]>
Commit | Line | Data |
---|---|---|
0ad19a3f | 1 | /* |
2 | * lxc: linux Container library | |
3 | * | |
4 | * (C) Copyright IBM Corp. 2007, 2008 | |
5 | * | |
6 | * Authors: | |
9afe19d6 | 7 | * Daniel Lezcano <daniel.lezcano at free.fr> |
0ad19a3f | 8 | * |
9 | * This library is free software; you can redistribute it and/or | |
10 | * modify it under the terms of the GNU Lesser General Public | |
11 | * License as published by the Free Software Foundation; either | |
12 | * version 2.1 of the License, or (at your option) any later version. | |
13 | * | |
14 | * This library is distributed in the hope that it will be useful, | |
15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
17 | * Lesser General Public License for more details. | |
18 | * | |
19 | * You should have received a copy of the GNU Lesser General Public | |
20 | * License along with this library; if not, write to the Free Software | |
250b1eec | 21 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
0ad19a3f | 22 | */ |
23 | #define _GNU_SOURCE | |
24 | #include <stdio.h> | |
25 | #undef _GNU_SOURCE | |
26 | #include <stdlib.h> | |
e3b4c4c4 | 27 | #include <stdarg.h> |
0ad19a3f | 28 | #include <errno.h> |
29 | #include <string.h> | |
30 | #include <dirent.h> | |
0ad19a3f | 31 | #include <unistd.h> |
e3b4c4c4 | 32 | #include <sys/wait.h> |
2d76d1d7 | 33 | #include <sys/syscall.h> |
4a0ba80d | 34 | #include <time.h> |
e827ff7e | 35 | |
4ba0d9af SG |
36 | #if HAVE_IFADDRS_H |
37 | #include <ifaddrs.h> | |
38 | #else | |
39 | #include <../include/ifaddrs.h> | |
40 | #endif | |
41 | ||
e827ff7e | 42 | #if HAVE_PTY_H |
b0a33c1e | 43 | #include <pty.h> |
e827ff7e SG |
44 | #else |
45 | #include <../include/openpty.h> | |
46 | #endif | |
0ad19a3f | 47 | |
b3ecde1e DL |
48 | #include <linux/loop.h> |
49 | ||
0ad19a3f | 50 | #include <sys/types.h> |
51 | #include <sys/utsname.h> | |
52 | #include <sys/param.h> | |
53 | #include <sys/stat.h> | |
54 | #include <sys/socket.h> | |
55 | #include <sys/mount.h> | |
56 | #include <sys/mman.h> | |
81810dd1 | 57 | #include <sys/prctl.h> |
0ad19a3f | 58 | |
59 | #include <arpa/inet.h> | |
60 | #include <fcntl.h> | |
61 | #include <netinet/in.h> | |
62 | #include <net/if.h> | |
6f4a3756 | 63 | #include <libgen.h> |
0ad19a3f | 64 | |
e5bda9ee | 65 | #include "network.h" |
66 | #include "error.h" | |
b2718c72 | 67 | #include "parse.h" |
881450bb | 68 | #include "config.h" |
1b09f2c0 DL |
69 | #include "utils.h" |
70 | #include "conf.h" | |
71 | #include "log.h" | |
72 | #include "lxc.h" /* for lxc_cgroup_set() */ | |
d55bc1ad | 73 | #include "caps.h" /* for lxc_caps_last_cap() */ |
9be53773 | 74 | #include "bdev.h" |
368bbc02 | 75 | #include "cgroup.h" |
025ed0f3 | 76 | #include "lxclock.h" |
fe4de9a6 | 77 | #include "lsm/lsm.h" |
d0a36f2c | 78 | |
495d2046 SG |
79 | #if HAVE_SYS_CAPABILITY_H |
80 | #include <sys/capability.h> | |
81 | #endif | |
82 | ||
6ff05e18 SG |
83 | #if HAVE_SYS_PERSONALITY_H |
84 | #include <sys/personality.h> | |
85 | #endif | |
86 | ||
edaf8b1b SG |
87 | #if IS_BIONIC |
88 | #include <../include/lxcmntent.h> | |
89 | #else | |
90 | #include <mntent.h> | |
91 | #endif | |
92 | ||
769872f9 SH |
93 | #include "lxcseccomp.h" |
94 | ||
36eb9bde | 95 | lxc_log_define(lxc_conf, lxc); |
e5bda9ee | 96 | |
0ad19a3f | 97 | #define MAXHWLEN 18 |
98 | #define MAXINDEXLEN 20 | |
442cbbe6 | 99 | #define MAXMTULEN 16 |
0ad19a3f | 100 | #define MAXLINELEN 128 |
101 | ||
495d2046 | 102 | #if HAVE_SYS_CAPABILITY_H |
b09094da MN |
103 | #ifndef CAP_SETFCAP |
104 | #define CAP_SETFCAP 31 | |
105 | #endif | |
106 | ||
107 | #ifndef CAP_MAC_OVERRIDE | |
108 | #define CAP_MAC_OVERRIDE 32 | |
109 | #endif | |
110 | ||
111 | #ifndef CAP_MAC_ADMIN | |
112 | #define CAP_MAC_ADMIN 33 | |
113 | #endif | |
495d2046 | 114 | #endif |
b09094da MN |
115 | |
116 | #ifndef PR_CAPBSET_DROP | |
117 | #define PR_CAPBSET_DROP 24 | |
118 | #endif | |
119 | ||
9818cae4 SG |
120 | #ifndef LO_FLAGS_AUTOCLEAR |
121 | #define LO_FLAGS_AUTOCLEAR 4 | |
122 | #endif | |
123 | ||
2d76d1d7 SG |
124 | /* Define pivot_root() if missing from the C library */ |
125 | #ifndef HAVE_PIVOT_ROOT | |
126 | static int pivot_root(const char * new_root, const char * put_old) | |
127 | { | |
128 | #ifdef __NR_pivot_root | |
129 | return syscall(__NR_pivot_root, new_root, put_old); | |
130 | #else | |
131 | errno = ENOSYS; | |
132 | return -1; | |
133 | #endif | |
134 | } | |
135 | #else | |
136 | extern int pivot_root(const char * new_root, const char * put_old); | |
137 | #endif | |
138 | ||
139 | /* Define sethostname() if missing from the C library */ | |
140 | #ifndef HAVE_SETHOSTNAME | |
141 | static int sethostname(const char * name, size_t len) | |
142 | { | |
143 | #ifdef __NR_sethostname | |
144 | return syscall(__NR_sethostname, name, len); | |
145 | #else | |
146 | errno = ENOSYS; | |
147 | return -1; | |
148 | #endif | |
149 | } | |
150 | #endif | |
151 | ||
72f919c4 SG |
152 | /* Define __S_ISTYPE if missing from the C library */ |
153 | #ifndef __S_ISTYPE | |
154 | #define __S_ISTYPE(mode, mask) (((mode) & S_IFMT) == (mask)) | |
155 | #endif | |
156 | ||
72d0e1cb | 157 | char *lxchook_names[NUM_LXC_HOOKS] = { |
148e91f5 | 158 | "pre-start", "pre-mount", "mount", "autodev", "start", "post-stop", "clone" }; |
72d0e1cb | 159 | |
e3b4c4c4 | 160 | typedef int (*instanciate_cb)(struct lxc_handler *, struct lxc_netdev *); |
0ad19a3f | 161 | |
998ac676 RT |
162 | struct mount_opt { |
163 | char *name; | |
164 | int clear; | |
165 | int flag; | |
166 | }; | |
167 | ||
81810dd1 DL |
168 | struct caps_opt { |
169 | char *name; | |
170 | int value; | |
171 | }; | |
172 | ||
e3b4c4c4 ST |
173 | static int instanciate_veth(struct lxc_handler *, struct lxc_netdev *); |
174 | static int instanciate_macvlan(struct lxc_handler *, struct lxc_netdev *); | |
175 | static int instanciate_vlan(struct lxc_handler *, struct lxc_netdev *); | |
176 | static int instanciate_phys(struct lxc_handler *, struct lxc_netdev *); | |
177 | static int instanciate_empty(struct lxc_handler *, struct lxc_netdev *); | |
82d5ae15 | 178 | |
24654103 DL |
179 | static instanciate_cb netdev_conf[LXC_NET_MAXCONFTYPE + 1] = { |
180 | [LXC_NET_VETH] = instanciate_veth, | |
181 | [LXC_NET_MACVLAN] = instanciate_macvlan, | |
182 | [LXC_NET_VLAN] = instanciate_vlan, | |
183 | [LXC_NET_PHYS] = instanciate_phys, | |
184 | [LXC_NET_EMPTY] = instanciate_empty, | |
0ad19a3f | 185 | }; |
186 | ||
74a2b586 JK |
187 | static int shutdown_veth(struct lxc_handler *, struct lxc_netdev *); |
188 | static int shutdown_macvlan(struct lxc_handler *, struct lxc_netdev *); | |
189 | static int shutdown_vlan(struct lxc_handler *, struct lxc_netdev *); | |
190 | static int shutdown_phys(struct lxc_handler *, struct lxc_netdev *); | |
191 | static int shutdown_empty(struct lxc_handler *, struct lxc_netdev *); | |
192 | ||
193 | static instanciate_cb netdev_deconf[LXC_NET_MAXCONFTYPE + 1] = { | |
194 | [LXC_NET_VETH] = shutdown_veth, | |
195 | [LXC_NET_MACVLAN] = shutdown_macvlan, | |
196 | [LXC_NET_VLAN] = shutdown_vlan, | |
197 | [LXC_NET_PHYS] = shutdown_phys, | |
198 | [LXC_NET_EMPTY] = shutdown_empty, | |
199 | }; | |
200 | ||
998ac676 | 201 | static struct mount_opt mount_opt[] = { |
88d413d5 SW |
202 | { "defaults", 0, 0 }, |
203 | { "ro", 0, MS_RDONLY }, | |
204 | { "rw", 1, MS_RDONLY }, | |
205 | { "suid", 1, MS_NOSUID }, | |
206 | { "nosuid", 0, MS_NOSUID }, | |
207 | { "dev", 1, MS_NODEV }, | |
208 | { "nodev", 0, MS_NODEV }, | |
209 | { "exec", 1, MS_NOEXEC }, | |
210 | { "noexec", 0, MS_NOEXEC }, | |
211 | { "sync", 0, MS_SYNCHRONOUS }, | |
212 | { "async", 1, MS_SYNCHRONOUS }, | |
213 | { "dirsync", 0, MS_DIRSYNC }, | |
214 | { "remount", 0, MS_REMOUNT }, | |
215 | { "mand", 0, MS_MANDLOCK }, | |
216 | { "nomand", 1, MS_MANDLOCK }, | |
217 | { "atime", 1, MS_NOATIME }, | |
218 | { "noatime", 0, MS_NOATIME }, | |
219 | { "diratime", 1, MS_NODIRATIME }, | |
220 | { "nodiratime", 0, MS_NODIRATIME }, | |
221 | { "bind", 0, MS_BIND }, | |
222 | { "rbind", 0, MS_BIND|MS_REC }, | |
223 | { "relatime", 0, MS_RELATIME }, | |
224 | { "norelatime", 1, MS_RELATIME }, | |
225 | { "strictatime", 0, MS_STRICTATIME }, | |
226 | { "nostrictatime", 1, MS_STRICTATIME }, | |
227 | { NULL, 0, 0 }, | |
998ac676 RT |
228 | }; |
229 | ||
495d2046 | 230 | #if HAVE_SYS_CAPABILITY_H |
81810dd1 | 231 | static struct caps_opt caps_opt[] = { |
a6afdde9 | 232 | { "chown", CAP_CHOWN }, |
1e11be34 DL |
233 | { "dac_override", CAP_DAC_OVERRIDE }, |
234 | { "dac_read_search", CAP_DAC_READ_SEARCH }, | |
235 | { "fowner", CAP_FOWNER }, | |
236 | { "fsetid", CAP_FSETID }, | |
81810dd1 DL |
237 | { "kill", CAP_KILL }, |
238 | { "setgid", CAP_SETGID }, | |
239 | { "setuid", CAP_SETUID }, | |
240 | { "setpcap", CAP_SETPCAP }, | |
241 | { "linux_immutable", CAP_LINUX_IMMUTABLE }, | |
242 | { "net_bind_service", CAP_NET_BIND_SERVICE }, | |
243 | { "net_broadcast", CAP_NET_BROADCAST }, | |
244 | { "net_admin", CAP_NET_ADMIN }, | |
245 | { "net_raw", CAP_NET_RAW }, | |
246 | { "ipc_lock", CAP_IPC_LOCK }, | |
247 | { "ipc_owner", CAP_IPC_OWNER }, | |
248 | { "sys_module", CAP_SYS_MODULE }, | |
249 | { "sys_rawio", CAP_SYS_RAWIO }, | |
250 | { "sys_chroot", CAP_SYS_CHROOT }, | |
251 | { "sys_ptrace", CAP_SYS_PTRACE }, | |
252 | { "sys_pacct", CAP_SYS_PACCT }, | |
253 | { "sys_admin", CAP_SYS_ADMIN }, | |
254 | { "sys_boot", CAP_SYS_BOOT }, | |
255 | { "sys_nice", CAP_SYS_NICE }, | |
256 | { "sys_resource", CAP_SYS_RESOURCE }, | |
257 | { "sys_time", CAP_SYS_TIME }, | |
258 | { "sys_tty_config", CAP_SYS_TTY_CONFIG }, | |
259 | { "mknod", CAP_MKNOD }, | |
260 | { "lease", CAP_LEASE }, | |
9527e566 | 261 | #ifdef CAP_AUDIT_WRITE |
81810dd1 | 262 | { "audit_write", CAP_AUDIT_WRITE }, |
9527e566 FW |
263 | #endif |
264 | #ifdef CAP_AUDIT_CONTROL | |
81810dd1 | 265 | { "audit_control", CAP_AUDIT_CONTROL }, |
9527e566 | 266 | #endif |
81810dd1 DL |
267 | { "setfcap", CAP_SETFCAP }, |
268 | { "mac_override", CAP_MAC_OVERRIDE }, | |
269 | { "mac_admin", CAP_MAC_ADMIN }, | |
5170c716 CS |
270 | #ifdef CAP_SYSLOG |
271 | { "syslog", CAP_SYSLOG }, | |
272 | #endif | |
273 | #ifdef CAP_WAKE_ALARM | |
274 | { "wake_alarm", CAP_WAKE_ALARM }, | |
275 | #endif | |
81810dd1 | 276 | }; |
495d2046 SG |
277 | #else |
278 | static struct caps_opt caps_opt[] = {}; | |
279 | #endif | |
81810dd1 | 280 | |
4a0ba80d SG |
281 | static char padchar[] = |
282 | "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"; | |
283 | ||
284 | static char *mkifname(char *template) | |
285 | { | |
286 | char *name = NULL; | |
287 | int i = 0; | |
288 | FILE *urandom; | |
289 | unsigned int seed; | |
4a0ba80d SG |
290 | struct ifaddrs *ifaddr, *ifa; |
291 | int ifexists = 0; | |
292 | ||
293 | /* Get all the network interfaces */ | |
294 | getifaddrs(&ifaddr); | |
295 | ||
296 | /* Initialize the random number generator */ | |
025ed0f3 | 297 | process_lock(); |
4a0ba80d | 298 | urandom = fopen ("/dev/urandom", "r"); |
025ed0f3 | 299 | process_unlock(); |
4a0ba80d SG |
300 | if (urandom != NULL) { |
301 | if (fread (&seed, sizeof(seed), 1, urandom) <= 0) | |
302 | seed = time(0); | |
025ed0f3 | 303 | process_lock(); |
4a0ba80d | 304 | fclose(urandom); |
025ed0f3 | 305 | process_unlock(); |
4a0ba80d SG |
306 | } |
307 | else | |
308 | seed = time(0); | |
7f3e12f3 SG |
309 | |
310 | #ifndef HAVE_RAND_R | |
311 | srand(seed); | |
312 | #endif | |
4a0ba80d SG |
313 | |
314 | /* Generate random names until we find one that doesn't exist */ | |
315 | while(1) { | |
316 | ifexists = 0; | |
317 | name = strdup(template); | |
318 | ||
319 | if (name == NULL) | |
320 | return NULL; | |
321 | ||
322 | for (i = 0; i < strlen(name); i++) { | |
323 | if (name[i] == 'X') { | |
7f3e12f3 SG |
324 | #ifdef HAVE_RAND_R |
325 | name[i] = padchar[rand_r(&seed) % (strlen(padchar) - 1)]; | |
326 | #else | |
327 | name[i] = padchar[rand() % (strlen(padchar) - 1)]; | |
328 | #endif | |
4a0ba80d SG |
329 | } |
330 | } | |
331 | ||
332 | for (ifa = ifaddr; ifa != NULL; ifa = ifa->ifa_next) { | |
333 | if (strcmp(ifa->ifa_name, name) == 0) { | |
334 | ifexists = 1; | |
335 | break; | |
336 | } | |
337 | } | |
338 | ||
339 | if (ifexists == 0) | |
340 | break; | |
341 | ||
342 | free(name); | |
343 | } | |
344 | ||
345 | freeifaddrs(ifaddr); | |
346 | return name; | |
347 | } | |
348 | ||
91c3830e SH |
349 | static int run_buffer(char *buffer) |
350 | { | |
351 | FILE *f; | |
352 | char *output; | |
8e7da691 | 353 | int ret; |
91c3830e | 354 | |
025ed0f3 | 355 | process_lock(); |
91c3830e | 356 | f = popen(buffer, "r"); |
025ed0f3 | 357 | process_unlock(); |
91c3830e SH |
358 | if (!f) { |
359 | SYSERROR("popen failed"); | |
360 | return -1; | |
361 | } | |
362 | ||
363 | output = malloc(LXC_LOG_BUFFER_SIZE); | |
364 | if (!output) { | |
365 | ERROR("failed to allocate memory for script output"); | |
025ed0f3 | 366 | process_lock(); |
00b6be44 | 367 | pclose(f); |
025ed0f3 | 368 | process_unlock(); |
91c3830e SH |
369 | return -1; |
370 | } | |
371 | ||
372 | while(fgets(output, LXC_LOG_BUFFER_SIZE, f)) | |
373 | DEBUG("script output: %s", output); | |
374 | ||
375 | free(output); | |
376 | ||
025ed0f3 | 377 | process_lock(); |
8e7da691 | 378 | ret = pclose(f); |
025ed0f3 | 379 | process_unlock(); |
8e7da691 | 380 | if (ret == -1) { |
91c3830e SH |
381 | SYSERROR("Script exited on error"); |
382 | return -1; | |
8e7da691 DE |
383 | } else if (WIFEXITED(ret) && WEXITSTATUS(ret) != 0) { |
384 | ERROR("Script exited with status %d", WEXITSTATUS(ret)); | |
385 | return -1; | |
386 | } else if (WIFSIGNALED(ret)) { | |
387 | ERROR("Script terminated by signal %d (%s)", WTERMSIG(ret), | |
388 | strsignal(WTERMSIG(ret))); | |
389 | return -1; | |
91c3830e SH |
390 | } |
391 | ||
392 | return 0; | |
393 | } | |
394 | ||
148e91f5 | 395 | static int run_script_argv(const char *name, const char *section, |
283678ed SH |
396 | const char *script, const char *hook, const char *lxcpath, |
397 | char **argsin) | |
148e91f5 SH |
398 | { |
399 | int ret, i; | |
400 | char *buffer; | |
401 | size_t size = 0; | |
402 | ||
403 | INFO("Executing script '%s' for container '%s', config section '%s'", | |
404 | script, name, section); | |
405 | ||
406 | for (i=0; argsin && argsin[i]; i++) | |
407 | size += strlen(argsin[i]) + 1; | |
408 | ||
409 | size += strlen(hook) + 1; | |
410 | ||
411 | size += strlen(script); | |
412 | size += strlen(name); | |
413 | size += strlen(section); | |
414 | size += 3; | |
415 | ||
416 | if (size > INT_MAX) | |
417 | return -1; | |
418 | ||
419 | buffer = alloca(size); | |
420 | if (!buffer) { | |
421 | ERROR("failed to allocate memory"); | |
422 | return -1; | |
423 | } | |
424 | ||
425 | ret = snprintf(buffer, size, "%s %s %s %s", script, name, section, hook); | |
426 | if (ret < 0 || ret >= size) { | |
427 | ERROR("Script name too long"); | |
428 | return -1; | |
429 | } | |
430 | ||
431 | for (i=0; argsin && argsin[i]; i++) { | |
432 | int len = size-ret; | |
433 | int rc; | |
434 | rc = snprintf(buffer + ret, len, " %s", argsin[i]); | |
435 | if (rc < 0 || rc >= len) { | |
436 | ERROR("Script args too long"); | |
437 | return -1; | |
438 | } | |
439 | ret += rc; | |
