lustre/obdclass/llog_lvfs.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  *  Copyright (C) 2001-2003 Cluster File Systems, Inc.
   5  *   Author: Andreas Dilger <adilger@clusterfs.com>
   6  *
   7  *   This file is part of the Lustre file system, http://www.lustre.org
   8  *   Lustre is a trademark of Cluster File Systems, Inc.
   9  *
  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.
  14  *
  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.
  19  *
  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.
  24  *
  25  * OST<->MDS recovery logging infrastructure.
  26  *
  27  * Invariants in implementation:
  28  * - we do not share logs among different OST<->MDS connections, so that
  29  *   if an OST or MDS fails it need only look at log(s) relevant to itself
  30  */
  31
  32 #define DEBUG_SUBSYSTEM S_LOG
  33
  34 #ifndef EXPORT_SYMTAB
  35 #define EXPORT_SYMTAB
  36 #endif
  37
  38 #ifndef __KERNEL__
  39 #include <liblustre.h>
  40 #endif
  41
  42 #include <obd.h>
  43 #include <obd_class.h>
  44 #include <lustre_log.h>
  45 #include <obd_ost.h>
  46 #include <libcfs/list.h>
  47 #include <lvfs.h>
  48 #include <lustre_fsfilt.h>
  49 #include <lustre_disk.h>
  50 #include "llog_internal.h"
  51
  52 #if defined(__KERNEL__) && defined(LLOG_LVFS)
  53
  54 static int llog_lvfs_pad(struct obd_device *obd, struct l_file *file,
  55                                 int len, int index)
  56 {
  57         struct llog_rec_hdr rec = { 0 };
  58         struct llog_rec_tail tail;
  59         int rc;
  60         ENTRY;
  61
  62         LASSERT(len >= LLOG_MIN_REC_SIZE && (len & 0x7) == 0);
  63
  64         tail.lrt_len = rec.lrh_len = len;
  65         tail.lrt_index = rec.lrh_index = index;
  66         rec.lrh_type = LLOG_PAD_MAGIC;
  67
  68         rc = fsfilt_write_record(obd, file, &rec, sizeof(rec), &file->f_pos, 0);
  69         if (rc) {
  70                 CERROR("error writing padding record: rc %d\n", rc);
  71                 goto out;
  72         }
  73
  74         file->f_pos += len - sizeof(rec) - sizeof(tail);
  75         rc = fsfilt_write_record(obd, file, &tail, sizeof(tail),&file->f_pos,0);
  76         if (rc) {
  77                 CERROR("error writing padding record: rc %d\n", rc);
  78                 goto out;
  79         }
  80
  81  out:
  82         RETURN(rc);
  83 }
  84
  85 static int llog_lvfs_write_blob(struct obd_device *obd, struct l_file *file,
  86                                 struct llog_rec_hdr *rec, void *buf, loff_t off)
  87 {
  88         int rc;
  89         struct llog_rec_tail end;
  90         loff_t saved_off = file->f_pos;
  91         int buflen = rec->lrh_len;
  92
  93         ENTRY;
  94         file->f_pos = off;
  95
  96         if (!buf) {
  97                 rc = fsfilt_write_record(obd, file, rec, buflen,&file->f_pos,0);
  98                 if (rc) {
  99                         CERROR("error writing log record: rc %d\n", rc);
 100                         goto out;
 101                 }
 102                 GOTO(out, rc = 0);
 103         }
 104
 105         /* the buf case */
 106         rec->lrh_len = sizeof(*rec) + buflen + sizeof(end);
 107         rc = fsfilt_write_record(obd, file, rec, sizeof(*rec), &file->f_pos, 0);
 108         if (rc) {
 109                 CERROR("error writing log hdr: rc %d\n", rc);
 110                 goto out;
 111         }
 112
 113         rc = fsfilt_write_record(obd, file, buf, buflen, &file->f_pos, 0);
 114         if (rc) {
 115                 CERROR("error writing log buffer: rc %d\n", rc);
 116                 goto out;
 117         }
 118
 119         end.lrt_len = rec->lrh_len;
 120         end.lrt_index = rec->lrh_index;
 121         rc = fsfilt_write_record(obd, file, &end, sizeof(end), &file->f_pos, 0);
 122         if (rc) {
 123                 CERROR("error writing log tail: rc %d\n", rc);
 124                 goto out;
 125         }
 126
 127         rc = 0;
 128  out:
 129         if (saved_off > file->f_pos)
 130                 file->f_pos = saved_off;
 131         LASSERT(rc <= 0);
 132         RETURN(rc);
 133 }
 134
 135 static int llog_lvfs_read_blob(struct obd_device *obd, struct l_file *file,
 136                                 void *buf, int size, loff_t off)
 137 {
 138         loff_t offset = off;
 139         int rc;
 140         ENTRY;
 141
 142         rc = fsfilt_read_record(obd, file, buf, size, &offset);
 143         if (rc) {
 144                 CERROR("error reading log record: rc %d\n", rc);
 145                 RETURN(rc);
 146         }
 147         RETURN(0);
 148 }
 149
