1 // SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
2 /*
3  * Copyright (c) 2016-2018 Oracle. All rights reserved.
4  * Copyright (c) 2014 Open Grid Computing, Inc. All rights reserved.
5  * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
6  *
7  * This software is available to you under a choice of one of two
8  * licenses.  You may choose to be licensed under the terms of the GNU
9  * General Public License (GPL) Version 2, available from the file
10  * COPYING in the main directory of this source tree, or the BSD-type
11  * license below:
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  *
17  *      Redistributions of source code must retain the above copyright
18  *      notice, this list of conditions and the following disclaimer.
19  *
20  *      Redistributions in binary form must reproduce the above
21  *      copyright notice, this list of conditions and the following
22  *      disclaimer in the documentation and/or other materials provided
23  *      with the distribution.
24  *
25  *      Neither the name of the Network Appliance, Inc. nor the names of
26  *      its contributors may be used to endorse or promote products
27  *      derived from this software without specific prior written
28  *      permission.
29  *
30  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
31  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
32  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
33  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
34  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
35  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
36  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
37  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
38  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
39  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
40  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
41  *
42  * Author: Tom Tucker <tom@opengridcomputing.com>
43  */
44 
45 /* Operation
46  *
47  * The main entry point is svc_rdma_recvfrom. This is called from
48  * svc_recv when the transport indicates there is incoming data to
49  * be read. "Data Ready" is signaled when an RDMA Receive completes,
50  * or when a set of RDMA Reads complete.
51  *
52  * An svc_rqst is passed in. This structure contains an array of
53  * free pages (rq_pages) that will contain the incoming RPC message.
54  *
55  * Short messages are moved directly into svc_rqst::rq_arg, and
56  * the RPC Call is ready to be processed by the Upper Layer.
57  * svc_rdma_recvfrom returns the length of the RPC Call message,
58  * completing the reception of the RPC Call.
59  *
60  * However, when an incoming message has Read chunks,
61  * svc_rdma_recvfrom must post RDMA Reads to pull the RPC Call's
62  * data payload from the client. svc_rdma_recvfrom sets up the
63  * RDMA Reads using pages in svc_rqst::rq_pages, which are
64  * transferred to an svc_rdma_recv_ctxt for the duration of the
65  * I/O. svc_rdma_recvfrom then returns zero, since the RPC message
66  * is still not yet ready.
67  *
68  * When the Read chunk payloads have become available on the
69  * server, "Data Ready" is raised again, and svc_recv calls
70  * svc_rdma_recvfrom again. This second call may use a different
71  * svc_rqst than the first one, thus any information that needs
72  * to be preserved across these two calls is kept in an
73  * svc_rdma_recv_ctxt.
74  *
75  * The second call to svc_rdma_recvfrom performs final assembly
76  * of the RPC Call message, using the RDMA Read sink pages kept in
77  * the svc_rdma_recv_ctxt. The xdr_buf is copied from the
78  * svc_rdma_recv_ctxt to the second svc_rqst. The second call returns
79  * the length of the completed RPC Call message.
80  *
81  * Page Management
82  *
83  * Pages under I/O must be transferred from the first svc_rqst to an
84  * svc_rdma_recv_ctxt before the first svc_rdma_recvfrom call returns.
85  *
86  * The first svc_rqst supplies pages for RDMA Reads. These are moved
87  * from rqstp::rq_pages into ctxt::pages. The consumed elements of
88  * the rq_pages array are set to NULL and refilled with the first
89  * svc_rdma_recvfrom call returns.
90  *
91  * During the second svc_rdma_recvfrom call, RDMA Read sink pages
92  * are transferred from the svc_rdma_recv_ctxt to the second svc_rqst.
