git.oblomov.eu Git - 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_call().
  14  *  -   xprt_call transmits the message and installs the caller on the
  15  *      socket'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 socket. 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  *  TCP callback races fixes (C) 1998 Red Hat Software <alan@redhat.com>
  37  *  TCP send fixes (C) 1998 Red Hat Software <alan@redhat.com>
  38  *  TCP NFS related read + write fixes
  39  *   (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie>
  40  *
  41  *  Rewrite of larges part of the code in order to stabilize TCP stuff.
  42  *  Fix behaviour when socket buffer is full.
  43  *   (C) 1999 Trond Myklebust <trond.myklebust@fys.uio.no>
  44  */
  45
  46 #include <linux/types.h>
  47 #include <linux/slab.h>
  48 #include <linux/capability.h>
  49 #include <linux/sched.h>
  50 #include <linux/errno.h>
  51 #include <linux/socket.h>
  52 #include <linux/in.h>
  53 #include <linux/net.h>
  54 #include <linux/mm.h>
  55 #include <linux/udp.h>
  56 #include <linux/tcp.h>
  57 #include <linux/sunrpc/clnt.h>
  58 #include <linux/file.h>
  59 #include <linux/workqueue.h>
  60 #include <linux/random.h>
  61
  62 #include <net/sock.h>
  63 #include <net/checksum.h>
  64 #include <net/udp.h>
  65 #include <net/tcp.h>
  66
  67 /*
  68  * Local variables
  69  */
  70
  71 #ifdef RPC_DEBUG
  72 # undef  RPC_DEBUG_DATA
  73 # define RPCDBG_FACILITY        RPCDBG_XPRT
  74 #endif
  75
  76 #define XPRT_MAX_BACKOFF        (8)
  77 #define XPRT_IDLE_TIMEOUT       (5*60*HZ)
  78 #define XPRT_MAX_RESVPORT       (800)
  79
  80 /*
  81  * Local functions
  82  */
  83 static void     xprt_request_init(struct rpc_task *, struct rpc_xprt *);
  84 static inline void      do_xprt_reserve(struct rpc_task *);
  85 static void     xprt_disconnect(struct rpc_xprt *);
  86 static void     xprt_connect_status(struct rpc_task *task);
  87 static struct rpc_xprt * xprt_setup(int proto, struct sockaddr_in *ap,
  88                                                 struct rpc_timeout *to);
  89 static struct socket *xprt_create_socket(struct rpc_xprt *, int, int);
  90 static void     xprt_bind_socket(struct rpc_xprt *, struct socket *);
  91 static int      __xprt_get_cong(struct rpc_xprt *, struct rpc_task *);
  92
  93 static int      xprt_clear_backlog(struct rpc_xprt *xprt);
  94
  95 #ifdef RPC_DEBUG_DATA
  96 /*
  97  * Print the buffer contents (first 128 bytes only--just enough for
  98  * diropres return).
  99  */
 100 static void
 101 xprt_pktdump(char *msg, u32 *packet, unsigned int count)
 102 {
 103         u8      *buf = (u8 *) packet;
 104         int     j;
 105
 106         dprintk("RPC:      %s\n", msg);
 107         for (j = 0; j < count && j < 128; j += 4) {
 108                 if (!(j & 31)) {
 109                         if (j)
 110                                 dprintk("\n");
 111                         dprintk("0x%04x ", j);
 112                 }
 113                 dprintk("%02x%02x%02x%02x ",
 114                         buf[j], buf[j+1], buf[j+2], buf[j+3]);
 115         }
 116         dprintk("\n");
 117 }
 118 #else
 119 static inline void
 120 xprt_pktdump(char *msg, u32 *packet, unsigned int count)
 121 {
 122         /* NOP */
 123 }
 124 #endif
 125
 126 /*
 127  * Look up RPC transport given an INET socket
 128  */
 129 static inline struct rpc_xprt *
 130 xprt_from_sock(struct sock *sk)
 131 {
 132         return (struct rpc_xprt *) sk->sk_user_data;
 133 }
 134
 135 /*
 136  * Serialize write access to sockets, in order to prevent different
 137  * requests from interfering with each other.
 138  * Also prevents TCP socket connects from colliding with writes.
 139  */
 140 static int
 141 __xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 142 {
 143         struct rpc_rqst *req = task->tk_rqstp;
 144
 145         if (test_and_set_bit(XPRT_LOCKED, &xprt->sockstate)) {
 146                 if (task == xprt->snd_task)
 147                         return 1;
 148                 goto out_sleep;
 149         }
 150         if (xprt->nocong || __xprt_get_cong(xprt, task)) {
 151                 xprt->snd_task = task;
 152                 if (req) {
 153                         req->rq_bytes_sent = 0;
 154                         req->rq_ntrans++;
 155                 }
 156                 return 1;
 157         }
 158         smp_mb__before_clear_bit();
 159         clear_bit(XPRT_LOCKED, &xprt->sockstate);
 160         smp_mb__after_clear_bit();
 161 out_sleep:
 162         dprintk("RPC: %4d failed to lock socket %p\n", task->tk_pid, xprt);
 163         task->tk_timeout = 0;
 164         task->tk_status = -EAGAIN;
 165         if (req && req->rq_ntrans)
 166                 rpc_sleep_on(&xprt->resend, task, NULL, NULL);
 167         else
 168                 rpc_sleep_on(&xprt->sending, task, NULL, NULL);
 169         return 0;
 170 }
 171
 172 static inline int
 173 xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 174 {
 175         int retval;
 176
 177         spin_lock_bh(&xprt->sock_lock);
 178         retval = __xprt_lock_write(xprt, task);
 179         spin_unlock_bh(&xprt->sock_lock);
 180         return retval;
 181 }
 182
 183
 184 static void
 185 __xprt_lock_write_next(struct rpc_xprt *xprt)
 186 {
 187         struct rpc_task *task;
 188
 189         if (test_and_set_bit(XPRT_LOCKED, &xprt->sockstate))
 190                 return;
 191         if (!xprt->nocong && RPCXPRT_CONGESTED(xprt))
 192                 goto out_unlock;
 193         task = rpc_wake_up_next(&xprt->resend);
 194         if (!task) {
 195                 task = rpc_wake_up_next(&xprt->sending);
 196                 if (!task)
 197                         goto out_unlock;
 198         }
 199         if (xprt->nocong || __xprt_get_cong(xprt, task)) {
 200                 struct rpc_rqst *req = task->tk_rqstp;
 201                 xprt->snd_task = task;
 202                 if (req) {
 203                         req->rq_bytes_sent = 0;
 204                         req->rq_ntrans++;
 205                 }
 206                 return;
 207         }
 208 out_unlock:
 209         smp_mb__before_clear_bit();
 210         clear_bit(XPRT_LOCKED, &xprt->sockstate);
 211         smp_mb__after_clear_bit();
 212 }
 213
 214 /*
 215  * Releases the socket for use by other requests.
 216  */
 217 static void
 218 __xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 219 {
 220         if (xprt->snd_task == task) {
 221                 xprt->snd_task = NULL;
 222                 smp_mb__before_clear_bit();
 223                 clear_bit(XPRT_LOCKED, &xprt->sockstate);
 224                 smp_mb__after_clear_bit();
 225                 __xprt_lock_write_next(xprt);
 226         }
 227 }
 228
 229 static inline void
 230 xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
 231 {
 232         spin_lock_bh(&xprt->sock_lock);
 233         __xprt_release_write(xprt, task);
 234         spin_unlock_bh(&xprt->sock_lock);
 235 }
 236
 237 /*
 238  * Write data to socket.
