4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.gnu.org/licenses/gpl-2.0.html
23 * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Use is subject to license terms.
26 * Copyright (c) 2011, 2017, Intel Corporation.
29 * This file is part of Lustre, http://www.lustre.org/
30 * Lustre is a trademark of Sun Microsystems, Inc.
33 #define DEBUG_SUBSYSTEM S_MDC
35 #include <linux/module.h>
38 #include <obd_class.h>
39 #include <lustre_dlm.h>
40 #include <lustre_fid.h>
41 #include <lustre_intent.h>
42 #include <lustre_mdc.h>
43 #include <lustre_net.h>
44 #include <lustre_req_layout.h>
45 #include <lustre_swab.h>
46 #include <lustre_acl.h>
48 #include "mdc_internal.h"
50 struct mdc_getattr_args {
51 struct obd_export *ga_exp;
52 struct md_enqueue_info *ga_minfo;
55 int it_open_error(int phase, struct lookup_intent *it)
57 if (it_disposition(it, DISP_OPEN_LEASE)) {
58 if (phase >= DISP_OPEN_LEASE)
63 if (it_disposition(it, DISP_OPEN_OPEN)) {
64 if (phase >= DISP_OPEN_OPEN)
70 if (it_disposition(it, DISP_OPEN_CREATE)) {
71 if (phase >= DISP_OPEN_CREATE)
77 if (it_disposition(it, DISP_LOOKUP_EXECD)) {
78 if (phase >= DISP_LOOKUP_EXECD)
84 if (it_disposition(it, DISP_IT_EXECD)) {
85 if (phase >= DISP_IT_EXECD)
91 CERROR("it disp: %X, status: %d\n", it->it_disposition, it->it_status);
96 EXPORT_SYMBOL(it_open_error);
98 /* this must be called on a lockh that is known to have a referenced lock */
99 int mdc_set_lock_data(struct obd_export *exp, const struct lustre_handle *lockh,
100 void *data, __u64 *bits)
102 struct ldlm_lock *lock;
103 struct inode *new_inode = data;
109 if (!lustre_handle_is_used(lockh))
112 lock = ldlm_handle2lock(lockh);
114 LASSERT(lock != NULL);
115 lock_res_and_lock(lock);
116 if (lock->l_resource->lr_lvb_inode &&
117 lock->l_resource->lr_lvb_inode != data) {
118 struct inode *old_inode = lock->l_resource->lr_lvb_inode;
120 LASSERTF(old_inode->i_state & I_FREEING,
121 "Found existing inode %p/%lu/%u state %lu in lock: setting data to %p/%lu/%u\n",
122 old_inode, old_inode->i_ino, old_inode->i_generation,
124 new_inode, new_inode->i_ino, new_inode->i_generation);
126 lock->l_resource->lr_lvb_inode = new_inode;
128 *bits = lock->l_policy_data.l_inodebits.bits;
130 unlock_res_and_lock(lock);
136 enum ldlm_mode mdc_lock_match(struct obd_export *exp, __u64 flags,
137 const struct lu_fid *fid, enum ldlm_type type,
138 union ldlm_policy_data *policy,
139 enum ldlm_mode mode, struct lustre_handle *lockh)
141 struct ldlm_res_id res_id;
145 fid_build_reg_res_name(fid, &res_id);
146 /* LU-4405: Clear bits not supported by server */
147 policy->l_inodebits.bits &= exp_connect_ibits(exp);
148 rc = ldlm_lock_match(class_exp2obd(exp)->obd_namespace, flags,
149 &res_id, type, policy, mode, lockh);
153 int mdc_cancel_unused(struct obd_export *exp, const struct lu_fid *fid,
154 union ldlm_policy_data *policy, enum ldlm_mode mode,
155 enum ldlm_cancel_flags flags, void *opaque)
157 struct obd_device *obd = class_exp2obd(exp);
158 struct ldlm_res_id res_id;
162 fid_build_reg_res_name(fid, &res_id);
163 rc = ldlm_cli_cancel_unused_resource(obd->obd_namespace, &res_id,
164 policy, mode, flags, opaque);
168 int mdc_null_inode(struct obd_export *exp,
169 const struct lu_fid *fid)
171 struct ldlm_res_id res_id;
172 struct ldlm_resource *res;
173 struct ldlm_namespace *ns = class_exp2obd(exp)->obd_namespace;
176 LASSERTF(ns != NULL, "no namespace passed\n");
178 fid_build_reg_res_name(fid, &res_id);
180 res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
185 res->lr_lvb_inode = NULL;
188 ldlm_resource_putref(res);
192 static inline void mdc_clear_replay_flag(struct ptlrpc_request *req, int rc)
194 /* Don't hold error requests for replay. */
195 if (req->rq_replay) {
196 spin_lock(&req->rq_lock);
198 spin_unlock(&req->rq_lock);
200 if (rc && req->rq_transno != 0) {
201 DEBUG_REQ(D_ERROR, req, "transno returned on error: rc = %d",
208 * Save a large LOV EA into the request buffer so that it is available
209 * for replay. We don't do this in the initial request because the
210 * original request doesn't need this buffer (at most it sends just the
211 * lov_mds_md) and it is a waste of RAM/bandwidth to send the empty
212 * buffer and may also be difficult to allocate and save a very large
213 * request buffer for each open. (b=5707)
215 * OOM here may cause recovery failure if lmm is needed (only for the
216 * original open if the MDS crashed just when this client also OOM'd)
217 * but this is incredibly unlikely, and questionable whether the client
218 * could do MDS recovery under OOM anyways...
