xref: /freebsd/sys/net/if_loop.c (revision d0b2dbfa0ecf2bbc9709efc5e20baf8e4b44bbbf)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 1993
5  *	The Regents of the University of California.  All rights reserved.
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  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	@(#)if_loop.c	8.2 (Berkeley) 1/9/95
32  */
33 
34 /*
35  * Loopback interface driver for protocol testing and timing.
36  */
37 
38 #include "opt_inet.h"
39 #include "opt_inet6.h"
40 #include "opt_rss.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_private.h>
56 #include <net/if_clone.h>
57 #include <net/if_types.h>
58 #include <net/netisr.h>
59 #include <net/route.h>
60 #include <net/bpf.h>
61 #include <net/vnet.h>
62 
63 #ifdef	INET
64 #include <netinet/in.h>
65 #include <netinet/in_var.h>
66 #endif
67 
68 #ifdef INET6
69 #ifndef INET
70 #include <netinet/in.h>
71 #endif
72 #include <netinet6/in6_var.h>
73 #include <netinet/ip6.h>
74 #endif
75 
76 #include <security/mac/mac_framework.h>
77 
78 #ifdef TINY_LOMTU
79 #define	LOMTU	(1024+512)
80 #elif defined(LARGE_LOMTU)
81 #define LOMTU	131072
82 #else
83 #define LOMTU	16384
84 #endif
85 
86 #define	LO_CSUM_FEATURES	(CSUM_IP | CSUM_TCP | CSUM_UDP | CSUM_SCTP)
87 #define	LO_CSUM_FEATURES6	(CSUM_TCP_IPV6 | CSUM_UDP_IPV6 | CSUM_SCTP_IPV6)
88 #define	LO_CSUM_SET		(CSUM_DATA_VALID | CSUM_DATA_VALID_IPV6 | \
89 				    CSUM_PSEUDO_HDR | \
90 				    CSUM_IP_CHECKED | CSUM_IP_VALID | \
91 				    CSUM_SCTP_VALID)
92 
93 static int	loioctl(struct ifnet *, u_long, caddr_t);
94 static int	looutput(struct ifnet *ifp, struct mbuf *m,
95 		    const struct sockaddr *dst, struct route *ro);
96 
97 VNET_DEFINE(struct ifnet *, loif);	/* Used externally */
98 
99 #ifdef VIMAGE
100 VNET_DEFINE_STATIC(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 int
108 lo_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
109 {
110 	if (ifp->if_dunit == 0 && (flags & IFC_F_FORCE) == 0)
111 		return (EINVAL);
112 
113 #ifndef VIMAGE
114 	/* XXX: destroying lo0 will lead to panics. */
115 	KASSERT(V_loif != ifp, ("%s: destroying lo0", __func__));
116 #endif
117 
118 	bpfdetach(ifp);
119 	if_detach(ifp);
120 	if_free(ifp);
121 
122 	return (0);
123 }
124 
125 static int
126 lo_clone_create(struct if_clone *ifc, char *name, size_t len,
127     struct ifc_data *ifd, struct ifnet **ifpp)
128 {
129 	struct ifnet *ifp;
130 
131 	ifp = if_alloc(IFT_LOOP);
132 	if (ifp == NULL)
133 		return (ENOSPC);
134 
135 	if_initname(ifp, loname, ifd->unit);
136 	ifp->if_mtu = LOMTU;
137 	ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST;
138 	ifp->if_ioctl = loioctl;
139 	ifp->if_output = looutput;
140 	ifp->if_snd.ifq_maxlen = ifqmaxlen;
141 	ifp->if_capabilities = ifp->if_capenable =
142 	    IFCAP_HWCSUM | IFCAP_HWCSUM_IPV6 | IFCAP_LINKSTATE;
143 	ifp->if_hwassist = LO_CSUM_FEATURES | LO_CSUM_FEATURES6;
144 	if_attach(ifp);
145 	bpfattach(ifp, DLT_NULL, sizeof(u_int32_t));
146 	if (V_loif == NULL)
147 		V_loif = ifp;
148 	*ifpp = ifp;
149 
150 	return (0);
151 }
152 
153 static void
154 vnet_loif_init(const void *unused __unused)
155 {
156 	struct if_clone_addreq req = {
157 		.create_f = lo_clone_create,
158 		.destroy_f = lo_clone_destroy,
159 		.flags = IFC_F_AUTOUNIT,
160 	};
161 	lo_cloner = ifc_attach_cloner(loname, &req);
162 #ifdef VIMAGE
163 	V_lo_cloner = lo_cloner;
164 #endif
165 	struct ifc_data ifd = { .unit = 0 };
166 	ifc_create_ifp(loname, &ifd, NULL);
167 }
168 VNET_SYSINIT(vnet_loif_init, SI_SUB_PSEUDO, SI_ORDER_ANY,
169     vnet_loif_init, NULL);
170 
171 #ifdef VIMAGE
172 static void
173 vnet_loif_uninit(const void *unused __unused)
174 {
175 
176 	ifc_detach_cloner(V_lo_cloner);
177 	V_loif = NULL;
178 }
179 VNET_SYSUNINIT(vnet_loif_uninit, SI_SUB_INIT_IF, SI_ORDER_SECOND,
180     vnet_loif_uninit, NULL);
181 #endif
182 
183 static int
184 loop_modevent(module_t mod, int type, void *data)
