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, if a reply is expected,
  16  *      it installs a timer that is run after the packet's timeout has
  17  *      expired.
  18  *  -   When a packet arrives, the data_ready handler walks the list of
  19  *      pending requests for that transport. If a matching XID is found, the
  20  *      caller is woken up, and the timer removed.
  21  *  -   When no reply arrives within the timeout interval, the timer is
  22  *      fired by the kernel and runs xprt_timer(). It either adjusts the
  23  *      timeout values (minor timeout) or wakes up the caller with a status
  24  *      of -ETIMEDOUT.
  25  *  -   When the caller receives a notification from RPC that a reply arrived,
  26  *      it should release the RPC slot, and process the reply.
  27  *      If the call timed out, it may choose to retry the operation by
  28  *      adjusting the initial timeout value, and simply calling rpc_call
  29  *      again.
  30  *
  31  *  Support for async RPC is done through a set of RPC-specific scheduling
  32  *  primitives that `transparently' work for processes as well as async
  33  *  tasks that rely on callbacks.
  34  *
  35  *  Copyright (C) 1995-1997, Olaf Kirch <okir@monad.swb.de>
  36  *
  37  *  Transport switch API copyright (C) 2005, Chuck Lever <cel@netapp.com>
  38  */
  39
  40 #include <linux/module.h>
  41
  42 #include <linux/types.h>
  43 #include <linux/interrupt.h>
  44 #include <linux/workqueue.h>
  45 #include <linux/net.h>
  46 #include <linux/ktime.h>
  47
  48 #include <linux/sunrpc/clnt.h>
  49 #include <linux/sunrpc/metrics.h>
  50 #include <linux/sunrpc/bc_xprt.h>
  51
  52 #include "sunrpc.h"
  53
  54 /*
  55  * Local variables
  56  */
  57
  58 #ifdef RPC_DEBUG
  59 # define RPCDBG_FACILITY        RPCDBG_XPRT
  60 #endif
  61
  62 /*
  63  * Local functions
  64  */
  65 static void      xprt_init(struct rpc_xprt *xprt, struct net *net);
  66 static void     xprt_request_init(struct rpc_task *, struct rpc_xprt *);
  67 static void     xprt_connect_status(struct rpc_task *task);
  68 static int      __xprt_get_cong(struct rpc_xprt *, struct rpc_task *);
  69 static void      xprt_destroy(struct rpc_xprt *xprt);
  70
  71 static DEFINE_SPINLOCK(xprt_list_lock);
  72 static LIST_HEAD(xprt_list);
  73
  74 /*
  75  * The transport code maintains an estimate on the maximum number of out-
  76  * standing RPC requests, using a smoothed version of the congestion
  77  * avoidance implemented in 44BSD. This is basically the Van Jacobson
  78  * congestion algorithm: If a retransmit occurs, the congestion window is
  79  * halved; otherwise, it is incremented by 1/cwnd when
  80  *
  81  *      -       a reply is received and
  82  *      -       a full number of requests are outstanding and
  83  *      -       the congestion window hasn't been updated recently.
  84  */
  85 #define RPC_CWNDSHIFT           (8U)
  86 #define RPC_CWNDSCALE           (1U << RPC_CWNDSHIFT)
  87 #define RPC_INITCWND            RPC_CWNDSCALE
  88 #define RPC_MAXCWND(xprt)       ((xprt)->max_reqs << RPC_CWNDSHIFT)
  89
  90 #define RPCXPRT_CONGESTED(xprt) ((xprt)->cong >= (xprt)->cwnd)
  91
  92 /**
  93  * xprt_register_transport - register a transport implementation
  94  * @transport: transport to register
  95  *
  96  * If a transport implementation is loaded as a kernel module, it can
  97  * call this interface to make itself known to the RPC client.
  98  *
  99  * Returns:
 100  * 0:           transport successfully registered
 101  * -EEXIST:     transport already registered
 102  * -EINVAL:     transport module being unloaded
 103  */
 104 int xprt_register_transport(struct xprt_class *transport)
 105 {
 106         struct xprt_class *t;
 107         int result;
 108
 109         result = -EEXIST;
 110         spin_lock(&xprt_list_lock);
 111         list_for_each_entry(t, &xprt_list, list) {
 112                 /* don't register the same transport class twice */
 113                 if (t->ident == transport->ident)
 114                         goto out;
 115         }
 116
 117         list_add_tail(&transport->list, &xprt_list);
 118         printk(KERN_INFO "RPC: Registered %s transport module.\n",
 119                transport->name);
 120         result = 0;
 121
 122 out:
 123         spin_unlock(&xprt_list_lock);
 124         return result;
 125 }
 126 EXPORT_SYMBOL_GPL(xprt_register_transport);
 127
 128 /**
 129  * xprt_unregister_transport - unregister a transport implementation
 130  * @transport: transport to unregister
 131  *
 132  * Returns:
 133  * 0:           transport successfully unregistered
 134  * -ENOENT:     transport never registered
 135  */
 136 int xprt_unregister_transport(struct xprt_class *transport)
 137 {
 138         struct xprt_class *t;
 139         int result;
 140
 141         result = 0;
 142         spin_lock(&xprt_list_lock);
 143         list_for_each_entry(t, &xprt_list, list) {
 144                 if (t == transport) {
 145                         printk(KERN_INFO
 146                                 "RPC: Unregistered %s transport module.\n",
 147                                 transport->name);
 148                         list_del_init(&transport->list);
 149                         goto out;
 150                 }
 151         }
 152         result = -ENOENT;
 153
 154 out:
 155         spin_unlock(&xprt_list_lock);
 156         return result;
 157 }
 158 EXPORT_SYMBOL_GPL(xprt_unregister_transport);
 159
 160 /**
 161  * xprt_load_transport - load a transport implementation
 162  * @transport_name: transport to load
 163  *
 164  * Returns:
 165  * 0:           transport successfully loaded
 166  * -ENOENT:     transport module not available
 167  */
 168 int xprt_load_transport(const char *transport_name)
 169 {
 170         struct xprt_class *t;
 171         int result;
 172
 173         result = 0;
 174         spin_lock(&xprt_list_lock);
 175         list_for_each_entry(t, &xprt_list, list) {
 176                 if (strcmp(t->name, transport_name) == 0) {
 177                         spin_unlock(&xprt_list_lock);
 178                         goto out;
 179                 }
 180         }
 181         spin_unlock(&xprt_list_lock);
 182         result = request_module("xprt%s", transport_name);
 183 out:
 184         return result;
 185 }
 186 EXPORT_SYMBOL_GPL(xprt_load_transport);
 187
 188 /**
 189  * xprt_reserve_xprt - serialize write access to transports
 190  * @task: task that is requesting access to the transport
 191  * @xprt: pointer to the target transport
 192  *
 193  * This prevents mixing the payload of separate requests, and prevents
 194  * transport connects from colliding with writes.  No congestion control
 195  * is provided.
