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> 17de57606cSSage Weil 183d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 19de57606cSSage Weil 2099ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 2199ccbd22SMilosz Tanski #include <linux/fscache.h> 2299ccbd22SMilosz Tanski #endif 2399ccbd22SMilosz Tanski 24de57606cSSage Weil /* f_type in struct statfs */ 25de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400 26de57606cSSage Weil 27de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 28de57606cSSage Weil * large volume sizes on 32-bit machines. */ 2992a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 30de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 31de57606cSSage Weil 323d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 333d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 343d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 35ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 36a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 3799ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 3810183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 39*e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 406a259382SSage Weil 41133e9156SYan, Zheng #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE 42de57606cSSage Weil 433d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 443d14c5d2SYehuda Sadeh (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; 453d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 463d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 47de57606cSSage Weil 4883817e35SSage Weil #define CEPH_RSIZE_DEFAULT 0 /* max read size */ 4983817e35SSage Weil #define CEPH_RASIZE_DEFAULT (8192*1024) /* readahead */ 503d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 513d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 523d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 53de57606cSSage Weil 543d14c5d2SYehuda Sadeh struct ceph_mount_options { 55de57606cSSage Weil int flags; 563d14c5d2SYehuda Sadeh int sb_flags; 573d14c5d2SYehuda Sadeh 5883817e35SSage Weil int wsize; /* max write size */ 5983817e35SSage Weil int rsize; /* max read size */ 6083817e35SSage Weil int rasize; /* max readahead */ 616e19a16eSSage Weil int congestion_kb; /* max writeback in flight */ 626e19a16eSSage Weil int caps_wanted_delay_min, caps_wanted_delay_max; 636e19a16eSSage Weil int cap_release_safety; 6423804d91SSage Weil int max_readdir; /* max readdir result (entires) */ 6523804d91SSage Weil int max_readdir_bytes; /* max readdir result (bytes) */ 663d14c5d2SYehuda Sadeh 673d14c5d2SYehuda Sadeh /* 683d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 693d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 703d14c5d2SYehuda Sadeh */ 713d14c5d2SYehuda Sadeh 72de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 73430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 743f384954SYan, Zheng char *server_path; /* default "/" */ 75de57606cSSage Weil }; 76de57606cSSage Weil 773d14c5d2SYehuda Sadeh struct ceph_fs_client { 78de57606cSSage Weil struct super_block *sb; 79de57606cSSage Weil 803d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 813d14c5d2SYehuda Sadeh struct ceph_client *client; 823d14c5d2SYehuda Sadeh 83de57606cSSage Weil unsigned long mount_state; 8485ccce43SSage Weil int min_caps; /* min caps i added */ 8585ccce43SSage Weil 863d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 87de57606cSSage Weil 88de57606cSSage Weil /* writeback */ 89de57606cSSage Weil mempool_t *wb_pagevec_pool; 90de57606cSSage Weil struct workqueue_struct *wb_wq; 91de57606cSSage Weil struct workqueue_struct *pg_inv_wq; 92de57606cSSage Weil struct workqueue_struct *trunc_wq; 932baba250SYehuda Sadeh atomic_long_t writeback_count; 94de57606cSSage Weil 95de57606cSSage Weil struct backing_dev_info backing_dev_info; 960743304dSSage 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 { 165de57606cSSage Weil atomic_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; 184de57606cSSage Weil int writing; /* a sync write is still in progress */ 185de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 186e20d258dSYan, Zheng bool inline_data; 18786056090SYan, Zheng bool need_flush; 188de57606cSSage Weil }; 189de57606cSSage Weil 190de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 191de57606cSSage Weil { 1924a625be4SSage Weil if (atomic_dec_and_test(&capsnap->nref)) { 1934a625be4SSage Weil if (capsnap->xattr_blob) 1944a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 195de57606cSSage Weil kfree(capsnap); 196de57606cSSage Weil } 1974a625be4SSage Weil } 198de57606cSSage Weil 199de57606cSSage Weil /* 200de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 201de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 202de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 203de57606cSSage Weil * 204de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 205de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 206de57606cSSage Weil */ 207de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 208de57606cSSage Weil 209de57606cSSage Weil struct ceph_inode_frag { 210de57606cSSage Weil struct rb_node node; 211de57606cSSage Weil 212de57606cSSage Weil /* fragtree state */ 213de57606cSSage Weil u32 frag; 214de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 215de57606cSSage Weil 216de57606cSSage Weil /* delegation and replication info */ 217de57606cSSage Weil int mds; /* -1 if same authority as parent */ 218de57606cSSage Weil int ndist; /* >0 if replicated */ 219de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 220de57606cSSage Weil }; 221de57606cSSage Weil 222de57606cSSage Weil /* 223de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 224de57606cSSage Weil * at which point we parse them into an rbtree. 