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> 20bc899ee1SDavid Howells #include <linux/netfs.h> 21bc899ee1SDavid Howells #include <linux/fscache.h> 224868e537SXiubo Li #include <linux/hashtable.h> 23de57606cSSage Weil 243d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 25de57606cSSage Weil 26de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 27de57606cSSage Weil * large volume sizes on 32-bit machines. */ 2892a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 29de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 308e55ba8cSKotresh HR #define CEPH_4K_BLOCK_SHIFT 12 /* 4 KB */ 31de57606cSSage Weil 320b98acd6SIlya Dryomov #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blocklisted */ 333d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 343d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 353d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 36ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 37a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 3899ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 3910183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 40e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 419122eed5SLuis Henriques #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ 42ea4cdc54SLuis Henriques #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ 432ccb4546SJeff Layton #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ 4494cc0877SJeff Layton #define CEPH_MOUNT_OPT_NOPAGECACHE (1<<16) /* bypass pagecache altogether */ 456a259382SSage Weil 466f9718feSLuis Henriques #define CEPH_MOUNT_OPT_DEFAULT \ 476f9718feSLuis Henriques (CEPH_MOUNT_OPT_DCACHE | \ 48f7a67b46SJeff Layton CEPH_MOUNT_OPT_NOCOPYFROM | \ 49f7a67b46SJeff Layton CEPH_MOUNT_OPT_ASYNC_DIROPS) 50de57606cSSage Weil 513d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 522ccb4546SJeff Layton (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt 532ccb4546SJeff Layton #define ceph_clear_mount_opt(fsc, opt) \ 542ccb4546SJeff Layton (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt 553d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 563d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 57de57606cSSage Weil 58aa187926SYan, Zheng /* max size of osd read request, limited by libceph */ 59aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN 6095cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size. 6195cca2b4SYan, Zheng * CEPH_MSG_MAX_DATA_LEN should be small enough. */ 6295cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN 637c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 643d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 653d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 663d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 67de57606cSSage Weil 684214fb15SYan, Zheng /* 694214fb15SYan, Zheng * Delay telling the MDS we no longer want caps, in case we reopen 704214fb15SYan, Zheng * the file. Delay a minimum amount of time, even if we send a cap 714214fb15SYan, Zheng * message for some other reason. Otherwise, take the oppotunity to 724214fb15SYan, Zheng * update the mds to avoid sending another message later. 734214fb15SYan, Zheng */ 744214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ 754214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ 764214fb15SYan, Zheng 773d14c5d2SYehuda Sadeh struct ceph_mount_options { 78ad8c28a9SJeff Layton unsigned int flags; 793d14c5d2SYehuda Sadeh 80ad8c28a9SJeff Layton unsigned int wsize; /* max write size */ 81ad8c28a9SJeff Layton unsigned int rsize; /* max read size */ 82ad8c28a9SJeff Layton unsigned int rasize; /* max readahead */ 83ad8c28a9SJeff Layton unsigned int congestion_kb; /* max writeback in flight */ 84ad8c28a9SJeff Layton unsigned int caps_wanted_delay_min, caps_wanted_delay_max; 85fe33032dSYan, Zheng int caps_max; 86ad8c28a9SJeff Layton unsigned int max_readdir; /* max readdir result (entries) */ 87ad8c28a9SJeff Layton unsigned int max_readdir_bytes; /* max readdir result (bytes) */ 883d14c5d2SYehuda Sadeh 897b19b4dbSVenky Shankar bool new_dev_syntax; 907b19b4dbSVenky Shankar 913d14c5d2SYehuda Sadeh /* 923d14c5d2SYehuda Sadeh * everything above this point can be memcmp'd; everything below 933d14c5d2SYehuda Sadeh * is handled in compare_mount_options() 943d14c5d2SYehuda Sadeh */ 953d14c5d2SYehuda Sadeh 96de57606cSSage Weil char *snapdir_name; /* default ".snap" */ 97430afbadSYan, Zheng char *mds_namespace; /* default NULL */ 98b27a939eSIlya Dryomov char *server_path; /* default NULL (means "/") */ 991d8f8360SYan, Zheng char *fscache_uniq; /* default NULL */ 1007b19b4dbSVenky Shankar char *mon_addr; 101de57606cSSage Weil }; 102de57606cSSage Weil 1034868e537SXiubo Li #define CEPH_ASYNC_CREATE_CONFLICT_BITS 8 1044868e537SXiubo Li 1053d14c5d2SYehuda Sadeh struct ceph_fs_client { 106de57606cSSage Weil struct super_block *sb; 107de57606cSSage Weil 10818f473b3SXiubo Li struct list_head metric_wakeup; 10918f473b3SXiubo Li 1103d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 1113d14c5d2SYehuda Sadeh struct ceph_client *client; 1123d14c5d2SYehuda Sadeh 113aa5c7910SJeff Layton int mount_state; 11481f148a9SYan, Zheng 1150b98acd6SIlya Dryomov bool blocklisted; 116131d7eb4SYan, Zheng 11778beb0ffSLuis Henriques bool have_copy_from2; 11878beb0ffSLuis Henriques 11981f148a9SYan, Zheng u32 filp_gen; 120719784baSChengguang Xu loff_t max_file_size; 12185ccce43SSage Weil 1223d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 123de57606cSSage Weil 124e3ec8d68SYan, Zheng atomic_long_t writeback_count; 125503d4fa6SNeilBrown bool write_congested; 126e3ec8d68SYan, Zheng 1271cf89a8dSYan, Zheng struct workqueue_struct *inode_wq; 128e3ec8d68SYan, Zheng struct workqueue_struct *cap_wq; 129de57606cSSage Weil 1304868e537SXiubo Li DECLARE_HASHTABLE(async_unlink_conflict, CEPH_ASYNC_CREATE_CONFLICT_BITS); 1314868e537SXiubo Li spinlock_t async_unlink_conflict_lock; 1324868e537SXiubo Li 1330743304dSSage Weil #ifdef CONFIG_DEBUG_FS 1343d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1352baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 13606edf046SSage Weil struct dentry *debugfs_bdi; 1373d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 138247b1f19SXiubo Li struct dentry *debugfs_status; 13914ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 140cbed4ff7SLuís Henriques struct dentry *debugfs_metrics_dir; 1410743304dSSage Weil #endif 14299ccbd22SMilosz Tanski 14399ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 144400e1286SJeff Layton struct fscache_volume *fscache; 14599ccbd22SMilosz Tanski #endif 146de57606cSSage Weil }; 147de57606cSSage Weil 1483d14c5d2SYehuda Sadeh 149de57606cSSage Weil /* 150de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 151de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 152de57606cSSage Weil * and write data. For any given inode, we should have one or more 153de57606cSSage Weil * capabilities, one issued by each metadata server, and our 154de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 155de57606cSSage Weil * 156de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 157de57606cSSage Weil * session capability lists. 