From 2450dd660e2c3483e232176a09b0229851522807 Mon Sep 17 00:00:00 2001
From: Lennart Poettering <lennart@poettering.net>
Date: Mon, 13 Aug 2012 15:27:04 +0200
Subject: [PATCH] namespace: rework namespace support
- don't use pivot_root() anymore, just reuse root hierarchy
- first create all mounts, then mark them read-only so that we get the
right behaviour when people want writable mounts inside of
read-only mounts
- don't pass invalid combinations of MS_ constants to the kernel
(cherry picked from commit ac0930c892bc7979b4c9bc2a52e5e844650b025d)
Conflicts:
src/core/execute.c
---
man/systemd.exec.xml | 19 ++---
src/core/execute.c | 3 +-
src/core/namespace.c | 218 +++++++++++++++++++++----------------------------
src/core/switch-root.c | 6 ++
src/shared/util.c | 3 +-
src/test/test-ns.c | 5 +-
6 files changed, 113 insertions(+), 141 deletions(-)
diff --git a/man/systemd.exec.xml b/man/systemd.exec.xml
index 09e9f0f..bd29ddf 100644
--- a/man/systemd.exec.xml
+++ b/man/systemd.exec.xml
@@ -1039,20 +1039,13 @@
<option>shared</option>,
<option>slave</option> or
<option>private</option>, which
- control whether namespaces set up with
- <varname>ReadWriteDirectories=</varname>,
- <varname>ReadOnlyDirectories=</varname>
- and
- <varname>InaccessibleDirectories=</varname>
- receive or propagate new mounts
- from/to the main namespace. See
+ control whether the file system
+ namespace set up for this unit's
+ processes will receive or propagate
+ new mounts. See
<citerefentry><refentrytitle>mount</refentrytitle><manvolnum>1</manvolnum></citerefentry>
- for details. Defaults to
- <option>shared</option>, i.e. the new
- namespace will both receive new mount
- points from the main namespace as well
- as propagate new mounts to
- it.</para></listitem>
+ for details. Default to
+ <option>shared</option>.</para></listitem>
</varlistentry>
<varlistentry>
diff --git a/src/core/execute.c b/src/core/execute.c
index 8a04fe6..931a9b0 100644
--- a/src/core/execute.c
+++ b/src/core/execute.c
@@ -1246,7 +1246,7 @@ int exec_spawn(ExecCommand *command,
if (strv_length(context->read_write_dirs) > 0 ||
strv_length(context->read_only_dirs) > 0 ||
strv_length(context->inaccessible_dirs) > 0 ||
- context->mount_flags != MS_SHARED ||
+ context->mount_flags != 0 ||
context->private_tmp) {
err = setup_namespace(context->read_write_dirs,
context->read_only_dirs,
@@ -1467,7 +1467,6 @@ void exec_context_init(ExecContext *c) {
c->cpu_sched_policy = SCHED_OTHER;
c->syslog_priority = LOG_DAEMON|LOG_INFO;
c->syslog_level_prefix = true;
- c->mount_flags = MS_SHARED;
c->kill_signal = SIGTERM;
c->send_sigkill = true;
c->control_group_persistent = -1;
diff --git a/src/core/namespace.c b/src/core/namespace.c
index 0148f5f..d163762 100644
--- a/src/core/namespace.c
+++ b/src/core/namespace.c
@@ -41,13 +41,15 @@ typedef enum PathMode {
/* This is ordered by priority! */
INACCESSIBLE,
READONLY,
- PRIVATE,
+ PRIVATE_TMP,
+ PRIVATE_VAR_TMP,
READWRITE
} PathMode;
typedef struct Path {
const char *path;
PathMode mode;
+ bool done;
} Path;
static int append_paths(Path **p, char **strv, PathMode mode) {
@@ -91,25 +93,22 @@ static int path_compare(const void *a, const void *b) {
return 0;
}
-static void drop_duplicates(Path *p, unsigned *n, bool *need_inaccessible, bool *need_private) {
+static void drop_duplicates(Path *p, unsigned *n, bool *need_inaccessible) {
Path *f, *t, *previous;
assert(p);
assert(n);
assert(need_inaccessible);
- assert(need_private);
for (f = p, t = p, previous = NULL; f < p+*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 == PRIVATE)
- *need_private = true;
-
if (t->mode == INACCESSIBLE)
*need_inaccessible = true;
@@ -121,65 +120,62 @@ static void drop_duplicates(Path *p, unsigned *n, bool *need_inaccessible, bool
*n = t - p;
}
