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 313fb18a575SLuis Henriques /* quotas */ 314fb18a575SLuis Henriques u64 i_max_bytes, i_max_files; 315fb18a575SLuis Henriques 316de57606cSSage Weil struct rb_root i_fragtree; 3171b1bc16dSYan, Zheng int i_fragtree_nsplits; 318de57606cSSage Weil struct mutex i_fragtree_mutex; 319de57606cSSage Weil 320de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 321de57606cSSage Weil 322be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 323de57606cSSage Weil * s_mutex. */ 324de57606cSSage Weil struct rb_root i_caps; /* cap list */ 325de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 326de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 327de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 328de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 329de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 330de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 331f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 332e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 333de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 334de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 335de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 336de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 337de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 338de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3397d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3407d8cb26dSSage Weil dirty|flushing caps */ 341de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 342de57606cSSage Weil 343774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 344de57606cSSage Weil 345b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 346de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 347de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 348de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 349de57606cSSage Weil 350de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 351de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 352de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 353de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 354de57606cSSage Weil 355de57606cSSage Weil /* held references to caps */ 356de57606cSSage Weil int i_pin_ref; 357d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 358de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 35989aa5930SYan, Zheng atomic_t i_filelock_ref; 36097aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 361cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 362de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 363de57606cSSage Weil 364de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 36568cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 366de57606cSSage Weil spinlock_t i_unsafe_lock; 367de57606cSSage Weil 368de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 369de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 370de57606cSSage Weil struct list_head i_snap_realm_item; 371de57606cSSage Weil struct list_head i_snap_flush_item; 372de57606cSSage Weil 373de57606cSSage Weil struct work_struct i_wb_work; /* writeback work */ 374de57606cSSage Weil struct work_struct i_pg_inv_work; /* page invalidation work */ 375de57606cSSage Weil 376de57606cSSage Weil struct work_struct i_vmtruncate_work; 377de57606cSSage Weil 37899ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 37999ccbd22SMilosz Tanski struct fscache_cookie *fscache; 380f7f7e7a0SYan, Zheng u32 i_fscache_gen; 38199ccbd22SMilosz Tanski #endif 382de57606cSSage Weil struct inode vfs_inode; /* at end */ 383de57606cSSage Weil }; 384de57606cSSage Weil 385de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 386de57606cSSage Weil { 387fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 388de57606cSSage Weil } 389de57606cSSage Weil 390ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 391ad1fee96SYehuda Sadeh { 392ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 393ad1fee96SYehuda Sadeh } 394ad1fee96SYehuda Sadeh 395ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 396ad1fee96SYehuda Sadeh { 397ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 398ad1fee96SYehuda Sadeh } 399ad1fee96SYehuda Sadeh 4003d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 4013d14c5d2SYehuda Sadeh { 4023d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4033d14c5d2SYehuda Sadeh } 4043d14c5d2SYehuda Sadeh 4053d14c5d2SYehuda Sadeh /* 4063d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 4073d14c5d2SYehuda Sadeh * 408ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 409ad1fee96SYehuda Sadeh * i386 32 32 410ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 411ad1fee96SYehuda Sadeh * x86_64 64 64 412ad1fee96SYehuda Sadeh */ 4133310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 414ad1fee96SYehuda Sadeh { 4153310f754SAmon Ott u32 ino = vino & 0xffffffff; 4163310f754SAmon Ott ino ^= vino >> 32; 417ad1fee96SYehuda Sadeh if (!ino) 418a661fc56SAmon Ott ino = 2; 419ad1fee96SYehuda Sadeh return ino; 420ad1fee96SYehuda Sadeh } 421ad1fee96SYehuda Sadeh 422ad1fee96SYehuda Sadeh /* 