1/*	$NetBSD: svc_vc.c,v 1.7 2000/08/03 00:01:53 fvdl Exp $	*/
2
3/*-
4 * Copyright (c) 2009, Sun Microsystems, Inc.
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 are met:
9 * - Redistributions of source code must retain the above copyright notice,
10 *   this list of conditions and the following disclaimer.
11 * - Redistributions in binary form must reproduce the above copyright notice,
12 *   this list of conditions and the following disclaimer in the documentation
13 *   and/or other materials provided with the distribution.
14 * - Neither the name of Sun Microsystems, Inc. nor the names of its
15 *   contributors may be used to endorse or promote products derived
16 *   from this software without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
19 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
22 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28 * POSSIBILITY OF SUCH DAMAGE.
29 */
30
31#if defined(LIBC_SCCS) && !defined(lint)
32static char *sccsid2 = "@(#)svc_tcp.c 1.21 87/08/11 Copyr 1984 Sun Micro";
33static char *sccsid = "@(#)svc_tcp.c	2.2 88/08/01 4.0 RPCSRC";
34#endif
35#include <sys/cdefs.h>
36__FBSDID("$FreeBSD$");
37
38/*
39 * svc_vc.c, Server side for Connection Oriented based RPC.
40 *
41 * Actually implements two flavors of transporter -
42 * a tcp rendezvouser (a listner and connection establisher)
43 * and a record/tcp stream.
44 */
45
46#include <sys/param.h>
47#include <sys/limits.h>
48#include <sys/lock.h>
49#include <sys/kernel.h>
50#include <sys/malloc.h>
51#include <sys/mbuf.h>
52#include <sys/mutex.h>
53#include <sys/proc.h>
54#include <sys/protosw.h>
55#include <sys/queue.h>
56#include <sys/socket.h>
57#include <sys/socketvar.h>
58#include <sys/sx.h>
59#include <sys/systm.h>
60#include <sys/uio.h>
61
62#include <net/vnet.h>
63
64#include <netinet/tcp.h>
65
66#include <rpc/rpc.h>
67
68#include <rpc/krpc.h>
69#include <rpc/rpc_com.h>
70
71#include <security/mac/mac_framework.h>
72
73static bool_t svc_vc_rendezvous_recv(SVCXPRT *, struct rpc_msg *,
74    struct sockaddr **, struct mbuf **);
75static enum xprt_stat svc_vc_rendezvous_stat(SVCXPRT *);
76static void svc_vc_rendezvous_destroy(SVCXPRT *);
77static bool_t svc_vc_null(void);
78static void svc_vc_destroy(SVCXPRT *);
79static enum xprt_stat svc_vc_stat(SVCXPRT *);
80static bool_t svc_vc_ack(SVCXPRT *, uint32_t *);
81static bool_t svc_vc_recv(SVCXPRT *, struct rpc_msg *,
82    struct sockaddr **, struct mbuf **);
83static bool_t svc_vc_reply(SVCXPRT *, struct rpc_msg *,
84    struct sockaddr *, struct mbuf *, uint32_t *seq);
85static bool_t svc_vc_control(SVCXPRT *xprt, const u_int rq, void *in);
86static bool_t svc_vc_rendezvous_control (SVCXPRT *xprt, const u_int rq,
87    void *in);
88static void svc_vc_backchannel_destroy(SVCXPRT *);
89static enum xprt_stat svc_vc_backchannel_stat(SVCXPRT *);
90static bool_t svc_vc_backchannel_recv(SVCXPRT *, struct rpc_msg *,
91    struct sockaddr **, struct mbuf **);
92static bool_t svc_vc_backchannel_reply(SVCXPRT *, struct rpc_msg *,
93    struct sockaddr *, struct mbuf *, uint32_t *);
94static bool_t svc_vc_backchannel_control(SVCXPRT *xprt, const u_int rq,
