1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
4 * Copyright (C) 2001-2003 Cluster File Systems, Inc.
5 * Author Peter Braam <braam@clusterfs.com>
7 * This file is part of the Lustre file system, http://www.lustre.org
8 * Lustre is a trademark of Cluster File Systems, Inc.
10 * You may have signed or agreed to another license before downloading
11 * this software. If so, you are bound by the terms and conditions
12 * of that agreement, and the following does not apply to you. See the
13 * LICENSE file included with this distribution for more information.
15 * If you did not agree to a different license, then this copy of Lustre
16 * is open source software; you can redistribute it and/or modify it
17 * under the terms of version 2 of the GNU General Public License as
18 * published by the Free Software Foundation.
20 * In either case, Lustre is distributed in the hope that it will be
21 * useful, but WITHOUT ANY WARRANTY; without even the implied warranty
22 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23 * license text for more details.
25 * For testing and management it is treated as an obd_device,
26 * although * it does not export a full OBD method table (the
27 * requests are coming * in over the wire, so object target modules
28 * do not have a full * method table.)
33 # define EXPORT_SYMTAB
35 #define DEBUG_SUBSYSTEM S_OSC
38 # include <libcfs/libcfs.h>
39 #else /* __KERNEL__ */
40 # include <liblustre.h>
43 #include <lustre_dlm.h>
44 #include <libcfs/kp30.h>
45 #include <lustre_net.h>
46 #include <lustre/lustre_user.h>
54 #include <lustre_ha.h>
55 #include <lprocfs_status.h>
56 #include <lustre_log.h>
57 #include <lustre_debug.h>
58 #include <lustre_param.h>
59 #include "osc_internal.h"
61 static quota_interface_t *quota_interface = NULL;
62 extern quota_interface_t osc_quota_interface;
64 static void osc_release_ppga(struct brw_page **ppga, obd_count count);
66 /* Pack OSC object metadata for disk storage (LE byte order). */
67 static int osc_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,
68 struct lov_stripe_md *lsm)
73 lmm_size = sizeof(**lmmp);
78 OBD_FREE(*lmmp, lmm_size);
84 OBD_ALLOC(*lmmp, lmm_size);
90 LASSERT(lsm->lsm_object_id);
91 LASSERT(lsm->lsm_object_gr);
92 (*lmmp)->lmm_object_id = cpu_to_le64(lsm->lsm_object_id);
93 (*lmmp)->lmm_object_gr = cpu_to_le64(lsm->lsm_object_gr);
99 /* Unpack OSC object metadata from disk storage (LE byte order). */
100 static int osc_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
101 struct lov_mds_md *lmm, int lmm_bytes)
107 if (lmm_bytes < sizeof (*lmm)) {
108 CERROR("lov_mds_md too small: %d, need %d\n",
109 lmm_bytes, (int)sizeof(*lmm));
112 /* XXX LOV_MAGIC etc check? */
114 if (lmm->lmm_object_id == 0) {
115 CERROR("lov_mds_md: zero lmm_object_id\n");
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);
135 OBD_ALLOC((*lsmp)->lsm_oinfo[0], sizeof(struct lov_oinfo));
136 if ((*lsmp)->lsm_oinfo[0] == NULL) {
137 OBD_FREE(*lsmp, lsm_size);
140 loi_init((*lsmp)->lsm_oinfo[0]);
144 /* XXX zero *lsmp? */
145 (*lsmp)->lsm_object_id = le64_to_cpu (lmm->lmm_object_id);
146 (*lsmp)->lsm_object_gr = le64_to_cpu (lmm->lmm_object_gr);
147 LASSERT((*lsmp)->lsm_object_id);
148 LASSERT((*lsmp)->lsm_object_gr);
151 (*lsmp)->lsm_maxbytes = LUSTRE_STRIPE_MAXBYTES;
156 static inline void osc_pack_capa(struct ptlrpc_request *req, int offset,
157 struct ost_body *body, void *capa)
159 struct obd_capa *oc = (struct obd_capa *)capa;
160 struct lustre_capa *c;
165 c = lustre_msg_buf(req->rq_reqmsg, offset, sizeof(*c));
168 body->oa.o_valid |= OBD_MD_FLOSSCAPA;
169 DEBUG_CAPA(D_SEC, c, "pack");
172 static inline void osc_pack_req_body(struct ptlrpc_request *req, int offset,
173 struct obd_info *oinfo)
175 struct ost_body *body;
177 body = lustre_msg_buf(req->rq_reqmsg, offset, sizeof(*body));
178 body->oa = *oinfo->oi_oa;
179 osc_pack_capa(req, offset + 1, body, oinfo->oi_capa);
182 static int osc_getattr_interpret(struct ptlrpc_request *req,
183 struct osc_async_args *aa, int rc)
185 struct ost_body *body;
191 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
192 lustre_swab_ost_body);
194 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
195 memcpy(aa->aa_oi->oi_oa, &body->oa, sizeof(*aa->aa_oi->oi_oa));
197 /* This should really be sent by the OST */
198 aa->aa_oi->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
199 aa->aa_oi->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
201 CERROR("can't unpack ost_body\n");
203 aa->aa_oi->oi_oa->o_valid = 0;
206 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
210 static int osc_getattr_async(struct obd_export *exp, struct obd_info *oinfo,
211 struct ptlrpc_request_set *set)
213 struct ptlrpc_request *req;
214 struct ost_body *body;
215 int size[3] = { sizeof(struct ptlrpc_body), sizeof(*body) };
216 struct osc_async_args *aa;
219 size[REQ_REC_OFF + 1] = oinfo->oi_capa ? sizeof(struct lustre_capa) : 0;
220 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
221 OST_GETATTR, 3, size,NULL);
225 osc_pack_req_body(req, REQ_REC_OFF, oinfo);
227 ptlrpc_req_set_repsize(req, 2, size);
228 req->rq_interpret_reply = osc_getattr_interpret;
230 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
231 aa = (struct osc_async_args *)&req->rq_async_args;
234 ptlrpc_set_add_req(set, req);
238 static int osc_getattr(struct obd_export *exp, struct obd_info *oinfo)
240 struct ptlrpc_request *req;
241 struct ost_body *body;
242 int rc, size[3] = { sizeof(struct ptlrpc_body), sizeof(*body) };
245 size[REQ_REC_OFF + 1] = oinfo->oi_capa ? sizeof(struct lustre_capa) : 0;
246 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
247 OST_GETATTR, 3, size, NULL);
251 osc_pack_req_body(req, REQ_REC_OFF, oinfo);
253 ptlrpc_req_set_repsize(req, 2, size);
255 rc = ptlrpc_queue_wait(req);
257 CERROR("%s failed: rc = %d\n", __FUNCTION__, rc);
261 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
262 lustre_swab_ost_body);
264 CERROR ("can't unpack ost_body\n");
265 GOTO (out, rc = -EPROTO);
268 CDEBUG(D_INODE, "mode: %o\n", body->oa.o_mode);
269 *oinfo->oi_oa = body->oa;
271 /* This should really be sent by the OST */
272 oinfo->oi_oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
273 oinfo->oi_oa->o_valid |= OBD_MD_FLBLKSZ;
277 ptlrpc_req_finished(req);
281 static int osc_setattr(struct obd_export *exp, struct obd_info *oinfo,
282 struct obd_trans_info *oti)
284 struct ptlrpc_request *req;
285 struct ost_body *body;
286 int rc, size[3] = { sizeof(struct ptlrpc_body), sizeof(*body) };
289 LASSERT(!(oinfo->oi_oa->o_valid & OBD_MD_FLGROUP) ||
290 oinfo->oi_oa->o_gr > 0);
291 size[REQ_REC_OFF + 1] = oinfo->oi_capa ? sizeof(struct lustre_capa) : 0;
292 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
293 OST_SETATTR, 3, size, NULL);
297 osc_pack_req_body(req, REQ_REC_OFF, oinfo);
299 ptlrpc_req_set_repsize(req, 2, size);
301 rc = ptlrpc_queue_wait(req);
305 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
306 lustre_swab_ost_body);
308 GOTO(out, rc = -EPROTO);
310 *oinfo->oi_oa = body->oa;
314 ptlrpc_req_finished(req);
318 static int osc_setattr_interpret(struct ptlrpc_request *req,
319 struct osc_async_args *aa, int rc)
321 struct ost_body *body;
327 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
328 lustre_swab_ost_body);
330 CERROR("can't unpack ost_body\n");
331 GOTO(out, rc = -EPROTO);
334 *aa->aa_oi->oi_oa = body->oa;
336 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
340 static int osc_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
341 struct obd_trans_info *oti,
342 struct ptlrpc_request_set *rqset)
344 struct ptlrpc_request *req;
345 int size[3] = { sizeof(struct ptlrpc_body), sizeof(struct ost_body) };
346 struct osc_async_args *aa;
349 size[REQ_REC_OFF + 1] = oinfo->oi_capa ? sizeof(struct lustre_capa) : 0;
350 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
351 OST_SETATTR, 3, size, NULL);
355 osc_pack_req_body(req, REQ_REC_OFF, oinfo);
356 if (oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE) {
358 *obdo_logcookie(oinfo->oi_oa) = *oti->oti_logcookies;
361 ptlrpc_req_set_repsize(req, 2, size);
362 /* do mds to ost setattr asynchronouly */
364 /* Do not wait for response. */
365 ptlrpcd_add_req(req);
367 req->rq_interpret_reply = osc_setattr_interpret;
369 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
370 aa = (struct osc_async_args *)&req->rq_async_args;
373 ptlrpc_set_add_req(rqset, req);
379 int osc_real_create(struct obd_export *exp, struct obdo *oa,
380 struct lov_stripe_md **ea, struct obd_trans_info *oti)
382 struct ptlrpc_request *req;
383 struct ost_body *body;
384 struct lov_stripe_md *lsm;
385 int rc, size[2] = { sizeof(struct ptlrpc_body), sizeof(*body) };
393 rc = obd_alloc_memmd(exp, &lsm);
398 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
399 OST_CREATE, 2, size, NULL);
401 GOTO(out, rc = -ENOMEM);
403 body = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF, sizeof(*body));
406 ptlrpc_req_set_repsize(req, 2, size);
407 if (oa->o_valid & OBD_MD_FLINLINE) {
408 LASSERT((oa->o_valid & OBD_MD_FLFLAGS) &&
409 oa->o_flags == OBD_FL_DELORPHAN);
411 "delorphan from OST integration");
412 /* Don't resend the delorphan req */
413 req->rq_no_resend = req->rq_no_delay = 1;
416 rc = ptlrpc_queue_wait(req);
420 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
421 lustre_swab_ost_body);
423 CERROR ("can't unpack ost_body\n");
424 GOTO (out_req, rc = -EPROTO);
429 /* This should really be sent by the OST */
430 oa->o_blksize = PTLRPC_MAX_BRW_SIZE;
431 oa->o_valid |= OBD_MD_FLBLKSZ;
433 /* XXX LOV STACKING: the lsm that is passed to us from LOV does not
434 * have valid lsm_oinfo data structs, so don't go touching that.
435 * This needs to be fixed in a big way.
437 lsm->lsm_object_id = oa->o_id;
438 lsm->lsm_object_gr = oa->o_gr;
442 oti->oti_transno = lustre_msg_get_transno(req->rq_repmsg);
444 if (oa->o_valid & OBD_MD_FLCOOKIE) {
445 if (!oti->oti_logcookies)
446 oti_alloc_cookies(oti, 1);
447 *oti->oti_logcookies = *obdo_logcookie(oa);
451 CDEBUG(D_HA, "transno: "LPD64"\n",
452 lustre_msg_get_transno(req->rq_repmsg));
455 ptlrpc_req_finished(req);
458 obd_free_memmd(exp, &lsm);
462 static int osc_punch_interpret(struct ptlrpc_request *req,
463 struct osc_async_args *aa, int rc)
465 struct ost_body *body;
471 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof (*body),
472 lustre_swab_ost_body);
474 CERROR ("can't unpack ost_body\n");
475 GOTO(out, rc = -EPROTO);
478 *aa->aa_oi->oi_oa = body->oa;
480 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
484 static int osc_punch(struct obd_export *exp, struct obd_info *oinfo,
485 struct obd_trans_info *oti,
486 struct ptlrpc_request_set *rqset)
488 struct ptlrpc_request *req;
489 struct osc_async_args *aa;
490 struct ost_body *body;
491 int size[3] = { sizeof(struct ptlrpc_body), sizeof(*body) };
499 size[REQ_REC_OFF + 1] = oinfo->oi_capa? sizeof(struct lustre_capa) : 0;
500 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
501 OST_PUNCH, 3, size, NULL);
505 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
507 osc_pack_req_body(req, REQ_REC_OFF, oinfo);
508 /* overload the size and blocks fields in the oa with start/end */
509 body = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF, sizeof(*body));
510 body->oa.o_size = oinfo->oi_policy.l_extent.start;
511 body->oa.o_blocks = oinfo->oi_policy.l_extent.end;
512 body->oa.o_valid |= (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS);
514 ptlrpc_req_set_repsize(req, 2, size);
516 req->rq_interpret_reply = osc_punch_interpret;
517 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
518 aa = (struct osc_async_args *)&req->rq_async_args;
520 ptlrpc_set_add_req(rqset, req);
525 static int osc_sync(struct obd_export *exp, struct obdo *oa,
526 struct lov_stripe_md *md, obd_size start, obd_size end,
529 struct ptlrpc_request *req;
530 struct ost_body *body;
531 int rc, size[3] = { sizeof(struct ptlrpc_body), sizeof(*body) };
539 size[REQ_REC_OFF + 1] = capa ? sizeof(struct lustre_capa) : 0;
541 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
542 OST_SYNC, 3, size, NULL);
546 /* overload the size and blocks fields in the oa with start/end */
547 body = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF, sizeof(*body));
549 body->oa.o_size = start;
550 body->oa.o_blocks = end;
551 body->oa.o_valid |= (OBD_MD_FLSIZE | OBD_MD_FLBLOCKS);
553 osc_pack_capa(req, REQ_REC_OFF + 1, body, capa);
555 ptlrpc_req_set_repsize(req, 2, size);
557 rc = ptlrpc_queue_wait(req);
561 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
562 lustre_swab_ost_body);
564 CERROR ("can't unpack ost_body\n");
565 GOTO (out, rc = -EPROTO);
572 ptlrpc_req_finished(req);
576 /* Find and cancel locally locks matched by @mode in the resource found by
577 * @objid. Found locks are added into @cancel list. Returns the amount of
578 * locks added to @cancels list. */
579 static int osc_resource_get_unused(struct obd_export *exp, struct obdo *oa,
580 struct list_head *cancels, ldlm_mode_t mode,
583 struct ldlm_namespace *ns = exp->exp_obd->obd_namespace;
584 struct ldlm_res_id res_id = { .name = { oa->o_id, 0, oa->o_gr, 0 } };
585 struct ldlm_resource *res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
592 count = ldlm_cancel_resource_local(res, cancels, NULL, mode,
593 lock_flags, 0, NULL);
594 ldlm_resource_putref(res);
598 /* Destroy requests can be async always on the client, and we don't even really
599 * care about the return code since the client cannot do anything at all about
601 * When the MDS is unlinking a filename, it saves the file objects into a
602 * recovery llog, and these object records are cancelled when the OST reports
603 * they were destroyed and sync'd to disk (i.e. transaction committed).
