1 /* 2 * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. 3 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved. 4 * 5 * This copyrighted material is made available to anyone wishing to use, 6 * modify, copy, or redistribute it subject to the terms and conditions 7 * of the GNU General Public License version 2. 8 */ 9 10 #ifndef __INCORE_DOT_H__ 11 #define __INCORE_DOT_H__ 12 13 #include <linux/fs.h> 14 #include <linux/workqueue.h> 15 #include <linux/slow-work.h> 16 #include <linux/dlm.h> 17 #include <linux/buffer_head.h> 18 19 #define DIO_WAIT 0x00000010 20 #define DIO_METADATA 0x00000020 21 #define DIO_ALL 0x00000100 22 23 struct gfs2_log_operations; 24 struct gfs2_log_element; 25 struct gfs2_holder; 26 struct gfs2_glock; 27 struct gfs2_quota_data; 28 struct gfs2_trans; 29 struct gfs2_ail; 30 struct gfs2_jdesc; 31 struct gfs2_sbd; 32 struct lm_lockops; 33 34 typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret); 35 36 struct gfs2_log_header_host { 37 u64 lh_sequence; /* Sequence number of this transaction */ 38 u32 lh_flags; /* GFS2_LOG_HEAD_... */ 39 u32 lh_tail; /* Block number of log tail */ 40 u32 lh_blkno; 41 u32 lh_hash; 42 }; 43 44 /* 45 * Structure of operations that are associated with each 46 * type of element in the log. 47 */ 48 49 struct gfs2_log_operations { 50 void (*lo_add) (struct gfs2_sbd *sdp, struct gfs2_log_element *le); 51 void (*lo_before_commit) (struct gfs2_sbd *sdp); 52 void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_ail *ai); 53 void (*lo_before_scan) (struct gfs2_jdesc *jd, 54 struct gfs2_log_header_host *head, int pass); 55 int (*lo_scan_elements) (struct gfs2_jdesc *jd, unsigned int start, 56 struct gfs2_log_descriptor *ld, __be64 *ptr, 57 int pass); 58 void (*lo_after_scan) (struct gfs2_jdesc *jd, int error, int pass); 59 const char *lo_name; 60 }; 61 62 struct gfs2_log_element { 63 struct list_head le_list; 64 const struct gfs2_log_operations *le_ops; 65 }; 66 67 #define GBF_FULL 1 68 69 struct gfs2_bitmap { 70 struct buffer_head *bi_bh; 71 char *bi_clone; 72 unsigned long bi_flags; 73 u32 bi_offset; 74 u32 bi_start; 75 u32 bi_len; 76 }; 77 78 struct gfs2_rgrpd { 79 struct list_head rd_list; /* Link with superblock */ 80 struct list_head rd_list_mru; 81 struct gfs2_glock *rd_gl; /* Glock for this rgrp */ 82 u64 rd_addr; /* grp block disk address */ 83 u64 rd_data0; /* first data location */ 84 u32 rd_length; /* length of rgrp header in fs blocks */ 85 u32 rd_data; /* num of data blocks in rgrp */ 86 u32 rd_bitbytes; /* number of bytes in data bitmaps */ 87 u32 rd_free; 88 u32 rd_free_clone; 89 u32 rd_dinodes; 90 u64 rd_igeneration; 91 struct gfs2_bitmap *rd_bits; 92 struct mutex rd_mutex; 93 struct gfs2_log_element rd_le; 94 struct gfs2_sbd *rd_sbd; 95 unsigned int rd_bh_count; 96 u32 rd_last_alloc; 97 u32 rd_flags; 98 #define GFS2_RDF_CHECK 0x10000000 /* check for unlinked inodes */ 99 #define GFS2_RDF_UPTODATE 0x20000000 /* rg is up to date */ 100 #define GFS2_RDF_ERROR 0x40000000 /* error in rg */ 101 #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */ 102 }; 103 104 enum gfs2_state_bits { 105 BH_Pinned = BH_PrivateStart, 106 BH_Escaped = BH_PrivateStart + 1, 107 }; 108 109 BUFFER_FNS(Pinned, pinned) 110 TAS_BUFFER_FNS(Pinned, pinned) 111 BUFFER_FNS(Escaped, escaped) 112 TAS_BUFFER_FNS(Escaped, escaped) 113 114 struct gfs2_bufdata { 115 struct buffer_head *bd_bh; 116 struct gfs2_glock *bd_gl; 117 118 union { 119 struct list_head list_tr; 120 u64 blkno; 121 } u; 122 #define bd_list_tr u.list_tr 123 #define bd_blkno u.blkno 124 125 struct gfs2_log_element bd_le; 126 127 struct gfs2_ail *bd_ail; 128 struct list_head bd_ail_st_list; 129 struct list_head bd_ail_gl_list; 130 }; 131 132 /* 133 * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a 134 * prefix of lock_dlm_ gets awkward. 135 */ 136 137 #define GDLM_STRNAME_BYTES 25 138 #define GDLM_LVB_SIZE 32 139 140 enum { 141 DFL_BLOCK_LOCKS = 0, 142 }; 143 144 struct lm_lockname { 145 u64 ln_number; 146 unsigned int ln_type; 147 }; 148 149 #define lm_name_equal(name1, name2) \ 150 (((name1)->ln_number == (name2)->ln_number) && \ 151 ((name1)->ln_type == (name2)->ln_type)) 152 153 154 struct gfs2_glock_operations { 155 void (*go_xmote_th) (struct gfs2_glock *gl); 156 int (*go_xmote_bh) (struct gfs2_glock *gl, struct gfs2_holder *gh); 157 void (*go_inval) (struct gfs2_glock *gl, int flags); 158 int (*go_demote_ok) (const struct gfs2_glock *gl); 159 int (*go_lock) (struct gfs2_holder *gh); 160 void (*go_unlock) (struct gfs2_holder *gh); 161 int (*go_dump)(struct seq_file *seq, const struct gfs2_glock *gl); 162 void (*go_callback) (struct gfs2_glock *gl); 163 const int go_type; 164 const unsigned long go_min_hold_time; 165 }; 166 167 enum { 168 /* States */ 169 HIF_HOLDER = 6, /* Set for gh that "holds" the glock */ 170 HIF_FIRST = 7, 171 HIF_WAIT = 10, 172 }; 173 174 struct gfs2_holder { 175 struct list_head gh_list; 176 177 struct gfs2_glock *gh_gl; 178 struct pid *gh_owner_pid; 179 unsigned int gh_state; 180 unsigned gh_flags; 181 182 int gh_error; 183 unsigned long gh_iflags; /* HIF_... */ 184 unsigned long gh_ip; 185 }; 186 187 enum { 188 GLF_LOCK = 1, 189 GLF_DEMOTE = 3, 190 GLF_PENDING_DEMOTE = 4, 191 GLF_DEMOTE_IN_PROGRESS = 5, 192 GLF_DIRTY = 6, 193 GLF_LFLUSH = 7, 194 GLF_INVALIDATE_IN_PROGRESS = 8, 195 GLF_REPLY_PENDING = 9, 196 GLF_INITIAL = 10, 197 GLF_FROZEN = 11, 198 }; 199 200 struct gfs2_glock { 201 struct hlist_node gl_list; 202 unsigned long gl_flags; /* GLF_... */ 203 struct lm_lockname gl_name; 204 atomic_t gl_ref; 205 206 spinlock_t gl_spin; 207 208 unsigned int gl_state; 209 unsigned int gl_target; 210 unsigned int gl_reply; 211 unsigned int gl_hash; 212 unsigned int gl_req; 213 unsigned int gl_demote_state; /* state requested by remote node */ 214 unsigned long gl_demote_time; /* time of first demote request */ 215 struct list_head gl_holders; 216 217 const struct gfs2_glock_operations *gl_ops; 218 char gl_strname[GDLM_STRNAME_BYTES]; 219 struct dlm_lksb gl_lksb; 220 char gl_lvb[32]; 221 unsigned long gl_tchange; 222 void *gl_object; 223 224 struct list_head gl_lru; 225 226 struct gfs2_sbd *gl_sbd; 227 228 struct inode *gl_aspace; 229 struct list_head gl_ail_list; 230 atomic_t gl_ail_count; 231 struct delayed_work gl_work; 232 struct work_struct gl_delete; 233 }; 234 235 #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */ 236 237 struct gfs2_alloc { 238 /* Quota stuff */ 239 240 struct gfs2_quota_data *al_qd[2*MAXQUOTAS]; 241 struct gfs2_holder al_qd_ghs[2*MAXQUOTAS]; 242 unsigned int al_qd_num; 243 244 u32 al_requested; /* Filled in by caller of gfs2_inplace_reserve() */ 245 u32 al_alloced; /* Filled in by gfs2_alloc_*() */ 246 247 /* Filled in by gfs2_inplace_reserve() */ 248 249 unsigned int al_line; 250 char *al_file; 251 struct gfs2_holder al_ri_gh; 252 struct gfs2_holder al_rgd_gh; 253 struct gfs2_rgrpd *al_rgd; 254 255 }; 256 257 enum { 258 GIF_INVALID = 0, 259 GIF_QD_LOCKED = 1, 260 GIF_SW_PAGED = 3, 261 GIF_USER = 4, /* user inode, not metadata addr space */ 262 }; 263 264 265 struct gfs2_inode { 266 struct inode i_inode; 267 u64 i_no_addr; 268 u64 i_no_formal_ino; 269 u64 i_generation; 270 u64 i_eattr; 271 loff_t i_disksize; 272 unsigned long i_flags; /* GIF_... */ 273 struct gfs2_glock *i_gl; /* Move into i_gh? */ 274 struct gfs2_holder i_iopen_gh; 275 struct gfs2_holder i_gh; /* for prepare/commit_write only */ 276 struct gfs2_alloc *i_alloc; 277 u64 i_goal; /* goal block for allocations */ 278 struct rw_semaphore i_rw_mutex; 279 struct list_head i_trunc_list; 280 u32 i_entries; 281 u32 i_diskflags; 282 u8 i_height; 283 u8 i_depth; 284 }; 285 286 /* 287 * Since i_inode is the first element of struct gfs2_inode, 288 * this is effectively a cast. 289 */ 290 static inline struct gfs2_inode *GFS2_I(struct inode *inode) 291 { 292 return container_of(inode, struct gfs2_inode, i_inode); 293 } 294 295 static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode) 296 { 297 return inode->i_sb->s_fs_info; 298 } 299 300 struct gfs2_file { 301 struct mutex f_fl_mutex; 302 struct gfs2_holder f_fl_gh; 303 }; 304 305 struct gfs2_revoke_replay { 306 struct list_head rr_list; 307 u64 rr_blkno; 308 unsigned int rr_where; 309 }; 310 311 enum { 312 QDF_USER = 0, 313 QDF_CHANGE = 1, 314 QDF_LOCKED = 2, 315 }; 316 317 struct gfs2_quota_data { 318 struct list_head qd_list; 319 struct list_head qd_reclaim; 320 321 atomic_t qd_count; 322 323 u32 qd_id; 324 unsigned long qd_flags; /* QDF_... */ 325 326 s64 qd_change; 327 s64 qd_change_sync; 328 329 unsigned int qd_slot; 330 unsigned int qd_slot_count; 331 332 struct buffer_head *qd_bh; 333 struct gfs2_quota_change *qd_bh_qc; 334 unsigned int qd_bh_count; 335 336 struct gfs2_glock *qd_gl; 337 struct gfs2_quota_lvb qd_qb; 338 339 u64 qd_sync_gen; 340 unsigned long qd_last_warn; 341 }; 342 343 struct gfs2_trans { 344 unsigned long tr_ip; 345 346 unsigned int tr_blocks; 347 unsigned int tr_revokes; 348 unsigned int tr_reserved; 349 350 struct gfs2_holder tr_t_gh; 351 352 int tr_touched; 353 354 unsigned int tr_num_buf; 355 unsigned int tr_num_buf_new; 356 unsigned int tr_num_databuf_new; 357 unsigned int tr_num_buf_rm; 358 unsigned int tr_num_databuf_rm; 359 struct