440 | } | |
441 | ||
442 | return run_buffer(buffer); | |
443 | } | |
444 | ||
751d9dcd DL |
445 | static int run_script(const char *name, const char *section, |
446 | const char *script, ...) | |
e3b4c4c4 | 447 | { |
abbfd20b | 448 | int ret; |
91c3830e | 449 | char *buffer, *p; |
abbfd20b DL |
450 | size_t size = 0; |
451 | va_list ap; | |
751d9dcd DL |
452 | |
453 | INFO("Executing script '%s' for container '%s', config section '%s'", | |
454 | script, name, section); | |
e3b4c4c4 | 455 | |
abbfd20b DL |
456 | va_start(ap, script); |
457 | while ((p = va_arg(ap, char *))) | |
95642a10 | 458 | size += strlen(p) + 1; |
abbfd20b DL |
459 | va_end(ap); |
460 | ||
461 | size += strlen(script); | |
462 | size += strlen(name); | |
463 | size += strlen(section); | |
95642a10 | 464 | size += 3; |
abbfd20b | 465 | |
95642a10 MS |
466 | if (size > INT_MAX) |
467 | return -1; | |
468 | ||
469 | buffer = alloca(size); | |
abbfd20b DL |
470 | if (!buffer) { |
471 | ERROR("failed to allocate memory"); | |
751d9dcd DL |
472 | return -1; |
473 | } | |
474 | ||
9ba8130c SH |
475 | ret = snprintf(buffer, size, "%s %s %s", script, name, section); |
476 | if (ret < 0 || ret >= size) { | |
477 | ERROR("Script name too long"); | |
9ba8130c SH |
478 | return -1; |
479 | } | |
751d9dcd | 480 | |
abbfd20b | 481 | va_start(ap, script); |
9ba8130c SH |
482 | while ((p = va_arg(ap, char *))) { |
483 | int len = size-ret; | |
484 | int rc; | |
485 | rc = snprintf(buffer + ret, len, " %s", p); | |
486 | if (rc < 0 || rc >= len) { | |
9ba8130c SH |
487 | ERROR("Script args too long"); |
488 | return -1; | |
489 | } | |
490 | ret += rc; | |
491 | } | |
abbfd20b | 492 | va_end(ap); |
751d9dcd | 493 | |
91c3830e | 494 | return run_buffer(buffer); |
e3b4c4c4 ST |
495 | } |
496 | ||
a6afdde9 | 497 | static int find_fstype_cb(char* buffer, void *data) |
78ae2fcc | 498 | { |
499 | struct cbarg { | |
500 | const char *rootfs; | |
a6afdde9 | 501 | const char *target; |
78ae2fcc | 502 | int mntopt; |
503 | } *cbarg = data; | |
504 | ||
505 | char *fstype; | |
506 | ||
507 | /* we don't try 'nodev' entries */ | |
508 | if (strstr(buffer, "nodev")) | |
509 | return 0; | |
510 | ||
511 | fstype = buffer; | |
b2718c72 | 512 | fstype += lxc_char_left_gc(fstype, strlen(fstype)); |
513 | fstype[lxc_char_right_gc(fstype, strlen(fstype))] = '\0'; | |
78ae2fcc | 514 | |
a6afdde9 DL |
515 | DEBUG("trying to mount '%s'->'%s' with fstype '%s'", |
516 | cbarg->rootfs, cbarg->target, fstype); | |
517 | ||
518 | if (mount(cbarg->rootfs, cbarg->target, fstype, cbarg->mntopt, NULL)) { | |
519 | DEBUG("mount failed with error: %s", strerror(errno)); | |
78ae2fcc | 520 | return 0; |
a6afdde9 | 521 | } |
78ae2fcc | 522 | |
a6afdde9 DL |
523 | INFO("mounted '%s' on '%s', with fstype '%s'", |
524 | cbarg->rootfs, cbarg->target, fstype); | |
78ae2fcc | 525 | |
526 | return 1; | |
527 | } | |
528 | ||
2656d231 | 529 | static int mount_unknow_fs(const char *rootfs, const char *target, int mntopt) |
78ae2fcc | 530 | { |
a6afdde9 | 531 | int i; |
78ae2fcc | 532 | |
533 | struct cbarg { | |
534 | const char *rootfs; | |
a6afdde9 | 535 | const char *target; |
78ae2fcc | 536 | int mntopt; |
537 | } cbarg = { | |
538 | .rootfs = rootfs, | |
a6afdde9 | 539 | .target = target, |
78ae2fcc | 540 | .mntopt = mntopt, |
541 | }; | |
542 | ||
a6afdde9 DL |
543 | /* |
544 | * find the filesystem type with brute force: | |
545 | * first we check with /etc/filesystems, in case the modules | |
78ae2fcc | 546 | * are auto-loaded and fall back to the supported kernel fs |
547 | */ | |
548 | char *fsfile[] = { | |
549 | "/etc/filesystems", | |
550 | "/proc/filesystems", | |
551 | }; | |
552 | ||
a6afdde9 DL |
553 | for (i = 0; i < sizeof(fsfile)/sizeof(fsfile[0]); i++) { |
554 | ||
555 | int ret; | |
556 | ||
557 | if (access(fsfile[i], F_OK)) | |
558 | continue; | |
559 | ||
560 | ret = lxc_file_for_each_line(fsfile[i], find_fstype_cb, &cbarg); | |
561 | if (ret < 0) { | |
562 | ERROR("failed to parse '%s'", fsfile[i]); | |
563 | return -1; | |
564 | } | |
565 | ||
566 | if (ret) | |
567 | return 0; | |
78ae2fcc | 568 | } |
569 | ||
a6afdde9 DL |
570 | ERROR("failed to determine fs type for '%s'", rootfs); |
571 | return -1; | |
572 | } | |
573 | ||
2656d231 | 574 | static int mount_rootfs_dir(const char *rootfs, const char *target) |
a6afdde9 DL |
575 | { |
576 | return mount(rootfs, target, "none", MS_BIND | MS_REC, NULL); | |
577 | } | |
578 | ||
579 | static int setup_lodev(const char *rootfs, int fd, struct loop_info64 *loinfo) | |
580 | { | |
581 | int rfd; | |
582 | int ret = -1; | |
583 | ||
025ed0f3 | 584 | process_lock(); |
a6afdde9 | 585 | rfd = open(rootfs, O_RDWR); |
025ed0f3 | 586 | process_unlock(); |
a6afdde9 DL |
587 | if (rfd < 0) { |
588 | SYSERROR("failed to open '%s'", rootfs); | |
78ae2fcc | 589 | return -1; |
590 | } | |
591 | ||
a6afdde9 | 592 | memset(loinfo, 0, sizeof(*loinfo)); |
78ae2fcc | 593 | |
a6afdde9 | 594 | loinfo->lo_flags = LO_FLAGS_AUTOCLEAR; |
78ae2fcc | 595 | |
a6afdde9 DL |
596 | if (ioctl(fd, LOOP_SET_FD, rfd)) { |
597 | SYSERROR("failed to LOOP_SET_FD"); | |
598 | goto out; | |
78ae2fcc | 599 | } |
600 | ||
a6afdde9 DL |
601 | if (ioctl(fd, LOOP_SET_STATUS64, loinfo)) { |
602 | SYSERROR("failed to LOOP_SET_STATUS64"); | |
78ae2fcc | 603 | goto out; |
604 | } | |
605 | ||
a6afdde9 | 606 | ret = 0; |
78ae2fcc | 607 | out: |
025ed0f3 | 608 | process_lock(); |
a6afdde9 | 609 | close(rfd); |
025ed0f3 | 610 | process_unlock(); |
78ae2fcc | 611 | |
a6afdde9 | 612 | return ret; |
78ae2fcc | 613 | } |
614 | ||
2656d231 | 615 | static int mount_rootfs_file(const char *rootfs, const char *target) |
78ae2fcc | 616 | { |
a6afdde9 DL |
617 | struct dirent dirent, *direntp; |
618 | struct loop_info64 loinfo; | |
9ba8130c | 619 | int ret = -1, fd = -1, rc; |
a6afdde9 DL |
620 | DIR *dir; |
621 | char path[MAXPATHLEN]; | |
78ae2fcc | 622 | |
025ed0f3 | 623 | process_lock(); |
a6afdde9 | 624 | dir = opendir("/dev"); |
025ed0f3 | 625 | process_unlock(); |
a6afdde9 DL |
626 | if (!dir) { |
627 | SYSERROR("failed to open '/dev'"); | |
78ae2fcc | 628 | return -1; |
629 | } | |
630 | ||
a6afdde9 DL |
631 | while (!readdir_r(dir, &dirent, &direntp)) { |
632 | ||
633 | if (!direntp) | |
634 | break; | |
635 | ||
636 | if (!strcmp(direntp->d_name, ".")) | |
637 | continue; | |
638 | ||
639 | if (!strcmp(direntp->d_name, "..")) | |
640 | continue; | |
641 | ||
642 | if (strncmp(direntp->d_name, "loop", 4)) | |
643 | continue; | |
644 | ||
9ba8130c SH |
645 | rc = snprintf(path, MAXPATHLEN, "/dev/%s", direntp->d_name); |
646 | if (rc < 0 || rc >= MAXPATHLEN) | |
647 | continue; | |
648 | ||
025ed0f3 | 649 | process_lock(); |
a6afdde9 | 650 | fd = open(path, O_RDWR); |
025ed0f3 | 651 | process_unlock(); |
a6afdde9 DL |
652 | if (fd < 0) |
653 | continue; | |
654 | ||
655 | if (ioctl(fd, LOOP_GET_STATUS64, &loinfo) == 0) { | |
025ed0f3 | 656 | process_lock(); |
a6afdde9 | 657 | close(fd); |
025ed0f3 | 658 | process_unlock(); |
a6afdde9 DL |
659 | continue; |
660 | } | |
661 | ||
662 | if (errno != ENXIO) { | |
663 | WARN("unexpected error for ioctl on '%s': %m", | |
664 | direntp->d_name); | |
025ed0f3 | 665 | process_lock(); |
00b6be44 | 666 | close(fd); |
025ed0f3 | 667 | process_unlock(); |
a6afdde9 DL |
668 | continue; |
669 | } | |
670 | ||
671 | DEBUG("found '%s' free lodev", path); | |
672 | ||
673 | ret = setup_lodev(rootfs, fd, &loinfo); | |
674 | if (!ret) | |
2656d231 | 675 | ret = mount_unknow_fs(path, target, 0); |
025ed0f3 | 676 | process_lock(); |
a6afdde9 | 677 | close(fd); |
025ed0f3 | 678 | process_unlock(); |
a6afdde9 DL |
679 | |
680 | break; | |
681 | } | |
682 | ||
025ed0f3 | 683 | process_lock(); |
a6afdde9 DL |
684 | if (closedir(dir)) |
685 | WARN("failed to close directory"); | |
025ed0f3 | 686 | process_unlock(); |
a6afdde9 DL |
687 | |
688 | return ret; | |
78ae2fcc | 689 | } |
690 | ||
2656d231 | 691 | static int mount_rootfs_block(const char *rootfs, const char *target) |
a6afdde9 | 692 | { |
2656d231 | 693 | return mount_unknow_fs(rootfs, target, 0); |
a6afdde9 DL |
694 | } |
695 | ||
0c547523 SH |
696 | /* |
697 | * pin_rootfs | |
b7ed4bf0 CS |
698 | * if rootfs is a directory, then open ${rootfs}/lxc.hold for writing for |
699 | * the duration of the container run, to prevent the container from marking | |
700 | * the underlying fs readonly on shutdown. unlink the file immediately so | |
701 | * no name pollution is happens | |
0c547523 SH |
702 | * return -1 on error. |
703 | * return -2 if nothing needed to be pinned. | |
704 | * return an open fd (>=0) if we pinned it. | |
705 | */ | |
706 | int pin_rootfs(const char *rootfs) | |
707 | { | |
708 | char absrootfs[MAXPATHLEN]; | |
709 | char absrootfspin[MAXPATHLEN]; | |
710 | struct stat s; | |
711 | int ret, fd; | |
712 | ||
e99ee0de | 713 | if (rootfs == NULL || strlen(rootfs) == 0) |
0d03360a | 714 | return -2; |
e99ee0de | 715 | |
00ec333b | 716 | if (!realpath(rootfs, absrootfs)) |
9be53773 | 717 | return -2; |
0c547523 | 718 | |
00ec333b | 719 | if (access(absrootfs, F_OK)) |
0c547523 | 720 | return -1; |
0c547523 | 721 | |
00ec333b | 722 | if (stat(absrootfs, &s)) |
0c547523 | 723 | return -1; |
0c547523 | 724 | |
72f919c4 | 725 | if (!S_ISDIR(s.st_mode)) |
0c547523 SH |
726 | return -2; |
727 | ||
b7ed4bf0 | 728 | ret = snprintf(absrootfspin, MAXPATHLEN, "%s/lxc.hold", absrootfs); |
00ec333b | 729 | if (ret >= MAXPATHLEN) |
0c547523 | 730 | return -1; |
0c547523 | 731 | |
025ed0f3 | 732 | process_lock(); |
0c547523 | 733 | fd = open(absrootfspin, O_CREAT | O_RDWR, S_IWUSR|S_IRUSR); |
025ed0f3 | 734 | process_unlock(); |
b7ed4bf0 CS |
735 | if (fd < 0) |
736 | return fd; | |
737 | (void)unlink(absrootfspin); | |
0c547523 SH |
738 | return fd; |
739 | } | |
740 | ||
368bbc02 CS |
741 | static int lxc_mount_auto_mounts(struct lxc_conf *conf, int flags, struct cgroup_process_info *cgroup_info) |
742 | { | |
743 | char *path = NULL; | |
744 | char *dev_null = NULL; | |
745 | int r; | |
746 | ||
747 | dev_null = lxc_append_paths(conf->rootfs.mount, "/dev/null"); | |
748 | if (!dev_null) { | |
749 | SYSERROR("memory allocation error"); | |
750 | goto cleanup; | |
751 | } | |
752 | ||
753 | if (flags & LXC_AUTO_PROC) { | |
754 | path = lxc_append_paths(conf->rootfs.mount, "/proc"); | |
755 | if (!path) { | |
756 | SYSERROR("memory allocation error trying to automatically mount /proc"); | |
757 | goto cleanup; | |
758 | } | |
759 | ||
760 | r = mount("proc", path, "proc", MS_NODEV|MS_NOEXEC|MS_NOSUID, NULL); | |
761 | if (r < 0) { | |
762 | SYSERROR("error mounting /proc"); | |
763 | goto cleanup; | |
764 | } | |
765 | ||
766 | free(path); | |
767 | path = NULL; | |
768 | } | |
769 | ||
770 | if (flags & LXC_AUTO_PROC_SYSRQ) { | |
771 | path = lxc_append_paths(conf->rootfs.mount, "/proc/sysrq-trigger"); | |
772 | if (!path) { | |
773 | SYSERROR("memory allocation error trying to automatically mount /proc"); | |
774 | goto cleanup; | |
775 | } | |
776 | ||
777 | /* safety measure, mount /dev/null over /proc/sysrq-trigger, | |
778 | * otherwise, a container may trigger a host reboot or such | |
779 | */ | |
780 | r = mount(dev_null, path, NULL, MS_BIND, NULL); | |
781 | if (r < 0) | |
782 | WARN("error mounting /dev/null over /proc/sysrq-trigger: %s", strerror(errno)); | |
783 | ||
784 | free(path); | |
785 | path = NULL; | |
786 | } | |
787 | ||
788 | if (flags & LXC_AUTO_SYS) { | |
789 | path = lxc_append_paths(conf->rootfs.mount, "/sys"); | |
790 | if (!path) { | |
791 | SYSERROR("memory allocation error trying to automatically mount /sys"); | |
792 | goto cleanup; | |
793 | } | |
794 | ||
795 | r = mount("sysfs", path, "sysfs", MS_RDONLY, NULL); | |
796 | if (r < 0) { | |
797 | SYSERROR("error mounting /sys"); | |
798 | goto cleanup; | |
799 | } | |
800 | ||
801 | free(path); | |
802 | path = NULL; | |
803 | } | |
804 | ||
805 | if (flags & LXC_AUTO_CGROUP) { | |
806 | r = lxc_setup_mount_cgroup(conf->rootfs.mount, cgroup_info); | |
807 | if (r < 0) { | |
808 | SYSERROR("error mounting /sys/fs/cgroup"); | |
809 | goto cleanup; | |
810 | } | |
811 | } | |
812 | ||
813 | free(dev_null); | |
814 | free(path); | |
815 | ||
816 | return 0; | |
817 | ||
818 | cleanup: | |
819 | free(dev_null); | |
820 | free(path); | |
821 | return -1; | |
822 | } | |
823 | ||
2656d231 | 824 | static int mount_rootfs(const char *rootfs, const char *target) |
0ad19a3f | 825 | { |
b09ef133 | 826 | char absrootfs[MAXPATHLEN]; |
78ae2fcc | 827 | struct stat s; |
a6afdde9 | 828 | int i; |
78ae2fcc | 829 | |
a6afdde9 | 830 | typedef int (*rootfs_cb)(const char *, const char *); |
78ae2fcc | 831 | |
832 | struct rootfs_type { | |
833 | int type; | |
834 | rootfs_cb cb; | |
835 | } rtfs_type[] = { | |
2656d231 DL |
836 | { S_IFDIR, mount_rootfs_dir }, |
837 | { S_IFBLK, mount_rootfs_block }, | |
838 | { S_IFREG, mount_rootfs_file }, | |
78ae2fcc | 839 | }; |
0ad19a3f | 840 | |
4c8ab83b | 841 | if (!realpath(rootfs, absrootfs)) { |
36eb9bde | 842 | SYSERROR("failed to get real path for '%s'", rootfs); |
4c8ab83b | 843 | return -1; |
844 | } | |
b09ef133 | 845 | |
b09ef133 | 846 | if (access(absrootfs, F_OK)) { |
36eb9bde | 847 | SYSERROR("'%s' is not accessible", absrootfs); |
b09ef133 | 848 | return -1; |
849 | } | |
850 | ||
78ae2fcc | 851 | if (stat(absrootfs, &s)) { |
36eb9bde | 852 | SYSERROR("failed to stat '%s'", absrootfs); |
9b0f0477 | 853 | return -1; |
854 | } | |
855 | ||
78ae2fcc | 856 | for (i = 0; i < sizeof(rtfs_type)/sizeof(rtfs_type[0]); i++) { |
9b0f0477 | 857 | |
78ae2fcc | 858 | if (!__S_ISTYPE(s.st_mode, rtfs_type[i].type)) |
859 | continue; | |
9b0f0477 | 860 | |
a6afdde9 | 861 | return rtfs_type[i].cb(absrootfs, target); |
78ae2fcc | 862 | } |
9b0f0477 | 863 | |
36eb9bde | 864 | ERROR("unsupported rootfs type for '%s'", absrootfs); |
78ae2fcc | 865 | return -1; |
0ad19a3f | 866 | } |
867 | ||
4e5440c6 | 868 | static int setup_utsname(struct utsname *utsname) |
0ad19a3f | 869 | { |
4e5440c6 DL |
870 | if (!utsname) |
871 | return 0; | |
0ad19a3f | 872 | |
4e5440c6 DL |
873 | if (sethostname(utsname->nodename, strlen(utsname->nodename))) { |
874 | SYSERROR("failed to set the hostname to '%s'", utsname->nodename); | |
0ad19a3f | 875 | return -1; |
876 | } | |
877 | ||
4e5440c6 | 878 | INFO("'%s' hostname has been setup", utsname->nodename); |
cd54d859 | 879 | |
0ad19a3f | 880 | return 0; |
881 | } | |
882 | ||
33fcb7a0 | 883 | static int setup_tty(const struct lxc_rootfs *rootfs, |
7c6ef2a2 | 884 | const struct lxc_tty_info *tty_info, char *ttydir) |
b0a33c1e | 885 | { |
7c6ef2a2 SH |
886 | char path[MAXPATHLEN], lxcpath[MAXPATHLEN]; |
887 | int i, ret; | |
b0a33c1e | 888 | |
bc9bd0e3 DL |
889 | if (!rootfs->path) |
890 | return 0; | |
891 | ||
b0a33c1e | 892 | for (i = 0; i < tty_info->nbtty; i++) { |
893 | ||
894 | struct lxc_pty_info *pty_info = &tty_info->pty_info[i]; | |
895 | ||
7c6ef2a2 | 896 | ret = snprintf(path, sizeof(path), "%s/dev/tty%d", |
12297168 | 897 | rootfs->mount, i + 1); |
7c6ef2a2 SH |
898 | if (ret >= sizeof(path)) { |
899 | ERROR("pathname too long for ttys"); | |
900 | return -1; | |
901 | } | |
902 | if (ttydir) { | |
903 | /* create dev/lxc/tty%d" */ | |
