xref: /linux/fs/ceph/super.h (revision cbed4ff76bbb8083780dcd6ebb5bed2af25b9505)
1b2441318SGreg Kroah-Hartman /* SPDX-License-Identifier: GPL-2.0 */
2de57606cSSage Weil #ifndef _FS_CEPH_SUPER_H
3de57606cSSage Weil #define _FS_CEPH_SUPER_H
4de57606cSSage Weil 
53d14c5d2SYehuda Sadeh #include <linux/ceph/ceph_debug.h>
6de57606cSSage Weil 
7de57606cSSage Weil #include <asm/unaligned.h>
8de57606cSSage Weil #include <linux/backing-dev.h>
9de57606cSSage Weil #include <linux/completion.h>
10de57606cSSage Weil #include <linux/exportfs.h>
11de57606cSSage Weil #include <linux/fs.h>
12de57606cSSage Weil #include <linux/mempool.h>
13de57606cSSage Weil #include <linux/pagemap.h>
14de57606cSSage Weil #include <linux/wait.h>
15f1a3d572SStephen Rothwell #include <linux/writeback.h>
165a0e3ad6STejun Heo #include <linux/slab.h>
17c969d9bfSGuangliang Zhao #include <linux/posix_acl.h>
18805692d0SElena Reshetova #include <linux/refcount.h>
19668959a5SJeff Layton #include <linux/security.h>
20de57606cSSage Weil 
213d14c5d2SYehuda Sadeh #include <linux/ceph/libceph.h>
22de57606cSSage Weil 
2399ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE
247c46b318SJeff Layton #define FSCACHE_USE_NEW_IO_API
2599ccbd22SMilosz Tanski #include <linux/fscache.h>
2699ccbd22SMilosz Tanski #endif
2799ccbd22SMilosz Tanski 
28de57606cSSage Weil /* f_type in struct statfs */
29de57606cSSage Weil #define CEPH_SUPER_MAGIC 0x00c36400
30de57606cSSage Weil 
31de57606cSSage Weil /* large granularity for statfs utilization stats to facilitate
32de57606cSSage Weil  * large volume sizes on 32-bit machines. */
3392a49fb0SSage Weil #define CEPH_BLOCK_SHIFT   22  /* 4 MB */
34de57606cSSage Weil #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
35de57606cSSage Weil 
360b98acd6SIlya Dryomov #define CEPH_MOUNT_OPT_CLEANRECOVER    (1<<1) /* auto reonnect (clean mode) after blocklisted */
373d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_DIRSTAT         (1<<4) /* `cat dirname` for stats */
383d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
393d14c5d2SYehuda Sadeh #define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
40ad1fee96SYehuda Sadeh #define CEPH_MOUNT_OPT_INO32           (1<<8) /* 32 bit inos */
41a40dc6ccSSage Weil #define CEPH_MOUNT_OPT_DCACHE          (1<<9) /* use dcache for readdir etc */
4299ccbd22SMilosz Tanski #define CEPH_MOUNT_OPT_FSCACHE         (1<<10) /* use fscache */
4310183a69SYan, Zheng #define CEPH_MOUNT_OPT_NOPOOLPERM      (1<<11) /* no pool permission check */
44e9e427f0SYan, Zheng #define CEPH_MOUNT_OPT_MOUNTWAIT       (1<<12) /* mount waits if no mds is up */
459122eed5SLuis Henriques #define CEPH_MOUNT_OPT_NOQUOTADF       (1<<13) /* no root dir quota in statfs */
46ea4cdc54SLuis Henriques #define CEPH_MOUNT_OPT_NOCOPYFROM      (1<<14) /* don't use RADOS 'copy-from' op */
472ccb4546SJeff Layton #define CEPH_MOUNT_OPT_ASYNC_DIROPS    (1<<15) /* allow async directory ops */
486a259382SSage Weil 
496f9718feSLuis Henriques #define CEPH_MOUNT_OPT_DEFAULT			\
506f9718feSLuis Henriques 	(CEPH_MOUNT_OPT_DCACHE |		\
51f7a67b46SJeff Layton 	 CEPH_MOUNT_OPT_NOCOPYFROM |		\
52f7a67b46SJeff Layton 	 CEPH_MOUNT_OPT_ASYNC_DIROPS)
53de57606cSSage Weil 
543d14c5d2SYehuda Sadeh #define ceph_set_mount_opt(fsc, opt) \
552ccb4546SJeff Layton 	(fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt
562ccb4546SJeff Layton #define ceph_clear_mount_opt(fsc, opt) \
572ccb4546SJeff Layton 	(fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt
583d14c5d2SYehuda Sadeh #define ceph_test_mount_opt(fsc, opt) \
593d14c5d2SYehuda Sadeh 	(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
60de57606cSSage Weil 
61aa187926SYan, Zheng /* max size of osd read request, limited by libceph */
62aa187926SYan, Zheng #define CEPH_MAX_READ_SIZE              CEPH_MSG_MAX_DATA_LEN
6395cca2b4SYan, Zheng /* osd has a configurable limitaion of max write size.
6495cca2b4SYan, Zheng  * CEPH_MSG_MAX_DATA_LEN should be small enough. */
6595cca2b4SYan, Zheng #define CEPH_MAX_WRITE_SIZE		CEPH_MSG_MAX_DATA_LEN
667c94ba27SAndreas Gerstmayr #define CEPH_RASIZE_DEFAULT             (8192*1024)    /* max readahead */
673d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_DEFAULT        1024
683d14c5d2SYehuda Sadeh #define CEPH_MAX_READDIR_BYTES_DEFAULT  (512*1024)
693d14c5d2SYehuda Sadeh #define CEPH_SNAPDIRNAME_DEFAULT        ".snap"
70de57606cSSage Weil 
714214fb15SYan, Zheng /*
724214fb15SYan, Zheng  * Delay telling the MDS we no longer want caps, in case we reopen
734214fb15SYan, Zheng  * the file.  Delay a minimum amount of time, even if we send a cap
744214fb15SYan, Zheng  * message for some other reason.  Otherwise, take the oppotunity to
754214fb15SYan, Zheng  * update the mds to avoid sending another message later.