158de57606cSSage Weil */ 159de57606cSSage Weil struct ceph_cap { 160de57606cSSage Weil struct ceph_inode_info *ci; 161de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 162de57606cSSage Weil struct ceph_mds_session *session; 163de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 164de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 165745a8e3bSYan, Zheng union { 166745a8e3bSYan, Zheng /* in-use caps */ 167745a8e3bSYan, Zheng struct { 168de57606cSSage Weil int issued; /* latest, from the mds */ 169745a8e3bSYan, Zheng int implemented; /* implemented superset of 170745a8e3bSYan, Zheng issued (for revocation) */ 171c74d79afSJeff Layton int mds; /* mds index for this cap */ 172c74d79afSJeff Layton int mds_wanted; /* caps wanted from this mds */ 173745a8e3bSYan, Zheng }; 174745a8e3bSYan, Zheng /* caps to release */ 175745a8e3bSYan, Zheng struct { 176745a8e3bSYan, Zheng u64 cap_ino; 177745a8e3bSYan, Zheng int queue_release; 178745a8e3bSYan, Zheng }; 179745a8e3bSYan, Zheng }; 180685f9a5dSSage Weil u32 seq, issue_seq, mseq; 181685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 182de57606cSSage Weil unsigned long last_used; 183de57606cSSage Weil struct list_head caps_item; 184de57606cSSage Weil }; 185de57606cSSage Weil 186a0d93e32SYan, Zheng #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ 187a0d93e32SYan, Zheng #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ 188a0d93e32SYan, Zheng #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ 189de57606cSSage Weil 1900e294387SYan, Zheng struct ceph_cap_flush { 1910e294387SYan, Zheng u64 tid; 192b2f9fa1fSXiubo Li int caps; 193c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 194b2f9fa1fSXiubo Li bool is_capsnap; /* true means capsnap */ 1950e294387SYan, Zheng struct list_head g_list; // global 1960e294387SYan, Zheng struct list_head i_list; // per inode 1970e294387SYan, Zheng }; 1980e294387SYan, Zheng 199de57606cSSage Weil /* 200de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 201de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 202de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 203de57606cSSage Weil */ 204de57606cSSage Weil struct ceph_cap_snap { 205805692d0SElena Reshetova refcount_t nref; 2060e294387SYan, Zheng struct list_head ci_item; 207de57606cSSage Weil 2080e294387SYan, Zheng struct ceph_cap_flush cap_flush; 2090e294387SYan, Zheng 2100e294387SYan, Zheng u64 follows; 211de57606cSSage Weil int issued, dirty; 212de57606cSSage Weil struct ceph_snap_context *context; 213de57606cSSage Weil 2145706b27dSAl Viro umode_t mode; 21505cb11c1SEric W. Biederman kuid_t uid; 21605cb11c1SEric W. Biederman kgid_t gid; 217de57606cSSage Weil 2184a625be4SSage Weil struct ceph_buffer *xattr_blob; 219de57606cSSage Weil u64 xattr_version; 220de57606cSSage Weil 221de57606cSSage Weil u64 size; 222176c77c9SJeff Layton u64 change_attr; 223ec62b894SJeff Layton struct timespec64 mtime, atime, ctime, btime; 224de57606cSSage Weil u64 time_warp_seq; 2255f743e45SYan, Zheng u64 truncate_size; 2265f743e45SYan, Zheng u32 truncate_seq; 227de57606cSSage Weil int writing; /* a sync write is still in progress */ 228de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 229e20d258dSYan, Zheng bool inline_data; 23086056090SYan, Zheng bool need_flush; 231de57606cSSage Weil }; 232de57606cSSage Weil 233de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 234de57606cSSage Weil { 235805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 2364a625be4SSage Weil if (capsnap->xattr_blob) 2374a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 238ab58a5a1SXiubo Li kmem_cache_free(ceph_cap_snap_cachep, capsnap); 239de57606cSSage Weil } 2404a625be4SSage Weil } 241de57606cSSage Weil 242de57606cSSage Weil /* 243de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 244de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 245de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 246de57606cSSage Weil * 247de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 248de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 249de57606cSSage Weil */ 250de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 251de57606cSSage Weil 252de57606cSSage Weil struct ceph_inode_frag { 253de57606cSSage Weil struct rb_node node; 254de57606cSSage Weil 255de57606cSSage Weil /* fragtree state */ 256de57606cSSage Weil u32 frag; 257de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 258de57606cSSage Weil 259de57606cSSage Weil /* delegation and replication info */ 260de57606cSSage Weil int mds; /* -1 if same authority as parent */ 261de57606cSSage Weil int ndist; /* >0 if replicated */ 262de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 263de57606cSSage Weil }; 264de57606cSSage Weil 265de57606cSSage Weil /* 266de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 267de57606cSSage Weil * at which point we parse them into an rbtree. 268de57606cSSage Weil */ 269de57606cSSage Weil struct ceph_inode_xattr { 270de57606cSSage Weil struct rb_node node; 271de57606cSSage Weil 272de57606cSSage Weil const char *name; 273de57606cSSage Weil int name_len; 274de57606cSSage Weil const char *val; 275de57606cSSage Weil int val_len; 276de57606cSSage Weil int dirty; 277de57606cSSage Weil 278de57606cSSage Weil int should_free_name; 279de57606cSSage Weil int should_free_val; 280de57606cSSage Weil }; 281de57606cSSage Weil 2823d14c5d2SYehuda Sadeh /* 2833d14c5d2SYehuda Sadeh * Ceph dentry state 2843d14c5d2SYehuda Sadeh */ 2853d14c5d2SYehuda Sadeh struct ceph_dentry_info { 28637c4efc1SYan, Zheng struct dentry *dentry; 2873d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 28837c4efc1SYan, Zheng struct list_head lease_list; 2894868e537SXiubo Li struct hlist_node hnode; 2904868e537SXiubo Li unsigned long flags; 29197aeb6bfSYan, Zheng int lease_shared_gen; 29297aeb6bfSYan, Zheng u32 lease_gen; 2933d14c5d2SYehuda Sadeh u32 lease_seq; 2943d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2959b16f03cSMiklos Szeredi unsigned long time; 2963d14c5d2SYehuda Sadeh u64 offset; 2973d14c5d2SYehuda Sadeh }; 2983d14c5d2SYehuda Sadeh 2994868e537SXiubo Li #define CEPH_DENTRY_REFERENCED (1 << 0) 3004868e537SXiubo Li #define CEPH_DENTRY_LEASE_LIST (1 << 1) 3014868e537SXiubo Li #define CEPH_DENTRY_SHRINK_LIST (1 << 2) 3024868e537SXiubo Li #define CEPH_DENTRY_PRIMARY_LINK (1 << 3) 3034868e537SXiubo Li #define CEPH_DENTRY_ASYNC_UNLINK_BIT (4) 3044868e537SXiubo Li #define CEPH_DENTRY_ASYNC_UNLINK (1 << CEPH_DENTRY_ASYNC_UNLINK_BIT) 30500061645SXiubo Li #define CEPH_DENTRY_ASYNC_CREATE_BIT (5) 30600061645SXiubo Li #define CEPH_DENTRY_ASYNC_CREATE (1 << CEPH_DENTRY_ASYNC_CREATE_BIT) 30737c4efc1SYan, Zheng 308de57606cSSage Weil struct ceph_inode_xattrs_info { 309de57606cSSage Weil /* 310de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 311de57606cSSage Weil * this until someone actually calls getxattr, etc. 312de57606cSSage Weil * 313de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 314de57606cSSage Weil * NULL means we don't know. 315de57606cSSage Weil */ 316de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 317de57606cSSage Weil 318de57606cSSage Weil struct rb_root index; 319de57606cSSage Weil bool dirty; 320de57606cSSage Weil int count; 321de57606cSSage Weil int names_size; 322de57606cSSage Weil int vals_size; 323de57606cSSage Weil u64 version, index_version; 324de57606cSSage Weil }; 325de57606cSSage Weil 326de57606cSSage Weil /* 327de57606cSSage Weil * Ceph inode. 328de57606cSSage Weil */ 329de57606cSSage Weil struct ceph_inode_info { 330874c8ca1SDavid Howells struct netfs_inode netfs; /* Netfslib context and vfs inode */ 331de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 332de57606cSSage Weil 333be655596SSage Weil spinlock_t i_ceph_lock; 334be655596SSage Weil 335de57606cSSage Weil u64 i_version; 33631c542a1SYan, Zheng u64 i_inline_version; 337de57606cSSage Weil u32 i_time_warp_seq; 338de57606cSSage Weil 339891f3f5aSJeff Layton unsigned long i_ceph_flags; 340fdd4e158SYan, Zheng atomic64_t i_release_count; 341fdd4e158SYan, Zheng atomic64_t i_ordered_count; 342fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 343de57606cSSage Weil 3446c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 345de57606cSSage Weil struct ceph_file_layout i_layout; 346785892feSJeff Layton struct ceph_file_layout i_cached_layout; // for async creates 347de57606cSSage Weil char *i_symlink; 348de57606cSSage Weil 349de57606cSSage Weil /* for dirs */ 3509bbeab41SArnd Bergmann struct timespec64 i_rctime; 351e7f72952SYanhu Cao u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps; 352de57606cSSage Weil u64 i_files, i_subdirs; 353de57606cSSage Weil 354fb18a575SLuis Henriques /* quotas */ 355fb18a575SLuis Henriques u64 i_max_bytes, i_max_files; 356fb18a575SLuis Henriques 35708796873SYan, Zheng s32 i_dir_pin; 35808796873SYan, Zheng 359de57606cSSage Weil struct rb_root i_fragtree; 3601b1bc16dSYan, Zheng int i_fragtree_nsplits; 361de57606cSSage Weil struct mutex i_fragtree_mutex; 362de57606cSSage Weil 363de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 364de57606cSSage Weil 365be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 366de57606cSSage Weil * s_mutex. */ 367de57606cSSage Weil struct rb_root i_caps; /* cap list */ 368de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 369de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 3701cf03a68SJeff Layton 3711cf03a68SJeff Layton /* 372f1f565a2SRandy Dunlap * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty 3731cf03a68SJeff Layton * is protected by the mdsc->cap_dirty_lock, but each individual item 3741cf03a68SJeff Layton * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty 3751cf03a68SJeff Layton * requires the mdsc->cap_dirty_lock. List presence for an item can 3761cf03a68SJeff Layton * be tested under the i_ceph_lock. Changing anything requires both. 3774fb5dda3SJeff Layton */ 3781cf03a68SJeff Layton struct list_head i_dirty_item; 3791cf03a68SJeff Layton 3801cf03a68SJeff Layton /* 381829ad4dbSJeff Layton * Link to session's s_cap_flushing list. Protected in a similar 382829ad4dbSJeff Layton * fashion to i_dirty_item, but also by the s_mutex for changes. The 383829ad4dbSJeff Layton * s_cap_flushing list can be walked while holding either the s_mutex 384829ad4dbSJeff Layton * or msdc->cap_dirty_lock. List presence can also be checked while 385829ad4dbSJeff Layton * holding the i_ceph_lock for this inode. 