1 /* SPDX-License-Identifier: GPL-2.0-only */
2 /*
3  * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
4  * Copyright (C) 2004-2008 Red Hat, Inc.  All rights reserved.
5  */
6 
7 #ifndef __INCORE_DOT_H__
8 #define __INCORE_DOT_H__
9 
10 #include <linux/fs.h>
11 #include <linux/kobject.h>
12 #include <linux/workqueue.h>
13 #include <linux/dlm.h>
14 #include <linux/buffer_head.h>
15 #include <linux/rcupdate.h>
16 #include <linux/rculist_bl.h>
17 #include <linux/completion.h>
18 #include <linux/rbtree.h>
19 #include <linux/ktime.h>
20 #include <linux/percpu.h>
21 #include <linux/lockref.h>
22 #include <linux/rhashtable.h>
23 
24 #define DIO_WAIT	0x00000010
25 #define DIO_METADATA	0x00000020
26 
27 struct gfs2_log_operations;
28 struct gfs2_bufdata;
29 struct gfs2_holder;
30 struct gfs2_glock;
31 struct gfs2_quota_data;
32 struct gfs2_trans;
33 struct gfs2_jdesc;
34 struct gfs2_sbd;
35 struct lm_lockops;
36 
37 typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret);
38 
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 */
43 	u32 lh_blkno;
44 
45 	s64 lh_local_total;
46 	s64 lh_local_free;
47 	s64 lh_local_dinodes;
48 };
49 
50 /*
51  * Structure of operations that are associated with each
52  * type of element in the log.
53  */
54 
55 struct gfs2_log_operations {
56 	void (*lo_before_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
57 	void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
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,
62 				 int pass);
63 	void (*lo_after_scan) (struct gfs2_jdesc *jd, int error, int pass);
64 	const char *lo_name;
65 };
66 
67 #define GBF_FULL 1
68 
69 /**
70  * Clone bitmaps (bi_clone):
71  *
72  * - When a block is freed, we remember the previous state of the block in the
73  *   clone bitmap, and only mark the block as free in the real bitmap.
74  *
75  * - When looking for a block to allocate, we check for a free block in the
76  *   clone bitmap, and if no clone bitmap exists, in the real bitmap.
77  *
78  * - For allocating a block, we mark it as allocated in the real bitmap, and if
79  *   a clone bitmap exists, also in the clone bitmap.
80  *
81  * - At the end of a log_flush, we copy the real bitmap into the clone bitmap
82  *   to make the clone bitmap reflect the current allocation state.
83  *   (Alternatively, we could remove the clone bitmap.)
84  *
85  * The clone bitmaps are in-core only, and is never written to disk.
86  *
87  * These steps ensure that blocks which have been freed in a transaction cannot
88  * be reallocated in that same transaction.
