1
0
mirror of https://github.com/systemd/systemd.git synced 2025-01-24 06:04:05 +03:00

Merge pull request #4845 from poettering/various-smaller-fixes

Various smaller fixes
This commit is contained in:
Zbigniew Jędrzejewski-Szmek 2016-12-17 13:36:00 -05:00 committed by GitHub
commit 117d5a27a3
16 changed files with 307 additions and 89 deletions

View File

@ -27,7 +27,7 @@ Format=raw_btrfs
Bootable=yes
[Partitions]
RootSize=2G
RootSize=3G
[Packages]
BuildPackages=
@ -37,6 +37,7 @@ BuildPackages=
bzip2-devel
cryptsetup-devel
dbus-devel
diffutils
docbook-style-xsl
elfutils-devel
gcc
@ -69,4 +70,5 @@ BuildPackages=
python3-devel
python3-lxml
qrencode-devel
tree
xz-devel

View File

@ -149,7 +149,8 @@ TESTS =
endif
AM_TESTS_ENVIRONMENT = \
export SYSTEMD_KBD_MODEL_MAP=$(abs_top_srcdir)/src/locale/kbd-model-map; \
export SYSTEMD_LANGUAGE_FALLBACK_MAP=$(abs_top_srcdir)/src/locale/language-fallback-map;
export SYSTEMD_LANGUAGE_FALLBACK_MAP=$(abs_top_srcdir)/src/locale/language-fallback-map; \
export PATH=$(abs_top_builddir):$$PATH;
if ENABLE_BASH_COMPLETION
dist_bashcompletion_DATA = $(dist_bashcompletion_data)

37
TODO
View File

@ -12,9 +12,6 @@ Bugfixes:
Environment=ONE='one' "TWO='two two' too" THREE=
ExecStart=/bin/python3 -c 'import sys;print(sys.argv)' $ONE $TWO $THREE
* When systemctl --host is used, underlying ssh connection can remain open.
bus_close does not kill children?
External:
* Fedora: add an rpmlint check that verifies that all unit files in the RPM are listed in %systemd_post macros.
@ -34,6 +31,38 @@ Features:
* replace all canonicalize_file_name() invocations by chase_symlinks(), in
particulr those where a rootdir is relevant.
* maybe introduce gpt auto discovery for /var/tmp?
* set ProtectSystem=strict for all our usual services.
* maybe add gpt-partition-based user management: each user gets his own
LUKS-encrypted GPT partition with a new GPT type. A small nss module
enumerates users via udev partition enumeration. UIDs are assigned in a fixed
way: the partition index is added as offset to some fixed base uid. User name
is stored in GPT partition name. A PAM module authenticates the user via the
LUKS partition password. Benefits: strong per-user security, compatibility
with stateless/read-only/verity-enabled root. (other idea: do this based on
loopback files in /home, without GPT involvement)
* gpt-auto logic: introduce support for discovering /var matching an image. For
that, use a partition type UUID that is hashed from the OS name (as encoded
in /etc/os-release), the architecture, and 4 new bits from the gpt flags
field of the root partition. This way can easily support multiple OS
installations on the same GPT partition table, without problems with
unmatched /var partitions.
* gpt-auto logic: related to the above, maybe support a "secondary" root
partition, that is mounted to / and is writable, and where the actual root's
/usr is mounted into.
* add dm-verity boots, and in nspawn (libcryptsetup knows this, should be relatively straight-forward)
* machined: add apis to query /etc/machine-info data of a container
* .mount and .swap units: add Format=yes|no option that formats the partition before mounting/enabling it, implicitly
* gpt-auto logic: support encrypted swap, add kernel cmdline option to force it, and honour a gpt bit about it, plus maybe a configuration file
* drop nss-myhostname in favour of nss-resolve?
* drop internal dlopen() based nss-dns fallback in nss-resolve, and rely on the
@ -279,8 +308,6 @@ Features:
* For timer units: add some mechanisms so that timer units that trigger immediately on boot do not have the services
they run added to the initial transaction and thus confuse Type=idle.
* Run most system services with cgroupfs read-only and procfs with a more secure mode (doesn't work, since the hidepid= option is per-pid-namespace, not per-mount)
* add bus api to query unit file's X fields.
* gpt-auto-generator:

