1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21
22 /*
23 * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
24 * Use is subject to license terms.
25 */
26
27 /*
28 * Copyright 2018 Nexenta Systems, Inc.
29 * Copyright 2019 Nexenta by DDN, Inc.
30 */
31
32 #ifndef _NFS4_H
33 #define _NFS4_H
34
35 #include <sys/types.h>
36 #include <sys/vnode.h>
37 #include <sys/fem.h>
38 #include <rpc/rpc.h>
39 #include <nfs/nfs.h>
40
41 #ifdef _KERNEL
42 #include <nfs/nfs4_kprot.h>
43 #include <sys/nvpair.h>
44 #else
45 #include <rpcsvc/nfs4_prot.h>
46 #endif
47 #include <nfs/nfs4_attr.h>
48 #include <sys/acl.h>
49 #include <sys/list.h>
50 #include <nfs/nfs4x.h>
51
52 #ifdef __cplusplus
53 extern "C" {
54 #endif
55
56 #define NFS4_MAX_SECOID4 65536
57 #define NFS4_MAX_UTF8STRING 65536
58 #define NFS4_MAX_LINKTEXT4 65536
59 #define NFS4_MAX_PATHNAME4 65536
60
61 struct nfs_fsl_info {
62 uint_t netbuf_len;
63 uint_t netnm_len;
64 uint_t knconf_len;
65 char *netname;
66 struct netbuf *addr;
67 struct knetconfig *knconf;
68 };
69
70 #ifdef _KERNEL
71
72 typedef struct nfs4_fhandle {
73 int fh_len;
74 char fh_buf[NFS4_FHSIZE];
75 } nfs4_fhandle_t;
76
77 typedef uint8_t nfs4_minor_t;
78
79 #define NFS4_MINORVERSION 0
80 #define CB4_MINORVERSION 0
81
82 #define FIRST_NFS4_OP OP_ACCESS
83 #define LAST_NFS40_OP OP_RELEASE_LOCKOWNER
84 #define LAST_NFS41_OP OP_RECLAIM_COMPLETE
85 #define LAST_NFS42_OP OP_RECLAIM_COMPLETE
86 #define LAST_NFS4_OP LAST_NFS42_OP
87
88 /*
89 * Set the fattr4_change variable using a time struct. Note that change
90 * is 64 bits, but timestruc_t is 128 bits in a 64-bit kernel.
91 */
92 #define NFS4_SET_FATTR4_CHANGE(change, ts) \
93 { \
94 change = (ts).tv_sec; \
95 change <<= 32; \
96 change |= (uint32_t)((ts).tv_nsec); \
97 }
98
99 /*
100 * Server lease period. Value is in seconds; Also used for grace period
101 */
102 extern time_t rfs4_lease_time;
103
104 /*
105 * This set of typedefs and interfaces represent the core or base set
106 * of functionality that backs the NFSv4 server's state related data
107 * structures. Since the NFSv4 server needs inter-RPC state to be
108 * available that is unrelated to the filesystem (in other words,
109 * soft-state), this functionality is needed to maintain that and is
110 * written to be somewhat flexible to adapt to the various types of
111 * data structures contained within the server.
112 *
113 * The basic structure at this level is that the server maintains a
114 * global "database" which consists of a set of tables. Each table
115 * contains a set of like data structures. Each table is indexed by
116 * at least one hash function and in most cases two hashes. Each
117 * table's characteristics is set when it is created at run-time via
118 * rfs4_table_create(). All table creation and related functions are
119 * located in nfs4_state.c. The generic database functionality is
120 * located in nfs4_db.c.
121 */
122
123 typedef struct rfs4_dbe rfs4_dbe_t; /* basic opaque db entry */
124 typedef struct rfs4_table rfs4_table_t; /* basic table type */
125 typedef struct rfs4_index rfs4_index_t; /* index */
126 typedef struct rfs4_database rfs4_database_t; /* and database */
127
128 typedef struct { /* opaque entry type for later use */
129 rfs4_dbe_t *dbe;
130 } *rfs4_entry_t;
131
132 /*
133 * NFSv4 server state databases
134 *
135 * Initialized when the module is loaded and used by NFSv4 state tables.
136 * These kmem_cache free pools are used globally, the NFSv4 state tables
137 * which make use of these kmem_cache free pools are per zone.
138 */
139 extern kmem_cache_t *rfs4_client_mem_cache;
140 extern kmem_cache_t *rfs4_clntIP_mem_cache;
141 extern kmem_cache_t *rfs4_openown_mem_cache;
142 extern kmem_cache_t *rfs4_openstID_mem_cache;
143 extern kmem_cache_t *rfs4_lockstID_mem_cache;
144 extern kmem_cache_t *rfs4_lockown_mem_cache;
145 extern kmem_cache_t *rfs4_file_mem_cache;
146 extern kmem_cache_t *rfs4_delegstID_mem_cache;
147 extern kmem_cache_t *rfs4_session_mem_cache;
148
149 /* database, table, index creation entry points */
150 extern rfs4_database_t *rfs4_database_create(uint32_t);
151 extern void rfs4_database_shutdown(rfs4_database_t *);
152 extern void rfs4_database_destroy(rfs4_database_t *);
153
154 extern void rfs4_database_destroy(rfs4_database_t *);
155
156 extern kmem_cache_t *nfs4_init_mem_cache(char *, uint32_t, uint32_t,
157 uint32_t);
158 extern rfs4_table_t *rfs4_table_create(rfs4_database_t *, char *,
159 time_t, uint32_t,
160 bool_t (*create)(rfs4_entry_t, void *),
161 void (*destroy)(rfs4_entry_t),
162 bool_t (*expiry)(rfs4_entry_t),
163 uint32_t, uint32_t, uint32_t, id_t);
164 extern void rfs4_table_destroy(rfs4_database_t *, rfs4_table_t *);
165 extern rfs4_index_t *rfs4_index_create(rfs4_table_t *, char *,
166 uint32_t (*hash)(void *),
167 bool_t (compare)(rfs4_entry_t, void *),
168 void *(*mkkey)(rfs4_entry_t), bool_t);
169 extern void rfs4_index_destroy(rfs4_index_t *);
170
171 /* Type used to direct rfs4_dbsearch() in what types of records to inspect */
172 typedef enum {RFS4_DBS_VALID, RFS4_DBS_INVALID} rfs4_dbsearch_type_t;
173 /* search and db entry manipulation entry points */
174 extern rfs4_entry_t rfs4_dbsearch(rfs4_index_t *, void *,
175 bool_t *, void *, rfs4_dbsearch_type_t);
176 extern void rfs4_dbe_lock(rfs4_dbe_t *);
177 extern void rfs4_dbe_unlock(rfs4_dbe_t *);
178 extern clock_t rfs4_dbe_twait(rfs4_dbe_t *, clock_t);
179 extern void rfs4_dbe_cv_broadcast(rfs4_dbe_t *);
180 extern void rfs4_dbe_hold(rfs4_dbe_t *);
181 extern void rfs4_dbe_hold_nolock(rfs4_dbe_t *);
182 extern void rfs4_dbe_rele_nolock(rfs4_dbe_t *);
183 extern void rfs4_dbe_rele(rfs4_dbe_t *);
184 extern uint32_t rfs4_dbe_refcnt(rfs4_dbe_t *);
185 extern id_t rfs4_dbe_getid(rfs4_dbe_t *);
186 extern void rfs4_dbe_invalidate(rfs4_dbe_t *);
187 extern bool_t rfs4_dbe_is_invalid(rfs4_dbe_t *);
188 extern time_t rfs4_dbe_get_timerele(rfs4_dbe_t *);
189 extern void rfs4_dbe_hide(rfs4_dbe_t *);
190 extern void rfs4_dbe_unhide(rfs4_dbe_t *);
191 #ifdef DEBUG
192 extern bool_t rfs4_dbe_islocked(rfs4_dbe_t *);
193 #endif
194 extern void rfs4_dbe_walk(rfs4_table_t *,
195 void (*callout)(rfs4_entry_t, void *), void *);
196
197 /*
198 * Minimal server stable storage.
199 *
200 * Currently the NFSv4 server will only save the client
201 * ID (the long version) so that it will be able to
202 * grant possible reclaim requests during the infamous
203 * grace_period.
204 */
205
206 #define RFS4_SS_DIRSIZE 64 * 1024
207 #define NFS4_SS_VERSION 1
208
209 /* handy pathname structure */
210 typedef struct ss_pn {
211 char *leaf;
212 char pn[MAXPATHLEN];
213 } rfs4_ss_pn_t;
214
215 /*
216 * The server will build this link list on startup. It represents the
217 * clients that have had valid state on the server in a prior instance.
