1b2441318SGreg Kroah-Hartman /* SPDX-License-Identifier: GPL-2.0 */ 2de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H 3de57606cSSage Weil #define _FS_CEPH_SUPER_H 4de57606cSSage Weil 53d14c5d2SYehuda Sadeh #include <linux/ceph/ceph_debug.h> 6de57606cSSage Weil 7de57606cSSage Weil #include <asm/unaligned.h> 8de57606cSSage Weil #include <linux/backing-dev.h> 9de57606cSSage Weil #include <linux/completion.h> 10de57606cSSage Weil #include <linux/exportfs.h> 11de57606cSSage Weil #include <linux/fs.h> 12de57606cSSage Weil #include <linux/mempool.h> 13de57606cSSage Weil #include <linux/pagemap.h> 14de57606cSSage Weil #include <linux/wait.h> 15f1a3d572SStephen Rothwell #include <linux/writeback.h> 165a0e3ad6STejun Heo #include <linux/slab.h> 17c969d9bfSGuangliang Zhao #include <linux/posix_acl.h> 18805692d0SElena Reshetova #include <linux/refcount.h> 19de57606cSSage Weil 203d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 21de57606cSSage Weil 2299ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 2399ccbd22SMilosz Tanski #include <linux/fscache.h> 2499ccbd22SMilosz Tanski #endif 2599ccbd22SMilosz Tanski 26de57606cSSage Weil /* f_type in struct statfs */ 27de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400 28de57606cSSage Weil 29de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 30de57606cSSage Weil * large volume sizes on 32-bit machines. */ 3192a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 32de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 33de57606cSSage Weil 343d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 353d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 363d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 37ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 38a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 3999ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 4010183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 41e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 426a259382SSage Weil 43133e9156SYan, Zheng #define CEPH_MOUNT_OPT_DEFAULT CEPH_MOUNT_OPT_DCACHE 44de57606cSSage Weil 453d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 463d14c5d2SYehuda Sadeh (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; 473d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 483d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 49de57606cSSage Weil 50aa187926SYan, Zheng /* max size of osd read request, limited by libceph */ 51aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN 5295cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size. 5395cca2b4SYan, Zheng * CEPH_MSG_MAX_DATA_LEN should be small enough. */ 5495cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN 557c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 563d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 573d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 583d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 59de57606cSSage Weil 604214fb15SYan, Zheng /* 614214fb15SYan, Zheng * Delay telling the MDS we no longer want caps, in case we reopen 624214fb15SYan, Zheng * the file. Delay a minimum amount of time, even if we send a cap 634214fb15SYan, Zheng * message for some other reason. Otherwise, take the oppotunity to 644214fb15SYan, Zheng * update the mds to avoid sending another message later. 654214fb15SYan, Zheng */ 664214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ 674214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ 684214fb15SYan, Zheng 693d14c5d2SYehuda Sadeh struct ceph_mount_options { 70de57606cSSage Weil int flags; 713d14c5d2SYehuda Sadeh int sb_flags; 723d14c5d2SYehuda Sadeh 7383817e35SSage Weil int wsize; /* max write size */ 7483817e35SSage Weil int rsize; /* max read size */ 7583817e35SSage Weil int rasize; /* max readahead */ 766e19a16eSSage Weil int congestion_kb; /* max writeback in flight */ 776e19a16eSSage Weil int caps_wanted_delay_min, caps_wanted_delay_max; 7823804d91SSage Weil int max_readdir; /* max readdir result (entires) */ 7923804d91SSage Weil int max_readdir_bytes; /* max readdir result (bytes) */ 803d14c5d2SYehuda Sadeh 813d14c5d2SYehuda Sadeh /* 823d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 833d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 843d14c5d2SYehuda Sadeh */ 853d14c5d2SYehuda Sadeh 86de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 87430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 883f384954SYan, Zheng char *server_path; /* default "/" */ 891d8f8360SYan, Zheng char *fscache_uniq; /* default NULL */ 90de57606cSSage Weil }; 91de57606cSSage Weil 923d14c5d2SYehuda Sadeh struct ceph_fs_client { 93de57606cSSage Weil struct super_block *sb; 94de57606cSSage Weil 953d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 963d14c5d2SYehuda Sadeh struct ceph_client *client; 973d14c5d2SYehuda Sadeh 98de57606cSSage Weil unsigned long mount_state; 9985ccce43SSage Weil int min_caps; /* min caps i added */ 10085ccce43SSage Weil 1013d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 102de57606cSSage Weil 103de57606cSSage Weil /* writeback */ 104de57606cSSage Weil mempool_t *wb_pagevec_pool; 105de57606cSSage Weil struct workqueue_struct *wb_wq; 106de57606cSSage Weil struct workqueue_struct *pg_inv_wq; 107de57606cSSage Weil struct workqueue_struct *trunc_wq; 1082baba250SYehuda Sadeh atomic_long_t writeback_count; 109de57606cSSage Weil 1100743304dSSage Weil #ifdef CONFIG_DEBUG_FS 1113d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1122baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 11306edf046SSage Weil struct dentry *debugfs_bdi; 1143d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 11514ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 1160743304dSSage Weil #endif 11799ccbd22SMilosz Tanski 11899ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 11999ccbd22SMilosz Tanski struct fscache_cookie *fscache; 12099ccbd22SMilosz Tanski #endif 121de57606cSSage Weil }; 122de57606cSSage Weil 1233d14c5d2SYehuda Sadeh 124de57606cSSage Weil /* 125de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 126de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 127de57606cSSage Weil * and write data. For any given inode, we should have one or more 128de57606cSSage Weil * capabilities, one issued by each metadata server, and our 129de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 130de57606cSSage Weil * 131de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 132de57606cSSage Weil * session capability lists. 