net/sunrpc/xprtsock.c

   1 /*
   2  * linux/net/sunrpc/xprtsock.c
   3  *
   4  * Client-side transport implementation for sockets.
   5  *
   6  * TCP callback races fixes (C) 1998 Red Hat Software <alan@redhat.com>
   7  * TCP send fixes (C) 1998 Red Hat Software <alan@redhat.com>
   8  * TCP NFS related read + write fixes
   9  *  (C) 1999 Dave Airlie, University of Limerick, Ireland <airlied@linux.ie>
  10  *
  11  * Rewrite of larges part of the code in order to stabilize TCP stuff.
  12  * Fix behaviour when socket buffer is full.
  13  *  (C) 1999 Trond Myklebust <trond.myklebust@fys.uio.no>
  14  *
  15  * IP socket transport implementation, (C) 2005 Chuck Lever <cel@netapp.com>
  16  *
  17  * IPv6 support contributed by Gilles Quillard, Bull Open Source, 2005.
  18  *   <gilles.quillard@bull.net>
  19  */
  20
  21 #include <linux/types.h>
  22 #include <linux/slab.h>
  23 #include <linux/capability.h>
  24 #include <linux/pagemap.h>
  25 #include <linux/errno.h>
  26 #include <linux/socket.h>
  27 #include <linux/in.h>
  28 #include <linux/net.h>
  29 #include <linux/mm.h>
  30 #include <linux/udp.h>
  31 #include <linux/tcp.h>
  32 #include <linux/sunrpc/clnt.h>
  33 #include <linux/sunrpc/sched.h>
  34 #include <linux/file.h>
  35
  36 #include <net/sock.h>
  37 #include <net/checksum.h>
  38 #include <net/udp.h>
  39 #include <net/tcp.h>
  40
  41 /*
  42  * xprtsock tunables
  43  */
  44 unsigned int xprt_udp_slot_table_entries = RPC_DEF_SLOT_TABLE;
  45 unsigned int xprt_tcp_slot_table_entries = RPC_DEF_SLOT_TABLE;
  46
  47 unsigned int xprt_min_resvport = RPC_DEF_MIN_RESVPORT;
  48 unsigned int xprt_max_resvport = RPC_DEF_MAX_RESVPORT;
  49
  50 /*
  51  * We can register our own files under /proc/sys/sunrpc by
  52  * calling register_sysctl_table() again.  The files in that
  53  * directory become the union of all files registered there.
  54  *
  55  * We simply need to make sure that we don't collide with
  56  * someone else's file names!
  57  */
  58
  59 #ifdef RPC_DEBUG
  60
  61 static unsigned int min_slot_table_size = RPC_MIN_SLOT_TABLE;
  62 static unsigned int max_slot_table_size = RPC_MAX_SLOT_TABLE;
  63 static unsigned int xprt_min_resvport_limit = RPC_MIN_RESVPORT;
  64 static unsigned int xprt_max_resvport_limit = RPC_MAX_RESVPORT;
  65
  66 static struct ctl_table_header *sunrpc_table_header;
  67
  68 /*
  69  * FIXME: changing the UDP slot table size should also resize the UDP
  70  *        socket buffers for existing UDP transports
  71  */
  72 static ctl_table xs_tunables_table[] = {
  73         {
  74                 .ctl_name       = CTL_SLOTTABLE_UDP,
  75                 .procname       = "udp_slot_table_entries",
  76                 .data           = &xprt_udp_slot_table_entries,
  77                 .maxlen         = sizeof(unsigned int),
  78                 .mode           = 0644,
  79                 .proc_handler   = &proc_dointvec_minmax,
  80                 .strategy       = &sysctl_intvec,
  81                 .extra1         = &min_slot_table_size,
  82                 .extra2         = &max_slot_table_size
  83         },
  84         {
  85                 .ctl_name       = CTL_SLOTTABLE_TCP,
  86                 .procname       = "tcp_slot_table_entries",
  87                 .data           = &xprt_tcp_slot_table_entries,
  88                 .maxlen         = sizeof(unsigned int),
  89                 .mode           = 0644,
  90                 .proc_handler   = &proc_dointvec_minmax,
  91                 .strategy       = &sysctl_intvec,
  92                 .extra1         = &min_slot_table_size,
  93                 .extra2         = &max_slot_table_size
  94         },
  95         {
  96                 .ctl_name       = CTL_MIN_RESVPORT,
  97                 .procname       = "min_resvport",
  98                 .data           = &xprt_min_resvport,
  99                 .maxlen         = sizeof(unsigned int),
 100                 .mode           = 0644,
 101                 .proc_handler   = &proc_dointvec_minmax,
 102                 .strategy       = &sysctl_intvec,
 103                 .extra1         = &xprt_min_resvport_limit,
 104                 .extra2         = &xprt_max_resvport_limit
 105         },
 106         {
 107                 .ctl_name       = CTL_MAX_RESVPORT,
 108                 .procname       = "max_resvport",
 109                 .data           = &xprt_max_resvport,
 110                 .maxlen         = sizeof(unsigned int),
 111                 .mode           = 0644,
 112                 .proc_handler   = &proc_dointvec_minmax,
 113                 .strategy       = &sysctl_intvec,
 114                 .extra1         = &xprt_min_resvport_limit,
 115                 .extra2         = &xprt_max_resvport_limit
 116         },
 117         {
 118                 .ctl_name = 0,
 119         },
 120 };
 121
 122 static ctl_table sunrpc_table[] = {
 123         {
 124                 .ctl_name       = CTL_SUNRPC,
 125                 .procname       = "sunrpc",
 126                 .mode           = 0555,
 127                 .child          = xs_tunables_table
 128         },
 129         {
 130                 .ctl_name = 0,
 131         },
 132 };
 133
 134 #endif
 135
 136 /*
 137  * How many times to try sending a request on a socket before waiting
 138  * for the socket buffer to clear.
 139  */
 140 #define XS_SENDMSG_RETRY        (10U)
 141
 142 /*
 143  * Time out for an RPC UDP socket connect.  UDP socket connects are
 144  * synchronous, but we set a timeout anyway in case of resource
 145  * exhaustion on the local host.
 146  */
 147 #define XS_UDP_CONN_TO          (5U * HZ)
 148
 149 /*
 150  * Wait duration for an RPC TCP connection to be established.  Solaris
 151  * NFS over TCP uses 60 seconds, for example, which is in line with how
 152  * long a server takes to reboot.
 153  */
 154 #define XS_TCP_CONN_TO          (60U * HZ)
 155
 156 /*
 157  * Wait duration for a reply from the RPC portmapper.
 158  */
 159 #define XS_BIND_TO              (60U * HZ)
 160
 161 /*
 162  * Delay if a UDP socket connect error occurs.  This is most likely some
 163  * kind of resource problem on the local host.
 164  */
 165 #define XS_UDP_REEST_TO         (2U * HZ)
 166
 167 /*
 168  * The reestablish timeout allows clients to delay for a bit before attempting
 169  * to reconnect to a server that just dropped our connection.
 170  *
 171  * We implement an exponential backoff when trying to reestablish a TCP
 172  * transport connection with the server.  Some servers like to drop a TCP
 173  * connection when they are overworked, so we start with a short timeout and
 174  * increase over time if the server is down or not responding.
 175  */
 176 #define XS_TCP_INIT_REEST_TO    (3U * HZ)
 177 #define XS_TCP_MAX_REEST_TO     (5U * 60 * HZ)
 178
 179 /*
 180  * TCP idle timeout; client drops the transport socket if it is idle
 181  * for this long.  Note that we also timeout UDP sockets to prevent
 182  * holding port numbers when there is no RPC traffic.
