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 2008 Sun Microsystems, Inc. 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_MDS_GROUP(lsm->lsm_object_gr);
100 (*lmmp)->lmm_object_id = cpu_to_le64(lsm->lsm_object_id);
101 (*lmmp)->lmm_object_gr = cpu_to_le64(lsm->lsm_object_gr);
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_gr = le64_to_cpu (lmm->lmm_object_gr);
155 LASSERT((*lsmp)->lsm_object_id);
156 LASSERT_MDS_GROUP((*lsmp)->lsm_object_gr);
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 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 = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
214 lustre_swab_ost_body);
216 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
217 memcpy(aa->aa_oi->oi_oa, &body->oa, sizeof(*aa->aa_oi->oi_oa));
219 /* This should really be sent by the OST */
220 aa->aa_oi->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
221 aa->aa_oi->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
223 CDEBUG(D_INFO, "can't unpack ost_body\n");
225 aa->aa_oi->oi_oa->o_valid = 0;
228 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
232 static int osc_getattr_async(struct obd_export *exp, struct obd_info *oinfo,
233 struct ptlrpc_request_set *set)
235 struct ptlrpc_request *req;
236 struct osc_async_args *aa;
240 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
244 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
245 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
247 ptlrpc_request_free(req);
251 osc_pack_req_body(req, oinfo);
253 ptlrpc_request_set_replen(req);
254 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_getattr_interpret;
256 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
257 aa = ptlrpc_req_async_args(req);
260 ptlrpc_set_add_req(set, req);
264 static int osc_getattr(struct obd_export *exp, struct obd_info *oinfo)
266 struct ptlrpc_request *req;
267 struct ost_body *body;
271 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_GETATTR);
275 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
276 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GETATTR);
278 ptlrpc_request_free(req);
282 osc_pack_req_body(req, oinfo);
284 ptlrpc_request_set_replen(req);
286 rc = ptlrpc_queue_wait(req);
290 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
292 GOTO(out, rc = -EPROTO);
294 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
295 *oinfo->oi_oa = body->oa;
297 /* This should really be sent by the OST */
298 oinfo->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
299 oinfo->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
303 ptlrpc_req_finished(req);
307 static int osc_setattr(struct obd_export *exp, struct obd_info *oinfo,
308 struct obd_trans_info *oti)
310 struct ptlrpc_request *req;
311 struct ost_body *body;
315 LASSERTF(!(oinfo->oi_oa->o_valid & OBD_MD_FLGROUP) ||
316 CHECK_MDS_GROUP(oinfo->oi_oa->o_gr),
317 "oinfo->oi_oa->o_valid="LPU64" oinfo->oi_oa->o_gr="LPU64"\n",
318 oinfo->oi_oa->o_valid, oinfo->oi_oa->o_gr);
320 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
324 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
325 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
327 ptlrpc_request_free(req);
331 osc_pack_req_body(req, oinfo);
333 ptlrpc_request_set_replen(req);
335 rc = ptlrpc_queue_wait(req);
339 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
341 GOTO(out, rc = -EPROTO);
343 *oinfo->oi_oa = body->oa;
347 ptlrpc_req_finished(req);
351 static int osc_setattr_interpret(const struct lu_env *env,
352 struct ptlrpc_request *req,
353 struct osc_async_args *aa, int rc)
355 struct ost_body *body;
361 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
363 GOTO(out, rc = -EPROTO);
365 *aa->aa_oi->oi_oa = body->oa;
367 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
371 static int osc_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
372 struct obd_trans_info *oti,
373 struct ptlrpc_request_set *rqset)
375 struct ptlrpc_request *req;
376 struct osc_async_args *aa;
380 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SETATTR);
384 osc_set_capa_size(req, &RMF_CAPA1, oinfo->oi_capa);
385 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SETATTR);
387 ptlrpc_request_free(req);
391 osc_pack_req_body(req, oinfo);
393 ptlrpc_request_set_replen(req);
395 if (oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE) {
397 oinfo->oi_oa->o_lcookie = *oti->oti_logcookies;
400 /* do mds to ost setattr asynchronously */
402 /* Do not wait for response. */
403 ptlrpcd_add_req(req, PSCOPE_OTHER);
405 req->rq_interpret_reply =
406 (ptlrpc_interpterer_t)osc_setattr_interpret;
408 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
409 aa = ptlrpc_req_async_args(req);
412 ptlrpc_set_add_req(rqset, req);
418 int osc_real_create(struct obd_export *exp, struct obdo *oa,
419 struct lov_stripe_md **ea, struct obd_trans_info *oti)
421 struct ptlrpc_request *req;
422 struct ost_body *body;
423 struct lov_stripe_md *lsm;
432 rc = obd_alloc_memmd(exp, &lsm);
437 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_CREATE);
439 GOTO(out, rc = -ENOMEM);
441 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_CREATE);
443 ptlrpc_request_free(req);
447 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
451 ptlrpc_request_set_replen(req);
453 if ((oa->o_valid & OBD_MD_FLFLAGS) &&
454 oa->o_flags == OBD_FL_DELORPHAN) {
456 "delorphan from OST integration");
457 /* Don't resend the delorphan req */
458 req->rq_no_resend = req->rq_no_delay = 1;
461 rc = ptlrpc_queue_wait(req);
465 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
467 GOTO(out_req, rc = -EPROTO);
471 /* This should really be sent by the OST */
472 oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
473 oa->o_valid |= OBD_MD_FLBLKSZ;
475 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
476 * have valid lsm_oinfo data structs, so don't go touching that.
477 * This needs to be fixed in a big way.
479 lsm->lsm_object_id = oa->o_id;
480 lsm->lsm_object_gr = oa->o_gr;
484 oti->oti_transno = lustre_msg_get_transno(req->rq_repmsg);
486 if (oa->o_valid & OBD_MD_FLCOOKIE) {
487 if (!oti->oti_logcookies)
488 oti_alloc_cookies(oti, 1);
489 *oti->oti_logcookies = oa->o_lcookie;
493 CDEBUG(D_HA, "transno: "LPD64"\n",
494 lustre_msg_get_transno(req->rq_repmsg));
496 ptlrpc_req_finished(req);
499 obd_free_memmd(exp, &lsm);
503 static int osc_punch_interpret(const struct lu_env *env,
504 struct ptlrpc_request *req,
505 struct osc_punch_args *aa, int rc)
507 struct ost_body *body;
513 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
515 GOTO(out, rc = -EPROTO);
517 *aa->pa_oa = body->oa;
519 rc = aa->pa_upcall(aa->pa_cookie, rc);
523 int osc_punch_base(struct obd_export *exp, struct obdo *oa,
524 struct obd_capa *capa,
525 obd_enqueue_update_f upcall, void *cookie,
526 struct ptlrpc_request_set *rqset)
528 struct ptlrpc_request *req;
529 struct osc_punch_args *aa;
530 struct ost_body *body;
534 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_PUNCH);
538 osc_set_capa_size(req, &RMF_CAPA1, capa);
539 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_PUNCH);
541 ptlrpc_request_free(req);
544 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
545 ptlrpc_at_set_req_timeout(req);
547 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
550 osc_pack_capa(req, body, capa);
552 ptlrpc_request_set_replen(req);
555 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_punch_interpret;
556 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
557 aa = ptlrpc_req_async_args(req);
559 aa->pa_upcall = upcall;
560 aa->pa_cookie = cookie;
561 if (rqset == PTLRPCD_SET)
562 ptlrpcd_add_req(req, PSCOPE_OTHER);
564 ptlrpc_set_add_req(rqset, req);
569 static int osc_punch(struct obd_export *exp, struct obd_info *oinfo,
570 struct obd_trans_info *oti,
571 struct ptlrpc_request_set *rqset)
573 oinfo->oi_oa->o_size = oinfo->oi_policy.l_extent.start;
574 oinfo->oi_oa->o_blocks = oinfo->oi_policy.l_extent.end;
575 oinfo->oi_oa->o_valid |= OBD_MD_FLSIZE | OBD_MD_FLBLOCKS;
576 return osc_punch_base(exp, oinfo->oi_oa, oinfo->oi_capa,
577 oinfo->oi_cb_up, oinfo, rqset);
580 static int osc_sync(struct obd_export *exp, struct obdo *oa,
581 struct lov_stripe_md *md, obd_size start, obd_size end,
584 struct ptlrpc_request *req;
585 struct ost_body *body;
590 CDEBUG(D_INFO, "oa NULL\n");
594 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_SYNC);
598 osc_set_capa_size(req, &RMF_CAPA1, capa);
599 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SYNC);
601 ptlrpc_request_free(req);
605 /* overload the size and blocks fields in the oa with start/end */
606 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
609 body->oa.o_size = start;
610 body->oa.o_blocks = end;
611 body->oa.o_valid |= (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS);
612 osc_pack_capa(req, body, capa);
614 ptlrpc_request_set_replen(req);
616 rc = ptlrpc_queue_wait(req);
620 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
622 GOTO(out, rc = -EPROTO);
628 ptlrpc_req_finished(req);
632 /* Find and cancel locally locks matched by @mode in the resource found by
633 * @objid. Found locks are added into @cancel list. Returns the amount of
634 * locks added to @cancels list. */
635 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
636 struct list_head *cancels, ldlm_mode_t mode,
639 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
640 struct ldlm_res_id res_id;
641 struct ldlm_resource *res;
645 osc_build_res_name(oa->o_id, oa->o_gr, &res_id);
646 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
650 LDLM_RESOURCE_ADDREF(res);
651 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
652 lock_flags, 0, NULL);
653 LDLM_RESOURCE_DELREF(res);
654 ldlm_resource_putref(res);
658 static int osc_destroy_interpret(const struct lu_env *env,
659 struct ptlrpc_request *req, void *data,
662 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
664 atomic_dec(&cli->cl_destroy_in_flight);
665 cfs_waitq_signal(&cli->cl_destroy_waitq);
669 static int osc_can_send_destroy(struct client_obd *cli)
671 if (atomic_inc_return(&cli->cl_destroy_in_flight) <=
672 cli->cl_max_rpcs_in_flight) {
673 /* The destroy request can be sent */
676 if (atomic_dec_return(&cli->cl_destroy_in_flight) <
677 cli->cl_max_rpcs_in_flight) {
679 * The counter has been modified between the two atomic
682 cfs_waitq_signal(&cli->cl_destroy_waitq);
687 /* Destroy requests can be async always on the client, and we don't even really
688 * care about the return code since the client cannot do anything at all about
690 * When the MDS is unlinking a filename, it saves the file objects into a
691 * recovery llog, and these object records are cancelled when the OST reports
692 * they were destroyed and sync'd to disk (i.e. transaction committed).
693 * If the client dies, or the OST is down when the object should be destroyed,
694 * the records are not cancelled, and when the OST reconnects to the MDS next,
695 * it will retrieve the llog unlink logs and then sends the log cancellation
696 * cookies to the MDS after committing destroy transactions. */
697 static int osc_destroy(struct obd_export *exp, struct obdo *oa,
698 struct lov_stripe_md *ea, struct obd_trans_info *oti,
699 struct obd_export *md_export, void *capa)
701 struct client_obd *cli = &exp->exp_obd->u.cli;
702 struct ptlrpc_request *req;
703 struct ost_body *body;
704 CFS_LIST_HEAD(cancels);
709 CDEBUG(D_INFO, "oa NULL\n");
713 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
714 LDLM_FL_DISCARD_DATA);
716 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_OST_DESTROY);
718 ldlm_lock_list_put(&cancels, l_bl_ast, count);
722 osc_set_capa_size(req, &RMF_CAPA1, (struct obd_capa *)capa);
723 rc = ldlm_prep_elc_req(exp, req, LUSTRE_OST_VERSION, OST_DESTROY,
726 ptlrpc_request_free(req);
730 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
731 ptlrpc_at_set_req_timeout(req);
733 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
734 oa->o_lcookie = *oti->oti_logcookies;
735 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
739 osc_pack_capa(req, body, (struct obd_capa *)capa);
740 ptlrpc_request_set_replen(req);
742 /* don't throttle destroy RPCs for the MDT */
743 if (!(cli->cl_import->imp_connect_flags_orig & OBD_CONNECT_MDS)) {
744 req->rq_interpret_reply = osc_destroy_interpret;
745 if (!osc_can_send_destroy(cli)) {
746 struct l_wait_info lwi = { 0 };
749 * Wait until the number of on-going destroy RPCs drops
750 * under max_rpc_in_flight
752 l_wait_event_exclusive(cli->cl_destroy_waitq,
753 osc_can_send_destroy(cli), &lwi);
757 /* Do not wait for response */
758 ptlrpcd_add_req(req, PSCOPE_OTHER);
762 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
765 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
767 LASSERT(!(oa->o_valid & bits));
770 client_obd_list_lock(&cli->cl_loi_list_lock);
771 oa->o_dirty = cli->cl_dirty;
772 if (cli->cl_dirty - cli->cl_dirty_transit > cli->cl_dirty_max) {
773 CERROR("dirty %lu - %lu > dirty_max %lu\n",
774 cli->cl_dirty, cli->cl_dirty_transit, cli->cl_dirty_max);
776 } else if (atomic_read(&obd_dirty_pages) -
777 atomic_read(&obd_dirty_transit_pages) > obd_max_dirty_pages){
778 CERROR("dirty %d - %d > system dirty_max %d\n",
779 atomic_read(&obd_dirty_pages),
780 atomic_read(&obd_dirty_transit_pages),
781 obd_max_dirty_pages);
783 } else if (cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff) {
784 CERROR("dirty %lu - dirty_max %lu too big???\n",
785 cli->cl_dirty, cli->cl_dirty_max);
788 long max_in_flight = (cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT)*
789 (cli->cl_max_rpcs_in_flight + 1);
790 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
792 oa->o_grant = cli->cl_avail_grant;
793 oa->o_dropped = cli->cl_lost_grant;
794 cli->cl_lost_grant = 0;
795 client_obd_list_unlock(&cli->cl_loi_list_lock);
796 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
797 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
801 static void osc_update_next_shrink(struct client_obd *cli)
803 int time = GRANT_SHRINK_INTERVAL;
804 cli->cl_next_shrink_grant = cfs_time_shift(time);
805 CDEBUG(D_CACHE, "next time %ld to shrink grant \n",
806 cli->cl_next_shrink_grant);
809 /* caller must hold loi_list_lock */
810 static void osc_consume_write_grant(struct client_obd *cli,
811 struct brw_page *pga)
813 LASSERT_SPIN_LOCKED(&cli->cl_loi_list_lock);
814 LASSERT(!(pga->flag & OBD_BRW_FROM_GRANT));
815 atomic_inc(&obd_dirty_pages);
816 cli->cl_dirty += CFS_PAGE_SIZE;
817 cli->cl_avail_grant -= CFS_PAGE_SIZE;
818 pga->flag |= OBD_BRW_FROM_GRANT;
819 CDEBUG(D_CACHE, "using %lu grant credits for brw %p page %p\n",
820 CFS_PAGE_SIZE, pga, pga->pg);
821 LASSERT(cli->cl_avail_grant >= 0);
822 osc_update_next_shrink(cli);
825 /* the companion to osc_consume_write_grant, called when a brw has completed.