423ad1fee96SYehuda Sadeh * kernel i_ino value 4243d14c5d2SYehuda Sadeh */ 4253d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4263d14c5d2SYehuda Sadeh { 4273d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4283310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4293310f754SAmon Ott #else 4303310f754SAmon Ott return (ino_t)vino.ino; 4313d14c5d2SYehuda Sadeh #endif 4323d14c5d2SYehuda Sadeh } 4333d14c5d2SYehuda Sadeh 434ad1fee96SYehuda Sadeh /* 435ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 436ad1fee96SYehuda Sadeh */ 437ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 438ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 439ad1fee96SYehuda Sadeh { 440ad1fee96SYehuda Sadeh return ino; 441ad1fee96SYehuda Sadeh } 442ad1fee96SYehuda Sadeh #else 443ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 444ad1fee96SYehuda Sadeh { 445ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 446ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 447ad1fee96SYehuda Sadeh return ino; 448ad1fee96SYehuda Sadeh } 449ad1fee96SYehuda Sadeh #endif 450ad1fee96SYehuda Sadeh 451ad1fee96SYehuda Sadeh 4523d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4533d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4543d14c5d2SYehuda Sadeh 4553d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4563d14c5d2SYehuda Sadeh { 4573d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4583d14c5d2SYehuda Sadeh } 4593d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4603d14c5d2SYehuda Sadeh { 4613d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4623d14c5d2SYehuda Sadeh } 4633d14c5d2SYehuda Sadeh 4643d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4653d14c5d2SYehuda Sadeh { 4663d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4673d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4683d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4693d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4703d14c5d2SYehuda Sadeh } 4713d14c5d2SYehuda Sadeh 4723d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4733d14c5d2SYehuda Sadeh struct ceph_vino vino) 4743d14c5d2SYehuda Sadeh { 4753d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 4763d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 4773d14c5d2SYehuda Sadeh } 4783d14c5d2SYehuda Sadeh 4793d14c5d2SYehuda Sadeh 4803d14c5d2SYehuda Sadeh /* 4813d14c5d2SYehuda Sadeh * Ceph inode. 4823d14c5d2SYehuda Sadeh */ 48310183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 48410183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 48510183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 48610183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 48710183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 48810183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 48910183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 490315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 49177310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 49213c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 49370220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 49426544c62SJeff Layton #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ 495b3f8d68fSYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ 496b3f8d68fSYan, Zheng 49726544c62SJeff Layton 49826544c62SJeff Layton /* 49926544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 50026544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 50126544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 50226544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 50326544c62SJeff Layton * worried if a few slip through in either direction. 50426544c62SJeff Layton */ 50526544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 50626544c62SJeff Layton { 50726544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 50826544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 50926544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 51026544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 51126544c62SJeff Layton } 51226544c62SJeff Layton } 51326544c62SJeff Layton 51426544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 51526544c62SJeff Layton { 51626544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 51726544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 51826544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 51926544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 52026544c62SJeff Layton } 52126544c62SJeff Layton } 5223d14c5d2SYehuda Sadeh 5232f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 524fdd4e158SYan, Zheng long long release_count, 525fdd4e158SYan, Zheng long long ordered_count) 526de57606cSSage Weil { 527fdd4e158SYan, Zheng smp_mb__before_atomic(); 528fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 529fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 530de57606cSSage Weil } 531de57606cSSage Weil 5322f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 533de57606cSSage Weil { 534fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 535fdd4e158SYan, Zheng } 536fdd4e158SYan, Zheng 537fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 