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 34*131d7eb4SYan, Zheng #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blacklisted */ 353d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 363d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 373d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 38ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 39a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 4099ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 4110183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 42e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 439122eed5SLuis Henriques #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ 44ea4cdc54SLuis Henriques #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ 456a259382SSage Weil 466f9718feSLuis Henriques #define CEPH_MOUNT_OPT_DEFAULT \ 476f9718feSLuis Henriques (CEPH_MOUNT_OPT_DCACHE | \ 486f9718feSLuis Henriques CEPH_MOUNT_OPT_NOCOPYFROM) 49de57606cSSage Weil 503d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 513d14c5d2SYehuda Sadeh (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt; 523d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 533d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 54de57606cSSage Weil 55aa187926SYan, Zheng /* max size of osd read request, limited by libceph */ 56aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN 5795cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size. 5895cca2b4SYan, Zheng * CEPH_MSG_MAX_DATA_LEN should be small enough. */ 5995cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN 607c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 613d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 623d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 633d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 64de57606cSSage Weil 654214fb15SYan, Zheng /* 664214fb15SYan, Zheng * Delay telling the MDS we no longer want caps, in case we reopen 674214fb15SYan, Zheng * the file. Delay a minimum amount of time, even if we send a cap 684214fb15SYan, Zheng * message for some other reason. Otherwise, take the oppotunity to 694214fb15SYan, Zheng * update the mds to avoid sending another message later. 704214fb15SYan, Zheng */ 714214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ 724214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ 734214fb15SYan, Zheng 743d14c5d2SYehuda Sadeh struct ceph_mount_options { 75de57606cSSage Weil int flags; 763d14c5d2SYehuda Sadeh int sb_flags; 773d14c5d2SYehuda Sadeh 7883817e35SSage Weil int wsize; /* max write size */ 7983817e35SSage Weil int rsize; /* max read size */ 8083817e35SSage Weil int rasize; /* max readahead */ 816e19a16eSSage Weil int congestion_kb; /* max writeback in flight */ 826e19a16eSSage Weil int caps_wanted_delay_min, caps_wanted_delay_max; 83fe33032dSYan, Zheng int caps_max; 8423804d91SSage Weil int max_readdir; /* max readdir result (entires) */ 8523804d91SSage Weil int max_readdir_bytes; /* max readdir result (bytes) */ 863d14c5d2SYehuda Sadeh 873d14c5d2SYehuda Sadeh /* 883d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 893d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 903d14c5d2SYehuda Sadeh */ 913d14c5d2SYehuda Sadeh 92de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 93430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 943f384954SYan, Zheng char *server_path; /* default "/" */ 951d8f8360SYan, Zheng char *fscache_uniq; /* default NULL */ 96de57606cSSage Weil }; 97de57606cSSage Weil 983d14c5d2SYehuda Sadeh struct ceph_fs_client { 99de57606cSSage Weil struct super_block *sb; 100de57606cSSage Weil 1013d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 1023d14c5d2SYehuda Sadeh struct ceph_client *client; 1033d14c5d2SYehuda Sadeh 104de57606cSSage Weil unsigned long mount_state; 10581f148a9SYan, Zheng 106*131d7eb4SYan, Zheng unsigned long last_auto_reconnect; 107*131d7eb4SYan, Zheng bool blacklisted; 108*131d7eb4SYan, Zheng 10981f148a9SYan, Zheng u32 filp_gen; 110719784baSChengguang Xu loff_t max_file_size; 11185ccce43SSage Weil 1123d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 113de57606cSSage Weil 114de57606cSSage Weil /* writeback */ 115de57606cSSage Weil mempool_t *wb_pagevec_pool; 116e3ec8d68SYan, Zheng atomic_long_t writeback_count; 117e3ec8d68SYan, Zheng 1181cf89a8dSYan, Zheng struct workqueue_struct *inode_wq; 119e3ec8d68SYan, Zheng struct workqueue_struct *cap_wq; 120de57606cSSage Weil 1210743304dSSage Weil #ifdef CONFIG_DEBUG_FS 1223d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1232baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 12406edf046SSage Weil struct dentry *debugfs_bdi; 1253d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 12614ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 1270743304dSSage Weil #endif 12899ccbd22SMilosz Tanski 12999ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 13099ccbd22SMilosz Tanski struct fscache_cookie *fscache; 13199ccbd22SMilosz Tanski #endif 132de57606cSSage Weil }; 133de57606cSSage Weil 1343d14c5d2SYehuda Sadeh 135de57606cSSage Weil /* 136de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 137de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 138de57606cSSage Weil * and write data. For any given inode, we should have one or more 139de57606cSSage Weil * capabilities, one issued by each metadata server, and our 140de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 141de57606cSSage Weil * 142de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 143de57606cSSage Weil * session capability lists. 144de57606cSSage Weil */ 145de57606cSSage Weil struct ceph_cap { 146de57606cSSage Weil struct ceph_inode_info *ci; 147de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 148de57606cSSage Weil struct ceph_mds_session *session; 149de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 150de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 151745a8e3bSYan, Zheng union { 152745a8e3bSYan, Zheng /* in-use caps */ 153745a8e3bSYan, Zheng struct { 154de57606cSSage Weil int issued; /* latest, from the mds */ 155745a8e3bSYan, Zheng int implemented; /* implemented superset of 156745a8e3bSYan, Zheng issued (for revocation) */ 157745a8e3bSYan, Zheng int mds, mds_wanted; 158745a8e3bSYan, Zheng }; 159745a8e3bSYan, Zheng /* caps to release */ 160745a8e3bSYan, Zheng struct { 161745a8e3bSYan, Zheng u64 cap_ino; 162745a8e3bSYan, Zheng int queue_release; 163745a8e3bSYan, Zheng }; 164745a8e3bSYan, Zheng }; 165685f9a5dSSage Weil u32 seq, issue_seq, mseq; 166685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 167de57606cSSage Weil unsigned long last_used; 168de57606cSSage Weil struct list_head caps_item; 169de57606cSSage Weil }; 170de57606cSSage Weil 171de57606cSSage Weil #define CHECK_CAPS_NODELAY 1 /* do not delay any further */ 172de57606cSSage Weil #define CHECK_CAPS_AUTHONLY 2 /* only check auth cap */ 173de57606cSSage Weil #define CHECK_CAPS_FLUSH 4 /* flush any dirty caps */ 174de57606cSSage Weil 1750e294387SYan, Zheng struct ceph_cap_flush { 1760e294387SYan, Zheng u64 tid; 1770e294387SYan, Zheng int caps; /* 0 means capsnap */ 178c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 1790e294387SYan, Zheng struct list_head g_list; // global 1800e294387SYan, Zheng struct list_head i_list; // per inode 1810e294387SYan, Zheng }; 1820e294387SYan, Zheng 183de57606cSSage Weil /* 184de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 185de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 186de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 187de57606cSSage Weil */ 188de57606cSSage Weil struct ceph_cap_snap { 189805692d0SElena Reshetova refcount_t nref; 1900e294387SYan, Zheng struct list_head ci_item; 191de57606cSSage Weil 1920e294387SYan, Zheng struct ceph_cap_flush cap_flush; 1930e294387SYan, Zheng 1940e294387SYan, Zheng u64 follows; 195de57606cSSage Weil int issued, dirty; 196de57606cSSage Weil struct ceph_snap_context *context; 197de57606cSSage Weil 1985706b27dSAl Viro umode_t mode; 19905cb11c1SEric W. Biederman kuid_t uid; 20005cb11c1SEric W. Biederman kgid_t gid; 201de57606cSSage Weil 2024a625be4SSage Weil struct ceph_buffer *xattr_blob; 203de57606cSSage Weil u64 xattr_version; 204de57606cSSage Weil 205de57606cSSage Weil u64 size; 206176c77c9SJeff Layton u64 change_attr; 207ec62b894SJeff Layton struct timespec64 mtime, atime, ctime, btime; 208de57606cSSage Weil u64 time_warp_seq; 2095f743e45SYan, Zheng u64 truncate_size; 2105f743e45SYan, Zheng u32 truncate_seq; 211de57606cSSage Weil int writing; /* a sync write is still in progress */ 212de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 213e20d258dSYan, Zheng bool inline_data; 21486056090SYan, Zheng bool need_flush; 215de57606cSSage Weil }; 216de57606cSSage Weil 217de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 218de57606cSSage Weil { 219805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 2204a625be4SSage Weil if (capsnap->xattr_blob) 2214a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 222de57606cSSage Weil kfree(capsnap); 223de57606cSSage Weil } 2244a625be4SSage Weil } 225de57606cSSage Weil 226de57606cSSage Weil /* 227de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 228de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 229de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 230de57606cSSage Weil * 231de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 232de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 233de57606cSSage Weil */ 234de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 235de57606cSSage Weil 236de57606cSSage Weil struct ceph_inode_frag { 237de57606cSSage Weil struct rb_node node; 238de57606cSSage Weil 239de57606cSSage Weil /* fragtree state */ 240de57606cSSage Weil u32 frag; 241de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 242de57606cSSage Weil 243de57606cSSage Weil /* delegation and replication info */ 244de57606cSSage Weil int mds; /* -1 if same authority as parent */ 245de57606cSSage Weil int ndist; /* >0 if replicated */ 246de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 247de57606cSSage Weil }; 248de57606cSSage Weil 249de57606cSSage Weil /* 250de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 251de57606cSSage Weil * at which point we parse them into an rbtree. 252de57606cSSage Weil */ 253de57606cSSage Weil struct ceph_inode_xattr { 254de57606cSSage Weil struct rb_node node; 255de57606cSSage Weil 256de57606cSSage Weil const char *name; 257de57606cSSage Weil int name_len; 258de57606cSSage Weil const char *val; 259de57606cSSage Weil int val_len; 260de57606cSSage Weil int dirty; 261de57606cSSage Weil 262de57606cSSage Weil int should_free_name; 263de57606cSSage Weil int should_free_val; 264de57606cSSage Weil }; 265de57606cSSage Weil 2663d14c5d2SYehuda Sadeh /* 2673d14c5d2SYehuda Sadeh * Ceph dentry state 2683d14c5d2SYehuda Sadeh */ 2693d14c5d2SYehuda Sadeh struct ceph_dentry_info { 27037c4efc1SYan, Zheng struct dentry *dentry; 2713d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 27237c4efc1SYan, Zheng struct list_head lease_list; 27337c4efc1SYan, Zheng unsigned flags; 27497aeb6bfSYan, Zheng int lease_shared_gen; 27597aeb6bfSYan, Zheng u32 lease_gen; 2763d14c5d2SYehuda Sadeh u32 lease_seq; 2773d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2789b16f03cSMiklos Szeredi unsigned long time; 2793d14c5d2SYehuda Sadeh u64 offset; 2803d14c5d2SYehuda Sadeh }; 2813d14c5d2SYehuda Sadeh 28237c4efc1SYan, Zheng #define CEPH_DENTRY_REFERENCED 1 28337c4efc1SYan, Zheng #define CEPH_DENTRY_LEASE_LIST 2 28437c4efc1SYan, Zheng #define CEPH_DENTRY_SHRINK_LIST 4 28537c4efc1SYan, Zheng 286de57606cSSage Weil struct ceph_inode_xattrs_info { 287de57606cSSage Weil /* 288de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 289de57606cSSage Weil * this until someone actually calls getxattr, etc. 290de57606cSSage Weil * 291de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 292de57606cSSage Weil * NULL means we don't know. 293de57606cSSage Weil */ 294de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 295de57606cSSage Weil 296de57606cSSage Weil struct rb_root index; 297de57606cSSage Weil bool dirty; 298de57606cSSage Weil int count; 299de57606cSSage Weil int names_size; 300de57606cSSage Weil int vals_size; 301de57606cSSage Weil u64 version, index_version; 302de57606cSSage Weil }; 303de57606cSSage Weil 304de57606cSSage Weil /* 305de57606cSSage Weil * Ceph inode. 306de57606cSSage Weil */ 307de57606cSSage Weil struct ceph_inode_info { 308de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 309de57606cSSage Weil 310be655596SSage Weil spinlock_t i_ceph_lock; 311be655596SSage Weil 312de57606cSSage Weil u64 i_version; 31331c542a1SYan, Zheng u64 i_inline_version; 314de57606cSSage Weil u32 i_time_warp_seq; 315de57606cSSage Weil 316de57606cSSage Weil unsigned i_ceph_flags; 317fdd4e158SYan, Zheng atomic64_t i_release_count; 318fdd4e158SYan, Zheng atomic64_t i_ordered_count; 319fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 320de57606cSSage Weil 3216c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 322de57606cSSage Weil struct ceph_file_layout i_layout; 323de57606cSSage Weil char *i_symlink; 324de57606cSSage Weil 325de57606cSSage Weil /* for dirs */ 3269bbeab41SArnd Bergmann struct timespec64 i_rctime; 327de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 328de57606cSSage Weil u64 i_files, i_subdirs; 329de57606cSSage Weil 330fb18a575SLuis Henriques /* quotas */ 331fb18a575SLuis Henriques u64 i_max_bytes, i_max_files; 