]>
Commit | Line | Data |
---|---|---|
b3b94faa DT |
1 | /* |
2 | * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. | |
29d38cd1 | 3 | * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved. |
b3b94faa DT |
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 | |
e9fc2aa0 | 7 | * of the GNU General Public License version 2. |
b3b94faa DT |
8 | */ |
9 | ||
10 | #ifndef __INCORE_DOT_H__ | |
11 | #define __INCORE_DOT_H__ | |
12 | ||
f2f7ba52 | 13 | #include <linux/fs.h> |
57cc7215 | 14 | #include <linux/kobject.h> |
c4f68a13 | 15 | #include <linux/workqueue.h> |
f057f6cd SW |
16 | #include <linux/dlm.h> |
17 | #include <linux/buffer_head.h> | |
bc015cb8 SW |
18 | #include <linux/rcupdate.h> |
19 | #include <linux/rculist_bl.h> | |
3942ae53 | 20 | #include <linux/completion.h> |
7c9ca621 | 21 | #include <linux/rbtree.h> |
a245769f SW |
22 | #include <linux/ktime.h> |
23 | #include <linux/percpu.h> | |
e66cf161 | 24 | #include <linux/lockref.h> |
88ffbf3e | 25 | #include <linux/rhashtable.h> |
f2f7ba52 | 26 | |
b3b94faa DT |
27 | #define DIO_WAIT 0x00000010 |
28 | #define DIO_METADATA 0x00000020 | |
b3b94faa DT |
29 | |
30 | struct gfs2_log_operations; | |
c0752aa7 | 31 | struct gfs2_bufdata; |
b3b94faa DT |
32 | struct gfs2_holder; |
33 | struct gfs2_glock; | |
b3b94faa | 34 | struct gfs2_quota_data; |
b3b94faa | 35 | struct gfs2_trans; |
b3b94faa | 36 | struct gfs2_jdesc; |
b3b94faa | 37 | struct gfs2_sbd; |
f057f6cd | 38 | struct lm_lockops; |
b3b94faa DT |
39 | |
40 | typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret); | |
41 | ||
bb8d8a6f SW |
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; | |
bb8d8a6f SW |
47 | }; |
48 | ||
b3b94faa DT |
49 | /* |
50 | * Structure of operations that are associated with each | |
51 | * type of element in the log. | |
52 | */ | |
53 | ||
54 | struct gfs2_log_operations { | |
d69a3c65 | 55 | void (*lo_before_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr); |
16ca9412 | 56 | void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr); |
b3b94faa | 57 | void (*lo_before_scan) (struct gfs2_jdesc *jd, |
55167622 | 58 | struct gfs2_log_header_host *head, int pass); |
b3b94faa DT |
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); | |
b09e593d | 63 | const char *lo_name; |
b3b94faa DT |
64 | }; |
65 | ||
60a0b8f9 SW |
66 | #define GBF_FULL 1 |
67 | ||
dffe12a8 BP |
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 | */ | |
b3b94faa DT |
89 | struct gfs2_bitmap { |
90 | struct buffer_head *bi_bh; | |
91 | char *bi_clone; | |
60a0b8f9 | 92 | unsigned long bi_flags; |
cd915493 SW |
93 | u32 bi_offset; |
94 | u32 bi_start; | |
281b4952 | 95 | u32 bi_bytes; |
7e230f57 | 96 | u32 bi_blocks; |
b3b94faa DT |
97 | }; |
98 | ||
99 | struct gfs2_rgrpd { | |
7c9ca621 | 100 | struct rb_node rd_node; /* Link with superblock */ |
b3b94faa | 101 | struct gfs2_glock *rd_gl; /* Glock for this rgrp */ |
bb8d8a6f SW |
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 */ | |
cfc8b549 | 107 | u32 rd_free; |
8e2e0047 | 108 | u32 rd_reserved; /* number of blocks reserved */ |
73f74948 SW |
109 | u32 rd_free_clone; |
110 | u32 rd_dinodes; | |
d8b71f73 | 111 | u64 rd_igeneration; |
b3b94faa | 112 | struct gfs2_bitmap *rd_bits; |
b3b94faa | 113 | struct gfs2_sbd *rd_sbd; |
90306c41 | 114 | struct gfs2_rgrp_lvb *rd_rgl; |
73f74948 | 115 | u32 rd_last_alloc; |
1ce97e56 | 116 | u32 rd_flags; |
5ea5050c | 117 | u32 rd_extfail_pt; /* extent failure point */ |
09010978 SW |
