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 <nfs/nfs4_drc.h> 44 #include <sys/nvpair.h> 45 #else 46 #include <rpcsvc/nfs4_prot.h> 47 #endif 48 #include <nfs/nfs4_attr.h> 49 #include <sys/acl.h> 50 #include <sys/list.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 #define NFS4_MINORVERSION 0 78 #define CB4_MINORVERSION 0 79 80 /* 81 * Set the fattr4_change variable using a time struct. Note that change 82 * is 64 bits, but timestruc_t is 128 bits in a 64-bit kernel. 83 */ 84 #define NFS4_SET_FATTR4_CHANGE(change, ts) \ 85 { \ 86 change = (ts).tv_sec; \ 87 change <<= 32; \ 88 change |= (uint32_t)((ts).tv_nsec); \ 89 } 90 91 /* 92 * Server lease period. Value is in seconds; Also used for grace period 93 */ 94 extern time_t rfs4_lease_time; 95 96 /* 97 * This set of typedefs and interfaces represent the core or base set 98 * of functionality that backs the NFSv4 server's state related data 99 * structures. Since the NFSv4 server needs inter-RPC state to be 100 * available that is unrelated to the filesystem (in other words, 101 * soft-state), this functionality is needed to maintain that and is 102 * written to be somewhat flexible to adapt to the various types of 103 * data structures contained within the server. 104 * 105 * The basic structure at this level is that the server maintains a 106 * global "database" which consists of a set of tables. Each table 107 * contains a set of like data structures. Each table is indexed by 108 * at least one hash function and in most cases two hashes. Each 109 * table's characteristics is set when it is created at run-time via 110 * rfs4_table_create(). All table creation and related functions are 111 * located in nfs4_state.c. The generic database functionality is 112 * located in nfs4_db.c. 113 */ 114 115 typedef struct rfs4_dbe rfs4_dbe_t; /* basic opaque db entry */ 116 typedef struct rfs4_table rfs4_table_t; /* basic table type */ 117 typedef struct rfs4_index rfs4_index_t; /* index */ 118 typedef struct rfs4_database rfs4_database_t; /* and database */ 119 120 typedef struct { /* opaque entry type for later use */ 121 rfs4_dbe_t *dbe; 122 } *rfs4_entry_t; 123 124 /* 125 * NFSv4 server state databases 126 * 127 * Initialized when the module is loaded and used by NFSv4 state tables. 128 * These kmem_cache free pools are used globally, the NFSv4 state tables 129 * which make use of these kmem_cache free pools are per zone. 130 */ 131 extern kmem_cache_t *rfs4_client_mem_cache; 132 extern kmem_cache_t *rfs4_clntIP_mem_cache; 133 extern kmem_cache_t *rfs4_openown_mem_cache; 134 extern kmem_cache_t *rfs4_openstID_mem_cache; 135 extern kmem_cache_t *rfs4_lockstID_mem_cache; 136 extern kmem_cache_t *rfs4_lockown_mem_cache; 137 extern kmem_cache_t *rfs4_file_mem_cache; 138 extern kmem_cache_t *rfs4_delegstID_mem_cache; 139 140 /* database, table, index creation entry points */ 141 extern rfs4_database_t *rfs4_database_create(uint32_t); 142 extern void rfs4_database_shutdown(rfs4_database_t *); 143 extern void rfs4_database_destroy(rfs4_database_t *); 144 145 extern void rfs4_database_destroy(rfs4_database_t *); 146 147 extern kmem_cache_t *nfs4_init_mem_cache(char *, uint32_t, uint32_t, 148 uint32_t); 149 extern rfs4_table_t *rfs4_table_create(rfs4_database_t *, char *, 150 time_t, uint32_t, 151 bool_t (*create)(rfs4_entry_t, void *), 152 void (*destroy)(rfs4_entry_t), 153 bool_t (*expiry)(rfs4_entry_t), 154 uint32_t, uint32_t, uint32_t, id_t); 155 extern void rfs4_table_destroy(rfs4_database_t *, rfs4_table_t *); 156 extern rfs4_index_t *rfs4_index_create(rfs4_table_t *, char *, 157 uint32_t (*hash)(void *), 158 bool_t (compare)(rfs4_entry_t, void *), 159 void *(*mkkey)(rfs4_entry_t), bool_t); 160 extern void rfs4_index_destroy(rfs4_index_t *); 161 162 /* Type used to direct rfs4_dbsearch() in what types of records to inspect */ 163 typedef enum {RFS4_DBS_VALID, RFS4_DBS_INVALID} rfs4_dbsearch_type_t; 164 /* search and db entry manipulation entry points */ 165 extern rfs4_entry_t rfs4_dbsearch(rfs4_index_t *, void *, 166 bool_t *, void *, rfs4_dbsearch_type_t); 167 extern void rfs4_dbe_lock(rfs4_dbe_t *); 168 extern void rfs4_dbe_unlock(rfs4_dbe_t *); 169 extern clock_t rfs4_dbe_twait(rfs4_dbe_t *, clock_t); 170 extern void rfs4_dbe_cv_broadcast(rfs4_dbe_t *); 171 extern void rfs4_dbe_hold(rfs4_dbe_t *); 172 extern void rfs4_dbe_hold_nolock(rfs4_dbe_t *); 173 extern void rfs4_dbe_rele_nolock(rfs4_dbe_t *); 174 extern void rfs4_dbe_rele(rfs4_dbe_t *); 175 extern uint32_t rfs4_dbe_refcnt(rfs4_dbe_t *); 176 extern id_t rfs4_dbe_getid(rfs4_dbe_t *); 177 extern void rfs4_dbe_invalidate(rfs4_dbe_t *); 178 extern bool_t rfs4_dbe_is_invalid(rfs4_dbe_t *); 179 extern time_t rfs4_dbe_get_timerele(rfs4_dbe_t *); 180 extern void rfs4_dbe_hide(rfs4_dbe_t *); 181 extern void rfs4_dbe_unhide(rfs4_dbe_t *); 182 #ifdef DEBUG 183 extern bool_t rfs4_dbe_islocked(rfs4_dbe_t *); 184 #endif 185 extern void rfs4_dbe_walk(rfs4_table_t *, 186 void (*callout)(rfs4_entry_t, void *), void *); 187 188 /* 189 * Minimal server stable storage. 190 * 191 * Currently the NFSv4 server will only save the client 192 * ID (the long version) so that it will be able to 193 * grant possible reclaim requests during the infamous 194 * grace_period. 195 */ 196 197 #define RFS4_SS_DIRSIZE 64 * 1024 198 #define NFS4_SS_VERSION 1 199 200 /* handy pathname structure */ 201 typedef struct ss_pn { 202 char *leaf; 203 char pn[MAXPATHLEN]; 204 } rfs4_ss_pn_t; 205 206 /* 207 * The server will build this link list on startup. It represents the 208 * clients that have had valid state on the server in a prior instance. 209 * 210 */ 211 typedef struct rfs4_oldstate { 212 struct rfs4_oldstate *next; 213 struct rfs4_oldstate *prev; 214 rfs4_ss_pn_t *ss_pn; 215 nfs_client_id4 cl_id4; 216 } rfs4_oldstate_t; 217 218 /* 219 * This union is used to overlay the server's internal treatment of 220 * the protocols stateid4 datatype. Therefore, "bits" must not exceed 221 * the size of stateid4 and more importantly should match the size of 222 * stateid4. The chgseq field must the first entry since it overlays 223 * stateid4.seqid. 224 */ 225 typedef union { 226 stateid4 stateid; 227 struct { 228 uint32_t chgseq; /* State changes / protocol's seqid */ 229 uint32_t boottime; /* boot time */ 230 uint32_t type:2; /* stateid_type_t as define below */ 231 uint32_t clnodeid:8; /* cluster server nodeid */ 232 uint32_t ident:22; /* 2^22-1 openowner x fhs */ 233 pid_t pid; /* pid of corresponding lock owner */ 234 } bits; 235 } stateid_t; 236 /* 237 * Note that the way the type field above is defined, this enum must 238 * not have more than 4 members. 239 */ 240 typedef enum {OPENID, LOCKID, DELEGID} stateid_type_t; 241 242 243 /* 244 * Set of RPC credentials used for a particular operation. 245 * Used for operations like SETCLIENTID_CONFIRM where the 246 * credentials needs to match those used at SETCLIENTID. 