604 * If the client dies, or the OST is down when the object should be destroyed,
605 * the records are not cancelled, and when the OST reconnects to the MDS next,
606 * it will retrieve the llog unlink logs and then sends the log cancellation
607 * cookies to the MDS after committing destroy transactions. */
608 static int osc_destroy(struct obd_export *exp, struct obdo *oa,
609 struct lov_stripe_md *ea, struct obd_trans_info *oti,
610 struct obd_export *md_export)
612 CFS_LIST_HEAD(cancels);
613 struct ptlrpc_request *req;
614 struct ost_body *body;
615 int size[3] = { sizeof(struct ptlrpc_body), sizeof(*body), 0 };
616 int count, bufcount = 2;
624 count = osc_resource_get_unused(exp, oa, &cancels, LCK_PW,
625 LDLM_FL_DISCARD_DATA);
626 if (exp_connect_cancelset(exp) && count) {
628 size[REQ_REC_OFF + 1] = ldlm_request_bufsize(count,OST_DESTROY);
630 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
631 OST_DESTROY, bufcount, size, NULL);
632 if (exp_connect_cancelset(exp) && req)
633 ldlm_cli_cancel_list(&cancels, count, req, REQ_REC_OFF + 1, 0);
635 ldlm_lock_list_put(&cancels, l_bl_ast, count);
640 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
642 body = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF, sizeof(*body));
643 if (oti != NULL && oa->o_valid & OBD_MD_FLCOOKIE)
644 memcpy(obdo_logcookie(oa), oti->oti_logcookies,
645 sizeof(*oti->oti_logcookies));
648 ptlrpc_req_set_repsize(req, 2, size);
650 ptlrpcd_add_req(req);
654 static void osc_announce_cached(struct client_obd *cli, struct obdo *oa,
657 obd_flag bits = OBD_MD_FLBLOCKS|OBD_MD_FLGRANT;
659 LASSERT(!(oa->o_valid & bits));
662 client_obd_list_lock(&cli->cl_loi_list_lock);
663 oa->o_dirty = cli->cl_dirty;
664 if (cli->cl_dirty > cli->cl_dirty_max) {
665 CERROR("dirty %lu > dirty_max %lu\n",
666 cli->cl_dirty, cli->cl_dirty_max);
668 } else if (atomic_read(&obd_dirty_pages) > obd_max_dirty_pages) {
669 CERROR("dirty %d > system dirty_max %d\n",
670 atomic_read(&obd_dirty_pages), obd_max_dirty_pages);
672 } else if (cli->cl_dirty_max - cli->cl_dirty > 0x7fffffff) {
673 CERROR("dirty %lu - dirty_max %lu too big???\n",
674 cli->cl_dirty, cli->cl_dirty_max);
677 long max_in_flight = (cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT)*
678 (cli->cl_max_rpcs_in_flight + 1);
679 oa->o_undirty = max(cli->cl_dirty_max, max_in_flight);
681 oa->o_grant = cli->cl_avail_grant;
682 oa->o_dropped = cli->cl_lost_grant;
683 cli->cl_lost_grant = 0;
684 client_obd_list_unlock(&cli->cl_loi_list_lock);
685 CDEBUG(D_CACHE,"dirty: "LPU64" undirty: %u dropped %u grant: "LPU64"\n",
686 oa->o_dirty, oa->o_undirty, oa->o_dropped, oa->o_grant);
689 /* caller must hold loi_list_lock */
690 static void osc_consume_write_grant(struct client_obd *cli,
691 struct brw_page *pga)
693 atomic_inc(&obd_dirty_pages);
694 cli->cl_dirty += CFS_PAGE_SIZE;
695 cli->cl_avail_grant -= CFS_PAGE_SIZE;
696 pga->flag |= OBD_BRW_FROM_GRANT;
697 CDEBUG(D_CACHE, "using %lu grant credits for brw %p page %p\n",
698 CFS_PAGE_SIZE, pga, pga->pg);
699 LASSERT(cli->cl_avail_grant >= 0);
702 /* the companion to osc_consume_write_grant, called when a brw has completed.
703 * must be called with the loi lock held. */
704 static void osc_release_write_grant(struct client_obd *cli,
705 struct brw_page *pga, int sent)
707 int blocksize = cli->cl_import->imp_obd->obd_osfs.os_bsize ? : 4096;
710 if (!(pga->flag & OBD_BRW_FROM_GRANT)) {
715 pga->flag &= ~OBD_BRW_FROM_GRANT;
716 atomic_dec(&obd_dirty_pages);
717 cli->cl_dirty -= CFS_PAGE_SIZE;
719 cli->cl_lost_grant += CFS_PAGE_SIZE;
720 CDEBUG(D_CACHE, "lost grant: %lu avail grant: %lu dirty: %lu\n",
721 cli->cl_lost_grant, cli->cl_avail_grant, cli->cl_dirty);
722 } else if (CFS_PAGE_SIZE != blocksize && pga->count != CFS_PAGE_SIZE) {
723 /* For short writes we shouldn't count parts of pages that
724 * span a whole block on the OST side, or our accounting goes
725 * wrong. Should match the code in filter_grant_check. */
726 int offset = pga->off & ~CFS_PAGE_MASK;
727 int count = pga->count + (offset & (blocksize - 1));
728 int end = (offset + pga->count) & (blocksize - 1);
730 count += blocksize - end;
732 cli->cl_lost_grant += CFS_PAGE_SIZE - count;
733 CDEBUG(D_CACHE, "lost %lu grant: %lu avail: %lu dirty: %lu\n",
734 CFS_PAGE_SIZE - count, cli->cl_lost_grant,
735 cli->cl_avail_grant, cli->cl_dirty);
741 static unsigned long rpcs_in_flight(struct client_obd *cli)
743 return cli->cl_r_in_flight + cli->cl_w_in_flight;
746 /* caller must hold loi_list_lock */
747 void osc_wake_cache_waiters(struct client_obd *cli)
749 struct list_head *l, *tmp;
750 struct osc_cache_waiter *ocw;
753 list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
754 /* if we can't dirty more, we must wait until some is written */
755 if ((cli->cl_dirty + CFS_PAGE_SIZE > cli->cl_dirty_max) ||
756 (atomic_read(&obd_dirty_pages) + 1 > obd_max_dirty_pages)) {
757 CDEBUG(D_CACHE, "no dirty room: dirty: %ld "
758 "osc max %ld, sys max %d\n", cli->cl_dirty,
759 cli->cl_dirty_max, obd_max_dirty_pages);
763 /* if still dirty cache but no grant wait for pending RPCs that
764 * may yet return us some grant before doing sync writes */
765 if (cli->cl_w_in_flight && cli->cl_avail_grant < CFS_PAGE_SIZE) {
766 CDEBUG(D_CACHE, "%u BRW writes in flight, no grant\n",
767 cli->cl_w_in_flight);
771 ocw = list_entry(l, struct osc_cache_waiter, ocw_entry);
772 list_del_init(&ocw->ocw_entry);
773 if (cli->cl_avail_grant < CFS_PAGE_SIZE) {
774 /* no more RPCs in flight to return grant, do sync IO */
775 ocw->ocw_rc = -EDQUOT;
776 CDEBUG(D_INODE, "wake oap %p for sync\n", ocw->ocw_oap);
778 osc_consume_write_grant(cli,
779 &ocw->ocw_oap->oap_brw_page);
782 cfs_waitq_signal(&ocw->ocw_waitq);
788 static void osc_init_grant(struct client_obd *cli, struct obd_connect_data *ocd)
790 client_obd_list_lock(&cli->cl_loi_list_lock);
791 cli->cl_avail_grant = ocd->ocd_grant;
792 client_obd_list_unlock(&cli->cl_loi_list_lock);
794 CDEBUG(D_CACHE, "setting cl_avail_grant: %ld cl_lost_grant: %ld\n",
795 cli->cl_avail_grant, cli->cl_lost_grant);
796 LASSERT(cli->cl_avail_grant >= 0);
799 static void osc_update_grant(struct client_obd *cli, struct ost_body *body)
801 client_obd_list_lock(&cli->cl_loi_list_lock);
802 CDEBUG(D_CACHE, "got "LPU64" extra grant\n", body->oa.o_grant);
803 cli->cl_avail_grant += body->oa.o_grant;
804 /* waiters are woken in brw_interpret_oap */
805 client_obd_list_unlock(&cli->cl_loi_list_lock);
808 /* We assume that the reason this OSC got a short read is because it read
809 * beyond the end of a stripe file; i.e. lustre is reading a sparse file
810 * via the LOV, and it _knows_ it's reading inside the file, it's just that
811 * this stripe never got written at or beyond this stripe offset yet. */
812 static void handle_short_read(int nob_read, obd_count page_count,
813 struct brw_page **pga)
818 /* skip bytes read OK */
819 while (nob_read > 0) {
820 LASSERT (page_count > 0);
822 if (pga[i]->count > nob_read) {
823 /* EOF inside this page */
824 ptr = cfs_kmap(pga[i]->pg) +
825 (pga[i]->off & ~CFS_PAGE_MASK);
826 memset(ptr + nob_read, 0, pga[i]->count - nob_read);
827 cfs_kunmap(pga[i]->pg);
833 nob_read -= pga[i]->count;
838 /* zero remaining pages */
839 while (page_count-- > 0) {
840 ptr = cfs_kmap(pga[i]->pg) + (pga[i]->off & ~CFS_PAGE_MASK);
841 memset(ptr, 0, pga[i]->count);
842 cfs_kunmap(pga[i]->pg);
847 static int check_write_rcs(struct ptlrpc_request *req,
848 int requested_nob, int niocount,
849 obd_count page_count, struct brw_page **pga)
853 /* return error if any niobuf was in error */
854 remote_rcs = lustre_swab_repbuf(req, REQ_REC_OFF + 1,
855 sizeof(*remote_rcs) * niocount, NULL);
856 if (remote_rcs == NULL) {
857 CERROR("Missing/short RC vector on BRW_WRITE reply\n");
860 if (lustre_msg_swabbed(req->rq_repmsg))
861 for (i = 0; i < niocount; i++)
862 __swab32s(&remote_rcs[i]);
864 for (i = 0; i < niocount; i++) {
865 if (remote_rcs[i] < 0)
866 return(remote_rcs[i]);
868 if (remote_rcs[i] != 0) {
869 CERROR("rc[%d] invalid (%d) req %p\n",
870 i, remote_rcs[i], req);
875 if (req->rq_bulk->bd_nob_transferred != requested_nob) {
876 CERROR("Unexpected # bytes transferred: %d (requested %d)\n",
877 requested_nob, req->rq_bulk->bd_nob_transferred);
884 static inline int can_merge_pages(struct brw_page *p1, struct brw_page *p2)
886 if (p1->flag != p2->flag) {
887 unsigned mask = ~OBD_BRW_FROM_GRANT;
889 /* warn if we try to combine flags that we don't know to be
891 if ((p1->flag & mask) != (p2->flag & mask))
892 CERROR("is it ok to have flags 0x%x and 0x%x in the "
893 "same brw?\n", p1->flag, p2->flag);
897 return (p1->off + p1->count == p2->off);
900 static obd_count osc_checksum_bulk(int nob, obd_count pg_count,
901 struct brw_page **pga)
906 LASSERT (pg_count > 0);
907 while (nob > 0 && pg_count > 0) {
908 char *ptr = cfs_kmap(pga[i]->pg);
909 int off = pga[i]->off & ~CFS_PAGE_MASK;
910 int count = pga[i]->count > nob ? nob : pga[i]->count;
912 /* corrupt the data before we compute the checksum, to
913 * simulate an OST->client data error */
915 OBD_FAIL_CHECK_ONCE(OBD_FAIL_OSC_CHECKSUM_RECEIVE))
916 memcpy(ptr + off, "bad1", min(4, nob));
917 cksum = crc32_le(cksum, ptr + off, count);
918 cfs_kunmap(pga[i]->pg);
919 LL_CDEBUG_PAGE(D_PAGE, pga[i]->pg, "off %d checksum %x\n",
922 nob -= pga[i]->count;
926 /* For sending we only compute the wrong checksum instead
927 * of corrupting the data so it is still correct on a redo */
928 if (OBD_FAIL_CHECK_ONCE(OBD_FAIL_OSC_CHECKSUM_SEND))
934 static int osc_brw_prep_request(int cmd, struct client_obd *cli,struct obdo *oa,
935 struct lov_stripe_md *lsm, obd_count page_count,
936 struct brw_page **pga,
937 struct ptlrpc_request **reqp,
938 struct obd_capa *ocapa)
940 struct ptlrpc_request *req;
941 struct ptlrpc_bulk_desc *desc;
942 struct ost_body *body;
943 struct obd_ioobj *ioobj;
944 struct niobuf_remote *niobuf;
945 int size[5] = { sizeof(struct ptlrpc_body), sizeof(*body) };
946 int niocount, i, requested_nob, opc, rc;
947 struct ptlrpc_request_pool *pool;
948 struct lustre_capa *capa;
949 struct osc_brw_async_args *aa;
952 if ((cmd & OBD_BRW_WRITE) != 0) {
954 pool = cli->cl_import->imp_rq_pool;
960 for (niocount = i = 1; i < page_count; i++) {
961 if (!can_merge_pages(pga[i - 1], pga[i]))
965 size[REQ_REC_OFF + 1] = sizeof(*ioobj);
966 size[REQ_REC_OFF + 2] = niocount * sizeof(*niobuf);
968 size[REQ_REC_OFF + 3] = sizeof(*capa);
970 OBD_FAIL_RETURN(OBD_FAIL_OSC_BRW_PREP_REQ, -ENOMEM);
971 req = ptlrpc_prep_req_pool(cli->cl_import, LUSTRE_OST_VERSION, opc, 5,
972 size, NULL, pool, NULL);
976 req->rq_request_portal = OST_IO_PORTAL; /* bug 7198 */
978 if (opc == OST_WRITE)
979 desc = ptlrpc_prep_bulk_imp (req, page_count,
980 BULK_GET_SOURCE, OST_BULK_PORTAL);
982 desc = ptlrpc_prep_bulk_imp (req, page_count,
983 BULK_PUT_SINK, OST_BULK_PORTAL);
985 GOTO(out, rc = -ENOMEM);
986 /* NB request now owns desc and will free it when it gets freed */
988 body = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF, sizeof(*body));
989 ioobj = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF + 1, sizeof(*ioobj));
990 niobuf = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF + 2,
991 niocount * sizeof(*niobuf));
995 obdo_to_ioobj(oa, ioobj);
996 ioobj->ioo_bufcnt = niocount;
998 capa = lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF + 3,
1000 capa_cpy(capa, ocapa);
1001 body->oa.o_valid |= OBD_MD_FLOSSCAPA;
1004 LASSERT (page_count > 0);
1005 for (requested_nob = i = 0; i < page_count; i++, niobuf++) {
1006 struct brw_page *pg = pga[i];
