4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
27 * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
30 * Copyright (c) 2011, 2012, Intel Corporation.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
37 #define DEBUG_SUBSYSTEM S_OSC
39 #include <linux/libcfs/libcfs.h>
42 #include <lustre_dlm.h>
43 #include <lustre_net.h>
44 #include <lustre/lustre_user.h>
45 #include <obd_cksum.h>
53 #include <lustre_ha.h>
54 #include <lprocfs_status.h>
55 #include <lustre_log.h>
56 #include <lustre_debug.h>
57 #include <lustre_param.h>
58 #include <lustre_fid.h>
59 #include "osc_internal.h"
60 #include "osc_cl_internal.h"
62 static void osc_release_ppga(struct brw_page
**ppga
, obd_count count
);
63 static int brw_interpret(const struct lu_env
*env
,
64 struct ptlrpc_request
*req
, void *data
, int rc
);
65 int osc_cleanup(struct obd_device
*obd
);
67 /* Pack OSC object metadata for disk storage (LE byte order). */
68 static int osc_packmd(struct obd_export
*exp
, struct lov_mds_md
**lmmp
,
69 struct lov_stripe_md
*lsm
)
73 lmm_size
= sizeof(**lmmp
);
77 if (*lmmp
!= NULL
&& lsm
== NULL
) {
78 OBD_FREE(*lmmp
, lmm_size
);
81 } else if (unlikely(lsm
!= NULL
&& ostid_id(&lsm
->lsm_oi
) == 0)) {
86 OBD_ALLOC(*lmmp
, lmm_size
);
92 ostid_cpu_to_le(&lsm
->lsm_oi
, &(*lmmp
)->lmm_oi
);
97 /* Unpack OSC object metadata from disk storage (LE byte order). */
98 static int osc_unpackmd(struct obd_export
*exp
, struct lov_stripe_md
**lsmp
,
99 struct lov_mds_md
*lmm
, int lmm_bytes
)
102 struct obd_import
*imp
= class_exp2cliimp(exp
);
105 if (lmm_bytes
< sizeof(*lmm
)) {
106 CERROR("%s: lov_mds_md too small: %d, need %d\n",
107 exp
->exp_obd
->obd_name
, lmm_bytes
,
111 /* XXX LOV_MAGIC etc check? */
113 if (unlikely(ostid_id(&lmm
->lmm_oi
) == 0)) {
114 CERROR("%s: zero lmm_object_id: rc = %d\n",
115 exp
->exp_obd
->obd_name
, -EINVAL
);
120 lsm_size
= lov_stripe_md_size(1);
124 if (*lsmp
!= NULL
&& lmm
== NULL
) {
125 OBD_FREE((*lsmp
)->lsm_oinfo
[0], sizeof(struct lov_oinfo
));
126 OBD_FREE(*lsmp
, lsm_size
);
132 OBD_ALLOC(*lsmp
, lsm_size
);
133 if (unlikely(*lsmp
== NULL
))
135 OBD_ALLOC((*lsmp
)->lsm_oinfo
[0], sizeof(struct lov_oinfo
));
136 if (unlikely((*lsmp
)->lsm_oinfo
[0] == NULL
)) {
137 OBD_FREE(*lsmp
, lsm_size
);
140 loi_init((*lsmp
)->lsm_oinfo
[0]);
141 } else if (unlikely(ostid_id(&(*lsmp
)->lsm_oi
) == 0)) {
146 /* XXX zero *lsmp? */
147 ostid_le_to_cpu(&lmm
->lmm_oi
, &(*lsmp
)->lsm_oi
);
150 (imp
->imp_connect_data
.ocd_connect_flags
& OBD_CONNECT_MAXBYTES
))
151 (*lsmp
)->lsm_maxbytes
= imp
->imp_connect_data
.ocd_maxbytes
;
153 (*lsmp
)->lsm_maxbytes
= LUSTRE_STRIPE_MAXBYTES
;
158 static inline void osc_pack_capa(struct ptlrpc_request
*req
,
159 struct ost_body
*body
, void *capa
)
161 struct obd_capa
*oc
= (struct obd_capa
*)capa
;
162 struct lustre_capa
*c
;
167 c
= req_capsule_client_get(&req
->rq_pill
, &RMF_CAPA1
);
170 body
->oa
.o_valid
|= OBD_MD_FLOSSCAPA
;
171 DEBUG_CAPA(D_SEC
, c
, "pack");
174 static inline void osc_pack_req_body(struct ptlrpc_request
*req
,
175 struct obd_info
*oinfo
)
177 struct ost_body
*body
;
179 body
= req_capsule_client_get(&req
->rq_pill
, &RMF_OST_BODY
);
182 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
,
184 osc_pack_capa(req
, body
, oinfo
->oi_capa
);
187 static inline void osc_set_capa_size(struct ptlrpc_request
*req
,
188 const struct req_msg_field
*field
,
192 req_capsule_set_size(&req
->rq_pill
, field
, RCL_CLIENT
, 0);
194 /* it is already calculated as sizeof struct obd_capa */
198 static int osc_getattr_interpret(const struct lu_env
*env
,
199 struct ptlrpc_request
*req
,
200 struct osc_async_args
*aa
, int rc
)
202 struct ost_body
*body
;
207 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
209 CDEBUG(D_INODE
, "mode: %o\n", body
->oa
.o_mode
);
210 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
,
211 aa
->aa_oi
->oi_oa
, &body
->oa
);
213 /* This should really be sent by the OST */
214 aa
->aa_oi
->oi_oa
->o_blksize
= DT_MAX_BRW_SIZE
;
215 aa
->aa_oi
->oi_oa
->o_valid
|= OBD_MD_FLBLKSZ
;
217 CDEBUG(D_INFO
, "can't unpack ost_body\n");
219 aa
->aa_oi
->oi_oa
->o_valid
= 0;
222 rc
= aa
->aa_oi
->oi_cb_up(aa
->aa_oi
, rc
);
226 static int osc_getattr_async(struct obd_export
*exp
, struct obd_info
*oinfo
,
227 struct ptlrpc_request_set
*set
)
229 struct ptlrpc_request
*req
;
230 struct osc_async_args
*aa
;
233 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_GETATTR
);
237 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
238 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_GETATTR
);
240 ptlrpc_request_free(req
);
244 osc_pack_req_body(req
, oinfo
);
246 ptlrpc_request_set_replen(req
);
247 req
->rq_interpret_reply
= (ptlrpc_interpterer_t
)osc_getattr_interpret
;
249 CLASSERT(sizeof(*aa
) <= sizeof(req
->rq_async_args
));
250 aa
= ptlrpc_req_async_args(req
);
253 ptlrpc_set_add_req(set
, req
);
257 static int osc_getattr(const struct lu_env
*env
, struct obd_export
*exp
,
258 struct obd_info
*oinfo
)
260 struct ptlrpc_request
*req
;
261 struct ost_body
*body
;
264 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_GETATTR
);
268 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
269 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_GETATTR
);
271 ptlrpc_request_free(req
);
275 osc_pack_req_body(req
, oinfo
);
277 ptlrpc_request_set_replen(req
);
279 rc
= ptlrpc_queue_wait(req
);
283 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
285 GOTO(out
, rc
= -EPROTO
);
287 CDEBUG(D_INODE
, "mode: %o\n", body
->oa
.o_mode
);
288 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
, oinfo
->oi_oa
,
291 oinfo
->oi_oa
->o_blksize
= cli_brw_size(exp
->exp_obd
);
292 oinfo
->oi_oa
->o_valid
|= OBD_MD_FLBLKSZ
;
295 ptlrpc_req_finished(req
);
299 static int osc_setattr(const struct lu_env
*env
, struct obd_export
*exp
,
300 struct obd_info
*oinfo
, struct obd_trans_info
*oti
)
302 struct ptlrpc_request
*req
;
303 struct ost_body
*body
;
306 LASSERT(oinfo
->oi_oa
->o_valid
& OBD_MD_FLGROUP
);
308 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_SETATTR
);
312 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
313 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_SETATTR
);
315 ptlrpc_request_free(req
);
319 osc_pack_req_body(req
, oinfo
);
321 ptlrpc_request_set_replen(req
);
323 rc
= ptlrpc_queue_wait(req
);
327 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
329 GOTO(out
, rc
= -EPROTO
);
331 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
, oinfo
->oi_oa
,
335 ptlrpc_req_finished(req
);
339 static int osc_setattr_interpret(const struct lu_env
*env
,
340 struct ptlrpc_request
*req
,
341 struct osc_setattr_args
*sa
, int rc
)
343 struct ost_body
*body
;
348 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
350 GOTO(out
, rc
= -EPROTO
);
352 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
, sa
->sa_oa
,
355 rc
= sa
->sa_upcall(sa
->sa_cookie
, rc
);
359 int osc_setattr_async_base(struct obd_export
*exp
, struct obd_info
*oinfo
,
360 struct obd_trans_info
*oti
,
361 obd_enqueue_update_f upcall
, void *cookie
,
362 struct ptlrpc_request_set
*rqset
)
364 struct ptlrpc_request
*req
;
365 struct osc_setattr_args
*sa
;
368 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_SETATTR
);
372 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
373 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_SETATTR
);
375 ptlrpc_request_free(req
);
379 if (oti
&& oinfo
->oi_oa
->o_valid
& OBD_MD_FLCOOKIE
)
380 oinfo
->oi_oa
->o_lcookie
= *oti
->oti_logcookies
;
382 osc_pack_req_body(req
, oinfo
);
384 ptlrpc_request_set_replen(req
);
386 /* do mds to ost setattr asynchronously */
388 /* Do not wait for response. */
389 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
391 req
->rq_interpret_reply
=
392 (ptlrpc_interpterer_t
)osc_setattr_interpret
;
394 CLASSERT (sizeof(*sa
) <= sizeof(req
->rq_async_args
));
395 sa
= ptlrpc_req_async_args(req
);
396 sa
->sa_oa
= oinfo
->oi_oa
;
397 sa
->sa_upcall
= upcall
;
398 sa
->sa_cookie
= cookie
;
400 if (rqset
== PTLRPCD_SET
)
401 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
403 ptlrpc_set_add_req(rqset
, req
);
409 static int osc_setattr_async(struct obd_export
*exp
, struct obd_info
*oinfo
,
410 struct obd_trans_info
*oti
,
411 struct ptlrpc_request_set
*rqset
)
413 return osc_setattr_async_base(exp
, oinfo
, oti
,
414 oinfo
->oi_cb_up
, oinfo
, rqset
);
417 int osc_real_create(struct obd_export
*exp
, struct obdo
*oa
,
418 struct lov_stripe_md
**ea
, struct obd_trans_info
*oti
)
420 struct ptlrpc_request
*req
;
421 struct ost_body
*body
;
422 struct lov_stripe_md
*lsm
;
430 rc
= obd_alloc_memmd(exp
, &lsm
);
435 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_CREATE
);
437 GOTO(out
, rc
= -ENOMEM
);
439 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_CREATE
);
441 ptlrpc_request_free(req
);
445 body
= req_capsule_client_get(&req
->rq_pill
, &RMF_OST_BODY
);
448 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
, oa
);
450 ptlrpc_request_set_replen(req
);
452 if ((oa
->o_valid
& OBD_MD_FLFLAGS
) &&
453 oa
->o_flags
== OBD_FL_DELORPHAN
) {
455 "delorphan from OST integration");
456 /* Don't resend the delorphan req */
457 req
->rq_no_resend
= req
->rq_no_delay
= 1;
460 rc
= ptlrpc_queue_wait(req
);
464 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
466 GOTO(out_req
, rc
= -EPROTO
);
468 CDEBUG(D_INFO
, "oa flags %x\n", oa
->o_flags
);
469 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
, oa
, &body
->oa
);
471 oa
->o_blksize
= cli_brw_size(exp
->exp_obd
);
472 oa
->o_valid
|= OBD_MD_FLBLKSZ
;
474 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
475 * have valid lsm_oinfo data structs, so don't go touching that.
476 * This needs to be fixed in a big way.
478 lsm
->lsm_oi
= oa
->o_oi
;
482 oti
->oti_transno
= lustre_msg_get_transno(req
->rq_repmsg
);
484 if (oa
->o_valid
& OBD_MD_FLCOOKIE
) {
485 if (!oti
->oti_logcookies
)
486 oti_alloc_cookies(oti
, 1);
487 *oti
->oti_logcookies
= oa
->o_lcookie
;
491 CDEBUG(D_HA
, "transno: "LPD64
"\n",
492 lustre_msg_get_transno(req
->rq_repmsg
));
494 ptlrpc_req_finished(req
);
497 obd_free_memmd(exp
, &lsm
);
501 int osc_punch_base(struct obd_export
*exp
, struct obd_info
*oinfo
,
502 obd_enqueue_update_f upcall
, void *cookie
,
503 struct ptlrpc_request_set
*rqset
)
505 struct ptlrpc_request
*req
;
506 struct osc_setattr_args
*sa
;
507 struct ost_body
*body
;
510 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_PUNCH
);
514 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
515 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_PUNCH
);
517 ptlrpc_request_free(req
);
520 req
->rq_request_portal
= OST_IO_PORTAL
; /* bug 7198 */
521 ptlrpc_at_set_req_timeout(req
);
523 body
= req_capsule_client_get(&req
->rq_pill
, &RMF_OST_BODY
);
525 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
,
527 osc_pack_capa(req
, body
, oinfo
->oi_capa
);
529 ptlrpc_request_set_replen(req
);
531 req
->rq_interpret_reply
= (ptlrpc_interpterer_t
)osc_setattr_interpret
;
532 CLASSERT (sizeof(*sa
) <= sizeof(req
->rq_async_args
));
533 sa
= ptlrpc_req_async_args(req
);
534 sa
->sa_oa
= oinfo
->oi_oa
;
535 sa
->sa_upcall
= upcall
;
536 sa
->sa_cookie
= cookie
;
537 if (rqset
== PTLRPCD_SET
)
538 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
540 ptlrpc_set_add_req(rqset
, req
);
545 static int osc_punch(const struct lu_env
*env
, struct obd_export
*exp
,
546 struct obd_info
*oinfo
, struct obd_trans_info
*oti
,
547 struct ptlrpc_request_set
*rqset
)
549 oinfo
->oi_oa
->o_size
= oinfo
->oi_policy
.l_extent
.start
;
550 oinfo
->oi_oa
->o_blocks
= oinfo
->oi_policy
.l_extent
.end
;
551 oinfo
->oi_oa
->o_valid
|= OBD_MD_FLSIZE
| OBD_MD_FLBLOCKS
;
552 return osc_punch_base(exp
, oinfo
,
553 oinfo
->oi_cb_up
, oinfo
, rqset
);
556 static int osc_sync_interpret(const struct lu_env
*env
,
557 struct ptlrpc_request
*req
,
560 struct osc_fsync_args
*fa
= arg
;
561 struct ost_body
*body
;
566 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
568 CERROR ("can't unpack ost_body\n");
569 GOTO(out
, rc
= -EPROTO
);
572 *fa
->fa_oi
->oi_oa
= body
->oa
;
574 rc
= fa
->fa_upcall(fa
->fa_cookie
, rc
);
578 int osc_sync_base(struct obd_export
*exp
, struct obd_info
*oinfo
,
579 obd_enqueue_update_f upcall
, void *cookie
,
580 struct ptlrpc_request_set
*rqset
)
582 struct ptlrpc_request
*req
;
583 struct ost_body
*body
;
584 struct osc_fsync_args
*fa
;
587 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_SYNC
);
591 osc_set_capa_size(req
, &RMF_CAPA1
, oinfo
->oi_capa
);
592 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_SYNC
);
594 ptlrpc_request_free(req
);
598 /* overload the size and blocks fields in the oa with start/end */
599 body
= req_capsule_client_get(&req
->rq_pill
, &RMF_OST_BODY
);
601 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
,
603 osc_pack_capa(req
, body
, oinfo
->oi_capa
);
605 ptlrpc_request_set_replen(req
);
606 req
->rq_interpret_reply
= osc_sync_interpret
;
608 CLASSERT(sizeof(*fa
) <= sizeof(req
->rq_async_args
));
609 fa
= ptlrpc_req_async_args(req
);
611 fa
->fa_upcall
= upcall
;
612 fa
->fa_cookie
= cookie
;
614 if (rqset
== PTLRPCD_SET
)
615 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
617 ptlrpc_set_add_req(rqset
, req
);
622 static int osc_sync(const struct lu_env
*env
, struct obd_export
*exp
,
623 struct obd_info
*oinfo
, obd_size start
, obd_size end
,
624 struct ptlrpc_request_set
*set
)
627 CDEBUG(D_INFO
, "oa NULL\n");
631 oinfo
->oi_oa
->o_size
= start
;
632 oinfo
->oi_oa
->o_blocks
= end
;
633 oinfo
->oi_oa
->o_valid
|= (OBD_MD_FLSIZE
| OBD_MD_FLBLOCKS
);
635 return osc_sync_base(exp
, oinfo
, oinfo
->oi_cb_up
, oinfo
, set
);
638 /* Find and cancel locally locks matched by @mode in the resource found by
639 * @objid. Found locks are added into @cancel list. Returns the amount of
640 * locks added to @cancels list. */
641 static int osc_resource_get_unused(struct obd_export
*exp
, struct obdo
*oa
,
642 struct list_head
*cancels
,
643 ldlm_mode_t mode
, int lock_flags
)
645 struct ldlm_namespace
*ns
= exp
->exp_obd
->obd_namespace
;
646 struct ldlm_res_id res_id
;
647 struct ldlm_resource
*res
;
650 /* Return, i.e. cancel nothing, only if ELC is supported (flag in
651 * export) but disabled through procfs (flag in NS).
