if_loop.c revision 97289 
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 * 3. All advertising materials mentioning features or use of this software
14 *    must display the following acknowledgement:
15 *	This product includes software developed by the University of
16 *	California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 *    may be used to endorse or promote products derived from this software
19 *    without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 *	@(#)if_loop.c	8.2 (Berkeley) 1/9/95
34 * $FreeBSD: head/sys/net/if_loop.c 97289 2002-05-25 20:17:04Z brooks $
35 */
36
37/*
38 * Loopback interface driver for protocol testing and timing.
39 */
40
41#include "opt_atalk.h"
42#include "opt_inet.h"
43#include "opt_inet6.h"
44#include "opt_ipx.h"
45
46#include <sys/param.h>
47#include <sys/systm.h>
48#include <sys/kernel.h>
49#include <sys/malloc.h>
50#include <sys/mbuf.h>
51#include <sys/module.h>
52#include <machine/bus.h>
53#include <sys/rman.h>
54#include <sys/socket.h>
55#include <sys/sockio.h>
56#include <sys/sysctl.h>
57
58#include <net/if.h>
59#include <net/if_types.h>
60#include <net/netisr.h>
61#include <net/route.h>
62#include <net/bpf.h>
63#include <net/bpfdesc.h>
64
65#ifdef	INET
66#include <netinet/in.h>
67#include <netinet/in_var.h>
68#endif
69
70#ifdef IPX
71#include <netipx/ipx.h>
72#include <netipx/ipx_if.h>
73#endif
74
75#ifdef INET6
76#ifndef INET
77#include <netinet/in.h>
78#endif
79#include <netinet6/in6_var.h>
80#include <netinet/ip6.h>
81#endif
82
83#ifdef NS
84#include <netns/ns.h>
85#include <netns/ns_if.h>
86#endif
87
88#ifdef NETATALK
89#include <netatalk/at.h>
90#include <netatalk/at_var.h>
91#endif
92
93#ifdef TINY_LOMTU
94#define	LOMTU	(1024+512)
95#elif defined(LARGE_LOMTU)
96#define LOMTU	131072
97#else
98#define LOMTU	16384
99#endif
100
101#define LONAME	"lo"
102
103struct lo_softc {
104	struct	ifnet sc_if;		/* network-visible interface */
105	LIST_ENTRY(lo_softc) sc_next;
106};
107
108int		loioctl(struct ifnet *, u_long, caddr_t);
109static void	lortrequest(int, struct rtentry *, struct rt_addrinfo *);
110int		looutput(struct ifnet *ifp, struct mbuf *m,
111		    struct sockaddr *dst, struct rtentry *rt);
112int		lo_clone_create(struct if_clone *, int);
113void		lo_clone_destroy(struct ifnet *);
114
115struct ifnet *loif = NULL;			/* Used externally */
116
117static MALLOC_DEFINE(M_LO, LONAME, "Loopback Interface");
118
119static LIST_HEAD(lo_list, lo_softc) lo_list;
120
121struct if_clone lo_cloner = IF_CLONE_INITIALIZER(LONAME,
122    lo_clone_create, lo_clone_destroy, 1, IF_MAXUNIT);
123
124void
125lo_clone_destroy(ifp)
126	struct ifnet *ifp;
127{
128	struct lo_softc *sc;
129
130	sc = ifp->if_softc;
131
132	/* XXX: destroying lo0 will lead to panics. */
133	KASSERT(loif != ifp, ("%s: destroying lo0", __func__));
134
135	bpfdetach(ifp);
136	if_detach(ifp);
137	LIST_REMOVE(sc, sc_next);
138	free(sc, M_LO);
139}
140
141int
142lo_clone_create(ifc, unit)
143	struct if_clone *ifc;
144	int unit;
145{
146	struct lo_softc *sc;
147
148	MALLOC(sc, struct lo_softc *, sizeof(*sc), M_LO, M_WAITOK | M_ZERO);
149
150	sc->sc_if.if_name = LONAME;
151	sc->sc_if.if_unit = unit;
152	sc->sc_if.if_mtu = LOMTU;
153	sc->sc_if.if_flags = IFF_LOOPBACK | IFF_MULTICAST;
154	sc->sc_if.if_ioctl = loioctl;
155	sc->sc_if.if_output = looutput;
156	sc->sc_if.if_type = IFT_LOOP;
157	sc->sc_if.if_snd.ifq_maxlen = ifqmaxlen;
158	sc->sc_if.if_softc = sc;
159	if_attach(&sc->sc_if);
160	bpfattach(&sc->sc_if, DLT_NULL, sizeof(u_int));
161	LIST_INSERT_HEAD(&lo_list, sc, sc_next);
162	if (loif == NULL)