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 */ | |
0e27c18c | 121 | #define GFS2_RDF_PREFERRED 0x80000000 /* This rgrp is preferred */ |
09010978 | 122 | #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */ |
8e2e0047 BP |
123 | spinlock_t rd_rsspin; /* protects reservation related vars */ |
124 | struct rb_root rd_rstree; /* multi-block reservation tree */ | |
b3b94faa DT |
125 | }; |
126 | ||
4a993fb1 SW |
127 | struct gfs2_rbm { |
128 | struct gfs2_rgrpd *rgd; | |
4a993fb1 | 129 | u32 offset; /* The offset is bitmap relative */ |
e579ed4f | 130 | int bii; /* Bitmap index */ |
4a993fb1 SW |
131 | }; |
132 | ||
e579ed4f BP |
133 | static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm) |
134 | { | |
135 | return rbm->rgd->rd_bits + rbm->bii; | |
136 | } | |
137 | ||
4a993fb1 SW |
138 | static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm) |
139 | { | |
dc8fbb03 | 140 | BUG_ON(rbm->offset >= rbm->rgd->rd_data); |
e579ed4f BP |
141 | return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) + |
142 | rbm->offset; | |
4a993fb1 SW |
143 | } |
144 | ||
5b924ae2 SW |
145 | static inline bool gfs2_rbm_eq(const struct gfs2_rbm *rbm1, |
146 | const struct gfs2_rbm *rbm2) | |
147 | { | |
e579ed4f | 148 | return (rbm1->rgd == rbm2->rgd) && (rbm1->bii == rbm2->bii) && |
5b924ae2 SW |
149 | (rbm1->offset == rbm2->offset); |
150 | } | |
151 | ||
b3b94faa DT |
152 | enum gfs2_state_bits { |
153 | BH_Pinned = BH_PrivateStart, | |
f42faf4f | 154 | BH_Escaped = BH_PrivateStart + 1, |
b3b94faa DT |
155 | }; |
156 | ||
157 | BUFFER_FNS(Pinned, pinned) | |
158 | TAS_BUFFER_FNS(Pinned, pinned) | |
f42faf4f SW |
159 | BUFFER_FNS(Escaped, escaped) |
160 | TAS_BUFFER_FNS(Escaped, escaped) | |
b3b94faa DT |
161 | |
162 | struct gfs2_bufdata { | |
163 | struct buffer_head *bd_bh; | |
164 | struct gfs2_glock *bd_gl; | |
c50b91c4 | 165 | u64 bd_blkno; |
82e86087 | 166 | |
c0752aa7 | 167 | struct list_head bd_list; |
b3b94faa | 168 | |
16ca9412 | 169 | struct gfs2_trans *bd_tr; |
b3b94faa DT |
170 | struct list_head bd_ail_st_list; |
171 | struct list_head bd_ail_gl_list; | |
172 | }; | |
173 | ||
f057f6cd SW |
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 | ||
e0c2a9aa DT |
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 | ||
f057f6cd SW |
213 | enum { |
214 | DFL_BLOCK_LOCKS = 0, | |
e0c2a9aa DT |
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, | |
f057f6cd SW |
221 | }; |
222 | ||
972b044e AG |
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 | */ | |
f057f6cd SW |
227 | struct lm_lockname { |
228 | u64 ln_number; | |
972b044e | 229 | struct gfs2_sbd *ln_sbd; |
f057f6cd | 230 | unsigned int ln_type; |
972b044e | 231 | }; |
f057f6cd SW |
232 | |
233 | #define lm_name_equal(name1, name2) \ | |
15562c43 BP |
234 | (((name1)->ln_number == (name2)->ln_number) && \ |
235 | ((name1)->ln_type == (name2)->ln_type) && \ | |
236 | ((name1)->ln_sbd == (name2)->ln_sbd)) | |
f057f6cd SW |
237 | |
238 | ||
b3b94faa | 239 | struct gfs2_glock_operations { |
06dfc306 | 240 | void (*go_sync) (struct gfs2_glock *gl); |
6802e340 | 241 | int (*go_xmote_bh) (struct gfs2_glock *gl, struct gfs2_holder *gh); |
1a14d3a6 | 242 | void (*go_inval) (struct gfs2_glock *gl, int flags); |
97cc1025 | 243 | int (*go_demote_ok) (const struct gfs2_glock *gl); |
1a14d3a6 SW |
244 | int (*go_lock) (struct gfs2_holder *gh); |
245 | void (*go_unlock) (struct gfs2_holder *gh); | |
27a2660f | 246 | void (*go_dump)(struct seq_file *seq, struct gfs2_glock *gl); |
