sys/rpc/clnt_dg.c

   1 /*      $NetBSD: clnt_dg.c,v 1.4 2000/07/14 08:40:41 fvdl Exp $ */
   2
   3 /*-
   4  * SPDX-License-Identifier: BSD-3-Clause
   5  *
   6  * Copyright (c) 2009, Sun Microsystems, Inc.
   7  * All rights reserved.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions are met:
  11  * - Redistributions of source code must retain the above copyright notice,
  12  *   this list of conditions and the following disclaimer.
  13  * - Redistributions in binary form must reproduce the above copyright notice,
  14  *   this list of conditions and the following disclaimer in the documentation
  15  *   and/or other materials provided with the distribution.
  16  * - Neither the name of Sun Microsystems, Inc. nor the names of its
  17  *   contributors may be used to endorse or promote products derived
  18  *   from this software without specific prior written permission.
  19  *
  20  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
  21  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  23  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
  24  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
  25  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
  26  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  27  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  28  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
  29  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  30  * POSSIBILITY OF SUCH DAMAGE.
  31  */
  32 /*
  33  * Copyright (c) 1986-1991 by Sun Microsystems Inc.
  34  */
  35
  36 #if defined(LIBC_SCCS) && !defined(lint)
  37 #ident  "@(#)clnt_dg.c  1.23    94/04/22 SMI"
  38 static char sccsid[] = "@(#)clnt_dg.c 1.19 89/03/16 Copyr 1988 Sun Micro";
  39 #endif
  40 #include <sys/cdefs.h>
  41 __FBSDID("$FreeBSD$");
  42
  43 /*
  44  * Implements a connectionless client side RPC.
  45  */
  46
  47 #include <sys/param.h>
  48 #include <sys/systm.h>
  49 #include <sys/kernel.h>
  50 #include <sys/lock.h>
  51 #include <sys/malloc.h>
  52 #include <sys/mbuf.h>
  53 #include <sys/mutex.h>
  54 #include <sys/pcpu.h>
  55 #include <sys/proc.h>
  56 #include <sys/socket.h>
  57 #include <sys/socketvar.h>
  58 #include <sys/time.h>
  59 #include <sys/uio.h>
  60
  61 #include <net/vnet.h>
  62
  63 #include <rpc/rpc.h>
  64 #include <rpc/rpc_com.h>
  65
  66
  67 #ifdef _FREEFALL_CONFIG
  68 /*
  69  * Disable RPC exponential back-off for FreeBSD.org systems.
  70  */
  71 #define RPC_MAX_BACKOFF         1 /* second */
  72 #else
  73 #define RPC_MAX_BACKOFF         30 /* seconds */
  74 #endif
  75
  76 static bool_t time_not_ok(struct timeval *);
  77 static enum clnt_stat clnt_dg_call(CLIENT *, struct rpc_callextra *,
  78     rpcproc_t, struct mbuf *, struct mbuf **, struct timeval);
  79 static void clnt_dg_geterr(CLIENT *, struct rpc_err *);
  80 static bool_t clnt_dg_freeres(CLIENT *, xdrproc_t, void *);
  81 static void clnt_dg_abort(CLIENT *);
  82 static bool_t clnt_dg_control(CLIENT *, u_int, void *);
  83 static void clnt_dg_close(CLIENT *);
  84 static void clnt_dg_destroy(CLIENT *);
  85 static int clnt_dg_soupcall(struct socket *so, void *arg, int waitflag);
  86
  87 static struct clnt_ops clnt_dg_ops = {
  88         .cl_call =      clnt_dg_call,
  89         .cl_abort =     clnt_dg_abort,
  90         .cl_geterr =    clnt_dg_geterr,
  91         .cl_freeres =   clnt_dg_freeres,
  92         .cl_close =     clnt_dg_close,
  93         .cl_destroy =   clnt_dg_destroy,
  94         .cl_control =   clnt_dg_control
  95 };
  96
  97 /*
  98  * A pending RPC request which awaits a reply. Requests which have
  99  * received their reply will have cr_xid set to zero and cr_mrep to
 100  * the mbuf chain of the reply.
 101  */
 102 struct cu_request {
 103         TAILQ_ENTRY(cu_request) cr_link;
 104         CLIENT                  *cr_client;     /* owner */
 105         uint32_t                cr_xid;         /* XID of request */
 106         struct mbuf             *cr_mrep;       /* reply received by upcall */
 107         int                     cr_error;       /* any error from upcall */
 108         char                    cr_verf[MAX_AUTH_BYTES]; /* reply verf */
 109 };
 110
 111 TAILQ_HEAD(cu_request_list, cu_request);
 112
 113 #define MCALL_MSG_SIZE 24
 114
 115 /*
 116  * This structure is pointed to by the socket buffer's sb_upcallarg
 117  * member. It is separate from the client private data to facilitate
 118  * multiple clients sharing the same socket. The cs_lock mutex is used
 119  * to protect all fields of this structure, the socket's receive
 120  * buffer SOCKBUF_LOCK is used to ensure that exactly one of these
 121  * structures is installed on the socket.
