blob: a7077a0c989fb33cde837a41e6b4d8b9951d2c43 [file] [log] [blame]
Greg Kroah-Hartmanb2441312017-11-01 15:07:57 +01001/* SPDX-License-Identifier: GPL-2.0 */
Sage Weilde576062009-10-06 11:31:07 -07002#ifndef _FS_CEPH_SUPER_H
3#define _FS_CEPH_SUPER_H
4
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07005#include <linux/ceph/ceph_debug.h>
Sage Weilde576062009-10-06 11:31:07 -07006
7#include <asm/unaligned.h>
8#include <linux/backing-dev.h>
9#include <linux/completion.h>
10#include <linux/exportfs.h>
11#include <linux/fs.h>
12#include <linux/mempool.h>
13#include <linux/pagemap.h>
14#include <linux/wait.h>
Stephen Rothwellf1a3d572010-01-18 11:53:08 +110015#include <linux/writeback.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090016#include <linux/slab.h>
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -080017#include <linux/posix_acl.h>
Elena Reshetova805692d2017-03-03 11:15:07 +020018#include <linux/refcount.h>
Sage Weilde576062009-10-06 11:31:07 -070019
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070020#include <linux/ceph/libceph.h>
Sage Weilde576062009-10-06 11:31:07 -070021
Milosz Tanski99ccbd22013-08-21 17:29:54 -040022#ifdef CONFIG_CEPH_FSCACHE
23#include <linux/fscache.h>
24#endif
25
Sage Weilde576062009-10-06 11:31:07 -070026/* f_type in struct statfs */
27#define CEPH_SUPER_MAGIC 0x00c36400
28
29/* large granularity for statfs utilization stats to facilitate
30 * large volume sizes on 32-bit machines. */
Sage Weil92a49fb2013-02-22 15:31:00 -080031#define CEPH_BLOCK_SHIFT 22 /* 4 MB */
Sage Weilde576062009-10-06 11:31:07 -070032#define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
33
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070034#define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */
35#define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
36#define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
Yehuda Sadehad1fee92011-01-21 16:44:03 -080037#define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
Sage Weila40dc6c2012-01-10 09:12:55 -080038#define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
Milosz Tanski99ccbd22013-08-21 17:29:54 -040039#define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
Yan, Zheng10183a62015-04-27 15:33:28 +080040#define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
Yan, Zhenge9e427f2016-11-10 16:02:06 +080041#define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */
Luis Henriques9122eed2018-01-31 10:53:13 +000042#define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */
Sage Weil6a259382010-07-07 09:06:08 -070043
Yan, Zheng133e9152016-01-25 10:44:33 +080044#define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE
Sage Weilde576062009-10-06 11:31:07 -070045
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070046#define ceph_set_mount_opt(fsc, opt) \
47 (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt;
48#define ceph_test_mount_opt(fsc, opt) \
49 (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
Sage Weilde576062009-10-06 11:31:07 -070050
Yan, Zhengaa187922017-07-11 15:56:09 +080051/* max size of osd read request, limited by libceph */
52#define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN
Yan, Zheng95cca2b2017-07-11 17:34:46 +080053/* osd has a configurable limitaion of max write size.
54 * CEPH_MSG_MAX_DATA_LEN should be small enough. */
55#define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN
Andreas Gerstmayr7c94ba22017-01-10 14:17:56 +010056#define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070057#define CEPH_MAX_READDIR_DEFAULT 1024
58#define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
59#define CEPH_SNAPDIRNAME_DEFAULT ".snap"
Sage Weilde576062009-10-06 11:31:07 -070060
Yan, Zheng4214fb12017-07-11 18:49:44 +080061/*
62 * Delay telling the MDS we no longer want caps, in case we reopen
63 * the file. Delay a minimum amount of time, even if we send a cap
64 * message for some other reason. Otherwise, take the oppotunity to
65 * update the mds to avoid sending another message later.
66 */
67#define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
68#define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
69
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070070struct ceph_mount_options {
Sage Weilde576062009-10-06 11:31:07 -070071 int flags;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070072 int sb_flags;
73
Sage Weil83817e32011-08-04 08:03:44 -070074 int wsize; /* max write size */
75 int rsize; /* max read size */
76 int rasize; /* max readahead */
Sage Weil6e19a162010-04-29 16:38:32 -070077 int congestion_kb; /* max writeback in flight */
78 int caps_wanted_delay_min, caps_wanted_delay_max;
Sage Weil23804d92010-05-14 13:06:30 -070079 int max_readdir; /* max readdir result (entires) */
80 int max_readdir_bytes; /* max readdir result (bytes) */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070081
82 /*
83 * everything above this point can be memcmp'd; everything below
84 * is handled in compare_mount_options()
85 */
86
Sage Weilde576062009-10-06 11:31:07 -070087 char *snapdir_name; /* default ".snap" */
Yan, Zheng430afba2016-07-08 11:25:38 +080088 char *mds_namespace; /* default NULL */
Yan, Zheng3f384952016-04-21 11:09:55 +080089 char *server_path; /* default "/" */
Yan, Zheng1d8f8362017-06-27 11:57:56 +080090 char *fscache_uniq; /* default NULL */
Sage Weilde576062009-10-06 11:31:07 -070091};
92
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070093struct ceph_fs_client {
Sage Weilde576062009-10-06 11:31:07 -070094 struct super_block *sb;
95
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070096 struct ceph_mount_options *mount_options;
97 struct ceph_client *client;
98
Sage Weilde576062009-10-06 11:31:07 -070099 unsigned long mount_state;
Sage Weil85ccce42010-02-17 10:02:43 -0800100 int min_caps; /* min caps i added */
101
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700102 struct ceph_mds_client *mdsc;
Sage Weilde576062009-10-06 11:31:07 -0700103
104 /* writeback */
105 mempool_t *wb_pagevec_pool;
106 struct workqueue_struct *wb_wq;
107 struct workqueue_struct *pg_inv_wq;
108 struct workqueue_struct *trunc_wq;
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800109 atomic_long_t writeback_count;
Sage Weilde576062009-10-06 11:31:07 -0700110
Sage Weil07433042009-11-18 16:50:41 -0800111#ifdef CONFIG_DEBUG_FS
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700112 struct dentry *debugfs_dentry_lru, *debugfs_caps;
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800113 struct dentry *debugfs_congestion_kb;
Sage Weil06edf042009-12-15 14:44:32 -0800114 struct dentry *debugfs_bdi;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700115 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
John Spray14ed9702014-09-12 16:58:49 +0100116 struct dentry *debugfs_mds_sessions;
Sage Weil07433042009-11-18 16:50:41 -0800117#endif
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400118
119#ifdef CONFIG_CEPH_FSCACHE
120 struct fscache_cookie *fscache;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400121#endif
Sage Weilde576062009-10-06 11:31:07 -0700122};
123
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700124
Sage Weilde576062009-10-06 11:31:07 -0700125/*
126 * File i/o capability. This tracks shared state with the metadata
127 * server that allows us to cache or writeback attributes or to read
128 * and write data. For any given inode, we should have one or more
129 * capabilities, one issued by each metadata server, and our
130 * cumulative access is the OR of all issued capabilities.
