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>
22 #include <linux/ktime.h>
23 #include <linux/percpu.h>
25 #define DIO_WAIT 0x00000010
26 #define DIO_METADATA 0x00000020
28 struct gfs2_log_operations
;
29 struct gfs2_log_element
;
32 struct gfs2_quota_data
;
39 typedef void (*gfs2_glop_bh_t
) (struct gfs2_glock
*gl
, unsigned int ret
);
41 struct gfs2_log_header_host
{
42 u64 lh_sequence
; /* Sequence number of this transaction */
43 u32 lh_flags
; /* GFS2_LOG_HEAD_... */
44 u32 lh_tail
; /* Block number of log tail */
50 * Structure of operations that are associated with each
51 * type of element in the log.
54 struct gfs2_log_operations
{
55 void (*lo_add
) (struct gfs2_sbd
*sdp
, struct gfs2_log_element
*le
);
56 void (*lo_before_commit
) (struct gfs2_sbd
*sdp
);
57 void (*lo_after_commit
) (struct gfs2_sbd
*sdp
, struct gfs2_ail
*ai
);
58 void (*lo_before_scan
) (struct gfs2_jdesc
*jd
,
59 struct gfs2_log_header_host
*head
, int pass
);
60 int (*lo_scan_elements
) (struct gfs2_jdesc
*jd
, unsigned int start
,
61 struct gfs2_log_descriptor
*ld
, __be64
*ptr
,
63 void (*lo_after_scan
) (struct gfs2_jdesc
*jd
, int error
, int pass
);
67 struct gfs2_log_element
{
68 struct list_head le_list
;
69 const struct gfs2_log_operations
*le_ops
;
75 struct buffer_head
*bi_bh
;
77 unsigned long bi_flags
;
84 struct rb_node rd_node
; /* Link with superblock */
85 struct gfs2_glock
*rd_gl
; /* Glock for this rgrp */
86 u64 rd_addr
; /* grp block disk address */
87 u64 rd_data0
; /* first data location */
88 u32 rd_length
; /* length of rgrp header in fs blocks */
89 u32 rd_data
; /* num of data blocks in rgrp */
90 u32 rd_bitbytes
; /* number of bytes in data bitmaps */
95 struct gfs2_bitmap
*rd_bits
;
96 struct gfs2_sbd
*rd_sbd
;
99 #define GFS2_RDF_CHECK 0x10000000 /* check for unlinked inodes */
100 #define GFS2_RDF_UPTODATE 0x20000000 /* rg is up to date */
101 #define GFS2_RDF_ERROR 0x40000000 /* error in rg */
102 #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */
105 enum gfs2_state_bits
{
106 BH_Pinned
= BH_PrivateStart
,
107 BH_Escaped
= BH_PrivateStart
+ 1,
108 BH_Zeronew
= BH_PrivateStart
+ 2,
111 BUFFER_FNS(Pinned
, pinned
)
112 TAS_BUFFER_FNS(Pinned
, pinned
)
113 BUFFER_FNS(Escaped
, escaped
)
114 TAS_BUFFER_FNS(Escaped
, escaped
)
115 BUFFER_FNS(Zeronew
, zeronew
)
116 TAS_BUFFER_FNS(Zeronew
, zeronew
)
118 struct gfs2_bufdata
{
119 struct buffer_head
*bd_bh
;
120 struct gfs2_glock
*bd_gl
;
123 struct list_head list_tr
;
126 #define bd_list_tr u.list_tr
127 #define bd_blkno u.blkno
129 struct gfs2_log_element bd_le
;
131 struct gfs2_ail
*bd_ail
;
132 struct list_head bd_ail_st_list
;
133 struct list_head bd_ail_gl_list
;
137 * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
138 * prefix of lock_dlm_ gets awkward.
141 #define GDLM_STRNAME_BYTES 25
142 #define GDLM_LVB_SIZE 32
147 * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been
148 * held by failed nodes whose journals need recovery. Those locks should
149 * only be used for journal recovery until the journal recovery is done.
