linux/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
<<
>>
Prefs
   1// SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
   2/*
   3 * Copyright (c) 2016-2018 Oracle. All rights reserved.
   4 * Copyright (c) 2014 Open Grid Computing, Inc. All rights reserved.
   5 * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
   6 *
   7 * This software is available to you under a choice of one of two
   8 * licenses.  You may choose to be licensed under the terms of the GNU
   9 * General Public License (GPL) Version 2, available from the file
  10 * COPYING in the main directory of this source tree, or the BSD-type
  11 * license below:
  12 *
  13 * Redistribution and use in source and binary forms, with or without
  14 * modification, are permitted provided that the following conditions
  15 * are met:
  16 *
  17 *      Redistributions of source code must retain the above copyright
  18 *      notice, this list of conditions and the following disclaimer.
  19 *
  20 *      Redistributions in binary form must reproduce the above
  21 *      copyright notice, this list of conditions and the following
  22 *      disclaimer in the documentation and/or other materials provided
  23 *      with the distribution.
  24 *
  25 *      Neither the name of the Network Appliance, Inc. nor the names of
  26 *      its contributors may be used to endorse or promote products
  27 *      derived from this software without specific prior written
  28 *      permission.
  29 *
  30 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  31 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  32 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  33 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  34 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  35 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  36 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  37 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  38 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  39 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  40 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  41 *
  42 * Author: Tom Tucker <tom@opengridcomputing.com>
  43 */
  44
  45/* Operation
  46 *
  47 * The main entry point is svc_rdma_recvfrom. This is called from
  48 * svc_recv when the transport indicates there is incoming data to
  49 * be read. "Data Ready" is signaled when an RDMA Receive completes,
  50 * or when a set of RDMA Reads complete.
  51 *
  52 * An svc_rqst is passed in. This structure contains an array of
  53 * free pages (rq_pages) that will contain the incoming RPC message.
  54 *
  55 * Short messages are moved directly into svc_rqst::rq_arg, and
  56 * the RPC Call is ready to be processed by the Upper Layer.
  57 * svc_rdma_recvfrom returns the length of the RPC Call message,
  58 * completing the reception of the RPC Call.
  59 *
  60 * However, when an incoming message has Read chunks,
  61 * svc_rdma_recvfrom must post RDMA Reads to pull the RPC Call's
  62 * data payload from the client. svc_rdma_recvfrom sets up the
  63 * RDMA Reads using pages in svc_rqst::rq_pages, which are
  64 * transferred to an svc_rdma_recv_ctxt for the duration of the
  65 * I/O. svc_rdma_recvfrom then returns zero, since the RPC message
  66 * is still not yet ready.
  67 *
  68 * When the Read chunk payloads have become available on the
  69 * server, "Data Ready" is raised again, and svc_recv calls
  70 * svc_rdma_recvfrom again. This second call may use a different
  71 * svc_rqst than the first one, thus any information that needs
  72 * to be preserved across these two calls is kept in an
  73 * svc_rdma_recv_ctxt.
  74 *
  75 * The second call to svc_rdma_recvfrom performs final assembly
  76 * of the RPC Call message, using the RDMA Read sink pages kept in
  77 * the svc_rdma_recv_ctxt. The xdr_buf is copied from the
  78 * svc_rdma_recv_ctxt to the second svc_rqst. The second call returns
  79 * the length of the completed RPC Call message.
  80 *
  81 * Page Management
  82 *
  83 * Pages under I/O must be transferred from the first svc_rqst to an
  84 * svc_rdma_recv_ctxt before the first svc_rdma_recvfrom call returns.
  85 *
  86 * The first svc_rqst supplies pages for RDMA Reads. These are moved
  87 * from rqstp::rq_pages into ctxt::pages. The consumed elements of
  88 * the rq_pages array are set to NULL and refilled with the first
  89 * svc_rdma_recvfrom call returns.
  90 *
  91 * During the second svc_rdma_recvfrom call, RDMA Read sink pages
  92 * are transferred from the svc_rdma_recv_ctxt to the second svc_rqst.
