Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1 | #ifndef _FS_CEPH_SUPER_H |
| 2 | #define _FS_CEPH_SUPER_H |
| 3 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 4 | #include <linux/ceph/ceph_debug.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 5 | |
| 6 | #include <asm/unaligned.h> |
| 7 | #include <linux/backing-dev.h> |
| 8 | #include <linux/completion.h> |
| 9 | #include <linux/exportfs.h> |
| 10 | #include <linux/fs.h> |
| 11 | #include <linux/mempool.h> |
| 12 | #include <linux/pagemap.h> |
| 13 | #include <linux/wait.h> |
Stephen Rothwell | f1a3d57 | 2010-01-18 11:53:08 +1100 | [diff] [blame] | 14 | #include <linux/writeback.h> |
Tejun Heo | 5a0e3ad | 2010-03-24 17:04:11 +0900 | [diff] [blame] | 15 | #include <linux/slab.h> |
Guangliang Zhao | c969d9b | 2014-02-16 08:35:52 -0800 | [diff] [blame] | 16 | #include <linux/posix_acl.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 17 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 18 | #include <linux/ceph/libceph.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 19 | |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 20 | #ifdef CONFIG_CEPH_FSCACHE |
| 21 | #include <linux/fscache.h> |
| 22 | #endif |
| 23 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 24 | /* f_type in struct statfs */ |
| 25 | #define CEPH_SUPER_MAGIC 0x00c36400 |
| 26 | |
| 27 | /* large granularity for statfs utilization stats to facilitate |
| 28 | * large volume sizes on 32-bit machines. */ |
Sage Weil | 92a49fb | 2013-02-22 15:31:00 -0800 | [diff] [blame] | 29 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 30 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
| 31 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 32 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
| 33 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ |
| 34 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 35 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
Sage Weil | a40dc6c | 2012-01-10 09:12:55 -0800 | [diff] [blame] | 36 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 37 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
Sage Weil | 6a25938 | 2010-07-07 09:06:08 -0700 | [diff] [blame] | 38 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 39 | #define CEPH_MOUNT_OPT_DEFAULT (CEPH_MOUNT_OPT_RBYTES) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 40 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 41 | #define ceph_set_mount_opt(fsc, opt) \ |
| 42 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; |
| 43 | #define ceph_test_mount_opt(fsc, opt) \ |
| 44 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 45 | |
Sage Weil | 83817e3 | 2011-08-04 08:03:44 -0700 | [diff] [blame] | 46 | #define CEPH_RSIZE_DEFAULT 0 /* max read size */ |
| 47 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 48 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
| 49 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) |
| 50 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 51 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 52 | struct ceph_mount_options { |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 53 | int flags; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 54 | int sb_flags; |
| 55 | |
Sage Weil | 83817e3 | 2011-08-04 08:03:44 -0700 | [diff] [blame] | 56 | int wsize; /* max write size */ |
| 57 | int rsize; /* max read size */ |
| 58 | int rasize; /* max readahead */ |
Sage Weil | 6e19a16 | 2010-04-29 16:38:32 -0700 | [diff] [blame] | 59 | int congestion_kb; /* max writeback in flight */ |
| 60 | int caps_wanted_delay_min, caps_wanted_delay_max; |
| 61 | int cap_release_safety; |
Sage Weil | 23804d9 | 2010-05-14 13:06:30 -0700 | [diff] [blame] | 62 | int max_readdir; /* max readdir result (entires) */ |
| 63 | int max_readdir_bytes; /* max readdir result (bytes) */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 64 | |
| 65 | /* |
| 66 | * everything above this point can be memcmp'd; everything below |
| 67 | * is handled in compare_mount_options() |
| 68 | */ |
| 69 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 70 | char *snapdir_name; /* default ".snap" */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 71 | }; |
| 72 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 73 | struct ceph_fs_client { |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 74 | struct super_block *sb; |
| 75 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 76 | struct ceph_mount_options *mount_options; |
| 77 | struct ceph_client *client; |
| 78 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 79 | unsigned long mount_state; |
Sage Weil | 85ccce4 | 2010-02-17 10:02:43 -0800 | [diff] [blame] | 80 | int min_caps; /* min caps i added */ |
| 81 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 82 | struct ceph_mds_client *mdsc; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 83 | |
| 84 | /* writeback */ |
| 85 | mempool_t *wb_pagevec_pool; |
| 86 | struct workqueue_struct *wb_wq; |
| 87 | struct workqueue_struct *pg_inv_wq; |
| 88 | struct workqueue_struct *trunc_wq; |
Yehuda Sadeh | 2baba25 | 2009-12-18 13:51:57 -0800 | [diff] [blame] | 89 | atomic_long_t writeback_count; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 90 | |
| 91 | struct backing_dev_info backing_dev_info; |
Sage Weil | 0743304 | 2009-11-18 16:50:41 -0800 | [diff] [blame] | 92 | |
| 93 | #ifdef CONFIG_DEBUG_FS |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 94 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
Yehuda Sadeh | 2baba25 | 2009-12-18 13:51:57 -0800 | [diff] [blame] | 95 | struct dentry *debugfs_congestion_kb; |
