2 * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
3 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
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.
10 #ifndef __INCORE_DOT_H__
11 #define __INCORE_DOT_H__
14 #include <linux/kobject.h>
15 #include <linux/workqueue.h>
16 #include <linux/dlm.h>
17 #include <linux/buffer_head.h>
18 #include <linux/rcupdate.h>
19 #include <linux/rculist_bl.h>
20 #include <linux/completion.h>
21 #include <linux/rbtree.h>
23 #define DIO_WAIT 0x00000010
24 #define DIO_METADATA 0x00000020
26 struct gfs2_log_operations
;
27 struct gfs2_log_element
;
30 struct gfs2_quota_data
;
37 typedef void (*gfs2_glop_bh_t
) (struct gfs2_glock
*gl
, unsigned int ret
);
39 struct gfs2_log_header_host
{
40 u64 lh_sequence
; /* Sequence number of this transaction */
41 u32 lh_flags
; /* GFS2_LOG_HEAD_... */
42 u32 lh_tail
; /* Block number of log tail */
48 * Structure of operations that are associated with each
49 * type of element in the log.
52 struct gfs2_log_operations
{
53 void (*lo_add
) (struct gfs2_sbd
*sdp
, struct gfs2_log_element
*le
);
54 void (*lo_before_commit
) (struct gfs2_sbd
*sdp
);
55 void (*lo_after_commit
) (struct gfs2_sbd
*sdp
, struct gfs2_ail
*ai
);
56 void (*lo_before_scan
) (struct gfs2_jdesc
*jd
,
57 struct gfs2_log_header_host
*head
, int pass
);
58 int (*lo_scan_elements
) (struct gfs2_jdesc
*jd
, unsigned int start
,
59 struct gfs2_log_descriptor
*ld
, __be64
*ptr
,
61 void (*lo_after_scan
) (struct gfs2_jdesc
*jd
, int error
, int pass
);
65 struct gfs2_log_element
{
66 struct list_head le_list
;
67 const struct gfs2_log_operations
*le_ops
;
73 struct buffer_head
*bi_bh
;
75 unsigned long bi_flags
;
82 struct rb_node rd_node
; /* Link with superblock */
83 struct gfs2_glock
*rd_gl
; /* Glock for this rgrp */
84 u64 rd_addr
; /* grp block disk address */
85 u64 rd_data0
; /* first data location */
86 u32 rd_length
; /* length of rgrp header in fs blocks */
87 u32 rd_data
; /* num of data blocks in rgrp */
88 u32 rd_bitbytes
; /* number of bytes in data bitmaps */
93 struct gfs2_bitmap
*rd_bits
;
94 struct gfs2_sbd
*rd_sbd
;
97 #define GFS2_RDF_CHECK 0x10000000 /* check for unlinked inodes */
98 #define GFS2_RDF_UPTODATE 0x20000000 /* rg is up to date */
99 #define GFS2_RDF_ERROR 0x40000000 /* error in rg */
100 #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */
103 enum gfs2_state_bits
{
104 BH_Pinned
= BH_PrivateStart
,
105 BH_Escaped
= BH_PrivateStart
+ 1,
106 BH_Zeronew
= BH_PrivateStart
+ 2,
109 BUFFER_FNS(Pinned
, pinned
)
110 TAS_BUFFER_FNS(Pinned
, pinned
)
111 BUFFER_FNS(Escaped
, escaped
)
112 TAS_BUFFER_FNS(Escaped
, escaped
)
113 BUFFER_FNS(Zeronew
, zeronew
)
114 TAS_BUFFER_FNS(Zeronew
, zeronew
)
116 struct gfs2_bufdata
{
117 struct buffer_head
*bd_bh
;
118 struct gfs2_glock
*bd_gl
;
121 struct list_head list_tr
;
124 #define bd_list_tr u.list_tr
125 #define bd_blkno u.blkno
127 struct gfs2_log_element bd_le
;
129 struct gfs2_ail
*bd_ail
;
130 struct list_head bd_ail_st_list
;
131 struct list_head bd_ail_gl_list
;
135 * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
136 * prefix of lock_dlm_ gets awkward.
139 #define GDLM_STRNAME_BYTES 25
140 #define GDLM_LVB_SIZE 32
145 * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been
146 * held by failed nodes whose journals need recovery. Those locks should
147 * only be used for journal recovery until the journal recovery is done.
