Loading...
1/*
2 * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the BSD-type
8 * license below:
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 *
14 * Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 *
17 * Redistributions in binary form must reproduce the above
18 * copyright notice, this list of conditions and the following
19 * disclaimer in the documentation and/or other materials provided
20 * with the distribution.
21 *
22 * Neither the name of the Network Appliance, Inc. nor the names of
23 * its contributors may be used to endorse or promote products
24 * derived from this software without specific prior written
25 * permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
30 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
31 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
32 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
33 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
34 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
35 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
36 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
37 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
38 */
39
40/*
41 * rpc_rdma.c
42 *
43 * This file contains the guts of the RPC RDMA protocol, and
44 * does marshaling/unmarshaling, etc. It is also where interfacing
45 * to the Linux RPC framework lives.
46 */
47
48#include "xprt_rdma.h"
49
50#include <linux/highmem.h>
51
52#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
53# define RPCDBG_FACILITY RPCDBG_TRANS
54#endif
55
56enum rpcrdma_chunktype {
57 rpcrdma_noch = 0,
58 rpcrdma_readch,
59 rpcrdma_areadch,
60 rpcrdma_writech,
61 rpcrdma_replych
62};
63
64#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
65static const char transfertypes[][12] = {
66 "pure inline", /* no chunks */
67 " read chunk", /* some argument via rdma read */
68 "*read chunk", /* entire request via rdma read */
69 "write chunk", /* some result via rdma write */
70 "reply chunk" /* entire reply via rdma write */
71};
72#endif
73
74/* The client can send a request inline as long as the RPCRDMA header
75 * plus the RPC call fit under the transport's inline limit. If the
76 * combined call message size exceeds that limit, the client must use
77 * the read chunk list for this operation.
78 */
79static bool rpcrdma_args_inline(struct rpc_rqst *rqst)
80{
81 unsigned int callsize = RPCRDMA_HDRLEN_MIN + rqst->rq_snd_buf.len;
82
83 return callsize <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst);
84}
85
86/* The client can't know how large the actual reply will be. Thus it
87 * plans for the largest possible reply for that particular ULP
88 * operation. If the maximum combined reply message size exceeds that
89 * limit, the client must provide a write list or a reply chunk for
90 * this request.
91 */
92static bool rpcrdma_results_inline(struct rpc_rqst *rqst)
93{
94 unsigned int repsize = RPCRDMA_HDRLEN_MIN + rqst->rq_rcv_buf.buflen;
95
96 return repsize <= RPCRDMA_INLINE_READ_THRESHOLD(rqst);
97}
98
99static int
100rpcrdma_tail_pullup(struct xdr_buf *buf)
101{
102 size_t tlen = buf->tail[0].iov_len;
103 size_t skip = tlen & 3;
104
105 /* Do not include the tail if it is only an XDR pad */
106 if (tlen < 4)
107 return 0;
108
109 /* xdr_write_pages() adds a pad at the beginning of the tail
110 * if the content in "buf->pages" is unaligned. Force the
111 * tail's actual content to land at the next XDR position
112 * after the head instead.
113 */
114 if (skip) {
115 unsigned char *src, *dst;
116 unsigned int count;
117
118 src = buf->tail[0].iov_base;
119 dst = buf->head[0].iov_base;
120 dst += buf->head[0].iov_len;
121
122 src += skip;
123 tlen -= skip;
124
125 dprintk("RPC: %s: skip=%zu, memmove(%p, %p, %zu)\n",
126 __func__, skip, dst, src, tlen);
127
128 for (count = tlen; count; count--)
129 *dst++ = *src++;
130 }
131
132 return tlen;
133}
134
135/* Split "vec" on page boundaries into segments. FMR registers pages,
136 * not a byte range. Other modes coalesce these segments into a single
137 * MR when they can.
138 */
139static int
140rpcrdma_convert_kvec(struct kvec *vec, struct rpcrdma_mr_seg *seg,
141 int n, int nsegs)
142{
143 size_t page_offset;
144 u32 remaining;
145 char *base;
146
147 base = vec->iov_base;
148 page_offset = offset_in_page(base);
149 remaining = vec->iov_len;
150 while (remaining && n < nsegs) {
151 seg[n].mr_page = NULL;
152 seg[n].mr_offset = base;
153 seg[n].mr_len = min_t(u32, PAGE_SIZE - page_offset, remaining);
154 remaining -= seg[n].mr_len;
155 base += seg[n].mr_len;
156 ++n;
157 page_offset = 0;
158 }
159 return n;
160}
161
162/*
163 * Chunk assembly from upper layer xdr_buf.
164 *
165 * Prepare the passed-in xdr_buf into representation as RPC/RDMA chunk
166 * elements. Segments are then coalesced when registered, if possible
167 * within the selected memreg mode.
168 *
169 * Returns positive number of segments converted, or a negative errno.
170 */
171
172static int
173rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos,
174 enum rpcrdma_chunktype type, struct rpcrdma_mr_seg *seg, int nsegs)
175{
176 int len, n = 0, p;
177 int page_base;
178 struct page **ppages;
179
180 if (pos == 0) {
181 n = rpcrdma_convert_kvec(&xdrbuf->head[0], seg, n, nsegs);
182 if (n == nsegs)
183 return -EIO;
184 }
185
186 len = xdrbuf->page_len;
187 ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT);
188 page_base = xdrbuf->page_base & ~PAGE_MASK;
189 p = 0;
190 while (len && n < nsegs) {
191 if (!ppages[p]) {
192 /* alloc the pagelist for receiving buffer */
193 ppages[p] = alloc_page(GFP_ATOMIC);
194 if (!ppages[p])
195 return -ENOMEM;
196 }
197 seg[n].mr_page = ppages[p];
198 seg[n].mr_offset = (void *)(unsigned long) page_base;
199 seg[n].mr_len = min_t(u32, PAGE_SIZE - page_base, len);
200 if (seg[n].mr_len > PAGE_SIZE)
201 return -EIO;
202 len -= seg[n].mr_len;
203 ++n;
204 ++p;
205 page_base = 0; /* page offset only applies to first page */
206 }
207
208 /* Message overflows the seg array */
209 if (len && n == nsegs)
210 return -EIO;
211
212 /* When encoding the read list, the tail is always sent inline */
213 if (type == rpcrdma_readch)
214 return n;
215
216 if (xdrbuf->tail[0].iov_len) {
217 /* the rpcrdma protocol allows us to omit any trailing
218 * xdr pad bytes, saving the server an RDMA operation. */
219 if (xdrbuf->tail[0].iov_len < 4 && xprt_rdma_pad_optimize)
220 return n;
221 n = rpcrdma_convert_kvec(&xdrbuf->tail[0], seg, n, nsegs);
222 if (n == nsegs)
223 return -EIO;
224 }
225
226 return n;
227}
228
229/*
230 * Create read/write chunk lists, and reply chunks, for RDMA
231 *
232 * Assume check against THRESHOLD has been done, and chunks are required.
233 * Assume only encoding one list entry for read|write chunks. The NFSv3
234 * protocol is simple enough to allow this as it only has a single "bulk
235 * result" in each procedure - complicated NFSv4 COMPOUNDs are not. (The
236 * RDMA/Sessions NFSv4 proposal addresses this for future v4 revs.)
237 *
238 * When used for a single reply chunk (which is a special write
239 * chunk used for the entire reply, rather than just the data), it
240 * is used primarily for READDIR and READLINK which would otherwise
241 * be severely size-limited by a small rdma inline read max. The server
242 * response will come back as an RDMA Write, followed by a message
243 * of type RDMA_NOMSG carrying the xid and length. As a result, reply
244 * chunks do not provide data alignment, however they do not require
245 * "fixup" (moving the response to the upper layer buffer) either.
246 *
247 * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64):
248 *
249 * Read chunklist (a linked list):
250 * N elements, position P (same P for all chunks of same arg!):
251 * 1 - PHLOO - 1 - PHLOO - ... - 1 - PHLOO - 0
252 *
253 * Write chunklist (a list of (one) counted array):
254 * N elements:
255 * 1 - N - HLOO - HLOO - ... - HLOO - 0
256 *
257 * Reply chunk (a counted array):
258 * N elements:
259 * 1 - N - HLOO - HLOO - ... - HLOO
260 *
261 * Returns positive RPC/RDMA header size, or negative errno.
