incore.h 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886
  1. /*
  2. * Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
  3. * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
  4. *
  5. * This copyrighted material is made available to anyone wishing to use,
  6. * modify, copy, or redistribute it subject to the terms and conditions
  7. * of the GNU General Public License version 2.
  8. */
  9. #ifndef __INCORE_DOT_H__
  10. #define __INCORE_DOT_H__
  11. #include <linux/fs.h>
  12. #include <linux/kobject.h>
  13. #include <linux/workqueue.h>
  14. #include <linux/dlm.h>
  15. #include <linux/buffer_head.h>
  16. #include <linux/rcupdate.h>
  17. #include <linux/rculist_bl.h>
  18. #include <linux/completion.h>
  19. #include <linux/rbtree.h>
  20. #include <linux/ktime.h>
  21. #include <linux/percpu.h>
  22. #include <linux/lockref.h>
  23. #include <linux/rhashtable.h>
  24. #define DIO_WAIT 0x00000010
  25. #define DIO_METADATA 0x00000020
  26. struct gfs2_log_operations;
  27. struct gfs2_bufdata;
  28. struct gfs2_holder;
  29. struct gfs2_glock;
  30. struct gfs2_quota_data;
  31. struct gfs2_trans;
  32. struct gfs2_jdesc;
  33. struct gfs2_sbd;
  34. struct lm_lockops;
  35. typedef void (*gfs2_glop_bh_t) (struct gfs2_glock *gl, unsigned int ret);
  36. struct gfs2_log_header_host {
  37. u64 lh_sequence; /* Sequence number of this transaction */
  38. u32 lh_flags; /* GFS2_LOG_HEAD_... */
  39. u32 lh_tail; /* Block number of log tail */
  40. u32 lh_blkno;
  41. };
  42. /*
  43. * Structure of operations that are associated with each
  44. * type of element in the log.
  45. */
  46. struct gfs2_log_operations {
  47. void (*lo_before_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
  48. void (*lo_after_commit) (struct gfs2_sbd *sdp, struct gfs2_trans *tr);
  49. void (*lo_before_scan) (struct gfs2_jdesc *jd,
  50. struct gfs2_log_header_host *head, int pass);
  51. int (*lo_scan_elements) (struct gfs2_jdesc *jd, unsigned int start,
  52. struct gfs2_log_descriptor *ld, __be64 *ptr,
  53. int pass);
  54. void (*lo_after_scan) (struct gfs2_jdesc *jd, int error, int pass);
  55. const char *lo_name;
  56. };
  57. #define GBF_FULL 1
  58. /**
  59. * Clone bitmaps (bi_clone):
  60. *
  61. * - When a block is freed, we remember the previous state of the block in the
  62. * clone bitmap, and only mark the block as free in the real bitmap.
  63. *
  64. * - When looking for a block to allocate, we check for a free block in the
  65. * clone bitmap, and if no clone bitmap exists, in the real bitmap.
  66. *
  67. * - For allocating a block, we mark it as allocated in the real bitmap, and if
  68. * a clone bitmap exists, also in the clone bitmap.
  69. *
  70. * - At the end of a log_flush, we copy the real bitmap into the clone bitmap
  71. * to make the clone bitmap reflect the current allocation state.
  72. * (Alternatively, we could remove the clone bitmap.)
  73. *
  74. * The clone bitmaps are in-core only, and is never written to disk.
  75. *
  76. * These steps ensure that blocks which have been freed in a transaction cannot
  77. * be reallocated in that same transaction.
