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> 17805692d0SElena 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 "/" */ 76*1d8f8360SYan, Zheng char *fscache_uniq; /* default NULL */ 77de57606cSSage Weil }; 78de57606cSSage Weil 793d14c5d2SYehuda Sadeh struct ceph_fs_client { 80de57606cSSage Weil struct super_block *sb; 81de57606cSSage Weil 823d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 833d14c5d2SYehuda Sadeh struct ceph_client *client; 843d14c5d2SYehuda Sadeh 85de57606cSSage Weil unsigned long mount_state; 8685ccce43SSage Weil int min_caps; /* min caps i added */ 8785ccce43SSage Weil 883d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 89de57606cSSage Weil 90de57606cSSage Weil /* writeback */ 91de57606cSSage Weil mempool_t *wb_pagevec_pool; 92de57606cSSage Weil struct workqueue_struct *wb_wq; 93de57606cSSage Weil struct workqueue_struct *pg_inv_wq; 94de57606cSSage Weil struct workqueue_struct *trunc_wq; 952baba250SYehuda Sadeh atomic_long_t writeback_count; 96de57606cSSage Weil 970743304dSSage Weil #ifdef CONFIG_DEBUG_FS 983d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 992baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 10006edf046SSage Weil struct dentry *debugfs_bdi; 1013d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 10214ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 1030743304dSSage Weil #endif 10499ccbd22SMilosz Tanski 10599ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 10699ccbd22SMilosz Tanski struct fscache_cookie *fscache; 10799ccbd22SMilosz Tanski #endif 108de57606cSSage Weil }; 109de57606cSSage Weil 1103d14c5d2SYehuda Sadeh 111de57606cSSage Weil /* 112de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 113de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 114de57606cSSage Weil * and write data. For any given inode, we should have one or more 115de57606cSSage Weil * capabilities, one issued by each metadata server, and our 116de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 117de57606cSSage Weil * 118de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 119de57606cSSage Weil * session capability lists. 120de57606cSSage Weil */ 121de57606cSSage Weil struct ceph_cap { 122de57606cSSage Weil struct ceph_inode_info *ci; 123de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 124de57606cSSage Weil struct ceph_mds_session *session; 125de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 126de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 127745a8e3bSYan, Zheng union { 128745a8e3bSYan, Zheng /* in-use caps */ 129745a8e3bSYan, Zheng struct { 130de57606cSSage Weil int issued; /* latest, from the mds */ 131745a8e3bSYan, Zheng int implemented; /* implemented superset of 132745a8e3bSYan, Zheng issued (for revocation) */ 133745a8e3bSYan, Zheng int mds, mds_wanted; 134745a8e3bSYan, Zheng }; 135745a8e3bSYan, Zheng /* caps to release */ 136745a8e3bSYan, Zheng struct { 137745a8e3bSYan, Zheng u64 cap_ino; 138745a8e3bSYan, Zheng int queue_release; 139745a8e3bSYan, Zheng }; 140745a8e3bSYan, Zheng }; 141685f9a5dSSage Weil u32 seq, issue_seq, mseq; 142685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 143de57606cSSage Weil unsigned long last_used; 144de57606cSSage Weil struct list_head caps_item; 145de57606cSSage Weil }; 146de57606cSSage Weil 147de57606cSSage Weil #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ 148de57606cSSage Weil #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ 149de57606cSSage Weil #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ 150de57606cSSage Weil 1510e294387SYan, Zheng struct ceph_cap_flush { 1520e294387SYan, Zheng u64 tid; 1530e294387SYan, Zheng int caps; /* 0 means capsnap */ 154c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 1550e294387SYan, Zheng struct list_head g_list; // global 1560e294387SYan, Zheng struct list_head i_list; // per inode 1570e294387SYan, Zheng }; 1580e294387SYan, Zheng 159de57606cSSage Weil /* 160de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 161de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 162de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 163de57606cSSage Weil */ 164de57606cSSage Weil struct ceph_cap_snap { 165805692d0SElena Reshetova refcount_t nref; 1660e294387SYan, Zheng struct list_head ci_item; 167de57606cSSage Weil 1680e294387SYan, Zheng struct ceph_cap_flush cap_flush; 1690e294387SYan, Zheng 1700e294387SYan, Zheng u64 follows; 171de57606cSSage Weil int issued, dirty; 172de57606cSSage Weil struct ceph_snap_context *context; 173de57606cSSage Weil 1745706b27dSAl Viro umode_t mode; 17505cb11c1SEric W. Biederman kuid_t uid; 17605cb11c1SEric W. Biederman kgid_t gid; 177de57606cSSage Weil 1784a625be4SSage Weil struct ceph_buffer *xattr_blob; 179de57606cSSage Weil u64 xattr_version; 180de57606cSSage Weil 181de57606cSSage Weil u64 size; 182de57606cSSage Weil struct timespec mtime, atime, ctime; 183de57606cSSage Weil u64 time_warp_seq; 1845f743e45SYan, Zheng u64 truncate_size; 1855f743e45SYan, Zheng u32 truncate_seq; 186de57606cSSage Weil int writing; /* a sync write is still in progress */ 187de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 188e20d258dSYan, Zheng bool inline_data; 18986056090SYan, Zheng bool need_flush; 190de57606cSSage Weil }; 191de57606cSSage Weil 