1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
6 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 2 only,
10 * as published by the Free Software Foundation.
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License version 2 for more details (a copy is included
16 * in the LICENSE file that accompanied this code).
18 * You should have received a copy of the GNU General Public License
19 * version 2 along with this program; If not, see
20 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
22 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
23 * CA 95054 USA or visit www.sun.com if you need additional information or
29 * Copyright 2008 Sun Microsystems, Inc. All rights reserved
30 * Use is subject to license terms.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
36 * lustre/cmm/cmm_split.c
38 * Lustre splitting dir
40 * Author: Alex Thomas <alex@clusterfs.com>
41 * Author: Wang Di <wangdi@clusterfs.com>
42 * Author: Yury Umanets <umka@clusterfs.com>
46 # define EXPORT_SYMTAB
49 #define DEBUG_SUBSYSTEM S_MDS
51 #include <obd_class.h>
52 #include <lustre_fid.h>
53 #include <lustre_mds.h>
54 #include <lustre/lustre_idl.h>
55 #include "cmm_internal.h"
56 #include "mdc_internal.h"
59 CMM_SPLIT_SIZE = 128 * 1024
63 * This function checks if passed @name come to correct server (local MDT). If
64 * not - return -ERESTART and let client know that dir was split and client
65 * needs to chose correct stripe.
67 int cmm_split_check(const struct lu_env *env, struct md_object *mp,
70 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mp));
71 struct md_attr *ma = &cmm_env_info(env)->cmi_ma;
72 struct cml_object *clo = md2cml_obj(mp);
76 cmm_lprocfs_time_start(env);
79 if (clo->clo_split == CMM_SPLIT_NONE ||
80 clo->clo_split == CMM_SPLIT_DENIED)
83 lmv_size = CMM_MD_SIZE(cmm->cmm_tgt_count + 1);
85 /* Try to get the LMV EA */
86 memset(ma, 0, sizeof(*ma));
89 ma->ma_lmv_size = lmv_size;
90 OBD_ALLOC(ma->ma_lmv, lmv_size);
91 if (ma->ma_lmv == NULL)
92 GOTO(out, rc = -ENOMEM);
94 /* Get LMV EA, Note: refresh valid here for getting LMV_EA */
95 rc = mo_attr_get(env, mp, ma);
99 /* No LMV just return */
100 if (!(ma->ma_valid & MA_LMV)) {
101 /* update split state if unknown */
102 if (clo->clo_split == CMM_SPLIT_UNKNOWN)
103 clo->clo_split = CMM_SPLIT_NONE;
104 GOTO(cleanup, rc = 0);
107 /* Skip checking the slave dirs (mea_count is 0) */
108 if (ma->ma_lmv->mea_count != 0) {
112 * Get stripe by name to check the name belongs to master dir,
113 * otherwise return the -ERESTART
115 idx = mea_name2idx(ma->ma_lmv, name, strlen(name));
118 * Check if name came to correct MDT server. We suppose that if
119 * client does not know about split, it sends create operation
120 * to master MDT. And this is master job to say it that dir got
121 * split and client should orward request to correct MDT. This
122 * is why we check here if stripe zero or not. Zero stripe means
123 * master stripe. If stripe calculated from name is not zero -
129 /* update split state to DONE if unknown */
130 if (clo->clo_split == CMM_SPLIT_UNKNOWN)
131 clo->clo_split = CMM_SPLIT_DONE;
133 /* split is denied for slave dir */
134 clo->clo_split = CMM_SPLIT_DENIED;
138 OBD_FREE(ma->ma_lmv, lmv_size);
140 cmm_lprocfs_time_end(env, cmm, LPROC_CMM_SPLIT_CHECK);
145 * Return preferable access mode to caller taking into account possible split
146 * and the fact of existing not splittable dirs in principle.
148 int cmm_split_access(const struct lu_env *env, struct md_object *mo,
151 struct md_attr *ma = &cmm_env_info(env)->cmi_ma;
155 memset(ma, 0, sizeof(*ma));
158 * Check only if we need protection from split. If not - mdt handles
161 rc = cmm_split_expect(env, mo, ma, &split);
163 CERROR("Can't check for possible split, rc %d\n", rc);
168 * Do not take PDO lock on non-splittable objects if this is not PW,
169 * this should speed things up a bit.
