xref: /netbsd-src/sys/net/if_loop.c (revision 6cd39ddb8550f6fa1bff3fed32053d7f19fd0453) 
1 /*	$NetBSD: if_loop.c,v 1.83 2015/08/24 22:21:26 pooka Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * 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 project 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 PROJECT 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 PROJECT 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  * Copyright (c) 1982, 1986, 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * Redistribution and use in source and binary forms, with or without
37  * modification, are permitted provided that the following conditions
38  * are met:
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. Neither the name of the University nor the names of its contributors
45  *    may be used to endorse or promote products derived from this software
46  *    without specific prior written permission.
47  *
48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58  * SUCH DAMAGE.
59  *
60  *	@(#)if_loop.c	8.2 (Berkeley) 1/9/95
61  */
62 
63 /*
64  * Loopback interface driver for protocol testing and timing.
65  */
66 
67 #include <sys/cdefs.h>
68 __KERNEL_RCSID(0, "$NetBSD: if_loop.c,v 1.83 2015/08/24 22:21:26 pooka Exp $");
69 
70 #ifdef _KERNEL_OPT
71 #include "opt_inet.h"
72 #include "opt_atalk.h"
73 #include "opt_mbuftrace.h"
74 #include "opt_mpls.h"
75 #include "opt_net_mpsafe.h"
76 #endif
77 
78 #include <sys/param.h>
79 #include <sys/systm.h>
80 #include <sys/kernel.h>
81 #include <sys/mbuf.h>
82 #include <sys/socket.h>
83 #include <sys/errno.h>
84 #include <sys/ioctl.h>
85 #include <sys/time.h>
86 
87 #include <sys/cpu.h>
88 
89 #include <net/if.h>
90 #include <net/if_types.h>
91 #include <net/netisr.h>
92 #include <net/route.h>
93 
94 #ifdef	INET
95 #include <netinet/in.h>
96 #include <netinet/in_systm.h>
97 #include <netinet/in_var.h>
98 #include <netinet/in_offload.h>
99 #include <netinet/ip.h>
100 #endif
101 
102 #ifdef INET6
103 #ifndef INET
104 #include <netinet/in.h>
105 #endif
106 #include <netinet6/in6_var.h>
107 #include <netinet6/in6_offload.h>
108 #include <netinet/ip6.h>
109 #endif
110 
111 #ifdef MPLS
112 #include <netmpls/mpls.h>
113 #include <netmpls/mpls_var.h>
114 #endif
115 
116 #ifdef NETATALK
117 #include <netatalk/at.h>
118 #include <netatalk/at_var.h>
119 #endif
120 
121 #include <net/bpf.h>
122 
123 #if defined(LARGE_LOMTU)
124 #define LOMTU	(131072 +  MHLEN + MLEN)
125 #define LOMTU_MAX LOMTU
126 #else
127 #define	LOMTU	(32768 +  MHLEN + MLEN)
128 #define	LOMTU_MAX	(65536 +  MHLEN + MLEN)
129 #endif
130 
131 #ifdef ALTQ
132 static void	lostart(struct ifnet *);
133 #endif
134 
135 static int	loop_clone_create(struct if_clone *, int);
136 static int	loop_clone_destroy(struct ifnet *);
137 
138 static struct if_clone loop_cloner =
139     IF_CLONE_INITIALIZER("lo", loop_clone_create, loop_clone_destroy);
140 
141 void
142 loopattach(int n)
143 {
144 
145 	(void)loop_clone_create(&loop_cloner, 0);	/* lo0 always exists */
146 	if_clone_attach(&loop_cloner);
147 }
148 
149 static int
150 loop_clone_create(struct if_clone *ifc, int unit)
151 {
152 	struct ifnet *ifp;
153 
154 	ifp = if_alloc(IFT_LOOP);
155 
156 	if_initname(ifp, ifc->ifc_name, unit);
157 
158 	ifp->if_mtu = LOMTU;
159 	ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST | IFF_RUNNING;
160 	ifp->if_ioctl = loioctl;
161 	ifp->if_output = looutput;
162 #ifdef ALTQ
163 	ifp->if_start = lostart;
164 #endif
165 	ifp->if_type = IFT_LOOP;
166 	ifp->if_hdrlen = 0;
167 	ifp->if_addrlen = 0;
168 	ifp->if_dlt = DLT_NULL;
169 	IFQ_SET_READY(&ifp->if_snd);
170 	if (unit == 0)