826 * must be called with the loi lock held. */
827 static void osc_release_write_grant(struct client_obd *cli,
828 struct brw_page *pga, int sent)
830 int blocksize = cli->cl_import->imp_obd->obd_osfs.os_bsize ? : 4096;
833 LASSERT_SPIN_LOCKED(&cli->cl_loi_list_lock);
834 if (!(pga->flag & OBD_BRW_FROM_GRANT)) {
839 pga->flag &= ~OBD_BRW_FROM_GRANT;
840 atomic_dec(&obd_dirty_pages);
841 cli->cl_dirty -= CFS_PAGE_SIZE;
842 if (pga->flag & OBD_BRW_NOCACHE) {
843 pga->flag &= ~OBD_BRW_NOCACHE;
844 atomic_dec(&obd_dirty_transit_pages);
845 cli->cl_dirty_transit -= CFS_PAGE_SIZE;
848 cli->cl_lost_grant += CFS_PAGE_SIZE;
849 CDEBUG(D_CACHE, "lost grant: %lu avail grant: %lu dirty: %lu\n",
850 cli->cl_lost_grant, cli->cl_avail_grant, cli->cl_dirty);
851 } else if (CFS_PAGE_SIZE != blocksize && pga->count != CFS_PAGE_SIZE) {
852 /* For short writes we shouldn't count parts of pages that
853 * span a whole block on the OST side, or our accounting goes
854 * wrong. Should match the code in filter_grant_check. */
855 int offset = pga->off & ~CFS_PAGE_MASK;
856 int count = pga->count + (offset & (blocksize - 1));
857 int end = (offset + pga->count) & (blocksize - 1);
859 count += blocksize - end;
861 cli->cl_lost_grant += CFS_PAGE_SIZE - count;
862 CDEBUG(D_CACHE, "lost %lu grant: %lu avail: %lu dirty: %lu\n",
863 CFS_PAGE_SIZE - count, cli->cl_lost_grant,
864 cli->cl_avail_grant, cli->cl_dirty);
870 static unsigned long rpcs_in_flight(struct client_obd *cli)
872 return cli->cl_r_in_flight + cli->cl_w_in_flight;
875 /* caller must hold loi_list_lock */
876 void osc_wake_cache_waiters(struct client_obd *cli)
878 struct list_head *l, *tmp;
879 struct osc_cache_waiter *ocw;
882 list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
883 /* if we can't dirty more, we must wait until some is written */
884 if ((cli->cl_dirty + CFS_PAGE_SIZE > cli->cl_dirty_max) ||
885 (atomic_read(&obd_dirty_pages) + 1 > obd_max_dirty_pages)) {
886 CDEBUG(D_CACHE, "no dirty room: dirty: %ld "
887 "osc max %ld, sys max %d\n", cli->cl_dirty,
888 cli->cl_dirty_max, obd_max_dirty_pages);
892 /* if still dirty cache but no grant wait for pending RPCs that
893 * may yet return us some grant before doing sync writes */
894 if (cli->cl_w_in_flight && cli->cl_avail_grant < CFS_PAGE_SIZE) {
895 CDEBUG(D_CACHE, "%u BRW writes in flight, no grant\n",
896 cli->cl_w_in_flight);
900 ocw = list_entry(l, struct osc_cache_waiter, ocw_entry);
901 list_del_init(&ocw->ocw_entry);
902 if (cli->cl_avail_grant < CFS_PAGE_SIZE) {
903 /* no more RPCs in flight to return grant, do sync IO */
904 ocw->ocw_rc = -EDQUOT;
905 CDEBUG(D_INODE, "wake oap %p for sync\n", ocw->ocw_oap);
907 osc_consume_write_grant(cli,
908 &ocw->ocw_oap->oap_brw_page);
911 cfs_waitq_signal(&ocw->ocw_waitq);
917 static void __osc_update_grant(struct client_obd *cli, obd_size grant)
919 client_obd_list_lock(&cli->cl_loi_list_lock);
920 cli->cl_avail_grant += grant;
921 client_obd_list_unlock(&cli->cl_loi_list_lock);
924 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
926 if (body->oa.o_valid & OBD_MD_FLGRANT) {
927 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
928 __osc_update_grant(cli, body->oa.o_grant);
932 static int osc_set_info_async(struct obd_export *exp, obd_count keylen,
933 void *key, obd_count vallen, void *val,
934 struct ptlrpc_request_set *set);
936 static int osc_shrink_grant_interpret(const struct lu_env *env,
937 struct ptlrpc_request *req,
940 struct client_obd *cli = &req->rq_import->imp_obd->u.cli;
941 struct obdo *oa = ((struct osc_grant_args *)aa)->aa_oa;
942 struct ost_body *body;
945 __osc_update_grant(cli, oa->o_grant);
949 body = req_capsule_server_get(&req->rq_pill, &RMF_OST_BODY);
951 osc_update_grant(cli, body);
957 static void osc_shrink_grant_local(struct client_obd *cli, struct obdo *oa)
959 client_obd_list_lock(&cli->cl_loi_list_lock);
960 oa->o_grant = cli->cl_avail_grant / 4;
961 cli->cl_avail_grant -= oa->o_grant;
962 client_obd_list_unlock(&cli->cl_loi_list_lock);
963 oa->o_flags |= OBD_FL_SHRINK_GRANT;
964 osc_update_next_shrink(cli);
967 static int osc_shrink_grant(struct client_obd *cli)
970 struct ost_body *body;
977 osc_announce_cached(cli, &body->oa, 0);
978 osc_shrink_grant_local(cli, &body->oa);
979 rc = osc_set_info_async(cli->cl_import->imp_obd->obd_self_export,
980 sizeof(KEY_GRANT_SHRINK), KEY_GRANT_SHRINK,
981 sizeof(*body), body, NULL);
983 __osc_update_grant(cli, body->oa.o_grant);
989 #define GRANT_SHRINK_LIMIT PTLRPC_MAX_BRW_SIZE
990 static int osc_should_shrink_grant(struct client_obd *client)
992 cfs_time_t time = cfs_time_current();
993 cfs_time_t next_shrink = client->cl_next_shrink_grant;
994 if (cfs_time_aftereq(time, next_shrink - 5 * CFS_TICK)) {
995 if (client->cl_import->imp_state == LUSTRE_IMP_FULL &&
996 client->cl_avail_grant > GRANT_SHRINK_LIMIT)
999 osc_update_next_shrink(client);
1004 static int osc_grant_shrink_grant_cb(struct timeout_item *item, void *data)
1006 struct client_obd *client;
1008 list_for_each_entry(client, &item->ti_obd_list, cl_grant_shrink_list) {
1009 if (osc_should_shrink_grant(client))
1010 osc_shrink_grant(client);
1015 static int osc_add_shrink_grant(struct client_obd *client)
1019 rc = ptlrpc_add_timeout_client(GRANT_SHRINK_INTERVAL,
1021 osc_grant_shrink_grant_cb, NULL,
1022 &client->cl_grant_shrink_list);
1024 CERROR("add grant client %s error %d\n",
1025 client->cl_import->imp_obd->obd_name, rc);
1028 CDEBUG(D_CACHE, "add grant client %s \n",
1029 client->cl_import->imp_obd->obd_name);
1030 osc_update_next_shrink(client);
1034 static int osc_del_shrink_grant(struct client_obd *client)
1036 return ptlrpc_del_timeout_client(&client->cl_grant_shrink_list,
1040 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
1042 client_obd_list_lock(&cli->cl_loi_list_lock);
1043 cli->cl_avail_grant = ocd->ocd_grant;
1044 client_obd_list_unlock(&cli->cl_loi_list_lock);
1046 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT_SHRINK &&
1047 list_empty(&cli->cl_grant_shrink_list))
1048 osc_add_shrink_grant(cli);
1050 CDEBUG(D_CACHE, "setting cl_avail_grant: %ld cl_lost_grant: %ld \n",
1051 cli->cl_avail_grant, cli->cl_lost_grant);
1052 LASSERT(cli->cl_avail_grant >= 0);
1055 /* We assume that the reason this OSC got a short read is because it read
1056 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1057 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1058 * this stripe never got written at or beyond this stripe offset yet. */
1059 static void handle_short_read(int nob_read, obd_count page_count,
1060 struct brw_page **pga)
1065 /* skip bytes read OK */
1066 while (nob_read > 0) {
1067 LASSERT (page_count > 0);
1069 if (pga[i]->count > nob_read) {
1070 /* EOF inside this page */
1071 ptr = cfs_kmap(pga[i]->pg) +
1072 (pga[i]->off & ~CFS_PAGE_MASK);
1073 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
1074 cfs_kunmap(pga[i]->pg);
1080 nob_read -= pga[i]->count;
1085 /* zero remaining pages */
1086 while (page_count-- > 0) {
1087 ptr = cfs_kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
1088 memset(ptr, 0, pga[i]->count);
1089 cfs_kunmap(pga[i]->pg);
1094 static int check_write_rcs(struct ptlrpc_request *req,
1095 int requested_nob, int niocount,
1096 obd_count page_count, struct brw_page **pga)
1100 /* return error if any niobuf was in error */
1101 remote_rcs = lustre_swab_repbuf(req, REQ_REC_OFF + 1,
1102 sizeof(*remote_rcs) * niocount, NULL);
1103 if (remote_rcs == NULL) {
1104 CDEBUG(D_INFO, "Missing/short RC vector on BRW_WRITE reply\n");
1107 if (lustre_msg_swabbed(req->rq_repmsg))
1108 for (i = 0; i < niocount; i++)
1109 __swab32s(&remote_rcs[i]);
1111 for (i = 0; i < niocount; i++) {
1112 if (remote_rcs[i] < 0)
1113 return(remote_rcs[i]);
1115 if (remote_rcs[i] != 0) {
1116 CDEBUG(D_INFO, "rc[%d] invalid (%d) req %p\n",
1117 i, remote_rcs[i], req);
1122 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
1123 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1124 req->rq_bulk->bd_nob_transferred, requested_nob);
1131 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
1133 if (p1->flag != p2->flag) {
1134 unsigned mask = ~(OBD_BRW_FROM_GRANT|
1135 OBD_BRW_NOCACHE|OBD_BRW_SYNC);
1137 /* warn if we try to combine flags that we don't know to be
1138 * safe to combine */
1139 if ((p1->flag & mask) != (p2->flag & mask))
1140 CERROR("is it ok to have flags 0x%x and 0x%x in the "
1141 "same brw?\n", p1->flag, p2->flag);
1145 return (p1->off + p1->count == p2->off);
1148 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
1149 struct brw_page **pga, int opc,
1150 cksum_type_t cksum_type)
1155 LASSERT (pg_count > 0);
1156 cksum = init_checksum(cksum_type);
1157 while (nob > 0 && pg_count > 0) {
1158 unsigned char *ptr = cfs_kmap(pga[i]->pg);
1159 int off = pga[i]->off & ~CFS_PAGE_MASK;
1160 int count = pga[i]->count > nob ? nob : pga[i]->count;
1162 /* corrupt the data before we compute the checksum, to
1163 * simulate an OST->client data error */
1164 if (i == 0 && opc == OST_READ &&
1165 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE))
1166 memcpy(ptr + off, "bad1", min(4, nob));
1167 cksum = compute_checksum(cksum, ptr + off, count, cksum_type);
1168 cfs_kunmap(pga[i]->pg);
1169 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d checksum %x\n",
1172 nob -= pga[i]->count;
1176 /* For sending we only compute the wrong checksum instead
1177 * of corrupting the data so it is still correct on a redo */
1178 if (opc == OST_WRITE && OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND))
1184 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
1185 struct lov_stripe_md *lsm, obd_count page_count,
1186 struct brw_page **pga,
1187 struct ptlrpc_request **reqp,
1188 struct obd_capa *ocapa, int reserve)
1190 struct ptlrpc_request *req;
1191 struct ptlrpc_bulk_desc *desc;
1192 struct ost_body *body;
1193 struct obd_ioobj *ioobj;
1194 struct niobuf_remote *niobuf;
1195 int niocount, i, requested_nob, opc, rc;
1196 struct osc_brw_async_args *aa;
1197 struct req_capsule *pill;
1198 struct brw_page *pg_prev;
1201 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ))
1202 RETURN(-ENOMEM); /* Recoverable */
1203 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2))
1204 RETURN(-EINVAL); /* Fatal */
1206 if ((cmd & OBD_BRW_WRITE) != 0) {
1208 req = ptlrpc_request_alloc_pool(cli->cl_import,
1209 cli->cl_import->imp_rq_pool,
1213 req = ptlrpc_request_alloc(cli->cl_import, &RQF_OST_BRW);
1218 for (niocount = i = 1; i < page_count; i++) {
1219 if (!can_merge_pages(pga[i - 1], pga[i]))
1223 pill = &req->rq_pill;
1224 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_CLIENT,
1225 niocount * sizeof(*niobuf));
1226 osc_set_capa_size(req, &RMF_CAPA1, ocapa);
1228 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, opc);
1230 ptlrpc_request_free(req);
1233 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
1234 ptlrpc_at_set_req_timeout(req);
1236 if (opc == OST_WRITE)
1237 desc = ptlrpc_prep_bulk_imp(req, page_count,
1238 BULK_GET_SOURCE, OST_BULK_PORTAL);
1240 desc = ptlrpc_prep_bulk_imp(req, page_count,
1241 BULK_PUT_SINK, OST_BULK_PORTAL);
1244 GOTO(out, rc = -ENOMEM);
1245 /* NB request now owns desc and will free it when it gets freed */
1247 body = req_capsule_client_get(pill, &RMF_OST_BODY);
1248 ioobj = req_capsule_client_get(pill, &RMF_OBD_IOOBJ);
1249 niobuf = req_capsule_client_get(pill, &RMF_NIOBUF_REMOTE);
1250 LASSERT(body && ioobj && niobuf);
1254 obdo_to_ioobj(oa, ioobj);
1255 ioobj->ioo_bufcnt = niocount;
1256 osc_pack_capa(req, body, ocapa);
1257 LASSERT (page_count > 0);
1259 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1260 struct brw_page *pg = pga[i];
1262 LASSERT(pg->count > 0);
1263 LASSERTF((pg->off & ~CFS_PAGE_MASK) + pg->count <= CFS_PAGE_SIZE,
1264 "i: %d pg: %p off: "LPU64", count: %u\n", i, pg,
1265 pg->off, pg->count);
1267 LASSERTF(i == 0 || pg->off > pg_prev->off,
1268 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1269 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1271 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1272 pg_prev->pg, page_private(pg_prev->pg),
1273 pg_prev->pg->index, pg_prev->off);
1275 LASSERTF(i == 0 || pg->off > pg_prev->off,
1276 "i %d p_c %u\n", i, page_count);
1278 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1279 (pg->flag & OBD_BRW_SRVLOCK));
1281 ptlrpc_prep_bulk_page(desc, pg->pg, pg->off & ~CFS_PAGE_MASK,
1283 requested_nob += pg->count;
1285 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1287 niobuf->len += pg->count;
1289 niobuf->offset = pg->off;
1290 niobuf->len = pg->count;
1291 niobuf->flags = pg->flag;
1296 LASSERTF((void *)(niobuf - niocount) ==
1297 lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF + 2,
1298 niocount * sizeof(*niobuf)),
1299 "want %p - real %p\n", lustre_msg_buf(req->rq_reqmsg,
1300 REQ_REC_OFF + 2, niocount * sizeof(*niobuf)),
1301 (void *)(niobuf - niocount));
1303 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1304 if (osc_should_shrink_grant(cli))
1305 osc_shrink_grant_local(cli, &body->oa);
1307 /* size[REQ_REC_OFF] still sizeof (*body) */
1308 if (opc == OST_WRITE) {
1309 if (unlikely(cli->cl_checksum) &&
1310 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1311 /* store cl_cksum_type in a local variable since
1312 * it can be changed via lprocfs */
1313 cksum_type_t cksum_type = cli->cl_cksum_type;
1315 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1316 oa->o_flags = body->oa.o_flags = 0;
1317 body->oa.o_flags |= cksum_type_pack(cksum_type);
1318 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1319 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1323 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1325 /* save this in 'oa', too, for later checking */
1326 oa->o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1327 oa->o_flags |= cksum_type_pack(cksum_type);
1329 /* clear out the checksum flag, in case this is a
1330 * resend but cl_checksum is no longer set. b=11238 */
1331 oa->o_valid &= ~OBD_MD_FLCKSUM;
1333 oa->o_cksum = body->oa.o_cksum;
1334 /* 1 RC per niobuf */
1335 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_SERVER,
1336 sizeof(__u32) * niocount);
1338 if (unlikely(cli->cl_checksum) &&
1339 !sptlrpc_flavor_has_bulk(&req->rq_flvr)) {
1340 if ((body->oa.o_valid & OBD_MD_FLFLAGS) == 0)
1341 body->oa.o_flags = 0;
1342 body->oa.o_flags |= cksum_type_pack(cli->cl_cksum_type);
1343 body->oa.o_valid |= OBD_MD_FLCKSUM | OBD_MD_FLFLAGS;
1345 req_capsule_set_size(pill, &RMF_NIOBUF_REMOTE, RCL_SERVER, 0);
1346 /* 1 RC for the whole I/O */
1348 ptlrpc_request_set_replen(req);
1350 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1351 aa = ptlrpc_req_async_args(req);
1353 aa->aa_requested_nob = requested_nob;
1354 aa->aa_nio_count = niocount;
1355 aa->aa_page_count = page_count;
1359 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1360 if (ocapa && reserve)
1361 aa->aa_ocapa = capa_get(ocapa);
1367 ptlrpc_req_finished(req);
1371 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1372 __u32 client_cksum, __u32 server_cksum, int nob,
1373 obd_count page_count, struct brw_page **pga,
1374 cksum_type_t client_cksum_type)
1378 cksum_type_t cksum_type;
1380 if (server_cksum == client_cksum) {
1381 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1385 if (oa->o_valid & OBD_MD_FLFLAGS)
1386 cksum_type = cksum_type_unpack(oa->o_flags);
1388 cksum_type = OBD_CKSUM_CRC32;
1390 new_cksum = osc_checksum_bulk(nob, page_count, pga, OST_WRITE,
1393 if (cksum_type != client_cksum_type)
1394 msg = "the server did not use the checksum type specified in "
1395 "the original request - likely a protocol problem";
1396 else if (new_cksum == server_cksum)
1397 msg = "changed on the client after we checksummed it - "
1398 "likely false positive due to mmap IO (bug 11742)";
1399 else if (new_cksum == client_cksum)
1400 msg = "changed in transit before arrival at OST";