150 * This is set by the dlm recover_prep callback and cleared by the
151 * gfs2_control thread when journal recovery is complete. To avoid
152 * races between recover_prep setting and gfs2_control clearing, recover_spin
153 * is held while changing this bit and reading/writing recover_block
156 * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used.
158 * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing
159 * recovery of all journals before allowing other nodes to mount the fs.
160 * This is cleared when FIRST_MOUNT_DONE is set.
162 * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished
163 * recovery of all journals, and now allows other nodes to mount the fs.
165 * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared
166 * BLOCK_LOCKS for the first time. The gfs2_control thread should now
167 * control clearing BLOCK_LOCKS for further recoveries.
169 * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq.
171 * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep()
172 * and recover_done(), i.e. set while recover_block == recover_start.
179 DFL_FIRST_MOUNT_DONE
= 3,
182 DFL_DLM_RECOVERY
= 6,
187 unsigned int ln_type
;
190 #define lm_name_equal(name1, name2) \
191 (((name1)->ln_number == (name2)->ln_number) && \
192 ((name1)->ln_type == (name2)->ln_type))
195 struct gfs2_glock_operations
{
196 void (*go_xmote_th
) (struct gfs2_glock
*gl
);
197 int (*go_xmote_bh
) (struct gfs2_glock
*gl
, struct gfs2_holder
*gh
);
198 void (*go_inval
) (struct gfs2_glock
*gl
, int flags
);
199 int (*go_demote_ok
) (const struct gfs2_glock
*gl
);
200 int (*go_lock
) (struct gfs2_holder
*gh
);
201 void (*go_unlock
) (struct gfs2_holder
*gh
);
202 int (*go_dump
)(struct seq_file
*seq
, const struct gfs2_glock
*gl
);
203 void (*go_callback
) (struct gfs2_glock
*gl
);
205 const unsigned long go_flags
;
206 #define GLOF_ASPACE 1
210 GFS2_LKS_SRTT
= 0, /* Non blocking smoothed round trip time */
211 GFS2_LKS_SRTTVAR
= 1, /* Non blocking smoothed variance */
212 GFS2_LKS_SRTTB
= 2, /* Blocking smoothed round trip time */
213 GFS2_LKS_SRTTVARB
= 3, /* Blocking smoothed variance */
214 GFS2_LKS_SIRT
= 4, /* Smoothed Inter-request time */
215 GFS2_LKS_SIRTVAR
= 5, /* Smoothed Inter-request variance */
216 GFS2_LKS_DCOUNT
= 6, /* Count of dlm requests */
217 GFS2_LKS_QCOUNT
= 7, /* Count of gfs2_holder queues */
221 struct gfs2_lkstats
{
222 s64 stats
[GFS2_NR_LKSTATS
];
227 HIF_HOLDER
= 6, /* Set for gh that "holds" the glock */
233 struct list_head gh_list
;
235 struct gfs2_glock
*gh_gl
;
236 struct pid
*gh_owner_pid
;
237 unsigned int gh_state
;
241 unsigned long gh_iflags
; /* HIF_... */
248 GLF_PENDING_DEMOTE
= 4,
249 GLF_DEMOTE_IN_PROGRESS
= 5,
252 GLF_INVALIDATE_IN_PROGRESS
= 8,
253 GLF_REPLY_PENDING
= 9,
258 GLF_OBJECT
= 14, /* Used only for tracing */
263 struct hlist_bl_node gl_list
;
264 struct gfs2_sbd
*gl_sbd
;
265 unsigned long gl_flags
; /* GLF_... */
266 struct lm_lockname gl_name
;
271 /* State fields protected by gl_spin */
272 unsigned int gl_state
:2, /* Current state */
273 gl_target