  78. */
  79. struct gfs2_bitmap {
  80. struct buffer_head *bi_bh;
  81. char *bi_clone;
  82. unsigned long bi_flags;
  83. u32 bi_offset;
  84. u32 bi_start;
  85. u32 bi_len;
  86. u32 bi_blocks;
  87. };
  88. struct gfs2_rgrpd {
  89. struct rb_node rd_node; /* Link with superblock */
  90. struct gfs2_glock *rd_gl; /* Glock for this rgrp */
  91. u64 rd_addr; /* grp block disk address */
  92. u64 rd_data0; /* first data location */
  93. u32 rd_length; /* length of rgrp header in fs blocks */
  94. u32 rd_data; /* num of data blocks in rgrp */
  95. u32 rd_bitbytes; /* number of bytes in data bitmaps */
  96. u32 rd_free;
  97. u32 rd_reserved; /* number of blocks reserved */
  98. u32 rd_free_clone;
  99. u32 rd_dinodes;
  100. u64 rd_igeneration;
  101. struct gfs2_bitmap *rd_bits;
  102. struct gfs2_sbd *rd_sbd;
  103. struct gfs2_rgrp_lvb *rd_rgl;
  104. u32 rd_last_alloc;
  105. u32 rd_flags;
  106. u32 rd_extfail_pt; /* extent failure point */
  107. #define GFS2_RDF_CHECK 0x10000000 /* check for unlinked inodes */
  108. #define GFS2_RDF_UPTODATE 0x20000000 /* rg is up to date */
  109. #define GFS2_RDF_ERROR 0x40000000 /* error in rg */
  110. #define GFS2_RDF_PREFERRED 0x80000000 /* This rgrp is preferred */
  111. #define GFS2_RDF_MASK 0xf0000000 /* mask for internal flags */
  112. spinlock_t rd_rsspin; /* protects reservation related vars */
  113. struct rb_root rd_rstree; /* multi-block reservation tree */
  114. };
  115. struct gfs2_rbm {
  116. struct gfs2_rgrpd *rgd;
  117. u32 offset; /* The offset is bitmap relative */
  118. int bii; /* Bitmap index */
  119. };
  120. static inline struct gfs2_bitmap *rbm_bi(const struct gfs2_rbm *rbm)
  121. {
  122. return rbm->rgd->rd_bits + rbm->bii;
  123. }
  124. static inline u64 gfs2_rbm_to_block(const struct gfs2_rbm *rbm)
  125. {
  126. BUG_ON(rbm->offset >= rbm->rgd->rd_data);
  127. return rbm->rgd->rd_data0 + (rbm_bi(rbm)->bi_start * GFS2_NBBY) +
  128. rbm->offset;
  129. }
  130. static inline bool gfs2_rbm_eq(const struct gfs2_rbm *rbm1,
  131. const struct gfs2_rbm *rbm2)
  132. {
  133. return (rbm1->rgd == rbm2->rgd) && (rbm1->bii == rbm2->bii) &&
  134. (rbm1->offset == rbm2->offset);
  135. }
  136. enum gfs2_state_bits {
  137. BH_Pinned = BH_PrivateStart,
  138. BH_Escaped = BH_PrivateStart + 1,
  139. };
  140. BUFFER_FNS(Pinned, pinned)
  141. TAS_BUFFER_FNS(Pinned, pinned)
  142. BUFFER_FNS(Escaped, escaped)
  143. TAS_BUFFER_FNS(Escaped, escaped)
  144. struct gfs2_bufdata {
  145. struct buffer_head *bd_bh;
  146. struct gfs2_glock *bd_gl;
  147. u64 bd_blkno;
  148. struct list_head bd_list;
  149. const struct gfs2_log_operations *bd_ops;
  150. struct gfs2_trans *bd_tr;
  151. struct list_head bd_ail_st_list;
  152. struct list_head bd_ail_gl_list;
  153. };
  154. /*
  155. * Internally, we prefix things with gdlm_ and GDLM_ (for gfs-dlm) since a
  156. * prefix of lock_dlm_ gets awkward.
  157. */
  158. #define GDLM_STRNAME_BYTES 25
  159. #define GDLM_LVB_SIZE 32
  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. enum {
  191. DFL_BLOCK_LOCKS = 0,
  192. DFL_NO_DLM_OPS = 1,
  193. DFL_FIRST_MOUNT = 2,
  194. DFL_FIRST_MOUNT_DONE = 3,
  195. DFL_MOUNT_DONE = 4,
  196. DFL_UNMOUNT = 5,
  197. DFL_DLM_RECOVERY = 6,
  198. };
  199. /*
  200. * We are using struct lm_lockname as an rhashtable key. Avoid holes within
  201. * the struct; padding at the end is fine.