218 *
219 */
220 typedef struct rfs4_oldstate {
221 struct rfs4_oldstate *next;
222 struct rfs4_oldstate *prev;
223 rfs4_ss_pn_t *ss_pn;
224 nfs_client_id4 cl_id4;
225 } rfs4_oldstate_t;
226
227 /*
228 * This union is used to overlay the server's internal treatment of
229 * the protocols stateid4 datatype. Therefore, "bits" must not exceed
230 * the size of stateid4 and more importantly should match the size of
231 * stateid4. The chgseq field must the first entry since it overlays
232 * stateid4.seqid.
233 */
234 typedef union {
235 stateid4 stateid;
236 struct {
237 uint32_t chgseq; /* State changes / protocol's seqid */
238 uint32_t boottime; /* boot time */
239 uint32_t type:2; /* stateid_type_t as define below */
240 uint32_t clnodeid:8; /* cluster server nodeid */
241 uint32_t ident:22; /* 2^22-1 openowner x fhs */
242 pid_t pid; /* pid of corresponding lock owner */
243 } bits;
244 } stateid_t;
245 /*
246 * Note that the way the type field above is defined, this enum must
247 * not have more than 4 members.
248 */
249 typedef enum {OPENID, LOCKID, DELEGID} stateid_type_t;
250
251 /*
252 * "wait" struct for use in the open open and lock owner state
253 * structures to provide serialization between server threads that are
254 * handling requests for the same open owner or lock stateid. This
255 * way only one thread will be updating things like sequence ids,
256 * replay cache and stateid at a time.
257 */
258 typedef struct rfs4_state_wait {
259 uint32_t sw_active;
260 uint32_t sw_wait_count;
261 kmutex_t sw_cv_lock[1];
262 kcondvar_t sw_cv[1];
263 } rfs4_state_wait_t;
264
265 extern void rfs4_sw_enter(rfs4_state_wait_t *);
266 extern void rfs4_sw_exit(rfs4_state_wait_t *);
267
268 /*
269 * This enum and the following rfs4_cbinfo_t struct are used to
270 * maintain information about the callback path used from the server
271 * to client for operations like CB_GETATTR and CB_RECALL. The
272 * rfs4_cbinfo_t struct is meant to be encompassed in the client
273 * struct and managed within that structure's locking scheme.
274 *
275 * The various states of the callback path are used by the server to
276 * determine if delegations should initially be provided to a client
277 * and then later on if connectivity has been lost and delegations
278 * should be revoked.
279 */
280
281 /*
282 * CB_NOCHANGE - Special value used for interfaces within the delegation
283 * code to signify that "no change" has occurred to the
284 * callback path
285 * CB_UNINIT - No callback info provided by the client
286 * CB_NONE - Callback info provided but CB_NULL call
287 * has yet to be attempted
288 * CB_OK - Callback path tested with CB_NULL with success
289 * CB_INPROG - Callback path currently being tested with CB_NULL
290 * CB_FAILED - Callback path was == CB_OK but has failed
291 * with timeout/rpc error
292 * CB_BAD - Callback info provided but CB_NULL failed
293 */
294 typedef enum {
295 CB_NOCHANGE = 0,
296 CB_UNINIT = 1,
297 CB_NONE = 2,
298 CB_OK = 3,
299 CB_INPROG = 4,
300 CB_FAILED = 5,
301 CB_BAD = 6
302 } rfs4_cbstate_t;
303
304 #define RFS4_CBCH_MAX 10 /* size callback client handle cache */
305 /*
306 * Callback info for a client.
307 * Client only provides: cb_client4 and cb_ident
308 * The rest of the information is used to track callback path status
309 * and usage.
310 *
311 * cb_state - used as comments for the rfs4_cbstate_t enum indicate
312 * cb_notified_of_cb_path_down - if the callback path was once CB_OK and
313 * has hence CB_FAILED, the client needs to be notified via RENEW.
314 * cb_timefailed - current time when cb_state transitioned from
315 * CB_OK -> CB_FAILED. Meant for observability. When did that happen?
316 * cb_chc_free/cb_chc - cache of client handles for the callback path
317 * cb_ident - SETCLIENTID provided callback_ident value
318 * callback - SETCLIENTID provided cb_client4 value
319 * cb_refcnt - current number of users of this structure's content
320 * protected by cb_lock
321 * cb_badbehavior - how many times did a client do something we didn't like?
322 * cb_lock - lock for contents of cbinfo
323 * cb_cv - used to allow threads to wait on CB_NULL completion
324 * cb_nullcaller - is there a thread currently taking care of
325 * new callback information?
326 * cb_cv_nullcaller - used by the thread doing CB_NULL to wait on
327 * threads that may be using client handles of the current
328 * client handle cache.
329 * newer - new callback info provided by a client and awaiting
330 * CB_NULL testing and move to regular cbinfo.
331 */
332 typedef struct {
333 rfs4_cbstate_t cb_state;
334 unsigned cb_notified_of_cb_path_down:1;
335 time_t cb_timefailed;
336 int cb_chc_free;
337 CLIENT *cb_chc[RFS4_CBCH_MAX];
338 uint32_t cb_ident;
339 cb_client4 cb_callback;
340 uint32_t cb_refcnt;
341 uint32_t cb_badbehavior;
342 kmutex_t cb_lock[1];
343 kcondvar_t cb_cv[1];
344 bool_t cb_nullcaller;
345 kcondvar_t cb_cv_nullcaller[1];
346 struct {
347 bool_t cb_new;
348 bool_t cb_confirmed;
349 uint32_t cb_ident;
350 cb_client4 cb_callback;
351 } cb_newer;
352 } rfs4_cbinfo_t;
353
354 /*
355 * A server instance. We can associate sets of clients - via a pointer in
356 * rfs4_client_t - with a given server instance, allowing us to treat clients
357 * in the set differently to clients in other sets.
358 *
359 * Currently used only for Sun Cluster HA-NFS support, to group clients
360 * on NFS resource failover so each set of clients gets its own dedicated
361 * grace period and distributed stable storage data.
362 */
363 typedef struct rfs4_servinst {
364 int dss_npaths;
365 krwlock_t rwlock;
366 krwlock_t oldstate_lock;
367 time_t start_time;
368 time_t grace_period;
369 uint_t nreclaim; /* number reclaim clients */
370 rfs4_oldstate_t *oldstate;
371 struct rfs4_dss_path **dss_paths;
372 struct rfs4_servinst *next;
373 struct rfs4_servinst *prev;
374 } rfs4_servinst_t;
375
376 /*
377 * DSS: distributed stable storage
378 */
379
380 typedef struct rfs4_dss_path {
381 struct rfs4_dss_path *next; /* for insque/remque */
382 struct rfs4_dss_path *prev; /* for insque/remque */
383 char *path;
384 struct rfs4_servinst *sip;
385 unsigned index; /* offset in servinst's array */
386 } rfs4_dss_path_t;
387
388 /* array of paths passed-in from nfsd command-line; stored in nvlist */
389 extern char **rfs4_dss_newpaths;
390 extern uint_t rfs4_dss_numnewpaths;
391
392 /* nvlists of all DSS paths: current, and before last warmstart */
393 extern nvlist_t *rfs4_dss_paths, *rfs4_dss_oldpaths;
394
395 /*
396 * The server maintains a set of state on a per client basis that
397 * matches that of the protocol requirements. A client's state is
398 * rooted with the rfs4_client_t struct of which there is one per
399 * client and is created when SETCLIENTID/SETCLIENTID_CONFIRM are
400 * received. From there, the server then creates rfs4_openowner_t
401 * structs for each new open owner from that client and are initiated
402 * at OPEN/OPEN_CONFIRM (when the open owner is new to the server).
403 * At OPEN, at least two other structures are created, and potentially a
404 * third. rfs4_state_t is created to track the association between an
405 * open owner and a particular file. An rfs4_file_t struct may be
406 * created (if the file is not already open) at OPEN as well. The
407 * rfs4_file_t struct is the only one that is per server and not per
408 * client. The rfs4_deleg_state_t struct is created in the
409 * instance that the server is going to provide a delegation for the
410 * file being OPENed. Finally, the rfs4_lockowner_t is created at the
411 * first use of a lock owner at the server and is a result of the LOCK
412 * operation. The rfs4_lo_state_t struct is then created to represent
413 * the relation between the lock owner and the file.
414 *
415 */
416 /*
417 * The following ascii art represents each of these data structs and
418 * their references to each other. Note: "<-(x)->" represents the
419 * doubly link lists (list_t).
420 *
421 * ____________________
422 * | |
423 * | rfs4_client_t |
424 * ->| (1) |<-
425 * / |____________________| \
426 * / ^ \
427 * / | \
428 * ____________________ ____________________ ____________________
429 * | | | | | |
430 * | rfs4_lockowner_t | | rfs4_openowner_t | | rfs4_deleg_state_t |
431 * | | | (3) <-(1)-> | | <-(2)-> |
432 * |____________________| |____________________| |____________________|
433 * ^ ^ |
434 * | | V
435 * ____________________ ____________________ ____________________
436 * | | | | | |
437 * | rfs4_lo_state_t |->| rfs4_state_t |->| rfs4_file_t |
438 * | <-(4)-> | | (4) <-(3)-> | | (2) |
439 * |____________________| |____________________| |____________________|
440 */
441 /*
442 * Each of these data types are kept in a separate rfs4_table_t and is
443 * actually encapsulated within a rfs4_dbe_t struct. The various
444 * tables and their construction is done in nfs4_state.c but
445 * documented here to completeness.
