blob: ac331aa07cfa595a1b088998c43675fad27c7978 [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>
Jeff Layton668959a2019-08-06 09:07:51 -040019#include <linux/security.h>
Sage Weilde576062009-10-06 11:31:07 -070020
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070021#include <linux/ceph/libceph.h>
Sage Weilde576062009-10-06 11:31:07 -070022
Milosz Tanski99ccbd22013-08-21 17:29:54 -040023#ifdef CONFIG_CEPH_FSCACHE
Jeff Layton7c46b312021-01-21 16:27:14 -050024#define FSCACHE_USE_NEW_IO_API
Milosz Tanski99ccbd22013-08-21 17:29:54 -040025#include <linux/fscache.h>
26#endif
27
Sage Weilde576062009-10-06 11:31:07 -070028/* f_type in struct statfs */
29#define CEPH_SUPER_MAGIC 0x00c36400
30
31/* large granularity for statfs utilization stats to facilitate
32 * large volume sizes on 32-bit machines. */
Sage Weil92a49fb2013-02-22 15:31:00 -080033#define CEPH_BLOCK_SHIFT 22 /* 4 MB */
Sage Weilde576062009-10-06 11:31:07 -070034#define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT)
35
Ilya Dryomov0b98acd2020-09-14 13:39:19 +020036#define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blocklisted */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070037#define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */
38#define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */
39#define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */
Yehuda Sadehad1fee92011-01-21 16:44:03 -080040#define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */
Sage Weila40dc6c2012-01-10 09:12:55 -080041#define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */
Milosz Tanski99ccbd22013-08-21 17:29:54 -040042#define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */
Yan, Zheng10183a62015-04-27 15:33:28 +080043#define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */
Yan, Zhenge9e427f2016-11-10 16:02:06 +080044#define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */
Luis Henriques9122eed2018-01-31 10:53:13 +000045#define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */
Luis Henriquesea4cdc52018-10-15 16:46:00 +010046#define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */
Jeff Layton2ccb4542019-04-02 15:35:56 -040047#define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */
Sage Weil6a259382010-07-07 09:06:08 -070048
Luis Henriques6f9718f2018-12-10 10:23:12 +000049#define CEPH_MOUNT_OPT_DEFAULT \
50 (CEPH_MOUNT_OPT_DCACHE | \
Jeff Laytonf7a67b42021-08-09 11:55:15 -040051 CEPH_MOUNT_OPT_NOCOPYFROM | \
52 CEPH_MOUNT_OPT_ASYNC_DIROPS)
Sage Weilde576062009-10-06 11:31:07 -070053
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070054#define ceph_set_mount_opt(fsc, opt) \
Jeff Layton2ccb4542019-04-02 15:35:56 -040055 (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt
56#define ceph_clear_mount_opt(fsc, opt) \
57 (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070058#define ceph_test_mount_opt(fsc, opt) \
59 (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
Sage Weilde576062009-10-06 11:31:07 -070060
Yan, Zhengaa187922017-07-11 15:56:09 +080061/* max size of osd read request, limited by libceph */
62#define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN
Yan, Zheng95cca2b2017-07-11 17:34:46 +080063/* osd has a configurable limitaion of max write size.
64 * CEPH_MSG_MAX_DATA_LEN should be small enough. */
65#define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN
Andreas Gerstmayr7c94ba22017-01-10 14:17:56 +010066#define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070067#define CEPH_MAX_READDIR_DEFAULT 1024
68#define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024)
69#define CEPH_SNAPDIRNAME_DEFAULT ".snap"
Sage Weilde576062009-10-06 11:31:07 -070070
Yan, Zheng4214fb12017-07-11 18:49:44 +080071/*
72 * Delay telling the MDS we no longer want caps, in case we reopen
73 * the file. Delay a minimum amount of time, even if we send a cap
74 * message for some other reason. Otherwise, take the oppotunity to
75 * update the mds to avoid sending another message later.
76 */
77#define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */
78#define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */
79
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070080struct ceph_mount_options {
Jeff Laytonad8c28a2019-09-09 15:58:55 -040081 unsigned int flags;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070082
Jeff Laytonad8c28a2019-09-09 15:58:55 -040083 unsigned int wsize; /* max write size */
84 unsigned int rsize; /* max read size */
85 unsigned int rasize; /* max readahead */
86 unsigned int congestion_kb; /* max writeback in flight */
87 unsigned int caps_wanted_delay_min, caps_wanted_delay_max;
Yan, Zhengfe330322019-02-01 14:57:15 +080088 int caps_max;
Jeff Laytonad8c28a2019-09-09 15:58:55 -040089 unsigned int max_readdir; /* max readdir result (entries) */
90 unsigned int max_readdir_bytes; /* max readdir result (bytes) */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070091
92 /*
93 * everything above this point can be memcmp'd; everything below
94 * is handled in compare_mount_options()
95 */
96
Sage Weilde576062009-10-06 11:31:07 -070097 char *snapdir_name; /* default ".snap" */
Yan, Zheng430afba2016-07-08 11:25:38 +080098 char *mds_namespace; /* default NULL */
Ilya Dryomovb27a9392020-02-10 22:51:08 +010099 char *server_path; /* default NULL (means "/") */
Yan, Zheng1d8f8362017-06-27 11:57:56 +0800100 char *fscache_uniq; /* default NULL */
Sage Weilde576062009-10-06 11:31:07 -0700101};
102
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700103struct ceph_fs_client {
Sage Weilde576062009-10-06 11:31:07 -0700104 struct super_block *sb;
105
Xiubo Li18f473b2020-07-16 10:05:57 -0400106 struct list_head metric_wakeup;
107
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700108 struct ceph_mount_options *mount_options;
109 struct ceph_client *client;
110
Jeff Laytonaa5c7912020-10-06 08:38:20 -0400111 int mount_state;
Yan, Zheng81f148a2019-07-25 20:16:46 +0800112
Ilya Dryomov0b98acd2020-09-14 13:39:19 +0200113 bool blocklisted;
Yan, Zheng131d7eb2019-07-25 20:16:47 +0800114
Luis Henriques78beb0f2020-01-08 10:03:53 +0000115 bool have_copy_from2;
116
Yan, Zheng81f148a2019-07-25 20:16:46 +0800117 u32 filp_gen;
Chengguang Xu719784b2018-07-19 22:15:24 +0800118 loff_t max_file_size;
Sage Weil85ccce42010-02-17 10:02:43 -0800119
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700120 struct ceph_mds_client *mdsc;
Sage Weilde576062009-10-06 11:31:07 -0700121
Yan, Zhenge3ec8d62019-01-14 17:21:19 +0800122 atomic_long_t writeback_count;
123
Yan, Zheng1cf89a82019-05-18 11:18:44 +0800124 struct workqueue_struct *inode_wq;
Yan, Zhenge3ec8d62019-01-14 17:21:19 +0800125 struct workqueue_struct *cap_wq;
Sage Weilde576062009-10-06 11:31:07 -0700126
Sage Weil07433042009-11-18 16:50:41 -0800127#ifdef CONFIG_DEBUG_FS
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700128 struct dentry *debugfs_dentry_lru, *debugfs_caps;
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800129 struct dentry *debugfs_congestion_kb;
Sage Weil06edf042009-12-15 14:44:32 -0800130 struct dentry *debugfs_bdi;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700131 struct dentry *debugfs_mdsc, *debugfs_mdsmap;
Xiubo Li247b1f12020-11-11 09:29:39 +0800132 struct dentry *debugfs_status;
John Spray14ed9702014-09-12 16:58:49 +0100133 struct dentry *debugfs_mds_sessions;
Luís Henriquescbed4ff2021-10-27 11:01:30 +0100134 struct dentry *debugfs_metrics_dir;
Sage Weil07433042009-11-18 16:50:41 -0800135#endif
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400136
137#ifdef CONFIG_CEPH_FSCACHE
138 struct fscache_cookie *fscache;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400139#endif
Sage Weilde576062009-10-06 11:31:07 -0700140};
141
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700142
Sage Weilde576062009-10-06 11:31:07 -0700143/*
144 * File i/o capability. This tracks shared state with the metadata
145 * server that allows us to cache or writeback attributes or to read
146 * and write data. For any given inode, we should have one or more
147 * capabilities, one issued by each metadata server, and our
148 * cumulative access is the OR of all issued capabilities.
149 *
150 * Each cap is referenced by the inode's i_caps rbtree and by per-mds
151 * session capability lists.
