lustre/mds/mds_unlink_open.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  * GPL HEADER START
   5  *
   6  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   7  *
   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.
  11  *
  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).
  17  *
  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
  21  *
  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
  24  * have any questions.
  25  *
  26  * GPL HEADER END
  27  */
  28 /*
  29  * Copyright  2008 Sun Microsystems, Inc. All rights reserved
  30  * Use is subject to license terms.
  31  */
  32 /*
  33  * This file is part of Lustre, http://www.lustre.org/
  34  * Lustre is a trademark of Sun Microsystems, Inc.
  35  *
  36  * lustre/mds/mds_unlink_open.c
  37  *
  38  * Author: Peter Braam <braam@clusterfs.com>
  39  * Author: Andreas Dilger <adilger@clusterfs.com>
  40  * Author: Phil Schwan <phil@clusterfs.com>
  41  */
  42
  43 /* code for handling open unlinked files */
  44
  45 #define DEBUG_SUBSYSTEM S_MDS
  46
  47 #ifndef AUTOCONF_INCLUDED
  48 #include <linux/config.h>
  49 #endif
  50 #include <linux/module.h>
  51 #include <linux/version.h>
  52
  53 #include <libcfs/list.h>
  54 #include <obd_class.h>
  55 #include <lustre_fsfilt.h>
  56 #include <lustre_mds.h>
  57 #include <lvfs.h>
  58
  59 #include "mds_internal.h"
  60
  61 int mds_osc_destroy_orphan(struct obd_device *obd,
  62                            umode_t mode,
  63                            struct lov_mds_md *lmm,
  64                            int lmm_size,
  65                            struct llog_cookie *logcookies,
  66                            int log_unlink)
  67 {
  68         struct mds_obd *mds = &obd->u.mds;
  69         struct lov_stripe_md *lsm = NULL;
  70         struct obd_trans_info oti = { 0 };
  71         struct obdo *oa;
  72         int rc;
  73         ENTRY;
  74
  75         if (lmm_size == 0)
  76                 RETURN(0);
  77
  78         rc = obd_unpackmd(mds->mds_lov_exp, &lsm, lmm, lmm_size);
  79         if (rc < 0) {
  80                 CERROR("Error unpack md %p\n", lmm);
  81                 RETURN(rc);
  82         } else {
  83                 LASSERT(rc >= sizeof(*lsm));
  84                 rc = 0;
  85         }
  86
  87         rc = obd_checkmd(mds->mds_lov_exp, obd->obd_self_export, lsm);
  88         if (rc)
  89                 GOTO(out_free_memmd, rc);
  90
  91         OBDO_ALLOC(oa);
  92         if (oa == NULL)
  93                 GOTO(out_free_memmd, rc = -ENOMEM);
  94         oa->o_id = lsm->lsm_object_id;
  95         oa->o_gr = 0;
  96         oa->o_mode = mode & S_IFMT;
  97         oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE | OBD_MD_FLGROUP;
  98
  99         if (log_unlink && logcookies) {
 100                 oa->o_valid |= OBD_MD_FLCOOKIE;
 101                 oti.oti_logcookies = logcookies;
 102         }
 103         rc = obd_destroy(mds->mds_lov_exp, oa, lsm, &oti, obd->obd_self_export);
 104         OBDO_FREE(oa);
 105         if (rc)
 106                 CDEBUG(D_INODE, "destroy orphan objid 0x"LPX64" on ost error "
 107                        "%d\n", lsm->lsm_object_id, rc);
 108 out_free_memmd:
 109         obd_free_memmd(mds->mds_lov_exp, &lsm);
 110         RETURN(rc);
 111 }
 112
 113 static int mds_unlink_orphan(struct obd_device *obd, struct dentry *dchild,
 114                              struct inode *inode, struct inode *pending_dir)
 115 {
 116         struct mds_obd *mds = &obd->u.mds;
 117         struct lov_mds_md *lmm = NULL;
 118         struct llog_cookie *logcookies = NULL;
 119         int lmm_size, log_unlink = 0, cookie_size = 0;
 120         void *handle = NULL;
 121         umode_t mode;
 122         int rc, err;
 123         ENTRY;
 124
 125         LASSERT(mds->mds_lov_obd != NULL);
 126
 127         /* We don't need to do any of these other things for orhpan dirs,
 128          * especially not mds_get_md (may get a default LOV EA, bug 4554) */
 129         mode = inode->i_mode;
 130         if (S_ISDIR(mode)) {
 131                 rc = ll_vfs_rmdir(pending_dir, dchild, mds->mds_vfsmnt);
 132                 if (rc)
 133                         CERROR("error %d unlinking dir %*s from PENDING\n",
 134                                rc, dchild->d_name.len, dchild->d_name.name);
