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> 22de57606cSSage Weil 233d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h> 24de57606cSSage Weil 25de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate 26de57606cSSage Weil * large volume sizes on 32-bit machines. */ 2792a49fb0SSage Weil #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ 28de57606cSSage Weil #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) 298e55ba8cSKotresh HR #define CEPH_4K_BLOCK_SHIFT 12 /* 4 KB */ 30de57606cSSage Weil 310b98acd6SIlya Dryomov #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blocklisted */ 323d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ 333d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ 343d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ 35ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ 36a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ 3799ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ 3810183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ 39e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ 409122eed5SLuis Henriques #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ 41ea4cdc54SLuis Henriques #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ 422ccb4546SJeff Layton #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ 4394cc0877SJeff Layton #define CEPH_MOUNT_OPT_NOPAGECACHE (1<<16) /* bypass pagecache altogether */ 446a259382SSage Weil 456f9718feSLuis Henriques #define CEPH_MOUNT_OPT_DEFAULT \ 466f9718feSLuis Henriques (CEPH_MOUNT_OPT_DCACHE | \ 47f7a67b46SJeff Layton CEPH_MOUNT_OPT_NOCOPYFROM | \ 48f7a67b46SJeff Layton CEPH_MOUNT_OPT_ASYNC_DIROPS) 49de57606cSSage Weil 503d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \ 512ccb4546SJeff Layton (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt 522ccb4546SJeff Layton #define ceph_clear_mount_opt(fsc, opt) \ 532ccb4546SJeff Layton (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt 543d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \ 553d14c5d2SYehuda Sadeh (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) 56de57606cSSage Weil 57aa187926SYan, Zheng /* max size of osd read request, limited by libceph */ 58aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN 5995cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size. 6095cca2b4SYan, Zheng * CEPH_MSG_MAX_DATA_LEN should be small enough. */ 6195cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN 627c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ 633d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT 1024 643d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) 653d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT ".snap" 66de57606cSSage Weil 674214fb15SYan, Zheng /* 684214fb15SYan, Zheng * Delay telling the MDS we no longer want caps, in case we reopen 694214fb15SYan, Zheng * the file. Delay a minimum amount of time, even if we send a cap 704214fb15SYan, Zheng * message for some other reason. Otherwise, take the oppotunity to 714214fb15SYan, Zheng * update the mds to avoid sending another message later. 724214fb15SYan, Zheng */ 734214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ 744214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ 754214fb15SYan, Zheng 763d14c5d2SYehuda Sadeh struct ceph_mount_options { 77ad8c28a9SJeff Layton unsigned int flags; 783d14c5d2SYehuda Sadeh 79ad8c28a9SJeff Layton unsigned int wsize; /* max write size */ 80ad8c28a9SJeff Layton unsigned int rsize; /* max read size */ 81ad8c28a9SJeff Layton unsigned int rasize; /* max readahead */ 82ad8c28a9SJeff Layton unsigned int congestion_kb; /* max writeback in flight */ 83ad8c28a9SJeff Layton unsigned int caps_wanted_delay_min, caps_wanted_delay_max; 84fe33032dSYan, Zheng int caps_max; 85ad8c28a9SJeff Layton unsigned int max_readdir; /* max readdir result (entries) */ 86ad8c28a9SJeff Layton unsigned int max_readdir_bytes; /* max readdir result (bytes) */ 873d14c5d2SYehuda Sadeh 887b19b4dbSVenky Shankar bool new_dev_syntax; 897b19b4dbSVenky Shankar 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 */ 997b19b4dbSVenky Shankar char *mon_addr; 100de57606cSSage Weil }; 101de57606cSSage Weil 1023d14c5d2SYehuda Sadeh struct ceph_fs_client { 103de57606cSSage Weil struct super_block *sb; 104de57606cSSage Weil 10518f473b3SXiubo Li struct list_head metric_wakeup; 10618f473b3SXiubo Li 1073d14c5d2SYehuda Sadeh struct ceph_mount_options *mount_options; 1083d14c5d2SYehuda Sadeh struct ceph_client *client; 1093d14c5d2SYehuda Sadeh 110aa5c7910SJeff Layton int mount_state; 11181f148a9SYan, Zheng 1120b98acd6SIlya Dryomov bool blocklisted; 113131d7eb4SYan, Zheng 11478beb0ffSLuis Henriques bool have_copy_from2; 11578beb0ffSLuis Henriques 11681f148a9SYan, Zheng u32 filp_gen; 117719784baSChengguang Xu loff_t max_file_size; 11885ccce43SSage Weil 1193d14c5d2SYehuda Sadeh struct ceph_mds_client *mdsc; 120de57606cSSage Weil 121e3ec8d68SYan, Zheng atomic_long_t writeback_count; 122503d4fa6SNeilBrown bool write_congested; 123e3ec8d68SYan, Zheng 1241cf89a8dSYan, Zheng struct workqueue_struct *inode_wq; 125e3ec8d68SYan, Zheng struct workqueue_struct *cap_wq; 126de57606cSSage Weil 1270743304dSSage Weil #ifdef CONFIG_DEBUG_FS 1283d14c5d2SYehuda Sadeh struct dentry *debugfs_dentry_lru, *debugfs_caps; 1292baba250SYehuda Sadeh struct dentry *debugfs_congestion_kb; 13006edf046SSage Weil struct dentry *debugfs_bdi; 1313d14c5d2SYehuda Sadeh struct dentry *debugfs_mdsc, *debugfs_mdsmap; 132247b1f19SXiubo Li struct dentry *debugfs_status; 13314ed9703SJohn Spray struct dentry *debugfs_mds_sessions; 134cbed4ff7SLuís Henriques struct dentry *debugfs_metrics_dir; 1350743304dSSage Weil #endif 13699ccbd22SMilosz Tanski 13799ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE 138400e1286SJeff Layton struct fscache_volume *fscache; 13999ccbd22SMilosz Tanski #endif 140de57606cSSage Weil }; 141de57606cSSage Weil 1423d14c5d2SYehuda Sadeh 143de57606cSSage Weil /* 144de57606cSSage Weil * File i/o capability. This tracks shared state with the metadata 145de57606cSSage Weil * server that allows us to cache or writeback attributes or to read 146de57606cSSage Weil * and write data. For any given inode, we should have one or more 147de57606cSSage Weil * capabilities, one issued by each metadata server, and our 148de57606cSSage Weil * cumulative access is the OR of all issued capabilities. 149de57606cSSage Weil * 150de57606cSSage Weil * Each cap is referenced by the inode's i_caps rbtree and by per-mds 151de57606cSSage Weil * session capability lists. 152de57606cSSage Weil */ 153de57606cSSage Weil struct ceph_cap { 154de57606cSSage Weil struct ceph_inode_info *ci; 155de57606cSSage Weil struct rb_node ci_node; /* per-ci cap tree */ 156de57606cSSage Weil struct ceph_mds_session *session; 157de57606cSSage Weil struct list_head session_caps; /* per-session caplist */ 158de57606cSSage Weil u64 cap_id; /* unique cap id (mds provided) */ 159745a8e3bSYan, Zheng union { 160745a8e3bSYan, Zheng /* in-use caps */ 161745a8e3bSYan, Zheng struct { 162de57606cSSage Weil int issued; /* latest, from the mds */ 163745a8e3bSYan, Zheng int implemented; /* implemented superset of 164745a8e3bSYan, Zheng issued (for revocation) */ 165c74d79afSJeff Layton int mds; /* mds index for this cap */ 166c74d79afSJeff Layton int mds_wanted; /* caps wanted from this mds */ 167745a8e3bSYan, Zheng }; 168745a8e3bSYan, Zheng /* caps to release */ 169745a8e3bSYan, Zheng struct { 170745a8e3bSYan, Zheng u64 cap_ino; 171745a8e3bSYan, Zheng int queue_release; 172745a8e3bSYan, Zheng }; 173745a8e3bSYan, Zheng }; 174685f9a5dSSage Weil u32 seq, issue_seq, mseq; 175685f9a5dSSage Weil u32 cap_gen; /* active/stale cycle */ 176de57606cSSage Weil unsigned long last_used; 177de57606cSSage Weil struct list_head caps_item; 178de57606cSSage Weil }; 179de57606cSSage Weil 180a0d93e32SYan, Zheng #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ 181a0d93e32SYan, Zheng #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ 182a0d93e32SYan, Zheng #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ 183de57606cSSage Weil 1840e294387SYan, Zheng struct ceph_cap_flush { 1850e294387SYan, Zheng u64 tid; 186b2f9fa1fSXiubo