4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
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.
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).
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.gnu.org/licenses/gpl-2.0.html
23 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Use is subject to license terms.
26 * Copyright (c) 2011, 2017, Intel Corporation.
29 * This file is part of Lustre, http://www.lustre.org/
30 * Lustre is a trademark of Sun Microsystems, Inc.
33 #define DEBUG_SUBSYSTEM S_OSC
35 #include <linux/workqueue.h>
36 #include <libcfs/libcfs.h>
37 #include <linux/falloc.h>
38 #include <lprocfs_status.h>
39 #include <lustre_debug.h>
40 #include <lustre_dlm.h>
41 #include <lustre_fid.h>
42 #include <lustre_ha.h>
43 #include <uapi/linux/lustre/lustre_ioctl.h>
44 #include <lustre_net.h>
45 #include <lustre_obdo.h>
47 #include <obd_cksum.h>
48 #include <obd_class.h>
49 #include <lustre_osc.h>
50 #include <linux/falloc.h>
52 #include "osc_internal.h"
54 atomic_t osc_pool_req_count;
55 unsigned int osc_reqpool_maxreqcount;
56 struct ptlrpc_request_pool *osc_rq_pool;
58 /* max memory used for request pool, unit is MB */
59 static unsigned int osc_reqpool_mem_max = 5;
60 module_param(osc_reqpool_mem_max, uint, 0444);
62 static int osc_idle_timeout = 20;
63 module_param(osc_idle_timeout, uint, 0644);
65 #define osc_grant_args osc_brw_async_args
67 struct osc_setattr_args {
69 obd_enqueue_update_f sa_upcall;
73 struct osc_fsync_args {
74 struct osc_object *fa_obj;
76 obd_enqueue_update_f fa_upcall;
80 struct osc_ladvise_args {
82 obd_enqueue_update_f la_upcall;
86 static void osc_release_ppga(struct brw_page **ppga, size_t count);
87 static int brw_interpret(const struct lu_env *env, struct ptlrpc_request *req,
90 void osc_pack_req_body(struct ptlrpc_request *req, struct obdo *oa)
92 struct ost_body *body;
94 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
97 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
100 static int osc_getattr(const struct lu_env *env, struct obd_export *exp,
103 struct ptlrpc_request *req;
104 struct ost_body *body;
108 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
112 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
114 ptlrpc_request_free(req);
118 osc_pack_req_body(req, oa);
120 ptlrpc_request_set_replen(req);
122 rc = ptlrpc_queue_wait(req);
126 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
128 GOTO(out, rc = -EPROTO);
130 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
131 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
133 oa->o_blksize = cli_brw_size(exp->exp_obd);
134 oa->o_valid |= OBD_MD_FLBLKSZ;
138 ptlrpc_req_finished(req);
143 static int osc_setattr(const struct lu_env *env, struct obd_export *exp,
146 struct ptlrpc_request *req;
147 struct ost_body *body;
151 LASSERT(oa->o_valid & OBD_MD_FLGROUP);
153 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
157 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
159 ptlrpc_request_free(req);
163 osc_pack_req_body(req, oa);
165 ptlrpc_request_set_replen(req);
167 rc = ptlrpc_queue_wait(req);
171 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
173 GOTO(out, rc = -EPROTO);
175 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
179 ptlrpc_req_finished(req);
184 static int osc_setattr_interpret(const struct lu_env *env,
185 struct ptlrpc_request *req, void *args, int rc)
187 struct osc_setattr_args *sa = args;
188 struct ost_body *body;
195 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
197 GOTO(out, rc = -EPROTO);
199 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, sa->sa_oa,
202 rc = sa->sa_upcall(sa->sa_cookie, rc);
206 int osc_setattr_async(struct obd_export *exp, struct obdo *oa,
207 obd_enqueue_update_f upcall, void *cookie,
208 struct ptlrpc_request_set *rqset)
210 struct ptlrpc_request *req;
211 struct osc_setattr_args *sa;
216 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
220 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
222 ptlrpc_request_free(req);
226 osc_pack_req_body(req, oa);
228 ptlrpc_request_set_replen(req);
230 /* do mds to ost setattr asynchronously */
232 /* Do not wait for response. */
233 ptlrpcd_add_req(req);
235 req->rq_interpret_reply = osc_setattr_interpret;
237 sa = ptlrpc_req_async_args(sa, req);
239 sa->sa_upcall = upcall;
240 sa->sa_cookie = cookie;
242 ptlrpc_set_add_req(rqset, req);
248 static int osc_ladvise_interpret(const struct lu_env *env,
249 struct ptlrpc_request *req,
252 struct osc_ladvise_args *la = arg;
253 struct ost_body *body;
259 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
261 GOTO(out, rc = -EPROTO);
263 *la->la_oa = body->oa;
265 rc = la->la_upcall(la->la_cookie, rc);
270 * If rqset is NULL, do not wait for response. Upcall and cookie could also
271 * be NULL in this case
273 int osc_ladvise_base(struct obd_export *exp, struct obdo *oa,
274 struct ladvise_hdr *ladvise_hdr,
275 obd_enqueue_update_f upcall, void *cookie,
276 struct ptlrpc_request_set *rqset)
278 struct ptlrpc_request *req;
279 struct ost_body *body;
280 struct osc_ladvise_args *la;
282 struct lu_ladvise *req_ladvise;
283 struct lu_ladvise *ladvise = ladvise_hdr->lah_advise;
284 int num_advise = ladvise_hdr->lah_count;
285 struct ladvise_hdr *req_ladvise_hdr;
288 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_LADVISE);
292 req_capsule_set_size(&req->rq_pill, &RMF_OST_LADVISE, RCL_CLIENT,
293 num_advise * sizeof(*ladvise));
294 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_LADVISE);
296 ptlrpc_request_free(req);
299 req->rq_request_portal = OST_IO_PORTAL;
300 ptlrpc_at_set_req_timeout(req);
302 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
304 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa,
307 req_ladvise_hdr = req_capsule_client_get(&req->rq_pill,
308 &RMF_OST_LADVISE_HDR);
309 memcpy(req_ladvise_hdr, ladvise_hdr, sizeof(*ladvise_hdr));
311 req_ladvise = req_capsule_client_get(&req->rq_pill, &RMF_OST_LADVISE);
312 memcpy(req_ladvise, ladvise, sizeof(*ladvise) * num_advise);
313 ptlrpc_request_set_replen(req);
316 /* Do not wait for response. */
317 ptlrpcd_add_req(req);
321 req->rq_interpret_reply = osc_ladvise_interpret;
322 la = ptlrpc_req_async_args(la, req);
324 la->la_upcall = upcall;
325 la->la_cookie = cookie;
327 ptlrpc_set_add_req(rqset, req);
332 static int osc_create(const struct lu_env *env, struct obd_export *exp,
335 struct ptlrpc_request *req;
336 struct ost_body *body;
341 LASSERT(oa->o_valid & OBD_MD_FLGROUP);
342 LASSERT(fid_seq_is_echo(ostid_seq(&oa->o_oi)));
344 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
346 GOTO(out, rc = -ENOMEM);
348 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
350 ptlrpc_request_free(req);
354 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
357 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
359 ptlrpc_request_set_replen(req);
361 rc = ptlrpc_queue_wait(req);
365 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
367 GOTO(out_req, rc = -EPROTO);
369 CDEBUG(D_INFO, "oa flags %x\n", oa->o_flags);
370 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
372 oa->o_blksize = cli_brw_size(exp->exp_obd);
373 oa->o_valid |= OBD_MD_FLBLKSZ;
375 CDEBUG(D_HA, "transno: %lld\n",
376 lustre_msg_get_transno(req->rq_repmsg));
378 ptlrpc_req_finished(req);
383 int osc_punch_send(struct obd_export *exp, struct obdo *oa,
384 obd_enqueue_update_f upcall, void *cookie)
386 struct ptlrpc_request *req;
387 struct osc_setattr_args *sa;
388 struct obd_import *imp = class_exp2cliimp(exp);
389 struct ost_body *body;
394 req = ptlrpc_request_alloc(imp, &RQF_OST_PUNCH);
398 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
400 ptlrpc_request_free(req);
404 osc_set_io_portal(req);
406 ptlrpc_at_set_req_timeout(req);
408 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
410 lustre_set_wire_obdo(&imp->imp_connect_data, &body->oa, oa);
412 ptlrpc_request_set_replen(req);
414 req->rq_interpret_reply = osc_setattr_interpret;
415 sa = ptlrpc_req_async_args(sa, req);
417 sa->sa_upcall = upcall;
418 sa->sa_cookie = cookie;
420 ptlrpcd_add_req(req);
424 EXPORT_SYMBOL(osc_punch_send);
427 * osc_fallocate_base() - Handles fallocate request.
429 * @exp: Export structure
430 * @oa: Attributes passed to OSS from client (obdo structure)
431 * @upcall: Primary & supplementary group information
432 * @cookie: Exclusive identifier
433 * @rqset: Request list.
434 * @mode: Operation done on given range.
436 * osc_fallocate_base() - Handles fallocate requests only. Only block
437 * allocation or standard preallocate operation is supported currently.
438 * Other mode flags is not supported yet. ftruncate(2) or truncate(2)
439 * is supported via SETATTR request.
441 * Return: Non-zero on failure and O on success.
443 int osc_fallocate_base(struct obd_export *exp, struct obdo *oa,
444 obd_enqueue_update_f upcall, void *cookie, int mode)
446 struct ptlrpc_request *req;
447 struct osc_setattr_args *sa;
448 struct ost_body *body;
449 struct obd_import *imp = class_exp2cliimp(exp);
454 * Only mode == 0 (which is standard prealloc) is supported now.
455 * Punch is not supported yet.
457 if (mode & ~FALLOC_FL_KEEP_SIZE)
459 oa->o_falloc_mode = mode;
461 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
466 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_FALLOCATE);
468 ptlrpc_request_free(req);
472 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
475 lustre_set_wire_obdo(&imp->imp_connect_data, &body->oa, oa);
477 ptlrpc_request_set_replen(req);
479 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_setattr_interpret;
480 BUILD_BUG_ON(sizeof(*sa) > sizeof(req->rq_async_args));
481 sa = ptlrpc_req_async_args(sa, req);
483 sa->sa_upcall = upcall;
484 sa->sa_cookie = cookie;
486 ptlrpcd_add_req(req);
491 static int osc_sync_interpret(const struct lu_env *env,
492 struct ptlrpc_request *req, void *args, int rc)
494 struct osc_fsync_args *fa = args;
495 struct ost_body *body;
496 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
497 unsigned long valid = 0;
498 struct cl_object *obj;
504 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
506 CERROR("can't unpack ost_body\n");
507 GOTO(out, rc = -EPROTO);
510 *fa->fa_oa = body->oa;
511 obj = osc2cl(fa->fa_obj);
513 /* Update osc object's blocks attribute */
514 cl_object_attr_lock(obj);
515 if (body->oa.o_valid & OBD_MD_FLBLOCKS) {
516 attr->cat_blocks = body->oa.o_blocks;
521 cl_object_attr_update(env, obj, attr, valid);
522 cl_object_attr_unlock(obj);
525 rc = fa->fa_upcall(fa->fa_cookie, rc);
529 int osc_sync_base(struct osc_object *obj, struct obdo *oa,
530 obd_enqueue_update_f upcall, void *cookie,
531 struct ptlrpc_request_set *rqset)
533 struct obd_export *exp = osc_export(obj);
534 struct ptlrpc_request *req;
535 struct ost_body *body;
536 struct osc_fsync_args *fa;
540 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
544 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
546 ptlrpc_request_free(req);
550 /* overload the size and blocks fields in the oa with start/end */
551 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
553 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
555 ptlrpc_request_set_replen(req);
556 req->rq_interpret_reply = osc_sync_interpret;
558 fa = ptlrpc_req_async_args(fa, req);
561 fa->fa_upcall = upcall;
562 fa->fa_cookie = cookie;
564 ptlrpc_set_add_req(rqset, req);
569 /* Find and cancel locally locks matched by @mode in the resource found by
570 * @objid. Found locks are added into @cancel list. Returns the amount of
571 * locks added to @cancels list. */
572 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
573 struct list_head *cancels,
574 enum ldlm_mode mode, __u64 lock_flags)
576 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
577 struct ldlm_res_id res_id;
578 struct ldlm_resource *res;
582 /* Return, i.e. cancel nothing, only if ELC is supported (flag in
583 * export) but disabled through procfs (flag in NS).