152 */
153struct ceph_cap {
154 struct ceph_inode_info *ci;
155 struct rb_node ci_node; /* per-ci cap tree */
156 struct ceph_mds_session *session;
157 struct list_head session_caps; /* per-session caplist */
Sage Weilde576062009-10-06 11:31:07 -0700158 u64 cap_id; /* unique cap id (mds provided) */
Yan, Zheng745a8e32015-05-14 17:22:42 +0800159 union {
160 /* in-use caps */
161 struct {
162 int issued; /* latest, from the mds */
163 int implemented; /* implemented superset of
164 issued (for revocation) */
Jeff Laytonc74d79a2020-10-06 12:24:19 -0400165 int mds; /* mds index for this cap */
166 int mds_wanted; /* caps wanted from this mds */
Yan, Zheng745a8e32015-05-14 17:22:42 +0800167 };
168 /* caps to release */
169 struct {
170 u64 cap_ino;
171 int queue_release;
172 };
173 };
Sage Weil685f9a5d2009-11-09 12:05:48 -0800174 u32 seq, issue_seq, mseq;
175 u32 cap_gen; /* active/stale cycle */
Sage Weilde576062009-10-06 11:31:07 -0700176 unsigned long last_used;
177 struct list_head caps_item;
178};
179
Yan, Zhenga0d93e32020-03-05 20:21:01 +0800180#define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */
181#define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */
182#define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */
Sage Weilde576062009-10-06 11:31:07 -0700183
Yan, Zheng0e294382016-07-04 18:06:41 +0800184struct ceph_cap_flush {
185 u64 tid;
Xiubo Lib2f9fa12021-08-18 21:38:42 +0800186 int caps;
Yan, Zhengc8799fc2016-07-07 15:22:38 +0800187 bool wake; /* wake up flush waiters when finish ? */
Xiubo Lib2f9fa12021-08-18 21:38:42 +0800188 bool is_capsnap; /* true means capsnap */
Yan, Zheng0e294382016-07-04 18:06:41 +0800189 struct list_head g_list; // global
190 struct list_head i_list; // per inode
191};
192
Sage Weilde576062009-10-06 11:31:07 -0700193/*
194 * Snapped cap state that is pending flush to mds. When a snapshot occurs,
195 * we first complete any in-process sync writes and writeback any dirty
196 * data before flushing the snapped state (tracked here) back to the MDS.
197 */
198struct ceph_cap_snap {
Elena Reshetova805692d2017-03-03 11:15:07 +0200199 refcount_t nref;
Yan, Zheng0e294382016-07-04 18:06:41 +0800200 struct list_head ci_item;
Sage Weilde576062009-10-06 11:31:07 -0700201
Yan, Zheng0e294382016-07-04 18:06:41 +0800202 struct ceph_cap_flush cap_flush;
203
204 u64 follows;
Sage Weilde576062009-10-06 11:31:07 -0700205 int issued, dirty;
206 struct ceph_snap_context *context;
207
Al Viro5706b272011-07-26 04:52:22 -0400208 umode_t mode;
Eric W. Biederman05cb11c2013-01-31 02:56:19 -0800209 kuid_t uid;
210 kgid_t gid;
Sage Weilde576062009-10-06 11:31:07 -0700211
Sage Weil4a625be2010-08-22 15:03:56 -0700212 struct ceph_buffer *xattr_blob;
Sage Weilde576062009-10-06 11:31:07 -0700213 u64 xattr_version;
214
215 u64 size;
Jeff Layton176c77c2019-06-06 08:06:40 -0400216 u64 change_attr;
Jeff Laytonec62b892019-05-29 12:23:14 -0400217 struct timespec64 mtime, atime, ctime, btime;
Sage Weilde576062009-10-06 11:31:07 -0700218 u64 time_warp_seq;
Yan, Zheng5f743e42016-11-15 16:04:37 +0800219 u64 truncate_size;
220 u32 truncate_seq;
Sage Weilde576062009-10-06 11:31:07 -0700221 int writing; /* a sync write is still in progress */
222 int dirty_pages; /* dirty pages awaiting writeback */
Yan, Zhenge20d2582014-11-14 22:39:13 +0800223 bool inline_data;
Yan, Zheng86056092015-05-01 16:57:16 +0800224 bool need_flush;
Sage Weilde576062009-10-06 11:31:07 -0700225};
226
227static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
228{
Elena Reshetova805692d2017-03-03 11:15:07 +0200229 if (refcount_dec_and_test(&capsnap->nref)) {
Sage Weil4a625be2010-08-22 15:03:56 -0700230 if (capsnap->xattr_blob)
231 ceph_buffer_put(capsnap->xattr_blob);
Sage Weilde576062009-10-06 11:31:07 -0700232 kfree(capsnap);
Sage Weil4a625be2010-08-22 15:03:56 -0700233 }
Sage Weilde576062009-10-06 11:31:07 -0700234}
235
236/*
237 * The frag tree describes how a directory is fragmented, potentially across
238 * multiple metadata servers. It is also used to indicate points where
239 * metadata authority is delegated, and whether/where metadata is replicated.
240 *
241 * A _leaf_ frag will be present in the i_fragtree IFF there is
242 * delegation info. That is, if mds >= 0 || ndist > 0.
243 */
244#define CEPH_MAX_DIRFRAG_REP 4
245
246struct ceph_inode_frag {
247 struct rb_node node;
248
249 /* fragtree state */
250 u32 frag;
251 int split_by; /* i.e. 2^(split_by) children */
252
253 /* delegation and replication info */
254 int mds; /* -1 if same authority as parent */
255 int ndist; /* >0 if replicated */
256 int dist[CEPH_MAX_DIRFRAG_REP];
257};
258
259/*
260 * We cache inode xattrs as an encoded blob until they are first used,
261 * at which point we parse them into an rbtree.
262 */
263struct ceph_inode_xattr {
264 struct rb_node node;
265
266 const char *name;
267 int name_len;
268 const char *val;
269 int val_len;
270 int dirty;
271
272 int should_free_name;
273 int should_free_val;
274};
275
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700276/*
277 * Ceph dentry state
278 */
279struct ceph_dentry_info {
Yan, Zheng37c4efc2019-01-31 16:55:51 +0800280 struct dentry *dentry;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700281 struct ceph_mds_session *lease_session;
Yan, Zheng37c4efc2019-01-31 16:55:51 +0800282 struct list_head lease_list;
283 unsigned flags;
Yan, Zheng97aeb6b2017-11-27 10:47:46 +0800284 int lease_shared_gen;
285 u32 lease_gen;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700286 u32 lease_seq;
287 unsigned long lease_renew_after, lease_renew_from;
Miklos Szeredi9b16f03c2016-06-22 16:35:04 +0200288 unsigned long time;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700289 u64 offset;
290};
291
Yan, Zheng37c4efc2019-01-31 16:55:51 +0800292#define CEPH_DENTRY_REFERENCED 1
293#define CEPH_DENTRY_LEASE_LIST 2
294#define CEPH_DENTRY_SHRINK_LIST 4
Jeff Laytonf5e17ae2020-02-18 14:12:32 -0500295#define CEPH_DENTRY_PRIMARY_LINK 8
Yan, Zheng37c4efc2019-01-31 16:55:51 +0800296
Sage Weilde576062009-10-06 11:31:07 -0700297struct ceph_inode_xattrs_info {
298 /*
299 * (still encoded) xattr blob. we avoid the overhead of parsing
300 * this until someone actually calls getxattr, etc.
301 *
302 * blob->vec.iov_len == 4 implies there are no xattrs; blob ==
303 * NULL means we don't know.
304 */
305 struct ceph_buffer *blob, *prealloc_blob;
306
307 struct rb_root index;
308 bool dirty;
309 int count;
310 int names_size;
311 int vals_size;
312 u64 version, index_version;
313};
314
315/*
316 * Ceph inode.