247 */ 248 typedef void *cred_set_t; /* For now XXX */ 249 250 /* 251 * "wait" struct for use in the open open and lock owner state 252 * structures to provide serialization between server threads that are 253 * handling requests for the same open owner or lock stateid. This 254 * way only one thread will be updating things like sequence ids, 255 * replay cache and stateid at a time. 256 */ 257 typedef struct rfs4_state_wait { 258 uint32_t sw_active; 259 uint32_t sw_wait_count; 260 kmutex_t sw_cv_lock[1]; 261 kcondvar_t sw_cv[1]; 262 } rfs4_state_wait_t; 263 264 extern void rfs4_sw_enter(rfs4_state_wait_t *); 265 extern void rfs4_sw_exit(rfs4_state_wait_t *); 266 267 /* 268 * This enum and the following rfs4_cbinfo_t struct are used to 269 * maintain information about the callback path used from the server 270 * to client for operations like CB_GETATTR and CB_RECALL. The 271 * rfs4_cbinfo_t struct is meant to be encompassed in the client 272 * struct and managed within that structure's locking scheme. 273 * 274 * The various states of the callback path are used by the server to 275 * determine if delegations should initially be provided to a client 276 * and then later on if connectivity has been lost and delegations 277 * should be revoked. 278 */ 279 280 /* 281 * CB_NOCHANGE - Special value used for interfaces within the delegation 282 * code to signify that "no change" has occurred to the 283 * callback path 284 * CB_UNINIT - No callback info provided by the client 285 * CB_NONE - Callback info provided but CB_NULL call 286 * has yet to be attempted 287 * CB_OK - Callback path tested with CB_NULL with success 288 * CB_INPROG - Callback path currently being tested with CB_NULL 289 * CB_FAILED - Callback path was == CB_OK but has failed 290 * with timeout/rpc error 291 * CB_BAD - Callback info provided but CB_NULL failed 292 */ 293 typedef enum { 294 CB_NOCHANGE = 0, 295 CB_UNINIT = 1, 296 CB_NONE = 2, 297 CB_OK = 3, 298 CB_INPROG = 4, 299 CB_FAILED = 5, 300 CB_BAD = 6 301 } rfs4_cbstate_t; 302 303 #define RFS4_CBCH_MAX 10 /* size callback client handle cache */ 304 /* 305 * Callback info for a client. 306 * Client only provides: cb_client4 and cb_ident 307 * The rest of the information is used to track callback path status 308 * and usage. 309 * 310 * cb_state - used as comments for the rfs4_cbstate_t enum indicate 311 * cb_notified_of_cb_path_down - if the callback path was once CB_OK and 312 * has hence CB_FAILED, the client needs to be notified via RENEW. 313 * cb_timefailed - current time when cb_state transitioned from 314 * CB_OK -> CB_FAILED. Meant for observability. When did that happen? 315 * cb_chc_free/cb_chc - cache of client handles for the callback path 316 * cb_ident - SETCLIENTID provided callback_ident value 317 * callback - SETCLIENTID provided cb_client4 value 318 * cb_refcnt - current number of users of this structure's content 319 * protected by cb_lock 320 * cb_badbehavior - how many times did a client do something we didn't like? 321 * cb_lock - lock for contents of cbinfo 322 * cb_cv - used to allow threads to wait on CB_NULL completion 323 * cb_nullcaller - is there a thread currently taking care of 324 * new callback information? 325 * cb_cv_nullcaller - used by the thread doing CB_NULL to wait on 326 * threads that may be using client handles of the current 327 * client handle cache. 328 * newer - new callback info provided by a client and awaiting 329 * CB_NULL testing and move to regular cbinfo. 330 */ 331 typedef struct { 332 rfs4_cbstate_t cb_state; 333 unsigned cb_notified_of_cb_path_down:1; 334 time_t cb_timefailed; 335 int cb_chc_free; 336 CLIENT *cb_chc[RFS4_CBCH_MAX]; 337 uint32_t cb_ident; 338 cb_client4 cb_callback; 339 uint32_t cb_refcnt; 340 uint32_t cb_badbehavior; 341 kmutex_t cb_lock[1]; 342 kcondvar_t cb_cv[1]; 343 bool_t cb_nullcaller; 344 kcondvar_t cb_cv_nullcaller[1]; 345 struct { 346 bool_t cb_new; 347 bool_t cb_confirmed; 348 uint32_t cb_ident; 349 cb_client4 cb_callback; 350 } cb_newer; 351 } rfs4_cbinfo_t; 352 353 /* 354 * A server instance. We can associate sets of clients - via a pointer in 355 * rfs4_client_t - with a given server instance, allowing us to treat clients 356 * in the set differently to clients in other sets. 357 * 358 * Currently used only for Sun Cluster HA-NFS support, to group clients 359 * on NFS resource failover so each set of clients gets its own dedicated 360 * grace period and distributed stable storage data. 361 */ 362 typedef struct rfs4_servinst { 363 int dss_npaths; 364 krwlock_t rwlock; 365 krwlock_t oldstate_lock; 366 time_t start_time; 367 time_t grace_period; 368 rfs4_oldstate_t *oldstate; 369 struct rfs4_dss_path **dss_paths; 370 struct rfs4_servinst *next; 371 struct rfs4_servinst *prev; 372 } rfs4_servinst_t; 373 374 /* 375 * DSS: distributed stable storage 376 */ 377 378 typedef struct rfs4_dss_path { 379 struct rfs4_dss_path *next; /* for insque/remque */ 380 struct rfs4_dss_path *prev; /* for insque/remque */ 381 char *path; 382 struct rfs4_servinst *sip; 383 unsigned index; /* offset in servinst's array */ 384 } rfs4_dss_path_t; 385 386 /* array of paths passed-in from nfsd command-line; stored in nvlist */ 387 char **rfs4_dss_newpaths; 388 uint_t rfs4_dss_numnewpaths; 389 390 /* nvlists of all DSS paths: current, and before last warmstart */ 391 nvlist_t *rfs4_dss_paths, *rfs4_dss_oldpaths; 392 393 /* 394 * The server maintains a set of state on a per client basis that 395 * matches that of the protocol requirements. A client's state is 396 * rooted with the rfs4_client_t struct of which there is one per 397 * client and is created when SETCLIENTID/SETCLIENTID_CONFIRM are 398 * received. From there, the server then creates rfs4_openowner_t 399 * structs for each new open owner from that client and are initiated 400 * at OPEN/OPEN_CONFIRM (when the open owner is new to the server). 401 * At OPEN, at least two other structures are created, and potentially a 402 * third. rfs4_state_t is created to track the association between an 403 * open owner and a particular file. An rfs4_file_t struct may be 404 * created (if the file is not already open) at OPEN as well. The 405 * rfs4_file_t struct is the only one that is per server and not per 406 * client. The rfs4_deleg_state_t struct is created in the 407 * instance that the server is going to provide a delegation for the 408 * file being OPENed. Finally, the rfs4_lockowner_t is created at the 409 * first use of a lock owner at the server and is a result of the LOCK 410 * operation. The rfs4_lo_state_t struct is then created to represent 411 * the relation between the lock owner and the file. 412 * 413 */ 414 /* 415 * The following ascii art represents each of these data structs and 416 * their references to each other. Note: "<-(x)->" represents the 417 * doubly link lists (list_t). 