653 * This distinguishes from a case when ELC is not supported originally,
654 * when we still want to cancel locks in advance and just cancel them
655 * locally, without sending any RPC. */
656 if (exp_connect_cancelset(exp
) && !ns_connect_cancelset(ns
))
659 ostid_build_res_name(&oa
->o_oi
, &res_id
);
660 res
= ldlm_resource_get(ns
, NULL
, &res_id
, 0, 0);
664 LDLM_RESOURCE_ADDREF(res
);
665 count
= ldlm_cancel_resource_local(res
, cancels
, NULL
, mode
,
666 lock_flags
, 0, NULL
);
667 LDLM_RESOURCE_DELREF(res
);
668 ldlm_resource_putref(res
);
672 static int osc_destroy_interpret(const struct lu_env
*env
,
673 struct ptlrpc_request
*req
, void *data
,
676 struct client_obd
*cli
= &req
->rq_import
->imp_obd
->u
.cli
;
678 atomic_dec(&cli
->cl_destroy_in_flight
);
679 wake_up(&cli
->cl_destroy_waitq
);
683 static int osc_can_send_destroy(struct client_obd
*cli
)
685 if (atomic_inc_return(&cli
->cl_destroy_in_flight
) <=
686 cli
->cl_max_rpcs_in_flight
) {
687 /* The destroy request can be sent */
690 if (atomic_dec_return(&cli
->cl_destroy_in_flight
) <
691 cli
->cl_max_rpcs_in_flight
) {
693 * The counter has been modified between the two atomic
696 wake_up(&cli
->cl_destroy_waitq
);
701 int osc_create(const struct lu_env
*env
, struct obd_export
*exp
,
702 struct obdo
*oa
, struct lov_stripe_md
**ea
,
703 struct obd_trans_info
*oti
)
709 LASSERT(oa
->o_valid
& OBD_MD_FLGROUP
);
711 if ((oa
->o_valid
& OBD_MD_FLFLAGS
) &&
712 oa
->o_flags
== OBD_FL_RECREATE_OBJS
) {
713 return osc_real_create(exp
, oa
, ea
, oti
);
716 if (!fid_seq_is_mdt(ostid_seq(&oa
->o_oi
)))
717 return osc_real_create(exp
, oa
, ea
, oti
);
719 /* we should not get here anymore */
725 /* Destroy requests can be async always on the client, and we don't even really
726 * care about the return code since the client cannot do anything at all about
728 * When the MDS is unlinking a filename, it saves the file objects into a
729 * recovery llog, and these object records are cancelled when the OST reports
730 * they were destroyed and sync'd to disk (i.e. transaction committed).
731 * If the client dies, or the OST is down when the object should be destroyed,
732 * the records are not cancelled, and when the OST reconnects to the MDS next,
733 * it will retrieve the llog unlink logs and then sends the log cancellation
734 * cookies to the MDS after committing destroy transactions. */
735 static int osc_destroy(const struct lu_env
*env
, struct obd_export
*exp
,
736 struct obdo
*oa
, struct lov_stripe_md
*ea
,
737 struct obd_trans_info
*oti
, struct obd_export
*md_export
,
740 struct client_obd
*cli
= &exp
->exp_obd
->u
.cli
;
741 struct ptlrpc_request
*req
;
742 struct ost_body
*body
;
747 CDEBUG(D_INFO
, "oa NULL\n");
751 count
= osc_resource_get_unused(exp
, oa
, &cancels
, LCK_PW
,
752 LDLM_FL_DISCARD_DATA
);
754 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
), &RQF_OST_DESTROY
);
756 ldlm_lock_list_put(&cancels
, l_bl_ast
, count
);
760 osc_set_capa_size(req
, &RMF_CAPA1
, (struct obd_capa
*)capa
);
761 rc
= ldlm_prep_elc_req(exp
, req
, LUSTRE_OST_VERSION
, OST_DESTROY
,
764 ptlrpc_request_free(req
);
768 req
->rq_request_portal
= OST_IO_PORTAL
; /* bug 7198 */
769 ptlrpc_at_set_req_timeout(req
);
771 if (oti
!= NULL
&& oa
->o_valid
& OBD_MD_FLCOOKIE
)
772 oa
->o_lcookie
= *oti
->oti_logcookies
;
773 body
= req_capsule_client_get(&req
->rq_pill
, &RMF_OST_BODY
);
775 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
, oa
);
777 osc_pack_capa(req
, body
, (struct obd_capa
*)capa
);
778 ptlrpc_request_set_replen(req
);
780 /* If osc_destory is for destroying the unlink orphan,
781 * sent from MDT to OST, which should not be blocked here,
782 * because the process might be triggered by ptlrpcd, and
783 * it is not good to block ptlrpcd thread (b=16006)*/
784 if (!(oa
->o_flags
& OBD_FL_DELORPHAN
)) {
785 req
->rq_interpret_reply
= osc_destroy_interpret
;
786 if (!osc_can_send_destroy(cli
)) {
787 struct l_wait_info lwi
= LWI_INTR(LWI_ON_SIGNAL_NOOP
,
791 * Wait until the number of on-going destroy RPCs drops
792 * under max_rpc_in_flight
794 l_wait_event_exclusive(cli
->cl_destroy_waitq
,
795 osc_can_send_destroy(cli
), &lwi
);
799 /* Do not wait for response */
800 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
804 static void osc_announce_cached(struct client_obd
*cli
, struct obdo
*oa
,
807 obd_flag bits
= OBD_MD_FLBLOCKS
|OBD_MD_FLGRANT
;
809 LASSERT(!(oa
->o_valid
& bits
));
812 client_obd_list_lock(&cli
->cl_loi_list_lock
);
813 oa
->o_dirty
= cli
->cl_dirty
;
814 if (unlikely(cli
->cl_dirty
- cli
->cl_dirty_transit
>
815 cli
->cl_dirty_max
)) {
816 CERROR("dirty %lu - %lu > dirty_max %lu\n",
817 cli
->cl_dirty
, cli
->cl_dirty_transit
, cli
->cl_dirty_max
);
819 } else if (unlikely(atomic_read(&obd_dirty_pages
) -
820 atomic_read(&obd_dirty_transit_pages
) >
821 (long)(obd_max_dirty_pages
+ 1))) {
822 /* The atomic_read() allowing the atomic_inc() are
823 * not covered by a lock thus they may safely race and trip
824 * this CERROR() unless we add in a small fudge factor (+1). */
825 CERROR("dirty %d - %d > system dirty_max %d\n",
826 atomic_read(&obd_dirty_pages
),
827 atomic_read(&obd_dirty_transit_pages
),
828 obd_max_dirty_pages
);
830 } else if (unlikely(cli
->cl_dirty_max
- cli
->cl_dirty
> 0x7fffffff)) {
831 CERROR("dirty %lu - dirty_max %lu too big???\n",
832 cli
->cl_dirty
, cli
->cl_dirty_max
);
835 long max_in_flight
= (cli
->cl_max_pages_per_rpc
<<
837 (cli
->cl_max_rpcs_in_flight
+ 1);
838 oa
->o_undirty
= max(cli
->cl_dirty_max
, max_in_flight
);
840 oa
->o_grant
= cli
->cl_avail_grant
+ cli
->cl_reserved_grant
;
841 oa
->o_dropped
= cli
->cl_lost_grant
;
842 cli
->cl_lost_grant
= 0;
843 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
844 CDEBUG(D_CACHE
,"dirty: "LPU64
" undirty: %u dropped %u grant: "LPU64
"\n",
845 oa
->o_dirty
, oa
->o_undirty
, oa
->o_dropped
, oa
->o_grant
);
849 void osc_update_next_shrink(struct client_obd
*cli
)
851 cli
->cl_next_shrink_grant
=
852 cfs_time_shift(cli
->cl_grant_shrink_interval
);
853 CDEBUG(D_CACHE
, "next time %ld to shrink grant \n",
854 cli
->cl_next_shrink_grant
);
857 static void __osc_update_grant(struct client_obd
*cli
, obd_size grant
)
859 client_obd_list_lock(&cli
->cl_loi_list_lock
);
860 cli
->cl_avail_grant
+= grant
;
861 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
864 static void osc_update_grant(struct client_obd
*cli
, struct ost_body
*body
)
866 if (body
->oa
.o_valid
& OBD_MD_FLGRANT
) {
867 CDEBUG(D_CACHE
, "got "LPU64
" extra grant\n", body
->oa
.o_grant
);
868 __osc_update_grant(cli
, body
->oa
.o_grant
);
872 static int osc_set_info_async(const struct lu_env
*env
, struct obd_export
*exp
,
873 obd_count keylen
, void *key
, obd_count vallen
,
874 void *val
, struct ptlrpc_request_set
*set
);
876 static int osc_shrink_grant_interpret(const struct lu_env
*env
,
877 struct ptlrpc_request
*req
,
880 struct client_obd
*cli
= &req
->rq_import
->imp_obd
->u
.cli
;
881 struct obdo
*oa
= ((struct osc_grant_args
*)aa
)->aa_oa
;
882 struct ost_body
*body
;
885 __osc_update_grant(cli
, oa
->o_grant
);
889 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
891 osc_update_grant(cli
, body
);
897 static void osc_shrink_grant_local(struct client_obd
*cli
, struct obdo
*oa
)
899 client_obd_list_lock(&cli
->cl_loi_list_lock
);
900 oa
->o_grant
= cli
->cl_avail_grant
/ 4;
901 cli
->cl_avail_grant
-= oa
->o_grant
;
902 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
903 if (!(oa
->o_valid
& OBD_MD_FLFLAGS
)) {
904 oa
->o_valid
|= OBD_MD_FLFLAGS
;
907 oa
->o_flags
|= OBD_FL_SHRINK_GRANT
;
908 osc_update_next_shrink(cli
);
911 /* Shrink the current grant, either from some large amount to enough for a
912 * full set of in-flight RPCs, or if we have already shrunk to that limit
913 * then to enough for a single RPC. This avoids keeping more grant than
914 * needed, and avoids shrinking the grant piecemeal. */
915 static int osc_shrink_grant(struct client_obd
*cli
)
917 __u64 target_bytes
= (cli
->cl_max_rpcs_in_flight
+ 1) *
918 (cli
->cl_max_pages_per_rpc
<< PAGE_CACHE_SHIFT
);
920 client_obd_list_lock(&cli
->cl_loi_list_lock
);
921 if (cli
->cl_avail_grant
<= target_bytes
)
922 target_bytes
= cli
->cl_max_pages_per_rpc
<< PAGE_CACHE_SHIFT
;
923 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
925 return osc_shrink_grant_to_target(cli
, target_bytes
);
928 int osc_shrink_grant_to_target(struct client_obd
*cli
, __u64 target_bytes
)
931 struct ost_body
*body
;
933 client_obd_list_lock(&cli
->cl_loi_list_lock
);
934 /* Don't shrink if we are already above or below the desired limit
935 * We don't want to shrink below a single RPC, as that will negatively
936 * impact block allocation and long-term performance. */
937 if (target_bytes
< cli
->cl_max_pages_per_rpc
<< PAGE_CACHE_SHIFT
)
938 target_bytes
= cli
->cl_max_pages_per_rpc
<< PAGE_CACHE_SHIFT
;
940 if (target_bytes
>= cli
->cl_avail_grant
) {
941 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
944 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
950 osc_announce_cached(cli
, &body
->oa
, 0);
952 client_obd_list_lock(&cli
->cl_loi_list_lock
);
953 body
->oa
.o_grant
= cli
->cl_avail_grant
- target_bytes
;
954 cli
->cl_avail_grant
= target_bytes
;
955 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
956 if (!(body
->oa
.o_valid
& OBD_MD_FLFLAGS
)) {
957 body
->oa
.o_valid
|= OBD_MD_FLFLAGS
;
958 body
->oa
.o_flags
= 0;
960 body
->oa
.o_flags
|= OBD_FL_SHRINK_GRANT
;
961 osc_update_next_shrink(cli
);
963 rc
= osc_set_info_async(NULL
, cli
->cl_import
->imp_obd
->obd_self_export
,
964 sizeof(KEY_GRANT_SHRINK
), KEY_GRANT_SHRINK
,
965 sizeof(*body
), body
, NULL
);
967 __osc_update_grant(cli
, body
->oa
.o_grant
);
972 static int osc_should_shrink_grant(struct client_obd
*client
)
974 cfs_time_t time
= cfs_time_current();
975 cfs_time_t next_shrink
= client
->cl_next_shrink_grant
;
977 if ((client
->cl_import
->imp_connect_data
.ocd_connect_flags
&
978 OBD_CONNECT_GRANT_SHRINK
) == 0)
981 if (cfs_time_aftereq(time
, next_shrink
- 5 * CFS_TICK
)) {
982 /* Get the current RPC size directly, instead of going via:
983 * cli_brw_size(obd->u.cli.cl_import->imp_obd->obd_self_export)
984 * Keep comment here so that it can be found by searching. */
985 int brw_size
= client
->cl_max_pages_per_rpc
<< PAGE_CACHE_SHIFT
;
987 if (client
->cl_import
->imp_state
== LUSTRE_IMP_FULL
&&
988 client
->cl_avail_grant
> brw_size
)
991 osc_update_next_shrink(client
);
996 static int osc_grant_shrink_grant_cb(struct timeout_item
*item
, void *data
)
998 struct client_obd
*client
;
1000 list_for_each_entry(client
, &item
->ti_obd_list
,
1001 cl_grant_shrink_list
) {
1002 if (osc_should_shrink_grant(client
))
1003 osc_shrink_grant(client
);
1008 static int osc_add_shrink_grant(struct client_obd
*client
)
1012 rc
= ptlrpc_add_timeout_client(client
->cl_grant_shrink_interval
,
1014 osc_grant_shrink_grant_cb
, NULL
,
1015 &client
->cl_grant_shrink_list
);
1017 CERROR("add grant client %s error %d\n",
1018 client
->cl_import
->imp_obd
->obd_name
, rc
);
1021 CDEBUG(D_CACHE
, "add grant client %s \n",
1022 client
->cl_import
->imp_obd
->obd_name
);
1023 osc_update_next_shrink(client
);
1027 static int osc_del_shrink_grant(struct client_obd
*client
)
1029 return ptlrpc_del_timeout_client(&client
->cl_grant_shrink_list
,
1033 static void osc_init_grant(struct client_obd
*cli
, struct obd_connect_data
*ocd
)
1036 * ocd_grant is the total grant amount we're expect to hold: if we've
1037 * been evicted, it's the new avail_grant amount, cl_dirty will drop
1038 * to 0 as inflight RPCs fail out; otherwise, it's avail_grant + dirty.
1040 * race is tolerable here: if we're evicted, but imp_state already
1041 * left EVICTED state, then cl_dirty must be 0 already.
1043 client_obd_list_lock(&cli
->cl_loi_list_lock
);
1044 if (cli
->cl_import
->imp_state
== LUSTRE_IMP_EVICTED
)
1045 cli
->cl_avail_grant
= ocd
->ocd_grant
;
1047 cli
->cl_avail_grant
= ocd
->ocd_grant
- cli
->cl_dirty
;
1049 if (cli
->cl_avail_grant
< 0) {
1050 CWARN("%s: available grant < 0: avail/ocd/dirty %ld/%u/%ld\n",
1051 cli
->cl_import
->imp_obd
->obd_name
, cli
->cl_avail_grant
,
1052 ocd
->ocd_grant
, cli
->cl_dirty
);
1053 /* workaround for servers which do not have the patch from
1055 cli
->cl_avail_grant
= ocd
->ocd_grant
;
1058 /* determine the appropriate chunk size used by osc_extent. */
1059 cli
->cl_chunkbits
= max_t(int, PAGE_CACHE_SHIFT
, ocd
->ocd_blocksize
);
1060 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
1062 CDEBUG(D_CACHE
, "%s, setting cl_avail_grant: %ld cl_lost_grant: %ld."
