xref: /freebsd/sys/rpc/clnt_dg.c (revision d184218c)
1 /*	$NetBSD: clnt_dg.c,v 1.4 2000/07/14 08:40:41 fvdl Exp $	*/
2 
3 /*
4  * Sun RPC is a product of Sun Microsystems, Inc. and is provided for
5  * unrestricted use provided that this legend is included on all tape
6  * media and as a part of the software program in whole or part.  Users
7  * may copy or modify Sun RPC without charge, but are not authorized
8  * to license or distribute it to anyone else except as part of a product or
9  * program developed by the user.
10  *
11  * SUN RPC IS PROVIDED AS IS WITH NO WARRANTIES OF ANY KIND INCLUDING THE
12  * WARRANTIES OF DESIGN, MERCHANTIBILITY AND FITNESS FOR A PARTICULAR
13  * PURPOSE, OR ARISING FROM A COURSE OF DEALING, USAGE OR TRADE PRACTICE.
14  *
15  * Sun RPC is provided with no support and without any obligation on the
16  * part of Sun Microsystems, Inc. to assist in its use, correction,
17  * modification or enhancement.
18  *
19  * SUN MICROSYSTEMS, INC. SHALL HAVE NO LIABILITY WITH RESPECT TO THE
20  * INFRINGEMENT OF COPYRIGHTS, TRADE SECRETS OR ANY PATENTS BY SUN RPC
21  * OR ANY PART THEREOF.
22  *
23  * In no event will Sun Microsystems, Inc. be liable for any lost revenue
24  * or profits or other special, indirect and consequential damages, even if
25  * Sun has been advised of the possibility of such damages.
26  *
27  * Sun Microsystems, Inc.
28  * 2550 Garcia Avenue
29  * Mountain View, California  94043
30  */
31 /*
32  * Copyright (c) 1986-1991 by Sun Microsystems Inc.
33  */
34 
35 #if defined(LIBC_SCCS) && !defined(lint)
36 #ident	"@(#)clnt_dg.c	1.23	94/04/22 SMI"
37 static char sccsid[] = "@(#)clnt_dg.c 1.19 89/03/16 Copyr 1988 Sun Micro";
38 #endif
39 #include <sys/cdefs.h>
40 __FBSDID("$FreeBSD$");
41 
42 /*
43  * Implements a connectionless client side RPC.
44  */
45 
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/kernel.h>
49 #include <sys/lock.h>
50 #include <sys/malloc.h>
51 #include <sys/mbuf.h>
52 #include <sys/mutex.h>
53 #include <sys/pcpu.h>
54 #include <sys/proc.h>
55 #include <sys/socket.h>
56 #include <sys/socketvar.h>
57 #include <sys/time.h>
58 #include <sys/uio.h>
59 
60 #include <net/vnet.h>
61 
62 #include <rpc/rpc.h>
63 #include <rpc/rpc_com.h>
64 
65 
66 #ifdef _FREEFALL_CONFIG
67 /*
68  * Disable RPC exponential back-off for FreeBSD.org systems.
69  */
70 #define	RPC_MAX_BACKOFF		1 /* second */
71 #else
72 #define	RPC_MAX_BACKOFF		30 /* seconds */
73 #endif
74 
75 static bool_t time_not_ok(struct timeval *);
76 static enum clnt_stat clnt_dg_call(CLIENT *, struct rpc_callextra *,
77     rpcproc_t, struct mbuf *, struct mbuf **, struct timeval);
78 static void clnt_dg_geterr(CLIENT *, struct rpc_err *);
79 static bool_t clnt_dg_freeres(CLIENT *, xdrproc_t, void *);
80 static void clnt_dg_abort(CLIENT *);
81 static bool_t clnt_dg_control(CLIENT *, u_int, void *);
82 static void clnt_dg_close(CLIENT *);
83 static void clnt_dg_destroy(CLIENT *);
84 static int clnt_dg_soupcall(struct socket *so, void *arg, int waitflag);
85 
86 static struct clnt_ops clnt_dg_ops = {
87 	.cl_call =	clnt_dg_call,
88 	.cl_abort =	clnt_dg_abort,
89 	.cl_geterr =	clnt_dg_geterr,
90 	.cl_freeres =	clnt_dg_freeres,
91 	.cl_close =	clnt_dg_close,
92 	.cl_destroy =	clnt_dg_destroy,
93 	.cl_control =	clnt_dg_control
94 };
95 
96 static const char mem_err_clnt_dg[] = "clnt_dg_create: out of memory";
97 
98 /*
99  * A pending RPC request which awaits a reply. Requests which have
100  * received their reply will have cr_xid set to zero and cr_mrep to
101  * the mbuf chain of the reply.
102  */
103 struct cu_request {
104 	TAILQ_ENTRY(cu_request) cr_link;
105 	CLIENT			*cr_client;	/* owner */
106 	uint32_t		cr_xid;		/* XID of request */
107 	struct mbuf		*cr_mrep;	/* reply received by upcall */
108 	int			cr_error;	/* any error from upcall */
109 	char			cr_verf[MAX_AUTH_BYTES]; /* reply verf */
110 };
111 
112 TAILQ_HEAD(cu_request_list, cu_request);
113 
114 #define MCALL_MSG_SIZE 24
115 
116 /*
117  * This structure is pointed to by the socket buffer's sb_upcallarg
118  * member. It is separate from the client private data to facilitate
119  * multiple clients sharing the same socket. The cs_lock mutex is used
120  * to protect all fields of this structure, the socket's receive
121  * buffer SOCKBUF_LOCK is used to ensure that exactly one of these
122  * structures is installed on the socket.