Li int caps; 187c8799fc4SYan, Zheng bool wake; /* wake up flush waiters when finish ? */ 188b2f9fa1fSXiubo Li bool is_capsnap; /* true means capsnap */ 1890e294387SYan, Zheng struct list_head g_list; // global 1900e294387SYan, Zheng struct list_head i_list; // per inode 1910e294387SYan, Zheng }; 1920e294387SYan, Zheng 193de57606cSSage Weil /* 194de57606cSSage Weil * Snapped cap state that is pending flush to mds. When a snapshot occurs, 195de57606cSSage Weil * we first complete any in-process sync writes and writeback any dirty 196de57606cSSage Weil * data before flushing the snapped state (tracked here) back to the MDS. 197de57606cSSage Weil */ 198de57606cSSage Weil struct ceph_cap_snap { 199805692d0SElena Reshetova refcount_t nref; 2000e294387SYan, Zheng struct list_head ci_item; 201de57606cSSage Weil 2020e294387SYan, Zheng struct ceph_cap_flush cap_flush; 2030e294387SYan, Zheng 2040e294387SYan, Zheng u64 follows; 205de57606cSSage Weil int issued, dirty; 206de57606cSSage Weil struct ceph_snap_context *context; 207de57606cSSage Weil 2085706b27dSAl Viro umode_t mode; 20905cb11c1SEric W. Biederman kuid_t uid; 21005cb11c1SEric W. Biederman kgid_t gid; 211de57606cSSage Weil 2124a625be4SSage Weil struct ceph_buffer *xattr_blob; 213de57606cSSage Weil u64 xattr_version; 214de57606cSSage Weil 215de57606cSSage Weil u64 size; 216176c77c9SJeff Layton u64 change_attr; 217ec62b894SJeff Layton struct timespec64 mtime, atime, ctime, btime; 218de57606cSSage Weil u64 time_warp_seq; 2195f743e45SYan, Zheng u64 truncate_size; 2205f743e45SYan, Zheng u32 truncate_seq; 221de57606cSSage Weil int writing; /* a sync write is still in progress */ 222de57606cSSage Weil int dirty_pages; /* dirty pages awaiting writeback */ 223e20d258dSYan, Zheng bool inline_data; 22486056090SYan, Zheng bool need_flush; 225de57606cSSage Weil }; 226de57606cSSage Weil 227de57606cSSage Weil static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) 228de57606cSSage Weil { 229805692d0SElena Reshetova if (refcount_dec_and_test(&capsnap->nref)) { 2304a625be4SSage Weil if (capsnap->xattr_blob) 2314a625be4SSage Weil ceph_buffer_put(capsnap->xattr_blob); 232ab58a5a1SXiubo Li kmem_cache_free(ceph_cap_snap_cachep, capsnap); 233de57606cSSage Weil } 2344a625be4SSage Weil } 235de57606cSSage Weil 236de57606cSSage Weil /* 237de57606cSSage Weil * The frag tree describes how a directory is fragmented, potentially across 238de57606cSSage Weil * multiple metadata servers. It is also used to indicate points where 239de57606cSSage Weil * metadata authority is delegated, and whether/where metadata is replicated. 240de57606cSSage Weil * 241de57606cSSage Weil * A _leaf_ frag will be present in the i_fragtree IFF there is 242de57606cSSage Weil * delegation info. That is, if mds >= 0 || ndist > 0. 243de57606cSSage Weil */ 244de57606cSSage Weil #define CEPH_MAX_DIRFRAG_REP 4 245de57606cSSage Weil 246de57606cSSage Weil struct ceph_inode_frag { 247de57606cSSage Weil struct rb_node node; 248de57606cSSage Weil 249de57606cSSage Weil /* fragtree state */ 250de57606cSSage Weil u32 frag; 251de57606cSSage Weil int split_by; /* i.e. 2^(split_by) children */ 252de57606cSSage Weil 253de57606cSSage Weil /* delegation and replication info */ 254de57606cSSage Weil int mds; /* -1 if same authority as parent */ 255de57606cSSage Weil int ndist; /* >0 if replicated */ 256de57606cSSage Weil int dist[CEPH_MAX_DIRFRAG_REP]; 257de57606cSSage Weil }; 258de57606cSSage Weil 259de57606cSSage Weil /* 260de57606cSSage Weil * We cache inode xattrs as an encoded blob until they are first used, 261de57606cSSage Weil * at which point we parse them into an rbtree. 262de57606cSSage Weil */ 263de57606cSSage Weil struct ceph_inode_xattr { 264de57606cSSage Weil struct rb_node node; 265de57606cSSage Weil 266de57606cSSage Weil const char *name; 267de57606cSSage Weil int name_len; 268de57606cSSage Weil const char *val; 269de57606cSSage Weil int val_len; 270de57606cSSage Weil int dirty; 271de57606cSSage Weil 272de57606cSSage Weil int should_free_name; 273de57606cSSage Weil int should_free_val; 274de57606cSSage Weil }; 275de57606cSSage Weil 2763d14c5d2SYehuda Sadeh /* 2773d14c5d2SYehuda Sadeh * Ceph dentry state 2783d14c5d2SYehuda Sadeh */ 2793d14c5d2SYehuda Sadeh struct ceph_dentry_info { 28037c4efc1SYan, Zheng struct dentry *dentry; 2813d14c5d2SYehuda Sadeh struct ceph_mds_session *lease_session; 28237c4efc1SYan, Zheng struct list_head lease_list; 28337c4efc1SYan, Zheng unsigned flags; 28497aeb6bfSYan, Zheng int lease_shared_gen; 28597aeb6bfSYan, Zheng u32 lease_gen; 2863d14c5d2SYehuda Sadeh u32 lease_seq; 2873d14c5d2SYehuda Sadeh unsigned long lease_renew_after, lease_renew_from; 2889b16f03cSMiklos Szeredi unsigned long time; 2893d14c5d2SYehuda Sadeh u64 offset; 2903d14c5d2SYehuda Sadeh }; 2913d14c5d2SYehuda Sadeh 29237c4efc1SYan, Zheng #define CEPH_DENTRY_REFERENCED 1 29337c4efc1SYan, Zheng #define CEPH_DENTRY_LEASE_LIST 2 29437c4efc1SYan, Zheng #define CEPH_DENTRY_SHRINK_LIST 4 295f5e17aedSJeff Layton #define CEPH_DENTRY_PRIMARY_LINK 8 29637c4efc1SYan, Zheng 297de57606cSSage Weil struct ceph_inode_xattrs_info { 298de57606cSSage Weil /* 299de57606cSSage Weil * (still encoded) xattr blob. we avoid the overhead of parsing 300de57606cSSage Weil * this until someone actually calls getxattr, etc. 301de57606cSSage Weil * 302de57606cSSage Weil * blob->vec.iov_len == 4 implies there are no xattrs; blob == 303de57606cSSage Weil * NULL means we don't know. 304de57606cSSage Weil */ 305de57606cSSage Weil struct ceph_buffer *blob, *prealloc_blob; 306de57606cSSage Weil 307de57606cSSage Weil struct rb_root index; 308de57606cSSage Weil bool dirty; 309de57606cSSage Weil int count; 310de57606cSSage Weil int names_size; 311de57606cSSage Weil int vals_size; 312de57606cSSage Weil u64 version, index_version; 313de57606cSSage Weil }; 314de57606cSSage Weil 315de57606cSSage Weil /* 316de57606cSSage Weil * Ceph inode. 317de57606cSSage Weil */ 318de57606cSSage Weil struct ceph_inode_info { 319bc899ee1SDavid Howells struct { 320bc899ee1SDavid Howells /* These must be contiguous */ 321bc899ee1SDavid Howells struct inode vfs_inode; 322bc899ee1SDavid Howells struct netfs_i_context netfs_ctx; /* Netfslib context */ 323bc899ee1SDavid Howells }; 324de57606cSSage Weil struct ceph_vino i_vino; /* ceph ino + snap */ 325de57606cSSage Weil 326be655596SSage Weil spinlock_t i_ceph_lock; 327be655596SSage Weil 328de57606cSSage Weil u64 i_version; 32931c542a1SYan, Zheng u64 i_inline_version; 330de57606cSSage Weil u32 i_time_warp_seq; 331de57606cSSage Weil 332891f3f5aSJeff Layton unsigned long i_ceph_flags; 333fdd4e158SYan, Zheng atomic64_t i_release_count; 334fdd4e158SYan, Zheng atomic64_t i_ordered_count; 335fdd4e158SYan, Zheng atomic64_t i_complete_seq[2]; 336de57606cSSage Weil 3376c0f3af7SSage Weil struct ceph_dir_layout i_dir_layout; 338de57606cSSage Weil struct ceph_file_layout i_layout; 339785892feSJeff Layton struct ceph_file_layout i_cached_layout; // for async creates 340de57606cSSage Weil char *i_symlink; 341de57606cSSage Weil 342de57606cSSage Weil /* for dirs */ 3439bbeab41SArnd Bergmann struct timespec64 i_rctime; 344e7f72952SYanhu Cao u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps; 345de57606cSSage Weil u64 i_files, i_subdirs; 346de57606cSSage Weil 347fb18a575SLuis Henriques /* quotas */ 348fb18a575SLuis Henriques u64 i_max_bytes, i_max_files; 349fb18a575SLuis Henriques 35008796873SYan, Zheng s32 i_dir_pin; 35108796873SYan, Zheng 352de57606cSSage Weil struct rb_root i_fragtree; 3531b1bc16dSYan, Zheng int i_fragtree_nsplits; 354de57606cSSage Weil struct mutex i_fragtree_mutex; 355de57606cSSage Weil 356de57606cSSage Weil struct ceph_inode_xattrs_info i_xattrs; 357de57606cSSage Weil 358be655596SSage Weil /* capabilities. protected _both_ by i_ceph_lock and cap->session's 359de57606cSSage Weil * s_mutex. */ 360de57606cSSage Weil struct rb_root i_caps; /* cap list */ 361de57606cSSage Weil struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ 362de57606cSSage Weil unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ 3631cf03a68SJeff Layton 3641cf03a68SJeff Layton /* 365f1f565a2SRandy Dunlap * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty 3661cf03a68SJeff Layton * is protected by the mdsc->cap_dirty_lock, but each individual item 3671cf03a68SJeff Layton * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty 3681cf03a68SJeff Layton * requires the mdsc->cap_dirty_lock. List presence for an item can 3691cf03a68SJeff Layton * be tested under the i_ceph_lock. Changing anything requires both. 3704fb5dda3SJeff Layton */ 3711cf03a68SJeff Layton struct list_head i_dirty_item; 3721cf03a68SJeff Layton 3731cf03a68SJeff Layton /* 374829ad4dbSJeff Layton * Link to session's s_cap_flushing list. Protected in a similar 375829ad4dbSJeff Layton * fashion to i_dirty_item, but also by the s_mutex for changes. The 376829ad4dbSJeff Layton * s_cap_flushing list can be walked while holding either the s_mutex 377829ad4dbSJeff Layton * or msdc->cap_dirty_lock. List presence can also be checked while 378829ad4dbSJeff Layton * holding the i_ceph_lock for this inode. 