585 * This distinguishes from a case when ELC is not supported originally,
586 * when we still want to cancel locks in advance and just cancel them
587 * locally, without sending any RPC. */
588 if (exp_connect_cancelset(exp) && !ns_connect_cancelset(ns))
591 ostid_build_res_name(&oa->o_oi, &res_id);
592 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
596 LDLM_RESOURCE_ADDREF(res);
597 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
598 lock_flags, 0, NULL);
599 LDLM_RESOURCE_DELREF(res);
600 ldlm_resource_putref(res);
604 static int osc_destroy_interpret(const struct lu_env *env,
605 struct ptlrpc_request *req, void *args, int rc)
607 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
609 atomic_dec(&cli->cl_destroy_in_flight);
610 wake_up(&cli->cl_destroy_waitq);
615 static int osc_can_send_destroy(struct client_obd *cli)
617 if (atomic_inc_return(&cli->cl_destroy_in_flight) <=
618 cli->cl_max_rpcs_in_flight) {
619 /* The destroy request can be sent */
622 if (atomic_dec_return(&cli->cl_destroy_in_flight) <
623 cli->cl_max_rpcs_in_flight) {
625 * The counter has been modified between the two atomic
628 wake_up(&cli->cl_destroy_waitq);
633 static int osc_destroy(const struct lu_env *env, struct obd_export *exp,
636 struct client_obd *cli = &exp->exp_obd->u.cli;
637 struct ptlrpc_request *req;
638 struct ost_body *body;
644 CDEBUG(D_INFO, "oa NULL\n");
648 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
649 LDLM_FL_DISCARD_DATA);
651 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
653 ldlm_lock_list_put(&cancels, l_bl_ast, count);
657 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
660 ptlrpc_request_free(req);
664 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
665 ptlrpc_at_set_req_timeout(req);
667 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
669 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
671 ptlrpc_request_set_replen(req);
673 req->rq_interpret_reply = osc_destroy_interpret;
674 if (!osc_can_send_destroy(cli)) {
676 * Wait until the number of on-going destroy RPCs drops
677 * under max_rpc_in_flight
679 rc = l_wait_event_abortable_exclusive(
680 cli->cl_destroy_waitq,
681 osc_can_send_destroy(cli));
683 ptlrpc_req_finished(req);
688 /* Do not wait for response */
689 ptlrpcd_add_req(req);
693 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
696 u64 bits = OBD_MD_FLBLOCKS | OBD_MD_FLGRANT;
698 LASSERT(!(oa->o_valid & bits));
701 spin_lock(&cli->cl_loi_list_lock);
702 if (OCD_HAS_FLAG(&cli->cl_import->imp_connect_data, GRANT_PARAM))
703 oa->o_dirty = cli->cl_dirty_grant;
705 oa->o_dirty = cli->cl_dirty_pages << PAGE_SHIFT;
706 if (unlikely(cli->cl_dirty_pages > cli->cl_dirty_max_pages)) {
707 CERROR("dirty %lu > dirty_max %lu\n",
709 cli->cl_dirty_max_pages);
711 } else if (unlikely(atomic_long_read(&obd_dirty_pages) >
712 (long)(obd_max_dirty_pages + 1))) {
713 /* The atomic_read() allowing the atomic_inc() are
714 * not covered by a lock thus they may safely race and trip
715 * this CERROR() unless we add in a small fudge factor (+1). */
716 CERROR("%s: dirty %ld > system dirty_max %ld\n",
717 cli_name(cli), atomic_long_read(&obd_dirty_pages),
718 obd_max_dirty_pages);
720 } else if (unlikely(cli->cl_dirty_max_pages - cli->cl_dirty_pages >
722 CERROR("dirty %lu - dirty_max %lu too big???\n",
723 cli->cl_dirty_pages, cli->cl_dirty_max_pages);
726 unsigned long nrpages;
727 unsigned long undirty;
729 nrpages = cli->cl_max_pages_per_rpc;
730 nrpages *= cli->cl_max_rpcs_in_flight + 1;
731 nrpages = max(nrpages, cli->cl_dirty_max_pages);
732 undirty = nrpages << PAGE_SHIFT;
733 if (OCD_HAS_FLAG(&cli->cl_import->imp_connect_data,
737 /* take extent tax into account when asking for more
739 nrextents = (nrpages + cli->cl_max_extent_pages - 1) /
740 cli->cl_max_extent_pages;
741 undirty += nrextents * cli->cl_grant_extent_tax;
743 /* Do not ask for more than OBD_MAX_GRANT - a margin for server
744 * to add extent tax, etc.
746 oa->o_undirty = min(undirty, OBD_MAX_GRANT &
747 ~(PTLRPC_MAX_BRW_SIZE * 4UL));
749 oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
750 oa->o_dropped = cli->cl_lost_grant;
751 cli->cl_lost_grant = 0;
752 spin_unlock(&cli->cl_loi_list_lock);
753 CDEBUG(D_CACHE, "dirty: %llu undirty: %u dropped %u grant: %llu\n",
754 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
757 void osc_update_next_shrink(struct client_obd *cli)
759 cli->cl_next_shrink_grant = ktime_get_seconds() +
760 cli->cl_grant_shrink_interval;
762 CDEBUG(D_CACHE, "next time %lld to shrink grant\n",
763 cli->cl_next_shrink_grant);
766 static void __osc_update_grant(struct client_obd *cli, u64 grant)
768 spin_lock(&cli->cl_loi_list_lock);
769 cli->cl_avail_grant += grant;
770 spin_unlock(&cli->cl_loi_list_lock);
773 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
775 if (body->oa.o_valid & OBD_MD_FLGRANT) {
776 CDEBUG(D_CACHE, "got %llu extra grant\n", body->oa.o_grant);
777 __osc_update_grant(cli, body->oa.o_grant);
782 * grant thread data for shrinking space.
784 struct grant_thread_data {
785 struct list_head gtd_clients;
786 struct mutex gtd_mutex;
787 unsigned long gtd_stopped:1;
789 static struct grant_thread_data client_gtd;
791 static int osc_shrink_grant_interpret(const struct lu_env *env,
792 struct ptlrpc_request *req,
795 struct osc_grant_args *aa = args;
796 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
797 struct ost_body *body;
800 __osc_update_grant(cli, aa->aa_oa->o_grant);
804 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
806 osc_update_grant(cli, body);
808 OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
814 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
816 spin_lock(&cli->cl_loi_list_lock);
817 oa->o_grant = cli->cl_avail_grant / 4;
818 cli->cl_avail_grant -= oa->o_grant;
819 spin_unlock(&cli->cl_loi_list_lock);
820 if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
821 oa->o_valid |= OBD_MD_FLFLAGS;
824 oa->o_flags |= OBD_FL_SHRINK_GRANT;
825 osc_update_next_shrink(cli);
828 /* Shrink the current grant, either from some large amount to enough for a
829 * full set of in-flight RPCs, or if we have already shrunk to that limit
830 * then to enough for a single RPC. This avoids keeping more grant than
831 * needed, and avoids shrinking the grant piecemeal. */
832 static int osc_shrink_grant(struct client_obd *cli)
834 __u64 target_bytes = (cli->cl_max_rpcs_in_flight + 1) *
835 (cli->cl_max_pages_per_rpc << PAGE_SHIFT);
837 spin_lock(&cli->cl_loi_list_lock);
838 if (cli->cl_avail_grant <= target_bytes)
839 target_bytes = cli->cl_max_pages_per_rpc << PAGE_SHIFT;
840 spin_unlock(&cli->cl_loi_list_lock);
842 return osc_shrink_grant_to_target(cli, target_bytes);
845 int osc_shrink_grant_to_target(struct client_obd *cli, __u64 target_bytes)
848 struct ost_body *body;
851 spin_lock(&cli->cl_loi_list_lock);
852 /* Don't shrink if we are already above or below the desired limit
853 * We don't want to shrink below a single RPC, as that will negatively
854 * impact block allocation and long-term performance. */
855 if (target_bytes < cli->cl_max_pages_per_rpc << PAGE_SHIFT)
856 target_bytes = cli->cl_max_pages_per_rpc << PAGE_SHIFT;
858 if (target_bytes >= cli->cl_avail_grant) {
859 spin_unlock(&cli->cl_loi_list_lock);
862 spin_unlock(&cli->cl_loi_list_lock);
868 osc_announce_cached(cli, &body->oa, 0);
870 spin_lock(&cli->cl_loi_list_lock);
871 if (target_bytes >= cli->cl_avail_grant) {
872 /* available grant has changed since target calculation */
873 spin_unlock(&cli->cl_loi_list_lock);
874 GOTO(out_free, rc = 0);
876 body->oa.o_grant = cli->cl_avail_grant - target_bytes;
877 cli->cl_avail_grant = target_bytes;
878 spin_unlock(&cli->cl_loi_list_lock);
879 if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
880 body->oa.o_valid |= OBD_MD_FLFLAGS;
881 body->oa.o_flags = 0;
883 body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
884 osc_update_next_shrink(cli);
886 rc = osc_set_info_async(NULL, cli->cl_import->imp_obd->obd_self_export,
887 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
888 sizeof(*body), body, NULL);
890 __osc_update_grant(cli, body->oa.o_grant);
896 static int osc_should_shrink_grant(struct client_obd *client)
898 time64_t next_shrink = client->cl_next_shrink_grant;
900 if (client->cl_import == NULL)
903 if (!OCD_HAS_FLAG(&client->cl_import->imp_connect_data, GRANT_SHRINK) ||
904 client->cl_import->imp_grant_shrink_disabled) {
905 osc_update_next_shrink(client);
909 if (ktime_get_seconds() >= next_shrink - 5) {
910 /* Get the current RPC size directly, instead of going via:
911 * cli_brw_size(obd->u.cli.cl_import->imp_obd->obd_self_export)
912 * Keep comment here so that it can be found by searching. */
913 int brw_size = client->cl_max_pages_per_rpc << PAGE_SHIFT;
915 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
916 client->cl_avail_grant > brw_size)
919 osc_update_next_shrink(client);
924 #define GRANT_SHRINK_RPC_BATCH 100
926 static struct delayed_work work;
928 static void osc_grant_work_handler(struct work_struct *data)
930 struct client_obd *cli;
932 bool init_next_shrink = true;
933 time64_t next_shrink = ktime_get_seconds() + GRANT_SHRINK_INTERVAL;
936 mutex_lock(&client_gtd.gtd_mutex);
937 list_for_each_entry(cli, &client_gtd.gtd_clients,
939 if (rpc_sent < GRANT_SHRINK_RPC_BATCH &&
940 osc_should_shrink_grant(cli)) {
941 osc_shrink_grant(cli);
945 if (!init_next_shrink) {
946 if (cli->cl_next_shrink_grant < next_shrink &&
947 cli->cl_next_shrink_grant > ktime_get_seconds())
948 next_shrink = cli->cl_next_shrink_grant;
950 init_next_shrink = false;
951 next_shrink = cli->cl_next_shrink_grant;
954 mutex_unlock(&client_gtd.gtd_mutex);
956 if (client_gtd.gtd_stopped == 1)
959 if (next_shrink > ktime_get_seconds()) {
960 time64_t delay = next_shrink - ktime_get_seconds();
962 schedule_delayed_work(&work, cfs_time_seconds(delay));
964 schedule_work(&work.work);
968 void osc_schedule_grant_work(void)
970 cancel_delayed_work_sync(&work);
971 schedule_work(&work.work);
975 * Start grant thread for returing grant to server for idle clients.
977 static int osc_start_grant_work(void)
979 client_gtd.gtd_stopped = 0;
980 mutex_init(&client_gtd.gtd_mutex);
981 INIT_LIST_HEAD(&client_gtd.gtd_clients);
983 INIT_DELAYED_WORK(&work, osc_grant_work_handler);
984 schedule_work(&work.work);
989 static void osc_stop_grant_work(void)
991 client_gtd.gtd_stopped = 1;
992 cancel_delayed_work_sync(&work);
995 static void osc_add_grant_list(struct client_obd *client)
997 mutex_lock(&client_gtd.gtd_mutex);
998 list_add(&client->cl_grant_chain, &client_gtd.gtd_clients);
999 mutex_unlock(&client_gtd.gtd_mutex);
1002 static void osc_del_grant_list(struct client_obd *client)
1004 if (list_empty(&client->cl_grant_chain))
1007 mutex_lock(&client_gtd.gtd_mutex);
1008 list_del_init(&client->cl_grant_chain);
1009 mutex_unlock(&client_gtd.gtd_mutex);
1012 void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1015 * ocd_grant is the total grant amount we're expect to hold: if we've
1016 * been evicted, it's the new avail_grant amount, cl_dirty_pages will
1017 * drop to 0 as inflight RPCs fail out; otherwise, it's avail_grant +
1020 * race is tolerable here: if we're evicted, but imp_state already
1021 * left EVICTED state, then cl_dirty_pages must be 0 already.
