npf.c revision 1.12 1 /* $NetBSD: npf.c,v 1.12 2012/07/15 00:23:00 rmind Exp $ */
2
3 /*-
4 * Copyright (c) 2009-2010 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This material is based upon work partially supported by The
8 * NetBSD Foundation under a contract with Mindaugas Rasiukevicius.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
32 /*
33 * NPF main: dynamic load/initialisation and unload routines.
34 */
35
36 #include <sys/cdefs.h>
37 __KERNEL_RCSID(0, "$NetBSD: npf.c,v 1.12 2012/07/15 00:23:00 rmind Exp $");
38
39 #include <sys/param.h>
40 #include <sys/types.h>
41
42 #include <sys/atomic.h>
43 #include <sys/conf.h>
44 #include <sys/kauth.h>
45 #include <sys/kmem.h>
46 #include <sys/lwp.h>
47 #include <sys/module.h>
48 #include <sys/percpu.h>
49 #include <sys/rwlock.h>
50 #include <sys/socketvar.h>
51 #include <sys/sysctl.h>
52 #include <sys/uio.h>
53
54 #include "npf_impl.h"
55
56 /*
57 * Module and device structures.
58 */
59 MODULE(MODULE_CLASS_DRIVER, npf, NULL);
60
61 void npfattach(int);
62
63 static int npf_fini(void);
64 static int npf_dev_open(dev_t, int, int, lwp_t *);
65 static int npf_dev_close(dev_t, int, int, lwp_t *);
66 static int npf_dev_ioctl(dev_t, u_long, void *, int, lwp_t *);
67 static int npf_dev_poll(dev_t, int, lwp_t *);
68 static int npf_dev_read(dev_t, struct uio *, int);
69
70 typedef struct {
71 npf_ruleset_t * n_rules;
72 npf_tableset_t * n_tables;
73 npf_ruleset_t * n_nat_rules;
74 prop_dictionary_t n_dict;
75 bool n_default_pass;
76 } npf_core_t;
77
78 static void npf_core_destroy(npf_core_t *);
79 static int npfctl_stats(void *);
80
81 static krwlock_t npf_lock __cacheline_aligned;
82 static npf_core_t * npf_core __cacheline_aligned;
83 static percpu_t * npf_stats_percpu __read_mostly;
84 static struct sysctllog * npf_sysctl __read_mostly;
85
86 const struct cdevsw npf_cdevsw = {
87 npf_dev_open, npf_dev_close, npf_dev_read, nowrite, npf_dev_ioctl,
88 nostop, notty, npf_dev_poll, nommap, nokqfilter, D_OTHER | D_MPSAFE
89 };
90
91 static int
92 npf_init(void)
93 {
94 #ifdef _MODULE
95 devmajor_t bmajor = NODEVMAJOR, cmajor = NODEVMAJOR;
96 #endif
97 npf_ruleset_t *rset, *nset;
98 npf_tableset_t *tset;
99 prop_dictionary_t dict;
100 int error = 0;
101
102 rw_init(&npf_lock);
103 npf_stats_percpu = percpu_alloc(NPF_STATS_SIZE);
104 npf_sysctl = NULL;
105
106 npf_tableset_sysinit();
107 npf_session_sysinit();
108 npf_nat_sysinit();
109 npf_alg_sysinit();
110 npflogattach(1);
111
112 /* Load empty configuration. */
113 dict = prop_dictionary_create();
114 rset = npf_ruleset_create();
115 tset = npf_tableset_create();
116 nset = npf_ruleset_create();
117 npf_reload(dict, rset, tset, nset, true);
118 KASSERT(npf_core != NULL);
119
120 #ifdef _MODULE
121 /* Attach /dev/npf device. */
122 error = devsw_attach("npf", NULL, &bmajor, &npf_cdevsw, &cmajor);
123 if (error) {
124 /* It will call devsw_detach(), which is safe. */
125 (void)npf_fini();
126 }
127 #endif
128 return error;
129 }
130
131 static int
132 npf_fini(void)
133 {
134
135 /* At first, detach device and remove pfil hooks. */
136 #ifdef _MODULE
137 devsw_detach(NULL, &npf_cdevsw);
138 #endif
139 npflogdetach();
140 npf_pfil_unregister();
141
142 /* Flush all sessions, destroy configuration (ruleset, etc). */
143 npf_session_tracking(false);
144 npf_core_destroy(npf_core);
145
146 /* Finally, safe to destroy the subsystems. */
147 npf_alg_sysfini();
148 npf_nat_sysfini();
149 npf_session_sysfini();
150 npf_tableset_sysfini();
151
152 if (npf_sysctl) {
153 sysctl_teardown(&npf_sysctl);
154 }
155 percpu_free(npf_stats_percpu, NPF_STATS_SIZE);
156 rw_destroy(&npf_lock);
157
158 return 0;
159 }
160
161 /*
162 * Module interface.
