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/workqueue.h>
15 #include <linux/dlm.h>
16 #include <linux/buffer_head.h>
18 #define DIO_WAIT 0x00000010
19 #define DIO_METADATA 0x00000020
20 #define DIO_ALL 0x00000100
22 struct gfs2_log_operations
;
23 struct gfs2_log_element
;
26 struct gfs2_quota_data
;
33 typedef void (*gfs2_glop_bh_t
) (struct gfs2_glock
*gl
, unsigned int ret
);
35 struct gfs2_log_header_host
{
36 u64 lh_sequence
; /* Sequence number of this transaction */
37 u32 lh_flags
; /* GFS2_LOG_HEAD_... */
38 u32 lh_tail
; /* Block number of log tail */
44 * Structure of operations that are associated with each
45 * type of element in the log.
48 struct gfs2_log_operations
{
49 void (*lo_add
) (struct gfs2_sbd
*sdp
, struct gfs2_log_element
*le
);
50 void (*lo_before_commit
) (struct gfs2_sbd
*sdp
);
51 void (*lo_after_commit
) (struct gfs2_sbd
*sdp
, struct gfs2_ail
*ai
);
52 void (*lo_before_scan
) (struct gfs2_jdesc
*jd
,
53 struct gfs2_log_header_host
*head
, int pass
);
54 int (*lo_scan_elements
) (struct gfs2_jdesc
*jd
, unsigned int start
,
55 struct gfs2_log_descriptor
*ld
, __be64
*ptr
,
57 void (*lo_after_scan
) (struct gfs2_jdesc
*jd
, int error
, int pass
);
61 struct gfs2_log_element
{
62 struct list_head le_list
;
63 const struct gfs2_log_operations
*le_ops
;
69 struct buffer_head
*bi_bh
;
71 unsigned long bi_flags
;
78 struct list_head rd_list
; /* Link with superblock */
79 struct list_head rd_list_mru
;
80 struct gfs2_glock
*rd_gl
; /* Glock for this rgrp */
81 u64 rd_addr
; /* grp block disk address */
82 u64 rd_data0
; /* first data location */
83 u32 rd_length
; /* length of rgrp header in fs blocks */
84 u32 rd_data
; /* num of data blocks in rgrp */
85 u32 rd_bitbytes
; /* number of bytes in data bitmaps */
90 struct gfs2_bitmap
*rd_bits
;
91 struct mutex rd_mutex
;
92 struct gfs2_log_element rd_le
;
93 struct gfs2_sbd
*rd_sbd
;
94 unsigned int rd_bh_count
;
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,
108 BUFFER_FNS(Pinned
, pinned
)
109 TAS_BUFFER_FNS(Pinned
, pinned
)
110 BUFFER_FNS(Escaped
, escaped
)
111 TAS_BUFFER_FNS(Escaped
, escaped
)
113 struct gfs2_bufdata
{
114 struct buffer_head
*bd_bh
;
115 struct gfs2_glock
*bd_gl
;
118 struct list_head list_tr
;
121 #define bd_list_tr u.list_tr
122 #define bd_blkno u.blkno
124 struct gfs2_log_element bd_le
;
126 struct gfs2_ail
*bd_ail
;
127 struct list_head bd_ail_st_list
;
128 struct list_head bd_ail_gl_list
;
132 * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
133 * prefix of lock_dlm_ gets awkward.