 150 static int llog_lvfs_read_header(struct llog_handle *handle)
 151 {
 152         struct obd_device *obd;
 153         int rc;
 154         ENTRY;
 155
 156         LASSERT(sizeof(*handle->lgh_hdr) == LLOG_CHUNK_SIZE);
 157
 158         obd = handle->lgh_ctxt->loc_exp->exp_obd;
 159
 160         if (handle->lgh_file->f_dentry->d_inode->i_size == 0) {
 161                 CDEBUG(D_HA, "not reading header from 0-byte log\n");
 162                 RETURN(LLOG_EEMPTY);
 163         }
 164
 165         rc = llog_lvfs_read_blob(obd, handle->lgh_file, handle->lgh_hdr,
 166                                  LLOG_CHUNK_SIZE, 0);
 167         if (rc) {
 168                 CERROR("error reading log header from %.*s\n",
 169                        handle->lgh_file->f_dentry->d_name.len,
 170                        handle->lgh_file->f_dentry->d_name.name);
 171         } else {
 172                 struct llog_rec_hdr *llh_hdr = &handle->lgh_hdr->llh_hdr;
 173
 174                 if (LLOG_REC_HDR_NEEDS_SWABBING(llh_hdr))
 175                         lustre_swab_llog_hdr(handle->lgh_hdr);
 176
 177                 if (llh_hdr->lrh_type != LLOG_HDR_MAGIC) {
 178                         CERROR("bad log %.*s header magic: %#x (expected %#x)\n",
 179                                handle->lgh_file->f_dentry->d_name.len,
 180                                handle->lgh_file->f_dentry->d_name.name,
 181                                llh_hdr->lrh_type, LLOG_HDR_MAGIC);
 182                         rc = -EIO;
 183                 } else if (llh_hdr->lrh_len != LLOG_CHUNK_SIZE) {
 184                         CERROR("incorrectly sized log %.*s header: %#x "
 185                                "(expected %#x)\n",
 186                                handle->lgh_file->f_dentry->d_name.len,
 187                                handle->lgh_file->f_dentry->d_name.name,
 188                                llh_hdr->lrh_len, LLOG_CHUNK_SIZE);
 189                         CERROR("you may need to re-run lconf --write_conf.\n");
 190                         rc = -EIO;
 191                 }
 192         }
 193
 194         handle->lgh_last_idx = handle->lgh_hdr->llh_tail.lrt_index;
 195         handle->lgh_file->f_pos = handle->lgh_file->f_dentry->d_inode->i_size;
 196
 197         RETURN(rc);
 198 }
 199
 200 /* returns negative in on error; 0 if success && reccookie == 0; 1 otherwise */
 201 /* appends if idx == -1, otherwise overwrites record idx. */
 202 static int llog_lvfs_write_rec(struct llog_handle *loghandle,
 203                                struct llog_rec_hdr *rec,
 204                                struct llog_cookie *reccookie, int cookiecount,
 205                                void *buf, int idx)
 206 {
 207         struct llog_log_hdr *llh;
 208         int reclen = rec->lrh_len, index, rc;
 209         struct llog_rec_tail *lrt;
 210         struct obd_device *obd;
 211         struct file *file;
 212         size_t left;
 213         ENTRY;
 214
 215         llh = loghandle->lgh_hdr;
 216         file = loghandle->lgh_file;
 217         obd = loghandle->lgh_ctxt->loc_exp->exp_obd;
 218
 219         /* record length should not bigger than LLOG_CHUNK_SIZE */
 220         if (buf)
 221                 rc = (reclen > LLOG_CHUNK_SIZE - sizeof(struct llog_rec_hdr) -
 222                       sizeof(struct llog_rec_tail)) ? -E2BIG : 0;
 223         else
 224                 rc = (reclen > LLOG_CHUNK_SIZE) ? -E2BIG : 0;
 225         if (rc)
 226                 RETURN(rc);
 227
 228         if (buf)
 229                 /* write_blob adds header and tail to lrh_len. */
 230                 reclen = sizeof(*rec) + rec->lrh_len +
 231                         sizeof(struct llog_rec_tail);
 232
 233         if (idx != -1) {
 234                 loff_t saved_offset;
 235
 236                 /* no header: only allowed to insert record 1 */
 237                 if (idx != 1 && !file->f_dentry->d_inode->i_size) {
 238                         CERROR("idx != -1 in empty log\n");
 239                         LBUG();
 240                 }
 241
 242                 if (idx && llh->llh_size && llh->llh_size != rec->lrh_len)
 243                         RETURN(-EINVAL);
 244
 245                 if (!ext2_test_bit(idx, llh->llh_bitmap))
 246                         CERROR("Modify unset record %u\n", idx);
 247                 if (idx != rec->lrh_index)
 248                         CERROR("Index mismatch %d %u\n", idx, rec->lrh_index);
 249
 250                 rc = llog_lvfs_write_blob(obd, file, &llh->llh_hdr, NULL, 0);
 251                 /* we are done if we only write the header or on error */
 252                 if (rc || idx == 0)
