nfs.h revision 1.53 
1/*	$OpenBSD: nfs.h,v 1.53 2017/02/22 11:42:46 mpi Exp $	*/
2/*	$NetBSD: nfs.h,v 1.10.4.1 1996/05/27 11:23:56 fvdl Exp $	*/
3
4/*
5 * Copyright (c) 1989, 1993, 1995
6 *	The Regents of the University of California.  All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Rick Macklem at The University of Guelph.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 *    notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 *    notice, this list of conditions and the following disclaimer in the
18 *    documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the University nor the names of its contributors
20 *    may be used to endorse or promote products derived from this software
21 *    without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 *	@(#)nfs.h	8.4 (Berkeley) 5/1/95
36 */
37
38#ifndef _NFS_NFS_H_
39#define _NFS_NFS_H_
40
41#define NFS_TICKINTVL	5		/* Desired time for a tick (msec) */
42#define NFS_HZ		(hz / nfs_ticks) /* Ticks/sec */
43#define	NFS_TIMEO	(1 * NFS_HZ)	/* Default timeout = 1 second */
44#define	NFS_MINTIMEO	(1 * NFS_HZ)	/* Min timeout to use */
45#define	NFS_MAXTIMEO	(60 * NFS_HZ)	/* Max timeout to backoff to */
46#define	NFS_MINIDEMTIMEO (5 * NFS_HZ)	/* Min timeout for non-idempotent ops*/
47#define	NFS_TIMEOUTMUL	2		/* Timeout/Delay multiplier */
48#define	NFS_MAXREXMIT	100		/* Stop counting after this many */
49#define	NFS_RETRANS	10		/* Num of retrans for soft mounts */
50#define	NFS_MAXGRPS	16		/* Max. size of groups list */
51#define	NFS_MINATTRTIMO 5		/* Attribute cache timeout in sec */
52#define	NFS_MAXATTRTIMO 60
53#define	NFS_WSIZE	8192		/* Def. write data size <= 8192 */
54#define	NFS_RSIZE	8192		/* Def. read data size <= 8192 */
55#define NFS_READDIRSIZE	8192		/* Def. readdir size */
56#define	NFS_DEFRAHEAD	1		/* Def. read ahead # blocks */
57#define	NFS_MAXRAHEAD	4		/* Max. read ahead # blocks */
58#define	NFS_MAXASYNCDAEMON 	20	/* Max. number async_daemons runable */
59
60/*
61 * Ideally, NFS_DIRBLKSIZ should be bigger, but I've seen servers with
62 * broken NFS/ethernet drivers that won't work with anything bigger (Linux..)
63 */
64#define	NFS_DIRBLKSIZ	1024		/* Must be a multiple of DIRBLKSIZ */
65#define NFS_READDIRBLKSIZ	512	/* Size of read dir blocks. XXX */
66
67/*
68 * Oddballs
69 */
70#define NFS_CMPFH(n, f, s) \
71	((n)->n_fhsize == (s) && !bcmp((caddr_t)(n)->n_fhp, (caddr_t)(f), (s)))
72#define NFS_ISV3(v)	(VFSTONFS((v)->v_mount)->nm_flag & NFSMNT_NFSV3)
73#define NFS_SRVMAXDATA(n) \
74		(((n)->nd_flag & ND_NFSV3) ? (((n)->nd_nam2) ? \
75		 NFS_MAXDGRAMDATA : NFS_MAXDATA) : NFS_V2MAXDATA)
76
77/*
78 * sys/malloc.h needs M_NFSDIROFF, M_NFSRVDESC and M_NFSBIGFH added.
79 */
80#ifndef M_NFSRVDESC
81#define M_NFSRVDESC	M_TEMP
82#endif
83#ifndef M_NFSDIROFF
84#define M_NFSDIROFF	M_TEMP
85#endif
86#ifndef M_NFSBIGFH
87#define M_NFSBIGFH	M_TEMP
88#endif
89
90/*
91 * The B_INVAFTERWRITE flag should be set to whatever is required by the
92 * buffer cache code to say "Invalidate the block after it is written back".
93 */
94#define	B_INVAFTERWRITE	B_INVAL
95
96/*
97 * Structures for the nfssvc(2) syscall.
98 * Not that anyone besides nfsd(8) should ever use it.
