xref: /openbmc/linux/net/sunrpc/xprtrdma/transport.c (revision b08918fb)
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  * transport.c
44  *
45  * This file contains the top-level implementation of an RPC RDMA
46  * transport.
47  *
48  * Naming convention: functions beginning with xprt_ are part of the
49  * transport switch. All others are RPC RDMA internal.
50  */
51 
52 #include <linux/module.h>
53 #include <linux/slab.h>
54 #include <linux/seq_file.h>
55 #include <linux/smp.h>
56 
57 #include <linux/sunrpc/addr.h>
58 #include <linux/sunrpc/svc_rdma.h>
59 
60 #include "xprt_rdma.h"
61 #include <trace/events/rpcrdma.h>
62 
63 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
64 # define RPCDBG_FACILITY	RPCDBG_TRANS
65 #endif
66 
67 /*
68  * tunables
69  */
70 
71 unsigned int xprt_rdma_slot_table_entries = RPCRDMA_DEF_SLOT_TABLE;
72 unsigned int xprt_rdma_max_inline_read = RPCRDMA_DEF_INLINE;
73 unsigned int xprt_rdma_max_inline_write = RPCRDMA_DEF_INLINE;
74 unsigned int xprt_rdma_memreg_strategy		= RPCRDMA_FRWR;
75 int xprt_rdma_pad_optimize;
76 
77 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
78 
79 static unsigned int min_slot_table_size = RPCRDMA_MIN_SLOT_TABLE;
80 static unsigned int max_slot_table_size = RPCRDMA_MAX_SLOT_TABLE;
81 static unsigned int min_inline_size = RPCRDMA_MIN_INLINE;
82 static unsigned int max_inline_size = RPCRDMA_MAX_INLINE;
83 static unsigned int max_padding = PAGE_SIZE;
84 static unsigned int min_memreg = RPCRDMA_BOUNCEBUFFERS;
85 static unsigned int max_memreg = RPCRDMA_LAST - 1;
86 static unsigned int dummy;
87 
88 static struct ctl_table_header *sunrpc_table_header;
89 
90 static struct ctl_table xr_tunables_table[] = {
91 	{
92 		.procname	= "rdma_slot_table_entries",
93 		.data		= &xprt_rdma_slot_table_entries,
94 		.maxlen		= sizeof(unsigned int),
95 		.mode		= 0644,
96 		.proc_handler	= proc_dointvec_minmax,
97 		.extra1		= &min_slot_table_size,
98 		.extra2		= &max_slot_table_size
99 	},
100 	{
101 		.procname	= "rdma_max_inline_read",
102 		.data		= &xprt_rdma_max_inline_read,
103 		.maxlen		= sizeof(unsigned int),
104 		.mode		= 0644,
105 		.proc_handler	= proc_dointvec_minmax,
106 		.extra1		= &min_inline_size,
107 		.extra2		= &max_inline_size,
108 	},
109 	{
110 		.procname	= "rdma_max_inline_write",
111 		.data		= &xprt_rdma_max_inline_write,
112 		.maxlen		= sizeof(unsigned int),
113 		.mode		= 0644,
114 		.proc_handler	= proc_dointvec_minmax,
115 		.extra1		= &min_inline_size,
116 		.extra2		= &max_inline_size,
117 	},
118 	{
119 		.procname	= "rdma_inline_write_padding",
120 		.data		= &dummy,
121 		.maxlen		= sizeof(unsigned int),
122 		.mode		= 0644,
123 		.proc_handler	= proc_dointvec_minmax,
124 		.extra1		= SYSCTL_ZERO,
125 		.extra2		= &max_padding,
126 	},
127 	{
128 		.procname	= "rdma_memreg_strategy",
129 		.data		= &xprt_rdma_memreg_strategy,
130 		.maxlen		= sizeof(unsigned int),
131 		.mode		= 0644,
132 		.proc_handler	= proc_dointvec_minmax,
133 		.extra1		= &min_memreg,
134 		.extra2		= &max_memreg,
135 	},
136 	{
137 		.procname	= "rdma_pad_optimize",
138 		.data		= &xprt_rdma_pad_optimize,
139 		.maxlen		= sizeof(unsigned int),
140 		.mode		= 0644,
141 		.proc_handler	= proc_dointvec,
142 	},
143 	{ },
144 };
145 
146 static struct ctl_table sunrpc_table[] = {
147 	{
148 		.procname	= "sunrpc",
149 		.mode		= 0555,