 183  */
 184 #define XS_IDLE_DISC_TO         (5U * 60 * HZ)
 185
 186 #ifdef RPC_DEBUG
 187 # undef  RPC_DEBUG_DATA
 188 # define RPCDBG_FACILITY        RPCDBG_TRANS
 189 #endif
 190
 191 #ifdef RPC_DEBUG_DATA
 192 static void xs_pktdump(char *msg, u32 *packet, unsigned int count)
 193 {
 194         u8 *buf = (u8 *) packet;
 195         int j;
 196
 197         dprintk("RPC:       %s\n", msg);
 198         for (j = 0; j < count && j < 128; j += 4) {
 199                 if (!(j & 31)) {
 200                         if (j)
 201                                 dprintk("\n");
 202                         dprintk("0x%04x ", j);
 203                 }
 204                 dprintk("%02x%02x%02x%02x ",
 205                         buf[j], buf[j+1], buf[j+2], buf[j+3]);
 206         }
 207         dprintk("\n");
 208 }
 209 #else
 210 static inline void xs_pktdump(char *msg, u32 *packet, unsigned int count)
 211 {
 212         /* NOP */
 213 }
 214 #endif
 215
 216 struct sock_xprt {
 217         struct rpc_xprt         xprt;
 218
 219         /*
 220          * Network layer
 221          */
 222         struct socket *         sock;
 223         struct sock *           inet;
 224
 225         /*
 226          * State of TCP reply receive
 227          */
 228         __be32                  tcp_fraghdr,
 229                                 tcp_xid;
 230
 231         u32                     tcp_offset,
 232                                 tcp_reclen;
 233
 234         unsigned long           tcp_copied,
 235                                 tcp_flags;
 236
 237         /*
 238          * Connection of transports
 239          */
 240         struct delayed_work     connect_worker;
 241         struct sockaddr_storage addr;
 242         unsigned short          port;
 243
 244         /*
 245          * UDP socket buffer size parameters
 246          */
 247         size_t                  rcvsize,
 248                                 sndsize;
 249
 250         /*
 251          * Saved socket callback addresses
 252          */
 253         void                    (*old_data_ready)(struct sock *, int);
 254         void                    (*old_state_change)(struct sock *);
 255         void                    (*old_write_space)(struct sock *);
 256 };
 257
 258 /*
 259  * TCP receive state flags
 260  */
 261 #define TCP_RCV_LAST_FRAG       (1UL << 0)
 262 #define TCP_RCV_COPY_FRAGHDR    (1UL << 1)
 263 #define TCP_RCV_COPY_XID        (1UL << 2)
 264 #define TCP_RCV_COPY_DATA       (1UL << 3)
 265
 266 static inline struct sockaddr *xs_addr(struct rpc_xprt *xprt)
 267 {
 268         return (struct sockaddr *) &xprt->addr;
 269 }
 270
 271 static inline struct sockaddr_in *xs_addr_in(struct rpc_xprt *xprt)
 272 {
 273         return (struct sockaddr_in *) &xprt->addr;
 274 }
 275
 276 static inline struct sockaddr_in6 *xs_addr_in6(struct rpc_xprt *xprt)
 277 {
 278         return (struct sockaddr_in6 *) &xprt->addr;
 279 }
 280
 281 static void xs_format_ipv4_peer_addresses(struct rpc_xprt *xprt)
 282 {
 283         struct sockaddr_in *addr = xs_addr_in(xprt);
 284         char *buf;
 285
 286         buf = kzalloc(20, GFP_KERNEL);
 287         if (buf) {
 288                 snprintf(buf, 20, NIPQUAD_FMT,
 289                                 NIPQUAD(addr->sin_addr.s_addr));
 290         }
 291         xprt->address_strings[RPC_DISPLAY_ADDR] = buf;
 292
 293         buf = kzalloc(8, GFP_KERNEL);
 294         if (buf) {
 295                 snprintf(buf, 8, "%u",
 296                                 ntohs(addr->sin_port));
 297         }
 298         xprt->address_strings[RPC_DISPLAY_PORT] = buf;
 299
 300         buf = kzalloc(8, GFP_KERNEL);
 301         if (buf) {
 302                 if (xprt->prot == IPPROTO_UDP)
 303                         snprintf(buf, 8, "udp");
 304                 else
 305                         snprintf(buf, 8, "tcp");
 306         }
 307         xprt->address_strings[RPC_DISPLAY_PROTO] = buf;
 308
 309         buf = kzalloc(48, GFP_KERNEL);
 310         if (buf) {
 311                 snprintf(buf, 48, "addr="NIPQUAD_FMT" port=%u proto=%s",
 312                         NIPQUAD(addr->sin_addr.s_addr),
 313                         ntohs(addr->sin_port),
 314                         xprt->prot == IPPROTO_UDP ? "udp" : "tcp");
 315         }
 316         xprt->address_strings[RPC_DISPLAY_ALL] = buf;
 317
 318         buf = kzalloc(10, GFP_KERNEL);
 319         if (buf) {
 320                 snprintf(buf, 10, "%02x%02x%02x%02x",
 321                                 NIPQUAD(addr->sin_addr.s_addr));
 322         }
 323         xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = buf;
 324
 325         buf = kzalloc(8, GFP_KERNEL);
 326         if (buf) {
 327                 snprintf(buf, 8, "%4hx",
 328                                 ntohs(addr->sin_port));
 329         }
 330         xprt->address_strings[RPC_DISPLAY_HEX_PORT] = buf;
 331
 332         buf = kzalloc(30, GFP_KERNEL);
 333         if (buf) {
 334                 snprintf(buf, 30, NIPQUAD_FMT".%u.%u",
 335                                 NIPQUAD(addr->sin_addr.s_addr),
 336                                 ntohs(addr->sin_port) >> 8,
 337                                 ntohs(addr->sin_port) & 0xff);
 338         }
 339         xprt->address_strings[RPC_DISPLAY_UNIVERSAL_ADDR] = buf;
 340
 341         xprt->address_strings[RPC_DISPLAY_NETID] =
 342                 kstrdup(xprt->prot == IPPROTO_UDP ?
 343                         RPCBIND_NETID_UDP : RPCBIND_NETID_TCP, GFP_KERNEL);
 344 }
 345
 346 static void xs_format_ipv6_peer_addresses(struct rpc_xprt *xprt)
 347 {
 348         struct sockaddr_in6 *addr = xs_addr_in6(xprt);
 349         char *buf;
 350
 351         buf = kzalloc(40, GFP_KERNEL);
 352         if (buf) {
 353                 snprintf(buf, 40, NIP6_FMT,
 354                                 NIP6(addr->sin6_addr));
 355         }
 356         xprt->address_strings[RPC_DISPLAY_ADDR] = buf;
 357
 358         buf = kzalloc(8, GFP_KERNEL);
 359         if (buf) {
 360                 snprintf(buf, 8, "%u",
 361                                 ntohs(addr->sin6_port));
 362         }
 363         xprt->address_strings[RPC_DISPLAY_PORT] = buf;
 364
 365         buf = kzalloc(8, GFP_KERNEL);
 366         if (buf) {
 367                 if (xprt->prot == IPPROTO_UDP)
 368                         snprintf(buf, 8, "udp");
 369                 else
 370                         snprintf(buf, 8, "tcp");
 371         }
 372         xprt->address_strings[RPC_DISPLAY_PROTO] = buf;
 373
 374         buf = kzalloc(64, GFP_KERNEL);
 375         if (buf) {
 376                 snprintf(buf, 64, "addr="NIP6_FMT" port=%u proto=%s",
 377                                 NIP6(addr->sin6_addr),
 378                                 ntohs(addr->sin6_port),
 379                                 xprt->prot == IPPROTO_UDP ? "udp" : "tcp");
 380         }
 381         xprt->address_strings[RPC_DISPLAY_ALL] = buf;
 382
 383         buf = kzalloc(36, GFP_KERNEL);
 384         if (buf) {
 385                 snprintf(buf, 36, NIP6_SEQFMT,
 386                                 NIP6(addr->sin6_addr));
 387         }
 388         xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = buf;
 389
 390         buf = kzalloc(8, GFP_KERNEL);
 391         if (buf) {
 392                 snprintf(buf, 8, "%4hx",
 393                                 ntohs(addr->sin6_port));
 394         }
 395         xprt->address_strings[RPC_DISPLAY_HEX_PORT] = buf;
 396
 397         buf = kzalloc(50, GFP_KERNEL);
 398         if (buf) {
 399                 snprintf(buf, 50, NIP6_FMT".%u.%u",
 400                                 NIP6(addr->sin6_addr),
 401                                 ntohs(addr->sin6_port) >> 8,
 402                                 ntohs(addr->sin6_port) & 0xff);
 403         }
 404         xprt->address_strings[RPC_DISPLAY_UNIVERSAL_ADDR] = buf;
 405
 406         xprt->address_strings[RPC_DISPLAY_NETID] =
 407                 kstrdup(xprt->prot == IPPROTO_UDP ?