  93 */
  94
  95#include <linux/slab.h>
  96#include <linux/spinlock.h>
  97#include <asm/unaligned.h>
  98#include <rdma/ib_verbs.h>
  99#include <rdma/rdma_cm.h>
 100
 101#include <linux/sunrpc/xdr.h>
 102#include <linux/sunrpc/debug.h>
 103#include <linux/sunrpc/rpc_rdma.h>
 104#include <linux/sunrpc/svc_rdma.h>
 105
 106#include "xprt_rdma.h"
 107#include <trace/events/rpcrdma.h>
 108
 109static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc);
 110
 111static inline struct svc_rdma_recv_ctxt *
 112svc_rdma_next_recv_ctxt(struct list_head *list)
 113{
 114        return list_first_entry_or_null(list, struct svc_rdma_recv_ctxt,
 115                                        rc_list);
 116}
 117
 118static void svc_rdma_recv_cid_init(struct svcxprt_rdma *rdma,
 119                                   struct rpc_rdma_cid *cid)
 120{
 121        cid->ci_queue_id = rdma->sc_rq_cq->res.id;
 122        cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
 123}
 124
 125static struct svc_rdma_recv_ctxt *
 126svc_rdma_recv_ctxt_alloc(struct svcxprt_rdma *rdma)
 127{
 128        struct svc_rdma_recv_ctxt *ctxt;
 129        dma_addr_t addr;
 130        void *buffer;
 131
 132        ctxt = kmalloc(sizeof(*ctxt), GFP_KERNEL);
 133        if (!ctxt)
 134                goto fail0;
 135        buffer = kmalloc(rdma->sc_max_req_size, GFP_KERNEL);
 136        if (!buffer)
 137                goto fail1;
 138        addr = ib_dma_map_single(rdma->sc_pd->device, buffer,
 139                                 rdma->sc_max_req_size, DMA_FROM_DEVICE);
 140        if (ib_dma_mapping_error(rdma->sc_pd->device, addr))
 141                goto fail2;
 142
 143        svc_rdma_recv_cid_init(rdma, &ctxt->rc_cid);
 144        pcl_init(&ctxt->rc_call_pcl);
 145        pcl_init(&ctxt->rc_read_pcl);
 146        pcl_init(&ctxt->rc_write_pcl);
 147        pcl_init(&ctxt->rc_reply_pcl);
 148
 149        ctxt->rc_recv_wr.next = NULL;
 150        ctxt->rc_recv_wr.wr_cqe = &ctxt->rc_cqe;
 151        ctxt->rc_recv_wr.sg_list = &ctxt->rc_recv_sge;
 152        ctxt->rc_recv_wr.num_sge = 1;
 153        ctxt->rc_cqe.done = svc_rdma_wc_receive;
 154        ctxt->rc_recv_sge.addr = addr;
 155        ctxt->rc_recv_sge.length = rdma->sc_max_req_size;
 156        ctxt->rc_recv_sge.lkey = rdma->sc_pd->local_dma_lkey;
 157        ctxt->rc_recv_buf = buffer;
 158        ctxt->rc_temp = false;
 159        return ctxt;
 160
 161fail2:
 162        kfree(buffer);
 163fail1:
 164        kfree(ctxt);
 165fail0:
 166        return NULL;
 167}
 168
 169static void svc_rdma_recv_ctxt_destroy(struct svcxprt_rdma *rdma,
 170                                       struct svc_rdma_recv_ctxt *ctxt)
 171{
 172        ib_dma_unmap_single(rdma->sc_pd->device, ctxt->rc_recv_sge.addr,
 173                            ctxt->rc_recv_sge.length, DMA_FROM_DEVICE);
 174        kfree(ctxt->rc_recv_buf);
 175        kfree(ctxt);
 176}
 177
 178/**
 179 * svc_rdma_recv_ctxts_destroy - Release all recv_ctxt's for an xprt
 180 * @rdma: svcxprt_rdma being torn down
 181 *
 182 */
 183void svc_rdma_recv_ctxts_destroy(struct svcxprt_rdma *rdma)
 184{
 185        struct svc_rdma_recv_ctxt *ctxt;
 186        struct llist_node *node;
 187
 188        while ((node = llist_del_first(&rdma->sc_recv_ctxts))) {
 189                ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
 190                svc_rdma_recv_ctxt_destroy(rdma, ctxt);
 191        }
 192}
 193
 194/**
 195 * svc_rdma_recv_ctxt_get - Allocate a recv_ctxt
 196 * @rdma: controlling svcxprt_rdma
 197 *
 198 * Returns a recv_ctxt or (rarely) NULL if none are available.