 239  */
 240 static inline int
 241 xprt_sendmsg(struct rpc_xprt *xprt, struct rpc_rqst *req)
 242 {
 243         struct socket   *sock = xprt->sock;
 244         struct xdr_buf  *xdr = &req->rq_snd_buf;
 245         struct sockaddr *addr = NULL;
 246         int addrlen = 0;
 247         unsigned int    skip;
 248         int             result;
 249
 250         if (!sock)
 251                 return -ENOTCONN;
 252
 253         xprt_pktdump("packet data:",
 254                                 req->rq_svec->iov_base,
 255                                 req->rq_svec->iov_len);
 256
 257         /* For UDP, we need to provide an address */
 258         if (!xprt->stream) {
 259                 addr = (struct sockaddr *) &xprt->addr;
 260                 addrlen = sizeof(xprt->addr);
 261         }
 262         /* Dont repeat bytes */
 263         skip = req->rq_bytes_sent;
 264
 265         clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags);
 266         result = xdr_sendpages(sock, addr, addrlen, xdr, skip, MSG_DONTWAIT);
 267
 268         dprintk("RPC:      xprt_sendmsg(%d) = %d\n", xdr->len - skip, result);
 269
 270         if (result >= 0)
 271                 return result;
 272
 273         switch (result) {
 274         case -ECONNREFUSED:
 275                 /* When the server has died, an ICMP port unreachable message
 276                  * prompts ECONNREFUSED.
 277                  */
 278         case -EAGAIN:
 279                 break;
 280         case -ECONNRESET:
 281         case -ENOTCONN:
 282         case -EPIPE:
 283                 /* connection broken */
 284                 if (xprt->stream)
 285                         result = -ENOTCONN;
 286                 break;
 287         default:
 288                 printk(KERN_NOTICE "RPC: sendmsg returned error %d\n", -result);
 289         }
 290         return result;
 291 }
 292
 293 /*
 294  * Van Jacobson congestion avoidance. Check if the congestion window
 295  * overflowed. Put the task to sleep if this is the case.
 296  */
 297 static int
 298 __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_task *task)
 299 {
 300         struct rpc_rqst *req = task->tk_rqstp;
 301
 302         if (req->rq_cong)
 303                 return 1;
 304         dprintk("RPC: %4d xprt_cwnd_limited cong = %ld cwnd = %ld\n",
 305                         task->tk_pid, xprt->cong, xprt->cwnd);
 306         if (RPCXPRT_CONGESTED(xprt))
 307                 return 0;
 308         req->rq_cong = 1;
 309         xprt->cong += RPC_CWNDSCALE;
 310         return 1;
 311 }
 312
 313 /*
 314  * Adjust the congestion window, and wake up the next task
 315  * that has been sleeping due to congestion
 316  */
 317 static void
 318 __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
 319 {
 320         if (!req->rq_cong)
 321                 return;
 322         req->rq_cong = 0;
 323         xprt->cong -= RPC_CWNDSCALE;
 324         __xprt_lock_write_next(xprt);
 325 }
 326
 327 /*
 328  * Adjust RPC congestion window
 329  * We use a time-smoothed congestion estimator to avoid heavy oscillation.
 330  */
 331 static void
 332 xprt_adjust_cwnd(struct rpc_xprt *xprt, int result)
 333 {
 334         unsigned long   cwnd;
 335
 336         cwnd = xprt->cwnd;
 337         if (result >= 0 && cwnd <= xprt->cong) {
 338                 /* The (cwnd >> 1) term makes sure
 339                  * the result gets rounded properly. */
 340                 cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
 341                 if (cwnd > RPC_MAXCWND(xprt))
 342                         cwnd = RPC_MAXCWND(xprt);
 343                 __xprt_lock_write_next(xprt);
 344         } else if (result == -ETIMEDOUT) {
 345                 cwnd >>= 1;
 346                 if (cwnd < RPC_CWNDSCALE)
 347                         cwnd = RPC_CWNDSCALE;
 348         }
 349         dprintk("RPC:      cong %ld, cwnd was %ld, now %ld\n",
 350                         xprt->cong, xprt->cwnd, cwnd);
 351         xprt->cwnd = cwnd;
 352 }
 353
 354 /*
 355  * Reset the major timeout value
 356  */
 357 static void xprt_reset_majortimeo(struct rpc_rqst *req)
 358 {
 359         struct rpc_timeout *to = &req->rq_xprt->timeout;
 360
 361         req->rq_majortimeo = req->rq_timeout;
 362         if (to->to_exponential)
 363                 req->rq_majortimeo <<= to->to_retries;
 364         else
 365                 req->rq_majortimeo += to->to_increment * to->to_retries;
 366         if (req->rq_majortimeo > to->to_maxval || req->rq_majortimeo == 0)
 367                 req->rq_majortimeo = to->to_maxval;
 368         req->rq_majortimeo += jiffies;
 369 }
 370
 371 /*
 372  * Adjust timeout values etc for next retransmit
 373  */
 374 int xprt_adjust_timeout(struct rpc_rqst *req)
 375 {
 376         struct rpc_xprt *xprt = req->rq_xprt;
 377         struct rpc_timeout *to = &xprt->timeout;
 378         int status = 0;
 379
 380         if (time_before(jiffies, req->rq_majortimeo)) {
 381                 if (to->to_exponential)
 382                         req->rq_timeout <<= 1;
 383                 else
 384                         req->rq_timeout += to->to_increment;
 385                 if (to->to_maxval && req->rq_timeout >= to->to_maxval)
 386                         req->rq_timeout = to->to_maxval;
 387                 req->rq_retries++;
 388                 pprintk("RPC: %lu retrans\n", jiffies);
 389         } else {
 390                 req->rq_timeout = to->to_initval;
 391                 req->rq_retries = 0;
 392                 xprt_reset_majortimeo(req);
 393                 /* Reset the RTT counters == "slow start" */
 394                 spin_lock_bh(&xprt->sock_lock);
 395                 rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
 396                 spin_unlock_bh(&xprt->sock_lock);
 397                 pprintk("RPC: %lu timeout\n", jiffies);
 398                 status = -ETIMEDOUT;
 399         }
 400
 401         if (req->rq_timeout == 0) {
 402                 printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
 403                 req->rq_timeout = 5 * HZ;
 404         }
 405         return status;
 406 }
 407
 408 /*
 409  * Close down a transport socket
 410  */
 411 static void
 412 xprt_close(struct rpc_xprt *xprt)
 413 {
 414         struct socket   *sock = xprt->sock;
 415         struct sock     *sk = xprt->inet;
 416
 417         if (!sk)
 418                 return;
 419
 420         write_lock_bh(&sk->sk_callback_lock);
 421         xprt->inet = NULL;
 422         xprt->sock = NULL;
 423
 424         sk->sk_user_data    = NULL;
 425         sk->sk_data_ready   = xprt->old_data_ready;
 426         sk->sk_state_change = xprt->old_state_change;
 427         sk->sk_write_space  = xprt->old_write_space;
 428         write_unlock_bh(&sk->sk_callback_lock);
 429
 430         sk->sk_no_check  = 0;
 431
 432         sock_release(sock);
 433 }
 434
 435 static void
 436 xprt_socket_autoclose(void *args)
 437 {
 438         struct rpc_xprt *xprt = (struct rpc_xprt *)args;
 439
 440         xprt_disconnect(xprt);
 441         xprt_close(xprt);
 442         xprt_release_write(xprt, NULL);
 443 }
 444
 445 /*
 446  * Mark a transport as disconnected
 447  */
 448 static void
 449 xprt_disconnect(struct rpc_xprt *xprt)
 450 {
 451         dprintk("RPC:      disconnected transport %p\n", xprt);
 452         spin_lock_bh(&xprt->sock_lock);
 453         xprt_clear_connected(xprt);
 454         rpc_wake_up_status(&xprt->pending, -ENOTCONN);
 455         spin_unlock_bh(&xprt->sock_lock);
 456 }
 457
 458 /*
 459  * Used to allow disconnection when we've been idle
 460  */
 461 static void
 462 xprt_init_autodisconnect(unsigned long data)
 463 {
 464         struct rpc_xprt *xprt = (struct rpc_xprt *)data;
 465
 466         spin_lock(&xprt->sock_lock);
 467         if (!list_empty(&xprt->recv) || xprt->shutdown)
 468                 goto out_abort;
 469         if (test_and_set_bit(XPRT_LOCKED, &xprt->sockstate))
 470                 goto out_abort;
 471         spin_unlock(&xprt->sock_lock);
 472         /* Let keventd close the socket */
 473         if (test_bit(XPRT_CONNECTING, &xprt->sockstate) != 0)
 474                 xprt_release_write(xprt, NULL);
 475         else
 476                 schedule_work(&xprt->task_cleanup);
 477         return;
 478 out_abort:
 479         spin_unlock(&xprt->sock_lock);
 480 }
 481
 482 static void xprt_socket_connect(void *args)
 483 {
 484         struct rpc_xprt *xprt = (struct rpc_xprt *)args;
 485         struct socket *sock = xprt->sock;
 486         int status = -EIO;
 487
 488         if (xprt->shutdown || xprt->addr.sin_port == 0)
 489                 goto out;
 490
 491         /*
 492          * Start by resetting any existing state
 493          */
 494         xprt_close(xprt);
 495         sock = xprt_create_socket(xprt, xprt->prot, xprt->resvport);
 496         if (sock == NULL) {
 497                 /* couldn't create socket or bind to reserved port;
 498                  * this is likely a permanent error, so cause an abort */
 499                 goto out;
 500         }
 501         xprt_bind_socket(xprt, sock);
 502         xprt_sock_setbufsize(xprt);
 503
 504         status = 0;
 505         if (!xprt->stream)
 506                 goto out;
 507
 508         /*
 509          * Tell the socket layer to start connecting...