93  */
94 
95 #include <linux/slab.h>
96 #include <linux/spinlock.h>
97 #include <asm/unaligned.h>
98 #include <rdma/ib_verbs.h>
99 #include <rdma/rdma_cm.h>
100 
101 #include <linux/sunrpc/xdr.h>
102 #include <linux/sunrpc/debug.h>
103 #include <linux/sunrpc/rpc_rdma.h>
104 #include <linux/sunrpc/svc_rdma.h>
105 
106 #include "xprt_rdma.h"
107 #include <trace/events/rpcrdma.h>
108 
109 static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc);
110 
111 static inline struct svc_rdma_recv_ctxt *
112 svc_rdma_next_recv_ctxt(struct list_head *list)
113 {
114 	return list_first_entry_or_null(list, struct svc_rdma_recv_ctxt,
115 					rc_list);
116 }
117 
118 static void svc_rdma_recv_cid_init(struct svcxprt_rdma *rdma,
119 				   struct rpc_rdma_cid *cid)
120 {
121 	cid->ci_queue_id = rdma->sc_rq_cq->res.id;
122 	cid->ci_completion_id = atomic_inc_return(&rdma->sc_completion_ids);
123 }
124 
125 static struct svc_rdma_recv_ctxt *
126 svc_rdma_recv_ctxt_alloc(struct svcxprt_rdma *rdma)
127 {
128 	struct svc_rdma_recv_ctxt *ctxt;
129 	dma_addr_t addr;
130 	void *buffer;
131 
132 	ctxt = kmalloc(sizeof(*ctxt), GFP_KERNEL);
133 	if (!ctxt)
134 		goto fail0;
135 	buffer = kmalloc(rdma->sc_max_req_size, GFP_KERNEL);
136 	if (!buffer)
137 		goto fail1;
138 	addr = ib_dma_map_single(rdma->sc_pd->device, buffer,
139 				 rdma->sc_max_req_size, DMA_FROM_DEVICE);
140 	if (ib_dma_mapping_error(rdma->sc_pd->device, addr))
141 		goto fail2;
142 
143 	svc_rdma_recv_cid_init(rdma, &ctxt->rc_cid);
144 	pcl_init(&ctxt->rc_call_pcl);
145 	pcl_init(&ctxt->rc_read_pcl);
146 	pcl_init(&ctxt->rc_write_pcl);
147 	pcl_init(&ctxt->rc_reply_pcl);
148 
149 	ctxt->rc_recv_wr.next = NULL;
150 	ctxt->rc_recv_wr.wr_cqe = &ctxt->rc_cqe;
151 	ctxt->rc_recv_wr.sg_list = &ctxt->rc_recv_sge;
152 	ctxt->rc_recv_wr.num_sge = 1;
153 	ctxt->rc_cqe.done = svc_rdma_wc_receive;
154 	ctxt->rc_recv_sge.addr = addr;
155 	ctxt->rc_recv_sge.length = rdma->sc_max_req_size;
156 	ctxt->rc_recv_sge.lkey = rdma->sc_pd->local_dma_lkey;
157 	ctxt->rc_recv_buf = buffer;
158 	ctxt->rc_temp = false;
159 	return ctxt;
160 
161 fail2:
162 	kfree(buffer);
163 fail1:
164 	kfree(ctxt);
165 fail0:
166 	return NULL;
167 }
168 
169 static void svc_rdma_recv_ctxt_destroy(struct svcxprt_rdma *rdma,
170 				       struct svc_rdma_recv_ctxt *ctxt)
171 {
172 	ib_dma_unmap_single(rdma->sc_pd->device, ctxt->rc_recv_sge.addr,
173 			    ctxt->rc_recv_sge.length, DMA_FROM_DEVICE);
174 	kfree(ctxt->rc_recv_buf);
175 	kfree(ctxt);
176 }
177 
178 /**
179  * svc_rdma_recv_ctxts_destroy - Release all recv_ctxt's for an xprt
180  * @rdma: svcxprt_rdma being torn down
181  *
182  */
183 void svc_rdma_recv_ctxts_destroy(struct svcxprt_rdma *rdma)
184 {
185 	struct svc_rdma_recv_ctxt *ctxt;
186 	struct llist_node *node;
187 
188 	while ((node = llist_del_first(&rdma->sc_recv_ctxts))) {
189 		ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
190 		svc_rdma_recv_ctxt_destroy(rdma, ctxt);
191 	}
192 }
193 
194 /**
195  * svc_rdma_recv_ctxt_get - Allocate a recv_ctxt
196  * @rdma: controlling svcxprt_rdma
197  *
198  * Returns a recv_ctxt or (rarely) NULL if none are available.
