Commit | Line | Data |
---|---|---|
7336d0e6 | 1 | /* SPDX-License-Identifier: GPL-2.0-only */ |
b3b94faa DT |
2 | /* |
3 | * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved. | |
29d38cd1 | 4 | * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved. |
b3b94faa DT |
5 | */ |
6 | ||
7 | #ifndef __INCORE_DOT_H__ | |
8 | #define __INCORE_DOT_H__ | |
9 | ||
f2f7ba52 | 10 | #include <linux/fs.h> |
57cc7215 | 11 | #include <linux/kobject.h> |
c4f68a13 | 12 | #include <linux/workqueue.h> |
f057f6cd SW |
13 | #include <linux/dlm.h> |
14 | #include <linux/buffer_head.h> | |
bc015cb8 SW |
15 | #include <linux/rcupdate.h> |
16 | #include <linux/rculist_bl.h> | |
3942ae53 | 17 | #include <linux/completion.h> |
7c9ca621 | 18 | #include <linux/rbtree.h> |
a245769f SW |
19 | #include <linux/ktime.h> |
20 | #include <linux/percpu.h> | |
e66cf161 | 21 | #include <linux/lockref.h> |
88ffbf3e | 22 | #include <linux/rhashtable.h> |
9e514605 | 23 | #include <linux/mutex.h> |
f2f7ba52 | 24 | |
b3b94faa DT |
25 | #define DIO_WAIT 0x00000010 |
26 | #define DIO_METADATA 0x00000020 | |
b3b94faa DT |
27 | |
28 | struct gfs2_log_operations; | |
c0752aa7 | 29 | struct gfs2_bufdata; |
b3b94faa DT |
30 | struct gfs2_holder; |
31 | struct gfs2_glock; | |
b3b94faa | 32 | struct gfs2_quota_data; |
b3b94faa | 33 | struct gfs2_trans; |
b3b94faa | 34 | struct gfs2_jdesc; |
b3b94faa | 35 | struct gfs2_sbd; |
f057f6cd | 36 | struct lm_lockops; |
b3b94faa DT |
37 | |
38 | typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret); | |
39 | ||
bb8d8a6f SW |
40 | struct gfs2_log_header_host { |
41 | u64 lh_sequence; /* Sequence number of this transaction */ | |
42 | u32 lh_flags; /* GFS2_LOG_HEAD_... */ | |
43 | u32 lh_tail; /* Block number of log tail */ | |
44 | u32 lh_blkno; | |
73092698 AD |
45 | |
46 | s64 lh_local_total; | |
47 | s64 lh_local_free; | |
48 | s64 lh_local_dinodes; | |
bb8d8a6f SW |
49 | }; |
50 | ||
b3b94faa DT |
51 | /* |
52 | * Structure of operations that are associated with each | |
53 | * type of element in the log. | |
54 | */ | |
55 | ||
56 | struct gfs2_log_operations { | |
d69a3c65 | 57 | void (*lo_before_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr); |
16ca9412 | 58 | void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr); |
b3b94faa | 59 | void (*lo_before_scan) (struct gfs2_jdesc *jd, |
55167622 | 60 | struct gfs2_log_header_host *head, int pass); |
b3b94faa DT |
61 | int (*lo_scan_elements) (struct gfs2_jdesc *jd, unsigned int start, |
62 | struct gfs2_log_descriptor *ld, __be64 *ptr, | |
63 | int pass); | |
64 | void (*lo_after_scan) (struct gfs2_jdesc *jd, int error, int pass); | |
b09e593d | 65 | const char *lo_name; |
b3b94faa DT |
66 | }; |
67 | ||
60a0b8f9 SW |
68 | #define GBF_FULL 1 |
69 | ||
dffe12a8 BP |
70 | /** |
71 | * Clone bitmaps (bi_clone): | |
72 | * | |
73 | * - When a block is freed, we remember the previous state of the block in the | |
74 | * clone bitmap, and only mark the block as free in the real bitmap. | |
75 | * | |
76 | * - When looking for a block to allocate, we check for a free block in the | |
77 | * clone bitmap, and if no clone bitmap exists, in the real bitmap. | |
78 | * | |
79 | * - For allocating a block, we mark it as allocated in the real bitmap, and if | |
80 | * a clone bitmap exists, also in the clone bitmap. | |
81 | * | |
82 | * - At the end of a log_flush, we copy the real bitmap into the clone bitmap | |
83 | * to make the clone bitmap reflect the current allocation state. | |
84 | * (Alternatively, we could remove the clone bitmap.) | |
85 | * | |
86 | * The clone bitmaps are in-core only, and is never written to disk. | |
87 | * | |
88 | * These steps ensure that blocks which have been freed in a transaction cannot | |
89 | * be reallocated in that same transaction. | |
90 | */ | |
b3b94faa DT |
91 | struct gfs2_bitmap { |
92 | struct buffer_head *bi_bh; | |
93 | char *bi_clone; | |
60a0b8f9 | 94 | unsigned long bi_flags; |
cd915493 SW |
95 | u32 bi_offset; |
96 | u32 bi_start; | |
281b4952 | 97 | u32 bi_bytes; |
7e230f57 | 98 | u32 bi_blocks; |
b3b94faa DT |
99 | }; |
100 | ||
101 | struct gfs2_rgrpd { | |
7c9ca621 | 102 | struct rb_node rd_node; /* Link with superblock */ |
b3b94faa | 103 | struct gfs2_glock *rd_gl; /* Glock for this rgrp */ |
bb8d8a6f SW |
104 | u64 rd_addr; /* grp block disk address */ |