 135                 RETURN(rc);
 136         }
 137
 138         lmm_size = mds->mds_max_mdsize;
 139         OBD_ALLOC(lmm, lmm_size);
 140         if (lmm == NULL)
 141                 RETURN(-ENOMEM);
 142
 143         rc = mds_get_md(obd, inode, lmm, &lmm_size, 1, 0, 0);
 144         if (rc < 0)
 145                 GOTO(out_free_lmm, rc);
 146
 147         handle = fsfilt_start_log(obd, pending_dir, FSFILT_OP_UNLINK, NULL,
 148                                   le32_to_cpu(lmm->lmm_stripe_count));
 149         if (IS_ERR(handle)) {
 150                 rc = PTR_ERR(handle);
 151                 CERROR("error fsfilt_start: %d\n", rc);
 152                 handle = NULL;
 153                 GOTO(out_free_lmm, rc);
 154         }
 155
 156         rc = ll_vfs_unlink(pending_dir, dchild, mds->mds_vfsmnt);
 157         if (rc) {
 158                 CERROR("error %d unlinking orphan %.*s from PENDING\n",
 159                        rc, dchild->d_name.len, dchild->d_name.name);
 160         } else if (lmm_size) {
 161                 cookie_size = mds_get_cookie_size(obd, lmm);
 162                 OBD_ALLOC(logcookies, cookie_size);
 163                 if (logcookies == NULL)
 164                         rc = -ENOMEM;
 165                 else if (mds_log_op_unlink(obd, lmm,lmm_size,logcookies,
 166                                            cookie_size) > 0)
 167                         log_unlink = 1;
 168         }
 169
 170         err = fsfilt_commit(obd, pending_dir, handle, 0);
 171         if (err) {
 172                 CERROR("error committing orphan unlink: %d\n", err);
 173                 if (!rc)
 174                         rc = err;
 175         } else if (!rc) {
 176                 rc = mds_osc_destroy_orphan(obd, mode, lmm, lmm_size,
 177                                             logcookies, log_unlink);
 178         }
 179
 180         if (logcookies != NULL)
 181                 OBD_FREE(logcookies, cookie_size);
 182 out_free_lmm:
 183         OBD_FREE(lmm, mds->mds_max_mdsize);
 184         RETURN(rc);
 185 }
 186
 187 static __u64 mds_orphans_max_version(struct obd_device *obd)
 188 {
 189         struct obd_export *exp;
 190         __u32 epoch = lr_epoch(obd->u.mds.mds_last_transno);
 191         spin_lock(&obd->obd_dev_lock);
 192         list_for_each_entry(exp, &obd->obd_delayed_exports, exp_obd_chain) {
 193                 struct lu_export_data *led = &exp->exp_target_data;
 194                 epoch = min(epoch, le32_to_cpu(led->led_lcd->lcd_first_epoch));
 195         }
 196         spin_unlock(&obd->obd_dev_lock);
 197         return (__u64)epoch << LR_EPOCH_BITS;
 198 }
 199
 200 /* Delete inodes which were previously open-unlinked but were not reopened
 201  * during MDS recovery for whatever reason (e.g. client also failed, recovery
 202  * aborted, etc). */
 203 int mds_cleanup_pending(struct obd_device *obd)
 204 {
 205         struct mds_obd *mds = &obd->u.mds;
 206         struct lvfs_run_ctxt saved;
 207         struct file *file;
 208         struct dentry *dchild, *dentry;
 209         struct vfsmount *mnt;
 210         struct inode *child_inode, *pending_dir = mds->mds_pending_dir->d_inode;
 211         struct l_linux_dirent *dirent, *n;
 212         struct list_head dentry_list;
 213         char d_name[LL_FID_NAMELEN];
 214         unsigned long inum;
 215         __u64 max_version;
 216         int i = 0, rc = 0, item = 0, namlen;
 217         ENTRY;
 218
 219         push_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 220         /* dentry and mnt ref dropped in dentry_open() on error, or
 221          * in filp_close() if dentry_open() succeeds */
 222         dentry = dget(mds->mds_pending_dir);
 223         if (IS_ERR(dentry))
 224                 GOTO(err_pop, rc = PTR_ERR(dentry));
 225         mnt = mntget(mds->mds_vfsmnt);
 226         if (IS_ERR(mnt))
 227                 GOTO(err_mntget, rc = PTR_ERR(mnt));
 228
 229         file = ll_dentry_open(mds->mds_pending_dir, mds->mds_vfsmnt,
 230                            O_RDONLY | O_LARGEFILE, current_cred());
 231         if (IS_ERR(file))
 232                 GOTO(err_pop, rc = PTR_ERR(file));
 233
 234         CFS_INIT_LIST_HEAD(&dentry_list);
 235         rc = l_readdir(file, &dentry_list);
 236         filp_close(file, 0);
 237         if (rc < 0)
 238                 GOTO(err_out, rc);
