mirror of
https://github.com/systemd/systemd.git
synced 2024-11-14 15:21:37 +03:00
f69614f811
This makes it possible to use them from systemctl without linking against the core.
513 lines
11 KiB
C
513 lines
11 KiB
C
/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
|
|
|
|
/***
|
|
This file is part of systemd.
|
|
|
|
Copyright 2010 Lennart Poettering
|
|
|
|
systemd is free software; you can redistribute it and/or modify it
|
|
under the terms of the GNU Lesser General Public License as published by
|
|
the Free Software Foundation; either version 2.1 of the License, or
|
|
(at your option) any later version.
|
|
|
|
systemd is distributed in the hope that it will be useful, but
|
|
WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
Lesser General Public License for more details.
|
|
|
|
You should have received a copy of the GNU Lesser General Public License
|
|
along with systemd; If not, see <http://www.gnu.org/licenses/>.
|
|
***/
|
|
|
|
#include <errno.h>
|
|
#include <string.h>
|
|
#include <assert.h>
|
|
|
|
#include "path-util.h"
|
|
#include "util.h"
|
|
#include "unit-name.h"
|
|
|
|
#define VALID_CHARS \
|
|
"0123456789" \
|
|
"abcdefghijklmnopqrstuvwxyz" \
|
|
"ABCDEFGHIJKLMNOPQRSTUVWXYZ" \
|
|
":-_.\\"
|
|
|
|
static const char* const unit_type_table[_UNIT_TYPE_MAX] = {
|
|
[UNIT_SERVICE] = "service",
|
|
[UNIT_SOCKET] = "socket",
|
|
[UNIT_TARGET] = "target",
|
|
[UNIT_DEVICE] = "device",
|
|
[UNIT_MOUNT] = "mount",
|
|
[UNIT_AUTOMOUNT] = "automount",
|
|
[UNIT_SNAPSHOT] = "snapshot",
|
|
[UNIT_TIMER] = "timer",
|
|
[UNIT_SWAP] = "swap",
|
|
[UNIT_PATH] = "path",
|
|
};
|
|
|
|
DEFINE_STRING_TABLE_LOOKUP(unit_type, UnitType);
|
|
|
|
static const char* const unit_load_state_table[_UNIT_LOAD_STATE_MAX] = {
|
|
[UNIT_STUB] = "stub",
|
|
[UNIT_LOADED] = "loaded",
|
|
[UNIT_ERROR] = "error",
|
|
[UNIT_MERGED] = "merged",
|
|
[UNIT_MASKED] = "masked"
|
|
};
|
|
|
|
DEFINE_STRING_TABLE_LOOKUP(unit_load_state, UnitLoadState);
|
|
|
|
bool unit_name_is_valid(const char *n, bool template_ok) {
|
|
const char *e, *i, *at;
|
|
|
|
/* Valid formats:
|
|
*
|
|
* string@instance.suffix
|
|
* string.suffix
|
|
*/
|
|
|
|
assert(n);
|
|
|
|
if (strlen(n) >= UNIT_NAME_MAX)
|
|
return false;
|
|
|
|
e = strrchr(n, '.');
|
|
if (!e || e == n)
|
|
return false;
|
|
|
|
if (unit_type_from_string(e + 1) < 0)
|
|
return false;
|
|
|
|
for (i = n, at = NULL; i < e; i++) {
|
|
|
|
if (*i == '@' && !at)
|
|
at = i;
|
|
|
|
if (!strchr("@" VALID_CHARS, *i))
|
|
return false;
|
|
}
|
|
|
|
if (at) {
|
|
if (at == n)
|
|
return false;
|
|
|
|
if (!template_ok && at+1 == e)
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
bool unit_instance_is_valid(const char *i) {
|
|
assert(i);
|
|
|
|
/* The max length depends on the length of the string, so we
|
|
* don't really check this here. */
|
|
|
|
if (i[0] == 0)
|
|
return false;
|
|
|
|
/* We allow additional @ in the instance string, we do not
|
|
* allow them in the prefix! */
|
|
|
|
for (; *i; i++)
|
|
if (!strchr("@" VALID_CHARS, *i))
|
|
return false;
|
|
|
|
return true;
|
|
}
|
|
|
|
bool unit_prefix_is_valid(const char *p) {
|
|
|
|
/* We don't allow additional @ in the instance string */
|
|
|
|
if (p[0] == 0)
|
|
return false;
|
|
|
|
for (; *p; p++)
|
|
if (!strchr(VALID_CHARS, *p))
|
|
return false;
|
|
|
|
return true;
|
|
}
|
|
|
|
int unit_name_to_instance(const char *n, char **instance) {
|
|
const char *p, *d;
|
|
