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_len;
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 
rbm_bi(const struct gfs2_rbm * rbm)133 static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
134 {
135 	return rbm->rgd->rd_bits + rbm->bii;
136 }
137 
gfs2_rbm_to_block(const struct gfs2_rbm * rbm)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 
gfs2_rbm_eq(const struct gfs2_rbm * rbm1,const struct gfs2_rbm * rbm2)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 	const struct gfs2_log_operations *bd_ops;
169 
170 	struct gfs2_trans *bd_tr;
171 	struct list_head bd_ail_st_list;
172 	struct list_head bd_ail_gl_list;
173 };
174 
175 /*
176  * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
177  * prefix of lock_dlm_ gets awkward.
178  */
179 
180 #define GDLM_STRNAME_BYTES	25
181 #define GDLM_LVB_SIZE		32
182 
183 /*
184  * ls_recover_flags:
185  *
186  * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been
187  * held by failed nodes whose journals need recovery.  Those locks should
188  * only be used for journal recovery until the journal recovery is done.
189  * This is set by the dlm recover_prep callback and cleared by the
190  * gfs2_control thread when journal recovery is complete.  To avoid
191  * races between recover_prep setting and gfs2_control clearing, recover_spin
192  * is held while changing this bit and reading/writing recover_block
193  * and recover_start.
194  *
195  * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used.
196  *
197  * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing
198  * recovery of all journals before allowing other nodes to mount the fs.
199  * This is cleared when FIRST_MOUNT_DONE is set.
200  *
201  * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished
202  * recovery of all journals, and now allows other nodes to mount the fs.
203  *
204  * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared
205  * BLOCK_LOCKS for the first time.  The gfs2_control thread should now
206  * control clearing BLOCK_LOCKS for further recoveries.
207  *
208  * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq.
209  *
210  * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep()
211  * and recover_done(), i.e. set while recover_block == recover_start.
212  */
213 
214 enum {
215 	DFL_BLOCK_LOCKS		= 0,
216 	DFL_NO_DLM_OPS		= 1,
217 	DFL_FIRST_MOUNT		= 2,
218 	DFL_FIRST_MOUNT_DONE	= 3,
219 	DFL_MOUNT_DONE		= 4,
220 	DFL_UNMOUNT		= 5,
221 	DFL_DLM_RECOVERY	= 6,
222 };
223 
224 /*
225  * We are using struct lm_lockname as an rhashtable key.  Avoid holes within
226  * the struct; padding at the end is fine.
227  */
228 struct lm_lockname {
229 	u64 ln_number;
230 	struct gfs2_sbd *ln_sbd;
231 	unsigned int ln_type;
232 };
233 
234 #define lm_name_equal(name1, name2) \
235         (((name1)->ln_number == (name2)->ln_number) &&	\
236 	 ((name1)->ln_type == (name2)->ln_type) &&	\
237 	 ((name1)->ln_sbd == (name2)->ln_sbd))
238 
239 
240 struct gfs2_glock_operations {
241 	void (*go_sync) (struct gfs2_glock *gl);
242 	int (*go_xmote_bh) (struct gfs2_glock *gl, struct gfs2_holder *gh);
243 	void (*go_inval) (struct gfs2_glock *gl, int flags);
244 	int (*go_demote_ok) (const struct gfs2_glock *gl);