192de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 193de57606cSSage Weil { 194805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 1954a625be4SSage Weil if (capsnap->xattr_blob) 1964a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 197de57606cSSage Weil kfree(capsnap); 198de57606cSSage Weil } 1994a625be4SSage Weil } 200de57606cSSage Weil 201de57606cSSage Weil /* 202de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 203de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 204de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 205de57606cSSage Weil * 206de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 207de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 208de57606cSSage Weil */ 209de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 210de57606cSSage Weil 211de57606cSSage Weil struct ceph_inode_frag { 212de57606cSSage Weil struct rb_node node; 213de57606cSSage Weil 214de57606cSSage Weil /* fragtree state */ 215de57606cSSage Weil u32 frag; 216de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 217de57606cSSage Weil 218de57606cSSage Weil /* delegation and replication info */ 219de57606cSSage Weil int mds; /* -1 if same authority as parent */ 220de57606cSSage Weil int ndist; /* >0 if replicated */ 221de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 222de57606cSSage Weil }; 223de57606cSSage Weil 224de57606cSSage Weil /* 225de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 226de57606cSSage Weil * at which point we parse them into an rbtree. 227de57606cSSage Weil */ 228de57606cSSage Weil struct ceph_inode_xattr { 229de57606cSSage Weil struct rb_node node; 230de57606cSSage Weil 231de57606cSSage Weil const char *name; 232de57606cSSage Weil int name_len; 233de57606cSSage Weil const char *val; 234de57606cSSage Weil int val_len; 235de57606cSSage Weil int dirty; 236de57606cSSage Weil 237de57606cSSage Weil int should_free_name; 238de57606cSSage Weil int should_free_val; 239de57606cSSage Weil }; 240de57606cSSage Weil 2413d14c5d2SYehuda Sadeh /* 2423d14c5d2SYehuda Sadeh * Ceph dentry state 2433d14c5d2SYehuda Sadeh */ 2443d14c5d2SYehuda Sadeh struct ceph_dentry_info { 2453d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 2463d14c5d2SYehuda Sadeh u32 lease_gen, lease_shared_gen; 2473d14c5d2SYehuda Sadeh u32 lease_seq; 2483d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2493d14c5d2SYehuda Sadeh struct list_head lru; 2503d14c5d2SYehuda Sadeh struct dentry *dentry; 2519b16f03cSMiklos Szeredi unsigned long time; 2523d14c5d2SYehuda Sadeh u64 offset; 2533d14c5d2SYehuda Sadeh }; 2543d14c5d2SYehuda Sadeh 255de57606cSSage Weil struct ceph_inode_xattrs_info { 256de57606cSSage Weil /* 257de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 258de57606cSSage Weil * this until someone actually calls getxattr, etc. 259de57606cSSage Weil * 260de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 261de57606cSSage Weil * NULL means we don't know. 262de57606cSSage Weil */ 263de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 264de57606cSSage Weil 265de57606cSSage Weil struct rb_root index; 266de57606cSSage Weil bool dirty; 267de57606cSSage Weil int count; 268de57606cSSage Weil int names_size; 269de57606cSSage Weil int vals_size; 270de57606cSSage Weil u64 version, index_version; 271de57606cSSage Weil }; 272de57606cSSage Weil 273de57606cSSage Weil /* 274de57606cSSage Weil * Ceph inode. 275de57606cSSage Weil */ 276de57606cSSage Weil struct ceph_inode_info { 277de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 278de57606cSSage Weil 279be655596SSage Weil spinlock_t i_ceph_lock; 280be655596SSage Weil 281de57606cSSage Weil u64 i_version; 28231c542a1SYan, Zheng u64 i_inline_version; 283de57606cSSage Weil u32 i_time_warp_seq; 284de57606cSSage Weil 285de57606cSSage Weil unsigned i_ceph_flags; 286fdd4e158SYan, Zheng atomic64_t i_release_count; 287fdd4e158SYan, Zheng atomic64_t i_ordered_count; 288fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 289de57606cSSage Weil 2906c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 291de57606cSSage Weil struct ceph_file_layout i_layout; 292de57606cSSage Weil char *i_symlink; 293de57606cSSage Weil 294de57606cSSage Weil /* for dirs */ 295de57606cSSage Weil struct timespec i_rctime; 296de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 297de57606cSSage Weil u64 i_files, i_subdirs; 298de57606cSSage Weil 299de57606cSSage Weil struct rb_root i_fragtree; 3001b1bc16dSYan, Zheng int i_fragtree_nsplits; 301de57606cSSage Weil struct mutex i_fragtree_mutex; 302de57606cSSage Weil 303de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 304de57606cSSage Weil 305be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 306de57606cSSage Weil * s_mutex. */ 307de57606cSSage Weil struct rb_root i_caps; /* cap list */ 308de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 309de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 310de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 311de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 312de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 313de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 314f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 315e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 316de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 