1023 spin_lock(&cli->cl_loi_list_lock);
1024 cli->cl_avail_grant = ocd->ocd_grant;
1025 if (cli->cl_import->imp_state != LUSTRE_IMP_EVICTED) {
1026 unsigned long consumed = cli->cl_reserved_grant;
1028 if (OCD_HAS_FLAG(ocd, GRANT_PARAM))
1029 consumed += cli->cl_dirty_grant;
1031 consumed += cli->cl_dirty_pages << PAGE_SHIFT;
1032 if (cli->cl_avail_grant < consumed) {
1033 CERROR("%s: granted %ld but already consumed %ld\n",
1034 cli_name(cli), cli->cl_avail_grant, consumed);
1035 cli->cl_avail_grant = 0;
1037 cli->cl_avail_grant -= consumed;
1041 if (OCD_HAS_FLAG(ocd, GRANT_PARAM)) {
1045 /* overhead for each extent insertion */
1046 cli->cl_grant_extent_tax = ocd->ocd_grant_tax_kb << 10;
1047 /* determine the appropriate chunk size used by osc_extent. */
1048 cli->cl_chunkbits = max_t(int, PAGE_SHIFT,
1049 ocd->ocd_grant_blkbits);
1050 /* max_pages_per_rpc must be chunk aligned */
1051 chunk_mask = ~((1 << (cli->cl_chunkbits - PAGE_SHIFT)) - 1);
1052 cli->cl_max_pages_per_rpc = (cli->cl_max_pages_per_rpc +
1053 ~chunk_mask) & chunk_mask;
1054 /* determine maximum extent size, in #pages */
1055 size = (u64)ocd->ocd_grant_max_blks << ocd->ocd_grant_blkbits;
1056 cli->cl_max_extent_pages = size >> PAGE_SHIFT;
1057 if (cli->cl_max_extent_pages == 0)
1058 cli->cl_max_extent_pages = 1;
1060 cli->cl_grant_extent_tax = 0;
1061 cli->cl_chunkbits = PAGE_SHIFT;
1062 cli->cl_max_extent_pages = DT_MAX_BRW_PAGES;
1064 spin_unlock(&cli->cl_loi_list_lock);
1067 "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld. chunk bits: %d cl_max_extent_pages: %d\n",
1069 cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits,
1070 cli->cl_max_extent_pages);
1072 if (OCD_HAS_FLAG(ocd, GRANT_SHRINK) && list_empty(&cli->cl_grant_chain))
1073 osc_add_grant_list(cli);
1075 EXPORT_SYMBOL(osc_init_grant);
1077 /* We assume that the reason this OSC got a short read is because it read
1078 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1079 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1080 * this stripe never got written at or beyond this stripe offset yet. */
1081 static void handle_short_read(int nob_read, size_t page_count,
1082 struct brw_page **pga)
1087 /* skip bytes read OK */
1088 while (nob_read > 0) {
1089 LASSERT (page_count > 0);
1091 if (pga[i]->count > nob_read) {
1092 /* EOF inside this page */
1093 ptr = kmap(pga[i]->pg) +
1094 (pga[i]->off & ~PAGE_MASK);
1095 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1102 nob_read -= pga[i]->count;
1107 /* zero remaining pages */
1108 while (page_count-- > 0) {
1109 ptr = kmap(pga[i]->pg) + (pga[i]->off & ~PAGE_MASK);
1110 memset(ptr, 0, pga[i]->count);
1116 static int check_write_rcs(struct ptlrpc_request *req,
1117 int requested_nob, int niocount,
1118 size_t page_count, struct brw_page **pga)
1123 remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1124 sizeof(*remote_rcs) *
1126 if (remote_rcs == NULL) {
1127 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1131 /* return error if any niobuf was in error */
1132 for (i = 0; i < niocount; i++) {
1133 if ((int)remote_rcs[i] < 0) {
1134 CDEBUG(D_INFO, "rc[%d]: %d req %p\n",
1135 i, remote_rcs[i], req);
1136 return remote_rcs[i];
1139 if (remote_rcs[i] != 0) {
1140 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1141 i, remote_rcs[i], req);
1145 if (req->rq_bulk != NULL &&
1146 req->rq_bulk->bd_nob_transferred != requested_nob) {
1147 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1148 req->rq_bulk->bd_nob_transferred, requested_nob);
1155 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1157 if (p1->flag != p2->flag) {
1158 unsigned mask = ~(OBD_BRW_FROM_GRANT | OBD_BRW_NOCACHE |
1159 OBD_BRW_SYNC | OBD_BRW_ASYNC |
1160 OBD_BRW_NOQUOTA | OBD_BRW_SOFT_SYNC);
1162 /* warn if we try to combine flags that we don't know to be
1163 * safe to combine */
1164 if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
1165 CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1166 "report this at https://jira.whamcloud.com/\n",
1167 p1->flag, p2->flag);
1172 return (p1->off + p1->count == p2->off);
1175 #if IS_ENABLED(CONFIG_CRC_T10DIF)
1176 static int osc_checksum_bulk_t10pi(const char *obd_name, int nob,
1177 size_t pg_count, struct brw_page **pga,
1178 int opc, obd_dif_csum_fn *fn,
1182 struct ahash_request *req;
1183 /* Used Adler as the default checksum type on top of DIF tags */
1184 unsigned char cfs_alg = cksum_obd2cfs(OBD_CKSUM_T10_TOP);
1185 struct page *__page;
1186 unsigned char *buffer;
1188 unsigned int bufsize;
1190 int used_number = 0;
1196 LASSERT(pg_count > 0);
1198 __page = alloc_page(GFP_KERNEL);
1202 req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1205 CERROR("%s: unable to initialize checksum hash %s: rc = %d\n",
1206 obd_name, cfs_crypto_hash_name(cfs_alg), rc);
1210 buffer = kmap(__page);
1211 guard_start = (__u16 *)buffer;
1212 guard_number = PAGE_SIZE / sizeof(*guard_start);
1213 while (nob > 0 && pg_count > 0) {
1214 unsigned int count = pga[i]->count > nob ? nob : pga[i]->count;
1216 /* corrupt the data before we compute the checksum, to
1217 * simulate an OST->client data error */
1218 if (unlikely(i == 0 && opc == OST_READ &&
1219 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE))) {
1220 unsigned char *ptr = kmap(pga[i]->pg);
1221 int off = pga[i]->off & ~PAGE_MASK;
1223 memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
1228 * The left guard number should be able to hold checksums of a
1231 rc = obd_page_dif_generate_buffer(obd_name, pga[i]->pg,
1232 pga[i]->off & ~PAGE_MASK,
1234 guard_start + used_number,
1235 guard_number - used_number,
1241 used_number += used;
1242 if (used_number == guard_number) {
1243 cfs_crypto_hash_update_page(req, __page, 0,
1244 used_number * sizeof(*guard_start));
1248 nob -= pga[i]->count;
1256 if (used_number != 0)
1257 cfs_crypto_hash_update_page(req, __page, 0,
1258 used_number * sizeof(*guard_start));
1260 bufsize = sizeof(cksum);
1261 cfs_crypto_hash_final(req, (unsigned char *)&cksum, &bufsize);
1263 /* For sending we only compute the wrong checksum instead
1264 * of corrupting the data so it is still correct on a redo */
1265 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1270 __free_page(__page);
1273 #else /* !CONFIG_CRC_T10DIF */
1274 #define obd_dif_ip_fn NULL
1275 #define obd_dif_crc_fn NULL
1276 #define osc_checksum_bulk_t10pi(name, nob, pgc, pga, opc, fn, ssize, csum) \
1278 #endif /* CONFIG_CRC_T10DIF */
1280 static int osc_checksum_bulk(int nob, size_t pg_count,
1281 struct brw_page **pga, int opc,
1282 enum cksum_types cksum_type,
1286 struct ahash_request *req;
1287 unsigned int bufsize;
1288 unsigned char cfs_alg = cksum_obd2cfs(cksum_type);
1290 LASSERT(pg_count > 0);
1292 req = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1294 CERROR("Unable to initialize checksum hash %s\n",
1295 cfs_crypto_hash_name(cfs_alg));
1296 return PTR_ERR(req);
1299 while (nob > 0 && pg_count > 0) {
1300 unsigned int count = pga[i]->count > nob ? nob : pga[i]->count;
1302 /* corrupt the data before we compute the checksum, to
1303 * simulate an OST->client data error */
1304 if (i == 0 && opc == OST_READ &&
1305 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE)) {
1306 unsigned char *ptr = kmap(pga[i]->pg);
1307 int off = pga[i]->off & ~PAGE_MASK;
1309 memcpy(ptr + off, "bad1", min_t(typeof(nob), 4, nob));
1312 cfs_crypto_hash_update_page(req, pga[i]->pg,
1313 pga[i]->off & ~PAGE_MASK,
1315 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d\n",
1316 (int)(pga[i]->off & ~PAGE_MASK));
1318 nob -= pga[i]->count;
1323 bufsize = sizeof(*cksum);
1324 cfs_crypto_hash_final(req, (unsigned char *)cksum, &bufsize);
1326 /* For sending we only compute the wrong checksum instead
1327 * of corrupting the data so it is still correct on a redo */
1328 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1334 static int osc_checksum_bulk_rw(const char *obd_name,
1335 enum cksum_types cksum_type,
1336 int nob, size_t pg_count,
1337 struct brw_page **pga, int opc,
1340 obd_dif_csum_fn *fn = NULL;
1341 int sector_size = 0;
1345 obd_t10_cksum2dif(cksum_type, &fn, §or_size);
1348 rc = osc_checksum_bulk_t10pi(obd_name, nob, pg_count, pga,
1349 opc, fn, sector_size, check_sum);
1351 rc = osc_checksum_bulk(nob, pg_count, pga, opc, cksum_type,
1357 static inline void osc_release_bounce_pages(struct brw_page **pga,
1360 #ifdef HAVE_LUSTRE_CRYPTO
1363 for (i = 0; i < page_count; i++) {
1364 /* Bounce pages allocated by a call to
1365 * llcrypt_encrypt_pagecache_blocks() in osc_brw_prep_request()
1366 * are identified thanks to the PageChecked flag.
1368 if (PageChecked(pga[i]->pg))
1369 llcrypt_finalize_bounce_page(&pga[i]->pg);
1370 pga[i]->count -= pga[i]->bp_count_diff;
1371 pga[i]->off += pga[i]->bp_off_diff;
1377 osc_brw_prep_request(int cmd, struct client_obd *cli, struct obdo *oa,
1378 u32 page_count, struct brw_page **pga,
1379 struct ptlrpc_request **reqp, int resend)
1381 struct ptlrpc_request *req;
1382 struct ptlrpc_bulk_desc *desc;
1383 struct ost_body *body;
1384 struct obd_ioobj *ioobj;
1385 struct niobuf_remote *niobuf;
1386 int niocount, i, requested_nob, opc, rc, short_io_size = 0;
1387 struct osc_brw_async_args *aa;
1388 struct req_capsule *pill;
1389 struct brw_page *pg_prev;
1391 const char *obd_name = cli->cl_import->imp_obd->obd_name;
1392 struct inode *inode;
1395 inode = page2inode(pga[0]->pg);
1396 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1397 RETURN(-ENOMEM); /* Recoverable */
1398 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1399 RETURN(-EINVAL); /* Fatal */
1401 if ((cmd & OBD_BRW_WRITE) != 0) {
1403 req = ptlrpc_request_alloc_pool(cli->cl_import,
1405 &RQF_OST_BRW_WRITE);
1408 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1413 if (opc == OST_WRITE && inode && IS_ENCRYPTED(inode)) {
1414 for (i = 0; i < page_count; i++) {
1415 struct brw_page *pg = pga[i];
1416 struct page *data_page = NULL;
1417 bool retried = false;
1418 bool lockedbymyself;
1419 u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
1422 if (nunits & ~LUSTRE_ENCRYPTION_MASK)
1423 nunits = (nunits & LUSTRE_ENCRYPTION_MASK) +
1424 LUSTRE_ENCRYPTION_UNIT_SIZE;
1425 /* The page can already be locked when we arrive here.
1426 * This is possible when cl_page_assume/vvp_page_assume
1427 * is stuck on wait_on_page_writeback with page lock
1428 * held. In this case there is no risk for the lock to
1429 * be released while we are doing our encryption
1430 * processing, because writeback against that page will
1431 * end in vvp_page_completion_write/cl_page_completion,
1432 * which means only once the page is fully processed.
1434 lockedbymyself = trylock_page(pg->pg);
1436 llcrypt_encrypt_pagecache_blocks(pg->pg,
1440 unlock_page(pg->pg);
1441 if (IS_ERR(data_page)) {
1442 rc = PTR_ERR(data_page);
1443 if (rc == -ENOMEM && !retried) {
1448 ptlrpc_request_free(req);
1451 /* Set PageChecked flag on bounce page for
1452 * disambiguation in osc_release_bounce_pages().
1454 SetPageChecked(data_page);
1456 /* there should be no gap in the middle of page array */
1457 if (i == page_count - 1) {
1458 struct osc_async_page *oap = brw_page2oap(pg);
1460 oa->o_size = oap->oap_count +
1461 oap->oap_obj_off + oap->oap_page_off;
1463 /* len is forced to nunits, and relative offset to 0
1464 * so store the old, clear text info
1466 pg->bp_count_diff = nunits - pg->count;
1468 pg->bp_off_diff = pg->off & ~PAGE_MASK;
1469 pg->off = pg->off & PAGE_MASK;
1471 } else if (opc == OST_READ && inode && IS_ENCRYPTED(inode)) {
1472 for (i = 0; i < page_count; i++) {
1473 struct brw_page *pg = pga[i];
1474 u32 nunits = (pg->off & ~PAGE_MASK) + pg->count;
1476 if (nunits & ~LUSTRE_ENCRYPTION_MASK)
1477 nunits = (nunits & LUSTRE_ENCRYPTION_MASK) +
1478 LUSTRE_ENCRYPTION_UNIT_SIZE;
1479 /* count/off are forced to cover the whole encryption
1480 * unit size so that all encrypted data is stored on the
1481 * OST, so adjust bp_{count,off}_diff for the size of
1484 pg->bp_count_diff = nunits - pg->count;
1486 pg->bp_off_diff = pg->off & ~PAGE_MASK;
1487 pg->off = pg->off & PAGE_MASK;
1491 for (niocount = i = 1; i < page_count; i++) {
1492 if (!can_merge_pages(pga[i - 1], pga[i]))
1496 pill = &req->rq_pill;
1497 req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1499 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1500 niocount * sizeof(*niobuf));
1502 for (i = 0; i < page_count; i++) {
1503 short_io_size += pga[i]->count;
1504 if (!inode || !IS_ENCRYPTED(inode)) {
1505 pga[i]->bp_count_diff = 0;
1506 pga[i]->bp_off_diff = 0;
1510 /* Check if read/write is small enough to be a short io. */
1511 if (short_io_size > cli->cl_max_short_io_bytes || niocount > 1 ||
1512 !imp_connect_shortio(cli->cl_import))
1515 req_capsule_set_size(pill, &RMF_SHORT_IO, RCL_CLIENT,
1516 opc == OST_READ ? 0 : short_io_size);
1517 if (opc == OST_READ)
1518 req_capsule_set_size(pill, &RMF_SHORT_IO, RCL_SERVER,
1521 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1523 ptlrpc_request_free(req);
1526 osc_set_io_portal(req);
1528 ptlrpc_at_set_req_timeout(req);
1529 /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1531 req->rq_no_retry_einprogress = 1;
1533 if (short_io_size != 0) {
1535 short_io_buf = NULL;
1539 desc = ptlrpc_prep_bulk_imp(req, page_count,
1540 cli->cl_import->imp_connect_data.ocd_brw_size >> LNET_MTU_BITS,
1541 (opc == OST_WRITE ? PTLRPC_BULK_GET_SOURCE :
1542 PTLRPC_BULK_PUT_SINK),
1544 &ptlrpc_bulk_kiov_pin_ops);
1547 GOTO(out, rc = -ENOMEM);
1548 /* NB request now owns desc and will free it when it gets freed */
1550 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1551 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1552 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1553 LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1555 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
1557 /* For READ and WRITE, we can't fill o_uid and o_gid using from_kuid()
1558 * and from_kgid(), because they are asynchronous. Fortunately, variable
1559 * oa contains valid o_uid and o_gid in these two operations.