764214fb15SYan, Zheng  */
774214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT      5  /* cap release delay */
784214fb15SYan, Zheng #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT     60  /* cap release delay */
794214fb15SYan, Zheng 
803d14c5d2SYehuda Sadeh struct ceph_mount_options {
81ad8c28a9SJeff Layton 	unsigned int flags;
823d14c5d2SYehuda Sadeh 
83ad8c28a9SJeff Layton 	unsigned int wsize;            /* max write size */
84ad8c28a9SJeff Layton 	unsigned int rsize;            /* max read size */
85ad8c28a9SJeff Layton 	unsigned int rasize;           /* max readahead */
86ad8c28a9SJeff Layton 	unsigned int congestion_kb;    /* max writeback in flight */
87ad8c28a9SJeff Layton 	unsigned int caps_wanted_delay_min, caps_wanted_delay_max;
88fe33032dSYan, Zheng 	int caps_max;
89ad8c28a9SJeff Layton 	unsigned int max_readdir;       /* max readdir result (entries) */
90ad8c28a9SJeff Layton 	unsigned int max_readdir_bytes; /* max readdir result (bytes) */
913d14c5d2SYehuda Sadeh 
923d14c5d2SYehuda Sadeh 	/*
933d14c5d2SYehuda Sadeh 	 * everything above this point can be memcmp'd; everything below
943d14c5d2SYehuda Sadeh 	 * is handled in compare_mount_options()
953d14c5d2SYehuda Sadeh 	 */
963d14c5d2SYehuda Sadeh 
97de57606cSSage Weil 	char *snapdir_name;   /* default ".snap" */
98430afbadSYan, Zheng 	char *mds_namespace;  /* default NULL */
99b27a939eSIlya Dryomov 	char *server_path;    /* default NULL (means "/") */
1001d8f8360SYan, Zheng 	char *fscache_uniq;   /* default NULL */
101de57606cSSage Weil };
102de57606cSSage Weil 
1033d14c5d2SYehuda Sadeh struct ceph_fs_client {
104de57606cSSage Weil 	struct super_block *sb;
105de57606cSSage Weil 
10618f473b3SXiubo Li 	struct list_head metric_wakeup;
10718f473b3SXiubo Li 
1083d14c5d2SYehuda Sadeh 	struct ceph_mount_options *mount_options;
1093d14c5d2SYehuda Sadeh 	struct ceph_client *client;
1103d14c5d2SYehuda Sadeh 
111aa5c7910SJeff Layton 	int mount_state;
11281f148a9SYan, Zheng 
1130b98acd6SIlya Dryomov 	bool blocklisted;
114131d7eb4SYan, Zheng 
11578beb0ffSLuis Henriques 	bool have_copy_from2;
11678beb0ffSLuis Henriques 
11781f148a9SYan, Zheng 	u32 filp_gen;
118719784baSChengguang Xu 	loff_t max_file_size;
11985ccce43SSage Weil 
1203d14c5d2SYehuda Sadeh 	struct ceph_mds_client *mdsc;
121de57606cSSage Weil 
122e3ec8d68SYan, Zheng 	atomic_long_t writeback_count;
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;
134*cbed4ff7SLuís Henriques 	struct dentry *debugfs_metrics_dir;
1350743304dSSage Weil #endif
13699ccbd22SMilosz Tanski 
13799ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE
13899ccbd22SMilosz Tanski 	struct fscache_cookie *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);
232de57606cSSage Weil 		kfree(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 {
319de57606cSSage Weil 	struct ceph_vino i_vino;   /* ceph ino + snap */
320de57606cSSage Weil 
321be655596SSage Weil 	spinlock_t i_ceph_lock;
322be655596SSage Weil 
323de57606cSSage Weil 	u64 i_version;
32431c542a1SYan, Zheng 	u64 i_inline_version;
325de57606cSSage Weil 	u32 i_time_warp_seq;
326de57606cSSage Weil 
327891f3f5aSJeff Layton 	unsigned long i_ceph_flags;
328fdd4e158SYan, Zheng 	atomic64_t i_release_count;
329fdd4e158SYan, Zheng 	atomic64_t i_ordered_count;
330fdd4e158SYan, Zheng 	atomic64_t i_complete_seq[2];
331de57606cSSage Weil 
3326c0f3af7SSage Weil 	struct ceph_dir_layout i_dir_layout;
333de57606cSSage Weil 	struct ceph_file_layout i_layout;
334785892feSJeff Layton 	struct ceph_file_layout i_cached_layout;	// for async creates
335de57606cSSage Weil 	char *i_symlink;
336de57606cSSage Weil 
337de57606cSSage Weil 	/* for dirs */
3389bbeab41SArnd Bergmann 	struct timespec64 i_rctime;
339e7f72952SYanhu Cao 	u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps;
340de57606cSSage Weil 	u64 i_files, i_subdirs;
341de57606cSSage Weil 
342fb18a575SLuis Henriques 	/* quotas */
343fb18a575SLuis Henriques 	u64 i_max_bytes, i_max_files;
344fb18a575SLuis Henriques 
34508796873SYan, Zheng 	s32 i_dir_pin;
34608796873SYan, Zheng 
347de57606cSSage Weil 	struct rb_root i_fragtree;
3481b1bc16dSYan, Zheng 	int i_fragtree_nsplits;
349de57606cSSage Weil 	struct mutex i_fragtree_mutex;
350de57606cSSage Weil 
351de57606cSSage Weil 	struct ceph_inode_xattrs_info i_xattrs;
352de57606cSSage Weil 
353be655596SSage Weil 	/* capabilities.  protected _both_ by i_ceph_lock and cap->session's
354de57606cSSage Weil 	 * s_mutex. */
355de57606cSSage Weil 	struct rb_root i_caps;           /* cap list */
356de57606cSSage Weil 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
357de57606cSSage Weil 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
3581cf03a68SJeff Layton 
3591cf03a68SJeff Layton 	/*
360f1f565a2SRandy Dunlap 	 * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty
3611cf03a68SJeff Layton 	 * is protected by the mdsc->cap_dirty_lock, but each individual item
3621cf03a68SJeff Layton 	 * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty
3631cf03a68SJeff Layton 	 * requires the mdsc->cap_dirty_lock. List presence for an item can
3641cf03a68SJeff Layton 	 * be tested under the i_ceph_lock. Changing anything requires both.
3654fb5dda3SJeff Layton 	 */
3661cf03a68SJeff Layton 	struct list_head i_dirty_item;
3671cf03a68SJeff Layton 
3681cf03a68SJeff Layton 	/*
369829ad4dbSJeff Layton 	 * Link to session's s_cap_flushing list. Protected in a similar
370829ad4dbSJeff Layton 	 * fashion to i_dirty_item, but also by the s_mutex for changes. The
371829ad4dbSJeff Layton 	 * s_cap_flushing list can be walked while holding either the s_mutex
372829ad4dbSJeff Layton 	 * or msdc->cap_dirty_lock. List presence can also be checked while
373829ad4dbSJeff Layton 	 * holding the i_ceph_lock for this inode.
3741cf03a68SJeff Layton 	 */
3751cf03a68SJeff Layton 	struct list_head i_flushing_item;
3761cf03a68SJeff Layton 
377de57606cSSage Weil 	/* we need to track cap writeback on a per-cap-bit basis, to allow
378de57606cSSage Weil 	 * overlapping, pipelined cap flushes to the mds.  we can probably
379de57606cSSage Weil 	 * reduce the tid to 8 bits if we're concerned about inode size. */
380f66fd9f0SYan, Zheng 	struct ceph_cap_flush *i_prealloc_cap_flush;
381e4500b5eSYan, Zheng 	struct list_head i_cap_flush_list;
382de57606cSSage Weil 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
383de57606cSSage Weil 	unsigned long i_hold_caps_max; /* jiffies */
384de57606cSSage Weil 	struct list_head i_cap_delay_list;  /* for delayed cap release to mds */
385de57606cSSage Weil 	struct ceph_cap_reservation i_cap_migration_resv;
386de57606cSSage Weil 	struct list_head i_cap_snaps;   /* snapped state pending flush to mds */
3877d8cb26dSSage Weil 	struct ceph_snap_context *i_head_snapc;  /* set if wr_buffer_head > 0 or
3887d8cb26dSSage Weil 						    dirty|flushing caps */
389de57606cSSage Weil 	unsigned i_snap_caps;           /* cap bits for snapped files */
390de57606cSSage Weil 
391719a2514SYan, Zheng 	unsigned long i_last_rd;
392719a2514SYan, Zheng 	unsigned long i_last_wr;
393774a6a11SYan, Zheng 	int i_nr_by_mode[CEPH_FILE_MODE_BITS];  /* open file counts */
394de57606cSSage Weil 
395b0d7c223SYan, Zheng 	struct mutex i_truncate_mutex;
396de57606cSSage Weil 	u32 i_truncate_seq;        /* last truncate to smaller size */
397de57606cSSage Weil 	u64 i_truncate_size;       /*  and the size we last truncated down to */