 199 */
 200struct svc_rdma_recv_ctxt *svc_rdma_recv_ctxt_get(struct svcxprt_rdma *rdma)
 201{
 202        struct svc_rdma_recv_ctxt *ctxt;
 203        struct llist_node *node;
 204
 205        node = llist_del_first(&rdma->sc_recv_ctxts);
 206        if (!node)
 207                goto out_empty;
 208        ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
 209
 210out:
 211        ctxt->rc_page_count = 0;
 212        return ctxt;
 213
 214out_empty:
 215        ctxt = svc_rdma_recv_ctxt_alloc(rdma);
 216        if (!ctxt)
 217                return NULL;
 218        goto out;
 219}
 220
 221/**
 222 * svc_rdma_recv_ctxt_put - Return recv_ctxt to free list
 223 * @rdma: controlling svcxprt_rdma
 224 * @ctxt: object to return to the free list
 225 *
 226 */
 227void svc_rdma_recv_ctxt_put(struct svcxprt_rdma *rdma,
 228                            struct svc_rdma_recv_ctxt *ctxt)
 229{
 230        pcl_free(&ctxt->rc_call_pcl);
 231        pcl_free(&ctxt->rc_read_pcl);
 232        pcl_free(&ctxt->rc_write_pcl);
 233        pcl_free(&ctxt->rc_reply_pcl);
 234
 235        if (!ctxt->rc_temp)
 236                llist_add(&ctxt->rc_node, &rdma->sc_recv_ctxts);
 237        else
 238                svc_rdma_recv_ctxt_destroy(rdma, ctxt);
 239}
 240
 241/**
 242 * svc_rdma_release_rqst - Release transport-specific per-rqst resources
 243 * @rqstp: svc_rqst being released
 244 *
 245 * Ensure that the recv_ctxt is released whether or not a Reply
 246 * was sent. For example, the client could close the connection,
 247 * or svc_process could drop an RPC, before the Reply is sent.
 248 */
 249void svc_rdma_release_rqst(struct svc_rqst *rqstp)
 250{
 251        struct svc_rdma_recv_ctxt *ctxt = rqstp->rq_xprt_ctxt;
 252        struct svc_xprt *xprt = rqstp->rq_xprt;
 253        struct svcxprt_rdma *rdma =
 254                container_of(xprt, struct svcxprt_rdma, sc_xprt);
 255
 256        rqstp->rq_xprt_ctxt = NULL;
 257        if (ctxt)
 258                svc_rdma_recv_ctxt_put(rdma, ctxt);
 259}
 260
 261static bool svc_rdma_refresh_recvs(struct svcxprt_rdma *rdma,
 262                                   unsigned int wanted, bool temp)
 263{
 264        const struct ib_recv_wr *bad_wr = NULL;
 265        struct svc_rdma_recv_ctxt *ctxt;
 266        struct ib_recv_wr *recv_chain;
 267        int ret;
 268
 269        if (test_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags))
 270                return false;
 271
 272        recv_chain = NULL;
 273        while (wanted--) {
 274                ctxt = svc_rdma_recv_ctxt_get(rdma);
 275                if (!ctxt)
 276                        break;
 277
 278                trace_svcrdma_post_recv(ctxt);
 279                ctxt->rc_temp = temp;
 280                ctxt->rc_recv_wr.next = recv_chain;
 281                recv_chain = &ctxt->rc_recv_wr;
 282                rdma->sc_pending_recvs++;
 283        }
 284        if (!recv_chain)
 285                return false;
 286
 287        ret = ib_post_recv(rdma->sc_qp, recv_chain, &bad_wr);
 288        if (ret)
 289                goto err_free;
 290        return true;
 291
 292err_free:
 293        trace_svcrdma_rq_post_err(rdma, ret);
 294        while (bad_wr) {
 295                ctxt = container_of(bad_wr, struct svc_rdma_recv_ctxt,
 296                                    rc_recv_wr);
 297                bad_wr = bad_wr->next;
 298                svc_rdma_recv_ctxt_put(rdma, ctxt);
 299        }
 300        /* Since we're destroying the xprt, no need to reset
 301         * sc_pending_recvs. */
 302        return false;
 303}
 304
 305/**
 306 * svc_rdma_post_recvs - Post initial set of Recv WRs
 307 * @rdma: fresh svcxprt_rdma
 308 *
 309 * Returns true if successful, otherwise false.