133de57606cSSage Weil */ 134de57606cSSage Weil struct ceph_cap { 135de57606cSSage Weil struct ceph_inode_info *ci; 136de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 137de57606cSSage Weil struct ceph_mds_session *session; 138de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 139de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 140745a8e3bSYan, Zheng union { 141745a8e3bSYan, Zheng /* in-use caps */ 142745a8e3bSYan, Zheng struct { 143de57606cSSage Weil int issued; /* latest, from the mds */ 144745a8e3bSYan, Zheng int implemented; /* implemented superset of 145745a8e3bSYan, Zheng issued (for revocation) */ 146745a8e3bSYan, Zheng int mds, mds_wanted; 147745a8e3bSYan, Zheng }; 148745a8e3bSYan, Zheng /* caps to release */ 149745a8e3bSYan, Zheng struct { 150745a8e3bSYan, Zheng u64 cap_ino; 151745a8e3bSYan, Zheng int queue_release; 152745a8e3bSYan, Zheng }; 153745a8e3bSYan, Zheng }; 154685f9a5dSSage Weil u32 seq, issue_seq, mseq; 155685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 156de57606cSSage Weil unsigned long last_used; 157de57606cSSage Weil struct list_head caps_item; 158de57606cSSage Weil }; 159de57606cSSage Weil 160de57606cSSage Weil #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ 161de57606cSSage Weil #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ 162de57606cSSage Weil #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ 163de57606cSSage Weil 1640e294387SYan, Zheng struct ceph_cap_flush { 1650e294387SYan, Zheng u64 tid; 1660e294387SYan, Zheng int caps; /* 0 means capsnap */ 167c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 1680e294387SYan, Zheng struct list_head g_list; // global 1690e294387SYan, Zheng struct list_head i_list; // per inode 1700e294387SYan, Zheng }; 1710e294387SYan, Zheng 172de57606cSSage Weil /* 173de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 174de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 175de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 176de57606cSSage Weil */ 177de57606cSSage Weil struct ceph_cap_snap { 178805692d0SElena Reshetova refcount_t nref; 1790e294387SYan, Zheng struct list_head ci_item; 180de57606cSSage Weil 1810e294387SYan, Zheng struct ceph_cap_flush cap_flush; 1820e294387SYan, Zheng 1830e294387SYan, Zheng u64 follows; 184de57606cSSage Weil int issued, dirty; 185de57606cSSage Weil struct ceph_snap_context *context; 186de57606cSSage Weil 1875706b27dSAl Viro umode_t mode; 18805cb11c1SEric W. Biederman kuid_t uid; 18905cb11c1SEric W. Biederman kgid_t gid; 190de57606cSSage Weil 1914a625be4SSage Weil struct ceph_buffer *xattr_blob; 192de57606cSSage Weil u64 xattr_version; 193de57606cSSage Weil 194de57606cSSage Weil u64 size; 195de57606cSSage Weil struct timespec mtime, atime, ctime; 196de57606cSSage Weil u64 time_warp_seq; 1975f743e45SYan, Zheng u64 truncate_size; 1985f743e45SYan, Zheng u32 truncate_seq; 199de57606cSSage Weil int writing; /* a sync write is still in progress */ 200de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 201e20d258dSYan, Zheng bool inline_data; 20286056090SYan, Zheng bool need_flush; 203de57606cSSage Weil }; 204de57606cSSage Weil 205de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 206de57606cSSage Weil { 207805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 2084a625be4SSage Weil if (capsnap->xattr_blob) 2094a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 210de57606cSSage Weil kfree(capsnap); 211de57606cSSage Weil } 2124a625be4SSage Weil } 213de57606cSSage Weil 214de57606cSSage Weil /* 215de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 216de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 217de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 218de57606cSSage Weil * 219de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 220de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 221de57606cSSage Weil */ 222de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 223de57606cSSage Weil 224de57606cSSage Weil struct ceph_inode_frag { 225de57606cSSage Weil struct rb_node node; 226de57606cSSage Weil 227de57606cSSage Weil /* fragtree state */ 228de57606cSSage Weil u32 frag; 229de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 230de57606cSSage Weil 231de57606cSSage Weil /* delegation and replication info */ 232de57606cSSage Weil int mds; /* -1 if same authority as parent */ 233de57606cSSage Weil int ndist; /* >0 if replicated */ 234de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 235de57606cSSage Weil }; 236de57606cSSage Weil 237de57606cSSage Weil /* 238de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 239de57606cSSage Weil * at which point we parse them into an rbtree. 240de57606cSSage Weil */ 241de57606cSSage Weil struct ceph_inode_xattr { 242de57606cSSage Weil struct rb_node node; 243de57606cSSage Weil 244de57606cSSage Weil const char *name; 245de57606cSSage Weil int name_len; 246de57606cSSage Weil const char *val; 247de57606cSSage Weil int val_len; 248de57606cSSage Weil int dirty; 249de57606cSSage Weil 250de57606cSSage Weil int should_free_name; 251de57606cSSage Weil int should_free_val; 252de57606cSSage Weil }; 253de57606cSSage Weil 2543d14c5d2SYehuda Sadeh /* 2553d14c5d2SYehuda Sadeh * Ceph dentry state 2563d14c5d2SYehuda Sadeh */ 2573d14c5d2SYehuda Sadeh struct ceph_dentry_info { 2583d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 25997aeb6bfSYan, Zheng int lease_shared_gen; 26097aeb6bfSYan, Zheng u32 lease_gen; 2613d14c5d2SYehuda Sadeh u32 lease_seq; 2623d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2633d14c5d2SYehuda Sadeh struct list_head lru; 2643d14c5d2SYehuda Sadeh struct dentry *dentry; 2659b16f03cSMiklos Szeredi unsigned long time; 2663d14c5d2SYehuda Sadeh u64 offset; 2673d14c5d2SYehuda Sadeh }; 2683d14c5d2SYehuda Sadeh 269de57606cSSage Weil struct ceph_inode_xattrs_info { 270de57606cSSage Weil /* 271de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 272de57606cSSage Weil * this until someone actually calls getxattr, etc. 273de57606cSSage Weil * 274de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 275de57606cSSage Weil * NULL means we don't know. 276de57606cSSage Weil */ 277de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 278de57606cSSage Weil 279de57606cSSage Weil struct rb_root index; 280de57606cSSage Weil bool dirty; 281de57606cSSage Weil int count; 282de57606cSSage Weil int names_size; 283de57606cSSage Weil int vals_size; 284de57606cSSage Weil u64 version, index_version; 285de57606cSSage Weil }; 286de57606cSSage Weil 287de57606cSSage Weil /* 288de57606cSSage Weil * Ceph inode. 