 122  */
 123 struct cu_socket {
 124         struct mtx              cs_lock;
 125         int                     cs_refs;        /* Count of clients */
 126         struct cu_request_list  cs_pending;     /* Requests awaiting replies */
 127         int                     cs_upcallrefs;  /* Refcnt of upcalls in prog.*/
 128 };
 129
 130 static void clnt_dg_upcallsdone(struct socket *, struct cu_socket *);
 131
 132 /*
 133  * Private data kept per client handle
 134  */
 135 struct cu_data {
 136         int                     cu_threads;     /* # threads in clnt_vc_call */
 137         bool_t                  cu_closing;     /* TRUE if we are closing */
 138         bool_t                  cu_closed;      /* TRUE if we are closed */
 139         struct socket           *cu_socket;     /* connection socket */
 140         bool_t                  cu_closeit;     /* opened by library */
 141         struct sockaddr_storage cu_raddr;       /* remote address */
 142         int                     cu_rlen;
 143         struct timeval          cu_wait;        /* retransmit interval */
 144         struct timeval          cu_total;       /* total time for the call */
 145         struct rpc_err          cu_error;
 146         uint32_t                cu_xid;
 147         char                    cu_mcallc[MCALL_MSG_SIZE]; /* marshalled callmsg */
 148         size_t                  cu_mcalllen;
 149         size_t                  cu_sendsz;      /* send size */
 150         size_t                  cu_recvsz;      /* recv size */
 151         int                     cu_async;
 152         int                     cu_connect;     /* Use connect(). */
 153         int                     cu_connected;   /* Have done connect(). */
 154         const char              *cu_waitchan;
 155         int                     cu_waitflag;
 156         int                     cu_cwnd;        /* congestion window */
 157         int                     cu_sent;        /* number of in-flight RPCs */
 158         bool_t                  cu_cwnd_wait;
 159 };
 160
 161 #define CWNDSCALE       256
 162 #define MAXCWND         (32 * CWNDSCALE)
 163
 164 /*
 165  * Connection less client creation returns with client handle parameters.
 166  * Default options are set, which the user can change using clnt_control().
 167  * fd should be open and bound.
 168  * NB: The rpch->cl_auth is initialized to null authentication.
 169  *      Caller may wish to set this something more useful.
 170  *
 171  * sendsz and recvsz are the maximum allowable packet sizes that can be
 172  * sent and received. Normally they are the same, but they can be
 173  * changed to improve the program efficiency and buffer allocation.
 174  * If they are 0, use the transport default.
 175  *
 176  * If svcaddr is NULL, returns NULL.
 177  */
 178 CLIENT *
 179 clnt_dg_create(
 180         struct socket *so,
 181         struct sockaddr *svcaddr,       /* servers address */
 182         rpcprog_t program,              /* program number */
 183         rpcvers_t version,              /* version number */
 184         size_t sendsz,                  /* buffer recv size */
 185         size_t recvsz)                  /* buffer send size */
 186 {
 187         CLIENT *cl = NULL;              /* client handle */
 188         struct cu_data *cu = NULL;      /* private data */
 189         struct cu_socket *cs = NULL;
 190         struct sockbuf *sb;
 191         struct timeval now;
 192         struct rpc_msg call_msg;
 193         struct __rpc_sockinfo si;
 194         XDR xdrs;
 195         int error;
 196
 197         if (svcaddr == NULL) {
 198                 rpc_createerr.cf_stat = RPC_UNKNOWNADDR;
 199                 return (NULL);
 200         }
 201
 202         if (!__rpc_socket2sockinfo(so, &si)) {
 203                 rpc_createerr.cf_stat = RPC_TLIERROR;
 204                 rpc_createerr.cf_error.re_errno = 0;
 205                 return (NULL);
 206         }
 207
 208         /*
 209          * Find the receive and the send size
 210          */
 211         sendsz = __rpc_get_t_size(si.si_af, si.si_proto, (int)sendsz);
 212         recvsz = __rpc_get_t_size(si.si_af, si.si_proto, (int)recvsz);
 213         if ((sendsz == 0) || (recvsz == 0)) {
 214                 rpc_createerr.cf_stat = RPC_TLIERROR; /* XXX */
 215                 rpc_createerr.cf_error.re_errno = 0;
 216                 return (NULL);
 217         }
 218
 219         cl = mem_alloc(sizeof (CLIENT));
 220
 221         /*
 222          * Should be multiple of 4 for XDR.
 223          */
 224         sendsz = rounddown(sendsz + 3, 4);
 225         recvsz = rounddown(recvsz + 3, 4);
 226         cu = mem_alloc(sizeof (*cu));
 227         cu->cu_threads = 0;
 228         cu->cu_closing = FALSE;
 229         cu->cu_closed = FALSE;
 230         (void) memcpy(&cu->cu_raddr, svcaddr, (size_t)svcaddr->sa_len);
 231         cu->cu_rlen = svcaddr->sa_len;
 232         /* Other values can also be set through clnt_control() */
 233         cu->cu_wait.tv_sec = 3; /* heuristically chosen */
 234         cu->cu_wait.tv_usec = 0;
 235         cu->cu_total.tv_sec = -1;
 236         cu->cu_total.tv_usec = -1;
 237         cu->cu_sendsz = sendsz;
 238         cu->cu_recvsz = recvsz;
 239         cu->cu_async = FALSE;
 240         cu->cu_connect = FALSE;
 241         cu->cu_connected = FALSE;
 242         cu->cu_waitchan = "rpcrecv";
 243         cu->cu_waitflag = 0;
 244         cu->cu_cwnd = MAXCWND / 2;
 245         cu->cu_sent = 0;
 246         cu->cu_cwnd_wait = FALSE;
 247         (void) getmicrotime(&now);
 248         cu->cu_xid = __RPC_GETXID(&now);
 249         call_msg.rm_xid = cu->cu_xid;
 250         call_msg.rm_call.cb_prog = program;
 251         call_msg.rm_call.cb_vers = version;
 252         xdrmem_create(&xdrs, cu->cu_mcallc, MCALL_MSG_SIZE, XDR_ENCODE);
 253         if (! xdr_callhdr(&xdrs, &call_msg)) {
 254                 rpc_createerr.cf_stat = RPC_CANTENCODEARGS;  /* XXX */
 255                 rpc_createerr.cf_error.re_errno = 0;
 256                 goto err2;
 257         }
 258         cu->cu_mcalllen = XDR_GETPOS(&xdrs);
 259
 260         /*
 261          * By default, closeit is always FALSE. It is users responsibility
 262          * to do a close on it, else the user may use clnt_control
 263          * to let clnt_destroy do it for him/her.