3861cf03a68SJeff Layton */ 3871cf03a68SJeff Layton struct list_head i_flushing_item; 3881cf03a68SJeff Layton 389de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 390de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 391de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 392f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 393e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 394de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 395de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 396de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 397de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 398de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3997d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 4007d8cb26dSSage Weil dirty|flushing caps */ 401de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 402de57606cSSage Weil 403719a2514SYan, Zheng unsigned long i_last_rd; 404719a2514SYan, Zheng unsigned long i_last_wr; 405774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 406de57606cSSage Weil 407b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 408de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 409de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 410de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 411de57606cSSage Weil 412de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 413de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 414de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 415de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 416de57606cSSage Weil 417de57606cSSage Weil /* held references to caps */ 418de57606cSSage Weil int i_pin_ref; 419f85122afSJeff Layton int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; 420de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 42189aa5930SYan, Zheng atomic_t i_filelock_ref; 42297aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 423cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 424de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 425de57606cSSage Weil 426de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 42768cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 428de57606cSSage Weil spinlock_t i_unsafe_lock; 429de57606cSSage Weil 43075c9627eSYan, Zheng union { 431de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 43275c9627eSYan, Zheng struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ 43375c9627eSYan, Zheng }; 434de57606cSSage Weil struct list_head i_snap_realm_item; 435de57606cSSage Weil struct list_head i_snap_flush_item; 436245ce991SJeff Layton struct timespec64 i_btime; 437193e7b37SDavid Disseldorp struct timespec64 i_snap_btime; 438de57606cSSage Weil 4391cf89a8dSYan, Zheng struct work_struct i_work; 4401cf89a8dSYan, Zheng unsigned long i_work_mask; 441de57606cSSage Weil }; 442de57606cSSage Weil 443721d5c13SJeff Layton static inline struct ceph_inode_info * 444721d5c13SJeff Layton ceph_inode(const struct inode *inode) 445de57606cSSage Weil { 446874c8ca1SDavid Howells return container_of(inode, struct ceph_inode_info, netfs.inode); 447de57606cSSage Weil } 448de57606cSSage Weil 449721d5c13SJeff Layton static inline struct ceph_fs_client * 450721d5c13SJeff Layton ceph_inode_to_client(const struct inode *inode) 451ad1fee96SYehuda Sadeh { 452ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 453ad1fee96SYehuda Sadeh } 454ad1fee96SYehuda Sadeh 455721d5c13SJeff Layton static inline struct ceph_fs_client * 456721d5c13SJeff Layton ceph_sb_to_client(const struct super_block *sb) 457ad1fee96SYehuda Sadeh { 458ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 459ad1fee96SYehuda Sadeh } 460ad1fee96SYehuda Sadeh 4612678da88SXiubo Li static inline struct ceph_mds_client * 4622678da88SXiubo Li ceph_sb_to_mdsc(const struct super_block *sb) 4632678da88SXiubo Li { 4642678da88SXiubo Li return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc; 4652678da88SXiubo Li } 4662678da88SXiubo Li 467721d5c13SJeff Layton static inline struct ceph_vino 468721d5c13SJeff Layton ceph_vino(const struct inode *inode) 4693d14c5d2SYehuda Sadeh { 4703d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4713d14c5d2SYehuda Sadeh } 4723d14c5d2SYehuda Sadeh 473ebce3eb2SJeff Layton static inline u32 ceph_ino_to_ino32(u64 vino) 474ad1fee96SYehuda Sadeh { 4753310f754SAmon Ott u32 ino = vino & 0xffffffff; 4763310f754SAmon Ott ino ^= vino >> 32; 477ad1fee96SYehuda Sadeh if (!ino) 478a661fc56SAmon Ott ino = 2; 479ad1fee96SYehuda Sadeh return ino; 480ad1fee96SYehuda Sadeh } 481ad1fee96SYehuda Sadeh 482ad1fee96SYehuda Sadeh /* 483ebce3eb2SJeff Layton * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on 484ebce3eb2SJeff Layton * some arches. We generally do not use this value inside the ceph driver, but 485ebce3eb2SJeff Layton * we do want to set it to something, so that generic vfs code has an 486ebce3eb2SJeff Layton * appropriate value for tracepoints and the like. 4873d14c5d2SYehuda Sadeh */ 488ebce3eb2SJeff Layton static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino) 4893d14c5d2SYehuda Sadeh { 490ebce3eb2SJeff Layton if (sizeof(ino_t) == sizeof(u32)) 4913310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4923310f754SAmon Ott return (ino_t)vino.ino; 4933d14c5d2SYehuda Sadeh } 4943d14c5d2SYehuda Sadeh 4953d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4963d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4973d14c5d2SYehuda Sadeh 4983d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4993d14c5d2SYehuda Sadeh { 5003d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 5013d14c5d2SYehuda Sadeh } 502ebce3eb2SJeff Layton 5033d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 5043d14c5d2SYehuda Sadeh { 5053d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 5063d14c5d2SYehuda Sadeh } 5073d14c5d2SYehuda Sadeh 508ebce3eb2SJeff Layton /** 509ebce3eb2SJeff Layton * ceph_present_ino - format an inode number for presentation to userland 510ebce3eb2SJeff Layton * @sb: superblock where the inode lives 511ebce3eb2SJeff Layton * @ino: inode number to (possibly) convert 512ebce3eb2SJeff Layton * 513ebce3eb2SJeff Layton * If the user mounted with the ino32 option, then the 64-bit value needs 514ebce3eb2SJeff Layton * to be converted to something that can fit inside 32 bits. Note that 515ebce3eb2SJeff Layton * internal kernel code never uses this value, so this is entirely for 516ebce3eb2SJeff Layton * userland consumption. 