1de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H 2de57606cSSage Weil #define _FS_CEPH_SUPER_H 3de57606cSSage Weil 43d14c5d2SYehuda Sadeh #include <linux/ceph/ceph_debug.h> 5de57606cSSage Weil 6de57606cSSage Weil #include <asm/unaligned.h> 7de57606cSSage Weil #include <linux/backing-dev.h> 8de57606cSSage Weil #include <linux/completion.h> 9de57606cSSage Weil #include <linux/exportfs.h> 10de57606cSSage Weil #include <linux/fs.h> 11de57606cSSage Weil #include <linux/mempool.h> 12de57606cSSage Weil #include <linux/pagemap.h> 13de57606cSSage Weil #include <linux/wait.h> 14f1a3d572SStephen Rothwell #include <linux/writeback.h> 155a0e3ad6STejun Heo #include <linux/slab.h> 16c969d9bfSGuangliang Zhao #include <linux/posix_acl.h> 17*805692d0SElena Reshetova #include <linux/refcount.h> 18de57606cSSage Weil 193d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 20de57606cSSage Weil 2199ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 2299ccbd22SMilosz Tanski #include <linux/fscache.h> 2399ccbd22SMilosz Tanski #endif 2499ccbd22SMilosz Tanski 25de57606cSSage Weil /* f_type in struct statfs */ 26de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400 27de57606cSSage Weil 28de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 29de57606cSSage Weil * large volume sizes on 32-bit machines. */ 3092a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 31de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 32de57606cSSage Weil 333d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 343d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 353d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 36ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 37a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 3899ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 3910183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 40e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 416a259382SSage Weil 42133e9156SYan, Zheng #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE 43de57606cSSage Weil 443d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 453d14c5d2SYehuda Sadeh (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; 463d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 473d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 48de57606cSSage Weil 497c94ba27SAndreas Gerstmayr #define CEPH_RSIZE_DEFAULT (64*1024*1024) /* max read size */ 507c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 513d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 523d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 533d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 54de57606cSSage Weil 553d14c5d2SYehuda Sadeh struct ceph_mount_options { 56de57606cSSage Weil int flags; 573d14c5d2SYehuda Sadeh int sb_flags; 583d14c5d2SYehuda Sadeh 5983817e35SSage Weil int wsize; /* max write size */ 6083817e35SSage Weil int rsize; /* max read size */ 6183817e35SSage Weil int rasize; /* max readahead */ 626e19a16eSSage Weil int congestion_kb; /* max writeback in flight */ 636e19a16eSSage Weil int caps_wanted_delay_min, caps_wanted_delay_max; 646e19a16eSSage Weil int cap_release_safety; 6523804d91SSage Weil int max_readdir; /* max readdir result (entires) */ 6623804d91SSage Weil int max_readdir_bytes; /* max readdir result (bytes) */ 673d14c5d2SYehuda Sadeh 683d14c5d2SYehuda Sadeh /* 693d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 703d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 713d14c5d2SYehuda Sadeh */ 723d14c5d2SYehuda Sadeh 73de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 74430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 753f384954SYan, Zheng char *server_path; /* default "/" */ 76de57606cSSage Weil }; 77de57606cSSage Weil 783d14c5d2SYehuda Sadeh struct ceph_fs_client { 79de57606cSSage Weil struct super_block *sb; 80de57606cSSage Weil 813d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 823d14c5d2SYehuda Sadeh struct ceph_client *client; 833d14c5d2SYehuda Sadeh 84de57606cSSage Weil unsigned long mount_state; 8585ccce43SSage Weil int min_caps; /* min caps i added */ 8685ccce43SSage Weil 873d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 88de57606cSSage Weil 89de57606cSSage Weil /* writeback */ 90de57606cSSage Weil mempool_t *wb_pagevec_pool; 91de57606cSSage Weil struct workqueue_struct *wb_wq; 92de57606cSSage Weil struct workqueue_struct *pg_inv_wq; 93de57606cSSage Weil struct workqueue_struct *trunc_wq; 942baba250SYehuda Sadeh atomic_long_t writeback_count; 95de57606cSSage Weil 96de57606cSSage Weil struct backing_dev_info backing_dev_info; 970743304dSSage Weil 980743304dSSage Weil #ifdef CONFIG_DEBUG_FS 993d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1002baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 10106edf046SSage Weil struct dentry *debugfs_bdi; 1023d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 10314ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 1040743304dSSage Weil #endif 10599ccbd22SMilosz Tanski 10699ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 10799ccbd22SMilosz Tanski struct fscache_cookie *fscache; 10899ccbd22SMilosz Tanski #endif 109de57606cSSage Weil }; 110de57606cSSage Weil 1113d14c5d2SYehuda Sadeh 112de57606cSSage Weil /* 113de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 114de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 115de57606cSSage Weil * and write data. For any given inode, we should have one or more 116de57606cSSage Weil * capabilities, one issued by each metadata server, and our 117de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 118de57606cSSage Weil * 119de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 120de57606cSSage Weil * session capability lists. 