View File

@ -189,10 +189,9 @@ sd_journal_send("MESSAGE=Hello World, this is PID %lu!", (unsigned long) getpid(
<refsect1>
<title>Return Value</title>
<para>The four calls return 0 on success or a negative errno-style
error code. The
<citerefentry project='man-pages'><refentrytitle>errno</refentrytitle><manvolnum>3</manvolnum></citerefentry>
variable itself is not altered.</para>
<para>The five calls return 0 on success or a negative errno-style error code. The <citerefentry
project='man-pages'><refentrytitle>errno</refentrytitle><manvolnum>3</manvolnum></citerefentry> variable itself is
not altered.</para>
<para>If
<citerefentry><refentrytitle>systemd-journald</refentrytitle><manvolnum>8</manvolnum></citerefentry>
@ -218,13 +217,11 @@ sd_journal_send("MESSAGE=Hello World, this is PID %lu!", (unsigned long) getpid(
<refsect1>
<title>Notes</title>
<para>The <function>sd_journal_print()</function>,
<function>sd_journal_printv()</function>,
<function>sd_journal_send()</function> and
<function>sd_journal_sendv()</function> interfaces are available
as a shared library, which can be compiled and linked to with the
<constant>libsystemd</constant> <citerefentry project='die-net'><refentrytitle>pkg-config</refentrytitle><manvolnum>1</manvolnum></citerefentry>
file.</para>
<para>The <function>sd_journal_print()</function>, <function>sd_journal_printv()</function>,
<function>sd_journal_send()</function>, <function>sd_journal_sendv()</function> and
<function>sd_journal_perror()</function> interfaces are available as a shared library, which can be compiled and
linked to with the <constant>libsystemd</constant> <citerefentry
project='die-net'><refentrytitle>pkg-config</refentrytitle><manvolnum>1</manvolnum></citerefentry> file.</para>
</refsect1>
<refsect1>

View File

@ -882,17 +882,21 @@ kobject-uevent 1 systemd-udevd-kernel.socket systemd-udevd.service
<term><command>show</command> <optional><replaceable>PATTERN</replaceable>…|<replaceable>JOB</replaceable></optional></term>
<listitem>
<para>Show properties of one or more units, jobs, or the
manager itself. If no argument is specified, properties of
the manager will be shown. If a unit name is specified,
properties of the unit are shown, and if a job ID is
specified, properties of the job are shown. By default, empty
properties are suppressed. Use <option>--all</option> to
show those too. To select specific properties to show, use
<option>--property=</option>. This command is intended to be
used whenever computer-parsable output is required. Use
<command>status</command> if you are looking for formatted
human-readable output.</para>
<para>Show properties of one or more units, jobs, or the manager itself. If no argument is specified,
properties of the manager will be shown. If a unit name is specified, properties of the unit are shown, and
if a job ID is specified, properties of the job are shown. By default, empty properties are suppressed. Use
<option>--all</option> to show those too. To select specific properties to show, use
<option>--property=</option>. This command is intended to be used whenever computer-parsable output is
required. Use <command>status</command> if you are looking for formatted human-readable output.</para>
<para>Many properties shown by <command>systemctl show</command> map directly to configuration settings of
the system and service manager and its unit files. Note that the properties shown by the command are
generally more low-level, normalized versions of the original configuration settings and expose runtime
state in addition to configuration. For example, properties shown for service units include the service's
current main process identifier as <literal>MainPID</literal> (which is runtime state), and time settings
are always exposed as properties ending in the <literal>…USec</literal> suffix even if a matching
configuration options end in <literal>…Sec</literal>, because microseconds is the normalized time unit used
by the system and service manager.</para>
</listitem>
</varlistentry>
<varlistentry>

View File

@ -22,4 +22,14 @@
./autogen.sh c
make -j `nproc`
make check
make install
mkdir -p $DESTDIR/etc
cat > $DESTDIR/etc/issue <<EOF
\S (built from systemd tree)
Kernel \r on an \m (\l)
EOF