list_head tr_list_buf; 360 361 unsigned int tr_num_revoke; 362 unsigned int tr_num_revoke_rm; 363 }; 364 365 struct gfs2_ail { 366 struct list_head ai_list; 367 368 unsigned int ai_first; 369 struct list_head ai_ail1_list; 370 struct list_head ai_ail2_list; 371 372 u64 ai_sync_gen; 373 }; 374 375 struct gfs2_journal_extent { 376 struct list_head extent_list; 377 378 unsigned int lblock; /* First logical block */ 379 u64 dblock; /* First disk block */ 380 u64 blocks; 381 }; 382 383 struct gfs2_jdesc { 384 struct list_head jd_list; 385 struct list_head extent_list; 386 struct slow_work jd_work; 387 struct inode *jd_inode; 388 unsigned long jd_flags; 389 #define JDF_RECOVERY 1 390 unsigned int jd_jid; 391 unsigned int jd_blocks; 392 }; 393 394 struct gfs2_statfs_change_host { 395 s64 sc_total; 396 s64 sc_free; 397 s64 sc_dinodes; 398 }; 399 400 #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF 401 #define GFS2_QUOTA_OFF 0 402 #define GFS2_QUOTA_ACCOUNT 1 403 #define GFS2_QUOTA_ON 2 404 405 #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED 406 #define GFS2_DATA_WRITEBACK 1 407 #define GFS2_DATA_ORDERED 2 408 409 #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW 410 #define GFS2_ERRORS_WITHDRAW 0 411 #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */ 412 #define GFS2_ERRORS_RO 2 /* place holder for future feature */ 413 #define GFS2_ERRORS_PANIC 3 414 415 struct gfs2_args { 416 char ar_lockproto[GFS2_LOCKNAME_LEN]; /* Name of the Lock Protocol */ 417 char ar_locktable[GFS2_LOCKNAME_LEN]; /* Name of the Lock Table */ 418 char ar_hostdata[GFS2_LOCKNAME_LEN]; /* Host specific data */ 419 unsigned int ar_spectator:1; /* Don't get a journal */ 420 unsigned int ar_ignore_local_fs:1; /* Ignore optimisations */ 421 unsigned int ar_localflocks:1; /* Let the VFS do flock|fcntl */ 422 unsigned int ar_localcaching:1; /* Local caching */ 423 unsigned int ar_debug:1; /* Oops on errors */ 424 unsigned int ar_upgrade:1; /* Upgrade ondisk format */ 425 unsigned int ar_posix_acl:1; /* Enable posix acls */ 426 unsigned int ar_quota:2; /* off/account/on */ 427 unsigned int ar_suiddir:1; /* suiddir support */ 428 unsigned int ar_data:2; /* ordered/writeback */ 429 unsigned int ar_meta:1; /* mount metafs */ 430 unsigned int ar_discard:1; /* discard requests */ 431 unsigned int ar_errors:2; /* errors=withdraw | panic */ 432 unsigned int ar_nobarrier:1; /* do not send barriers */ 433 int ar_commit; /* Commit interval */ 434 int ar_statfs_quantum; /* The fast statfs interval */ 435 int ar_quota_quantum; /* The quota interval */ 436 int ar_statfs_percent; /* The % change to force sync */ 437 }; 438 439 struct gfs2_tune { 440 spinlock_t gt_spin; 441 442 unsigned int gt_incore_log_blocks; 443 unsigned int gt_log_flush_secs; 444 445 unsigned int gt_logd_secs; 446 447 unsigned int gt_quota_simul_sync; /* Max quotavals to sync at once */ 448 unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */ 449 unsigned int gt_quota_scale_num; /* Numerator */ 450 unsigned int gt_quota_scale_den; /* Denominator */ 451 unsigned int