317 */
Sage Weilde576062009-10-06 11:31:07 -0700318struct ceph_inode_info {
319 struct ceph_vino i_vino; /* ceph ino + snap */
320
Sage Weilbe655592011-11-30 09:47:09 -0800321 spinlock_t i_ceph_lock;
322
Sage Weilde576062009-10-06 11:31:07 -0700323 u64 i_version;
Yan, Zheng31c542a2014-11-14 21:41:55 +0800324 u64 i_inline_version;
Sage Weilde576062009-10-06 11:31:07 -0700325 u32 i_time_warp_seq;
326
Jeff Layton891f3f52020-01-14 15:06:40 -0500327 unsigned long i_ceph_flags;
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800328 atomic64_t i_release_count;
329 atomic64_t i_ordered_count;
330 atomic64_t i_complete_seq[2];
Sage Weilde576062009-10-06 11:31:07 -0700331
Sage Weil6c0f3af2010-11-16 11:14:34 -0800332 struct ceph_dir_layout i_dir_layout;
Sage Weilde576062009-10-06 11:31:07 -0700333 struct ceph_file_layout i_layout;
Jeff Layton785892f2020-01-02 07:11:38 -0500334 struct ceph_file_layout i_cached_layout; // for async creates
Sage Weilde576062009-10-06 11:31:07 -0700335 char *i_symlink;
336
337 /* for dirs */
Arnd Bergmann9bbeab42018-07-13 22:18:36 +0200338 struct timespec64 i_rctime;
Yanhu Caoe7f72952020-08-28 09:28:44 +0800339 u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps;
Sage Weilde576062009-10-06 11:31:07 -0700340 u64 i_files, i_subdirs;
Sage Weilde576062009-10-06 11:31:07 -0700341
Luis Henriquesfb18a572018-01-05 10:47:18 +0000342 /* quotas */
343 u64 i_max_bytes, i_max_files;
344
Yan, Zheng08796872019-01-09 11:07:02 +0800345 s32 i_dir_pin;
346
Sage Weilde576062009-10-06 11:31:07 -0700347 struct rb_root i_fragtree;
Yan, Zheng1b1bc162016-05-04 11:40:30 +0800348 int i_fragtree_nsplits;
Sage Weilde576062009-10-06 11:31:07 -0700349 struct mutex i_fragtree_mutex;
350
351 struct ceph_inode_xattrs_info i_xattrs;
352
Sage Weilbe655592011-11-30 09:47:09 -0800353 /* capabilities. protected _both_ by i_ceph_lock and cap->session's
Sage Weilde576062009-10-06 11:31:07 -0700354 * s_mutex. */
355 struct rb_root i_caps; /* cap list */
356 struct ceph_cap *i_auth_cap; /* authoritative cap, if any */
357 unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */
Jeff Layton1cf03a62020-04-01 17:07:52 -0400358
359 /*
Randy Dunlapf1f565a2020-07-17 16:36:04 -0700360 * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty
Jeff Layton1cf03a62020-04-01 17:07:52 -0400361 * is protected by the mdsc->cap_dirty_lock, but each individual item
362 * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty
363 * requires the mdsc->cap_dirty_lock. List presence for an item can
364 * be tested under the i_ceph_lock. Changing anything requires both.
365 */
366 struct list_head i_dirty_item;
367
368 /*
Jeff Layton829ad4d2020-04-03 13:09:07 -0400369 * Link to session's s_cap_flushing list. Protected in a similar
370 * fashion to i_dirty_item, but also by the s_mutex for changes. The
371 * s_cap_flushing list can be walked while holding either the s_mutex
372 * or msdc->cap_dirty_lock. List presence can also be checked while
373 * holding the i_ceph_lock for this inode.
Jeff Layton1cf03a62020-04-01 17:07:52 -0400374 */
375 struct list_head i_flushing_item;
376
Sage Weilde576062009-10-06 11:31:07 -0700377 /* we need to track cap writeback on a per-cap-bit basis, to allow
378 * overlapping, pipelined cap flushes to the mds. we can probably
379 * reduce the tid to 8 bits if we're concerned about inode size. */
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800380 struct ceph_cap_flush *i_prealloc_cap_flush;
Yan, Zhenge4500b52016-07-06 11:12:56 +0800381 struct list_head i_cap_flush_list;
Sage Weilde576062009-10-06 11:31:07 -0700382 wait_queue_head_t i_cap_wq; /* threads waiting on a capability */
Sage Weilde576062009-10-06 11:31:07 -0700383 unsigned long i_hold_caps_max; /* jiffies */
384 struct list_head i_cap_delay_list; /* for delayed cap release to mds */
Sage Weilde576062009-10-06 11:31:07 -0700385 struct ceph_cap_reservation i_cap_migration_resv;
386 struct list_head i_cap_snaps; /* snapped state pending flush to mds */
Sage Weil7d8cb262010-08-24 08:44:16 -0700387 struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or
388 dirty|flushing caps */
Sage Weilde576062009-10-06 11:31:07 -0700389 unsigned i_snap_caps; /* cap bits for snapped files */
390
Yan, Zheng719a2512020-03-05 20:21:00 +0800391 unsigned long i_last_rd;
392 unsigned long i_last_wr;
Yan, Zheng774a6a12016-06-06 16:01:39 +0800393 int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */
Sage Weilde576062009-10-06 11:31:07 -0700394
Yan, Zhengb0d7c222013-08-12 21:42:15 -0700395 struct mutex i_truncate_mutex;
Sage Weilde576062009-10-06 11:31:07 -0700396 u32 i_truncate_seq; /* last truncate to smaller size */
397 u64 i_truncate_size; /* and the size we last truncated down to */
398 int i_truncate_pending; /* still need to call vmtruncate */
399
400 u64 i_max_size; /* max file size authorized by mds */
401 u64 i_reported_size; /* (max_)size reported to or requested of mds */
402 u64 i_wanted_max_size; /* offset we'd like to write too */
403 u64 i_requested_max_size; /* max_size we've requested */
404
405 /* held references to caps */
406 int i_pin_ref;
Jeff Laytonf85122a2019-04-02 08:04:30 -0400407 int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref;
Sage Weilde576062009-10-06 11:31:07 -0700408 int i_wrbuffer_ref, i_wrbuffer_ref_head;
Yan, Zheng89aa5932017-09-08 15:23:18 +0800409 atomic_t i_filelock_ref;
Yan, Zheng97aeb6b2017-11-27 10:47:46 +0800410 atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */
Sage Weilcd045cb2010-11-04 11:05:05 -0700411 u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */
Sage Weilde576062009-10-06 11:31:07 -0700412 u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
413
Sage Weilde576062009-10-06 11:31:07 -0700414 struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
Yan, Zheng68cd5b42015-10-27 18:36:06 +0800415 struct list_head i_unsafe_iops; /* uncommitted mds inode ops */
Sage Weilde576062009-10-06 11:31:07 -0700416 spinlock_t i_unsafe_lock;
417
Yan, Zheng75c96272017-12-14 15:11:09 +0800418 union {
419 struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
420 struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */
421 };
Sage Weilde576062009-10-06 11:31:07 -0700422 struct list_head i_snap_realm_item;
423 struct list_head i_snap_flush_item;
Jeff Layton245ce992019-05-29 11:19:42 -0400424 struct timespec64 i_btime;
David Disseldorp193e7b32019-04-18 14:15:46 +0200425 struct timespec64 i_snap_btime;
Sage Weilde576062009-10-06 11:31:07 -0700426
Yan, Zheng1cf89a82019-05-18 11:18:44 +0800427 struct work_struct i_work;
428 unsigned long i_work_mask;
Sage Weilde576062009-10-06 11:31:07 -0700429
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400430#ifdef CONFIG_CEPH_FSCACHE
431 struct fscache_cookie *fscache;
Milosz Tanski99ccbd22013-08-21 17:29:54 -0400432#endif
Sage Weilde576062009-10-06 11:31:07 -0700433 struct inode vfs_inode; /* at end */
434};
435
Jeff Layton721d5c12019-08-01 14:11:15 -0400436static inline struct ceph_inode_info *
437ceph_inode(const struct inode *inode)
Sage Weilde576062009-10-06 11:31:07 -0700438{
Noah Watkinsfbbccec2009-10-28 11:54:49 -0700439 return container_of(inode, struct ceph_inode_info, vfs_inode);
Sage Weilde576062009-10-06 11:31:07 -0700440}
441
Jeff Layton721d5c12019-08-01 14:11:15 -0400442static inline struct ceph_fs_client *
443ceph_inode_to_client(const struct inode *inode)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800444{
445 return (struct ceph_fs_client *)inode->i_sb->s_fs_info;
446}
447
Jeff Layton721d5c12019-08-01 14:11:15 -0400448static inline struct ceph_fs_client *
449ceph_sb_to_client(const struct super_block *sb)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800450{
451 return (struct ceph_fs_client *)sb->s_fs_info;
452}
453
Xiubo Li2678da82020-09-03 09:01:39 -0400454static inline struct ceph_mds_client *
455ceph_sb_to_mdsc(const struct super_block *sb)
456{
457 return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc;
458}
459
Jeff Layton721d5c12019-08-01 14:11:15 -0400460static inline struct ceph_vino
461ceph_vino(const struct inode *inode)
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700462{
463 return ceph_inode(inode)->i_vino;
464}
465
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400466static inline u32 ceph_ino_to_ino32(u64 vino)
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800467{
Amon Ott3310f752011-10-20 13:04:07 -0700468 u32 ino = vino & 0xffffffff;
469 ino ^= vino >> 32;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800470 if (!ino)
Amon Otta661fc52012-01-23 09:25:23 -0800471 ino = 2;
Yehuda Sadehad1fee92011-01-21 16:44:03 -0800472 return ino;
473}
474
475/*
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400476 * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on
477 * some arches. We generally do not use this value inside the ceph driver, but
478 * we do want to set it to something, so that generic vfs code has an
479 * appropriate value for tracepoints and the like.
