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.sun.com/software/products/lustre/docs/GPLv2.pdf
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
27 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
30 * Copyright (c) 2011, 2012, Whamcloud, Inc.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
37 #define DEBUG_SUBSYSTEM S_OSC
39 #include <libcfs/libcfs.h>
42 # include <liblustre.h>
45 #include <lustre_dlm.h>
46 #include <lustre_net.h>
47 #include <lustre/lustre_user.h>
48 #include <obd_cksum.h>
56 #include <lustre_ha.h>
57 #include <lprocfs_status.h>
58 #include <lustre_log.h>
59 #include <lustre_debug.h>
60 #include <lustre_param.h>
61 #include "osc_internal.h"
62 #include "osc_cl_internal.h"
64 static void osc_release_ppga(struct brw_page **ppga, obd_count count);
65 static int brw_interpret(const struct lu_env *env,
66 struct ptlrpc_request *req, void *data, int rc);
67 int osc_cleanup(struct obd_device *obd);
69 /* Pack OSC object metadata for disk storage (LE byte order). */
70 static int osc_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,
71 struct lov_stripe_md *lsm)
76 lmm_size = sizeof(**lmmp);
81 OBD_FREE(*lmmp, lmm_size);
87 OBD_ALLOC(*lmmp, lmm_size);
93 LASSERT(lsm->lsm_object_id);
94 LASSERT_SEQ_IS_MDT(lsm->lsm_object_seq);
95 (*lmmp)->lmm_object_id = cpu_to_le64(lsm->lsm_object_id);
96 (*lmmp)->lmm_object_seq = cpu_to_le64(lsm->lsm_object_seq);
102 /* Unpack OSC object metadata from disk storage (LE byte order). */
103 static int osc_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
104 struct lov_mds_md *lmm, int lmm_bytes)
107 struct obd_import *imp = class_exp2cliimp(exp);
111 if (lmm_bytes < sizeof (*lmm)) {
112 CERROR("lov_mds_md too small: %d, need %d\n",
113 lmm_bytes, (int)sizeof(*lmm));
116 /* XXX LOV_MAGIC etc check? */
118 if (lmm->lmm_object_id == 0) {
119 CERROR("lov_mds_md: zero lmm_object_id\n");
124 lsm_size = lov_stripe_md_size(1);
128 if (*lsmp != NULL && lmm == NULL) {
129 OBD_FREE((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
130 OBD_FREE(*lsmp, lsm_size);
136 OBD_ALLOC(*lsmp, lsm_size);
139 OBD_ALLOC((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
140 if ((*lsmp)->lsm_oinfo[0] == NULL) {
141 OBD_FREE(*lsmp, lsm_size);
144 loi_init((*lsmp)->lsm_oinfo[0]);
148 /* XXX zero *lsmp? */
149 (*lsmp)->lsm_object_id = le64_to_cpu (lmm->lmm_object_id);
150 (*lsmp)->lsm_object_seq = le64_to_cpu (lmm->lmm_object_seq);
151 LASSERT((*lsmp)->lsm_object_id);
152 LASSERT_SEQ_IS_MDT((*lsmp)->lsm_object_seq);
156 (imp->imp_connect_data.ocd_connect_flags & OBD_CONNECT_MAXBYTES))
157 (*lsmp)->lsm_maxbytes = imp->imp_connect_data.ocd_maxbytes;
159 (*lsmp)->lsm_maxbytes = LUSTRE_STRIPE_MAXBYTES;
164 static inline void osc_pack_capa(struct ptlrpc_request *req,
165 struct ost_body *body, void *capa)
167 struct obd_capa *oc = (struct obd_capa *)capa;
168 struct lustre_capa *c;
173 c = req_capsule_client_get(&req->rq_pill, &RMF_CAPA1);
176 body->oa.o_valid |= OBD_MD_FLOSSCAPA;
177 DEBUG_CAPA(D_SEC, c, "pack");
180 static inline void osc_pack_req_body(struct ptlrpc_request *req,
181 struct obd_info *oinfo)
183 struct ost_body *body;
185 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
188 lustre_set_wire_obdo(&body->oa, oinfo->oi_oa);
189 osc_pack_capa(req, body, oinfo->oi_capa);
192 static inline void osc_set_capa_size(struct ptlrpc_request *req,
193 const struct req_msg_field *field,
197 req_capsule_set_size(&req->rq_pill, field, RCL_CLIENT, 0);
199 /* it is already calculated as sizeof struct obd_capa */
203 static int osc_getattr_interpret(const struct lu_env *env,
204 struct ptlrpc_request *req,
205 struct osc_async_args *aa, int rc)
207 struct ost_body *body;
213 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
215 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
216 lustre_get_wire_obdo(aa->aa_oi->oi_oa, &body->oa);
218 /* This should really be sent by the OST */
219 aa->aa_oi->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
220 aa->aa_oi->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
222 CDEBUG(D_INFO, "can't unpack ost_body\n");
224 aa->aa_oi->oi_oa->o_valid = 0;
227 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
231 static int osc_getattr_async(struct obd_export *exp, struct obd_info *oinfo,
232 struct ptlrpc_request_set *set)
234 struct ptlrpc_request *req;
235 struct osc_async_args *aa;
239 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
243 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
244 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
246 ptlrpc_request_free(req);
250 osc_pack_req_body(req, oinfo);
252 ptlrpc_request_set_replen(req);
253 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_getattr_interpret;
255 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
256 aa = ptlrpc_req_async_args(req);
259 ptlrpc_set_add_req(set, req);
263 static int osc_getattr(const struct lu_env *env, struct obd_export *exp,
264 struct obd_info *oinfo)
266 struct ptlrpc_request *req;
267 struct ost_body *body;
271 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
275 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
276 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
278 ptlrpc_request_free(req);
282 osc_pack_req_body(req, oinfo);
284 ptlrpc_request_set_replen(req);
286 rc = ptlrpc_queue_wait(req);
290 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
292 GOTO(out, rc = -EPROTO);
294 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
295 lustre_get_wire_obdo(oinfo->oi_oa, &body->oa);
297 /* This should really be sent by the OST */
298 oinfo->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
299 oinfo->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
303 ptlrpc_req_finished(req);
307 static int osc_setattr(const struct lu_env *env, struct obd_export *exp,
308 struct obd_info *oinfo, struct obd_trans_info *oti)
310 struct ptlrpc_request *req;
311 struct ost_body *body;
315 LASSERT(oinfo->oi_oa->o_valid & OBD_MD_FLGROUP);
317 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
321 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
322 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
324 ptlrpc_request_free(req);
328 osc_pack_req_body(req, oinfo);
330 ptlrpc_request_set_replen(req);
332 rc = ptlrpc_queue_wait(req);
336 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
338 GOTO(out, rc = -EPROTO);
340 lustre_get_wire_obdo(oinfo->oi_oa, &body->oa);
344 ptlrpc_req_finished(req);
348 static int osc_setattr_interpret(const struct lu_env *env,
349 struct ptlrpc_request *req,
350 struct osc_setattr_args *sa, int rc)
352 struct ost_body *body;
358 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
360 GOTO(out, rc = -EPROTO);
362 lustre_get_wire_obdo(sa->sa_oa, &body->oa);
364 rc = sa->sa_upcall(sa->sa_cookie, rc);
368 int osc_setattr_async_base(struct obd_export *exp, struct obd_info *oinfo,
369 struct obd_trans_info *oti,
370 obd_enqueue_update_f upcall, void *cookie,
371 struct ptlrpc_request_set *rqset)
373 struct ptlrpc_request *req;
374 struct osc_setattr_args *sa;
378 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
382 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
383 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
385 ptlrpc_request_free(req);
389 if (oti && oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE)
390 oinfo->oi_oa->o_lcookie = *oti->oti_logcookies;
392 osc_pack_req_body(req, oinfo);
394 ptlrpc_request_set_replen(req);
396 /* do mds to ost setattr asynchronously */
398 /* Do not wait for response. */
399 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
401 req->rq_interpret_reply =
402 (ptlrpc_interpterer_t)osc_setattr_interpret;
404 CLASSERT (sizeof(*sa) <= sizeof(req->rq_async_args));
405 sa = ptlrpc_req_async_args(req);
406 sa->sa_oa = oinfo->oi_oa;
407 sa->sa_upcall = upcall;
408 sa->sa_cookie = cookie;
410 if (rqset == PTLRPCD_SET)
411 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
413 ptlrpc_set_add_req(rqset, req);
419 static int osc_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
420 struct obd_trans_info *oti,
421 struct ptlrpc_request_set *rqset)
423 return osc_setattr_async_base(exp, oinfo, oti,
424 oinfo->oi_cb_up, oinfo, rqset);
427 int osc_real_create(struct obd_export *exp, struct obdo *oa,
428 struct lov_stripe_md **ea, struct obd_trans_info *oti)
430 struct ptlrpc_request *req;
431 struct ost_body *body;
432 struct lov_stripe_md *lsm;
441 rc = obd_alloc_memmd(exp, &lsm);
446 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
448 GOTO(out, rc = -ENOMEM);
450 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
452 ptlrpc_request_free(req);
456 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
458 lustre_set_wire_obdo(&body->oa, oa);
460 ptlrpc_request_set_replen(req);
462 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
463 oa->o_flags == OBD_FL_DELORPHAN) {
465 "delorphan from OST integration");
466 /* Don't resend the delorphan req */
467 req->rq_no_resend = req->rq_no_delay = 1;
470 rc = ptlrpc_queue_wait(req);
474 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
476 GOTO(out_req, rc = -EPROTO);
478 lustre_get_wire_obdo(oa, &body->oa);
480 /* This should really be sent by the OST */
481 oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
482 oa->o_valid |= OBD_MD_FLBLKSZ;
484 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
485 * have valid lsm_oinfo data structs, so don't go touching that.
486 * This needs to be fixed in a big way.