171 		lo0ifp = ifp;
172 	if_attach(ifp);
173 	if_alloc_sadl(ifp);
174 	bpf_attach(ifp, DLT_NULL, sizeof(u_int));
175 #ifdef MBUFTRACE
176 	ifp->if_mowner = malloc(sizeof(struct mowner), M_DEVBUF,
177 	    M_WAITOK | M_ZERO);
178 	strlcpy(ifp->if_mowner->mo_name, ifp->if_xname,
179 	    sizeof(ifp->if_mowner->mo_name));
180 	MOWNER_ATTACH(ifp->if_mowner);
181 #endif
182 
183 	return (0);
184 }
185 
186 static int
187 loop_clone_destroy(struct ifnet *ifp)
188 {
189 
190 	if (ifp == lo0ifp)
191 		return (EPERM);
192 
193 #ifdef MBUFTRACE
194 	MOWNER_DETACH(ifp->if_mowner);
195 	free(ifp->if_mowner, M_DEVBUF);
196 #endif
197 
198 	bpf_detach(ifp);
199 	if_detach(ifp);
200 
201 	if_free(ifp);
202 
203 	return (0);
204 }
205 
206 int
207 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
208     struct rtentry *rt)
209 {
210 	pktqueue_t *pktq = NULL;
211 	struct ifqueue *ifq = NULL;
212 	int s, isr = -1;
213 	int csum_flags;
214 	size_t pktlen;
215 
216 	MCLAIM(m, ifp->if_mowner);
217 #ifndef NET_MPSAFE
218 	KASSERT(KERNEL_LOCKED_P());
219 #endif
220 
221 	if ((m->m_flags & M_PKTHDR) == 0)
222 		panic("looutput: no header mbuf");
223 	if (ifp->if_flags & IFF_LOOPBACK)
224 		bpf_mtap_af(ifp, dst->sa_family, m);
225 	m->m_pkthdr.rcvif = ifp;
226 
227 	if (rt && rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) {
228 		m_freem(m);
229 		return (rt->rt_flags & RTF_BLACKHOLE ? 0 :
230 			rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
231 	}
232 
233 	pktlen = m->m_pkthdr.len;
234 	ifp->if_opackets++;
235 	ifp->if_obytes += pktlen;
236 
237 #ifdef ALTQ
238 	/*
239 	 * ALTQ on the loopback interface is just for debugging.  It's
240 	 * used only for loopback interfaces, not for a simplex interface.
241 	 */
242 	if ((ALTQ_IS_ENABLED(&ifp->if_snd) || TBR_IS_ENABLED(&ifp->if_snd)) &&
243 	    ifp->if_start == lostart) {
244 		struct altq_pktattr pktattr;
245 		int error;
246 
247 		/*
248 		 * If the queueing discipline needs packet classification,
249 		 * do it before prepending the link headers.
250 		 */
251 		IFQ_CLASSIFY(&ifp->if_snd, m, dst->sa_family, &pktattr);
252 
253 		M_PREPEND(m, sizeof(uint32_t), M_DONTWAIT);
254 		if (m == NULL)
255 			return (ENOBUFS);
256 		*(mtod(m, uint32_t *)) = dst->sa_family;
257 
258 		s = splnet();
259 		IFQ_ENQUEUE(&ifp->if_snd, m, &pktattr, error);
260 		(*ifp->if_start)(ifp);
261 		splx(s);
262 		return (error);
263 	}
264 #endif /* ALTQ */
265 
266 	m_tag_delete_nonpersistent(m);
267 
268 #ifdef MPLS
269 	if (rt != NULL && rt_gettag(rt) != NULL &&
270 	    rt_gettag(rt)->sa_family == AF_MPLS &&
271 	    (m->m_flags & (M_MCAST | M_BCAST)) == 0) {
272 		union mpls_shim msh;
273 		msh.s_addr = MPLS_GETSADDR(rt);
274 		if (msh.shim.label != MPLS_LABEL_IMPLNULL) {
275 			ifq = &mplsintrq;
276 			isr = NETISR_MPLS;
277 		}
278 	}
279 	if (isr != NETISR_MPLS)
280 #endif
281 	switch (dst->sa_family) {
282 
283 #ifdef INET
284 	case AF_INET:
285 		csum_flags = m->m_pkthdr.csum_flags;
286 		KASSERT((csum_flags & ~(M_CSUM_IPv4|M_CSUM_UDPv4)) == 0);
287 		if (csum_flags != 0 && IN_LOOPBACK_NEED_CHECKSUM(csum_flags)) {
288 			ip_undefer_csum(m, 0, csum_flags);
289 		}
290 		m->m_pkthdr.csum_flags = 0;
291 		pktq = ip_pktq;
292 		break;
293 #endif
294 #ifdef INET6
295 	case AF_INET6:
296 		csum_flags = m->m_pkthdr.csum_flags;
297 		KASSERT((csum_flags & ~M_CSUM_UDPv6) == 0);
298 		if (csum_flags != 0 &&
299 		    IN6_LOOPBACK_NEED_CHECKSUM(csum_flags)) {
300 			ip6_undefer_csum(m, 0, csum_flags);
301 		}
302 		m->m_pkthdr.csum_flags = 0;
303 		m->m_flags |= M_LOOP;
304 		pktq = ip6_pktq;
305 		break;
306 #endif
307 #ifdef NETATALK
