xprt.c revision fe3aca290f17ae4978bd73d02aa4029f1c9c024c
1/*
2 *  linux/net/sunrpc/xprt.c
3 *
4 *  This is a generic RPC call interface supporting congestion avoidance,
5 *  and asynchronous calls.
6 *
7 *  The interface works like this:
8 *
9 *  -	When a process places a call, it allocates a request slot if
10 *	one is available. Otherwise, it sleeps on the backlog queue
11 *	(xprt_reserve).
12 *  -	Next, the caller puts together the RPC message, stuffs it into
13 *	the request struct, and calls xprt_transmit().
14 *  -	xprt_transmit sends the message and installs the caller on the
15 *	transport's wait list. At the same time, it installs a timer that
16 *	is run after the packet's timeout has expired.
17 *  -	When a packet arrives, the data_ready handler walks the list of
18 *	pending requests for that transport. If a matching XID is found, the
19 *	caller is woken up, and the timer removed.
20 *  -	When no reply arrives within the timeout interval, the timer is
21 *	fired by the kernel and runs xprt_timer(). It either adjusts the
22 *	timeout values (minor timeout) or wakes up the caller with a status
23 *	of -ETIMEDOUT.
24 *  -	When the caller receives a notification from RPC that a reply arrived,
25 *	it should release the RPC slot, and process the reply.
26 *	If the call timed out, it may choose to retry the operation by
27 *	adjusting the initial timeout value, and simply calling rpc_call
28 *	again.
29 *
30 *  Support for async RPC is done through a set of RPC-specific scheduling
31 *  primitives that `transparently' work for processes as well as async
32 *  tasks that rely on callbacks.
33 *
34 *  Copyright (C) 1995-1997, Olaf Kirch <okir@monad.swb.de>
35 *
36 *  Transport switch API copyright (C) 2005, Chuck Lever <cel@netapp.com>
37 */
38
39#include <linux/module.h>
40
41#include <linux/types.h>
42#include <linux/interrupt.h>
43#include <linux/workqueue.h>
44#include <linux/random.h>
45
46#include <linux/sunrpc/clnt.h>
47
48/*
49 * Local variables
50 */
51
52#ifdef RPC_DEBUG
53# undef  RPC_DEBUG_DATA
54# define RPCDBG_FACILITY	RPCDBG_XPRT
55#endif
56
57/*
58 * Local functions
59 */
60static void	xprt_request_init(struct rpc_task *, struct rpc_xprt *);
61static inline void	do_xprt_reserve(struct rpc_task *);
62static void	xprt_connect_status(struct rpc_task *task);
63static int      __xprt_get_cong(struct rpc_xprt *, struct rpc_task *);
64
65static int	xprt_clear_backlog(struct rpc_xprt *xprt);
66
67/*
68 * Serialize write access to transports, in order to prevent different
69 * requests from interfering with each other.
70 * Also prevents transport connects from colliding with writes.
