xref: /freebsd/sys/net/if_loop.c (revision 40a8ac8f62b535d30349faf28cf47106b7041b83)
1 /*-
2  * Copyright (c) 1982, 1986, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)if_loop.c	8.2 (Berkeley) 1/9/95
30  * $FreeBSD$
31  */
32 
33 /*
34  * Loopback interface driver for protocol testing and timing.
35  */
36 
37 #include "opt_inet.h"
38 #include "opt_inet6.h"
39 
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/kernel.h>
43 #include <sys/mbuf.h>
44 #include <sys/module.h>
45 #include <machine/bus.h>
46 #include <sys/rman.h>
47 #include <sys/socket.h>
48 #include <sys/sockio.h>
49 #include <sys/sysctl.h>
50 
51 #include <net/if.h>
52 #include <net/if_var.h>
53 #include <net/if_clone.h>
54 #include <net/if_types.h>
55 #include <net/netisr.h>
56 #include <net/route.h>
57 #include <net/bpf.h>
58 #include <net/vnet.h>
59 
60 #ifdef	INET
61 #include <netinet/in.h>
62 #include <netinet/in_var.h>
63 #endif
64 
65 #ifdef INET6
66 #ifndef INET
67 #include <netinet/in.h>
68 #endif
69 #include <netinet6/in6_var.h>
70 #include <netinet/ip6.h>
71 #endif
72 
73 #include <security/mac/mac_framework.h>
74 
75 #ifdef TINY_LOMTU
76 #define	LOMTU	(1024+512)
77 #elif defined(LARGE_LOMTU)
78 #define LOMTU	131072
79 #else
80 #define LOMTU	16384
81 #endif
82 
83 #define	LO_CSUM_FEATURES	(CSUM_IP | CSUM_TCP | CSUM_UDP | CSUM_SCTP)
84 #define	LO_CSUM_FEATURES6	(CSUM_TCP_IPV6 | CSUM_UDP_IPV6 | CSUM_SCTP_IPV6)
85 #define	LO_CSUM_SET		(CSUM_DATA_VALID | CSUM_DATA_VALID_IPV6 | \
86 				    CSUM_PSEUDO_HDR | \
87 				    CSUM_IP_CHECKED | CSUM_IP_VALID | \
88 				    CSUM_SCTP_VALID)
89 
90 int		loioctl(struct ifnet *, u_long, caddr_t);
91 static void	lortrequest(int, struct rtentry *, struct rt_addrinfo *);
92 int		looutput(struct ifnet *ifp, struct mbuf *m,
93 		    const struct sockaddr *dst, struct route *ro);
94 static int	lo_clone_create(struct if_clone *, int, caddr_t);
95 static void	lo_clone_destroy(struct ifnet *);
96 
97 VNET_DEFINE(struct ifnet *, loif);	/* Used externally */
98 
99 #ifdef VIMAGE
100 static VNET_DEFINE(struct if_clone *, lo_cloner);
101 #define	V_lo_cloner		VNET(lo_cloner)
102 #endif
103 
104 static struct if_clone *lo_cloner;
105 static const char loname[] = "lo";
106 
107 static void
108 lo_clone_destroy(struct ifnet *ifp)
109 {
110 
111 #ifndef VIMAGE
112 	/* XXX: destroying lo0 will lead to panics. */
113 	KASSERT(V_loif != ifp, ("%s: destroying lo0", __func__));
114 #endif
115 
116 	bpfdetach(ifp);
117 	if_detach(ifp);
118 	if_free(ifp);
119 }
120 
121 static int
122 lo_clone_create(struct if_clone *ifc, int unit, caddr_t params)
123 {
124 	struct ifnet *ifp;
125 
126 	ifp = if_alloc(IFT_LOOP);
127 	if (ifp == NULL)
128 		return (ENOSPC);
129 
130 	if_initname(ifp, loname, unit);
131 	ifp->if_mtu = LOMTU;
132 	ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST;
133 	ifp->if_ioctl = loioctl;
134 	ifp->if_output = looutput;
135 	ifp->if_snd.ifq_maxlen = ifqmaxlen;
136 	ifp->if_capabilities = ifp->if_capenable =
137 	    IFCAP_HWCSUM | IFCAP_HWCSUM_IPV6;
138 	ifp->if_hwassist = LO_CSUM_FEATURES | LO_CSUM_FEATURES6;
139 	if_attach(ifp);
140 	bpfattach(ifp, DLT_NULL, sizeof(u_int32_t));
141 	if (V_loif == NULL)
142 		V_loif = ifp;
143 
144 	return (0);
145 }
146 
147 static void
148 vnet_loif_init(const void *unused __unused)
149 {
150 
151 #ifdef VIMAGE
152 	lo_cloner = if_clone_simple(loname, lo_clone_create, lo_clone_destroy,
153 	    1);
154 	V_lo_cloner = lo_cloner;
155 #else
156 	lo_cloner = if_clone_simple(loname, lo_clone_create, lo_clone_destroy,
157 	    1);
158 #endif
159 }
160 VNET_SYSINIT(vnet_loif_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
161     vnet_loif_init, NULL);
162 
163 #ifdef VIMAGE
164 static void
165 vnet_loif_uninit(const void *unused __unused)
166 {
167 
168 	if_clone_detach(V_lo_cloner);
169 	V_loif = NULL;
170 }
171 VNET_SYSUNINIT(vnet_loif_uninit, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
172     vnet_loif_uninit, NULL);
173 #endif
174 
175 static int
