xref: /openbmc/linux/net/sunrpc/xprt.c (revision b38269ec)
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 #include "fail.h"
60 
61 /*
62  * Local variables
63  */
64 
65 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
66 # define RPCDBG_FACILITY	RPCDBG_XPRT
67 #endif
68 
69 /*
70  * Local functions
71  */
72 static void	 xprt_init(struct rpc_xprt *xprt, struct net *net);
73 static __be32	xprt_alloc_xid(struct rpc_xprt *xprt);
74 static void	 xprt_destroy(struct rpc_xprt *xprt);
75 static void	 xprt_request_init(struct rpc_task *task);
76 
77 static DEFINE_SPINLOCK(xprt_list_lock);
78 static LIST_HEAD(xprt_list);
79 
80 static unsigned long xprt_request_timeout(const struct rpc_rqst *req)
81 {
82 	unsigned long timeout = jiffies + req->rq_timeout;
83 
84 	if (time_before(timeout, req->rq_majortimeo))
85 		return timeout;
86 	return req->rq_majortimeo;
87 }
88 
89 /**
90  * xprt_register_transport - register a transport implementation
91  * @transport: transport to register
92  *
93  * If a transport implementation is loaded as a kernel module, it can
94  * call this interface to make itself known to the RPC client.
95  *
96  * Returns:
97  * 0:		transport successfully registered
98  * -EEXIST:	transport already registered
99  * -EINVAL:	transport module being unloaded
100  */
101 int xprt_register_transport(struct xprt_class *transport)
102 {
103 	struct xprt_class *t;
104 	int result;
105 
106 	result = -EEXIST;
107 	spin_lock(&xprt_list_lock);
108 	list_for_each_entry(t, &xprt_list, list) {
109 		/* don't register the same transport class twice */
110 		if (t->ident == transport->ident)
111 			goto out;
112 	}
113 
114 	list_add_tail(&transport->list, &xprt_list);
115 	printk(KERN_INFO "RPC: Registered %s transport module.\n",
116 	       transport->name);
117 	result = 0;
118 
119 out:
120 	spin_unlock(&xprt_list_lock);
121 	return result;
122 }
123 EXPORT_SYMBOL_GPL(xprt_register_transport);
124 
125 /**
126  * xprt_unregister_transport - unregister a transport implementation
127  * @transport: transport to unregister
128  *
129  * Returns:
130  * 0:		transport successfully unregistered
131  * -ENOENT:	transport never registered
132  */
133 int xprt_unregister_transport(struct xprt_class *transport)
134 {
135 	struct xprt_class *t;
136 	int result;
137 
138 	result = 0;
139 	spin_lock(&xprt_list_lock);
140 	list_for_each_entry(t, &xprt_list, list) {
141 		if (t == transport) {
142 			printk(KERN_INFO
143 				"RPC: Unregistered %s transport module.\n",
144 				transport->name);
145 			list_del_init(&transport->list);
146 			goto out;
147 		}
148 	}
149 	result = -ENOENT;
150 
151 out:
152 	spin_unlock(&xprt_list_lock);
153 	return result;
154 }
155 EXPORT_SYMBOL_GPL(xprt_unregister_transport);
156 
157 static void
158 xprt_class_release(const struct xprt_class *t)
159 {
160 	module_put(t->owner);
161 }
162 
163 static const struct xprt_class *
164 xprt_class_find_by_ident_locked(int ident)
165 {
166 	const struct xprt_class *t;
167 
168 	list_for_each_entry(t, &xprt_list, list) {
169 		if (t->ident != ident)
170 			continue;
171 		if (!try_module_get(t->owner))
172 			continue;
173 		return t;
174 	}
175 	return NULL;
176 }
177 
178 static const struct xprt_class *
179 xprt_class_find_by_ident(int ident)
180 {
181 	const struct xprt_class *t;
182 
183 	spin_lock(&xprt_list_lock);
184 	t = xprt_class_find_by_ident_locked(ident);
185 	spin_unlock(&xprt_list_lock);
186 	return t;
187 }
188 
189 static const struct xprt_class *
190 xprt_class_find_by_netid_locked(const char *netid)
191 {
192 	const struct xprt_class *t;
193 	unsigned int i;
194 
195 	list_for_each_entry(t, &xprt_list, list) {
196 		for (i = 0; t->netid[i][0] != '\0'; i++) {
197 			if (strcmp(t->netid[i], netid) != 0)
198 				continue;
199 			if (!try_module_get(t->owner))
200 				continue;
201 			return t;
202 		}
203 	}
204 	return NULL;
205 }
206 
207 static const struct xprt_class *
208 xprt_class_find_by_netid(const char *netid)
209 {
210 	const struct xprt_class *t;
211 
212 	spin_lock(&xprt_list_lock);
213 	t = xprt_class_find_by_netid_locked(netid);
214 	if (!t) {
215 		spin_unlock(&xprt_list_lock);
216 		request_module("rpc%s", netid);
217 		spin_lock(&xprt_list_lock);
218 		t = xprt_class_find_by_netid_locked(netid);
219 	}
220 	spin_unlock(&xprt_list_lock);
221 	return t;
222 }
223 
224 /**
225  * xprt_find_transport_ident - convert a netid into a transport identifier
226  * @netid: transport to load
227  *
228  * Returns:
229  * > 0:		transport identifier
230  * -ENOENT:	transport module not available
231  */
232 int xprt_find_transport_ident(const char *netid)
233 {
234 	const struct xprt_class *t;
235 	int ret;
236 
237 	t = xprt_class_find_by_netid(netid);
238 	if (!t)
239 		return -ENOENT;
240 	ret = t->ident;
241 	xprt_class_release(t);
242 	return ret;
243 }
244 EXPORT_SYMBOL_GPL(xprt_find_transport_ident);
245 
246 static void xprt_clear_locked(struct rpc_xprt *xprt)
247 {
248 	xprt->snd_task = NULL;
249 	if (!test_bit(XPRT_CLOSE_WAIT, &xprt->state)) {
250 		smp_mb__before_atomic();
251 		clear_bit(XPRT_LOCKED, &xprt->state);
252 		smp_mb__after_atomic();
253 	} else
254 		queue_work(xprtiod_workqueue, &xprt->task_cleanup);
255 }
256 
257 /**
258  * xprt_reserve_xprt - serialize write access to transports
259  * @task: task that is requesting access to the transport
260  * @xprt: pointer to the target transport
261  *
262  * This prevents mixing the payload of separate requests, and prevents
263  * transport connects from colliding with writes.  No congestion control
264  * is provided.
265  */
266 int xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
267 {
268 	struct rpc_rqst *req = task->tk_rqstp;
269 
270 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
271 		if (task == xprt->snd_task)
272 			goto out_locked;
273 		goto out_sleep;
274 	}
275 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
276 		goto out_unlock;
277 	xprt->snd_task = task;
278 
279 out_locked:
280 	trace_xprt_reserve_xprt(xprt, task);
281 	return 1;
282 
283 out_unlock:
284 	xprt_clear_locked(xprt);
285 out_sleep:
286 	task->tk_status = -EAGAIN;
287 	if  (RPC_IS_SOFT(task))
288 		rpc_sleep_on_timeout(&xprt->sending, task, NULL,
289 				xprt_request_timeout(req));
290 	else
291 		rpc_sleep_on(&xprt->sending, task, NULL);
292 	return 0;
293 }
294 EXPORT_SYMBOL_GPL(xprt_reserve_xprt);
295 
296 static bool
297 xprt_need_congestion_window_wait(struct rpc_xprt *xprt)
298 {
299 	return test_bit(XPRT_CWND_WAIT, &xprt->state);
300 }
301 
302 static void
303 xprt_set_congestion_window_wait(struct rpc_xprt *xprt)
304 {
305 	if (!list_empty(&xprt->xmit_queue)) {
306 		/* Peek at head of queue to see if it can make progress */
307 		if (list_first_entry(&xprt->xmit_queue, struct rpc_rqst,
308 					rq_xmit)->rq_cong)
309 			return;
310 	}
311 	set_bit(XPRT_CWND_WAIT, &xprt->state);
312 }
313 
314 static void
315 xprt_test_and_clear_congestion_window_wait(struct rpc_xprt *xprt)
316 {
317 	if (!RPCXPRT_CONGESTED(xprt))
318 		clear_bit(XPRT_CWND_WAIT, &xprt->state);
319 }
320 
321 /*
322  * xprt_reserve_xprt_cong - serialize write access to transports
323  * @task: task that is requesting access to the transport
324  *
325  * Same as xprt_reserve_xprt, but Van Jacobson congestion control is
326  * integrated into the decision of whether a request is allowed to be
327  * woken up and given access to the transport.
328  * Note that the lock is only granted if we know there are free slots.
