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> 19668959a5SJeff Layton #include <linux/security.h> 20de57606cSSage Weil 213d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 22de57606cSSage Weil 2399ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 2499ccbd22SMilosz Tanski #include <linux/fscache.h> 2599ccbd22SMilosz Tanski #endif 2699ccbd22SMilosz Tanski 27de57606cSSage Weil /* f_type in struct statfs */ 28de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400 29de57606cSSage Weil 30de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 31de57606cSSage Weil * large volume sizes on 32-bit machines. */ 3292a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 33de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 34de57606cSSage Weil 350b98acd6SIlya Dryomov #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blocklisted */ 363d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 373d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 383d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 39ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 40a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 4199ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 4210183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 43e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 449122eed5SLuis Henriques #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ 45ea4cdc54SLuis Henriques #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ 462ccb4546SJeff Layton #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ 476a259382SSage Weil 486f9718feSLuis Henriques #define CEPH_MOUNT_OPT_DEFAULT \ 496f9718feSLuis Henriques (CEPH_MOUNT_OPT_DCACHE | \ 506f9718feSLuis Henriques CEPH_MOUNT_OPT_NOCOPYFROM) 51de57606cSSage Weil 523d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 532ccb4546SJeff Layton (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt 542ccb4546SJeff Layton #define ceph_clear_mount_opt(fsc, opt) \ 552ccb4546SJeff Layton (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt 563d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 573d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 58de57606cSSage Weil 59aa187926SYan, Zheng /* max size of osd read request, limited by libceph */ 60aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN 6195cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size. 6295cca2b4SYan, Zheng * CEPH_MSG_MAX_DATA_LEN should be small enough. */ 6395cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN 647c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 653d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 663d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 673d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 68de57606cSSage Weil 694214fb15SYan, Zheng /* 704214fb15SYan, Zheng * Delay telling the MDS we no longer want caps, in case we reopen 714214fb15SYan, Zheng * the file. Delay a minimum amount of time, even if we send a cap 724214fb15SYan, Zheng * message for some other reason. Otherwise, take the oppotunity to 734214fb15SYan, Zheng * update the mds to avoid sending another message later. 744214fb15SYan, Zheng */ 754214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ 764214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ 774214fb15SYan, Zheng 783d14c5d2SYehuda Sadeh struct ceph_mount_options { 79ad8c28a9SJeff Layton unsigned int flags; 803d14c5d2SYehuda Sadeh 81ad8c28a9SJeff Layton unsigned int wsize; /* max write size */ 82ad8c28a9SJeff Layton unsigned int rsize; /* max read size */ 83ad8c28a9SJeff Layton unsigned int rasize; /* max readahead */ 84ad8c28a9SJeff Layton unsigned int congestion_kb; /* max writeback in flight */ 85ad8c28a9SJeff Layton unsigned int caps_wanted_delay_min, caps_wanted_delay_max; 86fe33032dSYan, Zheng int caps_max; 87ad8c28a9SJeff Layton unsigned int max_readdir; /* max readdir result (entries) */ 88ad8c28a9SJeff Layton unsigned int max_readdir_bytes; /* max readdir result (bytes) */ 893d14c5d2SYehuda Sadeh 903d14c5d2SYehuda Sadeh /* 913d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 923d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 933d14c5d2SYehuda Sadeh */ 943d14c5d2SYehuda Sadeh 95de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 96430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 97b27a939eSIlya Dryomov char *server_path; /* default NULL (means "/") */ 981d8f8360SYan, Zheng char *fscache_uniq; /* default NULL */ 99de57606cSSage Weil }; 100de57606cSSage Weil 1013d14c5d2SYehuda Sadeh struct ceph_fs_client { 102de57606cSSage Weil struct super_block *sb; 103de57606cSSage Weil 10418f473b3SXiubo Li struct list_head metric_wakeup; 10518f473b3SXiubo Li 1063d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 1073d14c5d2SYehuda Sadeh struct ceph_client *client; 1083d14c5d2SYehuda Sadeh 109aa5c7910SJeff Layton int mount_state; 11081f148a9SYan, Zheng 1110b98acd6SIlya Dryomov bool blocklisted; 112131d7eb4SYan, Zheng 11378beb0ffSLuis Henriques bool have_copy_from2; 11478beb0ffSLuis Henriques 11581f148a9SYan, Zheng u32 filp_gen; 116719784baSChengguang Xu loff_t max_file_size; 11785ccce43SSage Weil 1183d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 119de57606cSSage Weil 120e3ec8d68SYan, Zheng atomic_long_t writeback_count; 121e3ec8d68SYan, Zheng 1221cf89a8dSYan, Zheng struct workqueue_struct *inode_wq; 123e3ec8d68SYan, Zheng struct workqueue_struct *cap_wq; 124de57606cSSage Weil 1250743304dSSage Weil #ifdef CONFIG_DEBUG_FS 1263d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1272baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 12806edf046SSage Weil struct dentry *debugfs_bdi; 1293d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 130f9009efaSXiubo Li struct dentry *debugfs_metric; 131247b1f19SXiubo Li struct dentry *debugfs_status; 13214ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 1330743304dSSage Weil #endif 13499ccbd22SMilosz Tanski 13599ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 13699ccbd22SMilosz Tanski struct fscache_cookie *fscache; 13799ccbd22SMilosz Tanski #endif 138de57606cSSage Weil }; 139de57606cSSage Weil 1403d14c5d2SYehuda Sadeh 141de57606cSSage Weil /* 142de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 143de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 144de57606cSSage Weil * and write data. For any given inode, we should have one or more 145de57606cSSage Weil * capabilities, one issued by each metadata server, and our 146de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 147de57606cSSage Weil * 148de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 149de57606cSSage Weil * session capability lists. 