  202. */
  203. struct lm_lockname {
  204. u64 ln_number;
  205. struct gfs2_sbd *ln_sbd;
  206. unsigned int ln_type;
  207. };
  208. #define lm_name_equal(name1, name2) \
  209. (((name1)->ln_number == (name2)->ln_number) && \
  210. ((name1)->ln_type == (name2)->ln_type) && \
  211. ((name1)->ln_sbd == (name2)->ln_sbd))
  212. struct gfs2_glock_operations {
  213. void (*go_sync) (struct gfs2_glock *gl);
  214. int (*go_xmote_bh) (struct gfs2_glock *gl, struct gfs2_holder *gh);
  215. void (*go_inval) (struct gfs2_glock *gl, int flags);
  216. int (*go_demote_ok) (const struct gfs2_glock *gl);
  217. int (*go_lock) (struct gfs2_holder *gh);
  218. void (*go_unlock) (struct gfs2_holder *gh);
  219. void (*go_dump)(struct seq_file *seq, const struct gfs2_glock *gl);
  220. void (*go_callback)(struct gfs2_glock *gl, bool remote);
  221. const int go_type;
  222. const unsigned long go_flags;
  223. #define GLOF_ASPACE 1
  224. #define GLOF_LVB 2
  225. #define GLOF_LRU 4
  226. };
  227. enum {
  228. GFS2_LKS_SRTT = 0, /* Non blocking smoothed round trip time */
  229. GFS2_LKS_SRTTVAR = 1, /* Non blocking smoothed variance */
  230. GFS2_LKS_SRTTB = 2, /* Blocking smoothed round trip time */
  231. GFS2_LKS_SRTTVARB = 3, /* Blocking smoothed variance */
  232. GFS2_LKS_SIRT = 4, /* Smoothed Inter-request time */
  233. GFS2_LKS_SIRTVAR = 5, /* Smoothed Inter-request variance */
  234. GFS2_LKS_DCOUNT = 6, /* Count of dlm requests */
  235. GFS2_LKS_QCOUNT = 7, /* Count of gfs2_holder queues */
  236. GFS2_NR_LKSTATS
  237. };
  238. struct gfs2_lkstats {
  239. u64 stats[GFS2_NR_LKSTATS];
  240. };
  241. enum {
  242. /* States */
  243. HIF_HOLDER = 6, /* Set for gh that "holds" the glock */
  244. HIF_FIRST = 7,
  245. HIF_WAIT = 10,
  246. };
  247. struct gfs2_holder {
  248. struct list_head gh_list;
  249. struct gfs2_glock *gh_gl;
  250. struct pid *gh_owner_pid;
  251. u16 gh_flags;
  252. u16 gh_state;
  253. int gh_error;
  254. unsigned long gh_iflags; /* HIF_... */
  255. unsigned long gh_ip;
  256. };
  257. /* Number of quota types we support */
  258. #define GFS2_MAXQUOTAS 2
  259. struct gfs2_qadata { /* quota allocation data */
  260. /* Quota stuff */
  261. struct gfs2_quota_data *qa_qd[2 * GFS2_MAXQUOTAS];
  262. struct gfs2_holder qa_qd_ghs[2 * GFS2_MAXQUOTAS];
  263. unsigned int qa_qd_num;
  264. };
  265. /* Resource group multi-block reservation, in order of appearance:
  266. Step 1. Function prepares to write, allocates a mb, sets the size hint.
  267. Step 2. User calls inplace_reserve to target an rgrp, sets the rgrp info
  268. Step 3. Function get_local_rgrp locks the rgrp, determines which bits to use
  269. Step 4. Bits are assigned from the rgrp based on either the reservation
  270. or wherever it can.
  271. */
  272. struct gfs2_blkreserv {
  273. /* components used during write (step 1): */
  274. atomic_t rs_sizehint; /* hint of the write size */
  275. struct gfs2_holder rs_rgd_gh; /* Filled in by get_local_rgrp */
  276. struct rb_node rs_node; /* link to other block reservations */
  277. struct gfs2_rbm rs_rbm; /* Start of reservation */
  278. u32 rs_free; /* how many blocks are still free */
  279. };
  280. /*
  281. * Allocation parameters
  282. * @target: The number of blocks we'd ideally like to allocate
  283. * @aflags: The flags (e.g. Orlov flag)
  284. *
  285. * The intent is to gradually expand this structure over time in
  286. * order to give more information, e.g. alignment, min extent size
  287. * to the allocation code.