1402 msg = "changed in transit AND doesn't match the original - "
1403 "likely false positive due to mmap IO (bug 11742)";
1405 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inum "
1406 LPU64"/"LPU64" object "LPU64"/"LPU64" extent "
1407 "["LPU64"-"LPU64"]\n",
1408 msg, libcfs_nid2str(peer->nid),
1409 oa->o_valid & OBD_MD_FLFID ? oa->o_fid : (__u64)0,
1410 oa->o_valid & OBD_MD_FLFID ? oa->o_generation :
1413 oa->o_valid & OBD_MD_FLGROUP ? oa->o_gr : (__u64)0,
1415 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1416 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1417 "client csum now %x\n", client_cksum, client_cksum_type,
1418 server_cksum, cksum_type, new_cksum);
1422 /* Note rc enters this function as number of bytes transferred */
1423 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1425 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1426 const lnet_process_id_t *peer =
1427 &req->rq_import->imp_connection->c_peer;
1428 struct client_obd *cli = aa->aa_cli;
1429 struct ost_body *body;
1430 __u32 client_cksum = 0;
1433 if (rc < 0 && rc != -EDQUOT)
1436 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1437 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
1438 lustre_swab_ost_body);
1440 CDEBUG(D_INFO, "Can't unpack body\n");
1444 /* set/clear over quota flag for a uid/gid */
1445 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1446 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA)) {
1447 unsigned int qid[MAXQUOTAS] = { body->oa.o_uid, body->oa.o_gid };
1449 lquota_setdq(quota_interface, cli, qid, body->oa.o_valid,
1456 if (aa->aa_oa->o_valid & OBD_MD_FLCKSUM)
1457 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1459 osc_update_grant(cli, body);
1461 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1463 CERROR("Unexpected +ve rc %d\n", rc);
1466 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1468 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1471 if ((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) && client_cksum &&
1472 check_write_checksum(&body->oa, peer, client_cksum,
1473 body->oa.o_cksum, aa->aa_requested_nob,
1474 aa->aa_page_count, aa->aa_ppga,
1475 cksum_type_unpack(aa->aa_oa->o_flags)))
1478 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1479 aa->aa_page_count, aa->aa_ppga);
1483 /* The rest of this function executes only for OST_READs */
1485 /* if unwrap_bulk failed, return -EAGAIN to retry */
1486 rc = sptlrpc_cli_unwrap_bulk_read(req, req->rq_bulk, rc);
1488 GOTO(out, rc = -EAGAIN);
1490 if (rc > aa->aa_requested_nob) {
1491 CERROR("Unexpected rc %d (%d requested)\n", rc,
1492 aa->aa_requested_nob);
1496 if (rc != req->rq_bulk->bd_nob_transferred) {
1497 CERROR ("Unexpected rc %d (%d transferred)\n",
1498 rc, req->rq_bulk->bd_nob_transferred);
1502 if (rc < aa->aa_requested_nob)
1503 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1505 if (body->oa.o_valid & OBD_MD_FLCKSUM) {
1506 static int cksum_counter;
1507 __u32 server_cksum = body->oa.o_cksum;
1510 cksum_type_t cksum_type;
1512 if (body->oa.o_valid & OBD_MD_FLFLAGS)
1513 cksum_type = cksum_type_unpack(body->oa.o_flags);
1515 cksum_type = OBD_CKSUM_CRC32;
1516 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1517 aa->aa_ppga, OST_READ,
1520 if (peer->nid == req->rq_bulk->bd_sender) {
1524 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1527 if (server_cksum == ~0 && rc > 0) {
1528 CERROR("Protocol error: server %s set the 'checksum' "
1529 "bit, but didn't send a checksum. Not fatal, "
1530 "but please notify on http://bugzilla.lustre.org/\n",
1531 libcfs_nid2str(peer->nid));
1532 } else if (server_cksum != client_cksum) {
1533 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1534 "%s%s%s inum "LPU64"/"LPU64" object "
1535 LPU64"/"LPU64" extent "
1536 "["LPU64"-"LPU64"]\n",
1537 req->rq_import->imp_obd->obd_name,
1538 libcfs_nid2str(peer->nid),
1540 body->oa.o_valid & OBD_MD_FLFID ?
1541 body->oa.o_fid : (__u64)0,
1542 body->oa.o_valid & OBD_MD_FLFID ?
1543 body->oa.o_generation :(__u64)0,
1545 body->oa.o_valid & OBD_MD_FLGROUP ?
1546 body->oa.o_gr : (__u64)0,
1547 aa->aa_ppga[0]->off,
1548 aa->aa_ppga[aa->aa_page_count-1]->off +
1549 aa->aa_ppga[aa->aa_page_count-1]->count -
1551 CERROR("client %x, server %x, cksum_type %x\n",
1552 client_cksum, server_cksum, cksum_type);
1554 aa->aa_oa->o_cksum = client_cksum;
1558 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1561 } else if (unlikely(client_cksum)) {
1562 static int cksum_missed;
1565 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1566 CERROR("Checksum %u requested from %s but not sent\n",
1567 cksum_missed, libcfs_nid2str(peer->nid));
1573 *aa->aa_oa = body->oa;
1578 static int osc_brw_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1579 struct lov_stripe_md *lsm,
1580 obd_count page_count, struct brw_page **pga,
1581 struct obd_capa *ocapa)
1583 struct ptlrpc_request *req;
1587 struct l_wait_info lwi;
1591 cfs_waitq_init(&waitq);
1594 rc = osc_brw_prep_request(cmd, &exp->exp_obd->u.cli, oa, lsm,
1595 page_count, pga, &req, ocapa, 0);
1599 rc = ptlrpc_queue_wait(req);
1601 if (rc == -ETIMEDOUT && req->rq_resend) {
1602 DEBUG_REQ(D_HA, req, "BULK TIMEOUT");
1603 ptlrpc_req_finished(req);
1607 rc = osc_brw_fini_request(req, rc);
1609 ptlrpc_req_finished(req);
1610 if (osc_recoverable_error(rc)) {
1612 if (!osc_should_resend(resends, &exp->exp_obd->u.cli)) {
1613 CERROR("too many resend retries, returning error\n");
1617 lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(resends), NULL, NULL, NULL);
1618 l_wait_event(waitq, 0, &lwi);
1626 int osc_brw_redo_request(struct ptlrpc_request *request,
1627 struct osc_brw_async_args *aa)
1629 struct ptlrpc_request *new_req;
1630 struct ptlrpc_request_set *set = request->rq_set;
1631 struct osc_brw_async_args *new_aa;
1632 struct osc_async_page *oap;
1636 if (!osc_should_resend(aa->aa_resends, aa->aa_cli)) {
1637 CERROR("too many resend retries, returning error\n");
1641 DEBUG_REQ(D_ERROR, request, "redo for recoverable error");
1643 rc = osc_brw_prep_request(lustre_msg_get_opc(request->rq_reqmsg) ==
1644 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1645 aa->aa_cli, aa->aa_oa,
1646 NULL /* lsm unused by osc currently */,
1647 aa->aa_page_count, aa->aa_ppga,
1648 &new_req, aa->aa_ocapa, 0);
1652 client_obd_list_lock(&aa->aa_cli->cl_loi_list_lock);
1654 list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1655 if (oap->oap_request != NULL) {
1656 LASSERTF(request == oap->oap_request,
1657 "request %p != oap_request %p\n",
1658 request, oap->oap_request);
1659 if (oap->oap_interrupted) {
1660 client_obd_list_unlock(&aa->aa_cli->cl_loi_list_lock);
1661 ptlrpc_req_finished(new_req);
1666 /* New request takes over pga and oaps from old request.
1667 * Note that copying a list_head doesn't work, need to move it... */
1669 new_req->rq_interpret_reply = request->rq_interpret_reply;
1670 new_req->rq_async_args = request->rq_async_args;
1671 new_req->rq_sent = cfs_time_current_sec() + aa->aa_resends;
1673 new_aa = ptlrpc_req_async_args(new_req);
1675 CFS_INIT_LIST_HEAD(&new_aa->aa_oaps);
1676 list_splice(&aa->aa_oaps, &new_aa->aa_oaps);
1677 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
1679 list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1680 if (oap->oap_request) {
1681 ptlrpc_req_finished(oap->oap_request);
1682 oap->oap_request = ptlrpc_request_addref(new_req);
1686 new_aa->aa_ocapa = aa->aa_ocapa;
1687 aa->aa_ocapa = NULL;
1689 /* use ptlrpc_set_add_req is safe because interpret functions work
1690 * in check_set context. only one way exist with access to request
1691 * from different thread got -EINTR - this way protected with
1692 * cl_loi_list_lock */
1693 ptlrpc_set_add_req(set, new_req);
1695 client_obd_list_unlock(&aa->aa_cli->cl_loi_list_lock);
1697 DEBUG_REQ(D_INFO, new_req, "new request");
1702 * ugh, we want disk allocation on the target to happen in offset order. we'll
1703 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1704 * fine for our small page arrays and doesn't require allocation. its an
1705 * insertion sort that swaps elements that are strides apart, shrinking the
1706 * stride down until its '1' and the array is sorted.
1708 static void sort_brw_pages(struct brw_page **array, int num)
1711 struct brw_page *tmp;
1715 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1720 for (i = stride ; i < num ; i++) {
1723 while (j >= stride && array[j - stride]->off > tmp->off) {
1724 array[j] = array[j - stride];
1729 } while (stride > 1);
1732 static obd_count max_unfragmented_pages(struct brw_page **pg, obd_count pages)
1738 LASSERT (pages > 0);
1739 offset = pg[i]->off & ~CFS_PAGE_MASK;
1743 if (pages == 0) /* that's all */
1746 if (offset + pg[i]->count < CFS_PAGE_SIZE)
1747 return count; /* doesn't end on page boundary */
1750 offset = pg[i]->off & ~CFS_PAGE_MASK;
1751 if (offset != 0) /* doesn't start on page boundary */
1758 static struct brw_page **osc_build_ppga(struct brw_page *pga, obd_count count)
1760 struct brw_page **ppga;
1763 OBD_ALLOC(ppga, sizeof(*ppga) * count);
1767 for (i = 0; i < count; i++)
1772 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1774 LASSERT(ppga != NULL);
1775 OBD_FREE(ppga, sizeof(*ppga) * count);
1778 static int osc_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
1779 obd_count page_count, struct brw_page *pga,
1780 struct obd_trans_info *oti)
1782 struct obdo *saved_oa = NULL;
1783 struct brw_page **ppga, **orig;
1784 struct obd_import *imp = class_exp2cliimp(exp);
1785 struct client_obd *cli;
1786 int rc, page_count_orig;
1789 LASSERT((imp != NULL) && (imp->imp_obd != NULL));
1790 cli = &imp->imp_obd->u.cli;
1792 if (cmd & OBD_BRW_CHECK) {
1793 /* The caller just wants to know if there's a chance that this
1794 * I/O can succeed */
1796 if (imp->imp_invalid)
1801 /* test_brw with a failed create can trip this, maybe others. */
1802 LASSERT(cli->cl_max_pages_per_rpc);
1806 orig = ppga = osc_build_ppga(pga, page_count);
1809 page_count_orig = page_count;
1811 sort_brw_pages(ppga, page_count);
1812 while (page_count) {
1813 obd_count pages_per_brw;
1815 if (page_count > cli->cl_max_pages_per_rpc)
1816 pages_per_brw = cli->cl_max_pages_per_rpc;
1818 pages_per_brw = page_count;
1820 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1822 if (saved_oa != NULL) {
1823 /* restore previously saved oa */
1824 *oinfo->oi_oa = *saved_oa;
1825 } else if (page_count > pages_per_brw) {
1826 /* save a copy of oa (brw will clobber it) */
1827 OBDO_ALLOC(saved_oa);
1828 if (saved_oa == NULL)
1829 GOTO(out, rc = -ENOMEM);
1830 *saved_oa = *oinfo->oi_oa;
1833 rc = osc_brw_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1834 pages_per_brw, ppga, oinfo->oi_capa);
1839 page_count -= pages_per_brw;
1840 ppga += pages_per_brw;
1844 osc_release_ppga(orig, page_count_orig);
1846 if (saved_oa != NULL)
1847 OBDO_FREE(saved_oa);
1852 /* The companion to osc_enter_cache(), called when @oap is no longer part of
1853 * the dirty accounting. Writeback completes or truncate happens before
1854 * writing starts. Must be called with the loi lock held. */
1855 static void osc_exit_cache(struct client_obd *cli, struct osc_async_page *oap,
1858 osc_release_write_grant(cli, &oap->oap_brw_page, sent);
1862 /* This maintains the lists of pending pages to read/write for a given object
1863 * (lop). This is used by osc_check_rpcs->osc_next_loi() and loi_list_maint()
1864 * to quickly find objects that are ready to send an RPC. */
1865 static int lop_makes_rpc(struct client_obd *cli, struct loi_oap_pages *lop,
1871 if (lop->lop_num_pending == 0)
1874 /* if we have an invalid import we want to drain the queued pages
1875 * by forcing them through rpcs that immediately fail and complete
1876 * the pages. recovery relies on this to empty the queued pages
1877 * before canceling the locks and evicting down the llite pages */
1878 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
1881 /* stream rpcs in queue order as long as as there is an urgent page
1882 * queued. this is our cheap solution for good batching in the case
1883 * where writepage marks some random page in the middle of the file
1884 * as urgent because of, say, memory pressure */
1885 if (!list_empty(&lop->lop_urgent)) {
1886 CDEBUG(D_CACHE, "urgent request forcing RPC\n");
1889 /* fire off rpcs when we have 'optimal' rpcs as tuned for the wire. */
1890 optimal = cli->cl_max_pages_per_rpc;
1891 if (cmd & OBD_BRW_WRITE) {
1892 /* trigger a write rpc stream as long as there are dirtiers
1893 * waiting for space. as they're waiting, they're not going to
1894 * create more pages to coallesce with what's waiting.. */
1895 if (!list_empty(&cli->cl_cache_waiters)) {
1896 CDEBUG(D_CACHE, "cache waiters forcing RPC\n");
1899 /* +16 to avoid triggering rpcs that would want to include pages
1900 * that are being queued but which can't be made ready until
1901 * the queuer finishes with the page. this is a wart for
1902 * llite::commit_write() */
1905 if (lop->lop_num_pending >= optimal)
1911 static int lop_makes_hprpc(struct loi_oap_pages *lop)
1913 struct osc_async_page *oap;
1916 if (list_empty(&lop->lop_urgent))
1919 oap = list_entry(lop->lop_urgent.next,
1920 struct osc_async_page, oap_urgent_item);
1922 if (oap->oap_async_flags & ASYNC_HP) {
1923 CDEBUG(D_CACHE, "hp request forcing RPC\n");
1930 static void on_list(struct list_head *item, struct list_head *list,
1933 if (list_empty(item) && should_be_on)
1934 list_add_tail(item, list);
1935 else if (!list_empty(item) && !should_be_on)
1936 list_del_init(item);
1939 /* maintain the loi's cli list membership invariants so that osc_send_oap_rpc
1940 * can find pages to build into rpcs quickly */
1941 void loi_list_maint(struct client_obd *cli, struct lov_oinfo *loi)
1943 if (lop_makes_hprpc(&loi->loi_write_lop) ||
1944 lop_makes_hprpc(&loi->loi_read_lop)) {
1946 on_list(&loi->loi_ready_item, &cli->cl_loi_ready_list, 0);
1947 on_list(&loi->loi_hp_ready_item, &cli->cl_loi_hp_ready_list, 1);
1949 on_list(&loi->loi_hp_ready_item, &cli->cl_loi_hp_ready_list, 0);
1950 on_list(&loi->loi_ready_item, &cli->cl_loi_ready_list,
1951 lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE)||
1952 lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ));
1955 on_list(&loi->loi_write_item, &cli->cl_loi_write_list,
1956 loi->loi_write_lop.lop_num_pending);
1958 on_list(&loi->loi_read_item, &cli->cl_loi_read_list,
1959 loi->loi_read_lop.lop_num_pending);
1962 static void lop_update_pending(struct client_obd *cli,
1963 struct loi_oap_pages *lop, int cmd, int delta)
1965 lop->lop_num_pending += delta;
1966 if (cmd & OBD_BRW_WRITE)
1967 cli->cl_pending_w_pages += delta;
1969 cli->cl_pending_r_pages += delta;
1973 * this is called when a sync waiter receives an interruption. Its job is to
1974 * get the caller woken as soon as possible. If its page hasn't been put in an
1975 * rpc yet it can dequeue immediately. Otherwise it has to mark the rpc as
1976 * desiring interruption which will forcefully complete the rpc once the rpc
1979 int osc_oap_interrupted(const struct lu_env *env, struct osc_async_page *oap)
1981 struct loi_oap_pages *lop;
1982 struct lov_oinfo *loi;
1986 LASSERT(!oap->oap_interrupted);
1987 oap->oap_interrupted = 1;
1989 /* ok, it's been put in an rpc. only one oap gets a request reference */
1990 if (oap->oap_request != NULL) {
1991 ptlrpc_mark_interrupted(oap->oap_request);
1992 ptlrpcd_wake(oap->oap_request);
1993 ptlrpc_req_finished(oap->oap_request);
1994 oap->oap_request = NULL;
1998 * page completion may be called only if ->cpo_prep() method was
1999 * executed by osc_io_submit(), that also adds page the to pending list
2001 if (!list_empty(&oap->oap_pending_item)) {
2002 list_del_init(&oap->oap_pending_item);
2003 list_del_init(&oap->oap_urgent_item);
2006 lop = (oap->oap_cmd & OBD_BRW_WRITE) ?