329  */
330 int xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
331 {
332 	struct rpc_rqst *req = task->tk_rqstp;
333 
334 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
335 		if (task == xprt->snd_task)
336 			goto out_locked;
337 		goto out_sleep;
338 	}
339 	if (req == NULL) {
340 		xprt->snd_task = task;
341 		goto out_locked;
342 	}
343 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
344 		goto out_unlock;
345 	if (!xprt_need_congestion_window_wait(xprt)) {
346 		xprt->snd_task = task;
347 		goto out_locked;
348 	}
349 out_unlock:
350 	xprt_clear_locked(xprt);
351 out_sleep:
352 	task->tk_status = -EAGAIN;
353 	if (RPC_IS_SOFT(task))
354 		rpc_sleep_on_timeout(&xprt->sending, task, NULL,
355 				xprt_request_timeout(req));
356 	else
357 		rpc_sleep_on(&xprt->sending, task, NULL);
358 	return 0;
359 out_locked:
360 	trace_xprt_reserve_cong(xprt, task);
361 	return 1;
362 }
363 EXPORT_SYMBOL_GPL(xprt_reserve_xprt_cong);
364 
365 static inline int xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
366 {
367 	int retval;
368 
369 	if (test_bit(XPRT_LOCKED, &xprt->state) && xprt->snd_task == task)
370 		return 1;
371 	spin_lock(&xprt->transport_lock);
372 	retval = xprt->ops->reserve_xprt(xprt, task);
373 	spin_unlock(&xprt->transport_lock);
374 	return retval;
375 }
376 
377 static bool __xprt_lock_write_func(struct rpc_task *task, void *data)
378 {
379 	struct rpc_xprt *xprt = data;
380 
381 	xprt->snd_task = task;
382 	return true;
383 }
384 
385 static void __xprt_lock_write_next(struct rpc_xprt *xprt)
386 {
387 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
388 		return;
389 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
390 		goto out_unlock;
391 	if (rpc_wake_up_first_on_wq(xprtiod_workqueue, &xprt->sending,
392 				__xprt_lock_write_func, xprt))
393 		return;
394 out_unlock:
395 	xprt_clear_locked(xprt);
396 }
397 
398 static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt)
399 {
400 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
401 		return;
402 	if (test_bit(XPRT_WRITE_SPACE, &xprt->state))
403 		goto out_unlock;
404 	if (xprt_need_congestion_window_wait(xprt))
405 		goto out_unlock;
406 	if (rpc_wake_up_first_on_wq(xprtiod_workqueue, &xprt->sending,
407 				__xprt_lock_write_func, xprt))
408 		return;
409 out_unlock:
410 	xprt_clear_locked(xprt);
411 }
412 
413 /**
414  * xprt_release_xprt - allow other requests to use a transport
415  * @xprt: transport with other tasks potentially waiting
416  * @task: task that is releasing access to the transport
417  *
418  * Note that "task" can be NULL.  No congestion control is provided.
419  */
420 void xprt_release_xprt(struct rpc_xprt *xprt, struct rpc_task *task)
421 {
422 	if (xprt->snd_task == task) {
423 		xprt_clear_locked(xprt);
424 		__xprt_lock_write_next(xprt);
425 	}
426 	trace_xprt_release_xprt(xprt, task);
427 }
428 EXPORT_SYMBOL_GPL(xprt_release_xprt);
429 
430 /**
431  * xprt_release_xprt_cong - allow other requests to use a transport
432  * @xprt: transport with other tasks potentially waiting
433  * @task: task that is releasing access to the transport
434  *
435  * Note that "task" can be NULL.  Another task is awoken to use the
436  * transport if the transport's congestion window allows it.
437  */
438 void xprt_release_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task)
439 {
440 	if (xprt->snd_task == task) {
441 		xprt_clear_locked(xprt);
442 		__xprt_lock_write_next_cong(xprt);
443 	}
444 	trace_xprt_release_cong(xprt, task);
445 }
446 EXPORT_SYMBOL_GPL(xprt_release_xprt_cong);
447 
448 void xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
449 {
450 	if (xprt->snd_task != task)
451 		return;
452 	spin_lock(&xprt->transport_lock);
453 	xprt->ops->release_xprt(xprt, task);
454 	spin_unlock(&xprt->transport_lock);
455 }
456 
457 /*
458  * Van Jacobson congestion avoidance. Check if the congestion window
459  * overflowed. Put the task to sleep if this is the case.
460  */
461 static int
462 __xprt_get_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
463 {
464 	if (req->rq_cong)
465 		return 1;
466 	trace_xprt_get_cong(xprt, req->rq_task);
467 	if (RPCXPRT_CONGESTED(xprt)) {
468 		xprt_set_congestion_window_wait(xprt);
469 		return 0;
470 	}
471 	req->rq_cong = 1;
472 	xprt->cong += RPC_CWNDSCALE;
473 	return 1;
474 }
475 
476 /*
477  * Adjust the congestion window, and wake up the next task
478  * that has been sleeping due to congestion
479  */
480 static void
481 __xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
482 {
483 	if (!req->rq_cong)
484 		return;
485 	req->rq_cong = 0;
486 	xprt->cong -= RPC_CWNDSCALE;
487 	xprt_test_and_clear_congestion_window_wait(xprt);
488 	trace_xprt_put_cong(xprt, req->rq_task);
489 	__xprt_lock_write_next_cong(xprt);
490 }
491 
492 /**
493  * xprt_request_get_cong - Request congestion control credits
494  * @xprt: pointer to transport
495  * @req: pointer to RPC request
496  *
497  * Useful for transports that require congestion control.
498  */
499 bool
500 xprt_request_get_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
501 {
502 	bool ret = false;
503 
504 	if (req->rq_cong)
505 		return true;
506 	spin_lock(&xprt->transport_lock);
507 	ret = __xprt_get_cong(xprt, req) != 0;
508 	spin_unlock(&xprt->transport_lock);
509 	return ret;
510 }
511 EXPORT_SYMBOL_GPL(xprt_request_get_cong);
512 
513 /**
514  * xprt_release_rqst_cong - housekeeping when request is complete
515  * @task: RPC request that recently completed
516  *
517  * Useful for transports that require congestion control.
518  */
519 void xprt_release_rqst_cong(struct rpc_task *task)
520 {
521 	struct rpc_rqst *req = task->tk_rqstp;
522 
523 	__xprt_put_cong(req->rq_xprt, req);
524 }
525 EXPORT_SYMBOL_GPL(xprt_release_rqst_cong);
526 
527 static void xprt_clear_congestion_window_wait_locked(struct rpc_xprt *xprt)
528 {
529 	if (test_and_clear_bit(XPRT_CWND_WAIT, &xprt->state))
530 		__xprt_lock_write_next_cong(xprt);
531 }
532 
533 /*
534  * Clear the congestion window wait flag and wake up the next
535  * entry on xprt->sending
536  */
537 static void
538 xprt_clear_congestion_window_wait(struct rpc_xprt *xprt)
539 {
540 	if (test_and_clear_bit(XPRT_CWND_WAIT, &xprt->state)) {
541 		spin_lock(&xprt->transport_lock);
542 		__xprt_lock_write_next_cong(xprt);
543 		spin_unlock(&xprt->transport_lock);
544 	}
545 }
546 
547 /**
548  * xprt_adjust_cwnd - adjust transport congestion window
549  * @xprt: pointer to xprt
550  * @task: recently completed RPC request used to adjust window
551  * @result: result code of completed RPC request
552  *
553  * The transport code maintains an estimate on the maximum number of out-
554  * standing RPC requests, using a smoothed version of the congestion
555  * avoidance implemented in 44BSD. This is basically the Van Jacobson
556  * congestion algorithm: If a retransmit occurs, the congestion window is
557  * halved; otherwise, it is incremented by 1/cwnd when
558  *
559  *	-	a reply is received and
560  *	-	a full number of requests are outstanding and
561  *	-	the congestion window hasn't been updated recently.
562  */
563 void xprt_adjust_cwnd(struct rpc_xprt *xprt, struct rpc_task *task, int result)
564 {
565 	struct rpc_rqst *req = task->tk_rqstp;
566 	unsigned long cwnd = xprt->cwnd;
567 
568 	if (result >= 0 && cwnd <= xprt->cong) {
569 		/* The (cwnd >> 1) term makes sure
570 		 * the result gets rounded properly. */
571 		cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
572 		if (cwnd > RPC_MAXCWND(xprt))
573 			cwnd = RPC_MAXCWND(xprt);
574 		__xprt_lock_write_next_cong(xprt);
575 	} else if (result == -ETIMEDOUT) {
576 		cwnd >>= 1;
577 		if (cwnd < RPC_CWNDSCALE)
578 			cwnd = RPC_CWNDSCALE;
579 	}
580 	dprintk("RPC:       cong %ld, cwnd was %ld, now %ld\n",
581 			xprt->cong, xprt->cwnd, cwnd);
582 	xprt->cwnd = cwnd;
583 	__xprt_put_cong(xprt, req);
584 }
585 EXPORT_SYMBOL_GPL(xprt_adjust_cwnd);
586 
587 /**
588  * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue
589  * @xprt: transport with waiting tasks
590  * @status: result code to plant in each task before waking it
591  *
592  */
593 void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status)
594 {
595 	if (status < 0)
596 		rpc_wake_up_status(&xprt->pending, status);
597 	else
598 		rpc_wake_up(&xprt->pending);
599 }
600 EXPORT_SYMBOL_GPL(xprt_wake_pending_tasks);
601 
602 /**
603  * xprt_wait_for_buffer_space - wait for transport output buffer to clear
604  * @xprt: transport
605  *
606  * Note that we only set the timer for the case of RPC_IS_SOFT(), since
607  * we don't in general want to force a socket disconnection due to
608  * an incomplete RPC call transmission.
609  */
610 void xprt_wait_for_buffer_space(struct rpc_xprt *xprt)
611 {
612 	set_bit(XPRT_WRITE_SPACE, &xprt->state);
613 }
614 EXPORT_SYMBOL_GPL(xprt_wait_for_buffer_space);
615 
616 static bool
617 xprt_clear_write_space_locked(struct rpc_xprt *xprt)
618 {
619 	if (test_and_clear_bit(XPRT_WRITE_SPACE, &xprt->state)) {
620 		__xprt_lock_write_next(xprt);
621 		dprintk("RPC:       write space: waking waiting task on "
622 				"xprt %p\n", xprt);
623 		return true;
624 	}
625 	return false;
626 }
627 
628 /**
629  * xprt_write_space - wake the task waiting for transport output buffer space
630  * @xprt: transport with waiting tasks
631  *
632  * Can be called in a soft IRQ context, so xprt_write_space never sleeps.
