]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - fs/afs/rxrpc.c
rxrpc: Use negative error codes in rxrpc_call struct
[mirror_ubuntu-artful-kernel.git] / fs / afs / rxrpc.c
1 /* Maintain an RxRPC server socket to do AFS communications through
2 *
3 * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
4 * Written by David Howells (dhowells@redhat.com)
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 */
11
12 #include <linux/slab.h>
13 #include <linux/sched/signal.h>
14
15 #include <net/sock.h>
16 #include <net/af_rxrpc.h>
17 #include <rxrpc/packet.h>
18 #include "internal.h"
19 #include "afs_cm.h"
20
21 struct socket *afs_socket; /* my RxRPC socket */
22 static struct workqueue_struct *afs_async_calls;
23 static struct afs_call *afs_spare_incoming_call;
24 atomic_t afs_outstanding_calls;
25
26 static void afs_wake_up_call_waiter(struct sock *, struct rxrpc_call *, unsigned long);
27 static int afs_wait_for_call_to_complete(struct afs_call *);
28 static void afs_wake_up_async_call(struct sock *, struct rxrpc_call *, unsigned long);
29 static void afs_process_async_call(struct work_struct *);
30 static void afs_rx_new_call(struct sock *, struct rxrpc_call *, unsigned long);
31 static void afs_rx_discard_new_call(struct rxrpc_call *, unsigned long);
32 static int afs_deliver_cm_op_id(struct afs_call *);
33
34 /* asynchronous incoming call initial processing */
35 static const struct afs_call_type afs_RXCMxxxx = {
36 .name = "CB.xxxx",
37 .deliver = afs_deliver_cm_op_id,
38 .abort_to_error = afs_abort_to_error,
39 };
40
41 static void afs_charge_preallocation(struct work_struct *);
42
43 static DECLARE_WORK(afs_charge_preallocation_work, afs_charge_preallocation);
44
45 static int afs_wait_atomic_t(atomic_t *p)
46 {
47 schedule();
48 return 0;
49 }
50
51 /*
52 * open an RxRPC socket and bind it to be a server for callback notifications
53 * - the socket is left in blocking mode and non-blocking ops use MSG_DONTWAIT
54 */
55 int afs_open_socket(void)
56 {
57 struct sockaddr_rxrpc srx;
58 struct socket *socket;
59 int ret;
60
61 _enter("");
62
63 ret = -ENOMEM;
64 afs_async_calls = alloc_workqueue("kafsd", WQ_MEM_RECLAIM, 0);
65 if (!afs_async_calls)
66 goto error_0;
67
68 ret = sock_create_kern(&init_net, AF_RXRPC, SOCK_DGRAM, PF_INET, &socket);
69 if (ret < 0)
70 goto error_1;
71
72 socket->sk->sk_allocation = GFP_NOFS;
73
74 /* bind the callback manager's address to make this a server socket */
75 srx.srx_family = AF_RXRPC;
76 srx.srx_service = CM_SERVICE;
77 srx.transport_type = SOCK_DGRAM;
78 srx.transport_len = sizeof(srx.transport.sin);
79 srx.transport.sin.sin_family = AF_INET;
80 srx.transport.sin.sin_port = htons(AFS_CM_PORT);
81 memset(&srx.transport.sin.sin_addr, 0,
82 sizeof(srx.transport.sin.sin_addr));
83
84 ret = kernel_bind(socket, (struct sockaddr *) &srx, sizeof(srx));
85 if (ret < 0)
86 goto error_2;
87
88 rxrpc_kernel_new_call_notification(socket, afs_rx_new_call,
89 afs_rx_discard_new_call);
90
91 ret = kernel_listen(socket, INT_MAX);
92 if (ret < 0)
93 goto error_2;
94
95 afs_socket = socket;
96 afs_charge_preallocation(NULL);
97 _leave(" = 0");
98 return 0;
99
100 error_2:
101 sock_release(socket);
102 error_1:
103 destroy_workqueue(afs_async_calls);
104 error_0:
105 _leave(" = %d", ret);
106 return ret;
107 }
108
109 /*
110 * close the RxRPC socket AFS was using
111 */
112 void afs_close_socket(void)
113 {
114 _enter("");
115
116 kernel_listen(afs_socket, 0);
117 flush_workqueue(afs_async_calls);
118
119 if (afs_spare_incoming_call) {
120 afs_put_call(afs_spare_incoming_call);
121 afs_spare_incoming_call = NULL;
122 }
123
124 _debug("outstanding %u", atomic_read(&afs_outstanding_calls));
125 wait_on_atomic_t(&afs_outstanding_calls, afs_wait_atomic_t,
126 TASK_UNINTERRUPTIBLE);
127 _debug("no outstanding calls");
128
129 kernel_sock_shutdown(afs_socket, SHUT_RDWR);
130 flush_workqueue(afs_async_calls);
131 sock_release(afs_socket);
132
133 _debug("dework");
134 destroy_workqueue(afs_async_calls);
135 _leave("");
136 }
137
138 /*
139 * Allocate a call.