1063 "chunk bits: %d.\n", cli
->cl_import
->imp_obd
->obd_name
,
1064 cli
->cl_avail_grant
, cli
->cl_lost_grant
, cli
->cl_chunkbits
);
1066 if (ocd
->ocd_connect_flags
& OBD_CONNECT_GRANT_SHRINK
&&
1067 list_empty(&cli
->cl_grant_shrink_list
))
1068 osc_add_shrink_grant(cli
);
1071 /* We assume that the reason this OSC got a short read is because it read
1072 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
1073 * via the LOV, and it _knows_ it's reading inside the file, it's just that
1074 * this stripe never got written at or beyond this stripe offset yet. */
1075 static void handle_short_read(int nob_read
, obd_count page_count
,
1076 struct brw_page
**pga
)
1081 /* skip bytes read OK */
1082 while (nob_read
> 0) {
1083 LASSERT (page_count
> 0);
1085 if (pga
[i
]->count
> nob_read
) {
1086 /* EOF inside this page */
1087 ptr
= kmap(pga
[i
]->pg
) +
1088 (pga
[i
]->off
& ~CFS_PAGE_MASK
);
1089 memset(ptr
+ nob_read
, 0, pga
[i
]->count
- nob_read
);
1096 nob_read
-= pga
[i
]->count
;
1101 /* zero remaining pages */
1102 while (page_count
-- > 0) {
1103 ptr
= kmap(pga
[i
]->pg
) + (pga
[i
]->off
& ~CFS_PAGE_MASK
);
1104 memset(ptr
, 0, pga
[i
]->count
);
1110 static int check_write_rcs(struct ptlrpc_request
*req
,
1111 int requested_nob
, int niocount
,
1112 obd_count page_count
, struct brw_page
**pga
)
1117 remote_rcs
= req_capsule_server_sized_get(&req
->rq_pill
, &RMF_RCS
,
1118 sizeof(*remote_rcs
) *
1120 if (remote_rcs
== NULL
) {
1121 CDEBUG(D_INFO
, "Missing/short RC vector on BRW_WRITE reply\n");
1125 /* return error if any niobuf was in error */
1126 for (i
= 0; i
< niocount
; i
++) {
1127 if ((int)remote_rcs
[i
] < 0)
1128 return(remote_rcs
[i
]);
1130 if (remote_rcs
[i
] != 0) {
1131 CDEBUG(D_INFO
, "rc[%d] invalid (%d) req %p\n",
1132 i
, remote_rcs
[i
], req
);
1137 if (req
->rq_bulk
->bd_nob_transferred
!= requested_nob
) {
1138 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
1139 req
->rq_bulk
->bd_nob_transferred
, requested_nob
);
1146 static inline int can_merge_pages(struct brw_page
*p1
, struct brw_page
*p2
)
1148 if (p1
->flag
!= p2
->flag
) {
1149 unsigned mask
= ~(OBD_BRW_FROM_GRANT
| OBD_BRW_NOCACHE
|
1150 OBD_BRW_SYNC
|OBD_BRW_ASYNC
|OBD_BRW_NOQUOTA
);
1152 /* warn if we try to combine flags that we don't know to be
1153 * safe to combine */
1154 if (unlikely((p1
->flag
& mask
) != (p2
->flag
& mask
))) {
1155 CWARN("Saw flags 0x%x and 0x%x in the same brw, please "
1156 "report this at http://bugs.whamcloud.com/\n",
1157 p1
->flag
, p2
->flag
);
1162 return (p1
->off
+ p1
->count
== p2
->off
);
1165 static obd_count
osc_checksum_bulk(int nob
, obd_count pg_count
,
1166 struct brw_page
**pga
, int opc
,
1167 cksum_type_t cksum_type
)
1171 struct cfs_crypto_hash_desc
*hdesc
;
1172 unsigned int bufsize
;
1174 unsigned char cfs_alg
= cksum_obd2cfs(cksum_type
);
1176 LASSERT(pg_count
> 0);
1178 hdesc
= cfs_crypto_hash_init(cfs_alg
, NULL
, 0);
1179 if (IS_ERR(hdesc
)) {
1180 CERROR("Unable to initialize checksum hash %s\n",
1181 cfs_crypto_hash_name(cfs_alg
));
1182 return PTR_ERR(hdesc
);
1185 while (nob
> 0 && pg_count
> 0) {
1186 int count
= pga
[i
]->count
> nob
? nob
: pga
[i
]->count
;
1188 /* corrupt the data before we compute the checksum, to
1189 * simulate an OST->client data error */
1190 if (i
== 0 && opc
== OST_READ
&&
1191 OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_RECEIVE
)) {
1192 unsigned char *ptr
= kmap(pga
[i
]->pg
);
1193 int off
= pga
[i
]->off
& ~CFS_PAGE_MASK
;
1194 memcpy(ptr
+ off
, "bad1", min(4, nob
));
1197 cfs_crypto_hash_update_page(hdesc
, pga
[i
]->pg
,
1198 pga
[i
]->off
& ~CFS_PAGE_MASK
,
1200 LL_CDEBUG_PAGE(D_PAGE
, pga
[i
]->pg
, "off %d\n",
1201 (int)(pga
[i
]->off
& ~CFS_PAGE_MASK
));
1203 nob
-= pga
[i
]->count
;
1209 err
= cfs_crypto_hash_final(hdesc
, (unsigned char *)&cksum
, &bufsize
);
1212 cfs_crypto_hash_final(hdesc
, NULL
, NULL
);
1214 /* For sending we only compute the wrong checksum instead
1215 * of corrupting the data so it is still correct on a redo */
1216 if (opc
== OST_WRITE
&& OBD_FAIL_CHECK(OBD_FAIL_OSC_CHECKSUM_SEND
))
1222 static int osc_brw_prep_request(int cmd
, struct client_obd
*cli
,struct obdo
*oa
,
1223 struct lov_stripe_md
*lsm
, obd_count page_count
,
1224 struct brw_page
**pga
,
1225 struct ptlrpc_request
**reqp
,
1226 struct obd_capa
*ocapa
, int reserve
,
1229 struct ptlrpc_request
*req
;
1230 struct ptlrpc_bulk_desc
*desc
;
1231 struct ost_body
*body
;
1232 struct obd_ioobj
*ioobj
;
1233 struct niobuf_remote
*niobuf
;
1234 int niocount
, i
, requested_nob
, opc
, rc
;
1235 struct osc_brw_async_args
*aa
;
1236 struct req_capsule
*pill
;
1237 struct brw_page
*pg_prev
;
1239 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ
))
1240 return -ENOMEM
; /* Recoverable */
1241 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_BRW_PREP_REQ2
))
1242 return -EINVAL
; /* Fatal */
1244 if ((cmd
& OBD_BRW_WRITE
) != 0) {
1246 req
= ptlrpc_request_alloc_pool(cli
->cl_import
,
1247 cli
->cl_import
->imp_rq_pool
,
1248 &RQF_OST_BRW_WRITE
);
1251 req
= ptlrpc_request_alloc(cli
->cl_import
, &RQF_OST_BRW_READ
);
1256 for (niocount
= i
= 1; i
< page_count
; i
++) {
1257 if (!can_merge_pages(pga
[i
- 1], pga
[i
]))
1261 pill
= &req
->rq_pill
;
1262 req_capsule_set_size(pill
, &RMF_OBD_IOOBJ
, RCL_CLIENT
,
1264 req_capsule_set_size(pill
, &RMF_NIOBUF_REMOTE
, RCL_CLIENT
,
1265 niocount
* sizeof(*niobuf
));
1266 osc_set_capa_size(req
, &RMF_CAPA1
, ocapa
);
1268 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, opc
);
1270 ptlrpc_request_free(req
);
1273 req
->rq_request_portal
= OST_IO_PORTAL
; /* bug 7198 */
1274 ptlrpc_at_set_req_timeout(req
);
1275 /* ask ptlrpc not to resend on EINPROGRESS since BRWs have their own
1277 req
->rq_no_retry_einprogress
= 1;
1279 desc
= ptlrpc_prep_bulk_imp(req
, page_count
,
1280 cli
->cl_import
->imp_connect_data
.ocd_brw_size
>> LNET_MTU_BITS
,
1281 opc
== OST_WRITE
? BULK_GET_SOURCE
: BULK_PUT_SINK
,
1285 GOTO(out
, rc
= -ENOMEM
);
1286 /* NB request now owns desc and will free it when it gets freed */
1288 body
= req_capsule_client_get(pill
, &RMF_OST_BODY
);
1289 ioobj
= req_capsule_client_get(pill
, &RMF_OBD_IOOBJ
);
1290 niobuf
= req_capsule_client_get(pill
, &RMF_NIOBUF_REMOTE
);
1291 LASSERT(body
!= NULL
&& ioobj
!= NULL
&& niobuf
!= NULL
);
1293 lustre_set_wire_obdo(&req
->rq_import
->imp_connect_data
, &body
->oa
, oa
);
1295 obdo_to_ioobj(oa
, ioobj
);
1296 ioobj
->ioo_bufcnt
= niocount
;
1297 /* The high bits of ioo_max_brw tells server _maximum_ number of bulks
1298 * that might be send for this request. The actual number is decided
1299 * when the RPC is finally sent in ptlrpc_register_bulk(). It sends
1300 * "max - 1" for old client compatibility sending "0", and also so the
1301 * the actual maximum is a power-of-two number, not one less. LU-1431 */
1302 ioobj_max_brw_set(ioobj
, desc
->bd_md_max_brw
);
1303 osc_pack_capa(req
, body
, ocapa
);
1304 LASSERT(page_count
> 0);
1306 for (requested_nob
= i
= 0; i
< page_count
; i
++, niobuf
++) {
1307 struct brw_page
*pg
= pga
[i
];
1308 int poff
= pg
->off
& ~CFS_PAGE_MASK
;
1310 LASSERT(pg
->count
> 0);
1311 /* make sure there is no gap in the middle of page array */
1312 LASSERTF(page_count
== 1 ||
1313 (ergo(i
== 0, poff
+ pg
->count
== PAGE_CACHE_SIZE
) &&
1314 ergo(i
> 0 && i
< page_count
- 1,
1315 poff
== 0 && pg
->count
== PAGE_CACHE_SIZE
) &&
1316 ergo(i
== page_count
- 1, poff
== 0)),
1317 "i: %d/%d pg: %p off: "LPU64
", count: %u\n",
1318 i
, page_count
, pg
, pg
->off
, pg
->count
);
1319 LASSERTF(i
== 0 || pg
->off
> pg_prev
->off
,
1320 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1321 " prev_pg %p [pri %lu ind %lu] off "LPU64
"\n",
1323 pg
->pg
, page_private(pg
->pg
), pg
->pg
->index
, pg
->off
,
1324 pg_prev
->pg
, page_private(pg_prev
->pg
),
1325 pg_prev
->pg
->index
, pg_prev
->off
);
1326 LASSERT((pga
[0]->flag
& OBD_BRW_SRVLOCK
) ==
1327 (pg
->flag
& OBD_BRW_SRVLOCK
));
1329 ptlrpc_prep_bulk_page_pin(desc
, pg
->pg
, poff
, pg
->count
);
1330 requested_nob
+= pg
->count
;
1332 if (i
> 0 && can_merge_pages(pg_prev
, pg
)) {
1334 niobuf
->len
+= pg
->count
;
1336 niobuf
->offset
= pg
->off
;
1337 niobuf
->len
= pg
->count
;
1338 niobuf
->flags
= pg
->flag
;
1343 LASSERTF((void *)(niobuf
- niocount
) ==
1344 req_capsule_client_get(&req
->rq_pill
, &RMF_NIOBUF_REMOTE
),
1345 "want %p - real %p\n", req_capsule_client_get(&req
->rq_pill
,
1346 &RMF_NIOBUF_REMOTE
), (void *)(niobuf
- niocount
));
1348 osc_announce_cached(cli
, &body
->oa
, opc
== OST_WRITE
? requested_nob
:0);
1350 if ((body
->oa
.o_valid
& OBD_MD_FLFLAGS
) == 0) {
1351 body
->oa
.o_valid
|= OBD_MD_FLFLAGS
;
1352 body
->oa
.o_flags
= 0;
1354 body
->oa
.o_flags
|= OBD_FL_RECOV_RESEND
;
1357 if (osc_should_shrink_grant(cli
))
1358 osc_shrink_grant_local(cli
, &body
->oa
);
1360 /* size[REQ_REC_OFF] still sizeof (*body) */
1361 if (opc
== OST_WRITE
) {
1362 if (cli
->cl_checksum
&&
1363 !sptlrpc_flavor_has_bulk(&req
->rq_flvr
)) {
1364 /* store cl_cksum_type in a local variable since
1365 * it can be changed via lprocfs */
1366 cksum_type_t cksum_type
= cli
->cl_cksum_type
;
1368 if ((body
->oa
.o_valid
& OBD_MD_FLFLAGS
) == 0) {
1369 oa
->o_flags
&= OBD_FL_LOCAL_MASK
;
1370 body
->oa
.o_flags
= 0;
1372 body
->oa
.o_flags
|= cksum_type_pack(cksum_type
);
1373 body
->oa
.o_valid
|= OBD_MD_FLCKSUM
| OBD_MD_FLFLAGS
;
1374 body
->oa
.o_cksum
= osc_checksum_bulk(requested_nob
,
1378 CDEBUG(D_PAGE
, "checksum at write origin: %x\n",
1380 /* save this in 'oa', too, for later checking */
1381 oa
->o_valid
|= OBD_MD_FLCKSUM
| OBD_MD_FLFLAGS
;
1382 oa
->o_flags
|= cksum_type_pack(cksum_type
);
1384 /* clear out the checksum flag, in case this is a
1385 * resend but cl_checksum is no longer set. b=11238 */
1386 oa
->o_valid
&= ~OBD_MD_FLCKSUM
;
1388 oa
->o_cksum
= body
->oa
.o_cksum
;
1389 /* 1 RC per niobuf */
1390 req_capsule_set_size(pill
, &RMF_RCS
, RCL_SERVER
,
1391 sizeof(__u32
) * niocount
);
1393 if (cli
->cl_checksum
&&
1394 !sptlrpc_flavor_has_bulk(&req
->rq_flvr
)) {
1395 if ((body
->oa
.o_valid
& OBD_MD_FLFLAGS
) == 0)
1396 body
->oa
.o_flags
= 0;
1397 body
->oa
.o_flags
|= cksum_type_pack(cli
->cl_cksum_type
);
1398 body
->oa
.o_valid
|= OBD_MD_FLCKSUM
| OBD_MD_FLFLAGS
;
1401 ptlrpc_request_set_replen(req
);
1403 CLASSERT(sizeof(*aa
) <= sizeof(req
->rq_async_args
));
1404 aa
= ptlrpc_req_async_args(req
);
1406 aa
->aa_requested_nob
= requested_nob
;
1407 aa
->aa_nio_count
= niocount
;
1408 aa
->aa_page_count
= page_count
;
1412 INIT_LIST_HEAD(&aa
->aa_oaps
);
1413 if (ocapa
&& reserve
)
1414 aa
->aa_ocapa
= capa_get(ocapa
);
1420 ptlrpc_req_finished(req
);
1424 static int check_write_checksum(struct obdo
*oa
, const lnet_process_id_t
*peer
,
1425 __u32 client_cksum
, __u32 server_cksum
, int nob
,
1426 obd_count page_count
, struct brw_page
**pga
,
1427 cksum_type_t client_cksum_type
)
1431 cksum_type_t cksum_type
;
1433 if (server_cksum
== client_cksum
) {
1434 CDEBUG(D_PAGE
, "checksum %x confirmed\n", client_cksum
);
1438 cksum_type
= cksum_type_unpack(oa
->o_valid
& OBD_MD_FLFLAGS
?
