shaba/systemd
1
0
Fork 0
mirror of https://github.com/systemd/systemd.git synced 2025-03-19 22:50:17 +03:00
Code

Merge pull request #4962 from poettering/root-directory-2

Browse Source 
Add new MountAPIVFS= boolean unit file setting + RootImage=
This commit is contained in:
Zbigniew Jędrzejewski-Szmek 2017-02-08 23:05:05 -05:00 committed by GitHub
commit fc6149a6ce
17 changed files with 405 additions and 139 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
TODO
View File

@ -122,8 +122,6 @@ Features:
* switch to ProtectSystem=strict for all our long-running services where that's possible
* If RootDirectory= is used, mount /proc, /sys, /dev into it, if not mounted yet
* Permit masking specific netlink APIs with RestrictAddressFamily=
* nspawn: start UID allocation loop from hash of container name
@ -153,8 +151,6 @@ Features:
* Add DataDirectory=, CacheDirectory= and LogDirectory= to match
RuntimeDirectory=, and create it as necessary when starting a service, owned by the right user.
* Add RootImage= for mounting a disk image or file as root directory
* make sure the ratelimit object can deal with USEC_INFINITY as way to turn off things
* journalctl: make sure -f ends when the container indicated by -M terminates

12
man/systemd-nspawn.xml
View File

@ -256,10 +256,14 @@
<listitem><para>Takes a data integrity (dm-verity) root hash specified in hexadecimal. This option enables data
integrity checks using dm-verity, if the used image contains the appropriate integrity data (see above). The
specified hash must match the root hash of integrity data, and is usually at least 256bits (and hence 64
hexadecimal characters) long (in case of SHA256 for example). If this option is not specified, but a file with
the <filename>.roothash</filename> suffix is found next to the image file, bearing otherwise the same name the
root hash is read from it and automatically used.</para></listitem>
specified hash must match the root hash of integrity data, and is usually at least 256 bits (and hence 64
formatted hexadecimal characters) long (in case of SHA256 for example). If this option is not specified, but
the image file carries the <literal>user.verity.roothash</literal> extended file attribute (see <citerefentry
project='man-pages'><refentrytitle>xattr</refentrytitle><manvolnum>7</manvolnum></citerefentry>), then the root
hash is read from it, also as formatted hexadecimal characters. If the extended file attribute is not found (or
is not supported by the underlying file system), but a file with the <filename>.roothash</filename> suffix is
found next to the image file, bearing otherwise the same name, the root hash is read from it and automatically
used, also as formatted hexadecimal characters.</para></listitem>
</varlistentry>
<varlistentry>