95    void *in);
96static SVCXPRT *svc_vc_create_conn(SVCPOOL *pool, struct socket *so,
97    struct sockaddr *raddr);
98static int svc_vc_accept(struct socket *head, struct socket **sop);
99static int svc_vc_soupcall(struct socket *so, void *arg, int waitflag);
100
101static struct xp_ops svc_vc_rendezvous_ops = {
102	.xp_recv =	svc_vc_rendezvous_recv,
103	.xp_stat =	svc_vc_rendezvous_stat,
104	.xp_reply =	(bool_t (*)(SVCXPRT *, struct rpc_msg *,
105		struct sockaddr *, struct mbuf *, uint32_t *))svc_vc_null,
106	.xp_destroy =	svc_vc_rendezvous_destroy,
107	.xp_control =	svc_vc_rendezvous_control
108};
109
110static struct xp_ops svc_vc_ops = {
111	.xp_recv =	svc_vc_recv,
112	.xp_stat =	svc_vc_stat,
113	.xp_ack =	svc_vc_ack,
114	.xp_reply =	svc_vc_reply,
115	.xp_destroy =	svc_vc_destroy,
116	.xp_control =	svc_vc_control
117};
118
119static struct xp_ops svc_vc_backchannel_ops = {
120	.xp_recv =	svc_vc_backchannel_recv,
121	.xp_stat =	svc_vc_backchannel_stat,
122	.xp_reply =	svc_vc_backchannel_reply,
123	.xp_destroy =	svc_vc_backchannel_destroy,
124	.xp_control =	svc_vc_backchannel_control
125};
126
127/*
128 * Usage:
129 *	xprt = svc_vc_create(sock, send_buf_size, recv_buf_size);
130 *
131 * Creates, registers, and returns a (rpc) tcp based transporter.
132 * Once *xprt is initialized, it is registered as a transporter
133 * see (svc.h, xprt_register).  This routine returns
134 * a NULL if a problem occurred.
135 *
136 * The filedescriptor passed in is expected to refer to a bound, but
137 * not yet connected socket.
138 *
139 * Since streams do buffered io similar to stdio, the caller can specify
140 * how big the send and receive buffers are via the second and third parms;
141 * 0 => use the system default.
142 */
143SVCXPRT *
144svc_vc_create(SVCPOOL *pool, struct socket *so, size_t sendsize,
145    size_t recvsize)
146{
147	SVCXPRT *xprt;
148	struct sockaddr* sa;
149	int error;
150
151	SOCK_LOCK(so);
152	if (so->so_state & (SS_ISCONNECTED|SS_ISDISCONNECTED)) {
153		SOCK_UNLOCK(so);
154		error = so->so_proto->pr_usrreqs->pru_peeraddr(so, &sa);
155		if (error)
156			return (NULL);
157		xprt = svc_vc_create_conn(pool, so, sa);
158		free(sa, M_SONAME);
159		return (xprt);
160	}
161	SOCK_UNLOCK(so);
162
163	xprt = svc_xprt_alloc();
164	sx_init(&xprt->xp_lock, "xprt->xp_lock");
165	xprt->xp_pool = pool;
166	xprt->xp_socket = so;
167	xprt->xp_p1 = NULL;
168	xprt->xp_p2 = NULL;
169	xprt->xp_ops = &svc_vc_rendezvous_ops;
170
171	error = so->so_proto->pr_usrreqs->pru_sockaddr(so, &sa);
172	if (error) {
173		goto cleanup_svc_vc_create;
174	}
175
176	memcpy(&xprt->xp_ltaddr, sa, sa->sa_len);
177	free(sa, M_SONAME);
178
179	xprt_register(xprt);
180
181	solisten(so, SOMAXCONN, curthread);
182
183	SOCKBUF_LOCK(&so->so_rcv);
184	xprt->xp_upcallset = 1;
185	soupcall_set(so, SO_RCV, svc_vc_soupcall, xprt);
186	SOCKBUF_UNLOCK(&so->so_rcv);
187
188	return (xprt);
189cleanup_svc_vc_create:
190	if (xprt) {
191		sx_destroy(&xprt->xp_lock);
192		svc_xprt_free(xprt);
193	}
194	return (NULL);
195}
196
197/*
198 * Create a new transport for a socket optained via soaccept().
