xref: /dragonfly/usr.bin/systat/pftop.c (revision cfd1aba3)
1 /*
2  * Copyright (c) 2013 The DragonFly Project.  All rights reserved.
3  *
4  * This code is derived from software contributed to The DragonFly Project
5  * by Matthew Dillon <dillon@backplane.com>
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in
15  *    the documentation and/or other materials provided with the
16  *    distribution.
17  * 3. Neither the name of The DragonFly Project nor the names of its
18  *    contributors may be used to endorse or promote products derived
19  *    from this software without specific, prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 #include <sys/param.h>
35 #include <sys/queue.h>
36 #include <sys/tree.h>
37 #include <sys/socket.h>
38 #include <sys/socketvar.h>
39 #include <sys/protosw.h>
40 #include <sys/sysctl.h>
41 #include <sys/endian.h>
42 
43 #include <netinet/in.h>
44 #include <arpa/inet.h>
45 #include <net/route.h>
46 #include <net/if.h>
47 #include <net/pf/pfvar.h>
48 #include <netinet/in_systm.h>
49 #include <netinet/ip.h>
50 #ifdef INET6
51 #include <netinet/ip6.h>
52 #endif
53 #include <netinet/in_pcb.h>
54 #include <netinet/ip_icmp.h>
55 #include <netinet/icmp_var.h>
56 #include <netinet/ip_var.h>
57 #include <netinet/tcp.h>
58 #include <netinet/tcpip.h>
59 #include <netinet/tcp_seq.h>
60 #include <netinet/tcp_fsm.h>
61 #include <netinet/tcp_timer.h>
62 #include <netinet/tcp_var.h>
63 #include <netinet/tcp_debug.h>
64 #include <netinet/udp.h>
65 #include <netinet/udp_var.h>
66 
67 #include <stdio.h>
68 #include <stdlib.h>
69 #include <string.h>
70 #include <unistd.h>
71 #include <fcntl.h>
72 #include <nlist.h>
73 #include <paths.h>
74 #include <err.h>
75 #include <errno.h>
76 #include <netdb.h>
77 
78 #include "systat.h"
79 #include "extern.h"
80 
81 struct mypfstate {
82 	RB_ENTRY(mypfstate)	rb_node;
83 	int			seq;
84 	struct pfsync_state	state;
85 	struct pfsync_state	last_state;
86 };
87 
88 static int
89 mypfstate_cmp(struct mypfstate *pf1, struct mypfstate *pf2)
90 {
91 	struct pfsync_state_key *nk1, *nk2;
92 	int r;
93 
94 	if (pf1->state.proto < pf2->state.proto)
95 		return(-1);
96 	if (pf1->state.proto > pf2->state.proto)
97 		return(1);
98 
99 	if (pf1->state.direction == PF_OUT) {
100 		nk1 = &pf1->state.key[PF_SK_WIRE];
101 	} else {
102 		nk1 = &pf1->state.key[PF_SK_STACK];
103 	}
104 	if (pf2->state.direction == PF_OUT) {
105 		nk2 = &pf2->state.key[PF_SK_WIRE];
106 	} else {
107 		nk2 = &pf2->state.key[PF_SK_STACK];
108 	}
109 	if (pf1->state.proto == IPPROTO_TCP || pf1->state.proto == IPPROTO_UDP) {
110 		if (ntohs(nk1->port[0]) >= 1024 &&
111 		    ntohs(nk2->port[0]) >= 1024) {
112 			if (ntohs(nk1->port[1]) < ntohs(nk2->port[1]))
113 				return(-1);
114 			if (ntohs(nk1->port[1]) > ntohs(nk2->port[1]))
115 				return(1);
116 		}
117 		if (ntohs(nk1->port[0]) < ntohs(nk2->port[0]))
118 			return(-1);
119 		if (ntohs(nk1->port[0]) > ntohs(nk2->port[0]))
120 			return(1);
121 		if (ntohs(nk1->port[1]) < ntohs(nk2->port[1]))
122 			return(-1);
123 		if (ntohs(nk1->port[1]) > ntohs(nk2->port[1]))
124 			return(1);
125 	}
126 
127 	/*
128 	 * Sort IPV4 vs IPV6 addresses