69
man/systemd.exec.xml
View File

@ -86,12 +86,10 @@
<para>A few execution parameters result in additional, automatic
dependencies to be added.</para>
<para>Units with <varname>WorkingDirectory=</varname> or
<varname>RootDirectory=</varname> set automatically gain
dependencies of type <varname>Requires=</varname> and
<varname>After=</varname> on all mount units required to access
the specified paths. This is equivalent to having them listed
explicitly in <varname>RequiresMountsFor=</varname>.</para>
<para>Units with <varname>WorkingDirectory=</varname>, <varname>RootDirectory=</varname> or
<varname>RootImage=</varname> set automatically gain dependencies of type <varname>Requires=</varname> and
<varname>After=</varname> on all mount units required to access the specified paths. This is equivalent to having
them listed explicitly in <varname>RequiresMountsFor=</varname>.</para>
<para>Similar, units with <varname>PrivateTmp=</varname> enabled automatically get mount unit dependencies for all
mounts required to access <filename>/tmp</filename> and <filename>/var/tmp</filename>. They will also gain an
@ -117,9 +115,10 @@
<varname>User=</varname> is used. If not set, defaults to the root directory when systemd is running as a
system instance and the respective user's home directory if run as user. If the setting is prefixed with the
<literal>-</literal> character, a missing working directory is not considered fatal. If
<varname>RootDirectory=</varname> is not set, then <varname>WorkingDirectory=</varname> is relative to the root
of the system running the service manager. Note that setting this parameter might result in additional
dependencies to be added to the unit (see above).</para></listitem>
<varname>RootDirectory=</varname>/<varname>RootImage=</varname> is not set, then
<varname>WorkingDirectory=</varname> is relative to the root of the system running the service manager. Note
that setting this parameter might result in additional dependencies to be added to the unit (see
above).</para></listitem>
</varlistentry>
<varlistentry>
@ -132,8 +131,33 @@
the <function>chroot()</function> jail. Note that setting this parameter might result in additional
dependencies to be added to the unit (see above).</para>
<para>The <varname>PrivateUsers=</varname> setting is particularly useful in conjunction with
<varname>RootDirectory=</varname>. For details, see below.</para></listitem>
<para>The <varname>MountAPIVFS=</varname> and <varname>PrivateUsers=</varname> settings are particularly useful
in conjunction with <varname>RootDirectory=</varname>. For details, see below.</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>RootImage=</varname></term>
<listitem><para>Takes a path to a block device node or regular file as argument. This call is similar to
<varname>RootDirectory=</varname> however mounts a file system hierarchy from a block device node or loopack
file instead of a directory. The device node or file system image file needs to contain a file system without a
partition table, or a file system within an MBR/MS-DOS or GPT partition table with only a single
Linux-compatible partition, or a set of file systems within a GPT partition table that follows the <ulink
url="http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec/">Discoverable Partitions
Specification</ulink>.</para></listitem>
</varlistentry>
<varlistentry>
<term><varname>MountAPIVFS=</varname></term>
<listitem><para>Takes a boolean argument. If on, a private mount namespace for the unit's processes is created
and the API file systems <filename>/proc</filename>, <filename>/sys</filename>, and <filename>/dev</filename>
are mounted inside of it, unless they are already mounted. Note that this option has no effect unless used in
conjunction with <varname>RootDirectory=</varname>/<varname>RootImage=</varname> as these three mounts are
generally mounted in the host anyway, and unless the root directory is changed, the private mount namespace
will be a 1:1 copy of the host's, and include these three mounts. Note that the <filename>/dev</filename> file
system of the host is bind mounted if this option is used without <varname>PrivateDevices=</varname>. To run
the service with a private, minimal version of <filename>/dev/</filename>, combine this option with
<varname>PrivateDevices=</varname>.</para></listitem>
</varlistentry>
<varlistentry>
@ -938,7 +962,7 @@
access a process might have to the file system hierarchy. Each setting takes a space-separated list of paths
relative to the host's root directory (i.e. the system running the service manager). Note that if paths
contain symlinks, they are resolved relative to the root directory set with
<varname>RootDirectory=</varname>.</para>
<varname>RootDirectory=</varname>/<varname>RootImage=</varname>.</para>
<para>Paths listed in <varname>ReadWritePaths=</varname> are accessible from within the namespace with the same
access modes as from outside of it. Paths listed in <varname>ReadOnlyPaths=</varname> are accessible for
@ -957,9 +981,10 @@
<para>Paths in <varname>ReadWritePaths=</varname>, <varname>ReadOnlyPaths=</varname> and
<varname>InaccessiblePaths=</varname> may be prefixed with <literal>-</literal>, in which case they will be
ignored when they do not exist. If prefixed with <literal>+</literal> the paths are taken relative to the root
directory of the unit, as configured with <varname>RootDirectory=</varname>, instead of relative to the root
directory of the host (see above). When combining <literal>-</literal> and <literal>+</literal> on the same
path make sure to specify <literal>-</literal> first, and <literal>+</literal> second.</para>
directory of the unit, as configured with <varname>RootDirectory=</varname>/<varname>RootImage=</varname>,
instead of relative to the root directory of the host (see above). When combining <literal>-</literal> and
<literal>+</literal> on the same path make sure to specify <literal>-</literal> first, and <literal>+</literal>
second.</para>
<para>Note that using this setting will disconnect propagation of mounts from the service to the host
(propagation in the opposite direction continues to work). This means that this setting may not be used for
@ -990,9 +1015,9 @@
that in this case both read-only and regular bind mounts are reset, regardless which of the two settings is
used.</para>
<para>This option is particularly useful when <varname>RootDirectory=</varname> is used. In this case the
source path refers to a path on the host file system, while the destination path refers to a path below the
root directory of the unit.</para></listitem>
<para>This option is particularly useful when <varname>RootDirectory=</varname>/<varname>RootImage=</varname>
is used. In this case the source path refers to a path on the host file system, while the destination path
refers to a path below the root directory of the unit.</para></listitem>
</varlistentry>
<varlistentry>
@ -1080,10 +1105,10 @@
such as <varname>CapabilityBoundingSet=</varname> will affect only the latter, and there's no way to acquire
additional capabilities in the host's user namespace. Defaults to off.</para>
<para>This setting is particularly useful in conjunction with <varname>RootDirectory=</varname>, as the need to
synchronize the user and group databases in the root directory and on the host is reduced, as the only users
and groups who need to be matched are <literal>root</literal>, <literal>nobody</literal> and the unit's own
user and group.</para></listitem>
<para>This setting is particularly useful in conjunction with
<varname>RootDirectory=</varname>/<varname>RootImage=</varname>, as the need to synchronize the user and group
databases in the root directory and on the host is reduced, as the only users and groups who need to be matched
are <literal>root</literal>, <literal>nobody</literal> and the unit's own user and group.</para></listitem>
</varlistentry>
<varlistentry>

