src/lxc/utils.h

   1 /*
   2  * lxc: linux Container library
   3  *
   4  * (C) Copyright IBM Corp. 2007, 2008
   5  *
   6  * Authors:
   7  * Daniel Lezcano <daniel.lezcano at free.fr>
   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
  21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  22  */
  23 #ifndef __LXC_UTILS_H
  24 #define __LXC_UTILS_H
  25
  26 /* Properly support loop devices on 32bit systems. */
  27 #define _FILE_OFFSET_BITS 64
  28
  29 #include <errno.h>
  30 #include <linux/loop.h>
  31 #include <linux/types.h>
  32 #include <stdarg.h>
  33 #include <stdbool.h>
  34 #include <stdio.h>
  35 #include <sys/syscall.h>
  36 #include <sys/types.h>
  37 #include <sys/vfs.h>
  38 #include <unistd.h>
  39
  40 #include "file_utils.h"
  41 #include "initutils.h"
  42 #include "macro.h"
  43 #include "string_utils.h"
  44
  45 #ifdef HAVE_LINUX_MEMFD_H
  46 #include <linux/memfd.h>
  47 #endif
  48
  49 /* returns 1 on success, 0 if there were any failures */
  50 extern int lxc_rmdir_onedev(const char *path, const char *exclude);
  51 extern int get_u16(unsigned short *val, const char *arg, int base);
  52 extern int mkdir_p(const char *dir, mode_t mode);
  53 extern char *get_rundir(void);
  54
  55 /* Define getline() if missing from the C library */
  56 #ifndef HAVE_GETLINE
  57 #ifdef HAVE_FGETLN
  58 #include <../include/getline.h>
  59 #endif
  60 #endif
  61
  62 #if !defined(__NR_setns) && !defined(__NR_set_ns)
  63         #if defined(__x86_64__)
  64                 #define __NR_setns 308
  65         #elif defined(__i386__)
  66                 #define __NR_setns 346
  67         #elif defined(__arm__)
  68                 #define __NR_setns 375
  69         #elif defined(__aarch64__)
  70                 #define __NR_setns 375
  71         #elif defined(__powerpc__)
  72                 #define __NR_setns 350
  73         #elif defined(__s390__)
  74                 #define __NR_setns 339
  75         #endif
  76 #endif
  77
  78 /* Define setns() if missing from the C library */
  79 #ifndef HAVE_SETNS
  80 static inline int setns(int fd, int nstype)
  81 {
  82 #ifdef __NR_setns
  83         return syscall(__NR_setns, fd, nstype);
  84 #elif defined(__NR_set_ns)
  85         return syscall(__NR_set_ns, fd, nstype);
  86 #else
  87         errno = ENOSYS;
  88         return -1;
  89 #endif
  90 }
  91 #endif
  92
  93 /* Define sethostname() if missing from the C library */
  94 #ifndef HAVE_SETHOSTNAME
  95 static inline int sethostname(const char *name, size_t len)
  96 {
  97 #ifdef __NR_sethostname
  98 return syscall(__NR_sethostname, name, len);
  99 #else
 100 errno = ENOSYS;
 101 return -1;
 102 #endif
 103 }
 104 #endif
 105
 106 /* Define unshare() if missing from the C library */
 107 #ifndef HAVE_UNSHARE
 108 static inline int unshare(int flags)
 109 {
 110 #ifdef __NR_unshare
 111         return syscall(__NR_unshare, flags);
 112 #else
 113         errno = ENOSYS;
 114         return -1;
 115 #endif
 116 }
 117 #else
 118 extern int unshare(int);
 119 #endif
 120
 121 /* Define signalfd() if missing from the C library */
 122 #ifdef HAVE_SYS_SIGNALFD_H
 123 #  include <sys/signalfd.h>
 124 #else
 125 /* assume kernel headers are too old */
 126 #include <stdint.h>
 127 struct signalfd_siginfo
 128 {
 129         uint32_t ssi_signo;
 130         int32_t ssi_errno;
 131         int32_t ssi_code;
 132         uint32_t ssi_pid;
 133         uint32_t ssi_uid;
 134         int32_t ssi_fd;
 135         uint32_t ssi_tid;
 136         uint32_t ssi_band;
 137         uint32_t ssi_overrun;
 138         uint32_t ssi_trapno;
 139         int32_t ssi_status;
 140         int32_t ssi_int;
 141         uint64_t ssi_ptr;
 142         uint64_t ssi_utime;
 143         uint64_t ssi_stime;
 144         uint64_t ssi_addr;
 145         uint8_t __pad[48];
 146 };
 147