 264          */
 265         cu->cu_closeit = FALSE;
 266         cu->cu_socket = so;
 267         error = soreserve(so, (u_long)sendsz, (u_long)recvsz);
 268         if (error != 0) {
 269                 rpc_createerr.cf_stat = RPC_FAILED;
 270                 rpc_createerr.cf_error.re_errno = error;
 271                 goto err2;
 272         }
 273
 274         sb = &so->so_rcv;
 275         SOCKBUF_LOCK(&so->so_rcv);
 276 recheck_socket:
 277         if (sb->sb_upcall) {
 278                 if (sb->sb_upcall != clnt_dg_soupcall) {
 279                         SOCKBUF_UNLOCK(&so->so_rcv);
 280                         printf("clnt_dg_create(): socket already has an incompatible upcall\n");
 281                         goto err2;
 282                 }
 283                 cs = (struct cu_socket *) sb->sb_upcallarg;
 284                 mtx_lock(&cs->cs_lock);
 285                 cs->cs_refs++;
 286                 mtx_unlock(&cs->cs_lock);
 287         } else {
 288                 /*
 289                  * We are the first on this socket - allocate the
 290                  * structure and install it in the socket.
 291                  */
 292                 SOCKBUF_UNLOCK(&so->so_rcv);
 293                 cs = mem_alloc(sizeof(*cs));
 294                 SOCKBUF_LOCK(&so->so_rcv);
 295                 if (sb->sb_upcall) {
 296                         /*
 297                          * We have lost a race with some other client.
 298                          */
 299                         mem_free(cs, sizeof(*cs));
 300                         goto recheck_socket;
 301                 }
 302                 mtx_init(&cs->cs_lock, "cs->cs_lock", NULL, MTX_DEF);
 303                 cs->cs_refs = 1;
 304                 cs->cs_upcallrefs = 0;
 305                 TAILQ_INIT(&cs->cs_pending);
 306                 soupcall_set(so, SO_RCV, clnt_dg_soupcall, cs);
 307         }
 308         SOCKBUF_UNLOCK(&so->so_rcv);
 309
 310         cl->cl_refs = 1;
 311         cl->cl_ops = &clnt_dg_ops;
 312         cl->cl_private = (caddr_t)(void *)cu;
 313         cl->cl_auth = authnone_create();
 314         cl->cl_tp = NULL;
 315         cl->cl_netid = NULL;
 316         return (cl);
 317 err2:
 318         mem_free(cl, sizeof (CLIENT));
 319         mem_free(cu, sizeof (*cu));
 320
 321         return (NULL);
 322 }
 323
 324 static enum clnt_stat
 325 clnt_dg_call(
 326         CLIENT          *cl,            /* client handle */
 327         struct rpc_callextra *ext,      /* call metadata */
 328         rpcproc_t       proc,           /* procedure number */
 329         struct mbuf     *args,          /* pointer to args */
 330         struct mbuf     **resultsp,     /* pointer to results */
 331         struct timeval  utimeout)       /* seconds to wait before giving up */
 332 {
 333         struct cu_data *cu = (struct cu_data *)cl->cl_private;
 334         struct cu_socket *cs;
 335         struct rpc_timers *rt;
 336         AUTH *auth;
 337         struct rpc_err *errp;
 338         enum clnt_stat stat;
 339         XDR xdrs;
 340         struct rpc_msg reply_msg;
 341         bool_t ok;
 342         int retrans;                    /* number of re-transmits so far */
 343         int nrefreshes = 2;             /* number of times to refresh cred */
 344         struct timeval *tvp;
 345         int timeout;
 346         int retransmit_time;
 347         int next_sendtime, starttime, rtt, time_waited, tv = 0;
 348         struct sockaddr *sa;
 349         uint32_t xid = 0;
 350         struct mbuf *mreq = NULL, *results;
 351         struct cu_request *cr;
 352         int error;
 353
 354         cs = cu->cu_socket->so_rcv.sb_upcallarg;
 355         cr = malloc(sizeof(struct cu_request), M_RPC, M_WAITOK);
 356
 357         mtx_lock(&cs->cs_lock);
 358
 359         if (cu->cu_closing || cu->cu_closed) {
 360                 mtx_unlock(&cs->cs_lock);
 361                 free(cr, M_RPC);
 362                 return (RPC_CANTSEND);
 363         }
 364         cu->cu_threads++;
 365
 366         if (ext) {
 367                 auth = ext->rc_auth;
 368                 errp = &ext->rc_err;
 369         } else {
 370                 auth = cl->cl_auth;
 371                 errp = &cu->cu_error;
 372         }
 373
 374         cr->cr_client = cl;
 375         cr->cr_mrep = NULL;
 376         cr->cr_error = 0;
 377
 378         if (cu->cu_total.tv_usec == -1) {
 379                 tvp = &utimeout; /* use supplied timeout */
 380         } else {
 381                 tvp = &cu->cu_total; /* use default timeout */
 382         }
 383         if (tvp->tv_sec || tvp->tv_usec)
 384                 timeout = tvtohz(tvp);
 385         else
 386                 timeout = 0;
 387
 388         if (cu->cu_connect && !cu->cu_connected) {
 389                 mtx_unlock(&cs->cs_lock);
 390                 error = soconnect(cu->cu_socket,
 391                     (struct sockaddr *)&cu->cu_raddr, curthread);
 392                 mtx_lock(&cs->cs_lock);
 393                 if (error) {
 394                         errp->re_errno = error;
 395                         errp->re_status = stat = RPC_CANTSEND;
 396                         goto out;
 397                 }
 398                 cu->cu_connected = 1;
 399         }
 400         if (cu->cu_connected)
 401                 sa = NULL;
 402         else
 403                 sa = (struct sockaddr *)&cu->cu_raddr;
 404         time_waited = 0;
 405         retrans = 0;
 406         if (ext && ext->rc_timers) {
 407                 rt = ext->rc_timers;
 408                 if (!rt->rt_rtxcur)
 409                         rt->rt_rtxcur = tvtohz(&cu->cu_wait);
 410                 retransmit_time = next_sendtime = rt->rt_rtxcur;
 411         } else {
 412                 rt = NULL;
 413                 retransmit_time = next_sendtime = tvtohz(&cu->cu_wait);
 414         }
 415
 416         starttime = ticks;
 417
 418 call_again:
 419         mtx_assert(&cs->cs_lock, MA_OWNED);
 420
 421         cu->cu_xid++;
 422         xid = cu->cu_xid;
 423
 424 send_again:
 425         mtx_unlock(&cs->cs_lock);
 426
 427         mreq = m_gethdr(M_WAITOK, MT_DATA);
 428         KASSERT(cu->cu_mcalllen <= MHLEN, ("RPC header too big"));
 429         bcopy(cu->cu_mcallc, mreq->m_data, cu->cu_mcalllen);
 430         mreq->m_len = cu->cu_mcalllen;
 431
 432         /*
 433          * The XID is the first thing in the request.
