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, 2013, 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>
54 #include <lustre_ha.h>
55 #include <lprocfs_status.h>
56 #include <lustre_ioctl.h>
57 #include <lustre_log.h>
58 #include <lustre_debug.h>
59 #include <lustre_param.h>
60 #include <lustre_fid.h>
61 #include "osc_internal.h"
62 #include "osc_cl_internal.h"
64 struct osc_brw_async_args {
68 obd_count aa_page_count;
70 struct brw_page **aa_ppga;
71 struct client_obd *aa_cli;
72 struct list_head aa_oaps;
73 struct list_head aa_exts;
74 struct obd_capa *aa_ocapa;
75 struct cl_req *aa_clerq;
78 #define osc_grant_args osc_brw_async_args
80 struct osc_async_args {
81 struct obd_info *aa_oi;
84 struct osc_setattr_args {
86 obd_enqueue_update_f sa_upcall;
90 struct osc_fsync_args {
91 struct obd_info *fa_oi;
92 obd_enqueue_update_f fa_upcall;
96 struct osc_enqueue_args {
97 struct obd_export *oa_exp;
99 obd_enqueue_update_f oa_upcall;
101 struct ost_lvb *oa_lvb;
102 struct lustre_handle *oa_lockh;
103 struct ldlm_enqueue_info *oa_ei;
104 unsigned int oa_agl:1;
107 static void osc_release_ppga(struct brw_page **ppga, obd_count count);
108 static int brw_interpret(const struct lu_env *env, struct ptlrpc_request *req,
111 /* Unpack OSC object metadata from disk storage (LE byte order). */
112 static int osc_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
113 struct lov_mds_md *lmm, int lmm_bytes)
116 struct obd_import *imp = class_exp2cliimp(exp);
120 if (lmm_bytes < sizeof(*lmm)) {
121 CERROR("%s: lov_mds_md too small: %d, need %d\n",
122 exp->exp_obd->obd_name, lmm_bytes,
126 /* XXX LOV_MAGIC etc check? */
128 if (unlikely(ostid_id(&lmm->lmm_oi) == 0)) {
129 CERROR("%s: zero lmm_object_id: rc = %d\n",
130 exp->exp_obd->obd_name, -EINVAL);
135 lsm_size = lov_stripe_md_size(1);
139 if (*lsmp != NULL && lmm == NULL) {
140 OBD_FREE((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
141 OBD_FREE(*lsmp, lsm_size);
147 OBD_ALLOC(*lsmp, lsm_size);
148 if (unlikely(*lsmp == NULL))
150 OBD_ALLOC((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
151 if (unlikely((*lsmp)->lsm_oinfo[0] == NULL)) {
152 OBD_FREE(*lsmp, lsm_size);
155 loi_init((*lsmp)->lsm_oinfo[0]);
156 } else if (unlikely(ostid_id(&(*lsmp)->lsm_oi) == 0)) {
161 /* XXX zero *lsmp? */
162 ostid_le_to_cpu(&lmm->lmm_oi, &(*lsmp)->lsm_oi);
165 (imp->imp_connect_data.ocd_connect_flags & OBD_CONNECT_MAXBYTES))
166 (*lsmp)->lsm_maxbytes = imp->imp_connect_data.ocd_maxbytes;
168 (*lsmp)->lsm_maxbytes = LUSTRE_STRIPE_MAXBYTES;
173 static inline void osc_pack_capa(struct ptlrpc_request *req,
174 struct ost_body *body, void *capa)
176 struct obd_capa *oc = (struct obd_capa *)capa;
177 struct lustre_capa *c;
182 c = req_capsule_client_get(&req->rq_pill, &RMF_CAPA1);
185 body->oa.o_valid |= OBD_MD_FLOSSCAPA;
186 DEBUG_CAPA(D_SEC, c, "pack");
189 static inline void osc_pack_req_body(struct ptlrpc_request *req,
190 struct obd_info *oinfo)
192 struct ost_body *body;
194 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
197 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa,
199 osc_pack_capa(req, body, oinfo->oi_capa);
202 static inline void osc_set_capa_size(struct ptlrpc_request *req,
203 const struct req_msg_field *field,
207 req_capsule_set_size(&req->rq_pill, field, RCL_CLIENT, 0);
209 /* it is already calculated as sizeof struct obd_capa */
213 static int osc_getattr_interpret(const struct lu_env *env,
214 struct ptlrpc_request *req,
215 struct osc_async_args *aa, int rc)
217 struct ost_body *body;
223 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
225 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
226 lustre_get_wire_obdo(&req->rq_import->imp_connect_data,
227 aa->aa_oi->oi_oa, &body->oa);
229 /* This should really be sent by the OST */
230 aa->aa_oi->oi_oa->o_blksize = DT_MAX_BRW_SIZE;
231 aa->aa_oi->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
233 CDEBUG(D_INFO, "can't unpack ost_body\n");
235 aa->aa_oi->oi_oa->o_valid = 0;
238 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
242 static int osc_getattr_async(struct obd_export *exp, struct obd_info *oinfo,
243 struct ptlrpc_request_set *set)
245 struct ptlrpc_request *req;
246 struct osc_async_args *aa;
250 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
254 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
255 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
257 ptlrpc_request_free(req);
261 osc_pack_req_body(req, oinfo);
263 ptlrpc_request_set_replen(req);
264 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_getattr_interpret;
266 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
267 aa = ptlrpc_req_async_args(req);
270 ptlrpc_set_add_req(set, req);
274 static int osc_getattr(const struct lu_env *env, struct obd_export *exp,
275 struct obd_info *oinfo)
277 struct ptlrpc_request *req;
278 struct ost_body *body;
282 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
286 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
287 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
289 ptlrpc_request_free(req);
293 osc_pack_req_body(req, oinfo);
295 ptlrpc_request_set_replen(req);
297 rc = ptlrpc_queue_wait(req);
301 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
303 GOTO(out, rc = -EPROTO);
305 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
306 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oinfo->oi_oa,
309 oinfo->oi_oa->o_blksize = cli_brw_size(exp->exp_obd);
310 oinfo->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
314 ptlrpc_req_finished(req);
318 static int osc_setattr(const struct lu_env *env, struct obd_export *exp,
319 struct obd_info *oinfo, struct obd_trans_info *oti)
321 struct ptlrpc_request *req;
322 struct ost_body *body;
326 LASSERT(oinfo->oi_oa->o_valid & OBD_MD_FLGROUP);
328 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
332 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
333 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
335 ptlrpc_request_free(req);
339 osc_pack_req_body(req, oinfo);
341 ptlrpc_request_set_replen(req);
343 rc = ptlrpc_queue_wait(req);
347 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
349 GOTO(out, rc = -EPROTO);
351 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oinfo->oi_oa,
356 ptlrpc_req_finished(req);
360 static int osc_setattr_interpret(const struct lu_env *env,
361 struct ptlrpc_request *req,
362 struct osc_setattr_args *sa, int rc)
364 struct ost_body *body;
370 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
372 GOTO(out, rc = -EPROTO);
374 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, sa->sa_oa,
377 rc = sa->sa_upcall(sa->sa_cookie, rc);
381 int osc_setattr_async_base(struct obd_export *exp, struct obd_info *oinfo,
382 struct obd_trans_info *oti,
383 obd_enqueue_update_f upcall, void *cookie,
384 struct ptlrpc_request_set *rqset)
386 struct ptlrpc_request *req;
387 struct osc_setattr_args *sa;
391 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
395 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
396 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
398 ptlrpc_request_free(req);
402 if (oti && oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE)
403 oinfo->oi_oa->o_lcookie = *oti->oti_logcookies;
405 osc_pack_req_body(req, oinfo);
407 ptlrpc_request_set_replen(req);
409 /* do mds to ost setattr asynchronously */
411 /* Do not wait for response. */
412 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
414 req->rq_interpret_reply =
415 (ptlrpc_interpterer_t)osc_setattr_interpret;
417 CLASSERT (sizeof(*sa) <= sizeof(req->rq_async_args));
418 sa = ptlrpc_req_async_args(req);
419 sa->sa_oa = oinfo->oi_oa;
420 sa->sa_upcall = upcall;
421 sa->sa_cookie = cookie;
423 if (rqset == PTLRPCD_SET)
424 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
426 ptlrpc_set_add_req(rqset, req);
432 static int osc_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
433 struct obd_trans_info *oti,
434 struct ptlrpc_request_set *rqset)
436 return osc_setattr_async_base(exp, oinfo, oti,
437 oinfo->oi_cb_up, oinfo, rqset);
440 int osc_real_create(struct obd_export *exp, struct obdo *oa,
441 struct lov_stripe_md **ea, struct obd_trans_info *oti)
443 struct ptlrpc_request *req;
444 struct ost_body *body;
445 struct lov_stripe_md *lsm;
454 rc = obd_alloc_memmd(exp, &lsm);
459 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
461 GOTO(out, rc = -ENOMEM);
463 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
465 ptlrpc_request_free(req);
469 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
472 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
474 ptlrpc_request_set_replen(req);
476 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
477 oa->o_flags == OBD_FL_DELORPHAN) {
479 "delorphan from OST integration");
480 /* Don't resend the delorphan req */
481 req->rq_no_resend = req->rq_no_delay = 1;
484 rc = ptlrpc_queue_wait(req);
488 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
490 GOTO(out_req, rc = -EPROTO);
492 CDEBUG(D_INFO, "oa flags %x\n", oa->o_flags);
493 lustre_get_wire_obdo(&req->rq_import->imp_connect_data, oa, &body->oa);
495 oa->o_blksize = cli_brw_size(exp->exp_obd);
496 oa->o_valid |= OBD_MD_FLBLKSZ;
498 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
499 * have valid lsm_oinfo data structs, so don't go touching that.
500 * This needs to be fixed in a big way.
502 lsm->lsm_oi = oa->o_oi;
506 oti->oti_transno = lustre_msg_get_transno(req->rq_repmsg);
508 if (oa->o_valid & OBD_MD_FLCOOKIE) {
509 if (!oti->oti_logcookies)
510 oti_alloc_cookies(oti, 1);
511 *oti->oti_logcookies = oa->o_lcookie;
515 CDEBUG(D_HA, "transno: "LPD64"\n",
516 lustre_msg_get_transno(req->rq_repmsg));
518 ptlrpc_req_finished(req);
521 obd_free_memmd(exp, &lsm);
525 int osc_punch_base(struct obd_export *exp, struct obd_info *oinfo,
526 obd_enqueue_update_f upcall, void *cookie,
527 struct ptlrpc_request_set *rqset)
529 struct ptlrpc_request *req;
530 struct osc_setattr_args *sa;
531 struct ost_body *body;
535 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_PUNCH);
539 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
540 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
542 ptlrpc_request_free(req);
545 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
546 ptlrpc_at_set_req_timeout(req);
548 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
550 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa,
552 osc_pack_capa(req, body, oinfo->oi_capa);
554 ptlrpc_request_set_replen(req);
556 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_setattr_interpret;
557 CLASSERT (sizeof(*sa) <= sizeof(req->rq_async_args));
558 sa = ptlrpc_req_async_args(req);
559 sa->sa_oa = oinfo->oi_oa;
560 sa->sa_upcall = upcall;
561 sa->sa_cookie = cookie;
562 if (rqset == PTLRPCD_SET)
563 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
565 ptlrpc_set_add_req(rqset, req);
570 static int osc_sync_interpret(const struct lu_env *env,
571 struct ptlrpc_request *req,
574 struct osc_fsync_args *fa = arg;
575 struct ost_body *body;
581 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
583 CERROR ("can't unpack ost_body\n");
584 GOTO(out, rc = -EPROTO);
587 *fa->fa_oi->oi_oa = body->oa;
589 rc = fa->fa_upcall(fa->fa_cookie, rc);
593 int osc_sync_base(struct obd_export *exp, struct obd_info *oinfo,
594 obd_enqueue_update_f upcall, void *cookie,
595 struct ptlrpc_request_set *rqset)
597 struct ptlrpc_request *req;
598 struct ost_body *body;
599 struct osc_fsync_args *fa;
603 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
607 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
608 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
610 ptlrpc_request_free(req);
614 /* overload the size and blocks fields in the oa with start/end */
615 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
617 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa,
619 osc_pack_capa(req, body, oinfo->oi_capa);
621 ptlrpc_request_set_replen(req);
622 req->rq_interpret_reply = osc_sync_interpret;
624 CLASSERT(sizeof(*fa) <= sizeof(req->rq_async_args));
625 fa = ptlrpc_req_async_args(req);
627 fa->fa_upcall = upcall;
628 fa->fa_cookie = cookie;
630 if (rqset == PTLRPCD_SET)
631 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
633 ptlrpc_set_add_req(rqset, req);
638 /* Find and cancel locally locks matched by @mode in the resource found by
639 * @objid. Found locks are added into @cancel list. Returns the amount of
640 * locks added to @cancels list. */
641 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
643 ldlm_mode_t mode, __u64 lock_flags)
645 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
646 struct ldlm_res_id res_id;
647 struct ldlm_resource *res;
651 /* Return, i.e. cancel nothing, only if ELC is supported (flag in
652 * export) but disabled through procfs (flag in NS).