262 */
263
264static ssize_t
265rpcrdma_create_chunks(struct rpc_rqst *rqst, struct xdr_buf *target,
266 struct rpcrdma_msg *headerp, enum rpcrdma_chunktype type)
267{
268 struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
269 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
270 int n, nsegs, nchunks = 0;
271 unsigned int pos;
272 struct rpcrdma_mr_seg *seg = req->rl_segments;
273 struct rpcrdma_read_chunk *cur_rchunk = NULL;
274 struct rpcrdma_write_array *warray = NULL;
275 struct rpcrdma_write_chunk *cur_wchunk = NULL;
276 __be32 *iptr = headerp->rm_body.rm_chunks;
277 int (*map)(struct rpcrdma_xprt *, struct rpcrdma_mr_seg *, int, bool);
278
279 if (type == rpcrdma_readch || type == rpcrdma_areadch) {
280 /* a read chunk - server will RDMA Read our memory */
281 cur_rchunk = (struct rpcrdma_read_chunk *) iptr;
282 } else {
283 /* a write or reply chunk - server will RDMA Write our memory */
284 *iptr++ = xdr_zero; /* encode a NULL read chunk list */
285 if (type == rpcrdma_replych)
286 *iptr++ = xdr_zero; /* a NULL write chunk list */
287 warray = (struct rpcrdma_write_array *) iptr;
288 cur_wchunk = (struct rpcrdma_write_chunk *) (warray + 1);
289 }
290
291 if (type == rpcrdma_replych || type == rpcrdma_areadch)
292 pos = 0;
293 else
294 pos = target->head[0].iov_len;
295
296 nsegs = rpcrdma_convert_iovs(target, pos, type, seg, RPCRDMA_MAX_SEGS);
297 if (nsegs < 0)
298 return nsegs;
299
300 map = r_xprt->rx_ia.ri_ops->ro_map;
301 do {
302 n = map(r_xprt, seg, nsegs, cur_wchunk != NULL);
303 if (n <= 0)
304 goto out;
305 if (cur_rchunk) { /* read */
306 cur_rchunk->rc_discrim = xdr_one;
307 /* all read chunks have the same "position" */
308 cur_rchunk->rc_position = cpu_to_be32(pos);
309 cur_rchunk->rc_target.rs_handle =
310 cpu_to_be32(seg->mr_rkey);
311 cur_rchunk->rc_target.rs_length =
312 cpu_to_be32(seg->mr_len);
313 xdr_encode_hyper(
314 (__be32 *)&cur_rchunk->rc_target.rs_offset,
315 seg->mr_base);
316 dprintk("RPC: %s: read chunk "
317 "elem %d@0x%llx:0x%x pos %u (%s)\n", __func__,
318 seg->mr_len, (unsigned long long)seg->mr_base,
319 seg->mr_rkey, pos, n < nsegs ? "more" : "last");
320 cur_rchunk++;
321 r_xprt->rx_stats.read_chunk_count++;
322 } else { /* write/reply */
323 cur_wchunk->wc_target.rs_handle =
324 cpu_to_be32(seg->mr_rkey);
325 cur_wchunk->wc_target.rs_length =
326 cpu_to_be32(seg->mr_len);
327 xdr_encode_hyper(
328 (__be32 *)&cur_wchunk->wc_target.rs_offset,
329 seg->mr_base);
330 dprintk("RPC: %s: %s chunk "
331 "elem %d@0x%llx:0x%x (%s)\n", __func__,
332 (type == rpcrdma_replych) ? "reply" : "write",
333 seg->mr_len, (unsigned long long)seg->mr_base,
334 seg->mr_rkey, n < nsegs ? "more" : "last");
335 cur_wchunk++;
336 if (type == rpcrdma_replych)
337 r_xprt->rx_stats.reply_chunk_count++;
338 else
339 r_xprt->rx_stats.write_chunk_count++;
340 r_xprt->rx_stats.total_rdma_request += seg->mr_len;
341 }
342 nchunks++;
343 seg += n;
344 nsegs -= n;
345 } while (nsegs);
346
347 /* success. all failures return above */
348 req->rl_nchunks = nchunks;
349
350 /*
351 * finish off header. If write, marshal discrim and nchunks.
352 */
353 if (cur_rchunk) {
354 iptr = (__be32 *) cur_rchunk;
355 *iptr++ = xdr_zero; /* finish the read chunk list */
356 *iptr++ = xdr_zero; /* encode a NULL write chunk list */
357 *iptr++ = xdr_zero; /* encode a NULL reply chunk */
358 } else {
359 warray->wc_discrim = xdr_one;
360 warray->wc_nchunks = cpu_to_be32(nchunks);
361 iptr = (__be32 *) cur_wchunk;
362 if (type == rpcrdma_writech) {
363 *iptr++ = xdr_zero; /* finish the write chunk list */
364 *iptr++ = xdr_zero; /* encode a NULL reply chunk */
365 }
366 }
367
368 /*
369 * Return header size.
370 */
371 return (unsigned char *)iptr - (unsigned char *)headerp;
372
373out:
374 for (pos = 0; nchunks--;)
375 pos += r_xprt->rx_ia.ri_ops->ro_unmap(r_xprt,
376 &req->rl_segments[pos]);
377 return n;
378}
379
380/*
381 * Copy write data inline.
382 * This function is used for "small" requests. Data which is passed
383 * to RPC via iovecs (or page list) is copied directly into the
384 * pre-registered memory buffer for this request. For small amounts
385 * of data, this is efficient. The cutoff value is tunable.
386 */
387static void rpcrdma_inline_pullup(struct rpc_rqst *rqst)
388{
389 int i, npages, curlen;
390 int copy_len;
391 unsigned char *srcp, *destp;
392 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
393 int page_base;
394 struct page **ppages;
395
396 destp = rqst->rq_svec[0].iov_base;
397 curlen = rqst->rq_svec[0].iov_len;
398 destp += curlen;
399
400 dprintk("RPC: %s: destp 0x%p len %d hdrlen %d\n",
401 __func__, destp, rqst->rq_slen, curlen);
402
403 copy_len = rqst->rq_snd_buf.page_len;
404
405 if (rqst->rq_snd_buf.tail[0].iov_len) {
406 curlen = rqst->rq_snd_buf.tail[0].iov_len;
407 if (destp + copy_len != rqst->rq_snd_buf.tail[0].iov_base) {
408 memmove(destp + copy_len,
409 rqst->rq_snd_buf.tail[0].iov_base, curlen);
410 r_xprt->rx_stats.pullup_copy_count += curlen;
411 }
412 dprintk("RPC: %s: tail destp 0x%p len %d\n",
413 __func__, destp + copy_len, curlen);
414 rqst->rq_svec[0].iov_len += curlen;
415 }
416 r_xprt->rx_stats.pullup_copy_count += copy_len;
417
418 page_base = rqst->rq_snd_buf.page_base;
419 ppages = rqst->rq_snd_buf.pages + (page_base >> PAGE_SHIFT);
420 page_base &= ~PAGE_MASK;
421 npages = PAGE_ALIGN(page_base+copy_len) >> PAGE_SHIFT;
422 for (i = 0; copy_len && i < npages; i++) {
423 curlen = PAGE_SIZE - page_base;
424 if (curlen > copy_len)
425 curlen = copy_len;
426 dprintk("RPC: %s: page %d destp 0x%p len %d curlen %d\n",
427 __func__, i, destp, copy_len, curlen);
428 srcp = kmap_atomic(ppages[i]);
429 memcpy(destp, srcp+page_base, curlen);
430 kunmap_atomic(srcp);
431 rqst->rq_svec[0].iov_len += curlen;
432 destp += curlen;
433 copy_len -= curlen;
434 page_base = 0;
435 }
436 /* header now contains entire send message */
437}
438
439/*
440 * Marshal a request: the primary job of this routine is to choose
441 * the transfer modes. See comments below.
442 *
443 * Uses multiple RDMA IOVs for a request:
444 * [0] -- RPC RDMA header, which uses memory from the *start* of the
445 * preregistered buffer that already holds the RPC data in
446 * its middle.
447 * [1] -- the RPC header/data, marshaled by RPC and the NFS protocol.
448 * [2] -- optional padding.
449 * [3] -- if padded, header only in [1] and data here.
450 *
451 * Returns zero on success, otherwise a negative errno.
452 */
453
454int
455rpcrdma_marshal_req(struct rpc_rqst *rqst)
456{
457 struct rpc_xprt *xprt = rqst->rq_xprt;
458 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
459 struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
460 char *base;
461 size_t rpclen;
462 ssize_t hdrlen;
463 enum rpcrdma_chunktype rtype, wtype;
464 struct rpcrdma_msg *headerp;
465
466#if defined(CONFIG_SUNRPC_BACKCHANNEL)
467 if (test_bit(RPC_BC_PA_IN_USE, &rqst->rq_bc_pa_state))
468 return rpcrdma_bc_marshal_reply(rqst);
469#endif
470
471 /*
472 * rpclen gets amount of data in first buffer, which is the
473 * pre-registered buffer.
474 */
475 base = rqst->rq_svec[0].iov_base;
476 rpclen = rqst->rq_svec[0].iov_len;
477
478 headerp = rdmab_to_msg(req->rl_rdmabuf);
479 /* don't byte-swap XID, it's already done in request */
480 headerp->rm_xid = rqst->rq_xid;
481 headerp->rm_vers = rpcrdma_version;
482 headerp->rm_credit = cpu_to_be32(r_xprt->rx_buf.rb_max_requests);
483 headerp->rm_type = rdma_msg;
484
485 /*
486 * Chunks needed for results?
487 *
488 * o Read ops return data as write chunk(s), header as inline.
489 * o If the expected result is under the inline threshold, all ops
490 * return as inline.
491 * o Large non-read ops return as a single reply chunk.
492 */
493 if (rqst->rq_rcv_buf.flags & XDRBUF_READ)
494 wtype = rpcrdma_writech;
495 else if (rpcrdma_results_inline(rqst))
496 wtype = rpcrdma_noch;
497 else
498 wtype = rpcrdma_replych;
499
500 /*
501 * Chunks needed for arguments?
502 *
503 * o If the total request is under the inline threshold, all ops
504 * are sent as inline.