136 #define GDLM_STRNAME_BYTES 25
137 #define GDLM_LVB_SIZE 32
145 unsigned int ln_type
;
148 #define lm_name_equal(name1, name2) \
149 (((name1)->ln_number == (name2)->ln_number) && \
150 ((name1)->ln_type == (name2)->ln_type))
153 struct gfs2_glock_operations
{
154 void (*go_xmote_th
) (struct gfs2_glock
*gl
);
155 int (*go_xmote_bh
) (struct gfs2_glock
*gl
, struct gfs2_holder
*gh
);
156 void (*go_inval
) (struct gfs2_glock
*gl
, int flags
);
157 int (*go_demote_ok
) (const struct gfs2_glock
*gl
);
158 int (*go_lock
) (struct gfs2_holder
*gh
);
159 void (*go_unlock
) (struct gfs2_holder
*gh
);
160 int (*go_dump
)(struct seq_file
*seq
, const struct gfs2_glock
*gl
);
161 void (*go_callback
) (struct gfs2_glock
*gl
);
163 const unsigned long go_min_hold_time
;
164 const unsigned long go_flags
;
165 #define GLOF_ASPACE 1
170 HIF_HOLDER
= 6, /* Set for gh that "holds" the glock */
176 struct list_head gh_list
;
178 struct gfs2_glock
*gh_gl
;
179 struct pid
*gh_owner_pid
;
180 unsigned int gh_state
;
184 unsigned long gh_iflags
; /* HIF_... */
191 GLF_PENDING_DEMOTE
= 4,
192 GLF_DEMOTE_IN_PROGRESS
= 5,
195 GLF_INVALIDATE_IN_PROGRESS
= 8,
196 GLF_REPLY_PENDING
= 9,
203 struct hlist_node gl_list
;
204 unsigned long gl_flags
; /* GLF_... */
205 struct lm_lockname gl_name
;
210 /* State fields protected by gl_spin */
211 unsigned int gl_state
:2, /* Current state */
212 gl_target
:2, /* Target state */
213 gl_demote_state
:2, /* State requested by remote node */
214 gl_req
:2, /* State in last dlm request */
215 gl_reply
:8; /* Last reply from the dlm */
217 unsigned int gl_hash
;
218 unsigned long gl_demote_time
; /* time of first demote request */
219 struct list_head gl_holders
;
221 const struct gfs2_glock_operations
*gl_ops
;
222 char gl_strname
[GDLM_STRNAME_BYTES
];
223 struct dlm_lksb gl_lksb
;
225 unsigned long gl_tchange
;
228 struct list_head gl_lru
;
230 struct gfs2_sbd
*gl_sbd
;
232 struct list_head gl_ail_list
;
233 atomic_t gl_ail_count
;
234 struct delayed_work gl_work
;
235 struct work_struct gl_delete
;
238 #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */
243 struct gfs2_quota_data
*al_qd
[2*MAXQUOTAS
];
244 struct gfs2_holder al_qd_ghs
[2*MAXQUOTAS
];
245 unsigned int al_qd_num
;
247 u32 al_requested
; /* Filled in by caller of gfs2_inplace_reserve() */
248 u32 al_alloced
; /* Filled in by gfs2_alloc_*() */
250 /* Filled in by gfs2_inplace_reserve() */
252 unsigned int al_line
;
254 struct gfs2_holder al_ri_gh
;
255 struct gfs2_holder al_rgd_gh
;
256 struct gfs2_rgrpd
*al_rgd
;
268 struct inode i_inode
;
273 unsigned long i_flags
; /* GIF_... */
274 struct gfs2_glock
*i_gl
; /* Move into i_gh? */
275 struct gfs2_holder i_iopen_gh
;
276 struct gfs2_holder i_gh
; /* for prepare/commit_write only */
277 struct gfs2_alloc
*i_alloc
;
278 u64 i_goal
; /* goal block for allocations */
279 struct rw_semaphore i_rw_mutex
;
280 struct list_head i_trunc_list
;
288 * Since i_inode is the first element of struct gfs2_inode,
289 * this is effectively a cast.
291 static inline struct gfs2_inode
*GFS2_I(struct inode
*inode
)
293 return container_of(inode
, struct gfs2_inode
, i_inode
);
296 static inline struct gfs2_sbd
*GFS2_SB(const struct inode
*inode
)
298 return inode
->i_sb
->s_fs_info
;
302 struct mutex f_fl_mutex
;
303 struct gfs2_holder f_fl_gh
;
306 struct gfs2_revoke_replay
{
307 struct list_head rr_list
;
309 unsigned int rr_where
;
318 struct gfs2_quota_data
{
319 struct list_head qd_list
;
320 struct list_head qd_reclaim
;
325 unsigned long qd_flags
; /* QDF_... */
330 unsigned int qd_slot
;
331 unsigned int qd_slot_count
;
333 struct buffer_head
*qd_bh
;
334 struct gfs2_quota_change
*qd_bh_qc
;
335 unsigned int qd_bh_count
;
337 struct gfs2_glock
*qd_gl
;
338 struct gfs2_quota_lvb qd_qb
;
341 unsigned long qd_last_warn
;
347 unsigned int tr_blocks
;
348 unsigned int tr_revokes
;
349 unsigned int tr_reserved
;
351 struct gfs2_holder tr_t_gh
;
355 unsigned int tr_num_buf
;