538fdd4e158SYan, Zheng { 539fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 540de57606cSSage Weil } 541de57606cSSage Weil 5422f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 543de57606cSSage Weil { 544fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 545fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 5462f276c51SYan, Zheng } 547de57606cSSage Weil 54870db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 54970db4f36SYan, Zheng { 550fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 551fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 552fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 553fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 55470db4f36SYan, Zheng } 55570db4f36SYan, Zheng 5562f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5572f276c51SYan, Zheng { 5582f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5592f276c51SYan, Zheng } 5602f276c51SYan, Zheng 56170db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5622f276c51SYan, Zheng { 563fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 564de57606cSSage Weil } 565de57606cSSage Weil 56670db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 56770db4f36SYan, Zheng { 568fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 569fdd4e158SYan, Zheng smp_rmb(); 57070db4f36SYan, Zheng return ret; 57170db4f36SYan, Zheng } 572de57606cSSage Weil 573de57606cSSage Weil /* find a specific frag @f */ 574de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 575de57606cSSage Weil u32 f); 576de57606cSSage Weil 577de57606cSSage Weil /* 578de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 579de57606cSSage Weil * exists 580de57606cSSage Weil */ 581de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 582de57606cSSage Weil struct ceph_inode_frag *pfrag, 583de57606cSSage Weil int *found); 584de57606cSSage Weil 585de57606cSSage Weil static inline struct ceph_dentry_info *ceph_dentry(struct dentry *dentry) 586de57606cSSage Weil { 587de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 588de57606cSSage Weil } 589de57606cSSage Weil 590de57606cSSage Weil /* 591de57606cSSage Weil * caps helpers 592de57606cSSage Weil */ 593de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 594de57606cSSage Weil { 595de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 596de57606cSSage Weil } 597de57606cSSage Weil 598de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 599de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 600de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 601de57606cSSage Weil struct ceph_cap *cap); 602de57606cSSage Weil 603de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 604de57606cSSage Weil { 605de57606cSSage Weil int issued; 606be655596SSage Weil spin_lock(&ci->i_ceph_lock); 607de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 608be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 609de57606cSSage Weil return issued; 610de57606cSSage Weil } 611de57606cSSage Weil 612de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 613de57606cSSage Weil int touch) 614de57606cSSage Weil { 615de57606cSSage Weil int r; 616be655596SSage Weil spin_lock(&ci->i_ceph_lock); 617de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 618be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 619de57606cSSage Weil return r; 620de57606cSSage Weil } 621de57606cSSage Weil 622de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 623de57606cSSage Weil { 624de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 625de57606cSSage Weil } 626f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 627f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 628f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 629f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 630de57606cSSage Weil 6319563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 6329563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 633de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 634de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 635de57606cSSage Weil 636de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 637de57606cSSage Weil 638de57606cSSage Weil /* 639de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 640de57606cSSage Weil */ 641de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 642de57606cSSage Weil { 643de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 644de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 645de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 646de57606cSSage Weil return w; 647de57606cSSage Weil } 648de57606cSSage Weil 649de57606cSSage Weil /* what the mds thinks we want */ 650c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 651de57606cSSage Weil 65237151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 65337151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 65437151668SYehuda Sadeh extern void ceph_adjust_min_caps(struct ceph_mds_client *mdsc, int delta); 655e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 65637151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 65737151668SYehuda Sadeh extern int ceph_unreserve_caps(struct