225de57606cSSage Weil */ 226de57606cSSage Weil struct ceph_inode_xattr { 227de57606cSSage Weil struct rb_node node; 228de57606cSSage Weil 229de57606cSSage Weil const char *name; 230de57606cSSage Weil int name_len; 231de57606cSSage Weil const char *val; 232de57606cSSage Weil int val_len; 233de57606cSSage Weil int dirty; 234de57606cSSage Weil 235de57606cSSage Weil int should_free_name; 236de57606cSSage Weil int should_free_val; 237de57606cSSage Weil }; 238de57606cSSage Weil 2393d14c5d2SYehuda Sadeh /* 2403d14c5d2SYehuda Sadeh * Ceph dentry state 2413d14c5d2SYehuda Sadeh */ 2423d14c5d2SYehuda Sadeh struct ceph_dentry_info { 2433d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 2443d14c5d2SYehuda Sadeh u32 lease_gen, lease_shared_gen; 2453d14c5d2SYehuda Sadeh u32 lease_seq; 2463d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2473d14c5d2SYehuda Sadeh struct list_head lru; 2483d14c5d2SYehuda Sadeh struct dentry *dentry; 2499b16f03cSMiklos Szeredi unsigned long time; 2503d14c5d2SYehuda Sadeh u64 offset; 2513d14c5d2SYehuda Sadeh }; 2523d14c5d2SYehuda Sadeh 253de57606cSSage Weil struct ceph_inode_xattrs_info { 254de57606cSSage Weil /* 255de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 256de57606cSSage Weil * this until someone actually calls getxattr, etc. 257de57606cSSage Weil * 258de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 259de57606cSSage Weil * NULL means we don't know. 260de57606cSSage Weil */ 261de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 262de57606cSSage Weil 263de57606cSSage Weil struct rb_root index; 264de57606cSSage Weil bool dirty; 265de57606cSSage Weil int count; 266de57606cSSage Weil int names_size; 267de57606cSSage Weil int vals_size; 268de57606cSSage Weil u64 version, index_version; 269de57606cSSage Weil }; 270de57606cSSage Weil 271de57606cSSage Weil /* 272de57606cSSage Weil * Ceph inode. 273de57606cSSage Weil */ 274de57606cSSage Weil struct ceph_inode_info { 275de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 276de57606cSSage Weil 277be655596SSage Weil spinlock_t i_ceph_lock; 278be655596SSage Weil 279de57606cSSage Weil u64 i_version; 28031c542a1SYan, Zheng u64 i_inline_version; 281de57606cSSage Weil u32 i_time_warp_seq; 282de57606cSSage Weil 283de57606cSSage Weil unsigned i_ceph_flags; 284fdd4e158SYan, Zheng atomic64_t i_release_count; 285fdd4e158SYan, Zheng atomic64_t i_ordered_count; 286fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 287de57606cSSage Weil 2886c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 289de57606cSSage Weil struct ceph_file_layout i_layout; 290de57606cSSage Weil char *i_symlink; 291de57606cSSage Weil 292de57606cSSage Weil /* for dirs */ 293de57606cSSage Weil struct timespec i_rctime; 294de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 295de57606cSSage Weil u64 i_files, i_subdirs; 296de57606cSSage Weil 297de57606cSSage Weil struct rb_root i_fragtree; 2981b1bc16dSYan, Zheng int i_fragtree_nsplits; 299de57606cSSage Weil struct mutex i_fragtree_mutex; 300de57606cSSage Weil 301de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 302de57606cSSage Weil 303be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 304de57606cSSage Weil * s_mutex. */ 305de57606cSSage Weil struct rb_root i_caps; /* cap list */ 306de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 307de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 308de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 309de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 310de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 311de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 312f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 313e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 314de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 315de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 316de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 317de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 318de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 319de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3207d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3217d8cb26dSSage Weil dirty|flushing caps */ 322de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 323de57606cSSage Weil 324774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 325de57606cSSage Weil 326b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 327de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 328de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 329de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 330de57606cSSage Weil 331de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 332de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 333de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 334de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 335de57606cSSage Weil 336de57606cSSage Weil /* held references to caps */ 337de57606cSSage Weil int i_pin_ref; 338d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 339de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 340de57606cSSage Weil u32 i_shared_gen; /* increment each time we get FILE_SHARED */ 341cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 342de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 343de57606cSSage Weil 344de57606cSSage Weil struct list_head