2007 &loi->loi_write_lop : &loi->loi_read_lop;
2008 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, -1);
2009 loi_list_maint(oap->oap_cli, oap->oap_loi);
2010 rc = oap->oap_caller_ops->ap_completion(env,
2011 oap->oap_caller_data,
2012 oap->oap_cmd, NULL, -EINTR);
2018 /* this is trying to propogate async writeback errors back up to the
2019 * application. As an async write fails we record the error code for later if
2020 * the app does an fsync. As long as errors persist we force future rpcs to be
2021 * sync so that the app can get a sync error and break the cycle of queueing
2022 * pages for which writeback will fail. */
2023 static void osc_process_ar(struct osc_async_rc *ar, __u64 xid,
2030 ar->ar_force_sync = 1;
2031 ar->ar_min_xid = ptlrpc_sample_next_xid();
2036 if (ar->ar_force_sync && (xid >= ar->ar_min_xid))
2037 ar->ar_force_sync = 0;
2040 void osc_oap_to_pending(struct osc_async_page *oap)
2042 struct loi_oap_pages *lop;
2044 if (oap->oap_cmd & OBD_BRW_WRITE)
2045 lop = &oap->oap_loi->loi_write_lop;
2047 lop = &oap->oap_loi->loi_read_lop;
2049 if (oap->oap_async_flags & ASYNC_HP)
2050 list_add(&oap->oap_urgent_item, &lop->lop_urgent);
2051 else if (oap->oap_async_flags & ASYNC_URGENT)
2052 list_add_tail(&oap->oap_urgent_item, &lop->lop_urgent);
2053 list_add_tail(&oap->oap_pending_item, &lop->lop_pending);
2054 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, 1);
2057 /* this must be called holding the loi list lock to give coverage to exit_cache,
2058 * async_flag maintenance, and oap_request */
2059 static void osc_ap_completion(const struct lu_env *env,
2060 struct client_obd *cli, struct obdo *oa,
2061 struct osc_async_page *oap, int sent, int rc)
2066 if (oap->oap_request != NULL) {
2067 xid = ptlrpc_req_xid(oap->oap_request);
2068 ptlrpc_req_finished(oap->oap_request);
2069 oap->oap_request = NULL;
2072 oap->oap_async_flags = 0;
2073 oap->oap_interrupted = 0;
2075 if (oap->oap_cmd & OBD_BRW_WRITE) {
2076 osc_process_ar(&cli->cl_ar, xid, rc);
2077 osc_process_ar(&oap->oap_loi->loi_ar, xid, rc);
2080 if (rc == 0 && oa != NULL) {
2081 if (oa->o_valid & OBD_MD_FLBLOCKS)
2082 oap->oap_loi->loi_lvb.lvb_blocks = oa->o_blocks;
2083 if (oa->o_valid & OBD_MD_FLMTIME)
2084 oap->oap_loi->loi_lvb.lvb_mtime = oa->o_mtime;
2085 if (oa->o_valid & OBD_MD_FLATIME)
2086 oap->oap_loi->loi_lvb.lvb_atime = oa->o_atime;
2087 if (oa->o_valid & OBD_MD_FLCTIME)
2088 oap->oap_loi->loi_lvb.lvb_ctime = oa->o_ctime;
2091 rc = oap->oap_caller_ops->ap_completion(env, oap->oap_caller_data,
2092 oap->oap_cmd, oa, rc);
2094 /* ll_ap_completion (from llite) drops PG_locked. so, a new
2095 * I/O on the page could start, but OSC calls it under lock
2096 * and thus we can add oap back to pending safely */
2098 /* upper layer wants to leave the page on pending queue */
2099 osc_oap_to_pending(oap);
2101 osc_exit_cache(cli, oap, sent);
2105 static int brw_interpret(const struct lu_env *env,
2106 struct ptlrpc_request *req, void *data, int rc)
2108 struct osc_brw_async_args *aa = data;
2109 struct client_obd *cli;
2113 rc = osc_brw_fini_request(req, rc);
2114 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
2115 if (osc_recoverable_error(rc)) {
2116 rc = osc_brw_redo_request(req, aa);
2122 capa_put(aa->aa_ocapa);
2123 aa->aa_ocapa = NULL;
2128 client_obd_list_lock(&cli->cl_loi_list_lock);
2130 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2131 * is called so we know whether to go to sync BRWs or wait for more
2132 * RPCs to complete */
2133 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
2134 cli->cl_w_in_flight--;
2136 cli->cl_r_in_flight--;
2138 async = list_empty(&aa->aa_oaps);
2139 if (!async) { /* from osc_send_oap_rpc() */
2140 struct osc_async_page *oap, *tmp;
2141 /* the caller may re-use the oap after the completion call so
2142 * we need to clean it up a little */
2143 list_for_each_entry_safe(oap, tmp, &aa->aa_oaps, oap_rpc_item) {
2144 list_del_init(&oap->oap_rpc_item);
2145 osc_ap_completion(env, cli, aa->aa_oa, oap, 1, rc);
2147 OBDO_FREE(aa->aa_oa);
2148 } else { /* from async_internal() */
2150 for (i = 0; i < aa->aa_page_count; i++)
2151 osc_release_write_grant(aa->aa_cli, aa->aa_ppga[i], 1);
2153 osc_wake_cache_waiters(cli);
2154 osc_check_rpcs(env, cli);
2155 client_obd_list_unlock(&cli->cl_loi_list_lock);
2157 cl_req_completion(env, aa->aa_clerq, rc);
2158 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
2162 static struct ptlrpc_request *osc_build_req(const struct lu_env *env,
2163 struct client_obd *cli,
2164 struct list_head *rpc_list,
2165 int page_count, int cmd)
2167 struct ptlrpc_request *req;
2168 struct brw_page **pga = NULL;
2169 struct osc_brw_async_args *aa;
2170 struct obdo *oa = NULL;
2171 const struct obd_async_page_ops *ops = NULL;
2172 void *caller_data = NULL;
2173 struct osc_async_page *oap;
2174 struct osc_async_page *tmp;
2175 struct ost_body *body;
2176 struct cl_req *clerq = NULL;
2177 enum cl_req_type crt = (cmd & OBD_BRW_WRITE) ? CRT_WRITE : CRT_READ;
2178 struct ldlm_lock *lock = NULL;
2179 struct cl_req_attr crattr;
2183 LASSERT(!list_empty(rpc_list));
2185 memset(&crattr, 0, sizeof crattr);
2186 OBD_ALLOC(pga, sizeof(*pga) * page_count);
2188 GOTO(out, req = ERR_PTR(-ENOMEM));
2192 GOTO(out, req = ERR_PTR(-ENOMEM));
2195 list_for_each_entry(oap, rpc_list, oap_rpc_item) {
2196 struct cl_page *page = osc_oap2cl_page(oap);
2198 ops = oap->oap_caller_ops;
2199 caller_data = oap->oap_caller_data;
2201 clerq = cl_req_alloc(env, page, crt,
2202 1 /* only 1-object rpcs for
2205 GOTO(out, req = (void *)clerq);
2206 lock = oap->oap_ldlm_lock;
2208 pga[i] = &oap->oap_brw_page;
2209 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
2210 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
2211 pga[i]->pg, cfs_page_index(oap->oap_page), oap, pga[i]->flag);
2213 cl_req_page_add(env, clerq, page);
2216 /* always get the data for the obdo for the rpc */
2217 LASSERT(ops != NULL);
2219 crattr.cra_capa = NULL;
2220 cl_req_attr_set(env, clerq, &crattr, ~0ULL);
2222 oa->o_handle = lock->l_remote_handle;
2223 oa->o_valid |= OBD_MD_FLHANDLE;
2226 rc = cl_req_prep(env, clerq);
2228 CERROR("cl_req_prep failed: %d\n", rc);
2229 GOTO(out, req = ERR_PTR(rc));
2232 sort_brw_pages(pga, page_count);
2233 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
2234 pga, &req, crattr.cra_capa, 1);
2236 CERROR("prep_req failed: %d\n", rc);
2237 GOTO(out, req = ERR_PTR(rc));
2240 /* Need to update the timestamps after the request is built in case
2241 * we race with setattr (locally or in queue at OST). If OST gets
2242 * later setattr before earlier BRW (as determined by the request xid),
2243 * the OST will not use BRW timestamps. Sadly, there is no obvious
2244 * way to do this in a single call. bug 10150 */
2245 body = req_capsule_client_get(&req->rq_pill, &RMF_OST_BODY);
2246 cl_req_attr_set(env, clerq, &crattr,
2247 OBD_MD_FLMTIME|OBD_MD_FLCTIME|OBD_MD_FLATIME);
2249 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
2250 aa = ptlrpc_req_async_args(req);
2251 CFS_INIT_LIST_HEAD(&aa->aa_oaps);
2252 list_splice(rpc_list, &aa->aa_oaps);
2253 CFS_INIT_LIST_HEAD(rpc_list);
2254 aa->aa_clerq = clerq;
2256 capa_put(crattr.cra_capa);
2261 OBD_FREE(pga, sizeof(*pga) * page_count);
2262 /* this should happen rarely and is pretty bad, it makes the
2263 * pending list not follow the dirty order */
2264 client_obd_list_lock(&cli->cl_loi_list_lock);
2265 list_for_each_entry_safe(oap, tmp, rpc_list, oap_rpc_item) {
2266 list_del_init(&oap->oap_rpc_item);
2268 /* queued sync pages can be torn down while the pages
2269 * were between the pending list and the rpc */
2270 if (oap->oap_interrupted) {
2271 CDEBUG(D_INODE, "oap %p interrupted\n", oap);
2272 osc_ap_completion(env, cli, NULL, oap, 0,
2276 osc_ap_completion(env, cli, NULL, oap, 0, PTR_ERR(req));
2278 if (clerq && !IS_ERR(clerq))
2279 cl_req_completion(env, clerq, PTR_ERR(req));
2285 * prepare pages for ASYNC io and put pages in send queue.
2289 * \param cmd - OBD_BRW_* macroses
2290 * \param lop - pending pages
2292 * \return zero if pages successfully add to send queue.
2293 * \return not zere if error occurring.
2296 osc_send_oap_rpc(const struct lu_env *env, struct client_obd *cli,
2297 struct lov_oinfo *loi,
2298 int cmd, struct loi_oap_pages *lop)
2300 struct ptlrpc_request *req;
2301 obd_count page_count = 0;
2302 struct osc_async_page *oap = NULL, *tmp;
2303 struct osc_brw_async_args *aa;
2304 const struct obd_async_page_ops *ops;
2305 CFS_LIST_HEAD(rpc_list);
2306 unsigned int ending_offset;
2307 unsigned starting_offset = 0;
2309 struct cl_object *clob = NULL;
2312 /* If there are HP OAPs we need to handle at least 1 of them,
2313 * move it the beginning of the pending list for that. */
2314 if (!list_empty(&lop->lop_urgent)) {
2315 oap = list_entry(lop->lop_urgent.next,
2316 struct osc_async_page, oap_urgent_item);
2317 if (oap->oap_async_flags & ASYNC_HP)
2318 list_move(&oap->oap_pending_item, &lop->lop_pending);
2321 /* first we find the pages we're allowed to work with */
2322 list_for_each_entry_safe(oap, tmp, &lop->lop_pending,
2324 ops = oap->oap_caller_ops;
2326 LASSERTF(oap->oap_magic == OAP_MAGIC, "Bad oap magic: oap %p, "
2327 "magic 0x%x\n", oap, oap->oap_magic);
2330 /* pin object in memory, so that completion call-backs
2331 * can be safely called under client_obd_list lock. */
2332 clob = osc_oap2cl_page(oap)->cp_obj;
2333 cl_object_get(clob);
2336 if (page_count != 0 &&
2337 srvlock != !!(oap->oap_brw_flags & OBD_BRW_SRVLOCK)) {
2338 CDEBUG(D_PAGE, "SRVLOCK flag mismatch,"
2339 " oap %p, page %p, srvlock %u\n",
2340 oap, oap->oap_brw_page.pg, (unsigned)!srvlock);
2343 /* in llite being 'ready' equates to the page being locked
2344 * until completion unlocks it. commit_write submits a page
2345 * as not ready because its unlock will happen unconditionally
2346 * as the call returns. if we race with commit_write giving
2347 * us that page we dont' want to create a hole in the page
2348 * stream, so we stop and leave the rpc to be fired by
2349 * another dirtier or kupdated interval (the not ready page
2350 * will still be on the dirty list). we could call in
2351 * at the end of ll_file_write to process the queue again. */
2352 if (!(oap->oap_async_flags & ASYNC_READY)) {
2353 int rc = ops->ap_make_ready(env, oap->oap_caller_data,
2356 CDEBUG(D_INODE, "oap %p page %p returned %d "
2357 "instead of ready\n", oap,
2361 /* llite is telling us that the page is still
2362 * in commit_write and that we should try
2363 * and put it in an rpc again later. we
2364 * break out of the loop so we don't create
2365 * a hole in the sequence of pages in the rpc
2370 /* the io isn't needed.. tell the checks
2371 * below to complete the rpc with EINTR */
2372 oap->oap_async_flags |= ASYNC_COUNT_STABLE;
2373 oap->oap_count = -EINTR;
2376 oap->oap_async_flags |= ASYNC_READY;
2379 LASSERTF(0, "oap %p page %p returned %d "
2380 "from make_ready\n", oap,
2388 * Page submitted for IO has to be locked. Either by
2389 * ->ap_make_ready() or by higher layers.