163		loif = &sc->sc_if;
164
165	return (0);
166}
167
168static int
169loop_modevent(module_t mod, int type, void *data)
170{
171	switch (type) {
172	case MOD_LOAD:
173		LIST_INIT(&lo_list);
174		if_clone_attach(&lo_cloner);
175		break;
176	case MOD_UNLOAD:
177		printf("loop module unload - not possible for this module type\n");
178		return EINVAL;
179	}
180	return 0;
181}
182
183static moduledata_t loop_mod = {
184	"loop",
185	loop_modevent,
186	0
187};
188
189DECLARE_MODULE(loop, loop_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
190
191int
192looutput(ifp, m, dst, rt)
193	struct ifnet *ifp;
194	register struct mbuf *m;
195	struct sockaddr *dst;
196	register struct rtentry *rt;
197{
198	if ((m->m_flags & M_PKTHDR) == 0)
199		panic("looutput no HDR");
200
201	if (rt && rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) {
202		m_freem(m);
203		return (rt->rt_flags & RTF_BLACKHOLE ? 0 :
204		        rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
205	}
206	/*
207	 * KAME requires that the packet to be contiguous on the
208	 * mbuf.  We need to make that sure.
209	 * this kind of code should be avoided.
210	 * XXX: fails to join if interface MTU > MCLBYTES.  jumbogram?
211	 */
212	if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
213		struct mbuf *n;
214
215		MGETHDR(n, M_DONTWAIT, MT_HEADER);
216		if (!n)
217			goto contiguousfail;
218		MCLGET(n, M_DONTWAIT);
219		if (! (n->m_flags & M_EXT)) {
220			m_freem(n);
221			goto contiguousfail;
222		}
223
224		m_copydata(m, 0, m->m_pkthdr.len, mtod(n, caddr_t));
225		n->m_pkthdr = m->m_pkthdr;
226		n->m_len = m->m_pkthdr.len;
227		n->m_pkthdr.aux = m->m_pkthdr.aux;
228		m->m_pkthdr.aux = (struct mbuf *)NULL;
229		m_freem(m);
230		m = n;
231	}
232	if (0) {
233contiguousfail:
234		printf("looutput: mbuf allocation failed\n");
235	}
236
237	ifp->if_opackets++;
238	ifp->if_obytes += m->m_pkthdr.len;
239#if 1	/* XXX */
240	switch (dst->sa_family) {
241	case AF_INET:
242	case AF_INET6:
243	case AF_IPX:
244	case AF_NS:
245	case AF_APPLETALK:
246		break;
247	default:
248		printf("looutput: af=%d unexpected\n", dst->sa_family);
249		m_freem(m);
250		return (EAFNOSUPPORT);
251	}
252#endif
253	return(if_simloop(ifp, m, dst->sa_family, 0));
254}
255
256/*
257 * if_simloop()
258 *
259 * This function is to support software emulation of hardware loopback,
260 * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't
261 * hear their own broadcasts, we create a copy of the packet that we
262 * would normally receive via a hardware loopback.
263 *
264 * This function expects the packet to include the media header of length hlen.
265 */
266
267int
268if_simloop(ifp, m, af, hlen)
269	struct ifnet *ifp;
270	struct mbuf *m;
271	int af;
272	int hlen;
273{
274	int isr;
275	struct ifqueue *inq = 0;
276
277	KASSERT((m->m_flags & M_PKTHDR) != 0, ("if_simloop: no HDR"));
278	m->m_pkthdr.rcvif = ifp;
279
280	/* BPF write needs to be handled specially */
281	if (af == AF_UNSPEC) {
282		KASSERT(m->m_len >= sizeof(int), ("if_simloop: m_len"));
283		af = *(mtod(m, int *));
284		m->m_len -= sizeof(int);
285		m->m_pkthdr.len -= sizeof(int);
286		m->m_data += sizeof(int);
287	}
288
289	/* Let BPF see incoming packet */
290	if (ifp->if_bpf) {
291		struct mbuf m0, *n = m;
292
293		if (ifp->if_bpf->bif_dlt == DLT_NULL) {
294			/*
295			 * We need to prepend the address family as
296			 * a four byte field.  Cons up a dummy header
297			 * to pacify bpf.  This is safe because bpf
298			 * will only read from the mbuf (i.e., it won't
299			 * try to free it or keep a pointer a to it).