199 */
200SVCXPRT *
201svc_vc_create_conn(SVCPOOL *pool, struct socket *so, struct sockaddr *raddr)
202{
203	SVCXPRT *xprt = NULL;
204	struct cf_conn *cd = NULL;
205	struct sockaddr* sa = NULL;
206	struct sockopt opt;
207	int one = 1;
208	int error;
209
210	bzero(&opt, sizeof(struct sockopt));
211	opt.sopt_dir = SOPT_SET;
212	opt.sopt_level = SOL_SOCKET;
213	opt.sopt_name = SO_KEEPALIVE;
214	opt.sopt_val = &one;
215	opt.sopt_valsize = sizeof(one);
216	error = sosetopt(so, &opt);
217	if (error) {
218		return (NULL);
219	}
220
221	if (so->so_proto->pr_protocol == IPPROTO_TCP) {
222		bzero(&opt, sizeof(struct sockopt));
223		opt.sopt_dir = SOPT_SET;
224		opt.sopt_level = IPPROTO_TCP;
225		opt.sopt_name = TCP_NODELAY;
226		opt.sopt_val = &one;
227		opt.sopt_valsize = sizeof(one);
228		error = sosetopt(so, &opt);
229		if (error) {
230			return (NULL);
231		}
232	}
233
234	cd = mem_alloc(sizeof(*cd));
235	cd->strm_stat = XPRT_IDLE;
236
237	xprt = svc_xprt_alloc();
238	sx_init(&xprt->xp_lock, "xprt->xp_lock");
239	xprt->xp_pool = pool;
240	xprt->xp_socket = so;
241	xprt->xp_p1 = cd;
242	xprt->xp_p2 = NULL;
243	xprt->xp_ops = &svc_vc_ops;
244
245	/*
246	 * See http://www.connectathon.org/talks96/nfstcp.pdf - client
247	 * has a 5 minute timer, server has a 6 minute timer.
248	 */
249	xprt->xp_idletimeout = 6 * 60;
250
251	memcpy(&xprt->xp_rtaddr, raddr, raddr->sa_len);
252
253	error = so->so_proto->pr_usrreqs->pru_sockaddr(so, &sa);
254	if (error)
255		goto cleanup_svc_vc_create;
256
257	memcpy(&xprt->xp_ltaddr, sa, sa->sa_len);
258	free(sa, M_SONAME);
259
260	xprt_register(xprt);
261
262	SOCKBUF_LOCK(&so->so_rcv);
263	xprt->xp_upcallset = 1;
264	soupcall_set(so, SO_RCV, svc_vc_soupcall, xprt);
265	SOCKBUF_UNLOCK(&so->so_rcv);
266
267	/*
268	 * Throw the transport into the active list in case it already
269	 * has some data buffered.
270	 */
271	sx_xlock(&xprt->xp_lock);
272	xprt_active(xprt);
273	sx_xunlock(&xprt->xp_lock);
274
275	return (xprt);
276cleanup_svc_vc_create:
277	if (xprt) {
278		sx_destroy(&xprt->xp_lock);
279		svc_xprt_free(xprt);
280	}
281	if (cd)
282		mem_free(cd, sizeof(*cd));
283	return (NULL);
284}
285
286/*
287 * Create a new transport for a backchannel on a clnt_vc socket.
288 */
289SVCXPRT *
290svc_vc_create_backchannel(SVCPOOL *pool)
291{
292	SVCXPRT *xprt = NULL;
293	struct cf_conn *cd = NULL;
294
295	cd = mem_alloc(sizeof(*cd));
296	cd->strm_stat = XPRT_IDLE;
297
298	xprt = svc_xprt_alloc();
299	sx_init(&xprt->xp_lock, "xprt->xp_lock");
300	xprt->xp_pool = pool;
301	xprt->xp_socket = NULL;
302	xprt->xp_p1 = cd;
303	xprt->xp_p2 = NULL;
304	xprt->xp_ops = &svc_vc_backchannel_ops;
305	return (xprt);
306}
307
308/*
309 * This does all of the accept except the final call to soaccept. The
310 * caller will call soaccept after dropping its locks (soaccept may
311 * call malloc).