129 	 */
130 	if (pf1->state.af < pf2->state.af)
131 		return(-1);
132 	if (pf1->state.af > pf2->state.af)
133 		return(1);
134 
135 	/*
136 	 * Local and foreign addresses
137 	 */
138 	if (pf1->state.af == AF_INET) {
139 		if (ntohl(nk1->addr[0].v4.s_addr) <
140 		    ntohl(nk2->addr[0].v4.s_addr))
141 			return(-1);
142 		if (ntohl(nk1->addr[0].v4.s_addr) >
143 		    ntohl(nk2->addr[0].v4.s_addr))
144 			return(1);
145 		if (ntohl(nk1->addr[1].v4.s_addr) <
146 		    ntohl(nk2->addr[1].v4.s_addr))
147 			return(-1);
148 		if (ntohl(nk1->addr[1].v4.s_addr) >
149 		    ntohl(nk2->addr[1].v4.s_addr))
150 			return(1);
151 	} else if (pf1->state.af == AF_INET6) {
152 		r = bcmp(&nk1->addr[0].v6,
153 			 &nk2->addr[0].v6,
154 			 sizeof(nk1->addr[0].v6));
155 		if (r)
156 			return(r);
157 	} else {
158 		r = bcmp(&nk1->addr[0].v6,
159 			 &nk2->addr[0].v6,
160 			 sizeof(nk1->addr[0].v6));
161 		if (r)
162 			return(r);
163 	}
164 	return(0);
165 }
166 
167 struct mypfstate_tree;
168 RB_HEAD(mypfstate_tree, mypfstate);
169 RB_PROTOTYPE(mypfstate_tree, mypfstate, rb_node, mypfstate_cmp);
170 RB_GENERATE(mypfstate_tree, mypfstate, rb_node, mypfstate_cmp);
171 
172 static struct mypfstate_tree mypf_tree;
173 static struct timeval tv_curr;
174 static struct timeval tv_last;
175 static int tcp_pcb_seq;
176 
177 static const char *numtok(double value);
178 static const char *netaddrstr(sa_family_t af, struct pf_addr *addr,
179 			u_int16_t port);
180 static void updatestate(struct pfsync_state *state);
181 static int statebwcmp(const void *data1, const void *data2);
182 
183 #define DELTARATE(field)	\
184 	((double)(be64toh(*(uint64_t *)elm->state.field) - \
185 		  be64toh(*(uint64_t *)elm->last_state.field)) / delta_time)
186 
187 WINDOW *
188 openpftop(void)
189 {
190 	RB_INIT(&mypf_tree);
191 	return (subwin(stdscr, LINES-0-1, 0, 0, 0));
192 }
193 
194 void
195 closepftop(WINDOW *w)
196 {
197 	struct mypfstate *mypf;
198 
199 	while ((mypf = RB_ROOT(&mypf_tree)) != NULL) {
200 		RB_REMOVE(mypfstate_tree, &mypf_tree, mypf);
201 		free(mypf);
202 	}
203 
204         if (w != NULL) {
205 		wclear(w);
206 		wrefresh(w);
207 		delwin(w);
208 	}
209 }
210 
211 int
212 initpftop(void)
213 {
214 	return(1);
215 }
216 
217 void
218 fetchpftop(void)
219 {
220 	struct pfioc_states ps;
221 	struct pfsync_state *states;
222 	size_t nstates;
223 	size_t i;
224 	int fd;
225 
226 	fd = open("/dev/pf", O_RDONLY);
227 	if (fd < 0)
228 		return;
229 
230 	/*
231 	 * Extract PCB list
232 	 */
233 	bzero(&ps, sizeof(ps));
234 	if (ioctl(fd, DIOCGETSTATES, &ps) < 0) {
235 		close(fd);
236 		return;
237 	}
238 	ps.ps_len += 1024 * 1024;
239 	ps.ps_buf = malloc(ps.ps_len);
240 	if (ioctl(fd, DIOCGETSTATES, &ps) < 0) {
241 		free(ps.ps_buf);
242 		close(fd);
243 		return;
244 	}
245 
246 	states = (void *)ps.ps_buf;
247 	nstates = ps.ps_len / sizeof(*states);
248 
249 	++tcp_pcb_seq;
250 
251 	for (i = 0; i < nstates; ++i)
252 		updatestate(&states[i]);
253 	free(ps.ps_buf);
254 	close(fd);
255 	states = NULL;
256 	fd = -1;
257 
258 	tv_last = tv_curr;
259 	gettimeofday(&tv_curr, NULL);
260 }
261 
262 void