1440 new_cksum
= osc_checksum_bulk(nob
, page_count
, pga
, OST_WRITE
,
1443 if (cksum_type
!= client_cksum_type
)
1444 msg
= "the server did not use the checksum type specified in "
1445 "the original request - likely a protocol problem";
1446 else if (new_cksum
== server_cksum
)
1447 msg
= "changed on the client after we checksummed it - "
1448 "likely false positive due to mmap IO (bug 11742)";
1449 else if (new_cksum
== client_cksum
)
1450 msg
= "changed in transit before arrival at OST";
1452 msg
= "changed in transit AND doesn't match the original - "
1453 "likely false positive due to mmap IO (bug 11742)";
1455 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inode "DFID
1456 " object "DOSTID
" extent ["LPU64
"-"LPU64
"]\n",
1457 msg
, libcfs_nid2str(peer
->nid
),
1458 oa
->o_valid
& OBD_MD_FLFID
? oa
->o_parent_seq
: (__u64
)0,
1459 oa
->o_valid
& OBD_MD_FLFID
? oa
->o_parent_oid
: 0,
1460 oa
->o_valid
& OBD_MD_FLFID
? oa
->o_parent_ver
: 0,
1461 POSTID(&oa
->o_oi
), pga
[0]->off
,
1462 pga
[page_count
-1]->off
+ pga
[page_count
-1]->count
- 1);
1463 CERROR("original client csum %x (type %x), server csum %x (type %x), "
1464 "client csum now %x\n", client_cksum
, client_cksum_type
,
1465 server_cksum
, cksum_type
, new_cksum
);
1469 /* Note rc enters this function as number of bytes transferred */
1470 static int osc_brw_fini_request(struct ptlrpc_request
*req
, int rc
)
1472 struct osc_brw_async_args
*aa
= (void *)&req
->rq_async_args
;
1473 const lnet_process_id_t
*peer
=
1474 &req
->rq_import
->imp_connection
->c_peer
;
1475 struct client_obd
*cli
= aa
->aa_cli
;
1476 struct ost_body
*body
;
1477 __u32 client_cksum
= 0;
1479 if (rc
< 0 && rc
!= -EDQUOT
) {
1480 DEBUG_REQ(D_INFO
, req
, "Failed request with rc = %d\n", rc
);
1484 LASSERTF(req
->rq_repmsg
!= NULL
, "rc = %d\n", rc
);
1485 body
= req_capsule_server_get(&req
->rq_pill
, &RMF_OST_BODY
);
1487 DEBUG_REQ(D_INFO
, req
, "Can't unpack body\n");
1491 /* set/clear over quota flag for a uid/gid */
1492 if (lustre_msg_get_opc(req
->rq_reqmsg
) == OST_WRITE
&&
1493 body
->oa
.o_valid
& (OBD_MD_FLUSRQUOTA
| OBD_MD_FLGRPQUOTA
)) {
1494 unsigned int qid
[MAXQUOTAS
] = { body
->oa
.o_uid
, body
->oa
.o_gid
};
1496 CDEBUG(D_QUOTA
, "setdq for [%u %u] with valid "LPX64
", flags %x\n",
1497 body
->oa
.o_uid
, body
->oa
.o_gid
, body
->oa
.o_valid
,
1499 osc_quota_setdq(cli
, qid
, body
->oa
.o_valid
, body
->oa
.o_flags
);
1502 osc_update_grant(cli
, body
);
1507 if (aa
->aa_oa
->o_valid
& OBD_MD_FLCKSUM
)
1508 client_cksum
= aa
->aa_oa
->o_cksum
; /* save for later */
1510 if (lustre_msg_get_opc(req
->rq_reqmsg
) == OST_WRITE
) {
1512 CERROR("Unexpected +ve rc %d\n", rc
);
1515 LASSERT(req
->rq_bulk
->bd_nob
== aa
->aa_requested_nob
);
1517 if (sptlrpc_cli_unwrap_bulk_write(req
, req
->rq_bulk
))
1520 if ((aa
->aa_oa
->o_valid
& OBD_MD_FLCKSUM
) && client_cksum
&&
1521 check_write_checksum(&body
->oa
, peer
, client_cksum
,
1522 body
->oa
.o_cksum
, aa
->aa_requested_nob
,
1523 aa
->aa_page_count
, aa
->aa_ppga
,
1524 cksum_type_unpack(aa
->aa_oa
->o_flags
)))
1527 rc
= check_write_rcs(req
, aa
->aa_requested_nob
,aa
->aa_nio_count
,
1528 aa
->aa_page_count
, aa
->aa_ppga
);
1532 /* The rest of this function executes only for OST_READs */
1534 /* if unwrap_bulk failed, return -EAGAIN to retry */
1535 rc
= sptlrpc_cli_unwrap_bulk_read(req
, req
->rq_bulk
, rc
);
1537 GOTO(out
, rc
= -EAGAIN
);
1539 if (rc
> aa
->aa_requested_nob
) {
1540 CERROR("Unexpected rc %d (%d requested)\n", rc
,
1541 aa
->aa_requested_nob
);
1545 if (rc
!= req
->rq_bulk
->bd_nob_transferred
) {
1546 CERROR ("Unexpected rc %d (%d transferred)\n",
1547 rc
, req
->rq_bulk
->bd_nob_transferred
);
1551 if (rc
< aa
->aa_requested_nob
)
1552 handle_short_read(rc
, aa
->aa_page_count
, aa
->aa_ppga
);
1554 if (body
->oa
.o_valid
& OBD_MD_FLCKSUM
) {
1555 static int cksum_counter
;
1556 __u32 server_cksum
= body
->oa
.o_cksum
;
1559 cksum_type_t cksum_type
;
1561 cksum_type
= cksum_type_unpack(body
->oa
.o_valid
&OBD_MD_FLFLAGS
?
1562 body
->oa
.o_flags
: 0);
1563 client_cksum
= osc_checksum_bulk(rc
, aa
->aa_page_count
,
1564 aa
->aa_ppga
, OST_READ
,
1567 if (peer
->nid
== req
->rq_bulk
->bd_sender
) {
1571 router
= libcfs_nid2str(req
->rq_bulk
->bd_sender
);
1574 if (server_cksum
== ~0 && rc
> 0) {
1575 CERROR("Protocol error: server %s set the 'checksum' "
1576 "bit, but didn't send a checksum. Not fatal, "
1577 "but please notify on http://bugs.whamcloud.com/\n",
1578 libcfs_nid2str(peer
->nid
));
1579 } else if (server_cksum
!= client_cksum
) {
1580 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1581 "%s%s%s inode "DFID
" object "DOSTID
1582 " extent ["LPU64
"-"LPU64
"]\n",
1583 req
->rq_import
->imp_obd
->obd_name
,
1584 libcfs_nid2str(peer
->nid
),
1586 body
->oa
.o_valid
& OBD_MD_FLFID
?
1587 body
->oa
.o_parent_seq
: (__u64
)0,
1588 body
->oa
.o_valid
& OBD_MD_FLFID
?
1589 body
->oa
.o_parent_oid
: 0,
1590 body
->oa
.o_valid
& OBD_MD_FLFID
?
1591 body
->oa
.o_parent_ver
: 0,
1592 POSTID(&body
->oa
.o_oi
),
1593 aa
->aa_ppga
[0]->off
,
1594 aa
->aa_ppga
[aa
->aa_page_count
-1]->off
+
1595 aa
->aa_ppga
[aa
->aa_page_count
-1]->count
-
1597 CERROR("client %x, server %x, cksum_type %x\n",
1598 client_cksum
, server_cksum
, cksum_type
);
1600 aa
->aa_oa
->o_cksum
= client_cksum
;
1604 CDEBUG(D_PAGE
, "checksum %x confirmed\n", client_cksum
);
1607 } else if (unlikely(client_cksum
)) {
1608 static int cksum_missed
;
1611 if ((cksum_missed
& (-cksum_missed
)) == cksum_missed
)
1612 CERROR("Checksum %u requested from %s but not sent\n",
1613 cksum_missed
, libcfs_nid2str(peer
->nid
));
1619 lustre_get_wire_obdo(&req
->rq_import
->imp_connect_data
,
1620 aa
->aa_oa
, &body
->oa
);
1625 static int osc_brw_internal(int cmd
, struct obd_export
*exp
, struct obdo
*oa
,
1626 struct lov_stripe_md
*lsm
,
1627 obd_count page_count
, struct brw_page
**pga
,
1628 struct obd_capa
*ocapa
)
1630 struct ptlrpc_request
*req
;
1632 wait_queue_head_t waitq
;
1633 int generation
, resends
= 0;
1634 struct l_wait_info lwi
;
1636 init_waitqueue_head(&waitq
);
1637 generation
= exp
->exp_obd
->u
.cli
.cl_import
->imp_generation
;
1640 rc
= osc_brw_prep_request(cmd
, &exp
->exp_obd
->u
.cli
, oa
, lsm
,
1641 page_count
, pga
, &req
, ocapa
, 0, resends
);
1646 req
->rq_generation_set
= 1;
1647 req
->rq_import_generation
= generation
;
1648 req
->rq_sent
= cfs_time_current_sec() + resends
;
1651 rc
= ptlrpc_queue_wait(req
);
1653 if (rc
== -ETIMEDOUT
&& req
->rq_resend
) {
1654 DEBUG_REQ(D_HA
, req
, "BULK TIMEOUT");
1655 ptlrpc_req_finished(req
);
1659 rc
= osc_brw_fini_request(req
, rc
);
1661 ptlrpc_req_finished(req
);
1662 /* When server return -EINPROGRESS, client should always retry
1663 * regardless of the number of times the bulk was resent already.*/
1664 if (osc_recoverable_error(rc
)) {
1666 if (rc
!= -EINPROGRESS
&&
1667 !client_should_resend(resends
, &exp
->exp_obd
->u
.cli
)) {
1668 CERROR("%s: too many resend retries for object: "
1669 ""DOSTID
", rc = %d.\n", exp
->exp_obd
->obd_name
,
1670 POSTID(&oa
->o_oi
), rc
);
1674 exp
->exp_obd
->u
.cli
.cl_import
->imp_generation
) {
1675 CDEBUG(D_HA
, "%s: resend cross eviction for object: "
1676 ""DOSTID
", rc = %d.\n", exp
->exp_obd
->obd_name
,
1677 POSTID(&oa
->o_oi
), rc
);
1681 lwi
= LWI_TIMEOUT_INTR(cfs_time_seconds(resends
), NULL
, NULL
,
1683 l_wait_event(waitq
, 0, &lwi
);
1688 if (rc
== -EAGAIN
|| rc
== -EINPROGRESS
)
1693 static int osc_brw_redo_request(struct ptlrpc_request
*request
,
1694 struct osc_brw_async_args
*aa
, int rc
)
1696 struct ptlrpc_request
*new_req
;
1697 struct osc_brw_async_args
*new_aa
;
1698 struct osc_async_page
*oap
;
1700 DEBUG_REQ(rc
== -EINPROGRESS
? D_RPCTRACE
: D_ERROR
, request
,
1701 "redo for recoverable error %d", rc
);
1703 rc
= osc_brw_prep_request(lustre_msg_get_opc(request
->rq_reqmsg
) ==
1704 OST_WRITE
? OBD_BRW_WRITE
:OBD_BRW_READ
,
1705 aa
->aa_cli
, aa
->aa_oa
,
1706 NULL
/* lsm unused by osc currently */,
1707 aa
->aa_page_count
, aa
->aa_ppga
,
1708 &new_req
, aa
->aa_ocapa
, 0, 1);
1712 list_for_each_entry(oap
, &aa
->aa_oaps
, oap_rpc_item
) {
1713 if (oap
->oap_request
!= NULL
) {
1714 LASSERTF(request
== oap
->oap_request
,
1715 "request %p != oap_request %p\n",
1716 request
, oap
->oap_request
);
1717 if (oap
->oap_interrupted
) {
1718 ptlrpc_req_finished(new_req
);
1723 /* New request takes over pga and oaps from old request.
1724 * Note that copying a list_head doesn't work, need to move it... */
1726 new_req
->rq_interpret_reply
= request
->rq_interpret_reply
;
1727 new_req
->rq_async_args
= request
->rq_async_args
;
1728 /* cap resend delay to the current request timeout, this is similar to
1729 * what ptlrpc does (see after_reply()) */
1730 if (aa
->aa_resends
> new_req
->rq_timeout
)
1731 new_req
->rq_sent
= cfs_time_current_sec() + new_req
->rq_timeout
;
1733 new_req
->rq_sent
= cfs_time_current_sec() + aa
->aa_resends
;
1734 new_req
->rq_generation_set
= 1;
1735 new_req
->rq_import_generation
= request
->rq_import_generation
;
1737 new_aa
= ptlrpc_req_async_args(new_req
);
1739 INIT_LIST_HEAD(&new_aa
->aa_oaps
);
1740 list_splice_init(&aa
->aa_oaps
, &new_aa
->aa_oaps
);
1741 INIT_LIST_HEAD(&new_aa
->aa_exts
);
1742 list_splice_init(&aa
->aa_exts
, &new_aa
->aa_exts
);
1743 new_aa
->aa_resends
= aa
->aa_resends
;
1745 list_for_each_entry(oap
, &new_aa
->aa_oaps
, oap_rpc_item
) {
1746 if (oap
->oap_request
) {
1747 ptlrpc_req_finished(oap
->oap_request
);
1748 oap
->oap_request
= ptlrpc_request_addref(new_req
);
1752 new_aa
->aa_ocapa
= aa
->aa_ocapa
;
1753 aa
->aa_ocapa
= NULL
;
1755 /* XXX: This code will run into problem if we're going to support
1756 * to add a series of BRW RPCs into a self-defined ptlrpc_request_set
1757 * and wait for all of them to be finished. We should inherit request
1758 * set from old request. */
1759 ptlrpcd_add_req(new_req
, PDL_POLICY_SAME
, -1);
1761 DEBUG_REQ(D_INFO
, new_req
, "new request");
1766 * ugh, we want disk allocation on the target to happen in offset order. we'll
1767 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1768 * fine for our small page arrays and doesn't require allocation. its an
1769 * insertion sort that swaps elements that are strides apart, shrinking the
1770 * stride down until its '1' and the array is sorted.