199  */
200 struct svc_rdma_recv_ctxt *svc_rdma_recv_ctxt_get(struct svcxprt_rdma *rdma)
201 {
202 	struct svc_rdma_recv_ctxt *ctxt;
203 	struct llist_node *node;
204 
205 	node = llist_del_first(&rdma->sc_recv_ctxts);
206 	if (!node)
207 		goto out_empty;
208 	ctxt = llist_entry(node, struct svc_rdma_recv_ctxt, rc_node);
209 
210 out:
211 	ctxt->rc_page_count = 0;
212 	return ctxt;
213 
214 out_empty:
215 	ctxt = svc_rdma_recv_ctxt_alloc(rdma);
216 	if (!ctxt)
217 		return NULL;
218 	goto out;
219 }
220 
221 /**
222  * svc_rdma_recv_ctxt_put - Return recv_ctxt to free list
223  * @rdma: controlling svcxprt_rdma
224  * @ctxt: object to return to the free list
225  *
226  */
227 void svc_rdma_recv_ctxt_put(struct svcxprt_rdma *rdma,
228 			    struct svc_rdma_recv_ctxt *ctxt)
229 {
230 	pcl_free(&ctxt->rc_call_pcl);
231 	pcl_free(&ctxt->rc_read_pcl);
232 	pcl_free(&ctxt->rc_write_pcl);
233 	pcl_free(&ctxt->rc_reply_pcl);
234 
235 	if (!ctxt->rc_temp)
236 		llist_add(&ctxt->rc_node, &rdma->sc_recv_ctxts);
237 	else
238 		svc_rdma_recv_ctxt_destroy(rdma, ctxt);
239 }
240 
241 /**
242  * svc_rdma_release_rqst - Release transport-specific per-rqst resources
243  * @rqstp: svc_rqst being released
244  *
245  * Ensure that the recv_ctxt is released whether or not a Reply
246  * was sent. For example, the client could close the connection,
247  * or svc_process could drop an RPC, before the Reply is sent.
248  */
249 void svc_rdma_release_rqst(struct svc_rqst *rqstp)
250 {
251 	struct svc_rdma_recv_ctxt *ctxt = rqstp->rq_xprt_ctxt;
252 	struct svc_xprt *xprt = rqstp->rq_xprt;
253 	struct svcxprt_rdma *rdma =
254 		container_of(xprt, struct svcxprt_rdma, sc_xprt);
255 
256 	rqstp->rq_xprt_ctxt = NULL;
257 	if (ctxt)
258 		svc_rdma_recv_ctxt_put(rdma, ctxt);
259 }
260 
261 static bool svc_rdma_refresh_recvs(struct svcxprt_rdma *rdma,
262 				   unsigned int wanted, bool temp)
263 {
264 	const struct ib_recv_wr *bad_wr = NULL;
265 	struct svc_rdma_recv_ctxt *ctxt;
266 	struct ib_recv_wr *recv_chain;
267 	int ret;
268 
269 	if (test_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags))
270 		return false;
271 
272 	recv_chain = NULL;
273 	while (wanted--) {
274 		ctxt = svc_rdma_recv_ctxt_get(rdma);
275 		if (!ctxt)
276 			break;
277 
278 		trace_svcrdma_post_recv(ctxt);
279 		ctxt->rc_temp = temp;
280 		ctxt->rc_recv_wr.next = recv_chain;
281 		recv_chain = &ctxt->rc_recv_wr;
282 		rdma->sc_pending_recvs++;
283 	}
284 	if (!recv_chain)
285 		return false;
286 
287 	ret = ib_post_recv(rdma->sc_qp, recv_chain, &bad_wr);
288 	if (ret)
289 		goto err_free;
290 	return true;
291 
292 err_free:
293 	trace_svcrdma_rq_post_err(rdma, ret);
294 	while (bad_wr) {
295 		ctxt = container_of(bad_wr, struct svc_rdma_recv_ctxt,
296 				    rc_recv_wr);
297 		bad_wr = bad_wr->next;
298 		svc_rdma_recv_ctxt_put(rdma, ctxt);
299 	}
300 	/* Since we're destroying the xprt, no need to reset
301 	 * sc_pending_recvs. */
302 	return false;
303 }
304 
305 /**
306  * svc_rdma_post_recvs - Post initial set of Recv WRs
307  * @rdma: fresh svcxprt_rdma
308  *
309  * Returns true if successful, otherwise false.