Sage Weil | 06edf04 | 2009-12-15 14:44:32 -0800 | [diff] [blame] | 96 | struct dentry *debugfs_bdi; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 97 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
John Spray | 14ed970 | 2014-09-12 16:58:49 +0100 | [diff] [blame] | 98 | struct dentry *debugfs_mds_sessions; |
Sage Weil | 0743304 | 2009-11-18 16:50:41 -0800 | [diff] [blame] | 99 | #endif |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 100 | |
| 101 | #ifdef CONFIG_CEPH_FSCACHE |
| 102 | struct fscache_cookie *fscache; |
| 103 | struct workqueue_struct *revalidate_wq; |
| 104 | #endif |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 105 | }; |
| 106 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 107 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 108 | /* |
| 109 | * File i/o capability. This tracks shared state with the metadata |
| 110 | * server that allows us to cache or writeback attributes or to read |
| 111 | * and write data. For any given inode, we should have one or more |
| 112 | * capabilities, one issued by each metadata server, and our |
| 113 | * cumulative access is the OR of all issued capabilities. |
| 114 | * |
| 115 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds |
| 116 | * session capability lists. |
| 117 | */ |
| 118 | struct ceph_cap { |
| 119 | struct ceph_inode_info *ci; |
| 120 | struct rb_node ci_node; /* per-ci cap tree */ |
| 121 | struct ceph_mds_session *session; |
| 122 | struct list_head session_caps; /* per-session caplist */ |
| 123 | int mds; |
| 124 | u64 cap_id; /* unique cap id (mds provided) */ |
| 125 | int issued; /* latest, from the mds */ |
| 126 | int implemented; /* implemented superset of issued (for revocation) */ |
| 127 | int mds_wanted; |
Sage Weil | 685f9a5d | 2009-11-09 12:05:48 -0800 | [diff] [blame] | 128 | u32 seq, issue_seq, mseq; |
| 129 | u32 cap_gen; /* active/stale cycle */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 130 | unsigned long last_used; |
| 131 | struct list_head caps_item; |
| 132 | }; |
| 133 | |
| 134 | #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ |
| 135 | #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ |
| 136 | #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ |
| 137 | |
| 138 | /* |
| 139 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, |
| 140 | * we first complete any in-process sync writes and writeback any dirty |
| 141 | * data before flushing the snapped state (tracked here) back to the MDS. |
| 142 | */ |
| 143 | struct ceph_cap_snap { |
| 144 | atomic_t nref; |
| 145 | struct ceph_inode_info *ci; |
| 146 | struct list_head ci_item, flushing_item; |
| 147 | |
| 148 | u64 follows, flush_tid; |
| 149 | int issued, dirty; |
| 150 | struct ceph_snap_context *context; |
| 151 | |
Al Viro | 5706b27 | 2011-07-26 04:52:22 -0400 | [diff] [blame] | 152 | umode_t mode; |
Eric W. Biederman | 05cb11c | 2013-01-31 02:56:19 -0800 | [diff] [blame] | 153 | kuid_t uid; |
| 154 | kgid_t gid; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 155 | |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 156 | struct ceph_buffer *xattr_blob; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 157 | u64 xattr_version; |
| 158 | |
| 159 | u64 size; |
| 160 | struct timespec mtime, atime, ctime; |
| 161 | u64 time_warp_seq; |
| 162 | int writing; /* a sync write is still in progress */ |
| 163 | int dirty_pages; /* dirty pages awaiting writeback */ |
| 164 | }; |
| 165 | |
| 166 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) |
| 167 | { |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 168 | if (atomic_dec_and_test(&capsnap->nref)) { |
| 169 | if (capsnap->xattr_blob) |
| 170 | ceph_buffer_put(capsnap->xattr_blob); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 171 | kfree(capsnap); |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 172 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 173 | } |
| 174 | |
| 175 | /* |
| 176 | * The frag tree describes how a directory is fragmented, potentially across |
| 177 | * multiple metadata servers. It is also used to indicate points where |
| 178 | * metadata authority is delegated, and whether/where metadata is replicated. |
| 179 | * |
| 180 | * A _leaf_ frag will be present in the i_fragtree IFF there is |
| 181 | * delegation info. That is, if mds >= 0 || ndist > 0. |
| 182 | */ |
| 183 | #define CEPH_MAX_DIRFRAG_REP 4 |
| 184 | |
| 185 | struct ceph_inode_frag { |
| 186 | struct rb_node node; |
| 187 | |
| 188 | /* fragtree state */ |
| 189 | u32 frag; |
| 190 | int split_by; /* i.e. 2^(split_by) children */ |
| 191 | |
| 192 | /* delegation and replication info */ |
| 193 | int mds; /* -1 if same authority as parent */ |
| 194 | int ndist; /* >0 if replicated */ |
| 195 | int dist[CEPH_MAX_DIRFRAG_REP]; |
| 196 | }; |
| 197 | |
| 198 | /* |
| 199 | * We cache inode xattrs as an encoded blob until they are first used, |
| 200 | * at which point we parse them into an rbtree. |
| 201 | */ |
| 202 | struct ceph_inode_xattr { |
| 203 | struct rb_node node; |
| 204 | |
| 205 | const char *name; |
| 206 | int name_len; |
| 207 | const char *val; |
| 208 | int val_len; |
| 209 | int dirty; |
| 210 | |
| 211 | int should_free_name; |
| 212 | int should_free_val; |
| 213 | }; |
| 214 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 215 | /* |
| 216 | * Ceph dentry state |
| 217 | */ |
| 218 | struct ceph_dentry_info { |
| 219 | struct ceph_mds_session *lease_session; |
| 220 | u32 lease_gen, lease_shared_gen; |
| 221 | u32 lease_seq; |
| 222 | unsigned long lease_renew_after, lease_renew_from; |
| 223 | struct list_head lru; |
| 224 | struct dentry *dentry; |
| 225 | u64 time; |
| 226 | u64 offset; |