ceph_mds_client *mdsc, 65837151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6593d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 660de57606cSSage Weil int *total, int *avail, int *used, 66185ccce43SSage Weil int *reserved, int *min); 662de57606cSSage Weil 663de57606cSSage Weil 664de57606cSSage Weil 665de57606cSSage Weil /* 666de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 667de57606cSSage Weil */ 6684918b6d1SSage Weil #define CEPH_F_SYNC 1 6699cfa1098SSage Weil #define CEPH_F_ATEND 2 6704918b6d1SSage Weil 671de57606cSSage Weil struct ceph_file_info { 672252c6728SSage Weil short fmode; /* initialized on open */ 673252c6728SSage Weil short flags; /* CEPH_F_* */ 674de57606cSSage Weil 6755d988308SYan, Zheng spinlock_t rw_contexts_lock; 6765d988308SYan, Zheng struct list_head rw_contexts; 677bb48bd4dSChengguang Xu }; 678bb48bd4dSChengguang Xu 679bb48bd4dSChengguang Xu struct ceph_dir_file_info { 680bb48bd4dSChengguang Xu struct ceph_file_info file_info; 6815d988308SYan, Zheng 682de57606cSSage Weil /* readdir: position within the dir */ 683de57606cSSage Weil u32 frag; 684de57606cSSage Weil struct ceph_mds_request *last_readdir; 685de57606cSSage Weil 686de57606cSSage Weil /* readdir: position within a frag */ 687f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 688de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 689fdd4e158SYan, Zheng long long dir_release_count; 690fdd4e158SYan, Zheng long long dir_ordered_count; 691fdd4e158SYan, Zheng int readdir_cache_idx; 692de57606cSSage Weil 693de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 694de57606cSSage Weil char *dir_info; 695de57606cSSage Weil int dir_info_len; 696de57606cSSage Weil }; 697de57606cSSage Weil 6985d988308SYan, Zheng struct ceph_rw_context { 6995d988308SYan, Zheng struct list_head list; 7005d988308SYan, Zheng struct task_struct *thread; 7015d988308SYan, Zheng int caps; 7025d988308SYan, Zheng }; 7035d988308SYan, Zheng 7045d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 7055d988308SYan, Zheng struct ceph_rw_context _name = { \ 7065d988308SYan, Zheng .thread = current, \ 7075d988308SYan, Zheng .caps = _caps, \ 7085d988308SYan, Zheng } 7095d988308SYan, Zheng 7105d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 7115d988308SYan, Zheng struct ceph_rw_context *ctx) 7125d988308SYan, Zheng { 7135d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7145d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 7155d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7165d988308SYan, Zheng } 7175d988308SYan, Zheng 7185d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 7195d988308SYan, Zheng struct ceph_rw_context *ctx) 7205d988308SYan, Zheng { 7215d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7225d988308SYan, Zheng list_del(&ctx->list); 7235d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7245d988308SYan, Zheng } 7255d988308SYan, Zheng 7265d988308SYan, Zheng static inline struct ceph_rw_context* 7275d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 7285d988308SYan, Zheng { 7295d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 7305d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7315d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 7325d988308SYan, Zheng if (ctx->thread == current) { 7335d988308SYan, Zheng found = ctx; 7345d988308SYan, Zheng break; 7355d988308SYan, Zheng } 7365d988308SYan, Zheng } 7375d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7385d988308SYan, Zheng return found; 7395d988308SYan, Zheng } 7405d988308SYan, Zheng 741fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 742fdd4e158SYan, Zheng struct page *page; 743fdd4e158SYan, Zheng struct dentry **dentries; 744fdd4e158SYan, Zheng int index; 745fdd4e158SYan, Zheng }; 746de57606cSSage Weil 747de57606cSSage Weil /* 748de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 749de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 750de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 751de57606cSSage Weil * the snapshots of their parents. 752de57606cSSage Weil * 753de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 754de57606cSSage Weil * per-realm list. 755de57606cSSage Weil */ 756de57606cSSage Weil struct ceph_snap_realm { 757de57606cSSage Weil u64 ino; 758*e3161f17SLuis Henriques struct inode *inode; 759de57606cSSage Weil atomic_t nref; 760a105f00cSSage Weil struct rb_node node; 761a105f00cSSage Weil 762de57606cSSage Weil u64 created, seq; 763de57606cSSage Weil u64 parent_ino; 764de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 765de57606cSSage Weil 766de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 767aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 768de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 769aa711ee3SAlex Elder u32 num_snaps; 770de57606cSSage Weil 771de57606cSSage Weil struct ceph_snap_realm *parent; 772de57606cSSage Weil struct list_head children; /* list of child realms */ 773de57606cSSage Weil struct list_head child_item; 774de57606cSSage Weil 775de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 776de57606cSSage Weil 777ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 