176 loop_modevent(module_t mod, int type, void *data)
177 {
178 
179 	switch (type) {
180 	case MOD_LOAD:
181 		break;
182 
183 	case MOD_UNLOAD:
184 		printf("loop module unload - not possible for this module type\n");
185 		return (EINVAL);
186 
187 	default:
188 		return (EOPNOTSUPP);
189 	}
190 	return (0);
191 }
192 
193 static moduledata_t loop_mod = {
194 	"if_lo",
195 	loop_modevent,
196 	0
197 };
198 
199 DECLARE_MODULE(if_lo, loop_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
200 
201 int
202 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
203     struct route *ro)
204 {
205 	u_int32_t af;
206 	struct rtentry *rt = NULL;
207 #ifdef MAC
208 	int error;
209 #endif
210 
211 	M_ASSERTPKTHDR(m); /* check if we have the packet header */
212 
213 	if (ro != NULL)
214 		rt = ro->ro_rt;
215 #ifdef MAC
216 	error = mac_ifnet_check_transmit(ifp, m);
217 	if (error) {
218 		m_freem(m);
219 		return (error);
220 	}
221 #endif
222 
223 	if (rt && rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) {
224 		m_freem(m);
225 		return (rt->rt_flags & RTF_BLACKHOLE ? 0 :
226 		        rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
227 	}
228 
229 	if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
230 	if_inc_counter(ifp, IFCOUNTER_OBYTES, m->m_pkthdr.len);
231 
232 	/* BPF writes need to be handled specially. */
233 	if (dst->sa_family == AF_UNSPEC)
234 		bcopy(dst->sa_data, &af, sizeof(af));
235 	else
236 		af = dst->sa_family;
237 
238 #if 1	/* XXX */
239 	switch (af) {
240 	case AF_INET:
241 		if (ifp->if_capenable & IFCAP_RXCSUM) {
242 			m->m_pkthdr.csum_data = 0xffff;
243 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
244 		}
245 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES;
246 		break;
247 	case AF_INET6:
248 #if 0
249 		/*
250 		 * XXX-BZ for now always claim the checksum is good despite
251 		 * any interface flags.   This is a workaround for 9.1-R and
252 		 * a proper solution ought to be sought later.
253 		 */
254 		if (ifp->if_capenable & IFCAP_RXCSUM_IPV6) {
255 			m->m_pkthdr.csum_data = 0xffff;
256 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
257 		}
258 #else
259 		m->m_pkthdr.csum_data = 0xffff;
260 		m->m_pkthdr.csum_flags = LO_CSUM_SET;
261 #endif
262 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES6;
263 		break;
264 	default:
265 		printf("looutput: af=%d unexpected\n", af);
266 		m_freem(m);
267 		return (EAFNOSUPPORT);
268 	}
269 #endif
270 	return (if_simloop(ifp, m, af, 0));
271 }
272 
273 /*
274  * if_simloop()
275  *
276  * This function is to support software emulation of hardware loopback,
277  * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't
278  * hear their own broadcasts, we create a copy of the packet that we
279  * would normally receive via a hardware loopback.
280  *
281  * This function expects the packet to include the media header of length hlen.
282  */
283 int
284 if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen)
285 {
286 	int isr;
287 
288 	M_ASSERTPKTHDR(m);
289 	m_tag_delete_nonpersistent(m);
290 	m->m_pkthdr.rcvif = ifp;
291 
292 #ifdef MAC
293 	mac_ifnet_create_mbuf(ifp, m);
294 #endif
295 
296 	/*
297 	 * Let BPF see incoming packet in the following manner:
298 	 *  - Emulated packet loopback for a simplex interface
299 	 *    (net/if_ethersubr.c)
300 	 *	-> passes it to ifp's BPF
301 	 *  - IPv4/v6 multicast packet loopback (netinet(6)/ip(6)_output.c)
302 	 *	-> not passes it to any BPF
303 	 *  - Normal packet loopback from myself to myself (net/if_loop.c)
304 	 *	-> passes to lo0's BPF (even in case of IPv6, where ifp!=lo0)
305 	 */
306 	if (hlen > 0) {
307 		if (bpf_peers_present(ifp->if_bpf)) {
308 			bpf_mtap(ifp->if_bpf, m);
309 		}
310 	} else {
311 		if (bpf_peers_present(V_loif->if_bpf)) {
312 			if ((m->m_flags & M_MCAST) == 0 || V_loif == ifp) {
313 				/* XXX beware sizeof(af) != 4 */
314 				u_int32_t af1 = af;
315 
316 				/*
317 				 * We need to prepend the address family.
