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 | static int xprt_request_prepare(struct rpc_rqst *req, struct xdr_buf *buf); |
77 | |
78 | static DEFINE_SPINLOCK(xprt_list_lock); |
79 | static LIST_HEAD(xprt_list); |
80 | |
81 | static unsigned long xprt_request_timeout(const struct rpc_rqst *req) |
82 | { |
83 | unsigned long timeout = jiffies + req->rq_timeout; |
84 | |
85 | if (time_before(timeout, req->rq_majortimeo)) |
86 | return timeout; |
87 | return req->rq_majortimeo; |
88 | } |
89 | |
90 | /** |
91 | * xprt_register_transport - register a transport implementation |
92 | * @transport: transport to register |
93 | * |
94 | * If a transport implementation is loaded as a kernel module, it can |
95 | * call this interface to make itself known to the RPC client. |
96 | * |
97 | * Returns: |
98 | * 0: transport successfully registered |
99 | * -EEXIST: transport already registered |
100 | * -EINVAL: transport module being unloaded |
101 | */ |
102 | int xprt_register_transport(struct xprt_class *transport) |
103 | { |
104 | struct xprt_class *t; |
105 | int result; |
106 | |
107 | result = -EEXIST; |
108 | spin_lock(lock: &xprt_list_lock); |
109 | list_for_each_entry(t, &xprt_list, list) { |
110 | /* don't register the same transport class twice */ |
111 | if (t->ident == transport->ident) |
112 | goto out; |
113 | } |
114 | |
115 | list_add_tail(new: &transport->list, head: &xprt_list); |
116 | printk(KERN_INFO "RPC: Registered %s transport module.\n" , |
117 | transport->name); |
118 | result = 0; |
119 | |
120 | out: |
121 | spin_unlock(lock: &xprt_list_lock); |
122 | return result; |
123 | } |
124 | EXPORT_SYMBOL_GPL(xprt_register_transport); |
125 | |
126 | /** |
127 | * xprt_unregister_transport - unregister a transport implementation |
128 | * @transport: transport to unregister |
129 | * |
130 | * Returns: |
131 | * 0: transport successfully unregistered |
132 | * -ENOENT: transport never registered |
133 | */ |
134 | int xprt_unregister_transport(struct xprt_class *transport) |
135 | { |
136 | struct xprt_class *t; |
137 | int result; |
138 | |
139 | result = 0; |
140 | spin_lock(lock: &xprt_list_lock); |
141 | list_for_each_entry(t, &xprt_list, list) { |
142 | if (t == transport) { |
143 | printk(KERN_INFO |
144 | "RPC: Unregistered %s transport module.\n" , |
145 | transport->name); |
146 | list_del_init(entry: &transport->list); |
147 | goto out; |
148 | } |
149 | } |
150 | result = -ENOENT; |
151 | |
152 | out: |
153 | spin_unlock(lock: &xprt_list_lock); |
154 | return result; |
155 | } |
156 | EXPORT_SYMBOL_GPL(xprt_unregister_transport); |
157 | |
158 | static void |
159 | xprt_class_release(const struct xprt_class *t) |
160 | { |
161 | module_put(module: t->owner); |
162 | } |
163 | |
164 | static const struct xprt_class * |
165 | xprt_class_find_by_ident_locked(int ident) |
166 | { |
167 | const struct xprt_class *t; |
168 | |
169 | list_for_each_entry(t, &xprt_list, list) { |
170 | if (t->ident != ident) |
171 | continue; |
172 | if (!try_module_get(module: t->owner)) |
173 | continue; |
174 | return t; |
175 | } |
176 | return NULL; |
177 | } |
178 | |
179 | static const struct xprt_class * |
180 | xprt_class_find_by_ident(int ident) |
181 | { |
182 | const struct xprt_class *t; |
183 | |
184 | spin_lock(lock: &xprt_list_lock); |
185 | t = xprt_class_find_by_ident_locked(ident); |
186 | spin_unlock(lock: &xprt_list_lock); |
187 | return t; |
188 | } |
189 | |
190 | static const struct xprt_class * |
191 | xprt_class_find_by_netid_locked(const char *netid) |
192 | { |
193 | const struct xprt_class *t; |
194 | unsigned int i; |
195 | |
196 | list_for_each_entry(t, &xprt_list, list) { |
197 | for (i = 0; t->netid[i][0] != '\0'; i++) { |
198 | if (strcmp(t->netid[i], netid) != 0) |
199 | continue; |
200 | if (!try_module_get(module: t->owner)) |
201 | continue; |
202 | return t; |
203 | } |
204 | } |
205 | return NULL; |
206 | } |
207 | |
208 | static const struct xprt_class * |
209 | xprt_class_find_by_netid(const char *netid) |
210 | { |
211 | const struct xprt_class *t; |
212 | |
213 | spin_lock(lock: &xprt_list_lock); |
214 | t = xprt_class_find_by_netid_locked(netid); |
215 | if (!t) { |
216 | spin_unlock(lock: &xprt_list_lock); |
217 | request_module("rpc%s" , netid); |
218 | spin_lock(lock: &xprt_list_lock); |
219 | t = xprt_class_find_by_netid_locked(netid); |
220 | } |
221 | spin_unlock(lock: &xprt_list_lock); |
222 | return t; |
223 | } |
224 | |
225 | /** |
226 | * xprt_find_transport_ident - convert a netid into a transport identifier |
227 | * @netid: transport to load |
228 | * |
229 | * Returns: |
230 | * > 0: transport identifier |
231 | * -ENOENT: transport module not available |
232 | */ |
233 | int xprt_find_transport_ident(const char *netid) |
234 | { |
235 | const struct xprt_class *t; |
236 | int ret; |
237 | |
238 | t = xprt_class_find_by_netid(netid); |
239 | if (!t) |
240 | return -ENOENT; |
241 | ret = t->ident; |
242 | xprt_class_release(t); |
243 | return ret; |
244 | } |
245 | EXPORT_SYMBOL_GPL(xprt_find_transport_ident); |
246 | |
247 | static void xprt_clear_locked(struct rpc_xprt *xprt) |
248 | { |
249 | xprt->snd_task = NULL; |
250 | if (!test_bit(XPRT_CLOSE_WAIT, &xprt->state)) |
251 | clear_bit_unlock(XPRT_LOCKED, addr: &xprt->state); |
252 | else |
253 | queue_work(wq: xprtiod_workqueue, work: &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, addr: &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) || RPC_IS_SOFTCONN(task)) |
287 | rpc_sleep_on_timeout(queue: &xprt->sending, task, NULL, |
288 | timeout: 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(head: &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, addr: &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, addr: &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, addr: &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) || RPC_IS_SOFTCONN(task)) |
353 | rpc_sleep_on_timeout(queue: &xprt->sending, task, NULL, |
354 | timeout: 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(lock: &xprt->transport_lock); |
371 | retval = xprt->ops->reserve_xprt(xprt, task); |
372 | spin_unlock(lock: &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, addr: &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(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, addr: &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(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(lock: &xprt->transport_lock); |
452 | xprt->ops->release_xprt(xprt, task); |
453 | spin_unlock(lock: &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, task: 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, task: 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(lock: &xprt->transport_lock); |
