xref: /openbmc/linux/net/sunrpc/xprtrdma/rpc_rdma.c (revision 7e6f7d24)
1 // SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
2 /*
3  * Copyright (c) 2014-2017 Oracle.  All rights reserved.
4  * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved.
5  *
6  * This software is available to you under a choice of one of two
7  * licenses.  You may choose to be licensed under the terms of the GNU
8  * General Public License (GPL) Version 2, available from the file
9  * COPYING in the main directory of this source tree, or the BSD-type
10  * license below:
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  *
16  *      Redistributions of source code must retain the above copyright
17  *      notice, this list of conditions and the following disclaimer.
18  *
19  *      Redistributions in binary form must reproduce the above
20  *      copyright notice, this list of conditions and the following
21  *      disclaimer in the documentation and/or other materials provided
22  *      with the distribution.
23  *
24  *      Neither the name of the Network Appliance, Inc. nor the names of
25  *      its contributors may be used to endorse or promote products
26  *      derived from this software without specific prior written
27  *      permission.
28  *
29  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
30  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
31  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
32  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
33  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
34  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
35  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
36  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
37  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
38  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
39  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
40  */
41 
42 /*
43  * rpc_rdma.c
44  *
45  * This file contains the guts of the RPC RDMA protocol, and
46  * does marshaling/unmarshaling, etc. It is also where interfacing
47  * to the Linux RPC framework lives.
48  */
49 
50 #include <linux/highmem.h>
51 
52 #include <linux/sunrpc/svc_rdma.h>
53 
54 #include "xprt_rdma.h"
55 #include <trace/events/rpcrdma.h>
56 
57 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
58 # define RPCDBG_FACILITY	RPCDBG_TRANS
59 #endif
60 
61 /* Returns size of largest RPC-over-RDMA header in a Call message
62  *
63  * The largest Call header contains a full-size Read list and a
64  * minimal Reply chunk.
65  */
66 static unsigned int rpcrdma_max_call_header_size(unsigned int maxsegs)
67 {
68 	unsigned int size;
69 
70 	/* Fixed header fields and list discriminators */
71 	size = RPCRDMA_HDRLEN_MIN;
72 
73 	/* Maximum Read list size */
74 	maxsegs += 2;	/* segment for head and tail buffers */
75 	size = maxsegs * rpcrdma_readchunk_maxsz * sizeof(__be32);
76 
77 	/* Minimal Read chunk size */
78 	size += sizeof(__be32);	/* segment count */
79 	size += rpcrdma_segment_maxsz * sizeof(__be32);
80 	size += sizeof(__be32);	/* list discriminator */
81 
82 	dprintk("RPC:       %s: max call header size = %u\n",
83 		__func__, size);
84 	return size;
85 }
86 
87 /* Returns size of largest RPC-over-RDMA header in a Reply message
88  *
89  * There is only one Write list or one Reply chunk per Reply
90  * message.  The larger list is the Write list.
91  */
92 static unsigned int rpcrdma_max_reply_header_size(unsigned int maxsegs)
93 {
94 	unsigned int size;
95 
96 	/* Fixed header fields and list discriminators */
97 	size = RPCRDMA_HDRLEN_MIN;
98 
99 	/* Maximum Write list size */
100 	maxsegs += 2;	/* segment for head and tail buffers */
101 	size = sizeof(__be32);		/* segment count */
102 	size += maxsegs * rpcrdma_segment_maxsz * sizeof(__be32);
103 	size += sizeof(__be32);	/* list discriminator */
104 
105 	dprintk("RPC:       %s: max reply header size = %u\n",
106 		__func__, size);
107 	return size;
108 }
109 
110 void rpcrdma_set_max_header_sizes(struct rpcrdma_xprt *r_xprt)
111 {
112 	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
113 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
114 	unsigned int maxsegs = ia->ri_max_segs;
115 
116 	ia->ri_max_inline_write = cdata->inline_wsize -
117 				  rpcrdma_max_call_header_size(maxsegs);
118 	ia->ri_max_inline_read = cdata->inline_rsize -
119 				 rpcrdma_max_reply_header_size(maxsegs);
120 }
121 
122 /* The client can send a request inline as long as the RPCRDMA header
123  * plus the RPC call fit under the transport's inline limit. If the
124  * combined call message size exceeds that limit, the client must use
125  * a Read chunk for this operation.
126  *
127  * A Read chunk is also required if sending the RPC call inline would
128  * exceed this device's max_sge limit.
129  */
130 static bool rpcrdma_args_inline(struct rpcrdma_xprt *r_xprt,
131 				struct rpc_rqst *rqst)
132 {
133 	struct xdr_buf *xdr = &rqst->rq_snd_buf;
134 	unsigned int count, remaining, offset;
135 
136 	if (xdr->len > r_xprt->rx_ia.ri_max_inline_write)
137 		return false;
138 
139 	if (xdr->page_len) {
140 		remaining = xdr->page_len;
141 		offset = offset_in_page(xdr->page_base);
142 		count = RPCRDMA_MIN_SEND_SGES;
143 		while (remaining) {
144 			remaining -= min_t(unsigned int,
145 					   PAGE_SIZE - offset, remaining);
146 			offset = 0;
147 			if (++count > r_xprt->rx_ia.ri_max_send_sges)
148 				return false;
149 		}
150 	}
151 
152 	return true;
153 }
154 
155 /* The client can't know how large the actual reply will be. Thus it
156  * plans for the largest possible reply for that particular ULP
157  * operation. If the maximum combined reply message size exceeds that
158  * limit, the client must provide a write list or a reply chunk for
159  * this request.
160  */
161 static bool rpcrdma_results_inline(struct rpcrdma_xprt *r_xprt,
162 				   struct rpc_rqst *rqst)
163 {
164 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
165 
166 	return rqst->rq_rcv_buf.buflen <= ia->ri_max_inline_read;
167 }
168 
169 /* Split @vec on page boundaries into SGEs. FMR registers pages, not
170  * a byte range. Other modes coalesce these SGEs into a single MR
171  * when they can.
172  *
173  * Returns pointer to next available SGE, and bumps the total number
174  * of SGEs consumed.
175  */
176 static struct rpcrdma_mr_seg *
177 rpcrdma_convert_kvec(struct kvec *vec, struct rpcrdma_mr_seg *seg,
178 		     unsigned int *n)
179 {
180 	u32 remaining, page_offset;
181 	char *base;
182 
183 	base = vec->iov_base;
184 	page_offset = offset_in_page(base);
185 	remaining = vec->iov_len;
186 	while (remaining) {
187 		seg->mr_page = NULL;
188 		seg->mr_offset = base;
189 		seg->mr_len = min_t(u32, PAGE_SIZE - page_offset, remaining);
190 		remaining -= seg->mr_len;
191 		base += seg->mr_len;
192 		++seg;
193 		++(*n);
194 		page_offset = 0;
195 	}
196 	return seg;
197 }
198 
199 /* Convert @xdrbuf into SGEs no larger than a page each. As they
200  * are registered, these SGEs are then coalesced into RDMA segments
201  * when the selected memreg mode supports it.