1772 static void sort_brw_pages(struct brw_page
**array
, int num
)
1775 struct brw_page
*tmp
;
1779 for (stride
= 1; stride
< num
; stride
= (stride
* 3) + 1)
1784 for (i
= stride
; i
< num
; i
++) {
1787 while (j
>= stride
&& array
[j
- stride
]->off
> tmp
->off
) {
1788 array
[j
] = array
[j
- stride
];
1793 } while (stride
> 1);
1796 static obd_count
max_unfragmented_pages(struct brw_page
**pg
, obd_count pages
)
1802 LASSERT (pages
> 0);
1803 offset
= pg
[i
]->off
& ~CFS_PAGE_MASK
;
1807 if (pages
== 0) /* that's all */
1810 if (offset
+ pg
[i
]->count
< PAGE_CACHE_SIZE
)
1811 return count
; /* doesn't end on page boundary */
1814 offset
= pg
[i
]->off
& ~CFS_PAGE_MASK
;
1815 if (offset
!= 0) /* doesn't start on page boundary */
1822 static struct brw_page
**osc_build_ppga(struct brw_page
*pga
, obd_count count
)
1824 struct brw_page
**ppga
;
1827 OBD_ALLOC(ppga
, sizeof(*ppga
) * count
);
1831 for (i
= 0; i
< count
; i
++)
1836 static void osc_release_ppga(struct brw_page
**ppga
, obd_count count
)
1838 LASSERT(ppga
!= NULL
);
1839 OBD_FREE(ppga
, sizeof(*ppga
) * count
);
1842 static int osc_brw(int cmd
, struct obd_export
*exp
, struct obd_info
*oinfo
,
1843 obd_count page_count
, struct brw_page
*pga
,
1844 struct obd_trans_info
*oti
)
1846 struct obdo
*saved_oa
= NULL
;
1847 struct brw_page
**ppga
, **orig
;
1848 struct obd_import
*imp
= class_exp2cliimp(exp
);
1849 struct client_obd
*cli
;
1850 int rc
, page_count_orig
;
1852 LASSERT((imp
!= NULL
) && (imp
->imp_obd
!= NULL
));
1853 cli
= &imp
->imp_obd
->u
.cli
;
1855 if (cmd
& OBD_BRW_CHECK
) {
1856 /* The caller just wants to know if there's a chance that this
1857 * I/O can succeed */
1859 if (imp
->imp_invalid
)
1864 /* test_brw with a failed create can trip this, maybe others. */
1865 LASSERT(cli
->cl_max_pages_per_rpc
);
1869 orig
= ppga
= osc_build_ppga(pga
, page_count
);
1872 page_count_orig
= page_count
;
1874 sort_brw_pages(ppga
, page_count
);
1875 while (page_count
) {
1876 obd_count pages_per_brw
;
1878 if (page_count
> cli
->cl_max_pages_per_rpc
)
1879 pages_per_brw
= cli
->cl_max_pages_per_rpc
;
1881 pages_per_brw
= page_count
;
1883 pages_per_brw
= max_unfragmented_pages(ppga
, pages_per_brw
);
1885 if (saved_oa
!= NULL
) {
1886 /* restore previously saved oa */
1887 *oinfo
->oi_oa
= *saved_oa
;
1888 } else if (page_count
> pages_per_brw
) {
1889 /* save a copy of oa (brw will clobber it) */
1890 OBDO_ALLOC(saved_oa
);
1891 if (saved_oa
== NULL
)
1892 GOTO(out
, rc
= -ENOMEM
);
1893 *saved_oa
= *oinfo
->oi_oa
;
1896 rc
= osc_brw_internal(cmd
, exp
, oinfo
->oi_oa
, oinfo
->oi_md
,
1897 pages_per_brw
, ppga
, oinfo
->oi_capa
);
1902 page_count
-= pages_per_brw
;
1903 ppga
+= pages_per_brw
;
1907 osc_release_ppga(orig
, page_count_orig
);
1909 if (saved_oa
!= NULL
)
1910 OBDO_FREE(saved_oa
);
1915 static int brw_interpret(const struct lu_env
*env
,
1916 struct ptlrpc_request
*req
, void *data
, int rc
)
1918 struct osc_brw_async_args
*aa
= data
;
1919 struct osc_extent
*ext
;
1920 struct osc_extent
*tmp
;
1921 struct cl_object
*obj
= NULL
;
1922 struct client_obd
*cli
= aa
->aa_cli
;
1924 rc
= osc_brw_fini_request(req
, rc
);
1925 CDEBUG(D_INODE
, "request %p aa %p rc %d\n", req
, aa
, rc
);
1926 /* When server return -EINPROGRESS, client should always retry
1927 * regardless of the number of times the bulk was resent already. */
1928 if (osc_recoverable_error(rc
)) {
1929 if (req
->rq_import_generation
!=
1930 req
->rq_import
->imp_generation
) {
1931 CDEBUG(D_HA
, "%s: resend cross eviction for object: "
1932 ""DOSTID
", rc = %d.\n",
1933 req
->rq_import
->imp_obd
->obd_name
,
1934 POSTID(&aa
->aa_oa
->o_oi
), rc
);
1935 } else if (rc
== -EINPROGRESS
||
1936 client_should_resend(aa
->aa_resends
, aa
->aa_cli
)) {
1937 rc
= osc_brw_redo_request(req
, aa
, rc
);
1939 CERROR("%s: too many resent retries for object: "
1940 ""LPU64
":"LPU64
", rc = %d.\n",
1941 req
->rq_import
->imp_obd
->obd_name
,
1942 POSTID(&aa
->aa_oa
->o_oi
), rc
);
1947 else if (rc
== -EAGAIN
|| rc
== -EINPROGRESS
)
1952 capa_put(aa
->aa_ocapa
);
1953 aa
->aa_ocapa
= NULL
;
1956 list_for_each_entry_safe(ext
, tmp
, &aa
->aa_exts
, oe_link
) {
1957 if (obj
== NULL
&& rc
== 0) {
1958 obj
= osc2cl(ext
->oe_obj
);
1962 list_del_init(&ext
->oe_link
);
1963 osc_extent_finish(env
, ext
, 1, rc
);
1965 LASSERT(list_empty(&aa
->aa_exts
));
1966 LASSERT(list_empty(&aa
->aa_oaps
));
1969 struct obdo
*oa
= aa
->aa_oa
;
1970 struct cl_attr
*attr
= &osc_env_info(env
)->oti_attr
;
1971 unsigned long valid
= 0;
1974 if (oa
->o_valid
& OBD_MD_FLBLOCKS
) {
1975 attr
->cat_blocks
= oa
->o_blocks
;
1976 valid
|= CAT_BLOCKS
;
1978 if (oa
->o_valid
& OBD_MD_FLMTIME
) {
1979 attr
->cat_mtime
= oa
->o_mtime
;
1982 if (oa
->o_valid
& OBD_MD_FLATIME
) {
1983 attr
->cat_atime
= oa
->o_atime
;
1986 if (oa
->o_valid
& OBD_MD_FLCTIME
) {
1987 attr
->cat_ctime
= oa
->o_ctime
;
1991 cl_object_attr_lock(obj
);
1992 cl_object_attr_set(env
, obj
, attr
, valid
);
1993 cl_object_attr_unlock(obj
);
1995 cl_object_put(env
, obj
);
1997 OBDO_FREE(aa
->aa_oa
);
1999 cl_req_completion(env
, aa
->aa_clerq
, rc
< 0 ? rc
:
2000 req
->rq_bulk
->bd_nob_transferred
);
2001 osc_release_ppga(aa
->aa_ppga
, aa
->aa_page_count
);
2002 ptlrpc_lprocfs_brw(req
, req
->rq_bulk
->bd_nob_transferred
);
2004 client_obd_list_lock(&cli
->cl_loi_list_lock
);
2005 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
2006 * is called so we know whether to go to sync BRWs or wait for more
2007 * RPCs to complete */
2008 if (lustre_msg_get_opc(req
->rq_reqmsg
) == OST_WRITE
)
2009 cli
->cl_w_in_flight
--;
2011 cli
->cl_r_in_flight
--;
2012 osc_wake_cache_waiters(cli
);
2013 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
2015 osc_io_unplug(env
, cli
, NULL
, PDL_POLICY_SAME
);
2020 * Build an RPC by the list of extent @ext_list. The caller must ensure
2021 * that the total pages in this list are NOT over max pages per RPC.
2022 * Extents in the list must be in OES_RPC state.
2024 int osc_build_rpc(const struct lu_env
*env
, struct client_obd
*cli
,
2025 struct list_head
*ext_list
, int cmd
, pdl_policy_t pol
)
2027 struct ptlrpc_request
*req
= NULL
;
2028 struct osc_extent
*ext
;
2029 struct brw_page
**pga
= NULL
;
2030 struct osc_brw_async_args
*aa
= NULL
;
2031 struct obdo
*oa
= NULL
;
2032 struct osc_async_page
*oap
;
2033 struct osc_async_page
*tmp
;
2034 struct cl_req
*clerq
= NULL
;
2035 enum cl_req_type crt
= (cmd
& OBD_BRW_WRITE
) ? CRT_WRITE
:
2037 struct ldlm_lock
*lock
= NULL
;
2038 struct cl_req_attr
*crattr
= NULL
;
2039 obd_off starting_offset
= OBD_OBJECT_EOF
;
2040 obd_off ending_offset
= 0;
2046 LIST_HEAD(rpc_list
);
2048 LASSERT(!list_empty(ext_list
));
2050 /* add pages into rpc_list to build BRW rpc */
2051 list_for_each_entry(ext
, ext_list
, oe_link
) {
2052 LASSERT(ext
->oe_state
== OES_RPC
);
2053 mem_tight
|= ext
->oe_memalloc
;
2054 list_for_each_entry(oap
, &ext
->oe_pages
, oap_pending_item
) {
2056 list_add_tail(&oap
->oap_rpc_item
, &rpc_list
);
2057 if (starting_offset
> oap
->oap_obj_off
)
2058 starting_offset
= oap
->oap_obj_off
;
2060 LASSERT(oap
->oap_page_off
== 0);
2061 if (ending_offset
< oap
->oap_obj_off
+ oap
->oap_count
)
2062 ending_offset
= oap
->oap_obj_off
+
2065 LASSERT(oap
->oap_page_off
+ oap
->oap_count
==
2071 mpflag
= cfs_memory_pressure_get_and_set();
2073 OBD_ALLOC(crattr
, sizeof(*crattr
));
2075 GOTO(out
, rc
= -ENOMEM
);
2077 OBD_ALLOC(pga
, sizeof(*pga
) * page_count
);
2079 GOTO(out
, rc
= -ENOMEM
);
2083 GOTO(out
, rc
= -ENOMEM
);
2086 list_for_each_entry(oap
, &rpc_list
, oap_rpc_item
) {
2087 struct cl_page
*page
= oap2cl_page(oap
);
2088 if (clerq
== NULL
) {
2089 clerq
= cl_req_alloc(env
, page
, crt
,
2090 1 /* only 1-object rpcs for now */);
2092 GOTO(out
, rc
= PTR_ERR(clerq
));
2093 lock
= oap
->oap_ldlm_lock
;
2096 oap
->oap_brw_flags
|= OBD_BRW_MEMALLOC
;
2097 pga
[i
] = &oap
->oap_brw_page
;
2098 pga
[i
]->off
= oap
->oap_obj_off
+ oap
->oap_page_off
;
2099 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
2100 pga
[i
]->pg
, page_index(oap
->oap_page
), oap
,
2103 cl_req_page_add(env
, clerq
, page
);
2106 /* always get the data for the obdo for the rpc */
2107 LASSERT(clerq
!= NULL
);
2108 crattr
->cra_oa
= oa
;
2109 cl_req_attr_set(env
, clerq
, crattr
, ~0ULL);
2111 oa
->o_handle
= lock
->l_remote_handle
;
2112 oa
->o_valid
|= OBD_MD_FLHANDLE
;
2115 rc
= cl_req_prep(env
, clerq
);
2117 CERROR("cl_req_prep failed: %d\n", rc
);
2121 sort_brw_pages(pga
, page_count
);
2122 rc
= osc_brw_prep_request(cmd
, cli
, oa
, NULL
, page_count
,
2123 pga
, &req
, crattr
->cra_capa
, 1, 0);
2125 CERROR("prep_req failed: %d\n", rc
);
2129 req
->rq_interpret_reply
= brw_interpret
;
2132 req
->rq_memalloc
= 1;
2134 /* Need to update the timestamps after the request is built in case
2135 * we race with setattr (locally or in queue at OST). If OST gets
2136 * later setattr before earlier BRW (as determined by the request xid),
2137 * the OST will not use BRW timestamps. Sadly, there is no obvious
2138 * way to do this in a single call. bug 10150 */
2139 cl_req_attr_set(env
, clerq
, crattr
,
2140 OBD_MD_FLMTIME
|OBD_MD_FLCTIME
|OBD_MD_FLATIME
);
2142 lustre_msg_set_jobid(req
->rq_reqmsg
, crattr
->cra_jobid
);
2144 CLASSERT(sizeof(*aa
) <= sizeof(req
->rq_async_args
));
2145 aa
= ptlrpc_req_async_args(req
);
2146 INIT_LIST_HEAD(&aa
->aa_oaps
);
2147 list_splice_init(&rpc_list
, &aa
->aa_oaps
);
2148 INIT_LIST_HEAD(&aa
->aa_exts
);
2149 list_splice_init(ext_list
, &aa
->aa_exts
);
2150 aa
->aa_clerq
= clerq
;
2152 /* queued sync pages can be torn down while the pages
2153 * were between the pending list and the rpc */
2155 list_for_each_entry(oap
, &aa
->aa_oaps
, oap_rpc_item
) {
2156 /* only one oap gets a request reference */
2159 if (oap
->oap_interrupted
&& !req
->rq_intr
) {
2160 CDEBUG(D_INODE
, "oap %p in req %p interrupted\n",
2162 ptlrpc_mark_interrupted(req
);
2166 tmp
->oap_request
= ptlrpc_request_addref(req
);
2168 client_obd_list_lock(&cli
->cl_loi_list_lock
);
2169 starting_offset
>>= PAGE_CACHE_SHIFT
;
2170 if (cmd
== OBD_BRW_READ
) {
2171 cli
->cl_r_in_flight
++;
2172 lprocfs_oh_tally_log2(&cli
->cl_read_page_hist
, page_count
);
2173 lprocfs_oh_tally(&cli
->cl_read_rpc_hist
, cli
->cl_r_in_flight
);
2174 lprocfs_oh_tally_log2(&cli
->cl_read_offset_hist
,
2175 starting_offset
+ 1);
2177 cli
->cl_w_in_flight
++;
2178 lprocfs_oh_tally_log2(&cli
->cl_write_page_hist
, page_count
);
2179 lprocfs_oh_tally(&cli
->cl_write_rpc_hist
, cli
->cl_w_in_flight
);
2180 lprocfs_oh_tally_log2(&cli
->cl_write_offset_hist
,
2181 starting_offset
+ 1);
2183 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
2185 DEBUG_REQ(D_INODE
, req
, "%d pages, aa %p. now %dr/%dw in flight",
2186 page_count
, aa
, cli
->cl_r_in_flight
,
2187 cli
->cl_w_in_flight
);
2189 /* XXX: Maybe the caller can check the RPC bulk descriptor to
2190 * see which CPU/NUMA node the majority of pages were allocated
2191 * on, and try to assign the async RPC to the CPU core
2192 * (PDL_POLICY_PREFERRED) to reduce cross-CPU memory traffic.
2194 * But on the other hand, we expect that multiple ptlrpcd
2195 * threads and the initial write sponsor can run in parallel,
2196 * especially when data checksum is enabled, which is CPU-bound
2197 * operation and single ptlrpcd thread cannot process in time.
2198 * So more ptlrpcd threads sharing BRW load
2199 * (with PDL_POLICY_ROUND) seems better.