81ffbf65 | 247 | void (*go_callback)(struct gfs2_glock *gl, bool remote); |
8fb4b536 | 248 | const int go_type; |
009d8518 SW |
249 | const unsigned long go_flags; |
250 | #define GLOF_ASPACE 1 | |
dba2d70c | 251 | #define GLOF_LVB 2 |
e7ccaf5f | 252 | #define GLOF_LRU 4 |
b3b94faa DT |
253 | }; |
254 | ||
a245769f SW |
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 { | |
4d207133 | 268 | u64 stats[GFS2_NR_LKSTATS]; |
a245769f SW |
269 | }; |
270 | ||
b3b94faa | 271 | enum { |
b3b94faa | 272 | /* States */ |
6802e340 | 273 | HIF_HOLDER = 6, /* Set for gh that "holds" the glock */ |
b3b94faa | 274 | HIF_FIRST = 7, |
fee852e3 | 275 | HIF_WAIT = 10, |
b3b94faa DT |
276 | }; |
277 | ||
278 | struct gfs2_holder { | |
279 | struct list_head gh_list; | |
280 | ||
281 | struct gfs2_glock *gh_gl; | |
b1e058da | 282 | struct pid *gh_owner_pid; |
b58bf407 BP |
283 | u16 gh_flags; |
284 | u16 gh_state; | |
b3b94faa DT |
285 | |
286 | int gh_error; | |
6802e340 | 287 | unsigned long gh_iflags; /* HIF_... */ |
d0dc80db | 288 | unsigned long gh_ip; |
b3b94faa DT |
289 | }; |
290 | ||
a937cca2 JK |
291 | /* Number of quota types we support */ |
292 | #define GFS2_MAXQUOTAS 2 | |
293 | ||
b54e9a0b BP |
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 | ||
8e2e0047 BP |
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 { | |
8e2e0047 | 311 | struct rb_node rs_node; /* link to other block reservations */ |
9e733d39 | 312 | struct gfs2_rbm rs_rbm; /* Start of reservation */ |
8e2e0047 | 313 | u32 rs_free; /* how many blocks are still free */ |
8e2e0047 BP |
314 | }; |
315 | ||
7b9cff46 SW |
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 { | |
b8fbf471 | 326 | u64 target; |
25435e5e | 327 | u32 min_target; |
7b9cff46 | 328 | u32 aflags; |
25435e5e | 329 | u64 allowed; |
7b9cff46 SW |
330 | }; |
331 | ||
b3b94faa | 332 | enum { |
6802e340 | 333 | GLF_LOCK = 1, |
6802e340 SW |
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, | |
f057f6cd SW |
341 | GLF_INITIAL = 10, |
342 | GLF_FROZEN = 11, | |
7b5e3d5f | 343 | GLF_QUEUED = 12, |
627c10b7 SW |
344 | GLF_LRU = 13, |
345 | GLF_OBJECT = 14, /* Used only for tracing */ | |
a245769f | 346 | GLF_BLOCKING = 15, |
a4923865 | 347 | GLF_INODE_CREATING = 16, /* Inode creation occurring */ |
73118ca8 | 348 | GLF_REVOKES = 17, /* Glock has revokes in queue */ |
b3b94faa DT |
349 | }; |
350 | ||
351 | struct gfs2_glock { | |
b3b94faa DT |
352 | unsigned long gl_flags; /* GLF_... */ |
353 | struct lm_lockname gl_name; | |
b3b94faa | 354 | |
e66cf161 | 355 | struct lockref gl_lockref; |
b3b94faa | 356 | |
f3dd1649 | 357 | /* State fields protected by gl_lockref.lock */ |
47a25380 SW |
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 | ||
3b8249f6 | 364 | unsigned long gl_demote_time; /* time of first demote request */ |
7cf8dcd3 | 365 | long gl_hold_time; |
b3b94faa | 366 | struct list_head gl_holders; |
b3b94faa | 367 | |
8fb4b536 | 368 | const struct gfs2_glock_operations *gl_ops; |
a245769f SW |
369 | ktime_t gl_dstamp; |
370 | struct gfs2_lkstats gl_stats; | |
f057f6cd | 371 | struct dlm_lksb gl_lksb; |
c4f68a13 | 372 | unsigned long gl_tchange; |
b3b94faa DT |
373 | void *gl_object; |
374 | ||
97cc1025 | 375 | struct list_head gl_lru; |
b3b94faa DT |
376 | struct list_head gl_ail_list; |
377 | atomic_t gl_ail_count; | |
c4f68a13 | 378 | struct delayed_work gl_work; |
7005c3e4 SW |
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 | }; | |
961ae1d8 | 388 | struct rcu_head gl_rcu; |
88ffbf3e | 389 | struct rhash_head gl_node; |
