diff options
-rw-r--r-- | .gitignore | 1 | ||||
-rw-r--r-- | man/systemd-detect-virt.xml | 16 | ||||
-rw-r--r-- | man/systemd-vconsole-setup.service.xml | 28 | ||||
-rw-r--r-- | man/systemd.exec.xml | 23 | ||||
-rw-r--r-- | man/systemd.unit.xml | 3 | ||||
-rw-r--r-- | man/udev.xml | 4 | ||||
-rw-r--r-- | src/basic/virt.c | 70 | ||||
-rw-r--r-- | src/basic/virt.h | 1 | ||||
-rw-r--r-- | src/core/execute.c | 177 | ||||
-rw-r--r-- | src/core/load-fragment.c | 7 | ||||
-rw-r--r-- | src/detect-virt/detect-virt.c | 31 | ||||
-rw-r--r-- | src/hwdb/hwdb.c | 4 | ||||
-rw-r--r-- | src/network/networkd-network-gperf.gperf | 2 | ||||
-rw-r--r-- | src/network/networkd-network.c | 50 | ||||
-rw-r--r-- | src/network/networkd-network.h | 1 | ||||
-rw-r--r-- | src/shared/condition.c | 19 | ||||
-rw-r--r-- | src/systemctl/systemctl.c | 14 | ||||
-rw-r--r-- | src/test/test-condition.c | 60 | ||||
-rw-r--r-- | src/test/test-tables.c | 2 | ||||
-rw-r--r-- | src/udev/udev-rules.c | 8 | ||||
-rw-r--r-- | src/vconsole/90-vconsole.rules.in | 2 | ||||
-rw-r--r-- | src/vconsole/vconsole-setup.c | 58 | ||||
-rw-r--r-- | units/dev-hugepages.mount | 1 |
23 files changed, 435 insertions, 147 deletions
diff --git a/.gitignore b/.gitignore index c925f19f41..f8336063b7 100644 --- a/.gitignore +++ b/.gitignore @@ -261,6 +261,7 @@ /test-ring /test-rlimit-util /test-sched-prio +/test-seccomp /test-selinux /test-set /test-sizeof diff --git a/man/systemd-detect-virt.xml b/man/systemd-detect-virt.xml index 61a5f8937f..996c2fa256 100644 --- a/man/systemd-detect-virt.xml +++ b/man/systemd-detect-virt.xml @@ -50,7 +50,8 @@ <refsynopsisdiv> <cmdsynopsis> - <command>systemd-detect-virt <arg choice="opt" rep="repeat">OPTIONS</arg></command> + <command>systemd-detect-virt</command> + <arg choice="opt" rep="repeat">OPTIONS</arg> </cmdsynopsis> </refsynopsisdiv> @@ -218,6 +219,16 @@ </varlistentry> <varlistentry> + <term><option>--private-users</option></term> + + <listitem><para>Detect whether invoked in a user namespace. In this mode, no + output is written, but the return value indicates whether the process was invoked + inside of a user namespace or not. See + <citerefentry project='man-pages'><refentrytitle>user_namespaces</refentrytitle><manvolnum>7</manvolnum></citerefentry> + for more information.</para></listitem> + </varlistentry> + + <varlistentry> <term><option>-q</option></term> <term><option>--quiet</option></term> @@ -243,7 +254,8 @@ <para> <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>, <citerefentry><refentrytitle>systemd-nspawn</refentrytitle><manvolnum>1</manvolnum></citerefentry>, - <citerefentry><refentrytitle>chroot</refentrytitle><manvolnum>2</manvolnum></citerefentry> + <citerefentry><refentrytitle>chroot</refentrytitle><manvolnum>2</manvolnum></citerefentry>, + <citerefentry project='man-pages'><refentrytitle>namespaces</refentrytitle><manvolnum>7</manvolnum></citerefentry> </para> </refsect1> diff --git a/man/systemd-vconsole-setup.service.xml b/man/systemd-vconsole-setup.service.xml index e048258621..f2da2a7b77 100644 --- a/man/systemd-vconsole-setup.service.xml +++ b/man/systemd-vconsole-setup.service.xml @@ -43,23 +43,35 @@ <refnamediv> <refname>systemd-vconsole-setup.service</refname> <refname>systemd-vconsole-setup</refname> - <refpurpose>Configure the virtual console at boot</refpurpose> + <refpurpose>Configure the virtual consoles</refpurpose> </refnamediv> <refsynopsisdiv> <para><filename>systemd-vconsole-setup.service</filename></para> - <para><filename>/usr/lib/systemd/systemd-vconsole-setup</filename></para> + <cmdsynopsis> + <command>/usr/lib/systemd/systemd-vconsole-setup</command> + <arg choice="opt">TTY</arg> + </cmdsynopsis> </refsynopsisdiv> <refsect1> <title>Description</title> - <para><filename>systemd-vconsole-setup.service</filename> is an - early boot service that configures the virtual console font and - console keymap. Internally it calls - <citerefentry project='mankier'><refentrytitle>loadkeys</refentrytitle><manvolnum>1</manvolnum></citerefentry> - and - <citerefentry project='die-net'><refentrytitle>setfont</refentrytitle><manvolnum>8</manvolnum></citerefentry>.