nfs.h revision 1.24 1 /* $NetBSD: nfs.h,v 1.24 2000/09/19 22:05:55 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 32768 /* Def. write data size */
65 #define NFS_RSIZE 32768 /* Def. read data size */
66 #define NFS_READDIRSIZE 8192 /* Def. readdir size */
67 #define NFS_DEFRAHEAD 2 /* 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 #ifdef NFS_V2_ONLY
106 #define NFS_ISV3(v) (0)
107 #else
108 #define NFS_ISV3(v) (VFSTONFS((v)->v_mount)->nm_flag & NFSMNT_NFSV3)
109 #endif
110 #define NFS_SRVMAXDATA(n) \
111 (((n)->nd_flag & ND_NFSV3) ? (((n)->nd_nam2) ? \
112 NFS_MAXDGRAMDATA : NFS_MAXDATA) : NFS_V2MAXDATA)
113
114 /*
115 * The VA_EXCLUSIVE flag should be added for va_vaflags and set for an
116 * exclusive create.
117 */
118 #ifndef VA_EXCLUSIVE
119 #define VA_EXCLUSIVE 0
120 #endif
121
122 /*
123 * The B_INVAFTERWRITE flag should be set to whatever is required by the
124 * buffer cache code to say "Invalidate the block after it is written back".
125 */
126 #define B_INVAFTERWRITE B_INVAL
127
128 /*
129 * The IO_METASYNC flag should be implemented for local file systems.
130 * (Until then, it is nothin at all.)
131 */
132 #ifndef IO_METASYNC
133 #define IO_METASYNC 0
134 #endif
135
136 /*
137 * Set the attribute timeout based on how recently the file has been modified.
138 */
139 #define NFS_ATTRTIMEO(np) \
140 ((((np)->n_flag & NMODIFIED) || \
141 (time.tv_sec - (np)->n_mtime) / 10 < NFS_MINATTRTIMO) ? NFS_MINATTRTIMO : \
142 ((time.tv_sec - (np)->n_mtime) / 10 > NFS_MAXATTRTIMO ? NFS_MAXATTRTIMO : \
143 (time.tv_sec - (np)->n_mtime) / 10))
144
145 /*
146 * Expected allocation sizes for major data structures. If the actual size
147 * of the structure exceeds these sizes, then malloc() will be allocating
148 * almost twice the memory required. This is used in nfs_init() to warn
149 * the sysadmin that the size of a structure should be reduced.
150 * (These sizes are always a power of 2. If the kernel malloc() changes
151 * to one that does not allocate space in powers of 2 size, then this all
152 * becomes bunk!)
153 */
154 #define NFS_NODEALLOC 256
155 #define NFS_MNTALLOC 512
156 #define NFS_SVCALLOC 256
157 #define NFS_UIDALLOC 128
158
159 /*
160 * Structures for the nfssvc(2) syscall. Not that anyone but nfsd and mount_nfs
161 * should ever try and use it.
162 */
163 struct nfsd_args {
164 int sock; /* Socket to serve */
165 caddr_t name; /* Client addr for connection based sockets */
166 int namelen; /* Length of name */
167 };
168
169 struct nfsd_srvargs {
170 struct nfsd *nsd_nfsd; /* Pointer to in kernel nfsd struct */
171 uid_t nsd_uid; /* Effective uid mapped to cred */
172 u_int32_t nsd_haddr; /* Ip address of client */
173 struct ucred nsd_cr; /* Cred. uid maps to */
174 int nsd_authlen; /* Length of auth string (ret) */
175 u_char *nsd_authstr; /* Auth string (ret) */
176 int nsd_verflen; /* and the verfier */
177 u_char *nsd_verfstr;
178 struct timeval nsd_timestamp; /* timestamp from verifier */
179 u_int32_t nsd_ttl; /* credential ttl (sec) */
180 NFSKERBKEY_T nsd_key; /* Session key */
181 };
182
183 struct nfsd_cargs {
184 char *ncd_dirp; /* Mount dir path */
185 uid_t ncd_authuid; /* Effective uid */
186 int ncd_authtype; /* Type of authenticator */
187 int ncd_authlen; /* Length of authenticator string */
188 u_char *ncd_authstr; /* Authenticator string */
189 int ncd_verflen; /* and the verifier */
190 u_char *ncd_verfstr;
191 NFSKERBKEY_T ncd_key; /* Session key */
192 };
193
194 /*
195 * Stats structure
196 */
197 struct nfsstats {
198 int attrcache_hits;
199 int attrcache_misses;
200 int lookupcache_hits;
201 int lookupcache_misses;
202 int direofcache_hits;
203 int direofcache_misses;
204 int biocache_reads;
205 int read_bios;
206 int read_physios;
207 int biocache_writes;
208 int write_bios;
209 int write_physios;
210 int biocache_readlinks;
211 int readlink_bios;
212 int biocache_readdirs;
213 int readdir_bios;
214 int rpccnt[NFS_NPROCS];
215 int rpcretries;
216 int srvrpccnt[NFS_NPROCS];
217 int srvrpc_errs;
218 int srv_errs;
219 int rpcrequests;
220 int rpctimeouts;
221 int rpcunexpected;
222 int rpcinvalid;
223 int srvcache_inproghits;
224 int srvcache_idemdonehits;
225 int srvcache_nonidemdonehits;
226 int srvcache_misses;
227 int srvnqnfs_leases;
228 int srvnqnfs_maxleases;
229 int srvnqnfs_getleases;
230 int srvvop_writes;
231 };
232
233 /*
234 * Flags for nfssvc() system call.