778ae00d4f3SSage Weil 779de57606cSSage Weil /* the current set of snaps for this realm */ 780de57606cSSage Weil struct ceph_snap_context *cached_context; 781de57606cSSage Weil 782de57606cSSage Weil struct list_head inodes_with_caps; 783de57606cSSage Weil spinlock_t inodes_with_caps_lock; 784de57606cSSage Weil }; 785de57606cSSage Weil 7863d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 7873d14c5d2SYehuda Sadeh { 7883d14c5d2SYehuda Sadeh int congestion_kb; 789de57606cSSage Weil 790de57606cSSage Weil /* 7913d14c5d2SYehuda Sadeh * Copied from NFS 7923d14c5d2SYehuda Sadeh * 7933d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 7943d14c5d2SYehuda Sadeh * 7953d14c5d2SYehuda Sadeh * 64MB: 8192k 7963d14c5d2SYehuda Sadeh * 128MB: 11585k 7973d14c5d2SYehuda Sadeh * 256MB: 16384k 7983d14c5d2SYehuda Sadeh * 512MB: 23170k 7993d14c5d2SYehuda Sadeh * 1GB: 32768k 8003d14c5d2SYehuda Sadeh * 2GB: 46340k 8013d14c5d2SYehuda Sadeh * 4GB: 65536k 8023d14c5d2SYehuda Sadeh * 8GB: 92681k 8033d14c5d2SYehuda Sadeh * 16GB: 131072k 8043d14c5d2SYehuda Sadeh * 8053d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 8063d14c5d2SYehuda Sadeh * Limit the default to 256M 807de57606cSSage Weil */ 8083d14c5d2SYehuda Sadeh congestion_kb = (16*int_sqrt(totalram_pages)) << (PAGE_SHIFT-10); 8093d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 8103d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 8113d14c5d2SYehuda Sadeh 8123d14c5d2SYehuda Sadeh return congestion_kb; 813de57606cSSage Weil } 814de57606cSSage Weil 815de57606cSSage Weil 816de57606cSSage Weil 817de57606cSSage Weil /* snap.c */ 818de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 819de57606cSSage Weil u64 ino); 820de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 821de57606cSSage Weil struct ceph_snap_realm *realm); 822de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 823de57606cSSage Weil struct ceph_snap_realm *realm); 824de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 825982d6011SYan, Zheng void *p, void *e, bool deletion, 826982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 827de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 8282600d2ddSSage Weil struct ceph_mds_session *session, 829de57606cSSage Weil struct ceph_msg *msg); 830fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 831de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 832de57606cSSage Weil struct ceph_cap_snap *capsnap); 833de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 834de57606cSSage Weil 835de57606cSSage Weil /* 836de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 837de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 838de57606cSSage Weil */ 839de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 840de57606cSSage Weil { 841de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 84286056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 843de57606cSSage Weil ci_item)->writing; 844de57606cSSage Weil } 845de57606cSSage Weil 846de57606cSSage Weil /* inode.c */ 847de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 848de57606cSSage Weil 849de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 850de57606cSSage Weil extern void ceph_destroy_inode(struct inode *inode); 8519f12bd11SYan, Zheng extern int ceph_drop_inode(struct inode *inode); 852de57606cSSage Weil 853de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 854de57606cSSage Weil struct ceph_vino vino); 855de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 856de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 857de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 858de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 859de57606cSSage Weil u64 time_warp_seq, struct timespec *ctime, 860de57606cSSage Weil struct timespec *mtime, struct timespec *atime); 861de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 862f5a03b08SJeff Layton struct ceph_mds_request *req); 863de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 864de57606cSSage Weil struct ceph_mds_session *session); 865de57606cSSage Weil 866de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 867de57606cSSage Weil 868efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 869b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 8703c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 8713c6f6b79SSage Weil 8723c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 8733c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 874de57606cSSage Weil 87501deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 87601deead0SYan, Zheng int mask, bool force); 87701deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 87801deead0SYan, Zheng { 87901deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 88001deead0SYan, Zheng } 88110556cb2SAl Viro extern int ceph_permission(struct inode *inode, int mask); 882a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 883de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); 884a528d35eSDavid Howells extern int ceph_getattr(const struct path *path, struct kstat *stat, 885a528d35eSDavid Howells u32 request_mask, unsigned int flags); 886de57606cSSage