1007 struct brw_page *pg_prev = pga[i - 1];
1009 LASSERT(pg->count > 0);
1010 LASSERTF((pg->off & ~CFS_PAGE_MASK) + pg->count <= CFS_PAGE_SIZE,
1011 "i: %d pg: %p off: "LPU64", count: %u\n", i, pg,
1012 pg->off, pg->count);
1014 LASSERTF(i == 0 || pg->off > pg_prev->off,
1015 "i %d p_c %u pg %p [pri %lu ind %lu] off "LPU64
1016 " prev_pg %p [pri %lu ind %lu] off "LPU64"\n",
1018 pg->pg, page_private(pg->pg), pg->pg->index, pg->off,
1019 pg_prev->pg, page_private(pg_prev->pg),
1020 pg_prev->pg->index, pg_prev->off);
1022 LASSERTF(i == 0 || pg->off > pg_prev->off,
1023 "i %d p_c %u\n", i, page_count);
1025 LASSERT((pga[0]->flag & OBD_BRW_SRVLOCK) ==
1026 (pg->flag & OBD_BRW_SRVLOCK));
1028 ptlrpc_prep_bulk_page(desc, pg->pg, pg->off & ~CFS_PAGE_MASK,
1030 requested_nob += pg->count;
1032 if (i > 0 && can_merge_pages(pg_prev, pg)) {
1034 niobuf->len += pg->count;
1036 niobuf->offset = pg->off;
1037 niobuf->len = pg->count;
1038 niobuf->flags = pg->flag;
1042 LASSERT((void *)(niobuf - niocount) ==
1043 lustre_msg_buf(req->rq_reqmsg, REQ_REC_OFF + 2,
1044 niocount * sizeof(*niobuf)));
1045 osc_announce_cached(cli, &body->oa, opc == OST_WRITE ? requested_nob:0);
1047 /* size[REQ_REC_OFF] still sizeof (*body) */
1048 if (opc == OST_WRITE) {
1049 if (unlikely(cli->cl_checksum)) {
1050 body->oa.o_valid |= OBD_MD_FLCKSUM;
1051 body->oa.o_cksum = osc_checksum_bulk(requested_nob,
1053 CDEBUG(D_PAGE, "checksum at write origin: %x\n",
1055 /* save this in 'oa', too, for later checking */
1056 oa->o_valid |= OBD_MD_FLCKSUM;
1058 /* clear out the checksum flag, in case this is a
1059 * resend but cl_checksum is no longer set. b=11238 */
1060 oa->o_valid &= ~OBD_MD_FLCKSUM;
1062 oa->o_cksum = body->oa.o_cksum;
1063 /* 1 RC per niobuf */
1064 size[REPLY_REC_OFF + 1] = sizeof(__u32) * niocount;
1065 ptlrpc_req_set_repsize(req, 3, size);
1067 if (unlikely(cli->cl_checksum))
1068 body->oa.o_valid |= OBD_MD_FLCKSUM;
1069 /* 1 RC for the whole I/O */
1070 ptlrpc_req_set_repsize(req, 2, size);
1073 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1074 aa = (struct osc_brw_async_args *)&req->rq_async_args;
1076 aa->aa_requested_nob = requested_nob;
1077 aa->aa_nio_count = niocount;
1078 aa->aa_page_count = page_count;
1079 aa->aa_retries = 5; /*retry for checksum errors; lprocfs? */
1082 INIT_LIST_HEAD(&aa->aa_oaps);
1088 ptlrpc_req_finished (req);
1092 static int check_write_checksum(struct obdo *oa, const lnet_process_id_t *peer,
1093 __u32 client_cksum, __u32 server_cksum,
1094 int nob, obd_count page_count,
1095 struct brw_page **pga)
1100 if (server_cksum == client_cksum) {
1101 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1105 new_cksum = osc_checksum_bulk(nob, page_count, pga);
1107 if (new_cksum == server_cksum)
1108 msg = "changed on the client after we checksummed it - "
1109 "likely false positive due to mmap IO (bug 11742)";
1110 else if (new_cksum == client_cksum)
1111 msg = "changed in transit before arrival at OST";
1113 msg = "changed in transit AND doesn't match the original - "
1114 "likely false positive due to mmap IO (bug 11742)";
1116 LCONSOLE_ERROR_MSG(0x132, "BAD WRITE CHECKSUM: %s: from %s inum "
1117 LPU64"/"LPU64" object "LPU64"/"LPU64" extent "
1118 "["LPU64"-"LPU64"]\n",
1119 msg, libcfs_nid2str(peer->nid),
1120 oa->o_valid & OBD_MD_FLFID ? oa->o_fid : (__u64)0,
1121 oa->o_valid & OBD_MD_FLFID ? oa->o_generation :
1124 oa->o_valid & OBD_MD_FLGROUP ? oa->o_gr : (__u64)0,
1126 pga[page_count-1]->off + pga[page_count-1]->count - 1);
1127 CERROR("original client csum %x, server csum %x, client csum now %x\n",
1128 client_cksum, server_cksum, new_cksum);
1132 /* Note rc enters this function as number of bytes transferred */
1133 static int osc_brw_fini_request(struct ptlrpc_request *req, int rc)
1135 struct osc_brw_async_args *aa = (void *)&req->rq_async_args;
1136 const lnet_process_id_t *peer =
1137 &req->rq_import->imp_connection->c_peer;
1138 struct client_obd *cli = aa->aa_cli;
1139 struct ost_body *body;
1140 __u32 client_cksum = 0;
1143 if (rc < 0 && rc != -EDQUOT)
1146 LASSERTF(req->rq_repmsg != NULL, "rc = %d\n", rc);
1147 body = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*body),
1148 lustre_swab_ost_body);
1150 CERROR ("Can't unpack body\n");
1154 /* set/clear over quota flag for a uid/gid */
1155 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE &&
1156 body->oa.o_valid & (OBD_MD_FLUSRQUOTA | OBD_MD_FLGRPQUOTA))
1157 lquota_setdq(quota_interface, cli, body->oa.o_uid,
1158 body->oa.o_gid, body->oa.o_valid,
1164 if (unlikely(aa->aa_oa->o_valid & OBD_MD_FLCKSUM))
1165 client_cksum = aa->aa_oa->o_cksum; /* save for later */
1167 osc_update_grant(cli, body);
1169 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE) {
1171 CERROR ("Unexpected +ve rc %d\n", rc);
1174 LASSERT(req->rq_bulk->bd_nob == aa->aa_requested_nob);
1176 if (unlikely((aa->aa_oa->o_valid & OBD_MD_FLCKSUM) &&
1178 check_write_checksum(&body->oa, peer, client_cksum,
1180 aa->aa_requested_nob,
1185 if (sptlrpc_cli_unwrap_bulk_write(req, req->rq_bulk))
1188 rc = check_write_rcs(req, aa->aa_requested_nob,aa->aa_nio_count,
1189 aa->aa_page_count, aa->aa_ppga);
1193 /* The rest of this function executes only for OST_READs */
1194 if (rc > aa->aa_requested_nob) {
1195 CERROR("Unexpected rc %d (%d requested)\n", rc,
1196 aa->aa_requested_nob);
1200 if (rc != req->rq_bulk->bd_nob_transferred) {
1201 CERROR ("Unexpected rc %d (%d transferred)\n",
1202 rc, req->rq_bulk->bd_nob_transferred);
1206 if (rc < aa->aa_requested_nob)
1207 handle_short_read(rc, aa->aa_page_count, aa->aa_ppga);
1209 if (sptlrpc_cli_unwrap_bulk_read(req, rc, aa->aa_page_count,
1211 GOTO(out, rc = -EAGAIN);
1213 if (unlikely(body->oa.o_valid & OBD_MD_FLCKSUM)) {
1214 static int cksum_counter;
1215 __u32 server_cksum = body->oa.o_cksum;
1219 client_cksum = osc_checksum_bulk(rc, aa->aa_page_count,
1222 if (peer->nid == req->rq_bulk->bd_sender) {
1226 router = libcfs_nid2str(req->rq_bulk->bd_sender);
1229 if (server_cksum == ~0 && rc > 0) {
1230 CERROR("Protocol error: server %s set the 'checksum' "
1231 "bit, but didn't send a checksum. Not fatal, "
1232 "but please tell CFS.\n",
1233 libcfs_nid2str(peer->nid));
1234 } else if (server_cksum != client_cksum) {
1235 LCONSOLE_ERROR_MSG(0x133, "%s: BAD READ CHECKSUM: from "
1236 "%s%s%s inum "LPU64"/"LPU64" object "
1237 LPU64"/"LPU64" extent "
1238 "["LPU64"-"LPU64"]\n",
1239 req->rq_import->imp_obd->obd_name,
1240 libcfs_nid2str(peer->nid),
1242 body->oa.o_valid & OBD_MD_FLFID ?
1243 body->oa.o_fid : (__u64)0,
1244 body->oa.o_valid & OBD_MD_FLFID ?
1245 body->oa.o_generation :(__u64)0,
1247 body->oa.o_valid & OBD_MD_FLGROUP ?
1248 body->oa.o_gr : (__u64)0,
1249 aa->aa_ppga[0]->off,
1250 aa->aa_ppga[aa->aa_page_count-1]->off +
1251 aa->aa_ppga[aa->aa_page_count-1]->count -
1253 CERROR("client %x, server %x\n",
1254 client_cksum, server_cksum);
1256 aa->aa_oa->o_cksum = client_cksum;
1260 CDEBUG(D_PAGE, "checksum %x confirmed\n", client_cksum);
1263 } else if (unlikely(client_cksum)) {
1264 static int cksum_missed;
1267 if ((cksum_missed & (-cksum_missed)) == cksum_missed)
1268 CERROR("Checksum %u requested from %s but not sent\n",
1269 cksum_missed, libcfs_nid2str(peer->nid));
1275 *aa->aa_oa = body->oa;
1280 static int osc_brw_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1281 struct lov_stripe_md *lsm,
1282 obd_count page_count, struct brw_page **pga,
1283 struct obd_capa *ocapa)
1285 struct ptlrpc_request *req;
1286 int rc, retries = 5; /* lprocfs? */
1290 rc = osc_brw_prep_request(cmd, &exp->exp_obd->u.cli, oa, lsm,
1291 page_count, pga, &req, ocapa);
1295 rc = ptlrpc_queue_wait(req);
1297 if (rc == -ETIMEDOUT && req->rq_resend) {
1298 DEBUG_REQ(D_HA, req, "BULK TIMEOUT");
1299 ptlrpc_req_finished(req);
1303 rc = osc_brw_fini_request(req, rc);
1305 ptlrpc_req_finished(req);
1306 if (rc == -EAGAIN) {
1314 int osc_brw_redo_request(struct ptlrpc_request *req,
1315 struct osc_brw_async_args *aa)
1317 struct ptlrpc_request *new_req;
1318 struct ptlrpc_request_set *set = req->rq_set;
1319 struct osc_brw_async_args *new_aa;
1320 struct osc_async_page *oap;
1324 if (aa->aa_retries-- <= 0) {
1325 CERROR("too many checksum retries, returning error\n");
1329 DEBUG_REQ(D_ERROR, req, "redo for checksum error");
1330 list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
1331 if (oap->oap_request != NULL) {
1332 LASSERTF(req == oap->oap_request,
1333 "request %p != oap_request %p\n",
1334 req, oap->oap_request);
1335 if (oap->oap_interrupted) {
1336 ptlrpc_mark_interrupted(oap->oap_request);
1344 /* TODO-MERGE: and where to get ocapa?? */
1345 rc = osc_brw_prep_request(lustre_msg_get_opc(req->rq_reqmsg) ==
1346 OST_WRITE ? OBD_BRW_WRITE :OBD_BRW_READ,
1347 aa->aa_cli, aa->aa_oa,
1348 NULL /* lsm unused by osc currently */,
1349 aa->aa_page_count, aa->aa_ppga, &new_req,
1354 /* New request takes over pga and oaps from old request.
1355 * Note that copying a list_head doesn't work, need to move it... */
1356 new_req->rq_interpret_reply = req->rq_interpret_reply;
1357 new_req->rq_async_args = req->rq_async_args;
1358 new_aa = (struct osc_brw_async_args *)&new_req->rq_async_args;
1359 INIT_LIST_HEAD(&new_aa->aa_oaps);
1360 list_splice(&aa->aa_oaps, &new_aa->aa_oaps);
1361 INIT_LIST_HEAD(&aa->aa_oaps);
1363 list_for_each_entry(oap, &new_aa->aa_oaps, oap_rpc_item) {
1364 if (oap->oap_request) {
1365 ptlrpc_req_finished(oap->oap_request);
1366 oap->oap_request = ptlrpc_request_addref(new_req);
1370 ptlrpc_set_add_req(set, new_req);
1375 static int brw_interpret(struct ptlrpc_request *req, void *data, int rc)
1377 struct osc_brw_async_args *aa = data;
1382 rc = osc_brw_fini_request(req, rc);
1383 if (rc == -EAGAIN) {
1384 rc = osc_brw_redo_request(req, aa);
1388 if ((rc >= 0) && req->rq_set && req->rq_set->set_countp)
1389 atomic_add(nob, (atomic_t *)req->rq_set->set_countp);
1391 spin_lock(&aa->aa_cli->cl_loi_list_lock);
1392 for (i = 0; i < aa->aa_page_count; i++)
1393 osc_release_write_grant(aa->aa_cli, aa->aa_ppga[i], 1);
1394 spin_unlock(&aa->aa_cli->cl_loi_list_lock);
1396 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
1401 static int async_internal(int cmd, struct obd_export *exp, struct obdo *oa,
1402 struct lov_stripe_md *lsm, obd_count page_count,
1403 struct brw_page **pga, struct ptlrpc_request_set *set,
1404 struct obd_capa *ocapa)
1406 struct ptlrpc_request *req;
1407 struct client_obd *cli = &exp->exp_obd->u.cli;
1411 /* Consume write credits even if doing a sync write -
1412 * otherwise we may run out of space on OST due to grant. */
1413 if (cmd == OBD_BRW_WRITE) {
1414 spin_lock(&cli->cl_loi_list_lock);
1415 for (i = 0; i < page_count; i++) {
1416 if (cli->cl_avail_grant >= CFS_PAGE_SIZE)
1417 osc_consume_write_grant(cli, pga[i]);
1419 spin_unlock(&cli->cl_loi_list_lock);
1422 rc = osc_brw_prep_request(cmd, cli, oa, lsm, page_count, pga,
1425 req->rq_interpret_reply = brw_interpret;
1426 ptlrpc_set_add_req(set, req);
1427 } else if (cmd == OBD_BRW_WRITE) {
1428 spin_lock(&cli->cl_loi_list_lock);
1429 for (i = 0; i < page_count; i++)
1430 osc_release_write_grant(cli, pga[i], 0);
1431 spin_unlock(&cli->cl_loi_list_lock);
1437 * ugh, we want disk allocation on the target to happen in offset order. we'll
1438 * follow sedgewicks advice and stick to the dead simple shellsort -- it'll do
1439 * fine for our small page arrays and doesn't require allocation. its an
1440 * insertion sort that swaps elements that are strides apart, shrinking the
1441 * stride down until its '1' and the array is sorted.
