Greg Kroah-Hartman | b244131 | 2017-11-01 15:07:57 +0100 | [diff] [blame] | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 2 | #ifndef _FS_CEPH_SUPER_H |
| 3 | #define _FS_CEPH_SUPER_H |
| 4 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 5 | #include <linux/ceph/ceph_debug.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 6 | |
| 7 | #include <asm/unaligned.h> |
| 8 | #include <linux/backing-dev.h> |
| 9 | #include <linux/completion.h> |
| 10 | #include <linux/exportfs.h> |
| 11 | #include <linux/fs.h> |
| 12 | #include <linux/mempool.h> |
| 13 | #include <linux/pagemap.h> |
| 14 | #include <linux/wait.h> |
Stephen Rothwell | f1a3d57 | 2010-01-18 11:53:08 +1100 | [diff] [blame] | 15 | #include <linux/writeback.h> |
Tejun Heo | 5a0e3ad | 2010-03-24 17:04:11 +0900 | [diff] [blame] | 16 | #include <linux/slab.h> |
Guangliang Zhao | c969d9b | 2014-02-16 08:35:52 -0800 | [diff] [blame] | 17 | #include <linux/posix_acl.h> |
Elena Reshetova | 805692d | 2017-03-03 11:15:07 +0200 | [diff] [blame] | 18 | #include <linux/refcount.h> |
Jeff Layton | 668959a | 2019-08-06 09:07:51 -0400 | [diff] [blame] | 19 | #include <linux/security.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 20 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 21 | #include <linux/ceph/libceph.h> |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 22 | |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 23 | #ifdef CONFIG_CEPH_FSCACHE |
Jeff Layton | 7c46b31 | 2021-01-21 16:27:14 -0500 | [diff] [blame] | 24 | #define FSCACHE_USE_NEW_IO_API |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 25 | #include <linux/fscache.h> |
| 26 | #endif |
| 27 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 28 | /* f_type in struct statfs */ |
| 29 | #define CEPH_SUPER_MAGIC 0x00c36400 |
| 30 | |
| 31 | /* large granularity for statfs utilization stats to facilitate |
| 32 | * large volume sizes on 32-bit machines. */ |
Sage Weil | 92a49fb | 2013-02-22 15:31:00 -0800 | [diff] [blame] | 33 | #define CEPH_BLOCK_SHIFT 22 /* 4 MB */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 34 | #define CEPH_BLOCK (1 << CEPH_BLOCK_SHIFT) |
Kotresh HR | 8e55ba8 | 2021-11-10 23:30:21 +0530 | [diff] [blame] | 35 | #define CEPH_4K_BLOCK_SHIFT 12 /* 4 KB */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 36 | |
Ilya Dryomov | 0b98acd | 2020-09-14 13:39:19 +0200 | [diff] [blame] | 37 | #define CEPH_MOUNT_OPT_CLEANRECOVER (1<<1) /* auto reonnect (clean mode) after blocklisted */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 38 | #define CEPH_MOUNT_OPT_DIRSTAT (1<<4) /* `cat dirname` for stats */ |
| 39 | #define CEPH_MOUNT_OPT_RBYTES (1<<5) /* dir st_bytes = rbytes */ |
| 40 | #define CEPH_MOUNT_OPT_NOASYNCREADDIR (1<<7) /* no dcache readdir */ |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 41 | #define CEPH_MOUNT_OPT_INO32 (1<<8) /* 32 bit inos */ |
Sage Weil | a40dc6c | 2012-01-10 09:12:55 -0800 | [diff] [blame] | 42 | #define CEPH_MOUNT_OPT_DCACHE (1<<9) /* use dcache for readdir etc */ |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 43 | #define CEPH_MOUNT_OPT_FSCACHE (1<<10) /* use fscache */ |
Yan, Zheng | 10183a6 | 2015-04-27 15:33:28 +0800 | [diff] [blame] | 44 | #define CEPH_MOUNT_OPT_NOPOOLPERM (1<<11) /* no pool permission check */ |
Yan, Zheng | e9e427f | 2016-11-10 16:02:06 +0800 | [diff] [blame] | 45 | #define CEPH_MOUNT_OPT_MOUNTWAIT (1<<12) /* mount waits if no mds is up */ |
Luis Henriques | 9122eed | 2018-01-31 10:53:13 +0000 | [diff] [blame] | 46 | #define CEPH_MOUNT_OPT_NOQUOTADF (1<<13) /* no root dir quota in statfs */ |
Luis Henriques | ea4cdc5 | 2018-10-15 16:46:00 +0100 | [diff] [blame] | 47 | #define CEPH_MOUNT_OPT_NOCOPYFROM (1<<14) /* don't use RADOS 'copy-from' op */ |
Jeff Layton | 2ccb454 | 2019-04-02 15:35:56 -0400 | [diff] [blame] | 48 | #define CEPH_MOUNT_OPT_ASYNC_DIROPS (1<<15) /* allow async directory ops */ |
Sage Weil | 6a25938 | 2010-07-07 09:06:08 -0700 | [diff] [blame] | 49 | |
Luis Henriques | 6f9718f | 2018-12-10 10:23:12 +0000 | [diff] [blame] | 50 | #define CEPH_MOUNT_OPT_DEFAULT \ |
| 51 | (CEPH_MOUNT_OPT_DCACHE | \ |
Jeff Layton | f7a67b4 | 2021-08-09 11:55:15 -0400 | [diff] [blame] | 52 | CEPH_MOUNT_OPT_NOCOPYFROM | \ |
| 53 | CEPH_MOUNT_OPT_ASYNC_DIROPS) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 54 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 55 | #define ceph_set_mount_opt(fsc, opt) \ |
Jeff Layton | 2ccb454 | 2019-04-02 15:35:56 -0400 | [diff] [blame] | 56 | (fsc)->mount_options->flags |= CEPH_MOUNT_OPT_##opt |
| 57 | #define ceph_clear_mount_opt(fsc, opt) \ |
| 58 | (fsc)->mount_options->flags &= ~CEPH_MOUNT_OPT_##opt |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 59 | #define ceph_test_mount_opt(fsc, opt) \ |
| 60 | (!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt)) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 61 | |
Yan, Zheng | aa18792 | 2017-07-11 15:56:09 +0800 | [diff] [blame] | 62 | /* max size of osd read request, limited by libceph */ |
| 63 | #define CEPH_MAX_READ_SIZE CEPH_MSG_MAX_DATA_LEN |
Yan, Zheng | 95cca2b | 2017-07-11 17:34:46 +0800 | [diff] [blame] | 64 | /* osd has a configurable limitaion of max write size. |
| 65 | * CEPH_MSG_MAX_DATA_LEN should be small enough. */ |
| 66 | #define CEPH_MAX_WRITE_SIZE CEPH_MSG_MAX_DATA_LEN |
Andreas Gerstmayr | 7c94ba2 | 2017-01-10 14:17:56 +0100 | [diff] [blame] | 67 | #define CEPH_RASIZE_DEFAULT (8192*1024) /* max readahead */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 68 | #define CEPH_MAX_READDIR_DEFAULT 1024 |
| 69 | #define CEPH_MAX_READDIR_BYTES_DEFAULT (512*1024) |
| 70 | #define CEPH_SNAPDIRNAME_DEFAULT ".snap" |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 71 | |
Yan, Zheng | 4214fb1 | 2017-07-11 18:49:44 +0800 | [diff] [blame] | 72 | /* |
| 73 | * Delay telling the MDS we no longer want caps, in case we reopen |
| 74 | * the file. Delay a minimum amount of time, even if we send a cap |
| 75 | * message for some other reason. Otherwise, take the oppotunity to |
| 76 | * update the mds to avoid sending another message later. |
| 77 | */ |
| 78 | #define CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT 5 /* cap release delay */ |
| 79 | #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT 60 /* cap release delay */ |
| 80 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 81 | struct ceph_mount_options { |
Jeff Layton | ad8c28a | 2019-09-09 15:58:55 -0400 | [diff] [blame] | 82 | unsigned int flags; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 83 | |
Jeff Layton | ad8c28a | 2019-09-09 15:58:55 -0400 | [diff] [blame] | 84 | unsigned int wsize; /* max write size */ |
| 85 | unsigned int rsize; /* max read size */ |
| 86 | unsigned int rasize; /* max readahead */ |
| 87 | unsigned int congestion_kb; /* max writeback in flight */ |
| 88 | unsigned int caps_wanted_delay_min, caps_wanted_delay_max; |
Yan, Zheng | fe33032 | 2019-02-01 14:57:15 +0800 | [diff] [blame] | 89 | int caps_max; |
Jeff Layton | ad8c28a | 2019-09-09 15:58:55 -0400 | [diff] [blame] | 90 | unsigned int max_readdir; /* max readdir result (entries) */ |
