sys/rpc/svc.c

   1 /*      $NetBSD: svc.c,v 1.21 2000/07/06 03:10:35 christos 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 #if defined(LIBC_SCCS) && !defined(lint)
  34 static char *sccsid2 = "@(#)svc.c 1.44 88/02/08 Copyr 1984 Sun Micro";
  35 static char *sccsid = "@(#)svc.c        2.4 88/08/11 4.0 RPCSRC";
  36 #endif
  37 #include <sys/cdefs.h>
  38 __FBSDID("$FreeBSD$");
  39
  40 /*
  41  * svc.c, Server-side remote procedure call interface.
  42  *
  43  * There are two sets of procedures here.  The xprt routines are
  44  * for handling transport handles.  The svc routines handle the
  45  * list of service routines.
  46  *
  47  * Copyright (C) 1984, Sun Microsystems, Inc.
  48  */
  49
  50 #include <sys/param.h>
  51 #include <sys/lock.h>
  52 #include <sys/kernel.h>
  53 #include <sys/kthread.h>
  54 #include <sys/malloc.h>
  55 #include <sys/mbuf.h>
  56 #include <sys/mutex.h>
  57 #include <sys/proc.h>
  58 #include <sys/queue.h>
  59 #include <sys/socketvar.h>
  60 #include <sys/systm.h>
  61 #include <sys/smp.h>
  62 #include <sys/sx.h>
  63 #include <sys/ucred.h>
  64
  65 #include <rpc/rpc.h>
  66 #include <rpc/rpcb_clnt.h>
  67 #include <rpc/replay.h>
  68
  69 #include <rpc/rpc_com.h>
  70
  71 #define SVC_VERSQUIET 0x0001            /* keep quiet about vers mismatch */
  72 #define version_keepquiet(xp) (SVC_EXT(xp)->xp_flags & SVC_VERSQUIET)
  73
  74 static struct svc_callout *svc_find(SVCPOOL *pool, rpcprog_t, rpcvers_t,
  75     char *);
  76 static void svc_new_thread(SVCGROUP *grp);
  77 static void xprt_unregister_locked(SVCXPRT *xprt);
  78 static void svc_change_space_used(SVCPOOL *pool, long delta);
  79 static bool_t svc_request_space_available(SVCPOOL *pool);
  80 static void svcpool_cleanup(SVCPOOL *pool);
  81
  82 /* ***************  SVCXPRT related stuff **************** */
  83
  84 static int svcpool_minthread_sysctl(SYSCTL_HANDLER_ARGS);
  85 static int svcpool_maxthread_sysctl(SYSCTL_HANDLER_ARGS);
  86 static int svcpool_threads_sysctl(SYSCTL_HANDLER_ARGS);
  87
  88 SVCPOOL*
  89 svcpool_create(const char *name, struct sysctl_oid_list *sysctl_base)
  90 {
  91         SVCPOOL *pool;
  92         SVCGROUP *grp;
  93         int g;
  94
  95         pool = malloc(sizeof(SVCPOOL), M_RPC, M_WAITOK|M_ZERO);
  96
  97         mtx_init(&pool->sp_lock, "sp_lock", NULL, MTX_DEF);
  98         pool->sp_name = name;
  99         pool->sp_state = SVCPOOL_INIT;
 100         pool->sp_proc = NULL;
 101         TAILQ_INIT(&pool->sp_callouts);
 102         TAILQ_INIT(&pool->sp_lcallouts);
 103         pool->sp_minthreads = 1;
 104         pool->sp_maxthreads = 1;
 105         pool->sp_groupcount = 1;
 106         for (g = 0; g < SVC_MAXGROUPS; g++) {
 107                 grp = &pool->sp_groups[g];
 108                 mtx_init(&grp->sg_lock, "sg_lock", NULL, MTX_DEF);
 109                 grp->sg_pool = pool;
 110                 grp->sg_state = SVCPOOL_ACTIVE;
 111                 TAILQ_INIT(&grp->sg_xlist);
 112                 TAILQ_INIT(&grp->sg_active);
 113                 LIST_INIT(&grp->sg_idlethreads);
 114                 grp->sg_minthreads = 1;
 115                 grp->sg_maxthreads = 1;
 116         }
 117
 118         /*
 119          * Don't use more than a quarter of mbuf clusters.  Nota bene:
 120          * nmbclusters is an int, but nmbclusters*MCLBYTES may overflow
 121          * on LP64 architectures, so cast to u_long to avoid undefined
 122          * behavior.  (ILP32 architectures cannot have nmbclusters
 123          * large enough to overflow for other reasons.)
 124          */
 125         pool->sp_space_high = (u_long)nmbclusters * MCLBYTES / 4;
 126         pool->sp_space_low = (pool->sp_space_high / 3) * 2;
 127
 128         sysctl_ctx_init(&pool->sp_sysctl);
 129         if (sysctl_base) {
 130                 SYSCTL_ADD_PROC(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 131                     "minthreads", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
 132                     pool, 0, svcpool_minthread_sysctl, "I",
 133                     "Minimal number of threads");
 134                 SYSCTL_ADD_PROC(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 135                     "maxthreads", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
 136                     pool, 0, svcpool_maxthread_sysctl, "I",
 137                     "Maximal number of threads");
 138                 SYSCTL_ADD_PROC(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 139                     "threads", CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_MPSAFE,
 140                     pool, 0, svcpool_threads_sysctl, "I",
 141                     "Current number of threads");
 142                 SYSCTL_ADD_INT(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 143                     "groups", CTLFLAG_RD, &pool->sp_groupcount, 0,
 144                     "Number of thread groups");
 145
 146                 SYSCTL_ADD_ULONG(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 147                     "request_space_used", CTLFLAG_RD,
 148                     &pool->sp_space_used,
 149                     "Space in parsed but not handled requests.");
 150
 151                 SYSCTL_ADD_ULONG(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 152                     "request_space_used_highest", CTLFLAG_RD,
 153                     &pool->sp_space_used_highest,
 154                     "Highest space used since reboot.");
 155
 156                 SYSCTL_ADD_ULONG(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 157                     "request_space_high", CTLFLAG_RW,
 158                     &pool->sp_space_high,
 159                     "Maximum space in parsed but not handled requests.");
 160
 161                 SYSCTL_ADD_ULONG(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 162                     "request_space_low", CTLFLAG_RW,
 163                     &pool->sp_space_low,
 164                     "Low water mark for request space.");
 165
 166                 SYSCTL_ADD_INT(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 167                     "request_space_throttled", CTLFLAG_RD,
 168                     &pool->sp_space_throttled, 0,
 169                     "Whether nfs requests are currently throttled");
 170
 171                 SYSCTL_ADD_INT(&pool->sp_sysctl, sysctl_base, OID_AUTO,
 172                     "request_space_throttle_count", CTLFLAG_RD,
 173                     &pool->sp_space_throttle_count, 0,
 174                     "Count of times throttling based on request space has occurred");
 175         }
 176
 177         return pool;
 178 }
 179
 180 /*
 181  * Code common to svcpool_destroy() and svcpool_close(), which cleans up
 182  * the pool data structures.