488 lsm->lsm_object_id = oa->o_id;
489 lsm->lsm_object_seq = oa->o_seq;
493 oti->oti_transno = lustre_msg_get_transno(req->rq_repmsg);
495 if (oa->o_valid & OBD_MD_FLCOOKIE) {
496 if (!oti->oti_logcookies)
497 oti_alloc_cookies(oti, 1);
498 *oti->oti_logcookies = oa->o_lcookie;
502 CDEBUG(D_HA, "transno: "LPD64"\n",
503 lustre_msg_get_transno(req->rq_repmsg));
505 ptlrpc_req_finished(req);
508 obd_free_memmd(exp, &lsm);
512 int osc_punch_base(struct obd_export *exp, struct obd_info *oinfo,
513 obd_enqueue_update_f upcall, void *cookie,
514 struct ptlrpc_request_set *rqset)
516 struct ptlrpc_request *req;
517 struct osc_setattr_args *sa;
518 struct ost_body *body;
522 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_PUNCH);
526 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
527 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
529 ptlrpc_request_free(req);
532 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
533 ptlrpc_at_set_req_timeout(req);
535 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
537 lustre_set_wire_obdo(&body->oa, oinfo->oi_oa);
538 osc_pack_capa(req, body, oinfo->oi_capa);
540 ptlrpc_request_set_replen(req);
542 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_setattr_interpret;
543 CLASSERT (sizeof(*sa) <= sizeof(req->rq_async_args));
544 sa = ptlrpc_req_async_args(req);
545 sa->sa_oa = oinfo->oi_oa;
546 sa->sa_upcall = upcall;
547 sa->sa_cookie = cookie;
548 if (rqset == PTLRPCD_SET)
549 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
551 ptlrpc_set_add_req(rqset, req);
556 static int osc_punch(const struct lu_env *env, struct obd_export *exp,
557 struct obd_info *oinfo, struct obd_trans_info *oti,
558 struct ptlrpc_request_set *rqset)
560 oinfo->oi_oa->o_size = oinfo->oi_policy.l_extent.start;
561 oinfo->oi_oa->o_blocks = oinfo->oi_policy.l_extent.end;
562 oinfo->oi_oa->o_valid |= OBD_MD_FLSIZE | OBD_MD_FLBLOCKS;
563 return osc_punch_base(exp, oinfo,
564 oinfo->oi_cb_up, oinfo, rqset);
567 static int osc_sync_interpret(const struct lu_env *env,
568 struct ptlrpc_request *req,
571 struct osc_fsync_args *fa = arg;
572 struct ost_body *body;
578 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
580 CERROR ("can't unpack ost_body\n");
581 GOTO(out, rc = -EPROTO);
584 *fa->fa_oi->oi_oa = body->oa;
586 rc = fa->fa_upcall(fa->fa_cookie, rc);
590 int osc_sync_base(struct obd_export *exp, struct obd_info *oinfo,
591 obd_enqueue_update_f upcall, void *cookie,
592 struct ptlrpc_request_set *rqset)
594 struct ptlrpc_request *req;
595 struct ost_body *body;
596 struct osc_fsync_args *fa;
600 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
604 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
605 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
607 ptlrpc_request_free(req);
611 /* overload the size and blocks fields in the oa with start/end */
612 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
614 lustre_set_wire_obdo(&body->oa, oinfo->oi_oa);
615 osc_pack_capa(req, body, oinfo->oi_capa);
617 ptlrpc_request_set_replen(req);
618 req->rq_interpret_reply = osc_sync_interpret;
620 CLASSERT(sizeof(*fa) <= sizeof(req->rq_async_args));
621 fa = ptlrpc_req_async_args(req);
623 fa->fa_upcall = upcall;
624 fa->fa_cookie = cookie;
626 if (rqset == PTLRPCD_SET)
627 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
629 ptlrpc_set_add_req(rqset, req);
634 static int osc_sync(const struct lu_env *env, struct obd_export *exp,
635 struct obd_info *oinfo, obd_size start, obd_size end,
636 struct ptlrpc_request_set *set)
641 CDEBUG(D_INFO, "oa NULL\n");
645 oinfo->oi_oa->o_size = start;
646 oinfo->oi_oa->o_blocks = end;
647 oinfo->oi_oa->o_valid |= (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS);
649 RETURN(osc_sync_base(exp, oinfo, oinfo->oi_cb_up, oinfo, set));
652 /* Find and cancel locally locks matched by @mode in the resource found by
653 * @objid. Found locks are added into @cancel list. Returns the amount of
654 * locks added to @cancels list. */
655 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
657 ldlm_mode_t mode, int lock_flags)
659 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
660 struct ldlm_res_id res_id;
661 struct ldlm_resource *res;
665 osc_build_res_name(oa->o_id, oa->o_seq, &res_id);
666 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
670 LDLM_RESOURCE_ADDREF(res);
671 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
672 lock_flags, 0, NULL);
673 LDLM_RESOURCE_DELREF(res);
674 ldlm_resource_putref(res);
678 static int osc_destroy_interpret(const struct lu_env *env,
679 struct ptlrpc_request *req, void *data,
682 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
684 cfs_atomic_dec(&cli->cl_destroy_in_flight);
685 cfs_waitq_signal(&cli->cl_destroy_waitq);
689 static int osc_can_send_destroy(struct client_obd *cli)
691 if (cfs_atomic_inc_return(&cli->cl_destroy_in_flight) <=
692 cli->cl_max_rpcs_in_flight) {
693 /* The destroy request can be sent */
696 if (cfs_atomic_dec_return(&cli->cl_destroy_in_flight) <
697 cli->cl_max_rpcs_in_flight) {
699 * The counter has been modified between the two atomic
702 cfs_waitq_signal(&cli->cl_destroy_waitq);
707 int osc_create(const struct lu_env *env, struct obd_export *exp,
708 struct obdo *oa, struct lov_stripe_md **ea,
709 struct obd_trans_info *oti)
716 LASSERT(oa->o_valid & OBD_MD_FLGROUP);
718 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
719 oa->o_flags == OBD_FL_RECREATE_OBJS) {
720 RETURN(osc_real_create(exp, oa, ea, oti));
723 if (!fid_seq_is_mdt(oa->o_seq))
724 RETURN(osc_real_create(exp, oa, ea, oti));
726 /* we should not get here anymore */
732 /* Destroy requests can be async always on the client, and we don't even really
733 * care about the return code since the client cannot do anything at all about
735 * When the MDS is unlinking a filename, it saves the file objects into a
736 * recovery llog, and these object records are cancelled when the OST reports
737 * they were destroyed and sync'd to disk (i.e. transaction committed).
738 * If the client dies, or the OST is down when the object should be destroyed,
739 * the records are not cancelled, and when the OST reconnects to the MDS next,
740 * it will retrieve the llog unlink logs and then sends the log cancellation
741 * cookies to the MDS after committing destroy transactions. */
742 static int osc_destroy(const struct lu_env *env, struct obd_export *exp,
743 struct obdo *oa, struct lov_stripe_md *ea,
744 struct obd_trans_info *oti, struct obd_export *md_export,
747 struct client_obd *cli = &exp->exp_obd->u.cli;
748 struct ptlrpc_request *req;
749 struct ost_body *body;
750 CFS_LIST_HEAD(cancels);
755 CDEBUG(D_INFO, "oa NULL\n");
759 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
760 LDLM_FL_DISCARD_DATA);
762 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
764 ldlm_lock_list_put(&cancels, l_bl_ast, count);
768 osc_set_capa_size(req, &RMF_CAPA1, (struct obd_capa *)capa);
769 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
772 ptlrpc_request_free(req);
776 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
777 ptlrpc_at_set_req_timeout(req);
779 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
780 oa->o_lcookie = *oti->oti_logcookies;
781 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
783 lustre_set_wire_obdo(&body->oa, oa);
785 osc_pack_capa(req, body, (struct obd_capa *)capa);
786 ptlrpc_request_set_replen(req);
788 /* If osc_destory is for destroying the unlink orphan,
789 * sent from MDT to OST, which should not be blocked here,
790 * because the process might be triggered by ptlrpcd, and
791 * it is not good to block ptlrpcd thread (b=16006)*/
792 if (!(oa->o_flags & OBD_FL_DELORPHAN)) {
793 req->rq_interpret_reply = osc_destroy_interpret;
794 if (!osc_can_send_destroy(cli)) {
795 struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP,
799 * Wait until the number of on-going destroy RPCs drops
800 * under max_rpc_in_flight
802 l_wait_event_exclusive(cli->cl_destroy_waitq,
803 osc_can_send_destroy(cli), &lwi);
807 /* Do not wait for response */
808 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
812 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
815 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
817 LASSERT(!(oa->o_valid & bits));
820 client_obd_list_lock(&cli->cl_loi_list_lock);
821 oa->o_dirty = cli->cl_dirty;
822 if (cli->cl_dirty - cli->cl_dirty_transit > cli->cl_dirty_max) {
823 CERROR("dirty %lu - %lu > dirty_max %lu\n",
824 cli->cl_dirty, cli->cl_dirty_transit, cli->cl_dirty_max);
826 } else if (cfs_atomic_read(&obd_dirty_pages) -
827 cfs_atomic_read(&obd_dirty_transit_pages) >
828 obd_max_dirty_pages + 1){
829 /* The cfs_atomic_read() allowing the cfs_atomic_inc() are
830 * not covered by a lock thus they may safely race and trip
831 * this CERROR() unless we add in a small fudge factor (+1). */
832 CERROR("dirty %d - %d > system dirty_max %d\n",
833 cfs_atomic_read(&obd_dirty_pages),
834 cfs_atomic_read(&obd_dirty_transit_pages),
835 obd_max_dirty_pages);
837 } else if (cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff) {
838 CERROR("dirty %lu - dirty_max %lu too big???\n",
839 cli->cl_dirty, cli->cl_dirty_max);
842 long max_in_flight = (cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT)*
843 (cli->cl_max_rpcs_in_flight + 1);
844 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
846 oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
847 oa->o_dropped = cli->cl_lost_grant;
848 cli->cl_lost_grant = 0;
849 client_obd_list_unlock(&cli->cl_loi_list_lock);
850 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
851 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
855 void osc_update_next_shrink(struct client_obd *cli)
857 cli->cl_next_shrink_grant =
858 cfs_time_shift(cli->cl_grant_shrink_interval);
859 CDEBUG(D_CACHE, "next time %ld to shrink grant \n",
860 cli->cl_next_shrink_grant);
863 static void __osc_update_grant(struct client_obd *cli, obd_size grant)
865 client_obd_list_lock(&cli->cl_loi_list_lock);
866 cli->cl_avail_grant += grant;
867 client_obd_list_unlock(&cli->cl_loi_list_lock);
870 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
872 if (body->oa.o_valid & OBD_MD_FLGRANT) {
873 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
874 __osc_update_grant(cli, body->oa.o_grant);
878 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
879 obd_count keylen, void *key, obd_count vallen,
880 void *val, struct ptlrpc_request_set *set);
882 static int osc_shrink_grant_interpret(const struct lu_env *env,
883 struct ptlrpc_request *req,
886 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
887 struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
888 struct ost_body *body;
891 __osc_update_grant(cli, oa->o_grant);
895 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
897 osc_update_grant(cli, body);
903 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
905 client_obd_list_lock(&cli->cl_loi_list_lock);
906 oa->o_grant = cli->cl_avail_grant / 4;
907 cli->cl_avail_grant -= oa->o_grant;
908 client_obd_list_unlock(&cli->cl_loi_list_lock);
909 if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
910 oa->o_valid |= OBD_MD_FLFLAGS;
913 oa->o_flags |= OBD_FL_SHRINK_GRANT;
914 osc_update_next_shrink(cli);
917 /* Shrink the current grant, either from some large amount to enough for a
918 * full set of in-flight RPCs, or if we have already shrunk to that limit
919 * then to enough for a single RPC. This avoids keeping more grant than
920 * needed, and avoids shrinking the grant piecemeal. */
921 static int osc_shrink_grant(struct client_obd *cli)
923 long target = (cli->cl_max_rpcs_in_flight + 1) *
924 cli->cl_max_pages_per_rpc;
926 client_obd_list_lock(&cli->cl_loi_list_lock);
927 if (cli->cl_avail_grant <= target)
928 target = cli->cl_max_pages_per_rpc;
929 client_obd_list_unlock(&cli->cl_loi_list_lock);
931 return osc_shrink_grant_to_target(cli, target);
934 int osc_shrink_grant_to_target(struct client_obd *cli, long target)
937 struct ost_body *body;
940 client_obd_list_lock(&cli->cl_loi_list_lock);
941 /* Don't shrink if we are already above or below the desired limit
942 * We don't want to shrink below a single RPC, as that will negatively
943 * impact block allocation and long-term performance. */
944 if (target < cli->cl_max_pages_per_rpc)
945 target = cli->cl_max_pages_per_rpc;
947 if (target >= cli->cl_avail_grant) {
948 client_obd_list_unlock(&cli->cl_loi_list_lock);
951 client_obd_list_unlock(&cli->cl_loi_list_lock);
957 osc_announce_cached(cli, &body->oa, 0);
959 client_obd_list_lock(&cli->cl_loi_list_lock);
960 body->oa.o_grant = cli->cl_avail_grant - target;
961 cli->cl_avail_grant = target;
962 client_obd_list_unlock(&cli->cl_loi_list_lock);
963 if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
964 body->oa.o_valid |= OBD_MD_FLFLAGS;
965 body->oa.o_flags = 0;
967 body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
968 osc_update_next_shrink(cli);
970 rc = osc_set_info_async(NULL, cli->cl_import->imp_obd->obd_self_export,
971 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
972 sizeof(*body), body, NULL);
974 __osc_update_grant(cli, body->oa.o_grant);
979 #define GRANT_SHRINK_LIMIT PTLRPC_MAX_BRW_SIZE
980 static int osc_should_shrink_grant(struct client_obd *client)
982 cfs_time_t time = cfs_time_current();
983 cfs_time_t next_shrink = client->cl_next_shrink_grant;
985 if ((client->cl_import->imp_connect_data.ocd_connect_flags &
986 OBD_CONNECT_GRANT_SHRINK) == 0)
989 if (cfs_time_aftereq(time, next_shrink - 5 * CFS_TICK)) {
990 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
991 client->cl_avail_grant > GRANT_SHRINK_LIMIT)
994 osc_update_next_shrink(client);
999 static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
1001 struct client_obd *client;
1003 cfs_list_for_each_entry(client, &item->ti_obd_list,
1004 cl_grant_shrink_list) {
1005 if (osc_should_shrink_grant(client))
1006 osc_shrink_grant(client);
1011 static int osc_add_shrink_grant(struct client_obd *client)
1015 rc = ptlrpc_add_timeout_client(client->cl_grant_shrink_interval,
1017 osc_grant_shrink_grant_cb, NULL,
1018 &client->cl_grant_shrink_list);
1020 CERROR("add grant client %s error %d\n",
1021 client->cl_import->imp_obd->obd_name, rc);
1024 CDEBUG(D_CACHE, "add grant client %s \n",
1025 client->cl_import->imp_obd->obd_name);
1026 osc_update_next_shrink(client);
1030 static int osc_del_shrink_grant(struct client_obd *client)
1032 return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
1036 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1039 * ocd_grant is the total grant amount we're expect to hold: if we've
1040 * been evicted, it's the new avail_grant amount, cl_dirty will drop
1041 * to 0 as inflight RPCs fail out; otherwise, it's avail_grant + dirty.
