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/kobject.h> 15 #include <linux/workqueue.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 const unsigned long go_flags; 166 #define GLOF_ASPACE 1 167 }; 168 169 enum { 170 /* States */ 171 HIF_HOLDER = 6, /* Set for gh that "holds" the glock */ 172 HIF_FIRST = 7, 173 HIF_WAIT = 10, 174 }; 175 176 struct gfs2_holder { 177 struct list_head gh_list; 178 179 struct gfs2_glock *gh_gl; 180 struct pid *gh_owner_pid; 181 unsigned int gh_state; 182 unsigned gh_flags; 183 184 int gh_error; 185 unsigned long gh_iflags; /* HIF_... */ 186 unsigned long gh_ip; 187 }; 188 189 enum { 190 GLF_LOCK = 1, 191 GLF_DEMOTE = 3, 192 GLF_PENDING_DEMOTE = 4, 193 GLF_DEMOTE_IN_PROGRESS = 5, 194 GLF_DIRTY = 6, 195 GLF_LFLUSH = 7, 196 GLF_INVALIDATE_IN_PROGRESS = 8, 197 GLF_REPLY_PENDING = 9, 198 GLF_INITIAL = 10, 199 GLF_FROZEN = 11, 200 GLF_QUEUED = 12, 201 }; 202 203 struct gfs2_glock { 204 struct hlist_node gl_list; 205 unsigned long gl_flags; /* GLF_... */ 206 struct lm_lockname gl_name; 207 atomic_t gl_ref; 208 209 spinlock_t gl_spin; 210 211 /* State fields protected by gl_spin */ 212 unsigned int gl_state:2, /* Current state */ 213 gl_target:2, /* Target state */ 214 gl_demote_state:2, /* State requested by remote node */ 215 gl_req:2, /* State in last dlm request */ 216 gl_reply:8; /* Last reply from the dlm */ 217 218 unsigned int gl_hash; 219 unsigned long gl_demote_time; /* time of first demote request */ 220 struct list_head gl_holders; 221 222 const struct gfs2_glock_operations *gl_ops; 223 char gl_strname[GDLM_STRNAME_BYTES]; 224 struct dlm_lksb gl_lksb; 225 char gl_lvb[32]; 226 unsigned long gl_tchange; 227 void *gl_object; 228 229 struct list_head gl_lru; 230 231 struct gfs2_sbd *gl_sbd; 232 233 struct list_head gl_ail_list; 234 atomic_t gl_ail_count; 235 struct delayed_work gl_work; 236 struct work_struct gl_delete; 237 }; 238 239 #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */ 240 241 struct gfs2_alloc { 242 /* Quota stuff */ 243 244 struct gfs2_quota_data *al_qd[2*MAXQUOTAS]; 245 struct gfs2_holder al_qd_ghs[2*MAXQUOTAS]; 246 unsigned int al_qd_num; 247 248 u32 al_requested; /* Filled in by caller of gfs2_inplace_reserve() */ 249 u32 al_alloced; /* Filled in by gfs2_alloc_*() */ 250 251 /* Filled in by gfs2_inplace_reserve() */ 252 253 unsigned int al_line; 254 char *al_file; 255 struct gfs2_holder al_ri_gh; 256 struct gfs2_holder al_rgd_gh; 257 struct gfs2_rgrpd *al_rgd; 258 259 }; 260 261 enum { 262 GIF_INVALID = 0, 263 GIF_QD_LOCKED = 1, 264 GIF_SW_PAGED = 3, 265 }; 266 267 268 struct gfs2_inode { 269 struct inode i_inode; 270 u64 i_no_addr; 271 u64 i_no_formal_ino; 272 u64 i_generation; 273 u64 i_eattr; 274 unsigned long i_flags; /* GIF_... */ 275 struct gfs2_glock *i_gl; /* Move into i_gh? */ 276 struct gfs2_holder i_iopen_gh; 277 struct gfs2_holder i_gh; /* for prepare/commit_write only */ 278 struct gfs2_alloc *i_alloc; 279 u64 i_goal; /* goal block for allocations */ 280 struct rw_semaphore i_rw_mutex; 281 struct list_head i_trunc_list; 282 u32 i_entries; 283 u32 i_diskflags; 284 u8 i_height; 285 u8 i_depth; 286 }; 287 288 /* 289 * Since i_inode is the first element of struct gfs2_inode, 290 * this is effectively a cast. 291 */ 292 static inline struct gfs2_inode *GFS2_I(struct inode *inode) 293 { 294 return container_of(inode, struct gfs2_inode, i_inode); 295 } 296 297 static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode) 298 { 299 return inode->i_sb->s_fs_info; 300 } 301 302 struct gfs2_file { 303 struct mutex f_fl_mutex; 304 struct gfs2_holder f_fl_gh; 305 }; 306 307 struct gfs2_revoke_replay { 308 struct list_head rr_list; 309 u64 rr_blkno; 310 unsigned int rr_where; 311 }; 312 313 enum { 314 QDF_USER = 0, 315 QDF_CHANGE = 1, 316 QDF_LOCKED = 2, 317 }; 318 319 struct gfs2_quota_data { 320 struct list_head qd_list; 321 struct list_head qd_reclaim; 322 323 atomic_t qd_count; 324 325 u32 qd_id; 326 unsigned long qd_flags; /* QDF_... */ 327 328 s64 qd_change; 329 s64 qd_change_sync; 330 331 unsigned int qd_slot; 332 unsigned int qd_slot_count; 333 334 struct buffer_head *qd_bh; 335 struct gfs2_quota_change *qd_bh_qc; 336 unsigned int qd_bh_count; 337 338 struct gfs2_glock *qd_gl; 339 struct gfs2_quota_lvb qd_qb; 340 341 u64 qd_sync_gen; 342 unsigned long qd_last_warn; 343 }; 344 345 struct gfs2_trans { 346 unsigned long tr_ip; 347 348 unsigned int tr_blocks; 349 unsigned int tr_revokes; 350 unsigned int tr_reserved; 351 352 struct gfs2_holder tr_t_gh; 353 354 int tr_touched; 355 356 unsigned int tr_num_buf; 357 unsigned int tr_num_buf_new; 358 unsigned int tr_num_databuf_new; 359 unsigned int tr_num_buf_rm; 360 unsigned int tr_num_databuf_rm; 361 struct list_head tr_list_buf; 362 363 unsigned int tr_num_revoke; 364 unsigned int tr_num_revoke_rm; 365 }; 366 367 struct gfs2_ail { 368 struct list_head ai_list; 369 370 unsigned int ai_first; 371 struct list_head ai_ail1_list; 372 struct list_head ai_ail2_list; 373 374 u64 ai_sync_gen; 375 }; 376 377 struct gfs2_journal_extent { 378 struct list_head extent_list; 379 380 unsigned int lblock; /* First logical block */ 381 u64 dblock; /* First disk block */ 382 u64 blocks; 383 }; 384 385 struct gfs2_jdesc { 386 struct list_head jd_list; 387 struct list_head extent_list; 388 struct work_struct jd_work; 389 struct inode *jd_inode; 390 unsigned long jd_flags; 391 #define JDF_RECOVERY 1 392 unsigned int jd_jid; 393 unsigned int jd_blocks; 394 }; 395 396 struct gfs2_statfs_change_host { 397 s64 sc_total; 398 s64 sc_free; 399 s64 sc_dinodes; 400 }; 401 402 #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF 403 #define GFS2_QUOTA_OFF 0 404 #define GFS2_QUOTA_ACCOUNT 1 405 #define GFS2_QUOTA_ON 2 406 407 #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED 408 #define GFS2_DATA_WRITEBACK 1 409 #define GFS2_DATA_ORDERED 2 410 411 #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW 412 #define GFS2_ERRORS_WITHDRAW 0 413 #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */ 414 #define GFS2_ERRORS_RO 2 /* place holder for future feature */ 415 #define GFS2_ERRORS_PANIC 3 416 417 struct gfs2_args { 418 char ar_lockproto[GFS2_LOCKNAME_LEN]; /* Name of the Lock Protocol */ 419 char ar_locktable[GFS2_LOCKNAME_LEN]; /* Name of the Lock Table */ 420 char ar_hostdata[GFS2_LOCKNAME_LEN]; /* Host specific data */ 421 unsigned int ar_spectator:1; /* Don't get a journal */ 422 unsigned int