185 {
186 
187 	switch (type) {
188 	case MOD_LOAD:
189 		break;
190 
191 	case MOD_UNLOAD:
192 		printf("loop module unload - not possible for this module type\n");
193 		return (EINVAL);
194 
195 	default:
196 		return (EOPNOTSUPP);
197 	}
198 	return (0);
199 }
200 
201 static moduledata_t loop_mod = {
202 	"if_lo",
203 	loop_modevent,
204 	0
205 };
206 
207 DECLARE_MODULE(if_lo, loop_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
208 
209 static int
210 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
211     struct route *ro)
212 {
213 	u_int32_t af;
214 #ifdef MAC
215 	int error;
216 #endif
217 
218 	M_ASSERTPKTHDR(m); /* check if we have the packet header */
219 
220 #ifdef MAC
221 	error = mac_ifnet_check_transmit(ifp, m);
222 	if (error) {
223 		m_freem(m);
224 		return (error);
225 	}
226 #endif
227 
228 	if (ro != NULL && ro->ro_flags & (RT_REJECT|RT_BLACKHOLE)) {
229 		m_freem(m);
230 		return (ro->ro_flags & RT_BLACKHOLE ? 0 : EHOSTUNREACH);
231 	}
232 
233 	if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
234 	if_inc_counter(ifp, IFCOUNTER_OBYTES, m->m_pkthdr.len);
235 
236 #ifdef RSS
237 	M_HASHTYPE_CLEAR(m);
238 #endif
239 
240 	/* BPF writes need to be handled specially. */
241 	if (dst->sa_family == AF_UNSPEC || dst->sa_family == pseudo_AF_HDRCMPLT)
242 		bcopy(dst->sa_data, &af, sizeof(af));
243 	else
244 		af = RO_GET_FAMILY(ro, dst);
245 
246 #if 1	/* XXX */
247 	switch (af) {
248 	case AF_INET:
249 		if (ifp->if_capenable & IFCAP_RXCSUM) {
250 			m->m_pkthdr.csum_data = 0xffff;
251 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
252 		}
253 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES;
254 		break;
255 	case AF_INET6:
256 #if 0
257 		/*
258 		 * XXX-BZ for now always claim the checksum is good despite
259 		 * any interface flags.   This is a workaround for 9.1-R and
260 		 * a proper solution ought to be sought later.
261 		 */
262 		if (ifp->if_capenable & IFCAP_RXCSUM_IPV6) {
263 			m->m_pkthdr.csum_data = 0xffff;
264 			m->m_pkthdr.csum_flags = LO_CSUM_SET;
265 		}
266 #else
267 		m->m_pkthdr.csum_data = 0xffff;
268 		m->m_pkthdr.csum_flags = LO_CSUM_SET;
269 #endif
270 		m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES6;
271 		break;
272 	default:
273 		printf("looutput: af=%d unexpected\n", af);
274 		m_freem(m);
275 		return (EAFNOSUPPORT);
276 	}
277 #endif
278 	return (if_simloop(ifp, m, af, 0));
279 }
280 
281 /*
282  * if_simloop()
283  *
284  * This function is to support software emulation of hardware loopback,
285  * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't
286  * hear their own broadcasts, we create a copy of the packet that we
287  * would normally receive via a hardware loopback.
288  *
289  * This function expects the packet to include the media header of length hlen.
290  */
291 int
292 if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen)
293 {
294 	int isr;
295 
296 	M_ASSERTPKTHDR(m);
297 	m_tag_delete_nonpersistent(m);
298 	m->m_pkthdr.rcvif = ifp;
299 
300 #ifdef MAC
301 	mac_ifnet_create_mbuf(ifp, m);
302 #endif
303 
304 	/*
305 	 * Let BPF see incoming packet in the following manner:
306 	 *  - Emulated packet loopback for a simplex interface
307 	 *    (net/if_ethersubr.c)
308 	 *	-> passes it to ifp's BPF
309 	 *  - IPv4/v6 multicast packet loopback (netinet(6)/ip(6)_output.c)
310 	 *	-> not passes it to any BPF
311 	 *  - Normal packet loopback from myself to myself (net/if_loop.c)
312 	 *	-> passes to lo0's BPF (even in case of IPv6, where ifp!=lo0)
313 	 */
314 	if (hlen > 0) {
315 		if (bpf_peers_present(ifp->if_bpf)) {
316 			bpf_mtap(ifp->if_bpf, m);
317 		}
318 	} else {
319 		if (bpf_peers_present(V_loif->if_bpf)) {
320 			if ((m->m_flags & M_MCAST) == 0 || V_loif == ifp) {
321 				/* XXX beware sizeof(af) != 4 */
322 				u_int32_t af1 = af;
323 
324 				/*
325 				 * We need to prepend the address family.