  288. */
  289. struct gfs2_alloc_parms {
  290. u64 target;
  291. u32 min_target;
  292. u32 aflags;
  293. u64 allowed;
  294. };
  295. enum {
  296. GLF_LOCK = 1,
  297. GLF_DEMOTE = 3,
  298. GLF_PENDING_DEMOTE = 4,
  299. GLF_DEMOTE_IN_PROGRESS = 5,
  300. GLF_DIRTY = 6,
  301. GLF_LFLUSH = 7,
  302. GLF_INVALIDATE_IN_PROGRESS = 8,
  303. GLF_REPLY_PENDING = 9,
  304. GLF_INITIAL = 10,
  305. GLF_FROZEN = 11,
  306. GLF_QUEUED = 12,
  307. GLF_LRU = 13,
  308. GLF_OBJECT = 14, /* Used only for tracing */
  309. GLF_BLOCKING = 15,
  310. GLF_INODE_CREATING = 16, /* Inode creation occurring */
  311. };
  312. struct gfs2_glock {
  313. unsigned long gl_flags; /* GLF_... */
  314. struct lm_lockname gl_name;
  315. struct lockref gl_lockref;
  316. /* State fields protected by gl_lockref.lock */
  317. unsigned int gl_state:2, /* Current state */
  318. gl_target:2, /* Target state */
  319. gl_demote_state:2, /* State requested by remote node */
  320. gl_req:2, /* State in last dlm request */
  321. gl_reply:8; /* Last reply from the dlm */
  322. unsigned long gl_demote_time; /* time of first demote request */
  323. long gl_hold_time;
  324. struct list_head gl_holders;
  325. const struct gfs2_glock_operations *gl_ops;
  326. ktime_t gl_dstamp;
  327. struct gfs2_lkstats gl_stats;
  328. struct dlm_lksb gl_lksb;
  329. unsigned long gl_tchange;
  330. void *gl_object;
  331. struct list_head gl_lru;
  332. struct list_head gl_ail_list;
  333. atomic_t gl_ail_count;
  334. atomic_t gl_revokes;
  335. struct delayed_work gl_work;
  336. union {
  337. /* For inode and iopen glocks only */
  338. struct work_struct gl_delete;
  339. /* For rgrp glocks only */
  340. struct {
  341. loff_t start;
  342. loff_t end;
  343. } gl_vm;
  344. };
  345. struct rcu_head gl_rcu;
  346. struct rhash_head gl_node;
  347. };
  348. #define GFS2_MIN_LVB_SIZE 32 /* Min size of LVB that gfs2 supports */
  349. enum {
  350. GIF_INVALID = 0,
  351. GIF_QD_LOCKED = 1,
  352. GIF_ALLOC_FAILED = 2,
  353. GIF_SW_PAGED = 3,
  354. GIF_ORDERED = 4,
  355. GIF_FREE_VFS_INODE = 5,
  356. GIF_GLOP_PENDING = 6,
  357. };
  358. struct gfs2_inode {
  359. struct inode i_inode;
  360. u64 i_no_addr;
  361. u64 i_no_formal_ino;
  362. u64 i_generation;
  363. u64 i_eattr;
  364. unsigned long i_flags; /* GIF_... */
  365. struct gfs2_glock *i_gl; /* Move into i_gh? */
  366. struct gfs2_holder i_iopen_gh;
  367. struct gfs2_holder i_gh; /* for prepare/commit_write only */
  368. struct gfs2_qadata *i_qadata; /* quota allocation data */
  369. struct gfs2_blkreserv i_res; /* rgrp multi-block reservation */
  370. u64 i_goal; /* goal block for allocations */
  371. struct rw_semaphore i_rw_mutex;
  372. struct list_head i_ordered;
  373. struct list_head i_trunc_list;
  374. __be64 *i_hash_cache;
  375. u32 i_entries;
  376. u32 i_diskflags;
  377. u8 i_height;
  378. u8 i_depth;
  379. u16 i_rahead;
  380. };
  381. /*
  382. * Since i_inode is the first element of struct gfs2_inode,
  383. * this is effectively a cast.
