verbs.c 38.1 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
2
/*
C
Chuck Lever 已提交
3
 * Copyright (c) 2014-2017 Oracle.  All rights reserved.
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
 * Copyright (c) 2003-2007 Network Appliance, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the BSD-type
 * license below:
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 *      Redistributions of source code must retain the above copyright
 *      notice, this list of conditions and the following disclaimer.
 *
 *      Redistributions in binary form must reproduce the above
 *      copyright notice, this list of conditions and the following
 *      disclaimer in the documentation and/or other materials provided
 *      with the distribution.
 *
 *      Neither the name of the Network Appliance, Inc. nor the names of
 *      its contributors may be used to endorse or promote products
 *      derived from this software without specific prior written
 *      permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
40 41
 */

42 43 44 45 46 47 48 49 50 51
/*
 * verbs.c
 *
 * Encapsulates the major functions managing:
 *  o adapters
 *  o endpoints
 *  o connections
 *  o buffer memory
 */

52
#include <linux/interrupt.h>
53
#include <linux/slab.h>
54
#include <linux/sunrpc/addr.h>
55
#include <linux/sunrpc/svc_rdma.h>
56 57

#include <asm-generic/barrier.h>
58
#include <asm/bitops.h>
59

60
#include <rdma/ib_cm.h>
61

62
#include "xprt_rdma.h"
63
#include <trace/events/rpcrdma.h>
64

65 66 67 68
/*
 * Globals/Macros
 */

J
Jeff Layton 已提交
69
#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
70 71 72 73 74 75
# define RPCDBG_FACILITY	RPCDBG_TRANS
#endif

/*
 * internal functions
 */
76
static void rpcrdma_sendctx_put_locked(struct rpcrdma_sendctx *sc);
C
Chuck Lever 已提交
77 78
static void rpcrdma_mrs_create(struct rpcrdma_xprt *r_xprt);
static void rpcrdma_mrs_destroy(struct rpcrdma_buffer *buf);
79
static int rpcrdma_create_rep(struct rpcrdma_xprt *r_xprt, bool temp);
80
static void rpcrdma_dma_unmap_regbuf(struct rpcrdma_regbuf *rb);
81

82
struct workqueue_struct *rpcrdma_receive_wq __read_mostly;
83

84 85
int
rpcrdma_alloc_wq(void)
86
{
87
	struct workqueue_struct *recv_wq;
88

89
	recv_wq = alloc_workqueue("xprtrdma_receive",
90
				  WQ_MEM_RECLAIM | WQ_HIGHPRI,
91 92 93
				  0);
	if (!recv_wq)
		return -ENOMEM;
94

95 96
	rpcrdma_receive_wq = recv_wq;
	return 0;
97 98
}

99 100
void
rpcrdma_destroy_wq(void)
101
{
102
	struct workqueue_struct *wq;
103

104 105 106 107 108
	if (rpcrdma_receive_wq) {
		wq = rpcrdma_receive_wq;
		rpcrdma_receive_wq = NULL;
		destroy_workqueue(wq);
	}
109 110
}

111 112 113 114
static void
rpcrdma_qp_async_error_upcall(struct ib_event *event, void *context)
{
	struct rpcrdma_ep *ep = context;
115 116
	struct rpcrdma_xprt *r_xprt = container_of(ep, struct rpcrdma_xprt,
						   rx_ep);
117

118
	trace_xprtrdma_qp_error(r_xprt, event);
119 120 121
	pr_err("rpcrdma: %s on device %s ep %p\n",
	       ib_event_msg(event->event), event->device->name, context);

122 123
	if (ep->rep_connected == 1) {
		ep->rep_connected = -EIO;
124
		rpcrdma_conn_func(ep);
125 126 127 128
		wake_up_all(&ep->rep_connect_wait);
	}
}

129 130 131 132 133
/**
 * rpcrdma_wc_send - Invoked by RDMA provider for each polled Send WC
 * @cq:	completion queue (ignored)
 * @wc:	completed WR
 *
134 135
 */
static void
136
rpcrdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
137
{
138 139 140 141
	struct ib_cqe *cqe = wc->wr_cqe;
	struct rpcrdma_sendctx *sc =
		container_of(cqe, struct rpcrdma_sendctx, sc_cqe);

142
	/* WARNING: Only wr_cqe and status are reliable at this point */
143
	trace_xprtrdma_wc_send(sc, wc);
144 145 146 147
	if (wc->status != IB_WC_SUCCESS && wc->status != IB_WC_WR_FLUSH_ERR)
		pr_err("rpcrdma: Send: %s (%u/0x%x)\n",
		       ib_wc_status_msg(wc->status),
		       wc->status, wc->vendor_err);
148 149

	rpcrdma_sendctx_put_locked(sc);
150
}
151

152
/**
153
 * rpcrdma_wc_receive - Invoked by RDMA provider for each polled Receive WC
154 155 156 157
 * @cq:	completion queue (ignored)
 * @wc:	completed WR
 *
 */
158
static void
159
rpcrdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
160
{
161 162 163
	struct ib_cqe *cqe = wc->wr_cqe;
	struct rpcrdma_rep *rep = container_of(cqe, struct rpcrdma_rep,
					       rr_cqe);
164

165
	/* WARNING: Only wr_id and status are reliable at this point */
166
	trace_xprtrdma_wc_receive(wc);
167 168
	if (wc->status != IB_WC_SUCCESS)
		goto out_fail;
169

170
	/* status == SUCCESS means all fields in wc are trustworthy */
171
	rpcrdma_set_xdrlen(&rep->rr_hdrbuf, wc->byte_len);
172 173 174
	rep->rr_wc_flags = wc->wc_flags;
	rep->rr_inv_rkey = wc->ex.invalidate_rkey;

175
	ib_dma_sync_single_for_cpu(rdmab_device(rep->rr_rdmabuf),
176
				   rdmab_addr(rep->rr_rdmabuf),
177
				   wc->byte_len, DMA_FROM_DEVICE);
178

179
out_schedule:
180
	rpcrdma_reply_handler(rep);
181
	return;
182

183 184
out_fail:
	if (wc->status != IB_WC_WR_FLUSH_ERR)
185 186 187
		pr_err("rpcrdma: Recv: %s (%u/0x%x)\n",
		       ib_wc_status_msg(wc->status),
		       wc->status, wc->vendor_err);
188
	rpcrdma_set_xdrlen(&rep->rr_hdrbuf, 0);
189
	goto out_schedule;
190 191
}

192 193 194 195 196 197 198 199
static void
rpcrdma_update_connect_private(struct rpcrdma_xprt *r_xprt,
			       struct rdma_conn_param *param)
{
	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
	const struct rpcrdma_connect_private *pmsg = param->private_data;
	unsigned int rsize, wsize;

200
	/* Default settings for RPC-over-RDMA Version One */
201
	r_xprt->rx_ia.ri_implicit_roundup = xprt_rdma_pad_optimize;
202 203 204 205 206 207
	rsize = RPCRDMA_V1_DEF_INLINE_SIZE;
	wsize = RPCRDMA_V1_DEF_INLINE_SIZE;

	if (pmsg &&
	    pmsg->cp_magic == rpcrdma_cmp_magic &&
	    pmsg->cp_version == RPCRDMA_CMP_VERSION) {
208
		r_xprt->rx_ia.ri_implicit_roundup = true;
209 210 211 212 213 214 215 216
		rsize = rpcrdma_decode_buffer_size(pmsg->cp_send_size);
		wsize = rpcrdma_decode_buffer_size(pmsg->cp_recv_size);
	}

	if (rsize < cdata->inline_rsize)
		cdata->inline_rsize = rsize;
	if (wsize < cdata->inline_wsize)
		cdata->inline_wsize = wsize;
217 218
	dprintk("RPC:       %s: max send %u, max recv %u\n",
		__func__, cdata->inline_wsize, cdata->inline_rsize);
219 220 221
	rpcrdma_set_max_header_sizes(r_xprt);
}