418 * 419 * ____________________ 420 * | | 421 * | rfs4_client_t | 422 * ->| (1) |<- 423 * / |____________________| \ 424 * / ^ \ 425 * / | \ 426 * ____________________ ____________________ ____________________ 427 * | | | | | | 428 * | rfs4_lockowner_t | | rfs4_openowner_t | | rfs4_deleg_state_t | 429 * | | | (3) <-(1)-> | | <-(2)-> | 430 * |____________________| |____________________| |____________________| 431 * ^ ^ | 432 * | | V 433 * ____________________ ____________________ ____________________ 434 * | | | | | | 435 * | rfs4_lo_state_t |->| rfs4_state_t |->| rfs4_file_t | 436 * | <-(4)-> | | (4) <-(3)-> | | (2) | 437 * |____________________| |____________________| |____________________| 438 */ 439 /* 440 * Each of these data types are kept in a separate rfs4_table_t and is 441 * actually encapsulated within a rfs4_dbe_t struct. The various 442 * tables and their construction is done in nfs4_state.c but 443 * documented here to completeness. 444 * 445 * Table Data struct stored Indexed by 446 * ----- ------------------ ---------- 447 * rfs4_client_tab rfs4_client_t nfs_client_id4 448 * clientid4 449 * 450 * rfs4_openowner_tab rfs4_openowner_t open_owner4 451 * 452 * rfs4_state_tab rfs4_state_t open_owner4 | file 453 * stateid 454 * 455 * rfs4_lo_state_tab rfs4_lo_state_t lockowner | stateid 456 * lock_stateid 457 * 458 * rfs4_lockowner_tab rfs4_lockowner_t lockowner 459 * pid 460 * 461 * rfs4_file_tab rfs4_file_t filehandle 462 * 463 * rfs4_deleg_state_tab rfs4_deleg_state_t clientid4 | file 464 * deleg_stateid 465 */ 466 467 /* 468 * The client struct, it is the root of all state for a particular 469 * client. The client is identified by the nfs_client_id4 via 470 * SETCLIENTID and the server returns the clientid4 as short hand reference 471 */ 472 /* 473 * Client struct - as mentioned above it is the root of all state for 474 * a single client as identified by the client supplied nfs_client_id4 475 * 476 * dbe - encapsulation struct 477 * clientid - server assigned short hand reference to client 478 * nfs_client - client supplied identifier for itself 479 * confirm_verf - the value provided to the client for SETCLIENTID_CONFIRM 480 * need_confirm - does this client need to be SETCLIENTID_CONFIRMed? 481 * 482 * unlksys_completed - has an F_UNLKSYS been done for this client which 483 * says that the use of cleanlocks() on individual files 484 * is not required? 485 * can_reclaim - indicates if client is allowed to reclaim after server 486 * start-up (client had previous state at server) 487 * ss_remove - indicates that the rfs4_client_destroy function should 488 * clean up stable storage file. 489 * forced_expire - set if the sysadmin has used clear_locks for this client. 490 * no_referrals - set if the client is Solaris and pre-dates referrals 491 * deleg_revoked - how many delegations have been revoked for this client? 492 * 493 * cp_confirmed - this refers to a confirmed client struct that has 494 * the same nfs_client_id4 as this client struct. When/if this client 495 * struct is confirmed via SETCLINETID_CONFIRM, the previously 496 * confirmed client struct will be "closed" and hence this reference. 497 * 498 * last_access - used to determine if the client has let its lease expire 499 * cbinfo - struct containing all callback related information 500 * cr_set - credentials used for the SETCLIENTID/SETCLIENTID_CONFIRM pair 501 * sysid - the lock manager sysid allocated for this client's file locks 502 * openownerlist - root of openowners list associated with this client 503 * ss_pn - Pathname to the stable storage file. 504 * cl_addr - Clients network address. 505 * server_instance - pointer to the currently associated server instance 506 */ 507 typedef struct rfs4_client { 508 rfs4_dbe_t *rc_dbe; 509 clientid4 rc_clientid; 510 nfs_client_id4 rc_nfs_client; 511 verifier4 rc_confirm_verf; 512 unsigned rc_need_confirm:1; 513 unsigned rc_unlksys_completed:1; 514 unsigned rc_can_reclaim:1; 515 unsigned rc_ss_remove:1; 516 unsigned rc_forced_expire:1; 517 uint_t rc_deleg_revoked; 518 struct rfs4_client *rc_cp_confirmed; 519 time_t rc_last_access; 520 rfs4_cbinfo_t rc_cbinfo; 521 cred_set_t rc_cr_set; 522 sysid_t rc_sysidt; 523 list_t rc_openownerlist; 524 rfs4_ss_pn_t *rc_ss_pn; 525 struct sockaddr_storage rc_addr; 526 rfs4_servinst_t *rc_server_instance; 527 } rfs4_client_t; 528 529 /* 530 * ClntIP struct - holds the diagnosis about whether the client 531 * cannot support referrals. Set to true for old Solaris clients. 532 */ 533 534 typedef struct rfs4_clntip { 535 rfs4_dbe_t *ri_dbe; 536 struct sockaddr_storage ri_addr; 537 unsigned ri_no_referrals:1; 538 } rfs4_clntip_t; 539 540 /* 541 * The openowner contains the client supplied open_owner4 as well as 542 * the matching sequence id and is used to track the client's usage of 543 * the open_owner4. Note that a reply is saved here as well for 544 * processing of retransmissions. 545 * 546 * dbe - encapsulation struct 547 * client - reference to rfs4_client_t for this openowner 548 * owner - actual client supplied open_owner4 549 * need_confirm - does this openowner need to be OPEN_CONFIRMed 550 * postpone_confirm - set if error received on first use of open_owner 551 * state2confirm - what stateid4 should be used on the OPEN_CONFIRM 552 * open_seqid - what is the next open_seqid expected for this openowner 553 * oo_sw - used to serialize access to the open seqid/reply handling 554 * cr_set - credential used for the OPEN 555 * statelist - root of state struct list associated with this openowner 556 * node - node for client struct list of openowners 557 * reply_fh - open replay processing needs the filehandle so that it is 558 * able to reset the current filehandle for appropriate compound 559 * processing and reply. 560 * reply - last reply sent in relation to this openowner 561 */ 562 typedef struct rfs4_openowner { 563 rfs4_dbe_t *ro_dbe; 564 rfs4_client_t *ro_client; 565 open_owner4 ro_owner; 566 unsigned ro_need_confirm:1; 567 unsigned ro_postpone_confirm:1; 568 seqid4 ro_open_seqid; 569 rfs4_state_wait_t ro_sw; 570 cred_set_t ro_cr_set; 571 list_t ro_statelist; 572 list_node_t ro_node; 573 nfs_fh4 ro_reply_fh; 574 nfs_resop4 ro_reply; 575 } rfs4_openowner_t; 576 577 /* 578 * This state struct represents the association between an openowner 579 * and a file that has been OPENed by that openowner. 580 * 581 * dbe - encapsulation struct 582 * stateid - server provided stateid 583 * owner - reference back to the openowner for this state 584 * finfo - reference to the open file for this state 585 * open_access - how did the openowner OPEN the file (access) 586 * open_deny - how did the openowner OPEN the file (deny) 587 * share_access - what share reservation is on the file (access) 588 * share_deny - what share reservation is on the file (deny) 589 * closed - has this file been closed? 