| 91 | unsigned int max_readdir_bytes; /* max readdir result (bytes) */ |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 92 | |
Venky Shankar | 7b19b4d | 2021-07-14 15:35:52 +0530 | [diff] [blame] | 93 | bool new_dev_syntax; |
| 94 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 95 | /* |
| 96 | * everything above this point can be memcmp'd; everything below |
| 97 | * is handled in compare_mount_options() |
| 98 | */ |
| 99 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 100 | char *snapdir_name; /* default ".snap" */ |
Yan, Zheng | 430afba | 2016-07-08 11:25:38 +0800 | [diff] [blame] | 101 | char *mds_namespace; /* default NULL */ |
Ilya Dryomov | b27a939 | 2020-02-10 22:51:08 +0100 | [diff] [blame] | 102 | char *server_path; /* default NULL (means "/") */ |
Yan, Zheng | 1d8f836 | 2017-06-27 11:57:56 +0800 | [diff] [blame] | 103 | char *fscache_uniq; /* default NULL */ |
Venky Shankar | 7b19b4d | 2021-07-14 15:35:52 +0530 | [diff] [blame] | 104 | char *mon_addr; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 105 | }; |
| 106 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 107 | struct ceph_fs_client { |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 108 | struct super_block *sb; |
| 109 | |
Xiubo Li | 18f473b | 2020-07-16 10:05:57 -0400 | [diff] [blame] | 110 | struct list_head metric_wakeup; |
| 111 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 112 | struct ceph_mount_options *mount_options; |
| 113 | struct ceph_client *client; |
| 114 | |
Jeff Layton | aa5c791 | 2020-10-06 08:38:20 -0400 | [diff] [blame] | 115 | int mount_state; |
Yan, Zheng | 81f148a | 2019-07-25 20:16:46 +0800 | [diff] [blame] | 116 | |
Ilya Dryomov | 0b98acd | 2020-09-14 13:39:19 +0200 | [diff] [blame] | 117 | bool blocklisted; |
Yan, Zheng | 131d7eb | 2019-07-25 20:16:47 +0800 | [diff] [blame] | 118 | |
Luis Henriques | 78beb0f | 2020-01-08 10:03:53 +0000 | [diff] [blame] | 119 | bool have_copy_from2; |
| 120 | |
Yan, Zheng | 81f148a | 2019-07-25 20:16:46 +0800 | [diff] [blame] | 121 | u32 filp_gen; |
Chengguang Xu | 719784b | 2018-07-19 22:15:24 +0800 | [diff] [blame] | 122 | loff_t max_file_size; |
Sage Weil | 85ccce4 | 2010-02-17 10:02:43 -0800 | [diff] [blame] | 123 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 124 | struct ceph_mds_client *mdsc; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 125 | |
Yan, Zheng | e3ec8d6 | 2019-01-14 17:21:19 +0800 | [diff] [blame] | 126 | atomic_long_t writeback_count; |
| 127 | |
Yan, Zheng | 1cf89a8 | 2019-05-18 11:18:44 +0800 | [diff] [blame] | 128 | struct workqueue_struct *inode_wq; |
Yan, Zheng | e3ec8d6 | 2019-01-14 17:21:19 +0800 | [diff] [blame] | 129 | struct workqueue_struct *cap_wq; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 130 | |
Sage Weil | 0743304 | 2009-11-18 16:50:41 -0800 | [diff] [blame] | 131 | #ifdef CONFIG_DEBUG_FS |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 132 | struct dentry *debugfs_dentry_lru, *debugfs_caps; |
Yehuda Sadeh | 2baba25 | 2009-12-18 13:51:57 -0800 | [diff] [blame] | 133 | struct dentry *debugfs_congestion_kb; |
Sage Weil | 06edf04 | 2009-12-15 14:44:32 -0800 | [diff] [blame] | 134 | struct dentry *debugfs_bdi; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 135 | struct dentry *debugfs_mdsc, *debugfs_mdsmap; |
Xiubo Li | 247b1f1 | 2020-11-11 09:29:39 +0800 | [diff] [blame] | 136 | struct dentry *debugfs_status; |
John Spray | 14ed970 | 2014-09-12 16:58:49 +0100 | [diff] [blame] | 137 | struct dentry *debugfs_mds_sessions; |
LuÃs Henriques | cbed4ff | 2021-10-27 11:01:30 +0100 | [diff] [blame] | 138 | struct dentry *debugfs_metrics_dir; |
Sage Weil | 0743304 | 2009-11-18 16:50:41 -0800 | [diff] [blame] | 139 | #endif |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 140 | |
| 141 | #ifdef CONFIG_CEPH_FSCACHE |
| 142 | struct fscache_cookie *fscache; |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 143 | #endif |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 144 | }; |
| 145 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 146 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 147 | /* |
| 148 | * File i/o capability. This tracks shared state with the metadata |
| 149 | * server that allows us to cache or writeback attributes or to read |
| 150 | * and write data. For any given inode, we should have one or more |
| 151 | * capabilities, one issued by each metadata server, and our |
| 152 | * cumulative access is the OR of all issued capabilities. |
| 153 | * |
| 154 | * Each cap is referenced by the inode's i_caps rbtree and by per-mds |
| 155 | * session capability lists. |
| 156 | */ |
| 157 | struct ceph_cap { |
| 158 | struct ceph_inode_info *ci; |
| 159 | struct rb_node ci_node; /* per-ci cap tree */ |
| 160 | struct ceph_mds_session *session; |
| 161 | struct list_head session_caps; /* per-session caplist */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 162 | u64 cap_id; /* unique cap id (mds provided) */ |
Yan, Zheng | 745a8e3 | 2015-05-14 17:22:42 +0800 | [diff] [blame] | 163 | union { |
| 164 | /* in-use caps */ |
| 165 | struct { |
| 166 | int issued; /* latest, from the mds */ |
| 167 | int implemented; /* implemented superset of |
| 168 | issued (for revocation) */ |
Jeff Layton | c74d79a | 2020-10-06 12:24:19 -0400 | [diff] [blame] | 169 | int mds; /* mds index for this cap */ |
| 170 | int mds_wanted; /* caps wanted from this mds */ |
Yan, Zheng | 745a8e3 | 2015-05-14 17:22:42 +0800 | [diff] [blame] | 171 | }; |
| 172 | /* caps to release */ |
| 173 | struct { |
| 174 | u64 cap_ino; |
| 175 | int queue_release; |
| 176 | }; |
| 177 | }; |
Sage Weil | 685f9a5d | 2009-11-09 12:05:48 -0800 | [diff] [blame] | 178 | u32 seq, issue_seq, mseq; |
| 179 | u32 cap_gen; /* active/stale cycle */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 180 | unsigned long last_used; |
| 181 | struct list_head caps_item; |
| 182 | }; |
| 183 | |
Yan, Zheng | a0d93e3 | 2020-03-05 20:21:01 +0800 | [diff] [blame] | 184 | #define CHECK_CAPS_AUTHONLY 1 /* only check auth cap */ |
| 185 | #define CHECK_CAPS_FLUSH 2 /* flush any dirty caps */ |
| 186 | #define CHECK_CAPS_NOINVAL 4 /* don't invalidate pagecache */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 187 | |
Yan, Zheng | 0e29438 | 2016-07-04 18:06:41 +0800 | [diff] [blame] | 188 | struct ceph_cap_flush { |
| 189 | u64 tid; |
Xiubo Li | b2f9fa1 | 2021-08-18 21:38:42 +0800 | [diff] [blame] | 190 | int caps; |
Yan, Zheng | c8799fc | 2016-07-07 15:22:38 +0800 | [diff] [blame] | 191 | bool wake; /* wake up flush waiters when finish ? */ |
Xiubo Li | b2f9fa1 | 2021-08-18 21:38:42 +0800 | [diff] [blame] | 192 | bool is_capsnap; /* true means capsnap */ |
Yan, Zheng | 0e29438 | 2016-07-04 18:06:41 +0800 | [diff] [blame] | 193 | struct list_head g_list; // global |
| 194 | struct list_head i_list; // per inode |
| 195 | }; |
| 196 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 197 | /* |
| 198 | * Snapped cap state that is pending flush to mds. When a snapshot occurs, |