202  *
203  * Returns positive number of SGEs consumed, or a negative errno.
204  */
205 
206 static int
207 rpcrdma_convert_iovs(struct rpcrdma_xprt *r_xprt, struct xdr_buf *xdrbuf,
208 		     unsigned int pos, enum rpcrdma_chunktype type,
209 		     struct rpcrdma_mr_seg *seg)
210 {
211 	unsigned long page_base;
212 	unsigned int len, n;
213 	struct page **ppages;
214 
215 	n = 0;
216 	if (pos == 0)
217 		seg = rpcrdma_convert_kvec(&xdrbuf->head[0], seg, &n);
218 
219 	len = xdrbuf->page_len;
220 	ppages = xdrbuf->pages + (xdrbuf->page_base >> PAGE_SHIFT);
221 	page_base = offset_in_page(xdrbuf->page_base);
222 	while (len) {
223 		if (unlikely(!*ppages)) {
224 			/* XXX: Certain upper layer operations do
225 			 *	not provide receive buffer pages.
226 			 */
227 			*ppages = alloc_page(GFP_ATOMIC);
228 			if (!*ppages)
229 				return -ENOBUFS;
230 		}
231 		seg->mr_page = *ppages;
232 		seg->mr_offset = (char *)page_base;
233 		seg->mr_len = min_t(u32, PAGE_SIZE - page_base, len);
234 		len -= seg->mr_len;
235 		++ppages;
236 		++seg;
237 		++n;
238 		page_base = 0;
239 	}
240 
241 	/* When encoding a Read chunk, the tail iovec contains an
242 	 * XDR pad and may be omitted.
243 	 */
244 	if (type == rpcrdma_readch && r_xprt->rx_ia.ri_implicit_roundup)
245 		goto out;
246 
247 	/* When encoding a Write chunk, some servers need to see an
248 	 * extra segment for non-XDR-aligned Write chunks. The upper
249 	 * layer provides space in the tail iovec that may be used
250 	 * for this purpose.
251 	 */
252 	if (type == rpcrdma_writech && r_xprt->rx_ia.ri_implicit_roundup)
253 		goto out;
254 
255 	if (xdrbuf->tail[0].iov_len)
256 		seg = rpcrdma_convert_kvec(&xdrbuf->tail[0], seg, &n);
257 
258 out:
259 	if (unlikely(n > RPCRDMA_MAX_SEGS))
260 		return -EIO;
261 	return n;
262 }
263 
264 static inline int
265 encode_item_present(struct xdr_stream *xdr)
266 {
267 	__be32 *p;
268 
269 	p = xdr_reserve_space(xdr, sizeof(*p));
270 	if (unlikely(!p))
271 		return -EMSGSIZE;
272 
273 	*p = xdr_one;
274 	return 0;
275 }
276 
277 static inline int
278 encode_item_not_present(struct xdr_stream *xdr)
279 {
280 	__be32 *p;
281 
282 	p = xdr_reserve_space(xdr, sizeof(*p));
283 	if (unlikely(!p))
284 		return -EMSGSIZE;
285 
286 	*p = xdr_zero;
287 	return 0;
288 }
289 
290 static void
291 xdr_encode_rdma_segment(__be32 *iptr, struct rpcrdma_mr *mr)
292 {
293 	*iptr++ = cpu_to_be32(mr->mr_handle);
294 	*iptr++ = cpu_to_be32(mr->mr_length);
295 	xdr_encode_hyper(iptr, mr->mr_offset);
296 }
297 
298 static int
299 encode_rdma_segment(struct xdr_stream *xdr, struct rpcrdma_mr *mr)
300 {
301 	__be32 *p;
302 
303 	p = xdr_reserve_space(xdr, 4 * sizeof(*p));
304 	if (unlikely(!p))
305 		return -EMSGSIZE;
306 
307 	xdr_encode_rdma_segment(p, mr);
308 	return 0;
309 }
310 
311 static int
312 encode_read_segment(struct xdr_stream *xdr, struct rpcrdma_mr *mr,
313 		    u32 position)
314 {
315 	__be32 *p;
316 
317 	p = xdr_reserve_space(xdr, 6 * sizeof(*p));
318 	if (unlikely(!p))
319 		return -EMSGSIZE;
320 
321 	*p++ = xdr_one;			/* Item present */
322 	*p++ = cpu_to_be32(position);
323 	xdr_encode_rdma_segment(p, mr);
324 	return 0;
325 }
326 
327 /* Register and XDR encode the Read list. Supports encoding a list of read
328  * segments that belong to a single read chunk.
329  *
330  * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64):
331  *
332  *  Read chunklist (a linked list):
333  *   N elements, position P (same P for all chunks of same arg!):
334  *    1 - PHLOO - 1 - PHLOO - ... - 1 - PHLOO - 0
335  *
336  * Returns zero on success, or a negative errno if a failure occurred.
337  * @xdr is advanced to the next position in the stream.
338  *
339  * Only a single @pos value is currently supported.
340  */
341 static noinline int
342 rpcrdma_encode_read_list(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
343 			 struct rpc_rqst *rqst, enum rpcrdma_chunktype rtype)
344 {
345 	struct xdr_stream *xdr = &req->rl_stream;
346 	struct rpcrdma_mr_seg *seg;
347 	struct rpcrdma_mr *mr;
348 	unsigned int pos;
349 	int nsegs;
350 
351 	pos = rqst->rq_snd_buf.head[0].iov_len;
352 	if (rtype == rpcrdma_areadch)
353 		pos = 0;
354 	seg = req->rl_segments;
355 	nsegs = rpcrdma_convert_iovs(r_xprt, &rqst->rq_snd_buf, pos,
356 				     rtype, seg);
357 	if (nsegs < 0)
358 		return nsegs;
359 
360 	do {
361 		seg = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs,
362 						   false, &mr);
363 		if (IS_ERR(seg))
364 			return PTR_ERR(seg);
365 		rpcrdma_mr_push(mr, &req->rl_registered);
366 
367 		if (encode_read_segment(xdr, mr, pos) < 0)
368 			return -EMSGSIZE;
369 
370 		trace_xprtrdma_read_chunk(rqst->rq_task, pos, mr, nsegs);
371 		r_xprt->rx_stats.read_chunk_count++;
372 		nsegs -= mr->mr_nents;
373 	} while (nsegs);
374 
375 	return 0;
376 }
377 
378 /* Register and XDR encode the Write list. Supports encoding a list
379  * containing one array of plain segments that belong to a single
380  * write chunk.