317de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 318de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 319de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 320de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 321de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3227d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3237d8cb26dSSage Weil dirty|flushing caps */ 324de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 325de57606cSSage Weil 326774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 327de57606cSSage Weil 328b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 329de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 330de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 331de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 332de57606cSSage Weil 333de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 334de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 335de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 336de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 337de57606cSSage Weil 338de57606cSSage Weil /* held references to caps */ 339de57606cSSage Weil int i_pin_ref; 340d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 341de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 342de57606cSSage Weil u32 i_shared_gen; /* increment each time we get FILE_SHARED */ 343cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 344de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 345de57606cSSage Weil 346de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 34768cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 348de57606cSSage Weil spinlock_t i_unsafe_lock; 349de57606cSSage Weil 350de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 351de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 352de57606cSSage Weil struct list_head i_snap_realm_item; 353de57606cSSage Weil struct list_head i_snap_flush_item; 354de57606cSSage Weil 355de57606cSSage Weil struct work_struct i_wb_work; /* writeback work */ 356de57606cSSage Weil struct work_struct i_pg_inv_work; /* page invalidation work */ 357de57606cSSage Weil 358de57606cSSage Weil struct work_struct i_vmtruncate_work; 359de57606cSSage Weil 36099ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 36199ccbd22SMilosz Tanski struct fscache_cookie *fscache; 362f7f7e7a0SYan, Zheng u32 i_fscache_gen; 36399ccbd22SMilosz Tanski #endif 364de57606cSSage Weil struct inode vfs_inode; /* at end */ 365de57606cSSage Weil }; 366de57606cSSage Weil 367de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 368de57606cSSage Weil { 369fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 370de57606cSSage Weil } 371de57606cSSage Weil 372ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 373ad1fee96SYehuda Sadeh { 374ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 375ad1fee96SYehuda Sadeh } 376ad1fee96SYehuda Sadeh 377ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 378ad1fee96SYehuda Sadeh { 379ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 380ad1fee96SYehuda Sadeh } 381ad1fee96SYehuda Sadeh 3823d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 3833d14c5d2SYehuda Sadeh { 3843d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 3853d14c5d2SYehuda Sadeh } 3863d14c5d2SYehuda Sadeh 3873d14c5d2SYehuda Sadeh /* 3883d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 3893d14c5d2SYehuda Sadeh * 390ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 391ad1fee96SYehuda Sadeh * i386 32 32 392ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 393ad1fee96SYehuda Sadeh * x86_64 64 64 394ad1fee96SYehuda Sadeh */ 3953310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 396ad1fee96SYehuda Sadeh { 3973310f754SAmon Ott u32 ino = vino & 0xffffffff; 3983310f754SAmon Ott ino ^= vino >> 32; 399ad1fee96SYehuda Sadeh if (!ino) 400a661fc56SAmon Ott ino = 2; 401ad1fee96SYehuda Sadeh return ino; 402ad1fee96SYehuda Sadeh } 403ad1fee96SYehuda Sadeh 404ad1fee96SYehuda Sadeh /* 405ad1fee96SYehuda Sadeh * kernel i_ino value 4063d14c5d2SYehuda Sadeh */ 4073d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4083d14c5d2SYehuda Sadeh { 4093d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4103310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4113310f754SAmon Ott #else 4123310f754SAmon Ott return (ino_t)vino.ino; 4133d14c5d2SYehuda Sadeh #endif 4143d14c5d2SYehuda Sadeh } 4153d14c5d2SYehuda Sadeh 416ad1fee96SYehuda Sadeh /* 417ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 418ad1fee96SYehuda Sadeh */ 419ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 420ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 421ad1fee96SYehuda Sadeh { 422ad1fee96SYehuda Sadeh return ino; 423ad1fee96SYehuda Sadeh } 424ad1fee96SYehuda Sadeh #else 425ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 426ad1fee96SYehuda Sadeh { 427ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 428ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 429ad1fee96SYehuda Sadeh return ino; 430ad1fee96SYehuda Sadeh } 431ad1fee96SYehuda Sadeh #endif 432ad1fee96SYehuda Sadeh 433ad1fee96SYehuda Sadeh 4343d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4353d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4363d14c5d2SYehuda Sadeh 