xref: /openbmc/linux/fs/gfs2/incore.h (revision ae213c44)
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 #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>
24 #include <linux/lockref.h>
25 #include <linux/rhashtable.h>
26 
27 #define DIO_WAIT	0x00000010
28 #define DIO_METADATA	0x00000020
29 
30 struct gfs2_log_operations;
31 struct gfs2_bufdata;
32 struct gfs2_holder;
33 struct gfs2_glock;
34 struct gfs2_quota_data;
35 struct gfs2_trans;
36 struct gfs2_jdesc;
37 struct gfs2_sbd;
38 struct lm_lockops;
39 
40 typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret);
41 
42 struct gfs2_log_header_host {
43 	u64 lh_sequence;	/* Sequence number of this transaction */
44 	u32 lh_flags;		/* GFS2_LOG_HEAD_... */
45 	u32 lh_tail;		/* Block number of log tail */
46 	u32 lh_blkno;
47 };
48 
49 /*
50  * Structure of operations that are associated with each
51  * type of element in the log.
52  */
53 
54 struct gfs2_log_operations {
55 	void (*lo_before_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
56 	void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
57 	void (*lo_before_scan) (struct gfs2_jdesc *jd,
58 				struct gfs2_log_header_host *head, int pass);
59 	int (*lo_scan_elements) (struct gfs2_jdesc *jd, unsigned int start,
60 				 struct gfs2_log_descriptor *ld, __be64 *ptr,
61 				 int pass);
62 	void (*lo_after_scan) (struct gfs2_jdesc *jd, int error, int pass);
63 	const char *lo_name;
64 };
65 
66 #define GBF_FULL 1
67 
68 /**
69  * Clone bitmaps (bi_clone):
70  *
71  * - When a block is freed, we remember the previous state of the block in the
72  *   clone bitmap, and only mark the block as free in the real bitmap.
73  *
74  * - When looking for a block to allocate, we check for a free block in the
75  *   clone bitmap, and if no clone bitmap exists, in the real bitmap.
76  *
77  * - For allocating a block, we mark it as allocated in the real bitmap, and if
78  *   a clone bitmap exists, also in the clone bitmap.
79  *
80  * - At the end of a log_flush, we copy the real bitmap into the clone bitmap
81  *   to make the clone bitmap reflect the current allocation state.
82  *   (Alternatively, we could remove the clone bitmap.)
83  *
84  * The clone bitmaps are in-core only, and is never written to disk.
85  *
86  * These steps ensure that blocks which have been freed in a transaction cannot
87  * be reallocated in that same transaction.
88  */
89 struct gfs2_bitmap {
90 	struct buffer_head *bi_bh;
91 	char *bi_clone;
92 	unsigned long bi_flags;
93 	u32 bi_offset;
94 	u32 bi_start;
95 	u32 bi_bytes;
96 	u32 bi_blocks;
97 };
98 
99 struct gfs2_rgrpd {
100 	struct rb_node rd_node;		/* Link with superblock */
101 	struct gfs2_glock *rd_gl;	/* Glock for this rgrp */
102 	u64 rd_addr;			/* grp block disk address */
103 	u64 rd_data0;			/* first data location */
104 	u32 rd_length;			/* length of rgrp header in fs blocks */
105 	u32 rd_data;			/* num of data blocks in rgrp */
106 	u32 rd_bitbytes;		/* number of bytes in data bitmaps */
107 	u32 rd_free;
108 	u32 rd_reserved;                /* number of blocks reserved */
109 	u32 rd_free_clone;
110 	u32 rd_dinodes;
111 	u64 rd_igeneration;
112 	struct gfs2_bitmap *rd_bits;
113 	struct gfs2_sbd *rd_sbd;
114 	struct gfs2_rgrp_lvb *rd_rgl;
115 	u32 rd_last_alloc;
116 	u32 rd_flags;
117 	u32 rd_extfail_pt;		/* extent failure point */
118 #define GFS2_RDF_CHECK		0x10000000 /* check for unlinked inodes */
119 #define GFS2_RDF_UPTODATE	0x20000000 /* rg is up to date */
120 #define GFS2_RDF_ERROR		0x40000000 /* error in rg */