 310 */
 311bool svc_rdma_post_recvs(struct svcxprt_rdma *rdma)
 312{
 313        return svc_rdma_refresh_recvs(rdma, rdma->sc_max_requests, true);
 314}
 315
 316/**
 317 * svc_rdma_wc_receive - Invoked by RDMA provider for each polled Receive WC
 318 * @cq: Completion Queue context
 319 * @wc: Work Completion object
 320 *
 321 */
 322static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
 323{
 324        struct svcxprt_rdma *rdma = cq->cq_context;
 325        struct ib_cqe *cqe = wc->wr_cqe;
 326        struct svc_rdma_recv_ctxt *ctxt;
 327
 328        rdma->sc_pending_recvs--;
 329
 330        /* WARNING: Only wc->wr_cqe and wc->status are reliable */
 331        ctxt = container_of(cqe, struct svc_rdma_recv_ctxt, rc_cqe);
 332
 333        trace_svcrdma_wc_receive(wc, &ctxt->rc_cid);
 334        if (wc->status != IB_WC_SUCCESS)
 335                goto flushed;
 336
 337        /* If receive posting fails, the connection is about to be
 338         * lost anyway. The server will not be able to send a reply
 339         * for this RPC, and the client will retransmit this RPC
 340         * anyway when it reconnects.
 341         *
 342         * Therefore we drop the Receive, even if status was SUCCESS
 343         * to reduce the likelihood of replayed requests once the
 344         * client reconnects.
 345         */
 346        if (rdma->sc_pending_recvs < rdma->sc_max_requests)
 347                if (!svc_rdma_refresh_recvs(rdma, rdma->sc_recv_batch, false))
 348                        goto flushed;
 349
 350        /* All wc fields are now known to be valid */
 351        ctxt->rc_byte_len = wc->byte_len;
 352
 353        spin_lock(&rdma->sc_rq_dto_lock);
 354        list_add_tail(&ctxt->rc_list, &rdma->sc_rq_dto_q);
 355        /* Note the unlock pairs with the smp_rmb in svc_xprt_ready: */
 356        set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
 357        spin_unlock(&rdma->sc_rq_dto_lock);
 358        if (!test_bit(RDMAXPRT_CONN_PENDING, &rdma->sc_flags))
 359                svc_xprt_enqueue(&rdma->sc_xprt);
 360        return;
 361
 362flushed:
 363        svc_rdma_recv_ctxt_put(rdma, ctxt);
 364        svc_xprt_deferred_close(&rdma->sc_xprt);
 365}
 366
 367/**
 368 * svc_rdma_flush_recv_queues - Drain pending Receive work
 369 * @rdma: svcxprt_rdma being shut down
 370 *
 371 */
 372void svc_rdma_flush_recv_queues(struct svcxprt_rdma *rdma)
 373{
 374        struct svc_rdma_recv_ctxt *ctxt;
 375
 376        while ((ctxt = svc_rdma_next_recv_ctxt(&rdma->sc_rq_dto_q))) {
 377                list_del(&ctxt->rc_list);
 378                svc_rdma_recv_ctxt_put(rdma, ctxt);
 379        }
 380}
 381
 382static void svc_rdma_build_arg_xdr(struct svc_rqst *rqstp,
 383                                   struct svc_rdma_recv_ctxt *ctxt)
 384{
 385        struct xdr_buf *arg = &rqstp->rq_arg;
 386
 387        arg->head[0].iov_base = ctxt->rc_recv_buf;
 388        arg->head[0].iov_len = ctxt->rc_byte_len;
 389        arg->tail[0].iov_base = NULL;
 390        arg->tail[0].iov_len = 0;
 391        arg->page_len = 0;
 392        arg->page_base = 0;
 393        arg->buflen = ctxt->rc_byte_len;
 394        arg->len = ctxt->rc_byte_len;
 395}
 396
 397/**
 398 * xdr_count_read_segments - Count number of Read segments in Read list
 399 * @rctxt: Ingress receive context
 400 * @p: Start of an un-decoded Read list
 401 *
 402 * Before allocating anything, ensure the ingress Read list is safe
 403 * to use.
 404 *
 405 * The segment count is limited to how many segments can fit in the
 406 * transport header without overflowing the buffer. That's about 40
 407 * Read segments for a 1KB inline threshold.
 408 *
 409 * Return values:
 410 *   %true: Read list is valid. @rctxt's xdr_stream is updated to point
 411 *          to the first byte past the Read list. rc_read_pcl and
 412 *          rc_call_pcl cl_count fields are set to the number of
 413 *          Read segments in the list.
 414 *  %false: Read list is corrupt. @rctxt's xdr_stream is left in an
 415 *          unknown state.