 408                         RPCBIND_NETID_UDP6 : RPCBIND_NETID_TCP6, GFP_KERNEL);
 409 }
 410
 411 static void xs_free_peer_addresses(struct rpc_xprt *xprt)
 412 {
 413         int i;
 414
 415         for (i = 0; i < RPC_DISPLAY_MAX; i++)
 416                 kfree(xprt->address_strings[i]);
 417 }
 418
 419 #define XS_SENDMSG_FLAGS        (MSG_DONTWAIT | MSG_NOSIGNAL)
 420
 421 static int xs_send_kvec(struct socket *sock, struct sockaddr *addr, int addrlen, struct kvec *vec, unsigned int base, int more)
 422 {
 423         struct msghdr msg = {
 424                 .msg_name       = addr,
 425                 .msg_namelen    = addrlen,
 426                 .msg_flags      = XS_SENDMSG_FLAGS | (more ? MSG_MORE : 0),
 427         };
 428         struct kvec iov = {
 429                 .iov_base       = vec->iov_base + base,
 430                 .iov_len        = vec->iov_len - base,
 431         };
 432
 433         if (iov.iov_len != 0)
 434                 return kernel_sendmsg(sock, &msg, &iov, 1, iov.iov_len);
 435         return kernel_sendmsg(sock, &msg, NULL, 0, 0);
 436 }
 437
 438 static int xs_send_pagedata(struct socket *sock, struct xdr_buf *xdr, unsigned int base, int more)
 439 {
 440         struct page **ppage;
 441         unsigned int remainder;
 442         int err, sent = 0;
 443
 444         remainder = xdr->page_len - base;
 445         base += xdr->page_base;
 446         ppage = xdr->pages + (base >> PAGE_SHIFT);
 447         base &= ~PAGE_MASK;
 448         for(;;) {
 449                 unsigned int len = min_t(unsigned int, PAGE_SIZE - base, remainder);
 450                 int flags = XS_SENDMSG_FLAGS;
 451
 452                 remainder -= len;
 453                 if (remainder != 0 || more)
 454                         flags |= MSG_MORE;
 455                 err = sock->ops->sendpage(sock, *ppage, base, len, flags);
 456                 if (remainder == 0 || err != len)
 457                         break;
 458                 sent += err;
 459                 ppage++;
 460                 base = 0;
 461         }
 462         if (sent == 0)
 463                 return err;
 464         if (err > 0)
 465                 sent += err;
 466         return sent;
 467 }
 468
 469 /**
 470  * xs_sendpages - write pages directly to a socket
 471  * @sock: socket to send on
 472  * @addr: UDP only -- address of destination
 473  * @addrlen: UDP only -- length of destination address
 474  * @xdr: buffer containing this request
 475  * @base: starting position in the buffer
 476  *
 477  */
 478 static int xs_sendpages(struct socket *sock, struct sockaddr *addr, int addrlen, struct xdr_buf *xdr, unsigned int base)
 479 {
 480         unsigned int remainder = xdr->len - base;
 481         int err, sent = 0;
 482
 483         if (unlikely(!sock))
 484                 return -ENOTCONN;
 485
 486         clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags);
 487         if (base != 0) {
 488                 addr = NULL;
 489                 addrlen = 0;
 490         }
 491
 492         if (base < xdr->head[0].iov_len || addr != NULL) {
 493                 unsigned int len = xdr->head[0].iov_len - base;
 494                 remainder -= len;
 495                 err = xs_send_kvec(sock, addr, addrlen, &xdr->head[0], base, remainder != 0);
 496                 if (remainder == 0 || err != len)
 497                         goto out;
 498                 sent += err;
 499                 base = 0;
 500         } else
 501                 base -= xdr->head[0].iov_len;
 502
 503         if (base < xdr->page_len) {
 504                 unsigned int len = xdr->page_len - base;
 505                 remainder -= len;
 506                 err = xs_send_pagedata(sock, xdr, base, remainder != 0);
 507                 if (remainder == 0 || err != len)
 508                         goto out;
 509                 sent += err;
 510                 base = 0;
 511         } else
 512                 base -= xdr->page_len;
 513
 514         if (base >= xdr->tail[0].iov_len)
 515                 return sent;
 516         err = xs_send_kvec(sock, NULL, 0, &xdr->tail[0], base, 0);
 517 out:
 518         if (sent == 0)
 519                 return err;
 520         if (err > 0)
 521                 sent += err;
 522         return sent;
 523 }
 524
 525 /**
 526  * xs_nospace - place task on wait queue if transmit was incomplete
 527  * @task: task to put to sleep
 528  *
 529  */
 530 static void xs_nospace(struct rpc_task *task)
 531 {
 532         struct rpc_rqst *req = task->tk_rqstp;
 533         struct rpc_xprt *xprt = req->rq_xprt;
 534         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 535
 536         dprintk("RPC: %5u xmit incomplete (%u left of %u)\n",
 537                         task->tk_pid, req->rq_slen - req->rq_bytes_sent,
 538                         req->rq_slen);
 539
 540         if (test_bit(SOCK_ASYNC_NOSPACE, &transport->sock->flags)) {
 541                 /* Protect against races with write_space */
 542                 spin_lock_bh(&xprt->transport_lock);
 543
 544                 /* Don't race with disconnect */
 545                 if (!xprt_connected(xprt))
 546                         task->tk_status = -ENOTCONN;
 547                 else if (test_bit(SOCK_NOSPACE, &transport->sock->flags))
 548                         xprt_wait_for_buffer_space(task);
 549
 550                 spin_unlock_bh(&xprt->transport_lock);
 551         } else
 552                 /* Keep holding the socket if it is blocked */
 553                 rpc_delay(task, HZ>>4);
 554 }
 555
 556 /**
 557  * xs_udp_send_request - write an RPC request to a UDP socket
 558  * @task: address of RPC task that manages the state of an RPC request
 559  *
 560  * Return values:
 561  *        0:    The request has been sent
 562  *   EAGAIN:    The socket was blocked, please call again later to
 563  *              complete the request
 564  * ENOTCONN:    Caller needs to invoke connect logic then call again
 565  *    other:    Some other error occured, the request was not sent
 566  */
 567 static int xs_udp_send_request(struct rpc_task *task)
 568 {
 569         struct rpc_rqst *req = task->tk_rqstp;
 570         struct rpc_xprt *xprt = req->rq_xprt;
 571         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 572         struct xdr_buf *xdr = &req->rq_snd_buf;
 573         int status;
 574
 575         xs_pktdump("packet data:",
 576                                 req->rq_svec->iov_base,
 577                                 req->rq_svec->iov_len);
 578
 579         req->rq_xtime = jiffies;
 580         status = xs_sendpages(transport->sock,
 581                               xs_addr(xprt),
 582                               xprt->addrlen, xdr,
 583                               req->rq_bytes_sent);
 584
 585         dprintk("RPC:       xs_udp_send_request(%u) = %d\n",
 586                         xdr->len - req->rq_bytes_sent, status);
 587
 588         if (likely(status >= (int) req->rq_slen))
 589                 return 0;
 590
 591         /* Still some bytes left; set up for a retry later. */
 592         if (status > 0)
 593                 status = -EAGAIN;
 594
 595         switch (status) {
 596         case -ENETUNREACH:
 597         case -EPIPE:
 598         case -ECONNREFUSED:
 599                 /* When the server has died, an ICMP port unreachable message
 600                  * prompts ECONNREFUSED. */
 601                 break;
 602         case -EAGAIN:
 603                 xs_nospace(task);
 604                 break;
 605         default:
 606                 dprintk("RPC:       sendmsg returned unrecognized error %d\n",
 607                         -status);
 608                 break;
 609         }
 610
 611         return status;
 612 }
 613
 614 static inline void xs_encode_tcp_record_marker(struct xdr_buf *buf)
 615 {
 616         u32 reclen = buf->len - sizeof(rpc_fraghdr);
 617         rpc_fraghdr *base = buf->head[0].iov_base;
 618         *base = htonl(RPC_LAST_STREAM_FRAGMENT | reclen);
 619 }
 620
 621 /**
 622  * xs_tcp_send_request - write an RPC request to a TCP socket
 623  * @task: address of RPC task that manages the state of an RPC request
 624  *
 625  * Return values:
 626  *        0:    The request has been sent
 627  *   EAGAIN:    The socket was blocked, please call again later to
 628  *              complete the request
 629  * ENOTCONN:    Caller needs to invoke connect logic then call again
 630  *    other:    Some other error occured, the request was not sent
 631  *
 632  * XXX: In the case of soft timeouts, should we eventually give up
 633  *      if sendmsg is not able to make progress?
 634  */
 635 static int xs_tcp_send_request(struct rpc_task *task)
 636 {
 637         struct rpc_rqst *req = task->tk_rqstp;
 638         struct rpc_xprt *xprt = req->rq_xprt;
 639         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 640         struct xdr_buf *xdr = &req->rq_snd_buf;
 641         int status;
 642         unsigned int retry = 0;
 643
 644         xs_encode_tcp_record_marker(&req->rq_snd_buf);
 645
 646         xs_pktdump("packet data:",
 647                                 req->rq_svec->iov_base,
 648                                 req->rq_svec->iov_len);
 649
 650         /* Continue transmitting the packet/record. We must be careful
 651          * to cope with writespace callbacks arriving _after_ we have
 652          * called sendmsg(). */
 653         while (1) {
 654                 req->rq_xtime = jiffies;
 655                 status = xs_sendpages(transport->sock,
 656                                         NULL, 0, xdr, req->rq_bytes_sent);
 657
 658                 dprintk("RPC:       xs_tcp_send_request(%u) = %d\n",
 659                                 xdr->len - req->rq_bytes_sent, status);
 660
 661                 if (unlikely(status < 0))
 662                         break;
 663
 664                 /* If we've sent the entire packet, immediately
 665                  * reset the count of bytes sent. */
 666                 req->rq_bytes_sent += status;
 667                 task->tk_bytes_sent += status;
 668                 if (likely(req->rq_bytes_sent >= req->rq_slen)) {
 669                         req->rq_bytes_sent = 0;
 670                         return 0;
 671                 }
 672
 673                 status = -EAGAIN;
 674                 if (retry++ > XS_SENDMSG_RETRY)
 675                         break;
 676         }
 677
 678         switch (status) {
 679         case -EAGAIN:
 680                 xs_nospace(task);
 681                 break;
 682         case -ECONNREFUSED:
 683         case -ECONNRESET:
 684         case -ENOTCONN:
 685         case -EPIPE:
 686                 status = -ENOTCONN;
 687                 break;
 688         default:
 689                 dprintk("RPC:       sendmsg returned unrecognized error %d\n",
 690                         -status);
 691                 xprt_disconnect(xprt);
 692                 break;
 693         }
 694
 695         return status;
 696 }
 697
 698 /**
 699  * xs_tcp_release_xprt - clean up after a tcp transmission
 700  * @xprt: transport
 701  * @task: rpc task
 702  *
 703  * This cleans up if an error causes us to abort the transmission of a request.
 704  * In this case, the socket may need to be reset in order to avoid confusing
 705  * the server.
 706  */
 707 static void xs_tcp_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
 708 {
 709         struct rpc_rqst *req;
 710
 711         if (task != xprt->snd_task)
 712                 return;
 713         if (task == NULL)
 714                 goto out_release;
 715         req = task->tk_rqstp;
 716         if (req->rq_bytes_sent == 0)
 717                 goto out_release;
 718         if (req->rq_bytes_sent == req->rq_snd_buf.len)
 719                 goto out_release;
 720         set_bit(XPRT_CLOSE_WAIT, &task->tk_xprt->state);
 721 out_release:
 722         xprt_release_xprt(xprt, task);
 723 }
 724
 725 /**
 726  * xs_close - close a socket
 727  * @xprt: transport
 728  *
 729  * This is used when all requests are complete; ie, no DRC state remains
 730  * on the server we want to save.
