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, Intel Corporation.
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 /* Return, i.e. cancel nothing, only if ELC is supported (flag in
666 * export) but disabled through procfs (flag in NS).
668 * This distinguishes from a case when ELC is not supported originally,
669 * when we still want to cancel locks in advance and just cancel them
670 * locally, without sending any RPC. */
671 if (exp_connect_cancelset(exp) && !ns_connect_cancelset(ns))
674 osc_build_res_name(oa->o_id, oa->o_seq, &res_id);
675 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
679 LDLM_RESOURCE_ADDREF(res);
680 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
681 lock_flags, 0, NULL);
682 LDLM_RESOURCE_DELREF(res);
683 ldlm_resource_putref(res);
687 static int osc_destroy_interpret(const struct lu_env *env,
688 struct ptlrpc_request *req, void *data,
691 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
693 cfs_atomic_dec(&cli->cl_destroy_in_flight);
694 cfs_waitq_signal(&cli->cl_destroy_waitq);
698 static int osc_can_send_destroy(struct client_obd *cli)
700 if (cfs_atomic_inc_return(&cli->cl_destroy_in_flight) <=
701 cli->cl_max_rpcs_in_flight) {
702 /* The destroy request can be sent */
705 if (cfs_atomic_dec_return(&cli->cl_destroy_in_flight) <
706 cli->cl_max_rpcs_in_flight) {
708 * The counter has been modified between the two atomic
711 cfs_waitq_signal(&cli->cl_destroy_waitq);
716 int osc_create(const struct lu_env *env, struct obd_export *exp,
717 struct obdo *oa, struct lov_stripe_md **ea,
718 struct obd_trans_info *oti)
725 LASSERT(oa->o_valid & OBD_MD_FLGROUP);
727 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
728 oa->o_flags == OBD_FL_RECREATE_OBJS) {
729 RETURN(osc_real_create(exp, oa, ea, oti));
732 if (!fid_seq_is_mdt(oa->o_seq))
733 RETURN(osc_real_create(exp, oa, ea, oti));
735 /* we should not get here anymore */
741 /* Destroy requests can be async always on the client, and we don't even really
742 * care about the return code since the client cannot do anything at all about
744 * When the MDS is unlinking a filename, it saves the file objects into a
745 * recovery llog, and these object records are cancelled when the OST reports
746 * they were destroyed and sync'd to disk (i.e. transaction committed).
747 * If the client dies, or the OST is down when the object should be destroyed,
748 * the records are not cancelled, and when the OST reconnects to the MDS next,
749 * it will retrieve the llog unlink logs and then sends the log cancellation
750 * cookies to the MDS after committing destroy transactions. */
751 static int osc_destroy(const struct lu_env *env, struct obd_export *exp,
752 struct obdo *oa, struct lov_stripe_md *ea,
753 struct obd_trans_info *oti, struct obd_export *md_export,
756 struct client_obd *cli = &exp->exp_obd->u.cli;
757 struct ptlrpc_request *req;
758 struct ost_body *body;
759 CFS_LIST_HEAD(cancels);
764 CDEBUG(D_INFO, "oa NULL\n");
768 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
769 LDLM_FL_DISCARD_DATA);
771 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
773 ldlm_lock_list_put(&cancels, l_bl_ast, count);
777 osc_set_capa_size(req, &RMF_CAPA1, (struct obd_capa *)capa);
778 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
781 ptlrpc_request_free(req);
785 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
786 ptlrpc_at_set_req_timeout(req);
788 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
789 oa->o_lcookie = *oti->oti_logcookies;
790 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
792 lustre_set_wire_obdo(&body->oa, oa);
794 osc_pack_capa(req, body, (struct obd_capa *)capa);
795 ptlrpc_request_set_replen(req);
797 /* If osc_destory is for destroying the unlink orphan,
798 * sent from MDT to OST, which should not be blocked here,
799 * because the process might be triggered by ptlrpcd, and
800 * it is not good to block ptlrpcd thread (b=16006)*/
801 if (!(oa->o_flags & OBD_FL_DELORPHAN)) {
802 req->rq_interpret_reply = osc_destroy_interpret;
803 if (!osc_can_send_destroy(cli)) {
804 struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP,
808 * Wait until the number of on-going destroy RPCs drops
809 * under max_rpc_in_flight
811 l_wait_event_exclusive(cli->cl_destroy_waitq,
812 osc_can_send_destroy(cli), &lwi);
816 /* Do not wait for response */
817 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
821 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
824 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
826 LASSERT(!(oa->o_valid & bits));
829 client_obd_list_lock(&cli->cl_loi_list_lock);
830 oa->o_dirty = cli->cl_dirty;
831 if (unlikely(cli->cl_dirty - cli->cl_dirty_transit >
832 cli->cl_dirty_max)) {
833 CERROR("dirty %lu - %lu > dirty_max %lu\n",
834 cli->cl_dirty, cli->cl_dirty_transit, cli->cl_dirty_max);
836 } else if (unlikely(cfs_atomic_read(&obd_dirty_pages) -
837 cfs_atomic_read(&obd_dirty_transit_pages) >
838 (long)(obd_max_dirty_pages + 1))) {
839 /* The cfs_atomic_read() allowing the cfs_atomic_inc() are
840 * not covered by a lock thus they may safely race and trip
841 * this CERROR() unless we add in a small fudge factor (+1). */
842 CERROR("dirty %d - %d > system dirty_max %d\n",
843 cfs_atomic_read(&obd_dirty_pages),
844 cfs_atomic_read(&obd_dirty_transit_pages),
845 obd_max_dirty_pages);
847 } else if (unlikely(cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff)) {
848 CERROR("dirty %lu - dirty_max %lu too big???\n",
849 cli->cl_dirty, cli->cl_dirty_max);
852 long max_in_flight = (cli->cl_max_pages_per_rpc <<
854 (cli->cl_max_rpcs_in_flight + 1);
855 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
857 oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
858 oa->o_dropped = cli->cl_lost_grant;
859 cli->cl_lost_grant = 0;
860 client_obd_list_unlock(&cli->cl_loi_list_lock);
861 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
862 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
866 void osc_update_next_shrink(struct client_obd *cli)
868 cli->cl_next_shrink_grant =
869 cfs_time_shift(cli->cl_grant_shrink_interval);
870 CDEBUG(D_CACHE, "next time %ld to shrink grant \n",
871 cli->cl_next_shrink_grant);
874 static void __osc_update_grant(struct client_obd *cli, obd_size grant)
876 client_obd_list_lock(&cli->cl_loi_list_lock);
877 cli->cl_avail_grant += grant;
878 client_obd_list_unlock(&cli->cl_loi_list_lock);
881 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
883 if (body->oa.o_valid & OBD_MD_FLGRANT) {
884 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
885 __osc_update_grant(cli, body->oa.o_grant);
889 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
890 obd_count keylen, void *key, obd_count vallen,
891 void *val, struct ptlrpc_request_set *set);
893 static int osc_shrink_grant_interpret(const struct lu_env *env,
894 struct ptlrpc_request *req,
897 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
898 struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
899 struct ost_body *body;
902 __osc_update_grant(cli, oa->o_grant);
906 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
908 osc_update_grant(cli, body);
914 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
916 client_obd_list_lock(&cli->cl_loi_list_lock);
917 oa->o_grant = cli->cl_avail_grant / 4;
918 cli->cl_avail_grant -= oa->o_grant;
919 client_obd_list_unlock(&cli->cl_loi_list_lock);
920 if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
921 oa->o_valid |= OBD_MD_FLFLAGS;
924 oa->o_flags |= OBD_FL_SHRINK_GRANT;
925 osc_update_next_shrink(cli);
928 /* Shrink the current grant, either from some large amount to enough for a
929 * full set of in-flight RPCs, or if we have already shrunk to that limit
930 * then to enough for a single RPC. This avoids keeping more grant than
931 * needed, and avoids shrinking the grant piecemeal. */
932 static int osc_shrink_grant(struct client_obd *cli)
934 long target = (cli->cl_max_rpcs_in_flight + 1) *
935 cli->cl_max_pages_per_rpc;
937 client_obd_list_lock(&cli->cl_loi_list_lock);
938 if (cli->cl_avail_grant <= target)
939 target = cli->cl_max_pages_per_rpc;
940 client_obd_list_unlock(&cli->cl_loi_list_lock);
942 return osc_shrink_grant_to_target(cli, target);
945 int osc_shrink_grant_to_target(struct client_obd *cli, long target)
948 struct ost_body *body;
951 client_obd_list_lock(&cli->cl_loi_list_lock);
952 /* Don't shrink if we are already above or below the desired limit
953 * We don't want to shrink below a single RPC, as that will negatively
954 * impact block allocation and long-term performance. */
955 if (target < cli->cl_max_pages_per_rpc)
956 target = cli->cl_max_pages_per_rpc;
958 if (target >= cli->cl_avail_grant) {
959 client_obd_list_unlock(&cli->cl_loi_list_lock);
962 client_obd_list_unlock(&cli->cl_loi_list_lock);
968 osc_announce_cached(cli, &body->oa, 0);
970 client_obd_list_lock(&cli->cl_loi_list_lock);
971 body->oa.o_grant = cli->cl_avail_grant - target;
972 cli->cl_avail_grant = target;
973 client_obd_list_unlock(&cli->cl_loi_list_lock);
974 if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
975 body->oa.o_valid |= OBD_MD_FLFLAGS;
976 body->oa.o_flags = 0;
978 body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
979 osc_update_next_shrink(cli);
981 rc = osc_set_info_async(NULL, cli->cl_import->imp_obd->obd_self_export,
982 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
983 sizeof(*body), body, NULL);
985 __osc_update_grant(cli, body->oa.o_grant);
990 #define GRANT_SHRINK_LIMIT PTLRPC_MAX_BRW_SIZE
991 static int osc_should_shrink_grant(struct client_obd *client)
993 cfs_time_t time = cfs_time_current();
994 cfs_time_t next_shrink = client->cl_next_shrink_grant;
996 if ((client->cl_import->imp_connect_data.ocd_connect_flags &
997 OBD_CONNECT_GRANT_SHRINK) == 0)
1000 if (cfs_time_aftereq(time, next_shrink - 5 * CFS_TICK)) {
1001 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
1002 client->cl_avail_grant > GRANT_SHRINK_LIMIT)
1005 osc_update_next_shrink(client);
1010 static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
1012 struct client_obd *client;
1014 cfs_list_for_each_entry(client, &item->ti_obd_list,
1015 cl_grant_shrink_list) {
1016 if (osc_should_shrink_grant(client))
1017 osc_shrink_grant(client);
1022 static int osc_add_shrink_grant(struct client_obd *client)
1026 rc = ptlrpc_add_timeout_client(client->cl_grant_shrink_interval,
1028 osc_grant_shrink_grant_cb, NULL,
1029 &client->cl_grant_shrink_list);
1031 CERROR("add grant client %s error %d\n",
1032 client->cl_import->imp_obd->obd_name, rc);
1035 CDEBUG(D_CACHE, "add grant client %s \n",
1036 client->cl_import->imp_obd->obd_name);
1037 osc_update_next_shrink(client);
1041 static int osc_del_shrink_grant(struct client_obd *client)
1043 return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
1047 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1050 * ocd_grant is the total grant amount we're expect to hold: if we've
1051 * been evicted, it's the new avail_grant amount, cl_dirty will drop
1052 * to 0 as inflight RPCs fail out; otherwise, it's avail_grant + dirty.