590 * lostatelist - root of list of lo_state associated with this state/file 591 * node - node for state struct list of states 592 */ 593 typedef struct rfs4_state { 594 rfs4_dbe_t *rs_dbe; 595 stateid_t rs_stateid; 596 rfs4_openowner_t *rs_owner; 597 struct rfs4_file *rs_finfo; 598 uint32_t rs_open_access; 599 uint32_t rs_open_deny; 600 uint32_t rs_share_access; 601 uint32_t rs_share_deny; 602 unsigned rs_closed:1; 603 list_t rs_lostatelist; 604 list_node_t rs_node; 605 } rfs4_state_t; 606 607 /* 608 * Lockowner - track the lockowner and its related info 609 * 610 * dbe - encapsulation struct 611 * client - reference to the client 612 * owner - lockowner supplied by the client 613 * pid - local identifier used for file locking 614 */ 615 typedef struct rfs4_lockowner { 616 rfs4_dbe_t *rl_dbe; 617 rfs4_client_t *rl_client; 618 lock_owner4 rl_owner; 619 pid_t rl_pid; 620 } rfs4_lockowner_t; 621 622 /* 623 * Lockowner_state associated with a state struct and lockowner 624 * 625 * dbe - encapsulation struct 626 * state - reference back to state struct for open file 627 * lockid - stateid for this lockowner/state 628 * locker - reference to lockowner 629 * seqid - sequence id for this lockowner/state 630 * skip_seqid_check - used on initialization of struct 631 * locks_cleaned - have all locks been released for this lockowner/file? 632 * lock_completed - successful LOCK with lockowner/file? 633 * ls_sw - used to serialize update seqid/reply/stateid handling 634 * node - node for state struct list of lo_states 635 * reply - last reply sent in relation to this lockowner/state 636 */ 637 typedef struct rfs4_lo_state { 638 rfs4_dbe_t *rls_dbe; 639 rfs4_state_t *rls_state; 640 stateid_t rls_lockid; 641 rfs4_lockowner_t *rls_locker; 642 seqid4 rls_seqid; 643 unsigned rls_skip_seqid_check:1; 644 unsigned rls_locks_cleaned:1; 645 unsigned rls_lock_completed:1; 646 rfs4_state_wait_t rls_sw; 647 list_node_t rls_node; 648 nfs_resop4 rls_reply; 649 } rfs4_lo_state_t; 650 651 /* 652 * Delegation state - per client 653 * 654 * dbe - encapsulation struct 655 * dtype - type of delegation (NONE, READ, WRITE) 656 * delegid - stateid for this delegation 657 * time_granted - time this delegation was assigned to client 658 * time_recalled - time when the server started recall process 659 * time_revoked - if revoked, time that the revoke occurred 660 * finfo - reference to the file associated with this delegation 661 * client - reference to client for which this delegation is associated 662 * node - list of delegations for the file (WRITE == 1, READ == ) 663 */ 664 typedef struct rfs4_deleg_state { 665 rfs4_dbe_t *rds_dbe; 666 open_delegation_type4 rds_dtype; 667 stateid_t rds_delegid; 668 time_t rds_time_granted; 669 time_t rds_time_recalled; 670 time_t rds_time_revoked; 671 struct rfs4_file *rds_finfo; 672 rfs4_client_t *rds_client; 673 list_node_t rds_node; 674 } rfs4_deleg_state_t; 675 676 /* 677 * Delegation info associated with the file 678 * 679 * dtype - type of delegation for file (NONE, READ, WRITE) 680 * time_returned - time that last delegation was returned for file 681 * time_recalled - time that recall sequence started 682 * time_lastgrant - time that last delegation was provided to a client 683 * time_lastwrite - time of last write to use the delegation stateid 684 * time_rm_delayed - time of last remove/rename which was DELAYed 685 * rdgrants - how many read delegations have been provided for this file 686 * wrgrants - how many write delegations provided (can only be one) 687 * recall_count - how many recall threads are outstanding 688 * recall_lock - lock to protect contents of this struct 689 * recall_cv - condition var for the "parent" thread to wait upon 690 * deleg_change_grant - value for change attribute at time of write grant 691 * deleg_change - most recent value of change obtained from client 692 * deleg_change_ts - time of last deleg_change update 693 * ever_recalled - has this particular delegation ever been recalled? 694 * dont_grant - file deletion is impending, don't grant a delegation 695 * conflicted_client - clientid of the client that caused a CB_RECALL 696 * to occur. This is used for delegation policy (should a delegation 697 * be granted shortly after it has been returned?) 698 */ 699 typedef struct rfs4_dinfo { 700 open_delegation_type4 rd_dtype; 701 time_t rd_time_returned; 702 time_t rd_time_recalled; 703 time_t rd_time_lastgrant; 704 time_t rd_time_lastwrite; 705 time_t rd_time_rm_delayed; 706 uint32_t rd_rdgrants; 707 uint32_t rd_wrgrants; 708 int32_t rd_recall_count; 709 kmutex_t rd_recall_lock[1]; 710 kcondvar_t rd_recall_cv[1]; 711 bool_t rd_ever_recalled; 712 uint32_t rd_hold_grant; 713 clientid4 rd_conflicted_client; 714 } rfs4_dinfo_t; 715 716 /* 717 * File 718 * 719 * dbe - encapsulation struct 720 * vp - vnode for the file that is open or has a delegation 721 * filehandle - the filehandle generated by the server for this file 722 * delegstatelist - root of delegation list for this file 723 * dinfo - see struct definition above 724 * share_deny - union of all deny modes on file 725 * share_access - union of all access modes on file 726 * access_read - count of read access 727 * access_write - count of write access 728 * deny_read - count of deny reads 729 * deny_write - count of deny writes 730 * file_rwlock - lock for serializing the removal of a file while 731 * the state structures are active within the server 732 * 733 * The only requirement for locking file_rwlock is that the 734 * caller have a reference to the containing rfs4_file. The dbe 735 * lock may or may not be held for lock/unlock of file_rwlock. 736 * As mentioned above, the file_rwlock is used for serialization 737 * of file removal and more specifically reference to the held 738 * vnode (e.g. vp). 739 */ 740 typedef struct rfs4_file { 741 rfs4_dbe_t *rf_dbe; 742 vnode_t *rf_vp; 743 nfs_fh4 rf_filehandle; 744 list_t rf_delegstatelist; 745 rfs4_dinfo_t rf_dinfo; 746 uint32_t rf_share_deny; 747 uint32_t rf_share_access; 748 uint32_t rf_access_read; 749 uint32_t rf_access_write; 750 uint32_t rf_deny_read; 751 uint32_t rf_deny_write; 752 krwlock_t rf_file_rwlock; 753 } rfs4_file_t; 754 755 /* 756 * nfs4_deleg_policy is used to signify the server's global delegation 757 * policy. The default is to NEVER delegate files and the 758 * administrator must configure the server to enable delegations. 759 * 760 * The disable/enable delegation functions are used to eliminate a 761 * race with exclusive creates. 762 */ 763 typedef enum { 764 SRV_NEVER_DELEGATE = 0, 765 SRV_NORMAL_DELEGATE = 1 766 } srv_deleg_policy_t; 767 768 extern void rfs4_disable_delegation(void), rfs4_enable_delegation(void); 769 770 /* 771 * Request types for delegation. These correspond with 772 * open_delegation_type4 with the addition of a new value, DELEG_ANY, 773 * to reqequest any delegation. 