| 199 | * we first complete any in-process sync writes and writeback any dirty |
| 200 | * data before flushing the snapped state (tracked here) back to the MDS. |
| 201 | */ |
| 202 | struct ceph_cap_snap { |
Elena Reshetova | 805692d | 2017-03-03 11:15:07 +0200 | [diff] [blame] | 203 | refcount_t nref; |
Yan, Zheng | 0e29438 | 2016-07-04 18:06:41 +0800 | [diff] [blame] | 204 | struct list_head ci_item; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 205 | |
Yan, Zheng | 0e29438 | 2016-07-04 18:06:41 +0800 | [diff] [blame] | 206 | struct ceph_cap_flush cap_flush; |
| 207 | |
| 208 | u64 follows; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 209 | int issued, dirty; |
| 210 | struct ceph_snap_context *context; |
| 211 | |
Al Viro | 5706b27 | 2011-07-26 04:52:22 -0400 | [diff] [blame] | 212 | umode_t mode; |
Eric W. Biederman | 05cb11c | 2013-01-31 02:56:19 -0800 | [diff] [blame] | 213 | kuid_t uid; |
| 214 | kgid_t gid; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 215 | |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 216 | struct ceph_buffer *xattr_blob; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 217 | u64 xattr_version; |
| 218 | |
| 219 | u64 size; |
Jeff Layton | 176c77c | 2019-06-06 08:06:40 -0400 | [diff] [blame] | 220 | u64 change_attr; |
Jeff Layton | ec62b89 | 2019-05-29 12:23:14 -0400 | [diff] [blame] | 221 | struct timespec64 mtime, atime, ctime, btime; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 222 | u64 time_warp_seq; |
Yan, Zheng | 5f743e4 | 2016-11-15 16:04:37 +0800 | [diff] [blame] | 223 | u64 truncate_size; |
| 224 | u32 truncate_seq; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 225 | int writing; /* a sync write is still in progress */ |
| 226 | int dirty_pages; /* dirty pages awaiting writeback */ |
Yan, Zheng | e20d258 | 2014-11-14 22:39:13 +0800 | [diff] [blame] | 227 | bool inline_data; |
Yan, Zheng | 8605609 | 2015-05-01 16:57:16 +0800 | [diff] [blame] | 228 | bool need_flush; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 229 | }; |
| 230 | |
| 231 | static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap) |
| 232 | { |
Elena Reshetova | 805692d | 2017-03-03 11:15:07 +0200 | [diff] [blame] | 233 | if (refcount_dec_and_test(&capsnap->nref)) { |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 234 | if (capsnap->xattr_blob) |
| 235 | ceph_buffer_put(capsnap->xattr_blob); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 236 | kfree(capsnap); |
Sage Weil | 4a625be | 2010-08-22 15:03:56 -0700 | [diff] [blame] | 237 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 238 | } |
| 239 | |
| 240 | /* |
| 241 | * The frag tree describes how a directory is fragmented, potentially across |
| 242 | * multiple metadata servers. It is also used to indicate points where |
| 243 | * metadata authority is delegated, and whether/where metadata is replicated. |
| 244 | * |
| 245 | * A _leaf_ frag will be present in the i_fragtree IFF there is |
| 246 | * delegation info. That is, if mds >= 0 || ndist > 0. |
| 247 | */ |
| 248 | #define CEPH_MAX_DIRFRAG_REP 4 |
| 249 | |
| 250 | struct ceph_inode_frag { |
| 251 | struct rb_node node; |
| 252 | |
| 253 | /* fragtree state */ |
| 254 | u32 frag; |
| 255 | int split_by; /* i.e. 2^(split_by) children */ |
| 256 | |
| 257 | /* delegation and replication info */ |
| 258 | int mds; /* -1 if same authority as parent */ |
| 259 | int ndist; /* >0 if replicated */ |
| 260 | int dist[CEPH_MAX_DIRFRAG_REP]; |
| 261 | }; |
| 262 | |
| 263 | /* |
| 264 | * We cache inode xattrs as an encoded blob until they are first used, |
| 265 | * at which point we parse them into an rbtree. |
| 266 | */ |
| 267 | struct ceph_inode_xattr { |
| 268 | struct rb_node node; |
| 269 | |
| 270 | const char *name; |
| 271 | int name_len; |
| 272 | const char *val; |
| 273 | int val_len; |
| 274 | int dirty; |
| 275 | |
| 276 | int should_free_name; |
| 277 | int should_free_val; |
| 278 | }; |
| 279 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 280 | /* |
| 281 | * Ceph dentry state |
| 282 | */ |
| 283 | struct ceph_dentry_info { |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 284 | struct dentry *dentry; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 285 | struct ceph_mds_session *lease_session; |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 286 | struct list_head lease_list; |
| 287 | unsigned flags; |
Yan, Zheng | 97aeb6b | 2017-11-27 10:47:46 +0800 | [diff] [blame] | 288 | int lease_shared_gen; |
| 289 | u32 lease_gen; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 290 | u32 lease_seq; |
| 291 | unsigned long lease_renew_after, lease_renew_from; |
Miklos Szeredi | 9b16f03c | 2016-06-22 16:35:04 +0200 | [diff] [blame] | 292 | unsigned long time; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 293 | u64 offset; |
| 294 | }; |
| 295 | |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 296 | #define CEPH_DENTRY_REFERENCED 1 |
| 297 | #define CEPH_DENTRY_LEASE_LIST 2 |
| 298 | #define CEPH_DENTRY_SHRINK_LIST 4 |
Jeff Layton | f5e17ae | 2020-02-18 14:12:32 -0500 | [diff] [blame] | 299 | #define CEPH_DENTRY_PRIMARY_LINK 8 |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 300 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 301 | struct ceph_inode_xattrs_info { |
| 302 | /* |
| 303 | * (still encoded) xattr blob. we avoid the overhead of parsing |
| 304 | * this until someone actually calls getxattr, etc. |
| 305 | * |
| 306 | * blob->vec.iov_len == 4 implies there are no xattrs; blob == |
| 307 | * NULL means we don't know. |
| 308 | */ |
| 309 | struct ceph_buffer *blob, *prealloc_blob; |
| 310 | |
| 311 | struct rb_root index; |
| 312 | bool dirty; |
| 313 | int count; |
| 314 | int names_size; |
| 315 | int vals_size; |
| 316 | u64 version, index_version; |
| 317 | }; |
| 318 | |
| 319 | /* |
| 320 | * Ceph inode. |
| 321 | */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 322 | struct ceph_inode_info { |
| 323 | struct ceph_vino i_vino; /* ceph ino + snap */ |
| 324 | |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 325 | spinlock_t i_ceph_lock; |
| 326 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 327 | u64 i_version; |
Yan, Zheng | 31c542a | 2014-11-14 21:41:55 +0800 | [diff] [blame] | 328 | u64 i_inline_version; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 329 | u32 i_time_warp_seq; |
| 330 | |
Jeff Layton | 891f3f5 | 2020-01-14 15:06:40 -0500 | [diff] [blame] | 331 | unsigned long i_ceph_flags; |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 332 | atomic64_t i_release_count; |
| 333 | atomic64_t i_ordered_count; |
| 334 | atomic64_t i_complete_seq[2]; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 335 | |
Sage Weil | 6c0f3af | 2010-11-16 11:14:34 -0800 | [diff] [blame] | 336 | struct ceph_dir_layout i_dir_layout; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 337 | struct ceph_file_layout i_layout; |
Jeff Layton | 785892f | 2020-01-02 07:11:38 -0500 | [diff] [blame] | 338 | struct ceph_file_layout i_cached_layout; // for async creates |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 339 | char *i_symlink; |