4373d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4383d14c5d2SYehuda Sadeh { 4393d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4403d14c5d2SYehuda Sadeh } 4413d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4423d14c5d2SYehuda Sadeh { 4433d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4443d14c5d2SYehuda Sadeh } 4453d14c5d2SYehuda Sadeh 4463d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4473d14c5d2SYehuda Sadeh { 4483d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4493d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4503d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4513d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4523d14c5d2SYehuda Sadeh } 4533d14c5d2SYehuda Sadeh 4543d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4553d14c5d2SYehuda Sadeh struct ceph_vino vino) 4563d14c5d2SYehuda Sadeh { 4573d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 4583d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 4593d14c5d2SYehuda Sadeh } 4603d14c5d2SYehuda Sadeh 4613d14c5d2SYehuda Sadeh 4623d14c5d2SYehuda Sadeh /* 4633d14c5d2SYehuda Sadeh * Ceph inode. 4643d14c5d2SYehuda Sadeh */ 46510183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 46610183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 46710183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 46810183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 46910183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 47010183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 47110183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 472315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 47377310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 47413c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 47570220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 47626544c62SJeff Layton #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ 47726544c62SJeff Layton 47826544c62SJeff Layton /* 47926544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 48026544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 48126544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 48226544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 48326544c62SJeff Layton * worried if a few slip through in either direction. 48426544c62SJeff Layton */ 48526544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 48626544c62SJeff Layton { 48726544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 48826544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 48926544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 49026544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 49126544c62SJeff Layton } 49226544c62SJeff Layton } 49326544c62SJeff Layton 49426544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 49526544c62SJeff Layton { 49626544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 49726544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 49826544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 49926544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 50026544c62SJeff Layton } 50126544c62SJeff Layton } 5023d14c5d2SYehuda Sadeh 5032f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 504fdd4e158SYan, Zheng long long release_count, 505fdd4e158SYan, Zheng long long ordered_count) 506de57606cSSage Weil { 507fdd4e158SYan, Zheng smp_mb__before_atomic(); 508fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 509fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 510de57606cSSage Weil } 511de57606cSSage Weil 5122f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 513de57606cSSage Weil { 514fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 515fdd4e158SYan, Zheng } 516fdd4e158SYan, Zheng 517fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 518fdd4e158SYan, Zheng { 519fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 520de57606cSSage Weil } 521de57606cSSage Weil 5222f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 523de57606cSSage Weil { 524fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 525fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 5262f276c51SYan, Zheng } 527de57606cSSage Weil 52870db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 52970db4f36SYan, Zheng { 530fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 531fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 532fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 533fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 53470db4f36SYan, Zheng } 53570db4f36SYan, Zheng 5362f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5372f276c51SYan, Zheng { 5382f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5392f276c51SYan, Zheng } 5402f276c51SYan, Zheng 54170db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5422f276c51SYan, Zheng { 543fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 544de57606cSSage Weil } 545de57606cSSage Weil 54670db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 54770db4f36SYan, Zheng { 548fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 549fdd4e158SYan, Zheng smp_rmb(); 55070db4f36SYan, Zheng return ret; 55170db4f36SYan, Zheng } 552de57606cSSage Weil 553de57606cSSage Weil /* find a specific frag @f */ 554de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 555de57606cSSage