263 labelpftop(void)
264 {
265 	wmove(wnd, 0, 0);
266 	wclrtobot(wnd);
267 #if 0
268 	mvwaddstr(wnd, 0, LADDR, "Local Address");
269 	mvwaddstr(wnd, 0, FADDR, "Foreign Address");
270 	mvwaddstr(wnd, 0, PROTO, "Proto");
271 	mvwaddstr(wnd, 0, RCVCC, "Recv-Q");
272 	mvwaddstr(wnd, 0, SNDCC, "Send-Q");
273 	mvwaddstr(wnd, 0, STATE, "(state)");
274 #endif
275 }
276 
277 void
278 showpftop(void)
279 {
280 	double delta_time;
281 	struct mypfstate *elm;
282 	struct mypfstate *delm;
283 	struct mypfstate **array;
284 	size_t i;
285 	size_t n;
286 	struct pfsync_state_key *nk;
287 	int row;
288 
289 	delta_time = (double)(tv_curr.tv_sec - tv_last.tv_sec) - 1.0 +
290 		     (tv_curr.tv_usec + 1000000 - tv_last.tv_usec) / 1e6;
291 	if (delta_time < 0.1)
292 		return;
293 
294 	/*
295 	 * Delete and collect pass
296 	 */
297 	delm = NULL;
298 	i = 0;
299 	n = 1024;
300 	array = malloc(n * sizeof(*array));
301 	RB_FOREACH(elm, mypfstate_tree, &mypf_tree) {
302 		if (delm) {
303 			RB_REMOVE(mypfstate_tree, &mypf_tree, delm);
304 			free(delm);
305 			delm = NULL;
306 		}
307 		if (elm->seq == tcp_pcb_seq &&
308 		    (DELTARATE(bytes[0]) ||
309 		     DELTARATE(bytes[1]))
310 		) {
311 			array[i++] = elm;
312 			if (i == n) {
313 				n *= 2;
314 				array = realloc(array, n * sizeof(*array));
315 			}
316 		} else if (elm->seq != tcp_pcb_seq) {
317 			delm = elm;
318 		}
319 	}
320 	if (delm) {
321 		RB_REMOVE(mypfstate_tree, &mypf_tree, delm);
322 		free(delm);
323 		delm = NULL;
324 	}
325 	qsort(array, i, sizeof(array[0]), statebwcmp);
326 
327 	row = 2;
328 	n = i;
329 	for (i = 0; i < n; ++i) {
330 		elm = array[i];
331 		if (elm->state.direction == PF_OUT) {
332 			nk = &elm->state.key[PF_SK_WIRE];
333 		} else {
334 			nk = &elm->state.key[PF_SK_STACK];
335 		}
336 		mvwprintw(wnd, row, 0,
337 			  "%s %s "
338 			  /*"rxb %s txb %s "*/
339 			  "rcv %s snd %s ",
340 			  netaddrstr(elm->state.af, &nk->addr[0], nk->port[0]),
341 			  netaddrstr(elm->state.af, &nk->addr[1], nk->port[1]),
342 			  numtok(DELTARATE(bytes[0])),
343 			  numtok(DELTARATE(bytes[1]))
344 		);
345 		wclrtoeol(wnd);
346 		if (++row >= LINES-3)
347 			break;
348 	}
349 	free(array);
350 	wmove(wnd, row, 0);
351 	wclrtobot(wnd);
352 	mvwprintw(wnd, LINES-2, 0, "Rate bytes/sec, active pf states");
353 }
354 
355 /*
356  * Sort by total bytes transfered, highest first
357  */
358 static
359 int
360 statebwcmp(const void *data1, const void *data2)
361 {
362 	const struct mypfstate *elm1 = *__DECONST(struct mypfstate **, data1);
363 	const struct mypfstate *elm2 = *__DECONST(struct mypfstate **, data2);
364 	uint64_t v1;
365 	uint64_t v2;
366 
367 	v1 = be64toh(*(const uint64_t *)elm1->state.bytes[0]) +
368 	     be64toh(*(const uint64_t *)elm1->state.bytes[1]);
369 	v1 -= be64toh(*(const uint64_t *)elm1->last_state.bytes[0]) +
370 	     be64toh(*(const uint64_t *)elm1->last_state.bytes[1]);
371 	v2 = be64toh(*(const uint64_t *)elm2->state.bytes[0]) +
372 	     be64toh(*(const uint64_t *)elm2->state.bytes[1]);
373 	v2 -= be64toh(*(const uint64_t *)elm2->last_state.bytes[0]) +
374 	     be64toh(*(const uint64_t *)elm2->last_state.bytes[1]);