1443 static void sort_brw_pages(struct brw_page **array, int num)
1446 struct brw_page *tmp;
1450 for (stride = 1; stride < num ; stride = (stride * 3) + 1)
1455 for (i = stride ; i < num ; i++) {
1458 while (j >= stride && array[j - stride]->off > tmp->off) {
1459 array[j] = array[j - stride];
1464 } while (stride > 1);
1467 static obd_count max_unfragmented_pages(struct brw_page **pg, obd_count pages)
1473 LASSERT (pages > 0);
1474 offset = pg[i]->off & ~CFS_PAGE_MASK;
1478 if (pages == 0) /* that's all */
1481 if (offset + pg[i]->count < CFS_PAGE_SIZE)
1482 return count; /* doesn't end on page boundary */
1485 offset = pg[i]->off & ~CFS_PAGE_MASK;
1486 if (offset != 0) /* doesn't start on page boundary */
1493 static struct brw_page **osc_build_ppga(struct brw_page *pga, obd_count count)
1495 struct brw_page **ppga;
1498 OBD_ALLOC(ppga, sizeof(*ppga) * count);
1502 for (i = 0; i < count; i++)
1507 static void osc_release_ppga(struct brw_page **ppga, obd_count count)
1509 LASSERT(ppga != NULL);
1510 OBD_FREE(ppga, sizeof(*ppga) * count);
1513 static int osc_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
1514 obd_count page_count, struct brw_page *pga,
1515 struct obd_trans_info *oti)
1517 struct obdo *saved_oa = NULL;
1518 struct brw_page **ppga, **orig;
1519 struct obd_import *imp = class_exp2cliimp(exp);
1520 struct client_obd *cli = &imp->imp_obd->u.cli;
1521 int rc, page_count_orig;
1524 if (cmd & OBD_BRW_CHECK) {
1525 /* The caller just wants to know if there's a chance that this
1526 * I/O can succeed */
1528 if (imp == NULL || imp->imp_invalid)
1533 /* test_brw with a failed create can trip this, maybe others. */
1534 LASSERT(cli->cl_max_pages_per_rpc);
1538 orig = ppga = osc_build_ppga(pga, page_count);
1541 page_count_orig = page_count;
1543 sort_brw_pages(ppga, page_count);
1544 while (page_count) {
1545 obd_count pages_per_brw;
1547 if (page_count > cli->cl_max_pages_per_rpc)
1548 pages_per_brw = cli->cl_max_pages_per_rpc;
1550 pages_per_brw = page_count;
1552 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1554 if (saved_oa != NULL) {
1555 /* restore previously saved oa */
1556 *oinfo->oi_oa = *saved_oa;
1557 } else if (page_count > pages_per_brw) {
1558 /* save a copy of oa (brw will clobber it) */
1559 OBDO_ALLOC(saved_oa);
1560 if (saved_oa == NULL)
1561 GOTO(out, rc = -ENOMEM);
1562 *saved_oa = *oinfo->oi_oa;
1565 rc = osc_brw_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1566 pages_per_brw, ppga, oinfo->oi_capa);
1571 page_count -= pages_per_brw;
1572 ppga += pages_per_brw;
1576 osc_release_ppga(orig, page_count_orig);
1578 if (saved_oa != NULL)
1579 OBDO_FREE(saved_oa);
1584 static int osc_brw_async(int cmd, struct obd_export *exp,
1585 struct obd_info *oinfo, obd_count page_count,
1586 struct brw_page *pga, struct obd_trans_info *oti,
1587 struct ptlrpc_request_set *set)
1589 struct brw_page **ppga, **orig;
1590 struct client_obd *cli = &exp->exp_obd->u.cli;
1591 int page_count_orig;
1595 if (cmd & OBD_BRW_CHECK) {
1596 struct obd_import *imp = class_exp2cliimp(exp);
1597 /* The caller just wants to know if there's a chance that this
1598 * I/O can succeed */
1600 if (imp == NULL || imp->imp_invalid)
1605 orig = ppga = osc_build_ppga(pga, page_count);
1608 page_count_orig = page_count;
1610 sort_brw_pages(ppga, page_count);
1611 while (page_count) {
1612 struct brw_page **copy;
1613 obd_count pages_per_brw;
1615 pages_per_brw = min_t(obd_count, page_count,
1616 cli->cl_max_pages_per_rpc);
1618 pages_per_brw = max_unfragmented_pages(ppga, pages_per_brw);
1620 /* use ppga only if single RPC is going to fly */
1621 if (pages_per_brw != page_count_orig || ppga != orig) {
1622 OBD_ALLOC(copy, sizeof(*copy) * pages_per_brw);
1624 GOTO(out, rc = -ENOMEM);
1625 memcpy(copy, ppga, sizeof(*copy) * pages_per_brw);
1629 rc = async_internal(cmd, exp, oinfo->oi_oa, oinfo->oi_md,
1630 pages_per_brw, copy, set, oinfo->oi_capa);
1634 OBD_FREE(copy, sizeof(*copy) * pages_per_brw);
1638 /* we passed it to async_internal() which is
1639 * now responsible for releasing memory */
1643 page_count -= pages_per_brw;
1644 ppga += pages_per_brw;
1648 osc_release_ppga(orig, page_count_orig);
1652 static void osc_check_rpcs(struct client_obd *cli);
1654 /* The companion to osc_enter_cache(), called when @oap is no longer part of
1655 * the dirty accounting. Writeback completes or truncate happens before
1656 * writing starts. Must be called with the loi lock held. */
1657 static void osc_exit_cache(struct client_obd *cli, struct osc_async_page *oap,
1660 osc_release_write_grant(cli, &oap->oap_brw_page, sent);
1664 /* This maintains the lists of pending pages to read/write for a given object
1665 * (lop). This is used by osc_check_rpcs->osc_next_loi() and loi_list_maint()
1666 * to quickly find objects that are ready to send an RPC. */
1667 static int lop_makes_rpc(struct client_obd *cli, struct loi_oap_pages *lop,
1673 if (lop->lop_num_pending == 0)
1676 /* if we have an invalid import we want to drain the queued pages
1677 * by forcing them through rpcs that immediately fail and complete
1678 * the pages. recovery relies on this to empty the queued pages
1679 * before canceling the locks and evicting down the llite pages */
1680 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
1683 /* stream rpcs in queue order as long as as there is an urgent page
1684 * queued. this is our cheap solution for good batching in the case
1685 * where writepage marks some random page in the middle of the file
1686 * as urgent because of, say, memory pressure */
1687 if (!list_empty(&lop->lop_urgent)) {
1688 CDEBUG(D_CACHE, "urgent request forcing RPC\n");
1691 /* fire off rpcs when we have 'optimal' rpcs as tuned for the wire. */
1692 optimal = cli->cl_max_pages_per_rpc;
1693 if (cmd & OBD_BRW_WRITE) {
1694 /* trigger a write rpc stream as long as there are dirtiers
1695 * waiting for space. as they're waiting, they're not going to
1696 * create more pages to coallesce with what's waiting.. */
1697 if (!list_empty(&cli->cl_cache_waiters)) {
1698 CDEBUG(D_CACHE, "cache waiters forcing RPC\n");
1701 /* +16 to avoid triggering rpcs that would want to include pages
1702 * that are being queued but which can't be made ready until
1703 * the queuer finishes with the page. this is a wart for
1704 * llite::commit_write() */
1707 if (lop->lop_num_pending >= optimal)
1713 static void on_list(struct list_head *item, struct list_head *list,
1716 if (list_empty(item) && should_be_on)
1717 list_add_tail(item, list);
1718 else if (!list_empty(item) && !should_be_on)
1719 list_del_init(item);
1722 /* maintain the loi's cli list membership invariants so that osc_send_oap_rpc
1723 * can find pages to build into rpcs quickly */
1724 static void loi_list_maint(struct client_obd *cli, struct lov_oinfo *loi)
1726 on_list(&loi->loi_cli_item, &cli->cl_loi_ready_list,
1727 lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE) ||
1728 lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ));
1730 on_list(&loi->loi_write_item, &cli->cl_loi_write_list,
1731 loi->loi_write_lop.lop_num_pending);
1733 on_list(&loi->loi_read_item, &cli->cl_loi_read_list,
1734 loi->loi_read_lop.lop_num_pending);
1737 static void lop_update_pending(struct client_obd *cli,
1738 struct loi_oap_pages *lop, int cmd, int delta)
1740 lop->lop_num_pending += delta;
1741 if (cmd & OBD_BRW_WRITE)
1742 cli->cl_pending_w_pages += delta;
1744 cli->cl_pending_r_pages += delta;
1747 /* this is called when a sync waiter receives an interruption. Its job is to
1748 * get the caller woken as soon as possible. If its page hasn't been put in an
1749 * rpc yet it can dequeue immediately. Otherwise it has to mark the rpc as
1750 * desiring interruption which will forcefully complete the rpc once the rpc
1752 static void osc_occ_interrupted(struct oig_callback_context *occ)
1754 struct osc_async_page *oap;
1755 struct loi_oap_pages *lop;
1756 struct lov_oinfo *loi;
1759 /* XXX member_of() */
1760 oap = list_entry(occ, struct osc_async_page, oap_occ);
1762 client_obd_list_lock(&oap->oap_cli->cl_loi_list_lock);
1764 oap->oap_interrupted = 1;
1766 /* ok, it's been put in an rpc. only one oap gets a request reference */
1767 if (oap->oap_request != NULL) {
1768 ptlrpc_mark_interrupted(oap->oap_request);
1769 ptlrpcd_wake(oap->oap_request);
1773 /* we don't get interruption callbacks until osc_trigger_group_io()
1774 * has been called and put the sync oaps in the pending/urgent lists.*/
1775 if (!list_empty(&oap->oap_pending_item)) {
1776 list_del_init(&oap->oap_pending_item);
1777 list_del_init(&oap->oap_urgent_item);
1780 lop = (oap->oap_cmd & OBD_BRW_WRITE) ?
1781 &loi->loi_write_lop : &loi->loi_read_lop;
1782 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, -1);
1783 loi_list_maint(oap->oap_cli, oap->oap_loi);
1785 oig_complete_one(oap->oap_oig, &oap->oap_occ, -EINTR);
1786 oap->oap_oig = NULL;
1790 client_obd_list_unlock(&oap->oap_cli->cl_loi_list_lock);
1793 /* this is trying to propogate async writeback errors back up to the
1794 * application. As an async write fails we record the error code for later if
1795 * the app does an fsync. As long as errors persist we force future rpcs to be
1796 * sync so that the app can get a sync error and break the cycle of queueing
1797 * pages for which writeback will fail. */
1798 static void osc_process_ar(struct osc_async_rc *ar, struct ptlrpc_request *req,
1805 ar->ar_force_sync = 1;
1806 ar->ar_min_xid = ptlrpc_sample_next_xid();
1811 if (ar->ar_force_sync && req && (ptlrpc_req_xid(req) >= ar->ar_min_xid))
1812 ar->ar_force_sync = 0;
1815 static void osc_oap_to_pending(struct osc_async_page *oap)
1817 struct loi_oap_pages *lop;
1819 if (oap->oap_cmd & OBD_BRW_WRITE)
1820 lop = &oap->oap_loi->loi_write_lop;
1822 lop = &oap->oap_loi->loi_read_lop;
1824 if (oap->oap_async_flags & ASYNC_URGENT)
1825 list_add(&oap->oap_urgent_item, &lop->lop_urgent);
1826 list_add_tail(&oap->oap_pending_item, &lop->lop_pending);
1827 lop_update_pending(oap->oap_cli, lop, oap->oap_cmd, 1);
1830 /* this must be called holding the loi list lock to give coverage to exit_cache,
1831 * async_flag maintenance, and oap_request */
1832 static void osc_ap_completion(struct client_obd *cli, struct obdo *oa,
1833 struct osc_async_page *oap, int sent, int rc)
1836 oap->oap_async_flags = 0;
1837 oap->oap_interrupted = 0;
1839 if (oap->oap_cmd & OBD_BRW_WRITE) {
1840 osc_process_ar(&cli->cl_ar, oap->oap_request, rc);
1841 osc_process_ar(&oap->oap_loi->loi_ar, oap->oap_request, rc);
1844 if (oap->oap_request != NULL) {
1845 ptlrpc_req_finished(oap->oap_request);
1846 oap->oap_request = NULL;
1849 if (rc == 0 && oa != NULL) {
1850 if (oa->o_valid & OBD_MD_FLBLOCKS)
1851 oap->oap_loi->loi_lvb.lvb_blocks = oa->o_blocks;
1852 if (oa->o_valid & OBD_MD_FLMTIME)
1853 oap->oap_loi->loi_lvb.lvb_mtime = oa->o_mtime;
1854 if (oa->o_valid & OBD_MD_FLATIME)
1855 oap->oap_loi->loi_lvb.lvb_atime = oa->o_atime;
1856 if (oa->o_valid & OBD_MD_FLCTIME)
1857 oap->oap_loi->loi_lvb.lvb_ctime = oa->o_ctime;
1861 osc_exit_cache(cli, oap, sent);
1862 oig_complete_one(oap->oap_oig, &oap->oap_occ, rc);
1863 oap->oap_oig = NULL;
1868 rc = oap->oap_caller_ops->ap_completion(oap->oap_caller_data,
1869 oap->oap_cmd, oa, rc);
1871 /* ll_ap_completion (from llite) drops PG_locked. so, a new
1872 * I/O on the page could start, but OSC calls it under lock
1873 * and thus we can add oap back to pending safely */
1875 /* upper layer wants to leave the page on pending queue */
1876 osc_oap_to_pending(oap);
1878 osc_exit_cache(cli, oap, sent);
1882 static int brw_interpret_oap(struct ptlrpc_request *req, void *data, int rc)
1884 struct osc_async_page *oap, *tmp;
1885 struct osc_brw_async_args *aa = data;
1886 struct client_obd *cli;
1889 rc = osc_brw_fini_request(req, rc);
1890 CDEBUG(D_INODE, "request %p aa %p rc %d\n", req, aa, rc);
1891 if (rc == -EAGAIN) {
1892 rc = osc_brw_redo_request(req, aa);
1900 client_obd_list_lock(&cli->cl_loi_list_lock);
1902 /* We need to decrement before osc_ap_completion->osc_wake_cache_waiters
1903 * is called so we know whether to go to sync BRWs or wait for more
1904 * RPCs to complete */
1905 if (lustre_msg_get_opc(req->rq_reqmsg) == OST_WRITE)
1906 cli->cl_w_in_flight--;
1908 cli->cl_r_in_flight--;
1910 /* the caller may re-use the oap after the completion call so
1911 * we need to clean it up a little */
1912 list_for_each_entry_safe(oap, tmp, &aa->aa_oaps, oap_rpc_item) {
1913 list_del_init(&oap->oap_rpc_item);
1914 osc_ap_completion(cli, aa->aa_oa, oap, 1, rc);
1917 osc_wake_cache_waiters(cli);
1918 osc_check_rpcs(cli);
1920 client_obd_list_unlock(&cli->cl_loi_list_lock);
1922 OBDO_FREE(aa->aa_oa);
1925 osc_release_ppga(aa->aa_ppga, aa->aa_page_count);
1929 static struct ptlrpc_request *osc_build_req(struct client_obd *cli,
1930 struct list_head *rpc_list,
1931 int page_count, int cmd)
1933 struct ptlrpc_request *req;
1934 struct brw_page **pga = NULL;
1935 struct osc_brw_async_args *aa;
1936 struct obdo *oa = NULL;
1937 struct obd_async_page_ops *ops = NULL;
1938 void *caller_data = NULL;
1939 struct obd_capa *ocapa;
1940 struct osc_async_page *oap;
1944 LASSERT(!list_empty(rpc_list));
1946 OBD_ALLOC(pga, sizeof(*pga) * page_count);
1948 RETURN(ERR_PTR(-ENOMEM));
1952 GOTO(out, req = ERR_PTR(-ENOMEM));
1955 list_for_each_entry(oap, rpc_list, oap_rpc_item) {
1957 ops = oap->oap_caller_ops;
1958 caller_data = oap->oap_caller_data;
1960 pga[i] = &oap->oap_brw_page;
1961 pga[i]->off = oap->oap_obj_off + oap->oap_page_off;
1962 CDEBUG(0, "put page %p index %lu oap %p flg %x to pga\n",
1963 pga[i]->pg, cfs_page_index(oap->oap_page), oap, pga[i]->flag);
1967 /* always get the data for the obdo for the rpc */
1968 LASSERT(ops != NULL);
1969 ops->ap_fill_obdo(caller_data, cmd, oa);
1970 ocapa = ops->ap_lookup_capa(caller_data, cmd);
1972 sort_brw_pages(pga, page_count);
1973 rc = osc_brw_prep_request(cmd, cli, oa, NULL, page_count,
1977 CERROR("prep_req failed: %d\n", rc);
1978 GOTO(out, req = ERR_PTR(rc));
1981 /* Need to update the timestamps after the request is built in case
1982 * we race with setattr (locally or in queue at OST). If OST gets
1983 * later setattr before earlier BRW (as determined by the request xid),
1984 * the OST will not use BRW timestamps. Sadly, there is no obvious
1985 * way to do this in a single call. bug 10150 */
1986 ops->ap_update_obdo(caller_data, cmd, oa,
1987 OBD_MD_FLMTIME | OBD_MD_FLCTIME | OBD_MD_FLATIME);
1989 CLASSERT(sizeof(*aa) <= sizeof(req->rq_async_args));
1990 aa = (struct osc_brw_async_args *)&req->rq_async_args;
1991 INIT_LIST_HEAD(&aa->aa_oaps);
1992 list_splice(rpc_list, &aa->aa_oaps);
1993 INIT_LIST_HEAD(rpc_list);
2000 OBD_FREE(pga, sizeof(*pga) * page_count);
2005 /* the loi lock is held across this function but it's allowed to release
2006 * and reacquire it during its work */
2007 static int osc_send_oap_rpc(struct client_obd *cli, struct lov_oinfo *loi,
2008 int cmd, struct loi_oap_pages *lop)
2010 struct ptlrpc_request *req;
2011 obd_count page_count = 0;
2012 struct osc_async_page *oap = NULL, *tmp;
2013 struct osc_brw_async_args *aa;
2014 struct obd_async_page_ops *ops;
2015 CFS_LIST_HEAD(rpc_list);
2016 unsigned int ending_offset;
2017 unsigned starting_offset = 0;
2020 /* first we find the pages we're allowed to work with */
2021 list_for_each_entry_safe(oap, tmp, &lop->lop_pending,
2023 ops = oap->oap_caller_ops;
2025 LASSERT(oap->oap_magic == OAP_MAGIC);
2027 /* in llite being 'ready' equates to the page being locked
2028 * until completion unlocks it. commit_write submits a page
2029 * as not ready because its unlock will happen unconditionally
2030 * as the call returns. if we race with commit_write giving
2031 * us that page we dont' want to create a hole in the page
2032 * stream, so we stop and leave the rpc to be fired by
2033 * another dirtier or kupdated interval (the not ready page
2034 * will still be on the dirty list). we could call in
2035 * at the end of ll_file_write to process the queue again. */
2036 if (!(oap->oap_async_flags & ASYNC_READY)) {
2037 int rc = ops->ap_make_ready(oap->oap_caller_data, cmd);
2039 CDEBUG(D_INODE, "oap %p page %p returned %d "
2040 "instead of ready\n", oap,
2044 /* llite is telling us that the page is still
2045 * in commit_write and that we should try
2046 * and put it in an rpc again later. we
2047 * break out of the loop so we don't create
2048 * a hole in the sequence of pages in the rpc
2053 /* the io isn't needed.. tell the checks
2054 * below to complete the rpc with EINTR */
2055 oap->oap_async_flags |= ASYNC_COUNT_STABLE;
2056 oap->oap_count = -EINTR;
2059 oap->oap_async_flags |= ASYNC_READY;
2062 LASSERTF(0, "oap %p page %p returned %d "
2063 "from make_ready\n", oap,
2071 * Page submitted for IO has to be locked. Either by
2072 * ->ap_make_ready() or by higher layers.