398de57606cSSage Weil 	int i_truncate_pending;    /*  still need to call vmtruncate */
399de57606cSSage Weil 
400de57606cSSage Weil 	u64 i_max_size;            /* max file size authorized by mds */
401de57606cSSage Weil 	u64 i_reported_size; /* (max_)size reported to or requested of mds */
402de57606cSSage Weil 	u64 i_wanted_max_size;     /* offset we'd like to write too */
403de57606cSSage Weil 	u64 i_requested_max_size;  /* max_size we've requested */
404de57606cSSage Weil 
405de57606cSSage Weil 	/* held references to caps */
406de57606cSSage Weil 	int i_pin_ref;
407f85122afSJeff Layton 	int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref;
408de57606cSSage Weil 	int i_wrbuffer_ref, i_wrbuffer_ref_head;
40989aa5930SYan, Zheng 	atomic_t i_filelock_ref;
41097aeb6bfSYan, Zheng 	atomic_t i_shared_gen;       /* increment each time we get FILE_SHARED */
411cd045cb4SSage Weil 	u32 i_rdcache_gen;      /* incremented each time we get FILE_CACHE. */
412de57606cSSage Weil 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
413de57606cSSage Weil 
414de57606cSSage Weil 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
41568cd5b4bSYan, Zheng 	struct list_head i_unsafe_iops;   /* uncommitted mds inode ops */
416de57606cSSage Weil 	spinlock_t i_unsafe_lock;
417de57606cSSage Weil 
41875c9627eSYan, Zheng 	union {
419de57606cSSage Weil 		struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */
42075c9627eSYan, Zheng 		struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */
42175c9627eSYan, Zheng 	};
422de57606cSSage Weil 	struct list_head i_snap_realm_item;
423de57606cSSage Weil 	struct list_head i_snap_flush_item;
424245ce991SJeff Layton 	struct timespec64 i_btime;
425193e7b37SDavid Disseldorp 	struct timespec64 i_snap_btime;
426de57606cSSage Weil 
4271cf89a8dSYan, Zheng 	struct work_struct i_work;
4281cf89a8dSYan, Zheng 	unsigned long  i_work_mask;
429de57606cSSage Weil 
43099ccbd22SMilosz Tanski #ifdef CONFIG_CEPH_FSCACHE
43199ccbd22SMilosz Tanski 	struct fscache_cookie *fscache;
43299ccbd22SMilosz Tanski #endif
433de57606cSSage Weil 	struct inode vfs_inode; /* at end */
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)
541d4f6b31dSJeff Layton #define CEPH_INO_SYSTEM_BASE		((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY))
542d4f6b31dSJeff Layton 
543d4f6b31dSJeff Layton static inline bool ceph_vino_is_reserved(const struct ceph_vino vino)
544d4f6b31dSJeff Layton {
545d4f6b31dSJeff Layton 	if (vino.ino < CEPH_INO_SYSTEM_BASE &&
546d4f6b31dSJeff Layton 	    vino.ino >= CEPH_MDS_INO_MDSDIR_OFFSET) {
547d4f6b31dSJeff Layton 		WARN_RATELIMIT(1, "Attempt to access reserved inode number 0x%llx", vino.ino);
548d4f6b31dSJeff Layton 		return true;
549d4f6b31dSJeff Layton 	}
550d4f6b31dSJeff Layton 	return false;
551d4f6b31dSJeff Layton }
552ebce3eb2SJeff Layton 
5533d14c5d2SYehuda Sadeh static inline struct inode *ceph_find_inode(struct super_block *sb,
5543d14c5d2SYehuda Sadeh 					    struct ceph_vino vino)
5553d14c5d2SYehuda Sadeh {
556d4f6b31dSJeff Layton 	if (ceph_vino_is_reserved(vino))
557d4f6b31dSJeff Layton 		return NULL;
558d4f6b31dSJeff Layton 
559ebce3eb2SJeff Layton 	/*
560ebce3eb2SJeff Layton 	 * NB: The hashval will be run through the fs/inode.c hash function
561ebce3eb2SJeff Layton 	 * anyway, so there is no need to squash the inode number down to
562ebce3eb2SJeff Layton 	 * 32-bits first. Just use low-order bits on arches with 32-bit long.
563ebce3eb2SJeff Layton 	 */
564ebce3eb2SJeff Layton 	return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino);
5653d14c5d2SYehuda Sadeh }
5663d14c5d2SYehuda Sadeh 
5673d14c5d2SYehuda Sadeh 
5683d14c5d2SYehuda Sadeh /*
5693d14c5d2SYehuda Sadeh  * Ceph inode.
5703d14c5d2SYehuda Sadeh  */
57110183a69SYan, Zheng #define CEPH_I_DIR_ORDERED	(1 << 0)  /* dentries in dir are ordered */
57210183a69SYan, Zheng #define CEPH_I_FLUSH		(1 << 2)  /* do not delay flush of dirty metadata */
573daca8bdaSJeff Layton #define CEPH_I_POOL_PERM	(1 << 3)  /* pool rd/wr bits are valid */
574daca8bdaSJeff Layton #define CEPH_I_POOL_RD		(1 << 4)  /* can read from pool */
575daca8bdaSJeff Layton #define CEPH_I_POOL_WR		(1 << 5)  /* can write to pool */
576daca8bdaSJeff Layton #define CEPH_I_SEC_INITED	(1 << 6)  /* security initialized */
577c0e385b1SYan, Zheng #define CEPH_I_KICK_FLUSH	(1 << 7)  /* kick flushing caps */
578c0e385b1SYan, Zheng #define CEPH_I_FLUSH_SNAPS	(1 << 8)  /* need flush snapss */
579c0e385b1SYan, Zheng #define CEPH_I_ERROR_WRITE	(1 << 9) /* have seen write errors */
580c0e385b1SYan, Zheng #define CEPH_I_ERROR_FILELOCK	(1 << 10) /* have seen file lock errors */
581c0e385b1SYan, Zheng #define CEPH_I_ODIRECT		(1 << 11) /* inode in direct I/O mode */
582c0e385b1SYan, Zheng #define CEPH_ASYNC_CREATE_BIT	(12)	  /* async create in flight for this */
583891f3f5aSJeff Layton #define CEPH_I_ASYNC_CREATE	(1 << CEPH_ASYNC_CREATE_BIT)
5845d6451b1SJeff Layton #define CEPH_I_SHUTDOWN		(1 << 13) /* inode is no longer usable */
58526544c62SJeff Layton 
58626544c62SJeff Layton /*
5871cf89a8dSYan, Zheng  * Masks of ceph inode work.
5881cf89a8dSYan, Zheng  */
589a8810cdcSJeff Layton #define CEPH_I_WORK_WRITEBACK		0
590a8810cdcSJeff Layton #define CEPH_I_WORK_INVALIDATE_PAGES	1
591a8810cdcSJeff Layton #define CEPH_I_WORK_VMTRUNCATE		2
592a8810cdcSJeff Layton #define CEPH_I_WORK_CHECK_CAPS		3
593a8810cdcSJeff Layton #define CEPH_I_WORK_FLUSH_SNAPS		4
5941cf89a8dSYan, Zheng 
5951cf89a8dSYan, Zheng /*
59626544c62SJeff Layton  * We set the ERROR_WRITE bit when we start seeing write errors on an inode
59726544c62SJeff Layton  * and then clear it when they start succeeding. Note that we do a lockless
59826544c62SJeff Layton  * check first, and only take the lock if it looks like it needs to be changed.
59926544c62SJeff Layton  * The write submission code just takes this as a hint, so we're not too
60026544c62SJeff Layton  * worried if a few slip through in either direction.
60126544c62SJeff Layton  */
60226544c62SJeff Layton static inline void ceph_set_error_write(struct ceph_inode_info *ci)
60326544c62SJeff Layton {
60426544c62SJeff Layton 	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) {
60526544c62SJeff Layton 		spin_lock(&ci->i_ceph_lock);
60626544c62SJeff Layton 		ci->i_ceph_flags |= CEPH_I_ERROR_WRITE;
60726544c62SJeff Layton 		spin_unlock(&ci->i_ceph_lock);
60826544c62SJeff Layton 	}
60926544c62SJeff Layton }
61026544c62SJeff Layton 
61126544c62SJeff Layton static inline void ceph_clear_error_write(struct ceph_inode_info *ci)
61226544c62SJeff Layton {
61326544c62SJeff Layton 	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) {
61426544c62SJeff Layton 		spin_lock(&ci->i_ceph_lock);
61526544c62SJeff Layton 		ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE;
61626544c62SJeff Layton 		spin_unlock(&ci->i_ceph_lock);
61726544c62SJeff Layton 	}
61826544c62SJeff Layton }
6193d14c5d2SYehuda Sadeh 
6202f276c51SYan, Zheng static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
621fdd4e158SYan, Zheng 					   long long release_count,
622fdd4e158SYan, Zheng 					   long long ordered_count)
623de57606cSSage Weil {
62474960773SAndrea Parri 	/*
62574960773SAndrea Parri 	 * Makes sure operations that setup readdir cache (update page
62674960773SAndrea Parri 	 * cache and i_size) are strongly ordered w.r.t. the following
62774960773SAndrea Parri 	 * atomic64_set() operations.