505 * o Large write ops transmit data as read chunk(s), header as
506 * inline.
507 * o Large non-write ops are sent with the entire message as a
508 * single read chunk (protocol 0-position special case).
509 *
510 * This assumes that the upper layer does not present a request
511 * that both has a data payload, and whose non-data arguments
512 * by themselves are larger than the inline threshold.
513 */
514 if (rpcrdma_args_inline(rqst)) {
515 rtype = rpcrdma_noch;
516 } else if (rqst->rq_snd_buf.flags & XDRBUF_WRITE) {
517 rtype = rpcrdma_readch;
518 } else {
519 r_xprt->rx_stats.nomsg_call_count++;
520 headerp->rm_type = htonl(RDMA_NOMSG);
521 rtype = rpcrdma_areadch;
522 rpclen = 0;
523 }
524
525 /* The following simplification is not true forever */
526 if (rtype != rpcrdma_noch && wtype == rpcrdma_replych)
527 wtype = rpcrdma_noch;
528 if (rtype != rpcrdma_noch && wtype != rpcrdma_noch) {
529 dprintk("RPC: %s: cannot marshal multiple chunk lists\n",
530 __func__);
531 return -EIO;
532 }
533
534 hdrlen = RPCRDMA_HDRLEN_MIN;
535
536 /*
537 * Pull up any extra send data into the preregistered buffer.
538 * When padding is in use and applies to the transfer, insert
539 * it and change the message type.
540 */
541 if (rtype == rpcrdma_noch) {
542
543 rpcrdma_inline_pullup(rqst);
544
545 headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero;
546 headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero;
547 headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero;
548 /* new length after pullup */
549 rpclen = rqst->rq_svec[0].iov_len;
550 } else if (rtype == rpcrdma_readch)
551 rpclen += rpcrdma_tail_pullup(&rqst->rq_snd_buf);
552 if (rtype != rpcrdma_noch) {
553 hdrlen = rpcrdma_create_chunks(rqst, &rqst->rq_snd_buf,
554 headerp, rtype);
555 wtype = rtype; /* simplify dprintk */
556
557 } else if (wtype != rpcrdma_noch) {
558 hdrlen = rpcrdma_create_chunks(rqst, &rqst->rq_rcv_buf,
559 headerp, wtype);
560 }
561 if (hdrlen < 0)
562 return hdrlen;
563
564 dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd"
565 " headerp 0x%p base 0x%p lkey 0x%x\n",
566 __func__, transfertypes[wtype], hdrlen, rpclen,
567 headerp, base, rdmab_lkey(req->rl_rdmabuf));
568
569 /*
570 * initialize send_iov's - normally only two: rdma chunk header and
571 * single preregistered RPC header buffer, but if padding is present,
572 * then use a preregistered (and zeroed) pad buffer between the RPC
573 * header and any write data. In all non-rdma cases, any following
574 * data has been copied into the RPC header buffer.
575 */
576 req->rl_send_iov[0].addr = rdmab_addr(req->rl_rdmabuf);
577 req->rl_send_iov[0].length = hdrlen;
578 req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf);
579
580 req->rl_niovs = 1;
581 if (rtype == rpcrdma_areadch)
582 return 0;
583
584 req->rl_send_iov[1].addr = rdmab_addr(req->rl_sendbuf);
585 req->rl_send_iov[1].length = rpclen;
586 req->rl_send_iov[1].lkey = rdmab_lkey(req->rl_sendbuf);
587
588 req->rl_niovs = 2;
589 return 0;
590}
591
592/*
593 * Chase down a received write or reply chunklist to get length
594 * RDMA'd by server. See map at rpcrdma_create_chunks()! :-)
595 */
596static int
597rpcrdma_count_chunks(struct rpcrdma_rep *rep, unsigned int max, int wrchunk, __be32 **iptrp)
598{
599 unsigned int i, total_len;
600 struct rpcrdma_write_chunk *cur_wchunk;
601 char *base = (char *)rdmab_to_msg(rep->rr_rdmabuf);
602
603 i = be32_to_cpu(**iptrp);
604 if (i > max)
605 return -1;
606 cur_wchunk = (struct rpcrdma_write_chunk *) (*iptrp + 1);
607 total_len = 0;
608 while (i--) {
609 struct rpcrdma_segment *seg = &cur_wchunk->wc_target;
610 ifdebug(FACILITY) {
611 u64 off;
612 xdr_decode_hyper((__be32 *)&seg->rs_offset, &off);
613 dprintk("RPC: %s: chunk %d@0x%llx:0x%x\n",
614 __func__,
615 be32_to_cpu(seg->rs_length),
616 (unsigned long long)off,
617 be32_to_cpu(seg->rs_handle));
618 }
619 total_len += be32_to_cpu(seg->rs_length);
620 ++cur_wchunk;
621 }
622 /* check and adjust for properly terminated write chunk */
623 if (wrchunk) {
624 __be32 *w = (__be32 *) cur_wchunk;
625 if (*w++ != xdr_zero)
626 return -1;
627 cur_wchunk = (struct rpcrdma_write_chunk *) w;
628 }
629 if ((char *)cur_wchunk > base + rep->rr_len)
630 return -1;
631
632 *iptrp = (__be32 *) cur_wchunk;
633 return total_len;
634}
635
636/*
637 * Scatter inline received data back into provided iov's.
638 */
639static void
640rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad)
641{
642 int i, npages, curlen, olen;
643 char *destp;
644 struct page **ppages;
645 int page_base;
646
647 curlen = rqst->rq_rcv_buf.head[0].iov_len;
648 if (curlen > copy_len) { /* write chunk header fixup */
649 curlen = copy_len;
650 rqst->rq_rcv_buf.head[0].iov_len = curlen;
651 }
652
653 dprintk("RPC: %s: srcp 0x%p len %d hdrlen %d\n",
654 __func__, srcp, copy_len, curlen);
655
656 /* Shift pointer for first receive segment only */
657 rqst->rq_rcv_buf.head[0].iov_base = srcp;
658 srcp += curlen;
659 copy_len -= curlen;
660
661 olen = copy_len;
662 i = 0;
663 rpcx_to_rdmax(rqst->rq_xprt)->rx_stats.fixup_copy_count += olen;
664 page_base = rqst->rq_rcv_buf.page_base;
665 ppages = rqst->rq_rcv_buf.pages + (page_base >> PAGE_SHIFT);
666 page_base &= ~PAGE_MASK;
667
668 if (copy_len && rqst->rq_rcv_buf.page_len) {
669 npages = PAGE_ALIGN(page_base +
670 rqst->rq_rcv_buf.page_len) >> PAGE_SHIFT;
671 for (; i < npages; i++) {
672 curlen = PAGE_SIZE - page_base;
673 if (curlen > copy_len)
674 curlen = copy_len;
675 dprintk("RPC: %s: page %d"
676 " srcp 0x%p len %d curlen %d\n",
677 __func__, i, srcp, copy_len, curlen);
678 destp = kmap_atomic(ppages[i]);
679 memcpy(destp + page_base, srcp, curlen);
680 flush_dcache_page(ppages[i]);
681 kunmap_atomic(destp);
682 srcp += curlen;
683 copy_len -= curlen;
684 if (copy_len == 0)
685 break;
686 page_base = 0;
687 }
688 }
689
690 if (copy_len && rqst->rq_rcv_buf.tail[0].iov_len) {
691 curlen = copy_len;
692 if (curlen > rqst->rq_rcv_buf.tail[0].iov_len)
693 curlen = rqst->rq_rcv_buf.tail[0].iov_len;
694 if (rqst->rq_rcv_buf.tail[0].iov_base != srcp)
695 memmove(rqst->rq_rcv_buf.tail[0].iov_base, srcp, curlen);
696 dprintk("RPC: %s: tail srcp 0x%p len %d curlen %d\n",
697 __func__, srcp, copy_len, curlen);
698 rqst->rq_rcv_buf.tail[0].iov_len = curlen;
699 copy_len -= curlen; ++i;
700 } else
701 rqst->rq_rcv_buf.tail[0].iov_len = 0;
702
703 if (pad) {
704 /* implicit padding on terminal chunk */
705 unsigned char *p = rqst->rq_rcv_buf.tail[0].iov_base;
706 while (pad--)
707 p[rqst->rq_rcv_buf.tail[0].iov_len++] = 0;
708 }
709
710 if (copy_len)
711 dprintk("RPC: %s: %d bytes in"
712 " %d extra segments (%d lost)\n",
713 __func__, olen, i, copy_len);
714
715 /* TBD avoid a warning from call_decode() */
716 rqst->rq_private_buf = rqst->rq_rcv_buf;
717}
718
719void
720rpcrdma_connect_worker(struct work_struct *work)
721{
722 struct rpcrdma_ep *ep =
723 container_of(work, struct rpcrdma_ep, rep_connect_worker.work);
724 struct rpcrdma_xprt *r_xprt =
725 container_of(ep, struct rpcrdma_xprt, rx_ep);
726 struct rpc_xprt *xprt = &r_xprt->rx_xprt;
727
728 spin_lock_bh(&xprt->transport_lock);
729 if (++xprt->connect_cookie == 0) /* maintain a reserved value */
730 ++xprt->connect_cookie;
731 if (ep->rep_connected > 0) {
732 if (!xprt_test_and_set_connected(xprt))
733 xprt_wake_pending_tasks(xprt, 0);
734 } else {
735 if (xprt_test_and_clear_connected(xprt))
736 xprt_wake_pending_tasks(xprt, -ENOTCONN);
737 }
738 spin_unlock_bh(&xprt->transport_lock);
739}
740
741#if defined(CONFIG_SUNRPC_BACKCHANNEL)
742/* By convention, backchannel calls arrive via rdma_msg type
743 * messages, and never populate the chunk lists. This makes
744 * the RPC/RDMA header small and fixed in size, so it is
745 * straightforward to check the RPC header's direction field.