3791cf03a68SJeff Layton */ 3801cf03a68SJeff Layton struct list_head i_flushing_item; 3811cf03a68SJeff Layton 382de57606cSSage Weil /* we need to track cap writeback on a per-cap-bit basis, to allow 383de57606cSSage Weil * overlapping, pipelined cap flushes to the mds. we can probably 384de57606cSSage Weil * reduce the tid to 8 bits if we're concerned about inode size. */ 385f66fd9f0SYan, Zheng struct ceph_cap_flush *i_prealloc_cap_flush; 386e4500b5eSYan, Zheng struct list_head i_cap_flush_list; 387de57606cSSage Weil wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ 388de57606cSSage Weil unsigned long i_hold_caps_max; /* jiffies */ 389de57606cSSage Weil struct list_head i_cap_delay_list; /* for delayed cap release to mds */ 390de57606cSSage Weil struct ceph_cap_reservation i_cap_migration_resv; 391de57606cSSage Weil struct list_head i_cap_snaps; /* snapped state pending flush to mds */ 3927d8cb26dSSage Weil struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or 3937d8cb26dSSage Weil dirty|flushing caps */ 394de57606cSSage Weil unsigned i_snap_caps; /* cap bits for snapped files */ 395de57606cSSage Weil 396719a2514SYan, Zheng unsigned long i_last_rd; 397719a2514SYan, Zheng unsigned long i_last_wr; 398774a6a11SYan, Zheng int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ 399de57606cSSage Weil 400b0d7c223SYan, Zheng struct mutex i_truncate_mutex; 401de57606cSSage Weil u32 i_truncate_seq; /* last truncate to smaller size */ 402de57606cSSage Weil u64 i_truncate_size; /* and the size we last truncated down to */ 403de57606cSSage Weil int i_truncate_pending; /* still need to call vmtruncate */ 404de57606cSSage Weil 405de57606cSSage Weil u64 i_max_size; /* max file size authorized by mds */ 406de57606cSSage Weil u64 i_reported_size; /* (max_)size reported to or requested of mds */ 407de57606cSSage Weil u64 i_wanted_max_size; /* offset we'd like to write too */ 408de57606cSSage Weil u64 i_requested_max_size; /* max_size we've requested */ 409de57606cSSage Weil 410de57606cSSage Weil /* held references to caps */ 411de57606cSSage Weil int i_pin_ref; 412f85122afSJeff Layton int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; 413de57606cSSage Weil int i_wrbuffer_ref, i_wrbuffer_ref_head; 41489aa5930SYan, Zheng atomic_t i_filelock_ref; 41597aeb6bfSYan, Zheng atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ 416cd045cb4SSage Weil u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ 417de57606cSSage Weil u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ 418de57606cSSage Weil 419de57606cSSage Weil struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ 42068cd5b4bSYan, Zheng struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ 421de57606cSSage Weil spinlock_t i_unsafe_lock; 422de57606cSSage Weil 42375c9627eSYan, Zheng union { 424de57606cSSage Weil struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ 42575c9627eSYan, Zheng struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ 42675c9627eSYan, Zheng }; 427de57606cSSage Weil struct list_head i_snap_realm_item; 428de57606cSSage Weil struct list_head i_snap_flush_item; 429245ce991SJeff Layton struct timespec64 i_btime; 430193e7b37SDavid Disseldorp struct timespec64 i_snap_btime; 431de57606cSSage Weil 4321cf89a8dSYan, Zheng struct work_struct i_work; 4331cf89a8dSYan, Zheng unsigned long i_work_mask; 434de57606cSSage Weil }; 435de57606cSSage Weil 436721d5c13SJeff Layton static inline struct ceph_inode_info * 437721d5c13SJeff Layton ceph_inode(const struct inode *inode) 438de57606cSSage Weil { 439fbbccec9SNoah Watkins return container_of(inode, struct ceph_inode_info, vfs_inode); 440de57606cSSage Weil } 441de57606cSSage Weil 442721d5c13SJeff Layton static inline struct ceph_fs_client * 443721d5c13SJeff Layton ceph_inode_to_client(const struct inode *inode) 444ad1fee96SYehuda Sadeh { 445ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)inode->i_sb->s_fs_info; 446ad1fee96SYehuda Sadeh } 447ad1fee96SYehuda Sadeh 448721d5c13SJeff Layton static inline struct ceph_fs_client * 449721d5c13SJeff Layton ceph_sb_to_client(const struct super_block *sb) 450ad1fee96SYehuda Sadeh { 451ad1fee96SYehuda Sadeh return (struct ceph_fs_client *)sb->s_fs_info; 452ad1fee96SYehuda Sadeh } 453ad1fee96SYehuda Sadeh 4542678da88SXiubo Li static inline struct ceph_mds_client * 4552678da88SXiubo Li ceph_sb_to_mdsc(const struct super_block *sb) 4562678da88SXiubo Li { 4572678da88SXiubo Li return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc; 4582678da88SXiubo Li } 4592678da88SXiubo Li 460721d5c13SJeff Layton static inline struct ceph_vino 461721d5c13SJeff Layton ceph_vino(const struct inode *inode) 4623d14c5d2SYehuda Sadeh { 4633d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino; 4643d14c5d2SYehuda Sadeh } 4653d14c5d2SYehuda Sadeh 466ebce3eb2SJeff Layton static inline u32 ceph_ino_to_ino32(u64 vino) 467ad1fee96SYehuda Sadeh { 4683310f754SAmon Ott u32 ino = vino & 0xffffffff; 4693310f754SAmon Ott ino ^= vino >> 32; 470ad1fee96SYehuda Sadeh if (!ino) 471a661fc56SAmon Ott ino = 2; 472ad1fee96SYehuda Sadeh return ino; 473ad1fee96SYehuda Sadeh } 474ad1fee96SYehuda Sadeh 475ad1fee96SYehuda Sadeh /* 476ebce3eb2SJeff Layton * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on 477ebce3eb2SJeff Layton * some arches. We generally do not use this value inside the ceph driver, but 478ebce3eb2SJeff Layton * we do want to set it to something, so that generic vfs code has an 479ebce3eb2SJeff Layton * appropriate value for tracepoints and the like. 4803d14c5d2SYehuda Sadeh */ 481ebce3eb2SJeff Layton static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino) 4823d14c5d2SYehuda Sadeh { 483ebce3eb2SJeff Layton if (sizeof(ino_t) == sizeof(u32)) 4843310f754SAmon Ott return ceph_ino_to_ino32(vino.ino); 4853310f754SAmon Ott return (ino_t)vino.ino; 4863d14c5d2SYehuda Sadeh } 4873d14c5d2SYehuda Sadeh 4883d14c5d2SYehuda Sadeh /* for printf-style formatting */ 4893d14c5d2SYehuda Sadeh #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap 4903d14c5d2SYehuda Sadeh 4913d14c5d2SYehuda Sadeh static inline u64 ceph_ino(struct inode *inode) 4923d14c5d2SYehuda Sadeh { 4933d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.ino; 4943d14c5d2SYehuda Sadeh } 495ebce3eb2SJeff Layton 4963d14c5d2SYehuda Sadeh static inline u64 ceph_snap(struct inode *inode) 4973d14c5d2SYehuda Sadeh { 4983d14c5d2SYehuda Sadeh return ceph_inode(inode)->i_vino.snap; 4993d14c5d2SYehuda Sadeh } 5003d14c5d2SYehuda Sadeh 501ebce3eb2SJeff Layton /** 502ebce3eb2SJeff Layton * ceph_present_ino - format an inode number for presentation to userland 503ebce3eb2SJeff Layton * @sb: superblock where the inode lives 504ebce3eb2SJeff Layton * @ino: inode number to (possibly) convert 505ebce3eb2SJeff Layton * 506ebce3eb2SJeff Layton * If the user mounted with the ino32 option, then the 64-bit value needs 507ebce3eb2SJeff Layton * to be converted to something that can fit inside 32 bits. Note that 508ebce3eb2SJeff Layton * internal kernel code never uses this value, so this is entirely for 509ebce3eb2SJeff Layton * userland consumption. 