nfs.h revision 1.22 1 /* $NetBSD: nfs.h,v 1.22 2000/06/09 00:00:17 fvdl Exp $ */
2 /*
3 * Copyright (c) 1989, 1993, 1995
4 * The Regents of the University of California. All rights reserved.
5 *
6 * This code is derived from software contributed to Berkeley by
7 * Rick Macklem at The University of Guelph.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by the University of
20 * California, Berkeley and its contributors.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * SUCH DAMAGE.
36 *
37 * @(#)nfs.h 8.4 (Berkeley) 5/1/95
38 */
39
40 #ifndef _NFS_NFS_H_
41 #define _NFS_NFS_H_
42
43 /*
44 * Tunable constants for nfs
45 */
46
47 #define NFS_MAXIOVEC 34
48 #define NFS_TICKINTVL 5 /* Desired time for a tick (msec) */
49 #define NFS_HZ (hz / nfs_ticks) /* Ticks/sec */
50 #define NFS_TIMEO (3 * NFS_HZ) /* Default timeout = 3 seconds */
51 #define NFS_MINTIMEO (1 * NFS_HZ) /* Min timeout to use */
52 #define NFS_MAXTIMEO (60 * NFS_HZ) /* Max timeout to backoff to */
53 #define NFS_MINIDEMTIMEO (5 * NFS_HZ) /* Min timeout for non-idempotent ops*/
54 #define NFS_MAXREXMIT 100 /* Stop counting after this many */
55 #define NFS_MAXWINDOW 1024 /* Max number of outstanding requests */
56 #define NFS_RETRANS 10 /* Num of retrans for soft mounts */
57 #define NFS_MAXGRPS 16 /* Max. size of groups list */
58 #ifndef NFS_MINATTRTIMO
59 #define NFS_MINATTRTIMO 5 /* Attribute cache timeout in sec */
60 #endif
61 #ifndef NFS_MAXATTRTIMO
62 #define NFS_MAXATTRTIMO 60
63 #endif
64 #define NFS_WSIZE 8192 /* Def. write data size <= 8192 */
65 #define NFS_RSIZE 8192 /* Def. read data size <= 8192 */
66 #define NFS_READDIRSIZE 8192 /* Def. readdir size */
67 #define NFS_DEFRAHEAD 1 /* Def. read ahead # blocks */
68 #define NFS_MAXRAHEAD 4 /* Max. read ahead # blocks */
69 #define NFS_MAXUIDHASH 64 /* Max. # of hashed uid entries/mp */
70 #define NFS_MAXASYNCDAEMON 20 /* Max. number async_daemons runable */
71 #ifdef _KERNEL
72 extern int nfs_niothreads; /* Number of async_daemons desired */
73 #endif
74 #define NFS_MAXGATHERDELAY 100 /* Max. write gather delay (msec) */
75 #ifndef NFS_GATHERDELAY
76 #define NFS_GATHERDELAY 10 /* Default write gather delay (msec) */
77 #endif
78
79 /*
80 * NFS_DIRBLKSIZ is the size of buffers in the buffer cache used for
81 * NFS directory vnodes. NFS_DIRFRAGSIZ is the minimum aligned amount
82 * of data in those buffers, and thus the minimum amount of data
83 * that you can request. NFS_DIRFRAGSIZ should be no smaller than
84 * DIRBLKSIZ.
85 */
86
87 #define NFS_DIRBLKSIZ 8192 /* Must be a multiple of DIRBLKSIZ */
88 #define NFS_DIRFRAGSIZ 512 /* Same as DIRBLKSIZ, generally */
89
90 /*
91 * Maximum number of directory entries cached per NFS node, to avoid
92 * having this grow without bounds on very large directories. The
93 * minimum size to get reasonable performance for emulated binaries
94 * is the maximum number of entries that fits in NFS_DIRBLKSIZ.
95 * For NFS_DIRBLKSIZ = 512, this would be 512 / 14 = 36.
96 */
97 #define NFS_MAXDIRCACHE (NFS_DIRBLKSIZ / 14)
98
99 /*
100 * Oddballs
101 */
102 #define NMOD(a) ((a) % nfs_asyncdaemons)
103 #define NFS_CMPFH(n, f, s) \
104 ((n)->n_fhsize == (s) && !memcmp((caddr_t)(n)->n_fhp, (caddr_t)(f), (s)))
105 #define NFS_ISV3(v) (VFSTONFS((v)->v_mount)->nm_flag & NFSMNT_NFSV3)
106 #define NFS_SRVMAXDATA(n) \
107 (((n)->nd_flag & ND_NFSV3) ? (((n)->nd_nam2) ? \
108 NFS_MAXDGRAMDATA : NFS_MAXDATA) : NFS_V2MAXDATA)
109
110 /*
111 * The VA_EXCLUSIVE flag should be added for va_vaflags and set for an
112 * exclusive create.