1054 * race is tolerable here: if we're evicted, but imp_state already
1055 * left EVICTED state, then cl_dirty must be 0 already.
1057 client_obd_list_lock(&cli->cl_loi_list_lock);
1058 if (cli->cl_import->imp_state == LUSTRE_IMP_EVICTED)
1059 cli->cl_avail_grant = ocd->ocd_grant;
1061 cli->cl_avail_grant = ocd->ocd_grant - cli->cl_dirty;
1063 if (cli->cl_avail_grant < 0) {
1064 CWARN("%s: available grant < 0, the OSS is probably not running"
1065 " with patch from bug20278 (%ld) \n",
1066 cli->cl_import->imp_obd->obd_name, cli->cl_avail_grant);
1067 /* workaround for 1.6 servers which do not have
1068 * the patch from bug20278 */
1069 cli->cl_avail_grant = ocd->ocd_grant;
1072 /* determine the appropriate chunk size used by osc_extent. */
1073 cli->cl_chunkbits = max_t(int, CFS_PAGE_SHIFT, ocd->ocd_blocksize);
1074 client_obd_list_unlock(&cli->cl_loi_list_lock);
1076 CDEBUG(D_CACHE, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld."
1077 "chunk bits: %d.\n", cli->cl_import->imp_obd->obd_name,
1078 cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits);
1080 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
1081 cfs_list_empty(&cli->cl_grant_shrink_list))
1082 osc_add_shrink_grant(cli);
1085 /* We assume that the reason this OSC got a short read is because it read
1086 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1087 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1088 * this stripe never got written at or beyond this stripe offset yet. */
1089 static void handle_short_read(int nob_read, obd_count page_count,
1090 struct brw_page **pga)
1095 /* skip bytes read OK */
1096 while (nob_read > 0) {
1097 LASSERT (page_count > 0);
1099 if (pga[i]->count > nob_read) {
1100 /* EOF inside this page */
1101 ptr = cfs_kmap(pga[i]->pg) +
1102 (pga[i]->off & ~CFS_PAGE_MASK);
1103 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1104 cfs_kunmap(pga[i]->pg);
1110 nob_read -= pga[i]->count;
1115 /* zero remaining pages */
1116 while (page_count-- > 0) {
1117 ptr = cfs_kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
1118 memset(ptr, 0, pga[i]->count);
1119 cfs_kunmap(pga[i]->pg);
1124 static int check_write_rcs(struct ptlrpc_request *req,
1125 int requested_nob, int niocount,
1126 obd_count page_count, struct brw_page **pga)
1131 remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1132 sizeof(*remote_rcs) *
1134 if (remote_rcs == NULL) {
1135 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1139 /* return error if any niobuf was in error */
1140 for (i = 0; i < niocount; i++) {
1141 if ((int)remote_rcs[i] < 0)
1142 return(remote_rcs[i]);
1144 if (remote_rcs[i] != 0) {
1145 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1146 i, remote_rcs[i], req);
1151 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
1152 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1153 req->rq_bulk->bd_nob_transferred, requested_nob);
1160 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1162 if (p1->flag != p2->flag) {
1163 unsigned mask = ~(OBD_BRW_FROM_GRANT| OBD_BRW_NOCACHE|
1164 OBD_BRW_SYNC|OBD_BRW_ASYNC|OBD_BRW_NOQUOTA);
1166 /* warn if we try to combine flags that we don't know to be
1167 * safe to combine */
1168 if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
1169 CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1170 "report this at http://bugs.whamcloud.com/\n",
1171 p1->flag, p2->flag);
1176 return (p1->off + p1->count == p2->off);
1179 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
1180 struct brw_page **pga, int opc,
1181 cksum_type_t cksum_type)
1185 struct cfs_crypto_hash_desc *hdesc;
1186 unsigned int bufsize;
1188 unsigned char cfs_alg = cksum_obd2cfs(cksum_type);
1190 LASSERT(pg_count > 0);
1192 hdesc = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1193 if (IS_ERR(hdesc)) {
1194 CERROR("Unable to initialize checksum hash %s\n",
1195 cfs_crypto_hash_name(cfs_alg));
1196 return PTR_ERR(hdesc);
1199 while (nob > 0 && pg_count > 0) {
1200 int count = pga[i]->count > nob ? nob : pga[i]->count;
1202 /* corrupt the data before we compute the checksum, to
1203 * simulate an OST->client data error */
1204 if (i == 0 && opc == OST_READ &&
1205 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE)) {
1206 unsigned char *ptr = cfs_kmap(pga[i]->pg);
1207 int off = pga[i]->off & ~CFS_PAGE_MASK;
1208 memcpy(ptr + off, "bad1", min(4, nob));
1209 cfs_kunmap(pga[i]->pg);
1211 cfs_crypto_hash_update_page(hdesc, pga[i]->pg,
1212 pga[i]->off & ~CFS_PAGE_MASK,
1214 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d checksum %x\n",
1215 (int)(pga[i]->off & ~CFS_PAGE_MASK), cksum);
1217 nob -= pga[i]->count;
1223 err = cfs_crypto_hash_final(hdesc, (unsigned char *)&cksum, &bufsize);
1226 cfs_crypto_hash_final(hdesc, NULL, NULL);
1228 /* For sending we only compute the wrong checksum instead
1229 * of corrupting the data so it is still correct on a redo */
1230 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1236 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
1237 struct lov_stripe_md *lsm, obd_count page_count,
1238 struct brw_page **pga,
1239 struct ptlrpc_request **reqp,
1240 struct obd_capa *ocapa, int reserve,
1243 struct ptlrpc_request *req;
1244 struct ptlrpc_bulk_desc *desc;
1245 struct ost_body *body;
1246 struct obd_ioobj *ioobj;
1247 struct niobuf_remote *niobuf;
1248 int niocount, i, requested_nob, opc, rc;
1249 struct osc_brw_async_args *aa;
1250 struct req_capsule *pill;
1251 struct brw_page *pg_prev;
1254 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1255 RETURN(-ENOMEM); /* Recoverable */
1256 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1257 RETURN(-EINVAL); /* Fatal */
1259 if ((cmd & OBD_BRW_WRITE) != 0) {
1261 req = ptlrpc_request_alloc_pool(cli->cl_import,
1262 cli->cl_import->imp_rq_pool,
1263 &RQF_OST_BRW_WRITE);
1266 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1271 for (niocount = i = 1; i < page_count; i++) {
1272 if (!can_merge_pages(pga[i - 1], pga[i]))
1276 pill = &req->rq_pill;
1277 req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1279 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1280 niocount * sizeof(*niobuf));
1281 osc_set_capa_size(req, &RMF_CAPA1, ocapa);
1283 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1285 ptlrpc_request_free(req);
1288 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
1289 ptlrpc_at_set_req_timeout(req);
1290 /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1292 req->rq_no_retry_einprogress = 1;
1294 if (opc == OST_WRITE)
1295 desc = ptlrpc_prep_bulk_imp(req, page_count,
1296 BULK_GET_SOURCE, OST_BULK_PORTAL);
1298 desc = ptlrpc_prep_bulk_imp(req, page_count,
1299 BULK_PUT_SINK, OST_BULK_PORTAL);
1302 GOTO(out, rc = -ENOMEM);
1303 /* NB request now owns desc and will free it when it gets freed */
1305 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1306 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1307 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1308 LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1310 lustre_set_wire_obdo(&body->oa, oa);
1312 obdo_to_ioobj(oa, ioobj);
1313 ioobj->ioo_bufcnt = niocount;
1314 osc_pack_capa(req, body, ocapa);
1315 LASSERT (page_count > 0);
1317 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1318 struct brw_page *pg = pga[i];
1319 int poff = pg->off & ~CFS_PAGE_MASK;
1321 LASSERT(pg->count > 0);
1322 /* make sure there is no gap in the middle of page array */
1323 LASSERTF(page_count == 1 ||
1324 (ergo(i == 0, poff + pg->count == CFS_PAGE_SIZE) &&
1325 ergo(i > 0 && i < page_count - 1,
1326 poff == 0 && pg->count == CFS_PAGE_SIZE) &&
1327 ergo(i == page_count - 1, poff == 0)),
1328 "i: %d/%d pg: %p off: "LPU64", count: %u\n",
1329 i, page_count, pg, pg->off, pg->count);
1331 LASSERTF(i == 0 || pg->off > pg_prev->off,
1332 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1333 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1335 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1336 pg_prev->pg, page_private(pg_prev->pg),
1337 pg_prev->pg->index, pg_prev->off);
1339 LASSERTF(i == 0 || pg->off > pg_prev->off,
1340 "i %d p_c %u\n", i, page_count);
1342 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1343 (pg->flag & OBD_BRW_SRVLOCK));
1345 ptlrpc_prep_bulk_page_pin(desc, pg->pg, poff, pg->count);
1346 requested_nob += pg->count;
1348 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1350 niobuf->len += pg->count;
1352 niobuf->offset = pg->off;
1353 niobuf->len = pg->count;
1354 niobuf->flags = pg->flag;
1359 LASSERTF((void *)(niobuf - niocount) ==
1360 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1361 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1362 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1364 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1366 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1367 body->oa.o_valid |= OBD_MD_FLFLAGS;
1368 body->oa.o_flags = 0;
1370 body->oa.o_flags |= OBD_FL_RECOV_RESEND;
1373 if (osc_should_shrink_grant(cli))
1374 osc_shrink_grant_local(cli, &body->oa);
1376 /* size[REQ_REC_OFF] still sizeof (*body) */
1377 if (opc == OST_WRITE) {
1378 if (cli->cl_checksum &&
1379 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1380 /* store cl_cksum_type in a local variable since
1381 * it can be changed via lprocfs */
1382 cksum_type_t cksum_type = cli->cl_cksum_type;
1384 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1385 oa->o_flags &= OBD_FL_LOCAL_MASK;
1386 body->oa.o_flags = 0;
1388 body->oa.o_flags |= cksum_type_pack(cksum_type);
1389 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1390 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1394 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1396 /* save this in 'oa', too, for later checking */
1397 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1398 oa->o_flags |= cksum_type_pack(cksum_type);
1400 /* clear out the checksum flag, in case this is a
1401 * resend but cl_checksum is no longer set. b=11238 */
1402 oa->o_valid &= ~OBD_MD_FLCKSUM;
1404 oa->o_cksum = body->oa.o_cksum;
1405 /* 1 RC per niobuf */
1406 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1407 sizeof(__u32) * niocount);
1409 if (cli->cl_checksum &&
1410 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1411 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1412 body->oa.o_flags = 0;
1413 body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
1414 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1417 ptlrpc_request_set_replen(req);
1419 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1420 aa = ptlrpc_req_async_args(req);