332fb18a575SLuis Henriques 33308796873SYan, Zheng s32 i_dir_pin; 33408796873SYan, Zheng 335de57606cSSage Weil struct rb_root i_fragtree; 3361b1bc16dSYan, Zheng int i_fragtree_nsplits; 337de57606cSSage Weil struct mutex i_fragtree_mutex; 338de57606cSSage Weil 339de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 340de57606cSSage Weil 341be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 342de57606cSSage Weil * s_mutex. */ 343de57606cSSage Weil struct rb_root i_caps; /* cap list */ 344de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 345de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 346de57606cSSage Weil struct list_head i_dirty_item, i_flushing_item; 347de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 348de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 349de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 350f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 351e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 352de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 353de57606cSSage Weil unsigned long i_hold_caps_min; /* jiffies */ 354de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 355de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 356de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 357de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3587d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3597d8cb26dSSage Weil dirty|flushing caps */ 360de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 361de57606cSSage Weil 362774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 363de57606cSSage Weil 364b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 365de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 366de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 367de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 368de57606cSSage Weil 369de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 370de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 371de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 372de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 373de57606cSSage Weil 374de57606cSSage Weil /* held references to caps */ 375de57606cSSage Weil int i_pin_ref; 376d3d0720dSHenry C Chang int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref; 377de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 37889aa5930SYan, Zheng atomic_t i_filelock_ref; 37997aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 380cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 381de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 382de57606cSSage Weil 383de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 38468cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 385de57606cSSage Weil spinlock_t i_unsafe_lock; 386de57606cSSage Weil 38775c9627eSYan, Zheng union { 388de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 38975c9627eSYan, Zheng struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ 39075c9627eSYan, Zheng }; 391de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 392de57606cSSage Weil struct list_head i_snap_realm_item; 393de57606cSSage Weil struct list_head i_snap_flush_item; 394245ce991SJeff Layton struct timespec64 i_btime; 395193e7b37SDavid Disseldorp struct timespec64 i_snap_btime; 396de57606cSSage Weil 3971cf89a8dSYan, Zheng struct work_struct i_work; 3981cf89a8dSYan, Zheng unsigned long i_work_mask; 399de57606cSSage Weil 40099ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 40199ccbd22SMilosz Tanski struct fscache_cookie *fscache; 402f7f7e7a0SYan, Zheng u32 i_fscache_gen; 40399ccbd22SMilosz Tanski #endif 404f4b97866SYan, Zheng errseq_t i_meta_err; 405f4b97866SYan, Zheng 406de57606cSSage Weil struct inode vfs_inode; /* at end */ 407de57606cSSage Weil }; 408de57606cSSage Weil 409de57606cSSage Weil static inline struct ceph_inode_info *ceph_inode(struct inode *inode) 410de57606cSSage Weil { 411fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 412de57606cSSage Weil } 413de57606cSSage Weil 414ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_inode_to_client(struct inode *inode) 415ad1fee96SYehuda Sadeh { 416ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 417ad1fee96SYehuda Sadeh } 418ad1fee96SYehuda Sadeh 419ad1fee96SYehuda Sadeh static inline struct ceph_fs_client *ceph_sb_to_client(struct super_block *sb) 420ad1fee96SYehuda Sadeh { 421ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 422ad1fee96SYehuda Sadeh } 423ad1fee96SYehuda Sadeh 4243d14c5d2SYehuda Sadeh static inline struct ceph_vino ceph_vino(struct inode *inode) 4253d14c5d2SYehuda Sadeh { 4263d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4273d14c5d2SYehuda Sadeh } 4283d14c5d2SYehuda Sadeh 4293d14c5d2SYehuda Sadeh /* 4303d14c5d2SYehuda Sadeh * ino_t is <64 bits on many architectures, blech. 4313d14c5d2SYehuda Sadeh * 432ad1fee96SYehuda Sadeh * i_ino (kernel inode) st_ino (userspace) 433ad1fee96SYehuda Sadeh * i386 32 32 434ad1fee96SYehuda Sadeh * x86_64+ino32 64 32 435ad1fee96SYehuda Sadeh * x86_64 64 64 436ad1fee96SYehuda Sadeh */ 4373310f754SAmon Ott static inline u32 ceph_ino_to_ino32(__u64 vino) 438ad1fee96SYehuda Sadeh { 4393310f754SAmon Ott u32 ino = vino & 0xffffffff; 4403310f754SAmon Ott ino ^= vino >> 32; 441ad1fee96SYehuda Sadeh if (!ino) 442a661fc56SAmon Ott ino = 2; 443ad1fee96SYehuda Sadeh return ino; 444ad1fee96SYehuda Sadeh } 445ad1fee96SYehuda Sadeh 446ad1fee96SYehuda Sadeh /* 447ad1fee96SYehuda Sadeh * kernel i_ino value 4483d14c5d2SYehuda Sadeh */ 4493d14c5d2SYehuda Sadeh static inline ino_t ceph_vino_to_ino(struct ceph_vino vino) 4503d14c5d2SYehuda Sadeh { 4513d14c5d2SYehuda Sadeh #if BITS_PER_LONG == 32 4523310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4533310f754SAmon Ott #else 4543310f754SAmon Ott return (ino_t)vino.ino; 4553d14c5d2SYehuda Sadeh #endif 4563d14c5d2SYehuda Sadeh } 4573d14c5d2SYehuda Sadeh 458ad1fee96SYehuda Sadeh /* 459ad1fee96SYehuda Sadeh * user-visible ino (stat, filldir) 460ad1fee96SYehuda Sadeh */ 461ad1fee96SYehuda Sadeh #if BITS_PER_LONG == 32 462ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 463ad1fee96SYehuda Sadeh { 464ad1fee96SYehuda Sadeh return ino; 465ad1fee96SYehuda Sadeh } 466ad1fee96SYehuda Sadeh #else 467ad1fee96SYehuda Sadeh static inline ino_t ceph_translate_ino(struct super_block *sb, ino_t ino) 468ad1fee96SYehuda Sadeh { 469ad1fee96SYehuda Sadeh if (ceph_test_mount_opt(ceph_sb_to_client(sb), INO32)) 470ad1fee96SYehuda Sadeh ino = ceph_ino_to_ino32(ino); 471ad1fee96SYehuda Sadeh return ino; 472ad1fee96SYehuda Sadeh } 473ad1fee96SYehuda Sadeh #endif 474ad1fee96SYehuda Sadeh 475ad1fee96SYehuda Sadeh 4763d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4773d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4783d14c5d2SYehuda