 196  */
 197 int xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
 198 {
 199         struct rpc_rqst *req = task->tk_rqstp;
 200         int priority;
 201
 202         if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
 203                 if (task == xprt->snd_task)
 204                         return 1;
 205                 goto out_sleep;
 206         }
 207         xprt->snd_task = task;
 208         if (req != NULL)
 209                 req->rq_ntrans++;
 210
 211         return 1;
 212
 213 out_sleep:
 214         dprintk("RPC: %5u failed to lock transport %p\n",
 215                         task->tk_pid, xprt);
 216         task->tk_timeout = 0;
 217         task->tk_status = -EAGAIN;
 218         if (req == NULL)
 219                 priority = RPC_PRIORITY_LOW;
 220         else if (!req->rq_ntrans)
 221                 priority = RPC_PRIORITY_NORMAL;
 222         else
 223                 priority = RPC_PRIORITY_HIGH;
 224         rpc_sleep_on_priority(&xprt->sending, task, NULL, priority);
 225         return 0;
 226 }
 227 EXPORT_SYMBOL_GPL(xprt_reserve_xprt);
 228
 229 static void xprt_clear_locked(struct rpc_xprt *xprt)
 230 {
 231         xprt->snd_task = NULL;
 232         if (!test_bit(XPRT_CLOSE_WAIT, &xprt->state)) {
 233                 smp_mb__before_clear_bit();
 234                 clear_bit(XPRT_LOCKED, &xprt->state);
 235                 smp_mb__after_clear_bit();
 236         } else
 237                 queue_work(rpciod_workqueue, &xprt->task_cleanup);
 238 }
 239
 240 /*
 241  * xprt_reserve_xprt_cong - serialize write access to transports
 242  * @task: task that is requesting access to the transport
 243  *
 244  * Same as xprt_reserve_xprt, but Van Jacobson congestion control is
 245  * integrated into the decision of whether a request is allowed to be
 246  * woken up and given access to the transport.
 247  */
 248 int xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 249 {
 250         struct rpc_rqst *req = task->tk_rqstp;
 251         int priority;
 252
 253         if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
 254                 if (task == xprt->snd_task)
 255                         return 1;
 256                 goto out_sleep;
 257         }
 258         if (req == NULL) {
 259                 xprt->snd_task = task;
 260                 return 1;
 261         }
 262         if (__xprt_get_cong(xprt, task)) {
 263                 xprt->snd_task = task;
 264                 req->rq_ntrans++;
 265                 return 1;
 266         }
 267         xprt_clear_locked(xprt);
 268 out_sleep:
 269         dprintk("RPC: %5u failed to lock transport %p\n", task->tk_pid, xprt);
 270         task->tk_timeout = 0;
 271         task->tk_status = -EAGAIN;
 272         if (req == NULL)
 273                 priority = RPC_PRIORITY_LOW;
 274         else if (!req->rq_ntrans)
 275                 priority = RPC_PRIORITY_NORMAL;
 276         else
 277                 priority = RPC_PRIORITY_HIGH;
 278         rpc_sleep_on_priority(&xprt->sending, task, NULL, priority);
 279         return 0;
 280 }
 281 EXPORT_SYMBOL_GPL(xprt_reserve_xprt_cong);
 282
 283 static inline int xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 284 {
 285         int retval;
 286
 287         spin_lock_bh(&xprt->transport_lock);
 288         retval = xprt->ops->reserve_xprt(xprt, task);
 289         spin_unlock_bh(&xprt->transport_lock);
 290         return retval;
 291 }
 292
 293 static bool __xprt_lock_write_func(struct rpc_task *task, void *data)
 294 {
 295         struct rpc_xprt *xprt = data;
 296         struct rpc_rqst *req;
 297
 298         req = task->tk_rqstp;
 299         xprt->snd_task = task;
 300         if (req)
 301                 req->rq_ntrans++;
 302         return true;
 303 }
 304
 305 static void __xprt_lock_write_next(struct rpc_xprt *xprt)
 306 {
 307         if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 308                 return;
 309
 310         if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_func, xprt))
 311                 return;
 312         xprt_clear_locked(xprt);
 313 }
 314
 315 static bool __xprt_lock_write_cong_func(struct rpc_task *task, void *data)
 316 {
 317         struct rpc_xprt *xprt = data;
 318         struct rpc_rqst *req;
 319
 320         req = task->tk_rqstp;
 321         if (req == NULL) {
 322                 xprt->snd_task = task;
 323                 return true;
 324         }
 325         if (__xprt_get_cong(xprt, task)) {
 326                 xprt->snd_task = task;
 327                 req->rq_ntrans++;
 328                 return true;
 329         }
 330         return false;
 331 }
 332
 333 static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt)
 334 {
 335         if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 336                 return;
 337         if (RPCXPRT_CONGESTED(xprt))
 338                 goto out_unlock;
 339         if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_cong_func, xprt))
 340                 return;
 341 out_unlock:
 342         xprt_clear_locked(xprt);
 343 }
 344
 345 /**
 346  * xprt_release_xprt - allow other requests to use a transport
 347  * @xprt: transport with other tasks potentially waiting
 348  * @task: task that is releasing access to the transport
 349  *
 350  * Note that "task" can be NULL.  No congestion control is provided.
 351  */
 352 void xprt_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
 353 {
 354         if (xprt->snd_task == task) {
 355                 if (task != NULL) {
 356                         struct rpc_rqst *req = task->tk_rqstp;
 357                         if (req != NULL)
 358                                 req->rq_bytes_sent = 0;
 359                 }
 360                 xprt_clear_locked(xprt);
 361                 __xprt_lock_write_next(xprt);
 362         }
 363 }
 364 EXPORT_SYMBOL_GPL(xprt_release_xprt);
 365
 366 /**
 367  * xprt_release_xprt_cong - allow other requests to use a transport
 368  * @xprt: transport with other tasks potentially waiting
 369  * @task: task that is releasing access to the transport
 370  *
 371  * Note that "task" can be NULL.  Another task is awoken to use the
 372  * transport if the transport's congestion window allows it.