150de57606cSSage Weil */ 151de57606cSSage Weil struct ceph_cap { 152de57606cSSage Weil struct ceph_inode_info *ci; 153de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 154de57606cSSage Weil struct ceph_mds_session *session; 155de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 156de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 157745a8e3bSYan, Zheng union { 158745a8e3bSYan, Zheng /* in-use caps */ 159745a8e3bSYan, Zheng struct { 160de57606cSSage Weil int issued; /* latest, from the mds */ 161745a8e3bSYan, Zheng int implemented; /* implemented superset of 162745a8e3bSYan, Zheng issued (for revocation) */ 163c74d79afSJeff Layton int mds; /* mds index for this cap */ 164c74d79afSJeff Layton int mds_wanted; /* caps wanted from this mds */ 165745a8e3bSYan, Zheng }; 166745a8e3bSYan, Zheng /* caps to release */ 167745a8e3bSYan, Zheng struct { 168745a8e3bSYan, Zheng u64 cap_ino; 169745a8e3bSYan, Zheng int queue_release; 170745a8e3bSYan, Zheng }; 171745a8e3bSYan, Zheng }; 172685f9a5dSSage Weil u32 seq, issue_seq, mseq; 173685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 174de57606cSSage Weil unsigned long last_used; 175de57606cSSage Weil struct list_head caps_item; 176de57606cSSage Weil }; 177de57606cSSage Weil 178a0d93e32SYan, Zheng #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ 179a0d93e32SYan, Zheng #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ 180a0d93e32SYan, Zheng #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ 181de57606cSSage Weil 1820e294387SYan, Zheng struct ceph_cap_flush { 1830e294387SYan, Zheng u64 tid; 1840e294387SYan, Zheng int caps; /* 0 means capsnap */ 185c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 1860e294387SYan, Zheng struct list_head g_list; // global 1870e294387SYan, Zheng struct list_head i_list; // per inode 1880e294387SYan, Zheng }; 1890e294387SYan, Zheng 190de57606cSSage Weil /* 191de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 192de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 193de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 194de57606cSSage Weil */ 195de57606cSSage Weil struct ceph_cap_snap { 196805692d0SElena Reshetova refcount_t nref; 1970e294387SYan, Zheng struct list_head ci_item; 198de57606cSSage Weil 1990e294387SYan, Zheng struct ceph_cap_flush cap_flush; 2000e294387SYan, Zheng 2010e294387SYan, Zheng u64 follows; 202de57606cSSage Weil int issued, dirty; 203de57606cSSage Weil struct ceph_snap_context *context; 204de57606cSSage Weil 2055706b27dSAl Viro umode_t mode; 20605cb11c1SEric W. Biederman kuid_t uid; 20705cb11c1SEric W. Biederman kgid_t gid; 208de57606cSSage Weil 2094a625be4SSage Weil struct ceph_buffer *xattr_blob; 210de57606cSSage Weil u64 xattr_version; 211de57606cSSage Weil 212de57606cSSage Weil u64 size; 213176c77c9SJeff Layton u64 change_attr; 214ec62b894SJeff Layton struct timespec64 mtime, atime, ctime, btime; 215de57606cSSage Weil u64 time_warp_seq; 2165f743e45SYan, Zheng u64 truncate_size; 2175f743e45SYan, Zheng u32 truncate_seq; 218de57606cSSage Weil int writing; /* a sync write is still in progress */ 219de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 220e20d258dSYan, Zheng bool inline_data; 22186056090SYan, Zheng bool need_flush; 222de57606cSSage Weil }; 223de57606cSSage Weil 224de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 225de57606cSSage Weil { 226805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 2274a625be4SSage Weil if (capsnap->xattr_blob) 2284a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 229de57606cSSage Weil kfree(capsnap); 230de57606cSSage Weil } 2314a625be4SSage Weil } 232de57606cSSage Weil 233de57606cSSage Weil /* 234de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 235de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 236de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 237de57606cSSage Weil * 238de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 239de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 240de57606cSSage Weil */ 241de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 242de57606cSSage Weil 243de57606cSSage Weil struct ceph_inode_frag { 244de57606cSSage Weil struct rb_node node; 245de57606cSSage Weil 246de57606cSSage Weil /* fragtree state */ 247de57606cSSage Weil u32 frag; 248de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 249de57606cSSage Weil 250de57606cSSage Weil /* delegation and replication info */ 251de57606cSSage Weil int mds; /* -1 if same authority as parent */ 252de57606cSSage Weil int ndist; /* >0 if replicated */ 253de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 254de57606cSSage Weil }; 255de57606cSSage Weil 256de57606cSSage Weil /* 257de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 258de57606cSSage Weil * at which point we parse them into an rbtree. 259de57606cSSage Weil */ 260de57606cSSage Weil struct ceph_inode_xattr { 261de57606cSSage Weil struct rb_node node; 262de57606cSSage Weil 263de57606cSSage Weil const char *name; 264de57606cSSage Weil int name_len; 265de57606cSSage Weil const char *val; 266de57606cSSage Weil int val_len; 267de57606cSSage Weil int dirty; 268de57606cSSage Weil 269de57606cSSage Weil int should_free_name; 270de57606cSSage Weil int should_free_val; 271de57606cSSage Weil }; 272de57606cSSage Weil 2733d14c5d2SYehuda Sadeh /* 2743d14c5d2SYehuda Sadeh * Ceph dentry state 2753d14c5d2SYehuda Sadeh */ 2763d14c5d2SYehuda Sadeh struct ceph_dentry_info { 27737c4efc1SYan, Zheng struct dentry *dentry; 2783d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 27937c4efc1SYan, Zheng struct list_head lease_list; 28037c4efc1SYan, Zheng unsigned flags; 28197aeb6bfSYan, Zheng int lease_shared_gen; 28297aeb6bfSYan, Zheng u32 lease_gen; 2833d14c5d2SYehuda Sadeh u32 lease_seq; 2843d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2859b16f03cSMiklos Szeredi unsigned long time; 2863d14c5d2SYehuda Sadeh u64 offset; 2873d14c5d2SYehuda Sadeh }; 2883d14c5d2SYehuda Sadeh 28937c4efc1SYan, Zheng #define CEPH_DENTRY_REFERENCED 1 29037c4efc1SYan, Zheng #define CEPH_DENTRY_LEASE_LIST 2 29137c4efc1SYan, Zheng #define CEPH_DENTRY_SHRINK_LIST 4 292f5e17aedSJeff Layton #define CEPH_DENTRY_PRIMARY_LINK 8 29337c4efc1SYan, Zheng 294de57606cSSage Weil struct ceph_inode_xattrs_info { 295de57606cSSage Weil /* 296de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 297de57606cSSage Weil * this until someone actually calls getxattr, etc. 298de57606cSSage Weil * 299de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 300de57606cSSage Weil * NULL means we don't know. 301de57606cSSage Weil */ 302de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 303de57606cSSage Weil 304de57606cSSage Weil struct rb_root index; 305de57606cSSage Weil bool dirty; 306de57606cSSage Weil int count; 307de57606cSSage Weil int names_size; 308de57606cSSage Weil int vals_size; 309de57606cSSage Weil u64 version, index_version; 310de57606cSSage Weil }; 311de57606cSSage Weil 312de57606cSSage Weil /* 313de57606cSSage Weil * Ceph inode. 