projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
ata: add AMD Seattle platform driver
[deliverable/linux.git] / drivers / staging / lustre / lnet / selftest / rpc.c
1 /*
2 * GPL HEADER START
3 *
4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
9 *
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
19 *
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
22 * have any questions.
23 *
24 * GPL HEADER END
25 */
26 /*
27 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
29 *
30 * Copyright (c) 2012, 2015, Intel Corporation.
31 */
32 /*
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
35 *
36 * lnet/selftest/rpc.c
37 *
38 * Author: Isaac Huang <isaac@clusterfs.com>
39 *
40 * 2012-05-13: Liang Zhen <liang@whamcloud.com>
41 * - percpt data for service to improve smp performance
42 * - code cleanup
43 */
44
45 #define DEBUG_SUBSYSTEM S_LNET
46
47 #include "selftest.h"
48
49 typedef enum {
50 SRPC_STATE_NONE,
51 SRPC_STATE_NI_INIT,
52 SRPC_STATE_EQ_INIT,
53 SRPC_STATE_RUNNING,
54 SRPC_STATE_STOPPING,
55 } srpc_state_t;
56
57 static struct smoketest_rpc {
58 spinlock_t rpc_glock; /* global lock */
59 srpc_service_t *rpc_services[SRPC_SERVICE_MAX_ID + 1];
60 lnet_handle_eq_t rpc_lnet_eq; /* _the_ LNet event queue */
61 srpc_state_t rpc_state;
62 srpc_counters_t rpc_counters;
63 __u64 rpc_matchbits; /* matchbits counter */
64 } srpc_data;
65
66 static inline int
67 srpc_serv_portal(int svc_id)
68 {
69 return svc_id < SRPC_FRAMEWORK_SERVICE_MAX_ID ?
70 SRPC_FRAMEWORK_REQUEST_PORTAL : SRPC_REQUEST_PORTAL;
71 }
72
73 /* forward ref's */
74 int srpc_handle_rpc(swi_workitem_t *wi);
75
76 void srpc_get_counters(srpc_counters_t *cnt)
77 {
78 spin_lock(&srpc_data.rpc_glock);
79 *cnt = srpc_data.rpc_counters;
80 spin_unlock(&srpc_data.rpc_glock);
81 }
82
83 void srpc_set_counters(const srpc_counters_t *cnt)
84 {
85 spin_lock(&srpc_data.rpc_glock);
86 srpc_data.rpc_counters = *cnt;
87 spin_unlock(&srpc_data.rpc_glock);
88 }
89
90 static int
91 srpc_add_bulk_page(srpc_bulk_t *bk, struct page *pg, int i, int nob)
92 {
93 nob = min_t(int, nob, PAGE_CACHE_SIZE);
94
95 LASSERT(nob > 0);
96 LASSERT(i >= 0 && i < bk->bk_niov);
97
98 bk->bk_iovs[i].kiov_offset = 0;
99 bk->bk_iovs[i].kiov_page = pg;
100 bk->bk_iovs[i].kiov_len = nob;
101 return nob;
102 }
103
104 void
105 srpc_free_bulk(srpc_bulk_t *bk)
106 {
107 int i;
108 struct page *pg;
109
110 LASSERT(bk);
111
112 for (i = 0; i < bk->bk_niov; i++) {
113 pg = bk->bk_iovs[i].kiov_page;
114 if (!pg)
115 break;
116
117 __free_page(pg);
118 }
119
120 LIBCFS_FREE(bk, offsetof(srpc_bulk_t, bk_iovs[bk->bk_niov]));
121 }
122
123 srpc_bulk_t *
124 srpc_alloc_bulk(int cpt, unsigned bulk_npg, unsigned bulk_len, int sink)
125 {
126 srpc_bulk_t *bk;
127 int i;
128
129 LASSERT(bulk_npg > 0 && bulk_npg <= LNET_MAX_IOV);
130
131 LIBCFS_CPT_ALLOC(bk, lnet_cpt_table(), cpt,
132 offsetof(srpc_bulk_t, bk_iovs[bulk_npg]));
133 if (!bk) {
134 CERROR("Can't allocate descriptor for %d pages\n", bulk_npg);
135 return NULL;
136 }
137
138 memset(bk, 0, offsetof(srpc_bulk_t, bk_iovs[bulk_npg]));
139 bk->bk_sink = sink;
140 bk->bk_len = bulk_len;
141 bk->bk_niov = bulk_npg;
142
143 for (i = 0; i < bulk_npg; i++) {
144 struct page *pg;
145 int nob;
146
147 pg = alloc_pages_node(cfs_cpt_spread_node(lnet_cpt_table(), cpt),
148 GFP_KERNEL, 0);
149 if (!pg) {
150 CERROR("Can't allocate page %d of %d\n", i, bulk_npg);
151 srpc_free_bulk(bk);
152 return NULL;
153 }
154
155 nob = srpc_add_bulk_page(bk, pg, i, bulk_len);
156 bulk_len -= nob;
157 }
158
159 return bk;
160 }
161
162 static inline __u64
163 srpc_next_id(void)
164 {
165 __u64 id;
166
167 spin_lock(&srpc_data.rpc_glock);
168 id = srpc_data.rpc_matchbits++;
169 spin_unlock(&srpc_data.rpc_glock);
170 return id;
171 }
172
173 static void
174 srpc_init_server_rpc(struct srpc_server_rpc *rpc,
175 struct srpc_service_cd *scd,
176 struct srpc_buffer *buffer)
177 {
178 memset(rpc, 0, sizeof(*rpc));
179 swi_init_workitem(&rpc->srpc_wi, rpc, srpc_handle_rpc,
180 srpc_serv_is_framework(scd->scd_svc) ?
181 lst_sched_serial : lst_sched_test[scd->scd_cpt]);
182
183 rpc->srpc_ev.ev_fired = 1; /* no event expected now */
184
185 rpc->srpc_scd = scd;
186 rpc->srpc_reqstbuf = buffer;
187 rpc->srpc_peer = buffer->buf_peer;
188 rpc->srpc_self = buffer->buf_self;
189 LNetInvalidateHandle(&rpc->srpc_replymdh);
190 }
191
192 static void
193 srpc_service_fini(struct srpc_service *svc)
194 {
195 struct srpc_service_cd *scd;
196 struct srpc_server_rpc *rpc;
197 struct srpc_buffer *buf;
198 struct list_head *q;
199 int i;
200
201 if (!svc->sv_cpt_data)
202 return;
203
204 cfs_percpt_for_each(scd, i, svc->sv_cpt_data) {
205 while (1) {
206 if (!list_empty(&scd->scd_buf_posted))
207 q = &scd->scd_buf_posted;
208 else if (!list_empty(&scd->scd_buf_blocked))
209 q = &scd->scd_buf_blocked;
210 else
211 break;
212
213 while (!list_empty(q)) {
214 buf = list_entry(q->next, struct srpc_buffer,
215 buf_list);
216 list_del(&buf->buf_list);
217 LIBCFS_FREE(buf, sizeof(*buf));
218 }
219 }
220
221 LASSERT(list_empty(&scd->scd_rpc_active));
222
223 while (!list_empty(&scd->scd_rpc_free)) {
224 rpc = list_entry(scd->scd_rpc_free.next,
225 struct srpc_server_rpc,
226 srpc_list);
227 list_del(&rpc->srpc_list);
228 LIBCFS_FREE(rpc, sizeof(*rpc));
229 }
230 }
231
232 cfs_percpt_free(svc->sv_cpt_data);
233 svc->sv_cpt_data = NULL;
234 }
235
236 static int
237 srpc_service_nrpcs(struct srpc_service *svc)
238 {
239 int nrpcs = svc->sv_wi_total / svc->sv_ncpts;
240
241 return srpc_serv_is_framework(svc) ?