22
src/core/dbus-execute.c
View File

@ -758,6 +758,7 @@ const sd_bus_vtable bus_exec_vtable[] = {
SD_BUS_PROPERTY("LimitRTTIMESoft", "t", bus_property_get_rlimit, offsetof(ExecContext, rlimit[RLIMIT_RTTIME]), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("WorkingDirectory", "s", property_get_working_directory, 0, SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("RootDirectory", "s", NULL, offsetof(ExecContext, root_directory), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("RootImage", "s", NULL, offsetof(ExecContext, root_image), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("OOMScoreAdjust", "i", property_get_oom_score_adjust, 0, SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("Nice", "i", property_get_nice, 0, SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("IOScheduling", "i", property_get_ioprio, 0, SD_BUS_VTABLE_PROPERTY_CONST),
@ -828,6 +829,7 @@ const sd_bus_vtable bus_exec_vtable[] = {
SD_BUS_PROPERTY("RestrictNamespaces", "t", bus_property_get_ulong, offsetof(ExecContext, restrict_namespaces), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("BindPaths", "a(ssbt)", property_get_bind_paths, 0, SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("BindReadOnlyPaths", "a(ssbt)", property_get_bind_paths, 0, SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("MountAPIVFS", "b", bus_property_get_bool, offsetof(ExecContext, mount_apivfs), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_VTABLE_END
};
@ -1047,7 +1049,7 @@ int bus_exec_context_set_transient_property(
return 1;
} else if (STR_IN_SET(name, "TTYPath", "RootDirectory")) {
} else if (STR_IN_SET(name, "TTYPath", "RootDirectory", "RootImage")) {
const char *s;
r = sd_bus_message_read(message, "s", &s);
@ -1060,6 +1062,8 @@ int bus_exec_context_set_transient_property(
if (mode != UNIT_CHECK) {
if (streq(name, "TTYPath"))
r = free_and_strdup(&c->tty_path, s);
else if (streq(name, "RootImage"))
r = free_and_strdup(&c->root_image, s);
else {
assert(streq(name, "RootDirectory"));
r = free_and_strdup(&c->root_directory, s);
@ -1207,7 +1211,7 @@ int bus_exec_context_set_transient_property(
"PrivateTmp", "PrivateDevices", "PrivateNetwork", "PrivateUsers",
"NoNewPrivileges", "SyslogLevelPrefix", "MemoryDenyWriteExecute",
"RestrictRealtime", "DynamicUser", "RemoveIPC", "ProtectKernelTunables",
"ProtectKernelModules", "ProtectControlGroups")) {
"ProtectKernelModules", "ProtectControlGroups", "MountAPIVFS")) {
int b;
r = sd_bus_message_read(message, "b", &b);
@ -1247,6 +1251,8 @@ int bus_exec_context_set_transient_property(
c->protect_kernel_modules = b;
else if (streq(name, "ProtectControlGroups"))
c->protect_control_groups = b;
else if (streq(name, "MountAPIVFS"))
c->mount_apivfs = b;
unit_write_drop_in_private_format(u, mode, name, "%s=%s", name, yes_no(b));
}
@ -1495,12 +1501,15 @@ int bus_exec_context_set_transient_property(
return r;
STRV_FOREACH(p, l) {
int offset;
if (!utf8_is_valid(*p))
const char *i = *p;
size_t offset;
if (!utf8_is_valid(i))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid %s", name);
offset = **p == '-';
if (!path_is_absolute(*p + offset))
offset = i[0] == '-';
offset += i[offset] == '+';
if (!path_is_absolute(i + offset))
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid %s", name);
}
@ -1519,7 +1528,6 @@ int bus_exec_context_set_transient_property(
unit_write_drop_in_private_format(u, mode, name, "%s=", name);
} else {
r = strv_extend_strv(dirs, l, true);
if (r < 0)
return -ENOMEM;

39
src/core/execute.c
View File

@ -1640,6 +1640,9 @@ static bool exec_needs_mount_namespace(
assert(context);
assert(params);
if (context->root_image)
return true;
if (!strv_isempty(context->read_write_paths) ||
!strv_isempty(context->read_only_paths) ||
!strv_isempty(context->inaccessible_paths))
@ -1662,6 +1665,9 @@ static bool exec_needs_mount_namespace(
context->protect_control_groups)
return true;
if (context->mount_apivfs)
return true;
return false;
}
@ -1935,13 +1941,14 @@ static int apply_mount_namespace(Unit *u, const ExecContext *context,
int r;
_cleanup_strv_free_ char **rw = NULL;
char *tmp = NULL, *var = NULL;
const char *root_dir = NULL;
const char *root_dir = NULL, *root_image = NULL;
NameSpaceInfo ns_info = {
.ignore_protect_paths = false,
.private_dev = context->private_devices,
.protect_control_groups = context->protect_control_groups,
.protect_kernel_tunables = context->protect_kernel_tunables,
.protect_kernel_modules = context->protect_kernel_modules,
.mount_apivfs = context->mount_apivfs,
};
assert(context);
@ -1961,8 +1968,12 @@ static int apply_mount_namespace(Unit *u, const ExecContext *context,
if (r < 0)
return r;
if (params->flags & EXEC_APPLY_CHROOT)
root_dir = context->root_directory;
if (params->flags & EXEC_APPLY_CHROOT) {
root_image = context->root_image;
if (!root_image)
root_dir = context->root_directory;
}
/*
* If DynamicUser=no and RootDirectory= is set then lets pass a relaxed
@ -1972,7 +1983,8 @@ static int apply_mount_namespace(Unit *u, const ExecContext *context,
if (!context->dynamic_user && root_dir)
ns_info.ignore_protect_paths = true;
r = setup_namespace(root_dir, &ns_info, rw,
r = setup_namespace(root_dir, root_image,
&ns_info, rw,
context->read_only_paths,
context->inaccessible_paths,
context->bind_mounts,
@ -1981,7 +1993,8 @@ static int apply_mount_namespace(Unit *u, const ExecContext *context,
var,
context->protect_home,
context->protect_system,
context->mount_flags);
context->mount_flags,
DISSECT_IMAGE_DISCARD_ON_LOOP);
/* If we couldn't set up the namespace this is probably due to a
* missing capability. In this case, silently proceeed. */
@ -1995,10 +2008,12 @@ static int apply_mount_namespace(Unit *u, const ExecContext *context,
return r;
}
static int apply_working_directory(const ExecContext *context,
const ExecParameters *params,
const char *home,
const bool needs_mount_ns) {
static int apply_working_directory(
const ExecContext *context,
const ExecParameters *params,
const char *home,
const bool needs_mount_ns) {
const char *d;
const char *wd;
@ -2979,6 +2994,7 @@ void exec_context_done(ExecContext *c) {
c->working_directory = mfree(c->working_directory);
c->root_directory = mfree(c->root_directory);
c->root_image = mfree(c->root_image);
c->tty_path = mfree(c->tty_path);
c->syslog_identifier = mfree(c->syslog_identifier);
c->user = mfree(c->user);
@ -3294,6 +3310,7 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) {
"%sPrivateUsers: %s\n"
"%sProtectHome: %s\n"
"%sProtectSystem: %s\n"
"%sMountAPIVFS: %s\n"
"%sIgnoreSIGPIPE: %s\n"
"%sMemoryDenyWriteExecute: %s\n"
"%sRestrictRealtime: %s\n",
@ -3310,10 +3327,14 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) {
prefix, yes_no(c->private_users),
prefix, protect_home_to_string(c->protect_home),
prefix, protect_system_to_string(c->protect_system),
prefix, yes_no(c->mount_apivfs),
prefix, yes_no(c->ignore_sigpipe),
prefix, yes_no(c->memory_deny_write_execute),
prefix, yes_no(c->restrict_realtime));
if (c->root_image)
fprintf(f, "%sRootImage: %s\n", prefix, c->root_image);
STRV_FOREACH(e, c->environment)
fprintf(f, "%sEnvironment: %s\n", prefix, *e);