506 | ret = __xprt_get_cong(xprt, req) != 0; |
507 | spin_unlock(lock: &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(xprt: 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, addr: &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, addr: &xprt->state)) { |
540 | spin_lock(lock: &xprt->transport_lock); |
541 | __xprt_lock_write_next_cong(xprt); |
542 | spin_unlock(lock: &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, addr: &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, addr: &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(lock: &xprt->transport_lock); |
640 | ret = xprt_clear_write_space_locked(xprt); |
641 | spin_unlock(lock: &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(kt: ktime_get() - abstime); |
649 | return likely(delta >= 0) ? |
650 | jiffies - nsecs_to_jiffies(n: delta) : |
651 | jiffies + nsecs_to_jiffies(n: -delta); |
652 | } |
653 | |
654 | static unsigned long xprt_calc_majortimeo(struct rpc_rqst *req, |
655 | const struct rpc_timeout *to) |
656 | { |
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 | const struct rpc_timeout *to) |
670 | { |
671 | req->rq_majortimeo += xprt_calc_majortimeo(req, to); |
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 | const struct rpc_timeout *to) |
681 | { |
682 | unsigned long time_init; |
683 | struct rpc_xprt *xprt = req->rq_xprt; |
684 | |
685 | if (likely(xprt && xprt_connected(xprt))) |
686 | time_init = jiffies; |
687 | else |
688 | time_init = xprt_abs_ktime_to_jiffies(abstime: task->tk_start); |
689 | |
690 | req->rq_timeout = to->to_initval; |
691 | req->rq_majortimeo = time_init + xprt_calc_majortimeo(req, to); |
692 | req->rq_minortimeo = time_init + req->rq_timeout; |
693 | } |
694 | |
695 | /** |
696 | * xprt_adjust_timeout - adjust timeout values for next retransmit |
697 | * @req: RPC request containing parameters to use for the adjustment |
698 | * |
699 | */ |
700 | int xprt_adjust_timeout(struct rpc_rqst *req) |
701 | { |
702 | struct rpc_xprt *xprt = req->rq_xprt; |
703 | const struct rpc_timeout *to = req->rq_task->tk_client->cl_timeout; |
704 | int status = 0; |
705 | |
706 | if (time_before(jiffies, req->rq_majortimeo)) { |
707 | if (time_before(jiffies, req->rq_minortimeo)) |
708 | return status; |
709 | if (to->to_exponential) |
710 | req->rq_timeout <<= 1; |
711 | else |
712 | req->rq_timeout += to->to_increment; |
713 | if (to->to_maxval && req->rq_timeout >= to->to_maxval) |
714 | req->rq_timeout = to->to_maxval; |
715 | req->rq_retries++; |
716 | } else { |
717 | req->rq_timeout = to->to_initval; |
718 | req->rq_retries = 0; |
719 | xprt_reset_majortimeo(req, to); |
720 | /* Reset the RTT counters == "slow start" */ |
721 | spin_lock(lock: &xprt->transport_lock); |
722 | rpc_init_rtt(rt: req->rq_task->tk_client->cl_rtt, timeo: to->to_initval); |
723 | spin_unlock(lock: &xprt->transport_lock); |
724 | status = -ETIMEDOUT; |
725 | } |
726 | xprt_reset_minortimeo(req); |
727 | |
728 | if (req->rq_timeout == 0) { |
729 | printk(KERN_WARNING "xprt_adjust_timeout: rq_timeout = 0!\n" ); |
730 | req->rq_timeout = 5 * HZ; |
731 | } |
732 | return status; |
733 | } |
734 | |
735 | static void xprt_autoclose(struct work_struct *work) |
736 | { |
737 | struct rpc_xprt *xprt = |
738 | container_of(work, struct rpc_xprt, task_cleanup); |
739 | unsigned int pflags = memalloc_nofs_save(); |
740 | |
741 | trace_xprt_disconnect_auto(xprt); |
742 | xprt->connect_cookie++; |
743 | smp_mb__before_atomic(); |
744 | clear_bit(XPRT_CLOSE_WAIT, addr: &xprt->state); |
745 | xprt->ops->close(xprt); |
746 | xprt_release_write(xprt, NULL); |
747 | wake_up_bit(word: &xprt->state, XPRT_LOCKED); |
748 | memalloc_nofs_restore(flags: pflags); |
749 | } |
750 | |
751 | /** |
752 | * xprt_disconnect_done - mark a transport as disconnected |
753 | * @xprt: transport to flag for disconnect |
754 | * |
755 | */ |
756 | void xprt_disconnect_done(struct rpc_xprt *xprt) |
757 | { |
758 | trace_xprt_disconnect_done(xprt); |
759 | spin_lock(lock: &xprt->transport_lock); |
760 | xprt_clear_connected(xprt); |
761 | xprt_clear_write_space_locked(xprt); |
762 | xprt_clear_congestion_window_wait_locked(xprt); |
763 | xprt_wake_pending_tasks(xprt, -ENOTCONN); |
764 | spin_unlock(lock: &xprt->transport_lock); |
765 | } |
766 | EXPORT_SYMBOL_GPL(xprt_disconnect_done); |
767 | |
768 | /** |
769 | * xprt_schedule_autoclose_locked - Try to schedule an autoclose RPC call |
770 | * @xprt: transport to disconnect |
771 | */ |
772 | static void xprt_schedule_autoclose_locked(struct rpc_xprt *xprt) |
773 | { |
774 | if (test_and_set_bit(XPRT_CLOSE_WAIT, addr: &xprt->state)) |
775 | return; |
776 | if (test_and_set_bit(XPRT_LOCKED, addr: &xprt->state) == 0) |
777 | queue_work(wq: xprtiod_workqueue, work: &xprt->task_cleanup); |
778 | else if (xprt->snd_task && !test_bit(XPRT_SND_IS_COOKIE, &xprt->state)) |
779 | rpc_wake_up_queued_task_set_status(&xprt->pending, |
780 | xprt->snd_task, -ENOTCONN); |
781 | } |
782 | |
783 | /** |
784 | * xprt_force_disconnect - force a transport to disconnect |
785 | * @xprt: transport to disconnect |
786 | * |
787 | */ |
788 | void xprt_force_disconnect(struct rpc_xprt *xprt) |
789 | { |
790 | trace_xprt_disconnect_force(xprt); |
791 | |
792 | /* Don't race with the test_bit() in xprt_clear_locked() */ |
793 | spin_lock(lock: &xprt->transport_lock); |
794 | xprt_schedule_autoclose_locked(xprt); |
795 | spin_unlock(lock: &xprt->transport_lock); |
796 | } |
797 | EXPORT_SYMBOL_GPL(xprt_force_disconnect); |
798 | |
799 | static unsigned int |
800 | xprt_connect_cookie(struct rpc_xprt *xprt) |
801 | { |
802 | return READ_ONCE(xprt->connect_cookie); |
803 | } |
804 | |
805 | static bool |
806 | xprt_request_retransmit_after_disconnect(struct rpc_task *task) |
807 | { |
808 | struct rpc_rqst *req = task->tk_rqstp; |
809 | struct rpc_xprt *xprt = req->rq_xprt; |
810 | |
811 | return req->rq_connect_cookie != xprt_connect_cookie(xprt) || |
812 | !xprt_connected(xprt); |
813 | } |
814 | |
815 | /** |
816 | * xprt_conditional_disconnect - force a transport to disconnect |
817 | * @xprt: transport to disconnect |
818 | * @cookie: 'connection cookie' |
819 | * |
820 | * This attempts to break the connection if and only if 'cookie' matches |
821 | * the current transport 'connection cookie'. It ensures that we don't |
822 | * try to break the connection more than once when we need to retransmit |
823 | * a batch of RPC requests. |
824 | * |
825 | */ |
826 | void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie) |
827 | { |
828 | /* Don't race with the test_bit() in xprt_clear_locked() */ |
829 | spin_lock(lock: &xprt->transport_lock); |
830 | if (cookie != xprt->connect_cookie) |
831 | goto out; |
832 | if (test_bit(XPRT_CLOSING, &xprt->state)) |