235 */
236 #define NFSSVC_BIOD 0x002
237 #define NFSSVC_NFSD 0x004
238 #define NFSSVC_ADDSOCK 0x008
239 #define NFSSVC_AUTHIN 0x010
240 #define NFSSVC_GOTAUTH 0x040
241 #define NFSSVC_AUTHINFAIL 0x080
242 #define NFSSVC_MNTD 0x100
243
244 /*
245 * fs.nfs sysctl(3) identifiers
246 */
247 #define NFS_NFSSTATS 1 /* struct: struct nfsstats */
248 #define NFS_IOTHREADS 2 /* number of io threads */
249 #define NFS_MAXID 3
250
251 #define NFS_NAMES { \
252 { 0, 0 }, \
253 { "nfsstats", CTLTYPE_STRUCT }, \
254 { "iothreads", CTLTYPE_INT }, \
255 }
256
257 /*
258 * The set of signals the interrupt an I/O in progress for NFSMNT_INT mounts.
259 * What should be in this set is open to debate, but I believe that since
260 * I/O system calls on ufs are never interrupted by signals the set should
261 * be minimal. My reasoning is that many current programs that use signals
262 * such as SIGALRM will not expect file I/O system calls to be interrupted
263 * by them and break.
264 */
265 #ifdef _KERNEL
266
267 struct uio; struct buf; struct vattr; struct nameidata; /* XXX */
268
269 /*
270 * Socket errors ignored for connectionless sockets??
271 * For now, ignore them all
272 */
273 #define NFSIGNORE_SOERROR(s, e) \
274 ((e) != EINTR && (e) != ERESTART && (e) != EWOULDBLOCK && \
275 ((s) & PR_CONNREQUIRED) == 0)
276
277 /*
278 * Nfs outstanding request list element
279 */
280 struct nfsreq {
281 TAILQ_ENTRY(nfsreq) r_chain;
282 struct mbuf *r_mreq;
283 struct mbuf *r_mrep;
284 struct mbuf *r_md;
285 caddr_t r_dpos;
286 struct nfsmount *r_nmp;
287 struct vnode *r_vp;
288 u_int32_t r_xid;
289 int r_flags; /* flags on request, see below */
290 int r_retry; /* max retransmission count */
291 int r_rexmit; /* current retrans count */
292 int r_timer; /* tick counter on reply */
293 u_int32_t r_procnum; /* NFS procedure number */
294 int r_rtt; /* RTT for rpc */
295 struct proc *r_procp; /* Proc that did I/O system call */
296 };
297
298 /*
299 * Queue head for nfsreq's
300 */
301 TAILQ_HEAD(, nfsreq) nfs_reqq;
302
303 /* Flag values for r_flags */
304 #define R_TIMING 0x01 /* timing request (in mntp) */
305 #define R_SENT 0x02 /* request has been sent */
306 #define R_SOFTTERM 0x04 /* soft mnt, too many retries */
307 #define R_INTR 0x08 /* intr mnt, signal pending */
308 #define R_SOCKERR 0x10 /* Fatal error on socket */
309 #define R_TPRINTFMSG 0x20 /* Did a tprintf msg. */
310 #define R_MUSTRESEND 0x40 /* Must resend request */
311 #define R_GETONEREP 0x80 /* Probe for one reply only */
312
313 /*
314 * A list of nfssvc_sock structures is maintained with all the sockets
315 * that require service by the nfsd.
316 * The nfsuid structs hang off of the nfssvc_sock structs in both lru
317 * and uid hash lists.