150 		.child		= xr_tunables_table
151 	},
152 	{ },
153 };
154 
155 #endif
156 
157 static const struct rpc_xprt_ops xprt_rdma_procs;
158 
159 static void
160 xprt_rdma_format_addresses4(struct rpc_xprt *xprt, struct sockaddr *sap)
161 {
162 	struct sockaddr_in *sin = (struct sockaddr_in *)sap;
163 	char buf[20];
164 
165 	snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr));
166 	xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL);
167 
168 	xprt->address_strings[RPC_DISPLAY_NETID] = RPCBIND_NETID_RDMA;
169 }
170 
171 static void
172 xprt_rdma_format_addresses6(struct rpc_xprt *xprt, struct sockaddr *sap)
173 {
174 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap;
175 	char buf[40];
176 
177 	snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr);
178 	xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL);
179 
180 	xprt->address_strings[RPC_DISPLAY_NETID] = RPCBIND_NETID_RDMA6;
181 }
182 
183 void
184 xprt_rdma_format_addresses(struct rpc_xprt *xprt, struct sockaddr *sap)
185 {
186 	char buf[128];
187 
188 	switch (sap->sa_family) {
189 	case AF_INET:
190 		xprt_rdma_format_addresses4(xprt, sap);
191 		break;
192 	case AF_INET6:
193 		xprt_rdma_format_addresses6(xprt, sap);
194 		break;
195 	default:
196 		pr_err("rpcrdma: Unrecognized address family\n");
197 		return;
198 	}
199 
200 	(void)rpc_ntop(sap, buf, sizeof(buf));
201 	xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL);
202 
203 	snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap));
204 	xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL);
205 
206 	snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap));
207 	xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL);
208 
209 	xprt->address_strings[RPC_DISPLAY_PROTO] = "rdma";
210 }
211 
212 void
213 xprt_rdma_free_addresses(struct rpc_xprt *xprt)
214 {
215 	unsigned int i;
216 
217 	for (i = 0; i < RPC_DISPLAY_MAX; i++)
218 		switch (i) {
219 		case RPC_DISPLAY_PROTO:
220 		case RPC_DISPLAY_NETID:
221 			continue;
222 		default:
223 			kfree(xprt->address_strings[i]);
224 		}
225 }
226 
227 /**
228  * xprt_rdma_connect_worker - establish connection in the background
229  * @work: worker thread context
230  *
231  * Requester holds the xprt's send lock to prevent activity on this
232  * transport while a fresh connection is being established. RPC tasks
233  * sleep on the xprt's pending queue waiting for connect to complete.
234  */
235 static void
236 xprt_rdma_connect_worker(struct work_struct *work)
237 {
238 	struct rpcrdma_xprt *r_xprt = container_of(work, struct rpcrdma_xprt,
239 						   rx_connect_worker.work);
240 	struct rpc_xprt *xprt = &r_xprt->rx_xprt;
241 	int rc;
242 
243 	rc = rpcrdma_ep_connect(&r_xprt->rx_ep, &r_xprt->rx_ia);
244 	xprt_clear_connecting(xprt);
245 	if (r_xprt->rx_ep.rep_connected > 0) {
246 		if (!xprt_test_and_set_connected(xprt)) {
247 			xprt->stat.connect_count++;
248 			xprt->stat.connect_time += (long)jiffies -
249 						   xprt->stat.connect_start;
250 			xprt_wake_pending_tasks(xprt, -EAGAIN);
251 		}
252 	} else {
253 		if (xprt_test_and_clear_connected(xprt))
254 			xprt_wake_pending_tasks(xprt, rc);
255 	}
256 }
257 
258 /**
259  * xprt_rdma_inject_disconnect - inject a connection fault
260  * @xprt: transport context
261  *
262  * If @xprt is connected, disconnect it to simulate spurious connection
263  * loss.