833 | goto out; |
834 | xprt_schedule_autoclose_locked(xprt); |
835 | out: |
836 | spin_unlock(lock: &xprt->transport_lock); |
837 | } |
838 | |
839 | static bool |
840 | xprt_has_timer(const struct rpc_xprt *xprt) |
841 | { |
842 | return xprt->idle_timeout != 0; |
843 | } |
844 | |
845 | static void |
846 | xprt_schedule_autodisconnect(struct rpc_xprt *xprt) |
847 | __must_hold(&xprt->transport_lock) |
848 | { |
849 | xprt->last_used = jiffies; |
850 | if (RB_EMPTY_ROOT(&xprt->recv_queue) && xprt_has_timer(xprt)) |
851 | mod_timer(timer: &xprt->timer, expires: xprt->last_used + xprt->idle_timeout); |
852 | } |
853 | |
854 | static void |
855 | xprt_init_autodisconnect(struct timer_list *t) |
856 | { |
857 | struct rpc_xprt *xprt = from_timer(xprt, t, timer); |
858 | |
859 | if (!RB_EMPTY_ROOT(&xprt->recv_queue)) |
860 | return; |
861 | /* Reset xprt->last_used to avoid connect/autodisconnect cycling */ |
862 | xprt->last_used = jiffies; |
863 | if (test_and_set_bit(XPRT_LOCKED, addr: &xprt->state)) |
864 | return; |
865 | queue_work(wq: xprtiod_workqueue, work: &xprt->task_cleanup); |
866 | } |
867 | |
868 | #if IS_ENABLED(CONFIG_FAIL_SUNRPC) |
869 | static void xprt_inject_disconnect(struct rpc_xprt *xprt) |
870 | { |
871 | if (!fail_sunrpc.ignore_client_disconnect && |
872 | should_fail(attr: &fail_sunrpc.attr, size: 1)) |
873 | xprt->ops->inject_disconnect(xprt); |
874 | } |
875 | #else |
876 | static inline void xprt_inject_disconnect(struct rpc_xprt *xprt) |
877 | { |
878 | } |
879 | #endif |
880 | |
881 | bool xprt_lock_connect(struct rpc_xprt *xprt, |
882 | struct rpc_task *task, |
883 | void *cookie) |
884 | { |
885 | bool ret = false; |
886 | |
887 | spin_lock(lock: &xprt->transport_lock); |
888 | if (!test_bit(XPRT_LOCKED, &xprt->state)) |
889 | goto out; |
890 | if (xprt->snd_task != task) |
891 | goto out; |
892 | set_bit(XPRT_SND_IS_COOKIE, addr: &xprt->state); |
893 | xprt->snd_task = cookie; |
894 | ret = true; |
895 | out: |
896 | spin_unlock(lock: &xprt->transport_lock); |
897 | return ret; |
898 | } |
899 | EXPORT_SYMBOL_GPL(xprt_lock_connect); |
900 | |
901 | void xprt_unlock_connect(struct rpc_xprt *xprt, void *cookie) |
902 | { |
903 | spin_lock(lock: &xprt->transport_lock); |
904 | if (xprt->snd_task != cookie) |
905 | goto out; |
906 | if (!test_bit(XPRT_LOCKED, &xprt->state)) |
907 | goto out; |
908 | xprt->snd_task =NULL; |
909 | clear_bit(XPRT_SND_IS_COOKIE, addr: &xprt->state); |
910 | xprt->ops->release_xprt(xprt, NULL); |
911 | xprt_schedule_autodisconnect(xprt); |
912 | out: |
913 | spin_unlock(lock: &xprt->transport_lock); |
914 | wake_up_bit(word: &xprt->state, XPRT_LOCKED); |
915 | } |
916 | EXPORT_SYMBOL_GPL(xprt_unlock_connect); |
917 | |
918 | /** |
919 | * xprt_connect - schedule a transport connect operation |
920 | * @task: RPC task that is requesting the connect |
921 | * |
922 | */ |
923 | void xprt_connect(struct rpc_task *task) |
924 | { |
925 | struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; |
926 | |
927 | trace_xprt_connect(xprt); |
928 | |
929 | if (!xprt_bound(xprt)) { |
930 | task->tk_status = -EAGAIN; |
931 | return; |
932 | } |
933 | if (!xprt_lock_write(xprt, task)) |
934 | return; |
935 | |
936 | if (!xprt_connected(xprt) && !test_bit(XPRT_CLOSE_WAIT, &xprt->state)) { |
937 | task->tk_rqstp->rq_connect_cookie = xprt->connect_cookie; |
938 | rpc_sleep_on_timeout(queue: &xprt->pending, task, NULL, |
939 | timeout: xprt_request_timeout(req: 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(xid1: xid, xid2: 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(xid1: new->rq_xid, xid2: 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(node: &new->rq_recv, parent: n, rb_link: 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, status: 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, status: -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(v: &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(v: &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(v: &req->rq_pin); |
1114 | return; |
1115 | } |
1116 | if (atomic_dec_and_test(v: &req->rq_pin)) |
1117 | wake_up_var(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 | int |
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 | int ret; |
1151 | |
1152 | if (!xprt_request_need_enqueue_receive(task, req)) |
1153 | return 0; |
1154 | |
1155 | ret = xprt_request_prepare(req: task->tk_rqstp, buf: &req->rq_rcv_buf); |
1156 | if (ret) |
1157 | return ret; |
1158 | spin_lock(lock: &xprt->queue_lock); |
1159 | |
1160 | /* Update the softirq receive buffer */ |
1161 | memcpy(&req->rq_private_buf, &req->rq_rcv_buf, |
1162 | sizeof(req->rq_private_buf)); |
1163 | |
1164 | /* Add request to the receive list */ |
1165 | xprt_request_rb_insert(xprt, new: req); |
1166 | set_bit(RPC_TASK_NEED_RECV, addr: &task->tk_runstate); |
1167 | spin_unlock(lock: &xprt->queue_lock); |
1168 | |
1169 | /* Turn off autodisconnect */ |
1170 | del_timer_sync(timer: &xprt->timer); |
1171 | return 0; |
1172 | } |
1173 | |
1174 | /** |
1175 | * xprt_request_dequeue_receive_locked - Remove a request from the receive queue |
1176 | * @task: RPC task |
1177 | * |
1178 | * Caller must hold xprt->queue_lock. |
1179 | */ |
1180 | static void |
1181 | xprt_request_dequeue_receive_locked(struct rpc_task *task) |
1182 | { |
1183 | struct rpc_rqst *req = task->tk_rqstp; |
1184 | |
1185 | if (test_and_clear_bit(RPC_TASK_NEED_RECV, addr: &task->tk_runstate)) |
1186 | xprt_request_rb_remove(xprt: req->rq_xprt, req); |
1187 | } |
1188 | |
1189 | /** |
1190 | * xprt_update_rtt - Update RPC RTT statistics |
1191 | * @task: RPC request that recently completed |
1192 | * |
1193 | * Caller holds xprt->queue_lock. |
1194 | */ |
1195 | void xprt_update_rtt(struct rpc_task *task) |
1196 | { |
1197 | struct rpc_rqst *req = task->tk_rqstp; |
1198 | struct rpc_rtt *rtt = task->tk_client->cl_rtt; |
1199 | unsigned int timer = task->tk_msg.rpc_proc->p_timer; |
1200 | long m = usecs_to_jiffies(u: ktime_to_us(kt: req->rq_rtt)); |
1201 | |
1202 | if (timer) { |
1203 | if (req->rq_ntrans == 1) |
1204 | rpc_update_rtt(rt: rtt, timer, m); |
1205 | rpc_set_timeo(rt: rtt, timer, ntimeo: req->rq_ntrans - 1); |
1206 | } |
1207 | } |
1208 | EXPORT_SYMBOL_GPL(xprt_update_rtt); |
1209 | |
1210 | /** |
1211 | * xprt_complete_rqst - called when reply processing is complete |
1212 | * @task: RPC request that recently completed |
1213 | * @copied: actual number of bytes received from the transport |
1214 | * |
1215 | * Caller holds xprt->queue_lock. |
1216 | */ |
1217 | void xprt_complete_rqst(struct rpc_task *task, int copied) |
1218 | { |
1219 | struct rpc_rqst *req = task->tk_rqstp; |
1220 | struct rpc_xprt *xprt = req->rq_xprt; |
1221 | |
