1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
6 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 2 only,
10 * as published by the Free Software Foundation.
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License version 2 for more details (a copy is included
16 * in the LICENSE file that accompanied this code).
18 * You should have received a copy of the GNU General Public License
19 * version 2 along with this program; If not, see
20 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
22 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
23 * CA 95054 USA or visit www.sun.com if you need additional information or
29 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
30 * Use is subject to license terms.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
38 # define EXPORT_SYMTAB
40 #define DEBUG_SUBSYSTEM S_OSC
42 #include <libcfs/libcfs.h>
45 # include <liblustre.h>
48 #include <lustre_dlm.h>
49 #include <lustre_net.h>
50 #include <lustre/lustre_user.h>
51 #include <obd_cksum.h>
59 #include <lustre_ha.h>
60 #include <lprocfs_status.h>
61 #include <lustre_log.h>
62 #include <lustre_debug.h>
63 #include <lustre_param.h>
64 #include "osc_internal.h"
66 static quota_interface_t *quota_interface = NULL;
67 extern quota_interface_t osc_quota_interface;
69 static void osc_release_ppga(struct brw_page **ppga, obd_count count);
70 static int brw_interpret(const struct lu_env *env,
71 struct ptlrpc_request *req, void *data, int rc);
72 int osc_cleanup(struct obd_device *obd);
74 /* Pack OSC object metadata for disk storage (LE byte order). */
75 static int osc_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,
76 struct lov_stripe_md *lsm)
81 lmm_size = sizeof(**lmmp);
86 OBD_FREE(*lmmp, lmm_size);
92 OBD_ALLOC(*lmmp, lmm_size);
98 LASSERT(lsm->lsm_object_id);
99 LASSERT_SEQ_IS_MDT(lsm->lsm_object_seq);
100 (*lmmp)->lmm_object_id = cpu_to_le64(lsm->lsm_object_id);
101 (*lmmp)->lmm_object_seq = cpu_to_le64(lsm->lsm_object_seq);
107 /* Unpack OSC object metadata from disk storage (LE byte order). */
108 static int osc_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
109 struct lov_mds_md *lmm, int lmm_bytes)
115 if (lmm_bytes < sizeof (*lmm)) {
116 CERROR("lov_mds_md too small: %d, need %d\n",
117 lmm_bytes, (int)sizeof(*lmm));
120 /* XXX LOV_MAGIC etc check? */
122 if (lmm->lmm_object_id == 0) {
123 CERROR("lov_mds_md: zero lmm_object_id\n");
128 lsm_size = lov_stripe_md_size(1);
132 if (*lsmp != NULL && lmm == NULL) {
133 OBD_FREE((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
134 OBD_FREE(*lsmp, lsm_size);
140 OBD_ALLOC(*lsmp, lsm_size);
143 OBD_ALLOC((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
144 if ((*lsmp)->lsm_oinfo[0] == NULL) {
145 OBD_FREE(*lsmp, lsm_size);
148 loi_init((*lsmp)->lsm_oinfo[0]);
152 /* XXX zero *lsmp? */
153 (*lsmp)->lsm_object_id = le64_to_cpu (lmm->lmm_object_id);
154 (*lsmp)->lsm_object_seq = le64_to_cpu (lmm->lmm_object_seq);
155 LASSERT((*lsmp)->lsm_object_id);
156 LASSERT_SEQ_IS_MDT((*lsmp)->lsm_object_seq);
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(struct obd_export *exp, struct obd_info *oinfo)
265 struct ptlrpc_request *req;
266 struct ost_body *body;
270 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
274 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
275 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
277 ptlrpc_request_free(req);
281 osc_pack_req_body(req, oinfo);
283 ptlrpc_request_set_replen(req);
285 rc = ptlrpc_queue_wait(req);
289 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
291 GOTO(out, rc = -EPROTO);
293 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
294 lustre_get_wire_obdo(oinfo->oi_oa, &body->oa);
296 /* This should really be sent by the OST */
297 oinfo->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
298 oinfo->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
302 ptlrpc_req_finished(req);
306 static int osc_setattr(struct obd_export *exp, struct obd_info *oinfo,
307 struct obd_trans_info *oti)
309 struct ptlrpc_request *req;
310 struct ost_body *body;
314 LASSERT(oinfo->oi_oa->o_valid & OBD_MD_FLGROUP);
316 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
320 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
321 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
323 ptlrpc_request_free(req);
327 osc_pack_req_body(req, oinfo);
329 ptlrpc_request_set_replen(req);
331 rc = ptlrpc_queue_wait(req);
335 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
337 GOTO(out, rc = -EPROTO);
339 lustre_get_wire_obdo(oinfo->oi_oa, &body->oa);
343 ptlrpc_req_finished(req);
347 static int osc_setattr_interpret(const struct lu_env *env,
348 struct ptlrpc_request *req,
349 struct osc_setattr_args *sa, int rc)
351 struct ost_body *body;
357 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
359 GOTO(out, rc = -EPROTO);
361 lustre_get_wire_obdo(sa->sa_oa, &body->oa);
363 rc = sa->sa_upcall(sa->sa_cookie, rc);
367 int osc_setattr_async_base(struct obd_export *exp, struct obd_info *oinfo,
368 struct obd_trans_info *oti,
369 obd_enqueue_update_f upcall, void *cookie,
370 struct ptlrpc_request_set *rqset)
372 struct ptlrpc_request *req;
373 struct osc_setattr_args *sa;
377 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
381 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
382 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
384 ptlrpc_request_free(req);
388 if (oti && oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE)
389 oinfo->oi_oa->o_lcookie = *oti->oti_logcookies;
391 osc_pack_req_body(req, oinfo);
393 ptlrpc_request_set_replen(req);
395 /* do mds to ost setattr asynchronously */
397 /* Do not wait for response. */
398 ptlrpcd_add_req(req, PSCOPE_OTHER);
400 req->rq_interpret_reply =
401 (ptlrpc_interpterer_t)osc_setattr_interpret;
403 CLASSERT (sizeof(*sa) <= sizeof(req->rq_async_args));
404 sa = ptlrpc_req_async_args(req);
405 sa->sa_oa = oinfo->oi_oa;
406 sa->sa_upcall = upcall;
407 sa->sa_cookie = cookie;
409 if (rqset == PTLRPCD_SET)
410 ptlrpcd_add_req(req, PSCOPE_OTHER);
412 ptlrpc_set_add_req(rqset, req);
418 static int osc_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
419 struct obd_trans_info *oti,
420 struct ptlrpc_request_set *rqset)
422 return osc_setattr_async_base(exp, oinfo, oti,
423 oinfo->oi_cb_up, oinfo, rqset);
426 int osc_real_create(struct obd_export *exp, struct obdo *oa,
427 struct lov_stripe_md **ea, struct obd_trans_info *oti)
429 struct ptlrpc_request *req;
430 struct ost_body *body;
431 struct lov_stripe_md *lsm;
440 rc = obd_alloc_memmd(exp, &lsm);
445 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
447 GOTO(out, rc = -ENOMEM);
449 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
451 ptlrpc_request_free(req);
455 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
457 lustre_set_wire_obdo(&body->oa, oa);
459 ptlrpc_request_set_replen(req);
461 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
462 oa->o_flags == OBD_FL_DELORPHAN) {
464 "delorphan from OST integration");
465 /* Don't resend the delorphan req */
466 req->rq_no_resend = req->rq_no_delay = 1;
469 rc = ptlrpc_queue_wait(req);
473 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
475 GOTO(out_req, rc = -EPROTO);
477 lustre_get_wire_obdo(oa, &body->oa);
479 /* This should really be sent by the OST */
480 oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
481 oa->o_valid |= OBD_MD_FLBLKSZ;
483 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
484 * have valid lsm_oinfo data structs, so don't go touching that.
485 * This needs to be fixed in a big way.
487 lsm->lsm_object_id = oa->o_id;
488 lsm->lsm_object_seq = oa->o_seq;
492 oti->oti_transno = lustre_msg_get_transno(req->rq_repmsg);
494 if (oa->o_valid & OBD_MD_FLCOOKIE) {
495 if (!oti->oti_logcookies)
496 oti_alloc_cookies(oti, 1);
497 *oti->oti_logcookies = oa->o_lcookie;
501 CDEBUG(D_HA, "transno: "LPD64"\n",
502 lustre_msg_get_transno(req->rq_repmsg));
504 ptlrpc_req_finished(req);
507 obd_free_memmd(exp, &lsm);
511 int osc_punch_base(struct obd_export *exp, struct obd_info *oinfo,
512 obd_enqueue_update_f upcall, void *cookie,
513 struct ptlrpc_request_set *rqset)
515 struct ptlrpc_request *req;
516 struct osc_setattr_args *sa;
517 struct ost_body *body;
521 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_PUNCH);
525 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
526 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
528 ptlrpc_request_free(req);
531 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
532 ptlrpc_at_set_req_timeout(req);
534 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
536 lustre_set_wire_obdo(&body->oa, oinfo->oi_oa);
537 osc_pack_capa(req, body, oinfo->oi_capa);
539 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, PSCOPE_OTHER);
551 ptlrpc_set_add_req(rqset, req);
556 static int osc_punch(struct obd_export *exp, struct obd_info *oinfo,
557 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(struct obd_export *exp, struct obdo *oa,
568 struct lov_stripe_md *md, obd_size start, obd_size end,
571 struct ptlrpc_request *req;
572 struct ost_body *body;
577 CDEBUG(D_INFO, "oa NULL\n");
581 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
585 osc_set_capa_size(req, &RMF_CAPA1, capa);
586 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
588 ptlrpc_request_free(req);
592 /* overload the size and blocks fields in the oa with start/end */
593 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
595 lustre_set_wire_obdo(&body->oa, oa);
596 body->oa.o_size = start;
597 body->oa.o_blocks = end;
598 body->oa.o_valid |= (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS);
599 osc_pack_capa(req, body, capa);
601 ptlrpc_request_set_replen(req);
603 rc = ptlrpc_queue_wait(req);
607 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
609 GOTO(out, rc = -EPROTO);
611 lustre_get_wire_obdo(oa, &body->oa);
615 ptlrpc_req_finished(req);
619 /* Find and cancel locally locks matched by @mode in the resource found by
620 * @objid. Found locks are added into @cancel list. Returns the amount of
621 * locks added to @cancels list. */
622 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
624 ldlm_mode_t mode, int lock_flags)
626 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
627 struct ldlm_res_id res_id;
628 struct ldlm_resource *res;
632 osc_build_res_name(oa->o_id, oa->o_seq, &res_id);
633 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
637 LDLM_RESOURCE_ADDREF(res);
638 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
639 lock_flags, 0, NULL);
640 LDLM_RESOURCE_DELREF(res);
641 ldlm_resource_putref(res);
645 static int osc_destroy_interpret(const struct lu_env *env,
646 struct ptlrpc_request *req, void *data,
649 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
651 cfs_atomic_dec(&cli->cl_destroy_in_flight);
652 cfs_waitq_signal(&cli->cl_destroy_waitq);
656 static int osc_can_send_destroy(struct client_obd *cli)
658 if (cfs_atomic_inc_return(&cli->cl_destroy_in_flight) <=
659 cli->cl_max_rpcs_in_flight) {
660 /* The destroy request can be sent */
663 if (cfs_atomic_dec_return(&cli->cl_destroy_in_flight) <
664 cli->cl_max_rpcs_in_flight) {
666 * The counter has been modified between the two atomic
669 cfs_waitq_signal(&cli->cl_destroy_waitq);
674 /* Destroy requests can be async always on the client, and we don't even really
675 * care about the return code since the client cannot do anything at all about
677 * When the MDS is unlinking a filename, it saves the file objects into a
678 * recovery llog, and these object records are cancelled when the OST reports
679 * they were destroyed and sync'd to disk (i.e. transaction committed).
680 * If the client dies, or the OST is down when the object should be destroyed,
681 * the records are not cancelled, and when the OST reconnects to the MDS next,
682 * it will retrieve the llog unlink logs and then sends the log cancellation
683 * cookies to the MDS after committing destroy transactions. */
684 static int osc_destroy(struct obd_export *exp, struct obdo *oa,
685 struct lov_stripe_md *ea, struct obd_trans_info *oti,
686 struct obd_export *md_export, void *capa)
688 struct client_obd *cli = &exp->exp_obd->u.cli;
689 struct ptlrpc_request *req;
690 struct ost_body *body;
691 CFS_LIST_HEAD(cancels);
696 CDEBUG(D_INFO, "oa NULL\n");
700 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
701 LDLM_FL_DISCARD_DATA);
703 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
705 ldlm_lock_list_put(&cancels, l_bl_ast, count);
709 osc_set_capa_size(req, &RMF_CAPA1, (struct obd_capa *)capa);
710 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
713 ptlrpc_request_free(req);
717 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
718 ptlrpc_at_set_req_timeout(req);
720 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
721 oa->o_lcookie = *oti->oti_logcookies;
722 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
724 lustre_set_wire_obdo(&body->oa, oa);
726 osc_pack_capa(req, body, (struct obd_capa *)capa);
727 ptlrpc_request_set_replen(req);
729 /* don't throttle destroy RPCs for the MDT */
730 if (!(cli->cl_import->imp_connect_flags_orig & OBD_CONNECT_MDS)) {
731 req->rq_interpret_reply = osc_destroy_interpret;
732 if (!osc_can_send_destroy(cli)) {
733 struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP,
737 * Wait until the number of on-going destroy RPCs drops
738 * under max_rpc_in_flight
740 l_wait_event_exclusive(cli->cl_destroy_waitq,
741 osc_can_send_destroy(cli), &lwi);
745 /* Do not wait for response */
746 ptlrpcd_add_req(req, PSCOPE_OTHER);
750 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
753 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
755 LASSERT(!(oa->o_valid & bits));
758 client_obd_list_lock(&cli->cl_loi_list_lock);
759 oa->o_dirty = cli->cl_dirty;
760 if (cli->cl_dirty - cli->cl_dirty_transit > cli->cl_dirty_max) {
761 CERROR("dirty %lu - %lu > dirty_max %lu\n",
762 cli->cl_dirty, cli->cl_dirty_transit, cli->cl_dirty_max);
764 } else if (cfs_atomic_read(&obd_dirty_pages) -
765 cfs_atomic_read(&obd_dirty_transit_pages) >
766 obd_max_dirty_pages + 1){
767 /* The cfs_atomic_read() allowing the cfs_atomic_inc() are
768 * not covered by a lock thus they may safely race and trip
769 * this CERROR() unless we add in a small fudge factor (+1). */
770 CERROR("dirty %d - %d > system dirty_max %d\n",
771 cfs_atomic_read(&obd_dirty_pages),
772 cfs_atomic_read(&obd_dirty_transit_pages),
773 obd_max_dirty_pages);
775 } else if (cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff) {
776 CERROR("dirty %lu - dirty_max %lu too big???\n",
777 cli->cl_dirty, cli->cl_dirty_max);
780 long max_in_flight = (cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT)*
781 (cli->cl_max_rpcs_in_flight + 1);
782 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
784 oa->o_grant = cli->cl_avail_grant;
785 oa->o_dropped = cli->cl_lost_grant;
786 cli->cl_lost_grant = 0;
787 client_obd_list_unlock(&cli->cl_loi_list_lock);
788 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
789 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
793 static void osc_update_next_shrink(struct client_obd *cli)
795 cli->cl_next_shrink_grant =
796 cfs_time_shift(cli->cl_grant_shrink_interval);
797 CDEBUG(D_CACHE, "next time %ld to shrink grant \n",
798 cli->cl_next_shrink_grant);
801 /* caller must hold loi_list_lock */
802 static void osc_consume_write_grant(struct client_obd *cli,
803 struct brw_page *pga)
805 LASSERT_SPIN_LOCKED(&cli->cl_loi_list_lock.lock);
806 LASSERT(!(pga->flag & OBD_BRW_FROM_GRANT));
807 cfs_atomic_inc(&obd_dirty_pages);
808 cli->cl_dirty += CFS_PAGE_SIZE;
809 cli->cl_avail_grant -= CFS_PAGE_SIZE;
810 pga->flag |= OBD_BRW_FROM_GRANT;
811 CDEBUG(D_CACHE, "using %lu grant credits for brw %p page %p\n",
812 CFS_PAGE_SIZE, pga, pga->pg);
813 LASSERT(cli->cl_avail_grant >= 0);
814 osc_update_next_shrink(cli);
817 /* the companion to osc_consume_write_grant, called when a brw has completed.