171 if (split == CMM_SPLIT_DONE && lm != MDL_PW)
174 /* Protect splitting by exclusive lock. */
175 if (split == CMM_SPLIT_NEEDED && lm == MDL_PW)
179 * Have no idea about lock mode, let it be what higher layer wants.
184 /* Check if split is expected for current thread. */
185 int cmm_split_expect(const struct lu_env *env, struct md_object *mo,
186 struct md_attr *ma, int *split)
188 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
189 struct cml_object *clo = md2cml_obj(mo);
190 struct lu_fid root_fid;
194 if (clo->clo_split == CMM_SPLIT_DONE ||
195 clo->clo_split == CMM_SPLIT_DENIED) {
196 *split = clo->clo_split;
199 /* CMM_SPLIT_UNKNOWN case below */
201 /* No need to split root object. */
202 rc = cmm_child_ops(cmm)->mdo_root_get(env, cmm->cmm_child,
207 if (lu_fid_eq(&root_fid, cmm2fid(md2cmm_obj(mo)))) {
208 /* update split state */
209 *split = clo->clo_split == CMM_SPLIT_DENIED;
214 * Assumption: ma_valid = 0 here, we only need get inode and lmv_size
217 LASSERT(ma->ma_valid == 0);
218 ma->ma_need = MA_INODE | MA_LMV;
219 rc = mo_attr_get(env, mo, ma);
223 /* No need split for already split object */
224 if (ma->ma_valid & MA_LMV) {
225 LASSERT(ma->ma_lmv_size > 0);
226 *split = clo->clo_split = CMM_SPLIT_DONE;
230 /* No need split for object whose size < CMM_SPLIT_SIZE */
231 if (ma->ma_attr.la_size < CMM_SPLIT_SIZE) {
232 *split = clo->clo_split = CMM_SPLIT_NONE;
236 *split = clo->clo_split = CMM_SPLIT_NEEDED;
240 struct cmm_object *cmm_object_find(const struct lu_env *env,
241 struct cmm_device *d,
242 const struct lu_fid *f)
244 return md2cmm_obj(md_object_find_slice(env, &d->cmm_md_dev, fid));
247 static inline void cmm_object_put(const struct lu_env *env,
248 struct cmm_object *o)
250 lu_object_put(env, &o->cmo_obj.mo_lu);
254 * Allocate new on passed @mc for slave object which is going to create there
257 static int cmm_split_fid_alloc(const struct lu_env *env,
258 struct cmm_device *cmm,
259 struct mdc_device *mc,
265 LASSERT(cmm != NULL && mc != NULL && fid != NULL);
267 down(&mc->mc_fid_sem);
269 /* Alloc new fid on @mc. */
270 rc = obd_fid_alloc(mc->mc_desc.cl_exp, fid, NULL);
278 /* Allocate new slave object on passed @mc */
279 static int cmm_split_slave_create(const struct lu_env *env,
280 struct cmm_device *cmm,
281 struct mdc_device *mc,
284 struct lmv_stripe_md *lmv,
287 struct md_op_spec *spec = &cmm_env_info(env)->cmi_spec;
288 struct cmm_object *obj;
292 /* Allocate new fid and store it to @fid */
293 rc = cmm_split_fid_alloc(env, cmm, mc, fid);
295 CERROR("Can't alloc new fid on "LPU64
296 ", rc %d\n", mc->mc_num, rc);
300 /* Allocate new object on @mc */
301 obj = cmm_object_find(env, cmm, fid);
303 RETURN(PTR_ERR(obj));
305 memset(spec, 0, sizeof *spec);
306 spec->u.sp_ea.fid = fid;
307 spec->u.sp_ea.eadata = lmv;
308 spec->u.sp_ea.eadatalen = lmv_size;
309 spec->sp_cr_flags |= MDS_CREATE_SLAVE_OBJ;
310 rc = mo_object_create(env, md_object_next(&obj->cmo_obj),
312 cmm_object_put(env, obj);
317 * Create so many slaves as number of stripes. This is called in split time
318 * before sending pages to slaves.