121de57606cSSage Weil */ 122de57606cSSage Weil struct ceph_cap { 123de57606cSSage Weil struct ceph_inode_info *ci; 124de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 125de57606cSSage Weil struct ceph_mds_session *session; 126de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 127de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 128745a8e3bSYan, Zheng union { 129745a8e3bSYan, Zheng /* in-use caps */ 130745a8e3bSYan, Zheng struct { 131de57606cSSage Weil int issued; /* latest, from the mds */ 132745a8e3bSYan, Zheng int implemented; /* implemented superset of 133745a8e3bSYan, Zheng issued (for revocation) */ 134745a8e3bSYan, Zheng int mds, mds_wanted; 135745a8e3bSYan, Zheng }; 136745a8e3bSYan, Zheng /* caps to release */ 137745a8e3bSYan, Zheng struct { 138745a8e3bSYan, Zheng u64 cap_ino; 139745a8e3bSYan, Zheng int queue_release; 140745a8e3bSYan, Zheng }; 141745a8e3bSYan, Zheng }; 142685f9a5dSSage Weil u32 seq, issue_seq, mseq; 143685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 144de57606cSSage Weil unsigned long last_used; 145de57606cSSage Weil struct list_head caps_item; 146de57606cSSage Weil }; 147de57606cSSage Weil 148de57606cSSage Weil #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ 149de57606cSSage Weil #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ 150de57606cSSage Weil #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ 151de57606cSSage Weil 1520e294387SYan, Zheng struct ceph_cap_flush { 1530e294387SYan, Zheng u64 tid; 1540e294387SYan, Zheng int caps; /* 0 means capsnap */ 155c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 1560e294387SYan, Zheng struct list_head g_list; // global 1570e294387SYan, Zheng struct list_head i_list; // per inode 1580e294387SYan, Zheng }; 1590e294387SYan, Zheng 160de57606cSSage Weil /* 161de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 162de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 163de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 164de57606cSSage Weil */ 165de57606cSSage Weil struct ceph_cap_snap { 166*805692d0SElena Reshetova refcount_t nref; 1670e294387SYan, Zheng struct list_head ci_item; 168de57606cSSage Weil 1690e294387SYan, Zheng struct ceph_cap_flush cap_flush; 1700e294387SYan, Zheng 1710e294387SYan, Zheng u64 follows; 172de57606cSSage Weil int issued, dirty; 173de57606cSSage Weil struct ceph_snap_context *context; 174de57606cSSage Weil 1755706b27dSAl Viro umode_t mode; 17605cb11c1SEric W. Biederman kuid_t uid; 17705cb11c1SEric W. Biederman kgid_t gid; 178de57606cSSage Weil 1794a625be4SSage Weil struct ceph_buffer *xattr_blob; 180de57606cSSage Weil u64 xattr_version; 181de57606cSSage Weil 182de57606cSSage Weil u64 size; 183de57606cSSage Weil struct timespec mtime, atime, ctime; 184de57606cSSage Weil u64 time_warp_seq; 1855f743e45SYan, Zheng u64 truncate_size; 1865f743e45SYan, Zheng u32 truncate_seq; 187de57606cSSage Weil int writing; /* a sync write is still in progress */ 188de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 189e20d258dSYan, Zheng bool inline_data; 19086056090SYan, Zheng bool need_flush; 191de57606cSSage Weil }; 192de57606cSSage Weil 193de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 194de57606cSSage Weil { 195*805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 1964a625be4SSage Weil if (capsnap->xattr_blob) 1974a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 198de57606cSSage Weil kfree(capsnap); 199de57606cSSage Weil } 2004a625be4SSage Weil } 201de57606cSSage Weil 202de57606cSSage Weil /* 203de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 204de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 205de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 206de57606cSSage Weil * 207de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 208de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 209de57606cSSage Weil */ 210de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 211de57606cSSage Weil 212de57606cSSage Weil struct ceph_inode_frag { 213de57606cSSage Weil struct rb_node node; 214de57606cSSage Weil 215de57606cSSage Weil /* fragtree state */ 216de57606cSSage Weil u32 frag; 217de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 218de57606cSSage Weil 219de57606cSSage Weil /* delegation and replication info */ 220de57606cSSage Weil int mds; /* -1 if same authority as parent */ 221de57606cSSage Weil int ndist; /* >0 if replicated */ 222de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 223de57606cSSage Weil }; 224de57606cSSage Weil 225de57606cSSage Weil /* 226de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 227de57606cSSage Weil * at which point we parse them into an rbtree. 