1043 * race is tolerable here: if we're evicted, but imp_state already
1044 * left EVICTED state, then cl_dirty must be 0 already.
1046 client_obd_list_lock(&cli->cl_loi_list_lock);
1047 if (cli->cl_import->imp_state == LUSTRE_IMP_EVICTED)
1048 cli->cl_avail_grant = ocd->ocd_grant;
1050 cli->cl_avail_grant = ocd->ocd_grant - cli->cl_dirty;
1052 if (cli->cl_avail_grant < 0) {
1053 CWARN("%s: available grant < 0, the OSS is probably not running"
1054 " with patch from bug20278 (%ld) \n",
1055 cli->cl_import->imp_obd->obd_name, cli->cl_avail_grant);
1056 /* workaround for 1.6 servers which do not have
1057 * the patch from bug20278 */
1058 cli->cl_avail_grant = ocd->ocd_grant;
1061 /* determine the appropriate chunk size used by osc_extent. */
1062 cli->cl_chunkbits = max_t(int, CFS_PAGE_SHIFT, ocd->ocd_blocksize);
1063 client_obd_list_unlock(&cli->cl_loi_list_lock);
1065 CDEBUG(D_CACHE, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld."
1066 "chunk bits: %d.\n", cli->cl_import->imp_obd->obd_name,
1067 cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits);
1069 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
1070 cfs_list_empty(&cli->cl_grant_shrink_list))
1071 osc_add_shrink_grant(cli);
1074 /* We assume that the reason this OSC got a short read is because it read
1075 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1076 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1077 * this stripe never got written at or beyond this stripe offset yet. */
1078 static void handle_short_read(int nob_read, obd_count page_count,
1079 struct brw_page **pga)
1084 /* skip bytes read OK */
1085 while (nob_read > 0) {
1086 LASSERT (page_count > 0);
1088 if (pga[i]->count > nob_read) {
1089 /* EOF inside this page */
1090 ptr = cfs_kmap(pga[i]->pg) +
1091 (pga[i]->off & ~CFS_PAGE_MASK);
1092 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1093 cfs_kunmap(pga[i]->pg);
1099 nob_read -= pga[i]->count;
1104 /* zero remaining pages */
1105 while (page_count-- > 0) {
1106 ptr = cfs_kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
1107 memset(ptr, 0, pga[i]->count);
1108 cfs_kunmap(pga[i]->pg);
1113 static int check_write_rcs(struct ptlrpc_request *req,
1114 int requested_nob, int niocount,
1115 obd_count page_count, struct brw_page **pga)
1120 remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1121 sizeof(*remote_rcs) *
1123 if (remote_rcs == NULL) {
1124 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1128 /* return error if any niobuf was in error */
1129 for (i = 0; i < niocount; i++) {
1130 if ((int)remote_rcs[i] < 0)
1131 return(remote_rcs[i]);
1133 if (remote_rcs[i] != 0) {
1134 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1135 i, remote_rcs[i], req);
1140 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
1141 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1142 req->rq_bulk->bd_nob_transferred, requested_nob);
1149 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1151 if (p1->flag != p2->flag) {
1152 unsigned mask = ~(OBD_BRW_FROM_GRANT| OBD_BRW_NOCACHE|
1153 OBD_BRW_SYNC|OBD_BRW_ASYNC|OBD_BRW_NOQUOTA);
1155 /* warn if we try to combine flags that we don't know to be
1156 * safe to combine */
1157 if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
1158 CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1159 "report this at http://bugs.whamcloud.com/\n",
1160 p1->flag, p2->flag);
1165 return (p1->off + p1->count == p2->off);
1168 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
1169 struct brw_page **pga, int opc,
1170 cksum_type_t cksum_type)
1174 struct cfs_crypto_hash_desc *hdesc;
1175 unsigned int bufsize;
1177 unsigned char cfs_alg = cksum_obd2cfs(cksum_type);
1179 LASSERT(pg_count > 0);
1181 hdesc = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1182 if (IS_ERR(hdesc)) {
1183 CERROR("Unable to initialize checksum hash %s\n",
1184 cfs_crypto_hash_name(cfs_alg));
1185 return PTR_ERR(hdesc);
1188 while (nob > 0 && pg_count > 0) {
1189 int count = pga[i]->count > nob ? nob : pga[i]->count;
1191 /* corrupt the data before we compute the checksum, to
1192 * simulate an OST->client data error */
1193 if (i == 0 && opc == OST_READ &&
1194 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE)) {
1195 unsigned char *ptr = cfs_kmap(pga[i]->pg);
1196 int off = pga[i]->off & ~CFS_PAGE_MASK;
1197 memcpy(ptr + off, "bad1", min(4, nob));
1198 cfs_kunmap(pga[i]->pg);
1200 cfs_crypto_hash_update_page(hdesc, pga[i]->pg,
1201 pga[i]->off & ~CFS_PAGE_MASK,
1203 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d checksum %x\n",
1204 (int)(pga[i]->off & ~CFS_PAGE_MASK), cksum);
1206 nob -= pga[i]->count;
1212 err = cfs_crypto_hash_final(hdesc, (unsigned char *)&cksum, &bufsize);
1215 cfs_crypto_hash_final(hdesc, NULL, NULL);
1217 /* For sending we only compute the wrong checksum instead
1218 * of corrupting the data so it is still correct on a redo */
1219 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1225 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
1226 struct lov_stripe_md *lsm, obd_count page_count,
1227 struct brw_page **pga,
1228 struct ptlrpc_request **reqp,
1229 struct obd_capa *ocapa, int reserve,
1232 struct ptlrpc_request *req;
1233 struct ptlrpc_bulk_desc *desc;
1234 struct ost_body *body;
1235 struct obd_ioobj *ioobj;
1236 struct niobuf_remote *niobuf;
1237 int niocount, i, requested_nob, opc, rc;
1238 struct osc_brw_async_args *aa;
1239 struct req_capsule *pill;
1240 struct brw_page *pg_prev;
1243 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1244 RETURN(-ENOMEM); /* Recoverable */
1245 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1246 RETURN(-EINVAL); /* Fatal */
1248 if ((cmd & OBD_BRW_WRITE) != 0) {
1250 req = ptlrpc_request_alloc_pool(cli->cl_import,
1251 cli->cl_import->imp_rq_pool,
1252 &RQF_OST_BRW_WRITE);
1255 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1260 for (niocount = i = 1; i < page_count; i++) {
1261 if (!can_merge_pages(pga[i - 1], pga[i]))
1265 pill = &req->rq_pill;
1266 req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1268 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1269 niocount * sizeof(*niobuf));
1270 osc_set_capa_size(req, &RMF_CAPA1, ocapa);
1272 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1274 ptlrpc_request_free(req);
1277 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
1278 ptlrpc_at_set_req_timeout(req);
1279 /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1281 req->rq_no_retry_einprogress = 1;
1283 if (opc == OST_WRITE)
1284 desc = ptlrpc_prep_bulk_imp(req, page_count,
1285 BULK_GET_SOURCE, OST_BULK_PORTAL);
1287 desc = ptlrpc_prep_bulk_imp(req, page_count,
1288 BULK_PUT_SINK, OST_BULK_PORTAL);
1291 GOTO(out, rc = -ENOMEM);
1292 /* NB request now owns desc and will free it when it gets freed */
1294 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1295 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1296 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1297 LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1299 lustre_set_wire_obdo(&body->oa, oa);
1301 obdo_to_ioobj(oa, ioobj);
1302 ioobj->ioo_bufcnt = niocount;
1303 osc_pack_capa(req, body, ocapa);
1304 LASSERT (page_count > 0);
1306 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1307 struct brw_page *pg = pga[i];
1308 int poff = pg->off & ~CFS_PAGE_MASK;
1310 LASSERT(pg->count > 0);
1311 /* make sure there is no gap in the middle of page array */
1312 LASSERTF(page_count == 1 ||
1313 (ergo(i == 0, poff + pg->count == CFS_PAGE_SIZE) &&
1314 ergo(i > 0 && i < page_count - 1,
1315 poff == 0 && pg->count == CFS_PAGE_SIZE) &&
1316 ergo(i == page_count - 1, poff == 0)),
1317 "i: %d/%d pg: %p off: "LPU64", count: %u\n",
1318 i, page_count, pg, pg->off, pg->count);
1320 LASSERTF(i == 0 || pg->off > pg_prev->off,
1321 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1322 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1324 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1325 pg_prev->pg, page_private(pg_prev->pg),
1326 pg_prev->pg->index, pg_prev->off);
1328 LASSERTF(i == 0 || pg->off > pg_prev->off,
1329 "i %d p_c %u\n", i, page_count);
1331 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1332 (pg->flag & OBD_BRW_SRVLOCK));
1334 ptlrpc_prep_bulk_page_pin(desc, pg->pg, poff, pg->count);
1335 requested_nob += pg->count;
1337 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1339 niobuf->len += pg->count;
1341 niobuf->offset = pg->off;
1342 niobuf->len = pg->count;
1343 niobuf->flags = pg->flag;
1348 LASSERTF((void *)(niobuf - niocount) ==
1349 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1350 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1351 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1353 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1355 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1356 body->oa.o_valid |= OBD_MD_FLFLAGS;
1357 body->oa.o_flags = 0;
1359 body->oa.o_flags |= OBD_FL_RECOV_RESEND;
1362 if (osc_should_shrink_grant(cli))
1363 osc_shrink_grant_local(cli, &body->oa);
1365 /* size[REQ_REC_OFF] still sizeof (*body) */
1366 if (opc == OST_WRITE) {
1367 if (cli->cl_checksum &&
1368 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1369 /* store cl_cksum_type in a local variable since
1370 * it can be changed via lprocfs */
1371 cksum_type_t cksum_type = cli->cl_cksum_type;
1373 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1374 oa->o_flags &= OBD_FL_LOCAL_MASK;
1375 body->oa.o_flags = 0;
1377 body->oa.o_flags |= cksum_type_pack(cksum_type);
1378 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1379 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1383 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1385 /* save this in 'oa', too, for later checking */
1386 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1387 oa->o_flags |= cksum_type_pack(cksum_type);
1389 /* clear out the checksum flag, in case this is a
1390 * resend but cl_checksum is no longer set. b=11238 */
1391 oa->o_valid &= ~OBD_MD_FLCKSUM;
1393 oa->o_cksum = body->oa.o_cksum;
1394 /* 1 RC per niobuf */
1395 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1396 sizeof(__u32) * niocount);
1398 if (cli->cl_checksum &&
1399 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1400 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1401 body->oa.o_flags = 0;
1402 body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
1403 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1406 ptlrpc_request_set_replen(req);
1408 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1409 aa = ptlrpc_req_async_args(req);
1411 aa->aa_requested_nob = requested_nob;
1412 aa->aa_nio_count = niocount;
1413 aa->aa_page_count = page_count;
1417 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1418 if (ocapa && reserve)
1419 aa->aa_ocapa = capa_get(ocapa);
1425 ptlrpc_req_finished(req);
1429 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1430 __u32 client_cksum, __u32 server_cksum, int nob,
1431 obd_count page_count, struct brw_page **pga,
1432 cksum_type_t client_cksum_type)
1436 cksum_type_t cksum_type;
1438 if (server_cksum == client_cksum) {
1439 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1443 cksum_type = cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
1445 new_cksum = osc_checksum_bulk(nob, page_count, pga, OST_WRITE,
1448 if (cksum_type != client_cksum_type)
1449 msg = "the server did not use the checksum type specified in "
1450 "the original request - likely a protocol problem";
1451 else if (new_cksum == server_cksum)
1452 msg = "changed on the client after we checksummed it - "
1453 "likely false positive due to mmap IO (bug 11742)";
1454 else if (new_cksum == client_cksum)
1455 msg = "changed in transit before arrival at OST";
1457 msg = "changed in transit AND doesn't match the original - "
1458 "likely false positive due to mmap IO (bug 11742)";
1460 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inode "DFID
1461 " object "LPU64"/"LPU64" extent ["LPU64"-"LPU64"]\n",
1462 msg, libcfs_nid2str(peer->nid),
1463 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1464 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1465 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1467 oa->o_valid & OBD_MD_FLGROUP ? oa->o_seq : (__u64)0,
1469 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1470 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1471 "client csum now %x\n", client_cksum, client_cksum_type,
1472 server_cksum, cksum_type, new_cksum);
1476 /* Note rc enters this function as number of bytes transferred */
1477 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1479 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1480 const lnet_process_id_t *peer =
1481 &req->rq_import->imp_connection->c_peer;
1482 struct client_obd *cli = aa->aa_cli;
1483 struct ost_body *body;
1484 __u32 client_cksum = 0;
1487 if (rc < 0 && rc != -EDQUOT) {
1488 DEBUG_REQ(D_INFO, req, "Failed request with rc = %d\n", rc);
1492 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1493 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1495 DEBUG_REQ(D_INFO, req, "Can't unpack body\n");
1499 /* set/clear over quota flag for a uid/gid */
1500 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1501 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA)) {
1502 unsigned int qid[MAXQUOTAS] = { body->oa.o_uid, body->oa.o_gid };
1504 CDEBUG(D_QUOTA, "setdq for [%u %u] with valid "LPX64", flags %x\n",
1505 body->oa.o_uid, body->oa.o_gid, body->oa.o_valid,
1507 osc_quota_setdq(cli, qid, body->oa.o_valid, body->oa.o_flags);
1510 osc_update_grant(cli, body);
1515 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1516 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1518 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1520 CERROR("Unexpected +ve rc %d\n", rc);
1523 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1525 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1528 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1529 check_write_checksum(&body->oa, peer, client_cksum,
1530 body->oa.o_cksum, aa->aa_requested_nob,
1531 aa->aa_page_count, aa->aa_ppga,
1532 cksum_type_unpack(aa->aa_oa->o_flags)))
1535 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1536 aa->aa_page_count, aa->aa_ppga);
1540 /* The rest of this function executes only for OST_READs */
1542 /* if unwrap_bulk failed, return -EAGAIN to retry */
1543 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1545 GOTO(out, rc = -EAGAIN);
1547 if (rc > aa->aa_requested_nob) {
1548 CERROR("Unexpected rc %d (%d requested)\n", rc,
1549 aa->aa_requested_nob);
1553 if (rc != req->rq_bulk->bd_nob_transferred) {
1554 CERROR ("Unexpected rc %d (%d transferred)\n",
1555 rc, req->rq_bulk->bd_nob_transferred);
1559 if (rc < aa->aa_requested_nob)
1560 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1562 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
1563 static int cksum_counter;
1564 __u32 server_cksum = body->oa.o_cksum;
1567 cksum_type_t cksum_type;
1569 cksum_type = cksum_type_unpack(body->oa.o_valid &OBD_MD_FLFLAGS?