818 * must be called with the loi lock held. */
819 static void osc_release_write_grant(struct client_obd *cli,
820 struct brw_page *pga, int sent)
822 int blocksize = cli->cl_import->imp_obd->obd_osfs.os_bsize ? : 4096;
825 LASSERT_SPIN_LOCKED(&cli->cl_loi_list_lock.lock);
826 if (!(pga->flag & OBD_BRW_FROM_GRANT)) {
831 pga->flag &= ~OBD_BRW_FROM_GRANT;
832 cfs_atomic_dec(&obd_dirty_pages);
833 cli->cl_dirty -= CFS_PAGE_SIZE;
834 if (pga->flag & OBD_BRW_NOCACHE) {
835 pga->flag &= ~OBD_BRW_NOCACHE;
836 cfs_atomic_dec(&obd_dirty_transit_pages);
837 cli->cl_dirty_transit -= CFS_PAGE_SIZE;
840 cli->cl_lost_grant += CFS_PAGE_SIZE;
841 CDEBUG(D_CACHE, "lost grant: %lu avail grant: %lu dirty: %lu\n",
842 cli->cl_lost_grant, cli->cl_avail_grant, cli->cl_dirty);
843 } else if (CFS_PAGE_SIZE != blocksize && pga->count != CFS_PAGE_SIZE) {
844 /* For short writes we shouldn't count parts of pages that
845 * span a whole block on the OST side, or our accounting goes
846 * wrong. Should match the code in filter_grant_check. */
847 int offset = pga->off & ~CFS_PAGE_MASK;
848 int count = pga->count + (offset & (blocksize - 1));
849 int end = (offset + pga->count) & (blocksize - 1);
851 count += blocksize - end;
853 cli->cl_lost_grant += CFS_PAGE_SIZE - count;
854 CDEBUG(D_CACHE, "lost %lu grant: %lu avail: %lu dirty: %lu\n",
855 CFS_PAGE_SIZE - count, cli->cl_lost_grant,
856 cli->cl_avail_grant, cli->cl_dirty);
862 static unsigned long rpcs_in_flight(struct client_obd *cli)
864 return cli->cl_r_in_flight + cli->cl_w_in_flight;
867 /* caller must hold loi_list_lock */
868 void osc_wake_cache_waiters(struct client_obd *cli)
871 struct osc_cache_waiter *ocw;
874 cfs_list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
875 /* if we can't dirty more, we must wait until some is written */
876 if ((cli->cl_dirty + CFS_PAGE_SIZE > cli->cl_dirty_max) ||
877 (cfs_atomic_read(&obd_dirty_pages) + 1 >
878 obd_max_dirty_pages)) {
879 CDEBUG(D_CACHE, "no dirty room: dirty: %ld "
880 "osc max %ld, sys max %d\n", cli->cl_dirty,
881 cli->cl_dirty_max, obd_max_dirty_pages);
885 /* if still dirty cache but no grant wait for pending RPCs that
886 * may yet return us some grant before doing sync writes */
887 if (cli->cl_w_in_flight && cli->cl_avail_grant < CFS_PAGE_SIZE) {
888 CDEBUG(D_CACHE, "%u BRW writes in flight, no grant\n",
889 cli->cl_w_in_flight);
893 ocw = cfs_list_entry(l, struct osc_cache_waiter, ocw_entry);
894 cfs_list_del_init(&ocw->ocw_entry);
895 if (cli->cl_avail_grant < CFS_PAGE_SIZE) {
896 /* no more RPCs in flight to return grant, do sync IO */
897 ocw->ocw_rc = -EDQUOT;
898 CDEBUG(D_INODE, "wake oap %p for sync\n", ocw->ocw_oap);
900 osc_consume_write_grant(cli,
901 &ocw->ocw_oap->oap_brw_page);
904 cfs_waitq_signal(&ocw->ocw_waitq);
910 static void __osc_update_grant(struct client_obd *cli, obd_size grant)
912 client_obd_list_lock(&cli->cl_loi_list_lock);
913 cli->cl_avail_grant += grant;
914 client_obd_list_unlock(&cli->cl_loi_list_lock);
917 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
919 if (body->oa.o_valid & OBD_MD_FLGRANT) {
920 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
921 __osc_update_grant(cli, body->oa.o_grant);
925 static int osc_set_info_async(struct obd_export *exp, obd_count keylen,
926 void *key, obd_count vallen, void *val,
927 struct ptlrpc_request_set *set);
929 static int osc_shrink_grant_interpret(const struct lu_env *env,
930 struct ptlrpc_request *req,
933 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
934 struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
935 struct ost_body *body;
938 __osc_update_grant(cli, oa->o_grant);
942 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
944 osc_update_grant(cli, body);
950 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
952 client_obd_list_lock(&cli->cl_loi_list_lock);
953 oa->o_grant = cli->cl_avail_grant / 4;
954 cli->cl_avail_grant -= oa->o_grant;
955 client_obd_list_unlock(&cli->cl_loi_list_lock);
956 if (!(oa->o_valid & OBD_MD_FLFLAGS)) {
957 oa->o_valid |= OBD_MD_FLFLAGS;
960 oa->o_flags |= OBD_FL_SHRINK_GRANT;
961 osc_update_next_shrink(cli);
964 /* Shrink the current grant, either from some large amount to enough for a
965 * full set of in-flight RPCs, or if we have already shrunk to that limit
966 * then to enough for a single RPC. This avoids keeping more grant than
967 * needed, and avoids shrinking the grant piecemeal. */
968 static int osc_shrink_grant(struct client_obd *cli)
970 long target = (cli->cl_max_rpcs_in_flight + 1) *
971 cli->cl_max_pages_per_rpc;
973 client_obd_list_lock(&cli->cl_loi_list_lock);
974 if (cli->cl_avail_grant <= target)
975 target = cli->cl_max_pages_per_rpc;
976 client_obd_list_unlock(&cli->cl_loi_list_lock);
978 return osc_shrink_grant_to_target(cli, target);
981 int osc_shrink_grant_to_target(struct client_obd *cli, long target)
984 struct ost_body *body;
987 client_obd_list_lock(&cli->cl_loi_list_lock);
988 /* Don't shrink if we are already above or below the desired limit
989 * We don't want to shrink below a single RPC, as that will negatively
990 * impact block allocation and long-term performance. */
991 if (target < cli->cl_max_pages_per_rpc)
992 target = cli->cl_max_pages_per_rpc;
994 if (target >= cli->cl_avail_grant) {
995 client_obd_list_unlock(&cli->cl_loi_list_lock);
998 client_obd_list_unlock(&cli->cl_loi_list_lock);
1000 OBD_ALLOC_PTR(body);
1004 osc_announce_cached(cli, &body->oa, 0);
1006 client_obd_list_lock(&cli->cl_loi_list_lock);
1007 body->oa.o_grant = cli->cl_avail_grant - target;
1008 cli->cl_avail_grant = target;
1009 client_obd_list_unlock(&cli->cl_loi_list_lock);
1010 if (!(body->oa.o_valid & OBD_MD_FLFLAGS)) {
1011 body->oa.o_valid |= OBD_MD_FLFLAGS;
1012 body->oa.o_flags = 0;
1014 body->oa.o_flags |= OBD_FL_SHRINK_GRANT;
1015 osc_update_next_shrink(cli);
1017 rc = osc_set_info_async(cli->cl_import->imp_obd->obd_self_export,
1018 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
1019 sizeof(*body), body, NULL);
1021 __osc_update_grant(cli, body->oa.o_grant);
1026 #define GRANT_SHRINK_LIMIT PTLRPC_MAX_BRW_SIZE
1027 static int osc_should_shrink_grant(struct client_obd *client)
1029 cfs_time_t time = cfs_time_current();
1030 cfs_time_t next_shrink = client->cl_next_shrink_grant;
1032 if ((client->cl_import->imp_connect_data.ocd_connect_flags &
1033 OBD_CONNECT_GRANT_SHRINK) == 0)
1036 if (cfs_time_aftereq(time, next_shrink - 5 * CFS_TICK)) {
1037 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
1038 client->cl_avail_grant > GRANT_SHRINK_LIMIT)
1041 osc_update_next_shrink(client);
1046 static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
1048 struct client_obd *client;
1050 cfs_list_for_each_entry(client, &item->ti_obd_list,
1051 cl_grant_shrink_list) {
1052 if (osc_should_shrink_grant(client))
1053 osc_shrink_grant(client);
1058 static int osc_add_shrink_grant(struct client_obd *client)
1062 rc = ptlrpc_add_timeout_client(client->cl_grant_shrink_interval,
1064 osc_grant_shrink_grant_cb, NULL,
1065 &client->cl_grant_shrink_list);
1067 CERROR("add grant client %s error %d\n",
1068 client->cl_import->imp_obd->obd_name, rc);
1071 CDEBUG(D_CACHE, "add grant client %s \n",
1072 client->cl_import->imp_obd->obd_name);
1073 osc_update_next_shrink(client);
1077 static int osc_del_shrink_grant(struct client_obd *client)
1079 return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
1083 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1086 * ocd_grant is the total grant amount we're expect to hold: if we've
1087 * been evicted, it's the new avail_grant amount, cl_dirty will drop
1088 * to 0 as inflight RPCs fail out; otherwise, it's avail_grant + dirty.
1090 * race is tolerable here: if we're evicted, but imp_state already
1091 * left EVICTED state, then cl_dirty must be 0 already.
1093 client_obd_list_lock(&cli->cl_loi_list_lock);
1094 if (cli->cl_import->imp_state == LUSTRE_IMP_EVICTED)
1095 cli->cl_avail_grant = ocd->ocd_grant;
1097 cli->cl_avail_grant = ocd->ocd_grant - cli->cl_dirty;
1099 if (cli->cl_avail_grant < 0) {
1100 CWARN("%s: available grant < 0, the OSS is probably not running"
1101 " with patch from bug20278 (%ld) \n",
1102 cli->cl_import->imp_obd->obd_name, cli->cl_avail_grant);
1103 /* workaround for 1.6 servers which do not have
1104 * the patch from bug20278 */
1105 cli->cl_avail_grant = ocd->ocd_grant;
1108 client_obd_list_unlock(&cli->cl_loi_list_lock);
1110 CDEBUG(D_CACHE, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld \n",
1111 cli->cl_import->imp_obd->obd_name,
1112 cli->cl_avail_grant, cli->cl_lost_grant);
1114 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
1115 cfs_list_empty(&cli->cl_grant_shrink_list))
1116 osc_add_shrink_grant(cli);
1119 /* We assume that the reason this OSC got a short read is because it read
1120 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1121 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1122 * this stripe never got written at or beyond this stripe offset yet. */
1123 static void handle_short_read(int nob_read, obd_count page_count,
1124 struct brw_page **pga)
1129 /* skip bytes read OK */
1130 while (nob_read > 0) {
1131 LASSERT (page_count > 0);
1133 if (pga[i]->count > nob_read) {
1134 /* EOF inside this page */
1135 ptr = cfs_kmap(pga[i]->pg) +
1136 (pga[i]->off & ~CFS_PAGE_MASK);
1137 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1138 cfs_kunmap(pga[i]->pg);
1144 nob_read -= pga[i]->count;
1149 /* zero remaining pages */
1150 while (page_count-- > 0) {
1151 ptr = cfs_kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
1152 memset(ptr, 0, pga[i]->count);
1153 cfs_kunmap(pga[i]->pg);
1158 static int check_write_rcs(struct ptlrpc_request *req,
1159 int requested_nob, int niocount,
1160 obd_count page_count, struct brw_page **pga)
1165 remote_rcs = req_capsule_server_sized_get(&req->rq_pill, &RMF_RCS,
1166 sizeof(*remote_rcs) *
1168 if (remote_rcs == NULL) {
1169 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1173 /* return error if any niobuf was in error */
1174 for (i = 0; i < niocount; i++) {
1175 if (remote_rcs[i] < 0)
1176 return(remote_rcs[i]);
1178 if (remote_rcs[i] != 0) {
1179 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1180 i, remote_rcs[i], req);
1185 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
1186 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1187 req->rq_bulk->bd_nob_transferred, requested_nob);
1194 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1196 if (p1->flag != p2->flag) {
1197 unsigned mask = ~(OBD_BRW_FROM_GRANT|
1198 OBD_BRW_NOCACHE|OBD_BRW_SYNC|OBD_BRW_ASYNC);
1200 /* warn if we try to combine flags that we don't know to be
1201 * safe to combine */
1202 if ((p1->flag & mask) != (p2->flag & mask))
1203 CERROR("is it ok to have flags 0x%x and 0x%x in the "
1204 "same brw?\n", p1->flag, p2->flag);
1208 return (p1->off + p1->count == p2->off);
1211 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
1212 struct brw_page **pga, int opc,
1213 cksum_type_t cksum_type)
1218 LASSERT (pg_count > 0);
1219 cksum = init_checksum(cksum_type);
1220 while (nob > 0 && pg_count > 0) {
1221 unsigned char *ptr = cfs_kmap(pga[i]->pg);
1222 int off = pga[i]->off & ~CFS_PAGE_MASK;
1223 int count = pga[i]->count > nob ? nob : pga[i]->count;
1225 /* corrupt the data before we compute the checksum, to
1226 * simulate an OST->client data error */
1227 if (i == 0 && opc == OST_READ &&
1228 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE))
1229 memcpy(ptr + off, "bad1", min(4, nob));
1230 cksum = compute_checksum(cksum, ptr + off, count, cksum_type);
1231 cfs_kunmap(pga[i]->pg);
1232 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d checksum %x\n",
1235 nob -= pga[i]->count;
1239 /* For sending we only compute the wrong checksum instead
1240 * of corrupting the data so it is still correct on a redo */
1241 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1247 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
1248 struct lov_stripe_md *lsm, obd_count page_count,
1249 struct brw_page **pga,
1250 struct ptlrpc_request **reqp,
1251 struct obd_capa *ocapa, int reserve)
1253 struct ptlrpc_request *req;
1254 struct ptlrpc_bulk_desc *desc;
1255 struct ost_body *body;
1256 struct obd_ioobj *ioobj;
1257 struct niobuf_remote *niobuf;
1258 int niocount, i, requested_nob, opc, rc;
1259 struct osc_brw_async_args *aa;
1260 struct req_capsule *pill;
1261 struct brw_page *pg_prev;
1264 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1265 RETURN(-ENOMEM); /* Recoverable */
1266 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1267 RETURN(-EINVAL); /* Fatal */
1269 if ((cmd & OBD_BRW_WRITE) != 0) {
1271 req = ptlrpc_request_alloc_pool(cli->cl_import,
1272 cli->cl_import->imp_rq_pool,
1273 &RQF_OST_BRW_WRITE);
1276 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW_READ);
1281 for (niocount = i = 1; i < page_count; i++) {
1282 if (!can_merge_pages(pga[i - 1], pga[i]))
1286 pill = &req->rq_pill;
1287 req_capsule_set_size(pill, &RMF_OBD_IOOBJ, RCL_CLIENT,
1289 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1290 niocount * sizeof(*niobuf));
1291 osc_set_capa_size(req, &RMF_CAPA1, ocapa);
1293 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1295 ptlrpc_request_free(req);
1298 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
1299 ptlrpc_at_set_req_timeout(req);
1301 if (opc == OST_WRITE)
1302 desc = ptlrpc_prep_bulk_imp(req, page_count,
1303 BULK_GET_SOURCE, OST_BULK_PORTAL);
1305 desc = ptlrpc_prep_bulk_imp(req, page_count,
1306 BULK_PUT_SINK, OST_BULK_PORTAL);
1309 GOTO(out, rc = -ENOMEM);
1310 /* NB request now owns desc and will free it when it gets freed */
1312 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1313 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1314 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1315 LASSERT(body != NULL && ioobj != NULL && niobuf != NULL);
1317 lustre_set_wire_obdo(&body->oa, oa);
1319 obdo_to_ioobj(oa, ioobj);
1320 ioobj->ioo_bufcnt = niocount;
1321 osc_pack_capa(req, body, ocapa);
1322 LASSERT (page_count > 0);
1324 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1325 struct brw_page *pg = pga[i];
1327 LASSERT(pg->count > 0);
1328 LASSERTF((pg->off & ~CFS_PAGE_MASK) + pg->count <= CFS_PAGE_SIZE,
1329 "i: %d pg: %p off: "LPU64", count: %u\n", i, pg,
1330 pg->off, pg->count);
1332 LASSERTF(i == 0 || pg->off > pg_prev->off,
1333 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1334 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1336 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1337 pg_prev->pg, page_private(pg_prev->pg),
1338 pg_prev->pg->index, pg_prev->off);
1340 LASSERTF(i == 0 || pg->off > pg_prev->off,
1341 "i %d p_c %u\n", i, page_count);
1343 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1344 (pg->flag & OBD_BRW_SRVLOCK));
1346 ptlrpc_prep_bulk_page(desc, pg->pg, pg->off & ~CFS_PAGE_MASK,
1348 requested_nob += pg->count;
1350 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1352 niobuf->len += pg->count;
1354 niobuf->offset = pg->off;
1355 niobuf->len = pg->count;
1356 niobuf->flags = pg->flag;
1361 LASSERTF((void *)(niobuf - niocount) ==
1362 req_capsule_client_get(&req->rq_pill, &RMF_NIOBUF_REMOTE),
1363 "want %p - real %p\n", req_capsule_client_get(&req->rq_pill,
1364 &RMF_NIOBUF_REMOTE), (void *)(niobuf - niocount));
1366 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1367 if (osc_should_shrink_grant(cli))
1368 osc_shrink_grant_local(cli, &body->oa);
1370 /* size[REQ_REC_OFF] still sizeof (*body) */
1371 if (opc == OST_WRITE) {
1372 if (unlikely(cli->cl_checksum) &&
1373 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1374 /* store cl_cksum_type in a local variable since
1375 * it can be changed via lprocfs */
1376 cksum_type_t cksum_type = cli->cl_cksum_type;
1378 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0) {
1379 oa->o_flags &= OBD_FL_LOCAL_MASK;
1380 body->oa.o_flags = 0;
1382 body->oa.o_flags |= cksum_type_pack(cksum_type);
1383 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1384 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1388 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1390 /* save this in 'oa', too, for later checking */
1391 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1392 oa->o_flags |= cksum_type_pack(cksum_type);
1394 /* clear out the checksum flag, in case this is a
1395 * resend but cl_checksum is no longer set. b=11238 */
1396 oa->o_valid &= ~OBD_MD_FLCKSUM;
1398 oa->o_cksum = body->oa.o_cksum;
1399 /* 1 RC per niobuf */
1400 req_capsule_set_size(pill, &RMF_RCS, RCL_SERVER,
1401 sizeof(__u32) * niocount);
1403 if (unlikely(cli->cl_checksum) &&
1404 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1405 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1406 body->oa.o_flags = 0;
1407 body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
1408 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1411 ptlrpc_request_set_replen(req);
1413 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1414 aa = ptlrpc_req_async_args(req);
1416 aa->aa_requested_nob = requested_nob;
1417 aa->aa_nio_count = niocount;
1418 aa->aa_page_count = page_count;
1422 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1423 if (ocapa && reserve)
1424 aa->aa_ocapa = capa_get(ocapa);
1430 ptlrpc_req_finished(req);