 731  */
 732 static void xs_close(struct rpc_xprt *xprt)
 733 {
 734         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 735         struct socket *sock = transport->sock;
 736         struct sock *sk = transport->inet;
 737
 738         if (!sk)
 739                 goto clear_close_wait;
 740
 741         dprintk("RPC:       xs_close xprt %p\n", xprt);
 742
 743         write_lock_bh(&sk->sk_callback_lock);
 744         transport->inet = NULL;
 745         transport->sock = NULL;
 746
 747         sk->sk_user_data = NULL;
 748         sk->sk_data_ready = transport->old_data_ready;
 749         sk->sk_state_change = transport->old_state_change;
 750         sk->sk_write_space = transport->old_write_space;
 751         write_unlock_bh(&sk->sk_callback_lock);
 752
 753         sk->sk_no_check = 0;
 754
 755         sock_release(sock);
 756 clear_close_wait:
 757         smp_mb__before_clear_bit();
 758         clear_bit(XPRT_CLOSE_WAIT, &xprt->state);
 759         smp_mb__after_clear_bit();
 760 }
 761
 762 /**
 763  * xs_destroy - prepare to shutdown a transport
 764  * @xprt: doomed transport
 765  *
 766  */
 767 static void xs_destroy(struct rpc_xprt *xprt)
 768 {
 769         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 770
 771         dprintk("RPC:       xs_destroy xprt %p\n", xprt);
 772
 773         cancel_rearming_delayed_work(&transport->connect_worker);
 774
 775         xprt_disconnect(xprt);
 776         xs_close(xprt);
 777         xs_free_peer_addresses(xprt);
 778         kfree(xprt->slot);
 779         kfree(xprt);
 780 }
 781
 782 static inline struct rpc_xprt *xprt_from_sock(struct sock *sk)
 783 {
 784         return (struct rpc_xprt *) sk->sk_user_data;
 785 }
 786
 787 /**
 788  * xs_udp_data_ready - "data ready" callback for UDP sockets
 789  * @sk: socket with data to read
 790  * @len: how much data to read
 791  *
 792  */
 793 static void xs_udp_data_ready(struct sock *sk, int len)
 794 {
 795         struct rpc_task *task;
 796         struct rpc_xprt *xprt;
 797         struct rpc_rqst *rovr;
 798         struct sk_buff *skb;
 799         int err, repsize, copied;
 800         u32 _xid;
 801         __be32 *xp;
 802
 803         read_lock(&sk->sk_callback_lock);
 804         dprintk("RPC:       xs_udp_data_ready...\n");
 805         if (!(xprt = xprt_from_sock(sk)))
 806                 goto out;
 807
 808         if ((skb = skb_recv_datagram(sk, 0, 1, &err)) == NULL)
 809                 goto out;
 810
 811         if (xprt->shutdown)
 812                 goto dropit;
 813
 814         repsize = skb->len - sizeof(struct udphdr);
 815         if (repsize < 4) {
 816                 dprintk("RPC:       impossible RPC reply size %d!\n", repsize);
 817                 goto dropit;
 818         }
 819
 820         /* Copy the XID from the skb... */
 821         xp = skb_header_pointer(skb, sizeof(struct udphdr),
 822                                 sizeof(_xid), &_xid);
 823         if (xp == NULL)
 824                 goto dropit;
 825
 826         /* Look up and lock the request corresponding to the given XID */
 827         spin_lock(&xprt->transport_lock);
 828         rovr = xprt_lookup_rqst(xprt, *xp);
 829         if (!rovr)
 830                 goto out_unlock;
 831         task = rovr->rq_task;
 832
 833         if ((copied = rovr->rq_private_buf.buflen) > repsize)
 834                 copied = repsize;
 835
 836         /* Suck it into the iovec, verify checksum if not done by hw. */
 837         if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb))
 838                 goto out_unlock;
 839
 840         /* Something worked... */
 841         dst_confirm(skb->dst);
 842
 843         xprt_adjust_cwnd(task, copied);
 844         xprt_update_rtt(task);
 845         xprt_complete_rqst(task, copied);
 846
 847  out_unlock:
 848         spin_unlock(&xprt->transport_lock);
 849  dropit:
 850         skb_free_datagram(sk, skb);
 851  out:
 852         read_unlock(&sk->sk_callback_lock);
 853 }
 854
 855 static inline void xs_tcp_read_fraghdr(struct rpc_xprt *xprt, struct xdr_skb_reader *desc)
 856 {
 857         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 858         size_t len, used;
 859         char *p;
 860
 861         p = ((char *) &transport->tcp_fraghdr) + transport->tcp_offset;
 862         len = sizeof(transport->tcp_fraghdr) - transport->tcp_offset;
 863         used = xdr_skb_read_bits(desc, p, len);
 864         transport->tcp_offset += used;
 865         if (used != len)
 866                 return;
 867
 868         transport->tcp_reclen = ntohl(transport->tcp_fraghdr);
 869         if (transport->tcp_reclen & RPC_LAST_STREAM_FRAGMENT)
 870                 transport->tcp_flags |= TCP_RCV_LAST_FRAG;
 871         else
 872                 transport->tcp_flags &= ~TCP_RCV_LAST_FRAG;
 873         transport->tcp_reclen &= RPC_FRAGMENT_SIZE_MASK;
 874
 875         transport->tcp_flags &= ~TCP_RCV_COPY_FRAGHDR;
 876         transport->tcp_offset = 0;
 877
 878         /* Sanity check of the record length */
 879         if (unlikely(transport->tcp_reclen < 4)) {
 880                 dprintk("RPC:       invalid TCP record fragment length\n");
 881                 xprt_disconnect(xprt);
 882                 return;
 883         }
 884         dprintk("RPC:       reading TCP record fragment of length %d\n",
 885                         transport->tcp_reclen);
 886 }
 887
 888 static void xs_tcp_check_fraghdr(struct sock_xprt *transport)
 889 {
 890         if (transport->tcp_offset == transport->tcp_reclen) {
 891                 transport->tcp_flags |= TCP_RCV_COPY_FRAGHDR;
 892                 transport->tcp_offset = 0;
 893                 if (transport->tcp_flags & TCP_RCV_LAST_FRAG) {
 894                         transport->tcp_flags &= ~TCP_RCV_COPY_DATA;
 895                         transport->tcp_flags |= TCP_RCV_COPY_XID;
 896                         transport->tcp_copied = 0;
 897                 }
 898         }
 899 }
 900
 901 static inline void xs_tcp_read_xid(struct sock_xprt *transport, struct xdr_skb_reader *desc)
 902 {
 903         size_t len, used;
 904         char *p;
 905
 906         len = sizeof(transport->tcp_xid) - transport->tcp_offset;
 907         dprintk("RPC:       reading XID (%Zu bytes)\n", len);
 908         p = ((char *) &transport->tcp_xid) + transport->tcp_offset;
 909         used = xdr_skb_read_bits(desc, p, len);
 910         transport->tcp_offset += used;
 911         if (used != len)
 912                 return;
 913         transport->tcp_flags &= ~TCP_RCV_COPY_XID;
 914         transport->tcp_flags |= TCP_RCV_COPY_DATA;
 915         transport->tcp_copied = 4;
 916         dprintk("RPC:       reading reply for XID %08x\n",
 917                         ntohl(transport->tcp_xid));
 918         xs_tcp_check_fraghdr(transport);
 919 }
 920
 921 static inline void xs_tcp_read_request(struct rpc_xprt *xprt, struct xdr_skb_reader *desc)
 922 {
 923         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
 924         struct rpc_rqst *req;
 925         struct xdr_buf *rcvbuf;
 926         size_t len;
 927         ssize_t r;
 928
 929         /* Find and lock the request corresponding to this xid */
 930         spin_lock(&xprt->transport_lock);
 931         req = xprt_lookup_rqst(xprt, transport->tcp_xid);
 932         if (!req) {
 933                 transport->tcp_flags &= ~TCP_RCV_COPY_DATA;
 934                 dprintk("RPC:       XID %08x request not found!\n",
 935                                 ntohl(transport->tcp_xid));
 936                 spin_unlock(&xprt->transport_lock);
 937                 return;
 938         }
 939
 940         rcvbuf = &req->rq_private_buf;
 941         len = desc->count;
 942         if (len > transport->tcp_reclen - transport->tcp_offset) {
 943                 struct xdr_skb_reader my_desc;
 944
 945                 len = transport->tcp_reclen - transport->tcp_offset;
 946                 memcpy(&my_desc, desc, sizeof(my_desc));
 947                 my_desc.count = len;
 948                 r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied,
 949                                           &my_desc, xdr_skb_read_bits);
 950                 desc->count -= r;
 951                 desc->offset += r;
 952         } else
 953                 r = xdr_partial_copy_from_skb(rcvbuf, transport->tcp_copied,
 954                                           desc, xdr_skb_read_bits);
 955
 956         if (r > 0) {
 957                 transport->tcp_copied += r;
 958                 transport->tcp_offset += r;
 959         }
 960         if (r != len) {
 961                 /* Error when copying to the receive buffer,
 962                  * usually because we weren't able to allocate
 963                  * additional buffer pages. All we can do now
 964                  * is turn off TCP_RCV_COPY_DATA, so the request
 965                  * will not receive any additional updates,
 966                  * and time out.
 967                  * Any remaining data from this record will
 968                  * be discarded.