140 */
141 static struct afs_call *afs_alloc_call(const struct afs_call_type *type,
142 gfp_t gfp)
143 {
144 struct afs_call *call;
145 int o;
146
147 call = kzalloc(sizeof(*call), gfp);
148 if (!call)
149 return NULL;
150
151 call->type = type;
152 atomic_set(&call->usage, 1);
153 INIT_WORK(&call->async_work, afs_process_async_call);
154 init_waitqueue_head(&call->waitq);
155
156 o = atomic_inc_return(&afs_outstanding_calls);
157 trace_afs_call(call, afs_call_trace_alloc, 1, o,
158 __builtin_return_address(0));
159 return call;
160 }
161
162 /*
163 * Dispose of a reference on a call.
164 */
165 void afs_put_call(struct afs_call *call)
166 {
167 int n = atomic_dec_return(&call->usage);
168 int o = atomic_read(&afs_outstanding_calls);
169
170 trace_afs_call(call, afs_call_trace_put, n + 1, o,
171 __builtin_return_address(0));
172
173 ASSERTCMP(n, >=, 0);
174 if (n == 0) {
175 ASSERT(!work_pending(&call->async_work));
176 ASSERT(call->type->name != NULL);
177
178 if (call->rxcall) {
179 rxrpc_kernel_end_call(afs_socket, call->rxcall);
180 call->rxcall = NULL;
181 }
182 if (call->type->destructor)
183 call->type->destructor(call);
184
185 kfree(call->request);
186 kfree(call);
187
188 o = atomic_dec_return(&afs_outstanding_calls);
189 trace_afs_call(call, afs_call_trace_free, 0, o,
190 __builtin_return_address(0));
191 if (o == 0)
192 wake_up_atomic_t(&afs_outstanding_calls);
193 }
194 }
195
196 /*
197 * Queue the call for actual work. Returns 0 unconditionally for convenience.