89  */
90 struct gfs2_bitmap {
91 	struct buffer_head *bi_bh;
92 	char *bi_clone;
93 	unsigned long bi_flags;
94 	u32 bi_offset;
95 	u32 bi_start;
96 	u32 bi_bytes;
97 	u32 bi_blocks;
98 };
99 
100 struct gfs2_rgrpd {
101 	struct rb_node rd_node;		/* Link with superblock */
102 	struct gfs2_glock *rd_gl;	/* Glock for this rgrp */
103 	u64 rd_addr;			/* grp block disk address */
104 	u64 rd_data0;			/* first data location */
105 	u32 rd_length;			/* length of rgrp header in fs blocks */
106 	u32 rd_data;			/* num of data blocks in rgrp */
107 	u32 rd_bitbytes;		/* number of bytes in data bitmaps */
108 	u32 rd_free;
109 	u32 rd_reserved;                /* number of blocks reserved */
110 	u32 rd_free_clone;
111 	u32 rd_dinodes;
112 	u64 rd_igeneration;
113 	struct gfs2_bitmap *rd_bits;
114 	struct gfs2_sbd *rd_sbd;
115 	struct gfs2_rgrp_lvb *rd_rgl;
116 	u32 rd_last_alloc;
117 	u32 rd_flags;
118 	u32 rd_extfail_pt;		/* extent failure point */
119 #define GFS2_RDF_CHECK		0x10000000 /* check for unlinked inodes */
120 #define GFS2_RDF_UPTODATE	0x20000000 /* rg is up to date */
121 #define GFS2_RDF_ERROR		0x40000000 /* error in rg */
122 #define GFS2_RDF_PREFERRED	0x80000000 /* This rgrp is preferred */
123 #define GFS2_RDF_MASK		0xf0000000 /* mask for internal flags */
124 	spinlock_t rd_rsspin;           /* protects reservation related vars */
125 	struct rb_root rd_rstree;       /* multi-block reservation tree */
126 };
127 
128 struct gfs2_rbm {
129 	struct gfs2_rgrpd *rgd;
130 	u32 offset;		/* The offset is bitmap relative */
131 	int bii;		/* Bitmap index */
132 };
133 
rbm_bi(const struct gfs2_rbm * rbm)134 static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
135 {
136 	return rbm->rgd->rd_bits + rbm->bii;
137 }
138 
gfs2_rbm_to_block(const struct gfs2_rbm * rbm)139 static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm)
140 {
141 	BUG_ON(rbm->offset >= rbm->rgd->rd_data);
142 	return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) +
143 		rbm->offset;
144 }
145 
gfs2_rbm_eq(const struct gfs2_rbm * rbm1,const struct gfs2_rbm * rbm2)146 static inline bool gfs2_rbm_eq(const struct gfs2_rbm *rbm1,
147 			       const struct gfs2_rbm *rbm2)
148 {
149 	return (rbm1->rgd == rbm2->rgd) && (rbm1->bii == rbm2->bii) &&
150 	       (rbm1->offset == rbm2->offset);
151 }
152 
153 enum gfs2_state_bits {
154 	BH_Pinned = BH_PrivateStart,
155 	BH_Escaped = BH_PrivateStart + 1,
156 };
157 
158 BUFFER_FNS(Pinned, pinned)
159 TAS_BUFFER_FNS(Pinned, pinned)
160 BUFFER_FNS(Escaped, escaped)
161 TAS_BUFFER_FNS(Escaped, escaped)
162 
163 struct gfs2_bufdata {
164 	struct buffer_head *bd_bh;
165 	struct gfs2_glock *bd_gl;
166 	u64 bd_blkno;
167 
168 	struct list_head bd_list;
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 	int (*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_dump)(struct seq_file *seq, struct gfs2_glock *gl,
247 			const char *fs_id_buf);
248 	void (*go_callback)(struct gfs2_glock *gl, bool remote);
249 	void (*go_free)(struct gfs2_glock *gl);
250 	const int go_subclass;
251 	const int go_type;