 969                  */
 970                 transport->tcp_flags &= ~TCP_RCV_COPY_DATA;
 971                 dprintk("RPC:       XID %08x truncated request\n",
 972                                 ntohl(transport->tcp_xid));
 973                 dprintk("RPC:       xprt = %p, tcp_copied = %lu, "
 974                                 "tcp_offset = %u, tcp_reclen = %u\n",
 975                                 xprt, transport->tcp_copied,
 976                                 transport->tcp_offset, transport->tcp_reclen);
 977                 goto out;
 978         }
 979
 980         dprintk("RPC:       XID %08x read %Zd bytes\n",
 981                         ntohl(transport->tcp_xid), r);
 982         dprintk("RPC:       xprt = %p, tcp_copied = %lu, tcp_offset = %u, "
 983                         "tcp_reclen = %u\n", xprt, transport->tcp_copied,
 984                         transport->tcp_offset, transport->tcp_reclen);
 985
 986         if (transport->tcp_copied == req->rq_private_buf.buflen)
 987                 transport->tcp_flags &= ~TCP_RCV_COPY_DATA;
 988         else if (transport->tcp_offset == transport->tcp_reclen) {
 989                 if (transport->tcp_flags & TCP_RCV_LAST_FRAG)
 990                         transport->tcp_flags &= ~TCP_RCV_COPY_DATA;
 991         }
 992
 993 out:
 994         if (!(transport->tcp_flags & TCP_RCV_COPY_DATA))
 995                 xprt_complete_rqst(req->rq_task, transport->tcp_copied);
 996         spin_unlock(&xprt->transport_lock);
 997         xs_tcp_check_fraghdr(transport);
 998 }
 999
1000 static inline void xs_tcp_read_discard(struct sock_xprt *transport, struct xdr_skb_reader *desc)
1001 {
1002         size_t len;
1003
1004         len = transport->tcp_reclen - transport->tcp_offset;
1005         if (len > desc->count)
1006                 len = desc->count;
1007         desc->count -= len;
1008         desc->offset += len;
1009         transport->tcp_offset += len;
1010         dprintk("RPC:       discarded %Zu bytes\n", len);
1011         xs_tcp_check_fraghdr(transport);
1012 }
1013
1014 static int xs_tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb, unsigned int offset, size_t len)
1015 {
1016         struct rpc_xprt *xprt = rd_desc->arg.data;
1017         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1018         struct xdr_skb_reader desc = {
1019                 .skb    = skb,
1020                 .offset = offset,
1021                 .count  = len,
1022         };
1023
1024         dprintk("RPC:       xs_tcp_data_recv started\n");
1025         do {
1026                 /* Read in a new fragment marker if necessary */
1027                 /* Can we ever really expect to get completely empty fragments? */
1028                 if (transport->tcp_flags & TCP_RCV_COPY_FRAGHDR) {
1029                         xs_tcp_read_fraghdr(xprt, &desc);
1030                         continue;
1031                 }
1032                 /* Read in the xid if necessary */
1033                 if (transport->tcp_flags & TCP_RCV_COPY_XID) {
1034                         xs_tcp_read_xid(transport, &desc);
1035                         continue;
1036                 }
1037                 /* Read in the request data */
1038                 if (transport->tcp_flags & TCP_RCV_COPY_DATA) {
1039                         xs_tcp_read_request(xprt, &desc);
1040                         continue;
1041                 }
1042                 /* Skip over any trailing bytes on short reads */
1043                 xs_tcp_read_discard(transport, &desc);
1044         } while (desc.count);
1045         dprintk("RPC:       xs_tcp_data_recv done\n");
1046         return len - desc.count;
1047 }
1048
1049 /**
1050  * xs_tcp_data_ready - "data ready" callback for TCP sockets
1051  * @sk: socket with data to read
1052  * @bytes: how much data to read
1053  *
1054  */
1055 static void xs_tcp_data_ready(struct sock *sk, int bytes)
1056 {
1057         struct rpc_xprt *xprt;
1058         read_descriptor_t rd_desc;
1059
1060         dprintk("RPC:       xs_tcp_data_ready...\n");
1061
1062         read_lock(&sk->sk_callback_lock);
1063         if (!(xprt = xprt_from_sock(sk)))
1064                 goto out;
1065         if (xprt->shutdown)
1066                 goto out;
1067
1068         /* We use rd_desc to pass struct xprt to xs_tcp_data_recv */
1069         rd_desc.arg.data = xprt;
1070         rd_desc.count = 65536;
1071         tcp_read_sock(sk, &rd_desc, xs_tcp_data_recv);
1072 out:
1073         read_unlock(&sk->sk_callback_lock);
1074 }
1075
1076 /**
1077  * xs_tcp_state_change - callback to handle TCP socket state changes
1078  * @sk: socket whose state has changed
1079  *
1080  */
1081 static void xs_tcp_state_change(struct sock *sk)
1082 {
1083         struct rpc_xprt *xprt;
1084
1085         read_lock(&sk->sk_callback_lock);
1086         if (!(xprt = xprt_from_sock(sk)))
1087                 goto out;
1088         dprintk("RPC:       xs_tcp_state_change client %p...\n", xprt);
1089         dprintk("RPC:       state %x conn %d dead %d zapped %d\n",
1090                         sk->sk_state, xprt_connected(xprt),
1091                         sock_flag(sk, SOCK_DEAD),
1092                         sock_flag(sk, SOCK_ZAPPED));
1093
1094         switch (sk->sk_state) {
1095         case TCP_ESTABLISHED:
1096                 spin_lock_bh(&xprt->transport_lock);
1097                 if (!xprt_test_and_set_connected(xprt)) {
1098                         struct sock_xprt *transport = container_of(xprt,
1099                                         struct sock_xprt, xprt);
1100
1101                         /* Reset TCP record info */
1102                         transport->tcp_offset = 0;
1103                         transport->tcp_reclen = 0;
1104                         transport->tcp_copied = 0;
1105                         transport->tcp_flags =
1106                                 TCP_RCV_COPY_FRAGHDR | TCP_RCV_COPY_XID;
1107
1108                         xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
1109                         xprt_wake_pending_tasks(xprt, 0);
1110                 }
1111                 spin_unlock_bh(&xprt->transport_lock);
1112                 break;
1113         case TCP_SYN_SENT:
1114         case TCP_SYN_RECV:
1115                 break;
1116         case TCP_CLOSE_WAIT:
1117                 /* Try to schedule an autoclose RPC calls */
1118                 set_bit(XPRT_CLOSE_WAIT, &xprt->state);
1119                 if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
1120                         queue_work(rpciod_workqueue, &xprt->task_cleanup);
1121         default:
1122                 xprt_disconnect(xprt);
1123         }
1124  out:
1125         read_unlock(&sk->sk_callback_lock);
1126 }
1127
1128 /**
1129  * xs_udp_write_space - callback invoked when socket buffer space
1130  *                             becomes available
1131  * @sk: socket whose state has changed
1132  *
1133  * Called when more output buffer space is available for this socket.
1134  * We try not to wake our writers until they can make "significant"
1135  * progress, otherwise we'll waste resources thrashing kernel_sendmsg
1136  * with a bunch of small requests.
1137  */
1138 static void xs_udp_write_space(struct sock *sk)
1139 {
1140         read_lock(&sk->sk_callback_lock);
1141
1142         /* from net/core/sock.c:sock_def_write_space */
1143         if (sock_writeable(sk)) {
1144                 struct socket *sock;
1145                 struct rpc_xprt *xprt;
1146
1147                 if (unlikely(!(sock = sk->sk_socket)))
1148                         goto out;
1149                 if (unlikely(!(xprt = xprt_from_sock(sk))))
1150                         goto out;
1151                 if (unlikely(!test_and_clear_bit(SOCK_NOSPACE, &sock->flags)))
1152                         goto out;
1153
1154                 xprt_write_space(xprt);
1155         }
1156
1157  out:
1158         read_unlock(&sk->sk_callback_lock);
1159 }
1160
1161 /**
1162  * xs_tcp_write_space - callback invoked when socket buffer space
1163  *                             becomes available
1164  * @sk: socket whose state has changed
1165  *
1166  * Called when more output buffer space is available for this socket.
1167  * We try not to wake our writers until they can make "significant"
1168  * progress, otherwise we'll waste resources thrashing kernel_sendmsg
1169  * with a bunch of small requests.
1170  */
1171 static void xs_tcp_write_space(struct sock *sk)
1172 {
1173         read_lock(&sk->sk_callback_lock);
1174
1175         /* from net/core/stream.c:sk_stream_write_space */
1176         if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) {
1177                 struct socket *sock;
1178                 struct rpc_xprt *xprt;
1179
1180                 if (unlikely(!(sock = sk->sk_socket)))
1181                         goto out;
1182                 if (unlikely(!(xprt = xprt_from_sock(sk))))
1183                         goto out;
1184                 if (unlikely(!test_and_clear_bit(SOCK_NOSPACE, &sock->flags)))
1185                         goto out;
1186
1187                 xprt_write_space(xprt);
1188         }
1189
1190  out:
1191         read_unlock(&sk->sk_callback_lock);
1192 }
1193
1194 static void xs_udp_do_set_buffer_size(struct rpc_xprt *xprt)
1195 {
1196         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1197         struct sock *sk = transport->inet;
1198
1199         if (transport->rcvsize) {
1200                 sk->sk_userlocks |= SOCK_RCVBUF_LOCK;
1201                 sk->sk_rcvbuf = transport->rcvsize * xprt->max_reqs * 2;
1202         }
1203         if (transport->sndsize) {
1204                 sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
1205                 sk->sk_sndbuf = transport->sndsize * xprt->max_reqs * 2;
1206                 sk->sk_write_space(sk);
1207         }
1208 }
1209
1210 /**
1211  * xs_udp_set_buffer_size - set send and receive limits
1212  * @xprt: generic transport
1213  * @sndsize: requested size of send buffer, in bytes
1214  * @rcvsize: requested size of receive buffer, in bytes
1215  *
1216  * Set socket send and receive buffer size limits.
