3 * Copyright (C) 2003 Enrico Scholz <enrico.scholz@informatik.tu-chemnitz.de>
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2, or (at your option)
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
21 * \brief The public interface of the the libvserver library.
24 #ifndef H_VSERVER_SYSCALL_H
25 #define H_VSERVER_SYSCALL_H
30 #include <sys/types.h>
34 # define VC_ATTR_UNUSED __attribute__((__unused__))
35 # define VC_ATTR_NORETURN __attribute__((__noreturn__))
36 # define VC_ATTR_CONST __attribute__((__const__))
37 # define VC_ATTR_DEPRECATED __attribute__((__deprecated__))
38 # if __GNUC__*0x10000 + __GNUC_MINOR__*0x100 + __GNUC_PATCHLEVEL__ >= 0x30300
39 # define VC_ATTR_NONNULL(ARGS) __attribute__((__nonnull__ ARGS))
40 # define VC_ATTR_ALWAYSINLINE __attribute__((__always_inline__))
42 # define VC_ATTR_NONNULL(ARGS)
43 # define VC_ATTR_ALWAYSINLINE
45 # if __GNUC__*0x10000 + __GNUC_MINOR__*0x100 + __GNUC_PATCHLEVEL__ >= 0x30303
46 # define VC_ATTR_PURE __attribute__((__pure__))
51 # define VC_ATTR_NONNULL(ARGS)
52 # define VC_ATTR_UNUSED
53 # define VC_ATTR_NORETURN
54 # define VC_ATTR_ALWAYSINLINE
55 # define VC_ATTR_DEPRECATED
57 # define VC_ATTR_CONST
61 /** the value which is returned in error-case (no ctx found) */
62 #define VC_NOCTX ((xid_t)(-1))
63 /** the value which means a random (the next free) ctx */
64 #define VC_DYNAMIC_XID ((xid_t)(-1))
65 /** the value which means the current ctx */
66 #define VC_SAMECTX ((xid_t)(-2))
68 #define VC_NONID ((nid_t)(-1))
69 #define VC_DYNAMIC_NID ((nid_t)(-1))
71 #define VC_LIM_INFINITY (~0ULL)
72 #define VC_LIM_KEEP (~1ULL)
75 #ifndef S_CTX_INFO_LOCK
76 # define S_CTX_INFO_LOCK 1
79 #ifndef S_CTX_INFO_SCHED
80 # define S_CTX_INFO_SCHED 2
83 #ifndef S_CTX_INFO_NPROC
84 # define S_CTX_INFO_NPROC 4
87 #ifndef S_CTX_INFO_PRIVATE
88 # define S_CTX_INFO_PRIVATE 8
91 #ifndef S_CTX_INFO_INIT
92 # define S_CTX_INFO_INIT 16
95 #ifndef S_CTX_INFO_HIDEINFO
96 # define S_CTX_INFO_HIDEINFO 32
99 #ifndef S_CTX_INFO_ULIMIT
100 # define S_CTX_INFO_ULIMIT 64
103 #ifndef S_CTX_INFO_NAMESPACE
104 # define S_CTX_INFO_NAMESPACE 128
107 #define VC_CAP_CHOWN 0
108 #define VC_CAP_DAC_OVERRIDE 1
109 #define VC_CAP_DAC_READ_SEARCH 2
110 #define VC_CAP_FOWNER 3
111 #define VC_CAP_FSETID 4
112 #define VC_CAP_KILL 5
113 #define VC_CAP_SETGID 6
114 #define VC_CAP_SETUID 7
115 #define VC_CAP_SETPCAP 8
116 #define VC_CAP_LINUX_IMMUTABLE 9
117 #define VC_CAP_NET_BIND_SERVICE 10
118 #define VC_CAP_NET_BROADCAST 11
119 #define VC_CAP_NET_ADMIN 12
120 #define VC_CAP_NET_RAW 13
121 #define VC_CAP_IPC_LOCK 14
122 #define VC_CAP_IPC_OWNER 15
123 #define VC_CAP_SYS_MODULE 16
124 #define VC_CAP_SYS_RAWIO 17
125 #define VC_CAP_SYS_CHROOT 18
126 #define VC_CAP_SYS_PTRACE 19
127 #define VC_CAP_SYS_PACCT 20