| 340 | |
| 341 | /* for dirs */ |
Arnd Bergmann | 9bbeab4 | 2018-07-13 22:18:36 +0200 | [diff] [blame] | 342 | struct timespec64 i_rctime; |
Yanhu Cao | e7f7295 | 2020-08-28 09:28:44 +0800 | [diff] [blame] | 343 | u64 i_rbytes, i_rfiles, i_rsubdirs, i_rsnaps; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 344 | u64 i_files, i_subdirs; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 345 | |
Luis Henriques | fb18a57 | 2018-01-05 10:47:18 +0000 | [diff] [blame] | 346 | /* quotas */ |
| 347 | u64 i_max_bytes, i_max_files; |
| 348 | |
Yan, Zheng | 0879687 | 2019-01-09 11:07:02 +0800 | [diff] [blame] | 349 | s32 i_dir_pin; |
| 350 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 351 | struct rb_root i_fragtree; |
Yan, Zheng | 1b1bc16 | 2016-05-04 11:40:30 +0800 | [diff] [blame] | 352 | int i_fragtree_nsplits; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 353 | struct mutex i_fragtree_mutex; |
| 354 | |
| 355 | struct ceph_inode_xattrs_info i_xattrs; |
| 356 | |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 357 | /* capabilities. protected _both_ by i_ceph_lock and cap->session's |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 358 | * s_mutex. */ |
| 359 | struct rb_root i_caps; /* cap list */ |
| 360 | struct ceph_cap *i_auth_cap; /* authoritative cap, if any */ |
| 361 | unsigned i_dirty_caps, i_flushing_caps; /* mask of dirtied fields */ |
Jeff Layton | 1cf03a6 | 2020-04-01 17:07:52 -0400 | [diff] [blame] | 362 | |
| 363 | /* |
Randy Dunlap | f1f565a | 2020-07-17 16:36:04 -0700 | [diff] [blame] | 364 | * Link to the auth cap's session's s_cap_dirty list. s_cap_dirty |
Jeff Layton | 1cf03a6 | 2020-04-01 17:07:52 -0400 | [diff] [blame] | 365 | * is protected by the mdsc->cap_dirty_lock, but each individual item |
| 366 | * is also protected by the inode's i_ceph_lock. Walking s_cap_dirty |
| 367 | * requires the mdsc->cap_dirty_lock. List presence for an item can |
| 368 | * be tested under the i_ceph_lock. Changing anything requires both. |
| 369 | */ |
| 370 | struct list_head i_dirty_item; |
| 371 | |
| 372 | /* |
Jeff Layton | 829ad4d | 2020-04-03 13:09:07 -0400 | [diff] [blame] | 373 | * Link to session's s_cap_flushing list. Protected in a similar |
| 374 | * fashion to i_dirty_item, but also by the s_mutex for changes. The |
| 375 | * s_cap_flushing list can be walked while holding either the s_mutex |
| 376 | * or msdc->cap_dirty_lock. List presence can also be checked while |
| 377 | * holding the i_ceph_lock for this inode. |
Jeff Layton | 1cf03a6 | 2020-04-01 17:07:52 -0400 | [diff] [blame] | 378 | */ |
| 379 | struct list_head i_flushing_item; |
| 380 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 381 | /* we need to track cap writeback on a per-cap-bit basis, to allow |
| 382 | * overlapping, pipelined cap flushes to the mds. we can probably |
| 383 | * reduce the tid to 8 bits if we're concerned about inode size. */ |
Yan, Zheng | f66fd9f | 2015-06-10 17:26:13 +0800 | [diff] [blame] | 384 | struct ceph_cap_flush *i_prealloc_cap_flush; |
Yan, Zheng | e4500b5 | 2016-07-06 11:12:56 +0800 | [diff] [blame] | 385 | struct list_head i_cap_flush_list; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 386 | wait_queue_head_t i_cap_wq; /* threads waiting on a capability */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 387 | unsigned long i_hold_caps_max; /* jiffies */ |
| 388 | struct list_head i_cap_delay_list; /* for delayed cap release to mds */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 389 | struct ceph_cap_reservation i_cap_migration_resv; |
| 390 | struct list_head i_cap_snaps; /* snapped state pending flush to mds */ |
Sage Weil | 7d8cb26 | 2010-08-24 08:44:16 -0700 | [diff] [blame] | 391 | struct ceph_snap_context *i_head_snapc; /* set if wr_buffer_head > 0 or |
| 392 | dirty|flushing caps */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 393 | unsigned i_snap_caps; /* cap bits for snapped files */ |
| 394 | |
Yan, Zheng | 719a251 | 2020-03-05 20:21:00 +0800 | [diff] [blame] | 395 | unsigned long i_last_rd; |
| 396 | unsigned long i_last_wr; |
Yan, Zheng | 774a6a1 | 2016-06-06 16:01:39 +0800 | [diff] [blame] | 397 | int i_nr_by_mode[CEPH_FILE_MODE_BITS]; /* open file counts */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 398 | |
Yan, Zheng | b0d7c22 | 2013-08-12 21:42:15 -0700 | [diff] [blame] | 399 | struct mutex i_truncate_mutex; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 400 | u32 i_truncate_seq; /* last truncate to smaller size */ |
| 401 | u64 i_truncate_size; /* and the size we last truncated down to */ |
| 402 | int i_truncate_pending; /* still need to call vmtruncate */ |
| 403 | |
| 404 | u64 i_max_size; /* max file size authorized by mds */ |
| 405 | u64 i_reported_size; /* (max_)size reported to or requested of mds */ |
| 406 | u64 i_wanted_max_size; /* offset we'd like to write too */ |
| 407 | u64 i_requested_max_size; /* max_size we've requested */ |
| 408 | |
| 409 | /* held references to caps */ |
| 410 | int i_pin_ref; |
Jeff Layton | f85122a | 2019-04-02 08:04:30 -0400 | [diff] [blame] | 411 | int i_rd_ref, i_rdcache_ref, i_wr_ref, i_wb_ref, i_fx_ref; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 412 | int i_wrbuffer_ref, i_wrbuffer_ref_head; |
Yan, Zheng | 89aa593 | 2017-09-08 15:23:18 +0800 | [diff] [blame] | 413 | atomic_t i_filelock_ref; |
Yan, Zheng | 97aeb6b | 2017-11-27 10:47:46 +0800 | [diff] [blame] | 414 | atomic_t i_shared_gen; /* increment each time we get FILE_SHARED */ |
Sage Weil | cd045cb | 2010-11-04 11:05:05 -0700 | [diff] [blame] | 415 | u32 i_rdcache_gen; /* incremented each time we get FILE_CACHE. */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 416 | u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */ |
| 417 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 418 | struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */ |
Yan, Zheng | 68cd5b4 | 2015-10-27 18:36:06 +0800 | [diff] [blame] | 419 | struct list_head i_unsafe_iops; /* uncommitted mds inode ops */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 420 | spinlock_t i_unsafe_lock; |
| 421 | |
Yan, Zheng | 75c9627 | 2017-12-14 15:11:09 +0800 | [diff] [blame] | 422 | union { |
| 423 | struct ceph_snap_realm *i_snap_realm; /* snap realm (if caps) */ |
| 424 | struct ceph_snapid_map *i_snapid_map; /* snapid -> dev_t */ |
| 425 | }; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 426 | struct list_head i_snap_realm_item; |
| 427 | struct list_head i_snap_flush_item; |
Jeff Layton | 245ce99 | 2019-05-29 11:19:42 -0400 | [diff] [blame] | 428 | struct timespec64 i_btime; |
David Disseldorp | 193e7b3 | 2019-04-18 14:15:46 +0200 | [diff] [blame] | 429 | struct timespec64 i_snap_btime; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 430 | |
Yan, Zheng | 1cf89a8 | 2019-05-18 11:18:44 +0800 | [diff] [blame] | 431 | struct work_struct i_work; |
| 432 | unsigned long i_work_mask; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 433 | |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 434 | #ifdef CONFIG_CEPH_FSCACHE |