1217  */
1218 static void xs_udp_set_buffer_size(struct rpc_xprt *xprt, size_t sndsize, size_t rcvsize)
1219 {
1220         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1221
1222         transport->sndsize = 0;
1223         if (sndsize)
1224                 transport->sndsize = sndsize + 1024;
1225         transport->rcvsize = 0;
1226         if (rcvsize)
1227                 transport->rcvsize = rcvsize + 1024;
1228
1229         xs_udp_do_set_buffer_size(xprt);
1230 }
1231
1232 /**
1233  * xs_udp_timer - called when a retransmit timeout occurs on a UDP transport
1234  * @task: task that timed out
1235  *
1236  * Adjust the congestion window after a retransmit timeout has occurred.
1237  */
1238 static void xs_udp_timer(struct rpc_task *task)
1239 {
1240         xprt_adjust_cwnd(task, -ETIMEDOUT);
1241 }
1242
1243 static unsigned short xs_get_random_port(void)
1244 {
1245         unsigned short range = xprt_max_resvport - xprt_min_resvport;
1246         unsigned short rand = (unsigned short) net_random() % range;
1247         return rand + xprt_min_resvport;
1248 }
1249
1250 /**
1251  * xs_set_port - reset the port number in the remote endpoint address
1252  * @xprt: generic transport
1253  * @port: new port number
1254  *
1255  */
1256 static void xs_set_port(struct rpc_xprt *xprt, unsigned short port)
1257 {
1258         struct sockaddr *addr = xs_addr(xprt);
1259
1260         dprintk("RPC:       setting port for xprt %p to %u\n", xprt, port);
1261
1262         switch (addr->sa_family) {
1263         case AF_INET:
1264                 ((struct sockaddr_in *)addr)->sin_port = htons(port);
1265                 break;
1266         case AF_INET6:
1267                 ((struct sockaddr_in6 *)addr)->sin6_port = htons(port);
1268                 break;
1269         default:
1270                 BUG();
1271         }
1272 }
1273
1274 static int xs_bind4(struct sock_xprt *transport, struct socket *sock)
1275 {
1276         struct sockaddr_in myaddr = {
1277                 .sin_family = AF_INET,
1278         };
1279         struct sockaddr_in *sa;
1280         int err;
1281         unsigned short port = transport->port;
1282
1283         if (!transport->xprt.resvport)
1284                 port = 0;
1285         sa = (struct sockaddr_in *)&transport->addr;
1286         myaddr.sin_addr = sa->sin_addr;
1287         do {
1288                 myaddr.sin_port = htons(port);
1289                 err = kernel_bind(sock, (struct sockaddr *) &myaddr,
1290                                                 sizeof(myaddr));
1291                 if (!transport->xprt.resvport)
1292                         break;
1293                 if (err == 0) {
1294                         transport->port = port;
1295                         break;
1296                 }
1297                 if (port <= xprt_min_resvport)
1298                         port = xprt_max_resvport;
1299                 else
1300                         port--;
1301         } while (err == -EADDRINUSE && port != transport->port);
1302         dprintk("RPC:       %s "NIPQUAD_FMT":%u: %s (%d)\n",
1303                         __FUNCTION__, NIPQUAD(myaddr.sin_addr),
1304                         port, err ? "failed" : "ok", err);
1305         return err;
1306 }
1307
1308 static int xs_bind6(struct sock_xprt *transport, struct socket *sock)
1309 {
1310         struct sockaddr_in6 myaddr = {
1311                 .sin6_family = AF_INET6,
1312         };
1313         struct sockaddr_in6 *sa;
1314         int err;
1315         unsigned short port = transport->port;
1316
1317         if (!transport->xprt.resvport)
1318                 port = 0;
1319         sa = (struct sockaddr_in6 *)&transport->addr;
1320         myaddr.sin6_addr = sa->sin6_addr;
1321         do {
1322                 myaddr.sin6_port = htons(port);
1323                 err = kernel_bind(sock, (struct sockaddr *) &myaddr,
1324                                                 sizeof(myaddr));
1325                 if (!transport->xprt.resvport)
1326                         break;
1327                 if (err == 0) {
1328                         transport->port = port;
1329                         break;
1330                 }
1331                 if (port <= xprt_min_resvport)
1332                         port = xprt_max_resvport;
1333                 else
1334                         port--;
1335         } while (err == -EADDRINUSE && port != transport->port);
1336         dprintk("RPC:       xs_bind6 "NIP6_FMT":%u: %s (%d)\n",
1337                 NIP6(myaddr.sin6_addr), port, err ? "failed" : "ok", err);
1338         return err;
1339 }
1340
1341 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1342 static struct lock_class_key xs_key[2];
1343 static struct lock_class_key xs_slock_key[2];
1344
1345 static inline void xs_reclassify_socket4(struct socket *sock)
1346 {
1347         struct sock *sk = sock->sk;
1348
1349         BUG_ON(sk->sk_lock.owner != NULL);
1350         sock_lock_init_class_and_name(sk, "slock-AF_INET-RPC",
1351                 &xs_slock_key[0], "sk_lock-AF_INET-RPC", &xs_key[0]);
1352 }
1353
1354 static inline void xs_reclassify_socket6(struct socket *sock)
1355 {
1356         struct sock *sk = sock->sk;
1357
1358         BUG_ON(sk->sk_lock.owner != NULL);
1359         sock_lock_init_class_and_name(sk, "slock-AF_INET6-RPC",
1360                 &xs_slock_key[1], "sk_lock-AF_INET6-RPC", &xs_key[1]);
1361 }
1362 #else
1363 static inline void xs_reclassify_socket4(struct socket *sock)
1364 {
1365 }
1366
1367 static inline void xs_reclassify_socket6(struct socket *sock)
1368 {
1369 }
1370 #endif
1371
1372 static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
1373 {
1374         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1375
1376         if (!transport->inet) {
1377                 struct sock *sk = sock->sk;
1378
1379                 write_lock_bh(&sk->sk_callback_lock);
1380
1381                 sk->sk_user_data = xprt;
1382                 transport->old_data_ready = sk->sk_data_ready;
1383                 transport->old_state_change = sk->sk_state_change;
1384                 transport->old_write_space = sk->sk_write_space;
1385                 sk->sk_data_ready = xs_udp_data_ready;
1386                 sk->sk_write_space = xs_udp_write_space;
1387                 sk->sk_no_check = UDP_CSUM_NORCV;
1388                 sk->sk_allocation = GFP_ATOMIC;
1389
1390                 xprt_set_connected(xprt);
1391
1392                 /* Reset to new socket */
1393                 transport->sock = sock;
1394                 transport->inet = sk;
1395
1396                 write_unlock_bh(&sk->sk_callback_lock);
1397         }
1398         xs_udp_do_set_buffer_size(xprt);
1399 }
1400
1401 /**
1402  * xs_udp_connect_worker4 - set up a UDP socket
1403  * @work: RPC transport to connect
1404  *
1405  * Invoked by a work queue tasklet.
1406  */
1407 static void xs_udp_connect_worker4(struct work_struct *work)
1408 {
1409         struct sock_xprt *transport =
1410                 container_of(work, struct sock_xprt, connect_worker.work);
1411         struct rpc_xprt *xprt = &transport->xprt;
1412         struct socket *sock = transport->sock;
1413         int err, status = -EIO;
1414
1415         if (xprt->shutdown || !xprt_bound(xprt))
1416                 goto out;
1417
1418         /* Start by resetting any existing state */
1419         xs_close(xprt);
1420
1421         if ((err = sock_create_kern(PF_INET, SOCK_DGRAM, IPPROTO_UDP, &sock)) < 0) {
1422                 dprintk("RPC:       can't create UDP transport socket (%d).\n", -err);
1423                 goto out;
1424         }
1425         xs_reclassify_socket4(sock);
1426
1427         if (xs_bind4(transport, sock)) {
1428                 sock_release(sock);
1429                 goto out;
1430         }
1431
1432         dprintk("RPC:       worker connecting xprt %p to address: %s\n",
1433                         xprt, xprt->address_strings[RPC_DISPLAY_ALL]);
1434
1435         xs_udp_finish_connecting(xprt, sock);
1436         status = 0;
1437 out:
1438         xprt_wake_pending_tasks(xprt, status);
1439         xprt_clear_connecting(xprt);
1440 }
1441
1442 /**
1443  * xs_udp_connect_worker6 - set up a UDP socket
1444  * @work: RPC transport to connect
1445  *
1446  * Invoked by a work queue tasklet.
1447  */
1448 static void xs_udp_connect_worker6(struct work_struct *work)
1449 {
1450         struct sock_xprt *transport =
1451                 container_of(work, struct sock_xprt, connect_worker.work);
1452         struct rpc_xprt *xprt = &transport->xprt;
1453         struct socket *sock = transport->sock;
1454         int err, status = -EIO;
1455
1456         if (xprt->shutdown || !xprt_bound(xprt))
1457                 goto out;
1458
1459         /* Start by resetting any existing state */
1460         xs_close(xprt);
1461
1462         if ((err = sock_create_kern(PF_INET6, SOCK_DGRAM, IPPROTO_UDP, &sock)) < 0) {
1463                 dprintk("RPC:       can't create UDP transport socket (%d).\n", -err);
1464                 goto out;
1465         }
1466         xs_reclassify_socket6(sock);
1467
1468         if (xs_bind6(transport, sock) < 0) {
1469                 sock_release(sock);
1470                 goto out;
1471         }
1472
1473         dprintk("RPC:       worker connecting xprt %p to address: %s\n",
1474                         xprt, xprt->address_strings[RPC_DISPLAY_ALL]);
1475
1476         xs_udp_finish_connecting(xprt, sock);
1477         status = 0;
1478 out:
1479         xprt_wake_pending_tasks(xprt, status);
1480         xprt_clear_connecting(xprt);
1481 }
1482
1483 /*
1484  * We need to preserve the port number so the reply cache on the server can
1485  * find our cached RPC replies when we get around to reconnecting.