312 */
313int
314svc_vc_accept(struct socket *head, struct socket **sop)
315{
316	int error = 0;
317	struct socket *so;
318
319	if ((head->so_options & SO_ACCEPTCONN) == 0) {
320		error = EINVAL;
321		goto done;
322	}
323#ifdef MAC
324	error = mac_socket_check_accept(curthread->td_ucred, head);
325	if (error != 0)
326		goto done;
327#endif
328	ACCEPT_LOCK();
329	if (TAILQ_EMPTY(&head->so_comp)) {
330		ACCEPT_UNLOCK();
331		error = EWOULDBLOCK;
332		goto done;
333	}
334	so = TAILQ_FIRST(&head->so_comp);
335	KASSERT(!(so->so_qstate & SQ_INCOMP), ("svc_vc_accept: so SQ_INCOMP"));
336	KASSERT(so->so_qstate & SQ_COMP, ("svc_vc_accept: so not SQ_COMP"));
337
338	/*
339	 * Before changing the flags on the socket, we have to bump the
340	 * reference count.  Otherwise, if the protocol calls sofree(),
341	 * the socket will be released due to a zero refcount.
342	 * XXX might not need soref() since this is simpler than kern_accept.
343	 */
344	SOCK_LOCK(so);			/* soref() and so_state update */
345	soref(so);			/* file descriptor reference */
346
347	TAILQ_REMOVE(&head->so_comp, so, so_list);
348	head->so_qlen--;
349	so->so_state |= (head->so_state & SS_NBIO);
350	so->so_qstate &= ~SQ_COMP;
351	so->so_head = NULL;
352
353	SOCK_UNLOCK(so);
354	ACCEPT_UNLOCK();
355
356	*sop = so;
357
358	/* connection has been removed from the listen queue */
359	KNOTE_UNLOCKED(&head->so_rcv.sb_sel.si_note, 0);
360done:
361	return (error);
362}
363
364/*ARGSUSED*/
365static bool_t
366svc_vc_rendezvous_recv(SVCXPRT *xprt, struct rpc_msg *msg,
367    struct sockaddr **addrp, struct mbuf **mp)
368{
369	struct socket *so = NULL;
370	struct sockaddr *sa = NULL;
371	int error;
372	SVCXPRT *new_xprt;
373
374	/*
375	 * The socket upcall calls xprt_active() which will eventually
376	 * cause the server to call us here. We attempt to accept a
377	 * connection from the socket and turn it into a new
378	 * transport. If the accept fails, we have drained all pending
379	 * connections so we call xprt_inactive().
380	 */
381	sx_xlock(&xprt->xp_lock);
382
383	error = svc_vc_accept(xprt->xp_socket, &so);
384
385	if (error == EWOULDBLOCK) {
386		/*
387		 * We must re-test for new connections after taking
388		 * the lock to protect us in the case where a new
389		 * connection arrives after our call to accept fails
390		 * with EWOULDBLOCK.
391		 */
392		ACCEPT_LOCK();
393		if (TAILQ_EMPTY(&xprt->xp_socket->so_comp))
394			xprt_inactive_self(xprt);
395		ACCEPT_UNLOCK();
396		sx_xunlock(&xprt->xp_lock);
397		return (FALSE);
398	}
399
400	if (error) {
401		SOCKBUF_LOCK(&xprt->xp_socket->so_rcv);
402		if (xprt->xp_upcallset) {
403			xprt->xp_upcallset = 0;
404			soupcall_clear(xprt->xp_socket, SO_RCV);
405		}
406		SOCKBUF_UNLOCK(&xprt->xp_socket->so_rcv);
407		xprt_inactive_self(xprt);
408		sx_xunlock(&xprt->xp_lock);
409		return (FALSE);
410	}
411
412	sx_xunlock(&xprt->xp_lock);
413
414	sa = 0;
415	error = soaccept(so, &sa);
416
417	if (error) {
418		/*
419		 * XXX not sure if I need to call sofree or soclose here.
420		 */
421		if (sa)
422			free(sa, M_SONAME);
423		return (FALSE);
424	}
425
426	/*
427	 * svc_vc_create_conn will call xprt_register - we don't need
428	 * to do anything with the new connection except derefence it.