320 static int cmm_split_slaves_create(const struct lu_env *env,
321 struct md_object *mo,
324 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
325 struct lu_fid *lf = cmm2fid(md2cmm_obj(mo));
326 struct lmv_stripe_md *slave_lmv = &cmm_env_info(env)->cmi_lmv;
327 struct mdc_device *mc, *tmp;
328 struct lmv_stripe_md *lmv;
332 /* Init the split MEA */
334 lmv->mea_master = cmm->cmm_local_num;
335 lmv->mea_magic = MEA_MAGIC_HASH_SEGMENT;
336 lmv->mea_count = cmm->cmm_tgt_count + 1;
339 * Store master FID to local node idx number. Local node is always
340 * master and its stripe number if 0.
342 lmv->mea_ids[0] = *lf;
344 memset(slave_lmv, 0, sizeof *slave_lmv);
345 slave_lmv->mea_master = cmm->cmm_local_num;
346 slave_lmv->mea_magic = MEA_MAGIC_HASH_SEGMENT;
347 slave_lmv->mea_count = 0;
349 list_for_each_entry_safe(mc, tmp, &cmm->cmm_targets, mc_linkage) {
350 rc = cmm_split_slave_create(env, cmm, mc, &lmv->mea_ids[i],
351 ma, slave_lmv, sizeof(*slave_lmv));
361 static inline int cmm_split_special_entry(struct lu_dirent *ent)
363 if (!strncmp(ent->lde_name, ".", le16_to_cpu(ent->lde_namelen)) ||
364 !strncmp(ent->lde_name, "..", le16_to_cpu(ent->lde_namelen)))
369 static inline struct lu_name *cmm_name(const struct lu_env *env,
370 char *name, int buflen)
372 struct lu_name *lname;
373 struct cmm_thread_info *cmi;
376 LASSERT(name[buflen - 1] == '\0');
378 cmi = cmm_env_info(env);
379 lname = &cmi->cti_name;
380 lname->ln_name = name;
381 /* do NOT count the terminating '\0' of name for length */
382 lname->ln_namelen = buflen - 1;
387 * Remove one entry from local MDT. Do not corrupt byte order in page, it will
388 * be sent to remote MDT.
390 static int cmm_split_remove_entry(const struct lu_env *env,
391 struct md_object *mo,
392 struct lu_dirent *ent)
394 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
395 struct cmm_thread_info *cmi;
397 struct cmm_object *obj;
400 struct lu_name *lname;
403 if (cmm_split_special_entry(ent))
406 fid_le_to_cpu(&cmm_env_info(env)->cmi_fid, &ent->lde_fid);
407 obj = cmm_object_find(env, cmm, &cmm_env_info(env)->cmi_fid);
409 RETURN(PTR_ERR(obj));
411 cmi = cmm_env_info(env);
414 if (lu_object_exists(&obj->cmo_obj.mo_lu) > 0)
415 is_dir = S_ISDIR(lu_object_attr(&obj->cmo_obj.mo_lu));
418 * XXX: These days only cross-ref dirs are possible, so for the
419 * sake of simplicity, in split, we suppose that all cross-ref
420 * names point to directory and do not do additional getattr to
425 OBD_ALLOC(name, le16_to_cpu(ent->lde_namelen) + 1);
427 GOTO(cleanup, rc = -ENOMEM);
429 memcpy(name, ent->lde_name, le16_to_cpu(ent->lde_namelen));
430 lname = cmm_name(env, name, le16_to_cpu(ent->lde_namelen) + 1);
432 * When split, no need update parent's ctime,
433 * and no permission check for name_remove.
435 ma->ma_attr.la_ctime = 0;
437 ma->ma_attr.la_mode = S_IFDIR;
439 ma->ma_attr.la_mode = 0;
440 ma->ma_attr.la_valid = LA_MODE;
441 ma->ma_valid = MA_INODE;
443 ma->ma_attr_flags |= MDS_PERM_BYPASS;
444 rc = mdo_name_remove(env, md_object_next(mo), lname, ma);
445 OBD_FREE(name, le16_to_cpu(ent->lde_namelen) + 1);
450 * This @ent will be transferred to slave MDS and insert there, so in
451 * the slave MDS, we should know whether this object is dir or not, so
452 * use the highest bit of the hash to indicate that (because we do not
453 * use highest bit of hash).