252 	const unsigned long go_flags;
253 #define GLOF_ASPACE 1 /* address space attached */
254 #define GLOF_LVB    2 /* Lock Value Block attached */
255 #define GLOF_LRU    4 /* LRU managed */
256 #define GLOF_NONDISK   8 /* not I/O related */
257 };
258 
259 enum {
260 	GFS2_LKS_SRTT = 0,	/* Non blocking smoothed round trip time */
261 	GFS2_LKS_SRTTVAR = 1,	/* Non blocking smoothed variance */
262 	GFS2_LKS_SRTTB = 2,	/* Blocking smoothed round trip time */
263 	GFS2_LKS_SRTTVARB = 3,	/* Blocking smoothed variance */
264 	GFS2_LKS_SIRT = 4,	/* Smoothed Inter-request time */
265 	GFS2_LKS_SIRTVAR = 5,	/* Smoothed Inter-request variance */
266 	GFS2_LKS_DCOUNT = 6,	/* Count of dlm requests */
267 	GFS2_LKS_QCOUNT = 7,	/* Count of gfs2_holder queues */
268 	GFS2_NR_LKSTATS
269 };
270 
271 struct gfs2_lkstats {
272 	u64 stats[GFS2_NR_LKSTATS];
273 };
274 
275 enum {
276 	/* States */
277 	HIF_HOLDER		= 6,  /* Set for gh that "holds" the glock */
278 	HIF_FIRST		= 7,
279 	HIF_WAIT		= 10,
280 };
281 
282 struct gfs2_holder {
283 	struct list_head gh_list;
284 
285 	struct gfs2_glock *gh_gl;
286 	struct pid *gh_owner_pid;
287 	u16 gh_flags;
288 	u16 gh_state;
289 
290 	int gh_error;
291 	unsigned long gh_iflags; /* HIF_... */
292 	unsigned long gh_ip;
293 };
294 
295 /* Number of quota types we support */
296 #define GFS2_MAXQUOTAS 2
297 
298 struct gfs2_qadata { /* quota allocation data */
299 	/* Quota stuff */
300 	struct gfs2_quota_data *qa_qd[2 * GFS2_MAXQUOTAS];
301 	struct gfs2_holder qa_qd_ghs[2 * GFS2_MAXQUOTAS];
302 	unsigned int qa_qd_num;
303 	int qa_ref;
304 };
305 
306 /* Resource group multi-block reservation, in order of appearance:
307 
308    Step 1. Function prepares to write, allocates a mb, sets the size hint.
309    Step 2. User calls inplace_reserve to target an rgrp, sets the rgrp info
310    Step 3. Function get_local_rgrp locks the rgrp, determines which bits to use
311    Step 4. Bits are assigned from the rgrp based on either the reservation
312            or wherever it can.
313 */
314 
315 struct gfs2_blkreserv {
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_LRU				= 13,
349 	GLF_OBJECT			= 14, /* Used only for tracing */
350 	GLF_BLOCKING			= 15,
351 	GLF_INODE_CREATING		= 16, /* Inode creation occurring */
352 	GLF_PENDING_DELETE		= 17,
353 	GLF_FREEING			= 18, /* Wait for glock to be freed */
354 };
355 
356 struct gfs2_glock {
357 	unsigned long gl_flags;		/* GLF_... */
358 	struct lm_lockname gl_name;
359 
360 	struct lockref gl_lockref;
361 
362 	/* State fields protected by gl_lockref.lock */
363 	unsigned int gl_state:2,	/* Current state */
364 		     gl_target:2,	/* Target state */
365 		     gl_demote_state:2,	/* State requested by remote node */
366 		     gl_req:2,		/* State in last dlm request */
367 		     gl_reply:8;	/* Last reply from the dlm */
368 
369 	unsigned long gl_demote_time; /* time of first demote request */
370 	long gl_hold_time;
371 	struct list_head gl_holders;
372 
373 	const struct gfs2_glock_operations *gl_ops;
374 	ktime_t gl_dstamp;
375 	struct gfs2_lkstats gl_stats;
376 	struct dlm_lksb gl_lksb;
