SAFE public projects git trees. - safe/jmp/linux-2.6/blob - net/sunrpc/xprt.c

   1 /*
   2  *  linux/net/sunrpc/xprt.c
   3  *
   4  *  This is a generic RPC call interface supporting congestion avoidance,
   5  *  and asynchronous calls.
   6  *
   7  *  The interface works like this:
   8  *
   9  *  -   When a process places a call, it allocates a request slot if
  10  *      one is available. Otherwise, it sleeps on the backlog queue
  11  *      (xprt_reserve).
  12  *  -   Next, the caller puts together the RPC message, stuffs it into
  13  *      the request struct, and calls xprt_transmit().
  14  *  -   xprt_transmit sends the message and installs the caller on the
  15  *      transport's wait list. At the same time, it installs a timer that
  16  *      is run after the packet's timeout has expired.
  17  *  -   When a packet arrives, the data_ready handler walks the list of
  18  *      pending requests for that transport. If a matching XID is found, the
  19  *      caller is woken up, and the timer removed.
  20  *  -   When no reply arrives within the timeout interval, the timer is
  21  *      fired by the kernel and runs xprt_timer(). It either adjusts the
  22  *      timeout values (minor timeout) or wakes up the caller with a status
  23  *      of -ETIMEDOUT.
  24  *  -   When the caller receives a notification from RPC that a reply arrived,
  25  *      it should release the RPC slot, and process the reply.
  26  *      If the call timed out, it may choose to retry the operation by
  27  *      adjusting the initial timeout value, and simply calling rpc_call
  28  *      again.
  29  *
  30  *  Support for async RPC is done through a set of RPC-specific scheduling
  31  *  primitives that `transparently' work for processes as well as async
  32  *  tasks that rely on callbacks.
  33  *
  34  *  Copyright (C) 1995-1997, Olaf Kirch <okir@monad.swb.de>
  35  *
  36  *  Transport switch API copyright (C) 2005, Chuck Lever <cel@netapp.com>
  37  */
  38
  39 #include <linux/module.h>
  40
  41 #include <linux/types.h>
  42 #include <linux/interrupt.h>
  43 #include <linux/workqueue.h>
  44 #include <linux/random.h>
  45
  46 #include <linux/sunrpc/clnt.h>
  47
  48 /*
  49  * Local variables
  50  */
  51
  52 #ifdef RPC_DEBUG
  53 # undef  RPC_DEBUG_DATA
  54 # define RPCDBG_FACILITY        RPCDBG_XPRT
  55 #endif
  56
  57 /*
  58  * Local functions
  59  */
  60 static void     xprt_request_init(struct rpc_task *, struct rpc_xprt *);
  61 static inline void      do_xprt_reserve(struct rpc_task *);
  62 static void     xprt_connect_status(struct rpc_task *task);
  63 static int      __xprt_get_cong(struct rpc_xprt *, struct rpc_task *);
  64
  65 static int      xprt_clear_backlog(struct rpc_xprt *xprt);
  66
  67 /*
  68  * Serialize write access to transports, in order to prevent different
  69  * requests from interfering with each other.
  70  * Also prevents transport connects from colliding with writes.