9ba8130c | 904 | ret = snprintf(lxcpath, sizeof(lxcpath), "%s/dev/%s/tty%d", |
7c6ef2a2 SH |
905 | rootfs->mount, ttydir, i + 1); |
906 | if (ret >= sizeof(lxcpath)) { | |
907 | ERROR("pathname too long for ttys"); | |
908 | return -1; | |
909 | } | |
025ed0f3 | 910 | process_lock(); |
7c6ef2a2 | 911 | ret = creat(lxcpath, 0660); |
025ed0f3 | 912 | process_unlock(); |
7c6ef2a2 SH |
913 | if (ret==-1 && errno != EEXIST) { |
914 | SYSERROR("error creating %s\n", lxcpath); | |
915 | return -1; | |
916 | } | |
025ed0f3 | 917 | process_lock(); |
4d44e274 SH |
918 | if (ret >= 0) |
919 | close(ret); | |
025ed0f3 | 920 | process_unlock(); |
7c6ef2a2 SH |
921 | ret = unlink(path); |
922 | if (ret && errno != ENOENT) { | |
923 | SYSERROR("error unlinking %s\n", path); | |
924 | return -1; | |
925 | } | |
b0a33c1e | 926 | |
7c6ef2a2 SH |
927 | if (mount(pty_info->name, lxcpath, "none", MS_BIND, 0)) { |
928 | WARN("failed to mount '%s'->'%s'", | |
929 | pty_info->name, path); | |
930 | continue; | |
931 | } | |
13954cce | 932 | |
9ba8130c SH |
933 | ret = snprintf(lxcpath, sizeof(lxcpath), "%s/tty%d", ttydir, i+1); |
934 | if (ret >= sizeof(lxcpath)) { | |
935 | ERROR("tty pathname too long"); | |
936 | return -1; | |
937 | } | |
7c6ef2a2 SH |
938 | ret = symlink(lxcpath, path); |
939 | if (ret) { | |
940 | SYSERROR("failed to create symlink for tty %d\n", i+1); | |
941 | return -1; | |
942 | } | |
943 | } else { | |
c6883f38 SH |
944 | /* If we populated /dev, then we need to create /dev/ttyN */ |
945 | if (access(path, F_OK)) { | |
025ed0f3 | 946 | process_lock(); |
c6883f38 | 947 | ret = creat(path, 0660); |
025ed0f3 | 948 | process_unlock(); |
c6883f38 SH |
949 | if (ret==-1) { |
950 | SYSERROR("error creating %s\n", path); | |
951 | /* this isn't fatal, continue */ | |
025ed0f3 SH |
952 | } else { |
953 | process_lock(); | |
c6883f38 | 954 | close(ret); |
025ed0f3 SH |
955 | process_unlock(); |
956 | } | |
c6883f38 | 957 | } |
7c6ef2a2 SH |
958 | if (mount(pty_info->name, path, "none", MS_BIND, 0)) { |
959 | WARN("failed to mount '%s'->'%s'", | |
960 | pty_info->name, path); | |
961 | continue; | |
962 | } | |
b0a33c1e | 963 | } |
964 | } | |
965 | ||
cd54d859 DL |
966 | INFO("%d tty(s) has been setup", tty_info->nbtty); |
967 | ||
b0a33c1e | 968 | return 0; |
969 | } | |
970 | ||
7a7ff0c6 | 971 | static int setup_rootfs_pivot_root_cb(char *buffer, void *data) |
bf601689 MH |
972 | { |
973 | struct lxc_list *mountlist, *listentry, *iterator; | |
2c7d90ac | 974 | char *pivotdir, *mountpoint, *mountentry, *saveptr = NULL; |
bf601689 MH |
975 | int found; |
976 | void **cbparm; | |
977 | ||
978 | mountentry = buffer; | |
979 | cbparm = (void **)data; | |
980 | ||
981 | mountlist = cbparm[0]; | |
982 | pivotdir = cbparm[1]; | |
983 | ||
984 | /* parse entry, first field is mountname, ignore */ | |
2796cf79 | 985 | mountpoint = strtok_r(mountentry, " ", &saveptr); |
bf601689 MH |
986 | if (!mountpoint) |
987 | return -1; | |
988 | ||
989 | /* second field is mountpoint */ | |
2796cf79 | 990 | mountpoint = strtok_r(NULL, " ", &saveptr); |
bf601689 MH |
991 | if (!mountpoint) |
992 | return -1; | |
993 | ||
994 | /* only consider mountpoints below old root fs */ | |
995 | if (strncmp(mountpoint, pivotdir, strlen(pivotdir))) | |
996 | return 0; | |
997 | ||
998 | /* filter duplicate mountpoints */ | |
999 | found = 0; | |
1000 | lxc_list_for_each(iterator, mountlist) { | |
1001 | if (!strcmp(iterator->elem, mountpoint)) { | |
1002 | found = 1; | |
1003 | break; | |
1004 | } | |
1005 | } | |
1006 | if (found) | |
1007 | return 0; | |
1008 | ||
1009 | /* add entry to list */ | |
1010 | listentry = malloc(sizeof(*listentry)); | |
1011 | if (!listentry) { | |
1012 | SYSERROR("malloc for mountpoint listentry failed"); | |
1013 | return -1; | |
1014 | } | |
1015 | ||
1016 | listentry->elem = strdup(mountpoint); | |
1017 | if (!listentry->elem) { | |
1018 | SYSERROR("strdup failed"); | |
00b6be44 | 1019 | free(listentry); |
bf601689 MH |
1020 | return -1; |
1021 | } | |
1022 | lxc_list_add_tail(mountlist, listentry); | |
1023 | ||
1024 | return 0; | |
1025 | } | |
1026 | ||
cc6f6dd7 | 1027 | static int umount_oldrootfs(const char *oldrootfs) |
bf601689 | 1028 | { |
2382ecff | 1029 | char path[MAXPATHLEN]; |
bf601689 | 1030 | void *cbparm[2]; |
9ebb03ad | 1031 | struct lxc_list mountlist, *iterator, *next; |
bf601689 | 1032 | int ok, still_mounted, last_still_mounted; |
9ba8130c | 1033 | int rc; |
bf601689 MH |
1034 | |
1035 | /* read and parse /proc/mounts in old root fs */ | |
1036 | lxc_list_init(&mountlist); | |
1037 | ||
cc6f6dd7 | 1038 | /* oldrootfs is on the top tree directory now */ |
9ba8130c SH |
1039 | rc = snprintf(path, sizeof(path), "/%s", oldrootfs); |
1040 | if (rc >= sizeof(path)) { | |
1041 | ERROR("rootfs name too long"); | |
1042 | return -1; | |
1043 | } | |
bf601689 | 1044 | cbparm[0] = &mountlist; |
bf601689 | 1045 | |
cc6f6dd7 | 1046 | cbparm[1] = strdup(path); |
bf601689 MH |
1047 | if (!cbparm[1]) { |
1048 | SYSERROR("strdup failed"); | |
1049 | return -1; | |
1050 | } | |
1051 | ||
9ba8130c SH |
1052 | rc = snprintf(path, sizeof(path), "%s/proc/mounts", oldrootfs); |
1053 | if (rc >= sizeof(path)) { | |
1054 | ERROR("container proc/mounts name too long"); | |
1055 | return -1; | |
1056 | } | |
cc6f6dd7 DL |
1057 | |
1058 | ok = lxc_file_for_each_line(path, | |
1059 | setup_rootfs_pivot_root_cb, &cbparm); | |
bf601689 MH |
1060 | if (ok < 0) { |
1061 | SYSERROR("failed to read or parse mount list '%s'", path); | |
1062 | return -1; | |
1063 | } | |
1064 | ||
1065 | /* umount filesystems until none left or list no longer shrinks */ | |
1066 | still_mounted = 0; | |
1067 | do { | |
1068 | last_still_mounted = still_mounted; | |
1069 | still_mounted = 0; | |
1070 | ||
9ebb03ad | 1071 | lxc_list_for_each_safe(iterator, &mountlist, next) { |
bf601689 | 1072 | |
c08556c6 | 1073 | /* umount normally */ |
bf601689 MH |
1074 | if (!umount(iterator->elem)) { |
1075 | DEBUG("umounted '%s'", (char *)iterator->elem); | |
1076 | lxc_list_del(iterator); | |
1077 | continue; | |
1078 | } | |
1079 | ||
bf601689 MH |
1080 | still_mounted++; |
1081 | } | |
7df119ee | 1082 | |
bf601689 MH |
1083 | } while (still_mounted > 0 && still_mounted != last_still_mounted); |
1084 | ||
7df119ee | 1085 | |
c08556c6 DL |
1086 | lxc_list_for_each(iterator, &mountlist) { |
1087 | ||
1088 | /* let's try a lazy umount */ | |
1089 | if (!umount2(iterator->elem, MNT_DETACH)) { | |
1090 | INFO("lazy unmount of '%s'", (char *)iterator->elem); | |
1091 | continue; | |
1092 | } | |
1093 | ||
1094 | /* be more brutal (nfs) */ | |
1095 | if (!umount2(iterator->elem, MNT_FORCE)) { | |
1096 | INFO("forced unmount of '%s'", (char *)iterator->elem); | |
1097 | continue; | |
1098 | } | |
1099 | ||
7df119ee | 1100 | WARN("failed to unmount '%s'", (char *)iterator->elem); |
c08556c6 | 1101 | } |
bf601689 | 1102 | |
cc6f6dd7 DL |
1103 | return 0; |
1104 | } | |
1105 | ||
1106 | static int setup_rootfs_pivot_root(const char *rootfs, const char *pivotdir) | |
1107 | { | |
1108 | char path[MAXPATHLEN]; | |
1109 | int remove_pivotdir = 0; | |
9ba8130c | 1110 | int rc; |
cc6f6dd7 DL |
1111 | |
1112 | /* change into new root fs */ | |
1113 | if (chdir(rootfs)) { | |
1114 | SYSERROR("can't chdir to new rootfs '%s'", rootfs); | |
1115 | return -1; | |
1116 | } | |
1117 | ||
1118 | if (!pivotdir) | |
30c5d292 | 1119 | pivotdir = "lxc_putold"; |
cc6f6dd7 | 1120 | |
4f9293b1 | 1121 | /* compute the full path to pivotdir under rootfs */ |
9ba8130c SH |
1122 | rc = snprintf(path, sizeof(path), "%s/%s", rootfs, pivotdir); |
1123 | if (rc >= sizeof(path)) { | |
1124 | ERROR("pivot dir name too long"); | |
1125 | return -1; | |
1126 | } | |
cc6f6dd7 DL |
1127 | |
1128 | if (access(path, F_OK)) { | |
1129 | ||
1130 | if (mkdir_p(path, 0755)) { | |
1131 | SYSERROR("failed to create pivotdir '%s'", path); | |
1132 | return -1; | |
1133 | } | |
1134 | ||
1135 | remove_pivotdir = 1; | |
1136 | DEBUG("created '%s' directory", path); | |
1137 | } | |
1138 | ||
1139 | DEBUG("mountpoint for old rootfs is '%s'", path); | |
1140 | ||
1141 | /* pivot_root into our new root fs */ | |
1142 | if (pivot_root(".", path)) { | |
1143 | SYSERROR("pivot_root syscall failed"); | |
bf601689 MH |
1144 | return -1; |
1145 | } | |
cc6f6dd7 DL |
1146 | |
1147 | if (chdir("/")) { | |
1148 | SYSERROR("can't chdir to / after pivot_root"); | |
1149 | return -1; | |
1150 | } | |
1151 | ||
1152 | DEBUG("pivot_root syscall to '%s' successful", rootfs); | |
1153 | ||
1154 | /* we switch from absolute path to relative path */ | |
1155 | if (umount_oldrootfs(pivotdir)) | |
1156 | return -1; | |
bf601689 | 1157 | |
c08556c6 DL |
1158 | /* remove temporary mount point, we don't consider the removing |
1159 | * as fatal */ | |
a91d897a FW |
1160 | if (remove_pivotdir && rmdir(pivotdir)) |
1161 | WARN("can't remove mountpoint '%s': %m", pivotdir); | |
bf601689 | 1162 | |
bf601689 MH |
1163 | return 0; |
1164 | } | |
1165 | ||
91c3830e SH |
1166 | /* |
1167 | * Do we want to add options for max size of /dev and a file to | |
1168 | * specify which devices to create? | |
1169 | */ | |
1170 | static int mount_autodev(char *root) | |
1171 | { | |
1172 | int ret; | |
1173 | char path[MAXPATHLEN]; | |
1174 | ||
1175 | INFO("Mounting /dev under %s\n", root); | |
1176 | ret = snprintf(path, MAXPATHLEN, "%s/dev", root); | |
1177 | if (ret < 0 || ret > MAXPATHLEN) | |
1178 | return -1; | |
1179 | ret = mount("none", path, "tmpfs", 0, "size=100000"); | |
1180 | if (ret) { | |
1181 | SYSERROR("Failed to mount /dev at %s\n", root); | |
1182 | return -1; | |
1183 | } | |
1184 | ret = snprintf(path, MAXPATHLEN, "%s/dev/pts", root); | |
1185 | if (ret < 0 || ret >= MAXPATHLEN) | |
1186 | return -1; | |
1187 | ret = mkdir(path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH); | |
1188 | if (ret) { | |
1189 | SYSERROR("Failed to create /dev/pts in container"); | |
1190 | return -1; | |
1191 | } | |
1192 | ||
1193 | INFO("Mounted /dev under %s\n", root); | |
1194 | return 0; | |
1195 | } | |
1196 | ||
c6883f38 SH |
1197 | struct lxc_devs { |
1198 | char *name; | |
1199 | mode_t mode; | |
1200 | int maj; | |
1201 | int min; | |
1202 | }; | |
1203 | ||
1204 | struct lxc_devs lxc_devs[] = { | |
1205 | { "null", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 1, 3 }, | |
1206 | { "zero", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 1, 5 }, | |
1207 | { "full", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 1, 7 }, | |
1208 | { "urandom", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 1, 9 }, | |
1209 | { "random", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 1, 8 }, | |
1210 | { "tty", S_IFCHR | S_IRWXU | S_IRWXG | S_IRWXO, 5, 0 }, | |
1211 | { "console", S_IFCHR | S_IRUSR | S_IWUSR, 5, 1 }, | |
1212 | }; | |
1213 | ||
c6883f38 SH |
1214 | static int setup_autodev(char *root) |
1215 | { | |
1216 | int ret; | |
1217 | struct lxc_devs *d; | |
1218 | char path[MAXPATHLEN]; | |
1219 | int i; | |
3a32201c | 1220 | mode_t cmask; |
c6883f38 | 1221 | |
91c3830e SH |
1222 | INFO("Creating initial consoles under %s/dev\n", root); |
1223 | ||
c6883f38 | 1224 | ret = snprintf(path, MAXPATHLEN, "%s/dev", root); |
91c3830e SH |
1225 | if (ret < 0 || ret >= MAXPATHLEN) { |
1226 | ERROR("Error calculating container /dev location"); | |
c6883f38 | 1227 | return -1; |
f7bee6c6 | 1228 | } |
91c3830e SH |
1229 | |
1230 | INFO("Populating /dev under %s\n", root); | |
3a32201c | 1231 | cmask = umask(S_IXUSR | S_IXGRP | S_IXOTH); |
c6883f38 SH |
1232 | for (i = 0; i < sizeof(lxc_devs) / sizeof(lxc_devs[0]); i++) { |
1233 | d = &lxc_devs[i]; | |
1234 | ret = snprintf(path, MAXPATHLEN, "%s/dev/%s", root, d->name); | |
1235 | if (ret < 0 || ret >= MAXPATHLEN) | |
1236 | return -1; | |
1237 | ret = mknod(path, d->mode, makedev(d->maj, d->min)); | |
91c3830e | 1238 | if (ret && errno != EEXIST) { |
c6883f38 SH |
1239 | SYSERROR("Error creating %s\n", d->name); |
1240 | return -1; | |
1241 | } | |
1242 | } | |
3a32201c | 1243 | umask(cmask); |
c6883f38 SH |
1244 | |
1245 | INFO("Populated /dev under %s\n", root); | |
1246 | return 0; | |
1247 | } | |
1248 | ||
cc28d0b0 SH |
1249 | /* |
1250 | * Detect whether / is mounted MS_SHARED. The only way I know of to | |
1251 | * check that is through /proc/self/mountinfo. | |
1252 | * I'm only checking for /. If the container rootfs or mount location | |
1253 | * is MS_SHARED, but not '/', then you're out of luck - figuring that | |
1254 | * out would be too much work to be worth it. | |
1255 | */ | |
1256 | #define LINELEN 4096 | |
1257 | int detect_shared_rootfs(void) | |
1258 | { | |
1259 | char buf[LINELEN], *p; | |
1260 | FILE *f; | |
1261 | int i; | |
1262 | char *p2; | |
1263 | ||
025ed0f3 | 1264 | process_lock(); |
cc28d0b0 | 1265 | f = fopen("/proc/self/mountinfo", "r"); |
025ed0f3 | 1266 | process_unlock(); |
cc28d0b0 SH |
1267 | if (!f) |
1268 | return 0; | |
1269 | while ((p = fgets(buf, LINELEN, f))) { | |
1270 | INFO("looking at .%s.", p); | |
1271 | for (p = buf, i=0; p && i < 4; i++) | |
1272 | p = index(p+1, ' '); | |
1273 | if (!p) | |
1274 | continue; | |
1275 | p2 = index(p+1, ' '); | |
1276 | if (!p2) | |
1277 | continue; | |
1278 | *p2 = '\0'; | |
1279 | INFO("now p is .%s.", p); | |
1280 | if (strcmp(p+1, "/") == 0) { | |
1281 | // this is '/'. is it shared? | |
1282 | p = index(p2+1, ' '); | |
ab81cef0 | 1283 | if (p && strstr(p, "shared:")) { |
025ed0f3 | 1284 | process_lock(); |
00b6be44 | 1285 | fclose(f); |
025ed0f3 | 1286 | process_unlock(); |
cc28d0b0 | 1287 | return 1; |
00b6be44 | 1288 | } |
cc28d0b0 SH |
1289 | } |
1290 | } | |
025ed0f3 | 1291 | process_lock(); |
cc28d0b0 | 1292 | fclose(f); |
025ed0f3 | 1293 | process_unlock(); |
cc28d0b0 SH |
1294 | return 0; |
1295 | } | |
1296 | ||
1297 | /* | |
1298 | * I'll forgive you for asking whether all of this is needed :) The | |
1299 | * answer is yes. | |
1300 | * pivot_root will fail if the new root, the put_old dir, or the parent | |
1301 | * of current->fs->root are MS_SHARED. (parent of current->fs_root may | |
1302 | * or may not be current->fs_root - if we assumed it always was, we could | |
1303 | * just mount --make-rslave /). So, | |
1304 | * 1. mount a tiny tmpfs to be parent of current->fs->root. | |
1305 | * 2. make that MS_SLAVE | |
1306 | * 3. make a 'root' directory under that | |
1307 | * 4. mount --rbind / under the $tinyroot/root. | |
1308 | * 5. make that rslave | |
1309 | * 6. chdir and chroot into $tinyroot/root | |
1310 | * 7. $tinyroot will be unmounted by our parent in start.c | |
1311 | */ | |
1312 | static int chroot_into_slave(struct lxc_conf *conf) | |
1313 | { | |
1314 | char path[MAXPATHLEN]; | |
1315 | const char *destpath = conf->rootfs.mount; | |
1316 | int ret; | |
1317 | ||
1318 | if (mount(destpath, destpath, NULL, MS_BIND, 0)) { | |
1319 | SYSERROR("failed to mount %s bind", destpath); | |
1320 | return -1; | |
1321 | } | |
1322 | if (mount("", destpath, NULL, MS_SLAVE, 0)) { | |
1323 | SYSERROR("failed to make %s slave", destpath); | |
1324 | return -1; | |
1325 | } | |
1326 | if (mount("none", destpath, "tmpfs", 0, "size=10000")) { | |