 434          */
 435         *mtod(mreq, uint32_t *) = htonl(xid);
 436
 437         xdrmbuf_create(&xdrs, mreq, XDR_ENCODE);
 438
 439         if (cu->cu_async == TRUE && args == NULL)
 440                 goto get_reply;
 441
 442         if ((! XDR_PUTINT32(&xdrs, &proc)) ||
 443             (! AUTH_MARSHALL(auth, xid, &xdrs,
 444                 m_copym(args, 0, M_COPYALL, M_WAITOK)))) {
 445                 errp->re_status = stat = RPC_CANTENCODEARGS;
 446                 mtx_lock(&cs->cs_lock);
 447                 goto out;
 448         }
 449         mreq->m_pkthdr.len = m_length(mreq, NULL);
 450
 451         cr->cr_xid = xid;
 452         mtx_lock(&cs->cs_lock);
 453
 454         /*
 455          * Try to get a place in the congestion window.
 456          */
 457         while (cu->cu_sent >= cu->cu_cwnd) {
 458                 cu->cu_cwnd_wait = TRUE;
 459                 error = msleep(&cu->cu_cwnd_wait, &cs->cs_lock,
 460                     cu->cu_waitflag, "rpccwnd", 0);
 461                 if (error) {
 462                         errp->re_errno = error;
 463                         if (error == EINTR || error == ERESTART)
 464                                 errp->re_status = stat = RPC_INTR;
 465                         else
 466                                 errp->re_status = stat = RPC_CANTSEND;
 467                         goto out;
 468                 }
 469         }
 470         cu->cu_sent += CWNDSCALE;
 471
 472         TAILQ_INSERT_TAIL(&cs->cs_pending, cr, cr_link);
 473         mtx_unlock(&cs->cs_lock);
 474
 475         /*
 476          * sosend consumes mreq.
 477          */
 478         error = sosend(cu->cu_socket, sa, NULL, mreq, NULL, 0, curthread);
 479         mreq = NULL;
 480
 481         /*
 482          * sub-optimal code appears here because we have
 483          * some clock time to spare while the packets are in flight.
 484          * (We assume that this is actually only executed once.)
 485          */
 486         reply_msg.acpted_rply.ar_verf.oa_flavor = AUTH_NULL;
 487         reply_msg.acpted_rply.ar_verf.oa_base = cr->cr_verf;
 488         reply_msg.acpted_rply.ar_verf.oa_length = 0;
 489         reply_msg.acpted_rply.ar_results.where = NULL;
 490         reply_msg.acpted_rply.ar_results.proc = (xdrproc_t)xdr_void;
 491
 492         mtx_lock(&cs->cs_lock);
 493         if (error) {
 494                 TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
 495                 errp->re_errno = error;
 496                 errp->re_status = stat = RPC_CANTSEND;
 497                 cu->cu_sent -= CWNDSCALE;
 498                 if (cu->cu_cwnd_wait) {
 499                         cu->cu_cwnd_wait = FALSE;
 500                         wakeup(&cu->cu_cwnd_wait);
 501                 }
 502                 goto out;
 503         }
 504
 505         /*
 506          * Check to see if we got an upcall while waiting for the
 507          * lock.
 508          */
 509         if (cr->cr_error) {
 510                 TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
 511                 errp->re_errno = cr->cr_error;
 512                 errp->re_status = stat = RPC_CANTRECV;
 513                 cu->cu_sent -= CWNDSCALE;
 514                 if (cu->cu_cwnd_wait) {
 515                         cu->cu_cwnd_wait = FALSE;
 516                         wakeup(&cu->cu_cwnd_wait);
 517                 }
 518                 goto out;
 519         }
 520         if (cr->cr_mrep) {
 521                 TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
 522                 cu->cu_sent -= CWNDSCALE;
 523                 if (cu->cu_cwnd_wait) {
 524                         cu->cu_cwnd_wait = FALSE;
 525                         wakeup(&cu->cu_cwnd_wait);
 526                 }
 527                 goto got_reply;
 528         }
 529
 530         /*
 531          * Hack to provide rpc-based message passing
 532          */
 533         if (timeout == 0) {
 534                 TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
 535                 errp->re_status = stat = RPC_TIMEDOUT;
 536                 cu->cu_sent -= CWNDSCALE;
 537                 if (cu->cu_cwnd_wait) {
 538                         cu->cu_cwnd_wait = FALSE;
 539                         wakeup(&cu->cu_cwnd_wait);
 540                 }
 541                 goto out;
 542         }
 543
 544 get_reply:
 545         for (;;) {
 546                 /* Decide how long to wait. */
 547                 if (next_sendtime < timeout)
 548                         tv = next_sendtime;
 549                 else
 550                         tv = timeout;
 551                 tv -= time_waited;
 552
 553                 if (tv > 0) {
 554                         if (cu->cu_closing || cu->cu_closed) {
 555                                 error = 0;
 556                                 cr->cr_error = ESHUTDOWN;
 557                         } else {
 558                                 error = msleep(cr, &cs->cs_lock,
 559                                     cu->cu_waitflag, cu->cu_waitchan, tv);
 560                         }
 561                 } else {
 562                         error = EWOULDBLOCK;
 563                 }
 564
 565                 TAILQ_REMOVE(&cs->cs_pending, cr, cr_link);
 566                 cu->cu_sent -= CWNDSCALE;
 567                 if (cu->cu_cwnd_wait) {
 568                         cu->cu_cwnd_wait = FALSE;
 569                         wakeup(&cu->cu_cwnd_wait);
 570                 }
 571
 572                 if (!error) {
 573                         /*
 574                          * We were woken up by the upcall.  If the
 575                          * upcall had a receive error, report that,
 576                          * otherwise we have a reply.
