xref: /openbmc/linux/net/sunrpc/xprt.c (revision 808643ea)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  *  linux/net/sunrpc/xprt.c
4  *
5  *  This is a generic RPC call interface supporting congestion avoidance,
6  *  and asynchronous calls.
7  *
8  *  The interface works like this:
9  *
10  *  -	When a process places a call, it allocates a request slot if
11  *	one is available. Otherwise, it sleeps on the backlog queue
12  *	(xprt_reserve).
13  *  -	Next, the caller puts together the RPC message, stuffs it into
14  *	the request struct, and calls xprt_transmit().
15  *  -	xprt_transmit sends the message and installs the caller on the
16  *	transport's wait list. At the same time, if a reply is expected,
17  *	it installs a timer that is run after the packet's timeout has
18  *	expired.
19  *  -	When a packet arrives, the data_ready handler walks the list of
20  *	pending requests for that transport. If a matching XID is found, the
21  *	caller is woken up, and the timer removed.
22  *  -	When no reply arrives within the timeout interval, the timer is
23  *	fired by the kernel and runs xprt_timer(). It either adjusts the
24  *	timeout values (minor timeout) or wakes up the caller with a status
25  *	of -ETIMEDOUT.
26  *  -	When the caller receives a notification from RPC that a reply arrived,
27  *	it should release the RPC slot, and process the reply.
28  *	If the call timed out, it may choose to retry the operation by
29  *	adjusting the initial timeout value, and simply calling rpc_call
30  *	again.
31  *
32  *  Support for async RPC is done through a set of RPC-specific scheduling
33  *  primitives that `transparently' work for processes as well as async
34  *  tasks that rely on callbacks.
35  *
36  *  Copyright (C) 1995-1997, Olaf Kirch <okir@monad.swb.de>
37  *
38  *  Transport switch API copyright (C) 2005, Chuck Lever <cel@netapp.com>
39  */
40 
41 #include <linux/module.h>
42 
43 #include <linux/types.h>
44 #include <linux/interrupt.h>
45 #include <linux/workqueue.h>
46 #include <linux/net.h>
47 #include <linux/ktime.h>
48 
49 #include <linux/sunrpc/clnt.h>
50 #include <linux/sunrpc/metrics.h>
51 #include <linux/sunrpc/bc_xprt.h>
52 #include <linux/rcupdate.h>
53 #include <linux/sched/mm.h>
54 
55 #include <trace/events/sunrpc.h>
56 
57 #include "sunrpc.h"
58 #include "sysfs.h"
59 
60 /*
61  * Local variables
62  */
63 
64 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
65 # define RPCDBG_FACILITY	RPCDBG_XPRT
66 #endif
67 
68 /*
69  * Local functions
70  */
71 static void	 xprt_init(struct rpc_xprt *xprt, struct net *net);
72 static __be32	xprt_alloc_xid(struct rpc_xprt *xprt);
73 static void	 xprt_destroy(struct rpc_xprt *xprt);
74 static void	 xprt_request_init(struct rpc_task *task);
75 
76 static DEFINE_SPINLOCK(xprt_list_lock);
77 static LIST_HEAD(xprt_list);
78 
79 static unsigned long xprt_request_timeout(const struct rpc_rqst *req)
80 {
81 	unsigned long timeout = jiffies + req->rq_timeout;
82 
83 	if (time_before(timeout, req->rq_majortimeo))
84 		return timeout;
85 	return req->rq_majortimeo;
86 }
87 
88 /**
89  * xprt_register_transport - register a transport implementation
90  * @transport: transport to register
91  *
92  * If a transport implementation is loaded as a kernel module, it can
93  * call this interface to make itself known to the RPC client.
94  *
95  * Returns:
96  * 0:		transport successfully registered
97  * -EEXIST:	transport already registered
98  * -EINVAL:	transport module being unloaded
99  */
100 int xprt_register_transport(struct xprt_class *transport)
101 {
102 	struct xprt_class *t;
103 	int result;
104 
105 	result = -EEXIST;
106 	spin_lock(&xprt_list_lock);
107 	list_for_each_entry(t, &xprt_list, list) {
108 		/* don't register the same transport class twice */
109 		if (t->ident == transport->ident)
110 			goto out;
111 	}
112 
113 	list_add_tail(&transport->list, &xprt_list);
114 	printk(KERN_INFO "RPC: Registered %s transport module.\n",
115 	       transport->name);
116 	result = 0;
117 
118 out:
119 	spin_unlock(&xprt_list_lock);
120 	return result;
121 }
122 EXPORT_SYMBOL_GPL(xprt_register_transport);
123 
124 /**
125  * xprt_unregister_transport - unregister a transport implementation
126  * @transport: transport to unregister
127  *
128  * Returns:
129  * 0:		transport successfully unregistered
130  * -ENOENT:	transport never registered
131  */
132 int xprt_unregister_transport(struct xprt_class *transport)
133 {
134 	struct xprt_class *t;
135 	int result;
136 
137 	result = 0;
138 	spin_lock(&xprt_list_lock);
139 	list_for_each_entry(t, &xprt_list, list) {
140 		if (t == transport) {
141 			printk(KERN_INFO
142 				"RPC: Unregistered %s transport module.\n",
143 				transport->name);
144 			list_del_init(&transport->list);
145 			goto out;
146 		}
147 	}
148 	result = -ENOENT;
149 
150 out:
151 	spin_unlock(&xprt_list_lock);
152 	return result;
153 }
154 EXPORT_SYMBOL_GPL(xprt_unregister_transport);
155 
156 static void
157 xprt_class_release(const struct xprt_class *t)
158 {
159 	module_put(t->owner);
160 }
161 
162 static const struct xprt_class *
163 xprt_class_find_by_ident_locked(int ident)
164 {
165 	const struct xprt_class *t;
166 
167 	list_for_each_entry(t, &xprt_list, list) {
168 		if (t->ident != ident)
169 			continue;
170 		if (!try_module_get(t->owner))
171 			continue;
172 		return t;
173 	}
174 	return NULL;
175 }
176 
177 static const struct xprt_class *
178 xprt_class_find_by_ident(int ident)
179 {
180 	const struct xprt_class *t;
181 
182 	spin_lock(&xprt_list_lock);
183 	t = xprt_class_find_by_ident_locked(ident);
184 	spin_unlock(&xprt_list_lock);
185 	return t;
186 }
187 
188 static const struct xprt_class *
189 xprt_class_find_by_netid_locked(const char *netid)
190 {
191 	const struct xprt_class *t;
192 	unsigned int i;
193 
194 	list_for_each_entry(t, &xprt_list, list) {
195 		for (i = 0; t->netid[i][0] != '\0'; i++) {
196 			if (strcmp(t->netid[i], netid) != 0)
197 				continue;
198 			if (!try_module_get(t->owner))
199 				continue;
200 			return t;
201 		}
202 	}
203 	return NULL;
204 }
205 
206 static const struct xprt_class *
207 xprt_class_find_by_netid(const char *netid)
208 {
209 	const struct xprt_class *t;
210 
211 	spin_lock(&xprt_list_lock);
212 	t = xprt_class_find_by_netid_locked(netid);
213 	if (!t) {
214 		spin_unlock(&xprt_list_lock);
215 		request_module("rpc%s", netid);
216 		spin_lock(&xprt_list_lock);
217 		t = xprt_class_find_by_netid_locked(netid);
218 	}
219 	spin_unlock(&xprt_list_lock);
220 	return t;
221 }
222 
223 /**
224  * xprt_find_transport_ident - convert a netid into a transport identifier
225  * @netid: transport to load
226  *
227  * Returns:
228  * > 0:		transport identifier
229  * -ENOENT:	transport module not available
230  */
231 int xprt_find_transport_ident(const char *netid)
232 {
233 	const struct xprt_class *t;
234 	int ret;
235 
236 	t = xprt_class_find_by_netid(netid);
237 	if (!t)
238 		return -ENOENT;
239 	ret = t->ident;
240 	xprt_class_release(t);
241 	return ret;
242 }
243 EXPORT_SYMBOL_GPL(xprt_find_transport_ident);
244 
245 static void xprt_clear_locked(struct rpc_xprt *xprt)
246 {
247 	xprt->snd_task = NULL;
248 	if (!test_bit(XPRT_CLOSE_WAIT, &xprt->state)) {
249 		smp_mb__before_atomic();
250 		clear_bit(XPRT_LOCKED, &xprt->state);
251 		smp_mb__after_atomic();
252 	} else
253 		queue_work(xprtiod_workqueue, &xprt->task_cleanup);
254 }
255 
256 /**
257  * xprt_reserve_xprt - serialize write access to transports
258  * @task: task that is requesting access to the transport
259  * @xprt: pointer to the target transport
260  *
261  * This prevents mixing the payload of separate requests, and prevents
262  * transport connects from colliding with writes.  No congestion control
263  * is provided.