510ebce3eb2SJeff Layton */ 511ebce3eb2SJeff Layton static inline u64 ceph_present_ino(struct super_block *sb, u64 ino) 512ebce3eb2SJeff Layton { 513ebce3eb2SJeff Layton if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))) 514ebce3eb2SJeff Layton return ceph_ino_to_ino32(ino); 515ebce3eb2SJeff Layton return ino; 516ebce3eb2SJeff Layton } 517ebce3eb2SJeff Layton 518ebce3eb2SJeff Layton static inline u64 ceph_present_inode(struct inode *inode) 519ebce3eb2SJeff Layton { 520ebce3eb2SJeff Layton return ceph_present_ino(inode->i_sb, ceph_ino(inode)); 521ebce3eb2SJeff Layton } 522ebce3eb2SJeff Layton 5233d14c5d2SYehuda Sadeh static inline int ceph_ino_compare(struct inode *inode, void *data) 5243d14c5d2SYehuda Sadeh { 5253d14c5d2SYehuda Sadeh struct ceph_vino *pvino = (struct ceph_vino *)data; 5263d14c5d2SYehuda Sadeh struct ceph_inode_info *ci = ceph_inode(inode); 5273d14c5d2SYehuda Sadeh return ci->i_vino.ino == pvino->ino && 5283d14c5d2SYehuda Sadeh ci->i_vino.snap == pvino->snap; 5293d14c5d2SYehuda Sadeh } 5303d14c5d2SYehuda Sadeh 531d4f6b31dSJeff Layton /* 532d4f6b31dSJeff Layton * The MDS reserves a set of inodes for its own usage. These should never 533d4f6b31dSJeff Layton * be accessible by clients, and so the MDS has no reason to ever hand these 534d4f6b31dSJeff Layton * out. The range is CEPH_MDS_INO_MDSDIR_OFFSET..CEPH_INO_SYSTEM_BASE. 535d4f6b31dSJeff Layton * 536d4f6b31dSJeff Layton * These come from src/mds/mdstypes.h in the ceph sources. 537d4f6b31dSJeff Layton */ 538d4f6b31dSJeff Layton #define CEPH_MAX_MDS 0x100 539d4f6b31dSJeff Layton #define CEPH_NUM_STRAY 10 540d4f6b31dSJeff Layton #define CEPH_MDS_INO_MDSDIR_OFFSET (1 * CEPH_MAX_MDS) 5410078ea3bSJeff Layton #define CEPH_MDS_INO_LOG_OFFSET (2 * CEPH_MAX_MDS) 542d4f6b31dSJeff Layton #define CEPH_INO_SYSTEM_BASE ((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY)) 543d4f6b31dSJeff Layton 544d4f6b31dSJeff Layton static inline bool ceph_vino_is_reserved(const struct ceph_vino vino) 545d4f6b31dSJeff Layton { 5460078ea3bSJeff Layton if (vino.ino >= CEPH_INO_SYSTEM_BASE || 5470078ea3bSJeff Layton vino.ino < CEPH_MDS_INO_MDSDIR_OFFSET) 548d4f6b31dSJeff Layton return false; 5490078ea3bSJeff Layton 5500078ea3bSJeff Layton /* Don't warn on mdsdirs */ 5510078ea3bSJeff Layton WARN_RATELIMIT(vino.ino >= CEPH_MDS_INO_LOG_OFFSET, 5520078ea3bSJeff Layton "Attempt to access reserved inode number 0x%llx", 5530078ea3bSJeff Layton vino.ino); 5540078ea3bSJeff Layton return true; 555d4f6b31dSJeff Layton } 556ebce3eb2SJeff Layton 5573d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb, 5583d14c5d2SYehuda Sadeh struct ceph_vino vino) 5593d14c5d2SYehuda Sadeh { 560d4f6b31dSJeff Layton if (ceph_vino_is_reserved(vino)) 561d4f6b31dSJeff Layton return NULL; 562d4f6b31dSJeff Layton 563ebce3eb2SJeff Layton /* 564ebce3eb2SJeff Layton * NB: The hashval will be run through the fs/inode.c hash function 565ebce3eb2SJeff Layton * anyway, so there is no need to squash the inode number down to 566ebce3eb2SJeff Layton * 32-bits first. Just use low-order bits on arches with 32-bit long. 567ebce3eb2SJeff Layton */ 568ebce3eb2SJeff Layton return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino); 5693d14c5d2SYehuda Sadeh } 5703d14c5d2SYehuda Sadeh 5713d14c5d2SYehuda Sadeh 5723d14c5d2SYehuda Sadeh /* 5733d14c5d2SYehuda Sadeh * Ceph inode. 5743d14c5d2SYehuda Sadeh */ 57510183a69SYan, Zheng #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ 57610183a69SYan, Zheng #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ 577daca8bdaSJeff Layton #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ 578daca8bdaSJeff Layton #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ 579daca8bdaSJeff Layton #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ 580daca8bdaSJeff Layton #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ 581c0e385b1SYan, Zheng #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ 582c0e385b1SYan, Zheng #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ 583c0e385b1SYan, Zheng #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ 584c0e385b1SYan, Zheng #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ 585c0e385b1SYan, Zheng #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ 586c0e385b1SYan, Zheng #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ 587891f3f5aSJeff Layton #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) 5885d6451b1SJeff Layton #define CEPH_I_SHUTDOWN (1 << 13) /* inode is no longer usable */ 58926544c62SJeff Layton 59026544c62SJeff Layton /* 5911cf89a8dSYan, Zheng * Masks of ceph inode work. 5921cf89a8dSYan, Zheng */ 593a8810cdcSJeff Layton #define CEPH_I_WORK_WRITEBACK 0 594a8810cdcSJeff Layton #define CEPH_I_WORK_INVALIDATE_PAGES 1 595a8810cdcSJeff Layton #define CEPH_I_WORK_VMTRUNCATE 2 596a8810cdcSJeff Layton #define CEPH_I_WORK_CHECK_CAPS 3 597a8810cdcSJeff Layton #define CEPH_I_WORK_FLUSH_SNAPS 4 5981cf89a8dSYan, Zheng 5991cf89a8dSYan, Zheng /* 60026544c62SJeff Layton * We set the ERROR_WRITE bit when we start seeing write errors on an inode 60126544c62SJeff Layton * and then clear it when they start succeeding. Note that we do a lockless 60226544c62SJeff Layton * check first, and only take the lock if it looks like it needs to be changed. 60326544c62SJeff Layton * The write submission code just takes this as a hint, so we're not too 60426544c62SJeff Layton * worried if a few slip through in either direction. 60526544c62SJeff Layton */ 60626544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci) 60726544c62SJeff Layton { 60826544c62SJeff Layton if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { 60926544c62SJeff Layton spin_lock(&ci->i_ceph_lock); 61026544c62SJeff Layton ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; 61126544c62SJeff Layton spin_unlock(&ci->i_ceph_lock); 61226544c62SJeff Layton } 61326544c62SJeff Layton } 61426544c62SJeff Layton 61526544c62SJeff Layton static inline void ceph_clear_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 } 6233d14c5d2SYehuda Sadeh 6242f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, 625fdd4e158SYan, Zheng long long release_count, 626fdd4e158SYan, Zheng long long ordered_count) 627de57606cSSage Weil { 62874960773SAndrea Parri /* 62974960773SAndrea Parri * Makes sure operations that setup readdir cache (update page 63074960773SAndrea Parri * cache and i_size) are strongly ordered w.r.t. the following 63174960773SAndrea Parri * atomic64_set() operations. 63274960773SAndrea Parri */ 63374960773SAndrea Parri smp_mb(); 634fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[0], release_count); 635fdd4e158SYan, Zheng atomic64_set(&ci->i_complete_seq[1], ordered_count); 636de57606cSSage Weil } 637de57606cSSage Weil 6382f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) 639de57606cSSage Weil { 640fdd4e158SYan, Zheng atomic64_inc(&ci->i_release_count); 641fdd4e158SYan, Zheng } 642fdd4e158SYan, Zheng 643fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) 644fdd4e158SYan, Zheng { 645fdd4e158SYan, Zheng atomic64_inc(&ci->i_ordered_count); 646de57606cSSage Weil } 647de57606cSSage Weil 6482f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) 649de57606cSSage Weil { 650fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 651fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count); 6522f276c51SYan, Zheng } 653de57606cSSage Weil 65470db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) 65570db4f36SYan, Zheng { 656fdd4e158SYan, Zheng return atomic64_read(&ci->i_complete_seq[0]) == 657fdd4e158SYan, Zheng atomic64_read(&ci->i_release_count) && 658fdd4e158SYan, Zheng atomic64_read(&ci->i_complete_seq[1]) == 659fdd4e158SYan, Zheng atomic64_read(&ci->i_ordered_count); 66070db4f36SYan, Zheng } 66170db4f36SYan, Zheng 6622f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode) 6632f276c51SYan, Zheng { 6642f276c51SYan, Zheng __ceph_dir_clear_complete(ceph_inode(inode)); 6652f276c51SYan, Zheng } 6662f276c51SYan, Zheng 66770db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode) 6682f276c51SYan, Zheng { 669fdd4e158SYan, Zheng __ceph_dir_clear_ordered(ceph_inode(inode)); 670de57606cSSage Weil } 671de57606cSSage Weil 67270db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode) 67370db4f36SYan, Zheng { 674fdd4e158SYan, Zheng bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); 675fdd4e158SYan, Zheng