654 * This distinguishes from a case when ELC is not supported originally,
655 * when we still want to cancel locks in advance and just cancel them
656 * locally, without sending any RPC. */
657 if (exp_connect_cancelset(exp) && !ns_connect_cancelset(ns))
660 ostid_build_res_name(&oa->o_oi, &res_id);
661 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
665 LDLM_RESOURCE_ADDREF(res);
666 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
667 lock_flags, 0, NULL);
668 LDLM_RESOURCE_DELREF(res);
669 ldlm_resource_putref(res);
673 static int osc_destroy_interpret(const struct lu_env *env,
674 struct ptlrpc_request *req, void *data,
677 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
679 atomic_dec(&cli->cl_destroy_in_flight);
680 wake_up(&cli->cl_destroy_waitq);
684 static int osc_can_send_destroy(struct client_obd *cli)
686 if (atomic_inc_return(&cli->cl_destroy_in_flight) <=
687 cli->cl_max_rpcs_in_flight) {
688 /* The destroy request can be sent */
691 if (atomic_dec_return(&cli->cl_destroy_in_flight) <
692 cli->cl_max_rpcs_in_flight) {
694 * The counter has been modified between the two atomic
697 wake_up(&cli->cl_destroy_waitq);
702 int osc_create(const struct lu_env *env, struct obd_export *exp,
703 struct obdo *oa, struct lov_stripe_md **ea,
704 struct obd_trans_info *oti)
711 LASSERT(oa->o_valid & OBD_MD_FLGROUP);
713 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
714 oa->o_flags == OBD_FL_RECREATE_OBJS) {
715 RETURN(osc_real_create(exp, oa, ea, oti));
718 if (!fid_seq_is_mdt(ostid_seq(&oa->o_oi)))
719 RETURN(osc_real_create(exp, oa, ea, oti));
721 /* we should not get here anymore */
727 /* Destroy requests can be async always on the client, and we don't even really
728 * care about the return code since the client cannot do anything at all about
730 * When the MDS is unlinking a filename, it saves the file objects into a
731 * recovery llog, and these object records are cancelled when the OST reports
732 * they were destroyed and sync'd to disk (i.e. transaction committed).
733 * If the client dies, or the OST is down when the object should be destroyed,
734 * the records are not cancelled, and when the OST reconnects to the MDS next,
735 * it will retrieve the llog unlink logs and then sends the log cancellation
736 * cookies to the MDS after committing destroy transactions. */
737 static int osc_destroy(const struct lu_env *env, struct obd_export *exp,
738 struct obdo *oa, struct lov_stripe_md *ea,
739 struct obd_trans_info *oti, struct obd_export *md_export,
742 struct client_obd *cli = &exp->exp_obd->u.cli;
743 struct ptlrpc_request *req;
744 struct ost_body *body;
745 CFS_LIST_HEAD(cancels);
750 CDEBUG(D_INFO, "oa NULL\n");
754 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
755 LDLM_FL_DISCARD_DATA);
757 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
759 ldlm_lock_list_put(&cancels, l_bl_ast, count);
763 osc_set_capa_size(req, &RMF_CAPA1, (struct obd_capa *)capa);
764 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
767 ptlrpc_request_free(req);
771 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
772 ptlrpc_at_set_req_timeout(req);
774 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
775 oa->o_lcookie = *oti->oti_logcookies;
776 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
778 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
780 osc_pack_capa(req, body, (struct obd_capa *)capa);
781 ptlrpc_request_set_replen(req);
783 /* If osc_destory is for destroying the unlink orphan,
784 * sent from MDT to OST, which should not be blocked here,
785 * because the process might be triggered by ptlrpcd, and
786 * it is not good to block ptlrpcd thread (b=16006)*/
787 if (!(oa->o_flags & OBD_FL_DELORPHAN)) {
788 req->rq_interpret_reply = osc_destroy_interpret;
789 if (!osc_can_send_destroy(cli)) {
790 struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP,
794 * Wait until the number of on-going destroy RPCs drops
795 * under max_rpc_in_flight
797 l_wait_event_exclusive(cli->cl_destroy_waitq,
798 osc_can_send_destroy(cli), &lwi);
802 /* Do not wait for response */
803 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
807 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
810 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
812 LASSERT(!(oa->o_valid & bits));
815 client_obd_list_lock(&cli->cl_loi_list_lock);
816 oa->o_dirty = cli->cl_dirty;
817 if (unlikely(cli->cl_dirty - cli->cl_dirty_transit >
818 cli->cl_dirty_max)) {
819 CERROR("dirty %lu - %lu > dirty_max %lu\n",
820 cli->cl_dirty, cli->cl_dirty_transit, cli->cl_dirty_max);
822 } else if (unlikely(atomic_read(&obd_unstable_pages) +
823 atomic_read(&obd_dirty_pages) -
824 atomic_read(&obd_dirty_transit_pages) >
825 (long)(obd_max_dirty_pages + 1))) {
826 /* The atomic_read() allowing the atomic_inc() are
827 * not covered by a lock thus they may safely race and trip
828 * this CERROR() unless we add in a small fudge factor (+1). */
829 CERROR("%s: dirty %d + %d - %d > system dirty_max %d\n",
830 cli->cl_import->imp_obd->obd_name,
831 atomic_read(&obd_unstable_pages),
832 atomic_read(&obd_dirty_pages),
833 atomic_read(&obd_dirty_transit_pages),
834 obd_max_dirty_pages);
836 } else if (unlikely(cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff)) {
837 CERROR("dirty %lu - dirty_max %lu too big???\n",
838 cli->cl_dirty, cli->cl_dirty_max);
841 long max_in_flight = (cli->cl_max_pages_per_rpc <<
843 (cli->cl_max_rpcs_in_flight + 1);
844 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
846 oa->o_grant = cli->cl_avail_grant + cli->cl_reserved_grant;
847 oa->o_dropped = cli->cl_lost_grant;
848 cli->cl_lost_grant = 0;
849 client_obd_list_unlock(&cli->cl_loi_list_lock);
850 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
851 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
855 void osc_update_next_shrink(struct client_obd *cli)
857 cli->cl_next_shrink_grant =
858 cfs_time_shift(cli->cl_grant_shrink_interval);
859 CDEBUG(D_CACHE, "next time %ld to shrink grant \n",
860 cli->cl_next_shrink_grant);
863 static void __osc_update_grant(struct client_obd *cli, obd_size grant)
865 client_obd_list_lock(&cli->cl_loi_list_lock);
866 cli->cl_avail_grant += grant;
867 client_obd_list_unlock(&cli->cl_loi_list_lock);
870 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
872 if (body->oa.o_valid & OBD_MD_FLGRANT) {
873 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
874 __osc_update_grant(cli, body->oa.o_grant);
878 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
879 obd_count keylen, void *key, obd_count vallen,
880 void *val, struct ptlrpc_request_set *set);
882 static int osc_shrink_grant_interpret(const struct lu_env *env,
883 struct ptlrpc_request *req,
886 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
887 struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
888 struct ost_body *body;
891 __osc_update_grant(cli, oa->o_grant);
895 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
897 osc_update_grant(cli, body);
903 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
905 client_obd_list_lock(&cli->cl_loi_list_lock);
906 oa->o_grant = cli->cl_avail_grant / 4;
907 cli->cl_avail_grant -= oa->o_grant;
908 client_obd_list_unlock(&cli->cl_loi_list_lock);
909 if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
910 oa->o_valid |= OBD_MD_FLFLAGS;
913 oa->o_flags |= OBD_FL_SHRINK_GRANT;
914 osc_update_next_shrink(cli);
917 /* Shrink the current grant, either from some large amount to enough for a
918 * full set of in-flight RPCs, or if we have already shrunk to that limit
919 * then to enough for a single RPC. This avoids keeping more grant than
920 * needed, and avoids shrinking the grant piecemeal. */
921 static int osc_shrink_grant(struct client_obd *cli)
923 __u64 target_bytes = (cli->cl_max_rpcs_in_flight + 1) *
924 (cli->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT);
926 client_obd_list_lock(&cli->cl_loi_list_lock);
927 if (cli->cl_avail_grant <= target_bytes)
928 target_bytes = cli->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT;
929 client_obd_list_unlock(&cli->cl_loi_list_lock);
931 return osc_shrink_grant_to_target(cli, target_bytes);
934 int osc_shrink_grant_to_target(struct client_obd *cli, __u64 target_bytes)
937 struct ost_body *body;
940 client_obd_list_lock(&cli->cl_loi_list_lock);
941 /* Don't shrink if we are already above or below the desired limit
942 * We don't want to shrink below a single RPC, as that will negatively
943 * impact block allocation and long-term performance. */
944 if (target_bytes < cli->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT)
945 target_bytes = cli->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT;
947 if (target_bytes >= cli->cl_avail_grant) {
948 client_obd_list_unlock(&cli->cl_loi_list_lock);
951 client_obd_list_unlock(&cli->cl_loi_list_lock);
957 osc_announce_cached(cli, &body->oa, 0);
959 client_obd_list_lock(&cli->cl_loi_list_lock);
960 body->oa.o_grant = cli->cl_avail_grant - target_bytes;
961 cli->cl_avail_grant = target_bytes;
962 client_obd_list_unlock(&cli->cl_loi_list_lock);
963 if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
964 body->oa.o_valid |= OBD_MD_FLFLAGS;
965 body->oa.o_flags = 0;
967 body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
968 osc_update_next_shrink(cli);
970 rc = osc_set_info_async(NULL, cli->cl_import->imp_obd->obd_self_export,
971 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
972 sizeof(*body), body, NULL);
974 __osc_update_grant(cli, body->oa.o_grant);
979 static int osc_should_shrink_grant(struct client_obd *client)
981 cfs_time_t time = cfs_time_current();
982 cfs_time_t next_shrink = client->cl_next_shrink_grant;
984 if ((client->cl_import->imp_connect_data.ocd_connect_flags &
985 OBD_CONNECT_GRANT_SHRINK) == 0)
988 if (cfs_time_aftereq(time, next_shrink - 5 * CFS_TICK)) {
989 /* Get the current RPC size directly, instead of going via:
990 * cli_brw_size(obd->u.cli.cl_import->imp_obd->obd_self_export)
991 * Keep comment here so that it can be found by searching. */
992 int brw_size = client->cl_max_pages_per_rpc << PAGE_CACHE_SHIFT;
994 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
995 client->cl_avail_grant > brw_size)
998 osc_update_next_shrink(client);
1003 static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
1005 struct client_obd *client;
1007 cfs_list_for_each_entry(client, &item->ti_obd_list,
1008 cl_grant_shrink_list) {
1009 if (osc_should_shrink_grant(client))
1010 osc_shrink_grant(client);
1015 static int osc_add_shrink_grant(struct client_obd *client)
1019 rc = ptlrpc_add_timeout_client(client->cl_grant_shrink_interval,
1021 osc_grant_shrink_grant_cb, NULL,
1022 &client->cl_grant_shrink_list);
1024 CERROR("add grant client %s error %d\n",
1025 client->cl_import->imp_obd->obd_name, rc);
1028 CDEBUG(D_CACHE, "add grant client %s \n",
1029 client->cl_import->imp_obd->obd_name);
1030 osc_update_next_shrink(client);
1034 static int osc_del_shrink_grant(struct client_obd *client)
1036 return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
1040 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1043 * ocd_grant is the total grant amount we're expect to hold: if we've
1044 * been evicted, it's the new avail_grant amount, cl_dirty will drop
1045 * to 0 as inflight RPCs fail out; otherwise, it's avail_grant + dirty.