1422 aa->aa_requested_nob = requested_nob;
1423 aa->aa_nio_count = niocount;
1424 aa->aa_page_count = page_count;
1428 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1429 if (ocapa && reserve)
1430 aa->aa_ocapa = capa_get(ocapa);
1436 ptlrpc_req_finished(req);
1440 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1441 __u32 client_cksum, __u32 server_cksum, int nob,
1442 obd_count page_count, struct brw_page **pga,
1443 cksum_type_t client_cksum_type)
1447 cksum_type_t cksum_type;
1449 if (server_cksum == client_cksum) {
1450 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1454 cksum_type = cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
1456 new_cksum = osc_checksum_bulk(nob, page_count, pga, OST_WRITE,
1459 if (cksum_type != client_cksum_type)
1460 msg = "the server did not use the checksum type specified in "
1461 "the original request - likely a protocol problem";
1462 else if (new_cksum == server_cksum)
1463 msg = "changed on the client after we checksummed it - "
1464 "likely false positive due to mmap IO (bug 11742)";
1465 else if (new_cksum == client_cksum)
1466 msg = "changed in transit before arrival at OST";
1468 msg = "changed in transit AND doesn't match the original - "
1469 "likely false positive due to mmap IO (bug 11742)";
1471 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inode "DFID
1472 " object "LPU64"/"LPU64" extent ["LPU64"-"LPU64"]\n",
1473 msg, libcfs_nid2str(peer->nid),
1474 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1475 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1476 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1478 oa->o_valid & OBD_MD_FLGROUP ? oa->o_seq : (__u64)0,
1480 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1481 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1482 "client csum now %x\n", client_cksum, client_cksum_type,
1483 server_cksum, cksum_type, new_cksum);
1487 /* Note rc enters this function as number of bytes transferred */
1488 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1490 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1491 const lnet_process_id_t *peer =
1492 &req->rq_import->imp_connection->c_peer;
1493 struct client_obd *cli = aa->aa_cli;
1494 struct ost_body *body;
1495 __u32 client_cksum = 0;
1498 if (rc < 0 && rc != -EDQUOT) {
1499 DEBUG_REQ(D_INFO, req, "Failed request with rc = %d\n", rc);
1503 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1504 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1506 DEBUG_REQ(D_INFO, req, "Can't unpack body\n");
1510 /* set/clear over quota flag for a uid/gid */
1511 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1512 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA)) {
1513 unsigned int qid[MAXQUOTAS] = { body->oa.o_uid, body->oa.o_gid };
1515 CDEBUG(D_QUOTA, "setdq for [%u %u] with valid "LPX64", flags %x\n",
1516 body->oa.o_uid, body->oa.o_gid, body->oa.o_valid,
1518 osc_quota_setdq(cli, qid, body->oa.o_valid, body->oa.o_flags);
1521 osc_update_grant(cli, body);
1526 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1527 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1529 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1531 CERROR("Unexpected +ve rc %d\n", rc);
1534 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1536 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1539 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1540 check_write_checksum(&body->oa, peer, client_cksum,
1541 body->oa.o_cksum, aa->aa_requested_nob,
1542 aa->aa_page_count, aa->aa_ppga,
1543 cksum_type_unpack(aa->aa_oa->o_flags)))
1546 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1547 aa->aa_page_count, aa->aa_ppga);
1551 /* The rest of this function executes only for OST_READs */
1553 /* if unwrap_bulk failed, return -EAGAIN to retry */
1554 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1556 GOTO(out, rc = -EAGAIN);
1558 if (rc > aa->aa_requested_nob) {
1559 CERROR("Unexpected rc %d (%d requested)\n", rc,
1560 aa->aa_requested_nob);
1564 if (rc != req->rq_bulk->bd_nob_transferred) {
1565 CERROR ("Unexpected rc %d (%d transferred)\n",
1566 rc, req->rq_bulk->bd_nob_transferred);
1570 if (rc < aa->aa_requested_nob)
1571 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1573 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
1574 static int cksum_counter;
1575 __u32 server_cksum = body->oa.o_cksum;
1578 cksum_type_t cksum_type;
1580 cksum_type = cksum_type_unpack(body->oa.o_valid &OBD_MD_FLFLAGS?
1581 body->oa.o_flags : 0);
1582 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1583 aa->aa_ppga, OST_READ,
1586 if (peer->nid == req->rq_bulk->bd_sender) {
1590 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1593 if (server_cksum == ~0 && rc > 0) {
1594 CERROR("Protocol error: server %s set the 'checksum' "
1595 "bit, but didn't send a checksum. Not fatal, "
1596 "but please notify on http://bugs.whamcloud.com/\n",
1597 libcfs_nid2str(peer->nid));
1598 } else if (server_cksum != client_cksum) {
1599 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1600 "%s%s%s inode "DFID" object "
1601 LPU64"/"LPU64" extent "
1602 "["LPU64"-"LPU64"]\n",
1603 req->rq_import->imp_obd->obd_name,
1604 libcfs_nid2str(peer->nid),
1606 body->oa.o_valid & OBD_MD_FLFID ?
1607 body->oa.o_parent_seq : (__u64)0,
1608 body->oa.o_valid & OBD_MD_FLFID ?
1609 body->oa.o_parent_oid : 0,
1610 body->oa.o_valid & OBD_MD_FLFID ?
1611 body->oa.o_parent_ver : 0,
1613 body->oa.o_valid & OBD_MD_FLGROUP ?
1614 body->oa.o_seq : (__u64)0,
1615 aa->aa_ppga[0]->off,
1616 aa->aa_ppga[aa->aa_page_count-1]->off +
1617 aa->aa_ppga[aa->aa_page_count-1]->count -
1619 CERROR("client %x, server %x, cksum_type %x\n",
1620 client_cksum, server_cksum, cksum_type);
1622 aa->aa_oa->o_cksum = client_cksum;
1626 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1629 } else if (unlikely(client_cksum)) {
1630 static int cksum_missed;
1633 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1634 CERROR("Checksum %u requested from %s but not sent\n",
1635 cksum_missed, libcfs_nid2str(peer->nid));
1641 lustre_get_wire_obdo(aa->aa_oa, &body->oa);
1646 static int osc_brw_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1647 struct lov_stripe_md *lsm,
1648 obd_count page_count, struct brw_page **pga,
1649 struct obd_capa *ocapa)
1651 struct ptlrpc_request *req;
1654 int generation, resends = 0;
1655 struct l_wait_info lwi;
1659 cfs_waitq_init(&waitq);
1660 generation = exp->exp_obd->u.cli.cl_import->imp_generation;
1663 rc = osc_brw_prep_request(cmd, &exp->exp_obd->u.cli, oa, lsm,
1664 page_count, pga, &req, ocapa, 0, resends);
1669 req->rq_generation_set = 1;
1670 req->rq_import_generation = generation;
1671 req->rq_sent = cfs_time_current_sec() + resends;
1674 rc = ptlrpc_queue_wait(req);
1676 if (rc == -ETIMEDOUT && req->rq_resend) {
1677 DEBUG_REQ(D_HA, req, "BULK TIMEOUT");
1678 ptlrpc_req_finished(req);
1682 rc = osc_brw_fini_request(req, rc);
1684 ptlrpc_req_finished(req);
1685 /* When server return -EINPROGRESS, client should always retry
1686 * regardless of the number of times the bulk was resent already.*/
1687 if (osc_recoverable_error(rc)) {
1689 if (rc != -EINPROGRESS &&
1690 !client_should_resend(resends, &exp->exp_obd->u.cli)) {
1691 CERROR("%s: too many resend retries for object: "
1692 ""LPU64":"LPU64", rc = %d.\n",
1693 exp->exp_obd->obd_name, oa->o_id, oa->o_seq, rc);
1697 exp->exp_obd->u.cli.cl_import->imp_generation) {
1698 CDEBUG(D_HA, "%s: resend cross eviction for object: "
1699 ""LPU64":"LPU64", rc = %d.\n",
1700 exp->exp_obd->obd_name, oa->o_id, oa->o_seq, rc);
1704 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(resends), NULL, NULL,
1706 l_wait_event(waitq, 0, &lwi);
1711 if (rc == -EAGAIN || rc == -EINPROGRESS)
1716 static int osc_brw_redo_request(struct ptlrpc_request *request,
1717 struct osc_brw_async_args *aa, int rc)
1719 struct ptlrpc_request *new_req;
1720 struct osc_brw_async_args *new_aa;
1721 struct osc_async_page *oap;
1724 DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
1725 "redo for recoverable error %d", rc);
1727 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
1728 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1729 aa->aa_cli, aa->aa_oa,
1730 NULL /* lsm unused by osc currently */,
1731 aa->aa_page_count, aa->aa_ppga,
1732 &new_req, aa->aa_ocapa, 0, 1);
1736 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1737 if (oap->oap_request != NULL) {
1738 LASSERTF(request == oap->oap_request,
1739 "request %p != oap_request %p\n",
1740 request, oap->oap_request);
1741 if (oap->oap_interrupted) {
1742 ptlrpc_req_finished(new_req);
1747 /* New request takes over pga and oaps from old request.
1748 * Note that copying a list_head doesn't work, need to move it... */
1750 new_req->rq_interpret_reply = request->rq_interpret_reply;
1751 new_req->rq_async_args = request->rq_async_args;
1752 /* cap resend delay to the current request timeout, this is similar to
1753 * what ptlrpc does (see after_reply()) */
1754 if (aa->aa_resends > new_req->rq_timeout)
1755 new_req->rq_sent = cfs_time_current_sec() + new_req->rq_timeout;
1757 new_req->rq_sent = cfs_time_current_sec() + aa->aa_resends;
1758 new_req->rq_generation_set = 1;
1759 new_req->rq_import_generation = request->rq_import_generation;
1761 new_aa = ptlrpc_req_async_args(new_req);
1763 CFS_INIT_LIST_HEAD(&new_aa->aa_oaps);
1764 cfs_list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
1765 CFS_INIT_LIST_HEAD(&new_aa->aa_exts);
1766 cfs_list_splice_init(&aa->aa_exts, &new_aa->aa_exts);
1767 new_aa->aa_resends = aa->aa_resends;
1769 cfs_list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1770 if (oap->oap_request) {
1771 ptlrpc_req_finished(oap->oap_request);
1772 oap->oap_request = ptlrpc_request_addref(new_req);
1776 new_aa->aa_ocapa = aa->aa_ocapa;
1777 aa->aa_ocapa = NULL;
1779 /* XXX: This code will run into problem if we're going to support
1780 * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
1781 * and wait for all of them to be finished. We should inherit request
1782 * set from old request. */
1783 ptlrpcd_add_req(new_req, PDL_POLICY_SAME, -1);
1785 DEBUG_REQ(D_INFO, new_req, "new request");
1790 * ugh, we want disk allocation on the target to happen in offset order. we'll
1791 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1792 * fine for our small page arrays and doesn't require allocation. its an
1793 * insertion sort that swaps elements that are strides apart, shrinking the
1794 * stride down until its '1' and the array is sorted.