245 	int (*go_lock) (struct gfs2_holder *gh);
246 	void (*go_unlock) (struct gfs2_holder *gh);
247 	void (*go_dump)(struct seq_file *seq, const struct gfs2_glock *gl);
248 	void (*go_callback)(struct gfs2_glock *gl, bool remote);
249 	const int go_type;
250 	const unsigned long go_flags;
251 #define GLOF_ASPACE 1
252 #define GLOF_LVB    2
253 #define GLOF_LRU    4
254 };
255 
256 enum {
257 	GFS2_LKS_SRTT = 0,	/* Non blocking smoothed round trip time */
258 	GFS2_LKS_SRTTVAR = 1,	/* Non blocking smoothed variance */
259 	GFS2_LKS_SRTTB = 2,	/* Blocking smoothed round trip time */
260 	GFS2_LKS_SRTTVARB = 3,	/* Blocking smoothed variance */
261 	GFS2_LKS_SIRT = 4,	/* Smoothed Inter-request time */
262 	GFS2_LKS_SIRTVAR = 5,	/* Smoothed Inter-request variance */
263 	GFS2_LKS_DCOUNT = 6,	/* Count of dlm requests */
264 	GFS2_LKS_QCOUNT = 7,	/* Count of gfs2_holder queues */
265 	GFS2_NR_LKSTATS
266 };
267 
268 struct gfs2_lkstats {
269 	u64 stats[GFS2_NR_LKSTATS];
270 };
271 
272 enum {
273 	/* States */
274 	HIF_HOLDER		= 6,  /* Set for gh that "holds" the glock */
275 	HIF_FIRST		= 7,
276 	HIF_WAIT		= 10,
277 };
278 
279 struct gfs2_holder {
280 	struct list_head gh_list;
281 
282 	struct gfs2_glock *gh_gl;
283 	struct pid *gh_owner_pid;
284 	u16 gh_flags;
285 	u16 gh_state;
286 
287 	int gh_error;
288 	unsigned long gh_iflags; /* HIF_... */
289 	unsigned long gh_ip;
290 };
291 
292 /* Number of quota types we support */
293 #define GFS2_MAXQUOTAS 2
294 
295 struct gfs2_qadata { /* quota allocation data */
296 	/* Quota stuff */
297 	struct gfs2_quota_data *qa_qd[2 * GFS2_MAXQUOTAS];
298 	struct gfs2_holder qa_qd_ghs[2 * GFS2_MAXQUOTAS];
299 	unsigned int qa_qd_num;
300 };
301 
302 /* Resource group multi-block reservation, in order of appearance:
303 
304    Step 1. Function prepares to write, allocates a mb, sets the size hint.
305    Step 2. User calls inplace_reserve to target an rgrp, sets the rgrp info
306    Step 3. Function get_local_rgrp locks the rgrp, determines which bits to use
307    Step 4. Bits are assigned from the rgrp based on either the reservation
308            or wherever it can.
309 */
310 
311 struct gfs2_blkreserv {
312 	/* components used during write (step 1): */
313 	atomic_t rs_sizehint;         /* hint of the write size */
314 
315 	struct gfs2_holder rs_rgd_gh; /* Filled in by get_local_rgrp */
316 	struct rb_node rs_node;       /* link to other block reservations */
317 	struct gfs2_rbm rs_rbm;       /* Start of reservation */
318 	u32 rs_free;                  /* how many blocks are still free */
319 };
320 
321 /*
322  * Allocation parameters
323  * @target: The number of blocks we'd ideally like to allocate
324  * @aflags: The flags (e.g. Orlov flag)
325  *
326  * The intent is to gradually expand this structure over time in
327  * order to give more information, e.g. alignment, min extent size
328  * to the allocation code.
329  */
330 struct gfs2_alloc_parms {
331 	u64 target;
332 	u32 min_target;
333 	u32 aflags;
334 	u64 allowed;
335 };
336 
337 enum {
338 	GLF_LOCK			= 1,
339 	GLF_DEMOTE			= 3,
340 	GLF_PENDING_DEMOTE		= 4,
341 	GLF_DEMOTE_IN_PROGRESS		= 5,
342 	GLF_DIRTY			= 6,
343 	GLF_LFLUSH			= 7,
344 	GLF_INVALIDATE_IN_PROGRESS	= 8,