377 	unsigned long gl_tchange;
378 	void *gl_object;
379 
380 	struct list_head gl_lru;
381 	struct list_head gl_ail_list;
382 	atomic_t gl_ail_count;
383 	atomic_t gl_revokes;
384 	struct delayed_work gl_work;
385 	/* For iopen glocks only */
386 	struct {
387 		struct delayed_work gl_delete;
388 		u64 gl_no_formal_ino;
389 	};
390 	struct rcu_head gl_rcu;
391 	struct rhash_head gl_node;
392 };
393 
394 enum {
395 	GIF_INVALID		= 0,
396 	GIF_QD_LOCKED		= 1,
397 	GIF_ALLOC_FAILED	= 2,
398 	GIF_SW_PAGED		= 3,
399 	GIF_FREE_VFS_INODE      = 5,
400 	GIF_GLOP_PENDING	= 6,
401 	GIF_DEFERRED_DELETE	= 7,
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  */
GFS2_I(struct inode * inode)434 static inline struct gfs2_inode *GFS2_I(struct inode *inode)
435 {
436 	return container_of(inode, struct gfs2_inode, i_inode);
437 }
438 
GFS2_SB(const struct inode * inode)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 	u64 jd_no_addr;
550 };
551 
552 struct gfs2_statfs_change_host {
553 	s64 sc_total;
554 	s64 sc_free;
555 	s64 sc_dinodes;
556 };
557 
558 #define GFS2_QUOTA_DEFAULT	GFS2_QUOTA_OFF
559 #define GFS2_QUOTA_OFF		0
560 #define GFS2_QUOTA_ACCOUNT	1
561 #define GFS2_QUOTA_ON		2
562 
563 #define GFS2_DATA_DEFAULT	GFS2_DATA_ORDERED
564 #define GFS2_DATA_WRITEBACK	1
565 #define GFS2_DATA_ORDERED	2
566 
567 #define GFS2_ERRORS_DEFAULT     GFS2_ERRORS_WITHDRAW
568 #define GFS2_ERRORS_WITHDRAW    0
569 #define GFS2_ERRORS_CONTINUE    1 /* place holder for future feature */
570 #define GFS2_ERRORS_RO          2 /* place holder for future feature */
571 #define GFS2_ERRORS_PANIC       3
572 
573 struct gfs2_args {
574 	char ar_lockproto[GFS2_LOCKNAME_LEN];	/* Name of the Lock Protocol */
575 	char ar_locktable[GFS2_LOCKNAME_LEN];	/* Name of the Lock Table */
576 	char ar_hostdata[GFS2_LOCKNAME_LEN];	/* Host specific data */
577 	unsigned int ar_spectator:1;		/* Don't get a journal */
578 	unsigned int ar_localflocks:1;		/* Let the VFS do flock|fcntl */
579 	unsigned int ar_debug:1;		/* Oops on errors */
580 	unsigned int ar_posix_acl:1;		/* Enable posix acls */
581 	unsigned int ar_quota:2;		/* off/account/on */
582 	unsigned int ar_suiddir:1;		/* suiddir support */
583 	unsigned int ar_data:2;			/* ordered/writeback */
584 	unsigned int ar_meta:1;			/* mount metafs */
585 	unsigned int ar_discard:1;		/* discard requests */
586 	unsigned int ar_errors:2;               /* errors=withdraw | panic */
587 	unsigned int ar_nobarrier:1;            /* do not send barriers */
588 	unsigned int ar_rgrplvb:1;		/* use lvbs for rgrp info */
589 	unsigned int ar_loccookie:1;		/* use location based readdir
590 						   cookies */
591 	s32 ar_commit;				/* Commit interval */
592 	s32 ar_statfs_quantum;			/* The fast statfs interval */
593 	s32 ar_quota_quantum;			/* The quota interval */
594 	s32 ar_statfs_percent;			/* The % change to force sync */
595 };
596 
597 struct gfs2_tune {
598 	spinlock_t gt_spin;
599 
600 	unsigned int gt_logd_secs;
601 
602 	unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */
603 	unsigned int gt_quota_scale_num; /* Numerator */