131 *
132 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
133 * session capability lists.
134 */
135struct ceph_cap {
136 struct ceph_inode_info *ci;
137 struct rb_node ci_node; /* per-ci cap tree */
138 struct ceph_mds_session *session;
139 struct list_head session_caps; /* per-session caplist */
Sage Weilde576062009-10-06 11:31:07 -0700140 u64 cap_id; /* unique cap id (mds provided) */
Yan, Zheng745a8e32015-05-14 17:22:42 +0800141 union {
142 /* in-use caps */
143 struct {
144 int issued; /* latest, from the mds */
145 int implemented; /* implemented superset of
146 issued (for revocation) */
147 int mds, mds_wanted;
148 };
149 /* caps to release */
150 struct {
151 u64 cap_ino;
152 int queue_release;
153 };
154 };
Sage Weil685f9a5d2009-11-09 12:05:48 -0800155 u32 seq, issue_seq, mseq;
156 u32 cap_gen; /* active/stale cycle */
Sage Weilde576062009-10-06 11:31:07 -0700157 unsigned long last_used;
158 struct list_head caps_item;
159};
160
161#define CHECK_CAPS_NODELAY 1 /* do not delay any further */
162#define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */
163#define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */
164
Yan, Zheng0e294382016-07-04 18:06:41 +0800165struct ceph_cap_flush {
166 u64 tid;
167 int caps; /* 0 means capsnap */
Yan, Zhengc8799fc2016-07-07 15:22:38 +0800168 bool wake; /* wake up flush waiters when finish ? */
Yan, Zheng0e294382016-07-04 18:06:41 +0800169 struct list_head g_list; // global
170 struct list_head i_list; // per inode
171};
172
Sage Weilde576062009-10-06 11:31:07 -0700173/*
174 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
175 * we first complete any in-process sync writes and writeback any dirty
176 * data before flushing the snapped state (tracked here) back to the MDS.
177 */
178struct ceph_cap_snap {
Elena Reshetova805692d2017-03-03 11:15:07 +0200179 refcount_t nref;
Yan, Zheng0e294382016-07-04 18:06:41 +0800180 struct list_head ci_item;
Sage Weilde576062009-10-06 11:31:07 -0700181
Yan, Zheng0e294382016-07-04 18:06:41 +0800182 struct ceph_cap_flush cap_flush;
183
184 u64 follows;
Sage Weilde576062009-10-06 11:31:07 -0700185 int issued, dirty;
186 struct ceph_snap_context *context;
187
Al Viro5706b272011-07-26 04:52:22 -0400188 umode_t mode;
Eric W. Biederman05cb11c2013-01-31 02:56:19 -0800189 kuid_t uid;
190 kgid_t gid;
Sage Weilde576062009-10-06 11:31:07 -0700191
Sage Weil4a625be2010-08-22 15:03:56 -0700192 struct ceph_buffer *xattr_blob;
Sage Weilde576062009-10-06 11:31:07 -0700193 u64 xattr_version;
194
195 u64 size;
196 struct timespec mtime, atime, ctime;
197 u64 time_warp_seq;
Yan, Zheng5f743e42016-11-15 16:04:37 +0800198 u64 truncate_size;
199 u32 truncate_seq;
Sage Weilde576062009-10-06 11:31:07 -0700200 int writing; /* a sync write is still in progress */
201 int dirty_pages; /* dirty pages awaiting writeback */
Yan, Zhenge20d2582014-11-14 22:39:13 +0800202 bool inline_data;
Yan, Zheng86056092015-05-01 16:57:16 +0800203 bool need_flush;
Sage Weilde576062009-10-06 11:31:07 -0700204};
205
206static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
207{
Elena Reshetova805692d2017-03-03 11:15:07 +0200208 if (refcount_dec_and_test(&capsnap->nref)) {
Sage Weil4a625be2010-08-22 15:03:56 -0700209 if (capsnap->xattr_blob)
210 ceph_buffer_put(capsnap->xattr_blob);
Sage Weilde576062009-10-06 11:31:07 -0700211 kfree(capsnap);
Sage Weil4a625be2010-08-22 15:03:56 -0700212 }
Sage Weilde576062009-10-06 11:31:07 -0700213}
214
215/*
216 * The frag tree describes how a directory is fragmented, potentially across
217 * multiple metadata servers. It is also used to indicate points where
218 * metadata authority is delegated, and whether/where metadata is replicated.
219 *
220 * A _leaf_ frag will be present in the i_fragtree IFF there is
221 * delegation info. That is, if mds >= 0 || ndist > 0.
222 */
223#define CEPH_MAX_DIRFRAG_REP 4
224
225struct ceph_inode_frag {
226 struct rb_node node;
227
228 /* fragtree state */
229 u32 frag;
230 int split_by; /* i.e. 2^(split_by) children */
231
232 /* delegation and replication info */
233 int mds; /* -1 if same authority as parent */
234 int ndist; /* >0 if replicated */
235 int dist[CEPH_MAX_DIRFRAG_REP];
236};
237
238/*
239 * We cache inode xattrs as an encoded blob until they are first used,
240 * at which point we parse them into an rbtree.