  71  */
  72 static int
  73 __xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
  74 {
  75         struct rpc_rqst *req = task->tk_rqstp;
  76
  77         if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
  78                 if (task == xprt->snd_task)
  79                         return 1;
  80                 goto out_sleep;
  81         }
  82         if (xprt->nocong || __xprt_get_cong(xprt, task)) {
  83                 xprt->snd_task = task;
  84                 if (req) {
  85                         req->rq_bytes_sent = 0;
  86                         req->rq_ntrans++;
  87                 }
  88                 return 1;
  89         }
  90         smp_mb__before_clear_bit();
  91         clear_bit(XPRT_LOCKED, &xprt->state);
  92         smp_mb__after_clear_bit();
  93 out_sleep:
  94         dprintk("RPC: %4d failed to lock transport %p\n", task->tk_pid, xprt);
  95         task->tk_timeout = 0;
  96         task->tk_status = -EAGAIN;
  97         if (req && req->rq_ntrans)
  98                 rpc_sleep_on(&xprt->resend, task, NULL, NULL);
  99         else
 100                 rpc_sleep_on(&xprt->sending, task, NULL, NULL);
 101         return 0;
 102 }
 103
 104 static inline int
 105 xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 106 {
 107         int retval;
 108
 109         spin_lock_bh(&xprt->transport_lock);
 110         retval = __xprt_lock_write(xprt, task);
 111         spin_unlock_bh(&xprt->transport_lock);
 112         return retval;
 113 }
 114
 115
 116 static void
 117 __xprt_lock_write_next(struct rpc_xprt *xprt)
 118 {
 119         struct rpc_task *task;
 120
 121         if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 122                 return;
 123         if (!xprt->nocong && RPCXPRT_CONGESTED(xprt))
 124                 goto out_unlock;
 125         task = rpc_wake_up_next(&xprt->resend);
 126         if (!task) {
 127                 task = rpc_wake_up_next(&xprt->sending);
 128                 if (!task)
 129                         goto out_unlock;
 130         }
 131         if (xprt->nocong || __xprt_get_cong(xprt, task)) {
 132                 struct rpc_rqst *req = task->tk_rqstp;
 133                 xprt->snd_task = task;
 134                 if (req) {
 135                         req->rq_bytes_sent = 0;
 136                         req->rq_ntrans++;
 137                 }
 138                 return;
 139         }
 140 out_unlock:
 141         smp_mb__before_clear_bit();
 142         clear_bit(XPRT_LOCKED, &xprt->state);
 143         smp_mb__after_clear_bit();
 144 }
 145
 146 /*
 147  * Releases the transport for use by other requests.
 148  */
 149 static void
 150 __xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 151 {
 152         if (xprt->snd_task == task) {
 153                 xprt->snd_task = NULL;
 154                 smp_mb__before_clear_bit();
 155                 clear_bit(XPRT_LOCKED, &xprt->state);
 156                 smp_mb__after_clear_bit();
 157                 __xprt_lock_write_next(xprt);
 158         }
 159 }
 160
 161 static inline void
 162 xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 163 {
 164         spin_lock_bh(&xprt->transport_lock);
 165         __xprt_release_write(xprt, task);
 166         spin_unlock_bh(&xprt->transport_lock);
 167 }
 168
 169 /*
 170  * Van Jacobson congestion avoidance. Check if the congestion window
 171  * overflowed. Put the task to sleep if this is the case.
 172  */
 173 static int
 174 __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 175 {
 176         struct rpc_rqst *req = task->tk_rqstp;
 177
 178         if (req->rq_cong)
 179                 return 1;
 180         dprintk("RPC: %4d xprt_cwnd_limited cong = %ld cwnd = %ld\n",
 181                         task->tk_pid, xprt->cong, xprt->cwnd);
 182         if (RPCXPRT_CONGESTED(xprt))
 183                 return 0;
 184         req->rq_cong = 1;
 185         xprt->cong += RPC_CWNDSCALE;
 186         return 1;
 187 }
 188
 189 /*
 190  * Adjust the congestion window, and wake up the next task
 191  * that has been sleeping due to congestion
 192  */
 193 static void
 194 __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
 195 {
 196         if (!req->rq_cong)
 197                 return;
 198         req->rq_cong = 0;
 199         xprt->cong -= RPC_CWNDSCALE;
 200         __xprt_lock_write_next(xprt);
 201 }
 202
 203 /*
 204  * Adjust RPC congestion window
 205  * We use a time-smoothed congestion estimator to avoid heavy oscillation.