121 #define GFS2_RDF_PREFERRED	0x80000000 /* This rgrp is preferred */
122 #define GFS2_RDF_MASK		0xf0000000 /* mask for internal flags */
123 	spinlock_t rd_rsspin;           /* protects reservation related vars */
124 	struct rb_root rd_rstree;       /* multi-block reservation tree */
125 };
126 
127 struct gfs2_rbm {
128 	struct gfs2_rgrpd *rgd;
129 	u32 offset;		/* The offset is bitmap relative */
130 	int bii;		/* Bitmap index */
131 };
132 
133 static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
134 {
135 	return rbm->rgd->rd_bits + rbm->bii;
136 }
137 
138 static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm)
139 {
140 	BUG_ON(rbm->offset >= rbm->rgd->rd_data);
141 	return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) +
142 		rbm->offset;
143 }
144 
145 static inline bool gfs2_rbm_eq(const struct gfs2_rbm *rbm1,
146 			       const struct gfs2_rbm *rbm2)
147 {
148 	return (rbm1->rgd == rbm2->rgd) && (rbm1->bii == rbm2->bii) &&
149 	       (rbm1->offset == rbm2->offset);
150 }
151 
152 enum gfs2_state_bits {
153 	BH_Pinned = BH_PrivateStart,
154 	BH_Escaped = BH_PrivateStart + 1,
155 };
156 
157 BUFFER_FNS(Pinned, pinned)
158 TAS_BUFFER_FNS(Pinned, pinned)
159 BUFFER_FNS(Escaped, escaped)
160 TAS_BUFFER_FNS(Escaped, escaped)
161 
162 struct gfs2_bufdata {
163 	struct buffer_head *bd_bh;
164 	struct gfs2_glock *bd_gl;
165 	u64 bd_blkno;
166 
167 	struct list_head bd_list;
168 
169 	struct gfs2_trans *bd_tr;
170 	struct list_head bd_ail_st_list;
171 	struct list_head bd_ail_gl_list;
172 };
173 
174 /*
175  * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
176  * prefix of lock_dlm_ gets awkward.
177  */
178 
179 #define GDLM_STRNAME_BYTES	25
180 #define GDLM_LVB_SIZE		32
181 
182 /*
183  * ls_recover_flags:
184  *
185  * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been
186  * held by failed nodes whose journals need recovery.  Those locks should
187  * only be used for journal recovery until the journal recovery is done.
188  * This is set by the dlm recover_prep callback and cleared by the
189  * gfs2_control thread when journal recovery is complete.  To avoid
190  * races between recover_prep setting and gfs2_control clearing, recover_spin
191  * is held while changing this bit and reading/writing recover_block
192  * and recover_start.
193  *
194  * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used.
195  *
196  * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing
197  * recovery of all journals before allowing other nodes to mount the fs.
198  * This is cleared when FIRST_MOUNT_DONE is set.
199  *
200  * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished
201  * recovery of all journals, and now allows other nodes to mount the fs.
202  *
203  * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared
204  * BLOCK_LOCKS for the first time.  The gfs2_control thread should now
205  * control clearing BLOCK_LOCKS for further recoveries.
206  *
207  * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq.
208  *
209  * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep()
210  * and recover_done(), i.e. set while recover_block == recover_start.
211  */
212 
213 enum {
214 	DFL_BLOCK_LOCKS		= 0,
215 	DFL_NO_DLM_OPS		= 1,
216 	DFL_FIRST_MOUNT		= 2,
217 	DFL_FIRST_MOUNT_DONE	= 3,
218 	DFL_MOUNT_DONE		= 4,
219 	DFL_UNMOUNT		= 5,
220 	DFL_DLM_RECOVERY	= 6,
221 };
222 
223 /*
224  * We are using struct lm_lockname as an rhashtable key.  Avoid holes within
225  * the struct; padding at the end is fine.