 239
 240         /** Get maximum version for orphans to delete. All other orphans may be
 241          *  needed for delayed clients */
 242         max_version = mds_orphans_max_version(obd);
 243
 244         list_for_each_entry_safe(dirent, n, &dentry_list, lld_list) {
 245                 __u64 version;
 246
 247                 i++;
 248                 list_del(&dirent->lld_list);
 249
 250                 namlen = strlen(dirent->lld_name);
 251                 LASSERT(sizeof(d_name) >= namlen + 1);
 252                 strcpy(d_name, dirent->lld_name);
 253                 inum = dirent->lld_ino;
 254                 OBD_FREE_PTR(dirent);
 255
 256                 CDEBUG(D_INODE, "entry %d of PENDING DIR: %s\n", i, d_name);
 257
 258                 if (((namlen == 1) && !strcmp(d_name, ".")) ||
 259                     ((namlen == 2) && !strcmp(d_name, "..")) || inum == 0)
 260                         continue;
 261
 262                 LOCK_INODE_MUTEX(pending_dir);
 263                 dchild = lookup_one_len(d_name, mds->mds_pending_dir, namlen);
 264                 if (IS_ERR(dchild)) {
 265                         UNLOCK_INODE_MUTEX(pending_dir);
 266                         GOTO(err_out, rc = PTR_ERR(dchild));
 267                 }
 268                 if (!dchild->d_inode) {
 269                         CWARN("%s: orphan %s has already been removed\n",
 270                               obd->obd_name, d_name);
 271                         GOTO(next, rc = 0);
 272                 }
 273
 274                 if (is_bad_inode(dchild->d_inode)) {
 275                         CERROR("%s: bad orphan inode found %lu/%u\n",
 276                                obd->obd_name, dchild->d_inode->i_ino,
 277                                dchild->d_inode->i_generation);
 278                         GOTO(next, rc = -ENOENT);
 279                 }
 280
 281                 child_inode = dchild->d_inode;
 282                 MDS_DOWN_READ_ORPHAN_SEM(child_inode);
 283                 if (mds_inode_is_orphan(child_inode) &&
 284                     mds_orphan_open_count(child_inode)) {
 285                         MDS_UP_READ_ORPHAN_SEM(child_inode);
 286                         CWARN("%s: orphan %s re-opened during recovery\n",
 287                               obd->obd_name, d_name);
 288                         GOTO(next, rc = 0);
 289                 }
 290                 /** Keep orphans for possible use by delayed exports. Remove
 291                  * orphans with version lower than minimal one of all exports */
 292                 version = fsfilt_get_version(obd, child_inode);
 293                 if ((__s64)version != -EOPNOTSUPP &&
 294                     version >= max_version) {
 295                         MDS_UP_READ_ORPHAN_SEM(child_inode);
 296                         CDEBUG(D_INFO,
 297                                "%s: orphan %s is needed for delayed exports\n",
 298                                obd->obd_name, d_name);
 299                         GOTO(next, rc = 0);
 300                 }
 301                 MDS_UP_READ_ORPHAN_SEM(child_inode);
 302
 303                 rc = mds_unlink_orphan(obd, dchild, child_inode, pending_dir);
 304                 CDEBUG(D_INODE, "%s: removed orphan %s: rc %d\n",
 305                        obd->obd_name, d_name, rc);
 306                 if (rc == 0)
 307                         item++;
 308                 else
 309                         rc = 0;
 310 next:
 311                 l_dput(dchild);
 312                 UNLOCK_INODE_MUTEX(pending_dir);
 313         }
 314         rc = 0;
 315 err_out:
 316         list_for_each_entry_safe(dirent, n, &dentry_list, lld_list) {
 317                 list_del(&dirent->lld_list);
 318                 OBD_FREE(dirent, sizeof(*dirent));
 319         }
 320 err_pop:
 321         pop_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 322         if (item > 0)
 323                 CWARN("%s: removed %d pending open-unlinked files\n",
 324                       obd->obd_name, item);
 325         RETURN(rc);
 326
 327 err_mntget:
 328         l_dput(mds->mds_pending_dir);
 329         goto err_pop;
 330 }
 331
 332 /**
 333  * Determine there is no orphan with the same inode number. That may happens
 334  * since unlink replay don't delete inode but keep orphan for delayed clients.