 148 #  ifndef __NR_signalfd4
 149 /* assume kernel headers are too old */
 150 #    if __i386__
 151 #      define __NR_signalfd4 327
 152 #    elif __x86_64__
 153 #      define __NR_signalfd4 289
 154 #    elif __powerpc__
 155 #      define __NR_signalfd4 313
 156 #    elif __s390x__
 157 #      define __NR_signalfd4 322
 158 #    elif __arm__
 159 #      define __NR_signalfd4 355
 160 #    elif __mips__ && _MIPS_SIM == _ABIO32
 161 #      define __NR_signalfd4 4324
 162 #    elif __mips__ && _MIPS_SIM == _ABI64
 163 #      define __NR_signalfd4 5283
 164 #    elif __mips__ && _MIPS_SIM == _ABIN32
 165 #      define __NR_signalfd4 6287
 166 #    endif
 167 #endif
 168
 169 #  ifndef __NR_signalfd
 170 /* assume kernel headers are too old */
 171 #    if __i386__
 172 #      define __NR_signalfd 321
 173 #    elif __x86_64__
 174 #      define __NR_signalfd 282
 175 #    elif __powerpc__
 176 #      define __NR_signalfd 305
 177 #    elif __s390x__
 178 #      define __NR_signalfd 316
 179 #    elif __arm__
 180 #      define __NR_signalfd 349
 181 #    elif __mips__ && _MIPS_SIM == _ABIO32
 182 #      define __NR_signalfd 4317
 183 #    elif __mips__ && _MIPS_SIM == _ABI64
 184 #      define __NR_signalfd 5276
 185 #    elif __mips__ && _MIPS_SIM == _ABIN32
 186 #      define __NR_signalfd 6280
 187 #    endif
 188 #endif
 189
 190 static inline int signalfd(int fd, const sigset_t *mask, int flags)
 191 {
 192         int retval;
 193
 194         retval = syscall (__NR_signalfd4, fd, mask, _NSIG / 8, flags);
 195         if (errno == ENOSYS && flags == 0)
 196                 retval = syscall (__NR_signalfd, fd, mask, _NSIG / 8);
 197         return retval;
 198 }
 199 #endif
 200
 201 #ifndef HAVE_MEMFD_CREATE
 202 static inline int memfd_create(const char *name, unsigned int flags) {
 203         #ifndef __NR_memfd_create
 204                 #if defined __i386__
 205                         #define __NR_memfd_create 356
 206                 #elif defined __x86_64__
 207                         #define __NR_memfd_create 319
 208                 #elif defined __arm__
 209                         #define __NR_memfd_create 385
 210                 #elif defined __aarch64__
 211                         #define __NR_memfd_create 279
 212                 #elif defined __s390__
 213                         #define __NR_memfd_create 350
 214                 #elif defined __powerpc__
 215                         #define __NR_memfd_create 360
 216                 #elif defined __sparc__
 217                         #define __NR_memfd_create 348
 218                 #elif defined __blackfin__
 219                         #define __NR_memfd_create 390
 220                 #elif defined __ia64__
 221                         #define __NR_memfd_create 1340
 222                 #elif defined _MIPS_SIM
 223                         #if _MIPS_SIM == _MIPS_SIM_ABI32
 224                                 #define __NR_memfd_create 4354
 225                         #endif
 226                         #if _MIPS_SIM == _MIPS_SIM_NABI32
 227                                 #define __NR_memfd_create 6318
 228                         #endif
 229                         #if _MIPS_SIM == _MIPS_SIM_ABI64
 230                                 #define __NR_memfd_create 5314
 231                         #endif
 232                 #endif
 233         #endif
 234         #ifdef __NR_memfd_create
 235         return syscall(__NR_memfd_create, name, flags);
 236         #else
 237         errno = ENOSYS;
 238         return -1;
 239         #endif
 240 }
 241 #else
 242 extern int memfd_create(const char *name, unsigned int flags);
 243 #endif
 244
 245 static inline int lxc_set_cloexec(int fd)
 246 {
 247         return fcntl(fd, F_SETFD, FD_CLOEXEC);
 248 }
 249
 250 /* Struct to carry child pid from lxc_popen() to lxc_pclose().