226  */
227 struct lm_lockname {
228 	u64 ln_number;
229 	struct gfs2_sbd *ln_sbd;
230 	unsigned int ln_type;
231 };
232 
233 #define lm_name_equal(name1, name2) \
234         (((name1)->ln_number == (name2)->ln_number) &&	\
235 	 ((name1)->ln_type == (name2)->ln_type) &&	\
236 	 ((name1)->ln_sbd == (name2)->ln_sbd))
237 
238 
239 struct gfs2_glock_operations {
240 	void (*go_sync) (struct gfs2_glock *gl);
241 	int (*go_xmote_bh) (struct gfs2_glock *gl, struct gfs2_holder *gh);
242 	void (*go_inval) (struct gfs2_glock *gl, int flags);
243 	int (*go_demote_ok) (const struct gfs2_glock *gl);
244 	int (*go_lock) (struct gfs2_holder *gh);
245 	void (*go_unlock) (struct gfs2_holder *gh);
246 	void (*go_dump)(struct seq_file *seq, struct gfs2_glock *gl);
247 	void (*go_callback)(struct gfs2_glock *gl, bool remote);
248 	const int go_type;
249 	const unsigned long go_flags;
250 #define GLOF_ASPACE 1
251 #define GLOF_LVB    2
252 #define GLOF_LRU    4
253 };
254 
255 enum {
256 	GFS2_LKS_SRTT = 0,	/* Non blocking smoothed round trip time */
257 	GFS2_LKS_SRTTVAR = 1,	/* Non blocking smoothed variance */
258 	GFS2_LKS_SRTTB = 2,	/* Blocking smoothed round trip time */
259 	GFS2_LKS_SRTTVARB = 3,	/* Blocking smoothed variance */
260 	GFS2_LKS_SIRT = 4,	/* Smoothed Inter-request time */
261 	GFS2_LKS_SIRTVAR = 5,	/* Smoothed Inter-request variance */
262 	GFS2_LKS_DCOUNT = 6,	/* Count of dlm requests */
263 	GFS2_LKS_QCOUNT = 7,	/* Count of gfs2_holder queues */
264 	GFS2_NR_LKSTATS
265 };
266 
267 struct gfs2_lkstats {
268 	u64 stats[GFS2_NR_LKSTATS];
269 };
270 
271 enum {
272 	/* States */
273 	HIF_HOLDER		= 6,  /* Set for gh that "holds" the glock */
274 	HIF_FIRST		= 7,
275 	HIF_WAIT		= 10,
276 };
277 
278 struct gfs2_holder {
279 	struct list_head gh_list;
280 
281 	struct gfs2_glock *gh_gl;
282 	struct pid *gh_owner_pid;
283 	u16 gh_flags;
284 	u16 gh_state;
285 
286 	int gh_error;
287 	unsigned long gh_iflags; /* HIF_... */
288 	unsigned long gh_ip;
289 };
290 
291 /* Number of quota types we support */
292 #define GFS2_MAXQUOTAS 2
293 
294 struct gfs2_qadata { /* quota allocation data */
295 	/* Quota stuff */
296 	struct gfs2_quota_data *qa_qd[2 * GFS2_MAXQUOTAS];
297 	struct gfs2_holder qa_qd_ghs[2 * GFS2_MAXQUOTAS];
298 	unsigned int qa_qd_num;
299 };
300 
301 /* Resource group multi-block reservation, in order of appearance:
302 
303    Step 1. Function prepares to write, allocates a mb, sets the size hint.
304    Step 2. User calls inplace_reserve to target an rgrp, sets the rgrp info
305    Step 3. Function get_local_rgrp locks the rgrp, determines which bits to use
306    Step 4. Bits are assigned from the rgrp based on either the reservation
307            or wherever it can.
308 */
309 
310 struct gfs2_blkreserv {
311 	struct rb_node rs_node;       /* link to other block reservations */
312 	struct gfs2_rbm rs_rbm;       /* Start of reservation */
313 	u32 rs_free;                  /* how many blocks are still free */
314 };
315 
316 /*
317  * Allocation parameters
318  * @target: The number of blocks we'd ideally like to allocate
319  * @aflags: The flags (e.g. Orlov flag)
320  *
321  * The intent is to gradually expand this structure over time in
322  * order to give more information, e.g. alignment, min extent size
323  * to the allocation code.