C
Chuck Lever 已提交
222 223 224 225 226 227 228 229
/**
 * rpcrdma_cm_event_handler - Handle RDMA CM events
 * @id: rdma_cm_id on which an event has occurred
 * @event: details of the event
 *
 * Called with @id's mutex held. Returns 1 if caller should
 * destroy @id, otherwise 0.
 */
230
static int
C
Chuck Lever 已提交
231
rpcrdma_cm_event_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
232 233 234 235 236 237
{
	struct rpcrdma_xprt *xprt = id->context;
	struct rpcrdma_ia *ia = &xprt->rx_ia;
	struct rpcrdma_ep *ep = &xprt->rx_ep;
	int connstate = 0;

C
Chuck Lever 已提交
238 239 240
	might_sleep();

	trace_xprtrdma_cm_event(xprt, event);
241 242 243
	switch (event->event) {
	case RDMA_CM_EVENT_ADDR_RESOLVED:
	case RDMA_CM_EVENT_ROUTE_RESOLVED:
244
		ia->ri_async_rc = 0;
245 246 247
		complete(&ia->ri_done);
		break;
	case RDMA_CM_EVENT_ADDR_ERROR:
248
		ia->ri_async_rc = -EPROTO;
249 250 251 252 253 254
		complete(&ia->ri_done);
		break;
	case RDMA_CM_EVENT_ROUTE_ERROR:
		ia->ri_async_rc = -ENETUNREACH;
		complete(&ia->ri_done);
		break;
255 256
	case RDMA_CM_EVENT_DEVICE_REMOVAL:
#if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
257
		pr_info("rpcrdma: removing device %s for %s:%s\n",
258
			ia->ri_device->name,
259
			rpcrdma_addrstr(xprt), rpcrdma_portstr(xprt));
260 261 262 263 264 265 266 267 268 269
#endif
		set_bit(RPCRDMA_IAF_REMOVING, &ia->ri_flags);
		ep->rep_connected = -ENODEV;
		xprt_force_disconnect(&xprt->rx_xprt);
		wait_for_completion(&ia->ri_remove_done);

		ia->ri_id = NULL;
		ia->ri_device = NULL;
		/* Return 1 to ensure the core destroys the id. */
		return 1;
270
	case RDMA_CM_EVENT_ESTABLISHED:
271
		++xprt->rx_xprt.connect_cookie;
272
		connstate = 1;
273
		rpcrdma_update_connect_private(xprt, &event->param.conn);
274 275 276 277 278
		goto connected;
	case RDMA_CM_EVENT_CONNECT_ERROR:
		connstate = -ENOTCONN;
		goto connected;
	case RDMA_CM_EVENT_UNREACHABLE:
279
		connstate = -ENETUNREACH;
280 281
		goto connected;
	case RDMA_CM_EVENT_REJECTED:
282 283
		dprintk("rpcrdma: connection to %s:%s rejected: %s\n",
			rpcrdma_addrstr(xprt), rpcrdma_portstr(xprt),
284
			rdma_reject_msg(id, event->status));
285
		connstate = -ECONNREFUSED;
286 287
		if (event->status == IB_CM_REJ_STALE_CONN)
			connstate = -EAGAIN;
288 289
		goto connected;
	case RDMA_CM_EVENT_DISCONNECTED:
290
		++xprt->rx_xprt.connect_cookie;
291 292 293
		connstate = -ECONNABORTED;
connected:
		ep->rep_connected = connstate;
294
		rpcrdma_conn_func(ep);
295
		wake_up_all(&ep->rep_connect_wait);
296
		/*FALLTHROUGH*/
297
	default:
298 299 300
		dprintk("RPC:       %s: %s:%s on %s/%s (ep 0x%p): %s\n",
			__func__,
			rpcrdma_addrstr(xprt), rpcrdma_portstr(xprt),
301 302
			ia->ri_device->name, ia->ri_ops->ro_displayname,
			ep, rdma_event_msg(event->event));
303 304 305 306 307 308 309
		break;
	}

	return 0;
}

static struct rdma_cm_id *
310
rpcrdma_create_id(struct rpcrdma_xprt *xprt, struct rpcrdma_ia *ia)
311
{
312
	unsigned long wtimeout = msecs_to_jiffies(RDMA_RESOLVE_TIMEOUT) + 1;
313 314 315
	struct rdma_cm_id *id;
	int rc;

316 317
	trace_xprtrdma_conn_start(xprt);

318
	init_completion(&ia->ri_done);
319
	init_completion(&ia->ri_remove_done);
320

C
Chuck Lever 已提交
321
	id = rdma_create_id(xprt->rx_xprt.xprt_net, rpcrdma_cm_event_handler,
322
			    xprt, RDMA_PS_TCP, IB_QPT_RC);
323 324 325 326 327 328 329
	if (IS_ERR(id)) {
		rc = PTR_ERR(id);
		dprintk("RPC:       %s: rdma_create_id() failed %i\n",
			__func__, rc);
		return id;
	}

330
	ia->ri_async_rc = -ETIMEDOUT;
331 332 333
	rc = rdma_resolve_addr(id, NULL,
			       (struct sockaddr *)&xprt->rx_xprt.addr,
			       RDMA_RESOLVE_TIMEOUT);
334 335 336 337 338
	if (rc) {
		dprintk("RPC:       %s: rdma_resolve_addr() failed %i\n",
			__func__, rc);
		goto out;
	}
339 340
	rc = wait_for_completion_interruptible_timeout(&ia->ri_done, wtimeout);
	if (rc < 0) {
341
		trace_xprtrdma_conn_tout(xprt);
342 343
		goto out;
	}
344

345 346 347 348
	rc = ia->ri_async_rc;
	if (rc)
		goto out;

349
	ia->ri_async_rc = -ETIMEDOUT;
350 351 352 353
	rc = rdma_resolve_route(id, RDMA_RESOLVE_TIMEOUT);
	if (rc) {
		dprintk("RPC:       %s: rdma_resolve_route() failed %i\n",
			__func__, rc);
354
		goto out;
355
	}
356 357
	rc = wait_for_completion_interruptible_timeout(&ia->ri_done, wtimeout);
	if (rc < 0) {
358
		trace_xprtrdma_conn_tout(xprt);
359
		goto out;
360
	}
361 362
	rc = ia->ri_async_rc;
	if (rc)
363
		goto out;
364 365

	return id;
366

367 368 369 370 371 372 373 374 375
out:
	rdma_destroy_id(id);
	return ERR_PTR(rc);
}

/*
 * Exported functions.
 */

376 377
/**
 * rpcrdma_ia_open - Open and initialize an Interface Adapter.
378
 * @xprt: transport with IA to (re)initialize
379 380 381
 *
 * Returns 0 on success, negative errno if an appropriate
 * Interface Adapter could not be found and opened.
382 383
 */
int
384
rpcrdma_ia_open(struct rpcrdma_xprt *xprt)
385 386
{
	struct rpcrdma_ia *ia = &xprt->rx_ia;
387 388
	int rc;

389
	ia->ri_id = rpcrdma_create_id(xprt, ia);
390 391
	if (IS_ERR(ia->ri_id)) {
		rc = PTR_ERR(ia->ri_id);
392
		goto out_err;
393
	}
394
	ia->ri_device = ia->ri_id->device;
395

396
	ia->ri_pd = ib_alloc_pd(ia->ri_device, 0);
397 398
	if (IS_ERR(ia->ri_pd)) {
		rc = PTR_ERR(ia->ri_pd);
399
		pr_err("rpcrdma: ib_alloc_pd() returned %d\n", rc);
400
		goto out_err;
401 402
	}