71 */
72static int
73__xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
74{
75	struct rpc_rqst *req = task->tk_rqstp;
76
77	if (test_and_set_bit(XPRT_LOCKED, &xprt->state)) {
78		if (task == xprt->snd_task)
79			return 1;
80		goto out_sleep;
81	}
82	if (xprt->nocong || __xprt_get_cong(xprt, task)) {
83		xprt->snd_task = task;
84		if (req) {
85			req->rq_bytes_sent = 0;
86			req->rq_ntrans++;
87		}
88		return 1;
89	}
90	smp_mb__before_clear_bit();
91	clear_bit(XPRT_LOCKED, &xprt->state);
92	smp_mb__after_clear_bit();
93out_sleep:
94	dprintk("RPC: %4d failed to lock transport %p\n", task->tk_pid, xprt);
95	task->tk_timeout = 0;
96	task->tk_status = -EAGAIN;
97	if (req && req->rq_ntrans)
98		rpc_sleep_on(&xprt->resend, task, NULL, NULL);
99	else
100		rpc_sleep_on(&xprt->sending, task, NULL, NULL);
101	return 0;
102}
103
104static inline int
105xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
106{
107	int retval;
108
109	spin_lock_bh(&xprt->transport_lock);
110	retval = __xprt_lock_write(xprt, task);
111	spin_unlock_bh(&xprt->transport_lock);
112	return retval;
113}
114
115
116static void
117__xprt_lock_write_next(struct rpc_xprt *xprt)
118{
119	struct rpc_task *task;
120
121	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
122		return;
123	if (!xprt->nocong && RPCXPRT_CONGESTED(xprt))
124		goto out_unlock;
125	task = rpc_wake_up_next(&xprt->resend);
126	if (!task) {
127		task = rpc_wake_up_next(&xprt->sending);
128		if (!task)
129			goto out_unlock;
130	}
131	if (xprt->nocong || __xprt_get_cong(xprt, task)) {
132		struct rpc_rqst *req = task->tk_rqstp;
133		xprt->snd_task = task;
134		if (req) {
135			req->rq_bytes_sent = 0;
136			req->rq_ntrans++;
137		}
138		return;
139	}
140out_unlock:
141	smp_mb__before_clear_bit();
142	clear_bit(XPRT_LOCKED, &xprt->state);
143	smp_mb__after_clear_bit();
144}
145
146/*
147 * Releases the transport for use by other requests.
148 */
149static void
150__xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
151{
152	if (xprt->snd_task == task) {
153		xprt->snd_task = NULL;
154		smp_mb__before_clear_bit();
155		clear_bit(XPRT_LOCKED, &xprt->state);
156		smp_mb__after_clear_bit();
157		__xprt_lock_write_next(xprt);
158	}
159}
160
161static inline void
162xprt_release_write(struct rpc_xprt *xprt, struct rpc_task *task)
163{
164	spin_lock_bh(&xprt->transport_lock);
165	__xprt_release_write(xprt, task);
166	spin_unlock_bh(&xprt->transport_lock);
167}
168
169/*
170 * Van Jacobson congestion avoidance. Check if the congestion window
171 * overflowed. Put the task to sleep if this is the case.
172 */
173static int
174__xprt_get_cong(struct rpc_xprt *xprt, struct rpc_task *task)
175{
176	struct rpc_rqst *req = task->tk_rqstp;
177
178	if (req->rq_cong)
179		return 1;
180	dprintk("RPC: %4d xprt_cwnd_limited cong = %ld cwnd = %ld\n",
181			task->tk_pid, xprt->cong, xprt->cwnd);
182	if (RPCXPRT_CONGESTED(xprt))
183		return 0;
184	req->rq_cong = 1;
185	xprt->cong += RPC_CWNDSCALE;
186	return 1;
187}
188
189/*
190 * Adjust the congestion window, and wake up the next task
191 * that has been sleeping due to congestion
192 */
193static void
194__xprt_put_cong(struct rpc_xprt *xprt, struct rpc_rqst *req)
195{
196	if (!req->rq_cong)
197		return;
198	req->rq_cong = 0;
199	xprt->cong -= RPC_CWNDSCALE;
200	__xprt_lock_write_next(xprt);
201}
202
203/*
204 * Adjust RPC congestion window
205 * We use a time-smoothed congestion estimator to avoid heavy oscillation.