429	 */
430	new_xprt = svc_vc_create_conn(xprt->xp_pool, so, sa);
431	if (!new_xprt) {
432		soclose(so);
433	} else {
434		SVC_RELEASE(new_xprt);
435	}
436
437	free(sa, M_SONAME);
438
439	return (FALSE); /* there is never an rpc msg to be processed */
440}
441
442/*ARGSUSED*/
443static enum xprt_stat
444svc_vc_rendezvous_stat(SVCXPRT *xprt)
445{
446
447	return (XPRT_IDLE);
448}
449
450static void
451svc_vc_destroy_common(SVCXPRT *xprt)
452{
453	SOCKBUF_LOCK(&xprt->xp_socket->so_rcv);
454	if (xprt->xp_upcallset) {
455		xprt->xp_upcallset = 0;
456		soupcall_clear(xprt->xp_socket, SO_RCV);
457	}
458	SOCKBUF_UNLOCK(&xprt->xp_socket->so_rcv);
459
460	if (xprt->xp_socket)
461		(void)soclose(xprt->xp_socket);
462
463	if (xprt->xp_netid)
464		(void) mem_free(xprt->xp_netid, strlen(xprt->xp_netid) + 1);
465	svc_xprt_free(xprt);
466}
467
468static void
469svc_vc_rendezvous_destroy(SVCXPRT *xprt)
470{
471
472	svc_vc_destroy_common(xprt);
473}
474
475static void
476svc_vc_destroy(SVCXPRT *xprt)
477{
478	struct cf_conn *cd = (struct cf_conn *)xprt->xp_p1;
479
480	svc_vc_destroy_common(xprt);
481
482	if (cd->mreq)
483		m_freem(cd->mreq);
484	if (cd->mpending)
485		m_freem(cd->mpending);
486	mem_free(cd, sizeof(*cd));
487}
488
489static void
490svc_vc_backchannel_destroy(SVCXPRT *xprt)
491{
492	struct cf_conn *cd = (struct cf_conn *)xprt->xp_p1;
493	struct mbuf *m, *m2;
494
495	svc_xprt_free(xprt);
496	m = cd->mreq;
497	while (m != NULL) {
498		m2 = m;
499		m = m->m_nextpkt;
500		m_freem(m2);
501	}
502	mem_free(cd, sizeof(*cd));
503}
504
505/*ARGSUSED*/
506static bool_t
507svc_vc_control(SVCXPRT *xprt, const u_int rq, void *in)
508{
509	return (FALSE);
510}
511
512static bool_t
513svc_vc_rendezvous_control(SVCXPRT *xprt, const u_int rq, void *in)
514{
515
516	return (FALSE);
517}
518
519static bool_t
520svc_vc_backchannel_control(SVCXPRT *xprt, const u_int rq, void *in)
521{
522
523	return (FALSE);
524}
525
526static enum xprt_stat
527svc_vc_stat(SVCXPRT *xprt)
528{
529	struct cf_conn *cd;
530
531	cd = (struct cf_conn *)(xprt->xp_p1);
532
533	if (cd->strm_stat == XPRT_DIED)
534		return (XPRT_DIED);
535
536	if (cd->mreq != NULL && cd->resid == 0 && cd->eor)
537		return (XPRT_MOREREQS);
538
539	if (soreadable(xprt->xp_socket))
540		return (XPRT_MOREREQS);
541
542	return (XPRT_IDLE);
543}
544
545static bool_t
546svc_vc_ack(SVCXPRT *xprt, uint32_t *ack)
547{
548
549	*ack = atomic_load_acq_32(&xprt->xp_snt_cnt);
550	*ack -= xprt->xp_socket->so_snd.sb_cc;
551	return (TRUE);
552}
553
554static enum xprt_stat
555svc_vc_backchannel_stat(SVCXPRT *xprt)
556{
557	struct cf_conn *cd;
558
559	cd = (struct cf_conn *)(xprt->xp_p1);
560
561	if (cd->mreq != NULL)
562		return (XPRT_MOREREQS);
563
564	return (XPRT_IDLE);
565}
566
567/*
568 * If we have an mbuf chain in cd->mpending, try to parse a record from it,
569 * leaving the result in cd->mreq. If we don't have a complete record, leave
570 * the partial result in cd->mreq and try to read more from the socket.
571 */
572static int
573svc_vc_process_pending(SVCXPRT *xprt)
574{
575	struct cf_conn *cd = (struct cf_conn *) xprt->xp_p1;
576	struct socket *so = xprt->xp_socket;
577	struct mbuf *m;
578
579	/*
580	 * If cd->resid is non-zero, we have part of the
581	 * record already, otherwise we are expecting a record
582	 * marker.