446 *
447 * Table Data struct stored Indexed by
448 * ----- ------------------ ----------
449 * rfs4_client_tab rfs4_client_t nfs_client_id4
450 * clientid4
451 *
452 * rfs4_openowner_tab rfs4_openowner_t open_owner4
453 *
454 * rfs4_state_tab rfs4_state_t open_owner4 | file
455 * stateid
456 *
457 * rfs4_lo_state_tab rfs4_lo_state_t lockowner | stateid
458 * lock_stateid
459 *
460 * rfs4_lockowner_tab rfs4_lockowner_t lockowner
461 * pid
462 *
463 * rfs4_file_tab rfs4_file_t filehandle
464 *
465 * rfs4_deleg_state_tab rfs4_deleg_state_t clientid4 | file
466 * deleg_stateid
467 */
468
469 /*
470 * The client struct, it is the root of all state for a particular
471 * client. The client is identified by the nfs_client_id4 via
472 * SETCLIENTID and the server returns the clientid4 as short hand reference
473 */
474 /*
475 * Client struct - as mentioned above it is the root of all state for
476 * a single client as identified by the client supplied nfs_client_id4
477 *
478 * dbe - encapsulation struct
479 * clientid - server assigned short hand reference to client
480 * nfs_client - client supplied identifier for itself
481 * confirm_verf - the value provided to the client for SETCLIENTID_CONFIRM
482 * need_confirm - does this client need to be SETCLIENTID_CONFIRMed?
483 *
484 * unlksys_completed - has an F_UNLKSYS been done for this client which
485 * says that the use of cleanlocks() on individual files
486 * is not required?
487 * can_reclaim - indicates if client is allowed to reclaim after server
488 * start-up (client had previous state at server)
489 * ss_remove - indicates that the rfs4_client_destroy function should
490 * clean up stable storage file.
491 * forced_expire - set if the sysadmin has used clear_locks for this client.
492 * no_referrals - set if the client is Solaris and pre-dates referrals
493 * deleg_revoked - how many delegations have been revoked for this client?
494 *
495 * cp_confirmed - this refers to a confirmed client struct that has
496 * the same nfs_client_id4 as this client struct. When/if this client
497 * struct is confirmed via SETCLINETID_CONFIRM, the previously
498 * confirmed client struct will be "closed" and hence this reference.
499 *
500 * last_access - used to determine if the client has let its lease expire
501 * cbinfo - struct containing all callback related information
502 * cr_set - credentials used for the SETCLIENTID/SETCLIENTID_CONFIRM pair
503 * sysid - the lock manager sysid allocated for this client's file locks
504 * openownerlist - root of openowners list associated with this client
505 * ss_pn - Pathname to the stable storage file.
506 * cl_addr - Clients network address.
507 * server_instance - pointer to the currently associated server instance
508 */
509 typedef struct rfs4_client {
510 rfs4_dbe_t *rc_dbe;
511 clientid4 rc_clientid;
512 nfs_client_id4 rc_nfs_client;
513 verifier4 rc_confirm_verf;
514 unsigned rc_need_confirm:1;
515 unsigned rc_unlksys_completed:1;
516 unsigned rc_can_reclaim:1;
517 unsigned rc_ss_remove:1;
518 unsigned rc_forced_expire:1;
519 unsigned rc_reclaim_completed:1;
520 uint_t rc_deleg_revoked;
521 struct rfs4_client *rc_cp_confirmed;
522 time_t rc_last_access;
523 rfs4_cbinfo_t rc_cbinfo;
524 cred_set_t rc_cr_set;
525 sysid_t rc_sysidt;
526 list_t rc_openownerlist;
527 rfs4_ss_pn_t *rc_ss_pn;
528 struct sockaddr_storage rc_addr;
529 rfs4_servinst_t *rc_server_instance;
530
531 /* nfsv4.1 */
532 rfs41_csr_t rc_contrived;
533 rfs41_sprot_t rc_state_prot;
534 list_t rc_sessions;
535 unsigned rc_destroying:1; /* flag: going to destroy */
536 } rfs4_client_t;
537
538 /*
539 * ClntIP struct - holds the diagnosis about whether the client
540 * cannot support referrals. Set to true for old Solaris clients.
541 */
542
543 typedef struct rfs4_clntip {
544 rfs4_dbe_t *ri_dbe;
545 struct sockaddr_storage ri_addr;
546 unsigned ri_no_referrals:1;
547 } rfs4_clntip_t;
548
549 /*
550 * The openowner contains the client supplied open_owner4 as well as
551 * the matching sequence id and is used to track the client's usage of
552 * the open_owner4. Note that a reply is saved here as well for
553 * processing of retransmissions.
554 *
555 * dbe - encapsulation struct
556 * client - reference to rfs4_client_t for this openowner
557 * owner - actual client supplied open_owner4
558 * need_confirm - does this openowner need to be OPEN_CONFIRMed
559 * postpone_confirm - set if error received on first use of open_owner
560 * state2confirm - what stateid4 should be used on the OPEN_CONFIRM
561 * open_seqid - what is the next open_seqid expected for this openowner
562 * oo_sw - used to serialize access to the open seqid/reply handling
563 * statelist - root of state struct list associated with this openowner
564 * node - node for client struct list of openowners
565 * reply_fh - open replay processing needs the filehandle so that it is
566 * able to reset the current filehandle for appropriate compound
567 * processing and reply.
568 * reply - last reply sent in relation to this openowner
569 */
570 typedef struct rfs4_openowner {
571 rfs4_dbe_t *ro_dbe;
572 rfs4_client_t *ro_client;
573 open_owner4 ro_owner;
574 unsigned ro_need_confirm:1;
575 unsigned ro_postpone_confirm:1;
576 seqid4 ro_open_seqid;
577 rfs4_state_wait_t ro_sw;
578 list_t ro_statelist;
579 list_node_t ro_node;
580 nfs_fh4 ro_reply_fh;
581 nfs_resop4 ro_reply;
582 } rfs4_openowner_t;
583
584 /*
585 * This state struct represents the association between an openowner
586 * and a file that has been OPENed by that openowner.
587 *
588 * dbe - encapsulation struct
589 * stateid - server provided stateid
590 * owner - reference back to the openowner for this state
591 * finfo - reference to the open file for this state
592 * open_access - how did the openowner OPEN the file (access)
593 * open_deny - how did the openowner OPEN the file (deny)
594 * share_access - what share reservation is on the file (access)
595 * share_deny - what share reservation is on the file (deny)
596 * closed - has this file been closed?
597 * lostatelist - root of list of lo_state associated with this state/file
598 * node - node for state struct list of states
599 */
600 typedef struct rfs4_state {
601 rfs4_dbe_t *rs_dbe;
602 stateid_t rs_stateid;
603 rfs4_openowner_t *rs_owner;
604 struct rfs4_file *rs_finfo;
605 uint32_t rs_open_access;
606 uint32_t rs_open_deny;
607 uint32_t rs_share_access;
608 uint32_t rs_share_deny;
609 unsigned rs_closed:1;
610 list_t rs_lostatelist;
611 list_node_t rs_node;
612 } rfs4_state_t;
613
614 /*
615 * Lockowner - track the lockowner and its related info
616 *
617 * dbe - encapsulation struct
618 * client - reference to the client
619 * owner - lockowner supplied by the client
620 * pid - local identifier used for file locking
621 */
622 typedef struct rfs4_lockowner {
623 rfs4_dbe_t *rl_dbe;
624 rfs4_client_t *rl_client;
625 lock_owner4 rl_owner;
626 pid_t rl_pid;
627 } rfs4_lockowner_t;
628
629 /*
630 * Lockowner_state associated with a state struct and lockowner
631 *
632 * dbe - encapsulation struct
633 * state - reference back to state struct for open file
634 * lockid - stateid for this lockowner/state
635 * locker - reference to lockowner
636 * seqid - sequence id for this lockowner/state
637 * skip_seqid_check - used on initialization of struct
638 * locks_cleaned - have all locks been released for this lockowner/file?
639 * lock_completed - successful LOCK with lockowner/file?