ar_localflocks:1; /* Let the VFS do flock|fcntl */ 423 unsigned int ar_debug:1; /* Oops on errors */ 424 unsigned int ar_posix_acl:1; /* Enable posix acls */ 425 unsigned int ar_quota:2; /* off/account/on */ 426 unsigned int ar_suiddir:1; /* suiddir support */ 427 unsigned int ar_data:2; /* ordered/writeback */ 428 unsigned int ar_meta:1; /* mount metafs */ 429 unsigned int ar_discard:1; /* discard requests */ 430 unsigned int ar_errors:2; /* errors=withdraw | panic */ 431 unsigned int ar_nobarrier:1; /* do not send barriers */ 432 int ar_commit; /* Commit interval */ 433 int ar_statfs_quantum; /* The fast statfs interval */ 434 int ar_quota_quantum; /* The quota interval */ 435 int ar_statfs_percent; /* The % change to force sync */ 436 }; 437 438 struct gfs2_tune { 439 spinlock_t gt_spin; 440 441 unsigned int gt_logd_secs; 442 443 unsigned int gt_quota_simul_sync; /* Max quotavals to sync at once */ 444 unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */ 445 unsigned int gt_quota_scale_num; /* Numerator */ 446 unsigned int gt_quota_scale_den; /* Denominator */ 447 unsigned int gt_quota_quantum; /* Secs between syncs to quota file */ 448 unsigned int gt_new_files_jdata; 449 unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */ 450 unsigned int gt_complain_secs; 451 unsigned int gt_statfs_quantum; 452 unsigned int gt_statfs_slow; 453 }; 454 455 enum { 456 SDF_JOURNAL_CHECKED = 0, 457 SDF_JOURNAL_LIVE = 1, 458 SDF_SHUTDOWN = 2, 459 SDF_NOBARRIERS = 3, 460 SDF_NORECOVERY = 4, 461 SDF_DEMOTE = 5, 462 SDF_NOJOURNALID = 6, 463 }; 464 465 #define GFS2_FSNAME_LEN 256 466 467 struct gfs2_inum_host { 468 u64 no_formal_ino; 469 u64 no_addr; 470 }; 471 472 struct gfs2_sb_host { 473 u32 sb_magic; 474 u32 sb_type; 475 u32 sb_format; 476 477 u32 sb_fs_format; 478 u32 sb_multihost_format; 479 u32 sb_bsize; 480 u32 sb_bsize_shift; 481 482 struct gfs2_inum_host sb_master_dir; 483 struct gfs2_inum_host sb_root_dir; 484 485 char sb_lockproto[GFS2_LOCKNAME_LEN]; 486 char sb_locktable[GFS2_LOCKNAME_LEN]; 487 u8 sb_uuid[16]; 488 }; 489 490 /* 491 * lm_mount() return values 492 * 493 * ls_jid - the journal ID this node should use 494 * ls_first - this node is the first to mount the file system 495 * ls_lockspace - lock module's context for this file system 496 * ls_ops - lock module's functions 497 */ 498 499 struct lm_lockstruct { 500 int ls_jid; 501 unsigned int ls_first; 502 unsigned int ls_first_done; 503 unsigned int ls_nodir; 504 const struct lm_lockops *ls_ops; 505 unsigned long ls_flags; 506 dlm_lockspace_t *ls_dlm; 507 508 int ls_recover_jid_done; 509 int ls_recover_jid_status; 510 }; 511 512 struct gfs2_sbd { 513 struct super_block *sd_vfs; 514 struct kobject sd_kobj; 515 unsigned long sd_flags; /* SDF_... */ 516 struct gfs2_sb_host sd_sb; 517 518 /* Constants computed on mount */ 519 520 u32 sd_fsb2bb; 521 u32 sd_fsb2bb_shift; 522 u32 sd_diptrs; /* Number of pointers in a dinode */ 523 u32 sd_inptrs; /* Number of pointers in a indirect block */ 524 u32 sd_jbsize; /* Size of a journaled data block */ 525 u32 sd_hash_bsize; /* sizeof(exhash block) */ 526 u32 sd_hash_bsize_shift; 527 u32 sd_hash_ptrs; /* Number of pointers in a hash block */ 528 u32 sd_qc_per_block; 