b3b94faa DT |
390 | }; |
391 | ||
da755fdb SW |
392 | #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */ |
393 | ||
b3b94faa | 394 | enum { |
bfded27b | 395 | GIF_INVALID = 0, |
b3b94faa | 396 | GIF_QD_LOCKED = 1, |
40ac218f | 397 | GIF_ALLOC_FAILED = 2, |
b3b94faa | 398 | GIF_SW_PAGED = 3, |
45138990 | 399 | GIF_ORDERED = 4, |
05978803 | 400 | GIF_FREE_VFS_INODE = 5, |
4fd1a579 | 401 | GIF_GLOP_PENDING = 6, |
b3b94faa DT |
402 | }; |
403 | ||
404 | struct gfs2_inode { | |
320dd101 | 405 | struct inode i_inode; |
dbb7cae2 SW |
406 | u64 i_no_addr; |
407 | u64 i_no_formal_ino; | |
bcf0b5b3 | 408 | u64 i_generation; |
3767ac21 | 409 | u64 i_eattr; |
b3b94faa | 410 | unsigned long i_flags; /* GIF_... */ |
feaa7bba | 411 | struct gfs2_glock *i_gl; /* Move into i_gh? */ |
b3b94faa | 412 | struct gfs2_holder i_iopen_gh; |
f42faf4f | 413 | struct gfs2_holder i_gh; /* for prepare/commit_write only */ |
b54e9a0b | 414 | struct gfs2_qadata *i_qadata; /* quota allocation data */ |
21f09c43 | 415 | struct gfs2_holder i_rgd_gh; |
a097dc7e | 416 | struct gfs2_blkreserv i_res; /* rgrp multi-block reservation */ |
ce276b06 | 417 | u64 i_goal; /* goal block for allocations */ |
21f09c43 | 418 | atomic_t i_sizehint; /* hint of the write size */ |
b3b94faa | 419 | struct rw_semaphore i_rw_mutex; |
45138990 | 420 | struct list_head i_ordered; |
813e0c46 | 421 | struct list_head i_trunc_list; |
17d539f0 | 422 | __be64 *i_hash_cache; |
ad6203f2 | 423 | u32 i_entries; |
383f01fb | 424 | u32 i_diskflags; |
ecc30c79 | 425 | u8 i_height; |
9a004508 | 426 | u8 i_depth; |
c8d57703 | 427 | u16 i_rahead; |
b3b94faa DT |
428 | }; |
429 | ||
feaa7bba SW |
430 | /* |
431 | * Since i_inode is the first element of struct gfs2_inode, | |
432 | * this is effectively a cast. | |
433 | */ | |
320dd101 SW |
434 | static inline struct gfs2_inode *GFS2_I(struct inode *inode) |
435 | { | |
436 | return container_of(inode, struct gfs2_inode, i_inode); | |
437 | } | |
438 | ||
bf36a713 | 439 | static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode) |
feaa7bba SW |
440 | { |
441 | return inode->i_sb->s_fs_info; | |
442 | } | |
443 | ||
b3b94faa | 444 | struct gfs2_file { |
f55ab26a | 445 | struct mutex f_fl_mutex; |
b3b94faa | 446 | struct gfs2_holder f_fl_gh; |
b3b94faa DT |
447 | }; |
448 | ||
b3b94faa DT |
449 | struct gfs2_revoke_replay { |
450 | struct list_head rr_list; | |
cd915493 | 451 | u64 rr_blkno; |
b3b94faa DT |
452 | unsigned int rr_where; |
453 | }; | |
454 | ||
b3b94faa | 455 | enum { |
b3b94faa DT |
456 | QDF_CHANGE = 1, |
457 | QDF_LOCKED = 2, | |
662e3a55 | 458 | QDF_REFRESH = 3, |
9cde2898 | 459 | QDF_QMSG_QUIET = 4, |
b3b94faa DT |
460 | }; |
461 | ||
462 | struct gfs2_quota_data { | |
c754fbbb | 463 | struct hlist_bl_node qd_hlist; |
b3b94faa | 464 | struct list_head qd_list; |
2147dbfd | 465 | struct kqid qd_id; |
c754fbbb | 466 | struct gfs2_sbd *qd_sbd; |
9b9f039d | 467 | struct lockref qd_lockref; |
2147dbfd | 468 | struct list_head qd_lru; |
c754fbbb | 469 | unsigned qd_hash; |
b3b94faa | 470 | |
b3b94faa DT |
471 | unsigned long qd_flags; /* QDF_... */ |
472 | ||
cd915493 SW |
473 | s64 qd_change; |
474 | s64 qd_change_sync; | |
b3b94faa DT |
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 | ||
cd915493 | 486 | u64 qd_sync_gen; |
b3b94faa | 487 | unsigned long qd_last_warn; |
c754fbbb | 488 | struct rcu_head qd_rcu; |
b3b94faa DT |
489 | }; |
490 | ||
9862ca05 BP |
491 | enum { |
492 | TR_TOUCHED = 1, | |
493 | TR_ATTACHED = 2, | |