:2, /* Target state */
274 gl_demote_state
:2, /* State requested by remote node */
275 gl_req
:2, /* State in last dlm request */
276 gl_reply
:8; /* Last reply from the dlm */
278 unsigned int gl_hash
;
279 unsigned long gl_demote_time
; /* time of first demote request */
281 struct list_head gl_holders
;
283 const struct gfs2_glock_operations
*gl_ops
;
285 struct gfs2_lkstats gl_stats
;
286 struct dlm_lksb gl_lksb
;
288 unsigned long gl_tchange
;
291 struct list_head gl_lru
;
292 struct list_head gl_ail_list
;
293 atomic_t gl_ail_count
;
295 struct delayed_work gl_work
;
296 struct work_struct gl_delete
;
297 struct rcu_head gl_rcu
;
300 #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */
302 struct gfs2_qadata
{ /* quota allocation data */
304 struct gfs2_quota_data
*qa_qd
[2*MAXQUOTAS
];
305 struct gfs2_holder qa_qd_ghs
[2*MAXQUOTAS
];
306 unsigned int qa_qd_num
;
309 struct gfs2_blkreserv
{
310 u32 rs_requested
; /* Filled in by caller of gfs2_inplace_reserve() */
311 struct gfs2_holder rs_rgd_gh
; /* Filled in by gfs2_inplace_reserve() */
317 GIF_ALLOC_FAILED
= 2,
323 struct inode i_inode
;
328 unsigned long i_flags
; /* GIF_... */
329 struct gfs2_glock
*i_gl
; /* Move into i_gh? */
330 struct gfs2_holder i_iopen_gh
;
331 struct gfs2_holder i_gh
; /* for prepare/commit_write only */
332 struct gfs2_qadata
*i_qadata
; /* quota allocation data */
333 struct gfs2_blkreserv
*i_res
; /* resource group block reservation */
334 struct gfs2_rgrpd
*i_rgd
;
335 u64 i_goal
; /* goal block for allocations */
336 struct rw_semaphore i_rw_mutex
;
337 struct list_head i_trunc_list
;
338 __be64
*i_hash_cache
;
346 * Since i_inode is the first element of struct gfs2_inode,
347 * this is effectively a cast.
349 static inline struct gfs2_inode
*GFS2_I(struct inode
*inode
)
351 return container_of(inode
, struct gfs2_inode
, i_inode
);
354 static inline struct gfs2_sbd
*GFS2_SB(const struct inode
*inode
)
356 return inode
->i_sb
->s_fs_info
;
360 struct mutex f_fl_mutex
;
361 struct gfs2_holder f_fl_gh
;
364 struct gfs2_revoke_replay
{
365 struct list_head rr_list
;
367 unsigned int rr_where
;
377 struct gfs2_quota_data
{
378 struct list_head qd_list
;
379 struct list_head qd_reclaim
;
384 unsigned long qd_flags
; /* QDF_... */
389 unsigned int qd_slot
;
390 unsigned int qd_slot_count
;
392 struct buffer_head
*qd_bh
;
393 struct gfs2_quota_change
*qd_bh_qc
;
394 unsigned int qd_bh_count
;
396 struct gfs2_glock
*qd_gl
;
397 struct gfs2_quota_lvb qd_qb
;
400 unsigned long qd_last_warn
;
406 unsigned int tr_blocks
;
407 unsigned int tr_revokes
;
408 unsigned int tr_reserved
;
410 struct gfs2_holder tr_t_gh
;
414 unsigned int tr_num_buf
;
415 unsigned int tr_num_buf_new
;
416 unsigned int tr_num_databuf_new
;
417 unsigned int tr_num_buf_rm
;
418 unsigned int tr_num_databuf_rm
;
419 struct list_head tr_list_buf
;
421 unsigned int tr_num_revoke
;
422 unsigned int tr_num_revoke_rm
;
426 struct list_head ai_list
;
428 unsigned int ai_first
;
429 struct list_head ai_ail1_list
;
430 struct list_head ai_ail2_list
;
433 struct gfs2_journal_extent