1047 * race is tolerable here: if we're evicted, but imp_state already
1048 * left EVICTED state, then cl_dirty must be 0 already.
1050 client_obd_list_lock(&cli->cl_loi_list_lock);
1051 if (cli->cl_import->imp_state == LUSTRE_IMP_EVICTED)
1052 cli->cl_avail_grant = ocd->ocd_grant;
1054 cli->cl_avail_grant = ocd->ocd_grant - cli->cl_dirty;
1056 if (cli->cl_avail_grant < 0) {
1057 CWARN("%s: available grant < 0: avail/ocd/dirty %ld/%u/%ld\n",
1058 cli->cl_import->imp_obd->obd_name, cli->cl_avail_grant,
1059 ocd->ocd_grant, cli->cl_dirty);
1060 /* workaround for servers which do not have the patch from
1062 cli->cl_avail_grant = ocd->ocd_grant;
1065 /* determine the appropriate chunk size used by osc_extent. */
1066 cli->cl_chunkbits = max_t(int, PAGE_CACHE_SHIFT, ocd->ocd_blocksize);
1067 client_obd_list_unlock(&cli->cl_loi_list_lock);
1069 CDEBUG(D_CACHE, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld."
1070 "chunk bits: %d.\n", cli->cl_import->imp_obd->obd_name,
1071 cli->cl_avail_grant, cli->cl_lost_grant, cli->cl_chunkbits);
1073 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
1074 cfs_list_empty(&cli->cl_grant_shrink_list))
1075 osc_add_shrink_grant(cli);
1078 /* We assume that the reason this OSC got a short read is because it read
1079 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1080 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1081 * this stripe never got written at or beyond this stripe offset yet. */
1082 static void handle_short_read(int nob_read, obd_count page_count,
1083 struct brw_page **pga)
1088 /* skip bytes read OK */
1089 while (nob_read > 0) {
1090 LASSERT (page_count > 0);
1092 if (pga[i]->count > nob_read) {
1093 /* EOF inside this page */
1094 ptr = kmap(pga[i]->pg) +
1095 (pga[i]->off & ~CFS_PAGE_MASK);
1096 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1103 nob_read -= pga[i]->count;
1108 /* zero remaining pages */
1109 while (page_count-- > 0) {
1110 ptr = kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
1111 memset(ptr, 0, pga[i]->count);
1117 static int check_write_rcs(struct ptlrpc_request *req,
1118 int requested_nob, int niocount,
1119 obd_count page_count, struct brw_page **pga)
1124 remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1125 sizeof(*remote_rcs) *
1127 if (remote_rcs == NULL) {
1128 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1132 /* return error if any niobuf was in error */
1133 for (i = 0; i < niocount; i++) {
1134 if ((int)remote_rcs[i] < 0)
1135 return(remote_rcs[i]);
1137 if (remote_rcs[i] != 0) {
1138 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1139 i, remote_rcs[i], req);
1144 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
1145 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1146 req->rq_bulk->bd_nob_transferred, requested_nob);
1153 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1155 if (p1->flag != p2->flag) {
1156 unsigned mask = ~(OBD_BRW_FROM_GRANT | OBD_BRW_NOCACHE |
1157 OBD_BRW_SYNC | OBD_BRW_ASYNC |
1158 OBD_BRW_NOQUOTA | OBD_BRW_SOFT_SYNC);
1160 /* warn if we try to combine flags that we don't know to be
1161 * safe to combine */
1162 if (unlikely((p1->flag & mask) != (p2->flag & mask))) {
1163 CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1164 "report this at http://bugs.whamcloud.com/\n",
1165 p1->flag, p2->flag);
1170 return (p1->off + p1->count == p2->off);
1173 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
1174 struct brw_page **pga, int opc,
1175 cksum_type_t cksum_type)
1179 struct cfs_crypto_hash_desc *hdesc;
1180 unsigned int bufsize;
1182 unsigned char cfs_alg = cksum_obd2cfs(cksum_type);
1184 LASSERT(pg_count > 0);
1186 hdesc = cfs_crypto_hash_init(cfs_alg, NULL, 0);
1187 if (IS_ERR(hdesc)) {
1188 CERROR("Unable to initialize checksum hash %s\n",
1189 cfs_crypto_hash_name(cfs_alg));
1190 return PTR_ERR(hdesc);
1193 while (nob > 0 && pg_count > 0) {
1194 int count = pga[i]->count > nob ? nob : pga[i]->count;
1196 /* corrupt the data before we compute the checksum, to
1197 * simulate an OST->client data error */
1198 if (i == 0 && opc == OST_READ &&
1199 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE)) {
1200 unsigned char *ptr = kmap(pga[i]->pg);
1201 int off = pga[i]->off & ~CFS_PAGE_MASK;
1202 memcpy(ptr + off, "bad1", min(4, nob));
1205 cfs_crypto_hash_update_page(hdesc, pga[i]->pg,
1206 pga[i]->off & ~CFS_PAGE_MASK,
1208 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d\n",
1209 (int)(pga[i]->off & ~CFS_PAGE_MASK));
1211 nob -= pga[i]->count;
1217 err = cfs_crypto_hash_final(hdesc, (unsigned char *)&cksum, &bufsize);
1220 cfs_crypto_hash_final(hdesc, NULL, NULL);
1222 /* For sending we only compute the wrong checksum instead
1223 * of corrupting the data so it is still correct on a redo */
1224 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1230 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
1231 struct lov_stripe_md *lsm, obd_count page_count,
1232 struct brw_page **pga,
1233 struct ptlrpc_request **reqp,
1234 struct obd_capa *ocapa, int reserve,
1237 struct ptlrpc_request *req;
1238 struct ptlrpc_bulk_desc *desc;
1239 struct ost_body *body;
1240 struct obd_ioobj *ioobj;
1241 struct niobuf_remote *niobuf;
1242 int niocount, i, requested_nob, opc, rc;
1243 struct osc_brw_async_args *aa;
1244 struct req_capsule *pill;
1245 struct brw_page *pg_prev;
1248 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1249 RETURN(-ENOMEM); /* Recoverable */
1250 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1251 RETURN(-EINVAL); /* Fatal */
1253 if ((cmd & OBD_BRW_WRITE) != 0) {
1255 req = ptlrpc_request_alloc_pool(cli->cl_import,
1256 cli->cl_import->imp_rq_pool,
1257 &RQF_OST_BRW_WRITE);
1260 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1265 for (niocount = i = 1; i < page_count; i++) {
1266 if (!can_merge_pages(pga[i - 1], pga[i]))
1270 pill = &req->rq_pill;
1271 req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1273 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1274 niocount * sizeof(*niobuf));
1275 osc_set_capa_size(req, &RMF_CAPA1, ocapa);
1277 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1279 ptlrpc_request_free(req);
1282 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
1283 ptlrpc_at_set_req_timeout(req);
1284 /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1286 req->rq_no_retry_einprogress = 1;
1288 desc = ptlrpc_prep_bulk_imp(req, page_count,
1289 cli->cl_import->imp_connect_data.ocd_brw_size >> LNET_MTU_BITS,
1290 opc == OST_WRITE ? BULK_GET_SOURCE : BULK_PUT_SINK,
1294 GOTO(out, rc = -ENOMEM);
1295 /* NB request now owns desc and will free it when it gets freed */
1297 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1298 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1299 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1300 LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1302 lustre_set_wire_obdo(&req->rq_import->imp_connect_data, &body->oa, oa);
1304 obdo_to_ioobj(oa, ioobj);
1305 ioobj->ioo_bufcnt = niocount;
1306 /* The high bits of ioo_max_brw tells server _maximum_ number of bulks
1307 * that might be send for this request. The actual number is decided
1308 * when the RPC is finally sent in ptlrpc_register_bulk(). It sends
1309 * "max - 1" for old client compatibility sending "0", and also so the
1310 * the actual maximum is a power-of-two number, not one less. LU-1431 */
1311 ioobj_max_brw_set(ioobj, desc->bd_md_max_brw);
1312 osc_pack_capa(req, body, ocapa);
1313 LASSERT(page_count > 0);
1315 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1316 struct brw_page *pg = pga[i];
1317 int poff = pg->off & ~CFS_PAGE_MASK;
1319 LASSERT(pg->count > 0);
1320 /* make sure there is no gap in the middle of page array */
1321 LASSERTF(page_count == 1 ||
1322 (ergo(i == 0, poff + pg->count == PAGE_CACHE_SIZE) &&
1323 ergo(i > 0 && i < page_count - 1,
1324 poff == 0 && pg->count == PAGE_CACHE_SIZE) &&
1325 ergo(i == page_count - 1, poff == 0)),
1326 "i: %d/%d pg: %p off: "LPU64", count: %u\n",
1327 i, page_count, pg, pg->off, pg->count);
1329 LASSERTF(i == 0 || pg->off > pg_prev->off,
1330 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1331 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1333 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1334 pg_prev->pg, page_private(pg_prev->pg),
1335 pg_prev->pg->index, pg_prev->off);
1337 LASSERTF(i == 0 || pg->off > pg_prev->off,
1338 "i %d p_c %u\n", i, page_count);
1340 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1341 (pg->flag & OBD_BRW_SRVLOCK));
1343 ptlrpc_prep_bulk_page_pin(desc, pg->pg, poff, pg->count);
1344 requested_nob += pg->count;
1346 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1348 niobuf->len += pg->count;
1350 niobuf->offset = pg->off;
1351 niobuf->len = pg->count;
1352 niobuf->flags = pg->flag;
1357 LASSERTF((void *)(niobuf - niocount) ==
1358 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1359 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1360 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1362 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1364 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1365 body->oa.o_valid |= OBD_MD_FLFLAGS;
1366 body->oa.o_flags = 0;
1368 body->oa.o_flags |= OBD_FL_RECOV_RESEND;
1371 if (osc_should_shrink_grant(cli))
1372 osc_shrink_grant_local(cli, &body->oa);
1374 /* size[REQ_REC_OFF] still sizeof (*body) */
1375 if (opc == OST_WRITE) {
1376 if (cli->cl_checksum &&
1377 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1378 /* store cl_cksum_type in a local variable since
1379 * it can be changed via lprocfs */
1380 cksum_type_t cksum_type = cli->cl_cksum_type;
1382 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1383 oa->o_flags &= OBD_FL_LOCAL_MASK;
1384 body->oa.o_flags = 0;
1386 body->oa.o_flags |= cksum_type_pack(cksum_type);
1387 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1388 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1392 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1394 /* save this in 'oa', too, for later checking */
1395 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1396 oa->o_flags |= cksum_type_pack(cksum_type);
1398 /* clear out the checksum flag, in case this is a
1399 * resend but cl_checksum is no longer set. b=11238 */
1400 oa->o_valid &= ~OBD_MD_FLCKSUM;
1402 oa->o_cksum = body->oa.o_cksum;
1403 /* 1 RC per niobuf */
1404 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1405 sizeof(__u32) * niocount);
1407 if (cli->cl_checksum &&
1408 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1409 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1410 body->oa.o_flags = 0;
1411 body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
1412 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1415 ptlrpc_request_set_replen(req);
1417 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1418 aa = ptlrpc_req_async_args(req);
1420 aa->aa_requested_nob = requested_nob;
1421 aa->aa_nio_count = niocount;
1422 aa->aa_page_count = page_count;
1426 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1427 if (ocapa && reserve)
1428 aa->aa_ocapa = capa_get(ocapa);
1434 ptlrpc_req_finished(req);
1438 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1439 __u32 client_cksum, __u32 server_cksum, int nob,
1440 obd_count page_count, struct brw_page **pga,
1441 cksum_type_t client_cksum_type)
1445 cksum_type_t cksum_type;
1447 if (server_cksum == client_cksum) {
1448 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1452 cksum_type = cksum_type_unpack(oa->o_valid & OBD_MD_FLFLAGS ?