640 * ls_sw - used to serialize update seqid/reply/stateid handling
641 * node - node for state struct list of lo_states
642 * reply - last reply sent in relation to this lockowner/state
643 */
644 typedef struct rfs4_lo_state {
645 rfs4_dbe_t *rls_dbe;
646 rfs4_state_t *rls_state;
647 stateid_t rls_lockid;
648 rfs4_lockowner_t *rls_locker;
649 seqid4 rls_seqid;
650 unsigned rls_skip_seqid_check:1;
651 unsigned rls_locks_cleaned:1;
652 unsigned rls_lock_completed:1;
653 rfs4_state_wait_t rls_sw;
654 list_node_t rls_node;
655 nfs_resop4 rls_reply;
656 } rfs4_lo_state_t;
657
658 /*
659 * Delegation state - per client
660 *
661 * dbe - encapsulation struct
662 * dtype - type of delegation (NONE, READ, WRITE)
663 * delegid - stateid for this delegation
664 * time_granted - time this delegation was assigned to client
665 * time_recalled - time when the server started recall process
666 * time_revoked - if revoked, time that the revoke occurred
667 * finfo - reference to the file associated with this delegation
668 * client - reference to client for which this delegation is associated
669 * node - list of delegations for the file (WRITE == 1, READ == )
670 */
671 typedef struct rfs4_deleg_state {
672 rfs4_dbe_t *rds_dbe;
673 open_delegation_type4 rds_dtype;
674 stateid_t rds_delegid;
675 time_t rds_time_granted;
676 time_t rds_time_recalled;
677 time_t rds_time_revoked;
678 struct rfs4_file *rds_finfo;
679 rfs4_client_t *rds_client;
680 list_node_t rds_node;
681 } rfs4_deleg_state_t;
682
683 /*
684 * Delegation info associated with the file
685 *
686 * dtype - type of delegation for file (NONE, READ, WRITE)
687 * time_returned - time that last delegation was returned for file
688 * time_recalled - time that recall sequence started
689 * time_lastgrant - time that last delegation was provided to a client
690 * time_lastwrite - time of last write to use the delegation stateid
691 * time_rm_delayed - time of last remove/rename which was DELAYed
692 * rdgrants - how many read delegations have been provided for this file
693 * wrgrants - how many write delegations provided (can only be one)
694 * recall_count - how many recall threads are outstanding
695 * recall_lock - lock to protect contents of this struct
696 * recall_cv - condition var for the "parent" thread to wait upon
697 * deleg_change_grant - value for change attribute at time of write grant
698 * deleg_change - most recent value of change obtained from client
699 * deleg_change_ts - time of last deleg_change update
700 * ever_recalled - has this particular delegation ever been recalled?
701 * dont_grant - file deletion is impending, don't grant a delegation
702 * conflicted_client - clientid of the client that caused a CB_RECALL
703 * to occur. This is used for delegation policy (should a delegation
704 * be granted shortly after it has been returned?)
705 */
706 typedef struct rfs4_dinfo {
707 open_delegation_type4 rd_dtype;
708 time_t rd_time_returned;
709 time_t rd_time_recalled;
710 time_t rd_time_lastgrant;
711 time_t rd_time_lastwrite;
712 time_t rd_time_rm_delayed;
713 uint32_t rd_rdgrants;
714 uint32_t rd_wrgrants;
715 int32_t rd_recall_count;
716 kmutex_t rd_recall_lock[1];
717 kcondvar_t rd_recall_cv[1];
718 bool_t rd_ever_recalled;
719 uint32_t rd_hold_grant;
720 clientid4 rd_conflicted_client;
721 } rfs4_dinfo_t;
722
723 /*
724 * File
725 *
726 * dbe - encapsulation struct
727 * vp - vnode for the file that is open or has a delegation
728 * filehandle - the filehandle generated by the server for this file
729 * delegstatelist - root of delegation list for this file
730 * dinfo - see struct definition above
731 * share_deny - union of all deny modes on file
732 * share_access - union of all access modes on file
733 * access_read - count of read access
734 * access_write - count of write access
735 * deny_read - count of deny reads
736 * deny_write - count of deny writes
737 * file_rwlock - lock for serializing the removal of a file while
738 * the state structures are active within the server
739 *
740 * The only requirement for locking file_rwlock is that the
741 * caller have a reference to the containing rfs4_file. The dbe
742 * lock may or may not be held for lock/unlock of file_rwlock.
743 * As mentioned above, the file_rwlock is used for serialization
744 * of file removal and more specifically reference to the held
745 * vnode (e.g. vp).
746 */
747 typedef struct rfs4_file {
748 rfs4_dbe_t *rf_dbe;
749 vnode_t *rf_vp;
750 nfs_fh4 rf_filehandle;
751 list_t rf_delegstatelist;
752 rfs4_dinfo_t rf_dinfo;
753 uint32_t rf_share_deny;
754 uint32_t rf_share_access;
755 uint32_t rf_access_read;
756 uint32_t rf_access_write;
757 uint32_t rf_deny_read;
758 uint32_t rf_deny_write;
759 krwlock_t rf_file_rwlock;
760 } rfs4_file_t;
761
762 /*
763 * nfs4_deleg_policy is used to signify the server's global delegation
764 * policy. The default is to NEVER delegate files and the
765 * administrator must configure the server to enable delegations.
766 *
767 * The disable/enable delegation functions are used to eliminate a
768 * race with exclusive creates.
769 */
770 typedef enum {
771 SRV_NEVER_DELEGATE = 0,
772 SRV_NORMAL_DELEGATE = 1
773 } srv_deleg_policy_t;
774
775 extern void rfs4_disable_delegation(void), rfs4_enable_delegation(void);
776
777 /*
778 * Request types for delegation. These correspond with
779 * open_delegation_type4 with the addition of a new value, DELEG_ANY,
780 * to reqequest any delegation.
781 */
782 typedef enum {
783 DELEG_NONE = 0, /* Corresponds to OPEN_DELEG_NONE */
784 DELEG_READ = 1, /* Corresponds to OPEN_DELEG_READ */
785 DELEG_WRITE = 2, /* Corresponds to OPEN_DELEG_WRITE */
786 DELEG_ANY = -1 /* New value to request any delegation type */
787 } delegreq_t;
788
789 #define NFS4_DELEG4TYPE2REQTYPE(x) (delegreq_t)(x)
790
791 /*
792 * Zone global variables of NFSv4 server
793 */
794 typedef struct nfs4_srv {
795 /* Unique write verifier */
796 verifier4 write4verf;
797 /* Delegation lock */
798 kmutex_t deleg_lock;
799 /* Used to serialize create/destroy of nfs4_server_state database */
800 kmutex_t state_lock;
801 rfs4_database_t *nfs4_server_state;
802 /* Used to manage access to server instance linked list */
803 kmutex_t servinst_lock;
804 rfs4_servinst_t *nfs4_cur_servinst;
805 /* Used to manage access to nfs4_deleg_policy */
806 krwlock_t deleg_policy_lock;
807 srv_deleg_policy_t nfs4_deleg_policy;
808 /* Allowed minors, nfs4 minors start from 0, which is always allowed */
809 nfs4_minor_t nfs4_minor_max;
810 /* Set first time we see one */
811 int seen_first_compound;
812 /*
813 * Circular double-linked list of paths for currently-served RGs.
814 * No locking required -- only changed on server start.
815 * Managed with insque/remque.
