nfs.h revision 1.18 1 /* $NetBSD: nfs.h,v 1.18 1998/08/09 21:19:49 perry 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 #define NFS_MAXGATHERDELAY 100 /* Max. write gather delay (msec) */
72 #ifndef NFS_GATHERDELAY
73 #define NFS_GATHERDELAY 10 /* Default write gather delay (msec) */
74 #endif
75
76 /*
77 * NFS_DIRBLKSIZ is the size of buffers in the buffer cache used for
78 * NFS directory vnodes. NFS_DIRFRAGSIZ is the minimum aligned amount
79 * of data in those buffers, and thus the minimum amount of data
80 * that you can request. NFS_DIRFRAGSIZ should be no smaller than
81 * DIRBLKSIZ.
82 */
83
84 #define NFS_DIRBLKSIZ 8192 /* Must be a multiple of DIRBLKSIZ */
85 #define NFS_DIRFRAGSIZ 512 /* Same as DIRBLKSIZ, generally */
86
87 /*
88 * Maximum number of directory entries cached per NFS node, to avoid
89 * having this grow without bounds on very large directories. The
90 * minimum size to get reasonable performance for emulated binaries
91 * is the maximum number of entries that fits in NFS_DIRBLKSIZ.
92 * For NFS_DIRBLKSIZ = 512, this would be 512 / 14 = 36.
93 */
94 #define NFS_MAXDIRCACHE (NFS_DIRBLKSIZ / 14)
95
96 /*
97 * Oddballs
98 */
99 #define NMOD(a) ((a) % nfs_asyncdaemons)
100 #define NFS_CMPFH(n, f, s) \
101 ((n)->n_fhsize == (s) && !memcmp((caddr_t)(n)->n_fhp, (caddr_t)(f), (s)))
102 #define NFS_ISV3(v) (VFSTONFS((v)->v_mount)->nm_flag & NFSMNT_NFSV3)
103 #define NFS_SRVMAXDATA(n) \
104 (((n)->nd_flag & ND_NFSV3) ? (((n)->nd_nam2) ? \
105 NFS_MAXDGRAMDATA : NFS_MAXDATA) : NFS_V2MAXDATA)
106
107 /*
108 * The VA_EXCLUSIVE flag should be added for va_vaflags and set for an
109 * exclusive create.
110 */
111 #ifndef VA_EXCLUSIVE
112 #define VA_EXCLUSIVE 0
113 #endif
114
115 /*
116 * The B_INVAFTERWRITE flag should be set to whatever is required by the
117 * buffer cache code to say "Invalidate the block after it is written back".
118 */
119 #define B_INVAFTERWRITE B_INVAL
120
121 /*
122 * The IO_METASYNC flag should be implemented for local file systems.
123 * (Until then, it is nothin at all.)
124 */
125 #ifndef IO_METASYNC
126 #define IO_METASYNC 0
127 #endif
128
129 /*
130 * Set the attribute timeout based on how recently the file has been modified.
131 */
132 #define NFS_ATTRTIMEO(np) \
133 ((((np)->n_flag & NMODIFIED) || \
134 (time.tv_sec - (np)->n_mtime) / 10 < NFS_MINATTRTIMO) ? NFS_MINATTRTIMO : \
135 ((time.tv_sec - (np)->n_mtime) / 10 > NFS_MAXATTRTIMO ? NFS_MAXATTRTIMO : \
136 (time.tv_sec - (np)->n_mtime) / 10))
137
138 /*
139 * Expected allocation sizes for major data structures. If the actual size
140 * of the structure exceeds these sizes, then malloc() will be allocating
141 * almost twice the memory required. This is used in nfs_init() to warn
142 * the sysadmin that the size of a structure should be reduced.
143 * (These sizes are always a power of 2. If the kernel malloc() changes
144 * to one that does not allocate space in powers of 2 size, then this all
145 * becomes bunk!)
146 */
147 #define NFS_NODEALLOC 256
148 #define NFS_MNTALLOC 512
149 #define NFS_SVCALLOC 256
150 #define NFS_UIDALLOC 128
151
152 /*
153 * Structures for the nfssvc(2) syscall. Not that anyone but nfsd and mount_nfs
154 * should ever try and use it.
