cgroup: Add support for ProtectControlGroups= private and strict

This commit adds two settings private and strict to
the ProtectControlGroups= property. Private will unshare the cgroup
namespace and mount a read-write private cgroup2 filesystem at /sys/fs/cgroup.
Strict does the same except the mount is read-only. Since the unit is
running in a cgroup namespace, the new root of /sys/fs/cgroup is the unit's
own cgroup.

We also add a new dbus property ProtectControlGroupsEx which accepts strings
instead of boolean. This will allow users to use private/strict via dbus
and systemd-run in addition to service files.

Note private and strict fall back to no and yes respectively if the kernel
doesn't support cgroup2 or system is not using unified hierarchy.

Fixes: #34634
This commit is contained in:
Ryan Wilson
2024-10-28 08:37:36 -07:00
parent 5fe2923828
commit cd58b5a135
10 changed files with 323 additions and 34 deletions
+23
View File
@@ -61,6 +61,7 @@ static BUS_DEFINE_PROPERTY_GET2(property_get_ioprio_priority, "i", ExecContext,
static BUS_DEFINE_PROPERTY_GET_GLOBAL(property_get_empty_string, "s", NULL);
static BUS_DEFINE_PROPERTY_GET_REF(property_get_private_tmp_ex, "s", PrivateTmp, private_tmp_to_string);
static BUS_DEFINE_PROPERTY_GET_REF(property_get_private_users_ex, "s", PrivateUsers, private_users_to_string);
static BUS_DEFINE_PROPERTY_GET_REF(property_get_protect_control_groups_ex, "s", ProtectControlGroups, protect_control_groups_to_string);
static BUS_DEFINE_PROPERTY_GET_REF(property_get_syslog_level, "i", int, LOG_PRI);
static BUS_DEFINE_PROPERTY_GET_REF(property_get_syslog_facility, "i", int, LOG_FAC);
static BUS_DEFINE_PROPERTY_GET(property_get_cpu_affinity_from_numa, "b", ExecContext, exec_context_get_cpu_affinity_from_numa);
@@ -1179,6 +1180,7 @@ const sd_bus_vtable bus_exec_vtable[] = {
SD_BUS_PROPERTY("ProtectKernelModules", "b", bus_property_get_bool, offsetof(ExecContext, protect_kernel_modules), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("ProtectKernelLogs", "b", bus_property_get_bool, offsetof(ExecContext, protect_kernel_logs), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("ProtectControlGroups", "b", property_get_protect_control_groups, offsetof(ExecContext, protect_control_groups), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("ProtectControlGroupsEx", "s", property_get_protect_control_groups_ex, offsetof(ExecContext, protect_control_groups), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PrivateNetwork", "b", bus_property_get_bool, offsetof(ExecContext, private_network), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PrivateUsers", "b", property_get_private_users, offsetof(ExecContext, private_users), SD_BUS_VTABLE_PROPERTY_CONST),
SD_BUS_PROPERTY("PrivateUsersEx", "s", property_get_private_users_ex, offsetof(ExecContext, private_users), SD_BUS_VTABLE_PROPERTY_CONST),
@@ -1939,6 +1941,27 @@ int bus_exec_context_set_transient_property(
return 1;
}
if (streq(name, "ProtectControlGroupsEx")) {
const char *s;
ProtectControlGroups t;
r = sd_bus_message_read(message, "s", &s);
if (r < 0)
return r;
t = protect_control_groups_from_string(s);
if (t < 0)
return sd_bus_error_setf(error, SD_BUS_ERROR_INVALID_ARGS, "Invalid %s setting: %s", name, s);
if (!UNIT_WRITE_FLAGS_NOOP(flags)) {
c->protect_control_groups = t;
(void) unit_write_settingf(u, flags, name, "ProtectControlGroups=%s",
protect_control_groups_to_string(c->protect_control_groups));
}
return 1;
}
if (streq(name, "PrivateDevices"))
return bus_set_transient_bool(u, name, &c->private_devices, message, flags, error);