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700480 */
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400481static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino)
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700482{
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400483 if (sizeof(ino_t) == sizeof(u32))
484 return ceph_ino_to_ino32(vino.ino);
Amon Ott3310f752011-10-20 13:04:07 -0700485 return (ino_t)vino.ino;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700486}
487
488/* for printf-style formatting */
489#define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap
490
491static inline u64 ceph_ino(struct inode *inode)
492{
493 return ceph_inode(inode)->i_vino.ino;
494}
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400495
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700496static inline u64 ceph_snap(struct inode *inode)
497{
498 return ceph_inode(inode)->i_vino.snap;
499}
500
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400501/**
502 * ceph_present_ino - format an inode number for presentation to userland
503 * @sb: superblock where the inode lives
504 * @ino: inode number to (possibly) convert
505 *
506 * If the user mounted with the ino32 option, then the 64-bit value needs
507 * to be converted to something that can fit inside 32 bits. Note that
508 * internal kernel code never uses this value, so this is entirely for
509 * userland consumption.
510 */
511static inline u64 ceph_present_ino(struct super_block *sb, u64 ino)
512{
513 if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)))
514 return ceph_ino_to_ino32(ino);
515 return ino;
516}
517
518static inline u64 ceph_present_inode(struct inode *inode)
519{
520 return ceph_present_ino(inode->i_sb, ceph_ino(inode));
521}
522
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700523static inline int ceph_ino_compare(struct inode *inode, void *data)
524{
525 struct ceph_vino *pvino = (struct ceph_vino *)data;
526 struct ceph_inode_info *ci = ceph_inode(inode);
527 return ci->i_vino.ino == pvino->ino &&
528 ci->i_vino.snap == pvino->snap;
529}
530
Jeff Laytond4f6b312021-04-01 13:55:11 -0400531/*
532 * The MDS reserves a set of inodes for its own usage. These should never
533 * be accessible by clients, and so the MDS has no reason to ever hand these
534 * out. The range is CEPH_MDS_INO_MDSDIR_OFFSET..CEPH_INO_SYSTEM_BASE.
535 *
536 * These come from src/mds/mdstypes.h in the ceph sources.
537 */
538#define CEPH_MAX_MDS 0x100
539#define CEPH_NUM_STRAY 10
540#define CEPH_MDS_INO_MDSDIR_OFFSET (1 * CEPH_MAX_MDS)
541#define CEPH_INO_SYSTEM_BASE ((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY))
542
543static inline bool ceph_vino_is_reserved(const struct ceph_vino vino)
544{
545 if (vino.ino < CEPH_INO_SYSTEM_BASE &&
546 vino.ino >= CEPH_MDS_INO_MDSDIR_OFFSET) {
547 WARN_RATELIMIT(1, "Attempt to access reserved inode number 0x%llx", vino.ino);
548 return true;
549 }
550 return false;
551}
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400552
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700553static inline struct inode *ceph_find_inode(struct super_block *sb,
554 struct ceph_vino vino)
555{
Jeff Laytond4f6b312021-04-01 13:55:11 -0400556 if (ceph_vino_is_reserved(vino))
557 return NULL;
558
Jeff Laytonebce3eb2020-08-18 08:03:48 -0400559 /*
560 * NB: The hashval will be run through the fs/inode.c hash function
561 * anyway, so there is no need to squash the inode number down to
562 * 32-bits first. Just use low-order bits on arches with 32-bit long.
563 */
564 return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700565}
566
567
568/*
569 * Ceph inode.
570 */
Yan, Zheng10183a62015-04-27 15:33:28 +0800571#define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */
Yan, Zheng10183a62015-04-27 15:33:28 +0800572#define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */
Jeff Laytondaca8bd2019-07-05 10:55:38 -0400573#define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */
574#define CEPH_I_POOL_RD (1 << 4) /* can read from pool */
575#define CEPH_I_POOL_WR (1 << 5) /* can write to pool */
576#define CEPH_I_SEC_INITED (1 << 6) /* security initialized */
Yan, Zhengc0e385b2020-03-05 20:20:59 +0800577#define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */
578#define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */
579#define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */
580#define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */
581#define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */
582#define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */
Jeff Layton891f3f52020-01-14 15:06:40 -0500583#define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT)
Jeff Layton5d6451b2021-08-31 13:39:13 -0400584#define CEPH_I_SHUTDOWN (1 << 13) /* inode is no longer usable */
Jeff Layton26544c622017-04-04 08:39:46 -0400585
586/*
Yan, Zheng1cf89a82019-05-18 11:18:44 +0800587 * Masks of ceph inode work.
588 */
Jeff Laytona8810cd2020-12-10 14:39:26 -0500589#define CEPH_I_WORK_WRITEBACK 0
590#define CEPH_I_WORK_INVALIDATE_PAGES 1
591#define CEPH_I_WORK_VMTRUNCATE 2
592#define CEPH_I_WORK_CHECK_CAPS 3
593#define CEPH_I_WORK_FLUSH_SNAPS 4
Yan, Zheng1cf89a82019-05-18 11:18:44 +0800594
595/*
Jeff Layton26544c622017-04-04 08:39:46 -0400596 * We set the ERROR_WRITE bit when we start seeing write errors on an inode
597 * and then clear it when they start succeeding. Note that we do a lockless
598 * check first, and only take the lock if it looks like it needs to be changed.
599 * The write submission code just takes this as a hint, so we're not too
600 * worried if a few slip through in either direction.
601 */
602static inline void ceph_set_error_write(struct ceph_inode_info *ci)
603{
604 if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
605 spin_lock(&ci->i_ceph_lock);
606 ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
607 spin_unlock(&ci->i_ceph_lock);
608 }
609}
610
611static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
612{
613 if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
614 spin_lock(&ci->i_ceph_lock);
615 ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
616 spin_unlock(&ci->i_ceph_lock);
617 }
618}
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700619
Yan, Zheng2f276c52013-03-13 19:44:32 +0800620static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800621 long long release_count,
622 long long ordered_count)
Sage Weilde576062009-10-06 11:31:07 -0700623{
Andrea Parri74960772019-05-20 19:23:58 +0200624 /*
625 * Makes sure operations that setup readdir cache (update page
626 * cache and i_size) are strongly ordered w.r.t. the following
627 * atomic64_set() operations.