356 unsigned int tr_num_buf_new
;
357 unsigned int tr_num_databuf_new
;
358 unsigned int tr_num_buf_rm
;
359 unsigned int tr_num_databuf_rm
;
360 struct list_head tr_list_buf
;
362 unsigned int tr_num_revoke
;
363 unsigned int tr_num_revoke_rm
;
367 struct list_head ai_list
;
369 unsigned int ai_first
;
370 struct list_head ai_ail1_list
;
371 struct list_head ai_ail2_list
;
376 struct gfs2_journal_extent
{
377 struct list_head extent_list
;
379 unsigned int lblock
; /* First logical block */
380 u64 dblock
; /* First disk block */
385 struct list_head jd_list
;
386 struct list_head extent_list
;
387 struct work_struct jd_work
;
388 struct inode
*jd_inode
;
389 unsigned long jd_flags
;
390 #define JDF_RECOVERY 1
392 unsigned int jd_blocks
;
395 struct gfs2_statfs_change_host
{
401 #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF
402 #define GFS2_QUOTA_OFF 0
403 #define GFS2_QUOTA_ACCOUNT 1
404 #define GFS2_QUOTA_ON 2
406 #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED
407 #define GFS2_DATA_WRITEBACK 1
408 #define GFS2_DATA_ORDERED 2
410 #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW
411 #define GFS2_ERRORS_WITHDRAW 0
412 #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */
413 #define GFS2_ERRORS_RO 2 /* place holder for future feature */
414 #define GFS2_ERRORS_PANIC 3
417 char ar_lockproto
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Protocol */
418 char ar_locktable
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Table */
419 char ar_hostdata
[GFS2_LOCKNAME_LEN
]; /* Host specific data */
420 unsigned int ar_spectator
:1; /* Don't get a journal */
421 unsigned int ar_localflocks
:1; /* Let the VFS do flock|fcntl */
422 unsigned int ar_debug
:1; /* Oops on errors */
423 unsigned int ar_posix_acl
:1; /* Enable posix acls */
424 unsigned int ar_quota
:2; /* off/account/on */
425 unsigned int ar_suiddir
:1; /* suiddir support */
426 unsigned int ar_data
:2; /* ordered/writeback */
427 unsigned int ar_meta
:1; /* mount metafs */
428 unsigned int ar_discard
:1; /* discard requests */
429 unsigned int ar_errors
:2; /* errors=withdraw | panic */
430 unsigned int ar_nobarrier
:1; /* do not send barriers */
431 int ar_commit
; /* Commit interval */
432 int ar_statfs_quantum
; /* The fast statfs interval */
433 int ar_quota_quantum
; /* The quota interval */
434 int ar_statfs_percent
; /* The % change to force sync */
440 unsigned int gt_logd_secs
;
442 unsigned int gt_quota_simul_sync
; /* Max quotavals to sync at once */
443 unsigned int gt_quota_warn_period
; /* Secs between quota warn msgs */
444 unsigned int gt_quota_scale_num
; /* Numerator */
445 unsigned int gt_quota_scale_den
; /* Denominator */
446 unsigned int gt_quota_quantum
; /* Secs between syncs to quota file */
447 unsigned int gt_new_files_jdata
;
448 unsigned int gt_max_readahead
; /* Max bytes to read-ahead from disk */
449 unsigned int gt_complain_secs
;
450 unsigned int gt_statfs_quantum
;
451 unsigned int gt_statfs_slow
;
455 SDF_JOURNAL_CHECKED
= 0,
456 SDF_JOURNAL_LIVE
= 1,
464 #define GFS2_FSNAME_LEN 256
466 struct gfs2_inum_host
{
471 struct gfs2_sb_host
{
477 u32 sb_multihost_format
;
481 struct gfs2_inum_host sb_master_dir
;
482 struct gfs2_inum_host sb_root_dir
;
484 char sb_lockproto
[GFS2_LOCKNAME_LEN
];
485 char sb_locktable
[GFS2_LOCKNAME_LEN
];
490 * lm_mount() return values
492 * ls_jid - the journal ID this node should use
493 * ls_first - this node is the first to mount the file system
494 * ls_lockspace - lock module's context for this file system
495 * ls_ops - lock module's functions
498 struct lm_lockstruct
{
500 unsigned int ls_first
;
501 unsigned int ls_first_done
;
502 unsigned int ls_nodir
;
503 const struct lm_lockops
*ls_ops
;
504 unsigned long ls_flags
;
505 dlm_lockspace_t
*ls_dlm
;
507 int ls_recover_jid_done
;
508 int ls_recover_jid_status
;
512 struct super_block
*sd_vfs
;
513 struct kobject sd_kobj
;
514 unsigned long sd_flags
; /* SDF_... */
515 struct gfs2_sb_host sd_sb
;