345 	GLF_REPLY_PENDING		= 9,
346 	GLF_INITIAL			= 10,
347 	GLF_FROZEN			= 11,
348 	GLF_QUEUED			= 12,
349 	GLF_LRU				= 13,
350 	GLF_OBJECT			= 14, /* Used only for tracing */
351 	GLF_BLOCKING			= 15,
352 	GLF_INODE_CREATING		= 16, /* Inode creation occurring */
353 };
354 
355 struct gfs2_glock {
356 	unsigned long gl_flags;		/* GLF_... */
357 	struct lm_lockname gl_name;
358 
359 	struct lockref gl_lockref;
360 
361 	/* State fields protected by gl_lockref.lock */
362 	unsigned int gl_state:2,	/* Current state */
363 		     gl_target:2,	/* Target state */
364 		     gl_demote_state:2,	/* State requested by remote node */
365 		     gl_req:2,		/* State in last dlm request */
366 		     gl_reply:8;	/* Last reply from the dlm */
367 
368 	unsigned long gl_demote_time; /* time of first demote request */
369 	long gl_hold_time;
370 	struct list_head gl_holders;
371 
372 	const struct gfs2_glock_operations *gl_ops;
373 	ktime_t gl_dstamp;
374 	struct gfs2_lkstats gl_stats;
375 	struct dlm_lksb gl_lksb;
376 	unsigned long gl_tchange;
377 	void *gl_object;
378 
379 	struct list_head gl_lru;
380 	struct list_head gl_ail_list;
381 	atomic_t gl_ail_count;
382 	atomic_t gl_revokes;
383 	struct delayed_work gl_work;
384 	union {
385 		/* For inode and iopen glocks only */
386 		struct work_struct gl_delete;
387 		/* For rgrp glocks only */
388 		struct {
389 			loff_t start;
390 			loff_t end;
391 		} gl_vm;
392 	};
393 	struct rcu_head gl_rcu;
394 	struct rhash_head gl_node;
395 };
396 
397 #define GFS2_MIN_LVB_SIZE 32	/* Min size of LVB that gfs2 supports */
398 
399 enum {
400 	GIF_INVALID		= 0,
401 	GIF_QD_LOCKED		= 1,
402 	GIF_ALLOC_FAILED	= 2,
403 	GIF_SW_PAGED		= 3,
404 	GIF_ORDERED		= 4,
405 	GIF_FREE_VFS_INODE      = 5,
406 	GIF_GLOP_PENDING	= 6,
407 };
408 
409 struct gfs2_inode {
410 	struct inode i_inode;
411 	u64 i_no_addr;
412 	u64 i_no_formal_ino;
413 	u64 i_generation;
414 	u64 i_eattr;
415 	unsigned long i_flags;		/* GIF_... */
416 	struct gfs2_glock *i_gl; /* Move into i_gh? */
417 	struct gfs2_holder i_iopen_gh;
418 	struct gfs2_holder i_gh; /* for prepare/commit_write only */
419 	struct gfs2_qadata *i_qadata; /* quota allocation data */
420 	struct gfs2_blkreserv i_res; /* rgrp multi-block reservation */
421 	u64 i_goal;	/* goal block for allocations */
422 	struct rw_semaphore i_rw_mutex;
423 	struct list_head i_ordered;
424 	struct list_head i_trunc_list;
425 	__be64 *i_hash_cache;
426 	u32 i_entries;
427 	u32 i_diskflags;
428 	u8 i_height;
429 	u8 i_depth;
430 	u16 i_rahead;
431 };
432 
433 /*
434  * Since i_inode is the first element of struct gfs2_inode,
435  * this is effectively a cast.
436  */
GFS2_I(struct inode * inode)437 static inline struct gfs2_inode *GFS2_I(struct inode *inode)
438 {
439 	return container_of(inode, struct gfs2_inode, i_inode);
440 }
441 
GFS2_SB(const struct inode * inode)442 static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode)
443 {
444 	return inode->i_sb->s_fs_info;
445 }
446 
447 struct gfs2_file {
448 	struct mutex f_fl_mutex;
449 	struct gfs2_holder f_fl_gh;
450 };
451 
452 struct gfs2_revoke_replay {
453 	struct list_head rr_list;
454 	u64 rr_blkno;
455 	unsigned int rr_where;
456 };
457 