2391 #if defined(__KERNEL__) && defined(__linux__)
2393 struct cl_page *page;
2395 page = osc_oap2cl_page(oap);
2397 if (page->cp_type == CPT_CACHEABLE &&
2398 !(PageLocked(oap->oap_page) &&
2399 (CheckWriteback(oap->oap_page, cmd)))) {
2400 CDEBUG(D_PAGE, "page %p lost wb %lx/%x\n",
2402 (long)oap->oap_page->flags,
2403 oap->oap_async_flags);
2408 /* If there is a gap at the start of this page, it can't merge
2409 * with any previous page, so we'll hand the network a
2410 * "fragmented" page array that it can't transfer in 1 RDMA */
2411 if (page_count != 0 && oap->oap_page_off != 0)
2414 /* take the page out of our book-keeping */
2415 list_del_init(&oap->oap_pending_item);
2416 lop_update_pending(cli, lop, cmd, -1);
2417 list_del_init(&oap->oap_urgent_item);
2419 if (page_count == 0)
2420 starting_offset = (oap->oap_obj_off+oap->oap_page_off) &
2421 (PTLRPC_MAX_BRW_SIZE - 1);
2423 /* ask the caller for the size of the io as the rpc leaves. */
2424 if (!(oap->oap_async_flags & ASYNC_COUNT_STABLE)) {
2426 ops->ap_refresh_count(env, oap->oap_caller_data,
2428 LASSERT(oap->oap_page_off + oap->oap_count <= CFS_PAGE_SIZE);
2430 if (oap->oap_count <= 0) {
2431 CDEBUG(D_CACHE, "oap %p count %d, completing\n", oap,
2433 osc_ap_completion(env, cli, NULL,
2434 oap, 0, oap->oap_count);
2438 /* now put the page back in our accounting */
2439 list_add_tail(&oap->oap_rpc_item, &rpc_list);
2440 if (page_count == 0)
2441 srvlock = !!(oap->oap_brw_flags & OBD_BRW_SRVLOCK);
2442 if (++page_count >= cli->cl_max_pages_per_rpc)
2445 /* End on a PTLRPC_MAX_BRW_SIZE boundary. We want full-sized
2446 * RPCs aligned on PTLRPC_MAX_BRW_SIZE boundaries to help reads
2447 * have the same alignment as the initial writes that allocated
2448 * extents on the server. */
2449 ending_offset = (oap->oap_obj_off + oap->oap_page_off +
2450 oap->oap_count) & (PTLRPC_MAX_BRW_SIZE - 1);
2451 if (ending_offset == 0)
2454 /* If there is a gap at the end of this page, it can't merge
2455 * with any subsequent pages, so we'll hand the network a
2456 * "fragmented" page array that it can't transfer in 1 RDMA */
2457 if (oap->oap_page_off + oap->oap_count < CFS_PAGE_SIZE)
2461 osc_wake_cache_waiters(cli);
2463 loi_list_maint(cli, loi);
2465 client_obd_list_unlock(&cli->cl_loi_list_lock);
2468 cl_object_put(env, clob);
2470 if (page_count == 0) {
2471 client_obd_list_lock(&cli->cl_loi_list_lock);
2475 req = osc_build_req(env, cli, &rpc_list, page_count, cmd);
2477 LASSERT(list_empty(&rpc_list));
2478 loi_list_maint(cli, loi);
2479 RETURN(PTR_ERR(req));
2482 aa = ptlrpc_req_async_args(req);
2484 if (cmd == OBD_BRW_READ) {
2485 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2486 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2487 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2488 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2490 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2491 lprocfs_oh_tally(&cli->cl_write_rpc_hist,
2492 cli->cl_w_in_flight);
2493 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2494 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2496 ptlrpc_lprocfs_brw(req, aa->aa_requested_nob);
2498 client_obd_list_lock(&cli->cl_loi_list_lock);
2500 if (cmd == OBD_BRW_READ)
2501 cli->cl_r_in_flight++;
2503 cli->cl_w_in_flight++;
2505 /* queued sync pages can be torn down while the pages
2506 * were between the pending list and the rpc */
2508 list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2509 /* only one oap gets a request reference */
2512 if (oap->oap_interrupted && !req->rq_intr) {
2513 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2515 ptlrpc_mark_interrupted(req);
2519 tmp->oap_request = ptlrpc_request_addref(req);
2521 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2522 page_count, aa, cli->cl_r_in_flight, cli->cl_w_in_flight);
2524 req->rq_interpret_reply = brw_interpret;
2525 ptlrpcd_add_req(req, PSCOPE_BRW);
2529 #define LOI_DEBUG(LOI, STR, args...) \
2530 CDEBUG(D_INODE, "loi ready %d wr %d:%d rd %d:%d " STR, \
2531 !list_empty(&(LOI)->loi_ready_item) || \
2532 !list_empty(&(LOI)->loi_hp_ready_item), \
2533 (LOI)->loi_write_lop.lop_num_pending, \
2534 !list_empty(&(LOI)->loi_write_lop.lop_urgent), \
2535 (LOI)->loi_read_lop.lop_num_pending, \
2536 !list_empty(&(LOI)->loi_read_lop.lop_urgent), \
2539 /* This is called by osc_check_rpcs() to find which objects have pages that
2540 * we could be sending. These lists are maintained by lop_makes_rpc(). */
2541 struct lov_oinfo *osc_next_loi(struct client_obd *cli)
2545 /* First return objects that have blocked locks so that they
2546 * will be flushed quickly and other clients can get the lock,
2547 * then objects which have pages ready to be stuffed into RPCs */
2548 if (!list_empty(&cli->cl_loi_hp_ready_list))
2549 RETURN(list_entry(cli->cl_loi_hp_ready_list.next,
2550 struct lov_oinfo, loi_hp_ready_item));
2551 if (!list_empty(&cli->cl_loi_ready_list))
2552 RETURN(list_entry(cli->cl_loi_ready_list.next,
2553 struct lov_oinfo, loi_ready_item));
2555 /* then if we have cache waiters, return all objects with queued
2556 * writes. This is especially important when many small files
2557 * have filled up the cache and not been fired into rpcs because
2558 * they don't pass the nr_pending/object threshhold */
2559 if (!list_empty(&cli->cl_cache_waiters) &&
2560 !list_empty(&cli->cl_loi_write_list))
2561 RETURN(list_entry(cli->cl_loi_write_list.next,
2562 struct lov_oinfo, loi_write_item));
2564 /* then return all queued objects when we have an invalid import
2565 * so that they get flushed */
2566 if (cli->cl_import == NULL || cli->cl_import->imp_invalid) {
2567 if (!list_empty(&cli->cl_loi_write_list))
2568 RETURN(list_entry(cli->cl_loi_write_list.next,
2569 struct lov_oinfo, loi_write_item));
2570 if (!list_empty(&cli->cl_loi_read_list))
2571 RETURN(list_entry(cli->cl_loi_read_list.next,
2572 struct lov_oinfo, loi_read_item));
2577 static int osc_max_rpc_in_flight(struct client_obd *cli, struct lov_oinfo *loi)
2579 struct osc_async_page *oap;
2582 if (!list_empty(&loi->loi_write_lop.lop_urgent)) {
2583 oap = list_entry(loi->loi_write_lop.lop_urgent.next,
2584 struct osc_async_page, oap_urgent_item);
2585 hprpc = !!(oap->oap_async_flags & ASYNC_HP);
2588 if (!hprpc && !list_empty(&loi->loi_read_lop.lop_urgent)) {
2589 oap = list_entry(loi->loi_write_lop.lop_urgent.next,
2590 struct osc_async_page, oap_urgent_item);
2591 hprpc = !!(oap->oap_async_flags & ASYNC_HP);
2594 return rpcs_in_flight(cli) >= cli->cl_max_rpcs_in_flight + hprpc;
2597 /* called with the loi list lock held */
2598 void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli)
2600 struct lov_oinfo *loi;
2601 int rc = 0, race_counter = 0;
2604 while ((loi = osc_next_loi(cli)) != NULL) {
2605 LOI_DEBUG(loi, "%lu in flight\n", rpcs_in_flight(cli));
2607 if (osc_max_rpc_in_flight(cli, loi))
2610 /* attempt some read/write balancing by alternating between
2611 * reads and writes in an object. The makes_rpc checks here
2612 * would be redundant if we were getting read/write work items
2613 * instead of objects. we don't want send_oap_rpc to drain a
2614 * partial read pending queue when we're given this object to
2615 * do io on writes while there are cache waiters */
2616 if (lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE)) {
2617 rc = osc_send_oap_rpc(env, cli, loi, OBD_BRW_WRITE,
2618 &loi->loi_write_lop);
2626 if (lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ)) {
2627 rc = osc_send_oap_rpc(env, cli, loi, OBD_BRW_READ,
2628 &loi->loi_read_lop);
2637 /* attempt some inter-object balancing by issueing rpcs
2638 * for each object in turn */
2639 if (!list_empty(&loi->loi_hp_ready_item))
2640 list_del_init(&loi->loi_hp_ready_item);
2641 if (!list_empty(&loi->loi_ready_item))
2642 list_del_init(&loi->loi_ready_item);
2643 if (!list_empty(&loi->loi_write_item))
2644 list_del_init(&loi->loi_write_item);
2645 if (!list_empty(&loi->loi_read_item))
2646 list_del_init(&loi->loi_read_item);
2648 loi_list_maint(cli, loi);
2650 /* send_oap_rpc fails with 0 when make_ready tells it to
2651 * back off. llite's make_ready does this when it tries
2652 * to lock a page queued for write that is already locked.
2653 * we want to try sending rpcs from many objects, but we
2654 * don't want to spin failing with 0. */
2655 if (race_counter == 10)
2661 /* we're trying to queue a page in the osc so we're subject to the
2662 * 'cl_dirty_max' limit on the number of pages that can be queued in the osc.
2663 * If the osc's queued pages are already at that limit, then we want to sleep
2664 * until there is space in the osc's queue for us. We also may be waiting for
2665 * write credits from the OST if there are RPCs in flight that may return some
2666 * before we fall back to sync writes.