 373  */
 374 void xprt_release_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 375 {
 376         if (xprt->snd_task == task) {
 377                 if (task != NULL) {
 378                         struct rpc_rqst *req = task->tk_rqstp;
 379                         if (req != NULL)
 380                                 req->rq_bytes_sent = 0;
 381                 }
 382                 xprt_clear_locked(xprt);
 383                 __xprt_lock_write_next_cong(xprt);
 384         }
 385 }
 386 EXPORT_SYMBOL_GPL(xprt_release_xprt_cong);
 387
 388 static inline void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 389 {
 390         spin_lock_bh(&xprt->transport_lock);
 391         xprt->ops->release_xprt(xprt, task);
 392         spin_unlock_bh(&xprt->transport_lock);
 393 }
 394
 395 /*
 396  * Van Jacobson congestion avoidance. Check if the congestion window
 397  * overflowed. Put the task to sleep if this is the case.
 398  */
 399 static int
 400 __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 401 {
 402         struct rpc_rqst *req = task->tk_rqstp;
 403
 404         if (req->rq_cong)
 405                 return 1;
 406         dprintk("RPC: %5u xprt_cwnd_limited cong = %lu cwnd = %lu\n",
 407                         task->tk_pid, xprt->cong, xprt->cwnd);
 408         if (RPCXPRT_CONGESTED(xprt))
 409                 return 0;
 410         req->rq_cong = 1;
 411         xprt->cong += RPC_CWNDSCALE;
 412         return 1;
 413 }
 414
 415 /*
 416  * Adjust the congestion window, and wake up the next task
 417  * that has been sleeping due to congestion
 418  */
 419 static void
 420 __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
 421 {
 422         if (!req->rq_cong)
 423                 return;
 424         req->rq_cong = 0;
 425         xprt->cong -= RPC_CWNDSCALE;
 426         __xprt_lock_write_next_cong(xprt);
 427 }
 428
 429 /**
 430  * xprt_release_rqst_cong - housekeeping when request is complete
 431  * @task: RPC request that recently completed
 432  *
 433  * Useful for transports that require congestion control.
 434  */
 435 void xprt_release_rqst_cong(struct rpc_task *task)
 436 {
 437         struct rpc_rqst *req = task->tk_rqstp;
 438
 439         __xprt_put_cong(req->rq_xprt, req);
 440 }
 441 EXPORT_SYMBOL_GPL(xprt_release_rqst_cong);
 442
 443 /**
 444  * xprt_adjust_cwnd - adjust transport congestion window
 445  * @xprt: pointer to xprt
 446  * @task: recently completed RPC request used to adjust window
 447  * @result: result code of completed RPC request
 448  *
 449  * We use a time-smoothed congestion estimator to avoid heavy oscillation.
 450  */
 451 void xprt_adjust_cwnd(struct rpc_xprt *xprt, struct rpc_task *task, int result)
 452 {
 453         struct rpc_rqst *req = task->tk_rqstp;
 454         unsigned long cwnd = xprt->cwnd;
 455
 456         if (result >= 0 && cwnd <= xprt->cong) {
 457                 /* The (cwnd >> 1) term makes sure
 458                  * the result gets rounded properly. */
 459                 cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
 460                 if (cwnd > RPC_MAXCWND(xprt))
 461                         cwnd = RPC_MAXCWND(xprt);
 462                 __xprt_lock_write_next_cong(xprt);
 463         } else if (result == -ETIMEDOUT) {
 464                 cwnd >>= 1;
 465                 if (cwnd < RPC_CWNDSCALE)
 466                         cwnd = RPC_CWNDSCALE;
 467         }
 468         dprintk("RPC:       cong %ld, cwnd was %ld, now %ld\n",
 469                         xprt->cong, xprt->cwnd, cwnd);
 470         xprt->cwnd = cwnd;
 471         __xprt_put_cong(xprt, req);
 472 }
 473 EXPORT_SYMBOL_GPL(xprt_adjust_cwnd);
 474
 475 /**
 476  * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue
 477  * @xprt: transport with waiting tasks
 478  * @status: result code to plant in each task before waking it
 479  *
 480  */
 481 void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status)
 482 {
 483         if (status < 0)
 484                 rpc_wake_up_status(&xprt->pending, status);
 485         else
 486                 rpc_wake_up(&xprt->pending);
 487 }
 488 EXPORT_SYMBOL_GPL(xprt_wake_pending_tasks);
 489
 490 /**
 491  * xprt_wait_for_buffer_space - wait for transport output buffer to clear
 492  * @task: task to be put to sleep
 493  * @action: function pointer to be executed after wait
 494  *
 495  * Note that we only set the timer for the case of RPC_IS_SOFT(), since
 496  * we don't in general want to force a socket disconnection due to
 497  * an incomplete RPC call transmission.
 498  */
 499 void xprt_wait_for_buffer_space(struct rpc_task *task, rpc_action action)
 500 {
 501         struct rpc_rqst *req = task->tk_rqstp;
 502         struct rpc_xprt *xprt = req->rq_xprt;
 503
 504         task->tk_timeout = RPC_IS_SOFT(task) ? req->rq_timeout : 0;
 505         rpc_sleep_on(&xprt->pending, task, action);
 506 }
 507 EXPORT_SYMBOL_GPL(xprt_wait_for_buffer_space);
 508
 509 /**
 510  * xprt_write_space - wake the task waiting for transport output buffer space
 511  * @xprt: transport with waiting tasks
 512  *
 513  * Can be called in a soft IRQ context, so xprt_write_space never sleeps.
 514  */
 515 void xprt_write_space(struct rpc_xprt *xprt)
 516 {
 517         spin_lock_bh(&xprt->transport_lock);
 518         if (xprt->snd_task) {
 519                 dprintk("RPC:       write space: waking waiting task on "
 520                                 "xprt %p\n", xprt);
 521                 rpc_wake_up_queued_task(&xprt->pending, xprt->snd_task);
 522         }
 523         spin_unlock_bh(&xprt->transport_lock);
 524 }
 525 EXPORT_SYMBOL_GPL(xprt_write_space);
 526
 527 /**
 528  * xprt_set_retrans_timeout_def - set a request's retransmit timeout
 529  * @task: task whose timeout is to be set
 530  *
 531  * Set a request's retransmit timeout based on the transport's
 532  * default timeout parameters.  Used by transports that don't adjust
 533  * the retransmit timeout based on round-trip time estimation.
 534  */
 535 void xprt_set_retrans_timeout_def(struct rpc_task *task)
 536 {
 537         task->tk_timeout = task->tk_rqstp->rq_timeout;
 538 }
 539 EXPORT_SYMBOL_GPL(xprt_set_retrans_timeout_def);
 540
 541 /**
 542  * xprt_set_retrans_timeout_rtt - set a request's retransmit timeout
 543  * @task: task whose timeout is to be set
 544  *
 545  * Set a request's retransmit timeout using the RTT estimator.