458 enum {
459 	QDF_CHANGE		= 1,
460 	QDF_LOCKED		= 2,
461 	QDF_REFRESH		= 3,
462 	QDF_QMSG_QUIET          = 4,
463 };
464 
465 struct gfs2_quota_data {
466 	struct hlist_bl_node qd_hlist;
467 	struct list_head qd_list;
468 	struct kqid qd_id;
469 	struct gfs2_sbd *qd_sbd;
470 	struct lockref qd_lockref;
471 	struct list_head qd_lru;
472 	unsigned qd_hash;
473 
474 	unsigned long qd_flags;		/* QDF_... */
475 
476 	s64 qd_change;
477 	s64 qd_change_sync;
478 
479 	unsigned int qd_slot;
480 	unsigned int qd_slot_count;
481 
482 	struct buffer_head *qd_bh;
483 	struct gfs2_quota_change *qd_bh_qc;
484 	unsigned int qd_bh_count;
485 
486 	struct gfs2_glock *qd_gl;
487 	struct gfs2_quota_lvb qd_qb;
488 
489 	u64 qd_sync_gen;
490 	unsigned long qd_last_warn;
491 	struct rcu_head qd_rcu;
492 };
493 
494 enum {
495 	TR_TOUCHED = 1,
496 	TR_ATTACHED = 2,
497 	TR_ALLOCED = 3,
498 };
499 
500 struct gfs2_trans {
501 	unsigned long tr_ip;
502 
503 	unsigned int tr_blocks;
504 	unsigned int tr_revokes;
505 	unsigned int tr_reserved;
506 	unsigned long tr_flags;
507 
508 	unsigned int tr_num_buf_new;
509 	unsigned int tr_num_databuf_new;
510 	unsigned int tr_num_buf_rm;
511 	unsigned int tr_num_databuf_rm;
512 	unsigned int tr_num_revoke;
513 	unsigned int tr_num_revoke_rm;
514 
515 	struct list_head tr_list;
516 	struct list_head tr_databuf;
517 	struct list_head tr_buf;
518 
519 	unsigned int tr_first;
520 	struct list_head tr_ail1_list;
521 	struct list_head tr_ail2_list;
522 };
523 
524 struct gfs2_journal_extent {
525 	struct list_head list;
526 
527 	unsigned int lblock; /* First logical block */
528 	u64 dblock; /* First disk block */
529 	u64 blocks;
530 };
531 
532 struct gfs2_jdesc {
533 	struct list_head jd_list;
534 	struct list_head extent_list;
535 	unsigned int nr_extents;
536 	struct work_struct jd_work;
537 	struct inode *jd_inode;
538 	unsigned long jd_flags;
539 #define JDF_RECOVERY 1
540 	unsigned int jd_jid;
541 	unsigned int jd_blocks;
542 	int jd_recover_error;
543 	/* Replay stuff */
544 
545 	unsigned int jd_found_blocks;
546 	unsigned int jd_found_revokes;
547 	unsigned int jd_replayed_blocks;
548 
549 	struct list_head jd_revoke_list;
550 	unsigned int jd_replay_tail;
551 
552 };
553 
554 struct gfs2_statfs_change_host {
555 	s64 sc_total;
556 	s64 sc_free;
557 	s64 sc_dinodes;
558 };
559 
560 #define GFS2_QUOTA_DEFAULT	GFS2_QUOTA_OFF
561 #define GFS2_QUOTA_OFF		0
562 #define GFS2_QUOTA_ACCOUNT	1
563 #define GFS2_QUOTA_ON		2
564 
565 #define GFS2_DATA_DEFAULT	GFS2_DATA_ORDERED
566 #define GFS2_DATA_WRITEBACK	1
567 #define GFS2_DATA_ORDERED	2
568 
569 #define GFS2_ERRORS_DEFAULT     GFS2_ERRORS_WITHDRAW
570 #define GFS2_ERRORS_WITHDRAW    0
571 #define GFS2_ERRORS_CONTINUE    1 /* place holder for future feature */
572 #define GFS2_ERRORS_RO          2 /* place holder for future feature */
573 #define GFS2_ERRORS_PANIC       3
574 
575 struct gfs2_args {
576 	char ar_lockproto[GFS2_LOCKNAME_LEN];	/* Name of the Lock Protocol */
577 	char ar_locktable[GFS2_LOCKNAME_LEN];	/* Name of the Lock Table */
578 	char ar_hostdata[GFS2_LOCKNAME_LEN];	/* Host specific data */
579 	unsigned int ar_spectator:1;		/* Don't get a journal */
580 	unsigned int ar_localflocks:1;		/* Let the VFS do flock|fcntl */