| 227 | }; |
| 228 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 229 | struct ceph_inode_xattrs_info { |
| 230 | /* |
| 231 | * (still encoded) xattr blob. we avoid the overhead of parsing |
| 232 | * this until someone actually calls getxattr, etc. |
| 233 | * |
| 234 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == |
| 235 | * NULL means we don't know. |
| 236 | */ |
| 237 | struct ceph_buffer *blob, *prealloc_blob; |
| 238 | |
| 239 | struct rb_root index; |
| 240 | bool dirty; |
| 241 | int count; |
| 242 | int names_size; |
| 243 | int vals_size; |
| 244 | u64 version, index_version; |
| 245 | }; |
| 246 | |
| 247 | /* |
| 248 | * Ceph inode. |
| 249 | */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 250 | struct ceph_inode_info { |
| 251 | struct ceph_vino i_vino; /* ceph ino + snap */ |
| 252 | |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 253 | spinlock_t i_ceph_lock; |
| 254 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 255 | u64 i_version; |
| 256 | u32 i_time_warp_seq; |
| 257 | |
| 258 | unsigned i_ceph_flags; |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 259 | atomic_t i_release_count; |
| 260 | atomic_t i_complete_count; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 261 | |
Sage Weil | 6c0f3af | 2010-11-16 11:14:34 -0800 | [diff] [blame] | 262 | struct ceph_dir_layout i_dir_layout; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 263 | struct ceph_file_layout i_layout; |
| 264 | char *i_symlink; |
| 265 | |
| 266 | /* for dirs */ |
| 267 | struct timespec i_rctime; |
| 268 | u64 i_rbytes, i_rfiles, i_rsubdirs; |
| 269 | u64 i_files, i_subdirs; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 270 | |
| 271 | struct rb_root i_fragtree; |
| 272 | struct mutex i_fragtree_mutex; |
| 273 | |
| 274 | struct ceph_inode_xattrs_info i_xattrs; |
| 275 | |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 276 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 277 | * s_mutex. */ |
| 278 | struct rb_root i_caps; /* cap list */ |
| 279 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ |
| 280 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ |
| 281 | struct list_head i_dirty_item, i_flushing_item; |
| 282 | u64 i_cap_flush_seq; |
| 283 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
| 284 | * overlapping, pipelined cap flushes to the mds. we can probably |
| 285 | * reduce the tid to 8 bits if we're concerned about inode size. */ |
| 286 | u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS]; |
| 287 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
| 288 | unsigned long i_hold_caps_min; /* jiffies */ |
| 289 | unsigned long i_hold_caps_max; /* jiffies */ |
| 290 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 291 | struct ceph_cap_reservation i_cap_migration_resv; |
| 292 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ |
Sage Weil | 7d8cb26 | 2010-08-24 08:44:16 -0700 | [diff] [blame] | 293 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
| 294 | dirty|flushing caps */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 295 | unsigned i_snap_caps; /* cap bits for snapped files */ |
| 296 | |
| 297 | int i_nr_by_mode[CEPH_FILE_MODE_NUM]; /* open file counts */ |
| 298 | |
Yan, Zheng | b0d7c22 | 2013-08-12 21:42:15 -0700 | [diff] [blame] | 299 | struct mutex i_truncate_mutex; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 300 | u32 i_truncate_seq; /* last truncate to smaller size */ |
| 301 | u64 i_truncate_size; /* and the size we last truncated down to */ |
| 302 | int i_truncate_pending; /* still need to call vmtruncate */ |
| 303 | |
| 304 | u64 i_max_size; /* max file size authorized by mds */ |
| 305 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ |
| 306 | u64 i_wanted_max_size; /* offset we'd like to write too */ |
| 307 | u64 i_requested_max_size; /* max_size we've requested */ |
| 308 | |
| 309 | /* held references to caps */ |
| 310 | int i_pin_ref; |
Henry C Chang | d3d0720 | 2011-05-11 10:29:54 +0000 | [diff] [blame] | 311 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 312 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
| 313 | u32 i_shared_gen; /* increment each time we get FILE_SHARED */ |
Sage Weil | cd045cb | 2010-11-04 11:05:05 -0700 | [diff] [blame] | 314 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 315 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
| 316 | |
| 317 | struct list_head i_unsafe_writes; /* uncommitted sync writes */ |
| 318 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
| 319 | spinlock_t i_unsafe_lock; |
| 320 | |
| 321 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ |
| 322 | int i_snap_realm_counter; /* snap realm (if caps) */ |
| 323 | struct list_head i_snap_realm_item; |
| 324 | struct list_head i_snap_flush_item; |
| 325 | |
| 326 | struct work_struct i_wb_work; /* writeback work */ |
| 327 | struct work_struct i_pg_inv_work; /* page invalidation work */ |
| 328 | |
| 329 | struct work_struct i_vmtruncate_work; |
| 330 | |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 331 | #ifdef CONFIG_CEPH_FSCACHE |
| 332 | struct fscache_cookie *fscache; |
| 333 | u32 i_fscache_gen; /* sequence, for delayed fscache validate */ |
| 334 | struct work_struct i_revalidate_work; |
| 335 | #endif |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 336 | struct inode vfs_inode; /* at end */ |
| 337 | }; |
| 338 | |
| 339 | static inline struct ceph_inode_info *ceph_inode(struct inode *inode) |
| 340 | { |
Noah Watkins | fbbccec | 2009-10-28 11:54:49 -0700 | [diff] [blame] | 341 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 342 | } |