1560 * Besides, filling o_uid and o_gid is enough for nrs-tbf, see LU-9658.
1561 * OBD_MD_FLUID and OBD_MD_FLUID is not set in order to avoid breaking
1562 * other process logic */
1563 body->oa.o_uid = oa->o_uid;
1564 body->oa.o_gid = oa->o_gid;
1566 obdo_to_ioobj(oa, ioobj);
1567 ioobj->ioo_bufcnt = niocount;
1568 /* The high bits of ioo_max_brw tells server _maximum_ number of bulks
1569 * that might be send for this request. The actual number is decided
1570 * when the RPC is finally sent in ptlrpc_register_bulk(). It sends
1571 * "max - 1" for old client compatibility sending "0", and also so the
1572 * the actual maximum is a power-of-two number, not one less. LU-1431 */
1574 ioobj_max_brw_set(ioobj, desc->bd_md_max_brw);
1576 ioobj_max_brw_set(ioobj, 0);
1578 if (short_io_size != 0) {
1579 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1580 body->oa.o_valid |= OBD_MD_FLFLAGS;
1581 body->oa.o_flags = 0;
1583 body->oa.o_flags |= OBD_FL_SHORT_IO;
1584 CDEBUG(D_CACHE, "Using short io for data transfer, size = %d\n",
1586 if (opc == OST_WRITE) {
1587 short_io_buf = req_capsule_client_get(pill,
1589 LASSERT(short_io_buf != NULL);
1593 LASSERT(page_count > 0);
1595 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1596 struct brw_page *pg = pga[i];
1597 int poff = pg->off & ~PAGE_MASK;
1599 LASSERT(pg->count > 0);
1600 /* make sure there is no gap in the middle of page array */
1601 LASSERTF(page_count == 1 ||
1602 (ergo(i == 0, poff + pg->count == PAGE_SIZE) &&
1603 ergo(i > 0 && i < page_count - 1,
1604 poff == 0 && pg->count == PAGE_SIZE) &&
1605 ergo(i == page_count - 1, poff == 0)),
1606 "i: %d/%d pg: %p off: %llu, count: %u\n",
1607 i, page_count, pg, pg->off, pg->count);
1608 LASSERTF(i == 0 || pg->off > pg_prev->off,
1609 "i %d p_c %u pg %p [pri %lu ind %lu] off %llu"
1610 " prev_pg %p [pri %lu ind %lu] off %llu\n",
1612 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1613 pg_prev->pg, page_private(pg_prev->pg),
1614 pg_prev->pg->index, pg_prev->off);
1615 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1616 (pg->flag & OBD_BRW_SRVLOCK));
1617 if (short_io_size != 0 && opc == OST_WRITE) {
1618 unsigned char *ptr = kmap_atomic(pg->pg);
1620 LASSERT(short_io_size >= requested_nob + pg->count);
1621 memcpy(short_io_buf + requested_nob,
1625 } else if (short_io_size == 0) {
1626 desc->bd_frag_ops->add_kiov_frag(desc, pg->pg, poff,
1629 requested_nob += pg->count;
1631 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1633 niobuf->rnb_len += pg->count;
1635 niobuf->rnb_offset = pg->off;
1636 niobuf->rnb_len = pg->count;
1637 niobuf->rnb_flags = pg->flag;
1642 LASSERTF((void *)(niobuf - niocount) ==
1643 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1644 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1645 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1647 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1649 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1650 body->oa.o_valid |= OBD_MD_FLFLAGS;
1651 body->oa.o_flags = 0;
1653 body->oa.o_flags |= OBD_FL_RECOV_RESEND;
1656 if (osc_should_shrink_grant(cli))
1657 osc_shrink_grant_local(cli, &body->oa);
1659 /* size[REQ_REC_OFF] still sizeof (*body) */
1660 if (opc == OST_WRITE) {
1661 if (cli->cl_checksum &&
1662 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1663 /* store cl_cksum_type in a local variable since
1664 * it can be changed via lprocfs */
1665 enum cksum_types cksum_type = cli->cl_cksum_type;
1667 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1668 body->oa.o_flags = 0;
1670 body->oa.o_flags |= obd_cksum_type_pack(obd_name,
1672 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1674 rc = osc_checksum_bulk_rw(obd_name, cksum_type,
1675 requested_nob, page_count,
1679 CDEBUG(D_PAGE, "failed to checksum, rc = %d\n",
1683 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1686 /* save this in 'oa', too, for later checking */
1687 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1688 oa->o_flags |= obd_cksum_type_pack(obd_name,
1691 /* clear out the checksum flag, in case this is a
1692 * resend but cl_checksum is no longer set. b=11238 */
1693 oa->o_valid &= ~OBD_MD_FLCKSUM;
1695 oa->o_cksum = body->oa.o_cksum;
1696 /* 1 RC per niobuf */
1697 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1698 sizeof(__u32) * niocount);
1700 if (cli->cl_checksum &&
1701 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1702 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1703 body->oa.o_flags = 0;
1704 body->oa.o_flags |= obd_cksum_type_pack(obd_name,
1705 cli->cl_cksum_type);
1706 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1709 /* Client cksum has been already copied to wire obdo in previous
1710 * lustre_set_wire_obdo(), and in the case a bulk-read is being
1711 * resent due to cksum error, this will allow Server to
1712 * check+dump pages on its side */
1714 ptlrpc_request_set_replen(req);
1716 aa = ptlrpc_req_async_args(aa, req);
1718 aa->aa_requested_nob = requested_nob;
1719 aa->aa_nio_count = niocount;
1720 aa->aa_page_count = page_count;
1724 INIT_LIST_HEAD(&aa->aa_oaps);
1727 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1728 CDEBUG(D_RPCTRACE, "brw rpc %p - object "DOSTID" offset %lld<>%lld\n",
1729 req, POSTID(&oa->o_oi), niobuf[0].rnb_offset,
1730 niobuf[niocount - 1].rnb_offset + niobuf[niocount - 1].rnb_len);
1734 ptlrpc_req_finished(req);
1738 char dbgcksum_file_name[PATH_MAX];
1740 static void dump_all_bulk_pages(struct obdo *oa, __u32 page_count,
1741 struct brw_page **pga, __u32 server_cksum,
1749 /* will only keep dump of pages on first error for the same range in
1750 * file/fid, not during the resends/retries. */
1751 snprintf(dbgcksum_file_name, sizeof(dbgcksum_file_name),
1752 "%s-checksum_dump-osc-"DFID":[%llu-%llu]-%x-%x",
1753 (strncmp(libcfs_debug_file_path_arr, "NONE", 4) != 0 ?
1754 libcfs_debug_file_path_arr :
1755 LIBCFS_DEBUG_FILE_PATH_DEFAULT),
1756 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : 0ULL,
1757 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1758 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1760 pga[page_count-1]->off + pga[page_count-1]->count - 1,
1761 client_cksum, server_cksum);
1762 filp = filp_open(dbgcksum_file_name,
1763 O_CREAT | O_EXCL | O_WRONLY | O_LARGEFILE, 0600);
1767 CDEBUG(D_INFO, "%s: can't open to dump pages with "
1768 "checksum error: rc = %d\n", dbgcksum_file_name,
1771 CERROR("%s: can't open to dump pages with checksum "
1772 "error: rc = %d\n", dbgcksum_file_name, rc);
1776 for (i = 0; i < page_count; i++) {
1777 len = pga[i]->count;
1778 buf = kmap(pga[i]->pg);
1780 rc = cfs_kernel_write(filp, buf, len, &filp->f_pos);
1782 CERROR("%s: wanted to write %u but got %d "
1783 "error\n", dbgcksum_file_name, len, rc);
1788 CDEBUG(D_INFO, "%s: wrote %d bytes\n",
1789 dbgcksum_file_name, rc);
1794 rc = vfs_fsync_range(filp, 0, LLONG_MAX, 1);
1796 CERROR("%s: sync returns %d\n", dbgcksum_file_name, rc);
1797 filp_close(filp, NULL);
1801 check_write_checksum(struct obdo *oa, const struct lnet_process_id *peer,
1802 __u32 client_cksum, __u32 server_cksum,
1803 struct osc_brw_async_args *aa)
1805 const char *obd_name = aa->aa_cli->cl_import->imp_obd->obd_name;
1806 enum cksum_types cksum_type;
1807 obd_dif_csum_fn *fn = NULL;
1808 int sector_size = 0;
1813 if (server_cksum == client_cksum) {
1814 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1818 if (aa->aa_cli->cl_checksum_dump)
1819 dump_all_bulk_pages(oa, aa->aa_page_count, aa->aa_ppga,
1820 server_cksum, client_cksum);
1822 cksum_type = obd_cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
1825 switch (cksum_type) {
1826 case OBD_CKSUM_T10IP512:
1830 case OBD_CKSUM_T10IP4K:
1834 case OBD_CKSUM_T10CRC512:
1835 fn = obd_dif_crc_fn;
1838 case OBD_CKSUM_T10CRC4K:
1839 fn = obd_dif_crc_fn;
1847 rc = osc_checksum_bulk_t10pi(obd_name, aa->aa_requested_nob,
1848 aa->aa_page_count, aa->aa_ppga,
1849 OST_WRITE, fn, sector_size,
1852 rc = osc_checksum_bulk(aa->aa_requested_nob, aa->aa_page_count,
1853 aa->aa_ppga, OST_WRITE, cksum_type,
1857 msg = "failed to calculate the client write checksum";
1858 else if (cksum_type != obd_cksum_type_unpack(aa->aa_oa->o_flags))
1859 msg = "the server did not use the checksum type specified in "
1860 "the original request - likely a protocol problem";
1861 else if (new_cksum == server_cksum)
1862 msg = "changed on the client after we checksummed it - "
1863 "likely false positive due to mmap IO (bug 11742)";
1864 else if (new_cksum == client_cksum)
1865 msg = "changed in transit before arrival at OST";
1867 msg = "changed in transit AND doesn't match the original - "
1868 "likely false positive due to mmap IO (bug 11742)";
1870 LCONSOLE_ERROR_MSG(0x132, "%s: BAD WRITE CHECKSUM: %s: from %s inode "
1871 DFID " object "DOSTID" extent [%llu-%llu], original "
1872 "client csum %x (type %x), server csum %x (type %x),"
1873 " client csum now %x\n",
1874 obd_name, msg, libcfs_nid2str(peer->nid),
1875 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1876 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1877 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1878 POSTID(&oa->o_oi), aa->aa_ppga[0]->off,
1879 aa->aa_ppga[aa->aa_page_count - 1]->off +
1880 aa->aa_ppga[aa->aa_page_count-1]->count - 1,
1882 obd_cksum_type_unpack(aa->aa_oa->o_flags),
1883 server_cksum, cksum_type, new_cksum);
1887 /* Note rc enters this function as number of bytes transferred */
1888 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1890 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1891 struct client_obd *cli = aa->aa_cli;
1892 const char *obd_name = cli->cl_import->imp_obd->obd_name;
1893 const struct lnet_process_id *peer =
1894 &req->rq_import->imp_connection->c_peer;
1895 struct ost_body *body;
1896 u32 client_cksum = 0;
1897 struct inode *inode;
1901 if (rc < 0 && rc != -EDQUOT) {
1902 DEBUG_REQ(D_INFO, req, "Failed request: rc = %d", rc);
1906 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1907 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1909 DEBUG_REQ(D_INFO, req, "cannot unpack body");
1913 /* set/clear over quota flag for a uid/gid/projid */
1914 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1915 body->oa.o_valid & (OBD_MD_FLALLQUOTA)) {
1916 unsigned qid[LL_MAXQUOTAS] = {
1917 body->oa.o_uid, body->oa.o_gid,
1918 body->oa.o_projid };
1920 "setdq for [%u %u %u] with valid %#llx, flags %x\n",
1921 body->oa.o_uid, body->oa.o_gid, body->oa.o_projid,
1922 body->oa.o_valid, body->oa.o_flags);
1923 osc_quota_setdq(cli, req->rq_xid, qid, body->oa.o_valid,
1927 osc_update_grant(cli, body);
1932 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1933 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1935 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1937 CERROR("%s: unexpected positive size %d\n",
1942 if (req->rq_bulk != NULL &&
1943 sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1946 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1947 check_write_checksum(&body->oa, peer, client_cksum,
1948 body->oa.o_cksum, aa))
1951 rc = check_write_rcs(req, aa->aa_requested_nob,
1952 aa->aa_nio_count, aa->aa_page_count,
1957 /* The rest of this function executes only for OST_READs */
1959 if (req->rq_bulk == NULL) {
1960 rc = req_capsule_get_size(&req->rq_pill, &RMF_SHORT_IO,
1962 LASSERT(rc == req->rq_status);
1964 /* if unwrap_bulk failed, return -EAGAIN to retry */
1965 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1968 GOTO(out, rc = -EAGAIN);
1970 if (rc > aa->aa_requested_nob) {
1971 CERROR("%s: unexpected size %d, requested %d\n", obd_name,
1972 rc, aa->aa_requested_nob);
1976 if (req->rq_bulk != NULL && rc != req->rq_bulk->bd_nob_transferred) {
1977 CERROR("%s: unexpected size %d, transferred %d\n", obd_name,
1978 rc, req->rq_bulk->bd_nob_transferred);
1982 if (req->rq_bulk == NULL) {
1984 int nob, pg_count, i = 0;
1987 CDEBUG(D_CACHE, "Using short io read, size %d\n", rc);
1988 pg_count = aa->aa_page_count;
1989 buf = req_capsule_server_sized_get(&req->rq_pill, &RMF_SHORT_IO,
1992 while (nob > 0 && pg_count > 0) {
1994 int count = aa->aa_ppga[i]->count > nob ?