324  */
325 struct gfs2_alloc_parms {
326 	u64 target;
327 	u32 min_target;
328 	u32 aflags;
329 	u64 allowed;
330 };
331 
332 enum {
333 	GLF_LOCK			= 1,
334 	GLF_DEMOTE			= 3,
335 	GLF_PENDING_DEMOTE		= 4,
336 	GLF_DEMOTE_IN_PROGRESS		= 5,
337 	GLF_DIRTY			= 6,
338 	GLF_LFLUSH			= 7,
339 	GLF_INVALIDATE_IN_PROGRESS	= 8,
340 	GLF_REPLY_PENDING		= 9,
341 	GLF_INITIAL			= 10,
342 	GLF_FROZEN			= 11,
343 	GLF_QUEUED			= 12,
344 	GLF_LRU				= 13,
345 	GLF_OBJECT			= 14, /* Used only for tracing */
346 	GLF_BLOCKING			= 15,
347 	GLF_INODE_CREATING		= 16, /* Inode creation occurring */
348 	GLF_REVOKES			= 17, /* Glock has revokes in queue */
349 };
350 
351 struct gfs2_glock {
352 	unsigned long gl_flags;		/* GLF_... */
353 	struct lm_lockname gl_name;
354 
355 	struct lockref gl_lockref;
356 
357 	/* State fields protected by gl_lockref.lock */
358 	unsigned int gl_state:2,	/* Current state */
359 		     gl_target:2,	/* Target state */
360 		     gl_demote_state:2,	/* State requested by remote node */
361 		     gl_req:2,		/* State in last dlm request */
362 		     gl_reply:8;	/* Last reply from the dlm */
363 
364 	unsigned long gl_demote_time; /* time of first demote request */
365 	long gl_hold_time;
366 	struct list_head gl_holders;
367 
368 	const struct gfs2_glock_operations *gl_ops;
369 	ktime_t gl_dstamp;
370 	struct gfs2_lkstats gl_stats;
371 	struct dlm_lksb gl_lksb;
372 	unsigned long gl_tchange;
373 	void *gl_object;
374 
375 	struct list_head gl_lru;
376 	struct list_head gl_ail_list;
377 	atomic_t gl_ail_count;
378 	struct delayed_work gl_work;
379 	union {
380 		/* For inode and iopen glocks only */
381 		struct work_struct gl_delete;
382 		/* For rgrp glocks only */
383 		struct {
384 			loff_t start;
385 			loff_t end;
386 		} gl_vm;
387 	};
388 	struct rcu_head gl_rcu;
389 	struct rhash_head gl_node;
390 };
391 
392 #define GFS2_MIN_LVB_SIZE 32	/* Min size of LVB that gfs2 supports */
393 
394 enum {
395 	GIF_INVALID		= 0,
396 	GIF_QD_LOCKED		= 1,
397 	GIF_ALLOC_FAILED	= 2,
398 	GIF_SW_PAGED		= 3,
399 	GIF_ORDERED		= 4,
400 	GIF_FREE_VFS_INODE      = 5,
401 	GIF_GLOP_PENDING	= 6,
402 };
403 
404 struct gfs2_inode {
405 	struct inode i_inode;
406 	u64 i_no_addr;
407 	u64 i_no_formal_ino;
408 	u64 i_generation;
409 	u64 i_eattr;
410 	unsigned long i_flags;		/* GIF_... */
411 	struct gfs2_glock *i_gl; /* Move into i_gh? */
412 	struct gfs2_holder i_iopen_gh;
413 	struct gfs2_holder i_gh; /* for prepare/commit_write only */
414 	struct gfs2_qadata *i_qadata; /* quota allocation data */
415 	struct gfs2_holder i_rgd_gh;
416 	struct gfs2_blkreserv i_res; /* rgrp multi-block reservation */
417 	u64 i_goal;	/* goal block for allocations */
418 	atomic_t i_sizehint;  /* hint of the write size */
419 	struct rw_semaphore i_rw_mutex;
420 	struct list_head i_ordered;
421 	struct list_head i_trunc_list;
422 	__be64 *i_hash_cache;
423 	u32 i_entries;
424 	u32 i_diskflags;
425 	u8 i_height;
426 	u8 i_depth;
427 	u16 i_rahead;
428 };
429 
430 /*
431  * Since i_inode is the first element of struct gfs2_inode,
432  * this is effectively a cast.