99 */
100struct nfsd_args {
101	int	sock;		/* Socket to serve */
102	caddr_t	name;		/* Client addr for connection based sockets */
103	int	namelen;	/* Length of name */
104};
105
106struct nfsd_srvargs {
107	struct nfsd	*nsd_nfsd;	/* Pointer to in kernel nfsd struct */
108	uid_t		nsd_uid;	/* Effective uid mapped to cred */
109	u_int32_t	nsd_haddr;	/* IP address of client */
110	struct xucred	nsd_cr;		/* Cred. uid maps to */
111	int		nsd_authlen;	/* Length of auth string (ret) */
112	u_char		*nsd_authstr;	/* Auth string (ret) */
113	int		nsd_verflen;	/* and the verifier */
114	u_char		*nsd_verfstr;
115	struct timeval	nsd_timestamp;	/* timestamp from verifier */
116	u_int32_t	nsd_ttl;	/* credential ttl (sec) */
117};
118
119/*
120 * Stats structure
121 */
122struct nfsstats {
123	uint64_t	attrcache_hits;
124	uint64_t	attrcache_misses;
125	uint64_t	lookupcache_hits;
126	uint64_t	lookupcache_misses;
127	uint64_t	direofcache_hits;
128	uint64_t	direofcache_misses;
129	uint64_t	biocache_reads;
130	uint64_t	read_bios;
131	uint64_t	read_physios;
132	uint64_t	biocache_writes;
133	uint64_t	write_bios;
134	uint64_t	write_physios;
135	uint64_t	biocache_readlinks;
136	uint64_t	readlink_bios;
137	uint64_t	biocache_readdirs;
138	uint64_t	readdir_bios;
139	uint64_t	rpccnt[NFS_NPROCS];
140	uint64_t	rpcretries;
141	uint64_t	srvrpccnt[NFS_NPROCS];
142	uint64_t	srvrpc_errs;
143	uint64_t	srv_errs;
144	uint64_t	rpcrequests;
145	uint64_t	rpctimeouts;
146	uint64_t	rpcunexpected;
147	uint64_t	rpcinvalid;
148	uint64_t	srvcache_inproghits;
149	uint64_t	srvcache_idemdonehits;
150	uint64_t	srvcache_nonidemdonehits;
151	uint64_t	srvcache_misses;
152	uint64_t	forcedsync;
153	uint64_t	srvnqnfs_leases;
154	uint64_t	srvnqnfs_maxleases;
155	uint64_t	srvnqnfs_getleases;
156	uint64_t	srvvop_writes;
157};
158
159/*
160 * Flags for nfssvc() system call.
161 */
162#define	NFSSVC_BIOD	0x002
163#define	NFSSVC_NFSD	0x004
164#define	NFSSVC_ADDSOCK	0x008
165#define	NFSSVC_AUTHIN	0x010
166#define	NFSSVC_GOTAUTH	0x040
167#define	NFSSVC_AUTHINFAIL 0x080
168#define	NFSSVC_MNTD	0x100
169
170/*
171 * fs.nfs sysctl(3) identifiers
172 */
173#define	NFS_NFSSTATS	1	/* struct: struct nfsstats */
174#define	NFS_NIOTHREADS	2	/* number of i/o threads */
175#define	NFS_MAXID	3
176
177#define FS_NFS_NAMES { \
178			{ 0, 0 }, \
179			{ "nfsstats", CTLTYPE_STRUCT }, \
180			{ "iothreads", CTLTYPE_INT } \
181}
182
183/*
184 * The set of signals the interrupt an I/O in progress for NFSMNT_INT mounts.
185 * What should be in this set is open to debate, but I believe that since
186 * I/O system calls on ufs are never interrupted by signals the set should
187 * be minimal. My reasoning is that many current programs that use signals
188 * such as SIGALRM will not expect file I/O system calls to be interrupted
189 * by them and break.
190 */
191#ifdef _KERNEL
192extern int nfs_niothreads;
193
194struct uio; struct buf; struct vattr; struct nameidata;	/* XXX */
195
196#define	NFSINT_SIGMASK	(sigmask(SIGINT)|sigmask(SIGTERM)|sigmask(SIGKILL)| \
197			 sigmask(SIGHUP)|sigmask(SIGQUIT))
198
199/*
200 * Socket errors ignored for connectionless sockets??
201 * For now, ignore them all
202 */
203#define	NFSIGNORE_SOERROR(s, e) \
204		((e) != EINTR && (e) != ERESTART && (e) != EWOULDBLOCK && \
205		((s) & PR_CONNREQUIRED) == 0)
206
207/*
208 * Nfs outstanding request list element
209 */
210struct nfsreq {
211	TAILQ_ENTRY(nfsreq) r_chain;
212	struct mbuf	*r_mreq;
213	struct mbuf	*r_mrep;
214	struct mbuf	*r_md;
215	caddr_t		r_dpos;
216	struct nfsmount *r_nmp;
217	struct vnode	*r_vp;
218	u_int32_t	r_xid;
219	int		r_flags;	/* flags on request, see below */
220	int		r_rexmit;	/* current retrans count */
221	int		r_timer;	/* tick counter on reply */
222	int		r_procnum;	/* NFS procedure number */
223	int		r_rtt;		/* RTT for rpc */
224	struct proc	*r_procp;	/* Proc that did I/O system call */
225};
226
227/* Flag values for r_flags */
228#define R_TIMING	0x01		/* timing request (in mntp) */
229#define R_SENT		0x02		/* request has been sent */
230#define	R_SOFTTERM	0x04		/* soft mnt, too many retries */
231#define	R_INTR		0x08		/* intr mnt, signal pending */
232#define	R_SOCKERR	0x10		/* Fatal error on socket */
233#define	R_TPRINTFMSG	0x20		/* Did a tprintf msg. */
234#define	R_MUSTRESEND	0x40		/* Must resend request */
235
236/*
237 * On fast networks, the estimator will try to reduce the
238 * timeout lower than the latency of the server's disks,
239 * which results in too many timeouts, so cap the lower
240 * bound.
