1 /* $OpenBSD: nfs_kq.c,v 1.37 2024/05/01 13:15:59 jsg Exp $ */
2 /* $NetBSD: nfs_kq.c,v 1.7 2003/10/30 01:43:10 simonb Exp $ */
3
4 /*-
5 * Copyright (c) 2002 The NetBSD Foundation, Inc.
6 * All rights reserved.
7 *
8 * This code is derived from software contributed to The NetBSD Foundation
9 * by Jaromir Dolecek.
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 *
20 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
22 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
23 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
30 * POSSIBILITY OF SUCH DAMAGE.
31 */
32
33 #include <sys/param.h>
34 #include <sys/systm.h>
35 #include <sys/proc.h>
36 #include <sys/mount.h>
37 #include <sys/malloc.h>
38 #include <sys/vnode.h>
39 #include <sys/file.h>
40 #include <sys/kthread.h>
41 #include <sys/rwlock.h>
42 #include <sys/queue.h>
43
44 #include <nfs/nfsproto.h>
45 #include <nfs/nfs.h>
46 #include <nfs/nfsnode.h>
47 #include <nfs/nfs_var.h>
48
49 void nfs_kqpoll(void *);
50 int nfs_kqwatch(struct vnode *);
51 void nfs_kqunwatch(struct vnode *);
52
53 void filt_nfsdetach(struct knote *);
54 int filt_nfsread(struct knote *, long);
55 int filt_nfswrite(struct knote *, long);
56 int filt_nfsvnode(struct knote *, long);
57
58 struct kevq {
59 SLIST_ENTRY(kevq) kev_link;
60 struct vnode *vp;
61 u_int usecount;
62 u_int flags;
63 #define KEVQ_BUSY 0x01 /* currently being processed */
64 #define KEVQ_WANT 0x02 /* want to change this entry */
65 struct timespec omtime; /* old modification time */
66 struct timespec octime; /* old change time */
67 nlink_t onlink; /* old number of references to file */
68 };
69 SLIST_HEAD(kevqlist, kevq);
70
71 struct rwlock nfskevq_lock = RWLOCK_INITIALIZER("nfskqlk");
72 struct proc *pnfskq;
73 struct kevqlist kevlist = SLIST_HEAD_INITIALIZER(kevlist);
74
75 /*
76 * This quite simplistic routine periodically checks for server changes
77 * of any of the watched files every NFS_MINATTRTIMO/2 seconds.
78 * Only changes in size, modification time, change time and nlinks
79 * are being checked, everything else is ignored.
80 * The routine only calls VOP_GETATTR() when it's likely it would get
81 * some new data, i.e. when the vnode expires from attrcache. This
82 * should give same result as periodically running stat(2) from userland,
83 * while keeping CPU/network usage low, and still provide proper kevent
84 * semantics.
85 * The poller thread is created when first vnode is added to watch list,
86 * and exits when the watch list is empty. The overhead of thread creation
87 * isn't really important, neither speed of attach and detach of knote.
88 */
89 void
nfs_kqpoll(void * arg)90 nfs_kqpoll(void *arg)
91 {
92 struct kevq *ke;
93 struct vattr attr;
94 struct proc *p = pnfskq;
95 u_quad_t osize;
96 int error;
97
98 for(;;) {
99 rw_enter_write(&nfskevq_lock);
100 SLIST_FOREACH(ke, &kevlist, kev_link) {
101 struct nfsnode *np = VTONFS(ke->vp);
102
103 #ifdef DEBUG
104 printf("nfs_kqpoll on: ");
105 VOP_PRINT(ke->vp);
106 #endif
107 /* skip if still in attrcache */
108 if (nfs_getattrcache(ke->vp, &attr) != ENOENT)
109 continue;
110
111 /*
112 * Mark entry busy, release lock and check
113 * for changes.