148 * This is set by the dlm recover_prep callback and cleared by the
149 * gfs2_control thread when journal recovery is complete. To avoid
150 * races between recover_prep setting and gfs2_control clearing, recover_spin
151 * is held while changing this bit and reading/writing recover_block
154 * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used.
156 * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing
157 * recovery of all journals before allowing other nodes to mount the fs.
158 * This is cleared when FIRST_MOUNT_DONE is set.
160 * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished
161 * recovery of all journals, and now allows other nodes to mount the fs.
163 * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared
164 * BLOCK_LOCKS for the first time. The gfs2_control thread should now
165 * control clearing BLOCK_LOCKS for further recoveries.
167 * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq.
169 * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep()
170 * and recover_done(), i.e. set while recover_block == recover_start.
177 DFL_FIRST_MOUNT_DONE
= 3,
180 DFL_DLM_RECOVERY
= 6,
185 unsigned int ln_type
;
188 #define lm_name_equal(name1, name2) \
189 (((name1)->ln_number == (name2)->ln_number) && \
190 ((name1)->ln_type == (name2)->ln_type))
193 struct gfs2_glock_operations
{
194 void (*go_xmote_th
) (struct gfs2_glock
*gl
);
195 int (*go_xmote_bh
) (struct gfs2_glock
*gl
, struct gfs2_holder
*gh
);
196 void (*go_inval
) (struct gfs2_glock
*gl
, int flags
);
197 int (*go_demote_ok
) (const struct gfs2_glock
*gl
);
198 int (*go_lock
) (struct gfs2_holder
*gh
);
199 void (*go_unlock
) (struct gfs2_holder
*gh
);
200 int (*go_dump
)(struct seq_file
*seq
, const struct gfs2_glock
*gl
);
201 void (*go_callback
) (struct gfs2_glock
*gl
);
203 const unsigned long go_flags
;
204 #define GLOF_ASPACE 1
209 HIF_HOLDER
= 6, /* Set for gh that "holds" the glock */
215 struct list_head gh_list
;
217 struct gfs2_glock
*gh_gl
;
218 struct pid
*gh_owner_pid
;
219 unsigned int gh_state
;
223 unsigned long gh_iflags
; /* HIF_... */
230 GLF_PENDING_DEMOTE
= 4,
231 GLF_DEMOTE_IN_PROGRESS
= 5,
234 GLF_INVALIDATE_IN_PROGRESS
= 8,
235 GLF_REPLY_PENDING
= 9,
240 GLF_OBJECT
= 14, /* Used only for tracing */
244 struct hlist_bl_node gl_list
;
245 unsigned long gl_flags
; /* GLF_... */
246 struct lm_lockname gl_name
;
251 /* State fields protected by gl_spin */
252 unsigned int gl_state
:2, /* Current state */
253 gl_target
:2, /* Target state */
254 gl_demote_state
:2, /* State requested by remote node */
255 gl_req
:2, /* State in last dlm request */
256 gl_reply
:8; /* Last reply from the dlm */
258 unsigned int gl_hash
;
259 unsigned long gl_demote_time
; /* time of first demote request */
261 struct list_head gl_holders
;
263 const struct gfs2_glock_operations
*gl_ops
;
264 char gl_strname
[GDLM_STRNAME_BYTES
];
265 struct dlm_lksb gl_lksb
;
267 unsigned long gl_tchange
;
270 struct list_head gl_lru
;
272 struct gfs2_sbd
*gl_sbd
;
274 struct list_head gl_ail_list
;
275 atomic_t gl_ail_count
;
277 struct delayed_work gl_work
;
278 struct work_struct gl_delete
;
279 struct rcu_head gl_rcu
;
282 #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */
284 struct gfs2_qadata
{ /* quota allocation data */
286 struct gfs2_quota_data
*qa_qd
[2*MAXQUOTAS
];
287 struct gfs2_holder qa_qd_ghs
[2*MAXQUOTAS
];
288 unsigned int qa_qd_num
;
291 struct gfs2_blkreserv
{
292 u32 rs_requested
; /* Filled in by caller of gfs2_inplace_reserve() */
293 struct gfs2_holder rs_rgd_gh
; /* Filled in by gfs2_inplace_reserve() */
299 GIF_ALLOC_FAILED
= 2,
305 struct inode i_inode
;
310 unsigned long i_flags
; /* GIF_... */
311 struct gfs2_glock
*i_gl
; /* Move into i_gh? */
312 struct gfs2_holder i_iopen_gh
;
313 struct gfs2_holder i_gh
; /* for prepare/commit_write only */
314 struct gfs2_qadata
*i_qadata
; /* quota allocation data */
315 struct gfs2_blkreserv
*i_res
; /* resource group block reservation */
316 struct gfs2_rgrpd
*i_rgd
;
317 u64 i_goal
; /* goal block for allocations */
318 struct rw_semaphore i_rw_mutex
;
319 struct list_head i_trunc_list
;
320 __be64
*i_hash_cache
;
328 * Since i_inode is the first element of struct gfs2_inode,
329 * this is effectively a cast.
