                if (!t) {
+                                        log_error("Failed to build path");
+                                        r = -ENOMEM;
+                                        free(options);
+                                        goto finish;
+                                }
+
+                                r = symlink(options, t);
+                                free(t);
+
+                                if (r < 0 && errno != EEXIST) {
+                                        log_error("Failed to create symlink: %m");
+                                        r = -errno;
+                                        free(options);
+                                        goto finish;
+                                }
+                        }
+                }
+
+                free(options);
+        }
+
+        r = 0;
+
+finish:
+        set_free_free(controllers);
+
+        fclose(f);
+
+        return r;
+}
+
+static int symlink_and_label(const char *old_path, const char *new_path) {
+        int r;
+
+        assert(old_path);
+        assert(new_path);
+
+        if ((r = label_symlinkfile_set(new_path)) < 0)
+                return r;
+
+        if (symlink(old_path, new_path) < 0)
+                r = -errno;
+
+        label_file_clear();
+
+        return r;
+}
+
+static int nftw_cb(
+                const char *fpath,
+                const struct stat *sb,
+                int tflag,
+                struct FTW *ftwbuf) {
+
+        /* No need to label /dev twice in a row... */
+        if (_unlikely_(ftwbuf->level == 0))
+                return FTW_CONTINUE;
+
+        label_fix(fpath, true);
+
+        /* /run/initramfs is static data and big, no need to
+         * dynamically relabel its contents at boot... */
+        if (_unlikely_(ftwbuf->level == 1 &&
+                      tflag == FTW_D &&
+                      streq(fpath, "/run/initramfs")))
+                return FTW_SKIP_SUBTREE;
+
+        return FTW_CONTINUE;
+};
+
+int mount_setup(bool loaded_policy) {
+
+        static const char symlinks[] =
+                "/proc/kcore\0"      "/dev/core\0"
+                "/proc/self/fd\0"    "/dev/fd\0"
+                "/proc/self/fd/0\0"  "/dev/stdin\0"
+                "/proc/self/fd/1\0"  "/dev/stdout\0"
+                "/proc/self/fd/2\0"  "/dev/stderr\0";
+
+        static const char relabel[] =
+                "/run/initramfs/root-fsck\0"
+                "/run/initramfs/shutdown\0";
+
+        int r;
+        unsigned i;
+        const char *j, *k;
+
+        for (i = 0; i < ELEMENTSOF(mount_table); i ++) {
+                r = mount_one(mount_table + i, true);
+
+                if (r < 0)
+                        return r;
+        }
+
+        /* 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
+         * use the same label for all their files. */
+        if (loaded_policy) {
+                usec_t before_relabel, after_relabel;
+                char timespan[FORMAT_TIMESPAN_MAX];
+
+                before_relabel = now(CLOCK_MONOTONIC);
+
+                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));
+        }
+
+        /* Create a few default symlinks, which are normally created
+         * by udevd, but some scripts might need them before we start
+         * udevd. */
+        NULSTR_FOREACH_PAIR(j, k, symlinks)
+                symlink_and_label(j, k);
+
+        /* Create a few directories we always want around */
+        label_mkdir("/run/systemd", 0755);
+        label_mkdir("/run/systemd/system", 0755);
+
+        return 0;
+}
diff --git a/src/core/mount-setup.h b/src/core/mount-setup.h
new file mode 100644
index 0000000000..720b66f76c
--- /dev/null
+++ b/src/core/mount-setup.h
@@ -0,0 +1,36 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+#ifndef foomountsetuphfoo
+#define foomountsetuphfoo
+
+/***
+  This file is part of systemd.
+
+  Copyright 2010 Lennart Poettering
+
+  systemd is free software; you can redistribute it and/or modify it
+  under the terms of the GNU Lesser General Public License as published by
+  the Free Software Foundation; either version 2.1 of the License, or
+  (at your option) any later version.
+
+  systemd is distributed in the hope that it will be useful, but
+  WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+  Lesser General Public License for more details.
+
+  You should have received a copy of the GNU Lesser General Public License
+  along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stdbool.h>
+
+int mount_setup_early(void);
+
+int mount_setup(bool loaded_policy);
+
+int mount_cgroup_controllers(char ***join_controllers);
+
+bool mount_point_is_api(const char *path);
+bool mount_point_ignore(const char *path);
+
+#endif