Sadeh 4793d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4803d14c5d2SYehuda Sadeh { 4813d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4823d14c5d2SYehuda Sadeh } 4833d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4843d14c5d2SYehuda Sadeh { 4853d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4863d14c5d2SYehuda Sadeh } 4873d14c5d2SYehuda Sadeh 4883d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 4893d14c5d2SYehuda Sadeh { 4903d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 4913d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 4923d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 4933d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 4943d14c5d2SYehuda Sadeh } 4953d14c5d2SYehuda Sadeh 4963d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 4973d14c5d2SYehuda Sadeh struct ceph_vino vino) 4983d14c5d2SYehuda Sadeh { 4993d14c5d2SYehuda Sadeh ino_t t = ceph_vino_to_ino(vino); 5003d14c5d2SYehuda Sadeh return ilookup5(sb, t, ceph_ino_compare, &vino); 5013d14c5d2SYehuda Sadeh } 5023d14c5d2SYehuda Sadeh 5033d14c5d2SYehuda Sadeh 5043d14c5d2SYehuda Sadeh /* 5053d14c5d2SYehuda Sadeh * Ceph inode. 5063d14c5d2SYehuda Sadeh */ 50710183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 50810183a69SYan, Zheng #define CEPH_I_NODELAY (1 << 1) /* do not delay cap release */ 50910183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 51010183a69SYan, Zheng #define CEPH_I_NOFLUSH (1 << 3) /* do not flush dirty caps */ 51110183a69SYan, Zheng #define CEPH_I_POOL_PERM (1 << 4) /* pool rd/wr bits are valid */ 51210183a69SYan, Zheng #define CEPH_I_POOL_RD (1 << 5) /* can read from pool */ 51310183a69SYan, Zheng #define CEPH_I_POOL_WR (1 << 6) /* can write to pool */ 514315f2408SYan, Zheng #define CEPH_I_SEC_INITED (1 << 7) /* security initialized */ 51577310320SYan, Zheng #define CEPH_I_CAP_DROPPED (1 << 8) /* caps were forcibly dropped */ 51613c2b57dSYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 9) /* kick flushing caps */ 51770220ac8SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 10) /* need flush snapss */ 51826544c62SJeff Layton #define CEPH_I_ERROR_WRITE (1 << 11) /* have seen write errors */ 519b3f8d68fSYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 12) /* have seen file lock errors */ 520b3f8d68fSYan, Zheng 52126544c62SJeff Layton 52226544c62SJeff Layton /* 5231cf89a8dSYan, Zheng * Masks of ceph inode work. 5241cf89a8dSYan, Zheng */ 5251cf89a8dSYan, Zheng #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ 5261cf89a8dSYan, Zheng #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ 5271cf89a8dSYan, Zheng #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ 5281cf89a8dSYan, Zheng 5291cf89a8dSYan, Zheng /* 53026544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 53126544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 53226544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 53326544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 53426544c62SJeff Layton * worried if a few slip through in either direction. 53526544c62SJeff Layton */ 53626544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 53726544c62SJeff Layton { 53826544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 53926544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 54026544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 54126544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 54226544c62SJeff Layton } 54326544c62SJeff Layton } 54426544c62SJeff Layton 54526544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 54626544c62SJeff Layton { 54726544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 54826544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 54926544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 55026544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 55126544c62SJeff Layton } 55226544c62SJeff Layton } 5533d14c5d2SYehuda Sadeh 5542f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 555fdd4e158SYan, Zheng long long release_count, 556fdd4e158SYan, Zheng long long ordered_count) 557de57606cSSage Weil { 55874960773SAndrea Parri /* 55974960773SAndrea Parri * Makes sure operations that setup readdir cache (update page 56074960773SAndrea Parri * cache and i_size) are strongly ordered w.r.t. the following 56174960773SAndrea Parri * atomic64_set() operations. 56274960773SAndrea Parri */ 56374960773SAndrea Parri smp_mb(); 564fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 565fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 566de57606cSSage Weil } 567de57606cSSage Weil 5682f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 569de57606cSSage Weil { 570fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 571fdd4e158SYan, Zheng } 572fdd4e158SYan, Zheng 573fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 574fdd4e158SYan, Zheng { 575fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 576de57606cSSage Weil } 577de57606cSSage Weil 5782f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 579de57606cSSage Weil { 580fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 581fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 5822f276c51SYan, Zheng } 583de57606cSSage Weil 58470db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 58570db4f36SYan, Zheng { 586fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 587fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 588fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 589fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 59070db4f36SYan, Zheng } 59170db4f36SYan, Zheng 5922f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 5932f276c51SYan, Zheng { 5942f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 5952f276c51SYan, Zheng } 5962f276c51SYan, Zheng 59770db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 5982f276c51SYan, Zheng { 599fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 600de57606cSSage Weil } 601de57606cSSage Weil 60270db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 60370db4f36SYan, Zheng { 604fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 605fdd4e158SYan, Zheng smp_rmb(); 60670db4f36SYan, Zheng return ret; 60770db4f36SYan, Zheng } 608de57606cSSage Weil 609de57606cSSage Weil /* find a specific frag @f */ 610de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 611de57606cSSage Weil u32 f); 612de57606cSSage Weil 613de57606cSSage Weil /* 614de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 615de57606cSSage Weil * exists 616de57606cSSage Weil */ 617de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 618de57606cSSage Weil struct ceph_inode_frag *pfrag, 619de57606cSSage Weil int *found); 620de57606cSSage Weil 6211e9c2eb6SYan, Zheng static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) 