105 | u64 rd_data0; /* first data location */ | |
106 | u32 rd_length; /* length of rgrp header in fs blocks */ | |
107 | u32 rd_data; /* num of data blocks in rgrp */ | |
108 | u32 rd_bitbytes; /* number of bytes in data bitmaps */ | |
cfc8b549 | 109 | u32 rd_free; |
07974d2a | 110 | u32 rd_requested; /* number of blocks in rd_rstree */ |
725d0e9d | 111 | u32 rd_reserved; /* number of reserved blocks */ |
73f74948 SW |
112 | u32 rd_free_clone; |
113 | u32 rd_dinodes; | |
d8b71f73 | 114 | u64 rd_igeneration; |
b3b94faa | 115 | struct gfs2_bitmap *rd_bits; |
b3b94faa | 116 | struct gfs2_sbd *rd_sbd; |
90306c41 | 117 | struct gfs2_rgrp_lvb *rd_rgl; |
73f74948 | 118 | u32 rd_last_alloc; |
1ce97e56 | 119 | u32 rd_flags; |
5ea5050c | 120 | u32 rd_extfail_pt; /* extent failure point */ |
09010978 | 121 | #define GFS2_RDF_CHECK 0x10000000 /* check for unlinked inodes */ |
09010978 | 122 | #define GFS2_RDF_ERROR 0x40000000 /* error in rg */ |
0e27c18c | 123 | #define GFS2_RDF_PREFERRED 0x80000000 /* This rgrp is preferred */ |
09010978 | 124 | #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */ |
8e2e0047 | 125 | spinlock_t rd_rsspin; /* protects reservation related vars */ |
9e514605 | 126 | struct mutex rd_mutex; |
8e2e0047 | 127 | struct rb_root rd_rstree; /* multi-block reservation tree */ |
b3b94faa DT |
128 | }; |
129 | ||
130 | enum gfs2_state_bits { | |
131 | BH_Pinned = BH_PrivateStart, | |
f42faf4f | 132 | BH_Escaped = BH_PrivateStart + 1, |
b3b94faa DT |
133 | }; |
134 | ||
135 | BUFFER_FNS(Pinned, pinned) | |
136 | TAS_BUFFER_FNS(Pinned, pinned) | |
f42faf4f SW |
137 | BUFFER_FNS(Escaped, escaped) |
138 | TAS_BUFFER_FNS(Escaped, escaped) | |
b3b94faa DT |
139 | |
140 | struct gfs2_bufdata { | |
141 | struct buffer_head *bd_bh; | |
142 | struct gfs2_glock *bd_gl; | |
c50b91c4 | 143 | u64 bd_blkno; |
82e86087 | 144 | |
c0752aa7 | 145 | struct list_head bd_list; |
b3b94faa | 146 | |
16ca9412 | 147 | struct gfs2_trans *bd_tr; |
b3b94faa DT |
148 | struct list_head bd_ail_st_list; |
149 | struct list_head bd_ail_gl_list; | |
150 | }; | |
151 | ||
f057f6cd SW |
152 | /* |
153 | * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a | |
154 | * prefix of lock_dlm_ gets awkward. | |
155 | */ | |
156 | ||
157 | #define GDLM_STRNAME_BYTES 25 | |
158 | #define GDLM_LVB_SIZE 32 | |
159 | ||
e0c2a9aa DT |
160 | /* |
161 | * ls_recover_flags: | |
162 | * | |
163 | * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been | |
164 | * held by failed nodes whose journals need recovery. Those locks should | |
165 | * only be used for journal recovery until the journal recovery is done. | |
166 | * This is set by the dlm recover_prep callback and cleared by the | |
167 | * gfs2_control thread when journal recovery is complete. To avoid | |
168 | * races between recover_prep setting and gfs2_control clearing, recover_spin | |
169 | * is held while changing this bit and reading/writing recover_block | |
170 | * and recover_start. | |
171 | * | |
172 | * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used. | |
173 | * | |
174 | * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing | |
175 | * recovery of all journals before allowing other nodes to mount the fs. | |
176 | * This is cleared when FIRST_MOUNT_DONE is set. | |
177 | * | |
178 | * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished | |
179 | * recovery of all journals, and now allows other nodes to mount the fs. | |
180 | * | |
181 | * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared | |
182 | * BLOCK_LOCKS for the first time. The gfs2_control thread should now | |
183 | * control clearing BLOCK_LOCKS for further recoveries. | |
184 | * | |
185 | * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq. | |
186 | * | |
187 | * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep() | |
188 | * and recover_done(), i.e. set while recover_block == recover_start. | |
189 | */ | |
190 | ||
f057f6cd SW |
191 | enum { |
192 | DFL_BLOCK_LOCKS = 0, | |
e0c2a9aa DT |
193 | DFL_NO_DLM_OPS = 1, |
194 | DFL_FIRST_MOUNT = 2, | |
195 | DFL_FIRST_MOUNT_DONE = 3, | |
196 | DFL_MOUNT_DONE = 4, | |
197 | DFL_UNMOUNT = 5, | |
198 | DFL_DLM_RECOVERY = 6, | |
f057f6cd SW |
199 | }; |
200 | ||
972b044e AG |
201 | /* |
202 | * We are using struct lm_lockname as an rhashtable key. Avoid holes within | |
203 | * the struct; padding at the end is fine. | |