  384. */
  385. static inline struct gfs2_inode *GFS2_I(struct inode *inode)
  386. {
  387. return container_of(inode, struct gfs2_inode, i_inode);
  388. }
  389. static inline struct gfs2_sbd *GFS2_SB(const struct inode *inode)
  390. {
  391. return inode->i_sb->s_fs_info;
  392. }
  393. struct gfs2_file {
  394. struct mutex f_fl_mutex;
  395. struct gfs2_holder f_fl_gh;
  396. };
  397. struct gfs2_revoke_replay {
  398. struct list_head rr_list;
  399. u64 rr_blkno;
  400. unsigned int rr_where;
  401. };
  402. enum {
  403. QDF_CHANGE = 1,
  404. QDF_LOCKED = 2,
  405. QDF_REFRESH = 3,
  406. QDF_QMSG_QUIET = 4,
  407. };
  408. struct gfs2_quota_data {
  409. struct hlist_bl_node qd_hlist;
  410. struct list_head qd_list;
  411. struct kqid qd_id;
  412. struct gfs2_sbd *qd_sbd;
  413. struct lockref qd_lockref;
  414. struct list_head qd_lru;
  415. unsigned qd_hash;
  416. unsigned long qd_flags; /* QDF_... */
  417. s64 qd_change;
  418. s64 qd_change_sync;
  419. unsigned int qd_slot;
  420. unsigned int qd_slot_count;
  421. struct buffer_head *qd_bh;
  422. struct gfs2_quota_change *qd_bh_qc;
  423. unsigned int qd_bh_count;
  424. struct gfs2_glock *qd_gl;
  425. struct gfs2_quota_lvb qd_qb;
  426. u64 qd_sync_gen;
  427. unsigned long qd_last_warn;
  428. struct rcu_head qd_rcu;
  429. };
  430. enum {
  431. TR_TOUCHED = 1,
  432. TR_ATTACHED = 2,
  433. TR_ALLOCED = 3,
  434. };
  435. struct gfs2_trans {
  436. unsigned long tr_ip;
  437. unsigned int tr_blocks;
  438. unsigned int tr_revokes;
  439. unsigned int tr_reserved;
  440. unsigned long tr_flags;
  441. unsigned int tr_num_buf_new;
  442. unsigned int tr_num_databuf_new;
  443. unsigned int tr_num_buf_rm;
  444. unsigned int tr_num_databuf_rm;
  445. unsigned int tr_num_revoke;
  446. unsigned int tr_num_revoke_rm;
  447. struct list_head tr_list;
  448. struct list_head tr_databuf;
  449. struct list_head tr_buf;
  450. unsigned int tr_first;
  451. struct list_head tr_ail1_list;
  452. struct list_head tr_ail2_list;
  453. };
  454. struct gfs2_journal_extent {
  455. struct list_head list;
  456. unsigned int lblock; /* First logical block */
  457. u64 dblock; /* First disk block */
  458. u64 blocks;
  459. };
  460. struct gfs2_jdesc {
  461. struct list_head jd_list;
  462. struct list_head extent_list;
  463. unsigned int nr_extents;
  464. struct work_struct jd_work;
  465. struct inode *jd_inode;
  466. unsigned long jd_flags;
  467. #define JDF_RECOVERY 1
  468. unsigned int jd_jid;
  469. unsigned int jd_blocks;
  470. int jd_recover_error;
  471. /* Replay stuff */
  472. unsigned int jd_found_blocks;
  473. unsigned int jd_found_revokes;
  474. unsigned int jd_replayed_blocks;
  475. struct list_head jd_revoke_list;
  476. unsigned int jd_replay_tail;
  477. };
  478. struct gfs2_statfs_change_host {
  479. s64 sc_total;
  480. s64 sc_free;
  481. s64 sc_dinodes;
  482. };
  483. #define GFS2_QUOTA_DEFAULT GFS2_QUOTA_OFF
  484. #define GFS2_QUOTA_OFF 0
  485. #define GFS2_QUOTA_ACCOUNT 1
  486. #define GFS2_QUOTA_ON 2
  487. #define GFS2_DATA_DEFAULT GFS2_DATA_ORDERED
  488. #define GFS2_DATA_WRITEBACK 1
  489. #define GFS2_DATA_ORDERED 2
  490. #define GFS2_ERRORS_DEFAULT GFS2_ERRORS_WITHDRAW
  491. #define GFS2_ERRORS_WITHDRAW 0
  492. #define GFS2_ERRORS_CONTINUE 1 /* place holder for future feature */