816 */
817 rfs4_dss_path_t *dss_pathlist;
818 /* Duplicate request cache */
819 struct rfs4_drc *nfs4_drc;
820 /* nfsv4 server start time */
821 time_t rfs4_start_time;
822 /* Used to serialize lookups of clientids */
823 krwlock_t rfs4_findclient_lock;
824
825 /* NFSv4 server state client tables */
826 /* table expiry times */
827 time_t rfs4_client_cache_time;
828 time_t rfs4_openowner_cache_time;
829 time_t rfs4_state_cache_time;
830 time_t rfs4_lo_state_cache_time;
831 time_t rfs4_lockowner_cache_time;
832 time_t rfs4_file_cache_time;
833 time_t rfs4_deleg_state_cache_time;
834 time_t rfs4_clntip_cache_time;
835 /* tables and indexes */
836 /* client table */
837 rfs4_table_t *rfs4_client_tab;
838 rfs4_index_t *rfs4_clientid_idx;
839 rfs4_index_t *rfs4_nfsclnt_idx;
840 /* client IP table */
841 rfs4_table_t *rfs4_clntip_tab;
842 rfs4_index_t *rfs4_clntip_idx;
843 /* Open Owner table */
844 rfs4_table_t *rfs4_openowner_tab;
845 rfs4_index_t *rfs4_openowner_idx;
846 /* Open State ID table */
847 rfs4_table_t *rfs4_state_tab;
848 rfs4_index_t *rfs4_state_idx;
849 rfs4_index_t *rfs4_state_owner_file_idx;
850 rfs4_index_t *rfs4_state_file_idx;
851 /* Lock State ID table */
852 rfs4_table_t *rfs4_lo_state_tab;
853 rfs4_index_t *rfs4_lo_state_idx;
854 rfs4_index_t *rfs4_lo_state_owner_idx;
855 /* Lock owner table */
856 rfs4_table_t *rfs4_lockowner_tab;
857 rfs4_index_t *rfs4_lockowner_idx;
858 rfs4_index_t *rfs4_lockowner_pid_idx;
859 /* File table */
860 rfs4_table_t *rfs4_file_tab;
861 rfs4_index_t *rfs4_file_idx;
862 /* Deleg State table */
863 rfs4_table_t *rfs4_deleg_state_tab;
864 rfs4_index_t *rfs4_deleg_idx;
865 rfs4_index_t *rfs4_deleg_state_idx;
866
867 /* nfs4.x */
868 rfs4_table_t *rfs4_session_tab;
869 rfs4_index_t *rfs4_session_idx;
870
871 /* client stable storage */
872 int rfs4_ss_enabled;
873 } nfs4_srv_t;
874
875 /*
876 * max length of the NFSv4 server database name
877 */
878 #define RFS4_MAX_MEM_CACHE_NAME 48
879
880 /*
881 * global NFSv4 server kmem caches
882 * r_db_name - The name of the state database and the table that will use it
883 * These tables are defined in nfs4_srv_t
884 * r_db_mem_cache - The kmem cache associated with the state database name
885 */
886 typedef struct rfs4_db_mem_cache {
887 char r_db_name[RFS4_MAX_MEM_CACHE_NAME];
888 kmem_cache_t *r_db_mem_cache;
889 } rfs4_db_mem_cache_t;
890
891 #define RFS4_DB_MEM_CACHE_NUM 9
892
893 extern rfs4_db_mem_cache_t rfs4_db_mem_cache_table[RFS4_DB_MEM_CACHE_NUM];
894
895 extern srv_deleg_policy_t nfs4_get_deleg_policy();
896
897 extern void rfs4_servinst_create(nfs4_srv_t *, int, int, char **);
898 extern void rfs4_servinst_destroy_all(nfs4_srv_t *);
899 extern void rfs4_servinst_assign(nfs4_srv_t *, rfs4_client_t *,
900 rfs4_servinst_t *);
901 extern rfs4_servinst_t *rfs4_servinst(rfs4_client_t *);
902 extern int rfs4_clnt_in_grace(rfs4_client_t *);
903 extern int rfs4_servinst_in_grace(rfs4_servinst_t *);
904 extern int rfs4_servinst_grace_new(rfs4_servinst_t *);
905 extern void rfs4_grace_start(rfs4_servinst_t *);
906 extern void rfs4_grace_start_new(nfs4_srv_t *);
907 extern void rfs4_grace_reset_all(nfs4_srv_t *);
908 extern void rfs4_dss_readstate(nfs4_srv_t *, int, char **);
909
910 /*
911 * Various interfaces to manipulate the state structures introduced
912 * above
913 */
914 extern void rfs4_free_reply(nfs_resop4 *);
915 extern void rfs4_copy_reply(nfs_resop4 *, nfs_resop4 *);
916
917 /* rfs4_client_t handling */
918 extern rfs4_client_t *rfs4_findclient(nfs_client_id4 *,
919 bool_t *, rfs4_client_t *);
920 extern rfs4_client_t *rfs4_findclient_by_id(clientid4, bool_t);
921 extern rfs4_client_t *rfs4_findclient_by_addr(struct sockaddr *);
922 extern void rfs4_client_rele(rfs4_client_t *);
923 extern void rfs4_client_close(rfs4_client_t *);
924 extern void rfs4_client_state_remove(rfs4_client_t *);
925 extern void rfs4_client_scv_next(rfs4_client_t *);
926 extern void rfs4_update_lease(rfs4_client_t *);
927 extern bool_t rfs4_lease_expired(rfs4_client_t *);
928 extern nfsstat4 rfs4_check_clientid(clientid4 *, int);
929
930 /* rfs4_clntip_t handling */
931 extern rfs4_clntip_t *rfs4_find_clntip(struct sockaddr *, bool_t *);
932 extern void rfs4_invalidate_clntip(struct sockaddr *);
933
934 /* rfs4_openowner_t handling */
935 extern rfs4_openowner_t *rfs4_findopenowner(open_owner4 *, bool_t *, seqid4);
936 extern void rfs4_update_open_sequence(rfs4_openowner_t *);
937 extern void rfs4_update_open_resp(rfs4_openowner_t *,
938 nfs_resop4 *, nfs_fh4 *);
939 extern void rfs4_openowner_rele(rfs4_openowner_t *);
940 extern void rfs4_free_opens(rfs4_openowner_t *, bool_t, bool_t);
941
942 /* rfs4_lockowner_t handling */
943 extern rfs4_lockowner_t *rfs4_findlockowner(lock_owner4 *, bool_t *);
944 extern rfs4_lockowner_t *rfs4_findlockowner_by_pid(pid_t);
945 extern void rfs4_lockowner_rele(rfs4_lockowner_t *);
946
947 /* rfs4_state_t handling */
948 extern rfs4_state_t *rfs4_findstate_by_owner_file(rfs4_openowner_t *,
949 rfs4_file_t *, bool_t *);
950 extern void rfs4_state_rele(rfs4_state_t *);
951 extern void rfs4_state_close(rfs4_state_t *, bool_t,
952 bool_t, cred_t *);
953 extern void rfs4_release_share_lock_state(rfs4_state_t *,
954 cred_t *, bool_t);
955 extern void rfs4_close_all_state(rfs4_file_t *);
956
957 /* rfs4_lo_state_t handling */
958 extern rfs4_lo_state_t *rfs4_findlo_state_by_owner(rfs4_lockowner_t *,
959 rfs4_state_t *, bool_t *);
960 extern void rfs4_lo_state_rele(rfs4_lo_state_t *, bool_t);
961 extern void rfs4_update_lock_sequence(rfs4_lo_state_t *);
962 extern void rfs4_update_lock_resp(rfs4_lo_state_t *,
963 nfs_resop4 *);
964
965 /* rfs4_file_t handling */
966 extern rfs4_file_t *rfs4_findfile(vnode_t *, nfs_fh4 *, bool_t *);
967 extern rfs4_file_t *rfs4_findfile_withlock(vnode_t *, nfs_fh4 *,
968 bool_t *);
969 extern void rfs4_file_rele(rfs4_file_t *);
970
971 /* General collection of "get state" functions */
972 extern nfsstat4 rfs4_get_state(stateid4 *, rfs4_state_t **,
973 rfs4_dbsearch_type_t);
974 extern nfsstat4 rfs4_get_deleg_state(stateid4 *,
975 rfs4_deleg_state_t **);
976 extern nfsstat4 rfs4_get_lo_state(stateid4 *, rfs4_lo_state_t **,
977 bool_t);
978 struct compound_state;
979 extern nfsstat4 rfs4_check_stateid(int, vnode_t *, stateid4 *,
980 bool_t, bool_t *, bool_t,
981 caller_context_t *,
982 struct compound_state *);
983 extern int rfs4_check_stateid_seqid(rfs4_state_t *, stateid4 *,
984 const struct compound_state *);
985 extern int rfs4_check_lo_stateid_seqid(rfs4_lo_state_t *,
986 stateid4 *,
987 const struct compound_state *);
988
989 /* return values for rfs4_check_stateid_seqid() */
990 #define NFS4_CHECK_STATEID_OKAY 1
991 #define NFS4_CHECK_STATEID_OLD 2
992 #define NFS4_CHECK_STATEID_BAD 3
993 #define NFS4_CHECK_STATEID_EXPIRED 4
994 #define NFS4_CHECK_STATEID_REPLAY 5
995 #define NFS4_CHECK_STATEID_CLOSED 6
996 #define NFS4_CHECK_STATEID_UNCONFIRMED 7
997
998 /* delay() time that server is willing to briefly wait for a delegreturn */
999 #define NFS4_DELEGATION_CONFLICT_DELAY (hz/10)
1000
1001 /*
1002 * Interfaces for handling of callback's client handle cache and
1003 * callback interfaces themselves.
1004 */
1005 extern void rfs4_cbinfo_free(rfs4_cbinfo_t *);
1006 extern void rfs4_client_setcb(rfs4_client_t *, cb_client4 *,
1007 uint32_t);
1008 extern void rfs4_deleg_cb_check(rfs4_client_t *);
1009 extern nfsstat4 rfs4_vop_getattr(vnode_t *, vattr_t *, int, cred_t *);
1010
1011 /* rfs4_deleg_state_t handling and other delegation interfaces */
1012 extern rfs4_deleg_state_t *rfs4_finddeleg(rfs4_state_t *, bool_t *);
1013 extern rfs4_deleg_state_t *rfs4_finddelegstate(stateid_t *);
1014 extern bool_t rfs4_check_recall(rfs4_state_t *, uint32_t);
1015 extern void rfs4_recall_deleg(rfs4_file_t *,
1016 bool_t, rfs4_client_t *);
1017 extern int rfs4_get_deleg(rfs4_state_t *, open_delegation_type4,
1018 open_delegation_type4 (*policy)(rfs4_state_t *,
1019 open_delegation_type4 dtype));
1020 extern rfs4_deleg_state_t *rfs4_grant_delegation(delegreq_t, rfs4_state_t *,
1021 int *);
1022 extern void rfs4_set_deleg_response(rfs4_deleg_state_t *,
1023 open_delegation4 *, nfsace4 *, int);
1024 extern void rfs4_return_deleg(rfs4_deleg_state_t *, bool_t);
1025 extern bool_t rfs4_is_deleg(rfs4_state_t *);
1026 extern void rfs4_deleg_state_rele(rfs4_deleg_state_t *);
1027 extern bool_t rfs4_check_delegated_byfp(int, rfs4_file_t *,
1028 bool_t, bool_t, bool_t, clientid4 *);
1029 extern void rfs4_clear_dont_grant(rfs4_file_t *);
1030
1031 /*
1032 * nfs4 monitored operations.