622de57606cSSage Weil { 623de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 624de57606cSSage Weil } 625de57606cSSage Weil 626de57606cSSage Weil /* 627de57606cSSage Weil * caps helpers 628de57606cSSage Weil */ 629de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 630de57606cSSage Weil { 631de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 632de57606cSSage Weil } 633de57606cSSage Weil 634de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 635de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 636de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 637de57606cSSage Weil struct ceph_cap *cap); 638de57606cSSage Weil 639de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 640de57606cSSage Weil { 641de57606cSSage Weil int issued; 642be655596SSage Weil spin_lock(&ci->i_ceph_lock); 643de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 644be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 645de57606cSSage Weil return issued; 646de57606cSSage Weil } 647de57606cSSage Weil 648de57606cSSage Weil static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, 649de57606cSSage Weil int touch) 650de57606cSSage Weil { 651de57606cSSage Weil int r; 652be655596SSage Weil spin_lock(&ci->i_ceph_lock); 653de57606cSSage Weil r = __ceph_caps_issued_mask(ci, mask, touch); 654be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 655de57606cSSage Weil return r; 656de57606cSSage Weil } 657de57606cSSage Weil 658de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 659de57606cSSage Weil { 660de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 661de57606cSSage Weil } 662f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 663f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 664f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 665f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 666de57606cSSage Weil 6679563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 6689563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 669de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 670de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 671de57606cSSage Weil 672de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 673de57606cSSage Weil 674de57606cSSage Weil /* 675de57606cSSage Weil * wanted, by virtue of open file modes AND cap refs (buffered/cached data) 676de57606cSSage Weil */ 677de57606cSSage Weil static inline int __ceph_caps_wanted(struct ceph_inode_info *ci) 678de57606cSSage Weil { 679de57606cSSage Weil int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); 680de57606cSSage Weil if (w & CEPH_CAP_FILE_BUFFER) 681de57606cSSage Weil w |= CEPH_CAP_FILE_EXCL; /* we want EXCL if dirty data */ 682de57606cSSage Weil return w; 683de57606cSSage Weil } 684de57606cSSage Weil 685de57606cSSage Weil /* what the mds thinks we want */ 686c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 687de57606cSSage Weil 68837151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 68937151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 690fe33032dSYan, Zheng extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, 691fe33032dSYan, Zheng struct ceph_mount_options *fsopt); 692e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 69337151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 6947bf8f736SChengguang Xu extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, 69537151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 6963d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 697de57606cSSage Weil int *total, int *avail, int *used, 69885ccce43SSage Weil int *reserved, int *min); 699de57606cSSage Weil 700de57606cSSage Weil 701de57606cSSage Weil 702de57606cSSage Weil /* 703de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 704de57606cSSage Weil */ 7054918b6d1SSage Weil #define CEPH_F_SYNC 1 7069cfa1098SSage Weil #define CEPH_F_ATEND 2 7074918b6d1SSage Weil 708de57606cSSage Weil struct ceph_file_info { 709252c6728SSage Weil short fmode; /* initialized on open */ 710252c6728SSage Weil short flags; /* CEPH_F_* */ 711de57606cSSage Weil 7125d988308SYan, Zheng spinlock_t rw_contexts_lock; 7135d988308SYan, Zheng struct list_head rw_contexts; 714f4b97866SYan, Zheng 715f4b97866SYan, Zheng errseq_t meta_err; 71681f148a9SYan, Zheng u32 filp_gen; 717ff5d913dSYan, Zheng atomic_t num_locks; 718bb48bd4dSChengguang Xu }; 719bb48bd4dSChengguang Xu 720bb48bd4dSChengguang Xu struct ceph_dir_file_info { 721bb48bd4dSChengguang Xu struct ceph_file_info file_info; 7225d988308SYan, Zheng 723de57606cSSage Weil /* readdir: position within the dir */ 724de57606cSSage Weil u32 frag; 725de57606cSSage Weil struct ceph_mds_request *last_readdir; 726de57606cSSage Weil 727de57606cSSage Weil /* readdir: position within a frag */ 728f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 729de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 730fdd4e158SYan, Zheng long long dir_release_count; 731fdd4e158SYan, Zheng long long dir_ordered_count; 732fdd4e158SYan, Zheng int readdir_cache_idx; 733de57606cSSage Weil 734de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 735de57606cSSage Weil char *dir_info; 736de57606cSSage Weil int dir_info_len; 737de57606cSSage Weil }; 738de57606cSSage Weil 7395d988308SYan, Zheng struct ceph_rw_context { 7405d988308SYan, Zheng struct list_head list; 7415d988308SYan, Zheng struct task_struct *thread; 7425d988308SYan, Zheng int caps; 7435d988308SYan, Zheng }; 7445d988308SYan, Zheng 7455d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 7465d988308SYan, Zheng struct ceph_rw_context _name = { \ 7475d988308SYan, Zheng .thread = current, \ 7485d988308SYan, Zheng .caps = _caps, \ 7495d988308SYan, Zheng } 7505d988308SYan, Zheng 7515d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 7525d988308SYan, Zheng struct ceph_rw_context *ctx) 7535d988308SYan, Zheng { 7545d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7555d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 7565d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7575d988308SYan, Zheng } 7585d988308SYan, Zheng 7595d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 7605d988308SYan, Zheng struct ceph_rw_context *ctx) 7615d988308SYan, Zheng { 7625d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7635d988308SYan, Zheng list_del(&ctx->list); 7645d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7655d988308SYan, Zheng } 7665d988308SYan, Zheng 7675d988308SYan, Zheng static inline struct ceph_rw_context* 7685d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 7695d988308SYan, Zheng { 7705d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 7715d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7725d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 7735d988308SYan, Zheng if (ctx->thread == current) { 7745d988308SYan, Zheng found = ctx; 7755d988308SYan, Zheng break; 7765d988308SYan, Zheng } 7775d988308SYan, Zheng } 7785d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7795d988308SYan, Zheng return found; 7805d988308SYan, Zheng } 7815d988308SYan, Zheng 782fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 783fdd4e158SYan, Zheng struct page *page; 784fdd4e158SYan, Zheng struct dentry **dentries; 785fdd4e158SYan, Zheng int index; 786fdd4e158SYan, Zheng }; 787de57606cSSage Weil 788de57606cSSage Weil /* 789de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 790de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 791de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 792de57606cSSage Weil * the snapshots of their parents. 