220 static int mdc_save_lovea(struct ptlrpc_request *req, void *data, u32 size)
222 struct req_capsule *pill = &req->rq_pill;
226 if (req_capsule_get_size(pill, &RMF_EADATA, RCL_CLIENT) < size) {
227 rc = sptlrpc_cli_enlarge_reqbuf(req, &RMF_EADATA, size);
229 CERROR("%s: Can't enlarge ea size to %d: rc = %d\n",
230 req->rq_export->exp_obd->obd_name,
235 req_capsule_shrink(pill, &RMF_EADATA, size, RCL_CLIENT);
238 req_capsule_set_size(pill, &RMF_EADATA, RCL_CLIENT, size);
239 lovea = req_capsule_client_get(pill, &RMF_EADATA);
241 memcpy(lovea, data, size);
242 lov_fix_ea_for_replay(lovea);
248 static struct ptlrpc_request *
249 mdc_intent_open_pack(struct obd_export *exp, struct lookup_intent *it,
250 struct md_op_data *op_data, __u32 acl_bufsize)
252 struct ptlrpc_request *req;
253 struct obd_device *obd = class_exp2obd(exp);
254 struct ldlm_intent *lit;
255 const void *lmm = op_data->op_data;
256 __u32 lmmsize = op_data->op_data_size;
257 __u32 mdt_md_capsule_size;
261 int repsize, repsize_estimate;
266 mdt_md_capsule_size = obd->u.cli.cl_default_mds_easize;
268 it->it_create_mode = (it->it_create_mode & ~S_IFMT) | S_IFREG;
270 /* XXX: openlock is not cancelled for cross-refs. */
271 /* If inode is known, cancel conflicting OPEN locks. */
272 if (fid_is_sane(&op_data->op_fid2)) {
273 if (it->it_flags & MDS_OPEN_LEASE) { /* try to get lease */
274 if (it->it_flags & MDS_FMODE_WRITE)
279 if (it->it_flags & (MDS_FMODE_WRITE | MDS_OPEN_TRUNC))
282 else if (it->it_flags & FMODE_EXEC)
288 count = mdc_resource_get_unused(exp, &op_data->op_fid2,
293 /* If CREATE, cancel parent's UPDATE lock. */
294 if (it->it_op & IT_CREAT)
298 count += mdc_resource_get_unused(exp, &op_data->op_fid1,
300 MDS_INODELOCK_UPDATE);
302 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
303 &RQF_LDLM_INTENT_OPEN);
305 ldlm_lock_list_put(&cancels, l_bl_ast, count);
306 RETURN(ERR_PTR(-ENOMEM));
309 req_capsule_set_size(&req->rq_pill, &RMF_NAME, RCL_CLIENT,
310 op_data->op_namelen + 1);
311 if (cl_is_lov_delay_create(it->it_flags)) {
312 /* open(O_LOV_DELAY_CREATE) won't pack lmm */
313 LASSERT(lmmsize == 0);
314 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT, 0);
316 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT,
317 max(lmmsize, obd->u.cli.cl_default_mds_easize));
320 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
321 RCL_CLIENT, op_data->op_file_secctx_name != NULL ?
322 op_data->op_file_secctx_name_size : 0);
324 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX, RCL_CLIENT,
325 op_data->op_file_secctx_size);
327 req_capsule_set_size(&req->rq_pill, &RMF_FILE_ENCCTX, RCL_CLIENT,
328 op_data->op_file_encctx_size);
330 /* get SELinux policy info if any */
331 rc = sptlrpc_get_sepol(req);
333 ptlrpc_request_free(req);
336 req_capsule_set_size(&req->rq_pill, &RMF_SELINUX_POL, RCL_CLIENT,
337 strlen(req->rq_sepol) ?
338 strlen(req->rq_sepol) + 1 : 0);
340 rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
342 ptlrpc_request_free(req);
346 spin_lock(&req->rq_lock);
347 req->rq_replay = req->rq_import->imp_replayable;
348 spin_unlock(&req->rq_lock);
350 /* pack the intent */
351 lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
352 lit->opc = (__u64)it->it_op;
354 /* pack the intended request */
355 mdc_open_pack(req, op_data, it->it_create_mode, 0, it->it_flags, lmm,
358 req_capsule_set_size(&req->rq_pill, &RMF_MDT_MD, RCL_SERVER,
359 mdt_md_capsule_size);
360 req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, acl_bufsize);
362 if (!(it->it_op & IT_CREAT) && it->it_op & IT_OPEN &&
363 req_capsule_has_field(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
365 op_data->op_file_secctx_name_size > 0 &&
366 op_data->op_file_secctx_name != NULL) {
369 secctx_name = req_capsule_client_get(&req->rq_pill,
370 &RMF_FILE_SECCTX_NAME);
371 memcpy(secctx_name, op_data->op_file_secctx_name,
372 op_data->op_file_secctx_name_size);
373 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
375 obd->u.cli.cl_max_mds_easize);
377 CDEBUG(D_SEC, "packed '%.*s' as security xattr name\n",
378 op_data->op_file_secctx_name_size,
379 op_data->op_file_secctx_name);
382 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
386 if (exp_connect_encrypt(exp) && !(it->it_op & IT_CREAT) &&
388 req_capsule_set_size(&req->rq_pill, &RMF_FILE_ENCCTX,
390 obd->u.cli.cl_max_mds_easize);
392 req_capsule_set_size(&req->rq_pill, &RMF_FILE_ENCCTX,
396 * Inline buffer for possible data from Data-on-MDT files.
398 req_capsule_set_size(&req->rq_pill, &RMF_NIOBUF_INLINE, RCL_SERVER,
399 sizeof(struct niobuf_remote));
400 ptlrpc_request_set_replen(req);
402 /* Get real repbuf allocated size as rounded up power of 2 */
403 repsize = size_roundup_power2(req->rq_replen +
404 lustre_msg_early_size());
405 /* Estimate free space for DoM files in repbuf */
406 repsize_estimate = repsize - (req->rq_replen -
407 mdt_md_capsule_size +
408 sizeof(struct lov_comp_md_v1) +
409 sizeof(struct lov_comp_md_entry_v1) +
410 lov_mds_md_size(0, LOV_MAGIC_V3));
412 if (repsize_estimate < obd->u.cli.cl_dom_min_inline_repsize) {
413 repsize = obd->u.cli.cl_dom_min_inline_repsize -
414 repsize_estimate + sizeof(struct niobuf_remote);
415 req_capsule_set_size(&req->rq_pill, &RMF_NIOBUF_INLINE,
417 sizeof(struct niobuf_remote) + repsize);
418 ptlrpc_request_set_replen(req);
419 CDEBUG(D_INFO, "Increase repbuf by %d bytes, total: %d\n",
420 repsize, req->rq_replen);
421 repsize = size_roundup_power2(req->rq_replen +
422 lustre_msg_early_size());
424 /* The only way to report real allocated repbuf size to the server
425 * is the lm_repsize but it must be set prior buffer allocation itself
426 * due to security reasons - it is part of buffer used in signature
427 * calculation (see LU-11414). Therefore the saved size is predicted
428 * value as rq_replen rounded to the next higher power of 2.