gt_quota_quantum; /* Secs between syncs to quota file */ 452 unsigned int gt_new_files_jdata; 453 unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */ 454 unsigned int gt_stall_secs; /* Detects trouble! */ 455 unsigned int gt_complain_secs; 456 unsigned int gt_statfs_quantum; 457 unsigned int gt_statfs_slow; 458 }; 459 460 enum { 461 SDF_JOURNAL_CHECKED = 0, 462 SDF_JOURNAL_LIVE = 1, 463 SDF_SHUTDOWN = 2, 464 SDF_NOBARRIERS = 3, 465 SDF_NORECOVERY = 4, 466 }; 467 468 #define GFS2_FSNAME_LEN 256 469 470 struct gfs2_inum_host { 471 u64 no_formal_ino; 472 u64 no_addr; 473 }; 474 475 struct gfs2_sb_host { 476 u32 sb_magic; 477 u32 sb_type; 478 u32 sb_format; 479 480 u32 sb_fs_format; 481 u32 sb_multihost_format; 482 u32 sb_bsize; 483 u32 sb_bsize_shift; 484 485 struct gfs2_inum_host sb_master_dir; 486 struct gfs2_inum_host sb_root_dir; 487 488 char sb_lockproto[GFS2_LOCKNAME_LEN]; 489 char sb_locktable[GFS2_LOCKNAME_LEN]; 490 u8 sb_uuid[16]; 491 }; 492 493 /* 494 * lm_mount() return values 495 * 496 * ls_jid - the journal ID this node should use 497 * ls_first - this node is the first to mount the file system 498 * ls_lockspace - lock module's context for this file system 499 * ls_ops - lock module's functions 500 */ 501 502 struct lm_lockstruct { 503 unsigned int ls_jid; 504 unsigned int ls_first; 505 unsigned int ls_first_done; 506 unsigned int ls_nodir; 507 const struct lm_lockops *ls_ops; 508 unsigned long ls_flags; 509 dlm_lockspace_t *ls_dlm; 510 511 int ls_recover_jid_done; 512 int ls_recover_jid_status; 513 }; 514 515 struct gfs2_sbd { 516 struct super_block *sd_vfs; 517 struct kobject sd_kobj; 518 unsigned long sd_flags; /* SDF_... */ 519 struct gfs2_sb_host sd_sb; 520 521 /* Constants computed on mount */ 522 523 u32 sd_fsb2bb; 524 u32 sd_fsb2bb_shift; 525 u32 sd_diptrs; /* Number of pointers in a dinode */ 526 u32 sd_inptrs; /* Number of pointers in a indirect block */ 527 u32 sd_jbsize; /* Size of a journaled data block */ 528 u32 sd_hash_bsize; /* sizeof(exhash block) */ 529 u32 sd_hash_bsize_shift; 530 u32 sd_hash_ptrs; /* Number of pointers in a hash block */ 531 u32 sd_qc_per_block; 532 u32 sd_max_dirres; /* Max blocks needed to add a directory entry */ 533 u32 sd_max_height; /* Max height of a file's metadata tree */ 534 u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1]; 535 u32 sd_max_jheight; /* Max height of journaled file's meta tree */ 536 u64 sd_jheightsize[GFS2_MAX_META_HEIGHT + 1]; 537 538 struct gfs2_args sd_args; /* Mount arguments */ 539 struct gfs2_tune sd_tune; /* Filesystem tuning structure */ 540 541 /* Lock Stuff */ 542 543 struct lm_lockstruct sd_lockstruct; 544 struct gfs2_holder sd_live_gh; 545 struct gfs2_glock *sd_rename_gl; 546 struct gfs2_glock *sd_trans_gl; 547 548 /* Inode Stuff */ 549 550 struct dentry *sd_master_dir; 551 struct dentry *sd_root_dir; 552 553 struct inode *sd_jindex; 554 struct inode *sd_statfs_inode; 555 struct inode *sd_sc_inode; 556 struct inode *sd_qc_inode; 557 struct inode *sd_rindex; 558 struct inode *sd_quota_inode; 559 560 /* StatFS stuff */ 