381  *
382  * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64):
383  *
384  *  Write chunklist (a list of (one) counted array):
385  *   N elements:
386  *    1 - N - HLOO - HLOO - ... - HLOO - 0
387  *
388  * Returns zero on success, or a negative errno if a failure occurred.
389  * @xdr is advanced to the next position in the stream.
390  *
391  * Only a single Write chunk is currently supported.
392  */
393 static noinline int
394 rpcrdma_encode_write_list(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
395 			  struct rpc_rqst *rqst, enum rpcrdma_chunktype wtype)
396 {
397 	struct xdr_stream *xdr = &req->rl_stream;
398 	struct rpcrdma_mr_seg *seg;
399 	struct rpcrdma_mr *mr;
400 	int nsegs, nchunks;
401 	__be32 *segcount;
402 
403 	seg = req->rl_segments;
404 	nsegs = rpcrdma_convert_iovs(r_xprt, &rqst->rq_rcv_buf,
405 				     rqst->rq_rcv_buf.head[0].iov_len,
406 				     wtype, seg);
407 	if (nsegs < 0)
408 		return nsegs;
409 
410 	if (encode_item_present(xdr) < 0)
411 		return -EMSGSIZE;
412 	segcount = xdr_reserve_space(xdr, sizeof(*segcount));
413 	if (unlikely(!segcount))
414 		return -EMSGSIZE;
415 	/* Actual value encoded below */
416 
417 	nchunks = 0;
418 	do {
419 		seg = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs,
420 						   true, &mr);
421 		if (IS_ERR(seg))
422 			return PTR_ERR(seg);
423 		rpcrdma_mr_push(mr, &req->rl_registered);
424 
425 		if (encode_rdma_segment(xdr, mr) < 0)
426 			return -EMSGSIZE;
427 
428 		trace_xprtrdma_write_chunk(rqst->rq_task, mr, nsegs);
429 		r_xprt->rx_stats.write_chunk_count++;
430 		r_xprt->rx_stats.total_rdma_request += mr->mr_length;
431 		nchunks++;
432 		nsegs -= mr->mr_nents;
433 	} while (nsegs);
434 
435 	/* Update count of segments in this Write chunk */
436 	*segcount = cpu_to_be32(nchunks);
437 
438 	return 0;
439 }
440 
441 /* Register and XDR encode the Reply chunk. Supports encoding an array
442  * of plain segments that belong to a single write (reply) chunk.
443  *
444  * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64):
445  *
446  *  Reply chunk (a counted array):
447  *   N elements:
448  *    1 - N - HLOO - HLOO - ... - HLOO
449  *
450  * Returns zero on success, or a negative errno if a failure occurred.
451  * @xdr is advanced to the next position in the stream.
452  */
453 static noinline int
454 rpcrdma_encode_reply_chunk(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
455 			   struct rpc_rqst *rqst, enum rpcrdma_chunktype wtype)
456 {
457 	struct xdr_stream *xdr = &req->rl_stream;
458 	struct rpcrdma_mr_seg *seg;
459 	struct rpcrdma_mr *mr;
460 	int nsegs, nchunks;
461 	__be32 *segcount;
462 
463 	seg = req->rl_segments;
464 	nsegs = rpcrdma_convert_iovs(r_xprt, &rqst->rq_rcv_buf, 0, wtype, seg);
465 	if (nsegs < 0)
466 		return nsegs;
467 
468 	if (encode_item_present(xdr) < 0)
469 		return -EMSGSIZE;
470 	segcount = xdr_reserve_space(xdr, sizeof(*segcount));
471 	if (unlikely(!segcount))
472 		return -EMSGSIZE;
473 	/* Actual value encoded below */
474 
475 	nchunks = 0;
476 	do {
477 		seg = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs,
478 						   true, &mr);
479 		if (IS_ERR(seg))
480 			return PTR_ERR(seg);
481 		rpcrdma_mr_push(mr, &req->rl_registered);
482 
483 		if (encode_rdma_segment(xdr, mr) < 0)
484 			return -EMSGSIZE;
485 
486 		trace_xprtrdma_reply_chunk(rqst->rq_task, mr, nsegs);
487 		r_xprt->rx_stats.reply_chunk_count++;
488 		r_xprt->rx_stats.total_rdma_request += mr->mr_length;
489 		nchunks++;
490 		nsegs -= mr->mr_nents;
491 	} while (nsegs);
492 
493 	/* Update count of segments in the Reply chunk */
494 	*segcount = cpu_to_be32(nchunks);
495 
496 	return 0;
497 }
498 
499 /**
500  * rpcrdma_unmap_sendctx - DMA-unmap Send buffers
501  * @sc: sendctx containing SGEs to unmap
502  *
503  */
504 void
505 rpcrdma_unmap_sendctx(struct rpcrdma_sendctx *sc)
506 {
507 	struct rpcrdma_ia *ia = &sc->sc_xprt->rx_ia;
508 	struct ib_sge *sge;
509 	unsigned int count;
510 
511 	/* The first two SGEs contain the transport header and
512 	 * the inline buffer. These are always left mapped so
513 	 * they can be cheaply re-used.
514 	 */
515 	sge = &sc->sc_sges[2];
516 	for (count = sc->sc_unmap_count; count; ++sge, --count)
517 		ib_dma_unmap_page(ia->ri_device,
518 				  sge->addr, sge->length, DMA_TO_DEVICE);
519 
520 	if (test_and_clear_bit(RPCRDMA_REQ_F_TX_RESOURCES, &sc->sc_req->rl_flags)) {
521 		smp_mb__after_atomic();
522 		wake_up_bit(&sc->sc_req->rl_flags, RPCRDMA_REQ_F_TX_RESOURCES);
523 	}
524 }
525 
526 /* Prepare an SGE for the RPC-over-RDMA transport header.
527  */
528 static bool
529 rpcrdma_prepare_hdr_sge(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
530 			u32 len)
531 {
532 	struct rpcrdma_sendctx *sc = req->rl_sendctx;
533 	struct rpcrdma_regbuf *rb = req->rl_rdmabuf;
534 	struct ib_sge *sge = sc->sc_sges;
535 
536 	if (!rpcrdma_dma_map_regbuf(ia, rb))
537 		goto out_regbuf;
538 	sge->addr = rdmab_addr(rb);
539 	sge->length = len;
540 	sge->lkey = rdmab_lkey(rb);
541 
542 	ib_dma_sync_single_for_device(rdmab_device(rb), sge->addr,
543 				      sge->length, DMA_TO_DEVICE);
544 	sc->sc_wr.num_sge++;
545 	return true;
546 
547 out_regbuf:
548 	pr_err("rpcrdma: failed to DMA map a Send buffer\n");
549 	return false;
550 }
551 
552 /* Prepare the Send SGEs. The head and tail iovec, and each entry
553  * in the page list, gets its own SGE.