 183  */
 184 static void
 185 svcpool_cleanup(SVCPOOL *pool)
 186 {
 187         SVCGROUP *grp;
 188         SVCXPRT *xprt, *nxprt;
 189         struct svc_callout *s;
 190         struct svc_loss_callout *sl;
 191         struct svcxprt_list cleanup;
 192         int g;
 193
 194         TAILQ_INIT(&cleanup);
 195
 196         for (g = 0; g < SVC_MAXGROUPS; g++) {
 197                 grp = &pool->sp_groups[g];
 198                 mtx_lock(&grp->sg_lock);
 199                 while ((xprt = TAILQ_FIRST(&grp->sg_xlist)) != NULL) {
 200                         xprt_unregister_locked(xprt);
 201                         TAILQ_INSERT_TAIL(&cleanup, xprt, xp_link);
 202                 }
 203                 mtx_unlock(&grp->sg_lock);
 204         }
 205         TAILQ_FOREACH_SAFE(xprt, &cleanup, xp_link, nxprt) {
 206                 SVC_RELEASE(xprt);
 207         }
 208
 209         mtx_lock(&pool->sp_lock);
 210         while ((s = TAILQ_FIRST(&pool->sp_callouts)) != NULL) {
 211                 mtx_unlock(&pool->sp_lock);
 212                 svc_unreg(pool, s->sc_prog, s->sc_vers);
 213                 mtx_lock(&pool->sp_lock);
 214         }
 215         while ((sl = TAILQ_FIRST(&pool->sp_lcallouts)) != NULL) {
 216                 mtx_unlock(&pool->sp_lock);
 217                 svc_loss_unreg(pool, sl->slc_dispatch);
 218                 mtx_lock(&pool->sp_lock);
 219         }
 220         mtx_unlock(&pool->sp_lock);
 221 }
 222
 223 void
 224 svcpool_destroy(SVCPOOL *pool)
 225 {
 226         SVCGROUP *grp;
 227         int g;
 228
 229         svcpool_cleanup(pool);
 230
 231         for (g = 0; g < SVC_MAXGROUPS; g++) {
 232                 grp = &pool->sp_groups[g];
 233                 mtx_destroy(&grp->sg_lock);
 234         }
 235         mtx_destroy(&pool->sp_lock);
 236
 237         if (pool->sp_rcache)
 238                 replay_freecache(pool->sp_rcache);
 239
 240         sysctl_ctx_free(&pool->sp_sysctl);
 241         free(pool, M_RPC);
 242 }
 243
 244 /*
 245  * Similar to svcpool_destroy(), except that it does not destroy the actual
 246  * data structures.  As such, "pool" may be used again.
 247  */
 248 void
 249 svcpool_close(SVCPOOL *pool)
 250 {
 251         SVCGROUP *grp;
 252         int g;
 253
 254         svcpool_cleanup(pool);
 255
 256         /* Now, initialize the pool's state for a fresh svc_run() call. */
 257         mtx_lock(&pool->sp_lock);
 258         pool->sp_state = SVCPOOL_INIT;
 259         mtx_unlock(&pool->sp_lock);
 260         for (g = 0; g < SVC_MAXGROUPS; g++) {
 261                 grp = &pool->sp_groups[g];
 262                 mtx_lock(&grp->sg_lock);
 263                 grp->sg_state = SVCPOOL_ACTIVE;
 264                 mtx_unlock(&grp->sg_lock);
 265         }
 266 }
 267
 268 /*
 269  * Sysctl handler to get the present thread count on a pool
 270  */
 271 static int
 272 svcpool_threads_sysctl(SYSCTL_HANDLER_ARGS)
 273 {
 274         SVCPOOL *pool;
 275         int threads, error, g;
 276
 277         pool = oidp->oid_arg1;
 278         threads = 0;
 279         mtx_lock(&pool->sp_lock);
 280         for (g = 0; g < pool->sp_groupcount; g++)
 281                 threads += pool->sp_groups[g].sg_threadcount;
 282         mtx_unlock(&pool->sp_lock);
 283         error = sysctl_handle_int(oidp, &threads, 0, req);
 284         return (error);
 285 }
 286
 287 /*
 288  * Sysctl handler to set the minimum thread count on a pool
 289  */
 290 static int
 291 svcpool_minthread_sysctl(SYSCTL_HANDLER_ARGS)
 292 {
 293         SVCPOOL *pool;
 294         int newminthreads, error, g;
 295
 296         pool = oidp->oid_arg1;
 297         newminthreads = pool->sp_minthreads;
 298         error = sysctl_handle_int(oidp, &newminthreads, 0, req);
 299         if (error == 0 && newminthreads != pool->sp_minthreads) {
 300                 if (newminthreads > pool->sp_maxthreads)
 301                         return (EINVAL);
 302                 mtx_lock(&pool->sp_lock);
 303                 pool->sp_minthreads = newminthreads;
 304                 for (g = 0; g < pool->sp_groupcount; g++) {
 305                         pool->sp_groups[g].sg_minthreads = max(1,
 306                             pool->sp_minthreads / pool->sp_groupcount);
 307                 }
 308                 mtx_unlock(&pool->sp_lock);
 309         }
 310         return (error);
 311 }
 312
 313 /*
 314  * Sysctl handler to set the maximum thread count on a pool
 315  */
 316 static int
 317 svcpool_maxthread_sysctl(SYSCTL_HANDLER_ARGS)
 318 {
 319         SVCPOOL *pool;
 320         int newmaxthreads, error, g;
 321
 322         pool = oidp->oid_arg1;
 323         newmaxthreads = pool->sp_maxthreads;
 324         error = sysctl_handle_int(oidp, &newmaxthreads, 0, req);
 325         if (error == 0 && newmaxthreads != pool->sp_maxthreads) {
 326                 if (newmaxthreads < pool->sp_minthreads)
 327                         return (EINVAL);
 328                 mtx_lock(&pool->sp_lock);
 329                 pool->sp_maxthreads = newmaxthreads;
 330                 for (g = 0; g < pool->sp_groupcount; g++) {
 331                         pool->sp_groups[g].sg_maxthreads = max(1,
 332                             pool->sp_maxthreads / pool->sp_groupcount);
 333                 }
 334                 mtx_unlock(&pool->sp_lock);
 335         }
 336         return (error);
 337 }
 338
 339 /*
 340  * Activate a transport handle.
 341  */
 342 void
 343 xprt_register(SVCXPRT *xprt)
 344 {
 345         SVCPOOL *pool = xprt->xp_pool;
 346         SVCGROUP *grp;
 347         int g;
 348
 349         SVC_ACQUIRE(xprt);
 350         g = atomic_fetchadd_int(&pool->sp_nextgroup, 1) % pool->sp_groupcount;
 351         xprt->xp_group = grp = &pool->sp_groups[g];
 352         mtx_lock(&grp->sg_lock);
 353         xprt->xp_registered = TRUE;
 354         xprt->xp_active = FALSE;
 355         TAILQ_INSERT_TAIL(&grp->sg_xlist, xprt, xp_link);
 356         mtx_unlock(&grp->sg_lock);
 357 }
 358
 359 /*
 360  * De-activate a transport handle. Note: the locked version doesn't
 361  * release the transport - caller must do that after dropping the pool
 362  * lock.