633  */
634 bool xprt_write_space(struct rpc_xprt *xprt)
635 {
636 	bool ret;
637 
638 	if (!test_bit(XPRT_WRITE_SPACE, &xprt->state))
639 		return false;
640 	spin_lock(&xprt->transport_lock);
641 	ret = xprt_clear_write_space_locked(xprt);
642 	spin_unlock(&xprt->transport_lock);
643 	return ret;
644 }
645 EXPORT_SYMBOL_GPL(xprt_write_space);
646 
647 static unsigned long xprt_abs_ktime_to_jiffies(ktime_t abstime)
648 {
649 	s64 delta = ktime_to_ns(ktime_get() - abstime);
650 	return likely(delta >= 0) ?
651 		jiffies - nsecs_to_jiffies(delta) :
652 		jiffies + nsecs_to_jiffies(-delta);
653 }
654 
655 static unsigned long xprt_calc_majortimeo(struct rpc_rqst *req)
656 {
657 	const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
658 	unsigned long majortimeo = req->rq_timeout;
659 
660 	if (to->to_exponential)
661 		majortimeo <<= to->to_retries;
662 	else
663 		majortimeo += to->to_increment * to->to_retries;
664 	if (majortimeo > to->to_maxval || majortimeo == 0)
665 		majortimeo = to->to_maxval;
666 	return majortimeo;
667 }
668 
669 static void xprt_reset_majortimeo(struct rpc_rqst *req)
670 {
671 	req->rq_majortimeo += xprt_calc_majortimeo(req);
672 }
673 
674 static void xprt_reset_minortimeo(struct rpc_rqst *req)
675 {
676 	req->rq_minortimeo += req->rq_timeout;
677 }
678 
679 static void xprt_init_majortimeo(struct rpc_task *task, struct rpc_rqst *req)
680 {
681 	unsigned long time_init;
682 	struct rpc_xprt *xprt = req->rq_xprt;
683 
684 	if (likely(xprt && xprt_connected(xprt)))
685 		time_init = jiffies;
686 	else
687 		time_init = xprt_abs_ktime_to_jiffies(task->tk_start);
688 	req->rq_timeout = task->tk_client->cl_timeout->to_initval;
689 	req->rq_majortimeo = time_init + xprt_calc_majortimeo(req);
690 	req->rq_minortimeo = time_init + req->rq_timeout;
691 }
692 
693 /**
694  * xprt_adjust_timeout - adjust timeout values for next retransmit
695  * @req: RPC request containing parameters to use for the adjustment
696  *
697  */
698 int xprt_adjust_timeout(struct rpc_rqst *req)
699 {
700 	struct rpc_xprt *xprt = req->rq_xprt;
701 	const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout;
702 	int status = 0;
703 
704 	if (time_before(jiffies, req->rq_majortimeo)) {
705 		if (time_before(jiffies, req->rq_minortimeo))
706 			return status;
707 		if (to->to_exponential)
708 			req->rq_timeout <<= 1;
709 		else
710 			req->rq_timeout += to->to_increment;
711 		if (to->to_maxval && req->rq_timeout >= to->to_maxval)
712 			req->rq_timeout = to->to_maxval;
713 		req->rq_retries++;
714 	} else {
715 		req->rq_timeout = to->to_initval;
716 		req->rq_retries = 0;
717 		xprt_reset_majortimeo(req);
718 		/* Reset the RTT counters == "slow start" */
719 		spin_lock(&xprt->transport_lock);
720 		rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
721 		spin_unlock(&xprt->transport_lock);
722 		status = -ETIMEDOUT;
723 	}
724 	xprt_reset_minortimeo(req);
725 
726 	if (req->rq_timeout == 0) {
727 		printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
728 		req->rq_timeout = 5 * HZ;
729 	}
730 	return status;
731 }
732 
733 static void xprt_autoclose(struct work_struct *work)
734 {
735 	struct rpc_xprt *xprt =
736 		container_of(work, struct rpc_xprt, task_cleanup);
737 	unsigned int pflags = memalloc_nofs_save();
738 
739 	trace_xprt_disconnect_auto(xprt);
740 	clear_bit(XPRT_CLOSE_WAIT, &xprt->state);
741 	xprt->ops->close(xprt);
742 	xprt_release_write(xprt, NULL);
743 	wake_up_bit(&xprt->state, XPRT_LOCKED);
744 	memalloc_nofs_restore(pflags);
745 }
746 
747 /**
748  * xprt_disconnect_done - mark a transport as disconnected
749  * @xprt: transport to flag for disconnect
750  *
751  */
752 void xprt_disconnect_done(struct rpc_xprt *xprt)
753 {
754 	trace_xprt_disconnect_done(xprt);
755 	spin_lock(&xprt->transport_lock);
756 	xprt_clear_connected(xprt);
757 	xprt_clear_write_space_locked(xprt);
758 	xprt_clear_congestion_window_wait_locked(xprt);
759 	xprt_wake_pending_tasks(xprt, -ENOTCONN);
760 	spin_unlock(&xprt->transport_lock);
761 }
762 EXPORT_SYMBOL_GPL(xprt_disconnect_done);
763 
764 /**
765  * xprt_schedule_autoclose_locked - Try to schedule an autoclose RPC call
766  * @xprt: transport to disconnect
767  */
768 static void xprt_schedule_autoclose_locked(struct rpc_xprt *xprt)
769 {
770 	set_bit(XPRT_CLOSE_WAIT, &xprt->state);
771 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
772 		queue_work(xprtiod_workqueue, &xprt->task_cleanup);
773 	else if (xprt->snd_task && !test_bit(XPRT_SND_IS_COOKIE, &xprt->state))
774 		rpc_wake_up_queued_task_set_status(&xprt->pending,
775 						   xprt->snd_task, -ENOTCONN);
776 }
777 
778 /**
779  * xprt_force_disconnect - force a transport to disconnect
780  * @xprt: transport to disconnect
781  *
782  */
783 void xprt_force_disconnect(struct rpc_xprt *xprt)
784 {
785 	trace_xprt_disconnect_force(xprt);
786 
787 	/* Don't race with the test_bit() in xprt_clear_locked() */
788 	spin_lock(&xprt->transport_lock);
789 	xprt_schedule_autoclose_locked(xprt);
790 	spin_unlock(&xprt->transport_lock);
791 }
792 EXPORT_SYMBOL_GPL(xprt_force_disconnect);
793 
794 static unsigned int
795 xprt_connect_cookie(struct rpc_xprt *xprt)
796 {
797 	return READ_ONCE(xprt->connect_cookie);
798 }
799 
800 static bool
801 xprt_request_retransmit_after_disconnect(struct rpc_task *task)
802 {
803 	struct rpc_rqst *req = task->tk_rqstp;
804 	struct rpc_xprt *xprt = req->rq_xprt;
805 
806 	return req->rq_connect_cookie != xprt_connect_cookie(xprt) ||
807 		!xprt_connected(xprt);
808 }
809 
810 /**
811  * xprt_conditional_disconnect - force a transport to disconnect
812  * @xprt: transport to disconnect
813  * @cookie: 'connection cookie'
814  *
815  * This attempts to break the connection if and only if 'cookie' matches
816  * the current transport 'connection cookie'. It ensures that we don't
817  * try to break the connection more than once when we need to retransmit
818  * a batch of RPC requests.