 577                          */
 578                         if (cr->cr_error) {
 579                                 errp->re_errno = cr->cr_error;
 580                                 errp->re_status = stat = RPC_CANTRECV;
 581                                 goto out;
 582                         }
 583
 584                         cu->cu_cwnd += (CWNDSCALE * CWNDSCALE
 585                             + cu->cu_cwnd / 2) / cu->cu_cwnd;
 586                         if (cu->cu_cwnd > MAXCWND)
 587                                 cu->cu_cwnd = MAXCWND;
 588
 589                         if (rt) {
 590                                 /*
 591                                  * Add one to the time since a tick
 592                                  * count of N means that the actual
 593                                  * time taken was somewhere between N
 594                                  * and N+1.
 595                                  */
 596                                 rtt = ticks - starttime + 1;
 597
 598                                 /*
 599                                  * Update our estimate of the round
 600                                  * trip time using roughly the
 601                                  * algorithm described in RFC
 602                                  * 2988. Given an RTT sample R:
 603                                  *
 604                                  * RTTVAR = (1-beta) * RTTVAR + beta * |SRTT-R|
 605                                  * SRTT = (1-alpha) * SRTT + alpha * R
 606                                  *
 607                                  * where alpha = 0.125 and beta = 0.25.
 608                                  *
 609                                  * The initial retransmit timeout is
 610                                  * SRTT + 4*RTTVAR and doubles on each
 611                                  * retransmision.
 612                                  */
 613                                 if (rt->rt_srtt == 0) {
 614                                         rt->rt_srtt = rtt;
 615                                         rt->rt_deviate = rtt / 2;
 616                                 } else {
 617                                         int32_t error = rtt - rt->rt_srtt;
 618                                         rt->rt_srtt += error / 8;
 619                                         error = abs(error) - rt->rt_deviate;
 620                                         rt->rt_deviate += error / 4;
 621                                 }
 622                                 rt->rt_rtxcur = rt->rt_srtt + 4*rt->rt_deviate;
 623                         }
 624
 625                         break;
 626                 }
 627
 628                 /*
 629                  * The sleep returned an error so our request is still
 630                  * on the list. If we got EWOULDBLOCK, we may want to
 631                  * re-send the request.
 632                  */
 633                 if (error != EWOULDBLOCK) {
 634                         errp->re_errno = error;
 635                         if (error == EINTR || error == ERESTART)
 636                                 errp->re_status = stat = RPC_INTR;
 637                         else
 638                                 errp->re_status = stat = RPC_CANTRECV;
 639                         goto out;
 640                 }
 641
 642                 time_waited = ticks - starttime;
 643
 644                 /* Check for timeout. */
 645                 if (time_waited > timeout) {
 646                         errp->re_errno = EWOULDBLOCK;
 647                         errp->re_status = stat = RPC_TIMEDOUT;
 648                         goto out;
 649                 }
 650
 651                 /* Retransmit if necessary. */
 652                 if (time_waited >= next_sendtime) {
 653                         cu->cu_cwnd /= 2;
 654                         if (cu->cu_cwnd < CWNDSCALE)
 655                                 cu->cu_cwnd = CWNDSCALE;
 656                         if (ext && ext->rc_feedback) {
 657                                 mtx_unlock(&cs->cs_lock);
 658                                 if (retrans == 0)
 659                                         ext->rc_feedback(FEEDBACK_REXMIT1,
 660                                             proc, ext->rc_feedback_arg);
 661                                 else
 662                                         ext->rc_feedback(FEEDBACK_REXMIT2,
 663                                             proc, ext->rc_feedback_arg);
 664                                 mtx_lock(&cs->cs_lock);
 665                         }
 666                         if (cu->cu_closing || cu->cu_closed) {
 667                                 errp->re_errno = ESHUTDOWN;
 668                                 errp->re_status = stat = RPC_CANTRECV;
 669                                 goto out;
 670                         }
 671                         retrans++;
 672                         /* update retransmit_time */
 673                         if (retransmit_time < RPC_MAX_BACKOFF * hz)
 674                                 retransmit_time = 2 * retransmit_time;
 675                         next_sendtime += retransmit_time;
 676                         goto send_again;
 677                 }
 678                 cu->cu_sent += CWNDSCALE;
 679                 TAILQ_INSERT_TAIL(&cs->cs_pending, cr, cr_link);
 680         }
 681
 682 got_reply:
 683         /*
 684          * Now decode and validate the response. We need to drop the
 685          * lock since xdr_replymsg may end up sleeping in malloc.