62874960773SAndrea Parri 	 */
62974960773SAndrea Parri 	smp_mb();
630fdd4e158SYan, Zheng 	atomic64_set(&ci->i_complete_seq[0], release_count);
631fdd4e158SYan, Zheng 	atomic64_set(&ci->i_complete_seq[1], ordered_count);
632de57606cSSage Weil }
633de57606cSSage Weil 
6342f276c51SYan, Zheng static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
635de57606cSSage Weil {
636fdd4e158SYan, Zheng 	atomic64_inc(&ci->i_release_count);
637fdd4e158SYan, Zheng }
638fdd4e158SYan, Zheng 
639fdd4e158SYan, Zheng static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
640fdd4e158SYan, Zheng {
641fdd4e158SYan, Zheng 	atomic64_inc(&ci->i_ordered_count);
642de57606cSSage Weil }
643de57606cSSage Weil 
6442f276c51SYan, Zheng static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
645de57606cSSage Weil {
646fdd4e158SYan, Zheng 	return atomic64_read(&ci->i_complete_seq[0]) ==
647fdd4e158SYan, Zheng 		atomic64_read(&ci->i_release_count);
6482f276c51SYan, Zheng }
649de57606cSSage Weil 
65070db4f36SYan, Zheng static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
65170db4f36SYan, Zheng {
652fdd4e158SYan, Zheng 	return  atomic64_read(&ci->i_complete_seq[0]) ==
653fdd4e158SYan, Zheng 		atomic64_read(&ci->i_release_count) &&
654fdd4e158SYan, Zheng 		atomic64_read(&ci->i_complete_seq[1]) ==
655fdd4e158SYan, Zheng 		atomic64_read(&ci->i_ordered_count);
65670db4f36SYan, Zheng }
65770db4f36SYan, Zheng 
6582f276c51SYan, Zheng static inline void ceph_dir_clear_complete(struct inode *inode)
6592f276c51SYan, Zheng {
6602f276c51SYan, Zheng 	__ceph_dir_clear_complete(ceph_inode(inode));
6612f276c51SYan, Zheng }
6622f276c51SYan, Zheng 
66370db4f36SYan, Zheng static inline void ceph_dir_clear_ordered(struct inode *inode)
6642f276c51SYan, Zheng {
665fdd4e158SYan, Zheng 	__ceph_dir_clear_ordered(ceph_inode(inode));
666de57606cSSage Weil }
667de57606cSSage Weil 
66870db4f36SYan, Zheng static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
66970db4f36SYan, Zheng {
670fdd4e158SYan, Zheng 	bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
671fdd4e158SYan, Zheng 	smp_rmb();
67270db4f36SYan, Zheng 	return ret;
67370db4f36SYan, Zheng }
674de57606cSSage Weil 
675de57606cSSage Weil /* find a specific frag @f */
676de57606cSSage Weil extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci,
677de57606cSSage Weil 						u32 f);
678de57606cSSage Weil 
679de57606cSSage Weil /*
680de57606cSSage Weil  * choose fragment for value @v.  copy frag content to pfrag, if leaf
681de57606cSSage Weil  * exists
682de57606cSSage Weil  */
683de57606cSSage Weil extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
684de57606cSSage Weil 			    struct ceph_inode_frag *pfrag,
685de57606cSSage Weil 			    int *found);
686de57606cSSage Weil 
6871e9c2eb6SYan, Zheng static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry)
688de57606cSSage Weil {
689de57606cSSage Weil 	return (struct ceph_dentry_info *)dentry->d_fsdata;
690de57606cSSage Weil }
691de57606cSSage Weil 
692de57606cSSage Weil /*
693de57606cSSage Weil  * caps helpers
694de57606cSSage Weil  */
695de57606cSSage Weil static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci)
696de57606cSSage Weil {
697de57606cSSage Weil 	return !RB_EMPTY_ROOT(&ci->i_caps);
698de57606cSSage Weil }
699de57606cSSage Weil 
700de57606cSSage Weil extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented);
701de57606cSSage Weil extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t);
7021af16d54SXiubo Li extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask,
7031af16d54SXiubo Li 					  int t);
704de57606cSSage Weil extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
705de57606cSSage Weil 				    struct ceph_cap *cap);
706de57606cSSage Weil 
707de57606cSSage Weil static inline int ceph_caps_issued(struct ceph_inode_info *ci)
708de57606cSSage Weil {
709de57606cSSage Weil 	int issued;
710be655596SSage Weil 	spin_lock(&ci->i_ceph_lock);
711de57606cSSage Weil 	issued = __ceph_caps_issued(ci, NULL);
712be655596SSage Weil 	spin_unlock(&ci->i_ceph_lock);
713de57606cSSage Weil 	return issued;
714de57606cSSage Weil }
715de57606cSSage Weil 
7161af16d54SXiubo Li static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci,
7171af16d54SXiubo Li 					       int mask, int touch)
718de57606cSSage Weil {
719de57606cSSage Weil 	int r;
720be655596SSage Weil 	spin_lock(&ci->i_ceph_lock);
7211af16d54SXiubo Li 	r = __ceph_caps_issued_mask_metric(ci, mask, touch);
722be655596SSage Weil 	spin_unlock(&ci->i_ceph_lock);
723de57606cSSage Weil 	return r;
724de57606cSSage Weil }
725de57606cSSage Weil 
726de57606cSSage Weil static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
727de57606cSSage Weil {
728de57606cSSage Weil 	return ci->i_dirty_caps | ci->i_flushing_caps;
729de57606cSSage Weil }
730f66fd9f0SYan, Zheng extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
731f66fd9f0SYan, Zheng extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
732f66fd9f0SYan, Zheng extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
733f66fd9f0SYan, Zheng 				  struct ceph_cap_flush **pcf);
734de57606cSSage Weil 
7359563f88cSYan, Zheng extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
7369563f88cSYan, Zheng 				      struct ceph_cap *ocap, int mask);
737de57606cSSage Weil extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask);
738de57606cSSage Weil extern int __ceph_caps_used(struct ceph_inode_info *ci);
739de57606cSSage Weil 
740719a2514SYan, Zheng static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci)
741719a2514SYan, Zheng {
742719a2514SYan, Zheng 	return ci->i_nr_by_mode[0];
743719a2514SYan, Zheng }
744de57606cSSage Weil extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci);
745525d15e8SYan, Zheng extern int __ceph_caps_wanted(struct ceph_inode_info *ci);
746de57606cSSage Weil 
747de57606cSSage Weil /* what the mds thinks we want */
748c1944fedSYan, Zheng extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
749de57606cSSage Weil 
75037151668SYehuda Sadeh extern void ceph_caps_init(struct ceph_mds_client *mdsc);
75137151668SYehuda Sadeh extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
752fe33032dSYan, Zheng extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc,
753fe33032dSYan, Zheng 				     struct ceph_mount_options *fsopt);
754e30ee581SZhi Zhang extern int ceph_reserve_caps(struct ceph_mds_client *mdsc,
75537151668SYehuda Sadeh 			     struct ceph_cap_reservation *ctx, int need);
7567bf8f736SChengguang Xu extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc,
75737151668SYehuda Sadeh 			       struct ceph_cap_reservation *ctx);
7583d14c5d2SYehuda Sadeh extern void ceph_reservation_status(struct ceph_fs_client *client,
759de57606cSSage Weil 				    int *total, int *avail, int *used,
76085ccce43SSage Weil 				    int *reserved, int *min);
761de57606cSSage Weil 
762de57606cSSage Weil 