206 */
207static void
208xprt_adjust_cwnd(struct rpc_xprt *xprt, int result)
209{
210	unsigned long	cwnd;
211
212	cwnd = xprt->cwnd;
213	if (result >= 0 && cwnd <= xprt->cong) {
214		/* The (cwnd >> 1) term makes sure
215		 * the result gets rounded properly. */
216		cwnd += (RPC_CWNDSCALE * RPC_CWNDSCALE + (cwnd >> 1)) / cwnd;
217		if (cwnd > RPC_MAXCWND(xprt))
218			cwnd = RPC_MAXCWND(xprt);
219		__xprt_lock_write_next(xprt);
220	} else if (result == -ETIMEDOUT) {
221		cwnd >>= 1;
222		if (cwnd < RPC_CWNDSCALE)
223			cwnd = RPC_CWNDSCALE;
224	}
225	dprintk("RPC:      cong %ld, cwnd was %ld, now %ld\n",
226			xprt->cong, xprt->cwnd, cwnd);
227	xprt->cwnd = cwnd;
228}
229
230/**
231 * xprt_wake_pending_tasks - wake all tasks on a transport's pending queue
232 * @xprt: transport with waiting tasks
233 * @status: result code to plant in each task before waking it
234 *
235 */
236void xprt_wake_pending_tasks(struct rpc_xprt *xprt, int status)
237{
238	if (status < 0)
239		rpc_wake_up_status(&xprt->pending, status);
240	else
241		rpc_wake_up(&xprt->pending);
242}
243
244/**
245 * xprt_wait_for_buffer_space - wait for transport output buffer to clear
246 * @task: task to be put to sleep
247 *
248 */
249void xprt_wait_for_buffer_space(struct rpc_task *task)
250{
251	struct rpc_rqst *req = task->tk_rqstp;
252	struct rpc_xprt *xprt = req->rq_xprt;
253
254	task->tk_timeout = req->rq_timeout;
255	rpc_sleep_on(&xprt->pending, task, NULL, NULL);
256}
257
258/**
259 * xprt_write_space - wake the task waiting for transport output buffer space
260 * @xprt: transport with waiting tasks
261 *
262 * Can be called in a soft IRQ context, so xprt_write_space never sleeps.
263 */
264void xprt_write_space(struct rpc_xprt *xprt)
265{
266	if (unlikely(xprt->shutdown))
267		return;
268
269	spin_lock_bh(&xprt->transport_lock);
270	if (xprt->snd_task) {
271		dprintk("RPC:      write space: waking waiting task on xprt %p\n",
272				xprt);
273		rpc_wake_up_task(xprt->snd_task);
274	}
275	spin_unlock_bh(&xprt->transport_lock);
276}
277
278/**
279 * xprt_set_retrans_timeout_def - set a request's retransmit timeout
280 * @task: task whose timeout is to be set
281 *
282 * Set a request's retransmit timeout based on the transport's
283 * default timeout parameters.  Used by transports that don't adjust
284 * the retransmit timeout based on round-trip time estimation.
285 */
286void xprt_set_retrans_timeout_def(struct rpc_task *task)
287{
288	task->tk_timeout = task->tk_rqstp->rq_timeout;
289}
290
291/*
292 * xprt_set_retrans_timeout_rtt - set a request's retransmit timeout
293 * @task: task whose timeout is to be set
294 *
295 * Set a request's retransmit timeout using the RTT estimator.
296 */
297void xprt_set_retrans_timeout_rtt(struct rpc_task *task)
298{
299	int timer = task->tk_msg.rpc_proc->p_timer;
300	struct rpc_rtt *rtt = task->tk_client->cl_rtt;
301	struct rpc_rqst *req = task->tk_rqstp;
302	unsigned long max_timeout = req->rq_xprt->timeout.to_maxval;
303
304	task->tk_timeout = rpc_calc_rto(rtt, timer);
305	task->tk_timeout <<= rpc_ntimeo(rtt, timer) + req->rq_retries;
306	if (task->tk_timeout > max_timeout || task->tk_timeout == 0)
307		task->tk_timeout = max_timeout;
308}
309
310static void xprt_reset_majortimeo(struct rpc_rqst *req)
311{
312	struct rpc_timeout *to = &req->rq_xprt->timeout;
313
314	req->rq_majortimeo = req->rq_timeout;
315	if (to->to_exponential)
316		req->rq_majortimeo <<= to->to_retries;