1796 static void sort_brw_pages(struct brw_page **array, int num)
1799 struct brw_page *tmp;
1803 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1808 for (i = stride ; i < num ; i++) {
1811 while (j >= stride && array[j - stride]->off > tmp->off) {
1812 array[j] = array[j - stride];
1817 } while (stride > 1);
1820 static obd_count max_unfragmented_pages(struct brw_page **pg, obd_count pages)
1826 LASSERT (pages > 0);
1827 offset = pg[i]->off & ~CFS_PAGE_MASK;
1831 if (pages == 0) /* that's all */
1834 if (offset + pg[i]->count < CFS_PAGE_SIZE)
1835 return count; /* doesn't end on page boundary */
1838 offset = pg[i]->off & ~CFS_PAGE_MASK;
1839 if (offset != 0) /* doesn't start on page boundary */
1846 static struct brw_page **osc_build_ppga(struct brw_page *pga, obd_count count)
1848 struct brw_page **ppga;
1851 OBD_ALLOC(ppga, sizeof(*ppga) * count);
1855 for (i = 0; i < count; i++)
1860 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1862 LASSERT(ppga != NULL);
1863 OBD_FREE(ppga, sizeof(*ppga) * count);
1866 static int osc_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
1867 obd_count page_count, struct brw_page *pga,
1868 struct obd_trans_info *oti)
1870 struct obdo *saved_oa = NULL;
1871 struct brw_page **ppga, **orig;
1872 struct obd_import *imp = class_exp2cliimp(exp);
1873 struct client_obd *cli;
1874 int rc, page_count_orig;
1877 LASSERT((imp != NULL) && (imp->imp_obd != NULL));
1878 cli = &imp->imp_obd->u.cli;
1880 if (cmd & OBD_BRW_CHECK) {
1881 /* The caller just wants to know if there's a chance that this
1882 * I/O can succeed */
1884 if (imp->imp_invalid)
1889 /* test_brw with a failed create can trip this, maybe others. */
1890 LASSERT(cli->cl_max_pages_per_rpc);
1894 orig = ppga = osc_build_ppga(pga, page_count);
1897 page_count_orig = page_count;
1899 sort_brw_pages(ppga, page_count);
1900 while (page_count) {
1901 obd_count pages_per_brw;
1903 if (page_count > cli->cl_max_pages_per_rpc)
1904 pages_per_brw = cli->cl_max_pages_per_rpc;
1906 pages_per_brw = page_count;
1908 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1910 if (saved_oa != NULL) {
1911 /* restore previously saved oa */
1912 *oinfo->oi_oa = *saved_oa;
1913 } else if (page_count > pages_per_brw) {
1914 /* save a copy of oa (brw will clobber it) */
1915 OBDO_ALLOC(saved_oa);
1916 if (saved_oa == NULL)
1917 GOTO(out, rc = -ENOMEM);
1918 *saved_oa = *oinfo->oi_oa;
1921 rc = osc_brw_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1922 pages_per_brw, ppga, oinfo->oi_capa);
1927 page_count -= pages_per_brw;
1928 ppga += pages_per_brw;
1932 osc_release_ppga(orig, page_count_orig);
1934 if (saved_oa != NULL)
1935 OBDO_FREE(saved_oa);
1940 static int brw_interpret(const struct lu_env *env,
1941 struct ptlrpc_request *req, void *data, int rc)
1943 struct osc_brw_async_args *aa = data;
1944 struct osc_extent *ext;
1945 struct osc_extent *tmp;
1946 struct cl_object *obj = NULL;
1947 struct client_obd *cli = aa->aa_cli;
1950 rc = osc_brw_fini_request(req, rc);
1951 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
1952 /* When server return -EINPROGRESS, client should always retry
1953 * regardless of the number of times the bulk was resent already. */
1954 if (osc_recoverable_error(rc)) {
1955 if (req->rq_import_generation !=
1956 req->rq_import->imp_generation) {
1957 CDEBUG(D_HA, "%s: resend cross eviction for object: "
1958 ""LPU64":"LPU64", rc = %d.\n",
1959 req->rq_import->imp_obd->obd_name,
1960 aa->aa_oa->o_id, aa->aa_oa->o_seq, rc);
1961 } else if (rc == -EINPROGRESS ||
1962 client_should_resend(aa->aa_resends, aa->aa_cli)) {
1963 rc = osc_brw_redo_request(req, aa, rc);
1965 CERROR("%s: too many resent retries for object: "
1966 ""LPU64":"LPU64", rc = %d.\n",
1967 req->rq_import->imp_obd->obd_name,
1968 aa->aa_oa->o_id, aa->aa_oa->o_seq, rc);
1973 else if (rc == -EAGAIN || rc == -EINPROGRESS)
1978 capa_put(aa->aa_ocapa);
1979 aa->aa_ocapa = NULL;
1982 cfs_list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
1983 if (obj == NULL && rc == 0) {
1984 obj = osc2cl(ext->oe_obj);
1988 cfs_list_del_init(&ext->oe_link);
1989 osc_extent_finish(env, ext, 1, rc);
1991 LASSERT(cfs_list_empty(&aa->aa_exts));
1992 LASSERT(cfs_list_empty(&aa->aa_oaps));
1995 struct obdo *oa = aa->aa_oa;
1996 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
1997 unsigned long valid = 0;
2000 if (oa->o_valid & OBD_MD_FLBLOCKS) {
2001 attr->cat_blocks = oa->o_blocks;
2002 valid |= CAT_BLOCKS;
2004 if (oa->o_valid & OBD_MD_FLMTIME) {
2005 attr->cat_mtime = oa->o_mtime;
2008 if (oa->o_valid & OBD_MD_FLATIME) {
2009 attr->cat_atime = oa->o_atime;
2012 if (oa->o_valid & OBD_MD_FLCTIME) {
2013 attr->cat_ctime = oa->o_ctime;
2017 cl_object_attr_lock(obj);
2018 cl_object_attr_set(env, obj, attr, valid);
2019 cl_object_attr_unlock(obj);
2021 cl_object_put(env, obj);
2023 OBDO_FREE(aa->aa_oa);
2025 cl_req_completion(env, aa->aa_clerq, rc < 0 ? rc :
2026 req->rq_bulk->bd_nob_transferred);
2027 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2028 ptlrpc_lprocfs_brw(req, req->rq_bulk->bd_nob_transferred);
2030 client_obd_list_lock(&cli->cl_loi_list_lock);
2031 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2032 * is called so we know whether to go to sync BRWs or wait for more
2033 * RPCs to complete */
2034 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2035 cli->cl_w_in_flight--;
2037 cli->cl_r_in_flight--;
2038 osc_wake_cache_waiters(cli);
2039 client_obd_list_unlock(&cli->cl_loi_list_lock);
2041 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
2046 * Build an RPC by the list of extent @ext_list. The caller must ensure
2047 * that the total pages in this list are NOT over max pages per RPC.
2048 * Extents in the list must be in OES_RPC state.
2050 int osc_build_rpc(const struct lu_env *env, struct client_obd *cli,
2051 cfs_list_t *ext_list, int cmd, pdl_policy_t pol)
2053 struct ptlrpc_request *req = NULL;
2054 struct osc_extent *ext;
2055 CFS_LIST_HEAD(rpc_list);
2056 struct brw_page **pga = NULL;
2057 struct osc_brw_async_args *aa = NULL;
2058 struct obdo *oa = NULL;
2059 struct osc_async_page *oap;
2060 struct osc_async_page *tmp;
2061 struct cl_req *clerq = NULL;
2062 enum cl_req_type crt = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2063 struct ldlm_lock *lock = NULL;
2064 struct cl_req_attr crattr;
2065 obd_off starting_offset = OBD_OBJECT_EOF;
2066 obd_off ending_offset = 0;
2067 int i, rc, mpflag = 0, mem_tight = 0, page_count = 0;
2070 LASSERT(!cfs_list_empty(ext_list));
2072 /* add pages into rpc_list to build BRW rpc */
2073 cfs_list_for_each_entry(ext, ext_list, oe_link) {
2074 LASSERT(ext->oe_state == OES_RPC);
2075 mem_tight |= ext->oe_memalloc;
2076 cfs_list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
2078 cfs_list_add_tail(&oap->oap_rpc_item, &rpc_list);
2079 if (starting_offset > oap->oap_obj_off)
2080 starting_offset = oap->oap_obj_off;
2082 LASSERT(oap->oap_page_off == 0);
2083 if (ending_offset < oap->oap_obj_off + oap->oap_count)
2084 ending_offset = oap->oap_obj_off +
2087 LASSERT(oap->oap_page_off + oap->oap_count ==
2093 mpflag = cfs_memory_pressure_get_and_set();
2095 memset(&crattr, 0, sizeof crattr);
2096 OBD_ALLOC(pga, sizeof(*pga) * page_count);
2098 GOTO(out, rc = -ENOMEM);
2102 GOTO(out, rc = -ENOMEM);
2105 cfs_list_for_each_entry(oap, &rpc_list, oap_rpc_item) {
2106 struct cl_page *page = oap2cl_page(oap);
2107 if (clerq == NULL) {
2108 clerq = cl_req_alloc(env, page, crt,
2109 1 /* only 1-object rpcs for
2112 GOTO(out, rc = PTR_ERR(clerq));
2113 lock = oap->oap_ldlm_lock;
2116 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
2117 pga[i] = &oap->oap_brw_page;
2118 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2119 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
2120 pga[i]->pg, cfs_page_index(oap->oap_page), oap, pga[i]->flag);
2122 cl_req_page_add(env, clerq, page);
2125 /* always get the data for the obdo for the rpc */
2126 LASSERT(clerq != NULL);
2128 crattr.cra_capa = NULL;
2129 memset(crattr.cra_jobid, 0, JOBSTATS_JOBID_SIZE);
2130 cl_req_attr_set(env, clerq, &crattr, ~0ULL);
2132 oa->o_handle = lock->l_remote_handle;
2133 oa->o_valid |= OBD_MD_FLHANDLE;
2136 rc = cl_req_prep(env, clerq);
2138 CERROR("cl_req_prep failed: %d\n", rc);
2142 sort_brw_pages(pga, page_count);
2143 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
2144 pga, &req, crattr.cra_capa, 1, 0);
2146 CERROR("prep_req failed: %d\n", rc);
2150 req->rq_interpret_reply = brw_interpret;
2152 req->rq_memalloc = 1;
2154 /* Need to update the timestamps after the request is built in case
2155 * we race with setattr (locally or in queue at OST). If OST gets
2156 * later setattr before earlier BRW (as determined by the request xid),
2157 * the OST will not use BRW timestamps. Sadly, there is no obvious
2158 * way to do this in a single call. bug 10150 */
2159 cl_req_attr_set(env, clerq, &crattr,
2160 OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME);
2162 lustre_msg_set_jobid(req->rq_reqmsg, crattr.cra_jobid);
2164 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2165 aa = ptlrpc_req_async_args(req);
2166 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
2167 cfs_list_splice_init(&rpc_list, &aa->aa_oaps);
2168 CFS_INIT_LIST_HEAD(&aa->aa_exts);
2169 cfs_list_splice_init(ext_list, &aa->aa_exts);
2170 aa->aa_clerq = clerq;
2172 /* queued sync pages can be torn down while the pages
2173 * were between the pending list and the rpc */
2175 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2176 /* only one oap gets a request reference */
2179 if (oap->oap_interrupted && !req->rq_intr) {
2180 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2182 ptlrpc_mark_interrupted(req);
2186 tmp->oap_request = ptlrpc_request_addref(req);
2188 client_obd_list_lock(&cli->cl_loi_list_lock);
2189 starting_offset >>= CFS_PAGE_SHIFT;
2190 if (cmd == OBD_BRW_READ) {
2191 cli->cl_r_in_flight++;
2192 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2193 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2194 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2195 starting_offset + 1);
2197 cli->cl_w_in_flight++;
2198 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2199 lprocfs_oh_tally(&cli->cl_write_rpc_hist, cli->cl_w_in_flight);
2200 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2201 starting_offset + 1);
2203 client_obd_list_unlock(&cli->cl_loi_list_lock);
2205 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2206 page_count, aa, cli->cl_r_in_flight,
2207 cli->cl_w_in_flight);
2209 /* XXX: Maybe the caller can check the RPC bulk descriptor to
2210 * see which CPU/NUMA node the majority of pages were allocated
2211 * on, and try to assign the async RPC to the CPU core
2212 * (PDL_POLICY_PREFERRED) to reduce cross-CPU memory traffic.