 416 */
 417static bool xdr_count_read_segments(struct svc_rdma_recv_ctxt *rctxt, __be32 *p)
 418{
 419        rctxt->rc_call_pcl.cl_count = 0;
 420        rctxt->rc_read_pcl.cl_count = 0;
 421        while (xdr_item_is_present(p)) {
 422                u32 position, handle, length;
 423                u64 offset;
 424
 425                p = xdr_inline_decode(&rctxt->rc_stream,
 426                                      rpcrdma_readseg_maxsz * sizeof(*p));
 427                if (!p)
 428                        return false;
 429
 430                xdr_decode_read_segment(p, &position, &handle,
 431                                            &length, &offset);
 432                if (position) {
 433                        if (position & 3)
 434                                return false;
 435                        ++rctxt->rc_read_pcl.cl_count;
 436                } else {
 437                        ++rctxt->rc_call_pcl.cl_count;
 438                }
 439
 440                p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
 441                if (!p)
 442                        return false;
 443        }
 444        return true;
 445}
 446
 447/* Sanity check the Read list.
 448 *
 449 * Sanity checks:
 450 * - Read list does not overflow Receive buffer.
 451 * - Chunk size limited by largest NFS data payload.
 452 *
 453 * Return values:
 454 *   %true: Read list is valid. @rctxt's xdr_stream is updated
 455 *          to point to the first byte past the Read list.
 456 *  %false: Read list is corrupt. @rctxt's xdr_stream is left
 457 *          in an unknown state.
 458 */
 459static bool xdr_check_read_list(struct svc_rdma_recv_ctxt *rctxt)
 460{
 461        __be32 *p;
 462
 463        p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
 464        if (!p)
 465                return false;
 466        if (!xdr_count_read_segments(rctxt, p))
 467                return false;
 468        if (!pcl_alloc_call(rctxt, p))
 469                return false;
 470        return pcl_alloc_read(rctxt, p);
 471}
 472
 473static bool xdr_check_write_chunk(struct svc_rdma_recv_ctxt *rctxt)
 474{
 475        u32 segcount;
 476        __be32 *p;
 477
 478        if (xdr_stream_decode_u32(&rctxt->rc_stream, &segcount))
 479                return false;
 480
 481        /* A bogus segcount causes this buffer overflow check to fail. */
 482        p = xdr_inline_decode(&rctxt->rc_stream,
 483                              segcount * rpcrdma_segment_maxsz * sizeof(*p));
 484        return p != NULL;
 485}
 486
 487/**
 488 * xdr_count_write_chunks - Count number of Write chunks in Write list
 489 * @rctxt: Received header and decoding state
 490 * @p: start of an un-decoded Write list
 491 *
 492 * Before allocating anything, ensure the ingress Write list is
 493 * safe to use.
 494 *
 495 * Return values:
 496 *       %true: Write list is valid. @rctxt's xdr_stream is updated
 497 *              to point to the first byte past the Write list, and
 498 *              the number of Write chunks is in rc_write_pcl.cl_count.
 499 *      %false: Write list is corrupt. @rctxt's xdr_stream is left
 500 *              in an indeterminate state.
 501 */
 502static bool xdr_count_write_chunks(struct svc_rdma_recv_ctxt *rctxt, __be32 *p)
 503{
 504        rctxt->rc_write_pcl.cl_count = 0;
 505        while (xdr_item_is_present(p)) {
 506                if (!xdr_check_write_chunk(rctxt))
 507                        return false;
 508                ++rctxt->rc_write_pcl.cl_count;
 509                p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
 510                if (!p)
 511                        return false;
 512        }
 513        return true;
 514}
 515
 516/* Sanity check the Write list.
 517 *
 518 * Implementation limits:
 519 * - This implementation currently supports only one Write chunk.
 520 *
 521 * Sanity checks:
 522 * - Write list does not overflow Receive buffer.
 523 * - Chunk size limited by largest NFS data payload.
 524 *
 525 * Return values:
 526 *       %true: Write list is valid. @rctxt's xdr_stream is updated
 527 *              to point to the first byte past the Write list.
 528 *      %false: Write list is corrupt. @rctxt's xdr_stream is left
 529 *              in an unknown state.