628 */
629 smp_mb();
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800630 atomic64_set(&ci->i_complete_seq[0], release_count);
631 atomic64_set(&ci->i_complete_seq[1], ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700632}
633
Yan, Zheng2f276c52013-03-13 19:44:32 +0800634static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700635{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800636 atomic64_inc(&ci->i_release_count);
637}
638
639static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
640{
641 atomic64_inc(&ci->i_ordered_count);
Sage Weilde576062009-10-06 11:31:07 -0700642}
643
Yan, Zheng2f276c52013-03-13 19:44:32 +0800644static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
Sage Weilde576062009-10-06 11:31:07 -0700645{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800646 return atomic64_read(&ci->i_complete_seq[0]) ==
647 atomic64_read(&ci->i_release_count);
Yan, Zheng2f276c52013-03-13 19:44:32 +0800648}
Sage Weilde576062009-10-06 11:31:07 -0700649
Yan, Zheng70db4f32014-10-21 18:09:56 -0700650static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
651{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800652 return atomic64_read(&ci->i_complete_seq[0]) ==
653 atomic64_read(&ci->i_release_count) &&
654 atomic64_read(&ci->i_complete_seq[1]) ==
655 atomic64_read(&ci->i_ordered_count);
Yan, Zheng70db4f32014-10-21 18:09:56 -0700656}
657
Yan, Zheng2f276c52013-03-13 19:44:32 +0800658static inline void ceph_dir_clear_complete(struct inode *inode)
659{
660 __ceph_dir_clear_complete(ceph_inode(inode));
661}
662
Yan, Zheng70db4f32014-10-21 18:09:56 -0700663static inline void ceph_dir_clear_ordered(struct inode *inode)
Yan, Zheng2f276c52013-03-13 19:44:32 +0800664{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800665 __ceph_dir_clear_ordered(ceph_inode(inode));
Sage Weilde576062009-10-06 11:31:07 -0700666}
667
Yan, Zheng70db4f32014-10-21 18:09:56 -0700668static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
669{
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800670 bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
671 smp_rmb();
Yan, Zheng70db4f32014-10-21 18:09:56 -0700672 return ret;
673}
Sage Weilde576062009-10-06 11:31:07 -0700674
675/* find a specific frag @f */
676extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
677 u32 f);
678
679/*
680 * choose fragment for value @v. copy frag content to pfrag, if leaf
681 * exists
682 */
683extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
684 struct ceph_inode_frag *pfrag,
685 int *found);
686
Yan, Zheng1e9c2eb2019-01-28 20:43:55 +0800687static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry)
Sage Weilde576062009-10-06 11:31:07 -0700688{
689 return (struct ceph_dentry_info *)dentry->d_fsdata;
690}
691
Sage Weilde576062009-10-06 11:31:07 -0700692/*
693 * caps helpers
694 */
695static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
696{
697 return !RB_EMPTY_ROOT(&ci->i_caps);
698}
699
700extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
701extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
Xiubo Li1af16d52020-03-19 23:45:00 -0400702extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask,
703 int t);
Sage Weilde576062009-10-06 11:31:07 -0700704extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
705 struct ceph_cap *cap);
706
707static inline int ceph_caps_issued(struct ceph_inode_info *ci)
708{
709 int issued;
Sage Weilbe655592011-11-30 09:47:09 -0800710 spin_lock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700711 issued = __ceph_caps_issued(ci, NULL);
Sage Weilbe655592011-11-30 09:47:09 -0800712 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700713 return issued;
714}
715
Xiubo Li1af16d52020-03-19 23:45:00 -0400716static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci,
717 int mask, int touch)
Sage Weilde576062009-10-06 11:31:07 -0700718{
719 int r;
Sage Weilbe655592011-11-30 09:47:09 -0800720 spin_lock(&ci->i_ceph_lock);
Xiubo Li1af16d52020-03-19 23:45:00 -0400721 r = __ceph_caps_issued_mask_metric(ci, mask, touch);
Sage Weilbe655592011-11-30 09:47:09 -0800722 spin_unlock(&ci->i_ceph_lock);
Sage Weilde576062009-10-06 11:31:07 -0700723 return r;
724}
725
726static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
727{
728 return ci->i_dirty_caps | ci->i_flushing_caps;
729}
Yan, Zhengf66fd9f2015-06-10 17:26:13 +0800730extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
731extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
732extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
733 struct ceph_cap_flush **pcf);
Sage Weilde576062009-10-06 11:31:07 -0700734
Yan, Zheng9563f882013-11-22 13:50:45 +0800735extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
736 struct ceph_cap *ocap, int mask);
Sage Weilde576062009-10-06 11:31:07 -0700737extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
738extern int __ceph_caps_used(struct ceph_inode_info *ci);
739
Yan, Zheng719a2512020-03-05 20:21:00 +0800740static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci)
741{
742 return ci->i_nr_by_mode[0];
743}
Sage Weilde576062009-10-06 11:31:07 -0700744extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
Yan, Zheng525d15e2019-05-11 17:27:59 +0800745extern int __ceph_caps_wanted(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -0700746
747/* what the mds thinks we want */
Yan, Zhengc1944fe2017-01-29 22:15:47 +0800748extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
Sage Weilde576062009-10-06 11:31:07 -0700749
Yehuda Sadeh37151662010-06-17 16:16:12 -0700750extern void ceph_caps_init(struct ceph_mds_client *mdsc);
751extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
Yan, Zhengfe330322019-02-01 14:57:15 +0800752extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc,
753 struct ceph_mount_options *fsopt);
Zhi Zhange30ee582018-01-24 21:24:33 +0800754extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
Yehuda Sadeh37151662010-06-17 16:16:12 -0700755 struct ceph_cap_reservation *ctx, int need);
Chengguang Xu7bf8f732018-07-28 23:15:35 +0800756extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc,
Yehuda Sadeh37151662010-06-17 16:16:12 -0700757 struct ceph_cap_reservation *ctx);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700758extern void ceph_reservation_status(struct ceph_fs_client *client,
Sage Weilde576062009-10-06 11:31:07 -0700759 int *total, int *avail, int *used,
Sage Weil85ccce42010-02-17 10:02:43 -0800760 int *reserved, int *min);
Sage Weilde576062009-10-06 11:31:07 -0700761
Sage Weilde576062009-10-06 11:31:07 -0700762
Sage Weilde576062009-10-06 11:31:07 -0700763
764/*
765 * we keep buffered readdir results attached to file->private_data
766 */
Sage Weil4918b6d2011-07-26 11:26:07 -0700767#define CEPH_F_SYNC 1
Sage Weil9cfa1092011-07-26 11:26:18 -0700768#define CEPH_F_ATEND 2
Sage Weil4918b6d2011-07-26 11:26:07 -0700769
Sage Weilde576062009-10-06 11:31:07 -0700770struct ceph_file_info {
Sage Weil252c6722011-07-26 11:25:27 -0700771 short fmode; /* initialized on open */
772 short flags; /* CEPH_F_* */
Sage Weilde576062009-10-06 11:31:07 -0700773
Yan, Zheng5d988302017-12-15 11:15:36 +0800774 spinlock_t rw_contexts_lock;
775 struct list_head rw_contexts;
Yan, Zhengf4b97862019-07-25 20:16:42 +0800776
Yan, Zheng81f148a2019-07-25 20:16:46 +0800777 u32 filp_gen;
Yan, Zhengff5d9132019-07-25 20:16:45 +0800778 atomic_t num_locks;
Chengguang Xubb48bd42018-03-13 10:42:44 +0800779};
780
781struct ceph_dir_file_info {
782 struct ceph_file_info file_info;
Yan, Zheng5d988302017-12-15 11:15:36 +0800783
Sage Weilde576062009-10-06 11:31:07 -0700784 /* readdir: position within the dir */
785 u32 frag;
786 struct ceph_mds_request *last_readdir;
Sage Weilde576062009-10-06 11:31:07 -0700787
788 /* readdir: position within a frag */
Yan, Zhengf0494202014-02-27 16:26:24 +0800789 unsigned next_offset; /* offset of next chunk (last_name's + 1) */
Sage Weilde576062009-10-06 11:31:07 -0700790 char *last_name; /* last entry in previous chunk */
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800791 long long dir_release_count;
792 long long dir_ordered_count;
793 int readdir_cache_idx;
Sage Weilde576062009-10-06 11:31:07 -0700794
795 /* used for -o dirstat read() on directory thing */
796 char *dir_info;
797 int dir_info_len;
798};
799
Yan, Zheng5d988302017-12-15 11:15:36 +0800800struct ceph_rw_context {
801 struct list_head list;
802 struct task_struct *thread;
803 int caps;
804};
805
806#define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \
807 struct ceph_rw_context _name = { \
808 .thread = current, \
809 .caps = _caps, \
810 }
811
812static inline void ceph_add_rw_context(struct ceph_file_info *cf,
813 struct ceph_rw_context *ctx)
814{
815 spin_lock(&cf->rw_contexts_lock);
816 list_add(&ctx->list, &cf->rw_contexts);
817 spin_unlock(&cf->rw_contexts_lock);
818}
819
820static inline void ceph_del_rw_context(struct ceph_file_info *cf,
821 struct ceph_rw_context *ctx)
822{
823 spin_lock(&cf->rw_contexts_lock);
824 list_del(&ctx->list);
825 spin_unlock(&cf->rw_contexts_lock);
826}
827
828static inline struct ceph_rw_context*
829ceph_find_rw_context(struct ceph_file_info *cf)
830{
831 struct ceph_rw_context *ctx, *found = NULL;
832 spin_lock(&cf->rw_contexts_lock);
833 list_for_each_entry(ctx, &cf->rw_contexts, list) {
834 if (ctx->thread == current) {
835 found = ctx;
836 break;
837 }
838 }
839 spin_unlock(&cf->rw_contexts_lock);
840 return found;
841}
842
Yan, Zhengfdd4e152015-06-16 20:48:56 +0800843struct ceph_readdir_cache_control {
844 struct page *page;
845 struct dentry **dentries;
846 int index;
847};
Sage Weilde576062009-10-06 11:31:07 -0700848
849/*
Sage Weilde576062009-10-06 11:31:07 -0700850 * A "snap realm" describes a subset of the file hierarchy sharing
851 * the same set of snapshots that apply to it. The realms themselves
852 * are organized into a hierarchy, such that children inherit (some of)
853 * the snapshots of their parents.