746 */
747static bool
748rpcrdma_is_bcall(struct rpcrdma_msg *headerp)
749{
750 __be32 *p = (__be32 *)headerp;
751
752 if (headerp->rm_type != rdma_msg)
753 return false;
754 if (headerp->rm_body.rm_chunks[0] != xdr_zero)
755 return false;
756 if (headerp->rm_body.rm_chunks[1] != xdr_zero)
757 return false;
758 if (headerp->rm_body.rm_chunks[2] != xdr_zero)
759 return false;
760
761 /* sanity */
762 if (p[7] != headerp->rm_xid)
763 return false;
764 /* call direction */
765 if (p[8] != cpu_to_be32(RPC_CALL))
766 return false;
767
768 return true;
769}
770#endif /* CONFIG_SUNRPC_BACKCHANNEL */
771
772/*
773 * This function is called when an async event is posted to
774 * the connection which changes the connection state. All it
775 * does at this point is mark the connection up/down, the rpc
776 * timers do the rest.
777 */
778void
779rpcrdma_conn_func(struct rpcrdma_ep *ep)
780{
781 schedule_delayed_work(&ep->rep_connect_worker, 0);
782}
783
784/* Process received RPC/RDMA messages.
785 *
786 * Errors must result in the RPC task either being awakened, or
787 * allowed to timeout, to discover the errors at that time.
788 */
789void
790rpcrdma_reply_handler(struct rpcrdma_rep *rep)
791{
792 struct rpcrdma_msg *headerp;
793 struct rpcrdma_req *req;
794 struct rpc_rqst *rqst;
795 struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
796 struct rpc_xprt *xprt = &r_xprt->rx_xprt;
797 __be32 *iptr;
798 int rdmalen, status, rmerr;
799 unsigned long cwnd;
800
801 dprintk("RPC: %s: incoming rep %p\n", __func__, rep);
802
803 if (rep->rr_len == RPCRDMA_BAD_LEN)
804 goto out_badstatus;
805 if (rep->rr_len < RPCRDMA_HDRLEN_ERR)
806 goto out_shortreply;
807
808 headerp = rdmab_to_msg(rep->rr_rdmabuf);
809#if defined(CONFIG_SUNRPC_BACKCHANNEL)
810 if (rpcrdma_is_bcall(headerp))
811 goto out_bcall;
812#endif
813
814 /* Match incoming rpcrdma_rep to an rpcrdma_req to
815 * get context for handling any incoming chunks.
816 */
817 spin_lock_bh(&xprt->transport_lock);
818 rqst = xprt_lookup_rqst(xprt, headerp->rm_xid);
819 if (!rqst)
820 goto out_nomatch;
821
822 req = rpcr_to_rdmar(rqst);
823 if (req->rl_reply)
824 goto out_duplicate;
825
826 /* Sanity checking has passed. We are now committed
827 * to complete this transaction.
828 */
829 list_del_init(&rqst->rq_list);
830 spin_unlock_bh(&xprt->transport_lock);
831 dprintk("RPC: %s: reply %p completes request %p (xid 0x%08x)\n",
832 __func__, rep, req, be32_to_cpu(headerp->rm_xid));
833
834 /* from here on, the reply is no longer an orphan */
835 req->rl_reply = rep;
836 xprt->reestablish_timeout = 0;
837
838 if (headerp->rm_vers != rpcrdma_version)
839 goto out_badversion;
840
841 /* check for expected message types */
842 /* The order of some of these tests is important. */
843 switch (headerp->rm_type) {
844 case rdma_msg:
845 /* never expect read chunks */
846 /* never expect reply chunks (two ways to check) */
847 /* never expect write chunks without having offered RDMA */
848 if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
849 (headerp->rm_body.rm_chunks[1] == xdr_zero &&
850 headerp->rm_body.rm_chunks[2] != xdr_zero) ||
851 (headerp->rm_body.rm_chunks[1] != xdr_zero &&
852 req->rl_nchunks == 0))
853 goto badheader;
854 if (headerp->rm_body.rm_chunks[1] != xdr_zero) {
855 /* count any expected write chunks in read reply */
856 /* start at write chunk array count */
857 iptr = &headerp->rm_body.rm_chunks[2];
858 rdmalen = rpcrdma_count_chunks(rep,
859 req->rl_nchunks, 1, &iptr);
860 /* check for validity, and no reply chunk after */
861 if (rdmalen < 0 || *iptr++ != xdr_zero)
862 goto badheader;
863 rep->rr_len -=
864 ((unsigned char *)iptr - (unsigned char *)headerp);
865 status = rep->rr_len + rdmalen;
866 r_xprt->rx_stats.total_rdma_reply += rdmalen;
867 /* special case - last chunk may omit padding */
868 if (rdmalen &= 3) {
869 rdmalen = 4 - rdmalen;
870 status += rdmalen;
871 }
872 } else {
873 /* else ordinary inline */
874 rdmalen = 0;
875 iptr = (__be32 *)((unsigned char *)headerp +
876 RPCRDMA_HDRLEN_MIN);
877 rep->rr_len -= RPCRDMA_HDRLEN_MIN;
878 status = rep->rr_len;
879 }
880 /* Fix up the rpc results for upper layer */
881 rpcrdma_inline_fixup(rqst, (char *)iptr, rep->rr_len, rdmalen);
882 break;
883
884 case rdma_nomsg:
885 /* never expect read or write chunks, always reply chunks */
886 if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
887 headerp->rm_body.rm_chunks[1] != xdr_zero ||
888 headerp->rm_body.rm_chunks[2] != xdr_one ||
889 req->rl_nchunks == 0)
890 goto badheader;
891 iptr = (__be32 *)((unsigned char *)headerp +
892 RPCRDMA_HDRLEN_MIN);
893 rdmalen = rpcrdma_count_chunks(rep, req->rl_nchunks, 0, &iptr);
894 if (rdmalen < 0)
895 goto badheader;
896 r_xprt->rx_stats.total_rdma_reply += rdmalen;
897 /* Reply chunk buffer already is the reply vector - no fixup. */
898 status = rdmalen;
899 break;
900
901 case rdma_error:
902 goto out_rdmaerr;
903
904badheader:
905 default:
906 dprintk("%s: invalid rpcrdma reply header (type %d):"
907 " chunks[012] == %d %d %d"
908 " expected chunks <= %d\n",
909 __func__, be32_to_cpu(headerp->rm_type),
910 headerp->rm_body.rm_chunks[0],
911 headerp->rm_body.rm_chunks[1],
912 headerp->rm_body.rm_chunks[2],
913 req->rl_nchunks);
914 status = -EIO;
915 r_xprt->rx_stats.bad_reply_count++;
916 break;
917 }
918
919out:
920 /* Invalidate and flush the data payloads before waking the
921 * waiting application. This guarantees the memory region is
922 * properly fenced from the server before the application
923 * accesses the data. It also ensures proper send flow
924 * control: waking the next RPC waits until this RPC has
925 * relinquished all its Send Queue entries.
926 */
927 if (req->rl_nchunks)
928 r_xprt->rx_ia.ri_ops->ro_unmap_sync(r_xprt, req);
929
930 spin_lock_bh(&xprt->transport_lock);
931 cwnd = xprt->cwnd;
932 xprt->cwnd = atomic_read(&r_xprt->rx_buf.rb_credits) << RPC_CWNDSHIFT;
933 if (xprt->cwnd > cwnd)
934 xprt_release_rqst_cong(rqst->rq_task);
935
936 xprt_complete_rqst(rqst->rq_task, status);
937 spin_unlock_bh(&xprt->transport_lock);
938 dprintk("RPC: %s: xprt_complete_rqst(0x%p, 0x%p, %d)\n",
939 __func__, xprt, rqst, status);
940 return;
941
942out_badstatus:
943 rpcrdma_recv_buffer_put(rep);
944 if (r_xprt->rx_ep.rep_connected == 1) {
945 r_xprt->rx_ep.rep_connected = -EIO;
946 rpcrdma_conn_func(&r_xprt->rx_ep);
947 }
948 return;
949
950#if defined(CONFIG_SUNRPC_BACKCHANNEL)
951out_bcall:
952 rpcrdma_bc_receive_call(r_xprt, rep);
953 return;
954#endif
955
956/* If the incoming reply terminated a pending RPC, the next
957 * RPC call will post a replacement receive buffer as it is
958 * being marshaled.