2201 ptlrpcd_add_req(req
, pol
, -1);
2206 cfs_memory_pressure_restore(mpflag
);
2208 if (crattr
!= NULL
) {
2209 capa_put(crattr
->cra_capa
);
2210 OBD_FREE(crattr
, sizeof(*crattr
));
2214 LASSERT(req
== NULL
);
2219 OBD_FREE(pga
, sizeof(*pga
) * page_count
);
2220 /* this should happen rarely and is pretty bad, it makes the
2221 * pending list not follow the dirty order */
2222 while (!list_empty(ext_list
)) {
2223 ext
= list_entry(ext_list
->next
, struct osc_extent
,
2225 list_del_init(&ext
->oe_link
);
2226 osc_extent_finish(env
, ext
, 0, rc
);
2228 if (clerq
&& !IS_ERR(clerq
))
2229 cl_req_completion(env
, clerq
, rc
);
2234 static int osc_set_lock_data_with_check(struct ldlm_lock
*lock
,
2235 struct ldlm_enqueue_info
*einfo
)
2237 void *data
= einfo
->ei_cbdata
;
2240 LASSERT(lock
!= NULL
);
2241 LASSERT(lock
->l_blocking_ast
== einfo
->ei_cb_bl
);
2242 LASSERT(lock
->l_resource
->lr_type
== einfo
->ei_type
);
2243 LASSERT(lock
->l_completion_ast
== einfo
->ei_cb_cp
);
2244 LASSERT(lock
->l_glimpse_ast
== einfo
->ei_cb_gl
);
2246 lock_res_and_lock(lock
);
2247 spin_lock(&osc_ast_guard
);
2249 if (lock
->l_ast_data
== NULL
)
2250 lock
->l_ast_data
= data
;
2251 if (lock
->l_ast_data
== data
)
2254 spin_unlock(&osc_ast_guard
);
2255 unlock_res_and_lock(lock
);
2260 static int osc_set_data_with_check(struct lustre_handle
*lockh
,
2261 struct ldlm_enqueue_info
*einfo
)
2263 struct ldlm_lock
*lock
= ldlm_handle2lock(lockh
);
2267 set
= osc_set_lock_data_with_check(lock
, einfo
);
2268 LDLM_LOCK_PUT(lock
);
2270 CERROR("lockh %p, data %p - client evicted?\n",
2271 lockh
, einfo
->ei_cbdata
);
2275 static int osc_change_cbdata(struct obd_export
*exp
, struct lov_stripe_md
*lsm
,
2276 ldlm_iterator_t replace
, void *data
)
2278 struct ldlm_res_id res_id
;
2279 struct obd_device
*obd
= class_exp2obd(exp
);
2281 ostid_build_res_name(&lsm
->lsm_oi
, &res_id
);
2282 ldlm_resource_iterate(obd
->obd_namespace
, &res_id
, replace
, data
);
2286 /* find any ldlm lock of the inode in osc
2290 static int osc_find_cbdata(struct obd_export
*exp
, struct lov_stripe_md
*lsm
,
2291 ldlm_iterator_t replace
, void *data
)
2293 struct ldlm_res_id res_id
;
2294 struct obd_device
*obd
= class_exp2obd(exp
);
2297 ostid_build_res_name(&lsm
->lsm_oi
, &res_id
);
2298 rc
= ldlm_resource_iterate(obd
->obd_namespace
, &res_id
, replace
, data
);
2299 if (rc
== LDLM_ITER_STOP
)
2301 if (rc
== LDLM_ITER_CONTINUE
)
2306 static int osc_enqueue_fini(struct ptlrpc_request
*req
, struct ost_lvb
*lvb
,
2307 obd_enqueue_update_f upcall
, void *cookie
,
2308 __u64
*flags
, int agl
, int rc
)
2310 int intent
= *flags
& LDLM_FL_HAS_INTENT
;
2313 /* The request was created before ldlm_cli_enqueue call. */
2314 if (rc
== ELDLM_LOCK_ABORTED
) {
2315 struct ldlm_reply
*rep
;
2316 rep
= req_capsule_server_get(&req
->rq_pill
,
2319 LASSERT(rep
!= NULL
);
2320 rep
->lock_policy_res1
=
2321 ptlrpc_status_ntoh(rep
->lock_policy_res1
);
2322 if (rep
->lock_policy_res1
)
2323 rc
= rep
->lock_policy_res1
;
2327 if ((intent
!= 0 && rc
== ELDLM_LOCK_ABORTED
&& agl
== 0) ||
2329 *flags
|= LDLM_FL_LVB_READY
;
2330 CDEBUG(D_INODE
,"got kms "LPU64
" blocks "LPU64
" mtime "LPU64
"\n",
2331 lvb
->lvb_size
, lvb
->lvb_blocks
, lvb
->lvb_mtime
);
2334 /* Call the update callback. */
2335 rc
= (*upcall
)(cookie
, rc
);
2339 static int osc_enqueue_interpret(const struct lu_env
*env
,
2340 struct ptlrpc_request
*req
,
2341 struct osc_enqueue_args
*aa
, int rc
)
2343 struct ldlm_lock
*lock
;
2344 struct lustre_handle handle
;
2346 struct ost_lvb
*lvb
;
2348 __u64
*flags
= aa
->oa_flags
;
2350 /* Make a local copy of a lock handle and a mode, because aa->oa_*
2351 * might be freed anytime after lock upcall has been called. */
2352 lustre_handle_copy(&handle
, aa
->oa_lockh
);
2353 mode
= aa
->oa_ei
->ei_mode
;
2355 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2357 lock
= ldlm_handle2lock(&handle
);
2359 /* Take an additional reference so that a blocking AST that
2360 * ldlm_cli_enqueue_fini() might post for a failed lock, is guaranteed
2361 * to arrive after an upcall has been executed by
2362 * osc_enqueue_fini(). */
2363 ldlm_lock_addref(&handle
, mode
);
2365 /* Let CP AST to grant the lock first. */
2366 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_ENQ_RACE
, 1);
2368 if (aa
->oa_agl
&& rc
== ELDLM_LOCK_ABORTED
) {
2373 lvb_len
= sizeof(*aa
->oa_lvb
);
2376 /* Complete obtaining the lock procedure. */
2377 rc
= ldlm_cli_enqueue_fini(aa
->oa_exp
, req
, aa
->oa_ei
->ei_type
, 1,
2378 mode
, flags
, lvb
, lvb_len
, &handle
, rc
);
2379 /* Complete osc stuff. */
2380 rc
= osc_enqueue_fini(req
, aa
->oa_lvb
, aa
->oa_upcall
, aa
->oa_cookie
,
2381 flags
, aa
->oa_agl
, rc
);
2383 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_CP_CANCEL_RACE
, 10);
2385 /* Release the lock for async request. */
2386 if (lustre_handle_is_used(&handle
) && rc
== ELDLM_OK
)
2388 * Releases a reference taken by ldlm_cli_enqueue(), if it is
2389 * not already released by
2390 * ldlm_cli_enqueue_fini()->failed_lock_cleanup()
2392 ldlm_lock_decref(&handle
, mode
);
2394 LASSERTF(lock
!= NULL
, "lockh %p, req %p, aa %p - client evicted?\n",
2395 aa
->oa_lockh
, req
, aa
);
2396 ldlm_lock_decref(&handle
, mode
);
2397 LDLM_LOCK_PUT(lock
);
2401 void osc_update_enqueue(struct lustre_handle
*lov_lockhp
,
2402 struct lov_oinfo
*loi
, int flags
,
2403 struct ost_lvb
*lvb
, __u32 mode
, int rc
)
2405 struct ldlm_lock
*lock
= ldlm_handle2lock(lov_lockhp
);
2407 if (rc
== ELDLM_OK
) {
2410 LASSERT(lock
!= NULL
);
2411 loi
->loi_lvb
= *lvb
;
2412 tmp
= loi
->loi_lvb
.lvb_size
;
2413 /* Extend KMS up to the end of this lock and no further
2414 * A lock on [x,y] means a KMS of up to y + 1 bytes! */
2415 if (tmp
> lock
->l_policy_data
.l_extent
.end
)
2416 tmp
= lock
->l_policy_data
.l_extent
.end
+ 1;
2417 if (tmp
>= loi
->loi_kms
) {
2418 LDLM_DEBUG(lock
, "lock acquired, setting rss="LPU64
2419 ", kms="LPU64
, loi
->loi_lvb
.lvb_size
, tmp
);
2420 loi_kms_set(loi
, tmp
);
2422 LDLM_DEBUG(lock
, "lock acquired, setting rss="
2423 LPU64
"; leaving kms="LPU64
", end="LPU64
,
2424 loi
->loi_lvb
.lvb_size
, loi
->loi_kms
,
2425 lock
->l_policy_data
.l_extent
.end
);
2427 ldlm_lock_allow_match(lock
);
2428 } else if (rc
== ELDLM_LOCK_ABORTED
&& (flags
& LDLM_FL_HAS_INTENT
)) {
2429 LASSERT(lock
!= NULL
);
2430 loi
->loi_lvb
= *lvb
;
2431 ldlm_lock_allow_match(lock
);
2432 CDEBUG(D_INODE
, "glimpsed, setting rss="LPU64
"; leaving"
2433 " kms="LPU64
"\n", loi
->loi_lvb
.lvb_size
, loi
->loi_kms
);
2439 ldlm_lock_fail_match(lock
);
2441 LDLM_LOCK_PUT(lock
);
2444 EXPORT_SYMBOL(osc_update_enqueue
);
2446 struct ptlrpc_request_set
*PTLRPCD_SET
= (void *)1;
2448 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2449 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2450 * other synchronous requests, however keeping some locks and trying to obtain
2451 * others may take a considerable amount of time in a case of ost failure; and
2452 * when other sync requests do not get released lock from a client, the client
2453 * is excluded from the cluster -- such scenarious make the life difficult, so
2454 * release locks just after they are obtained. */
2455 int osc_enqueue_base(struct obd_export
*exp
, struct ldlm_res_id
*res_id
,
2456 __u64
*flags
, ldlm_policy_data_t
*policy
,
2457 struct ost_lvb
*lvb
, int kms_valid
,
2458 obd_enqueue_update_f upcall
, void *cookie
,
2459 struct ldlm_enqueue_info
*einfo
,
2460 struct lustre_handle
*lockh
,
2461 struct ptlrpc_request_set
*rqset
, int async
, int agl
)
2463 struct obd_device
*obd
= exp
->exp_obd
;
2464 struct ptlrpc_request
*req
= NULL
;
2465 int intent
= *flags
& LDLM_FL_HAS_INTENT
;
2466 int match_lvb
= (agl
!= 0 ? 0 : LDLM_FL_LVB_READY
);
2470 /* Filesystem lock extents are extended to page boundaries so that
2471 * dealing with the page cache is a little smoother. */
2472 policy
->l_extent
.start
-= policy
->l_extent
.start
& ~CFS_PAGE_MASK
;
2473 policy
->l_extent
.end
|= ~CFS_PAGE_MASK
;
2476 * kms is not valid when either object is completely fresh (so that no
2477 * locks are cached), or object was evicted. In the latter case cached
2478 * lock cannot be used, because it would prime inode state with
2479 * potentially stale LVB.
2484 /* Next, search for already existing extent locks that will cover us */
2485 /* If we're trying to read, we also search for an existing PW lock. The
2486 * VFS and page cache already protect us locally, so lots of readers/
2487 * writers can share a single PW lock.
2489 * There are problems with conversion deadlocks, so instead of
2490 * converting a read lock to a write lock, we'll just enqueue a new
2493 * At some point we should cancel the read lock instead of making them
2494 * send us a blocking callback, but there are problems with canceling
2495 * locks out from other users right now, too. */
2496 mode
= einfo
->ei_mode
;
2497 if (einfo
->ei_mode
== LCK_PR
)
2499 mode
= ldlm_lock_match(obd
->obd_namespace
, *flags
| match_lvb
, res_id
,
2500 einfo
->ei_type
, policy
, mode
, lockh
, 0);
2502 struct ldlm_lock
*matched
= ldlm_handle2lock(lockh
);
2504 if ((agl
!= 0) && !(matched
->l_flags
& LDLM_FL_LVB_READY
)) {
2505 /* For AGL, if enqueue RPC is sent but the lock is not
2506 * granted, then skip to process this strpe.
2507 * Return -ECANCELED to tell the caller. */
2508 ldlm_lock_decref(lockh
, mode
);
2509 LDLM_LOCK_PUT(matched
);
2511 } else if (osc_set_lock_data_with_check(matched
, einfo
)) {
2512 *flags
|= LDLM_FL_LVB_READY
;
2513 /* addref the lock only if not async requests and PW
2514 * lock is matched whereas we asked for PR. */
2515 if (!rqset
&& einfo
->ei_mode
!= mode
)
2516 ldlm_lock_addref(lockh
, LCK_PR
);
2518 /* I would like to be able to ASSERT here that
2519 * rss <= kms, but I can't, for reasons which
2520 * are explained in lov_enqueue() */
2523 /* We already have a lock, and it's referenced.
2525 * At this point, the cl_lock::cll_state is CLS_QUEUING,
2526 * AGL upcall may change it to CLS_HELD directly. */
2527 (*upcall
)(cookie
, ELDLM_OK
);
2529 if (einfo
->ei_mode
!= mode
)
2530 ldlm_lock_decref(lockh
, LCK_PW
);
2532 /* For async requests, decref the lock. */
2533 ldlm_lock_decref(lockh
, einfo
->ei_mode
);
2534 LDLM_LOCK_PUT(matched
);
2537 ldlm_lock_decref(lockh
, mode
);
2538 LDLM_LOCK_PUT(matched
);
2545 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
),
2546 &RQF_LDLM_ENQUEUE_LVB
);
2550 rc
= ldlm_prep_enqueue_req(exp
, req
, &cancels
, 0);
2552 ptlrpc_request_free(req
);
2556 req_capsule_set_size(&req
->rq_pill
, &RMF_DLM_LVB
, RCL_SERVER
,
2558 ptlrpc_request_set_replen(req
);
2561 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2562 *flags
&= ~LDLM_FL_BLOCK_GRANTED
;
2564 rc
= ldlm_cli_enqueue(exp
, &req
, einfo
, res_id
, policy
, flags
, lvb
,
2565 sizeof(*lvb
), LVB_T_OST
, lockh
, async
);
2568 struct osc_enqueue_args
*aa
;
2569 CLASSERT (sizeof(*aa
) <= sizeof(req
->rq_async_args
));
2570 aa
= ptlrpc_req_async_args(req
);
2573 aa
->oa_flags
= flags
;
2574 aa
->oa_upcall
= upcall
;
2575 aa
->oa_cookie
= cookie
;
2577 aa
->oa_lockh
= lockh
;
2580 req
->rq_interpret_reply
=
2581 (ptlrpc_interpterer_t
)osc_enqueue_interpret
;
2582 if (rqset
== PTLRPCD_SET
)
2583 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
2585 ptlrpc_set_add_req(rqset
, req
);
2586 } else if (intent
) {
2587 ptlrpc_req_finished(req
);
2592 rc
= osc_enqueue_fini(req
, lvb
, upcall
, cookie
, flags
, agl
, rc
);
2594 ptlrpc_req_finished(req
);
2599 static int osc_enqueue(struct obd_export
*exp
, struct obd_info
*oinfo
,
2600 struct ldlm_enqueue_info
*einfo
,
2601 struct ptlrpc_request_set
*rqset
)
2603 struct ldlm_res_id res_id
;
2606 ostid_build_res_name(&oinfo
->oi_md
->lsm_oi
, &res_id
);
2607 rc
= osc_enqueue_base(exp
, &res_id
, &oinfo
->oi_flags
, &oinfo
->oi_policy
,
2608 &oinfo
->oi_md
->lsm_oinfo
[0]->loi_lvb
,
2609 oinfo
->oi_md
->lsm_oinfo
[0]->loi_kms_valid
,
2610 oinfo
->oi_cb_up
, oinfo
, einfo
, oinfo
->oi_lockh
,
2611 rqset
, rqset
!= NULL
, 0);
2615 int osc_match_base(struct obd_export
*exp
, struct ldlm_res_id
*res_id
,
2616 __u32 type
, ldlm_policy_data_t
*policy
, __u32 mode
,
2617 int *flags
, void *data
, struct lustre_handle
*lockh
,
2620 struct obd_device
*obd
= exp
->exp_obd
;
2621 int lflags
= *flags
;
2624 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_MATCH
))
2627 /* Filesystem lock extents are extended to page boundaries so that
2628 * dealing with the page cache is a little smoother */
2629 policy
->l_extent
.start
-= policy
->l_extent
.start
& ~CFS_PAGE_MASK
;
2630 policy
->l_extent
.end
|= ~CFS_PAGE_MASK
;
2632 /* Next, search for already existing extent locks that will cover us */
2633 /* If we're trying to read, we also search for an existing PW lock. The
2634 * VFS and page cache already protect us locally, so lots of readers/
2635 * writers can share a single PW lock. */
2639 rc
= ldlm_lock_match(obd
->obd_namespace
, lflags
,
2640 res_id
, type
, policy
, rc
, lockh
, unref
);
2643 if (!osc_set_data_with_check(lockh
, data
)) {
2644 if (!(lflags
& LDLM_FL_TEST_LOCK
))
2645 ldlm_lock_decref(lockh
, rc
);
2649 if (!(lflags
& LDLM_FL_TEST_LOCK
) && mode
!= rc
) {
2650 ldlm_lock_addref(lockh
, LCK_PR
);
2651 ldlm_lock_decref(lockh
, LCK_PW
);
2658 int osc_cancel_base(struct lustre_handle
*lockh
, __u32 mode
)
2660 if (unlikely(mode
== LCK_GROUP
))
2661 ldlm_lock_decref_and_cancel(lockh
, mode
);
2663 ldlm_lock_decref(lockh
, mode
);
2668 static int osc_cancel(struct obd_export
*exp
, struct lov_stripe_md
*md
,
2669 __u32 mode
, struct lustre_handle
*lockh
)
2671 return osc_cancel_base(lockh
, mode
);
2674 static int osc_cancel_unused(struct obd_export
*exp
,
2675 struct lov_stripe_md
*lsm
,
2676 ldlm_cancel_flags_t flags
,
2679 struct obd_device
*obd
= class_exp2obd(exp
);
2680 struct ldlm_res_id res_id
, *resp
= NULL
;
2683 ostid_build_res_name(&lsm
->lsm_oi
, &res_id
);
2687 return ldlm_cli_cancel_unused(obd
->obd_namespace
, resp
, flags
, opaque
);
2690 static int osc_statfs_interpret(const struct lu_env
*env
,
2691 struct ptlrpc_request
*req
,
2692 struct osc_async_args
*aa
, int rc
)
2694 struct obd_statfs
*msfs
;
2697 /* The request has in fact never been sent
2698 * due to issues at a higher level (LOV).
2699 * Exit immediately since the caller is
2700 * aware of the problem and takes care
2701 * of the clean up */
2704 if ((rc
== -ENOTCONN
|| rc
== -EAGAIN
) &&
2705 (aa
->aa_oi
->oi_flags
& OBD_STATFS_NODELAY
))
2711 msfs
= req_capsule_server_get(&req
->rq_pill
, &RMF_OBD_STATFS
);
2713 GOTO(out
, rc
= -EPROTO
);
2716 *aa
->aa_oi
->oi_osfs
= *msfs
;
2718 rc
= aa
->aa_oi
->oi_cb_up(aa
->aa_oi
, rc
);
2722 static int osc_statfs_async(struct obd_export
*exp
,
2723 struct obd_info
*oinfo
, __u64 max_age
,
2724 struct ptlrpc_request_set
*rqset
)
2726 struct obd_device
*obd
= class_exp2obd(exp
);
2727 struct ptlrpc_request
*req
;
2728 struct osc_async_args
*aa
;
2731 /* We could possibly pass max_age in the request (as an absolute
2732 * timestamp or a "seconds.usec ago") so the target can avoid doing
2733 * extra calls into the filesystem if that isn't necessary (e.g.