114 */
115 ke->flags |= KEVQ_BUSY;
116 rw_exit_write(&nfskevq_lock);
117
118 /* save v_size, nfs_getattr() updates it */
119 osize = np->n_size;
120
121 error = VOP_GETATTR(ke->vp, &attr, p->p_ucred, p);
122 if (error == ESTALE) {
123 NFS_INVALIDATE_ATTRCACHE(np);
124 VN_KNOTE(ke->vp, NOTE_DELETE);
125 goto next;
126 }
127
128 /* following is a bit fragile, but about best
129 * we can get */
130 if (attr.va_size != osize) {
131 int flags = NOTE_WRITE;
132
133 if (attr.va_size > osize)
134 flags |= NOTE_EXTEND;
135 else
136 flags |= NOTE_TRUNCATE;
137
138 VN_KNOTE(ke->vp, flags);
139 ke->omtime = attr.va_mtime;
140 } else if (attr.va_mtime.tv_sec != ke->omtime.tv_sec
141 || attr.va_mtime.tv_nsec != ke->omtime.tv_nsec) {
142 VN_KNOTE(ke->vp, NOTE_WRITE);
143 ke->omtime = attr.va_mtime;
144 }
145
146 if (attr.va_ctime.tv_sec != ke->octime.tv_sec
147 || attr.va_ctime.tv_nsec != ke->octime.tv_nsec) {
148 VN_KNOTE(ke->vp, NOTE_ATTRIB);
149 ke->octime = attr.va_ctime;
150 }
151
152 if (attr.va_nlink != ke->onlink) {
153 VN_KNOTE(ke->vp, NOTE_LINK);
154 ke->onlink = attr.va_nlink;
155 }
156
157 next:
158 rw_enter_write(&nfskevq_lock);
159 ke->flags &= ~KEVQ_BUSY;
160 if (ke->flags & KEVQ_WANT) {
161 ke->flags &= ~KEVQ_WANT;
162 wakeup(ke);
163 }
164 }
165
166 if (SLIST_EMPTY(&kevlist)) {
167 /* Nothing more to watch, exit */
168 pnfskq = NULL;
169 rw_exit_write(&nfskevq_lock);
170 kthread_exit(0);
171 }
172 rw_exit_write(&nfskevq_lock);
173
174 /* wait a while before checking for changes again */
175 tsleep_nsec(pnfskq, PSOCK, "nfskqpw",
176 SEC_TO_NSEC(NFS_MINATTRTIMO) / 2);
177 }
178 }
179
180 void
filt_nfsdetach(struct knote * kn)181 filt_nfsdetach(struct knote *kn)
182 {
183 struct vnode *vp = (struct vnode *)kn->kn_hook;
184
185 klist_remove_locked(&vp->v_klist, kn);
186
187 /* Remove the vnode from watch list */
188 if ((kn->kn_flags & (__EV_POLL | __EV_SELECT)) == 0)
189 nfs_kqunwatch(vp);
190 }
191
192 void
nfs_kqunwatch(struct vnode * vp)193 nfs_kqunwatch(struct vnode *vp)
194 {
195 struct kevq *ke;
196
197 rw_enter_write(&nfskevq_lock);
198 SLIST_FOREACH(ke, &kevlist, kev_link) {
199 if (ke->vp == vp) {
200 while (ke->flags & KEVQ_BUSY) {
201 ke->flags |= KEVQ_WANT;
202 rw_exit_write(&nfskevq_lock);
203 tsleep_nsec(ke, PSOCK, "nfskqdet", INFSLP);
204 rw_enter_write(&nfskevq_lock);
205 }
206
207 if (ke->usecount > 1) {
208 /* keep, other kevents need this */
209 ke->usecount--;
210 } else {
211 /* last user, g/c */
212 SLIST_REMOVE(&kevlist, ke, kevq, kev_link);
213 free(ke, M_KEVENT, sizeof(*ke));
214 }
215 break;
216 }
217 }
218 rw_exit_write(&nfskevq_lock);
219 }
220
221 int
filt_nfsread(struct knote * kn,long hint)222 filt_nfsread(struct knote *kn, long hint)
223 {
224 struct vnode *vp = (struct vnode *)kn->kn_hook;
225 struct nfsnode *np = VTONFS(vp);
226
227 /*
228 * filesystem is gone, so set the EOF flag and schedule
229 * the knote for deletion.
230 */
231 if (hint == NOTE_REVOKE) {
232 kn->kn_flags |= (EV_EOF | EV_ONESHOT);
233 return (1);
234 }
235
236 kn->kn_data = np->n_size - foffset(kn->kn_fp);
237 #ifdef DEBUG
238 printf("nfsread event. %lld\n", kn->kn_data);
239 #endif
240 if (kn->kn_data == 0 && kn->kn_sfflags & NOTE_EOF) {
241 kn->kn_fflags |= NOTE_EOF;
242 return (1);
243 }
244
245 if (kn->kn_flags & (__EV_POLL | __EV_SELECT))
246 return (1);
247
248 return (kn->kn_data != 0);
249 }
250
251 int
filt_nfswrite(struct knote * kn,long hint)252 filt_nfswrite(struct knote *kn, long hint)
253 {
254 /*
255 * filesystem is gone, so set the EOF flag and schedule
256 * the knote for deletion.