1033 */
1034 extern int deleg_rd_open(femarg_t *, int, cred_t *, caller_context_t *);
1035 extern int deleg_wr_open(femarg_t *, int, cred_t *, caller_context_t *);
1036 extern int deleg_wr_read(femarg_t *, uio_t *, int, cred_t *,
1037 caller_context_t *);
1038 extern int deleg_rd_write(femarg_t *, uio_t *, int, cred_t *,
1039 caller_context_t *);
1040 extern int deleg_wr_write(femarg_t *, uio_t *, int, cred_t *,
1041 caller_context_t *);
1042 extern int deleg_rd_setattr(femarg_t *, vattr_t *, int, cred_t *,
1043 caller_context_t *);
1044 extern int deleg_wr_setattr(femarg_t *, vattr_t *, int, cred_t *,
1045 caller_context_t *);
1046 extern int deleg_rd_rwlock(femarg_t *, int, caller_context_t *);
1047 extern int deleg_wr_rwlock(femarg_t *, int, caller_context_t *);
1048 extern int deleg_rd_space(femarg_t *, int, flock64_t *, int, offset_t, cred_t *,
1049 caller_context_t *);
1050 extern int deleg_wr_space(femarg_t *, int, flock64_t *, int, offset_t, cred_t *,
1051 caller_context_t *);
1052 extern int deleg_rd_setsecattr(femarg_t *, vsecattr_t *, int, cred_t *,
1053 caller_context_t *);
1054 extern int deleg_wr_setsecattr(femarg_t *, vsecattr_t *, int, cred_t *,
1055 caller_context_t *);
1056 extern int deleg_rd_vnevent(femarg_t *, vnevent_t, vnode_t *, char *,
1057 caller_context_t *);
1058 extern int deleg_wr_vnevent(femarg_t *, vnevent_t, vnode_t *, char *,
1059 caller_context_t *);
1060
1061 extern void rfs4_mon_hold(void *);
1062 extern void rfs4_mon_rele(void *);
1063
1064 extern fem_t *deleg_rdops;
1065 extern fem_t *deleg_wrops;
1066
1067 extern int rfs4_share(rfs4_state_t *, uint32_t, uint32_t);
1068 extern int rfs4_unshare(rfs4_state_t *);
1069 extern void rfs4_set_deleg_policy(nfs4_srv_t *, srv_deleg_policy_t);
1070 extern void rfs4_hold_deleg_policy(nfs4_srv_t *);
1071 extern void rfs4_rele_deleg_policy(nfs4_srv_t *);
1072
1073 #ifdef DEBUG
1074 #define NFS4_DEBUG(var, args) if (var) cmn_err args
1075
1076 extern int rfs4_debug;
1077 extern int nfs4_client_attr_debug;
1078 extern int nfs4_client_state_debug;
1079 extern int nfs4_client_shadow_debug;
1080 extern int nfs4_client_lock_debug;
1081 extern int nfs4_client_lease_debug;
1082 extern int nfs4_seqid_sync;
1083 extern int nfs4_client_map_debug;
1084 extern int nfs4_client_inactive_debug;
1085 extern int nfs4_client_recov_debug;
1086 extern int nfs4_client_failover_debug;
1087 extern int nfs4_client_call_debug;
1088 extern int nfs4_client_foo_debug;
1089 extern int nfs4_client_zone_debug;
1090 extern int nfs4_lost_rqst_debug;
1091 extern int nfs4_open_stream_debug;
1092 extern int nfs4_client_open_dg;
1093 extern int nfs4_srvmnt_debug;
1094 extern int nfs4_utf8_debug;
1095
1096 void rfs4_dbe_debug(rfs4_dbe_t *e);
1097
1098 #ifdef NFS4_DEBUG_MUTEX
1099 void nfs4_debug_mutex_enter(kmutex_t *, char *, int);
1100 void nfs4_debug_mutex_exit(kmutex_t *, char *, int);
1101
1102 #define mutex_enter(m) nfs4_debug_mutex_enter((m), __FILE__, __LINE__)
1103 #define mutex_exit(m) nfs4_debug_mutex_exit((m), __FILE__, __LINE__)
1104 #endif /* NFS4_DEBUG_MUTEX */
1105
1106 #else /* ! DEBUG */
1107 #define NFS4_DEBUG(var, args)
1108 #endif /* DEBUG */
1109
1110 /*
1111 * XXX - temporary for testing of volatile fh
1112 */
1113
1114 #ifdef VOLATILE_FH_TEST
1115
1116 struct nfs_fh4_fmt {
1117 fhandle4_t fh4_i;
1118 uint32_t fh4_flag;
1119 uint32_t fh4_volatile_id;
1120 };
1121
1122 #else /* VOLATILE_FH_TEST */
1123
1124 struct nfs_fh4_fmt {
1125 fhandle4_t fh4_i;
1126 uint32_t fh4_flag;
1127 };
1128
1129 #endif /* VOLATILE_FH_TEST */
1130
1131 #define FH4_NAMEDATTR 1
1132 #define FH4_ATTRDIR 2
1133
1134 #define fh4_fsid fh4_i.fhx_fsid
1135 #define fh4_len fh4_i.fhx_len /* fid length */
1136 #define fh4_data fh4_i.fhx_data /* fid bytes */
1137 #define fh4_xlen fh4_i.fhx_xlen
1138 #define fh4_xdata fh4_i.fhx_xdata
1139 typedef struct nfs_fh4_fmt nfs_fh4_fmt_t;
1140
1141 #define fh4_to_fmt4(fh4p) ((nfs_fh4_fmt_t *)(fh4p)->nfs_fh4_val)
1142 #define get_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) & (flag))
1143 #define set_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) |= (flag))
1144 #define clr_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) &= ~(flag))
1145
1146 #define NFS_FH4_LEN sizeof (nfs_fh4_fmt_t)
1147
1148 /*
1149 * Copy fields from external (fhandle_t) to in-memory (nfs_fh4_fmt_t)
1150 * format to support export info checking. It does not copy over
1151 * the complete filehandle, just the fsid, xlen and xdata. It may
1152 * need to be changed to be used in other places.
1153 *
1154 * NOTE: The macro expects the space to be pre-allocated for
1155 * the contents of nfs_fh4_fmt_t.
1156 */
1157 #define FH_TO_FMT4(exifh, nfs_fmt) { \
1158 bzero((nfs_fmt), NFS_FH4_LEN); \
1159 (nfs_fmt)->fh4_fsid = (exifh)->fh_fsid; \
1160 (nfs_fmt)->fh4_xlen = (exifh)->fh_xlen; \
1161 bcopy((exifh)->fh_xdata, (nfs_fmt)->fh4_xdata, \
1162 (exifh)->fh_xlen); \
1163 }
1164
1165 /*
1166 * A few definitions of repeatedly used constructs for nfsv4
1167 */
1168 #define UTF8STRING_FREE(str) { \
1169 kmem_free((str).utf8string_val, (str).utf8string_len); \
1170 (str).utf8string_val = NULL; \
1171 (str).utf8string_len = 0; \
1172 }
1173
1174 /*
1175 * NFS4_VOLATILE_FH yields non-zero if the filesystem uses non-persistent
1176 * filehandles.
1177 */
1178 #define NFS4_VOLATILE_FH(mi) \
1179 ((mi)->mi_fh_expire_type & \
1180 (FH4_VOLATILE_ANY | FH4_VOL_MIGRATION | FH4_VOL_RENAME))
1181
1182 /*
1183 * NFS_IS_DOTNAME checks if the name given represents a dot or dotdot entry
1184 */
1185 #define NFS_IS_DOTNAME(name) \
1186 (((name)[0] == '.') && \
1187 (((name)[1] == '\0') || (((name)[1] == '.') && ((name)[2] == '\0'))))
1188
1189 /*
1190 * Define the number of bits in a bitmap word (uint32)
1191 */
1192 #define NFS4_BITMAP4_BITSPERWORD (sizeof (uint32_t) * 8)
1193
1194 /*
1195 * Define the value for the access field of the compound_state structure
1196 * based on the result of nfsauth access checking.