854 *
855 * All inodes within the realm that have capabilities are linked into a
856 * per-realm list.
857 */
858struct ceph_snap_realm {
859 u64 ino;
Luis Henriquese3161f12018-01-12 17:19:28 +0000860 struct inode *inode;
Sage Weilde576062009-10-06 11:31:07 -0700861 atomic_t nref;
Sage Weila105f002010-02-15 14:37:55 -0800862 struct rb_node node;
863
Sage Weilde576062009-10-06 11:31:07 -0700864 u64 created, seq;
865 u64 parent_ino;
866 u64 parent_since; /* snapid when our current parent became so */
867
868 u64 *prior_parent_snaps; /* snaps inherited from any parents we */
Alex Elderaa711ee32012-07-13 20:35:11 -0500869 u32 num_prior_parent_snaps; /* had prior to parent_since */
Sage Weilde576062009-10-06 11:31:07 -0700870 u64 *snaps; /* snaps specific to this realm */
Alex Elderaa711ee32012-07-13 20:35:11 -0500871 u32 num_snaps;
Sage Weilde576062009-10-06 11:31:07 -0700872
873 struct ceph_snap_realm *parent;
874 struct list_head children; /* list of child realms */
875 struct list_head child_item;
876
877 struct list_head empty_item; /* if i have ref==0 */
878
Sage Weilae00d4f2010-09-16 16:26:51 -0700879 struct list_head dirty_item; /* if realm needs new context */
880
Sage Weilde576062009-10-06 11:31:07 -0700881 /* the current set of snaps for this realm */
882 struct ceph_snap_context *cached_context;
883
884 struct list_head inodes_with_caps;
885 spinlock_t inodes_with_caps_lock;
886};
887
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700888static inline int default_congestion_kb(void)
Sage Weilde576062009-10-06 11:31:07 -0700889{
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700890 int congestion_kb;
891
892 /*
893 * Copied from NFS
894 *
895 * congestion size, scale with available memory.
896 *
897 * 64MB: 8192k
898 * 128MB: 11585k
899 * 256MB: 16384k
900 * 512MB: 23170k
901 * 1GB: 32768k
902 * 2GB: 46340k
903 * 4GB: 65536k
904 * 8GB: 92681k
905 * 16GB: 131072k
906 *
907 * This allows larger machines to have larger/more transfers.
908 * Limit the default to 256M
909 */
Arun KSca79b0c2018-12-28 00:34:29 -0800910 congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700911 if (congestion_kb > 256*1024)
912 congestion_kb = 256*1024;
913
914 return congestion_kb;
Sage Weilde576062009-10-06 11:31:07 -0700915}
916
917
Yan, Zhengd468e722019-07-25 20:16:44 +0800918/* super.c */
919extern int ceph_force_reconnect(struct super_block *sb);
Sage Weilde576062009-10-06 11:31:07 -0700920/* snap.c */
921struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
922 u64 ino);
923extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
924 struct ceph_snap_realm *realm);
925extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
926 struct ceph_snap_realm *realm);
927extern int ceph_update_snap_trace(struct ceph_mds_client *m,
Yan, Zheng982d6012014-12-23 15:30:54 +0800928 void *p, void *e, bool deletion,
929 struct ceph_snap_realm **realm_ret);
Jeff Layton0ba92e12021-08-02 11:01:26 -0400930void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm);
Sage Weilde576062009-10-06 11:31:07 -0700931extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
Sage Weil2600d2d2010-02-22 15:12:16 -0800932 struct ceph_mds_session *session,
Sage Weilde576062009-10-06 11:31:07 -0700933 struct ceph_msg *msg);
Sage Weilde576062009-10-06 11:31:07 -0700934extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
935 struct ceph_cap_snap *capsnap);
936extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
937
Yan, Zheng75c96272017-12-14 15:11:09 +0800938extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc,
939 u64 snap);
940extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc,
941 struct ceph_snapid_map *sm);
942extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc);
943extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc);
Jeff Layton631ed4b2021-10-14 11:10:47 -0400944void ceph_umount_begin(struct super_block *sb);
Yan, Zheng75c96272017-12-14 15:11:09 +0800945
946
Sage Weilde576062009-10-06 11:31:07 -0700947/*
948 * a cap_snap is "pending" if it is still awaiting an in-progress
949 * sync write (that may/may not still update size, mtime, etc.).
950 */
951static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
952{
953 return !list_empty(&ci->i_cap_snaps) &&
Yan, Zheng86056092015-05-01 16:57:16 +0800954 list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
955 ci_item)->writing;
Sage Weilde576062009-10-06 11:31:07 -0700956}
957
Sage Weilde576062009-10-06 11:31:07 -0700958/* inode.c */
Jeff Layton966c7162019-12-05 09:09:25 -0500959struct ceph_mds_reply_info_in;
960struct ceph_mds_reply_dirfrag;
961
Sage Weilde576062009-10-06 11:31:07 -0700962extern const struct inode_operations ceph_file_iops;
963
964extern struct inode *ceph_alloc_inode(struct super_block *sb);
Yan, Zheng87bc5b82019-06-02 09:45:38 +0800965extern void ceph_evict_inode(struct inode *inode);
Al Virocfa6d412019-04-10 15:18:50 -0400966extern void ceph_free_inode(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -0700967
968extern struct inode *ceph_get_inode(struct super_block *sb,
969 struct ceph_vino vino);
970extern struct inode *ceph_get_snapdir(struct inode *parent);
971extern int ceph_fill_file_size(struct inode *inode, int issued,
972 u32 truncate_seq, u64 truncate_size, u64 size);
973extern void ceph_fill_file_time(struct inode *inode, int issued,
Arnd Bergmann9bbeab42018-07-13 22:18:36 +0200974 u64 time_warp_seq, struct timespec64 *ctime,
975 struct timespec64 *mtime,
976 struct timespec64 *atime);
Jeff Layton966c7162019-12-05 09:09:25 -0500977extern int ceph_fill_inode(struct inode *inode, struct page *locked_page,
978 struct ceph_mds_reply_info_in *iinfo,
979 struct ceph_mds_reply_dirfrag *dirinfo,
980 struct ceph_mds_session *session, int cap_fmode,
981 struct ceph_cap_reservation *caps_reservation);
Sage Weilde576062009-10-06 11:31:07 -0700982extern int ceph_fill_trace(struct super_block *sb,
Jeff Laytonf5a03b02017-01-31 11:06:13 -0500983 struct ceph_mds_request *req);
Sage Weilde576062009-10-06 11:31:07 -0700984extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
985 struct ceph_mds_session *session);
986
987extern int ceph_inode_holds_cap(struct inode *inode, int mask);
988
Yan, Zhengefb0ca72017-05-22 12:03:32 +0800989extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
Yan, Zhengb415bf42013-07-02 12:40:19 +0800990extern void __ceph_do_pending_vmtruncate(struct inode *inode);
Jeff Layton64f28c62020-10-09 14:24:34 -0400991
Jeff Layton64f28c62020-10-09 14:24:34 -0400992void ceph_queue_inode_work(struct inode *inode, int work_bit);
993
994static inline void ceph_queue_vmtruncate(struct inode *inode)
995{
996 ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE);
997}
998
999static inline void ceph_queue_invalidate(struct inode *inode)
1000{
1001 ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES);
1002}
1003
1004static inline void ceph_queue_writeback(struct inode *inode)
1005{
1006 ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK);
1007}
1008
Jeff Laytona8810cd2020-12-10 14:39:26 -05001009static inline void ceph_queue_check_caps(struct inode *inode)
1010{
1011 ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS);
1012}
1013
1014static inline void ceph_queue_flush_snaps(struct inode *inode)
1015{
1016 ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS);
1017}
1018
Yan, Zheng01deead2014-11-14 21:56:29 +08001019extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
1020 int mask, bool force);
1021static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
1022{
1023 return __ceph_do_getattr(inode, NULL, mask, force);
1024}
Christian Brauner549c7292021-01-21 14:19:43 +01001025extern int ceph_permission(struct user_namespace *mnt_userns,
1026 struct inode *inode, int mask);
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +02001027extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
Christian Brauner549c7292021-01-21 14:19:43 +01001028extern int ceph_setattr(struct user_namespace *mnt_userns,
1029 struct dentry *dentry, struct iattr *attr);