331 static inline struct gfs2_inode
*GFS2_I(struct inode
*inode
)
333 return container_of(inode
, struct gfs2_inode
, i_inode
);
336 static inline struct gfs2_sbd
*GFS2_SB(const struct inode
*inode
)
338 return inode
->i_sb
->s_fs_info
;
342 struct mutex f_fl_mutex
;
343 struct gfs2_holder f_fl_gh
;
346 struct gfs2_revoke_replay
{
347 struct list_head rr_list
;
349 unsigned int rr_where
;
359 struct gfs2_quota_data
{
360 struct list_head qd_list
;
361 struct list_head qd_reclaim
;
366 unsigned long qd_flags
; /* QDF_... */
371 unsigned int qd_slot
;
372 unsigned int qd_slot_count
;
374 struct buffer_head
*qd_bh
;
375 struct gfs2_quota_change
*qd_bh_qc
;
376 unsigned int qd_bh_count
;
378 struct gfs2_glock
*qd_gl
;
379 struct gfs2_quota_lvb qd_qb
;
382 unsigned long qd_last_warn
;
388 unsigned int tr_blocks
;
389 unsigned int tr_revokes
;
390 unsigned int tr_reserved
;
392 struct gfs2_holder tr_t_gh
;
396 unsigned int tr_num_buf
;
397 unsigned int tr_num_buf_new
;
398 unsigned int tr_num_databuf_new
;
399 unsigned int tr_num_buf_rm
;
400 unsigned int tr_num_databuf_rm
;
401 struct list_head tr_list_buf
;
403 unsigned int tr_num_revoke
;
404 unsigned int tr_num_revoke_rm
;
408 struct list_head ai_list
;
410 unsigned int ai_first
;
411 struct list_head ai_ail1_list
;
412 struct list_head ai_ail2_list
;
415 struct gfs2_journal_extent
{
416 struct list_head extent_list
;
418 unsigned int lblock
; /* First logical block */
419 u64 dblock
; /* First disk block */
424 struct list_head jd_list
;
425 struct list_head extent_list
;
426 struct work_struct jd_work
;
427 struct inode
*jd_inode
;
428 unsigned long jd_flags
;
429 #define JDF_RECOVERY 1
431 unsigned int jd_blocks
;
432 int jd_recover_error
;
435 struct gfs2_statfs_change_host
{
441 #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF
442 #define GFS2_QUOTA_OFF 0
443 #define GFS2_QUOTA_ACCOUNT 1
444 #define GFS2_QUOTA_ON 2
446 #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED
447 #define GFS2_DATA_WRITEBACK 1
448 #define GFS2_DATA_ORDERED 2
450 #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW
451 #define GFS2_ERRORS_WITHDRAW 0
452 #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */
453 #define GFS2_ERRORS_RO 2 /* place holder for future feature */
454 #define GFS2_ERRORS_PANIC 3
457 char ar_lockproto
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Protocol */
458 char ar_locktable
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Table */
459 char ar_hostdata
[GFS2_LOCKNAME_LEN
]; /* Host specific data */
460 unsigned int ar_spectator
:1; /* Don't get a journal */
461 unsigned int ar_localflocks
:1; /* Let the VFS do flock|fcntl */
462 unsigned int ar_debug
:1; /* Oops on errors */
463 unsigned int ar_posix_acl
:1; /* Enable posix acls */
464 unsigned int ar_quota
:2; /* off/account/on */
465 unsigned int ar_suiddir
:1; /* suiddir support */
466 unsigned int ar_data
:2; /* ordered/writeback */
467 unsigned int ar_meta
:1; /* mount metafs */
468 unsigned int ar_discard
:1; /* discard requests */
469 unsigned int ar_errors
:2; /* errors=withdraw | panic */
470 unsigned int ar_nobarrier
:1; /* do not send barriers */
471 int ar_commit
; /* Commit interval */
472 int ar_statfs_quantum
; /* The fast statfs interval */
473 int ar_quota_quantum
; /* The quota interval */
474 int ar_statfs_percent
; /* The % change to force sync */