128 #define VC_CAP_SYS_ADMIN 21
129 #define VC_CAP_SYS_BOOT 22
130 #define VC_CAP_SYS_NICE 23
131 #define VC_CAP_SYS_RESOURCE 24
132 #define VC_CAP_SYS_TIME 25
133 #define VC_CAP_SYS_TTY_CONFIG 26
134 #define VC_CAP_MKNOD 27
135 #define VC_CAP_LEASE 28
136 #define VC_CAP_QUOTACTL 29
138 #define VC_IMMUTABLE_FILE_FL 0x0000010lu
139 #define VC_IMMUTABLE_LINK_FL 0x0008000lu
140 #define VC_IMMUTABLE_ALL (VC_IMMUTABLE_LINK_FL|VC_IMMUTABLE_FILE_FL)
142 #define VC_IATTR_XID 0x01000000u
144 #define VC_IATTR_ADMIN 0x00000001u
145 #define VC_IATTR_WATCH 0x00000002u
146 #define VC_IATTR_HIDE 0x00000004u
147 #define VC_IATTR_FLAGS 0x00000007u
149 #define VC_IATTR_BARRIER 0x00010000u
150 #define VC_IATTR_IUNLINK 0x00020000u
151 #define VC_IATTR_IMMUTABLE 0x00040000u
155 #define VC_VXF_INFO_LOCK 0x00000001ull
156 #define VC_VXF_INFO_NPROC 0x00000004ull
157 #define VC_VXF_INFO_PRIVATE 0x00000008ull
158 #define VC_VXF_INFO_INIT 0x00000010ull
160 #define VC_VXF_INFO_HIDEINFO 0x00000020ull
161 #define VC_VXF_INFO_ULIMIT 0x00000040ull
162 #define VC_VXF_INFO_NAMESPACE 0x00000080ull
164 #define VC_VXF_SCHED_HARD 0x00000100ull
165 #define VC_VXF_SCHED_PRIO 0x00000200ull
166 #define VC_VXF_SCHED_PAUSE 0x00000400ull
168 #define VC_VXF_VIRT_MEM 0x00010000ull
169 #define VC_VXF_VIRT_UPTIME 0x00020000ull
170 #define VC_VXF_VIRT_CPU 0x00040000ull
172 #define VC_VXF_HIDE_MOUNT 0x01000000ull
173 #define VC_VXF_HIDE_NETIF 0x02000000ull
175 #define VC_VXF_STATE_SETUP (1ULL<<32)
176 #define VC_VXF_STATE_INIT (1ULL<<33)
179 #define VC_VXC_SET_UTSNAME 0x00000001ull
180 #define VC_VXC_SET_RLIMIT 0x00000002ull
182 #define VC_VXC_ICMP_PING 0x00000100ull
184 #define VC_VXC_SECURE_MOUNT 0x00010000ull
187 /** \defgroup syscalls Syscall wrappers
188 * Functions which are calling the vserver syscall directly. */
190 /** \defgroup helper Helper functions
191 * Functions which are doing general helper tasks like parameter parsing. */
193 /** \typedef an_unsigned_integer_type xid_t
194 * The identifier of a context. */
197 typedef an_unsigned_integer_type xid_t;
198 typedef an_unsigned_integer_type nid_t;
205 struct vc_ip_mask_pair {
210 /** \brief Returns the version of the current kernel API.
212 * \returns The versionnumber of the kernel API
214 int vc_get_version();
216 /** \brief Moves current process into a context
219 * Puts current process into context \a ctx, removes the capabilities
220 * given in \a remove_cap and sets \a flags.
222 * \param ctx The new context; special values for are
223 * - VC_SAMECTX which means the current context (just for changing caps and flags)
224 * - VC_DYNAMIC_XID which means the next free context; this value can be used by
225 * ordinary users also
226 * \param remove_cap The linux capabilities which will be \b removed.