433  */
434 static inline struct gfs2_inode *GFS2_I(struct inode *inode)
435 {
436 	return container_of(inode, struct gfs2_inode, i_inode);
437 }
438 
439 static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode)
440 {
441 	return inode->i_sb->s_fs_info;
442 }
443 
444 struct gfs2_file {
445 	struct mutex f_fl_mutex;
446 	struct gfs2_holder f_fl_gh;
447 };
448 
449 struct gfs2_revoke_replay {
450 	struct list_head rr_list;
451 	u64 rr_blkno;
452 	unsigned int rr_where;
453 };
454 
455 enum {
456 	QDF_CHANGE		= 1,
457 	QDF_LOCKED		= 2,
458 	QDF_REFRESH		= 3,
459 	QDF_QMSG_QUIET          = 4,
460 };
461 
462 struct gfs2_quota_data {
463 	struct hlist_bl_node qd_hlist;
464 	struct list_head qd_list;
465 	struct kqid qd_id;
466 	struct gfs2_sbd *qd_sbd;
467 	struct lockref qd_lockref;
468 	struct list_head qd_lru;
469 	unsigned qd_hash;
470 
471 	unsigned long qd_flags;		/* QDF_... */
472 
473 	s64 qd_change;
474 	s64 qd_change_sync;
475 
476 	unsigned int qd_slot;
477 	unsigned int qd_slot_count;
478 
479 	struct buffer_head *qd_bh;
480 	struct gfs2_quota_change *qd_bh_qc;
481 	unsigned int qd_bh_count;
482 
483 	struct gfs2_glock *qd_gl;
484 	struct gfs2_quota_lvb qd_qb;
485 
486 	u64 qd_sync_gen;
487 	unsigned long qd_last_warn;
488 	struct rcu_head qd_rcu;
489 };
490 
491 enum {
492 	TR_TOUCHED = 1,
493 	TR_ATTACHED = 2,
494 	TR_ALLOCED = 3,
495 };
496 
497 struct gfs2_trans {
498 	unsigned long tr_ip;
499 
500 	unsigned int tr_blocks;
501 	unsigned int tr_revokes;
502 	unsigned int tr_reserved;
503 	unsigned long tr_flags;
504 
505 	unsigned int tr_num_buf_new;
506 	unsigned int tr_num_databuf_new;
507 	unsigned int tr_num_buf_rm;
508 	unsigned int tr_num_databuf_rm;
509 	unsigned int tr_num_revoke;
510 	unsigned int tr_num_revoke_rm;
511 
512 	struct list_head tr_list;
513 	struct list_head tr_databuf;
514 	struct list_head tr_buf;
515 
516 	unsigned int tr_first;
517 	struct list_head tr_ail1_list;
518 	struct list_head tr_ail2_list;
519 };
520 
521 struct gfs2_journal_extent {
522 	struct list_head list;
523 
524 	unsigned int lblock; /* First logical block */
525 	u64 dblock; /* First disk block */
526 	u64 blocks;
527 };
528 
529 struct gfs2_jdesc {
530 	struct list_head jd_list;
531 	struct list_head extent_list;
532 	unsigned int nr_extents;
533 	struct work_struct jd_work;
534 	struct inode *jd_inode;
535 	unsigned long jd_flags;
536 #define JDF_RECOVERY 1
537 	unsigned int jd_jid;
538 	u32 jd_blocks;
539 	int jd_recover_error;
540 	/* Replay stuff */
541 
542 	unsigned int jd_found_blocks;
543 	unsigned int jd_found_revokes;
544 	unsigned int jd_replayed_blocks;
545 
546 	struct list_head jd_revoke_list;
547 	unsigned int jd_replay_tail;
548 
549 };
550 
551 struct gfs2_statfs_change_host {
552 	s64 sc_total;
553 	s64 sc_free;
554 	s64 sc_dinodes;
555 };
556 
557 #define GFS2_QUOTA_DEFAULT	GFS2_QUOTA_OFF
558 #define GFS2_QUOTA_OFF		0
559 #define GFS2_QUOTA_ACCOUNT	1
560 #define GFS2_QUOTA_ON		2
561 
562 #define GFS2_DATA_DEFAULT	GFS2_DATA_ORDERED
563 #define GFS2_DATA_WRITEBACK	1
564 #define GFS2_DATA_ORDERED	2
565 
566 #define GFS2_ERRORS_DEFAULT     GFS2_ERRORS_WITHDRAW
567 #define GFS2_ERRORS_WITHDRAW    0
568 #define GFS2_ERRORS_CONTINUE    1 /* place holder for future feature */