123  */
124 struct cu_socket {
125 	struct mtx		cs_lock;
126 	int			cs_refs;	/* Count of clients */
127 	struct cu_request_list	cs_pending;	/* Requests awaiting replies */
128 	int			cs_upcallrefs;	/* Refcnt of upcalls in prog.*/
129 };
130 
131 static void clnt_dg_upcallsdone(struct socket *, struct cu_socket *);
132 
133 /*
134  * Private data kept per client handle
135  */
136 struct cu_data {
137 	int			cu_threads;	/* # threads in clnt_vc_call */
138 	bool_t			cu_closing;	/* TRUE if we are closing */
139 	bool_t			cu_closed;	/* TRUE if we are closed */
140 	struct socket		*cu_socket;	/* connection socket */
141 	bool_t			cu_closeit;	/* opened by library */
142 	struct sockaddr_storage	cu_raddr;	/* remote address */
143 	int			cu_rlen;
144 	struct timeval		cu_wait;	/* retransmit interval */
145 	struct timeval		cu_total;	/* total time for the call */
146 	struct rpc_err		cu_error;
147 	uint32_t		cu_xid;
148 	char			cu_mcallc[MCALL_MSG_SIZE]; /* marshalled callmsg */
149 	size_t			cu_mcalllen;
150 	size_t			cu_sendsz;	/* send size */
151 	size_t			cu_recvsz;	/* recv size */
152 	int			cu_async;
153 	int			cu_connect;	/* Use connect(). */
154 	int			cu_connected;	/* Have done connect(). */
155 	const char		*cu_waitchan;
156 	int			cu_waitflag;
157 	int			cu_cwnd;	/* congestion window */
158 	int			cu_sent;	/* number of in-flight RPCs */
159 	bool_t			cu_cwnd_wait;
160 };
161 
162 #define CWNDSCALE	256
163 #define MAXCWND		(32 * CWNDSCALE)
164 
165 /*
166  * Connection less client creation returns with client handle parameters.
167  * Default options are set, which the user can change using clnt_control().
168  * fd should be open and bound.
169  * NB: The rpch->cl_auth is initialized to null authentication.
170  * 	Caller may wish to set this something more useful.
171  *
172  * sendsz and recvsz are the maximum allowable packet sizes that can be
173  * sent and received. Normally they are the same, but they can be
174  * changed to improve the program efficiency and buffer allocation.
175  * If they are 0, use the transport default.
176  *
177  * If svcaddr is NULL, returns NULL.
178  */
179 CLIENT *
180 clnt_dg_create(
181 	struct socket *so,
182 	struct sockaddr *svcaddr,	/* servers address */
183 	rpcprog_t program,		/* program number */
184 	rpcvers_t version,		/* version number */
185 	size_t sendsz,			/* buffer recv size */
186 	size_t recvsz)			/* buffer send size */
187 {
188 	CLIENT *cl = NULL;		/* client handle */
189 	struct cu_data *cu = NULL;	/* private data */
190 	struct cu_socket *cs = NULL;
191 	struct sockbuf *sb;
192 	struct timeval now;
193 	struct rpc_msg call_msg;
194 	struct __rpc_sockinfo si;
195 	XDR xdrs;
196 	int error;
197 
198 	if (svcaddr == NULL) {
199 		rpc_createerr.cf_stat = RPC_UNKNOWNADDR;
200 		return (NULL);
201 	}
202 
203 	if (!__rpc_socket2sockinfo(so, &si)) {
204 		rpc_createerr.cf_stat = RPC_TLIERROR;
205 		rpc_createerr.cf_error.re_errno = 0;
206 		return (NULL);
207 	}
208 
209 	/*
210 	 * Find the receive and the send size
211 	 */
212 	sendsz = __rpc_get_t_size(si.si_af, si.si_proto, (int)sendsz);
213 	recvsz = __rpc_get_t_size(si.si_af, si.si_proto, (int)recvsz);
214 	if ((sendsz == 0) || (recvsz == 0)) {
215 		rpc_createerr.cf_stat = RPC_TLIERROR; /* XXX */
216 		rpc_createerr.cf_error.re_errno = 0;
217 		return (NULL);
218 	}
219 
220 	cl = mem_alloc(sizeof (CLIENT));
221 
222 	/*
223 	 * Should be multiple of 4 for XDR.
224 	 */
225 	sendsz = ((sendsz + 3) / 4) * 4;
226 	recvsz = ((recvsz + 3) / 4) * 4;
227 	cu = mem_alloc(sizeof (*cu));
228 	cu->cu_threads = 0;
229 	cu->cu_closing = FALSE;
230 	cu->cu_closed = FALSE;
231 	(void) memcpy(&cu->cu_raddr, svcaddr, (size_t)svcaddr->sa_len);
232 	cu->cu_rlen = svcaddr->sa_len;
233 	/* Other values can also be set through clnt_control() */
234 	cu->cu_wait.tv_sec = 3;	/* heuristically chosen */
235 	cu->cu_wait.tv_usec = 0;
236 	cu->cu_total.tv_sec = -1;
237 	cu->cu_total.tv_usec = -1;
238 	cu->cu_sendsz = sendsz;
239 	cu->cu_recvsz = recvsz;
240 	cu->cu_async = FALSE;
241 	cu->cu_connect = FALSE;
242 	cu->cu_connected = FALSE;
243 	cu->cu_waitchan = "rpcrecv";
244 	cu->cu_waitflag = 0;
245 	cu->cu_cwnd = MAXCWND / 2;
246 	cu->cu_sent = 0;
247 	cu->cu_cwnd_wait = FALSE;
248 	(void) getmicrotime(&now);
249 	cu->cu_xid = __RPC_GETXID(&now);
250 	call_msg.rm_xid = cu->cu_xid;
251 	call_msg.rm_call.cb_prog = program;
252 	call_msg.rm_call.cb_vers = version;
253 	xdrmem_create(&xdrs, cu->cu_mcallc, MCALL_MSG_SIZE, XDR_ENCODE);
254 	if (! xdr_callhdr(&xdrs, &call_msg)) {
255 		rpc_createerr.cf_stat = RPC_CANTENCODEARGS;  /* XXX */
256 		rpc_createerr.cf_error.re_errno = 0;
257 		goto err2;
258 	}
259 	cu->cu_mcalllen = XDR_GETPOS(&xdrs);
260 
261 	/*
262 	 * By default, closeit is always FALSE. It is users responsibility
263 	 * to do a close on it, else the user may use clnt_control
264 	 * to let clnt_destroy do it for him/her.