264  */
265 int xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
266 {
267 	struct rpc_rqst *req = task->tk_rqstp;
268 
269 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
270 		if (task == xprt->snd_task)
271 			goto out_locked;
272 		goto out_sleep;
273 	}
274 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
275 		goto out_unlock;
276 	xprt->snd_task = task;
277 
278 out_locked:
279 	trace_xprt_reserve_xprt(xprt, task);
280 	return 1;
281 
282 out_unlock:
283 	xprt_clear_locked(xprt);
284 out_sleep:
285 	task->tk_status = -EAGAIN;
286 	if  (RPC_IS_SOFT(task))
287 		rpc_sleep_on_timeout(&xprt->sending, task, NULL,
288 				xprt_request_timeout(req));
289 	else
290 		rpc_sleep_on(&xprt->sending, task, NULL);
291 	return 0;
292 }
293 EXPORT_SYMBOL_GPL(xprt_reserve_xprt);
294 
295 static bool
296 xprt_need_congestion_window_wait(struct rpc_xprt *xprt)
297 {
298 	return test_bit(XPRT_CWND_WAIT, &xprt->state);
299 }
300 
301 static void
302 xprt_set_congestion_window_wait(struct rpc_xprt *xprt)
303 {
304 	if (!list_empty(&xprt->xmit_queue)) {
305 		/* Peek at head of queue to see if it can make progress */
306 		if (list_first_entry(&xprt->xmit_queue, struct rpc_rqst,
307 					rq_xmit)->rq_cong)
308 			return;
309 	}
310 	set_bit(XPRT_CWND_WAIT, &xprt->state);
311 }
312 
313 static void
314 xprt_test_and_clear_congestion_window_wait(struct rpc_xprt *xprt)
315 {
316 	if (!RPCXPRT_CONGESTED(xprt))
317 		clear_bit(XPRT_CWND_WAIT, &xprt->state);
318 }
319 
320 /*
321  * xprt_reserve_xprt_cong - serialize write access to transports
322  * @task: task that is requesting access to the transport
323  *
324  * Same as xprt_reserve_xprt, but Van Jacobson congestion control is
325  * integrated into the decision of whether a request is allowed to be
326  * woken up and given access to the transport.
327  * Note that the lock is only granted if we know there are free slots.
328  */
329 int xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
330 {
331 	struct rpc_rqst *req = task->tk_rqstp;
332 
333 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
334 		if (task == xprt->snd_task)
335 			goto out_locked;
336 		goto out_sleep;
337 	}
338 	if (req == NULL) {
339 		xprt->snd_task = task;
340 		goto out_locked;
341 	}
342 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
343 		goto out_unlock;
344 	if (!xprt_need_congestion_window_wait(xprt)) {
345 		xprt->snd_task = task;
346 		goto out_locked;
347 	}
348 out_unlock:
349 	xprt_clear_locked(xprt);
350 out_sleep:
351 	task->tk_status = -EAGAIN;
352 	if (RPC_IS_SOFT(task))
353 		rpc_sleep_on_timeout(&xprt->sending, task, NULL,
354 				xprt_request_timeout(req));
355 	else
356 		rpc_sleep_on(&xprt->sending, task, NULL);
357 	return 0;
358 out_locked:
359 	trace_xprt_reserve_cong(xprt, task);
360 	return 1;
361 }
362 EXPORT_SYMBOL_GPL(xprt_reserve_xprt_cong);
363 
364 static inline int xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
365 {
366 	int retval;
367 
368 	if (test_bit(XPRT_LOCKED, &xprt->state) && xprt->snd_task == task)
369 		return 1;
370 	spin_lock(&xprt->transport_lock);
371 	retval = xprt->ops->reserve_xprt(xprt, task);
372 	spin_unlock(&xprt->transport_lock);
373 	return retval;
374 }
375 
376 static bool __xprt_lock_write_func(struct rpc_task *task, void *data)
377 {
378 	struct rpc_xprt *xprt = data;
379 
380 	xprt->snd_task = task;
381 	return true;
382 }
383 
384 static void __xprt_lock_write_next(struct rpc_xprt *xprt)
385 {
386 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
387 		return;
388 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
389 		goto out_unlock;
390 	if (rpc_wake_up_first_on_wq(xprtiod_workqueue, &xprt->sending,
391 				__xprt_lock_write_func, xprt))
392 		return;
393 out_unlock:
394 	xprt_clear_locked(xprt);
395 }
396 
397 static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt)
398 {
399 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
400 		return;
401 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
402 		goto out_unlock;
403 	if (xprt_need_congestion_window_wait(xprt))
404 		goto out_unlock;
405 	if (rpc_wake_up_first_on_wq(xprtiod_workqueue, &xprt->sending,
406 				__xprt_lock_write_func, xprt))
407 		return;
408 out_unlock:
409 	xprt_clear_locked(xprt);
410 }
411 
412 /**
413  * xprt_release_xprt - allow other requests to use a transport
414  * @xprt: transport with other tasks potentially waiting
415  * @task: task that is releasing access to the transport
416  *
417  * Note that "task" can be NULL.  No congestion control is provided.
418  */
419 void xprt_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
420 {
421 	if (xprt->snd_task == task) {
422 		xprt_clear_locked(xprt);
423 		__xprt_lock_write_next(xprt);
424 	}
425 	trace_xprt_release_xprt(xprt, task);
426 }
427 EXPORT_SYMBOL_GPL(xprt_release_xprt);
428 
429 /**
430  * xprt_release_xprt_cong - allow other requests to use a transport
431  * @xprt: transport with other tasks potentially waiting
432  * @task: task that is releasing access to the transport
433  *
434  * Note that "task" can be NULL.  Another task is awoken to use the
435  * transport if the transport's congestion window allows it.
436  */
437 void xprt_release_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
438 {
439 	if (xprt->snd_task == task) {
440 		xprt_clear_locked(xprt);
441 		__xprt_lock_write_next_cong(xprt);
442 	}
443 	trace_xprt_release_cong(xprt, task);
444 }
445 EXPORT_SYMBOL_GPL(xprt_release_xprt_cong);
446 
447 void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
448 {
449 	if (xprt->snd_task != task)
450 		return;
451 	spin_lock(&xprt->transport_lock);
452 	xprt->ops->release_xprt(xprt, task);
453 	spin_unlock(&xprt->transport_lock);
454 }
455 
456 /*
457  * Van Jacobson congestion avoidance. Check if the congestion window
458  * overflowed. Put the task to sleep if this is the case.
459  */
460 static int
461 __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
462 {
463 	if (req->rq_cong)
464 		return 1;
465 	trace_xprt_get_cong(xprt, req->rq_task);
466 	if (RPCXPRT_CONGESTED(xprt)) {
467 		xprt_set_congestion_window_wait(xprt);
468 		return 0;
469 	}
470 	req->rq_cong = 1;
471 	xprt->cong += RPC_CWNDSCALE;
472 	return 1;
473 }
474 
475 /*
476  * Adjust the congestion window, and wake up the next task
477  * that has been sleeping due to congestion
478  */
479 static void
480 __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
481 {
482 	if (!req->rq_cong)
483 		return;
484 	req->rq_cong = 0;
485 	xprt->cong -= RPC_CWNDSCALE;
486 	xprt_test_and_clear_congestion_window_wait(xprt);
487 	trace_xprt_put_cong(xprt, req->rq_task);
488 	__xprt_lock_write_next_cong(xprt);
489 }
490 
491 /**
492  * xprt_request_get_cong - Request congestion control credits
493  * @xprt: pointer to transport
494  * @req: pointer to RPC request
495  *
496  * Useful for transports that require congestion control.
497  */
498 bool
499 xprt_request_get_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
500 {
501 	bool ret = false;
502 
503 	if (req->rq_cong)
504 		return true;
505 	spin_lock(&xprt->transport_lock);
506 	ret = __xprt_get_cong(xprt, req) != 0;
507 	spin_unlock(&xprt->transport_lock);
508 	return ret;
509 }
510 EXPORT_SYMBOL_GPL(xprt_request_get_cong);
511 
512 /**
513  * xprt_release_rqst_cong - housekeeping when request is complete
514  * @task: RPC request that recently completed
515  *
516  * Useful for transports that require congestion control.
517  */
518 void xprt_release_rqst_cong(struct rpc_task *task)
519 {
520 	struct rpc_rqst *req = task->tk_rqstp;
521 
522 	__xprt_put_cong(req->rq_xprt, req);
523 }
524 EXPORT_SYMBOL_GPL(xprt_release_rqst_cong);
525 
526 static void xprt_clear_congestion_window_wait_locked(struct rpc_xprt *xprt)
527 {
528 	if (test_and_clear_bit(XPRT_CWND_WAIT, &xprt->state))
529 		__xprt_lock_write_next_cong(xprt);
530 }
531 
532 /*
533  * Clear the congestion window wait flag and wake up the next
534  * entry on xprt->sending
535  */
536 static void
537 xprt_clear_congestion_window_wait(struct rpc_xprt *xprt)
538 {
539 	if (test_and_clear_bit(XPRT_CWND_WAIT, &xprt->state)) {
540 		spin_lock(&xprt->transport_lock);
541 		__xprt_lock_write_next_cong(xprt);
542 		spin_unlock(&xprt->transport_lock);
543 	}
544 }
545 
546 /**
547  * xprt_adjust_cwnd - adjust transport congestion window
548  * @xprt: pointer to xprt
549  * @task: recently completed RPC request used to adjust window
550  * @result: result code of completed RPC request
551  *
552  * The transport code maintains an estimate on the maximum number of out-
553  * standing RPC requests, using a smoothed version of the congestion
554  * avoidance implemented in 44BSD. This is basically the Van Jacobson
555  * congestion algorithm: If a retransmit occurs, the congestion window is
556  * halved; otherwise, it is incremented by 1/cwnd when
557  *
558  *	-	a reply is received and
559  *	-	a full number of requests are outstanding and
560  *	-	the congestion window hasn't been updated recently.
561  */
562 void xprt_adjust_cwnd(struct rpc_xprt *xprt, struct rpc_task *task, int result)
563 {
564 	struct rpc_rqst *req = task->tk_rqstp;
565 	unsigned long cwnd = xprt->cwnd;
566 
567 	if (result >= 0 && cwnd <= xprt->cong) {
568 		/* The (cwnd >> 1) term makes sure
569 		 * the result gets rounded properly. */
570 		cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
571 		if (cwnd > RPC_MAXCWND(xprt))
572 			cwnd = RPC_MAXCWND(xprt);
573 		__xprt_lock_write_next_cong(xprt);
574 	} else if (result == -ETIMEDOUT) {
575 		cwnd >>= 1;
576 		if (cwnd < RPC_CWNDSCALE)
577 			cwnd = RPC_CWNDSCALE;
578 	}
579 	dprintk("RPC:       cong %ld, cwnd was %ld, now %ld\n",
580 			xprt->cong, xprt->cwnd, cwnd);
581 	xprt->cwnd = cwnd;
582 	__xprt_put_cong(xprt, req);
583 }
584 EXPORT_SYMBOL_GPL(xprt_adjust_cwnd);
585 
586 /**
587  * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue
588  * @xprt: transport with waiting tasks
589  * @status: result code to plant in each task before waking it
590  *
591  */
592 void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status)
593 {
594 	if (status < 0)
595 		rpc_wake_up_status(&xprt->pending, status);
596 	else
597 		rpc_wake_up(&xprt->pending);
598 }
599 EXPORT_SYMBOL_GPL(xprt_wake_pending_tasks);
600 
601 /**
602  * xprt_wait_for_buffer_space - wait for transport output buffer to clear
603  * @xprt: transport
604  *
605  * Note that we only set the timer for the case of RPC_IS_SOFT(), since
606  * we don't in general want to force a socket disconnection due to
607  * an incomplete RPC call transmission.