959 */
960out_badversion:
961 dprintk("RPC: %s: invalid version %d\n",
962 __func__, be32_to_cpu(headerp->rm_vers));
963 status = -EIO;
964 r_xprt->rx_stats.bad_reply_count++;
965 goto out;
966
967out_rdmaerr:
968 rmerr = be32_to_cpu(headerp->rm_body.rm_error.rm_err);
969 switch (rmerr) {
970 case ERR_VERS:
971 pr_err("%s: server reports header version error (%u-%u)\n",
972 __func__,
973 be32_to_cpu(headerp->rm_body.rm_error.rm_vers_low),
974 be32_to_cpu(headerp->rm_body.rm_error.rm_vers_high));
975 break;
976 case ERR_CHUNK:
977 pr_err("%s: server reports header decoding error\n",
978 __func__);
979 break;
980 default:
981 pr_err("%s: server reports unknown error %d\n",
982 __func__, rmerr);
983 }
984 status = -EREMOTEIO;
985 r_xprt->rx_stats.bad_reply_count++;
986 goto out;
987
988/* If no pending RPC transaction was matched, post a replacement
989 * receive buffer before returning.
990 */
991out_shortreply:
992 dprintk("RPC: %s: short/invalid reply\n", __func__);
993 goto repost;
994
995out_nomatch:
996 spin_unlock_bh(&xprt->transport_lock);
997 dprintk("RPC: %s: no match for incoming xid 0x%08x len %d\n",
998 __func__, be32_to_cpu(headerp->rm_xid),
999 rep->rr_len);
1000 goto repost;
1001
1002out_duplicate:
1003 spin_unlock_bh(&xprt->transport_lock);
1004 dprintk("RPC: %s: "
1005 "duplicate reply %p to RPC request %p: xid 0x%08x\n",
1006 __func__, rep, req, be32_to_cpu(headerp->rm_xid));
1007
1008repost:
1009 r_xprt->rx_stats.bad_reply_count++;
1010 if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, &r_xprt->rx_ep, rep))
1011 rpcrdma_recv_buffer_put(rep);
1012}
1/*
2 * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the BSD-type
8 * license below:
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 *
14 * Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 *
17 * Redistributions in binary form must reproduce the above
18 * copyright notice, this list of conditions and the following
19 * disclaimer in the documentation and/or other materials provided
20 * with the distribution.
21 *
22 * Neither the name of the Network Appliance, Inc. nor the names of
23 * its contributors may be used to endorse or promote products
24 * derived from this software without specific prior written
25 * permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
28 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
29 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
30 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
31 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
32 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
33 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
34 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
35 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
36 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
37 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
38 */
39
40/*
41 * rpc_rdma.c
42 *
43 * This file contains the guts of the RPC RDMA protocol, and
44 * does marshaling/unmarshaling, etc. It is also where interfacing
45 * to the Linux RPC framework lives.
46 */
47
48#include "xprt_rdma.h"
49
50#include <linux/highmem.h>
51
52#ifdef RPC_DEBUG
53# define RPCDBG_FACILITY RPCDBG_TRANS
54#endif
55
56enum rpcrdma_chunktype {
57 rpcrdma_noch = 0,
58 rpcrdma_readch,
59 rpcrdma_areadch,
60 rpcrdma_writech,
61 rpcrdma_replych
62};
63
64#ifdef RPC_DEBUG
65static const char transfertypes[][12] = {
66 "pure inline", /* no chunks */
67 " read chunk", /* some argument via rdma read */
68 "*read chunk", /* entire request via rdma read */
69 "write chunk", /* some result via rdma write */
70 "reply chunk" /* entire reply via rdma write */
71};
72#endif
73
74/*
75 * Chunk assembly from upper layer xdr_buf.
76 *
77 * Prepare the passed-in xdr_buf into representation as RPC/RDMA chunk
78 * elements. Segments are then coalesced when registered, if possible
79 * within the selected memreg mode.
80 *
81 * Note, this routine is never called if the connection's memory
82 * registration strategy is 0 (bounce buffers).
83 */
84
85static int
86rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos,
87 enum rpcrdma_chunktype type, struct rpcrdma_mr_seg *seg, int nsegs)
88{
89 int len, n = 0, p;
90 int page_base;
91 struct page **ppages;
92
93 if (pos == 0 && xdrbuf->head[0].iov_len) {
94 seg[n].mr_page = NULL;
95 seg[n].mr_offset = xdrbuf->head[0].iov_base;
96 seg[n].mr_len = xdrbuf->head[0].iov_len;
97 ++n;
98 }
99
100 len = xdrbuf->page_len;
101 ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT);
102 page_base = xdrbuf->page_base & ~PAGE_MASK;
103 p = 0;
104 while (len && n < nsegs) {
105 seg[n].mr_page = ppages[p];
106 seg[n].mr_offset = (void *)(unsigned long) page_base;
107 seg[n].mr_len = min_t(u32, PAGE_SIZE - page_base, len);
108 BUG_ON(seg[n].mr_len > PAGE_SIZE);
109 len -= seg[n].mr_len;
110 ++n;
111 ++p;
112 page_base = 0; /* page offset only applies to first page */
113 }
114
115 /* Message overflows the seg array */
116 if (len && n == nsegs)
117 return 0;
118
119 if (xdrbuf->tail[0].iov_len) {
120 /* the rpcrdma protocol allows us to omit any trailing
121 * xdr pad bytes, saving the server an RDMA operation. */
122 if (xdrbuf->tail[0].iov_len < 4 && xprt_rdma_pad_optimize)
123 return n;
124 if (n == nsegs)
125 /* Tail remains, but we're out of segments */
126 return 0;
127 seg[n].mr_page = NULL;
128 seg[n].mr_offset = xdrbuf->tail[0].iov_base;
129 seg[n].mr_len = xdrbuf->tail[0].iov_len;
130 ++n;
131 }
132
133 return n;
134}
135
136/*
137 * Create read/write chunk lists, and reply chunks, for RDMA
138 *
139 * Assume check against THRESHOLD has been done, and chunks are required.
140 * Assume only encoding one list entry for read|write chunks. The NFSv3
141 * protocol is simple enough to allow this as it only has a single "bulk
142 * result" in each procedure - complicated NFSv4 COMPOUNDs are not. (The
143 * RDMA/Sessions NFSv4 proposal addresses this for future v4 revs.)
144 *
145 * When used for a single reply chunk (which is a special write
146 * chunk used for the entire reply, rather than just the data), it
147 * is used primarily for READDIR and READLINK which would otherwise
148 * be severely size-limited by a small rdma inline read max. The server
149 * response will come back as an RDMA Write, followed by a message
150 * of type RDMA_NOMSG carrying the xid and length. As a result, reply
151 * chunks do not provide data alignment, however they do not require
152 * "fixup" (moving the response to the upper layer buffer) either.
153 *
154 * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64):
155 *
156 * Read chunklist (a linked list):
157 * N elements, position P (same P for all chunks of same arg!):
158 * 1 - PHLOO - 1 - PHLOO - ... - 1 - PHLOO - 0
159 *
160 * Write chunklist (a list of (one) counted array):
161 * N elements:
162 * 1 - N - HLOO - HLOO - ... - HLOO - 0
163 *
164 * Reply chunk (a counted array):
165 * N elements:
166 * 1 - N - HLOO - HLOO - ... - HLOO
167 */
168
169static unsigned int
170rpcrdma_create_chunks(struct rpc_rqst *rqst, struct xdr_buf *target,
171 struct rpcrdma_msg *headerp, enum rpcrdma_chunktype type)
172{
173 struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
174 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_task->tk_xprt);
175 int nsegs, nchunks = 0;
176 unsigned int pos;
177 struct rpcrdma_mr_seg *seg = req->rl_segments;
178 struct rpcrdma_read_chunk *cur_rchunk = NULL;
179 struct rpcrdma_write_array *warray = NULL;
180 struct rpcrdma_write_chunk *cur_wchunk = NULL;
181 __be32 *iptr = headerp->rm_body.rm_chunks;
182
183 if (type == rpcrdma_readch || type == rpcrdma_areadch) {
184 /* a read chunk - server will RDMA Read our memory */
185 cur_rchunk = (struct rpcrdma_read_chunk *) iptr;
186 } else {
187 /* a write or reply chunk - server will RDMA Write our memory */
188 *iptr++ = xdr_zero; /* encode a NULL read chunk list */
189 if (type == rpcrdma_replych)
190 *iptr++ = xdr_zero; /* a NULL write chunk list */
191 warray = (struct rpcrdma_write_array *) iptr;
192 cur_wchunk = (struct rpcrdma_write_chunk *) (warray + 1);
193 }
194
195 if (type == rpcrdma_replych || type == rpcrdma_areadch)
196 pos = 0;
197 else
198 pos = target->head[0].iov_len;
199
200 nsegs = rpcrdma_convert_iovs(target, pos, type, seg, RPCRDMA_MAX_SEGS);
201 if (nsegs == 0)
202 return 0;
203
204 do {
205 /* bind/register the memory, then build chunk from result. */
206 int n = rpcrdma_register_external(seg, nsegs,
207 cur_wchunk != NULL, r_xprt);
208 if (n <= 0)
209 goto out;
210 if (cur_rchunk) { /* read */
211 cur_rchunk->rc_discrim = xdr_one;
212 /* all read chunks have the same "position" */
213 cur_rchunk->rc_position = htonl(pos);
214 cur_rchunk->rc_target.rs_handle = htonl(seg->mr_rkey);
215 cur_rchunk->rc_target.rs_length = htonl(seg->mr_len);
216 xdr_encode_hyper(
217 (__be32 *)&cur_rchunk->rc_target.rs_offset,
218 seg->mr_base);
219 dprintk("RPC: %s: read chunk "
220 "elem %d@0x%llx:0x%x pos %u (%s)\n", __func__,
221 seg->mr_len, (unsigned long long)seg->mr_base,
222 seg->mr_rkey, pos, n < nsegs ? "more" : "last");
223 cur_rchunk++;
224 r_xprt->rx_stats.read_chunk_count++;
225 } else { /* write/reply */
226 cur_wchunk->wc_target.rs_handle = htonl(seg->mr_rkey);
227 cur_wchunk->wc_target.rs_length = htonl(seg->mr_len);
228 xdr_encode_hyper(
229 (__be32 *)&cur_wchunk->wc_target.rs_offset,
230 seg->mr_base);
231 dprintk("RPC: %s: %s chunk "
232 "elem %d@0x%llx:0x%x (%s)\n", __func__,
233 (type == rpcrdma_replych) ? "reply" : "write",
234 seg->mr_len, (unsigned long long)seg->mr_base,
235 seg->mr_rkey, n < nsegs ? "more" : "last");
236 cur_wchunk++;
237 if (type == rpcrdma_replych)
238 r_xprt->rx_stats.reply_chunk_count++;
239 else
240 r_xprt->rx_stats.write_chunk_count++;
241 r_xprt->rx_stats.total_rdma_request += seg->mr_len;
242 }
243 nchunks++;
244 seg += n;
245 nsegs -= n;
246 } while (nsegs);
247
248 /* success. all failures return above */
249 req->rl_nchunks = nchunks;
250
251 BUG_ON(nchunks == 0);
252 BUG_ON((r_xprt->rx_ia.ri_memreg_strategy == RPCRDMA_FRMR)
253 && (nchunks > 3));
254
255 /*
256 * finish off header. If write, marshal discrim and nchunks.