227 * \param flags Special flags which will be set.
229 * \returns The new context-id, or VC_NOCTX on errors; errno
230 * will be set appropriately
232 * See http://vserver.13thfloor.at/Stuff/Logic.txt for details */
233 xid_t vc_new_s_context(xid_t ctx, unsigned int remove_cap, unsigned int flags);
235 /** \brief Sets the ipv4root information.
237 * \pre \a nb < NB_IPV4ROOT && \a ips != 0 */
238 int vc_set_ipv4root(uint32_t bcast, size_t nb,
239 struct vc_ip_mask_pair const *ips) VC_ATTR_NONNULL((3));
241 /** \brief Returns the value of NB_IPV4ROOT.
244 * This function returns the value of NB_IPV4ROOT which was used when the
245 * library was built, but \b not the value which is used by the currently
247 size_t vc_get_nb_ipv4root() VC_ATTR_CONST VC_ATTR_PURE;
249 /** \brief Creates a context without starting it.
252 * This functions initializes a new context. When already in a freshly
253 * created context, this old context will be discarded.
255 * \param xid The new context; special values are:
256 * - VC_DYNAMIC_XID which means to create a dynamic context
258 * \returns the xid of the created context, or VC_NOCTX on errors. errno
259 * will be set appropriately. */
260 xid_t vc_ctx_create(xid_t xid);
262 /** \brief Moves the current process into the specified context.
265 * \param xid The new context
266 * \returns 0 on success, -1 on errors */
267 int vc_ctx_migrate(xid_t xid);
269 /* rlimit related functions */
271 /** \brief The type which is used for a single limit value.
274 * - VC_LIM_INFINITY ... which is the infinite value
275 * - VC_LIM_KEEP ... which is used to mark values which shall not be
276 * modified by the vc_set_rlimit() operation.
278 * Else, the interpretation of the value depends on the corresponding
279 * resource; it might be bytes, pages, seconds or litres of beer. */
280 typedef uint_least64_t vc_limit_t;
282 /** \brief The limits of a resources.
284 * This is a triple consisting of a minimum, soft and hardlimit. */
286 vc_limit_t min; ///< the guaranted minimum of a resources
287 vc_limit_t soft; ///< the softlimit of a resource
288 vc_limit_t hard; ///< the absolute hardlimit of a resource
291 /** \brief Masks describing the supported limits. */
292 struct vc_rlimit_mask {
293 uint_least32_t min; ///< masks the resources supporting a minimum limit
294 uint_least32_t soft; ///< masks the resources supporting a soft limit
295 uint_least32_t hard; ///< masks the resources supporting a hard limit
298 /** \brief Returns the limits of \a resource.
301 * \param xid The id of the context
302 * \param resource The resource which will be queried
303 * \param lim The result which will be filled with the limits
305 * \returns 0 on success, and -1 on errors. */
306 int vc_get_rlimit(xid_t xid, int resource,
307 struct vc_rlimit /*@out@*/ *lim) VC_ATTR_NONNULL((3));
308 /** \brief Sets the limits of \a resource.
311 * \param xid The id of the context
312 * \param resource The resource which will be queried
313 * \param lim The new limits
315 * \returns 0 on success, and -1 on errors. */
316 int vc_set_rlimit(xid_t xid, int resource,
317 struct vc_rlimit const /*@in@*/ *lim) VC_ATTR_NONNULL((3));
318 int vc_get_rlimit_mask(xid_t xid,
319 struct vc_rlimit_mask *lim) VC_ATTR_NONNULL((2));
320 /** \brief Parses a string describing a limit
323 * This function parses \a str and interprets special words like \p "inf"
324 * or suffixes. Valid suffixes are
330 * \param str The string which shall be parsed
331 * \param res Will be filled with the interpreted value; in errorcase,
332 * this value is undefined.
334 * \returns \a true, iff the string \a str could be parsed. \a res will
335 * be filled with the interpreted value in this case.