242 max(nrpcs, SFW_FRWK_WI_MIN) : max(nrpcs, SFW_TEST_WI_MIN);
243 }
244
245 int srpc_add_buffer(struct swi_workitem *wi);
246
247 static int
248 srpc_service_init(struct srpc_service *svc)
249 {
250 struct srpc_service_cd *scd;
251 struct srpc_server_rpc *rpc;
252 int nrpcs;
253 int i;
254 int j;
255
256 svc->sv_shuttingdown = 0;
257
258 svc->sv_cpt_data = cfs_percpt_alloc(lnet_cpt_table(),
259 sizeof(struct srpc_service_cd));
260 if (!svc->sv_cpt_data)
261 return -ENOMEM;
262
263 svc->sv_ncpts = srpc_serv_is_framework(svc) ?
264 1 : cfs_cpt_number(lnet_cpt_table());
265 nrpcs = srpc_service_nrpcs(svc);
266
267 cfs_percpt_for_each(scd, i, svc->sv_cpt_data) {
268 scd->scd_cpt = i;
269 scd->scd_svc = svc;
270 spin_lock_init(&scd->scd_lock);
271 INIT_LIST_HEAD(&scd->scd_rpc_free);
272 INIT_LIST_HEAD(&scd->scd_rpc_active);
273 INIT_LIST_HEAD(&scd->scd_buf_posted);
274 INIT_LIST_HEAD(&scd->scd_buf_blocked);
275
276 scd->scd_ev.ev_data = scd;
277 scd->scd_ev.ev_type = SRPC_REQUEST_RCVD;
278
279 /*
280 * NB: don't use lst_sched_serial for adding buffer,
281 * see details in srpc_service_add_buffers()
282 */
283 swi_init_workitem(&scd->scd_buf_wi, scd,
284 srpc_add_buffer, lst_sched_test[i]);
285
286 if (i && srpc_serv_is_framework(svc)) {
287 /*
288 * NB: framework service only needs srpc_service_cd for
289 * one partition, but we allocate for all to make
290 * it easier to implement, it will waste a little
291 * memory but nobody should care about this
292 */
293 continue;
294 }
295
296 for (j = 0; j < nrpcs; j++) {
297 LIBCFS_CPT_ALLOC(rpc, lnet_cpt_table(),
298 i, sizeof(*rpc));
299 if (!rpc) {
300 srpc_service_fini(svc);
301 return -ENOMEM;
302 }
303 list_add(&rpc->srpc_list, &scd->scd_rpc_free);
304 }
305 }
306
307 return 0;
308 }
309
310 int
311 srpc_add_service(struct srpc_service *sv)
312 {
313 int id = sv->sv_id;
314
315 LASSERT(0 <= id && id <= SRPC_SERVICE_MAX_ID);
316
317 if (srpc_service_init(sv))
318 return -ENOMEM;
319
320 spin_lock(&srpc_data.rpc_glock);
321
322 LASSERT(srpc_data.rpc_state == SRPC_STATE_RUNNING);
323
324 if (srpc_data.rpc_services[id]) {
325 spin_unlock(&srpc_data.rpc_glock);
326 goto failed;
327 }
328
329 srpc_data.rpc_services[id] = sv;
330 spin_unlock(&srpc_data.rpc_glock);
331
332 CDEBUG(D_NET, "Adding service: id %d, name %s\n", id, sv->sv_name);
333 return 0;
334
335 failed:
336 srpc_service_fini(sv);
337 return -EBUSY;
338 }
339
340 int
341 srpc_remove_service(srpc_service_t *sv)
342 {
343 int id = sv->sv_id;
344
345 spin_lock(&srpc_data.rpc_glock);
346
347 if (srpc_data.rpc_services[id] != sv) {
348 spin_unlock(&srpc_data.rpc_glock);
349 return -ENOENT;
350 }
351
352 srpc_data.rpc_services[id] = NULL;
353 spin_unlock(&srpc_data.rpc_glock);
354 return 0;
355 }
356
357 static int
358 srpc_post_passive_rdma(int portal, int local, __u64 matchbits, void *buf,
359 int len, int options, lnet_process_id_t peer,
360 lnet_handle_md_t *mdh, srpc_event_t *ev)
361 {
362 int rc;
363 lnet_md_t md;
364 lnet_handle_me_t meh;
365
366 rc = LNetMEAttach(portal, peer, matchbits, 0, LNET_UNLINK,
367 local ? LNET_INS_LOCAL : LNET_INS_AFTER, &meh);
368 if (rc) {
369 CERROR("LNetMEAttach failed: %d\n", rc);
370 LASSERT(rc == -ENOMEM);
371 return -ENOMEM;
372 }
373
374 md.threshold = 1;
375 md.user_ptr = ev;
376 md.start = buf;
377 md.length = len;
378 md.options = options;
379 md.eq_handle = srpc_data.rpc_lnet_eq;
380
381 rc = LNetMDAttach(meh, md, LNET_UNLINK, mdh);
382 if (rc) {
383 CERROR("LNetMDAttach failed: %d\n", rc);
384 LASSERT(rc == -ENOMEM);
385
386 rc = LNetMEUnlink(meh);
387 LASSERT(!rc);
388 return -ENOMEM;
389 }
390
391 CDEBUG(D_NET, "Posted passive RDMA: peer %s, portal %d, matchbits %#llx\n",
392 libcfs_id2str(peer), portal, matchbits);
393 return 0;
394 }
395
396 static int
397 srpc_post_active_rdma(int portal, __u64 matchbits, void *buf, int len,
398 int options, lnet_process_id_t peer, lnet_nid_t self,
399 lnet_handle_md_t *mdh, srpc_event_t *ev)
400 {
401 int rc;
402 lnet_md_t md;
403
404 md.user_ptr = ev;
405 md.start = buf;
406 md.length = len;
407 md.eq_handle = srpc_data.rpc_lnet_eq;
408 md.threshold = options & LNET_MD_OP_GET ? 2 : 1;
409 md.options = options & ~(LNET_MD_OP_PUT | LNET_MD_OP_GET);
410
411 rc = LNetMDBind(md, LNET_UNLINK, mdh);
412 if (rc) {
413 CERROR("LNetMDBind failed: %d\n", rc);
414 LASSERT(rc == -ENOMEM);
415 return -ENOMEM;
416 }
417
418 /*
419 * this is kind of an abuse of the LNET_MD_OP_{PUT,GET} options.
420 * they're only meaningful for MDs attached to an ME (i.e. passive
421 * buffers...
422 */
423 if (options & LNET_MD_OP_PUT) {
424 rc = LNetPut(self, *mdh, LNET_NOACK_REQ, peer,
425 portal, matchbits, 0, 0);
426 } else {
427 LASSERT(options & LNET_MD_OP_GET);
428
429 rc = LNetGet(self, *mdh, peer, portal, matchbits, 0);
430 }
431
432 if (rc) {
433 CERROR("LNet%s(%s, %d, %lld) failed: %d\n",
434 options & LNET_MD_OP_PUT ? "Put" : "Get",
435 libcfs_id2str(peer), portal, matchbits, rc);
436
437 /*
438 * The forthcoming unlink event will complete this operation
439 * with failure, so fall through and return success here.