314de57606cSSage Weil */ 315de57606cSSage Weil struct ceph_inode_info { 316de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 317de57606cSSage Weil 318be655596SSage Weil spinlock_t i_ceph_lock; 319be655596SSage Weil 320de57606cSSage Weil u64 i_version; 32131c542a1SYan, Zheng u64 i_inline_version; 322de57606cSSage Weil u32 i_time_warp_seq; 323de57606cSSage Weil 324891f3f5aSJeff Layton unsigned long i_ceph_flags; 325fdd4e158SYan, Zheng atomic64_t i_release_count; 326fdd4e158SYan, Zheng atomic64_t i_ordered_count; 327fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 328de57606cSSage Weil 3296c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 330de57606cSSage Weil struct ceph_file_layout i_layout; 331785892feSJeff Layton struct ceph_file_layout i_cached_layout; // for async creates 332de57606cSSage Weil char *i_symlink; 333de57606cSSage Weil 334de57606cSSage Weil /* for dirs */ 3359bbeab41SArnd Bergmann struct timespec64 i_rctime; 336de57606cSSage Weil u64 i_rbytes, i_rfiles, i_rsubdirs; 337de57606cSSage Weil u64 i_files, i_subdirs; 338de57606cSSage Weil 339fb18a575SLuis Henriques /* quotas */ 340fb18a575SLuis Henriques u64 i_max_bytes, i_max_files; 341fb18a575SLuis Henriques 34208796873SYan, Zheng s32 i_dir_pin; 34308796873SYan, Zheng 344de57606cSSage Weil struct rb_root i_fragtree; 3451b1bc16dSYan, Zheng int i_fragtree_nsplits; 346de57606cSSage Weil struct mutex i_fragtree_mutex; 347de57606cSSage Weil 348de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 349de57606cSSage Weil 350be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 351de57606cSSage Weil * s_mutex. */ 352de57606cSSage Weil struct rb_root i_caps; /* cap list */ 353de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 354de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 3551cf03a68SJeff Layton 3561cf03a68SJeff Layton /* 357f1f565a2SRandy Dunlap * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty 3581cf03a68SJeff Layton * is protected by the mdsc->cap_dirty_lock, but each individual item 3591cf03a68SJeff Layton * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty 3601cf03a68SJeff Layton * requires the mdsc->cap_dirty_lock. List presence for an item can 3611cf03a68SJeff Layton * be tested under the i_ceph_lock. Changing anything requires both. 3624fb5dda3SJeff Layton */ 3631cf03a68SJeff Layton struct list_head i_dirty_item; 3641cf03a68SJeff Layton 3651cf03a68SJeff Layton /* 366829ad4dbSJeff Layton * Link to session's s_cap_flushing list. Protected in a similar 367829ad4dbSJeff Layton * fashion to i_dirty_item, but also by the s_mutex for changes. The 368829ad4dbSJeff Layton * s_cap_flushing list can be walked while holding either the s_mutex 369829ad4dbSJeff Layton * or msdc->cap_dirty_lock. List presence can also be checked while 370829ad4dbSJeff Layton * holding the i_ceph_lock for this inode. 3711cf03a68SJeff Layton */ 3721cf03a68SJeff Layton struct list_head i_flushing_item; 3731cf03a68SJeff Layton 374de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 375de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 376de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 377f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 378e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 379de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 380de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 381de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 382de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 383de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3847d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3857d8cb26dSSage Weil dirty|flushing caps */ 386de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 387de57606cSSage Weil 388719a2514SYan, Zheng unsigned long i_last_rd; 389719a2514SYan, Zheng unsigned long i_last_wr; 390774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 391de57606cSSage Weil 392b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 393de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 394de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 395de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 396de57606cSSage Weil 397de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 398de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 399de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 400de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 401de57606cSSage Weil 402de57606cSSage Weil /* held references to caps */ 403de57606cSSage Weil int i_pin_ref; 404f85122afSJeff Layton int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; 405de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 40689aa5930SYan, Zheng atomic_t i_filelock_ref; 40797aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 408cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 409de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 410de57606cSSage Weil 411de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 41268cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 413de57606cSSage Weil spinlock_t i_unsafe_lock; 414de57606cSSage Weil 41575c9627eSYan, Zheng union { 416de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 41775c9627eSYan, Zheng struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ 41875c9627eSYan, Zheng }; 419de57606cSSage Weil int i_snap_realm_counter; /* snap realm (if caps) */ 420de57606cSSage Weil struct list_head i_snap_realm_item; 421de57606cSSage Weil struct list_head i_snap_flush_item; 422245ce991SJeff Layton struct timespec64 i_btime; 423193e7b37SDavid Disseldorp struct timespec64 i_snap_btime; 424de57606cSSage Weil 4251cf89a8dSYan, Zheng struct work_struct i_work; 4261cf89a8dSYan, Zheng unsigned long i_work_mask; 427de57606cSSage Weil 42899ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 42999ccbd22SMilosz Tanski struct fscache_cookie *fscache; 430f7f7e7a0SYan, Zheng u32 i_fscache_gen; 43199ccbd22SMilosz Tanski #endif 432f4b97866SYan, Zheng errseq_t i_meta_err; 433f4b97866SYan, Zheng 434de57606cSSage Weil struct inode vfs_inode; /* at end */ 435de57606cSSage Weil }; 436de57606cSSage Weil 437721d5c13SJeff Layton static inline struct ceph_inode_info * 438721d5c13SJeff Layton ceph_inode(const struct inode *inode) 439de57606cSSage Weil { 440fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 441de57606cSSage Weil } 442de57606cSSage Weil 443721d5c13SJeff Layton static inline struct ceph_fs_client * 444721d5c13SJeff Layton ceph_inode_to_client(const struct inode *inode) 445ad1fee96SYehuda Sadeh { 446ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 447ad1fee96SYehuda Sadeh } 448ad1fee96SYehuda Sadeh 449721d5c13SJeff Layton static inline struct ceph_fs_client * 450721d5c13SJeff Layton ceph_sb_to_client(const struct super_block *sb) 451ad1fee96SYehuda Sadeh { 452ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 453ad1fee96SYehuda Sadeh } 454ad1fee96SYehuda Sadeh 4552678da88SXiubo Li static inline struct ceph_mds_client * 4562678da88SXiubo Li ceph_sb_to_mdsc(const struct super_block *sb) 4572678da88SXiubo Li { 4582678da88SXiubo Li return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc; 4592678da88SXiubo Li } 4602678da88SXiubo Li 461721d5c13SJeff Layton static inline struct ceph_vino 462721d5c13SJeff Layton ceph_vino(const struct inode *inode) 4633d14c5d2SYehuda Sadeh { 4643d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4653d14c5d2SYehuda Sadeh } 4663d14c5d2SYehuda Sadeh 467ebce3eb2SJeff Layton static inline u32 ceph_ino_to_ino32(u64 vino) 468ad1fee96SYehuda Sadeh { 4693310f754SAmon Ott u32 ino = vino & 0xffffffff; 4703310f754SAmon Ott ino ^= vino >> 32; 471ad1fee96SYehuda Sadeh if (!ino) 472a661fc56SAmon Ott ino = 2; 473ad1fee96SYehuda Sadeh return ino; 474ad1fee96SYehuda Sadeh } 475ad1fee96SYehuda Sadeh 476ad1fee96SYehuda Sadeh /* 477ebce3eb2SJeff Layton * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on 478ebce3eb2SJeff Layton * some arches. We generally do not use this value inside the ceph driver, but 479ebce3eb2SJeff Layton * we do want to set it to something, so that generic vfs code has an 480ebce3eb2SJeff Layton * appropriate value for tracepoints and the like. 4813d14c5d2SYehuda Sadeh */ 482ebce3eb2SJeff Layton static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino) 4833d14c5d2SYehuda Sadeh { 484ebce3eb2SJeff Layton if (sizeof(ino_t) == sizeof(u32)) 4853310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4863310f754SAmon