1222 | xprt->stat.recvs++; |
1223 | |
1224 | xdr_free_bvec(buf: &req->rq_rcv_buf); |
1225 | req->rq_private_buf.bvec = NULL; |
1226 | req->rq_private_buf.len = copied; |
1227 | /* Ensure all writes are done before we update */ |
1228 | /* req->rq_reply_bytes_recvd */ |
1229 | smp_wmb(); |
1230 | req->rq_reply_bytes_recvd = copied; |
1231 | xprt_request_dequeue_receive_locked(task); |
1232 | rpc_wake_up_queued_task(&xprt->pending, task); |
1233 | } |
1234 | EXPORT_SYMBOL_GPL(xprt_complete_rqst); |
1235 | |
1236 | static void xprt_timer(struct rpc_task *task) |
1237 | { |
1238 | struct rpc_rqst *req = task->tk_rqstp; |
1239 | struct rpc_xprt *xprt = req->rq_xprt; |
1240 | |
1241 | if (task->tk_status != -ETIMEDOUT) |
1242 | return; |
1243 | |
1244 | trace_xprt_timer(xprt, xid: req->rq_xid, status: task->tk_status); |
1245 | if (!req->rq_reply_bytes_recvd) { |
1246 | if (xprt->ops->timer) |
1247 | xprt->ops->timer(xprt, task); |
1248 | } else |
1249 | task->tk_status = 0; |
1250 | } |
1251 | |
1252 | /** |
1253 | * xprt_wait_for_reply_request_def - wait for reply |
1254 | * @task: pointer to rpc_task |
1255 | * |
1256 | * Set a request's retransmit timeout based on the transport's |
1257 | * default timeout parameters. Used by transports that don't adjust |
1258 | * the retransmit timeout based on round-trip time estimation, |
1259 | * and put the task to sleep on the pending queue. |
1260 | */ |
1261 | void xprt_wait_for_reply_request_def(struct rpc_task *task) |
1262 | { |
1263 | struct rpc_rqst *req = task->tk_rqstp; |
1264 | |
1265 | rpc_sleep_on_timeout(queue: &req->rq_xprt->pending, task, action: xprt_timer, |
1266 | timeout: xprt_request_timeout(req)); |
1267 | } |
1268 | EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_def); |
1269 | |
1270 | /** |
1271 | * xprt_wait_for_reply_request_rtt - wait for reply using RTT estimator |
1272 | * @task: pointer to rpc_task |
1273 | * |
1274 | * Set a request's retransmit timeout using the RTT estimator, |
1275 | * and put the task to sleep on the pending queue. |
1276 | */ |
1277 | void xprt_wait_for_reply_request_rtt(struct rpc_task *task) |
1278 | { |
1279 | int timer = task->tk_msg.rpc_proc->p_timer; |
1280 | struct rpc_clnt *clnt = task->tk_client; |
1281 | struct rpc_rtt *rtt = clnt->cl_rtt; |
1282 | struct rpc_rqst *req = task->tk_rqstp; |
1283 | unsigned long max_timeout = clnt->cl_timeout->to_maxval; |
1284 | unsigned long timeout; |
1285 | |
1286 | timeout = rpc_calc_rto(rt: rtt, timer); |
1287 | timeout <<= rpc_ntimeo(rt: rtt, timer) + req->rq_retries; |
1288 | if (timeout > max_timeout || timeout == 0) |
1289 | timeout = max_timeout; |
1290 | rpc_sleep_on_timeout(queue: &req->rq_xprt->pending, task, action: xprt_timer, |
1291 | timeout: jiffies + timeout); |
1292 | } |
1293 | EXPORT_SYMBOL_GPL(xprt_wait_for_reply_request_rtt); |
1294 | |
1295 | /** |
1296 | * xprt_request_wait_receive - wait for the reply to an RPC request |
1297 | * @task: RPC task about to send a request |
1298 | * |
1299 | */ |
1300 | void xprt_request_wait_receive(struct rpc_task *task) |
1301 | { |
1302 | struct rpc_rqst *req = task->tk_rqstp; |
1303 | struct rpc_xprt *xprt = req->rq_xprt; |
1304 | |
1305 | if (!test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate)) |
1306 | return; |
1307 | /* |
1308 | * Sleep on the pending queue if we're expecting a reply. |
1309 | * The spinlock ensures atomicity between the test of |
1310 | * req->rq_reply_bytes_recvd, and the call to rpc_sleep_on(). |
1311 | */ |
1312 | spin_lock(lock: &xprt->queue_lock); |
1313 | if (test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate)) { |
1314 | xprt->ops->wait_for_reply_request(task); |
1315 | /* |
1316 | * Send an extra queue wakeup call if the |
1317 | * connection was dropped in case the call to |
1318 | * rpc_sleep_on() raced. |
1319 | */ |
1320 | if (xprt_request_retransmit_after_disconnect(task)) |
1321 | rpc_wake_up_queued_task_set_status(&xprt->pending, |
1322 | task, -ENOTCONN); |
1323 | } |
1324 | spin_unlock(lock: &xprt->queue_lock); |
1325 | } |
1326 | |
1327 | static bool |
1328 | xprt_request_need_enqueue_transmit(struct rpc_task *task, struct rpc_rqst *req) |
1329 | { |
1330 | return !test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate); |
1331 | } |
1332 | |
1333 | /** |
1334 | * xprt_request_enqueue_transmit - queue a task for transmission |
1335 | * @task: pointer to rpc_task |
1336 | * |
1337 | * Add a task to the transmission queue. |
1338 | */ |
1339 | void |
1340 | xprt_request_enqueue_transmit(struct rpc_task *task) |
1341 | { |
1342 | struct rpc_rqst *pos, *req = task->tk_rqstp; |
1343 | struct rpc_xprt *xprt = req->rq_xprt; |
1344 | int ret; |
1345 | |
1346 | if (xprt_request_need_enqueue_transmit(task, req)) { |
1347 | ret = xprt_request_prepare(req: task->tk_rqstp, buf: &req->rq_snd_buf); |
1348 | if (ret) { |
1349 | task->tk_status = ret; |
1350 | return; |
1351 | } |
1352 | req->rq_bytes_sent = 0; |
1353 | spin_lock(lock: &xprt->queue_lock); |
1354 | /* |
1355 | * Requests that carry congestion control credits are added |
1356 | * to the head of the list to avoid starvation issues. |
1357 | */ |
1358 | if (req->rq_cong) { |
1359 | xprt_clear_congestion_window_wait(xprt); |
1360 | list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) { |
1361 | if (pos->rq_cong) |
1362 | continue; |
1363 | /* Note: req is added _before_ pos */ |
1364 | list_add_tail(new: &req->rq_xmit, head: &pos->rq_xmit); |
1365 | INIT_LIST_HEAD(list: &req->rq_xmit2); |
1366 | goto out; |
1367 | } |
1368 | } else if (!req->rq_seqno) { |
1369 | list_for_each_entry(pos, &xprt->xmit_queue, rq_xmit) { |
1370 | if (pos->rq_task->tk_owner != task->tk_owner) |
1371 | continue; |
1372 | list_add_tail(new: &req->rq_xmit2, head: &pos->rq_xmit2); |
1373 | INIT_LIST_HEAD(list: &req->rq_xmit); |
1374 | goto out; |
1375 | } |
1376 | } |
1377 | list_add_tail(new: &req->rq_xmit, head: &xprt->xmit_queue); |
1378 | INIT_LIST_HEAD(list: &req->rq_xmit2); |
1379 | out: |
1380 | atomic_long_inc(v: &xprt->xmit_queuelen); |
1381 | set_bit(RPC_TASK_NEED_XMIT, addr: &task->tk_runstate); |
1382 | spin_unlock(lock: &xprt->queue_lock); |
1383 | } |
1384 | } |
1385 | |
1386 | /** |
1387 | * xprt_request_dequeue_transmit_locked - remove a task from the transmission queue |
1388 | * @task: pointer to rpc_task |
1389 | * |
1390 | * Remove a task from the transmission queue |
1391 | * Caller must hold xprt->queue_lock |
1392 | */ |
1393 | static void |
1394 | xprt_request_dequeue_transmit_locked(struct rpc_task *task) |
1395 | { |
1396 | struct rpc_rqst *req = task->tk_rqstp; |
1397 | |
1398 | if (!test_and_clear_bit(RPC_TASK_NEED_XMIT, addr: &task->tk_runstate)) |
1399 | return; |
1400 | if (!list_empty(head: &req->rq_xmit)) { |