440 */
441 rc = LNetMDUnlink(*mdh);
442 LASSERT(!rc);
443 } else {
444 CDEBUG(D_NET, "Posted active RDMA: peer %s, portal %u, matchbits %#llx\n",
445 libcfs_id2str(peer), portal, matchbits);
446 }
447 return 0;
448 }
449
450 static int
451 srpc_post_passive_rqtbuf(int service, int local, void *buf, int len,
452 lnet_handle_md_t *mdh, srpc_event_t *ev)
453 {
454 lnet_process_id_t any = { 0 };
455
456 any.nid = LNET_NID_ANY;
457 any.pid = LNET_PID_ANY;
458
459 return srpc_post_passive_rdma(srpc_serv_portal(service),
460 local, service, buf, len,
461 LNET_MD_OP_PUT, any, mdh, ev);
462 }
463
464 static int
465 srpc_service_post_buffer(struct srpc_service_cd *scd, struct srpc_buffer *buf)
466 __must_hold(&scd->scd_lock)
467 {
468 struct srpc_service *sv = scd->scd_svc;
469 struct srpc_msg *msg = &buf->buf_msg;
470 int rc;
471
472 LNetInvalidateHandle(&buf->buf_mdh);
473 list_add(&buf->buf_list, &scd->scd_buf_posted);
474 scd->scd_buf_nposted++;
475 spin_unlock(&scd->scd_lock);
476
477 rc = srpc_post_passive_rqtbuf(sv->sv_id,
478 !srpc_serv_is_framework(sv),
479 msg, sizeof(*msg), &buf->buf_mdh,
480 &scd->scd_ev);
481
482 /*
483 * At this point, a RPC (new or delayed) may have arrived in
484 * msg and its event handler has been called. So we must add
485 * buf to scd_buf_posted _before_ dropping scd_lock
486 */
487 spin_lock(&scd->scd_lock);
488
489 if (!rc) {
490 if (!sv->sv_shuttingdown)
491 return 0;
492
493 spin_unlock(&scd->scd_lock);
494 /*
495 * srpc_shutdown_service might have tried to unlink me
496 * when my buf_mdh was still invalid
497 */
498 LNetMDUnlink(buf->buf_mdh);
499 spin_lock(&scd->scd_lock);
500 return 0;
501 }
502
503 scd->scd_buf_nposted--;
504 if (sv->sv_shuttingdown)
505 return rc; /* don't allow to change scd_buf_posted */
506
507 list_del(&buf->buf_list);
508 spin_unlock(&scd->scd_lock);
509
510 LIBCFS_FREE(buf, sizeof(*buf));
511
512 spin_lock(&scd->scd_lock);
513 return rc;
514 }
515
516 int
517 srpc_add_buffer(struct swi_workitem *wi)
518 {
519 struct srpc_service_cd *scd = wi->swi_workitem.wi_data;
520 struct srpc_buffer *buf;
521 int rc = 0;
522
523 /*
524 * it's called by workitem scheduler threads, these threads
525 * should have been set CPT affinity, so buffers will be posted
526 * on CPT local list of Portal
527 */
528 spin_lock(&scd->scd_lock);
529
530 while (scd->scd_buf_adjust > 0 &&
531 !scd->scd_svc->sv_shuttingdown) {
532 scd->scd_buf_adjust--; /* consume it */
533 scd->scd_buf_posting++;
534
535 spin_unlock(&scd->scd_lock);
536
537 LIBCFS_ALLOC(buf, sizeof(*buf));
538 if (!buf) {
539 CERROR("Failed to add new buf to service: %s\n",
540 scd->scd_svc->sv_name);
541 spin_lock(&scd->scd_lock);
542 rc = -ENOMEM;
543 break;
544 }
545
546 spin_lock(&scd->scd_lock);
547 if (scd->scd_svc->sv_shuttingdown) {
548 spin_unlock(&scd->scd_lock);
549 LIBCFS_FREE(buf, sizeof(*buf));
550
551 spin_lock(&scd->scd_lock);
552 rc = -ESHUTDOWN;
553 break;
554 }
555
556 rc = srpc_service_post_buffer(scd, buf);
557 if (rc)
558 break; /* buf has been freed inside */
559
560 LASSERT(scd->scd_buf_posting > 0);
561 scd->scd_buf_posting--;
562 scd->scd_buf_total++;
563 scd->scd_buf_low = max(2, scd->scd_buf_total / 4);
564 }
565
566 if (rc) {
567 scd->scd_buf_err_stamp = ktime_get_real_seconds();
568 scd->scd_buf_err = rc;
569
570 LASSERT(scd->scd_buf_posting > 0);
571 scd->scd_buf_posting--;
572 }
573
574 spin_unlock(&scd->scd_lock);
575 return 0;
576 }
577
578 int
579 srpc_service_add_buffers(struct srpc_service *sv, int nbuffer)
580 {
581 struct srpc_service_cd *scd;
582 int rc = 0;
583 int i;
584
585 LASSERTF(nbuffer > 0, "nbuffer must be positive: %d\n", nbuffer);
586
587 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
588 spin_lock(&scd->scd_lock);
589
590 scd->scd_buf_err = 0;
591 scd->scd_buf_err_stamp = 0;
592 scd->scd_buf_posting = 0;
593 scd->scd_buf_adjust = nbuffer;
594 /* start to post buffers */
595 swi_schedule_workitem(&scd->scd_buf_wi);
596 spin_unlock(&scd->scd_lock);
597
598 /* framework service only post buffer for one partition */
599 if (srpc_serv_is_framework(sv))
600 break;
601 }
602
603 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
604 spin_lock(&scd->scd_lock);
605 /*
606 * NB: srpc_service_add_buffers() can be called inside
607 * thread context of lst_sched_serial, and we don't normally
608 * allow to sleep inside thread context of WI scheduler
609 * because it will block current scheduler thread from doing
610 * anything else, even worse, it could deadlock if it's
611 * waiting on result from another WI of the same scheduler.
612 * However, it's safe at here because scd_buf_wi is scheduled
613 * by thread in a different WI scheduler (lst_sched_test),
614 * so we don't have any risk of deadlock, though this could
615 * block all WIs pending on lst_sched_serial for a moment
616 * which is not good but not fatal.
617 */
618 lst_wait_until(scd->scd_buf_err ||
619 (!scd->scd_buf_adjust &&
620 !scd->scd_buf_posting),
621 scd->scd_lock, "waiting for adding buffer\n");
622
623 if (scd->scd_buf_err && !rc)
624 rc = scd->scd_buf_err;
625
626 spin_unlock(&scd->scd_lock);
627 }
628
629 return rc;
630 }
631
632 void
633 srpc_service_remove_buffers(struct srpc_service *sv, int nbuffer)
634 {
635 struct srpc_service_cd *scd;
636 int num;
637 int i;
638
639 LASSERT(!sv->sv_shuttingdown);
640
641 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
642 spin_lock(&scd->scd_lock);
643
644 num = scd->scd_buf_total + scd->scd_buf_posting;
645 scd->scd_buf_adjust -= min(nbuffer, num);
646
647 spin_unlock(&scd->scd_lock);
648 }
649 }
650
651 /* returns 1 if sv has finished, otherwise 0 */
652 int
653 srpc_finish_service(struct srpc_service *sv)
654 {
655 struct srpc_service_cd *scd;
656 struct srpc_server_rpc *rpc;
657 int i;
658
659 LASSERT(sv->sv_shuttingdown); /* srpc_shutdown_service called */
660
661 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
662 spin_lock(&scd->scd_lock);
663 if (!swi_deschedule_workitem(&scd->scd_buf_wi)) {
664 spin_unlock(&scd->scd_lock);
665 return 0;
666 }
667
668 if (scd->scd_buf_nposted > 0) {
669 CDEBUG(D_NET, "waiting for %d posted buffers to unlink\n",
670 scd->scd_buf_nposted);
671 spin_unlock(&scd->scd_lock);
672 return 0;
673 }
674
675 if (list_empty(&scd->scd_rpc_active)) {
676 spin_unlock(&scd->scd_lock);
677 continue;
678 }
679
680 rpc = list_entry(scd->scd_rpc_active.next,
681 struct srpc_server_rpc, srpc_list);
682 CNETERR("Active RPC %p on shutdown: sv %s, peer %s, wi %s scheduled %d running %d, ev fired %d type %d status %d lnet %d\n",
683 rpc, sv->sv_name, libcfs_id2str(rpc->srpc_peer),
684 swi_state2str(rpc->srpc_wi.swi_state),
685 rpc->srpc_wi.swi_workitem.wi_scheduled,
686 rpc->srpc_wi.swi_workitem.wi_running,