1570 body->oa.o_flags : 0);
1571 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1572 aa->aa_ppga, OST_READ,
1575 if (peer->nid == req->rq_bulk->bd_sender) {
1579 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1582 if (server_cksum == ~0 && rc > 0) {
1583 CERROR("Protocol error: server %s set the 'checksum' "
1584 "bit, but didn't send a checksum. Not fatal, "
1585 "but please notify on http://bugs.whamcloud.com/\n",
1586 libcfs_nid2str(peer->nid));
1587 } else if (server_cksum != client_cksum) {
1588 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1589 "%s%s%s inode "DFID" object "
1590 LPU64"/"LPU64" extent "
1591 "["LPU64"-"LPU64"]\n",
1592 req->rq_import->imp_obd->obd_name,
1593 libcfs_nid2str(peer->nid),
1595 body->oa.o_valid & OBD_MD_FLFID ?
1596 body->oa.o_parent_seq : (__u64)0,
1597 body->oa.o_valid & OBD_MD_FLFID ?
1598 body->oa.o_parent_oid : 0,
1599 body->oa.o_valid & OBD_MD_FLFID ?
1600 body->oa.o_parent_ver : 0,
1602 body->oa.o_valid & OBD_MD_FLGROUP ?
1603 body->oa.o_seq : (__u64)0,
1604 aa->aa_ppga[0]->off,
1605 aa->aa_ppga[aa->aa_page_count-1]->off +
1606 aa->aa_ppga[aa->aa_page_count-1]->count -
1608 CERROR("client %x, server %x, cksum_type %x\n",
1609 client_cksum, server_cksum, cksum_type);
1611 aa->aa_oa->o_cksum = client_cksum;
1615 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1618 } else if (unlikely(client_cksum)) {
1619 static int cksum_missed;
1622 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1623 CERROR("Checksum %u requested from %s but not sent\n",
1624 cksum_missed, libcfs_nid2str(peer->nid));
1630 lustre_get_wire_obdo(aa->aa_oa, &body->oa);
1635 static int osc_brw_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1636 struct lov_stripe_md *lsm,
1637 obd_count page_count, struct brw_page **pga,
1638 struct obd_capa *ocapa)
1640 struct ptlrpc_request *req;
1643 int generation, resends = 0;
1644 struct l_wait_info lwi;
1648 cfs_waitq_init(&waitq);
1649 generation = exp->exp_obd->u.cli.cl_import->imp_generation;
1652 rc = osc_brw_prep_request(cmd, &exp->exp_obd->u.cli, oa, lsm,
1653 page_count, pga, &req, ocapa, 0, resends);
1658 req->rq_generation_set = 1;
1659 req->rq_import_generation = generation;
1660 req->rq_sent = cfs_time_current_sec() + resends;
1663 rc = ptlrpc_queue_wait(req);
1665 if (rc == -ETIMEDOUT && req->rq_resend) {
1666 DEBUG_REQ(D_HA, req, "BULK TIMEOUT");
1667 ptlrpc_req_finished(req);
1671 rc = osc_brw_fini_request(req, rc);
1673 ptlrpc_req_finished(req);
1674 /* When server return -EINPROGRESS, client should always retry
1675 * regardless of the number of times the bulk was resent already.*/
1676 if (osc_recoverable_error(rc)) {
1678 if (rc != -EINPROGRESS &&
1679 !client_should_resend(resends, &exp->exp_obd->u.cli)) {
1680 CERROR("%s: too many resend retries for object: "
1681 ""LPU64":"LPU64", rc = %d.\n",
1682 exp->exp_obd->obd_name, oa->o_id, oa->o_seq, rc);
1686 exp->exp_obd->u.cli.cl_import->imp_generation) {
1687 CDEBUG(D_HA, "%s: resend cross eviction for object: "
1688 ""LPU64":"LPU64", rc = %d.\n",
1689 exp->exp_obd->obd_name, oa->o_id, oa->o_seq, rc);
1693 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(resends), NULL, NULL,
1695 l_wait_event(waitq, 0, &lwi);
1700 if (rc == -EAGAIN || rc == -EINPROGRESS)
1705 static int osc_brw_redo_request(struct ptlrpc_request *request,
1706 struct osc_brw_async_args *aa, int rc)
1708 struct ptlrpc_request *new_req;
1709 struct osc_brw_async_args *new_aa;
1710 struct osc_async_page *oap;
1713 DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
1714 "redo for recoverable error %d", rc);
1716 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
1717 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1718 aa->aa_cli, aa->aa_oa,
1719 NULL /* lsm unused by osc currently */,
1720 aa->aa_page_count, aa->aa_ppga,
1721 &new_req, aa->aa_ocapa, 0, 1);
1725 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1726 if (oap->oap_request != NULL) {
1727 LASSERTF(request == oap->oap_request,
1728 "request %p != oap_request %p\n",
1729 request, oap->oap_request);
1730 if (oap->oap_interrupted) {
1731 ptlrpc_req_finished(new_req);
1736 /* New request takes over pga and oaps from old request.
1737 * Note that copying a list_head doesn't work, need to move it... */
1739 new_req->rq_interpret_reply = request->rq_interpret_reply;
1740 new_req->rq_async_args = request->rq_async_args;
1741 /* cap resend delay to the current request timeout, this is similar to
1742 * what ptlrpc does (see after_reply()) */
1743 if (aa->aa_resends > new_req->rq_timeout)
1744 new_req->rq_sent = cfs_time_current_sec() + new_req->rq_timeout;
1746 new_req->rq_sent = cfs_time_current_sec() + aa->aa_resends;
1747 new_req->rq_generation_set = 1;
1748 new_req->rq_import_generation = request->rq_import_generation;
1750 new_aa = ptlrpc_req_async_args(new_req);
1752 CFS_INIT_LIST_HEAD(&new_aa->aa_oaps);
1753 cfs_list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
1754 CFS_INIT_LIST_HEAD(&new_aa->aa_exts);
1755 cfs_list_splice_init(&aa->aa_exts, &new_aa->aa_exts);
1757 cfs_list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1758 if (oap->oap_request) {
1759 ptlrpc_req_finished(oap->oap_request);
1760 oap->oap_request = ptlrpc_request_addref(new_req);
1764 new_aa->aa_ocapa = aa->aa_ocapa;
1765 aa->aa_ocapa = NULL;
1767 /* XXX: This code will run into problem if we're going to support
1768 * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
1769 * and wait for all of them to be finished. We should inherit request
1770 * set from old request. */
1771 ptlrpcd_add_req(new_req, PDL_POLICY_SAME, -1);
1773 DEBUG_REQ(D_INFO, new_req, "new request");
1778 * ugh, we want disk allocation on the target to happen in offset order. we'll
1779 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1780 * fine for our small page arrays and doesn't require allocation. its an
1781 * insertion sort that swaps elements that are strides apart, shrinking the
1782 * stride down until its '1' and the array is sorted.