| 343 | |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 344 | static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) |
| 345 | { |
| 346 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; |
| 347 | } |
| 348 | |
| 349 | static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) |
| 350 | { |
| 351 | return (struct ceph_fs_client *)sb->s_fs_info; |
| 352 | } |
| 353 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 354 | static inline struct ceph_vino ceph_vino(struct inode *inode) |
| 355 | { |
| 356 | return ceph_inode(inode)->i_vino; |
| 357 | } |
| 358 | |
| 359 | /* |
| 360 | * ino_t is <64 bits on many architectures, blech. |
| 361 | * |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 362 | * i_ino (kernel inode) st_ino (userspace) |
| 363 | * i386 32 32 |
| 364 | * x86_64+ino32 64 32 |
| 365 | * x86_64 64 64 |
| 366 | */ |
Amon Ott | 3310f75 | 2011-10-20 13:04:07 -0700 | [diff] [blame] | 367 | static inline u32 ceph_ino_to_ino32(__u64 vino) |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 368 | { |
Amon Ott | 3310f75 | 2011-10-20 13:04:07 -0700 | [diff] [blame] | 369 | u32 ino = vino & 0xffffffff; |
| 370 | ino ^= vino >> 32; |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 371 | if (!ino) |
Amon Ott | a661fc5 | 2012-01-23 09:25:23 -0800 | [diff] [blame] | 372 | ino = 2; |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 373 | return ino; |
| 374 | } |
| 375 | |
| 376 | /* |
| 377 | * kernel i_ino value |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 378 | */ |
| 379 | static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) |
| 380 | { |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 381 | #if BITS_PER_LONG == 32 |
Amon Ott | 3310f75 | 2011-10-20 13:04:07 -0700 | [diff] [blame] | 382 | return ceph_ino_to_ino32(vino.ino); |
| 383 | #else |
| 384 | return (ino_t)vino.ino; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 385 | #endif |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 386 | } |
| 387 | |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 388 | /* |
| 389 | * user-visible ino (stat, filldir) |
| 390 | */ |
| 391 | #if BITS_PER_LONG == 32 |
| 392 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) |
| 393 | { |
| 394 | return ino; |
| 395 | } |
| 396 | #else |
| 397 | static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) |
| 398 | { |
| 399 | if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) |
| 400 | ino = ceph_ino_to_ino32(ino); |
| 401 | return ino; |
| 402 | } |
| 403 | #endif |
| 404 | |
| 405 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 406 | /* for printf-style formatting */ |
| 407 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap |
| 408 | |
| 409 | static inline u64 ceph_ino(struct inode *inode) |
| 410 | { |
| 411 | return ceph_inode(inode)->i_vino.ino; |
| 412 | } |
| 413 | static inline u64 ceph_snap(struct inode *inode) |
| 414 | { |
| 415 | return ceph_inode(inode)->i_vino.snap; |
| 416 | } |
| 417 | |
| 418 | static inline int ceph_ino_compare(struct inode *inode, void *data) |
| 419 | { |
| 420 | struct ceph_vino *pvino = (struct ceph_vino *)data; |
| 421 | struct ceph_inode_info *ci = ceph_inode(inode); |
| 422 | return ci->i_vino.ino == pvino->ino && |
| 423 | ci->i_vino.snap == pvino->snap; |
| 424 | } |
| 425 | |
| 426 | static inline struct inode *ceph_find_inode(struct super_block *sb, |
| 427 | struct ceph_vino vino) |
| 428 | { |
| 429 | ino_t t = ceph_vino_to_ino(vino); |
| 430 | return ilookup5(sb, t, ceph_ino_compare, &vino); |
| 431 | } |
| 432 | |
| 433 | |
| 434 | /* |
| 435 | * Ceph inode. |
| 436 | */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 437 | #define CEPH_I_NODELAY 4 /* do not delay cap release */ |
| 438 | #define CEPH_I_FLUSH 8 /* do not delay flush of dirty metadata */ |
| 439 | #define CEPH_I_NOFLUSH 16 /* do not flush dirty caps */ |
| 440 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 441 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
| 442 | int release_count) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 443 | { |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 444 | atomic_set(&ci->i_complete_count, release_count); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 445 | } |
| 446 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 447 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 448 | { |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 449 | atomic_inc(&ci->i_release_count); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 450 | } |
| 451 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 452 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 453 | { |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 454 | return atomic_read(&ci->i_complete_count) == |
| 455 | atomic_read(&ci->i_release_count); |
| 456 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 457 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 458 | static inline void ceph_dir_clear_complete(struct inode *inode) |
| 459 | { |
| 460 | __ceph_dir_clear_complete(ceph_inode(inode)); |
| 461 | } |
| 462 | |
| 463 | static inline bool ceph_dir_is_complete(struct inode *inode) |
| 464 | { |
| 465 | return __ceph_dir_is_complete(ceph_inode(inode)); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 466 | } |
| 467 | |
| 468 | |
| 469 | /* find a specific frag @f */ |
| 470 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, |
| 471 | u32 f); |
| 472 | |
| 473 | /* |
| 474 | * choose fragment for value @v. copy frag content to pfrag, if leaf |
| 475 | * exists |
| 476 | */ |
| 477 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, |
| 478 | struct ceph_inode_frag *pfrag, |
| 479 | int *found); |