2734 * during mount that would help a bit). Having relative timestamps
2735 * is not so great if request processing is slow, while absolute
2736 * timestamps are not ideal because they need time synchronization. */
2737 req
= ptlrpc_request_alloc(obd
->u
.cli
.cl_import
, &RQF_OST_STATFS
);
2741 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_STATFS
);
2743 ptlrpc_request_free(req
);
2746 ptlrpc_request_set_replen(req
);
2747 req
->rq_request_portal
= OST_CREATE_PORTAL
;
2748 ptlrpc_at_set_req_timeout(req
);
2750 if (oinfo
->oi_flags
& OBD_STATFS_NODELAY
) {
2751 /* procfs requests not want stat in wait for avoid deadlock */
2752 req
->rq_no_resend
= 1;
2753 req
->rq_no_delay
= 1;
2756 req
->rq_interpret_reply
= (ptlrpc_interpterer_t
)osc_statfs_interpret
;
2757 CLASSERT (sizeof(*aa
) <= sizeof(req
->rq_async_args
));
2758 aa
= ptlrpc_req_async_args(req
);
2761 ptlrpc_set_add_req(rqset
, req
);
2765 static int osc_statfs(const struct lu_env
*env
, struct obd_export
*exp
,
2766 struct obd_statfs
*osfs
, __u64 max_age
, __u32 flags
)
2768 struct obd_device
*obd
= class_exp2obd(exp
);
2769 struct obd_statfs
*msfs
;
2770 struct ptlrpc_request
*req
;
2771 struct obd_import
*imp
= NULL
;
2774 /*Since the request might also come from lprocfs, so we need
2775 *sync this with client_disconnect_export Bug15684*/
2776 down_read(&obd
->u
.cli
.cl_sem
);
2777 if (obd
->u
.cli
.cl_import
)
2778 imp
= class_import_get(obd
->u
.cli
.cl_import
);
2779 up_read(&obd
->u
.cli
.cl_sem
);
2783 /* We could possibly pass max_age in the request (as an absolute
2784 * timestamp or a "seconds.usec ago") so the target can avoid doing
2785 * extra calls into the filesystem if that isn't necessary (e.g.
2786 * during mount that would help a bit). Having relative timestamps
2787 * is not so great if request processing is slow, while absolute
2788 * timestamps are not ideal because they need time synchronization. */
2789 req
= ptlrpc_request_alloc(imp
, &RQF_OST_STATFS
);
2791 class_import_put(imp
);
2796 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_STATFS
);
2798 ptlrpc_request_free(req
);
2801 ptlrpc_request_set_replen(req
);
2802 req
->rq_request_portal
= OST_CREATE_PORTAL
;
2803 ptlrpc_at_set_req_timeout(req
);
2805 if (flags
& OBD_STATFS_NODELAY
) {
2806 /* procfs requests not want stat in wait for avoid deadlock */
2807 req
->rq_no_resend
= 1;
2808 req
->rq_no_delay
= 1;
2811 rc
= ptlrpc_queue_wait(req
);
2815 msfs
= req_capsule_server_get(&req
->rq_pill
, &RMF_OBD_STATFS
);
2817 GOTO(out
, rc
= -EPROTO
);
2823 ptlrpc_req_finished(req
);
2827 /* Retrieve object striping information.
2829 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
2830 * the maximum number of OST indices which will fit in the user buffer.
2831 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
2833 static int osc_getstripe(struct lov_stripe_md
*lsm
, struct lov_user_md
*lump
)
2835 /* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
2836 struct lov_user_md_v3 lum
, *lumk
;
2837 struct lov_user_ost_data_v1
*lmm_objects
;
2838 int rc
= 0, lum_size
;
2843 /* we only need the header part from user space to get lmm_magic and
2844 * lmm_stripe_count, (the header part is common to v1 and v3) */
2845 lum_size
= sizeof(struct lov_user_md_v1
);
2846 if (copy_from_user(&lum
, lump
, lum_size
))
2849 if ((lum
.lmm_magic
!= LOV_USER_MAGIC_V1
) &&
2850 (lum
.lmm_magic
!= LOV_USER_MAGIC_V3
))
2853 /* lov_user_md_vX and lov_mds_md_vX must have the same size */
2854 LASSERT(sizeof(struct lov_user_md_v1
) == sizeof(struct lov_mds_md_v1
));
2855 LASSERT(sizeof(struct lov_user_md_v3
) == sizeof(struct lov_mds_md_v3
));
2856 LASSERT(sizeof(lum
.lmm_objects
[0]) == sizeof(lumk
->lmm_objects
[0]));
2858 /* we can use lov_mds_md_size() to compute lum_size
2859 * because lov_user_md_vX and lov_mds_md_vX have the same size */
2860 if (lum
.lmm_stripe_count
> 0) {
2861 lum_size
= lov_mds_md_size(lum
.lmm_stripe_count
, lum
.lmm_magic
);
2862 OBD_ALLOC(lumk
, lum_size
);
2866 if (lum
.lmm_magic
== LOV_USER_MAGIC_V1
)
2868 &(((struct lov_user_md_v1
*)lumk
)->lmm_objects
[0]);
2870 lmm_objects
= &(lumk
->lmm_objects
[0]);
2871 lmm_objects
->l_ost_oi
= lsm
->lsm_oi
;
2873 lum_size
= lov_mds_md_size(0, lum
.lmm_magic
);
2877 lumk
->lmm_oi
= lsm
->lsm_oi
;
2878 lumk
->lmm_stripe_count
= 1;
2880 if (copy_to_user(lump
, lumk
, lum_size
))
2884 OBD_FREE(lumk
, lum_size
);
2890 static int osc_iocontrol(unsigned int cmd
, struct obd_export
*exp
, int len
,
2891 void *karg
, void *uarg
)
2893 struct obd_device
*obd
= exp
->exp_obd
;
2894 struct obd_ioctl_data
*data
= karg
;
2897 if (!try_module_get(THIS_MODULE
)) {
2898 CERROR("Can't get module. Is it alive?");
2902 case OBD_IOC_LOV_GET_CONFIG
: {
2904 struct lov_desc
*desc
;
2905 struct obd_uuid uuid
;
2909 if (obd_ioctl_getdata(&buf
, &len
, (void *)uarg
))
2910 GOTO(out
, err
= -EINVAL
);
2912 data
= (struct obd_ioctl_data
*)buf
;
2914 if (sizeof(*desc
) > data
->ioc_inllen1
) {
2915 obd_ioctl_freedata(buf
, len
);
2916 GOTO(out
, err
= -EINVAL
);
2919 if (data
->ioc_inllen2
< sizeof(uuid
)) {
2920 obd_ioctl_freedata(buf
, len
);
2921 GOTO(out
, err
= -EINVAL
);
2924 desc
= (struct lov_desc
*)data
->ioc_inlbuf1
;
2925 desc
->ld_tgt_count
= 1;
2926 desc
->ld_active_tgt_count
= 1;
2927 desc
->ld_default_stripe_count
= 1;
2928 desc
->ld_default_stripe_size
= 0;
2929 desc
->ld_default_stripe_offset
= 0;
2930 desc
->ld_pattern
= 0;
2931 memcpy(&desc
->ld_uuid
, &obd
->obd_uuid
, sizeof(uuid
));
2933 memcpy(data
->ioc_inlbuf2
, &obd
->obd_uuid
, sizeof(uuid
));
2935 err
= copy_to_user((void *)uarg
, buf
, len
);
2938 obd_ioctl_freedata(buf
, len
);
2941 case LL_IOC_LOV_SETSTRIPE
:
2942 err
= obd_alloc_memmd(exp
, karg
);
2946 case LL_IOC_LOV_GETSTRIPE
:
2947 err
= osc_getstripe(karg
, uarg
);
2949 case OBD_IOC_CLIENT_RECOVER
:
2950 err
= ptlrpc_recover_import(obd
->u
.cli
.cl_import
,
2951 data
->ioc_inlbuf1
, 0);
2955 case IOC_OSC_SET_ACTIVE
:
2956 err
= ptlrpc_set_import_active(obd
->u
.cli
.cl_import
,
2959 case OBD_IOC_POLL_QUOTACHECK
:
2960 err
= osc_quota_poll_check(exp
, (struct if_quotacheck
*)karg
);
2962 case OBD_IOC_PING_TARGET
:
2963 err
= ptlrpc_obd_ping(obd
);
2966 CDEBUG(D_INODE
, "unrecognised ioctl %#x by %s\n",
2967 cmd
, current_comm());
2968 GOTO(out
, err
= -ENOTTY
);
2971 module_put(THIS_MODULE
);
2975 static int osc_get_info(const struct lu_env
*env
, struct obd_export
*exp
,
2976 obd_count keylen
, void *key
, __u32
*vallen
, void *val
,
2977 struct lov_stripe_md
*lsm
)
2979 if (!vallen
|| !val
)
2982 if (KEY_IS(KEY_LOCK_TO_STRIPE
)) {
2983 __u32
*stripe
= val
;
2984 *vallen
= sizeof(*stripe
);
2987 } else if (KEY_IS(KEY_LAST_ID
)) {
2988 struct ptlrpc_request
*req
;
2993 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
),
2994 &RQF_OST_GET_INFO_LAST_ID
);
2998 req_capsule_set_size(&req
->rq_pill
, &RMF_SETINFO_KEY
,
2999 RCL_CLIENT
, keylen
);
3000 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_GET_INFO
);
3002 ptlrpc_request_free(req
);
3006 tmp
= req_capsule_client_get(&req
->rq_pill
, &RMF_SETINFO_KEY
);
3007 memcpy(tmp
, key
, keylen
);
3009 req
->rq_no_delay
= req
->rq_no_resend
= 1;
3010 ptlrpc_request_set_replen(req
);
3011 rc
= ptlrpc_queue_wait(req
);
3015 reply
= req_capsule_server_get(&req
->rq_pill
, &RMF_OBD_ID
);
3017 GOTO(out
, rc
= -EPROTO
);
3019 *((obd_id
*)val
) = *reply
;
3021 ptlrpc_req_finished(req
);
3023 } else if (KEY_IS(KEY_FIEMAP
)) {
3024 struct ll_fiemap_info_key
*fm_key
=
3025 (struct ll_fiemap_info_key
*)key
;
3026 struct ldlm_res_id res_id
;
3027 ldlm_policy_data_t policy
;
3028 struct lustre_handle lockh
;
3029 ldlm_mode_t mode
= 0;
3030 struct ptlrpc_request
*req
;
3031 struct ll_user_fiemap
*reply
;
3035 if (!(fm_key
->fiemap
.fm_flags
& FIEMAP_FLAG_SYNC
))
3038 policy
.l_extent
.start
= fm_key
->fiemap
.fm_start
&
3041 if (OBD_OBJECT_EOF
- fm_key
->fiemap
.fm_length
<=
3042 fm_key
->fiemap
.fm_start
+ PAGE_CACHE_SIZE
- 1)
3043 policy
.l_extent
.end
= OBD_OBJECT_EOF
;
3045 policy
.l_extent
.end
= (fm_key
->fiemap
.fm_start
+
3046 fm_key
->fiemap
.fm_length
+
3047 PAGE_CACHE_SIZE
- 1) & CFS_PAGE_MASK
;
3049 ostid_build_res_name(&fm_key
->oa
.o_oi
, &res_id
);
3050 mode
= ldlm_lock_match(exp
->exp_obd
->obd_namespace
,
3051 LDLM_FL_BLOCK_GRANTED
|
3053 &res_id
, LDLM_EXTENT
, &policy
,
3054 LCK_PR
| LCK_PW
, &lockh
, 0);
3055 if (mode
) { /* lock is cached on client */
3056 if (mode
!= LCK_PR
) {
3057 ldlm_lock_addref(&lockh
, LCK_PR
);
3058 ldlm_lock_decref(&lockh
, LCK_PW
);
3060 } else { /* no cached lock, needs acquire lock on server side */
3061 fm_key
->oa
.o_valid
|= OBD_MD_FLFLAGS
;
3062 fm_key
->oa
.o_flags
|= OBD_FL_SRVLOCK
;
3066 req
= ptlrpc_request_alloc(class_exp2cliimp(exp
),
3067 &RQF_OST_GET_INFO_FIEMAP
);
3069 GOTO(drop_lock
, rc
= -ENOMEM
);
3071 req_capsule_set_size(&req
->rq_pill
, &RMF_FIEMAP_KEY
,
3072 RCL_CLIENT
, keylen
);
3073 req_capsule_set_size(&req
->rq_pill
, &RMF_FIEMAP_VAL
,
3074 RCL_CLIENT
, *vallen
);
3075 req_capsule_set_size(&req
->rq_pill
, &RMF_FIEMAP_VAL
,
3076 RCL_SERVER
, *vallen
);
3078 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_GET_INFO
);
3080 ptlrpc_request_free(req
);
3081 GOTO(drop_lock
, rc
);
3084 tmp
= req_capsule_client_get(&req
->rq_pill
, &RMF_FIEMAP_KEY
);
3085 memcpy(tmp
, key
, keylen
);
3086 tmp
= req_capsule_client_get(&req
->rq_pill
, &RMF_FIEMAP_VAL
);
3087 memcpy(tmp
, val
, *vallen
);
3089 ptlrpc_request_set_replen(req
);
3090 rc
= ptlrpc_queue_wait(req
);
3094 reply
= req_capsule_server_get(&req
->rq_pill
, &RMF_FIEMAP_VAL
);
3096 GOTO(fini_req
, rc
= -EPROTO
);
3098 memcpy(val
, reply
, *vallen
);
3100 ptlrpc_req_finished(req
);
3103 ldlm_lock_decref(&lockh
, LCK_PR
);
3110 static int osc_set_info_async(const struct lu_env
*env
, struct obd_export
*exp
,
3111 obd_count keylen
, void *key
, obd_count vallen
,
3112 void *val
, struct ptlrpc_request_set
*set
)
3114 struct ptlrpc_request
*req
;
3115 struct obd_device
*obd
= exp
->exp_obd
;
3116 struct obd_import
*imp
= class_exp2cliimp(exp
);
3120 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN
, 10);
3122 if (KEY_IS(KEY_CHECKSUM
)) {
3123 if (vallen
!= sizeof(int))
3125 exp
->exp_obd
->u
.cli
.cl_checksum
= (*(int *)val
) ? 1 : 0;
3129 if (KEY_IS(KEY_SPTLRPC_CONF
)) {
3130 sptlrpc_conf_client_adapt(obd
);
3134 if (KEY_IS(KEY_FLUSH_CTX
)) {
3135 sptlrpc_import_flush_my_ctx(imp
);
3139 if (KEY_IS(KEY_CACHE_SET
)) {
3140 struct client_obd
*cli
= &obd
->u
.cli
;
3142 LASSERT(cli
->cl_cache
== NULL
); /* only once */
3143 cli
->cl_cache
= (struct cl_client_cache
*)val
;
3144 atomic_inc(&cli
->cl_cache
->ccc_users
);
3145 cli
->cl_lru_left
= &cli
->cl_cache
->ccc_lru_left
;
3147 /* add this osc into entity list */
3148 LASSERT(list_empty(&cli
->cl_lru_osc
));
3149 spin_lock(&cli
->cl_cache
->ccc_lru_lock
);
3150 list_add(&cli
->cl_lru_osc
, &cli
->cl_cache
->ccc_lru
);
3151 spin_unlock(&cli
->cl_cache
->ccc_lru_lock
);
3156 if (KEY_IS(KEY_CACHE_LRU_SHRINK
)) {
3157 struct client_obd
*cli
= &obd
->u
.cli
;
3158 int nr
= atomic_read(&cli
->cl_lru_in_list
) >> 1;
3159 int target
= *(int *)val
;
3161 nr
= osc_lru_shrink(cli
, min(nr
, target
));
3166 if (!set
&& !KEY_IS(KEY_GRANT_SHRINK
))
3169 /* We pass all other commands directly to OST. Since nobody calls osc
3170 methods directly and everybody is supposed to go through LOV, we
3171 assume lov checked invalid values for us.
3172 The only recognised values so far are evict_by_nid and mds_conn.
3173 Even if something bad goes through, we'd get a -EINVAL from OST
3176 req
= ptlrpc_request_alloc(imp
, KEY_IS(KEY_GRANT_SHRINK
) ?