1197 */
1198 #define CS_ACCESS_OK 0x1
1199 #define CS_ACCESS_DENIED 0x2
1200 #define CS_ACCESS_LIMITED 0x4
1201
1202 /*
1203 * compound state in nfsv4 server
1204 */
1205 struct compound_state {
1206 struct exportinfo *exi;
1207 struct exportinfo *saved_exi; /* export struct for saved_vp */
1208 cred_t *basecr; /* UNIX cred: only RPC request */
1209 caddr_t principal;
1210 int nfsflavor;
1211 cred_t *cr; /* UNIX cred: RPC request and */
1212 /* target export */
1213 bool_t cont;
1214 uint_t access; /* access perm on vp per request */
1215 bool_t deleg; /* TRUE if current fh has */
1216 /* write delegated */
1217 vnode_t *vp; /* modified by PUTFH, and by ops that */
1218 /* input to GETFH */
1219 bool_t mandlock; /* Is mandatory locking in effect */
1220 /* for vp */
1221 vnode_t *saved_vp; /* modified by SAVEFH, copied to */
1222 /* vp by RESTOREFH */
1223 nfsstat4 *statusp;
1224 nfs_fh4 fh; /* ditto. valid only if vp != NULL */
1225 nfs_fh4 saved_fh; /* ditto. valid only if */
1226 /* saved_vp != NULL */
1227 struct svc_req *req;
1228 char fhbuf[NFS4_FHSIZE];
1229
1230 /* NFSv4.1 */
1231 uint8_t minorversion; /* NFS4 minor version */
1232 rfs4_session_t *sp; /* OP_SEQUENCE set it */
1233 slotid4 slotno;
1234 rfs4_slot_t *slot;
1235 rfs4_client_t *client;
1236 uint16_t op_pos;
1237 uint16_t op_len; /* number operations in compound req */
1238 #define RFS4_DISPATCH_DONE (1 << 0)
1239 uint8_t cs_flags;
1240 bool_t cachethis;
1241 COMPOUND4res *cmpresp;
1242 };
1243
1244 typedef struct compound_state compound_state_t;
1245
1246 static inline bool_t
rfs4_has_session(const compound_state_t * cs)1247 rfs4_has_session(const compound_state_t *cs)
1248 {
1249 return (cs->slot != NULL);
1250 }
1251
1252 /*
1253 * Conversion commands for nfsv4 server attr checking
1254 */
1255 enum nfs4_attr_cmd {
1256 NFS4ATTR_SUPPORTED = 0, /* check which attrs supported */
1257 NFS4ATTR_GETIT = 1, /* getattr - sys to fattr4 (r) */
1258 NFS4ATTR_SETIT = 2, /* setattr - fattr4 to sys (w) */
1259 NFS4ATTR_VERIT = 3, /* verify - fattr4 to sys (r) */
1260 NFS4ATTR_FREEIT = 4 /* free any alloc'd space for attr */
1261 };
1262
1263 typedef enum nfs4_attr_cmd nfs4_attr_cmd_t;
1264
1265 struct nfs4_svgetit_arg {
1266 nfs4_attr_cmd_t op; /* getit or setit */
1267 struct compound_state *cs;
1268 struct statvfs64 *sbp;
1269 uint_t flag; /* VOP_GETATTR/VOP_SETATTR flag */
1270 uint_t xattr; /* object is xattr */
1271 bool_t rdattr_error_req; /* if readdir & client wants */
1272 /* rdattr_error */
1273 nfsstat4 rdattr_error; /* used for per-entry status */
1274 /* (if rdattr_err) */
1275 bool_t is_referral; /* because sometimes we tell lies */
1276 bool_t mntdfid_set;
1277
1278 fattr4_mounted_on_fileid
1279 mounted_on_fileid;
1280 /* readdir op can always return */
1281 /* d_ino from server fs dirent */
1282 /* for mounted_on_fileid attr. */
1283 /* This field holds d_ino so */
1284 /* srv attr conv code can avoid */
1285 /* doing an untraverse. */
1286 vattr_t vap[1];
1287 };
1288
1289 struct nfs4_ntov_map {
1290 bitmap4 fbit; /* FATTR4_XXX_MASKY */
1291 uint_t vbit; /* AT_XXX */
1292 bool_t vfsstat;
1293 bool_t mandatory; /* attribute mandatory to implement? */
1294 uint_t nval;
1295 int xdr_size; /* Size of XDR'd attr */
1296 xdrproc_t xfunc;
1297 int (*sv_getit)(nfs4_attr_cmd_t, struct nfs4_svgetit_arg *,
1298 union nfs4_attr_u *); /* subroutine for getting attr. */
1299 char *prtstr; /* string attr for printing */
1300 };
1301
1302 struct nfs4attr_to_vattr {
1303 vnode_t *vp;
1304 vattr_t *vap;
1305 nfs_fh4 *fhp;
1306 nfsstat4 rdattr_error;
1307 uint32_t flag;
1308 fattr4_change change;
1309 fattr4_fsid srv_fsid;
1310 fattr4_mounted_on_fileid mntd_fid;
1311 };
1312
1313 typedef struct nfs4attr_to_vattr ntov4_t;
1314
1315 /*
1316 * nfs4attr_to_vattr flags
1317 */
1318 #define NTOV_FHP_VALID 0x01
1319 #define NTOV_RDATTR_ERROR_VALID 0x02
1320 #define NTOV_CHANGE_VALID 0x04
1321 #define NTOV_SUPP_VALID 0x08
1322 #define NTOV_SRV_FSID_VALID 0x10
1323 #define NTOV_MOUNTED_ON_FILEID_VALID 0x20
1324
1325
1326 #define FATTR4_MANDATTR_MASK ( \
1327 FATTR4_SUPPORTED_ATTRS_MASK | \
1328 FATTR4_TYPE_MASK | \
1329 FATTR4_FH_EXPIRE_TYPE_MASK | \
1330 FATTR4_CHANGE_MASK | \
1331 FATTR4_SIZE_MASK | \
1332 FATTR4_LINK_SUPPORT_MASK | \
1333 FATTR4_SYMLINK_SUPPORT_MASK | \
1334 FATTR4_NAMED_ATTR_MASK | \
1335 FATTR4_FSID_MASK | \
1336 FATTR4_UNIQUE_HANDLES_MASK | \
1337 FATTR4_LEASE_TIME_MASK | \
1338 FATTR4_RDATTR_ERROR_MASK | \
1339 FATTR4_FILEHANDLE_MASK)
1340
1341
1342 struct nfs4attr_to_osattr {
1343 void *attrconv_arg;
1344 uint_t mask;
1345 };
1346
1347 struct mntinfo4;
1348
1349 /*
1350 * lkp4_attr_setup lists the different options for attributes when calling
1351 * nfs4lookup_setup - either no attributes (just lookups - e.g., secinfo),
1352 * one component only (normal component lookup), get attributes for the
1353 * last component (e.g., mount), attributes for each component (e.g.,
1354 * failovers later), just the filehandle for the last component (e.g.,
1355 * volatile filehandle recovery), or stuff that needs OPENATTR (e.g.
1356 * looking up a named attribute or it's hidden directory).
1357 */
1358 enum lkp4_attr_setup {
1359 LKP4_NO_ATTRIBUTES = 0, /* no attrs or filehandles */
1360 LKP4_ALL_ATTRIBUTES = 3, /* multi-comp: attrs for all comps */
1361 LKP4_LAST_NAMED_ATTR = 5, /* multi-comp: named attr & attrdir */
1362 LKP4_LAST_ATTRDIR = 6, /* multi-comp: just attrdir */
1363 LKP4_ALL_ATTR_SECINFO = 7 /* multi-comp: attrs for all comp and */
1364 /* secinfo for last comp */
1365 };
1366
1367 /*
1368 * lookup4_param a set of parameters to nfs4lookup_setup -
1369 * used to setup a path lookup compound request.
1370 */
1371 typedef struct lookup4_param {
1372 enum lkp4_attr_setup l4_getattrs; /* (in) get attrs in the lookup? */
1373 int header_len; /* (in) num ops before first lookup */
1374 int trailer_len; /* (in) num ops after last */
1375 /* Lookup/Getattr */
1376 bitmap4 ga_bits; /* (in) Which attributes for Getattr */
1377 COMPOUND4args_clnt *argsp; /* (in/out) args for compound struct */
1378 COMPOUND4res_clnt *resp; /* (in/out) res for compound struct */
1379 int arglen; /* (out) argop buffer alloc'd length */
1380 struct mntinfo4 *mi;
1381 } lookup4_param_t;
1382
1383
1384 #define NFS4_FATTR4_FINISH -1 /* fattr4 index indicating finish */
1385
1386 typedef int (*nfs4attr_to_os_t)(int, union nfs4_attr_u *,
1387 struct nfs4attr_to_osattr *);
1388
1389 /*
1390 * The nfs4_error_t is the basic structure to return error values
1391 * from rfs4call. It encapsulates the unix errno
1392 * value, the nfsstat4 value and the rpc status value into a single
1393 * structure.