204 | */ | |
f057f6cd SW |
205 | struct lm_lockname { |
206 | u64 ln_number; | |
972b044e | 207 | struct gfs2_sbd *ln_sbd; |
f057f6cd | 208 | unsigned int ln_type; |
972b044e | 209 | }; |
f057f6cd SW |
210 | |
211 | #define lm_name_equal(name1, name2) \ | |
15562c43 BP |
212 | (((name1)->ln_number == (name2)->ln_number) && \ |
213 | ((name1)->ln_type == (name2)->ln_type) && \ | |
214 | ((name1)->ln_sbd == (name2)->ln_sbd)) | |
f057f6cd SW |
215 | |
216 | ||
b3b94faa | 217 | struct gfs2_glock_operations { |
1c634f94 | 218 | int (*go_sync) (struct gfs2_glock *gl); |
f68effb3 | 219 | int (*go_xmote_bh)(struct gfs2_glock *gl); |
1a14d3a6 | 220 | void (*go_inval) (struct gfs2_glock *gl, int flags); |
97cc1025 | 221 | int (*go_demote_ok) (const struct gfs2_glock *gl); |
5f38a4d3 | 222 | int (*go_instantiate) (struct gfs2_glock *gl); |
86c30a01 | 223 | int (*go_held)(struct gfs2_holder *gh); |
3792ce97 BP |
224 | void (*go_dump)(struct seq_file *seq, struct gfs2_glock *gl, |
225 | const char *fs_id_buf); | |
81ffbf65 | 226 | void (*go_callback)(struct gfs2_glock *gl, bool remote); |
601ef0d5 | 227 | void (*go_free)(struct gfs2_glock *gl); |
515b269d | 228 | const int go_subclass; |
8fb4b536 | 229 | const int go_type; |
009d8518 | 230 | const unsigned long go_flags; |
a72d2401 BP |
231 | #define GLOF_ASPACE 1 /* address space attached */ |
232 | #define GLOF_LVB 2 /* Lock Value Block attached */ | |
233 | #define GLOF_LRU 4 /* LRU managed */ | |
234 | #define GLOF_NONDISK 8 /* not I/O related */ | |
b3b94faa DT |
235 | }; |
236 | ||
a245769f SW |
237 | enum { |
238 | GFS2_LKS_SRTT = 0, /* Non blocking smoothed round trip time */ | |
239 | GFS2_LKS_SRTTVAR = 1, /* Non blocking smoothed variance */ | |
240 | GFS2_LKS_SRTTB = 2, /* Blocking smoothed round trip time */ | |
241 | GFS2_LKS_SRTTVARB = 3, /* Blocking smoothed variance */ | |
242 | GFS2_LKS_SIRT = 4, /* Smoothed Inter-request time */ | |
243 | GFS2_LKS_SIRTVAR = 5, /* Smoothed Inter-request variance */ | |
244 | GFS2_LKS_DCOUNT = 6, /* Count of dlm requests */ | |
245 | GFS2_LKS_QCOUNT = 7, /* Count of gfs2_holder queues */ | |
246 | GFS2_NR_LKSTATS | |
247 | }; | |
248 | ||
249 | struct gfs2_lkstats { | |
4d207133 | 250 | u64 stats[GFS2_NR_LKSTATS]; |
a245769f SW |
251 | }; |
252 | ||
b3b94faa | 253 | enum { |
b3b94faa | 254 | /* States */ |
6802e340 | 255 | HIF_HOLDER = 6, /* Set for gh that "holds" the glock */ |
fee852e3 | 256 | HIF_WAIT = 10, |
b3b94faa DT |
257 | }; |
258 | ||
259 | struct gfs2_holder { | |
260 | struct list_head gh_list; | |
261 | ||
262 | struct gfs2_glock *gh_gl; | |
b1e058da | 263 | struct pid *gh_owner_pid; |
b58bf407 BP |
264 | u16 gh_flags; |
265 | u16 gh_state; | |
b3b94faa DT |
266 | |
267 | int gh_error; | |
6802e340 | 268 | unsigned long gh_iflags; /* HIF_... */ |
d0dc80db | 269 | unsigned long gh_ip; |
b3b94faa DT |
270 | }; |
271 | ||
a937cca2 JK |
272 | /* Number of quota types we support */ |
273 | #define GFS2_MAXQUOTAS 2 | |
274 | ||
b54e9a0b BP |
275 | struct gfs2_qadata { /* quota allocation data */ |
276 | /* Quota stuff */ | |
277 | struct gfs2_quota_data *qa_qd[2 * GFS2_MAXQUOTAS]; | |
278 | struct gfs2_holder qa_qd_ghs[2 * GFS2_MAXQUOTAS]; | |
279 | unsigned int qa_qd_num; | |
2fba46a0 | 280 | int qa_ref; |
b54e9a0b BP |
281 | }; |
282 | ||
8e2e0047 BP |
283 | /* Resource group multi-block reservation, in order of appearance: |
284 | ||
285 | Step 1. Function prepares to write, allocates a mb, sets the size hint. | |
286 | Step 2. User calls inplace_reserve to target an rgrp, sets the rgrp info | |
287 | Step 3. Function get_local_rgrp locks the rgrp, determines which bits to use | |
288 | Step 4. Bits are assigned from the rgrp based on either the reservation | |
289 | or wherever it can. | |
290 | */ | |
291 | ||
292 | struct gfs2_blkreserv { | |
c65b76b8 AG |
293 | struct rb_node rs_node; /* node within rd_rstree */ |
294 | struct gfs2_rgrpd *rs_rgd; | |
07974d2a AG |
295 | u64 rs_start; |
296 | u32 rs_requested; | |
725d0e9d | 297 | u32 rs_reserved; /* number of reserved blocks */ |
8e2e0047 BP |
298 | }; |
299 | ||
7b9cff46 SW |
300 | /* |
301 | * Allocation parameters | |
302 | * @target: The number of blocks we'd ideally like to allocate | |
303 | * @aflags: The flags (e.g. Orlov flag) | |
304 | * | |
305 | * The intent is to gradually expand this structure over time in | |