241 */
242struct ceph_inode_xattr {
243 struct rb_node node;
244
245 const char *name;
246 int name_len;
247 const char *val;
248 int val_len;
249 int dirty;
250
251 int should_free_name;
252 int should_free_val;
253};
254
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700255/*
256 * Ceph dentry state
257 */
258struct ceph_dentry_info {
259 struct ceph_mds_session *lease_session;
Yan, Zheng97aeb6b2017-11-27 10:47:46 +0800260 int lease_shared_gen;
261 u32 lease_gen;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700262 u32 lease_seq;
263 unsigned long lease_renew_after, lease_renew_from;
264 struct list_head lru;
265 struct dentry *dentry;
Miklos Szeredi9b16f03c2016-06-22 16:35:04 +0200266 unsigned long time;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700267 u64 offset;
268};
269
Sage Weilde576062009-10-06 11:31:07 -0700270struct ceph_inode_xattrs_info {
271 /*
272 * (still encoded) xattr blob. we avoid the overhead of parsing
273 * this until someone actually calls getxattr, etc.
274 *
275 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
276 * NULL means we don't know.
277 */
278 struct ceph_buffer *blob, *prealloc_blob;
279
280 struct rb_root index;
281 bool dirty;
282 int count;
283 int names_size;
284 int vals_size;
285 u64 version, index_version;
286};
287
288/*
289 * Ceph inode.
290 */
Sage Weilde576062009-10-06 11:31:07 -0700291struct ceph_inode_info {
292 struct ceph_vino i_vino; /* ceph ino + snap */
293
Sage Weilbe655592011-11-30 09:47:09 -0800294 spinlock_t i_ceph_lock;
295
Sage Weilde576062009-10-06 11:31:07 -0700296 u64 i_version;
Yan, Zheng31c542a2014-11-14 21:41:55 +0800297 u64 i_inline_version;
Sage Weilde576062009-10-06 11:31:07 -0700298 u32 i_time_warp_seq;
299
300 unsigned i_ceph_flags;
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800301 atomic64_t i_release_count;
302 atomic64_t i_ordered_count;
303 atomic64_t i_complete_seq[2];
Sage Weilde576062009-10-06 11:31:07 -0700304
Sage Weil6c0f3af2010-11-16 11:14:34 -0800305 struct ceph_dir_layout i_dir_layout;
Sage Weilde576062009-10-06 11:31:07 -0700306 struct ceph_file_layout i_layout;
307 char *i_symlink;
308
309 /* for dirs */
310 struct timespec i_rctime;
311 u64 i_rbytes, i_rfiles, i_rsubdirs;
312 u64 i_files, i_subdirs;
Sage Weilde576062009-10-06 11:31:07 -0700313
Luis Henriquesfb18a572018-01-05 10:47:18 +0000314 /* quotas */
315 u64 i_max_bytes, i_max_files;
316
Sage Weilde576062009-10-06 11:31:07 -0700317 struct rb_root i_fragtree;
Yan, Zheng1b1bc162016-05-04 11:40:30 +0800318 int i_fragtree_nsplits;
Sage Weilde576062009-10-06 11:31:07 -0700319 struct mutex i_fragtree_mutex;
320
321 struct ceph_inode_xattrs_info i_xattrs;
322
Sage Weilbe655592011-11-30 09:47:09 -0800323 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
Sage Weilde576062009-10-06 11:31:07 -0700324 * s_mutex. */
325 struct rb_root i_caps; /* cap list */
326 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
327 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
328 struct list_head i_dirty_item, i_flushing_item;
Sage Weilde576062009-10-06 11:31:07 -0700329 /* we need to track cap writeback on a per-cap-bit basis, to allow
330 * overlapping, pipelined cap flushes to the mds. we can probably
331 * reduce the tid to 8 bits if we're concerned about inode size. */
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800332 struct ceph_cap_flush *i_prealloc_cap_flush;
Yan, Zhenge4500b52016-07-06 11:12:56 +0800333 struct list_head i_cap_flush_list;
Sage Weilde576062009-10-06 11:31:07 -0700334 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
335 unsigned long i_hold_caps_min; /* jiffies */
336 unsigned long i_hold_caps_max; /* jiffies */
337 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
Sage Weilde576062009-10-06 11:31:07 -0700338 struct ceph_cap_reservation i_cap_migration_resv;
339 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
Sage Weil7d8cb262010-08-24 08:44:16 -0700340 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
341 dirty|flushing caps */
Sage Weilde576062009-10-06 11:31:07 -0700342 unsigned i_snap_caps; /* cap bits for snapped files */
343
Yan, Zheng774a6a12016-06-06 16:01:39 +0800344 int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */
Sage Weilde576062009-10-06 11:31:07 -0700345
Yan, Zhengb0d7c222013-08-12 21:42:15 -0700346 struct mutex i_truncate_mutex;
Sage Weilde576062009-10-06 11:31:07 -0700347 u32 i_truncate_seq; /* last truncate to smaller size */
348 u64 i_truncate_size; /* and the size we last truncated down to */
349 int i_truncate_pending; /* still need to call vmtruncate */
350
351 u64 i_max_size; /* max file size authorized by mds */
352 u64 i_reported_size; /* (max_)size reported to or requested of mds */
353 u64 i_wanted_max_size; /* offset we'd like to write too */
354 u64 i_requested_max_size; /* max_size we've requested */
355
356 /* held references to caps */
357 int i_pin_ref;
Henry C Changd3d07202011-05-11 10:29:54 +0000358 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref;
Sage Weilde576062009-10-06 11:31:07 -0700359 int i_wrbuffer_ref, i_wrbuffer_ref_head;
Yan, Zheng89aa5932017-09-08 15:23:18 +0800360 atomic_t i_filelock_ref;
Yan, Zheng97aeb6b2017-11-27 10:47:46 +0800361 atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */
Sage Weilcd045cb2010-11-04 11:05:05 -0700362 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
Sage Weilde576062009-10-06 11:31:07 -0700363 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
364
Sage Weilde576062009-10-06 11:31:07 -0700365 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
Yan, Zheng68cd5b4b2015-10-27 18:36:06 +0800366 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
Sage Weilde576062009-10-06 11:31:07 -0700367 spinlock_t i_unsafe_lock;
368
369 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
370 int i_snap_realm_counter; /* snap realm (if caps) */
371 struct list_head i_snap_realm_item;
372 struct list_head i_snap_flush_item;
373
374 struct work_struct i_wb_work; /* writeback work */
375 struct work_struct i_pg_inv_work; /* page invalidation work */
376
377 struct work_struct i_vmtruncate_work;
378
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400379#ifdef CONFIG_CEPH_FSCACHE
380 struct fscache_cookie *fscache;
Yan, Zhengf7f7e7a2016-05-18 20:31:55 +0800381 u32 i_fscache_gen;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400382#endif
Sage Weilde576062009-10-06 11:31:07 -0700383 struct inode vfs_inode; /* at end */
384};
385
386static inline struct ceph_inode_info *ceph_inode(struct inode *inode)
387{
Noah Watkinsfbbccec2009-10-28 11:54:49 -0700388 return container_of(inode, struct ceph_inode_info, vfs_inode);
Sage Weilde576062009-10-06 11:31:07 -0700389}
390
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800391static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode)
392{
393 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
394}
395
396static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb)
397{
398 return (struct ceph_fs_client *)sb->s_fs_info;
399}
400
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700401static inline struct ceph_vino ceph_vino(struct inode *inode)
402{
403 return ceph_inode(inode)->i_vino;
404}
405
406/*
407 * ino_t is <64 bits on many architectures, blech.