494 | TR_ALLOCED = 3, | |
495 | }; | |
496 | ||
b3b94faa | 497 | struct gfs2_trans { |
d0dc80db | 498 | unsigned long tr_ip; |
b3b94faa DT |
499 | |
500 | unsigned int tr_blocks; | |
501 | unsigned int tr_revokes; | |
502 | unsigned int tr_reserved; | |
9862ca05 | 503 | unsigned long tr_flags; |
b3b94faa | 504 | |
b3b94faa | 505 | unsigned int tr_num_buf_new; |
2332c443 | 506 | unsigned int tr_num_databuf_new; |
b3b94faa | 507 | unsigned int tr_num_buf_rm; |
2332c443 | 508 | unsigned int tr_num_databuf_rm; |
b3b94faa DT |
509 | unsigned int tr_num_revoke; |
510 | unsigned int tr_num_revoke_rm; | |
b3b94faa | 511 | |
16ca9412 | 512 | struct list_head tr_list; |
d69a3c65 SW |
513 | struct list_head tr_databuf; |
514 | struct list_head tr_buf; | |
b3b94faa | 515 | |
16ca9412 BM |
516 | unsigned int tr_first; |
517 | struct list_head tr_ail1_list; | |
518 | struct list_head tr_ail2_list; | |
b3b94faa DT |
519 | }; |
520 | ||
da6dd40d | 521 | struct gfs2_journal_extent { |
b50f227b | 522 | struct list_head list; |
da6dd40d BP |
523 | |
524 | unsigned int lblock; /* First logical block */ | |
525 | u64 dblock; /* First disk block */ | |
526 | u64 blocks; | |
527 | }; | |
528 | ||
b3b94faa DT |
529 | struct gfs2_jdesc { |
530 | struct list_head jd_list; | |
da6dd40d | 531 | struct list_head extent_list; |
b50f227b | 532 | unsigned int nr_extents; |
6ecd7c2d | 533 | struct work_struct jd_work; |
7359a19c | 534 | struct inode *jd_inode; |
fe64d517 SW |
535 | unsigned long jd_flags; |
536 | #define JDF_RECOVERY 1 | |
b3b94faa | 537 | unsigned int jd_jid; |
7c70b896 | 538 | u32 jd_blocks; |
376d3778 | 539 | int jd_recover_error; |
a17d758b BP |
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 | ||
b3b94faa DT |
549 | }; |
550 | ||
bb8d8a6f SW |
551 | struct gfs2_statfs_change_host { |
552 | s64 sc_total; | |
553 | s64 sc_free; | |
554 | s64 sc_dinodes; | |
555 | }; | |
556 | ||
b3b94faa DT |
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 | ||
d34843d0 BP |
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 | ||
b3b94faa | 572 | struct gfs2_args { |
37ec89e8 SW |
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 */ | |
37ec89e8 | 577 | unsigned int ar_localflocks:1; /* Let the VFS do flock|fcntl */ |
37ec89e8 | 578 | unsigned int ar_debug:1; /* Oops on errors */ |
37ec89e8 SW |
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 */ | |
f15ab561 | 584 | unsigned int ar_discard:1; /* discard requests */ |
d34843d0 | 585 | unsigned int ar_errors:2; /* errors=withdraw | panic */ |
f25934c5 | 586 | unsigned int ar_nobarrier:1; /* do not send barriers */ |
90306c41 | 587 | unsigned int ar_rgrplvb:1; /* use lvbs for rgrp info */ |
471f3db2 BM |
588 | unsigned int ar_loccookie:1; /* use location based readdir |
589 | cookies */ | |
48c2b613 | 590 | int ar_commit; /* Commit interval */ |
3d3c10f2 BM |
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 */ | |
b3b94faa DT |
594 | }; |
595 | ||
596 | struct gfs2_tune { | |
597 | spinlock_t gt_spin; | |
598 | ||
b3b94faa | 599 | unsigned int gt_logd_secs; |
b3b94faa | 600 | |
b3b94faa DT |
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 */ | |
b3b94faa | 604 | unsigned int gt_quota_quantum; /* Secs between syncs to quota file */ |
b3b94faa | 605 | unsigned int gt_new_files_jdata; |
b3b94faa | 606 | unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */ |
b3b94faa | 607 | unsigned int gt_complain_secs; |
b3b94faa DT |
608 | unsigned int gt_statfs_quantum; |
609 | unsigned int gt_statfs_slow; | |