 206  */
 207 static void
 208 xprt_adjust_cwnd(struct rpc_xprt *xprt, int result)
 209 {
 210         unsigned long   cwnd;
 211
 212         cwnd = xprt->cwnd;
 213         if (result >= 0 && cwnd <= xprt->cong) {
 214                 /* The (cwnd >> 1) term makes sure
 215                  * the result gets rounded properly. */
 216                 cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
 217                 if (cwnd > RPC_MAXCWND(xprt))
 218                         cwnd = RPC_MAXCWND(xprt);
 219                 __xprt_lock_write_next(xprt);
 220         } else if (result == -ETIMEDOUT) {
 221                 cwnd >>= 1;
 222                 if (cwnd < RPC_CWNDSCALE)
 223                         cwnd = RPC_CWNDSCALE;
 224         }
 225         dprintk("RPC:      cong %ld, cwnd was %ld, now %ld\n",
 226                         xprt->cong, xprt->cwnd, cwnd);
 227         xprt->cwnd = cwnd;
 228 }
 229
 230 /**
 231  * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue
 232  * @xprt: transport with waiting tasks
 233  * @status: result code to plant in each task before waking it
 234  *
 235  */
 236 void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status)
 237 {
 238         if (status < 0)
 239                 rpc_wake_up_status(&xprt->pending, status);
 240         else
 241                 rpc_wake_up(&xprt->pending);
 242 }
 243
 244 /**
 245  * xprt_wait_for_buffer_space - wait for transport output buffer to clear
 246  * @task: task to be put to sleep
 247  *
 248  */
 249 void xprt_wait_for_buffer_space(struct rpc_task *task)
 250 {
 251         struct rpc_rqst *req = task->tk_rqstp;
 252         struct rpc_xprt *xprt = req->rq_xprt;
 253
 254         task->tk_timeout = req->rq_timeout;
 255         rpc_sleep_on(&xprt->pending, task, NULL, NULL);
 256 }
 257
 258 /**
 259  * xprt_write_space - wake the task waiting for transport output buffer space
 260  * @xprt: transport with waiting tasks
 261  *
 262  * Can be called in a soft IRQ context, so xprt_write_space never sleeps.
 263  */
 264 void xprt_write_space(struct rpc_xprt *xprt)
 265 {
 266         if (unlikely(xprt->shutdown))
 267                 return;
 268
 269         spin_lock_bh(&xprt->transport_lock);
 270         if (xprt->snd_task) {
 271                 dprintk("RPC:      write space: waking waiting task on xprt %p\n",
 272                                 xprt);
 273                 rpc_wake_up_task(xprt->snd_task);
 274         }
 275         spin_unlock_bh(&xprt->transport_lock);
 276 }
 277
 278 static void xprt_reset_majortimeo(struct rpc_rqst *req)
 279 {
 280         struct rpc_timeout *to = &req->rq_xprt->timeout;
 281
 282         req->rq_majortimeo = req->rq_timeout;
 283         if (to->to_exponential)
 284                 req->rq_majortimeo <<= to->to_retries;
 285         else
 286                 req->rq_majortimeo += to->to_increment * to->to_retries;
 287         if (req->rq_majortimeo > to->to_maxval || req->rq_majortimeo == 0)
 288                 req->rq_majortimeo = to->to_maxval;
 289         req->rq_majortimeo += jiffies;
 290 }
 291
 292 /**
 293  * xprt_adjust_timeout - adjust timeout values for next retransmit
 294  * @req: RPC request containing parameters to use for the adjustment
 295  *
 296  */
 297 int xprt_adjust_timeout(struct rpc_rqst *req)
 298 {
 299         struct rpc_xprt *xprt = req->rq_xprt;
 300         struct rpc_timeout *to = &xprt->timeout;
 301         int status = 0;
 302
 303         if (time_before(jiffies, req->rq_majortimeo)) {
 304                 if (to->to_exponential)
 305                         req->rq_timeout <<= 1;
 306                 else
 307                         req->rq_timeout += to->to_increment;
 308                 if (to->to_maxval && req->rq_timeout >= to->to_maxval)
 309                         req->rq_timeout = to->to_maxval;
 310                 req->rq_retries++;
 311                 pprintk("RPC: %lu retrans\n", jiffies);
 312         } else {