1486  */
1487 static void xs_tcp_reuse_connection(struct rpc_xprt *xprt)
1488 {
1489         int result;
1490         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1491         struct sockaddr any;
1492
1493         dprintk("RPC:       disconnecting xprt %p to reuse port\n", xprt);
1494
1495         /*
1496          * Disconnect the transport socket by doing a connect operation
1497          * with AF_UNSPEC.  This should return immediately...
1498          */
1499         memset(&any, 0, sizeof(any));
1500         any.sa_family = AF_UNSPEC;
1501         result = kernel_connect(transport->sock, &any, sizeof(any), 0);
1502         if (result)
1503                 dprintk("RPC:       AF_UNSPEC connect return code %d\n",
1504                                 result);
1505 }
1506
1507 static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
1508 {
1509         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1510
1511         if (!transport->inet) {
1512                 struct sock *sk = sock->sk;
1513
1514                 write_lock_bh(&sk->sk_callback_lock);
1515
1516                 sk->sk_user_data = xprt;
1517                 transport->old_data_ready = sk->sk_data_ready;
1518                 transport->old_state_change = sk->sk_state_change;
1519                 transport->old_write_space = sk->sk_write_space;
1520                 sk->sk_data_ready = xs_tcp_data_ready;
1521                 sk->sk_state_change = xs_tcp_state_change;
1522                 sk->sk_write_space = xs_tcp_write_space;
1523                 sk->sk_allocation = GFP_ATOMIC;
1524
1525                 /* socket options */
1526                 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
1527                 sock_reset_flag(sk, SOCK_LINGER);
1528                 tcp_sk(sk)->linger2 = 0;
1529                 tcp_sk(sk)->nonagle |= TCP_NAGLE_OFF;
1530
1531                 xprt_clear_connected(xprt);
1532
1533                 /* Reset to new socket */
1534                 transport->sock = sock;
1535                 transport->inet = sk;
1536
1537                 write_unlock_bh(&sk->sk_callback_lock);
1538         }
1539
1540         /* Tell the socket layer to start connecting... */
1541         xprt->stat.connect_count++;
1542         xprt->stat.connect_start = jiffies;
1543         return kernel_connect(sock, xs_addr(xprt), xprt->addrlen, O_NONBLOCK);
1544 }
1545
1546 /**
1547  * xs_tcp_connect_worker4 - connect a TCP socket to a remote endpoint
1548  * @work: RPC transport to connect
1549  *
1550  * Invoked by a work queue tasklet.
1551  */
1552 static void xs_tcp_connect_worker4(struct work_struct *work)
1553 {
1554         struct sock_xprt *transport =
1555                 container_of(work, struct sock_xprt, connect_worker.work);
1556         struct rpc_xprt *xprt = &transport->xprt;
1557         struct socket *sock = transport->sock;
1558         int err, status = -EIO;
1559
1560         if (xprt->shutdown || !xprt_bound(xprt))
1561                 goto out;
1562
1563         if (!sock) {
1564                 /* start from scratch */
1565                 if ((err = sock_create_kern(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock)) < 0) {
1566                         dprintk("RPC:       can't create TCP transport socket (%d).\n", -err);
1567                         goto out;
1568                 }
1569                 xs_reclassify_socket4(sock);
1570
1571                 if (xs_bind4(transport, sock) < 0) {
1572                         sock_release(sock);
1573                         goto out;
1574                 }
1575         } else
1576                 /* "close" the socket, preserving the local port */
1577                 xs_tcp_reuse_connection(xprt);
1578
1579         dprintk("RPC:       worker connecting xprt %p to address: %s\n",
1580                         xprt, xprt->address_strings[RPC_DISPLAY_ALL]);
1581
1582         status = xs_tcp_finish_connecting(xprt, sock);
1583         dprintk("RPC:       %p connect status %d connected %d sock state %d\n",
1584                         xprt, -status, xprt_connected(xprt),
1585                         sock->sk->sk_state);
1586         if (status < 0) {
1587                 switch (status) {
1588                         case -EINPROGRESS:
1589                         case -EALREADY:
1590                                 goto out_clear;
1591                         case -ECONNREFUSED:
1592                         case -ECONNRESET:
1593                                 /* retry with existing socket, after a delay */
1594                                 break;
1595                         default:
1596                                 /* get rid of existing socket, and retry */
1597                                 xs_close(xprt);
1598                                 break;
1599                 }
1600         }
1601 out:
1602         xprt_wake_pending_tasks(xprt, status);
1603 out_clear:
1604         xprt_clear_connecting(xprt);
1605 }
1606
1607 /**
1608  * xs_tcp_connect_worker6 - connect a TCP socket to a remote endpoint
1609  * @work: RPC transport to connect
1610  *
1611  * Invoked by a work queue tasklet.
1612  */
1613 static void xs_tcp_connect_worker6(struct work_struct *work)
1614 {
1615         struct sock_xprt *transport =
1616                 container_of(work, struct sock_xprt, connect_worker.work);
1617         struct rpc_xprt *xprt = &transport->xprt;
1618         struct socket *sock = transport->sock;
1619         int err, status = -EIO;
1620
1621         if (xprt->shutdown || !xprt_bound(xprt))
1622                 goto out;
1623
1624         if (!sock) {
1625                 /* start from scratch */
1626                 if ((err = sock_create_kern(PF_INET6, SOCK_STREAM, IPPROTO_TCP, &sock)) < 0) {
1627                         dprintk("RPC:       can't create TCP transport socket (%d).\n", -err);
1628                         goto out;
1629                 }
1630                 xs_reclassify_socket6(sock);
1631
1632                 if (xs_bind6(transport, sock) < 0) {
1633                         sock_release(sock);
1634                         goto out;
1635                 }
1636         } else
1637                 /* "close" the socket, preserving the local port */
1638                 xs_tcp_reuse_connection(xprt);
1639
1640         dprintk("RPC:       worker connecting xprt %p to address: %s\n",
1641                         xprt, xprt->address_strings[RPC_DISPLAY_ALL]);
1642
1643         status = xs_tcp_finish_connecting(xprt, sock);
1644         dprintk("RPC:       %p connect status %d connected %d sock state %d\n",
1645                         xprt, -status, xprt_connected(xprt), sock->sk->sk_state);
1646         if (status < 0) {
1647                 switch (status) {
1648                         case -EINPROGRESS:
1649                         case -EALREADY:
1650                                 goto out_clear;
1651                         case -ECONNREFUSED:
1652                         case -ECONNRESET:
1653                                 /* retry with existing socket, after a delay */
1654                                 break;
1655                         default:
1656                                 /* get rid of existing socket, and retry */
1657                                 xs_close(xprt);
1658                                 break;
1659                 }
1660         }
1661 out:
1662         xprt_wake_pending_tasks(xprt, status);
1663 out_clear:
1664         xprt_clear_connecting(xprt);
1665 }
1666
1667 /**
1668  * xs_connect - connect a socket to a remote endpoint
1669  * @task: address of RPC task that manages state of connect request
1670  *
1671  * TCP: If the remote end dropped the connection, delay reconnecting.
1672  *
1673  * UDP socket connects are synchronous, but we use a work queue anyway
1674  * to guarantee that even unprivileged user processes can set up a
1675  * socket on a privileged port.
1676  *
1677  * If a UDP socket connect fails, the delay behavior here prevents
1678  * retry floods (hard mounts).