608  */
609 void xprt_wait_for_buffer_space(struct rpc_xprt *xprt)
610 {
611 	set_bit(XPRT_WRITE_SPACE, &xprt->state);
612 }
613 EXPORT_SYMBOL_GPL(xprt_wait_for_buffer_space);
614 
615 static bool
616 xprt_clear_write_space_locked(struct rpc_xprt *xprt)
617 {
618 	if (test_and_clear_bit(XPRT_WRITE_SPACE, &xprt->state)) {
619 		__xprt_lock_write_next(xprt);
620 		dprintk("RPC:       write space: waking waiting task on "
621 				"xprt %p\n", xprt);
622 		return true;
623 	}
624 	return false;
625 }
626 
627 /**
628  * xprt_write_space - wake the task waiting for transport output buffer space
629  * @xprt: transport with waiting tasks
630  *
631  * Can be called in a soft IRQ context, so xprt_write_space never sleeps.
632  */
633 bool xprt_write_space(struct rpc_xprt *xprt)
634 {
635 	bool ret;
636 
637 	if (!test_bit(XPRT_WRITE_SPACE, &xprt->state))
638 		return false;
639 	spin_lock(&xprt->transport_lock);
640 	ret = xprt_clear_write_space_locked(xprt);
641 	spin_unlock(&xprt->transport_lock);
642 	return ret;
643 }
644 EXPORT_SYMBOL_GPL(xprt_write_space);
645 
646 static unsigned long xprt_abs_ktime_to_jiffies(ktime_t abstime)
647 {
648 	s64 delta = ktime_to_ns(ktime_get() - abstime);
649 	return likely(delta >= 0) ?
650 		jiffies - nsecs_to_jiffies(delta) :
651 		jiffies + nsecs_to_jiffies(-delta);
652 }
653 
654 static unsigned long xprt_calc_majortimeo(struct rpc_rqst *req)
655 {
656 	const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
657 	unsigned long majortimeo = req->rq_timeout;
658 
659 	if (to->to_exponential)
660 		majortimeo <<= to->to_retries;
661 	else
662 		majortimeo += to->to_increment * to->to_retries;
663 	if (majortimeo > to->to_maxval || majortimeo == 0)
664 		majortimeo = to->to_maxval;
665 	return majortimeo;
666 }
667 
668 static void xprt_reset_majortimeo(struct rpc_rqst *req)
669 {
670 	req->rq_majortimeo += xprt_calc_majortimeo(req);
671 }
672 
673 static void xprt_reset_minortimeo(struct rpc_rqst *req)
674 {
675 	req->rq_minortimeo += req->rq_timeout;
676 }
677 
678 static void xprt_init_majortimeo(struct rpc_task *task, struct rpc_rqst *req)
679 {
680 	unsigned long time_init;
681 	struct rpc_xprt *xprt = req->rq_xprt;
682 
683 	if (likely(xprt && xprt_connected(xprt)))
684 		time_init = jiffies;
685 	else
686 		time_init = xprt_abs_ktime_to_jiffies(task->tk_start);
687 	req->rq_timeout = task->tk_client->cl_timeout->to_initval;
688 	req->rq_majortimeo = time_init + xprt_calc_majortimeo(req);
689 	req->rq_minortimeo = time_init + req->rq_timeout;
690 }
691 
692 /**
693  * xprt_adjust_timeout - adjust timeout values for next retransmit
694  * @req: RPC request containing parameters to use for the adjustment
695  *
696  */
697 int xprt_adjust_timeout(struct rpc_rqst *req)
698 {
699 	struct rpc_xprt *xprt = req->rq_xprt;
700 	const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
701 	int status = 0;
702 
703 	if (time_before(jiffies, req->rq_majortimeo)) {
704 		if (time_before(jiffies, req->rq_minortimeo))
705 			return status;
706 		if (to->to_exponential)
707 			req->rq_timeout <<= 1;
708 		else
709 			req->rq_timeout += to->to_increment;
710 		if (to->to_maxval && req->rq_timeout >= to->to_maxval)
711 			req->rq_timeout = to->to_maxval;
712 		req->rq_retries++;
713 	} else {
714 		req->rq_timeout = to->to_initval;
715 		req->rq_retries = 0;
716 		xprt_reset_majortimeo(req);
717 		/* Reset the RTT counters == "slow start" */
718 		spin_lock(&xprt->transport_lock);
719 		rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
720 		spin_unlock(&xprt->transport_lock);
721 		status = -ETIMEDOUT;
722 	}
723 	xprt_reset_minortimeo(req);
724 
725 	if (req->rq_timeout == 0) {
726 		printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
727 		req->rq_timeout = 5 * HZ;
728 	}
729 	return status;
730 }
731 
732 static void xprt_autoclose(struct work_struct *work)
733 {
734 	struct rpc_xprt *xprt =
735 		container_of(work, struct rpc_xprt, task_cleanup);
736 	unsigned int pflags = memalloc_nofs_save();
737 
738 	trace_xprt_disconnect_auto(xprt);
739 	clear_bit(XPRT_CLOSE_WAIT, &xprt->state);
740 	xprt->ops->close(xprt);
741 	xprt_release_write(xprt, NULL);
742 	wake_up_bit(&xprt->state, XPRT_LOCKED);
743 	memalloc_nofs_restore(pflags);
744 }
745 
746 /**
747  * xprt_disconnect_done - mark a transport as disconnected
748  * @xprt: transport to flag for disconnect
749  *
750  */
751 void xprt_disconnect_done(struct rpc_xprt *xprt)
752 {
753 	trace_xprt_disconnect_done(xprt);
754 	spin_lock(&xprt->transport_lock);
755 	xprt_clear_connected(xprt);
756 	xprt_clear_write_space_locked(xprt);
757 	xprt_clear_congestion_window_wait_locked(xprt);
758 	xprt_wake_pending_tasks(xprt, -ENOTCONN);
759 	spin_unlock(&xprt->transport_lock);
760 }
761 EXPORT_SYMBOL_GPL(xprt_disconnect_done);
762 
763 /**
764  * xprt_force_disconnect - force a transport to disconnect
765  * @xprt: transport to disconnect
766  *
767  */
768 void xprt_force_disconnect(struct rpc_xprt *xprt)
769 {
770 	trace_xprt_disconnect_force(xprt);
771 
772 	/* Don't race with the test_bit() in xprt_clear_locked() */
773 	spin_lock(&xprt->transport_lock);
774 	set_bit(XPRT_CLOSE_WAIT, &xprt->state);
775 	/* Try to schedule an autoclose RPC call */
776 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
777 		queue_work(xprtiod_workqueue, &xprt->task_cleanup);
778 	else if (xprt->snd_task)
779 		rpc_wake_up_queued_task_set_status(&xprt->pending,
780 				xprt->snd_task, -ENOTCONN);
781 	spin_unlock(&xprt->transport_lock);
782 }
783 EXPORT_SYMBOL_GPL(xprt_force_disconnect);
784 
785 static unsigned int
786 xprt_connect_cookie(struct rpc_xprt *xprt)
787 {
788 	return READ_ONCE(xprt->connect_cookie);
789 }
790 
791 static bool
792 xprt_request_retransmit_after_disconnect(struct rpc_task *task)
793 {
794 	struct rpc_rqst *req = task->tk_rqstp;
795 	struct rpc_xprt *xprt = req->rq_xprt;
796 
797 	return req->rq_connect_cookie != xprt_connect_cookie(xprt) ||
798 		!xprt_connected(xprt);
799 }
800 
801 /**
802  * xprt_conditional_disconnect - force a transport to disconnect
803  * @xprt: transport to disconnect
804  * @cookie: 'connection cookie'
805  *
806  * This attempts to break the connection if and only if 'cookie' matches
807  * the current transport 'connection cookie'. It ensures that we don't
808  * try to break the connection more than once when we need to retransmit
809  * a batch of RPC requests.