 510          */
 511         status = sock->ops->connect(sock, (struct sockaddr *) &xprt->addr,
 512                         sizeof(xprt->addr), O_NONBLOCK);
 513         dprintk("RPC: %p  connect status %d connected %d sock state %d\n",
 514                         xprt, -status, xprt_connected(xprt), sock->sk->sk_state);
 515         if (status < 0) {
 516                 switch (status) {
 517                         case -EINPROGRESS:
 518                         case -EALREADY:
 519                                 goto out_clear;
 520                 }
 521         }
 522 out:
 523         if (status < 0)
 524                 rpc_wake_up_status(&xprt->pending, status);
 525         else
 526                 rpc_wake_up(&xprt->pending);
 527 out_clear:
 528         smp_mb__before_clear_bit();
 529         clear_bit(XPRT_CONNECTING, &xprt->sockstate);
 530         smp_mb__after_clear_bit();
 531 }
 532
 533 /*
 534  * Attempt to connect a TCP socket.
 535  *
 536  */
 537 void xprt_connect(struct rpc_task *task)
 538 {
 539         struct rpc_xprt *xprt = task->tk_xprt;
 540
 541         dprintk("RPC: %4d xprt_connect xprt %p %s connected\n", task->tk_pid,
 542                         xprt, (xprt_connected(xprt) ? "is" : "is not"));
 543
 544         if (xprt->shutdown) {
 545                 task->tk_status = -EIO;
 546                 return;
 547         }
 548         if (!xprt->addr.sin_port) {
 549                 task->tk_status = -EIO;
 550                 return;
 551         }
 552         if (!xprt_lock_write(xprt, task))
 553                 return;
 554         if (xprt_connected(xprt))
 555                 goto out_write;
 556
 557         if (task->tk_rqstp)
 558                 task->tk_rqstp->rq_bytes_sent = 0;
 559
 560         task->tk_timeout = RPC_CONNECT_TIMEOUT;
 561         rpc_sleep_on(&xprt->pending, task, xprt_connect_status, NULL);
 562         if (!test_and_set_bit(XPRT_CONNECTING, &xprt->sockstate)) {
 563                 /* Note: if we are here due to a dropped connection
 564                  *       we delay reconnecting by RPC_REESTABLISH_TIMEOUT/HZ
 565                  *       seconds
 566                  */
 567                 if (xprt->sock != NULL)
 568                         schedule_delayed_work(&xprt->sock_connect,
 569                                         RPC_REESTABLISH_TIMEOUT);
 570                 else {
 571                         schedule_work(&xprt->sock_connect);
 572                         if (!RPC_IS_ASYNC(task))
 573                                 flush_scheduled_work();
 574                 }
 575         }
 576         return;
 577  out_write:
 578         xprt_release_write(xprt, task);
 579 }
 580
 581 /*
 582  * We arrive here when awoken from waiting on connection establishment.
 583  */
 584 static void
 585 xprt_connect_status(struct rpc_task *task)
 586 {
 587         struct rpc_xprt *xprt = task->tk_xprt;
 588
 589         if (task->tk_status >= 0) {
 590                 dprintk("RPC: %4d xprt_connect_status: connection established\n",
 591                                 task->tk_pid);
 592                 return;
 593         }
 594
 595         /* if soft mounted, just cause this RPC to fail */
 596         if (RPC_IS_SOFT(task))
 597                 task->tk_status = -EIO;
 598
 599         switch (task->tk_status) {
 600         case -ECONNREFUSED:
 601         case -ECONNRESET:
 602         case -ENOTCONN:
 603                 return;
 604         case -ETIMEDOUT:
 605                 dprintk("RPC: %4d xprt_connect_status: timed out\n",
 606                                 task->tk_pid);
 607                 break;
 608         default:
 609                 printk(KERN_ERR "RPC: error %d connecting to server %s\n",
 610                                 -task->tk_status, task->tk_client->cl_server);
 611         }
 612         xprt_release_write(xprt, task);
 613 }
 614
 615 /*
 616  * Look up the RPC request corresponding to a reply, and then lock it.
 617  */
 618 static inline struct rpc_rqst *
 619 xprt_lookup_rqst(struct rpc_xprt *xprt, u32 xid)
 620 {
 621         struct list_head *pos;
 622         struct rpc_rqst *req = NULL;
 623
 624         list_for_each(pos, &xprt->recv) {
 625                 struct rpc_rqst *entry = list_entry(pos, struct rpc_rqst, rq_list);
 626                 if (entry->rq_xid == xid) {
 627                         req = entry;
 628                         break;
 629                 }
 630         }
 631         return req;
 632 }
 633
 634 /*
 635  * Complete reply received.
 636  * The TCP code relies on us to remove the request from xprt->pending.