408 *
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800409 * i_ino (kernel inode) st_ino (userspace)
410 * i386 32 32
411 * x86_64+ino32 64 32
412 * x86_64 64 64
413 */
Amon Ott3310f752011-10-20 13:04:07 -0700414static inline u32 ceph_ino_to_ino32(__u64 vino)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800415{
Amon Ott3310f752011-10-20 13:04:07 -0700416 u32 ino = vino & 0xffffffff;
417 ino ^= vino >> 32;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800418 if (!ino)
Amon Otta661fc52012-01-23 09:25:23 -0800419 ino = 2;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800420 return ino;
421}
422
423/*
424 * kernel i_ino value
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700425 */
426static inline ino_t ceph_vino_to_ino(struct ceph_vino vino)
427{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700428#if BITS_PER_LONG == 32
Amon Ott3310f752011-10-20 13:04:07 -0700429 return ceph_ino_to_ino32(vino.ino);
430#else
431 return (ino_t)vino.ino;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700432#endif
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700433}
434
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800435/*
436 * user-visible ino (stat, filldir)
437 */
438#if BITS_PER_LONG == 32
439static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
440{
441 return ino;
442}
443#else
444static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino)
445{
446 if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))
447 ino = ceph_ino_to_ino32(ino);
448 return ino;
449}
450#endif
451
452
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700453/* for printf-style formatting */
454#define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
455
456static inline u64 ceph_ino(struct inode *inode)
457{
458 return ceph_inode(inode)->i_vino.ino;
459}
460static inline u64 ceph_snap(struct inode *inode)
461{
462 return ceph_inode(inode)->i_vino.snap;
463}
464
465static inline int ceph_ino_compare(struct inode *inode, void *data)
466{
467 struct ceph_vino *pvino = (struct ceph_vino *)data;
468 struct ceph_inode_info *ci = ceph_inode(inode);
469 return ci->i_vino.ino == pvino->ino &&
470 ci->i_vino.snap == pvino->snap;
471}
472
473static inline struct inode *ceph_find_inode(struct super_block *sb,
474 struct ceph_vino vino)
475{
476 ino_t t = ceph_vino_to_ino(vino);
477 return ilookup5(sb, t, ceph_ino_compare, &vino);
478}
479
480
481/*
482 * Ceph inode.
483 */
Yan, Zheng10183a62015-04-27 15:33:28 +0800484#define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
485#define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */
486#define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
487#define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */
488#define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */
489#define CEPH_I_POOL_RD (1 << 5) /* can read from pool */
490#define CEPH_I_POOL_WR (1 << 6) /* can write to pool */
Yan, Zheng315f2402016-03-07 10:34:50 +0800491#define CEPH_I_SEC_INITED (1 << 7) /* security initialized */
Yan, Zheng77310322016-04-08 15:27:16 +0800492#define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */
Yan, Zheng13c2b572016-07-05 16:45:21 +0800493#define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */
Yan, Zheng70220ac2016-07-06 16:21:30 +0800494#define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */
Jeff Layton26544c622017-04-04 08:39:46 -0400495#define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */
Yan, Zhengb3f8d682017-09-11 10:58:55 +0800496#define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */
497
Jeff Layton26544c622017-04-04 08:39:46 -0400498
499/*
500 * We set the ERROR_WRITE bit when we start seeing write errors on an inode
501 * and then clear it when they start succeeding. Note that we do a lockless
502 * check first, and only take the lock if it looks like it needs to be changed.
503 * The write submission code just takes this as a hint, so we're not too
504 * worried if a few slip through in either direction.