1401 | struct rpc_xprt *xprt = req->rq_xprt; |
1402 | |
1403 | if (list_is_first(list: &req->rq_xmit, head: &xprt->xmit_queue) && |
1404 | xprt->ops->abort_send_request) |
1405 | xprt->ops->abort_send_request(req); |
1406 | |
1407 | list_del(entry: &req->rq_xmit); |
1408 | if (!list_empty(head: &req->rq_xmit2)) { |
1409 | struct rpc_rqst *next = list_first_entry(&req->rq_xmit2, |
1410 | struct rpc_rqst, rq_xmit2); |
1411 | list_del(entry: &req->rq_xmit2); |
1412 | list_add_tail(new: &next->rq_xmit, head: &next->rq_xprt->xmit_queue); |
1413 | } |
1414 | } else |
1415 | list_del(entry: &req->rq_xmit2); |
1416 | atomic_long_dec(v: &req->rq_xprt->xmit_queuelen); |
1417 | xdr_free_bvec(buf: &req->rq_snd_buf); |
1418 | } |
1419 | |
1420 | /** |
1421 | * xprt_request_dequeue_transmit - remove a task from the transmission queue |
1422 | * @task: pointer to rpc_task |
1423 | * |
1424 | * Remove a task from the transmission queue |
1425 | */ |
1426 | static void |
1427 | xprt_request_dequeue_transmit(struct rpc_task *task) |
1428 | { |
1429 | struct rpc_rqst *req = task->tk_rqstp; |
1430 | struct rpc_xprt *xprt = req->rq_xprt; |
1431 | |
1432 | spin_lock(lock: &xprt->queue_lock); |
1433 | xprt_request_dequeue_transmit_locked(task); |
1434 | spin_unlock(lock: &xprt->queue_lock); |
1435 | } |
1436 | |
1437 | /** |
1438 | * xprt_request_dequeue_xprt - remove a task from the transmit+receive queue |
1439 | * @task: pointer to rpc_task |
1440 | * |
1441 | * Remove a task from the transmit and receive queues, and ensure that |
1442 | * it is not pinned by the receive work item. |
1443 | */ |
1444 | void |
1445 | xprt_request_dequeue_xprt(struct rpc_task *task) |
1446 | { |
1447 | struct rpc_rqst *req = task->tk_rqstp; |
1448 | struct rpc_xprt *xprt = req->rq_xprt; |
1449 | |
1450 | if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate) || |
1451 | test_bit(RPC_TASK_NEED_RECV, &task->tk_runstate) || |
1452 | xprt_is_pinned_rqst(req)) { |
1453 | spin_lock(lock: &xprt->queue_lock); |
1454 | while (xprt_is_pinned_rqst(req)) { |
1455 | set_bit(RPC_TASK_MSG_PIN_WAIT, addr: &task->tk_runstate); |
1456 | spin_unlock(lock: &xprt->queue_lock); |
1457 | xprt_wait_on_pinned_rqst(req); |
1458 | spin_lock(lock: &xprt->queue_lock); |
1459 | clear_bit(RPC_TASK_MSG_PIN_WAIT, addr: &task->tk_runstate); |
1460 | } |
1461 | xprt_request_dequeue_transmit_locked(task); |
1462 | xprt_request_dequeue_receive_locked(task); |
1463 | spin_unlock(lock: &xprt->queue_lock); |
1464 | xdr_free_bvec(buf: &req->rq_rcv_buf); |
1465 | } |
1466 | } |
1467 | |
1468 | /** |
1469 | * xprt_request_prepare - prepare an encoded request for transport |
1470 | * @req: pointer to rpc_rqst |
1471 | * @buf: pointer to send/rcv xdr_buf |
1472 | * |
1473 | * Calls into the transport layer to do whatever is needed to prepare |
1474 | * the request for transmission or receive. |
1475 | * Returns error, or zero. |
1476 | */ |
1477 | static int |
1478 | xprt_request_prepare(struct rpc_rqst *req, struct xdr_buf *buf) |
1479 | { |
1480 | struct rpc_xprt *xprt = req->rq_xprt; |
1481 | |
1482 | if (xprt->ops->prepare_request) |
1483 | return xprt->ops->prepare_request(req, buf); |
1484 | return 0; |
1485 | } |
1486 | |
1487 | /** |
1488 | * xprt_request_need_retransmit - Test if a task needs retransmission |
1489 | * @task: pointer to rpc_task |
1490 | * |
1491 | * Test for whether a connection breakage requires the task to retransmit |
1492 | */ |
1493 | bool |
1494 | xprt_request_need_retransmit(struct rpc_task *task) |
1495 | { |
1496 | return xprt_request_retransmit_after_disconnect(task); |
1497 | } |
1498 | |
1499 | /** |
1500 | * xprt_prepare_transmit - reserve the transport before sending a request |
1501 | * @task: RPC task about to send a request |
1502 | * |
1503 | */ |
1504 | bool xprt_prepare_transmit(struct rpc_task *task) |
1505 | { |
1506 | struct rpc_rqst *req = task->tk_rqstp; |
1507 | struct rpc_xprt *xprt = req->rq_xprt; |
1508 | |
1509 | if (!xprt_lock_write(xprt, task)) { |
1510 | /* Race breaker: someone may have transmitted us */ |
1511 | if (!test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate)) |
1512 | rpc_wake_up_queued_task_set_status(&xprt->sending, |
1513 | task, 0); |
1514 | return false; |
1515 | |
1516 | } |
1517 | if (atomic_read(v: &xprt->swapper)) |
1518 | /* This will be clear in __rpc_execute */ |
1519 | current->flags |= PF_MEMALLOC; |
1520 | return true; |
1521 | } |
1522 | |
1523 | void xprt_end_transmit(struct rpc_task *task) |
1524 | { |
1525 | struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt; |
1526 | |
1527 | xprt_inject_disconnect(xprt); |
1528 | xprt_release_write(xprt, task); |
1529 | } |
1530 | |
1531 | /** |
1532 | * xprt_request_transmit - send an RPC request on a transport |
1533 | * @req: pointer to request to transmit |
1534 | * @snd_task: RPC task that owns the transport lock |
1535 | * |
1536 | * This performs the transmission of a single request. |
1537 | * Note that if the request is not the same as snd_task, then it |
1538 | * does need to be pinned. |
1539 | * Returns '0' on success. |
1540 | */ |
1541 | static int |
1542 | xprt_request_transmit(struct rpc_rqst *req, struct rpc_task *snd_task) |
1543 | { |
1544 | struct rpc_xprt *xprt = req->rq_xprt; |
1545 | struct rpc_task *task = req->rq_task; |
1546 | unsigned int connect_cookie; |
1547 | int is_retrans = RPC_WAS_SENT(task); |
1548 | int status; |
1549 | |
1550 | if (test_bit(XPRT_CLOSE_WAIT, &xprt->state)) |
1551 | return -ENOTCONN; |
1552 | |
1553 | if (!req->rq_bytes_sent) { |
1554 | if (xprt_request_data_received(task)) { |
1555 | status = 0; |
1556 | goto out_dequeue; |
1557 | } |
1558 | /* Verify that our message lies in the RPCSEC_GSS window */ |
1559 | if (rpcauth_xmit_need_reencode(task)) { |
1560 | status = -EBADMSG; |
1561 | goto out_dequeue; |
1562 | } |
1563 | if (RPC_SIGNALLED(task)) { |
1564 | status = -ERESTARTSYS; |
1565 | goto out_dequeue; |
1566 | } |
1567 | } |
1568 | |
1569 | /* |
1570 | * Update req->rq_ntrans before transmitting to avoid races with |
1571 | * xprt_update_rtt(), which needs to know that it is recording a |
1572 | * reply to the first transmission. |
1573 | */ |
1574 | req->rq_ntrans++; |
1575 | |
1576 | trace_rpc_xdr_sendto(task, xdr: &req->rq_snd_buf); |
1577 | connect_cookie = xprt->connect_cookie; |
1578 | status = xprt->ops->send_request(req); |
1579 | if (status != 0) { |
1580 | req->rq_ntrans--; |
1581 | trace_xprt_transmit(rqst: req, status); |
1582 | return status; |
1583 | } |
1584 | |
1585 | if (is_retrans) { |
1586 | task->tk_client->cl_stats->rpcretrans++; |
1587 | trace_xprt_retransmit(rqst: req); |
1588 | } |
1589 | |
1590 | xprt_inject_disconnect(xprt); |
1591 | |
1592 | task->tk_flags |= RPC_TASK_SENT; |