265 	 */
266 	cu->cu_closeit = FALSE;
267 	cu->cu_socket = so;
268 	error = soreserve(so, (u_long)sendsz, (u_long)recvsz);
269 	if (error != 0) {
270 		rpc_createerr.cf_stat = RPC_FAILED;
271 		rpc_createerr.cf_error.re_errno = error;
272 		goto err2;
273 	}
274 
275 	sb = &so->so_rcv;
276 	SOCKBUF_LOCK(&so->so_rcv);
277 recheck_socket:
278 	if (sb->sb_upcall) {
279 		if (sb->sb_upcall != clnt_dg_soupcall) {
280 			SOCKBUF_UNLOCK(&so->so_rcv);
281 			printf("clnt_dg_create(): socket already has an incompatible upcall\n");
282 			goto err2;
283 		}
284 		cs = (struct cu_socket *) sb->sb_upcallarg;
285 		mtx_lock(&cs->cs_lock);
286 		cs->cs_refs++;
287 		mtx_unlock(&cs->cs_lock);
288 	} else {
289 		/*
290 		 * We are the first on this socket - allocate the
291 		 * structure and install it in the socket.
292 		 */
293 		SOCKBUF_UNLOCK(&so->so_rcv);
294 		cs = mem_alloc(sizeof(*cs));
295 		SOCKBUF_LOCK(&so->so_rcv);
296 		if (sb->sb_upcall) {
297 			/*
298 			 * We have lost a race with some other client.
299 			 */
300 			mem_free(cs, sizeof(*cs));
301 			goto recheck_socket;
302 		}
303 		mtx_init(&cs->cs_lock, "cs->cs_lock", NULL, MTX_DEF);
304 		cs->cs_refs = 1;
305 		cs->cs_upcallrefs = 0;
306 		TAILQ_INIT(&cs->cs_pending);
307 		soupcall_set(so, SO_RCV, clnt_dg_soupcall, cs);
308 	}
309 	SOCKBUF_UNLOCK(&so->so_rcv);
310 
311 	cl->cl_refs = 1;
312 	cl->cl_ops = &clnt_dg_ops;
313 	cl->cl_private = (caddr_t)(void *)cu;
314 	cl->cl_auth = authnone_create();
315 	cl->cl_tp = NULL;
316 	cl->cl_netid = NULL;
317 	return (cl);
318 err2:
319 	if (cl) {
320 		mem_free(cl, sizeof (CLIENT));
321 		if (cu)
322 			mem_free(cu, sizeof (*cu));
323 	}
324 	return (NULL);
325 }
326 
327 static enum clnt_stat
328 clnt_dg_call(
329 	CLIENT		*cl,		/* client handle */
330 	struct rpc_callextra *ext,	/* call metadata */
331 	rpcproc_t	proc,		/* procedure number */
332 	struct mbuf	*args,		/* pointer to args */
333 	struct mbuf	**resultsp,	/* pointer to results */
334 	struct timeval	utimeout)	/* seconds to wait before giving up */
335 {
336 	struct cu_data *cu = (struct cu_data *)cl->cl_private;
337 	struct cu_socket *cs;
338 	struct rpc_timers *rt;
339 	AUTH *auth;
340 	struct rpc_err *errp;
341 	enum clnt_stat stat;
342 	XDR xdrs;
343 	struct rpc_msg reply_msg;
344 	bool_t ok;
345 	int retrans;			/* number of re-transmits so far */
346 	int nrefreshes = 2;		/* number of times to refresh cred */
347 	struct timeval *tvp;
348 	int timeout;
349 	int retransmit_time;
350 	int next_sendtime, starttime, rtt, time_waited, tv = 0;
351 	struct sockaddr *sa;
352 	socklen_t salen;
353 	uint32_t xid = 0;
354 	struct mbuf *mreq = NULL, *results;
355 	struct cu_request *cr;
356 	int error;
357 
358 	cs = cu->cu_socket->so_rcv.sb_upcallarg;
359 	cr = malloc(sizeof(struct cu_request), M_RPC, M_WAITOK);
360 
361 	mtx_lock(&cs->cs_lock);
362 
363 	if (cu->cu_closing || cu->cu_closed) {
364 		mtx_unlock(&cs->cs_lock);
365 		free(cr, M_RPC);
366 		return (RPC_CANTSEND);
367 	}
368 	cu->cu_threads++;
369 
370 	if (ext) {
371 		auth = ext->rc_auth;
372 		errp = &ext->rc_err;
373 	} else {
374 		auth = cl->cl_auth;
375 		errp = &cu->cu_error;
376 	}
377 
378 	cr->cr_client = cl;
379 	cr->cr_mrep = NULL;
380 	cr->cr_error = 0;
381 
382 	if (cu->cu_total.tv_usec == -1) {
383 		tvp = &utimeout; /* use supplied timeout */
384 	} else {
385 		tvp = &cu->cu_total; /* use default timeout */
386 	}
387 	if (tvp->tv_sec || tvp->tv_usec)
388 		timeout = tvtohz(tvp);
389 	else
390 		timeout = 0;