2668 * We need this know our allocation was granted in the presence of signals */
2669 static int ocw_granted(struct client_obd *cli, struct osc_cache_waiter *ocw)
2673 client_obd_list_lock(&cli->cl_loi_list_lock);
2674 rc = list_empty(&ocw->ocw_entry) || rpcs_in_flight(cli) == 0;
2675 client_obd_list_unlock(&cli->cl_loi_list_lock);
2680 * Non-blocking version of osc_enter_cache() that consumes grant only when it
2683 int osc_enter_cache_try(const struct lu_env *env,
2684 struct client_obd *cli, struct lov_oinfo *loi,
2685 struct osc_async_page *oap, int transient)
2689 has_grant = cli->cl_avail_grant >= CFS_PAGE_SIZE;
2691 osc_consume_write_grant(cli, &oap->oap_brw_page);
2693 cli->cl_dirty_transit += CFS_PAGE_SIZE;
2694 atomic_inc(&obd_dirty_transit_pages);
2695 oap->oap_brw_flags |= OBD_BRW_NOCACHE;
2701 /* Caller must hold loi_list_lock - we drop/regain it if we need to wait for
2702 * grant or cache space. */
2703 static int osc_enter_cache(const struct lu_env *env,
2704 struct client_obd *cli, struct lov_oinfo *loi,
2705 struct osc_async_page *oap)
2707 struct osc_cache_waiter ocw;
2708 struct l_wait_info lwi = { 0 };
2712 CDEBUG(D_CACHE, "dirty: %ld/%d dirty_max: %ld/%d dropped: %lu "
2713 "grant: %lu\n", cli->cl_dirty, atomic_read(&obd_dirty_pages),
2714 cli->cl_dirty_max, obd_max_dirty_pages,
2715 cli->cl_lost_grant, cli->cl_avail_grant);
2717 /* force the caller to try sync io. this can jump the list
2718 * of queued writes and create a discontiguous rpc stream */
2719 if (cli->cl_dirty_max < CFS_PAGE_SIZE || cli->cl_ar.ar_force_sync ||
2720 loi->loi_ar.ar_force_sync)
2723 /* Hopefully normal case - cache space and write credits available */
2724 if (cli->cl_dirty + CFS_PAGE_SIZE <= cli->cl_dirty_max &&
2725 atomic_read(&obd_dirty_pages) + 1 <= obd_max_dirty_pages &&
2726 osc_enter_cache_try(env, cli, loi, oap, 0))
2729 /* Make sure that there are write rpcs in flight to wait for. This
2730 * is a little silly as this object may not have any pending but
2731 * other objects sure might. */
2732 if (cli->cl_w_in_flight) {
2733 list_add_tail(&ocw.ocw_entry, &cli->cl_cache_waiters);
2734 cfs_waitq_init(&ocw.ocw_waitq);
2738 loi_list_maint(cli, loi);
2739 osc_check_rpcs(env, cli);
2740 client_obd_list_unlock(&cli->cl_loi_list_lock);
2742 CDEBUG(D_CACHE, "sleeping for cache space\n");
2743 l_wait_event(ocw.ocw_waitq, ocw_granted(cli, &ocw), &lwi);
2745 client_obd_list_lock(&cli->cl_loi_list_lock);
2746 if (!list_empty(&ocw.ocw_entry)) {
2747 list_del(&ocw.ocw_entry);
2757 int osc_prep_async_page(struct obd_export *exp, struct lov_stripe_md *lsm,
2758 struct lov_oinfo *loi, cfs_page_t *page,
2759 obd_off offset, const struct obd_async_page_ops *ops,
2760 void *data, void **res, int nocache,
2761 struct lustre_handle *lockh)
2763 struct osc_async_page *oap;
2768 return size_round(sizeof(*oap));
2771 oap->oap_magic = OAP_MAGIC;
2772 oap->oap_cli = &exp->exp_obd->u.cli;
2775 oap->oap_caller_ops = ops;
2776 oap->oap_caller_data = data;
2778 oap->oap_page = page;
2779 oap->oap_obj_off = offset;
2780 if (!client_is_remote(exp) &&
2781 cfs_capable(CFS_CAP_SYS_RESOURCE))
2782 oap->oap_brw_flags = OBD_BRW_NOQUOTA;
2784 LASSERT(!(offset & ~CFS_PAGE_MASK));
2786 CFS_INIT_LIST_HEAD(&oap->oap_pending_item);
2787 CFS_INIT_LIST_HEAD(&oap->oap_urgent_item);
2788 CFS_INIT_LIST_HEAD(&oap->oap_rpc_item);
2789 CFS_INIT_LIST_HEAD(&oap->oap_page_list);
2791 spin_lock_init(&oap->oap_lock);
2792 CDEBUG(D_CACHE, "oap %p page %p obj off "LPU64"\n", oap, page, offset);
2796 struct osc_async_page *oap_from_cookie(void *cookie)
2798 struct osc_async_page *oap = cookie;
2799 if (oap->oap_magic != OAP_MAGIC)
2800 return ERR_PTR(-EINVAL);
2804 int osc_queue_async_io(const struct lu_env *env,
2805 struct obd_export *exp, struct lov_stripe_md *lsm,
2806 struct lov_oinfo *loi, void *cookie,
2807 int cmd, obd_off off, int count,
2808 obd_flag brw_flags, enum async_flags async_flags)
2810 struct client_obd *cli = &exp->exp_obd->u.cli;
2811 struct osc_async_page *oap;
2815 oap = oap_from_cookie(cookie);
2817 RETURN(PTR_ERR(oap));
2819 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2822 if (!list_empty(&oap->oap_pending_item) ||
2823 !list_empty(&oap->oap_urgent_item) ||
2824 !list_empty(&oap->oap_rpc_item))
2827 /* check if the file's owner/group is over quota */
2828 if ((cmd & OBD_BRW_WRITE) && !(cmd & OBD_BRW_NOQUOTA)) {
2829 struct cl_object *obj;
2830 struct cl_attr attr; /* XXX put attr into thread info */
2831 unsigned int qid[MAXQUOTAS];
2833 obj = cl_object_top(osc_oap2cl_page(oap)->cp_obj);
2835 cl_object_attr_lock(obj);
2836 rc = cl_object_attr_get(env, obj, &attr);
2837 cl_object_attr_unlock(obj);
2839 qid[USRQUOTA] = attr.cat_uid;
2840 qid[GRPQUOTA] = attr.cat_gid;
2842 lquota_chkdq(quota_interface, cli, qid) == NO_QUOTA)
2849 loi = lsm->lsm_oinfo[0];
2851 client_obd_list_lock(&cli->cl_loi_list_lock);
2853 LASSERT(off + count <= CFS_PAGE_SIZE);
2855 oap->oap_page_off = off;
2856 oap->oap_count = count;
2857 oap->oap_brw_flags = brw_flags;
2858 oap->oap_async_flags = async_flags;
2860 if (cmd & OBD_BRW_WRITE) {
2861 rc = osc_enter_cache(env, cli, loi, oap);
2863 client_obd_list_unlock(&cli->cl_loi_list_lock);
2868 osc_oap_to_pending(oap);
2869 loi_list_maint(cli, loi);
2871 LOI_DEBUG(loi, "oap %p page %p added for cmd %d\n", oap, oap->oap_page,
2874 osc_check_rpcs(env, cli);
2875 client_obd_list_unlock(&cli->cl_loi_list_lock);
2880 /* aka (~was & now & flag), but this is more clear :) */
2881 #define SETTING(was, now, flag) (!(was & flag) && (now & flag))
2883 int osc_set_async_flags_base(struct client_obd *cli,
2884 struct lov_oinfo *loi, struct osc_async_page *oap,
2885 obd_flag async_flags)
2887 struct loi_oap_pages *lop;
2890 LASSERT(!list_empty(&oap->oap_pending_item));
2892 if (oap->oap_cmd & OBD_BRW_WRITE) {
2893 lop = &loi->loi_write_lop;
2895 lop = &loi->loi_read_lop;
2898 if ((oap->oap_async_flags & async_flags) == async_flags)
2901 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_READY))
2902 oap->oap_async_flags |= ASYNC_READY;
2904 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_URGENT) &&
2905 list_empty(&oap->oap_rpc_item)) {
2906 if (oap->oap_async_flags & ASYNC_HP)
2907 list_add(&oap->oap_urgent_item, &lop->lop_urgent);
2909 list_add_tail(&oap->oap_urgent_item, &lop->lop_urgent);
2910 oap->oap_async_flags |= ASYNC_URGENT;
2911 loi_list_maint(cli, loi);
2914 LOI_DEBUG(loi, "oap %p page %p has flags %x\n", oap, oap->oap_page,
2915 oap->oap_async_flags);
2919 int osc_teardown_async_page(struct obd_export *exp,
2920 struct lov_stripe_md *lsm,
2921 struct lov_oinfo *loi, void *cookie)
2923 struct client_obd *cli = &exp->exp_obd->u.cli;
2924 struct loi_oap_pages *lop;
2925 struct osc_async_page *oap;
2929 oap = oap_from_cookie(cookie);
2931 RETURN(PTR_ERR(oap));
2934 loi = lsm->lsm_oinfo[0];
2936 if (oap->oap_cmd & OBD_BRW_WRITE) {
2937 lop = &loi->loi_write_lop;
2939 lop = &loi->loi_read_lop;
2942 client_obd_list_lock(&cli->cl_loi_list_lock);
2944 if (!list_empty(&oap->oap_rpc_item))
2945 GOTO(out, rc = -EBUSY);
2947 osc_exit_cache(cli, oap, 0);
2948 osc_wake_cache_waiters(cli);
2950 if (!list_empty(&oap->oap_urgent_item)) {
2951 list_del_init(&oap->oap_urgent_item);
2952 oap->oap_async_flags &= ~(ASYNC_URGENT | ASYNC_HP);
2954 if (!list_empty(&oap->oap_pending_item)) {
2955 list_del_init(&oap->oap_pending_item);
2956 lop_update_pending(cli, lop, oap->oap_cmd, -1);
2958 loi_list_maint(cli, loi);
2959 LOI_DEBUG(loi, "oap %p page %p torn down\n", oap, oap->oap_page);
2961 client_obd_list_unlock(&cli->cl_loi_list_lock);
2965 static void osc_set_lock_data_with_check(struct ldlm_lock *lock,
2966 struct ldlm_enqueue_info *einfo,
2969 void *data = einfo->ei_cbdata;
2971 LASSERT(lock != NULL);
2972 LASSERT(lock->l_blocking_ast == einfo->ei_cb_bl);
2973 LASSERT(lock->l_resource->lr_type == einfo->ei_type);
2974 LASSERT(lock->l_completion_ast == einfo->ei_cb_cp);
2975 LASSERT(lock->l_glimpse_ast == einfo->ei_cb_gl);
2977 lock_res_and_lock(lock);
2978 spin_lock(&osc_ast_guard);
2979 LASSERT(lock->l_ast_data == NULL || lock->l_ast_data == data);
2980 lock->l_ast_data = data;
2981 spin_unlock(&osc_ast_guard);
2982 unlock_res_and_lock(lock);
2985 static void osc_set_data_with_check(struct lustre_handle *lockh,
2986 struct ldlm_enqueue_info *einfo,
2989 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2992 osc_set_lock_data_with_check(lock, einfo, flags);
2993 LDLM_LOCK_PUT(lock);
2995 CERROR("lockh %p, data %p - client evicted?\n",
2996 lockh, einfo->ei_cbdata);
2999 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
3000 ldlm_iterator_t replace, void *data)
3002 struct ldlm_res_id res_id;
3003 struct obd_device *obd = class_exp2obd(exp);
3005 osc_build_res_name(lsm->lsm_object_id, lsm->lsm_object_gr, &res_id);
3006 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
3010 static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
3011 obd_enqueue_update_f upcall, void *cookie,
3014 int intent = *flags & LDLM_FL_HAS_INTENT;
3018 /* The request was created before ldlm_cli_enqueue call. */
3019 if (rc == ELDLM_LOCK_ABORTED) {
3020 struct ldlm_reply *rep;
3021 rep = req_capsule_server_get(&req->rq_pill,
3024 LASSERT(rep != NULL);
3025 if (rep->lock_policy_res1)
3026 rc = rep->lock_policy_res1;
3030 if ((intent && rc == ELDLM_LOCK_ABORTED) || !rc) {
3031 *flags |= LDLM_FL_LVB_READY;
3032 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
3033 lvb->lvb_size, lvb->lvb_blocks, lvb->lvb_mtime);
3036 /* Call the update callback. */
3037 rc = (*upcall)(cookie, rc);
3041 static int osc_enqueue_interpret(const struct lu_env *env,
3042 struct ptlrpc_request *req,
3043 struct osc_enqueue_args *aa, int rc)
3045 struct ldlm_lock *lock;
3046 struct lustre_handle handle;
3049 /* Make a local copy of a lock handle and a mode, because aa->oa_*
3050 * might be freed anytime after lock upcall has been called. */
3051 lustre_handle_copy(&handle, aa->oa_lockh);
3052 mode = aa->oa_ei->ei_mode;
3054 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
3056 lock = ldlm_handle2lock(&handle);
3058 /* Take an additional reference so that a blocking AST that
3059 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
3060 * to arrive after an upcall has been executed by
3061 * osc_enqueue_fini(). */
3062 ldlm_lock_addref(&handle, mode);
3064 /* Complete obtaining the lock procedure. */
3065 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
3066 mode, aa->oa_flags, aa->oa_lvb,
3067 sizeof(*aa->oa_lvb), lustre_swab_ost_lvb,
3069 /* Complete osc stuff. */
3070 rc = osc_enqueue_fini(req, aa->oa_lvb,
3071 aa->oa_upcall, aa->oa_cookie, aa->oa_flags, rc);
3073 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE, 10);
3075 /* Release the lock for async request. */
3076 if (lustre_handle_is_used(&handle) && rc == ELDLM_OK)
3078 * Releases a reference taken by ldlm_cli_enqueue(), if it is
3079 * not already released by
3080 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
3082 ldlm_lock_decref(&handle, mode);
3084 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
3085 aa->oa_lockh, req, aa);
3086 ldlm_lock_decref(&handle, mode);
3087 LDLM_LOCK_PUT(lock);
3091 void osc_update_enqueue(struct lustre_handle *lov_lockhp,
3092 struct lov_oinfo *loi, int flags,
3093 struct ost_lvb *lvb, __u32 mode, int rc)
3095 if (rc == ELDLM_OK) {
3096 struct ldlm_lock *lock = ldlm_handle2lock(lov_lockhp);
3099 LASSERT(lock != NULL);
3100 loi->loi_lvb = *lvb;
3101 tmp = loi->loi_lvb.lvb_size;
3102 /* Extend KMS up to the end of this lock and no further
3103 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
3104 if (tmp > lock->l_policy_data.l_extent.end)
3105 tmp = lock->l_policy_data.l_extent.end + 1;
3106 if (tmp >= loi->loi_kms) {
3107 LDLM_DEBUG(lock, "lock acquired, setting rss="LPU64
3108 ", kms="LPU64, loi->loi_lvb.lvb_size, tmp);
3109 loi_kms_set(loi, tmp);
3111 LDLM_DEBUG(lock, "lock acquired, setting rss="
3112 LPU64"; leaving kms="LPU64", end="LPU64,
3113 loi->loi_lvb.lvb_size, loi->loi_kms,
3114 lock->l_policy_data.l_extent.end);
3116 ldlm_lock_allow_match(lock);
3117 LDLM_LOCK_PUT(lock);
3118 } else if (rc == ELDLM_LOCK_ABORTED && (flags & LDLM_FL_HAS_INTENT)) {
3119 loi->loi_lvb = *lvb;
3120 CDEBUG(D_INODE, "glimpsed, setting rss="LPU64"; leaving"
3121 " kms="LPU64"\n", loi->loi_lvb.lvb_size, loi->loi_kms);
3125 EXPORT_SYMBOL(osc_update_enqueue);
3127 struct ptlrpc_request_set *PTLRPCD_SET = (void *)1;
3129 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
3130 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
3131 * other synchronous requests, however keeping some locks and trying to obtain
3132 * others may take a considerable amount of time in a case of ost failure; and
3133 * when other sync requests do not get released lock from a client, the client
3134 * is excluded from the cluster -- such scenarious make the life difficult, so
3135 * release locks just after they are obtained. */
3136 int osc_enqueue_base(struct obd_export *exp, struct ldlm_res_id *res_id,
3137 int *flags, ldlm_policy_data_t *policy,
3138 struct ost_lvb *lvb, int kms_valid,
3139 obd_enqueue_update_f upcall, void *cookie,
3140 struct ldlm_enqueue_info *einfo,
3141 struct lustre_handle *lockh,
3142 struct ptlrpc_request_set *rqset, int async)
3144 struct obd_device *obd = exp->exp_obd;
3145 struct ptlrpc_request *req = NULL;
3146 int intent = *flags & LDLM_FL_HAS_INTENT;
3151 /* Filesystem lock extents are extended to page boundaries so that
3152 * dealing with the page cache is a little smoother. */
3153 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
3154 policy->l_extent.end |= ~CFS_PAGE_MASK;
3157 * kms is not valid when either object is completely fresh (so that no
3158 * locks are cached), or object was evicted. In the latter case cached
3159 * lock cannot be used, because it would prime inode state with
3160 * potentially stale LVB.
3165 /* Next, search for already existing extent locks that will cover us */
3166 /* If we're trying to read, we also search for an existing PW lock. The
3167 * VFS and page cache already protect us locally, so lots of readers/
3168 * writers can share a single PW lock.