1784 static void sort_brw_pages(struct brw_page **array, int num)
1787 struct brw_page *tmp;
1791 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1796 for (i = stride ; i < num ; i++) {
1799 while (j >= stride && array[j - stride]->off > tmp->off) {
1800 array[j] = array[j - stride];
1805 } while (stride > 1);
1808 static obd_count max_unfragmented_pages(struct brw_page **pg, obd_count pages)
1814 LASSERT (pages > 0);
1815 offset = pg[i]->off & ~CFS_PAGE_MASK;
1819 if (pages == 0) /* that's all */
1822 if (offset + pg[i]->count < CFS_PAGE_SIZE)
1823 return count; /* doesn't end on page boundary */
1826 offset = pg[i]->off & ~CFS_PAGE_MASK;
1827 if (offset != 0) /* doesn't start on page boundary */
1834 static struct brw_page **osc_build_ppga(struct brw_page *pga, obd_count count)
1836 struct brw_page **ppga;
1839 OBD_ALLOC(ppga, sizeof(*ppga) * count);
1843 for (i = 0; i < count; i++)
1848 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1850 LASSERT(ppga != NULL);
1851 OBD_FREE(ppga, sizeof(*ppga) * count);
1854 static int osc_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
1855 obd_count page_count, struct brw_page *pga,
1856 struct obd_trans_info *oti)
1858 struct obdo *saved_oa = NULL;
1859 struct brw_page **ppga, **orig;
1860 struct obd_import *imp = class_exp2cliimp(exp);
1861 struct client_obd *cli;
1862 int rc, page_count_orig;
1865 LASSERT((imp != NULL) && (imp->imp_obd != NULL));
1866 cli = &imp->imp_obd->u.cli;
1868 if (cmd & OBD_BRW_CHECK) {
1869 /* The caller just wants to know if there's a chance that this
1870 * I/O can succeed */
1872 if (imp->imp_invalid)
1877 /* test_brw with a failed create can trip this, maybe others. */
1878 LASSERT(cli->cl_max_pages_per_rpc);
1882 orig = ppga = osc_build_ppga(pga, page_count);
1885 page_count_orig = page_count;
1887 sort_brw_pages(ppga, page_count);
1888 while (page_count) {
1889 obd_count pages_per_brw;
1891 if (page_count > cli->cl_max_pages_per_rpc)
1892 pages_per_brw = cli->cl_max_pages_per_rpc;
1894 pages_per_brw = page_count;
1896 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1898 if (saved_oa != NULL) {
1899 /* restore previously saved oa */
1900 *oinfo->oi_oa = *saved_oa;
1901 } else if (page_count > pages_per_brw) {
1902 /* save a copy of oa (brw will clobber it) */
1903 OBDO_ALLOC(saved_oa);
1904 if (saved_oa == NULL)
1905 GOTO(out, rc = -ENOMEM);
1906 *saved_oa = *oinfo->oi_oa;
1909 rc = osc_brw_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1910 pages_per_brw, ppga, oinfo->oi_capa);
1915 page_count -= pages_per_brw;
1916 ppga += pages_per_brw;
1920 osc_release_ppga(orig, page_count_orig);
1922 if (saved_oa != NULL)
1923 OBDO_FREE(saved_oa);
1928 static int brw_interpret(const struct lu_env *env,
1929 struct ptlrpc_request *req, void *data, int rc)
1931 struct osc_brw_async_args *aa = data;
1932 struct osc_extent *ext;
1933 struct osc_extent *tmp;
1934 struct cl_object *obj = NULL;
1935 struct client_obd *cli = aa->aa_cli;
1938 rc = osc_brw_fini_request(req, rc);
1939 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
1940 /* When server return -EINPROGRESS, client should always retry
1941 * regardless of the number of times the bulk was resent already. */
1942 if (osc_recoverable_error(rc)) {
1943 if (req->rq_import_generation !=
1944 req->rq_import->imp_generation) {
1945 CDEBUG(D_HA, "%s: resend cross eviction for object: "
1946 ""LPU64":"LPU64", rc = %d.\n",
1947 req->rq_import->imp_obd->obd_name,
1948 aa->aa_oa->o_id, aa->aa_oa->o_seq, rc);
1949 } else if (rc == -EINPROGRESS ||
1950 client_should_resend(aa->aa_resends, aa->aa_cli)) {
1951 rc = osc_brw_redo_request(req, aa, rc);
1953 CERROR("%s: too many resent retries for object: "
1954 ""LPU64":"LPU64", rc = %d.\n",
1955 req->rq_import->imp_obd->obd_name,
1956 aa->aa_oa->o_id, aa->aa_oa->o_seq, rc);
1961 else if (rc == -EAGAIN || rc == -EINPROGRESS)
1966 capa_put(aa->aa_ocapa);
1967 aa->aa_ocapa = NULL;
1970 cfs_list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
1971 if (obj == NULL && rc == 0) {
1972 obj = osc2cl(ext->oe_obj);
1976 cfs_list_del_init(&ext->oe_link);
1977 osc_extent_finish(env, ext, 1, rc);
1979 LASSERT(cfs_list_empty(&aa->aa_exts));
1980 LASSERT(cfs_list_empty(&aa->aa_oaps));
1983 struct obdo *oa = aa->aa_oa;
1984 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
1985 unsigned long valid = 0;
1988 if (oa->o_valid & OBD_MD_FLBLOCKS) {
1989 attr->cat_blocks = oa->o_blocks;
1990 valid |= CAT_BLOCKS;
1992 if (oa->o_valid & OBD_MD_FLMTIME) {
1993 attr->cat_mtime = oa->o_mtime;
1996 if (oa->o_valid & OBD_MD_FLATIME) {
1997 attr->cat_atime = oa->o_atime;
2000 if (oa->o_valid & OBD_MD_FLCTIME) {
2001 attr->cat_ctime = oa->o_ctime;
2005 cl_object_attr_lock(obj);
2006 cl_object_attr_set(env, obj, attr, valid);
2007 cl_object_attr_unlock(obj);
2009 cl_object_put(env, obj);
2011 OBDO_FREE(aa->aa_oa);
2013 cl_req_completion(env, aa->aa_clerq, rc < 0 ? rc :
2014 req->rq_bulk->bd_nob_transferred);
2015 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2016 ptlrpc_lprocfs_brw(req, req->rq_bulk->bd_nob_transferred);
2018 client_obd_list_lock(&cli->cl_loi_list_lock);
2019 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2020 * is called so we know whether to go to sync BRWs or wait for more
2021 * RPCs to complete */
2022 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2023 cli->cl_w_in_flight--;
2025 cli->cl_r_in_flight--;
2026 osc_wake_cache_waiters(cli);
2027 client_obd_list_unlock(&cli->cl_loi_list_lock);
2029 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
2034 * Build an RPC by the list of extent @ext_list. The caller must ensure
2035 * that the total pages in this list are NOT over max pages per RPC.
2036 * Extents in the list must be in OES_RPC state.
2038 int osc_build_rpc(const struct lu_env *env, struct client_obd *cli,
2039 cfs_list_t *ext_list, int cmd, pdl_policy_t pol)
2041 struct ptlrpc_request *req = NULL;
2042 struct osc_extent *ext;
2043 CFS_LIST_HEAD(rpc_list);
2044 struct brw_page **pga = NULL;
2045 struct osc_brw_async_args *aa = NULL;
2046 struct obdo *oa = NULL;
2047 struct osc_async_page *oap;
2048 struct osc_async_page *tmp;
2049 struct cl_req *clerq = NULL;
2050 enum cl_req_type crt = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2051 struct ldlm_lock *lock = NULL;
2052 struct cl_req_attr crattr;
2053 obd_off starting_offset = OBD_OBJECT_EOF;
2054 obd_off ending_offset = 0;
2055 int i, rc, mpflag = 0, mem_tight = 0, page_count = 0;
2058 LASSERT(!cfs_list_empty(ext_list));
2060 /* add pages into rpc_list to build BRW rpc */
2061 cfs_list_for_each_entry(ext, ext_list, oe_link) {
2062 LASSERT(ext->oe_state == OES_RPC);
2063 mem_tight |= ext->oe_memalloc;
2064 cfs_list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
2066 cfs_list_add_tail(&oap->oap_rpc_item, &rpc_list);
2067 if (starting_offset > oap->oap_obj_off)
2068 starting_offset = oap->oap_obj_off;
2070 LASSERT(oap->oap_page_off == 0);
2071 if (ending_offset < oap->oap_obj_off + oap->oap_count)
2072 ending_offset = oap->oap_obj_off +
2075 LASSERT(oap->oap_page_off + oap->oap_count ==
2081 mpflag = cfs_memory_pressure_get_and_set();
2083 memset(&crattr, 0, sizeof crattr);
2084 OBD_ALLOC(pga, sizeof(*pga) * page_count);
2086 GOTO(out, rc = -ENOMEM);
2090 GOTO(out, rc = -ENOMEM);
2093 cfs_list_for_each_entry(oap, &rpc_list, oap_rpc_item) {
2094 struct cl_page *page = oap2cl_page(oap);
2095 if (clerq == NULL) {
2096 clerq = cl_req_alloc(env, page, crt,
2097 1 /* only 1-object rpcs for
2100 GOTO(out, rc = PTR_ERR(clerq));
2101 lock = oap->oap_ldlm_lock;
2104 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
2105 pga[i] = &oap->oap_brw_page;
2106 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2107 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
2108 pga[i]->pg, cfs_page_index(oap->oap_page), oap, pga[i]->flag);
2110 cl_req_page_add(env, clerq, page);
2113 /* always get the data for the obdo for the rpc */
2114 LASSERT(clerq != NULL);
2116 crattr.cra_capa = NULL;
2117 memset(crattr.cra_jobid, 0, JOBSTATS_JOBID_SIZE);
2118 cl_req_attr_set(env, clerq, &crattr, ~0ULL);
2120 oa->o_handle = lock->l_remote_handle;
2121 oa->o_valid |= OBD_MD_FLHANDLE;
2124 rc = cl_req_prep(env, clerq);
2126 CERROR("cl_req_prep failed: %d\n", rc);
2130 sort_brw_pages(pga, page_count);
2131 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
2132 pga, &req, crattr.cra_capa, 1, 0);
2134 CERROR("prep_req failed: %d\n", rc);
2138 req->rq_interpret_reply = brw_interpret;
2140 req->rq_memalloc = 1;
2142 /* Need to update the timestamps after the request is built in case
2143 * we race with setattr (locally or in queue at OST). If OST gets
2144 * later setattr before earlier BRW (as determined by the request xid),
2145 * the OST will not use BRW timestamps. Sadly, there is no obvious
2146 * way to do this in a single call. bug 10150 */
2147 cl_req_attr_set(env, clerq, &crattr,
2148 OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME);
2150 lustre_msg_set_jobid(req->rq_reqmsg, crattr.cra_jobid);
2152 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2153 aa = ptlrpc_req_async_args(req);
2154 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
2155 cfs_list_splice_init(&rpc_list, &aa->aa_oaps);
2156 CFS_INIT_LIST_HEAD(&aa->aa_exts);
2157 cfs_list_splice_init(ext_list, &aa->aa_exts);
2158 aa->aa_clerq = clerq;
2160 /* queued sync pages can be torn down while the pages
2161 * were between the pending list and the rpc */
2163 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2164 /* only one oap gets a request reference */
2167 if (oap->oap_interrupted && !req->rq_intr) {
2168 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2170 ptlrpc_mark_interrupted(req);
2174 tmp->oap_request = ptlrpc_request_addref(req);
2176 client_obd_list_lock(&cli->cl_loi_list_lock);
2177 starting_offset >>= CFS_PAGE_SHIFT;
2178 if (cmd == OBD_BRW_READ) {
2179 cli->cl_r_in_flight++;
2180 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2181 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2182 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2183 starting_offset + 1);
2185 cli->cl_w_in_flight++;
2186 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2187 lprocfs_oh_tally(&cli->cl_write_rpc_hist, cli->cl_w_in_flight);
2188 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2189 starting_offset + 1);
2191 client_obd_list_unlock(&cli->cl_loi_list_lock);
2193 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2194 page_count, aa, cli->cl_r_in_flight,
2195 cli->cl_w_in_flight);
2197 /* XXX: Maybe the caller can check the RPC bulk descriptor to
2198 * see which CPU/NUMA node the majority of pages were allocated
2199 * on, and try to assign the async RPC to the CPU core
2200 * (PDL_POLICY_PREFERRED) to reduce cross-CPU memory traffic.
2202 * But on the other hand, we expect that multiple ptlrpcd
2203 * threads and the initial write sponsor can run in parallel,
2204 * especially when data checksum is enabled, which is CPU-bound
2205 * operation and single ptlrpcd thread cannot process in time.
2206 * So more ptlrpcd threads sharing BRW load
2207 * (with PDL_POLICY_ROUND) seems better.