429 * Such estimation is safe. Though the final allocated buffer might
430 * be even larger, it is not possible to know that at this point.
432 req->rq_reqmsg->lm_repsize = repsize;
436 #define GA_DEFAULT_EA_NAME_LEN 20
437 #define GA_DEFAULT_EA_VAL_LEN 250
438 #define GA_DEFAULT_EA_NUM 10
440 static struct ptlrpc_request *
441 mdc_intent_getxattr_pack(struct obd_export *exp, struct lookup_intent *it,
442 struct md_op_data *op_data)
444 struct ptlrpc_request *req;
445 struct ldlm_intent *lit;
448 u32 ea_vals_buf_size = GA_DEFAULT_EA_VAL_LEN * GA_DEFAULT_EA_NUM;
451 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
452 &RQF_LDLM_INTENT_GETXATTR);
454 RETURN(ERR_PTR(-ENOMEM));
456 /* get SELinux policy info if any */
457 rc = sptlrpc_get_sepol(req);
459 ptlrpc_request_free(req);
462 req_capsule_set_size(&req->rq_pill, &RMF_SELINUX_POL, RCL_CLIENT,
463 strlen(req->rq_sepol) ?
464 strlen(req->rq_sepol) + 1 : 0);
466 rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
468 ptlrpc_request_free(req);
472 /* pack the intent */
473 lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
474 lit->opc = IT_GETXATTR;
475 /* Message below is checked in sanity-selinux test_20d
476 * and sanity-sec test_49
478 CDEBUG(D_INFO, "%s: get xattrs for "DFID"\n",
479 exp->exp_obd->obd_name, PFID(&op_data->op_fid1));
481 #if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(3, 0, 53, 0)
482 /* If the supplied buffer is too small then the server will return
483 * -ERANGE and llite will fallback to using non cached xattr
484 * operations. On servers before 2.10.1 a (non-cached) listxattr RPC
485 * for an orphan or dead file causes an oops. So let's try to avoid
486 * sending too small a buffer to too old a server. This is effectively
487 * undoing the memory conservation of LU-9417 when it would be *more*
488 * likely to crash the server. See LU-9856.
490 if (exp->exp_connect_data.ocd_version < OBD_OCD_VERSION(2, 10, 1, 0))
491 ea_vals_buf_size = max_t(u32, ea_vals_buf_size,
492 exp->exp_connect_data.ocd_max_easize);
495 /* pack the intended request */
496 mdc_pack_body(req, &op_data->op_fid1, op_data->op_valid,
497 ea_vals_buf_size, -1, 0);
499 /* get SELinux policy info if any */
500 mdc_file_sepol_pack(req);
502 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_SERVER,
503 GA_DEFAULT_EA_NAME_LEN * GA_DEFAULT_EA_NUM);
505 req_capsule_set_size(&req->rq_pill, &RMF_EAVALS, RCL_SERVER,
508 req_capsule_set_size(&req->rq_pill, &RMF_EAVALS_LENS, RCL_SERVER,
509 sizeof(u32) * GA_DEFAULT_EA_NUM);
511 req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, 0);
513 ptlrpc_request_set_replen(req);
518 static struct ptlrpc_request *
519 mdc_intent_getattr_pack(struct obd_export *exp, struct lookup_intent *it,
520 struct md_op_data *op_data, __u32 acl_bufsize)
522 struct ptlrpc_request *req;
523 struct obd_device *obd = class_exp2obd(exp);
524 u64 valid = OBD_MD_FLGETATTR | OBD_MD_FLEASIZE | OBD_MD_FLMODEASIZE |
525 OBD_MD_FLDIREA | OBD_MD_MEA | OBD_MD_FLACL |
527 struct ldlm_intent *lit;
529 bool have_secctx = false;
533 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
534 &RQF_LDLM_INTENT_GETATTR);
536 RETURN(ERR_PTR(-ENOMEM));
538 /* send name of security xattr to get upon intent */
539 if (it->it_op & (IT_LOOKUP | IT_GETATTR) &&
540 req_capsule_has_field(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
542 op_data->op_file_secctx_name_size > 0 &&
543 op_data->op_file_secctx_name != NULL) {
545 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX_NAME,
547 op_data->op_file_secctx_name_size);
550 req_capsule_set_size(&req->rq_pill, &RMF_NAME, RCL_CLIENT,
551 op_data->op_namelen + 1);
553 rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
555 ptlrpc_request_free(req);
559 /* pack the intent */
560 lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
561 lit->opc = (__u64)it->it_op;
563 if (obd->u.cli.cl_default_mds_easize > 0)
564 easize = obd->u.cli.cl_default_mds_easize;
566 easize = obd->u.cli.cl_max_mds_easize;
568 /* pack the intended request */
569 mdc_getattr_pack(req, valid, it->it_flags, op_data, easize);
571 req_capsule_set_size(&req->rq_pill, &RMF_MDT_MD, RCL_SERVER, easize);
572 req_capsule_set_size(&req->rq_pill, &RMF_ACL, RCL_SERVER, acl_bufsize);
573 req_capsule_set_size(&req->rq_pill, &RMF_DEFAULT_MDT_MD, RCL_SERVER,
574 sizeof(struct lmv_user_md));
579 secctx_name = req_capsule_client_get(&req->rq_pill,
580 &RMF_FILE_SECCTX_NAME);
581 memcpy(secctx_name, op_data->op_file_secctx_name,
582 op_data->op_file_secctx_name_size);
584 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
587 CDEBUG(D_SEC, "packed '%.*s' as security xattr name\n",
588 op_data->op_file_secctx_name_size,
589 op_data->op_file_secctx_name);
591 req_capsule_set_size(&req->rq_pill, &RMF_FILE_SECCTX,
595 if (exp_connect_encrypt(exp) && it->it_op & (IT_LOOKUP | IT_GETATTR))
596 req_capsule_set_size(&req->rq_pill, &RMF_FILE_ENCCTX,
599 req_capsule_set_size(&req->rq_pill, &RMF_FILE_ENCCTX,