819  *
820  */
821 void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie)
822 {
823 	/* Don't race with the test_bit() in xprt_clear_locked() */
824 	spin_lock(&xprt->transport_lock);
825 	if (cookie != xprt->connect_cookie)
826 		goto out;
827 	if (test_bit(XPRT_CLOSING, &xprt->state))
828 		goto out;
829 	xprt_schedule_autoclose_locked(xprt);
830 out:
831 	spin_unlock(&xprt->transport_lock);
832 }
833 
834 static bool
835 xprt_has_timer(const struct rpc_xprt *xprt)
836 {
837 	return xprt->idle_timeout != 0;
838 }
839 
840 static void
841 xprt_schedule_autodisconnect(struct rpc_xprt *xprt)
842 	__must_hold(&xprt->transport_lock)
843 {
844 	xprt->last_used = jiffies;
845 	if (RB_EMPTY_ROOT(&xprt->recv_queue) && xprt_has_timer(xprt))
846 		mod_timer(&xprt->timer, xprt->last_used + xprt->idle_timeout);
847 }
848 
849 static void
850 xprt_init_autodisconnect(struct timer_list *t)
851 {
852 	struct rpc_xprt *xprt = from_timer(xprt, t, timer);
853 
854 	if (!RB_EMPTY_ROOT(&xprt->recv_queue))
855 		return;
856 	/* Reset xprt->last_used to avoid connect/autodisconnect cycling */
857 	xprt->last_used = jiffies;
858 	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
859 		return;
860 	queue_work(xprtiod_workqueue, &xprt->task_cleanup);
861 }
862 
863 #if IS_ENABLED(CONFIG_FAIL_SUNRPC)
864 static void xprt_inject_disconnect(struct rpc_xprt *xprt)
865 {
866 	if (!fail_sunrpc.ignore_client_disconnect &&
867 	    should_fail(&fail_sunrpc.attr, 1))
868 		xprt->ops->inject_disconnect(xprt);
869 }
870 #else
871 static inline void xprt_inject_disconnect(struct rpc_xprt *xprt)
872 {
873 }
874 #endif
875 
876 bool xprt_lock_connect(struct rpc_xprt *xprt,
877 		struct rpc_task *task,
878 		void *cookie)
879 {
880 	bool ret = false;
881 
882 	spin_lock(&xprt->transport_lock);
883 	if (!test_bit(XPRT_LOCKED, &xprt->state))
884 		goto out;
885 	if (xprt->snd_task != task)
886 		goto out;
887 	set_bit(XPRT_SND_IS_COOKIE, &xprt->state);
888 	xprt->snd_task = cookie;
889 	ret = true;
890 out:
891 	spin_unlock(&xprt->transport_lock);
892 	return ret;
893 }
894 EXPORT_SYMBOL_GPL(xprt_lock_connect);
895 
896 void xprt_unlock_connect(struct rpc_xprt *xprt, void *cookie)
897 {
898 	spin_lock(&xprt->transport_lock);
899 	if (xprt->snd_task != cookie)
900 		goto out;
901 	if (!test_bit(XPRT_LOCKED, &xprt->state))
902 		goto out;
903 	xprt->snd_task =NULL;
904 	clear_bit(XPRT_SND_IS_COOKIE, &xprt->state);
905 	xprt->ops->release_xprt(xprt, NULL);
906 	xprt_schedule_autodisconnect(xprt);
907 out:
908 	spin_unlock(&xprt->transport_lock);
909 	wake_up_bit(&xprt->state, XPRT_LOCKED);
910 }
911 EXPORT_SYMBOL_GPL(xprt_unlock_connect);
912 
913 /**
914  * xprt_connect - schedule a transport connect operation
915  * @task: RPC task that is requesting the connect
916  *
917  */
918 void xprt_connect(struct rpc_task *task)
919 {
920 	struct rpc_xprt	*xprt = task->tk_rqstp->rq_xprt;
921 
922 	trace_xprt_connect(xprt);
923 
924 	if (!xprt_bound(xprt)) {
925 		task->tk_status = -EAGAIN;
926 		return;
927 	}
928 	if (!xprt_lock_write(xprt, task))
929 		return;
930 
931 	if (test_and_clear_bit(XPRT_CLOSE_WAIT, &xprt->state)) {
932 		trace_xprt_disconnect_cleanup(xprt);
933 		xprt->ops->close(xprt);
934 	}
935 
936 	if (!xprt_connected(xprt)) {
937 		task->tk_rqstp->rq_connect_cookie = xprt->connect_cookie;
938 		rpc_sleep_on_timeout(&xprt->pending, task, NULL,
939 				xprt_request_timeout(task->tk_rqstp));
940 
941 		if (test_bit(XPRT_CLOSING, &xprt->state))
942 			return;
943 		if (xprt_test_and_set_connecting(xprt))
944 			return;
945 		/* Race breaker */
946 		if (!xprt_connected(xprt)) {
947 			xprt->stat.connect_start = jiffies;
948 			xprt->ops->connect(xprt, task);
949 		} else {
950 			xprt_clear_connecting(xprt);
951 			task->tk_status = 0;
952 			rpc_wake_up_queued_task(&xprt->pending, task);
953 		}
954 	}
955 	xprt_release_write(xprt, task);
956 }
957 
958 /**
959  * xprt_reconnect_delay - compute the wait before scheduling a connect
960  * @xprt: transport instance
961  *
962  */
963 unsigned long xprt_reconnect_delay(const struct rpc_xprt *xprt)
964 {
965 	unsigned long start, now = jiffies;
966 
967 	start = xprt->stat.connect_start + xprt->reestablish_timeout;
968 	if (time_after(start, now))
969 		return start - now;
970 	return 0;
971 }
972 EXPORT_SYMBOL_GPL(xprt_reconnect_delay);
973 
974 /**
975  * xprt_reconnect_backoff - compute the new re-establish timeout
976  * @xprt: transport instance
977  * @init_to: initial reestablish timeout
978  *
979  */
980 void xprt_reconnect_backoff(struct rpc_xprt *xprt, unsigned long init_to)
981 {
982 	xprt->reestablish_timeout <<= 1;
983 	if (xprt->reestablish_timeout > xprt->max_reconnect_timeout)
984 		xprt->reestablish_timeout = xprt->max_reconnect_timeout;
985 	if (xprt->reestablish_timeout < init_to)
986 		xprt->reestablish_timeout = init_to;
987 }
988 EXPORT_SYMBOL_GPL(xprt_reconnect_backoff);
989 
990 enum xprt_xid_rb_cmp {
991 	XID_RB_EQUAL,
992 	XID_RB_LEFT,
993 	XID_RB_RIGHT,
994 };
995 static enum xprt_xid_rb_cmp
996 xprt_xid_cmp(__be32 xid1, __be32 xid2)
997 {
998 	if (xid1 == xid2)
999 		return XID_RB_EQUAL;
1000 	if ((__force u32)xid1 < (__force u32)xid2)
1001 		return XID_RB_LEFT;
1002 	return XID_RB_RIGHT;
1003 }
1004 
1005 static struct rpc_rqst *
1006 xprt_request_rb_find(struct rpc_xprt *xprt, __be32 xid)
1007 {
1008 	struct rb_node *n = xprt->recv_queue.rb_node;
1009 	struct rpc_rqst *req;
1010 
1011 	while (n != NULL) {
1012 		req = rb_entry(n, struct rpc_rqst, rq_recv);
1013 		switch (xprt_xid_cmp(xid, req->rq_xid)) {
1014 		case XID_RB_LEFT:
1015 			n = n->rb_left;
1016 			break;
1017 		case XID_RB_RIGHT:
1018 			n = n->rb_right;
1019 			break;
1020 		case XID_RB_EQUAL:
1021 			return req;
1022 		}
1023 	}
1024 	return NULL;
1025 }
1026 
1027 static void
1028 xprt_request_rb_insert(struct rpc_xprt *xprt, struct rpc_rqst *new)
1029 {
1030 	struct rb_node **p = &xprt->recv_queue.rb_node;
1031 	struct rb_node *n = NULL;
1032 	struct rpc_rqst *req;
1033 
1034 	while (*p != NULL) {
1035 		n = *p;
1036 		req = rb_entry(n, struct rpc_rqst, rq_recv);
1037 		switch(xprt_xid_cmp(new->rq_xid, req->rq_xid)) {
1038 		case XID_RB_LEFT:
1039 			p = &n->rb_left;
1040 			break;
1041 		case XID_RB_RIGHT:
1042 			p = &n->rb_right;
1043 			break;
1044 		case XID_RB_EQUAL:
1045 			WARN_ON_ONCE(new != req);
1046 			return;
1047 		}
1048 	}
1049 	rb_link_node(&new->rq_recv, n, p);
1050 	rb_insert_color(&new->rq_recv, &xprt->recv_queue);
1051 }
1052 
1053 static void
1054 xprt_request_rb_remove(struct rpc_xprt *xprt, struct rpc_rqst *req)
1055 {
1056 	rb_erase(&req->rq_recv, &xprt->recv_queue);
1057 }
1058 
1059 /**
1060  * xprt_lookup_rqst - find an RPC request corresponding to an XID
1061  * @xprt: transport on which the original request was transmitted
1062  * @xid: RPC XID of incoming reply
1063  *
1064  * Caller holds xprt->queue_lock.
1065  */
1066 struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, __be32 xid)
1067 {
1068 	struct rpc_rqst *entry;
1069 
1070 	entry = xprt_request_rb_find(xprt, xid);
1071 	if (entry != NULL) {
1072 		trace_xprt_lookup_rqst(xprt, xid, 0);
1073 		entry->rq_rtt = ktime_sub(ktime_get(), entry->rq_xtime);
1074 		return entry;
1075 	}
1076 
1077 	dprintk("RPC:       xprt_lookup_rqst did not find xid %08x\n",
1078 			ntohl(xid));
1079 	trace_xprt_lookup_rqst(xprt, xid, -ENOENT);
1080 	xprt->stat.bad_xids++;
1081 	return NULL;
1082 }
1083 EXPORT_SYMBOL_GPL(xprt_lookup_rqst);
1084 
1085 static bool
1086 xprt_is_pinned_rqst(struct rpc_rqst *req)
1087 {
1088 	return atomic_read(&req->rq_pin) != 0;
1089 }
1090 
1091 /**
1092  * xprt_pin_rqst - Pin a request on the transport receive list
1093  * @req: Request to pin
1094  *
1095  * Caller must ensure this is atomic with the call to xprt_lookup_rqst()
1096  * so should be holding xprt->queue_lock.
1097  */
1098 void xprt_pin_rqst(struct rpc_rqst *req)
1099 {
1100 	atomic_inc(&req->rq_pin);
1101 }
1102 EXPORT_SYMBOL_GPL(xprt_pin_rqst);
1103 
1104 /**
1105  * xprt_unpin_rqst - Unpin a request on the transport receive list
1106  * @req: Request to pin
1107  *
1108  * Caller should be holding xprt->queue_lock.