Ott return (ino_t)vino.ino; 4873d14c5d2SYehuda Sadeh } 4883d14c5d2SYehuda Sadeh 4893d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4903d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4913d14c5d2SYehuda Sadeh 4923d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4933d14c5d2SYehuda Sadeh { 4943d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4953d14c5d2SYehuda Sadeh } 496ebce3eb2SJeff Layton 4973d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4983d14c5d2SYehuda Sadeh { 4993d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 5003d14c5d2SYehuda Sadeh } 5013d14c5d2SYehuda Sadeh 502ebce3eb2SJeff Layton /** 503ebce3eb2SJeff Layton * ceph_present_ino - format an inode number for presentation to userland 504ebce3eb2SJeff Layton * @sb: superblock where the inode lives 505ebce3eb2SJeff Layton * @ino: inode number to (possibly) convert 506ebce3eb2SJeff Layton * 507ebce3eb2SJeff Layton * If the user mounted with the ino32 option, then the 64-bit value needs 508ebce3eb2SJeff Layton * to be converted to something that can fit inside 32 bits. Note that 509ebce3eb2SJeff Layton * internal kernel code never uses this value, so this is entirely for 510ebce3eb2SJeff Layton * userland consumption. 511ebce3eb2SJeff Layton */ 512ebce3eb2SJeff Layton static inline u64 ceph_present_ino(struct super_block *sb, u64 ino) 513ebce3eb2SJeff Layton { 514ebce3eb2SJeff Layton if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))) 515ebce3eb2SJeff Layton return ceph_ino_to_ino32(ino); 516ebce3eb2SJeff Layton return ino; 517ebce3eb2SJeff Layton } 518ebce3eb2SJeff Layton 519ebce3eb2SJeff Layton static inline u64 ceph_present_inode(struct inode *inode) 520ebce3eb2SJeff Layton { 521ebce3eb2SJeff Layton return ceph_present_ino(inode->i_sb, ceph_ino(inode)); 522ebce3eb2SJeff Layton } 523ebce3eb2SJeff Layton 5243d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 5253d14c5d2SYehuda Sadeh { 5263d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 5273d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 5283d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 5293d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 5303d14c5d2SYehuda Sadeh } 5313d14c5d2SYehuda Sadeh 532ebce3eb2SJeff Layton 5333d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 5343d14c5d2SYehuda Sadeh struct ceph_vino vino) 5353d14c5d2SYehuda Sadeh { 536ebce3eb2SJeff Layton /* 537ebce3eb2SJeff Layton * NB: The hashval will be run through the fs/inode.c hash function 538ebce3eb2SJeff Layton * anyway, so there is no need to squash the inode number down to 539ebce3eb2SJeff Layton * 32-bits first. Just use low-order bits on arches with 32-bit long. 540ebce3eb2SJeff Layton */ 541ebce3eb2SJeff Layton return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino); 5423d14c5d2SYehuda Sadeh } 5433d14c5d2SYehuda Sadeh 5443d14c5d2SYehuda Sadeh 5453d14c5d2SYehuda Sadeh /* 5463d14c5d2SYehuda Sadeh * Ceph inode. 5473d14c5d2SYehuda Sadeh */ 54810183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 54910183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 550daca8bdaSJeff Layton #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ 551daca8bdaSJeff Layton #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ 552daca8bdaSJeff Layton #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ 553daca8bdaSJeff Layton #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ 554c0e385b1SYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ 555c0e385b1SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ 556c0e385b1SYan, Zheng #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ 557c0e385b1SYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ 558c0e385b1SYan, Zheng #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ 559c0e385b1SYan, Zheng #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ 560891f3f5aSJeff Layton #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) 56126544c62SJeff Layton 56226544c62SJeff Layton /* 5631cf89a8dSYan, Zheng * Masks of ceph inode work. 5641cf89a8dSYan, Zheng */ 5651cf89a8dSYan, Zheng #define CEPH_I_WORK_WRITEBACK 0 /* writeback */ 5661cf89a8dSYan, Zheng #define CEPH_I_WORK_INVALIDATE_PAGES 1 /* invalidate pages */ 5671cf89a8dSYan, Zheng #define CEPH_I_WORK_VMTRUNCATE 2 /* vmtruncate */ 5681cf89a8dSYan, Zheng 5691cf89a8dSYan, Zheng /* 57026544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 57126544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 57226544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 57326544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 57426544c62SJeff Layton * worried if a few slip through in either direction. 57526544c62SJeff Layton */ 57626544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 57726544c62SJeff Layton { 57826544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 57926544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 58026544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 58126544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 58226544c62SJeff Layton } 58326544c62SJeff Layton } 58426544c62SJeff Layton 58526544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 58626544c62SJeff Layton { 58726544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 58826544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 58926544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 59026544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 59126544c62SJeff Layton } 59226544c62SJeff Layton } 5933d14c5d2SYehuda Sadeh 5942f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 595fdd4e158SYan, Zheng long long release_count, 596fdd4e158SYan, Zheng long long ordered_count) 597de57606cSSage Weil { 59874960773SAndrea Parri /* 59974960773SAndrea Parri * Makes sure operations that setup readdir cache (update page 60074960773SAndrea Parri * cache and i_size) are strongly ordered w.r.t. the following 60174960773SAndrea Parri * atomic64_set() operations. 60274960773SAndrea Parri */ 60374960773SAndrea Parri smp_mb(); 604fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 605fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 606de57606cSSage Weil } 607de57606cSSage Weil 6082f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 609de57606cSSage Weil { 610fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 611fdd4e158SYan, Zheng } 612fdd4e158SYan, Zheng 613fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 614fdd4e158SYan, Zheng { 615fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 616de57606cSSage Weil } 617de57606cSSage Weil 6182f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 619de57606cSSage Weil { 620fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 621fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 6222f276c51SYan, Zheng } 623de57606cSSage Weil 62470db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 62570db4f36SYan, Zheng { 626fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 627fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 628fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 629fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 63070db4f36SYan, Zheng } 63170db4f36SYan, Zheng 6322f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 6332f276c51SYan, Zheng { 6342f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 6352f276c51SYan, Zheng } 6362f276c51SYan, Zheng 