391 
392 	if (cu->cu_connect && !cu->cu_connected) {
393 		mtx_unlock(&cs->cs_lock);
394 		error = soconnect(cu->cu_socket,
395 		    (struct sockaddr *)&cu->cu_raddr, curthread);
396 		mtx_lock(&cs->cs_lock);
397 		if (error) {
398 			errp->re_errno = error;
399 			errp->re_status = stat = RPC_CANTSEND;
400 			goto out;
401 		}
402 		cu->cu_connected = 1;
403 	}
404 	if (cu->cu_connected) {
405 		sa = NULL;
406 		salen = 0;
407 	} else {
408 		sa = (struct sockaddr *)&cu->cu_raddr;
409 		salen = cu->cu_rlen;
410 	}
411 	time_waited = 0;
412 	retrans = 0;
413 	if (ext && ext->rc_timers) {
414 		rt = ext->rc_timers;
415 		if (!rt->rt_rtxcur)
416 			rt->rt_rtxcur = tvtohz(&cu->cu_wait);
417 		retransmit_time = next_sendtime = rt->rt_rtxcur;
418 	} else {
419 		rt = NULL;
420 		retransmit_time = next_sendtime = tvtohz(&cu->cu_wait);
421 	}
422 
423 	starttime = ticks;
424 
425 call_again:
426 	mtx_assert(&cs->cs_lock, MA_OWNED);
427 
428 	cu->cu_xid++;
429 	xid = cu->cu_xid;
430 
431 send_again:
432 	mtx_unlock(&cs->cs_lock);
433 
434 	mreq = m_gethdr(M_WAITOK, MT_DATA);
435 	KASSERT(cu->cu_mcalllen <= MHLEN, ("RPC header too big"));
436 	bcopy(cu->cu_mcallc, mreq->m_data, cu->cu_mcalllen);
437 	mreq->m_len = cu->cu_mcalllen;
438 
439 	/*
440 	 * The XID is the first thing in the request.
441 	 */
442 	*mtod(mreq, uint32_t *) = htonl(xid);
443 
444 	xdrmbuf_create(&xdrs, mreq, XDR_ENCODE);
445 
446 	if (cu->cu_async == TRUE && args == NULL)
447 		goto get_reply;
448 
449 	if ((! XDR_PUTINT32(&xdrs, &proc)) ||
450 	    (! AUTH_MARSHALL(auth, xid, &xdrs,
451 		m_copym(args, 0, M_COPYALL, M_WAITOK)))) {
452 		errp->re_status = stat = RPC_CANTENCODEARGS;
453 		mtx_lock(&cs->cs_lock);
454 		goto out;
455 	}
456 	mreq->m_pkthdr.len = m_length(mreq, NULL);
457 
458 	cr->cr_xid = xid;
459 	mtx_lock(&cs->cs_lock);
460 
461 	/*
462 	 * Try to get a place in the congestion window.
463 	 */
464 	while (cu->cu_sent >= cu->cu_cwnd) {
465 		cu->cu_cwnd_wait = TRUE;
466 		error = msleep(&cu->cu_cwnd_wait, &cs->cs_lock,
467 		    cu->cu_waitflag, "rpccwnd", 0);
468 		if (error) {
469 			errp->re_errno = error;
470 			if (error == EINTR || error == ERESTART)
471 				errp->re_status = stat = RPC_INTR;
472 			else
473 				errp->re_status = stat = RPC_CANTSEND;
474 			goto out;
475 		}
476 	}
477 	cu->cu_sent += CWNDSCALE;
478 
479 	TAILQ_INSERT_TAIL(&cs->cs_pending, cr, cr_link);
480 	mtx_unlock(&cs->cs_lock);
481 
482 	/*
483 	 * sosend consumes mreq.
484 	 */
485 	error = sosend(cu->cu_socket, sa, NULL, mreq, NULL, 0, curthread);
486 	mreq = NULL;
487 
488 	/*
489 	 * sub-optimal code appears here because we have
490 	 * some clock time to spare while the packets are in flight.
491 	 * (We assume that this is actually only executed once.)
492 	 */
493 	reply_msg.acpted_rply.ar_verf.oa_flavor = AUTH_NULL;
494 	reply_msg.acpted_rply.ar_verf.oa_base = cr->cr_verf;
495 	reply_msg.acpted_rply.ar_verf.oa_length = 0;
496 	reply_msg.acpted_rply.ar_results.where = NULL;
497 	reply_msg.acpted_rply.ar_results.proc = (xdrproc_t)xdr_void;
498 
499 	mtx_lock(&cs->cs_lock);
500 	if (error) {
501 		TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
502 		errp->re_errno = error;
503 		errp->re_status = stat = RPC_CANTSEND;
504 		cu->cu_sent -= CWNDSCALE;
505 		if (cu->cu_cwnd_wait) {
506 			cu->cu_cwnd_wait = FALSE;
507 			wakeup(&cu->cu_cwnd_wait);
508 		}
509 		goto out;
510 	}
511 
512 	/*
513 	 * Check to see if we got an upcall while waiting for the
514 	 * lock.