3177 &RQF_OST_SET_GRANT_INFO
:
3182 req_capsule_set_size(&req
->rq_pill
, &RMF_SETINFO_KEY
,
3183 RCL_CLIENT
, keylen
);
3184 if (!KEY_IS(KEY_GRANT_SHRINK
))
3185 req_capsule_set_size(&req
->rq_pill
, &RMF_SETINFO_VAL
,
3186 RCL_CLIENT
, vallen
);
3187 rc
= ptlrpc_request_pack(req
, LUSTRE_OST_VERSION
, OST_SET_INFO
);
3189 ptlrpc_request_free(req
);
3193 tmp
= req_capsule_client_get(&req
->rq_pill
, &RMF_SETINFO_KEY
);
3194 memcpy(tmp
, key
, keylen
);
3195 tmp
= req_capsule_client_get(&req
->rq_pill
, KEY_IS(KEY_GRANT_SHRINK
) ?
3198 memcpy(tmp
, val
, vallen
);
3200 if (KEY_IS(KEY_GRANT_SHRINK
)) {
3201 struct osc_grant_args
*aa
;
3204 CLASSERT(sizeof(*aa
) <= sizeof(req
->rq_async_args
));
3205 aa
= ptlrpc_req_async_args(req
);
3208 ptlrpc_req_finished(req
);
3211 *oa
= ((struct ost_body
*)val
)->oa
;
3213 req
->rq_interpret_reply
= osc_shrink_grant_interpret
;
3216 ptlrpc_request_set_replen(req
);
3217 if (!KEY_IS(KEY_GRANT_SHRINK
)) {
3218 LASSERT(set
!= NULL
);
3219 ptlrpc_set_add_req(set
, req
);
3220 ptlrpc_check_set(NULL
, set
);
3222 ptlrpcd_add_req(req
, PDL_POLICY_ROUND
, -1);
3228 static int osc_llog_init(struct obd_device
*obd
, struct obd_llog_group
*olg
,
3229 struct obd_device
*disk_obd
, int *index
)
3231 /* this code is not supposed to be used with LOD/OSP
3232 * to be removed soon */
3237 static int osc_llog_finish(struct obd_device
*obd
, int count
)
3239 struct llog_ctxt
*ctxt
;
3241 ctxt
= llog_get_context(obd
, LLOG_MDS_OST_ORIG_CTXT
);
3243 llog_cat_close(NULL
, ctxt
->loc_handle
);
3244 llog_cleanup(NULL
, ctxt
);
3247 ctxt
= llog_get_context(obd
, LLOG_SIZE_REPL_CTXT
);
3249 llog_cleanup(NULL
, ctxt
);
3253 static int osc_reconnect(const struct lu_env
*env
,
3254 struct obd_export
*exp
, struct obd_device
*obd
,
3255 struct obd_uuid
*cluuid
,
3256 struct obd_connect_data
*data
,
3259 struct client_obd
*cli
= &obd
->u
.cli
;
3261 if (data
!= NULL
&& (data
->ocd_connect_flags
& OBD_CONNECT_GRANT
)) {
3264 client_obd_list_lock(&cli
->cl_loi_list_lock
);
3265 data
->ocd_grant
= (cli
->cl_avail_grant
+ cli
->cl_dirty
) ?:
3266 2 * cli_brw_size(obd
);
3267 lost_grant
= cli
->cl_lost_grant
;
3268 cli
->cl_lost_grant
= 0;
3269 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
3271 CDEBUG(D_RPCTRACE
, "ocd_connect_flags: "LPX64
" ocd_version: %d"
3272 " ocd_grant: %d, lost: %ld.\n", data
->ocd_connect_flags
,
3273 data
->ocd_version
, data
->ocd_grant
, lost_grant
);
3279 static int osc_disconnect(struct obd_export
*exp
)
3281 struct obd_device
*obd
= class_exp2obd(exp
);
3282 struct llog_ctxt
*ctxt
;
3285 ctxt
= llog_get_context(obd
, LLOG_SIZE_REPL_CTXT
);
3287 if (obd
->u
.cli
.cl_conn_count
== 1) {
3288 /* Flush any remaining cancel messages out to the
3290 llog_sync(ctxt
, exp
, 0);
3292 llog_ctxt_put(ctxt
);
3294 CDEBUG(D_HA
, "No LLOG_SIZE_REPL_CTXT found in obd %p\n",
3298 rc
= client_disconnect_export(exp
);
3300 * Initially we put del_shrink_grant before disconnect_export, but it
3301 * causes the following problem if setup (connect) and cleanup
3302 * (disconnect) are tangled together.
3303 * connect p1 disconnect p2
3304 * ptlrpc_connect_import
3305 * ............... class_manual_cleanup
3308 * ptlrpc_connect_interrupt
3310 * add this client to shrink list
3312 * Bang! pinger trigger the shrink.
3313 * So the osc should be disconnected from the shrink list, after we
3314 * are sure the import has been destroyed. BUG18662
3316 if (obd
->u
.cli
.cl_import
== NULL
)
3317 osc_del_shrink_grant(&obd
->u
.cli
);
3321 static int osc_import_event(struct obd_device
*obd
,
3322 struct obd_import
*imp
,
3323 enum obd_import_event event
)
3325 struct client_obd
*cli
;
3328 LASSERT(imp
->imp_obd
== obd
);
3331 case IMP_EVENT_DISCON
: {
3333 client_obd_list_lock(&cli
->cl_loi_list_lock
);
3334 cli
->cl_avail_grant
= 0;
3335 cli
->cl_lost_grant
= 0;
3336 client_obd_list_unlock(&cli
->cl_loi_list_lock
);
3339 case IMP_EVENT_INACTIVE
: {
3340 rc
= obd_notify_observer(obd
, obd
, OBD_NOTIFY_INACTIVE
, NULL
);
3343 case IMP_EVENT_INVALIDATE
: {
3344 struct ldlm_namespace
*ns
= obd
->obd_namespace
;
3348 env
= cl_env_get(&refcheck
);
3352 /* all pages go to failing rpcs due to the invalid
3354 osc_io_unplug(env
, cli
, NULL
, PDL_POLICY_ROUND
);
3356 ldlm_namespace_cleanup(ns
, LDLM_FL_LOCAL_ONLY
);
3357 cl_env_put(env
, &refcheck
);
3362 case IMP_EVENT_ACTIVE
: {
3363 rc
= obd_notify_observer(obd
, obd
, OBD_NOTIFY_ACTIVE
, NULL
);
3366 case IMP_EVENT_OCD
: {
3367 struct obd_connect_data
*ocd
= &imp
->imp_connect_data
;
3369 if (ocd
->ocd_connect_flags
& OBD_CONNECT_GRANT
)
3370 osc_init_grant(&obd
->u
.cli
, ocd
);
3373 if (ocd
->ocd_connect_flags
& OBD_CONNECT_REQPORTAL
)
3374 imp
->imp_client
->cli_request_portal
=OST_REQUEST_PORTAL
;
3376 rc
= obd_notify_observer(obd
, obd
, OBD_NOTIFY_OCD
, NULL
);
3379 case IMP_EVENT_DEACTIVATE
: {
3380 rc
= obd_notify_observer(obd
, obd
, OBD_NOTIFY_DEACTIVATE
, NULL
);
3383 case IMP_EVENT_ACTIVATE
: {
3384 rc
= obd_notify_observer(obd
, obd
, OBD_NOTIFY_ACTIVATE
, NULL
);
3388 CERROR("Unknown import event %d\n", event
);
3395 * Determine whether the lock can be canceled before replaying the lock
3396 * during recovery, see bug16774 for detailed information.
3398 * \retval zero the lock can't be canceled
3399 * \retval other ok to cancel
3401 static int osc_cancel_for_recovery(struct ldlm_lock
*lock
)
3403 check_res_locked(lock
->l_resource
);
3406 * Cancel all unused extent lock in granted mode LCK_PR or LCK_CR.
3408 * XXX as a future improvement, we can also cancel unused write lock
3409 * if it doesn't have dirty data and active mmaps.
3411 if (lock
->l_resource
->lr_type
== LDLM_EXTENT
&&
3412 (lock
->l_granted_mode
== LCK_PR
||
3413 lock
->l_granted_mode
== LCK_CR
) &&
3414 (osc_dlm_lock_pageref(lock
) == 0))
3420 static int brw_queue_work(const struct lu_env
*env
, void *data
)
3422 struct client_obd
*cli
= data
;
3424 CDEBUG(D_CACHE
, "Run writeback work for client obd %p.\n", cli
);
3426 osc_io_unplug(env
, cli
, NULL
, PDL_POLICY_SAME
);
3430 int osc_setup(struct obd_device
*obd
, struct lustre_cfg
*lcfg
)
3432 struct lprocfs_static_vars lvars
= { 0 };
3433 struct client_obd
*cli
= &obd
->u
.cli
;
3437 rc
= ptlrpcd_addref();
3441 rc
= client_obd_setup(obd
, lcfg
);
3443 GOTO(out_ptlrpcd
, rc
);
3445 handler
= ptlrpcd_alloc_work(cli
->cl_import
, brw_queue_work
, cli
);
3446 if (IS_ERR(handler
))
3447 GOTO(out_client_setup
, rc
= PTR_ERR(handler
));
3448 cli
->cl_writeback_work
= handler
;
3450 rc
= osc_quota_setup(obd
);
3452 GOTO(out_ptlrpcd_work
, rc
);
3454 cli
->cl_grant_shrink_interval
= GRANT_SHRINK_INTERVAL
;
3455 lprocfs_osc_init_vars(&lvars
);
3456 if (lprocfs_obd_setup(obd
, lvars
.obd_vars
) == 0) {
3457 lproc_osc_attach_seqstat(obd
);
3458 sptlrpc_lprocfs_cliobd_attach(obd
);
3459 ptlrpc_lprocfs_register_obd(obd
);
3462 /* We need to allocate a few requests more, because
3463 * brw_interpret tries to create new requests before freeing
3464 * previous ones, Ideally we want to have 2x max_rpcs_in_flight
3465 * reserved, but I'm afraid that might be too much wasted RAM
3466 * in fact, so 2 is just my guess and still should work. */
3467 cli
->cl_import
->imp_rq_pool
=
3468 ptlrpc_init_rq_pool(cli
->cl_max_rpcs_in_flight
+ 2,
3470 ptlrpc_add_rqs_to_pool
);
3472 INIT_LIST_HEAD(&cli
->cl_grant_shrink_list
);
3473 ns_register_cancel(obd
->obd_namespace
, osc_cancel_for_recovery
);
3477 ptlrpcd_destroy_work(handler
);
3479 client_obd_cleanup(obd
);
3485 static int osc_precleanup(struct obd_device
*obd
, enum obd_cleanup_stage stage
)
3490 case OBD_CLEANUP_EARLY
: {
3491 struct obd_import
*imp
;
3492 imp
= obd
->u
.cli
.cl_import
;
3493 CDEBUG(D_HA
, "Deactivating import %s\n", obd
->obd_name
);
3494 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
3495 ptlrpc_deactivate_import(imp
);
3496 spin_lock(&imp
->imp_lock
);
3497 imp
->imp_pingable
= 0;
3498 spin_unlock(&imp
->imp_lock
);
3501 case OBD_CLEANUP_EXPORTS
: {
3502 struct client_obd
*cli
= &obd
->u
.cli
;
3504 * for echo client, export may be on zombie list, wait for
3505 * zombie thread to cull it, because cli.cl_import will be
3506 * cleared in client_disconnect_export():
3507 * class_export_destroy() -> obd_cleanup() ->
3508 * echo_device_free() -> echo_client_cleanup() ->
3509 * obd_disconnect() -> osc_disconnect() ->
3510 * client_disconnect_export()
3512 obd_zombie_barrier();
3513 if (cli
->cl_writeback_work
) {
3514 ptlrpcd_destroy_work(cli
->cl_writeback_work
);
3515 cli
->cl_writeback_work
= NULL
;
3517 obd_cleanup_client_import(obd
);
3518 ptlrpc_lprocfs_unregister_obd(obd
);
3519 lprocfs_obd_cleanup(obd
);
3520 rc
= obd_llog_finish(obd
, 0);
3522 CERROR("failed to cleanup llogging subsystems\n");
3529 int osc_cleanup(struct obd_device
*obd
)
3531 struct client_obd
*cli
= &obd
->u
.cli
;
3535 if (cli
->cl_cache
!= NULL
) {
3536 LASSERT(atomic_read(&cli
->cl_cache
->ccc_users
) > 0);
3537 spin_lock(&cli
->cl_cache
->ccc_lru_lock
);
3538 list_del_init(&cli
->cl_lru_osc
);
3539 spin_unlock(&cli
->cl_cache
->ccc_lru_lock
);
3540 cli
->cl_lru_left
= NULL
;
3541 atomic_dec(&cli
->cl_cache
->ccc_users
);
3542 cli
->cl_cache
= NULL
;
3545 /* free memory of osc quota cache */
3546 osc_quota_cleanup(obd
);
3548 rc
= client_obd_cleanup(obd
);
3554 int osc_process_config_base(struct obd_device
*obd
, struct lustre_cfg
*lcfg
)
3556 struct lprocfs_static_vars lvars
= { 0 };
3559 lprocfs_osc_init_vars(&lvars
);
3561 switch (lcfg
->lcfg_command
) {
3563 rc
= class_process_proc_param(PARAM_OSC
, lvars
.obd_vars
,
3573 static int osc_process_config(struct obd_device
*obd
, obd_count len
, void *buf
)
3575 return osc_process_config_base(obd
, buf
);
3578 struct obd_ops osc_obd_ops
= {
3579 .o_owner
= THIS_MODULE
,
3580 .o_setup
= osc_setup
,
3581 .o_precleanup
= osc_precleanup
,
3582 .o_cleanup
= osc_cleanup
,
3583 .o_add_conn
= client_import_add_conn
,
3584 .o_del_conn
= client_import_del_conn
,
3585 .o_connect
= client_connect_import
,
3586 .o_reconnect
= osc_reconnect
,
3587 .o_disconnect
= osc_disconnect
,
3588 .o_statfs
= osc_statfs
,
3589 .o_statfs_async
= osc_statfs_async
,
3590 .o_packmd
= osc_packmd
,
3591 .o_unpackmd
= osc_unpackmd
,
3592 .o_create
= osc_create
,
3593 .o_destroy
= osc_destroy
,
3594 .o_getattr
= osc_getattr
,
3595 .o_getattr_async
= osc_getattr_async
,
3596 .o_setattr
= osc_setattr
,
3597 .o_setattr_async
= osc_setattr_async
,
3599 .o_punch
= osc_punch
,
3601 .o_enqueue
= osc_enqueue
,
3602 .o_change_cbdata
= osc_change_cbdata
,
3603 .o_find_cbdata
= osc_find_cbdata
,
3604 .o_cancel
= osc_cancel
,
3605 .o_cancel_unused
= osc_cancel_unused
,
3606 .o_iocontrol
= osc_iocontrol
,
3607 .o_get_info
= osc_get_info
,
3608 .o_set_info_async
= osc_set_info_async
,
3609 .o_import_event
= osc_import_event
,
3610 .o_llog_init
= osc_llog_init
,
3611 .o_llog_finish
= osc_llog_finish
,
3612 .o_process_config
= osc_process_config
,
3613 .o_quotactl
= osc_quotactl
,
3614 .o_quotacheck
= osc_quotacheck
,
3617 extern struct lu_kmem_descr osc_caches
[];
3618 extern spinlock_t osc_ast_guard
;
3619 extern struct lock_class_key osc_ast_guard_class
;
3621 int __init
osc_init(void)
3623 struct lprocfs_static_vars lvars
= { 0 };
3626 /* print an address of _any_ initialized kernel symbol from this
3627 * module, to allow debugging with gdb that doesn't support data
3628 * symbols from modules.*/
3629 CDEBUG(D_INFO
, "Lustre OSC module (%p).\n", &osc_caches
);
3631 rc
= lu_kmem_init(osc_caches
);
3635 lprocfs_osc_init_vars(&lvars
);
3637 rc
= class_register_type(&osc_obd_ops
, NULL
, lvars
.module_vars
,
3638 LUSTRE_OSC_NAME
, &osc_device_type
);
3640 lu_kmem_fini(osc_caches
);
3644 spin_lock_init(&osc_ast_guard
);
3645 lockdep_set_class(&osc_ast_guard
, &osc_ast_guard_class
);
3650 static void /*__exit*/ osc_exit(void)
3652 class_unregister_type(LUSTRE_OSC_NAME
);
3653 lu_kmem_fini(osc_caches
);
3656 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
3657 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3658 MODULE_LICENSE("GPL");
3659 MODULE_VERSION(LUSTRE_VERSION_STRING
);
3661 module_init(osc_init
);
3662 module_exit(osc_exit
);