264  */
265 static void
266 xprt_rdma_inject_disconnect(struct rpc_xprt *xprt)
267 {
268 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
269 
270 	trace_xprtrdma_op_inject_dsc(r_xprt);
271 	rdma_disconnect(r_xprt->rx_ia.ri_id);
272 }
273 
274 /**
275  * xprt_rdma_destroy - Full tear down of transport
276  * @xprt: doomed transport context
277  *
278  * Caller guarantees there will be no more calls to us with
279  * this @xprt.
280  */
281 static void
282 xprt_rdma_destroy(struct rpc_xprt *xprt)
283 {
284 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
285 
286 	trace_xprtrdma_op_destroy(r_xprt);
287 
288 	cancel_delayed_work_sync(&r_xprt->rx_connect_worker);
289 
290 	rpcrdma_ep_destroy(r_xprt);
291 	rpcrdma_buffer_destroy(&r_xprt->rx_buf);
292 	rpcrdma_ia_close(&r_xprt->rx_ia);
293 
294 	xprt_rdma_free_addresses(xprt);
295 	xprt_free(xprt);
296 
297 	module_put(THIS_MODULE);
298 }
299 
300 /* 60 second timeout, no retries */
301 static const struct rpc_timeout xprt_rdma_default_timeout = {
302 	.to_initval = 60 * HZ,
303 	.to_maxval = 60 * HZ,
304 };
305 
306 /**
307  * xprt_setup_rdma - Set up transport to use RDMA
308  *
309  * @args: rpc transport arguments
310  */
311 static struct rpc_xprt *
312 xprt_setup_rdma(struct xprt_create *args)
313 {
314 	struct rpc_xprt *xprt;
315 	struct rpcrdma_xprt *new_xprt;
316 	struct sockaddr *sap;
317 	int rc;
318 
319 	if (args->addrlen > sizeof(xprt->addr))
320 		return ERR_PTR(-EBADF);
321 
322 	xprt = xprt_alloc(args->net, sizeof(struct rpcrdma_xprt), 0, 0);
323 	if (!xprt)
324 		return ERR_PTR(-ENOMEM);
325 
326 	xprt->timeout = &xprt_rdma_default_timeout;
327 	xprt->connect_timeout = xprt->timeout->to_initval;
328 	xprt->max_reconnect_timeout = xprt->timeout->to_maxval;
329 	xprt->bind_timeout = RPCRDMA_BIND_TO;
330 	xprt->reestablish_timeout = RPCRDMA_INIT_REEST_TO;
331 	xprt->idle_timeout = RPCRDMA_IDLE_DISC_TO;
332 
333 	xprt->resvport = 0;		/* privileged port not needed */
334 	xprt->ops = &xprt_rdma_procs;
335 
336 	/*
337 	 * Set up RDMA-specific connect data.