310  */
311 bool svc_rdma_post_recvs(struct svcxprt_rdma *rdma)
312 {
313 	return svc_rdma_refresh_recvs(rdma, rdma->sc_max_requests, true);
314 }
315 
316 /**
317  * svc_rdma_wc_receive - Invoked by RDMA provider for each polled Receive WC
318  * @cq: Completion Queue context
319  * @wc: Work Completion object
320  *
321  */
322 static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
323 {
324 	struct svcxprt_rdma *rdma = cq->cq_context;
325 	struct ib_cqe *cqe = wc->wr_cqe;
326 	struct svc_rdma_recv_ctxt *ctxt;
327 
328 	rdma->sc_pending_recvs--;
329 
330 	/* WARNING: Only wc->wr_cqe and wc->status are reliable */
331 	ctxt = container_of(cqe, struct svc_rdma_recv_ctxt, rc_cqe);
332 
333 	if (wc->status != IB_WC_SUCCESS)
334 		goto flushed;
335 	trace_svcrdma_wc_recv(wc, &ctxt->rc_cid);
336 
337 	/* If receive posting fails, the connection is about to be
338 	 * lost anyway. The server will not be able to send a reply
339 	 * for this RPC, and the client will retransmit this RPC
340 	 * anyway when it reconnects.
341 	 *
342 	 * Therefore we drop the Receive, even if status was SUCCESS
343 	 * to reduce the likelihood of replayed requests once the
344 	 * client reconnects.
345 	 */
346 	if (rdma->sc_pending_recvs < rdma->sc_max_requests)
347 		if (!svc_rdma_refresh_recvs(rdma, rdma->sc_recv_batch, false))
348 			goto dropped;
349 
350 	/* All wc fields are now known to be valid */
351 	ctxt->rc_byte_len = wc->byte_len;
352 
353 	spin_lock(&rdma->sc_rq_dto_lock);
354 	list_add_tail(&ctxt->rc_list, &rdma->sc_rq_dto_q);
355 	/* Note the unlock pairs with the smp_rmb in svc_xprt_ready: */
356 	set_bit(XPT_DATA, &rdma->sc_xprt.xpt_flags);
357 	spin_unlock(&rdma->sc_rq_dto_lock);
358 	if (!test_bit(RDMAXPRT_CONN_PENDING, &rdma->sc_flags))
359 		svc_xprt_enqueue(&rdma->sc_xprt);
360 	return;
361 
362 flushed:
363 	if (wc->status == IB_WC_WR_FLUSH_ERR)
364 		trace_svcrdma_wc_recv_flush(wc, &ctxt->rc_cid);
365 	else
366 		trace_svcrdma_wc_recv_err(wc, &ctxt->rc_cid);
367 dropped:
368 	svc_rdma_recv_ctxt_put(rdma, ctxt);
369 	svc_xprt_deferred_close(&rdma->sc_xprt);
370 }
371 
372 /**
373  * svc_rdma_flush_recv_queues - Drain pending Receive work
374  * @rdma: svcxprt_rdma being shut down
375  *
376  */
377 void svc_rdma_flush_recv_queues(struct svcxprt_rdma *rdma)
378 {
379 	struct svc_rdma_recv_ctxt *ctxt;
380 
381 	while ((ctxt = svc_rdma_next_recv_ctxt(&rdma->sc_rq_dto_q))) {
382 		list_del(&ctxt->rc_list);
383 		svc_rdma_recv_ctxt_put(rdma, ctxt);
384 	}
385 }
386 
387 static void svc_rdma_build_arg_xdr(struct svc_rqst *rqstp,
388 				   struct svc_rdma_recv_ctxt *ctxt)
389 {
390 	struct xdr_buf *arg = &rqstp->rq_arg;
391 
392 	arg->head[0].iov_base = ctxt->rc_recv_buf;
393 	arg->head[0].iov_len = ctxt->rc_byte_len;
394 	arg->tail[0].iov_base = NULL;
395 	arg->tail[0].iov_len = 0;
396 	arg->page_len = 0;
397 	arg->page_base = 0;
398 	arg->buflen = ctxt->rc_byte_len;
399 	arg->len = ctxt->rc_byte_len;
400 }
401 
402 /**
403  * xdr_count_read_segments - Count number of Read segments in Read list
404  * @rctxt: Ingress receive context
405  * @p: Start of an un-decoded Read list
406  *
407  * Before allocating anything, ensure the ingress Read list is safe
408  * to use.
409  *
410  * The segment count is limited to how many segments can fit in the
411  * transport header without overflowing the buffer. That's about 40
412  * Read segments for a 1KB inline threshold.
413  *
414  * Return values:
415  *   %true: Read list is valid. @rctxt's xdr_stream is updated to point
416  *	    to the first byte past the Read list. rc_read_pcl and
417  *	    rc_call_pcl cl_count fields are set to the number of
418  *	    Read segments in the list.