1434 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1435 __u32 client_cksum, __u32 server_cksum, int nob,
1436 obd_count page_count, struct brw_page **pga,
1437 cksum_type_t client_cksum_type)
1441 cksum_type_t cksum_type;
1443 if (server_cksum == client_cksum) {
1444 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1448 /* If this is mmaped file - it can be changed at any time */
1449 if (oa->o_valid & OBD_MD_FLFLAGS && oa->o_flags & OBD_FL_MMAP)
1452 if (oa->o_valid & OBD_MD_FLFLAGS)
1453 cksum_type = cksum_type_unpack(oa->o_flags);
1455 cksum_type = OBD_CKSUM_CRC32;
1457 new_cksum = osc_checksum_bulk(nob, page_count, pga, OST_WRITE,
1460 if (cksum_type != client_cksum_type)
1461 msg = "the server did not use the checksum type specified in "
1462 "the original request - likely a protocol problem";
1463 else if (new_cksum == server_cksum)
1464 msg = "changed on the client after we checksummed it - "
1465 "likely false positive due to mmap IO (bug 11742)";
1466 else if (new_cksum == client_cksum)
1467 msg = "changed in transit before arrival at OST";
1469 msg = "changed in transit AND doesn't match the original - "
1470 "likely false positive due to mmap IO (bug 11742)";
1472 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inode "DFID
1473 " object "LPU64"/"LPU64" extent ["LPU64"-"LPU64"]\n",
1474 msg, libcfs_nid2str(peer->nid),
1475 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_seq : (__u64)0,
1476 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_oid : 0,
1477 oa->o_valid & OBD_MD_FLFID ? oa->o_parent_ver : 0,
1479 oa->o_valid & OBD_MD_FLGROUP ? oa->o_seq : (__u64)0,
1481 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1482 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1483 "client csum now %x\n", client_cksum, client_cksum_type,
1484 server_cksum, cksum_type, new_cksum);
1488 /* Note rc enters this function as number of bytes transferred */
1489 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1491 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1492 const lnet_process_id_t *peer =
1493 &req->rq_import->imp_connection->c_peer;
1494 struct client_obd *cli = aa->aa_cli;
1495 struct ost_body *body;
1496 __u32 client_cksum = 0;
1499 if (rc < 0 && rc != -EDQUOT) {
1500 DEBUG_REQ(D_INFO, req, "Failed request with rc = %d\n", rc);
1504 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1505 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
1507 DEBUG_REQ(D_INFO, req, "Can't unpack body\n");
1511 #ifdef HAVE_QUOTA_SUPPORT
1512 /* set/clear over quota flag for a uid/gid */
1513 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1514 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA)) {
1515 unsigned int qid[MAXQUOTAS] = { body->oa.o_uid, body->oa.o_gid };
1517 CDEBUG(D_QUOTA, "setdq for [%u %u] with valid "LPX64", flags %x\n",
1518 body->oa.o_uid, body->oa.o_gid, body->oa.o_valid,
1520 lquota_setdq(quota_interface, cli, qid, body->oa.o_valid,
1525 osc_update_grant(cli, body);
1530 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1531 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1533 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1535 CERROR("Unexpected +ve rc %d\n", rc);
1538 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1540 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1543 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1544 check_write_checksum(&body->oa, peer, client_cksum,
1545 body->oa.o_cksum, aa->aa_requested_nob,
1546 aa->aa_page_count, aa->aa_ppga,
1547 cksum_type_unpack(aa->aa_oa->o_flags)))
1550 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1551 aa->aa_page_count, aa->aa_ppga);
1555 /* The rest of this function executes only for OST_READs */
1557 /* if unwrap_bulk failed, return -EAGAIN to retry */
1558 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1560 GOTO(out, rc = -EAGAIN);
1562 if (rc > aa->aa_requested_nob) {
1563 CERROR("Unexpected rc %d (%d requested)\n", rc,
1564 aa->aa_requested_nob);
1568 if (rc != req->rq_bulk->bd_nob_transferred) {
1569 CERROR ("Unexpected rc %d (%d transferred)\n",
1570 rc, req->rq_bulk->bd_nob_transferred);
1574 if (rc < aa->aa_requested_nob)
1575 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1577 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
1578 static int cksum_counter;
1579 __u32 server_cksum = body->oa.o_cksum;
1582 cksum_type_t cksum_type;
1584 if (body->oa.o_valid & OBD_MD_FLFLAGS)
1585 cksum_type = cksum_type_unpack(body->oa.o_flags);
1587 cksum_type = OBD_CKSUM_CRC32;
1588 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1589 aa->aa_ppga, OST_READ,
1592 if (peer->nid == req->rq_bulk->bd_sender) {
1596 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1599 if (server_cksum == ~0 && rc > 0) {
1600 CERROR("Protocol error: server %s set the 'checksum' "
1601 "bit, but didn't send a checksum. Not fatal, "
1602 "but please notify on http://bugzilla.lustre.org/\n",
1603 libcfs_nid2str(peer->nid));
1604 } else if (server_cksum != client_cksum) {
1605 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1606 "%s%s%s inode "DFID" object "
1607 LPU64"/"LPU64" extent "
1608 "["LPU64"-"LPU64"]\n",
1609 req->rq_import->imp_obd->obd_name,
1610 libcfs_nid2str(peer->nid),
1612 body->oa.o_valid & OBD_MD_FLFID ?
1613 body->oa.o_parent_seq : (__u64)0,
1614 body->oa.o_valid & OBD_MD_FLFID ?
1615 body->oa.o_parent_oid : 0,
1616 body->oa.o_valid & OBD_MD_FLFID ?
1617 body->oa.o_parent_ver : 0,
1619 body->oa.o_valid & OBD_MD_FLGROUP ?
1620 body->oa.o_seq : (__u64)0,
1621 aa->aa_ppga[0]->off,
1622 aa->aa_ppga[aa->aa_page_count-1]->off +
1623 aa->aa_ppga[aa->aa_page_count-1]->count -
1625 CERROR("client %x, server %x, cksum_type %x\n",
1626 client_cksum, server_cksum, cksum_type);
1628 aa->aa_oa->o_cksum = client_cksum;
1632 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1635 } else if (unlikely(client_cksum)) {
1636 static int cksum_missed;
1639 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1640 CERROR("Checksum %u requested from %s but not sent\n",
1641 cksum_missed, libcfs_nid2str(peer->nid));
1647 lustre_get_wire_obdo(aa->aa_oa, &body->oa);
1652 static int osc_brw_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1653 struct lov_stripe_md *lsm,
1654 obd_count page_count, struct brw_page **pga,
1655 struct obd_capa *ocapa)
1657 struct ptlrpc_request *req;
1661 struct l_wait_info lwi;
1665 cfs_waitq_init(&waitq);
1668 rc = osc_brw_prep_request(cmd, &exp->exp_obd->u.cli, oa, lsm,
1669 page_count, pga, &req, ocapa, 0);
1673 rc = ptlrpc_queue_wait(req);
1675 if (rc == -ETIMEDOUT && req->rq_resend) {
1676 DEBUG_REQ(D_HA, req, "BULK TIMEOUT");
1677 ptlrpc_req_finished(req);
1681 rc = osc_brw_fini_request(req, rc);
1683 ptlrpc_req_finished(req);
1684 if (osc_recoverable_error(rc)) {
1686 if (!osc_should_resend(resends, &exp->exp_obd->u.cli)) {
1687 CERROR("too many resend retries, returning error\n");
1691 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(resends), NULL, NULL, NULL);
1692 l_wait_event(waitq, 0, &lwi);
1700 int osc_brw_redo_request(struct ptlrpc_request *request,
1701 struct osc_brw_async_args *aa)
1703 struct ptlrpc_request *new_req;
1704 struct ptlrpc_request_set *set = request->rq_set;
1705 struct osc_brw_async_args *new_aa;
1706 struct osc_async_page *oap;
1710 if (!osc_should_resend(aa->aa_resends, aa->aa_cli)) {
1711 CERROR("too many resent retries, returning error\n");
1715 DEBUG_REQ(D_ERROR, request, "redo for recoverable error");
1717 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
1718 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1719 aa->aa_cli, aa->aa_oa,
1720 NULL /* lsm unused by osc currently */,
1721 aa->aa_page_count, aa->aa_ppga,
1722 &new_req, aa->aa_ocapa, 0);
1726 client_obd_list_lock(&aa->aa_cli->cl_loi_list_lock);
1728 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1729 if (oap->oap_request != NULL) {
1730 LASSERTF(request == oap->oap_request,
1731 "request %p != oap_request %p\n",
1732 request, oap->oap_request);
1733 if (oap->oap_interrupted) {
1734 client_obd_list_unlock(&aa->aa_cli->cl_loi_list_lock);
1735 ptlrpc_req_finished(new_req);
1740 /* New request takes over pga and oaps from old request.
1741 * Note that copying a list_head doesn't work, need to move it... */
1743 new_req->rq_interpret_reply = request->rq_interpret_reply;
1744 new_req->rq_async_args = request->rq_async_args;
1745 new_req->rq_sent = cfs_time_current_sec() + aa->aa_resends;
1747 new_aa = ptlrpc_req_async_args(new_req);
1749 CFS_INIT_LIST_HEAD(&new_aa->aa_oaps);
1750 cfs_list_splice(&aa->aa_oaps, &new_aa->aa_oaps);
1751 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1753 cfs_list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1754 if (oap->oap_request) {
1755 ptlrpc_req_finished(oap->oap_request);
1756 oap->oap_request = ptlrpc_request_addref(new_req);
1760 new_aa->aa_ocapa = aa->aa_ocapa;
1761 aa->aa_ocapa = NULL;
1763 /* use ptlrpc_set_add_req is safe because interpret functions work
1764 * in check_set context. only one way exist with access to request
1765 * from different thread got -EINTR - this way protected with
1766 * cl_loi_list_lock */
1767 ptlrpc_set_add_req(set, new_req);
1769 client_obd_list_unlock(&aa->aa_cli->cl_loi_list_lock);
1771 DEBUG_REQ(D_INFO, new_req, "new request");
1776 * ugh, we want disk allocation on the target to happen in offset order. we'll
1777 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1778 * fine for our small page arrays and doesn't require allocation. its an
1779 * insertion sort that swaps elements that are strides apart, shrinking the
1780 * stride down until its '1' and the array is sorted.
1782 static void sort_brw_pages(struct brw_page **array, int num)
1785 struct brw_page *tmp;
1789 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1794 for (i = stride ; i < num ; i++) {
1797 while (j >= stride && array[j - stride]->off > tmp->off) {
1798 array[j] = array[j - stride];
1803 } while (stride > 1);
1806 static obd_count max_unfragmented_pages(struct brw_page **pg, obd_count pages)
1812 LASSERT (pages > 0);
1813 offset = pg[i]->off & ~CFS_PAGE_MASK;
1817 if (pages == 0) /* that's all */
1820 if (offset + pg[i]->count < CFS_PAGE_SIZE)
1821 return count; /* doesn't end on page boundary */
1824 offset = pg[i]->off & ~CFS_PAGE_MASK;
1825 if (offset != 0) /* doesn't start on page boundary */
1832 static struct brw_page **osc_build_ppga(struct brw_page *pga, obd_count count)
1834 struct brw_page **ppga;
1837 OBD_ALLOC(ppga, sizeof(*ppga) * count);
1841 for (i = 0; i < count; i++)
1846 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1848 LASSERT(ppga != NULL);
1849 OBD_FREE(ppga, sizeof(*ppga) * count);
1852 static int osc_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
1853 obd_count page_count, struct brw_page *pga,
1854 struct obd_trans_info *oti)
1856 struct obdo *saved_oa = NULL;
1857 struct brw_page **ppga, **orig;
1858 struct obd_import *imp = class_exp2cliimp(exp);
1859 struct client_obd *cli;
1860 int rc, page_count_orig;
1863 LASSERT((imp != NULL) && (imp->imp_obd != NULL));
1864 cli = &imp->imp_obd->u.cli;
1866 if (cmd & OBD_BRW_CHECK) {
1867 /* The caller just wants to know if there's a chance that this
1868 * I/O can succeed */
1870 if (imp->imp_invalid)
1875 /* test_brw with a failed create can trip this, maybe others. */
1876 LASSERT(cli->cl_max_pages_per_rpc);
1880 orig = ppga = osc_build_ppga(pga, page_count);
1883 page_count_orig = page_count;
1885 sort_brw_pages(ppga, page_count);
1886 while (page_count) {
1887 obd_count pages_per_brw;
1889 if (page_count > cli->cl_max_pages_per_rpc)
1890 pages_per_brw = cli->cl_max_pages_per_rpc;
1892 pages_per_brw = page_count;
1894 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1896 if (saved_oa != NULL) {
1897 /* restore previously saved oa */
1898 *oinfo->oi_oa = *saved_oa;
1899 } else if (page_count > pages_per_brw) {
1900 /* save a copy of oa (brw will clobber it) */
1901 OBDO_ALLOC(saved_oa);
1902 if (saved_oa == NULL)
1903 GOTO(out, rc = -ENOMEM);
1904 *saved_oa = *oinfo->oi_oa;
1907 rc = osc_brw_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1908 pages_per_brw, ppga, oinfo->oi_capa);
1913 page_count -= pages_per_brw;
1914 ppga += pages_per_brw;
1918 osc_release_ppga(orig, page_count_orig);
1920 if (saved_oa != NULL)
1921 OBDO_FREE(saved_oa);
1926 /* The companion to osc_enter_cache(), called when @oap is no longer part of
1927 * the dirty accounting. Writeback completes or truncate happens before
1928 * writing starts. Must be called with the loi lock held. */
1929 static void osc_exit_cache(struct client_obd *cli, struct osc_async_page *oap,
1932 osc_release_write_grant(cli, &oap->oap_brw_page, sent);
1936 /* This maintains the lists of pending pages to read/write for a given object
1937 * (lop). This is used by osc_check_rpcs->osc_next_loi() and loi_list_maint()
1938 * to quickly find objects that are ready to send an RPC. */
1939 static int lop_makes_rpc(struct client_obd *cli, struct loi_oap_pages *lop,
1945 if (lop->lop_num_pending == 0)
1948 /* if we have an invalid import we want to drain the queued pages
1949 * by forcing them through rpcs that immediately fail and complete
1950 * the pages. recovery relies on this to empty the queued pages
1951 * before canceling the locks and evicting down the llite pages */
1952 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
1955 /* stream rpcs in queue order as long as as there is an urgent page
1956 * queued. this is our cheap solution for good batching in the case
1957 * where writepage marks some random page in the middle of the file
1958 * as urgent because of, say, memory pressure */
1959 if (!cfs_list_empty(&lop->lop_urgent)) {
1960 CDEBUG(D_CACHE, "urgent request forcing RPC\n");
1963 /* fire off rpcs when we have 'optimal' rpcs as tuned for the wire. */
1964 optimal = cli->cl_max_pages_per_rpc;
1965 if (cmd & OBD_BRW_WRITE) {
1966 /* trigger a write rpc stream as long as there are dirtiers
1967 * waiting for space. as they're waiting, they're not going to
1968 * create more pages to coalesce with what's waiting.. */
1969 if (!cfs_list_empty(&cli->cl_cache_waiters)) {
1970 CDEBUG(D_CACHE, "cache waiters forcing RPC\n");
1973 /* +16 to avoid triggering rpcs that would want to include pages
1974 * that are being queued but which can't be made ready until
1975 * the queuer finishes with the page. this is a wart for
1976 * llite::commit_write() */
1979 if (lop->lop_num_pending >= optimal)
1985 static int lop_makes_hprpc(struct loi_oap_pages *lop)
1987 struct osc_async_page *oap;
1990 if (cfs_list_empty(&lop->lop_urgent))
1993 oap = cfs_list_entry(lop->lop_urgent.next,
1994 struct osc_async_page, oap_urgent_item);
1996 if (oap->oap_async_flags & ASYNC_HP) {
1997 CDEBUG(D_CACHE, "hp request forcing RPC\n");
2004 static void on_list(cfs_list_t *item, cfs_list_t *list,
2007 if (cfs_list_empty(item) && should_be_on)
2008 cfs_list_add_tail(item, list);
2009 else if (!cfs_list_empty(item) && !should_be_on)
2010 cfs_list_del_init(item);
2013 /* maintain the loi's cli list membership invariants so that osc_send_oap_rpc
2014 * can find pages to build into rpcs quickly */
2015 void loi_list_maint(struct client_obd *cli, struct lov_oinfo *loi)
2017 if (lop_makes_hprpc(&loi->loi_write_lop) ||
2018 lop_makes_hprpc(&loi->loi_read_lop)) {
2020 on_list(&loi->loi_ready_item, &cli->cl_loi_ready_list, 0);
2021 on_list(&loi->loi_hp_ready_item, &cli->cl_loi_hp_ready_list, 1);
2023 on_list(&loi->loi_hp_ready_item, &cli->cl_loi_hp_ready_list, 0);
2024 on_list(&loi->loi_ready_item, &cli->cl_loi_ready_list,
2025 lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE)||
2026 lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ));
2029 on_list(&loi->loi_write_item, &cli->cl_loi_write_list,
2030 loi->loi_write_lop.lop_num_pending);
2032 on_list(&loi->loi_read_item, &cli->cl_loi_read_list,
2033 loi->loi_read_lop.lop_num_pending);
2036 static void lop_update_pending(struct client_obd *cli,
2037 struct loi_oap_pages *lop, int cmd, int delta)
2039 lop->lop_num_pending += delta;
2040 if (cmd & OBD_BRW_WRITE)
2041 cli->cl_pending_w_pages += delta;
2043 cli->cl_pending_r_pages += delta;
2047 * this is called when a sync waiter receives an interruption. Its job is to
2048 * get the caller woken as soon as possible. If its page hasn't been put in an
2049 * rpc yet it can dequeue immediately. Otherwise it has to mark the rpc as
2050 * desiring interruption which will forcefully complete the rpc once the rpc
2053 int osc_oap_interrupted(const struct lu_env *env, struct osc_async_page *oap)
2055 struct loi_oap_pages *lop;
2056 struct lov_oinfo *loi;
2060 LASSERT(!oap->oap_interrupted);
2061 oap->oap_interrupted = 1;
2063 /* ok, it's been put in an rpc. only one oap gets a request reference */
2064 if (oap->oap_request != NULL) {
2065 ptlrpc_mark_interrupted(oap->oap_request);
2066 ptlrpcd_wake(oap->oap_request);
2067 ptlrpc_req_finished(oap->oap_request);
2068 oap->oap_request = NULL;
2072 * page completion may be called only if ->cpo_prep() method was
2073 * executed by osc_io_submit(), that also adds page the to pending list
2075 if (!cfs_list_empty(&oap->oap_pending_item)) {
2076 cfs_list_del_init(&oap->oap_pending_item);
2077 cfs_list_del_init(&oap->oap_urgent_item);
2080 lop = (oap->oap_cmd & OBD_BRW_WRITE) ?
