xref: /freebsd/sys/net/if_loop.c (revision 13de33a5dc2304b13d595d75d48c51793958474f)
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_atalk.h"
38 #include "opt_inet.h"
39 #include "opt_inet6.h"
40 #include "opt_ipx.h"
41 
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/kernel.h>
45 #include <sys/mbuf.h>
46 #include <sys/module.h>
47 #include <machine/bus.h>
48 #include <sys/rman.h>
49 #include <sys/socket.h>
50 #include <sys/sockio.h>
51 #include <sys/sysctl.h>
52 
53 #include <net/if.h>
54 #include <net/if_var.h>
55 #include <net/if_clone.h>
56 #include <net/if_types.h>
57 #include <net/netisr.h>
58 #include <net/route.h>
59 #include <net/bpf.h>
60 #include <net/vnet.h>
61 
62 #ifdef	INET
63 #include <netinet/in.h>
64 #include <netinet/in_var.h>
65 #endif
66 
67 #ifdef IPX
68 #include <netipx/ipx.h>
69 #include <netipx/ipx_if.h>
70 #endif
71 
72 #ifdef INET6
73 #ifndef INET
74 #include <netinet/in.h>
75 #endif
76 #include <netinet6/in6_var.h>
77 #include <netinet/ip6.h>
78 #endif
79 
80 #ifdef NETATALK
81 #include <netatalk/at.h>
82 #include <netatalk/at_var.h>
83 #endif
84 
85 #include <security/mac/mac_framework.h>
86 
87 #ifdef TINY_LOMTU
88 #define	LOMTU	(1024+512)
89 #elif defined(LARGE_LOMTU)
90 #define LOMTU	131072
91 #else
92 #define LOMTU	16384
93 #endif
94 
95 #define	LO_CSUM_FEATURES	(CSUM_IP | CSUM_TCP | CSUM_UDP | CSUM_SCTP)
96 #define	LO_CSUM_FEATURES6	(CSUM_TCP_IPV6 | CSUM_UDP_IPV6 | CSUM_SCTP_IPV6)
97 #define	LO_CSUM_SET		(CSUM_DATA_VALID | CSUM_DATA_VALID_IPV6 | \
98 				    CSUM_PSEUDO_HDR | \
99 				    CSUM_IP_CHECKED | CSUM_IP_VALID | \
100 				    CSUM_SCTP_VALID)
101 
102 int		loioctl(struct ifnet *, u_long, caddr_t);
103 static void	lortrequest(int, struct rtentry *, struct rt_addrinfo *);
104 int		looutput(struct ifnet *ifp, struct mbuf *m,
105 		    const struct sockaddr *dst, struct route *ro);
106 static int	lo_clone_create(struct if_clone *, int, caddr_t);
107 static void	lo_clone_destroy(struct ifnet *);
108 
109 VNET_DEFINE(struct ifnet *, loif);	/* Used externally */
110 
111 #ifdef VIMAGE
112 static VNET_DEFINE(struct if_clone *, lo_cloner);
113 #define	V_lo_cloner		VNET(lo_cloner)
114 #endif
115 
116 static struct if_clone *lo_cloner;
117 static const char loname[] = "lo";
118 
119 static void
120 lo_clone_destroy(struct ifnet *ifp)
121 {
122 
123 #ifndef VIMAGE
124 	/* XXX: destroying lo0 will lead to panics. */
125 	KASSERT(V_loif != ifp, ("%s: destroying lo0", __func__));
126 #endif
127 
128 	bpfdetach(ifp);
129 	if_detach(ifp);
130 	if_free(ifp);
131 }
132 
133 static int
134 lo_clone_create(struct if_clone *ifc, int unit, caddr_t params)
135 {
136 	struct ifnet *ifp;
137 
138 	ifp = if_alloc(IFT_LOOP);
139 	if (ifp == NULL)
140 		return (ENOSPC);
141 
142 	if_initname(ifp, loname, unit);
143 	ifp->if_mtu = LOMTU;
144 	ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST;
145 	ifp->if_ioctl = loioctl;
146 	ifp->if_output = looutput;
147 	ifp->if_snd.ifq_maxlen = ifqmaxlen;
148 	ifp->if_capabilities = ifp->if_capenable =
149 	    IFCAP_HWCSUM | IFCAP_HWCSUM_IPV6;
150 	ifp->if_hwassist = LO_CSUM_FEATURES | LO_CSUM_FEATURES6;
151 	if_attach(ifp);
152 	bpfattach(ifp, DLT_NULL, sizeof(u_int32_t));
153 	if (V_loif == NULL)
154 		V_loif = ifp;
155 
156 	return (0);
157 }
158 
159 static void
160 vnet_loif_init(const void *unused __unused)
161 {
162 
163 #ifdef VIMAGE
164 	lo_cloner = if_clone_simple(loname, lo_clone_create, lo_clone_destroy,
165 	    1);
166 	V_lo_cloner = lo_cloner;
167 #else
168 	lo_cloner = if_clone_simple(loname, lo_clone_create, lo_clone_destroy,
169 	    1);
170 #endif
171 }
172 VNET_SYSINIT(vnet_loif_init, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
173     vnet_loif_init, NULL);
174 
175 #ifdef VIMAGE
176 static void
177 vnet_loif_uninit(const void *unused __unused)
178 {
179 
180 	if_clone_detach(V_lo_cloner);
181 	V_loif = NULL;
182 }
183 VNET_SYSUNINIT(vnet_loif_uninit, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY,
184     vnet_loif_uninit, NULL);