337 * \pre \a str!=0 && \a res!=0
339 bool vc_parseLimit(char const /*@in@*/ *str, vc_limit_t /*@out@*/ *res) VC_ATTR_NONNULL((1,2));
342 /** \brief Sends a signal to a context/pid
345 * Special values for \a pid are:
346 * - -1 which means every process in ctx except the init-process
347 * - 0 which means every process in ctx inclusive the init-process */
348 int vc_ctx_kill(xid_t ctx, pid_t pid, int sig);
355 nid_t vc_get_task_nid(pid_t pid);
356 int vc_get_nx_info(nid_t nid, struct vc_nx_info *) VC_ATTR_NONNULL((2));
358 typedef enum { vcNET_IPV4, vcNET_IPV6, vcNET_IPV4R, vcNET_IPV6R } vc_net_nx_type;
367 nid_t vc_net_create(nid_t nid);
368 int vc_net_migrate(nid_t nid);
370 int vc_net_add(nid_t nid, struct vc_net_nx const *info);
371 int vc_net_remove(nid_t nid, struct vc_net_nx const *info);
373 struct vc_net_flags {
374 uint_least64_t flagword;
378 int vc_get_nflags(nid_t, struct vc_net_flags *);
379 int vc_set_nflags(nid_t, struct vc_net_flags const *);
383 uint_least64_t ncaps;
384 uint_least64_t cmask;
387 int vc_get_ncaps(nid_t, struct vc_net_caps *);
388 int vc_set_ncaps(nid_t, struct vc_net_caps const *);
393 int vc_set_iattr(char const *filename, xid_t xid,
394 uint_least32_t flags, uint_least32_t mask) VC_ATTR_NONNULL((1));
395 int vc_get_iattr(char const *filename, xid_t * /*@null@*/ xid,
396 uint_least32_t * /*@null@*/ flags,
397 uint_least32_t * /*@null@*/ mask) VC_ATTR_NONNULL((1));
404 /** Returns the context of the given process. pid==0 means the current process. */
405 xid_t vc_get_task_xid(pid_t pid);
406 int vc_get_vx_info(xid_t xid, struct vc_vx_info *info) VC_ATTR_NONNULL((2));
409 typedef enum { vcVHI_CONTEXT, vcVHI_SYSNAME, vcVHI_NODENAME,
410 vcVHI_RELEASE, vcVHI_VERSION, vcVHI_MACHINE,
411 vcVHI_DOMAINNAME } vc_uts_type;
413 int vc_set_vhi_name(xid_t xid, vc_uts_type type,
414 char const *val, size_t len) VC_ATTR_NONNULL((3));
415 int vc_get_vhi_name(xid_t xid, vc_uts_type type,
416 char *val, size_t len) VC_ATTR_NONNULL((3));
418 /** Returns true iff \a xid is a dynamic xid */
419 bool vc_is_dynamic_xid(xid_t xid);
421 int vc_enter_namespace(xid_t xid);
422 int vc_set_namespace();
423 int vc_cleanup_namespace();
426 /** \brief Flags of process-contexts
428 struct vc_ctx_flags {
429 /** \brief Mask of set context flags */
430 uint_least64_t flagword;
431 /** \brief Mask of set and unset context flags when used by set
432 * operations, or modifiable flags when used by get
437 /** \brief Capabilities of process-contexts */
439 /** \brief Mask of set common system capabilities */
440 uint_least64_t bcaps;
441 /** \brief Mask of set and unset common system capabilities when used by
442 * set operations, or the modifiable capabilities when used by
444 uint_least64_t bmask;
445 /** \brief Mask of set process context capabilities */
446 uint_least64_t ccaps;
447 /** \brief Mask of set and unset process context capabilities when used
448 * by set operations, or the modifiable capabilities when used
449 * by get operations */
450 uint_least64_t cmask;
453 /** \brief Information about parsing errors
455 struct vc_err_listparser {
456 char const *ptr; ///< Pointer to the first character of an erroneous string
457 size_t len; ///< Length of the erroneous string
460 int vc_get_cflags(xid_t xid, struct vc_ctx_flags *) VC_ATTR_NONNULL((2));
461 int vc_set_cflags(xid_t xid, struct vc_ctx_flags const *) VC_ATTR_NONNULL((2));
463 int vc_get_ccaps(xid_t xid, struct vc_ctx_caps *);
464 int vc_set_ccaps(xid_t xid, struct vc_ctx_caps const *);
466 /** \brief Converts a single string into bcapability
469 * \param str The string to be parsed;
470 * both "CAP_xxx" and "xxx" will be accepted
471 * \param len The length of the string, or \c 0 for automatic detection
473 * \returns 0 on error; a bitmask on success
476 uint_least64_t vc_text2bcap(char const *str, size_t len);
478 /** \brief Converts the lowest bit of a bcapability or the entire value
479 * (when possible) to a textual representation
482 * \param val The string to be converted; on success, the detected bit(s)
483 * will be unset, in errorcase only the lowest set bit
485 * \returns A textual representation of \a val resp. of its lowest set bit;
486 * or \c NULL in errorcase.