687 rpc->srpc_ev.ev_fired, rpc->srpc_ev.ev_type,
688 rpc->srpc_ev.ev_status, rpc->srpc_ev.ev_lnet);
689 spin_unlock(&scd->scd_lock);
690 return 0;
691 }
692
693 /* no lock needed from now on */
694 srpc_service_fini(sv);
695 return 1;
696 }
697
698 /* called with sv->sv_lock held */
699 static void
700 srpc_service_recycle_buffer(struct srpc_service_cd *scd, srpc_buffer_t *buf)
701 __must_hold(&scd->scd_lock)
702 {
703 if (!scd->scd_svc->sv_shuttingdown && scd->scd_buf_adjust >= 0) {
704 if (srpc_service_post_buffer(scd, buf)) {
705 CWARN("Failed to post %s buffer\n",
706 scd->scd_svc->sv_name);
707 }
708 return;
709 }
710
711 /* service is shutting down, or we want to recycle some buffers */
712 scd->scd_buf_total--;
713
714 if (scd->scd_buf_adjust < 0) {
715 scd->scd_buf_adjust++;
716 if (scd->scd_buf_adjust < 0 &&
717 !scd->scd_buf_total && !scd->scd_buf_posting) {
718 CDEBUG(D_INFO,
719 "Try to recycle %d buffers but nothing left\n",
720 scd->scd_buf_adjust);
721 scd->scd_buf_adjust = 0;
722 }
723 }
724
725 spin_unlock(&scd->scd_lock);
726 LIBCFS_FREE(buf, sizeof(*buf));
727 spin_lock(&scd->scd_lock);
728 }
729
730 void
731 srpc_abort_service(struct srpc_service *sv)
732 {
733 struct srpc_service_cd *scd;
734 struct srpc_server_rpc *rpc;
735 int i;
736
737 CDEBUG(D_NET, "Aborting service: id %d, name %s\n",
738 sv->sv_id, sv->sv_name);
739
740 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
741 spin_lock(&scd->scd_lock);
742
743 /*
744 * schedule in-flight RPCs to notice the abort, NB:
745 * racing with incoming RPCs; complete fix should make test
746 * RPCs carry session ID in its headers
747 */
748 list_for_each_entry(rpc, &scd->scd_rpc_active, srpc_list) {
749 rpc->srpc_aborted = 1;
750 swi_schedule_workitem(&rpc->srpc_wi);
751 }
752
753 spin_unlock(&scd->scd_lock);
754 }
755 }
756
757 void
758 srpc_shutdown_service(srpc_service_t *sv)
759 {
760 struct srpc_service_cd *scd;
761 struct srpc_server_rpc *rpc;
762 srpc_buffer_t *buf;
763 int i;
764
765 CDEBUG(D_NET, "Shutting down service: id %d, name %s\n",
766 sv->sv_id, sv->sv_name);
767
768 cfs_percpt_for_each(scd, i, sv->sv_cpt_data)
769 spin_lock(&scd->scd_lock);
770
771 sv->sv_shuttingdown = 1; /* i.e. no new active RPC */
772
773 cfs_percpt_for_each(scd, i, sv->sv_cpt_data)
774 spin_unlock(&scd->scd_lock);
775
776 cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
777 spin_lock(&scd->scd_lock);
778
779 /* schedule in-flight RPCs to notice the shutdown */
780 list_for_each_entry(rpc, &scd->scd_rpc_active, srpc_list)
781 swi_schedule_workitem(&rpc->srpc_wi);
782
783 spin_unlock(&scd->scd_lock);
784
785 /*
786 * OK to traverse scd_buf_posted without lock, since no one
787 * touches scd_buf_posted now
788 */
789 list_for_each_entry(buf, &scd->scd_buf_posted, buf_list)
790 LNetMDUnlink(buf->buf_mdh);
791 }
792 }
793
794 static int
795 srpc_send_request(srpc_client_rpc_t *rpc)
796 {
797 srpc_event_t *ev = &rpc->crpc_reqstev;
798 int rc;
799
800 ev->ev_fired = 0;
801 ev->ev_data = rpc;
802 ev->ev_type = SRPC_REQUEST_SENT;
803
804 rc = srpc_post_active_rdma(srpc_serv_portal(rpc->crpc_service),
805 rpc->crpc_service, &rpc->crpc_reqstmsg,
806 sizeof(srpc_msg_t), LNET_MD_OP_PUT,
807 rpc->crpc_dest, LNET_NID_ANY,
808 &rpc->crpc_reqstmdh, ev);
809 if (rc) {
810 LASSERT(rc == -ENOMEM);
811 ev->ev_fired = 1; /* no more event expected */
812 }
813 return rc;
814 }
815
816 static int
817 srpc_prepare_reply(srpc_client_rpc_t *rpc)
818 {
819 srpc_event_t *ev = &rpc->crpc_replyev;
820 __u64 *id = &rpc->crpc_reqstmsg.msg_body.reqst.rpyid;
821 int rc;
822
823 ev->ev_fired = 0;
824 ev->ev_data = rpc;
825 ev->ev_type = SRPC_REPLY_RCVD;
826
827 *id = srpc_next_id();
828
829 rc = srpc_post_passive_rdma(SRPC_RDMA_PORTAL, 0, *id,
830 &rpc->crpc_replymsg, sizeof(srpc_msg_t),
831 LNET_MD_OP_PUT, rpc->crpc_dest,
832 &rpc->crpc_replymdh, ev);
833 if (rc) {
834 LASSERT(rc == -ENOMEM);
835 ev->ev_fired = 1; /* no more event expected */
836 }
837 return rc;
838 }
839
840 static int
841 srpc_prepare_bulk(srpc_client_rpc_t *rpc)
842 {
843 srpc_bulk_t *bk = &rpc->crpc_bulk;
844 srpc_event_t *ev = &rpc->crpc_bulkev;
845 __u64 *id = &rpc->crpc_reqstmsg.msg_body.reqst.bulkid;
846 int rc;
847 int opt;
848
849 LASSERT(bk->bk_niov <= LNET_MAX_IOV);
850
851 if (!bk->bk_niov)
852 return 0; /* nothing to do */
853
854 opt = bk->bk_sink ? LNET_MD_OP_PUT : LNET_MD_OP_GET;
855 opt |= LNET_MD_KIOV;
856
857 ev->ev_fired = 0;
858 ev->ev_data = rpc;
859 ev->ev_type = SRPC_BULK_REQ_RCVD;
860
861 *id = srpc_next_id();
862
863 rc = srpc_post_passive_rdma(SRPC_RDMA_PORTAL, 0, *id,
864 &bk->bk_iovs[0], bk->bk_niov, opt,
865 rpc->crpc_dest, &bk->bk_mdh, ev);
866 if (rc) {
867 LASSERT(rc == -ENOMEM);
868 ev->ev_fired = 1; /* no more event expected */
869 }
870 return rc;
871 }
872
873 static int
874 srpc_do_bulk(struct srpc_server_rpc *rpc)
875 {
876 srpc_event_t *ev = &rpc->srpc_ev;
877 srpc_bulk_t *bk = rpc->srpc_bulk;
878 __u64 id = rpc->srpc_reqstbuf->buf_msg.msg_body.reqst.bulkid;
879 int rc;
880 int opt;
881
882 LASSERT(bk);
883
884 opt = bk->bk_sink ? LNET_MD_OP_GET : LNET_MD_OP_PUT;
885 opt |= LNET_MD_KIOV;
886
887 ev->ev_fired = 0;
888 ev->ev_data = rpc;
889 ev->ev_type = bk->bk_sink ? SRPC_BULK_GET_RPLD : SRPC_BULK_PUT_SENT;
890
891 rc = srpc_post_active_rdma(SRPC_RDMA_PORTAL, id,
892 &bk->bk_iovs[0], bk->bk_niov, opt,
893 rpc->srpc_peer, rpc->srpc_self,
894 &bk->bk_mdh, ev);
895 if (rc)
896 ev->ev_fired = 1; /* no more event expected */
897 return rc;
898 }
899
900 /* only called from srpc_handle_rpc */
901 static void
902 srpc_server_rpc_done(struct srpc_server_rpc *rpc, int status)
903 {
904 struct srpc_service_cd *scd = rpc->srpc_scd;
905 struct srpc_service *sv = scd->scd_svc;
906 srpc_buffer_t *buffer;
907
908 LASSERT(status || rpc->srpc_wi.swi_state == SWI_STATE_DONE);
909
910 rpc->srpc_status = status;
911
912 CDEBUG_LIMIT(!status ? D_NET : D_NETERROR,
913 "Server RPC %p done: service %s, peer %s, status %s:%d\n",
914 rpc, sv->sv_name, libcfs_id2str(rpc->srpc_peer),
915 swi_state2str(rpc->srpc_wi.swi_state), status);
916
917 if (status) {
918 spin_lock(&srpc_data.rpc_glock);
919 srpc_data.rpc_counters.rpcs_dropped++;
920 spin_unlock(&srpc_data.rpc_glock);
921 }
922
923 if (rpc->srpc_done)
924 (*rpc->srpc_done) (rpc);
925 LASSERT(!rpc->srpc_bulk);
926
927 spin_lock(&scd->scd_lock);
928
929 if (rpc->srpc_reqstbuf) {
930 /*
931 * NB might drop sv_lock in srpc_service_recycle_buffer, but
932 * sv won't go away for scd_rpc_active must not be empty
933 */
934 srpc_service_recycle_buffer(scd, rpc->srpc_reqstbuf);
935 rpc->srpc_reqstbuf = NULL;
936 }
937
938 list_del(&rpc->srpc_list); /* from scd->scd_rpc_active */
939
940 /*
941 * No one can schedule me now since:
942 * - I'm not on scd_rpc_active.