2081 &loi->loi_write_lop : &loi->loi_read_lop;
2082 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, -1);
2083 loi_list_maint(oap->oap_cli, oap->oap_loi);
2084 rc = oap->oap_caller_ops->ap_completion(env,
2085 oap->oap_caller_data,
2086 oap->oap_cmd, NULL, -EINTR);
2092 /* this is trying to propogate async writeback errors back up to the
2093 * application. As an async write fails we record the error code for later if
2094 * the app does an fsync. As long as errors persist we force future rpcs to be
2095 * sync so that the app can get a sync error and break the cycle of queueing
2096 * pages for which writeback will fail. */
2097 static void osc_process_ar(struct osc_async_rc *ar, __u64 xid,
2104 ar->ar_force_sync = 1;
2105 ar->ar_min_xid = ptlrpc_sample_next_xid();
2110 if (ar->ar_force_sync && (xid >= ar->ar_min_xid))
2111 ar->ar_force_sync = 0;
2114 void osc_oap_to_pending(struct osc_async_page *oap)
2116 struct loi_oap_pages *lop;
2118 if (oap->oap_cmd & OBD_BRW_WRITE)
2119 lop = &oap->oap_loi->loi_write_lop;
2121 lop = &oap->oap_loi->loi_read_lop;
2123 if (oap->oap_async_flags & ASYNC_HP)
2124 cfs_list_add(&oap->oap_urgent_item, &lop->lop_urgent);
2125 else if (oap->oap_async_flags & ASYNC_URGENT)
2126 cfs_list_add_tail(&oap->oap_urgent_item, &lop->lop_urgent);
2127 cfs_list_add_tail(&oap->oap_pending_item, &lop->lop_pending);
2128 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, 1);
2131 /* this must be called holding the loi list lock to give coverage to exit_cache,
2132 * async_flag maintenance, and oap_request */
2133 static void osc_ap_completion(const struct lu_env *env,
2134 struct client_obd *cli, struct obdo *oa,
2135 struct osc_async_page *oap, int sent, int rc)
2140 if (oap->oap_request != NULL) {
2141 xid = ptlrpc_req_xid(oap->oap_request);
2142 ptlrpc_req_finished(oap->oap_request);
2143 oap->oap_request = NULL;
2146 cfs_spin_lock(&oap->oap_lock);
2147 oap->oap_async_flags = 0;
2148 cfs_spin_unlock(&oap->oap_lock);
2149 oap->oap_interrupted = 0;
2151 if (oap->oap_cmd & OBD_BRW_WRITE) {
2152 osc_process_ar(&cli->cl_ar, xid, rc);
2153 osc_process_ar(&oap->oap_loi->loi_ar, xid, rc);
2156 if (rc == 0 && oa != NULL) {
2157 if (oa->o_valid & OBD_MD_FLBLOCKS)
2158 oap->oap_loi->loi_lvb.lvb_blocks = oa->o_blocks;
2159 if (oa->o_valid & OBD_MD_FLMTIME)
2160 oap->oap_loi->loi_lvb.lvb_mtime = oa->o_mtime;
2161 if (oa->o_valid & OBD_MD_FLATIME)
2162 oap->oap_loi->loi_lvb.lvb_atime = oa->o_atime;
2163 if (oa->o_valid & OBD_MD_FLCTIME)
2164 oap->oap_loi->loi_lvb.lvb_ctime = oa->o_ctime;
2167 rc = oap->oap_caller_ops->ap_completion(env, oap->oap_caller_data,
2168 oap->oap_cmd, oa, rc);
2170 /* ll_ap_completion (from llite) drops PG_locked. so, a new
2171 * I/O on the page could start, but OSC calls it under lock
2172 * and thus we can add oap back to pending safely */
2174 /* upper layer wants to leave the page on pending queue */
2175 osc_oap_to_pending(oap);
2177 osc_exit_cache(cli, oap, sent);
2181 static int brw_interpret(const struct lu_env *env,
2182 struct ptlrpc_request *req, void *data, int rc)
2184 struct osc_brw_async_args *aa = data;
2185 struct client_obd *cli;
2189 rc = osc_brw_fini_request(req, rc);
2190 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
2191 if (osc_recoverable_error(rc)) {
2192 /* Only retry once for mmaped files since the mmaped page
2193 * might be modified at anytime. We have to retry at least
2194 * once in case there WAS really a corruption of the page
2195 * on the network, that was not caused by mmap() modifying
2196 * the page. Bug11742 */
2197 if ((rc == -EAGAIN) && (aa->aa_resends > 0) &&
2198 aa->aa_oa->o_valid & OBD_MD_FLFLAGS &&
2199 aa->aa_oa->o_flags & OBD_FL_MMAP) {
2202 rc = osc_brw_redo_request(req, aa);
2209 capa_put(aa->aa_ocapa);
2210 aa->aa_ocapa = NULL;
2215 client_obd_list_lock(&cli->cl_loi_list_lock);
2217 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2218 * is called so we know whether to go to sync BRWs or wait for more
2219 * RPCs to complete */
2220 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2221 cli->cl_w_in_flight--;
2223 cli->cl_r_in_flight--;
2225 async = cfs_list_empty(&aa->aa_oaps);
2226 if (!async) { /* from osc_send_oap_rpc() */
2227 struct osc_async_page *oap, *tmp;
2228 /* the caller may re-use the oap after the completion call so
2229 * we need to clean it up a little */
2230 cfs_list_for_each_entry_safe(oap, tmp, &aa->aa_oaps,
2232 cfs_list_del_init(&oap->oap_rpc_item);
2233 osc_ap_completion(env, cli, aa->aa_oa, oap, 1, rc);
2235 OBDO_FREE(aa->aa_oa);
2236 } else { /* from async_internal() */
2238 for (i = 0; i < aa->aa_page_count; i++)
2239 osc_release_write_grant(aa->aa_cli, aa->aa_ppga[i], 1);
2241 osc_wake_cache_waiters(cli);
2242 osc_check_rpcs(env, cli);
2243 client_obd_list_unlock(&cli->cl_loi_list_lock);
2245 cl_req_completion(env, aa->aa_clerq, rc);
2246 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2251 static struct ptlrpc_request *osc_build_req(const struct lu_env *env,
2252 struct client_obd *cli,
2253 cfs_list_t *rpc_list,
2254 int page_count, int cmd)
2256 struct ptlrpc_request *req;
2257 struct brw_page **pga = NULL;
2258 struct osc_brw_async_args *aa;
2259 struct obdo *oa = NULL;
2260 const struct obd_async_page_ops *ops = NULL;
2261 void *caller_data = NULL;
2262 struct osc_async_page *oap;
2263 struct osc_async_page *tmp;
2264 struct ost_body *body;
2265 struct cl_req *clerq = NULL;
2266 enum cl_req_type crt = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2267 struct ldlm_lock *lock = NULL;
2268 struct cl_req_attr crattr;
2269 int i, rc, mpflag = 0;
2272 LASSERT(!cfs_list_empty(rpc_list));
2274 if (cmd & OBD_BRW_MEMALLOC)
2275 mpflag = cfs_memory_pressure_get_and_set();
2277 memset(&crattr, 0, sizeof crattr);
2278 OBD_ALLOC(pga, sizeof(*pga) * page_count);
2280 GOTO(out, req = ERR_PTR(-ENOMEM));
2284 GOTO(out, req = ERR_PTR(-ENOMEM));
2287 cfs_list_for_each_entry(oap, rpc_list, oap_rpc_item) {
2288 struct cl_page *page = osc_oap2cl_page(oap);
2290 ops = oap->oap_caller_ops;
2291 caller_data = oap->oap_caller_data;
2293 clerq = cl_req_alloc(env, page, crt,
2294 1 /* only 1-object rpcs for
2297 GOTO(out, req = (void *)clerq);
2298 lock = oap->oap_ldlm_lock;
2300 pga[i] = &oap->oap_brw_page;
2301 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2302 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
2303 pga[i]->pg, cfs_page_index(oap->oap_page), oap, pga[i]->flag);
2305 cl_req_page_add(env, clerq, page);
2308 /* always get the data for the obdo for the rpc */
2309 LASSERT(ops != NULL);
2311 crattr.cra_capa = NULL;
2312 cl_req_attr_set(env, clerq, &crattr, ~0ULL);
2314 oa->o_handle = lock->l_remote_handle;
2315 oa->o_valid |= OBD_MD_FLHANDLE;
2318 rc = cl_req_prep(env, clerq);
2320 CERROR("cl_req_prep failed: %d\n", rc);
2321 GOTO(out, req = ERR_PTR(rc));
2324 sort_brw_pages(pga, page_count);
2325 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
2326 pga, &req, crattr.cra_capa, 1);
2328 CERROR("prep_req failed: %d\n", rc);
2329 GOTO(out, req = ERR_PTR(rc));
2332 if (cmd & OBD_BRW_MEMALLOC)
2333 req->rq_memalloc = 1;
2335 /* Need to update the timestamps after the request is built in case
2336 * we race with setattr (locally or in queue at OST). If OST gets
2337 * later setattr before earlier BRW (as determined by the request xid),
2338 * the OST will not use BRW timestamps. Sadly, there is no obvious
2339 * way to do this in a single call. bug 10150 */
2340 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
2341 cl_req_attr_set(env, clerq, &crattr,
2342 OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME);
2344 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2345 aa = ptlrpc_req_async_args(req);
2346 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
2347 cfs_list_splice(rpc_list, &aa->aa_oaps);
2348 CFS_INIT_LIST_HEAD(rpc_list);
2349 aa->aa_clerq = clerq;
2351 if (cmd & OBD_BRW_MEMALLOC)
2352 cfs_memory_pressure_restore(mpflag);
2354 capa_put(crattr.cra_capa);
2359 OBD_FREE(pga, sizeof(*pga) * page_count);
2360 /* this should happen rarely and is pretty bad, it makes the
2361 * pending list not follow the dirty order */
2362 client_obd_list_lock(&cli->cl_loi_list_lock);
2363 cfs_list_for_each_entry_safe(oap, tmp, rpc_list, oap_rpc_item) {
2364 cfs_list_del_init(&oap->oap_rpc_item);
2366 /* queued sync pages can be torn down while the pages
2367 * were between the pending list and the rpc */
2368 if (oap->oap_interrupted) {
2369 CDEBUG(D_INODE, "oap %p interrupted\n", oap);
2370 osc_ap_completion(env, cli, NULL, oap, 0,
2374 osc_ap_completion(env, cli, NULL, oap, 0, PTR_ERR(req));
2376 if (clerq && !IS_ERR(clerq))
2377 cl_req_completion(env, clerq, PTR_ERR(req));
2383 * prepare pages for ASYNC io and put pages in send queue.
2385 * \param cmd OBD_BRW_* macroses
2386 * \param lop pending pages
2388 * \return zero if no page added to send queue.
2389 * \return 1 if pages successfully added to send queue.
2390 * \return negative on errors.
2393 osc_send_oap_rpc(const struct lu_env *env, struct client_obd *cli,
2394 struct lov_oinfo *loi,
2395 int cmd, struct loi_oap_pages *lop)
2397 struct ptlrpc_request *req;
2398 obd_count page_count = 0;
2399 struct osc_async_page *oap = NULL, *tmp;
2400 struct osc_brw_async_args *aa;
2401 const struct obd_async_page_ops *ops;
2402 CFS_LIST_HEAD(rpc_list);
2403 CFS_LIST_HEAD(tmp_list);
2404 unsigned int ending_offset;
2405 unsigned starting_offset = 0;
2406 int srvlock = 0, mem_tight = 0;
2407 struct cl_object *clob = NULL;
2410 /* ASYNC_HP pages first. At present, when the lock the pages is
2411 * to be canceled, the pages covered by the lock will be sent out
2412 * with ASYNC_HP. We have to send out them as soon as possible. */
2413 cfs_list_for_each_entry_safe(oap, tmp, &lop->lop_urgent, oap_urgent_item) {
2414 if (oap->oap_async_flags & ASYNC_HP)
2415 cfs_list_move(&oap->oap_pending_item, &tmp_list);
2417 cfs_list_move_tail(&oap->oap_pending_item, &tmp_list);
2418 if (++page_count >= cli->cl_max_pages_per_rpc)
2422 cfs_list_splice(&tmp_list, &lop->lop_pending);
2425 /* first we find the pages we're allowed to work with */
2426 cfs_list_for_each_entry_safe(oap, tmp, &lop->lop_pending,
2428 ops = oap->oap_caller_ops;
2430 LASSERTF(oap->oap_magic == OAP_MAGIC, "Bad oap magic: oap %p, "
2431 "magic 0x%x\n", oap, oap->oap_magic);
2434 /* pin object in memory, so that completion call-backs
2435 * can be safely called under client_obd_list lock. */
2436 clob = osc_oap2cl_page(oap)->cp_obj;
2437 cl_object_get(clob);
2440 if (page_count != 0 &&
2441 srvlock != !!(oap->oap_brw_flags & OBD_BRW_SRVLOCK)) {
2442 CDEBUG(D_PAGE, "SRVLOCK flag mismatch,"
2443 " oap %p, page %p, srvlock %u\n",
2444 oap, oap->oap_brw_page.pg, (unsigned)!srvlock);
2448 /* If there is a gap at the start of this page, it can't merge
2449 * with any previous page, so we'll hand the network a
2450 * "fragmented" page array that it can't transfer in 1 RDMA */
2451 if (page_count != 0 && oap->oap_page_off != 0)
2454 /* in llite being 'ready' equates to the page being locked
2455 * until completion unlocks it. commit_write submits a page
2456 * as not ready because its unlock will happen unconditionally
2457 * as the call returns. if we race with commit_write giving
2458 * us that page we don't want to create a hole in the page
2459 * stream, so we stop and leave the rpc to be fired by
2460 * another dirtier or kupdated interval (the not ready page
2461 * will still be on the dirty list). we could call in
2462 * at the end of ll_file_write to process the queue again. */
2463 if (!(oap->oap_async_flags & ASYNC_READY)) {
2464 int rc = ops->ap_make_ready(env, oap->oap_caller_data,
2467 CDEBUG(D_INODE, "oap %p page %p returned %d "
2468 "instead of ready\n", oap,
2472 /* llite is telling us that the page is still
2473 * in commit_write and that we should try
2474 * and put it in an rpc again later. we
2475 * break out of the loop so we don't create
2476 * a hole in the sequence of pages in the rpc
2481 /* the io isn't needed.. tell the checks
2482 * below to complete the rpc with EINTR */
2483 cfs_spin_lock(&oap->oap_lock);
2484 oap->oap_async_flags |= ASYNC_COUNT_STABLE;
2485 cfs_spin_unlock(&oap->oap_lock);
2486 oap->oap_count = -EINTR;
2489 cfs_spin_lock(&oap->oap_lock);
2490 oap->oap_async_flags |= ASYNC_READY;
2491 cfs_spin_unlock(&oap->oap_lock);
2494 LASSERTF(0, "oap %p page %p returned %d "
2495 "from make_ready\n", oap,
2503 * Page submitted for IO has to be locked. Either by
2504 * ->ap_make_ready() or by higher layers.