583	 */
584	if (!cd->resid && cd->mpending) {
585		/*
586		 * See if there is enough data buffered to
587		 * make up a record marker. Make sure we can
588		 * handle the case where the record marker is
589		 * split across more than one mbuf.
590		 */
591		size_t n = 0;
592		uint32_t header;
593
594		m = cd->mpending;
595		while (n < sizeof(uint32_t) && m) {
596			n += m->m_len;
597			m = m->m_next;
598		}
599		if (n < sizeof(uint32_t)) {
600			so->so_rcv.sb_lowat = sizeof(uint32_t) - n;
601			return (FALSE);
602		}
603		m_copydata(cd->mpending, 0, sizeof(header),
604		    (char *)&header);
605		header = ntohl(header);
606		cd->eor = (header & 0x80000000) != 0;
607		cd->resid = header & 0x7fffffff;
608		m_adj(cd->mpending, sizeof(uint32_t));
609	}
610
611	/*
612	 * Start pulling off mbufs from cd->mpending
613	 * until we either have a complete record or
614	 * we run out of data. We use m_split to pull
615	 * data - it will pull as much as possible and
616	 * split the last mbuf if necessary.
617	 */
618	while (cd->mpending && cd->resid) {
619		m = cd->mpending;
620		if (cd->mpending->m_next
621		    || cd->mpending->m_len > cd->resid)
622			cd->mpending = m_split(cd->mpending,
623			    cd->resid, M_WAITOK);
624		else
625			cd->mpending = NULL;
626		if (cd->mreq)
627			m_last(cd->mreq)->m_next = m;
628		else
629			cd->mreq = m;
630		while (m) {
631			cd->resid -= m->m_len;
632			m = m->m_next;
633		}
634	}
635
636	/*
637	 * Block receive upcalls if we have more data pending,
638	 * otherwise report our need.
639	 */
640	if (cd->mpending)
641		so->so_rcv.sb_lowat = INT_MAX;
642	else
643		so->so_rcv.sb_lowat =
644		    imax(1, imin(cd->resid, so->so_rcv.sb_hiwat / 2));
645	return (TRUE);
646}
647
648static bool_t
649svc_vc_recv(SVCXPRT *xprt, struct rpc_msg *msg,
650    struct sockaddr **addrp, struct mbuf **mp)
651{
652	struct cf_conn *cd = (struct cf_conn *) xprt->xp_p1;
653	struct uio uio;
654	struct mbuf *m;
655	struct socket* so = xprt->xp_socket;
656	XDR xdrs;
657	int error, rcvflag;
658
659	/*
660	 * Serialise access to the socket and our own record parsing
661	 * state.
662	 */
663	sx_xlock(&xprt->xp_lock);
664
665	for (;;) {
666		/* If we have no request ready, check pending queue. */
667		while (cd->mpending &&
668		    (cd->mreq == NULL || cd->resid != 0 || !cd->eor)) {
669			if (!svc_vc_process_pending(xprt))
670				break;
671		}
672
673		/* Process and return complete request in cd->mreq. */
674		if (cd->mreq != NULL && cd->resid == 0 && cd->eor) {
675
676			xdrmbuf_create(&xdrs, cd->mreq, XDR_DECODE);
677			cd->mreq = NULL;
678
679			/* Check for next request in a pending queue. */
680			svc_vc_process_pending(xprt);
681			if (cd->mreq == NULL || cd->resid != 0) {
682				SOCKBUF_LOCK(&so->so_rcv);
683				if (!soreadable(so))
684					xprt_inactive_self(xprt);
685				SOCKBUF_UNLOCK(&so->so_rcv);
686			}
687
688			sx_xunlock(&xprt->xp_lock);
689
690			if (! xdr_callmsg(&xdrs, msg)) {
691				XDR_DESTROY(&xdrs);
692				return (FALSE);
693			}
694
695			*addrp = NULL;
696			*mp = xdrmbuf_getall(&xdrs);
697			XDR_DESTROY(&xdrs);
698
699			return (TRUE);
700		}
701
702		/*
703		 * The socket upcall calls xprt_active() which will eventually
704		 * cause the server to call us here. We attempt to
705		 * read as much as possible from the socket and put
706		 * the result in cd->mpending. If the read fails,
707		 * we have drained both cd->mpending and the socket so
708		 * we can call xprt_inactive().