810  *
811  */
812 void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie)
813 {
814 	/* Don't race with the test_bit() in xprt_clear_locked() */
815 	spin_lock(&xprt->transport_lock);
816 	if (cookie != xprt->connect_cookie)
817 		goto out;
818 	if (test_bit(XPRT_CLOSING, &xprt->state))
819 		goto out;
820 	set_bit(XPRT_CLOSE_WAIT, &xprt->state);
821 	/* Try to schedule an autoclose RPC call */
822 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
823 		queue_work(xprtiod_workqueue, &xprt->task_cleanup);
824 	xprt_wake_pending_tasks(xprt, -EAGAIN);
825 out:
826 	spin_unlock(&xprt->transport_lock);
827 }
828 
829 static bool
830 xprt_has_timer(const struct rpc_xprt *xprt)
831 {
832 	return xprt->idle_timeout != 0;
833 }
834 
835 static void
836 xprt_schedule_autodisconnect(struct rpc_xprt *xprt)
837 	__must_hold(&xprt->transport_lock)
838 {
839 	xprt->last_used = jiffies;
840 	if (RB_EMPTY_ROOT(&xprt->recv_queue) && xprt_has_timer(xprt))
841 		mod_timer(&xprt->timer, xprt->last_used + xprt->idle_timeout);
842 }
843 
844 static void
845 xprt_init_autodisconnect(struct timer_list *t)
846 {
847 	struct rpc_xprt *xprt = from_timer(xprt, t, timer);
848 
849 	if (!RB_EMPTY_ROOT(&xprt->recv_queue))
850 		return;
851 	/* Reset xprt->last_used to avoid connect/autodisconnect cycling */
852 	xprt->last_used = jiffies;
853 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
854 		return;
855 	queue_work(xprtiod_workqueue, &xprt->task_cleanup);
856 }
857 
858 bool xprt_lock_connect(struct rpc_xprt *xprt,
859 		struct rpc_task *task,
860 		void *cookie)
861 {
862 	bool ret = false;
863 
864 	spin_lock(&xprt->transport_lock);
865 	if (!test_bit(XPRT_LOCKED, &xprt->state))
866 		goto out;
867 	if (xprt->snd_task != task)
868 		goto out;
869 	xprt->snd_task = cookie;
870 	ret = true;
871 out:
872 	spin_unlock(&xprt->transport_lock);
873 	return ret;
874 }
875 
876 void xprt_unlock_connect(struct rpc_xprt *xprt, void *cookie)
877 {
878 	spin_lock(&xprt->transport_lock);
879 	if (xprt->snd_task != cookie)
880 		goto out;
881 	if (!test_bit(XPRT_LOCKED, &xprt->state))
882 		goto out;
883 	xprt->snd_task =NULL;
884 	xprt->ops->release_xprt(xprt, NULL);
885 	xprt_schedule_autodisconnect(xprt);
886 out:
887 	spin_unlock(&xprt->transport_lock);
888 	wake_up_bit(&xprt->state, XPRT_LOCKED);
889 }
890 
891 /**
892  * xprt_connect - schedule a transport connect operation
893  * @task: RPC task that is requesting the connect
894  *
895  */
896 void xprt_connect(struct rpc_task *task)
897 {
898 	struct rpc_xprt	*xprt = task->tk_rqstp->rq_xprt;
899 
900 	trace_xprt_connect(xprt);
901 
902 	if (!xprt_bound(xprt)) {
903 		task->tk_status = -EAGAIN;
904 		return;
905 	}
906 	if (!xprt_lock_write(xprt, task))
907 		return;
908 
909 	if (test_and_clear_bit(XPRT_CLOSE_WAIT, &xprt->state)) {
910 		trace_xprt_disconnect_cleanup(xprt);
911 		xprt->ops->close(xprt);
912 	}
913 
914 	if (!xprt_connected(xprt)) {
915 		task->tk_rqstp->rq_connect_cookie = xprt->connect_cookie;
916 		rpc_sleep_on_timeout(&xprt->pending, task, NULL,
917 				xprt_request_timeout(task->tk_rqstp));
918 
919 		if (test_bit(XPRT_CLOSING, &xprt->state))
920 			return;
921 		if (xprt_test_and_set_connecting(xprt))
922 			return;
923 		/* Race breaker */
924 		if (!xprt_connected(xprt)) {
925 			xprt->stat.connect_start = jiffies;
926 			xprt->ops->connect(xprt, task);
927 		} else {
928 			xprt_clear_connecting(xprt);
929 			task->tk_status = 0;
930 			rpc_wake_up_queued_task(&xprt->pending, task);
931 		}
932 	}
933 	xprt_release_write(xprt, task);
934 }
935 
936 /**
937  * xprt_reconnect_delay - compute the wait before scheduling a connect
938  * @xprt: transport instance
939  *
940  */
941 unsigned long xprt_reconnect_delay(const struct rpc_xprt *xprt)
942 {
943 	unsigned long start, now = jiffies;
944 
945 	start = xprt->stat.connect_start + xprt->reestablish_timeout;
946 	if (time_after(start, now))
947 		return start - now;
948 	return 0;
949 }
950 EXPORT_SYMBOL_GPL(xprt_reconnect_delay);
951 
952 /**
953  * xprt_reconnect_backoff - compute the new re-establish timeout
954  * @xprt: transport instance
955  * @init_to: initial reestablish timeout
956  *
957  */
958 void xprt_reconnect_backoff(struct rpc_xprt *xprt, unsigned long init_to)
959 {
960 	xprt->reestablish_timeout <<= 1;
961 	if (xprt->reestablish_timeout > xprt->max_reconnect_timeout)
962 		xprt->reestablish_timeout = xprt->max_reconnect_timeout;
963 	if (xprt->reestablish_timeout < init_to)
964 		xprt->reestablish_timeout = init_to;
965 }
966 EXPORT_SYMBOL_GPL(xprt_reconnect_backoff);
967 
968 enum xprt_xid_rb_cmp {
969 	XID_RB_EQUAL,
970 	XID_RB_LEFT,
971 	XID_RB_RIGHT,
972 };
973 static enum xprt_xid_rb_cmp
974 xprt_xid_cmp(__be32 xid1, __be32 xid2)
975 {
976 	if (xid1 == xid2)
977 		return XID_RB_EQUAL;
978 	if ((__force u32)xid1 < (__force u32)xid2)
979 		return XID_RB_LEFT;
980 	return XID_RB_RIGHT;
981 }
982 
983 static struct rpc_rqst *
984 xprt_request_rb_find(struct rpc_xprt *xprt, __be32 xid)
985 {
986 	struct rb_node *n = xprt->recv_queue.rb_node;
987 	struct rpc_rqst *req;
988 
989 	while (n != NULL) {
990 		req = rb_entry(n, struct rpc_rqst, rq_recv);
991 		switch (xprt_xid_cmp(xid, req->rq_xid)) {
992 		case XID_RB_LEFT:
993 			n = n->rb_left;
994 			break;
995 		case XID_RB_RIGHT:
996 			n = n->rb_right;
997 			break;
998 		case XID_RB_EQUAL:
999 			return req;
1000 		}
1001 	}
1002 	return NULL;
1003 }
1004 
1005 static void
1006 xprt_request_rb_insert(struct rpc_xprt *xprt, struct rpc_rqst *new)
1007 {
1008 	struct rb_node **p = &xprt->recv_queue.rb_node;
1009 	struct rb_node *n = NULL;
1010 	struct rpc_rqst *req;
1011 
1012 	while (*p != NULL) {
1013 		n = *p;
1014 		req = rb_entry(n, struct rpc_rqst, rq_recv);
1015 		switch(xprt_xid_cmp(new->rq_xid, req->rq_xid)) {
1016 		case XID_RB_LEFT:
1017 			p = &n->rb_left;
1018 			break;
1019 		case XID_RB_RIGHT:
1020 			p = &n->rb_right;
1021 			break;
1022 		case XID_RB_EQUAL:
1023 			WARN_ON_ONCE(new != req);
1024 			return;
1025 		}
1026 	}
1027 	rb_link_node(&new->rq_recv, n, p);
1028 	rb_insert_color(&new->rq_recv, &xprt->recv_queue);
1029 }
1030 
1031 static void
1032 xprt_request_rb_remove(struct rpc_xprt *xprt, struct rpc_rqst *req)
1033 {
1034 	rb_erase(&req->rq_recv, &xprt->recv_queue);
1035 }
1036 
1037 /**
1038  * xprt_lookup_rqst - find an RPC request corresponding to an XID
1039  * @xprt: transport on which the original request was transmitted
1040  * @xid: RPC XID of incoming reply
1041  *
1042  * Caller holds xprt->queue_lock.
1043  */
1044 struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, __be32 xid)
1045 {
1046 	struct rpc_rqst *entry;
1047 
1048 	entry = xprt_request_rb_find(xprt, xid);
1049 	if (entry != NULL) {
1050 		trace_xprt_lookup_rqst(xprt, xid, 0);
1051 		entry->rq_rtt = ktime_sub(ktime_get(), entry->rq_xtime);
1052 		return entry;
1053 	}
1054 
1055 	dprintk("RPC:       xprt_lookup_rqst did not find xid %08x\n",
1056 			ntohl(xid));
1057 	trace_xprt_lookup_rqst(xprt, xid, -ENOENT);
1058 	xprt->stat.bad_xids++;
1059 	return NULL;
1060 }
1061 EXPORT_SYMBOL_GPL(xprt_lookup_rqst);
1062 
1063 static bool
1064 xprt_is_pinned_rqst(struct rpc_rqst *req)
1065 {
1066 	return atomic_read(&req->rq_pin) != 0;
1067 }
1068 
1069 /**
1070  * xprt_pin_rqst - Pin a request on the transport receive list
1071  * @req: Request to pin
1072  *
1073  * Caller must ensure this is atomic with the call to xprt_lookup_rqst()
1074  * so should be holding xprt->queue_lock.
1075  */
1076 void xprt_pin_rqst(struct rpc_rqst *req)
1077 {
1078 	atomic_inc(&req->rq_pin);
1079 }
1080 EXPORT_SYMBOL_GPL(xprt_pin_rqst);
1081 
1082 /**
1083  * xprt_unpin_rqst - Unpin a request on the transport receive list
1084  * @req: Request to pin
1085  *
1086  * Caller should be holding xprt->queue_lock.