317	else
318		req->rq_majortimeo += to->to_increment * to->to_retries;
319	if (req->rq_majortimeo > to->to_maxval || req->rq_majortimeo == 0)
320		req->rq_majortimeo = to->to_maxval;
321	req->rq_majortimeo += jiffies;
322}
323
324/**
325 * xprt_adjust_timeout - adjust timeout values for next retransmit
326 * @req: RPC request containing parameters to use for the adjustment
327 *
328 */
329int xprt_adjust_timeout(struct rpc_rqst *req)
330{
331	struct rpc_xprt *xprt = req->rq_xprt;
332	struct rpc_timeout *to = &xprt->timeout;
333	int status = 0;
334
335	if (time_before(jiffies, req->rq_majortimeo)) {
336		if (to->to_exponential)
337			req->rq_timeout <<= 1;
338		else
339			req->rq_timeout += to->to_increment;
340		if (to->to_maxval && req->rq_timeout >= to->to_maxval)
341			req->rq_timeout = to->to_maxval;
342		req->rq_retries++;
343		pprintk("RPC: %lu retrans\n", jiffies);
344	} else {
345		req->rq_timeout = to->to_initval;
346		req->rq_retries = 0;
347		xprt_reset_majortimeo(req);
348		/* Reset the RTT counters == "slow start" */
349		spin_lock_bh(&xprt->transport_lock);
350		rpc_init_rtt(req->rq_task->tk_client->cl_rtt, to->to_initval);
351		spin_unlock_bh(&xprt->transport_lock);
352		pprintk("RPC: %lu timeout\n", jiffies);
353		status = -ETIMEDOUT;
354	}
355
356	if (req->rq_timeout == 0) {
357		printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n");
358		req->rq_timeout = 5 * HZ;
359	}
360	return status;
361}
362
363static void xprt_autoclose(void *args)
364{
365	struct rpc_xprt *xprt = (struct rpc_xprt *)args;
366
367	xprt_disconnect(xprt);
368	xprt->ops->close(xprt);
369	xprt_release_write(xprt, NULL);
370}
371
372/**
373 * xprt_disconnect - mark a transport as disconnected
374 * @xprt: transport to flag for disconnect
375 *
376 */
377void xprt_disconnect(struct rpc_xprt *xprt)
378{
379	dprintk("RPC:      disconnected transport %p\n", xprt);
380	spin_lock_bh(&xprt->transport_lock);
381	xprt_clear_connected(xprt);
382	xprt_wake_pending_tasks(xprt, -ENOTCONN);
383	spin_unlock_bh(&xprt->transport_lock);
384}
385
386static void
387xprt_init_autodisconnect(unsigned long data)
388{
389	struct rpc_xprt *xprt = (struct rpc_xprt *)data;
390
391	spin_lock(&xprt->transport_lock);
392	if (!list_empty(&xprt->recv) || xprt->shutdown)
393		goto out_abort;
394	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
395		goto out_abort;
396	spin_unlock(&xprt->transport_lock);
397	if (xprt_connecting(xprt))
398		xprt_release_write(xprt, NULL);
399	else
400		schedule_work(&xprt->task_cleanup);
401	return;
402out_abort:
403	spin_unlock(&xprt->transport_lock);
404}
405
406/**
407 * xprt_connect - schedule a transport connect operation
408 * @task: RPC task that is requesting the connect
409 *
410 */
411void xprt_connect(struct rpc_task *task)
412{
413	struct rpc_xprt	*xprt = task->tk_xprt;
414
415	dprintk("RPC: %4d xprt_connect xprt %p %s connected\n", task->tk_pid,
416			xprt, (xprt_connected(xprt) ? "is" : "is not"));
417
418	if (xprt->shutdown) {
419		task->tk_status = -EIO;
420		return;
421	}
422	if (!xprt->addr.sin_port) {
423		task->tk_status = -EIO;
424		return;
425	}
426	if (!xprt_lock_write(xprt, task))
427		return;
428	if (xprt_connected(xprt))
429		xprt_release_write(xprt, task);
430	else {
431		if (task->tk_rqstp)
432			task->tk_rqstp->rq_bytes_sent = 0;
433
434		task->tk_timeout = RPC_CONNECT_TIMEOUT;
435		rpc_sleep_on(&xprt->pending, task, xprt_connect_status, NULL);