-static int apply_mount(Path *p, const char *root_dir, const char *inaccessible_dir, const char *private_dir, unsigned long flags) {
+static int apply_mount(
+ Path *p,
+ const char *tmp_dir,
+ const char *var_tmp_dir,
+ const char *inaccessible_dir) {
+
const char *what;
- char *where;
int r;
assert(p);
- assert(root_dir);
- assert(inaccessible_dir);
- assert(private_dir);
-
- where = strappend(root_dir, p->path);
- if (!where)
- return -ENOMEM;
switch (p->mode) {
case INACCESSIBLE:
what = inaccessible_dir;
- flags |= MS_RDONLY;
break;
case READONLY:
- flags |= MS_RDONLY;
- /* Fall through */
-
case READWRITE:
what = p->path;
break;
- case PRIVATE:
- what = private_dir;
+ case PRIVATE_TMP:
+ what = tmp_dir;
+ break;
+
+ case PRIVATE_VAR_TMP:
+ what = var_tmp_dir;
break;
default:
assert_not_reached("Unknown mode");
}
- r = mount(what, where, NULL, MS_BIND|MS_REC, NULL);
- if (r >= 0) {
- log_debug("Successfully mounted %s to %s", what, where);
+ assert(what);
- /* The bind mount will always inherit the original
- * flags. If we want to set any flag we need
- * to do so in a second independent step. */
- if (flags)
- r = mount(NULL, where, NULL, MS_REMOUNT|MS_BIND|MS_REC|flags, NULL);
+ r = mount(what, p->path, NULL, MS_BIND, NULL);
+ if (r >= 0)
+ log_debug("Successfully mounted %s to %s", what, p->path);
- /* Avoid exponential growth of trees */
- if (r >= 0 && path_equal(p->path, "/"))
- r = mount(NULL, where, NULL, MS_REMOUNT|MS_BIND|flags, NULL);
+ return r;
+}
- if (r < 0) {
- r = -errno;
- umount2(where, MNT_DETACH);
- }
- }
+static int make_read_only(Path *p) {
+ int r;
- free(where);
- return r;
+ assert(p);
+
+ if (p->mode != INACCESSIBLE && p->mode != READONLY)
+ return 0;
+
+ r = mount(NULL, p->path, NULL, MS_BIND|MS_REMOUNT|MS_RDONLY, NULL);
+ if (r < 0)
+ return -errno;
+
+ return 0;
}
int setup_namespace(
@@ -190,30 +186,26 @@ int setup_namespace(
unsigned long flags) {
char
- tmp_dir[] = "/tmp/systemd-namespace-XXXXXX",
- root_dir[] = "/tmp/systemd-namespace-XXXXXX/root",
- old_root_dir[] = "/tmp/systemd-namespace-XXXXXX/root/tmp/old-root-XXXXXX",
- inaccessible_dir[] = "/tmp/systemd-namespace-XXXXXX/inaccessible",
- private_dir[] = "/tmp/systemd-namespace-XXXXXX/private";
+ 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_private = false, need_inaccessible = false;
- bool remove_tmp = false, remove_root = false, remove_old_root = false, remove_inaccessible = false, remove_private = false;
+ bool need_inaccessible = false;
+ bool remove_tmp = false, remove_var_tmp = false, remove_inaccessible = false;
int r;
- const char *t;
+
+ if (!flags)
+ flags = MS_SHARED;
n =
strv_length(writable) +
strv_length(readable) +
strv_length(inaccessible) +
- (private_tmp ? 3 : 1);
-
- paths = new(Path, n);
- if (!paths)
- return -ENOMEM;
+ (private_tmp ? 2 : 0);
- p = paths;
+ 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)
@@ -221,60 +213,70 @@ int setup_namespace(
if (private_tmp) {
p->path = "/tmp";
- p->mode = PRIVATE;
+ p->mode = PRIVATE_TMP;
p++;
p->path = "/var/tmp";
- p->mode = PRIVATE;
+ p->mode = PRIVATE_VAR_TMP;
p++;
}
- p->path = "/";
- p->mode = READWRITE;
- p++;
-
assert(paths + n == p);
qsort(paths, n, sizeof(Path), path_compare);
- drop_duplicates(paths, &n, &need_inaccessible, &need_private);
+ drop_duplicates(paths, &n, &need_inaccessible);
- if (!mkdtemp(tmp_dir)) {
- r = -errno;
- goto fail;
- }
- remove_tmp = true;
+ if (need_inaccessible) {
+ mode_t u;
+ char *d;
- memcpy(root_dir, tmp_dir, sizeof(tmp_dir)-1);
- if (mkdir(root_dir, 0777) < 0) {
- r = -errno;
- goto fail;
- }
- remove_root = true;
+ u = umask(0777);
+ d = mkdtemp(inaccessible_dir);
+ umask(u);
- if (need_inaccessible) {
- memcpy(inaccessible_dir, tmp_dir, sizeof(tmp_dir)-1);
- if (mkdir(inaccessible_dir, 0) < 0) {
+ if (!d) {
r = -errno;
goto fail;
}
+
remove_inaccessible = true;