Weil 887de57606cSSage Weil /* xattr.c */ 888a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 8897221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 890de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 891de57606cSSage Weil extern void __ceph_build_xattrs_blob(struct ceph_inode_info *ci); 892de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 8933ce6cd12SAlex Elder extern void __init ceph_xattr_init(void); 8943ce6cd12SAlex Elder extern void ceph_xattr_exit(void); 895b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 896de57606cSSage Weil 897315f2408SYan, Zheng #ifdef CONFIG_SECURITY 898315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 899315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 900315f2408SYan, Zheng #else 901315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 902315f2408SYan, Zheng { 903315f2408SYan, Zheng return false; 904315f2408SYan, Zheng } 905315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 906315f2408SYan, Zheng { 907315f2408SYan, Zheng return false; 908315f2408SYan, Zheng } 909315f2408SYan, Zheng #endif 910315f2408SYan, Zheng 9117221fe4cSGuangliang Zhao /* acl.c */ 912b1ee94aaSYan, Zheng struct ceph_acls_info { 913b1ee94aaSYan, Zheng void *default_acl; 914b1ee94aaSYan, Zheng void *acl; 915b1ee94aaSYan, Zheng struct ceph_pagelist *pagelist; 916b1ee94aaSYan, Zheng }; 9177221fe4cSGuangliang Zhao 9187221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 9197221fe4cSGuangliang Zhao 9207221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 92172466d0bSSage Weil int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 922b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 923b1ee94aaSYan, Zheng struct ceph_acls_info *info); 924b1ee94aaSYan, Zheng void ceph_init_inode_acls(struct inode *inode, struct ceph_acls_info *info); 925b1ee94aaSYan, Zheng void ceph_release_acls_info(struct ceph_acls_info *info); 926c969d9bfSGuangliang Zhao 927c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 928c969d9bfSGuangliang Zhao { 929c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 930c969d9bfSGuangliang Zhao } 9317221fe4cSGuangliang Zhao 9327221fe4cSGuangliang Zhao #else 9337221fe4cSGuangliang Zhao 9347221fe4cSGuangliang Zhao #define ceph_get_acl NULL 93572466d0bSSage Weil #define ceph_set_acl NULL 9367221fe4cSGuangliang Zhao 937b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 938b1ee94aaSYan, Zheng struct ceph_acls_info *info) 9397221fe4cSGuangliang Zhao { 9407221fe4cSGuangliang Zhao return 0; 9417221fe4cSGuangliang Zhao } 942b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 943b1ee94aaSYan, Zheng struct ceph_acls_info *info) 944b1ee94aaSYan, Zheng { 945b1ee94aaSYan, Zheng } 946b1ee94aaSYan, Zheng static inline void ceph_release_acls_info(struct ceph_acls_info *info) 947b1ee94aaSYan, Zheng { 948b1ee94aaSYan, Zheng } 9497221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 9507221fe4cSGuangliang Zhao { 9517221fe4cSGuangliang Zhao return 0; 9527221fe4cSGuangliang Zhao } 9537221fe4cSGuangliang Zhao 9547221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 9557221fe4cSGuangliang Zhao { 9567221fe4cSGuangliang Zhao } 9577221fe4cSGuangliang Zhao 9587221fe4cSGuangliang Zhao #endif 9597221fe4cSGuangliang Zhao 960de57606cSSage Weil /* caps.c */ 961de57606cSSage Weil extern const char *ceph_cap_string(int c); 962de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 963de57606cSSage Weil struct ceph_msg *msg); 964d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 965d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 966d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 967de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 968de57606cSSage Weil int fmode, unsigned issued, unsigned wanted, 969de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 970d9df2783SYan, Zheng struct ceph_cap **new_cap); 971a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 97237151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 97337151668SYehuda Sadeh struct ceph_cap *cap); 9749215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 975de57606cSSage Weil 976de57606cSSage Weil extern void ceph_queue_caps_release(struct inode *inode); 977f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 97802c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 97902c24a82SJosef Bacik int datasync); 980e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 981e548e9b9SYan, Zheng struct ceph_mds_session *session); 982de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 983de57606cSSage Weil struct ceph_mds_session *session); 9842bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 9852bc50259SGreg Farnum int mds); 986de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode); 987de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 988de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 989de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 990de57606cSSage Weil struct ceph_snap_context *snapc); 991ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 9920e294387SYan, Zheng struct ceph_mds_session **psession); 993efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 