2214 * But on the other hand, we expect that multiple ptlrpcd
2215 * threads and the initial write sponsor can run in parallel,
2216 * especially when data checksum is enabled, which is CPU-bound
2217 * operation and single ptlrpcd thread cannot process in time.
2218 * So more ptlrpcd threads sharing BRW load
2219 * (with PDL_POLICY_ROUND) seems better.
2221 ptlrpcd_add_req(req, pol, -1);
2227 cfs_memory_pressure_restore(mpflag);
2229 capa_put(crattr.cra_capa);
2231 LASSERT(req == NULL);
2236 OBD_FREE(pga, sizeof(*pga) * page_count);
2237 /* this should happen rarely and is pretty bad, it makes the
2238 * pending list not follow the dirty order */
2239 while (!cfs_list_empty(ext_list)) {
2240 ext = cfs_list_entry(ext_list->next, struct osc_extent,
2242 cfs_list_del_init(&ext->oe_link);
2243 osc_extent_finish(env, ext, 0, rc);
2245 if (clerq && !IS_ERR(clerq))
2246 cl_req_completion(env, clerq, rc);
2251 static int osc_set_lock_data_with_check(struct ldlm_lock *lock,
2252 struct ldlm_enqueue_info *einfo)
2254 void *data = einfo->ei_cbdata;
2257 LASSERT(lock != NULL);
2258 LASSERT(lock->l_blocking_ast == einfo->ei_cb_bl);
2259 LASSERT(lock->l_resource->lr_type == einfo->ei_type);
2260 LASSERT(lock->l_completion_ast == einfo->ei_cb_cp);
2261 LASSERT(lock->l_glimpse_ast == einfo->ei_cb_gl);
2263 lock_res_and_lock(lock);
2264 spin_lock(&osc_ast_guard);
2266 if (lock->l_ast_data == NULL)
2267 lock->l_ast_data = data;
2268 if (lock->l_ast_data == data)
2271 spin_unlock(&osc_ast_guard);
2272 unlock_res_and_lock(lock);
2277 static int osc_set_data_with_check(struct lustre_handle *lockh,
2278 struct ldlm_enqueue_info *einfo)
2280 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2284 set = osc_set_lock_data_with_check(lock, einfo);
2285 LDLM_LOCK_PUT(lock);
2287 CERROR("lockh %p, data %p - client evicted?\n",
2288 lockh, einfo->ei_cbdata);
2292 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2293 ldlm_iterator_t replace, void *data)
2295 struct ldlm_res_id res_id;
2296 struct obd_device *obd = class_exp2obd(exp);
2298 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
2299 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2303 /* find any ldlm lock of the inode in osc
2307 static int osc_find_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2308 ldlm_iterator_t replace, void *data)
2310 struct ldlm_res_id res_id;
2311 struct obd_device *obd = class_exp2obd(exp);
2314 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
2315 rc = ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2316 if (rc == LDLM_ITER_STOP)
2318 if (rc == LDLM_ITER_CONTINUE)
2323 static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
2324 obd_enqueue_update_f upcall, void *cookie,
2325 __u64 *flags, int agl, int rc)
2327 int intent = *flags & LDLM_FL_HAS_INTENT;
2331 /* The request was created before ldlm_cli_enqueue call. */
2332 if (rc == ELDLM_LOCK_ABORTED) {
2333 struct ldlm_reply *rep;
2334 rep = req_capsule_server_get(&req->rq_pill,
2337 LASSERT(rep != NULL);
2338 if (rep->lock_policy_res1)
2339 rc = rep->lock_policy_res1;
2343 if ((intent != 0 && rc == ELDLM_LOCK_ABORTED && agl == 0) ||
2345 *flags |= LDLM_FL_LVB_READY;
2346 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
2347 lvb->lvb_size, lvb->lvb_blocks, lvb->lvb_mtime);
2350 /* Call the update callback. */
2351 rc = (*upcall)(cookie, rc);
2355 static int osc_enqueue_interpret(const struct lu_env *env,
2356 struct ptlrpc_request *req,
2357 struct osc_enqueue_args *aa, int rc)
2359 struct ldlm_lock *lock;
2360 struct lustre_handle handle;
2362 struct ost_lvb *lvb;
2364 __u64 *flags = aa->oa_flags;
2366 /* Make a local copy of a lock handle and a mode, because aa->oa_*
2367 * might be freed anytime after lock upcall has been called. */
2368 lustre_handle_copy(&handle, aa->oa_lockh);
2369 mode = aa->oa_ei->ei_mode;
2371 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2373 lock = ldlm_handle2lock(&handle);
2375 /* Take an additional reference so that a blocking AST that
2376 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2377 * to arrive after an upcall has been executed by
2378 * osc_enqueue_fini(). */
2379 ldlm_lock_addref(&handle, mode);
2381 /* Let CP AST to grant the lock first. */
2382 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
2384 if (aa->oa_agl && rc == ELDLM_LOCK_ABORTED) {
2389 lvb_len = sizeof(*aa->oa_lvb);
2392 /* Complete obtaining the lock procedure. */
2393 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
2394 mode, flags, lvb, lvb_len, &handle, rc);
2395 /* Complete osc stuff. */
2396 rc = osc_enqueue_fini(req, aa->oa_lvb, aa->oa_upcall, aa->oa_cookie,
2397 flags, aa->oa_agl, rc);
2399 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
2401 /* Release the lock for async request. */
2402 if (lustre_handle_is_used(&handle) && rc == ELDLM_OK)
2404 * Releases a reference taken by ldlm_cli_enqueue(), if it is
2405 * not already released by
2406 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
2408 ldlm_lock_decref(&handle, mode);
2410 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
2411 aa->oa_lockh, req, aa);
2412 ldlm_lock_decref(&handle, mode);
2413 LDLM_LOCK_PUT(lock);
2417 void osc_update_enqueue(struct lustre_handle *lov_lockhp,
2418 struct lov_oinfo *loi, int flags,
2419 struct ost_lvb *lvb, __u32 mode, int rc)
2421 struct ldlm_lock *lock = ldlm_handle2lock(lov_lockhp);
2423 if (rc == ELDLM_OK) {
2426 LASSERT(lock != NULL);
2427 loi->loi_lvb = *lvb;
2428 tmp = loi->loi_lvb.lvb_size;
2429 /* Extend KMS up to the end of this lock and no further
2430 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
2431 if (tmp > lock->l_policy_data.l_extent.end)
2432 tmp = lock->l_policy_data.l_extent.end + 1;
2433 if (tmp >= loi->loi_kms) {
2434 LDLM_DEBUG(lock, "lock acquired, setting rss="LPU64
2435 ", kms="LPU64, loi->loi_lvb.lvb_size, tmp);
2436 loi_kms_set(loi, tmp);
2438 LDLM_DEBUG(lock, "lock acquired, setting rss="
2439 LPU64"; leaving kms="LPU64", end="LPU64,
2440 loi->loi_lvb.lvb_size, loi->loi_kms,
2441 lock->l_policy_data.l_extent.end);
2443 ldlm_lock_allow_match(lock);
2444 } else if (rc == ELDLM_LOCK_ABORTED && (flags & LDLM_FL_HAS_INTENT)) {
2445 LASSERT(lock != NULL);
2446 loi->loi_lvb = *lvb;
2447 ldlm_lock_allow_match(lock);
2448 CDEBUG(D_INODE, "glimpsed, setting rss="LPU64"; leaving"
2449 " kms="LPU64"\n", loi->loi_lvb.lvb_size, loi->loi_kms);
2455 ldlm_lock_fail_match(lock);
2457 LDLM_LOCK_PUT(lock);
2460 EXPORT_SYMBOL(osc_update_enqueue);
2462 struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
2464 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2465 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2466 * other synchronous requests, however keeping some locks and trying to obtain
2467 * others may take a considerable amount of time in a case of ost failure; and
2468 * when other sync requests do not get released lock from a client, the client
2469 * is excluded from the cluster -- such scenarious make the life difficult, so
2470 * release locks just after they are obtained. */
2471 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2472 __u64 *flags, ldlm_policy_data_t *policy,
2473 struct ost_lvb *lvb, int kms_valid,
2474 obd_enqueue_update_f upcall, void *cookie,
2475 struct ldlm_enqueue_info *einfo,
2476 struct lustre_handle *lockh,
2477 struct ptlrpc_request_set *rqset, int async, int agl)
2479 struct obd_device *obd = exp->exp_obd;
2480 struct ptlrpc_request *req = NULL;
2481 int intent = *flags & LDLM_FL_HAS_INTENT;
2482 int match_lvb = (agl != 0 ? 0 : LDLM_FL_LVB_READY);
2487 /* Filesystem lock extents are extended to page boundaries so that
2488 * dealing with the page cache is a little smoother. */
2489 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2490 policy->l_extent.end |= ~CFS_PAGE_MASK;
2493 * kms is not valid when either object is completely fresh (so that no
2494 * locks are cached), or object was evicted. In the latter case cached
2495 * lock cannot be used, because it would prime inode state with
2496 * potentially stale LVB.