1087  */
1088 void xprt_unpin_rqst(struct rpc_rqst *req)
1089 {
1090 	if (!test_bit(RPC_TASK_MSG_PIN_WAIT, &req->rq_task->tk_runstate)) {
1091 		atomic_dec(&req->rq_pin);
1092 		return;
1093 	}
1094 	if (atomic_dec_and_test(&req->rq_pin))
1095 		wake_up_var(&req->rq_pin);
1096 }
1097 EXPORT_SYMBOL_GPL(xprt_unpin_rqst);
1098 
1099 static void xprt_wait_on_pinned_rqst(struct rpc_rqst *req)
1100 {
1101 	wait_var_event(&req->rq_pin, !xprt_is_pinned_rqst(req));
1102 }
1103 
1104 static bool
1105 xprt_request_data_received(struct rpc_task *task)
1106 {
1107 	return !test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) &&
1108 		READ_ONCE(task->tk_rqstp->rq_reply_bytes_recvd) != 0;
1109 }
1110 
1111 static bool
1112 xprt_request_need_enqueue_receive(struct rpc_task *task, struct rpc_rqst *req)
1113 {
1114 	return !test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) &&
1115 		READ_ONCE(task->tk_rqstp->rq_reply_bytes_recvd) == 0;
1116 }
1117 
1118 /**
1119  * xprt_request_enqueue_receive - Add an request to the receive queue
1120  * @task: RPC task
1121  *
1122  */
1123 void
1124 xprt_request_enqueue_receive(struct rpc_task *task)
1125 {
1126 	struct rpc_rqst *req = task->tk_rqstp;
1127 	struct rpc_xprt *xprt = req->rq_xprt;
1128 
1129 	if (!xprt_request_need_enqueue_receive(task, req))
1130 		return;
1131 
1132 	xprt_request_prepare(task->tk_rqstp);
1133 	spin_lock(&xprt->queue_lock);
1134 
1135 	/* Update the softirq receive buffer */
1136 	memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
1137 			sizeof(req->rq_private_buf));
1138 
1139 	/* Add request to the receive list */
1140 	xprt_request_rb_insert(xprt, req);
1141 	set_bit(RPC_TASK_NEED_RECV, &task->tk_runstate);
1142 	spin_unlock(&xprt->queue_lock);
1143 
1144 	/* Turn off autodisconnect */
1145 	del_singleshot_timer_sync(&xprt->timer);
1146 }
1147 
1148 /**
1149  * xprt_request_dequeue_receive_locked - Remove a request from the receive queue
1150  * @task: RPC task
1151  *
1152  * Caller must hold xprt->queue_lock.
1153  */
1154 static void
1155 xprt_request_dequeue_receive_locked(struct rpc_task *task)
1156 {
1157 	struct rpc_rqst *req = task->tk_rqstp;
1158 
1159 	if (test_and_clear_bit(RPC_TASK_NEED_RECV, &task->tk_runstate))
1160 		xprt_request_rb_remove(req->rq_xprt, req);
1161 }
1162 
1163 /**
1164  * xprt_update_rtt - Update RPC RTT statistics
1165  * @task: RPC request that recently completed
1166  *
1167  * Caller holds xprt->queue_lock.
1168  */
1169 void xprt_update_rtt(struct rpc_task *task)
1170 {
1171 	struct rpc_rqst *req = task->tk_rqstp;
1172 	struct rpc_rtt *rtt = task->tk_client->cl_rtt;
1173 	unsigned int timer = task->tk_msg.rpc_proc->p_timer;
1174 	long m = usecs_to_jiffies(ktime_to_us(req->rq_rtt));
1175 
1176 	if (timer) {
1177 		if (req->rq_ntrans == 1)
1178 			rpc_update_rtt(rtt, timer, m);
1179 		rpc_set_timeo(rtt, timer, req->rq_ntrans - 1);
1180 	}
1181 }
1182 EXPORT_SYMBOL_GPL(xprt_update_rtt);
1183 
1184 /**
1185  * xprt_complete_rqst - called when reply processing is complete
1186  * @task: RPC request that recently completed
1187  * @copied: actual number of bytes received from the transport
1188  *
1189  * Caller holds xprt->queue_lock.
1190  */
1191 void xprt_complete_rqst(struct rpc_task *task, int copied)
1192 {
1193 	struct rpc_rqst *req = task->tk_rqstp;
1194 	struct rpc_xprt *xprt = req->rq_xprt;
1195 
1196 	xprt->stat.recvs++;
1197 
1198 	req->rq_private_buf.len = copied;
1199 	/* Ensure all writes are done before we update */
1200 	/* req->rq_reply_bytes_recvd */
1201 	smp_wmb();
1202 	req->rq_reply_bytes_recvd = copied;
1203 	xprt_request_dequeue_receive_locked(task);
1204 	rpc_wake_up_queued_task(&xprt->pending, task);
1205 }
1206 EXPORT_SYMBOL_GPL(xprt_complete_rqst);
1207 
1208 static void xprt_timer(struct rpc_task *task)
1209 {
1210 	struct rpc_rqst *req = task->tk_rqstp;
1211 	struct rpc_xprt *xprt = req->rq_xprt;
1212 
1213 	if (task->tk_status != -ETIMEDOUT)
1214 		return;
1215 
1216 	trace_xprt_timer(xprt, req->rq_xid, task->tk_status);
1217 	if (!req->rq_reply_bytes_recvd) {
1218 		if (xprt->ops->timer)
1219 			xprt->ops->timer(xprt, task);
1220 	} else
1221 		task->tk_status = 0;
1222 }
1223 
1224 /**
1225  * xprt_wait_for_reply_request_def - wait for reply
1226  * @task: pointer to rpc_task
1227  *
1228  * Set a request's retransmit timeout based on the transport's
1229  * default timeout parameters.  Used by transports that don't adjust
1230  * the retransmit timeout based on round-trip time estimation,
1231  * and put the task to sleep on the pending queue.
1232  */
1233 void xprt_wait_for_reply_request_def(struct rpc_task *task)
1234 {
1235 	struct rpc_rqst *req = task->tk_rqstp;
1236 
1237 	rpc_sleep_on_timeout(&req->rq_xprt->pending, task, xprt_timer,
1238 			xprt_request_timeout(req));
1239 }
1240 EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_def);
1241 
1242 /**
1243  * xprt_wait_for_reply_request_rtt - wait for reply using RTT estimator
1244  * @task: pointer to rpc_task
1245  *
1246  * Set a request's retransmit timeout using the RTT estimator,
1247  * and put the task to sleep on the pending queue.
1248  */
1249 void xprt_wait_for_reply_request_rtt(struct rpc_task *task)
1250 {
1251 	int timer = task->tk_msg.rpc_proc->p_timer;
1252 	struct rpc_clnt *clnt = task->tk_client;
1253 	struct rpc_rtt *rtt = clnt->cl_rtt;
1254 	struct rpc_rqst *req = task->tk_rqstp;
1255 	unsigned long max_timeout = clnt->cl_timeout->to_maxval;
1256 	unsigned long timeout;
1257 
1258 	timeout = rpc_calc_rto(rtt, timer);
1259 	timeout <<= rpc_ntimeo(rtt, timer) + req->rq_retries;
1260 	if (timeout > max_timeout || timeout == 0)
1261 		timeout = max_timeout;
1262 	rpc_sleep_on_timeout(&req->rq_xprt->pending, task, xprt_timer,
1263 			jiffies + timeout);
1264 }
1265 EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_rtt);
1266 
1267 /**
1268  * xprt_request_wait_receive - wait for the reply to an RPC request
1269  * @task: RPC task about to send a request
1270  *
1271  */
1272 void xprt_request_wait_receive(struct rpc_task *task)
1273 {
1274 	struct rpc_rqst *req = task->tk_rqstp;
1275 	struct rpc_xprt *xprt = req->rq_xprt;
1276 
1277 	if (!test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate))
1278 		return;
1279 	/*
1280 	 * Sleep on the pending queue if we're expecting a reply.
1281 	 * The spinlock ensures atomicity between the test of
1282 	 * req->rq_reply_bytes_recvd, and the call to rpc_sleep_on().
1283 	 */
1284 	spin_lock(&xprt->queue_lock);
1285 	if (test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate)) {
1286 		xprt->ops->wait_for_reply_request(task);
1287 		/*
1288 		 * Send an extra queue wakeup call if the
1289 		 * connection was dropped in case the call to
1290 		 * rpc_sleep_on() raced.
1291 		 */
1292 		if (xprt_request_retransmit_after_disconnect(task))
1293 			rpc_wake_up_queued_task_set_status(&xprt->pending,
1294 					task, -ENOTCONN);
1295 	}
1296 	spin_unlock(&xprt->queue_lock);
1297 }
1298 
1299 static bool
1300 xprt_request_need_enqueue_transmit(struct rpc_task *task, struct rpc_rqst *req)
1301 {
1302 	return !test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate);
1303 }
1304 
1305 /**
1306  * xprt_request_enqueue_transmit - queue a task for transmission
1307  * @task: pointer to rpc_task
1308  *
1309  * Add a task to the transmission queue.
1310  */
1311 void
1312 xprt_request_enqueue_transmit(struct rpc_task *task)
1313 {
1314 	struct rpc_rqst *pos, *req = task->tk_rqstp;
1315 	struct rpc_xprt *xprt = req->rq_xprt;
1316 
1317 	if (xprt_request_need_enqueue_transmit(task, req)) {
1318 		req->rq_bytes_sent = 0;
1319 		spin_lock(&xprt->queue_lock);
1320 		/*
1321 		 * Requests that carry congestion control credits are added
1322 		 * to the head of the list to avoid starvation issues.