289de57606cSSage Weil */ 290de57606cSSage Weil struct ceph_inode_info { 291de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 292de57606cSSage Weil 293be655596SSage Weil spinlock_t i_ceph_lock; 294be655596SSage Weil 295de57606cSSage Weil u64 i_version; 29631c542a1SYan, Zheng u64 i_inline_version; 297de57606cSSage Weil u32 i_time_warp_seq; 298de57606cSSage Weil 299de57606cSSage Weil unsigned i_ceph_flags; 300fdd4e158SYan, Zheng atomic64_t i_release_count; 301fdd4e158SYan, Zheng atomic64_t i_ordered_count; 302fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 303de57606cSSage Weil 3046c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 305de57606cSSage Weil struct ceph_file_layout i_layout; 306de57606cSSage Weil char *i_symlink; 307de57606cSSage Weil 308de57606cSSage Weil /* for dirs */ 309de57606cSSage Weil struct timespec i_rctime; 310de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 311de57606cSSage Weil u64 i_files, i_subdirs; 312de57606cSSage Weil 313de57606cSSage Weil struct rb_root i_fragtree; 3141b1bc16dSYan, Zheng int i_fragtree_nsplits; 315de57606cSSage Weil struct mutex i_fragtree_mutex; 316de57606cSSage Weil 317de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 318de57606cSSage Weil 319be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 320de57606cSSage Weil * s_mutex. */ 321de57606cSSage Weil struct rb_root i_caps; /* cap list */ 322de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 323de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 324de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 325de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 326de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 327de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 328f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 329e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 330de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 331de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 332de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 333de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 334de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 335de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3367d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3377d8cb26dSSage Weil dirty|flushing caps */ 338de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 339de57606cSSage Weil 340774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 341de57606cSSage Weil 342b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 343de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 344de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 345de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 346de57606cSSage Weil 347de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 348de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 349de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 350de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 351de57606cSSage Weil 352de57606cSSage Weil /* held references to caps */ 353de57606cSSage Weil int i_pin_ref; 354d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 355de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 35689aa5930SYan, Zheng atomic_t i_filelock_ref; 35797aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 358cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 359de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 360de57606cSSage Weil 361de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 36268cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 363de57606cSSage Weil spinlock_t i_unsafe_lock; 364de57606cSSage Weil 365de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 366de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 367de57606cSSage Weil struct list_head i_snap_realm_item; 368de57606cSSage Weil struct list_head i_snap_flush_item; 369de57606cSSage Weil 370de57606cSSage Weil struct work_struct i_wb_work; /* writeback work */ 371de57606cSSage Weil struct work_struct i_pg_inv_work; /* page invalidation work */ 372de57606cSSage Weil 373de57606cSSage Weil struct work_struct i_vmtruncate_work; 374de57606cSSage Weil 37599ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 37699ccbd22SMilosz Tanski struct fscache_cookie *fscache; 377f7f7e7a0SYan, Zheng u32 i_fscache_gen; 37899ccbd22SMilosz Tanski #endif 379de57606cSSage Weil struct inode vfs_inode; /* at end */ 380de57606cSSage Weil }; 381de57606cSSage Weil 382de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 383de57606cSSage Weil { 384fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 385de57606cSSage Weil } 386de57606cSSage Weil 387ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 388ad1fee96SYehuda Sadeh { 389ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 390ad1fee96SYehuda Sadeh } 391ad1fee96SYehuda Sadeh 392ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 393ad1fee96SYehuda Sadeh { 394ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 395ad1fee96SYehuda Sadeh } 396ad1fee96SYehuda Sadeh 3973d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 3983d14c5d2SYehuda Sadeh { 3993d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4003d14c5d2SYehuda