 335  * Therefore replays like 'create, unlink, create' will fail due to inode can't
 336  * be reused.
 337  */
 338 int mds_check_stale_orphan(struct obd_device *obd, struct ll_fid *fid)
 339 {
 340         struct mds_obd *mds = &obd->u.mds;
 341         char fidname[32];
 342         struct dentry *result;
 343         struct inode *inode, *pending_dir = mds->mds_pending_dir->d_inode;
 344         int fidlen = 0, rc = 0;
 345
 346         /* no need in checks*/
 347         if (fid->id == 0 || obd->obd_recovering == 0)
 348                 RETURN(0);
 349
 350         /** open by fid like mds_fid2dentry does */
 351         snprintf(fidname, sizeof(fidname), "0x%lx", (unsigned long)(fid->id));
 352         fidlen = strlen(fidname);
 353         result = mds_lookup(obd, fidname, mds->mds_fid_de, fidlen);
 354         if (IS_ERR(result))
 355                 RETURN(0);
 356         inode = result->d_inode;
 357         if (!inode)
 358                 GOTO(out, rc = 0);
 359
 360         LOCK_INODE_MUTEX(pending_dir);
 361         MDS_DOWN_READ_ORPHAN_SEM(inode);
 362         if (mds_inode_is_orphan(inode)) {
 363                 struct dentry *orphan;
 364
 365                 /* bz18927: The exactly same inode can be marked as orphan
 366                  * if there was open|creat replay and this is second one */
 367                 if (inode->i_generation == fid->generation)
 368                         GOTO(unlock_child, rc);
 369
 370                 if (mds_orphan_open_count(inode) > 0) {
 371                         CERROR("Orphan "LPU64"/%u is in use!\n",
 372                                fid->id, fid->generation);
 373                         GOTO(unlock_child, rc = -EFAULT);
 374                 }
 375
 376                 /** Found orphan in pending dir and delete it */
 377                 fidlen = ll_fid2str(fidname, fid->id, inode->i_generation);
 378                 orphan = lookup_one_len(fidname, mds->mds_pending_dir, fidlen);
 379                 if (IS_ERR(orphan)) {
 380                         rc = PTR_ERR(orphan);
 381                         CERROR("error looking up %s in PENDING: rc = %d\n",
 382                                 fidname, rc);
 383                         GOTO(unlock_child, rc);
 384                 }
 385                 if (orphan->d_inode != inode) {
 386                         l_dput(orphan);
 387                         CWARN("%s: Found wrong orphan %s %p/%p\n",
 388                               obd->obd_name, fidname, orphan->d_inode, inode);
 389                         GOTO(unlock_child, rc = -EFAULT);
 390                 }
 391                 MDS_UP_READ_ORPHAN_SEM(inode);
 392
 393                 rc = mds_unlink_orphan(obd, orphan, inode, pending_dir);
 394                 CDEBUG(D_INODE, "%s: removed orphan %s: rc %d\n",
 395                        obd->obd_name, fidname, rc);
 396                 l_dput(orphan);
 397                 GOTO(unlock, rc);
 398         }
 399 unlock_child:
 400         MDS_UP_READ_ORPHAN_SEM(inode);
 401 unlock:
 402         UNLOCK_INODE_MUTEX(pending_dir);
 403 out:
 404         l_dput(result);
 405         RETURN(0);
 406 }