163 */
164 static int
165 npf_modcmd(modcmd_t cmd, void *arg)
166 {
167
168 switch (cmd) {
169 case MODULE_CMD_INIT:
170 return npf_init();
171 case MODULE_CMD_FINI:
172 return npf_fini();
173 case MODULE_CMD_AUTOUNLOAD:
174 if (npf_pfil_registered_p() || !npf_default_pass()) {
175 return EBUSY;
176 }
177 break;
178 default:
179 return ENOTTY;
180 }
181 return 0;
182 }
183
184 void
185 npfattach(int nunits)
186 {
187
188 /* Void. */
189 }
190
191 static int
192 npf_dev_open(dev_t dev, int flag, int mode, lwp_t *l)
193 {
194
195 /* Available only for super-user. */
196 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FIREWALL,
197 KAUTH_REQ_NETWORK_FIREWALL_FW, NULL, NULL, NULL)) {
198 return EPERM;
199 }
200 return 0;
201 }
202
203 static int
204 npf_dev_close(dev_t dev, int flag, int mode, lwp_t *l)
205 {
206
207 return 0;
208 }
209
210 static int
211 npf_dev_ioctl(dev_t dev, u_long cmd, void *data, int flag, lwp_t *l)
212 {
213 int error;
214
215 /* Available only for super-user. */
216 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FIREWALL,
217 KAUTH_REQ_NETWORK_FIREWALL_FW, NULL, NULL, NULL)) {
218 return EPERM;
219 }
220
221 switch (cmd) {
222 case IOC_NPF_VERSION:
223 *(int *)data = NPF_VERSION;
224 error = 0;
225 break;
226 case IOC_NPF_SWITCH:
227 error = npfctl_switch(data);
228 break;
229 case IOC_NPF_RELOAD:
230 error = npfctl_reload(cmd, data);
231 break;
232 case IOC_NPF_GETCONF:
233 error = npfctl_getconf(cmd, data);
234 break;
235 case IOC_NPF_TABLE:
236 error = npfctl_table(data);
237 break;
238 case IOC_NPF_STATS:
239 error = npfctl_stats(data);
240 break;
241 case IOC_NPF_SESSIONS_SAVE:
242 error = npfctl_sessions_save(cmd, data);
243 break;
244 case IOC_NPF_SESSIONS_LOAD:
245 error = npfctl_sessions_load(cmd, data);
246 break;
247 case IOC_NPF_UPDATE_RULE:
248 error = npfctl_update_rule(cmd, data);
249 break;
250 default:
251 error = ENOTTY;
252 break;
253 }
254 return error;
255 }
256
257 static int
258 npf_dev_poll(dev_t dev, int events, lwp_t *l)
259 {
260
261 return ENOTSUP;
262 }
263
264 static int
265 npf_dev_read(dev_t dev, struct uio *uio, int flag)
266 {
267
268 return ENOTSUP;
269 }
270
271 /*
272 * NPF core loading/reloading/unloading mechanism.
273 */
274
275 static void
276 npf_core_destroy(npf_core_t *nc)
277 {
278
279 prop_object_release(nc->n_dict);
280 npf_ruleset_destroy(nc->n_rules);
281 npf_ruleset_destroy(nc->n_nat_rules);
282 npf_tableset_destroy(nc->n_tables);
283 kmem_free(nc, sizeof(npf_core_t));
284 }
285
286 /*
287 * npf_reload: atomically load new ruleset, tableset and NAT policies.