318 				 */
319 				bpf_mtap2(V_loif->if_bpf, &af1, sizeof(af1), m);
320 			}
321 		}
322 	}
323 
324 	/* Strip away media header */
325 	if (hlen > 0) {
326 		m_adj(m, hlen);
327 #ifndef __NO_STRICT_ALIGNMENT
328 		/*
329 		 * Some archs do not like unaligned data, so
330 		 * we move data down in the first mbuf.
331 		 */
332 		if (mtod(m, vm_offset_t) & 3) {
333 			KASSERT(hlen >= 3, ("if_simloop: hlen too small"));
334 			bcopy(m->m_data,
335 			    (char *)(mtod(m, vm_offset_t)
336 				- (mtod(m, vm_offset_t) & 3)),
337 			    m->m_len);
338 			m->m_data -= (mtod(m,vm_offset_t) & 3);
339 		}
340 #endif
341 	}
342 
343 	/* Deliver to upper layer protocol */
344 	switch (af) {
345 #ifdef INET
346 	case AF_INET:
347 		isr = NETISR_IP;
348 		break;
349 #endif
350 #ifdef INET6
351 	case AF_INET6:
352 		m->m_flags |= M_LOOP;
353 		isr = NETISR_IPV6;
354 		break;
355 #endif
356 	default:
357 		printf("if_simloop: can't handle af=%d\n", af);
358 		m_freem(m);
359 		return (EAFNOSUPPORT);
360 	}
361 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
362 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
363 	netisr_queue(isr, m);	/* mbuf is free'd on failure. */
364 	return (0);
365 }
366 
367 /* ARGSUSED */
368 static void
369 lortrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info)
370 {
371 
372 	RT_LOCK_ASSERT(rt);
373 	rt->rt_mtu = rt->rt_ifp->if_mtu;
374 }
375 
376 /*
377  * Process an ioctl request.
378  */
379 /* ARGSUSED */
380 int
381 loioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
382 {
383 	struct ifaddr *ifa;
384 	struct ifreq *ifr = (struct ifreq *)data;
385 	int error = 0, mask;
386 
387 	switch (cmd) {
388 	case SIOCSIFADDR:
389 		ifp->if_flags |= IFF_UP;
390 		ifp->if_drv_flags |= IFF_DRV_RUNNING;
391 		ifa = (struct ifaddr *)data;
392 		ifa->ifa_rtrequest = lortrequest;
393 		/*
394 		 * Everything else is done at a higher level.
395 		 */
396 		break;
397 
398 	case SIOCADDMULTI:
399 	case SIOCDELMULTI:
400 		if (ifr == 0) {
401 			error = EAFNOSUPPORT;		/* XXX */
402 			break;
403 		}
404 		switch (ifr->ifr_addr.sa_family) {
405 
406 #ifdef INET
407 		case AF_INET:
408 			break;
409 #endif
410 #ifdef INET6
411 		case AF_INET6:
412 			break;
413 #endif
414 
415 		default:
416 			error = EAFNOSUPPORT;
417 			break;
418 		}
419 		break;
420 
421 	case SIOCSIFMTU:
422 		ifp->if_mtu = ifr->ifr_mtu;
423 		break;
424 
425 	case SIOCSIFFLAGS:
426 		break;
427 
428 	case SIOCSIFCAP:
429 		mask = ifp->if_capenable ^ ifr->ifr_reqcap;
430 		if ((mask & IFCAP_RXCSUM) != 0)
431 			ifp->if_capenable ^= IFCAP_RXCSUM;
432 		if ((mask & IFCAP_TXCSUM) != 0)
433 			ifp->if_capenable ^= IFCAP_TXCSUM;
434 		if ((mask & IFCAP_RXCSUM_IPV6) != 0) {
435 #if 0
436 			ifp->if_capenable ^= IFCAP_RXCSUM_IPV6;
437 #else
438 			error = EOPNOTSUPP;
439 			break;
440 #endif
441 		}
442 		if ((mask & IFCAP_TXCSUM_IPV6) != 0) {
443 #if 0
444 			ifp->if_capenable ^= IFCAP_TXCSUM_IPV6;
445 #else
446 			error = EOPNOTSUPP;
447 			break;
448 #endif
449 		}
450 		ifp->if_hwassist = 0;
451 		if (ifp->if_capenable & IFCAP_TXCSUM)
452 			ifp->if_hwassist = LO_CSUM_FEATURES;
453 #if 0
454 		if (ifp->if_capenable & IFCAP_TXCSUM_IPV6)
455 			ifp->if_hwassist |= LO_CSUM_FEATURES6;
456 #endif
457 		break;
458 
459 	default:
460 		error = EINVAL;
461 	}
462 	return (error);
463 }
464