From 915e6d1676cf73c4f927f3bbfa21ee82640b1832 Mon Sep 17 00:00:00 2001 From: Lennart Poettering Date: Fri, 23 Dec 2016 14:26:05 +0100 Subject: [PATCH] core: add RootImage= setting for using a specific image file as root directory for a service This is similar to RootDirectory= but mounts the root file system from a block device or loopback file instead of another directory. This reuses the image dissector code now used by nspawn and gpt-auto-discovery. --- man/systemd.exec.xml | 53 ++++++++++++++++----------- src/core/dbus-execute.c | 5 ++- src/core/execute.c | 33 ++++++++++++----- src/core/execute.h | 2 +- src/core/load-fragment-gperf.gperf.m4 | 1 + src/core/namespace.c | 44 +++++++++++++++++++++- src/core/namespace.h | 5 ++- src/core/unit.c | 6 +++ src/shared/bus-unit-util.c | 2 +- src/test/test-ns.c | 2 + 10 files changed, 117 insertions(+), 36 deletions(-) diff --git a/man/systemd.exec.xml b/man/systemd.exec.xml index e594dc1b0cd..09e78c67867 100644 --- a/man/systemd.exec.xml +++ b/man/systemd.exec.xml @@ -86,12 +86,10 @@ A few execution parameters result in additional, automatic dependencies to be added. - Units with WorkingDirectory= or - RootDirectory= set automatically gain - dependencies of type Requires= and - After= on all mount units required to access - the specified paths. This is equivalent to having them listed - explicitly in RequiresMountsFor=. + Units with WorkingDirectory=, RootDirectory= or + RootImage= set automatically gain dependencies of type Requires= and + After= on all mount units required to access the specified paths. This is equivalent to having + them listed explicitly in RequiresMountsFor=. Similar, units with PrivateTmp= enabled automatically get mount unit dependencies for all mounts required to access /tmp and /var/tmp. They will also gain an @@ -117,9 +115,10 @@ User= 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 - character, a missing working directory is not considered fatal. If - RootDirectory= is not set, then WorkingDirectory= 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). + RootDirectory=/RootImage= is not set, then + WorkingDirectory= 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). @@ -136,13 +135,24 @@ in conjunction with RootDirectory=. For details, see below. + + RootImage= + Takes a path to a block device node or regular file as argument. This call is similar to + RootDirectory= 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 Discoverable Partitions + Specification. + + MountAPIVFS= Takes a boolean argument. If on, a private mount namespace for the unit's processes is created and the API file systems /proc, /sys and /dev will be mounted inside of it, unless they are already mounted. Note that this option has no effect unless used - in conjunction with RootDirectory= as these three mounts are generally mounted in the host + in conjunction with RootDirectory=/RootImage= 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 /dev file system of the host is bind mounted if this option is used without PrivateDevices=. To run the service with a private, minimal @@ -952,7 +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 - RootDirectory=. + RootDirectory=/RootImage=. Paths listed in ReadWritePaths= are accessible from within the namespace with the same access modes as from outside of it. Paths listed in ReadOnlyPaths= are accessible for @@ -971,9 +981,10 @@ Paths in ReadWritePaths=, ReadOnlyPaths= and InaccessiblePaths= may be prefixed with -, in which case they will be ignored when they do not exist. If prefixed with + the paths are taken relative to the root - directory of the unit, as configured with RootDirectory=, instead of relative to the root - directory of the host (see above). When combining - and + on the same - path make sure to specify - first, and + second. + directory of the unit, as configured with RootDirectory=/RootImage=, + instead of relative to the root directory of the host (see above). When combining - and + + on the same path make sure to specify - first, and + + second. 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 @@ -1004,9 +1015,9 @@ that in this case both read-only and regular bind mounts are reset, regardless which of the two settings is used. - This option is particularly useful when RootDirectory= 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. + This option is particularly useful when RootDirectory=/RootImage= + 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. @@ -1094,10 +1105,10 @@ such as CapabilityBoundingSet= will affect only the latter, and there's no way to acquire additional capabilities in the host's user namespace. Defaults to off. - This setting is particularly useful in conjunction with RootDirectory=, 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 root, nobody and the unit's own - user and group. + This setting is particularly useful in conjunction with + RootDirectory=/RootImage=, 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 root, nobody and the unit's own user and group. diff --git a/src/core/dbus-execute.c b/src/core/dbus-execute.c index c57af5aaafa..7df4cab3f6f 100644 --- a/src/core/dbus-execute.c +++ b/src/core/dbus-execute.c @@ -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), @@ -1048,7 +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); @@ -1061,6 +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); diff --git a/src/core/execute.c b/src/core/execute.c index 54f6418c5a8..f57eb263887 100644 --- a/src/core/execute.c +++ b/src/core/execute.c @@ -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)) @@ -1938,7 +1941,7 @@ 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, @@ -1965,8 +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 @@ -1976,7 +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, @@ -1985,7 +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. */ @@ -1999,10 +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; @@ -2983,6 +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); @@ -3320,6 +3332,9 @@ void exec_context_dump(ExecContext *c, FILE* f, const char *prefix) { 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); diff --git a/src/core/execute.h b/src/core/execute.h index 6fd5a6e5ce7..9f2b6fd39e3 100644 --- a/src/core/execute.h +++ b/src/core/execute.h @@ -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; diff --git a/src/core/load-fragment-gperf.gperf.m4 b/src/core/load-fragment-gperf.gperf.m4 index 07f2a70c8f8..cb9e6fea278 100644 --- a/src/core/load-fragment-gperf.gperf.m4 +++ b/src/core/load-fragment-gperf.gperf.m4 @@ -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) diff --git a/src/core/namespace.c b/src/core/namespace.c index 10917f7b70e..0ae5f704c77 100644 --- a/src/core/namespace.c +++ b/src/core/namespace.c @@ -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" @@ -867,6 +868,7 @@ static unsigned namespace_calculate_mounts( int setup_namespace( const char* root_directory, + const char* root_image, const NameSpaceInfo *ns_info, char** read_write_paths, char** read_only_paths, @@ -877,16 +879,46 @@ 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_(dissected_image_unrefp) DissectedImage *dissected_image = NULL; MountEntry *m, *mounts = NULL; 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 = dissect_image(loop_device->fd, NULL, 0, dissect_image_flags, &dissected_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, @@ -1001,7 +1033,15 @@ 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; + + 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) diff --git a/src/core/namespace.h b/src/core/namespace.h index bb9de9857c2..f54954bd864 100644 --- a/src/core/namespace.h +++ b/src/core/namespace.h @@ -25,6 +25,7 @@ typedef struct BindMount BindMount; #include +#include "dissect-image.h" #include "macro.h" typedef enum ProtectHome { @@ -63,6 +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, @@ -73,7 +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, diff --git a/src/core/unit.c b/src/core/unit.c index 44f1d5e206e..90d7eea9562 100644 --- a/src/core/unit.c +++ b/src/core/unit.c @@ -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; diff --git a/src/shared/bus-unit-util.c b/src/shared/bus-unit-util.c index a4677bef271..20c1085697a 100644 --- a/src/shared/bus-unit-util.c +++ b/src/shared/bus-unit-util.c @@ -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")) { diff --git a/src/test/test-ns.c b/src/test/test-ns.c index c99bcb371b9..0125d905a63 100644 --- a/src/test/test-ns.c +++ b/src/test/test-ns.c @@ -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"); -- 2.39.2