554  */
555 static bool
556 rpcrdma_prepare_msg_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
557 			 struct xdr_buf *xdr, enum rpcrdma_chunktype rtype)
558 {
559 	struct rpcrdma_sendctx *sc = req->rl_sendctx;
560 	unsigned int sge_no, page_base, len, remaining;
561 	struct rpcrdma_regbuf *rb = req->rl_sendbuf;
562 	struct ib_device *device = ia->ri_device;
563 	struct ib_sge *sge = sc->sc_sges;
564 	u32 lkey = ia->ri_pd->local_dma_lkey;
565 	struct page *page, **ppages;
566 
567 	/* The head iovec is straightforward, as it is already
568 	 * DMA-mapped. Sync the content that has changed.
569 	 */
570 	if (!rpcrdma_dma_map_regbuf(ia, rb))
571 		goto out_regbuf;
572 	sge_no = 1;
573 	sge[sge_no].addr = rdmab_addr(rb);
574 	sge[sge_no].length = xdr->head[0].iov_len;
575 	sge[sge_no].lkey = rdmab_lkey(rb);
576 	ib_dma_sync_single_for_device(rdmab_device(rb), sge[sge_no].addr,
577 				      sge[sge_no].length, DMA_TO_DEVICE);
578 
579 	/* If there is a Read chunk, the page list is being handled
580 	 * via explicit RDMA, and thus is skipped here. However, the
581 	 * tail iovec may include an XDR pad for the page list, as
582 	 * well as additional content, and may not reside in the
583 	 * same page as the head iovec.
584 	 */
585 	if (rtype == rpcrdma_readch) {
586 		len = xdr->tail[0].iov_len;
587 
588 		/* Do not include the tail if it is only an XDR pad */
589 		if (len < 4)
590 			goto out;
591 
592 		page = virt_to_page(xdr->tail[0].iov_base);
593 		page_base = offset_in_page(xdr->tail[0].iov_base);
594 
595 		/* If the content in the page list is an odd length,
596 		 * xdr_write_pages() has added a pad at the beginning
597 		 * of the tail iovec. Force the tail's non-pad content
598 		 * to land at the next XDR position in the Send message.
599 		 */
600 		page_base += len & 3;
601 		len -= len & 3;
602 		goto map_tail;
603 	}
604 
605 	/* If there is a page list present, temporarily DMA map
606 	 * and prepare an SGE for each page to be sent.
607 	 */
608 	if (xdr->page_len) {
609 		ppages = xdr->pages + (xdr->page_base >> PAGE_SHIFT);
610 		page_base = offset_in_page(xdr->page_base);
611 		remaining = xdr->page_len;
612 		while (remaining) {
613 			sge_no++;
614 			if (sge_no > RPCRDMA_MAX_SEND_SGES - 2)
615 				goto out_mapping_overflow;
616 
617 			len = min_t(u32, PAGE_SIZE - page_base, remaining);
618 			sge[sge_no].addr = ib_dma_map_page(device, *ppages,
619 							   page_base, len,
620 							   DMA_TO_DEVICE);
621 			if (ib_dma_mapping_error(device, sge[sge_no].addr))
622 				goto out_mapping_err;
623 			sge[sge_no].length = len;
624 			sge[sge_no].lkey = lkey;
625 
626 			sc->sc_unmap_count++;
627 			ppages++;
628 			remaining -= len;
629 			page_base = 0;
630 		}
631 	}
632 
633 	/* The tail iovec is not always constructed in the same
634 	 * page where the head iovec resides (see, for example,
635 	 * gss_wrap_req_priv). To neatly accommodate that case,
636 	 * DMA map it separately.
637 	 */
638 	if (xdr->tail[0].iov_len) {
639 		page = virt_to_page(xdr->tail[0].iov_base);
640 		page_base = offset_in_page(xdr->tail[0].iov_base);
641 		len = xdr->tail[0].iov_len;
642 
643 map_tail:
644 		sge_no++;
645 		sge[sge_no].addr = ib_dma_map_page(device, page,
646 						   page_base, len,
647 						   DMA_TO_DEVICE);
648 		if (ib_dma_mapping_error(device, sge[sge_no].addr))
649 			goto out_mapping_err;
650 		sge[sge_no].length = len;
651 		sge[sge_no].lkey = lkey;
652 		sc->sc_unmap_count++;
653 	}
654 
655 out:
656 	sc->sc_wr.num_sge += sge_no;
657 	if (sc->sc_unmap_count)
658 		__set_bit(RPCRDMA_REQ_F_TX_RESOURCES, &req->rl_flags);
659 	return true;
660 
661 out_regbuf:
662 	pr_err("rpcrdma: failed to DMA map a Send buffer\n");
663 	return false;
664 
665 out_mapping_overflow:
666 	rpcrdma_unmap_sendctx(sc);
667 	pr_err("rpcrdma: too many Send SGEs (%u)\n", sge_no);
668 	return false;
669 
670 out_mapping_err:
671 	rpcrdma_unmap_sendctx(sc);
672 	pr_err("rpcrdma: Send mapping error\n");
673 	return false;
674 }
675 
676 /**
677  * rpcrdma_prepare_send_sges - Construct SGEs for a Send WR
678  * @r_xprt: controlling transport
679  * @req: context of RPC Call being marshalled
680  * @hdrlen: size of transport header, in bytes
681  * @xdr: xdr_buf containing RPC Call
682  * @rtype: chunk type being encoded
683  *
684  * Returns 0 on success; otherwise a negative errno is returned.
685  */
686 int
687 rpcrdma_prepare_send_sges(struct rpcrdma_xprt *r_xprt,
688 			  struct rpcrdma_req *req, u32 hdrlen,
689 			  struct xdr_buf *xdr, enum rpcrdma_chunktype rtype)
690 {
691 	req->rl_sendctx = rpcrdma_sendctx_get_locked(&r_xprt->rx_buf);
692 	if (!req->rl_sendctx)
693 		return -EAGAIN;
694 	req->rl_sendctx->sc_wr.num_sge = 0;
695 	req->rl_sendctx->sc_unmap_count = 0;
696 	req->rl_sendctx->sc_req = req;
697 	__clear_bit(RPCRDMA_REQ_F_TX_RESOURCES, &req->rl_flags);
698 
699 	if (!rpcrdma_prepare_hdr_sge(&r_xprt->rx_ia, req, hdrlen))
700 		return -EIO;
701 
702 	if (rtype != rpcrdma_areadch)
703 		if (!rpcrdma_prepare_msg_sges(&r_xprt->rx_ia, req, xdr, rtype))
704 			return -EIO;
705 
706 	return 0;
707 }
708 
709 /**
710  * rpcrdma_marshal_req - Marshal and send one RPC request
711  * @r_xprt: controlling transport
712  * @rqst: RPC request to be marshaled
713  *
714  * For the RPC in "rqst", this function:
715  *  - Chooses the transfer mode (eg., RDMA_MSG or RDMA_NOMSG)
716  *  - Registers Read, Write, and Reply chunks
717  *  - Constructs the transport header
718  *  - Posts a Send WR to send the transport header and request
719  *
720  * Returns:
721  *	%0 if the RPC was sent successfully,
722  *	%-ENOTCONN if the connection was lost,
723  *	%-EAGAIN if the caller should call again with the same arguments,
724  *	%-ENOBUFS if the caller should call again after a delay,
725  *	%-EMSGSIZE if the transport header is too small,
726  *	%-EIO if a permanent problem occurred while marshaling.