i_unsafe_writes; /* uncommitted sync writes */ 345de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 34668cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 347de57606cSSage Weil spinlock_t i_unsafe_lock; 348de57606cSSage Weil 349de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 350de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 351de57606cSSage Weil struct list_head i_snap_realm_item; 352de57606cSSage Weil struct list_head i_snap_flush_item; 353de57606cSSage Weil 354de57606cSSage Weil struct work_struct i_wb_work; /* writeback work */ 355de57606cSSage Weil struct work_struct i_pg_inv_work; /* page invalidation work */ 356de57606cSSage Weil 357de57606cSSage Weil struct work_struct i_vmtruncate_work; 358de57606cSSage Weil 35999ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 36099ccbd22SMilosz Tanski struct fscache_cookie *fscache; 361f7f7e7a0SYan, Zheng u32 i_fscache_gen; 36299ccbd22SMilosz Tanski #endif 363de57606cSSage Weil struct inode vfs_inode; /* at end */ 364de57606cSSage Weil }; 365de57606cSSage Weil 366de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 367de57606cSSage Weil { 368fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 369de57606cSSage Weil } 370de57606cSSage Weil 371ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 372ad1fee96SYehuda Sadeh { 373ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 374ad1fee96SYehuda Sadeh } 375ad1fee96SYehuda Sadeh 376ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 377ad1fee96SYehuda Sadeh { 378ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 379ad1fee96SYehuda Sadeh } 380ad1fee96SYehuda Sadeh 3813d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 3823d14c5d2SYehuda Sadeh { 3833d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 3843d14c5d2SYehuda Sadeh } 3853d14c5d2SYehuda Sadeh 3863d14c5d2SYehuda Sadeh /* 3873d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 3883d14c5d2SYehuda Sadeh * 389ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 390ad1fee96SYehuda Sadeh * i386 32 32 391ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 392ad1fee96SYehuda Sadeh * x86_64 64 64 393ad1fee96SYehuda Sadeh */ 3943310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 395ad1fee96SYehuda Sadeh { 3963310f754SAmon Ott u32 ino = vino & 0xffffffff; 3973310f754SAmon Ott ino ^= vino >> 32; 398ad1fee96SYehuda Sadeh if (!ino) 399a661fc56SAmon Ott ino = 2; 400ad1fee96SYehuda Sadeh return ino; 401ad1fee96SYehuda Sadeh } 402ad1fee96SYehuda Sadeh 403ad1fee96SYehuda Sadeh /* 404ad1fee96SYehuda Sadeh * kernel i_ino value 4053d14c5d2SYehuda Sadeh */ 4063d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4073d14c5d2SYehuda Sadeh { 4083d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4093310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4103310f754SAmon Ott #else 4113310f754SAmon Ott return (ino_t)vino.ino; 4123d14c5d2SYehuda Sadeh #endif 4133d14c5d2SYehuda Sadeh } 4143d14c5d2SYehuda Sadeh 415ad1fee96SYehuda Sadeh /* 416ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 417ad1fee96SYehuda Sadeh */ 418ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 419ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 420ad1fee96SYehuda Sadeh { 421ad1fee96SYehuda Sadeh return ino; 422ad1fee96SYehuda Sadeh } 423ad1fee96SYehuda Sadeh #else 424ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 425ad1fee96SYehuda Sadeh { 426ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 427ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 428ad1fee96SYehuda Sadeh return ino; 429ad1fee96SYehuda Sadeh } 430ad1fee96SYehuda Sadeh #endif 431ad1fee96SYehuda Sadeh 432ad1fee96SYehuda Sadeh 4333d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4343d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4353d14c5d2SYehuda Sadeh 4363d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4373d14c5d2SYehuda Sadeh { 4383d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4393d14c5d2SYehuda Sadeh } 4403d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4413d14c5d2SYehuda Sadeh { 4423d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4433d14c5d2SYehuda Sadeh } 4443d14c5d2SYehuda Sadeh 4453d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4463d14c5d2SYehuda Sadeh { 4473d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4483d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4493d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4503d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4513d14c5d2SYehuda Sadeh } 4523d14c5d2SYehuda Sadeh 4533d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4543d14c5d2SYehuda Sadeh struct ceph_vino vino) 4553d14c5d2SYehuda Sadeh { 4563d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 4573d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 4583d14c5d2SYehuda Sadeh } 4593d14c5d2SYehuda Sadeh 4603d14c5d2SYehuda Sadeh 4613d14c5d2SYehuda Sadeh /* 4623d14c5d2SYehuda Sadeh * Ceph inode. 