436		xprt->ops->connect(task);
437	}
438	return;
439}
440
441static void xprt_connect_status(struct rpc_task *task)
442{
443	struct rpc_xprt	*xprt = task->tk_xprt;
444
445	if (task->tk_status >= 0) {
446		dprintk("RPC: %4d xprt_connect_status: connection established\n",
447				task->tk_pid);
448		return;
449	}
450
451	switch (task->tk_status) {
452	case -ECONNREFUSED:
453	case -ECONNRESET:
454		dprintk("RPC: %4d xprt_connect_status: server %s refused connection\n",
455				task->tk_pid, task->tk_client->cl_server);
456		break;
457	case -ENOTCONN:
458		dprintk("RPC: %4d xprt_connect_status: connection broken\n",
459				task->tk_pid);
460		break;
461	case -ETIMEDOUT:
462		dprintk("RPC: %4d xprt_connect_status: connect attempt timed out\n",
463				task->tk_pid);
464		break;
465	default:
466		dprintk("RPC: %4d xprt_connect_status: error %d connecting to server %s\n",
467				task->tk_pid, -task->tk_status, task->tk_client->cl_server);
468		xprt_release_write(xprt, task);
469		task->tk_status = -EIO;
470		return;
471	}
472
473	/* if soft mounted, just cause this RPC to fail */
474	if (RPC_IS_SOFT(task)) {
475		xprt_release_write(xprt, task);
476		task->tk_status = -EIO;
477	}
478}
479
480/**
481 * xprt_lookup_rqst - find an RPC request corresponding to an XID
482 * @xprt: transport on which the original request was transmitted
483 * @xid: RPC XID of incoming reply
484 *
485 */
486struct rpc_rqst *xprt_lookup_rqst(struct rpc_xprt *xprt, u32 xid)
487{
488	struct list_head *pos;
489	struct rpc_rqst	*req = NULL;
490
491	list_for_each(pos, &xprt->recv) {
492		struct rpc_rqst *entry = list_entry(pos, struct rpc_rqst, rq_list);
493		if (entry->rq_xid == xid) {
494			req = entry;
495			break;
496		}
497	}
498	return req;
499}
500
501/**
502 * xprt_complete_rqst - called when reply processing is complete
503 * @xprt: controlling transport
504 * @req: RPC request that just completed
505 * @copied: actual number of bytes received from the transport
506 *
507 */
508void xprt_complete_rqst(struct rpc_xprt *xprt, struct rpc_rqst *req, int copied)
509{
510	struct rpc_task	*task = req->rq_task;
511	struct rpc_clnt *clnt = task->tk_client;
512
513	/* Adjust congestion window */
514	if (!xprt->nocong) {
515		unsigned timer = task->tk_msg.rpc_proc->p_timer;
516		xprt_adjust_cwnd(xprt, copied);
517		__xprt_put_cong(xprt, req);
518		if (timer) {
519			if (req->rq_ntrans == 1)
520				rpc_update_rtt(clnt->cl_rtt, timer,
521						(long)jiffies - req->rq_xtime);
522			rpc_set_timeo(clnt->cl_rtt, timer, req->rq_ntrans - 1);
523		}
524	}
525
526#ifdef RPC_PROFILE
527	/* Profile only reads for now */
528	if (copied > 1024) {
529		static unsigned long	nextstat;
530		static unsigned long	pkt_rtt, pkt_len, pkt_cnt;
531
532		pkt_cnt++;
533		pkt_len += req->rq_slen + copied;
534		pkt_rtt += jiffies - req->rq_xtime;
535		if (time_before(nextstat, jiffies)) {
536			printk("RPC: %lu %ld cwnd\n", jiffies, xprt->cwnd);
537			printk("RPC: %ld %ld %ld %ld stat\n",
538					jiffies, pkt_cnt, pkt_len, pkt_rtt);
539			pkt_rtt = pkt_len = pkt_cnt = 0;
540			nextstat = jiffies + 5 * HZ;
541		}
542	}
543#endif
544
545	dprintk("RPC: %4d has input (%d bytes)\n", task->tk_pid, copied);
546	list_del_init(&req->rq_list);
547	req->rq_received = req->rq_private_buf.len = copied;
548
549	/* ... and wake up the process. */
550	rpc_wake_up_task(task);
551	return;
552}
553
554/*
555 * RPC receive timeout handler.