 637  */
 638 static void
 639 xprt_complete_rqst(struct rpc_xprt *xprt, struct rpc_rqst *req, int copied)
 640 {
 641         struct rpc_task *task = req->rq_task;
 642         struct rpc_clnt *clnt = task->tk_client;
 643
 644         /* Adjust congestion window */
 645         if (!xprt->nocong) {
 646                 unsigned timer = task->tk_msg.rpc_proc->p_timer;
 647                 xprt_adjust_cwnd(xprt, copied);
 648                 __xprt_put_cong(xprt, req);
 649                 if (timer) {
 650                         if (req->rq_ntrans == 1)
 651                                 rpc_update_rtt(clnt->cl_rtt, timer,
 652                                                 (long)jiffies - req->rq_xtime);
 653                         rpc_set_timeo(clnt->cl_rtt, timer, req->rq_ntrans - 1);
 654                 }
 655         }
 656
 657 #ifdef RPC_PROFILE
 658         /* Profile only reads for now */
 659         if (copied > 1024) {
 660                 static unsigned long    nextstat;
 661                 static unsigned long    pkt_rtt, pkt_len, pkt_cnt;
 662
 663                 pkt_cnt++;
 664                 pkt_len += req->rq_slen + copied;
 665                 pkt_rtt += jiffies - req->rq_xtime;
 666                 if (time_before(nextstat, jiffies)) {
 667                         printk("RPC: %lu %ld cwnd\n", jiffies, xprt->cwnd);
 668                         printk("RPC: %ld %ld %ld %ld stat\n",
 669                                         jiffies, pkt_cnt, pkt_len, pkt_rtt);
 670                         pkt_rtt = pkt_len = pkt_cnt = 0;
 671                         nextstat = jiffies + 5 * HZ;
 672                 }
 673         }
 674 #endif
 675
 676         dprintk("RPC: %4d has input (%d bytes)\n", task->tk_pid, copied);
 677         list_del_init(&req->rq_list);
 678         req->rq_received = req->rq_private_buf.len = copied;
 679
 680         /* ... and wake up the process. */
 681         rpc_wake_up_task(task);
 682         return;
 683 }
 684
 685 static size_t
 686 skb_read_bits(skb_reader_t *desc, void *to, size_t len)
 687 {
 688         if (len > desc->count)
 689                 len = desc->count;
 690         if (skb_copy_bits(desc->skb, desc->offset, to, len))
 691                 return 0;
 692         desc->count -= len;
 693         desc->offset += len;
 694         return len;
 695 }
 696
 697 static size_t
 698 skb_read_and_csum_bits(skb_reader_t *desc, void *to, size_t len)
 699 {
 700         unsigned int csum2, pos;
 701
 702         if (len > desc->count)
 703                 len = desc->count;
 704         pos = desc->offset;
 705         csum2 = skb_copy_and_csum_bits(desc->skb, pos, to, len, 0);
 706         desc->csum = csum_block_add(desc->csum, csum2, pos);
 707         desc->count -= len;
 708         desc->offset += len;
 709         return len;
 710 }
 711
 712 /*
 713  * We have set things up such that we perform the checksum of the UDP
 714  * packet in parallel with the copies into the RPC client iovec.  -DaveM
 715  */
 716 int
 717 csum_partial_copy_to_xdr(struct xdr_buf *xdr, struct sk_buff *skb)
 718 {
 719         skb_reader_t desc;
 720
 721         desc.skb = skb;
 722         desc.offset = sizeof(struct udphdr);
 723         desc.count = skb->len - desc.offset;
 724
 725         if (skb->ip_summed == CHECKSUM_UNNECESSARY)
 726                 goto no_checksum;
 727
 728         desc.csum = csum_partial(skb->data, desc.offset, skb->csum);
 729         if (xdr_partial_copy_from_skb(xdr, 0, &desc, skb_read_and_csum_bits) < 0)
 730                 return -1;
 731         if (desc.offset != skb->len) {
 732                 unsigned int csum2;
 733                 csum2 = skb_checksum(skb, desc.offset, skb->len - desc.offset, 0);
 734                 desc.csum = csum_block_add(desc.csum, csum2, desc.offset);
 735         }
 736         if (desc.count)
 737                 return -1;
 738         if ((unsigned short)csum_fold(desc.csum))
 739                 return -1;
 740         return 0;
 741 no_checksum:
 742         if (xdr_partial_copy_from_skb(xdr, 0, &desc, skb_read_bits) < 0)
 743                 return -1;
 744         if (desc.count)
 745                 return -1;
 746         return 0;
 747 }
 748
 749 /*
 750  * Input handler for RPC replies. Called from a bottom half and hence
 751  * atomic.
 752  */
 753 static void
 754 udp_data_ready(struct sock *sk, int len)
 755 {
 756         struct rpc_task *task;
 757         struct rpc_xprt *xprt;
 758         struct rpc_rqst *rovr;
 759         struct sk_buff  *skb;
 760         int err, repsize, copied;
 761         u32 _xid, *xp;
 762
 763         read_lock(&sk->sk_callback_lock);
 764         dprintk("RPC:      udp_data_ready...\n");
 765         if (!(xprt = xprt_from_sock(sk))) {
 766                 printk("RPC:      udp_data_ready request not found!\n");
 767                 goto out;
 768         }
 769
 770         dprintk("RPC:      udp_data_ready client %p\n", xprt);
 771
 772         if ((skb = skb_recv_datagram(sk, 0, 1, &err)) == NULL)
 773                 goto out;
 774
 775         if (xprt->shutdown)
 776                 goto dropit;
 777
 778         repsize = skb->len - sizeof(struct udphdr);
 779         if (repsize < 4) {
 780                 printk("RPC: impossible RPC reply size %d!\n", repsize);
 781                 goto dropit;
 782         }
 783
 784         /* Copy the XID from the skb... */
 785         xp = skb_header_pointer(skb, sizeof(struct udphdr),
 786                                 sizeof(_xid), &_xid);
 787         if (xp == NULL)
 788                 goto dropit;
 789
 790         /* Look up and lock the request corresponding to the given XID */
 791         spin_lock(&xprt->sock_lock);
 792         rovr = xprt_lookup_rqst(xprt, *xp);
 793         if (!rovr)
 794                 goto out_unlock;
 795         task = rovr->rq_task;
 796
 797         dprintk("RPC: %4d received reply\n", task->tk_pid);
 798
 799         if ((copied = rovr->rq_private_buf.buflen) > repsize)
 800                 copied = repsize;
 801
 802         /* Suck it into the iovec, verify checksum if not done by hw. */
 803         if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb))
 804                 goto out_unlock;
 805
 806         /* Something worked... */
 807         dst_confirm(skb->dst);
 808
 809         xprt_complete_rqst(xprt, rovr, copied);
 810
 811  out_unlock:
 812         spin_unlock(&xprt->sock_lock);
 813  dropit:
 814         skb_free_datagram(sk, skb);
 815  out:
 816         read_unlock(&sk->sk_callback_lock);
 817 }
 818
 819 /*
 820  * Copy from an skb into memory and shrink the skb.
 821  */
 822 static inline size_t
 823 tcp_copy_data(skb_reader_t *desc, void *p, size_t len)
 824 {
 825         if (len > desc->count)
 826                 len = desc->count;
 827         if (skb_copy_bits(desc->skb, desc->offset, p, len)) {
 828                 dprintk("RPC:      failed to copy %zu bytes from skb. %zu bytes remain\n",
 829                                 len, desc->count);
 830                 return 0;
 831         }
 832         desc->offset += len;
 833         desc->count -= len;
 834         dprintk("RPC:      copied %zu bytes from skb. %zu bytes remain\n",
 835                         len, desc->count);
 836         return len;
 837 }
 838
 839 /*
 840  * TCP read fragment marker
 841  */
 842 static inline void
 843 tcp_read_fraghdr(struct rpc_xprt *xprt, skb_reader_t *desc)
 844 {
 845         size_t len, used;
 846         char *p;
 847
 848         p = ((char *) &xprt->tcp_recm) + xprt->tcp_offset;
 849         len = sizeof(xprt->tcp_recm) - xprt->tcp_offset;
 850         used = tcp_copy_data(desc, p, len);
 851         xprt->tcp_offset += used;
 852         if (used != len)
 853                 return;
 854         xprt->tcp_reclen = ntohl(xprt->tcp_recm);
 855         if (xprt->tcp_reclen & 0x80000000)
 856                 xprt->tcp_flags |= XPRT_LAST_FRAG;
 857         else
 858                 xprt->tcp_flags &= ~XPRT_LAST_FRAG;
 859         xprt->tcp_reclen &= 0x7fffffff;
 860         xprt->tcp_flags &= ~XPRT_COPY_RECM;
 861         xprt->tcp_offset = 0;
 862         /* Sanity check of the record length */
 863         if (xprt->tcp_reclen < 4) {
 864                 printk(KERN_ERR "RPC: Invalid TCP record fragment length\n");
 865                 xprt_disconnect(xprt);
 866         }
 867         dprintk("RPC:      reading TCP record fragment of length %d\n",
 868                         xprt->tcp_reclen);
 869 }
 870
 871 static void
 872 tcp_check_recm(struct rpc_xprt *xprt)
 873 {
 874         dprintk("RPC:      xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u, tcp_flags = %lx\n",
 875                         xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen, xprt->tcp_flags);
 876         if (xprt->tcp_offset == xprt->tcp_reclen) {
 877                 xprt->tcp_flags |= XPRT_COPY_RECM;
 878                 xprt->tcp_offset = 0;
 879                 if (xprt->tcp_flags & XPRT_LAST_FRAG) {
 880                         xprt->tcp_flags &= ~XPRT_COPY_DATA;