515 	 */
516 	if (cr->cr_error) {
517 		TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
518 		errp->re_errno = cr->cr_error;
519 		errp->re_status = stat = RPC_CANTRECV;
520 		cu->cu_sent -= CWNDSCALE;
521 		if (cu->cu_cwnd_wait) {
522 			cu->cu_cwnd_wait = FALSE;
523 			wakeup(&cu->cu_cwnd_wait);
524 		}
525 		goto out;
526 	}
527 	if (cr->cr_mrep) {
528 		TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
529 		cu->cu_sent -= CWNDSCALE;
530 		if (cu->cu_cwnd_wait) {
531 			cu->cu_cwnd_wait = FALSE;
532 			wakeup(&cu->cu_cwnd_wait);
533 		}
534 		goto got_reply;
535 	}
536 
537 	/*
538 	 * Hack to provide rpc-based message passing
539 	 */
540 	if (timeout == 0) {
541 		TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
542 		errp->re_status = stat = RPC_TIMEDOUT;
543 		cu->cu_sent -= CWNDSCALE;
544 		if (cu->cu_cwnd_wait) {
545 			cu->cu_cwnd_wait = FALSE;
546 			wakeup(&cu->cu_cwnd_wait);
547 		}
548 		goto out;
549 	}
550 
551 get_reply:
552 	for (;;) {
553 		/* Decide how long to wait. */
554 		if (next_sendtime < timeout)
555 			tv = next_sendtime;
556 		else
557 			tv = timeout;
558 		tv -= time_waited;
559 
560 		if (tv > 0) {
561 			if (cu->cu_closing || cu->cu_closed) {
562 				error = 0;
563 				cr->cr_error = ESHUTDOWN;
564 			} else {
565 				error = msleep(cr, &cs->cs_lock,
566 				    cu->cu_waitflag, cu->cu_waitchan, tv);
567 			}
568 		} else {
569 			error = EWOULDBLOCK;
570 		}
571 
572 		TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
573 		cu->cu_sent -= CWNDSCALE;
574 		if (cu->cu_cwnd_wait) {
575 			cu->cu_cwnd_wait = FALSE;
576 			wakeup(&cu->cu_cwnd_wait);
577 		}
578 
579 		if (!error) {
580 			/*
581 			 * We were woken up by the upcall.  If the
582 			 * upcall had a receive error, report that,
583 			 * otherwise we have a reply.
584 			 */
585 			if (cr->cr_error) {
586 				errp->re_errno = cr->cr_error;
587 				errp->re_status = stat = RPC_CANTRECV;
588 				goto out;
589 			}
590 
591 			cu->cu_cwnd += (CWNDSCALE * CWNDSCALE
592 			    + cu->cu_cwnd / 2) / cu->cu_cwnd;
593 			if (cu->cu_cwnd > MAXCWND)
594 				cu->cu_cwnd = MAXCWND;
595 
596 			if (rt) {
597 				/*
598 				 * Add one to the time since a tick
599 				 * count of N means that the actual
600 				 * time taken was somewhere between N
601 				 * and N+1.
602 				 */
603 				rtt = ticks - starttime + 1;
604 
605 				/*
606 				 * Update our estimate of the round
607 				 * trip time using roughly the
608 				 * algorithm described in RFC
609 				 * 2988. Given an RTT sample R:
610 				 *
611 				 * RTTVAR = (1-beta) * RTTVAR + beta * |SRTT-R|
612 				 * SRTT = (1-alpha) * SRTT + alpha * R
613 				 *
614 				 * where alpha = 0.125 and beta = 0.25.
615 				 *
616 				 * The initial retransmit timeout is
617 				 * SRTT + 4*RTTVAR and doubles on each
618 				 * retransmision.
619 				 */
620 				if (rt->rt_srtt == 0) {
621 					rt->rt_srtt = rtt;
622 					rt->rt_deviate = rtt / 2;
623 				} else {
624 					int32_t error = rtt - rt->rt_srtt;
625 					rt->rt_srtt += error / 8;
626 					error = abs(error) - rt->rt_deviate;
627 					rt->rt_deviate += error / 4;
628 				}
629 				rt->rt_rtxcur = rt->rt_srtt + 4*rt->rt_deviate;
630 			}
631 
632 			break;
633 		}
634 
635 		/*
636 		 * The sleep returned an error so our request is still
637 		 * on the list. If we got EWOULDBLOCK, we may want to
638 		 * re-send the request.
639 		 */
640 		if (error != EWOULDBLOCK) {
641 			errp->re_errno = error;
642 			if (error == EINTR || error == ERESTART)
643 				errp->re_status = stat = RPC_INTR;
644 			else
645 				errp->re_status = stat = RPC_CANTRECV;
646 			goto out;
647 		}
648 
649 		time_waited = ticks - starttime;
650 
651 		/* Check for timeout. */
652 		if (time_waited > timeout) {
653 			errp->re_errno = EWOULDBLOCK;
654 			errp->re_status = stat = RPC_TIMEDOUT;
655 			goto out;
656 		}
657 
658 		/* Retransmit if necessary. */
659 		if (time_waited >= next_sendtime) {
660 			cu->cu_cwnd /= 2;
661 			if (cu->cu_cwnd < CWNDSCALE)
662 				cu->cu_cwnd = CWNDSCALE;
663 			if (ext && ext->rc_feedback) {
664 				mtx_unlock(&cs->cs_lock);
665 				if (retrans == 0)
666 					ext->rc_feedback(FEEDBACK_REXMIT1,
667 					    proc, ext->rc_feedback_arg);
668 				else
669 					ext->rc_feedback(FEEDBACK_REXMIT2,
670 					    proc, ext->rc_feedback_arg);
671 				mtx_lock(&cs->cs_lock);
672 			}
673 			if (cu->cu_closing || cu->cu_closed) {
674 				errp->re_errno = ESHUTDOWN;
675 				errp->re_status = stat = RPC_CANTRECV;
676 				goto out;
677 			}
678 			retrans++;
679 			/* update retransmit_time */
680 			if (retransmit_time < RPC_MAX_BACKOFF * hz)
681 				retransmit_time = 2 * retransmit_time;
682 			next_sendtime += retransmit_time;
683 			goto send_again;
684 		}
685 		TAILQ_INSERT_TAIL(&cs->cs_pending, cr, cr_link);
686 	}
687 
688 got_reply:
689 	/*
690 	 * Now decode and validate the response. We need to drop the
691 	 * lock since xdr_replymsg may end up sleeping in malloc.