2209 ptlrpcd_add_req(req, pol, -1);
2215 cfs_memory_pressure_restore(mpflag);
2217 capa_put(crattr.cra_capa);
2219 LASSERT(req == NULL);
2224 OBD_FREE(pga, sizeof(*pga) * page_count);
2225 /* this should happen rarely and is pretty bad, it makes the
2226 * pending list not follow the dirty order */
2227 while (!cfs_list_empty(ext_list)) {
2228 ext = cfs_list_entry(ext_list->next, struct osc_extent,
2230 cfs_list_del_init(&ext->oe_link);
2231 osc_extent_finish(env, ext, 0, rc);
2233 if (clerq && !IS_ERR(clerq))
2234 cl_req_completion(env, clerq, rc);
2239 static int osc_set_lock_data_with_check(struct ldlm_lock *lock,
2240 struct ldlm_enqueue_info *einfo)
2242 void *data = einfo->ei_cbdata;
2245 LASSERT(lock != NULL);
2246 LASSERT(lock->l_blocking_ast == einfo->ei_cb_bl);
2247 LASSERT(lock->l_resource->lr_type == einfo->ei_type);
2248 LASSERT(lock->l_completion_ast == einfo->ei_cb_cp);
2249 LASSERT(lock->l_glimpse_ast == einfo->ei_cb_gl);
2251 lock_res_and_lock(lock);
2252 cfs_spin_lock(&osc_ast_guard);
2254 if (lock->l_ast_data == NULL)
2255 lock->l_ast_data = data;
2256 if (lock->l_ast_data == data)
2259 cfs_spin_unlock(&osc_ast_guard);
2260 unlock_res_and_lock(lock);
2265 static int osc_set_data_with_check(struct lustre_handle *lockh,
2266 struct ldlm_enqueue_info *einfo)
2268 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2272 set = osc_set_lock_data_with_check(lock, einfo);
2273 LDLM_LOCK_PUT(lock);
2275 CERROR("lockh %p, data %p - client evicted?\n",
2276 lockh, einfo->ei_cbdata);
2280 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2281 ldlm_iterator_t replace, void *data)
2283 struct ldlm_res_id res_id;
2284 struct obd_device *obd = class_exp2obd(exp);
2286 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
2287 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2291 /* find any ldlm lock of the inode in osc
2295 static int osc_find_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2296 ldlm_iterator_t replace, void *data)
2298 struct ldlm_res_id res_id;
2299 struct obd_device *obd = class_exp2obd(exp);
2302 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
2303 rc = ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2304 if (rc == LDLM_ITER_STOP)
2306 if (rc == LDLM_ITER_CONTINUE)
2311 static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
2312 obd_enqueue_update_f upcall, void *cookie,
2313 __u64 *flags, int agl, int rc)
2315 int intent = *flags & LDLM_FL_HAS_INTENT;
2319 /* The request was created before ldlm_cli_enqueue call. */
2320 if (rc == ELDLM_LOCK_ABORTED) {
2321 struct ldlm_reply *rep;
2322 rep = req_capsule_server_get(&req->rq_pill,
2325 LASSERT(rep != NULL);
2326 if (rep->lock_policy_res1)
2327 rc = rep->lock_policy_res1;
2331 if ((intent != 0 && rc == ELDLM_LOCK_ABORTED && agl == 0) ||
2333 *flags |= LDLM_FL_LVB_READY;
2334 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
2335 lvb->lvb_size, lvb->lvb_blocks, lvb->lvb_mtime);
2338 /* Call the update callback. */
2339 rc = (*upcall)(cookie, rc);
2343 static int osc_enqueue_interpret(const struct lu_env *env,
2344 struct ptlrpc_request *req,
2345 struct osc_enqueue_args *aa, int rc)
2347 struct ldlm_lock *lock;
2348 struct lustre_handle handle;
2350 struct ost_lvb *lvb;
2352 __u64 *flags = aa->oa_flags;
2354 /* Make a local copy of a lock handle and a mode, because aa->oa_*
2355 * might be freed anytime after lock upcall has been called. */
2356 lustre_handle_copy(&handle, aa->oa_lockh);
2357 mode = aa->oa_ei->ei_mode;
2359 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2361 lock = ldlm_handle2lock(&handle);
2363 /* Take an additional reference so that a blocking AST that
2364 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2365 * to arrive after an upcall has been executed by
2366 * osc_enqueue_fini(). */
2367 ldlm_lock_addref(&handle, mode);
2369 /* Let CP AST to grant the lock first. */
2370 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
2372 if (aa->oa_agl && rc == ELDLM_LOCK_ABORTED) {
2377 lvb_len = sizeof(*aa->oa_lvb);
2380 /* Complete obtaining the lock procedure. */
2381 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
2382 mode, flags, lvb, lvb_len, &handle, rc);
2383 /* Complete osc stuff. */
2384 rc = osc_enqueue_fini(req, aa->oa_lvb, aa->oa_upcall, aa->oa_cookie,
2385 flags, aa->oa_agl, rc);
2387 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
2389 /* Release the lock for async request. */
2390 if (lustre_handle_is_used(&handle) && rc == ELDLM_OK)
2392 * Releases a reference taken by ldlm_cli_enqueue(), if it is
2393 * not already released by
2394 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
2396 ldlm_lock_decref(&handle, mode);
2398 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
2399 aa->oa_lockh, req, aa);
2400 ldlm_lock_decref(&handle, mode);
2401 LDLM_LOCK_PUT(lock);
2405 void osc_update_enqueue(struct lustre_handle *lov_lockhp,
2406 struct lov_oinfo *loi, int flags,
2407 struct ost_lvb *lvb, __u32 mode, int rc)
2409 struct ldlm_lock *lock = ldlm_handle2lock(lov_lockhp);
2411 if (rc == ELDLM_OK) {
2414 LASSERT(lock != NULL);
2415 loi->loi_lvb = *lvb;
2416 tmp = loi->loi_lvb.lvb_size;
2417 /* Extend KMS up to the end of this lock and no further
2418 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
2419 if (tmp > lock->l_policy_data.l_extent.end)
2420 tmp = lock->l_policy_data.l_extent.end + 1;
2421 if (tmp >= loi->loi_kms) {
2422 LDLM_DEBUG(lock, "lock acquired, setting rss="LPU64
2423 ", kms="LPU64, loi->loi_lvb.lvb_size, tmp);
2424 loi_kms_set(loi, tmp);
2426 LDLM_DEBUG(lock, "lock acquired, setting rss="
2427 LPU64"; leaving kms="LPU64", end="LPU64,
2428 loi->loi_lvb.lvb_size, loi->loi_kms,
2429 lock->l_policy_data.l_extent.end);
2431 ldlm_lock_allow_match(lock);
2432 } else if (rc == ELDLM_LOCK_ABORTED && (flags & LDLM_FL_HAS_INTENT)) {
2433 LASSERT(lock != NULL);
2434 loi->loi_lvb = *lvb;
2435 ldlm_lock_allow_match(lock);
2436 CDEBUG(D_INODE, "glimpsed, setting rss="LPU64"; leaving"
2437 " kms="LPU64"\n", loi->loi_lvb.lvb_size, loi->loi_kms);
2443 ldlm_lock_fail_match(lock);
2445 LDLM_LOCK_PUT(lock);
2448 EXPORT_SYMBOL(osc_update_enqueue);
2450 struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
2452 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2453 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2454 * other synchronous requests, however keeping some locks and trying to obtain
2455 * others may take a considerable amount of time in a case of ost failure; and
2456 * when other sync requests do not get released lock from a client, the client
2457 * is excluded from the cluster -- such scenarious make the life difficult, so
2458 * release locks just after they are obtained. */
2459 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2460 __u64 *flags, ldlm_policy_data_t *policy,
2461 struct ost_lvb *lvb, int kms_valid,
2462 obd_enqueue_update_f upcall, void *cookie,
2463 struct ldlm_enqueue_info *einfo,
2464 struct lustre_handle *lockh,
2465 struct ptlrpc_request_set *rqset, int async, int agl)
2467 struct obd_device *obd = exp->exp_obd;
2468 struct ptlrpc_request *req = NULL;
2469 int intent = *flags & LDLM_FL_HAS_INTENT;
2470 int match_lvb = (agl != 0 ? 0 : LDLM_FL_LVB_READY);
2475 /* Filesystem lock extents are extended to page boundaries so that
2476 * dealing with the page cache is a little smoother. */
2477 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2478 policy->l_extent.end |= ~CFS_PAGE_MASK;
2481 * kms is not valid when either object is completely fresh (so that no
2482 * locks are cached), or object was evicted. In the latter case cached
2483 * lock cannot be used, because it would prime inode state with
2484 * potentially stale LVB.
2489 /* Next, search for already existing extent locks that will cover us */
2490 /* If we're trying to read, we also search for an existing PW lock. The
2491 * VFS and page cache already protect us locally, so lots of readers/
2492 * writers can share a single PW lock.
2494 * There are problems with conversion deadlocks, so instead of
2495 * converting a read lock to a write lock, we'll just enqueue a new
2498 * At some point we should cancel the read lock instead of making them
2499 * send us a blocking callback, but there are problems with canceling
2500 * locks out from other users right now, too. */
2501 mode = einfo->ei_mode;
2502 if (einfo->ei_mode == LCK_PR)
2504 mode = ldlm_lock_match(obd->obd_namespace, *flags | match_lvb, res_id,
2505 einfo->ei_type, policy, mode, lockh, 0);
2507 struct ldlm_lock *matched = ldlm_handle2lock(lockh);
2509 if ((agl != 0) && !(matched->l_flags & LDLM_FL_LVB_READY)) {
2510 /* For AGL, if enqueue RPC is sent but the lock is not
2511 * granted, then skip to process this strpe.
2512 * Return -ECANCELED to tell the caller. */
2513 ldlm_lock_decref(lockh, mode);
2514 LDLM_LOCK_PUT(matched);
2516 } else if (osc_set_lock_data_with_check(matched, einfo)) {
2517 *flags |= LDLM_FL_LVB_READY;
2518 /* addref the lock only if not async requests and PW
2519 * lock is matched whereas we asked for PR. */
2520 if (!rqset && einfo->ei_mode != mode)
2521 ldlm_lock_addref(lockh, LCK_PR);
2523 /* I would like to be able to ASSERT here that
2524 * rss <= kms, but I can't, for reasons which
2525 * are explained in lov_enqueue() */
2528 /* We already have a lock, and it's referenced.
2530 * At this point, the cl_lock::cll_state is CLS_QUEUING,
2531 * AGL upcall may change it to CLS_HELD directly. */
2532 (*upcall)(cookie, ELDLM_OK);
2534 if (einfo->ei_mode != mode)
2535 ldlm_lock_decref(lockh, LCK_PW);
2537 /* For async requests, decref the lock. */
2538 ldlm_lock_decref(lockh, einfo->ei_mode);
2539 LDLM_LOCK_PUT(matched);
2542 ldlm_lock_decref(lockh, mode);
2543 LDLM_LOCK_PUT(matched);
2549 CFS_LIST_HEAD(cancels);
2550 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2551 &RQF_LDLM_ENQUEUE_LVB);
2555 rc = ldlm_prep_enqueue_req(exp, req, &cancels, 0);
2557 ptlrpc_request_free(req);
2561 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
2563 ptlrpc_request_set_replen(req);
2566 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2567 *flags &= ~LDLM_FL_BLOCK_GRANTED;
2569 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
2570 sizeof(*lvb), lockh, async);
2573 struct osc_enqueue_args *aa;
2574 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2575 aa = ptlrpc_req_async_args(req);
2578 aa->oa_flags = flags;
2579 aa->oa_upcall = upcall;
2580 aa->oa_cookie = cookie;
2582 aa->oa_lockh = lockh;
2585 req->rq_interpret_reply =
2586 (ptlrpc_interpterer_t)osc_enqueue_interpret;
2587 if (rqset == PTLRPCD_SET)
2588 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
2590 ptlrpc_set_add_req(rqset, req);
2591 } else if (intent) {
2592 ptlrpc_req_finished(req);
2597 rc = osc_enqueue_fini(req, lvb, upcall, cookie, flags, agl, rc);
2599 ptlrpc_req_finished(req);
2604 static int osc_enqueue(struct obd_export *exp, struct obd_info *oinfo,
2605 struct ldlm_enqueue_info *einfo,
2606 struct ptlrpc_request_set *rqset)
2608 struct ldlm_res_id res_id;
2612 osc_build_res_name(oinfo->oi_md->lsm_object_id,
2613 oinfo->oi_md->lsm_object_seq, &res_id);
2615 rc = osc_enqueue_base(exp, &res_id, &oinfo->oi_flags, &oinfo->oi_policy,
2616 &oinfo->oi_md->lsm_oinfo[0]->loi_lvb,
2617 oinfo->oi_md->lsm_oinfo[0]->loi_kms_valid,
2618 oinfo->oi_cb_up, oinfo, einfo, oinfo->oi_lockh,
2619 rqset, rqset != NULL, 0);
2623 int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2624 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
2625 int *flags, void *data, struct lustre_handle *lockh,
2628 struct obd_device *obd = exp->exp_obd;
2629 int lflags = *flags;
2633 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
2636 /* Filesystem lock extents are extended to page boundaries so that
2637 * dealing with the page cache is a little smoother */
2638 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2639 policy->l_extent.end |= ~CFS_PAGE_MASK;
2641 /* Next, search for already existing extent locks that will cover us */
2642 /* If we're trying to read, we also search for an existing PW lock. The
2643 * VFS and page cache already protect us locally, so lots of readers/
2644 * writers can share a single PW lock. */
2648 rc = ldlm_lock_match(obd->obd_namespace, lflags,
2649 res_id, type, policy, rc, lockh, unref);
2652 if (!osc_set_data_with_check(lockh, data)) {
2653 if (!(lflags & LDLM_FL_TEST_LOCK))
2654 ldlm_lock_decref(lockh, rc);
2658 if (!(lflags & LDLM_FL_TEST_LOCK) && mode != rc) {
2659 ldlm_lock_addref(lockh, LCK_PR);
2660 ldlm_lock_decref(lockh, LCK_PW);
2667 int osc_cancel_base(struct lustre_handle *lockh, __u32 mode)
2671 if (unlikely(mode == LCK_GROUP))
2672 ldlm_lock_decref_and_cancel(lockh, mode);
2674 ldlm_lock_decref(lockh, mode);
2679 static int osc_cancel(struct obd_export *exp, struct lov_stripe_md *md,
2680 __u32 mode, struct lustre_handle *lockh)
2683 RETURN(osc_cancel_base(lockh, mode));
2686 static int osc_cancel_unused(struct obd_export *exp,
2687 struct lov_stripe_md *lsm,
2688 ldlm_cancel_flags_t flags,
2691 struct obd_device *obd = class_exp2obd(exp);
2692 struct ldlm_res_id res_id, *resp = NULL;
2695 resp = osc_build_res_name(lsm->lsm_object_id,
2696 lsm->lsm_object_seq, &res_id);
2699 return ldlm_cli_cancel_unused(obd->obd_namespace, resp, flags, opaque);
2702 static int osc_statfs_interpret(const struct lu_env *env,
2703 struct ptlrpc_request *req,
2704 struct osc_async_args *aa, int rc)
2706 struct obd_statfs *msfs;
2710 /* The request has in fact never been sent
2711 * due to issues at a higher level (LOV).