 253                         RETURN(rc);
 254
 255                 /* Assumes constant lrh_len */
 256                 saved_offset = sizeof(*llh) + (idx - 1) * reclen;
 257
 258                 if (buf) {
 259                         struct llog_rec_hdr check;
 260
 261                         /* We assume that caller has set lgh_cur_* */
 262                         saved_offset = loghandle->lgh_cur_offset;
 263                         CDEBUG(D_OTHER,
 264                                "modify record "LPX64": idx:%d/%u/%d, len:%u "
 265                                "offset %llu\n",
 266                                loghandle->lgh_id.lgl_oid, idx, rec->lrh_index,
 267                                loghandle->lgh_cur_idx,
 268                                rec->lrh_len, saved_offset - sizeof(*llh));
 269                         if (rec->lrh_index != loghandle->lgh_cur_idx) {
 270                                 CERROR("modify idx mismatch %u/%d\n",
 271                                        idx, loghandle->lgh_cur_idx);
 272                                 RETURN(-EFAULT);
 273                         }
 274 #if 1  /* FIXME remove this safety check at some point */
 275                         /* Verify that the record we're modifying is the
 276                            right one. */
 277                         rc = llog_lvfs_read_blob(obd, file, &check,
 278                                                  sizeof(check), saved_offset);
 279                         if (check.lrh_index != idx || check.lrh_len != reclen) {
 280                                 CERROR("Bad modify idx %u/%u size %u/%u (%d)\n",
 281                                        idx, check.lrh_index, reclen,
 282                                        check.lrh_len, rc);
 283                                 RETURN(-EFAULT);
 284                         }
 285 #endif
 286                 }
 287
 288                 rc = llog_lvfs_write_blob(obd, file, rec, buf, saved_offset);
 289                 if (rc == 0 && reccookie) {
 290                         reccookie->lgc_lgl = loghandle->lgh_id;
 291                         reccookie->lgc_index = idx;
 292                         rc = 1;
 293                 }
 294                 RETURN(rc);
 295         }
 296
 297         /* Make sure that records don't cross a chunk boundary, so we can
 298          * process them page-at-a-time if needed.  If it will cross a chunk
 299          * boundary, write in a fake (but referenced) entry to pad the chunk.
 300          *
 301          * We know that llog_current_log() will return a loghandle that is
 302          * big enough to hold reclen, so all we care about is padding here.
 303          */
 304         left = LLOG_CHUNK_SIZE - (file->f_pos & (LLOG_CHUNK_SIZE - 1));
 305
 306         /* NOTE: padding is a record, but no bit is set */
 307         if (left != 0 && left != reclen &&
 308             left < (reclen + LLOG_MIN_REC_SIZE)) {
 309                 loghandle->lgh_last_idx++;
 310                 rc = llog_lvfs_pad(obd, file, left, loghandle->lgh_last_idx);
 311                 if (rc)
 312                         RETURN(rc);
 313                 /* if it's the last idx in log file, then return -ENOSPC */
 314                 if (loghandle->lgh_last_idx == LLOG_BITMAP_SIZE(llh) - 1)
 315                         RETURN(-ENOSPC);
 316         }
 317
 318         loghandle->lgh_last_idx++;
 319         index = loghandle->lgh_last_idx;
 320         LASSERT(index < LLOG_BITMAP_SIZE(llh));
 321         rec->lrh_index = index;
 322         if (buf == NULL) {
 323                 lrt = (struct llog_rec_tail *)
 324                         ((char *)rec + rec->lrh_len - sizeof(*lrt));
 325                 lrt->lrt_len = rec->lrh_len;
 326                 lrt->lrt_index = rec->lrh_index;
 327         }
 328         if (ext2_set_bit(index, llh->llh_bitmap)) {
 329                 CERROR("argh, index %u already set in log bitmap?\n", index);
 330                 LBUG(); /* should never happen */
 331         }
 332         llh->llh_count++;
 333         llh->llh_tail.lrt_index = index;
 334
 335         rc = llog_lvfs_write_blob(obd, file, &llh->llh_hdr, NULL, 0);
 336         if (rc)
 337                 RETURN(rc);
 338
 339         rc = llog_lvfs_write_blob(obd, file, rec, buf, file->f_pos);
 340         if (rc)
 341                 RETURN(rc);
 342
 343         CDEBUG(D_HA, "added record "LPX64": idx: %u, %u bytes\n",
 344                loghandle->lgh_id.lgl_oid, index, rec->lrh_len);
 345         if (rc == 0 && reccookie) {
 346                 reccookie->lgc_lgl = loghandle->lgh_id;