 363  */
 364 static void
 365 xprt_unregister_locked(SVCXPRT *xprt)
 366 {
 367         SVCGROUP *grp = xprt->xp_group;
 368
 369         mtx_assert(&grp->sg_lock, MA_OWNED);
 370         KASSERT(xprt->xp_registered == TRUE,
 371             ("xprt_unregister_locked: not registered"));
 372         xprt_inactive_locked(xprt);
 373         TAILQ_REMOVE(&grp->sg_xlist, xprt, xp_link);
 374         xprt->xp_registered = FALSE;
 375 }
 376
 377 void
 378 xprt_unregister(SVCXPRT *xprt)
 379 {
 380         SVCGROUP *grp = xprt->xp_group;
 381
 382         mtx_lock(&grp->sg_lock);
 383         if (xprt->xp_registered == FALSE) {
 384                 /* Already unregistered by another thread */
 385                 mtx_unlock(&grp->sg_lock);
 386                 return;
 387         }
 388         xprt_unregister_locked(xprt);
 389         mtx_unlock(&grp->sg_lock);
 390
 391         SVC_RELEASE(xprt);
 392 }
 393
 394 /*
 395  * Attempt to assign a service thread to this transport.
 396  */
 397 static int
 398 xprt_assignthread(SVCXPRT *xprt)
 399 {
 400         SVCGROUP *grp = xprt->xp_group;
 401         SVCTHREAD *st;
 402
 403         mtx_assert(&grp->sg_lock, MA_OWNED);
 404         st = LIST_FIRST(&grp->sg_idlethreads);
 405         if (st) {
 406                 LIST_REMOVE(st, st_ilink);
 407                 SVC_ACQUIRE(xprt);
 408                 xprt->xp_thread = st;
 409                 st->st_xprt = xprt;
 410                 cv_signal(&st->st_cond);
 411                 return (TRUE);
 412         } else {
 413                 /*
 414                  * See if we can create a new thread. The
 415                  * actual thread creation happens in
 416                  * svc_run_internal because our locking state
 417                  * is poorly defined (we are typically called
 418                  * from a socket upcall). Don't create more
 419                  * than one thread per second.
 420                  */
 421                 if (grp->sg_state == SVCPOOL_ACTIVE
 422                     && grp->sg_lastcreatetime < time_uptime
 423                     && grp->sg_threadcount < grp->sg_maxthreads) {
 424                         grp->sg_state = SVCPOOL_THREADWANTED;
 425                 }
 426         }
 427         return (FALSE);
 428 }
 429
 430 void
 431 xprt_active(SVCXPRT *xprt)
 432 {
 433         SVCGROUP *grp = xprt->xp_group;
 434
 435         mtx_lock(&grp->sg_lock);
 436
 437         if (!xprt->xp_registered) {
 438                 /*
 439                  * Race with xprt_unregister - we lose.
 440                  */
 441                 mtx_unlock(&grp->sg_lock);
 442                 return;
 443         }
 444
 445         if (!xprt->xp_active) {
 446                 xprt->xp_active = TRUE;
 447                 if (xprt->xp_thread == NULL) {
 448                         if (!svc_request_space_available(xprt->xp_pool) ||
 449                             !xprt_assignthread(xprt))
 450                                 TAILQ_INSERT_TAIL(&grp->sg_active, xprt,
 451                                     xp_alink);
 452                 }
 453         }
 454
 455         mtx_unlock(&grp->sg_lock);
 456 }
 457
 458 void
 459 xprt_inactive_locked(SVCXPRT *xprt)
 460 {
 461         SVCGROUP *grp = xprt->xp_group;
 462
 463         mtx_assert(&grp->sg_lock, MA_OWNED);
 464         if (xprt->xp_active) {
 465                 if (xprt->xp_thread == NULL)
 466                         TAILQ_REMOVE(&grp->sg_active, xprt, xp_alink);
 467                 xprt->xp_active = FALSE;
 468         }
 469 }
 470
 471 void
 472 xprt_inactive(SVCXPRT *xprt)
 473 {
 474         SVCGROUP *grp = xprt->xp_group;
 475
 476         mtx_lock(&grp->sg_lock);
 477         xprt_inactive_locked(xprt);
 478         mtx_unlock(&grp->sg_lock);
 479 }
 480
 481 /*
 482  * Variant of xprt_inactive() for use only when sure that port is
 483  * assigned to thread. For example, within receive handlers.
 484  */
 485 void
 486 xprt_inactive_self(SVCXPRT *xprt)
 487 {
 488
 489         KASSERT(xprt->xp_thread != NULL,
 490             ("xprt_inactive_self(%p) with NULL xp_thread", xprt));
 491         xprt->xp_active = FALSE;
 492 }
 493
 494 /*
 495  * Add a service program to the callout list.
 496  * The dispatch routine will be called when a rpc request for this
 497  * program number comes in.
 498  */
 499 bool_t
 500 svc_reg(SVCXPRT *xprt, const rpcprog_t prog, const rpcvers_t vers,
 501     void (*dispatch)(struct svc_req *, SVCXPRT *),
 502     const struct netconfig *nconf)
 503 {
 504         SVCPOOL *pool = xprt->xp_pool;
 505         struct svc_callout *s;
 506         char *netid = NULL;
 507         int flag = 0;
 508
 509 /* VARIABLES PROTECTED BY svc_lock: s, svc_head */
 510
 511         if (xprt->xp_netid) {
 512                 netid = strdup(xprt->xp_netid, M_RPC);
 513                 flag = 1;
 514         } else if (nconf && nconf->nc_netid) {
 515                 netid = strdup(nconf->nc_netid, M_RPC);
 516                 flag = 1;
 517         } /* must have been created with svc_raw_create */
 518         if ((netid == NULL) && (flag == 1)) {
 519                 return (FALSE);
 520         }
 521
 522         mtx_lock(&pool->sp_lock);
 523         if ((s = svc_find(pool, prog, vers, netid)) != NULL) {
 524                 if (netid)
 525                         free(netid, M_RPC);
 526                 if (s->sc_dispatch == dispatch)
 527                         goto rpcb_it; /* he is registering another xptr */
 528                 mtx_unlock(&pool->sp_lock);
 529                 return (FALSE);
 530         }
 531         s = malloc(sizeof (struct svc_callout), M_RPC, M_NOWAIT);
 532         if (s == NULL) {
 533                 if (netid)
 534                         free(netid, M_RPC);
 535                 mtx_unlock(&pool->sp_lock);
 536                 return (FALSE);
 537         }
 538
 539         s->sc_prog = prog;
 540         s->sc_vers = vers;
 541         s->sc_dispatch = dispatch;
 542         s->sc_netid = netid;
 543         TAILQ_INSERT_TAIL(&pool->sp_callouts, s, sc_link);
 544
 545         if ((xprt->xp_netid == NULL) && (flag == 1) && netid)
 546                 ((SVCXPRT *) xprt)->xp_netid = strdup(netid, M_RPC);
 547
 548 rpcb_it:
 549         mtx_unlock(&pool->sp_lock);
 550         /* now register the information with the local binder service */
 551         if (nconf) {
 552                 bool_t dummy;
 553                 struct netconfig tnc;
 554                 struct netbuf nb;
 555                 tnc = *nconf;
 556                 nb.buf = &xprt->xp_ltaddr;
 557                 nb.len = xprt->xp_ltaddr.ss_len;
 558                 dummy = rpcb_set(prog, vers, &tnc, &nb);
 559                 return (dummy);
 560         }
 561         return (TRUE);
 562 }
 563
 564 /*
 565  * Remove a service program from the callout list.