505 */
506static inline void ceph_set_error_write(struct ceph_inode_info *ci)
507{
508 if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
509 spin_lock(&ci->i_ceph_lock);
510 ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
511 spin_unlock(&ci->i_ceph_lock);
512 }
513}
514
515static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
516{
517 if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
518 spin_lock(&ci->i_ceph_lock);
519 ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
520 spin_unlock(&ci->i_ceph_lock);
521 }
522}
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700523
Yan, Zheng2f276c52013-03-13 19:44:32 +0800524static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800525 long long release_count,
526 long long ordered_count)
Sage Weilde576062009-10-06 11:31:07 -0700527{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800528 smp_mb__before_atomic();
529 atomic64_set(&ci->i_complete_seq[0], release_count);
530 atomic64_set(&ci->i_complete_seq[1], ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700531}
532
Yan, Zheng2f276c52013-03-13 19:44:32 +0800533static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700534{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800535 atomic64_inc(&ci->i_release_count);
536}
537
538static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
539{
540 atomic64_inc(&ci->i_ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700541}
542
Yan, Zheng2f276c52013-03-13 19:44:32 +0800543static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700544{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800545 return atomic64_read(&ci->i_complete_seq[0]) ==
546 atomic64_read(&ci->i_release_count);
Yan, Zheng2f276c52013-03-13 19:44:32 +0800547}
Sage Weilde576062009-10-06 11:31:07 -0700548
Yan, Zheng70db4f32014-10-21 18:09:56 -0700549static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
550{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800551 return atomic64_read(&ci->i_complete_seq[0]) ==
552 atomic64_read(&ci->i_release_count) &&
553 atomic64_read(&ci->i_complete_seq[1]) ==
554 atomic64_read(&ci->i_ordered_count);
Yan, Zheng70db4f32014-10-21 18:09:56 -0700555}
556
Yan, Zheng2f276c52013-03-13 19:44:32 +0800557static inline void ceph_dir_clear_complete(struct inode *inode)
558{
559 __ceph_dir_clear_complete(ceph_inode(inode));
560}
561
Yan, Zheng70db4f32014-10-21 18:09:56 -0700562static inline void ceph_dir_clear_ordered(struct inode *inode)
Yan, Zheng2f276c52013-03-13 19:44:32 +0800563{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800564 __ceph_dir_clear_ordered(ceph_inode(inode));
Sage Weilde576062009-10-06 11:31:07 -0700565}
566
Yan, Zheng70db4f32014-10-21 18:09:56 -0700567static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
568{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800569 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
570 smp_rmb();
Yan, Zheng70db4f32014-10-21 18:09:56 -0700571 return ret;
572}
Sage Weilde576062009-10-06 11:31:07 -0700573
574/* find a specific frag @f */
575extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
576 u32 f);
577
578/*
579 * choose fragment for value @v. copy frag content to pfrag, if leaf
580 * exists
581 */
582extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
583 struct ceph_inode_frag *pfrag,
584 int *found);
585
Sage Weilde576062009-10-06 11:31:07 -0700586static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry)
587{
588 return (struct ceph_dentry_info *)dentry->d_fsdata;
589}
590
Sage Weilde576062009-10-06 11:31:07 -0700591/*
592 * caps helpers
593 */
594static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
595{
596 return !RB_EMPTY_ROOT(&ci->i_caps);
597}
598
599extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
600extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
601extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
602 struct ceph_cap *cap);
603
604static inline int ceph_caps_issued(struct ceph_inode_info *ci)
605{
606 int issued;
Sage Weilbe655592011-11-30 09:47:09 -0800607 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700608 issued = __ceph_caps_issued(ci, NULL);
Sage Weilbe655592011-11-30 09:47:09 -0800609 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700610 return issued;
611}
612
613static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
614 int touch)
615{
616 int r;
Sage Weilbe655592011-11-30 09:47:09 -0800617 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700618 r = __ceph_caps_issued_mask(ci, mask, touch);
Sage Weilbe655592011-11-30 09:47:09 -0800619 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700620 return r;
621}
622
623static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
624{
625 return ci->i_dirty_caps | ci->i_flushing_caps;
626}
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800627extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
628extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
629extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
630 struct ceph_cap_flush **pcf);
Sage Weilde576062009-10-06 11:31:07 -0700631
Yan, Zheng9563f882013-11-22 13:50:45 +0800632extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
633 struct ceph_cap *ocap, int mask);
Sage Weilde576062009-10-06 11:31:07 -0700634extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
635extern int __ceph_caps_used(struct ceph_inode_info *ci);
636
637extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
638
639/*
640 * wanted, by virtue of open file modes AND cap refs (buffered/cached data)
641 */
642static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
643{
644 int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci);
645 if (w & CEPH_CAP_FILE_BUFFER)
646 w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */
647 return w;
648}
649
650/* what the mds thinks we want */
Yan, Zhengc1944fe2017-01-29 22:15:47 +0800651extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
Sage Weilde576062009-10-06 11:31:07 -0700652
Yehuda Sadeh37151662010-06-17 16:16:12 -0700653extern void ceph_caps_init(struct ceph_mds_client *mdsc);
654extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
655extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta);
Zhi Zhange30ee582018-01-24 21:24:33 +0800656extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
Yehuda Sadeh37151662010-06-17 16:16:12 -0700657 struct ceph_cap_reservation *ctx, int need);
658extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc,
659 struct ceph_cap_reservation *ctx);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700660extern void ceph_reservation_status(struct ceph_fs_client *client,
Sage Weilde576062009-10-06 11:31:07 -0700661 int *total, int *avail, int *used,
Sage Weil85ccce42010-02-17 10:02:43 -0800662 int *reserved, int *min);
Sage Weilde576062009-10-06 11:31:07 -0700663
Sage Weilde576062009-10-06 11:31:07 -0700664
Sage Weilde576062009-10-06 11:31:07 -0700665
666/*
667 * we keep buffered readdir results attached to file->private_data
668 */
Sage Weil4918b6d2011-07-26 11:26:07 -0700669#define CEPH_F_SYNC 1
Sage Weil9cfa1092011-07-26 11:26:18 -0700670#define CEPH_F_ATEND 2
Sage Weil4918b6d2011-07-26 11:26:07 -0700671
Sage Weilde576062009-10-06 11:31:07 -0700672struct ceph_file_info {
Sage Weil252c6722011-07-26 11:25:27 -0700673 short fmode; /* initialized on open */
674 short flags; /* CEPH_F_* */
Sage Weilde576062009-10-06 11:31:07 -0700675
Yan, Zheng5d988302017-12-15 11:15:36 +0800676 spinlock_t rw_contexts_lock;
677 struct list_head rw_contexts;
Chengguang Xubb48bd42018-03-13 10:42:44 +0800678};
679
680struct ceph_dir_file_info {
681 struct ceph_file_info file_info;
Yan, Zheng5d988302017-12-15 11:15:36 +0800682
Sage Weilde576062009-10-06 11:31:07 -0700683 /* readdir: position within the dir */
684 u32 frag;
685 struct ceph_mds_request *last_readdir;
Sage Weilde576062009-10-06 11:31:07 -0700686
687 /* readdir: position within a frag */
Yan, Zhengf0494202014-02-27 16:26:24 +0800688 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
Sage Weilde576062009-10-06 11:31:07 -0700689 char *last_name; /* last entry in previous chunk */
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800690 long long dir_release_count;
691 long long dir_ordered_count;
692 int readdir_cache_idx;
Sage Weilde576062009-10-06 11:31:07 -0700693
694 /* used for -o dirstat read() on directory thing */
695 char *dir_info;
696 int dir_info_len;
697};
698
Yan, Zheng5d988302017-12-15 11:15:36 +0800699struct ceph_rw_context {
700 struct list_head list;
701 struct task_struct *thread;
702 int caps;
703};
704
705#define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \
706 struct ceph_rw_context _name = { \
707 .thread = current, \
708 .caps = _caps, \
709 }
710
711static inline void ceph_add_rw_context(struct ceph_file_info *cf,
712 struct ceph_rw_context *ctx)
713{
714 spin_lock(&cf->rw_contexts_lock);
715 list_add(&ctx->list, &cf->rw_contexts);
716 spin_unlock(&cf->rw_contexts_lock);
717}
718
719static inline void ceph_del_rw_context(struct ceph_file_info *cf,
720 struct ceph_rw_context *ctx)
721{
722 spin_lock(&cf->rw_contexts_lock);
723 list_del(&ctx->list);
724 spin_unlock(&cf->rw_contexts_lock);
725}
726
727static inline struct ceph_rw_context*
728ceph_find_rw_context(struct ceph_file_info *cf)
729{
730 struct ceph_rw_context *ctx, *found = NULL;
731 spin_lock(&cf->rw_contexts_lock);
732 list_for_each_entry(ctx, &cf->rw_contexts, list) {
733 if (ctx->thread == current) {
734 found = ctx;
735 break;
736 }
737 }
738 spin_unlock(&cf->rw_contexts_lock);
739 return found;
740}
741
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800742struct ceph_readdir_cache_control {
743 struct page *page;
744 struct dentry **dentries;
745 int index;
746};
Sage Weilde576062009-10-06 11:31:07 -0700747
748/*
Sage Weilde576062009-10-06 11:31:07 -0700749 * A "snap realm" describes a subset of the file hierarchy sharing
750 * the same set of snapshots that apply to it. The realms themselves
751 * are organized into a hierarchy, such that children inherit (some of)
752 * the snapshots of their parents.