1327 | SYSERROR("Failed to mount tmpfs / at %s", destpath); | |
1328 | return -1; | |
1329 | } | |
1330 | ret = snprintf(path, MAXPATHLEN, "%s/root", destpath); | |
1331 | if (ret < 0 || ret >= MAXPATHLEN) { | |
1332 | ERROR("out of memory making root path"); | |
1333 | return -1; | |
1334 | } | |
1335 | if (mkdir(path, S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH)) { | |
1336 | SYSERROR("Failed to create /dev/pts in container"); | |
1337 | return -1; | |
1338 | } | |
1339 | if (mount("/", path, NULL, MS_BIND|MS_REC, 0)) { | |
1340 | SYSERROR("Failed to rbind mount / to %s", path); | |
1341 | return -1; | |
1342 | } | |
1343 | if (mount("", destpath, NULL, MS_SLAVE|MS_REC, 0)) { | |
1344 | SYSERROR("Failed to make tmp-/ at %s rslave", path); | |
1345 | return -1; | |
1346 | } | |
1347 | if (chdir(path)) { | |
1348 | SYSERROR("Failed to chdir into tmp-/"); | |
1349 | return -1; | |
1350 | } | |
1351 | if (chroot(path)) { | |
1352 | SYSERROR("Failed to chroot into tmp-/"); | |
1353 | return -1; | |
1354 | } | |
1355 | INFO("Chrooted into tmp-/ at %s\n", path); | |
1356 | return 0; | |
1357 | } | |
1358 | ||
1359 | static int setup_rootfs(struct lxc_conf *conf) | |
0ad19a3f | 1360 | { |
cc28d0b0 SH |
1361 | const struct lxc_rootfs *rootfs = &conf->rootfs; |
1362 | ||
a0f379bf DW |
1363 | if (!rootfs->path) { |
1364 | if (mount("", "/", NULL, MS_SLAVE|MS_REC, 0)) { | |
1365 | SYSERROR("Failed to make / rslave"); | |
1366 | return -1; | |
1367 | } | |
c69bd12f | 1368 | return 0; |
a0f379bf | 1369 | } |
0ad19a3f | 1370 | |
12297168 | 1371 | if (access(rootfs->mount, F_OK)) { |
b1789442 | 1372 | SYSERROR("failed to access to '%s', check it is present", |
12297168 | 1373 | rootfs->mount); |
b1789442 DL |
1374 | return -1; |
1375 | } | |
1376 | ||
cc28d0b0 SH |
1377 | if (detect_shared_rootfs()) { |
1378 | if (chroot_into_slave(conf)) { | |
1379 | ERROR("Failed to chroot into slave /"); | |
1380 | return -1; | |
1381 | } | |
1382 | } | |
1383 | ||
9be53773 SH |
1384 | // First try mounting rootfs using a bdev |
1385 | struct bdev *bdev = bdev_init(rootfs->path, rootfs->mount, NULL); | |
1386 | if (bdev && bdev->ops->mount(bdev) == 0) { | |
59d66af2 | 1387 | bdev_put(bdev); |
9be53773 SH |
1388 | DEBUG("mounted '%s' on '%s'", rootfs->path, rootfs->mount); |
1389 | return 0; | |
1390 | } | |
59d66af2 SH |
1391 | if (bdev) |
1392 | bdev_put(bdev); | |
2656d231 | 1393 | if (mount_rootfs(rootfs->path, rootfs->mount)) { |
a6afdde9 | 1394 | ERROR("failed to mount rootfs"); |
c3f0a28c | 1395 | return -1; |
1396 | } | |
0ad19a3f | 1397 | |
12297168 | 1398 | DEBUG("mounted '%s' on '%s'", rootfs->path, rootfs->mount); |
c69bd12f | 1399 | |
ac778708 DL |
1400 | return 0; |
1401 | } | |
1402 | ||
1403 | int setup_pivot_root(const struct lxc_rootfs *rootfs) | |
1404 | { | |
ac778708 DL |
1405 | if (!rootfs->path) |
1406 | return 0; | |
1407 | ||
12297168 | 1408 | if (setup_rootfs_pivot_root(rootfs->mount, rootfs->pivot)) { |
cc6f6dd7 | 1409 | ERROR("failed to setup pivot root"); |
25368b52 | 1410 | return -1; |
c69bd12f DL |
1411 | } |
1412 | ||
25368b52 | 1413 | return 0; |
0ad19a3f | 1414 | } |
1415 | ||
d852c78c | 1416 | static int setup_pts(int pts) |
3c26f34e | 1417 | { |
77890c6d SW |
1418 | char target[PATH_MAX]; |
1419 | ||
d852c78c DL |
1420 | if (!pts) |
1421 | return 0; | |
3c26f34e | 1422 | |
1423 | if (!access("/dev/pts/ptmx", F_OK) && umount("/dev/pts")) { | |
36eb9bde | 1424 | SYSERROR("failed to umount 'dev/pts'"); |
3c26f34e | 1425 | return -1; |
1426 | } | |
1427 | ||
a6afdde9 DL |
1428 | if (mount("devpts", "/dev/pts", "devpts", MS_MGC_VAL, |
1429 | "newinstance,ptmxmode=0666")) { | |
36eb9bde | 1430 | SYSERROR("failed to mount a new instance of '/dev/pts'"); |
3c26f34e | 1431 | return -1; |
1432 | } | |
1433 | ||
3c26f34e | 1434 | if (access("/dev/ptmx", F_OK)) { |
1435 | if (!symlink("/dev/pts/ptmx", "/dev/ptmx")) | |
1436 | goto out; | |
36eb9bde | 1437 | SYSERROR("failed to symlink '/dev/pts/ptmx'->'/dev/ptmx'"); |
3c26f34e | 1438 | return -1; |
1439 | } | |
1440 | ||
77890c6d SW |
1441 | if (realpath("/dev/ptmx", target) && !strcmp(target, "/dev/pts/ptmx")) |
1442 | goto out; | |
1443 | ||
3c26f34e | 1444 | /* fallback here, /dev/pts/ptmx exists just mount bind */ |
1445 | if (mount("/dev/pts/ptmx", "/dev/ptmx", "none", MS_BIND, 0)) { | |
36eb9bde | 1446 | SYSERROR("mount failed '/dev/pts/ptmx'->'/dev/ptmx'"); |
3c26f34e | 1447 | return -1; |
1448 | } | |
cd54d859 DL |
1449 | |
1450 | INFO("created new pts instance"); | |
d852c78c | 1451 | |
3c26f34e | 1452 | out: |
1453 | return 0; | |
1454 | } | |
1455 | ||
cccc74b5 DL |
1456 | static int setup_personality(int persona) |
1457 | { | |
6ff05e18 | 1458 | #if HAVE_SYS_PERSONALITY_H |
cccc74b5 DL |
1459 | if (persona == -1) |
1460 | return 0; | |
1461 | ||
1462 | if (personality(persona) < 0) { | |
1463 | SYSERROR("failed to set personality to '0x%x'", persona); | |
1464 | return -1; | |
1465 | } | |
1466 | ||
1467 | INFO("set personality to '0x%x'", persona); | |
6ff05e18 | 1468 | #endif |
cccc74b5 DL |
1469 | |
1470 | return 0; | |
1471 | } | |
1472 | ||
7c6ef2a2 | 1473 | static int setup_dev_console(const struct lxc_rootfs *rootfs, |
33fcb7a0 | 1474 | const struct lxc_console *console) |
6e590161 | 1475 | { |
63376d7d DL |
1476 | char path[MAXPATHLEN]; |
1477 | struct stat s; | |
7c6ef2a2 | 1478 | int ret; |
52e35957 | 1479 | |
7c6ef2a2 SH |
1480 | ret = snprintf(path, sizeof(path), "%s/dev/console", rootfs->mount); |
1481 | if (ret >= sizeof(path)) { | |
1482 | ERROR("console path too long\n"); | |
1483 | return -1; | |
1484 | } | |
52e35957 | 1485 | |
63376d7d | 1486 | if (access(path, F_OK)) { |
466978b0 | 1487 | WARN("rootfs specified but no console found at '%s'", path); |
63376d7d | 1488 | return 0; |
52e35957 DL |
1489 | } |
1490 | ||
b5159817 DE |
1491 | if (console->master < 0) { |
1492 | INFO("no console"); | |
f78a1f32 DL |
1493 | return 0; |
1494 | } | |
ed502555 | 1495 | |
63376d7d DL |
1496 | if (stat(path, &s)) { |
1497 | SYSERROR("failed to stat '%s'", path); | |
1498 | return -1; | |
1499 | } | |
1500 | ||
1501 | if (chmod(console->name, s.st_mode)) { | |
1502 | SYSERROR("failed to set mode '0%o' to '%s'", | |
1503 | s.st_mode, console->name); | |
1504 | return -1; | |
1505 | } | |
13954cce | 1506 | |
63376d7d DL |
1507 | if (mount(console->name, path, "none", MS_BIND, 0)) { |
1508 | ERROR("failed to mount '%s' on '%s'", console->name, path); | |
6e590161 | 1509 | return -1; |
1510 | } | |
1511 | ||
63376d7d | 1512 | INFO("console has been setup"); |
7c6ef2a2 SH |
1513 | return 0; |
1514 | } | |
1515 | ||
1516 | static int setup_ttydir_console(const struct lxc_rootfs *rootfs, | |
1517 | const struct lxc_console *console, | |
1518 | char *ttydir) | |
1519 | { | |
1520 | char path[MAXPATHLEN], lxcpath[MAXPATHLEN]; | |
1521 | int ret; | |
1522 | ||
1523 | /* create rootfs/dev/<ttydir> directory */ | |
1524 | ret = snprintf(path, sizeof(path), "%s/dev/%s", rootfs->mount, | |
1525 | ttydir); | |
1526 | if (ret >= sizeof(path)) | |
1527 | return -1; | |
1528 | ret = mkdir(path, 0755); | |
1529 | if (ret && errno != EEXIST) { | |
1530 | SYSERROR("failed with errno %d to create %s\n", errno, path); | |
1531 | return -1; | |
1532 | } | |
1533 | INFO("created %s\n", path); | |
1534 | ||
1535 | ret = snprintf(lxcpath, sizeof(lxcpath), "%s/dev/%s/console", | |
1536 | rootfs->mount, ttydir); | |
1537 | if (ret >= sizeof(lxcpath)) { | |
1538 | ERROR("console path too long\n"); | |
1539 | return -1; | |
1540 | } | |
1541 | ||
1542 | snprintf(path, sizeof(path), "%s/dev/console", rootfs->mount); | |
1543 | ret = unlink(path); | |
1544 | if (ret && errno != ENOENT) { | |
1545 | SYSERROR("error unlinking %s\n", path); | |
1546 | return -1; | |
1547 | } | |
1548 | ||
025ed0f3 | 1549 | process_lock(); |
7c6ef2a2 | 1550 | ret = creat(lxcpath, 0660); |
025ed0f3 | 1551 | process_unlock(); |
7c6ef2a2 SH |
1552 | if (ret==-1 && errno != EEXIST) { |
1553 | SYSERROR("error %d creating %s\n", errno, lxcpath); | |
1554 | return -1; | |
1555 | } | |
025ed0f3 | 1556 | process_lock(); |
4d44e274 SH |
1557 | if (ret >= 0) |
1558 | close(ret); | |
025ed0f3 | 1559 | process_unlock(); |
7c6ef2a2 | 1560 | |
b5159817 DE |
1561 | if (console->master < 0) { |
1562 | INFO("no console"); | |
7c6ef2a2 SH |
1563 | return 0; |
1564 | } | |
1565 | ||
1566 | if (mount(console->name, lxcpath, "none", MS_BIND, 0)) { | |
1567 | ERROR("failed to mount '%s' on '%s'", console->name, lxcpath); | |
1568 | return -1; | |
1569 | } | |
1570 | ||
1571 | /* create symlink from rootfs/dev/console to 'lxc/console' */ | |
9ba8130c SH |
1572 | ret = snprintf(lxcpath, sizeof(lxcpath), "%s/console", ttydir); |
1573 | if (ret >= sizeof(lxcpath)) { | |
1574 | ERROR("lxc/console path too long"); | |
1575 | return -1; | |
1576 | } | |
7c6ef2a2 SH |
1577 | ret = symlink(lxcpath, path); |
1578 | if (ret) { | |
1579 | SYSERROR("failed to create symlink for console"); | |
1580 | return -1; | |
1581 | } | |
1582 | ||
1583 | INFO("console has been setup on %s", lxcpath); | |
cd54d859 | 1584 | |
6e590161 | 1585 | return 0; |
1586 | } | |
1587 | ||
7c6ef2a2 SH |
1588 | static int setup_console(const struct lxc_rootfs *rootfs, |
1589 | const struct lxc_console *console, | |
1590 | char *ttydir) | |
1591 | { | |
1592 | /* We don't have a rootfs, /dev/console will be shared */ | |
1593 | if (!rootfs->path) | |
1594 | return 0; | |
1595 | if (!ttydir) | |
1596 | return setup_dev_console(rootfs, console); | |
1597 | ||
1598 | return setup_ttydir_console(rootfs, console, ttydir); | |
1599 | } | |
1600 | ||
1bd051a6 SH |
1601 | static int setup_kmsg(const struct lxc_rootfs *rootfs, |
1602 | const struct lxc_console *console) | |
1603 | { | |
1604 | char kpath[MAXPATHLEN]; | |
1605 | int ret; | |
1606 | ||
222fea5a DE |
1607 | if (!rootfs->path) |
1608 | return 0; | |
1bd051a6 SH |
1609 | ret = snprintf(kpath, sizeof(kpath), "%s/dev/kmsg", rootfs->mount); |
1610 | if (ret < 0 || ret >= sizeof(kpath)) | |
1611 | return -1; | |
1612 | ||
1613 | ret = unlink(kpath); | |
1614 | if (ret && errno != ENOENT) { | |
1615 | SYSERROR("error unlinking %s\n", kpath); | |
1616 | return -1; | |
1617 | } | |
1618 | ||
1619 | ret = symlink("console", kpath); | |
1620 | if (ret) { | |
1621 | SYSERROR("failed to create symlink for kmsg"); | |
1622 | return -1; | |
1623 | } | |
1624 | ||
1625 | return 0; | |
1626 | } | |
1627 | ||
998ac676 RT |
1628 | static void parse_mntopt(char *opt, unsigned long *flags, char **data) |
1629 | { | |
1630 | struct mount_opt *mo; | |
1631 | ||
1632 | /* If opt is found in mount_opt, set or clear flags. | |
1633 | * Otherwise append it to data. */ | |
1634 | ||
1635 | for (mo = &mount_opt[0]; mo->name != NULL; mo++) { | |
1636 | if (!strncmp(opt, mo->name, strlen(mo->name))) { | |
1637 | if (mo->clear) | |
1638 | *flags &= ~mo->flag; | |
1639 | else | |
1640 | *flags |= mo->flag; | |
1641 | return; | |
1642 | } | |
1643 | } | |
1644 | ||
1645 | if (strlen(*data)) | |
1646 | strcat(*data, ","); | |
1647 | strcat(*data, opt); | |
1648 | } | |
1649 | ||
911324ef | 1650 | static int parse_mntopts(const char *mntopts, unsigned long *mntflags, |
998ac676 RT |
1651 | char **mntdata) |
1652 | { | |
1653 | char *s, *data; | |
1654 | char *p, *saveptr = NULL; | |
1655 | ||
911324ef | 1656 | *mntdata = NULL; |
91656ce5 | 1657 | *mntflags = 0L; |
911324ef DL |
1658 | |
1659 | if (!mntopts) | |
998ac676 RT |
1660 | return 0; |
1661 | ||
911324ef | 1662 | s = strdup(mntopts); |
998ac676 | 1663 | if (!s) { |
36eb9bde | 1664 | SYSERROR("failed to allocate memory"); |
998ac676 RT |
1665 | return -1; |
1666 | } | |
1667 | ||
1668 | data = malloc(strlen(s) + 1); | |
1669 | if (!data) { | |
36eb9bde | 1670 | SYSERROR("failed to allocate memory"); |
998ac676 RT |
1671 | free(s); |
1672 | return -1; | |
1673 | } | |
1674 | *data = 0; | |
1675 | ||
1676 | for (p = strtok_r(s, ",", &saveptr); p != NULL; | |
1677 | p = strtok_r(NULL, ",", &saveptr)) | |
1678 | parse_mntopt(p, mntflags, &data); | |
1679 | ||
1680 | if (*data) | |
1681 | *mntdata = data; | |
1682 | else | |
1683 | free(data); | |
1684 | free(s); | |
1685 | ||
1686 | return 0; | |
1687 | } | |
1688 | ||
911324ef DL |
1689 | static int mount_entry(const char *fsname, const char *target, |
1690 | const char *fstype, unsigned long mountflags, | |
1691 | const char *data) | |
1692 | { | |
1693 | if (mount(fsname, target, fstype, mountflags & ~MS_REMOUNT, data)) { | |
1694 | SYSERROR("failed to mount '%s' on '%s'", fsname, target); | |
1695 | return -1; | |
1696 | } | |
1697 | ||
1698 | if ((mountflags & MS_REMOUNT) || (mountflags & MS_BIND)) { | |
1699 | ||
1700 | DEBUG("remounting %s on %s to respect bind or remount options", | |
1701 | fsname, target); | |
1702 | ||
1703 | if (mount(fsname, target, fstype, | |
1704 | mountflags | MS_REMOUNT, data)) { | |
1705 | SYSERROR("failed to mount '%s' on '%s'", | |
1706 | fsname, target); | |
1707 | return -1; | |
1708 | } | |
1709 | } | |
1710 | ||
1711 | DEBUG("mounted '%s' on '%s', type '%s'", fsname, target, fstype); | |
1712 | ||
1713 | return 0; | |
1714 | } | |
1715 | ||
1716 | static inline int mount_entry_on_systemfs(struct mntent *mntent) | |
0ad19a3f | 1717 | { |
998ac676 RT |
1718 | unsigned long mntflags; |
1719 | char *mntdata; | |
911324ef DL |
1720 | int ret; |
1721 | ||
1722 | if (parse_mntopts(mntent->mnt_opts, &mntflags, &mntdata) < 0) { | |
1723 | ERROR("failed to parse mount option '%s'", mntent->mnt_opts); | |
1724 | return -1; | |
1725 | } | |
1726 | ||
1727 | ret = mount_entry(mntent->mnt_fsname, mntent->mnt_dir, | |
1728 | mntent->mnt_type, mntflags, mntdata); | |
1729 | ||
68c152ef SH |
1730 | if (hasmntopt(mntent, "optional") != NULL) |
1731 | ret = 0; | |
1732 | ||
911324ef DL |
1733 | free(mntdata); |
1734 | ||
1735 | return ret; | |
1736 | } | |
1737 | ||
1738 | static int mount_entry_on_absolute_rootfs(struct mntent *mntent, | |
80a881b2 SH |
1739 | const struct lxc_rootfs *rootfs, |
1740 | const char *lxc_name) | |
911324ef | 1741 | { |
013bd428 | 1742 | char *aux; |
59760f5d | 1743 | char path[MAXPATHLEN]; |
911324ef DL |
1744 | unsigned long mntflags; |
1745 | char *mntdata; | |
80a881b2 | 1746 | int r, ret = 0, offset; |
67e571de | 1747 | const char *lxcpath; |
0ad19a3f | 1748 | |
911324ef DL |
1749 | if (parse_mntopts(mntent->mnt_opts, &mntflags, &mntdata) < 0) { |
1750 | ERROR("failed to parse mount option '%s'", mntent->mnt_opts); | |
1751 | return -1; | |
1752 | } | |
1bc60a65 | 1753 | |
2a59a681 SH |
1754 | lxcpath = default_lxc_path(); |
1755 | if (!lxcpath) { | |
1756 | ERROR("Out of memory"); | |
1757 | return -1; | |
1758 | } | |
1759 | ||
80a881b2 | 1760 | /* if rootfs->path is a blockdev path, allow container fstab to |
2a59a681 SH |
1761 | * use $lxcpath/CN/rootfs as the target prefix */ |
1762 | r = snprintf(path, MAXPATHLEN, "%s/%s/rootfs", lxcpath, lxc_name); | |
80a881b2 SH |
1763 | if (r < 0 || r >= MAXPATHLEN) |
1764 | goto skipvarlib; | |
1765 | ||
1766 | aux = strstr(mntent->mnt_dir, path); | |
1767 | if (aux) { | |
1768 | offset = strlen(path); | |
1769 | goto skipabs; | |
1770 | } | |
1771 | ||
1772 | skipvarlib: | |
013bd428 DL |