300			 */
301			m0.m_next = m;
302			m0.m_len = 4;
303			m0.m_data = (char *)&af;
304			n = &m0;
305		}
306		bpf_mtap(ifp, n);
307	}
308
309	/* Strip away media header */
310	if (hlen > 0) {
311		m_adj(m, hlen);
312#if defined(__alpha__) || defined(__ia64__) || defined(__sparc64__)
313		/* The alpha doesn't like unaligned data.
314		 * We move data down in the first mbuf */
315		if (mtod(m, vm_offset_t) & 3) {
316			KASSERT(hlen >= 3, ("if_simloop: hlen too small"));
317			bcopy(m->m_data,
318			    (char *)(mtod(m, vm_offset_t)
319				- (mtod(m, vm_offset_t) & 3)),
320			    m->m_len);
321			mtod(m,vm_offset_t) -= (mtod(m, vm_offset_t) & 3);
322		}
323#endif
324	}
325
326	/* Deliver to upper layer protocol */
327	switch (af) {
328#ifdef INET
329	case AF_INET:
330		inq = &ipintrq;
331		isr = NETISR_IP;
332		break;
333#endif
334#ifdef INET6
335	case AF_INET6:
336		m->m_flags |= M_LOOP;
337		inq = &ip6intrq;
338		isr = NETISR_IPV6;
339		break;
340#endif
341#ifdef IPX
342	case AF_IPX:
343		inq = &ipxintrq;
344		isr = NETISR_IPX;
345		break;
346#endif
347#ifdef NS
348	case AF_NS:
349		inq = &nsintrq;
350		isr = NETISR_NS;
351		break;
352#endif
353#ifdef NETATALK
354	case AF_APPLETALK:
355	        inq = &atintrq2;
356		isr = NETISR_ATALK;
357		break;
358#endif
359	default:
360		printf("if_simloop: can't handle af=%d\n", af);
361		m_freem(m);
362		return (EAFNOSUPPORT);
363	}
364	ifp->if_ipackets++;
365	ifp->if_ibytes += m->m_pkthdr.len;
366	(void) IF_HANDOFF(inq, m, NULL);
367	schednetisr(isr);
368	return (0);
369}
370
371/* ARGSUSED */
372static void
373lortrequest(cmd, rt, info)
374	int cmd;
375	struct rtentry *rt;
376	struct rt_addrinfo *info;
377{
378	if (rt) {
379		rt->rt_rmx.rmx_mtu = rt->rt_ifp->if_mtu; /* for ISO */
380		/*
381		 * For optimal performance, the send and receive buffers
382		 * should be at least twice the MTU plus a little more for
383		 * overhead.
384		 */
385		rt->rt_rmx.rmx_recvpipe =
386			rt->rt_rmx.rmx_sendpipe = 3 * LOMTU;
387	}
388}
389
390/*
391 * Process an ioctl request.
392 */
393/* ARGSUSED */
394int
395loioctl(ifp, cmd, data)
396	register struct ifnet *ifp;
397	u_long cmd;
398	caddr_t data;
399{
400	register struct ifaddr *ifa;
401	register struct ifreq *ifr = (struct ifreq *)data;
402	register int error = 0;
403
404	switch (cmd) {
405
406	case SIOCSIFADDR:
407		ifp->if_flags |= IFF_UP | IFF_RUNNING;
408		ifa = (struct ifaddr *)data;
409		ifa->ifa_rtrequest = lortrequest;
410		/*
411		 * Everything else is done at a higher level.
412		 */
413		break;
414
415	case SIOCADDMULTI:
416	case SIOCDELMULTI:
417		if (ifr == 0) {
418			error = EAFNOSUPPORT;		/* XXX */
419			break;
420		}
421		switch (ifr->ifr_addr.sa_family) {
422
423#ifdef INET
424		case AF_INET:
425			break;
426#endif
427#ifdef INET6
428		case AF_INET6:
429			break;
430#endif
431
432		default:
433			error = EAFNOSUPPORT;
434			break;
435		}
436		break;
437
438	case SIOCSIFMTU:
439		ifp->if_mtu = ifr->ifr_mtu;
440		break;
441
442	case SIOCSIFFLAGS:
443		break;
444
445	default:
446		error = EINVAL;
447	}
448	return (error);
449}
450