610 | }; | |
611 | ||
b3b94faa DT |
612 | enum { |
613 | SDF_JOURNAL_CHECKED = 0, | |
614 | SDF_JOURNAL_LIVE = 1, | |
615 | SDF_SHUTDOWN = 2, | |
254db57f | 616 | SDF_NOBARRIERS = 3, |
fe64d517 | 617 | SDF_NORECOVERY = 4, |
913a71d2 | 618 | SDF_DEMOTE = 5, |
ba6e9364 | 619 | SDF_NOJOURNALID = 6, |
e8ca5cc5 | 620 | SDF_RORECOVERY = 7, /* read only recovery */ |
fb6791d1 | 621 | SDF_SKIP_DLM_UNLOCK = 8, |
b066a4ee | 622 | SDF_FORCE_AIL_FLUSH = 9, |
b524abcc | 623 | SDF_AIL1_IO_ERROR = 10, |
b3b94faa DT |
624 | }; |
625 | ||
2e60d768 BM |
626 | enum gfs2_freeze_state { |
627 | SFS_UNFROZEN = 0, | |
628 | SFS_STARTING_FREEZE = 1, | |
629 | SFS_FROZEN = 2, | |
630 | }; | |
631 | ||
b3b94faa DT |
632 | #define GFS2_FSNAME_LEN 256 |
633 | ||
bb8d8a6f SW |
634 | struct gfs2_inum_host { |
635 | u64 no_formal_ino; | |
636 | u64 no_addr; | |
637 | }; | |
638 | ||
639 | struct gfs2_sb_host { | |
640 | u32 sb_magic; | |
641 | u32 sb_type; | |
642 | u32 sb_format; | |
643 | ||
644 | u32 sb_fs_format; | |
645 | u32 sb_multihost_format; | |
646 | u32 sb_bsize; | |
647 | u32 sb_bsize_shift; | |
648 | ||
649 | struct gfs2_inum_host sb_master_dir; | |
650 | struct gfs2_inum_host sb_root_dir; | |
651 | ||
652 | char sb_lockproto[GFS2_LOCKNAME_LEN]; | |
653 | char sb_locktable[GFS2_LOCKNAME_LEN]; | |
654 | }; | |
655 | ||
f057f6cd SW |
656 | /* |
657 | * lm_mount() return values | |
658 | * | |
659 | * ls_jid - the journal ID this node should use | |
660 | * ls_first - this node is the first to mount the file system | |
661 | * ls_lockspace - lock module's context for this file system | |
662 | * ls_ops - lock module's functions | |
663 | */ | |
664 | ||
665 | struct lm_lockstruct { | |
feb47ca9 | 666 | int ls_jid; |
f057f6cd | 667 | unsigned int ls_first; |
f057f6cd | 668 | const struct lm_lockops *ls_ops; |
f057f6cd SW |
669 | dlm_lockspace_t *ls_dlm; |
670 | ||
e0c2a9aa DT |
671 | int ls_recover_jid_done; /* These two are deprecated, */ |
672 | int ls_recover_jid_status; /* used previously by gfs_controld */ | |
673 | ||
674 | struct dlm_lksb ls_mounted_lksb; /* mounted_lock */ | |
675 | struct dlm_lksb ls_control_lksb; /* control_lock */ | |
676 | char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */ | |
677 | struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */ | |
57c7310b | 678 | char *ls_lvb_bits; |
e0c2a9aa DT |
679 | |
680 | spinlock_t ls_recover_spin; /* protects following fields */ | |
681 | unsigned long ls_recover_flags; /* DFL_ */ | |
682 | uint32_t ls_recover_mount; /* gen in first recover_done cb */ | |
683 | uint32_t ls_recover_start; /* gen in last recover_done cb */ | |
684 | uint32_t ls_recover_block; /* copy recover_start in last recover_prep */ | |
685 | uint32_t ls_recover_size; /* size of recover_submit, recover_result */ | |
686 | uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */ | |
687 | uint32_t *ls_recover_result; /* result of last jid recovery */ | |
f057f6cd SW |
688 | }; |
689 | ||
a245769f SW |
690 | struct gfs2_pcpu_lkstats { |
691 | /* One struct for each glock type */ | |
692 | struct gfs2_lkstats lkstats[10]; | |
693 | }; | |
694 | ||
b3b94faa DT |
695 | struct gfs2_sbd { |
696 | struct super_block *sd_vfs; | |
a245769f | 697 | struct gfs2_pcpu_lkstats __percpu *sd_lkstats; |
b3b94faa DT |
698 | struct kobject sd_kobj; |
699 | unsigned long sd_flags; /* SDF_... */ | |
f50dfaf7 | 700 | struct gfs2_sb_host sd_sb; |
b3b94faa DT |
701 | |
702 | /* Constants computed on mount */ | |
703 | ||