994de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 995de57606cSSage Weil struct ceph_mds_session *session); 996afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 997afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 9986ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 999de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 1000de57606cSSage Weil int mds, int drop, int unless, int force); 1001de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 1002ca6c8ae0SJeff Layton struct inode *dir, 1003de57606cSSage Weil int mds, int drop, int unless); 1004de57606cSSage Weil 1005de57606cSSage Weil extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want, 100631c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 10072b1ac852SYan, Zheng extern int ceph_try_get_caps(struct ceph_inode_info *ci, 10082b1ac852SYan, Zheng int need, int want, int *got); 1009de57606cSSage Weil 1010de57606cSSage Weil /* for counting open files by mode */ 1011774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 1012de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 1013de57606cSSage Weil 1014de57606cSSage Weil /* addr.c */ 1015de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1016de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 101710183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 101810183a69SYan, Zheng extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need); 101910183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 1020de57606cSSage Weil 1021de57606cSSage Weil /* file.c */ 1022de57606cSSage Weil extern const struct file_operations ceph_file_fops; 10239e0eb85dSAlex Elder 102477310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 1025de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 10265ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 10275ef50c3bSSage Weil struct file *file, unsigned flags, umode_t mode, 10285ef50c3bSSage Weil int *opened); 1029de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 103031c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 103131c542a1SYan, Zheng char *data, size_t len); 103255f2a045SIlya Dryomov 1033de57606cSSage Weil /* dir.c */ 1034de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 103538c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1036de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 103738c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 103818fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1039de57606cSSage Weil 1040f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1041de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1042468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 1043468640e3SSage Weil struct dentry *dentry, int err); 1044de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1045de57606cSSage Weil struct dentry *dentry, int err); 1046de57606cSSage Weil 1047de57606cSSage Weil extern void ceph_dentry_lru_add(struct dentry *dn); 1048de57606cSSage Weil extern void ceph_dentry_lru_touch(struct dentry *dn); 1049de57606cSSage Weil extern void ceph_dentry_lru_del(struct dentry *dn); 105081a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 1051e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1052fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1053de57606cSSage Weil 1054de57606cSSage Weil /* ioctl.c */ 1055de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1056de57606cSSage Weil 1057de57606cSSage Weil /* export.c */ 1058de57606cSSage Weil extern const struct export_operations ceph_export_ops; 1059de57606cSSage Weil 106040819f6fSGreg Farnum /* locks.c */ 1061eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 106240819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 106340819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 106440819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 106539be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 106639be95e9SJim Schutt struct ceph_filelock *flocks, 106739be95e9SJim Schutt int num_fcntl_locks, 106839be95e9SJim Schutt int num_flock_locks); 106939be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 107039be95e9SJim Schutt struct ceph_pagelist *pagelist, 107139be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 107240819f6fSGreg Farnum 10733d14c5d2SYehuda Sadeh /* debugfs.c */ 10743d14c5d2SYehuda Sadeh extern int ceph_fs_debugfs_init(struct ceph_fs_client *client); 10753d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 10763d14c5d2SYehuda Sadeh 1077fb18a575SLuis Henriques /* quota.c */ 1078fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc, 1079fb18a575SLuis Henriques struct ceph_mds_session *session, 1080fb18a575SLuis Henriques struct ceph_msg *msg); 1081b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); 1082cafe21a4SLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); 10832b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, 10842b83845fSLuis Henriques loff_t newlen); 10851ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, 10861ab302a0SLuis Henriques loff_t newlen); 1087fb18a575SLuis Henriques 1088de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1089