456 ent->lde_hash = le64_to_cpu(ent->lde_hash);
457 ent->lde_hash = cpu_to_le64(ent->lde_hash | MAX_HASH_HIGHEST_BIT);
461 cmm_object_put(env, obj);
466 * Remove all entries from passed page. These entries are going to remote MDT
467 * and thus should be removed locally.
469 static int cmm_split_remove_page(const struct lu_env *env,
470 struct md_object *mo,
471 struct lu_rdpg *rdpg,
472 __u64 hash_end, __u32 *len)
474 struct lu_dirpage *dp;
475 struct lu_dirent *ent;
480 cfs_kmap(rdpg->rp_pages[0]);
481 dp = page_address(rdpg->rp_pages[0]);
482 for (ent = lu_dirent_start(dp);
483 ent != NULL && le64_to_cpu(ent->lde_hash) < hash_end;
484 ent = lu_dirent_next(ent)) {
485 rc = cmm_split_remove_entry(env, mo, ent);
488 * XXX: Error handler to insert remove name back,
489 * currently we assumed it will success anyway in
492 CERROR("Can not del %*.*s, rc %d\n",
493 le16_to_cpu(ent->lde_namelen),
494 le16_to_cpu(ent->lde_namelen),
498 *len += lu_dirent_size(ent);
501 if (ent != lu_dirent_start(dp))
502 *len += sizeof(struct lu_dirpage);
505 cfs_kunmap(rdpg->rp_pages[0]);
509 /* Send one page to remote MDT for creating entries there. */
510 static int cmm_split_send_page(const struct lu_env *env,
511 struct md_object *mo,
512 struct lu_rdpg *rdpg,
513 struct lu_fid *fid, int len)
515 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
516 struct cmm_object *obj;
520 obj = cmm_object_find(env, cmm, fid);
522 RETURN(PTR_ERR(obj));
524 rc = mdc_send_page(cmm, env, md_object_next(&obj->cmo_obj),
525 rdpg->rp_pages[0], len);
526 cmm_object_put(env, obj);
530 /* Read one page of entries from local MDT. */
531 static int cmm_split_read_page(const struct lu_env *env,
532 struct md_object *mo,
533 struct lu_rdpg *rdpg)
537 memset(cfs_kmap(rdpg->rp_pages[0]), 0, CFS_PAGE_SIZE);
538 cfs_kunmap(rdpg->rp_pages[0]);
539 rc = mo_readpage(env, md_object_next(mo), rdpg);
544 * This function performs migration of all pages with entries which fit into one
545 * stripe and one hash segment.
547 static int cmm_split_process_stripe(const struct lu_env *env,
548 struct md_object *mo,
549 struct lu_rdpg *rdpg,
556 LASSERT(rdpg->rp_npages == 1);
558 struct lu_dirpage *ldp;
561 /* Read one page from local MDT. */
562 rc = cmm_split_read_page(env, mo, rdpg);
564 CERROR("Error in readpage: %d\n", rc);
568 /* Remove local entries which are going to remite MDT. */
569 rc = cmm_split_remove_page(env, mo, rdpg, end, &len);
571 CERROR("Error in remove stripe entries: %d\n", rc);
575 /* Send entries page to slave MDT. */
577 rc = cmm_split_send_page(env, mo, rdpg, lf, len);
579 CERROR("Error in sending page: %d\n", rc);
584 cfs_kmap(rdpg->rp_pages[0]);
585 ldp = page_address(rdpg->rp_pages[0]);
586 if (le64_to_cpu(ldp->ldp_hash_end) >= end)
589 rdpg->rp_hash = le64_to_cpu(ldp->ldp_hash_end);
590 cfs_kunmap(rdpg->rp_pages[0]);
596 static int cmm_split_process_dir(const struct lu_env *env,
597 struct md_object *mo,
600 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