  493. #define GFS2_ERRORS_RO 2 /* place holder for future feature */
  494. #define GFS2_ERRORS_PANIC 3
  495. struct gfs2_args {
  496. char ar_lockproto[GFS2_LOCKNAME_LEN]; /* Name of the Lock Protocol */
  497. char ar_locktable[GFS2_LOCKNAME_LEN]; /* Name of the Lock Table */
  498. char ar_hostdata[GFS2_LOCKNAME_LEN]; /* Host specific data */
  499. unsigned int ar_spectator:1; /* Don't get a journal */
  500. unsigned int ar_localflocks:1; /* Let the VFS do flock|fcntl */
  501. unsigned int ar_debug:1; /* Oops on errors */
  502. unsigned int ar_posix_acl:1; /* Enable posix acls */
  503. unsigned int ar_quota:2; /* off/account/on */
  504. unsigned int ar_suiddir:1; /* suiddir support */
  505. unsigned int ar_data:2; /* ordered/writeback */
  506. unsigned int ar_meta:1; /* mount metafs */
  507. unsigned int ar_discard:1; /* discard requests */
  508. unsigned int ar_errors:2; /* errors=withdraw | panic */
  509. unsigned int ar_nobarrier:1; /* do not send barriers */
  510. unsigned int ar_rgrplvb:1; /* use lvbs for rgrp info */
  511. unsigned int ar_loccookie:1; /* use location based readdir
  512. cookies */
  513. int ar_commit; /* Commit interval */
  514. int ar_statfs_quantum; /* The fast statfs interval */
  515. int ar_quota_quantum; /* The quota interval */
  516. int ar_statfs_percent; /* The % change to force sync */
  517. };
  518. struct gfs2_tune {
  519. spinlock_t gt_spin;
  520. unsigned int gt_logd_secs;
  521. unsigned int gt_quota_warn_period; /* Secs between quota warn msgs */
  522. unsigned int gt_quota_scale_num; /* Numerator */
  523. unsigned int gt_quota_scale_den; /* Denominator */
  524. unsigned int gt_quota_quantum; /* Secs between syncs to quota file */
  525. unsigned int gt_new_files_jdata;
  526. unsigned int gt_max_readahead; /* Max bytes to read-ahead from disk */
  527. unsigned int gt_complain_secs;
  528. unsigned int gt_statfs_quantum;
  529. unsigned int gt_statfs_slow;
  530. };
  531. enum {
  532. SDF_JOURNAL_CHECKED = 0,
  533. SDF_JOURNAL_LIVE = 1,
  534. SDF_SHUTDOWN = 2,
  535. SDF_NOBARRIERS = 3,
  536. SDF_NORECOVERY = 4,
  537. SDF_DEMOTE = 5,
  538. SDF_NOJOURNALID = 6,
  539. SDF_RORECOVERY = 7, /* read only recovery */
  540. SDF_SKIP_DLM_UNLOCK = 8,
  541. SDF_FORCE_AIL_FLUSH = 9,
  542. SDF_AIL1_IO_ERROR = 10,
  543. };
  544. enum gfs2_freeze_state {
  545. SFS_UNFROZEN = 0,
  546. SFS_STARTING_FREEZE = 1,
  547. SFS_FROZEN = 2,
  548. };
  549. #define GFS2_FSNAME_LEN 256
  550. struct gfs2_inum_host {
  551. u64 no_formal_ino;
  552. u64 no_addr;
  553. };
  554. struct gfs2_sb_host {
  555. u32 sb_magic;
  556. u32 sb_type;
  557. u32 sb_format;
  558. u32 sb_fs_format;
  559. u32 sb_multihost_format;
  560. u32 sb_bsize;
  561. u32 sb_bsize_shift;
  562. struct gfs2_inum_host sb_master_dir;
  563. struct gfs2_inum_host sb_root_dir;
  564. char sb_lockproto[GFS2_LOCKNAME_LEN];
  565. char sb_locktable[GFS2_LOCKNAME_LEN];
  566. };
  567. /*
  568. * lm_mount() return values
  569. *
  570. * ls_jid - the journal ID this node should use
  571. * ls_first - this node is the first to mount the file system
  572. * ls_lockspace - lock module's context for this file system
  573. * ls_ops - lock module's functions
  574. */
  575. struct lm_lockstruct {
  576. int ls_jid;
  577. unsigned int ls_first;
  578. const struct lm_lockops *ls_ops;