602 ptlrpc_request_set_replen(req);
606 static struct ptlrpc_request *mdc_intent_layout_pack(struct obd_export *exp,
607 struct lookup_intent *it,
608 struct md_op_data *op_data)
610 struct obd_device *obd = class_exp2obd(exp);
611 struct ptlrpc_request *req;
612 struct ldlm_intent *lit;
613 struct layout_intent *layout;
618 req = ptlrpc_request_alloc(class_exp2cliimp(exp),
619 &RQF_LDLM_INTENT_LAYOUT);
621 RETURN(ERR_PTR(-ENOMEM));
623 if (fid_is_sane(&op_data->op_fid2) && (it->it_op & IT_LAYOUT) &&
624 (it->it_flags & FMODE_WRITE)) {
625 count = mdc_resource_get_unused(exp, &op_data->op_fid2,
627 MDS_INODELOCK_LAYOUT);
630 req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT, 0);
631 rc = ldlm_prep_enqueue_req(exp, req, &cancels, count);
633 ptlrpc_request_free(req);
637 /* pack the intent */
638 lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
639 lit->opc = (__u64)it->it_op;
641 /* pack the layout intent request */
642 layout = req_capsule_client_get(&req->rq_pill, &RMF_LAYOUT_INTENT);
643 LASSERT(op_data->op_data != NULL);
644 LASSERT(op_data->op_data_size == sizeof(*layout));
645 memcpy(layout, op_data->op_data, sizeof(*layout));
647 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
648 obd->u.cli.cl_default_mds_easize);
649 ptlrpc_request_set_replen(req);
653 static struct ptlrpc_request *mdc_enqueue_pack(struct obd_export *exp,
656 struct ptlrpc_request *req;
660 req = ptlrpc_request_alloc(class_exp2cliimp(exp), &RQF_LDLM_ENQUEUE);
662 RETURN(ERR_PTR(-ENOMEM));
664 rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
666 ptlrpc_request_free(req);
670 req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER, lvb_len);
671 ptlrpc_request_set_replen(req);
675 static int mdc_finish_enqueue(struct obd_export *exp,
676 struct ptlrpc_request *req,
677 struct ldlm_enqueue_info *einfo,
678 struct lookup_intent *it,
679 struct lustre_handle *lockh, int rc)
681 struct req_capsule *pill = &req->rq_pill;
682 struct ldlm_request *lockreq;
683 struct ldlm_reply *lockrep;
684 struct ldlm_lock *lock;
685 struct mdt_body *body = NULL;
686 void *lvb_data = NULL;
691 /* Similarly, if we're going to replay this request, we don't want to
692 * actually get a lock, just perform the intent.
694 if (req->rq_transno || req->rq_replay) {
695 lockreq = req_capsule_client_get(pill, &RMF_DLM_REQ);
696 lockreq->lock_flags |= ldlm_flags_to_wire(LDLM_FL_INTENT_ONLY);
699 if (rc == ELDLM_LOCK_ABORTED) {
701 memset(lockh, 0, sizeof(*lockh));
703 } else { /* rc = 0 */
704 lock = ldlm_handle2lock(lockh);
705 LASSERT(lock != NULL);
707 /* If server returned a different lock mode, fix up variables */
708 if (lock->l_req_mode != einfo->ei_mode) {
709 ldlm_lock_addref(lockh, lock->l_req_mode);
710 ldlm_lock_decref(lockh, einfo->ei_mode);
711 einfo->ei_mode = lock->l_req_mode;
716 lockrep = req_capsule_server_get(pill, &RMF_DLM_REP);
717 LASSERT(lockrep != NULL); /* checked by ldlm_cli_enqueue() */
719 it->it_disposition = (int)lockrep->lock_policy_res1;
720 it->it_status = (int)lockrep->lock_policy_res2;
721 it->it_lock_mode = einfo->ei_mode;
722 it->it_lock_handle = lockh->cookie;
723 it->it_request = req;
725 /* Technically speaking rq_transno must already be zero if
726 * it_status is in error, so the check is a bit redundant.
728 if ((!req->rq_transno || it->it_status < 0) && req->rq_replay)
729 mdc_clear_replay_flag(req, it->it_status);
731 /* If we're doing an IT_OPEN which did not result in an actual
732 * successful open, then we need to remove the bit which saves
733 * this request for unconditional replay.
735 * It's important that we do this first! Otherwise we might exit the
736 * function without doing so, and try to replay a failed create.
739 if (it->it_op & IT_OPEN && req->rq_replay &&
740 (!it_disposition(it, DISP_OPEN_OPEN) || it->it_status != 0))
741 mdc_clear_replay_flag(req, it->it_status);
743 DEBUG_REQ(D_RPCTRACE, req, "op=%x disposition=%x, status=%d",
744 it->it_op, it->it_disposition, it->it_status);
746 /* We know what to expect, so we do any byte flipping required here */
747 if (it_has_reply_body(it)) {
748 body = req_capsule_server_get(pill, &RMF_MDT_BODY);
751 CERROR("%s: cannot swab mdt_body: rc = %d\n",
752 exp->exp_obd->obd_name, rc);
756 if (it_disposition(it, DISP_OPEN_OPEN) &&
757 !it_open_error(DISP_OPEN_OPEN, it)) {
759 * If this is a successful OPEN request, we need to set
760 * replay handler and data early, so that if replay
761 * happens immediately after swabbing below, new reply
762 * is swabbed by that handler correctly.
764 mdc_set_open_replay_data(NULL, NULL, it);
767 if (it_disposition(it, DISP_OPEN_CREATE) &&
768 !it_open_error(DISP_OPEN_CREATE, it)) {
769 lprocfs_counter_incr(exp->exp_obd->obd_md_stats,
773 if (body->mbo_valid & (OBD_MD_FLDIREA | OBD_MD_FLEASIZE)) {
776 mdc_update_max_ea_from_body(exp, body);
779 * The eadata is opaque; just check that it is there.
780 * Eventually, obd_unpackmd() will check the contents.