2712 * Exit immediately since the caller is
2713 * aware of the problem and takes care
2714 * of the clean up */
2717 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
2718 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
2724 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2726 GOTO(out, rc = -EPROTO);
2729 *aa->aa_oi->oi_osfs = *msfs;
2731 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
2735 static int osc_statfs_async(struct obd_export *exp,
2736 struct obd_info *oinfo, __u64 max_age,
2737 struct ptlrpc_request_set *rqset)
2739 struct obd_device *obd = class_exp2obd(exp);
2740 struct ptlrpc_request *req;
2741 struct osc_async_args *aa;
2745 /* We could possibly pass max_age in the request (as an absolute
2746 * timestamp or a "seconds.usec ago") so the target can avoid doing
2747 * extra calls into the filesystem if that isn't necessary (e.g.
2748 * during mount that would help a bit). Having relative timestamps
2749 * is not so great if request processing is slow, while absolute
2750 * timestamps are not ideal because they need time synchronization. */
2751 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
2755 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2757 ptlrpc_request_free(req);
2760 ptlrpc_request_set_replen(req);
2761 req->rq_request_portal = OST_CREATE_PORTAL;
2762 ptlrpc_at_set_req_timeout(req);
2764 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
2765 /* procfs requests not want stat in wait for avoid deadlock */
2766 req->rq_no_resend = 1;
2767 req->rq_no_delay = 1;
2770 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
2771 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2772 aa = ptlrpc_req_async_args(req);
2775 ptlrpc_set_add_req(rqset, req);
2779 static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
2780 struct obd_statfs *osfs, __u64 max_age, __u32 flags)
2782 struct obd_device *obd = class_exp2obd(exp);
2783 struct obd_statfs *msfs;
2784 struct ptlrpc_request *req;
2785 struct obd_import *imp = NULL;
2789 /*Since the request might also come from lprocfs, so we need
2790 *sync this with client_disconnect_export Bug15684*/
2791 cfs_down_read(&obd->u.cli.cl_sem);
2792 if (obd->u.cli.cl_import)
2793 imp = class_import_get(obd->u.cli.cl_import);
2794 cfs_up_read(&obd->u.cli.cl_sem);
2798 /* We could possibly pass max_age in the request (as an absolute
2799 * timestamp or a "seconds.usec ago") so the target can avoid doing
2800 * extra calls into the filesystem if that isn't necessary (e.g.
2801 * during mount that would help a bit). Having relative timestamps
2802 * is not so great if request processing is slow, while absolute
2803 * timestamps are not ideal because they need time synchronization. */
2804 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
2806 class_import_put(imp);
2811 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2813 ptlrpc_request_free(req);
2816 ptlrpc_request_set_replen(req);
2817 req->rq_request_portal = OST_CREATE_PORTAL;
2818 ptlrpc_at_set_req_timeout(req);
2820 if (flags & OBD_STATFS_NODELAY) {
2821 /* procfs requests not want stat in wait for avoid deadlock */
2822 req->rq_no_resend = 1;
2823 req->rq_no_delay = 1;
2826 rc = ptlrpc_queue_wait(req);
2830 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2832 GOTO(out, rc = -EPROTO);
2839 ptlrpc_req_finished(req);
2843 /* Retrieve object striping information.
2845 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
2846 * the maximum number of OST indices which will fit in the user buffer.
2847 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
2849 static int osc_getstripe(struct lov_stripe_md *lsm, struct lov_user_md *lump)
2851 /* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
2852 struct lov_user_md_v3 lum, *lumk;
2853 struct lov_user_ost_data_v1 *lmm_objects;
2854 int rc = 0, lum_size;
2860 /* we only need the header part from user space to get lmm_magic and
2861 * lmm_stripe_count, (the header part is common to v1 and v3) */
2862 lum_size = sizeof(struct lov_user_md_v1);
2863 if (cfs_copy_from_user(&lum, lump, lum_size))
2866 if ((lum.lmm_magic != LOV_USER_MAGIC_V1) &&
2867 (lum.lmm_magic != LOV_USER_MAGIC_V3))
2870 /* lov_user_md_vX and lov_mds_md_vX must have the same size */
2871 LASSERT(sizeof(struct lov_user_md_v1) == sizeof(struct lov_mds_md_v1));
2872 LASSERT(sizeof(struct lov_user_md_v3) == sizeof(struct lov_mds_md_v3));
2873 LASSERT(sizeof(lum.lmm_objects[0]) == sizeof(lumk->lmm_objects[0]));
2875 /* we can use lov_mds_md_size() to compute lum_size
2876 * because lov_user_md_vX and lov_mds_md_vX have the same size */
2877 if (lum.lmm_stripe_count > 0) {
2878 lum_size = lov_mds_md_size(lum.lmm_stripe_count, lum.lmm_magic);
2879 OBD_ALLOC(lumk, lum_size);
2883 if (lum.lmm_magic == LOV_USER_MAGIC_V1)
2884 lmm_objects = &(((struct lov_user_md_v1 *)lumk)->lmm_objects[0]);
2886 lmm_objects = &(lumk->lmm_objects[0]);
2887 lmm_objects->l_object_id = lsm->lsm_object_id;
2889 lum_size = lov_mds_md_size(0, lum.lmm_magic);
2893 lumk->lmm_object_id = lsm->lsm_object_id;
2894 lumk->lmm_object_seq = lsm->lsm_object_seq;
2895 lumk->lmm_stripe_count = 1;
2897 if (cfs_copy_to_user(lump, lumk, lum_size))
2901 OBD_FREE(lumk, lum_size);
2907 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
2908 void *karg, void *uarg)
2910 struct obd_device *obd = exp->exp_obd;
2911 struct obd_ioctl_data *data = karg;
2915 if (!cfs_try_module_get(THIS_MODULE)) {
2916 CERROR("Can't get module. Is it alive?");
2920 case OBD_IOC_LOV_GET_CONFIG: {
2922 struct lov_desc *desc;
2923 struct obd_uuid uuid;
2927 if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
2928 GOTO(out, err = -EINVAL);
2930 data = (struct obd_ioctl_data *)buf;
2932 if (sizeof(*desc) > data->ioc_inllen1) {
2933 obd_ioctl_freedata(buf, len);
2934 GOTO(out, err = -EINVAL);
2937 if (data->ioc_inllen2 < sizeof(uuid)) {
2938 obd_ioctl_freedata(buf, len);
2939 GOTO(out, err = -EINVAL);
2942 desc = (struct lov_desc *)data->ioc_inlbuf1;
2943 desc->ld_tgt_count = 1;
2944 desc->ld_active_tgt_count = 1;
2945 desc->ld_default_stripe_count = 1;
2946 desc->ld_default_stripe_size = 0;
2947 desc->ld_default_stripe_offset = 0;
2948 desc->ld_pattern = 0;
2949 memcpy(&desc->ld_uuid, &obd->obd_uuid, sizeof(uuid));
2951 memcpy(data->ioc_inlbuf2, &obd->obd_uuid, sizeof(uuid));
2953 err = cfs_copy_to_user((void *)uarg, buf, len);
2956 obd_ioctl_freedata(buf, len);
2959 case LL_IOC_LOV_SETSTRIPE:
2960 err = obd_alloc_memmd(exp, karg);
2964 case LL_IOC_LOV_GETSTRIPE:
2965 err = osc_getstripe(karg, uarg);
2967 case OBD_IOC_CLIENT_RECOVER:
2968 err = ptlrpc_recover_import(obd->u.cli.cl_import,
2969 data->ioc_inlbuf1, 0);
2973 case IOC_OSC_SET_ACTIVE:
2974 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
2977 case OBD_IOC_POLL_QUOTACHECK:
2978 err = osc_quota_poll_check(exp, (struct if_quotacheck *)karg);
2980 case OBD_IOC_PING_TARGET:
2981 err = ptlrpc_obd_ping(obd);
2984 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
2985 cmd, cfs_curproc_comm());
2986 GOTO(out, err = -ENOTTY);
2989 cfs_module_put(THIS_MODULE);
2993 static int osc_get_info(const struct lu_env *env, struct obd_export *exp,
2994 obd_count keylen, void *key, __u32 *vallen, void *val,
2995 struct lov_stripe_md *lsm)
2998 if (!vallen || !val)
3001 if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
3002 __u32 *stripe = val;
3003 *vallen = sizeof(*stripe);
3006 } else if (KEY_IS(KEY_LAST_ID)) {
3007 struct ptlrpc_request *req;
3012 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3013 &RQF_OST_GET_INFO_LAST_ID);
3017 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3018 RCL_CLIENT, keylen);
3019 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3021 ptlrpc_request_free(req);
3025 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3026 memcpy(tmp, key, keylen);
3028 req->rq_no_delay = req->rq_no_resend = 1;
3029 ptlrpc_request_set_replen(req);
3030 rc = ptlrpc_queue_wait(req);
3034 reply = req_capsule_server_get(&req->rq_pill, &RMF_OBD_ID);
3036 GOTO(out, rc = -EPROTO);
3038 *((obd_id *)val) = *reply;
3040 ptlrpc_req_finished(req);
3042 } else if (KEY_IS(KEY_FIEMAP)) {
3043 struct ptlrpc_request *req;
3044 struct ll_user_fiemap *reply;
3048 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3049 &RQF_OST_GET_INFO_FIEMAP);
3053 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_KEY,
3054 RCL_CLIENT, keylen);
3055 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3056 RCL_CLIENT, *vallen);
3057 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3058 RCL_SERVER, *vallen);
3060 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3062 ptlrpc_request_free(req);
3066 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_KEY);
3067 memcpy(tmp, key, keylen);
3068 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3069 memcpy(tmp, val, *vallen);
3071 ptlrpc_request_set_replen(req);
3072 rc = ptlrpc_queue_wait(req);
3076 reply = req_capsule_server_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3078 GOTO(out1, rc = -EPROTO);
3080 memcpy(val, reply, *vallen);
3082 ptlrpc_req_finished(req);
3090 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
3091 obd_count keylen, void *key, obd_count vallen,
3092 void *val, struct ptlrpc_request_set *set)
3094 struct ptlrpc_request *req;
3095 struct obd_device *obd = exp->exp_obd;
3096 struct obd_import *imp = class_exp2cliimp(exp);
3101 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3103 if (KEY_IS(KEY_CHECKSUM)) {
3104 if (vallen != sizeof(int))
3106 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3110 if (KEY_IS(KEY_SPTLRPC_CONF)) {
3111 sptlrpc_conf_client_adapt(obd);
3115 if (KEY_IS(KEY_FLUSH_CTX)) {
3116 sptlrpc_import_flush_my_ctx(imp);
3120 if (KEY_IS(KEY_CACHE_SET)) {
3121 struct client_obd *cli = &obd->u.cli;
3123 LASSERT(cli->cl_cache == NULL); /* only once */
3124 cli->cl_cache = (struct cl_client_cache *)val;
3125 cfs_atomic_inc(&cli->cl_cache->ccc_users);
3126 cli->cl_lru_left = &cli->cl_cache->ccc_lru_left;
3128 /* add this osc into entity list */
3129 LASSERT(cfs_list_empty(&cli->cl_lru_osc));
3130 cfs_spin_lock(&cli->cl_cache->ccc_lru_lock);
3131 cfs_list_add(&cli->cl_lru_osc, &cli->cl_cache->ccc_lru);
3132 cfs_spin_unlock(&cli->cl_cache->ccc_lru_lock);
3137 if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
3138 struct client_obd *cli = &obd->u.cli;
3139 int nr = cfs_atomic_read(&cli->cl_lru_in_list) >> 1;
3140 int target = *(int *)val;
3142 nr = osc_lru_shrink(cli, min(nr, target));
3147 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
3150 /* We pass all other commands directly to OST. Since nobody calls osc
3151 methods directly and everybody is supposed to go through LOV, we
3152 assume lov checked invalid values for us.