 313                 req->rq_timeout = to->to_initval;
 314                 req->rq_retries = 0;
 315                 xprt_reset_majortimeo(req);
 316                 /* Reset the RTT counters == "slow start" */
 317                 spin_lock_bh(&xprt->transport_lock);
 318                 rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
 319                 spin_unlock_bh(&xprt->transport_lock);
 320                 pprintk("RPC: %lu timeout\n", jiffies);
 321                 status = -ETIMEDOUT;
 322         }
 323
 324         if (req->rq_timeout == 0) {
 325                 printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
 326                 req->rq_timeout = 5 * HZ;
 327         }
 328         return status;
 329 }
 330
 331 static void xprt_autoclose(void *args)
 332 {
 333         struct rpc_xprt *xprt = (struct rpc_xprt *)args;
 334
 335         xprt_disconnect(xprt);
 336         xprt->ops->close(xprt);
 337         xprt_release_write(xprt, NULL);
 338 }
 339
 340 /**
 341  * xprt_disconnect - mark a transport as disconnected
 342  * @xprt: transport to flag for disconnect
 343  *
 344  */
 345 void xprt_disconnect(struct rpc_xprt *xprt)
 346 {
 347         dprintk("RPC:      disconnected transport %p\n", xprt);
 348         spin_lock_bh(&xprt->transport_lock);
 349         xprt_clear_connected(xprt);
 350         xprt_wake_pending_tasks(xprt, -ENOTCONN);
 351         spin_unlock_bh(&xprt->transport_lock);
 352 }
 353
 354 static void
 355 xprt_init_autodisconnect(unsigned long data)
 356 {
 357         struct rpc_xprt *xprt = (struct rpc_xprt *)data;
 358
 359         spin_lock(&xprt->transport_lock);
 360         if (!list_empty(&xprt->recv) || xprt->shutdown)
 361                 goto out_abort;
 362         if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 363                 goto out_abort;
 364         spin_unlock(&xprt->transport_lock);
 365         if (xprt_connecting(xprt))
 366                 xprt_release_write(xprt, NULL);
 367         else
 368                 schedule_work(&xprt->task_cleanup);
 369         return;
 370 out_abort:
 371         spin_unlock(&xprt->transport_lock);
 372 }
 373
 374 /**
 375  * xprt_connect - schedule a transport connect operation
 376  * @task: RPC task that is requesting the connect
 377  *
 378  */
 379 void xprt_connect(struct rpc_task *task)
 380 {
 381         struct rpc_xprt *xprt = task->tk_xprt;
 382
 383         dprintk("RPC: %4d xprt_connect xprt %p %s connected\n", task->tk_pid,
 384                         xprt, (xprt_connected(xprt) ? "is" : "is not"));
 385
 386         if (xprt->shutdown) {
 387                 task->tk_status = -EIO;
 388                 return;
 389         }
 390         if (!xprt->addr.sin_port) {
 391                 task->tk_status = -EIO;
 392                 return;
 393         }
 394         if (!xprt_lock_write(xprt, task))
 395                 return;
 396         if (xprt_connected(xprt))
 397                 xprt_release_write(xprt, task);
 398         else {
 399                 if (task->tk_rqstp)
 400                         task->tk_rqstp->rq_bytes_sent = 0;
 401
 402                 task->tk_timeout = RPC_CONNECT_TIMEOUT;
 403                 rpc_sleep_on(&xprt->pending, task, xprt_connect_status, NULL);
 404                 xprt->ops->connect(task);
 405         }
 406         return;
 407 }
 408
 409 static void xprt_connect_status(struct rpc_task *task)
 410 {
 411         struct rpc_xprt *xprt = task->tk_xprt;
 412
 413         if (task->tk_status >= 0) {
 414                 dprintk("RPC: %4d xprt_connect_status: connection established\n",
 415                                 task->tk_pid);
 416                 return;
 417         }
 418
 419         switch (task->tk_status) {
 420         case -ECONNREFUSED:
 421         case -ECONNRESET:
 422                 dprintk("RPC: %4d xprt_connect_status: server %s refused connection\n",
 423                                 task->tk_pid, task->tk_client->cl_server);