155 */
156 struct nfsd_args {
157 int sock; /* Socket to serve */
158 caddr_t name; /* Client addr for connection based sockets */
159 int namelen; /* Length of name */
160 };
161
162 struct nfsd_srvargs {
163 struct nfsd *nsd_nfsd; /* Pointer to in kernel nfsd struct */
164 uid_t nsd_uid; /* Effective uid mapped to cred */
165 u_int32_t nsd_haddr; /* Ip address of client */
166 struct ucred nsd_cr; /* Cred. uid maps to */
167 int nsd_authlen; /* Length of auth string (ret) */
168 u_char *nsd_authstr; /* Auth string (ret) */
169 int nsd_verflen; /* and the verfier */
170 u_char *nsd_verfstr;
171 struct timeval nsd_timestamp; /* timestamp from verifier */
172 u_int32_t nsd_ttl; /* credential ttl (sec) */
173 NFSKERBKEY_T nsd_key; /* Session key */
174 };
175
176 struct nfsd_cargs {
177 char *ncd_dirp; /* Mount dir path */
178 uid_t ncd_authuid; /* Effective uid */
179 int ncd_authtype; /* Type of authenticator */
180 int ncd_authlen; /* Length of authenticator string */
181 u_char *ncd_authstr; /* Authenticator string */
182 int ncd_verflen; /* and the verifier */
183 u_char *ncd_verfstr;
184 NFSKERBKEY_T ncd_key; /* Session key */
185 };
186
187 /*
188 * Stats structure
189 */
190 struct nfsstats {
191 int attrcache_hits;
192 int attrcache_misses;
193 int lookupcache_hits;
194 int lookupcache_misses;
195 int direofcache_hits;
196 int direofcache_misses;
197 int biocache_reads;
198 int read_bios;
199 int read_physios;
200 int biocache_writes;
201 int write_bios;
202 int write_physios;
203 int biocache_readlinks;
204 int readlink_bios;
205 int biocache_readdirs;
206 int readdir_bios;
207 int rpccnt[NFS_NPROCS];
208 int rpcretries;
209 int srvrpccnt[NFS_NPROCS];
210 int srvrpc_errs;
211 int srv_errs;
212 int rpcrequests;
213 int rpctimeouts;
214 int rpcunexpected;
215 int rpcinvalid;
216 int srvcache_inproghits;
217 int srvcache_idemdonehits;
218 int srvcache_nonidemdonehits;
219 int srvcache_misses;
220 int srvnqnfs_leases;
221 int srvnqnfs_maxleases;
222 int srvnqnfs_getleases;
223 int srvvop_writes;
224 };
225
226 /*
227 * Flags for nfssvc() system call.
228 */
229 #define NFSSVC_BIOD 0x002
230 #define NFSSVC_NFSD 0x004
231 #define NFSSVC_ADDSOCK 0x008
232 #define NFSSVC_AUTHIN 0x010
233 #define NFSSVC_GOTAUTH 0x040
234 #define NFSSVC_AUTHINFAIL 0x080
235 #define NFSSVC_MNTD 0x100
236
237 /*
238 * fs.nfs sysctl(3) identifiers
239 */
240 #define NFS_NFSSTATS 1 /* struct: struct nfsstats */
241
242 #define FS_NFS_NAMES { \
243 { 0, 0 }, \
244 { "nfsstats", CTLTYPE_STRUCT }, \
245 }
246
247 /*
248 * The set of signals the interrupt an I/O in progress for NFSMNT_INT mounts.
249 * What should be in this set is open to debate, but I believe that since
250 * I/O system calls on ufs are never interrupted by signals the set should
251 * be minimal. My reasoning is that many current programs that use signals
252 * such as SIGALRM will not expect file I/O system calls to be interrupted
253 * by them and break.