 546  */
 547 void xprt_set_retrans_timeout_rtt(struct rpc_task *task)
 548 {
 549         int timer = task->tk_msg.rpc_proc->p_timer;
 550         struct rpc_clnt *clnt = task->tk_client;
 551         struct rpc_rtt *rtt = clnt->cl_rtt;
 552         struct rpc_rqst *req = task->tk_rqstp;
 553         unsigned long max_timeout = clnt->cl_timeout->to_maxval;
 554
 555         task->tk_timeout = rpc_calc_rto(rtt, timer);
 556         task->tk_timeout <<= rpc_ntimeo(rtt, timer) + req->rq_retries;
 557         if (task->tk_timeout > max_timeout || task->tk_timeout == 0)
 558                 task->tk_timeout = max_timeout;
 559 }
 560 EXPORT_SYMBOL_GPL(xprt_set_retrans_timeout_rtt);
 561
 562 static void xprt_reset_majortimeo(struct rpc_rqst *req)
 563 {
 564         const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
 565
 566         req->rq_majortimeo = req->rq_timeout;
 567         if (to->to_exponential)
 568                 req->rq_majortimeo <<= to->to_retries;
 569         else
 570                 req->rq_majortimeo += to->to_increment * to->to_retries;
 571         if (req->rq_majortimeo > to->to_maxval || req->rq_majortimeo == 0)
 572                 req->rq_majortimeo = to->to_maxval;
 573         req->rq_majortimeo += jiffies;
 574 }
 575
 576 /**
 577  * xprt_adjust_timeout - adjust timeout values for next retransmit
 578  * @req: RPC request containing parameters to use for the adjustment
 579  *
 580  */
 581 int xprt_adjust_timeout(struct rpc_rqst *req)
 582 {
 583         struct rpc_xprt *xprt = req->rq_xprt;
 584         const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
 585         int status = 0;
 586
 587         if (time_before(jiffies, req->rq_majortimeo)) {
 588                 if (to->to_exponential)
 589                         req->rq_timeout <<= 1;
 590                 else
 591                         req->rq_timeout += to->to_increment;
 592                 if (to->to_maxval && req->rq_timeout >= to->to_maxval)
 593                         req->rq_timeout = to->to_maxval;
 594                 req->rq_retries++;
 595         } else {
 596                 req->rq_timeout = to->to_initval;
 597                 req->rq_retries = 0;
 598                 xprt_reset_majortimeo(req);
 599                 /* Reset the RTT counters == "slow start" */
 600                 spin_lock_bh(&xprt->transport_lock);
 601                 rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
 602                 spin_unlock_bh(&xprt->transport_lock);
 603                 status = -ETIMEDOUT;
 604         }
 605
 606         if (req->rq_timeout == 0) {
 607                 printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
 608                 req->rq_timeout = 5 * HZ;
 609         }
 610         return status;
 611 }
 612
 613 static void xprt_autoclose(struct work_struct *work)
 614 {
 615         struct rpc_xprt *xprt =
 616                 container_of(work, struct rpc_xprt, task_cleanup);
 617
 618         xprt->ops->close(xprt);
 619         clear_bit(XPRT_CLOSE_WAIT, &xprt->state);
 620         xprt_release_write(xprt, NULL);
 621 }
 622
 623 /**
 624  * xprt_disconnect_done - mark a transport as disconnected
 625  * @xprt: transport to flag for disconnect
 626  *
 627  */
 628 void xprt_disconnect_done(struct rpc_xprt *xprt)
 629 {
 630         dprintk("RPC:       disconnected transport %p\n", xprt);
 631         spin_lock_bh(&xprt->transport_lock);
 632         xprt_clear_connected(xprt);
 633         xprt_wake_pending_tasks(xprt, -EAGAIN);
 634         spin_unlock_bh(&xprt->transport_lock);
 635 }
 636 EXPORT_SYMBOL_GPL(xprt_disconnect_done);
 637
 638 /**
 639  * xprt_force_disconnect - force a transport to disconnect
 640  * @xprt: transport to disconnect
 641  *
 642  */
 643 void xprt_force_disconnect(struct rpc_xprt *xprt)
 644 {
 645         /* Don't race with the test_bit() in xprt_clear_locked() */
 646         spin_lock_bh(&xprt->transport_lock);
 647         set_bit(XPRT_CLOSE_WAIT, &xprt->state);
 648         /* Try to schedule an autoclose RPC call */
 649         if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
 650                 queue_work(rpciod_workqueue, &xprt->task_cleanup);
 651         xprt_wake_pending_tasks(xprt, -EAGAIN);
 652         spin_unlock_bh(&xprt->transport_lock);
 653 }
 654
 655 /**
 656  * xprt_conditional_disconnect - force a transport to disconnect
 657  * @xprt: transport to disconnect
 658  * @cookie: 'connection cookie'
 659  *
 660  * This attempts to break the connection if and only if 'cookie' matches
 661  * the current transport 'connection cookie'. It ensures that we don't
 662  * try to break the connection more than once when we need to retransmit
 663  * a batch of RPC requests.