 881                         xprt->tcp_flags |= XPRT_COPY_XID;
 882                         xprt->tcp_copied = 0;
 883                 }
 884         }
 885 }
 886
 887 /*
 888  * TCP read xid
 889  */
 890 static inline void
 891 tcp_read_xid(struct rpc_xprt *xprt, skb_reader_t *desc)
 892 {
 893         size_t len, used;
 894         char *p;
 895
 896         len = sizeof(xprt->tcp_xid) - xprt->tcp_offset;
 897         dprintk("RPC:      reading XID (%Zu bytes)\n", len);
 898         p = ((char *) &xprt->tcp_xid) + xprt->tcp_offset;
 899         used = tcp_copy_data(desc, p, len);
 900         xprt->tcp_offset += used;
 901         if (used != len)
 902                 return;
 903         xprt->tcp_flags &= ~XPRT_COPY_XID;
 904         xprt->tcp_flags |= XPRT_COPY_DATA;
 905         xprt->tcp_copied = 4;
 906         dprintk("RPC:      reading reply for XID %08x\n",
 907                                                 ntohl(xprt->tcp_xid));
 908         tcp_check_recm(xprt);
 909 }
 910
 911 /*
 912  * TCP read and complete request
 913  */
 914 static inline void
 915 tcp_read_request(struct rpc_xprt *xprt, skb_reader_t *desc)
 916 {
 917         struct rpc_rqst *req;
 918         struct xdr_buf *rcvbuf;
 919         size_t len;
 920         ssize_t r;
 921
 922         /* Find and lock the request corresponding to this xid */
 923         spin_lock(&xprt->sock_lock);
 924         req = xprt_lookup_rqst(xprt, xprt->tcp_xid);
 925         if (!req) {
 926                 xprt->tcp_flags &= ~XPRT_COPY_DATA;
 927                 dprintk("RPC:      XID %08x request not found!\n",
 928                                 ntohl(xprt->tcp_xid));
 929                 spin_unlock(&xprt->sock_lock);
 930                 return;
 931         }
 932
 933         rcvbuf = &req->rq_private_buf;
 934         len = desc->count;
 935         if (len > xprt->tcp_reclen - xprt->tcp_offset) {
 936                 skb_reader_t my_desc;
 937
 938                 len = xprt->tcp_reclen - xprt->tcp_offset;
 939                 memcpy(&my_desc, desc, sizeof(my_desc));
 940                 my_desc.count = len;
 941                 r = xdr_partial_copy_from_skb(rcvbuf, xprt->tcp_copied,
 942                                           &my_desc, tcp_copy_data);
 943                 desc->count -= r;
 944                 desc->offset += r;
 945         } else
 946                 r = xdr_partial_copy_from_skb(rcvbuf, xprt->tcp_copied,
 947                                           desc, tcp_copy_data);
 948
 949         if (r > 0) {
 950                 xprt->tcp_copied += r;
 951                 xprt->tcp_offset += r;
 952         }
 953         if (r != len) {
 954                 /* Error when copying to the receive buffer,
 955                  * usually because we weren't able to allocate
 956                  * additional buffer pages. All we can do now
 957                  * is turn off XPRT_COPY_DATA, so the request
 958                  * will not receive any additional updates,
 959                  * and time out.
 960                  * Any remaining data from this record will
 961                  * be discarded.
 962                  */
 963                 xprt->tcp_flags &= ~XPRT_COPY_DATA;
 964                 dprintk("RPC:      XID %08x truncated request\n",
 965                                 ntohl(xprt->tcp_xid));
 966                 dprintk("RPC:      xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u\n",
 967                                 xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen);
 968                 goto out;
 969         }
 970
 971         dprintk("RPC:      XID %08x read %Zd bytes\n",
 972                         ntohl(xprt->tcp_xid), r);
 973         dprintk("RPC:      xprt = %p, tcp_copied = %lu, tcp_offset = %u, tcp_reclen = %u\n",
 974                         xprt, xprt->tcp_copied, xprt->tcp_offset, xprt->tcp_reclen);
 975
 976         if (xprt->tcp_copied == req->rq_private_buf.buflen)
 977                 xprt->tcp_flags &= ~XPRT_COPY_DATA;
 978         else if (xprt->tcp_offset == xprt->tcp_reclen) {
 979                 if (xprt->tcp_flags & XPRT_LAST_FRAG)
 980                         xprt->tcp_flags &= ~XPRT_COPY_DATA;
 981         }
 982
 983 out:
 984         if (!(xprt->tcp_flags & XPRT_COPY_DATA)) {
 985                 dprintk("RPC: %4d received reply complete\n",
 986                                 req->rq_task->tk_pid);
 987                 xprt_complete_rqst(xprt, req, xprt->tcp_copied);
 988         }
 989         spin_unlock(&xprt->sock_lock);
 990         tcp_check_recm(xprt);
 991 }
 992
 993 /*
 994  * TCP discard extra bytes from a short read
 995  */
 996 static inline void
 997 tcp_read_discard(struct rpc_xprt *xprt, skb_reader_t *desc)
 998 {
 999         size_t len;
1000
1001         len = xprt->tcp_reclen - xprt->tcp_offset;
1002         if (len > desc->count)
1003                 len = desc->count;
1004         desc->count -= len;
1005         desc->offset += len;
1006         xprt->tcp_offset += len;
1007         dprintk("RPC:      discarded %Zu bytes\n", len);
1008         tcp_check_recm(xprt);
1009 }
1010
1011 /*
1012  * TCP record receive routine
1013  * We first have to grab the record marker, then the XID, then the data.
1014  */
1015 static int
1016 tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb,
1017                 unsigned int offset, size_t len)
1018 {
1019         struct rpc_xprt *xprt = rd_desc->arg.data;
1020         skb_reader_t desc = {
1021                 .skb    = skb,
1022                 .offset = offset,
1023                 .count  = len,
1024                 .csum   = 0
1025         };
1026
1027         dprintk("RPC:      tcp_data_recv\n");
1028         do {
1029                 /* Read in a new fragment marker if necessary */
1030                 /* Can we ever really expect to get completely empty fragments? */
1031                 if (xprt->tcp_flags & XPRT_COPY_RECM) {
1032                         tcp_read_fraghdr(xprt, &desc);
1033                         continue;
1034                 }
1035                 /* Read in the xid if necessary */
1036                 if (xprt->tcp_flags & XPRT_COPY_XID) {
1037                         tcp_read_xid(xprt, &desc);
1038                         continue;
1039                 }
1040                 /* Read in the request data */
1041                 if (xprt->tcp_flags & XPRT_COPY_DATA) {
1042                         tcp_read_request(xprt, &desc);
1043                         continue;
1044                 }
1045                 /* Skip over any trailing bytes on short reads */
1046                 tcp_read_discard(xprt, &desc);
1047         } while (desc.count);
1048         dprintk("RPC:      tcp_data_recv done\n");
1049         return len - desc.count;
1050 }
1051
1052 static void tcp_data_ready(struct sock *sk, int bytes)
1053 {
1054         struct rpc_xprt *xprt;
1055         read_descriptor_t rd_desc;
1056
1057         read_lock(&sk->sk_callback_lock);
1058         dprintk("RPC:      tcp_data_ready...\n");
1059         if (!(xprt = xprt_from_sock(sk))) {
1060                 printk("RPC:      tcp_data_ready socket info not found!\n");
1061                 goto out;
1062         }
1063         if (xprt->shutdown)
1064                 goto out;
1065
1066         /* We use rd_desc to pass struct xprt to tcp_data_recv */
1067         rd_desc.arg.data = xprt;
1068         rd_desc.count = 65536;
1069         tcp_read_sock(sk, &rd_desc, tcp_data_recv);
1070 out:
1071         read_unlock(&sk->sk_callback_lock);
1072 }
1073
1074 static void
1075 tcp_state_change(struct sock *sk)
1076 {
1077         struct rpc_xprt *xprt;
1078
1079         read_lock(&sk->sk_callback_lock);
1080         if (!(xprt = xprt_from_sock(sk)))
1081                 goto out;
1082         dprintk("RPC:      tcp_state_change client %p...\n", xprt);
1083         dprintk("RPC:      state %x conn %d dead %d zapped %d\n",
1084                                 sk->sk_state, xprt_connected(xprt),
1085                                 sock_flag(sk, SOCK_DEAD),
1086                                 sock_flag(sk, SOCK_ZAPPED));
1087
1088         switch (sk->sk_state) {
1089         case TCP_ESTABLISHED:
1090                 spin_lock_bh(&xprt->sock_lock);
1091                 if (!xprt_test_and_set_connected(xprt)) {
1092                         /* Reset TCP record info */
1093                         xprt->tcp_offset = 0;
1094                         xprt->tcp_reclen = 0;
1095                         xprt->tcp_copied = 0;
1096                         xprt->tcp_flags = XPRT_COPY_RECM | XPRT_COPY_XID;
1097                         rpc_wake_up(&xprt->pending);
1098                 }
1099                 spin_unlock_bh(&xprt->sock_lock);
1100                 break;
1101         case TCP_SYN_SENT:
1102         case TCP_SYN_RECV:
1103                 break;
1104         default:
1105                 xprt_disconnect(xprt);
1106                 break;
1107         }
1108  out:
1109         read_unlock(&sk->sk_callback_lock);
1110 }
1111
1112 /*
1113  * Called when more output buffer space is available for this socket.