488 * \post \a *val<sub>old</sub> \c != 0 \c <-->
489 * \a *val<sub>old</sub> > \a *val<sub>new</sub>
490 * \post \a *val<sub>old</sub> \c == 0 \c ---> \a result == 0
492 char const * vc_lobcap2text(uint_least64_t *val) VC_ATTR_NONNULL((1));
494 /** \brief Converts a string into a bcapability-bitmask
497 * Syntax of \a str: \verbinclude list2xxx.syntax
499 * When the \c `~' prefix is used, the bits will be unset and a `~' after
500 * another `~' will cancel both ones. The \c `^' prefix specifies a
501 * bitnumber instead of a bitmask.
503 * "literal name" is everything which will be accepted by the
504 * vc_text2bcap() function. The special values for \c NAME will be
505 * recognized case insensitively
507 * \param str The string to be parsed
508 * \param len The length of the string, or \c 0 for automatic detection
509 * \param err Pointer to a structure for error-information, or \c NULL.
510 * \param cap Pointer to a vc_ctx_caps structure holding the results;
511 * only the \a bcaps and \a bmask fields will be changed and
512 * already set values will not be honored. When an error
513 * occured, \a cap will have the value of all processed valid
516 * \returns 0 on success, -1 on error. In error case, \a err will hold
517 * position and length of the first not understood BCAP part
518 * \pre \a str != 0 && \a cap != 0;
519 * \a cap->bcaps and \a cap->bmask must be initialized
521 int vc_list2bcap(char const *str, size_t len,
522 struct vc_err_listparser *err,
523 struct vc_ctx_caps *cap) VC_ATTR_NONNULL((1,4));
525 uint_least64_t vc_text2ccap(char const *, size_t len);
526 char const * vc_loccap2text(uint_least64_t *);
527 int vc_list2ccap(char const *, size_t len,
528 struct vc_err_listparser *err,
529 struct vc_ctx_caps *);
531 int vc_list2cflag(char const *, size_t len,
532 struct vc_err_listparser *err,
533 struct vc_ctx_flags *flags);
534 uint_least64_t vc_text2cflag(char const *, size_t len);
535 char const * vc_locflag2text(uint_least64_t *);
537 uint_least32_t vc_list2cflag_compat(char const *, size_t len,
538 struct vc_err_listparser *err);
539 uint_least32_t vc_text2cflag_compat(char const *, size_t len);
540 char const * vc_hicflag2text_compat(uint_least32_t);
542 int vc_text2cap(char const *);
543 char const * vc_cap2text(unsigned int);
546 int vc_list2nflag(char const *, size_t len,
547 struct vc_err_listparser *err,
548 struct vc_net_flags *flags);
549 uint_least64_t vc_text2nflag(char const *, size_t len);
550 char const * vc_lonflag2text(uint_least64_t *);
552 uint_least64_t vc_text2ncap(char const *, size_t len);
553 char const * vc_loncap2text(uint_least64_t *);
554 int vc_list2ncap(char const *, size_t len,
555 struct vc_err_listparser *err,
556 struct vc_net_caps *);
558 uint_least64_t vc_get_insecurebcaps() VC_ATTR_CONST;
559 inline static uint_least64_t vc_get_insecureccaps() {
560 return ~(VC_VXC_SET_UTSNAME|VC_VXC_ICMP_PING);
563 inline static int vc_setfilecontext(char const *filename, xid_t xid) {
564 return vc_set_iattr(filename, xid, 0, VC_IATTR_XID);