4633d14c5d2SYehuda Sadeh */ 46410183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 46510183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 46610183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 46710183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 46810183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 46910183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 47010183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 471315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 47277310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 47313c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 47470220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 4753d14c5d2SYehuda Sadeh 4762f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 477fdd4e158SYan, Zheng long long release_count, 478fdd4e158SYan, Zheng long long ordered_count) 479de57606cSSage Weil { 480fdd4e158SYan, Zheng smp_mb__before_atomic(); 481fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 482fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 483de57606cSSage Weil } 484de57606cSSage Weil 4852f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 486de57606cSSage Weil { 487fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 488fdd4e158SYan, Zheng } 489fdd4e158SYan, Zheng 490fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 491fdd4e158SYan, Zheng { 492fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 493de57606cSSage Weil } 494de57606cSSage Weil 4952f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 496de57606cSSage Weil { 497fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 498fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 4992f276c51SYan, Zheng } 500de57606cSSage Weil 50170db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 50270db4f36SYan, Zheng { 503fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 504fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 505fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 506fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 50770db4f36SYan, Zheng } 50870db4f36SYan, Zheng 5092f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5102f276c51SYan, Zheng { 5112f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5122f276c51SYan, Zheng } 5132f276c51SYan, Zheng 51470db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5152f276c51SYan, Zheng { 516fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 517de57606cSSage Weil } 518de57606cSSage Weil 51970db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 52070db4f36SYan, Zheng { 521fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 522fdd4e158SYan, Zheng smp_rmb(); 52370db4f36SYan, Zheng return ret; 52470db4f36SYan, Zheng } 525de57606cSSage Weil 526de57606cSSage Weil /* find a specific frag @f */ 527de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 528de57606cSSage Weil u32 f); 529de57606cSSage Weil 530de57606cSSage Weil /* 531de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 532de57606cSSage Weil * exists 533de57606cSSage Weil */ 534de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 535de57606cSSage Weil struct ceph_inode_frag *pfrag, 536de57606cSSage Weil int *found); 537de57606cSSage Weil 538de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) 539de57606cSSage Weil { 540de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 541de57606cSSage Weil } 542de57606cSSage Weil 543de57606cSSage Weil /* 544de57606cSSage Weil * caps helpers 545de57606cSSage Weil */ 546de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 547de57606cSSage Weil { 548de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 549de57606cSSage Weil } 550de57606cSSage Weil 551de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 552de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 553de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 554de57606cSSage Weil struct ceph_cap *cap); 555de57606cSSage Weil 556de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 557de57606cSSage Weil { 558de57606cSSage Weil int issued; 559be655596SSage Weil spin_lock(&ci->i_ceph_lock); 560de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 561be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 562de57606cSSage Weil return issued; 563de57606cSSage Weil } 564de57606cSSage Weil 565de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 566de57606cSSage Weil int touch) 567de57606cSSage Weil { 568de57606cSSage Weil int r; 569be655596SSage Weil spin_lock(&ci->i_ceph_lock); 570de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 571be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 572de57606cSSage Weil return r; 573de57606cSSage Weil } 574de57606cSSage Weil 575de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 576de57606cSSage Weil { 577de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 578de57606cSSage Weil } 579f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 580f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 581f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 582f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 583de57606cSSage Weil 5849563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 5859563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 586de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 587de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 588de57606cSSage Weil 589de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 590de57606cSSage Weil 591de57606cSSage Weil /* 592de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 593de57606cSSage Weil */ 594de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 595de57606cSSage Weil { 596de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 597de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 598de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 599de57606cSSage Weil return w; 600de57606cSSage Weil } 601de57606cSSage Weil 602de57606cSSage Weil /* what the mds thinks we want */ 603de57606cSSage Weil extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci); 604de57606cSSage Weil 60537151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 60637151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 