782 eadata = req_capsule_server_sized_get(pill, &RMF_MDT_MD,
783 body->mbo_eadatasize);
787 /* save LVB data and length if for layout lock */
789 lvb_len = body->mbo_eadatasize;
792 * We save the reply LOV EA in case we have to replay a
793 * create for recovery. If we didn't allocate a large
794 * enough request buffer above we need to reallocate it
795 * here to hold the actual LOV EA.
797 * To not save LOV EA if request is not going to replay
798 * (for example error one).
800 if ((it->it_op & IT_OPEN) && req->rq_replay) {
801 rc = mdc_save_lovea(req, eadata,
802 body->mbo_eadatasize);
804 body->mbo_valid &= ~OBD_MD_FLEASIZE;
805 body->mbo_eadatasize = 0;
810 } else if (it->it_op & IT_LAYOUT) {
811 /* maybe the lock was granted right away and layout
812 * is packed into RMF_DLM_LVB of req
814 lvb_len = req_capsule_get_size(pill, &RMF_DLM_LVB, RCL_SERVER);
815 CDEBUG(D_INFO, "%s: layout return lvb %d transno %lld\n",
816 class_exp2obd(exp)->obd_name, lvb_len, req->rq_transno);
818 lvb_data = req_capsule_server_sized_get(pill,
819 &RMF_DLM_LVB, lvb_len);
820 if (lvb_data == NULL)
824 * save replied layout data to the request buffer for
825 * recovery consideration (lest MDS reinitialize
826 * another set of OST objects).
829 (void)mdc_save_lovea(req, lvb_data, lvb_len);
833 /* fill in stripe data for layout lock.
834 * LU-6581: trust layout data only if layout lock is granted. The MDT
835 * has stopped sending layout unless the layout lock is granted. The
836 * client still does this checking in case it's talking with an old
839 lock = ldlm_handle2lock(lockh);
843 if (ldlm_has_layout(lock) && lvb_data != NULL &&
844 !(lockrep->lock_flags & LDLM_FL_BLOCKED_MASK)) {
847 LDLM_DEBUG(lock, "layout lock returned by: %s, lvb_len: %d",
848 ldlm_it2str(it->it_op), lvb_len);
850 OBD_ALLOC_LARGE(lmm, lvb_len);
852 GOTO(out_lock, rc = -ENOMEM);
854 memcpy(lmm, lvb_data, lvb_len);
856 /* install lvb_data */
857 lock_res_and_lock(lock);
858 if (lock->l_lvb_data == NULL) {
859 lock->l_lvb_type = LVB_T_LAYOUT;
860 lock->l_lvb_data = lmm;
861 lock->l_lvb_len = lvb_len;
864 unlock_res_and_lock(lock);
866 OBD_FREE_LARGE(lmm, lvb_len);
869 if (ldlm_has_dom(lock)) {
870 LASSERT(lock->l_glimpse_ast == mdc_ldlm_glimpse_ast);
872 body = req_capsule_server_get(pill, &RMF_MDT_BODY);
873 if (!(body->mbo_valid & OBD_MD_DOM_SIZE)) {
874 LDLM_ERROR(lock, "%s: DoM lock without size.",
875 exp->exp_obd->obd_name);
876 GOTO(out_lock, rc = -EPROTO);
879 LDLM_DEBUG(lock, "DoM lock is returned by: %s, size: %llu",
880 ldlm_it2str(it->it_op), body->mbo_dom_size);
882 lock_res_and_lock(lock);
883 mdc_body2lvb(body, &lock->l_ost_lvb);
884 ldlm_lock_allow_match_locked(lock);
885 unlock_res_and_lock(lock);
893 static inline bool mdc_skip_mod_rpc_slot(const struct lookup_intent *it)
896 (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
897 it->it_op == IT_READDIR ||
898 (it->it_op == IT_LAYOUT && !(it->it_flags & MDS_FMODE_WRITE))))
903 /* We always reserve enough space in the reply packet for a stripe MD, because
904 * we don't know in advance the file type.
906 static int mdc_enqueue_base(struct obd_export *exp,
907 struct ldlm_enqueue_info *einfo,
908 const union ldlm_policy_data *policy,
909 struct lookup_intent *it,
910 struct md_op_data *op_data,
911 struct lustre_handle *lockh,
912 __u64 extra_lock_flags)
914 struct obd_device *obd = class_exp2obd(exp);
915 struct ptlrpc_request *req;
916 __u64 flags, saved_flags = extra_lock_flags;
917 struct ldlm_res_id res_id;
918 static const union ldlm_policy_data lookup_policy = {
919 .l_inodebits = { MDS_INODELOCK_LOOKUP } };
920 static const union ldlm_policy_data update_policy = {
921 .l_inodebits = { MDS_INODELOCK_UPDATE } };
922 static const union ldlm_policy_data layout_policy = {
923 .l_inodebits = { MDS_INODELOCK_LAYOUT } };
924 static const union ldlm_policy_data getxattr_policy = {
925 .l_inodebits = { MDS_INODELOCK_XATTR } };
926 int generation, resends = 0;
927 struct ldlm_reply *lockrep;
928 struct obd_import *imp = class_exp2cliimp(exp);
930 enum lvb_type lvb_type = 0;
934 LASSERTF(!it || einfo->ei_type == LDLM_IBITS, "lock type %d\n",
936 fid_build_reg_res_name(&op_data->op_fid1, &res_id);
939 LASSERT(policy == NULL);
941 saved_flags |= LDLM_FL_HAS_INTENT;
942 if (it->it_op & (IT_GETATTR | IT_READDIR))
943 policy = &update_policy;
944 else if (it->it_op & IT_LAYOUT)
945 policy = &layout_policy;
946 else if (it->it_op & IT_GETXATTR)
947 policy = &getxattr_policy;
949 policy = &lookup_policy;
952 generation = obd->u.cli.cl_import->imp_generation;
953 if (!it || (it->it_op & (IT_OPEN | IT_CREAT)))
954 acl_bufsize = min_t(__u32, imp->imp_connect_data.ocd_max_easize,
957 acl_bufsize = LUSTRE_POSIX_ACL_MAX_SIZE_OLD;
962 /* The only way right now is FLOCK. */
963 LASSERTF(einfo->ei_type == LDLM_FLOCK, "lock type %d\n",
965 res_id.name[3] = LDLM_FLOCK;
966 req = ldlm_enqueue_pack(exp, 0);
967 } else if (it->it_op & IT_OPEN) {
968 req = mdc_intent_open_pack(exp, it, op_data, acl_bufsize);
969 } else if (it->it_op & (IT_GETATTR | IT_LOOKUP)) {
970 req = mdc_intent_getattr_pack(exp, it, op_data, acl_bufsize);
971 } else if (it->it_op & IT_READDIR) {
972 req = mdc_enqueue_pack(exp, 0);
973 } else if (it->it_op & IT_LAYOUT) {
974 if (!imp_connect_lvb_type(imp))
976 req = mdc_intent_layout_pack(exp, it, op_data);
977 lvb_type = LVB_T_LAYOUT;
978 } else if (it->it_op & IT_GETXATTR) {
979 req = mdc_intent_getxattr_pack(exp, it, op_data);
986 RETURN(PTR_ERR(req));
989 req->rq_generation_set = 1;
990 req->rq_import_generation = generation;
991 req->rq_sent = ktime_get_real_seconds() + resends;
994 einfo->ei_enq_slot = !mdc_skip_mod_rpc_slot(it);
996 /* With Data-on-MDT the glimpse callback is needed too.