318 */
319 #ifndef NFS_UIDHASHSIZ
320 #define NFS_UIDHASHSIZ 29 /* Tune the size of nfssvc_sock with this */
321 #endif
322 #define NUIDHASH(sock, uid) \
323 (&(sock)->ns_uidhashtbl[(uid) % NFS_UIDHASHSIZ])
324 #ifndef NFS_WDELAYHASHSIZ
325 #define NFS_WDELAYHASHSIZ 16 /* and with this */
326 #endif
327 #define NWDELAYHASH(sock, f) \
328 (&(sock)->ns_wdelayhashtbl[(*((u_int32_t *)(f))) % NFS_WDELAYHASHSIZ])
329 #ifndef NFS_MUIDHASHSIZ
330 #define NFS_MUIDHASHSIZ 63 /* Tune the size of nfsmount with this */
331 #endif
332 #define NMUIDHASH(nmp, uid) \
333 (&(nmp)->nm_uidhashtbl[(uid) % NFS_MUIDHASHSIZ])
334 #define NFSNOHASH(fhsum) \
335 (&nfsnodehashtbl[(fhsum) & nfsnodehash])
336
337 #ifndef NFS_DIRHASHSIZ
338 #define NFS_DIRHASHSIZ 64
339 #endif
340 #define NFSDIRHASH(np, off) \
341 (&np->n_dircache[(nfs_dirhash((off)) & nfsdirhashmask)])
342
343 /*
344 * Macros for storing/retrieving cookies into directory buffers.
345 */
346 #define NFS_STASHCOOKIE(dp,off) \
347 *((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))) = off
348 #define NFS_GETCOOKIE(dp) \
349 (*((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))))
350 #define NFS_STASHCOOKIE32(dp, val) \
351 *((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
352 sizeof (int))) = val
353 #define NFS_GETCOOKIE32(dp) \
354 (*((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
355 sizeof (int))))
356
357 /*
358 * Flags passed to nfs_bioread().
359 */
360 #define NFSBIO_CACHECOOKIES 0x0001 /* Cache dir offset cookies */
361
362 /*
363 * Network address hash list element
364 */
365 union nethostaddr {
366 u_int32_t had_inetaddr;
367 struct mbuf *had_nam;
368 };
369
370 struct nfsuid {
371 TAILQ_ENTRY(nfsuid) nu_lru; /* LRU chain */
372 LIST_ENTRY(nfsuid) nu_hash; /* Hash list */
373 int nu_flag; /* Flags */
374 union nethostaddr nu_haddr; /* Host addr. for dgram sockets */
375 struct ucred nu_cr; /* Cred uid mapped to */
376 int nu_expire; /* Expiry time (sec) */
377 struct timeval nu_timestamp; /* Kerb. timestamp */
378 u_int32_t nu_nickname; /* Nickname on server */
379 NFSKERBKEY_T nu_key; /* and session key */
380 };
381
382 #define nu_inetaddr nu_haddr.had_inetaddr
383 #define nu_nam nu_haddr.had_nam
384 /* Bits for nu_flag */
385 #define NU_INETADDR 0x1
386 #define NU_NAM 0x2
387 #ifdef INET6
388 #define NU_NETFAM(u) \
389 (((u)->nu_flag & NU_INETADDR) ? \
390 (((u)->nu_flag & NU_NAM) ? AF_INET6 : AF_INET) : AF_ISO)
391 #else
392 #define NU_NETFAM(u) (((u)->nu_flag & NU_INETADDR) ? AF_INET : AF_ISO)
393 #endif
394
395 struct nfssvc_sock {
396 TAILQ_ENTRY(nfssvc_sock) ns_chain; /* List of all nfssvc_sock's */
397 TAILQ_HEAD(, nfsuid) ns_uidlruhead;
398 struct file *ns_fp;
399 struct socket *ns_so;
400 struct mbuf *ns_nam;
401 struct mbuf *ns_raw;
402 struct mbuf *ns_rawend;
403 struct mbuf *ns_rec;
404 struct mbuf *ns_recend;
405 struct mbuf *ns_frag;
406 int ns_flag;
407 int ns_solock;
408 int ns_cc;
409 int ns_reclen;
410 int ns_numuids;
411 u_int32_t ns_sref;
412 LIST_HEAD(, nfsrv_descript) ns_tq; /* Write gather lists */
413 LIST_HEAD(, nfsuid) ns_uidhashtbl[NFS_UIDHASHSIZ];
414 LIST_HEAD(nfsrvw_delayhash, nfsrv_descript) ns_wdelayhashtbl[NFS_WDELAYHASHSIZ];
415 };
416
417 /* Bits for "ns_flag" */
418 #define SLP_VALID 0x01
419 #define SLP_DOREC 0x02
420 #define SLP_NEEDQ 0x04
421 #define SLP_DISCONN 0x08
422 #define SLP_GETSTREAM 0x10
423 #define SLP_LASTFRAG 0x20
424 #define SLP_ALLFLAGS 0xff
425
426 TAILQ_HEAD(, nfssvc_sock) nfssvc_sockhead;
427 int nfssvc_sockhead_flag;
428 #define SLP_INIT 0x01
429 #define SLP_WANTINIT 0x02
430
431 /*
432 * One of these structures is allocated for each nfsd.