763de57606cSSage Weil 
764de57606cSSage Weil /*
765de57606cSSage Weil  * we keep buffered readdir results attached to file->private_data
766de57606cSSage Weil  */
7674918b6d1SSage Weil #define CEPH_F_SYNC     1
7689cfa1098SSage Weil #define CEPH_F_ATEND    2
7694918b6d1SSage Weil 
770de57606cSSage Weil struct ceph_file_info {
771252c6728SSage Weil 	short fmode;     /* initialized on open */
772252c6728SSage Weil 	short flags;     /* CEPH_F_* */
773de57606cSSage Weil 
7745d988308SYan, Zheng 	spinlock_t rw_contexts_lock;
7755d988308SYan, Zheng 	struct list_head rw_contexts;
776f4b97866SYan, Zheng 
77781f148a9SYan, Zheng 	u32 filp_gen;
778ff5d913dSYan, Zheng 	atomic_t num_locks;
779bb48bd4dSChengguang Xu };
780bb48bd4dSChengguang Xu 
781bb48bd4dSChengguang Xu struct ceph_dir_file_info {
782bb48bd4dSChengguang Xu 	struct ceph_file_info file_info;
7835d988308SYan, Zheng 
784de57606cSSage Weil 	/* readdir: position within the dir */
785de57606cSSage Weil 	u32 frag;
786de57606cSSage Weil 	struct ceph_mds_request *last_readdir;
787de57606cSSage Weil 
788de57606cSSage Weil 	/* readdir: position within a frag */
789f0494206SYan, Zheng 	unsigned next_offset;  /* offset of next chunk (last_name's + 1) */
790de57606cSSage Weil 	char *last_name;       /* last entry in previous chunk */
791fdd4e158SYan, Zheng 	long long dir_release_count;
792fdd4e158SYan, Zheng 	long long dir_ordered_count;
793fdd4e158SYan, Zheng 	int readdir_cache_idx;
794de57606cSSage Weil 
795de57606cSSage Weil 	/* used for -o dirstat read() on directory thing */
796de57606cSSage Weil 	char *dir_info;
797de57606cSSage Weil 	int dir_info_len;
798de57606cSSage Weil };
799de57606cSSage Weil 
8005d988308SYan, Zheng struct ceph_rw_context {
8015d988308SYan, Zheng 	struct list_head list;
8025d988308SYan, Zheng 	struct task_struct *thread;
8035d988308SYan, Zheng 	int caps;
8045d988308SYan, Zheng };
8055d988308SYan, Zheng 
8065d988308SYan, Zheng #define CEPH_DEFINE_RW_CONTEXT(_name, _caps)	\
8075d988308SYan, Zheng 	struct ceph_rw_context _name = {	\
8085d988308SYan, Zheng 		.thread = current,		\
8095d988308SYan, Zheng 		.caps = _caps,			\
8105d988308SYan, Zheng 	}
8115d988308SYan, Zheng 
8125d988308SYan, Zheng static inline void ceph_add_rw_context(struct ceph_file_info *cf,
8135d988308SYan, Zheng 				       struct ceph_rw_context *ctx)
8145d988308SYan, Zheng {
8155d988308SYan, Zheng 	spin_lock(&cf->rw_contexts_lock);
8165d988308SYan, Zheng 	list_add(&ctx->list, &cf->rw_contexts);
8175d988308SYan, Zheng 	spin_unlock(&cf->rw_contexts_lock);
8185d988308SYan, Zheng }
8195d988308SYan, Zheng 
8205d988308SYan, Zheng static inline void ceph_del_rw_context(struct ceph_file_info *cf,
8215d988308SYan, Zheng 				       struct ceph_rw_context *ctx)
8225d988308SYan, Zheng {
8235d988308SYan, Zheng 	spin_lock(&cf->rw_contexts_lock);
8245d988308SYan, Zheng 	list_del(&ctx->list);
8255d988308SYan, Zheng 	spin_unlock(&cf->rw_contexts_lock);
8265d988308SYan, Zheng }
8275d988308SYan, Zheng 
8285d988308SYan, Zheng static inline struct ceph_rw_context*
8295d988308SYan, Zheng ceph_find_rw_context(struct ceph_file_info *cf)
8305d988308SYan, Zheng {
8315d988308SYan, Zheng 	struct ceph_rw_context *ctx, *found = NULL;
8325d988308SYan, Zheng 	spin_lock(&cf->rw_contexts_lock);
8335d988308SYan, Zheng 	list_for_each_entry(ctx, &cf->rw_contexts, list) {
8345d988308SYan, Zheng 		if (ctx->thread == current) {
8355d988308SYan, Zheng 			found = ctx;
8365d988308SYan, Zheng 			break;
8375d988308SYan, Zheng 		}
8385d988308SYan, Zheng 	}
8395d988308SYan, Zheng 	spin_unlock(&cf->rw_contexts_lock);
8405d988308SYan, Zheng 	return found;
8415d988308SYan, Zheng }
8425d988308SYan, Zheng 
843fdd4e158SYan, Zheng struct ceph_readdir_cache_control {
844fdd4e158SYan, Zheng 	struct page  *page;
845fdd4e158SYan, Zheng 	struct dentry **dentries;
846fdd4e158SYan, Zheng 	int index;
847fdd4e158SYan, Zheng };
848de57606cSSage Weil 
849de57606cSSage Weil /*
850de57606cSSage Weil  * A "snap realm" describes a subset of the file hierarchy sharing
851de57606cSSage Weil  * the same set of snapshots that apply to it.  The realms themselves
852de57606cSSage Weil  * are organized into a hierarchy, such that children inherit (some of)
853de57606cSSage Weil  * the snapshots of their parents.
854de57606cSSage Weil  *
855de57606cSSage Weil  * All inodes within the realm that have capabilities are linked into a
856de57606cSSage Weil  * per-realm list.
857de57606cSSage Weil  */
858de57606cSSage Weil struct ceph_snap_realm {
859de57606cSSage Weil 	u64 ino;
860e3161f17SLuis Henriques 	struct inode *inode;
861de57606cSSage Weil 	atomic_t nref;
862a105f00cSSage Weil 	struct rb_node node;
863a105f00cSSage Weil 
864de57606cSSage Weil 	u64 created, seq;
865de57606cSSage Weil 	u64 parent_ino;
866de57606cSSage Weil 	u64 parent_since;   /* snapid when our current parent became so */
867de57606cSSage Weil 
868de57606cSSage Weil 	u64 *prior_parent_snaps;      /* snaps inherited from any parents we */
869aa711ee3SAlex Elder 	u32 num_prior_parent_snaps;   /*  had prior to parent_since */
870de57606cSSage Weil 	u64 *snaps;                   /* snaps specific to this realm */
871aa711ee3SAlex Elder 	u32 num_snaps;
872de57606cSSage Weil 
873de57606cSSage Weil 	struct ceph_snap_realm *parent;
874de57606cSSage Weil 	struct list_head children;       /* list of child realms */
875de57606cSSage Weil 	struct list_head child_item;
876de57606cSSage Weil 
877de57606cSSage Weil 	struct list_head empty_item;     /* if i have ref==0 */
878de57606cSSage Weil 
879ae00d4f3SSage Weil 	struct list_head dirty_item;     /* if realm needs new context */
880ae00d4f3SSage Weil 
881de57606cSSage Weil 	/* the current set of snaps for this realm */
882de57606cSSage Weil 	struct ceph_snap_context *cached_context;
883de57606cSSage Weil 
884de57606cSSage Weil 	struct list_head inodes_with_caps;
885de57606cSSage Weil 	spinlock_t inodes_with_caps_lock;
886de57606cSSage Weil };
887de57606cSSage Weil 
8883d14c5d2SYehuda Sadeh static inline int default_congestion_kb(void)
8893d14c5d2SYehuda Sadeh {
8903d14c5d2SYehuda Sadeh 	int congestion_kb;
891de57606cSSage Weil 
892de57606cSSage Weil 	/*
8933d14c5d2SYehuda Sadeh 	 * Copied from NFS
8943d14c5d2SYehuda Sadeh 	 *
8953d14c5d2SYehuda Sadeh 	 * congestion size, scale with available memory.
8963d14c5d2SYehuda Sadeh 	 *
8973d14c5d2SYehuda Sadeh 	 *  64MB:    8192k
8983d14c5d2SYehuda Sadeh 	 * 128MB:   11585k
8993d14c5d2SYehuda Sadeh 	 * 256MB:   16384k
9003d14c5d2SYehuda Sadeh 	 * 512MB:   23170k
9013d14c5d2SYehuda Sadeh 	 *   1GB:   32768k
9023d14c5d2SYehuda Sadeh 	 *   2GB:   46340k
9033d14c5d2SYehuda Sadeh 	 *   4GB:   65536k
9043d14c5d2SYehuda Sadeh 	 *   8GB:   92681k
9053d14c5d2SYehuda Sadeh 	 *  16GB:  131072k
9063d14c5d2SYehuda Sadeh 	 *
9073d14c5d2SYehuda Sadeh 	 * This allows larger machines to have larger/more transfers.