 566  */
 567 void
 568 svc_unreg(SVCPOOL *pool, const rpcprog_t prog, const rpcvers_t vers)
 569 {
 570         struct svc_callout *s;
 571
 572         /* unregister the information anyway */
 573         (void) rpcb_unset(prog, vers, NULL);
 574         mtx_lock(&pool->sp_lock);
 575         while ((s = svc_find(pool, prog, vers, NULL)) != NULL) {
 576                 TAILQ_REMOVE(&pool->sp_callouts, s, sc_link);
 577                 if (s->sc_netid)
 578                         mem_free(s->sc_netid, sizeof (s->sc_netid) + 1);
 579                 mem_free(s, sizeof (struct svc_callout));
 580         }
 581         mtx_unlock(&pool->sp_lock);
 582 }
 583
 584 /*
 585  * Add a service connection loss program to the callout list.
 586  * The dispatch routine will be called when some port in ths pool die.
 587  */
 588 bool_t
 589 svc_loss_reg(SVCXPRT *xprt, void (*dispatch)(SVCXPRT *))
 590 {
 591         SVCPOOL *pool = xprt->xp_pool;
 592         struct svc_loss_callout *s;
 593
 594         mtx_lock(&pool->sp_lock);
 595         TAILQ_FOREACH(s, &pool->sp_lcallouts, slc_link) {
 596                 if (s->slc_dispatch == dispatch)
 597                         break;
 598         }
 599         if (s != NULL) {
 600                 mtx_unlock(&pool->sp_lock);
 601                 return (TRUE);
 602         }
 603         s = malloc(sizeof(struct svc_loss_callout), M_RPC, M_NOWAIT);
 604         if (s == NULL) {
 605                 mtx_unlock(&pool->sp_lock);
 606                 return (FALSE);
 607         }
 608         s->slc_dispatch = dispatch;
 609         TAILQ_INSERT_TAIL(&pool->sp_lcallouts, s, slc_link);
 610         mtx_unlock(&pool->sp_lock);
 611         return (TRUE);
 612 }
 613
 614 /*
 615  * Remove a service connection loss program from the callout list.
 616  */
 617 void
 618 svc_loss_unreg(SVCPOOL *pool, void (*dispatch)(SVCXPRT *))
 619 {
 620         struct svc_loss_callout *s;
 621
 622         mtx_lock(&pool->sp_lock);
 623         TAILQ_FOREACH(s, &pool->sp_lcallouts, slc_link) {
 624                 if (s->slc_dispatch == dispatch) {
 625                         TAILQ_REMOVE(&pool->sp_lcallouts, s, slc_link);
 626                         free(s, M_RPC);
 627                         break;
 628                 }
 629         }
 630         mtx_unlock(&pool->sp_lock);
 631 }
 632
 633 /* ********************** CALLOUT list related stuff ************* */
 634
 635 /*
 636  * Search the callout list for a program number, return the callout
 637  * struct.
 638  */
 639 static struct svc_callout *
 640 svc_find(SVCPOOL *pool, rpcprog_t prog, rpcvers_t vers, char *netid)
 641 {
 642         struct svc_callout *s;
 643
 644         mtx_assert(&pool->sp_lock, MA_OWNED);
 645         TAILQ_FOREACH(s, &pool->sp_callouts, sc_link) {
 646                 if (s->sc_prog == prog && s->sc_vers == vers
 647                     && (netid == NULL || s->sc_netid == NULL ||
 648                         strcmp(netid, s->sc_netid) == 0))
 649                         break;
 650         }
 651
 652         return (s);
 653 }
 654
 655 /* ******************* REPLY GENERATION ROUTINES  ************ */
 656
 657 static bool_t
 658 svc_sendreply_common(struct svc_req *rqstp, struct rpc_msg *rply,
 659     struct mbuf *body)
 660 {
 661         SVCXPRT *xprt = rqstp->rq_xprt;
 662         bool_t ok;
 663
 664         if (rqstp->rq_args) {
 665                 m_freem(rqstp->rq_args);
 666                 rqstp->rq_args = NULL;
 667         }
 668
 669         if (xprt->xp_pool->sp_rcache)
 670                 replay_setreply(xprt->xp_pool->sp_rcache,
 671                     rply, svc_getrpccaller(rqstp), body);
 672
 673         if (!SVCAUTH_WRAP(&rqstp->rq_auth, &body))
 674                 return (FALSE);
 675
 676         ok = SVC_REPLY(xprt, rply, rqstp->rq_addr, body, &rqstp->rq_reply_seq);
 677         if (rqstp->rq_addr) {
 678                 free(rqstp->rq_addr, M_SONAME);
 679                 rqstp->rq_addr = NULL;
 680         }
 681
 682         return (ok);
 683 }
 684
 685 /*
 686  * Send a reply to an rpc request
 687  */
 688 bool_t
 689 svc_sendreply(struct svc_req *rqstp, xdrproc_t xdr_results, void * xdr_location)
 690 {
 691         struct rpc_msg rply;
 692         struct mbuf *m;
 693         XDR xdrs;
 694         bool_t ok;
 695
 696         rply.rm_xid = rqstp->rq_xid;
 697         rply.rm_direction = REPLY;
 698         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 699         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 700         rply.acpted_rply.ar_stat = SUCCESS;
 701         rply.acpted_rply.ar_results.where = NULL;
 702         rply.acpted_rply.ar_results.proc = (xdrproc_t) xdr_void;
 703
 704         m = m_getcl(M_WAITOK, MT_DATA, 0);
 705         xdrmbuf_create(&xdrs, m, XDR_ENCODE);
 706         ok = xdr_results(&xdrs, xdr_location);
 707         XDR_DESTROY(&xdrs);
 708
 709         if (ok) {
 710                 return (svc_sendreply_common(rqstp, &rply, m));
 711         } else {
 712                 m_freem(m);
 713                 return (FALSE);
 714         }
 715 }
 716
 717 bool_t
 718 svc_sendreply_mbuf(struct svc_req *rqstp, struct mbuf *m)
 719 {
 720         struct rpc_msg rply;
 721
 722         rply.rm_xid = rqstp->rq_xid;
 723         rply.rm_direction = REPLY;
 724         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 725         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 726         rply.acpted_rply.ar_stat = SUCCESS;
 727         rply.acpted_rply.ar_results.where = NULL;
 728         rply.acpted_rply.ar_results.proc = (xdrproc_t) xdr_void;
 729
 730         return (svc_sendreply_common(rqstp, &rply, m));
 731 }
 732
 733 /*
 734  * No procedure error reply