1995 nob : aa->aa_ppga[i]->count;
1997 CDEBUG(D_CACHE, "page %p count %d\n",
1998 aa->aa_ppga[i]->pg, count);
1999 ptr = kmap_atomic(aa->aa_ppga[i]->pg);
2000 memcpy(ptr + (aa->aa_ppga[i]->off & ~PAGE_MASK), buf,
2002 kunmap_atomic((void *) ptr);
2011 if (rc < aa->aa_requested_nob)
2012 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
2014 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
2015 static int cksum_counter;
2016 u32 server_cksum = body->oa.o_cksum;
2019 enum cksum_types cksum_type;
2020 u32 o_flags = body->oa.o_valid & OBD_MD_FLFLAGS ?
2021 body->oa.o_flags : 0;
2023 cksum_type = obd_cksum_type_unpack(o_flags);
2024 rc = osc_checksum_bulk_rw(obd_name, cksum_type, rc,
2025 aa->aa_page_count, aa->aa_ppga,
2026 OST_READ, &client_cksum);
2030 if (req->rq_bulk != NULL &&
2031 peer->nid != req->rq_bulk->bd_sender) {
2033 router = libcfs_nid2str(req->rq_bulk->bd_sender);
2036 if (server_cksum != client_cksum) {
2037 struct ost_body *clbody;
2038 u32 page_count = aa->aa_page_count;
2040 clbody = req_capsule_client_get(&req->rq_pill,
2042 if (cli->cl_checksum_dump)
2043 dump_all_bulk_pages(&clbody->oa, page_count,
2044 aa->aa_ppga, server_cksum,
2047 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
2048 "%s%s%s inode "DFID" object "DOSTID
2049 " extent [%llu-%llu], client %x, "
2050 "server %x, cksum_type %x\n",
2052 libcfs_nid2str(peer->nid),
2054 clbody->oa.o_valid & OBD_MD_FLFID ?
2055 clbody->oa.o_parent_seq : 0ULL,
2056 clbody->oa.o_valid & OBD_MD_FLFID ?
2057 clbody->oa.o_parent_oid : 0,
2058 clbody->oa.o_valid & OBD_MD_FLFID ?
2059 clbody->oa.o_parent_ver : 0,
2060 POSTID(&body->oa.o_oi),
2061 aa->aa_ppga[0]->off,
2062 aa->aa_ppga[page_count-1]->off +
2063 aa->aa_ppga[page_count-1]->count - 1,
2064 client_cksum, server_cksum,
2067 aa->aa_oa->o_cksum = client_cksum;
2071 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
2074 } else if (unlikely(client_cksum)) {
2075 static int cksum_missed;
2078 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
2079 CERROR("%s: checksum %u requested from %s but not sent\n",
2080 obd_name, cksum_missed,
2081 libcfs_nid2str(peer->nid));
2086 inode = page2inode(aa->aa_ppga[0]->pg);
2087 if (inode && IS_ENCRYPTED(inode)) {
2090 if (!llcrypt_has_encryption_key(inode)) {
2091 CDEBUG(D_SEC, "no enc key for ino %lu\n", inode->i_ino);
2094 for (idx = 0; idx < aa->aa_page_count; idx++) {
2095 struct brw_page *pg = aa->aa_ppga[idx];
2096 unsigned int offs = 0;
2098 while (offs < PAGE_SIZE) {
2099 /* do not decrypt if page is all 0s */
2100 if (memchr_inv(page_address(pg->pg) + offs, 0,
2101 LUSTRE_ENCRYPTION_UNIT_SIZE) == NULL) {
2102 /* if page is empty forward info to
2103 * upper layers (ll_io_zero_page) by
2104 * clearing PagePrivate2
2107 ClearPagePrivate2(pg->pg);
2111 /* The page is already locked when we arrive here,
2112 * except when we deal with a twisted page for
2113 * specific Direct IO support, in which case
2114 * PageChecked flag is set on page.
2116 if (PageChecked(pg->pg))
2118 rc = llcrypt_decrypt_pagecache_blocks(pg->pg,
2119 LUSTRE_ENCRYPTION_UNIT_SIZE,
2121 if (PageChecked(pg->pg))
2122 unlock_page(pg->pg);
2126 offs += LUSTRE_ENCRYPTION_UNIT_SIZE;
2133 lustre_get_wire_obdo(&req->rq_import->imp_connect_data,
2134 aa->aa_oa, &body->oa);
2139 static int osc_brw_redo_request(struct ptlrpc_request *request,
2140 struct osc_brw_async_args *aa, int rc)
2142 struct ptlrpc_request *new_req;
2143 struct osc_brw_async_args *new_aa;
2144 struct osc_async_page *oap;
2147 /* The below message is checked in replay-ost-single.sh test_8ae*/
2148 DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
2149 "redo for recoverable error %d", rc);
2151 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
2152 OST_WRITE ? OBD_BRW_WRITE : OBD_BRW_READ,
2153 aa->aa_cli, aa->aa_oa, aa->aa_page_count,
2154 aa->aa_ppga, &new_req, 1);
2158 list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2159 if (oap->oap_request != NULL) {
2160 LASSERTF(request == oap->oap_request,
2161 "request %p != oap_request %p\n",
2162 request, oap->oap_request);
2166 * New request takes over pga and oaps from old request.
2167 * Note that copying a list_head doesn't work, need to move it...
2170 new_req->rq_interpret_reply = request->rq_interpret_reply;
2171 new_req->rq_async_args = request->rq_async_args;
2172 new_req->rq_commit_cb = request->rq_commit_cb;
2173 /* cap resend delay to the current request timeout, this is similar to
2174 * what ptlrpc does (see after_reply()) */
2175 if (aa->aa_resends > new_req->rq_timeout)
2176 new_req->rq_sent = ktime_get_real_seconds() + new_req->rq_timeout;
2178 new_req->rq_sent = ktime_get_real_seconds() + aa->aa_resends;
2179 new_req->rq_generation_set = 1;
2180 new_req->rq_import_generation = request->rq_import_generation;
2182 new_aa = ptlrpc_req_async_args(new_aa, new_req);
2184 INIT_LIST_HEAD(&new_aa->aa_oaps);
2185 list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
2186 INIT_LIST_HEAD(&new_aa->aa_exts);
2187 list_splice_init(&aa->aa_exts, &new_aa->aa_exts);
2188 new_aa->aa_resends = aa->aa_resends;
2190 list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
2191 if (oap->oap_request) {
2192 ptlrpc_req_finished(oap->oap_request);
2193 oap->oap_request = ptlrpc_request_addref(new_req);
2197 /* XXX: This code will run into problem if we're going to support
2198 * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
2199 * and wait for all of them to be finished. We should inherit request
2200 * set from old request. */
2201 ptlrpcd_add_req(new_req);
2203 DEBUG_REQ(D_INFO, new_req, "new request");
2208 * ugh, we want disk allocation on the target to happen in offset order. we'll
2209 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
2210 * fine for our small page arrays and doesn't require allocation. its an
2211 * insertion sort that swaps elements that are strides apart, shrinking the
2212 * stride down until its '1' and the array is sorted.
2214 static void sort_brw_pages(struct brw_page **array, int num)
2217 struct brw_page *tmp;
2221 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
2226 for (i = stride ; i < num ; i++) {
2229 while (j >= stride && array[j - stride]->off > tmp->off) {
2230 array[j] = array[j - stride];
2235 } while (stride > 1);
2238 static void osc_release_ppga(struct brw_page **ppga, size_t count)
2240 LASSERT(ppga != NULL);
2241 OBD_FREE_PTR_ARRAY(ppga, count);
2244 static int brw_interpret(const struct lu_env *env,
2245 struct ptlrpc_request *req, void *args, int rc)
2247 struct osc_brw_async_args *aa = args;
2248 struct osc_extent *ext;
2249 struct osc_extent *tmp;
2250 struct client_obd *cli = aa->aa_cli;
2251 unsigned long transferred = 0;
2255 rc = osc_brw_fini_request(req, rc);
2256 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
2258 /* restore clear text pages */
2259 osc_release_bounce_pages(aa->aa_ppga, aa->aa_page_count);
2262 * When server returns -EINPROGRESS, client should always retry
2263 * regardless of the number of times the bulk was resent already.
2265 if (osc_recoverable_error(rc) && !req->rq_no_delay) {
2266 if (req->rq_import_generation !=
2267 req->rq_import->imp_generation) {
2268 CDEBUG(D_HA, "%s: resend cross eviction for object: "
2269 ""DOSTID", rc = %d.\n",
2270 req->rq_import->imp_obd->obd_name,
2271 POSTID(&aa->aa_oa->o_oi), rc);
2272 } else if (rc == -EINPROGRESS ||
2273 client_should_resend(aa->aa_resends, aa->aa_cli)) {
2274 rc = osc_brw_redo_request(req, aa, rc);
2276 CERROR("%s: too many resent retries for object: "
2277 "%llu:%llu, rc = %d.\n",
2278 req->rq_import->imp_obd->obd_name,
2279 POSTID(&aa->aa_oa->o_oi), rc);
2284 else if (rc == -EAGAIN || rc == -EINPROGRESS)
2289 struct obdo *oa = aa->aa_oa;
2290 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
2291 unsigned long valid = 0;
2292 struct cl_object *obj;
2293 struct osc_async_page *last;
2295 last = brw_page2oap(aa->aa_ppga[aa->aa_page_count - 1]);
2296 obj = osc2cl(last->oap_obj);
2298 cl_object_attr_lock(obj);
2299 if (oa->o_valid & OBD_MD_FLBLOCKS) {
2300 attr->cat_blocks = oa->o_blocks;
2301 valid |= CAT_BLOCKS;
2303 if (oa->o_valid & OBD_MD_FLMTIME) {
2304 attr->cat_mtime = oa->o_mtime;
2307 if (oa->o_valid & OBD_MD_FLATIME) {
2308 attr->cat_atime = oa->o_atime;
2311 if (oa->o_valid & OBD_MD_FLCTIME) {
2312 attr->cat_ctime = oa->o_ctime;
2316 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
2317 struct lov_oinfo *loi = cl2osc(obj)->oo_oinfo;
2318 loff_t last_off = last->oap_count + last->oap_obj_off +
2321 /* Change file size if this is an out of quota or
2322 * direct IO write and it extends the file size */
2323 if (loi->loi_lvb.lvb_size < last_off) {
2324 attr->cat_size = last_off;
2327 /* Extend KMS if it's not a lockless write */
2328 if (loi->loi_kms < last_off &&
2329 oap2osc_page(last)->ops_srvlock == 0) {
2330 attr->cat_kms = last_off;
2336 cl_object_attr_update(env, obj, attr, valid);
2337 cl_object_attr_unlock(obj);
2339 OBD_SLAB_FREE_PTR(aa->aa_oa, osc_obdo_kmem);
2342 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE && rc == 0)
2343 osc_inc_unstable_pages(req);
2345 list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
2346 list_del_init(&ext->oe_link);
2347 osc_extent_finish(env, ext, 1,
2348 rc && req->rq_no_delay ? -EWOULDBLOCK : rc);
2350 LASSERT(list_empty(&aa->aa_exts));
2351 LASSERT(list_empty(&aa->aa_oaps));
2353 transferred = (req->rq_bulk == NULL ? /* short io */
2354 aa->aa_requested_nob :
2355 req->rq_bulk->bd_nob_transferred);
2357 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2358 ptlrpc_lprocfs_brw(req, transferred);
2360 spin_lock(&cli->cl_loi_list_lock);
2361 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2362 * is called so we know whether to go to sync BRWs or wait for more
2363 * RPCs to complete */
2364 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2365 cli->cl_w_in_flight--;
2367 cli->cl_r_in_flight--;
2368 osc_wake_cache_waiters(cli);
2369 spin_unlock(&cli->cl_loi_list_lock);
2371 osc_io_unplug(env, cli, NULL);
2375 static void brw_commit(struct ptlrpc_request *req)
2377 /* If osc_inc_unstable_pages (via osc_extent_finish) races with
2378 * this called via the rq_commit_cb, I need to ensure
2379 * osc_dec_unstable_pages is still called. Otherwise unstable
2380 * pages may be leaked. */
2381 spin_lock(&req->rq_lock);
2382 if (likely(req->rq_unstable)) {
2383 req->rq_unstable = 0;
2384 spin_unlock(&req->rq_lock);
2386 osc_dec_unstable_pages(req);
2388 req->rq_committed = 1;
2389 spin_unlock(&req->rq_lock);
2394 * Build an RPC by the list of extent @ext_list. The caller must ensure
2395 * that the total pages in this list are NOT over max pages per RPC.