 347                 reccookie->lgc_index = index;
 348                 if ((rec->lrh_type == MDS_UNLINK_REC) ||
 349                                 (rec->lrh_type == MDS_SETATTR_REC))
 350                         reccookie->lgc_subsys = LLOG_MDS_OST_ORIG_CTXT;
 351                 else if (rec->lrh_type == OST_SZ_REC)
 352                         reccookie->lgc_subsys = LLOG_SIZE_ORIG_CTXT;
 353                 else if (rec->lrh_type == OST_RAID1_REC)
 354                         reccookie->lgc_subsys = LLOG_RD1_ORIG_CTXT;
 355                 else
 356                         reccookie->lgc_subsys = -1;
 357                 rc = 1;
 358         }
 359         if (rc == 0 && rec->lrh_type == LLOG_GEN_REC)
 360                 rc = 1;
 361
 362         RETURN(rc);
 363 }
 364
 365 /* We can skip reading at least as many log blocks as the number of
 366 * minimum sized log records we are skipping.  If it turns out
 367 * that we are not far enough along the log (because the
 368 * actual records are larger than minimum size) we just skip
 369 * some more records. */
 370
 371 static void llog_skip_over(__u64 *off, int curr, int goal)
 372 {
 373         if (goal <= curr)
 374                 return;
 375         *off = (*off + (goal-curr-1) * LLOG_MIN_REC_SIZE) &
 376                 ~(LLOG_CHUNK_SIZE - 1);
 377 }
 378
 379
 380 /* sets:
 381  *  - cur_offset to the furthest point read in the log file
 382  *  - cur_idx to the log index preceeding cur_offset
 383  * returns -EIO/-EINVAL on error
 384  */
 385 static int llog_lvfs_next_block(struct llog_handle *loghandle, int *cur_idx,
 386                                 int next_idx, __u64 *cur_offset, void *buf,
 387                                 int len)
 388 {
 389         int rc;
 390         ENTRY;
 391
 392         if (len == 0 || len & (LLOG_CHUNK_SIZE - 1))
 393                 RETURN(-EINVAL);
 394
 395         CDEBUG(D_OTHER, "looking for log index %u (cur idx %u off "LPU64")\n",
 396                next_idx, *cur_idx, *cur_offset);
 397
 398         while (*cur_offset < loghandle->lgh_file->f_dentry->d_inode->i_size) {
 399                 struct llog_rec_hdr *rec;
 400                 struct llog_rec_tail *tail;
 401                 loff_t ppos;
 402
 403                 llog_skip_over(cur_offset, *cur_idx, next_idx);
 404
 405                 ppos = *cur_offset;
 406                 rc = fsfilt_read_record(loghandle->lgh_ctxt->loc_exp->exp_obd,
 407                                         loghandle->lgh_file, buf, len,
 408                                         &ppos);
 409                 if (rc) {
 410                         CERROR("Cant read llog block at log id "LPU64
 411                                "/%u offset "LPU64"\n",
 412                                loghandle->lgh_id.lgl_oid,
 413                                loghandle->lgh_id.lgl_ogen,
 414                                *cur_offset);
 415                         RETURN(rc);
 416                 }
 417
 418                 /* put number of bytes read into rc to make code simpler */
 419                 rc = ppos - *cur_offset;
 420                 *cur_offset = ppos;
 421
 422                 if (rc < len) {
 423                         /* signal the end of the valid buffer to llog_process */
 424                         memset(buf + rc, 0, len - rc);
 425                 }
 426
 427                 if (rc == 0) /* end of file, nothing to do */
 428                         RETURN(0);
 429
 430                 if (rc < sizeof(*tail)) {
 431                         CERROR("Invalid llog block at log id "LPU64"/%u offset "
 432                                LPU64"\n", loghandle->lgh_id.lgl_oid,
 433                                loghandle->lgh_id.lgl_ogen, *cur_offset);
 434                         RETURN(-EINVAL);
 435                 }
 436
 437                 rec = buf;
 438                 tail = (struct llog_rec_tail *)((char *)buf + rc -
 439                                                 sizeof(struct llog_rec_tail));
 440
 441                 if (LLOG_REC_HDR_NEEDS_SWABBING(rec)) {
 442                         lustre_swab_llog_rec(rec, tail);
 443                 }
 444
 445                 *cur_idx = tail->lrt_index;
 446
 447                 /* this shouldn't happen */
 448                 if (tail->lrt_index == 0) {
 449                         CERROR("Invalid llog tail at log id "LPU64"/%u offset "
 450                                LPU64"\n", loghandle->lgh_id.lgl_oid,
 451                                loghandle->lgh_id.lgl_ogen, *cur_offset);
 452                         RETURN(-EINVAL);
 453                 }