 424                 break;
 425         case -ENOTCONN:
 426                 dprintk("RPC: %4d xprt_connect_status: connection broken\n",
 427                                 task->tk_pid);
 428                 break;
 429         case -ETIMEDOUT:
 430                 dprintk("RPC: %4d xprt_connect_status: connect attempt timed out\n",
 431                                 task->tk_pid);
 432                 break;
 433         default:
 434                 dprintk("RPC: %4d xprt_connect_status: error %d connecting to server %s\n",
 435                                 task->tk_pid, -task->tk_status, task->tk_client->cl_server);
 436                 xprt_release_write(xprt, task);
 437                 task->tk_status = -EIO;
 438                 return;
 439         }
 440
 441         /* if soft mounted, just cause this RPC to fail */
 442         if (RPC_IS_SOFT(task)) {
 443                 xprt_release_write(xprt, task);
 444                 task->tk_status = -EIO;
 445         }
 446 }
 447
 448 /**
 449  * xprt_lookup_rqst - find an RPC request corresponding to an XID
 450  * @xprt: transport on which the original request was transmitted
 451  * @xid: RPC XID of incoming reply
 452  *
 453  */
 454 struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, u32 xid)
 455 {
 456         struct list_head *pos;
 457         struct rpc_rqst *req = NULL;
 458
 459         list_for_each(pos, &xprt->recv) {
 460                 struct rpc_rqst *entry = list_entry(pos, struct rpc_rqst, rq_list);
 461                 if (entry->rq_xid == xid) {
 462                         req = entry;
 463                         break;
 464                 }
 465         }
 466         return req;
 467 }
 468
 469 /**
 470  * xprt_complete_rqst - called when reply processing is complete
 471  * @xprt: controlling transport
 472  * @req: RPC request that just completed
 473  * @copied: actual number of bytes received from the transport
 474  *
 475  */
 476 void xprt_complete_rqst(struct rpc_xprt *xprt, struct rpc_rqst *req, int copied)
 477 {
 478         struct rpc_task *task = req->rq_task;
 479         struct rpc_clnt *clnt = task->tk_client;
 480
 481         /* Adjust congestion window */
 482         if (!xprt->nocong) {
 483                 unsigned timer = task->tk_msg.rpc_proc->p_timer;
 484                 xprt_adjust_cwnd(xprt, copied);
 485                 __xprt_put_cong(xprt, req);
 486                 if (timer) {
 487                         if (req->rq_ntrans == 1)
 488                                 rpc_update_rtt(clnt->cl_rtt, timer,
 489                                                 (long)jiffies - req->rq_xtime);
 490                         rpc_set_timeo(clnt->cl_rtt, timer, req->rq_ntrans - 1);
 491                 }
 492         }
 493
 494 #ifdef RPC_PROFILE
 495         /* Profile only reads for now */
 496         if (copied > 1024) {
 497                 static unsigned long    nextstat;
 498                 static unsigned long    pkt_rtt, pkt_len, pkt_cnt;
 499
 500                 pkt_cnt++;
 501                 pkt_len += req->rq_slen + copied;
 502                 pkt_rtt += jiffies - req->rq_xtime;
 503                 if (time_before(nextstat, jiffies)) {
 504                         printk("RPC: %lu %ld cwnd\n", jiffies, xprt->cwnd);
 505                         printk("RPC: %ld %ld %ld %ld stat\n",
 506                                         jiffies, pkt_cnt, pkt_len, pkt_rtt);
 507                         pkt_rtt = pkt_len = pkt_cnt = 0;
 508                         nextstat = jiffies + 5 * HZ;
 509                 }
 510         }
 511 #endif
 512
 513         dprintk("RPC: %4d has input (%d bytes)\n", task->tk_pid, copied);
 514         list_del_init(&req->rq_list);
 515         req->rq_received = req->rq_private_buf.len = copied;
 516
 517         /* ... and wake up the process. */
 518         rpc_wake_up_task(task);
 519         return;
 520 }
 521
 522 /*
 523  * RPC receive timeout handler.