1114  * We try not to wake our writers until they can make "significant"
1115  * progress, otherwise we'll waste resources thrashing sock_sendmsg
1116  * with a bunch of small requests.
1117  */
1118 static void
1119 xprt_write_space(struct sock *sk)
1120 {
1121         struct rpc_xprt *xprt;
1122         struct socket   *sock;
1123
1124         read_lock(&sk->sk_callback_lock);
1125         if (!(xprt = xprt_from_sock(sk)) || !(sock = sk->sk_socket))
1126                 goto out;
1127         if (xprt->shutdown)
1128                 goto out;
1129
1130         /* Wait until we have enough socket memory */
1131         if (xprt->stream) {
1132                 /* from net/core/stream.c:sk_stream_write_space */
1133                 if (sk_stream_wspace(sk) < sk_stream_min_wspace(sk))
1134                         goto out;
1135         } else {
1136                 /* from net/core/sock.c:sock_def_write_space */
1137                 if (!sock_writeable(sk))
1138                         goto out;
1139         }
1140
1141         if (!test_and_clear_bit(SOCK_NOSPACE, &sock->flags))
1142                 goto out;
1143
1144         spin_lock_bh(&xprt->sock_lock);
1145         if (xprt->snd_task)
1146                 rpc_wake_up_task(xprt->snd_task);
1147         spin_unlock_bh(&xprt->sock_lock);
1148 out:
1149         read_unlock(&sk->sk_callback_lock);
1150 }
1151
1152 /*
1153  * RPC receive timeout handler.
1154  */
1155 static void
1156 xprt_timer(struct rpc_task *task)
1157 {
1158         struct rpc_rqst *req = task->tk_rqstp;
1159         struct rpc_xprt *xprt = req->rq_xprt;
1160
1161         spin_lock(&xprt->sock_lock);
1162         if (req->rq_received)
1163                 goto out;
1164
1165         xprt_adjust_cwnd(req->rq_xprt, -ETIMEDOUT);
1166         __xprt_put_cong(xprt, req);
1167
1168         dprintk("RPC: %4d xprt_timer (%s request)\n",
1169                 task->tk_pid, req ? "pending" : "backlogged");
1170
1171         task->tk_status  = -ETIMEDOUT;
1172 out:
1173         task->tk_timeout = 0;
1174         rpc_wake_up_task(task);
1175         spin_unlock(&xprt->sock_lock);
1176 }
1177
1178 /*
1179  * Place the actual RPC call.
1180  * We have to copy the iovec because sendmsg fiddles with its contents.
1181  */
1182 int
1183 xprt_prepare_transmit(struct rpc_task *task)
1184 {
1185         struct rpc_rqst *req = task->tk_rqstp;
1186         struct rpc_xprt *xprt = req->rq_xprt;
1187         int err = 0;
1188
1189         dprintk("RPC: %4d xprt_prepare_transmit\n", task->tk_pid);
1190
1191         if (xprt->shutdown)
1192                 return -EIO;
1193
1194         spin_lock_bh(&xprt->sock_lock);
1195         if (req->rq_received && !req->rq_bytes_sent) {
1196                 err = req->rq_received;
1197                 goto out_unlock;
1198         }
1199         if (!__xprt_lock_write(xprt, task)) {
1200                 err = -EAGAIN;
1201                 goto out_unlock;
1202         }
1203
1204         if (!xprt_connected(xprt)) {
1205                 err = -ENOTCONN;
1206                 goto out_unlock;
1207         }
1208 out_unlock:
1209         spin_unlock_bh(&xprt->sock_lock);
1210         return err;
1211 }
1212
1213 void
1214 xprt_transmit(struct rpc_task *task)
1215 {
1216         struct rpc_clnt *clnt = task->tk_client;
1217         struct rpc_rqst *req = task->tk_rqstp;
1218         struct rpc_xprt *xprt = req->rq_xprt;
1219         int status, retry = 0;
1220
1221
1222         dprintk("RPC: %4d xprt_transmit(%u)\n", task->tk_pid, req->rq_slen);
1223
1224         /* set up everything as needed. */
1225         /* Write the record marker */
1226         if (xprt->stream) {
1227                 u32     *marker = req->rq_svec[0].iov_base;
1228
1229                 *marker = htonl(0x80000000|(req->rq_slen-sizeof(*marker)));
1230         }
1231
1232         smp_rmb();
1233         if (!req->rq_received) {
1234                 if (list_empty(&req->rq_list)) {
1235                         spin_lock_bh(&xprt->sock_lock);
1236                         /* Update the softirq receive buffer */
1237                         memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
1238                                         sizeof(req->rq_private_buf));
1239                         /* Add request to the receive list */
1240                         list_add_tail(&req->rq_list, &xprt->recv);
1241                         spin_unlock_bh(&xprt->sock_lock);
1242                         xprt_reset_majortimeo(req);
1243                         /* Turn off autodisconnect */
1244                         del_singleshot_timer_sync(&xprt->timer);
1245                 }
1246         } else if (!req->rq_bytes_sent)
1247                 return;
1248
1249         /* Continue transmitting the packet/record. We must be careful
1250          * to cope with writespace callbacks arriving _after_ we have
1251          * called xprt_sendmsg().
1252          */
1253         while (1) {
1254                 req->rq_xtime = jiffies;
1255                 status = xprt_sendmsg(xprt, req);
1256
1257                 if (status < 0)
1258                         break;
1259
1260                 if (xprt->stream) {
1261                         req->rq_bytes_sent += status;
1262
1263                         /* If we've sent the entire packet, immediately
1264                          * reset the count of bytes sent. */
1265                         if (req->rq_bytes_sent >= req->rq_slen) {
1266                                 req->rq_bytes_sent = 0;
1267                                 goto out_receive;
1268                         }
1269                 } else {
1270                         if (status >= req->rq_slen)
1271                                 goto out_receive;
1272                         status = -EAGAIN;
1273                         break;
1274                 }
1275
1276                 dprintk("RPC: %4d xmit incomplete (%d left of %d)\n",
1277                                 task->tk_pid, req->rq_slen - req->rq_bytes_sent,
1278                                 req->rq_slen);
1279
1280                 status = -EAGAIN;
1281                 if (retry++ > 50)
1282                         break;
1283         }
1284
1285         /* Note: at this point, task->tk_sleeping has not yet been set,
1286          *       hence there is no danger of the waking up task being put on
1287          *       schedq, and being picked up by a parallel run of rpciod().