Sadeh } 4013d14c5d2SYehuda Sadeh 4023d14c5d2SYehuda Sadeh /* 4033d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 4043d14c5d2SYehuda Sadeh * 405ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 406ad1fee96SYehuda Sadeh * i386 32 32 407ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 408ad1fee96SYehuda Sadeh * x86_64 64 64 409ad1fee96SYehuda Sadeh */ 4103310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 411ad1fee96SYehuda Sadeh { 4123310f754SAmon Ott u32 ino = vino & 0xffffffff; 4133310f754SAmon Ott ino ^= vino >> 32; 414ad1fee96SYehuda Sadeh if (!ino) 415a661fc56SAmon Ott ino = 2; 416ad1fee96SYehuda Sadeh return ino; 417ad1fee96SYehuda Sadeh } 418ad1fee96SYehuda Sadeh 419ad1fee96SYehuda Sadeh /* 420ad1fee96SYehuda Sadeh * kernel i_ino value 4213d14c5d2SYehuda Sadeh */ 4223d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4233d14c5d2SYehuda Sadeh { 4243d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4253310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4263310f754SAmon Ott #else 4273310f754SAmon Ott return (ino_t)vino.ino; 4283d14c5d2SYehuda Sadeh #endif 4293d14c5d2SYehuda Sadeh } 4303d14c5d2SYehuda Sadeh 431ad1fee96SYehuda Sadeh /* 432ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 433ad1fee96SYehuda Sadeh */ 434ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 435ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 436ad1fee96SYehuda Sadeh { 437ad1fee96SYehuda Sadeh return ino; 438ad1fee96SYehuda Sadeh } 439ad1fee96SYehuda Sadeh #else 440ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 441ad1fee96SYehuda Sadeh { 442ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 443ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 444ad1fee96SYehuda Sadeh return ino; 445ad1fee96SYehuda Sadeh } 446ad1fee96SYehuda Sadeh #endif 447ad1fee96SYehuda Sadeh 448ad1fee96SYehuda Sadeh 4493d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4503d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4513d14c5d2SYehuda Sadeh 4523d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4533d14c5d2SYehuda Sadeh { 4543d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4553d14c5d2SYehuda Sadeh } 4563d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4573d14c5d2SYehuda Sadeh { 4583d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4593d14c5d2SYehuda Sadeh } 4603d14c5d2SYehuda Sadeh 4613d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4623d14c5d2SYehuda Sadeh { 4633d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4643d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4653d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4663d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4673d14c5d2SYehuda Sadeh } 4683d14c5d2SYehuda Sadeh 4693d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4703d14c5d2SYehuda Sadeh struct ceph_vino vino) 4713d14c5d2SYehuda Sadeh { 4723d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 4733d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 4743d14c5d2SYehuda Sadeh } 4753d14c5d2SYehuda Sadeh 4763d14c5d2SYehuda Sadeh 4773d14c5d2SYehuda Sadeh /* 4783d14c5d2SYehuda Sadeh * Ceph inode. 4793d14c5d2SYehuda Sadeh */ 48010183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 48110183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 48210183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 48310183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 48410183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 48510183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 48610183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 487315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 48877310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 48913c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 49070220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 49126544c62SJeff Layton #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ 492b3f8d68fSYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ 493b3f8d68fSYan, Zheng 49426544c62SJeff Layton 49526544c62SJeff Layton /* 49626544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 49726544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 49826544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 49926544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 50026544c62SJeff Layton * worried if a few slip through in either direction. 50126544c62SJeff Layton */ 50226544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 50326544c62SJeff Layton { 50426544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 50526544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 50626544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 50726544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 50826544c62SJeff Layton } 50926544c62SJeff Layton } 51026544c62SJeff Layton 51126544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 51226544c62SJeff Layton { 51326544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 51426544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 51526544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 51626544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 51726544c62SJeff Layton } 51826544c62SJeff Layton } 5193d14c5d2SYehuda Sadeh 5202f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 521fdd4e158SYan, Zheng long long release_count, 522fdd4e158SYan, Zheng long long ordered_count) 523de57606cSSage Weil { 524fdd4e158SYan, Zheng smp_mb__before_atomic(); 525fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 526fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 527de57606cSSage Weil } 528de57606cSSage Weil 5292f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 530de57606cSSage Weil { 531fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 532fdd4e158SYan, Zheng } 533fdd4e158SYan, Zheng 534fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 535fdd4e158SYan, Zheng { 536fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 537de57606cSSage Weil } 538de57606cSSage Weil 5392f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 540de57606cSSage Weil { 541fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 542fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 5432f276c51SYan, Zheng } 544de57606cSSage Weil 54570db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 54670db4f36SYan, Zheng { 547fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 548fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 549fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 550fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 55170db4f36SYan, Zheng } 55270db4f36SYan, Zheng 5532f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5542f276c51SYan, Zheng { 5552f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5562f276c51SYan, Zheng } 5572f276c51SYan, Zheng 55870db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5592f276c51SYan, Zheng { 560fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 561de57606cSSage Weil } 562de57606cSSage Weil 56370db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 56470db4f36SYan, Zheng { 565fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 566fdd4e158SYan, Zheng smp_rmb(); 56770db4f36SYan, Zheng return ret; 56870db4f36SYan, Zheng } 569de57606cSSage Weil 570de57606cSSage Weil /* find a specific frag @f */ 571de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 572de57606cSSage Weil u32 f); 573de57606cSSage Weil 574de57606cSSage Weil /* 575de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 576de57606cSSage Weil * exists 577de57606cSSage Weil */ 578de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 579de57606cSSage Weil struct ceph_inode_frag *pfrag, 580de57606cSSage Weil int *found); 581de57606cSSage Weil 582de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) 583de57606cSSage Weil { 584de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 585de57606cSSage Weil } 586de57606cSSage Weil 587de57606cSSage Weil /* 588de57606cSSage Weil * caps helpers 589de57606cSSage Weil */ 590de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 591de57606cSSage Weil { 592de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 593de57606cSSage Weil } 594de57606cSSage Weil 595de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 596de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 597de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 598de57606cSSage Weil struct ceph_cap *cap); 599de57606cSSage Weil 600de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 601de57606cSSage Weil { 602de57606cSSage Weil int issued; 603be655596SSage Weil spin_lock(&ci->i_ceph_lock); 604de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 605be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 606de57606cSSage Weil return issued; 607de57606cSSage Weil } 608de57606cSSage Weil 609de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 610de57606cSSage Weil int touch) 611de57606cSSage Weil { 612de57606cSSage Weil int r; 613be655596SSage Weil spin_lock(&ci->i_ceph_lock); 614de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 615be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 616de57606cSSage Weil return r; 617de57606cSSage Weil } 618de57606cSSage Weil 619de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 620de57606cSSage Weil { 621de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 622de57606cSSage Weil } 623f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 624f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 625f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 626f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 627de57606cSSage Weil 6289563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 6299563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 630de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 631de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 632de57606cSSage Weil 633de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 634de57606cSSage Weil 635de57606cSSage Weil /* 636de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 637de57606cSSage Weil */ 638de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 639de57606cSSage Weil { 640de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 641de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 642de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 643de57606cSSage Weil return w; 644de57606cSSage Weil } 645de57606cSSage Weil 646de57606cSSage Weil /* what the mds thinks we want */ 647c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 648de57606cSSage Weil 64937151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 65037151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 65137151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); 652e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 65337151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 65437151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, 65537151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6563d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 657de57606cSSage Weil int *total, int *avail, int *used, 65885ccce43SSage Weil int *reserved, int *min); 659de57606cSSage Weil 660de57606cSSage Weil 661de57606cSSage Weil 662de57606cSSage Weil /* 663de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 664de57606cSSage Weil */ 6654918b6d1SSage Weil #define CEPH_F_SYNC 1 6669cfa1098SSage Weil #define CEPH_F_ATEND 2 6674918b6d1SSage Weil 668de57606cSSage Weil struct ceph_file_info { 669252c6728SSage Weil short fmode; /* initialized on open */ 670252c6728SSage Weil short flags; /* CEPH_F_* */ 671de57606cSSage Weil 6725d988308SYan, Zheng spinlock_t rw_contexts_lock; 6735d988308SYan, Zheng struct list_head rw_contexts; 6745d988308SYan, Zheng 675de57606cSSage Weil /* readdir: position within the dir */ 676de57606cSSage Weil u32 frag; 677de57606cSSage Weil struct ceph_mds_request *last_readdir; 678de57606cSSage Weil 679de57606cSSage Weil /* readdir: position within a frag */ 680f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 681de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 682fdd4e158SYan, Zheng long long dir_release_count; 683fdd4e158SYan, Zheng long