692 	 */
693 	mtx_unlock(&cs->cs_lock);
694 
695 	if (ext && ext->rc_feedback)
696 		ext->rc_feedback(FEEDBACK_OK, proc, ext->rc_feedback_arg);
697 
698 	xdrmbuf_create(&xdrs, cr->cr_mrep, XDR_DECODE);
699 	ok = xdr_replymsg(&xdrs, &reply_msg);
700 	cr->cr_mrep = NULL;
701 
702 	if (ok) {
703 		if ((reply_msg.rm_reply.rp_stat == MSG_ACCEPTED) &&
704 		    (reply_msg.acpted_rply.ar_stat == SUCCESS))
705 			errp->re_status = stat = RPC_SUCCESS;
706 		else
707 			stat = _seterr_reply(&reply_msg, &(cu->cu_error));
708 
709 		if (errp->re_status == RPC_SUCCESS) {
710 			results = xdrmbuf_getall(&xdrs);
711 			if (! AUTH_VALIDATE(auth, xid,
712 				&reply_msg.acpted_rply.ar_verf,
713 				&results)) {
714 				errp->re_status = stat = RPC_AUTHERROR;
715 				errp->re_why = AUTH_INVALIDRESP;
716 				if (retrans &&
717 				    auth->ah_cred.oa_flavor == RPCSEC_GSS) {
718 					/*
719 					 * If we retransmitted, its
720 					 * possible that we will
721 					 * receive a reply for one of
722 					 * the earlier transmissions
723 					 * (which will use an older
724 					 * RPCSEC_GSS sequence
725 					 * number). In this case, just
726 					 * go back and listen for a
727 					 * new reply. We could keep a
728 					 * record of all the seq
729 					 * numbers we have transmitted
730 					 * so far so that we could
731 					 * accept a reply for any of
732 					 * them here.
733 					 */
734 					XDR_DESTROY(&xdrs);
735 					mtx_lock(&cs->cs_lock);
736 					TAILQ_INSERT_TAIL(&cs->cs_pending,
737 					    cr, cr_link);
738 					cr->cr_mrep = NULL;
739 					goto get_reply;
740 				}
741 			} else {
742 				*resultsp = results;
743 			}
744 		}		/* end successful completion */
745 		/*
746 		 * If unsuccesful AND error is an authentication error
747 		 * then refresh credentials and try again, else break
748 		 */
749 		else if (stat == RPC_AUTHERROR)
750 			/* maybe our credentials need to be refreshed ... */
751 			if (nrefreshes > 0 &&
752 			    AUTH_REFRESH(auth, &reply_msg)) {
753 				nrefreshes--;
754 				XDR_DESTROY(&xdrs);
755 				mtx_lock(&cs->cs_lock);
756 				goto call_again;
757 			}
758 		/* end of unsuccessful completion */
759 	}	/* end of valid reply message */
760 	else {
761 		errp->re_status = stat = RPC_CANTDECODERES;
762 
763 	}
764 	XDR_DESTROY(&xdrs);
765 	mtx_lock(&cs->cs_lock);
766 out:
767 	mtx_assert(&cs->cs_lock, MA_OWNED);
768 
769 	if (mreq)
770 		m_freem(mreq);
771 	if (cr->cr_mrep)
772 		m_freem(cr->cr_mrep);
773 
774 	cu->cu_threads--;
775 	if (cu->cu_closing)
776 		wakeup(cu);
777 
778 	mtx_unlock(&cs->cs_lock);
779 
780 	if (auth && stat != RPC_SUCCESS)
781 		AUTH_VALIDATE(auth, xid, NULL, NULL);
782 
783 	free(cr, M_RPC);
784 
785 	return (stat);
786 }
787 
788 static void
789 clnt_dg_geterr(CLIENT *cl, struct rpc_err *errp)
790 {
791 	struct cu_data *cu = (struct cu_data *)cl->cl_private;
792 
793 	*errp = cu->cu_error;
794 }
795 
796 static bool_t
797 clnt_dg_freeres(CLIENT *cl, xdrproc_t xdr_res, void *res_ptr)
798 {
799 	XDR xdrs;
800 	bool_t dummy;
801 
802 	xdrs.x_op = XDR_FREE;
803 	dummy = (*xdr_res)(&xdrs, res_ptr);
804 
805 	return (dummy);
806 }
807 
808 /*ARGSUSED*/
809 static void
810 clnt_dg_abort(CLIENT *h)
811 {
812 }
813 
814 static bool_t
815 clnt_dg_control(CLIENT *cl, u_int request, void *info)
816 {
817 	struct cu_data *cu = (struct cu_data *)cl->cl_private;
818 	struct cu_socket *cs;
819 	struct sockaddr *addr;
820 
821 	cs = cu->cu_socket->so_rcv.sb_upcallarg;
822 	mtx_lock(&cs->cs_lock);
823 
824 	switch (request) {
825 	case CLSET_FD_CLOSE:
826 		cu->cu_closeit = TRUE;
827 		mtx_unlock(&cs->cs_lock);
828 		return (TRUE);
829 	case CLSET_FD_NCLOSE:
830 		cu->cu_closeit = FALSE;
831 		mtx_unlock(&cs->cs_lock);
832 		return (TRUE);
833 	}
834 
835 	/* for other requests which use info */
836 	if (info == NULL) {
837 		mtx_unlock(&cs->cs_lock);
838 		return (FALSE);
839 	}
840 	switch (request) {
841 	case CLSET_TIMEOUT:
842 		if (time_not_ok((struct timeval *)info)) {
843 			mtx_unlock(&cs->cs_lock);
844 			return (FALSE);
845 		}
846 		cu->cu_total = *(struct timeval *)info;
847 		break;
848 	case CLGET_TIMEOUT:
849 		*(struct timeval *)info = cu->cu_total;
850 		break;
851 	case CLSET_RETRY_TIMEOUT:
852 		if (time_not_ok((struct timeval *)info)) {
853 			mtx_unlock(&cs->cs_lock);
854 			return (FALSE);
855 		}
856 		cu->cu_wait = *(struct timeval *)info;
857 		break;
858 	case CLGET_RETRY_TIMEOUT:
859 		*(struct timeval *)info = cu->cu_wait;
860 		break;
861 	case CLGET_SVC_ADDR:
862 		/*
863 		 * Slightly different semantics to userland - we use
864 		 * sockaddr instead of netbuf.