 524  */
 525 static void
 526 xprt_timer(struct rpc_task *task)
 527 {
 528         struct rpc_rqst *req = task->tk_rqstp;
 529         struct rpc_xprt *xprt = req->rq_xprt;
 530
 531         spin_lock(&xprt->transport_lock);
 532         if (req->rq_received)
 533                 goto out;
 534
 535         xprt_adjust_cwnd(req->rq_xprt, -ETIMEDOUT);
 536         __xprt_put_cong(xprt, req);
 537
 538         dprintk("RPC: %4d xprt_timer (%s request)\n",
 539                 task->tk_pid, req ? "pending" : "backlogged");
 540
 541         task->tk_status  = -ETIMEDOUT;
 542 out:
 543         task->tk_timeout = 0;
 544         rpc_wake_up_task(task);
 545         spin_unlock(&xprt->transport_lock);
 546 }
 547
 548 /**
 549  * xprt_prepare_transmit - reserve the transport before sending a request
 550  * @task: RPC task about to send a request
 551  *
 552  */
 553 int xprt_prepare_transmit(struct rpc_task *task)
 554 {
 555         struct rpc_rqst *req = task->tk_rqstp;
 556         struct rpc_xprt *xprt = req->rq_xprt;
 557         int err = 0;
 558
 559         dprintk("RPC: %4d xprt_prepare_transmit\n", task->tk_pid);
 560
 561         if (xprt->shutdown)
 562                 return -EIO;
 563
 564         spin_lock_bh(&xprt->transport_lock);
 565         if (req->rq_received && !req->rq_bytes_sent) {
 566                 err = req->rq_received;
 567                 goto out_unlock;
 568         }
 569         if (!__xprt_lock_write(xprt, task)) {
 570                 err = -EAGAIN;
 571                 goto out_unlock;
 572         }
 573
 574         if (!xprt_connected(xprt)) {
 575                 err = -ENOTCONN;
 576                 goto out_unlock;
 577         }
 578 out_unlock:
 579         spin_unlock_bh(&xprt->transport_lock);
 580         return err;
 581 }
 582
 583 /**
 584  * xprt_transmit - send an RPC request on a transport
 585  * @task: controlling RPC task
 586  *
 587  * We have to copy the iovec because sendmsg fiddles with its contents.
 588  */
 589 void xprt_transmit(struct rpc_task *task)
 590 {
 591         struct rpc_clnt *clnt = task->tk_client;
 592         struct rpc_rqst *req = task->tk_rqstp;
 593         struct rpc_xprt *xprt = req->rq_xprt;
 594         int status;
 595
 596         dprintk("RPC: %4d xprt_transmit(%u)\n", task->tk_pid, req->rq_slen);
 597
 598         smp_rmb();
 599         if (!req->rq_received) {
 600                 if (list_empty(&req->rq_list)) {
 601                         spin_lock_bh(&xprt->transport_lock);
 602                         /* Update the softirq receive buffer */
 603                         memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
 604                                         sizeof(req->rq_private_buf));
 605                         /* Add request to the receive list */
 606                         list_add_tail(&req->rq_list, &xprt->recv);
 607                         spin_unlock_bh(&xprt->transport_lock);
 608                         xprt_reset_majortimeo(req);
 609                         /* Turn off autodisconnect */
 610                         del_singleshot_timer_sync(&xprt->timer);
 611                 }
 612         } else if (!req->rq_bytes_sent)
 613                 return;
 614
 615         status = xprt->ops->send_request(task);
 616         if (!status)
 617                 goto out_receive;
 618
 619         /* Note: at this point, task->tk_sleeping has not yet been set,
 620          *       hence there is no danger of the waking up task being put on
 621          *       schedq, and being picked up by a parallel run of rpciod().
 622          */
 623         task->tk_status = status;
 624
 625         switch (status) {
 626         case -ECONNREFUSED:
 627                 task->tk_timeout = RPC_REESTABLISH_TIMEOUT;
 628                 rpc_sleep_on(&xprt->sending, task, NULL, NULL);
 629         case -EAGAIN:
 630         case -ENOTCONN:
 631                 return;
 632         default:
 633                 break;
 634         }
 635         xprt_release_write(xprt, task);
 636         return;
 637  out_receive:
 638         dprintk("RPC: %4d xmit complete\n", task->tk_pid);
 639         /* Set the task's receive timeout value */
 640         spin_lock_bh(&xprt->transport_lock);
 641         if (!xprt->nocong) {
 642                 int timer = task->tk_msg.rpc_proc->p_timer;
 643                 task->tk_timeout = rpc_calc_rto(clnt->cl_rtt, timer);
 644                 task->tk_timeout <<= rpc_ntimeo(clnt->cl_rtt, timer) + req->rq_retries;
 645                 if (task->tk_timeout > xprt->timeout.to_maxval || task->tk_timeout == 0)
 646                         task->tk_timeout = xprt->timeout.to_maxval;
 647         } else
 648                 task->tk_timeout = req->rq_timeout;
 649         /* Don't race with disconnect */
 650         if (!xprt_connected(xprt))
 651                 task->tk_status = -ENOTCONN;
 652         else if (!req->rq_received)
 653                 rpc_sleep_on(&xprt->pending, task, NULL, xprt_timer);
 654         __xprt_release_write(xprt, task);
 655         spin_unlock_bh(&xprt->transport_lock);
 656 }
 657
 658 static inline void do_xprt_reserve(struct rpc_task *task)
 659 {
 660         struct rpc_xprt *xprt = task->tk_xprt;
 661
 662         task->tk_status = 0;
 663         if (task->tk_rqstp)
 664                 return;
 665         if (!list_empty(&xprt->free)) {
 666                 struct rpc_rqst *req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
 667                 list_del_init(&req->rq_list);
 668                 task->tk_rqstp = req;
 669                 xprt_request_init(task, xprt);
 670                 return;
 671         }
 672         dprintk("RPC:      waiting for request slot\n");
 673         task->tk_status = -EAGAIN;
 674         task->tk_timeout = 0;
 675         rpc_sleep_on(&xprt->backlog, task, NULL, NULL);
 676 }
 677
 678 /**
 679  * xprt_reserve - allocate an RPC request slot
 680  * @task: RPC task requesting a slot allocation
 681  *
 682  * If no more slots are available, place the task on the transport's
 683  * backlog queue.