241 */
242#define NFS_MINRTO	(NFS_HZ >> 2)
243
244/*
245 * Keep the RTO from increasing to unreasonably large values
246 * when a server is not responding.
247 */
248#define NFS_MAXRTO	(20 * NFS_HZ)
249
250enum nfs_rto_timers {
251	NFS_DEFAULT_TIMER,
252	NFS_GETATTR_TIMER,
253	NFS_LOOKUP_TIMER,
254	NFS_READ_TIMER,
255	NFS_WRITE_TIMER,
256};
257#define NFS_MAX_TIMER	(NFS_WRITE_TIMER)
258
259#define NFS_INITRTT	(NFS_HZ << 3)
260
261/*
262 * Network address hash list element
263 */
264union nethostaddr {
265	u_int32_t had_inetaddr;
266	struct mbuf *had_nam;
267};
268
269struct nfssvc_sock {
270	TAILQ_ENTRY(nfssvc_sock) ns_chain; /* List of all nfssvc_sock's */
271	struct file	*ns_fp;		/* fp from the... */
272	struct socket	*ns_so;		/* ...socket this struct wraps */
273	struct mbuf	*ns_nam;	/* MT_SONAME of client */
274	struct mbuf	*ns_raw;	/* head of unpeeked mbufs */
275	struct mbuf	*ns_rawend;	/* tail of unpeeked mbufs */
276	struct mbuf	*ns_rec;	/* queued RPC records */
277	struct mbuf	*ns_recend;	/* last queued RPC record */
278	struct mbuf	*ns_frag;	/* end of record fragment */
279	int		ns_flag;	/* socket status flags */
280	int		ns_solock;	/* lock for connected socket */
281	int		ns_cc;		/* actual chars queued */
282	int		ns_reclen;	/* length of first queued record */
283	u_int32_t	ns_sref;	/* # of refs to this struct */
284};
285
286/* Bits for "ns_flag" */
287#define	SLP_VALID	0x01	/* connection is usable */
288#define	SLP_DOREC	0x02	/* receive operation required */
289#define	SLP_NEEDQ	0x04	/* connection has data to queue from socket */
290#define	SLP_DISCONN	0x08	/* connection is closed */
291#define	SLP_GETSTREAM	0x10	/* extracting RPC from TCP connection */
292#define	SLP_LASTFRAG	0x20	/* last fragment received on TCP connection */
293#define	SLP_ALLFLAGS	0xff	/* convenience */
294
295/*
296 * One of these structures is allocated for each nfsd.
297 */
298struct nfsd {
299	TAILQ_ENTRY(nfsd) nfsd_chain;	/* List of all nfsd's */
300	int		nfsd_flag;	/* NFSD_ flags */
301	struct nfssvc_sock *nfsd_slp;	/* Current socket */
302	struct proc	*nfsd_procp;	/* Proc ptr */
303	struct nfsrv_descript *nfsd_nd;	/* Associated nfsrv_descript */
304};
305
306/* Bits for "nfsd_flag" */
307#define	NFSD_WAITING	0x01
308#define	NFSD_REQINPROG	0x02
309#define	NFSD_NEEDAUTH	0x04
310#define	NFSD_AUTHFAIL	0x08
311
312/*
313 * This structure is used by the server for describing each request.
314 */
315struct nfsrv_descript {
316	struct mbuf		*nd_mrep;	/* Request mbuf list */
317	struct mbuf		*nd_md;		/* Current dissect mbuf */
318	struct mbuf		*nd_nam;	/* and socket addr */
319	struct mbuf		*nd_nam2;	/* return socket addr */
320	caddr_t			nd_dpos;	/* Current dissect pos */
321	unsigned int		nd_procnum;	/* RPC # */
322	int			nd_flag;	/* nd_flag */
323	int			nd_repstat;	/* Reply status */
324	u_int32_t		nd_retxid;	/* Reply xid */
325	struct ucred		nd_cr;		/* Credentials */
326};
327
328/* Bits for "nd_flag" */
329#define ND_NFSV3	0x08
330
331extern struct pool nfsreqpl;
332extern struct pool nfs_node_pool;
333extern TAILQ_HEAD(nfsdhead, nfsd) nfsd_head;
334extern int nfsd_head_flag;
335#define	NFSD_CHECKSLP	0x01
336
337#endif	/* _KERNEL */
338#endif /* _NFS_NFS_H */
339