char *i;
|
|
|
|
assert(n);
|
|
assert(instance);
|
|
|
|
/* Everything past the first @ and before the last . is the instance */
|
|
p = strchr(n, '@');
|
|
if (!p) {
|
|
*instance = NULL;
|
|
return 0;
|
|
}
|
|
|
|
assert_se(d = strrchr(n, '.'));
|
|
assert(p < d);
|
|
|
|
i = strndup(p+1, d-p-1);
|
|
if (!i)
|
|
return -ENOMEM;
|
|
|
|
*instance = i;
|
|
return 0;
|
|
}
|
|
|
|
char *unit_name_to_prefix_and_instance(const char *n) {
|
|
const char *d;
|
|
|
|
assert(n);
|
|
|
|
assert_se(d = strrchr(n, '.'));
|
|
|
|
return strndup(n, d - n);
|
|
}
|
|
|
|
char *unit_name_to_prefix(const char *n) {
|
|
const char *p;
|
|
|
|
p = strchr(n, '@');
|
|
if (p)
|
|
return strndup(n, p - n);
|
|
|
|
return unit_name_to_prefix_and_instance(n);
|
|
}
|
|
|
|
char *unit_name_change_suffix(const char *n, const char *suffix) {
|
|
char *e, *r;
|
|
size_t a, b;
|
|
|
|
assert(n);
|
|
assert(unit_name_is_valid(n, true));
|
|
assert(suffix);
|
|
|
|
assert_se(e = strrchr(n, '.'));
|
|
a = e - n;
|
|
b = strlen(suffix);
|
|
|
|
r = new(char, a + b + 1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
memcpy(r, n, a);
|
|
memcpy(r+a, suffix, b+1);
|
|
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_build(const char *prefix, const char *instance, const char *suffix) {
|
|
assert(prefix);
|
|
assert(unit_prefix_is_valid(prefix));
|
|
assert(!instance || unit_instance_is_valid(instance));
|
|
assert(suffix);
|
|
|
|
if (!instance)
|
|
return strappend(prefix, suffix);
|
|
|
|
return join(prefix, "@", instance, suffix, NULL);
|
|
}
|
|
|
|
static char *do_escape_char(char c, char *t) {
|
|
*(t++) = '\\';
|
|
*(t++) = 'x';
|
|
*(t++) = hexchar(c >> 4);
|
|
*(t++) = hexchar(c);
|
|
return t;
|
|
}
|
|
|
|
static char *do_escape(const char *f, char *t) {
|
|
assert(f);
|
|
assert(t);
|
|
|
|
/* do not create units with a leading '.', like for "/.dotdir" mount points */
|
|
if (*f == '.') {
|
|
t = do_escape_char(*f, t);
|
|
f++;
|
|
}
|
|
|
|
for (; *f; f++) {
|
|
if (*f == '/')
|
|
*(t++) = '-';
|
|
else if (*f == '-' || *f == '\\' || !strchr(VALID_CHARS, *f))
|
|
t = do_escape_char(*f, t);
|
|
else
|
|
*(t++) = *f;
|
|
}
|
|
|
|
return t;
|
|
}
|
|
|
|
char *unit_name_escape(const char *f) {
|
|
char *r, *t;
|
|
|
|
r = new(char, strlen(f)*4+1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
t = do_escape(f, r);
|
|
*t = 0;
|
|
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_unescape(const char *f) {
|
|
char *r, *t;
|
|
|
|
assert(f);
|
|
|
|
r = strdup(f);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
for (t = r; *f; f++) {
|
|
if (*f == '-')
|
|
*(t++) = '/';
|
|
else if (*f == '\\') {
|
|
int a, b;
|
|
|
|
if (f[1] != 'x' ||
|
|
(a = unhexchar(f[2])) < 0 ||
|
|
(b = unhexchar(f[3])) < 0) {
|
|
/* Invalid escape code, let's take it literal then */
|
|
*(t++) = '\\';
|
|
} else {
|
|
*(t++) = (char) ((a << 4) | b);
|
|
f += 3;
|
|
}
|
|
} else
|
|
*(t++) = *f;
|
|
}
|
|
|
|
*t = 0;
|
|
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_path_escape(const char *f) {
|
|
char *p, *e;
|
|
|
|
assert(f);
|
|
|
|
p = strdup(f);
|
|
if (!p)
|
|
return NULL;
|
|
|
|
path_kill_slashes(p);
|
|
|
|
if (streq(p, "/")) {
|
|
free(p);
|
|
return strdup("-");
|
|
}
|
|
|
|
e = unit_name_escape(p[0] == '/' ? p + 1 : p);
|
|
free(p);
|
|
|
|
return e;
|
|
}
|
|
|
|
char *unit_name_path_unescape(const char *f) {
|
|
char *e;
|
|
|
|
assert(f);
|
|
|
|
e = unit_name_unescape(f);
|
|
if (!e)
|
|