556 */
557static void
558xprt_timer(struct rpc_task *task)
559{
560	struct rpc_rqst	*req = task->tk_rqstp;
561	struct rpc_xprt *xprt = req->rq_xprt;
562
563	spin_lock(&xprt->transport_lock);
564	if (req->rq_received)
565		goto out;
566
567	xprt_adjust_cwnd(req->rq_xprt, -ETIMEDOUT);
568	__xprt_put_cong(xprt, req);
569
570	dprintk("RPC: %4d xprt_timer (%s request)\n",
571		task->tk_pid, req ? "pending" : "backlogged");
572
573	task->tk_status  = -ETIMEDOUT;
574out:
575	task->tk_timeout = 0;
576	rpc_wake_up_task(task);
577	spin_unlock(&xprt->transport_lock);
578}
579
580/**
581 * xprt_prepare_transmit - reserve the transport before sending a request
582 * @task: RPC task about to send a request
583 *
584 */
585int xprt_prepare_transmit(struct rpc_task *task)
586{
587	struct rpc_rqst	*req = task->tk_rqstp;
588	struct rpc_xprt	*xprt = req->rq_xprt;
589	int err = 0;
590
591	dprintk("RPC: %4d xprt_prepare_transmit\n", task->tk_pid);
592
593	if (xprt->shutdown)
594		return -EIO;
595
596	spin_lock_bh(&xprt->transport_lock);
597	if (req->rq_received && !req->rq_bytes_sent) {
598		err = req->rq_received;
599		goto out_unlock;
600	}
601	if (!__xprt_lock_write(xprt, task)) {
602		err = -EAGAIN;
603		goto out_unlock;
604	}
605
606	if (!xprt_connected(xprt)) {
607		err = -ENOTCONN;
608		goto out_unlock;
609	}
610out_unlock:
611	spin_unlock_bh(&xprt->transport_lock);
612	return err;
613}
614
615/**
616 * xprt_transmit - send an RPC request on a transport
617 * @task: controlling RPC task
618 *
619 * We have to copy the iovec because sendmsg fiddles with its contents.
620 */
621void xprt_transmit(struct rpc_task *task)
622{
623	struct rpc_rqst	*req = task->tk_rqstp;
624	struct rpc_xprt	*xprt = req->rq_xprt;
625	int status;
626
627	dprintk("RPC: %4d xprt_transmit(%u)\n", task->tk_pid, req->rq_slen);
628
629	smp_rmb();
630	if (!req->rq_received) {
631		if (list_empty(&req->rq_list)) {
632			spin_lock_bh(&xprt->transport_lock);
633			/* Update the softirq receive buffer */
634			memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
635					sizeof(req->rq_private_buf));
636			/* Add request to the receive list */
637			list_add_tail(&req->rq_list, &xprt->recv);
638			spin_unlock_bh(&xprt->transport_lock);
639			xprt_reset_majortimeo(req);
640			/* Turn off autodisconnect */
641			del_singleshot_timer_sync(&xprt->timer);
642		}
643	} else if (!req->rq_bytes_sent)
644		return;
645
646	status = xprt->ops->send_request(task);
647	if (status == 0) {
648		dprintk("RPC: %4d xmit complete\n", task->tk_pid);
649		spin_lock_bh(&xprt->transport_lock);
650		xprt->ops->set_retrans_timeout(task);
651		/* Don't race with disconnect */
652		if (!xprt_connected(xprt))
653			task->tk_status = -ENOTCONN;
654		else if (!req->rq_received)
655			rpc_sleep_on(&xprt->pending, task, NULL, xprt_timer);
656		__xprt_release_write(xprt, task);
657		spin_unlock_bh(&xprt->transport_lock);
658		return;
659	}
660
661	/* Note: at this point, task->tk_sleeping has not yet been set,
662	 *	 hence there is no danger of the waking up task being put on
663	 *	 schedq, and being picked up by a parallel run of rpciod().