113 */
114 #ifndef VA_EXCLUSIVE
115 #define VA_EXCLUSIVE 0
116 #endif
117
118 /*
119 * The B_INVAFTERWRITE flag should be set to whatever is required by the
120 * buffer cache code to say "Invalidate the block after it is written back".
121 */
122 #define B_INVAFTERWRITE B_INVAL
123
124 /*
125 * The IO_METASYNC flag should be implemented for local file systems.
126 * (Until then, it is nothin at all.)
127 */
128 #ifndef IO_METASYNC
129 #define IO_METASYNC 0
130 #endif
131
132 /*
133 * Set the attribute timeout based on how recently the file has been modified.
134 */
135 #define NFS_ATTRTIMEO(np) \
136 ((((np)->n_flag & NMODIFIED) || \
137 (time.tv_sec - (np)->n_mtime) / 10 < NFS_MINATTRTIMO) ? NFS_MINATTRTIMO : \
138 ((time.tv_sec - (np)->n_mtime) / 10 > NFS_MAXATTRTIMO ? NFS_MAXATTRTIMO : \
139 (time.tv_sec - (np)->n_mtime) / 10))
140
141 /*
142 * Expected allocation sizes for major data structures. If the actual size
143 * of the structure exceeds these sizes, then malloc() will be allocating
144 * almost twice the memory required. This is used in nfs_init() to warn
145 * the sysadmin that the size of a structure should be reduced.
146 * (These sizes are always a power of 2. If the kernel malloc() changes
147 * to one that does not allocate space in powers of 2 size, then this all
148 * becomes bunk!)
149 */
150 #define NFS_NODEALLOC 256
151 #define NFS_MNTALLOC 512
152 #define NFS_SVCALLOC 256
153 #define NFS_UIDALLOC 128
154
155 /*
156 * Structures for the nfssvc(2) syscall. Not that anyone but nfsd and mount_nfs
157 * should ever try and use it.
158 */
159 struct nfsd_args {
160 int sock; /* Socket to serve */
161 caddr_t name; /* Client addr for connection based sockets */
162 int namelen; /* Length of name */
163 };
164
165 struct nfsd_srvargs {
166 struct nfsd *nsd_nfsd; /* Pointer to in kernel nfsd struct */
167 uid_t nsd_uid; /* Effective uid mapped to cred */
168 u_int32_t nsd_haddr; /* Ip address of client */
169 struct ucred nsd_cr; /* Cred. uid maps to */
170 int nsd_authlen; /* Length of auth string (ret) */
171 u_char *nsd_authstr; /* Auth string (ret) */
172 int nsd_verflen; /* and the verfier */
173 u_char *nsd_verfstr;
174 struct timeval nsd_timestamp; /* timestamp from verifier */
175 u_int32_t nsd_ttl; /* credential ttl (sec) */
176 NFSKERBKEY_T nsd_key; /* Session key */
177 };
178
179 struct nfsd_cargs {
180 char *ncd_dirp; /* Mount dir path */
181 uid_t ncd_authuid; /* Effective uid */
182 int ncd_authtype; /* Type of authenticator */
183 int ncd_authlen; /* Length of authenticator string */
184 u_char *ncd_authstr; /* Authenticator string */
185 int ncd_verflen; /* and the verifier */
186 u_char *ncd_verfstr;
187 NFSKERBKEY_T ncd_key; /* Session key */
188 };
189
190 /*
191 * Stats structure
192 */
193 struct nfsstats {
194 int attrcache_hits;
195 int attrcache_misses;
196 int lookupcache_hits;
197 int lookupcache_misses;
198 int direofcache_hits;
199 int direofcache_misses;
200 int biocache_reads;
201 int read_bios;
202 int read_physios;
203 int biocache_writes;
204 int write_bios;
205 int write_physios;
206 int biocache_readlinks;
207 int readlink_bios;
208 int biocache_readdirs;
209 int readdir_bios;
210 int rpccnt[NFS_NPROCS];
211 int rpcretries;
212 int srvrpccnt[NFS_NPROCS];
213 int srvrpc_errs;
214 int srv_errs;
215 int rpcrequests;
216 int rpctimeouts;
217 int rpcunexpected;
218 int rpcinvalid;
219 int srvcache_inproghits;
220 int srvcache_idemdonehits;
221 int srvcache_nonidemdonehits;
222 int srvcache_misses;
223 int srvnqnfs_leases;
224 int srvnqnfs_maxleases;
225 int srvnqnfs_getleases;
226 int srvvop_writes;
227 };
228
229 /*
230 * Flags for nfssvc() system call.