254 */
255 #ifdef _KERNEL
256
257 struct uio; struct buf; struct vattr; struct nameidata; /* XXX */
258
259 #define NFSINT_SIGMASK (sigmask(SIGINT)|sigmask(SIGTERM)|sigmask(SIGKILL)| \
260 sigmask(SIGHUP)|sigmask(SIGQUIT))
261
262 /*
263 * Socket errors ignored for connectionless sockets??
264 * For now, ignore them all
265 */
266 #define NFSIGNORE_SOERROR(s, e) \
267 ((e) != EINTR && (e) != ERESTART && (e) != EWOULDBLOCK && \
268 ((s) & PR_CONNREQUIRED) == 0)
269
270 /*
271 * Nfs outstanding request list element
272 */
273 struct nfsreq {
274 TAILQ_ENTRY(nfsreq) r_chain;
275 struct mbuf *r_mreq;
276 struct mbuf *r_mrep;
277 struct mbuf *r_md;
278 caddr_t r_dpos;
279 struct nfsmount *r_nmp;
280 struct vnode *r_vp;
281 u_int32_t r_xid;
282 int r_flags; /* flags on request, see below */
283 int r_retry; /* max retransmission count */
284 int r_rexmit; /* current retrans count */
285 int r_timer; /* tick counter on reply */
286 u_int32_t r_procnum; /* NFS procedure number */
287 int r_rtt; /* RTT for rpc */
288 struct proc *r_procp; /* Proc that did I/O system call */
289 };
290
291 /*
292 * Queue head for nfsreq's
293 */
294 TAILQ_HEAD(, nfsreq) nfs_reqq;
295
296 /* Flag values for r_flags */
297 #define R_TIMING 0x01 /* timing request (in mntp) */
298 #define R_SENT 0x02 /* request has been sent */
299 #define R_SOFTTERM 0x04 /* soft mnt, too many retries */
300 #define R_INTR 0x08 /* intr mnt, signal pending */
301 #define R_SOCKERR 0x10 /* Fatal error on socket */
302 #define R_TPRINTFMSG 0x20 /* Did a tprintf msg. */
303 #define R_MUSTRESEND 0x40 /* Must resend request */
304 #define R_GETONEREP 0x80 /* Probe for one reply only */
305
306 /*
307 * A list of nfssvc_sock structures is maintained with all the sockets
308 * that require service by the nfsd.
309 * The nfsuid structs hang off of the nfssvc_sock structs in both lru
310 * and uid hash lists.
311 */
312 #ifndef NFS_UIDHASHSIZ
313 #define NFS_UIDHASHSIZ 29 /* Tune the size of nfssvc_sock with this */
314 #endif
315 #define NUIDHASH(sock, uid) \
316 (&(sock)->ns_uidhashtbl[(uid) % NFS_UIDHASHSIZ])
317 #ifndef NFS_WDELAYHASHSIZ
318 #define NFS_WDELAYHASHSIZ 16 /* and with this */
319 #endif
320 #define NWDELAYHASH(sock, f) \
321 (&(sock)->ns_wdelayhashtbl[(*((u_int32_t *)(f))) % NFS_WDELAYHASHSIZ])
322 #ifndef NFS_MUIDHASHSIZ
323 #define NFS_MUIDHASHSIZ 63 /* Tune the size of nfsmount with this */
324 #endif
325 #define NMUIDHASH(nmp, uid) \
326 (&(nmp)->nm_uidhashtbl[(uid) % NFS_MUIDHASHSIZ])
327 #define NFSNOHASH(fhsum) \
328 (&nfsnodehashtbl[(fhsum) & nfsnodehash])
329
330 #ifndef NFS_DIRHASHSIZ
331 #define NFS_DIRHASHSIZ 64
332 #endif
333 #define NFSDIRHASH(np, off) \
334 (&np->n_dircache[(nfs_dirhash((off)) & nfsdirhashmask)])
335
336 /*
337 * Macros for storing/retrieving cookies into directory buffers.