306 | * order to give more information, e.g. alignment, min extent size | |
307 | * to the allocation code. | |
308 | */ | |
309 | struct gfs2_alloc_parms { | |
b8fbf471 | 310 | u64 target; |
25435e5e | 311 | u32 min_target; |
7b9cff46 | 312 | u32 aflags; |
25435e5e | 313 | u64 allowed; |
7b9cff46 SW |
314 | }; |
315 | ||
b3b94faa | 316 | enum { |
6802e340 | 317 | GLF_LOCK = 1, |
f2e70d8f | 318 | GLF_INSTANTIATE_NEEDED = 2, /* needs instantiate */ |
6802e340 SW |
319 | GLF_DEMOTE = 3, |
320 | GLF_PENDING_DEMOTE = 4, | |
321 | GLF_DEMOTE_IN_PROGRESS = 5, | |
322 | GLF_DIRTY = 6, | |
323 | GLF_LFLUSH = 7, | |
324 | GLF_INVALIDATE_IN_PROGRESS = 8, | |
325 | GLF_REPLY_PENDING = 9, | |
f057f6cd SW |
326 | GLF_INITIAL = 10, |
327 | GLF_FROZEN = 11, | |
f2e70d8f | 328 | GLF_INSTANTIATE_IN_PROG = 12, /* instantiate happening now */ |
627c10b7 SW |
329 | GLF_LRU = 13, |
330 | GLF_OBJECT = 14, /* Used only for tracing */ | |
a245769f | 331 | GLF_BLOCKING = 15, |
3056dc46 | 332 | GLF_FREEING = 16, /* Wait for glock to be freed */ |
f0e56edc AG |
333 | GLF_TRY_TO_EVICT = 17, /* iopen glocks only */ |
334 | GLF_VERIFY_EVICT = 18, /* iopen glocks only */ | |
b3b94faa DT |
335 | }; |
336 | ||
337 | struct gfs2_glock { | |
b3b94faa DT |
338 | unsigned long gl_flags; /* GLF_... */ |
339 | struct lm_lockname gl_name; | |
b3b94faa | 340 | |
e66cf161 | 341 | struct lockref gl_lockref; |
b3b94faa | 342 | |
f3dd1649 | 343 | /* State fields protected by gl_lockref.lock */ |
47a25380 SW |
344 | unsigned int gl_state:2, /* Current state */ |
345 | gl_target:2, /* Target state */ | |
346 | gl_demote_state:2, /* State requested by remote node */ | |
347 | gl_req:2, /* State in last dlm request */ | |
348 | gl_reply:8; /* Last reply from the dlm */ | |
349 | ||
3b8249f6 | 350 | unsigned long gl_demote_time; /* time of first demote request */ |
7cf8dcd3 | 351 | long gl_hold_time; |
b3b94faa | 352 | struct list_head gl_holders; |
b3b94faa | 353 | |
8fb4b536 | 354 | const struct gfs2_glock_operations *gl_ops; |
a245769f SW |
355 | ktime_t gl_dstamp; |
356 | struct gfs2_lkstats gl_stats; | |
f057f6cd | 357 | struct dlm_lksb gl_lksb; |
c4f68a13 | 358 | unsigned long gl_tchange; |
b3b94faa DT |
359 | void *gl_object; |
360 | ||
97cc1025 | 361 | struct list_head gl_lru; |
b3b94faa DT |
362 | struct list_head gl_ail_list; |
363 | atomic_t gl_ail_count; | |
638803d4 | 364 | atomic_t gl_revokes; |
c4f68a13 | 365 | struct delayed_work gl_work; |
23cfb0c3 BP |
366 | /* For iopen glocks only */ |
367 | struct { | |
368 | struct delayed_work gl_delete; | |
369 | u64 gl_no_formal_ino; | |
7005c3e4 | 370 | }; |
961ae1d8 | 371 | struct rcu_head gl_rcu; |
88ffbf3e | 372 | struct rhash_head gl_node; |
b3b94faa DT |
373 | }; |
374 | ||
b3b94faa | 375 | enum { |
b3b94faa | 376 | GIF_QD_LOCKED = 1, |
40ac218f | 377 | GIF_ALLOC_FAILED = 2, |
b3b94faa | 378 | GIF_SW_PAGED = 3, |
05978803 | 379 | GIF_FREE_VFS_INODE = 5, |
4fd1a579 | 380 | GIF_GLOP_PENDING = 6, |
8c7b9262 | 381 | GIF_DEFERRED_DELETE = 7, |
b3b94faa DT |
382 | }; |
383 | ||
384 | struct gfs2_inode { | |
320dd101 | 385 | struct inode i_inode; |
dbb7cae2 SW |
386 | u64 i_no_addr; |
387 | u64 i_no_formal_ino; | |
bcf0b5b3 | 388 | u64 i_generation; |
3767ac21 | 389 | u64 i_eattr; |
b3b94faa | 390 | unsigned long i_flags; /* GIF_... */ |
1b223f70 | 391 | struct gfs2_glock *i_gl; |
b3b94faa | 392 | struct gfs2_holder i_iopen_gh; |
b54e9a0b | 393 | struct gfs2_qadata *i_qadata; /* quota allocation data */ |
21f09c43 | 394 | struct gfs2_holder i_rgd_gh; |
a097dc7e | 395 | struct gfs2_blkreserv i_res; /* rgrp multi-block reservation */ |
ce276b06 | 396 | u64 i_goal; /* goal block for allocations */ |
21f09c43 | 397 | atomic_t i_sizehint; /* hint of the write size */ |
b3b94faa | 398 | struct rw_semaphore i_rw_mutex; |
45138990 | 399 | struct list_head i_ordered; |
17d539f0 | 400 | __be64 *i_hash_cache; |
ad6203f2 | 401 | u32 i_entries; |
383f01fb | 402 | u32 i_diskflags; |
ecc30c79 | 403 | u8 i_height; |
9a004508 | 404 | u8 i_depth; |
c8d57703 | 405 | u16 i_rahead; |
b3b94faa DT |
406 | }; |
407 | ||
feaa7bba SW |
408 | /* |
409 | * Since i_inode is the first element of struct gfs2_inode, | |
410 | * this is effectively a cast. | |
411 | */ | |
320dd101 SW |
412 | static inline struct gfs2_inode *GFS2_I(struct inode *inode) |
413 | { | |