3
src/core/execute.h
View File

@ -106,7 +106,7 @@ struct ExecContext {
char **pass_environment;
struct rlimit *rlimit[_RLIMIT_MAX];
char *working_directory, *root_directory;
char *working_directory, *root_directory, *root_image;
bool working_directory_missing_ok;
bool working_directory_home;
@ -183,6 +183,7 @@ struct ExecContext {
bool protect_kernel_tunables;
bool protect_kernel_modules;
bool protect_control_groups;
bool mount_apivfs;
bool no_new_privileges;

2
src/core/load-fragment-gperf.gperf.m4
View File

@ -19,6 +19,7 @@ m4_dnl Define the context options only once
m4_define(`EXEC_CONTEXT_CONFIG_ITEMS',
`$1.WorkingDirectory, config_parse_working_directory, 0, offsetof($1, exec_context)
$1.RootDirectory, config_parse_unit_path_printf, 0, offsetof($1, exec_context.root_directory)
$1.RootImage, config_parse_unit_path_printf, 0, offsetof($1, exec_context.root_image)
$1.User, config_parse_user_group, 0, offsetof($1, exec_context.user)
$1.Group, config_parse_user_group, 0, offsetof($1, exec_context.group)
$1.SupplementaryGroups, config_parse_user_group_strv, 0, offsetof($1, exec_context.supplementary_groups)
@ -101,6 +102,7 @@ $1.PrivateUsers, config_parse_bool, 0,
$1.ProtectSystem, config_parse_protect_system, 0, offsetof($1, exec_context)
$1.ProtectHome, config_parse_protect_home, 0, offsetof($1, exec_context)
$1.MountFlags, config_parse_exec_mount_flags, 0, offsetof($1, exec_context)
$1.MountAPIVFS, config_parse_bool, 0, offsetof($1, exec_context.mount_apivfs)
$1.Personality, config_parse_personality, 0, offsetof($1, exec_context.personality)
$1.RuntimeDirectoryMode, config_parse_mode, 0, offsetof($1, exec_context.runtime_directory_mode)
$1.RuntimeDirectory, config_parse_runtime_directory, 0, offsetof($1, exec_context.runtime_directory)

19
src/core/load-fragment.c
View File

@ -3839,7 +3839,8 @@ int config_parse_namespace_path_strv(
cur = rvalue;
for (;;) {
_cleanup_free_ char *word = NULL, *resolved = NULL, *joined = NULL;
bool ignore_enoent;
const char *w;
bool ignore_enoent = false, shall_prefix = false;
r = extract_first_word(&cur, &word, NULL, EXTRACT_QUOTES);
if (r == 0)
@ -3856,9 +3857,17 @@ int config_parse_namespace_path_strv(
continue;
}
ignore_enoent = word[0] == '-';
w = word;
if (startswith(w, "-")) {
ignore_enoent = true;
w++;
}
if (startswith(w, "+")) {
shall_prefix = true;
w++;
}
r = unit_full_printf(u, word + ignore_enoent, &resolved);
r = unit_full_printf(u, w, &resolved);
if (r < 0) {
log_syntax(unit, LOG_ERR, filename, line, r, "Failed to resolve specifiers in %s: %m", word);
continue;
@ -3871,7 +3880,9 @@ int config_parse_namespace_path_strv(
path_kill_slashes(resolved);
joined = strjoin(ignore_enoent ? "-" : "", resolved);
joined = strjoin(ignore_enoent ? "-" : "",
shall_prefix ? "+" : "",
resolved);
r = strv_push(sv, joined);
if (r < 0)