1773 | aux = strstr(mntent->mnt_dir, rootfs->path); |
1774 | if (!aux) { | |
1775 | WARN("ignoring mount point '%s'", mntent->mnt_dir); | |
1776 | goto out; | |
1777 | } | |
80a881b2 SH |
1778 | offset = strlen(rootfs->path); |
1779 | ||
1780 | skipabs: | |
013bd428 | 1781 | |
9ba8130c | 1782 | r = snprintf(path, MAXPATHLEN, "%s/%s", rootfs->mount, |
80a881b2 SH |
1783 | aux + offset); |
1784 | if (r < 0 || r >= MAXPATHLEN) { | |
1785 | WARN("pathnme too long for '%s'", mntent->mnt_dir); | |
1786 | ret = -1; | |
1787 | goto out; | |
1788 | } | |
1789 | ||
d330fe7b | 1790 | |
013bd428 | 1791 | ret = mount_entry(mntent->mnt_fsname, path, mntent->mnt_type, |
911324ef | 1792 | mntflags, mntdata); |
0ad19a3f | 1793 | |
68c152ef SH |
1794 | if (hasmntopt(mntent, "optional") != NULL) |
1795 | ret = 0; | |
1796 | ||
013bd428 | 1797 | out: |
911324ef DL |
1798 | free(mntdata); |
1799 | return ret; | |
1800 | } | |
d330fe7b | 1801 | |
911324ef DL |
1802 | static int mount_entry_on_relative_rootfs(struct mntent *mntent, |
1803 | const char *rootfs) | |
1804 | { | |
1805 | char path[MAXPATHLEN]; | |
1806 | unsigned long mntflags; | |
1807 | char *mntdata; | |
1808 | int ret; | |
d330fe7b | 1809 | |
911324ef DL |
1810 | if (parse_mntopts(mntent->mnt_opts, &mntflags, &mntdata) < 0) { |
1811 | ERROR("failed to parse mount option '%s'", mntent->mnt_opts); | |
1812 | return -1; | |
1813 | } | |
d330fe7b | 1814 | |
911324ef | 1815 | /* relative to root mount point */ |
9ba8130c SH |
1816 | ret = snprintf(path, sizeof(path), "%s/%s", rootfs, mntent->mnt_dir); |
1817 | if (ret >= sizeof(path)) { | |
1818 | ERROR("path name too long"); | |
1819 | return -1; | |
1820 | } | |
911324ef DL |
1821 | |
1822 | ret = mount_entry(mntent->mnt_fsname, path, mntent->mnt_type, | |
1823 | mntflags, mntdata); | |
1824 | ||
68c152ef SH |
1825 | if (hasmntopt(mntent, "optional") != NULL) |
1826 | ret = 0; | |
1827 | ||
911324ef | 1828 | free(mntdata); |
998ac676 | 1829 | |
911324ef DL |
1830 | return ret; |
1831 | } | |
1832 | ||
80a881b2 SH |
1833 | static int mount_file_entries(const struct lxc_rootfs *rootfs, FILE *file, |
1834 | const char *lxc_name) | |
911324ef DL |
1835 | { |
1836 | struct mntent *mntent; | |
1837 | int ret = -1; | |
e76b8764 | 1838 | |
911324ef | 1839 | while ((mntent = getmntent(file))) { |
e76b8764 | 1840 | |
911324ef DL |
1841 | if (!rootfs->path) { |
1842 | if (mount_entry_on_systemfs(mntent)) | |
e76b8764 | 1843 | goto out; |
911324ef | 1844 | continue; |
e76b8764 CDC |
1845 | } |
1846 | ||
911324ef DL |
1847 | /* We have a separate root, mounts are relative to it */ |
1848 | if (mntent->mnt_dir[0] != '/') { | |
1849 | if (mount_entry_on_relative_rootfs(mntent, | |
1850 | rootfs->mount)) | |
1851 | goto out; | |
1852 | continue; | |
1853 | } | |
cd54d859 | 1854 | |
80a881b2 | 1855 | if (mount_entry_on_absolute_rootfs(mntent, rootfs, lxc_name)) |
911324ef | 1856 | goto out; |
0ad19a3f | 1857 | } |
cd54d859 | 1858 | |
0ad19a3f | 1859 | ret = 0; |
cd54d859 DL |
1860 | |
1861 | INFO("mount points have been setup"); | |
0ad19a3f | 1862 | out: |
e7938e9e MN |
1863 | return ret; |
1864 | } | |
1865 | ||
80a881b2 SH |
1866 | static int setup_mount(const struct lxc_rootfs *rootfs, const char *fstab, |
1867 | const char *lxc_name) | |
e7938e9e MN |
1868 | { |
1869 | FILE *file; | |
1870 | int ret; | |
1871 | ||
1872 | if (!fstab) | |
1873 | return 0; | |
1874 | ||
025ed0f3 | 1875 | process_lock(); |
e7938e9e | 1876 | file = setmntent(fstab, "r"); |
025ed0f3 | 1877 | process_unlock(); |
e7938e9e MN |
1878 | if (!file) { |
1879 | SYSERROR("failed to use '%s'", fstab); | |
1880 | return -1; | |
1881 | } | |
1882 | ||
80a881b2 | 1883 | ret = mount_file_entries(rootfs, file, lxc_name); |
e7938e9e | 1884 | |
025ed0f3 | 1885 | process_lock(); |
0ad19a3f | 1886 | endmntent(file); |
025ed0f3 | 1887 | process_unlock(); |
0ad19a3f | 1888 | return ret; |
1889 | } | |
1890 | ||
80a881b2 SH |
1891 | static int setup_mount_entries(const struct lxc_rootfs *rootfs, struct lxc_list *mount, |
1892 | const char *lxc_name) | |
e7938e9e MN |
1893 | { |
1894 | FILE *file; | |
1895 | struct lxc_list *iterator; | |
1896 | char *mount_entry; | |
1897 | int ret; | |
1898 | ||
025ed0f3 | 1899 | process_lock(); |
e7938e9e | 1900 | file = tmpfile(); |
025ed0f3 | 1901 | process_unlock(); |
e7938e9e MN |
1902 | if (!file) { |
1903 | ERROR("tmpfile error: %m"); | |
1904 | return -1; | |
1905 | } | |
1906 | ||
1907 | lxc_list_for_each(iterator, mount) { | |
1908 | mount_entry = iterator->elem; | |
1d6b1976 | 1909 | fprintf(file, "%s\n", mount_entry); |
e7938e9e MN |
1910 | } |
1911 | ||
1912 | rewind(file); | |
1913 | ||
80a881b2 | 1914 | ret = mount_file_entries(rootfs, file, lxc_name); |
e7938e9e | 1915 | |
025ed0f3 | 1916 | process_lock(); |
e7938e9e | 1917 | fclose(file); |
025ed0f3 | 1918 | process_unlock(); |
e7938e9e MN |
1919 | return ret; |
1920 | } | |
1921 | ||
81810dd1 DL |
1922 | static int setup_caps(struct lxc_list *caps) |
1923 | { | |
1924 | struct lxc_list *iterator; | |
1925 | char *drop_entry; | |
d55bc1ad | 1926 | char *ptr; |
81810dd1 DL |
1927 | int i, capid; |
1928 | ||
1929 | lxc_list_for_each(iterator, caps) { | |
1930 | ||
1931 | drop_entry = iterator->elem; | |
1932 | ||
1933 | capid = -1; | |
1934 | ||
1935 | for (i = 0; i < sizeof(caps_opt)/sizeof(caps_opt[0]); i++) { | |
1936 | ||
1937 | if (strcmp(drop_entry, caps_opt[i].name)) | |
1938 | continue; | |
1939 | ||
1940 | capid = caps_opt[i].value; | |
1941 | break; | |
1942 | } | |
1943 | ||
d55bc1ad CS |
1944 | if (capid < 0) { |
1945 | /* try to see if it's numeric, so the user may specify | |
1946 | * capabilities that the running kernel knows about but | |
1947 | * we don't */ | |
1948 | capid = strtol(drop_entry, &ptr, 10); | |
1949 | if (!ptr || *ptr != '\0' || | |
1950 | capid == LONG_MIN || capid == LONG_MAX) | |
1951 | /* not a valid number */ | |
1952 | capid = -1; | |
1953 | else if (capid > lxc_caps_last_cap()) | |
1954 | /* we have a number but it's not a valid | |
1955 | * capability */ | |
1956 | capid = -1; | |
1957 | } | |
1958 | ||
81810dd1 | 1959 | if (capid < 0) { |
1e11be34 DL |
1960 | ERROR("unknown capability %s", drop_entry); |
1961 | return -1; | |
81810dd1 DL |
1962 | } |
1963 | ||
1964 | DEBUG("drop capability '%s' (%d)", drop_entry, capid); | |
1965 | ||
1966 | if (prctl(PR_CAPBSET_DROP, capid, 0, 0, 0)) { | |
1967 | SYSERROR("failed to remove %s capability", drop_entry); | |
1968 | return -1; | |
1969 | } | |
1970 | ||
1971 | } | |
1972 | ||
1fb86a7c SH |
1973 | DEBUG("capabilities have been setup"); |
1974 | ||
1975 | return 0; | |
1976 | } | |
1977 | ||
1978 | static int dropcaps_except(struct lxc_list *caps) | |
1979 | { | |
1980 | struct lxc_list *iterator; | |
1981 | char *keep_entry; | |
1982 | char *ptr; | |
1983 | int i, capid; | |
1984 | int numcaps = lxc_caps_last_cap() + 1; | |
1985 | INFO("found %d capabilities\n", numcaps); | |
1986 | ||
2caf9a97 SH |
1987 | if (numcaps <= 0 || numcaps > 200) |
1988 | return -1; | |
1989 | ||
1fb86a7c SH |
1990 | // caplist[i] is 1 if we keep capability i |
1991 | int *caplist = alloca(numcaps * sizeof(int)); | |
1992 | memset(caplist, 0, numcaps * sizeof(int)); | |
1993 | ||
1994 | lxc_list_for_each(iterator, caps) { | |
1995 | ||
1996 | keep_entry = iterator->elem; | |
1997 | ||
1998 | capid = -1; | |
1999 | ||
2000 | for (i = 0; i < sizeof(caps_opt)/sizeof(caps_opt[0]); i++) { | |
2001 | ||
2002 | if (strcmp(keep_entry, caps_opt[i].name)) | |
2003 | continue; | |
2004 | ||
2005 | capid = caps_opt[i].value; | |
2006 | break; | |
2007 | } | |
2008 | ||
2009 | if (capid < 0) { | |
2010 | /* try to see if it's numeric, so the user may specify | |
2011 | * capabilities that the running kernel knows about but | |
2012 | * we don't */ | |
2013 | capid = strtol(keep_entry, &ptr, 10); | |
2014 | if (!ptr || *ptr != '\0' || | |
2015 | capid == LONG_MIN || capid == LONG_MAX) | |
2016 | /* not a valid number */ | |
2017 | capid = -1; | |
2018 | else if (capid > lxc_caps_last_cap()) | |
2019 | /* we have a number but it's not a valid | |
2020 | * capability */ | |
2021 | capid = -1; | |
2022 | } | |
2023 | ||
2024 | if (capid < 0) { | |
2025 | ERROR("unknown capability %s", keep_entry); | |
2026 | return -1; | |
2027 | } | |
2028 | ||
2029 | DEBUG("drop capability '%s' (%d)", keep_entry, capid); | |
2030 | ||
2031 | caplist[capid] = 1; | |
2032 | } | |
2033 | for (i=0; i<numcaps; i++) { | |
2034 | if (caplist[i]) | |
2035 | continue; | |
2036 | if (prctl(PR_CAPBSET_DROP, i, 0, 0, 0)) { | |
2037 | SYSERROR("failed to remove capability %d", i); | |
2038 | return -1; | |
2039 | } | |
2040 | } | |
2041 | ||
2042 | DEBUG("capabilities have been setup"); | |
81810dd1 DL |
2043 | |
2044 | return 0; | |
2045 | } | |
2046 | ||
0ad19a3f | 2047 | static int setup_hw_addr(char *hwaddr, const char *ifname) |
2048 | { | |
2049 | struct sockaddr sockaddr; | |
2050 | struct ifreq ifr; | |
2051 | int ret, fd; | |
2052 | ||
3cfc0f3a MN |
2053 | ret = lxc_convert_mac(hwaddr, &sockaddr); |
2054 | if (ret) { | |
2055 | ERROR("mac address '%s' conversion failed : %s", | |
2056 | hwaddr, strerror(-ret)); | |
0ad19a3f | 2057 | return -1; |
2058 | } | |
2059 | ||
2060 | memcpy(ifr.ifr_name, ifname, IFNAMSIZ); | |
2061 | memcpy((char *) &ifr.ifr_hwaddr, (char *) &sockaddr, sizeof(sockaddr)); | |
2062 | ||
025ed0f3 | 2063 | process_lock(); |
0ad19a3f | 2064 | fd = socket(AF_INET, SOCK_DGRAM, 0); |
025ed0f3 | 2065 | process_unlock(); |
0ad19a3f | 2066 | if (fd < 0) { |
3ab87b66 | 2067 | ERROR("socket failure : %s", strerror(errno)); |
0ad19a3f | 2068 | return -1; |
2069 | } | |
2070 | ||
2071 | ret = ioctl(fd, SIOCSIFHWADDR, &ifr); | |
025ed0f3 | 2072 | process_lock(); |
0ad19a3f | 2073 | close(fd); |
025ed0f3 | 2074 | process_unlock(); |
0ad19a3f | 2075 | if (ret) |
3ab87b66 | 2076 | ERROR("ioctl failure : %s", strerror(errno)); |
0ad19a3f | 2077 | |
cd54d859 DL |
2078 | DEBUG("mac address '%s' on '%s' has been setup", hwaddr, ifname); |
2079 | ||
0ad19a3f | 2080 | return ret; |
2081 | } | |
2082 | ||
82d5ae15 | 2083 | static int setup_ipv4_addr(struct lxc_list *ip, int ifindex) |
0ad19a3f | 2084 | { |
82d5ae15 DL |
2085 | struct lxc_list *iterator; |
2086 | struct lxc_inetdev *inetdev; | |
3cfc0f3a | 2087 | int err; |
0ad19a3f | 2088 | |
82d5ae15 DL |
2089 | lxc_list_for_each(iterator, ip) { |
2090 | ||
2091 | inetdev = iterator->elem; | |
2092 | ||
0093bb8c DL |
2093 | err = lxc_ipv4_addr_add(ifindex, &inetdev->addr, |
2094 | &inetdev->bcast, inetdev->prefix); | |
3cfc0f3a MN |
2095 | if (err) { |
2096 | ERROR("failed to setup_ipv4_addr ifindex %d : %s", | |
2097 | ifindex, strerror(-err)); | |
82d5ae15 DL |
2098 | return -1; |
2099 | } | |
2100 | } | |
2101 | ||
2102 | return 0; | |
0ad19a3f | 2103 | } |
2104 | ||
82d5ae15 | 2105 | static int setup_ipv6_addr(struct lxc_list *ip, int ifindex) |
0ad19a3f | 2106 | { |
82d5ae15 | 2107 | struct lxc_list *iterator; |
7fa9074f | 2108 | struct lxc_inet6dev *inet6dev; |
3cfc0f3a | 2109 | int err; |
0ad19a3f | 2110 | |
82d5ae15 DL |
2111 | lxc_list_for_each(iterator, ip) { |
2112 | ||
2113 | inet6dev = iterator->elem; | |
2114 | ||
b3df193c | 2115 | err = lxc_ipv6_addr_add(ifindex, &inet6dev->addr, |
0093bb8c DL |
2116 | &inet6dev->mcast, &inet6dev->acast, |
2117 | inet6dev->prefix); | |
3cfc0f3a MN |
2118 | if (err) { |
2119 | ERROR("failed to setup_ipv6_addr ifindex %d : %s", | |
2120 | ifindex, strerror(-err)); | |
82d5ae15 | 2121 | return -1; |
3cfc0f3a | 2122 | } |
82d5ae15 DL |
2123 | } |
2124 | ||
2125 | return 0; | |
0ad19a3f | 2126 | } |
2127 | ||
82d5ae15 | 2128 | static int setup_netdev(struct lxc_netdev *netdev) |
0ad19a3f | 2129 | { |
0ad19a3f | 2130 | char ifname[IFNAMSIZ]; |
0ad19a3f | 2131 | char *current_ifname = ifname; |
3cfc0f3a | 2132 | int err; |
0ad19a3f | 2133 | |
82d5ae15 DL |
2134 | /* empty network namespace */ |
2135 | if (!netdev->ifindex) { | |
b0efbac4 | 2136 | if (netdev->flags & IFF_UP) { |
d472214b | 2137 | err = lxc_netdev_up("lo"); |
3cfc0f3a MN |
2138 | if (err) { |
2139 | ERROR("failed to set the loopback up : %s", | |
2140 | strerror(-err)); | |
82d5ae15 DL |
2141 | return -1; |
2142 | } | |
82d5ae15 | 2143 | } |
7b57e8b6 | 2144 | return 0; |
0ad19a3f | 2145 | } |
13954cce | 2146 | |
82d5ae15 DL |
2147 | /* retrieve the name of the interface */ |
2148 | if (!if_indextoname(netdev->ifindex, current_ifname)) { | |
36eb9bde | 2149 | ERROR("no interface corresponding to index '%d'", |
82d5ae15 | 2150 | netdev->ifindex); |
0ad19a3f | 2151 | return -1; |
2152 | } | |
13954cce | 2153 | |
018ef520 | 2154 | /* default: let the system to choose one interface name */ |
9d083402 | 2155 | if (!netdev->name) |
fb6d9b2f DL |
2156 | netdev->name = netdev->type == LXC_NET_PHYS ? |
2157 | netdev->link : "eth%d"; | |
018ef520 | 2158 | |
82d5ae15 | 2159 | /* rename the interface name */ |
b84f58b9 | 2160 | err = lxc_netdev_rename_by_name(ifname, netdev->name); |
3cfc0f3a MN |
2161 | if (err) { |
2162 | ERROR("failed to rename %s->%s : %s", ifname, netdev->name, | |
2163 | strerror(-err)); | |
018ef520 DL |
2164 | return -1; |
2165 | } | |
2166 | ||
2167 | /* Re-read the name of the interface because its name has changed | |
2168 | * and would be automatically allocated by the system | |
2169 | */ | |
82d5ae15 | 2170 | if (!if_indextoname(netdev->ifindex, current_ifname)) { |
018ef520 | 2171 | ERROR("no interface corresponding to index '%d'", |
82d5ae15 | 2172 | netdev->ifindex); |
018ef520 | 2173 | return -1; |
0ad19a3f | 2174 | } |
2175 | ||
82d5ae15 DL |
2176 | /* set a mac address */ |
2177 | if (netdev->hwaddr) { | |
2178 | if (setup_hw_addr(netdev->hwaddr, current_ifname)) { | |
36eb9bde | 2179 | ERROR("failed to setup hw address for '%s'", |
82d5ae15 | 2180 | current_ifname); |
0ad19a3f | 2181 | return -1; |
2182 | } | |
2183 | } | |
2184 | ||
82d5ae15 DL |
2185 | /* setup ipv4 addresses on the interface */ |
2186 | if (setup_ipv4_addr(&netdev->ipv4, netdev->ifindex)) { | |
36eb9bde | 2187 | ERROR("failed to setup ip addresses for '%s'", |
0ad19a3f | 2188 | ifname); |
2189 | return -1; | |
2190 | } | |
2191 | ||
82d5ae15 DL |
2192 | /* setup ipv6 addresses on the interface */ |
2193 | if (setup_ipv6_addr(&netdev->ipv6, netdev->ifindex)) { | |
36eb9bde | 2194 | ERROR("failed to setup ipv6 addresses for '%s'", |
0ad19a3f | 2195 | ifname); |
2196 | return -1; | |
2197 | } | |
2198 | ||
82d5ae15 | 2199 | /* set the network device up */ |
b0efbac4 | 2200 | if (netdev->flags & IFF_UP) { |
3cfc0f3a MN |
2201 | int err; |
2202 | ||
d472214b | 2203 | err = lxc_netdev_up(current_ifname); |
3cfc0f3a MN |
2204 | if (err) { |
2205 | ERROR("failed to set '%s' up : %s", current_ifname, | |
2206 | strerror(-err)); | |
0ad19a3f | 2207 | return -1; |
2208 | } | |
2209 | ||
2210 | /* the network is up, make the loopback up too */ | |
d472214b | 2211 | err = lxc_netdev_up("lo"); |
3cfc0f3a MN |
2212 | if (err) { |
2213 | ERROR("failed to set the loopback up : %s", | |
2214 | strerror(-err)); | |
0ad19a3f | 2215 | return -1; |
2216 | } | |
2217 | } | |
2218 | ||
f8fee0e2 MK |
2219 | /* We can only set up the default routes after bringing |
2220 | * up the interface, sine bringing up the interface adds | |