709		 */
710		uio.uio_resid = 1000000000;
711		uio.uio_td = curthread;
712		m = NULL;
713		rcvflag = MSG_DONTWAIT;
714		error = soreceive(so, NULL, &uio, &m, NULL, &rcvflag);
715
716		if (error == EWOULDBLOCK) {
717			/*
718			 * We must re-test for readability after
719			 * taking the lock to protect us in the case
720			 * where a new packet arrives on the socket
721			 * after our call to soreceive fails with
722			 * EWOULDBLOCK.
723			 */
724			SOCKBUF_LOCK(&so->so_rcv);
725			if (!soreadable(so))
726				xprt_inactive_self(xprt);
727			SOCKBUF_UNLOCK(&so->so_rcv);
728			sx_xunlock(&xprt->xp_lock);
729			return (FALSE);
730		}
731
732		if (error) {
733			SOCKBUF_LOCK(&so->so_rcv);
734			if (xprt->xp_upcallset) {
735				xprt->xp_upcallset = 0;
736				soupcall_clear(so, SO_RCV);
737			}
738			SOCKBUF_UNLOCK(&so->so_rcv);
739			xprt_inactive_self(xprt);
740			cd->strm_stat = XPRT_DIED;
741			sx_xunlock(&xprt->xp_lock);
742			return (FALSE);
743		}
744
745		if (!m) {
746			/*
747			 * EOF - the other end has closed the socket.
748			 */
749			xprt_inactive_self(xprt);
750			cd->strm_stat = XPRT_DIED;
751			sx_xunlock(&xprt->xp_lock);
752			return (FALSE);
753		}
754
755		if (cd->mpending)
756			m_last(cd->mpending)->m_next = m;
757		else
758			cd->mpending = m;
759	}
760}
761
762static bool_t
763svc_vc_backchannel_recv(SVCXPRT *xprt, struct rpc_msg *msg,
764    struct sockaddr **addrp, struct mbuf **mp)
765{
766	struct cf_conn *cd = (struct cf_conn *) xprt->xp_p1;
767	struct ct_data *ct;
768	struct mbuf *m;
769	XDR xdrs;
770
771	sx_xlock(&xprt->xp_lock);
772	ct = (struct ct_data *)xprt->xp_p2;
773	if (ct == NULL) {
774		sx_xunlock(&xprt->xp_lock);
775		return (FALSE);
776	}
777	mtx_lock(&ct->ct_lock);
778	m = cd->mreq;
779	if (m == NULL) {
780		xprt_inactive_self(xprt);
781		mtx_unlock(&ct->ct_lock);
782		sx_xunlock(&xprt->xp_lock);
783		return (FALSE);
784	}
785	cd->mreq = m->m_nextpkt;
786	mtx_unlock(&ct->ct_lock);
787	sx_xunlock(&xprt->xp_lock);
788
789	xdrmbuf_create(&xdrs, m, XDR_DECODE);
790	if (! xdr_callmsg(&xdrs, msg)) {
791		XDR_DESTROY(&xdrs);
792		return (FALSE);
793	}
794	*addrp = NULL;
795	*mp = xdrmbuf_getall(&xdrs);
796	XDR_DESTROY(&xdrs);
797	return (TRUE);
798}
799
800static bool_t
801svc_vc_reply(SVCXPRT *xprt, struct rpc_msg *msg,
802    struct sockaddr *addr, struct mbuf *m, uint32_t *seq)
803{
804	XDR xdrs;
805	struct mbuf *mrep;
806	bool_t stat = TRUE;
807	int error, len;
808
809	/*
810	 * Leave space for record mark.
811	 */
812	MGETHDR(mrep, M_WAIT, MT_DATA);
813	mrep->m_len = 0;
814	mrep->m_data += sizeof(uint32_t);
815
816	xdrmbuf_create(&xdrs, mrep, XDR_ENCODE);
817
818	if (msg->rm_reply.rp_stat == MSG_ACCEPTED &&
819	    msg->rm_reply.rp_acpt.ar_stat == SUCCESS) {
820		if (!xdr_replymsg(&xdrs, msg))
821			stat = FALSE;
822		else
823			xdrmbuf_append(&xdrs, m);
824	} else {
825		stat = xdr_replymsg(&xdrs, msg);
826	}
827
828	if (stat) {
829		m_fixhdr(mrep);
830
831		/*
832		 * Prepend a record marker containing the reply length.