| 480 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 481 | static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) |
| 482 | { |
| 483 | return (struct ceph_dentry_info *)dentry->d_fsdata; |
| 484 | } |
| 485 | |
| 486 | static inline loff_t ceph_make_fpos(unsigned frag, unsigned off) |
| 487 | { |
| 488 | return ((loff_t)frag << 32) | (loff_t)off; |
| 489 | } |
| 490 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 491 | /* |
| 492 | * caps helpers |
| 493 | */ |
| 494 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) |
| 495 | { |
| 496 | return !RB_EMPTY_ROOT(&ci->i_caps); |
| 497 | } |
| 498 | |
| 499 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); |
| 500 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); |
| 501 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, |
| 502 | struct ceph_cap *cap); |
| 503 | |
| 504 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) |
| 505 | { |
| 506 | int issued; |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 507 | spin_lock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 508 | issued = __ceph_caps_issued(ci, NULL); |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 509 | spin_unlock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 510 | return issued; |
| 511 | } |
| 512 | |
| 513 | static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, |
| 514 | int touch) |
| 515 | { |
| 516 | int r; |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 517 | spin_lock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 518 | r = __ceph_caps_issued_mask(ci, mask, touch); |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 519 | spin_unlock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 520 | return r; |
| 521 | } |
| 522 | |
| 523 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) |
| 524 | { |
| 525 | return ci->i_dirty_caps | ci->i_flushing_caps; |
| 526 | } |
Sage Weil | fca65b4 | 2011-05-04 11:33:47 -0700 | [diff] [blame] | 527 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 528 | |
Yan, Zheng | 9563f88 | 2013-11-22 13:50:45 +0800 | [diff] [blame] | 529 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
| 530 | struct ceph_cap *ocap, int mask); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 531 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
| 532 | extern int __ceph_caps_used(struct ceph_inode_info *ci); |
| 533 | |
| 534 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); |
| 535 | |
| 536 | /* |
| 537 | * wanted, by virtue of open file modes AND cap refs (buffered/cached data) |
| 538 | */ |
| 539 | static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) |
| 540 | { |
| 541 | int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); |
| 542 | if (w & CEPH_CAP_FILE_BUFFER) |
| 543 | w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ |
| 544 | return w; |
| 545 | } |
| 546 | |
| 547 | /* what the mds thinks we want */ |
| 548 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci); |
| 549 | |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 550 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
| 551 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); |
| 552 | extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); |
majianpeng | 93faca6 | 2013-06-26 11:15:27 +0800 | [diff] [blame] | 553 | extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 554 | struct ceph_cap_reservation *ctx, int need); |
| 555 | extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
| 556 | struct ceph_cap_reservation *ctx); |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 557 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 558 | int *total, int *avail, int *used, |
Sage Weil | 85ccce4 | 2010-02-17 10:02:43 -0800 | [diff] [blame] | 559 | int *reserved, int *min); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 560 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 561 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 562 | |
| 563 | /* |
| 564 | * we keep buffered readdir results attached to file->private_data |
| 565 | */ |
Sage Weil | 4918b6d | 2011-07-26 11:26:07 -0700 | [diff] [blame] | 566 | #define CEPH_F_SYNC 1 |
Sage Weil | 9cfa109 | 2011-07-26 11:26:18 -0700 | [diff] [blame] | 567 | #define CEPH_F_ATEND 2 |
Sage Weil | 4918b6d | 2011-07-26 11:26:07 -0700 | [diff] [blame] | 568 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 569 | struct ceph_file_info { |
Sage Weil | 252c672 | 2011-07-26 11:25:27 -0700 | [diff] [blame] | 570 | short fmode; /* initialized on open */ |
| 571 | short flags; /* CEPH_F_* */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 572 | |
| 573 | /* readdir: position within the dir */ |
| 574 | u32 frag; |
| 575 | struct ceph_mds_request *last_readdir; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 576 | |
| 577 | /* readdir: position within a frag */ |
| 578 | unsigned offset; /* offset of last chunk, adjusted for . and .. */ |
Yan, Zheng | f049420 | 2014-02-27 16:26:24 +0800 | [diff] [blame] | 579 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 580 | char *last_name; /* last entry in previous chunk */ |
| 581 | struct dentry *dentry; /* next dentry (for dcache readdir) */ |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 582 | int dir_release_count; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 583 | |
| 584 | /* used for -o dirstat read() on directory thing */ |
| 585 | char *dir_info; |
| 586 | int dir_info_len; |
| 587 | }; |
| 588 | |
| 589 | |
| 590 | |
| 591 | /* |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 592 | * A "snap realm" describes a subset of the file hierarchy sharing |
| 593 | * the same set of snapshots that apply to it. The realms themselves |