419  *  %false: Read list is corrupt. @rctxt's xdr_stream is left in an
420  *	    unknown state.
421  */
422 static bool xdr_count_read_segments(struct svc_rdma_recv_ctxt *rctxt, __be32 *p)
423 {
424 	rctxt->rc_call_pcl.cl_count = 0;
425 	rctxt->rc_read_pcl.cl_count = 0;
426 	while (xdr_item_is_present(p)) {
427 		u32 position, handle, length;
428 		u64 offset;
429 
430 		p = xdr_inline_decode(&rctxt->rc_stream,
431 				      rpcrdma_readseg_maxsz * sizeof(*p));
432 		if (!p)
433 			return false;
434 
435 		xdr_decode_read_segment(p, &position, &handle,
436 					    &length, &offset);
437 		if (position) {
438 			if (position & 3)
439 				return false;
440 			++rctxt->rc_read_pcl.cl_count;
441 		} else {
442 			++rctxt->rc_call_pcl.cl_count;
443 		}
444 
445 		p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
446 		if (!p)
447 			return false;
448 	}
449 	return true;
450 }
451 
452 /* Sanity check the Read list.
453  *
454  * Sanity checks:
455  * - Read list does not overflow Receive buffer.
456  * - Chunk size limited by largest NFS data payload.
457  *
458  * Return values:
459  *   %true: Read list is valid. @rctxt's xdr_stream is updated
460  *	    to point to the first byte past the Read list.
461  *  %false: Read list is corrupt. @rctxt's xdr_stream is left
462  *	    in an unknown state.
463  */
464 static bool xdr_check_read_list(struct svc_rdma_recv_ctxt *rctxt)
465 {
466 	__be32 *p;
467 
468 	p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
469 	if (!p)
470 		return false;
471 	if (!xdr_count_read_segments(rctxt, p))
472 		return false;
473 	if (!pcl_alloc_call(rctxt, p))
474 		return false;
475 	return pcl_alloc_read(rctxt, p);
476 }
477 
478 static bool xdr_check_write_chunk(struct svc_rdma_recv_ctxt *rctxt)
479 {
480 	u32 segcount;
481 	__be32 *p;
482 
483 	if (xdr_stream_decode_u32(&rctxt->rc_stream, &segcount))
484 		return false;
485 
486 	/* A bogus segcount causes this buffer overflow check to fail. */
487 	p = xdr_inline_decode(&rctxt->rc_stream,
488 			      segcount * rpcrdma_segment_maxsz * sizeof(*p));
489 	return p != NULL;
490 }
491 
492 /**
493  * xdr_count_write_chunks - Count number of Write chunks in Write list
494  * @rctxt: Received header and decoding state
495  * @p: start of an un-decoded Write list
496  *
497  * Before allocating anything, ensure the ingress Write list is
498  * safe to use.
499  *
500  * Return values:
501  *       %true: Write list is valid. @rctxt's xdr_stream is updated
502  *		to point to the first byte past the Write list, and
503  *		the number of Write chunks is in rc_write_pcl.cl_count.
504  *      %false: Write list is corrupt. @rctxt's xdr_stream is left
505  *		in an indeterminate state.
506  */
507 static bool xdr_count_write_chunks(struct svc_rdma_recv_ctxt *rctxt, __be32 *p)
508 {
509 	rctxt->rc_write_pcl.cl_count = 0;
510 	while (xdr_item_is_present(p)) {
511 		if (!xdr_check_write_chunk(rctxt))
512 			return false;
513 		++rctxt->rc_write_pcl.cl_count;
514 		p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
515 		if (!p)
516 			return false;
517 	}
518 	return true;
519 }
520 
521 /* Sanity check the Write list.
522  *
523  * Implementation limits:
524  * - This implementation currently supports only one Write chunk.
525  *
526  * Sanity checks:
527  * - Write list does not overflow Receive buffer.
528  * - Chunk size limited by largest NFS data payload.
529  *
530  * Return values:
531  *       %true: Write list is valid. @rctxt's xdr_stream is updated
532  *		to point to the first byte past the Write list.
533  *      %false: Write list is corrupt. @rctxt's xdr_stream is left
534  *		in an unknown state.