2506 #if defined(__KERNEL__) && defined(__linux__)
2508 struct cl_page *page;
2510 page = osc_oap2cl_page(oap);
2512 if (page->cp_type == CPT_CACHEABLE &&
2513 !(PageLocked(oap->oap_page) &&
2514 (CheckWriteback(oap->oap_page, cmd)))) {
2515 CDEBUG(D_PAGE, "page %p lost wb %lx/%x\n",
2517 (long)oap->oap_page->flags,
2518 oap->oap_async_flags);
2524 /* take the page out of our book-keeping */
2525 cfs_list_del_init(&oap->oap_pending_item);
2526 lop_update_pending(cli, lop, cmd, -1);
2527 cfs_list_del_init(&oap->oap_urgent_item);
2529 if (page_count == 0)
2530 starting_offset = (oap->oap_obj_off+oap->oap_page_off) &
2531 (PTLRPC_MAX_BRW_SIZE - 1);
2533 /* ask the caller for the size of the io as the rpc leaves. */
2534 if (!(oap->oap_async_flags & ASYNC_COUNT_STABLE)) {
2536 ops->ap_refresh_count(env, oap->oap_caller_data,
2538 LASSERT(oap->oap_page_off + oap->oap_count <= CFS_PAGE_SIZE);
2540 if (oap->oap_count <= 0) {
2541 CDEBUG(D_CACHE, "oap %p count %d, completing\n", oap,
2543 osc_ap_completion(env, cli, NULL,
2544 oap, 0, oap->oap_count);
2548 /* now put the page back in our accounting */
2549 cfs_list_add_tail(&oap->oap_rpc_item, &rpc_list);
2550 if (oap->oap_brw_flags & OBD_BRW_MEMALLOC)
2552 if (page_count == 0)
2553 srvlock = !!(oap->oap_brw_flags & OBD_BRW_SRVLOCK);
2554 if (++page_count >= cli->cl_max_pages_per_rpc)
2557 /* End on a PTLRPC_MAX_BRW_SIZE boundary. We want full-sized
2558 * RPCs aligned on PTLRPC_MAX_BRW_SIZE boundaries to help reads
2559 * have the same alignment as the initial writes that allocated
2560 * extents on the server. */
2561 ending_offset = (oap->oap_obj_off + oap->oap_page_off +
2562 oap->oap_count) & (PTLRPC_MAX_BRW_SIZE - 1);
2563 if (ending_offset == 0)
2566 /* If there is a gap at the end of this page, it can't merge
2567 * with any subsequent pages, so we'll hand the network a
2568 * "fragmented" page array that it can't transfer in 1 RDMA */
2569 if (oap->oap_page_off + oap->oap_count < CFS_PAGE_SIZE)
2573 osc_wake_cache_waiters(cli);
2575 loi_list_maint(cli, loi);
2577 client_obd_list_unlock(&cli->cl_loi_list_lock);
2580 cl_object_put(env, clob);
2582 if (page_count == 0) {
2583 client_obd_list_lock(&cli->cl_loi_list_lock);
2587 req = osc_build_req(env, cli, &rpc_list, page_count,
2588 mem_tight ? (cmd | OBD_BRW_MEMALLOC) : cmd);
2590 LASSERT(cfs_list_empty(&rpc_list));
2591 loi_list_maint(cli, loi);
2592 RETURN(PTR_ERR(req));
2595 aa = ptlrpc_req_async_args(req);
2597 if (cmd == OBD_BRW_READ) {
2598 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2599 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2600 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2601 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2603 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2604 lprocfs_oh_tally(&cli->cl_write_rpc_hist,
2605 cli->cl_w_in_flight);
2606 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2607 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2609 ptlrpc_lprocfs_brw(req, aa->aa_requested_nob);
2611 client_obd_list_lock(&cli->cl_loi_list_lock);
2613 if (cmd == OBD_BRW_READ)
2614 cli->cl_r_in_flight++;
2616 cli->cl_w_in_flight++;
2618 /* queued sync pages can be torn down while the pages
2619 * were between the pending list and the rpc */
2621 cfs_list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2622 /* only one oap gets a request reference */
2625 if (oap->oap_interrupted && !req->rq_intr) {
2626 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2628 ptlrpc_mark_interrupted(req);
2632 tmp->oap_request = ptlrpc_request_addref(req);
2634 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2635 page_count, aa, cli->cl_r_in_flight, cli->cl_w_in_flight);
2637 req->rq_interpret_reply = brw_interpret;
2638 ptlrpcd_add_req(req, PSCOPE_BRW);
2642 #define LOI_DEBUG(LOI, STR, args...) \
2643 CDEBUG(D_INODE, "loi ready %d wr %d:%d rd %d:%d " STR, \
2644 !cfs_list_empty(&(LOI)->loi_ready_item) || \
2645 !cfs_list_empty(&(LOI)->loi_hp_ready_item), \
2646 (LOI)->loi_write_lop.lop_num_pending, \
2647 !cfs_list_empty(&(LOI)->loi_write_lop.lop_urgent), \
2648 (LOI)->loi_read_lop.lop_num_pending, \
2649 !cfs_list_empty(&(LOI)->loi_read_lop.lop_urgent), \
2652 /* This is called by osc_check_rpcs() to find which objects have pages that
2653 * we could be sending. These lists are maintained by lop_makes_rpc(). */
2654 struct lov_oinfo *osc_next_loi(struct client_obd *cli)
2658 /* First return objects that have blocked locks so that they
2659 * will be flushed quickly and other clients can get the lock,
2660 * then objects which have pages ready to be stuffed into RPCs */
2661 if (!cfs_list_empty(&cli->cl_loi_hp_ready_list))
2662 RETURN(cfs_list_entry(cli->cl_loi_hp_ready_list.next,
2663 struct lov_oinfo, loi_hp_ready_item));
2664 if (!cfs_list_empty(&cli->cl_loi_ready_list))
2665 RETURN(cfs_list_entry(cli->cl_loi_ready_list.next,
2666 struct lov_oinfo, loi_ready_item));
2668 /* then if we have cache waiters, return all objects with queued
2669 * writes. This is especially important when many small files
2670 * have filled up the cache and not been fired into rpcs because
2671 * they don't pass the nr_pending/object threshhold */
2672 if (!cfs_list_empty(&cli->cl_cache_waiters) &&
2673 !cfs_list_empty(&cli->cl_loi_write_list))
2674 RETURN(cfs_list_entry(cli->cl_loi_write_list.next,
2675 struct lov_oinfo, loi_write_item));
2677 /* then return all queued objects when we have an invalid import
2678 * so that they get flushed */
2679 if (cli->cl_import == NULL || cli->cl_import->imp_invalid) {
2680 if (!cfs_list_empty(&cli->cl_loi_write_list))
2681 RETURN(cfs_list_entry(cli->cl_loi_write_list.next,
2684 if (!cfs_list_empty(&cli->cl_loi_read_list))
2685 RETURN(cfs_list_entry(cli->cl_loi_read_list.next,
2686 struct lov_oinfo, loi_read_item));
2691 static int osc_max_rpc_in_flight(struct client_obd *cli, struct lov_oinfo *loi)
2693 struct osc_async_page *oap;
2696 if (!cfs_list_empty(&loi->loi_write_lop.lop_urgent)) {
2697 oap = cfs_list_entry(loi->loi_write_lop.lop_urgent.next,
2698 struct osc_async_page, oap_urgent_item);
2699 hprpc = !!(oap->oap_async_flags & ASYNC_HP);
2702 if (!hprpc && !cfs_list_empty(&loi->loi_read_lop.lop_urgent)) {
2703 oap = cfs_list_entry(loi->loi_read_lop.lop_urgent.next,
2704 struct osc_async_page, oap_urgent_item);
2705 hprpc = !!(oap->oap_async_flags & ASYNC_HP);
2708 return rpcs_in_flight(cli) >= cli->cl_max_rpcs_in_flight + hprpc;
2711 /* called with the loi list lock held */
2712 void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli)
2714 struct lov_oinfo *loi;
2715 int rc = 0, race_counter = 0;
2718 while ((loi = osc_next_loi(cli)) != NULL) {
2719 LOI_DEBUG(loi, "%lu in flight\n", rpcs_in_flight(cli));
2721 if (osc_max_rpc_in_flight(cli, loi))
2724 /* attempt some read/write balancing by alternating between
2725 * reads and writes in an object. The makes_rpc checks here
2726 * would be redundant if we were getting read/write work items
2727 * instead of objects. we don't want send_oap_rpc to drain a
2728 * partial read pending queue when we're given this object to
2729 * do io on writes while there are cache waiters */
2730 if (lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE)) {
2731 rc = osc_send_oap_rpc(env, cli, loi, OBD_BRW_WRITE,
2732 &loi->loi_write_lop);
2734 CERROR("Write request failed with %d\n", rc);
2736 /* osc_send_oap_rpc failed, mostly because of
2739 * It can't break here, because if:
2740 * - a page was submitted by osc_io_submit, so
2742 * - no request in flight
2743 * - no subsequent request
2744 * The system will be in live-lock state,
2745 * because there is no chance to call
2746 * osc_io_unplug() and osc_check_rpcs() any
2747 * more. pdflush can't help in this case,
2748 * because it might be blocked at grabbing
2749 * the page lock as we mentioned.
2751 * Anyway, continue to drain pages. */
2760 if (lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ)) {
2761 rc = osc_send_oap_rpc(env, cli, loi, OBD_BRW_READ,
2762 &loi->loi_read_lop);
2764 CERROR("Read request failed with %d\n", rc);
2772 /* attempt some inter-object balancing by issuing rpcs
2773 * for each object in turn */
2774 if (!cfs_list_empty(&loi->loi_hp_ready_item))
2775 cfs_list_del_init(&loi->loi_hp_ready_item);
2776 if (!cfs_list_empty(&loi->loi_ready_item))
2777 cfs_list_del_init(&loi->loi_ready_item);
2778 if (!cfs_list_empty(&loi->loi_write_item))
2779 cfs_list_del_init(&loi->loi_write_item);
2780 if (!cfs_list_empty(&loi->loi_read_item))
2781 cfs_list_del_init(&loi->loi_read_item);
2783 loi_list_maint(cli, loi);
2785 /* send_oap_rpc fails with 0 when make_ready tells it to
2786 * back off. llite's make_ready does this when it tries
2787 * to lock a page queued for write that is already locked.
2788 * we want to try sending rpcs from many objects, but we
2789 * don't want to spin failing with 0. */
2790 if (race_counter == 10)
2796 /* we're trying to queue a page in the osc so we're subject to the
2797 * 'cl_dirty_max' limit on the number of pages that can be queued in the osc.
2798 * If the osc's queued pages are already at that limit, then we want to sleep
2799 * until there is space in the osc's queue for us. We also may be waiting for
2800 * write credits from the OST if there are RPCs in flight that may return some
2801 * before we fall back to sync writes.
2803 * We need this know our allocation was granted in the presence of signals */
2804 static int ocw_granted(struct client_obd *cli, struct osc_cache_waiter *ocw)
2808 client_obd_list_lock(&cli->cl_loi_list_lock);
2809 rc = cfs_list_empty(&ocw->ocw_entry) || rpcs_in_flight(cli) == 0;
2810 client_obd_list_unlock(&cli->cl_loi_list_lock);
2815 * Non-blocking version of osc_enter_cache() that consumes grant only when it
2818 int osc_enter_cache_try(const struct lu_env *env,
2819 struct client_obd *cli, struct lov_oinfo *loi,
2820 struct osc_async_page *oap, int transient)
2824 has_grant = cli->cl_avail_grant >= CFS_PAGE_SIZE;
2826 osc_consume_write_grant(cli, &oap->oap_brw_page);
2828 cli->cl_dirty_transit += CFS_PAGE_SIZE;
2829 cfs_atomic_inc(&obd_dirty_transit_pages);
2830 oap->oap_brw_flags |= OBD_BRW_NOCACHE;
2836 /* Caller must hold loi_list_lock - we drop/regain it if we need to wait for
2837 * grant or cache space. */
2838 static int osc_enter_cache(const struct lu_env *env,
2839 struct client_obd *cli, struct lov_oinfo *loi,
2840 struct osc_async_page *oap)
2842 struct osc_cache_waiter ocw;
2843 struct l_wait_info lwi = { 0 };
2847 CDEBUG(D_CACHE, "dirty: %ld/%d dirty_max: %ld/%d dropped: %lu "
2848 "grant: %lu\n", cli->cl_dirty, cfs_atomic_read(&obd_dirty_pages),
2849 cli->cl_dirty_max, obd_max_dirty_pages,
2850 cli->cl_lost_grant, cli->cl_avail_grant);
2852 /* force the caller to try sync io. this can jump the list
2853 * of queued writes and create a discontiguous rpc stream */
2854 if (cli->cl_dirty_max < CFS_PAGE_SIZE || cli->cl_ar.ar_force_sync ||
2855 loi->loi_ar.ar_force_sync)
2858 /* Hopefully normal case - cache space and write credits available */
2859 if (cli->cl_dirty + CFS_PAGE_SIZE <= cli->cl_dirty_max &&
2860 cfs_atomic_read(&obd_dirty_pages) + 1 <= obd_max_dirty_pages &&
2861 osc_enter_cache_try(env, cli, loi, oap, 0))
2864 /* It is safe to block as a cache waiter as long as there is grant
2865 * space available or the hope of additional grant being returned
2866 * when an in flight write completes. Using the write back cache
2867 * if possible is preferable to sending the data synchronously
2868 * because write pages can then be merged in to large requests.