9083d14c5d2SYehuda Sadeh 	 * Limit the default to 256M
909de57606cSSage Weil 	 */
910ca79b0c2SArun KS 	congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10);
9113d14c5d2SYehuda Sadeh 	if (congestion_kb > 256*1024)
9123d14c5d2SYehuda Sadeh 		congestion_kb = 256*1024;
9133d14c5d2SYehuda Sadeh 
9143d14c5d2SYehuda Sadeh 	return congestion_kb;
915de57606cSSage Weil }
916de57606cSSage Weil 
917de57606cSSage Weil 
918d468e729SYan, Zheng /* super.c */
919d468e729SYan, Zheng extern int ceph_force_reconnect(struct super_block *sb);
920de57606cSSage Weil /* snap.c */
921de57606cSSage Weil struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
922de57606cSSage Weil 					       u64 ino);
923de57606cSSage Weil extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
924de57606cSSage Weil 				struct ceph_snap_realm *realm);
925de57606cSSage Weil extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc,
926de57606cSSage Weil 				struct ceph_snap_realm *realm);
927de57606cSSage Weil extern int ceph_update_snap_trace(struct ceph_mds_client *m,
928982d6011SYan, Zheng 				  void *p, void *e, bool deletion,
929982d6011SYan, Zheng 				  struct ceph_snap_realm **realm_ret);
9300ba92e1cSJeff Layton void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm);
931de57606cSSage Weil extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
9322600d2ddSSage Weil 			     struct ceph_mds_session *session,
933de57606cSSage Weil 			     struct ceph_msg *msg);
934de57606cSSage Weil extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
935de57606cSSage Weil 				  struct ceph_cap_snap *capsnap);
936de57606cSSage Weil extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
937de57606cSSage Weil 
93875c9627eSYan, Zheng extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc,
93975c9627eSYan, Zheng 						   u64 snap);
94075c9627eSYan, Zheng extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc,
94175c9627eSYan, Zheng 				struct ceph_snapid_map *sm);
94275c9627eSYan, Zheng extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc);
94375c9627eSYan, Zheng extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc);
944631ed4b0SJeff Layton void ceph_umount_begin(struct super_block *sb);
94575c9627eSYan, Zheng 
94675c9627eSYan, Zheng 
947de57606cSSage Weil /*
948de57606cSSage Weil  * a cap_snap is "pending" if it is still awaiting an in-progress
949de57606cSSage Weil  * sync write (that may/may not still update size, mtime, etc.).
950de57606cSSage Weil  */
951de57606cSSage Weil static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
952de57606cSSage Weil {
953de57606cSSage Weil 	return !list_empty(&ci->i_cap_snaps) &&
95486056090SYan, Zheng 	       list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
955de57606cSSage Weil 			       ci_item)->writing;
956de57606cSSage Weil }
957de57606cSSage Weil 
958de57606cSSage Weil /* inode.c */
959966c7160SJeff Layton struct ceph_mds_reply_info_in;
960966c7160SJeff Layton struct ceph_mds_reply_dirfrag;
961966c7160SJeff Layton 
962de57606cSSage Weil extern const struct inode_operations ceph_file_iops;
963de57606cSSage Weil 
964de57606cSSage Weil extern struct inode *ceph_alloc_inode(struct super_block *sb);
96587bc5b89SYan, Zheng extern void ceph_evict_inode(struct inode *inode);
966cfa6d412SAl Viro extern void ceph_free_inode(struct inode *inode);
967de57606cSSage Weil 
968de57606cSSage Weil extern struct inode *ceph_get_inode(struct super_block *sb,
969de57606cSSage Weil 				    struct ceph_vino vino);
970de57606cSSage Weil extern struct inode *ceph_get_snapdir(struct inode *parent);
971de57606cSSage Weil extern int ceph_fill_file_size(struct inode *inode, int issued,
972de57606cSSage Weil 			       u32 truncate_seq, u64 truncate_size, u64 size);
973de57606cSSage Weil extern void ceph_fill_file_time(struct inode *inode, int issued,
9749bbeab41SArnd Bergmann 				u64 time_warp_seq, struct timespec64 *ctime,
9759bbeab41SArnd Bergmann 				struct timespec64 *mtime,
9769bbeab41SArnd Bergmann 				struct timespec64 *atime);
977966c7160SJeff Layton extern int ceph_fill_inode(struct inode *inode, struct page *locked_page,
978966c7160SJeff Layton 		    struct ceph_mds_reply_info_in *iinfo,
979966c7160SJeff Layton 		    struct ceph_mds_reply_dirfrag *dirinfo,
980966c7160SJeff Layton 		    struct ceph_mds_session *session, int cap_fmode,
981966c7160SJeff Layton 		    struct ceph_cap_reservation *caps_reservation);
982de57606cSSage Weil extern int ceph_fill_trace(struct super_block *sb,
983f5a03b08SJeff Layton 			   struct ceph_mds_request *req);
984de57606cSSage Weil extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
985de57606cSSage Weil 				    struct ceph_mds_session *session);
986de57606cSSage Weil 
987de57606cSSage Weil extern int ceph_inode_holds_cap(struct inode *inode, int mask);
988de57606cSSage Weil 
989efb0ca76SYan, Zheng extern bool ceph_inode_set_size(struct inode *inode, loff_t size);
990b415bf4fSYan, Zheng extern void __ceph_do_pending_vmtruncate(struct inode *inode);
99164f28c62SJeff Layton 
99264f28c62SJeff Layton void ceph_queue_inode_work(struct inode *inode, int work_bit);
99364f28c62SJeff Layton 
99464f28c62SJeff Layton static inline void ceph_queue_vmtruncate(struct inode *inode)
99564f28c62SJeff Layton {
99664f28c62SJeff Layton 	ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE);
99764f28c62SJeff Layton }
99864f28c62SJeff Layton 
99964f28c62SJeff Layton static inline void ceph_queue_invalidate(struct inode *inode)
100064f28c62SJeff Layton {
100164f28c62SJeff Layton 	ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES);
100264f28c62SJeff Layton }
100364f28c62SJeff Layton 
100464f28c62SJeff Layton static inline void ceph_queue_writeback(struct inode *inode)
100564f28c62SJeff Layton {
100664f28c62SJeff Layton 	ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK);
100764f28c62SJeff Layton }
100864f28c62SJeff Layton 
1009a8810cdcSJeff Layton static inline void ceph_queue_check_caps(struct inode *inode)
1010a8810cdcSJeff Layton {
1011a8810cdcSJeff Layton 	ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS);
1012a8810cdcSJeff Layton }
1013a8810cdcSJeff Layton 
1014a8810cdcSJeff Layton static inline void ceph_queue_flush_snaps(struct inode *inode)
1015a8810cdcSJeff Layton {
1016a8810cdcSJeff Layton 	ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS);
1017a8810cdcSJeff Layton }
1018a8810cdcSJeff Layton 
101901deead0SYan, Zheng extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
102001deead0SYan, Zheng 			     int mask, bool force);
102101deead0SYan, Zheng static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
102201deead0SYan, Zheng {
102301deead0SYan, Zheng 	return __ceph_do_getattr(inode, NULL, mask, force);
102401deead0SYan, Zheng }
1025549c7297SChristian Brauner extern int ceph_permission(struct user_namespace *mnt_userns,
1026549c7297SChristian Brauner 			   struct inode *inode, int mask);
1027a26feccaSAndreas Gruenbacher extern int __ceph_setattr(struct inode *inode, struct iattr *attr);
1028549c7297SChristian Brauner extern int ceph_setattr(struct user_namespace *mnt_userns,
1029549c7297SChristian Brauner 			struct dentry *dentry, struct iattr *attr);
1030549c7297SChristian Brauner extern int ceph_getattr(struct user_namespace *mnt_userns,
1031549c7297SChristian Brauner 			const struct path *path, struct kstat *stat,
1032a528d35eSDavid Howells 			u32 request_mask, unsigned int flags);
10335d6451b1SJeff Layton void ceph_inode_shutdown(struct inode *inode);
10345d6451b1SJeff Layton 
10355d6451b1SJeff Layton static inline bool ceph_inode_is_shutdown(struct inode *inode)
10365d6451b1SJeff Layton {
10375d6451b1SJeff Layton 	unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags);
10385d6451b1SJeff Layton 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
10395d6451b1SJeff Layton 	int state = READ_ONCE(fsc->mount_state);
10405d6451b1SJeff Layton 
10415d6451b1SJeff Layton 	return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN;
10425d6451b1SJeff Layton }