480 unsigned int gt_logd_secs
;
482 unsigned int gt_quota_simul_sync
; /* Max quotavals to sync at once */
483 unsigned int gt_quota_warn_period
; /* Secs between quota warn msgs */
484 unsigned int gt_quota_scale_num
; /* Numerator */
485 unsigned int gt_quota_scale_den
; /* Denominator */
486 unsigned int gt_quota_quantum
; /* Secs between syncs to quota file */
487 unsigned int gt_new_files_jdata
;
488 unsigned int gt_max_readahead
; /* Max bytes to read-ahead from disk */
489 unsigned int gt_complain_secs
;
490 unsigned int gt_statfs_quantum
;
491 unsigned int gt_statfs_slow
;
495 SDF_JOURNAL_CHECKED
= 0,
496 SDF_JOURNAL_LIVE
= 1,
502 SDF_RORECOVERY
= 7, /* read only recovery */
505 #define GFS2_FSNAME_LEN 256
507 struct gfs2_inum_host
{
512 struct gfs2_sb_host
{
518 u32 sb_multihost_format
;
522 struct gfs2_inum_host sb_master_dir
;
523 struct gfs2_inum_host sb_root_dir
;
525 char sb_lockproto
[GFS2_LOCKNAME_LEN
];
526 char sb_locktable
[GFS2_LOCKNAME_LEN
];
530 * lm_mount() return values
532 * ls_jid - the journal ID this node should use
533 * ls_first - this node is the first to mount the file system
534 * ls_lockspace - lock module's context for this file system
535 * ls_ops - lock module's functions
538 struct lm_lockstruct
{
540 unsigned int ls_first
;
541 unsigned int ls_nodir
;
542 const struct lm_lockops
*ls_ops
;
543 dlm_lockspace_t
*ls_dlm
;
545 int ls_recover_jid_done
; /* These two are deprecated, */
546 int ls_recover_jid_status
; /* used previously by gfs_controld */
548 struct dlm_lksb ls_mounted_lksb
; /* mounted_lock */
549 struct dlm_lksb ls_control_lksb
; /* control_lock */
550 char ls_control_lvb
[GDLM_LVB_SIZE
]; /* control_lock lvb */
551 struct completion ls_sync_wait
; /* {control,mounted}_{lock,unlock} */
553 spinlock_t ls_recover_spin
; /* protects following fields */
554 unsigned long ls_recover_flags
; /* DFL_ */
555 uint32_t ls_recover_mount
; /* gen in first recover_done cb */
556 uint32_t ls_recover_start
; /* gen in last recover_done cb */
557 uint32_t ls_recover_block
; /* copy recover_start in last recover_prep */
558 uint32_t ls_recover_size
; /* size of recover_submit, recover_result */
559 uint32_t *ls_recover_submit
; /* gen in last recover_slot cb per jid */
560 uint32_t *ls_recover_result
; /* result of last jid recovery */
564 struct super_block
*sd_vfs
;
565 struct kobject sd_kobj
;
566 unsigned long sd_flags
; /* SDF_... */
567 struct gfs2_sb_host sd_sb
;
569 /* Constants computed on mount */
573 u32 sd_diptrs
; /* Number of pointers in a dinode */
574 u32 sd_inptrs
; /* Number of pointers in a indirect block */
575 u32 sd_jbsize
; /* Size of a journaled data block */
576 u32 sd_hash_bsize
; /* sizeof(exhash block) */
577 u32 sd_hash_bsize_shift
;
578 u32 sd_hash_ptrs
; /* Number of pointers in a hash block */
580 u32 sd_max_dirres
; /* Max blocks needed to add a directory entry */
581 u32 sd_max_height
; /* Max height of a file's metadata tree */
582 u64 sd_heightsize
[GFS2_MAX_META_HEIGHT
+ 1];
583 u32 sd_max_jheight
; /* Max height of journaled file's meta tree */
584 u64 sd_jheightsize
[GFS2_MAX_META_HEIGHT
+ 1];
586 struct gfs2_args sd_args
; /* Mount arguments */
587 struct gfs2_tune sd_tune
; /* Filesystem tuning structure */
591 struct lm_lockstruct sd_lockstruct
;
592 struct gfs2_holder sd_live_gh
;
593 struct gfs2_glock