338 */
339 #define NFS_STASHCOOKIE(dp,off) \
340 *((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))) = off
341 #define NFS_GETCOOKIE(dp) \
342 (*((off_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t))))
343 #define NFS_STASHCOOKIE32(dp, val) \
344 *((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
345 sizeof (int))) = val
346 #define NFS_GETCOOKIE32(dp) \
347 (*((u_int32_t *)((caddr_t)(dp) + (dp)->d_reclen - sizeof (off_t) - \
348 sizeof (int))))
349
350 /*
351 * Flags passed to nfs_bioread().
352 */
353 #define NFSBIO_CACHECOOKIES 0x0001 /* Cache dir offset cookies */
354
355 /*
356 * Network address hash list element
357 */
358 union nethostaddr {
359 u_int32_t had_inetaddr;
360 struct mbuf *had_nam;
361 };
362
363 struct nfsuid {
364 TAILQ_ENTRY(nfsuid) nu_lru; /* LRU chain */
365 LIST_ENTRY(nfsuid) nu_hash; /* Hash list */
366 int nu_flag; /* Flags */
367 union nethostaddr nu_haddr; /* Host addr. for dgram sockets */
368 struct ucred nu_cr; /* Cred uid mapped to */
369 int nu_expire; /* Expiry time (sec) */
370 struct timeval nu_timestamp; /* Kerb. timestamp */
371 u_int32_t nu_nickname; /* Nickname on server */
372 NFSKERBKEY_T nu_key; /* and session key */
373 };
374
375 #define nu_inetaddr nu_haddr.had_inetaddr
376 #define nu_nam nu_haddr.had_nam
377 /* Bits for nu_flag */
378 #define NU_INETADDR 0x1
379 #define NU_NAM 0x2
380 #define NU_NETFAM(u) (((u)->nu_flag & NU_INETADDR) ? AF_INET : AF_ISO)
381
382 struct nfssvc_sock {
383 TAILQ_ENTRY(nfssvc_sock) ns_chain; /* List of all nfssvc_sock's */
384 TAILQ_HEAD(, nfsuid) ns_uidlruhead;
385 struct file *ns_fp;
386 struct socket *ns_so;
387 struct mbuf *ns_nam;
388 struct mbuf *ns_raw;
389 struct mbuf *ns_rawend;
390 struct mbuf *ns_rec;
391 struct mbuf *ns_recend;
392 struct mbuf *ns_frag;
393 int ns_flag;
394 int ns_solock;
395 int ns_cc;
396 int ns_reclen;
397 int ns_numuids;
398 u_int32_t ns_sref;
399 LIST_HEAD(, nfsrv_descript) ns_tq; /* Write gather lists */
400 LIST_HEAD(, nfsuid) ns_uidhashtbl[NFS_UIDHASHSIZ];
401 LIST_HEAD(nfsrvw_delayhash, nfsrv_descript) ns_wdelayhashtbl[NFS_WDELAYHASHSIZ];
402 };
403
404 /* Bits for "ns_flag" */
405 #define SLP_VALID 0x01
406 #define SLP_DOREC 0x02
407 #define SLP_NEEDQ 0x04
408 #define SLP_DISCONN 0x08
409 #define SLP_GETSTREAM 0x10
410 #define SLP_LASTFRAG 0x20
411 #define SLP_ALLFLAGS 0xff
412
413 TAILQ_HEAD(, nfssvc_sock) nfssvc_sockhead;
414 int nfssvc_sockhead_flag;
415 #define SLP_INIT 0x01
416 #define SLP_WANTINIT 0x02
417
418 /*
419 * One of these structures is allocated for each nfsd.
420 */
421 struct nfsd {
422 TAILQ_ENTRY(nfsd) nfsd_chain; /* List of all nfsd's */
423 int nfsd_flag; /* NFSD_ flags */
424 struct nfssvc_sock *nfsd_slp; /* Current socket */
425 int nfsd_authlen; /* Authenticator len */
426 u_char nfsd_authstr[RPCAUTH_MAXSIZ]; /* Authenticator data */
427 int nfsd_verflen; /* and the Verifier */
428 u_char nfsd_verfstr[RPCVERF_MAXSIZ];
429 struct proc *nfsd_procp; /* Proc ptr */
430 struct nfsrv_descript *nfsd_nd; /* Associated nfsrv_descript */
431 };
432
433 /* Bits for "nfsd_flag" */
434 #define NFSD_WAITING 0x01
435 #define NFSD_REQINPROG 0x02
436 #define NFSD_NEEDAUTH 0x04
437 #define NFSD_AUTHFAIL 0x08
438
439 /*
440 * This structure is used by the server for describing each request.