185 #endif
186 
187 static int
188 loop_modevent(module_t mod, int type, void *data)
189 {
190 
191 	switch (type) {
192 	case MOD_LOAD:
193 		break;
194 
195 	case MOD_UNLOAD:
196 		printf("loop module unload - not possible for this module type\n");
197 		return (EINVAL);
198 
199 	default:
200 		return (EOPNOTSUPP);
201 	}
202 	return (0);
203 }
204 
205 static moduledata_t loop_mod = {
206 	"if_lo",
207 	loop_modevent,
208 	0
209 };
210 
211 DECLARE_MODULE(if_lo, loop_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
212 
213 int
214 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
215     struct route *ro)
216 {
217 	u_int32_t af;
218 	struct rtentry *rt = NULL;
219 #ifdef MAC
220 	int error;
221 #endif
222 
223 	M_ASSERTPKTHDR(m); /* check if we have the packet header */
224 
225 	if (ro != NULL)
226 		rt = ro->ro_rt;
227 #ifdef MAC
228 	error = mac_ifnet_check_transmit(ifp, m);
229 	if (error) {
230 		m_freem(m);
231 		return (error);
232 	}
233 #endif
234 
235 	if (rt && rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) {
236 		m_freem(m);
237 		return (rt->rt_flags & RTF_BLACKHOLE ? 0 :
238 		        rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
239 	}
240 
241 	ifp->if_opackets++;
242 	ifp->if_obytes += m->m_pkthdr.len;
243 
244 	/* BPF writes need to be handled specially. */
245 	if (dst->sa_family == AF_UNSPEC)
246 		bcopy(dst->sa_data, &af, sizeof(af));
247 	else
248 		af = dst->sa_family;
249 
250 #if 1	/* XXX */
251 	switch (af) {
252 	case AF_INET:
253 		if (ifp->if_capenable & IFCAP_RXCSUM) {
254 			m->m_pkthdr.csum_data = 0xffff;
255 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
256 		}
257 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES;
258 		break;
259 	case AF_INET6:
260 #if 0
261 		/*
262 		 * XXX-BZ for now always claim the checksum is good despite
263 		 * any interface flags.   This is a workaround for 9.1-R and
264 		 * a proper solution ought to be sought later.
265 		 */
266 		if (ifp->if_capenable & IFCAP_RXCSUM_IPV6) {
267 			m->m_pkthdr.csum_data = 0xffff;
268 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
269 		}
270 #else
271 		m->m_pkthdr.csum_data = 0xffff;
272 		m->m_pkthdr.csum_flags = LO_CSUM_SET;
273 #endif
274 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES6;
275 		break;
276 	case AF_IPX:
277 	case AF_APPLETALK:
278 		break;
279 	default:
280 		printf("looutput: af=%d unexpected\n", af);
281 		m_freem(m);
282 		return (EAFNOSUPPORT);
283 	}
284 #endif
285 	return (if_simloop(ifp, m, af, 0));
286 }
287 
288 /*
289  * if_simloop()
290  *
291  * This function is to support software emulation of hardware loopback,
292  * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't
293  * hear their own broadcasts, we create a copy of the packet that we
294  * would normally receive via a hardware loopback.
295  *
296  * This function expects the packet to include the media header of length hlen.
297  */
298 int
299 if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen)
300 {
301 	int isr;
302 
303 	M_ASSERTPKTHDR(m);
304 	m_tag_delete_nonpersistent(m);
305 	m->m_pkthdr.rcvif = ifp;
306 
307 #ifdef MAC
308 	mac_ifnet_create_mbuf(ifp, m);
309 #endif
310 
311 	/*
312 	 * Let BPF see incoming packet in the following manner:
313 	 *  - Emulated packet loopback for a simplex interface
314 	 *    (net/if_ethersubr.c)
315 	 *	-> passes it to ifp's BPF
316 	 *  - IPv4/v6 multicast packet loopback (netinet(6)/ip(6)_output.c)
317 	 *	-> not passes it to any BPF
318 	 *  - Normal packet loopback from myself to myself (net/if_loop.c)
319 	 *	-> passes to lo0's BPF (even in case of IPv6, where ifp!=lo0)
320 	 */
321 	if (hlen > 0) {
322 		if (bpf_peers_present(ifp->if_bpf)) {
323 			bpf_mtap(ifp->if_bpf, m);
324 		}
325 	} else {
326 		if (bpf_peers_present(V_loif->if_bpf)) {
327 			if ((m->m_flags & M_MCAST) == 0 || V_loif == ifp) {
328 				/* XXX beware sizeof(af) != 4 */
329 				u_int32_t af1 = af;
330 
331 				/*
332 				 * We need to prepend the address family.
