xref: /freebsd/sys/net/if_loop.c (revision d15f2551b25f79ddcbe289faa95e655100b952da)
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 
32 /*
33  * Loopback interface driver for protocol testing and timing.
34  */
35 
36 #include "opt_inet.h"
37 #include "opt_inet6.h"
38 #include "opt_rss.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_private.h>
54 #include <net/if_clone.h>
55 #include <net/if_types.h>
56 #include <net/netisr.h>
57 #include <net/route.h>
58 #include <net/bpf.h>
59 #include <net/vnet.h>
60 
61 #ifdef	INET
62 #include <netinet/in.h>
63 #include <netinet/in_var.h>
64 #endif
65 
66 #ifdef INET6
67 #ifndef INET
68 #include <netinet/in.h>
69 #endif
70 #include <netinet6/in6_var.h>
71 #include <netinet/ip6.h>
72 #endif
73 
74 #include <security/mac/mac_framework.h>
75 
76 #ifdef TINY_LOMTU
77 #define	LOMTU	(1024+512)
78 #elif defined(LARGE_LOMTU)
79 #define LOMTU	131072
80 #else
81 #define LOMTU	16384
82 #endif
83 
84 #define	LO_CSUM_FEATURES	(CSUM_IP | CSUM_IP_TCP | CSUM_IP_UDP | \
85 				    CSUM_IP_SCTP)
86 #define	LO_CSUM_FEATURES6	(CSUM_IP6_TCP | CSUM_IP6_UDP | CSUM_IP6_SCTP)
87 #define	LO_CSUM_SET		(CSUM_L3_CALC | CSUM_L3_VALID | \
88 				    CSUM_L4_CALC | CSUM_L4_VALID)
89 
90 static int	loioctl(struct ifnet *, u_long, caddr_t);
91 static int	looutput(struct ifnet *ifp, struct mbuf *m,
92 		    const struct sockaddr *dst, struct route *ro);
93 
94 VNET_DEFINE(struct ifnet *, loif);	/* Used externally */
95 VNET_DEFINE_STATIC(struct if_clone *, lo_cloner);
96 #define	V_lo_cloner		VNET(lo_cloner)
97 
98 static const char loname[] = "lo";
99 
100 static int
101 lo_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags)
102 {
103 	if (ifp->if_dunit == 0 && (flags & IFC_F_FORCE) == 0)
104 		return (EINVAL);
105 
106 #ifndef VIMAGE
107 	/* XXX: destroying lo0 will lead to panics. */
108 	KASSERT(V_loif != ifp, ("%s: destroying lo0", __func__));
109 #endif
110 
111 	bpfdetach(ifp);
112 	if_detach(ifp);
113 	if_free(ifp);
114 
115 	return (0);
116 }
117 
118 static int
119 lo_clone_create(struct if_clone *ifc, char *name, size_t len,
120     struct ifc_data *ifd, struct ifnet **ifpp)
121 {
122 	struct ifnet *ifp;
123 
124 	ifp = if_alloc(IFT_LOOP);
125 	if_initname(ifp, loname, ifd->unit);
126 	ifp->if_mtu = LOMTU;
127 	ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST;
128 	ifp->if_ioctl = loioctl;
129 	ifp->if_output = looutput;
130 	ifp->if_snd.ifq_maxlen = ifqmaxlen;
131 	ifp->if_capabilities = ifp->if_capenable =
132 	    IFCAP_HWCSUM | IFCAP_HWCSUM_IPV6 | IFCAP_LINKSTATE;
133 	ifp->if_hwassist = LO_CSUM_FEATURES | LO_CSUM_FEATURES6;
134 	if_attach(ifp);
135 	bpfattach(ifp, DLT_NULL, sizeof(u_int32_t));
136 	*ifpp = ifp;
137 
138 	return (0);
139 }
140 
141 static void
142 vnet_loif_init(const void *unused __unused)
143 {
144 	struct if_clone_addreq req = {
145 		.create_f = lo_clone_create,
146 		.destroy_f = lo_clone_destroy,
147 		.flags = IFC_F_AUTOUNIT,
148 	};
149 	V_lo_cloner = ifc_attach_cloner(loname, &req);
150 	struct ifc_data ifd = { .unit = 0 };
151 	ifc_create_ifp(loname, &ifd, &V_loif);
152 }
153 VNET_SYSINIT(vnet_loif_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
154     vnet_loif_init, NULL);
155 
156 #ifdef VIMAGE
157 static void
158 vnet_loif_uninit(const void *unused __unused)
159 {
160 
161 	ifc_detach_cloner(V_lo_cloner);
162 	V_loif = NULL;
163 }
164 VNET_SYSUNINIT(vnet_loif_uninit, SI_SUB_INIT_IF, SI_ORDER_SECOND,
165     vnet_loif_uninit, NULL);