943 * - all LNet events have been fired.
944 * Cancel pending schedules and prevent future schedule attempts:
945 */
946 LASSERT(rpc->srpc_ev.ev_fired);
947 swi_exit_workitem(&rpc->srpc_wi);
948
949 if (!sv->sv_shuttingdown && !list_empty(&scd->scd_buf_blocked)) {
950 buffer = list_entry(scd->scd_buf_blocked.next,
951 srpc_buffer_t, buf_list);
952 list_del(&buffer->buf_list);
953
954 srpc_init_server_rpc(rpc, scd, buffer);
955 list_add_tail(&rpc->srpc_list, &scd->scd_rpc_active);
956 swi_schedule_workitem(&rpc->srpc_wi);
957 } else {
958 list_add(&rpc->srpc_list, &scd->scd_rpc_free);
959 }
960
961 spin_unlock(&scd->scd_lock);
962 }
963
964 /* handles an incoming RPC */
965 int
966 srpc_handle_rpc(swi_workitem_t *wi)
967 {
968 struct srpc_server_rpc *rpc = wi->swi_workitem.wi_data;
969 struct srpc_service_cd *scd = rpc->srpc_scd;
970 struct srpc_service *sv = scd->scd_svc;
971 srpc_event_t *ev = &rpc->srpc_ev;
972 int rc = 0;
973
974 LASSERT(wi == &rpc->srpc_wi);
975
976 spin_lock(&scd->scd_lock);
977
978 if (sv->sv_shuttingdown || rpc->srpc_aborted) {
979 spin_unlock(&scd->scd_lock);
980
981 if (rpc->srpc_bulk)
982 LNetMDUnlink(rpc->srpc_bulk->bk_mdh);
983 LNetMDUnlink(rpc->srpc_replymdh);
984
985 if (ev->ev_fired) { /* no more event, OK to finish */
986 srpc_server_rpc_done(rpc, -ESHUTDOWN);
987 return 1;
988 }
989 return 0;
990 }
991
992 spin_unlock(&scd->scd_lock);
993
994 switch (wi->swi_state) {
995 default:
996 LBUG();
997 case SWI_STATE_NEWBORN: {
998 srpc_msg_t *msg;
999 srpc_generic_reply_t *reply;
1000
1001 msg = &rpc->srpc_reqstbuf->buf_msg;
1002 reply = &rpc->srpc_replymsg.msg_body.reply;
1003
1004 if (!msg->msg_magic) {
1005 /* moaned already in srpc_lnet_ev_handler */
1006 srpc_server_rpc_done(rpc, EBADMSG);
1007 return 1;
1008 }
1009
1010 srpc_unpack_msg_hdr(msg);
1011 if (msg->msg_version != SRPC_MSG_VERSION) {
1012 CWARN("Version mismatch: %u, %u expected, from %s\n",
1013 msg->msg_version, SRPC_MSG_VERSION,
1014 libcfs_id2str(rpc->srpc_peer));
1015 reply->status = EPROTO;
1016 /* drop through and send reply */
1017 } else {
1018 reply->status = 0;
1019 rc = (*sv->sv_handler)(rpc);
1020 LASSERT(!reply->status || !rpc->srpc_bulk);
1021 if (rc) {
1022 srpc_server_rpc_done(rpc, rc);
1023 return 1;
1024 }
1025 }
1026
1027 wi->swi_state = SWI_STATE_BULK_STARTED;
1028
1029 if (rpc->srpc_bulk) {
1030 rc = srpc_do_bulk(rpc);
1031 if (!rc)
1032 return 0; /* wait for bulk */
1033
1034 LASSERT(ev->ev_fired);
1035 ev->ev_status = rc;
1036 }
1037 }
1038 case SWI_STATE_BULK_STARTED:
1039 LASSERT(!rpc->srpc_bulk || ev->ev_fired);
1040
1041 if (rpc->srpc_bulk) {
1042 rc = ev->ev_status;
1043
1044 if (sv->sv_bulk_ready)
1045 rc = (*sv->sv_bulk_ready) (rpc, rc);
1046
1047 if (rc) {
1048 srpc_server_rpc_done(rpc, rc);
1049 return 1;
1050 }
1051 }
1052
1053 wi->swi_state = SWI_STATE_REPLY_SUBMITTED;
1054 rc = srpc_send_reply(rpc);
1055 if (!rc)
1056 return 0; /* wait for reply */
1057 srpc_server_rpc_done(rpc, rc);
1058 return 1;
1059
1060 case SWI_STATE_REPLY_SUBMITTED:
1061 if (!ev->ev_fired) {
1062 CERROR("RPC %p: bulk %p, service %d\n",
1063 rpc, rpc->srpc_bulk, sv->sv_id);
1064 CERROR("Event: status %d, type %d, lnet %d\n",
1065 ev->ev_status, ev->ev_type, ev->ev_lnet);
1066 LASSERT(ev->ev_fired);
1067 }
1068
1069 wi->swi_state = SWI_STATE_DONE;
1070 srpc_server_rpc_done(rpc, ev->ev_status);
1071 return 1;
1072 }
1073
1074 return 0;
1075 }
1076
1077 static void
1078 srpc_client_rpc_expired(void *data)
1079 {
1080 srpc_client_rpc_t *rpc = data;
1081
1082 CWARN("Client RPC expired: service %d, peer %s, timeout %d.\n",
1083 rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
1084 rpc->crpc_timeout);
1085
1086 spin_lock(&rpc->crpc_lock);
1087
1088 rpc->crpc_timeout = 0;
1089 srpc_abort_rpc(rpc, -ETIMEDOUT);
1090
1091 spin_unlock(&rpc->crpc_lock);
1092
1093 spin_lock(&srpc_data.rpc_glock);
1094 srpc_data.rpc_counters.rpcs_expired++;
1095 spin_unlock(&srpc_data.rpc_glock);
1096 }
1097
1098 static void
1099 srpc_add_client_rpc_timer(srpc_client_rpc_t *rpc)
1100 {
1101 struct stt_timer *timer = &rpc->crpc_timer;
1102
1103 if (!rpc->crpc_timeout)
1104 return;
1105
1106 INIT_LIST_HEAD(&timer->stt_list);
1107 timer->stt_data = rpc;
1108 timer->stt_func = srpc_client_rpc_expired;
1109 timer->stt_expires = ktime_get_real_seconds() + rpc->crpc_timeout;
1110 stt_add_timer(timer);
1111 }
1112
1113 /*
1114 * Called with rpc->crpc_lock held.