535  */
536 static bool xdr_check_write_list(struct svc_rdma_recv_ctxt *rctxt)
537 {
538 	__be32 *p;
539 
540 	p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
541 	if (!p)
542 		return false;
543 	if (!xdr_count_write_chunks(rctxt, p))
544 		return false;
545 	if (!pcl_alloc_write(rctxt, &rctxt->rc_write_pcl, p))
546 		return false;
547 
548 	rctxt->rc_cur_result_payload = pcl_first_chunk(&rctxt->rc_write_pcl);
549 	return true;
550 }
551 
552 /* Sanity check the Reply chunk.
553  *
554  * Sanity checks:
555  * - Reply chunk does not overflow Receive buffer.
556  * - Chunk size limited by largest NFS data payload.
557  *
558  * Return values:
559  *       %true: Reply chunk is valid. @rctxt's xdr_stream is updated
560  *		to point to the first byte past the Reply chunk.
561  *      %false: Reply chunk is corrupt. @rctxt's xdr_stream is left
562  *		in an unknown state.
563  */
564 static bool xdr_check_reply_chunk(struct svc_rdma_recv_ctxt *rctxt)
565 {
566 	__be32 *p;
567 
568 	p = xdr_inline_decode(&rctxt->rc_stream, sizeof(*p));
569 	if (!p)
570 		return false;
571 
572 	if (!xdr_item_is_present(p))
573 		return true;
574 	if (!xdr_check_write_chunk(rctxt))
575 		return false;
576 
577 	rctxt->rc_reply_pcl.cl_count = 1;
578 	return pcl_alloc_write(rctxt, &rctxt->rc_reply_pcl, p);
579 }
580 
581 /* RPC-over-RDMA Version One private extension: Remote Invalidation.
582  * Responder's choice: requester signals it can handle Send With
583  * Invalidate, and responder chooses one R_key to invalidate.
584  *
585  * If there is exactly one distinct R_key in the received transport
586  * header, set rc_inv_rkey to that R_key. Otherwise, set it to zero.
587  */
588 static void svc_rdma_get_inv_rkey(struct svcxprt_rdma *rdma,
589 				  struct svc_rdma_recv_ctxt *ctxt)
590 {
591 	struct svc_rdma_segment *segment;
592 	struct svc_rdma_chunk *chunk;
593 	u32 inv_rkey;
594 
595 	ctxt->rc_inv_rkey = 0;
596 
597 	if (!rdma->sc_snd_w_inv)
598 		return;
599 
600 	inv_rkey = 0;
601 	pcl_for_each_chunk(chunk, &ctxt->rc_call_pcl) {
602 		pcl_for_each_segment(segment, chunk) {
603 			if (inv_rkey == 0)
604 				inv_rkey = segment->rs_handle;
605 			else if (inv_rkey != segment->rs_handle)
606 				return;
607 		}
608 	}
609 	pcl_for_each_chunk(chunk, &ctxt->rc_read_pcl) {
610 		pcl_for_each_segment(segment, chunk) {
611 			if (inv_rkey == 0)
612 				inv_rkey = segment->rs_handle;
613 			else if (inv_rkey != segment->rs_handle)
614 				return;
615 		}
616 	}
617 	pcl_for_each_chunk(chunk, &ctxt->rc_write_pcl) {
618 		pcl_for_each_segment(segment, chunk) {
619 			if (inv_rkey == 0)
620 				inv_rkey = segment->rs_handle;
621 			else if (inv_rkey != segment->rs_handle)
622 				return;
623 		}
624 	}
625 	pcl_for_each_chunk(chunk, &ctxt->rc_reply_pcl) {
626 		pcl_for_each_segment(segment, chunk) {
627 			if (inv_rkey == 0)
628 				inv_rkey = segment->rs_handle;
629 			else if (inv_rkey != segment->rs_handle)
630 				return;
631 		}
632 	}
633 	ctxt->rc_inv_rkey = inv_rkey;
634 }
635 
636 /**
637  * svc_rdma_xdr_decode_req - Decode the transport header
638  * @rq_arg: xdr_buf containing ingress RPC/RDMA message
639  * @rctxt: state of decoding
640  *
641  * On entry, xdr->head[0].iov_base points to first byte of the
642  * RPC-over-RDMA transport header.
643  *
644  * On successful exit, head[0] points to first byte past the
645  * RPC-over-RDMA header. For RDMA_MSG, this is the RPC message.
646  *
647  * The length of the RPC-over-RDMA header is returned.
648  *
649  * Assumptions:
650  * - The transport header is entirely contained in the head iovec.