604 	unsigned int gt_quota_scale_den; /* Denominator */
605 	unsigned int gt_quota_quantum; /* Secs between syncs to quota file */
606 	unsigned int gt_new_files_jdata;
607 	unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */
608 	unsigned int gt_complain_secs;
609 	unsigned int gt_statfs_quantum;
610 	unsigned int gt_statfs_slow;
611 };
612 
613 enum {
614 	SDF_JOURNAL_CHECKED	= 0,
615 	SDF_JOURNAL_LIVE	= 1,
616 	SDF_WITHDRAWN		= 2,
617 	SDF_NOBARRIERS		= 3,
618 	SDF_NORECOVERY		= 4,
619 	SDF_DEMOTE		= 5,
620 	SDF_NOJOURNALID		= 6,
621 	SDF_RORECOVERY		= 7, /* read only recovery */
622 	SDF_SKIP_DLM_UNLOCK	= 8,
623 	SDF_FORCE_AIL_FLUSH     = 9,
624 	SDF_FS_FROZEN           = 10,
625 	SDF_WITHDRAWING		= 11, /* Will withdraw eventually */
626 	SDF_WITHDRAW_IN_PROG	= 12, /* Withdraw is in progress */
627 	SDF_REMOTE_WITHDRAW	= 13, /* Performing remote recovery */
628 	SDF_WITHDRAW_RECOVERY	= 14, /* Wait for journal recovery when we are
629 					 withdrawing */
630 };
631 
632 enum gfs2_freeze_state {
633 	SFS_UNFROZEN		= 0,
634 	SFS_STARTING_FREEZE	= 1,
635 	SFS_FROZEN		= 2,
636 };
637 
638 #define GFS2_FSNAME_LEN		256
639 
640 struct gfs2_inum_host {
641 	u64 no_formal_ino;
642 	u64 no_addr;
643 };
644 
645 struct gfs2_sb_host {
646 	u32 sb_magic;
647 	u32 sb_type;
648 	u32 sb_format;
649 
650 	u32 sb_fs_format;
651 	u32 sb_multihost_format;
652 	u32 sb_bsize;
653 	u32 sb_bsize_shift;
654 
655 	struct gfs2_inum_host sb_master_dir;
656 	struct gfs2_inum_host sb_root_dir;
657 
658 	char sb_lockproto[GFS2_LOCKNAME_LEN];
659 	char sb_locktable[GFS2_LOCKNAME_LEN];
660 };
661 
662 /*
663  * lm_mount() return values
664  *
665  * ls_jid - the journal ID this node should use
666  * ls_first - this node is the first to mount the file system
667  * ls_lockspace - lock module's context for this file system
668  * ls_ops - lock module's functions
669  */
670 
671 struct lm_lockstruct {
672 	int ls_jid;
673 	unsigned int ls_first;
674 	const struct lm_lockops *ls_ops;
675 	dlm_lockspace_t *ls_dlm;
676 
677 	int ls_recover_jid_done;   /* These two are deprecated, */
678 	int ls_recover_jid_status; /* used previously by gfs_controld */
679 
680 	struct dlm_lksb ls_mounted_lksb; /* mounted_lock */
681 	struct dlm_lksb ls_control_lksb; /* control_lock */
682 	char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
683 	struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
684 	char *ls_lvb_bits;
685 
686 	spinlock_t ls_recover_spin; /* protects following fields */
687 	unsigned long ls_recover_flags; /* DFL_ */
688 	uint32_t ls_recover_mount; /* gen in first recover_done cb */
689 	uint32_t ls_recover_start; /* gen in last recover_done cb */
690 	uint32_t ls_recover_block; /* copy recover_start in last recover_prep */
691 	uint32_t ls_recover_size; /* size of recover_submit, recover_result */
692 	uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */
693 	uint32_t *ls_recover_result; /* result of last jid recovery */
694 };
695 
696 struct gfs2_pcpu_lkstats {
697 	/* One struct for each glock type */
698 	struct gfs2_lkstats lkstats[10];