1115 *
1116 * Upon exit the RPC expiry timer is not queued and the handler is not
1117 * running on any CPU.
1118 */
1119 static void
1120 srpc_del_client_rpc_timer(srpc_client_rpc_t *rpc)
1121 {
1122 /* timer not planted or already exploded */
1123 if (!rpc->crpc_timeout)
1124 return;
1125
1126 /* timer successfully defused */
1127 if (stt_del_timer(&rpc->crpc_timer))
1128 return;
1129
1130 /* timer detonated, wait for it to explode */
1131 while (rpc->crpc_timeout) {
1132 spin_unlock(&rpc->crpc_lock);
1133
1134 schedule();
1135
1136 spin_lock(&rpc->crpc_lock);
1137 }
1138 }
1139
1140 static void
1141 srpc_client_rpc_done(srpc_client_rpc_t *rpc, int status)
1142 {
1143 swi_workitem_t *wi = &rpc->crpc_wi;
1144
1145 LASSERT(status || wi->swi_state == SWI_STATE_DONE);
1146
1147 spin_lock(&rpc->crpc_lock);
1148
1149 rpc->crpc_closed = 1;
1150 if (!rpc->crpc_status)
1151 rpc->crpc_status = status;
1152
1153 srpc_del_client_rpc_timer(rpc);
1154
1155 CDEBUG_LIMIT(!status ? D_NET : D_NETERROR,
1156 "Client RPC done: service %d, peer %s, status %s:%d:%d\n",
1157 rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
1158 swi_state2str(wi->swi_state), rpc->crpc_aborted, status);
1159
1160 /*
1161 * No one can schedule me now since:
1162 * - RPC timer has been defused.
1163 * - all LNet events have been fired.
1164 * - crpc_closed has been set, preventing srpc_abort_rpc from
1165 * scheduling me.
1166 * Cancel pending schedules and prevent future schedule attempts:
1167 */
1168 LASSERT(!srpc_event_pending(rpc));
1169 swi_exit_workitem(wi);
1170
1171 spin_unlock(&rpc->crpc_lock);
1172
1173 (*rpc->crpc_done)(rpc);
1174 }
1175
1176 /* sends an outgoing RPC */
1177 int
1178 srpc_send_rpc(swi_workitem_t *wi)
1179 {
1180 int rc = 0;
1181 srpc_client_rpc_t *rpc;
1182 srpc_msg_t *reply;
1183 int do_bulk;
1184
1185 LASSERT(wi);
1186
1187 rpc = wi->swi_workitem.wi_data;
1188
1189 LASSERT(rpc);
1190 LASSERT(wi == &rpc->crpc_wi);
1191
1192 reply = &rpc->crpc_replymsg;
1193 do_bulk = rpc->crpc_bulk.bk_niov > 0;
1194
1195 spin_lock(&rpc->crpc_lock);
1196
1197 if (rpc->crpc_aborted) {
1198 spin_unlock(&rpc->crpc_lock);
1199 goto abort;
1200 }
1201
1202 spin_unlock(&rpc->crpc_lock);
1203
1204 switch (wi->swi_state) {
1205 default:
1206 LBUG();
1207 case SWI_STATE_NEWBORN:
1208 LASSERT(!srpc_event_pending(rpc));
1209
1210 rc = srpc_prepare_reply(rpc);
1211 if (rc) {
1212 srpc_client_rpc_done(rpc, rc);
1213 return 1;
1214 }
1215
1216 rc = srpc_prepare_bulk(rpc);
1217 if (rc)
1218 break;
1219
1220 wi->swi_state = SWI_STATE_REQUEST_SUBMITTED;
1221 rc = srpc_send_request(rpc);
1222 break;
1223
1224 case SWI_STATE_REQUEST_SUBMITTED:
1225 /*
1226 * CAVEAT EMPTOR: rqtev, rpyev, and bulkev may come in any
1227 * order; however, they're processed in a strict order:
1228 * rqt, rpy, and bulk.
1229 */
1230 if (!rpc->crpc_reqstev.ev_fired)
1231 break;
1232
1233 rc = rpc->crpc_reqstev.ev_status;
1234 if (rc)
1235 break;
1236
1237 wi->swi_state = SWI_STATE_REQUEST_SENT;
1238 /* perhaps more events, fall thru */
1239 case SWI_STATE_REQUEST_SENT: {
1240 srpc_msg_type_t type = srpc_service2reply(rpc->crpc_service);
1241
1242 if (!rpc->crpc_replyev.ev_fired)
1243 break;
1244
1245 rc = rpc->crpc_replyev.ev_status;
1246 if (rc)
1247 break;
1248
1249 srpc_unpack_msg_hdr(reply);
1250 if (reply->msg_type != type ||
1251 (reply->msg_magic != SRPC_MSG_MAGIC &&
1252 reply->msg_magic != __swab32(SRPC_MSG_MAGIC))) {
1253 CWARN("Bad message from %s: type %u (%d expected), magic %u (%d expected).\n",
1254 libcfs_id2str(rpc->crpc_dest),
1255 reply->msg_type, type,
1256 reply->msg_magic, SRPC_MSG_MAGIC);
1257 rc = -EBADMSG;
1258 break;
1259 }
1260
1261 if (do_bulk && reply->msg_body.reply.status) {
1262 CWARN("Remote error %d at %s, unlink bulk buffer in case peer didn't initiate bulk transfer\n",
1263 reply->msg_body.reply.status,
1264 libcfs_id2str(rpc->crpc_dest));
1265 LNetMDUnlink(rpc->crpc_bulk.bk_mdh);
1266 }
1267
1268 wi->swi_state = SWI_STATE_REPLY_RECEIVED;
1269 }
1270 case SWI_STATE_REPLY_RECEIVED:
1271 if (do_bulk && !rpc->crpc_bulkev.ev_fired)
1272 break;
1273
1274 rc = do_bulk ? rpc->crpc_bulkev.ev_status : 0;
1275
1276 /*
1277 * Bulk buffer was unlinked due to remote error. Clear error
1278 * since reply buffer still contains valid data.
1279 * NB rpc->crpc_done shouldn't look into bulk data in case of
1280 * remote error.