187
src/core/namespace.c
View File

@ -30,6 +30,7 @@
#include "dev-setup.h"
#include "fd-util.h"
#include "fs-util.h"
#include "loop-util.h"
#include "loopback-setup.h"
#include "missing.h"
#include "mkdir.h"
@ -52,10 +53,13 @@ typedef enum MountMode {
INACCESSIBLE,
BIND_MOUNT,
BIND_MOUNT_RECURSIVE,
READONLY,
PRIVATE_TMP,
PRIVATE_VAR_TMP,
PRIVATE_DEV,
BIND_DEV,
SYSFS,
PROCFS,
READONLY,
READWRITE,
} MountMode;
@ -70,13 +74,13 @@ typedef struct MountEntry {
char *source_malloc;
} MountEntry;
/*
* The following Protect tables are to protect paths and mark some of them
* READONLY, in case a path is covered by an option from another table, then
* it is marked READWRITE in the current one, and the more restrictive mode is
* applied from that other table. This way all options can be combined in a
* safe and comprehensible way for users.
*/
/* If MountAPIVFS= is used, let's mount /sys and /proc into the it, but only as a fallback if the user hasn't mounted
* something there already. These mounts are hence overriden by any other explicitly configured mounts. */
static const MountEntry apivfs_table[] = {
{ "/proc", PROCFS, false },
{ "/dev", BIND_DEV, false },
{ "/sys", SYSFS, false },
};
/* ProtectKernelTunables= option and the related filesystem APIs */
static const MountEntry protect_kernel_tunables_table[] = {
@ -176,6 +180,13 @@ static const char *mount_entry_source(const MountEntry *p) {
return p->source_malloc ?: p->source_const;
}
static void mount_entry_done(MountEntry *p) {
assert(p);
p->path_malloc = mfree(p->path_malloc);
p->source_malloc = mfree(p->source_malloc);
}
static int append_access_mounts(MountEntry **p, char **strv, MountMode mode) {
char **i;
@ -351,7 +362,7 @@ static void drop_duplicates(MountEntry *m, unsigned *n) {
if (previous && path_equal(mount_entry_path(f), mount_entry_path(previous))) {
log_debug("%s is duplicate.", mount_entry_path(f));
previous->read_only = previous->read_only || mount_entry_read_only(f); /* Propagate the read-only flag to the remaining entry */
f->path_malloc = mfree(f->path_malloc);
mount_entry_done(f);
continue;
}
@ -379,7 +390,7 @@ static void drop_inaccessible(MountEntry *m, unsigned *n) {
* it, as inaccessible paths really should drop the entire subtree. */
if (clear && path_startswith(mount_entry_path(f), clear)) {
log_debug("%s is masked by %s.", mount_entry_path(f), clear);
f->path_malloc = mfree(f->path_malloc);
mount_entry_done(f);
continue;
}
@ -419,7 +430,7 @@ static void drop_nop(MountEntry *m, unsigned *n) {
/* We found it, let's see if it's the same mode, if so, we can drop this entry */
if (found && p->mode == f->mode) {
log_debug("%s is redundant by %s", mount_entry_path(f), mount_entry_path(p));
f->path_malloc = mfree(f->path_malloc);
mount_entry_done(f);
continue;
}
}
@ -447,7 +458,7 @@ static void drop_outside_root(const char *root_directory, MountEntry *m, unsigne
if (!path_startswith(mount_entry_path(f), root_directory)) {
log_debug("%s is outside of root directory.", mount_entry_path(f));
f->path_malloc = mfree(f->path_malloc);
mount_entry_done(f);
continue;
}
@ -458,7 +469,7 @@ static void drop_outside_root(const char *root_directory, MountEntry *m, unsigne
*n = t - m;
}
static int mount_dev(MountEntry *m) {
static int mount_private_dev(MountEntry *m) {
static const char devnodes[] =
"/dev/null\0"
"/dev/zero\0"
@ -597,6 +608,62 @@ fail:
return r;
}
static int mount_bind_dev(MountEntry *m) {
int r;
assert(m);
/* Implements the little brother of mount_private_dev(): simply bind mounts the host's /dev into the service's
* /dev. This is only used when RootDirectory= is set. */
r = path_is_mount_point(mount_entry_path(m), NULL, 0);
if (r < 0)
return log_debug_errno(r, "Unable to determine whether /dev is already mounted: %m");
if (r > 0) /* make this a NOP if /dev is already a mount point */
return 0;
if (mount("/dev", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0)
return log_debug_errno(errno, "Failed to bind mount %s: %m", mount_entry_path(m));
return 1;
}
static int mount_sysfs(MountEntry *m) {
int r;
assert(m);
r = path_is_mount_point(mount_entry_path(m), NULL, 0);
if (r < 0)
return log_debug_errno(r, "Unable to determine whether /sys is already mounted: %m");
if (r > 0) /* make this a NOP if /sys is already a mount point */
return 0;
/* Bind mount the host's version so that we get all child mounts of it, too. */
if (mount("/sys", mount_entry_path(m), NULL, MS_BIND|MS_REC, NULL) < 0)
return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m));
return 1;
}
static int mount_procfs(MountEntry *m) {
int r;
assert(m);
r = path_is_mount_point(mount_entry_path(m), NULL, 0);
if (r < 0)
return log_debug_errno(r, "Unable to determine whether /proc is already mounted: %m");
if (r > 0) /* make this a NOP if /proc is already a mount point */
return 0;
/* Mount a new instance, so that we get the one that matches our user namespace, if we are running in one */