1109  */
1110 void xprt_unpin_rqst(struct rpc_rqst *req)
1111 {
1112 	if (!test_bit(RPC_TASK_MSG_PIN_WAIT, &req->rq_task->tk_runstate)) {
1113 		atomic_dec(&req->rq_pin);
1114 		return;
1115 	}
1116 	if (atomic_dec_and_test(&req->rq_pin))
1117 		wake_up_var(&req->rq_pin);
1118 }
1119 EXPORT_SYMBOL_GPL(xprt_unpin_rqst);
1120 
1121 static void xprt_wait_on_pinned_rqst(struct rpc_rqst *req)
1122 {
1123 	wait_var_event(&req->rq_pin, !xprt_is_pinned_rqst(req));
1124 }
1125 
1126 static bool
1127 xprt_request_data_received(struct rpc_task *task)
1128 {
1129 	return !test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) &&
1130 		READ_ONCE(task->tk_rqstp->rq_reply_bytes_recvd) != 0;
1131 }
1132 
1133 static bool
1134 xprt_request_need_enqueue_receive(struct rpc_task *task, struct rpc_rqst *req)
1135 {
1136 	return !test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) &&
1137 		READ_ONCE(task->tk_rqstp->rq_reply_bytes_recvd) == 0;
1138 }
1139 
1140 /**
1141  * xprt_request_enqueue_receive - Add an request to the receive queue
1142  * @task: RPC task
1143  *
1144  */
1145 void
1146 xprt_request_enqueue_receive(struct rpc_task *task)
1147 {
1148 	struct rpc_rqst *req = task->tk_rqstp;
1149 	struct rpc_xprt *xprt = req->rq_xprt;
1150 
1151 	if (!xprt_request_need_enqueue_receive(task, req))
1152 		return;
1153 
1154 	xprt_request_prepare(task->tk_rqstp);
1155 	spin_lock(&xprt->queue_lock);
1156 
1157 	/* Update the softirq receive buffer */
1158 	memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
1159 			sizeof(req->rq_private_buf));
1160 
1161 	/* Add request to the receive list */
1162 	xprt_request_rb_insert(xprt, req);
1163 	set_bit(RPC_TASK_NEED_RECV, &task->tk_runstate);
1164 	spin_unlock(&xprt->queue_lock);
1165 
1166 	/* Turn off autodisconnect */
1167 	del_singleshot_timer_sync(&xprt->timer);
1168 }
1169 
1170 /**
1171  * xprt_request_dequeue_receive_locked - Remove a request from the receive queue
1172  * @task: RPC task
1173  *
1174  * Caller must hold xprt->queue_lock.
1175  */
1176 static void
1177 xprt_request_dequeue_receive_locked(struct rpc_task *task)
1178 {
1179 	struct rpc_rqst *req = task->tk_rqstp;
1180 
1181 	if (test_and_clear_bit(RPC_TASK_NEED_RECV, &task->tk_runstate))
1182 		xprt_request_rb_remove(req->rq_xprt, req);
1183 }
1184 
1185 /**
1186  * xprt_update_rtt - Update RPC RTT statistics
1187  * @task: RPC request that recently completed
1188  *
1189  * Caller holds xprt->queue_lock.
1190  */
1191 void xprt_update_rtt(struct rpc_task *task)
1192 {
1193 	struct rpc_rqst *req = task->tk_rqstp;
1194 	struct rpc_rtt *rtt = task->tk_client->cl_rtt;
1195 	unsigned int timer = task->tk_msg.rpc_proc->p_timer;
1196 	long m = usecs_to_jiffies(ktime_to_us(req->rq_rtt));
1197 
1198 	if (timer) {
1199 		if (req->rq_ntrans == 1)
1200 			rpc_update_rtt(rtt, timer, m);
1201 		rpc_set_timeo(rtt, timer, req->rq_ntrans - 1);
1202 	}
1203 }
1204 EXPORT_SYMBOL_GPL(xprt_update_rtt);
1205 
1206 /**
1207  * xprt_complete_rqst - called when reply processing is complete
1208  * @task: RPC request that recently completed
1209  * @copied: actual number of bytes received from the transport
1210  *
1211  * Caller holds xprt->queue_lock.
1212  */
1213 void xprt_complete_rqst(struct rpc_task *task, int copied)
1214 {
1215 	struct rpc_rqst *req = task->tk_rqstp;
1216 	struct rpc_xprt *xprt = req->rq_xprt;
1217 
1218 	xprt->stat.recvs++;
1219 
1220 	req->rq_private_buf.len = copied;
1221 	/* Ensure all writes are done before we update */
1222 	/* req->rq_reply_bytes_recvd */
1223 	smp_wmb();
1224 	req->rq_reply_bytes_recvd = copied;
1225 	xprt_request_dequeue_receive_locked(task);
1226 	rpc_wake_up_queued_task(&xprt->pending, task);
1227 }
1228 EXPORT_SYMBOL_GPL(xprt_complete_rqst);
1229 
1230 static void xprt_timer(struct rpc_task *task)
1231 {
1232 	struct rpc_rqst *req = task->tk_rqstp;
1233 	struct rpc_xprt *xprt = req->rq_xprt;
1234 
1235 	if (task->tk_status != -ETIMEDOUT)
1236 		return;
1237 
1238 	trace_xprt_timer(xprt, req->rq_xid, task->tk_status);
1239 	if (!req->rq_reply_bytes_recvd) {
1240 		if (xprt->ops->timer)
1241 			xprt->ops->timer(xprt, task);
1242 	} else
1243 		task->tk_status = 0;
1244 }
1245 
1246 /**
1247  * xprt_wait_for_reply_request_def - wait for reply
1248  * @task: pointer to rpc_task
1249  *
1250  * Set a request's retransmit timeout based on the transport's
1251  * default timeout parameters.  Used by transports that don't adjust
1252  * the retransmit timeout based on round-trip time estimation,
1253  * and put the task to sleep on the pending queue.
1254  */
1255 void xprt_wait_for_reply_request_def(struct rpc_task *task)
1256 {
1257 	struct rpc_rqst *req = task->tk_rqstp;
1258 
1259 	rpc_sleep_on_timeout(&req->rq_xprt->pending, task, xprt_timer,
1260 			xprt_request_timeout(req));
1261 }
1262 EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_def);
1263 
1264 /**
1265  * xprt_wait_for_reply_request_rtt - wait for reply using RTT estimator
1266  * @task: pointer to rpc_task
1267  *
1268  * Set a request's retransmit timeout using the RTT estimator,
1269  * and put the task to sleep on the pending queue.
1270  */
1271 void xprt_wait_for_reply_request_rtt(struct rpc_task *task)
1272 {
1273 	int timer = task->tk_msg.rpc_proc->p_timer;
1274 	struct rpc_clnt *clnt = task->tk_client;
1275 	struct rpc_rtt *rtt = clnt->cl_rtt;
1276 	struct rpc_rqst *req = task->tk_rqstp;
1277 	unsigned long max_timeout = clnt->cl_timeout->to_maxval;
1278 	unsigned long timeout;
1279 
1280 	timeout = rpc_calc_rto(rtt, timer);
1281 	timeout <<= rpc_ntimeo(rtt, timer) + req->rq_retries;
1282 	if (timeout > max_timeout || timeout == 0)
1283 		timeout = max_timeout;
1284 	rpc_sleep_on_timeout(&req->rq_xprt->pending, task, xprt_timer,
1285 			jiffies + timeout);
1286 }
1287 EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_rtt);
1288 
1289 /**
1290  * xprt_request_wait_receive - wait for the reply to an RPC request
1291  * @task: RPC task about to send a request
1292  *
1293  */
1294 void xprt_request_wait_receive(struct rpc_task *task)
1295 {
1296 	struct rpc_rqst *req = task->tk_rqstp;
1297 	struct rpc_xprt *xprt = req->rq_xprt;
1298 
1299 	if (!test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate))
1300 		return;
1301 	/*
1302 	 * Sleep on the pending queue if we're expecting a reply.
1303 	 * The spinlock ensures atomicity between the test of
1304 	 * req->rq_reply_bytes_recvd, and the call to rpc_sleep_on().
1305 	 */
1306 	spin_lock(&xprt->queue_lock);
1307 	if (test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate)) {
1308 		xprt->ops->wait_for_reply_request(task);
1309 		/*
1310 		 * Send an extra queue wakeup call if the
1311 		 * connection was dropped in case the call to
1312 		 * rpc_sleep_on() raced.
1313 		 */
1314 		if (xprt_request_retransmit_after_disconnect(task))
1315 			rpc_wake_up_queued_task_set_status(&xprt->pending,
1316 					task, -ENOTCONN);
1317 	}
1318 	spin_unlock(&xprt->queue_lock);
1319 }
1320 
1321 static bool
1322 xprt_request_need_enqueue_transmit(struct rpc_task *task, struct rpc_rqst *req)
1323 {
1324 	return !test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate);
1325 }
1326 
1327 /**
1328  * xprt_request_enqueue_transmit - queue a task for transmission
1329  * @task: pointer to rpc_task
1330  *
1331  * Add a task to the transmission queue.
1332  */
1333 void
1334 xprt_request_enqueue_transmit(struct rpc_task *task)
1335 {
1336 	struct rpc_rqst *pos, *req = task->tk_rqstp;
1337 	struct rpc_xprt *xprt = req->rq_xprt;
1338 
1339 	if (xprt_request_need_enqueue_transmit(task, req)) {
1340 		req->rq_bytes_sent = 0;
1341 		spin_lock(&xprt->queue_lock);
1342 		/*
1343 		 * Requests that carry congestion control credits are added
1344 		 * to the head of the list to avoid starvation issues.