 530 */
 531static bool xdr_check_write_list(struct svc_rdma_recv_ctxt *rctxt)
 532{
 533        __be32 *p;
 534
 535        p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
 536        if (!p)
 537                return false;
 538        if (!xdr_count_write_chunks(rctxt, p))
 539                return false;
 540        if (!pcl_alloc_write(rctxt, &rctxt->rc_write_pcl, p))
 541                return false;
 542
 543        rctxt->rc_cur_result_payload = pcl_first_chunk(&rctxt->rc_write_pcl);
 544        return true;
 545}
 546
 547/* Sanity check the Reply chunk.
 548 *
 549 * Sanity checks:
 550 * - Reply chunk does not overflow Receive buffer.
 551 * - Chunk size limited by largest NFS data payload.
 552 *
 553 * Return values:
 554 *       %true: Reply chunk is valid. @rctxt's xdr_stream is updated
 555 *              to point to the first byte past the Reply chunk.
 556 *      %false: Reply chunk is corrupt. @rctxt's xdr_stream is left
 557 *              in an unknown state.
 558 */
 559static bool xdr_check_reply_chunk(struct svc_rdma_recv_ctxt *rctxt)
 560{
 561        __be32 *p;
 562
 563        p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
 564        if (!p)
 565                return false;
 566
 567        if (!xdr_item_is_present(p))
 568                return true;
 569        if (!xdr_check_write_chunk(rctxt))
 570                return false;
 571
 572        rctxt->rc_reply_pcl.cl_count = 1;
 573        return pcl_alloc_write(rctxt, &rctxt->rc_reply_pcl, p);
 574}
 575
 576/* RPC-over-RDMA Version One private extension: Remote Invalidation.
 577 * Responder's choice: requester signals it can handle Send With
 578 * Invalidate, and responder chooses one R_key to invalidate.
 579 *
 580 * If there is exactly one distinct R_key in the received transport
 581 * header, set rc_inv_rkey to that R_key. Otherwise, set it to zero.
 582 */
 583static void svc_rdma_get_inv_rkey(struct svcxprt_rdma *rdma,
 584                                  struct svc_rdma_recv_ctxt *ctxt)
 585{
 586        struct svc_rdma_segment *segment;
 587        struct svc_rdma_chunk *chunk;
 588        u32 inv_rkey;
 589
 590        ctxt->rc_inv_rkey = 0;
 591
 592        if (!rdma->sc_snd_w_inv)
 593                return;
 594
 595        inv_rkey = 0;
 596        pcl_for_each_chunk(chunk, &ctxt->rc_call_pcl) {
 597                pcl_for_each_segment(segment, chunk) {
 598                        if (inv_rkey == 0)
 599                                inv_rkey = segment->rs_handle;
 600                        else if (inv_rkey != segment->rs_handle)
 601                                return;
 602                }
 603        }
 604        pcl_for_each_chunk(chunk, &ctxt->rc_read_pcl) {
 605                pcl_for_each_segment(segment, chunk) {
 606                        if (inv_rkey == 0)
 607                                inv_rkey = segment->rs_handle;
 608                        else if (inv_rkey != segment->rs_handle)
 609                                return;
 610                }
 611        }
 612        pcl_for_each_chunk(chunk, &ctxt->rc_write_pcl) {
 613                pcl_for_each_segment(segment, chunk) {
 614                        if (inv_rkey == 0)
 615                                inv_rkey = segment->rs_handle;
 616                        else if (inv_rkey != segment->rs_handle)
 617                                return;
 618                }
 619        }
 620        pcl_for_each_chunk(chunk, &ctxt->rc_reply_pcl) {
 621                pcl_for_each_segment(segment, chunk) {
 622                        if (inv_rkey == 0)
 623                                inv_rkey = segment->rs_handle;
 624                        else if (inv_rkey != segment->rs_handle)
 625                                return;
 626                }
 627        }
 628        ctxt->rc_inv_rkey = inv_rkey;
 629}
 630
 631/**
 632 * svc_rdma_xdr_decode_req - Decode the transport header
 633 * @rq_arg: xdr_buf containing ingress RPC/RDMA message
 634 * @rctxt: state of decoding
 635 *
 636 * On entry, xdr->head[0].iov_base points to first byte of the
 637 * RPC-over-RDMA transport header.
 638 *
 639 * On successful exit, head[0] points to first byte past the
 640 * RPC-over-RDMA header. For RDMA_MSG, this is the RPC message.
 641 *
 642 * The length of the RPC-over-RDMA header is returned.
 643 *
 644 * Assumptions:
 645 * - The transport header is entirely contained in the head iovec.