166 #endif
167 
168 static int
169 loop_modevent(module_t mod, int type, void *data)
170 {
171 
172 	switch (type) {
173 	case MOD_LOAD:
174 		break;
175 
176 	case MOD_UNLOAD:
177 		printf("loop module unload - not possible for this module type\n");
178 		return (EINVAL);
179 
180 	default:
181 		return (EOPNOTSUPP);
182 	}
183 	return (0);
184 }
185 
186 static moduledata_t loop_mod = {
187 	"if_lo",
188 	loop_modevent,
189 	0
190 };
191 
192 DECLARE_MODULE(if_lo, loop_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY);
193 
194 static int
195 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
196     struct route *ro)
197 {
198 	u_int32_t af;
199 #ifdef MAC
200 	int error;
201 #endif
202 
203 	M_ASSERTPKTHDR(m); /* check if we have the packet header */
204 
205 #ifdef MAC
206 	error = mac_ifnet_check_transmit(ifp, m);
207 	if (error) {
208 		m_freem(m);
209 		return (error);
210 	}
211 #endif
212 
213 	if (ro != NULL && ro->ro_flags & (RT_REJECT|RT_BLACKHOLE)) {
214 		m_freem(m);
215 		return (ro->ro_flags & RT_BLACKHOLE ? 0 : EHOSTUNREACH);
216 	}
217 
218 	if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
219 	if_inc_counter(ifp, IFCOUNTER_OBYTES, m->m_pkthdr.len);
220 
221 #ifdef RSS
222 	M_HASHTYPE_CLEAR(m);
223 #endif
224 
225 	/* BPF writes need to be handled specially. */
226 	if (dst->sa_family == AF_UNSPEC || dst->sa_family == pseudo_AF_HDRCMPLT)
227 		bcopy(dst->sa_data, &af, sizeof(af));
228 	else
229 		af = RO_GET_FAMILY(ro, dst);
230 
231 	switch (af) {
232 	case AF_INET:
233 		if ((ifp->if_capenable & IFCAP_RXCSUM) == 0) {
234 			m->m_pkthdr.csum_flags &= ~LO_CSUM_SET;
235 		}
236 		break;
237 	case AF_INET6:
238 		if ((ifp->if_capenable & IFCAP_RXCSUM_IPV6) == 0) {
239 			m->m_pkthdr.csum_flags &= ~LO_CSUM_SET;
240 		}
241 		break;
242 	default:
243 		printf("looutput: af=%d unexpected\n", af);
244 		m_freem(m);
245 		return (EAFNOSUPPORT);
246 	}
247 	return (if_simloop(ifp, m, af, 0));
248 }
249 
250 /*
251  * if_simloop()
252  *
253  * This function is to support software emulation of hardware loopback,
254  * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't
255  * hear their own broadcasts, we create a copy of the packet that we
256  * would normally receive via a hardware loopback.
257  *
258  * This function expects the packet to include the media header of length hlen.
259  */
260 int
261 if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen)
262 {
263 	int isr;
264 	int32_t len;
265 
266 	M_ASSERTPKTHDR(m);
267 	m_tag_delete_nonpersistent(m);
268 	m->m_pkthdr.rcvif = ifp;
269 
270 #ifdef MAC
271 	mac_ifnet_create_mbuf(ifp, m);
272 #endif
273 
274 	/*
275 	 * Let BPF see incoming packet in the following manner:
276 	 *  - Emulated packet loopback for a simplex interface
277 	 *    (net/if_ethersubr.c)
278 	 *	-> passes it to ifp's BPF
279 	 *  - IPv4/v6 multicast packet loopback (netinet(6)/ip(6)_output.c)
280 	 *	-> not passes it to any BPF
281 	 *  - Normal packet loopback from myself to myself (net/if_loop.c)
282 	 *	-> passes to lo0's BPF (even in case of IPv6, where ifp!=lo0)
283 	 */
284 	if (hlen > 0) {
285 		if (bpf_peers_present(ifp->if_bpf)) {
286 			bpf_mtap(ifp->if_bpf, m);
287 		}
288 	} else {
289 		if (bpf_peers_present(V_loif->if_bpf)) {
290 			if ((m->m_flags & M_MCAST) == 0 || V_loif == ifp) {
291 				/* XXX beware sizeof(af) != 4 */
292 				u_int32_t af1 = af;
293 
294 				/*
295 				 * We need to prepend the address family.