753 *
754 * All inodes within the realm that have capabilities are linked into a
755 * per-realm list.
756 */
757struct ceph_snap_realm {
758 u64 ino;
Luis Henriquese3161f12018-01-12 17:19:28 +0000759 struct inode *inode;
Sage Weilde576062009-10-06 11:31:07 -0700760 atomic_t nref;
Sage Weila105f002010-02-15 14:37:55 -0800761 struct rb_node node;
762
Sage Weilde576062009-10-06 11:31:07 -0700763 u64 created, seq;
764 u64 parent_ino;
765 u64 parent_since; /* snapid when our current parent became so */
766
767 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
Alex Elderaa711ee32012-07-13 20:35:11 -0500768 u32 num_prior_parent_snaps; /* had prior to parent_since */
Sage Weilde576062009-10-06 11:31:07 -0700769 u64 *snaps; /* snaps specific to this realm */
Alex Elderaa711ee32012-07-13 20:35:11 -0500770 u32 num_snaps;
Sage Weilde576062009-10-06 11:31:07 -0700771
772 struct ceph_snap_realm *parent;
773 struct list_head children; /* list of child realms */
774 struct list_head child_item;
775
776 struct list_head empty_item; /* if i have ref==0 */
777
Sage Weilae00d4f2010-09-16 16:26:51 -0700778 struct list_head dirty_item; /* if realm needs new context */
779
Sage Weilde576062009-10-06 11:31:07 -0700780 /* the current set of snaps for this realm */
781 struct ceph_snap_context *cached_context;
782
783 struct list_head inodes_with_caps;
784 spinlock_t inodes_with_caps_lock;
785};
786
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700787static inline int default_congestion_kb(void)
Sage Weilde576062009-10-06 11:31:07 -0700788{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700789 int congestion_kb;
790
791 /*
792 * Copied from NFS
793 *
794 * congestion size, scale with available memory.
795 *
796 * 64MB: 8192k
797 * 128MB: 11585k
798 * 256MB: 16384k
799 * 512MB: 23170k
800 * 1GB: 32768k
801 * 2GB: 46340k
802 * 4GB: 65536k
803 * 8GB: 92681k
804 * 16GB: 131072k
805 *
806 * This allows larger machines to have larger/more transfers.
807 * Limit the default to 256M
808 */
809 congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10);
810 if (congestion_kb > 256*1024)
811 congestion_kb = 256*1024;
812
813 return congestion_kb;
Sage Weilde576062009-10-06 11:31:07 -0700814}
815
816
817
818/* snap.c */
819struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
820 u64 ino);
821extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
822 struct ceph_snap_realm *realm);
823extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
824 struct ceph_snap_realm *realm);
825extern int ceph_update_snap_trace(struct ceph_mds_client *m,
Yan, Zheng982d6012014-12-23 15:30:54 +0800826 void *p, void *e, bool deletion,
827 struct ceph_snap_realm **realm_ret);
Sage Weilde576062009-10-06 11:31:07 -0700828extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
Sage Weil2600d2d2010-02-22 15:12:16 -0800829 struct ceph_mds_session *session,
Sage Weilde576062009-10-06 11:31:07 -0700830 struct ceph_msg *msg);
Sage Weilfc837c8f2010-04-13 11:41:22 -0700831extern void ceph_queue_cap_snap(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -0700832extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
833 struct ceph_cap_snap *capsnap);
834extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
835
836/*
837 * a cap_snap is "pending" if it is still awaiting an in-progress
838 * sync write (that may/may not still update size, mtime, etc.).