1345 		 */
1346 		if (req->rq_cong) {
1347 			xprt_clear_congestion_window_wait(xprt);
1348 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1349 				if (pos->rq_cong)
1350 					continue;
1351 				/* Note: req is added _before_ pos */
1352 				list_add_tail(&req->rq_xmit, &pos->rq_xmit);
1353 				INIT_LIST_HEAD(&req->rq_xmit2);
1354 				goto out;
1355 			}
1356 		} else if (RPC_IS_SWAPPER(task)) {
1357 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1358 				if (pos->rq_cong || pos->rq_bytes_sent)
1359 					continue;
1360 				if (RPC_IS_SWAPPER(pos->rq_task))
1361 					continue;
1362 				/* Note: req is added _before_ pos */
1363 				list_add_tail(&req->rq_xmit, &pos->rq_xmit);
1364 				INIT_LIST_HEAD(&req->rq_xmit2);
1365 				goto out;
1366 			}
1367 		} else if (!req->rq_seqno) {
1368 			list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) {
1369 				if (pos->rq_task->tk_owner != task->tk_owner)
1370 					continue;
1371 				list_add_tail(&req->rq_xmit2, &pos->rq_xmit2);
1372 				INIT_LIST_HEAD(&req->rq_xmit);
1373 				goto out;
1374 			}
1375 		}
1376 		list_add_tail(&req->rq_xmit, &xprt->xmit_queue);
1377 		INIT_LIST_HEAD(&req->rq_xmit2);
1378 out:
1379 		atomic_long_inc(&xprt->xmit_queuelen);
1380 		set_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate);
1381 		spin_unlock(&xprt->queue_lock);
1382 	}
1383 }
1384 
1385 /**
1386  * xprt_request_dequeue_transmit_locked - remove a task from the transmission queue
1387  * @task: pointer to rpc_task
1388  *
1389  * Remove a task from the transmission queue
1390  * Caller must hold xprt->queue_lock
1391  */
1392 static void
1393 xprt_request_dequeue_transmit_locked(struct rpc_task *task)
1394 {
1395 	struct rpc_rqst *req = task->tk_rqstp;
1396 
1397 	if (!test_and_clear_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1398 		return;
1399 	if (!list_empty(&req->rq_xmit)) {
1400 		list_del(&req->rq_xmit);
1401 		if (!list_empty(&req->rq_xmit2)) {
1402 			struct rpc_rqst *next = list_first_entry(&req->rq_xmit2,
1403 					struct rpc_rqst, rq_xmit2);
1404 			list_del(&req->rq_xmit2);
1405 			list_add_tail(&next->rq_xmit, &next->rq_xprt->xmit_queue);
1406 		}
1407 	} else
1408 		list_del(&req->rq_xmit2);
1409 	atomic_long_dec(&req->rq_xprt->xmit_queuelen);
1410 }
1411 
1412 /**
1413  * xprt_request_dequeue_transmit - remove a task from the transmission queue
1414  * @task: pointer to rpc_task
1415  *
1416  * Remove a task from the transmission queue
1417  */
1418 static void
1419 xprt_request_dequeue_transmit(struct rpc_task *task)
1420 {
1421 	struct rpc_rqst *req = task->tk_rqstp;
1422 	struct rpc_xprt *xprt = req->rq_xprt;
1423 
1424 	spin_lock(&xprt->queue_lock);
1425 	xprt_request_dequeue_transmit_locked(task);
1426 	spin_unlock(&xprt->queue_lock);
1427 }
1428 
1429 /**
1430  * xprt_request_dequeue_xprt - remove a task from the transmit+receive queue
1431  * @task: pointer to rpc_task
1432  *
1433  * Remove a task from the transmit and receive queues, and ensure that
1434  * it is not pinned by the receive work item.
1435  */
1436 void
1437 xprt_request_dequeue_xprt(struct rpc_task *task)
1438 {
1439 	struct rpc_rqst	*req = task->tk_rqstp;
1440 	struct rpc_xprt *xprt = req->rq_xprt;
1441 
1442 	if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate) ||
1443 	    test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) ||
1444 	    xprt_is_pinned_rqst(req)) {
1445 		spin_lock(&xprt->queue_lock);
1446 		xprt_request_dequeue_transmit_locked(task);
1447 		xprt_request_dequeue_receive_locked(task);
1448 		while (xprt_is_pinned_rqst(req)) {
1449 			set_bit(RPC_TASK_MSG_PIN_WAIT, &task->tk_runstate);
1450 			spin_unlock(&xprt->queue_lock);
1451 			xprt_wait_on_pinned_rqst(req);
1452 			spin_lock(&xprt->queue_lock);
1453 			clear_bit(RPC_TASK_MSG_PIN_WAIT, &task->tk_runstate);
1454 		}
1455 		spin_unlock(&xprt->queue_lock);
1456 	}
1457 }
1458 
1459 /**
1460  * xprt_request_prepare - prepare an encoded request for transport
1461  * @req: pointer to rpc_rqst
1462  *
1463  * Calls into the transport layer to do whatever is needed to prepare
1464  * the request for transmission or receive.
1465  */
1466 void
1467 xprt_request_prepare(struct rpc_rqst *req)
1468 {
1469 	struct rpc_xprt *xprt = req->rq_xprt;
1470 
1471 	if (xprt->ops->prepare_request)
1472 		xprt->ops->prepare_request(req);
1473 }
1474 
1475 /**
1476  * xprt_request_need_retransmit - Test if a task needs retransmission
1477  * @task: pointer to rpc_task
1478  *
1479  * Test for whether a connection breakage requires the task to retransmit
1480  */
1481 bool
1482 xprt_request_need_retransmit(struct rpc_task *task)
1483 {
1484 	return xprt_request_retransmit_after_disconnect(task);
1485 }
1486 
1487 /**
1488  * xprt_prepare_transmit - reserve the transport before sending a request
1489  * @task: RPC task about to send a request
1490  *
1491  */
1492 bool xprt_prepare_transmit(struct rpc_task *task)
1493 {
1494 	struct rpc_rqst	*req = task->tk_rqstp;
1495 	struct rpc_xprt	*xprt = req->rq_xprt;
1496 
1497 	if (!xprt_lock_write(xprt, task)) {
1498 		/* Race breaker: someone may have transmitted us */
1499 		if (!test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1500 			rpc_wake_up_queued_task_set_status(&xprt->sending,
1501 					task, 0);
1502 		return false;
1503 
1504 	}
1505 	return true;
1506 }
1507 
1508 void xprt_end_transmit(struct rpc_task *task)
1509 {
1510 	struct rpc_xprt	*xprt = task->tk_rqstp->rq_xprt;
1511 
1512 	xprt_inject_disconnect(xprt);
1513 	xprt_release_write(xprt, task);
1514 }
1515 
1516 /**
1517  * xprt_request_transmit - send an RPC request on a transport
1518  * @req: pointer to request to transmit
1519  * @snd_task: RPC task that owns the transport lock
1520  *
1521  * This performs the transmission of a single request.
1522  * Note that if the request is not the same as snd_task, then it
1523  * does need to be pinned.
1524  * Returns '0' on success.
1525  */
1526 static int
1527 xprt_request_transmit(struct rpc_rqst *req, struct rpc_task *snd_task)
1528 {
1529 	struct rpc_xprt *xprt = req->rq_xprt;
1530 	struct rpc_task *task = req->rq_task;
1531 	unsigned int connect_cookie;
1532 	int is_retrans = RPC_WAS_SENT(task);
1533 	int status;
1534 
1535 	if (!req->rq_bytes_sent) {
1536 		if (xprt_request_data_received(task)) {
1537 			status = 0;
1538 			goto out_dequeue;
1539 		}
1540 		/* Verify that our message lies in the RPCSEC_GSS window */
1541 		if (rpcauth_xmit_need_reencode(task)) {
1542 			status = -EBADMSG;
1543 			goto out_dequeue;
1544 		}
1545 		if (RPC_SIGNALLED(task)) {
1546 			status = -ERESTARTSYS;
1547 			goto out_dequeue;
1548 		}
1549 	}
1550 
1551 	/*
1552 	 * Update req->rq_ntrans before transmitting to avoid races with
1553 	 * xprt_update_rtt(), which needs to know that it is recording a
1554 	 * reply to the first transmission.
1555 	 */
1556 	req->rq_ntrans++;
1557 
1558 	trace_rpc_xdr_sendto(task, &req->rq_snd_buf);
1559 	connect_cookie = xprt->connect_cookie;
1560 	status = xprt->ops->send_request(req);
1561 	if (status != 0) {
1562 		req->rq_ntrans--;
1563 		trace_xprt_transmit(req, status);
1564 		return status;
1565 	}
1566 
1567 	if (is_retrans) {
1568 		task->tk_client->cl_stats->rpcretrans++;
1569 		trace_xprt_retransmit(req);
1570 	}
1571 
1572 	xprt_inject_disconnect(xprt);
1573 
1574 	task->tk_flags |= RPC_TASK_SENT;
1575 	spin_lock(&xprt->transport_lock);
1576 
1577 	xprt->stat.sends++;
1578 	xprt->stat.req_u += xprt->stat.sends - xprt->stat.recvs;
1579 	xprt->stat.bklog_u += xprt->backlog.qlen;
1580 	xprt->stat.sending_u += xprt->sending.qlen;
1581 	xprt->stat.pending_u += xprt->pending.qlen;
1582 	spin_unlock(&xprt->transport_lock);
1583 
1584 	req->rq_connect_cookie = connect_cookie;
1585 out_dequeue:
1586 	trace_xprt_transmit(req, status);
1587 	xprt_request_dequeue_transmit(task);
1588 	rpc_wake_up_queued_task_set_status(&xprt->sending, task, status);
1589 	return status;
1590 }
1591 
1592 /**
1593  * xprt_transmit - send an RPC request on a transport
1594  * @task: controlling RPC task
1595  *
1596  * Attempts to drain the transmit queue. On exit, either the transport
1597  * signalled an error that needs to be handled before transmission can
1598  * resume, or @task finished transmitting, and detected that it already
1599  * received a reply.