1323 		 */
1324 		if (req->rq_cong) {
1325 			xprt_clear_congestion_window_wait(xprt);
1326 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1327 				if (pos->rq_cong)
1328 					continue;
1329 				/* Note: req is added _before_ pos */
1330 				list_add_tail(&req->rq_xmit, &pos->rq_xmit);
1331 				INIT_LIST_HEAD(&req->rq_xmit2);
1332 				goto out;
1333 			}
1334 		} else if (RPC_IS_SWAPPER(task)) {
1335 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1336 				if (pos->rq_cong || pos->rq_bytes_sent)
1337 					continue;
1338 				if (RPC_IS_SWAPPER(pos->rq_task))
1339 					continue;
1340 				/* Note: req is added _before_ pos */
1341 				list_add_tail(&req->rq_xmit, &pos->rq_xmit);
1342 				INIT_LIST_HEAD(&req->rq_xmit2);
1343 				goto out;
1344 			}
1345 		} else if (!req->rq_seqno) {
1346 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1347 				if (pos->rq_task->tk_owner != task->tk_owner)
1348 					continue;
1349 				list_add_tail(&req->rq_xmit2, &pos->rq_xmit2);
1350 				INIT_LIST_HEAD(&req->rq_xmit);
1351 				goto out;
1352 			}
1353 		}
1354 		list_add_tail(&req->rq_xmit, &xprt->xmit_queue);
1355 		INIT_LIST_HEAD(&req->rq_xmit2);
1356 out:
1357 		atomic_long_inc(&xprt->xmit_queuelen);
1358 		set_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate);
1359 		spin_unlock(&xprt->queue_lock);
1360 	}
1361 }
1362 
1363 /**
1364  * xprt_request_dequeue_transmit_locked - remove a task from the transmission queue
1365  * @task: pointer to rpc_task
1366  *
1367  * Remove a task from the transmission queue
1368  * Caller must hold xprt->queue_lock
1369  */
1370 static void
1371 xprt_request_dequeue_transmit_locked(struct rpc_task *task)
1372 {
1373 	struct rpc_rqst *req = task->tk_rqstp;
1374 
1375 	if (!test_and_clear_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1376 		return;
1377 	if (!list_empty(&req->rq_xmit)) {
1378 		list_del(&req->rq_xmit);
1379 		if (!list_empty(&req->rq_xmit2)) {
1380 			struct rpc_rqst *next = list_first_entry(&req->rq_xmit2,
1381 					struct rpc_rqst, rq_xmit2);
1382 			list_del(&req->rq_xmit2);
1383 			list_add_tail(&next->rq_xmit, &next->rq_xprt->xmit_queue);
1384 		}
1385 	} else
1386 		list_del(&req->rq_xmit2);
1387 	atomic_long_dec(&req->rq_xprt->xmit_queuelen);
1388 }
1389 
1390 /**
1391  * xprt_request_dequeue_transmit - remove a task from the transmission queue
1392  * @task: pointer to rpc_task
1393  *
1394  * Remove a task from the transmission queue
1395  */
1396 static void
1397 xprt_request_dequeue_transmit(struct rpc_task *task)
1398 {
1399 	struct rpc_rqst *req = task->tk_rqstp;
1400 	struct rpc_xprt *xprt = req->rq_xprt;
1401 
1402 	spin_lock(&xprt->queue_lock);
1403 	xprt_request_dequeue_transmit_locked(task);
1404 	spin_unlock(&xprt->queue_lock);
1405 }
1406 
1407 /**
1408  * xprt_request_dequeue_xprt - remove a task from the transmit+receive queue
1409  * @task: pointer to rpc_task
1410  *
1411  * Remove a task from the transmit and receive queues, and ensure that
1412  * it is not pinned by the receive work item.
1413  */
1414 void
1415 xprt_request_dequeue_xprt(struct rpc_task *task)
1416 {
1417 	struct rpc_rqst	*req = task->tk_rqstp;
1418 	struct rpc_xprt *xprt = req->rq_xprt;
1419 
1420 	if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate) ||
1421 	    test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) ||
1422 	    xprt_is_pinned_rqst(req)) {
1423 		spin_lock(&xprt->queue_lock);
1424 		xprt_request_dequeue_transmit_locked(task);
1425 		xprt_request_dequeue_receive_locked(task);
1426 		while (xprt_is_pinned_rqst(req)) {
1427 			set_bit(RPC_TASK_MSG_PIN_WAIT, &task->tk_runstate);
1428 			spin_unlock(&xprt->queue_lock);
1429 			xprt_wait_on_pinned_rqst(req);
1430 			spin_lock(&xprt->queue_lock);
1431 			clear_bit(RPC_TASK_MSG_PIN_WAIT, &task->tk_runstate);
1432 		}
1433 		spin_unlock(&xprt->queue_lock);
1434 	}
1435 }
1436 
1437 /**
1438  * xprt_request_prepare - prepare an encoded request for transport
1439  * @req: pointer to rpc_rqst
1440  *
1441  * Calls into the transport layer to do whatever is needed to prepare
1442  * the request for transmission or receive.
1443  */
1444 void
1445 xprt_request_prepare(struct rpc_rqst *req)
1446 {
1447 	struct rpc_xprt *xprt = req->rq_xprt;
1448 
1449 	if (xprt->ops->prepare_request)
1450 		xprt->ops->prepare_request(req);
1451 }
1452 
1453 /**
1454  * xprt_request_need_retransmit - Test if a task needs retransmission
1455  * @task: pointer to rpc_task
1456  *
1457  * Test for whether a connection breakage requires the task to retransmit
1458  */
1459 bool
1460 xprt_request_need_retransmit(struct rpc_task *task)
1461 {
1462 	return xprt_request_retransmit_after_disconnect(task);
1463 }
1464 
1465 /**
1466  * xprt_prepare_transmit - reserve the transport before sending a request
1467  * @task: RPC task about to send a request
1468  *
1469  */
1470 bool xprt_prepare_transmit(struct rpc_task *task)
1471 {
1472 	struct rpc_rqst	*req = task->tk_rqstp;
1473 	struct rpc_xprt	*xprt = req->rq_xprt;
1474 
1475 	if (!xprt_lock_write(xprt, task)) {
1476 		/* Race breaker: someone may have transmitted us */
1477 		if (!test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1478 			rpc_wake_up_queued_task_set_status(&xprt->sending,
1479 					task, 0);
1480 		return false;
1481 
1482 	}
1483 	return true;
1484 }
1485 
1486 void xprt_end_transmit(struct rpc_task *task)
1487 {
1488 	struct rpc_xprt	*xprt = task->tk_rqstp->rq_xprt;
1489 
1490 	xprt_inject_disconnect(xprt);
1491 	xprt_release_write(xprt, task);
1492 }
1493 
1494 /**
1495  * xprt_request_transmit - send an RPC request on a transport
1496  * @req: pointer to request to transmit
1497  * @snd_task: RPC task that owns the transport lock
1498  *
1499  * This performs the transmission of a single request.
1500  * Note that if the request is not the same as snd_task, then it
1501  * does need to be pinned.
1502  * Returns '0' on success.
1503  */
1504 static int
1505 xprt_request_transmit(struct rpc_rqst *req, struct rpc_task *snd_task)
1506 {
1507 	struct rpc_xprt *xprt = req->rq_xprt;
1508 	struct rpc_task *task = req->rq_task;
1509 	unsigned int connect_cookie;
1510 	int is_retrans = RPC_WAS_SENT(task);
1511 	int status;
1512 
1513 	if (!req->rq_bytes_sent) {
1514 		if (xprt_request_data_received(task)) {
1515 			status = 0;
1516 			goto out_dequeue;
1517 		}
1518 		/* Verify that our message lies in the RPCSEC_GSS window */
1519 		if (rpcauth_xmit_need_reencode(task)) {
1520 			status = -EBADMSG;
1521 			goto out_dequeue;
1522 		}
1523 		if (RPC_SIGNALLED(task)) {
1524 			status = -ERESTARTSYS;
1525 			goto out_dequeue;
1526 		}
1527 	}
1528 
1529 	/*
1530 	 * Update req->rq_ntrans before transmitting to avoid races with
1531 	 * xprt_update_rtt(), which needs to know that it is recording a
1532 	 * reply to the first transmission.
1533 	 */
1534 	req->rq_ntrans++;
1535 
1536 	trace_rpc_xdr_sendto(task, &req->rq_snd_buf);
1537 	connect_cookie = xprt->connect_cookie;
1538 	status = xprt->ops->send_request(req);
1539 	if (status != 0) {
1540 		req->rq_ntrans--;
1541 		trace_xprt_transmit(req, status);
1542 		return status;
1543 	}
1544 
1545 	if (is_retrans) {
1546 		task->tk_client->cl_stats->rpcretrans++;
1547 		trace_xprt_retransmit(req);
1548 	}
1549 
1550 	xprt_inject_disconnect(xprt);
1551 
1552 	task->tk_flags |= RPC_TASK_SENT;
1553 	spin_lock(&xprt->transport_lock);
1554 
1555 	xprt->stat.sends++;
1556 	xprt->stat.req_u += xprt->stat.sends - xprt->stat.recvs;
1557 	xprt->stat.bklog_u += xprt->backlog.qlen;
1558 	xprt->stat.sending_u += xprt->sending.qlen;
1559 	xprt->stat.pending_u += xprt->pending.qlen;
1560 	spin_unlock(&xprt->transport_lock);
1561 
1562 	req->rq_connect_cookie = connect_cookie;
1563 out_dequeue:
1564 	trace_xprt_transmit(req, status);
1565 	xprt_request_dequeue_transmit(task);
1566 	rpc_wake_up_queued_task_set_status(&xprt->sending, task, status);
1567 	return status;
1568 }
1569 
1570 /**
1571  * xprt_transmit - send an RPC request on a transport
1572  * @task: controlling RPC task
1573  *
1574  * Attempts to drain the transmit queue. On exit, either the transport
1575  * signalled an error that needs to be handled before transmission can
1576  * resume, or @task finished transmitting, and detected that it already
1577  * received a reply.