smp_rmb(); 67670db4f36SYan, Zheng return ret; 67770db4f36SYan, Zheng } 678de57606cSSage Weil 679de57606cSSage Weil /* find a specific frag @f */ 680de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, 681de57606cSSage Weil u32 f); 682de57606cSSage Weil 683de57606cSSage Weil /* 684de57606cSSage Weil * choose fragment for value @v. copy frag content to pfrag, if leaf 685de57606cSSage Weil * exists 686de57606cSSage Weil */ 687de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, 688de57606cSSage Weil struct ceph_inode_frag *pfrag, 689de57606cSSage Weil int *found); 690de57606cSSage Weil 6911e9c2eb6SYan, Zheng static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) 692de57606cSSage Weil { 693de57606cSSage Weil return (struct ceph_dentry_info *)dentry->d_fsdata; 694de57606cSSage Weil } 695de57606cSSage Weil 696de57606cSSage Weil /* 697de57606cSSage Weil * caps helpers 698de57606cSSage Weil */ 699de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) 700de57606cSSage Weil { 701de57606cSSage Weil return !RB_EMPTY_ROOT(&ci->i_caps); 702de57606cSSage Weil } 703de57606cSSage Weil 704de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); 705de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); 7061af16d54SXiubo Li extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, 7071af16d54SXiubo Li int t); 708de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, 709de57606cSSage Weil struct ceph_cap *cap); 710de57606cSSage Weil 711de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci) 712de57606cSSage Weil { 713de57606cSSage Weil int issued; 714be655596SSage Weil spin_lock(&ci->i_ceph_lock); 715de57606cSSage Weil issued = __ceph_caps_issued(ci, NULL); 716be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 717de57606cSSage Weil return issued; 718de57606cSSage Weil } 719de57606cSSage Weil 7201af16d54SXiubo Li static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, 7211af16d54SXiubo Li int mask, int touch) 722de57606cSSage Weil { 723de57606cSSage Weil int r; 724be655596SSage Weil spin_lock(&ci->i_ceph_lock); 7251af16d54SXiubo Li r = __ceph_caps_issued_mask_metric(ci, mask, touch); 726be655596SSage Weil spin_unlock(&ci->i_ceph_lock); 727de57606cSSage Weil return r; 728de57606cSSage Weil } 729de57606cSSage Weil 730de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) 731de57606cSSage Weil { 732de57606cSSage Weil return ci->i_dirty_caps | ci->i_flushing_caps; 733de57606cSSage Weil } 734f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); 735f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); 736f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, 737f66fd9f0SYan, Zheng struct ceph_cap_flush **pcf); 738de57606cSSage Weil 7399563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, 7409563f88cSYan, Zheng struct ceph_cap *ocap, int mask); 741de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); 742de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci); 743de57606cSSage Weil 744719a2514SYan, Zheng static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) 745719a2514SYan, Zheng { 746719a2514SYan, Zheng return ci->i_nr_by_mode[0]; 747719a2514SYan, Zheng } 748de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); 749525d15e8SYan, Zheng extern int __ceph_caps_wanted(struct ceph_inode_info *ci); 750de57606cSSage Weil 751de57606cSSage Weil /* what the mds thinks we want */ 752c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); 753de57606cSSage Weil 75437151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc); 75537151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); 756fe33032dSYan, Zheng extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, 757fe33032dSYan, Zheng struct ceph_mount_options *fsopt); 758e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, 75937151668SYehuda Sadeh struct ceph_cap_reservation *ctx, int need); 7607bf8f736SChengguang Xu extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, 76137151668SYehuda Sadeh struct ceph_cap_reservation *ctx); 7623d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client, 763de57606cSSage Weil int *total, int *avail, int *used, 76485ccce43SSage Weil int *reserved, int *min); 765de57606cSSage Weil 766de57606cSSage Weil 767de57606cSSage Weil 768de57606cSSage Weil /* 769de57606cSSage Weil * we keep buffered readdir results attached to file->private_data 770de57606cSSage Weil */ 7714918b6d1SSage Weil #define CEPH_F_SYNC 1 7729cfa1098SSage Weil #define CEPH_F_ATEND 2 7734918b6d1SSage Weil 774de57606cSSage Weil struct ceph_file_info { 775252c6728SSage Weil short fmode; /* initialized on open */ 776252c6728SSage Weil short flags; /* CEPH_F_* */ 777de57606cSSage Weil 7785d988308SYan, Zheng spinlock_t rw_contexts_lock; 7795d988308SYan, Zheng struct list_head rw_contexts; 780f4b97866SYan, Zheng 78181f148a9SYan, Zheng u32 filp_gen; 782ff5d913dSYan, Zheng atomic_t num_locks; 783bb48bd4dSChengguang Xu }; 784bb48bd4dSChengguang Xu 785bb48bd4dSChengguang Xu struct ceph_dir_file_info { 786bb48bd4dSChengguang Xu struct ceph_file_info file_info; 7875d988308SYan, Zheng 788de57606cSSage Weil /* readdir: position within the dir */ 789de57606cSSage Weil u32 frag; 790de57606cSSage Weil struct ceph_mds_request *last_readdir; 791de57606cSSage Weil 792de57606cSSage Weil /* readdir: position within a frag */ 793f0494206SYan, Zheng unsigned next_offset; /* offset of next chunk (last_name's + 1) */ 794de57606cSSage Weil char *last_name; /* last entry in previous chunk */ 795fdd4e158SYan, Zheng long long dir_release_count; 796fdd4e158SYan, Zheng long long dir_ordered_count; 797fdd4e158SYan, Zheng int readdir_cache_idx; 798de57606cSSage Weil 799de57606cSSage Weil /* used for -o dirstat read() on directory thing */ 800de57606cSSage Weil char *dir_info; 801de57606cSSage Weil int dir_info_len; 802de57606cSSage Weil }; 803de57606cSSage Weil 8045d988308SYan, Zheng struct ceph_rw_context { 8055d988308SYan, Zheng struct list_head list; 8065d988308SYan, Zheng struct task_struct *thread; 8075d988308SYan, Zheng int caps; 8085d988308SYan, Zheng }; 8095d988308SYan, Zheng 8105d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ 8115d988308SYan, Zheng struct ceph_rw_context _name = { \ 8125d988308SYan, Zheng .thread = current, \ 8135d988308SYan, Zheng .caps = _caps, \ 8145d988308SYan, Zheng } 8155d988308SYan, Zheng 8165d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf, 8175d988308SYan, Zheng struct ceph_rw_context *ctx) 8185d988308SYan, Zheng { 8195d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8205d988308SYan, Zheng list_add(&ctx->list, &cf->rw_contexts); 8215d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8225d988308SYan, Zheng } 8235d988308SYan, Zheng 8245d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf, 8255d988308SYan, Zheng struct ceph_rw_context *ctx) 8265d988308SYan, Zheng { 8275d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8285d988308SYan, Zheng list_del(&ctx->list); 8295d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8305d988308SYan, Zheng } 8315d988308SYan, Zheng 8325d988308SYan, Zheng static inline struct ceph_rw_context* 8335d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf) 8345d988308SYan, Zheng { 8355d988308SYan, Zheng struct ceph_rw_context *ctx, *found = NULL; 8365d988308SYan, Zheng spin_lock(&cf->rw_contexts_lock); 8375d988308SYan, Zheng list_for_each_entry(ctx, &cf->rw_contexts, list) { 8385d988308SYan, Zheng if (ctx->thread == current) { 8395d988308SYan, Zheng found = ctx; 8405d988308SYan, Zheng break; 8415d988308SYan, Zheng } 8425d988308SYan, Zheng } 8435d988308SYan, Zheng spin_unlock(&cf->rw_contexts_lock); 8445d988308SYan, Zheng return found; 8455d988308SYan, Zheng } 8465d988308SYan, Zheng 847fdd4e158SYan, Zheng struct ceph_readdir_cache_control { 848fdd4e158SYan, Zheng struct page *page; 849fdd4e158SYan, Zheng struct dentry **dentries; 850fdd4e158SYan, Zheng int index; 851fdd4e158SYan, Zheng }; 852de57606cSSage Weil 853de57606cSSage Weil /* 854de57606cSSage Weil * A "snap realm" describes a subset of the file hierarchy sharing 855de57606cSSage Weil * the same set of snapshots that apply to it. The realms themselves 856de57606cSSage Weil * are organized into a hierarchy, such that children inherit (some of) 857de57606cSSage Weil * the snapshots of their parents. 