833		 */
834		M_PREPEND(mrep, sizeof(uint32_t), M_WAIT);
835		len = mrep->m_pkthdr.len;
836		*mtod(mrep, uint32_t *) =
837			htonl(0x80000000 | (len - sizeof(uint32_t)));
838		atomic_add_acq_32(&xprt->xp_snd_cnt, len);
839		error = sosend(xprt->xp_socket, NULL, NULL, mrep, NULL,
840		    0, curthread);
841		if (!error) {
842			atomic_add_rel_32(&xprt->xp_snt_cnt, len);
843			if (seq)
844				*seq = xprt->xp_snd_cnt;
845			stat = TRUE;
846		} else
847			atomic_subtract_32(&xprt->xp_snd_cnt, len);
848	} else {
849		m_freem(mrep);
850	}
851
852	XDR_DESTROY(&xdrs);
853	xprt->xp_p2 = NULL;
854
855	return (stat);
856}
857
858static bool_t
859svc_vc_backchannel_reply(SVCXPRT *xprt, struct rpc_msg *msg,
860    struct sockaddr *addr, struct mbuf *m, uint32_t *seq)
861{
862	struct ct_data *ct;
863	XDR xdrs;
864	struct mbuf *mrep;
865	bool_t stat = TRUE;
866	int error;
867
868	/*
869	 * Leave space for record mark.
870	 */
871	MGETHDR(mrep, M_WAITOK, MT_DATA);
872	mrep->m_len = 0;
873	mrep->m_data += sizeof(uint32_t);
874
875	xdrmbuf_create(&xdrs, mrep, XDR_ENCODE);
876
877	if (msg->rm_reply.rp_stat == MSG_ACCEPTED &&
878	    msg->rm_reply.rp_acpt.ar_stat == SUCCESS) {
879		if (!xdr_replymsg(&xdrs, msg))
880			stat = FALSE;
881		else
882			xdrmbuf_append(&xdrs, m);
883	} else {
884		stat = xdr_replymsg(&xdrs, msg);
885	}
886
887	if (stat) {
888		m_fixhdr(mrep);
889
890		/*
891		 * Prepend a record marker containing the reply length.
892		 */
893		M_PREPEND(mrep, sizeof(uint32_t), M_WAITOK);
894		*mtod(mrep, uint32_t *) =
895			htonl(0x80000000 | (mrep->m_pkthdr.len
896				- sizeof(uint32_t)));
897		sx_xlock(&xprt->xp_lock);
898		ct = (struct ct_data *)xprt->xp_p2;
899		if (ct != NULL)
900			error = sosend(ct->ct_socket, NULL, NULL, mrep, NULL,
901			    0, curthread);
902		else
903			error = EPIPE;
904		sx_xunlock(&xprt->xp_lock);
905		if (!error) {
906			stat = TRUE;
907		}
908	} else {
909		m_freem(mrep);
910	}
911
912	XDR_DESTROY(&xdrs);
913
914	return (stat);
915}
916
917static bool_t
918svc_vc_null()
919{
920
921	return (FALSE);
922}
923
924static int
925svc_vc_soupcall(struct socket *so, void *arg, int waitflag)
926{
927	SVCXPRT *xprt = (SVCXPRT *) arg;
928
929	if (soreadable(xprt->xp_socket))
930		xprt_active(xprt);
931	return (SU_OK);
932}
933
934#if 0
935/*
936 * Get the effective UID of the sending process. Used by rpcbind, keyserv
937 * and rpc.yppasswdd on AF_LOCAL.
938 */
939int
940__rpc_get_local_uid(SVCXPRT *transp, uid_t *uid) {
941	int sock, ret;
942	gid_t egid;
943	uid_t euid;
944	struct sockaddr *sa;
945
946	sock = transp->xp_fd;
947	sa = (struct sockaddr *)transp->xp_rtaddr;
948	if (sa->sa_family == AF_LOCAL) {
949		ret = getpeereid(sock, &euid, &egid);
950		if (ret == 0)
951			*uid = euid;
952		return (ret);
953	} else
954		return (-1);
955}
956#endif
957