569 #define GFS2_ERRORS_RO          2 /* place holder for future feature */
570 #define GFS2_ERRORS_PANIC       3
571 
572 struct gfs2_args {
573 	char ar_lockproto[GFS2_LOCKNAME_LEN];	/* Name of the Lock Protocol */
574 	char ar_locktable[GFS2_LOCKNAME_LEN];	/* Name of the Lock Table */
575 	char ar_hostdata[GFS2_LOCKNAME_LEN];	/* Host specific data */
576 	unsigned int ar_spectator:1;		/* Don't get a journal */
577 	unsigned int ar_localflocks:1;		/* Let the VFS do flock|fcntl */
578 	unsigned int ar_debug:1;		/* Oops on errors */
579 	unsigned int ar_posix_acl:1;		/* Enable posix acls */
580 	unsigned int ar_quota:2;		/* off/account/on */
581 	unsigned int ar_suiddir:1;		/* suiddir support */
582 	unsigned int ar_data:2;			/* ordered/writeback */
583 	unsigned int ar_meta:1;			/* mount metafs */
584 	unsigned int ar_discard:1;		/* discard requests */
585 	unsigned int ar_errors:2;               /* errors=withdraw | panic */
586 	unsigned int ar_nobarrier:1;            /* do not send barriers */
587 	unsigned int ar_rgrplvb:1;		/* use lvbs for rgrp info */
588 	unsigned int ar_loccookie:1;		/* use location based readdir
589 						   cookies */
590 	int ar_commit;				/* Commit interval */
591 	int ar_statfs_quantum;			/* The fast statfs interval */
592 	int ar_quota_quantum;			/* The quota interval */
593 	int ar_statfs_percent;			/* The % change to force sync */
594 };
595 
596 struct gfs2_tune {
597 	spinlock_t gt_spin;
598 
599 	unsigned int gt_logd_secs;
600 
601 	unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */
602 	unsigned int gt_quota_scale_num; /* Numerator */
603 	unsigned int gt_quota_scale_den; /* Denominator */
604 	unsigned int gt_quota_quantum; /* Secs between syncs to quota file */
605 	unsigned int gt_new_files_jdata;
606 	unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */
607 	unsigned int gt_complain_secs;
608 	unsigned int gt_statfs_quantum;
609 	unsigned int gt_statfs_slow;
610 };
611 
612 enum {
613 	SDF_JOURNAL_CHECKED	= 0,
614 	SDF_JOURNAL_LIVE	= 1,
615 	SDF_SHUTDOWN		= 2,
616 	SDF_NOBARRIERS		= 3,
617 	SDF_NORECOVERY		= 4,
618 	SDF_DEMOTE		= 5,
619 	SDF_NOJOURNALID		= 6,
620 	SDF_RORECOVERY		= 7, /* read only recovery */
621 	SDF_SKIP_DLM_UNLOCK	= 8,
622 	SDF_FORCE_AIL_FLUSH     = 9,
623 	SDF_AIL1_IO_ERROR	= 10,
624 	SDF_FS_FROZEN           = 11,
625 };
626 
627 enum gfs2_freeze_state {
628 	SFS_UNFROZEN		= 0,
629 	SFS_STARTING_FREEZE	= 1,
630 	SFS_FROZEN		= 2,
631 };
632 
633 #define GFS2_FSNAME_LEN		256
634 
635 struct gfs2_inum_host {
636 	u64 no_formal_ino;
637 	u64 no_addr;
638 };
639 
640 struct gfs2_sb_host {
641 	u32 sb_magic;
642 	u32 sb_type;
643 	u32 sb_format;
644 
645 	u32 sb_fs_format;
646 	u32 sb_multihost_format;
647 	u32 sb_bsize;
648 	u32 sb_bsize_shift;
649 
650 	struct gfs2_inum_host sb_master_dir;
651 	struct gfs2_inum_host sb_root_dir;
652 
653 	char sb_lockproto[GFS2_LOCKNAME_LEN];
654 	char sb_locktable[GFS2_LOCKNAME_LEN];
655 };
656 
657 /*
658  * lm_mount() return values
659  *
660  * ls_jid - the journal ID this node should use
661  * ls_first - this node is the first to mount the file system
662  * ls_lockspace - lock module's context for this file system
663  * ls_ops - lock module's functions
664  */
665 
666 struct lm_lockstruct {
667 	int ls_jid;
668 	unsigned int ls_first;