| 594 | * are organized into a hierarchy, such that children inherit (some of) |
| 595 | * the snapshots of their parents. |
| 596 | * |
| 597 | * All inodes within the realm that have capabilities are linked into a |
| 598 | * per-realm list. |
| 599 | */ |
| 600 | struct ceph_snap_realm { |
| 601 | u64 ino; |
| 602 | atomic_t nref; |
Sage Weil | a105f00 | 2010-02-15 14:37:55 -0800 | [diff] [blame] | 603 | struct rb_node node; |
| 604 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 605 | u64 created, seq; |
| 606 | u64 parent_ino; |
| 607 | u64 parent_since; /* snapid when our current parent became so */ |
| 608 | |
| 609 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ |
Alex Elder | aa711ee | 2012-07-13 20:35:11 -0500 | [diff] [blame] | 610 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 611 | u64 *snaps; /* snaps specific to this realm */ |
Alex Elder | aa711ee | 2012-07-13 20:35:11 -0500 | [diff] [blame] | 612 | u32 num_snaps; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 613 | |
| 614 | struct ceph_snap_realm *parent; |
| 615 | struct list_head children; /* list of child realms */ |
| 616 | struct list_head child_item; |
| 617 | |
| 618 | struct list_head empty_item; /* if i have ref==0 */ |
| 619 | |
Sage Weil | ae00d4f | 2010-09-16 16:26:51 -0700 | [diff] [blame] | 620 | struct list_head dirty_item; /* if realm needs new context */ |
| 621 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 622 | /* the current set of snaps for this realm */ |
| 623 | struct ceph_snap_context *cached_context; |
| 624 | |
| 625 | struct list_head inodes_with_caps; |
| 626 | spinlock_t inodes_with_caps_lock; |
| 627 | }; |
| 628 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 629 | static inline int default_congestion_kb(void) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 630 | { |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 631 | int congestion_kb; |
| 632 | |
| 633 | /* |
| 634 | * Copied from NFS |
| 635 | * |
| 636 | * congestion size, scale with available memory. |
| 637 | * |
| 638 | * 64MB: 8192k |
| 639 | * 128MB: 11585k |
| 640 | * 256MB: 16384k |
| 641 | * 512MB: 23170k |
| 642 | * 1GB: 32768k |
| 643 | * 2GB: 46340k |
| 644 | * 4GB: 65536k |
| 645 | * 8GB: 92681k |
| 646 | * 16GB: 131072k |
| 647 | * |
| 648 | * This allows larger machines to have larger/more transfers. |
| 649 | * Limit the default to 256M |
| 650 | */ |
| 651 | congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); |
| 652 | if (congestion_kb > 256*1024) |
| 653 | congestion_kb = 256*1024; |
| 654 | |
| 655 | return congestion_kb; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 656 | } |
| 657 | |
| 658 | |
| 659 | |
| 660 | /* snap.c */ |
| 661 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, |
| 662 | u64 ino); |
| 663 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, |
| 664 | struct ceph_snap_realm *realm); |
| 665 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, |
| 666 | struct ceph_snap_realm *realm); |
| 667 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, |
| 668 | void *p, void *e, bool deletion); |
| 669 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
Sage Weil | 2600d2d | 2010-02-22 15:12:16 -0800 | [diff] [blame] | 670 | struct ceph_mds_session *session, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 671 | struct ceph_msg *msg); |
Sage Weil | fc837c8f | 2010-04-13 11:41:22 -0700 | [diff] [blame] | 672 | extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 673 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
| 674 | struct ceph_cap_snap *capsnap); |
| 675 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); |
| 676 | |
| 677 | /* |
| 678 | * a cap_snap is "pending" if it is still awaiting an in-progress |
| 679 | * sync write (that may/may not still update size, mtime, etc.). |
| 680 | */ |
| 681 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) |
| 682 | { |
| 683 | return !list_empty(&ci->i_cap_snaps) && |
| 684 | list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap, |
| 685 | ci_item)->writing; |
| 686 | } |
| 687 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 688 | /* inode.c */ |
| 689 | extern const struct inode_operations ceph_file_iops; |
| 690 | |
| 691 | extern struct inode *ceph_alloc_inode(struct super_block *sb); |
| 692 | extern void ceph_destroy_inode(struct inode *inode); |
Yan, Zheng | 9f12bd1 | 2013-09-20 19:55:31 +0800 | [diff] [blame] | 693 | extern int ceph_drop_inode(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 694 | |
| 695 | extern struct inode *ceph_get_inode(struct super_block *sb, |
| 696 | struct ceph_vino vino); |
| 697 | extern struct inode *ceph_get_snapdir(struct inode *parent); |
| 698 | extern int ceph_fill_file_size(struct inode *inode, int issued, |
| 699 | u32 truncate_seq, u64 truncate_size, u64 size); |
| 700 | extern void ceph_fill_file_time(struct inode *inode, int issued, |
| 701 | u64 time_warp_seq, struct timespec *ctime, |
| 702 | struct timespec *mtime, struct timespec *atime); |
| 703 | extern int ceph_fill_trace(struct super_block *sb, |
| 704 | struct ceph_mds_request *req, |
| 705 | struct ceph_mds_session *session); |
| 706 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
| 707 | struct ceph_mds_session *session); |
| 708 | |
| 709 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); |
| 710 | |
| 711 | extern int ceph_inode_set_size(struct inode *inode, loff_t size); |
Yan, Zheng | b415bf4 | 2013-07-02 12:40:19 +0800 | [diff] [blame] | 712 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
Sage Weil | 3c6f6b7 | 2010-02-09 15:24:44 -0800 | [diff] [blame] | 713 | extern void ceph_queue_vmtruncate(struct inode *inode); |