338 	 */
339 	sap = args->dstaddr;
340 
341 	/* Ensure xprt->addr holds valid server TCP (not RDMA)
342 	 * address, for any side protocols which peek at it */
343 	xprt->prot = IPPROTO_TCP;
344 	xprt->addrlen = args->addrlen;
345 	memcpy(&xprt->addr, sap, xprt->addrlen);
346 
347 	if (rpc_get_port(sap))
348 		xprt_set_bound(xprt);
349 	xprt_rdma_format_addresses(xprt, sap);
350 
351 	new_xprt = rpcx_to_rdmax(xprt);
352 	rc = rpcrdma_ia_open(new_xprt);
353 	if (rc)
354 		goto out1;
355 
356 	rc = rpcrdma_ep_create(new_xprt);
357 	if (rc)
358 		goto out2;
359 
360 	rc = rpcrdma_buffer_create(new_xprt);
361 	if (rc)
362 		goto out3;
363 
364 	INIT_DELAYED_WORK(&new_xprt->rx_connect_worker,
365 			  xprt_rdma_connect_worker);
366 
367 	xprt->max_payload = frwr_maxpages(new_xprt);
368 	if (xprt->max_payload == 0)
369 		goto out4;
370 	xprt->max_payload <<= PAGE_SHIFT;
371 	dprintk("RPC:       %s: transport data payload maximum: %zu bytes\n",
372 		__func__, xprt->max_payload);
373 
374 	if (!try_module_get(THIS_MODULE))
375 		goto out4;
376 
377 	dprintk("RPC:       %s: %s:%s\n", __func__,
378 		xprt->address_strings[RPC_DISPLAY_ADDR],
379 		xprt->address_strings[RPC_DISPLAY_PORT]);
380 	trace_xprtrdma_create(new_xprt);
381 	return xprt;
382 
383 out4:
384 	rpcrdma_buffer_destroy(&new_xprt->rx_buf);
385 	rc = -ENODEV;
386 out3:
387 	rpcrdma_ep_destroy(new_xprt);
388 out2:
389 	rpcrdma_ia_close(&new_xprt->rx_ia);
390 out1:
391 	trace_xprtrdma_op_destroy(new_xprt);
392 	xprt_rdma_free_addresses(xprt);
393 	xprt_free(xprt);
394 	return ERR_PTR(rc);
395 }
396 
397 /**
398  * xprt_rdma_close - close a transport connection
399  * @xprt: transport context
400  *
401  * Called during autoclose or device removal.
402  *
403  * Caller holds @xprt's send lock to prevent activity on this
404  * transport while the connection is torn down.
405  */
406 void xprt_rdma_close(struct rpc_xprt *xprt)
407 {
408 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
409 	struct rpcrdma_ep *ep = &r_xprt->rx_ep;
410 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
411 
412 	might_sleep();
413 
414 	trace_xprtrdma_op_close(r_xprt);
415 
416 	/* Prevent marshaling and sending of new requests */
417 	xprt_clear_connected(xprt);
418 
419 	if (test_and_clear_bit(RPCRDMA_IAF_REMOVING, &ia->ri_flags)) {
420 		rpcrdma_ia_remove(ia);
421 		goto out;
422 	}
423 
424 	if (ep->rep_connected == -ENODEV)
425 		return;
426 	if (ep->rep_connected > 0)
427 		xprt->reestablish_timeout = 0;
428 	rpcrdma_ep_disconnect(ep, ia);
429 
430 	/* Prepare @xprt for the next connection by reinitializing
431 	 * its credit grant to one (see RFC 8166, Section 3.3.3).
432 	 */
433 	r_xprt->rx_buf.rb_credits = 1;
434 	xprt->cwnd = RPC_CWNDSHIFT;
435 
436 out:
437 	++xprt->connect_cookie;
438 	xprt_disconnect_done(xprt);
439 }
440 
441 /**
442  * xprt_rdma_set_port - update server port with rpcbind result
443  * @xprt: controlling RPC transport
444  * @port: new port value
445  *
446  * Transport connect status is unchanged.
447  */
448 static void
449 xprt_rdma_set_port(struct rpc_xprt *xprt, u16 port)
450 {
451 	struct sockaddr *sap = (struct sockaddr *)&xprt->addr;
452 	char buf[8];
453 
454 	dprintk("RPC:       %s: setting port for xprt %p (%s:%s) to %u\n",
455 		__func__, xprt,
456 		xprt->address_strings[RPC_DISPLAY_ADDR],
457 		xprt->address_strings[RPC_DISPLAY_PORT],
458 		port);
459 
460 	rpc_set_port(sap, port);
461 
462 	kfree(xprt->address_strings[RPC_DISPLAY_PORT]);
463 	snprintf(buf, sizeof(buf), "%u", port);
464 	xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL);
465 
466 	kfree(xprt->address_strings[RPC_DISPLAY_HEX_PORT]);
467 	snprintf(buf, sizeof(buf), "%4hx", port);
468 	xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL);
469 }
470 
471 /**
472  * xprt_rdma_timer - invoked when an RPC times out
473  * @xprt: controlling RPC transport
474  * @task: RPC task that timed out
475  *
476  * Invoked when the transport is still connected, but an RPC
477  * retransmit timeout occurs.