669 	const struct lm_lockops *ls_ops;
670 	dlm_lockspace_t *ls_dlm;
671 
672 	int ls_recover_jid_done;   /* These two are deprecated, */
673 	int ls_recover_jid_status; /* used previously by gfs_controld */
674 
675 	struct dlm_lksb ls_mounted_lksb; /* mounted_lock */
676 	struct dlm_lksb ls_control_lksb; /* control_lock */
677 	char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
678 	struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
679 	char *ls_lvb_bits;
680 
681 	spinlock_t ls_recover_spin; /* protects following fields */
682 	unsigned long ls_recover_flags; /* DFL_ */
683 	uint32_t ls_recover_mount; /* gen in first recover_done cb */
684 	uint32_t ls_recover_start; /* gen in last recover_done cb */
685 	uint32_t ls_recover_block; /* copy recover_start in last recover_prep */
686 	uint32_t ls_recover_size; /* size of recover_submit, recover_result */
687 	uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */
688 	uint32_t *ls_recover_result; /* result of last jid recovery */
689 };
690 
691 struct gfs2_pcpu_lkstats {
692 	/* One struct for each glock type */
693 	struct gfs2_lkstats lkstats[10];
694 };
695 
696 struct gfs2_sbd {
697 	struct super_block *sd_vfs;
698 	struct gfs2_pcpu_lkstats __percpu *sd_lkstats;
699 	struct kobject sd_kobj;
700 	unsigned long sd_flags;	/* SDF_... */
701 	struct gfs2_sb_host sd_sb;
702 
703 	/* Constants computed on mount */
704 
705 	u32 sd_fsb2bb;
706 	u32 sd_fsb2bb_shift;
707 	u32 sd_diptrs;	/* Number of pointers in a dinode */
708 	u32 sd_inptrs;	/* Number of pointers in a indirect block */
709 	u32 sd_jbsize;	/* Size of a journaled data block */
710 	u32 sd_hash_bsize;	/* sizeof(exhash block) */
711 	u32 sd_hash_bsize_shift;
712 	u32 sd_hash_ptrs;	/* Number of pointers in a hash block */
713 	u32 sd_qc_per_block;
714 	u32 sd_blocks_per_bitmap;
715 	u32 sd_max_dirres;	/* Max blocks needed to add a directory entry */
716 	u32 sd_max_height;	/* Max height of a file's metadata tree */
717 	u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1];
718 	u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */
719 
720 	struct gfs2_args sd_args;	/* Mount arguments */
721 	struct gfs2_tune sd_tune;	/* Filesystem tuning structure */
722 
723 	/* Lock Stuff */
724 
725 	struct lm_lockstruct sd_lockstruct;
726 	struct gfs2_holder sd_live_gh;
727 	struct gfs2_glock *sd_rename_gl;
728 	struct gfs2_glock *sd_freeze_gl;
729 	struct work_struct sd_freeze_work;
730 	wait_queue_head_t sd_glock_wait;
731 	atomic_t sd_glock_disposal;
732 	struct completion sd_locking_init;
733 	struct completion sd_wdack;
734 	struct delayed_work sd_control_work;
735 
736 	/* Inode Stuff */
737 
738 	struct dentry *sd_master_dir;
739 	struct dentry *sd_root_dir;
740 
741 	struct inode *sd_jindex;
742 	struct inode *sd_statfs_inode;
743 	struct inode *sd_sc_inode;
744 	struct inode *sd_qc_inode;
745 	struct inode *sd_rindex;
746 	struct inode *sd_quota_inode;
747 
748 	/* StatFS stuff */
749 
750 	spinlock_t sd_statfs_spin;
751 	struct gfs2_statfs_change_host sd_statfs_master;
752 	struct gfs2_statfs_change_host sd_statfs_local;
753 	int sd_statfs_force_sync;
754 
755 	/* Resource group stuff */
756 
757 	int sd_rindex_uptodate;
758 	spinlock_t sd_rindex_spin;
759 	struct rb_root sd_rindex_tree;