 735  */
 736 void
 737 svcerr_noproc(struct svc_req *rqstp)
 738 {
 739         SVCXPRT *xprt = rqstp->rq_xprt;
 740         struct rpc_msg rply;
 741
 742         rply.rm_xid = rqstp->rq_xid;
 743         rply.rm_direction = REPLY;
 744         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 745         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 746         rply.acpted_rply.ar_stat = PROC_UNAVAIL;
 747
 748         if (xprt->xp_pool->sp_rcache)
 749                 replay_setreply(xprt->xp_pool->sp_rcache,
 750                     &rply, svc_getrpccaller(rqstp), NULL);
 751
 752         svc_sendreply_common(rqstp, &rply, NULL);
 753 }
 754
 755 /*
 756  * Can't decode args error reply
 757  */
 758 void
 759 svcerr_decode(struct svc_req *rqstp)
 760 {
 761         SVCXPRT *xprt = rqstp->rq_xprt;
 762         struct rpc_msg rply;
 763
 764         rply.rm_xid = rqstp->rq_xid;
 765         rply.rm_direction = REPLY;
 766         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 767         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 768         rply.acpted_rply.ar_stat = GARBAGE_ARGS;
 769
 770         if (xprt->xp_pool->sp_rcache)
 771                 replay_setreply(xprt->xp_pool->sp_rcache,
 772                     &rply, (struct sockaddr *) &xprt->xp_rtaddr, NULL);
 773
 774         svc_sendreply_common(rqstp, &rply, NULL);
 775 }
 776
 777 /*
 778  * Some system error
 779  */
 780 void
 781 svcerr_systemerr(struct svc_req *rqstp)
 782 {
 783         SVCXPRT *xprt = rqstp->rq_xprt;
 784         struct rpc_msg rply;
 785
 786         rply.rm_xid = rqstp->rq_xid;
 787         rply.rm_direction = REPLY;
 788         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 789         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 790         rply.acpted_rply.ar_stat = SYSTEM_ERR;
 791
 792         if (xprt->xp_pool->sp_rcache)
 793                 replay_setreply(xprt->xp_pool->sp_rcache,
 794                     &rply, svc_getrpccaller(rqstp), NULL);
 795
 796         svc_sendreply_common(rqstp, &rply, NULL);
 797 }
 798
 799 /*
 800  * Authentication error reply
 801  */
 802 void
 803 svcerr_auth(struct svc_req *rqstp, enum auth_stat why)
 804 {
 805         SVCXPRT *xprt = rqstp->rq_xprt;
 806         struct rpc_msg rply;
 807
 808         rply.rm_xid = rqstp->rq_xid;
 809         rply.rm_direction = REPLY;
 810         rply.rm_reply.rp_stat = MSG_DENIED;
 811         rply.rjcted_rply.rj_stat = AUTH_ERROR;
 812         rply.rjcted_rply.rj_why = why;
 813
 814         if (xprt->xp_pool->sp_rcache)
 815                 replay_setreply(xprt->xp_pool->sp_rcache,
 816                     &rply, svc_getrpccaller(rqstp), NULL);
 817
 818         svc_sendreply_common(rqstp, &rply, NULL);
 819 }
 820
 821 /*
 822  * Auth too weak error reply
 823  */
 824 void
 825 svcerr_weakauth(struct svc_req *rqstp)
 826 {
 827
 828         svcerr_auth(rqstp, AUTH_TOOWEAK);
 829 }
 830
 831 /*
 832  * Program unavailable error reply
 833  */
 834 void
 835 svcerr_noprog(struct svc_req *rqstp)
 836 {
 837         SVCXPRT *xprt = rqstp->rq_xprt;
 838         struct rpc_msg rply;
 839
 840         rply.rm_xid = rqstp->rq_xid;
 841         rply.rm_direction = REPLY;
 842         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 843         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 844         rply.acpted_rply.ar_stat = PROG_UNAVAIL;
 845
 846         if (xprt->xp_pool->sp_rcache)
 847                 replay_setreply(xprt->xp_pool->sp_rcache,
 848                     &rply, svc_getrpccaller(rqstp), NULL);
 849
 850         svc_sendreply_common(rqstp, &rply, NULL);
 851 }
 852
 853 /*
 854  * Program version mismatch error reply
 855  */
 856 void
 857 svcerr_progvers(struct svc_req *rqstp, rpcvers_t low_vers, rpcvers_t high_vers)
 858 {
 859         SVCXPRT *xprt = rqstp->rq_xprt;
 860         struct rpc_msg rply;
 861
 862         rply.rm_xid = rqstp->rq_xid;
 863         rply.rm_direction = REPLY;
 864         rply.rm_reply.rp_stat = MSG_ACCEPTED;
 865         rply.acpted_rply.ar_verf = rqstp->rq_verf;
 866         rply.acpted_rply.ar_stat = PROG_MISMATCH;
 867         rply.acpted_rply.ar_vers.low = (uint32_t)low_vers;
 868         rply.acpted_rply.ar_vers.high = (uint32_t)high_vers;
 869
 870         if (xprt->xp_pool->sp_rcache)
 871                 replay_setreply(xprt->xp_pool->sp_rcache,
 872                     &rply, svc_getrpccaller(rqstp), NULL);
 873
 874         svc_sendreply_common(rqstp, &rply, NULL);
 875 }
 876
 877 /*
 878  * Allocate a new server transport structure. All fields are
 879  * initialized to zero and xp_p3 is initialized to point at an
 880  * extension structure to hold various flags and authentication
 881  * parameters.
 882  */
 883 SVCXPRT *
 884 svc_xprt_alloc(void)
 885 {
 886         SVCXPRT *xprt;
 887         SVCXPRT_EXT *ext;
 888
 889         xprt = mem_alloc(sizeof(SVCXPRT));
 890         ext = mem_alloc(sizeof(SVCXPRT_EXT));
 891         xprt->xp_p3 = ext;
 892         refcount_init(&xprt->xp_refs, 1);
 893
 894         return (xprt);
 895 }
 896
 897 /*
 898  * Free a server transport structure.
 899  */
 900 void
 901 svc_xprt_free(SVCXPRT *xprt)
 902 {
 903
 904         mem_free(xprt->xp_p3, sizeof(SVCXPRT_EXT));
 905         /* The size argument is ignored, so 0 is ok. */
 906         mem_free(xprt->xp_gidp, 0);
 907         mem_free(xprt, sizeof(SVCXPRT));
 908 }
 909
 910 /* ******************* SERVER INPUT STUFF ******************* */
 911
 912 /*
 913  * Read RPC requests from a transport and queue them to be
 914  * executed. We handle authentication and replay cache replies here.
 915  * Actually dispatching the RPC is deferred till svc_executereq.