}
- if (need_private) {
+ if (private_tmp) {
mode_t u;
-
- memcpy(private_dir, tmp_dir, sizeof(tmp_dir)-1);
+ char *d;
u = umask(0000);
- if (mkdir(private_dir, 0777 + S_ISVTX) < 0) {
- umask(u);
+ d = mkdtemp(tmp_dir);
+ umask(u);
+ if (!d) {
r = -errno;
goto fail;
}
+ remove_tmp = true;
+
+ u = umask(0000);
+ d = mkdtemp(var_tmp_dir);
umask(u);
- remove_private = true;
+
+ if (!d) {
+ r = -errno;
+ goto fail;
+ }
+
+ remove_var_tmp = true;
+
+ if (chmod(tmp_dir, 0777 + S_ISVTX) < 0) {
+ r = -errno;
+ goto fail;
+ }
+
+ if (chmod(var_tmp_dir, 0777 + S_ISVTX) < 0) {
+ r = -errno;
+ goto fail;
+ }
}
if (unshare(CLONE_NEWNS) < 0) {
@@ -282,7 +284,7 @@ int setup_namespace(
goto fail;
}
- /* Remount / as SLAVE so that nothing mounted in 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;
@@ -290,69 +292,39 @@ int setup_namespace(
}
for (p = paths; p < paths + n; p++) {
- r = apply_mount(p, root_dir, inaccessible_dir, private_dir, flags);
+ r = apply_mount(p, tmp_dir, var_tmp_dir, inaccessible_dir);
if (r < 0)
goto undo_mounts;
}
- memcpy(old_root_dir, tmp_dir, sizeof(tmp_dir)-1);
- if (!mkdtemp(old_root_dir)) {
- r = -errno;
- goto undo_mounts;
- }
- remove_old_root = true;
-
- if (chdir(root_dir) < 0) {
- r = -errno;
- goto undo_mounts;
+ for (p = paths; p < paths + n; p++) {
+ r = make_read_only(p);
+ if (r < 0)
+ goto undo_mounts;
}
- if (pivot_root(root_dir, old_root_dir) < 0) {
+ /* Remount / as the desired mode */
+ if (mount(NULL, "/", NULL, flags|MS_REC, NULL) < 0) {
r = -errno;
goto undo_mounts;
}
- free(paths);
-
- t = old_root_dir + sizeof(root_dir) - 1;
- if (umount2(t, MNT_DETACH) < 0)
- /* At this point it's too late to turn anything back,
- * since we are already in the new root. */
- return -errno;
-
- if (rmdir(t) < 0)
- return -errno;
-
return 0;
undo_mounts:
-
- for (p--; p >= paths; p--) {
- char full_path[PATH_MAX];
-
- snprintf(full_path, sizeof(full_path), "%s%s", root_dir, p->path);
- char_array_0(full_path);
-
- umount2(full_path, MNT_DETACH);
- }
+ for (p = paths; p < paths + n; p++)
+ if (p->done)
+ umount2(p->path, MNT_DETACH);
fail:
- if (remove_old_root)
- rmdir(old_root_dir);
-
if (remove_inaccessible)
rmdir(inaccessible_dir);
- if (remove_private)
- rmdir(private_dir);
-
- if (remove_root)
- rmdir(root_dir);
-
if (remove_tmp)
rmdir(tmp_dir);
- free(paths);
+ if (remove_var_tmp)
+ rmdir(var_tmp_dir);
return r;
}
diff --git a/src/core/switch-root.c b/src/core/switch-root.c
index efc7d34..150332a 100644
--- a/src/core/switch-root.c
+++ b/src/core/switch-root.c
@@ -115,6 +115,12 @@ int switch_root(const char *new_root) {
goto fail;
}
+ if (chdir("/") < 0) {
+ r = -errno;
+ log_error("Failed to change directory: %m");
+ goto fail;
+ }
+
if (old_root_fd >= 0) {
struct stat rb;
diff --git a/src/shared/util.c b/src/shared/util.c
index 0d48691..faf428f 100644
--- a/src/shared/util.c
+++ b/src/shared/util.c
@@ -3006,7 +3006,8 @@ unsigned long long random_ull(void) {
uint64_t ull;
ssize_t r;
- if ((fd = open("/dev/urandom", O_RDONLY|O_CLOEXEC|O_NOCTTY)) < 0)
+ fd = open("/dev/urandom", O_RDONLY|O_CLOEXEC|O_NOCTTY);
+ if (fd < 0)
goto fallback;
r = loop_read(fd, &ull, sizeof(ull), true);
diff --git a/src/test/test-ns.c b/src/test/test-ns.c
index e2bdfc5..a2724b4 100644
--- a/src/test/test-ns.c
+++ b/src/test/test-ns.c
@@ -34,7 +34,7 @@ int main(int argc, char *argv[]) {
NULL
};
- const char * const readable[] = {
+ const char * const readonly[] = {
"/",
"/usr",
"/boot",
@@ -48,7 +48,8 @@ int main(int argc, char *argv[]) {
int r;
- if ((r = setup_namespace((char**) writable, (char**) readable, (char**) inaccessible, true, MS_SHARED)) < 0) {
+ r = setup_namespace((char**) writable, (char**) readonly, (char**) inaccessible, true, 0);
+ if (r < 0) {
log_error("Failed to setup namespace: %s", strerror(-r));
return 1;
}