{
434 struct list_head extent_list
;
436 unsigned int lblock
; /* First logical block */
437 u64 dblock
; /* First disk block */
442 struct list_head jd_list
;
443 struct list_head extent_list
;
444 struct work_struct jd_work
;
445 struct inode
*jd_inode
;
446 unsigned long jd_flags
;
447 #define JDF_RECOVERY 1
449 unsigned int jd_blocks
;
450 int jd_recover_error
;
453 struct gfs2_statfs_change_host
{
459 #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF
460 #define GFS2_QUOTA_OFF 0
461 #define GFS2_QUOTA_ACCOUNT 1
462 #define GFS2_QUOTA_ON 2
464 #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED
465 #define GFS2_DATA_WRITEBACK 1
466 #define GFS2_DATA_ORDERED 2
468 #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW
469 #define GFS2_ERRORS_WITHDRAW 0
470 #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */
471 #define GFS2_ERRORS_RO 2 /* place holder for future feature */
472 #define GFS2_ERRORS_PANIC 3
475 char ar_lockproto
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Protocol */
476 char ar_locktable
[GFS2_LOCKNAME_LEN
]; /* Name of the Lock Table */
477 char ar_hostdata
[GFS2_LOCKNAME_LEN
]; /* Host specific data */
478 unsigned int ar_spectator
:1; /* Don't get a journal */
479 unsigned int ar_localflocks
:1; /* Let the VFS do flock|fcntl */
480 unsigned int ar_debug
:1; /* Oops on errors */
481 unsigned int ar_posix_acl
:1; /* Enable posix acls */
482 unsigned int ar_quota
:2; /* off/account/on */
483 unsigned int ar_suiddir
:1; /* suiddir support */
484 unsigned int ar_data
:2; /* ordered/writeback */
485 unsigned int ar_meta
:1; /* mount metafs */
486 unsigned int ar_discard
:1; /* discard requests */
487 unsigned int ar_errors
:2; /* errors=withdraw | panic */
488 unsigned int ar_nobarrier
:1; /* do not send barriers */
489 int ar_commit
; /* Commit interval */
490 int ar_statfs_quantum
; /* The fast statfs interval */
491 int ar_quota_quantum
; /* The quota interval */
492 int ar_statfs_percent
; /* The % change to force sync */
498 unsigned int gt_logd_secs
;
500 unsigned int gt_quota_simul_sync
; /* Max quotavals to sync at once */
501 unsigned int gt_quota_warn_period
; /* Secs between quota warn msgs */
502 unsigned int gt_quota_scale_num
; /* Numerator */
503 unsigned int gt_quota_scale_den
; /* Denominator */
504 unsigned int gt_quota_quantum
; /* Secs between syncs to quota file */
505 unsigned int gt_new_files_jdata
;
506 unsigned int gt_max_readahead
; /* Max bytes to read-ahead from disk */
507 unsigned int gt_complain_secs
;
508 unsigned int gt_statfs_quantum
;
509 unsigned int gt_statfs_slow
;
513 SDF_JOURNAL_CHECKED
= 0,
514 SDF_JOURNAL_LIVE
= 1,
520 SDF_RORECOVERY
= 7, /* read only recovery */
523 #define GFS2_FSNAME_LEN 256
525 struct gfs2_inum_host
{
530 struct gfs2_sb_host
{
536 u32 sb_multihost_format
;
540 struct gfs2_inum_host sb_master_dir
;
541 struct gfs2_inum_host sb_root_dir
;
543 char sb_lockproto
[GFS2_LOCKNAME_LEN
];
544 char sb_locktable
[GFS2_LOCKNAME_LEN
];
548 * lm_mount() return values
550 * ls_jid - the journal ID this node should use