3153 The only recognised values so far are evict_by_nid and mds_conn.
3154 Even if something bad goes through, we'd get a -EINVAL from OST
3157 if (KEY_IS(KEY_GRANT_SHRINK))
3158 req = ptlrpc_request_alloc(imp, &RQF_OST_SET_GRANT_INFO);
3160 req = ptlrpc_request_alloc(imp, &RQF_OBD_SET_INFO);
3165 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3166 RCL_CLIENT, keylen);
3167 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
3168 RCL_CLIENT, vallen);
3169 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
3171 ptlrpc_request_free(req);
3175 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3176 memcpy(tmp, key, keylen);
3177 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_VAL);
3178 memcpy(tmp, val, vallen);
3180 if (KEY_IS(KEY_GRANT_SHRINK)) {
3181 struct osc_grant_args *aa;
3184 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
3185 aa = ptlrpc_req_async_args(req);
3188 ptlrpc_req_finished(req);
3191 *oa = ((struct ost_body *)val)->oa;
3193 req->rq_interpret_reply = osc_shrink_grant_interpret;
3196 ptlrpc_request_set_replen(req);
3197 if (!KEY_IS(KEY_GRANT_SHRINK)) {
3198 LASSERT(set != NULL);
3199 ptlrpc_set_add_req(set, req);
3200 ptlrpc_check_set(NULL, set);
3202 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
3208 static int osc_llog_init(struct obd_device *obd, struct obd_llog_group *olg,
3209 struct obd_device *disk_obd, int *index)
3211 /* this code is not supposed to be used with LOD/OSP
3212 * to be removed soon */
3217 static int osc_llog_finish(struct obd_device *obd, int count)
3219 struct llog_ctxt *ctxt;
3223 ctxt = llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
3225 llog_cat_close(NULL, ctxt->loc_handle);
3226 llog_cleanup(NULL, ctxt);
3229 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3231 llog_cleanup(NULL, ctxt);
3235 static int osc_reconnect(const struct lu_env *env,
3236 struct obd_export *exp, struct obd_device *obd,
3237 struct obd_uuid *cluuid,
3238 struct obd_connect_data *data,
3241 struct client_obd *cli = &obd->u.cli;
3243 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3246 client_obd_list_lock(&cli->cl_loi_list_lock);
3247 data->ocd_grant = (cli->cl_avail_grant + cli->cl_dirty) ?:
3248 2 * cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT;
3249 lost_grant = cli->cl_lost_grant;
3250 cli->cl_lost_grant = 0;
3251 client_obd_list_unlock(&cli->cl_loi_list_lock);
3253 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
3254 " ocd_grant: %d, lost: %ld.\n", data->ocd_connect_flags,
3255 data->ocd_version, data->ocd_grant, lost_grant);
3261 static int osc_disconnect(struct obd_export *exp)
3263 struct obd_device *obd = class_exp2obd(exp);
3264 struct llog_ctxt *ctxt;
3267 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3269 if (obd->u.cli.cl_conn_count == 1) {
3270 /* Flush any remaining cancel messages out to the
3272 llog_sync(ctxt, exp, 0);
3274 llog_ctxt_put(ctxt);
3276 CDEBUG(D_HA, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
3280 rc = client_disconnect_export(exp);
3282 * Initially we put del_shrink_grant before disconnect_export, but it
3283 * causes the following problem if setup (connect) and cleanup
3284 * (disconnect) are tangled together.
3285 * connect p1 disconnect p2
3286 * ptlrpc_connect_import
3287 * ............... class_manual_cleanup
3290 * ptlrpc_connect_interrupt
3292 * add this client to shrink list
3294 * Bang! pinger trigger the shrink.
3295 * So the osc should be disconnected from the shrink list, after we
3296 * are sure the import has been destroyed. BUG18662
3298 if (obd->u.cli.cl_import == NULL)
3299 osc_del_shrink_grant(&obd->u.cli);
3303 static int osc_import_event(struct obd_device *obd,
3304 struct obd_import *imp,
3305 enum obd_import_event event)
3307 struct client_obd *cli;
3311 LASSERT(imp->imp_obd == obd);
3314 case IMP_EVENT_DISCON: {
3316 client_obd_list_lock(&cli->cl_loi_list_lock);
3317 cli->cl_avail_grant = 0;
3318 cli->cl_lost_grant = 0;
3319 client_obd_list_unlock(&cli->cl_loi_list_lock);
3322 case IMP_EVENT_INACTIVE: {
3323 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
3326 case IMP_EVENT_INVALIDATE: {
3327 struct ldlm_namespace *ns = obd->obd_namespace;
3331 env = cl_env_get(&refcheck);
3335 /* all pages go to failing rpcs due to the invalid
3337 osc_io_unplug(env, cli, NULL, PDL_POLICY_ROUND);
3339 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3340 cl_env_put(env, &refcheck);
3345 case IMP_EVENT_ACTIVE: {
3346 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
3349 case IMP_EVENT_OCD: {
3350 struct obd_connect_data *ocd = &imp->imp_connect_data;
3352 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3353 osc_init_grant(&obd->u.cli, ocd);
3356 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3357 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3359 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
3362 case IMP_EVENT_DEACTIVATE: {
3363 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_DEACTIVATE, NULL);
3366 case IMP_EVENT_ACTIVATE: {
3367 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVATE, NULL);
3371 CERROR("Unknown import event %d\n", event);
3378 * Determine whether the lock can be canceled before replaying the lock
3379 * during recovery, see bug16774 for detailed information.
3381 * \retval zero the lock can't be canceled
3382 * \retval other ok to cancel
3384 static int osc_cancel_for_recovery(struct ldlm_lock *lock)
3386 check_res_locked(lock->l_resource);
3389 * Cancel all unused extent lock in granted mode LCK_PR or LCK_CR.
3391 * XXX as a future improvement, we can also cancel unused write lock
3392 * if it doesn't have dirty data and active mmaps.
3394 if (lock->l_resource->lr_type == LDLM_EXTENT &&
3395 (lock->l_granted_mode == LCK_PR ||
3396 lock->l_granted_mode == LCK_CR) &&
3397 (osc_dlm_lock_pageref(lock) == 0))
3403 static int brw_queue_work(const struct lu_env *env, void *data)
3405 struct client_obd *cli = data;
3407 CDEBUG(D_CACHE, "Run writeback work for client obd %p.\n", cli);
3409 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
3413 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3415 struct lprocfs_static_vars lvars = { 0 };
3416 struct client_obd *cli = &obd->u.cli;
3421 rc = ptlrpcd_addref();
3425 rc = client_obd_setup(obd, lcfg);
3427 GOTO(out_ptlrpcd, rc);
3429 handler = ptlrpcd_alloc_work(cli->cl_import, brw_queue_work, cli);
3430 if (IS_ERR(handler))
3431 GOTO(out_client_setup, rc = PTR_ERR(handler));
3432 cli->cl_writeback_work = handler;
3434 rc = osc_quota_setup(obd);
3436 GOTO(out_ptlrpcd_work, rc);
3438 cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
3439 lprocfs_osc_init_vars(&lvars);
3440 if (lprocfs_obd_setup(obd, lvars.obd_vars) == 0) {
3441 lproc_osc_attach_seqstat(obd);
3442 sptlrpc_lprocfs_cliobd_attach(obd);
3443 ptlrpc_lprocfs_register_obd(obd);
3446 /* We need to allocate a few requests more, because
3447 * brw_interpret tries to create new requests before freeing
3448 * previous ones, Ideally we want to have 2x max_rpcs_in_flight
3449 * reserved, but I'm afraid that might be too much wasted RAM
3450 * in fact, so 2 is just my guess and still should work. */
3451 cli->cl_import->imp_rq_pool =
3452 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
3454 ptlrpc_add_rqs_to_pool);
3456 CFS_INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
3457 ns_register_cancel(obd->obd_namespace, osc_cancel_for_recovery);
3461 ptlrpcd_destroy_work(handler);
3463 client_obd_cleanup(obd);
3469 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
3475 case OBD_CLEANUP_EARLY: {
3476 struct obd_import *imp;
3477 imp = obd->u.cli.cl_import;
3478 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
3479 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
3480 ptlrpc_deactivate_import(imp);
3481 cfs_spin_lock(&imp->imp_lock);
3482 imp->imp_pingable = 0;
3483 cfs_spin_unlock(&imp->imp_lock);
3486 case OBD_CLEANUP_EXPORTS: {
3487 struct client_obd *cli = &obd->u.cli;
3489 * for echo client, export may be on zombie list, wait for
3490 * zombie thread to cull it, because cli.cl_import will be
3491 * cleared in client_disconnect_export():
3492 * class_export_destroy() -> obd_cleanup() ->
3493 * echo_device_free() -> echo_client_cleanup() ->
3494 * obd_disconnect() -> osc_disconnect() ->
3495 * client_disconnect_export()
3497 obd_zombie_barrier();
3498 if (cli->cl_writeback_work) {
3499 ptlrpcd_destroy_work(cli->cl_writeback_work);
3500 cli->cl_writeback_work = NULL;
3502 obd_cleanup_client_import(obd);
3503 ptlrpc_lprocfs_unregister_obd(obd);
3504 lprocfs_obd_cleanup(obd);
3505 rc = obd_llog_finish(obd, 0);
3507 CERROR("failed to cleanup llogging subsystems\n");
3514 int osc_cleanup(struct obd_device *obd)
3516 struct client_obd *cli = &obd->u.cli;
3522 if (cli->cl_cache != NULL) {
3523 LASSERT(cfs_atomic_read(&cli->cl_cache->ccc_users) > 0);
3524 cfs_spin_lock(&cli->cl_cache->ccc_lru_lock);
3525 cfs_list_del_init(&cli->cl_lru_osc);
3526 cfs_spin_unlock(&cli->cl_cache->ccc_lru_lock);
3527 cli->cl_lru_left = NULL;
3528 cfs_atomic_dec(&cli->cl_cache->ccc_users);
3529 cli->cl_cache = NULL;
3532 /* free memory of osc quota cache */
3533 osc_quota_cleanup(obd);
3535 rc = client_obd_cleanup(obd);
3541 int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
3543 struct lprocfs_static_vars lvars = { 0 };
3546 lprocfs_osc_init_vars(&lvars);
3548 switch (lcfg->lcfg_command) {
3550 rc = class_process_proc_param(PARAM_OSC, lvars.obd_vars,
3560 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
3562 return osc_process_config_base(obd, buf);
3565 struct obd_ops osc_obd_ops = {
3566 .o_owner = THIS_MODULE,
3567 .o_setup = osc_setup,
3568 .o_precleanup = osc_precleanup,
3569 .o_cleanup = osc_cleanup,
3570 .o_add_conn = client_import_add_conn,
3571 .o_del_conn = client_import_del_conn,
3572 .o_connect = client_connect_import,
3573 .o_reconnect = osc_reconnect,
3574 .o_disconnect = osc_disconnect,
3575 .o_statfs = osc_statfs,
3576 .o_statfs_async = osc_statfs_async,
3577 .o_packmd = osc_packmd,
3578 .o_unpackmd = osc_unpackmd,
3579 .o_create = osc_create,
3580 .o_destroy = osc_destroy,
3581 .o_getattr = osc_getattr,
3582 .o_getattr_async = osc_getattr_async,
3583 .o_setattr = osc_setattr,
3584 .o_setattr_async = osc_setattr_async,
3586 .o_punch = osc_punch,
3588 .o_enqueue = osc_enqueue,
3589 .o_change_cbdata = osc_change_cbdata,
3590 .o_find_cbdata = osc_find_cbdata,
3591 .o_cancel = osc_cancel,
3592 .o_cancel_unused = osc_cancel_unused,
3593 .o_iocontrol = osc_iocontrol,
3594 .o_get_info = osc_get_info,
3595 .o_set_info_async = osc_set_info_async,
3596 .o_import_event = osc_import_event,
3597 .o_llog_init = osc_llog_init,
3598 .o_llog_finish = osc_llog_finish,
3599 .o_process_config = osc_process_config,
3600 .o_quotactl = osc_quotactl,
3601 .o_quotacheck = osc_quotacheck,
3604 extern struct lu_kmem_descr osc_caches[];
3605 extern cfs_spinlock_t osc_ast_guard;
3606 extern cfs_lock_class_key_t osc_ast_guard_class;
3608 int __init osc_init(void)
3610 struct lprocfs_static_vars lvars = { 0 };
3614 /* print an address of _any_ initialized kernel symbol from this
3615 * module, to allow debugging with gdb that doesn't support data
3616 * symbols from modules.*/
3617 CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
3619 rc = lu_kmem_init(osc_caches);
3621 lprocfs_osc_init_vars(&lvars);
3623 rc = class_register_type(&osc_obd_ops, NULL, lvars.module_vars,
3624 LUSTRE_OSC_NAME, &osc_device_type);
3626 lu_kmem_fini(osc_caches);
3630 cfs_spin_lock_init(&osc_ast_guard);
3631 cfs_lockdep_set_class(&osc_ast_guard, &osc_ast_guard_class);
3637 static void /*__exit*/ osc_exit(void)
3639 class_unregister_type(LUSTRE_OSC_NAME);
3640 lu_kmem_fini(osc_caches);
3643 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
3644 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3645 MODULE_LICENSE("GPL");
3647 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);