441 * Some fields are used only when write request gathering is performed.
442 */
443 struct nfsrv_descript {
444 u_quad_t nd_time; /* Write deadline (usec) */
445 off_t nd_off; /* Start byte offset */
446 off_t nd_eoff; /* and end byte offset */
447 LIST_ENTRY(nfsrv_descript) nd_hash; /* Hash list */
448 LIST_ENTRY(nfsrv_descript) nd_tq; /* and timer list */
449 LIST_HEAD(,nfsrv_descript) nd_coalesce; /* coalesced writes */
450 struct mbuf *nd_mrep; /* Request mbuf list */
451 struct mbuf *nd_md; /* Current dissect mbuf */
452 struct mbuf *nd_mreq; /* Reply mbuf list */
453 struct mbuf *nd_nam; /* and socket addr */
454 struct mbuf *nd_nam2; /* return socket addr */
455 caddr_t nd_dpos; /* Current dissect pos */
456 u_int32_t nd_procnum; /* RPC # */
457 int nd_stable; /* storage type */
458 int nd_flag; /* nd_flag */
459 int nd_len; /* Length of this write */
460 int nd_repstat; /* Reply status */
461 u_int32_t nd_retxid; /* Reply xid */
462 u_int32_t nd_duration; /* Lease duration */
463 struct timeval nd_starttime; /* Time RPC initiated */
464 fhandle_t nd_fh; /* File handle */
465 struct ucred nd_cr; /* Credentials */
466 };
467
468 /* Bits for "nd_flag" */
469 #define ND_READ LEASE_READ
470 #define ND_WRITE LEASE_WRITE
471 #define ND_CHECK 0x04
472 #define ND_LEASE (ND_READ | ND_WRITE | ND_CHECK)
473 #define ND_NFSV3 0x08
474 #define ND_NQNFS 0x10
475 #define ND_KERBNICK 0x20
476 #define ND_KERBFULL 0x40
477 #define ND_KERBAUTH (ND_KERBNICK | ND_KERBFULL)
478
479 TAILQ_HEAD(, nfsd) nfsd_head;
480 int nfsd_head_flag;
481 #define NFSD_CHECKSLP 0x01
482
483 /*
484 * These macros compare nfsrv_descript structures.
485 */
486 #define NFSW_CONTIG(o, n) \
487 ((o)->nd_eoff >= (n)->nd_off && \
488 !memcmp((caddr_t)&(o)->nd_fh, (caddr_t)&(n)->nd_fh, NFSX_V3FH))
489
490 #define NFSW_SAMECRED(o, n) \
491 (((o)->nd_flag & ND_KERBAUTH) == ((n)->nd_flag & ND_KERBAUTH) && \
492 !memcmp((caddr_t)&(o)->nd_cr, (caddr_t)&(n)->nd_cr, \
493 sizeof (struct ucred)))
494
495 /*
496 * Defines for WebNFS
497 */
498
499 #define WEBNFS_ESC_CHAR '%'
500 #define WEBNFS_SPECCHAR_START 0x80
501
502 #define WEBNFS_NATIVE_CHAR 0x80
503 /*
504 * ..
505 * Possibly more here in the future.
506 */
507
508 /*
509 * Macro for converting escape characters in WebNFS pathnames.
510 * Should really be in libkern.
511 */
512
513 #define HEXTOC(c) \
514 ((c) >= 'a' ? ((c) - ('a' - 10)) : \
515 ((c) >= 'A' ? ((c) - ('A' - 10)) : ((c) - '0')))
516 #define HEXSTRTOI(p) \
517 ((HEXTOC(p[0]) << 4) + HEXTOC(p[1]))
518 #endif /* _KERNEL */
519
520 #endif /* _NFS_NFS_H */
521