 454                 if (tail->lrt_index < next_idx)
 455                         continue;
 456
 457                 /* sanity check that the start of the new buffer is no farther
 458                  * than the record that we wanted.  This shouldn't happen. */
 459                 if (rec->lrh_index > next_idx) {
 460                         CERROR("missed desired record? %u > %u\n",
 461                                rec->lrh_index, next_idx);
 462                         RETURN(-ENOENT);
 463                 }
 464                 RETURN(0);
 465         }
 466         RETURN(-EIO);
 467 }
 468
 469 static int llog_lvfs_prev_block(struct llog_handle *loghandle,
 470                                 int prev_idx, void *buf, int len)
 471 {
 472         __u64 cur_offset;
 473         int rc;
 474         ENTRY;
 475
 476         if (len == 0 || len & (LLOG_CHUNK_SIZE - 1))
 477                 RETURN(-EINVAL);
 478
 479         CDEBUG(D_OTHER, "looking for log index %u n", prev_idx);
 480
 481         cur_offset = LLOG_CHUNK_SIZE;
 482         llog_skip_over(&cur_offset, 0, prev_idx);
 483
 484         while (cur_offset < loghandle->lgh_file->f_dentry->d_inode->i_size) {
 485                 struct llog_rec_hdr *rec;
 486                 struct llog_rec_tail *tail;
 487                 loff_t ppos;
 488
 489                 ppos = cur_offset;
 490
 491                 rc = fsfilt_read_record(loghandle->lgh_ctxt->loc_exp->exp_obd,
 492                                         loghandle->lgh_file, buf, len,
 493                                         &ppos);
 494                 if (rc) {
 495                         CERROR("Cant read llog block at log id "LPU64
 496                                "/%u offset "LPU64"\n",
 497                                loghandle->lgh_id.lgl_oid,
 498                                loghandle->lgh_id.lgl_ogen,
 499                                cur_offset);
 500                         RETURN(rc);
 501                 }
 502
 503                 /* put number of bytes read into rc to make code simpler */
 504                 rc = ppos - cur_offset;
 505                 cur_offset = ppos;
 506
 507                 if (rc == 0) /* end of file, nothing to do */
 508                         RETURN(0);
 509
 510                 if (rc < sizeof(*tail)) {
 511                         CERROR("Invalid llog block at log id "LPU64"/%u offset "
 512                                LPU64"\n", loghandle->lgh_id.lgl_oid,
 513                                loghandle->lgh_id.lgl_ogen, cur_offset);
 514                         RETURN(-EINVAL);
 515                 }
 516
 517                 tail = buf + rc - sizeof(struct llog_rec_tail);
 518
 519                 /* this shouldn't happen */
 520                 if (tail->lrt_index == 0) {
 521                         CERROR("Invalid llog tail at log id "LPU64"/%u offset "
 522                                LPU64"\n", loghandle->lgh_id.lgl_oid,
 523                                loghandle->lgh_id.lgl_ogen, cur_offset);
 524                         RETURN(-EINVAL);
 525                 }
 526                 if (le32_to_cpu(tail->lrt_index) < prev_idx)
 527                         continue;
 528
 529                 /* sanity check that the start of the new buffer is no farther
 530                  * than the record that we wanted.  This shouldn't happen. */
 531                 rec = buf;
 532                 if (le32_to_cpu(rec->lrh_index) > prev_idx) {
 533                         CERROR("missed desired record? %u > %u\n",
 534                                le32_to_cpu(rec->lrh_index), prev_idx);
 535                         RETURN(-ENOENT);
 536                 }
 537                 RETURN(0);
 538         }
 539         RETURN(-EIO);
 540 }
 541
 542 static struct file *llog_filp_open(char *dir, char *name, int flags, int mode)
 543 {
 544         char *logname;
 545         struct file *filp;
 546         int len;
 547
 548         OBD_ALLOC(logname, PATH_MAX);
 549         if (logname == NULL)
 550                 return ERR_PTR(-ENOMEM);
 551
 552         len = snprintf(logname, PATH_MAX, "%s/%s", dir, name);
 553         if (len >= PATH_MAX - 1) {
 554                 filp = ERR_PTR(-ENAMETOOLONG);
 555         } else {
 556                 filp = l_filp_open(logname, flags, mode);
 557                 if (IS_ERR(filp))
 558                         CERROR("logfile creation %s: %ld\n", logname,
 559                                PTR_ERR(filp));
 560         }
 561         OBD_FREE(logname, PATH_MAX);
 562         return filp;
 563 }
 564
 565 /* This is a callback from the llog_* functions.