| 435 | struct fscache_cookie *fscache; |
Milosz Tanski | 99ccbd2 | 2013-08-21 17:29:54 -0400 | [diff] [blame] | 436 | #endif |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 437 | struct inode vfs_inode; /* at end */ |
| 438 | }; |
| 439 | |
Jeff Layton | 721d5c1 | 2019-08-01 14:11:15 -0400 | [diff] [blame] | 440 | static inline struct ceph_inode_info * |
| 441 | ceph_inode(const struct inode *inode) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 442 | { |
Noah Watkins | fbbccec | 2009-10-28 11:54:49 -0700 | [diff] [blame] | 443 | return container_of(inode, struct ceph_inode_info, vfs_inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 444 | } |
| 445 | |
Jeff Layton | 721d5c1 | 2019-08-01 14:11:15 -0400 | [diff] [blame] | 446 | static inline struct ceph_fs_client * |
| 447 | ceph_inode_to_client(const struct inode *inode) |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 448 | { |
| 449 | return (struct ceph_fs_client *)inode->i_sb->s_fs_info; |
| 450 | } |
| 451 | |
Jeff Layton | 721d5c1 | 2019-08-01 14:11:15 -0400 | [diff] [blame] | 452 | static inline struct ceph_fs_client * |
| 453 | ceph_sb_to_client(const struct super_block *sb) |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 454 | { |
| 455 | return (struct ceph_fs_client *)sb->s_fs_info; |
| 456 | } |
| 457 | |
Xiubo Li | 2678da8 | 2020-09-03 09:01:39 -0400 | [diff] [blame] | 458 | static inline struct ceph_mds_client * |
| 459 | ceph_sb_to_mdsc(const struct super_block *sb) |
| 460 | { |
| 461 | return (struct ceph_mds_client *)ceph_sb_to_client(sb)->mdsc; |
| 462 | } |
| 463 | |
Jeff Layton | 721d5c1 | 2019-08-01 14:11:15 -0400 | [diff] [blame] | 464 | static inline struct ceph_vino |
| 465 | ceph_vino(const struct inode *inode) |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 466 | { |
| 467 | return ceph_inode(inode)->i_vino; |
| 468 | } |
| 469 | |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 470 | static inline u32 ceph_ino_to_ino32(u64 vino) |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 471 | { |
Amon Ott | 3310f75 | 2011-10-20 13:04:07 -0700 | [diff] [blame] | 472 | u32 ino = vino & 0xffffffff; |
| 473 | ino ^= vino >> 32; |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 474 | if (!ino) |
Amon Ott | a661fc5 | 2012-01-23 09:25:23 -0800 | [diff] [blame] | 475 | ino = 2; |
Yehuda Sadeh | ad1fee9 | 2011-01-21 16:44:03 -0800 | [diff] [blame] | 476 | return ino; |
| 477 | } |
| 478 | |
| 479 | /* |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 480 | * Inode numbers in cephfs are 64 bits, but inode->i_ino is 32-bits on |
| 481 | * some arches. We generally do not use this value inside the ceph driver, but |
| 482 | * we do want to set it to something, so that generic vfs code has an |
| 483 | * appropriate value for tracepoints and the like. |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 484 | */ |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 485 | static inline ino_t ceph_vino_to_ino_t(struct ceph_vino vino) |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 486 | { |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 487 | if (sizeof(ino_t) == sizeof(u32)) |
| 488 | return ceph_ino_to_ino32(vino.ino); |
Amon Ott | 3310f75 | 2011-10-20 13:04:07 -0700 | [diff] [blame] | 489 | return (ino_t)vino.ino; |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 490 | } |
| 491 | |
| 492 | /* for printf-style formatting */ |
| 493 | #define ceph_vinop(i) ceph_inode(i)->i_vino.ino, ceph_inode(i)->i_vino.snap |
| 494 | |
| 495 | static inline u64 ceph_ino(struct inode *inode) |
| 496 | { |
| 497 | return ceph_inode(inode)->i_vino.ino; |
| 498 | } |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 499 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 500 | static inline u64 ceph_snap(struct inode *inode) |
| 501 | { |
| 502 | return ceph_inode(inode)->i_vino.snap; |
| 503 | } |
| 504 | |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 505 | /** |
| 506 | * ceph_present_ino - format an inode number for presentation to userland |
| 507 | * @sb: superblock where the inode lives |
| 508 | * @ino: inode number to (possibly) convert |
| 509 | * |
| 510 | * If the user mounted with the ino32 option, then the 64-bit value needs |
| 511 | * to be converted to something that can fit inside 32 bits. Note that |
| 512 | * internal kernel code never uses this value, so this is entirely for |
| 513 | * userland consumption. |
| 514 | */ |
| 515 | static inline u64 ceph_present_ino(struct super_block *sb, u64 ino) |
| 516 | { |
| 517 | if (unlikely(ceph_test_mount_opt(ceph_sb_to_client(sb), INO32))) |
| 518 | return ceph_ino_to_ino32(ino); |
| 519 | return ino; |
| 520 | } |
| 521 | |
| 522 | static inline u64 ceph_present_inode(struct inode *inode) |
| 523 | { |
| 524 | return ceph_present_ino(inode->i_sb, ceph_ino(inode)); |
| 525 | } |
| 526 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 527 | static inline int ceph_ino_compare(struct inode *inode, void *data) |
| 528 | { |
| 529 | struct ceph_vino *pvino = (struct ceph_vino *)data; |
| 530 | struct ceph_inode_info *ci = ceph_inode(inode); |
| 531 | return ci->i_vino.ino == pvino->ino && |
| 532 | ci->i_vino.snap == pvino->snap; |
| 533 | } |
| 534 | |
Jeff Layton | d4f6b31 | 2021-04-01 13:55:11 -0400 | [diff] [blame] | 535 | /* |
| 536 | * The MDS reserves a set of inodes for its own usage. These should never |
| 537 | * be accessible by clients, and so the MDS has no reason to ever hand these |
| 538 | * out. The range is CEPH_MDS_INO_MDSDIR_OFFSET..CEPH_INO_SYSTEM_BASE. |
| 539 | * |
| 540 | * These come from src/mds/mdstypes.h in the ceph sources. |
| 541 | */ |
Jeff Layton | 0078ea3 | 2021-11-09 09:54:49 -0500 | [diff] [blame^] | 542 | #define CEPH_MAX_MDS 0x100 |
| 543 | #define CEPH_NUM_STRAY 10 |
Jeff Layton | d4f6b31 | 2021-04-01 13:55:11 -0400 | [diff] [blame] | 544 | #define CEPH_MDS_INO_MDSDIR_OFFSET (1 * CEPH_MAX_MDS) |
Jeff Layton | 0078ea3 | 2021-11-09 09:54:49 -0500 | [diff] [blame^] | 545 | #define CEPH_MDS_INO_LOG_OFFSET (2 * CEPH_MAX_MDS) |
Jeff Layton | d4f6b31 | 2021-04-01 13:55:11 -0400 | [diff] [blame] | 546 | #define CEPH_INO_SYSTEM_BASE ((6*CEPH_MAX_MDS) + (CEPH_MAX_MDS * CEPH_NUM_STRAY)) |
| 547 | |
| 548 | static inline bool ceph_vino_is_reserved(const struct ceph_vino vino) |
| 549 | { |
Jeff Layton | 0078ea3 | 2021-11-09 09:54:49 -0500 | [diff] [blame^] | 550 | if (vino.ino >= CEPH_INO_SYSTEM_BASE || |
| 551 | vino.ino < CEPH_MDS_INO_MDSDIR_OFFSET) |
| 552 | return false; |
| 553 | |
| 554 | /* Don't warn on mdsdirs */ |
| 555 | WARN_RATELIMIT(vino.ino >= CEPH_MDS_INO_LOG_OFFSET, |
| 556 | "Attempt to access reserved inode number 0x%llx", |
| 557 | vino.ino); |
| 558 | return true; |
Jeff Layton | d4f6b31 | 2021-04-01 13:55:11 -0400 | [diff] [blame] | 559 | } |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 560 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 561 | static inline struct inode *ceph_find_inode(struct super_block *sb, |
| 562 | struct ceph_vino vino) |
| 563 | { |
Jeff Layton | d4f6b31 | 2021-04-01 13:55:11 -0400 | [diff] [blame] | 564 | if (ceph_vino_is_reserved(vino)) |