651  */
652 static int svc_rdma_xdr_decode_req(struct xdr_buf *rq_arg,
653 				   struct svc_rdma_recv_ctxt *rctxt)
654 {
655 	__be32 *p, *rdma_argp;
656 	unsigned int hdr_len;
657 
658 	rdma_argp = rq_arg->head[0].iov_base;
659 	xdr_init_decode(&rctxt->rc_stream, rq_arg, rdma_argp, NULL);
660 
661 	p = xdr_inline_decode(&rctxt->rc_stream,
662 			      rpcrdma_fixed_maxsz * sizeof(*p));
663 	if (unlikely(!p))
664 		goto out_short;
665 	p++;
666 	if (*p != rpcrdma_version)
667 		goto out_version;
668 	p += 2;
669 	rctxt->rc_msgtype = *p;
670 	switch (rctxt->rc_msgtype) {
671 	case rdma_msg:
672 		break;
673 	case rdma_nomsg:
674 		break;
675 	case rdma_done:
676 		goto out_drop;
677 	case rdma_error:
678 		goto out_drop;
679 	default:
680 		goto out_proc;
681 	}
682 
683 	if (!xdr_check_read_list(rctxt))
684 		goto out_inval;
685 	if (!xdr_check_write_list(rctxt))
686 		goto out_inval;
687 	if (!xdr_check_reply_chunk(rctxt))
688 		goto out_inval;
689 
690 	rq_arg->head[0].iov_base = rctxt->rc_stream.p;
691 	hdr_len = xdr_stream_pos(&rctxt->rc_stream);
692 	rq_arg->head[0].iov_len -= hdr_len;
693 	rq_arg->len -= hdr_len;
694 	trace_svcrdma_decode_rqst(rctxt, rdma_argp, hdr_len);
695 	return hdr_len;
696 
697 out_short:
698 	trace_svcrdma_decode_short_err(rctxt, rq_arg->len);
699 	return -EINVAL;
700 
701 out_version:
702 	trace_svcrdma_decode_badvers_err(rctxt, rdma_argp);
703 	return -EPROTONOSUPPORT;
704 
705 out_drop:
706 	trace_svcrdma_decode_drop_err(rctxt, rdma_argp);
707 	return 0;
708 
709 out_proc:
710 	trace_svcrdma_decode_badproc_err(rctxt, rdma_argp);
711 	return -EINVAL;
712 
713 out_inval:
714 	trace_svcrdma_decode_parse_err(rctxt, rdma_argp);
715 	return -EINVAL;
716 }
717 
718 static void svc_rdma_send_error(struct svcxprt_rdma *rdma,
719 				struct svc_rdma_recv_ctxt *rctxt,
720 				int status)
721 {
722 	struct svc_rdma_send_ctxt *sctxt;
723 
724 	sctxt = svc_rdma_send_ctxt_get(rdma);
725 	if (!sctxt)
726 		return;
727 	svc_rdma_send_error_msg(rdma, sctxt, rctxt, status);
728 }
729 
730 /* By convention, backchannel calls arrive via rdma_msg type
731  * messages, and never populate the chunk lists. This makes
732  * the RPC/RDMA header small and fixed in size, so it is
733  * straightforward to check the RPC header's direction field.
734  */
735 static bool svc_rdma_is_reverse_direction_reply(struct svc_xprt *xprt,
736 						struct svc_rdma_recv_ctxt *rctxt)
737 {
738 	__be32 *p = rctxt->rc_recv_buf;
739 
740 	if (!xprt->xpt_bc_xprt)
741 		return false;
742 
743 	if (rctxt->rc_msgtype != rdma_msg)
744 		return false;
745 
746 	if (!pcl_is_empty(&rctxt->rc_call_pcl))
747 		return false;
748 	if (!pcl_is_empty(&rctxt->rc_read_pcl))
749 		return false;
750 	if (!pcl_is_empty(&rctxt->rc_write_pcl))
751 		return false;
752 	if (!pcl_is_empty(&rctxt->rc_reply_pcl))
753 		return false;
754 
755 	/* RPC call direction */
756 	if (*(p + 8) == cpu_to_be32(RPC_CALL))
757 		return false;
758 
759 	return true;
760 }
761 
762 /**
763  * svc_rdma_recvfrom - Receive an RPC call
764  * @rqstp: request structure into which to receive an RPC Call
765  *
766  * Returns:
767  *	The positive number of bytes in the RPC Call message,
768  *	%0 if there were no Calls ready to return,
769  *	%-EINVAL if the Read chunk data is too large,
770  *	%-ENOMEM if rdma_rw context pool was exhausted,
771  *	%-ENOTCONN if posting failed (connection is lost),
772  *	%-EIO if rdma_rw initialization failed (DMA mapping, etc).