231 */
232 #define NFSSVC_BIOD 0x002
233 #define NFSSVC_NFSD 0x004
234 #define NFSSVC_ADDSOCK 0x008
235 #define NFSSVC_AUTHIN 0x010
236 #define NFSSVC_GOTAUTH 0x040
237 #define NFSSVC_AUTHINFAIL 0x080
238 #define NFSSVC_MNTD 0x100
239
240 /*
241 * fs.nfs sysctl(3) identifiers
242 */
243 #define NFS_NFSSTATS 1 /* struct: struct nfsstats */
244 #define NFS_IOTHREADS 2 /* number of io threads */
245 #define NFS_MAXID 3
246
247 #define NFS_NAMES { \
248 { 0, 0 }, \
249 { "nfsstats", CTLTYPE_STRUCT }, \
250 { "iothreads", CTLTYPE_INT }, \
251 }
252
253 /*
254 * The set of signals the interrupt an I/O in progress for NFSMNT_INT mounts.
255 * What should be in this set is open to debate, but I believe that since
256 * I/O system calls on ufs are never interrupted by signals the set should
257 * be minimal. My reasoning is that many current programs that use signals
258 * such as SIGALRM will not expect file I/O system calls to be interrupted
259 * by them and break.
260 */
261 #ifdef _KERNEL
262
263 struct uio; struct buf; struct vattr; struct nameidata; /* XXX */
264
265 /*
266 * Socket errors ignored for connectionless sockets??
267 * For now, ignore them all
268 */
269 #define NFSIGNORE_SOERROR(s, e) \
270 ((e) != EINTR && (e) != ERESTART && (e) != EWOULDBLOCK && \
271 ((s) & PR_CONNREQUIRED) == 0)
272
273 /*
274 * Nfs outstanding request list element
275 */
276 struct nfsreq {
277 TAILQ_ENTRY(nfsreq) r_chain;
278 struct mbuf *r_mreq;
279 struct mbuf *r_mrep;
280 struct mbuf *r_md;
281 caddr_t r_dpos;
282 struct nfsmount *r_nmp;
283 struct vnode *r_vp;
284 u_int32_t r_xid;
285 int r_flags; /* flags on request, see below */
286 int r_retry; /* max retransmission count */
287 int r_rexmit; /* current retrans count */
288 int r_timer; /* tick counter on reply */
289 u_int32_t r_procnum; /* NFS procedure number */
290 int r_rtt; /* RTT for rpc */
291 struct proc *r_procp; /* Proc that did I/O system call */
292 };
293
294 /*
295 * Queue head for nfsreq's
296 */
297 TAILQ_HEAD(, nfsreq) nfs_reqq;
298
299 /* Flag values for r_flags */
300 #define R_TIMING 0x01 /* timing request (in mntp) */
301 #define R_SENT 0x02 /* request has been sent */
302 #define R_SOFTTERM 0x04 /* soft mnt, too many retries */
303 #define R_INTR 0x08 /* intr mnt, signal pending */
304 #define R_SOCKERR 0x10 /* Fatal error on socket */
305 #define R_TPRINTFMSG 0x20 /* Did a tprintf msg. */
306 #define R_MUSTRESEND 0x40 /* Must resend request */
307 #define R_GETONEREP 0x80 /* Probe for one reply only */
308
309 /*
310 * A list of nfssvc_sock structures is maintained with all the sockets
311 * that require service by the nfsd.
312 * The nfsuid structs hang off of the nfssvc_sock structs in both lru
313 * and uid hash lists.
314 */
315 #ifndef NFS_UIDHASHSIZ
316 #define NFS_UIDHASHSIZ 29 /* Tune the size of nfssvc_sock with this */
317 #endif
318 #define NUIDHASH(sock, uid) \
319 (&(sock)->ns_uidhashtbl[(uid) % NFS_UIDHASHSIZ])
320 #ifndef NFS_WDELAYHASHSIZ
321 #define NFS_WDELAYHASHSIZ 16 /* and with this */
322 #endif
323 #define NWDELAYHASH(sock, f) \
324 (&(sock)->ns_wdelayhashtbl[(*((u_int32_t *)(f))) % NFS_WDELAYHASHSIZ])
325 #ifndef NFS_MUIDHASHSIZ
326 #define NFS_MUIDHASHSIZ 63 /* Tune the size of nfsmount with this */
327 #endif
328 #define NMUIDHASH(nmp, uid) \
329 (&(nmp)->nm_uidhashtbl[(uid) % NFS_MUIDHASHSIZ])
330 #define NFSNOHASH(fhsum) \
331 (&nfsnodehashtbl[(fhsum) & nfsnodehash])
332
333 #ifndef NFS_DIRHASHSIZ
334 #define NFS_DIRHASHSIZ 64
335 #endif
336 #define NFSDIRHASH(np, off) \
337 (&np->n_dircache[(nfs_dirhash((off)) & nfsdirhashmask)])
338
339 /*
340 * Macros for storing/retrieving cookies into directory buffers.