403
	switch (xprt_rdma_memreg_strategy) {
404
	case RPCRDMA_FRWR:
405 406 407 408 409
		if (frwr_is_supported(ia)) {
			ia->ri_ops = &rpcrdma_frwr_memreg_ops;
			break;
		}
		/*FALLTHROUGH*/
410
	case RPCRDMA_MTHCAFMR:
411 412 413 414 415
		if (fmr_is_supported(ia)) {
			ia->ri_ops = &rpcrdma_fmr_memreg_ops;
			break;
		}
		/*FALLTHROUGH*/
416
	default:
417 418
		pr_err("rpcrdma: Device %s does not support memreg mode %d\n",
		       ia->ri_device->name, xprt_rdma_memreg_strategy);
419
		rc = -EINVAL;
420
		goto out_err;
421 422 423
	}

	return 0;
424

425 426
out_err:
	rpcrdma_ia_close(ia);
427 428 429
	return rc;
}

430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
/**
 * rpcrdma_ia_remove - Handle device driver unload
 * @ia: interface adapter being removed
 *
 * Divest transport H/W resources associated with this adapter,
 * but allow it to be restored later.
 */
void
rpcrdma_ia_remove(struct rpcrdma_ia *ia)
{
	struct rpcrdma_xprt *r_xprt = container_of(ia, struct rpcrdma_xprt,
						   rx_ia);
	struct rpcrdma_ep *ep = &r_xprt->rx_ep;
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
	struct rpcrdma_req *req;
	struct rpcrdma_rep *rep;

	cancel_delayed_work_sync(&buf->rb_refresh_worker);

	/* This is similar to rpcrdma_ep_destroy, but:
	 * - Don't cancel the connect worker.
	 * - Don't call rpcrdma_ep_disconnect, which waits
	 *   for another conn upcall, which will deadlock.
	 * - rdma_disconnect is unneeded, the underlying
	 *   connection is already gone.
	 */
	if (ia->ri_id->qp) {
		ib_drain_qp(ia->ri_id->qp);
		rdma_destroy_qp(ia->ri_id);
		ia->ri_id->qp = NULL;
	}
	ib_free_cq(ep->rep_attr.recv_cq);
462
	ep->rep_attr.recv_cq = NULL;
463
	ib_free_cq(ep->rep_attr.send_cq);
464
	ep->rep_attr.send_cq = NULL;
465 466 467 468 469 470 471 472 473 474 475

	/* The ULP is responsible for ensuring all DMA
	 * mappings and MRs are gone.
	 */
	list_for_each_entry(rep, &buf->rb_recv_bufs, rr_list)
		rpcrdma_dma_unmap_regbuf(rep->rr_rdmabuf);
	list_for_each_entry(req, &buf->rb_allreqs, rl_all) {
		rpcrdma_dma_unmap_regbuf(req->rl_rdmabuf);
		rpcrdma_dma_unmap_regbuf(req->rl_sendbuf);
		rpcrdma_dma_unmap_regbuf(req->rl_recvbuf);
	}
C
Chuck Lever 已提交
476
	rpcrdma_mrs_destroy(buf);
477 478
	ib_dealloc_pd(ia->ri_pd);
	ia->ri_pd = NULL;
479 480 481

	/* Allow waiters to continue */
	complete(&ia->ri_remove_done);
482 483

	trace_xprtrdma_remove(r_xprt);
484 485
}

486 487 488 489
/**
 * rpcrdma_ia_close - Clean up/close an IA.
 * @ia: interface adapter to close
 *
490 491 492 493
 */
void
rpcrdma_ia_close(struct rpcrdma_ia *ia)
{
494 495 496
	if (ia->ri_id != NULL && !IS_ERR(ia->ri_id)) {
		if (ia->ri_id->qp)
			rdma_destroy_qp(ia->ri_id);
497
		rdma_destroy_id(ia->ri_id);
498
	}
499 500
	ia->ri_id = NULL;
	ia->ri_device = NULL;
501 502 503

	/* If the pd is still busy, xprtrdma missed freeing a resource */
	if (ia->ri_pd && !IS_ERR(ia->ri_pd))
504
		ib_dealloc_pd(ia->ri_pd);
505
	ia->ri_pd = NULL;
506 507 508 509 510 511 512
}

/*
 * Create unconnected endpoint.
 */
int
rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia,
513
		  struct rpcrdma_create_data_internal *cdata)
514
{
515
	struct rpcrdma_connect_private *pmsg = &ep->rep_cm_private;
516
	struct ib_cq *sendcq, *recvcq;
517
	unsigned int max_sge;
518
	int rc;
519

520
	max_sge = min_t(unsigned int, ia->ri_device->attrs.max_send_sge,
521
			RPCRDMA_MAX_SEND_SGES);
522 523
	if (max_sge < RPCRDMA_MIN_SEND_SGES) {
		pr_warn("rpcrdma: HCA provides only %d send SGEs\n", max_sge);
524 525
		return -ENOMEM;
	}
526
	ia->ri_max_send_sges = max_sge;
527

528 529 530
	rc = ia->ri_ops->ro_open(ia, ep, cdata);
	if (rc)
		return rc;
531 532 533 534

	ep->rep_attr.event_handler = rpcrdma_qp_async_error_upcall;
	ep->rep_attr.qp_context = ep;
	ep->rep_attr.srq = NULL;
535
	ep->rep_attr.cap.max_send_sge = max_sge;
536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
	ep->rep_attr.cap.max_recv_sge = 1;
	ep->rep_attr.cap.max_inline_data = 0;
	ep->rep_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
	ep->rep_attr.qp_type = IB_QPT_RC;
	ep->rep_attr.port_num = ~0;

	dprintk("RPC:       %s: requested max: dtos: send %d recv %d; "
		"iovs: send %d recv %d\n",
		__func__,
		ep->rep_attr.cap.max_send_wr,
		ep->rep_attr.cap.max_recv_wr,
		ep->rep_attr.cap.max_send_sge,
		ep->rep_attr.cap.max_recv_sge);

	/* set trigger for requesting send completion */
551 552 553
	ep->rep_send_batch = min_t(unsigned int, RPCRDMA_MAX_SEND_BATCH,
				   cdata->max_requests >> 2);
	ep->rep_send_count = ep->rep_send_batch;
554
	init_waitqueue_head(&ep->rep_connect_wait);
555
	INIT_DELAYED_WORK(&ep->rep_connect_worker, rpcrdma_connect_worker);
556

557 558
	sendcq = ib_alloc_cq(ia->ri_device, NULL,
			     ep->rep_attr.cap.max_send_wr + 1,
559
			     1, IB_POLL_WORKQUEUE);
560 561 562
	if (IS_ERR(sendcq)) {
		rc = PTR_ERR(sendcq);
		dprintk("RPC:       %s: failed to create send CQ: %i\n",
563 564 565 566
			__func__, rc);
		goto out1;
	}

567 568
	recvcq = ib_alloc_cq(ia->ri_device, NULL,
			     ep->rep_attr.cap.max_recv_wr + 1,
569
			     0, IB_POLL_WORKQUEUE);
570 571 572 573 574 575 576 577 578
	if (IS_ERR(recvcq)) {
		rc = PTR_ERR(recvcq);
		dprintk("RPC:       %s: failed to create recv CQ: %i\n",
			__func__, rc);
		goto out2;
	}

	ep->rep_attr.send_cq = sendcq;
	ep->rep_attr.recv_cq = recvcq;
579 580

	/* Initialize cma parameters */
581
	memset(&ep->rep_remote_cma, 0, sizeof(ep->rep_remote_cma));
582