198 */
199 int afs_queue_call_work(struct afs_call *call)
200 {
201 int u = atomic_inc_return(&call->usage);
202
203 trace_afs_call(call, afs_call_trace_work, u,
204 atomic_read(&afs_outstanding_calls),
205 __builtin_return_address(0));
206
207 INIT_WORK(&call->work, call->type->work);
208
209 if (!queue_work(afs_wq, &call->work))
210 afs_put_call(call);
211 return 0;
212 }
213
214 /*
215 * allocate a call with flat request and reply buffers
216 */
217 struct afs_call *afs_alloc_flat_call(const struct afs_call_type *type,
218 size_t request_size, size_t reply_max)
219 {
220 struct afs_call *call;
221
222 call = afs_alloc_call(type, GFP_NOFS);
223 if (!call)
224 goto nomem_call;
225
226 if (request_size) {
227 call->request_size = request_size;
228 call->request = kmalloc(request_size, GFP_NOFS);
229 if (!call->request)
230 goto nomem_free;
231 }
232
233 if (reply_max) {
234 call->reply_max = reply_max;
235 call->buffer = kmalloc(reply_max, GFP_NOFS);
236 if (!call->buffer)
237 goto nomem_free;
238 }
239
240 init_waitqueue_head(&call->waitq);
241 return call;
242
243 nomem_free:
244 afs_put_call(call);
245 nomem_call:
246 return NULL;
247 }
248
249 /*
250 * clean up a call with flat buffer
251 */
252 void afs_flat_call_destructor(struct afs_call *call)
253 {
254 _enter("");
255
256 kfree(call->request);
257 call->request = NULL;
258 kfree(call->buffer);
259 call->buffer = NULL;
260 }
261
262 #define AFS_BVEC_MAX 8
263
264 /*
265 * Load the given bvec with the next few pages.
266 */
267 static void afs_load_bvec(struct afs_call *call, struct msghdr *msg,
268 struct bio_vec *bv, pgoff_t first, pgoff_t last,
269 unsigned offset)
270 {
271 struct page *pages[AFS_BVEC_MAX];
272 unsigned int nr, n, i, to, bytes = 0;
273
274 nr = min_t(pgoff_t, last - first + 1, AFS_BVEC_MAX);
275 n = find_get_pages_contig(call->mapping, first, nr, pages);
276 ASSERTCMP(n, ==, nr);
277
278 msg->msg_flags |= MSG_MORE;
279 for (i = 0; i < nr; i++) {
280 to = PAGE_SIZE;
281 if (first + i >= last) {
282 to = call->last_to;
283 msg->msg_flags &= ~MSG_MORE;
284 }
285 bv[i].bv_page = pages[i];
286 bv[i].bv_len = to - offset;
287 bv[i].bv_offset = offset;
288 bytes += to - offset;
289 offset = 0;
290 }
291
292 iov_iter_bvec(&msg->msg_iter, WRITE | ITER_BVEC, bv, nr, bytes);
293 }
294
295 /*
296 * attach the data from a bunch of pages on an inode to a call
297 */
298 static int afs_send_pages(struct afs_call *call, struct msghdr *msg)
299 {
300 struct bio_vec bv[AFS_BVEC_MAX];
301 unsigned int bytes, nr, loop, offset;
302 pgoff_t first = call->first, last = call->last;
303 int ret;
304
305 offset = call->first_offset;
306 call->first_offset = 0;
307
308 do {
309 afs_load_bvec(call, msg, bv, first, last, offset);
310 offset = 0;
311 bytes = msg->msg_iter.count;
312 nr = msg->msg_iter.nr_segs;
313
314 /* Have to change the state *before* sending the last
315 * packet as RxRPC might give us the reply before it
316 * returns from sending the request.
317 */
318 if (first + nr - 1 >= last)
319 call->state = AFS_CALL_AWAIT_REPLY;
320 ret = rxrpc_kernel_send_data(afs_socket, call->rxcall,
321 msg, bytes);
322 for (loop = 0; loop < nr; loop++)
323 put_page(bv[loop].bv_page);
324 if (ret < 0)
325 break;
326
327 first += nr;
328 } while (first <= last);
329
330 return ret;
331 }
332
333 /*
334 * initiate a call
335 */
336 int afs_make_call(struct in_addr *addr, struct afs_call *call, gfp_t gfp,
337 bool async)
338 {
339 struct sockaddr_rxrpc srx;
340 struct rxrpc_call *rxcall;
341 struct msghdr msg;
342 struct kvec iov[1];
343 size_t offset;
344 u32 abort_code;
345 int ret;
346
347 _enter("%x,{%d},", addr->s_addr, ntohs(call->port));
348
349 ASSERT(call->type != NULL);
350 ASSERT(call->type->name != NULL);
351
352 _debug("____MAKE %p{%s,%x} [%d]____",
353 call, call->type->name, key_serial(call->key),
354 atomic_read(&afs_outstanding_calls));
355
356 call->async = async;
357
358 memset(&srx, 0, sizeof(srx));
359 srx.srx_family = AF_RXRPC;
360 srx.srx_service = call->service_id;
361 srx.transport_type = SOCK_DGRAM;
362 srx.transport_len = sizeof(srx.transport.sin);
363 srx.transport.sin.sin_family = AF_INET;
364 srx.transport.sin.sin_port = call->port;
365 memcpy(&srx.transport.sin.sin_addr, addr, 4);
366
367 /* create a call */
368 rxcall = rxrpc_kernel_begin_call(afs_socket, &srx, call->key,
369 (unsigned long) call, gfp,
370 (async ?