long dir_ordered_count; 684fdd4e158SYan, Zheng int readdir_cache_idx; 685de57606cSSage Weil 686de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 687de57606cSSage Weil char *dir_info; 688de57606cSSage Weil int dir_info_len; 689de57606cSSage Weil }; 690de57606cSSage Weil 6915d988308SYan, Zheng struct ceph_rw_context { 6925d988308SYan, Zheng struct list_head list; 6935d988308SYan, Zheng struct task_struct *thread; 6945d988308SYan, Zheng int caps; 6955d988308SYan, Zheng }; 6965d988308SYan, Zheng 6975d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 6985d988308SYan, Zheng struct ceph_rw_context _name = { \ 6995d988308SYan, Zheng .thread = current, \ 7005d988308SYan, Zheng .caps = _caps, \ 7015d988308SYan, Zheng } 7025d988308SYan, Zheng 7035d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 7045d988308SYan, Zheng struct ceph_rw_context *ctx) 7055d988308SYan, Zheng { 7065d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7075d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 7085d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7095d988308SYan, Zheng } 7105d988308SYan, Zheng 7115d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 7125d988308SYan, Zheng struct ceph_rw_context *ctx) 7135d988308SYan, Zheng { 7145d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7155d988308SYan, Zheng list_del(&ctx->list); 7165d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7175d988308SYan, Zheng } 7185d988308SYan, Zheng 7195d988308SYan, Zheng static inline struct ceph_rw_context* 7205d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 7215d988308SYan, Zheng { 7225d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 7235d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7245d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 7255d988308SYan, Zheng if (ctx->thread == current) { 7265d988308SYan, Zheng found = ctx; 7275d988308SYan, Zheng break; 7285d988308SYan, Zheng } 7295d988308SYan, Zheng } 7305d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7315d988308SYan, Zheng return found; 7325d988308SYan, Zheng } 7335d988308SYan, Zheng 734fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 735fdd4e158SYan, Zheng struct page *page; 736fdd4e158SYan, Zheng struct dentry **dentries; 737fdd4e158SYan, Zheng int index; 738fdd4e158SYan, Zheng }; 739de57606cSSage Weil 740de57606cSSage Weil /* 741de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 742de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 743de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 744de57606cSSage Weil * the snapshots of their parents. 745de57606cSSage Weil * 746de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 747de57606cSSage Weil * per-realm list. 748de57606cSSage Weil */ 749de57606cSSage Weil struct ceph_snap_realm { 750de57606cSSage Weil u64 ino; 751de57606cSSage Weil atomic_t nref; 752a105f00cSSage Weil struct rb_node node; 753a105f00cSSage Weil 754de57606cSSage Weil u64 created, seq; 755de57606cSSage Weil u64 parent_ino; 756de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 757de57606cSSage Weil 758de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 759aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 760de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 761aa711ee3SAlex Elder u32 num_snaps; 762de57606cSSage Weil 763de57606cSSage Weil struct ceph_snap_realm *parent; 764de57606cSSage Weil struct list_head children; /* list of child realms */ 765de57606cSSage Weil struct list_head child_item; 766de57606cSSage Weil 767de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 768de57606cSSage Weil 769ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 770ae00d4f3SSage Weil 771de57606cSSage Weil /* the current set of snaps for this realm */ 772de57606cSSage Weil struct ceph_snap_context *cached_context; 773de57606cSSage Weil 774de57606cSSage Weil struct list_head inodes_with_caps; 775de57606cSSage Weil spinlock_t inodes_with_caps_lock; 776de57606cSSage Weil }; 777de57606cSSage Weil 7783d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 7793d14c5d2SYehuda Sadeh { 7803d14c5d2SYehuda Sadeh int congestion_kb; 781de57606cSSage Weil 782de57606cSSage Weil /* 7833d14c5d2SYehuda Sadeh * Copied from NFS 7843d14c5d2SYehuda Sadeh * 7853d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 7863d14c5d2SYehuda Sadeh * 7873d14c5d2SYehuda Sadeh * 64MB: 8192k 7883d14c5d2SYehuda Sadeh * 128MB: 11585k 7893d14c5d2SYehuda Sadeh * 256MB: 16384k 7903d14c5d2SYehuda Sadeh * 512MB: 23170k 7913d14c5d2SYehuda Sadeh * 1GB: 32768k 7923d14c5d2SYehuda Sadeh * 2GB: 46340k 7933d14c5d2SYehuda Sadeh * 4GB: 65536k 7943d14c5d2SYehuda Sadeh * 8GB: 92681k 7953d14c5d2SYehuda Sadeh * 16GB: 131072k 7963d14c5d2SYehuda Sadeh * 7973d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 7983d14c5d2SYehuda Sadeh * Limit the default to 256M 799de57606cSSage Weil */ 8003d14c5d2SYehuda Sadeh congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); 8013d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 8023d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 8033d14c5d2SYehuda Sadeh 8043d14c5d2SYehuda Sadeh return congestion_kb; 805de57606cSSage Weil } 806de57606cSSage Weil 807de57606cSSage Weil 808de57606cSSage Weil 809de57606cSSage Weil /* snap.c */ 810de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 811de57606cSSage Weil u64 ino); 812de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 813de57606cSSage Weil struct ceph_snap_realm *realm); 814de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 815de57606cSSage Weil struct ceph_snap_realm *realm); 816de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 817982d6011SYan, Zheng void *p, void *e, bool deletion, 818982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 819de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 8202600d2ddSSage Weil struct ceph_mds_session *session, 821de57606cSSage Weil struct ceph_msg *msg); 822fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 823de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 824de57606cSSage Weil struct ceph_cap_snap *capsnap); 