581 	unsigned int ar_debug:1;		/* Oops on errors */
582 	unsigned int ar_posix_acl:1;		/* Enable posix acls */
583 	unsigned int ar_quota:2;		/* off/account/on */
584 	unsigned int ar_suiddir:1;		/* suiddir support */
585 	unsigned int ar_data:2;			/* ordered/writeback */
586 	unsigned int ar_meta:1;			/* mount metafs */
587 	unsigned int ar_discard:1;		/* discard requests */
588 	unsigned int ar_errors:2;               /* errors=withdraw | panic */
589 	unsigned int ar_nobarrier:1;            /* do not send barriers */
590 	unsigned int ar_rgrplvb:1;		/* use lvbs for rgrp info */
591 	unsigned int ar_loccookie:1;		/* use location based readdir
592 						   cookies */
593 	int ar_commit;				/* Commit interval */
594 	int ar_statfs_quantum;			/* The fast statfs interval */
595 	int ar_quota_quantum;			/* The quota interval */
596 	int ar_statfs_percent;			/* The % change to force sync */
597 };
598 
599 struct gfs2_tune {
600 	spinlock_t gt_spin;
601 
602 	unsigned int gt_logd_secs;
603 
604 	unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */
605 	unsigned int gt_quota_scale_num; /* Numerator */
606 	unsigned int gt_quota_scale_den; /* Denominator */
607 	unsigned int gt_quota_quantum; /* Secs between syncs to quota file */
608 	unsigned int gt_new_files_jdata;
609 	unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */
610 	unsigned int gt_complain_secs;
611 	unsigned int gt_statfs_quantum;
612 	unsigned int gt_statfs_slow;
613 };
614 
615 enum {
616 	SDF_JOURNAL_CHECKED	= 0,
617 	SDF_JOURNAL_LIVE	= 1,
618 	SDF_SHUTDOWN		= 2,
619 	SDF_NOBARRIERS		= 3,
620 	SDF_NORECOVERY		= 4,
621 	SDF_DEMOTE		= 5,
622 	SDF_NOJOURNALID		= 6,
623 	SDF_RORECOVERY		= 7, /* read only recovery */
624 	SDF_SKIP_DLM_UNLOCK	= 8,
625 	SDF_FORCE_AIL_FLUSH     = 9,
626 	SDF_AIL1_IO_ERROR	= 10,
627 };
628 
629 enum gfs2_freeze_state {
630 	SFS_UNFROZEN		= 0,
631 	SFS_STARTING_FREEZE	= 1,
632 	SFS_FROZEN		= 2,
633 };
634 
635 #define GFS2_FSNAME_LEN		256
636 
637 struct gfs2_inum_host {
638 	u64 no_formal_ino;
639 	u64 no_addr;
640 };
641 
642 struct gfs2_sb_host {
643 	u32 sb_magic;
644 	u32 sb_type;
645 	u32 sb_format;
646 
647 	u32 sb_fs_format;
648 	u32 sb_multihost_format;
649 	u32 sb_bsize;
650 	u32 sb_bsize_shift;
651 
652 	struct gfs2_inum_host sb_master_dir;
653 	struct gfs2_inum_host sb_root_dir;
654 
655 	char sb_lockproto[GFS2_LOCKNAME_LEN];
656 	char sb_locktable[GFS2_LOCKNAME_LEN];
657 };
658 
659 /*
660  * lm_mount() return values
661  *
662  * ls_jid - the journal ID this node should use
663  * ls_first - this node is the first to mount the file system
664  * ls_lockspace - lock module's context for this file system
665  * ls_ops - lock module's functions
666  */
667 
668 struct lm_lockstruct {
669 	int ls_jid;
670 	unsigned int ls_first;
671 	const struct lm_lockops *ls_ops;
672 	dlm_lockspace_t *ls_dlm;
673 
674 	int ls_recover_jid_done;   /* These two are deprecated, */
675 	int ls_recover_jid_status; /* used previously by gfs_controld */
676 
677 	struct dlm_lksb ls_mounted_lksb; /* mounted_lock */
678 	struct dlm_lksb ls_control_lksb; /* control_lock */
679 	char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