517 /* Constants computed on mount */
521 u32 sd_diptrs
; /* Number of pointers in a dinode */
522 u32 sd_inptrs
; /* Number of pointers in a indirect block */
523 u32 sd_jbsize
; /* Size of a journaled data block */
524 u32 sd_hash_bsize
; /* sizeof(exhash block) */
525 u32 sd_hash_bsize_shift
;
526 u32 sd_hash_ptrs
; /* Number of pointers in a hash block */
528 u32 sd_max_dirres
; /* Max blocks needed to add a directory entry */
529 u32 sd_max_height
; /* Max height of a file's metadata tree */
530 u64 sd_heightsize
[GFS2_MAX_META_HEIGHT
+ 1];
531 u32 sd_max_jheight
; /* Max height of journaled file's meta tree */
532 u64 sd_jheightsize
[GFS2_MAX_META_HEIGHT
+ 1];
534 struct gfs2_args sd_args
; /* Mount arguments */
535 struct gfs2_tune sd_tune
; /* Filesystem tuning structure */
539 struct lm_lockstruct sd_lockstruct
;
540 struct gfs2_holder sd_live_gh
;
541 struct gfs2_glock
*sd_rename_gl
;
542 struct gfs2_glock
*sd_trans_gl
;
543 wait_queue_head_t sd_glock_wait
;
544 atomic_t sd_glock_disposal
;
548 struct dentry
*sd_master_dir
;
549 struct dentry
*sd_root_dir
;
551 struct inode
*sd_jindex
;
552 struct inode
*sd_statfs_inode
;
553 struct inode
*sd_sc_inode
;
554 struct inode
*sd_qc_inode
;
555 struct inode
*sd_rindex
;
556 struct inode
*sd_quota_inode
;
560 spinlock_t sd_statfs_spin
;
561 struct gfs2_statfs_change_host sd_statfs_master
;
562 struct gfs2_statfs_change_host sd_statfs_local
;
563 int sd_statfs_force_sync
;
565 /* Resource group stuff */
567 int sd_rindex_uptodate
;
568 spinlock_t sd_rindex_spin
;
569 struct mutex sd_rindex_mutex
;
570 struct list_head sd_rindex_list
;
571 struct list_head sd_rindex_mru_list
;
572 struct gfs2_rgrpd
*sd_rindex_forward
;
573 unsigned int sd_rgrps
;
574 unsigned int sd_max_rg_data
;
576 /* Journal index stuff */
578 struct list_head sd_jindex_list
;
579 spinlock_t sd_jindex_spin
;
580 struct mutex sd_jindex_mutex
;
581 unsigned int sd_journals
;
583 struct gfs2_jdesc
*sd_jdesc
;
584 struct gfs2_holder sd_journal_gh
;
585 struct gfs2_holder sd_jinode_gh
;
587 struct gfs2_holder sd_sc_gh
;
588 struct gfs2_holder sd_qc_gh
;
592 struct task_struct
*sd_logd_process
;
593 struct task_struct
*sd_quotad_process
;
597 struct list_head sd_quota_list
;
598 atomic_t sd_quota_count
;
599 struct mutex sd_quota_mutex
;
600 wait_queue_head_t sd_quota_wait
;
601 struct list_head sd_trunc_list
;
602 spinlock_t sd_trunc_lock
;
604 unsigned int sd_quota_slots
;
605 unsigned int sd_quota_chunks
;
606 unsigned char **sd_quota_bitmap
;
608 u64 sd_quota_sync_gen
;
612 spinlock_t sd_log_lock
;
614 unsigned int sd_log_blks_reserved
;
615 unsigned int sd_log_commited_buf
;
616 unsigned int sd_log_commited_databuf
;
617 int sd_log_commited_revoke
;
619 atomic_t sd_log_pinned
;
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
;
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
;
631 atomic_t sd_log_thresh1
;
632 atomic_t sd_log_thresh2
;
633 atomic_t sd_log_blks_free
;
634 wait_queue_head_t sd_log_waitq
;
635 wait_queue_head_t sd_logd_waitq
;
638 unsigned int sd_log_head
;
639 unsigned int sd_log_tail
;
642 struct rw_semaphore sd_log_flush_lock
;
643 atomic_t sd_log_in_flight
;
644 wait_queue_head_t sd_log_flush_wait
;
646 unsigned int sd_log_flush_head
;
647 u64 sd_log_flush_wrapped
;
649 struct list_head sd_ail1_list
;
650 struct list_head sd_ail2_list
;
655 struct list_head sd_revoke_list
;
656 unsigned int sd_replay_tail
;
658 unsigned int sd_found_blocks
;
659 unsigned int sd_found_revokes
;
660 unsigned int sd_replayed_blocks
;
662 /* For quiescing the filesystem */
664 struct gfs2_holder sd_freeze_gh
;
665 struct mutex sd_freeze_lock
;
666 unsigned int sd_freeze_count
;
668 char sd_fsname
[GFS2_FSNAME_LEN
];
669 char sd_table_name
[GFS2_FSNAME_LEN
];
670 char sd_proto_name
[GFS2_FSNAME_LEN
];
674 unsigned long sd_last_warning
;
675 struct dentry
*debugfs_dir
; /* debugfs directory */
676 struct dentry
*debugfs_dentry_glocks
; /* for debugfs */
679 #endif /* __INCORE_DOT_H__ */