  579. dlm_lockspace_t *ls_dlm;
  580. int ls_recover_jid_done; /* These two are deprecated, */
  581. int ls_recover_jid_status; /* used previously by gfs_controld */
  582. struct dlm_lksb ls_mounted_lksb; /* mounted_lock */
  583. struct dlm_lksb ls_control_lksb; /* control_lock */
  584. char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
  585. struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
  586. char *ls_lvb_bits;
  587. spinlock_t ls_recover_spin; /* protects following fields */
  588. unsigned long ls_recover_flags; /* DFL_ */
  589. uint32_t ls_recover_mount; /* gen in first recover_done cb */
  590. uint32_t ls_recover_start; /* gen in last recover_done cb */
  591. uint32_t ls_recover_block; /* copy recover_start in last recover_prep */
  592. uint32_t ls_recover_size; /* size of recover_submit, recover_result */
  593. uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */
  594. uint32_t *ls_recover_result; /* result of last jid recovery */
  595. };
  596. struct gfs2_pcpu_lkstats {
  597. /* One struct for each glock type */
  598. struct gfs2_lkstats lkstats[10];
  599. };
  600. struct gfs2_sbd {
  601. struct super_block *sd_vfs;
  602. struct gfs2_pcpu_lkstats __percpu *sd_lkstats;
  603. struct kobject sd_kobj;
  604. unsigned long sd_flags; /* SDF_... */
  605. struct gfs2_sb_host sd_sb;
  606. /* Constants computed on mount */
  607. u32 sd_fsb2bb;
  608. u32 sd_fsb2bb_shift;
  609. u32 sd_diptrs; /* Number of pointers in a dinode */
  610. u32 sd_inptrs; /* Number of pointers in a indirect block */
  611. u32 sd_jbsize; /* Size of a journaled data block */
  612. u32 sd_hash_bsize; /* sizeof(exhash block) */
  613. u32 sd_hash_bsize_shift;
  614. u32 sd_hash_ptrs; /* Number of pointers in a hash block */
  615. u32 sd_qc_per_block;
  616. u32 sd_blocks_per_bitmap;
  617. u32 sd_max_dirres; /* Max blocks needed to add a directory entry */
  618. u32 sd_max_height; /* Max height of a file's metadata tree */
  619. u64 sd_heightsize[GFS2_MAX_META_HEIGHT + 1];
  620. u32 sd_max_dents_per_leaf; /* Max number of dirents in a leaf block */
  621. struct gfs2_args sd_args; /* Mount arguments */
  622. struct gfs2_tune sd_tune; /* Filesystem tuning structure */
  623. /* Lock Stuff */
  624. struct lm_lockstruct sd_lockstruct;
  625. struct gfs2_holder sd_live_gh;
  626. struct gfs2_glock *sd_rename_gl;
  627. struct gfs2_glock *sd_freeze_gl;
  628. struct work_struct sd_freeze_work;
  629. wait_queue_head_t sd_glock_wait;
  630. atomic_t sd_glock_disposal;
  631. struct completion sd_locking_init;
  632. struct completion sd_wdack;
  633. struct delayed_work sd_control_work;
  634. /* Inode Stuff */
  635. struct dentry *sd_master_dir;
  636. struct dentry *sd_root_dir;
  637. struct inode *sd_jindex;
  638. struct inode *sd_statfs_inode;
  639. struct inode *sd_sc_inode;
  640. struct inode *sd_qc_inode;
  641. struct inode *sd_rindex;
  642. struct inode *sd_quota_inode;
  643. /* StatFS stuff */
  644. spinlock_t sd_statfs_spin;
  645. struct gfs2_statfs_change_host sd_statfs_master;
  646. struct gfs2_statfs_change_host sd_statfs_local;
  647. int sd_statfs_force_sync;
  648. /* Resource group stuff */
  649. int sd_rindex_uptodate;
  650. spinlock_t sd_rindex_spin;
  651. struct rb_root sd_rindex_tree;
  652. unsigned int sd_rgrps;
  653. unsigned int sd_max_rg_data;
  654. /* Journal index stuff */