cd915493 SW |
704 | u32 sd_fsb2bb; |
705 | u32 sd_fsb2bb_shift; | |
706 | u32 sd_diptrs; /* Number of pointers in a dinode */ | |
707 | u32 sd_inptrs; /* Number of pointers in a indirect block */ | |
708 | u32 sd_jbsize; /* Size of a journaled data block */ | |
709 | u32 sd_hash_bsize; /* sizeof(exhash block) */ | |
710 | u32 sd_hash_bsize_shift; | |
711 | u32 sd_hash_ptrs; /* Number of pointers in a hash block */ | |
712 | u32 sd_qc_per_block; | |
a68a0a35 | 713 | u32 sd_blocks_per_bitmap; |
cd915493 SW |
714 | u32 sd_max_dirres; /* Max blocks needed to add a directory entry */ |
715 | u32 sd_max_height; /* Max height of a file's metadata tree */ | |
ecc30c79 | 716 | u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1]; |
471f3db2 | 717 | u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */ |
b3b94faa DT |
718 | |
719 | struct gfs2_args sd_args; /* Mount arguments */ | |
720 | struct gfs2_tune sd_tune; /* Filesystem tuning structure */ | |
721 | ||
722 | /* Lock Stuff */ | |
723 | ||
724 | struct lm_lockstruct sd_lockstruct; | |
b3b94faa DT |
725 | struct gfs2_holder sd_live_gh; |
726 | struct gfs2_glock *sd_rename_gl; | |
24972557 | 727 | struct gfs2_glock *sd_freeze_gl; |
2e60d768 | 728 | struct work_struct sd_freeze_work; |
e402746a SW |
729 | wait_queue_head_t sd_glock_wait; |
730 | atomic_t sd_glock_disposal; | |
3942ae53 | 731 | struct completion sd_locking_init; |
fd95e81c | 732 | struct completion sd_wdack; |
e0c2a9aa | 733 | struct delayed_work sd_control_work; |
b3b94faa DT |
734 | |
735 | /* Inode Stuff */ | |
736 | ||
9b8df98f SW |
737 | struct dentry *sd_master_dir; |
738 | struct dentry *sd_root_dir; | |
739 | ||
f42faf4f | 740 | struct inode *sd_jindex; |
f42faf4f | 741 | struct inode *sd_statfs_inode; |
f42faf4f | 742 | struct inode *sd_sc_inode; |
f42faf4f SW |
743 | struct inode *sd_qc_inode; |
744 | struct inode *sd_rindex; | |
745 | struct inode *sd_quota_inode; | |
b3b94faa | 746 | |
b3b94faa DT |
747 | /* StatFS stuff */ |
748 | ||
749 | spinlock_t sd_statfs_spin; | |
bd209cc0 AV |
750 | struct gfs2_statfs_change_host sd_statfs_master; |
751 | struct gfs2_statfs_change_host sd_statfs_local; | |
3d3c10f2 | 752 | int sd_statfs_force_sync; |
b3b94faa DT |
753 | |
754 | /* Resource group stuff */ | |
755 | ||
cf45b752 | 756 | int sd_rindex_uptodate; |
b3b94faa | 757 | spinlock_t sd_rindex_spin; |
7c9ca621 | 758 | struct rb_root sd_rindex_tree; |
b3b94faa | 759 | unsigned int sd_rgrps; |
3921120e | 760 | unsigned int sd_max_rg_data; |
b3b94faa DT |
761 | |
762 | /* Journal index stuff */ | |
763 | ||
764 | struct list_head sd_jindex_list; | |
765 | spinlock_t sd_jindex_spin; | |
f55ab26a | 766 | struct mutex sd_jindex_mutex; |
b3b94faa | 767 | unsigned int sd_journals; |
b3b94faa DT |
768 | |
769 | struct gfs2_jdesc *sd_jdesc; | |
770 | struct gfs2_holder sd_journal_gh; | |
771 | struct gfs2_holder sd_jinode_gh; | |
772 | ||
b3b94faa | 773 | struct gfs2_holder sd_sc_gh; |
b3b94faa DT |
774 | struct gfs2_holder sd_qc_gh; |
775 | ||
0e48e055 BP |
776 | struct completion sd_journal_ready; |
777 | ||
b3b94faa DT |
778 | /* Daemon stuff */ |
779 | ||
b3b94faa DT |
780 | struct task_struct *sd_logd_process; |
781 | struct task_struct *sd_quotad_process; | |
b3b94faa | 782 | |
b3b94faa DT |
783 | /* Quota stuff */ |
784 | ||
785 | struct list_head sd_quota_list; | |
786 | atomic_t sd_quota_count; | |
f55ab26a | 787 | struct mutex sd_quota_mutex; |
e46c772d | 788 | struct mutex sd_quota_sync_mutex; |
37b2c837 | 789 | wait_queue_head_t sd_quota_wait; |