2501 /* Next, search for already existing extent locks that will cover us */
2502 /* If we're trying to read, we also search for an existing PW lock. The
2503 * VFS and page cache already protect us locally, so lots of readers/
2504 * writers can share a single PW lock.
2506 * There are problems with conversion deadlocks, so instead of
2507 * converting a read lock to a write lock, we'll just enqueue a new
2510 * At some point we should cancel the read lock instead of making them
2511 * send us a blocking callback, but there are problems with canceling
2512 * locks out from other users right now, too. */
2513 mode = einfo->ei_mode;
2514 if (einfo->ei_mode == LCK_PR)
2516 mode = ldlm_lock_match(obd->obd_namespace, *flags | match_lvb, res_id,
2517 einfo->ei_type, policy, mode, lockh, 0);
2519 struct ldlm_lock *matched = ldlm_handle2lock(lockh);
2521 if ((agl != 0) && !(matched->l_flags & LDLM_FL_LVB_READY)) {
2522 /* For AGL, if enqueue RPC is sent but the lock is not
2523 * granted, then skip to process this strpe.
2524 * Return -ECANCELED to tell the caller. */
2525 ldlm_lock_decref(lockh, mode);
2526 LDLM_LOCK_PUT(matched);
2528 } else if (osc_set_lock_data_with_check(matched, einfo)) {
2529 *flags |= LDLM_FL_LVB_READY;
2530 /* addref the lock only if not async requests and PW
2531 * lock is matched whereas we asked for PR. */
2532 if (!rqset && einfo->ei_mode != mode)
2533 ldlm_lock_addref(lockh, LCK_PR);
2535 /* I would like to be able to ASSERT here that
2536 * rss <= kms, but I can't, for reasons which
2537 * are explained in lov_enqueue() */
2540 /* We already have a lock, and it's referenced.
2542 * At this point, the cl_lock::cll_state is CLS_QUEUING,
2543 * AGL upcall may change it to CLS_HELD directly. */
2544 (*upcall)(cookie, ELDLM_OK);
2546 if (einfo->ei_mode != mode)
2547 ldlm_lock_decref(lockh, LCK_PW);
2549 /* For async requests, decref the lock. */
2550 ldlm_lock_decref(lockh, einfo->ei_mode);
2551 LDLM_LOCK_PUT(matched);
2554 ldlm_lock_decref(lockh, mode);
2555 LDLM_LOCK_PUT(matched);
2561 CFS_LIST_HEAD(cancels);
2562 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2563 &RQF_LDLM_ENQUEUE_LVB);
2567 rc = ldlm_prep_enqueue_req(exp, req, &cancels, 0);
2569 ptlrpc_request_free(req);
2573 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
2575 ptlrpc_request_set_replen(req);
2578 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2579 *flags &= ~LDLM_FL_BLOCK_GRANTED;
2581 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
2582 sizeof(*lvb), LVB_T_OST, lockh, async);
2585 struct osc_enqueue_args *aa;
2586 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2587 aa = ptlrpc_req_async_args(req);
2590 aa->oa_flags = flags;
2591 aa->oa_upcall = upcall;
2592 aa->oa_cookie = cookie;
2594 aa->oa_lockh = lockh;
2597 req->rq_interpret_reply =
2598 (ptlrpc_interpterer_t)osc_enqueue_interpret;
2599 if (rqset == PTLRPCD_SET)
2600 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
2602 ptlrpc_set_add_req(rqset, req);
2603 } else if (intent) {
2604 ptlrpc_req_finished(req);
2609 rc = osc_enqueue_fini(req, lvb, upcall, cookie, flags, agl, rc);
2611 ptlrpc_req_finished(req);
2616 static int osc_enqueue(struct obd_export *exp, struct obd_info *oinfo,
2617 struct ldlm_enqueue_info *einfo,
2618 struct ptlrpc_request_set *rqset)
2620 struct ldlm_res_id res_id;
2624 osc_build_res_name(oinfo->oi_md->lsm_object_id,
2625 oinfo->oi_md->lsm_object_seq, &res_id);
2627 rc = osc_enqueue_base(exp, &res_id, &oinfo->oi_flags, &oinfo->oi_policy,
2628 &oinfo->oi_md->lsm_oinfo[0]->loi_lvb,
2629 oinfo->oi_md->lsm_oinfo[0]->loi_kms_valid,
2630 oinfo->oi_cb_up, oinfo, einfo, oinfo->oi_lockh,
2631 rqset, rqset != NULL, 0);
2635 int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2636 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
2637 int *flags, void *data, struct lustre_handle *lockh,
2640 struct obd_device *obd = exp->exp_obd;
2641 int lflags = *flags;
2645 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
2648 /* Filesystem lock extents are extended to page boundaries so that
2649 * dealing with the page cache is a little smoother */
2650 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2651 policy->l_extent.end |= ~CFS_PAGE_MASK;
2653 /* Next, search for already existing extent locks that will cover us */
2654 /* If we're trying to read, we also search for an existing PW lock. The
2655 * VFS and page cache already protect us locally, so lots of readers/
2656 * writers can share a single PW lock. */
2660 rc = ldlm_lock_match(obd->obd_namespace, lflags,
2661 res_id, type, policy, rc, lockh, unref);
2664 if (!osc_set_data_with_check(lockh, data)) {
2665 if (!(lflags & LDLM_FL_TEST_LOCK))
2666 ldlm_lock_decref(lockh, rc);
2670 if (!(lflags & LDLM_FL_TEST_LOCK) && mode != rc) {
2671 ldlm_lock_addref(lockh, LCK_PR);
2672 ldlm_lock_decref(lockh, LCK_PW);
2679 int osc_cancel_base(struct lustre_handle *lockh, __u32 mode)
2683 if (unlikely(mode == LCK_GROUP))
2684 ldlm_lock_decref_and_cancel(lockh, mode);
2686 ldlm_lock_decref(lockh, mode);
2691 static int osc_cancel(struct obd_export *exp, struct lov_stripe_md *md,
2692 __u32 mode, struct lustre_handle *lockh)
2695 RETURN(osc_cancel_base(lockh, mode));
2698 static int osc_cancel_unused(struct obd_export *exp,
2699 struct lov_stripe_md *lsm,
2700 ldlm_cancel_flags_t flags,
2703 struct obd_device *obd = class_exp2obd(exp);
2704 struct ldlm_res_id res_id, *resp = NULL;
2707 resp = osc_build_res_name(lsm->lsm_object_id,
2708 lsm->lsm_object_seq, &res_id);
2711 return ldlm_cli_cancel_unused(obd->obd_namespace, resp, flags, opaque);
2714 static int osc_statfs_interpret(const struct lu_env *env,
2715 struct ptlrpc_request *req,
2716 struct osc_async_args *aa, int rc)
2718 struct obd_statfs *msfs;
2722 /* The request has in fact never been sent
2723 * due to issues at a higher level (LOV).
2724 * Exit immediately since the caller is
2725 * aware of the problem and takes care
2726 * of the clean up */
2729 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
2730 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
2736 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2738 GOTO(out, rc = -EPROTO);
2741 *aa->aa_oi->oi_osfs = *msfs;
2743 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
2747 static int osc_statfs_async(struct obd_export *exp,
2748 struct obd_info *oinfo, __u64 max_age,
2749 struct ptlrpc_request_set *rqset)
2751 struct obd_device *obd = class_exp2obd(exp);
2752 struct ptlrpc_request *req;
2753 struct osc_async_args *aa;
2757 /* We could possibly pass max_age in the request (as an absolute
2758 * timestamp or a "seconds.usec ago") so the target can avoid doing
2759 * extra calls into the filesystem if that isn't necessary (e.g.
2760 * during mount that would help a bit). Having relative timestamps
2761 * is not so great if request processing is slow, while absolute
2762 * timestamps are not ideal because they need time synchronization. */
2763 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
2767 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2769 ptlrpc_request_free(req);
2772 ptlrpc_request_set_replen(req);
2773 req->rq_request_portal = OST_CREATE_PORTAL;
2774 ptlrpc_at_set_req_timeout(req);
2776 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
2777 /* procfs requests not want stat in wait for avoid deadlock */
2778 req->rq_no_resend = 1;
2779 req->rq_no_delay = 1;
2782 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
2783 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2784 aa = ptlrpc_req_async_args(req);
2787 ptlrpc_set_add_req(rqset, req);
2791 static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
2792 struct obd_statfs *osfs, __u64 max_age, __u32 flags)
2794 struct obd_device *obd = class_exp2obd(exp);
2795 struct obd_statfs *msfs;
2796 struct ptlrpc_request *req;
2797 struct obd_import *imp = NULL;
2801 /*Since the request might also come from lprocfs, so we need
2802 *sync this with client_disconnect_export Bug15684*/
2803 down_read(&obd->u.cli.cl_sem);
2804 if (obd->u.cli.cl_import)
2805 imp = class_import_get(obd->u.cli.cl_import);
2806 up_read(&obd->u.cli.cl_sem);
2810 /* We could possibly pass max_age in the request (as an absolute
2811 * timestamp or a "seconds.usec ago") so the target can avoid doing
2812 * extra calls into the filesystem if that isn't necessary (e.g.
2813 * during mount that would help a bit). Having relative timestamps
2814 * is not so great if request processing is slow, while absolute
2815 * timestamps are not ideal because they need time synchronization. */
2816 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
2818 class_import_put(imp);
2823 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2825 ptlrpc_request_free(req);
2828 ptlrpc_request_set_replen(req);
2829 req->rq_request_portal = OST_CREATE_PORTAL;
2830 ptlrpc_at_set_req_timeout(req);
2832 if (flags & OBD_STATFS_NODELAY) {
2833 /* procfs requests not want stat in wait for avoid deadlock */
2834 req->rq_no_resend = 1;
2835 req->rq_no_delay = 1;
2838 rc = ptlrpc_queue_wait(req);
2842 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2844 GOTO(out, rc = -EPROTO);
2851 ptlrpc_req_finished(req);
2855 /* Retrieve object striping information.
2857 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
2858 * the maximum number of OST indices which will fit in the user buffer.