257 */
258 if (hint == NOTE_REVOKE) {
259 kn->kn_flags |= (EV_EOF | EV_ONESHOT);
260 return (1);
261 }
262
263 kn->kn_data = 0;
264 return (1);
265 }
266
267 int
filt_nfsvnode(struct knote * kn,long hint)268 filt_nfsvnode(struct knote *kn, long hint)
269 {
270 if (kn->kn_sfflags & hint)
271 kn->kn_fflags |= hint;
272 if (hint == NOTE_REVOKE) {
273 kn->kn_flags |= EV_EOF;
274 return (1);
275 }
276 return (kn->kn_fflags != 0);
277 }
278
279 static const struct filterops nfsread_filtops = {
280 .f_flags = FILTEROP_ISFD,
281 .f_attach = NULL,
282 .f_detach = filt_nfsdetach,
283 .f_event = filt_nfsread,
284 };
285
286 static const struct filterops nfswrite_filtops = {
287 .f_flags = FILTEROP_ISFD,
288 .f_attach = NULL,
289 .f_detach = filt_nfsdetach,
290 .f_event = filt_nfswrite,
291 };
292
293 static const struct filterops nfsvnode_filtops = {
294 .f_flags = FILTEROP_ISFD,
295 .f_attach = NULL,
296 .f_detach = filt_nfsdetach,
297 .f_event = filt_nfsvnode,
298 };
299
300 int
nfs_kqfilter(void * v)301 nfs_kqfilter(void *v)
302 {
303 struct vop_kqfilter_args *ap = v;
304 struct vnode *vp;
305 struct knote *kn;
306
307 vp = ap->a_vp;
308 kn = ap->a_kn;
309
310 #ifdef DEBUG
311 printf("nfs_kqfilter(%d) on: ", kn->kn_filter);
312 VOP_PRINT(vp);
313 #endif
314
315 switch (kn->kn_filter) {
316 case EVFILT_READ:
317 kn->kn_fop = &nfsread_filtops;
318 break;
319 case EVFILT_WRITE:
320 kn->kn_fop = &nfswrite_filtops;
321 break;
322 case EVFILT_VNODE:
323 kn->kn_fop = &nfsvnode_filtops;
324 break;
325 default:
326 return (EINVAL);
327 }
328
329 kn->kn_hook = vp;
330
331 /*
332 * Put the vnode to watched list.
333 */
334 if ((kn->kn_flags & (__EV_POLL | __EV_SELECT)) == 0) {
335 int error;
336
337 error = nfs_kqwatch(vp);
338 if (error)
339 return (error);
340 }
341
342 klist_insert_locked(&vp->v_klist, kn);
343
344 return (0);
345 }
346
347 int
nfs_kqwatch(struct vnode * vp)348 nfs_kqwatch(struct vnode *vp)
349 {
350 struct proc *p = curproc; /* XXX */
351 struct vattr attr;
352 struct kevq *ke;
353 int error = 0;
354
355 /*
356 * Fetch current attributes. It's only needed when the vnode
357 * is not watched yet, but we need to do this without lock
358 * held. This is likely cheap due to attrcache, so do it now.
359 */
360 memset(&attr, 0, sizeof(attr));
361 (void) VOP_GETATTR(vp, &attr, p->p_ucred, p);
362
363 rw_enter_write(&nfskevq_lock);
364
365 /* ensure the poller is running */
366 if (!pnfskq) {
367 error = kthread_create(nfs_kqpoll, NULL, &pnfskq,
368 "nfskqpoll");
369 if (error)
370 goto out;
371 }
372
373 SLIST_FOREACH(ke, &kevlist, kev_link)
374 if (ke->vp == vp)
375 break;
376
377 if (ke) {
378 /* already watched, so just bump usecount */
379 ke->usecount++;
380 } else {
381 /* need a new one */
382 ke = malloc(sizeof(*ke), M_KEVENT, M_WAITOK);
383 ke->vp = vp;
384 ke->usecount = 1;
385 ke->flags = 0;
386 ke->omtime = attr.va_mtime;
387 ke->octime = attr.va_ctime;
388 ke->onlink = attr.va_nlink;
389 SLIST_INSERT_HEAD(&kevlist, ke, kev_link);
390 }
391
392 /* kick the poller */
393 wakeup(pnfskq);
394
395 out:
396 rw_exit_write(&nfskevq_lock);
397 return (error);
398 }
399