| 565 | return NULL; |
| 566 | |
Jeff Layton | ebce3eb | 2020-08-18 08:03:48 -0400 | [diff] [blame] | 567 | /* |
| 568 | * NB: The hashval will be run through the fs/inode.c hash function |
| 569 | * anyway, so there is no need to squash the inode number down to |
| 570 | * 32-bits first. Just use low-order bits on arches with 32-bit long. |
| 571 | */ |
| 572 | return ilookup5(sb, (unsigned long)vino.ino, ceph_ino_compare, &vino); |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 573 | } |
| 574 | |
| 575 | |
| 576 | /* |
| 577 | * Ceph inode. |
| 578 | */ |
Yan, Zheng | 10183a6 | 2015-04-27 15:33:28 +0800 | [diff] [blame] | 579 | #define CEPH_I_DIR_ORDERED (1 << 0) /* dentries in dir are ordered */ |
Yan, Zheng | 10183a6 | 2015-04-27 15:33:28 +0800 | [diff] [blame] | 580 | #define CEPH_I_FLUSH (1 << 2) /* do not delay flush of dirty metadata */ |
Jeff Layton | daca8bd | 2019-07-05 10:55:38 -0400 | [diff] [blame] | 581 | #define CEPH_I_POOL_PERM (1 << 3) /* pool rd/wr bits are valid */ |
| 582 | #define CEPH_I_POOL_RD (1 << 4) /* can read from pool */ |
| 583 | #define CEPH_I_POOL_WR (1 << 5) /* can write to pool */ |
| 584 | #define CEPH_I_SEC_INITED (1 << 6) /* security initialized */ |
Yan, Zheng | c0e385b | 2020-03-05 20:20:59 +0800 | [diff] [blame] | 585 | #define CEPH_I_KICK_FLUSH (1 << 7) /* kick flushing caps */ |
| 586 | #define CEPH_I_FLUSH_SNAPS (1 << 8) /* need flush snapss */ |
| 587 | #define CEPH_I_ERROR_WRITE (1 << 9) /* have seen write errors */ |
| 588 | #define CEPH_I_ERROR_FILELOCK (1 << 10) /* have seen file lock errors */ |
| 589 | #define CEPH_I_ODIRECT (1 << 11) /* inode in direct I/O mode */ |
| 590 | #define CEPH_ASYNC_CREATE_BIT (12) /* async create in flight for this */ |
Jeff Layton | 891f3f5 | 2020-01-14 15:06:40 -0500 | [diff] [blame] | 591 | #define CEPH_I_ASYNC_CREATE (1 << CEPH_ASYNC_CREATE_BIT) |
Jeff Layton | 5d6451b | 2021-08-31 13:39:13 -0400 | [diff] [blame] | 592 | #define CEPH_I_SHUTDOWN (1 << 13) /* inode is no longer usable */ |
Jeff Layton | 26544c62 | 2017-04-04 08:39:46 -0400 | [diff] [blame] | 593 | |
| 594 | /* |
Yan, Zheng | 1cf89a8 | 2019-05-18 11:18:44 +0800 | [diff] [blame] | 595 | * Masks of ceph inode work. |
| 596 | */ |
Jeff Layton | a8810cd | 2020-12-10 14:39:26 -0500 | [diff] [blame] | 597 | #define CEPH_I_WORK_WRITEBACK 0 |
| 598 | #define CEPH_I_WORK_INVALIDATE_PAGES 1 |
| 599 | #define CEPH_I_WORK_VMTRUNCATE 2 |
| 600 | #define CEPH_I_WORK_CHECK_CAPS 3 |
| 601 | #define CEPH_I_WORK_FLUSH_SNAPS 4 |
Yan, Zheng | 1cf89a8 | 2019-05-18 11:18:44 +0800 | [diff] [blame] | 602 | |
| 603 | /* |
Jeff Layton | 26544c62 | 2017-04-04 08:39:46 -0400 | [diff] [blame] | 604 | * We set the ERROR_WRITE bit when we start seeing write errors on an inode |
| 605 | * and then clear it when they start succeeding. Note that we do a lockless |
| 606 | * check first, and only take the lock if it looks like it needs to be changed. |
| 607 | * The write submission code just takes this as a hint, so we're not too |
| 608 | * worried if a few slip through in either direction. |
| 609 | */ |
| 610 | static inline void ceph_set_error_write(struct ceph_inode_info *ci) |
| 611 | { |
| 612 | if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE)) { |
| 613 | spin_lock(&ci->i_ceph_lock); |
| 614 | ci->i_ceph_flags |= CEPH_I_ERROR_WRITE; |
| 615 | spin_unlock(&ci->i_ceph_lock); |
| 616 | } |
| 617 | } |
| 618 | |
| 619 | static inline void ceph_clear_error_write(struct ceph_inode_info *ci) |
| 620 | { |
| 621 | if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ERROR_WRITE) { |
| 622 | spin_lock(&ci->i_ceph_lock); |
| 623 | ci->i_ceph_flags &= ~CEPH_I_ERROR_WRITE; |
| 624 | spin_unlock(&ci->i_ceph_lock); |
| 625 | } |
| 626 | } |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 627 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 628 | static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci, |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 629 | long long release_count, |
| 630 | long long ordered_count) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 631 | { |
Andrea Parri | 7496077 | 2019-05-20 19:23:58 +0200 | [diff] [blame] | 632 | /* |
| 633 | * Makes sure operations that setup readdir cache (update page |
| 634 | * cache and i_size) are strongly ordered w.r.t. the following |
| 635 | * atomic64_set() operations. |
| 636 | */ |
| 637 | smp_mb(); |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 638 | atomic64_set(&ci->i_complete_seq[0], release_count); |
| 639 | atomic64_set(&ci->i_complete_seq[1], ordered_count); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 640 | } |
| 641 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 642 | static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 643 | { |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 644 | atomic64_inc(&ci->i_release_count); |
| 645 | } |
| 646 | |
| 647 | static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci) |
| 648 | { |
| 649 | atomic64_inc(&ci->i_ordered_count); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 650 | } |
| 651 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 652 | static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 653 | { |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 654 | return atomic64_read(&ci->i_complete_seq[0]) == |
| 655 | atomic64_read(&ci->i_release_count); |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 656 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 657 | |
Yan, Zheng | 70db4f3 | 2014-10-21 18:09:56 -0700 | [diff] [blame] | 658 | static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci) |
| 659 | { |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 660 | return atomic64_read(&ci->i_complete_seq[0]) == |
| 661 | atomic64_read(&ci->i_release_count) && |
| 662 | atomic64_read(&ci->i_complete_seq[1]) == |
| 663 | atomic64_read(&ci->i_ordered_count); |
Yan, Zheng | 70db4f3 | 2014-10-21 18:09:56 -0700 | [diff] [blame] | 664 | } |
| 665 | |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 666 | static inline void ceph_dir_clear_complete(struct inode *inode) |
| 667 | { |
| 668 | __ceph_dir_clear_complete(ceph_inode(inode)); |
| 669 | } |
| 670 | |
Yan, Zheng | 70db4f3 | 2014-10-21 18:09:56 -0700 | [diff] [blame] | 671 | static inline void ceph_dir_clear_ordered(struct inode *inode) |
Yan, Zheng | 2f276c5 | 2013-03-13 19:44:32 +0800 | [diff] [blame] | 672 | { |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 673 | __ceph_dir_clear_ordered(ceph_inode(inode)); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 674 | } |
| 675 | |
Yan, Zheng | 70db4f3 | 2014-10-21 18:09:56 -0700 | [diff] [blame] | 676 | static inline bool ceph_dir_is_complete_ordered(struct inode *inode) |
| 677 | { |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 678 | bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode)); |
| 679 | smp_rmb(); |
Yan, Zheng | 70db4f3 | 2014-10-21 18:09:56 -0700 | [diff] [blame] | 680 | return ret; |
| 681 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 682 | |