793de57606cSSage Weil * 794de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 795de57606cSSage Weil * per-realm list. 796de57606cSSage Weil */ 797de57606cSSage Weil struct ceph_snap_realm { 798de57606cSSage Weil u64 ino; 799e3161f17SLuis Henriques struct inode *inode; 800de57606cSSage Weil atomic_t nref; 801a105f00cSSage Weil struct rb_node node; 802a105f00cSSage Weil 803de57606cSSage Weil u64 created, seq; 804de57606cSSage Weil u64 parent_ino; 805de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 806de57606cSSage Weil 807de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 808aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 809de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 810aa711ee3SAlex Elder u32 num_snaps; 811de57606cSSage Weil 812de57606cSSage Weil struct ceph_snap_realm *parent; 813de57606cSSage Weil struct list_head children; /* list of child realms */ 814de57606cSSage Weil struct list_head child_item; 815de57606cSSage Weil 816de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 817de57606cSSage Weil 818ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 819ae00d4f3SSage Weil 820de57606cSSage Weil /* the current set of snaps for this realm */ 821de57606cSSage Weil struct ceph_snap_context *cached_context; 822de57606cSSage Weil 823de57606cSSage Weil struct list_head inodes_with_caps; 824de57606cSSage Weil spinlock_t inodes_with_caps_lock; 825de57606cSSage Weil }; 826de57606cSSage Weil 8273d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 8283d14c5d2SYehuda Sadeh { 8293d14c5d2SYehuda Sadeh int congestion_kb; 830de57606cSSage Weil 831de57606cSSage Weil /* 8323d14c5d2SYehuda Sadeh * Copied from NFS 8333d14c5d2SYehuda Sadeh * 8343d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 8353d14c5d2SYehuda Sadeh * 8363d14c5d2SYehuda Sadeh * 64MB: 8192k 8373d14c5d2SYehuda Sadeh * 128MB: 11585k 8383d14c5d2SYehuda Sadeh * 256MB: 16384k 8393d14c5d2SYehuda Sadeh * 512MB: 23170k 8403d14c5d2SYehuda Sadeh * 1GB: 32768k 8413d14c5d2SYehuda Sadeh * 2GB: 46340k 8423d14c5d2SYehuda Sadeh * 4GB: 65536k 8433d14c5d2SYehuda Sadeh * 8GB: 92681k 8443d14c5d2SYehuda Sadeh * 16GB: 131072k 8453d14c5d2SYehuda Sadeh * 8463d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 8473d14c5d2SYehuda Sadeh * Limit the default to 256M 848de57606cSSage Weil */ 849ca79b0c2SArun KS congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); 8503d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 8513d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 8523d14c5d2SYehuda Sadeh 8533d14c5d2SYehuda Sadeh return congestion_kb; 854de57606cSSage Weil } 855de57606cSSage Weil 856de57606cSSage Weil 857d468e729SYan, Zheng /* super.c */ 858d468e729SYan, Zheng extern int ceph_force_reconnect(struct super_block *sb); 859de57606cSSage Weil /* snap.c */ 860de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 861de57606cSSage Weil u64 ino); 862de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 863de57606cSSage Weil struct ceph_snap_realm *realm); 864de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 865de57606cSSage Weil struct ceph_snap_realm *realm); 866de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 867982d6011SYan, Zheng void *p, void *e, bool deletion, 868982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 869de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 8702600d2ddSSage Weil struct ceph_mds_session *session, 871de57606cSSage Weil struct ceph_msg *msg); 872fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 873de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 874de57606cSSage Weil struct ceph_cap_snap *capsnap); 875de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 876de57606cSSage Weil 87775c9627eSYan, Zheng extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, 87875c9627eSYan, Zheng u64 snap); 87975c9627eSYan, Zheng extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, 88075c9627eSYan, Zheng struct ceph_snapid_map *sm); 88175c9627eSYan, Zheng extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); 88275c9627eSYan, Zheng extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); 88375c9627eSYan, Zheng 88475c9627eSYan, Zheng 885de57606cSSage Weil /* 886de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 887de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 888de57606cSSage Weil */ 889de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 890de57606cSSage Weil { 891de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 89286056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 893de57606cSSage Weil ci_item)->writing; 894de57606cSSage Weil } 895de57606cSSage Weil 896de57606cSSage Weil /* inode.c */ 897de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 898de57606cSSage Weil 899de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 90087bc5b89SYan, Zheng extern void ceph_evict_inode(struct inode *inode); 901cfa6d412SAl Viro extern void ceph_free_inode(struct inode *inode); 902de57606cSSage Weil 903de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 904de57606cSSage Weil struct ceph_vino vino); 905de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 906de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 907de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 908de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 9099bbeab41SArnd Bergmann u64 time_warp_seq, struct timespec64 *ctime, 9109bbeab41SArnd Bergmann struct timespec64 *mtime, 9119bbeab41SArnd Bergmann struct timespec64 *atime); 912de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 913f5a03b08SJeff Layton struct ceph_mds_request *req); 914de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 915de57606cSSage Weil struct ceph_mds_session *session); 916de57606cSSage Weil 917de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 918de57606cSSage Weil 919efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 920b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 9213c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 