1288          */
1289         task->tk_status = status;
1290
1291         switch (status) {
1292         case -EAGAIN:
1293                 if (test_bit(SOCK_ASYNC_NOSPACE, &xprt->sock->flags)) {
1294                         /* Protect against races with xprt_write_space */
1295                         spin_lock_bh(&xprt->sock_lock);
1296                         /* Don't race with disconnect */
1297                         if (!xprt_connected(xprt))
1298                                 task->tk_status = -ENOTCONN;
1299                         else if (test_bit(SOCK_NOSPACE, &xprt->sock->flags)) {
1300                                 task->tk_timeout = req->rq_timeout;
1301                                 rpc_sleep_on(&xprt->pending, task, NULL, NULL);
1302                         }
1303                         spin_unlock_bh(&xprt->sock_lock);
1304                         return;
1305                 }
1306                 /* Keep holding the socket if it is blocked */
1307                 rpc_delay(task, HZ>>4);
1308                 return;
1309         case -ECONNREFUSED:
1310                 task->tk_timeout = RPC_REESTABLISH_TIMEOUT;
1311                 rpc_sleep_on(&xprt->sending, task, NULL, NULL);
1312         case -ENOTCONN:
1313                 return;
1314         default:
1315                 if (xprt->stream)
1316                         xprt_disconnect(xprt);
1317         }
1318         xprt_release_write(xprt, task);
1319         return;
1320  out_receive:
1321         dprintk("RPC: %4d xmit complete\n", task->tk_pid);
1322         /* Set the task's receive timeout value */
1323         spin_lock_bh(&xprt->sock_lock);
1324         if (!xprt->nocong) {
1325                 int timer = task->tk_msg.rpc_proc->p_timer;
1326                 task->tk_timeout = rpc_calc_rto(clnt->cl_rtt, timer);
1327                 task->tk_timeout <<= rpc_ntimeo(clnt->cl_rtt, timer) + req->rq_retries;
1328                 if (task->tk_timeout > xprt->timeout.to_maxval || task->tk_timeout == 0)
1329                         task->tk_timeout = xprt->timeout.to_maxval;
1330         } else
1331                 task->tk_timeout = req->rq_timeout;
1332         /* Don't race with disconnect */
1333         if (!xprt_connected(xprt))
1334                 task->tk_status = -ENOTCONN;
1335         else if (!req->rq_received)
1336                 rpc_sleep_on(&xprt->pending, task, NULL, xprt_timer);
1337         __xprt_release_write(xprt, task);
1338         spin_unlock_bh(&xprt->sock_lock);
1339 }
1340
1341 /*
1342  * Reserve an RPC call slot.
1343  */
1344 static inline void
1345 do_xprt_reserve(struct rpc_task *task)
1346 {
1347         struct rpc_xprt *xprt = task->tk_xprt;
1348
1349         task->tk_status = 0;
1350         if (task->tk_rqstp)
1351                 return;
1352         if (!list_empty(&xprt->free)) {
1353                 struct rpc_rqst *req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
1354                 list_del_init(&req->rq_list);
1355                 task->tk_rqstp = req;
1356                 xprt_request_init(task, xprt);
1357                 return;
1358         }
1359         dprintk("RPC:      waiting for request slot\n");
1360         task->tk_status = -EAGAIN;
1361         task->tk_timeout = 0;
1362         rpc_sleep_on(&xprt->backlog, task, NULL, NULL);
1363 }
1364
1365 void
1366 xprt_reserve(struct rpc_task *task)
1367 {
1368         struct rpc_xprt *xprt = task->tk_xprt;
1369
1370         task->tk_status = -EIO;
1371         if (!xprt->shutdown) {
1372                 spin_lock(&xprt->xprt_lock);
1373                 do_xprt_reserve(task);
1374                 spin_unlock(&xprt->xprt_lock);
1375         }
1376 }
1377
1378 /*
1379  * Allocate a 'unique' XID
1380  */
1381 static inline u32 xprt_alloc_xid(struct rpc_xprt *xprt)
1382 {
1383         return xprt->xid++;
1384 }
1385
1386 static inline void xprt_init_xid(struct rpc_xprt *xprt)
1387 {
1388         get_random_bytes(&xprt->xid, sizeof(xprt->xid));
1389 }
1390
1391 /*
1392  * Initialize RPC request
1393  */
1394 static void
1395 xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt)
1396 {
1397         struct rpc_rqst *req = task->tk_rqstp;
1398
1399         req->rq_timeout = xprt->timeout.to_initval;
1400         req->rq_task    = task;
1401         req->rq_xprt    = xprt;
1402         req->rq_xid     = xprt_alloc_xid(xprt);
1403         dprintk("RPC: %4d reserved req %p xid %08x\n", task->tk_pid,
1404                         req, ntohl(req->rq_xid));
1405 }
1406
1407 /*
1408  * Release an RPC call slot
1409  */
1410 void
1411 xprt_release(struct rpc_task *task)
1412 {
1413         struct rpc_xprt *xprt = task->tk_xprt;
1414         struct rpc_rqst *req;
1415
1416         if (!(req = task->tk_rqstp))
1417                 return;
1418         spin_lock_bh(&xprt->sock_lock);
1419         __xprt_release_write(xprt, task);
1420         __xprt_put_cong(xprt, req);
1421         if (!list_empty(&req->rq_list))
1422                 list_del(&req->rq_list);
1423         xprt->last_used = jiffies;
1424         if (list_empty(&xprt->recv) && !xprt->shutdown)
1425                 mod_timer(&xprt->timer, xprt->last_used + XPRT_IDLE_TIMEOUT);
1426         spin_unlock_bh(&xprt->sock_lock);
1427         task->tk_rqstp = NULL;
1428         memset(req, 0, sizeof(*req));   /* mark unused */
1429
1430         dprintk("RPC: %4d release request %p\n", task->tk_pid, req);
1431
1432         spin_lock(&xprt->xprt_lock);
1433         list_add(&req->rq_list, &xprt->free);
1434         xprt_clear_backlog(xprt);
1435         spin_unlock(&xprt->xprt_lock);
1436 }
1437
1438 /*
1439  * Set default timeout parameters
1440  */
1441 static void
1442 xprt_default_timeout(struct rpc_timeout *to, int proto)
1443 {
1444         if (proto == IPPROTO_UDP)
1445                 xprt_set_timeout(to, 5,  5 * HZ);
1446         else
1447                 xprt_set_timeout(to, 5, 60 * HZ);
1448 }
1449
1450 /*
1451  * Set constant timeout
1452  */
1453 void
1454 xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long incr)
1455 {
1456         to->to_initval   =
1457         to->to_increment = incr;
1458         to->to_maxval    = incr * retr;
1459         to->to_retries   = retr;
1460         to->to_exponential = 0;
1461 }
1462
1463 unsigned int xprt_udp_slot_table_entries = RPC_DEF_SLOT_TABLE;
1464 unsigned int xprt_tcp_slot_table_entries = RPC_DEF_SLOT_TABLE;
1465
1466 /*
1467  * Initialize an RPC client
1468  */
1469 static struct rpc_xprt *
1470 xprt_setup(int proto, struct sockaddr_in *ap, struct rpc_timeout *to)
1471 {
1472         struct rpc_xprt *xprt;
1473         unsigned int entries;
1474         size_t slot_table_size;
1475         struct rpc_rqst *req;
1476
1477         dprintk("RPC:      setting up %s transport...\n",
1478                                 proto == IPPROTO_UDP? "UDP" : "TCP");
1479
1480         entries = (proto == IPPROTO_TCP)?