2074 * XXX nikita: this assertion should be adjusted when lustre
2075 * starts using PG_writeback for pages being written out.
2077 #if defined(__KERNEL__) && defined(__LINUX__)
2078 LASSERT(PageLocked(oap->oap_page));
2080 /* If there is a gap at the start of this page, it can't merge
2081 * with any previous page, so we'll hand the network a
2082 * "fragmented" page array that it can't transfer in 1 RDMA */
2083 if (page_count != 0 && oap->oap_page_off != 0)
2086 /* take the page out of our book-keeping */
2087 list_del_init(&oap->oap_pending_item);
2088 lop_update_pending(cli, lop, cmd, -1);
2089 list_del_init(&oap->oap_urgent_item);
2091 if (page_count == 0)
2092 starting_offset = (oap->oap_obj_off+oap->oap_page_off) &
2093 (PTLRPC_MAX_BRW_SIZE - 1);
2095 /* ask the caller for the size of the io as the rpc leaves. */
2096 if (!(oap->oap_async_flags & ASYNC_COUNT_STABLE))
2098 ops->ap_refresh_count(oap->oap_caller_data,cmd);
2099 if (oap->oap_count <= 0) {
2100 CDEBUG(D_CACHE, "oap %p count %d, completing\n", oap,
2102 osc_ap_completion(cli, NULL, oap, 0, oap->oap_count);
2106 /* now put the page back in our accounting */
2107 list_add_tail(&oap->oap_rpc_item, &rpc_list);
2108 if (++page_count >= cli->cl_max_pages_per_rpc)
2111 /* End on a PTLRPC_MAX_BRW_SIZE boundary. We want full-sized
2112 * RPCs aligned on PTLRPC_MAX_BRW_SIZE boundaries to help reads
2113 * have the same alignment as the initial writes that allocated
2114 * extents on the server. */
2115 ending_offset = (oap->oap_obj_off + oap->oap_page_off +
2116 oap->oap_count) & (PTLRPC_MAX_BRW_SIZE - 1);
2117 if (ending_offset == 0)
2120 /* If there is a gap at the end of this page, it can't merge
2121 * with any subsequent pages, so we'll hand the network a
2122 * "fragmented" page array that it can't transfer in 1 RDMA */
2123 if (oap->oap_page_off + oap->oap_count < CFS_PAGE_SIZE)
2127 osc_wake_cache_waiters(cli);
2129 if (page_count == 0)
2132 loi_list_maint(cli, loi);
2134 client_obd_list_unlock(&cli->cl_loi_list_lock);
2136 req = osc_build_req(cli, &rpc_list, page_count, cmd);
2138 /* this should happen rarely and is pretty bad, it makes the
2139 * pending list not follow the dirty order */
2140 client_obd_list_lock(&cli->cl_loi_list_lock);
2141 list_for_each_entry_safe(oap, tmp, &rpc_list, oap_rpc_item) {
2142 list_del_init(&oap->oap_rpc_item);
2144 /* queued sync pages can be torn down while the pages
2145 * were between the pending list and the rpc */
2146 if (oap->oap_interrupted) {
2147 CDEBUG(D_INODE, "oap %p interrupted\n", oap);
2148 osc_ap_completion(cli, NULL, oap, 0,
2152 osc_ap_completion(cli, NULL, oap, 0, PTR_ERR(req));
2154 loi_list_maint(cli, loi);
2155 RETURN(PTR_ERR(req));
2158 aa = (struct osc_brw_async_args *)&req->rq_async_args;
2160 if (cmd == OBD_BRW_READ) {
2161 lprocfs_oh_tally_log2(&cli->cl_read_page_hist, page_count);
2162 lprocfs_oh_tally(&cli->cl_read_rpc_hist, cli->cl_r_in_flight);
2163 lprocfs_oh_tally_log2(&cli->cl_read_offset_hist,
2164 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2165 ptlrpc_lprocfs_brw(req, OST_READ, aa->aa_requested_nob);
2167 lprocfs_oh_tally_log2(&cli->cl_write_page_hist, page_count);
2168 lprocfs_oh_tally(&cli->cl_write_rpc_hist,
2169 cli->cl_w_in_flight);
2170 lprocfs_oh_tally_log2(&cli->cl_write_offset_hist,
2171 (starting_offset >> CFS_PAGE_SHIFT) + 1);
2172 ptlrpc_lprocfs_brw(req, OST_WRITE, aa->aa_requested_nob);
2175 client_obd_list_lock(&cli->cl_loi_list_lock);
2177 if (cmd == OBD_BRW_READ)
2178 cli->cl_r_in_flight++;
2180 cli->cl_w_in_flight++;
2182 /* queued sync pages can be torn down while the pages
2183 * were between the pending list and the rpc */
2185 list_for_each_entry(oap, &aa->aa_oaps, oap_rpc_item) {
2186 /* only one oap gets a request reference */
2189 if (oap->oap_interrupted && !req->rq_intr) {
2190 CDEBUG(D_INODE, "oap %p in req %p interrupted\n",
2192 ptlrpc_mark_interrupted(req);
2196 tmp->oap_request = ptlrpc_request_addref(req);
2198 DEBUG_REQ(D_INODE, req, "%d pages, aa %p. now %dr/%dw in flight",
2199 page_count, aa, cli->cl_r_in_flight, cli->cl_w_in_flight);
2201 req->rq_interpret_reply = brw_interpret_oap;
2202 ptlrpcd_add_req(req);
2206 #define LOI_DEBUG(LOI, STR, args...) \
2207 CDEBUG(D_INODE, "loi ready %d wr %d:%d rd %d:%d " STR, \
2208 !list_empty(&(LOI)->loi_cli_item), \
2209 (LOI)->loi_write_lop.lop_num_pending, \
2210 !list_empty(&(LOI)->loi_write_lop.lop_urgent), \
2211 (LOI)->loi_read_lop.lop_num_pending, \
2212 !list_empty(&(LOI)->loi_read_lop.lop_urgent), \
2215 /* This is called by osc_check_rpcs() to find which objects have pages that
2216 * we could be sending. These lists are maintained by lop_makes_rpc(). */
2217 struct lov_oinfo *osc_next_loi(struct client_obd *cli)
2220 /* first return all objects which we already know to have
2221 * pages ready to be stuffed into rpcs */
2222 if (!list_empty(&cli->cl_loi_ready_list))
2223 RETURN(list_entry(cli->cl_loi_ready_list.next,
2224 struct lov_oinfo, loi_cli_item));
2226 /* then if we have cache waiters, return all objects with queued
2227 * writes. This is especially important when many small files
2228 * have filled up the cache and not been fired into rpcs because
2229 * they don't pass the nr_pending/object threshhold */
2230 if (!list_empty(&cli->cl_cache_waiters) &&
2231 !list_empty(&cli->cl_loi_write_list))
2232 RETURN(list_entry(cli->cl_loi_write_list.next,
2233 struct lov_oinfo, loi_write_item));
2235 /* then return all queued objects when we have an invalid import
2236 * so that they get flushed */
2237 if (cli->cl_import == NULL || cli->cl_import->imp_invalid) {
2238 if (!list_empty(&cli->cl_loi_write_list))
2239 RETURN(list_entry(cli->cl_loi_write_list.next,
2240 struct lov_oinfo, loi_write_item));
2241 if (!list_empty(&cli->cl_loi_read_list))
2242 RETURN(list_entry(cli->cl_loi_read_list.next,
2243 struct lov_oinfo, loi_read_item));
2248 /* called with the loi list lock held */
2249 static void osc_check_rpcs(struct client_obd *cli)
2251 struct lov_oinfo *loi;
2252 int rc = 0, race_counter = 0;
2255 while ((loi = osc_next_loi(cli)) != NULL) {
2256 LOI_DEBUG(loi, "%lu in flight\n", rpcs_in_flight(cli));
2258 if (rpcs_in_flight(cli) >= cli->cl_max_rpcs_in_flight)
2261 /* attempt some read/write balancing by alternating between
2262 * reads and writes in an object. The makes_rpc checks here
2263 * would be redundant if we were getting read/write work items
2264 * instead of objects. we don't want send_oap_rpc to drain a
2265 * partial read pending queue when we're given this object to
2266 * do io on writes while there are cache waiters */
2267 if (lop_makes_rpc(cli, &loi->loi_write_lop, OBD_BRW_WRITE)) {
2268 rc = osc_send_oap_rpc(cli, loi, OBD_BRW_WRITE,
2269 &loi->loi_write_lop);
2277 if (lop_makes_rpc(cli, &loi->loi_read_lop, OBD_BRW_READ)) {
2278 rc = osc_send_oap_rpc(cli, loi, OBD_BRW_READ,
2279 &loi->loi_read_lop);
2288 /* attempt some inter-object balancing by issueing rpcs
2289 * for each object in turn */
2290 if (!list_empty(&loi->loi_cli_item))
2291 list_del_init(&loi->loi_cli_item);
2292 if (!list_empty(&loi->loi_write_item))
2293 list_del_init(&loi->loi_write_item);
2294 if (!list_empty(&loi->loi_read_item))
2295 list_del_init(&loi->loi_read_item);
2297 loi_list_maint(cli, loi);
2299 /* send_oap_rpc fails with 0 when make_ready tells it to
2300 * back off. llite's make_ready does this when it tries
2301 * to lock a page queued for write that is already locked.
2302 * we want to try sending rpcs from many objects, but we
2303 * don't want to spin failing with 0. */
2304 if (race_counter == 10)
2310 /* we're trying to queue a page in the osc so we're subject to the
2311 * 'cl_dirty_max' limit on the number of pages that can be queued in the osc.
2312 * If the osc's queued pages are already at that limit, then we want to sleep
2313 * until there is space in the osc's queue for us. We also may be waiting for
2314 * write credits from the OST if there are RPCs in flight that may return some
2315 * before we fall back to sync writes.