517ebce3eb2SJeff Layton */ 518ebce3eb2SJeff Layton static inline u64 ceph_present_ino(struct super_block *sb, u64 ino) 519ebce3eb2SJeff Layton { 520ebce3eb2SJeff Layton if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))) 521ebce3eb2SJeff Layton return ceph_ino_to_ino32(ino); 522ebce3eb2SJeff Layton return ino; 523ebce3eb2SJeff Layton } 524ebce3eb2SJeff Layton 525ebce3eb2SJeff Layton static inline u64 ceph_present_inode(struct inode *inode) 526ebce3eb2SJeff Layton { 527ebce3eb2SJeff Layton return ceph_present_ino(inode->i_sb, ceph_ino(inode)); 528ebce3eb2SJeff Layton } 529ebce3eb2SJeff Layton 5303d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 5313d14c5d2SYehuda Sadeh { 5323d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 5333d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 5343d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 5353d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 5363d14c5d2SYehuda Sadeh } 5373d14c5d2SYehuda Sadeh 538d4f6b31dSJeff Layton /* 539d4f6b31dSJeff Layton * The MDS reserves a set of inodes for its own usage. These should never 540d4f6b31dSJeff Layton * be accessible by clients, and so the MDS has no reason to ever hand these 541d4f6b31dSJeff Layton * out. The range is CEPH_MDS_INO_MDSDIR_OFFSET..CEPH_INO_SYSTEM_BASE. 542d4f6b31dSJeff Layton * 543d4f6b31dSJeff Layton * These come from src/mds/mdstypes.h in the ceph sources. 544d4f6b31dSJeff Layton */ 545d4f6b31dSJeff Layton #define CEPH_MAX_MDS 0x100 546d4f6b31dSJeff Layton #define CEPH_NUM_STRAY 10 547d4f6b31dSJeff Layton #define CEPH_MDS_INO_MDSDIR_OFFSET (1 * CEPH_MAX_MDS) 5480078ea3bSJeff Layton #define CEPH_MDS_INO_LOG_OFFSET (2 * CEPH_MAX_MDS) 549d4f6b31dSJeff Layton #define CEPH_INO_SYSTEM_BASE ((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY)) 550d4f6b31dSJeff Layton 551d4f6b31dSJeff Layton static inline bool ceph_vino_is_reserved(const struct ceph_vino vino) 552d4f6b31dSJeff Layton { 5530078ea3bSJeff Layton if (vino.ino >= CEPH_INO_SYSTEM_BASE || 5540078ea3bSJeff Layton vino.ino < CEPH_MDS_INO_MDSDIR_OFFSET) 555d4f6b31dSJeff Layton return false; 5560078ea3bSJeff Layton 5570078ea3bSJeff Layton /* Don't warn on mdsdirs */ 5580078ea3bSJeff Layton WARN_RATELIMIT(vino.ino >= CEPH_MDS_INO_LOG_OFFSET, 5590078ea3bSJeff Layton "Attempt to access reserved inode number 0x%llx", 5600078ea3bSJeff Layton vino.ino); 5610078ea3bSJeff Layton return true; 562d4f6b31dSJeff Layton } 563ebce3eb2SJeff Layton 5643d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 5653d14c5d2SYehuda Sadeh struct ceph_vino vino) 5663d14c5d2SYehuda Sadeh { 567d4f6b31dSJeff Layton if (ceph_vino_is_reserved(vino)) 568d4f6b31dSJeff Layton return NULL; 569d4f6b31dSJeff Layton 570ebce3eb2SJeff Layton /* 571ebce3eb2SJeff Layton * NB: The hashval will be run through the fs/inode.c hash function 572ebce3eb2SJeff Layton * anyway, so there is no need to squash the inode number down to 573ebce3eb2SJeff Layton * 32-bits first. Just use low-order bits on arches with 32-bit long. 574ebce3eb2SJeff Layton */ 575ebce3eb2SJeff Layton return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino); 5763d14c5d2SYehuda Sadeh } 5773d14c5d2SYehuda Sadeh 5783d14c5d2SYehuda Sadeh 5793d14c5d2SYehuda Sadeh /* 5803d14c5d2SYehuda Sadeh * Ceph inode. 5813d14c5d2SYehuda Sadeh */ 58210183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 58310183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 584daca8bdaSJeff Layton #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ 585daca8bdaSJeff Layton #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ 586daca8bdaSJeff Layton #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ 587daca8bdaSJeff Layton #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ 588c0e385b1SYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ 589c0e385b1SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ 590c0e385b1SYan, Zheng #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ 591c0e385b1SYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ 592c0e385b1SYan, Zheng #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ 593c0e385b1SYan, Zheng #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ 594891f3f5aSJeff Layton #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) 5955d6451b1SJeff Layton #define CEPH_I_SHUTDOWN (1 << 13) /* inode is no longer usable */ 596*68c62beeSXiubo Li #define CEPH_I_ASYNC_CHECK_CAPS (1 << 14) /* check caps immediately after async 597*68c62beeSXiubo Li creating finishes */ 59826544c62SJeff Layton 59926544c62SJeff Layton /* 6001cf89a8dSYan, Zheng * Masks of ceph inode work. 6011cf89a8dSYan, Zheng */ 602a8810cdcSJeff Layton #define CEPH_I_WORK_WRITEBACK 0 603a8810cdcSJeff Layton #define CEPH_I_WORK_INVALIDATE_PAGES 1 604a8810cdcSJeff Layton #define CEPH_I_WORK_VMTRUNCATE 2 605a8810cdcSJeff Layton #define CEPH_I_WORK_CHECK_CAPS 3 606a8810cdcSJeff Layton #define CEPH_I_WORK_FLUSH_SNAPS 4 6071cf89a8dSYan, Zheng 6081cf89a8dSYan, Zheng /* 60926544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 61026544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 61126544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 61226544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 61326544c62SJeff Layton * worried if a few slip through in either direction. 61426544c62SJeff Layton */ 61526544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 61626544c62SJeff Layton { 61726544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 61826544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 61926544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 62026544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 62126544c62SJeff Layton } 62226544c62SJeff Layton } 62326544c62SJeff Layton 62426544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci) 62526544c62SJeff Layton { 62626544c62SJeff Layton if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { 62726544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 62826544c62SJeff Layton ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; 62926544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 63026544c62SJeff Layton } 63126544c62SJeff Layton } 6323d14c5d2SYehuda Sadeh 6332f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 634fdd4e158SYan, Zheng long long release_count, 635fdd4e158SYan, Zheng long long ordered_count) 636de57606cSSage Weil { 63774960773SAndrea Parri /* 63874960773SAndrea Parri * Makes sure operations that setup readdir cache (update page 63974960773SAndrea Parri * cache and i_size) are strongly ordered w.r.t. the following 64074960773SAndrea Parri * atomic64_set() operations. 64174960773SAndrea Parri */ 64274960773SAndrea Parri smp_mb(); 643fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 644fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 645de57606cSSage Weil } 646de57606cSSage Weil 6472f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 648de57606cSSage Weil { 649fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 650fdd4e158SYan, Zheng } 651fdd4e158SYan, Zheng 652fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 653fdd4e158SYan, Zheng { 654fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 655de57606cSSage Weil } 656de57606cSSage Weil 6572f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 658de57606cSSage Weil { 659fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 660fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 6612f276c51SYan, Zheng } 662de57606cSSage Weil 66370db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 66470db4f36SYan, Zheng { 665fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 666fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 667fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 668fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 66970db4f36SYan, Zheng } 67070db4f36SYan, Zheng 6712f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 6722f276c51SYan, Zheng { 6732f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 6742f276c51SYan, Zheng } 6752f276c51SYan, Zheng 67670db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 6772f276c51SYan, Zheng { 678fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 679de57606cSSage Weil } 680de57606cSSage Weil 68170db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 68270db4f36SYan, Zheng { 683fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 684fdd4e158SYan, Zheng smp_rmb(); 68570db4f36SYan, Zheng return ret; 68670db4f36SYan, Zheng } 687de57606cSSage Weil 688de57606cSSage Weil /* find a specific frag @f */ 689de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 690de57606cSSage Weil u32 f); 691de57606cSSage Weil 692de57606cSSage Weil /* 693de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 694de57606cSSage Weil * exists 695de57606cSSage Weil */ 