478  *
479  * Since RDMA connections don't have a keep-alive, forcibly
480  * disconnect and retry to connect. This drives full
481  * detection of the network path, and retransmissions of
482  * all pending RPCs.
483  */
484 static void
485 xprt_rdma_timer(struct rpc_xprt *xprt, struct rpc_task *task)
486 {
487 	xprt_force_disconnect(xprt);
488 }
489 
490 /**
491  * xprt_rdma_set_connect_timeout - set timeouts for establishing a connection
492  * @xprt: controlling transport instance
493  * @connect_timeout: reconnect timeout after client disconnects
494  * @reconnect_timeout: reconnect timeout after server disconnects
495  *
496  */
497 static void xprt_rdma_tcp_set_connect_timeout(struct rpc_xprt *xprt,
498 					      unsigned long connect_timeout,
499 					      unsigned long reconnect_timeout)
500 {
501 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
502 
503 	trace_xprtrdma_op_set_cto(r_xprt, connect_timeout, reconnect_timeout);
504 
505 	spin_lock(&xprt->transport_lock);
506 
507 	if (connect_timeout < xprt->connect_timeout) {
508 		struct rpc_timeout to;
509 		unsigned long initval;
510 
511 		to = *xprt->timeout;
512 		initval = connect_timeout;
513 		if (initval < RPCRDMA_INIT_REEST_TO << 1)
514 			initval = RPCRDMA_INIT_REEST_TO << 1;
515 		to.to_initval = initval;
516 		to.to_maxval = initval;
517 		r_xprt->rx_timeout = to;
518 		xprt->timeout = &r_xprt->rx_timeout;
519 		xprt->connect_timeout = connect_timeout;
520 	}
521 
522 	if (reconnect_timeout < xprt->max_reconnect_timeout)
523 		xprt->max_reconnect_timeout = reconnect_timeout;
524 
525 	spin_unlock(&xprt->transport_lock);
526 }
527 
528 /**
529  * xprt_rdma_connect - schedule an attempt to reconnect
530  * @xprt: transport state
531  * @task: RPC scheduler context (unused)
532  *
533  */
534 static void
535 xprt_rdma_connect(struct rpc_xprt *xprt, struct rpc_task *task)
536 {
537 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
538 	unsigned long delay;
539 
540 	trace_xprtrdma_op_connect(r_xprt);
541 
542 	delay = 0;
543 	if (r_xprt->rx_ep.rep_connected != 0) {
544 		delay = xprt_reconnect_delay(xprt);
545 		xprt_reconnect_backoff(xprt, RPCRDMA_INIT_REEST_TO);
546 	}
547 	queue_delayed_work(xprtiod_workqueue, &r_xprt->rx_connect_worker,
548 			   delay);
549 }
550 
551 /**
552  * xprt_rdma_alloc_slot - allocate an rpc_rqst
553  * @xprt: controlling RPC transport
554  * @task: RPC task requesting a fresh rpc_rqst
555  *
556  * tk_status values:
557  *	%0 if task->tk_rqstp points to a fresh rpc_rqst
558  *	%-EAGAIN if no rpc_rqst is available; queued on backlog
559  */
560 static void
561 xprt_rdma_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
562 {
563 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
564 	struct rpcrdma_req *req;
565 
566 	req = rpcrdma_buffer_get(&r_xprt->rx_buf);
567 	if (!req)
568 		goto out_sleep;
569 	task->tk_rqstp = &req->rl_slot;
570 	task->tk_status = 0;
571 	return;
572 
573 out_sleep:
574 	rpc_sleep_on(&xprt->backlog, task, NULL);