228de57606cSSage Weil */ 229de57606cSSage Weil struct ceph_inode_xattr { 230de57606cSSage Weil struct rb_node node; 231de57606cSSage Weil 232de57606cSSage Weil const char *name; 233de57606cSSage Weil int name_len; 234de57606cSSage Weil const char *val; 235de57606cSSage Weil int val_len; 236de57606cSSage Weil int dirty; 237de57606cSSage Weil 238de57606cSSage Weil int should_free_name; 239de57606cSSage Weil int should_free_val; 240de57606cSSage Weil }; 241de57606cSSage Weil 2423d14c5d2SYehuda Sadeh /* 2433d14c5d2SYehuda Sadeh * Ceph dentry state 2443d14c5d2SYehuda Sadeh */ 2453d14c5d2SYehuda Sadeh struct ceph_dentry_info { 2463d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 2473d14c5d2SYehuda Sadeh u32 lease_gen, lease_shared_gen; 2483d14c5d2SYehuda Sadeh u32 lease_seq; 2493d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2503d14c5d2SYehuda Sadeh struct list_head lru; 2513d14c5d2SYehuda Sadeh struct dentry *dentry; 2529b16f03cSMiklos Szeredi unsigned long time; 2533d14c5d2SYehuda Sadeh u64 offset; 2543d14c5d2SYehuda Sadeh }; 2553d14c5d2SYehuda Sadeh 256de57606cSSage Weil struct ceph_inode_xattrs_info { 257de57606cSSage Weil /* 258de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 259de57606cSSage Weil * this until someone actually calls getxattr, etc. 260de57606cSSage Weil * 261de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 262de57606cSSage Weil * NULL means we don't know. 263de57606cSSage Weil */ 264de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 265de57606cSSage Weil 266de57606cSSage Weil struct rb_root index; 267de57606cSSage Weil bool dirty; 268de57606cSSage Weil int count; 269de57606cSSage Weil int names_size; 270de57606cSSage Weil int vals_size; 271de57606cSSage Weil u64 version, index_version; 272de57606cSSage Weil }; 273de57606cSSage Weil 274de57606cSSage Weil /* 275de57606cSSage Weil * Ceph inode. 276de57606cSSage Weil */ 277de57606cSSage Weil struct ceph_inode_info { 278de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 279de57606cSSage Weil 280be655596SSage Weil spinlock_t i_ceph_lock; 281be655596SSage Weil 282de57606cSSage Weil u64 i_version; 28331c542a1SYan, Zheng u64 i_inline_version; 284de57606cSSage Weil u32 i_time_warp_seq; 285de57606cSSage Weil 286de57606cSSage Weil unsigned i_ceph_flags; 287fdd4e158SYan, Zheng atomic64_t i_release_count; 288fdd4e158SYan, Zheng atomic64_t i_ordered_count; 289fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 290de57606cSSage Weil 2916c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 292de57606cSSage Weil struct ceph_file_layout i_layout; 293de57606cSSage Weil char *i_symlink; 294de57606cSSage Weil 295de57606cSSage Weil /* for dirs */ 296de57606cSSage Weil struct timespec i_rctime; 297de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 298de57606cSSage Weil u64 i_files, i_subdirs; 299de57606cSSage Weil 300de57606cSSage Weil struct rb_root i_fragtree; 3011b1bc16dSYan, Zheng int i_fragtree_nsplits; 302de57606cSSage Weil struct mutex i_fragtree_mutex; 303de57606cSSage Weil 304de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 305de57606cSSage Weil 306be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 307de57606cSSage Weil * s_mutex. */ 308de57606cSSage Weil struct rb_root i_caps; /* cap list */ 309de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 310de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 311de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 312de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 313de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 314de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 315f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 316e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 317de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 318de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 319de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 320de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 321de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 322de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3237d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3247d8cb26dSSage Weil dirty|flushing caps */ 325de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 326de57606cSSage Weil 327774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 328de57606cSSage Weil 329b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 330de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 331de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 332de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 333de57606cSSage Weil 334de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 335de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 336de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 337de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 338de57606cSSage Weil 339de57606cSSage Weil /* held references to caps */ 340de57606cSSage Weil int i_pin_ref; 341d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 342de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 343de57606cSSage Weil u32 i_shared_gen; /* increment each time we get FILE_SHARED */ 344cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 345de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 346de57606cSSage Weil 347de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 34868cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 349de57606cSSage Weil spinlock_t i_unsafe_lock; 350de57606cSSage Weil 351de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 352de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 353de57606cSSage Weil struct list_head i_snap_realm_item; 354de57606cSSage Weil struct list_head i_snap_flush_item; 355de57606cSSage Weil 356de57606cSSage Weil struct work_struct i_wb_work; /* writeback work */ 357de57606cSSage Weil struct work_struct i_pg_inv_work; /* page invalidation work */ 358de57606cSSage Weil 359de57606cSSage Weil struct work_struct i_vmtruncate_work; 360de57606cSSage Weil 36199ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 36299ccbd22SMilosz Tanski struct fscache_cookie *fscache; 363f7f7e7a0SYan, Zheng u32 i_fscache_gen; 36499ccbd22SMilosz Tanski #endif 365de57606cSSage Weil struct inode vfs_inode; /* at end */ 366de57606cSSage Weil }; 367de57606cSSage Weil 