583 584 585
	/* Prepare RDMA-CM private message */
	pmsg->cp_magic = rpcrdma_cmp_magic;
	pmsg->cp_version = RPCRDMA_CMP_VERSION;
586
	pmsg->cp_flags |= ia->ri_ops->ro_send_w_inv_ok;
587 588 589 590
	pmsg->cp_send_size = rpcrdma_encode_buffer_size(cdata->inline_wsize);
	pmsg->cp_recv_size = rpcrdma_encode_buffer_size(cdata->inline_rsize);
	ep->rep_remote_cma.private_data = pmsg;
	ep->rep_remote_cma.private_data_len = sizeof(*pmsg);
591 592

	/* Client offers RDMA Read but does not initiate */
593
	ep->rep_remote_cma.initiator_depth = 0;
594 595
	ep->rep_remote_cma.responder_resources =
		min_t(int, U8_MAX, ia->ri_device->attrs.max_qp_rd_atom);
596

597 598 599 600 601 602 603 604 605 606
	/* Limit transport retries so client can detect server
	 * GID changes quickly. RPC layer handles re-establishing
	 * transport connection and retransmission.
	 */
	ep->rep_remote_cma.retry_count = 6;

	/* RPC-over-RDMA handles its own flow control. In addition,
	 * make all RNR NAKs visible so we know that RPC-over-RDMA
	 * flow control is working correctly (no NAKs should be seen).
	 */
607 608 609 610 611 612
	ep->rep_remote_cma.flow_control = 0;
	ep->rep_remote_cma.rnr_retry_count = 0;

	return 0;

out2:
613
	ib_free_cq(sendcq);
614 615 616 617 618 619 620 621 622 623 624
out1:
	return rc;
}

/*
 * rpcrdma_ep_destroy
 *
 * Disconnect and destroy endpoint. After this, the only
 * valid operations on the ep are to free it (if dynamically
 * allocated) or re-create it.
 */
625
void
626 627
rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
{
628 629
	cancel_delayed_work_sync(&ep->rep_connect_worker);

630
	if (ia->ri_id && ia->ri_id->qp) {
631
		rpcrdma_ep_disconnect(ep, ia);
632 633
		rdma_destroy_qp(ia->ri_id);
		ia->ri_id->qp = NULL;
634 635
	}

636 637 638 639
	if (ep->rep_attr.recv_cq)
		ib_free_cq(ep->rep_attr.recv_cq);
	if (ep->rep_attr.send_cq)
		ib_free_cq(ep->rep_attr.send_cq);
640 641
}

642 643 644 645 646 647 648 649 650 651
/* Re-establish a connection after a device removal event.
 * Unlike a normal reconnection, a fresh PD and a new set
 * of MRs and buffers is needed.
 */
static int
rpcrdma_ep_recreate_xprt(struct rpcrdma_xprt *r_xprt,
			 struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
{
	int rc, err;

652
	trace_xprtrdma_reinsert(r_xprt);
653 654

	rc = -EHOSTUNREACH;
655
	if (rpcrdma_ia_open(r_xprt))
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
		goto out1;

	rc = -ENOMEM;
	err = rpcrdma_ep_create(ep, ia, &r_xprt->rx_data);
	if (err) {
		pr_err("rpcrdma: rpcrdma_ep_create returned %d\n", err);
		goto out2;
	}

	rc = -ENETUNREACH;
	err = rdma_create_qp(ia->ri_id, ia->ri_pd, &ep->rep_attr);
	if (err) {
		pr_err("rpcrdma: rdma_create_qp returned %d\n", err);
		goto out3;
	}

C
Chuck Lever 已提交
672
	rpcrdma_mrs_create(r_xprt);
673 674 675 676 677 678 679 680 681 682
	return 0;

out3:
	rpcrdma_ep_destroy(ep, ia);
out2:
	rpcrdma_ia_close(ia);
out1:
	return rc;
}

683 684 685 686 687 688 689
static int
rpcrdma_ep_reconnect(struct rpcrdma_xprt *r_xprt, struct rpcrdma_ep *ep,
		     struct rpcrdma_ia *ia)
{
	struct rdma_cm_id *id, *old;
	int err, rc;

690
	trace_xprtrdma_reconnect(r_xprt);
691 692 693 694

	rpcrdma_ep_disconnect(ep, ia);

	rc = -EHOSTUNREACH;
695
	id = rpcrdma_create_id(r_xprt, ia);
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727
	if (IS_ERR(id))
		goto out;

	/* As long as the new ID points to the same device as the
	 * old ID, we can reuse the transport's existing PD and all
	 * previously allocated MRs. Also, the same device means
	 * the transport's previous DMA mappings are still valid.
	 *
	 * This is a sanity check only. There should be no way these
	 * point to two different devices here.
	 */
	old = id;
	rc = -ENETUNREACH;
	if (ia->ri_device != id->device) {
		pr_err("rpcrdma: can't reconnect on different device!\n");
		goto out_destroy;
	}

	err = rdma_create_qp(id, ia->ri_pd, &ep->rep_attr);
	if (err) {
		dprintk("RPC:       %s: rdma_create_qp returned %d\n",
			__func__, err);
		goto out_destroy;
	}

	/* Atomically replace the transport's ID and QP. */
	rc = 0;
	old = ia->ri_id;
	ia->ri_id = id;
	rdma_destroy_qp(old);

out_destroy:
728
	rdma_destroy_id(old);
729 730 731 732
out:
	return rc;
}

733 734 735 736 737 738
/*
 * Connect unconnected endpoint.
 */
int
rpcrdma_ep_connect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
{
739 740
	struct rpcrdma_xprt *r_xprt = container_of(ia, struct rpcrdma_xprt,
						   rx_ia);
741
	int rc;
742 743

retry:
744 745
	switch (ep->rep_connected) {
	case 0:
746 747 748 749 750
		dprintk("RPC:       %s: connecting...\n", __func__);
		rc = rdma_create_qp(ia->ri_id, ia->ri_pd, &ep->rep_attr);
		if (rc) {
			dprintk("RPC:       %s: rdma_create_qp failed %i\n",
				__func__, rc);
751 752
			rc = -ENETUNREACH;
			goto out_noupdate;
753
		}
754
		break;
755 756 757 758 759
	case -ENODEV:
		rc = rpcrdma_ep_recreate_xprt(r_xprt, ep, ia);
		if (rc)
			goto out_noupdate;
		break;
760 761 762 763
	default:
		rc = rpcrdma_ep_reconnect(r_xprt, ep, ia);
		if (rc)
			goto out;
764 765 766
	}

	ep->rep_connected = 0;
767
	rpcrdma_post_recvs(r_xprt, true);
768 769 770 771 772 773 774 775 776 777

	rc = rdma_connect(ia->ri_id, &ep->rep_remote_cma);
	if (rc) {
		dprintk("RPC:       %s: rdma_connect() failed with %i\n",
				__func__, rc);
		goto out;
	}

	wait_event_interruptible(ep->rep_connect_wait, ep->rep_connected != 0);
	if (ep->rep_connected <= 0) {
778
		if (ep->rep_connected == -EAGAIN)
779 780
			goto retry;
		rc = ep->rep_connected;
781
		goto out;
782 783
	}