9223c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 9233c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 9243e1d0452SYan, Zheng extern void ceph_async_iput(struct inode *inode); 925de57606cSSage Weil 92601deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 92701deead0SYan, Zheng int mask, bool force); 92801deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 92901deead0SYan, Zheng { 93001deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 93101deead0SYan, Zheng } 93210556cb2SAl Viro extern int ceph_permission(struct inode *inode, int mask); 933a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 934de57606cSSage Weil extern int ceph_setattr(struct dentry *dentry, struct iattr *attr); 935a528d35eSDavid Howells extern int ceph_getattr(const struct path *path, struct kstat *stat, 936a528d35eSDavid Howells u32 request_mask, unsigned int flags); 937de57606cSSage Weil 938de57606cSSage Weil /* xattr.c */ 939a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 9407221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 941de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 94212fe3ddaSLuis Henriques extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); 943de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 944b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 945de57606cSSage Weil 9465c31e92dSYan, Zheng struct ceph_acl_sec_ctx { 9475c31e92dSYan, Zheng #ifdef CONFIG_CEPH_FS_POSIX_ACL 9485c31e92dSYan, Zheng void *default_acl; 9495c31e92dSYan, Zheng void *acl; 9505c31e92dSYan, Zheng #endif 951ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 952ac6713ccSYan, Zheng void *sec_ctx; 953ac6713ccSYan, Zheng u32 sec_ctxlen; 954ac6713ccSYan, Zheng #endif 9555c31e92dSYan, Zheng struct ceph_pagelist *pagelist; 9565c31e92dSYan, Zheng }; 9575c31e92dSYan, Zheng 958315f2408SYan, Zheng #ifdef CONFIG_SECURITY 959315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 960315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 961315f2408SYan, Zheng #else 962315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 963315f2408SYan, Zheng { 964315f2408SYan, Zheng return false; 965315f2408SYan, Zheng } 966315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 967315f2408SYan, Zheng { 968315f2408SYan, Zheng return false; 969315f2408SYan, Zheng } 970315f2408SYan, Zheng #endif 971315f2408SYan, Zheng 972ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 973ac6713ccSYan, Zheng extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 974ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx); 975ac6713ccSYan, Zheng extern void ceph_security_invalidate_secctx(struct inode *inode); 976ac6713ccSYan, Zheng #else 977ac6713ccSYan, Zheng static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 978ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx) 979ac6713ccSYan, Zheng { 980ac6713ccSYan, Zheng return 0; 981ac6713ccSYan, Zheng } 982ac6713ccSYan, Zheng static inline void ceph_security_invalidate_secctx(struct inode *inode) 983ac6713ccSYan, Zheng { 984ac6713ccSYan, Zheng } 985ac6713ccSYan, Zheng #endif 9867221fe4cSGuangliang Zhao 9875c31e92dSYan, Zheng void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); 9887221fe4cSGuangliang Zhao 9895c31e92dSYan, Zheng /* acl.c */ 9907221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 9917221fe4cSGuangliang Zhao 9927221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 99372466d0bSSage Weil int ceph_set_acl(struct inode *inode, struct posix_acl *acl, int type); 994b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 9955c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 9965c31e92dSYan, Zheng void ceph_init_inode_acls(struct inode *inode, 9975c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 998c969d9bfSGuangliang Zhao 999c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 1000c969d9bfSGuangliang Zhao { 1001c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 1002c969d9bfSGuangliang Zhao } 10037221fe4cSGuangliang Zhao 10047221fe4cSGuangliang Zhao #else 10057221fe4cSGuangliang Zhao 10067221fe4cSGuangliang Zhao #define ceph_get_acl NULL 100772466d0bSSage Weil #define ceph_set_acl NULL 10087221fe4cSGuangliang Zhao 1009b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 10105c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 10117221fe4cSGuangliang Zhao { 10127221fe4cSGuangliang Zhao return 0; 10137221fe4cSGuangliang Zhao } 1014b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 10155c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 1016b1ee94aaSYan, Zheng { 1017b1ee94aaSYan, Zheng } 10187221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 10197221fe4cSGuangliang Zhao { 10207221fe4cSGuangliang Zhao return 0; 10217221fe4cSGuangliang Zhao } 10227221fe4cSGuangliang Zhao 10237221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 10247221fe4cSGuangliang Zhao { 10257221fe4cSGuangliang Zhao } 10267221fe4cSGuangliang Zhao 10277221fe4cSGuangliang Zhao #endif 10287221fe4cSGuangliang Zhao 1029de57606cSSage Weil /* caps.c */ 1030de57606cSSage Weil extern const char *ceph_cap_string(int c); 1031de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 1032de57606cSSage Weil struct ceph_msg *msg); 1033d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 1034d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 1035d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 1036de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 1037de57606cSSage Weil int fmode, unsigned issued, unsigned wanted, 1038de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 1039d9df2783SYan, Zheng struct ceph_cap **new_cap); 1040a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1041d6e47819SYan, Zheng extern void __ceph_remove_caps(struct ceph_inode_info *ci); 104237151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 104337151668SYehuda Sadeh struct ceph_cap *cap); 10449215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 1045de57606cSSage Weil 1046f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 104702c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 104802c24a82SJosef Bacik int datasync); 1049e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 1050e548e9b9SYan, Zheng struct ceph_mds_session *session); 1051de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 1052de57606cSSage Weil struct ceph_mds_session *session); 10532bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 10542bc50259SGreg Farnum int mds); 1055de57606cSSage Weil extern int ceph_get_cap_mds(struct inode *inode); 1056de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 1057de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 1058de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 1059de57606cSSage Weil struct ceph_snap_context *snapc); 