60737151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); 60893faca6eSmajianpeng extern void ceph_reserve_caps(struct ceph_mds_client *mdsc, 60937151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 61037151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, 61137151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6123d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 613de57606cSSage Weil int *total, int *avail, int *used, 61485ccce43SSage Weil int *reserved, int *min); 615de57606cSSage Weil 616de57606cSSage Weil 617de57606cSSage Weil 618de57606cSSage Weil /* 619de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 620de57606cSSage Weil */ 6214918b6d1SSage Weil #define CEPH_F_SYNC 1 6229cfa1098SSage Weil #define CEPH_F_ATEND 2 6234918b6d1SSage Weil 624de57606cSSage Weil struct ceph_file_info { 625252c6728SSage Weil short fmode; /* initialized on open */ 626252c6728SSage Weil short flags; /* CEPH_F_* */ 627de57606cSSage Weil 628de57606cSSage Weil /* readdir: position within the dir */ 629de57606cSSage Weil u32 frag; 630de57606cSSage Weil struct ceph_mds_request *last_readdir; 631de57606cSSage Weil 632de57606cSSage Weil /* readdir: position within a frag */ 633f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 634de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 635fdd4e158SYan, Zheng long long dir_release_count; 636fdd4e158SYan, Zheng long long dir_ordered_count; 637fdd4e158SYan, Zheng int readdir_cache_idx; 638de57606cSSage Weil 639de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 640de57606cSSage Weil char *dir_info; 641de57606cSSage Weil int dir_info_len; 642de57606cSSage Weil }; 643de57606cSSage Weil 644fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 645fdd4e158SYan, Zheng struct page *page; 646fdd4e158SYan, Zheng struct dentry **dentries; 647fdd4e158SYan, Zheng int index; 648fdd4e158SYan, Zheng }; 649de57606cSSage Weil 650de57606cSSage Weil /* 651de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 652de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 653de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 654de57606cSSage Weil * the snapshots of their parents. 655de57606cSSage Weil * 656de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 657de57606cSSage Weil * per-realm list. 658de57606cSSage Weil */ 659de57606cSSage Weil struct ceph_snap_realm { 660de57606cSSage Weil u64 ino; 661de57606cSSage Weil atomic_t nref; 662a105f00cSSage Weil struct rb_node node; 663a105f00cSSage Weil 664de57606cSSage Weil u64 created, seq; 665de57606cSSage Weil u64 parent_ino; 666de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 667de57606cSSage Weil 668de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 669aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 670de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 671aa711ee3SAlex Elder u32 num_snaps; 672de57606cSSage Weil 673de57606cSSage Weil struct ceph_snap_realm *parent; 674de57606cSSage Weil struct list_head children; /* list of child realms */ 675de57606cSSage Weil struct list_head child_item; 676de57606cSSage Weil 677de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 678de57606cSSage Weil 679ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 680ae00d4f3SSage Weil 681de57606cSSage Weil /* the current set of snaps for this realm */ 682de57606cSSage Weil struct ceph_snap_context *cached_context; 683de57606cSSage Weil 684de57606cSSage Weil struct list_head inodes_with_caps; 685de57606cSSage Weil spinlock_t inodes_with_caps_lock; 686de57606cSSage Weil }; 687de57606cSSage Weil 6883d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 6893d14c5d2SYehuda Sadeh { 6903d14c5d2SYehuda Sadeh int congestion_kb; 691de57606cSSage Weil 692de57606cSSage Weil /* 6933d14c5d2SYehuda Sadeh * Copied from NFS 6943d14c5d2SYehuda Sadeh * 6953d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 6963d14c5d2SYehuda Sadeh * 6973d14c5d2SYehuda Sadeh * 64MB: 8192k 6983d14c5d2SYehuda Sadeh * 128MB: 11585k 6993d14c5d2SYehuda Sadeh * 256MB: 16384k 7003d14c5d2SYehuda Sadeh * 512MB: 23170k 7013d14c5d2SYehuda Sadeh * 1GB: 32768k 7023d14c5d2SYehuda Sadeh * 2GB: 46340k 7033d14c5d2SYehuda Sadeh * 4GB: 65536k 7043d14c5d2SYehuda Sadeh * 8GB: 92681k 7053d14c5d2SYehuda Sadeh * 16GB: 131072k 7063d14c5d2SYehuda Sadeh * 7073d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 7083d14c5d2SYehuda Sadeh * Limit the default to 256M 709de57606cSSage Weil */ 7103d14c5d2SYehuda Sadeh congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); 7113d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 7123d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 7133d14c5d2SYehuda Sadeh 7143d14c5d2SYehuda Sadeh return congestion_kb; 715de57606cSSage Weil } 716de57606cSSage Weil 717de57606cSSage Weil 718de57606cSSage Weil 719de57606cSSage Weil /* snap.c */ 720de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 721de57606cSSage Weil u64 ino); 722de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 723de57606cSSage Weil struct ceph_snap_realm *realm); 724de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 725de57606cSSage Weil struct ceph_snap_realm *realm); 726de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 727982d6011SYan, Zheng void *p, void *e, bool deletion, 728982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 729de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 7302600d2ddSSage Weil struct ceph_mds_session *session, 731de57606cSSage Weil struct ceph_msg *msg); 732fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 733de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 