2869 * The addition of this cache waiter will causing pending write
2870 * pages to be sent immediately. */
2871 if (cli->cl_w_in_flight || cli->cl_avail_grant >= CFS_PAGE_SIZE) {
2872 cfs_list_add_tail(&ocw.ocw_entry, &cli->cl_cache_waiters);
2873 cfs_waitq_init(&ocw.ocw_waitq);
2877 loi_list_maint(cli, loi);
2878 osc_check_rpcs(env, cli);
2879 client_obd_list_unlock(&cli->cl_loi_list_lock);
2881 CDEBUG(D_CACHE, "sleeping for cache space\n");
2882 l_wait_event(ocw.ocw_waitq, ocw_granted(cli, &ocw), &lwi);
2884 client_obd_list_lock(&cli->cl_loi_list_lock);
2885 if (!cfs_list_empty(&ocw.ocw_entry)) {
2886 cfs_list_del(&ocw.ocw_entry);
2896 int osc_prep_async_page(struct obd_export *exp, struct lov_stripe_md *lsm,
2897 struct lov_oinfo *loi, cfs_page_t *page,
2898 obd_off offset, const struct obd_async_page_ops *ops,
2899 void *data, void **res, int nocache,
2900 struct lustre_handle *lockh)
2902 struct osc_async_page *oap;
2907 return cfs_size_round(sizeof(*oap));
2910 oap->oap_magic = OAP_MAGIC;
2911 oap->oap_cli = &exp->exp_obd->u.cli;
2914 oap->oap_caller_ops = ops;
2915 oap->oap_caller_data = data;
2917 oap->oap_page = page;
2918 oap->oap_obj_off = offset;
2919 if (!client_is_remote(exp) &&
2920 cfs_capable(CFS_CAP_SYS_RESOURCE))
2921 oap->oap_brw_flags = OBD_BRW_NOQUOTA;
2923 LASSERT(!(offset & ~CFS_PAGE_MASK));
2925 CFS_INIT_LIST_HEAD(&oap->oap_pending_item);
2926 CFS_INIT_LIST_HEAD(&oap->oap_urgent_item);
2927 CFS_INIT_LIST_HEAD(&oap->oap_rpc_item);
2928 CFS_INIT_LIST_HEAD(&oap->oap_page_list);
2930 cfs_spin_lock_init(&oap->oap_lock);
2931 CDEBUG(D_CACHE, "oap %p page %p obj off "LPU64"\n", oap, page, offset);
2935 struct osc_async_page *oap_from_cookie(void *cookie)
2937 struct osc_async_page *oap = cookie;
2938 if (oap->oap_magic != OAP_MAGIC)
2939 return ERR_PTR(-EINVAL);
2943 int osc_queue_async_io(const struct lu_env *env,
2944 struct obd_export *exp, struct lov_stripe_md *lsm,
2945 struct lov_oinfo *loi, void *cookie,
2946 int cmd, obd_off off, int count,
2947 obd_flag brw_flags, enum async_flags async_flags)
2949 struct client_obd *cli = &exp->exp_obd->u.cli;
2950 struct osc_async_page *oap;
2954 oap = oap_from_cookie(cookie);
2956 RETURN(PTR_ERR(oap));
2958 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2961 if (!cfs_list_empty(&oap->oap_pending_item) ||
2962 !cfs_list_empty(&oap->oap_urgent_item) ||
2963 !cfs_list_empty(&oap->oap_rpc_item))
2966 /* check if the file's owner/group is over quota */
2967 if ((cmd & OBD_BRW_WRITE) && !(cmd & OBD_BRW_NOQUOTA)) {
2968 struct cl_object *obj;
2969 struct cl_attr attr; /* XXX put attr into thread info */
2970 unsigned int qid[MAXQUOTAS];
2972 obj = cl_object_top(osc_oap2cl_page(oap)->cp_obj);
2974 cl_object_attr_lock(obj);
2975 rc = cl_object_attr_get(env, obj, &attr);
2976 cl_object_attr_unlock(obj);
2978 qid[USRQUOTA] = attr.cat_uid;
2979 qid[GRPQUOTA] = attr.cat_gid;
2981 lquota_chkdq(quota_interface, cli, qid) == NO_QUOTA)
2988 loi = lsm->lsm_oinfo[0];
2990 client_obd_list_lock(&cli->cl_loi_list_lock);
2992 LASSERT(off + count <= CFS_PAGE_SIZE);
2994 oap->oap_page_off = off;
2995 oap->oap_count = count;
2996 oap->oap_brw_flags = brw_flags;
2997 /* Give a hint to OST that requests are coming from kswapd - bug19529 */
2998 if (cfs_memory_pressure_get())
2999 oap->oap_brw_flags |= OBD_BRW_MEMALLOC;
3000 cfs_spin_lock(&oap->oap_lock);
3001 oap->oap_async_flags = async_flags;
3002 cfs_spin_unlock(&oap->oap_lock);
3004 if (cmd & OBD_BRW_WRITE) {
3005 rc = osc_enter_cache(env, cli, loi, oap);
3007 client_obd_list_unlock(&cli->cl_loi_list_lock);
3012 osc_oap_to_pending(oap);
3013 loi_list_maint(cli, loi);
3015 LOI_DEBUG(loi, "oap %p page %p added for cmd %d\n", oap, oap->oap_page,
3018 osc_check_rpcs(env, cli);
3019 client_obd_list_unlock(&cli->cl_loi_list_lock);
3024 /* aka (~was & now & flag), but this is more clear :) */
3025 #define SETTING(was, now, flag) (!(was & flag) && (now & flag))
3027 int osc_set_async_flags_base(struct client_obd *cli,
3028 struct lov_oinfo *loi, struct osc_async_page *oap,
3029 obd_flag async_flags)
3031 struct loi_oap_pages *lop;
3035 LASSERT(!cfs_list_empty(&oap->oap_pending_item));
3037 if (oap->oap_cmd & OBD_BRW_WRITE) {
3038 lop = &loi->loi_write_lop;
3040 lop = &loi->loi_read_lop;
3043 if ((oap->oap_async_flags & async_flags) == async_flags)
3046 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_READY))
3047 flags |= ASYNC_READY;
3049 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_URGENT) &&
3050 cfs_list_empty(&oap->oap_rpc_item)) {
3051 if (oap->oap_async_flags & ASYNC_HP)
3052 cfs_list_add(&oap->oap_urgent_item, &lop->lop_urgent);
3054 cfs_list_add_tail(&oap->oap_urgent_item,
3056 flags |= ASYNC_URGENT;
3057 loi_list_maint(cli, loi);
3059 cfs_spin_lock(&oap->oap_lock);
3060 oap->oap_async_flags |= flags;
3061 cfs_spin_unlock(&oap->oap_lock);
3063 LOI_DEBUG(loi, "oap %p page %p has flags %x\n", oap, oap->oap_page,
3064 oap->oap_async_flags);
3068 int osc_teardown_async_page(struct obd_export *exp,
3069 struct lov_stripe_md *lsm,
3070 struct lov_oinfo *loi, void *cookie)
3072 struct client_obd *cli = &exp->exp_obd->u.cli;
3073 struct loi_oap_pages *lop;
3074 struct osc_async_page *oap;
3078 oap = oap_from_cookie(cookie);
3080 RETURN(PTR_ERR(oap));
3083 loi = lsm->lsm_oinfo[0];
3085 if (oap->oap_cmd & OBD_BRW_WRITE) {
3086 lop = &loi->loi_write_lop;
3088 lop = &loi->loi_read_lop;
3091 client_obd_list_lock(&cli->cl_loi_list_lock);
3093 if (!cfs_list_empty(&oap->oap_rpc_item))
3094 GOTO(out, rc = -EBUSY);
3096 osc_exit_cache(cli, oap, 0);
3097 osc_wake_cache_waiters(cli);
3099 if (!cfs_list_empty(&oap->oap_urgent_item)) {
3100 cfs_list_del_init(&oap->oap_urgent_item);
3101 cfs_spin_lock(&oap->oap_lock);
3102 oap->oap_async_flags &= ~(ASYNC_URGENT | ASYNC_HP);
3103 cfs_spin_unlock(&oap->oap_lock);
3105 if (!cfs_list_empty(&oap->oap_pending_item)) {
3106 cfs_list_del_init(&oap->oap_pending_item);
3107 lop_update_pending(cli, lop, oap->oap_cmd, -1);
3109 loi_list_maint(cli, loi);
3110 LOI_DEBUG(loi, "oap %p page %p torn down\n", oap, oap->oap_page);
3112 client_obd_list_unlock(&cli->cl_loi_list_lock);
3116 static void osc_set_lock_data_with_check(struct ldlm_lock *lock,
3117 struct ldlm_enqueue_info *einfo,
3120 void *data = einfo->ei_cbdata;
3122 LASSERT(lock != NULL);
3123 LASSERT(lock->l_blocking_ast == einfo->ei_cb_bl);
3124 LASSERT(lock->l_resource->lr_type == einfo->ei_type);
3125 LASSERT(lock->l_completion_ast == einfo->ei_cb_cp);
3126 LASSERT(lock->l_glimpse_ast == einfo->ei_cb_gl);
3128 lock_res_and_lock(lock);
3129 cfs_spin_lock(&osc_ast_guard);
3130 LASSERT(lock->l_ast_data == NULL || lock->l_ast_data == data);
3131 lock->l_ast_data = data;
3132 cfs_spin_unlock(&osc_ast_guard);
3133 unlock_res_and_lock(lock);
3136 static void osc_set_data_with_check(struct lustre_handle *lockh,
3137 struct ldlm_enqueue_info *einfo,
3140 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
3143 osc_set_lock_data_with_check(lock, einfo, flags);
3144 LDLM_LOCK_PUT(lock);
3146 CERROR("lockh %p, data %p - client evicted?\n",
3147 lockh, einfo->ei_cbdata);
3150 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
3151 ldlm_iterator_t replace, void *data)
3153 struct ldlm_res_id res_id;
3154 struct obd_device *obd = class_exp2obd(exp);
3156 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
3157 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
3161 /* find any ldlm lock of the inode in osc
3165 static int osc_find_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
3166 ldlm_iterator_t replace, void *data)
3168 struct ldlm_res_id res_id;
3169 struct obd_device *obd = class_exp2obd(exp);
3172 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_seq, &res_id);
3173 rc = ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
3174 if (rc == LDLM_ITER_STOP)
3176 if (rc == LDLM_ITER_CONTINUE)
3181 static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
3182 obd_enqueue_update_f upcall, void *cookie,
3185 int intent = *flags & LDLM_FL_HAS_INTENT;
3189 /* The request was created before ldlm_cli_enqueue call. */
3190 if (rc == ELDLM_LOCK_ABORTED) {
3191 struct ldlm_reply *rep;
3192 rep = req_capsule_server_get(&req->rq_pill,
3195 LASSERT(rep != NULL);
3196 if (rep->lock_policy_res1)
3197 rc = rep->lock_policy_res1;
3201 if ((intent && rc == ELDLM_LOCK_ABORTED) || !rc) {
3202 *flags |= LDLM_FL_LVB_READY;
3203 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
3204 lvb->lvb_size, lvb->lvb_blocks, lvb->lvb_mtime);
3207 /* Call the update callback. */
3208 rc = (*upcall)(cookie, rc);
3212 static int osc_enqueue_interpret(const struct lu_env *env,
3213 struct ptlrpc_request *req,
3214 struct osc_enqueue_args *aa, int rc)
3216 struct ldlm_lock *lock;
3217 struct lustre_handle handle;
3220 /* Make a local copy of a lock handle and a mode, because aa->oa_*
3221 * might be freed anytime after lock upcall has been called. */
3222 lustre_handle_copy(&handle, aa->oa_lockh);
3223 mode = aa->oa_ei->ei_mode;
3225 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
3227 lock = ldlm_handle2lock(&handle);
3229 /* Take an additional reference so that a blocking AST that
3230 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
3231 * to arrive after an upcall has been executed by
3232 * osc_enqueue_fini(). */
3233 ldlm_lock_addref(&handle, mode);
3235 /* Let CP AST to grant the lock first. */
3236 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE, 1);
3238 /* Complete obtaining the lock procedure. */
3239 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
3240 mode, aa->oa_flags, aa->oa_lvb,
3241 sizeof(*aa->oa_lvb), &handle, rc);
3242 /* Complete osc stuff. */
3243 rc = osc_enqueue_fini(req, aa->oa_lvb,
3244 aa->oa_upcall, aa->oa_cookie, aa->oa_flags, rc);
3246 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
3248 /* Release the lock for async request. */
3249 if (lustre_handle_is_used(&handle) && rc == ELDLM_OK)
3251 * Releases a reference taken by ldlm_cli_enqueue(), if it is
3252 * not already released by
3253 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
3255 ldlm_lock_decref(&handle, mode);
3257 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
3258 aa->oa_lockh, req, aa);
3259 ldlm_lock_decref(&handle, mode);
3260 LDLM_LOCK_PUT(lock);
3264 void osc_update_enqueue(struct lustre_handle *lov_lockhp,
3265 struct lov_oinfo *loi, int flags,
3266 struct ost_lvb *lvb, __u32 mode, int rc)
3268 if (rc == ELDLM_OK) {
3269 struct ldlm_lock *lock = ldlm_handle2lock(lov_lockhp);
3272 LASSERT(lock != NULL);
3273 loi->loi_lvb = *lvb;
3274 tmp = loi->loi_lvb.lvb_size;
3275 /* Extend KMS up to the end of this lock and no further
3276 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
3277 if (tmp > lock->l_policy_data.l_extent.end)
3278 tmp = lock->l_policy_data.l_extent.end + 1;
3279 if (tmp >= loi->loi_kms) {
3280 LDLM_DEBUG(lock, "lock acquired, setting rss="LPU64
3281 ", kms="LPU64, loi->loi_lvb.lvb_size, tmp);
3282 loi_kms_set(loi, tmp);
3284 LDLM_DEBUG(lock, "lock acquired, setting rss="
3285 LPU64"; leaving kms="LPU64", end="LPU64,
3286 loi->loi_lvb.lvb_size, loi->loi_kms,
3287 lock->l_policy_data.l_extent.end);
3289 ldlm_lock_allow_match(lock);
3290 LDLM_LOCK_PUT(lock);
3291 } else if (rc == ELDLM_LOCK_ABORTED && (flags & LDLM_FL_HAS_INTENT)) {
3292 loi->loi_lvb = *lvb;
3293 CDEBUG(D_INODE, "glimpsed, setting rss="LPU64"; leaving"
3294 " kms="LPU64"\n", loi->loi_lvb.lvb_size, loi->loi_kms);
3298 EXPORT_SYMBOL(osc_update_enqueue);
3300 struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
3302 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
3303 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
3304 * other synchronous requests, however keeping some locks and trying to obtain
3305 * others may take a considerable amount of time in a case of ost failure; and
3306 * when other sync requests do not get released lock from a client, the client
3307 * is excluded from the cluster -- such scenarious make the life difficult, so
3308 * release locks just after they are obtained. */
3309 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
3310 int *flags, ldlm_policy_data_t *policy,
3311 struct ost_lvb *lvb, int kms_valid,
3312 obd_enqueue_update_f upcall, void *cookie,
3313 struct ldlm_enqueue_info *einfo,
3314 struct lustre_handle *lockh,
3315 struct ptlrpc_request_set *rqset, int async)
3317 struct obd_device *obd = exp->exp_obd;
3318 struct ptlrpc_request *req = NULL;
3319 int intent = *flags & LDLM_FL_HAS_INTENT;
3324 /* Filesystem lock extents are extended to page boundaries so that
3325 * dealing with the page cache is a little smoother. */
3326 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
3327 policy->l_extent.end |= ~CFS_PAGE_MASK;
3330 * kms is not valid when either object is completely fresh (so that no
3331 * locks are cached), or object was evicted. In the latter case cached
3332 * lock cannot be used, because it would prime inode state with
3333 * potentially stale LVB.
3338 /* Next, search for already existing extent locks that will cover us */
3339 /* If we're trying to read, we also search for an existing PW lock. The
3340 * VFS and page cache already protect us locally, so lots of readers/
3341 * writers can share a single PW lock.
3343 * There are problems with conversion deadlocks, so instead of
3344 * converting a read lock to a write lock, we'll just enqueue a new
3347 * At some point we should cancel the read lock instead of making them
3348 * send us a blocking callback, but there are problems with canceling
3349 * locks out from other users right now, too. */
3350 mode = einfo->ei_mode;
3351 if (einfo->ei_mode == LCK_PR)
3353 mode = ldlm_lock_match(obd->obd_namespace,
3354 *flags | LDLM_FL_LVB_READY, res_id,
3355 einfo->ei_type, policy, mode, lockh, 0);
3357 struct ldlm_lock *matched = ldlm_handle2lock(lockh);
3359 if (matched->l_ast_data == NULL ||
3360 matched->l_ast_data == einfo->ei_cbdata) {
3361 /* addref the lock only if not async requests and PW
3362 * lock is matched whereas we asked for PR. */
3363 if (!rqset && einfo->ei_mode != mode)
3364 ldlm_lock_addref(lockh, LCK_PR);
3365 osc_set_lock_data_with_check(matched, einfo, *flags);
3367 /* I would like to be able to ASSERT here that
3368 * rss <= kms, but I can't, for reasons which
3369 * are explained in lov_enqueue() */
3372 /* We already have a lock, and it's referenced */
3373 (*upcall)(cookie, ELDLM_OK);
3375 /* For async requests, decref the lock. */
3376 if (einfo->ei_mode != mode)
3377 ldlm_lock_decref(lockh, LCK_PW);
3379 ldlm_lock_decref(lockh, einfo->ei_mode);
3380 LDLM_LOCK_PUT(matched);
3383 ldlm_lock_decref(lockh, mode);
3384 LDLM_LOCK_PUT(matched);
3389 CFS_LIST_HEAD(cancels);
3390 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3391 &RQF_LDLM_ENQUEUE_LVB);
3395 rc = ldlm_prep_enqueue_req(exp, req, &cancels, 0);
3397 ptlrpc_request_free(req);
3401 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
3403 ptlrpc_request_set_replen(req);
3406 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
3407 *flags &= ~LDLM_FL_BLOCK_GRANTED;
3409 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
3410 sizeof(*lvb), lockh, async);
3413 struct osc_enqueue_args *aa;
3414 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
3415 aa = ptlrpc_req_async_args(req);
3418 aa->oa_flags = flags;
3419 aa->oa_upcall = upcall;
3420 aa->oa_cookie = cookie;
3422 aa->oa_lockh = lockh;
3424 req->rq_interpret_reply =
3425 (ptlrpc_interpterer_t)osc_enqueue_interpret;
3426 if (rqset == PTLRPCD_SET)
3427 ptlrpcd_add_req(req, PSCOPE_OTHER);
3429 ptlrpc_set_add_req(rqset, req);
3430 } else if (intent) {
3431 ptlrpc_req_finished(req);
3436 rc = osc_enqueue_fini(req, lvb, upcall, cookie, flags, rc);
3438 ptlrpc_req_finished(req);
3443 static int osc_enqueue(struct obd_export *exp, struct obd_info *oinfo,
3444 struct ldlm_enqueue_info *einfo,
3445 struct ptlrpc_request_set *rqset)
3447 struct ldlm_res_id res_id;
3451 osc_build_res_name(oinfo->oi_md->lsm_object_id,
3452 oinfo->oi_md->lsm_object_seq, &res_id);
3454 rc = osc_enqueue_base(exp, &res_id, &oinfo->oi_flags, &oinfo->oi_policy,
3455 &oinfo->oi_md->lsm_oinfo[0]->loi_lvb,
3456 oinfo->oi_md->lsm_oinfo[0]->loi_kms_valid,
3457 oinfo->oi_cb_up, oinfo, einfo, oinfo->oi_lockh,
3458 rqset, rqset != NULL);
3462 int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
3463 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
3464 int *flags, void *data, struct lustre_handle *lockh,
3467 struct obd_device *obd = exp->exp_obd;
3468 int lflags = *flags;
3472 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
3475 /* Filesystem lock extents are extended to page boundaries so that
3476 * dealing with the page cache is a little smoother */
3477 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
3478 policy->l_extent.end |= ~CFS_PAGE_MASK;
3480 /* Next, search for already existing extent locks that will cover us */
3481 /* If we're trying to read, we also search for an existing PW lock. The
3482 * VFS and page cache already protect us locally, so lots of readers/
3483 * writers can share a single PW lock. */
3487 rc = ldlm_lock_match(obd->obd_namespace, lflags,
3488 res_id, type, policy, rc, lockh, unref);
3491 osc_set_data_with_check(lockh, data, lflags);
3492 if (!(lflags & LDLM_FL_TEST_LOCK) && mode != rc) {
3493 ldlm_lock_addref(lockh, LCK_PR);
3494 ldlm_lock_decref(lockh, LCK_PW);
3501 int osc_cancel_base(struct lustre_handle *lockh, __u32 mode)
3505 if (unlikely(mode == LCK_GROUP))
3506 ldlm_lock_decref_and_cancel(lockh, mode);
3508 ldlm_lock_decref(lockh, mode);
3513 static int osc_cancel(struct obd_export *exp, struct lov_stripe_md *md,
3514 __u32 mode, struct lustre_handle *lockh)
3517 RETURN(osc_cancel_base(lockh, mode));
3520 static int osc_cancel_unused(struct obd_export *exp,
3521 struct lov_stripe_md *lsm,
3522 ldlm_cancel_flags_t flags,
3525 struct obd_device *obd = class_exp2obd(exp);
3526 struct ldlm_res_id res_id, *resp = NULL;
3529 resp = osc_build_res_name(lsm->lsm_object_id,
3530 lsm->lsm_object_seq, &res_id);
3533 return ldlm_cli_cancel_unused(obd->obd_namespace, resp, flags, opaque);
3536 static int osc_statfs_interpret(const struct lu_env *env,
3537 struct ptlrpc_request *req,
3538 struct osc_async_args *aa, int rc)
3540 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
3541 struct obd_statfs *msfs;
3546 /* The request has in fact never been sent
3547 * due to issues at a higher level (LOV).