774 */ 775 typedef enum { 776 DELEG_NONE = 0, /* Corresponds to OPEN_DELEG_NONE */ 777 DELEG_READ = 1, /* Corresponds to OPEN_DELEG_READ */ 778 DELEG_WRITE = 2, /* Corresponds to OPEN_DELEG_WRITE */ 779 DELEG_ANY = -1 /* New value to request any delegation type */ 780 } delegreq_t; 781 782 #define NFS4_DELEG4TYPE2REQTYPE(x) (delegreq_t)(x) 783 784 /* 785 * Zone global variables of NFSv4 server 786 */ 787 typedef struct nfs4_srv { 788 /* Unique write verifier */ 789 verifier4 write4verf; 790 /* Delegation lock */ 791 kmutex_t deleg_lock; 792 /* Used to serialize create/destroy of nfs4_server_state database */ 793 kmutex_t state_lock; 794 rfs4_database_t *nfs4_server_state; 795 /* Used to manage access to server instance linked list */ 796 kmutex_t servinst_lock; 797 rfs4_servinst_t *nfs4_cur_servinst; 798 /* Used to manage access to nfs4_deleg_policy */ 799 krwlock_t deleg_policy_lock; 800 srv_deleg_policy_t nfs4_deleg_policy; 801 /* Set first time we see one */ 802 int seen_first_compound; 803 /* 804 * Circular double-linked list of paths for currently-served RGs. 805 * No locking required -- only changed on server start. 806 * Managed with insque/remque. 807 */ 808 rfs4_dss_path_t *dss_pathlist; 809 /* Duplicate request cache */ 810 rfs4_drc_t *nfs4_drc; 811 /* nfsv4 server start time */ 812 time_t rfs4_start_time; 813 /* Used to serialize lookups of clientids */ 814 krwlock_t rfs4_findclient_lock; 815 816 /* NFSv4 server state client tables */ 817 /* table expiry times */ 818 time_t rfs4_client_cache_time; 819 time_t rfs4_openowner_cache_time; 820 time_t rfs4_state_cache_time; 821 time_t rfs4_lo_state_cache_time; 822 time_t rfs4_lockowner_cache_time; 823 time_t rfs4_file_cache_time; 824 time_t rfs4_deleg_state_cache_time; 825 time_t rfs4_clntip_cache_time; 826 /* tables and indexes */ 827 /* client table */ 828 rfs4_table_t *rfs4_client_tab; 829 rfs4_index_t *rfs4_clientid_idx; 830 rfs4_index_t *rfs4_nfsclnt_idx; 831 /* client IP table */ 832 rfs4_table_t *rfs4_clntip_tab; 833 rfs4_index_t *rfs4_clntip_idx; 834 /* Open Owner table */ 835 rfs4_table_t *rfs4_openowner_tab; 836 rfs4_index_t *rfs4_openowner_idx; 837 /* Open State ID table */ 838 rfs4_table_t *rfs4_state_tab; 839 rfs4_index_t *rfs4_state_idx; 840 rfs4_index_t *rfs4_state_owner_file_idx; 841 rfs4_index_t *rfs4_state_file_idx; 842 /* Lock State ID table */ 843 rfs4_table_t *rfs4_lo_state_tab; 844 rfs4_index_t *rfs4_lo_state_idx; 845 rfs4_index_t *rfs4_lo_state_owner_idx; 846 /* Lock owner table */ 847 rfs4_table_t *rfs4_lockowner_tab; 848 rfs4_index_t *rfs4_lockowner_idx; 849 rfs4_index_t *rfs4_lockowner_pid_idx; 850 /* File table */ 851 rfs4_table_t *rfs4_file_tab; 852 rfs4_index_t *rfs4_file_idx; 853 /* Deleg State table */ 854 rfs4_table_t *rfs4_deleg_state_tab; 855 rfs4_index_t *rfs4_deleg_idx; 856 rfs4_index_t *rfs4_deleg_state_idx; 857 858 /* client stable storage */ 859 int rfs4_ss_enabled; 860 } nfs4_srv_t; 861 862 /* 863 * max length of the NFSv4 server database name 864 */ 865 #define RFS4_MAX_MEM_CACHE_NAME 48 866 867 /* 868 * global NFSv4 server kmem caches 869 * r_db_name - The name of the state database and the table that will use it 870 * These tables are defined in nfs4_srv_t 871 * r_db_mem_cache - The kmem cache associated with the state database name 872 */ 873 typedef struct rfs4_db_mem_cache { 874 char r_db_name[RFS4_MAX_MEM_CACHE_NAME]; 875 kmem_cache_t *r_db_mem_cache; 876 } rfs4_db_mem_cache_t; 877 878 #define RFS4_DB_MEM_CACHE_NUM 8 879 880 rfs4_db_mem_cache_t rfs4_db_mem_cache_table[RFS4_DB_MEM_CACHE_NUM]; 881 882 883 extern srv_deleg_policy_t nfs4_get_deleg_policy(); 884 885 extern void rfs4_servinst_create(nfs4_srv_t *, int, int, char **); 886 extern void rfs4_servinst_destroy_all(nfs4_srv_t *); 887 extern void rfs4_servinst_assign(nfs4_srv_t *, rfs4_client_t *, 888 rfs4_servinst_t *); 889 extern rfs4_servinst_t *rfs4_servinst(rfs4_client_t *); 890 extern int rfs4_clnt_in_grace(rfs4_client_t *); 891 extern int rfs4_servinst_in_grace(rfs4_servinst_t *); 892 extern int rfs4_servinst_grace_new(rfs4_servinst_t *); 893 extern void rfs4_grace_start(rfs4_servinst_t *); 894 extern void rfs4_grace_start_new(nfs4_srv_t *); 895 extern void rfs4_grace_reset_all(nfs4_srv_t *); 896 extern void rfs4_ss_oldstate(rfs4_oldstate_t *, char *, char *); 897 extern void rfs4_dss_readstate(nfs4_srv_t *, int, char **); 898 899 /* 900 * Various interfaces to manipulate the state structures introduced 901 * above 902 */ 903 extern void rfs4_free_reply(nfs_resop4 *); 904 extern void rfs4_copy_reply(nfs_resop4 *, nfs_resop4 *); 905 906 /* rfs4_client_t handling */ 907 extern rfs4_client_t *rfs4_findclient(nfs_client_id4 *, 908 bool_t *, rfs4_client_t *); 909 extern rfs4_client_t *rfs4_findclient_by_id(clientid4, bool_t); 910 extern rfs4_client_t *rfs4_findclient_by_addr(struct sockaddr *); 911 extern void rfs4_client_rele(rfs4_client_t *); 912 extern void rfs4_client_close(rfs4_client_t *); 913 extern void rfs4_client_state_remove(rfs4_client_t *); 914 extern void rfs4_client_scv_next(rfs4_client_t *); 915 extern void rfs4_update_lease(rfs4_client_t *); 916 extern bool_t rfs4_lease_expired(rfs4_client_t *); 917 extern nfsstat4 rfs4_check_clientid(clientid4 *, int); 918 919 /* rfs4_clntip_t handling */ 920 extern rfs4_clntip_t *rfs4_find_clntip(struct sockaddr *, bool_t *); 921 extern void rfs4_invalidate_clntip(struct sockaddr *); 922 923 /* rfs4_openowner_t handling */ 924 extern rfs4_openowner_t *rfs4_findopenowner(open_owner4 *, bool_t *, seqid4); 925 extern void rfs4_update_open_sequence(rfs4_openowner_t *); 926 extern void rfs4_update_open_resp(rfs4_openowner_t *, 927 nfs_resop4 *, nfs_fh4 *); 928 extern void rfs4_openowner_rele(rfs4_openowner_t *); 929 extern void rfs4_free_opens(rfs4_openowner_t *, bool_t, bool_t); 930 931 /* rfs4_lockowner_t handling */ 932 extern rfs4_lockowner_t *rfs4_findlockowner(lock_owner4 *, bool_t *); 933 extern rfs4_lockowner_t *rfs4_findlockowner_by_pid(pid_t); 934 extern void rfs4_lockowner_rele(rfs4_lockowner_t *); 935 936 /* rfs4_state_t handling */ 937 extern rfs4_state_t *rfs4_findstate_by_owner_file(rfs4_openowner_t *, 938 rfs4_file_t *, bool_t *); 939 extern void rfs4_state_rele(rfs4_state_t *); 940 extern void rfs4_state_close(rfs4_state_t *, bool_t, 941 bool_t, cred_t *); 942 extern void rfs4_release_share_lock_state(rfs4_state_t *, 943 cred_t *, bool_t); 944 extern void rfs4_close_all_state(rfs4_file_t *); 945 946 /* rfs4_lo_state_t handling */ 947 extern rfs4_lo_state_t *rfs4_findlo_state_by_owner(rfs4_lockowner_t *, 948 rfs4_state_t *, bool_t *); 949 extern void rfs4_lo_state_rele(rfs4_lo_state_t *, bool_t); 950 extern void rfs4_update_lock_sequence(rfs4_lo_state_t *); 951 extern void rfs4_update_lock_resp(rfs4_lo_state_t *, 952 nfs_resop4 *); 953 954 /* rfs4_file_t handling */ 955 extern rfs4_file_t *rfs4_findfile(vnode_t *, nfs_fh4 *, bool_t *); 956 extern rfs4_file_t *rfs4_findfile_withlock(vnode_t *, nfs_fh4 *, 957 bool_t *); 958 extern void rfs4_file_rele(rfs4_file_t *); 959 960 /* General collection of "get state" functions */ 