825de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 826de57606cSSage Weil 827de57606cSSage Weil /* 828de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 829de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 830de57606cSSage Weil */ 831de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 832de57606cSSage Weil { 833de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 83486056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 835de57606cSSage Weil ci_item)->writing; 836de57606cSSage Weil } 837de57606cSSage Weil 838de57606cSSage Weil /* inode.c */ 839de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 840de57606cSSage Weil 841de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 842de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode); 8439f12bd11SYan, Zheng extern int ceph_drop_inode(struct inode *inode); 844de57606cSSage Weil 845de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 846de57606cSSage Weil struct ceph_vino vino); 847de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 848de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 849de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 850de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 851de57606cSSage Weil u64 time_warp_seq, struct timespec *ctime, 852de57606cSSage Weil struct timespec *mtime, struct timespec *atime); 853de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 854f5a03b08SJeff Layton struct ceph_mds_request *req); 855de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 856de57606cSSage Weil struct ceph_mds_session *session); 857de57606cSSage Weil 858de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 859de57606cSSage Weil 860efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 861b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 8623c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 8633c6f6b79SSage Weil 8643c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 8653c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 866de57606cSSage Weil 86701deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 86801deead0SYan, Zheng int mask, bool force); 86901deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 87001deead0SYan, Zheng { 87101deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 87201deead0SYan, Zheng } 87310556cb2SAl Viro extern int ceph_permission(struct inode *inode, int mask); 874a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 875de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); 876a528d35eSDavid Howells extern int ceph_getattr(const struct path *path, struct kstat *stat, 877a528d35eSDavid Howells u32 request_mask, unsigned int flags); 878de57606cSSage Weil 879de57606cSSage Weil /* xattr.c */ 880a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 8817221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 882de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 883de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); 884de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 8853ce6cd12SAlex Elder extern void __init ceph_xattr_init(void); 8863ce6cd12SAlex Elder extern void ceph_xattr_exit(void); 887b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 888de57606cSSage Weil 889315f2408SYan, Zheng #ifdef CONFIG_SECURITY 890315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 891315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 892315f2408SYan, Zheng #else 893315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 894315f2408SYan, Zheng { 895315f2408SYan, Zheng return false; 896315f2408SYan, Zheng } 897315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 898315f2408SYan, Zheng { 899315f2408SYan, Zheng return false; 900315f2408SYan, Zheng } 901315f2408SYan, Zheng #endif 902315f2408SYan, Zheng 9037221fe4cSGuangliang Zhao /* acl.c */ 904b1ee94aaSYan, Zheng struct ceph_acls_info { 905b1ee94aaSYan, Zheng void *default_acl; 906b1ee94aaSYan, Zheng void *acl; 907b1ee94aaSYan, Zheng struct ceph_pagelist *pagelist; 908b1ee94aaSYan, Zheng }; 9097221fe4cSGuangliang Zhao 9107221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 9117221fe4cSGuangliang Zhao 9127221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 91372466d0bSSage Weil int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 914b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 915b1ee94aaSYan, Zheng struct ceph_acls_info *info); 916b1ee94aaSYan, Zheng void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); 917b1ee94aaSYan, Zheng void ceph_release_acls_info(struct ceph_acls_info *info); 918c969d9bfSGuangliang Zhao 919c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 920c969d9bfSGuangliang Zhao { 921c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 922c969d9bfSGuangliang Zhao } 9237221fe4cSGuangliang Zhao 9247221fe4cSGuangliang Zhao #else 9257221fe4cSGuangliang Zhao 9267221fe4cSGuangliang Zhao #define ceph_get_acl NULL 92772466d0bSSage Weil #define ceph_set_acl NULL 9287221fe4cSGuangliang Zhao 929b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 930b1ee94aaSYan, Zheng struct ceph_acls_info *info) 9317221fe4cSGuangliang Zhao { 9327221fe4cSGuangliang Zhao return 0; 9337221fe4cSGuangliang Zhao } 934b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 935b1ee94aaSYan, Zheng struct ceph_acls_info *info) 936b1ee94aaSYan, Zheng { 937b1ee94aaSYan, Zheng } 938b1ee94aaSYan, Zheng static inline void ceph_release_acls_info(struct ceph_acls_info *info) 939b1ee94aaSYan, Zheng { 940b1ee94aaSYan, Zheng } 9417221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 9427221fe4cSGuangliang Zhao { 9437221fe4cSGuangliang Zhao return 0; 9447221fe4cSGuangliang Zhao } 9457221fe4cSGuangliang Zhao 9467221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 9477221fe4cSGuangliang Zhao { 9487221fe4cSGuangliang Zhao } 9497221fe4cSGuangliang Zhao 9507221fe4cSGuangliang Zhao #endif 9517221fe4cSGuangliang Zhao 952de57606cSSage Weil /* caps.c */ 953de57606cSSage Weil extern const char *ceph_cap_string(int c); 954de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 955de57606cSSage Weil struct ceph_msg *msg); 