 251  * Not an opaque struct to allow direct access to the underlying FILE *
 252  * (i.e., struct lxc_popen_FILE *file; fgets(buf, sizeof(buf), file->f))
 253  * without additional wrappers.
 254  */
 255 struct lxc_popen_FILE {
 256         int pipe;
 257         FILE *f;
 258         pid_t child_pid;
 259 };
 260
 261 /* popen(command, "re") replacement that restores default signal mask
 262  * via sigprocmask(2) (unblocks all signals) after fork(2) but prior to calling exec(3).
 263  * In short, popen(command, "re") does pipe() + fork()                 + exec()
 264  * while lxc_popen(command)       does pipe() + fork() + sigprocmask() + exec().
 265  * Returns pointer to struct lxc_popen_FILE, that should be freed with lxc_pclose().
 266  * On error returns NULL.
 267  */
 268 extern struct lxc_popen_FILE *lxc_popen(const char *command);
 269
 270 /* pclose() replacement to be used on struct lxc_popen_FILE *,
 271  * returned by lxc_popen().
 272  * Waits for associated process to terminate, returns its exit status and
 273  * frees resources, pointed to by struct lxc_popen_FILE *.
 274  */
 275 extern int lxc_pclose(struct lxc_popen_FILE *fp);
 276
 277 /*
 278  * wait on a child we forked
 279  */
 280 extern int wait_for_pid(pid_t pid);
 281 extern int lxc_wait_for_pid_status(pid_t pid);
 282
 283 #if HAVE_LIBGNUTLS
 284 #define SHA_DIGEST_LENGTH 20
 285 extern int sha1sum_file(char *fnam, unsigned char *md_value);
 286 #endif
 287
 288 /* initialize rand with urandom */
 289 extern int randseed(bool);
 290
 291 /* are we unprivileged with respect to our namespaces */
 292 inline static bool am_guest_unpriv(void) {
 293         return geteuid() != 0;
 294 }
 295
 296 /* are we unprivileged with respect to init_user_ns */
 297 inline static bool am_host_unpriv(void)
 298 {
 299         FILE *f;
 300         uid_t user, host, count;
 301         int ret;
 302
 303         if (geteuid() != 0)
 304                 return true;
 305
 306         /* Now: are we in a user namespace? Because then we're also
 307          * unprivileged.
 308          */
 309         f = fopen("/proc/self/uid_map", "r");
 310         if (!f) {
 311                 return false;
 312         }
 313
 314         ret = fscanf(f, "%u %u %u", &user, &host, &count);
 315         fclose(f);
 316         if (ret != 3) {
 317                 return false;
 318         }
 319
 320         if (user != 0 || host != 0 || count != UINT32_MAX)
 321                 return true;
 322         return false;
 323 }
 324
 325 /*
 326  * parse /proc/self/uid_map to find what @orig maps to
 327  */
 328 extern uid_t get_ns_uid(uid_t orig);
 329 /*
 330  * parse /proc/self/gid_map to find what @orig maps to
 331  */
 332 extern gid_t get_ns_gid(gid_t orig);
 333
 334 extern bool dir_exists(const char *path);
 335
 336 #define FNV1A_64_INIT ((uint64_t)0xcbf29ce484222325ULL)
 337 extern uint64_t fnv_64a_buf(void *buf, size_t len, uint64_t hval);
 338
 339 extern bool is_shared_mountpoint(const char *path);
 340 extern int detect_shared_rootfs(void);
 341 extern bool detect_ramfs_rootfs(void);
 342 extern char *on_path(const char *cmd, const char *rootfs);
 343 extern bool cgns_supported(void);
 344 extern char *choose_init(const char *rootfs);
 345 extern bool switch_to_ns(pid_t pid, const char *ns);
 346 extern char *get_template_path(const char *t);
 347 extern int safe_mount(const char *src, const char *dest, const char *fstype,
 348                       unsigned long flags, const void *data,
 349                       const char *rootfs);
 350 extern int lxc_mount_proc_if_needed(const char *rootfs);
 351 extern int open_devnull(void);
 352 extern int set_stdfds(int fd);
 353 extern int null_stdfds(void);
 354 extern int lxc_preserve_ns(const int pid, const char *ns);
 355
 356 /* Check whether a signal is blocked by a process. */
 357 extern bool task_blocks_signal(pid_t pid, int signal);
 358
 359 /* Switch to a new uid and gid.