696de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 697de57606cSSage Weil struct ceph_inode_frag *pfrag, 698de57606cSSage Weil int *found); 699de57606cSSage Weil 7001e9c2eb6SYan, Zheng static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) 701de57606cSSage Weil { 702de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 703de57606cSSage Weil } 704de57606cSSage Weil 705de57606cSSage Weil /* 706de57606cSSage Weil * caps helpers 707de57606cSSage Weil */ 708de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 709de57606cSSage Weil { 710de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 711de57606cSSage Weil } 712de57606cSSage Weil 713de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 714de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 7151af16d54SXiubo Li extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, 7161af16d54SXiubo Li int t); 717de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 718de57606cSSage Weil struct ceph_cap *cap); 719de57606cSSage Weil 720de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 721de57606cSSage Weil { 722de57606cSSage Weil int issued; 723be655596SSage Weil spin_lock(&ci->i_ceph_lock); 724de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 725be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 726de57606cSSage Weil return issued; 727de57606cSSage Weil } 728de57606cSSage Weil 7291af16d54SXiubo Li static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, 7301af16d54SXiubo Li int mask, int touch) 731de57606cSSage Weil { 732de57606cSSage Weil int r; 733be655596SSage Weil spin_lock(&ci->i_ceph_lock); 7341af16d54SXiubo Li r = __ceph_caps_issued_mask_metric(ci, mask, touch); 735be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 736de57606cSSage Weil return r; 737de57606cSSage Weil } 738de57606cSSage Weil 739de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 740de57606cSSage Weil { 741de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 742de57606cSSage Weil } 743f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 744f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 745f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 746f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 747de57606cSSage Weil 7489563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 7499563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 750de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 751de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 752de57606cSSage Weil 753719a2514SYan, Zheng static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) 754719a2514SYan, Zheng { 755719a2514SYan, Zheng return ci->i_nr_by_mode[0]; 756719a2514SYan, Zheng } 757de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 758525d15e8SYan, Zheng extern int __ceph_caps_wanted(struct ceph_inode_info *ci); 759de57606cSSage Weil 760de57606cSSage Weil /* what the mds thinks we want */ 761c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 762de57606cSSage Weil 76337151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 76437151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 765fe33032dSYan, Zheng extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, 766fe33032dSYan, Zheng struct ceph_mount_options *fsopt); 767e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 76837151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 7697bf8f736SChengguang Xu extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, 77037151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 7713d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 772de57606cSSage Weil int *total, int *avail, int *used, 77385ccce43SSage Weil int *reserved, int *min); 774e19feff9SXiubo Li extern void change_auth_cap_ses(struct ceph_inode_info *ci, 775e19feff9SXiubo Li struct ceph_mds_session *session); 776de57606cSSage Weil 777de57606cSSage Weil 778de57606cSSage Weil 779de57606cSSage Weil /* 780de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 781de57606cSSage Weil */ 7824918b6d1SSage Weil #define CEPH_F_SYNC 1 7839cfa1098SSage Weil #define CEPH_F_ATEND 2 7844918b6d1SSage Weil 785de57606cSSage Weil struct ceph_file_info { 786252c6728SSage Weil short fmode; /* initialized on open */ 787252c6728SSage Weil short flags; /* CEPH_F_* */ 788de57606cSSage Weil 7895d988308SYan, Zheng spinlock_t rw_contexts_lock; 7905d988308SYan, Zheng struct list_head rw_contexts; 791f4b97866SYan, Zheng 79281f148a9SYan, Zheng u32 filp_gen; 793ff5d913dSYan, Zheng atomic_t num_locks; 794bb48bd4dSChengguang Xu }; 795bb48bd4dSChengguang Xu 796bb48bd4dSChengguang Xu struct ceph_dir_file_info { 797bb48bd4dSChengguang Xu struct ceph_file_info file_info; 7985d988308SYan, Zheng 799de57606cSSage Weil /* readdir: position within the dir */ 800de57606cSSage Weil u32 frag; 801de57606cSSage Weil struct ceph_mds_request *last_readdir; 802de57606cSSage Weil 803de57606cSSage Weil /* readdir: position within a frag */ 804f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 805de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 806fdd4e158SYan, Zheng long long dir_release_count; 807fdd4e158SYan, Zheng long long dir_ordered_count; 808fdd4e158SYan, Zheng int readdir_cache_idx; 809de57606cSSage Weil 810de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 811de57606cSSage Weil char *dir_info; 812de57606cSSage Weil int dir_info_len; 813de57606cSSage Weil }; 814de57606cSSage Weil 8155d988308SYan, Zheng struct ceph_rw_context { 8165d988308SYan, Zheng struct list_head list; 8175d988308SYan, Zheng struct task_struct *thread; 8185d988308SYan, Zheng int caps; 8195d988308SYan, Zheng }; 8205d988308SYan, Zheng 8215d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 8225d988308SYan, Zheng struct ceph_rw_context _name = { \ 8235d988308SYan, Zheng .thread = current, \ 8245d988308SYan, Zheng .caps = _caps, \ 8255d988308SYan, Zheng } 8265d988308SYan, Zheng 8275d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 8285d988308SYan, Zheng struct ceph_rw_context *ctx) 8295d988308SYan, Zheng { 8305d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8315d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 8325d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8335d988308SYan, Zheng } 8345d988308SYan, Zheng 8355d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 8365d988308SYan, Zheng struct ceph_rw_context *ctx) 8375d988308SYan, Zheng { 8385d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8395d988308SYan, Zheng list_del(&ctx->list); 8405d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8415d988308SYan, Zheng } 8425d988308SYan, Zheng 8435d988308SYan, Zheng static inline struct ceph_rw_context* 8445d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 8455d988308SYan, Zheng { 8465d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 8475d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8485d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 8495d988308SYan, Zheng if (ctx->thread == current) { 8505d988308SYan, Zheng found = ctx; 8515d988308SYan, Zheng break; 8525d988308SYan, Zheng } 8535d988308SYan, Zheng } 8545d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8555d988308SYan, Zheng return found; 8565d988308SYan, Zheng } 8575d988308SYan, Zheng 858fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 859fdd4e158SYan, Zheng struct page *page; 860fdd4e158SYan, Zheng struct dentry **dentries; 861fdd4e158SYan, Zheng int index; 862fdd4e158SYan, Zheng }; 863de57606cSSage Weil 864de57606cSSage Weil /* 865de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 866de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 867de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 868de57606cSSage Weil * the snapshots of their parents. 