414 | return container_of(inode, struct gfs2_inode, i_inode); | |
415 | } | |
416 | ||
bf36a713 | 417 | static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode) |
feaa7bba SW |
418 | { |
419 | return inode->i_sb->s_fs_info; | |
420 | } | |
421 | ||
b3b94faa | 422 | struct gfs2_file { |
f55ab26a | 423 | struct mutex f_fl_mutex; |
b3b94faa | 424 | struct gfs2_holder f_fl_gh; |
b3b94faa DT |
425 | }; |
426 | ||
b3b94faa DT |
427 | struct gfs2_revoke_replay { |
428 | struct list_head rr_list; | |
cd915493 | 429 | u64 rr_blkno; |
b3b94faa DT |
430 | unsigned int rr_where; |
431 | }; | |
432 | ||
b3b94faa | 433 | enum { |
b3b94faa DT |
434 | QDF_CHANGE = 1, |
435 | QDF_LOCKED = 2, | |
662e3a55 | 436 | QDF_REFRESH = 3, |
9cde2898 | 437 | QDF_QMSG_QUIET = 4, |
b3b94faa DT |
438 | }; |
439 | ||
440 | struct gfs2_quota_data { | |
c754fbbb | 441 | struct hlist_bl_node qd_hlist; |
b3b94faa | 442 | struct list_head qd_list; |
2147dbfd | 443 | struct kqid qd_id; |
c754fbbb | 444 | struct gfs2_sbd *qd_sbd; |
9b9f039d | 445 | struct lockref qd_lockref; |
2147dbfd | 446 | struct list_head qd_lru; |
c754fbbb | 447 | unsigned qd_hash; |
b3b94faa | 448 | |
b3b94faa DT |
449 | unsigned long qd_flags; /* QDF_... */ |
450 | ||
cd915493 SW |
451 | s64 qd_change; |
452 | s64 qd_change_sync; | |
b3b94faa DT |
453 | |
454 | unsigned int qd_slot; | |
455 | unsigned int qd_slot_count; | |
456 | ||
457 | struct buffer_head *qd_bh; | |
458 | struct gfs2_quota_change *qd_bh_qc; | |
459 | unsigned int qd_bh_count; | |
460 | ||
461 | struct gfs2_glock *qd_gl; | |
462 | struct gfs2_quota_lvb qd_qb; | |
463 | ||
cd915493 | 464 | u64 qd_sync_gen; |
b3b94faa | 465 | unsigned long qd_last_warn; |
c754fbbb | 466 | struct rcu_head qd_rcu; |
b3b94faa DT |
467 | }; |
468 | ||
9862ca05 BP |
469 | enum { |
470 | TR_TOUCHED = 1, | |
471 | TR_ATTACHED = 2, | |
c968f578 | 472 | TR_ONSTACK = 3, |
9862ca05 BP |
473 | }; |
474 | ||
b3b94faa | 475 | struct gfs2_trans { |
d0dc80db | 476 | unsigned long tr_ip; |
b3b94faa DT |
477 | |
478 | unsigned int tr_blocks; | |
479 | unsigned int tr_revokes; | |
480 | unsigned int tr_reserved; | |
9862ca05 | 481 | unsigned long tr_flags; |
b3b94faa | 482 | |
b3b94faa | 483 | unsigned int tr_num_buf_new; |
2332c443 | 484 | unsigned int tr_num_databuf_new; |
b3b94faa | 485 | unsigned int tr_num_buf_rm; |
2332c443 | 486 | unsigned int tr_num_databuf_rm; |
b3b94faa | 487 | unsigned int tr_num_revoke; |
b3b94faa | 488 | |
16ca9412 | 489 | struct list_head tr_list; |
d69a3c65 SW |
490 | struct list_head tr_databuf; |
491 | struct list_head tr_buf; | |
b3b94faa | 492 | |
16ca9412 BM |
493 | unsigned int tr_first; |
494 | struct list_head tr_ail1_list; | |
495 | struct list_head tr_ail2_list; | |
b3b94faa DT |
496 | }; |
497 | ||
da6dd40d | 498 | struct gfs2_journal_extent { |
b50f227b | 499 | struct list_head list; |
da6dd40d BP |
500 | |
501 | unsigned int lblock; /* First logical block */ | |
502 | u64 dblock; /* First disk block */ | |
503 | u64 blocks; | |
504 | }; | |
505 | ||
b3b94faa DT |
506 | struct gfs2_jdesc { |
507 | struct list_head jd_list; | |
da6dd40d | 508 | struct list_head extent_list; |
b50f227b | 509 | unsigned int nr_extents; |
6ecd7c2d | 510 | struct work_struct jd_work; |
7359a19c | 511 | struct inode *jd_inode; |
82218943 | 512 | struct bio *jd_log_bio; |
fe64d517 SW |
513 | unsigned long jd_flags; |
514 | #define JDF_RECOVERY 1 | |
b3b94faa | 515 | unsigned int jd_jid; |
7c70b896 | 516 | u32 jd_blocks; |
376d3778 | 517 | int jd_recover_error; |
a17d758b BP |
518 | /* Replay stuff */ |
519 | ||
520 | unsigned int jd_found_blocks; | |
521 | unsigned int jd_found_revokes; | |
522 | unsigned int jd_replayed_blocks; | |
523 | ||
524 | struct list_head jd_revoke_list; | |
525 | unsigned int jd_replay_tail; | |
526 | ||
a72d2401 | 527 | u64 jd_no_addr; |
b3b94faa DT |
528 | }; |
529 | ||
bb8d8a6f SW |
530 | struct gfs2_statfs_change_host { |
531 | s64 sc_total; | |
532 | s64 sc_free; | |
533 | s64 sc_dinodes; | |
534 | }; | |
535 | ||
b3b94faa DT |
536 | #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF |
537 | #define GFS2_QUOTA_OFF 0 | |
538 | #define GFS2_QUOTA_ACCOUNT 1 | |
539 | #define GFS2_QUOTA_ON 2 | |
540 | ||
541 | #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED | |
542 | #define GFS2_DATA_WRITEBACK 1 | |
543 | #define GFS2_DATA_ORDERED 2 | |
544 | ||
d34843d0 BP |