1454 new_cksum = osc_checksum_bulk(nob, page_count, pga, OST_WRITE,
1457 if (cksum_type != client_cksum_type)
1458 msg = "the server did not use the checksum type specified in "
1459 "the original request - likely a protocol problem";
1460 else if (new_cksum == server_cksum)
1461 msg = "changed on the client after we checksummed it - "
1462 "likely false positive due to mmap IO (bug 11742)";
1463 else if (new_cksum == client_cksum)
1464 msg = "changed in transit before arrival at OST";
1466 msg = "changed in transit AND doesn't match the original - "
1467 "likely false positive due to mmap IO (bug 11742)";
1469 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inode "DFID
1470 " object "DOSTID" extent ["LPU64"-"LPU64"]\n",
1471 msg, libcfs_nid2str(peer->nid),
1472 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1473 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1474 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1475 POSTID(&oa->o_oi), pga[0]->off,
1476 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1477 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1478 "client csum now %x\n", client_cksum, client_cksum_type,
1479 server_cksum, cksum_type, new_cksum);
1483 /* Note rc enters this function as number of bytes transferred */
1484 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1486 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1487 const lnet_process_id_t *peer =
1488 &req->rq_import->imp_connection->c_peer;
1489 struct client_obd *cli = aa->aa_cli;
1490 struct ost_body *body;
1491 __u32 client_cksum = 0;
1494 if (rc < 0 && rc != -EDQUOT) {
1495 DEBUG_REQ(D_INFO, req, "Failed request with rc = %d\n", rc);
1499 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1500 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1502 DEBUG_REQ(D_INFO, req, "Can't unpack body\n");
1506 /* set/clear over quota flag for a uid/gid */
1507 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1508 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA)) {
1509 unsigned int qid[MAXQUOTAS] = { body->oa.o_uid, body->oa.o_gid };
1511 CDEBUG(D_QUOTA, "setdq for [%u %u] with valid "LPX64", flags %x\n",
1512 body->oa.o_uid, body->oa.o_gid, body->oa.o_valid,
1514 osc_quota_setdq(cli, qid, body->oa.o_valid, body->oa.o_flags);
1517 osc_update_grant(cli, body);
1522 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1523 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1525 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1527 CERROR("Unexpected +ve rc %d\n", rc);
1530 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1532 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1535 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1536 check_write_checksum(&body->oa, peer, client_cksum,
1537 body->oa.o_cksum, aa->aa_requested_nob,
1538 aa->aa_page_count, aa->aa_ppga,
1539 cksum_type_unpack(aa->aa_oa->o_flags)))
1542 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1543 aa->aa_page_count, aa->aa_ppga);
1547 /* The rest of this function executes only for OST_READs */
1549 /* if unwrap_bulk failed, return -EAGAIN to retry */
1550 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1552 GOTO(out, rc = -EAGAIN);
1554 if (rc > aa->aa_requested_nob) {
1555 CERROR("Unexpected rc %d (%d requested)\n", rc,
1556 aa->aa_requested_nob);
1560 if (rc != req->rq_bulk->bd_nob_transferred) {
1561 CERROR ("Unexpected rc %d (%d transferred)\n",
1562 rc, req->rq_bulk->bd_nob_transferred);
1566 if (rc < aa->aa_requested_nob)
1567 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1569 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
1570 static int cksum_counter;
1571 __u32 server_cksum = body->oa.o_cksum;
1574 cksum_type_t cksum_type;
1576 cksum_type = cksum_type_unpack(body->oa.o_valid &OBD_MD_FLFLAGS?
1577 body->oa.o_flags : 0);
1578 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1579 aa->aa_ppga, OST_READ,
1582 if (peer->nid == req->rq_bulk->bd_sender) {
1586 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1589 if (server_cksum == ~0 && rc > 0) {
1590 CERROR("Protocol error: server %s set the 'checksum' "
1591 "bit, but didn't send a checksum. Not fatal, "
1592 "but please notify on http://bugs.whamcloud.com/\n",
1593 libcfs_nid2str(peer->nid));
1594 } else if (server_cksum != client_cksum) {
1595 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1596 "%s%s%s inode "DFID" object "DOSTID
1597 " extent ["LPU64"-"LPU64"]\n",
1598 req->rq_import->imp_obd->obd_name,
1599 libcfs_nid2str(peer->nid),
1601 body->oa.o_valid & OBD_MD_FLFID ?
1602 body->oa.o_parent_seq : (__u64)0,
1603 body->oa.o_valid & OBD_MD_FLFID ?
1604 body->oa.o_parent_oid : 0,
1605 body->oa.o_valid & OBD_MD_FLFID ?
1606 body->oa.o_parent_ver : 0,
1607 POSTID(&body->oa.o_oi),
1608 aa->aa_ppga[0]->off,
1609 aa->aa_ppga[aa->aa_page_count-1]->off +
1610 aa->aa_ppga[aa->aa_page_count-1]->count -
1612 CERROR("client %x, server %x, cksum_type %x\n",
1613 client_cksum, server_cksum, cksum_type);
1615 aa->aa_oa->o_cksum = client_cksum;
1619 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1622 } else if (unlikely(client_cksum)) {
1623 static int cksum_missed;
1626 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1627 CERROR("Checksum %u requested from %s but not sent\n",
1628 cksum_missed, libcfs_nid2str(peer->nid));
1634 lustre_get_wire_obdo(&req->rq_import->imp_connect_data,
1635 aa->aa_oa, &body->oa);
1640 static int osc_brw_redo_request(struct ptlrpc_request *request,
1641 struct osc_brw_async_args *aa, int rc)
1643 struct ptlrpc_request *new_req;
1644 struct osc_brw_async_args *new_aa;
1645 struct osc_async_page *oap;
1648 DEBUG_REQ(rc == -EINPROGRESS ? D_RPCTRACE : D_ERROR, request,
1649 "redo for recoverable error %d", rc);
1651 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
1652 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1653 aa->aa_cli, aa->aa_oa,
1654 NULL /* lsm unused by osc currently */,
1655 aa->aa_page_count, aa->aa_ppga,
1656 &new_req, aa->aa_ocapa, 0, 1);
1660 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1661 if (oap->oap_request != NULL) {
1662 LASSERTF(request == oap->oap_request,
1663 "request %p != oap_request %p\n",
1664 request, oap->oap_request);
1665 if (oap->oap_interrupted) {
1666 ptlrpc_req_finished(new_req);
1671 /* New request takes over pga and oaps from old request.
1672 * Note that copying a list_head doesn't work, need to move it... */
1674 new_req->rq_interpret_reply = request->rq_interpret_reply;
1675 new_req->rq_async_args = request->rq_async_args;
1676 new_req->rq_commit_cb = request->rq_commit_cb;
1677 /* cap resend delay to the current request timeout, this is similar to
1678 * what ptlrpc does (see after_reply()) */
1679 if (aa->aa_resends > new_req->rq_timeout)
1680 new_req->rq_sent = cfs_time_current_sec() + new_req->rq_timeout;
1682 new_req->rq_sent = cfs_time_current_sec() + aa->aa_resends;
1683 new_req->rq_generation_set = 1;
1684 new_req->rq_import_generation = request->rq_import_generation;
1686 new_aa = ptlrpc_req_async_args(new_req);
1688 CFS_INIT_LIST_HEAD(&new_aa->aa_oaps);
1689 cfs_list_splice_init(&aa->aa_oaps, &new_aa->aa_oaps);
1690 CFS_INIT_LIST_HEAD(&new_aa->aa_exts);
1691 cfs_list_splice_init(&aa->aa_exts, &new_aa->aa_exts);
1692 new_aa->aa_resends = aa->aa_resends;
1694 cfs_list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1695 if (oap->oap_request) {
1696 ptlrpc_req_finished(oap->oap_request);
1697 oap->oap_request = ptlrpc_request_addref(new_req);
1701 new_aa->aa_ocapa = aa->aa_ocapa;
1702 aa->aa_ocapa = NULL;
1704 /* XXX: This code will run into problem if we're going to support
1705 * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
1706 * and wait for all of them to be finished. We should inherit request
1707 * set from old request. */
1708 ptlrpcd_add_req(new_req, PDL_POLICY_SAME, -1);
1710 DEBUG_REQ(D_INFO, new_req, "new request");
1715 * ugh, we want disk allocation on the target to happen in offset order. we'll
1716 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1717 * fine for our small page arrays and doesn't require allocation. its an
1718 * insertion sort that swaps elements that are strides apart, shrinking the
1719 * stride down until its '1' and the array is sorted.