865 		 */
866 		memcpy(info, &cu->cu_raddr, cu->cu_raddr.ss_len);
867 		break;
868 	case CLSET_SVC_ADDR:		/* set to new address */
869 		addr = (struct sockaddr *)info;
870 		(void) memcpy(&cu->cu_raddr, addr, addr->sa_len);
871 		break;
872 	case CLGET_XID:
873 		*(uint32_t *)info = cu->cu_xid;
874 		break;
875 
876 	case CLSET_XID:
877 		/* This will set the xid of the NEXT call */
878 		/* decrement by 1 as clnt_dg_call() increments once */
879 		cu->cu_xid = *(uint32_t *)info - 1;
880 		break;
881 
882 	case CLGET_VERS:
883 		/*
884 		 * This RELIES on the information that, in the call body,
885 		 * the version number field is the fifth field from the
886 		 * begining of the RPC header. MUST be changed if the
887 		 * call_struct is changed
888 		 */
889 		*(uint32_t *)info =
890 		    ntohl(*(uint32_t *)(void *)(cu->cu_mcallc +
891 		    4 * BYTES_PER_XDR_UNIT));
892 		break;
893 
894 	case CLSET_VERS:
895 		*(uint32_t *)(void *)(cu->cu_mcallc + 4 * BYTES_PER_XDR_UNIT)
896 			= htonl(*(uint32_t *)info);
897 		break;
898 
899 	case CLGET_PROG:
900 		/*
901 		 * This RELIES on the information that, in the call body,
902 		 * the program number field is the fourth field from the
903 		 * begining of the RPC header. MUST be changed if the
904 		 * call_struct is changed
905 		 */
906 		*(uint32_t *)info =
907 		    ntohl(*(uint32_t *)(void *)(cu->cu_mcallc +
908 		    3 * BYTES_PER_XDR_UNIT));
909 		break;
910 
911 	case CLSET_PROG:
912 		*(uint32_t *)(void *)(cu->cu_mcallc + 3 * BYTES_PER_XDR_UNIT)
913 			= htonl(*(uint32_t *)info);
914 		break;
915 	case CLSET_ASYNC:
916 		cu->cu_async = *(int *)info;
917 		break;
918 	case CLSET_CONNECT:
919 		cu->cu_connect = *(int *)info;
920 		break;
921 	case CLSET_WAITCHAN:
922 		cu->cu_waitchan = (const char *)info;
923 		break;
924 	case CLGET_WAITCHAN:
925 		*(const char **) info = cu->cu_waitchan;
926 		break;
927 	case CLSET_INTERRUPTIBLE:
928 		if (*(int *) info)
929 			cu->cu_waitflag = PCATCH;
930 		else
931 			cu->cu_waitflag = 0;
932 		break;
933 	case CLGET_INTERRUPTIBLE:
934 		if (cu->cu_waitflag)
935 			*(int *) info = TRUE;
936 		else
937 			*(int *) info = FALSE;
938 		break;
939 	default:
940 		mtx_unlock(&cs->cs_lock);
941 		return (FALSE);
942 	}
943 	mtx_unlock(&cs->cs_lock);
944 	return (TRUE);
945 }
946 
947 static void
948 clnt_dg_close(CLIENT *cl)
949 {
950 	struct cu_data *cu = (struct cu_data *)cl->cl_private;
951 	struct cu_socket *cs;
952 	struct cu_request *cr;
953 
954 	cs = cu->cu_socket->so_rcv.sb_upcallarg;
955 	mtx_lock(&cs->cs_lock);
956 
957 	if (cu->cu_closed) {
958 		mtx_unlock(&cs->cs_lock);
959 		return;
960 	}
961 
962 	if (cu->cu_closing) {
963 		while (cu->cu_closing)
964 			msleep(cu, &cs->cs_lock, 0, "rpcclose", 0);
965 		KASSERT(cu->cu_closed, ("client should be closed"));
966 		mtx_unlock(&cs->cs_lock);
967 		return;
968 	}
969 
970 	/*
971 	 * Abort any pending requests and wait until everyone
972 	 * has finished with clnt_vc_call.