545 | #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW |
546 | #define GFS2_ERRORS_WITHDRAW 0 | |
547 | #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */ | |
548 | #define GFS2_ERRORS_RO 2 /* place holder for future feature */ | |
549 | #define GFS2_ERRORS_PANIC 3 | |
550 | ||
b3b94faa | 551 | struct gfs2_args { |
37ec89e8 SW |
552 | char ar_lockproto[GFS2_LOCKNAME_LEN]; /* Name of the Lock Protocol */ |
553 | char ar_locktable[GFS2_LOCKNAME_LEN]; /* Name of the Lock Table */ | |
554 | char ar_hostdata[GFS2_LOCKNAME_LEN]; /* Host specific data */ | |
555 | unsigned int ar_spectator:1; /* Don't get a journal */ | |
37ec89e8 | 556 | unsigned int ar_localflocks:1; /* Let the VFS do flock|fcntl */ |
37ec89e8 | 557 | unsigned int ar_debug:1; /* Oops on errors */ |
37ec89e8 SW |
558 | unsigned int ar_posix_acl:1; /* Enable posix acls */ |
559 | unsigned int ar_quota:2; /* off/account/on */ | |
560 | unsigned int ar_suiddir:1; /* suiddir support */ | |
561 | unsigned int ar_data:2; /* ordered/writeback */ | |
562 | unsigned int ar_meta:1; /* mount metafs */ | |
f15ab561 | 563 | unsigned int ar_discard:1; /* discard requests */ |
d34843d0 | 564 | unsigned int ar_errors:2; /* errors=withdraw | panic */ |
f25934c5 | 565 | unsigned int ar_nobarrier:1; /* do not send barriers */ |
90306c41 | 566 | unsigned int ar_rgrplvb:1; /* use lvbs for rgrp info */ |
47b7ec1d | 567 | unsigned int ar_got_rgrplvb:1; /* Was the rgrplvb opt given? */ |
471f3db2 BM |
568 | unsigned int ar_loccookie:1; /* use location based readdir |
569 | cookies */ | |
1f52aa08 AP |
570 | s32 ar_commit; /* Commit interval */ |
571 | s32 ar_statfs_quantum; /* The fast statfs interval */ | |
572 | s32 ar_quota_quantum; /* The quota interval */ | |
573 | s32 ar_statfs_percent; /* The % change to force sync */ | |
b3b94faa DT |
574 | }; |
575 | ||
576 | struct gfs2_tune { | |
577 | spinlock_t gt_spin; | |
578 | ||
b3b94faa | 579 | unsigned int gt_logd_secs; |
b3b94faa | 580 | |
b3b94faa DT |
581 | unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */ |
582 | unsigned int gt_quota_scale_num; /* Numerator */ | |
583 | unsigned int gt_quota_scale_den; /* Denominator */ | |
b3b94faa | 584 | unsigned int gt_quota_quantum; /* Secs between syncs to quota file */ |
b3b94faa | 585 | unsigned int gt_new_files_jdata; |
b3b94faa | 586 | unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */ |
b3b94faa | 587 | unsigned int gt_complain_secs; |
b3b94faa DT |
588 | unsigned int gt_statfs_quantum; |
589 | unsigned int gt_statfs_slow; | |
590 | }; | |
591 | ||
b3b94faa DT |
592 | enum { |
593 | SDF_JOURNAL_CHECKED = 0, | |
594 | SDF_JOURNAL_LIVE = 1, | |
04aea0ca | 595 | SDF_WITHDRAWN = 2, |
254db57f | 596 | SDF_NOBARRIERS = 3, |
fe64d517 | 597 | SDF_NORECOVERY = 4, |
913a71d2 | 598 | SDF_DEMOTE = 5, |
ba6e9364 | 599 | SDF_NOJOURNALID = 6, |
e8ca5cc5 | 600 | SDF_RORECOVERY = 7, /* read only recovery */ |
fb6791d1 | 601 | SDF_SKIP_DLM_UNLOCK = 8, |
b066a4ee | 602 | SDF_FORCE_AIL_FLUSH = 9, |
036330c9 BP |
603 | SDF_FS_FROZEN = 10, |
604 | SDF_WITHDRAWING = 11, /* Will withdraw eventually */ | |
601ef0d5 BP |
605 | SDF_WITHDRAW_IN_PROG = 12, /* Withdraw is in progress */ |
606 | SDF_REMOTE_WITHDRAW = 13, /* Performing remote recovery */ | |
607 | SDF_WITHDRAW_RECOVERY = 14, /* Wait for journal recovery when we are | |
608 | withdrawing */ | |
1c900151 | 609 | SDF_DEACTIVATING = 15, |
b88beb9a | 610 | SDF_EVICTING = 16, |
b3b94faa DT |
611 | }; |
612 | ||
2e60d768 BM |
613 | enum gfs2_freeze_state { |
614 | SFS_UNFROZEN = 0, | |
615 | SFS_STARTING_FREEZE = 1, | |
616 | SFS_FROZEN = 2, | |
617 | }; | |
618 | ||
b3b94faa DT |
619 | #define GFS2_FSNAME_LEN 256 |
620 | ||
bb8d8a6f SW |
621 | struct gfs2_inum_host { |
622 | u64 no_formal_ino; | |
623 | u64 no_addr; | |
624 | }; | |
625 | ||
626 | struct gfs2_sb_host { | |
627 | u32 sb_magic; | |
628 | u32 sb_type; | |
bb8d8a6f SW |
629 | |
630 | u32 sb_fs_format; | |
631 | u32 sb_multihost_format; | |
632 | u32 sb_bsize; | |
633 | u32 sb_bsize_shift; | |
634 | ||
635 | struct gfs2_inum_host sb_master_dir; | |
636 | struct gfs2_inum_host sb_root_dir; | |
637 | ||
638 | char sb_lockproto[GFS2_LOCKNAME_LEN]; | |
639 | char sb_locktable[GFS2_LOCKNAME_LEN]; | |
640 | }; | |
641 | ||
f057f6cd SW |
642 | /* |
643 | * lm_mount() return values | |
644 | * | |
645 | * ls_jid - the journal ID this node should use | |