1600  */
1601 void
1602 xprt_transmit(struct rpc_task *task)
1603 {
1604 	struct rpc_rqst *next, *req = task->tk_rqstp;
1605 	struct rpc_xprt	*xprt = req->rq_xprt;
1606 	int counter, status;
1607 
1608 	spin_lock(&xprt->queue_lock);
1609 	counter = 0;
1610 	while (!list_empty(&xprt->xmit_queue)) {
1611 		if (++counter == 20)
1612 			break;
1613 		next = list_first_entry(&xprt->xmit_queue,
1614 				struct rpc_rqst, rq_xmit);
1615 		xprt_pin_rqst(next);
1616 		spin_unlock(&xprt->queue_lock);
1617 		status = xprt_request_transmit(next, task);
1618 		if (status == -EBADMSG && next != req)
1619 			status = 0;
1620 		spin_lock(&xprt->queue_lock);
1621 		xprt_unpin_rqst(next);
1622 		if (status == 0) {
1623 			if (!xprt_request_data_received(task) ||
1624 			    test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1625 				continue;
1626 		} else if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate))
1627 			task->tk_status = status;
1628 		break;
1629 	}
1630 	spin_unlock(&xprt->queue_lock);
1631 }
1632 
1633 static void xprt_complete_request_init(struct rpc_task *task)
1634 {
1635 	if (task->tk_rqstp)
1636 		xprt_request_init(task);
1637 }
1638 
1639 void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task)
1640 {
1641 	set_bit(XPRT_CONGESTED, &xprt->state);
1642 	rpc_sleep_on(&xprt->backlog, task, xprt_complete_request_init);
1643 }
1644 EXPORT_SYMBOL_GPL(xprt_add_backlog);
1645 
1646 static bool __xprt_set_rq(struct rpc_task *task, void *data)
1647 {
1648 	struct rpc_rqst *req = data;
1649 
1650 	if (task->tk_rqstp == NULL) {
1651 		memset(req, 0, sizeof(*req));	/* mark unused */
1652 		task->tk_rqstp = req;
1653 		return true;
1654 	}
1655 	return false;
1656 }
1657 
1658 bool xprt_wake_up_backlog(struct rpc_xprt *xprt, struct rpc_rqst *req)
1659 {
1660 	if (rpc_wake_up_first(&xprt->backlog, __xprt_set_rq, req) == NULL) {
1661 		clear_bit(XPRT_CONGESTED, &xprt->state);
1662 		return false;
1663 	}
1664 	return true;
1665 }
1666 EXPORT_SYMBOL_GPL(xprt_wake_up_backlog);
1667 
1668 static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task)
1669 {
1670 	bool ret = false;
1671 
1672 	if (!test_bit(XPRT_CONGESTED, &xprt->state))
1673 		goto out;
1674 	spin_lock(&xprt->reserve_lock);
1675 	if (test_bit(XPRT_CONGESTED, &xprt->state)) {
1676 		xprt_add_backlog(xprt, task);
1677 		ret = true;
1678 	}
1679 	spin_unlock(&xprt->reserve_lock);
1680 out:
1681 	return ret;
1682 }
1683 
1684 static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt)
1685 {
1686 	struct rpc_rqst *req = ERR_PTR(-EAGAIN);
1687 
1688 	if (xprt->num_reqs >= xprt->max_reqs)
1689 		goto out;
1690 	++xprt->num_reqs;
1691 	spin_unlock(&xprt->reserve_lock);
1692 	req = kzalloc(sizeof(struct rpc_rqst), GFP_NOFS);
1693 	spin_lock(&xprt->reserve_lock);
1694 	if (req != NULL)
1695 		goto out;
1696 	--xprt->num_reqs;
1697 	req = ERR_PTR(-ENOMEM);
1698 out:
1699 	return req;
1700 }
1701 
1702 static bool xprt_dynamic_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1703 {
1704 	if (xprt->num_reqs > xprt->min_reqs) {
1705 		--xprt->num_reqs;
1706 		kfree(req);
1707 		return true;
1708 	}
1709 	return false;
1710 }
1711 
1712 void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task)
1713 {
1714 	struct rpc_rqst *req;
1715 
1716 	spin_lock(&xprt->reserve_lock);
1717 	if (!list_empty(&xprt->free)) {
1718 		req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
1719 		list_del(&req->rq_list);
1720 		goto out_init_req;
1721 	}
1722 	req = xprt_dynamic_alloc_slot(xprt);
1723 	if (!IS_ERR(req))
1724 		goto out_init_req;
1725 	switch (PTR_ERR(req)) {
1726 	case -ENOMEM:
1727 		dprintk("RPC:       dynamic allocation of request slot "
1728 				"failed! Retrying\n");
1729 		task->tk_status = -ENOMEM;
1730 		break;
1731 	case -EAGAIN:
1732 		xprt_add_backlog(xprt, task);
1733 		dprintk("RPC:       waiting for request slot\n");
1734 		fallthrough;
1735 	default:
1736 		task->tk_status = -EAGAIN;
1737 	}
1738 	spin_unlock(&xprt->reserve_lock);
1739 	return;
1740 out_init_req:
1741 	xprt->stat.max_slots = max_t(unsigned int, xprt->stat.max_slots,
1742 				     xprt->num_reqs);
1743 	spin_unlock(&xprt->reserve_lock);
1744 
1745 	task->tk_status = 0;
1746 	task->tk_rqstp = req;
1747 }
1748 EXPORT_SYMBOL_GPL(xprt_alloc_slot);
1749 
1750 void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req)
1751 {
1752 	spin_lock(&xprt->reserve_lock);
1753 	if (!xprt_wake_up_backlog(xprt, req) &&
1754 	    !xprt_dynamic_free_slot(xprt, req)) {
1755 		memset(req, 0, sizeof(*req));	/* mark unused */
1756 		list_add(&req->rq_list, &xprt->free);
1757 	}
1758 	spin_unlock(&xprt->reserve_lock);
1759 }
1760 EXPORT_SYMBOL_GPL(xprt_free_slot);
1761 
1762 static void xprt_free_all_slots(struct rpc_xprt *xprt)
1763 {
1764 	struct rpc_rqst *req;
1765 	while (!list_empty(&xprt->free)) {
1766 		req = list_first_entry(&xprt->free, struct rpc_rqst, rq_list);
1767 		list_del(&req->rq_list);
1768 		kfree(req);
1769 	}
1770 }
1771 
1772 static DEFINE_IDA(rpc_xprt_ids);
1773 
1774 void xprt_cleanup_ids(void)
1775 {
1776 	ida_destroy(&rpc_xprt_ids);
1777 }
1778 
1779 static int xprt_alloc_id(struct rpc_xprt *xprt)
1780 {
1781 	int id;
1782 
1783 	id = ida_simple_get(&rpc_xprt_ids, 0, 0, GFP_KERNEL);
1784 	if (id < 0)
1785 		return id;
1786 
1787 	xprt->id = id;
1788 	return 0;
1789 }
1790 
1791 static void xprt_free_id(struct rpc_xprt *xprt)
1792 {
1793 	ida_simple_remove(&rpc_xprt_ids, xprt->id);
1794 }
1795 
1796 struct rpc_xprt *xprt_alloc(struct net *net, size_t size,
1797 		unsigned int num_prealloc,
1798 		unsigned int max_alloc)
1799 {
1800 	struct rpc_xprt *xprt;
1801 	struct rpc_rqst *req;
1802 	int i;
1803 
1804 	xprt = kzalloc(size, GFP_KERNEL);
1805 	if (xprt == NULL)
1806 		goto out;
1807 
1808 	xprt_alloc_id(xprt);
1809 	xprt_init(xprt, net);
1810 
1811 	for (i = 0; i < num_prealloc; i++) {
1812 		req = kzalloc(sizeof(struct rpc_rqst), GFP_KERNEL);
1813 		if (!req)
1814 			goto out_free;
1815 		list_add(&req->rq_list, &xprt->free);
1816 	}
1817 	if (max_alloc > num_prealloc)
1818 		xprt->max_reqs = max_alloc;
1819 	else
1820 		xprt->max_reqs = num_prealloc;
1821 	xprt->min_reqs = num_prealloc;
1822 	xprt->num_reqs = num_prealloc;
1823 
1824 	return xprt;
1825 
1826 out_free:
1827 	xprt_free(xprt);
1828 out:
1829 	return NULL;
1830 }
1831 EXPORT_SYMBOL_GPL(xprt_alloc);
1832 
1833 void xprt_free(struct rpc_xprt *xprt)
1834 {
1835 	put_net(xprt->xprt_net);
1836 	xprt_free_all_slots(xprt);
1837 	xprt_free_id(xprt);
1838 	rpc_sysfs_xprt_destroy(xprt);
1839 	kfree_rcu(xprt, rcu);
1840 }
1841 EXPORT_SYMBOL_GPL(xprt_free);
1842 
1843 static void
1844 xprt_init_connect_cookie(struct rpc_rqst *req, struct rpc_xprt *xprt)
1845 {
1846 	req->rq_connect_cookie = xprt_connect_cookie(xprt) - 1;
1847 }
1848 
1849 static __be32
1850 xprt_alloc_xid(struct rpc_xprt *xprt)
1851 {
1852 	__be32 xid;
1853 
1854 	spin_lock(&xprt->reserve_lock);
1855 	xid = (__force __be32)xprt->xid++;
1856 	spin_unlock(&xprt->reserve_lock);
1857 	return xid;
1858 }
1859 
1860 static void
1861 xprt_init_xid(struct rpc_xprt *xprt)
1862 {
1863 	xprt->xid = prandom_u32();
1864 }
1865 
1866 static void
1867 xprt_request_init(struct rpc_task *task)
1868 {
1869 	struct rpc_xprt *xprt = task->tk_xprt;
1870 	struct rpc_rqst	*req = task->tk_rqstp;
1871 
1872 	req->rq_task	= task;
1873 	req->rq_xprt    = xprt;
1874 	req->rq_buffer  = NULL;
1875 	req->rq_xid	= xprt_alloc_xid(xprt);
1876 	xprt_init_connect_cookie(req, xprt);
1877 	req->rq_snd_buf.len = 0;
1878 	req->rq_snd_buf.buflen = 0;
1879 	req->rq_rcv_buf.len = 0;
1880 	req->rq_rcv_buf.buflen = 0;
1881 	req->rq_snd_buf.bvec = NULL;
1882 	req->rq_rcv_buf.bvec = NULL;
1883 	req->rq_release_snd_buf = NULL;
1884 	xprt_init_majortimeo(task, req);
1885 
1886 	trace_xprt_reserve(req);
1887 }
1888 
1889 static void
1890 xprt_do_reserve(struct rpc_xprt *xprt, struct rpc_task *task)
1891 {
1892 	xprt->ops->alloc_slot(xprt, task);
1893 	if (task->tk_rqstp != NULL)
1894 		xprt_request_init(task);
1895 }
1896 
1897 /**
1898  * xprt_reserve - allocate an RPC request slot
1899  * @task: RPC task requesting a slot allocation
1900  *
1901  * If the transport is marked as being congested, or if no more
1902  * slots are available, place the task on the transport's
1903  * backlog queue.