 664  *
 665  */
 666 void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie)
 667 {
 668         /* Don't race with the test_bit() in xprt_clear_locked() */
 669         spin_lock_bh(&xprt->transport_lock);
 670         if (cookie != xprt->connect_cookie)
 671                 goto out;
 672         if (test_bit(XPRT_CLOSING, &xprt->state) || !xprt_connected(xprt))
 673                 goto out;
 674         set_bit(XPRT_CLOSE_WAIT, &xprt->state);
 675         /* Try to schedule an autoclose RPC call */
 676         if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
 677                 queue_work(rpciod_workqueue, &xprt->task_cleanup);
 678         xprt_wake_pending_tasks(xprt, -EAGAIN);
 679 out:
 680         spin_unlock_bh(&xprt->transport_lock);
 681 }
 682
 683 static void
 684 xprt_init_autodisconnect(unsigned long data)
 685 {
 686         struct rpc_xprt *xprt = (struct rpc_xprt *)data;
 687
 688         spin_lock(&xprt->transport_lock);
 689         if (!list_empty(&xprt->recv))
 690                 goto out_abort;
 691         if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
 692                 goto out_abort;
 693         spin_unlock(&xprt->transport_lock);
 694         set_bit(XPRT_CONNECTION_CLOSE, &xprt->state);
 695         queue_work(rpciod_workqueue, &xprt->task_cleanup);
 696         return;
 697 out_abort:
 698         spin_unlock(&xprt->transport_lock);
 699 }
 700
 701 /**
 702  * xprt_connect - schedule a transport connect operation
 703  * @task: RPC task that is requesting the connect
 704  *
 705  */
 706 void xprt_connect(struct rpc_task *task)
 707 {
 708         struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt;
 709
 710         dprintk("RPC: %5u xprt_connect xprt %p %s connected\n", task->tk_pid,
 711                         xprt, (xprt_connected(xprt) ? "is" : "is not"));
 712
 713         if (!xprt_bound(xprt)) {
 714                 task->tk_status = -EAGAIN;
 715                 return;
 716         }
 717         if (!xprt_lock_write(xprt, task))
 718                 return;
 719
 720         if (test_and_clear_bit(XPRT_CLOSE_WAIT, &xprt->state))
 721                 xprt->ops->close(xprt);
 722
 723         if (xprt_connected(xprt))
 724                 xprt_release_write(xprt, task);
 725         else {
 726                 task->tk_rqstp->rq_bytes_sent = 0;
 727                 task->tk_timeout = task->tk_rqstp->rq_timeout;
 728                 rpc_sleep_on(&xprt->pending, task, xprt_connect_status);
 729
 730                 if (test_bit(XPRT_CLOSING, &xprt->state))
 731                         return;
 732                 if (xprt_test_and_set_connecting(xprt))
 733                         return;
 734                 xprt->stat.connect_start = jiffies;
 735                 xprt->ops->connect(xprt, task);
 736         }
 737 }
 738
 739 static void xprt_connect_status(struct rpc_task *task)
 740 {
 741         struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt;
 742
 743         if (task->tk_status == 0) {
 744                 xprt->stat.connect_count++;
 745                 xprt->stat.connect_time += (long)jiffies - xprt->stat.connect_start;
 746                 dprintk("RPC: %5u xprt_connect_status: connection established\n",
 747                                 task->tk_pid);
 748                 return;
 749         }
 750
 751         switch (task->tk_status) {
 752         case -EAGAIN:
 753                 dprintk("RPC: %5u xprt_connect_status: retrying\n", task->tk_pid);
 754                 break;
 755         case -ETIMEDOUT:
 756                 dprintk("RPC: %5u xprt_connect_status: connect attempt timed "
 757                                 "out\n", task->tk_pid);
 758                 break;
 759         default:
 760                 dprintk("RPC: %5u xprt_connect_status: error %d connecting to "
 761                                 "server %s\n", task->tk_pid, -task->tk_status,
 762                                 xprt->servername);
 763                 xprt_release_write(xprt, task);
 764                 task->tk_status = -EIO;
 765         }
 766 }
 767
 768 /**
 769  * xprt_lookup_rqst - find an RPC request corresponding to an XID
 770  * @xprt: transport on which the original request was transmitted
 771  * @xid: RPC XID of incoming reply
 772  *
 773  */
 774 struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, __be32 xid)
 775 {
 776         struct rpc_rqst *entry;
 777
 778         list_for_each_entry(entry, &xprt->recv, rq_list)
 779                 if (entry->rq_xid == xid)
 780                         return entry;
 781
 782         dprintk("RPC:       xprt_lookup_rqst did not find xid %08x\n",
 783                         ntohl(xid));
 784         xprt->stat.bad_xids++;
 785         return NULL;
 786 }
 787 EXPORT_SYMBOL_GPL(xprt_lookup_rqst);
 788
 789 static void xprt_update_rtt(struct rpc_task *task)
 790 {
 791         struct rpc_rqst *req = task->tk_rqstp;
 792         struct rpc_rtt *rtt = task->tk_client->cl_rtt;
 793         unsigned int timer = task->tk_msg.rpc_proc->p_timer;
 794         long m = usecs_to_jiffies(ktime_to_us(req->rq_rtt));
 795
 796         if (timer) {
 797                 if (req->rq_ntrans == 1)
 798                         rpc_update_rtt(rtt, timer, m);
 799                 rpc_set_timeo(rtt, timer, req->rq_ntrans - 1);
 800         }
 801 }
 802
 803 /**
 804  * xprt_complete_rqst - called when reply processing is complete
 805  * @task: RPC request that recently completed
 806  * @copied: actual number of bytes received from the transport
 807  *
 808  * Caller holds transport lock.
 809  */
 810 void xprt_complete_rqst(struct rpc_task *task, int copied)
 811 {
 812         struct rpc_rqst *req = task->tk_rqstp;
 813         struct rpc_xprt *xprt = req->rq_xprt;
 814
 815         dprintk("RPC: %5u xid %08x complete (%d bytes received)\n",
 816                         task->tk_pid, ntohl(req->rq_xid), copied);
 817
 818         xprt->stat.recvs++;
 819         req->rq_rtt = ktime_sub(ktime_get(), req->rq_xtime);
 820         if (xprt->ops->timer != NULL)
 821                 xprt_update_rtt(task);
 822
 823         list_del_init(&req->rq_list);
 824         req->rq_private_buf.len = copied;
 825         /* Ensure all writes are done before we update */
 826         /* req->rq_reply_bytes_recvd */
 827         smp_wmb();
 828         req->rq_reply_bytes_recvd = copied;
 829         rpc_wake_up_queued_task(&xprt->pending, task);
 830 }
 831 EXPORT_SYMBOL_GPL(xprt_complete_rqst);
 832
 833 static void xprt_timer(struct rpc_task *task)
 834 {
 835         struct rpc_rqst *req = task->tk_rqstp;
 836         struct rpc_xprt *xprt = req->rq_xprt;
 837
 838         if (task->tk_status != -ETIMEDOUT)
 839                 return;
 840         dprintk("RPC: %5u xprt_timer\n", task->tk_pid);
 841
 842         spin_lock_bh(&xprt->transport_lock);
 843         if (!req->rq_reply_bytes_recvd) {
 844                 if (xprt->ops->timer)
 845                         xprt->ops->timer(xprt, task);
 846         } else
 847                 task->tk_status = 0;
 848         spin_unlock_bh(&xprt->transport_lock);
 849 }
 850
 851 static inline int xprt_has_timer(struct rpc_xprt *xprt)
 852 {
 853         return xprt->idle_timeout != 0;
 854 }
 855
 856 /**
 857  * xprt_prepare_transmit - reserve the transport before sending a request
 858  * @task: RPC task about to send a request
 859  *
 860  */
 861 bool xprt_prepare_transmit(struct rpc_task *task)
 862 {
 863         struct rpc_rqst *req = task->tk_rqstp;
 864         struct rpc_xprt *xprt = req->rq_xprt;
 865         bool ret = false;
 866
 867         dprintk("RPC: %5u xprt_prepare_transmit\n", task->tk_pid);
 868
 869         spin_lock_bh(&xprt->transport_lock);
 870         if (!req->rq_bytes_sent) {
 871                 if (req->rq_reply_bytes_recvd) {
 872                         task->tk_status = req->rq_reply_bytes_recvd;
 873                         goto out_unlock;
 874                 }
 875                 if ((task->tk_flags & RPC_TASK_NO_RETRANS_TIMEOUT)
 876                     && xprt_connected(xprt)
 877                     && req->rq_connect_cookie == xprt->connect_cookie) {
 878                         xprt->ops->set_retrans_timeout(task);
 879                         rpc_sleep_on(&xprt->pending, task, xprt_timer);
 880                         goto out_unlock;
 881                 }
 882         }
 883         if (!xprt->ops->reserve_xprt(xprt, task)) {
 884                 task->tk_status = -EAGAIN;
 885                 goto out_unlock;
 886         }
 887         ret = true;
 888 out_unlock:
 889         spin_unlock_bh(&xprt->transport_lock);
 890         return ret;
 891 }
 892
 893 void xprt_end_transmit(struct rpc_task *task)
 894 {
 895         xprt_release_write(task->tk_rqstp->rq_xprt, task);
 896 }
 897
 898 /**
 899  * xprt_transmit - send an RPC request on a transport
 900  * @task: controlling RPC task
 901  *
 902  * We have to copy the iovec because sendmsg fiddles with its contents.