 566  * Assumes caller has already pushed us into the kernel context. */
 567 static int llog_lvfs_create(struct llog_ctxt *ctxt, struct llog_handle **res,
 568                             struct llog_logid *logid, char *name)
 569 {
 570         struct llog_handle *handle;
 571         struct obd_device *obd;
 572         struct l_dentry *dchild = NULL;
 573         struct obdo *oa = NULL;
 574         int rc = 0, cleanup_phase = 1;
 575         int open_flags = O_RDWR | O_CREAT | O_LARGEFILE;
 576         ENTRY;
 577
 578         handle = llog_alloc_handle();
 579         if (handle == NULL)
 580                 RETURN(-ENOMEM);
 581         *res = handle;
 582
 583         LASSERT(ctxt);
 584         LASSERT(ctxt->loc_exp);
 585         obd = ctxt->loc_exp->exp_obd;
 586
 587         if (logid != NULL) {
 588                 dchild = obd_lvfs_fid2dentry(ctxt->loc_exp, logid->lgl_oid,
 589                                              logid->lgl_ogen, logid->lgl_ogr);
 590
 591                 if (IS_ERR(dchild)) {
 592                         rc = PTR_ERR(dchild);
 593                         CERROR("error looking up logfile "LPX64":0x%x: rc %d\n",
 594                                logid->lgl_oid, logid->lgl_ogen, rc);
 595                         GOTO(cleanup, rc);
 596                 }
 597
 598                 cleanup_phase = 2;
 599                 if (dchild->d_inode == NULL) {
 600                         rc = -ENOENT;
 601                         CERROR("nonexistent log file "LPX64":"LPX64": rc %d\n",
 602                                logid->lgl_oid, logid->lgl_ogr, rc);
 603                         GOTO(cleanup, rc);
 604                 }
 605
 606                 handle->lgh_file = l_dentry_open(&obd->obd_lvfs_ctxt, dchild,
 607                                                     O_RDWR | O_LARGEFILE);
 608                 if (IS_ERR(handle->lgh_file)) {
 609                         rc = PTR_ERR(handle->lgh_file);
 610                         CERROR("error opening logfile "LPX64"0x%x: rc %d\n",
 611                                logid->lgl_oid, logid->lgl_ogen, rc);
 612                         GOTO(cleanup, rc);
 613                 }
 614
 615                 /* assign the value of lgh_id for handle directly */
 616                 handle->lgh_id = *logid;
 617
 618         } else if (name) {
 619                 /* COMPAT_146 */
 620                 if (strcmp(obd->obd_type->typ_name, LUSTRE_MDS_NAME) == 0) {
 621                         handle->lgh_file = llog_filp_open(MDT_LOGS_DIR, name,
 622                                                           open_flags, 0644);
 623                 } else {
 624                         /* end COMPAT_146 */
 625                         handle->lgh_file = llog_filp_open(MOUNT_CONFIGS_DIR,
 626                                                           name, open_flags,
 627                                                           0644);
 628                 }
 629                 if (IS_ERR(handle->lgh_file))
 630                         GOTO(cleanup, rc = PTR_ERR(handle->lgh_file));
 631
 632                 handle->lgh_id.lgl_ogr = 1;
 633                 handle->lgh_id.lgl_oid =
 634                         handle->lgh_file->f_dentry->d_inode->i_ino;
 635                 handle->lgh_id.lgl_ogen =
 636                         handle->lgh_file->f_dentry->d_inode->i_generation;
 637         } else {
 638                 oa = obdo_alloc();
 639                 if (oa == NULL)
 640                         GOTO(cleanup, rc = -ENOMEM);
 641
 642                 oa->o_gr = FILTER_GROUP_LLOG;
 643                 oa->o_valid = OBD_MD_FLGENER | OBD_MD_FLGROUP;
 644
 645                 rc = obd_create(ctxt->loc_exp, oa, NULL, NULL);
 646                 if (rc)
 647                         GOTO(cleanup, rc);
 648
 649                 dchild = obd_lvfs_fid2dentry(ctxt->loc_exp, oa->o_id,
 650                                              oa->o_generation, oa->o_gr);
 651
 652                 if (IS_ERR(dchild))
 653                         GOTO(cleanup, rc = PTR_ERR(dchild));
 654                 cleanup_phase = 2;
 655                 handle->lgh_file = l_dentry_open(&obd->obd_lvfs_ctxt, dchild,
 656                                                  open_flags);
 657                 if (IS_ERR(handle->lgh_file))
 658                         GOTO(cleanup, rc = PTR_ERR(handle->lgh_file));
 659
 660                 handle->lgh_id.lgl_ogr = oa->o_gr;
