sys/rpc/clnt_dg.c

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