1721 static void sort_brw_pages(struct brw_page **array, int num)
1724 struct brw_page *tmp;
1728 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1733 for (i = stride ; i < num ; i++) {
1736 while (j >= stride && array[j - stride]->off > tmp->off) {
1737 array[j] = array[j - stride];
1742 } while (stride > 1);
1745 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1747 LASSERT(ppga != NULL);
1748 OBD_FREE(ppga, sizeof(*ppga) * count);
1751 static int brw_interpret(const struct lu_env *env,
1752 struct ptlrpc_request *req, void *data, int rc)
1754 struct osc_brw_async_args *aa = data;
1755 struct osc_extent *ext;
1756 struct osc_extent *tmp;
1757 struct client_obd *cli = aa->aa_cli;
1760 rc = osc_brw_fini_request(req, rc);
1761 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
1762 /* When server return -EINPROGRESS, client should always retry
1763 * regardless of the number of times the bulk was resent already. */
1764 if (osc_recoverable_error(rc)) {
1765 if (req->rq_import_generation !=
1766 req->rq_import->imp_generation) {
1767 CDEBUG(D_HA, "%s: resend cross eviction for object: "
1768 ""DOSTID", rc = %d.\n",
1769 req->rq_import->imp_obd->obd_name,
1770 POSTID(&aa->aa_oa->o_oi), rc);
1771 } else if (rc == -EINPROGRESS ||
1772 client_should_resend(aa->aa_resends, aa->aa_cli)) {
1773 rc = osc_brw_redo_request(req, aa, rc);
1775 CERROR("%s: too many resent retries for object: "
1776 ""LPU64":"LPU64", rc = %d.\n",
1777 req->rq_import->imp_obd->obd_name,
1778 POSTID(&aa->aa_oa->o_oi), rc);
1783 else if (rc == -EAGAIN || rc == -EINPROGRESS)
1788 capa_put(aa->aa_ocapa);
1789 aa->aa_ocapa = NULL;
1793 struct obdo *oa = aa->aa_oa;
1794 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
1795 unsigned long valid = 0;
1796 struct cl_object *obj;
1797 struct osc_async_page *last;
1799 last = brw_page2oap(aa->aa_ppga[aa->aa_page_count - 1]);
1800 obj = osc2cl(last->oap_obj);
1802 cl_object_attr_lock(obj);
1803 if (oa->o_valid & OBD_MD_FLBLOCKS) {
1804 attr->cat_blocks = oa->o_blocks;
1805 valid |= CAT_BLOCKS;
1807 if (oa->o_valid & OBD_MD_FLMTIME) {
1808 attr->cat_mtime = oa->o_mtime;
1811 if (oa->o_valid & OBD_MD_FLATIME) {
1812 attr->cat_atime = oa->o_atime;
1815 if (oa->o_valid & OBD_MD_FLCTIME) {
1816 attr->cat_ctime = oa->o_ctime;
1820 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1821 struct lov_oinfo *loi = cl2osc(obj)->oo_oinfo;
1822 loff_t last_off = last->oap_count + last->oap_obj_off;
1824 /* Change file size if this is an out of quota or
1825 * direct IO write and it extends the file size */
1826 if (loi->loi_lvb.lvb_size < last_off) {
1827 attr->cat_size = last_off;
1830 /* Extend KMS if it's not a lockless write */
1831 if (loi->loi_kms < last_off &&
1832 oap2osc_page(last)->ops_srvlock == 0) {
1833 attr->cat_kms = last_off;
1839 cl_object_attr_set(env, obj, attr, valid);
1840 cl_object_attr_unlock(obj);
1842 OBDO_FREE(aa->aa_oa);
1844 cfs_list_for_each_entry_safe(ext, tmp, &aa->aa_exts, oe_link) {
1845 cfs_list_del_init(&ext->oe_link);
1846 osc_extent_finish(env, ext, 1, rc);
1848 LASSERT(cfs_list_empty(&aa->aa_exts));
1849 LASSERT(cfs_list_empty(&aa->aa_oaps));
1851 cl_req_completion(env, aa->aa_clerq, rc < 0 ? rc :
1852 req->rq_bulk->bd_nob_transferred);
1853 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
1854 ptlrpc_lprocfs_brw(req, req->rq_bulk->bd_nob_transferred);
1856 client_obd_list_lock(&cli->cl_loi_list_lock);
1857 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
1858 * is called so we know whether to go to sync BRWs or wait for more
1859 * RPCs to complete */
1860 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
1861 cli->cl_w_in_flight--;
1863 cli->cl_r_in_flight--;
1864 osc_wake_cache_waiters(cli);
1865 client_obd_list_unlock(&cli->cl_loi_list_lock);
1867 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
1871 static void brw_commit(struct ptlrpc_request *req)
1873 /* If osc_inc_unstable_pages (via osc_extent_finish) races with
1874 * this called via the rq_commit_cb, I need to ensure
1875 * osc_dec_unstable_pages is still called. Otherwise unstable
1876 * pages may be leaked. */
1877 spin_lock(&req->rq_lock);
1878 if (likely(req->rq_unstable)) {
1879 req->rq_unstable = 0;
1880 spin_unlock(&req->rq_lock);
1882 osc_dec_unstable_pages(req);
1884 req->rq_committed = 1;
1885 spin_unlock(&req->rq_lock);
1890 * Build an RPC by the list of extent @ext_list. The caller must ensure
1891 * that the total pages in this list are NOT over max pages per RPC.
1892 * Extents in the list must be in OES_RPC state.
1894 int osc_build_rpc(const struct lu_env *env, struct client_obd *cli,
1895 cfs_list_t *ext_list, int cmd, pdl_policy_t pol)
1897 struct ptlrpc_request *req = NULL;
1898 struct osc_extent *ext;
1899 struct brw_page **pga = NULL;
1900 struct osc_brw_async_args *aa = NULL;
1901 struct obdo *oa = NULL;
1902 struct osc_async_page *oap;
1903 struct osc_async_page *tmp;
1904 struct cl_req *clerq = NULL;
1905 enum cl_req_type crt = (cmd & OBD_BRW_WRITE) ? CRT_WRITE :
1907 struct ldlm_lock *lock = NULL;
1908 struct cl_req_attr *crattr = NULL;
1909 obd_off starting_offset = OBD_OBJECT_EOF;
1910 obd_off ending_offset = 0;
1916 CFS_LIST_HEAD(rpc_list);
1919 LASSERT(!cfs_list_empty(ext_list));
1921 /* add pages into rpc_list to build BRW rpc */
1922 cfs_list_for_each_entry(ext, ext_list, oe_link) {
1923 LASSERT(ext->oe_state == OES_RPC);
1924 mem_tight |= ext->oe_memalloc;
1925 cfs_list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
1927 cfs_list_add_tail(&oap->oap_rpc_item, &rpc_list);
1928 if (starting_offset > oap->oap_obj_off)
1929 starting_offset = oap->oap_obj_off;
1931 LASSERT(oap->oap_page_off == 0);
1932 if (ending_offset < oap->oap_obj_off + oap->oap_count)
1933 ending_offset = oap->oap_obj_off +
1936 LASSERT(oap->oap_page_off + oap->oap_count ==
1942 mpflag = cfs_memory_pressure_get_and_set();
1944 OBD_ALLOC(crattr, sizeof(*crattr));
1946 GOTO(out, rc = -ENOMEM);
1948 OBD_ALLOC(pga, sizeof(*pga) * page_count);
1950 GOTO(out, rc = -ENOMEM);
1954 GOTO(out, rc = -ENOMEM);
1957 cfs_list_for_each_entry(oap, &rpc_list, oap_rpc_item) {
1958 struct cl_page *page = oap2cl_page(oap);
1959 if (clerq == NULL) {
1960 clerq = cl_req_alloc(env, page, crt,
1961 1 /* only 1-object rpcs for now */);
1963 GOTO(out, rc = PTR_ERR(clerq));
1964 lock = oap->oap_ldlm_lock;
1967 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
1968 pga[i] = &oap->oap_brw_page;
1969 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
1970 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
1971 pga[i]->pg, page_index(oap->oap_page), oap,
1974 cl_req_page_add(env, clerq, page);
1977 /* always get the data for the obdo for the rpc */
1978 LASSERT(clerq != NULL);
1979 crattr->cra_oa = oa;
1980 cl_req_attr_set(env, clerq, crattr, ~0ULL);
1982 oa->o_handle = lock->l_remote_handle;
1983 oa->o_valid |= OBD_MD_FLHANDLE;
1986 rc = cl_req_prep(env, clerq);
1988 CERROR("cl_req_prep failed: %d\n", rc);
1992 sort_brw_pages(pga, page_count);
1993 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
1994 pga, &req, crattr->cra_capa, 1, 0);
1996 CERROR("prep_req failed: %d\n", rc);
2000 req->rq_commit_cb = brw_commit;
2001 req->rq_interpret_reply = brw_interpret;
2004 req->rq_memalloc = 1;
2006 /* Need to update the timestamps after the request is built in case
2007 * we race with setattr (locally or in queue at OST). If OST gets
2008 * later setattr before earlier BRW (as determined by the request xid),
2009 * the OST will not use BRW timestamps. Sadly, there is no obvious
2010 * way to do this in a single call. bug 10150 */
2011 cl_req_attr_set(env, clerq, crattr,
2012 OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME);
2014 lustre_msg_set_jobid(req->rq_reqmsg, crattr->cra_jobid);
2016 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2017 aa = ptlrpc_req_async_args(req);
2018 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
2019 cfs_list_splice_init(&rpc_list, &aa->aa_oaps);
2020 CFS_INIT_LIST_HEAD(&aa->aa_exts);
2021 cfs_list_splice_init(ext_list, &aa->aa_exts);
2022 aa->aa_clerq = clerq;
2024 /* queued sync pages can be torn down while the pages
2025 * were between the pending list and the rpc */
2027 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2028 /* only one oap gets a request reference */
2031 if (oap->oap_interrupted && !req->rq_intr) {
2032 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2034 ptlrpc_mark_interrupted(req);
2038 tmp->oap_request = ptlrpc_request_addref(req);
2040 client_obd_list_lock(&cli->cl_loi_list_lock);
2041 starting_offset >>= PAGE_CACHE_SHIFT;
2042 if (cmd == OBD_BRW_READ) {
2043 cli->cl_r_in_flight++;
2044 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2045 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2046 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2047 starting_offset + 1);
2049 cli->cl_w_in_flight++;
2050 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2051 lprocfs_oh_tally(&cli->cl_write_rpc_hist, cli->cl_w_in_flight);
2052 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2053 starting_offset + 1);
2055 client_obd_list_unlock(&cli->cl_loi_list_lock);
2057 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2058 page_count, aa, cli->cl_r_in_flight,
2059 cli->cl_w_in_flight);
2061 /* XXX: Maybe the caller can check the RPC bulk descriptor to
2062 * see which CPU/NUMA node the majority of pages were allocated
2063 * on, and try to assign the async RPC to the CPU core
2064 * (PDL_POLICY_PREFERRED) to reduce cross-CPU memory traffic.
2066 * But on the other hand, we expect that multiple ptlrpcd
2067 * threads and the initial write sponsor can run in parallel,
2068 * especially when data checksum is enabled, which is CPU-bound
2069 * operation and single ptlrpcd thread cannot process in time.
2070 * So more ptlrpcd threads sharing BRW load
2071 * (with PDL_POLICY_ROUND) seems better.