 916  */
 917 static enum xprt_stat
 918 svc_getreq(SVCXPRT *xprt, struct svc_req **rqstp_ret)
 919 {
 920         SVCPOOL *pool = xprt->xp_pool;
 921         struct svc_req *r;
 922         struct rpc_msg msg;
 923         struct mbuf *args;
 924         struct svc_loss_callout *s;
 925         enum xprt_stat stat;
 926
 927         /* now receive msgs from xprtprt (support batch calls) */
 928         r = malloc(sizeof(*r), M_RPC, M_WAITOK|M_ZERO);
 929
 930         msg.rm_call.cb_cred.oa_base = r->rq_credarea;
 931         msg.rm_call.cb_verf.oa_base = &r->rq_credarea[MAX_AUTH_BYTES];
 932         r->rq_clntcred = &r->rq_credarea[2*MAX_AUTH_BYTES];
 933         if (SVC_RECV(xprt, &msg, &r->rq_addr, &args)) {
 934                 enum auth_stat why;
 935
 936                 /*
 937                  * Handle replays and authenticate before queuing the
 938                  * request to be executed.
 939                  */
 940                 SVC_ACQUIRE(xprt);
 941                 r->rq_xprt = xprt;
 942                 if (pool->sp_rcache) {
 943                         struct rpc_msg repmsg;
 944                         struct mbuf *repbody;
 945                         enum replay_state rs;
 946                         rs = replay_find(pool->sp_rcache, &msg,
 947                             svc_getrpccaller(r), &repmsg, &repbody);
 948                         switch (rs) {
 949                         case RS_NEW:
 950                                 break;
 951                         case RS_DONE:
 952                                 SVC_REPLY(xprt, &repmsg, r->rq_addr,
 953                                     repbody, &r->rq_reply_seq);
 954                                 if (r->rq_addr) {
 955                                         free(r->rq_addr, M_SONAME);
 956                                         r->rq_addr = NULL;
 957                                 }
 958                                 m_freem(args);
 959                                 goto call_done;
 960
 961                         default:
 962                                 m_freem(args);
 963                                 goto call_done;
 964                         }
 965                 }
 966
 967                 r->rq_xid = msg.rm_xid;
 968                 r->rq_prog = msg.rm_call.cb_prog;
 969                 r->rq_vers = msg.rm_call.cb_vers;
 970                 r->rq_proc = msg.rm_call.cb_proc;
 971                 r->rq_size = sizeof(*r) + m_length(args, NULL);
 972                 r->rq_args = args;
 973                 if ((why = _authenticate(r, &msg)) != AUTH_OK) {
 974                         /*
 975                          * RPCSEC_GSS uses this return code
 976                          * for requests that form part of its
 977                          * context establishment protocol and
 978                          * should not be dispatched to the
 979                          * application.
 980                          */
 981                         if (why != RPCSEC_GSS_NODISPATCH)
 982                                 svcerr_auth(r, why);
 983                         goto call_done;
 984                 }
 985
 986                 if (!SVCAUTH_UNWRAP(&r->rq_auth, &r->rq_args)) {
 987                         svcerr_decode(r);
 988                         goto call_done;
 989                 }
 990
 991                 /*
 992                  * Everything checks out, return request to caller.
 993                  */
 994                 *rqstp_ret = r;
 995                 r = NULL;
 996         }
 997 call_done:
 998         if (r) {
 999                 svc_freereq(r);
1000                 r = NULL;
1001         }
1002         if ((stat = SVC_STAT(xprt)) == XPRT_DIED) {
1003                 TAILQ_FOREACH(s, &pool->sp_lcallouts, slc_link)
1004                         (*s->slc_dispatch)(xprt);
1005                 xprt_unregister(xprt);
1006         }
1007
1008         return (stat);
1009 }
1010
1011 static void
1012 svc_executereq(struct svc_req *rqstp)
1013 {
1014         SVCXPRT *xprt = rqstp->rq_xprt;
1015         SVCPOOL *pool = xprt->xp_pool;
1016         int prog_found;
1017         rpcvers_t low_vers;
1018         rpcvers_t high_vers;
1019         struct svc_callout *s;
1020
1021         /* now match message with a registered service*/
1022         prog_found = FALSE;
1023         low_vers = (rpcvers_t) -1L;
1024         high_vers = (rpcvers_t) 0L;
1025         TAILQ_FOREACH(s, &pool->sp_callouts, sc_link) {
1026                 if (s->sc_prog == rqstp->rq_prog) {
1027                         if (s->sc_vers == rqstp->rq_vers) {
1028                                 /*
1029                                  * We hand ownership of r to the
1030                                  * dispatch method - they must call
1031                                  * svc_freereq.
1032                                  */
1033                                 (*s->sc_dispatch)(rqstp, xprt);
1034                                 return;
1035                         }  /* found correct version */
1036                         prog_found = TRUE;
1037                         if (s->sc_vers < low_vers)
1038                                 low_vers = s->sc_vers;
1039                         if (s->sc_vers > high_vers)
1040                                 high_vers = s->sc_vers;
1041                 }   /* found correct program */
1042         }
1043
1044         /*
1045          * if we got here, the program or version
1046          * is not served ...
1047          */
1048         if (prog_found)
1049                 svcerr_progvers(rqstp, low_vers, high_vers);
1050         else
1051                 svcerr_noprog(rqstp);
1052
1053         svc_freereq(rqstp);
1054 }
1055
1056 static void
1057 svc_checkidle(SVCGROUP *grp)
1058 {
1059         SVCXPRT *xprt, *nxprt;
1060         time_t timo;
1061         struct svcxprt_list cleanup;
1062
1063         TAILQ_INIT(&cleanup);
1064         TAILQ_FOREACH_SAFE(xprt, &grp->sg_xlist, xp_link, nxprt) {
1065                 /*
1066                  * Only some transports have idle timers. Don't time
1067                  * something out which is just waking up.