341 */
342 #define NFS_STASHCOOKIE(dp,off) \
343 *((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))) = off
344 #define NFS_GETCOOKIE(dp) \
345 (*((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))))
346 #define NFS_STASHCOOKIE32(dp, val) \
347 *((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
348 sizeof (int))) = val
349 #define NFS_GETCOOKIE32(dp) \
350 (*((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
351 sizeof (int))))
352
353 /*
354 * Flags passed to nfs_bioread().
355 */
356 #define NFSBIO_CACHECOOKIES 0x0001 /* Cache dir offset cookies */
357
358 /*
359 * Network address hash list element
360 */
361 union nethostaddr {
362 u_int32_t had_inetaddr;
363 struct mbuf *had_nam;
364 };
365
366 struct nfsuid {
367 TAILQ_ENTRY(nfsuid) nu_lru; /* LRU chain */
368 LIST_ENTRY(nfsuid) nu_hash; /* Hash list */
369 int nu_flag; /* Flags */
370 union nethostaddr nu_haddr; /* Host addr. for dgram sockets */
371 struct ucred nu_cr; /* Cred uid mapped to */
372 int nu_expire; /* Expiry time (sec) */
373 struct timeval nu_timestamp; /* Kerb. timestamp */
374 u_int32_t nu_nickname; /* Nickname on server */
375 NFSKERBKEY_T nu_key; /* and session key */
376 };
377
378 #define nu_inetaddr nu_haddr.had_inetaddr
379 #define nu_nam nu_haddr.had_nam
380 /* Bits for nu_flag */
381 #define NU_INETADDR 0x1
382 #define NU_NAM 0x2
383 #ifdef INET6
384 #define NU_NETFAM(u) \
385 (((u)->nu_flag & NU_INETADDR) ? \
386 (((u)->nu_flag & NU_NAM) ? AF_INET6 : AF_INET) : AF_ISO)
387 #else
388 #define NU_NETFAM(u) (((u)->nu_flag & NU_INETADDR) ? AF_INET : AF_ISO)
389 #endif
390
391 struct nfssvc_sock {
392 TAILQ_ENTRY(nfssvc_sock) ns_chain; /* List of all nfssvc_sock's */
393 TAILQ_HEAD(, nfsuid) ns_uidlruhead;
394 struct file *ns_fp;
395 struct socket *ns_so;
396 struct mbuf *ns_nam;
397 struct mbuf *ns_raw;
398 struct mbuf *ns_rawend;
399 struct mbuf *ns_rec;
400 struct mbuf *ns_recend;
401 struct mbuf *ns_frag;
402 int ns_flag;
403 int ns_solock;
404 int ns_cc;
405 int ns_reclen;
406 int ns_numuids;
407 u_int32_t ns_sref;
408 LIST_HEAD(, nfsrv_descript) ns_tq; /* Write gather lists */
409 LIST_HEAD(, nfsuid) ns_uidhashtbl[NFS_UIDHASHSIZ];
410 LIST_HEAD(nfsrvw_delayhash, nfsrv_descript) ns_wdelayhashtbl[NFS_WDELAYHASHSIZ];
411 };
412
413 /* Bits for "ns_flag" */
414 #define SLP_VALID 0x01
415 #define SLP_DOREC 0x02
416 #define SLP_NEEDQ 0x04
417 #define SLP_DISCONN 0x08
418 #define SLP_GETSTREAM 0x10
419 #define SLP_LASTFRAG 0x20
420 #define SLP_ALLFLAGS 0xff
421
422 TAILQ_HEAD(, nfssvc_sock) nfssvc_sockhead;
423 int nfssvc_sockhead_flag;
424 #define SLP_INIT 0x01
425 #define SLP_WANTINIT 0x02
426
427 /*
428 * One of these structures is allocated for each nfsd.