1030extern int ceph_getattr(struct user_namespace *mnt_userns,
1031 const struct path *path, struct kstat *stat,
David Howellsa528d352017-01-31 16:46:22 +00001032 u32 request_mask, unsigned int flags);
Jeff Layton5d6451b2021-08-31 13:39:13 -04001033void ceph_inode_shutdown(struct inode *inode);
1034
1035static inline bool ceph_inode_is_shutdown(struct inode *inode)
1036{
1037 unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags);
1038 struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1039 int state = READ_ONCE(fsc->mount_state);
1040
1041 return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN;
1042}
Sage Weilde576062009-10-06 11:31:07 -07001043
1044/* xattr.c */
Andreas Gruenbachera26fecc2016-04-14 00:30:16 +02001045int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001046ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
Sage Weilde576062009-10-06 11:31:07 -07001047extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
Luis Henriques12fe3dd2019-07-19 15:32:21 +01001048extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -07001049extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +08001050extern const struct xattr_handler *ceph_xattr_handlers[];
Sage Weilde576062009-10-06 11:31:07 -07001051
Yan, Zheng5c31e922019-05-26 15:35:39 +08001052struct ceph_acl_sec_ctx {
1053#ifdef CONFIG_CEPH_FS_POSIX_ACL
1054 void *default_acl;
1055 void *acl;
1056#endif
Yan, Zhengac6713c2019-05-26 16:27:56 +08001057#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
1058 void *sec_ctx;
1059 u32 sec_ctxlen;
1060#endif
Yan, Zheng5c31e922019-05-26 15:35:39 +08001061 struct ceph_pagelist *pagelist;
1062};
1063
Yan, Zheng315f2402016-03-07 10:34:50 +08001064#ifdef CONFIG_SECURITY
1065extern bool ceph_security_xattr_deadlock(struct inode *in);
1066extern bool ceph_security_xattr_wanted(struct inode *in);
1067#else
1068static inline bool ceph_security_xattr_deadlock(struct inode *in)
1069{
1070 return false;
1071}
1072static inline bool ceph_security_xattr_wanted(struct inode *in)
1073{
1074 return false;
1075}
1076#endif
1077
Yan, Zhengac6713c2019-05-26 16:27:56 +08001078#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
1079extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
1080 struct ceph_acl_sec_ctx *ctx);
Jeff Layton668959a2019-08-06 09:07:51 -04001081static inline void ceph_security_invalidate_secctx(struct inode *inode)
1082{
1083 security_inode_invalidate_secctx(inode);
1084}
Yan, Zhengac6713c2019-05-26 16:27:56 +08001085#else
1086static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
1087 struct ceph_acl_sec_ctx *ctx)
1088{
1089 return 0;
1090}
1091static inline void ceph_security_invalidate_secctx(struct inode *inode)
1092{
1093}
1094#endif
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001095
Yan, Zheng5c31e922019-05-26 15:35:39 +08001096void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx);
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001097
Yan, Zheng5c31e922019-05-26 15:35:39 +08001098/* acl.c */
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001099#ifdef CONFIG_CEPH_FS_POSIX_ACL
1100
Miklos Szeredi0cad6242021-08-18 22:08:24 +02001101struct posix_acl *ceph_get_acl(struct inode *, int, bool);
Christian Brauner549c7292021-01-21 14:19:43 +01001102int ceph_set_acl(struct user_namespace *mnt_userns,
1103 struct inode *inode, struct posix_acl *acl, int type);
Yan, Zhengb1ee94a2014-09-16 20:35:17 +08001104int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
Yan, Zheng5c31e922019-05-26 15:35:39 +08001105 struct ceph_acl_sec_ctx *as_ctx);
1106void ceph_init_inode_acls(struct inode *inode,
1107 struct ceph_acl_sec_ctx *as_ctx);
Guangliang Zhaoc969d9b2014-02-16 08:35:52 -08001108
1109static inline void ceph_forget_all_cached_acls(struct inode *inode)
1110{
1111 forget_all_cached_acls(inode);
1112}
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001113
1114#else
1115
1116#define ceph_get_acl NULL
Sage Weil72466d02014-01-29 06:22:25 -08001117#define ceph_set_acl NULL
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001118
Yan, Zhengb1ee94a2014-09-16 20:35:17 +08001119static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
Yan, Zheng5c31e922019-05-26 15:35:39 +08001120 struct ceph_acl_sec_ctx *as_ctx)
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001121{
1122 return 0;
1123}
Yan, Zhengb1ee94a2014-09-16 20:35:17 +08001124static inline void ceph_init_inode_acls(struct inode *inode,
Yan, Zheng5c31e922019-05-26 15:35:39 +08001125 struct ceph_acl_sec_ctx *as_ctx)
Yan, Zhengb1ee94a2014-09-16 20:35:17 +08001126{
1127}
Guangliang Zhao7221fe42013-11-11 15:18:03 +08001128static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
1129{
1130 return 0;
1131}
1132
1133static inline void ceph_forget_all_cached_acls(struct inode *inode)
1134{
1135}
1136
1137#endif
1138
Sage Weilde576062009-10-06 11:31:07 -07001139/* caps.c */
1140extern const char *ceph_cap_string(int c);
1141extern void ceph_handle_caps(struct ceph_mds_session *session,
1142 struct ceph_msg *msg);
Yan, Zhengd9df2782014-04-18 09:57:11 +08001143extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
1144 struct ceph_cap_reservation *ctx);
1145extern void ceph_add_cap(struct inode *inode,
1146 struct ceph_mds_session *session, u64 cap_id,
Yan, Zheng135e6712020-03-05 20:21:02 +08001147 unsigned issued, unsigned wanted,
Yan, Zhengd9df2782014-04-18 09:57:11 +08001148 unsigned cap, unsigned seq, u64 realmino, int flags,
1149 struct ceph_cap **new_cap);
Yan, Zhenga096b092013-09-22 10:15:58 +08001150extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
Xiubo Lia76d0a92021-08-25 21:45:45 +08001151extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
Yan, Zhengd6e47812019-05-23 11:01:37 +08001152extern void __ceph_remove_caps(struct ceph_inode_info *ci);
Yehuda Sadeh37151662010-06-17 16:16:12 -07001153extern void ceph_put_cap(struct ceph_mds_client *mdsc,
1154 struct ceph_cap *cap);
Yan, Zheng9215aee2013-11-30 12:47:41 +08001155extern int ceph_is_any_caps(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -07001156
Stephen Rothwellf1a3d572010-01-18 11:53:08 +11001157extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
Josef Bacik02c24a82011-07-16 20:44:56 -04001158extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
1159 int datasync);
Yan, Zhenge548e9b2015-06-10 15:17:56 +08001160extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
1161 struct ceph_mds_session *session);
Sage Weilde576062009-10-06 11:31:07 -07001162extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
1163 struct ceph_mds_session *session);
Jeff Laytone8a4d262020-02-25 11:49:53 -08001164void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session,
1165 struct ceph_inode_info *ci);
Greg Farnum2bc50252010-06-30 12:44:34 -07001166extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
1167 int mds);
Jeff Layton40dcf752020-01-14 09:23:49 -05001168extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps,
1169 bool snap_rwsem_locked);
Sage Weilde576062009-10-06 11:31:07 -07001170extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
1171extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
Jeff Laytona8810cd2020-12-10 14:39:26 -05001172extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had);
Xiubo Lie64f44a2020-05-27 09:09:27 -04001173extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci,
1174 int had);
Sage Weilde576062009-10-06 11:31:07 -07001175extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
1176 struct ceph_snap_context *snapc);
Xiubo Lia6d37cc2021-08-25 21:45:43 +08001177extern void __ceph_remove_capsnap(struct inode *inode,
1178 struct ceph_cap_snap *capsnap,
1179 bool *wake_ci, bool *wake_mdsc);
1180extern void ceph_remove_capsnap(struct inode *inode,
1181 struct ceph_cap_snap *capsnap,
1182 bool *wake_ci, bool *wake_mdsc);
Yan, Zhenged9b4302016-07-05 21:08:07 +08001183extern void ceph_flush_snaps(struct ceph_inode_info *ci,
1184 struct ceph_mds_session **psession);