257 */
258 if (cur_rchunk) {
259 iptr = (__be32 *) cur_rchunk;
260 *iptr++ = xdr_zero; /* finish the read chunk list */
261 *iptr++ = xdr_zero; /* encode a NULL write chunk list */
262 *iptr++ = xdr_zero; /* encode a NULL reply chunk */
263 } else {
264 warray->wc_discrim = xdr_one;
265 warray->wc_nchunks = htonl(nchunks);
266 iptr = (__be32 *) cur_wchunk;
267 if (type == rpcrdma_writech) {
268 *iptr++ = xdr_zero; /* finish the write chunk list */
269 *iptr++ = xdr_zero; /* encode a NULL reply chunk */
270 }
271 }
272
273 /*
274 * Return header size.
275 */
276 return (unsigned char *)iptr - (unsigned char *)headerp;
277
278out:
279 for (pos = 0; nchunks--;)
280 pos += rpcrdma_deregister_external(
281 &req->rl_segments[pos], r_xprt, NULL);
282 return 0;
283}
284
285/*
286 * Copy write data inline.
287 * This function is used for "small" requests. Data which is passed
288 * to RPC via iovecs (or page list) is copied directly into the
289 * pre-registered memory buffer for this request. For small amounts
290 * of data, this is efficient. The cutoff value is tunable.
291 */
292static int
293rpcrdma_inline_pullup(struct rpc_rqst *rqst, int pad)
294{
295 int i, npages, curlen;
296 int copy_len;
297 unsigned char *srcp, *destp;
298 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
299 int page_base;
300 struct page **ppages;
301
302 destp = rqst->rq_svec[0].iov_base;
303 curlen = rqst->rq_svec[0].iov_len;
304 destp += curlen;
305 /*
306 * Do optional padding where it makes sense. Alignment of write
307 * payload can help the server, if our setting is accurate.
308 */
309 pad -= (curlen + 36/*sizeof(struct rpcrdma_msg_padded)*/);
310 if (pad < 0 || rqst->rq_slen - curlen < RPCRDMA_INLINE_PAD_THRESH)
311 pad = 0; /* don't pad this request */
312
313 dprintk("RPC: %s: pad %d destp 0x%p len %d hdrlen %d\n",
314 __func__, pad, destp, rqst->rq_slen, curlen);
315
316 copy_len = rqst->rq_snd_buf.page_len;
317
318 if (rqst->rq_snd_buf.tail[0].iov_len) {
319 curlen = rqst->rq_snd_buf.tail[0].iov_len;
320 if (destp + copy_len != rqst->rq_snd_buf.tail[0].iov_base) {
321 memmove(destp + copy_len,
322 rqst->rq_snd_buf.tail[0].iov_base, curlen);
323 r_xprt->rx_stats.pullup_copy_count += curlen;
324 }
325 dprintk("RPC: %s: tail destp 0x%p len %d\n",
326 __func__, destp + copy_len, curlen);
327 rqst->rq_svec[0].iov_len += curlen;
328 }
329 r_xprt->rx_stats.pullup_copy_count += copy_len;
330
331 page_base = rqst->rq_snd_buf.page_base;
332 ppages = rqst->rq_snd_buf.pages + (page_base >> PAGE_SHIFT);
333 page_base &= ~PAGE_MASK;
334 npages = PAGE_ALIGN(page_base+copy_len) >> PAGE_SHIFT;
335 for (i = 0; copy_len && i < npages; i++) {
336 curlen = PAGE_SIZE - page_base;
337 if (curlen > copy_len)
338 curlen = copy_len;
339 dprintk("RPC: %s: page %d destp 0x%p len %d curlen %d\n",
340 __func__, i, destp, copy_len, curlen);
341 srcp = kmap_atomic(ppages[i], KM_SKB_SUNRPC_DATA);
342 memcpy(destp, srcp+page_base, curlen);
343 kunmap_atomic(srcp, KM_SKB_SUNRPC_DATA);
344 rqst->rq_svec[0].iov_len += curlen;
345 destp += curlen;
346 copy_len -= curlen;
347 page_base = 0;
348 }
349 /* header now contains entire send message */
350 return pad;
351}
352
353/*
354 * Marshal a request: the primary job of this routine is to choose
355 * the transfer modes. See comments below.
356 *
357 * Uses multiple RDMA IOVs for a request:
358 * [0] -- RPC RDMA header, which uses memory from the *start* of the
359 * preregistered buffer that already holds the RPC data in
360 * its middle.
361 * [1] -- the RPC header/data, marshaled by RPC and the NFS protocol.
362 * [2] -- optional padding.
363 * [3] -- if padded, header only in [1] and data here.
364 */
365
366int
367rpcrdma_marshal_req(struct rpc_rqst *rqst)
368{
369 struct rpc_xprt *xprt = rqst->rq_task->tk_xprt;
370 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
371 struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
372 char *base;
373 size_t hdrlen, rpclen, padlen;
374 enum rpcrdma_chunktype rtype, wtype;
375 struct rpcrdma_msg *headerp;
376
377 /*
378 * rpclen gets amount of data in first buffer, which is the
379 * pre-registered buffer.
380 */
381 base = rqst->rq_svec[0].iov_base;
382 rpclen = rqst->rq_svec[0].iov_len;
383
384 /* build RDMA header in private area at front */
385 headerp = (struct rpcrdma_msg *) req->rl_base;
386 /* don't htonl XID, it's already done in request */
387 headerp->rm_xid = rqst->rq_xid;
388 headerp->rm_vers = xdr_one;
389 headerp->rm_credit = htonl(r_xprt->rx_buf.rb_max_requests);
390 headerp->rm_type = htonl(RDMA_MSG);
391
392 /*
393 * Chunks needed for results?
394 *
395 * o If the expected result is under the inline threshold, all ops
396 * return as inline (but see later).
397 * o Large non-read ops return as a single reply chunk.
398 * o Large read ops return data as write chunk(s), header as inline.
399 *
400 * Note: the NFS code sending down multiple result segments implies
401 * the op is one of read, readdir[plus], readlink or NFSv4 getacl.
402 */
403
404 /*
405 * This code can handle read chunks, write chunks OR reply
406 * chunks -- only one type. If the request is too big to fit
407 * inline, then we will choose read chunks. If the request is
408 * a READ, then use write chunks to separate the file data
409 * into pages; otherwise use reply chunks.
410 */
411 if (rqst->rq_rcv_buf.buflen <= RPCRDMA_INLINE_READ_THRESHOLD(rqst))
412 wtype = rpcrdma_noch;
413 else if (rqst->rq_rcv_buf.page_len == 0)
414 wtype = rpcrdma_replych;
415 else if (rqst->rq_rcv_buf.flags & XDRBUF_READ)
416 wtype = rpcrdma_writech;
417 else
418 wtype = rpcrdma_replych;
419
420 /*
421 * Chunks needed for arguments?
422 *
423 * o If the total request is under the inline threshold, all ops
424 * are sent as inline.
425 * o Large non-write ops are sent with the entire message as a
426 * single read chunk (protocol 0-position special case).
427 * o Large write ops transmit data as read chunk(s), header as
428 * inline.
429 *
430 * Note: the NFS code sending down multiple argument segments
431 * implies the op is a write.