734de57606cSSage Weil struct ceph_cap_snap *capsnap); 735de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 736de57606cSSage Weil 737de57606cSSage Weil /* 738de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 739de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 740de57606cSSage Weil */ 741de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 742de57606cSSage Weil { 743de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 74486056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 745de57606cSSage Weil ci_item)->writing; 746de57606cSSage Weil } 747de57606cSSage Weil 748de57606cSSage Weil /* inode.c */ 749de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 750de57606cSSage Weil 751de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 752de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode); 7539f12bd11SYan, Zheng extern int ceph_drop_inode(struct inode *inode); 7549a5530c6SYan, Zheng extern void ceph_evict_inode(struct inode *inode); 755de57606cSSage Weil 756de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 757de57606cSSage Weil struct ceph_vino vino); 758de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 759de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 760de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 761de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 762de57606cSSage Weil u64 time_warp_seq, struct timespec *ctime, 763de57606cSSage Weil struct timespec *mtime, struct timespec *atime); 764de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 765de57606cSSage Weil struct ceph_mds_request *req, 766de57606cSSage Weil struct ceph_mds_session *session); 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); 788de57606cSSage Weil extern int ceph_getattr(struct vfsmount *mnt, struct dentry *dentry, 789de57606cSSage Weil struct kstat *stat); 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, 905de57606cSSage Weil int mds, int drop, int unless); 906de57606cSSage Weil 907de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, 90831c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 9092b1ac852SYan, Zheng extern int ceph_try_get_caps(struct ceph_inode_info *ci, 9102b1ac852SYan, Zheng int need, int want, int *got); 911de57606cSSage Weil 912de57606cSSage Weil /* for counting open files by mode */ 913774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 914de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 915de57606cSSage Weil 916de57606cSSage Weil /* addr.c */ 917de57606cSSage Weil extern const struct address_space_operations ceph_aops; 918de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 91910183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 92010183a69SYan, Zheng extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); 92110183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 922de57606cSSage Weil 923de57606cSSage Weil /* file.c */ 924de57606cSSage Weil extern const struct file_operations ceph_file_fops; 9259e0eb85dSAlex Elder 92677310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 927de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 9285ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 9295ef50c3bSSage Weil struct file *file, unsigned flags, umode_t mode, 9305ef50c3bSSage Weil int *opened); 931de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 93231c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 93331c542a1SYan, Zheng char *data, size_t len); 9349a5530c6SYan, Zheng extern void ceph_sync_write_wait(struct inode *inode); 935de57606cSSage Weil /* dir.c */ 936de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 93738c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 938de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 93938c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 94052dfb8acSSage Weil extern const struct dentry_operations ceph_dentry_ops, ceph_snap_dentry_ops, 941de57606cSSage Weil ceph_snapdir_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 /* 958de57606cSSage Weil * our d_ops vary depending on whether the inode is live, 959de57606cSSage Weil * snapshotted (read-only), or a virtual ".snap" directory. 960de57606cSSage Weil */ 961de57606cSSage Weil int ceph_init_dentry(struct dentry *dentry); 962de57606cSSage Weil 963de57606cSSage Weil 964de57606cSSage Weil /* ioctl.c */ 965de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 966de57606cSSage Weil 967de57606cSSage Weil /* export.c */ 968de57606cSSage Weil extern const struct export_operations ceph_export_ops; 969de57606cSSage Weil 97040819f6fSGreg Farnum /* locks.c */ 971eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 97240819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 97340819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 97440819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 97539be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 97639be95e9SJim Schutt struct ceph_filelock *flocks, 97739be95e9SJim Schutt int num_fcntl_locks, 97839be95e9SJim Schutt int num_flock_locks); 97939be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 98039be95e9SJim Schutt struct ceph_pagelist *pagelist, 98139be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 98240819f6fSGreg Farnum extern int lock_to_ceph_filelock(struct file_lock *fl, struct ceph_filelock *c); 98340819f6fSGreg Farnum 9843d14c5d2SYehuda Sadeh /* debugfs.c */ 9853d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); 9863d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 9873d14c5d2SYehuda Sadeh 988de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 989