1068                  */
1069                 if (!xprt->xp_idletimeout || xprt->xp_thread)
1070                         continue;
1071
1072                 timo = xprt->xp_lastactive + xprt->xp_idletimeout;
1073                 if (time_uptime > timo) {
1074                         xprt_unregister_locked(xprt);
1075                         TAILQ_INSERT_TAIL(&cleanup, xprt, xp_link);
1076                 }
1077         }
1078
1079         mtx_unlock(&grp->sg_lock);
1080         TAILQ_FOREACH_SAFE(xprt, &cleanup, xp_link, nxprt) {
1081                 SVC_RELEASE(xprt);
1082         }
1083         mtx_lock(&grp->sg_lock);
1084 }
1085
1086 static void
1087 svc_assign_waiting_sockets(SVCPOOL *pool)
1088 {
1089         SVCGROUP *grp;
1090         SVCXPRT *xprt;
1091         int g;
1092
1093         for (g = 0; g < pool->sp_groupcount; g++) {
1094                 grp = &pool->sp_groups[g];
1095                 mtx_lock(&grp->sg_lock);
1096                 while ((xprt = TAILQ_FIRST(&grp->sg_active)) != NULL) {
1097                         if (xprt_assignthread(xprt))
1098                                 TAILQ_REMOVE(&grp->sg_active, xprt, xp_alink);
1099                         else
1100                                 break;
1101                 }
1102                 mtx_unlock(&grp->sg_lock);
1103         }
1104 }
1105
1106 static void
1107 svc_change_space_used(SVCPOOL *pool, long delta)
1108 {
1109         unsigned long value;
1110
1111         value = atomic_fetchadd_long(&pool->sp_space_used, delta) + delta;
1112         if (delta > 0) {
1113                 if (value >= pool->sp_space_high && !pool->sp_space_throttled) {
1114                         pool->sp_space_throttled = TRUE;
1115                         pool->sp_space_throttle_count++;
1116                 }
1117                 if (value > pool->sp_space_used_highest)
1118                         pool->sp_space_used_highest = value;
1119         } else {
1120                 if (value < pool->sp_space_low && pool->sp_space_throttled) {
1121                         pool->sp_space_throttled = FALSE;
1122                         svc_assign_waiting_sockets(pool);
1123                 }
1124         }
1125 }
1126
1127 static bool_t
1128 svc_request_space_available(SVCPOOL *pool)
1129 {
1130
1131         if (pool->sp_space_throttled)
1132                 return (FALSE);
1133         return (TRUE);
1134 }
1135
1136 static void
1137 svc_run_internal(SVCGROUP *grp, bool_t ismaster)
1138 {
1139         SVCPOOL *pool = grp->sg_pool;
1140         SVCTHREAD *st, *stpref;
1141         SVCXPRT *xprt;
1142         enum xprt_stat stat;
1143         struct svc_req *rqstp;
1144         struct proc *p;
1145         long sz;
1146         int error;
1147
1148         st = mem_alloc(sizeof(*st));
1149         mtx_init(&st->st_lock, "st_lock", NULL, MTX_DEF);
1150         st->st_pool = pool;
1151         st->st_xprt = NULL;
1152         STAILQ_INIT(&st->st_reqs);
1153         cv_init(&st->st_cond, "rpcsvc");
1154
1155         mtx_lock(&grp->sg_lock);
1156
1157         /*
1158          * If we are a new thread which was spawned to cope with
1159          * increased load, set the state back to SVCPOOL_ACTIVE.
1160          */
1161         if (grp->sg_state == SVCPOOL_THREADSTARTING)
1162                 grp->sg_state = SVCPOOL_ACTIVE;
1163
1164         while (grp->sg_state != SVCPOOL_CLOSING) {
1165                 /*
1166                  * Create new thread if requested.
1167                  */
1168                 if (grp->sg_state == SVCPOOL_THREADWANTED) {
1169                         grp->sg_state = SVCPOOL_THREADSTARTING;
1170                         grp->sg_lastcreatetime = time_uptime;
1171                         mtx_unlock(&grp->sg_lock);
1172                         svc_new_thread(grp);
1173                         mtx_lock(&grp->sg_lock);
1174                         continue;
1175                 }
1176
1177                 /*
1178                  * Check for idle transports once per second.
1179                  */
1180                 if (time_uptime > grp->sg_lastidlecheck) {
1181                         grp->sg_lastidlecheck = time_uptime;
1182                         svc_checkidle(grp);
1183                 }
1184
1185                 xprt = st->st_xprt;
1186                 if (!xprt) {
1187                         /*
1188                          * Enforce maxthreads count.
1189                          */
1190                         if (!ismaster && grp->sg_threadcount >
1191                             grp->sg_maxthreads)
1192                                 break;
1193
1194                         /*
1195                          * Before sleeping, see if we can find an
1196                          * active transport which isn't being serviced
1197                          * by a thread.
1198                          */
1199                         if (svc_request_space_available(pool) &&
1200                             (xprt = TAILQ_FIRST(&grp->sg_active)) != NULL) {
1201                                 TAILQ_REMOVE(&grp->sg_active, xprt, xp_alink);
1202                                 SVC_ACQUIRE(xprt);
1203                                 xprt->xp_thread = st;
1204                                 st->st_xprt = xprt;
1205                                 continue;
1206                         }
1207
1208                         LIST_INSERT_HEAD(&grp->sg_idlethreads, st, st_ilink);
1209                         if (ismaster || (!ismaster &&
1210                             grp->sg_threadcount > grp->sg_minthreads))
1211                                 error = cv_timedwait_sig(&st->st_cond,
1212                                     &grp->sg_lock, 5 * hz);
1213                         else
1214                                 error = cv_wait_sig(&st->st_cond,
1215                                     &grp->sg_lock);
1216                         if (st->st_xprt == NULL)
1217                                 LIST_REMOVE(st, st_ilink);
1218
1219                         /*
1220                          * Reduce worker thread count when idle.
1221                          */
1222                         if (error == EWOULDBLOCK) {
1223                                 if (!ismaster
1224                                     && (grp->sg_threadcount
1225                                         > grp->sg_minthreads)
1226                                         && !st->st_xprt)
1227                                         break;
1228                         } else if (error != 0) {
1229                                 KASSERT(error == EINTR || error == ERESTART,
1230                                     ("non-signal error %d", error));
1231                                 mtx_unlock(&grp->sg_lock);
1232                                 p = curproc;
1233                                 PROC_LOCK(p);
1234                                 if (P_SHOULDSTOP(p) ||
1235                                     (p->p_flag & P_TOTAL_STOP) != 0) {
1236                                         thread_suspend_check(0);
1237                                         PROC_UNLOCK(p);
1238                                         mtx_lock(&grp->sg_lock);
1239                                 } else {
1240                                         PROC_UNLOCK(p);
1241                                         svc_exit(pool);
1242                                         mtx_lock(&grp->sg_lock);
1243                                         break;
1244                                 }
1245                         }
1246                         continue;
1247                 }
1248                 mtx_unlock(&grp->sg_lock);
1249
1250                 /*
1251                  * Drain the transport socket and queue up any RPCs.
1252                  */
1253                 xprt->xp_lastactive = time_uptime;
1254                 do {
1255                         if (!svc_request_space_available(pool))
1256                                 break;
1257                         rqstp = NULL;
1258                         stat = svc_getreq(xprt, &rqstp);
1259                         if (rqstp) {
1260                                 svc_change_space_used(pool, rqstp->rq_size);
1261                                 /*
1262                                  * See if the application has a preference
1263                                  * for some other thread.
1264                                  */
1265                                 if (pool->sp_assign) {
1266                                         stpref = pool->sp_assign(st, rqstp);
1267                                         rqstp->rq_thread = stpref;
1268                                         STAILQ_INSERT_TAIL(&stpref->st_reqs,
1269                                             rqstp, rq_link);
1270                                         mtx_unlock(&stpref->st_lock);
1271                                         if (stpref != st)
1272                                                 rqstp = NULL;
1273                                 } else {
1274                                         rqstp->rq_thread = st;
1275                                         STAILQ_INSERT_TAIL(&st->st_reqs,
1276                                             rqstp, rq_link);
1277                                 }
1278                         }
1279                 } while (rqstp == NULL && stat == XPRT_MOREREQS
1280                     && grp->sg_state != SVCPOOL_CLOSING);
1281
1282                 /*
1283                  * Move this transport to the end of the active list to
1284                  * ensure fairness when multiple transports are active.
1285                  * If this was the last queued request, svc_getreq will end
1286                  * up calling xprt_inactive to remove from the active list.