784
	dprintk("RPC:       %s: connected\n", __func__);
785

786 787 788
out:
	if (rc)
		ep->rep_connected = rc;
789 790

out_noupdate:
791 792 793 794 795 796 797 798 799 800 801 802
	return rc;
}

/*
 * rpcrdma_ep_disconnect
 *
 * This is separate from destroy to facilitate the ability
 * to reconnect without recreating the endpoint.
 *
 * This call is not reentrant, and must not be made in parallel
 * on the same endpoint.
 */
803
void
804 805 806 807 808
rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia)
{
	int rc;

	rc = rdma_disconnect(ia->ri_id);
809
	if (!rc)
810 811 812
		/* returns without wait if not connected */
		wait_event_interruptible(ep->rep_connect_wait,
							ep->rep_connected != 1);
813
	else
814
		ep->rep_connected = rc;
815 816
	trace_xprtrdma_disconnect(container_of(ep, struct rpcrdma_xprt,
					       rx_ep), rc);
817 818

	ib_drain_qp(ia->ri_id->qp);
819 820
}

821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889
/* Fixed-size circular FIFO queue. This implementation is wait-free and
 * lock-free.
 *
 * Consumer is the code path that posts Sends. This path dequeues a
 * sendctx for use by a Send operation. Multiple consumer threads
 * are serialized by the RPC transport lock, which allows only one
 * ->send_request call at a time.
 *
 * Producer is the code path that handles Send completions. This path
 * enqueues a sendctx that has been completed. Multiple producer
 * threads are serialized by the ib_poll_cq() function.
 */

/* rpcrdma_sendctxs_destroy() assumes caller has already quiesced
 * queue activity, and ib_drain_qp has flushed all remaining Send
 * requests.
 */
static void rpcrdma_sendctxs_destroy(struct rpcrdma_buffer *buf)
{
	unsigned long i;

	for (i = 0; i <= buf->rb_sc_last; i++)
		kfree(buf->rb_sc_ctxs[i]);
	kfree(buf->rb_sc_ctxs);
}

static struct rpcrdma_sendctx *rpcrdma_sendctx_create(struct rpcrdma_ia *ia)
{
	struct rpcrdma_sendctx *sc;

	sc = kzalloc(sizeof(*sc) +
		     ia->ri_max_send_sges * sizeof(struct ib_sge),
		     GFP_KERNEL);
	if (!sc)
		return NULL;

	sc->sc_wr.wr_cqe = &sc->sc_cqe;
	sc->sc_wr.sg_list = sc->sc_sges;
	sc->sc_wr.opcode = IB_WR_SEND;
	sc->sc_cqe.done = rpcrdma_wc_send;
	return sc;
}

static int rpcrdma_sendctxs_create(struct rpcrdma_xprt *r_xprt)
{
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
	struct rpcrdma_sendctx *sc;
	unsigned long i;

	/* Maximum number of concurrent outstanding Send WRs. Capping
	 * the circular queue size stops Send Queue overflow by causing
	 * the ->send_request call to fail temporarily before too many
	 * Sends are posted.
	 */
	i = buf->rb_max_requests + RPCRDMA_MAX_BC_REQUESTS;
	dprintk("RPC:       %s: allocating %lu send_ctxs\n", __func__, i);
	buf->rb_sc_ctxs = kcalloc(i, sizeof(sc), GFP_KERNEL);
	if (!buf->rb_sc_ctxs)
		return -ENOMEM;

	buf->rb_sc_last = i - 1;
	for (i = 0; i <= buf->rb_sc_last; i++) {
		sc = rpcrdma_sendctx_create(&r_xprt->rx_ia);
		if (!sc)
			goto out_destroy;

		sc->sc_xprt = r_xprt;
		buf->rb_sc_ctxs[i] = sc;
	}
890
	buf->rb_flags = 0;
891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947

	return 0;

out_destroy:
	rpcrdma_sendctxs_destroy(buf);
	return -ENOMEM;
}

/* The sendctx queue is not guaranteed to have a size that is a
 * power of two, thus the helpers in circ_buf.h cannot be used.
 * The other option is to use modulus (%), which can be expensive.
 */
static unsigned long rpcrdma_sendctx_next(struct rpcrdma_buffer *buf,
					  unsigned long item)
{
	return likely(item < buf->rb_sc_last) ? item + 1 : 0;
}

/**
 * rpcrdma_sendctx_get_locked - Acquire a send context
 * @buf: transport buffers from which to acquire an unused context
 *
 * Returns pointer to a free send completion context; or NULL if
 * the queue is empty.
 *
 * Usage: Called to acquire an SGE array before preparing a Send WR.
 *
 * The caller serializes calls to this function (per rpcrdma_buffer),
 * and provides an effective memory barrier that flushes the new value
 * of rb_sc_head.
 */
struct rpcrdma_sendctx *rpcrdma_sendctx_get_locked(struct rpcrdma_buffer *buf)
{
	struct rpcrdma_xprt *r_xprt;
	struct rpcrdma_sendctx *sc;
	unsigned long next_head;

	next_head = rpcrdma_sendctx_next(buf, buf->rb_sc_head);

	if (next_head == READ_ONCE(buf->rb_sc_tail))
		goto out_emptyq;

	/* ORDER: item must be accessed _before_ head is updated */
	sc = buf->rb_sc_ctxs[next_head];

	/* Releasing the lock in the caller acts as a memory
	 * barrier that flushes rb_sc_head.
	 */
	buf->rb_sc_head = next_head;

	return sc;

out_emptyq:
	/* The queue is "empty" if there have not been enough Send
	 * completions recently. This is a sign the Send Queue is
	 * backing up. Cause the caller to pause and try again.
	 */
948
	set_bit(RPCRDMA_BUF_F_EMPTY_SCQ, &buf->rb_flags);
949 950 951 952 953 954 955 956 957 958 959 960 961 962
	r_xprt = container_of(buf, struct rpcrdma_xprt, rx_buf);
	r_xprt->rx_stats.empty_sendctx_q++;
	return NULL;
}

/**
 * rpcrdma_sendctx_put_locked - Release a send context
 * @sc: send context to release
 *
 * Usage: Called from Send completion to return a sendctxt
 * to the queue.
 *
 * The caller serializes calls to this function (per rpcrdma_buffer).
 */
963 964
static void
rpcrdma_sendctx_put_locked(struct rpcrdma_sendctx *sc)
965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982
{
	struct rpcrdma_buffer *buf = &sc->sc_xprt->rx_buf;
	unsigned long next_tail;

	/* Unmap SGEs of previously completed by unsignaled
	 * Sends by walking up the queue until @sc is found.
	 */
	next_tail = buf->rb_sc_tail;
	do {
		next_tail = rpcrdma_sendctx_next(buf, next_tail);

		/* ORDER: item must be accessed _before_ tail is updated */
		rpcrdma_unmap_sendctx(buf->rb_sc_ctxs[next_tail]);

	} while (buf->rb_sc_ctxs[next_tail] != sc);

	/* Paired with READ_ONCE */
	smp_store_release(&buf->rb_sc_tail, next_tail);
983 984 985 986 987

	if (test_and_clear_bit(RPCRDMA_BUF_F_EMPTY_SCQ, &buf->rb_flags)) {
		smp_mb__after_atomic();
		xprt_write_space(&sc->sc_xprt->rx_xprt);
	}
988 989
}

C
Chuck Lever 已提交
990
static void
C
Chuck Lever 已提交
991
rpcrdma_mrs_create(struct rpcrdma_xprt *r_xprt)
C
Chuck Lever 已提交
992 993 994 995 996 997 998
{
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
	unsigned int count;
	LIST_HEAD(free);
	LIST_HEAD(all);

C
Chuck Lever 已提交
999
	for (count = 0; count < ia->ri_max_segs; count++) {
C
Chuck Lever 已提交
1000
		struct rpcrdma_mr *mr;
C
Chuck Lever 已提交
1001 1002
		int rc;

C
Chuck Lever 已提交
1003 1004
		mr = kzalloc(sizeof(*mr), GFP_KERNEL);
		if (!mr)
C
Chuck Lever 已提交
1005 1006
			break;