727  */
728 int
729 rpcrdma_marshal_req(struct rpcrdma_xprt *r_xprt, struct rpc_rqst *rqst)
730 {
731 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
732 	struct xdr_stream *xdr = &req->rl_stream;
733 	enum rpcrdma_chunktype rtype, wtype;
734 	bool ddp_allowed;
735 	__be32 *p;
736 	int ret;
737 
738 	rpcrdma_set_xdrlen(&req->rl_hdrbuf, 0);
739 	xdr_init_encode(xdr, &req->rl_hdrbuf,
740 			req->rl_rdmabuf->rg_base);
741 
742 	/* Fixed header fields */
743 	ret = -EMSGSIZE;
744 	p = xdr_reserve_space(xdr, 4 * sizeof(*p));
745 	if (!p)
746 		goto out_err;
747 	*p++ = rqst->rq_xid;
748 	*p++ = rpcrdma_version;
749 	*p++ = cpu_to_be32(r_xprt->rx_buf.rb_max_requests);
750 
751 	/* When the ULP employs a GSS flavor that guarantees integrity
752 	 * or privacy, direct data placement of individual data items
753 	 * is not allowed.
754 	 */
755 	ddp_allowed = !(rqst->rq_cred->cr_auth->au_flags &
756 						RPCAUTH_AUTH_DATATOUCH);
757 
758 	/*
759 	 * Chunks needed for results?
760 	 *
761 	 * o If the expected result is under the inline threshold, all ops
762 	 *   return as inline.
763 	 * o Large read ops return data as write chunk(s), header as
764 	 *   inline.
765 	 * o Large non-read ops return as a single reply chunk.
766 	 */
767 	if (rpcrdma_results_inline(r_xprt, rqst))
768 		wtype = rpcrdma_noch;
769 	else if (ddp_allowed && rqst->rq_rcv_buf.flags & XDRBUF_READ)
770 		wtype = rpcrdma_writech;
771 	else
772 		wtype = rpcrdma_replych;
773 
774 	/*
775 	 * Chunks needed for arguments?
776 	 *
777 	 * o If the total request is under the inline threshold, all ops
778 	 *   are sent as inline.
779 	 * o Large write ops transmit data as read chunk(s), header as
780 	 *   inline.
781 	 * o Large non-write ops are sent with the entire message as a
782 	 *   single read chunk (protocol 0-position special case).
783 	 *
784 	 * This assumes that the upper layer does not present a request
785 	 * that both has a data payload, and whose non-data arguments
786 	 * by themselves are larger than the inline threshold.
787 	 */
788 	if (rpcrdma_args_inline(r_xprt, rqst)) {
789 		*p++ = rdma_msg;
790 		rtype = rpcrdma_noch;
791 	} else if (ddp_allowed && rqst->rq_snd_buf.flags & XDRBUF_WRITE) {
792 		*p++ = rdma_msg;
793 		rtype = rpcrdma_readch;
794 	} else {
795 		r_xprt->rx_stats.nomsg_call_count++;
796 		*p++ = rdma_nomsg;
797 		rtype = rpcrdma_areadch;
798 	}
799 
800 	/* If this is a retransmit, discard previously registered
801 	 * chunks. Very likely the connection has been replaced,
802 	 * so these registrations are invalid and unusable.
803 	 */
804 	while (unlikely(!list_empty(&req->rl_registered))) {
805 		struct rpcrdma_mr *mr;
806 
807 		mr = rpcrdma_mr_pop(&req->rl_registered);
808 		rpcrdma_mr_defer_recovery(mr);
809 	}
810 
811 	/* This implementation supports the following combinations
812 	 * of chunk lists in one RPC-over-RDMA Call message:
813 	 *
814 	 *   - Read list
815 	 *   - Write list
816 	 *   - Reply chunk
817 	 *   - Read list + Reply chunk
818 	 *
819 	 * It might not yet support the following combinations:
820 	 *
821 	 *   - Read list + Write list
822 	 *
823 	 * It does not support the following combinations:
824 	 *
825 	 *   - Write list + Reply chunk
826 	 *   - Read list + Write list + Reply chunk
827 	 *
828 	 * This implementation supports only a single chunk in each
829 	 * Read or Write list. Thus for example the client cannot
830 	 * send a Call message with a Position Zero Read chunk and a
831 	 * regular Read chunk at the same time.
832 	 */
833 	if (rtype != rpcrdma_noch) {
834 		ret = rpcrdma_encode_read_list(r_xprt, req, rqst, rtype);
835 		if (ret)
836 			goto out_err;
837 	}
838 	ret = encode_item_not_present(xdr);
839 	if (ret)
840 		goto out_err;
841 
842 	if (wtype == rpcrdma_writech) {
843 		ret = rpcrdma_encode_write_list(r_xprt, req, rqst, wtype);
844 		if (ret)
845 			goto out_err;
846 	}
847 	ret = encode_item_not_present(xdr);
848 	if (ret)
849 		goto out_err;
850 
851 	if (wtype != rpcrdma_replych)
852 		ret = encode_item_not_present(xdr);
853 	else
854 		ret = rpcrdma_encode_reply_chunk(r_xprt, req, rqst, wtype);
855 	if (ret)
856 		goto out_err;
857 
858 	trace_xprtrdma_marshal(rqst, xdr_stream_pos(xdr), rtype, wtype);
859 
860 	ret = rpcrdma_prepare_send_sges(r_xprt, req, xdr_stream_pos(xdr),
861 					&rqst->rq_snd_buf, rtype);
862 	if (ret)
863 		goto out_err;
864 	return 0;
865 
866 out_err:
867 	switch (ret) {
868 	case -EAGAIN:
869 		xprt_wait_for_buffer_space(rqst->rq_task, NULL);
870 		break;
871 	case -ENOBUFS:
872 		break;
873 	default:
874 		r_xprt->rx_stats.failed_marshal_count++;
875 	}
876 	return ret;
877 }
878 
879 /**
880  * rpcrdma_inline_fixup - Scatter inline received data into rqst's iovecs
881  * @rqst: controlling RPC request
882  * @srcp: points to RPC message payload in receive buffer
883  * @copy_len: remaining length of receive buffer content
884  * @pad: Write chunk pad bytes needed (zero for pure inline)
885  *
886  * The upper layer has set the maximum number of bytes it can
887  * receive in each component of rq_rcv_buf. These values are set in
888  * the head.iov_len, page_len, tail.iov_len, and buflen fields.