1593 | spin_lock(lock: &xprt->transport_lock); |
1594 | |
1595 | xprt->stat.sends++; |
1596 | xprt->stat.req_u += xprt->stat.sends - xprt->stat.recvs; |
1597 | xprt->stat.bklog_u += xprt->backlog.qlen; |
1598 | xprt->stat.sending_u += xprt->sending.qlen; |
1599 | xprt->stat.pending_u += xprt->pending.qlen; |
1600 | spin_unlock(lock: &xprt->transport_lock); |
1601 | |
1602 | req->rq_connect_cookie = connect_cookie; |
1603 | out_dequeue: |
1604 | trace_xprt_transmit(rqst: req, status); |
1605 | xprt_request_dequeue_transmit(task); |
1606 | rpc_wake_up_queued_task_set_status(&xprt->sending, task, status); |
1607 | return status; |
1608 | } |
1609 | |
1610 | /** |
1611 | * xprt_transmit - send an RPC request on a transport |
1612 | * @task: controlling RPC task |
1613 | * |
1614 | * Attempts to drain the transmit queue. On exit, either the transport |
1615 | * signalled an error that needs to be handled before transmission can |
1616 | * resume, or @task finished transmitting, and detected that it already |
1617 | * received a reply. |
1618 | */ |
1619 | void |
1620 | xprt_transmit(struct rpc_task *task) |
1621 | { |
1622 | struct rpc_rqst *next, *req = task->tk_rqstp; |
1623 | struct rpc_xprt *xprt = req->rq_xprt; |
1624 | int status; |
1625 | |
1626 | spin_lock(lock: &xprt->queue_lock); |
1627 | for (;;) { |
1628 | next = list_first_entry_or_null(&xprt->xmit_queue, |
1629 | struct rpc_rqst, rq_xmit); |
1630 | if (!next) |
1631 | break; |
1632 | xprt_pin_rqst(next); |
1633 | spin_unlock(lock: &xprt->queue_lock); |
1634 | status = xprt_request_transmit(req: next, snd_task: task); |
1635 | if (status == -EBADMSG && next != req) |
1636 | status = 0; |
1637 | spin_lock(lock: &xprt->queue_lock); |
1638 | xprt_unpin_rqst(next); |
1639 | if (status < 0) { |
1640 | if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate)) |
1641 | task->tk_status = status; |
1642 | break; |
1643 | } |
1644 | /* Was @task transmitted, and has it received a reply? */ |
1645 | if (xprt_request_data_received(task) && |
1646 | !test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate)) |
1647 | break; |
1648 | cond_resched_lock(&xprt->queue_lock); |
1649 | } |
1650 | spin_unlock(lock: &xprt->queue_lock); |
1651 | } |
1652 | |
1653 | static void xprt_complete_request_init(struct rpc_task *task) |
1654 | { |
1655 | if (task->tk_rqstp) |
1656 | xprt_request_init(task); |
1657 | } |
1658 | |
1659 | void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task) |
1660 | { |
1661 | set_bit(XPRT_CONGESTED, addr: &xprt->state); |
1662 | rpc_sleep_on(&xprt->backlog, task, action: xprt_complete_request_init); |
1663 | } |
1664 | EXPORT_SYMBOL_GPL(xprt_add_backlog); |
1665 | |
1666 | static bool __xprt_set_rq(struct rpc_task *task, void *data) |
1667 | { |
1668 | struct rpc_rqst *req = data; |
1669 | |
1670 | if (task->tk_rqstp == NULL) { |
1671 | memset(req, 0, sizeof(*req)); /* mark unused */ |
1672 | task->tk_rqstp = req; |
1673 | return true; |
1674 | } |
1675 | return false; |
1676 | } |
1677 | |
1678 | bool xprt_wake_up_backlog(struct rpc_xprt *xprt, struct rpc_rqst *req) |
1679 | { |
1680 | if (rpc_wake_up_first(&xprt->backlog, __xprt_set_rq, req) == NULL) { |
1681 | clear_bit(XPRT_CONGESTED, addr: &xprt->state); |
1682 | return false; |
1683 | } |
1684 | return true; |
1685 | } |
1686 | EXPORT_SYMBOL_GPL(xprt_wake_up_backlog); |
1687 | |
1688 | static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task) |
1689 | { |
1690 | bool ret = false; |
1691 | |
1692 | if (!test_bit(XPRT_CONGESTED, &xprt->state)) |
1693 | goto out; |
1694 | spin_lock(lock: &xprt->reserve_lock); |
1695 | if (test_bit(XPRT_CONGESTED, &xprt->state)) { |
1696 | xprt_add_backlog(xprt, task); |
1697 | ret = true; |
1698 | } |
1699 | spin_unlock(lock: &xprt->reserve_lock); |
1700 | out: |
1701 | return ret; |
1702 | } |
1703 | |
1704 | static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt) |
1705 | { |
1706 | struct rpc_rqst *req = ERR_PTR(error: -EAGAIN); |
1707 | |
1708 | if (xprt->num_reqs >= xprt->max_reqs) |
1709 | goto out; |
1710 | ++xprt->num_reqs; |
1711 | spin_unlock(lock: &xprt->reserve_lock); |
1712 | req = kzalloc(size: sizeof(*req), flags: rpc_task_gfp_mask()); |
1713 | spin_lock(lock: &xprt->reserve_lock); |
1714 | if (req != NULL) |
1715 | goto out; |
1716 | --xprt->num_reqs; |
1717 | req = ERR_PTR(error: -ENOMEM); |
1718 | out: |
1719 | return req; |
1720 | } |
1721 | |
1722 | static bool xprt_dynamic_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req) |
1723 | { |
1724 | if (xprt->num_reqs > xprt->min_reqs) { |
1725 | --xprt->num_reqs; |
1726 | kfree(objp: req); |
1727 | return true; |
1728 | } |
1729 | return false; |
1730 | } |
1731 | |
1732 | void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task) |
1733 | { |
1734 | struct rpc_rqst *req; |
1735 | |
1736 | spin_lock(lock: &xprt->reserve_lock); |
1737 | if (!list_empty(head: &xprt->free)) { |
1738 | req = list_entry(xprt->free.next, struct rpc_rqst, rq_list); |
1739 | list_del(entry: &req->rq_list); |
1740 | goto out_init_req; |
1741 | } |
1742 | req = xprt_dynamic_alloc_slot(xprt); |
1743 | if (!IS_ERR(ptr: req)) |
1744 | goto out_init_req; |
1745 | switch (PTR_ERR(ptr: req)) { |
1746 | case -ENOMEM: |
1747 | dprintk("RPC: dynamic allocation of request slot " |
1748 | "failed! Retrying\n" ); |
1749 | task->tk_status = -ENOMEM; |
1750 | break; |
1751 | case -EAGAIN: |
1752 | xprt_add_backlog(xprt, task); |
1753 | dprintk("RPC: waiting for request slot\n" ); |
1754 | fallthrough; |
1755 | default: |
1756 | task->tk_status = -EAGAIN; |
1757 | } |
1758 | spin_unlock(lock: &xprt->reserve_lock); |
1759 | return; |
1760 | out_init_req: |
1761 | xprt->stat.max_slots = max_t(unsigned int, xprt->stat.max_slots, |
1762 | xprt->num_reqs); |
1763 | spin_unlock(lock: &xprt->reserve_lock); |
1764 | |
1765 | task->tk_status = 0; |
1766 | task->tk_rqstp = req; |
1767 | } |
1768 | EXPORT_SYMBOL_GPL(xprt_alloc_slot); |
1769 | |
1770 | void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req) |
1771 | { |
1772 | spin_lock(lock: &xprt->reserve_lock); |
1773 | if (!xprt_wake_up_backlog(xprt, req) && |
1774 | !xprt_dynamic_free_slot(xprt, req)) { |
1775 | memset(req, 0, sizeof(*req)); /* mark unused */ |
1776 | list_add(new: &req->rq_list, head: &xprt->free); |
1777 | } |
1778 | spin_unlock(lock: &xprt->reserve_lock); |
1779 | } |
1780 | EXPORT_SYMBOL_GPL(xprt_free_slot); |
1781 | |
1782 | static void xprt_free_all_slots(struct rpc_xprt *xprt) |
1783 | { |
1784 | struct rpc_rqst *req; |
1785 | while (!list_empty(head: &xprt->free)) { |
1786 | req = list_first_entry(&xprt->free, struct rpc_rqst, rq_list); |
1787 | list_del(entry: &req->rq_list); |
1788 | kfree(objp: req); |
1789 | } |
1790 | } |
1791 | |