551 * ls_first - this node is the first to mount the file system
552 * ls_lockspace - lock module's context for this file system
553 * ls_ops - lock module's functions
556 struct lm_lockstruct
{
558 unsigned int ls_first
;
559 const struct lm_lockops
*ls_ops
;
560 dlm_lockspace_t
*ls_dlm
;
562 int ls_recover_jid_done
; /* These two are deprecated, */
563 int ls_recover_jid_status
; /* used previously by gfs_controld */
565 struct dlm_lksb ls_mounted_lksb
; /* mounted_lock */
566 struct dlm_lksb ls_control_lksb
; /* control_lock */
567 char ls_control_lvb
[GDLM_LVB_SIZE
]; /* control_lock lvb */
568 struct completion ls_sync_wait
; /* {control,mounted}_{lock,unlock} */
570 spinlock_t ls_recover_spin
; /* protects following fields */
571 unsigned long ls_recover_flags
; /* DFL_ */
572 uint32_t ls_recover_mount
; /* gen in first recover_done cb */
573 uint32_t ls_recover_start
; /* gen in last recover_done cb */
574 uint32_t ls_recover_block
; /* copy recover_start in last recover_prep */
575 uint32_t ls_recover_size
; /* size of recover_submit, recover_result */
576 uint32_t *ls_recover_submit
; /* gen in last recover_slot cb per jid */
577 uint32_t *ls_recover_result
; /* result of last jid recovery */
580 struct gfs2_pcpu_lkstats
{
581 /* One struct for each glock type */
582 struct gfs2_lkstats lkstats
[10];
586 struct super_block
*sd_vfs
;
587 struct gfs2_pcpu_lkstats __percpu
*sd_lkstats
;
588 struct kobject sd_kobj
;
589 unsigned long sd_flags
; /* SDF_... */
590 struct gfs2_sb_host sd_sb
;
592 /* Constants computed on mount */
596 u32 sd_diptrs
; /* Number of pointers in a dinode */
597 u32 sd_inptrs
; /* Number of pointers in a indirect block */
598 u32 sd_jbsize
; /* Size of a journaled data block */
599 u32 sd_hash_bsize
; /* sizeof(exhash block) */
600 u32 sd_hash_bsize_shift
;
601 u32 sd_hash_ptrs
; /* Number of pointers in a hash block */
603 u32 sd_max_dirres
; /* Max blocks needed to add a directory entry */
604 u32 sd_max_height
; /* Max height of a file's metadata tree */
605 u64 sd_heightsize
[GFS2_MAX_META_HEIGHT
+ 1];
606 u32 sd_max_jheight
; /* Max height of journaled file's meta tree */
607 u64 sd_jheightsize
[GFS2_MAX_META_HEIGHT
+ 1];
609 struct gfs2_args sd_args
; /* Mount arguments */
610 struct gfs2_tune sd_tune
; /* Filesystem tuning structure */
614 struct lm_lockstruct sd_lockstruct
;
615 struct gfs2_holder sd_live_gh
;
616 struct gfs2_glock
*sd_rename_gl
;
617 struct gfs2_glock
*sd_trans_gl
;
618 wait_queue_head_t sd_glock_wait
;
619 atomic_t sd_glock_disposal
;
620 struct completion sd_locking_init
;
621 struct delayed_work sd_control_work
;
625 struct dentry
*sd_master_dir
;
626 struct dentry
*sd_root_dir
;
628 struct inode
*sd_jindex
;
629 struct inode
*sd_statfs_inode
;
630 struct inode
*sd_sc_inode
;
631 struct inode
*sd_qc_inode
;
632 struct inode
*sd_rindex
;
633 struct inode
*sd_quota_inode
;
637 spinlock_t sd_statfs_spin
;
638 struct gfs2_statfs_change_host sd_statfs_master
;
639 struct gfs2_statfs_change_host sd_statfs_local
;
640 int sd_statfs_force_sync
;