Weil u32 f); 556de57606cSSage Weil 557de57606cSSage Weil /* 558de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 559de57606cSSage Weil * exists 560de57606cSSage Weil */ 561de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 562de57606cSSage Weil struct ceph_inode_frag *pfrag, 563de57606cSSage Weil int *found); 564de57606cSSage Weil 565de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) 566de57606cSSage Weil { 567de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 568de57606cSSage Weil } 569de57606cSSage Weil 570de57606cSSage Weil /* 571de57606cSSage Weil * caps helpers 572de57606cSSage Weil */ 573de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 574de57606cSSage Weil { 575de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 576de57606cSSage Weil } 577de57606cSSage Weil 578de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 579de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 580de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 581de57606cSSage Weil struct ceph_cap *cap); 582de57606cSSage Weil 583de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 584de57606cSSage Weil { 585de57606cSSage Weil int issued; 586be655596SSage Weil spin_lock(&ci->i_ceph_lock); 587de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 588be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 589de57606cSSage Weil return issued; 590de57606cSSage Weil } 591de57606cSSage Weil 592de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 593de57606cSSage Weil int touch) 594de57606cSSage Weil { 595de57606cSSage Weil int r; 596be655596SSage Weil spin_lock(&ci->i_ceph_lock); 597de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 598be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 599de57606cSSage Weil return r; 600de57606cSSage Weil } 601de57606cSSage Weil 602de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 603de57606cSSage Weil { 604de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 605de57606cSSage Weil } 606f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 607f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 608f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 609f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 610de57606cSSage Weil 6119563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 6129563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 613de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 614de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 615de57606cSSage Weil 616de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 617de57606cSSage Weil 618de57606cSSage Weil /* 619de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 620de57606cSSage Weil */ 621de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 622de57606cSSage Weil { 623de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 624de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 625de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 626de57606cSSage Weil return w; 627de57606cSSage Weil } 628de57606cSSage Weil 629de57606cSSage Weil /* what the mds thinks we want */ 630c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 631de57606cSSage Weil 63237151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 63337151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 63437151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); 63593faca6eSmajianpeng extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, 63637151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 63737151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, 63837151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6393d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 640de57606cSSage Weil int *total, int *avail, int *used, 64185ccce43SSage Weil int *reserved, int *min); 642de57606cSSage Weil 643de57606cSSage Weil 644de57606cSSage Weil 645de57606cSSage Weil /* 646de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 647de57606cSSage Weil */ 6484918b6d1SSage Weil #define CEPH_F_SYNC 1 6499cfa1098SSage Weil #define CEPH_F_ATEND 2 6504918b6d1SSage Weil 651de57606cSSage Weil struct ceph_file_info { 652252c6728SSage Weil short fmode; /* initialized on open */ 653252c6728SSage Weil short flags; /* CEPH_F_* */ 654de57606cSSage Weil 655de57606cSSage Weil /* readdir: position within the dir */ 656de57606cSSage Weil u32 frag; 657de57606cSSage Weil struct ceph_mds_request *last_readdir; 658de57606cSSage Weil 659de57606cSSage Weil /* readdir: position within a frag */ 660f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 661de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 662fdd4e158SYan, Zheng long long dir_release_count; 663fdd4e158SYan, Zheng long long dir_ordered_count; 664fdd4e158SYan, Zheng int readdir_cache_idx; 665de57606cSSage Weil 666de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 667de57606cSSage Weil char *dir_info; 668de57606cSSage Weil int dir_info_len; 669de57606cSSage Weil }; 670de57606cSSage Weil 671fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 672fdd4e158SYan, Zheng struct page *page; 673fdd4e158SYan, Zheng struct dentry **dentries; 674fdd4e158SYan, Zheng int index; 675fdd4e158SYan, Zheng }; 676de57606cSSage Weil 677de57606cSSage Weil /* 678de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 679de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 680de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 681de57606cSSage Weil * the snapshots of their parents. 