1792 | static DEFINE_IDA(rpc_xprt_ids); |
1793 | |
1794 | void xprt_cleanup_ids(void) |
1795 | { |
1796 | ida_destroy(ida: &rpc_xprt_ids); |
1797 | } |
1798 | |
1799 | static int xprt_alloc_id(struct rpc_xprt *xprt) |
1800 | { |
1801 | int id; |
1802 | |
1803 | id = ida_alloc(ida: &rpc_xprt_ids, GFP_KERNEL); |
1804 | if (id < 0) |
1805 | return id; |
1806 | |
1807 | xprt->id = id; |
1808 | return 0; |
1809 | } |
1810 | |
1811 | static void xprt_free_id(struct rpc_xprt *xprt) |
1812 | { |
1813 | ida_free(&rpc_xprt_ids, id: xprt->id); |
1814 | } |
1815 | |
1816 | struct rpc_xprt *xprt_alloc(struct net *net, size_t size, |
1817 | unsigned int num_prealloc, |
1818 | unsigned int max_alloc) |
1819 | { |
1820 | struct rpc_xprt *xprt; |
1821 | struct rpc_rqst *req; |
1822 | int i; |
1823 | |
1824 | xprt = kzalloc(size, GFP_KERNEL); |
1825 | if (xprt == NULL) |
1826 | goto out; |
1827 | |
1828 | xprt_alloc_id(xprt); |
1829 | xprt_init(xprt, net); |
1830 | |
1831 | for (i = 0; i < num_prealloc; i++) { |
1832 | req = kzalloc(size: sizeof(struct rpc_rqst), GFP_KERNEL); |
1833 | if (!req) |
1834 | goto out_free; |
1835 | list_add(new: &req->rq_list, head: &xprt->free); |
1836 | } |
1837 | xprt->max_reqs = max_t(unsigned int, max_alloc, num_prealloc); |
1838 | xprt->min_reqs = num_prealloc; |
1839 | xprt->num_reqs = num_prealloc; |
1840 | |
1841 | return xprt; |
1842 | |
1843 | out_free: |
1844 | xprt_free(xprt); |
1845 | out: |
1846 | return NULL; |
1847 | } |
1848 | EXPORT_SYMBOL_GPL(xprt_alloc); |
1849 | |
1850 | void xprt_free(struct rpc_xprt *xprt) |
1851 | { |
1852 | put_net_track(net: xprt->xprt_net, tracker: &xprt->ns_tracker); |
1853 | xprt_free_all_slots(xprt); |
1854 | xprt_free_id(xprt); |
1855 | rpc_sysfs_xprt_destroy(xprt); |
1856 | kfree_rcu(xprt, rcu); |
1857 | } |
1858 | EXPORT_SYMBOL_GPL(xprt_free); |
1859 | |
1860 | static void |
1861 | xprt_init_connect_cookie(struct rpc_rqst *req, struct rpc_xprt *xprt) |
1862 | { |
1863 | req->rq_connect_cookie = xprt_connect_cookie(xprt) - 1; |
1864 | } |
1865 | |
1866 | static __be32 |
1867 | xprt_alloc_xid(struct rpc_xprt *xprt) |
1868 | { |
1869 | __be32 xid; |
1870 | |
1871 | spin_lock(lock: &xprt->reserve_lock); |
1872 | xid = (__force __be32)xprt->xid++; |
1873 | spin_unlock(lock: &xprt->reserve_lock); |
1874 | return xid; |
1875 | } |
1876 | |
1877 | static void |
1878 | xprt_init_xid(struct rpc_xprt *xprt) |
1879 | { |
1880 | xprt->xid = get_random_u32(); |
1881 | } |
1882 | |
1883 | static void |
1884 | xprt_request_init(struct rpc_task *task) |
1885 | { |
1886 | struct rpc_xprt *xprt = task->tk_xprt; |
1887 | struct rpc_rqst *req = task->tk_rqstp; |
1888 | |
1889 | req->rq_task = task; |
1890 | req->rq_xprt = xprt; |
1891 | req->rq_buffer = NULL; |
1892 | req->rq_xid = xprt_alloc_xid(xprt); |
1893 | xprt_init_connect_cookie(req, xprt); |
1894 | req->rq_snd_buf.len = 0; |
1895 | req->rq_snd_buf.buflen = 0; |
1896 | req->rq_rcv_buf.len = 0; |
1897 | req->rq_rcv_buf.buflen = 0; |
1898 | req->rq_snd_buf.bvec = NULL; |
1899 | req->rq_rcv_buf.bvec = NULL; |
1900 | req->rq_release_snd_buf = NULL; |
1901 | xprt_init_majortimeo(task, req, to: task->tk_client->cl_timeout); |
1902 | |
1903 | trace_xprt_reserve(rqst: req); |
1904 | } |
1905 | |
1906 | static void |
1907 | xprt_do_reserve(struct rpc_xprt *xprt, struct rpc_task *task) |
1908 | { |
1909 | xprt->ops->alloc_slot(xprt, task); |
1910 | if (task->tk_rqstp != NULL) |
1911 | xprt_request_init(task); |
1912 | } |
1913 | |
1914 | /** |
1915 | * xprt_reserve - allocate an RPC request slot |
1916 | * @task: RPC task requesting a slot allocation |
1917 | * |
1918 | * If the transport is marked as being congested, or if no more |
1919 | * slots are available, place the task on the transport's |
1920 | * backlog queue. |
1921 | */ |
1922 | void xprt_reserve(struct rpc_task *task) |
1923 | { |
1924 | struct rpc_xprt *xprt = task->tk_xprt; |
1925 | |
1926 | task->tk_status = 0; |
1927 | if (task->tk_rqstp != NULL) |
1928 | return; |
1929 | |
1930 | task->tk_status = -EAGAIN; |
1931 | if (!xprt_throttle_congested(xprt, task)) |
1932 | xprt_do_reserve(xprt, task); |
1933 | } |
1934 | |
1935 | /** |
1936 | * xprt_retry_reserve - allocate an RPC request slot |
1937 | * @task: RPC task requesting a slot allocation |
1938 | * |
1939 | * If no more slots are available, place the task on the transport's |
1940 | * backlog queue. |
1941 | * Note that the only difference with xprt_reserve is that we now |
1942 | * ignore the value of the XPRT_CONGESTED flag. |
1943 | */ |
1944 | void xprt_retry_reserve(struct rpc_task *task) |
1945 | { |
1946 | struct rpc_xprt *xprt = task->tk_xprt; |
1947 | |
1948 | task->tk_status = 0; |
1949 | if (task->tk_rqstp != NULL) |
1950 | return; |
1951 | |
1952 | task->tk_status = -EAGAIN; |
1953 | xprt_do_reserve(xprt, task); |
1954 | } |
1955 | |
1956 | /** |
1957 | * xprt_release - release an RPC request slot |
1958 | * @task: task which is finished with the slot |
1959 | * |
1960 | */ |
1961 | void xprt_release(struct rpc_task *task) |
1962 | { |
1963 | struct rpc_xprt *xprt; |
1964 | struct rpc_rqst *req = task->tk_rqstp; |
1965 | |
1966 | if (req == NULL) { |
1967 | if (task->tk_client) { |
1968 | xprt = task->tk_xprt; |
1969 | xprt_release_write(xprt, task); |
1970 | } |
1971 | return; |
1972 | } |
1973 | |
1974 | xprt = req->rq_xprt; |
1975 | xprt_request_dequeue_xprt(task); |
1976 | spin_lock(lock: &xprt->transport_lock); |
1977 | xprt->ops->release_xprt(xprt, task); |
1978 | if (xprt->ops->release_request) |
1979 | xprt->ops->release_request(task); |
1980 | xprt_schedule_autodisconnect(xprt); |
1981 | spin_unlock(lock: &xprt->transport_lock); |
1982 | if (req->rq_buffer) |
1983 | xprt->ops->buf_free(task); |
1984 | if (req->rq_cred != NULL) |
1985 | put_rpccred(req->rq_cred); |
1986 | if (req->rq_release_snd_buf) |
1987 | req->rq_release_snd_buf(req); |
1988 | |
1989 | task->tk_rqstp = NULL; |
1990 | if (likely(!bc_prealloc(req))) |
1991 | xprt->ops->free_slot(xprt, req); |
1992 | else |
1993 | xprt_free_bc_request(req); |
1994 | } |
1995 | |
1996 | #ifdef CONFIG_SUNRPC_BACKCHANNEL |
1997 | void |
1998 | xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task, |
1999 | const struct rpc_timeout *to) |
2000 | { |
2001 | struct xdr_buf *xbufp = &req->rq_snd_buf; |
2002 | |
2003 | task->tk_rqstp = req; |
2004 | req->rq_task = task; |
2005 | xprt_init_connect_cookie(req, xprt: req->rq_xprt); |
2006 | /* |
2007 | * Set up the xdr_buf length. |
2008 | * This also indicates that the buffer is XDR encoded already. |
2009 | */ |
2010 | xbufp->len = xbufp->head[0].iov_len + xbufp->page_len + |
2011 | xbufp->tail[0].iov_len; |
2012 | /* |
2013 | * Backchannel Replies are sent with !RPC_TASK_SOFT and |