997 * It is set here in advance but not in mdc_finish_enqueue()
998 * to avoid possible races. It is safe to have glimpse handler
999 * for non-DOM locks and costs nothing.
1001 if (einfo->ei_cb_gl == NULL)
1002 einfo->ei_cb_gl = mdc_ldlm_glimpse_ast;
1004 rc = ldlm_cli_enqueue(exp, &req, einfo, &res_id, policy, &flags, NULL,
1005 0, lvb_type, lockh, 0);
1008 /* For flock requests we immediatelly return without further
1009 * delay and let caller deal with the rest, since rest of
1010 * this function metadata processing makes no sense for flock
1011 * requests anyway. But in case of problem during comms with
1012 * server (-ETIMEDOUT) or any signal/kill attempt (-EINTR),
1013 * we cannot rely on caller and this mainly for F_UNLCKs
1014 * (explicits or automatically generated by kernel to clean
1015 * current flocks upon exit) that can't be trashed.
1017 ptlrpc_req_finished(req);
1018 if (((rc == -EINTR) || (rc == -ETIMEDOUT)) &&
1019 (einfo->ei_type == LDLM_FLOCK) &&
1020 (einfo->ei_mode == LCK_NL))
1027 "%s: ldlm_cli_enqueue "DFID":"DFID"=%s failed: rc = %d\n",
1028 obd->obd_name, PFID(&op_data->op_fid1),
1029 PFID(&op_data->op_fid2), op_data->op_name ?: "", rc);
1031 mdc_clear_replay_flag(req, rc);
1032 ptlrpc_req_finished(req);
1036 lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1037 LASSERT(lockrep != NULL);
1039 lockrep->lock_policy_res2 =
1040 ptlrpc_status_ntoh(lockrep->lock_policy_res2);
1042 /* Retry infinitely when the server returns -EINPROGRESS for the
1043 * intent operation, when server returns -EINPROGRESS for acquiring
1044 * intent lock, we'll retry in after_reply().
1046 if (it && (int)lockrep->lock_policy_res2 == -EINPROGRESS) {
1047 mdc_clear_replay_flag(req, rc);
1048 ptlrpc_req_finished(req);
1049 if (generation == obd->u.cli.cl_import->imp_generation) {
1050 if (signal_pending(current))
1054 CDEBUG(D_HA, "%s: resend:%d op:%d "DFID"/"DFID"\n",
1055 obd->obd_name, resends, it->it_op,
1056 PFID(&op_data->op_fid1),
1057 PFID(&op_data->op_fid2));
1060 CDEBUG(D_HA, "resend cross eviction\n");
1065 if ((int)lockrep->lock_policy_res2 == -ERANGE &&
1066 it->it_op & (IT_OPEN | IT_GETATTR | IT_LOOKUP) &&
1067 acl_bufsize == LUSTRE_POSIX_ACL_MAX_SIZE_OLD) {
1068 mdc_clear_replay_flag(req, -ERANGE);
1069 ptlrpc_req_finished(req);
1070 acl_bufsize = min_t(__u32, imp->imp_connect_data.ocd_max_easize,
1075 rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
1077 if (lustre_handle_is_used(lockh)) {
1078 ldlm_lock_decref(lockh, einfo->ei_mode);
1079 memset(lockh, 0, sizeof(*lockh));
1081 ptlrpc_req_finished(req);
1083 it->it_lock_handle = 0;
1084 it->it_lock_mode = 0;
1085 it->it_request = NULL;
1091 int mdc_enqueue(struct obd_export *exp, struct ldlm_enqueue_info *einfo,
1092 const union ldlm_policy_data *policy,
1093 struct md_op_data *op_data,
1094 struct lustre_handle *lockh, __u64 extra_lock_flags)
1096 return mdc_enqueue_base(exp, einfo, policy, NULL,
1097 op_data, lockh, extra_lock_flags);
1100 static int mdc_finish_intent_lock(struct obd_export *exp,
1101 struct ptlrpc_request *request,
1102 struct md_op_data *op_data,
1103 struct lookup_intent *it,
1104 struct lustre_handle *lockh)
1106 struct lustre_handle old_lock;
1107 struct ldlm_lock *lock;
1111 LASSERT(request != NULL);
1112 LASSERT(request != LP_POISON);
1113 LASSERT(request->rq_repmsg != LP_POISON);
1115 if (it->it_op & IT_READDIR)
1118 if (it->it_op & (IT_GETXATTR | IT_LAYOUT)) {
1119 if (it->it_status != 0)
1120 GOTO(out, rc = it->it_status);
1122 if (!it_disposition(it, DISP_IT_EXECD)) {
1123 /* The server failed before it even started executing
1124 * the intent, i.e. because it couldn't unpack the
1127 LASSERT(it->it_status != 0);
1128 GOTO(out, rc = it->it_status);
1130 rc = it_open_error(DISP_IT_EXECD, it);
1134 rc = it_open_error(DISP_LOOKUP_EXECD, it);
1138 /* keep requests around for the multiple phases of the call
1139 * this shows the DISP_XX must guarantee we make it into the
1142 if (!it_disposition(it, DISP_ENQ_CREATE_REF) &&
1143 it_disposition(it, DISP_OPEN_CREATE) &&
1144 !it_open_error(DISP_OPEN_CREATE, it)) {
1145 it_set_disposition(it, DISP_ENQ_CREATE_REF);
1146 /* balanced in ll_create_node */
1147 ptlrpc_request_addref(request);
1149 if (!it_disposition(it, DISP_ENQ_OPEN_REF) &&
1150 it_disposition(it, DISP_OPEN_OPEN) &&
1151 !it_open_error(DISP_OPEN_OPEN, it)) {
1152 it_set_disposition(it, DISP_ENQ_OPEN_REF);
1153 /* balanced in ll_file_open */
1154 ptlrpc_request_addref(request);
1155 /* eviction in middle of open RPC processing b=11546 */
1156 OBD_FAIL_TIMEOUT(OBD_FAIL_MDC_ENQUEUE_PAUSE,
1160 if (it->it_op & IT_CREAT) {
1161 /* XXX this belongs in ll_create_it */
1162 } else if (it->it_op == IT_OPEN) {
1163 LASSERT(!it_disposition(it, DISP_OPEN_CREATE));
1165 LASSERT(it->it_op & (IT_GETATTR | IT_LOOKUP));
1169 /* If we already have a matching lock, then cancel the new
1170 * one. We have to set the data here instead of in
1171 * mdc_enqueue, because we need to use the child's inode as
1172 * the l_ast_data to match, and that's not available until
1173 * intent_finish has performed the iget().