296 				 */
297 				bpf_mtap2(V_loif->if_bpf, &af1, sizeof(af1), m);
298 			}
299 		}
300 	}
301 
302 	/* Strip away media header */
303 	if (hlen > 0) {
304 		m_adj(m, hlen);
305 #ifndef __NO_STRICT_ALIGNMENT
306 		/*
307 		 * Some archs do not like unaligned data, so
308 		 * we move data down in the first mbuf.
309 		 */
310 		if (mtod(m, vm_offset_t) & 3) {
311 			KASSERT(hlen >= 3, ("if_simloop: hlen too small"));
312 			bcopy(m->m_data,
313 			    (char *)(mtod(m, vm_offset_t)
314 				- (mtod(m, vm_offset_t) & 3)),
315 			    m->m_len);
316 			m->m_data -= (mtod(m,vm_offset_t) & 3);
317 		}
318 #endif
319 	}
320 
321 	/* Deliver to upper layer protocol */
322 	switch (af) {
323 #ifdef INET
324 	case AF_INET:
325 		isr = NETISR_IP;
326 		break;
327 #endif
328 #ifdef INET6
329 	case AF_INET6:
330 		m->m_flags |= M_LOOP;
331 		isr = NETISR_IPV6;
332 		break;
333 #endif
334 	default:
335 		printf("if_simloop: can't handle af=%d\n", af);
336 		m_freem(m);
337 		return (EAFNOSUPPORT);
338 	}
339 	len = m->m_pkthdr.len;
340 	if (netisr_queue(isr, m) == 0) {
341 		if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
342 		if_inc_counter(ifp, IFCOUNTER_IBYTES, len);
343 	} else {
344 		/* mbuf is free'd on failure. */
345 		if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
346 	}
347 	return (0);
348 }
349 
350 /*
351  * Process an ioctl request.
352  */
353 static int
354 loioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
355 {
356 	struct ifreq *ifr = (struct ifreq *)data;
357 	int error = 0, mask;
358 
359 	switch (cmd) {
360 	case SIOCSIFADDR:
361 		ifp->if_flags |= IFF_UP;
362 		ifp->if_drv_flags |= IFF_DRV_RUNNING;
363 		if_link_state_change(ifp, LINK_STATE_UP);
364 		/*
365 		 * Everything else is done at a higher level.
366 		 */
367 		break;
368 
369 	case SIOCADDMULTI:
370 	case SIOCDELMULTI:
371 		if (ifr == NULL) {
372 			error = EAFNOSUPPORT;		/* XXX */
373 			break;
374 		}
375 		switch (ifr->ifr_addr.sa_family) {
376 #ifdef INET
377 		case AF_INET:
378 			break;
379 #endif
380 #ifdef INET6
381 		case AF_INET6:
382 			break;
383 #endif
384 
385 		default:
386 			error = EAFNOSUPPORT;
387 			break;
388 		}
389 		break;
390 
391 	case SIOCSIFMTU:
392 		ifp->if_mtu = ifr->ifr_mtu;
393 		break;
394 
395 	case SIOCSIFFLAGS:
396 		if_link_state_change(ifp, (ifp->if_flags & IFF_UP) ?
397 		    LINK_STATE_UP: LINK_STATE_DOWN);
398 		break;
399 
400 	case SIOCSIFCAP:
401 		mask = ifp->if_capenable ^ ifr->ifr_reqcap;
402 		if ((mask & IFCAP_RXCSUM) != 0)
403 			ifp->if_capenable ^= IFCAP_RXCSUM;
404 		if ((mask & IFCAP_TXCSUM) != 0)
405 			ifp->if_capenable ^= IFCAP_TXCSUM;
406 		if ((mask & IFCAP_RXCSUM_IPV6) != 0)
407 			ifp->if_capenable ^= IFCAP_RXCSUM_IPV6;
408 		if ((mask & IFCAP_TXCSUM_IPV6) != 0)
409 			ifp->if_capenable ^= IFCAP_TXCSUM_IPV6;
410 		ifp->if_hwassist = 0;
411 		if (ifp->if_capenable & IFCAP_TXCSUM)
412 			ifp->if_hwassist = LO_CSUM_FEATURES;
413 		if (ifp->if_capenable & IFCAP_TXCSUM_IPV6)
414 			ifp->if_hwassist |= LO_CSUM_FEATURES6;
415 		break;
416 
417 	default:
418 		error = EINVAL;
419 	}
420 	return (error);
421 }
422