2221 | * the link-local routes and we can't add a default | |
2222 | * route if the gateway is not reachable. */ | |
2223 | ||
2224 | /* setup ipv4 gateway on the interface */ | |
2225 | if (netdev->ipv4_gateway) { | |
2226 | if (!(netdev->flags & IFF_UP)) { | |
2227 | ERROR("Cannot add ipv4 gateway for %s when not bringing up the interface", ifname); | |
2228 | return -1; | |
2229 | } | |
2230 | ||
2231 | if (lxc_list_empty(&netdev->ipv4)) { | |
2232 | ERROR("Cannot add ipv4 gateway for %s when not assigning an address", ifname); | |
2233 | return -1; | |
2234 | } | |
2235 | ||
2236 | err = lxc_ipv4_gateway_add(netdev->ifindex, netdev->ipv4_gateway); | |
2237 | if (err) { | |
2238 | ERROR("failed to setup ipv4 gateway for '%s': %s", | |
2239 | ifname, strerror(-err)); | |
19a26f82 MK |
2240 | if (netdev->ipv4_gateway_auto) { |
2241 | char buf[INET_ADDRSTRLEN]; | |
2242 | inet_ntop(AF_INET, netdev->ipv4_gateway, buf, sizeof(buf)); | |
2243 | ERROR("tried to set autodetected ipv4 gateway '%s'", buf); | |
2244 | } | |
f8fee0e2 MK |
2245 | return -1; |
2246 | } | |
2247 | } | |
2248 | ||
2249 | /* setup ipv6 gateway on the interface */ | |
2250 | if (netdev->ipv6_gateway) { | |
2251 | if (!(netdev->flags & IFF_UP)) { | |
2252 | ERROR("Cannot add ipv6 gateway for %s when not bringing up the interface", ifname); | |
2253 | return -1; | |
2254 | } | |
2255 | ||
2256 | if (lxc_list_empty(&netdev->ipv6) && !IN6_IS_ADDR_LINKLOCAL(netdev->ipv6_gateway)) { | |
2257 | ERROR("Cannot add ipv6 gateway for %s when not assigning an address", ifname); | |
2258 | return -1; | |
2259 | } | |
2260 | ||
2261 | err = lxc_ipv6_gateway_add(netdev->ifindex, netdev->ipv6_gateway); | |
2262 | if (err) { | |
2263 | ERROR("failed to setup ipv6 gateway for '%s': %s", | |
2264 | ifname, strerror(-err)); | |
19a26f82 MK |
2265 | if (netdev->ipv6_gateway_auto) { |
2266 | char buf[INET6_ADDRSTRLEN]; | |
72d0e1cb | 2267 | inet_ntop(AF_INET6, netdev->ipv6_gateway, buf, sizeof(buf)); |
19a26f82 MK |
2268 | ERROR("tried to set autodetected ipv6 gateway '%s'", buf); |
2269 | } | |
f8fee0e2 MK |
2270 | return -1; |
2271 | } | |
2272 | } | |
2273 | ||
cd54d859 DL |
2274 | DEBUG("'%s' has been setup", current_ifname); |
2275 | ||
0ad19a3f | 2276 | return 0; |
2277 | } | |
2278 | ||
5f4535a3 | 2279 | static int setup_network(struct lxc_list *network) |
0ad19a3f | 2280 | { |
82d5ae15 | 2281 | struct lxc_list *iterator; |
82d5ae15 | 2282 | struct lxc_netdev *netdev; |
0ad19a3f | 2283 | |
5f4535a3 | 2284 | lxc_list_for_each(iterator, network) { |
cd54d859 | 2285 | |
5f4535a3 | 2286 | netdev = iterator->elem; |
82d5ae15 DL |
2287 | |
2288 | if (setup_netdev(netdev)) { | |
2289 | ERROR("failed to setup netdev"); | |
2290 | return -1; | |
2291 | } | |
2292 | } | |
cd54d859 | 2293 | |
5f4535a3 DL |
2294 | if (!lxc_list_empty(network)) |
2295 | INFO("network has been setup"); | |
cd54d859 DL |
2296 | |
2297 | return 0; | |
0ad19a3f | 2298 | } |
2299 | ||
7b35f3d6 SH |
2300 | void lxc_rename_phys_nics_on_shutdown(struct lxc_conf *conf) |
2301 | { | |
2302 | int i; | |
2303 | ||
2304 | INFO("running to reset %d nic names", conf->num_savednics); | |
2305 | for (i=0; i<conf->num_savednics; i++) { | |
2306 | struct saved_nic *s = &conf->saved_nics[i]; | |
2307 | INFO("resetting nic %d to %s\n", s->ifindex, s->orig_name); | |
2308 | lxc_netdev_rename_by_index(s->ifindex, s->orig_name); | |
2309 | free(s->orig_name); | |
2310 | } | |
2311 | conf->num_savednics = 0; | |
2312 | free(conf->saved_nics); | |
2313 | } | |
2314 | ||
49684c0b CS |
2315 | static int setup_private_host_hw_addr(char *veth1) |
2316 | { | |
2317 | struct ifreq ifr; | |
2318 | int err; | |
2319 | int sockfd; | |
2320 | ||
025ed0f3 | 2321 | process_lock(); |
49684c0b | 2322 | sockfd = socket(AF_INET, SOCK_DGRAM, 0); |
025ed0f3 | 2323 | process_unlock(); |
49684c0b CS |
2324 | if (sockfd < 0) |
2325 | return -errno; | |
2326 | ||
2327 | snprintf((char *)ifr.ifr_name, IFNAMSIZ, "%s", veth1); | |
2328 | err = ioctl(sockfd, SIOCGIFHWADDR, &ifr); | |
2329 | if (err < 0) { | |
025ed0f3 | 2330 | process_lock(); |
49684c0b | 2331 | close(sockfd); |
025ed0f3 | 2332 | process_unlock(); |
49684c0b CS |
2333 | return -errno; |
2334 | } | |
2335 | ||
2336 | ifr.ifr_hwaddr.sa_data[0] = 0xfe; | |
2337 | err = ioctl(sockfd, SIOCSIFHWADDR, &ifr); | |
025ed0f3 | 2338 | process_lock(); |
49684c0b | 2339 | close(sockfd); |
025ed0f3 | 2340 | process_unlock(); |
49684c0b CS |
2341 | if (err < 0) |
2342 | return -errno; | |
2343 | ||
7ad84da7 DL |
2344 | DEBUG("mac address of host interface '%s' changed to private " |
2345 | "%02x:%02x:%02x:%02x:%02x:%02x", veth1, | |
2346 | ifr.ifr_hwaddr.sa_data[0] & 0xff, | |
2347 | ifr.ifr_hwaddr.sa_data[1] & 0xff, | |
2348 | ifr.ifr_hwaddr.sa_data[2] & 0xff, | |
2349 | ifr.ifr_hwaddr.sa_data[3] & 0xff, | |
2350 | ifr.ifr_hwaddr.sa_data[4] & 0xff, | |
2351 | ifr.ifr_hwaddr.sa_data[5] & 0xff); | |
49684c0b CS |
2352 | |
2353 | return 0; | |
2354 | } | |
2355 | ||
ae9242c8 SH |
2356 | static char *default_rootfs_mount = LXCROOTFSMOUNT; |
2357 | ||
7b379ab3 | 2358 | struct lxc_conf *lxc_conf_init(void) |
089cd8b8 | 2359 | { |
7b379ab3 | 2360 | struct lxc_conf *new; |
26ddeedd | 2361 | int i; |
7b379ab3 MN |
2362 | |
2363 | new = malloc(sizeof(*new)); | |
2364 | if (!new) { | |
2365 | ERROR("lxc_conf_init : %m"); | |
2366 | return NULL; | |
2367 | } | |
2368 | memset(new, 0, sizeof(*new)); | |
2369 | ||
b40a606e | 2370 | new->loglevel = LXC_LOG_PRIORITY_NOTSET; |
cccc74b5 | 2371 | new->personality = -1; |
596a818d DE |
2372 | new->console.log_path = NULL; |
2373 | new->console.log_fd = -1; | |
28a4b0e5 | 2374 | new->console.path = NULL; |
63376d7d | 2375 | new->console.peer = -1; |
b5159817 DE |
2376 | new->console.peerpty.busy = -1; |
2377 | new->console.peerpty.master = -1; | |
2378 | new->console.peerpty.slave = -1; | |
63376d7d DL |
2379 | new->console.master = -1; |
2380 | new->console.slave = -1; | |
2381 | new->console.name[0] = '\0'; | |
d2e30e99 | 2382 | new->maincmd_fd = -1; |
54c30e29 | 2383 | new->rootfs.mount = strdup(default_rootfs_mount); |
53f3f048 SH |
2384 | if (!new->rootfs.mount) { |
2385 | ERROR("lxc_conf_init : %m"); | |
2386 | free(new); | |
2387 | return NULL; | |
2388 | } | |
2f3f41d0 | 2389 | new->kmsg = 1; |
7b379ab3 MN |
2390 | lxc_list_init(&new->cgroup); |
2391 | lxc_list_init(&new->network); | |
2392 | lxc_list_init(&new->mount_list); | |
81810dd1 | 2393 | lxc_list_init(&new->caps); |
1fb86a7c | 2394 | lxc_list_init(&new->keepcaps); |
f6d3e3e4 | 2395 | lxc_list_init(&new->id_map); |
26ddeedd SH |
2396 | for (i=0; i<NUM_LXC_HOOKS; i++) |
2397 | lxc_list_init(&new->hooks[i]); | |
fe4de9a6 DE |
2398 | new->lsm_aa_profile = NULL; |
2399 | new->lsm_se_context = NULL; | |
e075f5d9 | 2400 | new->lsm_umount_proc = 0; |
7b379ab3 MN |
2401 | |
2402 | return new; | |
089cd8b8 DL |
2403 | } |
2404 | ||
e3b4c4c4 | 2405 | static int instanciate_veth(struct lxc_handler *handler, struct lxc_netdev *netdev) |
0ad19a3f | 2406 | { |
8634bc19 | 2407 | char veth1buf[IFNAMSIZ], *veth1; |
0e391e57 | 2408 | char veth2buf[IFNAMSIZ], *veth2; |
3cfc0f3a | 2409 | int err; |
13954cce | 2410 | |
e892973e DL |
2411 | if (netdev->priv.veth_attr.pair) |
2412 | veth1 = netdev->priv.veth_attr.pair; | |
8634bc19 | 2413 | else { |
9ba8130c SH |
2414 | err = snprintf(veth1buf, sizeof(veth1buf), "vethXXXXXX"); |
2415 | if (err >= sizeof(veth1buf)) { /* can't *really* happen, but... */ | |
2416 | ERROR("veth1 name too long"); | |
2417 | return -1; | |
2418 | } | |
4a0ba80d | 2419 | veth1 = mkifname(veth1buf); |
74a2b586 JK |
2420 | /* store away for deconf */ |
2421 | memcpy(netdev->priv.veth_attr.veth1, veth1, IFNAMSIZ); | |
8634bc19 | 2422 | } |
82d5ae15 | 2423 | |
0e391e57 | 2424 | snprintf(veth2buf, sizeof(veth2buf), "vethXXXXXX"); |
4a0ba80d | 2425 | veth2 = mkifname(veth2buf); |
82d5ae15 DL |
2426 | |
2427 | if (!strlen(veth1) || !strlen(veth2)) { | |
2428 | ERROR("failed to allocate a temporary name"); | |
2429 | return -1; | |
0ad19a3f | 2430 | } |
2431 | ||
3cfc0f3a MN |
2432 | err = lxc_veth_create(veth1, veth2); |
2433 | if (err) { | |
2434 | ERROR("failed to create %s-%s : %s", veth1, veth2, | |
2435 | strerror(-err)); | |
6ab9ab6d | 2436 | return -1; |
0ad19a3f | 2437 | } |
13954cce | 2438 | |
49684c0b CS |
2439 | /* changing the high byte of the mac address to 0xfe, the bridge interface |
2440 | * will always keep the host's mac address and not take the mac address | |
2441 | * of a container */ | |
2442 | err = setup_private_host_hw_addr(veth1); | |
2443 | if (err) { | |
2444 | ERROR("failed to change mac address of host interface '%s' : %s", | |
2445 | veth1, strerror(-err)); | |
2446 | goto out_delete; | |
2447 | } | |
2448 | ||
82d5ae15 | 2449 | if (netdev->mtu) { |
d472214b | 2450 | err = lxc_netdev_set_mtu(veth1, atoi(netdev->mtu)); |
3cfc0f3a | 2451 | if (!err) |
d472214b | 2452 | err = lxc_netdev_set_mtu(veth2, atoi(netdev->mtu)); |
3cfc0f3a MN |
2453 | if (err) { |
2454 | ERROR("failed to set mtu '%s' for %s-%s : %s", | |
2455 | netdev->mtu, veth1, veth2, strerror(-err)); | |
eb14c10a | 2456 | goto out_delete; |
75d09f83 DL |
2457 | } |
2458 | } | |
2459 | ||
3cfc0f3a MN |
2460 | if (netdev->link) { |
2461 | err = lxc_bridge_attach(netdev->link, veth1); | |
2462 | if (err) { | |
2463 | ERROR("failed to attach '%s' to the bridge '%s' : %s", | |
2464 | veth1, netdev->link, strerror(-err)); | |
2465 | goto out_delete; | |
2466 | } | |
eb14c10a DL |
2467 | } |
2468 | ||
82d5ae15 DL |
2469 | netdev->ifindex = if_nametoindex(veth2); |
2470 | if (!netdev->ifindex) { | |
36eb9bde | 2471 | ERROR("failed to retrieve the index for %s", veth2); |
eb14c10a DL |
2472 | goto out_delete; |
2473 | } | |
2474 | ||
d472214b | 2475 | err = lxc_netdev_up(veth1); |
6e35af2e DL |
2476 | if (err) { |
2477 | ERROR("failed to set %s up : %s", veth1, strerror(-err)); | |
2478 | goto out_delete; | |
0ad19a3f | 2479 | } |
2480 | ||
e3b4c4c4 | 2481 | if (netdev->upscript) { |
751d9dcd DL |
2482 | err = run_script(handler->name, "net", netdev->upscript, "up", |
2483 | "veth", veth1, (char*) NULL); | |
2484 | if (err) | |
e3b4c4c4 | 2485 | goto out_delete; |
e3b4c4c4 ST |
2486 | } |
2487 | ||
82d5ae15 DL |
2488 | DEBUG("instanciated veth '%s/%s', index is '%d'", |
2489 | veth1, veth2, netdev->ifindex); | |
2490 | ||
6ab9ab6d | 2491 | return 0; |
eb14c10a DL |
2492 | |
2493 | out_delete: | |
b84f58b9 | 2494 | lxc_netdev_delete_by_name(veth1); |
6ab9ab6d | 2495 | return -1; |
13954cce | 2496 | } |
d957ae2d | 2497 | |
74a2b586 JK |
2498 | static int shutdown_veth(struct lxc_handler *handler, struct lxc_netdev *netdev) |
2499 | { | |
2500 | char *veth1; | |
2501 | int err; | |
2502 | ||
2503 | if (netdev->priv.veth_attr.pair) | |
2504 | veth1 = netdev->priv.veth_attr.pair; | |
2505 | else | |
2506 | veth1 = netdev->priv.veth_attr.veth1; | |
2507 | ||
2508 | if (netdev->downscript) { | |
2509 | err = run_script(handler->name, "net", netdev->downscript, | |
2510 | "down", "veth", veth1, (char*) NULL); | |
2511 | if (err) | |
2512 | return -1; | |
2513 | } | |
2514 | return 0; | |
2515 | } | |
2516 | ||
e3b4c4c4 | 2517 | static int instanciate_macvlan(struct lxc_handler *handler, struct lxc_netdev *netdev) |
0ad19a3f | 2518 | { |
0e391e57 | 2519 | char peerbuf[IFNAMSIZ], *peer; |
3cfc0f3a | 2520 | int err; |
d957ae2d MT |
2521 | |
2522 | if (!netdev->link) { | |
2523 | ERROR("no link specified for macvlan netdev"); | |
2524 | return -1; | |
2525 | } | |
13954cce | 2526 | |
9ba8130c SH |
2527 | err = snprintf(peerbuf, sizeof(peerbuf), "mcXXXXXX"); |
2528 | if (err >= sizeof(peerbuf)) | |
2529 | return -1; | |
82d5ae15 | 2530 | |
4a0ba80d | 2531 | peer = mkifname(peerbuf); |
82d5ae15 DL |
2532 | if (!strlen(peer)) { |
2533 | ERROR("failed to make a temporary name"); | |
2534 | return -1; | |
0ad19a3f | 2535 | } |
2536 | ||
3cfc0f3a MN |
2537 | err = lxc_macvlan_create(netdev->link, peer, |
2538 | netdev->priv.macvlan_attr.mode); | |
2539 | if (err) { | |
2540 | ERROR("failed to create macvlan interface '%s' on '%s' : %s", | |
2541 | peer, netdev->link, strerror(-err)); | |
d957ae2d | 2542 | return -1; |
0ad19a3f | 2543 | } |
2544 | ||
82d5ae15 DL |
2545 | netdev->ifindex = if_nametoindex(peer); |
2546 | if (!netdev->ifindex) { | |
36eb9bde | 2547 | ERROR("failed to retrieve the index for %s", peer); |
b84f58b9 | 2548 | lxc_netdev_delete_by_name(peer); |
d957ae2d | 2549 | return -1; |
22ebac19 | 2550 | } |
2551 | ||
e3b4c4c4 | 2552 | if (netdev->upscript) { |
751d9dcd DL |
2553 | err = run_script(handler->name, "net", netdev->upscript, "up", |
2554 | "macvlan", netdev->link, (char*) NULL); | |
2555 | if (err) | |
e3b4c4c4 | 2556 | return -1; |
e3b4c4c4 ST |
2557 | } |
2558 | ||
e892973e DL |
2559 | DEBUG("instanciated macvlan '%s', index is '%d' and mode '%d'", |
2560 | peer, netdev->ifindex, netdev->priv.macvlan_attr.mode); | |
0ad19a3f | 2561 | |
d957ae2d | 2562 | return 0; |
0ad19a3f | 2563 | } |
2564 | ||
74a2b586 JK |
2565 | static int shutdown_macvlan(struct lxc_handler *handler, struct lxc_netdev *netdev) |
2566 | { | |
2567 | int err; | |
2568 | ||
2569 | if (netdev->downscript) { | |
2570 | err = run_script(handler->name, "net", netdev->downscript, | |
2571 | "down", "macvlan", netdev->link, | |
2572 | (char*) NULL); | |
2573 | if (err) | |
2574 | return -1; | |
2575 | } | |
2576 | return 0; | |
2577 | } | |
2578 | ||
26c39028 | 2579 | /* XXX: merge with instanciate_macvlan */ |
e3b4c4c4 | 2580 | static int instanciate_vlan(struct lxc_handler *handler, struct lxc_netdev *netdev) |
26c39028 JHS |
2581 | { |
2582 | char peer[IFNAMSIZ]; | |
3cfc0f3a | 2583 | int err; |
26c39028 JHS |
2584 | |
2585 | if (!netdev->link) { | |
2586 | ERROR("no link specified for vlan netdev"); | |
2587 | return -1; | |
2588 | } | |
2589 | ||
9ba8130c SH |
2590 | err = snprintf(peer, sizeof(peer), "vlan%d", netdev->priv.vlan_attr.vid); |
2591 | if (err >= sizeof(peer)) { | |
2592 | ERROR("peer name too long"); | |
2593 | return -1; | |
2594 | } | |
26c39028 | 2595 | |
3cfc0f3a MN |
2596 | err = lxc_vlan_create(netdev->link, peer, netdev->priv.vlan_attr.vid); |
2597 | if (err) { | |
2598 | ERROR("failed to create vlan interface '%s' on '%s' : %s", | |
2599 | peer, netdev->link, strerror(-err)); | |
26c39028 JHS |
2600 | return -1; |
2601 | } | |
2602 | ||
2603 | netdev->ifindex = if_nametoindex(peer); | |
2604 | if (!netdev->ifindex) { | |
2605 | ERROR("failed to retrieve the ifindex for %s", peer); | |
b84f58b9 | 2606 | lxc_netdev_delete_by_name(peer); |
26c39028 JHS |
2607 | return -1; |
2608 | } | |
2609 | ||
e892973e DL |
2610 | DEBUG("instanciated vlan '%s', ifindex is '%d'", " vlan1000", |
2611 | netdev->ifindex); | |
2612 | ||
26c39028 JHS |
2613 | return 0; |
2614 | } | |
2615 | ||
74a2b586 JK |
2616 | static int shutdown_vlan(struct lxc_handler *handler, struct lxc_netdev *netdev) |
2617 | { | |
2618 | return 0; | |
2619 | } | |
2620 | ||
e3b4c4c4 | 2621 | static int instanciate_phys(struct lxc_handler *handler, struct lxc_netdev *netdev) |
0ad19a3f | 2622 | { |
6168e99f DL |
2623 | if (!netdev->link) { |
2624 | ERROR("no link specified for the physical interface"); | |
2625 | return -1; | |