773  *
774  * Called in a loop when XPT_DATA is set. XPT_DATA is cleared only
775  * when there are no remaining ctxt's to process.
776  *
777  * The next ctxt is removed from the "receive" lists.
778  *
779  * - If the ctxt completes a Read, then finish assembling the Call
780  *   message and return the number of bytes in the message.
781  *
782  * - If the ctxt completes a Receive, then construct the Call
783  *   message from the contents of the Receive buffer.
784  *
785  *   - If there are no Read chunks in this message, then finish
786  *     assembling the Call message and return the number of bytes
787  *     in the message.
788  *
789  *   - If there are Read chunks in this message, post Read WRs to
790  *     pull that payload and return 0.
791  */
792 int svc_rdma_recvfrom(struct svc_rqst *rqstp)
793 {
794 	struct svc_xprt *xprt = rqstp->rq_xprt;
795 	struct svcxprt_rdma *rdma_xprt =
796 		container_of(xprt, struct svcxprt_rdma, sc_xprt);
797 	struct svc_rdma_recv_ctxt *ctxt;
798 	int ret;
799 
800 	rqstp->rq_xprt_ctxt = NULL;
801 
802 	ctxt = NULL;
803 	spin_lock(&rdma_xprt->sc_rq_dto_lock);
804 	ctxt = svc_rdma_next_recv_ctxt(&rdma_xprt->sc_rq_dto_q);
805 	if (ctxt)
806 		list_del(&ctxt->rc_list);
807 	else
808 		/* No new incoming requests, terminate the loop */
809 		clear_bit(XPT_DATA, &xprt->xpt_flags);
810 	spin_unlock(&rdma_xprt->sc_rq_dto_lock);
811 
812 	/* Unblock the transport for the next receive */
813 	svc_xprt_received(xprt);
814 	if (!ctxt)
815 		return 0;
816 
817 	percpu_counter_inc(&svcrdma_stat_recv);
818 	ib_dma_sync_single_for_cpu(rdma_xprt->sc_pd->device,
819 				   ctxt->rc_recv_sge.addr, ctxt->rc_byte_len,
820 				   DMA_FROM_DEVICE);
821 	svc_rdma_build_arg_xdr(rqstp, ctxt);
822 
823 	/* Prevent svc_xprt_release from releasing pages in rq_pages
824 	 * if we return 0 or an error.
825 	 */
826 	rqstp->rq_respages = rqstp->rq_pages;
827 	rqstp->rq_next_page = rqstp->rq_respages;
828 
829 	ret = svc_rdma_xdr_decode_req(&rqstp->rq_arg, ctxt);
830 	if (ret < 0)
831 		goto out_err;
832 	if (ret == 0)
833 		goto out_drop;
834 
835 	if (svc_rdma_is_reverse_direction_reply(xprt, ctxt))
836 		goto out_backchannel;
837 
838 	svc_rdma_get_inv_rkey(rdma_xprt, ctxt);
839 
840 	if (!pcl_is_empty(&ctxt->rc_read_pcl) ||
841 	    !pcl_is_empty(&ctxt->rc_call_pcl)) {
842 		ret = svc_rdma_process_read_list(rdma_xprt, rqstp, ctxt);
843 		if (ret < 0)
844 			goto out_readfail;
845 	}
846 
847 	rqstp->rq_xprt_ctxt = ctxt;
848 	rqstp->rq_prot = IPPROTO_MAX;
849 	svc_xprt_copy_addrs(rqstp, xprt);
850 	set_bit(RQ_SECURE, &rqstp->rq_flags);
851 	return rqstp->rq_arg.len;
852 
853 out_err:
854 	svc_rdma_send_error(rdma_xprt, ctxt, ret);
855 	svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
856 	return 0;
857 
858 out_readfail:
859 	if (ret == -EINVAL)
860 		svc_rdma_send_error(rdma_xprt, ctxt, ret);
861 	svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
862 	return ret;
863 
864 out_backchannel:
865 	svc_rdma_handle_bc_reply(rqstp, ctxt);
866 out_drop:
867 	svc_rdma_recv_ctxt_put(rdma_xprt, ctxt);
868 	return 0;
869 }
870