C
Chuck Lever 已提交
1007
		rc = ia->ri_ops->ro_init_mr(ia, mr);
C
Chuck Lever 已提交
1008
		if (rc) {
C
Chuck Lever 已提交
1009
			kfree(mr);
C
Chuck Lever 已提交
1010 1011 1012
			break;
		}

C
Chuck Lever 已提交
1013
		mr->mr_xprt = r_xprt;
C
Chuck Lever 已提交
1014

C
Chuck Lever 已提交
1015 1016
		list_add(&mr->mr_list, &free);
		list_add(&mr->mr_all, &all);
C
Chuck Lever 已提交
1017 1018
	}

C
Chuck Lever 已提交
1019 1020
	spin_lock(&buf->rb_mrlock);
	list_splice(&free, &buf->rb_mrs);
C
Chuck Lever 已提交
1021 1022
	list_splice(&all, &buf->rb_all);
	r_xprt->rx_stats.mrs_allocated += count;
C
Chuck Lever 已提交
1023
	spin_unlock(&buf->rb_mrlock);
1024
	trace_xprtrdma_createmrs(r_xprt, count);
1025 1026

	xprt_write_space(&r_xprt->rx_xprt);
C
Chuck Lever 已提交
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
}

static void
rpcrdma_mr_refresh_worker(struct work_struct *work)
{
	struct rpcrdma_buffer *buf = container_of(work, struct rpcrdma_buffer,
						  rb_refresh_worker.work);
	struct rpcrdma_xprt *r_xprt = container_of(buf, struct rpcrdma_xprt,
						   rx_buf);

C
Chuck Lever 已提交
1037
	rpcrdma_mrs_create(r_xprt);
C
Chuck Lever 已提交
1038 1039
}

1040
struct rpcrdma_req *
1041 1042
rpcrdma_create_req(struct rpcrdma_xprt *r_xprt)
{
1043
	struct rpcrdma_buffer *buffer = &r_xprt->rx_buf;
1044
	struct rpcrdma_regbuf *rb;
1045 1046
	struct rpcrdma_req *req;

1047
	req = kzalloc(sizeof(*req), GFP_KERNEL);
1048
	if (req == NULL)
1049
		return ERR_PTR(-ENOMEM);
1050

1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
	rb = rpcrdma_alloc_regbuf(RPCRDMA_HDRBUF_SIZE,
				  DMA_TO_DEVICE, GFP_KERNEL);
	if (IS_ERR(rb)) {
		kfree(req);
		return ERR_PTR(-ENOMEM);
	}
	req->rl_rdmabuf = rb;
	xdr_buf_init(&req->rl_hdrbuf, rb->rg_base, rdmab_length(rb));
	req->rl_buffer = buffer;
	INIT_LIST_HEAD(&req->rl_registered);

1062 1063 1064
	spin_lock(&buffer->rb_reqslock);
	list_add(&req->rl_all, &buffer->rb_allreqs);
	spin_unlock(&buffer->rb_reqslock);
1065 1066 1067
	return req;
}

1068 1069
static int
rpcrdma_create_rep(struct rpcrdma_xprt *r_xprt, bool temp)
1070 1071
{
	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
1072
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
1073 1074 1075 1076
	struct rpcrdma_rep *rep;
	int rc;

	rc = -ENOMEM;
1077
	rep = kzalloc(sizeof(*rep), GFP_KERNEL);
1078 1079 1080
	if (rep == NULL)
		goto out;

1081
	rep->rr_rdmabuf = rpcrdma_alloc_regbuf(cdata->inline_rsize,
1082
					       DMA_FROM_DEVICE, GFP_KERNEL);
1083 1084
	if (IS_ERR(rep->rr_rdmabuf)) {
		rc = PTR_ERR(rep->rr_rdmabuf);
1085
		goto out_free;
1086
	}
1087 1088
	xdr_buf_init(&rep->rr_hdrbuf, rep->rr_rdmabuf->rg_base,
		     rdmab_length(rep->rr_rdmabuf));
1089

1090
	rep->rr_cqe.done = rpcrdma_wc_receive;
1091
	rep->rr_rxprt = r_xprt;
1092
	INIT_WORK(&rep->rr_work, rpcrdma_deferred_completion);
1093 1094 1095 1096
	rep->rr_recv_wr.next = NULL;
	rep->rr_recv_wr.wr_cqe = &rep->rr_cqe;
	rep->rr_recv_wr.sg_list = &rep->rr_rdmabuf->rg_iov;
	rep->rr_recv_wr.num_sge = 1;
1097
	rep->rr_temp = temp;
1098 1099 1100 1101 1102

	spin_lock(&buf->rb_lock);
	list_add(&rep->rr_list, &buf->rb_recv_bufs);
	spin_unlock(&buf->rb_lock);
	return 0;
1103 1104 1105 1106

out_free:
	kfree(rep);
out:
1107 1108 1109
	dprintk("RPC:       %s: reply buffer %d alloc failed\n",
		__func__, rc);
	return rc;
1110 1111
}

1112
int
1113
rpcrdma_buffer_create(struct rpcrdma_xprt *r_xprt)
1114
{
1115
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
1116 1117
	int i, rc;

1118
	buf->rb_max_requests = r_xprt->rx_data.max_requests;
1119
	buf->rb_bc_srv_max_requests = 0;
C
Chuck Lever 已提交
1120
	spin_lock_init(&buf->rb_mrlock);
1121
	spin_lock_init(&buf->rb_lock);
C
Chuck Lever 已提交
1122
	INIT_LIST_HEAD(&buf->rb_mrs);
C
Chuck Lever 已提交
1123 1124 1125
	INIT_LIST_HEAD(&buf->rb_all);
	INIT_DELAYED_WORK(&buf->rb_refresh_worker,
			  rpcrdma_mr_refresh_worker);
1126

C
Chuck Lever 已提交
1127
	rpcrdma_mrs_create(r_xprt);
1128

1129
	INIT_LIST_HEAD(&buf->rb_send_bufs);
1130 1131
	INIT_LIST_HEAD(&buf->rb_allreqs);
	spin_lock_init(&buf->rb_reqslock);
1132 1133 1134
	for (i = 0; i < buf->rb_max_requests; i++) {
		struct rpcrdma_req *req;

1135 1136
		req = rpcrdma_create_req(r_xprt);
		if (IS_ERR(req)) {
1137 1138
			dprintk("RPC:       %s: request buffer %d alloc"
				" failed\n", __func__, i);
1139
			rc = PTR_ERR(req);
1140 1141
			goto out;
		}
1142
		list_add(&req->rl_list, &buf->rb_send_bufs);
1143 1144
	}

1145
	buf->rb_credits = 1;
1146
	buf->rb_posted_receives = 0;
1147
	INIT_LIST_HEAD(&buf->rb_recv_bufs);
1148

1149 1150 1151 1152
	rc = rpcrdma_sendctxs_create(r_xprt);
	if (rc)
		goto out;

1153 1154 1155 1156 1157 1158
	return 0;
out:
	rpcrdma_buffer_destroy(buf);
	return rc;
}

1159
static void
1160
rpcrdma_destroy_rep(struct rpcrdma_rep *rep)
1161
{
1162
	rpcrdma_free_regbuf(rep->rr_rdmabuf);
1163 1164 1165
	kfree(rep);
}

1166
void
1167
rpcrdma_destroy_req(struct rpcrdma_req *req)
1168
{
1169 1170 1171
	rpcrdma_free_regbuf(req->rl_recvbuf);
	rpcrdma_free_regbuf(req->rl_sendbuf);
	rpcrdma_free_regbuf(req->rl_rdmabuf);
1172 1173 1174
	kfree(req);
}