 903  */
 904 void xprt_transmit(struct rpc_task *task)
 905 {
 906         struct rpc_rqst *req = task->tk_rqstp;
 907         struct rpc_xprt *xprt = req->rq_xprt;
 908         int status, numreqs;
 909
 910         dprintk("RPC: %5u xprt_transmit(%u)\n", task->tk_pid, req->rq_slen);
 911
 912         if (!req->rq_reply_bytes_recvd) {
 913                 if (list_empty(&req->rq_list) && rpc_reply_expected(task)) {
 914                         /*
 915                          * Add to the list only if we're expecting a reply
 916                          */
 917                         spin_lock_bh(&xprt->transport_lock);
 918                         /* Update the softirq receive buffer */
 919                         memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
 920                                         sizeof(req->rq_private_buf));
 921                         /* Add request to the receive list */
 922                         list_add_tail(&req->rq_list, &xprt->recv);
 923                         spin_unlock_bh(&xprt->transport_lock);
 924                         xprt_reset_majortimeo(req);
 925                         /* Turn off autodisconnect */
 926                         del_singleshot_timer_sync(&xprt->timer);
 927                 }
 928         } else if (!req->rq_bytes_sent)
 929                 return;
 930
 931         req->rq_xtime = ktime_get();
 932         status = xprt->ops->send_request(task);
 933         if (status != 0) {
 934                 task->tk_status = status;
 935                 return;
 936         }
 937
 938         dprintk("RPC: %5u xmit complete\n", task->tk_pid);
 939         task->tk_flags |= RPC_TASK_SENT;
 940         spin_lock_bh(&xprt->transport_lock);
 941
 942         xprt->ops->set_retrans_timeout(task);
 943
 944         numreqs = atomic_read(&xprt->num_reqs);
 945         if (numreqs > xprt->stat.max_slots)
 946                 xprt->stat.max_slots = numreqs;
 947         xprt->stat.sends++;
 948         xprt->stat.req_u += xprt->stat.sends - xprt->stat.recvs;
 949         xprt->stat.bklog_u += xprt->backlog.qlen;
 950         xprt->stat.sending_u += xprt->sending.qlen;
 951         xprt->stat.pending_u += xprt->pending.qlen;
 952
 953         /* Don't race with disconnect */
 954         if (!xprt_connected(xprt))
 955                 task->tk_status = -ENOTCONN;
 956         else {
 957                 /*
 958                  * Sleep on the pending queue since
 959                  * we're expecting a reply.
 960                  */
 961                 if (!req->rq_reply_bytes_recvd && rpc_reply_expected(task))
 962                         rpc_sleep_on(&xprt->pending, task, xprt_timer);
 963                 req->rq_connect_cookie = xprt->connect_cookie;
 964         }
 965         spin_unlock_bh(&xprt->transport_lock);
 966 }
 967
 968 static void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task)
 969 {
 970         set_bit(XPRT_CONGESTED, &xprt->state);
 971         rpc_sleep_on(&xprt->backlog, task, NULL);
 972 }
 973
 974 static void xprt_wake_up_backlog(struct rpc_xprt *xprt)
 975 {
 976         if (rpc_wake_up_next(&xprt->backlog) == NULL)
 977                 clear_bit(XPRT_CONGESTED, &xprt->state);
 978 }
 979
 980 static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task)
 981 {
 982         bool ret = false;
 983
 984         if (!test_bit(XPRT_CONGESTED, &xprt->state))
 985                 goto out;
 986         spin_lock(&xprt->reserve_lock);
 987         if (test_bit(XPRT_CONGESTED, &xprt->state)) {
 988                 rpc_sleep_on(&xprt->backlog, task, NULL);
 989                 ret = true;
 990         }
 991         spin_unlock(&xprt->reserve_lock);
 992 out:
 993         return ret;
 994 }
 995
 996 static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt, gfp_t gfp_flags)
 997 {
 998         struct rpc_rqst *req = ERR_PTR(-EAGAIN);
 999
1000         if (!atomic_add_unless(&xprt->num_reqs, 1, xprt->max_reqs))
1001                 goto out;
1002         req = kzalloc(sizeof(struct rpc_rqst), gfp_flags);
1003         if (req != NULL)
1004                 goto out;
1005         atomic_dec(&xprt->num_reqs);
1006         req = ERR_PTR(-ENOMEM);
1007 out:
1008         return req;
1009 }
1010
1011 static bool xprt_dynamic_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1012 {
1013         if (atomic_add_unless(&xprt->num_reqs, -1, xprt->min_reqs)) {
1014                 kfree(req);
1015                 return true;
1016         }
1017         return false;
1018 }
1019
1020 void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
1021 {
1022         struct rpc_rqst *req;
1023
1024         spin_lock(&xprt->reserve_lock);
1025         if (!list_empty(&xprt->free)) {
1026                 req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
1027                 list_del(&req->rq_list);
1028                 goto out_init_req;
1029         }
1030         req = xprt_dynamic_alloc_slot(xprt, GFP_NOWAIT|__GFP_NOWARN);
1031         if (!IS_ERR(req))
1032                 goto out_init_req;
1033         switch (PTR_ERR(req)) {
1034         case -ENOMEM:
1035                 dprintk("RPC:       dynamic allocation of request slot "
1036                                 "failed! Retrying\n");
1037                 task->tk_status = -ENOMEM;
1038                 break;
1039         case -EAGAIN:
1040                 xprt_add_backlog(xprt, task);
1041                 dprintk("RPC:       waiting for request slot\n");
1042         default:
1043                 task->tk_status = -EAGAIN;
1044         }
1045         spin_unlock(&xprt->reserve_lock);
1046         return;
1047 out_init_req:
1048         task->tk_status = 0;
1049         task->tk_rqstp = req;
1050         xprt_request_init(task, xprt);
1051         spin_unlock(&xprt->reserve_lock);
1052 }
1053 EXPORT_SYMBOL_GPL(xprt_alloc_slot);
1054
1055 void xprt_lock_and_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
1056 {
1057         /* Note: grabbing the xprt_lock_write() ensures that we throttle
1058          * new slot allocation if the transport is congested (i.e. when
1059          * reconnecting a stream transport or when out of socket write
1060          * buffer space).