1060ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 10610e294387SYan, Zheng struct ceph_mds_session **psession); 1062efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 1063de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 1064de57606cSSage Weil struct ceph_mds_session *session); 1065afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 1066afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 10676ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 1068de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 1069de57606cSSage Weil int mds, int drop, int unless, int force); 1070de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 1071ca6c8ae0SJeff Layton struct inode *dir, 1072de57606cSSage Weil int mds, int drop, int unless); 1073de57606cSSage Weil 10745e3ded1bSYan, Zheng extern int ceph_get_caps(struct file *filp, int need, int want, 107531c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 10765e3ded1bSYan, Zheng extern int ceph_try_get_caps(struct inode *inode, 10772ee9dd95SLuis Henriques int need, int want, bool nonblock, int *got); 1078de57606cSSage Weil 1079de57606cSSage Weil /* for counting open files by mode */ 1080774a6a11SYan, Zheng extern void __ceph_get_fmode(struct ceph_inode_info *ci, int mode); 1081de57606cSSage Weil extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode); 1082de57606cSSage Weil 1083de57606cSSage Weil /* addr.c */ 1084de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1085de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 108610183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 10875e3ded1bSYan, Zheng extern int ceph_pool_perm_check(struct inode *inode, int need); 108810183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 1089de57606cSSage Weil 1090de57606cSSage Weil /* file.c */ 1091de57606cSSage Weil extern const struct file_operations ceph_file_fops; 10929e0eb85dSAlex Elder 109377310320SYan, Zheng extern int ceph_renew_caps(struct inode *inode); 1094de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 10955ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 109644907d79SAl Viro struct file *file, unsigned flags, umode_t mode); 1097de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 109831c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 109931c542a1SYan, Zheng char *data, size_t len); 110055f2a045SIlya Dryomov 1101de57606cSSage Weil /* dir.c */ 1102de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 110338c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1104de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 110538c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 110618fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1107de57606cSSage Weil 1108f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1109de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1110468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 1111468640e3SSage Weil struct dentry *dentry, int err); 1112de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1113de57606cSSage Weil struct dentry *dentry, int err); 1114de57606cSSage Weil 111537c4efc1SYan, Zheng extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); 111637c4efc1SYan, Zheng extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); 111781a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 111837c4efc1SYan, Zheng extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); 1119e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1120fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1121de57606cSSage Weil 1122de57606cSSage Weil /* ioctl.c */ 1123de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1124de57606cSSage Weil 1125de57606cSSage Weil /* export.c */ 1126de57606cSSage Weil extern const struct export_operations ceph_export_ops; 11273886274aSLuis Henriques struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); 1128de57606cSSage Weil 112940819f6fSGreg Farnum /* locks.c */ 1130eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 113140819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 113240819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 113340819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 113439be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 113539be95e9SJim Schutt struct ceph_filelock *flocks, 113639be95e9SJim Schutt int num_fcntl_locks, 113739be95e9SJim Schutt int num_flock_locks); 113839be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 113939be95e9SJim Schutt struct ceph_pagelist *pagelist, 114039be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 114140819f6fSGreg Farnum 11423d14c5d2SYehuda Sadeh /* debugfs.c */ 11431a829ff2SGreg Kroah-Hartman extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); 11443d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 11453d14c5d2SYehuda Sadeh 1146fb18a575SLuis Henriques /* quota.c */ 1147d557c48dSLuis Henriques static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) 1148d557c48dSLuis Henriques { 1149d557c48dSLuis Henriques return ci->i_max_files || ci->i_max_bytes; 1150d557c48dSLuis Henriques } 1151d557c48dSLuis Henriques 1152d557c48dSLuis Henriques extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); 1153d557c48dSLuis Henriques 1154d557c48dSLuis Henriques static inline void __ceph_update_quota(struct ceph_inode_info *ci, 1155d557c48dSLuis Henriques u64 max_bytes, u64 max_files) 1156d557c48dSLuis Henriques { 1157d557c48dSLuis Henriques bool had_quota, has_quota; 1158d557c48dSLuis Henriques had_quota = __ceph_has_any_quota(ci); 1159d557c48dSLuis Henriques ci->i_max_bytes = max_bytes; 1160d557c48dSLuis Henriques ci->i_max_files = max_files; 1161d557c48dSLuis Henriques has_quota = __ceph_has_any_quota(ci); 1162d557c48dSLuis Henriques 1163d557c48dSLuis Henriques if (had_quota != has_quota) 1164d557c48dSLuis Henriques ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); 1165d557c48dSLuis Henriques } 1166d557c48dSLuis Henriques 1167fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc, 1168fb18a575SLuis Henriques struct ceph_mds_session *session, 1169fb18a575SLuis Henriques struct ceph_msg *msg); 1170b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); 1171cafe21a4SLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); 11722b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, 11732b83845fSLuis Henriques loff_t newlen); 11741ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, 11751ab302a0SLuis Henriques loff_t newlen); 11769122eed5SLuis Henriques extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, 11779122eed5SLuis Henriques struct kstatfs *buf); 11780c44a8e0SLuis Henriques extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); 1179fb18a575SLuis Henriques 1180de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1181