889  *
890  * Unlike the TCP equivalent (xdr_partial_copy_from_skb), in
891  * many cases this function simply updates iov_base pointers in
892  * rq_rcv_buf to point directly to the received reply data, to
893  * avoid copying reply data.
894  *
895  * Returns the count of bytes which had to be memcopied.
896  */
897 static unsigned long
898 rpcrdma_inline_fixup(struct rpc_rqst *rqst, char *srcp, int copy_len, int pad)
899 {
900 	unsigned long fixup_copy_count;
901 	int i, npages, curlen;
902 	char *destp;
903 	struct page **ppages;
904 	int page_base;
905 
906 	/* The head iovec is redirected to the RPC reply message
907 	 * in the receive buffer, to avoid a memcopy.
908 	 */
909 	rqst->rq_rcv_buf.head[0].iov_base = srcp;
910 	rqst->rq_private_buf.head[0].iov_base = srcp;
911 
912 	/* The contents of the receive buffer that follow
913 	 * head.iov_len bytes are copied into the page list.
914 	 */
915 	curlen = rqst->rq_rcv_buf.head[0].iov_len;
916 	if (curlen > copy_len)
917 		curlen = copy_len;
918 	trace_xprtrdma_fixup(rqst, copy_len, curlen);
919 	srcp += curlen;
920 	copy_len -= curlen;
921 
922 	ppages = rqst->rq_rcv_buf.pages +
923 		(rqst->rq_rcv_buf.page_base >> PAGE_SHIFT);
924 	page_base = offset_in_page(rqst->rq_rcv_buf.page_base);
925 	fixup_copy_count = 0;
926 	if (copy_len && rqst->rq_rcv_buf.page_len) {
927 		int pagelist_len;
928 
929 		pagelist_len = rqst->rq_rcv_buf.page_len;
930 		if (pagelist_len > copy_len)
931 			pagelist_len = copy_len;
932 		npages = PAGE_ALIGN(page_base + pagelist_len) >> PAGE_SHIFT;
933 		for (i = 0; i < npages; i++) {
934 			curlen = PAGE_SIZE - page_base;
935 			if (curlen > pagelist_len)
936 				curlen = pagelist_len;
937 
938 			trace_xprtrdma_fixup_pg(rqst, i, srcp,
939 						copy_len, curlen);
940 			destp = kmap_atomic(ppages[i]);
941 			memcpy(destp + page_base, srcp, curlen);
942 			flush_dcache_page(ppages[i]);
943 			kunmap_atomic(destp);
944 			srcp += curlen;
945 			copy_len -= curlen;
946 			fixup_copy_count += curlen;
947 			pagelist_len -= curlen;
948 			if (!pagelist_len)
949 				break;
950 			page_base = 0;
951 		}
952 
953 		/* Implicit padding for the last segment in a Write
954 		 * chunk is inserted inline at the front of the tail
955 		 * iovec. The upper layer ignores the content of
956 		 * the pad. Simply ensure inline content in the tail
957 		 * that follows the Write chunk is properly aligned.
958 		 */
959 		if (pad)
960 			srcp -= pad;
961 	}
962 
963 	/* The tail iovec is redirected to the remaining data
964 	 * in the receive buffer, to avoid a memcopy.
965 	 */
966 	if (copy_len || pad) {
967 		rqst->rq_rcv_buf.tail[0].iov_base = srcp;
968 		rqst->rq_private_buf.tail[0].iov_base = srcp;
969 	}
970 
971 	return fixup_copy_count;
972 }
973 
974 /* By convention, backchannel calls arrive via rdma_msg type
975  * messages, and never populate the chunk lists. This makes
976  * the RPC/RDMA header small and fixed in size, so it is
977  * straightforward to check the RPC header's direction field.
978  */
979 static bool
980 rpcrdma_is_bcall(struct rpcrdma_xprt *r_xprt, struct rpcrdma_rep *rep)
981 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
982 {
983 	struct xdr_stream *xdr = &rep->rr_stream;
984 	__be32 *p;
985 
986 	if (rep->rr_proc != rdma_msg)
987 		return false;
988 
989 	/* Peek at stream contents without advancing. */
990 	p = xdr_inline_decode(xdr, 0);
991 
992 	/* Chunk lists */
993 	if (*p++ != xdr_zero)
994 		return false;
995 	if (*p++ != xdr_zero)
996 		return false;
997 	if (*p++ != xdr_zero)
998 		return false;
999 
1000 	/* RPC header */
1001 	if (*p++ != rep->rr_xid)
1002 		return false;
1003 	if (*p != cpu_to_be32(RPC_CALL))
1004 		return false;
1005 
1006 	/* Now that we are sure this is a backchannel call,
1007 	 * advance to the RPC header.
1008 	 */
1009 	p = xdr_inline_decode(xdr, 3 * sizeof(*p));
1010 	if (unlikely(!p))
1011 		goto out_short;
1012 
1013 	rpcrdma_bc_receive_call(r_xprt, rep);
1014 	return true;
1015 
1016 out_short:
1017 	pr_warn("RPC/RDMA short backward direction call\n");
1018 	return true;
1019 }
1020 #else	/* CONFIG_SUNRPC_BACKCHANNEL */
1021 {
1022 	return false;
1023 }
1024 #endif	/* CONFIG_SUNRPC_BACKCHANNEL */
1025 
1026 static int decode_rdma_segment(struct xdr_stream *xdr, u32 *length)
1027 {
1028 	u32 handle;
1029 	u64 offset;
1030 	__be32 *p;
1031 
1032 	p = xdr_inline_decode(xdr, 4 * sizeof(*p));
1033 	if (unlikely(!p))
1034 		return -EIO;
1035 
1036 	handle = be32_to_cpup(p++);
1037 	*length = be32_to_cpup(p++);
1038 	xdr_decode_hyper(p, &offset);
1039 
1040 	trace_xprtrdma_decode_seg(handle, *length, offset);
1041 	return 0;
1042 }
1043 
1044 static int decode_write_chunk(struct xdr_stream *xdr, u32 *length)
1045 {
1046 	u32 segcount, seglength;
1047 	__be32 *p;
1048 
1049 	p = xdr_inline_decode(xdr, sizeof(*p));
1050 	if (unlikely(!p))
1051 		return -EIO;
1052 
1053 	*length = 0;
1054 	segcount = be32_to_cpup(p);
1055 	while (segcount--) {
1056 		if (decode_rdma_segment(xdr, &seglength))
1057 			return -EIO;
1058 		*length += seglength;
1059 	}
1060 
1061 	return 0;
1062 }
1063 
1064 /* In RPC-over-RDMA Version One replies, a Read list is never
1065  * expected. This decoder is a stub that returns an error if
1066  * a Read list is present.