Yan, Zhengefb0ca72017-05-22 12:03:32 +08001185extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
Sage Weilde576062009-10-06 11:31:07 -07001186extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
1187 struct ceph_mds_session *session);
Luis Henriquesbf2ba432021-07-06 14:52:41 +01001188extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
Sage Weilafcdaea2009-10-14 14:27:38 -07001189extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
Zhi Zhang6ef0bc62018-01-24 21:24:33 +08001190extern int ceph_drop_caps_for_unlink(struct inode *inode);
Sage Weilde576062009-10-06 11:31:07 -07001191extern int ceph_encode_inode_release(void **p, struct inode *inode,
1192 int mds, int drop, int unless, int force);
1193extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
Jeff Laytonca6c8ae2016-12-15 08:37:59 -05001194 struct inode *dir,
Sage Weilde576062009-10-06 11:31:07 -07001195 int mds, int drop, int unless);
1196
Yan, Zheng5e3ded12019-07-25 20:16:43 +08001197extern int ceph_get_caps(struct file *filp, int need, int want,
Jeff Laytone72968e2021-04-05 12:19:35 -04001198 loff_t endoff, int *got);
Yan, Zheng5e3ded12019-07-25 20:16:43 +08001199extern int ceph_try_get_caps(struct inode *inode,
Luis Henriques2ee9dd92018-10-15 16:45:57 +01001200 int need, int want, bool nonblock, int *got);
Sage Weilde576062009-10-06 11:31:07 -07001201
1202/* for counting open files by mode */
Yan, Zheng719a2512020-03-05 20:21:00 +08001203extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count);
1204extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count);
1205extern void __ceph_touch_fmode(struct ceph_inode_info *ci,
1206 struct ceph_mds_client *mdsc, int fmode);
Sage Weilde576062009-10-06 11:31:07 -07001207
1208/* addr.c */
1209extern const struct address_space_operations ceph_aops;
1210extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
Yan, Zheng10183a62015-04-27 15:33:28 +08001211extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
Yan, Zheng5e3ded12019-07-25 20:16:43 +08001212extern int ceph_pool_perm_check(struct inode *inode, int need);
Yan, Zheng10183a62015-04-27 15:33:28 +08001213extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
Jeff Layton36e6da92021-09-02 13:06:57 -04001214int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate);
Sage Weilde576062009-10-06 11:31:07 -07001215
1216/* file.c */
1217extern const struct file_operations ceph_file_fops;
Alex Elder9e0eb852013-02-06 13:11:38 -06001218
Yan, Zheng719a2512020-03-05 20:21:00 +08001219extern int ceph_renew_caps(struct inode *inode, int fmode);
Sage Weilde576062009-10-06 11:31:07 -07001220extern int ceph_open(struct inode *inode, struct file *file);
Sage Weil5ef50c32012-07-31 11:27:36 -07001221extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
Al Viro44907d72018-06-08 13:32:02 -04001222 struct file *file, unsigned flags, umode_t mode);
Sage Weilde576062009-10-06 11:31:07 -07001223extern int ceph_release(struct inode *inode, struct file *filp);
Yan, Zheng31c542a2014-11-14 21:41:55 +08001224extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
1225 char *data, size_t len);
Ilya Dryomov55f2a042017-02-13 14:44:19 +01001226
Sage Weilde576062009-10-06 11:31:07 -07001227/* dir.c */
1228extern const struct file_operations ceph_dir_fops;
Yan, Zheng38c48b52015-01-14 13:46:04 +08001229extern const struct file_operations ceph_snapdir_fops;
Sage Weilde576062009-10-06 11:31:07 -07001230extern const struct inode_operations ceph_dir_iops;
Yan, Zheng38c48b52015-01-14 13:46:04 +08001231extern const struct inode_operations ceph_snapdir_iops;
Al Viro18fc8ab2016-10-28 21:52:50 -04001232extern const struct dentry_operations ceph_dentry_ops;
Sage Weilde576062009-10-06 11:31:07 -07001233
Yan, Zhengf3c4ebe2016-04-29 11:27:30 +08001234extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
Sage Weilde576062009-10-06 11:31:07 -07001235extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
Jeff Laytonaa60cfc2021-03-01 08:01:54 -05001236extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req,
Jeff Layton7a971e22021-06-02 12:46:07 -04001237 struct dentry *dentry);
Sage Weilde576062009-10-06 11:31:07 -07001238extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
1239 struct dentry *dentry, int err);
1240
Yan, Zheng37c4efc2019-01-31 16:55:51 +08001241extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di);
1242extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di);
Sage Weil81a6cf22010-05-14 09:35:38 -07001243extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
Yan, Zheng37c4efc2019-01-31 16:55:51 +08001244extern int ceph_trim_dentries(struct ceph_mds_client *mdsc);
Sage Weile5f86dc2011-07-26 11:30:55 -07001245extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
Yan, Zhengfdd4e152015-06-16 20:48:56 +08001246extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
Sage Weilde576062009-10-06 11:31:07 -07001247
Sage Weilde576062009-10-06 11:31:07 -07001248/* ioctl.c */
1249extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1250
1251/* export.c */
1252extern const struct export_operations ceph_export_ops;
Luis Henriques38862742019-03-21 10:20:09 +00001253struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino);
Sage Weilde576062009-10-06 11:31:07 -07001254
Greg Farnum40819f62010-08-02 15:34:23 -07001255/* locks.c */
Yan, Zhengeb13e832014-03-09 23:16:40 +08001256extern __init void ceph_flock_init(void);
Greg Farnum40819f62010-08-02 15:34:23 -07001257extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
1258extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
1259extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
Jim Schutt39be95e2013-05-15 13:03:35 -05001260extern int ceph_encode_locks_to_buffer(struct inode *inode,
1261 struct ceph_filelock *flocks,
1262 int num_fcntl_locks,
1263 int num_flock_locks);
1264extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
1265 struct ceph_pagelist *pagelist,
1266 int num_fcntl_locks, int num_flock_locks);
Greg Farnum40819f62010-08-02 15:34:23 -07001267
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001268/* debugfs.c */
Greg Kroah-Hartman1a829ff2019-06-12 16:55:38 +02001269extern void ceph_fs_debugfs_init(struct ceph_fs_client *client);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001270extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
1271
Luis Henriquesfb18a572018-01-05 10:47:18 +00001272/* quota.c */
Luis Henriquesd557c482018-01-12 17:19:29 +00001273static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci)
1274{
1275 return ci->i_max_files || ci->i_max_bytes;
1276}
1277
1278extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
1279
1280static inline void __ceph_update_quota(struct ceph_inode_info *ci,
1281 u64 max_bytes, u64 max_files)
1282{
1283 bool had_quota, has_quota;
1284 had_quota = __ceph_has_any_quota(ci);
1285 ci->i_max_bytes = max_bytes;
1286 ci->i_max_files = max_files;
1287 has_quota = __ceph_has_any_quota(ci);
1288
1289 if (had_quota != has_quota)
1290 ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota);
1291}
1292
Luis Henriquesfb18a572018-01-05 10:47:18 +00001293extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
1294 struct ceph_mds_session *session,
1295 struct ceph_msg *msg);
Luis Henriquesb7a29212018-01-05 10:47:19 +00001296extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
Luis Henriques6646ea12020-11-12 15:23:21 +00001297extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
Luis Henriques2b838452018-01-05 10:47:21 +00001298extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
1299 loff_t newlen);
Luis Henriques1ab302a2018-01-05 10:47:22 +00001300extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
1301 loff_t newlen);
Luis Henriques9122eed2018-01-31 10:53:13 +00001302extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
1303 struct kstatfs *buf);
Luis Henriques0c44a8e2019-03-21 10:20:10 +00001304extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc);
Luis Henriquesfb18a572018-01-05 10:47:18 +00001305
Sage Weilde576062009-10-06 11:31:07 -07001306#endif /* _FS_CEPH_SUPER_H */