2396 * Extents in the list must be in OES_RPC state.
2398 int osc_build_rpc(const struct lu_env *env, struct client_obd *cli,
2399 struct list_head *ext_list, int cmd)
2401 struct ptlrpc_request *req = NULL;
2402 struct osc_extent *ext;
2403 struct brw_page **pga = NULL;
2404 struct osc_brw_async_args *aa = NULL;
2405 struct obdo *oa = NULL;
2406 struct osc_async_page *oap;
2407 struct osc_object *obj = NULL;
2408 struct cl_req_attr *crattr = NULL;
2409 loff_t starting_offset = OBD_OBJECT_EOF;
2410 loff_t ending_offset = 0;
2411 /* '1' for consistency with code that checks !mpflag to restore */
2415 bool soft_sync = false;
2416 bool ndelay = false;
2420 __u32 layout_version = 0;
2421 LIST_HEAD(rpc_list);
2422 struct ost_body *body;
2424 LASSERT(!list_empty(ext_list));
2426 /* add pages into rpc_list to build BRW rpc */
2427 list_for_each_entry(ext, ext_list, oe_link) {
2428 LASSERT(ext->oe_state == OES_RPC);
2429 mem_tight |= ext->oe_memalloc;
2430 grant += ext->oe_grants;
2431 page_count += ext->oe_nr_pages;
2432 layout_version = max(layout_version, ext->oe_layout_version);
2437 soft_sync = osc_over_unstable_soft_limit(cli);
2439 mpflag = memalloc_noreclaim_save();
2441 OBD_ALLOC_PTR_ARRAY(pga, page_count);
2443 GOTO(out, rc = -ENOMEM);
2445 OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
2447 GOTO(out, rc = -ENOMEM);
2450 list_for_each_entry(ext, ext_list, oe_link) {
2451 list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
2453 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
2455 oap->oap_brw_flags |= OBD_BRW_SOFT_SYNC;
2456 pga[i] = &oap->oap_brw_page;
2457 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2460 list_add_tail(&oap->oap_rpc_item, &rpc_list);
2461 if (starting_offset == OBD_OBJECT_EOF ||
2462 starting_offset > oap->oap_obj_off)
2463 starting_offset = oap->oap_obj_off;
2465 LASSERT(oap->oap_page_off == 0);
2466 if (ending_offset < oap->oap_obj_off + oap->oap_count)
2467 ending_offset = oap->oap_obj_off +
2470 LASSERT(oap->oap_page_off + oap->oap_count ==
2477 /* first page in the list */
2478 oap = list_entry(rpc_list.next, typeof(*oap), oap_rpc_item);
2480 crattr = &osc_env_info(env)->oti_req_attr;
2481 memset(crattr, 0, sizeof(*crattr));
2482 crattr->cra_type = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2483 crattr->cra_flags = ~0ULL;
2484 crattr->cra_page = oap2cl_page(oap);
2485 crattr->cra_oa = oa;
2486 cl_req_attr_set(env, osc2cl(obj), crattr);
2488 if (cmd == OBD_BRW_WRITE) {
2489 oa->o_grant_used = grant;
2490 if (layout_version > 0) {
2491 CDEBUG(D_LAYOUT, DFID": write with layout version %u\n",
2492 PFID(&oa->o_oi.oi_fid), layout_version);
2494 oa->o_layout_version = layout_version;
2495 oa->o_valid |= OBD_MD_LAYOUT_VERSION;
2499 sort_brw_pages(pga, page_count);
2500 rc = osc_brw_prep_request(cmd, cli, oa, page_count, pga, &req, 0);
2502 CERROR("prep_req failed: %d\n", rc);
2506 req->rq_commit_cb = brw_commit;
2507 req->rq_interpret_reply = brw_interpret;
2508 req->rq_memalloc = mem_tight != 0;
2509 oap->oap_request = ptlrpc_request_addref(req);
2511 req->rq_no_resend = req->rq_no_delay = 1;
2512 /* probably set a shorter timeout value.
2513 * to handle ETIMEDOUT in brw_interpret() correctly. */
2514 /* lustre_msg_set_timeout(req, req->rq_timeout / 2); */
2517 /* Need to update the timestamps after the request is built in case
2518 * we race with setattr (locally or in queue at OST). If OST gets
2519 * later setattr before earlier BRW (as determined by the request xid),
2520 * the OST will not use BRW timestamps. Sadly, there is no obvious
2521 * way to do this in a single call. bug 10150 */
2522 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
2523 crattr->cra_oa = &body->oa;
2524 crattr->cra_flags = OBD_MD_FLMTIME | OBD_MD_FLCTIME | OBD_MD_FLATIME;
2525 cl_req_attr_set(env, osc2cl(obj), crattr);
2526 lustre_msg_set_jobid(req->rq_reqmsg, crattr->cra_jobid);
2528 aa = ptlrpc_req_async_args(aa, req);
2529 INIT_LIST_HEAD(&aa->aa_oaps);
2530 list_splice_init(&rpc_list, &aa->aa_oaps);
2531 INIT_LIST_HEAD(&aa->aa_exts);
2532 list_splice_init(ext_list, &aa->aa_exts);
2534 spin_lock(&cli->cl_loi_list_lock);
2535 starting_offset >>= PAGE_SHIFT;
2536 if (cmd == OBD_BRW_READ) {
2537 cli->cl_r_in_flight++;
2538 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2539 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2540 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2541 starting_offset + 1);
2543 cli->cl_w_in_flight++;
2544 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2545 lprocfs_oh_tally(&cli->cl_write_rpc_hist, cli->cl_w_in_flight);
2546 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2547 starting_offset + 1);
2549 spin_unlock(&cli->cl_loi_list_lock);
2551 DEBUG_REQ(D_INODE, req, "%d pages, aa %p, now %ur/%uw in flight",
2552 page_count, aa, cli->cl_r_in_flight,
2553 cli->cl_w_in_flight);
2554 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_DELAY_IO, cfs_fail_val);
2556 ptlrpcd_add_req(req);
2562 memalloc_noreclaim_restore(mpflag);
2565 LASSERT(req == NULL);
2568 OBD_SLAB_FREE_PTR(oa, osc_obdo_kmem);
2570 osc_release_bounce_pages(pga, page_count);
2571 osc_release_ppga(pga, page_count);
2573 /* this should happen rarely and is pretty bad, it makes the
2574 * pending list not follow the dirty order */
2575 while (!list_empty(ext_list)) {
2576 ext = list_entry(ext_list->next, struct osc_extent,
2578 list_del_init(&ext->oe_link);
2579 osc_extent_finish(env, ext, 0, rc);
2585 static int osc_set_lock_data(struct ldlm_lock *lock, void *data)
2589 LASSERT(lock != NULL);
2591 lock_res_and_lock(lock);
2593 if (lock->l_ast_data == NULL)
2594 lock->l_ast_data = data;
2595 if (lock->l_ast_data == data)
2598 unlock_res_and_lock(lock);
2603 int osc_enqueue_fini(struct ptlrpc_request *req, osc_enqueue_upcall_f upcall,
2604 void *cookie, struct lustre_handle *lockh,
2605 enum ldlm_mode mode, __u64 *flags, bool speculative,
2608 bool intent = *flags & LDLM_FL_HAS_INTENT;
2612 /* The request was created before ldlm_cli_enqueue call. */
2613 if (intent && errcode == ELDLM_LOCK_ABORTED) {
2614 struct ldlm_reply *rep;
2616 rep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
2617 LASSERT(rep != NULL);
2619 rep->lock_policy_res1 =
2620 ptlrpc_status_ntoh(rep->lock_policy_res1);
2621 if (rep->lock_policy_res1)
2622 errcode = rep->lock_policy_res1;
2624 *flags |= LDLM_FL_LVB_READY;
2625 } else if (errcode == ELDLM_OK) {
2626 *flags |= LDLM_FL_LVB_READY;
2629 /* Call the update callback. */
2630 rc = (*upcall)(cookie, lockh, errcode);
2632 /* release the reference taken in ldlm_cli_enqueue() */
2633 if (errcode == ELDLM_LOCK_MATCHED)
2635 if (errcode == ELDLM_OK && lustre_handle_is_used(lockh))
2636 ldlm_lock_decref(lockh, mode);
2641 int osc_enqueue_interpret(const struct lu_env *env, struct ptlrpc_request *req,
2644 struct osc_enqueue_args *aa = args;
2645 struct ldlm_lock *lock;
2646 struct lustre_handle *lockh = &aa->oa_lockh;
2647 enum ldlm_mode mode = aa->oa_mode;
2648 struct ost_lvb *lvb = aa->oa_lvb;
2649 __u32 lvb_len = sizeof(*lvb);
2654 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2656 lock = ldlm_handle2lock(lockh);
2657 LASSERTF(lock != NULL,
2658 "lockh %#llx, req %p, aa %p - client evicted?\n",
2659 lockh->cookie, req, aa);
2661 /* Take an additional reference so that a blocking AST that
2662 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2663 * to arrive after an upcall has been executed by
2664 * osc_enqueue_fini(). */
2665 ldlm_lock_addref(lockh, mode);
2667 /* Let cl_lock_state_wait fail with -ERESTARTSYS to unuse sublocks. */
2668 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_ENQUEUE_HANG, 2);
2670 /* Let CP AST to grant the lock first. */
2671 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
2673 if (aa->oa_speculative) {
2674 LASSERT(aa->oa_lvb == NULL);
2675 LASSERT(aa->oa_flags == NULL);
2676 aa->oa_flags = &flags;
2679 /* Complete obtaining the lock procedure. */
2680 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_type, 1,
2681 aa->oa_mode, aa->oa_flags, lvb, lvb_len,
2683 /* Complete osc stuff. */
2684 rc = osc_enqueue_fini(req, aa->oa_upcall, aa->oa_cookie, lockh, mode,
2685 aa->oa_flags, aa->oa_speculative, rc);
2687 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
2689 ldlm_lock_decref(lockh, mode);
2690 LDLM_LOCK_PUT(lock);
2694 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2695 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2696 * other synchronous requests, however keeping some locks and trying to obtain
2697 * others may take a considerable amount of time in a case of ost failure; and
2698 * when other sync requests do not get released lock from a client, the client
2699 * is evicted from the cluster -- such scenarious make the life difficult, so
2700 * release locks just after they are obtained. */
2701 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2702 __u64 *flags, union ldlm_policy_data *policy,
2703 struct ost_lvb *lvb, osc_enqueue_upcall_f upcall,
2704 void *cookie, struct ldlm_enqueue_info *einfo,
2705 struct ptlrpc_request_set *rqset, int async,
2708 struct obd_device *obd = exp->exp_obd;
2709 struct lustre_handle lockh = { 0 };
2710 struct ptlrpc_request *req = NULL;
2711 int intent = *flags & LDLM_FL_HAS_INTENT;
2712 __u64 match_flags = *flags;
2713 enum ldlm_mode mode;
2717 /* Filesystem lock extents are extended to page boundaries so that
2718 * dealing with the page cache is a little smoother. */
2719 policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
2720 policy->l_extent.end |= ~PAGE_MASK;
2722 /* Next, search for already existing extent locks that will cover us */
2723 /* If we're trying to read, we also search for an existing PW lock. The
2724 * VFS and page cache already protect us locally, so lots of readers/
2725 * writers can share a single PW lock.