2317 * We need this know our allocation was granted in the presence of signals */
2318 static int ocw_granted(struct client_obd *cli, struct osc_cache_waiter *ocw)
2322 client_obd_list_lock(&cli->cl_loi_list_lock);
2323 rc = list_empty(&ocw->ocw_entry) || rpcs_in_flight(cli) == 0;
2324 client_obd_list_unlock(&cli->cl_loi_list_lock);
2328 /* Caller must hold loi_list_lock - we drop/regain it if we need to wait for
2329 * grant or cache space. */
2330 static int osc_enter_cache(struct client_obd *cli, struct lov_oinfo *loi,
2331 struct osc_async_page *oap)
2333 struct osc_cache_waiter ocw;
2334 struct l_wait_info lwi = { 0 };
2338 CDEBUG(D_CACHE, "dirty: %ld/%d dirty_max: %ld/%d dropped: %lu "
2339 "grant: %lu\n", cli->cl_dirty, atomic_read(&obd_dirty_pages),
2340 cli->cl_dirty_max, obd_max_dirty_pages,
2341 cli->cl_lost_grant, cli->cl_avail_grant);
2343 /* force the caller to try sync io. this can jump the list
2344 * of queued writes and create a discontiguous rpc stream */
2345 if (cli->cl_dirty_max < CFS_PAGE_SIZE || cli->cl_ar.ar_force_sync ||
2346 loi->loi_ar.ar_force_sync)
2349 /* Hopefully normal case - cache space and write credits available */
2350 if ((cli->cl_dirty + CFS_PAGE_SIZE <= cli->cl_dirty_max) &&
2351 (atomic_read(&obd_dirty_pages) + 1 <= obd_max_dirty_pages) &&
2352 (cli->cl_avail_grant >= CFS_PAGE_SIZE)) {
2353 /* account for ourselves */
2354 osc_consume_write_grant(cli, &oap->oap_brw_page);
2358 /* Make sure that there are write rpcs in flight to wait for. This
2359 * is a little silly as this object may not have any pending but
2360 * other objects sure might. */
2361 if (cli->cl_w_in_flight) {
2362 list_add_tail(&ocw.ocw_entry, &cli->cl_cache_waiters);
2363 cfs_waitq_init(&ocw.ocw_waitq);
2367 loi_list_maint(cli, loi);
2368 osc_check_rpcs(cli);
2369 client_obd_list_unlock(&cli->cl_loi_list_lock);
2371 CDEBUG(D_CACHE, "sleeping for cache space\n");
2372 l_wait_event(ocw.ocw_waitq, ocw_granted(cli, &ocw), &lwi);
2374 client_obd_list_lock(&cli->cl_loi_list_lock);
2375 if (!list_empty(&ocw.ocw_entry)) {
2376 list_del(&ocw.ocw_entry);
2385 int osc_prep_async_page(struct obd_export *exp, struct lov_stripe_md *lsm,
2386 struct lov_oinfo *loi, cfs_page_t *page,
2387 obd_off offset, struct obd_async_page_ops *ops,
2388 void *data, void **res)
2390 struct osc_async_page *oap;
2394 return size_round(sizeof(*oap));
2397 oap->oap_magic = OAP_MAGIC;
2398 oap->oap_cli = &exp->exp_obd->u.cli;
2401 oap->oap_caller_ops = ops;
2402 oap->oap_caller_data = data;
2404 oap->oap_page = page;
2405 oap->oap_obj_off = offset;
2407 CFS_INIT_LIST_HEAD(&oap->oap_pending_item);
2408 CFS_INIT_LIST_HEAD(&oap->oap_urgent_item);
2409 CFS_INIT_LIST_HEAD(&oap->oap_rpc_item);
2411 oap->oap_occ.occ_interrupted = osc_occ_interrupted;
2413 CDEBUG(D_CACHE, "oap %p page %p obj off "LPU64"\n", oap, page, offset);
2417 struct osc_async_page *oap_from_cookie(void *cookie)
2419 struct osc_async_page *oap = cookie;
2420 if (oap->oap_magic != OAP_MAGIC)
2421 return ERR_PTR(-EINVAL);
2425 static int osc_queue_async_io(struct obd_export *exp, struct lov_stripe_md *lsm,
2426 struct lov_oinfo *loi, void *cookie,
2427 int cmd, obd_off off, int count,
2428 obd_flag brw_flags, enum async_flags async_flags)
2430 struct client_obd *cli = &exp->exp_obd->u.cli;
2431 struct osc_async_page *oap;
2435 oap = oap_from_cookie(cookie);
2437 RETURN(PTR_ERR(oap));
2439 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2442 if (!list_empty(&oap->oap_pending_item) ||
2443 !list_empty(&oap->oap_urgent_item) ||
2444 !list_empty(&oap->oap_rpc_item))
2447 /* check if the file's owner/group is over quota */
2448 #ifdef HAVE_QUOTA_SUPPORT
2449 if ((cmd & OBD_BRW_WRITE) && !(cmd & OBD_BRW_NOQUOTA)){
2450 struct obd_async_page_ops *ops;
2457 ops = oap->oap_caller_ops;
2458 ops->ap_fill_obdo(oap->oap_caller_data, cmd, oa);
2459 if (lquota_chkdq(quota_interface, cli, oa->o_uid, oa->o_gid) ==
2470 loi = lsm->lsm_oinfo[0];
2472 client_obd_list_lock(&cli->cl_loi_list_lock);
2475 oap->oap_page_off = off;
2476 oap->oap_count = count;
2477 oap->oap_brw_flags = brw_flags;
2478 oap->oap_async_flags = async_flags;
2480 if (cmd & OBD_BRW_WRITE) {
2481 rc = osc_enter_cache(cli, loi, oap);
2483 client_obd_list_unlock(&cli->cl_loi_list_lock);
2488 osc_oap_to_pending(oap);
2489 loi_list_maint(cli, loi);
2491 LOI_DEBUG(loi, "oap %p page %p added for cmd %d\n", oap, oap->oap_page,
2494 osc_check_rpcs(cli);
2495 client_obd_list_unlock(&cli->cl_loi_list_lock);
2500 /* aka (~was & now & flag), but this is more clear :) */
2501 #define SETTING(was, now, flag) (!(was & flag) && (now & flag))
2503 static int osc_set_async_flags(struct obd_export *exp,
2504 struct lov_stripe_md *lsm,
2505 struct lov_oinfo *loi, void *cookie,
2506 obd_flag async_flags)
2508 struct client_obd *cli = &exp->exp_obd->u.cli;
2509 struct loi_oap_pages *lop;
2510 struct osc_async_page *oap;
2514 oap = oap_from_cookie(cookie);
2516 RETURN(PTR_ERR(oap));
2519 * bug 7311: OST-side locking is only supported for liblustre for now
2520 * (and liblustre never calls obd_set_async_flags(). I hope.), generic
2521 * implementation has to handle case where OST-locked page was picked
2522 * up by, e.g., ->writepage().
2524 LASSERT(!(oap->oap_brw_flags & OBD_BRW_SRVLOCK));
2525 LASSERT(!LIBLUSTRE_CLIENT); /* check that liblustre angels do fear to
2528 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2532 loi = lsm->lsm_oinfo[0];
2534 if (oap->oap_cmd & OBD_BRW_WRITE) {
2535 lop = &loi->loi_write_lop;
2537 lop = &loi->loi_read_lop;
2540 client_obd_list_lock(&cli->cl_loi_list_lock);
2542 if (list_empty(&oap->oap_pending_item))
2543 GOTO(out, rc = -EINVAL);
2545 if ((oap->oap_async_flags & async_flags) == async_flags)
2548 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_READY))
2549 oap->oap_async_flags |= ASYNC_READY;
2551 if (SETTING(oap->oap_async_flags, async_flags, ASYNC_URGENT)) {
2552 if (list_empty(&oap->oap_rpc_item)) {
2553 list_add(&oap->oap_urgent_item, &lop->lop_urgent);
2554 loi_list_maint(cli, loi);
2558 LOI_DEBUG(loi, "oap %p page %p has flags %x\n", oap, oap->oap_page,
2559 oap->oap_async_flags);
2561 osc_check_rpcs(cli);
2562 client_obd_list_unlock(&cli->cl_loi_list_lock);
2566 static int osc_queue_group_io(struct obd_export *exp, struct lov_stripe_md *lsm,
2567 struct lov_oinfo *loi,
2568 struct obd_io_group *oig, void *cookie,
2569 int cmd, obd_off off, int count,
2571 obd_flag async_flags)
2573 struct client_obd *cli = &exp->exp_obd->u.cli;
2574 struct osc_async_page *oap;
2575 struct loi_oap_pages *lop;
2579 oap = oap_from_cookie(cookie);
2581 RETURN(PTR_ERR(oap));
2583 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2586 if (!list_empty(&oap->oap_pending_item) ||
2587 !list_empty(&oap->oap_urgent_item) ||
2588 !list_empty(&oap->oap_rpc_item))
2592 loi = lsm->lsm_oinfo[0];
2594 client_obd_list_lock(&cli->cl_loi_list_lock);
2597 oap->oap_page_off = off;
2598 oap->oap_count = count;
2599 oap->oap_brw_flags = brw_flags;
2600 oap->oap_async_flags = async_flags;
2602 if (cmd & OBD_BRW_WRITE)
2603 lop = &loi->loi_write_lop;
2605 lop = &loi->loi_read_lop;
2607 list_add_tail(&oap->oap_pending_item, &lop->lop_pending_group);
2608 if (oap->oap_async_flags & ASYNC_GROUP_SYNC) {
2610 rc = oig_add_one(oig, &oap->oap_occ);
2613 LOI_DEBUG(loi, "oap %p page %p on group pending: rc %d\n",
2614 oap, oap->oap_page, rc);
2616 client_obd_list_unlock(&cli->cl_loi_list_lock);
2621 static void osc_group_to_pending(struct client_obd *cli, struct lov_oinfo *loi,
2622 struct loi_oap_pages *lop, int cmd)
2624 struct list_head *pos, *tmp;
2625 struct osc_async_page *oap;
2627 list_for_each_safe(pos, tmp, &lop->lop_pending_group) {
2628 oap = list_entry(pos, struct osc_async_page, oap_pending_item);
2629 list_del(&oap->oap_pending_item);
2630 osc_oap_to_pending(oap);
2632 loi_list_maint(cli, loi);
2635 static int osc_trigger_group_io(struct obd_export *exp,
2636 struct lov_stripe_md *lsm,
2637 struct lov_oinfo *loi,
2638 struct obd_io_group *oig)
2640 struct client_obd *cli = &exp->exp_obd->u.cli;
2644 loi = lsm->lsm_oinfo[0];
2646 client_obd_list_lock(&cli->cl_loi_list_lock);
2648 osc_group_to_pending(cli, loi, &loi->loi_write_lop, OBD_BRW_WRITE);
2649 osc_group_to_pending(cli, loi, &loi->loi_read_lop, OBD_BRW_READ);
2651 osc_check_rpcs(cli);
2652 client_obd_list_unlock(&cli->cl_loi_list_lock);
2657 static int osc_teardown_async_page(struct obd_export *exp,
2658 struct lov_stripe_md *lsm,
2659 struct lov_oinfo *loi, void *cookie)
2661 struct client_obd *cli = &exp->exp_obd->u.cli;
2662 struct loi_oap_pages *lop;
2663 struct osc_async_page *oap;
2667 oap = oap_from_cookie(cookie);
2669 RETURN(PTR_ERR(oap));
2672 loi = lsm->lsm_oinfo[0];
2674 if (oap->oap_cmd & OBD_BRW_WRITE) {
2675 lop = &loi->loi_write_lop;
2677 lop = &loi->loi_read_lop;
2680 client_obd_list_lock(&cli->cl_loi_list_lock);
2682 if (!list_empty(&oap->oap_rpc_item))
2683 GOTO(out, rc = -EBUSY);
2685 osc_exit_cache(cli, oap, 0);
2686 osc_wake_cache_waiters(cli);
2688 if (!list_empty(&oap->oap_urgent_item)) {
2689 list_del_init(&oap->oap_urgent_item);
2690 oap->oap_async_flags &= ~ASYNC_URGENT;
2692 if (!list_empty(&oap->oap_pending_item)) {
2693 list_del_init(&oap->oap_pending_item);
2694 lop_update_pending(cli, lop, oap->oap_cmd, -1);
2696 loi_list_maint(cli, loi);
2698 LOI_DEBUG(loi, "oap %p page %p torn down\n", oap, oap->oap_page);
2700 client_obd_list_unlock(&cli->cl_loi_list_lock);
2704 static void osc_set_data_with_check(struct lustre_handle *lockh, void *data,
2707 struct ldlm_lock *lock = ldlm_handle2lock(lockh);
2710 CERROR("lockh %p, data %p - client evicted?\n", lockh, data);
2713 lock_res_and_lock(lock);
2716 /* Liang XXX: Darwin and Winnt checking should be added */
2717 if (lock->l_ast_data && lock->l_ast_data != data) {
2718 struct inode *new_inode = data;
2719 struct inode *old_inode = lock->l_ast_data;
2720 if (!(old_inode->i_state & I_FREEING))
2721 LDLM_ERROR(lock, "inconsistent l_ast_data found");
2722 LASSERTF(old_inode->i_state & I_FREEING,
2723 "Found existing inode %p/%lu/%u state %lu in lock: "
2724 "setting data to %p/%lu/%u\n", old_inode,
2725 old_inode->i_ino, old_inode->i_generation,
2727 new_inode, new_inode->i_ino, new_inode->i_generation);
2731 lock->l_ast_data = data;
2732 lock->l_flags |= (flags & LDLM_FL_NO_LRU);
2733 unlock_res_and_lock(lock);
2734 LDLM_LOCK_PUT(lock);
2737 static int osc_change_cbdata(struct obd_export *exp, struct lov_stripe_md *lsm,
2738 ldlm_iterator_t replace, void *data)
2740 struct ldlm_res_id res_id = { .name = {0} };
2741 struct obd_device *obd = class_exp2obd(exp);
2743 res_id.name[0] = lsm->lsm_object_id;
2744 res_id.name[2] = lsm->lsm_object_gr;
2746 ldlm_resource_iterate(obd->obd_namespace, &res_id, replace, data);
2750 static int osc_enqueue_fini(struct ptlrpc_request *req, struct obd_info *oinfo,
2756 /* The request was created before ldlm_cli_enqueue call. */
2757 if (rc == ELDLM_LOCK_ABORTED) {
2758 struct ldlm_reply *rep;
2760 /* swabbed by ldlm_cli_enqueue() */
2761 LASSERT_REPSWABBED(req, DLM_LOCKREPLY_OFF);
2762 rep = lustre_msg_buf(req->rq_repmsg, DLM_LOCKREPLY_OFF,
2764 LASSERT(rep != NULL);
2765 if (rep->lock_policy_res1)
2766 rc = rep->lock_policy_res1;
2770 if ((intent && rc == ELDLM_LOCK_ABORTED) || !rc) {
2771 CDEBUG(D_INODE,"got kms "LPU64" blocks "LPU64" mtime "LPU64"\n",
2772 oinfo->oi_md->lsm_oinfo[0]->loi_lvb.lvb_size,
2773 oinfo->oi_md->lsm_oinfo[0]->loi_lvb.lvb_blocks,
2774 oinfo->oi_md->lsm_oinfo[0]->loi_lvb.lvb_mtime);
2777 /* Call the update callback. */
2778 rc = oinfo->oi_cb_up(oinfo, rc);
2782 static int osc_enqueue_interpret(struct ptlrpc_request *req,
2783 struct osc_enqueue_args *aa, int rc)
2785 int intent = aa->oa_oi->oi_flags & LDLM_FL_HAS_INTENT;
2786 struct lov_stripe_md *lsm = aa->oa_oi->oi_md;
2787 struct ldlm_lock *lock;
2789 /* ldlm_cli_enqueue is holding a reference on the lock, so it must
2791 lock = ldlm_handle2lock(aa->oa_oi->oi_lockh);
2793 /* Complete obtaining the lock procedure. */
2794 rc = ldlm_cli_enqueue_fini(aa->oa_exp, req, aa->oa_ei->ei_type, 1,
2796 &aa->oa_oi->oi_flags,
2797 &lsm->lsm_oinfo[0]->loi_lvb,
2798 sizeof(lsm->lsm_oinfo[0]->loi_lvb),
2799 lustre_swab_ost_lvb,
2800 aa->oa_oi->oi_lockh, rc);
2802 /* Complete osc stuff. */
2803 rc = osc_enqueue_fini(req, aa->oa_oi, intent, rc);
2805 /* Release the lock for async request. */
2806 if (lustre_handle_is_used(aa->oa_oi->oi_lockh) && rc == ELDLM_OK)
2807 ldlm_lock_decref(aa->oa_oi->oi_lockh, aa->oa_ei->ei_mode);
2809 LASSERTF(lock != NULL, "lockh %p, req %p, aa %p - client evicted?\n",
2810 aa->oa_oi->oi_lockh, req, aa);
2811 LDLM_LOCK_PUT(lock);
2815 /* When enqueuing asynchronously, locks are not ordered, we can obtain a lock
2816 * from the 2nd OSC before a lock from the 1st one. This does not deadlock with
2817 * other synchronous requests, however keeping some locks and trying to obtain
2818 * others may take a considerable amount of time in a case of ost failure; and
2819 * when other sync requests do not get released lock from a client, the client
2820 * is excluded from the cluster -- such scenarious make the life difficult, so
2821 * release locks just after they are obtained. */
2822 static int osc_enqueue(struct obd_export *exp, struct obd_info *oinfo,
2823 struct ldlm_enqueue_info *einfo,
2824 struct ptlrpc_request_set *rqset)
2826 struct ldlm_res_id res_id = { .name = {0} };
2827 struct obd_device *obd = exp->exp_obd;
2828 struct ldlm_reply *rep;
2829 struct ptlrpc_request *req = NULL;
2830 int intent = oinfo->oi_flags & LDLM_FL_HAS_INTENT;
2834 res_id.name[0] = oinfo->oi_md->lsm_object_id;
2835 res_id.name[2] = oinfo->oi_md->lsm_object_gr;
2837 /* Filesystem lock extents are extended to page boundaries so that
2838 * dealing with the page cache is a little smoother. */
2839 oinfo->oi_policy.l_extent.start -=
2840 oinfo->oi_policy.l_extent.start & ~CFS_PAGE_MASK;
2841 oinfo->oi_policy.l_extent.end |= ~CFS_PAGE_MASK;
2843 if (oinfo->oi_md->lsm_oinfo[0]->loi_kms_valid == 0)
2846 /* Next, search for already existing extent locks that will cover us */
2847 rc = ldlm_lock_match(obd->obd_namespace,
2848 oinfo->oi_flags | LDLM_FL_LVB_READY, &res_id,
2849 einfo->ei_type, &oinfo->oi_policy, einfo->ei_mode,
2852 osc_set_data_with_check(oinfo->oi_lockh, einfo->ei_cbdata,
2855 /* I would like to be able to ASSERT here that rss <=
2856 * kms, but I can't, for reasons which are explained in
2860 /* We already have a lock, and it's referenced */
2861 oinfo->oi_cb_up(oinfo, ELDLM_OK);
2863 /* For async requests, decref the lock. */
2865 ldlm_lock_decref(oinfo->oi_lockh, einfo->ei_mode);
2870 /* If we're trying to read, we also search for an existing PW lock. The
2871 * VFS and page cache already protect us locally, so lots of readers/
2872 * writers can share a single PW lock.