961 extern nfsstat4 rfs4_get_state(stateid4 *, rfs4_state_t **, 962 rfs4_dbsearch_type_t); 963 extern nfsstat4 rfs4_get_deleg_state(stateid4 *, 964 rfs4_deleg_state_t **); 965 extern nfsstat4 rfs4_get_lo_state(stateid4 *, rfs4_lo_state_t **, 966 bool_t); 967 extern nfsstat4 rfs4_check_stateid(int, vnode_t *, stateid4 *, 968 bool_t, bool_t *, bool_t, 969 caller_context_t *); 970 extern int rfs4_check_stateid_seqid(rfs4_state_t *, stateid4 *); 971 extern int rfs4_check_lo_stateid_seqid(rfs4_lo_state_t *, 972 stateid4 *); 973 974 /* return values for rfs4_check_stateid_seqid() */ 975 #define NFS4_CHECK_STATEID_OKAY 1 976 #define NFS4_CHECK_STATEID_OLD 2 977 #define NFS4_CHECK_STATEID_BAD 3 978 #define NFS4_CHECK_STATEID_EXPIRED 4 979 #define NFS4_CHECK_STATEID_REPLAY 5 980 #define NFS4_CHECK_STATEID_CLOSED 6 981 #define NFS4_CHECK_STATEID_UNCONFIRMED 7 982 983 /* delay() time that server is willing to briefly wait for a delegreturn */ 984 #define NFS4_DELEGATION_CONFLICT_DELAY (hz/10) 985 986 /* 987 * Interfaces for handling of callback's client handle cache and 988 * callback interfaces themselves. 989 */ 990 extern void rfs4_cbinfo_free(rfs4_cbinfo_t *); 991 extern void rfs4_client_setcb(rfs4_client_t *, cb_client4 *, 992 uint32_t); 993 extern void rfs4_deleg_cb_check(rfs4_client_t *); 994 extern nfsstat4 rfs4_vop_getattr(vnode_t *, vattr_t *, int, cred_t *); 995 996 /* rfs4_deleg_state_t handling and other delegation interfaces */ 997 extern rfs4_deleg_state_t *rfs4_finddeleg(rfs4_state_t *, bool_t *); 998 extern rfs4_deleg_state_t *rfs4_finddelegstate(stateid_t *); 999 extern bool_t rfs4_check_recall(rfs4_state_t *, uint32_t); 1000 extern void rfs4_recall_deleg(rfs4_file_t *, 1001 bool_t, rfs4_client_t *); 1002 extern int rfs4_get_deleg(rfs4_state_t *, open_delegation_type4, 1003 open_delegation_type4 (*policy)(rfs4_state_t *, 1004 open_delegation_type4 dtype)); 1005 extern rfs4_deleg_state_t *rfs4_grant_delegation(delegreq_t, rfs4_state_t *, 1006 int *); 1007 extern void rfs4_set_deleg_response(rfs4_deleg_state_t *, 1008 open_delegation4 *, nfsace4 *, int); 1009 extern void rfs4_return_deleg(rfs4_deleg_state_t *, bool_t); 1010 extern bool_t rfs4_is_deleg(rfs4_state_t *); 1011 extern void rfs4_deleg_state_rele(rfs4_deleg_state_t *); 1012 extern bool_t rfs4_check_delegated_byfp(int, rfs4_file_t *, 1013 bool_t, bool_t, bool_t, clientid4 *); 1014 extern void rfs4_clear_dont_grant(rfs4_file_t *); 1015 1016 /* 1017 * nfs4 monitored operations. 1018 */ 1019 extern int deleg_rd_open(femarg_t *, int, cred_t *, caller_context_t *); 1020 extern int deleg_wr_open(femarg_t *, int, cred_t *, caller_context_t *); 1021 extern int deleg_wr_read(femarg_t *, uio_t *, int, cred_t *, 1022 caller_context_t *); 1023 extern int deleg_rd_write(femarg_t *, uio_t *, int, cred_t *, 1024 caller_context_t *); 1025 extern int deleg_wr_write(femarg_t *, uio_t *, int, cred_t *, 1026 caller_context_t *); 1027 extern int deleg_rd_setattr(femarg_t *, vattr_t *, int, cred_t *, 1028 caller_context_t *); 1029 extern int deleg_wr_setattr(femarg_t *, vattr_t *, int, cred_t *, 1030 caller_context_t *); 1031 extern int deleg_rd_rwlock(femarg_t *, int, caller_context_t *); 1032 extern int deleg_wr_rwlock(femarg_t *, int, caller_context_t *); 1033 extern int deleg_rd_space(femarg_t *, int, flock64_t *, int, offset_t, cred_t *, 1034 caller_context_t *); 1035 extern int deleg_wr_space(femarg_t *, int, flock64_t *, int, offset_t, cred_t *, 1036 caller_context_t *); 1037 extern int deleg_rd_setsecattr(femarg_t *, vsecattr_t *, int, cred_t *, 1038 caller_context_t *); 1039 extern int deleg_wr_setsecattr(femarg_t *, vsecattr_t *, int, cred_t *, 1040 caller_context_t *); 1041 extern int deleg_rd_vnevent(femarg_t *, vnevent_t, vnode_t *, char *, 1042 caller_context_t *); 1043 extern int deleg_wr_vnevent(femarg_t *, vnevent_t, vnode_t *, char *, 1044 caller_context_t *); 1045 1046 extern void rfs4_mon_hold(void *); 1047 extern void rfs4_mon_rele(void *); 1048 1049 extern fem_t *deleg_rdops; 1050 extern fem_t *deleg_wrops; 1051 1052 extern int rfs4_share(rfs4_state_t *, uint32_t, uint32_t); 1053 extern int rfs4_unshare(rfs4_state_t *); 1054 extern void rfs4_set_deleg_policy(nfs4_srv_t *, srv_deleg_policy_t); 1055 extern void rfs4_hold_deleg_policy(nfs4_srv_t *); 1056 extern void rfs4_rele_deleg_policy(nfs4_srv_t *); 1057 1058 #ifdef DEBUG 1059 #define NFS4_DEBUG(var, args) if (var) cmn_err args 1060 1061 extern int rfs4_debug; 1062 extern int nfs4_client_attr_debug; 1063 extern int nfs4_client_state_debug; 1064 extern int nfs4_client_shadow_debug; 1065 extern int nfs4_client_lock_debug; 1066 extern int nfs4_client_lease_debug; 1067 extern int nfs4_seqid_sync; 1068 extern int nfs4_client_map_debug; 1069 extern int nfs4_client_inactive_debug; 1070 extern int nfs4_client_recov_debug; 1071 extern int nfs4_client_failover_debug; 1072 extern int nfs4_client_call_debug; 1073 extern int nfs4_client_foo_debug; 1074 extern int nfs4_client_zone_debug; 1075 extern int nfs4_lost_rqst_debug; 1076 extern int nfs4_open_stream_debug; 1077 extern int nfs4_client_open_dg; 1078 extern int nfs4_srvmnt_debug; 1079 extern int nfs4_utf8_debug; 1080 1081 void rfs4_dbe_debug(rfs4_dbe_t *e); 1082 1083 #ifdef NFS4_DEBUG_MUTEX 1084 void nfs4_debug_mutex_enter(kmutex_t *, char *, int); 1085 void nfs4_debug_mutex_exit(kmutex_t *, char *, int); 1086 1087 #define mutex_enter(m) nfs4_debug_mutex_enter((m), __FILE__, __LINE__) 1088 #define mutex_exit(m) nfs4_debug_mutex_exit((m), __FILE__, __LINE__) 1089 #endif /* NFS4_DEBUG_MUTEX */ 1090 1091 #else /* ! DEBUG */ 1092 #define NFS4_DEBUG(var, args) 1093 #endif /* DEBUG */ 1094 1095 /* 1096 * XXX - temporary for testing of volatile fh 1097 */ 1098 1099 #ifdef VOLATILE_FH_TEST 1100 1101 struct nfs_fh4_fmt { 1102 fhandle4_t fh4_i; 1103 uint32_t fh4_flag; 1104 uint32_t fh4_volatile_id; 1105 }; 1106 1107 #else /* VOLATILE_FH_TEST */ 1108 1109 struct nfs_fh4_fmt { 1110 fhandle4_t fh4_i; 1111 uint32_t fh4_flag; 1112 }; 1113 1114 #endif /* VOLATILE_FH_TEST */ 1115 1116 #define FH4_NAMEDATTR 1 1117 #define FH4_ATTRDIR 2 1118 1119 #define fh4_fsid fh4_i.fhx_fsid 1120 #define fh4_len fh4_i.fhx_len /* fid length */ 1121 #define fh4_data fh4_i.fhx_data /* fid bytes */ 1122 #define fh4_xlen fh4_i.fhx_xlen 1123 #define fh4_xdata fh4_i.fhx_xdata 1124 typedef struct nfs_fh4_fmt nfs_fh4_fmt_t; 1125 1126 #define fh4_to_fmt4(fh4p) ((nfs_fh4_fmt_t *)(fh4p)->nfs_fh4_val) 1127 #define get_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) & (flag)) 1128 #define set_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) |= (flag)) 1129 #define clr_fh4_flag(fh4p, flag) ((fh4_to_fmt4(fh4p)->fh4_flag) &= ~(flag)) 1130 1131 #define NFS_FH4_LEN sizeof (nfs_fh4_fmt_t) 1132 1133 /* 1134 * Copy fields from external (fhandle_t) to in-memory (nfs_fh4_fmt_t) 1135 * format to support export info checking. It does not copy over 1136 * the complete filehandle, just the fsid, xlen and xdata. It may 1137 * need to be changed to be used in other places. 