839 */
840static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
841{
842 return !list_empty(&ci->i_cap_snaps) &&
Yan, Zheng86056092015-05-01 16:57:16 +0800843 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
844 ci_item)->writing;
Sage Weilde576062009-10-06 11:31:07 -0700845}
846
Sage Weilde576062009-10-06 11:31:07 -0700847/* inode.c */
848extern const struct inode_operations ceph_file_iops;
849
850extern struct inode *ceph_alloc_inode(struct super_block *sb);
851extern void ceph_destroy_inode(struct inode *inode);
Yan, Zheng9f12bd12013-09-20 19:55:31 +0800852extern int ceph_drop_inode(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700853
854extern struct inode *ceph_get_inode(struct super_block *sb,
855 struct ceph_vino vino);
856extern struct inode *ceph_get_snapdir(struct inode *parent);
857extern int ceph_fill_file_size(struct inode *inode, int issued,
858 u32 truncate_seq, u64 truncate_size, u64 size);
859extern void ceph_fill_file_time(struct inode *inode, int issued,
860 u64 time_warp_seq, struct timespec *ctime,
861 struct timespec *mtime, struct timespec *atime);
862extern int ceph_fill_trace(struct super_block *sb,
Jeff Laytonf5a03b02017-01-31 11:06:13 -0500863 struct ceph_mds_request *req);
Sage Weilde576062009-10-06 11:31:07 -0700864extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
865 struct ceph_mds_session *session);
866
867extern int ceph_inode_holds_cap(struct inode *inode, int mask);
868
Yan, Zhengefb0ca72017-05-22 12:03:32 +0800869extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
Yan, Zhengb415bf42013-07-02 12:40:19 +0800870extern void __ceph_do_pending_vmtruncate(struct inode *inode);
Sage Weil3c6f6b72010-02-09 15:24:44 -0800871extern void ceph_queue_vmtruncate(struct inode *inode);
872
873extern void ceph_queue_invalidate(struct inode *inode);
874extern void ceph_queue_writeback(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700875
Yan, Zheng01deead2014-11-14 21:56:29 +0800876extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
877 int mask, bool force);
878static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
879{
880 return __ceph_do_getattr(inode, NULL, mask, force);
881}
Al Viro10556cb22011-06-20 19:28:19 -0400882extern int ceph_permission(struct inode *inode, int mask);
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +0200883extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
Sage Weilde576062009-10-06 11:31:07 -0700884extern int ceph_setattr(struct dentry *dentry, struct iattr *attr);
David Howellsa528d352017-01-31 16:46:22 +0000885extern int ceph_getattr(const struct path *path, struct kstat *stat,
886 u32 request_mask, unsigned int flags);
Sage Weilde576062009-10-06 11:31:07 -0700887
888/* xattr.c */
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +0200889int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800890ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
Sage Weilde576062009-10-06 11:31:07 -0700891extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
Sage Weilde576062009-10-06 11:31:07 -0700892extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci);
893extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
Alex Elder3ce6cd12012-01-23 15:49:28 -0600894extern void __init ceph_xattr_init(void);
895extern void ceph_xattr_exit(void);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800896extern const struct xattr_handler *ceph_xattr_handlers[];
Sage Weilde576062009-10-06 11:31:07 -0700897
Yan, Zheng315f2402016-03-07 10:34:50 +0800898#ifdef CONFIG_SECURITY
899extern bool ceph_security_xattr_deadlock(struct inode *in);
900extern bool ceph_security_xattr_wanted(struct inode *in);
901#else
902static inline bool ceph_security_xattr_deadlock(struct inode *in)
903{
904 return false;
905}
906static inline bool ceph_security_xattr_wanted(struct inode *in)
907{
908 return false;
909}
910#endif
911
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800912/* acl.c */
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800913struct ceph_acls_info {
914 void *default_acl;
915 void *acl;
916 struct ceph_pagelist *pagelist;
917};
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800918
919#ifdef CONFIG_CEPH_FS_POSIX_ACL
920
921struct posix_acl *ceph_get_acl(struct inode *, int);
Sage Weil72466d02014-01-29 06:22:25 -0800922int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800923int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
924 struct ceph_acls_info *info);
925void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info);
926void ceph_release_acls_info(struct ceph_acls_info *info);
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -0800927
928static inline void ceph_forget_all_cached_acls(struct inode *inode)
929{
930 forget_all_cached_acls(inode);
931}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800932
933#else
934
935#define ceph_get_acl NULL
Sage Weil72466d02014-01-29 06:22:25 -0800936#define ceph_set_acl NULL
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800937
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800938static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
939 struct ceph_acls_info *info)
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800940{
941 return 0;
942}
Yan, Zhengb1ee94a2014-09-16 20:35:17 +0800943static inline void ceph_init_inode_acls(struct inode *inode,
944 struct ceph_acls_info *info)
945{
946}
947static inline void ceph_release_acls_info(struct ceph_acls_info *info)
948{
949}
Guangliang Zhao7221fe42013-11-11 15:18:03 +0800950static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
951{
952 return 0;
953}
954
955static inline void ceph_forget_all_cached_acls(struct inode *inode)
956{
957}
958
959#endif
960
Sage Weilde576062009-10-06 11:31:07 -0700961/* caps.c */
962extern const char *ceph_cap_string(int c);
963extern void ceph_handle_caps(struct ceph_mds_session *session,
964 struct ceph_msg *msg);
Yan, Zhengd9df2782014-04-18 09:57:11 +0800965extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
966 struct ceph_cap_reservation *ctx);
967extern void ceph_add_cap(struct inode *inode,
968 struct ceph_mds_session *session, u64 cap_id,
969 int fmode, unsigned issued, unsigned wanted,
970 unsigned cap, unsigned seq, u64 realmino, int flags,
971 struct ceph_cap **new_cap);
Yan, Zhenga096b092013-09-22 10:15:58 +0800972extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
Yehuda Sadeh37151662010-06-17 16:16:12 -0700973extern void ceph_put_cap(struct ceph_mds_client *mdsc,
974 struct ceph_cap *cap);
Yan, Zheng9215aee2013-11-30 12:47:41 +0800975extern int ceph_is_any_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700976
977extern void ceph_queue_caps_release(struct inode *inode);
Stephen Rothwellf1a3d572010-01-18 11:53:08 +1100978extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
Josef Bacik02c24a82011-07-16 20:44:56 -0400979extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
980 int datasync);
Yan, Zhenge548e9b2015-06-10 15:17:56 +0800981extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
982 struct ceph_mds_session *session);
Sage Weilde576062009-10-06 11:31:07 -0700983extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
984 struct ceph_mds_session *session);
Greg Farnum2bc50252010-06-30 12:44:34 -0700985extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
986 int mds);
Sage Weilde576062009-10-06 11:31:07 -0700987extern int ceph_get_cap_mds(struct inode *inode);
988extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
989extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
990extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
991 struct ceph_snap_context *snapc);
Yan, Zhenged9b4302016-07-05 21:08:07 +0800992extern void ceph_flush_snaps(struct ceph_inode_info *ci,
993 struct ceph_mds_session **psession);
Yan, Zhengefb0ca72017-05-22 12:03:32 +0800994extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -0700995extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
996 struct ceph_mds_session *session);
Sage Weilafcdaea2009-10-14 14:27:38 -0700997extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
998extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
Zhi Zhang6ef0bc62018-01-24 21:24:33 +0800999extern int ceph_drop_caps_for_unlink(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -07001000extern int ceph_encode_inode_release(void **p, struct inode *inode,
1001 int mds, int drop, int unless, int force);
1002extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
Jeff Laytonca6c8ae2016-12-15 08:37:59 -05001003 struct inode *dir,
Sage Weilde576062009-10-06 11:31:07 -07001004 int mds, int drop, int unless);
1005
1006extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
Yan, Zheng31c542a2014-11-14 21:41:55 +08001007 loff_t endoff, int *got, struct page **pinned_page);
Yan, Zheng2b1ac852016-10-25 10:51:55 +08001008extern int ceph_try_get_caps(struct ceph_inode_info *ci,
1009 int need, int want, int *got);
Sage Weilde576062009-10-06 11:31:07 -07001010
1011/* for counting open files by mode */
Yan, Zheng774a6a12016-06-06 16:01:39 +08001012extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode);
Sage Weilde576062009-10-06 11:31:07 -07001013extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
1014
1015/* addr.c */
1016extern const struct address_space_operations ceph_aops;
1017extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
Yan, Zheng10183a62015-04-27 15:33:28 +08001018extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
1019extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
1020extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
Sage Weilde576062009-10-06 11:31:07 -07001021
1022/* file.c */
1023extern const struct file_operations ceph_file_fops;
Alex Elder9e0eb852013-02-06 13:11:38 -06001024
Yan, Zheng77310322016-04-08 15:27:16 +08001025extern int ceph_renew_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -07001026extern int ceph_open(struct inode *inode, struct file *file);
Sage Weil5ef50c32012-07-31 11:27:36 -07001027extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
1028 struct file *file, unsigned flags, umode_t mode,
1029 int *opened);
Sage Weilde576062009-10-06 11:31:07 -07001030extern int ceph_release(struct inode *inode, struct file *filp);
Yan, Zheng31c542a2014-11-14 21:41:55 +08001031extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
1032 char *data, size_t len);
Ilya Dryomov55f2a042017-02-13 14:44:19 +01001033
Sage Weilde576062009-10-06 11:31:07 -07001034/* dir.c */
1035extern const struct file_operations ceph_dir_fops;
Yan, Zheng38c48b52015-01-14 13:46:04 +08001036extern const struct file_operations ceph_snapdir_fops;
Sage Weilde576062009-10-06 11:31:07 -07001037extern const struct inode_operations ceph_dir_iops;
Yan, Zheng38c48b52015-01-14 13:46:04 +08001038extern const struct inode_operations ceph_snapdir_iops;
Al Viro18fc8ab2016-10-28 21:52:50 -04001039extern const struct dentry_operations ceph_dentry_ops;
Sage Weilde576062009-10-06 11:31:07 -07001040
Yan, Zhengf3c4ebe2016-04-29 11:27:30 +08001041extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
Sage Weilde576062009-10-06 11:31:07 -07001042extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
Sage Weil468640e2011-07-26 11:28:11 -07001043extern int ceph_handle_snapdir(struct ceph_mds_request *req,
1044 struct dentry *dentry, int err);
Sage Weilde576062009-10-06 11:31:07 -07001045extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
1046 struct dentry *dentry, int err);
1047
1048extern void ceph_dentry_lru_add(struct dentry *dn);
1049extern void ceph_dentry_lru_touch(struct dentry *dn);
1050extern void ceph_dentry_lru_del(struct dentry *dn);
Sage Weil81a6cf22010-05-14 09:35:38 -07001051extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
Sage Weile5f86dc2011-07-26 11:30:55 -07001052extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
Yan, Zhengfdd4e152015-06-16 20:48:56 +08001053extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
Sage Weilde576062009-10-06 11:31:07 -07001054
Sage Weilde576062009-10-06 11:31:07 -07001055/* ioctl.c */
1056extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1057
1058/* export.c */
1059extern const struct export_operations ceph_export_ops;
1060
Greg Farnum40819f62010-08-02 15:34:23 -07001061/* locks.c */
Yan, Zhengeb13e832014-03-09 23:16:40 +08001062extern __init void ceph_flock_init(void);
Greg Farnum40819f62010-08-02 15:34:23 -07001063extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
1064extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
1065extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
Jim Schutt39be95e2013-05-15 13:03:35 -05001066extern int ceph_encode_locks_to_buffer(struct inode *inode,
1067 struct ceph_filelock *flocks,
1068 int num_fcntl_locks,
1069 int num_flock_locks);
1070extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
1071 struct ceph_pagelist *pagelist,
1072 int num_fcntl_locks, int num_flock_locks);
Greg Farnum40819f62010-08-02 15:34:23 -07001073
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001074/* debugfs.c */
1075extern int ceph_fs_debugfs_init(struct ceph_fs_client *client);
1076extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
1077
Luis Henriquesfb18a572018-01-05 10:47:18 +00001078/* quota.c */
Luis Henriquesd557c482018-01-12 17:19:29 +00001079static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci)
1080{
1081 return ci->i_max_files || ci->i_max_bytes;
1082}
1083
1084extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
1085
1086static inline void __ceph_update_quota(struct ceph_inode_info *ci,
1087 u64 max_bytes, u64 max_files)
1088{
1089 bool had_quota, has_quota;
1090 had_quota = __ceph_has_any_quota(ci);
1091 ci->i_max_bytes = max_bytes;
1092 ci->i_max_files = max_files;
1093 has_quota = __ceph_has_any_quota(ci);
1094
1095 if (had_quota != has_quota)
1096 ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota);
1097}
1098
Luis Henriquesfb18a572018-01-05 10:47:18 +00001099extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
1100 struct ceph_mds_session *session,
1101 struct ceph_msg *msg);
Luis Henriquesb7a29212018-01-05 10:47:19 +00001102extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
Luis Henriquescafe21a2018-01-05 10:47:20 +00001103extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
Luis Henriques2b838452018-01-05 10:47:21 +00001104extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
1105 loff_t newlen);
Luis Henriques1ab302a2018-01-05 10:47:22 +00001106extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
1107 loff_t newlen);
Luis Henriques9122eed2018-01-31 10:53:13 +00001108extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
1109 struct kstatfs *buf);
Luis Henriquesfb18a572018-01-05 10:47:18 +00001110
Sage Weilde576062009-10-06 11:31:07 -07001111#endif /* _FS_CEPH_SUPER_H */