432 * TBD check NFSv4 setacl
433 */
434 if (rqst->rq_snd_buf.len <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst))
435 rtype = rpcrdma_noch;
436 else if (rqst->rq_snd_buf.page_len == 0)
437 rtype = rpcrdma_areadch;
438 else
439 rtype = rpcrdma_readch;
440
441 /* The following simplification is not true forever */
442 if (rtype != rpcrdma_noch && wtype == rpcrdma_replych)
443 wtype = rpcrdma_noch;
444 BUG_ON(rtype != rpcrdma_noch && wtype != rpcrdma_noch);
445
446 if (r_xprt->rx_ia.ri_memreg_strategy == RPCRDMA_BOUNCEBUFFERS &&
447 (rtype != rpcrdma_noch || wtype != rpcrdma_noch)) {
448 /* forced to "pure inline"? */
449 dprintk("RPC: %s: too much data (%d/%d) for inline\n",
450 __func__, rqst->rq_rcv_buf.len, rqst->rq_snd_buf.len);
451 return -1;
452 }
453
454 hdrlen = 28; /*sizeof *headerp;*/
455 padlen = 0;
456
457 /*
458 * Pull up any extra send data into the preregistered buffer.
459 * When padding is in use and applies to the transfer, insert
460 * it and change the message type.
461 */
462 if (rtype == rpcrdma_noch) {
463
464 padlen = rpcrdma_inline_pullup(rqst,
465 RPCRDMA_INLINE_PAD_VALUE(rqst));
466
467 if (padlen) {
468 headerp->rm_type = htonl(RDMA_MSGP);
469 headerp->rm_body.rm_padded.rm_align =
470 htonl(RPCRDMA_INLINE_PAD_VALUE(rqst));
471 headerp->rm_body.rm_padded.rm_thresh =
472 htonl(RPCRDMA_INLINE_PAD_THRESH);
473 headerp->rm_body.rm_padded.rm_pempty[0] = xdr_zero;
474 headerp->rm_body.rm_padded.rm_pempty[1] = xdr_zero;
475 headerp->rm_body.rm_padded.rm_pempty[2] = xdr_zero;
476 hdrlen += 2 * sizeof(u32); /* extra words in padhdr */
477 BUG_ON(wtype != rpcrdma_noch);
478
479 } else {
480 headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero;
481 headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero;
482 headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero;
483 /* new length after pullup */
484 rpclen = rqst->rq_svec[0].iov_len;
485 /*
486 * Currently we try to not actually use read inline.
487 * Reply chunks have the desirable property that
488 * they land, packed, directly in the target buffers
489 * without headers, so they require no fixup. The
490 * additional RDMA Write op sends the same amount
491 * of data, streams on-the-wire and adds no overhead
492 * on receive. Therefore, we request a reply chunk
493 * for non-writes wherever feasible and efficient.
494 */
495 if (wtype == rpcrdma_noch &&
496 r_xprt->rx_ia.ri_memreg_strategy > RPCRDMA_REGISTER)
497 wtype = rpcrdma_replych;
498 }
499 }
500
501 /*
502 * Marshal chunks. This routine will return the header length
503 * consumed by marshaling.
504 */
505 if (rtype != rpcrdma_noch) {
506 hdrlen = rpcrdma_create_chunks(rqst,
507 &rqst->rq_snd_buf, headerp, rtype);
508 wtype = rtype; /* simplify dprintk */
509
510 } else if (wtype != rpcrdma_noch) {
511 hdrlen = rpcrdma_create_chunks(rqst,
512 &rqst->rq_rcv_buf, headerp, wtype);
513 }
514
515 if (hdrlen == 0)
516 return -1;
517
518 dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd padlen %zd"
519 " headerp 0x%p base 0x%p lkey 0x%x\n",
520 __func__, transfertypes[wtype], hdrlen, rpclen, padlen,
521 headerp, base, req->rl_iov.lkey);
522
523 /*
524 * initialize send_iov's - normally only two: rdma chunk header and
525 * single preregistered RPC header buffer, but if padding is present,
526 * then use a preregistered (and zeroed) pad buffer between the RPC
527 * header and any write data. In all non-rdma cases, any following
528 * data has been copied into the RPC header buffer.
529 */
530 req->rl_send_iov[0].addr = req->rl_iov.addr;
531 req->rl_send_iov[0].length = hdrlen;
532 req->rl_send_iov[0].lkey = req->rl_iov.lkey;
533
534 req->rl_send_iov[1].addr = req->rl_iov.addr + (base - req->rl_base);
535 req->rl_send_iov[1].length = rpclen;
536 req->rl_send_iov[1].lkey = req->rl_iov.lkey;
537
538 req->rl_niovs = 2;
539
540 if (padlen) {
541 struct rpcrdma_ep *ep = &r_xprt->rx_ep;
542
543 req->rl_send_iov[2].addr = ep->rep_pad.addr;
544 req->rl_send_iov[2].length = padlen;
545 req->rl_send_iov[2].lkey = ep->rep_pad.lkey;
546
547 req->rl_send_iov[3].addr = req->rl_send_iov[1].addr + rpclen;
548 req->rl_send_iov[3].length = rqst->rq_slen - rpclen;
549 req->rl_send_iov[3].lkey = req->rl_iov.lkey;
550
551 req->rl_niovs = 4;
552 }
553
554 return 0;
555}
556
557/*
558 * Chase down a received write or reply chunklist to get length
559 * RDMA'd by server. See map at rpcrdma_create_chunks()! :-)
560 */
561static int
562rpcrdma_count_chunks(struct rpcrdma_rep *rep, unsigned int max, int wrchunk, __be32 **iptrp)
563{
564 unsigned int i, total_len;
565 struct rpcrdma_write_chunk *cur_wchunk;
566
567 i = ntohl(**iptrp); /* get array count */
568 if (i > max)
569 return -1;
570 cur_wchunk = (struct rpcrdma_write_chunk *) (*iptrp + 1);
571 total_len = 0;
572 while (i--) {
573 struct rpcrdma_segment *seg = &cur_wchunk->wc_target;
574 ifdebug(FACILITY) {
575 u64 off;
576 xdr_decode_hyper((__be32 *)&seg->rs_offset, &off);
577 dprintk("RPC: %s: chunk %d@0x%llx:0x%x\n",
578 __func__,
579 ntohl(seg->rs_length),
580 (unsigned long long)off,
581 ntohl(seg->rs_handle));
582 }
583 total_len += ntohl(seg->rs_length);
584 ++cur_wchunk;
585 }
586 /* check and adjust for properly terminated write chunk */
587 if (wrchunk) {
588 __be32 *w = (__be32 *) cur_wchunk;
589 if (*w++ != xdr_zero)
590 return -1;
591 cur_wchunk = (struct rpcrdma_write_chunk *) w;
592 }
593 if ((char *) cur_wchunk > rep->rr_base + rep->rr_len)
594 return -1;
595
596 *iptrp = (__be32 *) cur_wchunk;
597 return total_len;
598}
599
600/*
601 * Scatter inline received data back into provided iov's.
602 */
603static void
604rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad)
605{
606 int i, npages, curlen, olen;
607 char *destp;
608 struct page **ppages;
609 int page_base;
610
611 curlen = rqst->rq_rcv_buf.head[0].iov_len;
612 if (curlen > copy_len) { /* write chunk header fixup */
613 curlen = copy_len;
614 rqst->rq_rcv_buf.head[0].iov_len = curlen;
615 }
616
617 dprintk("RPC: %s: srcp 0x%p len %d hdrlen %d\n",
618 __func__, srcp, copy_len, curlen);
619
620 /* Shift pointer for first receive segment only */
621 rqst->rq_rcv_buf.head[0].iov_base = srcp;
622 srcp += curlen;
623 copy_len -= curlen;
624
625 olen = copy_len;
626 i = 0;
627 rpcx_to_rdmax(rqst->rq_xprt)->rx_stats.fixup_copy_count += olen;
628 page_base = rqst->rq_rcv_buf.page_base;
629 ppages = rqst->rq_rcv_buf.pages + (page_base >> PAGE_SHIFT);
630 page_base &= ~PAGE_MASK;
631
632 if (copy_len && rqst->rq_rcv_buf.page_len) {
633 npages = PAGE_ALIGN(page_base +
634 rqst->rq_rcv_buf.page_len) >> PAGE_SHIFT;
635 for (; i < npages; i++) {
636 curlen = PAGE_SIZE - page_base;
637 if (curlen > copy_len)
638 curlen = copy_len;
639 dprintk("RPC: %s: page %d"
640 " srcp 0x%p len %d curlen %d\n",
641 __func__, i, srcp, copy_len, curlen);
642 destp = kmap_atomic(ppages[i], KM_SKB_SUNRPC_DATA);
643 memcpy(destp + page_base, srcp, curlen);
644 flush_dcache_page(ppages[i]);
645 kunmap_atomic(destp, KM_SKB_SUNRPC_DATA);
646 srcp += curlen;
647 copy_len -= curlen;
648 if (copy_len == 0)
649 break;
650 page_base = 0;
651 }
652 rqst->rq_rcv_buf.page_len = olen - copy_len;
653 } else
654 rqst->rq_rcv_buf.page_len = 0;
655
656 if (copy_len && rqst->rq_rcv_buf.tail[0].iov_len) {
657 curlen = copy_len;
658 if (curlen > rqst->rq_rcv_buf.tail[0].iov_len)
659 curlen = rqst->rq_rcv_buf.tail[0].iov_len;
660 if (rqst->rq_rcv_buf.tail[0].iov_base != srcp)
661 memmove(rqst->rq_rcv_buf.tail[0].iov_base, srcp, curlen);
662 dprintk("RPC: %s: tail srcp 0x%p len %d curlen %d\n",
663 __func__, srcp, copy_len, curlen);
664 rqst->rq_rcv_buf.tail[0].iov_len = curlen;
665 copy_len -= curlen; ++i;
666 } else
667 rqst->rq_rcv_buf.tail[0].iov_len = 0;
668
669 if (pad) {
670 /* implicit padding on terminal chunk */
671 unsigned char *p = rqst->rq_rcv_buf.tail[0].iov_base;
672 while (pad--)
673 p[rqst->rq_rcv_buf.tail[0].iov_len++] = 0;
674 }
675
676 if (copy_len)
677 dprintk("RPC: %s: %d bytes in"
678 " %d extra segments (%d lost)\n",
679 __func__, olen, i, copy_len);
680
681 /* TBD avoid a warning from call_decode() */
682 rqst->rq_private_buf = rqst->rq_rcv_buf;
683}
684
685/*
686 * This function is called when an async event is posted to
687 * the connection which changes the connection state. All it
688 * does at this point is mark the connection up/down, the rpc
689 * timers do the rest.