682de57606cSSage Weil * 683de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 684de57606cSSage Weil * per-realm list. 685de57606cSSage Weil */ 686de57606cSSage Weil struct ceph_snap_realm { 687de57606cSSage Weil u64 ino; 688de57606cSSage Weil atomic_t nref; 689a105f00cSSage Weil struct rb_node node; 690a105f00cSSage Weil 691de57606cSSage Weil u64 created, seq; 692de57606cSSage Weil u64 parent_ino; 693de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 694de57606cSSage Weil 695de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 696aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 697de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 698aa711ee3SAlex Elder u32 num_snaps; 699de57606cSSage Weil 700de57606cSSage Weil struct ceph_snap_realm *parent; 701de57606cSSage Weil struct list_head children; /* list of child realms */ 702de57606cSSage Weil struct list_head child_item; 703de57606cSSage Weil 704de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 705de57606cSSage Weil 706ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 707ae00d4f3SSage Weil 708de57606cSSage Weil /* the current set of snaps for this realm */ 709de57606cSSage Weil struct ceph_snap_context *cached_context; 710de57606cSSage Weil 711de57606cSSage Weil struct list_head inodes_with_caps; 712de57606cSSage Weil spinlock_t inodes_with_caps_lock; 713de57606cSSage Weil }; 714de57606cSSage Weil 7153d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 7163d14c5d2SYehuda Sadeh { 7173d14c5d2SYehuda Sadeh int congestion_kb; 718de57606cSSage Weil 719de57606cSSage Weil /* 7203d14c5d2SYehuda Sadeh * Copied from NFS 7213d14c5d2SYehuda Sadeh * 7223d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 7233d14c5d2SYehuda Sadeh * 7243d14c5d2SYehuda Sadeh * 64MB: 8192k 7253d14c5d2SYehuda Sadeh * 128MB: 11585k 7263d14c5d2SYehuda Sadeh * 256MB: 16384k 7273d14c5d2SYehuda Sadeh * 512MB: 23170k 7283d14c5d2SYehuda Sadeh * 1GB: 32768k 7293d14c5d2SYehuda Sadeh * 2GB: 46340k 7303d14c5d2SYehuda Sadeh * 4GB: 65536k 7313d14c5d2SYehuda Sadeh * 8GB: 92681k 7323d14c5d2SYehuda Sadeh * 16GB: 131072k 7333d14c5d2SYehuda Sadeh * 7343d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 7353d14c5d2SYehuda Sadeh * Limit the default to 256M 736de57606cSSage Weil */ 7373d14c5d2SYehuda Sadeh congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); 7383d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 7393d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 7403d14c5d2SYehuda Sadeh 7413d14c5d2SYehuda Sadeh return congestion_kb; 742de57606cSSage Weil } 743de57606cSSage Weil 744de57606cSSage Weil 745de57606cSSage Weil 746de57606cSSage Weil /* snap.c */ 747de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 748de57606cSSage Weil u64 ino); 749de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 750de57606cSSage Weil struct ceph_snap_realm *realm); 751de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 752de57606cSSage Weil struct ceph_snap_realm *realm); 753de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 754982d6011SYan, Zheng void *p, void *e, bool deletion, 755982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 756de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 7572600d2ddSSage Weil struct ceph_mds_session *session, 758de57606cSSage Weil struct ceph_msg *msg); 759fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 760de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 761de57606cSSage Weil struct ceph_cap_snap *capsnap); 762de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 763de57606cSSage Weil 764de57606cSSage Weil /* 765de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 766de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 767de57606cSSage Weil */ 768de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 769de57606cSSage Weil { 770de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 77186056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 772de57606cSSage Weil ci_item)->writing; 773de57606cSSage Weil } 774de57606cSSage Weil 775de57606cSSage Weil /* inode.c */ 776de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 777de57606cSSage Weil 778de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 779de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode); 7809f12bd11SYan, Zheng extern int ceph_drop_inode(struct inode *inode); 781de57606cSSage Weil 782de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 783de57606cSSage Weil struct ceph_vino vino); 784de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 785de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 786de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 787de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 788de57606cSSage Weil u64 time_warp_seq, struct timespec *ctime, 789de57606cSSage Weil struct timespec *mtime, struct timespec *atime); 790de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 791f5a03b08SJeff Layton struct ceph_mds_request *req); 792de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 793de57606cSSage Weil struct ceph_mds_session *session); 794de57606cSSage Weil 