680 	struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
681 	char *ls_lvb_bits;
682 
683 	spinlock_t ls_recover_spin; /* protects following fields */
684 	unsigned long ls_recover_flags; /* DFL_ */
685 	uint32_t ls_recover_mount; /* gen in first recover_done cb */
686 	uint32_t ls_recover_start; /* gen in last recover_done cb */
687 	uint32_t ls_recover_block; /* copy recover_start in last recover_prep */
688 	uint32_t ls_recover_size; /* size of recover_submit, recover_result */
689 	uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */
690 	uint32_t *ls_recover_result; /* result of last jid recovery */
691 };
692 
693 struct gfs2_pcpu_lkstats {
694 	/* One struct for each glock type */
695 	struct gfs2_lkstats lkstats[10];
696 };
697 
698 struct gfs2_sbd {
699 	struct super_block *sd_vfs;
700 	struct gfs2_pcpu_lkstats __percpu *sd_lkstats;
701 	struct kobject sd_kobj;
702 	unsigned long sd_flags;	/* SDF_... */
703 	struct gfs2_sb_host sd_sb;
704 
705 	/* Constants computed on mount */
706 
707 	u32 sd_fsb2bb;
708 	u32 sd_fsb2bb_shift;
709 	u32 sd_diptrs;	/* Number of pointers in a dinode */
710 	u32 sd_inptrs;	/* Number of pointers in a indirect block */
711 	u32 sd_jbsize;	/* Size of a journaled data block */
712 	u32 sd_hash_bsize;	/* sizeof(exhash block) */
713 	u32 sd_hash_bsize_shift;
714 	u32 sd_hash_ptrs;	/* Number of pointers in a hash block */
715 	u32 sd_qc_per_block;
716 	u32 sd_blocks_per_bitmap;
717 	u32 sd_max_dirres;	/* Max blocks needed to add a directory entry */
718 	u32 sd_max_height;	/* Max height of a file's metadata tree */
719 	u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1];
720 	u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */
721 
722 	struct gfs2_args sd_args;	/* Mount arguments */
723 	struct gfs2_tune sd_tune;	/* Filesystem tuning structure */
724 
725 	/* Lock Stuff */
726 
727 	struct lm_lockstruct sd_lockstruct;
728 	struct gfs2_holder sd_live_gh;
729 	struct gfs2_glock *sd_rename_gl;
730 	struct gfs2_glock *sd_freeze_gl;
731 	struct work_struct sd_freeze_work;
732 	wait_queue_head_t sd_glock_wait;
733 	atomic_t sd_glock_disposal;
734 	struct completion sd_locking_init;
735 	struct completion sd_wdack;
736 	struct delayed_work sd_control_work;
737 
738 	/* Inode Stuff */
739 
740 	struct dentry *sd_master_dir;
741 	struct dentry *sd_root_dir;
742 
743 	struct inode *sd_jindex;
744 	struct inode *sd_statfs_inode;
745 	struct inode *sd_sc_inode;
746 	struct inode *sd_qc_inode;
747 	struct inode *sd_rindex;
748 	struct inode *sd_quota_inode;
749 
750 	/* StatFS stuff */
751 
752 	spinlock_t sd_statfs_spin;
753 	struct gfs2_statfs_change_host sd_statfs_master;
754 	struct gfs2_statfs_change_host sd_statfs_local;
755 	int sd_statfs_force_sync;
756 
757 	/* Resource group stuff */
758 
759 	int sd_rindex_uptodate;
760 	spinlock_t sd_rindex_spin;
761 	struct rb_root sd_rindex_tree;
762 	unsigned int sd_rgrps;
763 	unsigned int sd_max_rg_data;
764 
765 	/* Journal index stuff */
766 
767 	struct list_head sd_jindex_list;
768 	spinlock_t sd_jindex_spin;
769 	struct mutex sd_jindex_mutex;
770 	unsigned int sd_journals;
771 
772 	struct gfs2_jdesc *sd_jdesc;
773 	struct gfs2_holder sd_journal_gh;
774 	struct gfs2_holder sd_jinode_gh;
775 