1138 * 1139 * NOTE: The macro expects the space to be pre-allocated for 1140 * the contents of nfs_fh4_fmt_t. 1141 */ 1142 #define FH_TO_FMT4(exifh, nfs_fmt) { \ 1143 bzero((nfs_fmt), NFS_FH4_LEN); \ 1144 (nfs_fmt)->fh4_fsid = (exifh)->fh_fsid; \ 1145 (nfs_fmt)->fh4_xlen = (exifh)->fh_xlen; \ 1146 bcopy((exifh)->fh_xdata, (nfs_fmt)->fh4_xdata, \ 1147 (exifh)->fh_xlen); \ 1148 } 1149 1150 /* 1151 * A few definitions of repeatedly used constructs for nfsv4 1152 */ 1153 #define UTF8STRING_FREE(str) { \ 1154 kmem_free((str).utf8string_val, (str).utf8string_len); \ 1155 (str).utf8string_val = NULL; \ 1156 (str).utf8string_len = 0; \ 1157 } 1158 1159 /* 1160 * NFS4_VOLATILE_FH yields non-zero if the filesystem uses non-persistent 1161 * filehandles. 1162 */ 1163 #define NFS4_VOLATILE_FH(mi) \ 1164 ((mi)->mi_fh_expire_type & \ 1165 (FH4_VOLATILE_ANY | FH4_VOL_MIGRATION | FH4_VOL_RENAME)) 1166 1167 /* 1168 * NFS_IS_DOTNAME checks if the name given represents a dot or dotdot entry 1169 */ 1170 #define NFS_IS_DOTNAME(name) \ 1171 (((name)[0] == '.') && \ 1172 (((name)[1] == '\0') || (((name)[1] == '.') && ((name)[2] == '\0')))) 1173 1174 /* 1175 * Define the number of bits in a bitmap word (uint32) 1176 */ 1177 #define NFS4_BITMAP4_BITSPERWORD (sizeof (uint32_t) * 8) 1178 1179 /* 1180 * Define the value for the access field of the compound_state structure 1181 * based on the result of nfsauth access checking. 1182 */ 1183 #define CS_ACCESS_OK 0x1 1184 #define CS_ACCESS_DENIED 0x2 1185 #define CS_ACCESS_LIMITED 0x4 1186 1187 /* 1188 * compound state in nfsv4 server 1189 */ 1190 struct compound_state { 1191 struct exportinfo *exi; 1192 struct exportinfo *saved_exi; /* export struct for saved_vp */ 1193 cred_t *basecr; /* UNIX cred: only RPC request */ 1194 caddr_t principal; 1195 int nfsflavor; 1196 cred_t *cr; /* UNIX cred: RPC request and */ 1197 /* target export */ 1198 bool_t cont; 1199 uint_t access; /* access perm on vp per request */ 1200 bool_t deleg; /* TRUE if current fh has */ 1201 /* write delegated */ 1202 vnode_t *vp; /* modified by PUTFH, and by ops that */ 1203 /* input to GETFH */ 1204 bool_t mandlock; /* Is mandatory locking in effect */ 1205 /* for vp */ 1206 vnode_t *saved_vp; /* modified by SAVEFH, copied to */ 1207 /* vp by RESTOREFH */ 1208 nfsstat4 *statusp; 1209 nfs_fh4 fh; /* ditto. valid only if vp != NULL */ 1210 nfs_fh4 saved_fh; /* ditto. valid only if */ 1211 /* saved_vp != NULL */ 1212 struct svc_req *req; 1213 char fhbuf[NFS4_FHSIZE]; 1214 }; 1215 1216 /* 1217 * Conversion commands for nfsv4 server attr checking 1218 */ 1219 enum nfs4_attr_cmd { 1220 NFS4ATTR_SUPPORTED = 0, /* check which attrs supported */ 1221 NFS4ATTR_GETIT = 1, /* getattr - sys to fattr4 (r) */ 1222 NFS4ATTR_SETIT = 2, /* setattr - fattr4 to sys (w) */ 1223 NFS4ATTR_VERIT = 3, /* verify - fattr4 to sys (r) */ 1224 NFS4ATTR_FREEIT = 4 /* free any alloc'd space for attr */ 1225 }; 1226 1227 typedef enum nfs4_attr_cmd nfs4_attr_cmd_t; 1228 1229 struct nfs4_svgetit_arg { 1230 nfs4_attr_cmd_t op; /* getit or setit */ 1231 struct compound_state *cs; 1232 struct statvfs64 *sbp; 1233 uint_t flag; /* VOP_GETATTR/VOP_SETATTR flag */ 1234 uint_t xattr; /* object is xattr */ 1235 bool_t rdattr_error_req; /* if readdir & client wants */ 1236 /* rdattr_error */ 1237 nfsstat4 rdattr_error; /* used for per-entry status */ 1238 /* (if rdattr_err) */ 1239 bool_t is_referral; /* because sometimes we tell lies */ 1240 bool_t mntdfid_set; 1241 fattr4_mounted_on_fileid 1242 mounted_on_fileid; 1243 /* readdir op can always return */ 1244 /* d_ino from server fs dirent */ 1245 /* for mounted_on_fileid attr. */ 1246 /* This field holds d_ino so */ 1247 /* srv attr conv code can avoid */ 1248 /* doing an untraverse. */ 1249 vattr_t vap[1]; 1250 }; 1251 1252 struct nfs4_ntov_map { 1253 bitmap4 fbit; /* FATTR4_XXX_MASKY */ 1254 uint_t vbit; /* AT_XXX */ 1255 bool_t vfsstat; 1256 bool_t mandatory; /* attribute mandatory to implement? */ 1257 uint_t nval; 1258 int xdr_size; /* Size of XDR'd attr */ 1259 xdrproc_t xfunc; 1260 int (*sv_getit)(nfs4_attr_cmd_t, struct nfs4_svgetit_arg *, 1261 union nfs4_attr_u *); /* subroutine for getting attr. */ 1262 char *prtstr; /* string attr for printing */ 1263 }; 1264 1265 struct nfs4attr_to_vattr { 1266 vnode_t *vp; 1267 vattr_t *vap; 1268 nfs_fh4 *fhp; 1269 nfsstat4 rdattr_error; 1270 uint32_t flag; 1271 fattr4_change change; 1272 fattr4_fsid srv_fsid; 1273 fattr4_mounted_on_fileid mntd_fid; 1274 }; 1275 1276 typedef struct nfs4attr_to_vattr ntov4_t; 1277 1278 /* 1279 * nfs4attr_to_vattr flags 1280 */ 1281 #define NTOV_FHP_VALID 0x01 1282 #define NTOV_RDATTR_ERROR_VALID 0x02 1283 #define NTOV_CHANGE_VALID 0x04 1284 #define NTOV_SUPP_VALID 0x08 1285 #define NTOV_SRV_FSID_VALID 0x10 1286 #define NTOV_MOUNTED_ON_FILEID_VALID 0x20 1287 1288 1289 #define FATTR4_MANDATTR_MASK ( \ 1290 FATTR4_SUPPORTED_ATTRS_MASK | \ 1291 FATTR4_TYPE_MASK | \ 1292 FATTR4_FH_EXPIRE_TYPE_MASK | \ 1293 FATTR4_CHANGE_MASK | \ 1294 FATTR4_SIZE_MASK | \ 1295 FATTR4_LINK_SUPPORT_MASK | \ 1296 FATTR4_SYMLINK_SUPPORT_MASK | \ 1297 FATTR4_NAMED_ATTR_MASK | \ 1298 FATTR4_FSID_MASK | \ 1299 FATTR4_UNIQUE_HANDLES_MASK | \ 1300 FATTR4_LEASE_TIME_MASK | \ 1301 FATTR4_RDATTR_ERROR_MASK | \ 1302 FATTR4_FILEHANDLE_MASK) 1303 1304 1305 struct nfs4attr_to_osattr { 1306 void *attrconv_arg; 1307 uint_t mask; 1308 }; 1309 1310 struct mntinfo4; 1311 1312 /* 1313 * lkp4_attr_setup lists the different options for attributes when calling 1314 * nfs4lookup_setup - either no attributes (just lookups - e.g., secinfo), 1315 * one component only (normal component lookup), get attributes for the 1316 * last component (e.g., mount), attributes for each component (e.g., 1317 * failovers later), just the filehandle for the last component (e.g., 1318 * volatile filehandle recovery), or stuff that needs OPENATTR (e.g. 1319 * looking up a named attribute or it's hidden directory). 1320 */ 1321 enum lkp4_attr_setup { 1322 LKP4_NO_ATTRIBUTES = 0, /* no attrs or filehandles */ 1323 LKP4_ALL_ATTRIBUTES = 3, /* multi-comp: attrs for all comps */ 1324 LKP4_LAST_NAMED_ATTR = 5, /* multi-comp: named attr & attrdir */ 1325 LKP4_LAST_ATTRDIR = 6, /* multi-comp: just attrdir */ 1326 LKP4_ALL_ATTR_SECINFO = 7 /* multi-comp: attrs for all comp and */ 1327 /* secinfo for last comp */ 1328 }; 1329 1330 /* 1331 * lookup4_param a set of parameters to nfs4lookup_setup - 1332 * used to setup a path lookup compound request. 