*sd_rename_gl
;
594 struct gfs2_glock
*sd_trans_gl
;
595 wait_queue_head_t sd_glock_wait
;
596 atomic_t sd_glock_disposal
;
597 struct completion sd_locking_init
;
598 struct delayed_work sd_control_work
;
602 struct dentry
*sd_master_dir
;
603 struct dentry
*sd_root_dir
;
605 struct inode
*sd_jindex
;
606 struct inode
*sd_statfs_inode
;
607 struct inode
*sd_sc_inode
;
608 struct inode
*sd_qc_inode
;
609 struct inode
*sd_rindex
;
610 struct inode
*sd_quota_inode
;
614 spinlock_t sd_statfs_spin
;
615 struct gfs2_statfs_change_host sd_statfs_master
;
616 struct gfs2_statfs_change_host sd_statfs_local
;
617 int sd_statfs_force_sync
;
619 /* Resource group stuff */
621 int sd_rindex_uptodate
;
622 spinlock_t sd_rindex_spin
;
623 struct mutex sd_rindex_mutex
;
624 struct rb_root sd_rindex_tree
;
625 unsigned int sd_rgrps
;
626 unsigned int sd_max_rg_data
;
628 /* Journal index stuff */
630 struct list_head sd_jindex_list
;
631 spinlock_t sd_jindex_spin
;
632 struct mutex sd_jindex_mutex
;
633 unsigned int sd_journals
;
635 struct gfs2_jdesc
*sd_jdesc
;
636 struct gfs2_holder sd_journal_gh
;
637 struct gfs2_holder sd_jinode_gh
;
639 struct gfs2_holder sd_sc_gh
;
640 struct gfs2_holder sd_qc_gh
;
644 struct task_struct
*sd_logd_process
;
645 struct task_struct
*sd_quotad_process
;
649 struct list_head sd_quota_list
;
650 atomic_t sd_quota_count
;
651 struct mutex sd_quota_mutex
;
652 wait_queue_head_t sd_quota_wait
;
653 struct list_head sd_trunc_list
;
654 spinlock_t sd_trunc_lock
;
656 unsigned int sd_quota_slots
;
657 unsigned int sd_quota_chunks
;
658 unsigned char **sd_quota_bitmap
;
660 u64 sd_quota_sync_gen
;
664 spinlock_t sd_log_lock
;
666 unsigned int sd_log_blks_reserved
;
667 unsigned int sd_log_commited_buf
;
668 unsigned int sd_log_commited_databuf
;
669 int sd_log_commited_revoke
;
671 atomic_t sd_log_pinned
;
672 unsigned int sd_log_num_buf
;
673 unsigned int sd_log_num_revoke
;
674 unsigned int sd_log_num_rg
;
675 unsigned int sd_log_num_databuf
;
677 struct list_head sd_log_le_buf
;
678 struct list_head sd_log_le_revoke
;
679 struct list_head sd_log_le_rg
;
680 struct list_head sd_log_le_databuf
;
681 struct list_head sd_log_le_ordered
;
683 atomic_t sd_log_thresh1
;
684 atomic_t sd_log_thresh2
;
685 atomic_t sd_log_blks_free
;
686 wait_queue_head_t sd_log_waitq
;
687 wait_queue_head_t sd_logd_waitq
;
690 unsigned int sd_log_head
;
691 unsigned int sd_log_tail
;
694 struct rw_semaphore sd_log_flush_lock
;
695 atomic_t sd_log_in_flight
;
696 wait_queue_head_t sd_log_flush_wait
;
698 unsigned int sd_log_flush_head
;
699 u64 sd_log_flush_wrapped
;
701 spinlock_t sd_ail_lock
;
702 struct list_head sd_ail1_list
;
703 struct list_head sd_ail2_list
;
707 struct list_head sd_revoke_list
;
708 unsigned int sd_replay_tail
;
710 unsigned int sd_found_blocks
;
711 unsigned int sd_found_revokes
;
712 unsigned int sd_replayed_blocks
;
714 /* For quiescing the filesystem */
716 struct gfs2_holder sd_freeze_gh
;
717 struct mutex sd_freeze_lock
;
718 unsigned int sd_freeze_count
;
720 char sd_fsname
[GFS2_FSNAME_LEN
];
721 char sd_table_name
[GFS2_FSNAME_LEN
];
722 char sd_proto_name
[GFS2_FSNAME_LEN
];
726 unsigned long sd_last_warning
;
727 struct dentry
*debugfs_dir
; /* debugfs directory */
728 struct dentry
*debugfs_dentry_glocks
; /* for debugfs */
731 #endif /* __INCORE_DOT_H__ */