795de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 796de57606cSSage Weil 797efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 798b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 7993c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 8003c6f6b79SSage Weil 8013c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 8023c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 803de57606cSSage Weil 80401deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 80501deead0SYan, Zheng int mask, bool force); 80601deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 80701deead0SYan, Zheng { 80801deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 80901deead0SYan, Zheng } 81010556cb2SAl Viro extern int ceph_permission(struct inode *inode, int mask); 811a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 812de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); 813a528d35eSDavid Howells extern int ceph_getattr(const struct path *path, struct kstat *stat, 814a528d35eSDavid Howells u32 request_mask, unsigned int flags); 815de57606cSSage Weil 816de57606cSSage Weil /* xattr.c */ 817a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 8187221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 819de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 820de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); 821de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 8223ce6cd12SAlex Elder extern void __init ceph_xattr_init(void); 8233ce6cd12SAlex Elder extern void ceph_xattr_exit(void); 824b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 825de57606cSSage Weil 826315f2408SYan, Zheng #ifdef CONFIG_SECURITY 827315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 828315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 829315f2408SYan, Zheng #else 830315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 831315f2408SYan, Zheng { 832315f2408SYan, Zheng return false; 833315f2408SYan, Zheng } 834315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 835315f2408SYan, Zheng { 836315f2408SYan, Zheng return false; 837315f2408SYan, Zheng } 838315f2408SYan, Zheng #endif 839315f2408SYan, Zheng 8407221fe4cSGuangliang Zhao /* acl.c */ 841b1ee94aaSYan, Zheng struct ceph_acls_info { 842b1ee94aaSYan, Zheng void *default_acl; 843b1ee94aaSYan, Zheng void *acl; 844b1ee94aaSYan, Zheng struct ceph_pagelist *pagelist; 845b1ee94aaSYan, Zheng }; 8467221fe4cSGuangliang Zhao 8477221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 8487221fe4cSGuangliang Zhao 8497221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 85072466d0bSSage Weil int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 851b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 852b1ee94aaSYan, Zheng struct ceph_acls_info *info); 853b1ee94aaSYan, Zheng void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); 854b1ee94aaSYan, Zheng void ceph_release_acls_info(struct ceph_acls_info *info); 855c969d9bfSGuangliang Zhao 856c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 857c969d9bfSGuangliang Zhao { 858c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 859c969d9bfSGuangliang Zhao } 8607221fe4cSGuangliang Zhao 8617221fe4cSGuangliang Zhao #else 8627221fe4cSGuangliang Zhao 8637221fe4cSGuangliang Zhao #define ceph_get_acl NULL 86472466d0bSSage Weil #define ceph_set_acl NULL 8657221fe4cSGuangliang Zhao 866b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 867b1ee94aaSYan, Zheng struct ceph_acls_info *info) 8687221fe4cSGuangliang Zhao { 8697221fe4cSGuangliang Zhao return 0; 8707221fe4cSGuangliang Zhao } 871b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 872b1ee94aaSYan, Zheng struct ceph_acls_info *info) 873b1ee94aaSYan, Zheng { 874b1ee94aaSYan, Zheng } 875b1ee94aaSYan, Zheng static inline void ceph_release_acls_info(struct ceph_acls_info *info) 876b1ee94aaSYan, Zheng { 877b1ee94aaSYan, Zheng } 8787221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 8797221fe4cSGuangliang Zhao { 8807221fe4cSGuangliang Zhao return 0; 8817221fe4cSGuangliang Zhao } 8827221fe4cSGuangliang Zhao 8837221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 8847221fe4cSGuangliang Zhao { 8857221fe4cSGuangliang Zhao } 8867221fe4cSGuangliang Zhao 8877221fe4cSGuangliang Zhao #endif 8887221fe4cSGuangliang Zhao 889de57606cSSage Weil /* caps.c */ 890de57606cSSage Weil extern const char *ceph_cap_string(int c); 891de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 892de57606cSSage Weil struct ceph_msg *msg); 893d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 894d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 895d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 896de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 897de57606cSSage Weil int fmode, unsigned issued, unsigned wanted, 898de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 899d9df2783SYan, Zheng struct ceph_cap **new_cap); 900a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 90137151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 90237151668SYehuda Sadeh struct ceph_cap *cap); 9039215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 904de57606cSSage Weil 905de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode); 906f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 90702c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 