| 714 | |
| 715 | extern void ceph_queue_invalidate(struct inode *inode); |
| 716 | extern void ceph_queue_writeback(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 717 | |
Yan, Zheng | 508b32d | 2014-09-16 21:46:17 +0800 | [diff] [blame] | 718 | extern int ceph_do_getattr(struct inode *inode, int mask, bool force); |
Al Viro | 10556cb | 2011-06-20 19:28:19 -0400 | [diff] [blame] | 719 | extern int ceph_permission(struct inode *inode, int mask); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 720 | extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); |
| 721 | extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry, |
| 722 | struct kstat *stat); |
| 723 | |
| 724 | /* xattr.c */ |
| 725 | extern int ceph_setxattr(struct dentry *, const char *, const void *, |
| 726 | size_t, int); |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 727 | int __ceph_setxattr(struct dentry *, const char *, const void *, size_t, int); |
| 728 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
| 729 | int __ceph_removexattr(struct dentry *, const char *); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 730 | extern ssize_t ceph_getxattr(struct dentry *, const char *, void *, size_t); |
| 731 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
| 732 | extern int ceph_removexattr(struct dentry *, const char *); |
| 733 | extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
| 734 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); |
Alex Elder | 3ce6cd1 | 2012-01-23 15:49:28 -0600 | [diff] [blame] | 735 | extern void __init ceph_xattr_init(void); |
| 736 | extern void ceph_xattr_exit(void); |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 737 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 738 | |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 739 | /* acl.c */ |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 740 | struct ceph_acls_info { |
| 741 | void *default_acl; |
| 742 | void *acl; |
| 743 | struct ceph_pagelist *pagelist; |
| 744 | }; |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 745 | |
| 746 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
| 747 | |
| 748 | struct posix_acl *ceph_get_acl(struct inode *, int); |
Sage Weil | 72466d0 | 2014-01-29 06:22:25 -0800 | [diff] [blame] | 749 | int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 750 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
| 751 | struct ceph_acls_info *info); |
| 752 | void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); |
| 753 | void ceph_release_acls_info(struct ceph_acls_info *info); |
Guangliang Zhao | c969d9b | 2014-02-16 08:35:52 -0800 | [diff] [blame] | 754 | |
| 755 | static inline void ceph_forget_all_cached_acls(struct inode *inode) |
| 756 | { |
| 757 | forget_all_cached_acls(inode); |
| 758 | } |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 759 | |
| 760 | #else |
| 761 | |
| 762 | #define ceph_get_acl NULL |
Sage Weil | 72466d0 | 2014-01-29 06:22:25 -0800 | [diff] [blame] | 763 | #define ceph_set_acl NULL |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 764 | |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 765 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
| 766 | struct ceph_acls_info *info) |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 767 | { |
| 768 | return 0; |
| 769 | } |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 770 | static inline void ceph_init_inode_acls(struct inode *inode, |
| 771 | struct ceph_acls_info *info) |
| 772 | { |
| 773 | } |
| 774 | static inline void ceph_release_acls_info(struct ceph_acls_info *info) |
| 775 | { |
| 776 | } |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 777 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
| 778 | { |
| 779 | return 0; |
| 780 | } |
| 781 | |
| 782 | static inline void ceph_forget_all_cached_acls(struct inode *inode) |
| 783 | { |
| 784 | } |
| 785 | |
| 786 | #endif |
| 787 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 788 | /* caps.c */ |
| 789 | extern const char *ceph_cap_string(int c); |
| 790 | extern void ceph_handle_caps(struct ceph_mds_session *session, |
| 791 | struct ceph_msg *msg); |
Yan, Zheng | d9df278 | 2014-04-18 09:57:11 +0800 | [diff] [blame] | 792 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
| 793 | struct ceph_cap_reservation *ctx); |
| 794 | extern void ceph_add_cap(struct inode *inode, |
| 795 | struct ceph_mds_session *session, u64 cap_id, |
| 796 | int fmode, unsigned issued, unsigned wanted, |
| 797 | unsigned cap, unsigned seq, u64 realmino, int flags, |
| 798 | struct ceph_cap **new_cap); |
Yan, Zheng | a096b09 | 2013-09-22 10:15:58 +0800 | [diff] [blame] | 799 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 800 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
| 801 | struct ceph_cap *cap); |
Yan, Zheng | 9215aee | 2013-11-30 12:47:41 +0800 | [diff] [blame] | 802 | extern int ceph_is_any_caps(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 803 | |
Yan, Zheng | d40ee0d | 2013-02-18 13:43:43 +0800 | [diff] [blame] | 804 | extern void __queue_cap_release(struct ceph_mds_session *session, u64 ino, |
| 805 | u64 cap_id, u32 migrate_seq, u32 issue_seq); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 806 | extern void ceph_queue_caps_release(struct inode *inode); |
Stephen Rothwell | f1a3d57 | 2010-01-18 11:53:08 +1100 | [diff] [blame] | 807 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
Josef Bacik | 02c24a8 | 2011-07-16 20:44:56 -0400 | [diff] [blame] | 808 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
| 809 | int datasync); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 810 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