288 * Then destroy old (unloaded) structures.
289 */
290 void
291 npf_reload(prop_dictionary_t dict, npf_ruleset_t *rset,
292 npf_tableset_t *tset, npf_ruleset_t *nset, bool flush)
293 {
294 npf_core_t *nc, *onc;
295
296 /* Setup a new core structure. */
297 nc = kmem_zalloc(sizeof(npf_core_t), KM_SLEEP);
298 nc->n_rules = rset;
299 nc->n_tables = tset;
300 nc->n_nat_rules = nset;
301 nc->n_dict = dict;
302 nc->n_default_pass = flush;
303
304 /* Lock and load the core structure. */
305 rw_enter(&npf_lock, RW_WRITER);
306 onc = atomic_swap_ptr(&npf_core, nc);
307 if (onc) {
308 /* Reload only necessary NAT policies. */
309 npf_ruleset_natreload(nset, onc->n_nat_rules);
310 }
311 /* Unlock. Everything goes "live" now. */
312 rw_exit(&npf_lock);
313
314 if (onc) {
315 /* Destroy unloaded structures. */
316 npf_core_destroy(onc);
317 }
318 }
319
320 void
321 npf_core_enter(void)
322 {
323 rw_enter(&npf_lock, RW_READER);
324 }
325
326 npf_ruleset_t *
327 npf_core_ruleset(void)
328 {
329 KASSERT(rw_lock_held(&npf_lock));
330 return npf_core->n_rules;
331 }
332
333 npf_ruleset_t *
334 npf_core_natset(void)
335 {
336 KASSERT(rw_lock_held(&npf_lock));
337 return npf_core->n_nat_rules;
338 }
339
340 npf_tableset_t *
341 npf_core_tableset(void)
342 {
343 KASSERT(rw_lock_held(&npf_lock));
344 return npf_core->n_tables;
345 }
346
347 void
348 npf_core_exit(void)
349 {
350 rw_exit(&npf_lock);
351 }
352
353 bool
354 npf_core_locked(void)
355 {
356 return rw_lock_held(&npf_lock);
357 }
358
359 prop_dictionary_t
360 npf_core_dict(void)
361 {
362 KASSERT(rw_lock_held(&npf_lock));
363 return npf_core->n_dict;
364 }
365
366 bool
367 npf_default_pass(void)
368 {
369 KASSERT(rw_lock_held(&npf_lock));
370 return npf_core->n_default_pass;
371 }
372
373 /*
374 * NPF statistics interface.
375 */
376
377 void
378 npf_stats_inc(npf_stats_t st)
379 {
380 uint64_t *stats = percpu_getref(npf_stats_percpu);
381 stats[st]++;
382 percpu_putref(npf_stats_percpu);
383 }
384
385 void
386 npf_stats_dec(npf_stats_t st)
387 {
388 uint64_t *stats = percpu_getref(npf_stats_percpu);
389 stats[st]--;
390 percpu_putref(npf_stats_percpu);
391 }
392
393 static void
394 npf_stats_collect(void *mem, void *arg, struct cpu_info *ci)
395 {
396 uint64_t *percpu_stats = mem, *full_stats = arg;
397 int i;
398
399 for (i = 0; i < NPF_STATS_COUNT; i++) {
400 full_stats[i] += percpu_stats[i];
401 }
402 }
403
404 /*
405 * npfctl_stats: export collected statistics.
406 */
407 static int
408 npfctl_stats(void *data)
409 {
410 uint64_t *fullst, *uptr = *(uint64_t **)data;
411 int error;
412
413 fullst = kmem_zalloc(NPF_STATS_SIZE, KM_SLEEP);
414 percpu_foreach(npf_stats_percpu, npf_stats_collect, fullst);
415 error = copyout(fullst, uptr, NPF_STATS_SIZE);
416 kmem_free(fullst, NPF_STATS_SIZE);
417 return error;
418 }
419