90802c24a82SJosef Bacik int datasync); 909e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 910e548e9b9SYan, Zheng struct ceph_mds_session *session); 911de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 912de57606cSSage Weil struct ceph_mds_session *session); 9132bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 9142bc50259SGreg Farnum int mds); 915de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode); 916de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 917de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 918de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 919de57606cSSage Weil struct ceph_snap_context *snapc); 920ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 9210e294387SYan, Zheng struct ceph_mds_session **psession); 922efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 923de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 924de57606cSSage Weil struct ceph_mds_session *session); 925afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 926afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 927de57606cSSage Weil 928de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 929de57606cSSage Weil int mds, int drop, int unless, int force); 930de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 931ca6c8ae0SJeff Layton struct inode *dir, 932de57606cSSage Weil int mds, int drop, int unless); 933de57606cSSage Weil 934de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, 93531c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 9362b1ac852SYan, Zheng extern int ceph_try_get_caps(struct ceph_inode_info *ci, 9372b1ac852SYan, Zheng int need, int want, int *got); 938de57606cSSage Weil 939de57606cSSage Weil /* for counting open files by mode */ 940774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 941de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 942de57606cSSage Weil 943de57606cSSage Weil /* addr.c */ 944de57606cSSage Weil extern const struct address_space_operations ceph_aops; 945de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 94610183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 94710183a69SYan, Zheng extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); 94810183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 949de57606cSSage Weil 950de57606cSSage Weil /* file.c */ 951de57606cSSage Weil extern const struct file_operations ceph_file_fops; 9529e0eb85dSAlex Elder 95377310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 954de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 9555ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 9565ef50c3bSSage Weil struct file *file, unsigned flags, umode_t mode, 9575ef50c3bSSage Weil int *opened); 958de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 95931c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 96031c542a1SYan, Zheng char *data, size_t len); 96155f2a045SIlya Dryomov 962de57606cSSage Weil /* dir.c */ 963de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 96438c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 965de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 96638c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 96718fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 968de57606cSSage Weil 969f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 970de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 971468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 972468640e3SSage Weil struct dentry *dentry, int err); 973de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 974de57606cSSage Weil struct dentry *dentry, int err); 975de57606cSSage Weil 976de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn); 977de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn); 978de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn); 97981a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 980e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 981fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 982de57606cSSage Weil 983de57606cSSage Weil /* ioctl.c */ 984de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 985de57606cSSage Weil 986de57606cSSage Weil /* export.c */ 987de57606cSSage Weil extern const struct export_operations ceph_export_ops; 988de57606cSSage Weil 98940819f6fSGreg Farnum /* locks.c */ 990eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 99140819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 99240819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 99340819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 99439be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 99539be95e9SJim Schutt struct ceph_filelock *flocks, 99639be95e9SJim Schutt int num_fcntl_locks, 99739be95e9SJim Schutt int num_flock_locks); 99839be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 99939be95e9SJim Schutt struct ceph_pagelist *pagelist, 100039be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 100140819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c); 100240819f6fSGreg Farnum 10033d14c5d2SYehuda Sadeh /* debugfs.c */ 10043d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); 10053d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 10063d14c5d2SYehuda Sadeh 1007de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1008