869de57606cSSage Weil * 870de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 871de57606cSSage Weil * per-realm list. 872de57606cSSage Weil */ 873de57606cSSage Weil struct ceph_snap_realm { 874de57606cSSage Weil u64 ino; 875e3161f17SLuis Henriques struct inode *inode; 876de57606cSSage Weil atomic_t nref; 877a105f00cSSage Weil struct rb_node node; 878a105f00cSSage Weil 879de57606cSSage Weil u64 created, seq; 880de57606cSSage Weil u64 parent_ino; 881de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 882de57606cSSage Weil 883de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 884aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 885de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 886aa711ee3SAlex Elder u32 num_snaps; 887de57606cSSage Weil 888de57606cSSage Weil struct ceph_snap_realm *parent; 889de57606cSSage Weil struct list_head children; /* list of child realms */ 890de57606cSSage Weil struct list_head child_item; 891de57606cSSage Weil 892de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 893de57606cSSage Weil 894ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 895ae00d4f3SSage Weil 89674a31df4SXiubo Li struct list_head rebuild_item; /* rebuild snap realms _downward_ in hierarchy */ 89774a31df4SXiubo Li 898de57606cSSage Weil /* the current set of snaps for this realm */ 899de57606cSSage Weil struct ceph_snap_context *cached_context; 900de57606cSSage Weil 901de57606cSSage Weil struct list_head inodes_with_caps; 902de57606cSSage Weil spinlock_t inodes_with_caps_lock; 903de57606cSSage Weil }; 904de57606cSSage Weil 9053d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 9063d14c5d2SYehuda Sadeh { 9073d14c5d2SYehuda Sadeh int congestion_kb; 908de57606cSSage Weil 909de57606cSSage Weil /* 9103d14c5d2SYehuda Sadeh * Copied from NFS 9113d14c5d2SYehuda Sadeh * 9123d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 9133d14c5d2SYehuda Sadeh * 9143d14c5d2SYehuda Sadeh * 64MB: 8192k 9153d14c5d2SYehuda Sadeh * 128MB: 11585k 9163d14c5d2SYehuda Sadeh * 256MB: 16384k 9173d14c5d2SYehuda Sadeh * 512MB: 23170k 9183d14c5d2SYehuda Sadeh * 1GB: 32768k 9193d14c5d2SYehuda Sadeh * 2GB: 46340k 9203d14c5d2SYehuda Sadeh * 4GB: 65536k 9213d14c5d2SYehuda Sadeh * 8GB: 92681k 9223d14c5d2SYehuda Sadeh * 16GB: 131072k 9233d14c5d2SYehuda Sadeh * 9243d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 9253d14c5d2SYehuda Sadeh * Limit the default to 256M 926de57606cSSage Weil */ 927ca79b0c2SArun KS congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); 9283d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 9293d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 9303d14c5d2SYehuda Sadeh 9313d14c5d2SYehuda Sadeh return congestion_kb; 932de57606cSSage Weil } 933de57606cSSage Weil 934de57606cSSage Weil 935d468e729SYan, Zheng /* super.c */ 936d468e729SYan, Zheng extern int ceph_force_reconnect(struct super_block *sb); 937de57606cSSage Weil /* snap.c */ 938de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 939de57606cSSage Weil u64 ino); 940de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 941de57606cSSage Weil struct ceph_snap_realm *realm); 942de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 943de57606cSSage Weil struct ceph_snap_realm *realm); 944de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 945982d6011SYan, Zheng void *p, void *e, bool deletion, 946982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 9470ba92e1cSJeff Layton void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm); 948de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 9492600d2ddSSage Weil struct ceph_mds_session *session, 950de57606cSSage Weil struct ceph_msg *msg); 951de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 952de57606cSSage Weil struct ceph_cap_snap *capsnap); 9535ed91587SXiubo Li extern void ceph_cleanup_global_and_empty_realms(struct ceph_mds_client *mdsc); 954de57606cSSage Weil 95575c9627eSYan, Zheng extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, 95675c9627eSYan, Zheng u64 snap); 95775c9627eSYan, Zheng extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, 95875c9627eSYan, Zheng struct ceph_snapid_map *sm); 95975c9627eSYan, Zheng extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); 96075c9627eSYan, Zheng extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); 961631ed4b0SJeff Layton void ceph_umount_begin(struct super_block *sb); 96275c9627eSYan, Zheng 96375c9627eSYan, Zheng 964de57606cSSage Weil /* 965de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 966de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 967de57606cSSage Weil */ 968de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 969de57606cSSage Weil { 970de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 97186056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 972de57606cSSage Weil ci_item)->writing; 973de57606cSSage Weil } 974de57606cSSage Weil 975de57606cSSage Weil /* inode.c */ 976966c7160SJeff Layton struct ceph_mds_reply_info_in; 977966c7160SJeff Layton struct ceph_mds_reply_dirfrag; 978966c7160SJeff Layton 979de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 980de57606cSSage Weil 981de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 98287bc5b89SYan, Zheng extern void ceph_evict_inode(struct inode *inode); 983cfa6d412SAl Viro extern void ceph_free_inode(struct inode *inode); 984de57606cSSage Weil 985de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 986de57606cSSage Weil struct ceph_vino vino); 987de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 988de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 989de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 990de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 9919bbeab41SArnd Bergmann u64 time_warp_seq, struct timespec64 *ctime, 9929bbeab41SArnd Bergmann struct timespec64 *mtime, 9939bbeab41SArnd Bergmann struct timespec64 *atime); 994966c7160SJeff Layton extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, 995966c7160SJeff Layton struct ceph_mds_reply_info_in *iinfo, 996966c7160SJeff Layton struct ceph_mds_reply_dirfrag *dirinfo, 997966c7160SJeff Layton struct ceph_mds_session *session, int cap_fmode, 998966c7160SJeff Layton struct ceph_cap_reservation *caps_reservation); 999de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 1000f5a03b08SJeff Layton struct ceph_mds_request *req); 1001de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 1002de57606cSSage Weil struct ceph_mds_session *session); 1003de57606cSSage Weil 1004de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 1005de57606cSSage Weil 1006efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 1007b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 100864f28c62SJeff Layton 100964f28c62SJeff Layton void ceph_queue_inode_work(struct inode *inode, int work_bit); 101064f28c62SJeff Layton 101164f28c62SJeff Layton static inline void ceph_queue_vmtruncate(struct inode *inode) 101264f28c62SJeff Layton { 101364f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE); 101464f28c62SJeff Layton } 101564f28c62SJeff Layton 101664f28c62SJeff Layton static inline void ceph_queue_invalidate(struct inode *inode) 101764f28c62SJeff Layton { 101864f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES); 101964f28c62SJeff Layton } 102064f28c62SJeff Layton 102164f28c62SJeff Layton static inline void ceph_queue_writeback(struct inode *inode) 102264f28c62SJeff Layton { 102364f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK); 102464f28c62SJeff Layton } 102564f28c62SJeff Layton 1026a8810cdcSJeff Layton static inline void ceph_queue_check_caps(struct inode *inode) 1027a8810cdcSJeff Layton { 1028a8810cdcSJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS); 1029a8810cdcSJeff Layton } 1030a8810cdcSJeff Layton 1031a8810cdcSJeff Layton static inline void ceph_queue_flush_snaps(struct inode *inode) 1032a8810cdcSJeff Layton { 1033a8810cdcSJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS); 1034a8810cdcSJeff Layton } 1035a8810cdcSJeff Layton 10365eed80fbSXiubo Li extern int ceph_try_to_choose_auth_mds(struct inode *inode, int mask); 103701deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 103801deead0SYan, Zheng int mask, bool force); 103901deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 104001deead0SYan, Zheng { 104101deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 104201deead0SYan, Zheng } 1043549c7297SChristian Brauner extern int ceph_permission(struct user_namespace *mnt_userns, 1044549c7297SChristian Brauner struct inode *inode, int mask); 1045a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 1046549c7297SChristian Brauner extern int ceph_setattr(struct user_namespace *mnt_userns, 