View File

@ -27,6 +27,7 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/personality.h>
#include <sys/prctl.h>
#include <sys/types.h>
@ -274,27 +275,100 @@ int get_process_cmdline(pid_t pid, size_t max_length, bool comm_fallback, char *
return 0;
}
void rename_process(const char name[8]) {
assert(name);
int rename_process(const char name[]) {
static size_t mm_size = 0;
static char *mm = NULL;
bool truncated = false;
size_t l;
/* This is a like a poor man's setproctitle(). It changes the
* comm field, argv[0], and also the glibc's internally used
* name of the process. For the first one a limit of 16 chars
* applies, to the second one usually one of 10 (i.e. length
* of "/sbin/init"), to the third one one of 7 (i.e. length of
* "systemd"). If you pass a longer string it will be
* truncated */
/* This is a like a poor man's setproctitle(). It changes the comm field, argv[0], and also the glibc's
* internally used name of the process. For the first one a limit of 16 chars applies; to the second one in
* many cases one of 10 (i.e. length of "/sbin/init") however if we have CAP_SYS_RESOURCES it is unbounded;
* to the third one 7 (i.e. the length of "systemd". If you pass a longer string it will likely be
* truncated.
*
* Returns 0 if a name was set but truncated, > 0 if it was set but not truncated. */
if (isempty(name))
return -EINVAL; /* let's not confuse users unnecessarily with an empty name */
l = strlen(name);
/* First step, change the comm field. */
(void) prctl(PR_SET_NAME, name);
if (l > 15) /* Linux process names can be 15 chars at max */
truncated = true;
if (program_invocation_name)
strncpy(program_invocation_name, name, strlen(program_invocation_name));
/* Second step, change glibc's ID of the process name. */
if (program_invocation_name) {
size_t k;
k = strlen(program_invocation_name);
strncpy(program_invocation_name, name, k);
if (l > k)
truncated = true;
}
/* Third step, completely replace the argv[] array the kernel maintains for us. This requires privileges, but
* has the advantage that the argv[] array is exactly what we want it to be, and not filled up with zeros at
* the end. This is the best option for changing /proc/self/cmdline.*/
if (mm_size < l+1) {
size_t nn_size;
char *nn;
/* Let's not bother with this if we don't have euid == 0. Strictly speaking if people do weird stuff
* with capabilities this could work even for euid != 0, but our own code generally doesn't do that,
* hence let's use this as quick bypass check, to avoid calling mmap() if PR_SET_MM_ARG_START fails
* with EPERM later on anyway. After all geteuid() is dead cheap to call, but mmap() is not. */
if (geteuid() != 0) {
log_debug("Skipping PR_SET_MM_ARG_START, as we don't have privileges.");
goto use_saved_argv;
}
nn_size = PAGE_ALIGN(l+1);
nn = mmap(NULL, nn_size, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
if (nn == MAP_FAILED) {
log_debug_errno(errno, "mmap() failed: %m");
goto use_saved_argv;
}
strncpy(nn, name, nn_size);
/* Now, let's tell the kernel about this new memory */
if (prctl(PR_SET_MM, PR_SET_MM_ARG_START, (unsigned long) nn, 0, 0) < 0) {
log_debug_errno(errno, "PR_SET_MM_ARG_START failed, proceeding without: %m");
(void) munmap(nn, nn_size);
goto use_saved_argv;
}
/* And update the end pointer to the new end, too. If this fails, we don't really know what to do, it's
* pretty unlikely that we can rollback, hence we'll just accept the failure, and continue. */
if (prctl(PR_SET_MM, PR_SET_MM_ARG_END, (unsigned long) nn + l + 1, 0, 0) < 0)
log_debug_errno(errno, "PR_SET_MM_ARG_END failed, proceeding without: %m");
if (mm)
(void) munmap(mm, mm_size);
mm = nn;
mm_size = nn_size;
} else
strncpy(mm, name, mm_size);
use_saved_argv:
/* Fourth step: in all cases we'll also update the original argv[], so that our own code gets it right too if
* it still looks here */
if (saved_argc > 0) {
int i;
if (saved_argv[0])
strncpy(saved_argv[0], name, strlen(saved_argv[0]));
if (saved_argv[0]) {
size_t k;
k = strlen(saved_argv[0]);
strncpy(saved_argv[0], name, k);
if (l > k)
truncated = true;
}
for (i = 1; i < saved_argc; i++) {
if (!saved_argv[i])
@ -303,6 +377,8 @@ void rename_process(const char name[8]) {
memzero(saved_argv[i], strlen(saved_argv[i]));
}
}
return !truncated;
}
int is_kernel_thread(pid_t pid) {

View File

@ -64,7 +64,7 @@ void sigkill_waitp(pid_t *pid);
int kill_and_sigcont(pid_t pid, int sig);
void rename_process(const char name[8]);
int rename_process(const char name[]);
int is_kernel_thread(pid_t pid);
int getenv_for_pid(pid_t pid, const char *field, char **_value);