1175 lock = ldlm_handle2lock(lockh);
1177 union ldlm_policy_data policy = lock->l_policy_data;
1179 LDLM_DEBUG(lock, "matching against this");
1181 if (it_has_reply_body(it)) {
1182 struct mdt_body *body;
1184 body = req_capsule_server_get(&request->rq_pill,
1186 /* mdc_enqueue checked */
1187 LASSERT(body != NULL);
1188 LASSERTF(fid_res_name_eq(&body->mbo_fid1,
1189 &lock->l_resource->lr_name),
1190 "Lock res_id: "DLDLMRES", fid: "DFID"\n",
1191 PLDLMRES(lock->l_resource),
1192 PFID(&body->mbo_fid1));
1194 LDLM_LOCK_PUT(lock);
1196 memcpy(&old_lock, lockh, sizeof(*lockh));
1197 if (ldlm_lock_match(NULL, LDLM_FL_BLOCK_GRANTED, NULL,
1198 LDLM_IBITS, &policy, LCK_NL, &old_lock)) {
1199 ldlm_lock_decref_and_cancel(lockh, it->it_lock_mode);
1200 memcpy(lockh, &old_lock, sizeof(old_lock));
1201 it->it_lock_handle = lockh->cookie;
1208 "D_IT dentry=%.*s intent=%s status=%d disp=%x: rc = %d\n",
1209 (int)op_data->op_namelen, op_data->op_name,
1210 ldlm_it2str(it->it_op), it->it_status, it->it_disposition, rc);
1215 int mdc_revalidate_lock(struct obd_export *exp, struct lookup_intent *it,
1216 struct lu_fid *fid, __u64 *bits)
1218 /* We could just return 1 immediately, but as we should only be called
1219 * in revalidate_it if we already have a lock, let's verify that.
1221 struct ldlm_res_id res_id;
1222 struct lustre_handle lockh;
1223 union ldlm_policy_data policy;
1224 enum ldlm_mode mode;
1227 if (it->it_lock_handle) {
1228 lockh.cookie = it->it_lock_handle;
1229 mode = ldlm_revalidate_lock_handle(&lockh, bits);
1231 fid_build_reg_res_name(fid, &res_id);
1232 switch (it->it_op) {
1234 /* File attributes are held under multiple bits:
1235 * nlink is under lookup lock, size and times are
1236 * under UPDATE lock and recently we've also got
1237 * a separate permissions lock for owner/group/acl that
1238 * were protected by lookup lock before.
1239 * Getattr must provide all of that information,
1240 * so we need to ensure we have all of those locks.
1241 * Unfortunately, if the bits are split across multiple
1242 * locks, there's no easy way to match all of them here,
1243 * so an extra RPC would be performed to fetch all
1244 * of those bits at once for now.
1246 /* For new MDTs(> 2.4), UPDATE|PERM should be enough,
1247 * but for old MDTs (< 2.4), permission is covered
1248 * by LOOKUP lock, so it needs to match all bits here.
1250 policy.l_inodebits.bits = MDS_INODELOCK_UPDATE |
1254 policy.l_inodebits.bits = MDS_INODELOCK_UPDATE;
1257 policy.l_inodebits.bits = MDS_INODELOCK_LAYOUT;
1260 policy.l_inodebits.bits = MDS_INODELOCK_LOOKUP;
1264 mode = mdc_lock_match(exp, LDLM_FL_BLOCK_GRANTED, fid,
1265 LDLM_IBITS, &policy,
1266 LCK_CR | LCK_CW | LCK_PR | LCK_PW,
1271 it->it_lock_handle = lockh.cookie;
1272 it->it_lock_mode = mode;
1274 it->it_lock_handle = 0;
1275 it->it_lock_mode = 0;
1282 * This long block is all about fixing up the lock and request state
1283 * so that it is correct as of the moment _before_ the operation was
1284 * applied; that way, the VFS will think that everything is normal and
1285 * call Lustre's regular VFS methods.
1287 * If we're performing a creation, that means that unless the creation
1288 * failed with EEXIST, we should fake up a negative dentry.
1290 * For everything else, we want to lookup to succeed.
1292 * One additional note: if CREATE or OPEN succeeded, we add an extra
1293 * reference to the request because we need to keep it around until
1294 * ll_create/ll_open gets called.
1296 * The server will return to us, in it_disposition, an indication of
1297 * exactly what it_status refers to.