  655. struct list_head sd_jindex_list;
  656. spinlock_t sd_jindex_spin;
  657. struct mutex sd_jindex_mutex;
  658. unsigned int sd_journals;
  659. struct gfs2_jdesc *sd_jdesc;
  660. struct gfs2_holder sd_journal_gh;
  661. struct gfs2_holder sd_jinode_gh;
  662. struct gfs2_holder sd_sc_gh;
  663. struct gfs2_holder sd_qc_gh;
  664. struct completion sd_journal_ready;
  665. /* Daemon stuff */
  666. struct task_struct *sd_logd_process;
  667. struct task_struct *sd_quotad_process;
  668. /* Quota stuff */
  669. struct list_head sd_quota_list;
  670. atomic_t sd_quota_count;
  671. struct mutex sd_quota_mutex;
  672. struct mutex sd_quota_sync_mutex;
  673. wait_queue_head_t sd_quota_wait;
  674. struct list_head sd_trunc_list;
  675. spinlock_t sd_trunc_lock;
  676. unsigned int sd_quota_slots;
  677. unsigned long *sd_quota_bitmap;
  678. spinlock_t sd_bitmap_lock;
  679. u64 sd_quota_sync_gen;
  680. /* Log stuff */
  681. struct address_space sd_aspace;
  682. spinlock_t sd_log_lock;
  683. struct gfs2_trans *sd_log_tr;
  684. unsigned int sd_log_blks_reserved;
  685. int sd_log_commited_revoke;
  686. atomic_t sd_log_pinned;
  687. unsigned int sd_log_num_revoke;
  688. struct list_head sd_log_le_revoke;
  689. struct list_head sd_log_le_ordered;
  690. spinlock_t sd_ordered_lock;
  691. atomic_t sd_log_thresh1;
  692. atomic_t sd_log_thresh2;
  693. atomic_t sd_log_blks_free;
  694. atomic_t sd_log_blks_needed;
  695. wait_queue_head_t sd_log_waitq;
  696. wait_queue_head_t sd_logd_waitq;
  697. u64 sd_log_sequence;
  698. unsigned int sd_log_head;
  699. unsigned int sd_log_tail;
  700. int sd_log_idle;
  701. struct rw_semaphore sd_log_flush_lock;
  702. atomic_t sd_log_in_flight;
  703. struct bio *sd_log_bio;
  704. wait_queue_head_t sd_log_flush_wait;
  705. int sd_log_error;
  706. atomic_t sd_reserving_log;
  707. wait_queue_head_t sd_reserving_log_wait;
  708. unsigned int sd_log_flush_head;
  709. spinlock_t sd_ail_lock;
  710. struct list_head sd_ail1_list;
  711. struct list_head sd_ail2_list;
  712. /* For quiescing the filesystem */
  713. struct gfs2_holder sd_freeze_gh;
  714. atomic_t sd_freeze_state;
  715. struct mutex sd_freeze_mutex;
  716. char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2];
  717. char sd_table_name[GFS2_FSNAME_LEN];
  718. char sd_proto_name[GFS2_FSNAME_LEN];
  719. /* Debugging crud */
  720. unsigned long sd_last_warning;
  721. struct dentry *debugfs_dir; /* debugfs directory */
  722. struct dentry *debugfs_dentry_glocks;
  723. struct dentry *debugfs_dentry_glstats;
  724. struct dentry *debugfs_dentry_sbstats;
  725. };
  726. static inline void gfs2_glstats_inc(struct gfs2_glock *gl, int which)
  727. {
  728. gl->gl_stats.stats[which]++;
  729. }
  730. static inline void gfs2_sbstats_inc(const struct gfs2_glock *gl, int which)
  731. {
  732. const struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
  733. preempt_disable();
  734. this_cpu_ptr(sdp->sd_lkstats)->lkstats[gl->gl_name.ln_type].stats[which]++;
  735. preempt_enable();
  736. }
  737. extern struct gfs2_rgrpd *gfs2_glock2rgrp(struct gfs2_glock *gl);
  738. static inline unsigned gfs2_max_stuffed_size(const struct gfs2_inode *ip)
  739. {
  740. return GFS2_SB(&ip->i_inode)->sd_sb.sb_bsize - sizeof(struct gfs2_dinode);
  741. }
  742. #endif /* __INCORE_DOT_H__ */