 646 */
 647static int svc_rdma_xdr_decode_req(struct xdr_buf *rq_arg,
 648                                   struct svc_rdma_recv_ctxt *rctxt)
 649{
 650        __be32 *p, *rdma_argp;
 651        unsigned int hdr_len;
 652
 653        rdma_argp = rq_arg->head[0].iov_base;
 654        xdr_init_decode(&rctxt->rc_stream, rq_arg, rdma_argp, NULL);
 655
 656        p = xdr_inline_decode(&rctxt->rc_stream,
 657                              rpcrdma_fixed_maxsz * sizeof(*p));
 658        if (unlikely(!p))
 659                goto out_short;
 660        p++;
 661        if (*p != rpcrdma_version)
 662                goto out_version;
 663        p += 2;
 664        rctxt->rc_msgtype = *p;
 665        switch (rctxt->rc_msgtype) {
 666        case rdma_msg:
 667                break;
 668        case rdma_nomsg:
 669                break;
 670        case rdma_done:
 671                goto out_drop;
 672        case rdma_error:
 673                goto out_drop;
 674        default:
 675                goto out_proc;
 676        }
 677
 678        if (!xdr_check_read_list(rctxt))
 679                goto out_inval;
 680        if (!xdr_check_write_list(rctxt))
 681                goto out_inval;
 682        if (!xdr_check_reply_chunk(rctxt))
 683                goto out_inval;
 684
 685        rq_arg->head[0].iov_base = rctxt->rc_stream.p;
 686        hdr_len = xdr_stream_pos(&rctxt->rc_stream);
 687        rq_arg->head[0].iov_len -= hdr_len;
 688        rq_arg->len -= hdr_len;
 689        trace_svcrdma_decode_rqst(rctxt, rdma_argp, hdr_len);
 690        return hdr_len;
 691
 692out_short:
 693        trace_svcrdma_decode_short_err(rctxt, rq_arg->len);
 694        return -EINVAL;
 695
 696out_version:
 697        trace_svcrdma_decode_badvers_err(rctxt, rdma_argp);
 698        return -EPROTONOSUPPORT;
 699
 700out_drop:
 701        trace_svcrdma_decode_drop_err(rctxt, rdma_argp);
 702        return 0;
 703
 704out_proc:
 705        trace_svcrdma_decode_badproc_err(rctxt, rdma_argp);
 706        return -EINVAL;
 707
 708out_inval:
 709        trace_svcrdma_decode_parse_err(rctxt, rdma_argp);
 710        return -EINVAL;
 711}
 712
 713static void svc_rdma_send_error(struct svcxprt_rdma *rdma,
 714                                struct svc_rdma_recv_ctxt *rctxt,
 715                                int status)
 716{
 717        struct svc_rdma_send_ctxt *sctxt;
 718
 719        sctxt = svc_rdma_send_ctxt_get(rdma);
 720        if (!sctxt)
 721                return;
 722        svc_rdma_send_error_msg(rdma, sctxt, rctxt, status);
 723}
 724
 725/* By convention, backchannel calls arrive via rdma_msg type
 726 * messages, and never populate the chunk lists. This makes
 727 * the RPC/RDMA header small and fixed in size, so it is
 728 * straightforward to check the RPC header's direction field.
 729 */
 730static bool svc_rdma_is_reverse_direction_reply(struct svc_xprt *xprt,
 731                                                struct svc_rdma_recv_ctxt *rctxt)
 732{
 733        __be32 *p = rctxt->rc_recv_buf;
 734
 735        if (!xprt->xpt_bc_xprt)
 736                return false;
 737
 738        if (rctxt->rc_msgtype != rdma_msg)
 739                return false;
 740
 741        if (!pcl_is_empty(&rctxt->rc_call_pcl))
 742                return false;
 743        if (!pcl_is_empty(&rctxt->rc_read_pcl))
 744                return false;
 745        if (!pcl_is_empty(&rctxt->rc_write_pcl))
 746                return false;
 747        if (!pcl_is_empty(&rctxt->rc_reply_pcl))
 748                return false;
 749
 750        /* RPC call direction */
 751        if (*(p + 8) == cpu_to_be32(RPC_CALL))
 752                return false;
 753
 754        return true;
 755}
 756
 757/**
 758 * svc_rdma_recvfrom - Receive an RPC call
 759 * @rqstp: request structure into which to receive an RPC Call
 760 *
 761 * Returns:
 762 *      The positive number of bytes in the RPC Call message,
 763 *      %0 if there were no Calls ready to return,
 764 *      %-EINVAL if the Read chunk data is too large,
 765 *      %-ENOMEM if rdma_rw context pool was exhausted,
 766 *      %-ENOTCONN if posting failed (connection is lost),
 767 *      %-EIO if rdma_rw initialization failed (DMA mapping, etc).