1333 */ 1334 typedef struct lookup4_param { 1335 enum lkp4_attr_setup l4_getattrs; /* (in) get attrs in the lookup? */ 1336 int header_len; /* (in) num ops before first lookup */ 1337 int trailer_len; /* (in) num ops after last */ 1338 /* Lookup/Getattr */ 1339 bitmap4 ga_bits; /* (in) Which attributes for Getattr */ 1340 COMPOUND4args_clnt *argsp; /* (in/out) args for compound struct */ 1341 COMPOUND4res_clnt *resp; /* (in/out) res for compound struct */ 1342 int arglen; /* (out) argop buffer alloc'd length */ 1343 struct mntinfo4 *mi; 1344 } lookup4_param_t; 1345 1346 1347 #define NFS4_FATTR4_FINISH -1 /* fattr4 index indicating finish */ 1348 1349 typedef int (*nfs4attr_to_os_t)(int, union nfs4_attr_u *, 1350 struct nfs4attr_to_osattr *); 1351 1352 /* 1353 * The nfs4_error_t is the basic structure to return error values 1354 * from rfs4call. It encapsulates the unix errno 1355 * value, the nfsstat4 value and the rpc status value into a single 1356 * structure. 1357 * 1358 * If error is set, then stat is ignored and rpc_status may be 1359 * set if the error occurred as the result of a CLNT_CALL. If 1360 * stat is set, then rpc request succeeded, error and 1361 * rpc_status are set to 0 and stat contains the result of 1362 * operation, NFS4_OK or one of the NFS4ERR_* values. 1363 * 1364 * Functions which want to generate errors independently from 1365 * rfs4call should set error to the desired errno value and 1366 * set stat and rpc_status to 0. nfs4_error_init() is a 1367 * convenient function to do this. 1368 */ 1369 typedef struct { 1370 int error; 1371 nfsstat4 stat; 1372 enum clnt_stat rpc_status; 1373 } nfs4_error_t; 1374 1375 /* 1376 * Shared functions 1377 */ 1378 extern void rfs4_op_readdir(nfs_argop4 *, nfs_resop4 *, 1379 struct svc_req *, struct compound_state *); 1380 extern void nfs_fh4_copy(nfs_fh4 *, nfs_fh4 *); 1381 1382 extern void nfs4_fattr4_free(fattr4 *); 1383 1384 extern int nfs4lookup_setup(char *, lookup4_param_t *, int); 1385 extern void nfs4_getattr_otw_norecovery(vnode_t *, 1386 nfs4_ga_res_t *, nfs4_error_t *, cred_t *, int); 1387 extern int nfs4_getattr_otw(vnode_t *, nfs4_ga_res_t *, cred_t *, int); 1388 extern int nfs4cmpfh(const nfs_fh4 *, const nfs_fh4 *); 1389 extern int nfs4cmpfhandle(nfs4_fhandle_t *, nfs4_fhandle_t *); 1390 extern int nfs4getattr(vnode_t *, struct vattr *, cred_t *); 1391 extern int nfs4_waitfor_purge_complete(vnode_t *); 1392 extern int nfs4_validate_caches(vnode_t *, cred_t *); 1393 extern int nfs4init(int, char *); 1394 extern void nfs4fini(void); 1395 extern int nfs4_vfsinit(void); 1396 extern void nfs4_vfsfini(void); 1397 1398 extern void nfs4_vnops_init(void); 1399 extern void nfs4_vnops_fini(void); 1400 extern void nfs_idmap_init(void); 1401 extern void nfs_idmap_flush(int); 1402 extern void nfs_idmap_fini(void); 1403 extern int nfs4_rnode_init(void); 1404 extern int nfs4_rnode_fini(void); 1405 extern int nfs4_shadow_init(void); 1406 extern int nfs4_shadow_fini(void); 1407 extern int nfs4_acache_init(void); 1408 extern int nfs4_acache_fini(void); 1409 extern int nfs4_subr_init(void); 1410 extern int nfs4_subr_fini(void); 1411 extern void nfs4_acl_init(void); 1412 extern void nfs4_acl_free_cache(vsecattr_t *); 1413 1414 extern int geterrno4(nfsstat4); 1415 extern nfsstat4 puterrno4(int); 1416 extern int nfs4_need_to_bump_seqid(COMPOUND4res_clnt *); 1417 extern int nfs4tsize(void); 1418 extern int checkauth4(struct compound_state *, struct svc_req *); 1419 extern nfsstat4 call_checkauth4(struct compound_state *, struct svc_req *); 1420 extern int is_exported_sec(int, struct exportinfo *); 1421 extern void nfs4_vmask_to_nmask(uint_t, bitmap4 *); 1422 extern void nfs4_vmask_to_nmask_set(uint_t, bitmap4 *); 1423 extern int nfs_idmap_str_uid(utf8string *u8s, uid_t *, bool_t); 1424 extern int nfs_idmap_str_gid(utf8string *u8s, gid_t *, bool_t); 1425 extern int nfs_idmap_uid_str(uid_t, utf8string *u8s, bool_t); 1426 extern int nfs_idmap_gid_str(gid_t gid, utf8string *u8s, bool_t); 1427 extern int nfs4_time_ntov(nfstime4 *, timestruc_t *); 1428 extern int nfs4_time_vton(timestruc_t *, nfstime4 *); 1429 extern char *utf8_to_str(utf8string *, uint_t *, char *); 1430 extern char *utf8_to_fn(utf8string *, uint_t *, char *); 1431 extern utf8string *str_to_utf8(char *, utf8string *); 1432 extern utf8string *utf8_copy(utf8string *, utf8string *); 1433 extern int utf8_compare(const utf8string *, const utf8string *); 1434 extern nfsstat4 utf8_dir_verify(utf8string *); 1435 extern char *utf8_strchr(utf8string *, const char); 1436 extern int ln_ace4_cmp(nfsace4 *, nfsace4 *, int); 1437 extern int vs_aent_to_ace4(vsecattr_t *, vsecattr_t *, int, int); 1438 extern int vs_ace4_to_aent(vsecattr_t *, vsecattr_t *, uid_t, gid_t, 1439 int, int); 1440 extern int vs_ace4_to_acet(vsecattr_t *, vsecattr_t *, uid_t, gid_t, 1441 int); 1442 extern int vs_acet_to_ace4(vsecattr_t *, vsecattr_t *, int); 1443 extern void vs_acet_destroy(vsecattr_t *); 1444 extern void vs_ace4_destroy(vsecattr_t *); 1445 extern void vs_aent_destroy(vsecattr_t *); 1446 1447 extern int vn_find_nfs_record(vnode_t *, nvlist_t **, char **, char **); 1448 extern int vn_is_nfs_reparse(vnode_t *, cred_t *); 1449 extern fs_locations4 *fetch_referral(vnode_t *, cred_t *); 1450 extern char *build_symlink(vnode_t *, cred_t *, size_t *); 1451 1452 extern int stateid4_cmp(stateid4 *, stateid4 *); 1453 1454 extern vtype_t nf4_to_vt[]; 1455 1456 extern struct nfs4_ntov_map nfs4_ntov_map[]; 1457 extern uint_t nfs4_ntov_map_size; 1458 1459 extern struct vfsops *nfs4_vfsops; 1460 extern struct vnodeops *nfs4_vnodeops; 1461 extern const struct fs_operation_def nfs4_vnodeops_template[]; 1462 extern vnodeops_t *nfs4_trigger_vnodeops; 1463 extern const struct fs_operation_def nfs4_trigger_vnodeops_template[]; 1464 1465 extern uint_t nfs4_tsize(struct knetconfig *); 1466 extern uint_t rfs4_tsize(struct svc_req *); 1467 1468 extern bool_t xdr_inline_decode_nfs_fh4(uint32_t *, nfs_fh4_fmt_t *, 1469 uint32_t); 1470 extern bool_t xdr_inline_encode_nfs_fh4(uint32_t **, uint32_t *, 1471 nfs_fh4_fmt_t *); 1472 1473 #ifdef DEBUG 1474 extern int rfs4_do_pre_op_attr; 1475 extern int rfs4_do_post_op_attr; 1476 #endif 1477 1478 extern stateid4 clnt_special0; 1479 extern stateid4 clnt_special1; 1480 #define CLNT_ISSPECIAL(id) (stateid4_cmp(id, &clnt_special0) || \ 1481 stateid4_cmp(id, &clnt_special1)) 1482 1483 /* 1484 * The NFS Version 4 service procedures. 1485 */ 1486 1487 extern void rfs4_do_server_start(int, int, int); 1488 extern void rfs4_compound(COMPOUND4args *, COMPOUND4res *, 1489 struct exportinfo *, struct svc_req *, cred_t *, int *); 1490 extern void rfs4_compound_free(COMPOUND4res *); 1491 extern void rfs4_compound_flagproc(COMPOUND4args *, int *); 1492 1493 extern void rfs4_srvrinit(void); 1494 extern void rfs4_srvrfini(void); 1495 extern void rfs4_srv_zone_init(nfs_globals_t *); 1496 extern void rfs4_srv_zone_fini(nfs_globals_t *); 1497 extern void rfs4_state_g_init(void); 1498 extern void rfs4_state_zone_init(nfs4_srv_t *); 1499 extern void rfs4_state_g_fini(void); 1500 extern void rfs4_state_zone_fini(void); 1501 extern nfs4_srv_t *nfs4_get_srv(void); 1502 1503 #endif 1504 #ifdef __cplusplus 1505 } 1506 #endif 1507 1508 #endif /* _NFS4_H */ 1509