2874 * There are problems with conversion deadlocks, so instead of
2875 * converting a read lock to a write lock, we'll just enqueue a new
2878 * At some point we should cancel the read lock instead of making them
2879 * send us a blocking callback, but there are problems with canceling
2880 * locks out from other users right now, too. */
2882 if (einfo->ei_mode == LCK_PR) {
2883 rc = ldlm_lock_match(obd->obd_namespace,
2884 oinfo->oi_flags | LDLM_FL_LVB_READY,
2885 &res_id, einfo->ei_type, &oinfo->oi_policy,
2886 LCK_PW, oinfo->oi_lockh);
2888 /* FIXME: This is not incredibly elegant, but it might
2889 * be more elegant than adding another parameter to
2890 * lock_match. I want a second opinion. */
2891 /* addref the lock only if not async requests. */
2893 ldlm_lock_addref(oinfo->oi_lockh, LCK_PR);
2894 osc_set_data_with_check(oinfo->oi_lockh,
2897 oinfo->oi_cb_up(oinfo, ELDLM_OK);
2898 ldlm_lock_decref(oinfo->oi_lockh, LCK_PW);
2906 [MSG_PTLRPC_BODY_OFF] = sizeof(struct ptlrpc_body),
2907 [DLM_LOCKREQ_OFF] = sizeof(struct ldlm_request),
2908 [DLM_LOCKREQ_OFF + 1] = 0 };
2910 req = ldlm_prep_enqueue_req(exp, 2, size, NULL, 0);
2914 size[DLM_LOCKREPLY_OFF] = sizeof(*rep);
2915 size[DLM_REPLY_REC_OFF] =
2916 sizeof(oinfo->oi_md->lsm_oinfo[0]->loi_lvb);
2917 ptlrpc_req_set_repsize(req, 3, size);
2920 /* users of osc_enqueue() can pass this flag for ldlm_lock_match() */
2921 oinfo->oi_flags &= ~LDLM_FL_BLOCK_GRANTED;
2923 rc = ldlm_cli_enqueue(exp, &req, einfo, &res_id,
2924 &oinfo->oi_policy, &oinfo->oi_flags,
2925 &oinfo->oi_md->lsm_oinfo[0]->loi_lvb,
2926 sizeof(oinfo->oi_md->lsm_oinfo[0]->loi_lvb),
2927 lustre_swab_ost_lvb, oinfo->oi_lockh,
2931 struct osc_enqueue_args *aa;
2932 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
2933 aa = (struct osc_enqueue_args *)&req->rq_async_args;
2938 req->rq_interpret_reply = osc_enqueue_interpret;
2939 ptlrpc_set_add_req(rqset, req);
2940 } else if (intent) {
2941 ptlrpc_req_finished(req);
2946 rc = osc_enqueue_fini(req, oinfo, intent, rc);
2948 ptlrpc_req_finished(req);
2953 static int osc_match(struct obd_export *exp, struct lov_stripe_md *lsm,
2954 __u32 type, ldlm_policy_data_t *policy, __u32 mode,
2955 int *flags, void *data, struct lustre_handle *lockh)
2957 struct ldlm_res_id res_id = { .name = {0} };
2958 struct obd_device *obd = exp->exp_obd;
2960 int lflags = *flags;
2963 res_id.name[0] = lsm->lsm_object_id;
2964 res_id.name[2] = lsm->lsm_object_gr;
2966 OBD_FAIL_RETURN(OBD_FAIL_OSC_MATCH, -EIO);
2968 /* Filesystem lock extents are extended to page boundaries so that
2969 * dealing with the page cache is a little smoother */
2970 policy->l_extent.start -= policy->l_extent.start & ~CFS_PAGE_MASK;
2971 policy->l_extent.end |= ~CFS_PAGE_MASK;
2973 /* Next, search for already existing extent locks that will cover us */
2974 rc = ldlm_lock_match(obd->obd_namespace, lflags | LDLM_FL_LVB_READY,
2975 &res_id, type, policy, mode, lockh);
2977 //if (!(*flags & LDLM_FL_TEST_LOCK))
2978 osc_set_data_with_check(lockh, data, lflags);
2981 /* If we're trying to read, we also search for an existing PW lock. The
2982 * VFS and page cache already protect us locally, so lots of readers/
2983 * writers can share a single PW lock. */
2984 if (mode == LCK_PR) {
2985 rc = ldlm_lock_match(obd->obd_namespace,
2986 lflags | LDLM_FL_LVB_READY, &res_id,
2987 type, policy, LCK_PW, lockh);
2988 if (rc == 1 && !(lflags & LDLM_FL_TEST_LOCK)) {
2989 /* FIXME: This is not incredibly elegant, but it might
2990 * be more elegant than adding another parameter to
2991 * lock_match. I want a second opinion. */
2992 osc_set_data_with_check(lockh, data, lflags);
2993 ldlm_lock_addref(lockh, LCK_PR);
2994 ldlm_lock_decref(lockh, LCK_PW);
3000 static int osc_cancel(struct obd_export *exp, struct lov_stripe_md *md,
3001 __u32 mode, struct lustre_handle *lockh)
3005 if (unlikely(mode == LCK_GROUP))
3006 ldlm_lock_decref_and_cancel(lockh, mode);
3008 ldlm_lock_decref(lockh, mode);
3013 static int osc_cancel_unused(struct obd_export *exp,
3014 struct lov_stripe_md *lsm, int flags,
3017 struct obd_device *obd = class_exp2obd(exp);
3018 struct ldlm_res_id res_id = { .name = {0} }, *resp = NULL;
3021 res_id.name[0] = lsm->lsm_object_id;
3022 res_id.name[2] = lsm->lsm_object_gr;
3026 return ldlm_cli_cancel_unused(obd->obd_namespace, resp, flags, opaque);
3029 static int osc_join_lru(struct obd_export *exp,
3030 struct lov_stripe_md *lsm, int join)
3032 struct obd_device *obd = class_exp2obd(exp);
3033 struct ldlm_res_id res_id = { .name = {0} }, *resp = NULL;
3036 res_id.name[0] = lsm->lsm_object_id;
3037 res_id.name[2] = lsm->lsm_object_gr;
3041 return ldlm_cli_join_lru(obd->obd_namespace, resp, join);
3044 static int osc_statfs_interpret(struct ptlrpc_request *req,
3045 struct osc_async_args *aa, int rc)
3047 struct obd_statfs *msfs;
3053 msfs = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*msfs),
3054 lustre_swab_obd_statfs);
3056 CERROR("Can't unpack obd_statfs\n");
3057 GOTO(out, rc = -EPROTO);
3060 memcpy(aa->aa_oi->oi_osfs, msfs, sizeof(*msfs));
3062 rc = aa->aa_oi->oi_cb_up(aa->aa_oi, rc);
3066 static int osc_statfs_async(struct obd_device *obd, struct obd_info *oinfo,
3067 __u64 max_age, struct ptlrpc_request_set *rqset)
3069 struct ptlrpc_request *req;
3070 struct osc_async_args *aa;
3071 int size[2] = { sizeof(struct ptlrpc_body), sizeof(*oinfo->oi_osfs) };
3074 /* We could possibly pass max_age in the request (as an absolute
3075 * timestamp or a "seconds.usec ago") so the target can avoid doing
3076 * extra calls into the filesystem if that isn't necessary (e.g.
3077 * during mount that would help a bit). Having relative timestamps
3078 * is not so great if request processing is slow, while absolute
3079 * timestamps are not ideal because they need time synchronization. */
3080 req = ptlrpc_prep_req(obd->u.cli.cl_import, LUSTRE_OST_VERSION,
3081 OST_STATFS, 1, NULL, NULL);
3085 ptlrpc_req_set_repsize(req, 2, size);
3086 req->rq_request_portal = OST_CREATE_PORTAL; //XXX FIXME bug 249
3088 req->rq_interpret_reply = osc_statfs_interpret;
3089 CLASSERT (sizeof(*aa) <= sizeof(req->rq_async_args));
3090 aa = (struct osc_async_args *)&req->rq_async_args;
3093 ptlrpc_set_add_req(rqset, req);
3097 static int osc_statfs(struct obd_device *obd, struct obd_statfs *osfs,
3100 struct obd_statfs *msfs;
3101 struct ptlrpc_request *req;
3102 int rc, size[2] = { sizeof(struct ptlrpc_body), sizeof(*osfs) };
3105 /* We could possibly pass max_age in the request (as an absolute
3106 * timestamp or a "seconds.usec ago") so the target can avoid doing
3107 * extra calls into the filesystem if that isn't necessary (e.g.
3108 * during mount that would help a bit). Having relative timestamps
3109 * is not so great if request processing is slow, while absolute
3110 * timestamps are not ideal because they need time synchronization. */
3111 req = ptlrpc_prep_req(obd->u.cli.cl_import, LUSTRE_OST_VERSION,
3112 OST_STATFS, 1, NULL, NULL);
3116 ptlrpc_req_set_repsize(req, 2, size);
3117 req->rq_request_portal = OST_CREATE_PORTAL; //XXX FIXME bug 249
3119 rc = ptlrpc_queue_wait(req);
3123 msfs = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*msfs),
3124 lustre_swab_obd_statfs);
3126 CERROR("Can't unpack obd_statfs\n");
3127 GOTO(out, rc = -EPROTO);
3130 memcpy(osfs, msfs, sizeof(*osfs));
3134 ptlrpc_req_finished(req);
3138 /* Retrieve object striping information.
3140 * @lmmu is a pointer to an in-core struct with lmm_ost_count indicating
3141 * the maximum number of OST indices which will fit in the user buffer.
3142 * lmm_magic must be LOV_MAGIC (we only use 1 slot here).
3144 static int osc_getstripe(struct lov_stripe_md *lsm, struct lov_user_md *lump)
3146 struct lov_user_md lum, *lumk;
3147 int rc = 0, lum_size;
3153 if (copy_from_user(&lum, lump, sizeof(lum)))
3156 if (lum.lmm_magic != LOV_USER_MAGIC)
3159 if (lum.lmm_stripe_count > 0) {
3160 lum_size = sizeof(lum) + sizeof(lum.lmm_objects[0]);
3161 OBD_ALLOC(lumk, lum_size);
3165 lumk->lmm_objects[0].l_object_id = lsm->lsm_object_id;
3166 lumk->lmm_objects[0].l_object_gr = lsm->lsm_object_gr;
3168 lum_size = sizeof(lum);
3172 lumk->lmm_object_id = lsm->lsm_object_id;
3173 lumk->lmm_object_gr = lsm->lsm_object_gr;
3174 lumk->lmm_stripe_count = 1;
3176 if (copy_to_user(lump, lumk, lum_size))
3180 OBD_FREE(lumk, lum_size);
3186 static int osc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
3187 void *karg, void *uarg)
3189 struct obd_device *obd = exp->exp_obd;
3190 struct obd_ioctl_data *data = karg;
3194 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
3197 if (!try_module_get(THIS_MODULE)) {
3198 CERROR("Can't get module. Is it alive?");
3203 case OBD_IOC_LOV_GET_CONFIG: {
3205 struct lov_desc *desc;
3206 struct obd_uuid uuid;
3210 if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
3211 GOTO(out, err = -EINVAL);
3213 data = (struct obd_ioctl_data *)buf;
3215 if (sizeof(*desc) > data->ioc_inllen1) {
3216 obd_ioctl_freedata(buf, len);
3217 GOTO(out, err = -EINVAL);
3220 if (data->ioc_inllen2 < sizeof(uuid)) {
3221 obd_ioctl_freedata(buf, len);
3222 GOTO(out, err = -EINVAL);
3225 desc = (struct lov_desc *)data->ioc_inlbuf1;
3226 desc->ld_tgt_count = 1;
3227 desc->ld_active_tgt_count = 1;
3228 desc->ld_default_stripe_count = 1;
3229 desc->ld_default_stripe_size = 0;
3230 desc->ld_default_stripe_offset = 0;
3231 desc->ld_pattern = 0;
3232 memcpy(&desc->ld_uuid, &obd->obd_uuid, sizeof(uuid));
3234 memcpy(data->ioc_inlbuf2, &obd->obd_uuid, sizeof(uuid));
3236 err = copy_to_user((void *)uarg, buf, len);
3239 obd_ioctl_freedata(buf, len);
3242 case LL_IOC_LOV_SETSTRIPE:
3243 err = obd_alloc_memmd(exp, karg);
3247 case LL_IOC_LOV_GETSTRIPE:
3248 err = osc_getstripe(karg, uarg);
3250 case OBD_IOC_CLIENT_RECOVER:
3251 err = ptlrpc_recover_import(obd->u.cli.cl_import,
3256 case IOC_OSC_SET_ACTIVE:
3257 err = ptlrpc_set_import_active(obd->u.cli.cl_import,
3260 case OBD_IOC_POLL_QUOTACHECK:
3261 err = lquota_poll_check(quota_interface, exp,
3262 (struct if_quotacheck *)karg);
3265 CDEBUG(D_INODE, "unrecognised ioctl %#x by %s\n",
3266 cmd, cfs_curproc_comm());
3267 GOTO(out, err = -ENOTTY);
3270 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
3273 module_put(THIS_MODULE);
3278 static int osc_get_info(struct obd_export *exp, obd_count keylen,
3279 void *key, __u32 *vallen, void *val)
3282 if (!vallen || !val)
3285 if (keylen > strlen("lock_to_stripe") &&
3286 strcmp(key, "lock_to_stripe") == 0) {
3287 __u32 *stripe = val;
3288 *vallen = sizeof(*stripe);
3291 } else if (keylen >= strlen("last_id") && strcmp(key, "last_id") == 0) {
3292 struct ptlrpc_request *req;
3294 char *bufs[2] = { NULL, key };
3295 int rc, size[2] = { sizeof(struct ptlrpc_body), keylen };
3297 req = ptlrpc_prep_req(class_exp2cliimp(exp), LUSTRE_OST_VERSION,
3298 OST_GET_INFO, 2, size, bufs);
3302 size[REPLY_REC_OFF] = *vallen;
3303 ptlrpc_req_set_repsize(req, 2, size);
3304 rc = ptlrpc_queue_wait(req);
3308 reply = lustre_swab_repbuf(req, REPLY_REC_OFF, sizeof(*reply),
3309 lustre_swab_ost_last_id);
3310 if (reply == NULL) {
3311 CERROR("Can't unpack OST last ID\n");
3312 GOTO(out, rc = -EPROTO);
3314 *((obd_id *)val) = *reply;
3316 ptlrpc_req_finished(req);
3322 static int osc_setinfo_mds_conn_interpret(struct ptlrpc_request *req,
3325 struct llog_ctxt *ctxt;
3326 struct obd_import *imp = req->rq_import;
3332 ctxt = llog_get_context(imp->imp_obd, LLOG_MDS_OST_ORIG_CTXT);
3335 rc = llog_initiator_connect(ctxt);
3337 CERROR("cannot establish connection for "
3338 "ctxt %p: %d\n", ctxt, rc);
3341 spin_lock(&imp->imp_lock);
3342 imp->imp_server_timeout = 1;
3343 imp->imp_pingable = 1;
3344 spin_unlock(&imp->imp_lock);
3345 CDEBUG(D_HA, "pinging OST %s\n", obd2cli_tgt(imp->imp_obd));
3350 static int osc_set_info_async(struct obd_export *exp, obd_count keylen,
3351 void *key, obd_count vallen, void *val,
3352 struct ptlrpc_request_set *set)
3354 struct ptlrpc_request *req;
3355 struct obd_device *obd = exp->exp_obd;
3356 struct obd_import *imp = class_exp2cliimp(exp);
3357 int size[3] = { sizeof(struct ptlrpc_body), keylen, vallen };
3358 char *bufs[3] = { NULL, key, val };
3361 OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_SHUTDOWN, 10);
3363 if (KEY_IS(KEY_NEXT_ID)) {
3364 if (vallen != sizeof(obd_id))
3366 obd->u.cli.cl_oscc.oscc_next_id = *((obd_id*)val) + 1;
3367 CDEBUG(D_HA, "%s: set oscc_next_id = "LPU64"\n",
3368 exp->exp_obd->obd_name,
3369 obd->u.cli.cl_oscc.oscc_next_id);
3374 if (KEY_IS("unlinked")) {
3375 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3376 spin_lock(&oscc->oscc_lock);
3377 oscc->oscc_flags &= ~OSCC_FLAG_NOSPC;
3378 spin_unlock(&oscc->oscc_lock);
3382 if (KEY_IS(KEY_INIT_RECOV)) {
3383 if (vallen != sizeof(int))
3385 spin_lock(&imp->imp_lock);
3386 imp->imp_initial_recov = *(int *)val;
3387 spin_unlock(&imp->imp_lock);
3388 CDEBUG(D_HA, "%s: set imp_initial_recov = %d\n",
3389 exp->exp_obd->obd_name,
3390 imp->imp_initial_recov);
3394 if (KEY_IS("checksum")) {
3395 if (vallen != sizeof(int))
3397 exp->exp_obd->u.cli.cl_checksum = (*(int *)val) ? 1 : 0;
3401 if (KEY_IS(KEY_FLUSH_CTX)) {
3402 sptlrpc_import_flush_my_ctx(imp);
3409 /* We pass all other commands directly to OST. Since nobody calls osc
3410 methods directly and everybody is supposed to go through LOV, we
3411 assume lov checked invalid values for us.