1679  */
1680 static void xs_connect(struct rpc_task *task)
1681 {
1682         struct rpc_xprt *xprt = task->tk_xprt;
1683         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1684
1685         if (xprt_test_and_set_connecting(xprt))
1686                 return;
1687
1688         if (transport->sock != NULL) {
1689                 dprintk("RPC:       xs_connect delayed xprt %p for %lu "
1690                                 "seconds\n",
1691                                 xprt, xprt->reestablish_timeout / HZ);
1692                 queue_delayed_work(rpciod_workqueue,
1693                                    &transport->connect_worker,
1694                                    xprt->reestablish_timeout);
1695                 xprt->reestablish_timeout <<= 1;
1696                 if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO)
1697                         xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO;
1698         } else {
1699                 dprintk("RPC:       xs_connect scheduled xprt %p\n", xprt);
1700                 queue_delayed_work(rpciod_workqueue,
1701                                    &transport->connect_worker, 0);
1702         }
1703 }
1704
1705 /**
1706  * xs_udp_print_stats - display UDP socket-specifc stats
1707  * @xprt: rpc_xprt struct containing statistics
1708  * @seq: output file
1709  *
1710  */
1711 static void xs_udp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
1712 {
1713         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1714
1715         seq_printf(seq, "\txprt:\tudp %u %lu %lu %lu %lu %Lu %Lu\n",
1716                         transport->port,
1717                         xprt->stat.bind_count,
1718                         xprt->stat.sends,
1719                         xprt->stat.recvs,
1720                         xprt->stat.bad_xids,
1721                         xprt->stat.req_u,
1722                         xprt->stat.bklog_u);
1723 }
1724
1725 /**
1726  * xs_tcp_print_stats - display TCP socket-specifc stats
1727  * @xprt: rpc_xprt struct containing statistics
1728  * @seq: output file
1729  *
1730  */
1731 static void xs_tcp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
1732 {
1733         struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
1734         long idle_time = 0;
1735
1736         if (xprt_connected(xprt))
1737                 idle_time = (long)(jiffies - xprt->last_used) / HZ;
1738
1739         seq_printf(seq, "\txprt:\ttcp %u %lu %lu %lu %ld %lu %lu %lu %Lu %Lu\n",
1740                         transport->port,
1741                         xprt->stat.bind_count,
1742                         xprt->stat.connect_count,
1743                         xprt->stat.connect_time,
1744                         idle_time,
1745                         xprt->stat.sends,
1746                         xprt->stat.recvs,
1747                         xprt->stat.bad_xids,
1748                         xprt->stat.req_u,
1749                         xprt->stat.bklog_u);
1750 }
1751
1752 static struct rpc_xprt_ops xs_udp_ops = {
1753         .set_buffer_size        = xs_udp_set_buffer_size,
1754         .reserve_xprt           = xprt_reserve_xprt_cong,
1755         .release_xprt           = xprt_release_xprt_cong,
1756         .rpcbind                = rpcb_getport_async,
1757         .set_port               = xs_set_port,
1758         .connect                = xs_connect,
1759         .buf_alloc              = rpc_malloc,
1760         .buf_free               = rpc_free,
1761         .send_request           = xs_udp_send_request,
1762         .set_retrans_timeout    = xprt_set_retrans_timeout_rtt,
1763         .timer                  = xs_udp_timer,
1764         .release_request        = xprt_release_rqst_cong,
1765         .close                  = xs_close,
1766         .destroy                = xs_destroy,
1767         .print_stats            = xs_udp_print_stats,
1768 };
1769
1770 static struct rpc_xprt_ops xs_tcp_ops = {
1771         .reserve_xprt           = xprt_reserve_xprt,
1772         .release_xprt           = xs_tcp_release_xprt,
1773         .rpcbind                = rpcb_getport_async,
1774         .set_port               = xs_set_port,
1775         .connect                = xs_connect,
1776         .buf_alloc              = rpc_malloc,
1777         .buf_free               = rpc_free,
1778         .send_request           = xs_tcp_send_request,
1779         .set_retrans_timeout    = xprt_set_retrans_timeout_def,
1780         .close                  = xs_close,
1781         .destroy                = xs_destroy,
1782         .print_stats            = xs_tcp_print_stats,
1783 };
1784
1785 static struct rpc_xprt *xs_setup_xprt(struct rpc_xprtsock_create *args, unsigned int slot_table_size)
1786 {
1787         struct rpc_xprt *xprt;
1788         struct sock_xprt *new;
1789
1790         if (args->addrlen > sizeof(xprt->addr)) {
1791                 dprintk("RPC:       xs_setup_xprt: address too large\n");
1792                 return ERR_PTR(-EBADF);
1793         }
1794
1795         new = kzalloc(sizeof(*new), GFP_KERNEL);
1796         if (new == NULL) {
1797                 dprintk("RPC:       xs_setup_xprt: couldn't allocate "
1798                                 "rpc_xprt\n");
1799                 return ERR_PTR(-ENOMEM);
1800         }
1801         xprt = &new->xprt;
1802
1803         xprt->max_reqs = slot_table_size;
1804         xprt->slot = kcalloc(xprt->max_reqs, sizeof(struct rpc_rqst), GFP_KERNEL);
1805         if (xprt->slot == NULL) {
1806                 kfree(xprt);
1807                 dprintk("RPC:       xs_setup_xprt: couldn't allocate slot "
1808                                 "table\n");
1809                 return ERR_PTR(-ENOMEM);
1810         }
1811
1812         memcpy(&xprt->addr, args->dstaddr, args->addrlen);
1813         xprt->addrlen = args->addrlen;
1814         if (args->srcaddr)
1815                 memcpy(&new->addr, args->srcaddr, args->addrlen);
1816         new->port = xs_get_random_port();
1817
1818         return xprt;
1819 }
1820
1821 /**
1822  * xs_setup_udp - Set up transport to use a UDP socket
1823  * @args: rpc transport creation arguments
1824  *
1825  */
1826 struct rpc_xprt *xs_setup_udp(struct rpc_xprtsock_create *args)
1827 {
1828         struct sockaddr *addr = args->dstaddr;
1829         struct rpc_xprt *xprt;
1830         struct sock_xprt *transport;
1831
1832         xprt = xs_setup_xprt(args, xprt_udp_slot_table_entries);
1833         if (IS_ERR(xprt))
1834                 return xprt;
1835         transport = container_of(xprt, struct sock_xprt, xprt);
1836
1837         xprt->prot = IPPROTO_UDP;
1838         xprt->tsh_size = 0;
1839         /* XXX: header size can vary due to auth type, IPv6, etc. */
1840         xprt->max_payload = (1U << 16) - (MAX_HEADER << 3);
1841
1842         xprt->bind_timeout = XS_BIND_TO;
1843         xprt->connect_timeout = XS_UDP_CONN_TO;
1844         xprt->reestablish_timeout = XS_UDP_REEST_TO;
1845         xprt->idle_timeout = XS_IDLE_DISC_TO;
1846
1847         xprt->ops = &xs_udp_ops;
1848
1849         if (args->timeout)
1850                 xprt->timeout = *args->timeout;
1851         else
1852                 xprt_set_timeout(&xprt->timeout, 5, 5 * HZ);
1853
1854         switch (addr->sa_family) {
1855         case AF_INET:
1856                 if (((struct sockaddr_in *)addr)->sin_port != htons(0))
1857                         xprt_set_bound(xprt);
1858
1859                 INIT_DELAYED_WORK(&transport->connect_worker,
1860                                         xs_udp_connect_worker4);
1861                 xs_format_ipv4_peer_addresses(xprt);
1862                 break;
1863         case AF_INET6:
1864                 if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0))
1865                         xprt_set_bound(xprt);
1866
1867                 INIT_DELAYED_WORK(&transport->connect_worker,
1868                                         xs_udp_connect_worker6);
1869                 xs_format_ipv6_peer_addresses(xprt);
1870                 break;
1871         default:
1872                 kfree(xprt);
1873                 return ERR_PTR(-EAFNOSUPPORT);
1874         }
1875
1876         dprintk("RPC:       set up transport to address %s\n",
1877                         xprt->address_strings[RPC_DISPLAY_ALL]);
1878
1879         return xprt;
1880 }
1881
1882 /**
1883  * xs_setup_tcp - Set up transport to use a TCP socket
1884  * @args: rpc transport creation arguments
1885  *
1886  */
1887 struct rpc_xprt *xs_setup_tcp(struct rpc_xprtsock_create *args)
1888 {
1889         struct sockaddr *addr = args->dstaddr;
1890         struct rpc_xprt *xprt;
1891         struct sock_xprt *transport;
1892
1893         xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries);
1894         if (IS_ERR(xprt))
1895                 return xprt;
1896         transport = container_of(xprt, struct sock_xprt, xprt);
1897
1898         xprt->prot = IPPROTO_TCP;
1899         xprt->tsh_size = sizeof(rpc_fraghdr) / sizeof(u32);
1900         xprt->max_payload = RPC_MAX_FRAGMENT_SIZE;
1901
1902         xprt->bind_timeout = XS_BIND_TO;
1903         xprt->connect_timeout = XS_TCP_CONN_TO;
1904         xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
1905         xprt->idle_timeout = XS_IDLE_DISC_TO;
1906
1907         xprt->ops = &xs_tcp_ops;
1908
1909         if (args->timeout)
1910                 xprt->timeout = *args->timeout;
1911         else
1912                 xprt_set_timeout(&xprt->timeout, 2, 60 * HZ);
1913
1914         switch (addr->sa_family) {
1915         case AF_INET:
1916                 if (((struct sockaddr_in *)addr)->sin_port != htons(0))
1917                         xprt_set_bound(xprt);
1918
1919                 INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_connect_worker4);
1920                 xs_format_ipv4_peer_addresses(xprt);
1921                 break;
1922         case AF_INET6:
1923                 if (((struct sockaddr_in6 *)addr)->sin6_port != htons(0))
1924                         xprt_set_bound(xprt);
1925
1926                 INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_connect_worker6);
1927                 xs_format_ipv6_peer_addresses(xprt);
1928                 break;
1929         default:
1930                 kfree(xprt);
1931                 return ERR_PTR(-EAFNOSUPPORT);
1932         }
1933
1934         dprintk("RPC:       set up transport to address %s\n",
1935                         xprt->address_strings[RPC_DISPLAY_ALL]);
1936
1937         return xprt;
1938 }
1939
1940 /**
1941  * init_socket_xprt - set up xprtsock's sysctls
1942  *
1943  */
1944 int init_socket_xprt(void)
1945 {
1946 #ifdef RPC_DEBUG
1947         if (!sunrpc_table_header)
1948                 sunrpc_table_header = register_sysctl_table(sunrpc_table);
1949 #endif
1950
1951         return 0;
1952 }
1953
1954 /**
1955  * cleanup_socket_xprt - remove xprtsock's sysctls
1956  *
1957  */
1958 void cleanup_socket_xprt(void)
1959 {
1960 #ifdef RPC_DEBUG
1961         if (sunrpc_table_header) {
1962                 unregister_sysctl_table(sunrpc_table_header);
1963                 sunrpc_table_header = NULL;
1964         }
1965 #endif
1966 }