646 | * ls_first - this node is the first to mount the file system | |
647 | * ls_lockspace - lock module's context for this file system | |
648 | * ls_ops - lock module's functions | |
649 | */ | |
650 | ||
651 | struct lm_lockstruct { | |
feb47ca9 | 652 | int ls_jid; |
f057f6cd | 653 | unsigned int ls_first; |
f057f6cd | 654 | const struct lm_lockops *ls_ops; |
f057f6cd SW |
655 | dlm_lockspace_t *ls_dlm; |
656 | ||
e0c2a9aa DT |
657 | int ls_recover_jid_done; /* These two are deprecated, */ |
658 | int ls_recover_jid_status; /* used previously by gfs_controld */ | |
659 | ||
660 | struct dlm_lksb ls_mounted_lksb; /* mounted_lock */ | |
661 | struct dlm_lksb ls_control_lksb; /* control_lock */ | |
662 | char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */ | |
663 | struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */ | |
57c7310b | 664 | char *ls_lvb_bits; |
e0c2a9aa DT |
665 | |
666 | spinlock_t ls_recover_spin; /* protects following fields */ | |
667 | unsigned long ls_recover_flags; /* DFL_ */ | |
668 | uint32_t ls_recover_mount; /* gen in first recover_done cb */ | |
669 | uint32_t ls_recover_start; /* gen in last recover_done cb */ | |
670 | uint32_t ls_recover_block; /* copy recover_start in last recover_prep */ | |
671 | uint32_t ls_recover_size; /* size of recover_submit, recover_result */ | |
672 | uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */ | |
673 | uint32_t *ls_recover_result; /* result of last jid recovery */ | |
f057f6cd SW |
674 | }; |
675 | ||
a245769f SW |
676 | struct gfs2_pcpu_lkstats { |
677 | /* One struct for each glock type */ | |
678 | struct gfs2_lkstats lkstats[10]; | |
679 | }; | |
680 | ||
97fd734b AD |
681 | /* List of local (per node) statfs inodes */ |
682 | struct local_statfs_inode { | |
683 | struct list_head si_list; | |
684 | struct inode *si_sc_inode; | |
685 | unsigned int si_jid; /* journal id this statfs inode corresponds to */ | |
686 | }; | |
687 | ||
b3b94faa DT |
688 | struct gfs2_sbd { |
689 | struct super_block *sd_vfs; | |
a245769f | 690 | struct gfs2_pcpu_lkstats __percpu *sd_lkstats; |
b3b94faa | 691 | struct kobject sd_kobj; |
c2a04b02 | 692 | struct completion sd_kobj_unregister; |
b3b94faa | 693 | unsigned long sd_flags; /* SDF_... */ |
f50dfaf7 | 694 | struct gfs2_sb_host sd_sb; |
b3b94faa DT |
695 | |
696 | /* Constants computed on mount */ | |
697 | ||
cd915493 SW |
698 | u32 sd_fsb2bb; |
699 | u32 sd_fsb2bb_shift; | |
700 | u32 sd_diptrs; /* Number of pointers in a dinode */ | |
701 | u32 sd_inptrs; /* Number of pointers in a indirect block */ | |
2e9eeaa1 | 702 | u32 sd_ldptrs; /* Number of pointers in a log descriptor block */ |
cd915493 SW |
703 | u32 sd_jbsize; /* Size of a journaled data block */ |
704 | u32 sd_hash_bsize; /* sizeof(exhash block) */ | |
705 | u32 sd_hash_bsize_shift; | |
706 | u32 sd_hash_ptrs; /* Number of pointers in a hash block */ | |
707 | u32 sd_qc_per_block; | |
a68a0a35 | 708 | u32 sd_blocks_per_bitmap; |
cd915493 SW |
709 | u32 sd_max_dirres; /* Max blocks needed to add a directory entry */ |
710 | u32 sd_max_height; /* Max height of a file's metadata tree */ | |
ecc30c79 | 711 | u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1]; |
471f3db2 | 712 | u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */ |
b3b94faa DT |
713 | |
714 | struct gfs2_args sd_args; /* Mount arguments */ | |
715 | struct gfs2_tune sd_tune; /* Filesystem tuning structure */ | |
716 | ||
717 | /* Lock Stuff */ | |
718 | ||
719 | struct lm_lockstruct sd_lockstruct; | |
b3b94faa DT |
720 | struct gfs2_holder sd_live_gh; |
721 | struct gfs2_glock *sd_rename_gl; | |
24972557 | 722 | struct gfs2_glock *sd_freeze_gl; |
2e60d768 | 723 | struct work_struct sd_freeze_work; |
e402746a | 724 | wait_queue_head_t sd_glock_wait; |
ad26967b | 725 | wait_queue_head_t sd_async_glock_wait; |
e402746a | 726 | atomic_t sd_glock_disposal; |
3942ae53 | 727 | struct completion sd_locking_init; |
fd95e81c | 728 | struct completion sd_wdack; |
e0c2a9aa | 729 | struct delayed_work sd_control_work; |
b3b94faa DT |
730 | |
731 | /* Inode Stuff */ | |
732 | ||
9b8df98f SW |
733 | struct dentry *sd_master_dir; |
734 | struct dentry *sd_root_dir; | |
735 | ||
f42faf4f | 736 | struct inode *sd_jindex; |
f42faf4f | 737 | struct inode *sd_statfs_inode; |
f42faf4f | 738 | struct inode *sd_sc_inode; |