2073 ptlrpcd_add_req(req, pol, -1);
2079 cfs_memory_pressure_restore(mpflag);
2081 if (crattr != NULL) {
2082 capa_put(crattr->cra_capa);
2083 OBD_FREE(crattr, sizeof(*crattr));
2087 LASSERT(req == NULL);
2092 OBD_FREE(pga, sizeof(*pga) * page_count);
2093 /* this should happen rarely and is pretty bad, it makes the
2094 * pending list not follow the dirty order */
2095 while (!cfs_list_empty(ext_list)) {
2096 ext = cfs_list_entry(ext_list->next, struct osc_extent,
2098 cfs_list_del_init(&ext->oe_link);
2099 osc_extent_finish(env, ext, 0, rc);
2101 if (clerq && !IS_ERR(clerq))
2102 cl_req_completion(env, clerq, rc);
2107 static int osc_set_lock_data_with_check(struct ldlm_lock *lock,
2108 struct ldlm_enqueue_info *einfo)
2110 void *data = einfo->ei_cbdata;
2113 LASSERT(lock != NULL);
2114 LASSERT(lock->l_blocking_ast == einfo->ei_cb_bl);
2115 LASSERT(lock->l_resource->lr_type == einfo->ei_type);
2116 LASSERT(lock->l_completion_ast == einfo->ei_cb_cp);
2117 LASSERT(lock->l_glimpse_ast == einfo->ei_cb_gl);
2119 lock_res_and_lock(lock);
2120 spin_lock(&osc_ast_guard);
2122 if (lock->l_ast_data == NULL)
2123 lock->l_ast_data = data;
2124 if (lock->l_ast_data == data)
2127 spin_unlock(&osc_ast_guard);
2128 unlock_res_and_lock(lock);
2133 static int osc_set_data_with_check(struct lustre_handle *lockh,
2134 struct ldlm_enqueue_info *einfo)
2136 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2140 set = osc_set_lock_data_with_check(lock, einfo);
2141 LDLM_LOCK_PUT(lock);
2143 CERROR("lockh %p, data %p - client evicted?\n",
2144 lockh, einfo->ei_cbdata);
2148 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2149 ldlm_iterator_t replace, void *data)
2151 struct ldlm_res_id res_id;
2152 struct obd_device *obd = class_exp2obd(exp);
2154 ostid_build_res_name(&lsm->lsm_oi, &res_id);
2155 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2159 /* find any ldlm lock of the inode in osc
2163 static int osc_find_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2164 ldlm_iterator_t replace, void *data)
2166 struct ldlm_res_id res_id;
2167 struct obd_device *obd = class_exp2obd(exp);
2170 ostid_build_res_name(&lsm->lsm_oi, &res_id);
2171 rc = ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2172 if (rc == LDLM_ITER_STOP)
2174 if (rc == LDLM_ITER_CONTINUE)
2179 static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
2180 obd_enqueue_update_f upcall, void *cookie,
2181 __u64 *flags, int agl, int rc)
2183 int intent = *flags & LDLM_FL_HAS_INTENT;
2187 /* The request was created before ldlm_cli_enqueue call. */
2188 if (rc == ELDLM_LOCK_ABORTED) {
2189 struct ldlm_reply *rep;
2190 rep = req_capsule_server_get(&req->rq_pill,
2193 LASSERT(rep != NULL);
2194 rep->lock_policy_res1 =
2195 ptlrpc_status_ntoh(rep->lock_policy_res1);
2196 if (rep->lock_policy_res1)
2197 rc = rep->lock_policy_res1;
2201 if ((intent != 0 && rc == ELDLM_LOCK_ABORTED && agl == 0) ||
2203 *flags |= LDLM_FL_LVB_READY;
2204 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
2205 lvb->lvb_size, lvb->lvb_blocks, lvb->lvb_mtime);
2208 /* Call the update callback. */
2209 rc = (*upcall)(cookie, rc);
2213 static int osc_enqueue_interpret(const struct lu_env *env,
2214 struct ptlrpc_request *req,
2215 struct osc_enqueue_args *aa, int rc)
2217 struct ldlm_lock *lock;
2218 struct lustre_handle handle;
2220 struct ost_lvb *lvb;
2222 __u64 *flags = aa->oa_flags;
2224 /* Make a local copy of a lock handle and a mode, because aa->oa_*
2225 * might be freed anytime after lock upcall has been called. */
2226 lustre_handle_copy(&handle, aa->oa_lockh);
2227 mode = aa->oa_ei->ei_mode;
2229 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2231 lock = ldlm_handle2lock(&handle);
2233 /* Take an additional reference so that a blocking AST that
2234 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2235 * to arrive after an upcall has been executed by
2236 * osc_enqueue_fini(). */
2237 ldlm_lock_addref(&handle, mode);
2239 /* Let cl_lock_state_wait fail with -ERESTARTSYS to unuse sublocks. */
2240 OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_ENQUEUE_HANG, 2);
2242 /* Let CP AST to grant the lock first. */
2243 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
2245 if (aa->oa_agl && rc == ELDLM_LOCK_ABORTED) {
2250 lvb_len = sizeof(*aa->oa_lvb);
2253 /* Complete obtaining the lock procedure. */
2254 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
2255 mode, flags, lvb, lvb_len, &handle, rc);
2256 /* Complete osc stuff. */
2257 rc = osc_enqueue_fini(req, aa->oa_lvb, aa->oa_upcall, aa->oa_cookie,
2258 flags, aa->oa_agl, rc);
2260 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
2262 /* Release the lock for async request. */
2263 if (lustre_handle_is_used(&handle) && rc == ELDLM_OK)
2265 * Releases a reference taken by ldlm_cli_enqueue(), if it is
2266 * not already released by
2267 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
2269 ldlm_lock_decref(&handle, mode);
2271 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
2272 aa->oa_lockh, req, aa);
2273 ldlm_lock_decref(&handle, mode);
2274 LDLM_LOCK_PUT(lock);
2278 struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
2280 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2281 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2282 * other synchronous requests, however keeping some locks and trying to obtain
2283 * others may take a considerable amount of time in a case of ost failure; and
2284 * when other sync requests do not get released lock from a client, the client
2285 * is excluded from the cluster -- such scenarious make the life difficult, so
2286 * release locks just after they are obtained. */
2287 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2288 __u64 *flags, ldlm_policy_data_t *policy,
2289 struct ost_lvb *lvb, int kms_valid,
2290 obd_enqueue_update_f upcall, void *cookie,
2291 struct ldlm_enqueue_info *einfo,
2292 struct lustre_handle *lockh,
2293 struct ptlrpc_request_set *rqset, int async, int agl)
2295 struct obd_device *obd = exp->exp_obd;
2296 struct ptlrpc_request *req = NULL;
2297 int intent = *flags & LDLM_FL_HAS_INTENT;
2298 __u64 match_lvb = (agl != 0 ? 0 : LDLM_FL_LVB_READY);
2303 /* Filesystem lock extents are extended to page boundaries so that
2304 * dealing with the page cache is a little smoother. */
2305 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2306 policy->l_extent.end |= ~CFS_PAGE_MASK;
2309 * kms is not valid when either object is completely fresh (so that no
2310 * locks are cached), or object was evicted. In the latter case cached
2311 * lock cannot be used, because it would prime inode state with
2312 * potentially stale LVB.
2317 /* Next, search for already existing extent locks that will cover us */
2318 /* If we're trying to read, we also search for an existing PW lock. The
2319 * VFS and page cache already protect us locally, so lots of readers/
2320 * writers can share a single PW lock.
2322 * There are problems with conversion deadlocks, so instead of
2323 * converting a read lock to a write lock, we'll just enqueue a new
2326 * At some point we should cancel the read lock instead of making them
2327 * send us a blocking callback, but there are problems with canceling
2328 * locks out from other users right now, too. */
2329 mode = einfo->ei_mode;
2330 if (einfo->ei_mode == LCK_PR)
2332 mode = ldlm_lock_match(obd->obd_namespace, *flags | match_lvb, res_id,
2333 einfo->ei_type, policy, mode, lockh, 0);
2335 struct ldlm_lock *matched = ldlm_handle2lock(lockh);
2337 if ((agl != 0) && !ldlm_is_lvb_ready(matched)) {
2338 /* For AGL, if enqueue RPC is sent but the lock is not
2339 * granted, then skip to process this strpe.
2340 * Return -ECANCELED to tell the caller. */
2341 ldlm_lock_decref(lockh, mode);
2342 LDLM_LOCK_PUT(matched);
2344 } else if (osc_set_lock_data_with_check(matched, einfo)) {
2345 *flags |= LDLM_FL_LVB_READY;
2346 /* addref the lock only if not async requests and PW
2347 * lock is matched whereas we asked for PR. */
2348 if (!rqset && einfo->ei_mode != mode)
2349 ldlm_lock_addref(lockh, LCK_PR);
2351 /* I would like to be able to ASSERT here that
2352 * rss <= kms, but I can't, for reasons which
2353 * are explained in lov_enqueue() */
2356 /* We already have a lock, and it's referenced.
2358 * At this point, the cl_lock::cll_state is CLS_QUEUING,
2359 * AGL upcall may change it to CLS_HELD directly. */
2360 (*upcall)(cookie, ELDLM_OK);
2362 if (einfo->ei_mode != mode)
2363 ldlm_lock_decref(lockh, LCK_PW);
2365 /* For async requests, decref the lock. */
2366 ldlm_lock_decref(lockh, einfo->ei_mode);
2367 LDLM_LOCK_PUT(matched);
2370 ldlm_lock_decref(lockh, mode);
2371 LDLM_LOCK_PUT(matched);
2377 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2378 &RQF_LDLM_ENQUEUE_LVB);
2382 rc = ptlrpc_request_pack(req, LUSTRE_DLM_VERSION, LDLM_ENQUEUE);
2384 ptlrpc_request_free(req);
2388 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
2390 ptlrpc_request_set_replen(req);
2393 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2394 *flags &= ~LDLM_FL_BLOCK_GRANTED;
2396 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
2397 sizeof(*lvb), LVB_T_OST, lockh, async);
2400 struct osc_enqueue_args *aa;
2401 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2402 aa = ptlrpc_req_async_args(req);
2405 aa->oa_flags = flags;
2406 aa->oa_upcall = upcall;
2407 aa->oa_cookie = cookie;
2409 aa->oa_lockh = lockh;
2412 req->rq_interpret_reply =
2413 (ptlrpc_interpterer_t)osc_enqueue_interpret;
2414 if (rqset == PTLRPCD_SET)
2415 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
2417 ptlrpc_set_add_req(rqset, req);
2418 } else if (intent) {
2419 ptlrpc_req_finished(req);
2424 rc = osc_enqueue_fini(req, lvb, upcall, cookie, flags, agl, rc);
2426 ptlrpc_req_finished(req);
2431 int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
2432 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
2433 __u64 *flags, void *data, struct lustre_handle *lockh,
2436 struct obd_device *obd = exp->exp_obd;
2437 __u64 lflags = *flags;
2441 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
2444 /* Filesystem lock extents are extended to page boundaries so that
2445 * dealing with the page cache is a little smoother */
2446 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2447 policy->l_extent.end |= ~CFS_PAGE_MASK;
2449 /* Next, search for already existing extent locks that will cover us */
2450 /* If we're trying to read, we also search for an existing PW lock. The
2451 * VFS and page cache already protect us locally, so lots of readers/
2452 * writers can share a single PW lock. */
2456 rc = ldlm_lock_match(obd->obd_namespace, lflags,
2457 res_id, type, policy, rc, lockh, unref);
2460 if (!osc_set_data_with_check(lockh, data)) {
2461 if (!(lflags & LDLM_FL_TEST_LOCK))
2462 ldlm_lock_decref(lockh, rc);
2466 if (!(lflags & LDLM_FL_TEST_LOCK) && mode != rc) {
2467 ldlm_lock_addref(lockh, LCK_PR);
2468 ldlm_lock_decref(lockh, LCK_PW);
2475 int osc_cancel_base(struct lustre_handle *lockh, __u32 mode)
2479 if (unlikely(mode == LCK_GROUP))
2480 ldlm_lock_decref_and_cancel(lockh, mode);
2482 ldlm_lock_decref(lockh, mode);
2487 static int osc_statfs_interpret(const struct lu_env *env,
2488 struct ptlrpc_request *req,
2489 struct osc_async_args *aa, int rc)
2491 struct obd_statfs *msfs;
2495 /* The request has in fact never been sent
2496 * due to issues at a higher level (LOV).
2497 * Exit immediately since the caller is
2498 * aware of the problem and takes care
2499 * of the clean up */
2502 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
2503 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
2509 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2511 GOTO(out, rc = -EPROTO);
2514 *aa->aa_oi->oi_osfs = *msfs;
2516 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
2520 static int osc_statfs_async(struct obd_export *exp,
2521 struct obd_info *oinfo, __u64 max_age,
2522 struct ptlrpc_request_set *rqset)
2524 struct obd_device *obd = class_exp2obd(exp);
2525 struct ptlrpc_request *req;
2526 struct osc_async_args *aa;
2530 /* We could possibly pass max_age in the request (as an absolute
2531 * timestamp or a "seconds.usec ago") so the target can avoid doing
2532 * extra calls into the filesystem if that isn't necessary (e.g.