2014 | * RPC_TASK_NO_RETRANS_TIMEOUT. The major timeout setting |
2015 | * affects only how long each Reply waits to be sent when |
2016 | * a transport connection cannot be established. |
2017 | */ |
2018 | xprt_init_majortimeo(task, req, to); |
2019 | } |
2020 | #endif |
2021 | |
2022 | static void xprt_init(struct rpc_xprt *xprt, struct net *net) |
2023 | { |
2024 | kref_init(kref: &xprt->kref); |
2025 | |
2026 | spin_lock_init(&xprt->transport_lock); |
2027 | spin_lock_init(&xprt->reserve_lock); |
2028 | spin_lock_init(&xprt->queue_lock); |
2029 | |
2030 | INIT_LIST_HEAD(list: &xprt->free); |
2031 | xprt->recv_queue = RB_ROOT; |
2032 | INIT_LIST_HEAD(list: &xprt->xmit_queue); |
2033 | #if defined(CONFIG_SUNRPC_BACKCHANNEL) |
2034 | spin_lock_init(&xprt->bc_pa_lock); |
2035 | INIT_LIST_HEAD(list: &xprt->bc_pa_list); |
2036 | #endif /* CONFIG_SUNRPC_BACKCHANNEL */ |
2037 | INIT_LIST_HEAD(list: &xprt->xprt_switch); |
2038 | |
2039 | xprt->last_used = jiffies; |
2040 | xprt->cwnd = RPC_INITCWND; |
2041 | xprt->bind_index = 0; |
2042 | |
2043 | rpc_init_wait_queue(&xprt->binding, "xprt_binding" ); |
2044 | rpc_init_wait_queue(&xprt->pending, "xprt_pending" ); |
2045 | rpc_init_wait_queue(&xprt->sending, "xprt_sending" ); |
2046 | rpc_init_priority_wait_queue(&xprt->backlog, "xprt_backlog" ); |
2047 | |
2048 | xprt_init_xid(xprt); |
2049 | |
2050 | xprt->xprt_net = get_net_track(net, tracker: &xprt->ns_tracker, GFP_KERNEL); |
2051 | } |
2052 | |
2053 | /** |
2054 | * xprt_create_transport - create an RPC transport |
2055 | * @args: rpc transport creation arguments |
2056 | * |
2057 | */ |
2058 | struct rpc_xprt *xprt_create_transport(struct xprt_create *args) |
2059 | { |
2060 | struct rpc_xprt *xprt; |
2061 | const struct xprt_class *t; |
2062 | |
2063 | t = xprt_class_find_by_ident(ident: args->ident); |
2064 | if (!t) { |
2065 | dprintk("RPC: transport (%d) not supported\n" , args->ident); |
2066 | return ERR_PTR(error: -EIO); |
2067 | } |
2068 | |
2069 | xprt = t->setup(args); |
2070 | xprt_class_release(t); |
2071 | |
2072 | if (IS_ERR(ptr: xprt)) |
2073 | goto out; |
2074 | if (args->flags & XPRT_CREATE_NO_IDLE_TIMEOUT) |
2075 | xprt->idle_timeout = 0; |
2076 | INIT_WORK(&xprt->task_cleanup, xprt_autoclose); |
2077 | if (xprt_has_timer(xprt)) |
2078 | timer_setup(&xprt->timer, xprt_init_autodisconnect, 0); |
2079 | else |
2080 | timer_setup(&xprt->timer, NULL, 0); |
2081 | |
2082 | if (strlen(args->servername) > RPC_MAXNETNAMELEN) { |
2083 | xprt_destroy(xprt); |
2084 | return ERR_PTR(error: -EINVAL); |
2085 | } |
2086 | xprt->servername = kstrdup(s: args->servername, GFP_KERNEL); |
2087 | if (xprt->servername == NULL) { |
2088 | xprt_destroy(xprt); |
2089 | return ERR_PTR(error: -ENOMEM); |
2090 | } |
2091 | |
2092 | rpc_xprt_debugfs_register(xprt); |
2093 | |
2094 | trace_xprt_create(xprt); |
2095 | out: |
2096 | return xprt; |
2097 | } |
2098 | |
2099 | static void xprt_destroy_cb(struct work_struct *work) |
2100 | { |
2101 | struct rpc_xprt *xprt = |
2102 | container_of(work, struct rpc_xprt, task_cleanup); |
2103 | |
2104 | trace_xprt_destroy(xprt); |
2105 | |
2106 | rpc_xprt_debugfs_unregister(xprt); |
2107 | rpc_destroy_wait_queue(&xprt->binding); |
2108 | rpc_destroy_wait_queue(&xprt->pending); |
2109 | rpc_destroy_wait_queue(&xprt->sending); |
2110 | rpc_destroy_wait_queue(&xprt->backlog); |
2111 | kfree(objp: xprt->servername); |
2112 | /* |
2113 | * Destroy any existing back channel |
2114 | */ |
2115 | xprt_destroy_backchannel(xprt, UINT_MAX); |
2116 | |
2117 | /* |
2118 | * Tear down transport state and free the rpc_xprt |
2119 | */ |
2120 | xprt->ops->destroy(xprt); |
2121 | } |
2122 | |
2123 | /** |
2124 | * xprt_destroy - destroy an RPC transport, killing off all requests. |
2125 | * @xprt: transport to destroy |
2126 | * |
2127 | */ |
2128 | static void xprt_destroy(struct rpc_xprt *xprt) |
2129 | { |
2130 | /* |
2131 | * Exclude transport connect/disconnect handlers and autoclose |
2132 | */ |
2133 | wait_on_bit_lock(word: &xprt->state, XPRT_LOCKED, TASK_UNINTERRUPTIBLE); |
2134 | |
2135 | /* |
2136 | * xprt_schedule_autodisconnect() can run after XPRT_LOCKED |
2137 | * is cleared. We use ->transport_lock to ensure the mod_timer() |
2138 | * can only run *before* del_time_sync(), never after. |
2139 | */ |
2140 | spin_lock(lock: &xprt->transport_lock); |
2141 | del_timer_sync(timer: &xprt->timer); |
2142 | spin_unlock(lock: &xprt->transport_lock); |
2143 | |
2144 | /* |
2145 | * Destroy sockets etc from the system workqueue so they can |
2146 | * safely flush receive work running on rpciod. |
2147 | */ |
2148 | INIT_WORK(&xprt->task_cleanup, xprt_destroy_cb); |
2149 | schedule_work(work: &xprt->task_cleanup); |
2150 | } |
2151 | |
2152 | static void xprt_destroy_kref(struct kref *kref) |
2153 | { |
2154 | xprt_destroy(container_of(kref, struct rpc_xprt, kref)); |
2155 | } |
2156 | |
2157 | /** |
2158 | * xprt_get - return a reference to an RPC transport. |
2159 | * @xprt: pointer to the transport |
2160 | * |
2161 | */ |
2162 | struct rpc_xprt *xprt_get(struct rpc_xprt *xprt) |
2163 | { |
2164 | if (xprt != NULL && kref_get_unless_zero(kref: &xprt->kref)) |
2165 | return xprt; |
2166 | return NULL; |
2167 | } |
2168 | EXPORT_SYMBOL_GPL(xprt_get); |
2169 | |
2170 | /** |
2171 | * xprt_put - release a reference to an RPC transport. |
2172 | * @xprt: pointer to the transport |
2173 | * |
2174 | */ |
2175 | void xprt_put(struct rpc_xprt *xprt) |
2176 | { |
2177 | if (xprt != NULL) |
2178 | kref_put(kref: &xprt->kref, release: xprt_destroy_kref); |
2179 | } |
2180 | EXPORT_SYMBOL_GPL(xprt_put); |
2181 | |
2182 | void xprt_set_offline_locked(struct rpc_xprt *xprt, struct rpc_xprt_switch *xps) |
2183 | { |
2184 | if (!test_and_set_bit(XPRT_OFFLINE, addr: &xprt->state)) { |
2185 | spin_lock(lock: &xps->xps_lock); |
2186 | xps->xps_nactive--; |
2187 | spin_unlock(lock: &xps->xps_lock); |
2188 | } |
2189 | } |
2190 | |
2191 | void xprt_set_online_locked(struct rpc_xprt *xprt, struct rpc_xprt_switch *xps) |
2192 | { |
2193 | if (test_and_clear_bit(XPRT_OFFLINE, addr: &xprt->state)) { |
2194 | spin_lock(lock: &xps->xps_lock); |
2195 | xps->xps_nactive++; |
2196 | spin_unlock(lock: &xps->xps_lock); |
2197 | } |
2198 | } |
2199 | |
2200 | void xprt_delete_locked(struct rpc_xprt *xprt, struct rpc_xprt_switch *xps) |
2201 | { |
2202 | if (test_and_set_bit(XPRT_REMOVE, addr: &xprt->state)) |
2203 | return; |
2204 | |
2205 | xprt_force_disconnect(xprt); |
2206 | if (!test_bit(XPRT_CONNECTED, &xprt->state)) |
2207 | return; |
2208 | |
2209 | if (!xprt->sending.qlen && !xprt->pending.qlen && |
2210 | !xprt->backlog.qlen && !atomic_long_read(v: &xprt->queuelen)) |
2211 | rpc_xprt_switch_remove_xprt(xps, xprt, offline: true); |
2212 | } |
2213 | |