1299 * If DISP_OPEN_OPEN is set, then it_status refers to the open() call,
1300 * otherwise if DISP_OPEN_CREATE is set, then it status is the
1301 * creation failure mode. In either case, one of DISP_LOOKUP_NEG or
1302 * DISP_LOOKUP_POS will be set, indicating whether the child lookup
1305 * Else, if DISP_LOOKUP_EXECD then it_status is the rc of the
1308 int mdc_intent_lock(struct obd_export *exp, struct md_op_data *op_data,
1309 struct lookup_intent *it, struct ptlrpc_request **reqp,
1310 ldlm_blocking_callback cb_blocking, __u64 extra_lock_flags)
1312 struct ldlm_enqueue_info einfo = {
1313 .ei_type = LDLM_IBITS,
1314 .ei_mode = it_to_lock_mode(it),
1315 .ei_cb_bl = cb_blocking,
1316 .ei_cb_cp = ldlm_completion_ast,
1317 .ei_cb_gl = mdc_ldlm_glimpse_ast,
1319 struct lustre_handle lockh;
1324 CDEBUG(D_DLMTRACE, "(name: %.*s,"DFID") in obj "DFID
1325 ", intent: %s flags %#llo\n", (int)op_data->op_namelen,
1326 op_data->op_name, PFID(&op_data->op_fid2),
1327 PFID(&op_data->op_fid1), ldlm_it2str(it->it_op),
1331 if (fid_is_sane(&op_data->op_fid2) &&
1332 (it->it_op & (IT_LOOKUP | IT_GETATTR | IT_READDIR))) {
1333 /* We could just return 1 immediately, but since we should only
1334 * be called in revalidate_it if we already have a lock, let's
1337 it->it_lock_handle = 0;
1338 rc = mdc_revalidate_lock(exp, it, &op_data->op_fid2, NULL);
1339 /* Only return failure if it was not GETATTR by cfid
1340 * (from inode_revalidate()).
1342 if (rc || op_data->op_namelen != 0)
1346 /* For case if upper layer did not alloc fid, do it now. */
1347 if (!fid_is_sane(&op_data->op_fid2) && it->it_op & IT_CREAT) {
1348 rc = mdc_fid_alloc(NULL, exp, &op_data->op_fid2, op_data);
1350 CERROR("%s: cannot allocate new FID: rc=%d\n",
1351 exp->exp_obd->obd_name, rc);
1356 rc = mdc_enqueue_base(exp, &einfo, NULL, it, op_data, &lockh,
1361 *reqp = it->it_request;
1362 rc = mdc_finish_intent_lock(exp, *reqp, op_data, it, &lockh);
1366 static int mdc_intent_getattr_async_interpret(const struct lu_env *env,
1367 struct ptlrpc_request *req,
1370 struct mdc_getattr_args *ga = args;
1371 struct obd_export *exp = ga->ga_exp;
1372 struct md_enqueue_info *minfo = ga->ga_minfo;
1373 struct ldlm_enqueue_info *einfo = &minfo->mi_einfo;
1374 struct lookup_intent *it = &minfo->mi_it;
1375 struct lustre_handle *lockh = &minfo->mi_lockh;
1376 struct ldlm_reply *lockrep;
1377 __u64 flags = LDLM_FL_HAS_INTENT;
1380 if (OBD_FAIL_CHECK(OBD_FAIL_MDC_GETATTR_ENQUEUE))
1383 rc = ldlm_cli_enqueue_fini(exp, req, einfo, 1, &flags, NULL, 0,
1386 CERROR("%s: ldlm_cli_enqueue_fini() failed: rc = %d\n",
1387 exp->exp_obd->obd_name, rc);
1388 mdc_clear_replay_flag(req, rc);
1392 lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
1393 LASSERT(lockrep != NULL);
1395 lockrep->lock_policy_res2 =
1396 ptlrpc_status_ntoh(lockrep->lock_policy_res2);
1398 rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
1402 rc = mdc_finish_intent_lock(exp, req, &minfo->mi_data, it, lockh);
1406 minfo->mi_cb(req, minfo, rc);
1410 int mdc_intent_getattr_async(struct obd_export *exp,
1411 struct md_enqueue_info *minfo)
1413 struct md_op_data *op_data = &minfo->mi_data;
1414 struct lookup_intent *it = &minfo->mi_it;
1415 struct ptlrpc_request *req;
1416 struct mdc_getattr_args *ga;
1417 struct ldlm_res_id res_id;
1418 union ldlm_policy_data policy = {
1419 .l_inodebits = { MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE }
1421 __u64 flags = LDLM_FL_HAS_INTENT;
1426 "name: %.*s in inode "DFID", intent: %s flags %#llo\n",
1427 (int)op_data->op_namelen, op_data->op_name,
1428 PFID(&op_data->op_fid1), ldlm_it2str(it->it_op), it->it_flags);
1430 fid_build_reg_res_name(&op_data->op_fid1, &res_id);
1431 /* If the MDT return -ERANGE because of large ACL, then the sponsor
1432 * of the async getattr RPC will handle that by itself.
1434 req = mdc_intent_getattr_pack(exp, it, op_data,
1435 LUSTRE_POSIX_ACL_MAX_SIZE_OLD);
1437 RETURN(PTR_ERR(req));
1439 /* With Data-on-MDT the glimpse callback is needed too.
1440 * It is set here in advance but not in mdc_finish_enqueue()
1441 * to avoid possible races. It is safe to have glimpse handler
1442 * for non-DOM locks and costs nothing.
1444 if (minfo->mi_einfo.ei_cb_gl == NULL)
1445 minfo->mi_einfo.ei_cb_gl = mdc_ldlm_glimpse_ast;
1447 rc = ldlm_cli_enqueue(exp, &req, &minfo->mi_einfo, &res_id, &policy,
1448 &flags, NULL, 0, LVB_T_NONE, &minfo->mi_lockh, 1);
1450 ptlrpc_req_finished(req);
1454 ga = ptlrpc_req_async_args(ga, req);
1456 ga->ga_minfo = minfo;
1458 req->rq_interpret_reply = mdc_intent_getattr_async_interpret;
1459 ptlrpcd_add_req(req);