2727 * There are problems with conversion deadlocks, so instead of
2728 * converting a read lock to a write lock, we'll just enqueue a new
2731 * At some point we should cancel the read lock instead of making them
2732 * send us a blocking callback, but there are problems with canceling
2733 * locks out from other users right now, too. */
2734 mode = einfo->ei_mode;
2735 if (einfo->ei_mode == LCK_PR)
2737 /* Normal lock requests must wait for the LVB to be ready before
2738 * matching a lock; speculative lock requests do not need to,
2739 * because they will not actually use the lock. */
2741 match_flags |= LDLM_FL_LVB_READY;
2743 match_flags |= LDLM_FL_BLOCK_GRANTED;
2744 mode = ldlm_lock_match(obd->obd_namespace, match_flags, res_id,
2745 einfo->ei_type, policy, mode, &lockh);
2747 struct ldlm_lock *matched;
2749 if (*flags & LDLM_FL_TEST_LOCK)
2752 matched = ldlm_handle2lock(&lockh);
2754 /* This DLM lock request is speculative, and does not
2755 * have an associated IO request. Therefore if there
2756 * is already a DLM lock, it wll just inform the
2757 * caller to cancel the request for this stripe.*/
2758 lock_res_and_lock(matched);
2759 if (ldlm_extent_equal(&policy->l_extent,
2760 &matched->l_policy_data.l_extent))
2764 unlock_res_and_lock(matched);
2766 ldlm_lock_decref(&lockh, mode);
2767 LDLM_LOCK_PUT(matched);
2769 } else if (osc_set_lock_data(matched, einfo->ei_cbdata)) {
2770 *flags |= LDLM_FL_LVB_READY;
2772 /* We already have a lock, and it's referenced. */
2773 (*upcall)(cookie, &lockh, ELDLM_LOCK_MATCHED);
2775 ldlm_lock_decref(&lockh, mode);
2776 LDLM_LOCK_PUT(matched);
2779 ldlm_lock_decref(&lockh, mode);
2780 LDLM_LOCK_PUT(matched);
2784 if (*flags & (LDLM_FL_TEST_LOCK | LDLM_FL_MATCH_LOCK))
2788 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2789 &RQF_LDLM_ENQUEUE_LVB);
2793 rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
2795 ptlrpc_request_free(req);
2799 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
2801 ptlrpc_request_set_replen(req);
2804 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2805 *flags &= ~LDLM_FL_BLOCK_GRANTED;
2807 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
2808 sizeof(*lvb), LVB_T_OST, &lockh, async);
2811 struct osc_enqueue_args *aa;
2812 aa = ptlrpc_req_async_args(aa, req);
2814 aa->oa_mode = einfo->ei_mode;
2815 aa->oa_type = einfo->ei_type;
2816 lustre_handle_copy(&aa->oa_lockh, &lockh);
2817 aa->oa_upcall = upcall;
2818 aa->oa_cookie = cookie;
2819 aa->oa_speculative = speculative;
2821 aa->oa_flags = flags;
2824 /* speculative locks are essentially to enqueue
2825 * a DLM lock in advance, so we don't care
2826 * about the result of the enqueue. */
2828 aa->oa_flags = NULL;
2831 req->rq_interpret_reply = osc_enqueue_interpret;
2832 ptlrpc_set_add_req(rqset, req);
2833 } else if (intent) {
2834 ptlrpc_req_finished(req);
2839 rc = osc_enqueue_fini(req, upcall, cookie, &lockh, einfo->ei_mode,
2840 flags, speculative, rc);
2842 ptlrpc_req_finished(req);
2847 int osc_match_base(const struct lu_env *env, struct obd_export *exp,
2848 struct ldlm_res_id *res_id, enum ldlm_type type,
2849 union ldlm_policy_data *policy, enum ldlm_mode mode,
2850 __u64 *flags, struct osc_object *obj,
2851 struct lustre_handle *lockh, enum ldlm_match_flags match_flags)
2853 struct obd_device *obd = exp->exp_obd;
2854 __u64 lflags = *flags;
2858 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
2861 /* Filesystem lock extents are extended to page boundaries so that
2862 * dealing with the page cache is a little smoother */
2863 policy->l_extent.start -= policy->l_extent.start & ~PAGE_MASK;
2864 policy->l_extent.end |= ~PAGE_MASK;
2866 /* Next, search for already existing extent locks that will cover us */
2867 /* If we're trying to read, we also search for an existing PW lock. The
2868 * VFS and page cache already protect us locally, so lots of readers/
2869 * writers can share a single PW lock. */
2874 rc = ldlm_lock_match_with_skip(obd->obd_namespace, lflags, 0,
2875 res_id, type, policy, rc, lockh,
2877 if (rc == 0 || lflags & LDLM_FL_TEST_LOCK)
2881 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2883 LASSERT(lock != NULL);
2884 if (osc_set_lock_data(lock, obj)) {
2885 lock_res_and_lock(lock);
2886 if (!ldlm_is_lvb_cached(lock)) {
2887 LASSERT(lock->l_ast_data == obj);
2888 osc_lock_lvb_update(env, obj, lock, NULL);
2889 ldlm_set_lvb_cached(lock);
2891 unlock_res_and_lock(lock);
2893 ldlm_lock_decref(lockh, rc);
2896 LDLM_LOCK_PUT(lock);
2901 static int osc_statfs_interpret(const struct lu_env *env,
2902 struct ptlrpc_request *req, void *args, int rc)
2904 struct osc_async_args *aa = args;
2905 struct obd_statfs *msfs;
2910 * The request has in fact never been sent due to issues at
2911 * a higher level (LOV). Exit immediately since the caller
2912 * is aware of the problem and takes care of the clean up.
2916 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
2917 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
2923 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2925 GOTO(out, rc = -EPROTO);
2927 *aa->aa_oi->oi_osfs = *msfs;
2929 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
2934 static int osc_statfs_async(struct obd_export *exp,
2935 struct obd_info *oinfo, time64_t max_age,
2936 struct ptlrpc_request_set *rqset)
2938 struct obd_device *obd = class_exp2obd(exp);
2939 struct ptlrpc_request *req;
2940 struct osc_async_args *aa;
2944 if (obd->obd_osfs_age >= max_age) {
2946 "%s: use %p cache blocks %llu/%llu objects %llu/%llu\n",
2947 obd->obd_name, &obd->obd_osfs,
2948 obd->obd_osfs.os_bavail, obd->obd_osfs.os_blocks,
2949 obd->obd_osfs.os_ffree, obd->obd_osfs.os_files);
2950 spin_lock(&obd->obd_osfs_lock);
2951 memcpy(oinfo->oi_osfs, &obd->obd_osfs, sizeof(*oinfo->oi_osfs));
2952 spin_unlock(&obd->obd_osfs_lock);
2953 oinfo->oi_flags |= OBD_STATFS_FROM_CACHE;
2954 if (oinfo->oi_cb_up)
2955 oinfo->oi_cb_up(oinfo, 0);
2960 /* We could possibly pass max_age in the request (as an absolute
2961 * timestamp or a "seconds.usec ago") so the target can avoid doing
2962 * extra calls into the filesystem if that isn't necessary (e.g.
2963 * during mount that would help a bit). Having relative timestamps
2964 * is not so great if request processing is slow, while absolute
2965 * timestamps are not ideal because they need time synchronization. */
2966 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
2970 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2972 ptlrpc_request_free(req);
2975 ptlrpc_request_set_replen(req);
2976 req->rq_request_portal = OST_CREATE_PORTAL;
2977 ptlrpc_at_set_req_timeout(req);
2979 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
2980 /* procfs requests not want stat in wait for avoid deadlock */
2981 req->rq_no_resend = 1;
2982 req->rq_no_delay = 1;
2985 req->rq_interpret_reply = osc_statfs_interpret;
2986 aa = ptlrpc_req_async_args(aa, req);
2989 ptlrpc_set_add_req(rqset, req);
2993 static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
2994 struct obd_statfs *osfs, time64_t max_age, __u32 flags)
2996 struct obd_device *obd = class_exp2obd(exp);
2997 struct obd_statfs *msfs;
2998 struct ptlrpc_request *req;
2999 struct obd_import *imp = NULL;
3004 /*Since the request might also come from lprocfs, so we need
3005 *sync this with client_disconnect_export Bug15684*/
3006 down_read(&obd->u.cli.cl_sem);
3007 if (obd->u.cli.cl_import)
3008 imp = class_import_get(obd->u.cli.cl_import);
3009 up_read(&obd->u.cli.cl_sem);
3013 /* We could possibly pass max_age in the request (as an absolute
3014 * timestamp or a "seconds.usec ago") so the target can avoid doing
3015 * extra calls into the filesystem if that isn't necessary (e.g.
3016 * during mount that would help a bit). Having relative timestamps
3017 * is not so great if request processing is slow, while absolute
3018 * timestamps are not ideal because they need time synchronization. */
3019 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
3021 class_import_put(imp);
3026 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3028 ptlrpc_request_free(req);
3031 ptlrpc_request_set_replen(req);
3032 req->rq_request_portal = OST_CREATE_PORTAL;
3033 ptlrpc_at_set_req_timeout(req);
3035 if (flags & OBD_STATFS_NODELAY) {
3036 /* procfs requests not want stat in wait for avoid deadlock */
3037 req->rq_no_resend = 1;
3038 req->rq_no_delay = 1;
3041 rc = ptlrpc_queue_wait(req);
3045 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3047 GOTO(out, rc = -EPROTO);
3053 ptlrpc_req_finished(req);
3057 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
3058 void *karg, void __user *uarg)
3060 struct obd_device *obd = exp->exp_obd;
3061 struct obd_ioctl_data *data = karg;
3065 if (!try_module_get(THIS_MODULE)) {
3066 CERROR("%s: cannot get module '%s'\n", obd->obd_name,
3067 module_name(THIS_MODULE));
3071 case OBD_IOC_CLIENT_RECOVER:
3072 rc = ptlrpc_recover_import(obd->u.cli.cl_import,
3073 data->ioc_inlbuf1, 0);
3077 case IOC_OSC_SET_ACTIVE:
3078 rc = ptlrpc_set_import_active(obd->u.cli.cl_import,
3083 CDEBUG(D_INODE, "%s: unrecognised ioctl %#x by %s: rc = %d\n",
3084 obd->obd_name, cmd, current->comm, rc);
3088 module_put(THIS_MODULE);
3092 int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
3093 u32 keylen, void *key, u32 vallen, void *val,
3094 struct ptlrpc_request_set *set)
3096 struct ptlrpc_request *req;
3097 struct obd_device *obd = exp->exp_obd;
3098 struct obd_import *imp = class_exp2cliimp(exp);
3103 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3105 if (KEY_IS(KEY_CHECKSUM)) {
3106 if (vallen != sizeof(int))
3108 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3112 if (KEY_IS(KEY_SPTLRPC_CONF)) {
3113 sptlrpc_conf_client_adapt(obd);
3117 if (KEY_IS(KEY_FLUSH_CTX)) {
3118 sptlrpc_import_flush_my_ctx(imp);
3122 if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
3123 struct client_obd *cli = &obd->u.cli;
3124 long nr = atomic_long_read(&cli->cl_lru_in_list) >> 1;
3125 long target = *(long *)val;
3127 nr = osc_lru_shrink(env, cli, min(nr, target), true);
3132 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
3135 /* We pass all other commands directly to OST. Since nobody calls osc
3136 methods directly and everybody is supposed to go through LOV, we
3137 assume lov checked invalid values for us.
3138 The only recognised values so far are evict_by_nid and mds_conn.
3139 Even if something bad goes through, we'd get a -EINVAL from OST
3142 req = ptlrpc_request_alloc(imp, KEY_IS(KEY_GRANT_SHRINK) ?
3143 &RQF_OST_SET_GRANT_INFO :
3148 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3149 RCL_CLIENT, keylen);
3150 if (!KEY_IS(KEY_GRANT_SHRINK))
3151 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
3152 RCL_CLIENT, vallen);
3153 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
3155 ptlrpc_request_free(req);
3159 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3160 memcpy(tmp, key, keylen);
3161 tmp = req_capsule_client_get(&req->rq_pill, KEY_IS(KEY_GRANT_SHRINK) ?
3164 memcpy(tmp, val, vallen);
3166 if (KEY_IS(KEY_GRANT_SHRINK)) {
3167 struct osc_grant_args *aa;
3170 aa = ptlrpc_req_async_args(aa, req);
3171 OBD_SLAB_ALLOC_PTR_GFP(oa, osc_obdo_kmem, GFP_NOFS);
3173 ptlrpc_req_finished(req);
3176 *oa = ((struct ost_body *)val)->oa;
3178 req->rq_interpret_reply = osc_shrink_grant_interpret;
3181 ptlrpc_request_set_replen(req);
3182 if (!KEY_IS(KEY_GRANT_SHRINK)) {
3183 LASSERT(set != NULL);
3184 ptlrpc_set_add_req(set, req);
3185 ptlrpc_check_set(NULL, set);
3187 ptlrpcd_add_req(req);
3192 EXPORT_SYMBOL(osc_set_info_async);
3194 int osc_reconnect(const struct lu_env *env, struct obd_export *exp,
3195 struct obd_device *obd, struct obd_uuid *cluuid,
3196 struct obd_connect_data *data, void *localdata)
3198 struct client_obd *cli = &obd->u.cli;
3200 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3204 spin_lock(&cli->cl_loi_list_lock);
3205 grant = cli->cl_avail_grant + cli->cl_reserved_grant;
3206 if (data->ocd_connect_flags & OBD_CONNECT_GRANT_PARAM) {
3207 /* restore ocd_grant_blkbits as client page bits */
3208 data->ocd_grant_blkbits = PAGE_SHIFT;
3209 grant += cli->cl_dirty_grant;
3211 grant += cli->cl_dirty_pages << PAGE_SHIFT;
3213 data->ocd_grant = grant ? : 2 * cli_brw_size(obd);
3214 lost_grant = cli->cl_lost_grant;
3215 cli->cl_lost_grant = 0;
3216 spin_unlock(&cli->cl_loi_list_lock);
3218 CDEBUG(D_RPCTRACE, "ocd_connect_flags: %#llx ocd_version: %d"
3219 " ocd_grant: %d, lost: %ld.\n", data->ocd_connect_flags,
3220 data->ocd_version, data->ocd_grant, lost_grant);
3225 EXPORT_SYMBOL(osc_reconnect);
3227 int osc_disconnect(struct obd_export *exp)
3229 struct obd_device *obd = class_exp2obd(exp);
3232 rc = client_disconnect_export(exp);
3234 * Initially we put del_shrink_grant before disconnect_export, but it
3235 * causes the following problem if setup (connect) and cleanup
3236 * (disconnect) are tangled together.