664	 */
665	task->tk_status = status;
666
667	switch (status) {
668	case -ECONNREFUSED:
669		task->tk_timeout = RPC_REESTABLISH_TIMEOUT;
670		rpc_sleep_on(&xprt->sending, task, NULL, NULL);
671	case -EAGAIN:
672	case -ENOTCONN:
673		return;
674	default:
675		break;
676	}
677	xprt_release_write(xprt, task);
678	return;
679}
680
681static inline void do_xprt_reserve(struct rpc_task *task)
682{
683	struct rpc_xprt	*xprt = task->tk_xprt;
684
685	task->tk_status = 0;
686	if (task->tk_rqstp)
687		return;
688	if (!list_empty(&xprt->free)) {
689		struct rpc_rqst	*req = list_entry(xprt->free.next, struct rpc_rqst, rq_list);
690		list_del_init(&req->rq_list);
691		task->tk_rqstp = req;
692		xprt_request_init(task, xprt);
693		return;
694	}
695	dprintk("RPC:      waiting for request slot\n");
696	task->tk_status = -EAGAIN;
697	task->tk_timeout = 0;
698	rpc_sleep_on(&xprt->backlog, task, NULL, NULL);
699}
700
701/**
702 * xprt_reserve - allocate an RPC request slot
703 * @task: RPC task requesting a slot allocation
704 *
705 * If no more slots are available, place the task on the transport's
706 * backlog queue.
707 */
708void xprt_reserve(struct rpc_task *task)
709{
710	struct rpc_xprt	*xprt = task->tk_xprt;
711
712	task->tk_status = -EIO;
713	if (!xprt->shutdown) {
714		spin_lock(&xprt->reserve_lock);
715		do_xprt_reserve(task);
716		spin_unlock(&xprt->reserve_lock);
717	}
718}
719
720static inline u32 xprt_alloc_xid(struct rpc_xprt *xprt)
721{
722	return xprt->xid++;
723}
724
725static inline void xprt_init_xid(struct rpc_xprt *xprt)
726{
727	get_random_bytes(&xprt->xid, sizeof(xprt->xid));
728}
729
730static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt)
731{
732	struct rpc_rqst	*req = task->tk_rqstp;
733
734	req->rq_timeout = xprt->timeout.to_initval;
735	req->rq_task	= task;
736	req->rq_xprt    = xprt;
737	req->rq_xid     = xprt_alloc_xid(xprt);
738	dprintk("RPC: %4d reserved req %p xid %08x\n", task->tk_pid,
739			req, ntohl(req->rq_xid));
740}
741
742/**
743 * xprt_release - release an RPC request slot
744 * @task: task which is finished with the slot
745 *
746 */
747void xprt_release(struct rpc_task *task)
748{
749	struct rpc_xprt	*xprt = task->tk_xprt;
750	struct rpc_rqst	*req;
751
752	if (!(req = task->tk_rqstp))
753		return;
754	spin_lock_bh(&xprt->transport_lock);
755	__xprt_release_write(xprt, task);
756	__xprt_put_cong(xprt, req);
757	if (!list_empty(&req->rq_list))
758		list_del(&req->rq_list);
759	xprt->last_used = jiffies;
760	if (list_empty(&xprt->recv) && !xprt->shutdown)
761		mod_timer(&xprt->timer,
762				xprt->last_used + RPC_IDLE_DISCONNECT_TIMEOUT);
763	spin_unlock_bh(&xprt->transport_lock);
764	task->tk_rqstp = NULL;
765	memset(req, 0, sizeof(*req));	/* mark unused */
766
767	dprintk("RPC: %4d release request %p\n", task->tk_pid, req);
768
769	spin_lock(&xprt->reserve_lock);
770	list_add(&req->rq_list, &xprt->free);
771	xprt_clear_backlog(xprt);
772	spin_unlock(&xprt->reserve_lock);
773}
774
775/**
776 * xprt_set_timeout - set constant RPC timeout
777 * @to: RPC timeout parameters to set up
778 * @retr: number of retries
779 * @incr: amount of increase after each retry
780 *
781 */
782void xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long incr)