371 afs_wake_up_async_call :
372 afs_wake_up_call_waiter));
373 call->key = NULL;
374 if (IS_ERR(rxcall)) {
375 ret = PTR_ERR(rxcall);
376 goto error_kill_call;
377 }
378
379 call->rxcall = rxcall;
380
381 /* send the request */
382 iov[0].iov_base = call->request;
383 iov[0].iov_len = call->request_size;
384
385 msg.msg_name = NULL;
386 msg.msg_namelen = 0;
387 iov_iter_kvec(&msg.msg_iter, WRITE | ITER_KVEC, iov, 1,
388 call->request_size);
389 msg.msg_control = NULL;
390 msg.msg_controllen = 0;
391 msg.msg_flags = (call->send_pages ? MSG_MORE : 0);
392
393 /* We have to change the state *before* sending the last packet as
394 * rxrpc might give us the reply before it returns from sending the
395 * request. Further, if the send fails, we may already have been given
396 * a notification and may have collected it.
397 */
398 if (!call->send_pages)
399 call->state = AFS_CALL_AWAIT_REPLY;
400 ret = rxrpc_kernel_send_data(afs_socket, rxcall,
401 &msg, call->request_size);
402 if (ret < 0)
403 goto error_do_abort;
404
405 if (call->send_pages) {
406 ret = afs_send_pages(call, &msg);
407 if (ret < 0)
408 goto error_do_abort;
409 }
410
411 /* at this point, an async call may no longer exist as it may have
412 * already completed */
413 if (call->async)
414 return -EINPROGRESS;
415
416 return afs_wait_for_call_to_complete(call);
417
418 error_do_abort:
419 call->state = AFS_CALL_COMPLETE;
420 if (ret != -ECONNABORTED) {
421 rxrpc_kernel_abort_call(afs_socket, rxcall, RX_USER_ABORT,
422 ret, "KSD");
423 } else {
424 abort_code = 0;
425 offset = 0;
426 rxrpc_kernel_recv_data(afs_socket, rxcall, NULL, 0, &offset,
427 false, &abort_code);
428 ret = call->type->abort_to_error(abort_code);
429 }
430 error_kill_call:
431 afs_put_call(call);
432 _leave(" = %d", ret);
433 return ret;
434 }
435
436 /*
437 * deliver messages to a call
438 */
439 static void afs_deliver_to_call(struct afs_call *call)
440 {
441 u32 abort_code;
442 int ret;
443
444 _enter("%s", call->type->name);
445
446 while (call->state == AFS_CALL_AWAIT_REPLY ||
447 call->state == AFS_CALL_AWAIT_OP_ID ||
448 call->state == AFS_CALL_AWAIT_REQUEST ||
449 call->state == AFS_CALL_AWAIT_ACK
450 ) {
451 if (call->state == AFS_CALL_AWAIT_ACK) {
452 size_t offset = 0;
453 ret = rxrpc_kernel_recv_data(afs_socket, call->rxcall,
454 NULL, 0, &offset, false,
455 &call->abort_code);
456 trace_afs_recv_data(call, 0, offset, false, ret);
457
458 if (ret == -EINPROGRESS || ret == -EAGAIN)
459 return;
460 if (ret == 1 || ret < 0) {
461 call->state = AFS_CALL_COMPLETE;
462 goto done;
463 }
464 return;
465 }
466
467 ret = call->type->deliver(call);
468 switch (ret) {
469 case 0:
470 if (call->state == AFS_CALL_AWAIT_REPLY)
471 call->state = AFS_CALL_COMPLETE;
472 goto done;
473 case -EINPROGRESS:
474 case -EAGAIN:
475 goto out;
476 case -ECONNABORTED:
477 goto call_complete;
478 case -ENOTCONN:
479 abort_code = RX_CALL_DEAD;
480 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