575 	task->tk_status = -EAGAIN;
576 }
577 
578 /**
579  * xprt_rdma_free_slot - release an rpc_rqst
580  * @xprt: controlling RPC transport
581  * @rqst: rpc_rqst to release
582  *
583  */
584 static void
585 xprt_rdma_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *rqst)
586 {
587 	struct rpcrdma_xprt *r_xprt =
588 		container_of(xprt, struct rpcrdma_xprt, rx_xprt);
589 
590 	memset(rqst, 0, sizeof(*rqst));
591 	rpcrdma_buffer_put(&r_xprt->rx_buf, rpcr_to_rdmar(rqst));
592 	rpc_wake_up_next(&xprt->backlog);
593 }
594 
595 static bool rpcrdma_check_regbuf(struct rpcrdma_xprt *r_xprt,
596 				 struct rpcrdma_regbuf *rb, size_t size,
597 				 gfp_t flags)
598 {
599 	if (unlikely(rdmab_length(rb) < size)) {
600 		if (!rpcrdma_regbuf_realloc(rb, size, flags))
601 			return false;
602 		r_xprt->rx_stats.hardway_register_count += size;
603 	}
604 	return true;
605 }
606 
607 /**
608  * xprt_rdma_allocate - allocate transport resources for an RPC
609  * @task: RPC task
610  *
611  * Return values:
612  *        0:	Success; rq_buffer points to RPC buffer to use
613  *   ENOMEM:	Out of memory, call again later
614  *      EIO:	A permanent error occurred, do not retry
615  */
616 static int
617 xprt_rdma_allocate(struct rpc_task *task)
618 {
619 	struct rpc_rqst *rqst = task->tk_rqstp;
620 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
621 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
622 	gfp_t flags;
623 
624 	flags = RPCRDMA_DEF_GFP;
625 	if (RPC_IS_SWAPPER(task))
626 		flags = __GFP_MEMALLOC | GFP_NOWAIT | __GFP_NOWARN;
627 
628 	if (!rpcrdma_check_regbuf(r_xprt, req->rl_sendbuf, rqst->rq_callsize,
629 				  flags))
630 		goto out_fail;
631 	if (!rpcrdma_check_regbuf(r_xprt, req->rl_recvbuf, rqst->rq_rcvsize,
632 				  flags))
633 		goto out_fail;
634 
635 	rqst->rq_buffer = rdmab_data(req->rl_sendbuf);
636 	rqst->rq_rbuffer = rdmab_data(req->rl_recvbuf);
637 	trace_xprtrdma_op_allocate(task, req);
638 	return 0;
639 
640 out_fail:
641 	trace_xprtrdma_op_allocate(task, NULL);
642 	return -ENOMEM;
643 }
644 
645 /**
646  * xprt_rdma_free - release resources allocated by xprt_rdma_allocate
647  * @task: RPC task
648  *
649  * Caller guarantees rqst->rq_buffer is non-NULL.
650  */
651 static void
652 xprt_rdma_free(struct rpc_task *task)
653 {
654 	struct rpc_rqst *rqst = task->tk_rqstp;
655 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
656 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
657 
658 	trace_xprtrdma_op_free(task, req);
659 
660 	if (!list_empty(&req->rl_registered))
661 		frwr_unmap_sync(r_xprt, req);
662 
663 	/* XXX: If the RPC is completing because of a signal and
664 	 * not because a reply was received, we ought to ensure
665 	 * that the Send completion has fired, so that memory
666 	 * involved with the Send is not still visible to the NIC.
667 	 */
668 }
669 
670 /**
671  * xprt_rdma_send_request - marshal and send an RPC request
672  * @rqst: RPC message in rq_snd_buf
673  *
674  * Caller holds the transport's write lock.