2859 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
2861 static int osc_getstripe(struct lov_stripe_md *lsm, struct lov_user_md *lump)
2863 /* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
2864 struct lov_user_md_v3 lum, *lumk;
2865 struct lov_user_ost_data_v1 *lmm_objects;
2866 int rc = 0, lum_size;
2872 /* we only need the header part from user space to get lmm_magic and
2873 * lmm_stripe_count, (the header part is common to v1 and v3) */
2874 lum_size = sizeof(struct lov_user_md_v1);
2875 if (cfs_copy_from_user(&lum, lump, lum_size))
2878 if ((lum.lmm_magic != LOV_USER_MAGIC_V1) &&
2879 (lum.lmm_magic != LOV_USER_MAGIC_V3))
2882 /* lov_user_md_vX and lov_mds_md_vX must have the same size */
2883 LASSERT(sizeof(struct lov_user_md_v1) == sizeof(struct lov_mds_md_v1));
2884 LASSERT(sizeof(struct lov_user_md_v3) == sizeof(struct lov_mds_md_v3));
2885 LASSERT(sizeof(lum.lmm_objects[0]) == sizeof(lumk->lmm_objects[0]));
2887 /* we can use lov_mds_md_size() to compute lum_size
2888 * because lov_user_md_vX and lov_mds_md_vX have the same size */
2889 if (lum.lmm_stripe_count > 0) {
2890 lum_size = lov_mds_md_size(lum.lmm_stripe_count, lum.lmm_magic);
2891 OBD_ALLOC(lumk, lum_size);
2895 if (lum.lmm_magic == LOV_USER_MAGIC_V1)
2896 lmm_objects = &(((struct lov_user_md_v1 *)lumk)->lmm_objects[0]);
2898 lmm_objects = &(lumk->lmm_objects[0]);
2899 lmm_objects->l_object_id = lsm->lsm_object_id;
2901 lum_size = lov_mds_md_size(0, lum.lmm_magic);
2905 lumk->lmm_object_id = lsm->lsm_object_id;
2906 lumk->lmm_object_seq = lsm->lsm_object_seq;
2907 lumk->lmm_stripe_count = 1;
2909 if (cfs_copy_to_user(lump, lumk, lum_size))
2913 OBD_FREE(lumk, lum_size);
2919 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
2920 void *karg, void *uarg)
2922 struct obd_device *obd = exp->exp_obd;
2923 struct obd_ioctl_data *data = karg;
2927 if (!cfs_try_module_get(THIS_MODULE)) {
2928 CERROR("Can't get module. Is it alive?");
2932 case OBD_IOC_LOV_GET_CONFIG: {
2934 struct lov_desc *desc;
2935 struct obd_uuid uuid;
2939 if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
2940 GOTO(out, err = -EINVAL);
2942 data = (struct obd_ioctl_data *)buf;
2944 if (sizeof(*desc) > data->ioc_inllen1) {
2945 obd_ioctl_freedata(buf, len);
2946 GOTO(out, err = -EINVAL);
2949 if (data->ioc_inllen2 < sizeof(uuid)) {
2950 obd_ioctl_freedata(buf, len);
2951 GOTO(out, err = -EINVAL);
2954 desc = (struct lov_desc *)data->ioc_inlbuf1;
2955 desc->ld_tgt_count = 1;
2956 desc->ld_active_tgt_count = 1;
2957 desc->ld_default_stripe_count = 1;
2958 desc->ld_default_stripe_size = 0;
2959 desc->ld_default_stripe_offset = 0;
2960 desc->ld_pattern = 0;
2961 memcpy(&desc->ld_uuid, &obd->obd_uuid, sizeof(uuid));
2963 memcpy(data->ioc_inlbuf2, &obd->obd_uuid, sizeof(uuid));
2965 err = cfs_copy_to_user((void *)uarg, buf, len);
2968 obd_ioctl_freedata(buf, len);
2971 case LL_IOC_LOV_SETSTRIPE:
2972 err = obd_alloc_memmd(exp, karg);
2976 case LL_IOC_LOV_GETSTRIPE:
2977 err = osc_getstripe(karg, uarg);
2979 case OBD_IOC_CLIENT_RECOVER:
2980 err = ptlrpc_recover_import(obd->u.cli.cl_import,
2981 data->ioc_inlbuf1, 0);
2985 case IOC_OSC_SET_ACTIVE:
2986 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
2989 case OBD_IOC_POLL_QUOTACHECK:
2990 err = osc_quota_poll_check(exp, (struct if_quotacheck *)karg);
2992 case OBD_IOC_PING_TARGET:
2993 err = ptlrpc_obd_ping(obd);
2996 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
2997 cmd, cfs_curproc_comm());
2998 GOTO(out, err = -ENOTTY);
3001 cfs_module_put(THIS_MODULE);
3005 static int osc_get_info(const struct lu_env *env, struct obd_export *exp,
3006 obd_count keylen, void *key, __u32 *vallen, void *val,
3007 struct lov_stripe_md *lsm)
3010 if (!vallen || !val)
3013 if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
3014 __u32 *stripe = val;
3015 *vallen = sizeof(*stripe);
3018 } else if (KEY_IS(KEY_LAST_ID)) {
3019 struct ptlrpc_request *req;
3024 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3025 &RQF_OST_GET_INFO_LAST_ID);
3029 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3030 RCL_CLIENT, keylen);
3031 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3033 ptlrpc_request_free(req);
3037 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3038 memcpy(tmp, key, keylen);
3040 req->rq_no_delay = req->rq_no_resend = 1;
3041 ptlrpc_request_set_replen(req);
3042 rc = ptlrpc_queue_wait(req);
3046 reply = req_capsule_server_get(&req->rq_pill, &RMF_OBD_ID);
3048 GOTO(out, rc = -EPROTO);
3050 *((obd_id *)val) = *reply;
3052 ptlrpc_req_finished(req);
3054 } else if (KEY_IS(KEY_FIEMAP)) {
3055 struct ptlrpc_request *req;
3056 struct ll_user_fiemap *reply;
3060 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3061 &RQF_OST_GET_INFO_FIEMAP);
3065 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_KEY,
3066 RCL_CLIENT, keylen);
3067 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3068 RCL_CLIENT, *vallen);
3069 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3070 RCL_SERVER, *vallen);
3072 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3074 ptlrpc_request_free(req);
3078 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_KEY);
3079 memcpy(tmp, key, keylen);
3080 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3081 memcpy(tmp, val, *vallen);
3083 ptlrpc_request_set_replen(req);
3084 rc = ptlrpc_queue_wait(req);
3088 reply = req_capsule_server_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3090 GOTO(out1, rc = -EPROTO);
3092 memcpy(val, reply, *vallen);
3094 ptlrpc_req_finished(req);
3102 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
3103 obd_count keylen, void *key, obd_count vallen,
3104 void *val, struct ptlrpc_request_set *set)
3106 struct ptlrpc_request *req;
3107 struct obd_device *obd = exp->exp_obd;
3108 struct obd_import *imp = class_exp2cliimp(exp);
3113 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3115 if (KEY_IS(KEY_CHECKSUM)) {
3116 if (vallen != sizeof(int))
3118 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3122 if (KEY_IS(KEY_SPTLRPC_CONF)) {
3123 sptlrpc_conf_client_adapt(obd);
3127 if (KEY_IS(KEY_FLUSH_CTX)) {
3128 sptlrpc_import_flush_my_ctx(imp);
3132 if (KEY_IS(KEY_CACHE_SET)) {
3133 struct client_obd *cli = &obd->u.cli;
3135 LASSERT(cli->cl_cache == NULL); /* only once */
3136 cli->cl_cache = (struct cl_client_cache *)val;
3137 cfs_atomic_inc(&cli->cl_cache->ccc_users);
3138 cli->cl_lru_left = &cli->cl_cache->ccc_lru_left;
3140 /* add this osc into entity list */
3141 LASSERT(cfs_list_empty(&cli->cl_lru_osc));
3142 spin_lock(&cli->cl_cache->ccc_lru_lock);
3143 cfs_list_add(&cli->cl_lru_osc, &cli->cl_cache->ccc_lru);
3144 spin_unlock(&cli->cl_cache->ccc_lru_lock);
3149 if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
3150 struct client_obd *cli = &obd->u.cli;
3151 int nr = cfs_atomic_read(&cli->cl_lru_in_list) >> 1;
3152 int target = *(int *)val;
3154 nr = osc_lru_shrink(cli, min(nr, target));
3159 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
3162 /* We pass all other commands directly to OST. Since nobody calls osc
3163 methods directly and everybody is supposed to go through LOV, we
3164 assume lov checked invalid values for us.
3165 The only recognised values so far are evict_by_nid and mds_conn.
3166 Even if something bad goes through, we'd get a -EINVAL from OST
3169 req = ptlrpc_request_alloc(imp, KEY_IS(KEY_GRANT_SHRINK) ?
3170 &RQF_OST_SET_GRANT_INFO :
3175 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3176 RCL_CLIENT, keylen);
3177 if (!KEY_IS(KEY_GRANT_SHRINK))
3178 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
3179 RCL_CLIENT, vallen);
3180 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
3182 ptlrpc_request_free(req);
3186 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3187 memcpy(tmp, key, keylen);
3188 tmp = req_capsule_client_get(&req->rq_pill, KEY_IS(KEY_GRANT_SHRINK) ?
3191 memcpy(tmp, val, vallen);
3193 if (KEY_IS(KEY_GRANT_SHRINK)) {
3194 struct osc_grant_args *aa;
3197 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
3198 aa = ptlrpc_req_async_args(req);
3201 ptlrpc_req_finished(req);
3204 *oa = ((struct ost_body *)val)->oa;
3206 req->rq_interpret_reply = osc_shrink_grant_interpret;
3209 ptlrpc_request_set_replen(req);
3210 if (!KEY_IS(KEY_GRANT_SHRINK)) {
3211 LASSERT(set != NULL);
3212 ptlrpc_set_add_req(set, req);
3213 ptlrpc_check_set(NULL, set);
3215 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
3221 static int osc_llog_init(struct obd_device *obd, struct obd_llog_group *olg,
3222 struct obd_device *disk_obd, int *index)
3224 /* this code is not supposed to be used with LOD/OSP
3225 * to be removed soon */
3230 static int osc_llog_finish(struct obd_device *obd, int count)
3232 struct llog_ctxt *ctxt;
3236 ctxt = llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
3238 llog_cat_close(NULL, ctxt->loc_handle);
3239 llog_cleanup(NULL, ctxt);
3242 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3244 llog_cleanup(NULL, ctxt);
3248 static int osc_reconnect(const struct lu_env *env,
3249 struct obd_export *exp, struct obd_device *obd,
3250 struct obd_uuid *cluuid,
3251 struct obd_connect_data *data,
3254 struct client_obd *cli = &obd->u.cli;
3256 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3259 client_obd_list_lock(&cli->cl_loi_list_lock);
3260 data->ocd_grant = (cli->cl_avail_grant + cli->cl_dirty) ?:
3261 2 * cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT;
3262 lost_grant = cli->cl_lost_grant;
3263 cli->cl_lost_grant = 0;
3264 client_obd_list_unlock(&cli->cl_loi_list_lock);
3266 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
3267 " ocd_grant: %d, lost: %ld.\n", data->ocd_connect_flags,
3268 data->ocd_version, data->ocd_grant, lost_grant);
3274 static int osc_disconnect(struct obd_export *exp)
3276 struct obd_device *obd = class_exp2obd(exp);
3277 struct llog_ctxt *ctxt;
3280 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3282 if (obd->u.cli.cl_conn_count == 1) {
3283 /* Flush any remaining cancel messages out to the
3285 llog_sync(ctxt, exp, 0);
3287 llog_ctxt_put(ctxt);
3289 CDEBUG(D_HA, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
3293 rc = client_disconnect_export(exp);
3295 * Initially we put del_shrink_grant before disconnect_export, but it
3296 * causes the following problem if setup (connect) and cleanup
3297 * (disconnect) are tangled together.