760 	unsigned int sd_rgrps;
761 	unsigned int sd_max_rg_data;
762 
763 	/* Journal index stuff */
764 
765 	struct list_head sd_jindex_list;
766 	spinlock_t sd_jindex_spin;
767 	struct mutex sd_jindex_mutex;
768 	unsigned int sd_journals;
769 
770 	struct gfs2_jdesc *sd_jdesc;
771 	struct gfs2_holder sd_journal_gh;
772 	struct gfs2_holder sd_jinode_gh;
773 
774 	struct gfs2_holder sd_sc_gh;
775 	struct gfs2_holder sd_qc_gh;
776 
777 	struct completion sd_journal_ready;
778 
779 	/* Daemon stuff */
780 
781 	struct task_struct *sd_logd_process;
782 	struct task_struct *sd_quotad_process;
783 
784 	/* Quota stuff */
785 
786 	struct list_head sd_quota_list;
787 	atomic_t sd_quota_count;
788 	struct mutex sd_quota_mutex;
789 	struct mutex sd_quota_sync_mutex;
790 	wait_queue_head_t sd_quota_wait;
791 	struct list_head sd_trunc_list;
792 	spinlock_t sd_trunc_lock;
793 
794 	unsigned int sd_quota_slots;
795 	unsigned long *sd_quota_bitmap;
796 	spinlock_t sd_bitmap_lock;
797 
798 	u64 sd_quota_sync_gen;
799 
800 	/* Log stuff */
801 
802 	struct address_space sd_aspace;
803 
804 	spinlock_t sd_log_lock;
805 
806 	struct gfs2_trans *sd_log_tr;
807 	unsigned int sd_log_blks_reserved;
808 	int sd_log_commited_revoke;
809 
810 	atomic_t sd_log_pinned;
811 	unsigned int sd_log_num_revoke;
812 
813 	struct list_head sd_log_revokes;
814 	struct list_head sd_log_ordered;
815 	spinlock_t sd_ordered_lock;
816 
817 	atomic_t sd_log_thresh1;
818 	atomic_t sd_log_thresh2;
819 	atomic_t sd_log_blks_free;
820 	atomic_t sd_log_blks_needed;
821 	wait_queue_head_t sd_log_waitq;
822 	wait_queue_head_t sd_logd_waitq;
823 
824 	u64 sd_log_sequence;
825 	unsigned int sd_log_head;
826 	unsigned int sd_log_tail;
827 	int sd_log_idle;
828 
829 	struct rw_semaphore sd_log_flush_lock;
830 	atomic_t sd_log_in_flight;
831 	struct bio *sd_log_bio;
832 	wait_queue_head_t sd_log_flush_wait;
833 	int sd_log_error;
834 
835 	atomic_t sd_reserving_log;
836 	wait_queue_head_t sd_reserving_log_wait;
837 
838 	unsigned int sd_log_flush_head;
839 
840 	spinlock_t sd_ail_lock;
841 	struct list_head sd_ail1_list;
842 	struct list_head sd_ail2_list;
843 
844 	/* For quiescing the filesystem */
845 	struct gfs2_holder sd_freeze_gh;
846 	atomic_t sd_freeze_state;
847 	struct mutex sd_freeze_mutex;
848 
849 	char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2];
850 	char sd_table_name[GFS2_FSNAME_LEN];
851 	char sd_proto_name[GFS2_FSNAME_LEN];
852 
853 	/* Debugging crud */
854 
855 	unsigned long sd_last_warning;
856 	struct dentry *debugfs_dir;    /* debugfs directory */
857 };
858 
859 static inline void gfs2_glstats_inc(struct gfs2_glock *gl, int which)
860 {
861 	gl->gl_stats.stats[which]++;
862 }
863 
864 static inline void gfs2_sbstats_inc(const struct gfs2_glock *gl, int which)
865 {
866 	const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
867 	preempt_disable();
868 	this_cpu_ptr(sdp->sd_lkstats)->lkstats[gl->gl_name.ln_type].stats[which]++;
869 	preempt_enable();
870 }
871 
872 extern struct gfs2_rgrpd *gfs2_glock2rgrp(struct gfs2_glock *gl);
873 
874 static inline unsigned gfs2_max_stuffed_size(const struct gfs2_inode *ip)
875 {
876 	return GFS2_SB(&ip->i_inode)->sd_sb.sb_bsize - sizeof(struct gfs2_dinode);
877 }
878 
879 #endif /* __INCORE_DOT_H__ */
880 
881