675  *
676  * Returns:
677  *	%0 if the RPC message has been sent
678  *	%-ENOTCONN if the caller should reconnect and call again
679  *	%-EAGAIN if the caller should call again
680  *	%-ENOBUFS if the caller should call again after a delay
681  *	%-EMSGSIZE if encoding ran out of buffer space. The request
682  *		was not sent. Do not try to send this message again.
683  *	%-EIO if an I/O error occurred. The request was not sent.
684  *		Do not try to send this message again.
685  */
686 static int
687 xprt_rdma_send_request(struct rpc_rqst *rqst)
688 {
689 	struct rpc_xprt *xprt = rqst->rq_xprt;
690 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
691 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
692 	int rc = 0;
693 
694 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
695 	if (unlikely(!rqst->rq_buffer))
696 		return xprt_rdma_bc_send_reply(rqst);
697 #endif	/* CONFIG_SUNRPC_BACKCHANNEL */
698 
699 	if (!xprt_connected(xprt))
700 		return -ENOTCONN;
701 
702 	if (!xprt_request_get_cong(xprt, rqst))
703 		return -EBADSLT;
704 
705 	rc = rpcrdma_marshal_req(r_xprt, rqst);
706 	if (rc < 0)
707 		goto failed_marshal;
708 
709 	/* Must suppress retransmit to maintain credits */
710 	if (rqst->rq_connect_cookie == xprt->connect_cookie)
711 		goto drop_connection;
712 	rqst->rq_xtime = ktime_get();
713 
714 	if (rpcrdma_ep_post(&r_xprt->rx_ia, &r_xprt->rx_ep, req))
715 		goto drop_connection;
716 
717 	rqst->rq_xmit_bytes_sent += rqst->rq_snd_buf.len;
718 
719 	/* An RPC with no reply will throw off credit accounting,
720 	 * so drop the connection to reset the credit grant.
721 	 */
722 	if (!rpc_reply_expected(rqst->rq_task))
723 		goto drop_connection;
724 	return 0;
725 
726 failed_marshal:
727 	if (rc != -ENOTCONN)
728 		return rc;
729 drop_connection:
730 	xprt_rdma_close(xprt);
731 	return -ENOTCONN;
732 }
733 
734 void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
735 {
736 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
737 	long idle_time = 0;
738 
739 	if (xprt_connected(xprt))
740 		idle_time = (long)(jiffies - xprt->last_used) / HZ;
741 
742 	seq_puts(seq, "\txprt:\trdma ");
743 	seq_printf(seq, "%u %lu %lu %lu %ld %lu %lu %lu %llu %llu ",
744 		   0,	/* need a local port? */
745 		   xprt->stat.bind_count,
746 		   xprt->stat.connect_count,
747 		   xprt->stat.connect_time / HZ,
748 		   idle_time,
749 		   xprt->stat.sends,
750 		   xprt->stat.recvs,
751 		   xprt->stat.bad_xids,
752 		   xprt->stat.req_u,
753 		   xprt->stat.bklog_u);
754 	seq_printf(seq, "%lu %lu %lu %llu %llu %llu %llu %lu %lu %lu %lu ",
755 		   r_xprt->rx_stats.read_chunk_count,
756 		   r_xprt->rx_stats.write_chunk_count,
757 		   r_xprt->rx_stats.reply_chunk_count,
758 		   r_xprt->rx_stats.total_rdma_request,
759 		   r_xprt->rx_stats.total_rdma_reply,
760 		   r_xprt->rx_stats.pullup_copy_count,
761 		   r_xprt->rx_stats.fixup_copy_count,
762 		   r_xprt->rx_stats.hardway_register_count,
763 		   r_xprt->rx_stats.failed_marshal_count,