1047549c7297SChristian Brauner struct dentry *dentry, struct iattr *attr); 1048549c7297SChristian Brauner extern int ceph_getattr(struct user_namespace *mnt_userns, 1049549c7297SChristian Brauner const struct path *path, struct kstat *stat, 1050a528d35eSDavid Howells u32 request_mask, unsigned int flags); 10515d6451b1SJeff Layton void ceph_inode_shutdown(struct inode *inode); 10525d6451b1SJeff Layton 10535d6451b1SJeff Layton static inline bool ceph_inode_is_shutdown(struct inode *inode) 10545d6451b1SJeff Layton { 10555d6451b1SJeff Layton unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags); 10565d6451b1SJeff Layton struct ceph_fs_client *fsc = ceph_inode_to_client(inode); 10575d6451b1SJeff Layton int state = READ_ONCE(fsc->mount_state); 10585d6451b1SJeff Layton 10595d6451b1SJeff Layton return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN; 10605d6451b1SJeff Layton } 1061de57606cSSage Weil 1062de57606cSSage Weil /* xattr.c */ 1063a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 10646ddf5f16SMilind Changire int ceph_do_getvxattr(struct inode *inode, const char *name, void *value, size_t size); 10657221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 1066de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 106712fe3ddaSLuis Henriques extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); 1068de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 1069b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 1070de57606cSSage Weil 10715c31e92dSYan, Zheng struct ceph_acl_sec_ctx { 10725c31e92dSYan, Zheng #ifdef CONFIG_CEPH_FS_POSIX_ACL 10735c31e92dSYan, Zheng void *default_acl; 10745c31e92dSYan, Zheng void *acl; 10755c31e92dSYan, Zheng #endif 1076ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 1077ac6713ccSYan, Zheng void *sec_ctx; 1078ac6713ccSYan, Zheng u32 sec_ctxlen; 1079ac6713ccSYan, Zheng #endif 10805c31e92dSYan, Zheng struct ceph_pagelist *pagelist; 10815c31e92dSYan, Zheng }; 10825c31e92dSYan, Zheng 1083315f2408SYan, Zheng #ifdef CONFIG_SECURITY 1084315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 1085315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 1086315f2408SYan, Zheng #else 1087315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 1088315f2408SYan, Zheng { 1089315f2408SYan, Zheng return false; 1090315f2408SYan, Zheng } 1091315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 1092315f2408SYan, Zheng { 1093315f2408SYan, Zheng return false; 1094315f2408SYan, Zheng } 1095315f2408SYan, Zheng #endif 1096315f2408SYan, Zheng 1097ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 1098ac6713ccSYan, Zheng extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1099ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx); 1100668959a5SJeff Layton static inline void ceph_security_invalidate_secctx(struct inode *inode) 1101668959a5SJeff Layton { 1102668959a5SJeff Layton security_inode_invalidate_secctx(inode); 1103668959a5SJeff Layton } 1104ac6713ccSYan, Zheng #else 1105ac6713ccSYan, Zheng static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1106ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx) 1107ac6713ccSYan, Zheng { 1108ac6713ccSYan, Zheng return 0; 1109ac6713ccSYan, Zheng } 1110ac6713ccSYan, Zheng static inline void ceph_security_invalidate_secctx(struct inode *inode) 1111ac6713ccSYan, Zheng { 1112ac6713ccSYan, Zheng } 1113ac6713ccSYan, Zheng #endif 11147221fe4cSGuangliang Zhao 11155c31e92dSYan, Zheng void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); 11167221fe4cSGuangliang Zhao 11175c31e92dSYan, Zheng /* acl.c */ 11187221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 11197221fe4cSGuangliang Zhao 11200cad6246SMiklos Szeredi struct posix_acl *ceph_get_acl(struct inode *, int, bool); 1121549c7297SChristian Brauner int ceph_set_acl(struct user_namespace *mnt_userns, 1122549c7297SChristian Brauner struct inode *inode, struct posix_acl *acl, int type); 1123b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 11245c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 11255c31e92dSYan, Zheng void ceph_init_inode_acls(struct inode *inode, 11265c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 1127c969d9bfSGuangliang Zhao 1128c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 1129c969d9bfSGuangliang Zhao { 1130c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 1131c969d9bfSGuangliang Zhao } 11327221fe4cSGuangliang Zhao 11337221fe4cSGuangliang Zhao #else 11347221fe4cSGuangliang Zhao 11357221fe4cSGuangliang Zhao #define ceph_get_acl NULL 113672466d0bSSage Weil #define ceph_set_acl NULL 11377221fe4cSGuangliang Zhao 1138b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 11395c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 11407221fe4cSGuangliang Zhao { 11417221fe4cSGuangliang Zhao return 0; 11427221fe4cSGuangliang Zhao } 1143b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 11445c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 1145b1ee94aaSYan, Zheng { 1146b1ee94aaSYan, Zheng } 11477221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 11487221fe4cSGuangliang Zhao { 11497221fe4cSGuangliang Zhao return 0; 11507221fe4cSGuangliang Zhao } 11517221fe4cSGuangliang Zhao 11527221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 11537221fe4cSGuangliang Zhao { 11547221fe4cSGuangliang Zhao } 11557221fe4cSGuangliang Zhao 11567221fe4cSGuangliang Zhao #endif 11577221fe4cSGuangliang Zhao 1158de57606cSSage Weil /* caps.c */ 1159de57606cSSage Weil extern const char *ceph_cap_string(int c); 1160de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 1161de57606cSSage Weil struct ceph_msg *msg); 1162d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 1163d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 1164d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 1165de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 1166135e671eSYan, Zheng unsigned issued, unsigned wanted, 1167de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 1168d9df2783SYan, Zheng struct ceph_cap **new_cap); 1169a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1170a76d0a9cSXiubo Li extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1171d6e47819SYan, Zheng extern void __ceph_remove_caps(struct ceph_inode_info *ci); 117237151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 117337151668SYehuda Sadeh struct ceph_cap *cap); 11749215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 1175de57606cSSage Weil 1176f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 117702c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 117802c24a82SJosef Bacik int datasync); 1179e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 1180e548e9b9SYan, Zheng struct ceph_mds_session *session); 1181de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 1182de57606cSSage Weil struct ceph_mds_session *session); 1183e8a4d267SJeff Layton void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, 1184e8a4d267SJeff Layton struct ceph_inode_info *ci); 11852bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 11862bc50259SGreg Farnum int mds); 118740dcf75eSJeff Layton extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, 118840dcf75eSJeff Layton bool snap_rwsem_locked); 1189de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 1190de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 1191a8810cdcSJeff Layton extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had); 1192e64f44a8SXiubo Li extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, 1193e64f44a8SXiubo Li int had); 1194de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 1195de57606cSSage Weil struct ceph_snap_context *snapc); 1196a6d37ccdSXiubo Li extern void __ceph_remove_capsnap(struct inode *inode, 1197a6d37ccdSXiubo Li struct ceph_cap_snap *capsnap, 1198a6d37ccdSXiubo Li bool *wake_ci, bool *wake_mdsc); 1199a6d37ccdSXiubo Li extern void ceph_remove_capsnap(struct inode *inode, 1200a6d37ccdSXiubo Li struct ceph_cap_snap *capsnap, 1201a6d37ccdSXiubo Li bool *wake_ci, bool *wake_mdsc); 1202ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 12030e294387SYan, Zheng struct ceph_mds_session **psession); 1204efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 