368de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 369de57606cSSage Weil { 370fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 371de57606cSSage Weil } 372de57606cSSage Weil 373ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 374ad1fee96SYehuda Sadeh { 375ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 376ad1fee96SYehuda Sadeh } 377ad1fee96SYehuda Sadeh 378ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 379ad1fee96SYehuda Sadeh { 380ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 381ad1fee96SYehuda Sadeh } 382ad1fee96SYehuda Sadeh 3833d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 3843d14c5d2SYehuda Sadeh { 3853d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 3863d14c5d2SYehuda Sadeh } 3873d14c5d2SYehuda Sadeh 3883d14c5d2SYehuda Sadeh /* 3893d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 3903d14c5d2SYehuda Sadeh * 391ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 392ad1fee96SYehuda Sadeh * i386 32 32 393ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 394ad1fee96SYehuda Sadeh * x86_64 64 64 395ad1fee96SYehuda Sadeh */ 3963310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 397ad1fee96SYehuda Sadeh { 3983310f754SAmon Ott u32 ino = vino & 0xffffffff; 3993310f754SAmon Ott ino ^= vino >> 32; 400ad1fee96SYehuda Sadeh if (!ino) 401a661fc56SAmon Ott ino = 2; 402ad1fee96SYehuda Sadeh return ino; 403ad1fee96SYehuda Sadeh } 404ad1fee96SYehuda Sadeh 405ad1fee96SYehuda Sadeh /* 406ad1fee96SYehuda Sadeh * kernel i_ino value 4073d14c5d2SYehuda Sadeh */ 4083d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4093d14c5d2SYehuda Sadeh { 4103d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4113310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4123310f754SAmon Ott #else 4133310f754SAmon Ott return (ino_t)vino.ino; 4143d14c5d2SYehuda Sadeh #endif 4153d14c5d2SYehuda Sadeh } 4163d14c5d2SYehuda Sadeh 417ad1fee96SYehuda Sadeh /* 418ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 419ad1fee96SYehuda Sadeh */ 420ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 421ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 422ad1fee96SYehuda Sadeh { 423ad1fee96SYehuda Sadeh return ino; 424ad1fee96SYehuda Sadeh } 425ad1fee96SYehuda Sadeh #else 426ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 427ad1fee96SYehuda Sadeh { 428ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 429ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 430ad1fee96SYehuda Sadeh return ino; 431ad1fee96SYehuda Sadeh } 432ad1fee96SYehuda Sadeh #endif 433ad1fee96SYehuda Sadeh 434ad1fee96SYehuda Sadeh 4353d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4363d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4373d14c5d2SYehuda Sadeh 4383d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4393d14c5d2SYehuda Sadeh { 4403d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4413d14c5d2SYehuda Sadeh } 4423d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4433d14c5d2SYehuda Sadeh { 4443d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4453d14c5d2SYehuda Sadeh } 4463d14c5d2SYehuda Sadeh 4473d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4483d14c5d2SYehuda Sadeh { 4493d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4503d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4513d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4523d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4533d14c5d2SYehuda Sadeh } 4543d14c5d2SYehuda Sadeh 4553d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4563d14c5d2SYehuda Sadeh struct ceph_vino vino) 4573d14c5d2SYehuda Sadeh { 4583d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 4593d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 4603d14c5d2SYehuda Sadeh } 4613d14c5d2SYehuda Sadeh 4623d14c5d2SYehuda Sadeh 4633d14c5d2SYehuda Sadeh /* 4643d14c5d2SYehuda Sadeh * Ceph inode. 4653d14c5d2SYehuda Sadeh */ 46610183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 46710183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 46810183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 46910183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 47010183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 47110183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 47210183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 473315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 47477310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 47513c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 47670220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 4773d14c5d2SYehuda Sadeh 4782f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 479fdd4e158SYan, Zheng long long release_count, 480fdd4e158SYan, Zheng long long ordered_count) 481de57606cSSage Weil { 482fdd4e158SYan, Zheng smp_mb__before_atomic(); 483fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 484fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 485de57606cSSage Weil } 486de57606cSSage Weil 4872f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 488de57606cSSage Weil { 489fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 490fdd4e158SYan, Zheng } 491fdd4e158SYan, Zheng 492fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 493fdd4e158SYan, Zheng { 494fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 495de57606cSSage Weil } 496de57606cSSage Weil 4972f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 498de57606cSSage Weil { 499fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 500fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 