 686          */
 687         mtx_unlock(&cs->cs_lock);
 688
 689         if (ext && ext->rc_feedback)
 690                 ext->rc_feedback(FEEDBACK_OK, proc, ext->rc_feedback_arg);
 691
 692         xdrmbuf_create(&xdrs, cr->cr_mrep, XDR_DECODE);
 693         ok = xdr_replymsg(&xdrs, &reply_msg);
 694         cr->cr_mrep = NULL;
 695
 696         if (ok) {
 697                 if ((reply_msg.rm_reply.rp_stat == MSG_ACCEPTED) &&
 698                     (reply_msg.acpted_rply.ar_stat == SUCCESS))
 699                         errp->re_status = stat = RPC_SUCCESS;
 700                 else
 701                         stat = _seterr_reply(&reply_msg, &(cu->cu_error));
 702
 703                 if (errp->re_status == RPC_SUCCESS) {
 704                         results = xdrmbuf_getall(&xdrs);
 705                         if (! AUTH_VALIDATE(auth, xid,
 706                                 &reply_msg.acpted_rply.ar_verf,
 707                                 &results)) {
 708                                 errp->re_status = stat = RPC_AUTHERROR;
 709                                 errp->re_why = AUTH_INVALIDRESP;
 710                                 if (retrans &&
 711                                     auth->ah_cred.oa_flavor == RPCSEC_GSS) {
 712                                         /*
 713                                          * If we retransmitted, its
 714                                          * possible that we will
 715                                          * receive a reply for one of
 716                                          * the earlier transmissions
 717                                          * (which will use an older
 718                                          * RPCSEC_GSS sequence
 719                                          * number). In this case, just
 720                                          * go back and listen for a
 721                                          * new reply. We could keep a
 722                                          * record of all the seq
 723                                          * numbers we have transmitted
 724                                          * so far so that we could
 725                                          * accept a reply for any of
 726                                          * them here.
 727                                          */
 728                                         XDR_DESTROY(&xdrs);
 729                                         mtx_lock(&cs->cs_lock);
 730                                         cu->cu_sent += CWNDSCALE;
 731                                         TAILQ_INSERT_TAIL(&cs->cs_pending,
 732                                             cr, cr_link);
 733                                         cr->cr_mrep = NULL;
 734                                         goto get_reply;
 735                                 }
 736                         } else {
 737                                 *resultsp = results;
 738                         }
 739                 }               /* end successful completion */
 740                 /*
 741                  * If unsuccessful AND error is an authentication error
 742                  * then refresh credentials and try again, else break
 743                  */
 744                 else if (stat == RPC_AUTHERROR)
 745                         /* maybe our credentials need to be refreshed ... */
 746                         if (nrefreshes > 0 &&
 747                             AUTH_REFRESH(auth, &reply_msg)) {
 748                                 nrefreshes--;
 749                                 XDR_DESTROY(&xdrs);
 750                                 mtx_lock(&cs->cs_lock);
 751                                 goto call_again;
 752                         }
 753                 /* end of unsuccessful completion */
 754         }       /* end of valid reply message */
 755         else {
 756                 errp->re_status = stat = RPC_CANTDECODERES;
 757
 758         }
 759         XDR_DESTROY(&xdrs);
 760         mtx_lock(&cs->cs_lock);
 761 out:
 762         mtx_assert(&cs->cs_lock, MA_OWNED);
 763
 764         if (mreq)
 765                 m_freem(mreq);
 766         if (cr->cr_mrep)
 767                 m_freem(cr->cr_mrep);
 768
 769         cu->cu_threads--;
 770         if (cu->cu_closing)
 771                 wakeup(cu);
 772
 773         mtx_unlock(&cs->cs_lock);
 774
 775         if (auth && stat != RPC_SUCCESS)
 776                 AUTH_VALIDATE(auth, xid, NULL, NULL);
 777
 778         free(cr, M_RPC);
 779
 780         return (stat);
 781 }
 782
 783 static void
 784 clnt_dg_geterr(CLIENT *cl, struct rpc_err *errp)
 785 {
 786         struct cu_data *cu = (struct cu_data *)cl->cl_private;
 787
 788         *errp = cu->cu_error;
 789 }
 790
 791 static bool_t
 792 clnt_dg_freeres(CLIENT *cl, xdrproc_t xdr_res, void *res_ptr)
 793 {
 794         XDR xdrs;
 795         bool_t dummy;
 796
 797         xdrs.x_op = XDR_FREE;
 798         dummy = (*xdr_res)(&xdrs, res_ptr);
 799
 800         return (dummy);
 801 }
 802
 803 /*ARGSUSED*/
 804 static void
 805 clnt_dg_abort(CLIENT *h)
 806 {
 807 }
 808
 809 static bool_t
 810 clnt_dg_control(CLIENT *cl, u_int request, void *info)
 811 {
 812         struct cu_data *cu = (struct cu_data *)cl->cl_private;
 813         struct cu_socket *cs;
 814         struct sockaddr *addr;
 815
 816         cs = cu->cu_socket->so_rcv.sb_upcallarg;
 817         mtx_lock(&cs->cs_lock);
 818
 819         switch (request) {
 820         case CLSET_FD_CLOSE:
 821                 cu->cu_closeit = TRUE;
 822                 mtx_unlock(&cs->cs_lock);
 823                 return (TRUE);
 824         case CLSET_FD_NCLOSE:
 825                 cu->cu_closeit = FALSE;
 826                 mtx_unlock(&cs->cs_lock);
 827                 return (TRUE);
 828         }
 829
 830         /* for other requests which use info */
 831         if (info == NULL) {
 832                 mtx_unlock(&cs->cs_lock);
 833                 return (FALSE);
 834         }
 835         switch (request) {
 836         case CLSET_TIMEOUT:
 837                 if (time_not_ok((struct timeval *)info)) {
 838                         mtx_unlock(&cs->cs_lock);
 839                         return (FALSE);
 840                 }
 841                 cu->cu_total = *(struct timeval *)info;
 842                 break;
 843         case CLGET_TIMEOUT:
 844                 *(struct timeval *)info = cu->cu_total;
 845                 break;
 846         case CLSET_RETRY_TIMEOUT:
 847                 if (time_not_ok((struct timeval *)info)) {
 848                         mtx_unlock(&cs->cs_lock);
 849                         return (FALSE);
 850                 }
 851                 cu->cu_wait = *(struct timeval *)info;
 852                 break;
 853         case CLGET_RETRY_TIMEOUT:
 854                 *(struct timeval *)info = cu->cu_wait;
 855                 break;
 856         case CLGET_SVC_ADDR:
 857                 /*
 858                  * Slightly different semantics to userland - we use
 859                  * sockaddr instead of netbuf.