326 				 */
327 				bpf_mtap2(V_loif->if_bpf, &af1, sizeof(af1), m);
328 			}
329 		}
330 	}
331 
332 	/* Strip away media header */
333 	if (hlen > 0) {
334 		m_adj(m, hlen);
335 #ifndef __NO_STRICT_ALIGNMENT
336 		/*
337 		 * Some archs do not like unaligned data, so
338 		 * we move data down in the first mbuf.
339 		 */
340 		if (mtod(m, vm_offset_t) & 3) {
341 			KASSERT(hlen >= 3, ("if_simloop: hlen too small"));
342 			bcopy(m->m_data,
343 			    (char *)(mtod(m, vm_offset_t)
344 				- (mtod(m, vm_offset_t) & 3)),
345 			    m->m_len);
346 			m->m_data -= (mtod(m,vm_offset_t) & 3);
347 		}
348 #endif
349 	}
350 
351 	/* Deliver to upper layer protocol */
352 	switch (af) {
353 #ifdef INET
354 	case AF_INET:
355 		isr = NETISR_IP;
356 		break;
357 #endif
358 #ifdef INET6
359 	case AF_INET6:
360 		m->m_flags |= M_LOOP;
361 		isr = NETISR_IPV6;
362 		break;
363 #endif
364 	default:
365 		printf("if_simloop: can't handle af=%d\n", af);
366 		m_freem(m);
367 		return (EAFNOSUPPORT);
368 	}
369 	if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
370 	if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
371 	netisr_queue(isr, m);	/* mbuf is free'd on failure. */
372 	return (0);
373 }
374 
375 /*
376  * Process an ioctl request.
377  */
378 static int
379 loioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
380 {
381 	struct ifreq *ifr = (struct ifreq *)data;
382 	int error = 0, mask;
383 
384 	switch (cmd) {
385 	case SIOCSIFADDR:
386 		ifp->if_flags |= IFF_UP;
387 		ifp->if_drv_flags |= IFF_DRV_RUNNING;
388 		if_link_state_change(ifp, LINK_STATE_UP);
389 		/*
390 		 * Everything else is done at a higher level.
391 		 */
392 		break;
393 
394 	case SIOCADDMULTI:
395 	case SIOCDELMULTI:
396 		if (ifr == NULL) {
397 			error = EAFNOSUPPORT;		/* XXX */
398 			break;
399 		}
400 		switch (ifr->ifr_addr.sa_family) {
401 #ifdef INET
402 		case AF_INET:
403 			break;
404 #endif
405 #ifdef INET6
406 		case AF_INET6:
407 			break;
408 #endif
409 
410 		default:
411 			error = EAFNOSUPPORT;
412 			break;
413 		}
414 		break;
415 
416 	case SIOCSIFMTU:
417 		ifp->if_mtu = ifr->ifr_mtu;
418 		break;
419 
420 	case SIOCSIFFLAGS:
421 		if_link_state_change(ifp, (ifp->if_flags & IFF_UP) ?
422 		    LINK_STATE_UP: LINK_STATE_DOWN);
423 		break;
424 
425 	case SIOCSIFCAP:
426 		mask = ifp->if_capenable ^ ifr->ifr_reqcap;
427 		if ((mask & IFCAP_RXCSUM) != 0)
428 			ifp->if_capenable ^= IFCAP_RXCSUM;
429 		if ((mask & IFCAP_TXCSUM) != 0)
430 			ifp->if_capenable ^= IFCAP_TXCSUM;
431 		if ((mask & IFCAP_RXCSUM_IPV6) != 0) {
432 #if 0
433 			ifp->if_capenable ^= IFCAP_RXCSUM_IPV6;
434 #else
435 			error = EOPNOTSUPP;
436 			break;
437 #endif
438 		}
439 		if ((mask & IFCAP_TXCSUM_IPV6) != 0) {
440 #if 0
441 			ifp->if_capenable ^= IFCAP_TXCSUM_IPV6;
442 #else
443 			error = EOPNOTSUPP;
444 			break;
445 #endif
446 		}
447 		ifp->if_hwassist = 0;
448 		if (ifp->if_capenable & IFCAP_TXCSUM)
449 			ifp->if_hwassist = LO_CSUM_FEATURES;
450 #if 0
451 		if (ifp->if_capenable & IFCAP_TXCSUM_IPV6)
452 			ifp->if_hwassist |= LO_CSUM_FEATURES6;
453 #endif
454 		break;
455 
456 	default:
457 		error = EINVAL;
458 	}
459 	return (error);
460 }
461