601 struct lu_rdpg *rdpg = &cmm_env_info(env)->cmi_rdpg;
606 memset(rdpg, 0, sizeof *rdpg);
607 rdpg->rp_npages = CMM_SPLIT_PAGE_COUNT;
608 rdpg->rp_count = CFS_PAGE_SIZE * rdpg->rp_npages;
609 rdpg->rp_pages = cmm_env_info(env)->cmi_pages;
611 for (i = 0; i < rdpg->rp_npages; i++) {
612 rdpg->rp_pages[i] = cfs_alloc_page(CFS_ALLOC_STD);
613 if (rdpg->rp_pages[i] == NULL)
614 GOTO(cleanup, rc = -ENOMEM);
617 hash_segment = MAX_HASH_SIZE;
618 do_div(hash_segment, cmm->cmm_tgt_count + 1);
619 for (i = 1; i < cmm->cmm_tgt_count + 1; i++) {
623 lf = &ma->ma_lmv->mea_ids[i];
625 rdpg->rp_hash = i * hash_segment;
626 if (i == cmm->cmm_tgt_count)
627 hash_end = MAX_HASH_SIZE;
629 hash_end = rdpg->rp_hash + hash_segment;
630 rc = cmm_split_process_stripe(env, mo, rdpg, lf, hash_end);
632 CERROR("Error (rc = %d) while splitting for %d: fid="
633 DFID", "LPX64":"LPX64"\n", rc, i, PFID(lf),
634 rdpg->rp_hash, hash_end);
640 for (i = 0; i < rdpg->rp_npages; i++)
641 if (rdpg->rp_pages[i] != NULL)
642 __cfs_free_page(rdpg->rp_pages[i]);
646 int cmm_split_dir(const struct lu_env *env, struct md_object *mo)
648 struct cmm_device *cmm = cmm_obj2dev(md2cmm_obj(mo));
649 struct md_attr *ma = &cmm_env_info(env)->cmi_ma;
654 cmm_lprocfs_time_start(env);
656 LASSERT(S_ISDIR(lu_object_attr(&mo->mo_lu)));
657 memset(ma, 0, sizeof(*ma));
659 /* Step1: Checking whether the dir needs to be split. */
660 rc = cmm_split_expect(env, mo, ma, &split);
664 if (split != CMM_SPLIT_NEEDED) {
665 /* No split is needed, caller may proceed with create. */
669 /* Split should be done now, let's do it. */
670 CWARN("Dir "DFID" is going to split (size: "LPU64")\n",
671 PFID(lu_object_fid(&mo->mo_lu)), ma->ma_attr.la_size);
674 * Disable transacrions for split, since there will be so many trans in
675 * this one ops, conflict with current recovery design.
677 rc = cmm_upcall(env, &cmm->cmm_md_dev, MD_NO_TRANS);
679 CERROR("Can't disable trans for split, rc %d\n", rc);
683 /* Step2: Prepare the md memory */
684 ma->ma_lmv_size = CMM_MD_SIZE(cmm->cmm_tgt_count + 1);
685 OBD_ALLOC(ma->ma_lmv, ma->ma_lmv_size);
686 if (ma->ma_lmv == NULL)
687 GOTO(out, rc = -ENOMEM);
689 /* Step3: Create slave objects and fill the ma->ma_lmv */
690 rc = cmm_split_slaves_create(env, mo, ma);
692 CERROR("Can't create slaves for split, rc %d\n", rc);
696 /* Step4: Scan and split the object. */
697 rc = cmm_split_process_dir(env, mo, ma);
699 CERROR("Can't scan and split, rc %d\n", rc);
703 /* Step5: Set mea to the master object. */
704 buf = cmm_buf_get(env, ma->ma_lmv, ma->ma_lmv_size);
705 rc = mo_xattr_set(env, md_object_next(mo), buf,
708 CERROR("Can't set MEA to master dir, " "rc %d\n", rc);
712 /* set flag in cmm_object */
713 md2cml_obj(mo)->clo_split = CMM_SPLIT_DONE;
716 * Finally, split succeed, tell client to repeat opetartion on correct
719 CWARN("Dir "DFID" has been split\n", PFID(lu_object_fid(&mo->mo_lu)));
723 OBD_FREE(ma->ma_lmv, ma->ma_lmv_size);
725 cmm_lprocfs_time_end(env, cmm, LPROC_CMM_SPLIT);