1481                 xprt_tcp_slot_table_entries : xprt_udp_slot_table_entries;
1482
1483         if ((xprt = kmalloc(sizeof(struct rpc_xprt), GFP_KERNEL)) == NULL)
1484                 return ERR_PTR(-ENOMEM);
1485         memset(xprt, 0, sizeof(*xprt)); /* Nnnngh! */
1486         xprt->max_reqs = entries;
1487         slot_table_size = entries * sizeof(xprt->slot[0]);
1488         xprt->slot = kmalloc(slot_table_size, GFP_KERNEL);
1489         if (xprt->slot == NULL) {
1490                 kfree(xprt);
1491                 return ERR_PTR(-ENOMEM);
1492         }
1493         memset(xprt->slot, 0, slot_table_size);
1494
1495         xprt->addr = *ap;
1496         xprt->prot = proto;
1497         xprt->stream = (proto == IPPROTO_TCP)? 1 : 0;
1498         if (xprt->stream) {
1499                 xprt->cwnd = RPC_MAXCWND(xprt);
1500                 xprt->nocong = 1;
1501                 xprt->max_payload = (1U << 31) - 1;
1502         } else {
1503                 xprt->cwnd = RPC_INITCWND;
1504                 xprt->max_payload = (1U << 16) - (MAX_HEADER << 3);
1505         }
1506         spin_lock_init(&xprt->sock_lock);
1507         spin_lock_init(&xprt->xprt_lock);
1508         init_waitqueue_head(&xprt->cong_wait);
1509
1510         INIT_LIST_HEAD(&xprt->free);
1511         INIT_LIST_HEAD(&xprt->recv);
1512         INIT_WORK(&xprt->sock_connect, xprt_socket_connect, xprt);
1513         INIT_WORK(&xprt->task_cleanup, xprt_socket_autoclose, xprt);
1514         init_timer(&xprt->timer);
1515         xprt->timer.function = xprt_init_autodisconnect;
1516         xprt->timer.data = (unsigned long) xprt;
1517         xprt->last_used = jiffies;
1518         xprt->port = XPRT_MAX_RESVPORT;
1519
1520         /* Set timeout parameters */
1521         if (to) {
1522                 xprt->timeout = *to;
1523         } else
1524                 xprt_default_timeout(&xprt->timeout, xprt->prot);
1525
1526         rpc_init_wait_queue(&xprt->pending, "xprt_pending");
1527         rpc_init_wait_queue(&xprt->sending, "xprt_sending");
1528         rpc_init_wait_queue(&xprt->resend, "xprt_resend");
1529         rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
1530
1531         /* initialize free list */
1532         for (req = &xprt->slot[entries-1]; req >= &xprt->slot[0]; req--)
1533                 list_add(&req->rq_list, &xprt->free);
1534
1535         xprt_init_xid(xprt);
1536
1537         /* Check whether we want to use a reserved port */
1538         xprt->resvport = capable(CAP_NET_BIND_SERVICE) ? 1 : 0;
1539
1540         dprintk("RPC:      created transport %p with %u slots\n", xprt,
1541                         xprt->max_reqs);
1542
1543         return xprt;
1544 }
1545
1546 /*
1547  * Bind to a reserved port
1548  */
1549 static inline int xprt_bindresvport(struct rpc_xprt *xprt, struct socket *sock)
1550 {
1551         struct sockaddr_in myaddr = {
1552                 .sin_family = AF_INET,
1553         };
1554         int             err, port;
1555
1556         /* Were we already bound to a given port? Try to reuse it */
1557         port = xprt->port;
1558         do {
1559                 myaddr.sin_port = htons(port);
1560                 err = sock->ops->bind(sock, (struct sockaddr *) &myaddr,
1561                                                 sizeof(myaddr));
1562                 if (err == 0) {
1563                         xprt->port = port;
1564                         return 0;
1565                 }
1566                 if (--port == 0)
1567                         port = XPRT_MAX_RESVPORT;
1568         } while (err == -EADDRINUSE && port != xprt->port);
1569
1570         printk("RPC: Can't bind to reserved port (%d).\n", -err);
1571         return err;
1572 }
1573
1574 static void
1575 xprt_bind_socket(struct rpc_xprt *xprt, struct socket *sock)
1576 {
1577         struct sock     *sk = sock->sk;
1578
1579         if (xprt->inet)
1580                 return;
1581
1582         write_lock_bh(&sk->sk_callback_lock);
1583         sk->sk_user_data = xprt;
1584         xprt->old_data_ready = sk->sk_data_ready;
1585         xprt->old_state_change = sk->sk_state_change;
1586         xprt->old_write_space = sk->sk_write_space;
1587         if (xprt->prot == IPPROTO_UDP) {
1588                 sk->sk_data_ready = udp_data_ready;
1589                 sk->sk_no_check = UDP_CSUM_NORCV;
1590                 xprt_set_connected(xprt);
1591         } else {
1592                 tcp_sk(sk)->nonagle = 1;        /* disable Nagle's algorithm */
1593                 sk->sk_data_ready = tcp_data_ready;
1594                 sk->sk_state_change = tcp_state_change;
1595                 xprt_clear_connected(xprt);
1596         }
1597         sk->sk_write_space = xprt_write_space;
1598
1599         /* Reset to new socket */
1600         xprt->sock = sock;
1601         xprt->inet = sk;
1602         write_unlock_bh(&sk->sk_callback_lock);
1603
1604         return;
1605 }
1606
1607 /*
1608  * Set socket buffer length
1609  */
1610 void
1611 xprt_sock_setbufsize(struct rpc_xprt *xprt)
1612 {
1613         struct sock *sk = xprt->inet;
1614
1615         if (xprt->stream)
1616                 return;
1617         if (xprt->rcvsize) {
1618                 sk->sk_userlocks |= SOCK_RCVBUF_LOCK;
1619                 sk->sk_rcvbuf = xprt->rcvsize * xprt->max_reqs *  2;
1620         }
1621         if (xprt->sndsize) {
1622                 sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
1623                 sk->sk_sndbuf = xprt->sndsize * xprt->max_reqs * 2;
1624                 sk->sk_write_space(sk);
1625         }
1626 }
1627
1628 /*
1629  * Datastream sockets are created here, but xprt_connect will create
1630  * and connect stream sockets.
1631  */
1632 static struct socket * xprt_create_socket(struct rpc_xprt *xprt, int proto, int resvport)
1633 {
1634         struct socket   *sock;
1635         int             type, err;
1636
1637         dprintk("RPC:      xprt_create_socket(%s %d)\n",
1638                            (proto == IPPROTO_UDP)? "udp" : "tcp", proto);
1639
1640         type = (proto == IPPROTO_UDP)? SOCK_DGRAM : SOCK_STREAM;
1641
1642         if ((err = sock_create_kern(PF_INET, type, proto, &sock)) < 0) {
1643                 printk("RPC: can't create socket (%d).\n", -err);
1644                 return NULL;
1645         }
1646
1647         /* If the caller has the capability, bind to a reserved port */
1648         if (resvport && xprt_bindresvport(xprt, sock) < 0) {
1649                 printk("RPC: can't bind to reserved port.\n");
1650                 goto failed;
1651         }
1652
1653         return sock;
1654
1655 failed:
1656         sock_release(sock);
1657         return NULL;
1658 }
1659
1660 /*
1661  * Create an RPC client transport given the protocol and peer address.
1662  */
1663 struct rpc_xprt *
1664 xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to)
1665 {
1666         struct rpc_xprt *xprt;
1667
1668         xprt = xprt_setup(proto, sap, to);
1669         if (IS_ERR(xprt))
1670                 dprintk("RPC:      xprt_create_proto failed\n");
1671         else
1672                 dprintk("RPC:      xprt_create_proto created xprt %p\n", xprt);
1673         return xprt;
1674 }
1675
1676 /*
1677  * Prepare for transport shutdown.
1678  */
1679 static void
1680 xprt_shutdown(struct rpc_xprt *xprt)
1681 {
1682         xprt->shutdown = 1;
1683         rpc_wake_up(&xprt->sending);
1684         rpc_wake_up(&xprt->resend);
1685         rpc_wake_up(&xprt->pending);
1686         rpc_wake_up(&xprt->backlog);
1687         wake_up(&xprt->cong_wait);
1688         del_timer_sync(&xprt->timer);
1689
1690         /* synchronously wait for connect worker to finish */
1691         cancel_delayed_work(&xprt->sock_connect);
1692         flush_scheduled_work();
1693 }
1694
1695 /*
1696  * Clear the xprt backlog queue
1697  */
1698 static int
1699 xprt_clear_backlog(struct rpc_xprt *xprt) {
1700         rpc_wake_up_next(&xprt->backlog);
1701         wake_up(&xprt->cong_wait);
1702         return 1;
1703 }
1704
1705 /*
1706  * Destroy an RPC transport, killing off all requests.
1707  */
1708 int
1709 xprt_destroy(struct rpc_xprt *xprt)
1710 {
1711         dprintk("RPC:      destroying transport %p\n", xprt);
1712         xprt_shutdown(xprt);
1713         xprt_disconnect(xprt);
1714         xprt_close(xprt);
1715         kfree(xprt->slot);
1716         kfree(xprt);
1717
1718         return 0;
1719 }