1904  */
1905 void xprt_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 	if (!xprt_throttle_congested(xprt, task))
1915 		xprt_do_reserve(xprt, task);
1916 }
1917 
1918 /**
1919  * xprt_retry_reserve - allocate an RPC request slot
1920  * @task: RPC task requesting a slot allocation
1921  *
1922  * If no more slots are available, place the task on the transport's
1923  * backlog queue.
1924  * Note that the only difference with xprt_reserve is that we now
1925  * ignore the value of the XPRT_CONGESTED flag.
1926  */
1927 void xprt_retry_reserve(struct rpc_task *task)
1928 {
1929 	struct rpc_xprt *xprt = task->tk_xprt;
1930 
1931 	task->tk_status = 0;
1932 	if (task->tk_rqstp != NULL)
1933 		return;
1934 
1935 	task->tk_status = -EAGAIN;
1936 	xprt_do_reserve(xprt, task);
1937 }
1938 
1939 /**
1940  * xprt_release - release an RPC request slot
1941  * @task: task which is finished with the slot
1942  *
1943  */
1944 void xprt_release(struct rpc_task *task)
1945 {
1946 	struct rpc_xprt	*xprt;
1947 	struct rpc_rqst	*req = task->tk_rqstp;
1948 
1949 	if (req == NULL) {
1950 		if (task->tk_client) {
1951 			xprt = task->tk_xprt;
1952 			xprt_release_write(xprt, task);
1953 		}
1954 		return;
1955 	}
1956 
1957 	xprt = req->rq_xprt;
1958 	xprt_request_dequeue_xprt(task);
1959 	spin_lock(&xprt->transport_lock);
1960 	xprt->ops->release_xprt(xprt, task);
1961 	if (xprt->ops->release_request)
1962 		xprt->ops->release_request(task);
1963 	xprt_schedule_autodisconnect(xprt);
1964 	spin_unlock(&xprt->transport_lock);
1965 	if (req->rq_buffer)
1966 		xprt->ops->buf_free(task);
1967 	xdr_free_bvec(&req->rq_rcv_buf);
1968 	xdr_free_bvec(&req->rq_snd_buf);
1969 	if (req->rq_cred != NULL)
1970 		put_rpccred(req->rq_cred);
1971 	if (req->rq_release_snd_buf)
1972 		req->rq_release_snd_buf(req);
1973 
1974 	task->tk_rqstp = NULL;
1975 	if (likely(!bc_prealloc(req)))
1976 		xprt->ops->free_slot(xprt, req);
1977 	else
1978 		xprt_free_bc_request(req);
1979 }
1980 
1981 #ifdef CONFIG_SUNRPC_BACKCHANNEL
1982 void
1983 xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task)
1984 {
1985 	struct xdr_buf *xbufp = &req->rq_snd_buf;
1986 
1987 	task->tk_rqstp = req;
1988 	req->rq_task = task;
1989 	xprt_init_connect_cookie(req, req->rq_xprt);
1990 	/*
1991 	 * Set up the xdr_buf length.
1992 	 * This also indicates that the buffer is XDR encoded already.
1993 	 */
1994 	xbufp->len = xbufp->head[0].iov_len + xbufp->page_len +
1995 		xbufp->tail[0].iov_len;
1996 }
1997 #endif
1998 
1999 static void xprt_init(struct rpc_xprt *xprt, struct net *net)
2000 {
2001 	kref_init(&xprt->kref);
2002 
2003 	spin_lock_init(&xprt->transport_lock);
2004 	spin_lock_init(&xprt->reserve_lock);
2005 	spin_lock_init(&xprt->queue_lock);
2006 
2007 	INIT_LIST_HEAD(&xprt->free);
2008 	xprt->recv_queue = RB_ROOT;
2009 	INIT_LIST_HEAD(&xprt->xmit_queue);
2010 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
2011 	spin_lock_init(&xprt->bc_pa_lock);
2012 	INIT_LIST_HEAD(&xprt->bc_pa_list);
2013 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
2014 	INIT_LIST_HEAD(&xprt->xprt_switch);
2015 
2016 	xprt->last_used = jiffies;
2017 	xprt->cwnd = RPC_INITCWND;
2018 	xprt->bind_index = 0;
2019 
2020 	rpc_init_wait_queue(&xprt->binding, "xprt_binding");
2021 	rpc_init_wait_queue(&xprt->pending, "xprt_pending");
2022 	rpc_init_wait_queue(&xprt->sending, "xprt_sending");
2023 	rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
2024 
2025 	xprt_init_xid(xprt);
2026 
2027 	xprt->xprt_net = get_net(net);
2028 }
2029 
2030 /**
2031  * xprt_create_transport - create an RPC transport
2032  * @args: rpc transport creation arguments
2033  *
2034  */
2035 struct rpc_xprt *xprt_create_transport(struct xprt_create *args)
2036 {
2037 	struct rpc_xprt	*xprt;
2038 	const struct xprt_class *t;
2039 
2040 	t = xprt_class_find_by_ident(args->ident);
2041 	if (!t) {
2042 		dprintk("RPC: transport (%d) not supported\n", args->ident);
2043 		return ERR_PTR(-EIO);
2044 	}
2045 
2046 	xprt = t->setup(args);
2047 	xprt_class_release(t);
2048 
2049 	if (IS_ERR(xprt))
2050 		goto out;
2051 	if (args->flags & XPRT_CREATE_NO_IDLE_TIMEOUT)
2052 		xprt->idle_timeout = 0;
2053 	INIT_WORK(&xprt->task_cleanup, xprt_autoclose);
2054 	if (xprt_has_timer(xprt))
2055 		timer_setup(&xprt->timer, xprt_init_autodisconnect, 0);
2056 	else
2057 		timer_setup(&xprt->timer, NULL, 0);
2058 
2059 	if (strlen(args->servername) > RPC_MAXNETNAMELEN) {
2060 		xprt_destroy(xprt);
2061 		return ERR_PTR(-EINVAL);
2062 	}
2063 	xprt->servername = kstrdup(args->servername, GFP_KERNEL);
2064 	if (xprt->servername == NULL) {
2065 		xprt_destroy(xprt);
2066 		return ERR_PTR(-ENOMEM);
2067 	}
2068 
2069 	rpc_xprt_debugfs_register(xprt);
2070 
2071 	trace_xprt_create(xprt);
2072 out:
2073 	return xprt;
2074 }
2075 
2076 static void xprt_destroy_cb(struct work_struct *work)
2077 {
2078 	struct rpc_xprt *xprt =
2079 		container_of(work, struct rpc_xprt, task_cleanup);
2080 
2081 	trace_xprt_destroy(xprt);
2082 
2083 	rpc_xprt_debugfs_unregister(xprt);
2084 	rpc_destroy_wait_queue(&xprt->binding);
2085 	rpc_destroy_wait_queue(&xprt->pending);
2086 	rpc_destroy_wait_queue(&xprt->sending);
2087 	rpc_destroy_wait_queue(&xprt->backlog);
2088 	kfree(xprt->servername);
2089 	/*
2090 	 * Destroy any existing back channel
2091 	 */
2092 	xprt_destroy_backchannel(xprt, UINT_MAX);
2093 
2094 	/*
2095 	 * Tear down transport state and free the rpc_xprt
2096 	 */
2097 	xprt->ops->destroy(xprt);
2098 }
2099 
2100 /**
2101  * xprt_destroy - destroy an RPC transport, killing off all requests.
2102  * @xprt: transport to destroy
2103  *
2104  */
2105 static void xprt_destroy(struct rpc_xprt *xprt)
2106 {
2107 	/*
2108 	 * Exclude transport connect/disconnect handlers and autoclose
2109 	 */
2110 	wait_on_bit_lock(&xprt->state, XPRT_LOCKED, TASK_UNINTERRUPTIBLE);
2111 
2112 	del_timer_sync(&xprt->timer);
2113 
2114 	/*
2115 	 * Destroy sockets etc from the system workqueue so they can
2116 	 * safely flush receive work running on rpciod.
2117 	 */
2118 	INIT_WORK(&xprt->task_cleanup, xprt_destroy_cb);
2119 	schedule_work(&xprt->task_cleanup);
2120 }
2121 
2122 static void xprt_destroy_kref(struct kref *kref)
2123 {
2124 	xprt_destroy(container_of(kref, struct rpc_xprt, kref));
2125 }
2126 
2127 /**
2128  * xprt_get - return a reference to an RPC transport.
2129  * @xprt: pointer to the transport
2130  *
2131  */
2132 struct rpc_xprt *xprt_get(struct rpc_xprt *xprt)
2133 {
2134 	if (xprt != NULL && kref_get_unless_zero(&xprt->kref))
2135 		return xprt;
2136 	return NULL;
2137 }
2138 EXPORT_SYMBOL_GPL(xprt_get);
2139 
2140 /**
2141  * xprt_put - release a reference to an RPC transport.
2142  * @xprt: pointer to the transport
2143  *
2144  */
2145 void xprt_put(struct rpc_xprt *xprt)
2146 {
2147 	if (xprt != NULL)
2148 		kref_put(&xprt->kref, xprt_destroy_kref);
2149 }
2150 EXPORT_SYMBOL_GPL(xprt_put);
2151