97fd734b | 739 | struct list_head sd_sc_inodes_list; |
f42faf4f SW |
740 | struct inode *sd_qc_inode; |
741 | struct inode *sd_rindex; | |
742 | struct inode *sd_quota_inode; | |
b3b94faa | 743 | |
b3b94faa DT |
744 | /* StatFS stuff */ |
745 | ||
746 | spinlock_t sd_statfs_spin; | |
bd209cc0 AV |
747 | struct gfs2_statfs_change_host sd_statfs_master; |
748 | struct gfs2_statfs_change_host sd_statfs_local; | |
3d3c10f2 | 749 | int sd_statfs_force_sync; |
b3b94faa DT |
750 | |
751 | /* Resource group stuff */ | |
752 | ||
cf45b752 | 753 | int sd_rindex_uptodate; |
b3b94faa | 754 | spinlock_t sd_rindex_spin; |
7c9ca621 | 755 | struct rb_root sd_rindex_tree; |
b3b94faa | 756 | unsigned int sd_rgrps; |
3921120e | 757 | unsigned int sd_max_rg_data; |
b3b94faa DT |
758 | |
759 | /* Journal index stuff */ | |
760 | ||
761 | struct list_head sd_jindex_list; | |
762 | spinlock_t sd_jindex_spin; | |
f55ab26a | 763 | struct mutex sd_jindex_mutex; |
b3b94faa | 764 | unsigned int sd_journals; |
b3b94faa DT |
765 | |
766 | struct gfs2_jdesc *sd_jdesc; | |
767 | struct gfs2_holder sd_journal_gh; | |
768 | struct gfs2_holder sd_jinode_gh; | |
601ef0d5 | 769 | struct gfs2_glock *sd_jinode_gl; |
b3b94faa | 770 | |
b3b94faa | 771 | struct gfs2_holder sd_sc_gh; |
70c11ba8 | 772 | struct buffer_head *sd_sc_bh; |
b3b94faa DT |
773 | struct gfs2_holder sd_qc_gh; |
774 | ||
0e48e055 BP |
775 | struct completion sd_journal_ready; |
776 | ||
0247f4e9 AG |
777 | /* Workqueue stuff */ |
778 | ||
779 | struct workqueue_struct *sd_delete_wq; | |
780 | ||
b3b94faa DT |
781 | /* Daemon stuff */ |
782 | ||
b3b94faa DT |
783 | struct task_struct *sd_logd_process; |
784 | struct task_struct *sd_quotad_process; | |
b3b94faa | 785 | |
b3b94faa DT |
786 | /* Quota stuff */ |
787 | ||
788 | struct list_head sd_quota_list; | |
789 | atomic_t sd_quota_count; | |
f55ab26a | 790 | struct mutex sd_quota_mutex; |
e46c772d | 791 | struct mutex sd_quota_sync_mutex; |
37b2c837 | 792 | wait_queue_head_t sd_quota_wait; |
b3b94faa DT |
793 | |
794 | unsigned int sd_quota_slots; | |
ee2411a8 | 795 | unsigned long *sd_quota_bitmap; |
2d9e7230 | 796 | spinlock_t sd_bitmap_lock; |
b3b94faa | 797 | |
cd915493 | 798 | u64 sd_quota_sync_gen; |
b3b94faa DT |
799 | |
800 | /* Log stuff */ | |
801 | ||
70d4ee94 SW |
802 | struct address_space sd_aspace; |
803 | ||
b3b94faa | 804 | spinlock_t sd_log_lock; |
b3b94faa | 805 | |
16ca9412 | 806 | struct gfs2_trans *sd_log_tr; |
b3b94faa | 807 | unsigned int sd_log_blks_reserved; |
b3b94faa | 808 | |
5e687eac | 809 | atomic_t sd_log_pinned; |
b3b94faa | 810 | unsigned int sd_log_num_revoke; |
f42faf4f | 811 | |
a5b1d3fc AG |
812 | struct list_head sd_log_revokes; |
813 | struct list_head sd_log_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; |
2129b428 | 820 | atomic_t sd_log_revokes_available; |
5e687eac BM |
821 | wait_queue_head_t sd_log_waitq; |
822 | wait_queue_head_t sd_logd_waitq; | |
b3b94faa | 823 | |
cd915493 | 824 | u64 sd_log_sequence; |
b3b94faa DT |
825 | int sd_log_idle; |
826 | ||
484adff8 | 827 | struct rw_semaphore sd_log_flush_lock; |
16615be1 SW |
828 | atomic_t sd_log_in_flight; |
829 | wait_queue_head_t sd_log_flush_wait; | |
036330c9 | 830 | int sd_log_error; /* First log error */ |
601ef0d5 | 831 | wait_queue_head_t sd_withdraw_wait; |
b3b94faa | 832 | |
5cb738b5 AG |
833 | unsigned int sd_log_tail; |
834 | unsigned int sd_log_flush_tail; | |
835 | unsigned int sd_log_head; | |
b3b94faa | 836 | unsigned int sd_log_flush_head; |
b3b94faa | 837 | |
d6a079e8 | 838 | spinlock_t sd_ail_lock; |
b3b94faa DT |
839 | struct list_head sd_ail1_list; |
840 | struct list_head sd_ail2_list; | |
b3b94faa | 841 | |
b3b94faa | 842 | /* For quiescing the filesystem */ |
b3b94faa | 843 | struct gfs2_holder sd_freeze_gh; |
2e60d768 BM |
844 | atomic_t sd_freeze_state; |
845 | struct mutex sd_freeze_mutex; | |
b3b94faa | 846 | |
561b7969 | 847 | char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2]; |
b3b94faa DT |
848 | char sd_table_name[GFS2_FSNAME_LEN]; |
849 | char sd_proto_name[GFS2_FSNAME_LEN]; | |
850 | ||
851 | /* Debugging crud */ | |
852 | ||
853 | unsigned long sd_last_warning; | |
5f882096 | 854 | struct dentry *debugfs_dir; /* debugfs directory */ |
601ef0d5 | 855 | unsigned long sd_glock_dqs_held; |
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 |