783{
784	to->to_initval   =
785	to->to_increment = incr;
786	to->to_maxval    = to->to_initval + (incr * retr);
787	to->to_retries   = retr;
788	to->to_exponential = 0;
789}
790
791static struct rpc_xprt *xprt_setup(int proto, struct sockaddr_in *ap, struct rpc_timeout *to)
792{
793	int result;
794	struct rpc_xprt	*xprt;
795	struct rpc_rqst	*req;
796
797	if ((xprt = kmalloc(sizeof(struct rpc_xprt), GFP_KERNEL)) == NULL)
798		return ERR_PTR(-ENOMEM);
799	memset(xprt, 0, sizeof(*xprt)); /* Nnnngh! */
800
801	xprt->addr = *ap;
802
803	switch (proto) {
804	case IPPROTO_UDP:
805		result = xs_setup_udp(xprt, to);
806		break;
807	case IPPROTO_TCP:
808		result = xs_setup_tcp(xprt, to);
809		break;
810	default:
811		printk(KERN_ERR "RPC: unrecognized transport protocol: %d\n",
812				proto);
813		result = -EIO;
814		break;
815	}
816	if (result) {
817		kfree(xprt);
818		return ERR_PTR(result);
819	}
820
821	spin_lock_init(&xprt->transport_lock);
822	spin_lock_init(&xprt->reserve_lock);
823	init_waitqueue_head(&xprt->cong_wait);
824
825	INIT_LIST_HEAD(&xprt->free);
826	INIT_LIST_HEAD(&xprt->recv);
827	INIT_WORK(&xprt->task_cleanup, xprt_autoclose, xprt);
828	init_timer(&xprt->timer);
829	xprt->timer.function = xprt_init_autodisconnect;
830	xprt->timer.data = (unsigned long) xprt;
831	xprt->last_used = jiffies;
832
833	rpc_init_wait_queue(&xprt->pending, "xprt_pending");
834	rpc_init_wait_queue(&xprt->sending, "xprt_sending");
835	rpc_init_wait_queue(&xprt->resend, "xprt_resend");
836	rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog");
837
838	/* initialize free list */
839	for (req = &xprt->slot[xprt->max_reqs-1]; req >= &xprt->slot[0]; req--)
840		list_add(&req->rq_list, &xprt->free);
841
842	xprt_init_xid(xprt);
843
844	dprintk("RPC:      created transport %p with %u slots\n", xprt,
845			xprt->max_reqs);
846
847	return xprt;
848}
849
850/**
851 * xprt_create_proto - create an RPC client transport
852 * @proto: requested transport protocol
853 * @sap: remote peer's address
854 * @to: timeout parameters for new transport
855 *
856 */
857struct rpc_xprt *xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to)
858{
859	struct rpc_xprt	*xprt;
860
861	xprt = xprt_setup(proto, sap, to);
862	if (IS_ERR(xprt))
863		dprintk("RPC:      xprt_create_proto failed\n");
864	else
865		dprintk("RPC:      xprt_create_proto created xprt %p\n", xprt);
866	return xprt;
867}
868
869static void xprt_shutdown(struct rpc_xprt *xprt)
870{
871	xprt->shutdown = 1;
872	rpc_wake_up(&xprt->sending);
873	rpc_wake_up(&xprt->resend);
874	xprt_wake_pending_tasks(xprt, -EIO);
875	rpc_wake_up(&xprt->backlog);
876	wake_up(&xprt->cong_wait);
877	del_timer_sync(&xprt->timer);
878}
879
880static int xprt_clear_backlog(struct rpc_xprt *xprt) {
881	rpc_wake_up_next(&xprt->backlog);
882	wake_up(&xprt->cong_wait);
883	return 1;
884}
885
886/**
887 * xprt_destroy - destroy an RPC transport, killing off all requests.
888 * @xprt: transport to destroy
889 *
890 */
891int xprt_destroy(struct rpc_xprt *xprt)
892{
893	dprintk("RPC:      destroying transport %p\n", xprt);
894	xprt_shutdown(xprt);
895	xprt->ops->destroy(xprt);
896	kfree(xprt);
897
898	return 0;
899}
900