1061          */
1062         if (xprt_lock_write(xprt, task)) {
1063                 xprt_alloc_slot(xprt, task);
1064                 xprt_release_write(xprt, task);
1065         }
1066 }
1067 EXPORT_SYMBOL_GPL(xprt_lock_and_alloc_slot);
1068
1069 static void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1070 {
1071         spin_lock(&xprt->reserve_lock);
1072         if (!xprt_dynamic_free_slot(xprt, req)) {
1073                 memset(req, 0, sizeof(*req));   /* mark unused */
1074                 list_add(&req->rq_list, &xprt->free);
1075         }
1076         xprt_wake_up_backlog(xprt);
1077         spin_unlock(&xprt->reserve_lock);
1078 }
1079
1080 static void xprt_free_all_slots(struct rpc_xprt *xprt)
1081 {
1082         struct rpc_rqst *req;
1083         while (!list_empty(&xprt->free)) {
1084                 req = list_first_entry(&xprt->free, struct rpc_rqst, rq_list);
1085                 list_del(&req->rq_list);
1086                 kfree(req);
1087         }
1088 }
1089
1090 struct rpc_xprt *xprt_alloc(struct net *net, size_t size,
1091                 unsigned int num_prealloc,
1092                 unsigned int max_alloc)
1093 {
1094         struct rpc_xprt *xprt;
1095         struct rpc_rqst *req;
1096         int i;
1097
1098         xprt = kzalloc(size, GFP_KERNEL);
1099         if (xprt == NULL)
1100                 goto out;
1101
1102         xprt_init(xprt, net);
1103
1104         for (i = 0; i < num_prealloc; i++) {
1105                 req = kzalloc(sizeof(struct rpc_rqst), GFP_KERNEL);
1106                 if (!req)
1107                         break;
1108                 list_add(&req->rq_list, &xprt->free);
1109         }
1110         if (i < num_prealloc)
1111                 goto out_free;
1112         if (max_alloc > num_prealloc)
1113                 xprt->max_reqs = max_alloc;
1114         else
1115                 xprt->max_reqs = num_prealloc;
1116         xprt->min_reqs = num_prealloc;
1117         atomic_set(&xprt->num_reqs, num_prealloc);
1118
1119         return xprt;
1120
1121 out_free:
1122         xprt_free(xprt);
1123 out:
1124         return NULL;
1125 }
1126 EXPORT_SYMBOL_GPL(xprt_alloc);
1127
1128 void xprt_free(struct rpc_xprt *xprt)
1129 {
1130         put_net(xprt->xprt_net);
1131         xprt_free_all_slots(xprt);
1132         kfree(xprt);
1133 }
1134 EXPORT_SYMBOL_GPL(xprt_free);
1135
1136 /**
1137  * xprt_reserve - allocate an RPC request slot
1138  * @task: RPC task requesting a slot allocation
1139  *
1140  * If the transport is marked as being congested, or if no more
1141  * slots are available, place the task on the transport's
1142  * backlog queue.
1143  */
1144 void xprt_reserve(struct rpc_task *task)
1145 {
1146         struct rpc_xprt *xprt;
1147
1148         task->tk_status = 0;
1149         if (task->tk_rqstp != NULL)
1150                 return;
1151
1152         task->tk_timeout = 0;
1153         task->tk_status = -EAGAIN;
1154         rcu_read_lock();
1155         xprt = rcu_dereference(task->tk_client->cl_xprt);
1156         if (!xprt_throttle_congested(xprt, task))
1157                 xprt->ops->alloc_slot(xprt, task);
1158         rcu_read_unlock();
1159 }
1160
1161 /**
1162  * xprt_retry_reserve - allocate an RPC request slot
1163  * @task: RPC task requesting a slot allocation
1164  *
1165  * If no more slots are available, place the task on the transport's
1166  * backlog queue.
1167  * Note that the only difference with xprt_reserve is that we now
1168  * ignore the value of the XPRT_CONGESTED flag.
1169  */
1170 void xprt_retry_reserve(struct rpc_task *task)
1171 {
1172         struct rpc_xprt *xprt;
1173
1174         task->tk_status = 0;
1175         if (task->tk_rqstp != NULL)
1176                 return;
1177
1178         task->tk_timeout = 0;
1179         task->tk_status = -EAGAIN;
1180         rcu_read_lock();
1181         xprt = rcu_dereference(task->tk_client->cl_xprt);
1182         xprt->ops->alloc_slot(xprt, task);
1183         rcu_read_unlock();
1184 }
1185
1186 static inline __be32 xprt_alloc_xid(struct rpc_xprt *xprt)
1187 {
1188         return (__force __be32)xprt->xid++;
1189 }
1190
1191 static inline void xprt_init_xid(struct rpc_xprt *xprt)
1192 {
1193         xprt->xid = net_random();
1194 }
1195
1196 static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt)
1197 {
1198         struct rpc_rqst *req = task->tk_rqstp;
1199
1200         INIT_LIST_HEAD(&req->rq_list);
1201         req->rq_timeout = task->tk_client->cl_timeout->to_initval;
1202         req->rq_task    = task;
1203         req->rq_xprt    = xprt;
1204         req->rq_buffer  = NULL;
1205         req->rq_xid     = xprt_alloc_xid(xprt);
1206         req->rq_connect_cookie = xprt->connect_cookie - 1;
1207         req->rq_bytes_sent = 0;
1208         req->rq_snd_buf.len = 0;
1209         req->rq_snd_buf.buflen = 0;
1210         req->rq_rcv_buf.len = 0;
1211         req->rq_rcv_buf.buflen = 0;
1212         req->rq_release_snd_buf = NULL;
1213         xprt_reset_majortimeo(req);
1214         dprintk("RPC: %5u reserved req %p xid %08x\n", task->tk_pid,
1215                         req, ntohl(req->rq_xid));
1216 }
1217
1218 /**
1219  * xprt_release - release an RPC request slot
1220  * @task: task which is finished with the slot
1221  *
1222  */
1223 void xprt_release(struct rpc_task *task)
1224 {
1225         struct rpc_xprt *xprt;
1226         struct rpc_rqst *req = task->tk_rqstp;
1227
1228         if (req == NULL) {
1229                 if (task->tk_client) {
1230                         rcu_read_lock();
1231                         xprt = rcu_dereference(task->tk_client->cl_xprt);
1232                         if (xprt->snd_task == task)
1233                                 xprt_release_write(xprt, task);
1234                         rcu_read_unlock();