5012f276c51SYan, Zheng } 502de57606cSSage Weil 50370db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 50470db4f36SYan, Zheng { 505fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 506fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 507fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 508fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 50970db4f36SYan, Zheng } 51070db4f36SYan, Zheng 5112f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5122f276c51SYan, Zheng { 5132f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5142f276c51SYan, Zheng } 5152f276c51SYan, Zheng 51670db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5172f276c51SYan, Zheng { 518fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 519de57606cSSage Weil } 520de57606cSSage Weil 52170db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 52270db4f36SYan, Zheng { 523fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 524fdd4e158SYan, Zheng smp_rmb(); 52570db4f36SYan, Zheng return ret; 52670db4f36SYan, Zheng } 527de57606cSSage Weil 528de57606cSSage Weil /* find a specific frag @f */ 529de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 530de57606cSSage Weil u32 f); 531de57606cSSage Weil 532de57606cSSage Weil /* 533de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 534de57606cSSage Weil * exists 535de57606cSSage Weil */ 536de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 537de57606cSSage Weil struct ceph_inode_frag *pfrag, 538de57606cSSage Weil int *found); 539de57606cSSage Weil 540de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) 541de57606cSSage Weil { 542de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 543de57606cSSage Weil } 544de57606cSSage Weil 545de57606cSSage Weil /* 546de57606cSSage Weil * caps helpers 547de57606cSSage Weil */ 548de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 549de57606cSSage Weil { 550de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 551de57606cSSage Weil } 552de57606cSSage Weil 553de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 554de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 555de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 556de57606cSSage Weil struct ceph_cap *cap); 557de57606cSSage Weil 558de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 559de57606cSSage Weil { 560de57606cSSage Weil int issued; 561be655596SSage Weil spin_lock(&ci->i_ceph_lock); 562de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 563be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 564de57606cSSage Weil return issued; 565de57606cSSage Weil } 566de57606cSSage Weil 567de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 568de57606cSSage Weil int touch) 569de57606cSSage Weil { 570de57606cSSage Weil int r; 571be655596SSage Weil spin_lock(&ci->i_ceph_lock); 572de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 573be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 574de57606cSSage Weil return r; 575de57606cSSage Weil } 576de57606cSSage Weil 577de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 578de57606cSSage Weil { 579de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 580de57606cSSage Weil } 581f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 582f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 583f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 584f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 585de57606cSSage Weil 5869563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 5879563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 588de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 589de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 590de57606cSSage Weil 591de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 592de57606cSSage Weil 593de57606cSSage Weil /* 594de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 595de57606cSSage Weil */ 596de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 597de57606cSSage Weil { 598de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 599de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 600de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 601de57606cSSage Weil return w; 602de57606cSSage Weil } 603de57606cSSage Weil 604de57606cSSage Weil /* what the mds thinks we want */ 605c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 606de57606cSSage Weil 60737151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 60837151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 60937151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); 61093faca6eSmajianpeng extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, 61137151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 61237151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, 61337151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6143d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 615de57606cSSage Weil int *total, int *avail, int *used, 61685ccce43SSage Weil int *reserved, int *min); 617de57606cSSage Weil 618de57606cSSage Weil 619de57606cSSage Weil 620de57606cSSage Weil /* 621de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 622de57606cSSage Weil */ 6234918b6d1SSage Weil #define CEPH_F_SYNC 1 6249cfa1098SSage Weil #define CEPH_F_ATEND 2 6254918b6d1SSage Weil 626de57606cSSage Weil struct ceph_file_info { 627252c6728SSage Weil short fmode; /* initialized on open */ 628252c6728SSage Weil short flags; /* CEPH_F_* */ 629de57606cSSage Weil 630de57606cSSage Weil /* readdir: position within the dir */ 631de57606cSSage Weil u32 frag; 632de57606cSSage Weil struct ceph_mds_request *last_readdir; 633de57606cSSage Weil 634de57606cSSage Weil /* readdir: position within a frag */ 635f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 636de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 637fdd4e158SYan, Zheng long long dir_release_count; 638fdd4e158SYan, Zheng long long dir_ordered_count; 639fdd4e158SYan, Zheng int readdir_cache_idx; 640de57606cSSage Weil 641de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 642de57606cSSage Weil char *dir_info; 643de57606cSSage Weil int dir_info_len; 644de57606cSSage Weil }; 645de57606cSSage Weil 646fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 647fdd4e158SYan, Zheng struct page *page; 648fdd4e158SYan, Zheng struct dentry **dentries; 649fdd4e158SYan, Zheng int index; 650fdd4e158SYan, Zheng }; 651de57606cSSage Weil 652de57606cSSage Weil /* 653de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 654de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 655de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 656de57606cSSage Weil * the snapshots of their parents. 657de57606cSSage Weil * 658de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 659de57606cSSage Weil * per-realm list. 660de57606cSSage Weil */ 661de57606cSSage Weil struct ceph_snap_realm { 662de57606cSSage Weil u64 ino; 663de57606cSSage Weil atomic_t nref; 664a105f00cSSage Weil struct rb_node node; 665a105f00cSSage Weil 666de57606cSSage Weil u64 created, seq; 667de57606cSSage Weil u64 parent_ino; 668de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 669de57606cSSage Weil 670de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 671aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 672de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 673aa711ee3SAlex Elder u32 num_snaps; 674de57606cSSage Weil 675de57606cSSage Weil struct ceph_snap_realm *parent; 676de57606cSSage Weil struct list_head children; /* list of child realms */ 677de57606cSSage Weil struct list_head child_item; 678de57606cSSage Weil 679de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 680de57606cSSage Weil 681ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 682ae00d4f3SSage Weil 683de57606cSSage Weil /* the current set of snaps for this realm */ 684de57606cSSage Weil struct ceph_snap_context *cached_context; 685de57606cSSage Weil 686de57606cSSage Weil struct list_head inodes_with_caps; 687de57606cSSage Weil spinlock_t inodes_with_caps_lock; 688de57606cSSage Weil }; 689de57606cSSage Weil 6903d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 6913d14c5d2SYehuda Sadeh { 6923d14c5d2SYehuda Sadeh int congestion_kb; 693de57606cSSage Weil 694de57606cSSage Weil /* 6953d14c5d2SYehuda Sadeh * Copied from NFS 6963d14c5d2SYehuda Sadeh * 6973d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 6983d14c5d2SYehuda Sadeh * 6993d14c5d2SYehuda Sadeh * 64MB: 8192k 7003d14c5d2SYehuda Sadeh * 128MB: 11585k 7013d14c5d2SYehuda Sadeh * 256MB: 16384k 7023d14c5d2SYehuda Sadeh * 512MB: 23170k 7033d14c5d2SYehuda Sadeh * 1GB: 32768k 7043d14c5d2SYehuda Sadeh * 2GB: 46340k 7053d14c5d2SYehuda Sadeh * 4GB: 65536k 7063d14c5d2SYehuda Sadeh * 8GB: 92681k 7073d14c5d2SYehuda Sadeh * 16GB: 131072k 7083d14c5d2SYehuda Sadeh * 7093d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 7103d14c5d2SYehuda Sadeh * Limit the default to 256M 711de57606cSSage Weil */ 7123d14c5d2SYehuda Sadeh congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); 7133d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 7143d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 7153d14c5d2SYehuda Sadeh 7163d14c5d2SYehuda Sadeh return congestion_kb; 717de57606cSSage Weil } 718de57606cSSage Weil 719de57606cSSage Weil 720de57606cSSage Weil 721de57606cSSage Weil /* snap.c */ 722de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 723de57606cSSage Weil u64 ino); 724de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 725de57606cSSage Weil struct ceph_snap_realm *realm); 726de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 727de57606cSSage Weil struct ceph_snap_realm *realm); 728de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 729982d6011SYan, Zheng void *p, void *e, bool deletion, 730982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 731de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 7322600d2ddSSage Weil struct ceph_mds_session *session, 733de57606cSSage Weil struct ceph_msg *msg); 734fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 735de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 736de57606cSSage Weil struct ceph_cap_snap *capsnap); 737de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 738de57606cSSage Weil 739de57606cSSage Weil /* 740de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 741de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 742de57606cSSage Weil */ 743de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 744de57606cSSage Weil { 745de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 74686056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 747de57606cSSage Weil ci_item)->writing; 748de57606cSSage Weil } 749de57606cSSage Weil 750de57606cSSage Weil /* inode.c */ 751de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 752de57606cSSage Weil 753de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 754de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode); 7559f12bd11SYan, Zheng extern int ceph_drop_inode(struct inode *inode); 756de57606cSSage Weil 757de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 758de57606cSSage Weil struct ceph_vino vino); 759de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 