3237 * connect p1 disconnect p2
3238 * ptlrpc_connect_import
3239 * ............... class_manual_cleanup
3242 * ptlrpc_connect_interrupt
3244 * add this client to shrink list
3246 * Bang! grant shrink thread trigger the shrink. BUG18662
3248 osc_del_grant_list(&obd->u.cli);
3251 EXPORT_SYMBOL(osc_disconnect);
3253 int osc_ldlm_resource_invalidate(struct cfs_hash *hs, struct cfs_hash_bd *bd,
3254 struct hlist_node *hnode, void *arg)
3256 struct lu_env *env = arg;
3257 struct ldlm_resource *res = cfs_hash_object(hs, hnode);
3258 struct ldlm_lock *lock;
3259 struct osc_object *osc = NULL;
3263 list_for_each_entry(lock, &res->lr_granted, l_res_link) {
3264 if (lock->l_ast_data != NULL && osc == NULL) {
3265 osc = lock->l_ast_data;
3266 cl_object_get(osc2cl(osc));
3269 /* clear LDLM_FL_CLEANED flag to make sure it will be canceled
3270 * by the 2nd round of ldlm_namespace_clean() call in
3271 * osc_import_event(). */
3272 ldlm_clear_cleaned(lock);
3277 osc_object_invalidate(env, osc);
3278 cl_object_put(env, osc2cl(osc));
3283 EXPORT_SYMBOL(osc_ldlm_resource_invalidate);
3285 static int osc_import_event(struct obd_device *obd,
3286 struct obd_import *imp,
3287 enum obd_import_event event)
3289 struct client_obd *cli;
3293 LASSERT(imp->imp_obd == obd);
3296 case IMP_EVENT_DISCON: {
3298 spin_lock(&cli->cl_loi_list_lock);
3299 cli->cl_avail_grant = 0;
3300 cli->cl_lost_grant = 0;
3301 spin_unlock(&cli->cl_loi_list_lock);
3304 case IMP_EVENT_INACTIVE: {
3305 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE);
3308 case IMP_EVENT_INVALIDATE: {
3309 struct ldlm_namespace *ns = obd->obd_namespace;
3313 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3315 env = cl_env_get(&refcheck);
3317 osc_io_unplug(env, &obd->u.cli, NULL);
3319 cfs_hash_for_each_nolock(ns->ns_rs_hash,
3320 osc_ldlm_resource_invalidate,
3322 cl_env_put(env, &refcheck);
3324 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3329 case IMP_EVENT_ACTIVE: {
3330 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE);
3333 case IMP_EVENT_OCD: {
3334 struct obd_connect_data *ocd = &imp->imp_connect_data;
3336 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3337 osc_init_grant(&obd->u.cli, ocd);
3340 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3341 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3343 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD);
3346 case IMP_EVENT_DEACTIVATE: {
3347 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_DEACTIVATE);
3350 case IMP_EVENT_ACTIVATE: {
3351 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVATE);
3355 CERROR("Unknown import event %d\n", event);
3362 * Determine whether the lock can be canceled before replaying the lock
3363 * during recovery, see bug16774 for detailed information.
3365 * \retval zero the lock can't be canceled
3366 * \retval other ok to cancel
3368 static int osc_cancel_weight(struct ldlm_lock *lock)
3371 * Cancel all unused and granted extent lock.
3373 if (lock->l_resource->lr_type == LDLM_EXTENT &&
3374 ldlm_is_granted(lock) &&
3375 osc_ldlm_weigh_ast(lock) == 0)
3381 static int brw_queue_work(const struct lu_env *env, void *data)
3383 struct client_obd *cli = data;
3385 CDEBUG(D_CACHE, "Run writeback work for client obd %p.\n", cli);
3387 osc_io_unplug(env, cli, NULL);
3391 int osc_setup_common(struct obd_device *obd, struct lustre_cfg *lcfg)
3393 struct client_obd *cli = &obd->u.cli;
3399 rc = ptlrpcd_addref();
3403 rc = client_obd_setup(obd, lcfg);
3405 GOTO(out_ptlrpcd, rc);
3408 handler = ptlrpcd_alloc_work(cli->cl_import, brw_queue_work, cli);
3409 if (IS_ERR(handler))
3410 GOTO(out_ptlrpcd_work, rc = PTR_ERR(handler));
3411 cli->cl_writeback_work = handler;
3413 handler = ptlrpcd_alloc_work(cli->cl_import, lru_queue_work, cli);
3414 if (IS_ERR(handler))
3415 GOTO(out_ptlrpcd_work, rc = PTR_ERR(handler));
3416 cli->cl_lru_work = handler;
3418 rc = osc_quota_setup(obd);
3420 GOTO(out_ptlrpcd_work, rc);
3422 cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
3423 osc_update_next_shrink(cli);
3428 if (cli->cl_writeback_work != NULL) {
3429 ptlrpcd_destroy_work(cli->cl_writeback_work);
3430 cli->cl_writeback_work = NULL;
3432 if (cli->cl_lru_work != NULL) {
3433 ptlrpcd_destroy_work(cli->cl_lru_work);
3434 cli->cl_lru_work = NULL;
3436 client_obd_cleanup(obd);
3441 EXPORT_SYMBOL(osc_setup_common);
3443 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3445 struct client_obd *cli = &obd->u.cli;
3453 rc = osc_setup_common(obd, lcfg);
3457 rc = osc_tunables_init(obd);
3462 * We try to control the total number of requests with a upper limit
3463 * osc_reqpool_maxreqcount. There might be some race which will cause
3464 * over-limit allocation, but it is fine.
3466 req_count = atomic_read(&osc_pool_req_count);
3467 if (req_count < osc_reqpool_maxreqcount) {
3468 adding = cli->cl_max_rpcs_in_flight + 2;
3469 if (req_count + adding > osc_reqpool_maxreqcount)
3470 adding = osc_reqpool_maxreqcount - req_count;
3472 added = ptlrpc_add_rqs_to_pool(osc_rq_pool, adding);
3473 atomic_add(added, &osc_pool_req_count);
3476 ns_register_cancel(obd->obd_namespace, osc_cancel_weight);
3478 spin_lock(&osc_shrink_lock);
3479 list_add_tail(&cli->cl_shrink_list, &osc_shrink_list);
3480 spin_unlock(&osc_shrink_lock);
3481 cli->cl_import->imp_idle_timeout = osc_idle_timeout;
3482 cli->cl_import->imp_idle_debug = D_HA;
3487 int osc_precleanup_common(struct obd_device *obd)
3489 struct client_obd *cli = &obd->u.cli;
3493 * for echo client, export may be on zombie list, wait for
3494 * zombie thread to cull it, because cli.cl_import will be
3495 * cleared in client_disconnect_export():
3496 * class_export_destroy() -> obd_cleanup() ->
3497 * echo_device_free() -> echo_client_cleanup() ->
3498 * obd_disconnect() -> osc_disconnect() ->
3499 * client_disconnect_export()
3501 obd_zombie_barrier();
3502 if (cli->cl_writeback_work) {
3503 ptlrpcd_destroy_work(cli->cl_writeback_work);
3504 cli->cl_writeback_work = NULL;
3507 if (cli->cl_lru_work) {
3508 ptlrpcd_destroy_work(cli->cl_lru_work);
3509 cli->cl_lru_work = NULL;
3512 obd_cleanup_client_import(obd);
3515 EXPORT_SYMBOL(osc_precleanup_common);
3517 static int osc_precleanup(struct obd_device *obd)
3521 osc_precleanup_common(obd);
3523 ptlrpc_lprocfs_unregister_obd(obd);
3527 int osc_cleanup_common(struct obd_device *obd)
3529 struct client_obd *cli = &obd->u.cli;
3534 spin_lock(&osc_shrink_lock);
3535 list_del(&cli->cl_shrink_list);
3536 spin_unlock(&osc_shrink_lock);
3539 if (cli->cl_cache != NULL) {
3540 LASSERT(atomic_read(&cli->cl_cache->ccc_users) > 0);
3541 spin_lock(&cli->cl_cache->ccc_lru_lock);
3542 list_del_init(&cli->cl_lru_osc);
3543 spin_unlock(&cli->cl_cache->ccc_lru_lock);
3544 cli->cl_lru_left = NULL;
3545 cl_cache_decref(cli->cl_cache);
3546 cli->cl_cache = NULL;
3549 /* free memory of osc quota cache */
3550 osc_quota_cleanup(obd);
3552 rc = client_obd_cleanup(obd);
3557 EXPORT_SYMBOL(osc_cleanup_common);
3559 static const struct obd_ops osc_obd_ops = {
3560 .o_owner = THIS_MODULE,
3561 .o_setup = osc_setup,
3562 .o_precleanup = osc_precleanup,
3563 .o_cleanup = osc_cleanup_common,
3564 .o_add_conn = client_import_add_conn,
3565 .o_del_conn = client_import_del_conn,
3566 .o_connect = client_connect_import,
3567 .o_reconnect = osc_reconnect,
3568 .o_disconnect = osc_disconnect,
3569 .o_statfs = osc_statfs,
3570 .o_statfs_async = osc_statfs_async,
3571 .o_create = osc_create,
3572 .o_destroy = osc_destroy,
3573 .o_getattr = osc_getattr,
3574 .o_setattr = osc_setattr,
3575 .o_iocontrol = osc_iocontrol,
3576 .o_set_info_async = osc_set_info_async,
3577 .o_import_event = osc_import_event,
3578 .o_quotactl = osc_quotactl,
3581 static struct shrinker *osc_cache_shrinker;
3582 LIST_HEAD(osc_shrink_list);
3583 DEFINE_SPINLOCK(osc_shrink_lock);
3585 #ifndef HAVE_SHRINKER_COUNT
3586 static int osc_cache_shrink(SHRINKER_ARGS(sc, nr_to_scan, gfp_mask))
3588 struct shrink_control scv = {
3589 .nr_to_scan = shrink_param(sc, nr_to_scan),
3590 .gfp_mask = shrink_param(sc, gfp_mask)
3592 (void)osc_cache_shrink_scan(shrinker, &scv);
3594 return osc_cache_shrink_count(shrinker, &scv);
3598 static int __init osc_init(void)
3600 unsigned int reqpool_size;
3601 unsigned int reqsize;
3603 DEF_SHRINKER_VAR(osc_shvar, osc_cache_shrink,
3604 osc_cache_shrink_count, osc_cache_shrink_scan);
3607 /* print an address of _any_ initialized kernel symbol from this
3608 * module, to allow debugging with gdb that doesn't support data
3609 * symbols from modules.*/
3610 CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
3612 rc = lu_kmem_init(osc_caches);
3616 rc = class_register_type(&osc_obd_ops, NULL, true, NULL,
3617 LUSTRE_OSC_NAME, &osc_device_type);
3621 osc_cache_shrinker = set_shrinker(DEFAULT_SEEKS, &osc_shvar);
3623 /* This is obviously too much memory, only prevent overflow here */
3624 if (osc_reqpool_mem_max >= 1 << 12 || osc_reqpool_mem_max == 0)
3625 GOTO(out_type, rc = -EINVAL);
3627 reqpool_size = osc_reqpool_mem_max << 20;
3630 while (reqsize < OST_IO_MAXREQSIZE)
3631 reqsize = reqsize << 1;
3634 * We don't enlarge the request count in OSC pool according to
3635 * cl_max_rpcs_in_flight. The allocation from the pool will only be
3636 * tried after normal allocation failed. So a small OSC pool won't
3637 * cause much performance degression in most of cases.
3639 osc_reqpool_maxreqcount = reqpool_size / reqsize;
3641 atomic_set(&osc_pool_req_count, 0);
3642 osc_rq_pool = ptlrpc_init_rq_pool(0, OST_IO_MAXREQSIZE,
3643 ptlrpc_add_rqs_to_pool);
3645 if (osc_rq_pool == NULL)
3646 GOTO(out_type, rc = -ENOMEM);
3648 rc = osc_start_grant_work();
3650 GOTO(out_req_pool, rc);
3655 ptlrpc_free_rq_pool(osc_rq_pool);
3657 class_unregister_type(LUSTRE_OSC_NAME);
3659 lu_kmem_fini(osc_caches);
3664 static void __exit osc_exit(void)
3666 osc_stop_grant_work();
3667 remove_shrinker(osc_cache_shrinker);
3668 class_unregister_type(LUSTRE_OSC_NAME);
3669 lu_kmem_fini(osc_caches);
3670 ptlrpc_free_rq_pool(osc_rq_pool);
3673 MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
3674 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3675 MODULE_VERSION(LUSTRE_VERSION_STRING);
3676 MODULE_LICENSE("GPL");
3678 module_init(osc_init);
3679 module_exit(osc_exit);