1578  */
1579 void
1580 xprt_transmit(struct rpc_task *task)
1581 {
1582 	struct rpc_rqst *next, *req = task->tk_rqstp;
1583 	struct rpc_xprt	*xprt = req->rq_xprt;
1584 	int counter, status;
1585 
1586 	spin_lock(&xprt->queue_lock);
1587 	counter = 0;
1588 	while (!list_empty(&xprt->xmit_queue)) {
1589 		if (++counter == 20)
1590 			break;
1591 		next = list_first_entry(&xprt->xmit_queue,
1592 				struct rpc_rqst, rq_xmit);
1593 		xprt_pin_rqst(next);
1594 		spin_unlock(&xprt->queue_lock);
1595 		status = xprt_request_transmit(next, task);
1596 		if (status == -EBADMSG && next != req)
1597 			status = 0;
1598 		spin_lock(&xprt->queue_lock);
1599 		xprt_unpin_rqst(next);
1600 		if (status == 0) {
1601 			if (!xprt_request_data_received(task) ||
1602 			    test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1603 				continue;
1604 		} else if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1605 			task->tk_status = status;
1606 		break;
1607 	}
1608 	spin_unlock(&xprt->queue_lock);
1609 }
1610 
1611 static void xprt_complete_request_init(struct rpc_task *task)
1612 {
1613 	if (task->tk_rqstp)
1614 		xprt_request_init(task);
1615 }
1616 
1617 void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task)
1618 {
1619 	set_bit(XPRT_CONGESTED, &xprt->state);
1620 	rpc_sleep_on(&xprt->backlog, task, xprt_complete_request_init);
1621 }
1622 EXPORT_SYMBOL_GPL(xprt_add_backlog);
1623 
1624 static bool __xprt_set_rq(struct rpc_task *task, void *data)
1625 {
1626 	struct rpc_rqst *req = data;
1627 
1628 	if (task->tk_rqstp == NULL) {
1629 		memset(req, 0, sizeof(*req));	/* mark unused */
1630 		task->tk_rqstp = req;
1631 		return true;
1632 	}
1633 	return false;
1634 }
1635 
1636 bool xprt_wake_up_backlog(struct rpc_xprt *xprt, struct rpc_rqst *req)
1637 {
1638 	if (rpc_wake_up_first(&xprt->backlog, __xprt_set_rq, req) == NULL) {
1639 		clear_bit(XPRT_CONGESTED, &xprt->state);
1640 		return false;
1641 	}
1642 	return true;
1643 }
1644 EXPORT_SYMBOL_GPL(xprt_wake_up_backlog);
1645 
1646 static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task)
1647 {
1648 	bool ret = false;
1649 
1650 	if (!test_bit(XPRT_CONGESTED, &xprt->state))
1651 		goto out;
1652 	spin_lock(&xprt->reserve_lock);
1653 	if (test_bit(XPRT_CONGESTED, &xprt->state)) {
1654 		xprt_add_backlog(xprt, task);
1655 		ret = true;
1656 	}
1657 	spin_unlock(&xprt->reserve_lock);
1658 out:
1659 	return ret;
1660 }
1661 
1662 static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt)
1663 {
1664 	struct rpc_rqst *req = ERR_PTR(-EAGAIN);
1665 
1666 	if (xprt->num_reqs >= xprt->max_reqs)
1667 		goto out;
1668 	++xprt->num_reqs;
1669 	spin_unlock(&xprt->reserve_lock);
1670 	req = kzalloc(sizeof(struct rpc_rqst), GFP_NOFS);
1671 	spin_lock(&xprt->reserve_lock);
1672 	if (req != NULL)
1673 		goto out;
1674 	--xprt->num_reqs;
1675 	req = ERR_PTR(-ENOMEM);
1676 out:
1677 	return req;
1678 }
1679 
1680 static bool xprt_dynamic_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1681 {
1682 	if (xprt->num_reqs > xprt->min_reqs) {
1683 		--xprt->num_reqs;
1684 		kfree(req);
1685 		return true;
1686 	}
1687 	return false;
1688 }
1689 
1690 void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
1691 {
1692 	struct rpc_rqst *req;
1693 
1694 	spin_lock(&xprt->reserve_lock);
1695 	if (!list_empty(&xprt->free)) {
1696 		req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
1697 		list_del(&req->rq_list);
1698 		goto out_init_req;
1699 	}
1700 	req = xprt_dynamic_alloc_slot(xprt);
1701 	if (!IS_ERR(req))
1702 		goto out_init_req;
1703 	switch (PTR_ERR(req)) {
1704 	case -ENOMEM:
1705 		dprintk("RPC:       dynamic allocation of request slot "
1706 				"failed! Retrying\n");
1707 		task->tk_status = -ENOMEM;
1708 		break;
1709 	case -EAGAIN:
1710 		xprt_add_backlog(xprt, task);
1711 		dprintk("RPC:       waiting for request slot\n");
1712 		fallthrough;
1713 	default:
1714 		task->tk_status = -EAGAIN;
1715 	}
1716 	spin_unlock(&xprt->reserve_lock);
1717 	return;
1718 out_init_req:
1719 	xprt->stat.max_slots = max_t(unsigned int, xprt->stat.max_slots,
1720 				     xprt->num_reqs);
1721 	spin_unlock(&xprt->reserve_lock);
1722 
1723 	task->tk_status = 0;
1724 	task->tk_rqstp = req;
1725 }
1726 EXPORT_SYMBOL_GPL(xprt_alloc_slot);
1727 
1728 void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1729 {
1730 	spin_lock(&xprt->reserve_lock);
1731 	if (!xprt_wake_up_backlog(xprt, req) &&
1732 	    !xprt_dynamic_free_slot(xprt, req)) {
1733 		memset(req, 0, sizeof(*req));	/* mark unused */
1734 		list_add(&req->rq_list, &xprt->free);
1735 	}
1736 	spin_unlock(&xprt->reserve_lock);
1737 }
1738 EXPORT_SYMBOL_GPL(xprt_free_slot);
1739 
1740 static void xprt_free_all_slots(struct rpc_xprt *xprt)
1741 {
1742 	struct rpc_rqst *req;
1743 	while (!list_empty(&xprt->free)) {
1744 		req = list_first_entry(&xprt->free, struct rpc_rqst, rq_list);
1745 		list_del(&req->rq_list);
1746 		kfree(req);
1747 	}
1748 }
1749 
1750 static DEFINE_IDA(rpc_xprt_ids);
1751 
1752 void xprt_cleanup_ids(void)
1753 {
1754 	ida_destroy(&rpc_xprt_ids);
1755 }
1756 
1757 static int xprt_alloc_id(struct rpc_xprt *xprt)
1758 {
1759 	int id;
1760 
1761 	id = ida_simple_get(&rpc_xprt_ids, 0, 0, GFP_KERNEL);
1762 	if (id < 0)
1763 		return id;
1764 
1765 	xprt->id = id;
1766 	return 0;
1767 }
1768 
1769 static void xprt_free_id(struct rpc_xprt *xprt)
1770 {
1771 	ida_simple_remove(&rpc_xprt_ids, xprt->id);
1772 }
1773 
1774 struct rpc_xprt *xprt_alloc(struct net *net, size_t size,
1775 		unsigned int num_prealloc,
1776 		unsigned int max_alloc)
1777 {
1778 	struct rpc_xprt *xprt;
1779 	struct rpc_rqst *req;
1780 	int i;
1781 
1782 	xprt = kzalloc(size, GFP_KERNEL);
1783 	if (xprt == NULL)
1784 		goto out;
1785 
1786 	xprt_alloc_id(xprt);
1787 	xprt_init(xprt, net);
1788 
1789 	for (i = 0; i < num_prealloc; i++) {
1790 		req = kzalloc(sizeof(struct rpc_rqst), GFP_KERNEL);
1791 		if (!req)
1792 			goto out_free;
1793 		list_add(&req->rq_list, &xprt->free);
1794 	}
1795 	if (max_alloc > num_prealloc)
1796 		xprt->max_reqs = max_alloc;
1797 	else
1798 		xprt->max_reqs = num_prealloc;
1799 	xprt->min_reqs = num_prealloc;
1800 	xprt->num_reqs = num_prealloc;
1801 
1802 	return xprt;
1803 
1804 out_free:
1805 	xprt_free(xprt);
1806 out:
1807 	return NULL;
1808 }
1809 EXPORT_SYMBOL_GPL(xprt_alloc);
1810 
1811 void xprt_free(struct rpc_xprt *xprt)
1812 {
1813 	put_net(xprt->xprt_net);
1814 	xprt_free_all_slots(xprt);
1815 	xprt_free_id(xprt);
1816 	rpc_sysfs_xprt_destroy(xprt);
1817 	kfree_rcu(xprt, rcu);
1818 }
1819 EXPORT_SYMBOL_GPL(xprt_free);
1820 
1821 static void
1822 xprt_init_connect_cookie(struct rpc_rqst *req, struct rpc_xprt *xprt)
1823 {
1824 	req->rq_connect_cookie = xprt_connect_cookie(xprt) - 1;
1825 }
1826 
1827 static __be32
1828 xprt_alloc_xid(struct rpc_xprt *xprt)
1829 {
1830 	__be32 xid;
1831 
1832 	spin_lock(&xprt->reserve_lock);
1833 	xid = (__force __be32)xprt->xid++;
1834 	spin_unlock(&xprt->reserve_lock);
1835 	return xid;
1836 }
1837 
1838 static void
1839 xprt_init_xid(struct rpc_xprt *xprt)
1840 {
1841 	xprt->xid = prandom_u32();
1842 }
1843 
1844 static void
1845 xprt_request_init(struct rpc_task *task)
1846 {
1847 	struct rpc_xprt *xprt = task->tk_xprt;
1848 	struct rpc_rqst	*req = task->tk_rqstp;
1849 
1850 	req->rq_task	= task;
1851 	req->rq_xprt    = xprt;
1852 	req->rq_buffer  = NULL;
1853 	req->rq_xid	= xprt_alloc_xid(xprt);
1854 	xprt_init_connect_cookie(req, xprt);
1855 	req->rq_snd_buf.len = 0;
1856 	req->rq_snd_buf.buflen = 0;
1857 	req->rq_rcv_buf.len = 0;
1858 	req->rq_rcv_buf.buflen = 0;
1859 	req->rq_snd_buf.bvec = NULL;
1860 	req->rq_rcv_buf.bvec = NULL;
1861 	req->rq_release_snd_buf = NULL;
1862 	xprt_init_majortimeo(task, req);
1863 
1864 	trace_xprt_reserve(req);
1865 }
1866 
1867 static void
1868 xprt_do_reserve(struct rpc_xprt *xprt, struct rpc_task *task)
1869 {
1870 	xprt->ops->alloc_slot(xprt, task);
1871 	if (task->tk_rqstp != NULL)
1872 		xprt_request_init(task);
1873 }
1874 
1875 /**
1876  * xprt_reserve - allocate an RPC request slot
1877  * @task: RPC task requesting a slot allocation
1878  *
1879  * If the transport is marked as being congested, or if no more
1880  * slots are available, place the task on the transport's
1881  * backlog queue.