375 	if (v1 < v2)
376 		return(1);
377 	if (v1 > v2)
378 		return(-1);
379 	return(0);
380 }
381 
382 #if 0
383 int
384 cmdpftop(const char *cmd __unused, char *args __unused)
385 {
386 	fetchpftop();
387 	showpftop();
388 	refresh();
389 
390 	return (0);
391 }
392 #endif
393 
394 #define MAXINDEXES 8
395 
396 static
397 const char *
398 numtok(double value)
399 {
400 	static char buf[MAXINDEXES][32];
401 	static int nexti;
402 	static const char *suffixes[] = { " ", "K", "M", "G", "T", NULL };
403 	int suffix = 0;
404 	const char *fmt;
405 
406 	while (value >= 1000.0 && suffixes[suffix+1]) {
407 		value /= 1000.0;
408 		++suffix;
409 	}
410 	nexti = (nexti + 1) % MAXINDEXES;
411 	if (value < 0.001) {
412 		fmt = "      ";
413 	} else if (value < 1.0) {
414 		fmt = "%5.3f%s";
415 	} else if (value < 10.0) {
416 		fmt = "%5.3f%s";
417 	} else if (value < 100.0) {
418 		fmt = "%5.2f%s";
419 	} else if (value < 1000.0) {
420 		fmt = "%5.1f%s";
421 	} else {
422 		fmt = "<huge>";
423 	}
424 	snprintf(buf[nexti], sizeof(buf[nexti]),
425 		 fmt, value, suffixes[suffix]);
426 	return (buf[nexti]);
427 }
428 
429 static const char *
430 netaddrstr(sa_family_t af, struct pf_addr *addr, u_int16_t port)
431 {
432 	static char buf[MAXINDEXES][64];
433 	static int nexta;
434 	char bufip[64];
435 
436 	nexta = (nexta + 1) % MAXINDEXES;
437 
438 	port = ntohs(port);
439 
440 	if (af == AF_INET) {
441 		snprintf(bufip, sizeof(bufip),
442 			 "%d.%d.%d.%d",
443 			 (ntohl(addr->v4.s_addr) >> 24) & 255,
444 			 (ntohl(addr->v4.s_addr) >> 16) & 255,
445 			 (ntohl(addr->v4.s_addr) >> 8) & 255,
446 			 (ntohl(addr->v4.s_addr) >> 0) & 255);
447 		snprintf(buf[nexta], sizeof(buf[nexta]),
448 			 "%15s:%-5d", bufip, port);
449 	} else if (af == AF_INET6) {
450 		snprintf(bufip, sizeof(bufip),
451 			 "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x",
452 			 ntohs(addr->v6.s6_addr16[0]),
453 			 ntohs(addr->v6.s6_addr16[1]),
454 			 ntohs(addr->v6.s6_addr16[2]),
455 			 ntohs(addr->v6.s6_addr16[3]),
456 			 ntohs(addr->v6.s6_addr16[4]),
457 			 ntohs(addr->v6.s6_addr16[5]),
458 			 ntohs(addr->v6.s6_addr16[6]),
459 			 ntohs(addr->v6.s6_addr16[7]));
460 		snprintf(buf[nexta], sizeof(buf[nexta]),
461 			 "%39s:%-5d", bufip, port);
462 	} else {
463 		snprintf(bufip, sizeof(bufip), "<unknown>:%-5d", port);
464 		snprintf(buf[nexta], sizeof(buf[nexta]),
465 			 "%15s:%-5d", bufip, port);
466 	}
467 	return (buf[nexta]);
468 }
469 
470 static
471 void
472 updatestate(struct pfsync_state *state)
473 {
474 	struct mypfstate dummy;
475 	struct mypfstate *elm;
476 
477 	dummy.state = *state;
478 	if ((elm = RB_FIND(mypfstate_tree, &mypf_tree, &dummy)) == NULL) {
479 		elm = malloc(sizeof(*elm));
480 		bzero(elm, sizeof(*elm));
481 		elm->state = *state;
482 		elm->last_state = *state;
483 		bzero(elm->last_state.bytes,
484 			sizeof(elm->last_state.bytes));
485 		bzero(elm->last_state.packets,
486 			sizeof(elm->last_state.packets));
487 		RB_INSERT(mypfstate_tree, &mypf_tree, elm);
488 	} else {
489 		elm->last_state = elm->state;
490 		elm->state = *state;
491 	}
492 	elm->seq = tcp_pcb_seq;
493 }
494