63770db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 6382f276c51SYan, Zheng { 639fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 640de57606cSSage Weil } 641de57606cSSage Weil 64270db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 64370db4f36SYan, Zheng { 644fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 645fdd4e158SYan, Zheng smp_rmb(); 64670db4f36SYan, Zheng return ret; 64770db4f36SYan, Zheng } 648de57606cSSage Weil 649de57606cSSage Weil /* find a specific frag @f */ 650de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 651de57606cSSage Weil u32 f); 652de57606cSSage Weil 653de57606cSSage Weil /* 654de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 655de57606cSSage Weil * exists 656de57606cSSage Weil */ 657de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 658de57606cSSage Weil struct ceph_inode_frag *pfrag, 659de57606cSSage Weil int *found); 660de57606cSSage Weil 6611e9c2eb6SYan, Zheng static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) 662de57606cSSage Weil { 663de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 664de57606cSSage Weil } 665de57606cSSage Weil 666de57606cSSage Weil /* 667de57606cSSage Weil * caps helpers 668de57606cSSage Weil */ 669de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 670de57606cSSage Weil { 671de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 672de57606cSSage Weil } 673de57606cSSage Weil 674de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 675de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 6761af16d54SXiubo Li extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, 6771af16d54SXiubo Li int t); 678de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 679de57606cSSage Weil struct ceph_cap *cap); 680de57606cSSage Weil 681de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 682de57606cSSage Weil { 683de57606cSSage Weil int issued; 684be655596SSage Weil spin_lock(&ci->i_ceph_lock); 685de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 686be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 687de57606cSSage Weil return issued; 688de57606cSSage Weil } 689de57606cSSage Weil 6901af16d54SXiubo Li static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, 6911af16d54SXiubo Li int mask, int touch) 692de57606cSSage Weil { 693de57606cSSage Weil int r; 694be655596SSage Weil spin_lock(&ci->i_ceph_lock); 6951af16d54SXiubo Li r = __ceph_caps_issued_mask_metric(ci, mask, touch); 696be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 697de57606cSSage Weil return r; 698de57606cSSage Weil } 699de57606cSSage Weil 700de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 701de57606cSSage Weil { 702de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 703de57606cSSage Weil } 704f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 705f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 706f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 707f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 708de57606cSSage Weil 7099563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 7109563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 711de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 712de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 713de57606cSSage Weil 714719a2514SYan, Zheng static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) 715719a2514SYan, Zheng { 716719a2514SYan, Zheng return ci->i_nr_by_mode[0]; 717719a2514SYan, Zheng } 718de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 719525d15e8SYan, Zheng extern int __ceph_caps_wanted(struct ceph_inode_info *ci); 720de57606cSSage Weil 721de57606cSSage Weil /* what the mds thinks we want */ 722c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 723de57606cSSage Weil 72437151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 72537151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 726fe33032dSYan, Zheng extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, 727fe33032dSYan, Zheng struct ceph_mount_options *fsopt); 728e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 72937151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 7307bf8f736SChengguang Xu extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, 73137151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 7323d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 733de57606cSSage Weil int *total, int *avail, int *used, 73485ccce43SSage Weil int *reserved, int *min); 735de57606cSSage Weil 736de57606cSSage Weil 737de57606cSSage Weil 738de57606cSSage Weil /* 739de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 740de57606cSSage Weil */ 7414918b6d1SSage Weil #define CEPH_F_SYNC 1 7429cfa1098SSage Weil #define CEPH_F_ATEND 2 7434918b6d1SSage Weil 744de57606cSSage Weil struct ceph_file_info { 745252c6728SSage Weil short fmode; /* initialized on open */ 746252c6728SSage Weil short flags; /* CEPH_F_* */ 747de57606cSSage Weil 7485d988308SYan, Zheng spinlock_t rw_contexts_lock; 7495d988308SYan, Zheng struct list_head rw_contexts; 750f4b97866SYan, Zheng 751f4b97866SYan, Zheng errseq_t meta_err; 75281f148a9SYan, Zheng u32 filp_gen; 753ff5d913dSYan, Zheng atomic_t num_locks; 754bb48bd4dSChengguang Xu }; 755bb48bd4dSChengguang Xu 756bb48bd4dSChengguang Xu struct ceph_dir_file_info { 757bb48bd4dSChengguang Xu struct ceph_file_info file_info; 7585d988308SYan, Zheng 759de57606cSSage Weil /* readdir: position within the dir */ 760de57606cSSage Weil u32 frag; 761de57606cSSage Weil struct ceph_mds_request *last_readdir; 762de57606cSSage Weil 763de57606cSSage Weil /* readdir: position within a frag */ 764f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 765de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 766fdd4e158SYan, Zheng long long dir_release_count; 767fdd4e158SYan, Zheng long long dir_ordered_count; 768fdd4e158SYan, Zheng int readdir_cache_idx; 769de57606cSSage Weil 770de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 771de57606cSSage Weil char *dir_info; 772de57606cSSage Weil int dir_info_len; 773de57606cSSage Weil }; 774de57606cSSage Weil 7755d988308SYan, Zheng struct ceph_rw_context { 7765d988308SYan, Zheng struct list_head list; 7775d988308SYan, Zheng struct task_struct *thread; 7785d988308SYan, Zheng int caps; 7795d988308SYan, Zheng }; 7805d988308SYan, Zheng 7815d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 7825d988308SYan, Zheng struct ceph_rw_context _name = { \ 7835d988308SYan, Zheng .thread = current, \ 7845d988308SYan, Zheng .caps = _caps, \ 7855d988308SYan, Zheng } 7865d988308SYan, Zheng 7875d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 7885d988308SYan, Zheng struct ceph_rw_context *ctx) 7895d988308SYan, Zheng { 7905d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7915d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 7925d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 7935d988308SYan, Zheng } 7945d988308SYan, Zheng 7955d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 7965d988308SYan, Zheng struct ceph_rw_context *ctx) 7975d988308SYan, Zheng { 7985d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 7995d988308SYan, Zheng list_del(&ctx->list); 8005d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8015d988308SYan, Zheng } 8025d988308SYan, Zheng 8035d988308SYan, Zheng static inline struct ceph_rw_context* 8045d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 8055d988308SYan, Zheng { 8065d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 8075d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8085d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 8095d988308SYan, Zheng if (ctx->thread == current) { 8105d988308SYan, Zheng found = ctx; 8115d988308SYan, Zheng break; 8125d988308SYan, Zheng } 8135d988308SYan, Zheng } 8145d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8155d988308SYan, Zheng return found; 8165d988308SYan, Zheng } 8175d988308SYan, Zheng 818fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 819fdd4e158SYan, Zheng struct page *page; 820fdd4e158SYan, Zheng struct dentry **dentries; 821fdd4e158SYan, Zheng int index; 822fdd4e158SYan, Zheng }; 823de57606cSSage Weil 824de57606cSSage Weil /* 825de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 826de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 827de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 828de57606cSSage Weil * the snapshots of their parents. 