760de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 761de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 762de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 763de57606cSSage Weil u64 time_warp_seq, struct timespec *ctime, 764de57606cSSage Weil struct timespec *mtime, struct timespec *atime); 765de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 766f5a03b08SJeff Layton struct ceph_mds_request *req); 767de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 768de57606cSSage Weil struct ceph_mds_session *session); 769de57606cSSage Weil 770de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 771de57606cSSage Weil 772de57606cSSage Weil extern int ceph_inode_set_size(struct inode *inode, loff_t size); 773b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 7743c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 7753c6f6b79SSage Weil 7763c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 7773c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 778de57606cSSage Weil 77901deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 78001deead0SYan, Zheng int mask, bool force); 78101deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 78201deead0SYan, Zheng { 78301deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 78401deead0SYan, Zheng } 78510556cb2SAl Viro extern int ceph_permission(struct inode *inode, int mask); 786a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 787de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); 788a528d35eSDavid Howells extern int ceph_getattr(const struct path *path, struct kstat *stat, 789a528d35eSDavid Howells u32 request_mask, unsigned int flags); 790de57606cSSage Weil 791de57606cSSage Weil /* xattr.c */ 792a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 7937221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 794de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 795de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); 796de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 7973ce6cd12SAlex Elder extern void __init ceph_xattr_init(void); 7983ce6cd12SAlex Elder extern void ceph_xattr_exit(void); 799b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 800de57606cSSage Weil 801315f2408SYan, Zheng #ifdef CONFIG_SECURITY 802315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 803315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 804315f2408SYan, Zheng #else 805315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 806315f2408SYan, Zheng { 807315f2408SYan, Zheng return false; 808315f2408SYan, Zheng } 809315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 810315f2408SYan, Zheng { 811315f2408SYan, Zheng return false; 812315f2408SYan, Zheng } 813315f2408SYan, Zheng #endif 814315f2408SYan, Zheng 8157221fe4cSGuangliang Zhao /* acl.c */ 816b1ee94aaSYan, Zheng struct ceph_acls_info { 817b1ee94aaSYan, Zheng void *default_acl; 818b1ee94aaSYan, Zheng void *acl; 819b1ee94aaSYan, Zheng struct ceph_pagelist *pagelist; 820b1ee94aaSYan, Zheng }; 8217221fe4cSGuangliang Zhao 8227221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 8237221fe4cSGuangliang Zhao 8247221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 82572466d0bSSage Weil int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 826b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 827b1ee94aaSYan, Zheng struct ceph_acls_info *info); 828b1ee94aaSYan, Zheng void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); 829b1ee94aaSYan, Zheng void ceph_release_acls_info(struct ceph_acls_info *info); 830c969d9bfSGuangliang Zhao 831c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 832c969d9bfSGuangliang Zhao { 833c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 834c969d9bfSGuangliang Zhao } 8357221fe4cSGuangliang Zhao 8367221fe4cSGuangliang Zhao #else 8377221fe4cSGuangliang Zhao 8387221fe4cSGuangliang Zhao #define ceph_get_acl NULL 83972466d0bSSage Weil #define ceph_set_acl NULL 8407221fe4cSGuangliang Zhao 841b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 842b1ee94aaSYan, Zheng struct ceph_acls_info *info) 8437221fe4cSGuangliang Zhao { 8447221fe4cSGuangliang Zhao return 0; 8457221fe4cSGuangliang Zhao } 846b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 847b1ee94aaSYan, Zheng struct ceph_acls_info *info) 848b1ee94aaSYan, Zheng { 849b1ee94aaSYan, Zheng } 850b1ee94aaSYan, Zheng static inline void ceph_release_acls_info(struct ceph_acls_info *info) 851b1ee94aaSYan, Zheng { 852b1ee94aaSYan, Zheng } 8537221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 8547221fe4cSGuangliang Zhao { 8557221fe4cSGuangliang Zhao return 0; 8567221fe4cSGuangliang Zhao } 8577221fe4cSGuangliang Zhao 8587221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 8597221fe4cSGuangliang Zhao { 8607221fe4cSGuangliang Zhao } 8617221fe4cSGuangliang Zhao 8627221fe4cSGuangliang Zhao #endif 8637221fe4cSGuangliang Zhao 864de57606cSSage Weil /* caps.c */ 865de57606cSSage Weil extern const char *ceph_cap_string(int c); 866de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 867de57606cSSage Weil struct ceph_msg *msg); 868d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 869d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 870d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 871de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 872de57606cSSage Weil int fmode, unsigned issued, unsigned wanted, 873de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 874d9df2783SYan, Zheng struct ceph_cap **new_cap); 875a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 