| 683 | /* find a specific frag @f */ |
| 684 | extern struct ceph_inode_frag *__ceph_find_frag(struct ceph_inode_info *ci, |
| 685 | u32 f); |
| 686 | |
| 687 | /* |
| 688 | * choose fragment for value @v. copy frag content to pfrag, if leaf |
| 689 | * exists |
| 690 | */ |
| 691 | extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v, |
| 692 | struct ceph_inode_frag *pfrag, |
| 693 | int *found); |
| 694 | |
Yan, Zheng | 1e9c2eb | 2019-01-28 20:43:55 +0800 | [diff] [blame] | 695 | static inline struct ceph_dentry_info *ceph_dentry(const struct dentry *dentry) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 696 | { |
| 697 | return (struct ceph_dentry_info *)dentry->d_fsdata; |
| 698 | } |
| 699 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 700 | /* |
| 701 | * caps helpers |
| 702 | */ |
| 703 | static inline bool __ceph_is_any_real_caps(struct ceph_inode_info *ci) |
| 704 | { |
| 705 | return !RB_EMPTY_ROOT(&ci->i_caps); |
| 706 | } |
| 707 | |
| 708 | extern int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented); |
| 709 | extern int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int t); |
Xiubo Li | 1af16d5 | 2020-03-19 23:45:00 -0400 | [diff] [blame] | 710 | extern int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, |
| 711 | int t); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 712 | extern int __ceph_caps_issued_other(struct ceph_inode_info *ci, |
| 713 | struct ceph_cap *cap); |
| 714 | |
| 715 | static inline int ceph_caps_issued(struct ceph_inode_info *ci) |
| 716 | { |
| 717 | int issued; |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 718 | spin_lock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 719 | issued = __ceph_caps_issued(ci, NULL); |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 720 | spin_unlock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 721 | return issued; |
| 722 | } |
| 723 | |
Xiubo Li | 1af16d5 | 2020-03-19 23:45:00 -0400 | [diff] [blame] | 724 | static inline int ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, |
| 725 | int mask, int touch) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 726 | { |
| 727 | int r; |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 728 | spin_lock(&ci->i_ceph_lock); |
Xiubo Li | 1af16d5 | 2020-03-19 23:45:00 -0400 | [diff] [blame] | 729 | r = __ceph_caps_issued_mask_metric(ci, mask, touch); |
Sage Weil | be65559 | 2011-11-30 09:47:09 -0800 | [diff] [blame] | 730 | spin_unlock(&ci->i_ceph_lock); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 731 | return r; |
| 732 | } |
| 733 | |
| 734 | static inline int __ceph_caps_dirty(struct ceph_inode_info *ci) |
| 735 | { |
| 736 | return ci->i_dirty_caps | ci->i_flushing_caps; |
| 737 | } |
Yan, Zheng | f66fd9f | 2015-06-10 17:26:13 +0800 | [diff] [blame] | 738 | extern struct ceph_cap_flush *ceph_alloc_cap_flush(void); |
| 739 | extern void ceph_free_cap_flush(struct ceph_cap_flush *cf); |
| 740 | extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, |
| 741 | struct ceph_cap_flush **pcf); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 742 | |
Yan, Zheng | 9563f88 | 2013-11-22 13:50:45 +0800 | [diff] [blame] | 743 | extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci, |
| 744 | struct ceph_cap *ocap, int mask); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 745 | extern int ceph_caps_revoking(struct ceph_inode_info *ci, int mask); |
| 746 | extern int __ceph_caps_used(struct ceph_inode_info *ci); |
| 747 | |
Yan, Zheng | 719a251 | 2020-03-05 20:21:00 +0800 | [diff] [blame] | 748 | static inline bool __ceph_is_file_opened(struct ceph_inode_info *ci) |
| 749 | { |
| 750 | return ci->i_nr_by_mode[0]; |
| 751 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 752 | extern int __ceph_caps_file_wanted(struct ceph_inode_info *ci); |
Yan, Zheng | 525d15e | 2019-05-11 17:27:59 +0800 | [diff] [blame] | 753 | extern int __ceph_caps_wanted(struct ceph_inode_info *ci); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 754 | |
| 755 | /* what the mds thinks we want */ |
Yan, Zheng | c1944fe | 2017-01-29 22:15:47 +0800 | [diff] [blame] | 756 | extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 757 | |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 758 | extern void ceph_caps_init(struct ceph_mds_client *mdsc); |
| 759 | extern void ceph_caps_finalize(struct ceph_mds_client *mdsc); |
Yan, Zheng | fe33032 | 2019-02-01 14:57:15 +0800 | [diff] [blame] | 760 | extern void ceph_adjust_caps_max_min(struct ceph_mds_client *mdsc, |
| 761 | struct ceph_mount_options *fsopt); |
Zhi Zhang | e30ee58 | 2018-01-24 21:24:33 +0800 | [diff] [blame] | 762 | extern int ceph_reserve_caps(struct ceph_mds_client *mdsc, |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 763 | struct ceph_cap_reservation *ctx, int need); |
Chengguang Xu | 7bf8f73 | 2018-07-28 23:15:35 +0800 | [diff] [blame] | 764 | extern void ceph_unreserve_caps(struct ceph_mds_client *mdsc, |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 765 | struct ceph_cap_reservation *ctx); |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 766 | extern void ceph_reservation_status(struct ceph_fs_client *client, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 767 | int *total, int *avail, int *used, |
Sage Weil | 85ccce4 | 2010-02-17 10:02:43 -0800 | [diff] [blame] | 768 | int *reserved, int *min); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 769 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 770 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 771 | |
| 772 | /* |
| 773 | * we keep buffered readdir results attached to file->private_data |
| 774 | */ |
Sage Weil | 4918b6d | 2011-07-26 11:26:07 -0700 | [diff] [blame] | 775 | #define CEPH_F_SYNC 1 |
Sage Weil | 9cfa109 | 2011-07-26 11:26:18 -0700 | [diff] [blame] | 776 | #define CEPH_F_ATEND 2 |
Sage Weil | 4918b6d | 2011-07-26 11:26:07 -0700 | [diff] [blame] | 777 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 778 | struct ceph_file_info { |
Sage Weil | 252c672 | 2011-07-26 11:25:27 -0700 | [diff] [blame] | 779 | short fmode; /* initialized on open */ |
| 780 | short flags; /* CEPH_F_* */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 781 | |
Yan, Zheng | 5d98830 | 2017-12-15 11:15:36 +0800 | [diff] [blame] | 782 | spinlock_t rw_contexts_lock; |
| 783 | struct list_head rw_contexts; |
Yan, Zheng | f4b9786 | 2019-07-25 20:16:42 +0800 | [diff] [blame] | 784 | |
Yan, Zheng | 81f148a | 2019-07-25 20:16:46 +0800 | [diff] [blame] | 785 | u32 filp_gen; |
Yan, Zheng | ff5d913 | 2019-07-25 20:16:45 +0800 | [diff] [blame] | 786 | atomic_t num_locks; |
Chengguang Xu | bb48bd4 | 2018-03-13 10:42:44 +0800 | [diff] [blame] | 787 | }; |
| 788 | |
| 789 | struct ceph_dir_file_info { |
| 790 | struct ceph_file_info file_info; |
Yan, Zheng | 5d98830 | 2017-12-15 11:15:36 +0800 | [diff] [blame] | 791 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 792 | /* readdir: position within the dir */ |
| 793 | u32 frag; |
| 794 | struct ceph_mds_request *last_readdir; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 795 | |