481 abort_code, ret, "KNC");
482 goto save_error;
483 case -ENOTSUPP:
484 abort_code = RXGEN_OPCODE;
485 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
486 abort_code, ret, "KIV");
487 goto save_error;
488 case -ENODATA:
489 case -EBADMSG:
490 case -EMSGSIZE:
491 default:
492 abort_code = RXGEN_CC_UNMARSHAL;
493 if (call->state != AFS_CALL_AWAIT_REPLY)
494 abort_code = RXGEN_SS_UNMARSHAL;
495 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
496 abort_code, -EBADMSG, "KUM");
497 goto save_error;
498 }
499 }
500
501 done:
502 if (call->state == AFS_CALL_COMPLETE && call->incoming)
503 afs_put_call(call);
504 out:
505 _leave("");
506 return;
507
508 save_error:
509 call->error = ret;
510 call_complete:
511 call->state = AFS_CALL_COMPLETE;
512 goto done;
513 }
514
515 /*
516 * wait synchronously for a call to complete
517 */
518 static int afs_wait_for_call_to_complete(struct afs_call *call)
519 {
520 int ret;
521
522 DECLARE_WAITQUEUE(myself, current);
523
524 _enter("");
525
526 add_wait_queue(&call->waitq, &myself);
527 for (;;) {
528 set_current_state(TASK_INTERRUPTIBLE);
529
530 /* deliver any messages that are in the queue */
531 if (call->state < AFS_CALL_COMPLETE && call->need_attention) {
532 call->need_attention = false;
533 __set_current_state(TASK_RUNNING);
534 afs_deliver_to_call(call);
535 continue;
536 }
537
538 if (call->state == AFS_CALL_COMPLETE ||
539 signal_pending(current))
540 break;
541 schedule();
542 }
543
544 remove_wait_queue(&call->waitq, &myself);
545 __set_current_state(TASK_RUNNING);
546
547 /* Kill off the call if it's still live. */
548 if (call->state < AFS_CALL_COMPLETE) {
549 _debug("call interrupted");
550 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
551 RX_USER_ABORT, -EINTR, "KWI");
552 }
553
554 ret = call->error;
555 _debug("call complete");
556 afs_put_call(call);
557 _leave(" = %d", ret);
558 return ret;
559 }
560
561 /*
562 * wake up a waiting call
563 */
564 static void afs_wake_up_call_waiter(struct sock *sk, struct rxrpc_call *rxcall,
565 unsigned long call_user_ID)
566 {
567 struct afs_call *call = (struct afs_call *)call_user_ID;
568
569 call->need_attention = true;
570 wake_up(&call->waitq);
571 }
572
573 /*
574 * wake up an asynchronous call
575 */
576 static void afs_wake_up_async_call(struct sock *sk, struct rxrpc_call *rxcall,
577 unsigned long call_user_ID)
578 {
579 struct afs_call *call = (struct afs_call *)call_user_ID;
580 int u;
581
582 trace_afs_notify_call(rxcall, call);
583 call->need_attention = true;
584
585 u = __atomic_add_unless(&call->usage, 1, 0);
586 if (u != 0) {
587 trace_afs_call(call, afs_call_trace_wake, u,
588 atomic_read(&afs_outstanding_calls),
589 __builtin_return_address(0));
590
591 if (!queue_work(afs_async_calls, &call->async_work))
592 afs_put_call(call);
593 }
594 }
595
596 /*
597 * Delete an asynchronous call. The work item carries a ref to the call struct
598 * that we need to release.