87637151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 87737151668SYehuda Sadeh struct ceph_cap *cap); 8789215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 879de57606cSSage Weil 880de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode); 881f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 88202c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 88302c24a82SJosef Bacik int datasync); 884e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 885e548e9b9SYan, Zheng struct ceph_mds_session *session); 886de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 887de57606cSSage Weil struct ceph_mds_session *session); 8882bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 8892bc50259SGreg Farnum int mds); 890de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode); 891de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 892de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 893de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 894de57606cSSage Weil struct ceph_snap_context *snapc); 895ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 8960e294387SYan, Zheng struct ceph_mds_session **psession); 897de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 898de57606cSSage Weil struct ceph_mds_session *session); 899afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 900afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 901de57606cSSage Weil 902de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 903de57606cSSage Weil int mds, int drop, int unless, int force); 904de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 905ca6c8ae0SJeff Layton struct inode *dir, 906de57606cSSage Weil int mds, int drop, int unless); 907de57606cSSage Weil 908de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, 90931c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 9102b1ac852SYan, Zheng extern int ceph_try_get_caps(struct ceph_inode_info *ci, 9112b1ac852SYan, Zheng int need, int want, int *got); 912de57606cSSage Weil 913de57606cSSage Weil /* for counting open files by mode */ 914774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 915de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 916de57606cSSage Weil 917de57606cSSage Weil /* addr.c */ 918de57606cSSage Weil extern const struct address_space_operations ceph_aops; 919de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 92010183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 92110183a69SYan, Zheng extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); 92210183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 923de57606cSSage Weil 924de57606cSSage Weil /* file.c */ 925de57606cSSage Weil extern const struct file_operations ceph_file_fops; 9269e0eb85dSAlex Elder 92777310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 928de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 9295ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 9305ef50c3bSSage Weil struct file *file, unsigned flags, umode_t mode, 9315ef50c3bSSage Weil int *opened); 932de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 93331c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 93431c542a1SYan, Zheng char *data, size_t len); 93555f2a045SIlya Dryomov 936de57606cSSage Weil /* dir.c */ 937de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 93838c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 939de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 94038c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 94118fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 942de57606cSSage Weil 943f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 944de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 945468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 946468640e3SSage Weil struct dentry *dentry, int err); 947de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 948de57606cSSage Weil struct dentry *dentry, int err); 949de57606cSSage Weil 950de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn); 951de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn); 952de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn); 95381a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 954e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 955fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 956de57606cSSage Weil 957de57606cSSage Weil /* ioctl.c */ 958de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 959de57606cSSage Weil 960de57606cSSage Weil /* export.c */ 961de57606cSSage Weil extern const struct export_operations ceph_export_ops; 962de57606cSSage Weil 96340819f6fSGreg Farnum /* locks.c */ 964eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 96540819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 96640819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 96740819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 96839be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 96939be95e9SJim Schutt struct ceph_filelock *flocks, 97039be95e9SJim Schutt int num_fcntl_locks, 97139be95e9SJim Schutt int num_flock_locks); 97239be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 97339be95e9SJim Schutt struct ceph_pagelist *pagelist, 97439be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 97540819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c); 97640819f6fSGreg Farnum 9773d14c5d2SYehuda Sadeh /* debugfs.c */ 9783d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); 9793d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 9803d14c5d2SYehuda Sadeh 981de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 982