690 */
691void
692rpcrdma_conn_func(struct rpcrdma_ep *ep)
693{
694 struct rpc_xprt *xprt = ep->rep_xprt;
695
696 spin_lock_bh(&xprt->transport_lock);
697 if (++xprt->connect_cookie == 0) /* maintain a reserved value */
698 ++xprt->connect_cookie;
699 if (ep->rep_connected > 0) {
700 if (!xprt_test_and_set_connected(xprt))
701 xprt_wake_pending_tasks(xprt, 0);
702 } else {
703 if (xprt_test_and_clear_connected(xprt))
704 xprt_wake_pending_tasks(xprt, -ENOTCONN);
705 }
706 spin_unlock_bh(&xprt->transport_lock);
707}
708
709/*
710 * This function is called when memory window unbind which we are waiting
711 * for completes. Just use rr_func (zeroed by upcall) to signal completion.
712 */
713static void
714rpcrdma_unbind_func(struct rpcrdma_rep *rep)
715{
716 wake_up(&rep->rr_unbind);
717}
718
719/*
720 * Called as a tasklet to do req/reply match and complete a request
721 * Errors must result in the RPC task either being awakened, or
722 * allowed to timeout, to discover the errors at that time.
723 */
724void
725rpcrdma_reply_handler(struct rpcrdma_rep *rep)
726{
727 struct rpcrdma_msg *headerp;
728 struct rpcrdma_req *req;
729 struct rpc_rqst *rqst;
730 struct rpc_xprt *xprt = rep->rr_xprt;
731 struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
732 __be32 *iptr;
733 int i, rdmalen, status;
734
735 /* Check status. If bad, signal disconnect and return rep to pool */
736 if (rep->rr_len == ~0U) {
737 rpcrdma_recv_buffer_put(rep);
738 if (r_xprt->rx_ep.rep_connected == 1) {
739 r_xprt->rx_ep.rep_connected = -EIO;
740 rpcrdma_conn_func(&r_xprt->rx_ep);
741 }
742 return;
743 }
744 if (rep->rr_len < 28) {
745 dprintk("RPC: %s: short/invalid reply\n", __func__);
746 goto repost;
747 }
748 headerp = (struct rpcrdma_msg *) rep->rr_base;
749 if (headerp->rm_vers != xdr_one) {
750 dprintk("RPC: %s: invalid version %d\n",
751 __func__, ntohl(headerp->rm_vers));
752 goto repost;
753 }
754
755 /* Get XID and try for a match. */
756 spin_lock(&xprt->transport_lock);
757 rqst = xprt_lookup_rqst(xprt, headerp->rm_xid);
758 if (rqst == NULL) {
759 spin_unlock(&xprt->transport_lock);
760 dprintk("RPC: %s: reply 0x%p failed "
761 "to match any request xid 0x%08x len %d\n",
762 __func__, rep, headerp->rm_xid, rep->rr_len);
763repost:
764 r_xprt->rx_stats.bad_reply_count++;
765 rep->rr_func = rpcrdma_reply_handler;
766 if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, &r_xprt->rx_ep, rep))
767 rpcrdma_recv_buffer_put(rep);
768
769 return;
770 }
771
772 /* get request object */
773 req = rpcr_to_rdmar(rqst);
774
775 dprintk("RPC: %s: reply 0x%p completes request 0x%p\n"
776 " RPC request 0x%p xid 0x%08x\n",
777 __func__, rep, req, rqst, headerp->rm_xid);
778
779 BUG_ON(!req || req->rl_reply);
780
781 /* from here on, the reply is no longer an orphan */
782 req->rl_reply = rep;
783
784 /* check for expected message types */
785 /* The order of some of these tests is important. */
786 switch (headerp->rm_type) {
787 case htonl(RDMA_MSG):
788 /* never expect read chunks */
789 /* never expect reply chunks (two ways to check) */
790 /* never expect write chunks without having offered RDMA */
791 if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
792 (headerp->rm_body.rm_chunks[1] == xdr_zero &&
793 headerp->rm_body.rm_chunks[2] != xdr_zero) ||
794 (headerp->rm_body.rm_chunks[1] != xdr_zero &&
795 req->rl_nchunks == 0))
796 goto badheader;
797 if (headerp->rm_body.rm_chunks[1] != xdr_zero) {
798 /* count any expected write chunks in read reply */
799 /* start at write chunk array count */
800 iptr = &headerp->rm_body.rm_chunks[2];
801 rdmalen = rpcrdma_count_chunks(rep,
802 req->rl_nchunks, 1, &iptr);
803 /* check for validity, and no reply chunk after */
804 if (rdmalen < 0 || *iptr++ != xdr_zero)
805 goto badheader;
806 rep->rr_len -=
807 ((unsigned char *)iptr - (unsigned char *)headerp);
808 status = rep->rr_len + rdmalen;
809 r_xprt->rx_stats.total_rdma_reply += rdmalen;
810 /* special case - last chunk may omit padding */
811 if (rdmalen &= 3) {
812 rdmalen = 4 - rdmalen;
813 status += rdmalen;
814 }
815 } else {
816 /* else ordinary inline */
817 rdmalen = 0;
818 iptr = (__be32 *)((unsigned char *)headerp + 28);
819 rep->rr_len -= 28; /*sizeof *headerp;*/
820 status = rep->rr_len;
821 }
822 /* Fix up the rpc results for upper layer */
823 rpcrdma_inline_fixup(rqst, (char *)iptr, rep->rr_len, rdmalen);
824 break;
825
826 case htonl(RDMA_NOMSG):
827 /* never expect read or write chunks, always reply chunks */
828 if (headerp->rm_body.rm_chunks[0] != xdr_zero ||
829 headerp->rm_body.rm_chunks[1] != xdr_zero ||
830 headerp->rm_body.rm_chunks[2] != xdr_one ||
831 req->rl_nchunks == 0)
832 goto badheader;
833 iptr = (__be32 *)((unsigned char *)headerp + 28);
834 rdmalen = rpcrdma_count_chunks(rep, req->rl_nchunks, 0, &iptr);
835 if (rdmalen < 0)
836 goto badheader;
837 r_xprt->rx_stats.total_rdma_reply += rdmalen;
838 /* Reply chunk buffer already is the reply vector - no fixup. */
839 status = rdmalen;
840 break;
841
842badheader:
843 default:
844 dprintk("%s: invalid rpcrdma reply header (type %d):"
845 " chunks[012] == %d %d %d"
846 " expected chunks <= %d\n",
847 __func__, ntohl(headerp->rm_type),
848 headerp->rm_body.rm_chunks[0],
849 headerp->rm_body.rm_chunks[1],
850 headerp->rm_body.rm_chunks[2],
851 req->rl_nchunks);
852 status = -EIO;
853 r_xprt->rx_stats.bad_reply_count++;
854 break;
855 }
856
857 /* If using mw bind, start the deregister process now. */
858 /* (Note: if mr_free(), cannot perform it here, in tasklet context) */
859 if (req->rl_nchunks) switch (r_xprt->rx_ia.ri_memreg_strategy) {
860 case RPCRDMA_MEMWINDOWS:
861 for (i = 0; req->rl_nchunks-- > 1;)
862 i += rpcrdma_deregister_external(
863 &req->rl_segments[i], r_xprt, NULL);
864 /* Optionally wait (not here) for unbinds to complete */
865 rep->rr_func = rpcrdma_unbind_func;
866 (void) rpcrdma_deregister_external(&req->rl_segments[i],
867 r_xprt, rep);
868 break;
869 case RPCRDMA_MEMWINDOWS_ASYNC:
870 for (i = 0; req->rl_nchunks--;)
871 i += rpcrdma_deregister_external(&req->rl_segments[i],
872 r_xprt, NULL);
873 break;
874 default:
875 break;
876 }
877
878 dprintk("RPC: %s: xprt_complete_rqst(0x%p, 0x%p, %d)\n",
879 __func__, xprt, rqst, status);
880 xprt_complete_rqst(rqst->rq_task, status);
881 spin_unlock(&xprt->transport_lock);
882}