956d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 957d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 958d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 959de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 960de57606cSSage Weil int fmode, unsigned issued, unsigned wanted, 961de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 962d9df2783SYan, Zheng struct ceph_cap **new_cap); 963a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 96437151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 96537151668SYehuda Sadeh struct ceph_cap *cap); 9669215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 967de57606cSSage Weil 968de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode); 969f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 97002c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 97102c24a82SJosef Bacik int datasync); 972e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 973e548e9b9SYan, Zheng struct ceph_mds_session *session); 974de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 975de57606cSSage Weil struct ceph_mds_session *session); 9762bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 9772bc50259SGreg Farnum int mds); 978de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode); 979de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 980de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 981de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 982de57606cSSage Weil struct ceph_snap_context *snapc); 983ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 9840e294387SYan, Zheng struct ceph_mds_session **psession); 985efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 986de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 987de57606cSSage Weil struct ceph_mds_session *session); 988afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 989afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 990*6ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 991de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 992de57606cSSage Weil int mds, int drop, int unless, int force); 993de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 994ca6c8ae0SJeff Layton struct inode *dir, 995de57606cSSage Weil int mds, int drop, int unless); 996de57606cSSage Weil 997de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, 99831c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 9992b1ac852SYan, Zheng extern int ceph_try_get_caps(struct ceph_inode_info *ci, 10002b1ac852SYan, Zheng int need, int want, int *got); 1001de57606cSSage Weil 1002de57606cSSage Weil /* for counting open files by mode */ 1003774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 1004de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 1005de57606cSSage Weil 1006de57606cSSage Weil /* addr.c */ 1007de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1008de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 100910183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 101010183a69SYan, Zheng extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); 101110183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 1012de57606cSSage Weil 1013de57606cSSage Weil /* file.c */ 1014de57606cSSage Weil extern const struct file_operations ceph_file_fops; 10159e0eb85dSAlex Elder 101677310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 1017de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 10185ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 10195ef50c3bSSage Weil struct file *file, unsigned flags, umode_t mode, 10205ef50c3bSSage Weil int *opened); 1021de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 102231c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 102331c542a1SYan, Zheng char *data, size_t len); 102455f2a045SIlya Dryomov 1025de57606cSSage Weil /* dir.c */ 1026de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 102738c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1028de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 102938c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 103018fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1031de57606cSSage Weil 1032f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1033de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1034468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 1035468640e3SSage Weil struct dentry *dentry, int err); 1036de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1037de57606cSSage Weil struct dentry *dentry, int err); 1038de57606cSSage Weil 1039de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn); 1040de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn); 1041de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn); 104281a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 1043e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1044fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1045de57606cSSage Weil 1046de57606cSSage Weil /* ioctl.c */ 1047de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1048de57606cSSage Weil 1049de57606cSSage Weil /* export.c */ 1050de57606cSSage Weil extern const struct export_operations ceph_export_ops; 1051de57606cSSage Weil 105240819f6fSGreg Farnum /* locks.c */ 1053eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 105440819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 105540819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 105640819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 105739be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 105839be95e9SJim Schutt struct ceph_filelock *flocks, 105939be95e9SJim Schutt int num_fcntl_locks, 106039be95e9SJim Schutt int num_flock_locks); 106139be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 106239be95e9SJim Schutt struct ceph_pagelist *pagelist, 106339be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 106440819f6fSGreg Farnum 10653d14c5d2SYehuda Sadeh /* debugfs.c */ 10663d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); 10673d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 10683d14c5d2SYehuda Sadeh 1069de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1070