858de57606cSSage Weil * 859de57606cSSage Weil * All inodes within the realm that have capabilities are linked into a 860de57606cSSage Weil * per-realm list. 861de57606cSSage Weil */ 862de57606cSSage Weil struct ceph_snap_realm { 863de57606cSSage Weil u64 ino; 864e3161f17SLuis Henriques struct inode *inode; 865de57606cSSage Weil atomic_t nref; 866a105f00cSSage Weil struct rb_node node; 867a105f00cSSage Weil 868de57606cSSage Weil u64 created, seq; 869de57606cSSage Weil u64 parent_ino; 870de57606cSSage Weil u64 parent_since; /* snapid when our current parent became so */ 871de57606cSSage Weil 872de57606cSSage Weil u64 *prior_parent_snaps; /* snaps inherited from any parents we */ 873aa711ee3SAlex Elder u32 num_prior_parent_snaps; /* had prior to parent_since */ 874de57606cSSage Weil u64 *snaps; /* snaps specific to this realm */ 875aa711ee3SAlex Elder u32 num_snaps; 876de57606cSSage Weil 877de57606cSSage Weil struct ceph_snap_realm *parent; 878de57606cSSage Weil struct list_head children; /* list of child realms */ 879de57606cSSage Weil struct list_head child_item; 880de57606cSSage Weil 881de57606cSSage Weil struct list_head empty_item; /* if i have ref==0 */ 882de57606cSSage Weil 883ae00d4f3SSage Weil struct list_head dirty_item; /* if realm needs new context */ 884ae00d4f3SSage Weil 88574a31df4SXiubo Li struct list_head rebuild_item; /* rebuild snap realms _downward_ in hierarchy */ 88674a31df4SXiubo Li 887de57606cSSage Weil /* the current set of snaps for this realm */ 888de57606cSSage Weil struct ceph_snap_context *cached_context; 889de57606cSSage Weil 890de57606cSSage Weil struct list_head inodes_with_caps; 891de57606cSSage Weil spinlock_t inodes_with_caps_lock; 892de57606cSSage Weil }; 893de57606cSSage Weil 8943d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void) 8953d14c5d2SYehuda Sadeh { 8963d14c5d2SYehuda Sadeh int congestion_kb; 897de57606cSSage Weil 898de57606cSSage Weil /* 8993d14c5d2SYehuda Sadeh * Copied from NFS 9003d14c5d2SYehuda Sadeh * 9013d14c5d2SYehuda Sadeh * congestion size, scale with available memory. 9023d14c5d2SYehuda Sadeh * 9033d14c5d2SYehuda Sadeh * 64MB: 8192k 9043d14c5d2SYehuda Sadeh * 128MB: 11585k 9053d14c5d2SYehuda Sadeh * 256MB: 16384k 9063d14c5d2SYehuda Sadeh * 512MB: 23170k 9073d14c5d2SYehuda Sadeh * 1GB: 32768k 9083d14c5d2SYehuda Sadeh * 2GB: 46340k 9093d14c5d2SYehuda Sadeh * 4GB: 65536k 9103d14c5d2SYehuda Sadeh * 8GB: 92681k 9113d14c5d2SYehuda Sadeh * 16GB: 131072k 9123d14c5d2SYehuda Sadeh * 9133d14c5d2SYehuda Sadeh * This allows larger machines to have larger/more transfers. 9143d14c5d2SYehuda Sadeh * Limit the default to 256M 915de57606cSSage Weil */ 916ca79b0c2SArun KS congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); 9173d14c5d2SYehuda Sadeh if (congestion_kb > 256*1024) 9183d14c5d2SYehuda Sadeh congestion_kb = 256*1024; 9193d14c5d2SYehuda Sadeh 9203d14c5d2SYehuda Sadeh return congestion_kb; 921de57606cSSage Weil } 922de57606cSSage Weil 923de57606cSSage Weil 924d468e729SYan, Zheng /* super.c */ 925d468e729SYan, Zheng extern int ceph_force_reconnect(struct super_block *sb); 926de57606cSSage Weil /* snap.c */ 927de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, 928de57606cSSage Weil u64 ino); 929de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, 930de57606cSSage Weil struct ceph_snap_realm *realm); 931de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, 932de57606cSSage Weil struct ceph_snap_realm *realm); 933de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m, 934982d6011SYan, Zheng void *p, void *e, bool deletion, 935982d6011SYan, Zheng struct ceph_snap_realm **realm_ret); 9360ba92e1cSJeff Layton void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm); 937de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc, 9382600d2ddSSage Weil struct ceph_mds_session *session, 939de57606cSSage Weil struct ceph_msg *msg); 940de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 941de57606cSSage Weil struct ceph_cap_snap *capsnap); 9425ed91587SXiubo Li extern void ceph_cleanup_global_and_empty_realms(struct ceph_mds_client *mdsc); 943de57606cSSage Weil 94475c9627eSYan, Zheng extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, 94575c9627eSYan, Zheng u64 snap); 94675c9627eSYan, Zheng extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, 94775c9627eSYan, Zheng struct ceph_snapid_map *sm); 94875c9627eSYan, Zheng extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); 94975c9627eSYan, Zheng extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); 950631ed4b0SJeff Layton void ceph_umount_begin(struct super_block *sb); 95175c9627eSYan, Zheng 95275c9627eSYan, Zheng 953de57606cSSage Weil /* 954de57606cSSage Weil * a cap_snap is "pending" if it is still awaiting an in-progress 955de57606cSSage Weil * sync write (that may/may not still update size, mtime, etc.). 956de57606cSSage Weil */ 957de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) 958de57606cSSage Weil { 959de57606cSSage Weil return !list_empty(&ci->i_cap_snaps) && 96086056090SYan, Zheng list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, 961de57606cSSage Weil ci_item)->writing; 962de57606cSSage Weil } 963de57606cSSage Weil 964de57606cSSage Weil /* inode.c */ 965966c7160SJeff Layton struct ceph_mds_reply_info_in; 966966c7160SJeff Layton struct ceph_mds_reply_dirfrag; 967966c7160SJeff Layton 968de57606cSSage Weil extern const struct inode_operations ceph_file_iops; 969de57606cSSage Weil 970de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb); 97187bc5b89SYan, Zheng extern void ceph_evict_inode(struct inode *inode); 972cfa6d412SAl Viro extern void ceph_free_inode(struct inode *inode); 973de57606cSSage Weil 974de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb, 975de57606cSSage Weil struct ceph_vino vino); 976de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent); 977de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued, 978de57606cSSage Weil u32 truncate_seq, u64 truncate_size, u64 size); 979de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued, 9809bbeab41SArnd Bergmann u64 time_warp_seq, struct timespec64 *ctime, 9819bbeab41SArnd Bergmann struct timespec64 *mtime, 9829bbeab41SArnd Bergmann struct timespec64 *atime); 983966c7160SJeff Layton extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, 984966c7160SJeff Layton struct ceph_mds_reply_info_in *iinfo, 985966c7160SJeff Layton struct ceph_mds_reply_dirfrag *dirinfo, 986966c7160SJeff Layton struct ceph_mds_session *session, int cap_fmode, 987966c7160SJeff Layton struct ceph_cap_reservation *caps_reservation); 988de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb, 989f5a03b08SJeff Layton struct ceph_mds_request *req); 990de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, 991de57606cSSage Weil struct ceph_mds_session *session); 992de57606cSSage Weil 993de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask); 994de57606cSSage Weil 995efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size); 996b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode); 99764f28c62SJeff Layton 99864f28c62SJeff Layton void ceph_queue_inode_work(struct inode *inode, int work_bit); 99964f28c62SJeff Layton 100064f28c62SJeff Layton static inline void ceph_queue_vmtruncate(struct inode *inode) 100164f28c62SJeff Layton { 100264f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE); 100364f28c62SJeff Layton } 100464f28c62SJeff Layton 100564f28c62SJeff Layton static inline void ceph_queue_invalidate(struct inode *inode) 100664f28c62SJeff Layton { 100764f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES); 100864f28c62SJeff Layton } 100964f28c62SJeff Layton 101064f28c62SJeff Layton static inline void ceph_queue_writeback(struct inode *inode) 101164f28c62SJeff Layton { 101264f28c62SJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK); 101364f28c62SJeff Layton } 101464f28c62SJeff Layton 1015a8810cdcSJeff Layton static inline void ceph_queue_check_caps(struct inode *inode) 1016a8810cdcSJeff Layton { 1017a8810cdcSJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS); 1018a8810cdcSJeff Layton } 