776 	struct gfs2_holder sd_sc_gh;
777 	struct gfs2_holder sd_qc_gh;
778 
779 	struct completion sd_journal_ready;
780 
781 	/* Daemon stuff */
782 
783 	struct task_struct *sd_logd_process;
784 	struct task_struct *sd_quotad_process;
785 
786 	/* Quota stuff */
787 
788 	struct list_head sd_quota_list;
789 	atomic_t sd_quota_count;
790 	struct mutex sd_quota_mutex;
791 	struct mutex sd_quota_sync_mutex;
792 	wait_queue_head_t sd_quota_wait;
793 	struct list_head sd_trunc_list;
794 	spinlock_t sd_trunc_lock;
795 
796 	unsigned int sd_quota_slots;
797 	unsigned long *sd_quota_bitmap;
798 	spinlock_t sd_bitmap_lock;
799 
800 	u64 sd_quota_sync_gen;
801 
802 	/* Log stuff */
803 
804 	struct address_space sd_aspace;
805 
806 	spinlock_t sd_log_lock;
807 
808 	struct gfs2_trans *sd_log_tr;
809 	unsigned int sd_log_blks_reserved;
810 	int sd_log_commited_revoke;
811 
812 	atomic_t sd_log_pinned;
813 	unsigned int sd_log_num_revoke;
814 
815 	struct list_head sd_log_le_revoke;
816 	struct list_head sd_log_le_ordered;
817 	spinlock_t sd_ordered_lock;
818 
819 	atomic_t sd_log_thresh1;
820 	atomic_t sd_log_thresh2;
821 	atomic_t sd_log_blks_free;
822 	atomic_t sd_log_blks_needed;
823 	wait_queue_head_t sd_log_waitq;
824 	wait_queue_head_t sd_logd_waitq;
825 
826 	u64 sd_log_sequence;
827 	unsigned int sd_log_head;
828 	unsigned int sd_log_tail;
829 	int sd_log_idle;
830 
831 	struct rw_semaphore sd_log_flush_lock;
832 	atomic_t sd_log_in_flight;
833 	struct bio *sd_log_bio;
834 	wait_queue_head_t sd_log_flush_wait;
835 	int sd_log_error;
836 
837 	atomic_t sd_reserving_log;
838 	wait_queue_head_t sd_reserving_log_wait;
839 
840 	unsigned int sd_log_flush_head;
841 
842 	spinlock_t sd_ail_lock;
843 	struct list_head sd_ail1_list;
844 	struct list_head sd_ail2_list;
845 
846 	/* For quiescing the filesystem */
847 	struct gfs2_holder sd_freeze_gh;
848 	atomic_t sd_freeze_state;
849 	struct mutex sd_freeze_mutex;
850 
851 	char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2];
852 	char sd_table_name[GFS2_FSNAME_LEN];
853 	char sd_proto_name[GFS2_FSNAME_LEN];
854 
855 	/* Debugging crud */
856 
857 	unsigned long sd_last_warning;
858 	struct dentry *debugfs_dir;    /* debugfs directory */
859 	struct dentry *debugfs_dentry_glocks;
860 	struct dentry *debugfs_dentry_glstats;
861 	struct dentry *debugfs_dentry_sbstats;
862 };
863 
gfs2_glstats_inc(struct gfs2_glock * gl,int which)864 static inline void gfs2_glstats_inc(struct gfs2_glock *gl, int which)
865 {
866 	gl->gl_stats.stats[which]++;
867 }
868 
gfs2_sbstats_inc(const struct gfs2_glock * gl,int which)869 static inline void gfs2_sbstats_inc(const struct gfs2_glock *gl, int which)
870 {
871 	const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
872 	preempt_disable();
873 	this_cpu_ptr(sdp->sd_lkstats)->lkstats[gl->gl_name.ln_type].stats[which]++;
874 	preempt_enable();
875 }
876 
877 extern struct gfs2_rgrpd *gfs2_glock2rgrp(struct gfs2_glock *gl);
878 
gfs2_max_stuffed_size(const struct gfs2_inode * ip)879 static inline unsigned gfs2_max_stuffed_size(const struct gfs2_inode *ip)
880 {
881 	return GFS2_SB(&ip->i_inode)->sd_sb.sb_bsize - sizeof(struct gfs2_dinode);
882 }
883 
884 #endif /* __INCORE_DOT_H__ */
885 
886