1043de57606cSSage Weil 
1044de57606cSSage Weil /* xattr.c */
1045a26feccaSAndreas Gruenbacher int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
10467221fe4cSGuangliang Zhao ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
1047de57606cSSage Weil extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
104812fe3ddaSLuis Henriques extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci);
1049de57606cSSage Weil extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci);
1050b1ee94aaSYan, Zheng extern const struct xattr_handler *ceph_xattr_handlers[];
1051de57606cSSage Weil 
10525c31e92dSYan, Zheng struct ceph_acl_sec_ctx {
10535c31e92dSYan, Zheng #ifdef CONFIG_CEPH_FS_POSIX_ACL
10545c31e92dSYan, Zheng 	void *default_acl;
10555c31e92dSYan, Zheng 	void *acl;
10565c31e92dSYan, Zheng #endif
1057ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL
1058ac6713ccSYan, Zheng 	void *sec_ctx;
1059ac6713ccSYan, Zheng 	u32 sec_ctxlen;
1060ac6713ccSYan, Zheng #endif
10615c31e92dSYan, Zheng 	struct ceph_pagelist *pagelist;
10625c31e92dSYan, Zheng };
10635c31e92dSYan, Zheng 
1064315f2408SYan, Zheng #ifdef CONFIG_SECURITY
1065315f2408SYan, Zheng extern bool ceph_security_xattr_deadlock(struct inode *in);
1066315f2408SYan, Zheng extern bool ceph_security_xattr_wanted(struct inode *in);
1067315f2408SYan, Zheng #else
1068315f2408SYan, Zheng static inline bool ceph_security_xattr_deadlock(struct inode *in)
1069315f2408SYan, Zheng {
1070315f2408SYan, Zheng 	return false;
1071315f2408SYan, Zheng }
1072315f2408SYan, Zheng static inline bool ceph_security_xattr_wanted(struct inode *in)
1073315f2408SYan, Zheng {
1074315f2408SYan, Zheng 	return false;
1075315f2408SYan, Zheng }
1076315f2408SYan, Zheng #endif
1077315f2408SYan, Zheng 
1078ac6713ccSYan, Zheng #ifdef CONFIG_CEPH_FS_SECURITY_LABEL
1079ac6713ccSYan, Zheng extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
1080ac6713ccSYan, Zheng 				     struct ceph_acl_sec_ctx *ctx);
1081668959a5SJeff Layton static inline void ceph_security_invalidate_secctx(struct inode *inode)
1082668959a5SJeff Layton {
1083668959a5SJeff Layton 	security_inode_invalidate_secctx(inode);
1084668959a5SJeff Layton }
1085ac6713ccSYan, Zheng #else
1086ac6713ccSYan, Zheng static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
1087ac6713ccSYan, Zheng 					    struct ceph_acl_sec_ctx *ctx)
1088ac6713ccSYan, Zheng {
1089ac6713ccSYan, Zheng 	return 0;
1090ac6713ccSYan, Zheng }
1091ac6713ccSYan, Zheng static inline void ceph_security_invalidate_secctx(struct inode *inode)
1092ac6713ccSYan, Zheng {
1093ac6713ccSYan, Zheng }
1094ac6713ccSYan, Zheng #endif
10957221fe4cSGuangliang Zhao 
10965c31e92dSYan, Zheng void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx);
10977221fe4cSGuangliang Zhao 
10985c31e92dSYan, Zheng /* acl.c */
10997221fe4cSGuangliang Zhao #ifdef CONFIG_CEPH_FS_POSIX_ACL
11007221fe4cSGuangliang Zhao 
11010cad6246SMiklos Szeredi struct posix_acl *ceph_get_acl(struct inode *, int, bool);
1102549c7297SChristian Brauner int ceph_set_acl(struct user_namespace *mnt_userns,
1103549c7297SChristian Brauner 		 struct inode *inode, struct posix_acl *acl, int type);
1104b1ee94aaSYan, Zheng int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
11055c31e92dSYan, Zheng 		       struct ceph_acl_sec_ctx *as_ctx);
11065c31e92dSYan, Zheng void ceph_init_inode_acls(struct inode *inode,
11075c31e92dSYan, Zheng 			  struct ceph_acl_sec_ctx *as_ctx);
1108c969d9bfSGuangliang Zhao 
1109c969d9bfSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode)
1110c969d9bfSGuangliang Zhao {
1111c969d9bfSGuangliang Zhao        forget_all_cached_acls(inode);
1112c969d9bfSGuangliang Zhao }
11137221fe4cSGuangliang Zhao 
11147221fe4cSGuangliang Zhao #else
11157221fe4cSGuangliang Zhao 
11167221fe4cSGuangliang Zhao #define ceph_get_acl NULL
111772466d0bSSage Weil #define ceph_set_acl NULL
11187221fe4cSGuangliang Zhao 
1119b1ee94aaSYan, Zheng static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
11205c31e92dSYan, Zheng 				     struct ceph_acl_sec_ctx *as_ctx)
11217221fe4cSGuangliang Zhao {
11227221fe4cSGuangliang Zhao 	return 0;
11237221fe4cSGuangliang Zhao }
1124b1ee94aaSYan, Zheng static inline void ceph_init_inode_acls(struct inode *inode,
11255c31e92dSYan, Zheng 					struct ceph_acl_sec_ctx *as_ctx)
1126b1ee94aaSYan, Zheng {
1127b1ee94aaSYan, Zheng }
11287221fe4cSGuangliang Zhao static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode)
11297221fe4cSGuangliang Zhao {
11307221fe4cSGuangliang Zhao 	return 0;
11317221fe4cSGuangliang Zhao }
11327221fe4cSGuangliang Zhao 
11337221fe4cSGuangliang Zhao static inline void ceph_forget_all_cached_acls(struct inode *inode)
11347221fe4cSGuangliang Zhao {
11357221fe4cSGuangliang Zhao }
11367221fe4cSGuangliang Zhao 
11377221fe4cSGuangliang Zhao #endif
11387221fe4cSGuangliang Zhao 
1139de57606cSSage Weil /* caps.c */
1140de57606cSSage Weil extern const char *ceph_cap_string(int c);
1141de57606cSSage Weil extern void ceph_handle_caps(struct ceph_mds_session *session,
1142de57606cSSage Weil 			     struct ceph_msg *msg);
1143d9df2783SYan, Zheng extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc,
1144d9df2783SYan, Zheng 				     struct ceph_cap_reservation *ctx);
1145d9df2783SYan, Zheng extern void ceph_add_cap(struct inode *inode,
1146de57606cSSage Weil 			 struct ceph_mds_session *session, u64 cap_id,
1147135e671eSYan, Zheng 			 unsigned issued, unsigned wanted,
1148de57606cSSage Weil 			 unsigned cap, unsigned seq, u64 realmino, int flags,
1149d9df2783SYan, Zheng 			 struct ceph_cap **new_cap);
1150a096b09aSYan, Zheng extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
1151a76d0a9cSXiubo Li extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release);
1152d6e47819SYan, Zheng extern void __ceph_remove_caps(struct ceph_inode_info *ci);
115337151668SYehuda Sadeh extern void ceph_put_cap(struct ceph_mds_client *mdsc,
115437151668SYehuda Sadeh 			 struct ceph_cap *cap);
11559215aeeaSYan, Zheng extern int ceph_is_any_caps(struct inode *inode);
1156de57606cSSage Weil 
1157f1a3d572SStephen Rothwell extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
115802c24a82SJosef Bacik extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
115902c24a82SJosef Bacik 		      int datasync);
1160e548e9b9SYan, Zheng extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
1161e548e9b9SYan, Zheng 					  struct ceph_mds_session *session);
1162de57606cSSage Weil extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
1163de57606cSSage Weil 				    struct ceph_mds_session *session);
1164e8a4d267SJeff Layton void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session,
1165e8a4d267SJeff Layton 				   struct ceph_inode_info *ci);
11662bc50259SGreg Farnum extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
11672bc50259SGreg Farnum 					     int mds);
116840dcf75eSJeff Layton extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps,
116940dcf75eSJeff Layton 				bool snap_rwsem_locked);
1170de57606cSSage Weil extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
1171de57606cSSage Weil extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
1172a8810cdcSJeff Layton extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had);
1173e64f44a8SXiubo Li extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci,
1174e64f44a8SXiubo Li 					    int had);
1175de57606cSSage Weil extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
1176de57606cSSage Weil 				       struct ceph_snap_context *snapc);
1177a6d37ccdSXiubo Li extern void __ceph_remove_capsnap(struct inode *inode,
1178a6d37ccdSXiubo Li 				  struct ceph_cap_snap *capsnap,
1179a6d37ccdSXiubo Li 				  bool *wake_ci, bool *wake_mdsc);
1180a6d37ccdSXiubo Li extern void ceph_remove_capsnap(struct inode *inode,
1181a6d37ccdSXiubo Li 				struct ceph_cap_snap *capsnap,
1182a6d37ccdSXiubo Li 				bool *wake_ci, bool *wake_mdsc);
1183ed9b430cSYan, Zheng extern void ceph_flush_snaps(struct ceph_inode_info *ci,
11840e294387SYan, Zheng 			     struct ceph_mds_session **psession);