C
Chuck Lever 已提交
1175
static void
C
Chuck Lever 已提交
1176
rpcrdma_mrs_destroy(struct rpcrdma_buffer *buf)
C
Chuck Lever 已提交
1177 1178 1179 1180
{
	struct rpcrdma_xprt *r_xprt = container_of(buf, struct rpcrdma_xprt,
						   rx_buf);
	struct rpcrdma_ia *ia = rdmab_to_ia(buf);
C
Chuck Lever 已提交
1181
	struct rpcrdma_mr *mr;
C
Chuck Lever 已提交
1182 1183 1184
	unsigned int count;

	count = 0;
C
Chuck Lever 已提交
1185
	spin_lock(&buf->rb_mrlock);
C
Chuck Lever 已提交
1186
	while (!list_empty(&buf->rb_all)) {
C
Chuck Lever 已提交
1187 1188
		mr = list_entry(buf->rb_all.next, struct rpcrdma_mr, mr_all);
		list_del(&mr->mr_all);
C
Chuck Lever 已提交
1189

C
Chuck Lever 已提交
1190
		spin_unlock(&buf->rb_mrlock);
1191 1192 1193 1194 1195

		/* Ensure MW is not on any rl_registered list */
		if (!list_empty(&mr->mr_list))
			list_del(&mr->mr_list);

C
Chuck Lever 已提交
1196
		ia->ri_ops->ro_release_mr(mr);
C
Chuck Lever 已提交
1197
		count++;
C
Chuck Lever 已提交
1198
		spin_lock(&buf->rb_mrlock);
C
Chuck Lever 已提交
1199
	}
C
Chuck Lever 已提交
1200
	spin_unlock(&buf->rb_mrlock);
C
Chuck Lever 已提交
1201 1202 1203 1204 1205
	r_xprt->rx_stats.mrs_allocated = 0;

	dprintk("RPC:       %s: released %u MRs\n", __func__, count);
}

1206 1207 1208
void
rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
{
1209
	cancel_delayed_work_sync(&buf->rb_refresh_worker);
1210

1211 1212
	rpcrdma_sendctxs_destroy(buf);

1213 1214
	while (!list_empty(&buf->rb_recv_bufs)) {
		struct rpcrdma_rep *rep;
1215

1216 1217 1218
		rep = list_first_entry(&buf->rb_recv_bufs,
				       struct rpcrdma_rep, rr_list);
		list_del(&rep->rr_list);
1219
		rpcrdma_destroy_rep(rep);
1220 1221
	}

1222 1223
	spin_lock(&buf->rb_reqslock);
	while (!list_empty(&buf->rb_allreqs)) {
1224
		struct rpcrdma_req *req;
A
Allen Andrews 已提交
1225

1226 1227 1228 1229 1230
		req = list_first_entry(&buf->rb_allreqs,
				       struct rpcrdma_req, rl_all);
		list_del(&req->rl_all);

		spin_unlock(&buf->rb_reqslock);
1231
		rpcrdma_destroy_req(req);
1232
		spin_lock(&buf->rb_reqslock);
1233
	}
1234
	spin_unlock(&buf->rb_reqslock);
A
Allen Andrews 已提交
1235

C
Chuck Lever 已提交
1236
	rpcrdma_mrs_destroy(buf);
1237 1238
}

C
Chuck Lever 已提交
1239 1240 1241 1242 1243 1244 1245 1246 1247
/**
 * rpcrdma_mr_get - Allocate an rpcrdma_mr object
 * @r_xprt: controlling transport
 *
 * Returns an initialized rpcrdma_mr or NULL if no free
 * rpcrdma_mr objects are available.
 */
struct rpcrdma_mr *
rpcrdma_mr_get(struct rpcrdma_xprt *r_xprt)
1248
{
1249
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
C
Chuck Lever 已提交
1250
	struct rpcrdma_mr *mr = NULL;
1251

C
Chuck Lever 已提交
1252 1253 1254 1255
	spin_lock(&buf->rb_mrlock);
	if (!list_empty(&buf->rb_mrs))
		mr = rpcrdma_mr_pop(&buf->rb_mrs);
	spin_unlock(&buf->rb_mrlock);
1256

C
Chuck Lever 已提交
1257 1258 1259
	if (!mr)
		goto out_nomrs;
	return mr;
C
Chuck Lever 已提交
1260

C
Chuck Lever 已提交
1261
out_nomrs:
1262
	trace_xprtrdma_nomrs(r_xprt);
1263 1264
	if (r_xprt->rx_ep.rep_connected != -ENODEV)
		schedule_delayed_work(&buf->rb_refresh_worker, 0);
C
Chuck Lever 已提交
1265 1266 1267 1268 1269

	/* Allow the reply handler and refresh worker to run */
	cond_resched();

	return NULL;
1270 1271
}

1272 1273 1274 1275 1276 1277 1278 1279
static void
__rpcrdma_mr_put(struct rpcrdma_buffer *buf, struct rpcrdma_mr *mr)
{
	spin_lock(&buf->rb_mrlock);
	rpcrdma_mr_push(mr, &buf->rb_mrs);
	spin_unlock(&buf->rb_mrlock);
}

C
Chuck Lever 已提交
1280 1281 1282 1283 1284
/**
 * rpcrdma_mr_put - Release an rpcrdma_mr object
 * @mr: object to release
 *
 */
1285
void
C
Chuck Lever 已提交
1286
rpcrdma_mr_put(struct rpcrdma_mr *mr)
1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
{
	__rpcrdma_mr_put(&mr->mr_xprt->rx_buf, mr);
}

/**
 * rpcrdma_mr_unmap_and_put - DMA unmap an MR and release it
 * @mr: object to release
 *
 */
void
rpcrdma_mr_unmap_and_put(struct rpcrdma_mr *mr)
1298
{
C
Chuck Lever 已提交
1299
	struct rpcrdma_xprt *r_xprt = mr->mr_xprt;
1300

1301
	trace_xprtrdma_mr_unmap(mr);
1302 1303 1304
	ib_dma_unmap_sg(r_xprt->rx_ia.ri_device,
			mr->mr_sg, mr->mr_nents, mr->mr_dir);
	__rpcrdma_mr_put(&r_xprt->rx_buf, mr);
1305 1306
}

1307 1308 1309
/**
 * rpcrdma_buffer_get - Get a request buffer
 * @buffers: Buffer pool from which to obtain a buffer
1310
 *
1311
 * Returns a fresh rpcrdma_req, or NULL if none are available.
1312 1313 1314 1315 1316
 */
struct rpcrdma_req *
rpcrdma_buffer_get(struct rpcrdma_buffer *buffers)
{
	struct rpcrdma_req *req;
1317

1318
	spin_lock(&buffers->rb_lock);
1319 1320 1321 1322
	req = list_first_entry_or_null(&buffers->rb_send_bufs,
				       struct rpcrdma_req, rl_list);
	if (req)
		list_del_init(&req->rl_list);
1323
	spin_unlock(&buffers->rb_lock);
1324
	return req;
1325 1326
}

1327 1328 1329 1330
/**
 * rpcrdma_buffer_put - Put request/reply buffers back into pool
 * @req: object to return
 *
1331 1332 1333 1334 1335
 */
void
rpcrdma_buffer_put(struct rpcrdma_req *req)
{
	struct rpcrdma_buffer *buffers = req->rl_buffer;
1336
	struct rpcrdma_rep *rep = req->rl_reply;
1337

1338 1339
	req->rl_reply = NULL;

1340
	spin_lock(&buffers->rb_lock);
1341
	list_add(&req->rl_list, &buffers->rb_send_bufs);
1342
	if (rep) {
1343 1344 1345 1346
		if (!rep->rr_temp) {
			list_add(&rep->rr_list, &buffers->rb_recv_bufs);
			rep = NULL;
		}
1347
	}
1348
	spin_unlock(&buffers->rb_lock);
1349 1350
	if (rep)
		rpcrdma_destroy_rep(rep);
1351 1352 1353 1354
}