3170 * There are problems with conversion deadlocks, so instead of
3171 * converting a read lock to a write lock, we'll just enqueue a new
3174 * At some point we should cancel the read lock instead of making them
3175 * send us a blocking callback, but there are problems with canceling
3176 * locks out from other users right now, too. */
3177 mode = einfo->ei_mode;
3178 if (einfo->ei_mode == LCK_PR)
3180 mode = ldlm_lock_match(obd->obd_namespace,
3181 *flags | LDLM_FL_LVB_READY, res_id,
3182 einfo->ei_type, policy, mode, lockh, 0);
3184 struct ldlm_lock *matched = ldlm_handle2lock(lockh);
3186 if (matched->l_ast_data == NULL ||
3187 matched->l_ast_data == einfo->ei_cbdata) {
3188 /* addref the lock only if not async requests and PW
3189 * lock is matched whereas we asked for PR. */
3190 if (!rqset && einfo->ei_mode != mode)
3191 ldlm_lock_addref(lockh, LCK_PR);
3192 osc_set_lock_data_with_check(matched, einfo, *flags);
3194 /* I would like to be able to ASSERT here that
3195 * rss <= kms, but I can't, for reasons which
3196 * are explained in lov_enqueue() */
3199 /* We already have a lock, and it's referenced */
3200 (*upcall)(cookie, ELDLM_OK);
3202 /* For async requests, decref the lock. */
3203 if (einfo->ei_mode != mode)
3204 ldlm_lock_decref(lockh, LCK_PW);
3206 ldlm_lock_decref(lockh, einfo->ei_mode);
3207 LDLM_LOCK_PUT(matched);
3210 ldlm_lock_decref(lockh, mode);
3211 LDLM_LOCK_PUT(matched);
3216 CFS_LIST_HEAD(cancels);
3217 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3218 &RQF_LDLM_ENQUEUE_LVB);
3222 rc = ldlm_prep_enqueue_req(exp, req, &cancels, 0);
3226 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
3228 ptlrpc_request_set_replen(req);
3231 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
3232 *flags &= ~LDLM_FL_BLOCK_GRANTED;
3234 rc = ldlm_cli_enqueue(exp, &req, einfo, res_id, policy, flags, lvb,
3235 sizeof(*lvb), lustre_swab_ost_lvb, lockh, async);
3238 struct osc_enqueue_args *aa;
3239 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
3240 aa = ptlrpc_req_async_args(req);
3243 aa->oa_flags = flags;
3244 aa->oa_upcall = upcall;
3245 aa->oa_cookie = cookie;
3247 aa->oa_lockh = lockh;
3249 req->rq_interpret_reply =
3250 (ptlrpc_interpterer_t)osc_enqueue_interpret;
3251 if (rqset == PTLRPCD_SET)
3252 ptlrpcd_add_req(req, PSCOPE_OTHER);
3254 ptlrpc_set_add_req(rqset, req);
3255 } else if (intent) {
3256 ptlrpc_req_finished(req);
3261 rc = osc_enqueue_fini(req, lvb, upcall, cookie, flags, rc);
3263 ptlrpc_req_finished(req);
3268 static int osc_enqueue(struct obd_export *exp, struct obd_info *oinfo,
3269 struct ldlm_enqueue_info *einfo,
3270 struct ptlrpc_request_set *rqset)
3272 struct ldlm_res_id res_id;
3276 osc_build_res_name(oinfo->oi_md->lsm_object_id,
3277 oinfo->oi_md->lsm_object_gr, &res_id);
3279 rc = osc_enqueue_base(exp, &res_id, &oinfo->oi_flags, &oinfo->oi_policy,
3280 &oinfo->oi_md->lsm_oinfo[0]->loi_lvb,
3281 oinfo->oi_md->lsm_oinfo[0]->loi_kms_valid,
3282 oinfo->oi_cb_up, oinfo, einfo, oinfo->oi_lockh,
3283 rqset, rqset != NULL);
3287 int osc_match_base(struct obd_export *exp, struct ldlm_res_id *res_id,
3288 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
3289 int *flags, void *data, struct lustre_handle *lockh,
3292 struct obd_device *obd = exp->exp_obd;
3293 int lflags = *flags;
3297 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH))
3300 /* Filesystem lock extents are extended to page boundaries so that
3301 * dealing with the page cache is a little smoother */
3302 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
3303 policy->l_extent.end |= ~CFS_PAGE_MASK;
3305 /* Next, search for already existing extent locks that will cover us */
3306 /* If we're trying to read, we also search for an existing PW lock. The
3307 * VFS and page cache already protect us locally, so lots of readers/
3308 * writers can share a single PW lock. */
3312 rc = ldlm_lock_match(obd->obd_namespace, lflags | LDLM_FL_LVB_READY,
3313 res_id, type, policy, rc, lockh, unref);
3316 osc_set_data_with_check(lockh, data, lflags);
3317 if (!(lflags & LDLM_FL_TEST_LOCK) && mode != rc) {
3318 ldlm_lock_addref(lockh, LCK_PR);
3319 ldlm_lock_decref(lockh, LCK_PW);
3326 int osc_cancel_base(struct lustre_handle *lockh, __u32 mode)
3330 if (unlikely(mode == LCK_GROUP))
3331 ldlm_lock_decref_and_cancel(lockh, mode);
3333 ldlm_lock_decref(lockh, mode);
3338 static int osc_cancel(struct obd_export *exp, struct lov_stripe_md *md,
3339 __u32 mode, struct lustre_handle *lockh)
3342 RETURN(osc_cancel_base(lockh, mode));
3345 static int osc_cancel_unused(struct obd_export *exp,
3346 struct lov_stripe_md *lsm, int flags,
3349 struct obd_device *obd = class_exp2obd(exp);
3350 struct ldlm_res_id res_id, *resp = NULL;
3353 resp = osc_build_res_name(lsm->lsm_object_id,
3354 lsm->lsm_object_gr, &res_id);
3357 return ldlm_cli_cancel_unused(obd->obd_namespace, resp, flags, opaque);
3360 static int osc_statfs_interpret(const struct lu_env *env,
3361 struct ptlrpc_request *req,
3362 struct osc_async_args *aa, int rc)
3364 struct obd_statfs *msfs;
3367 if ((rc == -ENOTCONN || rc == -EAGAIN) &&
3368 (aa->aa_oi->oi_flags & OBD_STATFS_NODELAY))
3374 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3376 GOTO(out, rc = -EPROTO);
3379 *aa->aa_oi->oi_osfs = *msfs;
3381 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
3385 static int osc_statfs_async(struct obd_device *obd, struct obd_info *oinfo,
3386 __u64 max_age, struct ptlrpc_request_set *rqset)
3388 struct ptlrpc_request *req;
3389 struct osc_async_args *aa;
3393 /* We could possibly pass max_age in the request (as an absolute
3394 * timestamp or a "seconds.usec ago") so the target can avoid doing
3395 * extra calls into the filesystem if that isn't necessary (e.g.
3396 * during mount that would help a bit). Having relative timestamps
3397 * is not so great if request processing is slow, while absolute
3398 * timestamps are not ideal because they need time synchronization. */
3399 req = ptlrpc_request_alloc(obd->u.cli.cl_import, &RQF_OST_STATFS);
3403 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3405 ptlrpc_request_free(req);
3408 ptlrpc_request_set_replen(req);
3409 req->rq_request_portal = OST_CREATE_PORTAL;
3410 ptlrpc_at_set_req_timeout(req);
3412 if (oinfo->oi_flags & OBD_STATFS_NODELAY) {
3413 /* procfs requests not want stat in wait for avoid deadlock */
3414 req->rq_no_resend = 1;
3415 req->rq_no_delay = 1;
3418 req->rq_interpret_reply = (ptlrpc_interpterer_t)osc_statfs_interpret;
3419 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
3420 aa = ptlrpc_req_async_args(req);
3423 ptlrpc_set_add_req(rqset, req);
3427 static int osc_statfs(struct obd_device *obd, struct obd_statfs *osfs,
3428 __u64 max_age, __u32 flags)
3430 struct obd_statfs *msfs;
3431 struct ptlrpc_request *req;
3432 struct obd_import *imp = NULL;
3436 /*Since the request might also come from lprocfs, so we need
3437 *sync this with client_disconnect_export Bug15684*/
3438 down_read(&obd->u.cli.cl_sem);
3439 if (obd->u.cli.cl_import)
3440 imp = class_import_get(obd->u.cli.cl_import);
3441 up_read(&obd->u.cli.cl_sem);
3445 /* We could possibly pass max_age in the request (as an absolute
3446 * timestamp or a "seconds.usec ago") so the target can avoid doing
3447 * extra calls into the filesystem if that isn't necessary (e.g.
3448 * during mount that would help a bit). Having relative timestamps
3449 * is not so great if request processing is slow, while absolute
3450 * timestamps are not ideal because they need time synchronization. */
3451 req = ptlrpc_request_alloc(imp, &RQF_OST_STATFS);
3453 class_import_put(imp);
3458 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_STATFS);
3460 ptlrpc_request_free(req);
3463 ptlrpc_request_set_replen(req);
3464 req->rq_request_portal = OST_CREATE_PORTAL;
3465 ptlrpc_at_set_req_timeout(req);
3467 if (flags & OBD_STATFS_NODELAY) {
3468 /* procfs requests not want stat in wait for avoid deadlock */
3469 req->rq_no_resend = 1;
3470 req->rq_no_delay = 1;
3473 rc = ptlrpc_queue_wait(req);
3477 msfs = req_capsule_server_get(&req->rq_pill, &RMF_OBD_STATFS);
3479 GOTO(out, rc = -EPROTO);
3486 ptlrpc_req_finished(req);
3490 /* Retrieve object striping information.
3492 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
3493 * the maximum number of OST indices which will fit in the user buffer.
3494 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
3496 static int osc_getstripe(struct lov_stripe_md *lsm, struct lov_user_md *lump)
3498 /* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
3499 struct lov_user_md_v3 lum, *lumk;
3500 struct lov_user_ost_data_v1 *lmm_objects;
3501 int rc = 0, lum_size;
3507 /* we only need the header part from user space to get lmm_magic and
3508 * lmm_stripe_count, (the header part is common to v1 and v3) */
3509 lum_size = sizeof(struct lov_user_md_v1);
3510 if (copy_from_user(&lum, lump, lum_size))
3513 if ((lum.lmm_magic != LOV_USER_MAGIC_V1) &&
3514 (lum.lmm_magic != LOV_USER_MAGIC_V3))
3517 /* lov_user_md_vX and lov_mds_md_vX must have the same size */
3518 LASSERT(sizeof(struct lov_user_md_v1) == sizeof(struct lov_mds_md_v1));
3519 LASSERT(sizeof(struct lov_user_md_v3) == sizeof(struct lov_mds_md_v3));
3520 LASSERT(sizeof(lum.lmm_objects[0]) == sizeof(lumk->lmm_objects[0]));
3522 /* we can use lov_mds_md_size() to compute lum_size
3523 * because lov_user_md_vX and lov_mds_md_vX have the same size */
3524 if (lum.lmm_stripe_count > 0) {
3525 lum_size = lov_mds_md_size(lum.lmm_stripe_count, lum.lmm_magic);
3526 OBD_ALLOC(lumk, lum_size);
3530 if (lum.lmm_magic == LOV_USER_MAGIC_V1)
3531 lmm_objects = &(((struct lov_user_md_v1 *)lumk)->lmm_objects[0]);
3533 lmm_objects = &(lumk->lmm_objects[0]);
3534 lmm_objects->l_object_id = lsm->lsm_object_id;
3536 lum_size = lov_mds_md_size(0, lum.lmm_magic);
3540 lumk->lmm_object_id = lsm->lsm_object_id;
3541 lumk->lmm_object_gr = lsm->lsm_object_gr;
3542 lumk->lmm_stripe_count = 1;
3544 if (copy_to_user(lump, lumk, lum_size))
3548 OBD_FREE(lumk, lum_size);
3554 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
3555 void *karg, void *uarg)
3557 struct obd_device *obd = exp->exp_obd;
3558 struct obd_ioctl_data *data = karg;
3562 if (!try_module_get(THIS_MODULE)) {
3563 CERROR("Can't get module. Is it alive?");
3567 case OBD_IOC_LOV_GET_CONFIG: {
3569 struct lov_desc *desc;
3570 struct obd_uuid uuid;
3574 if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
3575 GOTO(out, err = -EINVAL);
3577 data = (struct obd_ioctl_data *)buf;
3579 if (sizeof(*desc) > data->ioc_inllen1) {
3580 obd_ioctl_freedata(buf, len);
3581 GOTO(out, err = -EINVAL);
3584 if (data->ioc_inllen2 < sizeof(uuid)) {
3585 obd_ioctl_freedata(buf, len);
3586 GOTO(out, err = -EINVAL);
3589 desc = (struct lov_desc *)data->ioc_inlbuf1;
3590 desc->ld_tgt_count = 1;
3591 desc->ld_active_tgt_count = 1;
3592 desc->ld_default_stripe_count = 1;
3593 desc->ld_default_stripe_size = 0;
3594 desc->ld_default_stripe_offset = 0;
3595 desc->ld_pattern = 0;
3596 memcpy(&desc->ld_uuid, &obd->obd_uuid, sizeof(uuid));
3598 memcpy(data->ioc_inlbuf2, &obd->obd_uuid, sizeof(uuid));
3600 err = copy_to_user((void *)uarg, buf, len);
3603 obd_ioctl_freedata(buf, len);
3606 case LL_IOC_LOV_SETSTRIPE:
3607 err = obd_alloc_memmd(exp, karg);
3611 case LL_IOC_LOV_GETSTRIPE:
3612 err = osc_getstripe(karg, uarg);
3614 case OBD_IOC_CLIENT_RECOVER:
3615 err = ptlrpc_recover_import(obd->u.cli.cl_import,
3620 case IOC_OSC_SET_ACTIVE:
3621 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
3624 case OBD_IOC_POLL_QUOTACHECK:
3625 err = lquota_poll_check(quota_interface, exp,
3626 (struct if_quotacheck *)karg);
3628 case OBD_IOC_PING_TARGET:
3629 err = ptlrpc_obd_ping(obd);
3632 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
3633 cmd, cfs_curproc_comm());
3634 GOTO(out, err = -ENOTTY);
3637 module_put(THIS_MODULE);
3641 static int osc_get_info(struct obd_export *exp, obd_count keylen,
3642 void *key, __u32 *vallen, void *val,
3643 struct lov_stripe_md *lsm)
3646 if (!vallen || !val)
3649 if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
3650 __u32 *stripe = val;
3651 *vallen = sizeof(*stripe);
3654 } else if (KEY_IS(KEY_LAST_ID)) {
3655 struct ptlrpc_request *req;
3660 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3661 &RQF_OST_GET_INFO_LAST_ID);
3665 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3666 RCL_CLIENT, keylen);
3667 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3669 ptlrpc_request_free(req);
3673 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3674 memcpy(tmp, key, keylen);
3676 req->rq_no_delay = req->rq_no_resend = 1;
3677 ptlrpc_request_set_replen(req);
3678 rc = ptlrpc_queue_wait(req);
3682 reply = req_capsule_server_get(&req->rq_pill, &RMF_OBD_ID);
3684 GOTO(out, rc = -EPROTO);
3686 *((obd_id *)val) = *reply;
3688 ptlrpc_req_finished(req);
3690 } else if (KEY_IS(KEY_FIEMAP)) {
3691 struct ptlrpc_request *req;
3692 struct ll_user_fiemap *reply;
3696 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
3697 &RQF_OST_GET_INFO_FIEMAP);
3701 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_KEY,
3702 RCL_CLIENT, keylen);
3703 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3704 RCL_CLIENT, *vallen);
3705 req_capsule_set_size(&req->rq_pill, &RMF_FIEMAP_VAL,
3706 RCL_SERVER, *vallen);
3708 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_GET_INFO);
3710 ptlrpc_request_free(req);
3714 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_KEY);
3715 memcpy(tmp, key, keylen);
3716 tmp = req_capsule_client_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3717 memcpy(tmp, val, *vallen);
3719 ptlrpc_request_set_replen(req);
3720 rc = ptlrpc_queue_wait(req);
3724 reply = req_capsule_server_get(&req->rq_pill, &RMF_FIEMAP_VAL);
3726 GOTO(out1, rc = -EPROTO);
3728 memcpy(val, reply, *vallen);
3730 ptlrpc_req_finished(req);
3738 static int osc_setinfo_mds_connect_import(struct obd_import *imp)
3740 struct llog_ctxt *ctxt;
3744 ctxt = llog_get_context(imp->imp_obd, LLOG_MDS_OST_ORIG_CTXT);
3746 rc = llog_initiator_connect(ctxt);
3747 llog_ctxt_put(ctxt);
3749 /* XXX return an error? skip setting below flags? */
3752 spin_lock(&imp->imp_lock);
3753 imp->imp_server_timeout = 1;
3754 imp->imp_pingable = 1;
3755 spin_unlock(&imp->imp_lock);
3756 CDEBUG(D_RPCTRACE, "pinging OST %s\n", obd2cli_tgt(imp->imp_obd));
3761 static int osc_setinfo_mds_conn_interpret(const struct lu_env *env,
3762 struct ptlrpc_request *req,
3769 RETURN(osc_setinfo_mds_connect_import(req->rq_import));
3772 static int osc_set_info_async(struct obd_export *exp, obd_count keylen,
3773 void *key, obd_count vallen, void *val,
3774 struct ptlrpc_request_set *set)
3776 struct ptlrpc_request *req;
3777 struct obd_device *obd = exp->exp_obd;
3778 struct obd_import *imp = class_exp2cliimp(exp);
3783 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3785 if (KEY_IS(KEY_NEXT_ID)) {
3786 if (vallen != sizeof(obd_id))
3790 obd->u.cli.cl_oscc.oscc_next_id = *((obd_id*)val) + 1;
3791 CDEBUG(D_HA, "%s: set oscc_next_id = "LPU64"\n",
3792 exp->exp_obd->obd_name,
3793 obd->u.cli.cl_oscc.oscc_next_id);
3798 if (KEY_IS(KEY_UNLINKED)) {
3799 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3800 spin_lock(&oscc->oscc_lock);
3801 oscc->oscc_flags &= ~OSCC_FLAG_NOSPC;
3802 spin_unlock(&oscc->oscc_lock);
3806 if (KEY_IS(KEY_INIT_RECOV)) {
3807 if (vallen != sizeof(int))
3809 spin_lock(&imp->imp_lock);
3810 imp->imp_initial_recov = *(int *)val;
3811 spin_unlock(&imp->imp_lock);
3812 CDEBUG(D_HA, "%s: set imp_initial_recov = %d\n",
3813 exp->exp_obd->obd_name,
3814 imp->imp_initial_recov);
3818 if (KEY_IS(KEY_CHECKSUM)) {
3819 if (vallen != sizeof(int))
3821 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3825 if (KEY_IS(KEY_SPTLRPC_CONF)) {
3826 sptlrpc_conf_client_adapt(obd);
3830 if (KEY_IS(KEY_FLUSH_CTX)) {
3831 sptlrpc_import_flush_my_ctx(imp);
3835 if (!set && !KEY_IS(KEY_GRANT_SHRINK))
3838 /* We pass all other commands directly to OST. Since nobody calls osc
3839 methods directly and everybody is supposed to go through LOV, we
3840 assume lov checked invalid values for us.