 661                 handle->lgh_id.lgl_oid = oa->o_id;
 662                 handle->lgh_id.lgl_ogen = oa->o_generation;
 663         }
 664
 665         handle->lgh_ctxt = ctxt;
 666  finish:
 667         if (oa)
 668                 obdo_free(oa);
 669         RETURN(rc);
 670 cleanup:
 671         switch (cleanup_phase) {
 672         case 2:
 673                 l_dput(dchild);
 674         case 1:
 675                 llog_free_handle(handle);
 676         }
 677         goto finish;
 678 }
 679
 680 static int llog_lvfs_close(struct llog_handle *handle)
 681 {
 682         int rc;
 683         ENTRY;
 684
 685         rc = filp_close(handle->lgh_file, 0);
 686         if (rc)
 687                 CERROR("error closing log: rc %d\n", rc);
 688         RETURN(rc);
 689 }
 690
 691 static int llog_lvfs_destroy(struct llog_handle *handle)
 692 {
 693         struct dentry *fdentry;
 694         struct obdo *oa;
 695         struct obd_device *obd = handle->lgh_ctxt->loc_exp->exp_obd;
 696         char *dir;
 697         int rc;
 698         ENTRY;
 699
 700         /* COMPAT_146 */
 701         if (strcmp(obd->obd_type->typ_name, LUSTRE_MDS_NAME) == 0)
 702                 dir = MDT_LOGS_DIR;
 703         else
 704                 /* end COMPAT_146 */
 705                 dir = MOUNT_CONFIGS_DIR;
 706
 707         fdentry = handle->lgh_file->f_dentry;
 708         if (strcmp(fdentry->d_parent->d_name.name, dir) == 0) {
 709                 struct inode *inode = fdentry->d_parent->d_inode;
 710                 struct lvfs_run_ctxt saved;
 711
 712                 push_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 713                 dget(fdentry);
 714                 rc = llog_lvfs_close(handle);
 715
 716                 if (rc == 0) {
 717                         LOCK_INODE_MUTEX(inode);
 718                         rc = vfs_unlink(inode, fdentry);
 719                         UNLOCK_INODE_MUTEX(inode);
 720                 }
 721
 722                 dput(fdentry);
 723                 pop_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 724                 RETURN(rc);
 725         }
 726
 727         oa = obdo_alloc();
 728         if (oa == NULL)
 729                 RETURN(-ENOMEM);
 730
 731         oa->o_id = handle->lgh_id.lgl_oid;
 732         oa->o_gr = handle->lgh_id.lgl_ogr;
 733         oa->o_generation = handle->lgh_id.lgl_ogen;
 734         oa->o_valid = OBD_MD_FLID | OBD_MD_FLGROUP | OBD_MD_FLGENER;
 735
 736         rc = llog_lvfs_close(handle);
 737         if (rc)
 738                 GOTO(out, rc);
 739
 740         rc = obd_destroy(handle->lgh_ctxt->loc_exp, oa, NULL, NULL, NULL);
 741  out:
 742         obdo_free(oa);
 743         RETURN(rc);
 744 }
 745
 746 /* reads the catalog list */
 747 int llog_get_cat_list(struct obd_device *obd, struct obd_device *disk_obd,
 748                       char *name, int count, struct llog_catid *idarray)
 749 {
 750         struct lvfs_run_ctxt saved;
 751         struct l_file *file;
 752         int rc;
 753         int size = sizeof(*idarray) * count;
 754         loff_t off = 0;
 755         ENTRY;
 756
 757         if (!count)
 758                 RETURN(0);
 759
 760         push_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 761         file = filp_open(name, O_RDWR | O_CREAT | O_LARGEFILE, 0700);
 762         if (!file || IS_ERR(file)) {
 763                 rc = PTR_ERR(file);
 764                 CERROR("OBD filter: cannot open/create %s: rc = %d\n",
 765                        name, rc);
 766                 GOTO(out, rc);
 767         }
 768
 769         if (!S_ISREG(file->f_dentry->d_inode->i_mode)) {
 770                 CERROR("%s is not a regular file!: mode = %o\n", name,
 771                        file->f_dentry->d_inode->i_mode);
 772                 GOTO(out, rc = -ENOENT);
 773         }
 774
 775         CDEBUG(D_CONFIG, "cat list: disk size=%d, read=%d\n",
 776                (int)file->f_dentry->d_inode->i_size, size);
 777
 778         rc = fsfilt_read_record(disk_obd, file, idarray, size, &off);
 779         if (rc) {
 780                 CERROR("OBD filter: error reading %s: rc %d\n", name, rc);
 781                 GOTO(out, rc);
 782         }
 783
 784         EXIT;
 785  out:
 786         pop_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 787         if (file && !IS_ERR(file))
 788                 rc = filp_close(file, 0);
 789         return rc;