1067  */
1068 static int decode_read_list(struct xdr_stream *xdr)
1069 {
1070 	__be32 *p;
1071 
1072 	p = xdr_inline_decode(xdr, sizeof(*p));
1073 	if (unlikely(!p))
1074 		return -EIO;
1075 	if (unlikely(*p != xdr_zero))
1076 		return -EIO;
1077 	return 0;
1078 }
1079 
1080 /* Supports only one Write chunk in the Write list
1081  */
1082 static int decode_write_list(struct xdr_stream *xdr, u32 *length)
1083 {
1084 	u32 chunklen;
1085 	bool first;
1086 	__be32 *p;
1087 
1088 	*length = 0;
1089 	first = true;
1090 	do {
1091 		p = xdr_inline_decode(xdr, sizeof(*p));
1092 		if (unlikely(!p))
1093 			return -EIO;
1094 		if (*p == xdr_zero)
1095 			break;
1096 		if (!first)
1097 			return -EIO;
1098 
1099 		if (decode_write_chunk(xdr, &chunklen))
1100 			return -EIO;
1101 		*length += chunklen;
1102 		first = false;
1103 	} while (true);
1104 	return 0;
1105 }
1106 
1107 static int decode_reply_chunk(struct xdr_stream *xdr, u32 *length)
1108 {
1109 	__be32 *p;
1110 
1111 	p = xdr_inline_decode(xdr, sizeof(*p));
1112 	if (unlikely(!p))
1113 		return -EIO;
1114 
1115 	*length = 0;
1116 	if (*p != xdr_zero)
1117 		if (decode_write_chunk(xdr, length))
1118 			return -EIO;
1119 	return 0;
1120 }
1121 
1122 static int
1123 rpcrdma_decode_msg(struct rpcrdma_xprt *r_xprt, struct rpcrdma_rep *rep,
1124 		   struct rpc_rqst *rqst)
1125 {
1126 	struct xdr_stream *xdr = &rep->rr_stream;
1127 	u32 writelist, replychunk, rpclen;
1128 	char *base;
1129 
1130 	/* Decode the chunk lists */
1131 	if (decode_read_list(xdr))
1132 		return -EIO;
1133 	if (decode_write_list(xdr, &writelist))
1134 		return -EIO;
1135 	if (decode_reply_chunk(xdr, &replychunk))
1136 		return -EIO;
1137 
1138 	/* RDMA_MSG sanity checks */
1139 	if (unlikely(replychunk))
1140 		return -EIO;
1141 
1142 	/* Build the RPC reply's Payload stream in rqst->rq_rcv_buf */
1143 	base = (char *)xdr_inline_decode(xdr, 0);
1144 	rpclen = xdr_stream_remaining(xdr);
1145 	r_xprt->rx_stats.fixup_copy_count +=
1146 		rpcrdma_inline_fixup(rqst, base, rpclen, writelist & 3);
1147 
1148 	r_xprt->rx_stats.total_rdma_reply += writelist;
1149 	return rpclen + xdr_align_size(writelist);
1150 }
1151 
1152 static noinline int
1153 rpcrdma_decode_nomsg(struct rpcrdma_xprt *r_xprt, struct rpcrdma_rep *rep)
1154 {
1155 	struct xdr_stream *xdr = &rep->rr_stream;
1156 	u32 writelist, replychunk;
1157 
1158 	/* Decode the chunk lists */
1159 	if (decode_read_list(xdr))
1160 		return -EIO;
1161 	if (decode_write_list(xdr, &writelist))
1162 		return -EIO;
1163 	if (decode_reply_chunk(xdr, &replychunk))
1164 		return -EIO;
1165 
1166 	/* RDMA_NOMSG sanity checks */
1167 	if (unlikely(writelist))
1168 		return -EIO;
1169 	if (unlikely(!replychunk))
1170 		return -EIO;
1171 
1172 	/* Reply chunk buffer already is the reply vector */
1173 	r_xprt->rx_stats.total_rdma_reply += replychunk;
1174 	return replychunk;
1175 }
1176 
1177 static noinline int
1178 rpcrdma_decode_error(struct rpcrdma_xprt *r_xprt, struct rpcrdma_rep *rep,
1179 		     struct rpc_rqst *rqst)
1180 {
1181 	struct xdr_stream *xdr = &rep->rr_stream;
1182 	__be32 *p;
1183 
1184 	p = xdr_inline_decode(xdr, sizeof(*p));
1185 	if (unlikely(!p))
1186 		return -EIO;
1187 
1188 	switch (*p) {
1189 	case err_vers:
1190 		p = xdr_inline_decode(xdr, 2 * sizeof(*p));
1191 		if (!p)
1192 			break;
1193 		dprintk("RPC: %5u: %s: server reports version error (%u-%u)\n",
1194 			rqst->rq_task->tk_pid, __func__,
1195 			be32_to_cpup(p), be32_to_cpu(*(p + 1)));
1196 		break;
1197 	case err_chunk:
1198 		dprintk("RPC: %5u: %s: server reports header decoding error\n",
1199 			rqst->rq_task->tk_pid, __func__);
1200 		break;
1201 	default:
1202 		dprintk("RPC: %5u: %s: server reports unrecognized error %d\n",
1203 			rqst->rq_task->tk_pid, __func__, be32_to_cpup(p));
1204 	}
1205 
1206 	r_xprt->rx_stats.bad_reply_count++;
1207 	return -EREMOTEIO;
1208 }
1209 
1210 /* Perform XID lookup, reconstruction of the RPC reply, and
1211  * RPC completion while holding the transport lock to ensure
1212  * the rep, rqst, and rq_task pointers remain stable.
1213  */
1214 void rpcrdma_complete_rqst(struct rpcrdma_rep *rep)
1215 {
1216 	struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
1217 	struct rpc_xprt *xprt = &r_xprt->rx_xprt;
1218 	struct rpc_rqst *rqst = rep->rr_rqst;
1219 	unsigned long cwnd;
1220 	int status;
1221 
1222 	xprt->reestablish_timeout = 0;
1223 
1224 	switch (rep->rr_proc) {
1225 	case rdma_msg:
1226 		status = rpcrdma_decode_msg(r_xprt, rep, rqst);
1227 		break;
1228 	case rdma_nomsg:
1229 		status = rpcrdma_decode_nomsg(r_xprt, rep);
1230 		break;
1231 	case rdma_error:
1232 		status = rpcrdma_decode_error(r_xprt, rep, rqst);
1233 		break;
1234 	default:
1235 		status = -EIO;
1236 	}
1237 	if (status < 0)
1238 		goto out_badheader;
1239 
1240 out:
1241 	spin_lock(&xprt->recv_lock);
1242 	cwnd = xprt->cwnd;
1243 	xprt->cwnd = r_xprt->rx_buf.rb_credits << RPC_CWNDSHIFT;
1244 	if (xprt->cwnd > cwnd)
1245 		xprt_release_rqst_cong(rqst->rq_task);
1246 
1247 	xprt_complete_rqst(rqst->rq_task, status);
1248 	xprt_unpin_rqst(rqst);
1249 	spin_unlock(&xprt->recv_lock);
1250 	return;
1251 
1252 /* If the incoming reply terminated a pending RPC, the next
1253  * RPC call will post a replacement receive buffer as it is
1254  * being marshaled.