308 	case AF_APPLETALK:
309 	        ifq = &atintrq2;
310 		isr = NETISR_ATALK;
311 		break;
312 #endif
313 	default:
314 		printf("%s: can't handle af%d\n", ifp->if_xname,
315 		    dst->sa_family);
316 		m_freem(m);
317 		return (EAFNOSUPPORT);
318 	}
319 
320 	s = splnet();
321 	if (__predict_true(pktq)) {
322 		int error = 0;
323 
324 		if (__predict_true(pktq_enqueue(pktq, m, 0))) {
325 			ifp->if_ipackets++;
326 			ifp->if_ibytes += pktlen;
327 		} else {
328 			m_freem(m);
329 			error = ENOBUFS;
330 		}
331 		splx(s);
332 		return error;
333 	}
334 	if (IF_QFULL(ifq)) {
335 		IF_DROP(ifq);
336 		m_freem(m);
337 		splx(s);
338 		return (ENOBUFS);
339 	}
340 	IF_ENQUEUE(ifq, m);
341 	schednetisr(isr);
342 	ifp->if_ipackets++;
343 	ifp->if_ibytes += m->m_pkthdr.len;
344 	splx(s);
345 	return (0);
346 }
347 
348 #ifdef ALTQ
349 static void
350 lostart(struct ifnet *ifp)
351 {
352 	for (;;) {
353 		pktqueue_t *pktq = NULL;
354 		struct ifqueue *ifq = NULL;
355 		struct mbuf *m;
356 		size_t pktlen;
357 		uint32_t af;
358 		int s, isr = 0;
359 
360 		IFQ_DEQUEUE(&ifp->if_snd, m);
361 		if (m == NULL)
362 			return;
363 
364 		af = *(mtod(m, uint32_t *));
365 		m_adj(m, sizeof(uint32_t));
366 
367 		switch (af) {
368 #ifdef INET
369 		case AF_INET:
370 			pktq = ip_pktq;
371 			break;
372 #endif
373 #ifdef INET6
374 		case AF_INET6:
375 			m->m_flags |= M_LOOP;
376 			pktq = ip6_pktq;
377 			break;
378 #endif
379 #ifdef NETATALK
380 		case AF_APPLETALK:
381 			ifq = &atintrq2;
382 			isr = NETISR_ATALK;
383 			break;
384 #endif
385 		default:
386 			printf("%s: can't handle af%d\n", ifp->if_xname, af);
387 			m_freem(m);
388 			return;
389 		}
390 		pktlen = m->m_pkthdr.len;
391 
392 		s = splnet();
393 		if (__predict_true(pktq)) {
394 			if (__predict_false(pktq_enqueue(pktq, m, 0))) {
395 				m_freem(m);
396 				splx(s);
397 				return;
398 			}
399 			ifp->if_ipackets++;
400 			ifp->if_ibytes += pktlen;
401 			splx(s);
402 			continue;
403 		}
404 		if (IF_QFULL(ifq)) {
405 			IF_DROP(ifq);
406 			splx(s);
407 			m_freem(m);
408 			return;
409 		}
410 		IF_ENQUEUE(ifq, m);
411 		schednetisr(isr);
412 		ifp->if_ipackets++;
413 		ifp->if_ibytes += pktlen;
414 		splx(s);
415 	}
416 }
417 #endif /* ALTQ */
418 
419 /* ARGSUSED */
420 void
421 lortrequest(int cmd, struct rtentry *rt,
422     const struct rt_addrinfo *info)
423 {
424 
425 	if (rt)
426 		rt->rt_rmx.rmx_mtu = lo0ifp->if_mtu;
427 }
428 
429 /*
430  * Process an ioctl request.
431  */
432 /* ARGSUSED */
433 int
434 loioctl(struct ifnet *ifp, u_long cmd, void *data)
435 {
436 	struct ifaddr *ifa;
437 	struct ifreq *ifr = data;
438 	int error = 0;
439 
440 	switch (cmd) {
441 
442 	case SIOCINITIFADDR:
443 		ifp->if_flags |= IFF_UP;
444 		ifa = (struct ifaddr *)data;
445 		if (ifa != NULL)
446 			ifa->ifa_rtrequest = lortrequest;
447 		/*
448 		 * Everything else is done at a higher level.
449 		 */
450 		break;
451 
452 	case SIOCSIFMTU:
453 		if ((unsigned)ifr->ifr_mtu > LOMTU_MAX)
454 			error = EINVAL;
455 		else if ((error = ifioctl_common(ifp, cmd, data)) == ENETRESET){
456 			error = 0;
457 		}
458 		break;
459 
460 	case SIOCADDMULTI:
461 	case SIOCDELMULTI:
462 		if (ifr == NULL) {
463 			error = EAFNOSUPPORT;		/* XXX */
464 			break;
465 		}
466 		switch (ifreq_getaddr(cmd, ifr)->sa_family) {
467 
468 #ifdef INET
469 		case AF_INET:
470 			break;
471 #endif
472 #ifdef INET6
473 		case AF_INET6:
474 			break;
475 #endif
476 
477 		default:
478 			error = EAFNOSUPPORT;
479 			break;
480 		}
481 		break;
482 
483 	default:
484 		error = ifioctl_common(ifp, cmd, data);
485 	}
486 	return (error);
487 }
488