1281 */
1282 if (do_bulk && rpc->crpc_bulkev.ev_lnet == LNET_EVENT_UNLINK &&
1283 !rpc->crpc_status && reply->msg_body.reply.status)
1284 rc = 0;
1285
1286 wi->swi_state = SWI_STATE_DONE;
1287 srpc_client_rpc_done(rpc, rc);
1288 return 1;
1289 }
1290
1291 if (rc) {
1292 spin_lock(&rpc->crpc_lock);
1293 srpc_abort_rpc(rpc, rc);
1294 spin_unlock(&rpc->crpc_lock);
1295 }
1296
1297 abort:
1298 if (rpc->crpc_aborted) {
1299 LNetMDUnlink(rpc->crpc_reqstmdh);
1300 LNetMDUnlink(rpc->crpc_replymdh);
1301 LNetMDUnlink(rpc->crpc_bulk.bk_mdh);
1302
1303 if (!srpc_event_pending(rpc)) {
1304 srpc_client_rpc_done(rpc, -EINTR);
1305 return 1;
1306 }
1307 }
1308 return 0;
1309 }
1310
1311 srpc_client_rpc_t *
1312 srpc_create_client_rpc(lnet_process_id_t peer, int service,
1313 int nbulkiov, int bulklen,
1314 void (*rpc_done)(srpc_client_rpc_t *),
1315 void (*rpc_fini)(srpc_client_rpc_t *), void *priv)
1316 {
1317 srpc_client_rpc_t *rpc;
1318
1319 LIBCFS_ALLOC(rpc, offsetof(srpc_client_rpc_t,
1320 crpc_bulk.bk_iovs[nbulkiov]));
1321 if (!rpc)
1322 return NULL;
1323
1324 srpc_init_client_rpc(rpc, peer, service, nbulkiov,
1325 bulklen, rpc_done, rpc_fini, priv);
1326 return rpc;
1327 }
1328
1329 /* called with rpc->crpc_lock held */
1330 void
1331 srpc_abort_rpc(srpc_client_rpc_t *rpc, int why)
1332 {
1333 LASSERT(why);
1334
1335 if (rpc->crpc_aborted || /* already aborted */
1336 rpc->crpc_closed) /* callback imminent */
1337 return;
1338
1339 CDEBUG(D_NET, "Aborting RPC: service %d, peer %s, state %s, why %d\n",
1340 rpc->crpc_service, libcfs_id2str(rpc->crpc_dest),
1341 swi_state2str(rpc->crpc_wi.swi_state), why);
1342
1343 rpc->crpc_aborted = 1;
1344 rpc->crpc_status = why;
1345 swi_schedule_workitem(&rpc->crpc_wi);
1346 }
1347
1348 /* called with rpc->crpc_lock held */
1349 void
1350 srpc_post_rpc(srpc_client_rpc_t *rpc)
1351 {
1352 LASSERT(!rpc->crpc_aborted);
1353 LASSERT(srpc_data.rpc_state == SRPC_STATE_RUNNING);
1354
1355 CDEBUG(D_NET, "Posting RPC: peer %s, service %d, timeout %d\n",
1356 libcfs_id2str(rpc->crpc_dest), rpc->crpc_service,
1357 rpc->crpc_timeout);
1358
1359 srpc_add_client_rpc_timer(rpc);
1360 swi_schedule_workitem(&rpc->crpc_wi);
1361 }
1362
1363 int
1364 srpc_send_reply(struct srpc_server_rpc *rpc)
1365 {
1366 srpc_event_t *ev = &rpc->srpc_ev;
1367 struct srpc_msg *msg = &rpc->srpc_replymsg;
1368 struct srpc_buffer *buffer = rpc->srpc_reqstbuf;
1369 struct srpc_service_cd *scd = rpc->srpc_scd;
1370 struct srpc_service *sv = scd->scd_svc;
1371 __u64 rpyid;
1372 int rc;
1373
1374 LASSERT(buffer);
1375 rpyid = buffer->buf_msg.msg_body.reqst.rpyid;
1376
1377 spin_lock(&scd->scd_lock);
1378
1379 if (!sv->sv_shuttingdown && !srpc_serv_is_framework(sv)) {
1380 /*
1381 * Repost buffer before replying since test client
1382 * might send me another RPC once it gets the reply
1383 */
1384 if (srpc_service_post_buffer(scd, buffer))
1385 CWARN("Failed to repost %s buffer\n", sv->sv_name);
1386 rpc->srpc_reqstbuf = NULL;
1387 }
1388
1389 spin_unlock(&scd->scd_lock);
1390
1391 ev->ev_fired = 0;
1392 ev->ev_data = rpc;
1393 ev->ev_type = SRPC_REPLY_SENT;
1394
1395 msg->msg_magic = SRPC_MSG_MAGIC;
1396 msg->msg_version = SRPC_MSG_VERSION;
1397 msg->msg_type = srpc_service2reply(sv->sv_id);
1398
1399 rc = srpc_post_active_rdma(SRPC_RDMA_PORTAL, rpyid, msg,
1400 sizeof(*msg), LNET_MD_OP_PUT,
1401 rpc->srpc_peer, rpc->srpc_self,
1402 &rpc->srpc_replymdh, ev);
1403 if (rc)
1404 ev->ev_fired = 1; /* no more event expected */
1405 return rc;
1406 }
1407
1408 /* when in kernel always called with LNET_LOCK() held, and in thread context */
1409 static void
1410 srpc_lnet_ev_handler(lnet_event_t *ev)
1411 {
1412 struct srpc_service_cd *scd;
1413 srpc_event_t *rpcev = ev->md.user_ptr;
1414 srpc_client_rpc_t *crpc;
1415 struct srpc_server_rpc *srpc;
1416 srpc_buffer_t *buffer;
1417 srpc_service_t *sv;
1418 srpc_msg_t *msg;
1419 srpc_msg_type_t type;
1420
1421 LASSERT(!in_interrupt());
1422
1423 if (ev->status) {
1424 __u32 errors;
1425
1426 spin_lock(&srpc_data.rpc_glock);
1427 if (ev->status != -ECANCELED) /* cancellation is not error */
1428 srpc_data.rpc_counters.errors++;
1429 errors = srpc_data.rpc_counters.errors;
1430 spin_unlock(&srpc_data.rpc_glock);
1431
1432 CNETERR("LNet event status %d type %d, RPC errors %u\n",
1433 ev->status, ev->type, errors);
1434 }
1435
1436 rpcev->ev_lnet = ev->type;
1437
1438 switch (rpcev->ev_type) {
1439 default:
1440 CERROR("Unknown event: status %d, type %d, lnet %d\n",
1441 rpcev->ev_status, rpcev->ev_type, rpcev->ev_lnet);
1442 LBUG();
1443 case SRPC_REQUEST_SENT:
1444 if (!ev->status && ev->type != LNET_EVENT_UNLINK) {
1445 spin_lock(&srpc_data.rpc_glock);
1446 srpc_data.rpc_counters.rpcs_sent++;
1447 spin_unlock(&srpc_data.rpc_glock);
1448 }
1449 case SRPC_REPLY_RCVD:
1450 case SRPC_BULK_REQ_RCVD:
1451 crpc = rpcev->ev_data;
1452
1453 if (rpcev != &crpc->crpc_reqstev &&
1454 rpcev != &crpc->crpc_replyev &&
1455 rpcev != &crpc->crpc_bulkev) {
1456 CERROR("rpcev %p, crpc %p, reqstev %p, replyev %p, bulkev %p\n",
1457 rpcev, crpc, &crpc->crpc_reqstev,
1458 &crpc->crpc_replyev, &crpc->crpc_bulkev);
1459 CERROR("Bad event: status %d, type %d, lnet %d\n",
1460 rpcev->ev_status, rpcev->ev_type, rpcev->ev_lnet);
1461 LBUG();
1462 }
1463
1464 spin_lock(&crpc->crpc_lock);
1465
1466 LASSERT(!rpcev->ev_fired);
1467 rpcev->ev_fired = 1;
1468 rpcev->ev_status = (ev->type == LNET_EVENT_UNLINK) ?
1469 -EINTR : ev->status;
1470 swi_schedule_workitem(&crpc->crpc_wi);
1471
1472 spin_unlock(&crpc->crpc_lock);
1473 break;
1474
1475 case SRPC_REQUEST_RCVD:
1476 scd = rpcev->ev_data;
1477 sv = scd->scd_svc;
1478
1479 LASSERT(rpcev == &scd->scd_ev);
1480
1481 spin_lock(&scd->scd_lock);
1482
1483 LASSERT(ev->unlinked);
1484 LASSERT(ev->type == LNET_EVENT_PUT ||
1485 ev->type == LNET_EVENT_UNLINK);
1486 LASSERT(ev->type != LNET_EVENT_UNLINK ||
1487 sv->sv_shuttingdown);
1488
1489 buffer = container_of(ev->md.start, srpc_buffer_t, buf_msg);
1490 buffer->buf_peer = ev->initiator;
1491 buffer->buf_self = ev->target.nid;
1492
1493 LASSERT(scd->scd_buf_nposted > 0);
1494 scd->scd_buf_nposted--;
1495
1496 if (sv->sv_shuttingdown) {
1497 /*
1498 * Leave buffer on scd->scd_buf_nposted since
1499 * srpc_finish_service needs to traverse it.