if (mount("proc", mount_entry_path(m), "proc", MS_NOSUID|MS_NOEXEC|MS_NODEV, NULL) < 0)
return log_debug_errno(errno, "Failed to mount %s: %m", mount_entry_path(m));
return 1;
}
static int mount_entry_chase(
const char *root_directory,
MountEntry *m,
@ -684,6 +751,7 @@ static int apply_mount(
case BIND_MOUNT_RECURSIVE:
/* Also chase the source mount */
r = mount_entry_chase(root_directory, m, mount_entry_source(m), &m->source_malloc);
if (r <= 0)
return r;
@ -700,7 +768,16 @@ static int apply_mount(
break;
case PRIVATE_DEV:
return mount_dev(m);
return mount_private_dev(m);
case BIND_DEV:
return mount_bind_dev(m);
case SYSFS:
return mount_sysfs(m);
case PROCFS:
return mount_procfs(m);
default:
assert_not_reached("Unknown mode");
@ -722,7 +799,7 @@ static int make_read_only(MountEntry *m, char **blacklist) {
if (mount_entry_read_only(m))
r = bind_remount_recursive(mount_entry_path(m), true, blacklist);
else if (m->mode == PRIVATE_DEV) { /* Can be readonly but the submounts can't*/
else if (m->mode == PRIVATE_DEV) { /* Superblock can be readonly but the submounts can't*/
if (mount(NULL, mount_entry_path(m), NULL, MS_REMOUNT|DEV_MOUNT_OPTIONS|MS_RDONLY, NULL) < 0)
r = -errno;
} else
@ -738,6 +815,17 @@ static int make_read_only(MountEntry *m, char **blacklist) {
return r;
}
static bool namespace_info_mount_apivfs(const NameSpaceInfo *ns_info) {
assert(ns_info);
/* ProtectControlGroups= and ProtectKernelTunables= imply MountAPIVFS=, since to protect the API VFS mounts,
* they need to be around in the first place... */
return ns_info->mount_apivfs ||
ns_info->protect_control_groups ||
ns_info->protect_kernel_tunables;
}
static unsigned namespace_calculate_mounts(
const NameSpaceInfo *ns_info,
char** read_write_paths,
@ -774,11 +862,13 @@ static unsigned namespace_calculate_mounts(
(ns_info->protect_kernel_tunables ? ELEMENTSOF(protect_kernel_tunables_table) : 0) +
(ns_info->protect_control_groups ? 1 : 0) +
(ns_info->protect_kernel_modules ? ELEMENTSOF(protect_kernel_modules_table) : 0) +
protect_home_cnt + protect_system_cnt;
protect_home_cnt + protect_system_cnt +
(namespace_info_mount_apivfs(ns_info) ? ELEMENTSOF(apivfs_table) : 0);
}
int setup_namespace(
const char* root_directory,
const char* root_image,
const NameSpaceInfo *ns_info,
char** read_write_paths,
char** read_only_paths,
@ -789,16 +879,57 @@ int setup_namespace(
const char* var_tmp_dir,
ProtectHome protect_home,
ProtectSystem protect_system,
unsigned long mount_flags) {
unsigned long mount_flags,
DissectImageFlags dissect_image_flags) {
_cleanup_(loop_device_unrefp) LoopDevice *loop_device = NULL;
_cleanup_(decrypted_image_unrefp) DecryptedImage *decrypted_image = NULL;
_cleanup_(dissected_image_unrefp) DissectedImage *dissected_image = NULL;
_cleanup_free_ void *root_hash = NULL;
MountEntry *m, *mounts = NULL;
size_t root_hash_size = 0;
bool make_slave = false;
unsigned n_mounts;
int r = 0;
assert(ns_info);
if (mount_flags == 0)
mount_flags = MS_SHARED;
if (root_image) {
dissect_image_flags |= DISSECT_IMAGE_REQUIRE_ROOT;
if (protect_system == PROTECT_SYSTEM_STRICT && strv_isempty(read_write_paths))
dissect_image_flags |= DISSECT_IMAGE_READ_ONLY;
r = loop_device_make_by_path(root_image,
dissect_image_flags & DISSECT_IMAGE_READ_ONLY ? O_RDONLY : O_RDWR,
&loop_device);
if (r < 0)
return r;
r = root_hash_load(root_image, &root_hash, &root_hash_size);
if (r < 0)
return r;
r = dissect_image(loop_device->fd, root_hash, root_hash_size, dissect_image_flags, &dissected_image);
if (r < 0)
return r;
r = dissected_image_decrypt(dissected_image, NULL, root_hash, root_hash_size, dissect_image_flags, &decrypted_image);
if (r < 0)
return r;
if (!root_directory) {
/* Create a mount point for the image, if it's still missing. We use the same mount point for
* all images, which is safe, since they all live in their own namespaces after all, and hence
* won't see each other. */
root_directory = "/run/systemd/unit-root";
(void) mkdir(root_directory, 0700);
}
}
n_mounts = namespace_calculate_mounts(
ns_info,
read_write_paths,
@ -878,6 +1009,12 @@ int setup_namespace(
if (r < 0)
goto finish;
if (namespace_info_mount_apivfs(ns_info)) {
r = append_static_mounts(&m, apivfs_table, ELEMENTSOF(apivfs_table), ns_info->ignore_protect_paths);
if (r < 0)
goto finish;
}
assert(mounts + n_mounts == m);
/* Prepend the root directory where that's necessary */
@ -907,7 +1044,19 @@ int setup_namespace(
}
}
if (root_directory) {
if (root_image) {
r = dissected_image_mount(dissected_image, root_directory, dissect_image_flags);
if (r < 0)
goto finish;
r = decrypted_image_relinquish(decrypted_image);
if (r < 0)
goto finish;
loop_device_relinquish(loop_device);
} else if (root_directory) {
/* Turn directory into bind mount, if it isn't one yet */
r = path_is_mount_point(root_directory, NULL, AT_SYMLINK_FOLLOW);
if (r < 0)
@ -964,7 +1113,7 @@ int setup_namespace(
finish:
for (m = mounts; m < mounts + n_mounts; m++)
free(m->path_malloc);
mount_entry_done(m);
return r;
}