599 */
600 static void afs_delete_async_call(struct work_struct *work)
601 {
602 struct afs_call *call = container_of(work, struct afs_call, async_work);
603
604 _enter("");
605
606 afs_put_call(call);
607
608 _leave("");
609 }
610
611 /*
612 * Perform I/O processing on an asynchronous call. The work item carries a ref
613 * to the call struct that we either need to release or to pass on.
614 */
615 static void afs_process_async_call(struct work_struct *work)
616 {
617 struct afs_call *call = container_of(work, struct afs_call, async_work);
618
619 _enter("");
620
621 if (call->state < AFS_CALL_COMPLETE && call->need_attention) {
622 call->need_attention = false;
623 afs_deliver_to_call(call);
624 }
625
626 if (call->state == AFS_CALL_COMPLETE) {
627 call->reply = NULL;
628
629 /* We have two refs to release - one from the alloc and one
630 * queued with the work item - and we can't just deallocate the
631 * call because the work item may be queued again.
632 */
633 call->async_work.func = afs_delete_async_call;
634 if (!queue_work(afs_async_calls, &call->async_work))
635 afs_put_call(call);
636 }
637
638 afs_put_call(call);
639 _leave("");
640 }
641
642 static void afs_rx_attach(struct rxrpc_call *rxcall, unsigned long user_call_ID)
643 {
644 struct afs_call *call = (struct afs_call *)user_call_ID;
645
646 call->rxcall = rxcall;
647 }
648
649 /*
650 * Charge the incoming call preallocation.
651 */
652 static void afs_charge_preallocation(struct work_struct *work)
653 {
654 struct afs_call *call = afs_spare_incoming_call;
655
656 for (;;) {
657 if (!call) {
658 call = afs_alloc_call(&afs_RXCMxxxx, GFP_KERNEL);
659 if (!call)
660 break;
661
662 call->async = true;
663 call->state = AFS_CALL_AWAIT_OP_ID;
664 init_waitqueue_head(&call->waitq);
665 }
666
667 if (rxrpc_kernel_charge_accept(afs_socket,
668 afs_wake_up_async_call,
669 afs_rx_attach,
670 (unsigned long)call,
671 GFP_KERNEL) < 0)
672 break;
673 call = NULL;
674 }
675 afs_spare_incoming_call = call;
676 }
677
678 /*
679 * Discard a preallocated call when a socket is shut down.
680 */
681 static void afs_rx_discard_new_call(struct rxrpc_call *rxcall,
682 unsigned long user_call_ID)
683 {
684 struct afs_call *call = (struct afs_call *)user_call_ID;
685
686 call->rxcall = NULL;
687 afs_put_call(call);
688 }
689
690 /*
691 * Notification of an incoming call.
692 */
693 static void afs_rx_new_call(struct sock *sk, struct rxrpc_call *rxcall,
694 unsigned long user_call_ID)
695 {
696 queue_work(afs_wq, &afs_charge_preallocation_work);
697 }
698
699 /*
700 * Grab the operation ID from an incoming cache manager call. The socket
701 * buffer is discarded on error or if we don't yet have sufficient data.