View File

@ -409,8 +409,7 @@ int detect_container(void) {
if (cached_found >= 0)
return cached_found;
/* /proc/vz exists in container and outside of the container,
* /proc/bc only outside of the container. */
/* /proc/vz exists in container and outside of the container, /proc/bc only outside of the container. */
if (access("/proc/vz", F_OK) >= 0 &&
access("/proc/bc", F_OK) < 0) {
r = VIRTUALIZATION_OPENVZ;
@ -418,50 +417,58 @@ int detect_container(void) {
}
if (getpid() == 1) {
/* If we are PID 1 we can just check our own
* environment variable */
/* If we are PID 1 we can just check our own environment variable, and that's authoritative. */
e = getenv("container");
if (isempty(e)) {
r = VIRTUALIZATION_NONE;
goto finish;
}
} else {
/* Otherwise, PID 1 dropped this information into a
* file in /run. This is better than accessing
* /proc/1/environ, since we don't need CAP_SYS_PTRACE
* for that. */
r = read_one_line_file("/run/systemd/container", &m);
if (r == -ENOENT) {
/* Fallback for cases where PID 1 was not
* systemd (for example, cases where
* init=/bin/sh is used. */
r = getenv_for_pid(1, "container", &m);
if (r <= 0) {
/* If that didn't work, give up,
* assume no container manager.
*
* Note: This means we still cannot
* detect containers if init=/bin/sh
* is passed but privileges dropped,
* as /proc/1/environ is only readable
* with privileges. */
r = VIRTUALIZATION_NONE;
goto finish;
}
}
if (r < 0)
return r;
e = m;
goto translate_name;
}
/* Otherwise, PID 1 might have dropped this information into a file in /run. This is better than accessing
* /proc/1/environ, since we don't need CAP_SYS_PTRACE for that. */
r = read_one_line_file("/run/systemd/container", &m);
if (r >= 0) {
e = m;
goto translate_name;
}
if (r != -ENOENT)
return log_debug_errno(r, "Failed to read /run/systemd/container: %m");
/* Fallback for cases where PID 1 was not systemd (for example, cases where init=/bin/sh is used. */
r = getenv_for_pid(1, "container", &m);
if (r > 0) {
e = m;
goto translate_name;
}
if (r < 0) /* This only works if we have CAP_SYS_PTRACE, hence let's better ignore failures here */
log_debug_errno(r, "Failed to read $container of PID 1, ignoring: %m");
/* Interestingly /proc/1/sched actually shows the host's PID for what we see as PID 1. Hence, if the PID shown
* there is not 1, we know we are in a PID namespace. and hence a container. */
r = read_one_line_file("/proc/1/sched", &m);
if (r >= 0) {
const char *t;
t = strrchr(m, '(');
if (!t)
return -EIO;
if (!startswith(t, "(1,")) {
r = VIRTUALIZATION_CONTAINER_OTHER;
goto finish;
}
} else if (r != -ENOENT)
return r;
/* If that didn't work, give up, assume no container manager. */
r = VIRTUALIZATION_NONE;
goto finish;
translate_name:
for (j = 0; j < ELEMENTSOF(value_table); j++)
if (streq(e, value_table[j].value)) {
r = value_table[j].id;
@ -471,7 +478,7 @@ int detect_container(void) {
r = VIRTUALIZATION_CONTAINER_OTHER;
finish:
log_debug("Found container virtualization %s", virtualization_to_string(r));
log_debug("Found container virtualization %s.", virtualization_to_string(r));
cached_found = r;
return r;
}

View File

@ -2318,7 +2318,7 @@ int main(int argc, char *argv[]) {
if (arg_boot_offset != 0 &&
sd_journal_has_runtime_files(j) > 0 &&
sd_journal_has_persistent_files(j) == 0) {
log_info("Specifying boot ID has no effect, no persistent journal was found");
log_info("Specifying boot ID or boot offset has no effect, no persistent journal was found.");
r = 0;
goto finish;
}

View File

@ -2769,7 +2769,7 @@ static int parse_argv(int argc, char *argv[]) {
assert(argv);
for (;;) {
static const char option_string[] = "-hp:als:H:M:qn:o:";
static const char option_string[] = "-hp:als:H:M:qn:o:E:";
c = getopt_long(argc, argv, option_string + reorder, options, NULL);
if (c < 0)