6
src/core/namespace.h
View File

@ -25,6 +25,7 @@ typedef struct BindMount BindMount;
#include <stdbool.h>
#include "dissect-image.h"
#include "macro.h"
typedef enum ProtectHome {
@ -50,6 +51,7 @@ struct NameSpaceInfo {
bool protect_control_groups:1;
bool protect_kernel_tunables:1;
bool protect_kernel_modules:1;
bool mount_apivfs:1;
};
struct BindMount {
@ -62,6 +64,7 @@ struct BindMount {
int setup_namespace(
const char *root_directory,
const char *root_image,
const NameSpaceInfo *ns_info,
char **read_write_paths,
char **read_only_paths,
@ -72,7 +75,8 @@ int setup_namespace(
const char *var_tmp_dir,
ProtectHome protect_home,
ProtectSystem protect_system,
unsigned long mount_flags);
unsigned long mount_flags,
DissectImageFlags dissected_image_flags);
int setup_tmp_dirs(
const char *id,

6
src/core/unit.c
View File

@ -862,6 +862,12 @@ int unit_add_exec_dependencies(Unit *u, ExecContext *c) {
return r;
}
if (c->root_image) {
r = unit_require_mounts_for(u, c->root_image);
if (r < 0)
return r;
}
if (!MANAGER_IS_SYSTEM(u->manager))
return 0;

8
src/dissect/dissect.c
View File

@ -191,6 +191,14 @@ int main(int argc, char *argv[]) {
goto finish;
}
if (!arg_root_hash) {
r = root_hash_load(arg_image, &arg_root_hash, &arg_root_hash_size);
if (r < 0) {
log_error_errno(r, "Failed to read root hash file for %s: %m", arg_image);
goto finish;
}
}
r = dissect_image(d->fd, arg_root_hash, arg_root_hash_size, arg_flags, &m);
if (r == -ENOPKG) {
log_error_errno(r, "Couldn't identify a suitable partition table or file system in %s.", arg_image);

57
src/nspawn/nspawn.c
View File

@ -3507,53 +3507,6 @@ static int run(int master,
return 1; /* loop again */
}
static int load_root_hash(const char *image) {
_cleanup_free_ char *text = NULL, *fn = NULL;
char *n, *e;
void *k;
size_t l;
int r;
assert_se(image);
/* Try to load the root hash from a file next to the image file if it exists. */
if (arg_root_hash)
return 0;
fn = new(char, strlen(image) + strlen(".roothash") + 1);
if (!fn)
return log_oom();
n = stpcpy(fn, image);
e = endswith(fn, ".raw");
if (e)
n = e;
strcpy(n, ".roothash");
r = read_one_line_file(fn, &text);
if (r == -ENOENT)
return 0;
if (r < 0) {
log_warning_errno(r, "Failed to read %s, ignoring: %m", fn);
return 0;
}
r = unhexmem(text, strlen(text), &k, &l);
if (r < 0)
return log_error_errno(r, "Invalid root hash: %s", text);
if (l < sizeof(sd_id128_t)) {
free(k);
return log_error_errno(r, "Root hash too short: %s", text);
}
arg_root_hash = k;
arg_root_hash_size = l;
return 0;
}
int main(int argc, char *argv[]) {
_cleanup_free_ char *console = NULL;
@ -3769,9 +3722,13 @@ int main(int argc, char *argv[]) {
goto finish;
}
r = load_root_hash(arg_image);
if (r < 0)
goto finish;
if (!arg_root_hash) {
r = root_hash_load(arg_image, &arg_root_hash, &arg_root_hash_size);
if (r < 0) {
log_error_errno(r, "Failed to load root hash file for %s: %m", arg_image);
goto finish;
}
}
}
if (!mkdtemp(tmprootdir)) {

8
src/shared/bus-unit-util.c
View File

@ -208,7 +208,7 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen
"PrivateTmp", "PrivateDevices", "PrivateNetwork", "PrivateUsers", "NoNewPrivileges",
"SyslogLevelPrefix", "Delegate", "RemainAfterElapse", "MemoryDenyWriteExecute",
"RestrictRealtime", "DynamicUser", "RemoveIPC", "ProtectKernelTunables",
"ProtectKernelModules", "ProtectControlGroups")) {
"ProtectKernelModules", "ProtectControlGroups", "MountAPIVFS")) {
r = parse_boolean(eq);
if (r < 0)
@ -266,7 +266,7 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen
"StandardInput", "StandardOutput", "StandardError",
"Description", "Slice", "Type", "WorkingDirectory",
"RootDirectory", "SyslogIdentifier", "ProtectSystem",
"ProtectHome", "SELinuxContext", "Restart"))
"ProtectHome", "SELinuxContext", "Restart", "RootImage"))
r = sd_bus_message_append(m, "v", "s", eq);
else if (streq(field, "SyslogLevel")) {
@ -484,7 +484,7 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen
for (p = eq;;) {
_cleanup_free_ char *word = NULL;
int offset;
size_t offset;
r = extract_first_word(&p, &word, NULL, EXTRACT_QUOTES);
if (r < 0) {
@ -500,6 +500,8 @@ int bus_append_unit_property_assignment(sd_bus_message *m, const char *assignmen
}
offset = word[0] == '-';
offset += word[offset] == '+';
if (!path_is_absolute(word + offset)) {
log_error("Failed to parse %s value %s", field, eq);
return -EINVAL;