 684  */
 685 void xprt_reserve(struct rpc_task *task)
 686 {
 687         struct rpc_xprt *xprt = task->tk_xprt;
 688
 689         task->tk_status = -EIO;
 690         if (!xprt->shutdown) {
 691                 spin_lock(&xprt->reserve_lock);
 692                 do_xprt_reserve(task);
 693                 spin_unlock(&xprt->reserve_lock);
 694         }
 695 }
 696
 697 static inline u32 xprt_alloc_xid(struct rpc_xprt *xprt)
 698 {
 699         return xprt->xid++;
 700 }
 701
 702 static inline void xprt_init_xid(struct rpc_xprt *xprt)
 703 {
 704         get_random_bytes(&xprt->xid, sizeof(xprt->xid));
 705 }
 706
 707 static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt)
 708 {
 709         struct rpc_rqst *req = task->tk_rqstp;
 710
 711         req->rq_timeout = xprt->timeout.to_initval;
 712         req->rq_task    = task;
 713         req->rq_xprt    = xprt;
 714         req->rq_xid     = xprt_alloc_xid(xprt);
 715         dprintk("RPC: %4d reserved req %p xid %08x\n", task->tk_pid,
 716                         req, ntohl(req->rq_xid));
 717 }
 718
 719 /**
 720  * xprt_release - release an RPC request slot
 721  * @task: task which is finished with the slot
 722  *
 723  */
 724 void xprt_release(struct rpc_task *task)
 725 {
 726         struct rpc_xprt *xprt = task->tk_xprt;
 727         struct rpc_rqst *req;
 728
 729         if (!(req = task->tk_rqstp))
 730                 return;
 731         spin_lock_bh(&xprt->transport_lock);
 732         __xprt_release_write(xprt, task);
 733         __xprt_put_cong(xprt, req);
 734         if (!list_empty(&req->rq_list))
 735                 list_del(&req->rq_list);
 736         xprt->last_used = jiffies;
 737         if (list_empty(&xprt->recv) && !xprt->shutdown)
 738                 mod_timer(&xprt->timer,
 739                                 xprt->last_used + RPC_IDLE_DISCONNECT_TIMEOUT);
 740         spin_unlock_bh(&xprt->transport_lock);
 741         task->tk_rqstp = NULL;
 742         memset(req, 0, sizeof(*req));   /* mark unused */
 743
 744         dprintk("RPC: %4d release request %p\n", task->tk_pid, req);
 745
 746         spin_lock(&xprt->reserve_lock);
 747         list_add(&req->rq_list, &xprt->free);
 748         xprt_clear_backlog(xprt);
 749         spin_unlock(&xprt->reserve_lock);
 750 }
 751
 752 /**
 753  * xprt_set_timeout - set constant RPC timeout
 754  * @to: RPC timeout parameters to set up
 755  * @retr: number of retries
 756  * @incr: amount of increase after each retry
 757  *
 758  */
 759 void xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long incr)
 760 {
 761         to->to_initval   =
 762         to->to_increment = incr;
 763         to->to_maxval    = to->to_initval + (incr * retr);
 764         to->to_retries   = retr;
 765         to->to_exponential = 0;
 766 }
 767
 768 static struct rpc_xprt *xprt_setup(int proto, struct sockaddr_in *ap, struct rpc_timeout *to)
 769 {
 770         int result;
 771         struct rpc_xprt *xprt;
 772         struct rpc_rqst *req;
 773
 774         if ((xprt = kmalloc(sizeof(struct rpc_xprt), GFP_KERNEL)) == NULL)
 775                 return ERR_PTR(-ENOMEM);
 776         memset(xprt, 0, sizeof(*xprt)); /* Nnnngh! */
 777
 778         xprt->addr = *ap;
 779
 780         switch (proto) {
 781         case IPPROTO_UDP:
 782                 result = xs_setup_udp(xprt, to);
 783                 break;
 784         case IPPROTO_TCP:
 785                 result = xs_setup_tcp(xprt, to);
 786                 break;
 787         default:
 788                 printk(KERN_ERR "RPC: unrecognized transport protocol: %d\n",
 789                                 proto);
 790                 result = -EIO;
 791                 break;
 792         }
 793         if (result) {
 794                 kfree(xprt);
 795                 return ERR_PTR(result);
 796         }
 797
 798         spin_lock_init(&xprt->transport_lock);
 799         spin_lock_init(&xprt->reserve_lock);
 800         init_waitqueue_head(&xprt->cong_wait);
 801
 802         INIT_LIST_HEAD(&xprt->free);
 803         INIT_LIST_HEAD(&xprt->recv);
 804         INIT_WORK(&xprt->task_cleanup, xprt_autoclose, xprt);
 805         init_timer(&xprt->timer);
 806         xprt->timer.function = xprt_init_autodisconnect;
 807         xprt->timer.data = (unsigned long) xprt;
 808         xprt->last_used = jiffies;
 809
 810         rpc_init_wait_queue(&xprt->pending, "xprt_pending");
 811         rpc_init_wait_queue(&xprt->sending, "xprt_sending");
 812         rpc_init_wait_queue(&xprt->resend, "xprt_resend");
 813         rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
 814
 815         /* initialize free list */
 816         for (req = &xprt->slot[xprt->max_reqs-1]; req >= &xprt->slot[0]; req--)
 817                 list_add(&req->rq_list, &xprt->free);
 818
 819         xprt_init_xid(xprt);
 820
 821         dprintk("RPC:      created transport %p with %u slots\n", xprt,
 822                         xprt->max_reqs);
 823
 824         return xprt;
 825 }
 826
 827 /**
 828  * xprt_create_proto - create an RPC client transport
 829  * @proto: requested transport protocol
 830  * @sap: remote peer's address
 831  * @to: timeout parameters for new transport
 832  *
 833  */
 834 struct rpc_xprt *xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to)
 835 {
 836         struct rpc_xprt *xprt;
 837
 838         xprt = xprt_setup(proto, sap, to);
 839         if (IS_ERR(xprt))
 840                 dprintk("RPC:      xprt_create_proto failed\n");
 841         else
 842                 dprintk("RPC:      xprt_create_proto created xprt %p\n", xprt);
 843         return xprt;
 844 }
 845
 846 static void xprt_shutdown(struct rpc_xprt *xprt)
 847 {
 848         xprt->shutdown = 1;
 849         rpc_wake_up(&xprt->sending);
 850         rpc_wake_up(&xprt->resend);
 851         xprt_wake_pending_tasks(xprt, -EIO);
 852         rpc_wake_up(&xprt->backlog);
 853         wake_up(&xprt->cong_wait);
 854         del_timer_sync(&xprt->timer);
 855 }
 856
 857 static int xprt_clear_backlog(struct rpc_xprt *xprt) {
 858         rpc_wake_up_next(&xprt->backlog);
 859         wake_up(&xprt->cong_wait);
 860         return 1;
 861 }
 862
 863 /**
 864  * xprt_destroy - destroy an RPC transport, killing off all requests.
 865  * @xprt: transport to destroy
 866  *
 867  */
 868 int xprt_destroy(struct rpc_xprt *xprt)
 869 {
 870         dprintk("RPC:      destroying transport %p\n", xprt);
 871         xprt_shutdown(xprt);
 872         xprt->ops->destroy(xprt);
 873         kfree(xprt);
 874
 875         return 0;
 876 }