333 				 */
334 				bpf_mtap2(V_loif->if_bpf, &af1, sizeof(af1), m);
335 			}
336 		}
337 	}
338 
339 	/* Strip away media header */
340 	if (hlen > 0) {
341 		m_adj(m, hlen);
342 #ifndef __NO_STRICT_ALIGNMENT
343 		/*
344 		 * Some archs do not like unaligned data, so
345 		 * we move data down in the first mbuf.
346 		 */
347 		if (mtod(m, vm_offset_t) & 3) {
348 			KASSERT(hlen >= 3, ("if_simloop: hlen too small"));
349 			bcopy(m->m_data,
350 			    (char *)(mtod(m, vm_offset_t)
351 				- (mtod(m, vm_offset_t) & 3)),
352 			    m->m_len);
353 			m->m_data -= (mtod(m,vm_offset_t) & 3);
354 		}
355 #endif
356 	}
357 
358 	/* Deliver to upper layer protocol */
359 	switch (af) {
360 #ifdef INET
361 	case AF_INET:
362 		isr = NETISR_IP;
363 		break;
364 #endif
365 #ifdef INET6
366 	case AF_INET6:
367 		m->m_flags |= M_LOOP;
368 		isr = NETISR_IPV6;
369 		break;
370 #endif
371 #ifdef IPX
372 	case AF_IPX:
373 		isr = NETISR_IPX;
374 		break;
375 #endif
376 #ifdef NETATALK
377 	case AF_APPLETALK:
378 		isr = NETISR_ATALK2;
379 		break;
380 #endif
381 	default:
382 		printf("if_simloop: can't handle af=%d\n", af);
383 		m_freem(m);
384 		return (EAFNOSUPPORT);
385 	}
386 	ifp->if_ipackets++;
387 	ifp->if_ibytes += m->m_pkthdr.len;
388 	netisr_queue(isr, m);	/* mbuf is free'd on failure. */
389 	return (0);
390 }
391 
392 /* ARGSUSED */
393 static void
394 lortrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info)
395 {
396 
397 	RT_LOCK_ASSERT(rt);
398 	rt->rt_rmx.rmx_mtu = rt->rt_ifp->if_mtu;
399 }
400 
401 /*
402  * Process an ioctl request.
403  */
404 /* ARGSUSED */
405 int
406 loioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
407 {
408 	struct ifaddr *ifa;
409 	struct ifreq *ifr = (struct ifreq *)data;
410 	int error = 0, mask;
411 
412 	switch (cmd) {
413 	case SIOCSIFADDR:
414 		ifp->if_flags |= IFF_UP;
415 		ifp->if_drv_flags |= IFF_DRV_RUNNING;
416 		ifa = (struct ifaddr *)data;
417 		ifa->ifa_rtrequest = lortrequest;
418 		/*
419 		 * Everything else is done at a higher level.
420 		 */
421 		break;
422 
423 	case SIOCADDMULTI:
424 	case SIOCDELMULTI:
425 		if (ifr == 0) {
426 			error = EAFNOSUPPORT;		/* XXX */
427 			break;
428 		}
429 		switch (ifr->ifr_addr.sa_family) {
430 
431 #ifdef INET
432 		case AF_INET:
433 			break;
434 #endif
435 #ifdef INET6
436 		case AF_INET6:
437 			break;
438 #endif
439 
440 		default:
441 			error = EAFNOSUPPORT;
442 			break;
443 		}
444 		break;
445 
446 	case SIOCSIFMTU:
447 		ifp->if_mtu = ifr->ifr_mtu;
448 		break;
449 
450 	case SIOCSIFFLAGS:
451 		break;
452 
453 	case SIOCSIFCAP:
454 		mask = ifp->if_capenable ^ ifr->ifr_reqcap;
455 		if ((mask & IFCAP_RXCSUM) != 0)
456 			ifp->if_capenable ^= IFCAP_RXCSUM;
457 		if ((mask & IFCAP_TXCSUM) != 0)
458 			ifp->if_capenable ^= IFCAP_TXCSUM;
459 		if ((mask & IFCAP_RXCSUM_IPV6) != 0) {
460 #if 0
461 			ifp->if_capenable ^= IFCAP_RXCSUM_IPV6;
462 #else
463 			error = EOPNOTSUPP;
464 			break;
465 #endif
466 		}
467 		if ((mask & IFCAP_TXCSUM_IPV6) != 0) {
468 #if 0
469 			ifp->if_capenable ^= IFCAP_TXCSUM_IPV6;
470 #else
471 			error = EOPNOTSUPP;
472 			break;
473 #endif
474 		}
475 		ifp->if_hwassist = 0;
476 		if (ifp->if_capenable & IFCAP_TXCSUM)
477 			ifp->if_hwassist = LO_CSUM_FEATURES;
478 #if 0
479 		if (ifp->if_capenable & IFCAP_TXCSUM_IPV6)
480 			ifp->if_hwassist |= LO_CSUM_FEATURES6;
481 #endif
482 		break;
483 
484 	default:
485 		error = EINVAL;
486 	}
487 	return (error);
488 }
489