 860                  */
 861                 memcpy(info, &cu->cu_raddr, cu->cu_raddr.ss_len);
 862                 break;
 863         case CLSET_SVC_ADDR:            /* set to new address */
 864                 addr = (struct sockaddr *)info;
 865                 (void) memcpy(&cu->cu_raddr, addr, addr->sa_len);
 866                 break;
 867         case CLGET_XID:
 868                 *(uint32_t *)info = cu->cu_xid;
 869                 break;
 870
 871         case CLSET_XID:
 872                 /* This will set the xid of the NEXT call */
 873                 /* decrement by 1 as clnt_dg_call() increments once */
 874                 cu->cu_xid = *(uint32_t *)info - 1;
 875                 break;
 876
 877         case CLGET_VERS:
 878                 /*
 879                  * This RELIES on the information that, in the call body,
 880                  * the version number field is the fifth field from the
 881                  * beginning of the RPC header. MUST be changed if the
 882                  * call_struct is changed
 883                  */
 884                 *(uint32_t *)info =
 885                     ntohl(*(uint32_t *)(void *)(cu->cu_mcallc +
 886                     4 * BYTES_PER_XDR_UNIT));
 887                 break;
 888
 889         case CLSET_VERS:
 890                 *(uint32_t *)(void *)(cu->cu_mcallc + 4 * BYTES_PER_XDR_UNIT)
 891                         = htonl(*(uint32_t *)info);
 892                 break;
 893
 894         case CLGET_PROG:
 895                 /*
 896                  * This RELIES on the information that, in the call body,
 897                  * the program number field is the fourth field from the
 898                  * beginning of the RPC header. MUST be changed if the
 899                  * call_struct is changed
 900                  */
 901                 *(uint32_t *)info =
 902                     ntohl(*(uint32_t *)(void *)(cu->cu_mcallc +
 903                     3 * BYTES_PER_XDR_UNIT));
 904                 break;
 905
 906         case CLSET_PROG:
 907                 *(uint32_t *)(void *)(cu->cu_mcallc + 3 * BYTES_PER_XDR_UNIT)
 908                         = htonl(*(uint32_t *)info);
 909                 break;
 910         case CLSET_ASYNC:
 911                 cu->cu_async = *(int *)info;
 912                 break;
 913         case CLSET_CONNECT:
 914                 cu->cu_connect = *(int *)info;
 915                 break;
 916         case CLSET_WAITCHAN:
 917                 cu->cu_waitchan = (const char *)info;
 918                 break;
 919         case CLGET_WAITCHAN:
 920                 *(const char **) info = cu->cu_waitchan;
 921                 break;
 922         case CLSET_INTERRUPTIBLE:
 923                 if (*(int *) info)
 924                         cu->cu_waitflag = PCATCH;
 925                 else
 926                         cu->cu_waitflag = 0;
 927                 break;
 928         case CLGET_INTERRUPTIBLE:
 929                 if (cu->cu_waitflag)
 930                         *(int *) info = TRUE;
 931                 else
 932                         *(int *) info = FALSE;
 933                 break;
 934         default:
 935                 mtx_unlock(&cs->cs_lock);
 936                 return (FALSE);
 937         }
 938         mtx_unlock(&cs->cs_lock);
 939         return (TRUE);
 940 }
 941
 942 static void
 943 clnt_dg_close(CLIENT *cl)
 944 {
 945         struct cu_data *cu = (struct cu_data *)cl->cl_private;
 946         struct cu_socket *cs;
 947         struct cu_request *cr;
 948
 949         cs = cu->cu_socket->so_rcv.sb_upcallarg;
 950         mtx_lock(&cs->cs_lock);
 951
 952         if (cu->cu_closed) {
 953                 mtx_unlock(&cs->cs_lock);
 954                 return;
 955         }
 956
 957         if (cu->cu_closing) {
 958                 while (cu->cu_closing)
 959                         msleep(cu, &cs->cs_lock, 0, "rpcclose", 0);
 960                 KASSERT(cu->cu_closed, ("client should be closed"));
 961                 mtx_unlock(&cs->cs_lock);
 962                 return;
 963         }
 964
 965         /*
 966          * Abort any pending requests and wait until everyone
 967          * has finished with clnt_vc_call.