98
src/shared/dissect-image.c
View File

@ -28,14 +28,19 @@
#include "blkid-util.h"
#include "dissect-image.h"
#include "fd-util.h"
#include "fileio.h"
#include "fs-util.h"
#include "gpt.h"
#include "hexdecoct.h"
#include "mount-util.h"
#include "path-util.h"
#include "stat-util.h"
#include "stdio-util.h"
#include "string-table.h"
#include "string-util.h"
#include "strv.h"
#include "udev-util.h"
#include "xattr-util.h"
static int probe_filesystem(const char *node, char **ret_fstype) {
#ifdef HAVE_BLKID
@ -657,7 +662,9 @@ static int mount_partition(
DissectImageFlags flags) {
const char *p, *options = NULL, *node, *fstype;
_cleanup_free_ char *chased = NULL;
bool rw;
int r;
assert(m);
assert(where);
@ -674,9 +681,13 @@ static int mount_partition(
rw = m->rw && !(flags & DISSECT_IMAGE_READ_ONLY);
if (directory)
p = strjoina(where, directory);
else
if (directory) {
r = chase_symlinks(directory, where, CHASE_PREFIX_ROOT, &chased);
if (r < 0)
return r;
p = chased;
} else
p = where;
/* If requested, turn on discard support. */
@ -710,22 +721,23 @@ int dissected_image_mount(DissectedImage *m, const char *where, DissectImageFlag
return r;
if (m->partitions[PARTITION_ESP].found) {
const char *mp, *x;
const char *mp;
/* Mount the ESP to /efi if it exists and is empty. If it doesn't exist, use /boot instead. */
mp = "/efi";
x = strjoina(where, mp);
r = dir_is_empty(x);
if (r == -ENOENT) {
mp = "/boot";
x = strjoina(where, mp);
r = dir_is_empty(x);
}
if (r > 0) {
r = mount_partition(m->partitions + PARTITION_ESP, where, mp, flags);
FOREACH_STRING(mp, "/efi", "/boot") {
_cleanup_free_ char *p = NULL;
r = chase_symlinks(mp, where, CHASE_PREFIX_ROOT, &p);
if (r < 0)
return r;
continue;
r = dir_is_empty(p);
if (r > 0) {
r = mount_partition(m->partitions + PARTITION_ESP, where, mp, flags);
if (r < 0)
return r;
}
}
}
@ -1111,6 +1123,62 @@ int decrypted_image_relinquish(DecryptedImage *d) {
return 0;
}
int root_hash_load(const char *image, void **ret, size_t *ret_size) {
_cleanup_free_ char *text = NULL;
_cleanup_free_ void *k = NULL;
size_t l;
int r;
assert(image);
assert(ret);
assert(ret_size);
if (is_device_path(image)) {
/* If we are asked to load the root hash for a device node, exit early */
*ret = NULL;
*ret_size = 0;
return 0;
}
r = getxattr_malloc(image, "user.verity.roothash", &text, true);
if (r < 0) {
char *fn, *e, *n;
if (!IN_SET(r, -ENODATA, -EOPNOTSUPP, -ENOENT))
return r;
fn = newa(char, strlen(image) + strlen(".roothash") + 1);
n = stpcpy(fn, image);
e = endswith(fn, ".raw");
if (e)
n = e;
strcpy(n, ".roothash");
r = read_one_line_file(fn, &text);
if (r == -ENOENT) {
*ret = NULL;
*ret_size = 0;
return 0;
}
if (r < 0)
return r;
}
r = unhexmem(text, strlen(text), &k, &l);
if (r < 0)
return r;
if (l < sizeof(sd_id128_t))
return -EINVAL;
*ret = k;
*ret_size = l;
k = NULL;
return 1;
}
static const char *const partition_designator_table[] = {
[PARTITION_ROOT] = "root",
[PARTITION_ROOT_SECONDARY] = "root-secondary",

2
src/shared/dissect-image.h
View File

@ -94,3 +94,5 @@ int decrypted_image_relinquish(DecryptedImage *d);
const char* partition_designator_to_string(int i) _const_;
int partition_designator_from_string(const char *name) _pure_;
int root_hash_load(const char *image, void **ret, size_t *ret_size);

2
src/test/test-ns.c
View File

@ -77,6 +77,7 @@ int main(int argc, char *argv[]) {
log_info("Not chrooted");
r = setup_namespace(root_directory,
NULL,
&ns_info,
(char **) writable,
(char **) readonly,
@ -86,6 +87,7 @@ int main(int argc, char *argv[]) {
var_tmp_dir,
PROTECT_HOME_NO,
PROTECT_SYSTEM_NO,
0,
0);
if (r < 0) {
log_error_errno(r, "Failed to setup namespace: %m");