| 811 | struct ceph_mds_session *session); |
Greg Farnum | 2bc5025 | 2010-06-30 12:44:34 -0700 | [diff] [blame] | 812 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
| 813 | int mds); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 814 | extern int ceph_get_cap_mds(struct inode *inode); |
| 815 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); |
| 816 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); |
| 817 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, |
| 818 | struct ceph_snap_context *snapc); |
| 819 | extern void __ceph_flush_snaps(struct ceph_inode_info *ci, |
Sage Weil | e835124 | 2010-09-17 08:03:08 -0700 | [diff] [blame] | 820 | struct ceph_mds_session **psession, |
| 821 | int again); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 822 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
| 823 | struct ceph_mds_session *session); |
Sage Weil | afcdaea | 2009-10-14 14:27:38 -0700 | [diff] [blame] | 824 | extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
| 825 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 826 | |
| 827 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
| 828 | int mds, int drop, int unless, int force); |
| 829 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, |
| 830 | int mds, int drop, int unless); |
| 831 | |
| 832 | extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, |
| 833 | int *got, loff_t endoff); |
| 834 | |
| 835 | /* for counting open files by mode */ |
| 836 | static inline void __ceph_get_fmode(struct ceph_inode_info *ci, int mode) |
| 837 | { |
| 838 | ci->i_nr_by_mode[mode]++; |
| 839 | } |
| 840 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); |
| 841 | |
| 842 | /* addr.c */ |
| 843 | extern const struct address_space_operations ceph_aops; |
| 844 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); |
| 845 | |
| 846 | /* file.c */ |
| 847 | extern const struct file_operations ceph_file_fops; |
| 848 | extern const struct address_space_operations ceph_aops; |
Alex Elder | 9e0eb85 | 2013-02-06 13:11:38 -0600 | [diff] [blame] | 849 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 850 | extern int ceph_open(struct inode *inode, struct file *file); |
Sage Weil | 5ef50c3 | 2012-07-31 11:27:36 -0700 | [diff] [blame] | 851 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
| 852 | struct file *file, unsigned flags, umode_t mode, |
| 853 | int *opened); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 854 | extern int ceph_release(struct inode *inode, struct file *filp); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 855 | |
| 856 | /* dir.c */ |
| 857 | extern const struct file_operations ceph_dir_fops; |
| 858 | extern const struct inode_operations ceph_dir_iops; |
Sage Weil | 52dfb8a | 2010-08-03 10:25:30 -0700 | [diff] [blame] | 859 | extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 860 | ceph_snapdir_dentry_ops; |
| 861 | |
| 862 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
Sage Weil | 468640e | 2011-07-26 11:28:11 -0700 | [diff] [blame] | 863 | extern int ceph_handle_snapdir(struct ceph_mds_request *req, |
| 864 | struct dentry *dentry, int err); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 865 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
| 866 | struct dentry *dentry, int err); |
| 867 | |
| 868 | extern void ceph_dentry_lru_add(struct dentry *dn); |
| 869 | extern void ceph_dentry_lru_touch(struct dentry *dn); |
| 870 | extern void ceph_dentry_lru_del(struct dentry *dn); |
Sage Weil | 81a6cf2 | 2010-05-14 09:35:38 -0700 | [diff] [blame] | 871 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
Sage Weil | e5f86dc | 2011-07-26 11:30:55 -0700 | [diff] [blame] | 872 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
Sage Weil | 5f21c96 | 2011-07-26 11:30:29 -0700 | [diff] [blame] | 873 | extern struct inode *ceph_get_dentry_parent_inode(struct dentry *dentry); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 874 | |
| 875 | /* |
| 876 | * our d_ops vary depending on whether the inode is live, |
| 877 | * snapshotted (read-only), or a virtual ".snap" directory. |
| 878 | */ |
| 879 | int ceph_init_dentry(struct dentry *dentry); |
| 880 | |
| 881 | |
| 882 | /* ioctl.c */ |
| 883 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
| 884 | |
| 885 | /* export.c */ |
| 886 | extern const struct export_operations ceph_export_ops; |
| 887 | |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 888 | /* locks.c */ |
Yan, Zheng | eb13e83 | 2014-03-09 23:16:40 +0800 | [diff] [blame] | 889 | extern __init void ceph_flock_init(void); |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 890 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
| 891 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); |
| 892 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); |
Jim Schutt | 39be95e | 2013-05-15 13:03:35 -0500 | [diff] [blame] | 893 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
| 894 | struct ceph_filelock *flocks, |
| 895 | int num_fcntl_locks, |
| 896 | int num_flock_locks); |
| 897 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, |
| 898 | struct ceph_pagelist *pagelist, |
| 899 | int num_fcntl_locks, int num_flock_locks); |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 900 | extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c); |
| 901 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 902 | /* debugfs.c */ |
| 903 | extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); |
| 904 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); |
| 905 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 906 | #endif /* _FS_CEPH_SUPER_H */ |