2626 | } | |
2627 | ||
9d083402 | 2628 | netdev->ifindex = if_nametoindex(netdev->link); |
82d5ae15 | 2629 | if (!netdev->ifindex) { |
9d083402 | 2630 | ERROR("failed to retrieve the index for %s", netdev->link); |
0ad19a3f | 2631 | return -1; |
2632 | } | |
2633 | ||
e3b4c4c4 ST |
2634 | if (netdev->upscript) { |
2635 | int err; | |
751d9dcd DL |
2636 | err = run_script(handler->name, "net", netdev->upscript, |
2637 | "up", "phys", netdev->link, (char*) NULL); | |
2638 | if (err) | |
e3b4c4c4 | 2639 | return -1; |
e3b4c4c4 ST |
2640 | } |
2641 | ||
82d5ae15 | 2642 | return 0; |
0ad19a3f | 2643 | } |
2644 | ||
74a2b586 JK |
2645 | static int shutdown_phys(struct lxc_handler *handler, struct lxc_netdev *netdev) |
2646 | { | |
2647 | int err; | |
2648 | ||
2649 | if (netdev->downscript) { | |
2650 | err = run_script(handler->name, "net", netdev->downscript, | |
2651 | "down", "phys", netdev->link, (char*) NULL); | |
2652 | if (err) | |
2653 | return -1; | |
2654 | } | |
2655 | return 0; | |
2656 | } | |
2657 | ||
e3b4c4c4 | 2658 | static int instanciate_empty(struct lxc_handler *handler, struct lxc_netdev *netdev) |
0ad19a3f | 2659 | { |
82d5ae15 | 2660 | netdev->ifindex = 0; |
e3b4c4c4 ST |
2661 | if (netdev->upscript) { |
2662 | int err; | |
751d9dcd DL |
2663 | err = run_script(handler->name, "net", netdev->upscript, |
2664 | "up", "empty", (char*) NULL); | |
2665 | if (err) | |
e3b4c4c4 | 2666 | return -1; |
e3b4c4c4 | 2667 | } |
82d5ae15 | 2668 | return 0; |
0ad19a3f | 2669 | } |
2670 | ||
74a2b586 JK |
2671 | static int shutdown_empty(struct lxc_handler *handler, struct lxc_netdev *netdev) |
2672 | { | |
2673 | int err; | |
2674 | ||
2675 | if (netdev->downscript) { | |
2676 | err = run_script(handler->name, "net", netdev->downscript, | |
2677 | "down", "empty", (char*) NULL); | |
2678 | if (err) | |
2679 | return -1; | |
2680 | } | |
2681 | return 0; | |
2682 | } | |
2683 | ||
e3b4c4c4 | 2684 | int lxc_create_network(struct lxc_handler *handler) |
0ad19a3f | 2685 | { |
e3b4c4c4 | 2686 | struct lxc_list *network = &handler->conf->network; |
82d5ae15 | 2687 | struct lxc_list *iterator; |
82d5ae15 | 2688 | struct lxc_netdev *netdev; |
0ad19a3f | 2689 | |
5f4535a3 | 2690 | lxc_list_for_each(iterator, network) { |
0ad19a3f | 2691 | |
5f4535a3 | 2692 | netdev = iterator->elem; |
13954cce | 2693 | |
24654103 | 2694 | if (netdev->type < 0 || netdev->type > LXC_NET_MAXCONFTYPE) { |
82d5ae15 | 2695 | ERROR("invalid network configuration type '%d'", |
5f4535a3 | 2696 | netdev->type); |
82d5ae15 DL |
2697 | return -1; |
2698 | } | |
0ad19a3f | 2699 | |
e3b4c4c4 | 2700 | if (netdev_conf[netdev->type](handler, netdev)) { |
82d5ae15 DL |
2701 | ERROR("failed to create netdev"); |
2702 | return -1; | |
2703 | } | |
e3b4c4c4 | 2704 | |
0ad19a3f | 2705 | } |
2706 | ||
2707 | return 0; | |
2708 | } | |
2709 | ||
74a2b586 | 2710 | void lxc_delete_network(struct lxc_handler *handler) |
7fef7a06 | 2711 | { |
74a2b586 | 2712 | struct lxc_list *network = &handler->conf->network; |
7fef7a06 DL |
2713 | struct lxc_list *iterator; |
2714 | struct lxc_netdev *netdev; | |
2715 | ||
2716 | lxc_list_for_each(iterator, network) { | |
2717 | netdev = iterator->elem; | |
d472214b | 2718 | |
74a2b586 | 2719 | if (netdev->ifindex != 0 && netdev->type == LXC_NET_PHYS) { |
d8f8e352 DL |
2720 | if (lxc_netdev_rename_by_index(netdev->ifindex, netdev->link)) |
2721 | WARN("failed to rename to the initial name the " \ | |
2722 | "netdev '%s'", netdev->link); | |
d472214b | 2723 | continue; |
d8f8e352 | 2724 | } |
d472214b | 2725 | |
74a2b586 JK |
2726 | if (netdev_deconf[netdev->type](handler, netdev)) { |
2727 | WARN("failed to destroy netdev"); | |
2728 | } | |
2729 | ||
d8f8e352 DL |
2730 | /* Recent kernel remove the virtual interfaces when the network |
2731 | * namespace is destroyed but in case we did not moved the | |
2732 | * interface to the network namespace, we have to destroy it | |
2733 | */ | |
74a2b586 JK |
2734 | if (netdev->ifindex != 0 && |
2735 | lxc_netdev_delete_by_index(netdev->ifindex)) | |
d8f8e352 | 2736 | WARN("failed to remove interface '%s'", netdev->name); |
7fef7a06 DL |
2737 | } |
2738 | } | |
2739 | ||
5f4535a3 | 2740 | int lxc_assign_network(struct lxc_list *network, pid_t pid) |
0ad19a3f | 2741 | { |
82d5ae15 | 2742 | struct lxc_list *iterator; |
82d5ae15 | 2743 | struct lxc_netdev *netdev; |
3cfc0f3a | 2744 | int err; |
0ad19a3f | 2745 | |
5f4535a3 | 2746 | lxc_list_for_each(iterator, network) { |
82d5ae15 | 2747 | |
5f4535a3 | 2748 | netdev = iterator->elem; |
82d5ae15 | 2749 | |
236087a6 DL |
2750 | /* empty network namespace, nothing to move */ |
2751 | if (!netdev->ifindex) | |
2752 | continue; | |
2753 | ||
d472214b | 2754 | err = lxc_netdev_move_by_index(netdev->ifindex, pid); |
3cfc0f3a MN |
2755 | if (err) { |
2756 | ERROR("failed to move '%s' to the container : %s", | |
2757 | netdev->link, strerror(-err)); | |
82d5ae15 DL |
2758 | return -1; |
2759 | } | |
2760 | ||
c1c75c04 | 2761 | DEBUG("move '%s' to '%d'", netdev->name, pid); |
0ad19a3f | 2762 | } |
2763 | ||
2764 | return 0; | |
2765 | } | |
2766 | ||
251d0d2a DE |
2767 | static int write_id_mapping(enum idtype idtype, pid_t pid, const char *buf, |
2768 | size_t buf_size) | |
f6d3e3e4 SH |
2769 | { |
2770 | char path[PATH_MAX]; | |
e4ccd113 | 2771 | int ret, closeret; |
f6d3e3e4 SH |
2772 | FILE *f; |
2773 | ||
2774 | ret = snprintf(path, PATH_MAX, "/proc/%d/%cid_map", pid, idtype == ID_TYPE_UID ? 'u' : 'g'); | |
2775 | if (ret < 0 || ret >= PATH_MAX) { | |
2776 | fprintf(stderr, "%s: path name too long", __func__); | |
2777 | return -E2BIG; | |
2778 | } | |
025ed0f3 | 2779 | process_lock(); |
f6d3e3e4 | 2780 | f = fopen(path, "w"); |
025ed0f3 | 2781 | process_unlock(); |
f6d3e3e4 SH |
2782 | if (!f) { |
2783 | perror("open"); | |
2784 | return -EINVAL; | |
2785 | } | |
251d0d2a | 2786 | ret = fwrite(buf, buf_size, 1, f); |
f6d3e3e4 | 2787 | if (ret < 0) |
e4ccd113 | 2788 | SYSERROR("writing id mapping"); |
025ed0f3 | 2789 | process_lock(); |
e4ccd113 | 2790 | closeret = fclose(f); |
025ed0f3 | 2791 | process_unlock(); |
e4ccd113 SH |
2792 | if (closeret) |
2793 | SYSERROR("writing id mapping"); | |
2794 | return ret < 0 ? ret : closeret; | |
f6d3e3e4 SH |
2795 | } |
2796 | ||
2797 | int lxc_map_ids(struct lxc_list *idmap, pid_t pid) | |
2798 | { | |
2799 | struct lxc_list *iterator; | |
2800 | struct id_map *map; | |
2801 | int ret = 0; | |
251d0d2a | 2802 | enum idtype type; |
4f7521b4 | 2803 | char *buf = NULL, *pos; |
251d0d2a DE |
2804 | |
2805 | for(type = ID_TYPE_UID; type <= ID_TYPE_GID; type++) { | |
4f7521b4 SH |
2806 | int left, fill; |
2807 | ||
2808 | pos = buf; | |
251d0d2a | 2809 | lxc_list_for_each(iterator, idmap) { |
4f7521b4 SH |
2810 | /* The kernel only takes <= 4k for writes to /proc/<nr>/[ug]id_map */ |
2811 | if (!buf) | |
2812 | buf = pos = malloc(4096); | |
2813 | if (!buf) | |
2814 | return -ENOMEM; | |
2815 | ||
251d0d2a DE |
2816 | map = iterator->elem; |
2817 | if (map->idtype == type) { | |
2818 | left = 4096 - (pos - buf); | |
2819 | fill = snprintf(pos, left, "%lu %lu %lu\n", | |
2820 | map->nsid, map->hostid, map->range); | |
2821 | if (fill <= 0 || fill >= left) | |
2822 | SYSERROR("snprintf failed, too many mappings"); | |
2823 | pos += fill; | |
2824 | } | |
2825 | } | |
4f7521b4 SH |
2826 | if (pos == buf) // no mappings were found |
2827 | continue; | |
251d0d2a | 2828 | ret = write_id_mapping(type, pid, buf, pos-buf); |
f6d3e3e4 SH |
2829 | if (ret) |
2830 | break; | |
2831 | } | |
251d0d2a | 2832 | |
4f7521b4 SH |
2833 | if (buf) |
2834 | free(buf); | |
f6d3e3e4 SH |
2835 | return ret; |
2836 | } | |
2837 | ||
19a26f82 MK |
2838 | int lxc_find_gateway_addresses(struct lxc_handler *handler) |
2839 | { | |
2840 | struct lxc_list *network = &handler->conf->network; | |
2841 | struct lxc_list *iterator; | |
2842 | struct lxc_netdev *netdev; | |
2843 | int link_index; | |
2844 | ||
2845 | lxc_list_for_each(iterator, network) { | |
2846 | netdev = iterator->elem; | |
2847 | ||
2848 | if (!netdev->ipv4_gateway_auto && !netdev->ipv6_gateway_auto) | |
2849 | continue; | |
2850 | ||
2851 | if (netdev->type != LXC_NET_VETH && netdev->type != LXC_NET_MACVLAN) { | |
2852 | ERROR("gateway = auto only supported for " | |
2853 | "veth and macvlan"); | |
2854 | return -1; | |
2855 | } | |
2856 | ||
2857 | if (!netdev->link) { | |
2858 | ERROR("gateway = auto needs a link interface"); | |
2859 | return -1; | |
2860 | } | |
2861 | ||
2862 | link_index = if_nametoindex(netdev->link); | |
2863 | if (!link_index) | |
2864 | return -EINVAL; | |
2865 | ||
2866 | if (netdev->ipv4_gateway_auto) { | |
2867 | if (lxc_ipv4_addr_get(link_index, &netdev->ipv4_gateway)) { | |
2868 | ERROR("failed to automatically find ipv4 gateway " | |
2869 | "address from link interface '%s'", netdev->link); | |
2870 | return -1; | |
2871 | } | |
2872 | } | |
2873 | ||
2874 | if (netdev->ipv6_gateway_auto) { | |
2875 | if (lxc_ipv6_addr_get(link_index, &netdev->ipv6_gateway)) { | |
2876 | ERROR("failed to automatically find ipv6 gateway " | |
2877 | "address from link interface '%s'", netdev->link); | |
2878 | return -1; | |
2879 | } | |
2880 | } | |
2881 | } | |
2882 | ||
2883 | return 0; | |
2884 | } | |
2885 | ||
5e4a62bf | 2886 | int lxc_create_tty(const char *name, struct lxc_conf *conf) |
b0a33c1e | 2887 | { |
5e4a62bf | 2888 | struct lxc_tty_info *tty_info = &conf->tty_info; |
025ed0f3 | 2889 | int i, ret; |
b0a33c1e | 2890 | |
5e4a62bf DL |
2891 | /* no tty in the configuration */ |
2892 | if (!conf->tty) | |
b0a33c1e | 2893 | return 0; |
2894 | ||
13954cce | 2895 | tty_info->pty_info = |
e4e7d59d | 2896 | malloc(sizeof(*tty_info->pty_info)*conf->tty); |
b0a33c1e | 2897 | if (!tty_info->pty_info) { |
36eb9bde | 2898 | SYSERROR("failed to allocate pty_info"); |
985d15b1 | 2899 | return -1; |
b0a33c1e | 2900 | } |
2901 | ||
985d15b1 | 2902 | for (i = 0; i < conf->tty; i++) { |
13954cce | 2903 | |
b0a33c1e | 2904 | struct lxc_pty_info *pty_info = &tty_info->pty_info[i]; |
2905 | ||
025ed0f3 SH |
2906 | process_lock(); |
2907 | ret = openpty(&pty_info->master, &pty_info->slave, | |
2908 | pty_info->name, NULL, NULL); | |
2909 | process_unlock(); | |
2910 | if (ret) { | |
36eb9bde | 2911 | SYSERROR("failed to create pty #%d", i); |
985d15b1 MT |
2912 | tty_info->nbtty = i; |
2913 | lxc_delete_tty(tty_info); | |
2914 | return -1; | |
b0a33c1e | 2915 | } |
2916 | ||
5332bb84 DL |
2917 | DEBUG("allocated pty '%s' (%d/%d)", |
2918 | pty_info->name, pty_info->master, pty_info->slave); | |
2919 | ||
b035ad62 MS |
2920 | /* Prevent leaking the file descriptors to the container */ |
2921 | fcntl(pty_info->master, F_SETFD, FD_CLOEXEC); | |
2922 | fcntl(pty_info->slave, F_SETFD, FD_CLOEXEC); | |
2923 | ||
b0a33c1e | 2924 | pty_info->busy = 0; |
2925 | } | |
2926 | ||
985d15b1 | 2927 | tty_info->nbtty = conf->tty; |
1ac470c0 DL |
2928 | |
2929 | INFO("tty's configured"); | |
2930 | ||
985d15b1 | 2931 | return 0; |
b0a33c1e | 2932 | } |
2933 | ||
2934 | void lxc_delete_tty(struct lxc_tty_info *tty_info) | |
2935 | { | |
2936 | int i; | |
2937 | ||
2938 | for (i = 0; i < tty_info->nbtty; i++) { | |
2939 | struct lxc_pty_info *pty_info = &tty_info->pty_info[i]; | |
2940 | ||
025ed0f3 | 2941 | process_lock(); |
b0a33c1e | 2942 | close(pty_info->master); |
2943 | close(pty_info->slave); | |
025ed0f3 | 2944 | process_unlock(); |
b0a33c1e | 2945 | } |
2946 | ||
2947 | free(tty_info->pty_info); | |
2948 | tty_info->nbtty = 0; | |
2949 | } | |
2950 | ||
f6d3e3e4 SH |
2951 | /* |
2952 | * given a host uid, return the ns uid if it is mapped. | |
2953 | * if it is not mapped, return the original host id. | |
2954 | */ | |
2955 | static int shiftid(struct lxc_conf *c, int uid, enum idtype w) | |
2956 | { | |
2957 | struct lxc_list *iterator; | |
2958 | struct id_map *map; | |
2959 | int low, high; | |
2960 | ||
2961 | lxc_list_for_each(iterator, &c->id_map) { | |
2962 | map = iterator->elem; | |
2963 | if (map->idtype != w) | |
2964 | continue; | |
2965 | ||
2966 | low = map->nsid; | |
2967 | high = map->nsid + map->range; | |
2968 | if (uid < low || uid >= high) | |
2969 | continue; | |
2970 | ||
2971 | return uid - low + map->hostid; | |
2972 | } | |
2973 | ||
2974 | return uid; | |
2975 | } | |
2976 | ||
2977 | /* | |
2978 | * Take a pathname for a file created on the host, and map the uid and gid | |
2979 | * into the container if needed. (Used for ttys) | |
2980 | */ | |
2981 | static int uid_shift_file(char *path, struct lxc_conf *c) | |
2982 | { | |
2983 | struct stat statbuf; | |
2984 | int newuid, newgid; | |
2985 | ||
2986 | if (stat(path, &statbuf)) { | |
2987 | SYSERROR("stat(%s)", path); | |
2988 | return -1; | |
2989 | } | |
2990 | ||
2991 | newuid = shiftid(c, statbuf.st_uid, ID_TYPE_UID); | |
2992 | newgid = shiftid(c, statbuf.st_gid, ID_TYPE_GID); | |
2993 | if (newuid != statbuf.st_uid || newgid != statbuf.st_gid) { | |
20087962 | 2994 | DEBUG("chowning %s from %d:%d to %d:%d\n", path, (int)statbuf.st_uid, (int)statbuf.st_gid, newuid, newgid); |
f6d3e3e4 SH |
2995 | if (chown(path, newuid, newgid)) { |
2996 | SYSERROR("chown(%s)", path); | |
2997 | return -1; | |
2998 | } | |
2999 | } | |
3000 | return 0; | |
3001 | } | |
3002 | ||
3003 | int uid_shift_ttys(int pid, struct lxc_conf *conf) | |
3004 | { | |
3005 | int i, ret; | |
3006 | struct lxc_tty_info *tty_info = &conf->tty_info; | |
3007 | char path[MAXPATHLEN]; | |
3008 | char *ttydir = conf->ttydir; | |
3009 | ||
3010 | if (!conf->rootfs.path) | |
3011 | return 0; | |
3012 | /* first the console */ | |
3013 | ret = snprintf(path, sizeof(path), "/proc/%d/root/dev/%s/console", pid, ttydir ? ttydir : ""); | |
3014 | if (ret < 0 || ret >= sizeof(path)) { | |
3015 | ERROR("console path too long\n"); | |
3016 | return -1; | |
3017 | } | |
3018 | if (uid_shift_file(path, conf)) { | |
3019 | DEBUG("Failed to chown the console %s.\n", path); | |
3020 | return -1; | |
3021 | } | |
3022 | for (i=0; i< tty_info->nbtty; i++) { | |
3023 | ret = snprintf(path, sizeof(path), "/proc/%d/root/dev/%s/tty%d", | |
3024 | pid, ttydir ? ttydir : "", i + 1); | |
3025 | if (ret < 0 || ret >= sizeof(path)) { | |
3026 | ERROR("pathname too long for ttys"); | |
3027 | return -1; | |
3028 | } | |
3029 | if (uid_shift_file(path, conf)) { | |
3030 | DEBUG("Failed to chown pty %s.\n", path); | |
3031 | return -1; | |
3032 | } | |
3033 | } | |
3034 | ||
3035 | return 0; | |
3036 | } | |
3037 | ||
368bbc02 | 3038 | int lxc_setup(const char *name, struct lxc_conf *lxc_conf, const char *lxcpath, struct cgroup_process_info *cgroup_info) |
0ad19a3f | 3039 | { |
571e6ec8 | 3040 | if (setup_utsname(lxc_conf->utsname)) { |
36eb9bde | 3041 | ERROR("failed to setup the utsname for '%s'", name); |
95b5ffaf | 3042 | return -1; |
0ad19a3f | 3043 | } |
3044 | ||
5f4535a3 | 3045 | if (setup_network(&lxc_conf->network)) { |
36eb9bde | 3046 | ERROR("failed to setup the network for '%s'", name); |
95b5ffaf | 3047 | return -1; |
0ad19a3f | 3048 | } |
3049 | ||
283678ed | 3050 | if (run_lxc_hooks(name, "pre-mount", lxc_conf, lxcpath, NULL)) { |
89eaa05e SH |
3051 | ERROR("failed to run pre-mount hooks for container '%s'.", name); |