561 562 spinlock_t sd_statfs_spin; 563 struct gfs2_statfs_change_host sd_statfs_master; 564 struct gfs2_statfs_change_host sd_statfs_local; 565 int sd_statfs_force_sync; 566 567 /* Resource group stuff */ 568 569 int sd_rindex_uptodate; 570 spinlock_t sd_rindex_spin; 571 struct mutex sd_rindex_mutex; 572 struct list_head sd_rindex_list; 573 struct list_head sd_rindex_mru_list; 574 struct gfs2_rgrpd *sd_rindex_forward; 575 unsigned int sd_rgrps; 576 577 /* Journal index stuff */ 578 579 struct list_head sd_jindex_list; 580 spinlock_t sd_jindex_spin; 581 struct mutex sd_jindex_mutex; 582 unsigned int sd_journals; 583 584 struct gfs2_jdesc *sd_jdesc; 585 struct gfs2_holder sd_journal_gh; 586 struct gfs2_holder sd_jinode_gh; 587 588 struct gfs2_holder sd_sc_gh; 589 struct gfs2_holder sd_qc_gh; 590 591 /* Daemon stuff */ 592 593 struct task_struct *sd_logd_process; 594 struct task_struct *sd_quotad_process; 595 596 /* Quota stuff */ 597 598 struct list_head sd_quota_list; 599 atomic_t sd_quota_count; 600 struct mutex sd_quota_mutex; 601 wait_queue_head_t sd_quota_wait; 602 struct list_head sd_trunc_list; 603 spinlock_t sd_trunc_lock; 604 605 unsigned int sd_quota_slots; 606 unsigned int sd_quota_chunks; 607 unsigned char **sd_quota_bitmap; 608 609 u64 sd_quota_sync_gen; 610 611 /* Log stuff */ 612 613 spinlock_t sd_log_lock; 614 615 unsigned int sd_log_blks_reserved; 616 unsigned int sd_log_commited_buf; 617 unsigned int sd_log_commited_databuf; 618 unsigned int sd_log_commited_revoke; 619 620 unsigned int sd_log_num_buf; 621 unsigned int sd_log_num_revoke; 622 unsigned int sd_log_num_rg; 623 unsigned int sd_log_num_databuf; 624 625 struct list_head sd_log_le_buf; 626 struct list_head sd_log_le_revoke; 627 struct list_head sd_log_le_rg; 628 struct list_head sd_log_le_databuf; 629 struct list_head sd_log_le_ordered; 630 631 atomic_t sd_log_blks_free; 632 struct mutex sd_log_reserve_mutex; 633 634 u64 sd_log_sequence; 635 unsigned int sd_log_head; 636 unsigned int sd_log_tail; 637 int sd_log_idle; 638 639 unsigned long sd_log_flush_time; 640 struct rw_semaphore sd_log_flush_lock; 641 atomic_t sd_log_in_flight; 642 wait_queue_head_t sd_log_flush_wait; 643 644 unsigned int sd_log_flush_head; 645 u64 sd_log_flush_wrapped; 646 647 struct list_head sd_ail1_list; 648 struct list_head sd_ail2_list; 649 u64 sd_ail_sync_gen; 650 651 /* Replay stuff */ 652 653 struct list_head sd_revoke_list; 654 unsigned int sd_replay_tail; 655 656 unsigned int sd_found_blocks; 657 unsigned int sd_found_revokes; 658 unsigned int sd_replayed_blocks; 659 660 /* For quiescing the filesystem */ 661 662 struct gfs2_holder sd_freeze_gh; 663 struct mutex sd_freeze_lock; 664 unsigned int sd_freeze_count; 665 666 char sd_fsname[GFS2_FSNAME_LEN]; 667 char sd_table_name[GFS2_FSNAME_LEN]; 668 char sd_proto_name[GFS2_FSNAME_LEN]; 669 670 /* Debugging crud */ 671 672 unsigned long sd_last_warning; 673 struct dentry *debugfs_dir; /* debugfs directory */ 674 struct dentry *debugfs_dentry_glocks; /* for debugfs */ 675 }; 676 677 #endif /* __INCORE_DOT_H__ */ 678 679