1185efb0ca76SYan, Zheng extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
1186de57606cSSage Weil extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
1187de57606cSSage Weil 			    struct ceph_mds_session *session);
1188bf2ba432SLuis Henriques extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
1189afcdaea3SSage Weil extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
11906ef0bc6dSZhi Zhang extern int  ceph_drop_caps_for_unlink(struct inode *inode);
1191de57606cSSage Weil extern int ceph_encode_inode_release(void **p, struct inode *inode,
1192de57606cSSage Weil 				     int mds, int drop, int unless, int force);
1193de57606cSSage Weil extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
1194ca6c8ae0SJeff Layton 				      struct inode *dir,
1195de57606cSSage Weil 				      int mds, int drop, int unless);
1196de57606cSSage Weil 
11975e3ded1bSYan, Zheng extern int ceph_get_caps(struct file *filp, int need, int want,
1198e72968e1SJeff Layton 			 loff_t endoff, int *got);
11995e3ded1bSYan, Zheng extern int ceph_try_get_caps(struct inode *inode,
12002ee9dd95SLuis Henriques 			     int need, int want, bool nonblock, int *got);
1201de57606cSSage Weil 
1202de57606cSSage Weil /* for counting open files by mode */
1203719a2514SYan, Zheng extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count);
1204719a2514SYan, Zheng extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count);
1205719a2514SYan, Zheng extern void __ceph_touch_fmode(struct ceph_inode_info *ci,
1206719a2514SYan, Zheng 			       struct ceph_mds_client *mdsc, int fmode);
1207de57606cSSage Weil 
1208de57606cSSage Weil /* addr.c */
1209de57606cSSage Weil extern const struct address_space_operations ceph_aops;
1210de57606cSSage Weil extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
121110183a69SYan, Zheng extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
12125e3ded1bSYan, Zheng extern int ceph_pool_perm_check(struct inode *inode, int need);
121310183a69SYan, Zheng extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
121436e6da98SJeff Layton int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate);
1215de57606cSSage Weil 
1216de57606cSSage Weil /* file.c */
1217de57606cSSage Weil extern const struct file_operations ceph_file_fops;
12189e0eb85dSAlex Elder 
1219719a2514SYan, Zheng extern int ceph_renew_caps(struct inode *inode, int fmode);
1220de57606cSSage Weil extern int ceph_open(struct inode *inode, struct file *file);
12215ef50c3bSSage Weil extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
122244907d79SAl Viro 			    struct file *file, unsigned flags, umode_t mode);
1223de57606cSSage Weil extern int ceph_release(struct inode *inode, struct file *filp);
122431c542a1SYan, Zheng extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
122531c542a1SYan, Zheng 				  char *data, size_t len);
122655f2a045SIlya Dryomov 
1227de57606cSSage Weil /* dir.c */
1228de57606cSSage Weil extern const struct file_operations ceph_dir_fops;
122938c48b5fSYan, Zheng extern const struct file_operations ceph_snapdir_fops;
1230de57606cSSage Weil extern const struct inode_operations ceph_dir_iops;
123138c48b5fSYan, Zheng extern const struct inode_operations ceph_snapdir_iops;
123218fc8abdSAl Viro extern const struct dentry_operations ceph_dentry_ops;
1233de57606cSSage Weil 
1234f3c4ebe6SYan, Zheng extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order);
1235de57606cSSage Weil extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry);
1236aa60cfc3SJeff Layton extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req,
12377a971e2cSJeff Layton 			       struct dentry *dentry);
1238de57606cSSage Weil extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req,
1239de57606cSSage Weil 					 struct dentry *dentry, int err);
1240de57606cSSage Weil 
124137c4efc1SYan, Zheng extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di);
124237c4efc1SYan, Zheng extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di);
124381a6cf2dSSage Weil extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
124437c4efc1SYan, Zheng extern int ceph_trim_dentries(struct ceph_mds_client *mdsc);
1245e5f86dc3SSage Weil extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
1246fdd4e158SYan, Zheng extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
1247de57606cSSage Weil 
1248de57606cSSage Weil /* ioctl.c */
1249de57606cSSage Weil extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1250de57606cSSage Weil 
1251de57606cSSage Weil /* export.c */
1252de57606cSSage Weil extern const struct export_operations ceph_export_ops;
12533886274aSLuis Henriques struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino);
1254de57606cSSage Weil 
125540819f6fSGreg Farnum /* locks.c */
1256eb13e832SYan, Zheng extern __init void ceph_flock_init(void);
125740819f6fSGreg Farnum extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl);
125840819f6fSGreg Farnum extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl);
125940819f6fSGreg Farnum extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num);
126039be95e9SJim Schutt extern int ceph_encode_locks_to_buffer(struct inode *inode,
126139be95e9SJim Schutt 				       struct ceph_filelock *flocks,
126239be95e9SJim Schutt 				       int num_fcntl_locks,
126339be95e9SJim Schutt 				       int num_flock_locks);
126439be95e9SJim Schutt extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
126539be95e9SJim Schutt 				  struct ceph_pagelist *pagelist,
126639be95e9SJim Schutt 				  int num_fcntl_locks, int num_flock_locks);
126740819f6fSGreg Farnum 
12683d14c5d2SYehuda Sadeh /* debugfs.c */
12691a829ff2SGreg Kroah-Hartman extern void ceph_fs_debugfs_init(struct ceph_fs_client *client);
12703d14c5d2SYehuda Sadeh extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client);
12713d14c5d2SYehuda Sadeh 
1272fb18a575SLuis Henriques /* quota.c */
1273d557c48dSLuis Henriques static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci)
1274d557c48dSLuis Henriques {
1275d557c48dSLuis Henriques 	return ci->i_max_files || ci->i_max_bytes;
1276d557c48dSLuis Henriques }
1277d557c48dSLuis Henriques 
1278d557c48dSLuis Henriques extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc);
1279d557c48dSLuis Henriques 
1280d557c48dSLuis Henriques static inline void __ceph_update_quota(struct ceph_inode_info *ci,
1281d557c48dSLuis Henriques 				       u64 max_bytes, u64 max_files)
1282d557c48dSLuis Henriques {
1283d557c48dSLuis Henriques 	bool had_quota, has_quota;
1284d557c48dSLuis Henriques 	had_quota = __ceph_has_any_quota(ci);
1285d557c48dSLuis Henriques 	ci->i_max_bytes = max_bytes;
1286d557c48dSLuis Henriques 	ci->i_max_files = max_files;
1287d557c48dSLuis Henriques 	has_quota = __ceph_has_any_quota(ci);
1288d557c48dSLuis Henriques 
1289d557c48dSLuis Henriques 	if (had_quota != has_quota)
1290d557c48dSLuis Henriques 		ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota);
1291d557c48dSLuis Henriques }
1292d557c48dSLuis Henriques 
1293fb18a575SLuis Henriques extern void ceph_handle_quota(struct ceph_mds_client *mdsc,
1294fb18a575SLuis Henriques 			      struct ceph_mds_session *session,
1295fb18a575SLuis Henriques 			      struct ceph_msg *msg);
1296b7a29217SLuis Henriques extern bool ceph_quota_is_max_files_exceeded(struct inode *inode);
12976646ea1cSLuis Henriques extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new);
12982b83845fSLuis Henriques extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode,
12992b83845fSLuis Henriques 					     loff_t newlen);
13001ab302a0SLuis Henriques extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode,
13011ab302a0SLuis Henriques 						loff_t newlen);
13029122eed5SLuis Henriques extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc,
13039122eed5SLuis Henriques 				     struct kstatfs *buf);
13040c44a8e0SLuis Henriques extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc);
1305fb18a575SLuis Henriques 
1306de57606cSSage Weil #endif /* _FS_CEPH_SUPER_H */
1307