813e0c46 SW |
790 | struct list_head sd_trunc_list; |
791 | spinlock_t sd_trunc_lock; | |
b3b94faa DT |
792 | |
793 | unsigned int sd_quota_slots; | |
ee2411a8 | 794 | unsigned long *sd_quota_bitmap; |
2d9e7230 | 795 | spinlock_t sd_bitmap_lock; |
b3b94faa | 796 | |
cd915493 | 797 | u64 sd_quota_sync_gen; |
b3b94faa DT |
798 | |
799 | /* Log stuff */ | |
800 | ||
70d4ee94 SW |
801 | struct address_space sd_aspace; |
802 | ||
b3b94faa | 803 | spinlock_t sd_log_lock; |
b3b94faa | 804 | |
16ca9412 | 805 | struct gfs2_trans *sd_log_tr; |
b3b94faa | 806 | unsigned int sd_log_blks_reserved; |
2e95e3f6 | 807 | int sd_log_commited_revoke; |
b3b94faa | 808 | |
5e687eac | 809 | atomic_t sd_log_pinned; |
b3b94faa | 810 | unsigned int sd_log_num_revoke; |
f42faf4f | 811 | |
b3b94faa | 812 | struct list_head sd_log_le_revoke; |
d7b616e2 | 813 | struct list_head sd_log_le_ordered; |
45138990 | 814 | spinlock_t sd_ordered_lock; |
b3b94faa | 815 | |
5e687eac BM |
816 | atomic_t sd_log_thresh1; |
817 | atomic_t sd_log_thresh2; | |
fd041f0b | 818 | atomic_t sd_log_blks_free; |
f07b3520 | 819 | atomic_t sd_log_blks_needed; |
5e687eac BM |
820 | wait_queue_head_t sd_log_waitq; |
821 | wait_queue_head_t sd_logd_waitq; | |
b3b94faa | 822 | |
cd915493 | 823 | u64 sd_log_sequence; |
b3b94faa DT |
824 | unsigned int sd_log_head; |
825 | unsigned int sd_log_tail; | |
b3b94faa DT |
826 | int sd_log_idle; |
827 | ||
484adff8 | 828 | struct rw_semaphore sd_log_flush_lock; |
16615be1 | 829 | atomic_t sd_log_in_flight; |
e8c92ed7 | 830 | struct bio *sd_log_bio; |
16615be1 | 831 | wait_queue_head_t sd_log_flush_wait; |
942b0cdd | 832 | int sd_log_error; |
b3b94faa | 833 | |
2e60d768 BM |
834 | atomic_t sd_reserving_log; |
835 | wait_queue_head_t sd_reserving_log_wait; | |
836 | ||
b3b94faa | 837 | unsigned int sd_log_flush_head; |
b3b94faa | 838 | |
d6a079e8 | 839 | spinlock_t sd_ail_lock; |
b3b94faa DT |
840 | struct list_head sd_ail1_list; |
841 | struct list_head sd_ail2_list; | |
b3b94faa | 842 | |
b3b94faa | 843 | /* For quiescing the filesystem */ |
b3b94faa | 844 | struct gfs2_holder sd_freeze_gh; |
2e60d768 BM |
845 | atomic_t sd_freeze_state; |
846 | struct mutex sd_freeze_mutex; | |
b3b94faa | 847 | |
561b7969 | 848 | char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2]; |
b3b94faa DT |
849 | char sd_table_name[GFS2_FSNAME_LEN]; |
850 | char sd_proto_name[GFS2_FSNAME_LEN]; | |
851 | ||
852 | /* Debugging crud */ | |
853 | ||
854 | unsigned long sd_last_warning; | |
5f882096 | 855 | struct dentry *debugfs_dir; /* debugfs directory */ |
b3b94faa DT |
856 | }; |
857 | ||
a245769f SW |
858 | static inline void gfs2_glstats_inc(struct gfs2_glock *gl, int which) |
859 | { | |
860 | gl->gl_stats.stats[which]++; | |
861 | } | |
862 | ||
863 | static inline void gfs2_sbstats_inc(const struct gfs2_glock *gl, int which) | |
864 | { | |
15562c43 | 865 | const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd; |
a245769f SW |
866 | preempt_disable(); |
867 | this_cpu_ptr(sdp->sd_lkstats)->lkstats[gl->gl_name.ln_type].stats[which]++; | |
868 | preempt_enable(); | |
869 | } | |
870 | ||
6f6597ba AG |
871 | extern struct gfs2_rgrpd *gfs2_glock2rgrp(struct gfs2_glock *gl); |
872 | ||
235628c5 AG |
873 | static inline unsigned gfs2_max_stuffed_size(const struct gfs2_inode *ip) |
874 | { | |
875 | return GFS2_SB(&ip->i_inode)->sd_sb.sb_bsize - sizeof(struct gfs2_dinode); | |
876 | } | |
877 | ||
b3b94faa DT |
878 | #endif /* __INCORE_DOT_H__ */ |
879 |