1394 *
1395 * If error is set, then stat is ignored and rpc_status may be
1396 * set if the error occurred as the result of a CLNT_CALL. If
1397 * stat is set, then rpc request succeeded, error and
1398 * rpc_status are set to 0 and stat contains the result of
1399 * operation, NFS4_OK or one of the NFS4ERR_* values.
1400 *
1401 * Functions which want to generate errors independently from
1402 * rfs4call should set error to the desired errno value and
1403 * set stat and rpc_status to 0. nfs4_error_init() is a
1404 * convenient function to do this.
1405 */
1406 typedef struct {
1407 int error;
1408 nfsstat4 stat;
1409 enum clnt_stat rpc_status;
1410 } nfs4_error_t;
1411
1412 /*
1413 * Shared functions
1414 */
1415 extern void rfs4_op_readdir(nfs_argop4 *, nfs_resop4 *,
1416 struct svc_req *, struct compound_state *);
1417 extern void nfs_fh4_copy(nfs_fh4 *, nfs_fh4 *);
1418
1419 extern void nfs4_fattr4_free(fattr4 *);
1420
1421 extern int nfs4lookup_setup(char *, lookup4_param_t *, int);
1422 extern void nfs4_getattr_otw_norecovery(vnode_t *,
1423 nfs4_ga_res_t *, nfs4_error_t *, cred_t *, int);
1424 extern int nfs4_getattr_otw(vnode_t *, nfs4_ga_res_t *, cred_t *, int);
1425 extern int nfs4cmpfh(const nfs_fh4 *, const nfs_fh4 *);
1426 extern int nfs4cmpfhandle(nfs4_fhandle_t *, nfs4_fhandle_t *);
1427 extern int nfs4getattr(vnode_t *, struct vattr *, cred_t *);
1428 extern int nfs4_waitfor_purge_complete(vnode_t *);
1429 extern int nfs4_validate_caches(vnode_t *, cred_t *);
1430 extern int nfs4init(int, char *);
1431 extern void nfs4fini(void);
1432 extern int nfs4_vfsinit(void);
1433 extern void nfs4_vfsfini(void);
1434
1435 extern void nfs4_vnops_init(void);
1436 extern void nfs4_vnops_fini(void);
1437 extern void nfs_idmap_init(void);
1438 extern void nfs_idmap_flush(int);
1439 extern void nfs_idmap_fini(void);
1440 extern int nfs4_rnode_init(void);
1441 extern int nfs4_rnode_fini(void);
1442 extern int nfs4_shadow_init(void);
1443 extern int nfs4_shadow_fini(void);
1444 extern int nfs4_acache_init(void);
1445 extern int nfs4_acache_fini(void);
1446 extern int nfs4_subr_init(void);
1447 extern int nfs4_subr_fini(void);
1448 extern void nfs4_acl_init(void);
1449 extern void nfs4_acl_free_cache(vsecattr_t *);
1450
1451 extern int geterrno4(nfsstat4);
1452 extern nfsstat4 puterrno4(int);
1453 extern int nfs4_need_to_bump_seqid(COMPOUND4res_clnt *);
1454 extern int nfs4tsize(void);
1455 extern int checkauth4(struct compound_state *, struct svc_req *);
1456 extern nfsstat4 call_checkauth4(struct compound_state *, struct svc_req *);
1457 extern int is_exported_sec(int, struct exportinfo *);
1458 extern void nfs4_vmask_to_nmask(uint_t, bitmap4 *);
1459 extern void nfs4_vmask_to_nmask_set(uint_t, bitmap4 *);
1460 extern int nfs_idmap_str_uid(utf8string *u8s, uid_t *, bool_t);
1461 extern int nfs_idmap_str_gid(utf8string *u8s, gid_t *, bool_t);
1462 extern int nfs_idmap_uid_str(uid_t, utf8string *u8s, bool_t);
1463 extern int nfs_idmap_gid_str(gid_t gid, utf8string *u8s, bool_t);
1464 extern int nfs4_time_ntov(nfstime4 *, timestruc_t *);
1465 extern int nfs4_time_vton(timestruc_t *, nfstime4 *);
1466 extern char *utf8_to_str(utf8string *, uint_t *, char *);
1467 extern char *utf8_to_fn(utf8string *, uint_t *, char *);
1468 extern utf8string *str_to_utf8(char *, utf8string *);
1469 extern utf8string *utf8_copy(utf8string *, utf8string *);
1470 extern int utf8_compare(const utf8string *, const utf8string *);
1471 extern nfsstat4 utf8_dir_verify(utf8string *);
1472 extern char *utf8_strchr(utf8string *, const char);
1473 extern int ln_ace4_cmp(nfsace4 *, nfsace4 *, int);
1474 extern int vs_aent_to_ace4(vsecattr_t *, vsecattr_t *, int, int);
1475 extern int vs_ace4_to_aent(vsecattr_t *, vsecattr_t *, uid_t, gid_t,
1476 int, int);
1477 extern int vs_ace4_to_acet(vsecattr_t *, vsecattr_t *, uid_t, gid_t,
1478 int);
1479 extern int vs_acet_to_ace4(vsecattr_t *, vsecattr_t *, int);
1480 extern void vs_acet_destroy(vsecattr_t *);
1481 extern void vs_ace4_destroy(vsecattr_t *);
1482 extern void vs_aent_destroy(vsecattr_t *);
1483
1484 extern int vn_find_nfs_record(vnode_t *, nvlist_t **, char **, char **);
1485 extern int vn_is_nfs_reparse(vnode_t *, cred_t *);
1486 extern fs_locations4 *fetch_referral(vnode_t *, cred_t *);
1487 extern char *build_symlink(vnode_t *, cred_t *, size_t *);
1488
1489 extern int stateid4_cmp(stateid4 *, stateid4 *);
1490
1491 extern vtype_t nf4_to_vt[];
1492
1493 extern struct nfs4_ntov_map nfs4_ntov_map[];
1494 extern uint_t nfs4_ntov_map_size;
1495
1496 extern struct vfsops *nfs4_vfsops;
1497 extern struct vnodeops *nfs4_vnodeops;
1498 extern const struct fs_operation_def nfs4_vnodeops_template[];
1499 extern vnodeops_t *nfs4_trigger_vnodeops;
1500 extern const struct fs_operation_def nfs4_trigger_vnodeops_template[];
1501
1502 extern uint_t nfs4_tsize(struct knetconfig *);
1503 extern uint_t rfs4_tsize(struct svc_req *);
1504
1505 extern bool_t xdr_inline_decode_nfs_fh4(uint32_t *, nfs_fh4_fmt_t *,
1506 uint32_t);
1507 extern bool_t xdr_inline_encode_nfs_fh4(uint32_t **, uint32_t *,
1508 nfs_fh4_fmt_t *);
1509
1510 #ifdef DEBUG
1511 extern int rfs4_do_pre_op_attr;
1512 extern int rfs4_do_post_op_attr;
1513 #endif
1514
1515 extern stateid4 clnt_special0;
1516 extern stateid4 clnt_special1;
1517 #define CLNT_ISSPECIAL(id) (stateid4_cmp(id, &clnt_special0) || \
1518 stateid4_cmp(id, &clnt_special1))
1519
1520 /* State's functions */
1521 extern void rfs4_ss_clid(nfs4_srv_t *nsrv4, rfs4_client_t *);
1522 extern void rfs4_ss_chkclid(nfs4_srv_t *nsrv4, rfs4_client_t *);
1523
1524 /* Declarations for nfs4.x */
1525 nfsstat4 do_rfs4_op_secinfo(struct compound_state *, char *, SECINFO4res *);
1526
1527 /*
1528 * The NFS Version 4 service procedures.
1529 */
1530
1531 extern void rfs4_do_server_start(int, int, nfs4_minor_t, int);
1532 extern void rfs4_compound(COMPOUND4args *, COMPOUND4res *,
1533 compound_state_t *, struct svc_req *, int *);
1534 extern void rfs4_init_compound_state(struct compound_state *);
1535 extern void rfs4_fini_compound_state(struct compound_state *);
1536
1537 struct rpcdisp;
1538 extern int rfs4_dispatch(struct rpcdisp *, struct svc_req *, SVCXPRT *, char *);
1539 extern void rfs4_compound_free(COMPOUND4res *);
1540 extern void rfs4_compound_flagproc(COMPOUND4args *, int *);
1541
1542 extern void rfs4_srvrinit(void);
1543 extern void rfs4_srvrfini(void);
1544 extern void rfs4_srv_zone_init(nfs_globals_t *);
1545 extern void rfs4_srv_zone_fini(nfs_globals_t *);
1546 extern void rfs4_state_g_init(void);
1547 extern void rfs4_state_zone_init(nfs4_srv_t *);
1548 extern void rfs4_state_g_fini(void);
1549 extern void rfs4_state_zone_fini(void);
1550 extern nfs4_srv_t *nfs4_get_srv(void);
1551
1552 #endif
1553 #ifdef __cplusplus
1554 }
1555 #endif
1556
1557 #endif /* _NFS4_H */
1558