702 */
703 static int afs_deliver_cm_op_id(struct afs_call *call)
704 {
705 int ret;
706
707 _enter("{%zu}", call->offset);
708
709 ASSERTCMP(call->offset, <, 4);
710
711 /* the operation ID forms the first four bytes of the request data */
712 ret = afs_extract_data(call, &call->tmp, 4, true);
713 if (ret < 0)
714 return ret;
715
716 call->operation_ID = ntohl(call->tmp);
717 call->state = AFS_CALL_AWAIT_REQUEST;
718 call->offset = 0;
719
720 /* ask the cache manager to route the call (it'll change the call type
721 * if successful) */
722 if (!afs_cm_incoming_call(call))
723 return -ENOTSUPP;
724
725 trace_afs_cb_call(call);
726
727 /* pass responsibility for the remainer of this message off to the
728 * cache manager op */
729 return call->type->deliver(call);
730 }
731
732 /*
733 * send an empty reply
734 */
735 void afs_send_empty_reply(struct afs_call *call)
736 {
737 struct msghdr msg;
738
739 _enter("");
740
741 msg.msg_name = NULL;
742 msg.msg_namelen = 0;
743 iov_iter_kvec(&msg.msg_iter, WRITE | ITER_KVEC, NULL, 0, 0);
744 msg.msg_control = NULL;
745 msg.msg_controllen = 0;
746 msg.msg_flags = 0;
747
748 call->state = AFS_CALL_AWAIT_ACK;
749 switch (rxrpc_kernel_send_data(afs_socket, call->rxcall, &msg, 0)) {
750 case 0:
751 _leave(" [replied]");
752 return;
753
754 case -ENOMEM:
755 _debug("oom");
756 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
757 RX_USER_ABORT, -ENOMEM, "KOO");
758 default:
759 _leave(" [error]");
760 return;
761 }
762 }
763
764 /*
765 * send a simple reply
766 */
767 void afs_send_simple_reply(struct afs_call *call, const void *buf, size_t len)
768 {
769 struct msghdr msg;
770 struct kvec iov[1];
771 int n;
772
773 _enter("");
774
775 iov[0].iov_base = (void *) buf;
776 iov[0].iov_len = len;
777 msg.msg_name = NULL;
778 msg.msg_namelen = 0;
779 iov_iter_kvec(&msg.msg_iter, WRITE | ITER_KVEC, iov, 1, len);
780 msg.msg_control = NULL;
781 msg.msg_controllen = 0;
782 msg.msg_flags = 0;
783
784 call->state = AFS_CALL_AWAIT_ACK;
785 n = rxrpc_kernel_send_data(afs_socket, call->rxcall, &msg, len);
786 if (n >= 0) {
787 /* Success */
788 _leave(" [replied]");
789 return;
790 }
791
792 if (n == -ENOMEM) {
793 _debug("oom");
794 rxrpc_kernel_abort_call(afs_socket, call->rxcall,
795 RX_USER_ABORT, -ENOMEM, "KOO");
796 }
797 _leave(" [error]");
798 }
799
800 /*
801 * Extract a piece of data from the received data socket buffers.
802 */
803 int afs_extract_data(struct afs_call *call, void *buf, size_t count,
804 bool want_more)
805 {
806 int ret;
807
808 _enter("{%s,%zu},,%zu,%d",
809 call->type->name, call->offset, count, want_more);
810
811 ASSERTCMP(call->offset, <=, count);
812
813 ret = rxrpc_kernel_recv_data(afs_socket, call->rxcall,
814 buf, count, &call->offset,
815 want_more, &call->abort_code);
816 trace_afs_recv_data(call, count, call->offset, want_more, ret);
817 if (ret == 0 || ret == -EAGAIN)
818 return ret;
819
820 if (ret == 1) {
821 switch (call->state) {
822 case AFS_CALL_AWAIT_REPLY:
823 call->state = AFS_CALL_COMPLETE;
824 break;
825 case AFS_CALL_AWAIT_REQUEST:
826 call->state = AFS_CALL_REPLYING;
827 break;
828 default:
829 break;
830 }
831 return 0;
832 }
833
834 if (ret == -ECONNABORTED)
835 call->error = call->type->abort_to_error(call->abort_code);
836 else
837 call->error = ret;
838 call->state = AFS_CALL_COMPLETE;
839 return ret;
840 }