1205e4b731ccSXiubo Li extern void ceph_check_caps(struct ceph_inode_info *ci, int flags); 1206bf2ba432SLuis Henriques extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 1207afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 12086ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 1209de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 1210de57606cSSage Weil int mds, int drop, int unless, int force); 1211de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 1212ca6c8ae0SJeff Layton struct inode *dir, 1213de57606cSSage Weil int mds, int drop, int unless); 1214de57606cSSage Weil 12155e3ded1bSYan, Zheng extern int ceph_get_caps(struct file *filp, int need, int want, 1216e72968e1SJeff Layton loff_t endoff, int *got); 12175e3ded1bSYan, Zheng extern int ceph_try_get_caps(struct inode *inode, 12182ee9dd95SLuis Henriques int need, int want, bool nonblock, int *got); 1219de57606cSSage Weil 1220de57606cSSage Weil /* for counting open files by mode */ 1221719a2514SYan, Zheng extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); 1222719a2514SYan, Zheng extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); 1223719a2514SYan, Zheng extern void __ceph_touch_fmode(struct ceph_inode_info *ci, 1224719a2514SYan, Zheng struct ceph_mds_client *mdsc, int fmode); 1225de57606cSSage Weil 1226de57606cSSage Weil /* addr.c */ 1227de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1228bc899ee1SDavid Howells extern const struct netfs_request_ops ceph_netfs_ops; 1229de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 1230083db6fdSDavid Howells extern int ceph_uninline_data(struct file *file); 12315e3ded1bSYan, Zheng extern int ceph_pool_perm_check(struct inode *inode, int need); 123210183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 123336e6da98SJeff Layton int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate); 1234de57606cSSage Weil 123548490776SXiubo Li static inline bool ceph_has_inline_data(struct ceph_inode_info *ci) 123648490776SXiubo Li { 123748490776SXiubo Li if (ci->i_inline_version == CEPH_INLINE_NONE || 123848490776SXiubo Li ci->i_inline_version == 1) /* initial version, no data */ 123948490776SXiubo Li return false; 124048490776SXiubo Li return true; 124148490776SXiubo Li } 124248490776SXiubo Li 1243de57606cSSage Weil /* file.c */ 1244de57606cSSage Weil extern const struct file_operations ceph_file_fops; 12459e0eb85dSAlex Elder 1246719a2514SYan, Zheng extern int ceph_renew_caps(struct inode *inode, int fmode); 1247de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 12485ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 124944907d79SAl Viro struct file *file, unsigned flags, umode_t mode); 1250de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 125131c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 125231c542a1SYan, Zheng char *data, size_t len); 125355f2a045SIlya Dryomov 1254de57606cSSage Weil /* dir.c */ 1255de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 125638c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1257de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 125838c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 125918fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1260de57606cSSage Weil 1261f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1262de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1263aa60cfc3SJeff Layton extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req, 12647a971e2cSJeff Layton struct dentry *dentry); 1265de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1266de57606cSSage Weil struct dentry *dentry, int err); 1267de57606cSSage Weil 126837c4efc1SYan, Zheng extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); 126937c4efc1SYan, Zheng extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); 127081a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 127137c4efc1SYan, Zheng extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); 1272e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1273fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1274de57606cSSage Weil 1275de57606cSSage Weil /* ioctl.c */ 1276de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1277de57606cSSage Weil 1278de57606cSSage Weil /* export.c */ 1279de57606cSSage Weil extern const struct export_operations ceph_export_ops; 12803886274aSLuis Henriques struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); 1281de57606cSSage Weil 128240819f6fSGreg Farnum /* locks.c */ 1283eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 128440819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 128540819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 128640819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 128739be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 128839be95e9SJim Schutt struct ceph_filelock *flocks, 128939be95e9SJim Schutt int num_fcntl_locks, 129039be95e9SJim Schutt int num_flock_locks); 129139be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 129239be95e9SJim Schutt struct ceph_pagelist *pagelist, 129339be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 129440819f6fSGreg Farnum 12953d14c5d2SYehuda Sadeh /* debugfs.c */ 12961a829ff2SGreg Kroah-Hartman extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); 12973d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 12983d14c5d2SYehuda Sadeh 1299fb18a575SLuis Henriques /* quota.c */ 130055ab5520SLuís Henriques 130155ab5520SLuís Henriques enum quota_get_realm { 130255ab5520SLuís Henriques QUOTA_GET_MAX_FILES, 130355ab5520SLuís Henriques QUOTA_GET_MAX_BYTES, 130455ab5520SLuís Henriques QUOTA_GET_ANY 130555ab5520SLuís Henriques }; 130655ab5520SLuís Henriques 130755ab5520SLuís Henriques static inline bool __ceph_has_quota(struct ceph_inode_info *ci, 130855ab5520SLuís Henriques enum quota_get_realm which) 1309d557c48dSLuis Henriques { 131055ab5520SLuís Henriques bool has_quota = false; 131155ab5520SLuís Henriques 131255ab5520SLuís Henriques switch (which) { 131355ab5520SLuís Henriques case QUOTA_GET_MAX_BYTES: 131455ab5520SLuís Henriques has_quota = !!ci->i_max_bytes; 131555ab5520SLuís Henriques break; 131655ab5520SLuís Henriques case QUOTA_GET_MAX_FILES: 131755ab5520SLuís Henriques has_quota = !!ci->i_max_files; 131855ab5520SLuís Henriques break; 131955ab5520SLuís Henriques default: 132055ab5520SLuís Henriques has_quota = !!(ci->i_max_files || ci->i_max_bytes); 132155ab5520SLuís Henriques } 132255ab5520SLuís Henriques return has_quota; 1323d557c48dSLuis Henriques } 1324d557c48dSLuis Henriques 1325d557c48dSLuis Henriques extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); 1326d557c48dSLuis Henriques 1327d557c48dSLuis Henriques static inline void __ceph_update_quota(struct ceph_inode_info *ci, 1328d557c48dSLuis Henriques u64 max_bytes, u64 max_files) 1329d557c48dSLuis Henriques { 1330d557c48dSLuis Henriques bool had_quota, has_quota; 133155ab5520SLuís Henriques had_quota = __ceph_has_quota(ci, QUOTA_GET_ANY); 1332d557c48dSLuis Henriques ci->i_max_bytes = max_bytes; 1333d557c48dSLuis Henriques ci->i_max_files = max_files; 133455ab5520SLuís Henriques has_quota = __ceph_has_quota(ci, QUOTA_GET_ANY); 1335d557c48dSLuis Henriques 1336d557c48dSLuis Henriques if (had_quota != has_quota) 1337874c8ca1SDavid Howells ceph_adjust_quota_realms_count(&ci->netfs.inode, has_quota); 1338d557c48dSLuis Henriques } 1339d557c48dSLuis Henriques 1340fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc, 1341fb18a575SLuis Henriques struct ceph_mds_session *session, 1342fb18a575SLuis Henriques struct ceph_msg *msg); 1343b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); 13446646ea1cSLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); 13452b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, 13462b83845fSLuis Henriques loff_t newlen); 13471ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, 13481ab302a0SLuis Henriques loff_t newlen); 13499122eed5SLuis Henriques extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, 13509122eed5SLuis Henriques struct kstatfs *buf); 13510c44a8e0SLuis Henriques extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); 1352fb18a575SLuis Henriques 1353de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1354