View File

@ -20,6 +20,7 @@
#include <sys/reboot.h>
#include <sys/unistd.h>
#include <sys/wait.h>
#include <sys/prctl.h>
#include "fd-util.h"
#include "log.h"
@ -29,7 +30,22 @@
#include "time-util.h"
#include "def.h"
int stub_pid1(void) {
static int reset_environ(const char *new_environment, size_t length) {
unsigned long start, end;
start = (unsigned long) new_environment;
end = start + length;
if (prctl(PR_SET_MM, PR_SET_MM_ENV_START, start, 0, 0) < 0)
return -errno;
if (prctl(PR_SET_MM, PR_SET_MM_ENV_END, end, 0, 0) < 0)
return -errno;
return 0;
}
int stub_pid1(sd_id128_t uuid) {
enum {
STATE_RUNNING,
STATE_REBOOT,
@ -41,6 +57,11 @@ int stub_pid1(void) {
pid_t pid;
int r;
/* The new environment we set up, on the stack. */
char new_environment[] =
"container=systemd-nspawn\0"
"container_uuid=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX";
/* Implements a stub PID 1, that reaps all processes and processes a couple of standard signals. This is useful
* for allowing arbitrary processes run in a container, and still have all zombies reaped. */
@ -64,6 +85,12 @@ int stub_pid1(void) {
close_all_fds(NULL, 0);
log_open();
/* Flush out /proc/self/environ, so that we don't leak the environment from the host into the container. Also,
* set $container= and $container_uuid= so that clients in the container that query it from /proc/1/environ
* find them set set. */
sd_id128_to_string(uuid, new_environment + sizeof(new_environment) - SD_ID128_STRING_MAX);
reset_environ(new_environment, sizeof(new_environment));
rename_process("STUBINIT");
assert_se(sigemptyset(&waitmask) >= 0);

View File

@ -19,4 +19,6 @@
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
int stub_pid1(void);
#include "sd-id128.h"
int stub_pid1(sd_id128_t uuid);

View File

@ -2278,7 +2278,7 @@ static int inner_child(
return log_error_errno(errno, "Failed to change to specified working directory %s: %m", arg_chdir);
if (arg_start_mode == START_PID2) {
r = stub_pid1();
r = stub_pid1(arg_uuid);
if (r < 0)
return r;
}

View File

@ -292,8 +292,10 @@ static int sysv_translate_facility(SysvStub *s, unsigned line, const char *name,
if (!streq(table[i], n))
continue;
if (!table[i+1])
if (!table[i+1]) {
*ret = NULL;
return 0;
}
m = strdup(table[i+1]);
if (!m)
@ -312,7 +314,7 @@ static int sysv_translate_facility(SysvStub *s, unsigned line, const char *name,
if (r < 0)
return log_error_errno(r, "[%s:%u] Could not build name for facility %s: %m", s->path, line, name);
return r;
return 1;
}
/* Strip ".sh" suffix from file name for comparison */
@ -324,8 +326,10 @@ static int sysv_translate_facility(SysvStub *s, unsigned line, const char *name,
}
/* Names equaling the file name of the services are redundant */
if (streq_ptr(n, filename))
if (streq_ptr(n, filename)) {
*ret = NULL;
return 0;
}
/* Everything else we assume to be normal service names */
m = sysv_translate_name(n);

View File

@ -355,10 +355,70 @@ static void test_get_process_cmdline_harder(void) {
_exit(0);
}
static void test_rename_process_one(const char *p, int ret) {
_cleanup_free_ char *comm = NULL, *cmdline = NULL;
pid_t pid;
int r;
pid = fork();
assert_se(pid >= 0);
if (pid > 0) {
siginfo_t si;
assert_se(wait_for_terminate(pid, &si) >= 0);
assert_se(si.si_code == CLD_EXITED);
assert_se(si.si_status == EXIT_SUCCESS);
return;
}
/* child */
r = rename_process(p);
assert_se(r == ret ||
(ret == 0 && r >= 0) ||
(ret > 0 && r > 0));
if (r < 0)
goto finish;
#ifdef HAVE_VALGRIND_VALGRIND_H
/* see above, valgrind is weird, we can't verify what we are doing here */
if (RUNNING_ON_VALGRIND)
goto finish;
#endif
assert_se(get_process_comm(0, &comm) >= 0);
log_info("comm = <%s>", comm);
assert_se(strneq(comm, p, 15));
assert_se(get_process_cmdline(0, 0, false, &cmdline) >= 0);
log_info("cmdline = <%s>", cmdline);
assert_se(strneq(p, cmdline, strlen("test-process-util")));
assert_se(startswith(p, cmdline));
finish:
_exit(EXIT_SUCCESS);
}
static void test_rename_process(void) {
test_rename_process_one(NULL, -EINVAL);
test_rename_process_one("", -EINVAL);
test_rename_process_one("foo", 1); /* should always fit */
test_rename_process_one("this is a really really long process name, followed by some more words", 0); /* unlikely to fit */
test_rename_process_one("1234567", 1); /* should always fit */
}
int main(int argc, char *argv[]) {
log_set_max_level(LOG_DEBUG);
log_parse_environment();
log_open();
saved_argc = argc;
saved_argv = argv;
if (argc > 1) {
pid_t pid = 0;
@ -373,6 +433,7 @@ int main(int argc, char *argv[]) {
test_pid_is_alive();
test_personality();
test_get_process_cmdline_harder();
test_rename_process();
return 0;
}