3052 | return -1; | |
3053 | } | |
5ea6163a | 3054 | |
cc28d0b0 | 3055 | if (setup_rootfs(lxc_conf)) { |
ac778708 | 3056 | ERROR("failed to setup rootfs for '%s'", name); |
95b5ffaf | 3057 | return -1; |
0ad19a3f | 3058 | } |
3059 | ||
c6883f38 | 3060 | if (lxc_conf->autodev) { |
91c3830e SH |
3061 | if (mount_autodev(lxc_conf->rootfs.mount)) { |
3062 | ERROR("failed to mount /dev in the container"); | |
c6883f38 SH |
3063 | return -1; |
3064 | } | |
3065 | } | |
3066 | ||
368bbc02 CS |
3067 | /* do automatic mounts (mainly /proc and /sys), but exclude |
3068 | * those that need to wait until other stuff has finished | |
3069 | */ | |
3070 | if (lxc_mount_auto_mounts(lxc_conf, lxc_conf->auto_mounts & ~LXC_AUTO_CGROUP & ~LXC_AUTO_PROC_SYSRQ, cgroup_info) < 0) { | |
3071 | ERROR("failed to setup the automatic mounts for '%s'", name); | |
3072 | return -1; | |
3073 | } | |
3074 | ||
80a881b2 | 3075 | if (setup_mount(&lxc_conf->rootfs, lxc_conf->fstab, name)) { |
36eb9bde | 3076 | ERROR("failed to setup the mounts for '%s'", name); |
95b5ffaf | 3077 | return -1; |
576f946d | 3078 | } |
3079 | ||
c1dc38c2 | 3080 | if (!lxc_list_empty(&lxc_conf->mount_list) && setup_mount_entries(&lxc_conf->rootfs, &lxc_conf->mount_list, name)) { |
e7938e9e MN |
3081 | ERROR("failed to setup the mount entries for '%s'", name); |
3082 | return -1; | |
3083 | } | |
3084 | ||
368bbc02 CS |
3085 | /* now mount only cgroup, if wanted; |
3086 | * before, /sys could not have been mounted | |
3087 | * (is either mounted automatically or via fstab entries) | |
3088 | */ | |
3089 | if (lxc_mount_auto_mounts(lxc_conf, lxc_conf->auto_mounts & LXC_AUTO_CGROUP, cgroup_info) < 0) { | |
3090 | ERROR("failed to setup the automatic mounts for '%s'", name); | |
3091 | return -1; | |
3092 | } | |
3093 | ||
283678ed | 3094 | if (run_lxc_hooks(name, "mount", lxc_conf, lxcpath, NULL)) { |
773fb9ca SH |
3095 | ERROR("failed to run mount hooks for container '%s'.", name); |
3096 | return -1; | |
3097 | } | |
3098 | ||
91c3830e | 3099 | if (lxc_conf->autodev) { |
283678ed | 3100 | if (run_lxc_hooks(name, "autodev", lxc_conf, lxcpath, NULL)) { |
f7bee6c6 MW |
3101 | ERROR("failed to run autodev hooks for container '%s'.", name); |
3102 | return -1; | |
3103 | } | |
91c3830e SH |
3104 | if (setup_autodev(lxc_conf->rootfs.mount)) { |
3105 | ERROR("failed to populate /dev in the container"); | |
3106 | return -1; | |
3107 | } | |
3108 | } | |
368bbc02 CS |
3109 | |
3110 | /* over-mount /proc/sysrq-trigger with /dev/null now, if wanted; | |
3111 | * before /dev/null did not necessarily exist | |
3112 | */ | |
3113 | if (lxc_mount_auto_mounts(lxc_conf, lxc_conf->auto_mounts & LXC_AUTO_PROC_SYSRQ, cgroup_info) < 0) { | |
3114 | ERROR("failed to setup the automatic mounts for '%s'", name); | |
3115 | return -1; | |
3116 | } | |
91c3830e | 3117 | |
37903589 | 3118 | if (!lxc_conf->is_execute && setup_console(&lxc_conf->rootfs, &lxc_conf->console, lxc_conf->ttydir)) { |
36eb9bde | 3119 | ERROR("failed to setup the console for '%s'", name); |
95b5ffaf | 3120 | return -1; |
6e590161 | 3121 | } |
3122 | ||
7e0e1d94 AV |
3123 | if (lxc_conf->kmsg) { |
3124 | if (setup_kmsg(&lxc_conf->rootfs, &lxc_conf->console)) // don't fail | |
3125 | ERROR("failed to setup kmsg for '%s'", name); | |
3126 | } | |
1bd051a6 | 3127 | |
37903589 | 3128 | if (!lxc_conf->is_execute && setup_tty(&lxc_conf->rootfs, &lxc_conf->tty_info, lxc_conf->ttydir)) { |
36eb9bde | 3129 | ERROR("failed to setup the ttys for '%s'", name); |
95b5ffaf | 3130 | return -1; |
b0a33c1e | 3131 | } |
3132 | ||
fe4de9a6 DE |
3133 | /* mount /proc if needed for LSM transition */ |
3134 | if (lsm_proc_mount(lxc_conf) < 0) { | |
3135 | ERROR("failed to LSM mount proc for '%s'", name); | |
e075f5d9 | 3136 | return -1; |
e075f5d9 | 3137 | } |
e075f5d9 | 3138 | |
ac778708 | 3139 | if (setup_pivot_root(&lxc_conf->rootfs)) { |
36eb9bde | 3140 | ERROR("failed to set rootfs for '%s'", name); |
95b5ffaf | 3141 | return -1; |
ed502555 | 3142 | } |
3143 | ||
571e6ec8 | 3144 | if (setup_pts(lxc_conf->pts)) { |
36eb9bde | 3145 | ERROR("failed to setup the new pts instance"); |
95b5ffaf | 3146 | return -1; |
3c26f34e | 3147 | } |
3148 | ||
cccc74b5 DL |
3149 | if (setup_personality(lxc_conf->personality)) { |
3150 | ERROR("failed to setup personality"); | |
3151 | return -1; | |
3152 | } | |
3153 | ||
f6d3e3e4 | 3154 | if (lxc_list_empty(&lxc_conf->id_map)) { |
1fb86a7c SH |
3155 | if (!lxc_list_empty(&lxc_conf->keepcaps)) { |
3156 | if (!lxc_list_empty(&lxc_conf->caps)) { | |
3157 | ERROR("Simultaneously requested dropping and keeping caps"); | |
3158 | return -1; | |
3159 | } | |
3160 | if (dropcaps_except(&lxc_conf->keepcaps)) { | |
3161 | ERROR("failed to keep requested caps\n"); | |
3162 | return -1; | |
3163 | } | |
3164 | } else if (setup_caps(&lxc_conf->caps)) { | |
f6d3e3e4 SH |
3165 | ERROR("failed to drop capabilities"); |
3166 | return -1; | |
3167 | } | |
81810dd1 DL |
3168 | } |
3169 | ||
cd54d859 DL |
3170 | NOTICE("'%s' is setup.", name); |
3171 | ||
0ad19a3f | 3172 | return 0; |
3173 | } | |
26ddeedd | 3174 | |
283678ed SH |
3175 | int run_lxc_hooks(const char *name, char *hook, struct lxc_conf *conf, |
3176 | const char *lxcpath, char *argv[]) | |
26ddeedd SH |
3177 | { |
3178 | int which = -1; | |
3179 | struct lxc_list *it; | |
3180 | ||
3181 | if (strcmp(hook, "pre-start") == 0) | |
3182 | which = LXCHOOK_PRESTART; | |
5ea6163a SH |
3183 | else if (strcmp(hook, "pre-mount") == 0) |
3184 | which = LXCHOOK_PREMOUNT; | |
26ddeedd SH |
3185 | else if (strcmp(hook, "mount") == 0) |
3186 | which = LXCHOOK_MOUNT; | |
f7bee6c6 MW |
3187 | else if (strcmp(hook, "autodev") == 0) |
3188 | which = LXCHOOK_AUTODEV; | |
26ddeedd SH |
3189 | else if (strcmp(hook, "start") == 0) |
3190 | which = LXCHOOK_START; | |
3191 | else if (strcmp(hook, "post-stop") == 0) | |
3192 | which = LXCHOOK_POSTSTOP; | |
148e91f5 SH |
3193 | else if (strcmp(hook, "clone") == 0) |
3194 | which = LXCHOOK_CLONE; | |
26ddeedd SH |
3195 | else |
3196 | return -1; | |
3197 | lxc_list_for_each(it, &conf->hooks[which]) { | |
3198 | int ret; | |
3199 | char *hookname = it->elem; | |
283678ed | 3200 | ret = run_script_argv(name, "lxc", hookname, hook, lxcpath, argv); |
26ddeedd SH |
3201 | if (ret) |
3202 | return ret; | |
3203 | } | |
3204 | return 0; | |
3205 | } | |
72d0e1cb | 3206 | |
427b3a21 | 3207 | static void lxc_remove_nic(struct lxc_list *it) |
72d0e1cb SG |
3208 | { |
3209 | struct lxc_netdev *netdev = it->elem; | |
9ebb03ad | 3210 | struct lxc_list *it2,*next; |
72d0e1cb SG |
3211 | |
3212 | lxc_list_del(it); | |
3213 | ||
3214 | if (netdev->link) | |
3215 | free(netdev->link); | |
3216 | if (netdev->name) | |
3217 | free(netdev->name); | |
3218 | if (netdev->upscript) | |
3219 | free(netdev->upscript); | |
3220 | if (netdev->hwaddr) | |
3221 | free(netdev->hwaddr); | |
3222 | if (netdev->mtu) | |
3223 | free(netdev->mtu); | |
3224 | if (netdev->ipv4_gateway) | |
3225 | free(netdev->ipv4_gateway); | |
3226 | if (netdev->ipv6_gateway) | |
3227 | free(netdev->ipv6_gateway); | |
9ebb03ad | 3228 | lxc_list_for_each_safe(it2, &netdev->ipv4, next) { |
72d0e1cb SG |
3229 | lxc_list_del(it2); |
3230 | free(it2->elem); | |
3231 | free(it2); | |
3232 | } | |
9ebb03ad | 3233 | lxc_list_for_each_safe(it2, &netdev->ipv6, next) { |
72d0e1cb SG |
3234 | lxc_list_del(it2); |
3235 | free(it2->elem); | |
3236 | free(it2); | |
3237 | } | |
d95db067 | 3238 | free(netdev); |
72d0e1cb SG |
3239 | free(it); |
3240 | } | |
3241 | ||
3242 | /* we get passed in something like '0', '0.ipv4' or '1.ipv6' */ | |
12a50cc6 | 3243 | int lxc_clear_nic(struct lxc_conf *c, const char *key) |
72d0e1cb SG |
3244 | { |
3245 | char *p1; | |
3246 | int ret, idx, i; | |
3247 | struct lxc_list *it; | |
3248 | struct lxc_netdev *netdev; | |
3249 | ||
3250 | p1 = index(key, '.'); | |
3251 | if (!p1 || *(p1+1) == '\0') | |
3252 | p1 = NULL; | |
3253 | ||
3254 | ret = sscanf(key, "%d", &idx); | |
3255 | if (ret != 1) return -1; | |
3256 | if (idx < 0) | |
3257 | return -1; | |
3258 | ||
3259 | i = 0; | |
3260 | lxc_list_for_each(it, &c->network) { | |
3261 | if (i == idx) | |
3262 | break; | |
3263 | i++; | |
3264 | } | |
3265 | if (i < idx) // we don't have that many nics defined | |
3266 | return -1; | |
3267 | ||
3268 | if (!it || !it->elem) | |
3269 | return -1; | |
3270 | ||
3271 | netdev = it->elem; | |
3272 | ||
3273 | if (!p1) { | |
3274 | lxc_remove_nic(it); | |
3275 | } else if (strcmp(p1, "ipv4") == 0) { | |
9ebb03ad DE |
3276 | struct lxc_list *it2,*next; |
3277 | lxc_list_for_each_safe(it2, &netdev->ipv4, next) { | |
72d0e1cb SG |
3278 | lxc_list_del(it2); |
3279 | free(it2->elem); | |
3280 | free(it2); | |
3281 | } | |
3282 | } else if (strcmp(p1, "ipv6") == 0) { | |
9ebb03ad DE |
3283 | struct lxc_list *it2,*next; |
3284 | lxc_list_for_each_safe(it2, &netdev->ipv6, next) { | |
72d0e1cb SG |
3285 | lxc_list_del(it2); |
3286 | free(it2->elem); | |
3287 | free(it2); | |
3288 | } | |
3289 | } else if (strcmp(p1, "link") == 0) { | |
3290 | if (netdev->link) { | |
3291 | free(netdev->link); | |
3292 | netdev->link = NULL; | |
3293 | } | |
3294 | } else if (strcmp(p1, "name") == 0) { | |
3295 | if (netdev->name) { | |
3296 | free(netdev->name); | |
3297 | netdev->name = NULL; | |
3298 | } | |
3299 | } else if (strcmp(p1, "script.up") == 0) { | |
3300 | if (netdev->upscript) { | |
3301 | free(netdev->upscript); | |
3302 | netdev->upscript = NULL; | |
3303 | } | |
3304 | } else if (strcmp(p1, "hwaddr") == 0) { | |
3305 | if (netdev->hwaddr) { | |
3306 | free(netdev->hwaddr); | |
3307 | netdev->hwaddr = NULL; | |
3308 | } | |
3309 | } else if (strcmp(p1, "mtu") == 0) { | |
3310 | if (netdev->mtu) { | |
3311 | free(netdev->mtu); | |
3312 | netdev->mtu = NULL; | |
3313 | } | |
3314 | } else if (strcmp(p1, "ipv4_gateway") == 0) { | |
3315 | if (netdev->ipv4_gateway) { | |
3316 | free(netdev->ipv4_gateway); | |
3317 | netdev->ipv4_gateway = NULL; | |
3318 | } | |
3319 | } else if (strcmp(p1, "ipv6_gateway") == 0) { | |
3320 | if (netdev->ipv6_gateway) { | |
3321 | free(netdev->ipv6_gateway); | |
3322 | netdev->ipv6_gateway = NULL; | |
3323 | } | |
3324 | } | |
3325 | else return -1; | |
3326 | ||
3327 | return 0; | |
3328 | } | |
3329 | ||
3330 | int lxc_clear_config_network(struct lxc_conf *c) | |
3331 | { | |
9ebb03ad DE |
3332 | struct lxc_list *it,*next; |
3333 | lxc_list_for_each_safe(it, &c->network, next) { | |
72d0e1cb SG |
3334 | lxc_remove_nic(it); |
3335 | } | |
3336 | return 0; | |
3337 | } | |
3338 | ||
3339 | int lxc_clear_config_caps(struct lxc_conf *c) | |
3340 | { | |
9ebb03ad | 3341 | struct lxc_list *it,*next; |
72d0e1cb | 3342 | |
9ebb03ad | 3343 | lxc_list_for_each_safe(it, &c->caps, next) { |
72d0e1cb SG |
3344 | lxc_list_del(it); |
3345 | free(it->elem); | |
3346 | free(it); | |
3347 | } | |
3348 | return 0; | |
3349 | } | |
3350 | ||
27c27d73 SH |
3351 | int lxc_clear_idmaps(struct lxc_conf *c) |
3352 | { | |
3353 | struct lxc_list *it, *next; | |
3354 | ||
3355 | lxc_list_for_each_safe(it, &c->id_map, next) { | |
3356 | lxc_list_del(it); | |
3357 | free(it->elem); | |
3358 | free(it); | |
3359 | } | |
3360 | return 0; | |
3361 | } | |
3362 | ||
1fb86a7c SH |
3363 | int lxc_clear_config_keepcaps(struct lxc_conf *c) |
3364 | { | |
3365 | struct lxc_list *it,*next; | |
3366 | ||
3367 | lxc_list_for_each_safe(it, &c->keepcaps, next) { | |
3368 | lxc_list_del(it); | |
3369 | free(it->elem); | |
3370 | free(it); | |
3371 | } | |
3372 | return 0; | |
3373 | } | |
3374 | ||
12a50cc6 | 3375 | int lxc_clear_cgroups(struct lxc_conf *c, const char *key) |
72d0e1cb | 3376 | { |
9ebb03ad | 3377 | struct lxc_list *it,*next; |
72d0e1cb | 3378 | bool all = false; |
12a50cc6 | 3379 | const char *k = key + 11; |
72d0e1cb SG |
3380 | |
3381 | if (strcmp(key, "lxc.cgroup") == 0) | |
3382 | all = true; | |
3383 | ||
9ebb03ad | 3384 | lxc_list_for_each_safe(it, &c->cgroup, next) { |
72d0e1cb SG |
3385 | struct lxc_cgroup *cg = it->elem; |
3386 | if (!all && strcmp(cg->subsystem, k) != 0) | |
3387 | continue; | |
3388 | lxc_list_del(it); | |
3389 | free(cg->subsystem); | |
3390 | free(cg->value); | |
3391 | free(cg); | |
3392 | free(it); | |
3393 | } | |
3394 | return 0; | |
3395 | } | |
3396 | ||
3397 | int lxc_clear_mount_entries(struct lxc_conf *c) | |
3398 | { | |
9ebb03ad | 3399 | struct lxc_list *it,*next; |
72d0e1cb | 3400 | |
9ebb03ad | 3401 | lxc_list_for_each_safe(it, &c->mount_list, next) { |
72d0e1cb SG |
3402 | lxc_list_del(it); |
3403 | free(it->elem); | |
3404 | free(it); | |
3405 | } | |
3406 | return 0; | |
3407 | } | |
3408 | ||
12a50cc6 | 3409 | int lxc_clear_hooks(struct lxc_conf *c, const char *key) |
72d0e1cb | 3410 | { |
9ebb03ad | 3411 | struct lxc_list *it,*next; |
17ed13a3 | 3412 | bool all = false, done = false; |
12a50cc6 | 3413 | const char *k = key + 9; |
72d0e1cb SG |
3414 | int i; |
3415 | ||
17ed13a3 SH |
3416 | if (strcmp(key, "lxc.hook") == 0) |
3417 | all = true; | |
3418 | ||
72d0e1cb | 3419 | for (i=0; i<NUM_LXC_HOOKS; i++) { |
17ed13a3 | 3420 | if (all || strcmp(k, lxchook_names[i]) == 0) { |
9ebb03ad | 3421 | lxc_list_for_each_safe(it, &c->hooks[i], next) { |
17ed13a3 SH |
3422 | lxc_list_del(it); |
3423 | free(it->elem); | |
3424 | free(it); | |
3425 | } | |
3426 | done = true; | |
72d0e1cb SG |
3427 | } |
3428 | } | |
17ed13a3 SH |
3429 | |
3430 | if (!done) { | |
3431 | ERROR("Invalid hook key: %s", key); | |
3432 | return -1; | |
3433 | } | |
72d0e1cb SG |
3434 | return 0; |
3435 | } | |
8eb5694b | 3436 | |
7b35f3d6 SH |
3437 | void lxc_clear_saved_nics(struct lxc_conf *conf) |
3438 | { | |
3439 | int i; | |
3440 | ||
3441 | if (!conf->num_savednics) | |
3442 | return; | |
3443 | for (i=0; i < conf->num_savednics; i++) | |
3444 | free(conf->saved_nics[i].orig_name); | |
3445 | conf->saved_nics = 0; | |
3446 | free(conf->saved_nics); | |
3447 | } | |
3448 | ||
8eb5694b SH |
3449 | void lxc_conf_free(struct lxc_conf *conf) |
3450 | { | |
3451 | if (!conf) | |
3452 | return; | |
3453 | if (conf->console.path) | |
3454 | free(conf->console.path); | |
54c30e29 | 3455 | if (conf->rootfs.mount) |
8eb5694b | 3456 | free(conf->rootfs.mount); |
d95db067 DE |
3457 | if (conf->rootfs.path) |
3458 | free(conf->rootfs.path); | |
3459 | if (conf->utsname) | |
3460 | free(conf->utsname); | |
3461 | if (conf->ttydir) | |
3462 | free(conf->ttydir); | |
3463 | if (conf->fstab) | |
3464 | free(conf->fstab); | |
fc7e8864 WM |
3465 | if (conf->rcfile) |
3466 | free(conf->rcfile); | |
8eb5694b | 3467 | lxc_clear_config_network(conf); |
fe4de9a6 DE |
3468 | if (conf->lsm_aa_profile) |
3469 | free(conf->lsm_aa_profile); | |
3470 | if (conf->lsm_se_context) | |
3471 | free(conf->lsm_se_context); | |
769872f9 | 3472 | lxc_seccomp_free(conf); |
8eb5694b | 3473 | lxc_clear_config_caps(conf); |
1fb86a7c | 3474 | lxc_clear_config_keepcaps(conf); |
8eb5694b | 3475 | lxc_clear_cgroups(conf, "lxc.cgroup"); |
17ed13a3 | 3476 | lxc_clear_hooks(conf, "lxc.hook"); |
8eb5694b | 3477 | lxc_clear_mount_entries(conf); |
7b35f3d6 | 3478 | lxc_clear_saved_nics(conf); |
27c27d73 | 3479 | lxc_clear_idmaps(conf); |
8eb5694b SH |
3480 | free(conf); |
3481 | } |