1500 */
1501 spin_unlock(&scd->scd_lock);
1502 break;
1503 }
1504
1505 if (scd->scd_buf_err_stamp &&
1506 scd->scd_buf_err_stamp < ktime_get_real_seconds()) {
1507 /* re-enable adding buffer */
1508 scd->scd_buf_err_stamp = 0;
1509 scd->scd_buf_err = 0;
1510 }
1511
1512 if (!scd->scd_buf_err && /* adding buffer is enabled */
1513 !scd->scd_buf_adjust &&
1514 scd->scd_buf_nposted < scd->scd_buf_low) {
1515 scd->scd_buf_adjust = max(scd->scd_buf_total / 2,
1516 SFW_TEST_WI_MIN);
1517 swi_schedule_workitem(&scd->scd_buf_wi);
1518 }
1519
1520 list_del(&buffer->buf_list); /* from scd->scd_buf_posted */
1521 msg = &buffer->buf_msg;
1522 type = srpc_service2request(sv->sv_id);
1523
1524 if (ev->status || ev->mlength != sizeof(*msg) ||
1525 (msg->msg_type != type &&
1526 msg->msg_type != __swab32(type)) ||
1527 (msg->msg_magic != SRPC_MSG_MAGIC &&
1528 msg->msg_magic != __swab32(SRPC_MSG_MAGIC))) {
1529 CERROR("Dropping RPC (%s) from %s: status %d mlength %d type %u magic %u.\n",
1530 sv->sv_name, libcfs_id2str(ev->initiator),
1531 ev->status, ev->mlength,
1532 msg->msg_type, msg->msg_magic);
1533
1534 /*
1535 * NB can't call srpc_service_recycle_buffer here since
1536 * it may call LNetM[DE]Attach. The invalid magic tells
1537 * srpc_handle_rpc to drop this RPC
1538 */
1539 msg->msg_magic = 0;
1540 }
1541
1542 if (!list_empty(&scd->scd_rpc_free)) {
1543 srpc = list_entry(scd->scd_rpc_free.next,
1544 struct srpc_server_rpc,
1545 srpc_list);
1546 list_del(&srpc->srpc_list);
1547
1548 srpc_init_server_rpc(srpc, scd, buffer);
1549 list_add_tail(&srpc->srpc_list,
1550 &scd->scd_rpc_active);
1551 swi_schedule_workitem(&srpc->srpc_wi);
1552 } else {
1553 list_add_tail(&buffer->buf_list,
1554 &scd->scd_buf_blocked);
1555 }
1556
1557 spin_unlock(&scd->scd_lock);
1558
1559 spin_lock(&srpc_data.rpc_glock);
1560 srpc_data.rpc_counters.rpcs_rcvd++;
1561 spin_unlock(&srpc_data.rpc_glock);
1562 break;
1563
1564 case SRPC_BULK_GET_RPLD:
1565 LASSERT(ev->type == LNET_EVENT_SEND ||
1566 ev->type == LNET_EVENT_REPLY ||
1567 ev->type == LNET_EVENT_UNLINK);
1568
1569 if (!ev->unlinked)
1570 break; /* wait for final event */
1571
1572 case SRPC_BULK_PUT_SENT:
1573 if (!ev->status && ev->type != LNET_EVENT_UNLINK) {
1574 spin_lock(&srpc_data.rpc_glock);
1575
1576 if (rpcev->ev_type == SRPC_BULK_GET_RPLD)
1577 srpc_data.rpc_counters.bulk_get += ev->mlength;
1578 else
1579 srpc_data.rpc_counters.bulk_put += ev->mlength;
1580
1581 spin_unlock(&srpc_data.rpc_glock);
1582 }
1583 case SRPC_REPLY_SENT:
1584 srpc = rpcev->ev_data;
1585 scd = srpc->srpc_scd;
1586
1587 LASSERT(rpcev == &srpc->srpc_ev);
1588
1589 spin_lock(&scd->scd_lock);
1590
1591 rpcev->ev_fired = 1;
1592 rpcev->ev_status = (ev->type == LNET_EVENT_UNLINK) ?
1593 -EINTR : ev->status;
1594 swi_schedule_workitem(&srpc->srpc_wi);
1595
1596 spin_unlock(&scd->scd_lock);
1597 break;
1598 }
1599 }
1600
1601 int
1602 srpc_startup(void)
1603 {
1604 int rc;
1605
1606 memset(&srpc_data, 0, sizeof(struct smoketest_rpc));
1607 spin_lock_init(&srpc_data.rpc_glock);
1608
1609 /* 1 second pause to avoid timestamp reuse */
1610 set_current_state(TASK_UNINTERRUPTIBLE);
1611 schedule_timeout(cfs_time_seconds(1));
1612 srpc_data.rpc_matchbits = ((__u64)ktime_get_real_seconds()) << 48;
1613
1614 srpc_data.rpc_state = SRPC_STATE_NONE;
1615
1616 rc = LNetNIInit(LNET_PID_LUSTRE);
1617 if (rc < 0) {
1618 CERROR("LNetNIInit() has failed: %d\n", rc);
1619 return rc;
1620 }
1621
1622 srpc_data.rpc_state = SRPC_STATE_NI_INIT;
1623
1624 LNetInvalidateHandle(&srpc_data.rpc_lnet_eq);
1625 rc = LNetEQAlloc(0, srpc_lnet_ev_handler, &srpc_data.rpc_lnet_eq);
1626 if (rc) {
1627 CERROR("LNetEQAlloc() has failed: %d\n", rc);
1628 goto bail;
1629 }
1630
1631 rc = LNetSetLazyPortal(SRPC_FRAMEWORK_REQUEST_PORTAL);
1632 LASSERT(!rc);
1633 rc = LNetSetLazyPortal(SRPC_REQUEST_PORTAL);
1634 LASSERT(!rc);
1635
1636 srpc_data.rpc_state = SRPC_STATE_EQ_INIT;
1637
1638 rc = stt_startup();
1639
1640 bail:
1641 if (rc)
1642 srpc_shutdown();
1643 else
1644 srpc_data.rpc_state = SRPC_STATE_RUNNING;
1645
1646 return rc;
1647 }
1648
1649 void
1650 srpc_shutdown(void)
1651 {
1652 int i;
1653 int rc;
1654 int state;
1655
1656 state = srpc_data.rpc_state;
1657 srpc_data.rpc_state = SRPC_STATE_STOPPING;
1658
1659 switch (state) {
1660 default:
1661 LBUG();
1662 case SRPC_STATE_RUNNING:
1663 spin_lock(&srpc_data.rpc_glock);
1664
1665 for (i = 0; i <= SRPC_SERVICE_MAX_ID; i++) {
1666 srpc_service_t *sv = srpc_data.rpc_services[i];
1667
1668 LASSERTF(!sv, "service not empty: id %d, name %s\n",
1669 i, sv->sv_name);
1670 }
1671
1672 spin_unlock(&srpc_data.rpc_glock);
1673
1674 stt_shutdown();
1675
1676 case SRPC_STATE_EQ_INIT:
1677 rc = LNetClearLazyPortal(SRPC_FRAMEWORK_REQUEST_PORTAL);
1678 rc = LNetClearLazyPortal(SRPC_REQUEST_PORTAL);
1679 LASSERT(!rc);
1680 rc = LNetEQFree(srpc_data.rpc_lnet_eq);
1681 LASSERT(!rc); /* the EQ should have no user by now */
1682
1683 case SRPC_STATE_NI_INIT:
1684 LNetNIFini();
1685 }
1686 }
Linux kernel - Deliverables
This page took 0.064677 seconds and 5 git commands to generate.