529 u32 sd_max_dirres; /* Max blocks needed to add a directory entry */ 530 u32 sd_max_height; /* Max height of a file's metadata tree */ 531 u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1]; 532 u32 sd_max_jheight; /* Max height of journaled file's meta tree */ 533 u64 sd_jheightsize[GFS2_MAX_META_HEIGHT + 1]; 534 535 struct gfs2_args sd_args; /* Mount arguments */ 536 struct gfs2_tune sd_tune; /* Filesystem tuning structure */ 537 538 /* Lock Stuff */ 539 540 struct lm_lockstruct sd_lockstruct; 541 struct gfs2_holder sd_live_gh; 542 struct gfs2_glock *sd_rename_gl; 543 struct gfs2_glock *sd_trans_gl; 544 wait_queue_head_t sd_glock_wait; 545 atomic_t sd_glock_disposal; 546 547 /* Inode Stuff */ 548 549 struct dentry *sd_master_dir; 550 struct dentry *sd_root_dir; 551 552 struct inode *sd_jindex; 553 struct inode *sd_statfs_inode; 554 struct inode *sd_sc_inode; 555 struct inode *sd_qc_inode; 556 struct inode *sd_rindex; 557 struct inode *sd_quota_inode; 558 559 /* StatFS stuff */ 560 561 spinlock_t sd_statfs_spin; 562 struct gfs2_statfs_change_host sd_statfs_master; 563 struct gfs2_statfs_change_host sd_statfs_local; 564 int sd_statfs_force_sync; 565 566 /* Resource group stuff */ 567 568 int sd_rindex_uptodate; 569 spinlock_t sd_rindex_spin; 570 struct mutex sd_rindex_mutex; 571 struct list_head sd_rindex_list; 572 struct list_head sd_rindex_mru_list; 573 struct gfs2_rgrpd *sd_rindex_forward; 574 unsigned int sd_rgrps; 575 unsigned int sd_max_rg_data; 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 int sd_log_commited_revoke; 619 620 atomic_t sd_log_pinned; 621 unsigned int sd_log_num_buf; 622 unsigned int sd_log_num_revoke; 623 unsigned int sd_log_num_rg; 624 unsigned int sd_log_num_databuf; 625 626 struct list_head sd_log_le_buf; 627 struct list_head sd_log_le_revoke; 628 struct list_head sd_log_le_rg; 629 struct list_head sd_log_le_databuf; 630 struct list_head sd_log_le_ordered; 631 632 atomic_t sd_log_thresh1; 633 atomic_t sd_log_thresh2; 634 atomic_t sd_log_blks_free; 635 wait_queue_head_t sd_log_waitq; 636 wait_queue_head_t sd_logd_waitq; 637 638 u64 sd_log_sequence; 639 unsigned int sd_log_head; 640 unsigned int sd_log_tail; 641 int sd_log_idle; 642 643 struct rw_semaphore sd_log_flush_lock; 644 atomic_t sd_log_in_flight; 645 wait_queue_head_t sd_log_flush_wait; 646 647 unsigned int sd_log_flush_head; 648 u64 sd_log_flush_wrapped; 649 650 struct list_head sd_ail1_list; 651 struct list_head sd_ail2_list; 652 u64 sd_ail_sync_gen; 653 654 /* Replay stuff */ 655 656 struct list_head sd_revoke_list; 657 unsigned int sd_replay_tail; 658 659 unsigned int sd_found_blocks; 660 unsigned int sd_found_revokes; 661 unsigned int sd_replayed_blocks; 662 663 /* For quiescing the filesystem */ 664 665 struct gfs2_holder sd_freeze_gh; 666 struct mutex sd_freeze_lock; 667 unsigned int sd_freeze_count; 668 669 char sd_fsname[GFS2_FSNAME_LEN]; 670 char sd_table_name[GFS2_FSNAME_LEN]; 671 char sd_proto_name[GFS2_FSNAME_LEN]; 672 673 /* Debugging crud */ 674 675 unsigned long sd_last_warning; 676 struct dentry *debugfs_dir; /* debugfs directory */ 677 struct dentry *debugfs_dentry_glocks; /* for debugfs */ 678 }; 679 680 #endif /* __INCORE_DOT_H__ */ 681 682