433 */
434 struct nfsd {
435 TAILQ_ENTRY(nfsd) nfsd_chain; /* List of all nfsd's */
436 int nfsd_flag; /* NFSD_ flags */
437 struct nfssvc_sock *nfsd_slp; /* Current socket */
438 int nfsd_authlen; /* Authenticator len */
439 u_char nfsd_authstr[RPCAUTH_MAXSIZ]; /* Authenticator data */
440 int nfsd_verflen; /* and the Verifier */
441 u_char nfsd_verfstr[RPCVERF_MAXSIZ];
442 struct proc *nfsd_procp; /* Proc ptr */
443 struct nfsrv_descript *nfsd_nd; /* Associated nfsrv_descript */
444 };
445
446 /* Bits for "nfsd_flag" */
447 #define NFSD_WAITING 0x01
448 #define NFSD_REQINPROG 0x02
449 #define NFSD_NEEDAUTH 0x04
450 #define NFSD_AUTHFAIL 0x08
451
452 /*
453 * This structure is used by the server for describing each request.
454 * Some fields are used only when write request gathering is performed.
455 */
456 struct nfsrv_descript {
457 u_quad_t nd_time; /* Write deadline (usec) */
458 off_t nd_off; /* Start byte offset */
459 off_t nd_eoff; /* and end byte offset */
460 LIST_ENTRY(nfsrv_descript) nd_hash; /* Hash list */
461 LIST_ENTRY(nfsrv_descript) nd_tq; /* and timer list */
462 LIST_HEAD(,nfsrv_descript) nd_coalesce; /* coalesced writes */
463 struct mbuf *nd_mrep; /* Request mbuf list */
464 struct mbuf *nd_md; /* Current dissect mbuf */
465 struct mbuf *nd_mreq; /* Reply mbuf list */
466 struct mbuf *nd_nam; /* and socket addr */
467 struct mbuf *nd_nam2; /* return socket addr */
468 caddr_t nd_dpos; /* Current dissect pos */
469 u_int32_t nd_procnum; /* RPC # */
470 int nd_stable; /* storage type */
471 int nd_flag; /* nd_flag */
472 int nd_len; /* Length of this write */
473 int nd_repstat; /* Reply status */
474 u_int32_t nd_retxid; /* Reply xid */
475 u_int32_t nd_duration; /* Lease duration */
476 struct timeval nd_starttime; /* Time RPC initiated */
477 fhandle_t nd_fh; /* File handle */
478 struct ucred nd_cr; /* Credentials */
479 };
480
481 /* Bits for "nd_flag" */
482 #define ND_READ LEASE_READ
483 #define ND_WRITE LEASE_WRITE
484 #define ND_CHECK 0x04
485 #define ND_LEASE (ND_READ | ND_WRITE | ND_CHECK)
486 #define ND_NFSV3 0x08
487 #define ND_NQNFS 0x10
488 #define ND_KERBNICK 0x20
489 #define ND_KERBFULL 0x40
490 #define ND_KERBAUTH (ND_KERBNICK | ND_KERBFULL)
491
492 TAILQ_HEAD(, nfsd) nfsd_head;
493 int nfsd_head_flag;
494 #define NFSD_CHECKSLP 0x01
495
496 /*
497 * These macros compare nfsrv_descript structures.
498 */
499 #define NFSW_CONTIG(o, n) \
500 ((o)->nd_eoff >= (n)->nd_off && \
501 !memcmp((caddr_t)&(o)->nd_fh, (caddr_t)&(n)->nd_fh, NFSX_V3FH))
502
503 #define NFSW_SAMECRED(o, n) \
504 (((o)->nd_flag & ND_KERBAUTH) == ((n)->nd_flag & ND_KERBAUTH) && \
505 !memcmp((caddr_t)&(o)->nd_cr, (caddr_t)&(n)->nd_cr, \
506 sizeof (struct ucred)))
507
508 /*
509 * Defines for WebNFS
510 */
511
512 #define WEBNFS_ESC_CHAR '%'
513 #define WEBNFS_SPECCHAR_START 0x80
514
515 #define WEBNFS_NATIVE_CHAR 0x80
516 /*
517 * ..
518 * Possibly more here in the future.
519 */
520
521 /*
522 * Macro for converting escape characters in WebNFS pathnames.
523 * Should really be in libkern.
524 */
525
526 #define HEXTOC(c) \
527 ((c) >= 'a' ? ((c) - ('a' - 10)) : \
528 ((c) >= 'A' ? ((c) - ('A' - 10)) : ((c) - '0')))
529 #define HEXSTRTOI(p) \
530 ((HEXTOC(p[0]) << 4) + HEXTOC(p[1]))
531 #endif /* _KERNEL */
532
533 #endif /* _NFS_NFS_H */
534