/*
 * Put reply buffers back into pool when not attached to
1355
 * request. This happens in error conditions.
1356 1357 1358 1359
 */
void
rpcrdma_recv_buffer_put(struct rpcrdma_rep *rep)
{
1360
	struct rpcrdma_buffer *buffers = &rep->rr_rxprt->rx_buf;
1361

1362 1363 1364 1365 1366 1367 1368
	if (!rep->rr_temp) {
		spin_lock(&buffers->rb_lock);
		list_add(&rep->rr_list, &buffers->rb_recv_bufs);
		spin_unlock(&buffers->rb_lock);
	} else {
		rpcrdma_destroy_rep(rep);
	}
1369 1370
}

1371
/**
1372
 * rpcrdma_alloc_regbuf - allocate and DMA-map memory for SEND/RECV buffers
1373
 * @size: size of buffer to be allocated, in bytes
1374
 * @direction: direction of data movement
1375 1376
 * @flags: GFP flags
 *
1377 1378
 * Returns an ERR_PTR, or a pointer to a regbuf, a buffer that
 * can be persistently DMA-mapped for I/O.
1379 1380
 *
 * xprtrdma uses a regbuf for posting an outgoing RDMA SEND, or for
1381 1382
 * receiving the payload of RDMA RECV operations. During Long Calls
 * or Replies they may be registered externally via ro_map.
1383 1384
 */
struct rpcrdma_regbuf *
1385 1386
rpcrdma_alloc_regbuf(size_t size, enum dma_data_direction direction,
		     gfp_t flags)
1387 1388 1389 1390 1391
{
	struct rpcrdma_regbuf *rb;

	rb = kmalloc(sizeof(*rb) + size, flags);
	if (rb == NULL)
1392
		return ERR_PTR(-ENOMEM);
1393

1394
	rb->rg_device = NULL;
1395
	rb->rg_direction = direction;
1396
	rb->rg_iov.length = size;
1397 1398

	return rb;
1399
}
1400

1401 1402 1403 1404 1405 1406 1407 1408
/**
 * __rpcrdma_map_regbuf - DMA-map a regbuf
 * @ia: controlling rpcrdma_ia
 * @rb: regbuf to be mapped
 */
bool
__rpcrdma_dma_map_regbuf(struct rpcrdma_ia *ia, struct rpcrdma_regbuf *rb)
{
1409 1410
	struct ib_device *device = ia->ri_device;

1411 1412 1413
	if (rb->rg_direction == DMA_NONE)
		return false;

1414
	rb->rg_iov.addr = ib_dma_map_single(device,
1415 1416 1417
					    (void *)rb->rg_base,
					    rdmab_length(rb),
					    rb->rg_direction);
1418
	if (ib_dma_mapping_error(device, rdmab_addr(rb)))
1419 1420
		return false;

1421
	rb->rg_device = device;
1422 1423 1424 1425 1426 1427 1428
	rb->rg_iov.lkey = ia->ri_pd->local_dma_lkey;
	return true;
}

static void
rpcrdma_dma_unmap_regbuf(struct rpcrdma_regbuf *rb)
{
1429 1430 1431
	if (!rb)
		return;

1432 1433 1434 1435 1436 1437
	if (!rpcrdma_regbuf_is_mapped(rb))
		return;

	ib_dma_unmap_single(rb->rg_device, rdmab_addr(rb),
			    rdmab_length(rb), rb->rg_direction);
	rb->rg_device = NULL;
1438 1439 1440 1441 1442 1443 1444
}

/**
 * rpcrdma_free_regbuf - deregister and free registered buffer
 * @rb: regbuf to be deregistered and freed
 */
void
1445
rpcrdma_free_regbuf(struct rpcrdma_regbuf *rb)
1446
{
1447
	rpcrdma_dma_unmap_regbuf(rb);
1448
	kfree(rb);
1449 1450
}

1451 1452 1453 1454 1455 1456 1457 1458 1459 1460
/*
 * Prepost any receive buffer, then post send.
 *
 * Receive buffer is donated to hardware, reclaimed upon recv completion.
 */
int
rpcrdma_ep_post(struct rpcrdma_ia *ia,
		struct rpcrdma_ep *ep,
		struct rpcrdma_req *req)
{
1461
	struct ib_send_wr *send_wr = &req->rl_sendctx->sc_wr;
1462
	int rc;
1463

1464 1465
	if (!ep->rep_send_count ||
	    test_bit(RPCRDMA_REQ_F_TX_RESOURCES, &req->rl_flags)) {
1466 1467 1468 1469 1470 1471
		send_wr->send_flags |= IB_SEND_SIGNALED;
		ep->rep_send_count = ep->rep_send_batch;
	} else {
		send_wr->send_flags &= ~IB_SEND_SIGNALED;
		--ep->rep_send_count;
	}
1472

1473
	rc = ia->ri_ops->ro_send(ia, req);
1474
	trace_xprtrdma_post_send(req, rc);
1475
	if (rc)
1476
		return -ENOTCONN;
1477
	return 0;
1478 1479
}

1480
/**
1481 1482 1483
 * rpcrdma_post_recvs - Maybe post some Receive buffers
 * @r_xprt: controlling transport
 * @temp: when true, allocate temp rpcrdma_rep objects
1484 1485
 *
 */
1486 1487
void
rpcrdma_post_recvs(struct rpcrdma_xprt *r_xprt, bool temp)
1488
{
1489 1490 1491
	struct rpcrdma_buffer *buf = &r_xprt->rx_buf;
	struct ib_recv_wr *wr, *bad_wr;
	int needed, count, rc;
1492

1493 1494 1495 1496
	needed = buf->rb_credits + (buf->rb_bc_srv_max_requests << 1);
	if (buf->rb_posted_receives > needed)
		return;
	needed -= buf->rb_posted_receives;
1497

1498 1499 1500 1501 1502
	count = 0;
	wr = NULL;
	while (needed) {
		struct rpcrdma_regbuf *rb;
		struct rpcrdma_rep *rep;
1503

1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514
		spin_lock(&buf->rb_lock);
		rep = list_first_entry_or_null(&buf->rb_recv_bufs,
					       struct rpcrdma_rep, rr_list);
		if (likely(rep))
			list_del(&rep->rr_list);
		spin_unlock(&buf->rb_lock);
		if (!rep) {
			if (rpcrdma_create_rep(r_xprt, temp))
				break;
			continue;
		}
1515

1516 1517 1518 1519 1520 1521 1522
		rb = rep->rr_rdmabuf;
		if (!rpcrdma_regbuf_is_mapped(rb)) {
			if (!__rpcrdma_dma_map_regbuf(&r_xprt->rx_ia, rb)) {
				rpcrdma_recv_buffer_put(rep);
				break;
			}
		}
1523

1524 1525 1526 1527 1528 1529 1530 1531 1532
		trace_xprtrdma_post_recv(rep->rr_recv_wr.wr_cqe);
		rep->rr_recv_wr.next = wr;
		wr = &rep->rr_recv_wr;
		++count;
		--needed;
	}
	if (!count)
		return;

1533 1534
	rc = ib_post_recv(r_xprt->rx_ia.ri_id->qp, wr,
			  (const struct ib_recv_wr **)&bad_wr);
1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545
	if (rc) {
		for (wr = bad_wr; wr; wr = wr->next) {
			struct rpcrdma_rep *rep;

			rep = container_of(wr, struct rpcrdma_rep, rr_recv_wr);
			rpcrdma_recv_buffer_put(rep);
			--count;
		}
	}
	buf->rb_posted_receives += count;
	trace_xprtrdma_post_recvs(r_xprt, count, rc);
1546
}