973 	 */
974 	cu->cu_closing = TRUE;
975 	TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
976 		if (cr->cr_client == cl) {
977 			cr->cr_xid = 0;
978 			cr->cr_error = ESHUTDOWN;
979 			wakeup(cr);
980 		}
981 	}
982 
983 	while (cu->cu_threads)
984 		msleep(cu, &cs->cs_lock, 0, "rpcclose", 0);
985 
986 	cu->cu_closing = FALSE;
987 	cu->cu_closed = TRUE;
988 
989 	mtx_unlock(&cs->cs_lock);
990 	wakeup(cu);
991 }
992 
993 static void
994 clnt_dg_destroy(CLIENT *cl)
995 {
996 	struct cu_data *cu = (struct cu_data *)cl->cl_private;
997 	struct cu_socket *cs;
998 	struct socket *so = NULL;
999 	bool_t lastsocketref;
1000 
1001 	cs = cu->cu_socket->so_rcv.sb_upcallarg;
1002 	clnt_dg_close(cl);
1003 
1004 	SOCKBUF_LOCK(&cu->cu_socket->so_rcv);
1005 	mtx_lock(&cs->cs_lock);
1006 
1007 	cs->cs_refs--;
1008 	if (cs->cs_refs == 0) {
1009 		mtx_unlock(&cs->cs_lock);
1010 		soupcall_clear(cu->cu_socket, SO_RCV);
1011 		clnt_dg_upcallsdone(cu->cu_socket, cs);
1012 		SOCKBUF_UNLOCK(&cu->cu_socket->so_rcv);
1013 		mtx_destroy(&cs->cs_lock);
1014 		mem_free(cs, sizeof(*cs));
1015 		lastsocketref = TRUE;
1016 	} else {
1017 		mtx_unlock(&cs->cs_lock);
1018 		SOCKBUF_UNLOCK(&cu->cu_socket->so_rcv);
1019 		lastsocketref = FALSE;
1020 	}
1021 
1022 	if (cu->cu_closeit && lastsocketref) {
1023 		so = cu->cu_socket;
1024 		cu->cu_socket = NULL;
1025 	}
1026 
1027 	if (so)
1028 		soclose(so);
1029 
1030 	if (cl->cl_netid && cl->cl_netid[0])
1031 		mem_free(cl->cl_netid, strlen(cl->cl_netid) +1);
1032 	if (cl->cl_tp && cl->cl_tp[0])
1033 		mem_free(cl->cl_tp, strlen(cl->cl_tp) +1);
1034 	mem_free(cu, sizeof (*cu));
1035 	mem_free(cl, sizeof (CLIENT));
1036 }
1037 
1038 /*
1039  * Make sure that the time is not garbage.  -1 value is allowed.
1040  */
1041 static bool_t
1042 time_not_ok(struct timeval *t)
1043 {
1044 	return (t->tv_sec < -1 || t->tv_sec > 100000000 ||
1045 		t->tv_usec < -1 || t->tv_usec > 1000000);
1046 }
1047 
1048 int
1049 clnt_dg_soupcall(struct socket *so, void *arg, int waitflag)
1050 {
1051 	struct cu_socket *cs = (struct cu_socket *) arg;
1052 	struct uio uio;
1053 	struct mbuf *m;
1054 	struct mbuf *control;
1055 	struct cu_request *cr;
1056 	int error, rcvflag, foundreq;
1057 	uint32_t xid;
1058 
1059 	cs->cs_upcallrefs++;
1060 	uio.uio_resid = 1000000000;
1061 	uio.uio_td = curthread;
1062 	do {
1063 		SOCKBUF_UNLOCK(&so->so_rcv);
1064 		m = NULL;
1065 		control = NULL;
1066 		rcvflag = MSG_DONTWAIT;
1067 		error = soreceive(so, NULL, &uio, &m, &control, &rcvflag);
1068 		if (control)
1069 			m_freem(control);
1070 		SOCKBUF_LOCK(&so->so_rcv);
1071 
1072 		if (error == EWOULDBLOCK)
1073 			break;
1074 
1075 		/*
1076 		 * If there was an error, wake up all pending
1077 		 * requests.
1078 		 */
1079 		if (error) {
1080 			mtx_lock(&cs->cs_lock);
1081 			TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
1082 				cr->cr_xid = 0;
1083 				cr->cr_error = error;
1084 				wakeup(cr);
1085 			}
1086 			mtx_unlock(&cs->cs_lock);
1087 			break;
1088 		}
1089 
1090 		/*
1091 		 * The XID is in the first uint32_t of the reply.
1092 		 */
1093 		if (m->m_len < sizeof(xid) && m_length(m, NULL) < sizeof(xid)) {
1094 			/*
1095 			 * Should never happen.
1096 			 */
1097 			m_freem(m);
1098 			continue;
1099 		}
1100 
1101 		m_copydata(m, 0, sizeof(xid), (char *)&xid);
1102 		xid = ntohl(xid);
1103 
1104 		/*
1105 		 * Attempt to match this reply with a pending request.
1106 		 */
1107 		mtx_lock(&cs->cs_lock);
1108 		foundreq = 0;
1109 		TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
1110 			if (cr->cr_xid == xid) {
1111 				/*
1112 				 * This one matches. We leave the
1113 				 * reply mbuf in cr->cr_mrep. Set the
1114 				 * XID to zero so that we will ignore
1115 				 * any duplicated replies that arrive
1116 				 * before clnt_dg_call removes it from
1117 				 * the queue.
1118 				 */
1119 				cr->cr_xid = 0;
1120 				cr->cr_mrep = m;
1121 				cr->cr_error = 0;
1122 				foundreq = 1;
1123 				wakeup(cr);
1124 				break;
1125 			}
1126 		}
1127 		mtx_unlock(&cs->cs_lock);
1128 
1129 		/*
1130 		 * If we didn't find the matching request, just drop
1131 		 * it - its probably a repeated reply.
1132 		 */
1133 		if (!foundreq)
1134 			m_freem(m);
1135 	} while (m);
1136 	cs->cs_upcallrefs--;
1137 	if (cs->cs_upcallrefs < 0)
1138 		panic("rpcdg upcall refcnt");
1139 	if (cs->cs_upcallrefs == 0)
1140 		wakeup(&cs->cs_upcallrefs);
1141 	return (SU_OK);
1142 }
1143 
1144 /*
1145  * Wait for all upcalls in progress to complete.
1146  */
1147 static void
1148 clnt_dg_upcallsdone(struct socket *so, struct cu_socket *cs)
1149 {
1150 
1151 	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
1152 
1153 	while (cs->cs_upcallrefs > 0)
1154 		(void) msleep(&cs->cs_upcallrefs, SOCKBUF_MTX(&so->so_rcv), 0,
1155 		    "rpcdgup", 0);
1156 }
1157