1235                 }
1236                 return;
1237         }
1238
1239         xprt = req->rq_xprt;
1240         if (task->tk_ops->rpc_count_stats != NULL)
1241                 task->tk_ops->rpc_count_stats(task, task->tk_calldata);
1242         else if (task->tk_client)
1243                 rpc_count_iostats(task, task->tk_client->cl_metrics);
1244         spin_lock_bh(&xprt->transport_lock);
1245         xprt->ops->release_xprt(xprt, task);
1246         if (xprt->ops->release_request)
1247                 xprt->ops->release_request(task);
1248         if (!list_empty(&req->rq_list))
1249                 list_del(&req->rq_list);
1250         xprt->last_used = jiffies;
1251         if (list_empty(&xprt->recv) && xprt_has_timer(xprt))
1252                 mod_timer(&xprt->timer,
1253                                 xprt->last_used + xprt->idle_timeout);
1254         spin_unlock_bh(&xprt->transport_lock);
1255         if (req->rq_buffer)
1256                 xprt->ops->buf_free(req->rq_buffer);
1257         if (req->rq_cred != NULL)
1258                 put_rpccred(req->rq_cred);
1259         task->tk_rqstp = NULL;
1260         if (req->rq_release_snd_buf)
1261                 req->rq_release_snd_buf(req);
1262
1263         dprintk("RPC: %5u release request %p\n", task->tk_pid, req);
1264         if (likely(!bc_prealloc(req)))
1265                 xprt_free_slot(xprt, req);
1266         else
1267                 xprt_free_bc_request(req);
1268 }
1269
1270 static void xprt_init(struct rpc_xprt *xprt, struct net *net)
1271 {
1272         atomic_set(&xprt->count, 1);
1273
1274         spin_lock_init(&xprt->transport_lock);
1275         spin_lock_init(&xprt->reserve_lock);
1276
1277         INIT_LIST_HEAD(&xprt->free);
1278         INIT_LIST_HEAD(&xprt->recv);
1279 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
1280         spin_lock_init(&xprt->bc_pa_lock);
1281         INIT_LIST_HEAD(&xprt->bc_pa_list);
1282 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
1283
1284         xprt->last_used = jiffies;
1285         xprt->cwnd = RPC_INITCWND;
1286         xprt->bind_index = 0;
1287
1288         rpc_init_wait_queue(&xprt->binding, "xprt_binding");
1289         rpc_init_wait_queue(&xprt->pending, "xprt_pending");
1290         rpc_init_priority_wait_queue(&xprt->sending, "xprt_sending");
1291         rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
1292
1293         xprt_init_xid(xprt);
1294
1295         xprt->xprt_net = get_net(net);
1296 }
1297
1298 /**
1299  * xprt_create_transport - create an RPC transport
1300  * @args: rpc transport creation arguments
1301  *
1302  */
1303 struct rpc_xprt *xprt_create_transport(struct xprt_create *args)
1304 {
1305         struct rpc_xprt *xprt;
1306         struct xprt_class *t;
1307
1308         spin_lock(&xprt_list_lock);
1309         list_for_each_entry(t, &xprt_list, list) {
1310                 if (t->ident == args->ident) {
1311                         spin_unlock(&xprt_list_lock);
1312                         goto found;
1313                 }
1314         }
1315         spin_unlock(&xprt_list_lock);
1316         printk(KERN_ERR "RPC: transport (%d) not supported\n", args->ident);
1317         return ERR_PTR(-EIO);
1318
1319 found:
1320         xprt = t->setup(args);
1321         if (IS_ERR(xprt)) {
1322                 dprintk("RPC:       xprt_create_transport: failed, %ld\n",
1323                                 -PTR_ERR(xprt));
1324                 goto out;
1325         }
1326         if (args->flags & XPRT_CREATE_NO_IDLE_TIMEOUT)
1327                 xprt->idle_timeout = 0;
1328         INIT_WORK(&xprt->task_cleanup, xprt_autoclose);
1329         if (xprt_has_timer(xprt))
1330                 setup_timer(&xprt->timer, xprt_init_autodisconnect,
1331                             (unsigned long)xprt);
1332         else
1333                 init_timer(&xprt->timer);
1334
1335         if (strlen(args->servername) > RPC_MAXNETNAMELEN) {
1336                 xprt_destroy(xprt);
1337                 return ERR_PTR(-EINVAL);
1338         }
1339         xprt->servername = kstrdup(args->servername, GFP_KERNEL);
1340         if (xprt->servername == NULL) {
1341                 xprt_destroy(xprt);
1342                 return ERR_PTR(-ENOMEM);
1343         }
1344
1345         dprintk("RPC:       created transport %p with %u slots\n", xprt,
1346                         xprt->max_reqs);
1347 out:
1348         return xprt;
1349 }
1350
1351 /**
1352  * xprt_destroy - destroy an RPC transport, killing off all requests.
1353  * @xprt: transport to destroy
1354  *
1355  */
1356 static void xprt_destroy(struct rpc_xprt *xprt)
1357 {
1358         dprintk("RPC:       destroying transport %p\n", xprt);
1359         del_timer_sync(&xprt->timer);
1360
1361         rpc_destroy_wait_queue(&xprt->binding);
1362         rpc_destroy_wait_queue(&xprt->pending);
1363         rpc_destroy_wait_queue(&xprt->sending);
1364         rpc_destroy_wait_queue(&xprt->backlog);
1365         cancel_work_sync(&xprt->task_cleanup);
1366         kfree(xprt->servername);
1367         /*
1368          * Tear down transport state and free the rpc_xprt
1369          */
1370         xprt->ops->destroy(xprt);
1371 }
1372
1373 /**
1374  * xprt_put - release a reference to an RPC transport.
1375  * @xprt: pointer to the transport
1376  *
1377  */
1378 void xprt_put(struct rpc_xprt *xprt)
1379 {
1380         if (atomic_dec_and_test(&xprt->count))
1381                 xprt_destroy(xprt);
1382 }
1383
1384 /**
1385  * xprt_get - return a reference to an RPC transport.
1386  * @xprt: pointer to the transport
1387  *
1388  */
1389 struct rpc_xprt *xprt_get(struct rpc_xprt *xprt)
1390 {
1391         if (atomic_inc_not_zero(&xprt->count))
1392                 return xprt;
1393         return NULL;
1394 }