 968          */
 969         cu->cu_closing = TRUE;
 970         TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
 971                 if (cr->cr_client == cl) {
 972                         cr->cr_xid = 0;
 973                         cr->cr_error = ESHUTDOWN;
 974                         wakeup(cr);
 975                 }
 976         }
 977
 978         while (cu->cu_threads)
 979                 msleep(cu, &cs->cs_lock, 0, "rpcclose", 0);
 980
 981         cu->cu_closing = FALSE;
 982         cu->cu_closed = TRUE;
 983
 984         mtx_unlock(&cs->cs_lock);
 985         wakeup(cu);
 986 }
 987
 988 static void
 989 clnt_dg_destroy(CLIENT *cl)
 990 {
 991         struct cu_data *cu = (struct cu_data *)cl->cl_private;
 992         struct cu_socket *cs;
 993         struct socket *so = NULL;
 994         bool_t lastsocketref;
 995
 996         cs = cu->cu_socket->so_rcv.sb_upcallarg;
 997         clnt_dg_close(cl);
 998
 999         SOCKBUF_LOCK(&cu->cu_socket->so_rcv);
1000         mtx_lock(&cs->cs_lock);
1001
1002         cs->cs_refs--;
1003         if (cs->cs_refs == 0) {
1004                 mtx_unlock(&cs->cs_lock);
1005                 soupcall_clear(cu->cu_socket, SO_RCV);
1006                 clnt_dg_upcallsdone(cu->cu_socket, cs);
1007                 SOCKBUF_UNLOCK(&cu->cu_socket->so_rcv);
1008                 mtx_destroy(&cs->cs_lock);
1009                 mem_free(cs, sizeof(*cs));
1010                 lastsocketref = TRUE;
1011         } else {
1012                 mtx_unlock(&cs->cs_lock);
1013                 SOCKBUF_UNLOCK(&cu->cu_socket->so_rcv);
1014                 lastsocketref = FALSE;
1015         }
1016
1017         if (cu->cu_closeit && lastsocketref) {
1018                 so = cu->cu_socket;
1019                 cu->cu_socket = NULL;
1020         }
1021
1022         if (so)
1023                 soclose(so);
1024
1025         if (cl->cl_netid && cl->cl_netid[0])
1026                 mem_free(cl->cl_netid, strlen(cl->cl_netid) +1);
1027         if (cl->cl_tp && cl->cl_tp[0])
1028                 mem_free(cl->cl_tp, strlen(cl->cl_tp) +1);
1029         mem_free(cu, sizeof (*cu));
1030         mem_free(cl, sizeof (CLIENT));
1031 }
1032
1033 /*
1034  * Make sure that the time is not garbage.  -1 value is allowed.
1035  */
1036 static bool_t
1037 time_not_ok(struct timeval *t)
1038 {
1039         return (t->tv_sec < -1 || t->tv_sec > 100000000 ||
1040                 t->tv_usec < -1 || t->tv_usec > 1000000);
1041 }
1042
1043 int
1044 clnt_dg_soupcall(struct socket *so, void *arg, int waitflag)
1045 {
1046         struct cu_socket *cs = (struct cu_socket *) arg;
1047         struct uio uio;
1048         struct mbuf *m;
1049         struct mbuf *control;
1050         struct cu_request *cr;
1051         int error, rcvflag, foundreq;
1052         uint32_t xid;
1053
1054         cs->cs_upcallrefs++;
1055         uio.uio_resid = 1000000000;
1056         uio.uio_td = curthread;
1057         do {
1058                 SOCKBUF_UNLOCK(&so->so_rcv);
1059                 m = NULL;
1060                 control = NULL;
1061                 rcvflag = MSG_DONTWAIT;
1062                 error = soreceive(so, NULL, &uio, &m, &control, &rcvflag);
1063                 if (control)
1064                         m_freem(control);
1065                 SOCKBUF_LOCK(&so->so_rcv);
1066
1067                 if (error == EWOULDBLOCK)
1068                         break;
1069
1070                 /*
1071                  * If there was an error, wake up all pending
1072                  * requests.
1073                  */
1074                 if (error) {
1075                         mtx_lock(&cs->cs_lock);
1076                         TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
1077                                 cr->cr_xid = 0;
1078                                 cr->cr_error = error;
1079                                 wakeup(cr);
1080                         }
1081                         mtx_unlock(&cs->cs_lock);
1082                         break;
1083                 }
1084
1085                 /*
1086                  * The XID is in the first uint32_t of the reply.
1087                  */
1088                 if (m->m_len < sizeof(xid) && m_length(m, NULL) < sizeof(xid)) {
1089                         /*
1090                          * Should never happen.
1091                          */
1092                         m_freem(m);
1093                         continue;
1094                 }
1095
1096                 m_copydata(m, 0, sizeof(xid), (char *)&xid);
1097                 xid = ntohl(xid);
1098
1099                 /*
1100                  * Attempt to match this reply with a pending request.
1101                  */
1102                 mtx_lock(&cs->cs_lock);
1103                 foundreq = 0;
1104                 TAILQ_FOREACH(cr, &cs->cs_pending, cr_link) {
1105                         if (cr->cr_xid == xid) {
1106                                 /*
1107                                  * This one matches. We leave the
1108                                  * reply mbuf in cr->cr_mrep. Set the
1109                                  * XID to zero so that we will ignore
1110                                  * any duplicated replies that arrive
1111                                  * before clnt_dg_call removes it from
1112                                  * the queue.
1113                                  */
1114                                 cr->cr_xid = 0;
1115                                 cr->cr_mrep = m;
1116                                 cr->cr_error = 0;
1117                                 foundreq = 1;
1118                                 wakeup(cr);
1119                                 break;
1120                         }
1121                 }
1122                 mtx_unlock(&cs->cs_lock);
1123
1124                 /*
1125                  * If we didn't find the matching request, just drop
1126                  * it - its probably a repeated reply.
1127                  */
1128                 if (!foundreq)
1129                         m_freem(m);
1130         } while (m);
1131         cs->cs_upcallrefs--;
1132         if (cs->cs_upcallrefs < 0)
1133                 panic("rpcdg upcall refcnt");
1134         if (cs->cs_upcallrefs == 0)
1135                 wakeup(&cs->cs_upcallrefs);
1136         return (SU_OK);
1137 }
1138
1139 /*
1140  * Wait for all upcalls in progress to complete.
1141  */
1142 static void
1143 clnt_dg_upcallsdone(struct socket *so, struct cu_socket *cs)
1144 {
1145
1146         SOCKBUF_LOCK_ASSERT(&so->so_rcv);
1147
1148         while (cs->cs_upcallrefs > 0)
1149                 (void) msleep(&cs->cs_upcallrefs, SOCKBUF_MTX(&so->so_rcv), 0,
1150                     "rpcdgup", 0);
1151 }