 360  * If LXC_INVALID_{G,U}ID is passed then the set{g,u}id() will not be called.
 361  */
 362 extern bool lxc_switch_uid_gid(uid_t uid, gid_t gid);
 363 extern bool lxc_setgroups(int size, gid_t list[]);
 364
 365 /* Find an unused loop device and associate it with source. */
 366 extern int lxc_prepare_loop_dev(const char *source, char *loop_dev, int flags);
 367
 368 /* Clear all mounts on a given node.
 369  * >= 0 successfully cleared. The number returned is the number of umounts
 370  *      performed.
 371  * < 0  error umounting. Return -errno.
 372  */
 373 extern int lxc_unstack_mountpoint(const char *path, bool lazy);
 374
 375 /*
 376  * run_command runs a command and collect it's std{err,out} output in buf.
 377  *
 378  * @param[out] buf     The buffer where the commands std{err,out] output will be
 379  *                     read into. If no output was produced, buf will be memset
 380  *                     to 0.
 381  * @param[in] buf_size The size of buf. This function will reserve one byte for
 382  *                     \0-termination.
 383  * @param[in] child_fn The function to be run in the child process. This
 384  *                     function must exec.
 385  * @param[in] args     Arguments to be passed to child_fn.
 386  */
 387 extern int run_command(char *buf, size_t buf_size, int (*child_fn)(void *),
 388                        void *args);
 389
 390 /* Concatenate all passed-in strings into one path. Do not fail. If any piece
 391  * is not prefixed with '/', add a '/'.
 392  */
 393 __attribute__((sentinel)) extern char *must_concat(const char *first, ...);
 394 __attribute__((sentinel)) extern char *must_make_path(const char *first, ...);
 395 __attribute__((sentinel)) extern char *must_append_path(char *first, ...);
 396
 397 /* return copy of string @entry;  do not fail. */
 398 extern char *must_copy_string(const char *entry);
 399
 400 /* Re-alllocate a pointer, do not fail */
 401 extern void *must_realloc(void *orig, size_t sz);
 402
 403 extern bool lxc_nic_exists(char *nic);
 404
 405 static inline uint64_t lxc_getpagesize(void)
 406 {
 407         int64_t pgsz;
 408
 409         pgsz = sysconf(_SC_PAGESIZE);
 410         if (pgsz <= 0)
 411                 pgsz = 1 << 12;
 412
 413         return pgsz;
 414 }
 415
 416 /* If n is not a power of 2 this function will return the next power of 2
 417  * greater than that number. Note that this function always returns the *next*
 418  * power of 2 *greater* that number not the *nearest*. For example, passing 1025
 419  * as argument this function will return 2048 although the closest power of 2
 420  * would be 1024.
 421  * If the caller passes in 0 they will receive 0 in return since this is invalid
 422  * input and 0 is not a power of 2.
 423  */
 424 extern uint64_t lxc_find_next_power2(uint64_t n);
 425
 426 static inline pid_t lxc_raw_gettid(void)
 427 {
 428 #ifdef SYS_gettid
 429         return syscall(SYS_gettid);
 430 #else
 431         return lxc_raw_getpid();
 432 #endif
 433 }
 434
 435 /* Set a signal the child process will receive after the parent has died. */
 436 extern int lxc_set_death_signal(int signal);
 437 extern int fd_cloexec(int fd, bool cloexec);
 438 extern int recursive_destroy(char *dirname);
 439
 440 #endif /* __LXC_UTILS_H */