return NULL;
|
|
|
|
if (e[0] != '/') {
|
|
char *w;
|
|
|
|
w = strappend("/", e);
|
|
free(e);
|
|
|
|
return w;
|
|
}
|
|
|
|
return e;
|
|
}
|
|
|
|
bool unit_name_is_template(const char *n) {
|
|
const char *p;
|
|
|
|
assert(n);
|
|
|
|
if (!(p = strchr(n, '@')))
|
|
return false;
|
|
|
|
return p[1] == '.';
|
|
}
|
|
|
|
char *unit_name_replace_instance(const char *f, const char *i) {
|
|
const char *p, *e;
|
|
char *r, *k;
|
|
size_t a;
|
|
|
|
assert(f);
|
|
|
|
p = strchr(f, '@');
|
|
assert_se(e = strrchr(f, '.'));
|
|
|
|
a = p - f;
|
|
|
|
if (p) {
|
|
size_t b;
|
|
|
|
b = strlen(i);
|
|
|
|
r = new(char, a + 1 + b + strlen(e) + 1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
k = mempcpy(r, f, a + 1);
|
|
k = mempcpy(k, i, b);
|
|
} else {
|
|
|
|
r = new(char, a + strlen(e) + 1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
k = mempcpy(r, f, a);
|
|
}
|
|
|
|
strcpy(k, e);
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_template(const char *f) {
|
|
const char *p, *e;
|
|
char *r;
|
|
size_t a;
|
|
|
|
p = strchr(f, '@');
|
|
if (!p)
|
|
return strdup(f);
|
|
|
|
assert_se(e = strrchr(f, '.'));
|
|
a = p - f + 1;
|
|
|
|
r = new(char, a + strlen(e) + 1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
strcpy(mempcpy(r, f, a), e);
|
|
return r;
|
|
|
|
}
|
|
|
|
char *unit_name_from_path(const char *path, const char *suffix) {
|
|
char *p, *r;
|
|
|
|
assert(path);
|
|
assert(suffix);
|
|
|
|
p = unit_name_path_escape(path);
|
|
if (!p)
|
|
return NULL;
|
|
|
|
r = strappend(p, suffix);
|
|
free(p);
|
|
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_from_path_instance(const char *prefix, const char *path, const char *suffix) {
|
|
char *p, *r;
|
|
|
|
assert(prefix);
|
|
assert(path);
|
|
assert(suffix);
|
|
|
|
p = unit_name_path_escape(path);
|
|
if (!p)
|
|
return NULL;
|
|
|
|
r = join(prefix, "@", p, suffix, NULL);
|
|
free(p);
|
|
|
|
return r;
|
|
}
|
|
|
|
char *unit_name_to_path(const char *name) {
|
|
char *w, *e;
|
|
|
|
assert(name);
|
|
|
|
w = unit_name_to_prefix(name);
|
|
if (!w)
|
|
return NULL;
|
|
|
|
e = unit_name_path_unescape(w);
|
|
free(w);
|
|
|
|
return e;
|
|
}
|
|
|
|
char *unit_dbus_path_from_name(const char *name) {
|
|
char *e, *p;
|
|
|
|
assert(name);
|
|
|
|
e = bus_path_escape(name);
|
|
if (!e)
|
|
return NULL;
|
|
|
|
p = strappend("/org/freedesktop/systemd1/unit/", e);
|
|
free(e);
|
|
|
|
return p;
|
|
}
|
|
|
|
char *unit_name_mangle(const char *name) {
|
|
char *r, *t;
|
|
const char *f;
|
|
|
|
assert(name);
|
|
|
|
/* Try to turn a string that might not be a unit name into a
|
|
* sensible unit name. */
|
|
|
|
if (path_startswith(name, "/dev/") ||
|
|
path_startswith(name, "/sys/"))
|
|
return unit_name_from_path(name, ".device");
|
|
|
|
if (path_is_absolute(name))
|
|
return unit_name_from_path(name, ".mount");
|
|
|
|
/* We'll only escape the obvious characters here, to play
|
|
* safe. */
|
|
|
|
r = new(char, strlen(name) * 4 + 1);
|
|
if (!r)
|
|
return NULL;
|
|
|
|
for (f = name, t = r; *f; f++) {
|
|
|
|
if (*f == '/')
|
|
*(t++) = '-';
|
|
else if (!strchr("@" VALID_CHARS, *f))
|
|
t = do_escape_char(*f, t);
|
|
else
|
|
*(t++) = *f;
|
|
}
|
|
|
|
*t = 0;
|
|
|
|
return r;
|
|
}
|
|
|
|
UnitType unit_name_to_type(const char *n) {
|
|
const char *e;
|
|
|
|
assert(n);
|
|
|
|
e = strrchr(n, '.');
|
|
if (!e)
|
|
return _UNIT_TYPE_INVALID;
|
|
|
|
return unit_type_from_string(e + 1);
|
|
}
|