764 		   r_xprt->rx_stats.bad_reply_count,
765 		   r_xprt->rx_stats.nomsg_call_count);
766 	seq_printf(seq, "%lu %lu %lu %lu %lu %lu\n",
767 		   r_xprt->rx_stats.mrs_recycled,
768 		   r_xprt->rx_stats.mrs_orphaned,
769 		   r_xprt->rx_stats.mrs_allocated,
770 		   r_xprt->rx_stats.local_inv_needed,
771 		   r_xprt->rx_stats.empty_sendctx_q,
772 		   r_xprt->rx_stats.reply_waits_for_send);
773 }
774 
775 static int
776 xprt_rdma_enable_swap(struct rpc_xprt *xprt)
777 {
778 	return 0;
779 }
780 
781 static void
782 xprt_rdma_disable_swap(struct rpc_xprt *xprt)
783 {
784 }
785 
786 /*
787  * Plumbing for rpc transport switch and kernel module
788  */
789 
790 static const struct rpc_xprt_ops xprt_rdma_procs = {
791 	.reserve_xprt		= xprt_reserve_xprt_cong,
792 	.release_xprt		= xprt_release_xprt_cong, /* sunrpc/xprt.c */
793 	.alloc_slot		= xprt_rdma_alloc_slot,
794 	.free_slot		= xprt_rdma_free_slot,
795 	.release_request	= xprt_release_rqst_cong,       /* ditto */
796 	.wait_for_reply_request	= xprt_wait_for_reply_request_def, /* ditto */
797 	.timer			= xprt_rdma_timer,
798 	.rpcbind		= rpcb_getport_async,	/* sunrpc/rpcb_clnt.c */
799 	.set_port		= xprt_rdma_set_port,
800 	.connect		= xprt_rdma_connect,
801 	.buf_alloc		= xprt_rdma_allocate,
802 	.buf_free		= xprt_rdma_free,
803 	.send_request		= xprt_rdma_send_request,
804 	.close			= xprt_rdma_close,
805 	.destroy		= xprt_rdma_destroy,
806 	.set_connect_timeout	= xprt_rdma_tcp_set_connect_timeout,
807 	.print_stats		= xprt_rdma_print_stats,
808 	.enable_swap		= xprt_rdma_enable_swap,
809 	.disable_swap		= xprt_rdma_disable_swap,
810 	.inject_disconnect	= xprt_rdma_inject_disconnect,
811 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
812 	.bc_setup		= xprt_rdma_bc_setup,
813 	.bc_maxpayload		= xprt_rdma_bc_maxpayload,
814 	.bc_num_slots		= xprt_rdma_bc_max_slots,
815 	.bc_free_rqst		= xprt_rdma_bc_free_rqst,
816 	.bc_destroy		= xprt_rdma_bc_destroy,
817 #endif
818 };
819 
820 static struct xprt_class xprt_rdma = {
821 	.list			= LIST_HEAD_INIT(xprt_rdma.list),
822 	.name			= "rdma",
823 	.owner			= THIS_MODULE,
824 	.ident			= XPRT_TRANSPORT_RDMA,
825 	.setup			= xprt_setup_rdma,
826 };
827 
828 void xprt_rdma_cleanup(void)
829 {
830 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
831 	if (sunrpc_table_header) {
832 		unregister_sysctl_table(sunrpc_table_header);
833 		sunrpc_table_header = NULL;
834 	}
835 #endif
836 
837 	xprt_unregister_transport(&xprt_rdma);
838 	xprt_unregister_transport(&xprt_rdma_bc);
839 }
840 
841 int xprt_rdma_init(void)
842 {
843 	int rc;
844 
845 	rc = xprt_register_transport(&xprt_rdma);
846 	if (rc)
847 		return rc;
848 
849 	rc = xprt_register_transport(&xprt_rdma_bc);
850 	if (rc) {
851 		xprt_unregister_transport(&xprt_rdma);
852 		return rc;
853 	}
854 
855 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
856 	if (!sunrpc_table_header)
857 		sunrpc_table_header = register_sysctl_table(sunrpc_table);
858 #endif
859 	return 0;
860 }
861