1882  */
1883 void xprt_reserve(struct rpc_task *task)
1884 {
1885 	struct rpc_xprt *xprt = task->tk_xprt;
1886 
1887 	task->tk_status = 0;
1888 	if (task->tk_rqstp != NULL)
1889 		return;
1890 
1891 	task->tk_status = -EAGAIN;
1892 	if (!xprt_throttle_congested(xprt, task))
1893 		xprt_do_reserve(xprt, task);
1894 }
1895 
1896 /**
1897  * xprt_retry_reserve - allocate an RPC request slot
1898  * @task: RPC task requesting a slot allocation
1899  *
1900  * If no more slots are available, place the task on the transport's
1901  * backlog queue.
1902  * Note that the only difference with xprt_reserve is that we now
1903  * ignore the value of the XPRT_CONGESTED flag.
1904  */
1905 void xprt_retry_reserve(struct rpc_task *task)
1906 {
1907 	struct rpc_xprt *xprt = task->tk_xprt;
1908 
1909 	task->tk_status = 0;
1910 	if (task->tk_rqstp != NULL)
1911 		return;
1912 
1913 	task->tk_status = -EAGAIN;
1914 	xprt_do_reserve(xprt, task);
1915 }
1916 
1917 /**
1918  * xprt_release - release an RPC request slot
1919  * @task: task which is finished with the slot
1920  *
1921  */
1922 void xprt_release(struct rpc_task *task)
1923 {
1924 	struct rpc_xprt	*xprt;
1925 	struct rpc_rqst	*req = task->tk_rqstp;
1926 
1927 	if (req == NULL) {
1928 		if (task->tk_client) {
1929 			xprt = task->tk_xprt;
1930 			xprt_release_write(xprt, task);
1931 		}
1932 		return;
1933 	}
1934 
1935 	xprt = req->rq_xprt;
1936 	xprt_request_dequeue_xprt(task);
1937 	spin_lock(&xprt->transport_lock);
1938 	xprt->ops->release_xprt(xprt, task);
1939 	if (xprt->ops->release_request)
1940 		xprt->ops->release_request(task);
1941 	xprt_schedule_autodisconnect(xprt);
1942 	spin_unlock(&xprt->transport_lock);
1943 	if (req->rq_buffer)
1944 		xprt->ops->buf_free(task);
1945 	xdr_free_bvec(&req->rq_rcv_buf);
1946 	xdr_free_bvec(&req->rq_snd_buf);
1947 	if (req->rq_cred != NULL)
1948 		put_rpccred(req->rq_cred);
1949 	if (req->rq_release_snd_buf)
1950 		req->rq_release_snd_buf(req);
1951 
1952 	task->tk_rqstp = NULL;
1953 	if (likely(!bc_prealloc(req)))
1954 		xprt->ops->free_slot(xprt, req);
1955 	else
1956 		xprt_free_bc_request(req);
1957 }
1958 
1959 #ifdef CONFIG_SUNRPC_BACKCHANNEL
1960 void
1961 xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task)
1962 {
1963 	struct xdr_buf *xbufp = &req->rq_snd_buf;
1964 
1965 	task->tk_rqstp = req;
1966 	req->rq_task = task;
1967 	xprt_init_connect_cookie(req, req->rq_xprt);
1968 	/*
1969 	 * Set up the xdr_buf length.
1970 	 * This also indicates that the buffer is XDR encoded already.
1971 	 */
1972 	xbufp->len = xbufp->head[0].iov_len + xbufp->page_len +
1973 		xbufp->tail[0].iov_len;
1974 }
1975 #endif
1976 
1977 static void xprt_init(struct rpc_xprt *xprt, struct net *net)
1978 {
1979 	kref_init(&xprt->kref);
1980 
1981 	spin_lock_init(&xprt->transport_lock);
1982 	spin_lock_init(&xprt->reserve_lock);
1983 	spin_lock_init(&xprt->queue_lock);
1984 
1985 	INIT_LIST_HEAD(&xprt->free);
1986 	xprt->recv_queue = RB_ROOT;
1987 	INIT_LIST_HEAD(&xprt->xmit_queue);
1988 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
1989 	spin_lock_init(&xprt->bc_pa_lock);
1990 	INIT_LIST_HEAD(&xprt->bc_pa_list);
1991 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
1992 	INIT_LIST_HEAD(&xprt->xprt_switch);
1993 
1994 	xprt->last_used = jiffies;
1995 	xprt->cwnd = RPC_INITCWND;
1996 	xprt->bind_index = 0;
1997 
1998 	rpc_init_wait_queue(&xprt->binding, "xprt_binding");
1999 	rpc_init_wait_queue(&xprt->pending, "xprt_pending");
2000 	rpc_init_wait_queue(&xprt->sending, "xprt_sending");
2001 	rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
2002 
2003 	xprt_init_xid(xprt);
2004 
2005 	xprt->xprt_net = get_net(net);
2006 }
2007 
2008 /**
2009  * xprt_create_transport - create an RPC transport
2010  * @args: rpc transport creation arguments
2011  *
2012  */
2013 struct rpc_xprt *xprt_create_transport(struct xprt_create *args)
2014 {
2015 	struct rpc_xprt	*xprt;
2016 	const struct xprt_class *t;
2017 
2018 	t = xprt_class_find_by_ident(args->ident);
2019 	if (!t) {
2020 		dprintk("RPC: transport (%d) not supported\n", args->ident);
2021 		return ERR_PTR(-EIO);
2022 	}
2023 
2024 	xprt = t->setup(args);
2025 	xprt_class_release(t);
2026 
2027 	if (IS_ERR(xprt))
2028 		goto out;
2029 	if (args->flags & XPRT_CREATE_NO_IDLE_TIMEOUT)
2030 		xprt->idle_timeout = 0;
2031 	INIT_WORK(&xprt->task_cleanup, xprt_autoclose);
2032 	if (xprt_has_timer(xprt))
2033 		timer_setup(&xprt->timer, xprt_init_autodisconnect, 0);
2034 	else
2035 		timer_setup(&xprt->timer, NULL, 0);
2036 
2037 	if (strlen(args->servername) > RPC_MAXNETNAMELEN) {
2038 		xprt_destroy(xprt);
2039 		return ERR_PTR(-EINVAL);
2040 	}
2041 	xprt->servername = kstrdup(args->servername, GFP_KERNEL);
2042 	if (xprt->servername == NULL) {
2043 		xprt_destroy(xprt);
2044 		return ERR_PTR(-ENOMEM);
2045 	}
2046 
2047 	rpc_xprt_debugfs_register(xprt);
2048 
2049 	trace_xprt_create(xprt);
2050 out:
2051 	return xprt;
2052 }
2053 
2054 static void xprt_destroy_cb(struct work_struct *work)
2055 {
2056 	struct rpc_xprt *xprt =
2057 		container_of(work, struct rpc_xprt, task_cleanup);
2058 
2059 	trace_xprt_destroy(xprt);
2060 
2061 	rpc_xprt_debugfs_unregister(xprt);
2062 	rpc_destroy_wait_queue(&xprt->binding);
2063 	rpc_destroy_wait_queue(&xprt->pending);
2064 	rpc_destroy_wait_queue(&xprt->sending);
2065 	rpc_destroy_wait_queue(&xprt->backlog);
2066 	kfree(xprt->servername);
2067 	/*
2068 	 * Destroy any existing back channel
2069 	 */
2070 	xprt_destroy_backchannel(xprt, UINT_MAX);
2071 
2072 	/*
2073 	 * Tear down transport state and free the rpc_xprt
2074 	 */
2075 	xprt->ops->destroy(xprt);
2076 }
2077 
2078 /**
2079  * xprt_destroy - destroy an RPC transport, killing off all requests.
2080  * @xprt: transport to destroy
2081  *
2082  */
2083 static void xprt_destroy(struct rpc_xprt *xprt)
2084 {
2085 	/*
2086 	 * Exclude transport connect/disconnect handlers and autoclose
2087 	 */
2088 	wait_on_bit_lock(&xprt->state, XPRT_LOCKED, TASK_UNINTERRUPTIBLE);
2089 
2090 	del_timer_sync(&xprt->timer);
2091 
2092 	/*
2093 	 * Destroy sockets etc from the system workqueue so they can
2094 	 * safely flush receive work running on rpciod.
2095 	 */
2096 	INIT_WORK(&xprt->task_cleanup, xprt_destroy_cb);
2097 	schedule_work(&xprt->task_cleanup);
2098 }
2099 
2100 static void xprt_destroy_kref(struct kref *kref)
2101 {
2102 	xprt_destroy(container_of(kref, struct rpc_xprt, kref));
2103 }
2104 
2105 /**
2106  * xprt_get - return a reference to an RPC transport.
2107  * @xprt: pointer to the transport
2108  *
2109  */
2110 struct rpc_xprt *xprt_get(struct rpc_xprt *xprt)
2111 {
2112 	if (xprt != NULL && kref_get_unless_zero(&xprt->kref))
2113 		return xprt;
2114 	return NULL;
2115 }
2116 EXPORT_SYMBOL_GPL(xprt_get);
2117 
2118 /**
2119  * xprt_put - release a reference to an RPC transport.
2120  * @xprt: pointer to the transport
2121  *
2122  */
2123 void xprt_put(struct rpc_xprt *xprt)
2124 {
2125 	if (xprt != NULL)
2126 		kref_put(&xprt->kref, xprt_destroy_kref);
2127 }
2128 EXPORT_SYMBOL_GPL(xprt_put);
2129