1287                  */
1288                 mtx_lock(&grp->sg_lock);
1289                 xprt->xp_thread = NULL;
1290                 st->st_xprt = NULL;
1291                 if (xprt->xp_active) {
1292                         if (!svc_request_space_available(pool) ||
1293                             !xprt_assignthread(xprt))
1294                                 TAILQ_INSERT_TAIL(&grp->sg_active,
1295                                     xprt, xp_alink);
1296                 }
1297                 mtx_unlock(&grp->sg_lock);
1298                 SVC_RELEASE(xprt);
1299
1300                 /*
1301                  * Execute what we have queued.
1302                  */
1303                 mtx_lock(&st->st_lock);
1304                 while ((rqstp = STAILQ_FIRST(&st->st_reqs)) != NULL) {
1305                         STAILQ_REMOVE_HEAD(&st->st_reqs, rq_link);
1306                         mtx_unlock(&st->st_lock);
1307                         sz = (long)rqstp->rq_size;
1308                         svc_executereq(rqstp);
1309                         svc_change_space_used(pool, -sz);
1310                         mtx_lock(&st->st_lock);
1311                 }
1312                 mtx_unlock(&st->st_lock);
1313                 mtx_lock(&grp->sg_lock);
1314         }
1315
1316         if (st->st_xprt) {
1317                 xprt = st->st_xprt;
1318                 st->st_xprt = NULL;
1319                 SVC_RELEASE(xprt);
1320         }
1321         KASSERT(STAILQ_EMPTY(&st->st_reqs), ("stray reqs on exit"));
1322         mtx_destroy(&st->st_lock);
1323         cv_destroy(&st->st_cond);
1324         mem_free(st, sizeof(*st));
1325
1326         grp->sg_threadcount--;
1327         if (!ismaster)
1328                 wakeup(grp);
1329         mtx_unlock(&grp->sg_lock);
1330 }
1331
1332 static void
1333 svc_thread_start(void *arg)
1334 {
1335
1336         svc_run_internal((SVCGROUP *) arg, FALSE);
1337         kthread_exit();
1338 }
1339
1340 static void
1341 svc_new_thread(SVCGROUP *grp)
1342 {
1343         SVCPOOL *pool = grp->sg_pool;
1344         struct thread *td;
1345
1346         mtx_lock(&grp->sg_lock);
1347         grp->sg_threadcount++;
1348         mtx_unlock(&grp->sg_lock);
1349         kthread_add(svc_thread_start, grp, pool->sp_proc, &td, 0, 0,
1350             "%s: service", pool->sp_name);
1351 }
1352
1353 void
1354 svc_run(SVCPOOL *pool)
1355 {
1356         int g, i;
1357         struct proc *p;
1358         struct thread *td;
1359         SVCGROUP *grp;
1360
1361         p = curproc;
1362         td = curthread;
1363         snprintf(td->td_name, sizeof(td->td_name),
1364             "%s: master", pool->sp_name);
1365         pool->sp_state = SVCPOOL_ACTIVE;
1366         pool->sp_proc = p;
1367
1368         /* Choose group count based on number of threads and CPUs. */
1369         pool->sp_groupcount = max(1, min(SVC_MAXGROUPS,
1370             min(pool->sp_maxthreads / 2, mp_ncpus) / 6));
1371         for (g = 0; g < pool->sp_groupcount; g++) {
1372                 grp = &pool->sp_groups[g];
1373                 grp->sg_minthreads = max(1,
1374                     pool->sp_minthreads / pool->sp_groupcount);
1375                 grp->sg_maxthreads = max(1,
1376                     pool->sp_maxthreads / pool->sp_groupcount);
1377                 grp->sg_lastcreatetime = time_uptime;
1378         }
1379
1380         /* Starting threads */
1381         pool->sp_groups[0].sg_threadcount++;
1382         for (g = 0; g < pool->sp_groupcount; g++) {
1383                 grp = &pool->sp_groups[g];
1384                 for (i = ((g == 0) ? 1 : 0); i < grp->sg_minthreads; i++)
1385                         svc_new_thread(grp);
1386         }
1387         svc_run_internal(&pool->sp_groups[0], TRUE);
1388
1389         /* Waiting for threads to stop. */
1390         for (g = 0; g < pool->sp_groupcount; g++) {
1391                 grp = &pool->sp_groups[g];
1392                 mtx_lock(&grp->sg_lock);
1393                 while (grp->sg_threadcount > 0)
1394                         msleep(grp, &grp->sg_lock, 0, "svcexit", 0);
1395                 mtx_unlock(&grp->sg_lock);
1396         }
1397 }
1398
1399 void
1400 svc_exit(SVCPOOL *pool)
1401 {
1402         SVCGROUP *grp;
1403         SVCTHREAD *st;
1404         int g;
1405
1406         pool->sp_state = SVCPOOL_CLOSING;
1407         for (g = 0; g < pool->sp_groupcount; g++) {
1408                 grp = &pool->sp_groups[g];
1409                 mtx_lock(&grp->sg_lock);
1410                 if (grp->sg_state != SVCPOOL_CLOSING) {
1411                         grp->sg_state = SVCPOOL_CLOSING;
1412                         LIST_FOREACH(st, &grp->sg_idlethreads, st_ilink)
1413                                 cv_signal(&st->st_cond);
1414                 }
1415                 mtx_unlock(&grp->sg_lock);
1416         }
1417 }
1418
1419 bool_t
1420 svc_getargs(struct svc_req *rqstp, xdrproc_t xargs, void *args)
1421 {
1422         struct mbuf *m;
1423         XDR xdrs;
1424         bool_t stat;
1425
1426         m = rqstp->rq_args;
1427         rqstp->rq_args = NULL;
1428
1429         xdrmbuf_create(&xdrs, m, XDR_DECODE);
1430         stat = xargs(&xdrs, args);
1431         XDR_DESTROY(&xdrs);
1432
1433         return (stat);
1434 }
1435
1436 bool_t
1437 svc_freeargs(struct svc_req *rqstp, xdrproc_t xargs, void *args)
1438 {
1439         XDR xdrs;
1440
1441         if (rqstp->rq_addr) {
1442                 free(rqstp->rq_addr, M_SONAME);
1443                 rqstp->rq_addr = NULL;
1444         }
1445
1446         xdrs.x_op = XDR_FREE;
1447         return (xargs(&xdrs, args));
1448 }
1449
1450 void
1451 svc_freereq(struct svc_req *rqstp)
1452 {
1453         SVCTHREAD *st;
1454         SVCPOOL *pool;
1455
1456         st = rqstp->rq_thread;
1457         if (st) {
1458                 pool = st->st_pool;
1459                 if (pool->sp_done)
1460                         pool->sp_done(st, rqstp);
1461         }
1462
1463         if (rqstp->rq_auth.svc_ah_ops)
1464                 SVCAUTH_RELEASE(&rqstp->rq_auth);
1465
1466         if (rqstp->rq_xprt) {
1467                 SVC_RELEASE(rqstp->rq_xprt);
1468         }
1469
1470         if (rqstp->rq_addr)
1471                 free(rqstp->rq_addr, M_SONAME);
1472
1473         if (rqstp->rq_args)
1474                 m_freem(rqstp->rq_args);
1475
1476         free(rqstp, M_RPC);
1477 }