1255  */
1256 out_badheader:
1257 	trace_xprtrdma_reply_hdr(rep);
1258 	r_xprt->rx_stats.bad_reply_count++;
1259 	status = -EIO;
1260 	goto out;
1261 }
1262 
1263 void rpcrdma_release_rqst(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
1264 {
1265 	/* Invalidate and unmap the data payloads before waking
1266 	 * the waiting application. This guarantees the memory
1267 	 * regions are properly fenced from the server before the
1268 	 * application accesses the data. It also ensures proper
1269 	 * send flow control: waking the next RPC waits until this
1270 	 * RPC has relinquished all its Send Queue entries.
1271 	 */
1272 	if (!list_empty(&req->rl_registered))
1273 		r_xprt->rx_ia.ri_ops->ro_unmap_sync(r_xprt,
1274 						    &req->rl_registered);
1275 
1276 	/* Ensure that any DMA mapped pages associated with
1277 	 * the Send of the RPC Call have been unmapped before
1278 	 * allowing the RPC to complete. This protects argument
1279 	 * memory not controlled by the RPC client from being
1280 	 * re-used before we're done with it.
1281 	 */
1282 	if (test_bit(RPCRDMA_REQ_F_TX_RESOURCES, &req->rl_flags)) {
1283 		r_xprt->rx_stats.reply_waits_for_send++;
1284 		out_of_line_wait_on_bit(&req->rl_flags,
1285 					RPCRDMA_REQ_F_TX_RESOURCES,
1286 					bit_wait,
1287 					TASK_UNINTERRUPTIBLE);
1288 	}
1289 }
1290 
1291 /* Reply handling runs in the poll worker thread. Anything that
1292  * might wait is deferred to a separate workqueue.
1293  */
1294 void rpcrdma_deferred_completion(struct work_struct *work)
1295 {
1296 	struct rpcrdma_rep *rep =
1297 			container_of(work, struct rpcrdma_rep, rr_work);
1298 	struct rpcrdma_req *req = rpcr_to_rdmar(rep->rr_rqst);
1299 	struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
1300 
1301 	trace_xprtrdma_defer_cmp(rep);
1302 	if (rep->rr_wc_flags & IB_WC_WITH_INVALIDATE)
1303 		r_xprt->rx_ia.ri_ops->ro_reminv(rep, &req->rl_registered);
1304 	rpcrdma_release_rqst(r_xprt, req);
1305 	rpcrdma_complete_rqst(rep);
1306 }
1307 
1308 /* Process received RPC/RDMA messages.
1309  *
1310  * Errors must result in the RPC task either being awakened, or
1311  * allowed to timeout, to discover the errors at that time.
1312  */
1313 void rpcrdma_reply_handler(struct rpcrdma_rep *rep)
1314 {
1315 	struct rpcrdma_xprt *r_xprt = rep->rr_rxprt;
1316 	struct rpc_xprt *xprt = &r_xprt->rx_xprt;
1317 	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
1318 	struct rpcrdma_req *req;
1319 	struct rpc_rqst *rqst;
1320 	u32 credits;
1321 	__be32 *p;
1322 
1323 	--buf->rb_posted_receives;
1324 
1325 	if (rep->rr_hdrbuf.head[0].iov_len == 0)
1326 		goto out_badstatus;
1327 
1328 	/* Fixed transport header fields */
1329 	xdr_init_decode(&rep->rr_stream, &rep->rr_hdrbuf,
1330 			rep->rr_hdrbuf.head[0].iov_base);
1331 	p = xdr_inline_decode(&rep->rr_stream, 4 * sizeof(*p));
1332 	if (unlikely(!p))
1333 		goto out_shortreply;
1334 	rep->rr_xid = *p++;
1335 	rep->rr_vers = *p++;
1336 	credits = be32_to_cpu(*p++);
1337 	rep->rr_proc = *p++;
1338 
1339 	if (rep->rr_vers != rpcrdma_version)
1340 		goto out_badversion;
1341 
1342 	if (rpcrdma_is_bcall(r_xprt, rep))
1343 		return;
1344 
1345 	/* Match incoming rpcrdma_rep to an rpcrdma_req to
1346 	 * get context for handling any incoming chunks.
1347 	 */
1348 	spin_lock(&xprt->recv_lock);
1349 	rqst = xprt_lookup_rqst(xprt, rep->rr_xid);
1350 	if (!rqst)
1351 		goto out_norqst;
1352 	xprt_pin_rqst(rqst);
1353 
1354 	if (credits == 0)
1355 		credits = 1;	/* don't deadlock */
1356 	else if (credits > buf->rb_max_requests)
1357 		credits = buf->rb_max_requests;
1358 	buf->rb_credits = credits;
1359 
1360 	spin_unlock(&xprt->recv_lock);
1361 
1362 	req = rpcr_to_rdmar(rqst);
1363 	req->rl_reply = rep;
1364 	rep->rr_rqst = rqst;
1365 	clear_bit(RPCRDMA_REQ_F_PENDING, &req->rl_flags);
1366 
1367 	trace_xprtrdma_reply(rqst->rq_task, rep, req, credits);
1368 
1369 	rpcrdma_post_recvs(r_xprt, false);
1370 	queue_work(rpcrdma_receive_wq, &rep->rr_work);
1371 	return;
1372 
1373 out_badversion:
1374 	trace_xprtrdma_reply_vers(rep);
1375 	goto repost;
1376 
1377 /* The RPC transaction has already been terminated, or the header
1378  * is corrupt.
1379  */
1380 out_norqst:
1381 	spin_unlock(&xprt->recv_lock);
1382 	trace_xprtrdma_reply_rqst(rep);
1383 	goto repost;
1384 
1385 out_shortreply:
1386 	trace_xprtrdma_reply_short(rep);
1387 
1388 /* If no pending RPC transaction was matched, post a replacement
1389  * receive buffer before returning.
1390  */
1391 repost:
1392 	rpcrdma_post_recvs(r_xprt, false);
1393 out_badstatus:
1394 	rpcrdma_recv_buffer_put(rep);
1395 }
1396