642 /* Resource group stuff */
644 int sd_rindex_uptodate
;
645 spinlock_t sd_rindex_spin
;
646 struct rb_root sd_rindex_tree
;
647 unsigned int sd_rgrps
;
648 unsigned int sd_max_rg_data
;
650 /* Journal index stuff */
652 struct list_head sd_jindex_list
;
653 spinlock_t sd_jindex_spin
;
654 struct mutex sd_jindex_mutex
;
655 unsigned int sd_journals
;
657 struct gfs2_jdesc
*sd_jdesc
;
658 struct gfs2_holder sd_journal_gh
;
659 struct gfs2_holder sd_jinode_gh
;
661 struct gfs2_holder sd_sc_gh
;
662 struct gfs2_holder sd_qc_gh
;
666 struct task_struct
*sd_logd_process
;
667 struct task_struct
*sd_quotad_process
;
671 struct list_head sd_quota_list
;
672 atomic_t sd_quota_count
;
673 struct mutex sd_quota_mutex
;
674 wait_queue_head_t sd_quota_wait
;
675 struct list_head sd_trunc_list
;
676 spinlock_t sd_trunc_lock
;
678 unsigned int sd_quota_slots
;
679 unsigned int sd_quota_chunks
;
680 unsigned char **sd_quota_bitmap
;
682 u64 sd_quota_sync_gen
;
686 spinlock_t sd_log_lock
;
688 unsigned int sd_log_blks_reserved
;
689 unsigned int sd_log_commited_buf
;
690 unsigned int sd_log_commited_databuf
;
691 int sd_log_commited_revoke
;
693 atomic_t sd_log_pinned
;
694 unsigned int sd_log_num_buf
;
695 unsigned int sd_log_num_revoke
;
696 unsigned int sd_log_num_rg
;
697 unsigned int sd_log_num_databuf
;
699 struct list_head sd_log_le_buf
;
700 struct list_head sd_log_le_revoke
;
701 struct list_head sd_log_le_rg
;
702 struct list_head sd_log_le_databuf
;
703 struct list_head sd_log_le_ordered
;
705 atomic_t sd_log_thresh1
;
706 atomic_t sd_log_thresh2
;
707 atomic_t sd_log_blks_free
;
708 wait_queue_head_t sd_log_waitq
;
709 wait_queue_head_t sd_logd_waitq
;
712 unsigned int sd_log_head
;
713 unsigned int sd_log_tail
;
716 struct rw_semaphore sd_log_flush_lock
;
717 atomic_t sd_log_in_flight
;
718 wait_queue_head_t sd_log_flush_wait
;
720 unsigned int sd_log_flush_head
;
721 u64 sd_log_flush_wrapped
;
723 spinlock_t sd_ail_lock
;
724 struct list_head sd_ail1_list
;
725 struct list_head sd_ail2_list
;
729 struct list_head sd_revoke_list
;
730 unsigned int sd_replay_tail
;
732 unsigned int sd_found_blocks
;
733 unsigned int sd_found_revokes
;
734 unsigned int sd_replayed_blocks
;
736 /* For quiescing the filesystem */
738 struct gfs2_holder sd_freeze_gh
;
739 struct mutex sd_freeze_lock
;
740 unsigned int sd_freeze_count
;
742 char sd_fsname
[GFS2_FSNAME_LEN
];
743 char sd_table_name
[GFS2_FSNAME_LEN
];
744 char sd_proto_name
[GFS2_FSNAME_LEN
];
748 unsigned long sd_last_warning
;
749 struct dentry
*debugfs_dir
; /* debugfs directory */
750 struct dentry
*debugfs_dentry_glocks
;
751 struct dentry
*debugfs_dentry_glstats
;
752 struct dentry
*debugfs_dentry_sbstats
;
755 static inline void gfs2_glstats_inc(struct gfs2_glock
*gl
, int which
)
757 gl
->gl_stats
.stats
[which
]++;
760 static inline void gfs2_sbstats_inc(const struct gfs2_glock
*gl
, int which
)
762 const struct gfs2_sbd
*sdp
= gl
->gl_sbd
;
764 this_cpu_ptr(sdp
->sd_lkstats
)->lkstats
[gl
->gl_name
.ln_type
].stats
[which
]++;
768 #endif /* __INCORE_DOT_H__ */