3548 * Exit immediately since the caller is
3549 * aware of the problem and takes care
3550 * of the clean up */
3553 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
3554 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
3560 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3562 GOTO(out, rc = -EPROTO);
3565 /* Reinitialize the RDONLY and DEGRADED flags at the client
3566 * on each statfs, so they don't stay set permanently. */
3567 cfs_spin_lock(&cli->cl_oscc.oscc_lock);
3569 if (unlikely(msfs->os_state & OS_STATE_DEGRADED))
3570 cli->cl_oscc.oscc_flags |= OSCC_FLAG_DEGRADED;
3571 else if (unlikely(cli->cl_oscc.oscc_flags & OSCC_FLAG_DEGRADED))
3572 cli->cl_oscc.oscc_flags &= ~OSCC_FLAG_DEGRADED;
3574 if (unlikely(msfs->os_state & OS_STATE_READONLY))
3575 cli->cl_oscc.oscc_flags |= OSCC_FLAG_RDONLY;
3576 else if (unlikely(cli->cl_oscc.oscc_flags & OSCC_FLAG_RDONLY))
3577 cli->cl_oscc.oscc_flags &= ~OSCC_FLAG_RDONLY;
3579 /* Add a bit of hysteresis so this flag isn't continually flapping,
3580 * and ensure that new files don't get extremely fragmented due to
3581 * only a small amount of available space in the filesystem.
3582 * We want to set the NOSPC flag when there is less than ~0.1% free
3583 * and clear it when there is at least ~0.2% free space, so:
3584 * avail < ~0.1% max max = avail + used
3585 * 1025 * avail < avail + used used = blocks - free
3586 * 1024 * avail < used
3587 * 1024 * avail < blocks - free
3588 * avail < ((blocks - free) >> 10)
3590 * On very large disk, say 16TB 0.1% will be 16 GB. We don't want to
3591 * lose that amount of space so in those cases we report no space left
3592 * if their is less than 1 GB left. */
3593 used = min_t(__u64,(msfs->os_blocks - msfs->os_bfree) >> 10, 1 << 30);
3594 if (unlikely(((cli->cl_oscc.oscc_flags & OSCC_FLAG_NOSPC) == 0) &&
3595 ((msfs->os_ffree < 32) || (msfs->os_bavail < used))))
3596 cli->cl_oscc.oscc_flags |= OSCC_FLAG_NOSPC;
3597 else if (unlikely(((cli->cl_oscc.oscc_flags & OSCC_FLAG_NOSPC) != 0) &&
3598 (msfs->os_ffree > 64) && (msfs->os_bavail > (used << 1))))
3599 cli->cl_oscc.oscc_flags &= ~OSCC_FLAG_NOSPC;
3601 cfs_spin_unlock(&cli->cl_oscc.oscc_lock);
3603 *aa->aa_oi->oi_osfs = *msfs;
3605 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
3609 static int osc_statfs_async(struct obd_device *obd, struct obd_info *oinfo,
3610 __u64 max_age, struct ptlrpc_request_set *rqset)
3612 struct ptlrpc_request *req;
3613 struct osc_async_args *aa;
3617 /* We could possibly pass max_age in the request (as an absolute
3618 * timestamp or a "seconds.usec ago") so the target can avoid doing
3619 * extra calls into the filesystem if that isn't necessary (e.g.
3620 * during mount that would help a bit). Having relative timestamps
3621 * is not so great if request processing is slow, while absolute
3622 * timestamps are not ideal because they need time synchronization. */
3623 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
3627 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3629 ptlrpc_request_free(req);
3632 ptlrpc_request_set_replen(req);
3633 req->rq_request_portal = OST_CREATE_PORTAL;
3634 ptlrpc_at_set_req_timeout(req);
3636 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
3637 /* procfs requests not want stat in wait for avoid deadlock */
3638 req->rq_no_resend = 1;
3639 req->rq_no_delay = 1;
3642 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
3643 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
3644 aa = ptlrpc_req_async_args(req);
3647 ptlrpc_set_add_req(rqset, req);
3651 static int osc_statfs(struct obd_device *obd, struct obd_statfs *osfs,
3652 __u64 max_age, __u32 flags)
3654 struct obd_statfs *msfs;
3655 struct ptlrpc_request *req;
3656 struct obd_import *imp = NULL;
3660 /*Since the request might also come from lprocfs, so we need
3661 *sync this with client_disconnect_export Bug15684*/
3662 cfs_down_read(&obd->u.cli.cl_sem);
3663 if (obd->u.cli.cl_import)
3664 imp = class_import_get(obd->u.cli.cl_import);
3665 cfs_up_read(&obd->u.cli.cl_sem);
3669 /* We could possibly pass max_age in the request (as an absolute
3670 * timestamp or a "seconds.usec ago") so the target can avoid doing
3671 * extra calls into the filesystem if that isn't necessary (e.g.
3672 * during mount that would help a bit). Having relative timestamps
3673 * is not so great if request processing is slow, while absolute
3674 * timestamps are not ideal because they need time synchronization. */
3675 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
3677 class_import_put(imp);
3682 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3684 ptlrpc_request_free(req);
3687 ptlrpc_request_set_replen(req);
3688 req->rq_request_portal = OST_CREATE_PORTAL;
3689 ptlrpc_at_set_req_timeout(req);
3691 if (flags & OBD_STATFS_NODELAY) {
3692 /* procfs requests not want stat in wait for avoid deadlock */
3693 req->rq_no_resend = 1;
3694 req->rq_no_delay = 1;
3697 rc = ptlrpc_queue_wait(req);
3701 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3703 GOTO(out, rc = -EPROTO);
3710 ptlrpc_req_finished(req);
3714 /* Retrieve object striping information.
3716 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
3717 * the maximum number of OST indices which will fit in the user buffer.
3718 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
3720 static int osc_getstripe(struct lov_stripe_md *lsm, struct lov_user_md *lump)
3722 /* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
3723 struct lov_user_md_v3 lum, *lumk;
3724 struct lov_user_ost_data_v1 *lmm_objects;
3725 int rc = 0, lum_size;
3731 /* we only need the header part from user space to get lmm_magic and
3732 * lmm_stripe_count, (the header part is common to v1 and v3) */
3733 lum_size = sizeof(struct lov_user_md_v1);
3734 if (cfs_copy_from_user(&lum, lump, lum_size))
3737 if ((lum.lmm_magic != LOV_USER_MAGIC_V1) &&
3738 (lum.lmm_magic != LOV_USER_MAGIC_V3))
3741 /* lov_user_md_vX and lov_mds_md_vX must have the same size */
3742 LASSERT(sizeof(struct lov_user_md_v1) == sizeof(struct lov_mds_md_v1));
3743 LASSERT(sizeof(struct lov_user_md_v3) == sizeof(struct lov_mds_md_v3));
3744 LASSERT(sizeof(lum.lmm_objects[0]) == sizeof(lumk->lmm_objects[0]));
3746 /* we can use lov_mds_md_size() to compute lum_size
3747 * because lov_user_md_vX and lov_mds_md_vX have the same size */
3748 if (lum.lmm_stripe_count > 0) {
3749 lum_size = lov_mds_md_size(lum.lmm_stripe_count, lum.lmm_magic);
3750 OBD_ALLOC(lumk, lum_size);
3754 if (lum.lmm_magic == LOV_USER_MAGIC_V1)
3755 lmm_objects = &(((struct lov_user_md_v1 *)lumk)->lmm_objects[0]);
3757 lmm_objects = &(lumk->lmm_objects[0]);
3758 lmm_objects->l_object_id = lsm->lsm_object_id;
3760 lum_size = lov_mds_md_size(0, lum.lmm_magic);
3764 lumk->lmm_object_id = lsm->lsm_object_id;
3765 lumk->lmm_object_seq = lsm->lsm_object_seq;
3766 lumk->lmm_stripe_count = 1;
3768 if (cfs_copy_to_user(lump, lumk, lum_size))
3772 OBD_FREE(lumk, lum_size);
3778 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
3779 void *karg, void *uarg)
3781 struct obd_device *obd = exp->exp_obd;
3782 struct obd_ioctl_data *data = karg;
3786 if (!cfs_try_module_get(THIS_MODULE)) {
3787 CERROR("Can't get module. Is it alive?");
3791 case OBD_IOC_LOV_GET_CONFIG: {
3793 struct lov_desc *desc;
3794 struct obd_uuid uuid;
3798 if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
3799 GOTO(out, err = -EINVAL);
3801 data = (struct obd_ioctl_data *)buf;
3803 if (sizeof(*desc) > data->ioc_inllen1) {
3804 obd_ioctl_freedata(buf, len);
3805 GOTO(out, err = -EINVAL);
3808 if (data->ioc_inllen2 < sizeof(uuid)) {
3809 obd_ioctl_freedata(buf, len);
3810 GOTO(out, err = -EINVAL);
3813 desc = (struct lov_desc *)data->ioc_inlbuf1;
3814 desc->ld_tgt_count = 1;
3815 desc->ld_active_tgt_count = 1;
3816 desc->ld_default_stripe_count = 1;
3817 desc->ld_default_stripe_size = 0;
3818 desc->ld_default_stripe_offset = 0;
3819 desc->ld_pattern = 0;
3820 memcpy(&desc->ld_uuid, &obd->obd_uuid, sizeof(uuid));
3822 memcpy(data->ioc_inlbuf2, &obd->obd_uuid, sizeof(uuid));
3824 err = cfs_copy_to_user((void *)uarg, buf, len);
3827 obd_ioctl_freedata(buf, len);
3830 case LL_IOC_LOV_SETSTRIPE:
3831 err = obd_alloc_memmd(exp, karg);
3835 case LL_IOC_LOV_GETSTRIPE:
3836 err = osc_getstripe(karg, uarg);
3838 case OBD_IOC_CLIENT_RECOVER:
3839 err = ptlrpc_recover_import(obd->u.cli.cl_import,
3844 case IOC_OSC_SET_ACTIVE:
3845 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
3848 case OBD_IOC_POLL_QUOTACHECK:
3849 err = lquota_poll_check(quota_interface, exp,
3850 (struct if_quotacheck *)karg);
3852 case OBD_IOC_PING_TARGET:
3853 err = ptlrpc_obd_ping(obd);
3856 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
3857 cmd, cfs_curproc_comm());
3858 GOTO(out, err = -ENOTTY);
3861 cfs_module_put(THIS_MODULE);
3865 static int osc_get_info(struct obd_export *exp, obd_count keylen,
3866 void *key, __u32 *vallen, void *val,
3867 struct lov_stripe_md *lsm)
3870 if (!vallen || !val)
3873 if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
3874 __u32 *stripe = val;
3875 *vallen = sizeof(*stripe);
3878 } else if (KEY_IS(KEY_LAST_ID)) {
3879 struct ptlrpc_request *req;
3884 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3885 &RQF_OST_GET_INFO_LAST_ID);
3889 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3890 RCL_CLIENT, keylen);
3891 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3893 ptlrpc_request_free(req);
3897 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3898 memcpy(tmp, key, keylen);
3900 req->rq_no_delay = req->rq_no_resend = 1;
3901 ptlrpc_request_set_replen(req);
3902 rc = ptlrpc_queue_wait(req);
3906 reply = req_capsule_server_get(&req->rq_pill, &RMF_OBD_ID);
3908 GOTO(out, rc = -EPROTO);
3910 *((obd_id *)val) = *reply;
3912 ptlrpc_req_finished(req);
3914 } else if (KEY_IS(KEY_FIEMAP)) {
3915 struct ptlrpc_request *req;
3916 struct ll_user_fiemap *reply;
3920 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3921 &RQF_OST_GET_INFO_FIEMAP);
3925 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_KEY,
3926 RCL_CLIENT, keylen);
3927 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3928 RCL_CLIENT, *vallen);
3929 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3930 RCL_SERVER, *vallen);
3932 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3934 ptlrpc_request_free(req);
3938 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_KEY);
3939 memcpy(tmp, key, keylen);
3940 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3941 memcpy(tmp, val, *vallen);
3943 ptlrpc_request_set_replen(req);
3944 rc = ptlrpc_queue_wait(req);
3948 reply = req_capsule_server_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3950 GOTO(out1, rc = -EPROTO);
3952 memcpy(val, reply, *vallen);
3954 ptlrpc_req_finished(req);
3962 static int osc_setinfo_mds_connect_import(struct obd_import *imp)
3964 struct llog_ctxt *ctxt;
3968 ctxt = llog_get_context(imp->imp_obd, LLOG_MDS_OST_ORIG_CTXT);
3970 rc = llog_initiator_connect(ctxt);
3971 llog_ctxt_put(ctxt);
3973 /* XXX return an error? skip setting below flags? */
3976 cfs_spin_lock(&imp->imp_lock);
3977 imp->imp_server_timeout = 1;
3978 imp->imp_pingable = 1;
3979 cfs_spin_unlock(&imp->imp_lock);
3980 CDEBUG(D_RPCTRACE, "pinging OST %s\n", obd2cli_tgt(imp->imp_obd));
3985 static int osc_setinfo_mds_conn_interpret(const struct lu_env *env,
3986 struct ptlrpc_request *req,
3993 RETURN(osc_setinfo_mds_connect_import(req->rq_import));
3996 static int osc_set_info_async(struct obd_export *exp, obd_count keylen,
3997 void *key, obd_count vallen, void *val,
3998 struct ptlrpc_request_set *set)
4000 struct ptlrpc_request *req;
4001 struct obd_device *obd = exp->exp_obd;
4002 struct obd_import *imp = class_exp2cliimp(exp);
4007 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
4009 if (KEY_IS(KEY_NEXT_ID)) {
4011 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4013 if (vallen != sizeof(obd_id))
4018 if (vallen != sizeof(obd_id))
4021 /* avoid race between allocate new object and set next id
4022 * from ll_sync thread */
4023 cfs_spin_lock(&oscc->oscc_lock);
4024 new_val = *((obd_id*)val) + 1;
4025 if (new_val > oscc->oscc_next_id)
4026 oscc->oscc_next_id = new_val;
4027 cfs_spin_unlock(&oscc->oscc_lock);
4028 CDEBUG(D_HA, "%s: set oscc_next_id = "LPU64"\n",
4029 exp->exp_obd->obd_name,
4030 obd->u.cli.cl_oscc.oscc_next_id);
4035 if (KEY_IS(KEY_CHECKSUM)) {
4036 if (vallen != sizeof(int))
4038 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
4042 if (KEY_IS(KEY_SPTLRPC_CONF)) {
4043 sptlrpc_conf_client_adapt(obd);
4047 if (KEY_IS(KEY_FLUSH_CTX)) {
4048 sptlrpc_import_flush_my_ctx(imp);
4052 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
4055 /* We pass all other commands directly to OST. Since nobody calls osc
4056 methods directly and everybody is supposed to go through LOV, we
4057 assume lov checked invalid values for us.