3412 The only recognised values so far are evict_by_nid and mds_conn.
3413 Even if something bad goes through, we'd get a -EINVAL from OST
3416 req = ptlrpc_prep_req(imp, LUSTRE_OST_VERSION, OST_SET_INFO, 3, size,
3421 if (KEY_IS("mds_conn")) {
3422 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3424 oscc->oscc_oa.o_gr = (*(__u32 *)val);
3425 oscc->oscc_oa.o_valid |= OBD_MD_FLGROUP;
3426 LASSERT(oscc->oscc_oa.o_gr > 0);
3427 req->rq_interpret_reply = osc_setinfo_mds_conn_interpret;
3430 ptlrpc_req_set_repsize(req, 1, NULL);
3431 ptlrpc_set_add_req(set, req);
3432 ptlrpc_check_set(set);
3438 static struct llog_operations osc_size_repl_logops = {
3439 lop_cancel: llog_obd_repl_cancel
3442 static struct llog_operations osc_mds_ost_orig_logops;
3443 static int osc_llog_init(struct obd_device *obd, struct obd_llogs *llogs,
3444 struct obd_device *tgt, int count,
3445 struct llog_catid *catid, struct obd_uuid *uuid)
3450 spin_lock(&obd->obd_dev_lock);
3451 if (osc_mds_ost_orig_logops.lop_setup != llog_obd_origin_setup) {
3452 osc_mds_ost_orig_logops = llog_lvfs_ops;
3453 osc_mds_ost_orig_logops.lop_setup = llog_obd_origin_setup;
3454 osc_mds_ost_orig_logops.lop_cleanup = llog_obd_origin_cleanup;
3455 osc_mds_ost_orig_logops.lop_add = llog_obd_origin_add;
3456 osc_mds_ost_orig_logops.lop_connect = llog_origin_connect;
3458 spin_unlock(&obd->obd_dev_lock);
3460 rc = llog_setup(obd, llogs, LLOG_MDS_OST_ORIG_CTXT, tgt, count,
3461 &catid->lci_logid, &osc_mds_ost_orig_logops);
3463 CERROR("failed LLOG_MDS_OST_ORIG_CTXT\n");
3467 rc = llog_setup(obd, llogs, LLOG_SIZE_REPL_CTXT, tgt, count, NULL,
3468 &osc_size_repl_logops);
3470 CERROR("failed LLOG_SIZE_REPL_CTXT\n");
3473 CERROR("osc '%s' tgt '%s' cnt %d catid %p rc=%d\n",
3474 obd->obd_name, tgt->obd_name, count, catid, rc);
3475 CERROR("logid "LPX64":0x%x\n",
3476 catid->lci_logid.lgl_oid, catid->lci_logid.lgl_ogen);
3481 static int osc_llog_finish(struct obd_device *obd, int count)
3483 struct llog_ctxt *ctxt;
3484 int rc = 0, rc2 = 0;
3487 ctxt = llog_get_context(obd, LLOG_MDS_OST_ORIG_CTXT);
3489 rc = llog_cleanup(ctxt);
3491 ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3493 rc2 = llog_cleanup(ctxt);
3500 static int osc_reconnect(struct obd_export *exp, struct obd_device *obd,
3501 struct obd_uuid *cluuid,
3502 struct obd_connect_data *data)
3504 struct client_obd *cli = &obd->u.cli;
3506 if (data != NULL && (data->ocd_connect_flags & OBD_CONNECT_GRANT)) {
3509 client_obd_list_lock(&cli->cl_loi_list_lock);
3510 data->ocd_grant = cli->cl_avail_grant ?:
3511 2 * cli->cl_max_pages_per_rpc << CFS_PAGE_SHIFT;
3512 lost_grant = cli->cl_lost_grant;
3513 cli->cl_lost_grant = 0;
3514 client_obd_list_unlock(&cli->cl_loi_list_lock);
3516 CDEBUG(D_CACHE, "request ocd_grant: %d cl_avail_grant: %ld "
3517 "cl_lost_grant: %ld\n", data->ocd_grant,
3518 cli->cl_avail_grant, lost_grant);
3519 CDEBUG(D_RPCTRACE, "ocd_connect_flags: "LPX64" ocd_version: %d"
3520 " ocd_grant: %d\n", data->ocd_connect_flags,
3521 data->ocd_version, data->ocd_grant);
3527 static int osc_disconnect(struct obd_export *exp)
3529 struct obd_device *obd = class_exp2obd(exp);
3530 struct llog_ctxt *ctxt = llog_get_context(obd, LLOG_SIZE_REPL_CTXT);
3533 if (obd->u.cli.cl_conn_count == 1)
3534 /* flush any remaining cancel messages out to the target */
3535 llog_sync(ctxt, exp);
3537 rc = client_disconnect_export(exp);
3541 static int osc_import_event(struct obd_device *obd,
3542 struct obd_import *imp,
3543 enum obd_import_event event)
3545 struct client_obd *cli;
3549 LASSERT(imp->imp_obd == obd);
3552 case IMP_EVENT_DISCON: {
3553 /* Only do this on the MDS OSC's */
3554 if (imp->imp_server_timeout) {
3555 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3557 spin_lock(&oscc->oscc_lock);
3558 oscc->oscc_flags |= OSCC_FLAG_RECOVERING;
3559 spin_unlock(&oscc->oscc_lock);
3562 client_obd_list_lock(&cli->cl_loi_list_lock);
3563 cli->cl_avail_grant = 0;
3564 cli->cl_lost_grant = 0;
3565 client_obd_list_unlock(&cli->cl_loi_list_lock);
3568 case IMP_EVENT_INACTIVE: {
3569 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_INACTIVE, NULL);
3572 case IMP_EVENT_INVALIDATE: {
3573 struct ldlm_namespace *ns = obd->obd_namespace;
3577 client_obd_list_lock(&cli->cl_loi_list_lock);
3578 /* all pages go to failing rpcs due to the invalid import */
3579 osc_check_rpcs(cli);
3580 client_obd_list_unlock(&cli->cl_loi_list_lock);
3582 ldlm_namespace_cleanup(ns, LDLM_FL_LOCAL_ONLY);
3586 case IMP_EVENT_ACTIVE: {
3587 /* Only do this on the MDS OSC's */
3588 if (imp->imp_server_timeout) {
3589 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3591 spin_lock(&oscc->oscc_lock);
3592 oscc->oscc_flags &= ~OSCC_FLAG_NOSPC;
3593 spin_unlock(&oscc->oscc_lock);
3595 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
3598 case IMP_EVENT_OCD: {
3599 struct obd_connect_data *ocd = &imp->imp_connect_data;
3601 if (ocd->ocd_connect_flags & OBD_CONNECT_GRANT)
3602 osc_init_grant(&obd->u.cli, ocd);
3605 if (ocd->ocd_connect_flags & OBD_CONNECT_REQPORTAL)
3606 imp->imp_client->cli_request_portal =OST_REQUEST_PORTAL;
3608 rc = obd_notify_observer(obd, obd, OBD_NOTIFY_OCD, NULL);
3612 CERROR("Unknown import event %d\n", event);
3618 int osc_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
3624 rc = ptlrpcd_addref();
3628 rc = client_obd_setup(obd, lcfg);
3632 struct lprocfs_static_vars lvars;
3633 struct client_obd *cli = &obd->u.cli;
3635 lprocfs_init_vars(osc, &lvars);
3636 if (lprocfs_obd_setup(obd, lvars.obd_vars) == 0) {
3637 lproc_osc_attach_seqstat(obd);
3638 ptlrpc_lprocfs_register_obd(obd);
3642 /* We need to allocate a few requests more, because
3643 brw_interpret_oap tries to create new requests before freeing
3644 previous ones. Ideally we want to have 2x max_rpcs_in_flight
3645 reserved, but I afraid that might be too much wasted RAM
3646 in fact, so 2 is just my guess and still should work. */
3647 cli->cl_import->imp_rq_pool =
3648 ptlrpc_init_rq_pool(cli->cl_max_rpcs_in_flight + 2,
3650 ptlrpc_add_rqs_to_pool);
3656 static int osc_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
3662 case OBD_CLEANUP_EARLY: {
3663 struct obd_import *imp;
3664 imp = obd->u.cli.cl_import;
3665 CDEBUG(D_HA, "Deactivating import %s\n", obd->obd_name);
3666 /* ptlrpc_abort_inflight to stop an mds_lov_synchronize */
3667 ptlrpc_deactivate_import(imp);
3668 spin_lock(&imp->imp_lock);
3669 imp->imp_pingable = 0;
3670 spin_unlock(&imp->imp_lock);
3673 case OBD_CLEANUP_EXPORTS: {
3674 /* If we set up but never connected, the
3675 client import will not have been cleaned. */
3676 if (obd->u.cli.cl_import) {
3677 struct obd_import *imp;
3678 imp = obd->u.cli.cl_import;
3679 CDEBUG(D_CONFIG, "%s: client import never connected\n",
3681 ptlrpc_invalidate_import(imp);
3682 ptlrpc_free_rq_pool(imp->imp_rq_pool);
3683 class_destroy_import(imp);
3684 obd->u.cli.cl_import = NULL;
3688 case OBD_CLEANUP_SELF_EXP:
3689 rc = obd_llog_finish(obd, 0);
3691 CERROR("failed to cleanup llogging subsystems\n");
3693 case OBD_CLEANUP_OBD:
3699 int osc_cleanup(struct obd_device *obd)
3701 struct osc_creator *oscc = &obd->u.cli.cl_oscc;
3705 ptlrpc_lprocfs_unregister_obd(obd);
3706 lprocfs_obd_cleanup(obd);
3708 spin_lock(&oscc->oscc_lock);
3709 oscc->oscc_flags &= ~OSCC_FLAG_RECOVERING;
3710 oscc->oscc_flags |= OSCC_FLAG_EXITING;
3711 spin_unlock(&oscc->oscc_lock);
3713 /* free memory of osc quota cache */
3714 lquota_cleanup(quota_interface, obd);
3716 rc = client_obd_cleanup(obd);
3722 static int osc_process_config(struct obd_device *obd, obd_count len, void *buf)
3724 struct lustre_cfg *lcfg = buf;
3725 struct lprocfs_static_vars lvars;
3728 lprocfs_init_vars(osc, &lvars);
3730 rc = class_process_proc_param(PARAM_OSC, lvars.obd_vars, lcfg, obd);
3734 struct obd_ops osc_obd_ops = {
3735 .o_owner = THIS_MODULE,
3736 .o_setup = osc_setup,
3737 .o_precleanup = osc_precleanup,
3738 .o_cleanup = osc_cleanup,
3739 .o_add_conn = client_import_add_conn,
3740 .o_del_conn = client_import_del_conn,
3741 .o_connect = client_connect_import,
3742 .o_reconnect = osc_reconnect,
3743 .o_disconnect = osc_disconnect,
3744 .o_statfs = osc_statfs,
3745 .o_statfs_async = osc_statfs_async,
3746 .o_packmd = osc_packmd,
3747 .o_unpackmd = osc_unpackmd,
3748 .o_precreate = osc_precreate,
3749 .o_create = osc_create,
3750 .o_destroy = osc_destroy,
3751 .o_getattr = osc_getattr,
3752 .o_getattr_async = osc_getattr_async,
3753 .o_setattr = osc_setattr,
3754 .o_setattr_async = osc_setattr_async,
3756 .o_brw_async = osc_brw_async,
3757 .o_prep_async_page = osc_prep_async_page,
3758 .o_queue_async_io = osc_queue_async_io,
3759 .o_set_async_flags = osc_set_async_flags,
3760 .o_queue_group_io = osc_queue_group_io,
3761 .o_trigger_group_io = osc_trigger_group_io,
3762 .o_teardown_async_page = osc_teardown_async_page,
3763 .o_punch = osc_punch,
3765 .o_enqueue = osc_enqueue,
3766 .o_match = osc_match,
3767 .o_change_cbdata = osc_change_cbdata,
3768 .o_cancel = osc_cancel,
3769 .o_cancel_unused = osc_cancel_unused,
3770 .o_join_lru = osc_join_lru,
3771 .o_iocontrol = osc_iocontrol,
3772 .o_get_info = osc_get_info,
3773 .o_set_info_async = osc_set_info_async,
3774 .o_import_event = osc_import_event,
3775 .o_llog_init = osc_llog_init,
3776 .o_llog_finish = osc_llog_finish,
3777 .o_process_config = osc_process_config,
3780 int __init osc_init(void)
3782 struct lprocfs_static_vars lvars;
3786 lprocfs_init_vars(osc, &lvars);
3788 request_module("lquota");
3789 quota_interface = PORTAL_SYMBOL_GET(osc_quota_interface);
3790 lquota_init(quota_interface);
3791 init_obd_quota_ops(quota_interface, &osc_obd_ops);
3793 rc = class_register_type(&osc_obd_ops, NULL, lvars.module_vars,
3794 LUSTRE_OSC_NAME, NULL);
3796 if (quota_interface)
3797 PORTAL_SYMBOL_PUT(osc_quota_interface);
3805 static void /*__exit*/ osc_exit(void)
3807 lquota_exit(quota_interface);
3808 if (quota_interface)
3809 PORTAL_SYMBOL_PUT(osc_quota_interface);
3811 class_unregister_type(LUSTRE_OSC_NAME);
3814 MODULE_AUTHOR("Cluster File Systems, Inc. <info@clusterfs.com>");
3815 MODULE_DESCRIPTION("Lustre Object Storage Client (OSC)");
3816 MODULE_LICENSE("GPL");
3818 cfs_module(osc, LUSTRE_VERSION_STRING, osc_init, osc_exit);
git://git.whamcloud.com / fs / lustre-release.git / blob