699 };
700 
701 /* List of local (per node) statfs inodes */
702 struct local_statfs_inode {
703 	struct list_head si_list;
704 	struct inode *si_sc_inode;
705 	unsigned int si_jid; /* journal id this statfs inode corresponds to */
706 };
707 
708 struct gfs2_sbd {
709 	struct super_block *sd_vfs;
710 	struct gfs2_pcpu_lkstats __percpu *sd_lkstats;
711 	struct kobject sd_kobj;
712 	struct completion sd_kobj_unregister;
713 	unsigned long sd_flags;	/* SDF_... */
714 	struct gfs2_sb_host sd_sb;
715 
716 	/* Constants computed on mount */
717 
718 	u32 sd_fsb2bb;
719 	u32 sd_fsb2bb_shift;
720 	u32 sd_diptrs;	/* Number of pointers in a dinode */
721 	u32 sd_inptrs;	/* Number of pointers in a indirect block */
722 	u32 sd_ldptrs;  /* Number of pointers in a log descriptor block */
723 	u32 sd_jbsize;	/* Size of a journaled data block */
724 	u32 sd_hash_bsize;	/* sizeof(exhash block) */
725 	u32 sd_hash_bsize_shift;
726 	u32 sd_hash_ptrs;	/* Number of pointers in a hash block */
727 	u32 sd_qc_per_block;
728 	u32 sd_blocks_per_bitmap;
729 	u32 sd_max_dirres;	/* Max blocks needed to add a directory entry */
730 	u32 sd_max_height;	/* Max height of a file's metadata tree */
731 	u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1];
732 	u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */
733 
734 	struct gfs2_args sd_args;	/* Mount arguments */
735 	struct gfs2_tune sd_tune;	/* Filesystem tuning structure */
736 
737 	/* Lock Stuff */
738 
739 	struct lm_lockstruct sd_lockstruct;
740 	struct gfs2_holder sd_live_gh;
741 	struct gfs2_glock *sd_rename_gl;
742 	struct gfs2_glock *sd_freeze_gl;
743 	struct work_struct sd_freeze_work;
744 	wait_queue_head_t sd_glock_wait;
745 	wait_queue_head_t sd_async_glock_wait;
746 	atomic_t sd_glock_disposal;
747 	struct completion sd_locking_init;
748 	struct completion sd_wdack;
749 	struct delayed_work sd_control_work;
750 
751 	/* Inode Stuff */
752 
753 	struct dentry *sd_master_dir;
754 	struct dentry *sd_root_dir;
755 
756 	struct inode *sd_jindex;
757 	struct inode *sd_statfs_inode;
758 	struct inode *sd_sc_inode;
759 	struct list_head sd_sc_inodes_list;
760 	struct inode *sd_qc_inode;
761 	struct inode *sd_rindex;
762 	struct inode *sd_quota_inode;
763 
764 	/* StatFS stuff */
765 
766 	spinlock_t sd_statfs_spin;
767 	struct gfs2_statfs_change_host sd_statfs_master;
768 	struct gfs2_statfs_change_host sd_statfs_local;
769 	int sd_statfs_force_sync;
770 
771 	/* Resource group stuff */
772 
773 	int sd_rindex_uptodate;
774 	spinlock_t sd_rindex_spin;
775 	struct rb_root sd_rindex_tree;
776 	unsigned int sd_rgrps;
777 	unsigned int sd_max_rg_data;
778 
779 	/* Journal index stuff */
780 
781 	struct list_head sd_jindex_list;
782 	spinlock_t sd_jindex_spin;
783 	struct mutex sd_jindex_mutex;
784 	unsigned int sd_journals;
785 
786 	struct gfs2_jdesc *sd_jdesc;
787 	struct gfs2_holder sd_journal_gh;
788 	struct gfs2_holder sd_jinode_gh;
789 	struct gfs2_glock *sd_jinode_gl;
790 
791 	struct gfs2_holder sd_sc_gh;
792 	struct gfs2_holder sd_qc_gh;
793 
794 	struct completion sd_journal_ready;