1019a8810cdcSJeff Layton 1020a8810cdcSJeff Layton static inline void ceph_queue_flush_snaps(struct inode *inode) 1021a8810cdcSJeff Layton { 1022a8810cdcSJeff Layton ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS); 1023a8810cdcSJeff Layton } 1024a8810cdcSJeff Layton 1025*5eed80fbSXiubo Li extern int ceph_try_to_choose_auth_mds(struct inode *inode, int mask); 102601deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, 102701deead0SYan, Zheng int mask, bool force); 102801deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) 102901deead0SYan, Zheng { 103001deead0SYan, Zheng return __ceph_do_getattr(inode, NULL, mask, force); 103101deead0SYan, Zheng } 1032549c7297SChristian Brauner extern int ceph_permission(struct user_namespace *mnt_userns, 1033549c7297SChristian Brauner struct inode *inode, int mask); 1034a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr); 1035549c7297SChristian Brauner extern int ceph_setattr(struct user_namespace *mnt_userns, 1036549c7297SChristian Brauner struct dentry *dentry, struct iattr *attr); 1037549c7297SChristian Brauner extern int ceph_getattr(struct user_namespace *mnt_userns, 1038549c7297SChristian Brauner const struct path *path, struct kstat *stat, 1039a528d35eSDavid Howells u32 request_mask, unsigned int flags); 10405d6451b1SJeff Layton void ceph_inode_shutdown(struct inode *inode); 10415d6451b1SJeff Layton 10425d6451b1SJeff Layton static inline bool ceph_inode_is_shutdown(struct inode *inode) 10435d6451b1SJeff Layton { 10445d6451b1SJeff Layton unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags); 10455d6451b1SJeff Layton struct ceph_fs_client *fsc = ceph_inode_to_client(inode); 10465d6451b1SJeff Layton int state = READ_ONCE(fsc->mount_state); 10475d6451b1SJeff Layton 10485d6451b1SJeff Layton return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN; 10495d6451b1SJeff Layton } 1050de57606cSSage Weil 1051de57606cSSage Weil /* xattr.c */ 1052a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); 10536ddf5f16SMilind Changire int ceph_do_getvxattr(struct inode *inode, const char *name, void *value, size_t size); 10547221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); 1055de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); 105612fe3ddaSLuis Henriques extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); 1057de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); 1058b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[]; 1059de57606cSSage Weil 10605c31e92dSYan, Zheng struct ceph_acl_sec_ctx { 10615c31e92dSYan, Zheng #ifdef CONFIG_CEPH_FS_POSIX_ACL 10625c31e92dSYan, Zheng void *default_acl; 10635c31e92dSYan, Zheng void *acl; 10645c31e92dSYan, Zheng #endif 1065ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 1066ac6713ccSYan, Zheng void *sec_ctx; 1067ac6713ccSYan, Zheng u32 sec_ctxlen; 1068ac6713ccSYan, Zheng #endif 10695c31e92dSYan, Zheng struct ceph_pagelist *pagelist; 10705c31e92dSYan, Zheng }; 10715c31e92dSYan, Zheng 1072315f2408SYan, Zheng #ifdef CONFIG_SECURITY 1073315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in); 1074315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in); 1075315f2408SYan, Zheng #else 1076315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in) 1077315f2408SYan, Zheng { 1078315f2408SYan, Zheng return false; 1079315f2408SYan, Zheng } 1080315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in) 1081315f2408SYan, Zheng { 1082315f2408SYan, Zheng return false; 1083315f2408SYan, Zheng } 1084315f2408SYan, Zheng #endif 1085315f2408SYan, Zheng 1086ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL 1087ac6713ccSYan, Zheng extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1088ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx); 1089668959a5SJeff Layton static inline void ceph_security_invalidate_secctx(struct inode *inode) 1090668959a5SJeff Layton { 1091668959a5SJeff Layton security_inode_invalidate_secctx(inode); 1092668959a5SJeff Layton } 1093ac6713ccSYan, Zheng #else 1094ac6713ccSYan, Zheng static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, 1095ac6713ccSYan, Zheng struct ceph_acl_sec_ctx *ctx) 1096ac6713ccSYan, Zheng { 1097ac6713ccSYan, Zheng return 0; 1098ac6713ccSYan, Zheng } 1099ac6713ccSYan, Zheng static inline void ceph_security_invalidate_secctx(struct inode *inode) 1100ac6713ccSYan, Zheng { 1101ac6713ccSYan, Zheng } 1102ac6713ccSYan, Zheng #endif 11037221fe4cSGuangliang Zhao 11045c31e92dSYan, Zheng void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); 11057221fe4cSGuangliang Zhao 11065c31e92dSYan, Zheng /* acl.c */ 11077221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL 11087221fe4cSGuangliang Zhao 11090cad6246SMiklos Szeredi struct posix_acl *ceph_get_acl(struct inode *, int, bool); 1110549c7297SChristian Brauner int ceph_set_acl(struct user_namespace *mnt_userns, 1111549c7297SChristian Brauner struct inode *inode, struct posix_acl *acl, int type); 1112b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 11135c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 11145c31e92dSYan, Zheng void ceph_init_inode_acls(struct inode *inode, 11155c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx); 1116c969d9bfSGuangliang Zhao 1117c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 1118c969d9bfSGuangliang Zhao { 1119c969d9bfSGuangliang Zhao forget_all_cached_acls(inode); 1120c969d9bfSGuangliang Zhao } 11217221fe4cSGuangliang Zhao 11227221fe4cSGuangliang Zhao #else 11237221fe4cSGuangliang Zhao 11247221fe4cSGuangliang Zhao #define ceph_get_acl NULL 112572466d0bSSage Weil #define ceph_set_acl NULL 11267221fe4cSGuangliang Zhao 1127b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, 11285c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 11297221fe4cSGuangliang Zhao { 11307221fe4cSGuangliang Zhao return 0; 11317221fe4cSGuangliang Zhao } 1132b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode, 11335c31e92dSYan, Zheng struct ceph_acl_sec_ctx *as_ctx) 1134b1ee94aaSYan, Zheng { 1135b1ee94aaSYan, Zheng } 11367221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) 11377221fe4cSGuangliang Zhao { 11387221fe4cSGuangliang Zhao return 0; 11397221fe4cSGuangliang Zhao } 11407221fe4cSGuangliang Zhao 11417221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode) 11427221fe4cSGuangliang Zhao { 11437221fe4cSGuangliang Zhao } 11447221fe4cSGuangliang Zhao 11457221fe4cSGuangliang Zhao #endif 11467221fe4cSGuangliang Zhao 1147de57606cSSage Weil /* caps.c */ 1148de57606cSSage Weil extern const char *ceph_cap_string(int c); 1149de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session, 1150de57606cSSage Weil struct ceph_msg *msg); 1151d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, 1152d9df2783SYan, Zheng struct ceph_cap_reservation *ctx); 1153d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode, 1154de57606cSSage Weil struct ceph_mds_session *session, u64 cap_id, 1155135e671eSYan, Zheng unsigned issued, unsigned wanted, 1156de57606cSSage Weil unsigned cap, unsigned seq, u64 realmino, int flags, 1157d9df2783SYan, Zheng struct ceph_cap **new_cap); 1158a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1159a76d0a9cSXiubo Li extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release); 1160d6e47819SYan, Zheng extern void __ceph_remove_caps(struct ceph_inode_info *ci); 116137151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc, 116237151668SYehuda Sadeh struct ceph_cap *cap); 11639215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode); 1164de57606cSSage Weil 1165f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); 116602c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end, 116702c24a82SJosef Bacik int datasync); 1168e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, 1169e548e9b9SYan, Zheng struct ceph_mds_session *session); 1170de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, 1171de57606cSSage Weil struct ceph_mds_session *session); 1172e8a4d267SJeff Layton void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, 1173e8a4d267SJeff Layton struct ceph_inode_info *ci); 11742bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, 11752bc50259SGreg Farnum int mds); 117640dcf75eSJeff Layton extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, 117740dcf75eSJeff Layton bool snap_rwsem_locked); 1178de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); 1179de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); 1180a8810cdcSJeff Layton extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had); 1181e64f44a8SXiubo Li extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, 1182e64f44a8SXiubo Li int had); 1183de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, 1184de57606cSSage Weil struct ceph_snap_context *snapc); 1185a6d37ccdSXiubo Li extern void __ceph_remove_capsnap(struct inode *inode, 1186a6d37ccdSXiubo Li struct ceph_cap_snap *capsnap, 1187a6d37ccdSXiubo Li bool *wake_ci, bool *wake_mdsc); 1188a6d37ccdSXiubo Li extern void ceph_remove_capsnap(struct inode *inode, 1189a6d37ccdSXiubo Li struct ceph_cap_snap *capsnap, 1190a6d37ccdSXiubo Li bool *wake_ci, bool *wake_mdsc); 1191ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci, 11920e294387SYan, Zheng struct ceph_mds_session **psession); 1193efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci); 1194de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, 1195de57606cSSage Weil struct ceph_mds_session *session); 1196bf2ba432SLuis Henriques extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc); 1197afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); 11986ef0bc6dSZhi Zhang extern int ceph_drop_caps_for_unlink(struct inode *inode); 1199de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode, 1200de57606cSSage Weil int mds, int drop, int unless, int force); 1201de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn, 1202ca6c8ae0SJeff Layton struct inode *dir, 1203de57606cSSage Weil int mds, int drop, int unless); 1204de57606cSSage Weil 12055e3ded1bSYan, Zheng extern int ceph_get_caps(struct file *filp, int need, int want, 1206e72968e1SJeff Layton loff_t endoff, int *got); 12075e3ded1bSYan, Zheng extern int ceph_try_get_caps(struct inode *inode, 12082ee9dd95SLuis Henriques int need, int want, bool nonblock, int *got); 1209de57606cSSage Weil 1210de57606cSSage Weil /* for counting open files by mode */ 1211719a2514SYan, Zheng extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); 1212719a2514SYan, Zheng extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); 1213719a2514SYan, Zheng extern void __ceph_touch_fmode(struct ceph_inode_info *ci, 1214719a2514SYan, Zheng struct ceph_mds_client *mdsc, int fmode); 1215de57606cSSage Weil 1216de57606cSSage Weil /* addr.c */ 1217de57606cSSage Weil extern const struct address_space_operations ceph_aops; 1218bc899ee1SDavid Howells extern const struct netfs_request_ops ceph_netfs_ops; 1219de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); 1220083db6fdSDavid Howells extern int ceph_uninline_data(struct file *file); 12215e3ded1bSYan, Zheng extern int ceph_pool_perm_check(struct inode *inode, int need); 122210183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); 122336e6da98SJeff Layton int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate); 1224de57606cSSage Weil 1225de57606cSSage Weil /* file.c */ 1226de57606cSSage Weil extern const struct file_operations ceph_file_fops; 12279e0eb85dSAlex Elder 1228719a2514SYan, Zheng extern int ceph_renew_caps(struct inode *inode, int fmode); 1229de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file); 12305ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 123144907d79SAl Viro struct file *file, unsigned flags, umode_t mode); 1232de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp); 123331c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, 123431c542a1SYan, Zheng char *data, size_t len); 123555f2a045SIlya Dryomov 1236de57606cSSage Weil /* dir.c */ 1237de57606cSSage Weil extern const struct file_operations ceph_dir_fops; 123838c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops; 1239de57606cSSage Weil extern const struct inode_operations ceph_dir_iops; 124038c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops; 124118fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops; 1242de57606cSSage Weil 1243f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); 1244de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); 1245aa60cfc3SJeff Layton extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req, 12467a971e2cSJeff Layton struct dentry *dentry); 1247de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, 1248de57606cSSage Weil struct dentry *dentry, int err); 1249de57606cSSage Weil 125037c4efc1SYan, Zheng extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); 125137c4efc1SYan, Zheng extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); 125281a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry); 125337c4efc1SYan, Zheng extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); 1254e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); 1255fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); 1256de57606cSSage Weil 1257de57606cSSage Weil /* ioctl.c */ 1258de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 1259de57606cSSage Weil 1260de57606cSSage Weil /* export.c */ 1261de57606cSSage Weil extern const struct export_operations ceph_export_ops; 12623886274aSLuis Henriques struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); 1263de57606cSSage Weil 126440819f6fSGreg Farnum /* locks.c */ 1265eb13e832SYan, Zheng extern __init void ceph_flock_init(void); 126640819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); 126740819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); 126840819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); 126939be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode, 127039be95e9SJim Schutt struct ceph_filelock *flocks, 127139be95e9SJim Schutt int num_fcntl_locks, 127239be95e9SJim Schutt int num_flock_locks); 127339be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, 127439be95e9SJim Schutt struct ceph_pagelist *pagelist, 127539be95e9SJim Schutt int num_fcntl_locks, int num_flock_locks); 127640819f6fSGreg Farnum 12773d14c5d2SYehuda Sadeh /* debugfs.c */ 12781a829ff2SGreg Kroah-Hartman extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); 12793d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); 12803d14c5d2SYehuda Sadeh 1281fb18a575SLuis Henriques /* quota.c */ 1282d557c48dSLuis Henriques static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) 1283d557c48dSLuis Henriques { 1284d557c48dSLuis Henriques return ci->i_max_files || ci->i_max_bytes; 1285d557c48dSLuis Henriques } 1286d557c48dSLuis Henriques 1287d557c48dSLuis Henriques extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); 1288d557c48dSLuis Henriques 1289d557c48dSLuis Henriques static inline void __ceph_update_quota(struct ceph_inode_info *ci, 1290d557c48dSLuis Henriques u64 max_bytes, u64 max_files) 1291d557c48dSLuis Henriques { 1292d557c48dSLuis Henriques bool had_quota, has_quota; 1293d557c48dSLuis Henriques had_quota = __ceph_has_any_quota(ci); 1294d557c48dSLuis Henriques ci->i_max_bytes = max_bytes; 1295d557c48dSLuis Henriques ci->i_max_files = max_files; 1296d557c48dSLuis Henriques has_quota = __ceph_has_any_quota(ci); 1297d557c48dSLuis Henriques 1298d557c48dSLuis Henriques if (had_quota != has_quota) 1299d557c48dSLuis Henriques ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); 1300d557c48dSLuis Henriques } 1301d557c48dSLuis Henriques 1302fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc, 1303fb18a575SLuis Henriques struct ceph_mds_session *session, 1304fb18a575SLuis Henriques struct ceph_msg *msg); 1305b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); 13066646ea1cSLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); 13072b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, 13082b83845fSLuis Henriques loff_t newlen); 13091ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, 13101ab302a0SLuis Henriques loff_t newlen); 13119122eed5SLuis Henriques extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, 13129122eed5SLuis Henriques struct kstatfs *buf); 13130c44a8e0SLuis Henriques extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); 1314fb18a575SLuis Henriques 1315de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */ 1316