3841 The only recognised values so far are evict_by_nid and mds_conn.
3842 Even if something bad goes through, we'd get a -EINVAL from OST
3845 if (KEY_IS(KEY_GRANT_SHRINK))
3846 req = ptlrpc_request_alloc(imp, &RQF_OST_SET_GRANT_INFO);
3848 req = ptlrpc_request_alloc(imp, &RQF_OST_SET_INFO);
3853 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_KEY,
3854 RCL_CLIENT, keylen);
3855 req_capsule_set_size(&req->rq_pill, &RMF_SETINFO_VAL,
3856 RCL_CLIENT, vallen);
3857 rc = ptlrpc_request_pack(req, LUSTRE_OST_VERSION, OST_SET_INFO);
3859 ptlrpc_request_free(req);
3863 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_KEY);
3864 memcpy(tmp, key, keylen);
3865 tmp = req_capsule_client_get(&req->rq_pill, &RMF_SETINFO_VAL);
3866 memcpy(tmp, val, vallen);
3868 if (KEY_IS(KEY_MDS_CONN)) {
3869 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3871 oscc->oscc_oa.o_gr = (*(__u32 *)val);
3872 oscc->oscc_oa.o_valid |= OBD_MD_FLGROUP;
3873 LASSERT_MDS_GROUP(oscc->oscc_oa.o_gr);
3874 req->rq_no_delay = req->rq_no_resend = 1;
3875 req->rq_interpret_reply = osc_setinfo_mds_conn_interpret;
3876 } else if (KEY_IS(KEY_GRANT_SHRINK)) {
3877 struct osc_grant_args *aa;
3880 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
3881 aa = ptlrpc_req_async_args(req);
3884 ptlrpc_req_finished(req);
3887 *oa = ((struct ost_body *)val)->oa;
3889 req->rq_interpret_reply = osc_shrink_grant_interpret;
3892 ptlrpc_request_set_replen(req);
3893 if (!KEY_IS(KEY_GRANT_SHRINK)) {
3894 LASSERT(set != NULL);
3895 ptlrpc_set_add_req(set, req);
3896 ptlrpc_check_set(NULL, set);
3898 ptlrpcd_add_req(req, PSCOPE_OTHER);
3904 static struct llog_operations osc_size_repl_logops = {
3905 lop_cancel: llog_obd_repl_cancel
3908 static struct llog_operations osc_mds_ost_orig_logops;
3909 static int osc_llog_init(struct obd_device *obd, struct obd_llog_group *olg,
3910 struct obd_device *tgt, int count,
3911 struct llog_catid *catid, struct obd_uuid *uuid)
3916 LASSERT(olg == &obd->obd_olg);
3917 spin_lock(&obd->obd_dev_lock);
3918 if (osc_mds_ost_orig_logops.lop_setup != llog_obd_origin_setup) {
3919 osc_mds_ost_orig_logops = llog_lvfs_ops;
3920 osc_mds_ost_orig_logops.lop_setup = llog_obd_origin_setup;
3921 osc_mds_ost_orig_logops.lop_cleanup = llog_obd_origin_cleanup;
3922 osc_mds_ost_orig_logops.lop_add = llog_obd_origin_add;
3923 osc_mds_ost_orig_logops.lop_connect = llog_origin_connect;
3925 spin_unlock(&obd->obd_dev_lock);
3927 rc = llog_setup(obd, &obd->obd_olg, LLOG_MDS_OST_ORIG_CTXT, tgt, count,
3928 &catid->lci_logid, &osc_mds_ost_orig_logops);
3930 CERROR("failed LLOG_MDS_OST_ORIG_CTXT\n");
3934 rc = llog_setup(obd, &obd->obd_olg, LLOG_SIZE_REPL_CTXT, tgt, count,
3935 NULL, &osc_size_repl_logops);
3937 struct llog_ctxt *ctxt =
3938 llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
3941 CERROR("failed LLOG_SIZE_REPL_CTXT\n");
3946 CERROR("osc '%s' tgt '%s' cnt %d catid %p rc=%d\n",
3947 obd->obd_name, tgt->obd_name, count, catid, rc);
3948 CERROR("logid "LPX64":0x%x\n",
3949 catid->lci_logid.lgl_oid, catid->lci_logid.lgl_ogen);
3954 static int osc_llog_finish(struct obd_device *obd, int count)
3956 struct llog_ctxt *ctxt;
3957 int rc = 0, rc2 = 0;
3960 ctxt = llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
3962 rc = llog_cleanup(ctxt);
3964 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3966 rc2 = llog_cleanup(ctxt);
3973 static int osc_reconnect(const struct lu_env *env,
3974 struct obd_export *exp, struct obd_device *obd,
3975 struct obd_uuid *cluuid,
3976 struct obd_connect_data *data,
3979 struct client_obd *cli = &obd->u.cli;
3981 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3984 client_obd_list_lock(&cli->cl_loi_list_lock);
3985 data->ocd_grant = cli->cl_avail_grant ?:
3986 2 * cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT;
3987 lost_grant = cli->cl_lost_grant;
3988 cli->cl_lost_grant = 0;
3989 client_obd_list_unlock(&cli->cl_loi_list_lock);
3991 CDEBUG(D_CACHE, "request ocd_grant: %d cl_avail_grant: %ld "
3992 "cl_lost_grant: %ld\n", data->ocd_grant,
3993 cli->cl_avail_grant, lost_grant);
3994 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
3995 " ocd_grant: %d\n", data->ocd_connect_flags,
3996 data->ocd_version, data->ocd_grant);
4002 static int osc_disconnect(struct obd_export *exp)
4004 struct obd_device *obd = class_exp2obd(exp);
4005 struct llog_ctxt *ctxt;
4008 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
4010 if (obd->u.cli.cl_conn_count == 1) {
4011 /* Flush any remaining cancel messages out to the
4013 llog_sync(ctxt, exp);
4015 llog_ctxt_put(ctxt);
4017 CDEBUG(D_HA, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
4021 rc = client_disconnect_export(exp);
4023 * Initially we put del_shrink_grant before disconnect_export, but it
4024 * causes the following problem if setup (connect) and cleanup
4025 * (disconnect) are tangled together.
4026 * connect p1 disconnect p2
4027 * ptlrpc_connect_import
4028 * ............... class_manual_cleanup
4031 * ptlrpc_connect_interrupt
4033 * add this client to shrink list
4035 * Bang! pinger trigger the shrink.
4036 * So the osc should be disconnected from the shrink list, after we
4037 * are sure the import has been destroyed. BUG18662
4039 if (obd->u.cli.cl_import == NULL)
4040 osc_del_shrink_grant(&obd->u.cli);
4044 static int osc_import_event(struct obd_device *obd,
4045 struct obd_import *imp,
4046 enum obd_import_event event)
4048 struct client_obd *cli;
4052 LASSERT(imp->imp_obd == obd);
4055 case IMP_EVENT_DISCON: {
4056 /* Only do this on the MDS OSC's */
4057 if (imp->imp_server_timeout) {
4058 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4060 spin_lock(&oscc->oscc_lock);
4061 oscc->oscc_flags |= OSCC_FLAG_RECOVERING;
4062 spin_unlock(&oscc->oscc_lock);
4065 client_obd_list_lock(&cli->cl_loi_list_lock);
4066 cli->cl_avail_grant = 0;
4067 cli->cl_lost_grant = 0;
4068 client_obd_list_unlock(&cli->cl_loi_list_lock);
4071 case IMP_EVENT_INACTIVE: {
4072 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
4075 case IMP_EVENT_INVALIDATE: {
4076 struct ldlm_namespace *ns = obd->obd_namespace;
4080 env = cl_env_get(&refcheck);
4084 client_obd_list_lock(&cli->cl_loi_list_lock);
4085 /* all pages go to failing rpcs due to the invalid
4087 osc_check_rpcs(env, cli);
4088 client_obd_list_unlock(&cli->cl_loi_list_lock);
4090 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
4091 cl_env_put(env, &refcheck);
4096 case IMP_EVENT_ACTIVE: {
4097 /* Only do this on the MDS OSC's */
4098 if (imp->imp_server_timeout) {
4099 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4101 spin_lock(&oscc->oscc_lock);
4102 oscc->oscc_flags &= ~OSCC_FLAG_NOSPC;
4103 spin_unlock(&oscc->oscc_lock);
4105 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
4108 case IMP_EVENT_OCD: {
4109 struct obd_connect_data *ocd = &imp->imp_connect_data;
4111 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
4112 osc_init_grant(&obd->u.cli, ocd);
4115 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
4116 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
4118 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
4122 CERROR("Unknown import event %d\n", event);
4128 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
4134 rc = ptlrpcd_addref();
4138 rc = client_obd_setup(obd, lcfg);
4142 struct lprocfs_static_vars lvars = { 0 };
4143 struct client_obd *cli = &obd->u.cli;
4145 lprocfs_osc_init_vars(&lvars);
4146 if (lprocfs_obd_setup(obd, lvars.obd_vars) == 0) {
4147 lproc_osc_attach_seqstat(obd);
4148 sptlrpc_lprocfs_cliobd_attach(obd);
4149 ptlrpc_lprocfs_register_obd(obd);
4153 /* We need to allocate a few requests more, because
4154 brw_interpret tries to create new requests before freeing
4155 previous ones. Ideally we want to have 2x max_rpcs_in_flight
4156 reserved, but I afraid that might be too much wasted RAM
4157 in fact, so 2 is just my guess and still should work. */
4158 cli->cl_import->imp_rq_pool =
4159 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
4161 ptlrpc_add_rqs_to_pool);
4163 CFS_INIT_LIST_HEAD(&cli->cl_grant_shrink_list);
4164 sema_init(&cli->cl_grant_sem, 1);
4170 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
4176 case OBD_CLEANUP_EARLY: {
4177 struct obd_import *imp;
4178 imp = obd->u.cli.cl_import;
4179 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
4180 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
4181 ptlrpc_deactivate_import(imp);
4182 spin_lock(&imp->imp_lock);
4183 imp->imp_pingable = 0;
4184 spin_unlock(&imp->imp_lock);
4187 case OBD_CLEANUP_EXPORTS: {
4188 /* If we set up but never connected, the
4189 client import will not have been cleaned. */
4190 if (obd->u.cli.cl_import) {
4191 struct obd_import *imp;
4192 down_write(&obd->u.cli.cl_sem);
4193 imp = obd->u.cli.cl_import;
4194 CDEBUG(D_CONFIG, "%s: client import never connected\n",
4196 ptlrpc_invalidate_import(imp);
4197 if (imp->imp_rq_pool) {
4198 ptlrpc_free_rq_pool(imp->imp_rq_pool);
4199 imp->imp_rq_pool = NULL;
4201 class_destroy_import(imp);
4202 up_write(&obd->u.cli.cl_sem);
4203 obd->u.cli.cl_import = NULL;
4205 rc = obd_llog_finish(obd, 0);
4207 CERROR("failed to cleanup llogging subsystems\n");
4214 int osc_cleanup(struct obd_device *obd)
4216 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
4220 ptlrpc_lprocfs_unregister_obd(obd);
4221 lprocfs_obd_cleanup(obd);
4223 spin_lock(&oscc->oscc_lock);
4224 oscc->oscc_flags &= ~OSCC_FLAG_RECOVERING;
4225 oscc->oscc_flags |= OSCC_FLAG_EXITING;
4226 spin_unlock(&oscc->oscc_lock);
4228 /* free memory of osc quota cache */
4229 lquota_cleanup(quota_interface, obd);
4231 rc = client_obd_cleanup(obd);
4237 int osc_process_config_base(struct obd_device *obd, struct lustre_cfg *lcfg)
4239 struct lprocfs_static_vars lvars = { 0 };
4242 lprocfs_osc_init_vars(&lvars);
4244 switch (lcfg->lcfg_command) {
4246 rc = class_process_proc_param(PARAM_OSC, lvars.obd_vars,
4256 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
4258 return osc_process_config_base(obd, buf);
4261 struct obd_ops osc_obd_ops = {
4262 .o_owner = THIS_MODULE,
4263 .o_setup = osc_setup,
4264 .o_precleanup = osc_precleanup,
4265 .o_cleanup = osc_cleanup,
4266 .o_add_conn = client_import_add_conn,
4267 .o_del_conn = client_import_del_conn,
4268 .o_connect = client_connect_import,
4269 .o_reconnect = osc_reconnect,
4270 .o_disconnect = osc_disconnect,
4271 .o_statfs = osc_statfs,
4272 .o_statfs_async = osc_statfs_async,
4273 .o_packmd = osc_packmd,
4274 .o_unpackmd = osc_unpackmd,
4275 .o_precreate = osc_precreate,
4276 .o_create = osc_create,
4277 .o_destroy = osc_destroy,
4278 .o_getattr = osc_getattr,
4279 .o_getattr_async = osc_getattr_async,
4280 .o_setattr = osc_setattr,
4281 .o_setattr_async = osc_setattr_async,
4283 .o_punch = osc_punch,
4285 .o_enqueue = osc_enqueue,
4286 .o_change_cbdata = osc_change_cbdata,
4287 .o_cancel = osc_cancel,
4288 .o_cancel_unused = osc_cancel_unused,
4289 .o_iocontrol = osc_iocontrol,
4290 .o_get_info = osc_get_info,
4291 .o_set_info_async = osc_set_info_async,
4292 .o_import_event = osc_import_event,
4293 .o_llog_init = osc_llog_init,
4294 .o_llog_finish = osc_llog_finish,
4295 .o_process_config = osc_process_config,
4298 extern struct lu_kmem_descr osc_caches[];
4299 extern spinlock_t osc_ast_guard;
4300 extern struct lock_class_key osc_ast_guard_class;
4302 int __init osc_init(void)
4304 struct lprocfs_static_vars lvars = { 0 };
4308 /* print an address of _any_ initialized kernel symbol from this
4309 * module, to allow debugging with gdb that doesn't support data
4310 * symbols from modules.*/
4311 CDEBUG(D_CONSOLE, "Lustre OSC module (%p).\n", &osc_caches);
4313 rc = lu_kmem_init(osc_caches);
4315 lprocfs_osc_init_vars(&lvars);
4317 request_module("lquota");
4318 quota_interface = PORTAL_SYMBOL_GET(osc_quota_interface);
4319 lquota_init(quota_interface);
4320 init_obd_quota_ops(quota_interface, &osc_obd_ops);
4322 rc = class_register_type(&osc_obd_ops, NULL, lvars.module_vars,
4323 LUSTRE_OSC_NAME, &osc_device_type);
4325 if (quota_interface)
4326 PORTAL_SYMBOL_PUT(osc_quota_interface);
4327 lu_kmem_fini(osc_caches);
4331 spin_lock_init(&osc_ast_guard);
4332 lockdep_set_class(&osc_ast_guard, &osc_ast_guard_class);
4338 static void /*__exit*/ osc_exit(void)
4340 lu_device_type_fini(&osc_device_type);
4342 lquota_exit(quota_interface);
4343 if (quota_interface)
4344 PORTAL_SYMBOL_PUT(osc_quota_interface);
4346 class_unregister_type(LUSTRE_OSC_NAME);
4347 lu_kmem_fini(osc_caches);
4350 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
4351 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
4352 MODULE_LICENSE("GPL");
4354 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);