795 
796 	/* Daemon stuff */
797 
798 	struct task_struct *sd_logd_process;
799 	struct task_struct *sd_quotad_process;
800 
801 	/* Quota stuff */
802 
803 	struct list_head sd_quota_list;
804 	atomic_t sd_quota_count;
805 	struct mutex sd_quota_mutex;
806 	struct mutex sd_quota_sync_mutex;
807 	wait_queue_head_t sd_quota_wait;
808 	struct list_head sd_trunc_list;
809 	spinlock_t sd_trunc_lock;
810 
811 	unsigned int sd_quota_slots;
812 	unsigned long *sd_quota_bitmap;
813 	spinlock_t sd_bitmap_lock;
814 
815 	u64 sd_quota_sync_gen;
816 
817 	/* Log stuff */
818 
819 	struct address_space sd_aspace;
820 
821 	spinlock_t sd_log_lock;
822 
823 	struct gfs2_trans *sd_log_tr;
824 	unsigned int sd_log_blks_reserved;
825 	int sd_log_committed_revoke;
826 
827 	atomic_t sd_log_pinned;
828 	unsigned int sd_log_num_revoke;
829 
830 	struct list_head sd_log_revokes;
831 	struct list_head sd_log_ordered;
832 	spinlock_t sd_ordered_lock;
833 
834 	atomic_t sd_log_thresh1;
835 	atomic_t sd_log_thresh2;
836 	atomic_t sd_log_blks_free;
837 	atomic_t sd_log_blks_needed;
838 	wait_queue_head_t sd_log_waitq;
839 	wait_queue_head_t sd_logd_waitq;
840 
841 	u64 sd_log_sequence;
842 	unsigned int sd_log_head;
843 	unsigned int sd_log_tail;
844 	int sd_log_idle;
845 
846 	struct rw_semaphore sd_log_flush_lock;
847 	atomic_t sd_log_in_flight;
848 	struct bio *sd_log_bio;
849 	wait_queue_head_t sd_log_flush_wait;
850 	int sd_log_error; /* First log error */
851 	wait_queue_head_t sd_withdraw_wait;
852 
853 	atomic_t sd_reserving_log;
854 	wait_queue_head_t sd_reserving_log_wait;
855 
856 	unsigned int sd_log_flush_head;
857 
858 	spinlock_t sd_ail_lock;
859 	struct list_head sd_ail1_list;
860 	struct list_head sd_ail2_list;
861 
862 	/* For quiescing the filesystem */
863 	struct gfs2_holder sd_freeze_gh;
864 	atomic_t sd_freeze_state;
865 	struct mutex sd_freeze_mutex;
866 
867 	char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2];
868 	char sd_table_name[GFS2_FSNAME_LEN];
869 	char sd_proto_name[GFS2_FSNAME_LEN];
870 
871 	/* Debugging crud */
872 
873 	unsigned long sd_last_warning;
874 	struct dentry *debugfs_dir;    /* debugfs directory */
875 	unsigned long sd_glock_dqs_held;
876 };
877 
gfs2_glstats_inc(struct gfs2_glock * gl,int which)878 static inline void gfs2_glstats_inc(struct gfs2_glock *gl, int which)
879 {
880 	gl->gl_stats.stats[which]++;
881 }
882 
gfs2_sbstats_inc(const struct gfs2_glock * gl,int which)883 static inline void gfs2_sbstats_inc(const struct gfs2_glock *gl, int which)
884 {
885 	const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
886 	preempt_disable();
887 	this_cpu_ptr(sdp->sd_lkstats)->lkstats[gl->gl_name.ln_type].stats[which]++;
888 	preempt_enable();
889 }
890 
891 extern struct gfs2_rgrpd *gfs2_glock2rgrp(struct gfs2_glock *gl);
892 
gfs2_max_stuffed_size(const struct gfs2_inode * ip)893 static inline unsigned gfs2_max_stuffed_size(const struct gfs2_inode *ip)
894 {
895 	return GFS2_SB(&ip->i_inode)->sd_sb.sb_bsize - sizeof(struct gfs2_dinode);
896 }
897 
898 #endif /* __INCORE_DOT_H__ */
899 
900