2533 * during mount that would help a bit). Having relative timestamps
2534 * is not so great if request processing is slow, while absolute
2535 * timestamps are not ideal because they need time synchronization. */
2536 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
2540 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2542 ptlrpc_request_free(req);
2545 ptlrpc_request_set_replen(req);
2546 req->rq_request_portal = OST_CREATE_PORTAL;
2547 ptlrpc_at_set_req_timeout(req);
2549 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
2550 /* procfs requests not want stat in wait for avoid deadlock */
2551 req->rq_no_resend = 1;
2552 req->rq_no_delay = 1;
2555 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
2556 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2557 aa = ptlrpc_req_async_args(req);
2560 ptlrpc_set_add_req(rqset, req);
2564 static int osc_statfs(const struct lu_env *env, struct obd_export *exp,
2565 struct obd_statfs *osfs, __u64 max_age, __u32 flags)
2567 struct obd_device *obd = class_exp2obd(exp);
2568 struct obd_statfs *msfs;
2569 struct ptlrpc_request *req;
2570 struct obd_import *imp = NULL;
2574 /*Since the request might also come from lprocfs, so we need
2575 *sync this with client_disconnect_export Bug15684*/
2576 down_read(&obd->u.cli.cl_sem);
2577 if (obd->u.cli.cl_import)
2578 imp = class_import_get(obd->u.cli.cl_import);
2579 up_read(&obd->u.cli.cl_sem);
2583 /* We could possibly pass max_age in the request (as an absolute
2584 * timestamp or a "seconds.usec ago") so the target can avoid doing
2585 * extra calls into the filesystem if that isn't necessary (e.g.
2586 * during mount that would help a bit). Having relative timestamps
2587 * is not so great if request processing is slow, while absolute
2588 * timestamps are not ideal because they need time synchronization. */
2589 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
2591 class_import_put(imp);
2596 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
2598 ptlrpc_request_free(req);
2601 ptlrpc_request_set_replen(req);
2602 req->rq_request_portal = OST_CREATE_PORTAL;
2603 ptlrpc_at_set_req_timeout(req);
2605 if (flags & OBD_STATFS_NODELAY) {
2606 /* procfs requests not want stat in wait for avoid deadlock */
2607 req->rq_no_resend = 1;
2608 req->rq_no_delay = 1;
2611 rc = ptlrpc_queue_wait(req);
2615 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
2617 GOTO(out, rc = -EPROTO);
2624 ptlrpc_req_finished(req);
2628 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
2629 void *karg, void *uarg)
2631 struct obd_device *obd = exp->exp_obd;
2632 struct obd_ioctl_data *data = karg;
2636 if (!try_module_get(THIS_MODULE)) {
2637 CERROR("Can't get module. Is it alive?");
2641 case OBD_IOC_CLIENT_RECOVER:
2642 err = ptlrpc_recover_import(obd->u.cli.cl_import,
2643 data->ioc_inlbuf1, 0);
2647 case IOC_OSC_SET_ACTIVE:
2648 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
2651 case OBD_IOC_POLL_QUOTACHECK:
2652 err = osc_quota_poll_check(exp, (struct if_quotacheck *)karg);
2654 case OBD_IOC_PING_TARGET:
2655 err = ptlrpc_obd_ping(obd);
2658 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
2659 cmd, current_comm());
2660 GOTO(out, err = -ENOTTY);
2663 module_put(THIS_MODULE);
2667 static int osc_get_info(const struct lu_env *env, struct obd_export *exp,
2668 obd_count keylen, void *key, __u32 *vallen, void *val,
2669 struct lov_stripe_md *lsm)
2672 if (!vallen || !val)
2675 if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
2676 __u32 *stripe = val;
2677 *vallen = sizeof(*stripe);
2680 } else if (KEY_IS(KEY_LAST_ID)) {
2681 struct ptlrpc_request *req;
2686 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2687 &RQF_OST_GET_INFO_LAST_ID);
2691 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
2692 RCL_CLIENT, keylen);
2693 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
2695 ptlrpc_request_free(req);
2699 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
2700 memcpy(tmp, key, keylen);
2702 req->rq_no_delay = req->rq_no_resend = 1;
2703 ptlrpc_request_set_replen(req);
2704 rc = ptlrpc_queue_wait(req);
2708 reply = req_capsule_server_get(&req->rq_pill, &RMF_OBD_ID);
2710 GOTO(out, rc = -EPROTO);
2712 *((obd_id *)val) = *reply;
2714 ptlrpc_req_finished(req);
2716 } else if (KEY_IS(KEY_FIEMAP)) {
2717 struct ll_fiemap_info_key *fm_key =
2718 (struct ll_fiemap_info_key *)key;
2719 struct ldlm_res_id res_id;
2720 ldlm_policy_data_t policy;
2721 struct lustre_handle lockh;
2722 ldlm_mode_t mode = 0;
2723 struct ptlrpc_request *req;
2724 struct ll_user_fiemap *reply;
2728 if (!(fm_key->fiemap.fm_flags & FIEMAP_FLAG_SYNC))
2731 policy.l_extent.start = fm_key->fiemap.fm_start &
2734 if (OBD_OBJECT_EOF - fm_key->fiemap.fm_length <=
2735 fm_key->fiemap.fm_start + PAGE_CACHE_SIZE - 1)
2736 policy.l_extent.end = OBD_OBJECT_EOF;
2738 policy.l_extent.end = (fm_key->fiemap.fm_start +
2739 fm_key->fiemap.fm_length +
2740 PAGE_CACHE_SIZE - 1) & CFS_PAGE_MASK;
2742 ostid_build_res_name(&fm_key->oa.o_oi, &res_id);
2743 mode = ldlm_lock_match(exp->exp_obd->obd_namespace,
2744 LDLM_FL_BLOCK_GRANTED |
2746 &res_id, LDLM_EXTENT, &policy,
2747 LCK_PR | LCK_PW, &lockh, 0);
2748 if (mode) { /* lock is cached on client */
2749 if (mode != LCK_PR) {
2750 ldlm_lock_addref(&lockh, LCK_PR);
2751 ldlm_lock_decref(&lockh, LCK_PW);
2753 } else { /* no cached lock, needs acquire lock on server side */
2754 fm_key->oa.o_valid |= OBD_MD_FLFLAGS;
2755 fm_key->oa.o_flags |= OBD_FL_SRVLOCK;
2759 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
2760 &RQF_OST_GET_INFO_FIEMAP);
2762 GOTO(drop_lock, rc = -ENOMEM);
2764 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_KEY,
2765 RCL_CLIENT, keylen);
2766 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
2767 RCL_CLIENT, *vallen);
2768 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
2769 RCL_SERVER, *vallen);
2771 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
2773 ptlrpc_request_free(req);
2774 GOTO(drop_lock, rc);
2777 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_KEY);
2778 memcpy(tmp, key, keylen);
2779 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_VAL);
2780 memcpy(tmp, val, *vallen);
2782 ptlrpc_request_set_replen(req);
2783 rc = ptlrpc_queue_wait(req);
2787 reply = req_capsule_server_get(&req->rq_pill, &RMF_FIEMAP_VAL);
2789 GOTO(fini_req, rc = -EPROTO);
2791 memcpy(val, reply, *vallen);
2793 ptlrpc_req_finished(req);
2796 ldlm_lock_decref(&lockh, LCK_PR);
2803 static int osc_set_info_async(const struct lu_env *env, struct obd_export *exp,
2804 obd_count keylen, void *key, obd_count vallen,
2805 void *val, struct ptlrpc_request_set *set)
2807 struct ptlrpc_request *req;
2808 struct obd_device *obd = exp->exp_obd;
2809 struct obd_import *imp = class_exp2cliimp(exp);
2814 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
2816 if (KEY_IS(KEY_CHECKSUM)) {
2817 if (vallen != sizeof(int))
2819 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
2823 if (KEY_IS(KEY_SPTLRPC_CONF)) {
2824 sptlrpc_conf_client_adapt(obd);
2828 if (KEY_IS(KEY_FLUSH_CTX)) {
2829 sptlrpc_import_flush_my_ctx(imp);
2833 if (KEY_IS(KEY_CACHE_SET)) {
2834 struct client_obd *cli = &obd->u.cli;
2836 LASSERT(cli->cl_cache == NULL); /* only once */
2837 cli->cl_cache = (struct cl_client_cache *)val;
2838 atomic_inc(&cli->cl_cache->ccc_users);
2839 cli->cl_lru_left = &cli->cl_cache->ccc_lru_left;
2841 /* add this osc into entity list */
2842 LASSERT(cfs_list_empty(&cli->cl_lru_osc));
2843 spin_lock(&cli->cl_cache->ccc_lru_lock);
2844 cfs_list_add(&cli->cl_lru_osc, &cli->cl_cache->ccc_lru);
2845 spin_unlock(&cli->cl_cache->ccc_lru_lock);
2850 if (KEY_IS(KEY_CACHE_LRU_SHRINK)) {
2851 struct client_obd *cli = &obd->u.cli;
2852 int nr = atomic_read(&cli->cl_lru_in_list) >> 1;
2853 int target = *(int *)val;
2855 nr = osc_lru_shrink(env, cli, min(nr, target), true);
2860 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
2863 /* We pass all other commands directly to OST. Since nobody calls osc
2864 methods directly and everybody is supposed to go through LOV, we
2865 assume lov checked invalid values for us.
2866 The only recognised values so far are evict_by_nid and mds_conn.
2867 Even if something bad goes through, we'd get a -EINVAL from OST
2870 req = ptlrpc_request_alloc(imp, KEY_IS(KEY_GRANT_SHRINK) ?
2871 &RQF_OST_SET_GRANT_INFO :
2876 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
2877 RCL_CLIENT, keylen);
2878 if (!KEY_IS(KEY_GRANT_SHRINK))
2879 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
2880 RCL_CLIENT, vallen);
2881 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
2883 ptlrpc_request_free(req);
2887 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
2888 memcpy(tmp, key, keylen);
2889 tmp = req_capsule_client_get(&req->rq_pill, KEY_IS(KEY_GRANT_SHRINK) ?
2892 memcpy(tmp, val, vallen);
2894 if (KEY_IS(KEY_GRANT_SHRINK)) {
2895 struct osc_grant_args *aa;
2898 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2899 aa = ptlrpc_req_async_args(req);
2902 ptlrpc_req_finished(req);
2905 *oa = ((struct ost_body *)val)->oa;
2907 req->rq_interpret_reply = osc_shrink_grant_interpret;
2910 ptlrpc_request_set_replen(req);
2911 if (!KEY_IS(KEY_GRANT_SHRINK)) {
2912 LASSERT(set != NULL);
2913 ptlrpc_set_add_req(set, req);
2914 ptlrpc_check_set(NULL, set);
2916 ptlrpcd_add_req(req, PDL_POLICY_ROUND, -1);
2921 static int osc_reconnect(const struct lu_env *env,
2922 struct obd_export *exp, struct obd_device *obd,
2923 struct obd_uuid *cluuid,
2924 struct obd_connect_data *data,
2927 struct client_obd *cli = &obd->u.cli;
2929 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
2932 client_obd_list_lock(&cli->cl_loi_list_lock);
2933 data->ocd_grant = (cli->cl_avail_grant + cli->cl_dirty) ?:
2934 2 * cli_brw_size(obd);
2935 lost_grant = cli->cl_lost_grant;
2936 cli->cl_lost_grant = 0;
2937 client_obd_list_unlock(&cli->cl_loi_list_lock);
2939 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
2940 " ocd_grant: %d, lost: %ld.\n", data->ocd_connect_flags,
2941 data->ocd_version, data->ocd_grant, lost_grant);
2947 static int osc_disconnect(struct obd_export *exp)
2949 struct obd_device *obd = class_exp2obd(exp);
2950 struct llog_ctxt *ctxt;
2953 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
2955 if (obd->u.cli.cl_conn_count == 1) {
2956 /* Flush any remaining cancel messages out to the
2958 llog_sync(ctxt, exp, 0);
2960 llog_ctxt_put(ctxt);
2962 CDEBUG(D_HA, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
2966 rc = client_disconnect_export(exp);
2968 * Initially we put del_shrink_grant before disconnect_export, but it
2969 * causes the following problem if setup (connect) and cleanup
2970 * (disconnect) are tangled together.