829de57606cSSage Weil * 830de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 831de57606cSSage Weil * per-realm list. 832de57606cSSage Weil */ 833de57606cSSage Weil struct ceph_snap_realm { 834de57606cSSage Weil u64 ino; 835e3161f17SLuis Henriques struct inode *inode; 836de57606cSSage Weil atomic_t nref; 837a105f00cSSage Weil struct rb_node node; 838a105f00cSSage Weil 839de57606cSSage Weil u64 created, seq; 840de57606cSSage Weil u64 parent_ino; 841de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 842de57606cSSage Weil 843de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 844aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 845de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 846aa711ee3SAlex Elder u32 num_snaps; 847de57606cSSage Weil 848de57606cSSage Weil struct ceph_snap_realm *parent; 849de57606cSSage Weil struct list_head children; /* list of child realms */ 850de57606cSSage Weil struct list_head child_item; 851de57606cSSage Weil 852de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 853de57606cSSage Weil 854ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 855ae00d4f3SSage Weil 856de57606cSSage Weil /* the current set of snaps for this realm */ 857de57606cSSage Weil struct ceph_snap_context *cached_context; 858de57606cSSage Weil 859de57606cSSage Weil struct list_head inodes_with_caps; 860de57606cSSage Weil spinlock_t inodes_with_caps_lock; 861de57606cSSage Weil }; 862de57606cSSage Weil 8633d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 8643d14c5d2SYehuda Sadeh { 8653d14c5d2SYehuda Sadeh int congestion_kb; 866de57606cSSage Weil 867de57606cSSage Weil /* 8683d14c5d2SYehuda Sadeh * Copied from NFS 8693d14c5d2SYehuda Sadeh * 8703d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 8713d14c5d2SYehuda Sadeh * 8723d14c5d2SYehuda Sadeh * 64MB: 8192k 8733d14c5d2SYehuda Sadeh * 128MB: 11585k 8743d14c5d2SYehuda Sadeh * 256MB: 16384k 8753d14c5d2SYehuda Sadeh * 512MB: 23170k 8763d14c5d2SYehuda Sadeh * 1GB: 32768k 8773d14c5d2SYehuda Sadeh * 2GB: 46340k 8783d14c5d2SYehuda Sadeh * 4GB: 65536k 8793d14c5d2SYehuda Sadeh * 8GB: 92681k 8803d14c5d2SYehuda Sadeh * 16GB: 131072k 8813d14c5d2SYehuda Sadeh * 8823d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 8833d14c5d2SYehuda Sadeh * Limit the default to 256M 884de57606cSSage Weil */ 885ca79b0c2SArun KS congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); 8863d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 8873d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 8883d14c5d2SYehuda Sadeh 8893d14c5d2SYehuda Sadeh return congestion_kb; 890de57606cSSage Weil } 891de57606cSSage Weil 892de57606cSSage Weil 893d468e729SYan, Zheng /* super.c */ 894d468e729SYan, Zheng extern int ceph_force_reconnect(struct super_block *sb); 895de57606cSSage Weil /* snap.c */ 896de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 897de57606cSSage Weil u64 ino); 898de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 899de57606cSSage Weil struct ceph_snap_realm *realm); 900de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 901de57606cSSage Weil struct ceph_snap_realm *realm); 902de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 903982d6011SYan, Zheng void *p, void *e, bool deletion, 904982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 905de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 9062600d2ddSSage Weil struct ceph_mds_session *session, 907de57606cSSage Weil struct ceph_msg *msg); 908fc837c8fSSage Weil extern void ceph_queue_cap_snap(struct ceph_inode_info *ci); 909de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 910de57606cSSage Weil struct ceph_cap_snap *capsnap); 911de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); 912de57606cSSage Weil 91375c9627eSYan, Zheng extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, 91475c9627eSYan, Zheng u64 snap); 91575c9627eSYan, Zheng extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, 91675c9627eSYan, Zheng struct ceph_snapid_map *sm); 91775c9627eSYan, Zheng extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); 91875c9627eSYan, Zheng extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); 91975c9627eSYan, Zheng 92075c9627eSYan, Zheng 921de57606cSSage Weil /* 922de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 923de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 924de57606cSSage Weil */ 925de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 926de57606cSSage Weil { 927de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 92886056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 929de57606cSSage Weil ci_item)->writing; 930de57606cSSage Weil } 931de57606cSSage Weil 932de57606cSSage Weil /* inode.c */ 933966c7160SJeff Layton struct ceph_mds_reply_info_in; 934966c7160SJeff Layton struct ceph_mds_reply_dirfrag; 935966c7160SJeff Layton 936de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 937de57606cSSage Weil 938de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 93987bc5b89SYan, Zheng extern void ceph_evict_inode(struct inode *inode); 940cfa6d412SAl Viro extern void ceph_free_inode(struct inode *inode); 941de57606cSSage Weil 942de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 943de57606cSSage Weil struct ceph_vino vino); 944de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 945de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 946de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 947de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 9489bbeab41SArnd Bergmann u64 time_warp_seq, struct timespec64 *ctime, 9499bbeab41SArnd Bergmann struct timespec64 *mtime, 9509bbeab41SArnd Bergmann struct timespec64 *atime); 951966c7160SJeff Layton extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, 952966c7160SJeff Layton struct ceph_mds_reply_info_in *iinfo, 953966c7160SJeff Layton struct ceph_mds_reply_dirfrag *dirinfo, 954966c7160SJeff Layton struct ceph_mds_session *session, int cap_fmode, 955966c7160SJeff Layton struct ceph_cap_reservation *caps_reservation); 956de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 957f5a03b08SJeff Layton struct ceph_mds_request *req); 958de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 959de57606cSSage Weil struct ceph_mds_session *session); 960de57606cSSage Weil 961de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 962de57606cSSage Weil 963efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 964b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 9653c6f6b79SSage Weil extern void ceph_queue_vmtruncate(struct inode *inode); 9663c6f6b79SSage Weil extern void ceph_queue_invalidate(struct inode *inode); 9673c6f6b79SSage Weil extern void ceph_queue_writeback(struct inode *inode); 9683e1d0452SYan, Zheng extern void ceph_async_iput(struct inode *inode); 969de57606cSSage Weil 97001deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 97101deead0SYan, Zheng int mask, bool force); 97201deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 97301deead0SYan, Zheng { 97401deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 97501deead0SYan, Zheng } 976*549c7297SChristian Brauner extern int ceph_permission(struct user_namespace *mnt_userns, 977*549c7297SChristian Brauner struct inode *inode, int mask); 978a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 979*549c7297SChristian Brauner extern int ceph_setattr(struct user_namespace *mnt_userns, 980*549c7297SChristian Brauner struct dentry *dentry, struct iattr *attr); 981*549c7297SChristian Brauner extern int ceph_getattr(struct user_namespace *mnt_userns, 982*549c7297SChristian Brauner const struct path *path, struct kstat *stat, 983a528d35eSDavid Howells u32 request_mask, unsigned int flags); 984de57606cSSage Weil 985de57606cSSage Weil /* xattr.c */ 986a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 9877221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 988de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 98912fe3ddaSLuis Henriques extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); 990de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 991b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 992de57606cSSage Weil 9935c31e92dSYan, Zheng struct ceph_acl_sec_ctx { 9945c31e92dSYan, Zheng #ifdef CONFIG_CEPH_FS_POSIX_ACL 9955c31e92dSYan, Zheng void *default_acl; 9965c31e92dSYan, Zheng void *acl; 9975c31e92dSYan, Zheng #endif 998ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 999ac6713ccSYan, Zheng void *sec_ctx; 1000ac6713ccSYan, Zheng u32 sec_ctxlen; 1001ac6713ccSYan, Zheng #endif 10025c31e92dSYan, Zheng struct ceph_pagelist *pagelist; 10035c31e92dSYan, Zheng }; 10045c31e92dSYan, Zheng 1005315f2408SYan, Zheng #ifdef CONFIG_SECURITY 1006315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 1007315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 1008315f2408SYan, Zheng #else 1009315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 1010315f2408SYan, Zheng { 1011315f2408SYan, Zheng return false; 1012315f2408SYan, Zheng } 1013315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 1014315f2408SYan, Zheng { 1015315f2408SYan, Zheng return false; 1016315f2408SYan, Zheng } 1017315f2408SYan, Zheng #endif 1018315f2408SYan, Zheng 1019ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 1020ac6713ccSYan, Zheng extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1021ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx); 1022668959a5SJeff Layton static inline void ceph_security_invalidate_secctx(struct inode *inode) 1023668959a5SJeff Layton { 1024668959a5SJeff Layton security_inode_invalidate_secctx(inode); 1025668959a5SJeff Layton } 1026ac6713ccSYan, Zheng #else 1027ac6713ccSYan, Zheng static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1028ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx) 1029ac6713ccSYan, Zheng { 1030ac6713ccSYan, Zheng return 0; 1031ac6713ccSYan, Zheng } 1032ac6713ccSYan, Zheng static inline void ceph_security_invalidate_secctx(struct inode *inode) 1033ac6713ccSYan, Zheng { 1034ac6713ccSYan, Zheng } 1035ac6713ccSYan, Zheng #endif 10367221fe4cSGuangliang Zhao 10375c31e92dSYan, Zheng void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); 10387221fe4cSGuangliang Zhao 10395c31e92dSYan, Zheng /* acl.c */ 10407221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 10417221fe4cSGuangliang Zhao 10427221fe4cSGuangliang Zhao struct posix_acl *ceph_get_acl(struct inode *, int); 1043*549c7297SChristian Brauner int ceph_set_acl(struct user_namespace *mnt_userns, 1044*549c7297SChristian Brauner struct inode *inode, struct posix_acl *acl, int type); 1045b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 10465c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 10475c31e92dSYan, Zheng void ceph_init_inode_acls(struct inode *inode, 10485c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 1049c969d9bfSGuangliang Zhao 1050c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 1051c969d9bfSGuangliang Zhao { 1052c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 1053c969d9bfSGuangliang Zhao } 10547221fe4cSGuangliang Zhao 10557221fe4cSGuangliang Zhao #else 10567221fe4cSGuangliang Zhao 10577221fe4cSGuangliang Zhao #define ceph_get_acl NULL 105872466d0bSSage Weil #define ceph_set_acl NULL 10597221fe4cSGuangliang Zhao 1060b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 10615c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 10627221fe4cSGuangliang Zhao { 10637221fe4cSGuangliang Zhao return 0; 10647221fe4cSGuangliang Zhao } 1065b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 10665c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 1067b1ee94aaSYan, Zheng { 1068b1ee94aaSYan, Zheng } 10697221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 10707221fe4cSGuangliang Zhao { 10717221fe4cSGuangliang Zhao return 0; 10727221fe4cSGuangliang Zhao } 10737221fe4cSGuangliang Zhao 10747221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 10757221fe4cSGuangliang Zhao { 10767221fe4cSGuangliang Zhao } 10777221fe4cSGuangliang Zhao 10787221fe4cSGuangliang Zhao #endif 10797221fe4cSGuangliang Zhao 1080de57606cSSage Weil /* caps.c */ 1081de57606cSSage Weil extern const char *ceph_cap_string(int c); 1082de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 1083de57606cSSage Weil struct ceph_msg *msg); 1084d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 1085d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 1086d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 1087de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 1088135e671eSYan, Zheng unsigned issued, unsigned wanted, 1089de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 1090d9df2783SYan, Zheng struct ceph_cap **new_cap); 1091a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1092d6e47819SYan, Zheng extern void __ceph_remove_caps(struct ceph_inode_info *ci); 109337151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 109437151668SYehuda Sadeh struct ceph_cap *cap); 10959215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 1096de57606cSSage Weil 1097f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 109802c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 109902c24a82SJosef Bacik int datasync); 1100e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 1101e548e9b9SYan, Zheng struct ceph_mds_session *session); 1102de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 1103de57606cSSage Weil struct ceph_mds_session *session); 1104e8a4d267SJeff Layton void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, 1105e8a4d267SJeff Layton struct ceph_inode_info *ci); 11062bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 11072bc50259SGreg Farnum int mds); 110840dcf75eSJeff Layton extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, 