3298 * connect p1 disconnect p2
3299 * ptlrpc_connect_import
3300 * ............... class_manual_cleanup
3303 * ptlrpc_connect_interrupt
3305 * add this client to shrink list
3307 * Bang! pinger trigger the shrink.
3308 * So the osc should be disconnected from the shrink list, after we
3309 * are sure the import has been destroyed. BUG18662
3311 if (obd->u.cli.cl_import == NULL)
3312 osc_del_shrink_grant(&obd->u.cli);
3316 static int osc_import_event(struct obd_device *obd,
3317 struct obd_import *imp,
3318 enum obd_import_event event)
3320 struct client_obd *cli;
3324 LASSERT(imp->imp_obd == obd);
3327 case IMP_EVENT_DISCON: {
3329 client_obd_list_lock(&cli->cl_loi_list_lock);
3330 cli->cl_avail_grant = 0;
3331 cli->cl_lost_grant = 0;
3332 client_obd_list_unlock(&cli->cl_loi_list_lock);
3335 case IMP_EVENT_INACTIVE: {
3336 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
3339 case IMP_EVENT_INVALIDATE: {
3340 struct ldlm_namespace *ns = obd->obd_namespace;
3344 env = cl_env_get(&refcheck);
3348 /* all pages go to failing rpcs due to the invalid
3350 osc_io_unplug(env, cli, NULL, PDL_POLICY_ROUND);
3352 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3353 cl_env_put(env, &refcheck);
3358 case IMP_EVENT_ACTIVE: {
3359 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
3362 case IMP_EVENT_OCD: {
3363 struct obd_connect_data *ocd = &imp->imp_connect_data;
3365 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3366 osc_init_grant(&obd->u.cli, ocd);
3369 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3370 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3372 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
3375 case IMP_EVENT_DEACTIVATE: {
3376 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_DEACTIVATE, NULL);
3379 case IMP_EVENT_ACTIVATE: {
3380 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVATE, NULL);
3384 CERROR("Unknown import event %d\n", event);
3391 * Determine whether the lock can be canceled before replaying the lock
3392 * during recovery, see bug16774 for detailed information.
3394 * \retval zero the lock can't be canceled
3395 * \retval other ok to cancel
3397 static int osc_cancel_for_recovery(struct ldlm_lock *lock)
3399 check_res_locked(lock->l_resource);
3402 * Cancel all unused extent lock in granted mode LCK_PR or LCK_CR.
3404 * XXX as a future improvement, we can also cancel unused write lock
3405 * if it doesn't have dirty data and active mmaps.
3407 if (lock->l_resource->lr_type == LDLM_EXTENT &&
3408 (lock->l_granted_mode == LCK_PR ||
3409 lock->l_granted_mode == LCK_CR) &&
3410 (osc_dlm_lock_pageref(lock) == 0))
3416 static int brw_queue_work(const struct lu_env *env, void *data)
3418 struct client_obd *cli = data;
3420 CDEBUG(D_CACHE, "Run writeback work for client obd %p.\n", cli);
3422 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
3426 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3428 struct lprocfs_static_vars lvars = { 0 };
3429 struct client_obd *cli = &obd->u.cli;
3434 rc = ptlrpcd_addref();
3438 rc = client_obd_setup(obd, lcfg);
3440 GOTO(out_ptlrpcd, rc);
3442 handler = ptlrpcd_alloc_work(cli->cl_import, brw_queue_work, cli);
3443 if (IS_ERR(handler))
3444 GOTO(out_client_setup, rc = PTR_ERR(handler));
3445 cli->cl_writeback_work = handler;
3447 rc = osc_quota_setup(obd);
3449 GOTO(out_ptlrpcd_work, rc);
3451 cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
3452 lprocfs_osc_init_vars(&lvars);
3453 if (lprocfs_obd_setup(obd, lvars.obd_vars) == 0) {
3454 lproc_osc_attach_seqstat(obd);
3455 sptlrpc_lprocfs_cliobd_attach(obd);
3456 ptlrpc_lprocfs_register_obd(obd);
3459 /* We need to allocate a few requests more, because
3460 * brw_interpret tries to create new requests before freeing
3461 * previous ones, Ideally we want to have 2x max_rpcs_in_flight
3462 * reserved, but I'm afraid that might be too much wasted RAM
3463 * in fact, so 2 is just my guess and still should work. */
3464 cli->cl_import->imp_rq_pool =
3465 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
3467 ptlrpc_add_rqs_to_pool);
3469 CFS_INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
3470 ns_register_cancel(obd->obd_namespace, osc_cancel_for_recovery);
3474 ptlrpcd_destroy_work(handler);
3476 client_obd_cleanup(obd);
3482 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
3488 case OBD_CLEANUP_EARLY: {
3489 struct obd_import *imp;
3490 imp = obd->u.cli.cl_import;
3491 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
3492 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
3493 ptlrpc_deactivate_import(imp);
3494 spin_lock(&imp->imp_lock);
3495 imp->imp_pingable = 0;
3496 spin_unlock(&imp->imp_lock);
3499 case OBD_CLEANUP_EXPORTS: {
3500 struct client_obd *cli = &obd->u.cli;
3502 * for echo client, export may be on zombie list, wait for
3503 * zombie thread to cull it, because cli.cl_import will be
3504 * cleared in client_disconnect_export():
3505 * class_export_destroy() -> obd_cleanup() ->
3506 * echo_device_free() -> echo_client_cleanup() ->
3507 * obd_disconnect() -> osc_disconnect() ->
3508 * client_disconnect_export()
3510 obd_zombie_barrier();
3511 if (cli->cl_writeback_work) {
3512 ptlrpcd_destroy_work(cli->cl_writeback_work);
3513 cli->cl_writeback_work = NULL;
3515 obd_cleanup_client_import(obd);
3516 ptlrpc_lprocfs_unregister_obd(obd);
3517 lprocfs_obd_cleanup(obd);
3518 rc = obd_llog_finish(obd, 0);
3520 CERROR("failed to cleanup llogging subsystems\n");
3527 int osc_cleanup(struct obd_device *obd)
3529 struct client_obd *cli = &obd->u.cli;
3535 if (cli->cl_cache != NULL) {
3536 LASSERT(cfs_atomic_read(&cli->cl_cache->ccc_users) > 0);
3537 spin_lock(&cli->cl_cache->ccc_lru_lock);
3538 cfs_list_del_init(&cli->cl_lru_osc);
3539 spin_unlock(&cli->cl_cache->ccc_lru_lock);
3540 cli->cl_lru_left = NULL;
3541 cfs_atomic_dec(&cli->cl_cache->ccc_users);
3542 cli->cl_cache = NULL;
3545 /* free memory of osc quota cache */
3546 osc_quota_cleanup(obd);
3548 rc = client_obd_cleanup(obd);
3554 int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
3556 struct lprocfs_static_vars lvars = { 0 };
3559 lprocfs_osc_init_vars(&lvars);
3561 switch (lcfg->lcfg_command) {
3563 rc = class_process_proc_param(PARAM_OSC, lvars.obd_vars,
3573 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
3575 return osc_process_config_base(obd, buf);
3578 struct obd_ops osc_obd_ops = {
3579 .o_owner = THIS_MODULE,
3580 .o_setup = osc_setup,
3581 .o_precleanup = osc_precleanup,
3582 .o_cleanup = osc_cleanup,
3583 .o_add_conn = client_import_add_conn,
3584 .o_del_conn = client_import_del_conn,
3585 .o_connect = client_connect_import,
3586 .o_reconnect = osc_reconnect,
3587 .o_disconnect = osc_disconnect,
3588 .o_statfs = osc_statfs,
3589 .o_statfs_async = osc_statfs_async,
3590 .o_packmd = osc_packmd,
3591 .o_unpackmd = osc_unpackmd,
3592 .o_create = osc_create,
3593 .o_destroy = osc_destroy,
3594 .o_getattr = osc_getattr,
3595 .o_getattr_async = osc_getattr_async,
3596 .o_setattr = osc_setattr,
3597 .o_setattr_async = osc_setattr_async,
3599 .o_punch = osc_punch,
3601 .o_enqueue = osc_enqueue,
3602 .o_change_cbdata = osc_change_cbdata,
3603 .o_find_cbdata = osc_find_cbdata,
3604 .o_cancel = osc_cancel,
3605 .o_cancel_unused = osc_cancel_unused,
3606 .o_iocontrol = osc_iocontrol,
3607 .o_get_info = osc_get_info,
3608 .o_set_info_async = osc_set_info_async,
3609 .o_import_event = osc_import_event,
3610 .o_llog_init = osc_llog_init,
3611 .o_llog_finish = osc_llog_finish,
3612 .o_process_config = osc_process_config,
3613 .o_quotactl = osc_quotactl,
3614 .o_quotacheck = osc_quotacheck,
3617 extern struct lu_kmem_descr osc_caches[];
3618 extern spinlock_t osc_ast_guard;
3619 extern struct lock_class_key osc_ast_guard_class;
3621 int __init osc_init(void)
3623 struct lprocfs_static_vars lvars = { 0 };
3627 /* print an address of _any_ initialized kernel symbol from this
3628 * module, to allow debugging with gdb that doesn't support data
3629 * symbols from modules.*/
3630 CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
3632 rc = lu_kmem_init(osc_caches);
3634 lprocfs_osc_init_vars(&lvars);
3636 rc = class_register_type(&osc_obd_ops, NULL, lvars.module_vars,
3637 LUSTRE_OSC_NAME, &osc_device_type);
3639 lu_kmem_fini(osc_caches);
3643 spin_lock_init(&osc_ast_guard);
3644 lockdep_set_class(&osc_ast_guard, &osc_ast_guard_class);
3650 static void /*__exit*/ osc_exit(void)
3652 class_unregister_type(LUSTRE_OSC_NAME);
3653 lu_kmem_fini(osc_caches);
3656 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
3657 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3658 MODULE_LICENSE("GPL");
3660 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);
git://git.whamcloud.com / fs / lustre-release.git / blob