567 inline static xid_t vc_getfilecontext(char const *filename) {
569 if (vc_get_iattr(filename, &res, 0,0)==-1) return VC_NOCTX;
574 struct vc_set_sched {
583 int vc_set_sched(xid_t xid, struct vc_set_sched const *);
586 typedef enum { vcFEATURE_VKILL, vcFEATURE_IATTR, vcFEATURE_RLIMIT,
587 vcFEATURE_COMPAT, vcFEATURE_MIGRATE, vcFEATURE_NAMESPACE,
588 vcFEATURE_SCHED, vcFEATURE_VINFO, vcFEATURE_VHI,
589 vcFEATURE_VSHELPER0, vcFEATURE_VSHELPER }
592 bool vc_isSupported(vcFeatureSet) VC_ATTR_CONST;
593 bool vc_isSupportedString(char const *);
595 /* The management part */
597 #define VC_LIMIT_VSERVER_NAME_LEN 1024
599 typedef enum { vcCFG_NONE, vcCFG_AUTO,
602 vcCFG_RECENT_FULL } vcCfgStyle;
605 /** Maps an xid given at '--xid' options to an xid_t */
606 xid_t vc_xidopt2xid(char const *, bool honor_static, char const **err_info);
608 vcCfgStyle vc_getVserverCfgStyle(char const *id);
610 /** Resolves the name of the vserver. The result will be allocated and must
611 be freed by the caller. */
612 char * vc_getVserverName(char const *id, vcCfgStyle style);
614 /** Returns the path of the vserver configuration directory. When the given
615 * vserver does not exist, or when it does not have such a directory, NULL
616 * will be returned. Else, the result will be allocated and must be freed
618 char * vc_getVserverCfgDir(char const *id, vcCfgStyle style);
620 /** Returns the path of the configuration directory for the given
621 * application. The result will be allocated and must be freed by the
623 char * vc_getVserverAppDir(char const *id, vcCfgStyle style, char const *app);
625 /** Returns the path to the vserver root-directory. The result will be
626 * allocated and must be freed by the caller. */
627 char * vc_getVserverVdir(char const *id, vcCfgStyle style, bool physical);
629 /** Returns the ctx of the given vserver. When vserver is not running and
630 * 'honor_static' is false, VC_NOCTX will be returned. Else, when
631 * 'honor_static' is true and a static assignment exists, those value will
632 * be returned. Else, the result will be VC_NOCTX.
634 * When 'is_running' is not null, the status of the vserver will be
635 * assigned to this variable. */
636 xid_t vc_getVserverCtx(char const *id, vcCfgStyle style,
637 bool honor_static, bool /*@null@*/ *is_running);
639 /** Resolves the cfg-path of the vserver owning the given ctx. 'revdir' will
640 be used as the directory holding the mapping-links; when NULL, the
641 default value will be assumed. The result will be allocated and must be
642 freed by the caller. */
643 char * vc_getVserverByCtx(xid_t ctx, /*@null@*/vcCfgStyle *style,
644 /*@null@*/char const *revdir);
646 #define vcSKEL_INTERFACES 1u
647 #define vcSKEL_PKGMGMT 2u
648 #define vcSKEL_FILESYSTEM 4u
650 /** Create a basic configuration skeleton for a vserver plus toplevel
651 * directories for pkgmanagemt and filesystem (when requested). */
652 int vc_createSkeleton(char const *id, vcCfgStyle style, int flags);
659 #undef VC_ATTR_ALWAYSINLINE
660 #undef VC_ATTR_NORETURN
661 #undef VC_ATTR_UNUSED
662 #undef VC_ATTR_NONNULL