429 */
430 struct nfsd {
431 TAILQ_ENTRY(nfsd) nfsd_chain; /* List of all nfsd's */
432 int nfsd_flag; /* NFSD_ flags */
433 struct nfssvc_sock *nfsd_slp; /* Current socket */
434 int nfsd_authlen; /* Authenticator len */
435 u_char nfsd_authstr[RPCAUTH_MAXSIZ]; /* Authenticator data */
436 int nfsd_verflen; /* and the Verifier */
437 u_char nfsd_verfstr[RPCVERF_MAXSIZ];
438 struct proc *nfsd_procp; /* Proc ptr */
439 struct nfsrv_descript *nfsd_nd; /* Associated nfsrv_descript */
440 };
441
442 /* Bits for "nfsd_flag" */
443 #define NFSD_WAITING 0x01
444 #define NFSD_REQINPROG 0x02
445 #define NFSD_NEEDAUTH 0x04
446 #define NFSD_AUTHFAIL 0x08
447
448 /*
449 * This structure is used by the server for describing each request.
450 * Some fields are used only when write request gathering is performed.
451 */
452 struct nfsrv_descript {
453 u_quad_t nd_time; /* Write deadline (usec) */
454 off_t nd_off; /* Start byte offset */
455 off_t nd_eoff; /* and end byte offset */
456 LIST_ENTRY(nfsrv_descript) nd_hash; /* Hash list */
457 LIST_ENTRY(nfsrv_descript) nd_tq; /* and timer list */
458 LIST_HEAD(,nfsrv_descript) nd_coalesce; /* coalesced writes */
459 struct mbuf *nd_mrep; /* Request mbuf list */
460 struct mbuf *nd_md; /* Current dissect mbuf */
461 struct mbuf *nd_mreq; /* Reply mbuf list */
462 struct mbuf *nd_nam; /* and socket addr */
463 struct mbuf *nd_nam2; /* return socket addr */
464 caddr_t nd_dpos; /* Current dissect pos */
465 u_int32_t nd_procnum; /* RPC # */
466 int nd_stable; /* storage type */
467 int nd_flag; /* nd_flag */
468 int nd_len; /* Length of this write */
469 int nd_repstat; /* Reply status */
470 u_int32_t nd_retxid; /* Reply xid */
471 u_int32_t nd_duration; /* Lease duration */
472 struct timeval nd_starttime; /* Time RPC initiated */
473 fhandle_t nd_fh; /* File handle */
474 struct ucred nd_cr; /* Credentials */
475 };
476
477 /* Bits for "nd_flag" */
478 #define ND_READ LEASE_READ
479 #define ND_WRITE LEASE_WRITE
480 #define ND_CHECK 0x04
481 #define ND_LEASE (ND_READ | ND_WRITE | ND_CHECK)
482 #define ND_NFSV3 0x08
483 #define ND_NQNFS 0x10
484 #define ND_KERBNICK 0x20
485 #define ND_KERBFULL 0x40
486 #define ND_KERBAUTH (ND_KERBNICK | ND_KERBFULL)
487
488 TAILQ_HEAD(, nfsd) nfsd_head;
489 int nfsd_head_flag;
490 #define NFSD_CHECKSLP 0x01
491
492 /*
493 * These macros compare nfsrv_descript structures.
494 */
495 #define NFSW_CONTIG(o, n) \
496 ((o)->nd_eoff >= (n)->nd_off && \
497 !memcmp((caddr_t)&(o)->nd_fh, (caddr_t)&(n)->nd_fh, NFSX_V3FH))
498
499 #define NFSW_SAMECRED(o, n) \
500 (((o)->nd_flag & ND_KERBAUTH) == ((n)->nd_flag & ND_KERBAUTH) && \
501 !memcmp((caddr_t)&(o)->nd_cr, (caddr_t)&(n)->nd_cr, \
502 sizeof (struct ucred)))
503
504 /*
505 * Defines for WebNFS
506 */
507
508 #define WEBNFS_ESC_CHAR '%'
509 #define WEBNFS_SPECCHAR_START 0x80
510
511 #define WEBNFS_NATIVE_CHAR 0x80
512 /*
513 * ..
514 * Possibly more here in the future.
515 */
516
517 /*
518 * Macro for converting escape characters in WebNFS pathnames.
519 * Should really be in libkern.
520 */
521
522 #define HEXTOC(c) \
523 ((c) >= 'a' ? ((c) - ('a' - 10)) : \
524 ((c) >= 'A' ? ((c) - ('A' - 10)) : ((c) - '0')))
525 #define HEXSTRTOI(p) \
526 ((HEXTOC(p[0]) << 4) + HEXTOC(p[1]))
527 #endif /* _KERNEL */
528
529 #endif /* _NFS_NFS_H */
530