2971 * connect p1 disconnect p2
2972 * ptlrpc_connect_import
2973 * ............... class_manual_cleanup
2976 * ptlrpc_connect_interrupt
2978 * add this client to shrink list
2980 * Bang! pinger trigger the shrink.
2981 * So the osc should be disconnected from the shrink list, after we
2982 * are sure the import has been destroyed. BUG18662
2984 if (obd->u.cli.cl_import == NULL)
2985 osc_del_shrink_grant(&obd->u.cli);
2989 static int osc_import_event(struct obd_device *obd,
2990 struct obd_import *imp,
2991 enum obd_import_event event)
2993 struct client_obd *cli;
2997 LASSERT(imp->imp_obd == obd);
3000 case IMP_EVENT_DISCON: {
3002 client_obd_list_lock(&cli->cl_loi_list_lock);
3003 cli->cl_avail_grant = 0;
3004 cli->cl_lost_grant = 0;
3005 client_obd_list_unlock(&cli->cl_loi_list_lock);
3008 case IMP_EVENT_INACTIVE: {
3009 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
3012 case IMP_EVENT_INVALIDATE: {
3013 struct ldlm_namespace *ns = obd->obd_namespace;
3017 env = cl_env_get(&refcheck);
3021 /* all pages go to failing rpcs due to the invalid
3023 osc_io_unplug(env, cli, NULL, PDL_POLICY_ROUND);
3025 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3026 cl_env_put(env, &refcheck);
3031 case IMP_EVENT_ACTIVE: {
3032 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
3035 case IMP_EVENT_OCD: {
3036 struct obd_connect_data *ocd = &imp->imp_connect_data;
3038 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3039 osc_init_grant(&obd->u.cli, ocd);
3042 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3043 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3045 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
3048 case IMP_EVENT_DEACTIVATE: {
3049 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_DEACTIVATE, NULL);
3052 case IMP_EVENT_ACTIVATE: {
3053 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVATE, NULL);
3057 CERROR("Unknown import event %d\n", event);
3064 * Determine whether the lock can be canceled before replaying the lock
3065 * during recovery, see bug16774 for detailed information.
3067 * \retval zero the lock can't be canceled
3068 * \retval other ok to cancel
3070 static int osc_cancel_weight(struct ldlm_lock *lock)
3073 * Cancel all unused and granted extent lock.
3075 if (lock->l_resource->lr_type == LDLM_EXTENT &&
3076 lock->l_granted_mode == lock->l_req_mode &&
3077 osc_ldlm_weigh_ast(lock) == 0)
3083 static int brw_queue_work(const struct lu_env *env, void *data)
3085 struct client_obd *cli = data;
3087 CDEBUG(D_CACHE, "Run writeback work for client obd %p.\n", cli);
3089 osc_io_unplug(env, cli, NULL, PDL_POLICY_SAME);
3093 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3095 struct client_obd *cli = &obd->u.cli;
3096 struct obd_type *type;
3101 rc = ptlrpcd_addref();
3105 rc = client_obd_setup(obd, lcfg);
3107 GOTO(out_ptlrpcd, rc);
3109 handler = ptlrpcd_alloc_work(cli->cl_import, brw_queue_work, cli);
3110 if (IS_ERR(handler))
3111 GOTO(out_client_setup, rc = PTR_ERR(handler));
3112 cli->cl_writeback_work = handler;
3114 handler = ptlrpcd_alloc_work(cli->cl_import, lru_queue_work, cli);
3115 if (IS_ERR(handler))
3116 GOTO(out_ptlrpcd_work, rc = PTR_ERR(handler));
3117 cli->cl_lru_work = handler;
3119 rc = osc_quota_setup(obd);
3121 GOTO(out_ptlrpcd_work, rc);
3123 cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
3126 obd->obd_vars = lprocfs_osc_obd_vars;
3128 /* If this is true then both client (osc) and server (osp) are on the
3129 * same node. The osp layer if loaded first will register the osc proc
3130 * directory. In that case this obd_device will be attached its proc
3131 * tree to type->typ_procsym instead of obd->obd_type->typ_procroot. */
3132 type = class_search_type(LUSTRE_OSP_NAME);
3133 if (type && type->typ_procsym) {
3134 obd->obd_proc_entry = lprocfs_seq_register(obd->obd_name,
3136 obd->obd_vars, obd);
3137 if (IS_ERR(obd->obd_proc_entry)) {
3138 rc = PTR_ERR(obd->obd_proc_entry);
3139 CERROR("error %d setting up lprocfs for %s\n", rc,
3141 obd->obd_proc_entry = NULL;
3144 rc = lprocfs_seq_obd_setup(obd);
3147 /* If the basic OSC proc tree construction succeeded then
3148 * lets do the rest. */
3150 lproc_osc_attach_seqstat(obd);
3151 sptlrpc_lprocfs_cliobd_attach(obd);
3152 ptlrpc_lprocfs_register_obd(obd);
3155 /* We need to allocate a few requests more, because
3156 * brw_interpret tries to create new requests before freeing
3157 * previous ones, Ideally we want to have 2x max_rpcs_in_flight
3158 * reserved, but I'm afraid that might be too much wasted RAM
3159 * in fact, so 2 is just my guess and still should work. */
3160 cli->cl_import->imp_rq_pool =
3161 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
3163 ptlrpc_add_rqs_to_pool);
3165 CFS_INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
3166 ns_register_cancel(obd->obd_namespace, osc_cancel_weight);
3170 if (cli->cl_writeback_work != NULL) {
3171 ptlrpcd_destroy_work(cli->cl_writeback_work);
3172 cli->cl_writeback_work = NULL;
3174 if (cli->cl_lru_work != NULL) {
3175 ptlrpcd_destroy_work(cli->cl_lru_work);
3176 cli->cl_lru_work = NULL;
3179 client_obd_cleanup(obd);
3185 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
3191 case OBD_CLEANUP_EARLY: {
3192 struct obd_import *imp;
3193 imp = obd->u.cli.cl_import;
3194 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
3195 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
3196 ptlrpc_deactivate_import(imp);
3197 spin_lock(&imp->imp_lock);
3198 imp->imp_pingable = 0;
3199 spin_unlock(&imp->imp_lock);
3202 case OBD_CLEANUP_EXPORTS: {
3203 struct client_obd *cli = &obd->u.cli;
3205 * for echo client, export may be on zombie list, wait for
3206 * zombie thread to cull it, because cli.cl_import will be
3207 * cleared in client_disconnect_export():
3208 * class_export_destroy() -> obd_cleanup() ->
3209 * echo_device_free() -> echo_client_cleanup() ->
3210 * obd_disconnect() -> osc_disconnect() ->
3211 * client_disconnect_export()
3213 obd_zombie_barrier();
3214 if (cli->cl_writeback_work) {
3215 ptlrpcd_destroy_work(cli->cl_writeback_work);
3216 cli->cl_writeback_work = NULL;
3218 if (cli->cl_lru_work) {
3219 ptlrpcd_destroy_work(cli->cl_lru_work);
3220 cli->cl_lru_work = NULL;
3222 obd_cleanup_client_import(obd);
3223 ptlrpc_lprocfs_unregister_obd(obd);
3224 lprocfs_obd_cleanup(obd);
3225 rc = obd_llog_finish(obd, 0);
3227 CERROR("failed to cleanup llogging subsystems\n");
3234 int osc_cleanup(struct obd_device *obd)
3236 struct client_obd *cli = &obd->u.cli;
3242 if (cli->cl_cache != NULL) {
3243 LASSERT(atomic_read(&cli->cl_cache->ccc_users) > 0);
3244 spin_lock(&cli->cl_cache->ccc_lru_lock);
3245 cfs_list_del_init(&cli->cl_lru_osc);
3246 spin_unlock(&cli->cl_cache->ccc_lru_lock);
3247 cli->cl_lru_left = NULL;
3248 atomic_dec(&cli->cl_cache->ccc_users);
3249 cli->cl_cache = NULL;
3252 /* free memory of osc quota cache */
3253 osc_quota_cleanup(obd);
3255 rc = client_obd_cleanup(obd);
3261 int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
3263 int rc = class_process_proc_seq_param(PARAM_OSC, obd->obd_vars,
3265 return rc > 0 ? 0: rc;
3268 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
3270 return osc_process_config_base(obd, buf);
3273 struct obd_ops osc_obd_ops = {
3274 .o_owner = THIS_MODULE,
3275 .o_setup = osc_setup,
3276 .o_precleanup = osc_precleanup,
3277 .o_cleanup = osc_cleanup,
3278 .o_add_conn = client_import_add_conn,
3279 .o_del_conn = client_import_del_conn,
3280 .o_connect = client_connect_import,
3281 .o_reconnect = osc_reconnect,
3282 .o_disconnect = osc_disconnect,
3283 .o_statfs = osc_statfs,
3284 .o_statfs_async = osc_statfs_async,
3285 .o_unpackmd = osc_unpackmd,
3286 .o_create = osc_create,
3287 .o_destroy = osc_destroy,
3288 .o_getattr = osc_getattr,
3289 .o_getattr_async = osc_getattr_async,
3290 .o_setattr = osc_setattr,
3291 .o_setattr_async = osc_setattr_async,
3292 .o_change_cbdata = osc_change_cbdata,
3293 .o_find_cbdata = osc_find_cbdata,
3294 .o_iocontrol = osc_iocontrol,
3295 .o_get_info = osc_get_info,
3296 .o_set_info_async = osc_set_info_async,
3297 .o_import_event = osc_import_event,
3298 .o_process_config = osc_process_config,
3299 .o_quotactl = osc_quotactl,
3300 .o_quotacheck = osc_quotacheck,
3303 extern struct lu_kmem_descr osc_caches[];
3304 extern spinlock_t osc_ast_guard;
3305 extern struct lock_class_key osc_ast_guard_class;
3307 int __init osc_init(void)
3309 bool enable_proc = true;
3310 struct obd_type *type;
3314 /* print an address of _any_ initialized kernel symbol from this
3315 * module, to allow debugging with gdb that doesn't support data
3316 * symbols from modules.*/
3317 CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
3319 rc = lu_kmem_init(osc_caches);
3323 type = class_search_type(LUSTRE_OSP_NAME);
3324 if (type != NULL && type->typ_procsym != NULL)
3325 enable_proc = false;
3327 rc = class_register_type(&osc_obd_ops, NULL, enable_proc, NULL,
3328 #ifndef HAVE_ONLY_PROCFS_SEQ
3331 LUSTRE_OSC_NAME, &osc_device_type);
3333 lu_kmem_fini(osc_caches);
3337 spin_lock_init(&osc_ast_guard);
3338 lockdep_set_class(&osc_ast_guard, &osc_ast_guard_class);
3344 static void /*__exit*/ osc_exit(void)
3346 class_unregister_type(LUSTRE_OSC_NAME);
3347 lu_kmem_fini(osc_caches);
3350 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
3351 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3352 MODULE_LICENSE("GPL");
3354 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);