110940dcf75eSJeff Layton bool snap_rwsem_locked); 1110de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 1111de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 1112e64f44a8SXiubo Li extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, 1113e64f44a8SXiubo Li int had); 1114de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 1115de57606cSSage Weil struct ceph_snap_context *snapc); 1116ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 11170e294387SYan, Zheng struct ceph_mds_session **psession); 1118efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 1119de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 1120de57606cSSage Weil struct ceph_mds_session *session); 1121afcdaea3SSage Weil extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 1122afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 11236ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 1124de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 1125de57606cSSage Weil int mds, int drop, int unless, int force); 1126de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 1127ca6c8ae0SJeff Layton struct inode *dir, 1128de57606cSSage Weil int mds, int drop, int unless); 1129de57606cSSage Weil 11305e3ded1bSYan, Zheng extern int ceph_get_caps(struct file *filp, int need, int want, 113131c542a1SYan, Zheng loff_t endoff, int *got, struct page **pinned_page); 11325e3ded1bSYan, Zheng extern int ceph_try_get_caps(struct inode *inode, 11332ee9dd95SLuis Henriques int need, int want, bool nonblock, int *got); 1134de57606cSSage Weil 1135de57606cSSage Weil /* for counting open files by mode */ 1136719a2514SYan, Zheng extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); 1137719a2514SYan, Zheng extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); 1138719a2514SYan, Zheng extern void __ceph_touch_fmode(struct ceph_inode_info *ci, 1139719a2514SYan, Zheng struct ceph_mds_client *mdsc, int fmode); 1140de57606cSSage Weil 1141de57606cSSage Weil /* addr.c */ 1142de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1143de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 114410183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page); 11455e3ded1bSYan, Zheng extern int ceph_pool_perm_check(struct inode *inode, int need); 114610183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 1147de57606cSSage Weil 1148de57606cSSage Weil /* file.c */ 1149de57606cSSage Weil extern const struct file_operations ceph_file_fops; 11509e0eb85dSAlex Elder 1151719a2514SYan, Zheng extern int ceph_renew_caps(struct inode *inode, int fmode); 1152de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 11535ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 115444907d79SAl Viro struct file *file, unsigned flags, umode_t mode); 1155de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 115631c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 115731c542a1SYan, Zheng char *data, size_t len); 115855f2a045SIlya Dryomov 1159de57606cSSage Weil /* dir.c */ 1160de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 116138c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1162de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 116338c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 116418fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1165de57606cSSage Weil 1166f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1167de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1168468640e3SSage Weil extern int ceph_handle_snapdir(struct ceph_mds_request *req, 1169468640e3SSage Weil struct dentry *dentry, int err); 1170de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1171de57606cSSage Weil struct dentry *dentry, int err); 1172de57606cSSage Weil 117337c4efc1SYan, Zheng extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); 117437c4efc1SYan, Zheng extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); 117581a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 117637c4efc1SYan, Zheng extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); 1177e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1178fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1179de57606cSSage Weil 1180de57606cSSage Weil /* ioctl.c */ 1181de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1182de57606cSSage Weil 1183de57606cSSage Weil /* export.c */ 1184de57606cSSage Weil extern const struct export_operations ceph_export_ops; 11853886274aSLuis Henriques struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); 1186de57606cSSage Weil 118740819f6fSGreg Farnum /* locks.c */ 1188eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 118940819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 119040819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 119140819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 119239be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 119339be95e9SJim Schutt struct ceph_filelock *flocks, 119439be95e9SJim Schutt int num_fcntl_locks, 119539be95e9SJim Schutt int num_flock_locks); 119639be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 119739be95e9SJim Schutt struct ceph_pagelist *pagelist, 119839be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 119940819f6fSGreg Farnum 12003d14c5d2SYehuda Sadeh /* debugfs.c */ 12011a829ff2SGreg Kroah-Hartman extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); 12023d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 12033d14c5d2SYehuda Sadeh 1204fb18a575SLuis Henriques /* quota.c */ 1205d557c48dSLuis Henriques static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) 1206d557c48dSLuis Henriques { 1207d557c48dSLuis Henriques return ci->i_max_files || ci->i_max_bytes; 1208d557c48dSLuis Henriques } 1209d557c48dSLuis Henriques 1210d557c48dSLuis Henriques extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); 1211d557c48dSLuis Henriques 1212d557c48dSLuis Henriques static inline void __ceph_update_quota(struct ceph_inode_info *ci, 1213d557c48dSLuis Henriques u64 max_bytes, u64 max_files) 1214d557c48dSLuis Henriques { 1215d557c48dSLuis Henriques bool had_quota, has_quota; 1216d557c48dSLuis Henriques had_quota = __ceph_has_any_quota(ci); 1217d557c48dSLuis Henriques ci->i_max_bytes = max_bytes; 1218d557c48dSLuis Henriques ci->i_max_files = max_files; 1219d557c48dSLuis Henriques has_quota = __ceph_has_any_quota(ci); 1220d557c48dSLuis Henriques 1221d557c48dSLuis Henriques if (had_quota != has_quota) 1222d557c48dSLuis Henriques ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); 1223d557c48dSLuis Henriques } 1224d557c48dSLuis Henriques 1225fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc, 1226fb18a575SLuis Henriques struct ceph_mds_session *session, 1227fb18a575SLuis Henriques struct ceph_msg *msg); 1228b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); 12296646ea1cSLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); 12302b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, 12312b83845fSLuis Henriques loff_t newlen); 12321ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, 12331ab302a0SLuis Henriques loff_t newlen); 12349122eed5SLuis Henriques extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, 12359122eed5SLuis Henriques struct kstatfs *buf); 12360c44a8e0SLuis Henriques extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); 1237fb18a575SLuis Henriques 1238de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1239