</para> + <para><filename>systemd-vconsole-setup</filename> is a helper used to prepare either all virtual consoles, or — if + the optional <replaceable>TTY</replaceable> parameter is provided — a specific one. When the system is booting up + it's called by <citerefentry><command>udev</command></citerefentry> during vtconsole subsystem initialization. + <productname>Systemd</productname> also calls it internally as needed via + <filename>systemd-vconsole-setup.service</filename>. The helper calls + <citerefentry project='mankier'><refentrytitle>loadkeys</refentrytitle><manvolnum>1</manvolnum></citerefentry> and + <citerefentry project='die-net'><refentrytitle>setfont</refentrytitle><manvolnum>8</manvolnum></citerefentry> + internally. + </para> + + <para> + You may want to use this helper whenever you change <filename>vconsole.conf</filename> to + refresh the settings on your consoles — either through the <command>systemctl restart</command> / + <command>systemctl start</command> command or directly through the executable. + </para> <para>See <citerefentry><refentrytitle>vconsole.conf</refentrytitle><manvolnum>5</manvolnum></citerefentry> diff --git a/man/systemd.exec.xml b/man/systemd.exec.xml index f9a15d8db0..54ec7e29ca 100644 --- a/man/systemd.exec.xml +++ b/man/systemd.exec.xml @@ -1234,13 +1234,22 @@ <varlistentry> <term><varname>NoNewPrivileges=</varname></term> - <listitem><para>Takes a boolean argument. If true, ensures - that the service process and all its children can never gain - new privileges. This option is more powerful than the - respective secure bits flags (see above), as it also prohibits - UID changes of any kind. This is the simplest, most effective - way to ensure that a process and its children can never - elevate privileges again.</para></listitem> + <listitem><para>Takes a boolean argument. If true, ensures that the service + process and all its children can never gain new privileges. This option is more + powerful than the respective secure bits flags (see above), as it also prohibits + UID changes of any kind. This is the simplest and most effective way to ensure that + a process and its children can never elevate privileges again. Defaults to false, + but in the user manager instance certain settings force + <varname>NoNewPrivileges=yes</varname>, ignoring the value of this setting. + Those is the case when <varname>SystemCallFilter=</varname>, + <varname>SystemCallArchitectures=</varname>, + <varname>RestrictAddressFamilies=</varname>, + <varname>PrivateDevices=</varname>, + <varname>ProtectKernelTunables=</varname>, + <varname>ProtectKernelModules=</varname>, + <varname>MemoryDenyWriteExecute=</varname>, or + <varname>RestrictRealtime=</varname> are specified. + </para></listitem> </varlistentry> <varlistentry> diff --git a/man/systemd.unit.xml b/man/systemd.unit.xml index 04efee2891..40c4cfd854 100644 --- a/man/systemd.unit.xml +++ b/man/systemd.unit.xml @@ -908,7 +908,8 @@ <varname>systemd-nspawn</varname>, <varname>docker</varname>, <varname>rkt</varname> to test - against a specific implementation. See + against a specific implementation, or + <varname>private-users</varname> to check whether we are running in a user namespace. See <citerefentry><refentrytitle>systemd-detect-virt</refentrytitle><manvolnum>1</manvolnum></citerefentry> for a full list of known virtualization technologies and their identifiers. If multiple virtualization technologies are diff --git a/man/udev.xml b/man/udev.xml index dd5563605c..3359fb0865 100644 --- a/man/udev.xml +++ b/man/udev.xml @@ -577,8 +577,8 @@ <para>The <varname>NAME</varname>, <varname>SYMLINK</varname>, <varname>PROGRAM</varname>, <varname>OWNER</varname>, - <varname>GROUP</varname>, <varname>MODE</varname>, and - <varname>RUN</varname> fields support simple string substitutions. + <varname>GROUP</varname>, <varname>MODE</varname>, <varname>SECLABEL</varname>, + and <varname>RUN</varname> fields support simple string substitutions. The <varname>RUN</varname> substitutions are performed after all rules have been processed, right before the program is executed, allowing for the use of device properties set by earlier matching rules. For all other diff --git a/src/basic/virt.c b/src/basic/virt.c index 41012d52a0..69b0f96183 100644 --- a/src/basic/virt.c +++ b/src/basic/virt.c @@ -485,6 +485,76 @@ int detect_virtualization(void) { return r; } +static int userns_has_mapping(const char *name) { + _cleanup_fclose_ FILE *f = NULL; + _cleanup_free_ char *buf = NULL; + size_t n_allocated = 0; + ssize_t n; + uint32_t a, b, c; + int r; + + f = fopen(name, "re"); + if (!f) { + log_debug_errno(errno, "Failed to open %s: %m", name); + return errno == -ENOENT ? false : -errno; + } + + n = getline(&buf, &n_allocated, f); + if (n < 0) { + if (feof(f)) { + log_debug("%s is empty, we're in an uninitialized user namespace", name); + return true; + } + + return log_debug_errno(errno, "Failed to read %s: %m", name); + } + + r = sscanf(buf, "%"PRIu32" %"PRIu32" %"PRIu32, &a, &b, &c); + if (r < 3) + return log_debug_errno(errno, "Failed to parse %s: %m", name); + + if (a == 0 && b == 0 && c == UINT32_MAX) { + /* The kernel calls mappings_overlap() and does not allow overlaps */ + log_debug("%s has a full 1:1 mapping", name); + return false; + } + + /* Anything else implies that we are in a user namespace */ + log_debug("Mapping found in %s, we're in a user namespace", name); + return true; +} + +int running_in_userns(void) { + _cleanup_free_ char *line = NULL; + int r; + + r = userns_has_mapping("/proc/self/uid_map"); + if (r != 0) + return r; + + r = userns_has_mapping("/proc/self/gid_map"); + if (r != 0) + return r; + + /* "setgroups" file was added in kernel v3.18-rc6-15-g9cc46516dd. It is also + * possible to compile a kernel without CONFIG_USER_NS, in which case "setgroups" + * also does not exist. We cannot distinguish those two cases, so assume that + * we're running on a stripped-down recent kernel, rather than on an old one, + * and if the file is not found, return false. + */ + r = read_one_line_file("/proc/self/setgroups", &line); + if (r < 0) { + log_debug_errno(r, "/proc/self/setgroups: %m"); + return r == -ENOENT ? false : r; + } + + truncate_nl(line); + r = streq(line, "deny"); + /* See user_namespaces(7) for a description of this "setgroups" contents. */ + log_debug("/proc/self/setgroups contains \"%s\", %s user namespace", line, r ? "in" : "not in"); + return r; +} + int running_in_chroot(void) { int ret; diff --git a/src/basic/virt.h b/src/basic/virt.h index bc5b3ae94d..7d15169112 100644 --- a/src/basic/virt.h +++ b/src/basic/virt.h @@ -67,6 +67,7 @@ int detect_vm(void); int detect_container(void); int detect_virtualization(void); +int running_in_userns(void); int running_in_chroot(void); const char *virtualization_to_string(int v) _const_; diff --git a/src/core/execute.c b/src/core/execute.c index 7b42ac7bdc..ae9df41b99 100644 --- a/src/core/execute.c +++ b/src/core/execute.c @@ -1481,7 +1481,7 @@ finish: return r; } -static int apply_protect_sysctl(Unit *u, const ExecContext *c) { +static int apply_protect_sysctl(const Unit *u, const ExecContext *c) { scmp_filter_ctx seccomp; int r; @@ -1512,7 +1512,7 @@ finish: return r; } -static int apply_protect_kernel_modules(Unit *u, const ExecContext *c) { +static int apply_protect_kernel_modules(const Unit *u, const ExecContext *c) { assert(c); /* Turn off module syscalls on ProtectKernelModules=yes */ @@ -1523,7 +1523,7 @@ static int apply_protect_kernel_modules(Unit *u, const ExecContext *c) { return seccomp_load_filter_set(SCMP_ACT_ALLOW, syscall_filter_sets + SYSCALL_FILTER_SET_MODULE, SCMP_ACT_ERRNO(EPERM)); } -static int apply_private_devices(Unit *u, const ExecContext *c) { +static int apply_private_devices(const Unit *u, const ExecContext *c) { assert(c); /* If PrivateDevices= is set, also turn off iopl and all @raw-io syscalls. */ @@ -2014,6 +2014,92 @@ static int compile_read_write_paths( return 0; } +static int apply_mount_namespace(Unit *u, const ExecContext *context, + const ExecParameters *params, + ExecRuntime *runtime) { + int r; + _cleanup_free_ char **rw = NULL; + char *tmp = NULL, *var = NULL; + const char *root_dir = NULL; + NameSpaceInfo ns_info = { + .