4058 The only recognised values so far are evict_by_nid and mds_conn.
4059 Even if something bad goes through, we'd get a -EINVAL from OST
4062 if (KEY_IS(KEY_GRANT_SHRINK))
4063 req = ptlrpc_request_alloc(imp, &RQF_OST_SET_GRANT_INFO);
4065 req = ptlrpc_request_alloc(imp, &RQF_OBD_SET_INFO);
4070 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
4071 RCL_CLIENT, keylen);
4072 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
4073 RCL_CLIENT, vallen);
4074 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
4076 ptlrpc_request_free(req);
4080 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
4081 memcpy(tmp, key, keylen);
4082 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_VAL);
4083 memcpy(tmp, val, vallen);
4085 if (KEY_IS(KEY_MDS_CONN)) {
4086 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4088 oscc->oscc_oa.o_seq = (*(__u32 *)val);
4089 oscc->oscc_oa.o_valid |= OBD_MD_FLGROUP;
4090 LASSERT_SEQ_IS_MDT(oscc->oscc_oa.o_seq);
4091 req->rq_no_delay = req->rq_no_resend = 1;
4092 req->rq_interpret_reply = osc_setinfo_mds_conn_interpret;
4093 } else if (KEY_IS(KEY_GRANT_SHRINK)) {
4094 struct osc_grant_args *aa;
4097 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
4098 aa = ptlrpc_req_async_args(req);
4101 ptlrpc_req_finished(req);
4104 *oa = ((struct ost_body *)val)->oa;
4106 req->rq_interpret_reply = osc_shrink_grant_interpret;
4109 ptlrpc_request_set_replen(req);
4110 if (!KEY_IS(KEY_GRANT_SHRINK)) {
4111 LASSERT(set != NULL);
4112 ptlrpc_set_add_req(set, req);
4113 ptlrpc_check_set(NULL, set);
4115 ptlrpcd_add_req(req, PSCOPE_OTHER);
4121 static struct llog_operations osc_size_repl_logops = {
4122 lop_cancel: llog_obd_repl_cancel
4125 static struct llog_operations osc_mds_ost_orig_logops;
4127 static int __osc_llog_init(struct obd_device *obd, struct obd_llog_group *olg,
4128 struct obd_device *tgt, struct llog_catid *catid)
4133 rc = llog_setup(obd, &obd->obd_olg, LLOG_MDS_OST_ORIG_CTXT, tgt, 1,
4134 &catid->lci_logid, &osc_mds_ost_orig_logops);
4136 CERROR("failed LLOG_MDS_OST_ORIG_CTXT\n");
4140 rc = llog_setup(obd, &obd->obd_olg, LLOG_SIZE_REPL_CTXT, tgt, 1,
4141 NULL, &osc_size_repl_logops);
4143 struct llog_ctxt *ctxt =
4144 llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
4147 CERROR("failed LLOG_SIZE_REPL_CTXT\n");
4152 CERROR("osc '%s' tgt '%s' catid %p rc=%d\n",
4153 obd->obd_name, tgt->obd_name, catid, rc);
4154 CERROR("logid "LPX64":0x%x\n",
4155 catid->lci_logid.lgl_oid, catid->lci_logid.lgl_ogen);
4160 static int osc_llog_init(struct obd_device *obd, struct obd_llog_group *olg,
4161 struct obd_device *disk_obd, int *index)
4163 struct llog_catid catid;
4164 static char name[32] = CATLIST;
4168 LASSERT(olg == &obd->obd_olg);
4170 cfs_mutex_down(&olg->olg_cat_processing);
4171 rc = llog_get_cat_list(disk_obd, name, *index, 1, &catid);
4173 CERROR("rc: %d\n", rc);
4177 CDEBUG(D_INFO, "%s: Init llog for %d - catid "LPX64"/"LPX64":%x\n",
4178 obd->obd_name, *index, catid.lci_logid.lgl_oid,
4179 catid.lci_logid.lgl_oseq, catid.lci_logid.lgl_ogen);
4181 rc = __osc_llog_init(obd, olg, disk_obd, &catid);
4183 CERROR("rc: %d\n", rc);
4187 rc = llog_put_cat_list(disk_obd, name, *index, 1, &catid);
4189 CERROR("rc: %d\n", rc);
4194 cfs_mutex_up(&olg->olg_cat_processing);
4199 static int osc_llog_finish(struct obd_device *obd, int count)
4201 struct llog_ctxt *ctxt;
4202 int rc = 0, rc2 = 0;
4205 ctxt = llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
4207 rc = llog_cleanup(ctxt);
4209 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
4211 rc2 = llog_cleanup(ctxt);
4218 static int osc_reconnect(const struct lu_env *env,
4219 struct obd_export *exp, struct obd_device *obd,
4220 struct obd_uuid *cluuid,
4221 struct obd_connect_data *data,
4224 struct client_obd *cli = &obd->u.cli;
4226 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
4229 client_obd_list_lock(&cli->cl_loi_list_lock);
4230 data->ocd_grant = (cli->cl_avail_grant + cli->cl_dirty) ?:
4231 2 * cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT;
4232 lost_grant = cli->cl_lost_grant;
4233 cli->cl_lost_grant = 0;
4234 client_obd_list_unlock(&cli->cl_loi_list_lock);
4236 CDEBUG(D_CACHE, "request ocd_grant: %d cl_avail_grant: %ld "
4237 "cl_dirty: %ld cl_lost_grant: %ld\n", data->ocd_grant,
4238 cli->cl_avail_grant, cli->cl_dirty, lost_grant);
4239 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
4240 " ocd_grant: %d\n", data->ocd_connect_flags,
4241 data->ocd_version, data->ocd_grant);
4247 static int osc_disconnect(struct obd_export *exp)
4249 struct obd_device *obd = class_exp2obd(exp);
4250 struct llog_ctxt *ctxt;
4253 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
4255 if (obd->u.cli.cl_conn_count == 1) {
4256 /* Flush any remaining cancel messages out to the
4258 llog_sync(ctxt, exp);
4260 llog_ctxt_put(ctxt);
4262 CDEBUG(D_HA, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
4266 rc = client_disconnect_export(exp);
4268 * Initially we put del_shrink_grant before disconnect_export, but it
4269 * causes the following problem if setup (connect) and cleanup
4270 * (disconnect) are tangled together.
4271 * connect p1 disconnect p2
4272 * ptlrpc_connect_import
4273 * ............... class_manual_cleanup
4276 * ptlrpc_connect_interrupt
4278 * add this client to shrink list
4280 * Bang! pinger trigger the shrink.
4281 * So the osc should be disconnected from the shrink list, after we
4282 * are sure the import has been destroyed. BUG18662
4284 if (obd->u.cli.cl_import == NULL)
4285 osc_del_shrink_grant(&obd->u.cli);
4289 static int osc_import_event(struct obd_device *obd,
4290 struct obd_import *imp,
4291 enum obd_import_event event)
4293 struct client_obd *cli;
4297 LASSERT(imp->imp_obd == obd);
4300 case IMP_EVENT_DISCON: {
4301 /* Only do this on the MDS OSC's */
4302 if (imp->imp_server_timeout) {
4303 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4305 cfs_spin_lock(&oscc->oscc_lock);
4306 oscc->oscc_flags |= OSCC_FLAG_RECOVERING;
4307 cfs_spin_unlock(&oscc->oscc_lock);
4310 client_obd_list_lock(&cli->cl_loi_list_lock);
4311 cli->cl_avail_grant = 0;
4312 cli->cl_lost_grant = 0;
4313 client_obd_list_unlock(&cli->cl_loi_list_lock);
4316 case IMP_EVENT_INACTIVE: {
4317 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
4320 case IMP_EVENT_INVALIDATE: {
4321 struct ldlm_namespace *ns = obd->obd_namespace;
4325 env = cl_env_get(&refcheck);
4329 client_obd_list_lock(&cli->cl_loi_list_lock);
4330 /* all pages go to failing rpcs due to the invalid
4332 osc_check_rpcs(env, cli);
4333 client_obd_list_unlock(&cli->cl_loi_list_lock);
4335 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
4336 cl_env_put(env, &refcheck);
4341 case IMP_EVENT_ACTIVE: {
4342 /* Only do this on the MDS OSC's */
4343 if (imp->imp_server_timeout) {
4344 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4346 cfs_spin_lock(&oscc->oscc_lock);
4347 oscc->oscc_flags &= ~OSCC_FLAG_NOSPC;
4348 cfs_spin_unlock(&oscc->oscc_lock);
4350 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
4353 case IMP_EVENT_OCD: {
4354 struct obd_connect_data *ocd = &imp->imp_connect_data;
4356 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
4357 osc_init_grant(&obd->u.cli, ocd);
4360 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
4361 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
4363 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
4367 CERROR("Unknown import event %d\n", event);
4374 * Determine whether the lock can be canceled before replaying the lock
4375 * during recovery, see bug16774 for detailed information.
4377 * \retval zero the lock can't be canceled
4378 * \retval other ok to cancel
4380 static int osc_cancel_for_recovery(struct ldlm_lock *lock)
4382 check_res_locked(lock->l_resource);
4385 * Cancel all unused extent lock in granted mode LCK_PR or LCK_CR.
4387 * XXX as a future improvement, we can also cancel unused write lock
4388 * if it doesn't have dirty data and active mmaps.
4390 if (lock->l_resource->lr_type == LDLM_EXTENT &&
4391 (lock->l_granted_mode == LCK_PR ||
4392 lock->l_granted_mode == LCK_CR) &&
4393 (osc_dlm_lock_pageref(lock) == 0))
4399 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
4405 rc = ptlrpcd_addref();
4409 rc = client_obd_setup(obd, lcfg);
4413 struct lprocfs_static_vars lvars = { 0 };
4414 struct client_obd *cli = &obd->u.cli;
4416 cli->cl_grant_shrink_interval = GRANT_SHRINK_INTERVAL;
4417 lprocfs_osc_init_vars(&lvars);
4418 if (lprocfs_obd_setup(obd, lvars.obd_vars) == 0) {
4419 lproc_osc_attach_seqstat(obd);
4420 sptlrpc_lprocfs_cliobd_attach(obd);
4421 ptlrpc_lprocfs_register_obd(obd);
4425 /* We need to allocate a few requests more, because
4426 brw_interpret tries to create new requests before freeing
4427 previous ones. Ideally we want to have 2x max_rpcs_in_flight
4428 reserved, but I afraid that might be too much wasted RAM
4429 in fact, so 2 is just my guess and still should work. */
4430 cli->cl_import->imp_rq_pool =
4431 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
4433 ptlrpc_add_rqs_to_pool);
4435 CFS_INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
4436 cfs_sema_init(&cli->cl_grant_sem, 1);
4438 ns_register_cancel(obd->obd_namespace, osc_cancel_for_recovery);
4444 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
4450 case OBD_CLEANUP_EARLY: {
4451 struct obd_import *imp;
4452 imp = obd->u.cli.cl_import;
4453 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
4454 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
4455 ptlrpc_deactivate_import(imp);
4456 cfs_spin_lock(&imp->imp_lock);
4457 imp->imp_pingable = 0;
4458 cfs_spin_unlock(&imp->imp_lock);
4461 case OBD_CLEANUP_EXPORTS: {
4462 /* If we set up but never connected, the
4463 client import will not have been cleaned. */
4464 if (obd->u.cli.cl_import) {
4465 struct obd_import *imp;
4466 cfs_down_write(&obd->u.cli.cl_sem);
4467 imp = obd->u.cli.cl_import;
4468 CDEBUG(D_CONFIG, "%s: client import never connected\n",
4470 ptlrpc_invalidate_import(imp);
4471 if (imp->imp_rq_pool) {
4472 ptlrpc_free_rq_pool(imp->imp_rq_pool);
4473 imp->imp_rq_pool = NULL;
4475 class_destroy_import(imp);
4476 cfs_up_write(&obd->u.cli.cl_sem);
4477 obd->u.cli.cl_import = NULL;
4479 rc = obd_llog_finish(obd, 0);
4481 CERROR("failed to cleanup llogging subsystems\n");
4488 int osc_cleanup(struct obd_device *obd)
4493 ptlrpc_lprocfs_unregister_obd(obd);
4494 lprocfs_obd_cleanup(obd);
4496 /* free memory of osc quota cache */
4497 lquota_cleanup(quota_interface, obd);
4499 rc = client_obd_cleanup(obd);
4505 int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
4507 struct lprocfs_static_vars lvars = { 0 };
4510 lprocfs_osc_init_vars(&lvars);
4512 switch (lcfg->lcfg_command) {
4514 rc = class_process_proc_param(PARAM_OSC, lvars.obd_vars,
4524 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
4526 return osc_process_config_base(obd, buf);
4529 struct obd_ops osc_obd_ops = {
4530 .o_owner = THIS_MODULE,
4531 .o_setup = osc_setup,
4532 .o_precleanup = osc_precleanup,
4533 .o_cleanup = osc_cleanup,
4534 .o_add_conn = client_import_add_conn,
4535 .o_del_conn = client_import_del_conn,
4536 .o_connect = client_connect_import,
4537 .o_reconnect = osc_reconnect,
4538 .o_disconnect = osc_disconnect,
4539 .o_statfs = osc_statfs,
4540 .o_statfs_async = osc_statfs_async,
4541 .o_packmd = osc_packmd,
4542 .o_unpackmd = osc_unpackmd,
4543 .o_precreate = osc_precreate,
4544 .o_create = osc_create,
4545 .o_create_async = osc_create_async,
4546 .o_destroy = osc_destroy,
4547 .o_getattr = osc_getattr,
4548 .o_getattr_async = osc_getattr_async,
4549 .o_setattr = osc_setattr,
4550 .o_setattr_async = osc_setattr_async,
4552 .o_punch = osc_punch,
4554 .o_enqueue = osc_enqueue,
4555 .o_change_cbdata = osc_change_cbdata,
4556 .o_find_cbdata = osc_find_cbdata,
4557 .o_cancel = osc_cancel,
4558 .o_cancel_unused = osc_cancel_unused,
4559 .o_iocontrol = osc_iocontrol,
4560 .o_get_info = osc_get_info,
4561 .o_set_info_async = osc_set_info_async,
4562 .o_import_event = osc_import_event,
4563 .o_llog_init = osc_llog_init,
4564 .o_llog_finish = osc_llog_finish,
4565 .o_process_config = osc_process_config,
4568 extern struct lu_kmem_descr osc_caches[];
4569 extern cfs_spinlock_t osc_ast_guard;
4570 extern cfs_lock_class_key_t osc_ast_guard_class;
4572 int __init osc_init(void)
4574 struct lprocfs_static_vars lvars = { 0 };
4578 /* print an address of _any_ initialized kernel symbol from this
4579 * module, to allow debugging with gdb that doesn't support data
4580 * symbols from modules.*/
4581 CDEBUG(D_CONSOLE, "Lustre OSC module (%p).\n", &osc_caches);
4583 rc = lu_kmem_init(osc_caches);
4585 lprocfs_osc_init_vars(&lvars);
4587 cfs_request_module("lquota");
4588 quota_interface = PORTAL_SYMBOL_GET(osc_quota_interface);
4589 lquota_init(quota_interface);
4590 init_obd_quota_ops(quota_interface, &osc_obd_ops);
4592 rc = class_register_type(&osc_obd_ops, NULL, lvars.module_vars,
4593 LUSTRE_OSC_NAME, &osc_device_type);
4595 if (quota_interface)
4596 PORTAL_SYMBOL_PUT(osc_quota_interface);
4597 lu_kmem_fini(osc_caches);
4601 cfs_spin_lock_init(&osc_ast_guard);
4602 cfs_lockdep_set_class(&osc_ast_guard, &osc_ast_guard_class);
4604 osc_mds_ost_orig_logops = llog_lvfs_ops;
4605 osc_mds_ost_orig_logops.lop_setup = llog_obd_origin_setup;
4606 osc_mds_ost_orig_logops.lop_cleanup = llog_obd_origin_cleanup;
4607 osc_mds_ost_orig_logops.lop_add = llog_obd_origin_add;
4608 osc_mds_ost_orig_logops.lop_connect = llog_origin_connect;
4614 static void /*__exit*/ osc_exit(void)
4616 lu_device_type_fini(&osc_device_type);
4618 lquota_exit(quota_interface);
4619 if (quota_interface)
4620 PORTAL_SYMBOL_PUT(osc_quota_interface);
4622 class_unregister_type(LUSTRE_OSC_NAME);
4623 lu_kmem_fini(osc_caches);
4626 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
4627 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
4628 MODULE_LICENSE("GPL");
4630 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);