 768 *
 769 * Called in a loop when XPT_DATA is set. XPT_DATA is cleared only
 770 * when there are no remaining ctxt's to process.
 771 *
 772 * The next ctxt is removed from the "receive" lists.
 773 *
 774 * - If the ctxt completes a Read, then finish assembling the Call
 775 *   message and return the number of bytes in the message.
 776 *
 777 * - If the ctxt completes a Receive, then construct the Call
 778 *   message from the contents of the Receive buffer.
 779 *
 780 *   - If there are no Read chunks in this message, then finish
 781 *     assembling the Call message and return the number of bytes
 782 *     in the message.
 783 *
 784 *   - If there are Read chunks in this message, post Read WRs to
 785 *     pull that payload and return 0.
 786 */
 787int svc_rdma_recvfrom(struct svc_rqst *rqstp)
 788{
 789        struct svc_xprt *xprt = rqstp->rq_xprt;
 790        struct svcxprt_rdma *rdma_xprt =
 791                container_of(xprt, struct svcxprt_rdma, sc_xprt);
 792        struct svc_rdma_recv_ctxt *ctxt;
 793        int ret;
 794
 795        rqstp->rq_xprt_ctxt = NULL;
 796
 797        ctxt = NULL;
 798        spin_lock(&rdma_xprt->sc_rq_dto_lock);
 799        ctxt = svc_rdma_next_recv_ctxt(&rdma_xprt->sc_rq_dto_q);
 800        if (ctxt)
 801                list_del(&ctxt->rc_list);
 802        else
 803                /* No new incoming requests, terminate the loop */
 804                clear_bit(XPT_DATA, &xprt->xpt_flags);
 805        spin_unlock(&rdma_xprt->sc_rq_dto_lock);
 806
 807        /* Unblock the transport for the next receive */
 808        svc_xprt_received(xprt);
 809        if (!ctxt)
 810                return 0;
 811
 812        percpu_counter_inc(&svcrdma_stat_recv);
 813        ib_dma_sync_single_for_cpu(rdma_xprt->sc_pd->device,
 814                                   ctxt->rc_recv_sge.addr, ctxt->rc_byte_len,
 815                                   DMA_FROM_DEVICE);
 816        svc_rdma_build_arg_xdr(rqstp, ctxt);
 817
 818        /* Prevent svc_xprt_release from releasing pages in rq_pages
 819         * if we return 0 or an error.
 820         */
 821        rqstp->rq_respages = rqstp->rq_pages;
 822        rqstp->rq_next_page = rqstp->rq_respages;
 823
 824        ret = svc_rdma_xdr_decode_req(&rqstp->rq_arg, ctxt);
 825        if (ret < 0)
 826                goto out_err;
 827        if (ret == 0)
 828                goto out_drop;
 829        rqstp->rq_xprt_hlen = ret;
 830
 831        if (svc_rdma_is_reverse_direction_reply(xprt, ctxt))
 832                goto out_backchannel;
 833
 834        svc_rdma_get_inv_rkey(rdma_xprt, ctxt);
 835
 836        if (!pcl_is_empty(&ctxt->rc_read_pcl) ||
 837            !pcl_is_empty(&ctxt->rc_call_pcl)) {
 838                ret = svc_rdma_process_read_list(rdma_xprt, rqstp, ctxt);
 839                if (ret < 0)
 840                        goto out_readfail;
 841        }
 842
 843        rqstp->rq_xprt_ctxt = ctxt;
 844        rqstp->rq_prot = IPPROTO_MAX;
 845        svc_xprt_copy_addrs(rqstp, xprt);
 846        return rqstp->rq_arg.len;
 847
 848out_err:
 849        svc_rdma_send_error(rdma_xprt, ctxt, ret);
 850        svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
 851        return 0;
 852
 853out_readfail:
 854        if (ret == -EINVAL)
 855                svc_rdma_send_error(rdma_xprt, ctxt, ret);
 856        svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
 857        return ret;
 858
 859out_backchannel:
 860        svc_rdma_handle_bc_reply(rqstp, ctxt);
 861out_drop:
 862        svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
 863        return 0;
 864}
 865