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, + }; + + assert(context); + + /* The runtime struct only contains the parent of the private /tmp, + * which is non-accessible to world users. Inside of it there's a /tmp + * that is sticky, and that's the one we want to use here. */ + + if (context->private_tmp && runtime) { + if (runtime->tmp_dir) + tmp = strjoina(runtime->tmp_dir, "/tmp"); + if (runtime->var_tmp_dir) + var = strjoina(runtime->var_tmp_dir, "/tmp"); + } + + r = compile_read_write_paths(context, params, &rw); + if (r < 0) + return r; + + if (params->flags & EXEC_APPLY_CHROOT) + root_dir = context->root_directory; + + r = setup_namespace(root_dir, &ns_info, rw, + context->read_only_paths, + context->inaccessible_paths, + tmp, + var, + context->protect_home, + context->protect_system, + context->mount_flags); + + /* If we couldn't set up the namespace this is probably due to a + * missing capability. In this case, silently proceeed. */ + if (IN_SET(r, -EPERM, -EACCES)) { + log_open(); + log_unit_debug_errno(u, r, "Failed to set up namespace, assuming containerized execution, ignoring: %m"); + log_close(); + r = 0; + } + + return r; +} + +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; + + assert(context); + + if (context->working_directory_home) + wd = home; + else if (context->working_directory) + wd = context->working_directory; + else + wd = "/"; + + if (params->flags & EXEC_APPLY_CHROOT) { + if (!needs_mount_ns && context->root_directory) + if (chroot(context->root_directory) < 0) + return -errno; + + d = wd; + } else + d = strjoina(strempty(context->root_directory), "/", strempty(wd)); + + if (chdir(d) < 0 && !context->working_directory_missing_ok) + return -errno; + + return 0; +} + static void append_socket_pair(int *array, unsigned *n, int pair[2]) { assert(array); assert(n); @@ -2150,7 +2236,7 @@ static int exec_child( _cleanup_free_ char *mac_selinux_context_net = NULL; _cleanup_free_ gid_t *supplementary_gids = NULL; const char *username = NULL, *groupname = NULL; - const char *home = NULL, *shell = NULL, *wd; + const char *home = NULL, *shell = NULL; dev_t journal_stream_dev = 0; ino_t journal_stream_ino = 0; bool needs_mount_namespace; @@ -2477,65 +2563,19 @@ static int exec_child( needs_mount_namespace = exec_needs_mount_namespace(context, params, runtime); if (needs_mount_namespace) { - _cleanup_free_ char **rw = NULL; - char *tmp = NULL, *var = NULL; - NameSpaceInfo ns_info = { - .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, - }; - - /* The runtime struct only contains the parent - * of the private /tmp, which is - * non-accessible to world users. Inside of it - * there's a /tmp that is sticky, and that's - * the one we want to use here. */ - - if (context->private_tmp && runtime) { - if (runtime->tmp_dir) - tmp = strjoina(runtime->tmp_dir, "/tmp"); - if (runtime->var_tmp_dir) - var = strjoina(runtime->var_tmp_dir, "/tmp"); - } - - r = compile_read_write_paths(context, params, &rw); + r = apply_mount_namespace(unit, context, params, runtime); if (r < 0) { *exit_status = EXIT_NAMESPACE; return r; } - - r = setup_namespace( - (params->flags & EXEC_APPLY_CHROOT) ? context->root_directory : NULL, - &ns_info, - rw, - context->read_only_paths, - context->inaccessible_paths, - tmp, - var, - context->protect_home, - context->protect_system, - context->mount_flags); - - /* If we couldn't set up the namespace this is - * probably due to a missing capability. In this case, - * silently proceeed. */ - if (r == -EPERM || r == -EACCES) { - log_open(); - log_unit_debug_errno(unit, r, "Failed to set up namespace, assuming containerized execution, ignoring: %m"); - log_close(); - } else if (r < 0) { - *exit_status = EXIT_NAMESPACE; - return r; - } } - if (context->working_directory_home) - wd = home; - else if (context->working_directory) - wd = context->working_directory; - else - wd = "/"; + /* Apply just after mount namespace setup */ + r = apply_working_directory(context, params, home, needs_mount_namespace); + if (r < 0) { + *exit_status = EXIT_CHROOT; + return r; + } /* Drop group as early as possbile */ if ((params->flags & EXEC_APPLY_PERMISSIONS) && !command->privileged) { @@ -2546,29 +2586,6 @@ static int exec_child( } } - if (params->flags & EXEC_APPLY_CHROOT) { - if (!needs_mount_namespace && context->root_directory) - if (chroot(context->root_directory) < 0) { - *exit_status = EXIT_CHROOT; - return -errno; - } - - if (chdir(wd) < 0 && - !context->working_directory_missing_ok) { - *exit_status = EXIT_CHDIR; - return -errno; - } - } else { - const char *d; - - d = strjoina(strempty(context->root_directory), "/", strempty(wd)); - if (chdir(d) < 0 && - !context->working_directory_missing_ok) { - *exit_status = EXIT_CHDIR; - return -errno; - } - } - #ifdef HAVE_SELINUX if ((params->flags & EXEC_APPLY_PERMISSIONS) && mac_selinux_use() && diff --git a/src/core/load-fragment.c b/src/core/load-fragment.c index 118b39c1cf..cbc826809e 100644 --- a/src/core/load-fragment.c +++ b/src/core/load-fragment.c @@ -2744,11 +2744,6 @@ int config_parse_syscall_filter( return r; } - /* Turn on NNP, but only if it wasn't configured explicitly - * before, and only if we are in user mode. */ - if (!c->no_new_privileges_set && MANAGER_IS_USER(u->manager)) - c->no_new_privileges = true; - return 0; } @@ -3837,7 +3832,7 @@ int config_parse_no_new_privileges( return 0; } - c->no_new_privileges = !!k; + c->no_new_privileges = k; c->no_new_privileges_set = true; return 0; diff --git a/src/detect-virt/detect-virt.c b/src/detect-virt/detect-virt.c index 5d51589a31..4b8956f0ad 100644 --- a/src/detect-virt/detect-virt.c +++ b/src/detect-virt/detect-virt.c @@ -31,6 +31,7 @@ static enum { ONLY_VM, ONLY_CONTAINER, ONLY_CHROOT, + ONLY_PRIVATE_USERS, } arg_mode = ANY_VIRTUALIZATION; static void help(void) { @@ -41,6 +42,7 @@ static void help(void) { " -c --container Only detect whether we are run in a container\n" " -v --vm Only detect whether we are run in a VM\n" " -r --chroot Detect whether we are run in a chroot() environment\n" + " --private-users Only detect whether we are running in a user namespace\n" " -q --quiet Don't output anything, just set return value\n" , program_invocation_short_name); } @@ -48,16 +50,18 @@ static void help(void) { static int parse_argv(int argc, char *argv[]) { enum { - ARG_VERSION = 0x100 + ARG_VERSION = 0x100, + ARG_PRIVATE_USERS, }; static const struct option options[] = { - { "help", no_argument, NULL, 'h' }, - { "version", no_argument, NULL, ARG_VERSION }, - { "container", no_argument, NULL, 'c' }, - { "vm", no_argument, NULL, 'v' }, - { "chroot", no_argument, NULL, 'r' }, - { "quiet", no_argument, NULL, 'q' }, + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "container", no_argument, NULL, 'c' }, + { "vm", no_argument, NULL, 'v' }, + { "chroot", no_argument, NULL, 'r' }, + { "private-users", no_argument, NULL, ARG_PRIVATE_USERS }, + { "quiet", no_argument, NULL, 'q' }, {} }; @@ -85,6 +89,10 @@ static int parse_argv(int argc, char *argv[]) { arg_mode = ONLY_CONTAINER; break; + case ARG_PRIVATE_USERS: + arg_mode = ONLY_PRIVATE_USERS; + break; + case 'v': arg_mode = ONLY_VM; break; @@ -151,6 +159,15 @@ int main(int argc, char *argv[]) { return r ? EXIT_SUCCESS : EXIT_FAILURE; + case ONLY_PRIVATE_USERS: + r = running_in_userns(); + if (r < 0) { + log_error_errno(r, "Failed to check for user namespace: %m"); + return EXIT_FAILURE; + } + + return r ? EXIT_SUCCESS : EXIT_FAILURE; + case ANY_VIRTUALIZATION: default: r = detect_virtualization(); diff --git a/src/hwdb/hwdb.c b/src/hwdb/hwdb.c index be4ef5f9e9..ab1feb435b 100644 --- a/src/hwdb/hwdb.c +++ b/src/hwdb/hwdb.c @@ -172,8 +172,8 @@ static int trie_node_add_value(struct trie *trie, struct trie_node *node, if (v < 0) return v; fn = strbuf_add_string(trie->strings, filename, strlen(filename)); - if (v < 0) - return v; + if (fn < 0) + return fn; if (node->values_count) { struct trie_value_entry search = { diff --git a/src/network/networkd-network-gperf.gperf b/src/network/networkd-network-gperf.gperf index 5587961b9f..bcf8186c33 100644 --- a/src/network/networkd-network-gperf.gperf +++ b/src/network/networkd-network-gperf.gperf @@ -49,7 +49,7 @@ Network.EmitLLDP, config_parse_lldp_emit, Network.Address, config_parse_address, 0, 0 Network.Gateway, config_parse_gateway, 0, 0 Network.Domains, config_parse_domains, 0, 0 -Network.DNS, config_parse_strv, 0, offsetof(Network, dns) +Network.DNS, config_parse_dns, 0, 0 Network.LLMNR, config_parse_resolve_support, 0, offsetof(Network, llmnr) Network.MulticastDNS, config_parse_resolve_support, 0, offsetof(Network, mdns) Network.DNSSEC, config_parse_dnssec_mode, 0, offsetof(Network, dnssec_mode) diff --git a/src/network/networkd-network.c b/src/network/networkd-network.c index 584cb96979..042232fcac 100644 --- a/src/network/networkd-network.c +++ b/src/network/networkd-network.c @@ -979,6 +979,56 @@ int config_parse_dhcp_server_ntp( } } +int config_parse_dns( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + Network *n = userdata; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + + for (;;) { + _cleanup_free_ char *w = NULL; + union in_addr_union a; + int family; + + r = extract_first_word(&rvalue, &w, WHITESPACE, EXTRACT_QUOTES|EXTRACT_RETAIN_ESCAPE); + if (r == 0) + break; + if (r == -ENOMEM) + return log_oom(); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Invalid syntax, ignoring: %s", rvalue); + break; + } + + r = in_addr_from_string_auto(w, &family, &a); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, 0, "Failed to parse dns server address, ignoring: %s", w); + continue; + } + + r = strv_consume(&n->dns, w); + if (r < 0) + return log_oom(); + + w = NULL; + } + + return 0; +} + int config_parse_dnssec_negative_trust_anchors( const char *unit, const char *filename, diff --git a/src/network/networkd-network.h b/src/network/networkd-network.h index ef4b499ab9..42fc82d392 100644 --- a/src/network/networkd-network.h +++ b/src/network/networkd-network.h @@ -220,6 +220,7 @@ int config_parse_netdev(const char *unit, const char *filename, unsigned line, c int config_parse_domains(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_tunnel(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_dhcp(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_dns(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_dhcp_client_identifier(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_ipv6token(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); int config_parse_ipv6_privacy_extensions(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); diff --git a/src/shared/condition.c b/src/shared/condition.c index 69b4837e1f..8bd6a51a99 100644 --- a/src/shared/condition.c +++ b/src/shared/condition.c @@ -146,25 +146,24 @@ static int condition_test_virtualization(Condition *c) { assert(c->parameter); assert(c->type == CONDITION_VIRTUALIZATION); + if (streq(c->parameter, "private-users")) + return running_in_userns(); + v = detect_virtualization(); if (v < 0) return v; /* First, compare with yes/no */ b = parse_boolean(c->parameter); - - if (v > 0 && b > 0) - return true; - - if (v == 0 && b == 0) - return true; + if (b >= 0) + return b == !!v; /* Then, compare categorization */ - if (VIRTUALIZATION_IS_VM(v) && streq(c->parameter, "vm")) - return true; + if (streq(c->parameter, "vm")) + return VIRTUALIZATION_IS_VM(v); - if (VIRTUALIZATION_IS_CONTAINER(v) && streq(c->parameter, "container")) - return true; + if (streq(c->parameter, "container")) + return VIRTUALIZATION_IS_CONTAINER(v); /* Finally compare id */ return v != VIRTUALIZATION_NONE && streq(c->parameter, virtualization_to_string(v)); diff --git a/src/systemctl/systemctl.c b/src/systemctl/systemctl.c index 35d5c11cc7..d311bbec1a 100644 --- a/src/systemctl/systemctl.c +++ b/src/systemctl/systemctl.c @@ -5272,6 +5272,20 @@ static int cat(int argc, char *argv[], void *userdata) { else puts(""); + if (need_daemon_reload(bus, *name)) + fprintf(stderr, + "%s# Warning: %s changed on disk, the version systemd has loaded is outdated.