 790 }
 791 EXPORT_SYMBOL(llog_get_cat_list);
 792
 793 /* writes the cat list */
 794 int llog_put_cat_list(struct obd_device *obd, struct obd_device *disk_obd,
 795                       char *name, int count, struct llog_catid *idarray)
 796 {
 797         struct lvfs_run_ctxt saved;
 798         struct l_file *file;
 799         int rc;
 800         int size = sizeof(*idarray) * count;
 801         loff_t off = 0;
 802
 803         if (!count)
 804                 return (0);
 805
 806         push_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 807         file = filp_open(name, O_RDWR | O_CREAT | O_LARGEFILE, 0700);
 808         if (!file || IS_ERR(file)) {
 809                 rc = PTR_ERR(file);
 810                 CERROR("OBD filter: cannot open/create %s: rc = %d\n",
 811                        name, rc);
 812                 GOTO(out, rc);
 813         }
 814
 815         if (!S_ISREG(file->f_dentry->d_inode->i_mode)) {
 816                 CERROR("%s is not a regular file!: mode = %o\n", name,
 817                        file->f_dentry->d_inode->i_mode);
 818                 GOTO(out, rc = -ENOENT);
 819         }
 820
 821         rc = fsfilt_write_record(disk_obd, file, idarray, size, &off, 1);
 822         if (rc) {
 823                 CDEBUG(D_INODE,"OBD filter: error reading %s: rc %d\n",
 824                        name, rc);
 825                 GOTO(out, rc);
 826         }
 827
 828  out:
 829         pop_ctxt(&saved, &obd->obd_lvfs_ctxt, NULL);
 830         if (file && !IS_ERR(file))
 831                 rc = filp_close(file, 0);
 832         RETURN(rc);
 833 }
 834
 835 struct llog_operations llog_lvfs_ops = {
 836         lop_write_rec:   llog_lvfs_write_rec,
 837         lop_next_block:  llog_lvfs_next_block,
 838         lop_prev_block:  llog_lvfs_prev_block,
 839         lop_read_header: llog_lvfs_read_header,
 840         lop_create:      llog_lvfs_create,
 841         lop_destroy:     llog_lvfs_destroy,
 842         lop_close:       llog_lvfs_close,
 843         //        lop_cancel: llog_lvfs_cancel,
 844 };
 845
 846 EXPORT_SYMBOL(llog_lvfs_ops);
 847
 848 #else /* !__KERNEL__ */
 849
 850 static int llog_lvfs_read_header(struct llog_handle *handle)
 851 {
 852         LBUG();
 853         return 0;
 854 }
 855
 856 static int llog_lvfs_write_rec(struct llog_handle *loghandle,
 857                                struct llog_rec_hdr *rec,
 858                                struct llog_cookie *reccookie, int cookiecount,
 859                                void *buf, int idx)
 860 {
 861         LBUG();
 862         return 0;
 863 }
 864
 865 static int llog_lvfs_next_block(struct llog_handle *loghandle, int *cur_idx,
 866                                 int next_idx, __u64 *cur_offset, void *buf,
 867                                 int len)
 868 {
 869         LBUG();
 870         return 0;
 871 }
 872
 873 static int llog_lvfs_prev_block(struct llog_handle *loghandle,
 874                                 int prev_idx, void *buf, int len)
 875 {
 876         LBUG();
 877         return 0;
 878 }
 879
 880 static int llog_lvfs_create(struct llog_ctxt *ctxt, struct llog_handle **res,
 881                             struct llog_logid *logid, char *name)
 882 {
 883         LBUG();
 884         return 0;
 885 }
 886
 887 static int llog_lvfs_close(struct llog_handle *handle)
 888 {
 889         LBUG();
 890         return 0;
 891 }
 892
 893 static int llog_lvfs_destroy(struct llog_handle *handle)
 894 {
 895         LBUG();
 896         return 0;
 897 }
 898
 899 int llog_get_cat_list(struct obd_device *obd, struct obd_device *disk_obd,
 900                       char *name, int count, struct llog_catid *idarray)
 901 {
 902         LBUG();
 903         return 0;
 904 }
 905
 906 int llog_put_cat_list(struct obd_device *obd, struct obd_device *disk_obd,
 907                       char *name, int count, struct llog_catid *idarray)
 908 {
 909         LBUG();
 910         return 0;
 911 }
 912
 913 struct llog_operations llog_lvfs_ops = {
 914         lop_write_rec:   llog_lvfs_write_rec,
 915         lop_next_block:  llog_lvfs_next_block,
 916         lop_prev_block:  llog_lvfs_prev_block,
 917         lop_read_header: llog_lvfs_read_header,
 918         lop_create:      llog_lvfs_create,
 919         lop_destroy:     llog_lvfs_destroy,
 920         lop_close:       llog_lvfs_close,
 921 //        lop_cancel:      llog_lvfs_cancel,
 922 };
 923 #endif