\n" + "%s# This output shows the current version of the unit's original fragment and drop-in files.\n" + "%s# If fragments or drop-ins were added or removed, they are not properly reflected in this output.\n" + "%s# Run 'systemctl%s daemon-reload' to reload units.%s\n", + ansi_highlight_red(), + *name, + ansi_highlight_red(), + ansi_highlight_red(), + ansi_highlight_red(), + arg_scope == UNIT_FILE_SYSTEM ? "" : " --user", + ansi_normal()); + if (fragment_path) { r = cat_file(fragment_path, false); if (r < 0) diff --git a/src/test/test-condition.c b/src/test/test-condition.c index 6f7d71ef9a..dd985f5863 100644 --- a/src/test/test-condition.c +++ b/src/test/test-condition.c @@ -31,6 +31,8 @@ #include "macro.h" #include "selinux-util.h" #include "smack-util.h" +#include "strv.h" +#include "virt.h" #include "util.h" static void test_condition_test_path(void) { @@ -265,7 +267,64 @@ static void test_condition_test_security(void) { condition_free(condition); } +static void test_condition_test_virtualization(void) { + Condition *condition; + const char *virt; + int r; + + condition = condition_new(CONDITION_VIRTUALIZATION, "garbage oifdsjfoidsjoj", false, false); + assert_se(condition); + r = condition_test(condition); + log_info("ConditionVirtualization=garbage → %i", r); + assert_se(r == 0); + condition_free(condition); + + condition = condition_new(CONDITION_VIRTUALIZATION, "container", false, false); + assert_se(condition); + r = condition_test(condition); + log_info("ConditionVirtualization=container → %i", r); + assert_se(r == !!detect_container()); + condition_free(condition); + + condition = condition_new(CONDITION_VIRTUALIZATION, "vm", false, false); + assert_se(condition); + r = condition_test(condition); + log_info("ConditionVirtualization=vm → %i", r); + assert_se(r == (detect_vm() && !detect_container())); + condition_free(condition); + + condition = condition_new(CONDITION_VIRTUALIZATION, "private-users", false, false); + assert_se(condition); + r = condition_test(condition); + log_info("ConditionVirtualization=private-users → %i", r); + assert_se(r == !!running_in_userns()); + condition_free(condition); + + NULSTR_FOREACH(virt, + "kvm\0" + "qemu\0" + "bochs\0" + "xen\0" + "uml\0" + "vmware\0" + "oracle\0" + "microsoft\0" + "zvm\0" + "parallels\0" + "bhyve\0" + "vm_other\0") { + + condition = condition_new(CONDITION_VIRTUALIZATION, virt, false, false); + assert_se(condition); + r = condition_test(condition); + log_info("ConditionVirtualization=%s → %i", virt, r); + assert_se(r >= 0); + condition_free(condition); + } +} + int main(int argc, char *argv[]) { + log_set_max_level(LOG_DEBUG); log_parse_environment(); log_open(); @@ -276,6 +335,7 @@ int main(int argc, char *argv[]) { test_condition_test_kernel_command_line(); test_condition_test_null(); test_condition_test_security(); + test_condition_test_virtualization(); return 0; } diff --git a/src/test/test-tables.c b/src/test/test-tables.c index 8d4622694e..294d219869 100644 --- a/src/test/test-tables.c +++ b/src/test/test-tables.c @@ -48,6 +48,7 @@ #include "unit-name.h" #include "unit.h" #include "util.h" +#include "virt.h" int main(int argc, char **argv) { test_table(architecture, ARCHITECTURE); @@ -114,6 +115,7 @@ int main(int argc, char **argv) { test_table(unit_load_state, UNIT_LOAD_STATE); test_table(unit_type, UNIT_TYPE); test_table(locale_variable, VARIABLE_LC); + test_table(virtualization, VIRTUALIZATION); test_table_sparse(object_compressed, OBJECT_COMPRESSED); diff --git a/src/udev/udev-rules.c b/src/udev/udev-rules.c index 7619c8371b..f6c416bf70 100644 --- a/src/udev/udev-rules.c +++ b/src/udev/udev-rules.c @@ -2218,10 +2218,16 @@ void udev_rules_apply_to_event(struct udev_rules *rules, rule->rule.filename_line); break; case TK_A_SECLABEL: { + char label_str[UTIL_LINE_SIZE] = {}; const char *name, *label; name = rules_str(rules, cur->key.attr_off); - label = rules_str(rules, cur->key.value_off); + udev_event_apply_format(event, rules_str(rules, cur->key.value_off), label_str, sizeof(label_str)); + if (label_str[0] != '\0') + label = label_str; + else + label = rules_str(rules, cur->key.value_off); + if (cur->key.op == OP_ASSIGN || cur->key.op == OP_ASSIGN_FINAL) udev_list_cleanup(&event->seclabel_list); udev_list_entry_add(&event->seclabel_list, name, label); diff --git a/src/vconsole/90-vconsole.rules.in b/src/vconsole/90-vconsole.rules.in index 35b9ad5151..84b4d575bd 100644 --- a/src/vconsole/90-vconsole.rules.in +++ b/src/vconsole/90-vconsole.rules.in @@ -7,4 +7,4 @@ # Each vtcon keeps its own state of fonts. # -ACTION=="add", SUBSYSTEM=="vtconsole", KERNEL=="vtcon*", RUN+="@rootlibexecdir@/systemd-vconsole-setup" +ACTION=="add", SUBSYSTEM=="vtconsole", KERNEL=="vtcon*", ATTR{name}!="*dummy device", RUN+="@rootlibexecdir@/systemd-vconsole-setup" diff --git a/src/vconsole/vconsole-setup.c b/src/vconsole/vconsole-setup.c index ac4ceb1486..a0ab5990fc 100644 --- a/src/vconsole/vconsole-setup.c +++ b/src/vconsole/vconsole-setup.c @@ -208,9 +208,9 @@ static int font_load_and_wait(const char *vc, const char *font, const char *map, */ static void setup_remaining_vcs(int fd, bool utf8) { struct console_font_op cfo = { - .op = KD_FONT_OP_GET, .flags = 0, - .width = 32, .height = 32, - .charcount = 512, + .op = KD_FONT_OP_GET, + .width = UINT_MAX, .height = UINT_MAX, + .charcount = UINT_MAX, }; struct vt_stat vcs = {}; struct unimapinit adv = {}; @@ -225,12 +225,6 @@ static void setup_remaining_vcs(int fd, bool utf8) { return; } - fontbuf = malloc(cfo.width * cfo.height * cfo.charcount / 8); - if (!fontbuf) { - log_oom(); - return; - } - /* get active, and 16 bit mask of used VT numbers */ r = ioctl(fd, VT_GETSTATE, &vcs); if (r < 0) { @@ -238,21 +232,47 @@ static void setup_remaining_vcs(int fd, bool utf8) { return; } - /* get fonts from source console */ - cfo.data = fontbuf; + /* get metadata of the current font (width, height, count) */ r = ioctl(fd, KDFONTOP, &cfo); if (r < 0) - log_warning_errno(errno, "KD_FONT_OP_GET failed, fonts will not be copied: %m"); + log_warning_errno(errno, "KD_FONT_OP_GET failed while trying to get the font metadata: %m"); else { - unimapd.entries = unipairs; - unimapd.entry_ct = USHRT_MAX; - r = ioctl(fd, GIO_UNIMAP, &unimapd); - if (r < 0) - log_warning_errno(errno, "GIO_UNIMAP failed, fonts will not be copied: %m"); - else - cfo.op = KD_FONT_OP_SET; + /* verify parameter sanity first */ + if (cfo.width > 32 || cfo.height > 32 || cfo.charcount > 512) + log_warning("Invalid font metadata - width: %u (max 32), height: %u (max 32), count: %u (max 512)", + cfo.width, cfo.height, cfo.charcount); + else { + /* + * Console fonts supported by the kernel are limited in size to 32 x 32 and maximum 512 + * characters. Thus with 1 bit per pixel it requires up to 65536 bytes. The height always + * requries 32 per glyph, regardless of the actual height - see the comment above #define + * max_font_size 65536 in drivers/tty/vt/vt.c for more details. + */ + fontbuf = malloc((cfo.width + 7) / 8 * 32 * cfo.charcount); + if (!fontbuf) { + log_oom(); + return; + } + /* get fonts from source console */ + cfo.data = fontbuf; + r = ioctl(fd, KDFONTOP, &cfo); + if (r < 0) + log_warning_errno(errno, "KD_FONT_OP_GET failed while trying to read the font data: %m"); + else { + unimapd.entries = unipairs; + unimapd.entry_ct = USHRT_MAX; + r = ioctl(fd, GIO_UNIMAP, &unimapd); + if (r < 0) + log_warning_errno(errno, "GIO_UNIMAP failed while trying to read unicode mappings: %m"); + else + cfo.op = KD_FONT_OP_SET; + } + } } + if (cfo.op != KD_FONT_OP_SET) + log_warning("Fonts will not be copied to remaining consoles"); + for (i = 1; i <= 63; i++) { char ttyname[strlen("/dev/tty") + DECIMAL_STR_MAX(int)]; _cleanup_close_ int fd_d = -1; diff --git a/units/dev-hugepages.mount b/units/dev-hugepages.mount index 882adb4545..489cc777e4 100644 --- a/units/dev-hugepages.mount +++ b/units/dev-hugepages.mount @@ -13,6 +13,7 @@ DefaultDependencies=no Before=sysinit.target ConditionPathExists=/sys/kernel/mm/hugepages ConditionCapability=CAP_SYS_ADMIN +ConditionVirtualization=!private-users [Mount] What=hugetlbfs |