| 796 | /* readdir: position within a frag */ |
Yan, Zheng | f049420 | 2014-02-27 16:26:24 +0800 | [diff] [blame] | 797 | unsigned next_offset; /* offset of next chunk (last_name's + 1) */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 798 | char *last_name; /* last entry in previous chunk */ |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 799 | long long dir_release_count; |
| 800 | long long dir_ordered_count; |
| 801 | int readdir_cache_idx; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 802 | |
| 803 | /* used for -o dirstat read() on directory thing */ |
| 804 | char *dir_info; |
| 805 | int dir_info_len; |
| 806 | }; |
| 807 | |
Yan, Zheng | 5d98830 | 2017-12-15 11:15:36 +0800 | [diff] [blame] | 808 | struct ceph_rw_context { |
| 809 | struct list_head list; |
| 810 | struct task_struct *thread; |
| 811 | int caps; |
| 812 | }; |
| 813 | |
| 814 | #define CEPH_DEFINE_RW_CONTEXT(_name, _caps) \ |
| 815 | struct ceph_rw_context _name = { \ |
| 816 | .thread = current, \ |
| 817 | .caps = _caps, \ |
| 818 | } |
| 819 | |
| 820 | static inline void ceph_add_rw_context(struct ceph_file_info *cf, |
| 821 | struct ceph_rw_context *ctx) |
| 822 | { |
| 823 | spin_lock(&cf->rw_contexts_lock); |
| 824 | list_add(&ctx->list, &cf->rw_contexts); |
| 825 | spin_unlock(&cf->rw_contexts_lock); |
| 826 | } |
| 827 | |
| 828 | static inline void ceph_del_rw_context(struct ceph_file_info *cf, |
| 829 | struct ceph_rw_context *ctx) |
| 830 | { |
| 831 | spin_lock(&cf->rw_contexts_lock); |
| 832 | list_del(&ctx->list); |
| 833 | spin_unlock(&cf->rw_contexts_lock); |
| 834 | } |
| 835 | |
| 836 | static inline struct ceph_rw_context* |
| 837 | ceph_find_rw_context(struct ceph_file_info *cf) |
| 838 | { |
| 839 | struct ceph_rw_context *ctx, *found = NULL; |
| 840 | spin_lock(&cf->rw_contexts_lock); |
| 841 | list_for_each_entry(ctx, &cf->rw_contexts, list) { |
| 842 | if (ctx->thread == current) { |
| 843 | found = ctx; |
| 844 | break; |
| 845 | } |
| 846 | } |
| 847 | spin_unlock(&cf->rw_contexts_lock); |
| 848 | return found; |
| 849 | } |
| 850 | |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 851 | struct ceph_readdir_cache_control { |
| 852 | struct page *page; |
| 853 | struct dentry **dentries; |
| 854 | int index; |
| 855 | }; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 856 | |
| 857 | /* |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 858 | * A "snap realm" describes a subset of the file hierarchy sharing |
| 859 | * the same set of snapshots that apply to it. The realms themselves |
| 860 | * are organized into a hierarchy, such that children inherit (some of) |
| 861 | * the snapshots of their parents. |
| 862 | * |
| 863 | * All inodes within the realm that have capabilities are linked into a |
| 864 | * per-realm list. |
| 865 | */ |
| 866 | struct ceph_snap_realm { |
| 867 | u64 ino; |
Luis Henriques | e3161f1 | 2018-01-12 17:19:28 +0000 | [diff] [blame] | 868 | struct inode *inode; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 869 | atomic_t nref; |
Sage Weil | a105f00 | 2010-02-15 14:37:55 -0800 | [diff] [blame] | 870 | struct rb_node node; |
| 871 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 872 | u64 created, seq; |
| 873 | u64 parent_ino; |
| 874 | u64 parent_since; /* snapid when our current parent became so */ |
| 875 | |
| 876 | u64 *prior_parent_snaps; /* snaps inherited from any parents we */ |
Alex Elder | aa711ee3 | 2012-07-13 20:35:11 -0500 | [diff] [blame] | 877 | u32 num_prior_parent_snaps; /* had prior to parent_since */ |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 878 | u64 *snaps; /* snaps specific to this realm */ |
Alex Elder | aa711ee3 | 2012-07-13 20:35:11 -0500 | [diff] [blame] | 879 | u32 num_snaps; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 880 | |
| 881 | struct ceph_snap_realm *parent; |
| 882 | struct list_head children; /* list of child realms */ |
| 883 | struct list_head child_item; |
| 884 | |
| 885 | struct list_head empty_item; /* if i have ref==0 */ |
| 886 | |
Sage Weil | ae00d4f | 2010-09-16 16:26:51 -0700 | [diff] [blame] | 887 | struct list_head dirty_item; /* if realm needs new context */ |
| 888 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 889 | /* the current set of snaps for this realm */ |
| 890 | struct ceph_snap_context *cached_context; |
| 891 | |
| 892 | struct list_head inodes_with_caps; |
| 893 | spinlock_t inodes_with_caps_lock; |
| 894 | }; |
| 895 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 896 | static inline int default_congestion_kb(void) |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 897 | { |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 898 | int congestion_kb; |
| 899 | |
| 900 | /* |
| 901 | * Copied from NFS |
| 902 | * |
| 903 | * congestion size, scale with available memory. |
| 904 | * |
| 905 | * 64MB: 8192k |
| 906 | * 128MB: 11585k |
| 907 | * 256MB: 16384k |
| 908 | * 512MB: 23170k |
| 909 | * 1GB: 32768k |
| 910 | * 2GB: 46340k |
| 911 | * 4GB: 65536k |
| 912 | * 8GB: 92681k |
| 913 | * 16GB: 131072k |
| 914 | * |
| 915 | * This allows larger machines to have larger/more transfers. |
| 916 | * Limit the default to 256M |
| 917 | */ |
Arun KS | ca79b0c | 2018-12-28 00:34:29 -0800 | [diff] [blame] | 918 | congestion_kb = (16*int_sqrt(totalram_pages())) << (PAGE_SHIFT-10); |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 919 | if (congestion_kb > 256*1024) |
| 920 | congestion_kb = 256*1024; |
| 921 | |
| 922 | return congestion_kb; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 923 | } |
| 924 | |
| 925 | |
Yan, Zheng | d468e72 | 2019-07-25 20:16:44 +0800 | [diff] [blame] | 926 | /* super.c */ |
| 927 | extern int ceph_force_reconnect(struct super_block *sb); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 928 | /* snap.c */ |
| 929 | struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc, |
| 930 | u64 ino); |
| 931 | extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc, |
| 932 | struct ceph_snap_realm *realm); |
| 933 | extern void ceph_put_snap_realm(struct ceph_mds_client *mdsc, |
| 934 | struct ceph_snap_realm *realm); |
| 935 | extern int ceph_update_snap_trace(struct ceph_mds_client *m, |
Yan, Zheng | 982d601 | 2014-12-23 15:30:54 +0800 | [diff] [blame] | 936 | void *p, void *e, bool deletion, |
| 937 | struct ceph_snap_realm **realm_ret); |
Jeff Layton | 0ba92e1 | 2021-08-02 11:01:26 -0400 | [diff] [blame] | 938 | void ceph_change_snap_realm(struct inode *inode, struct ceph_snap_realm *realm); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 939 | extern void ceph_handle_snap(struct ceph_mds_client *mdsc, |
Sage Weil | 2600d2d | 2010-02-22 15:12:16 -0800 | [diff] [blame] | 940 | struct ceph_mds_session *session, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 941 | struct ceph_msg *msg); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 942 | extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci, |
| 943 | struct ceph_cap_snap *capsnap); |
| 944 | extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc); |
| 945 | |
Yan, Zheng | 75c9627 | 2017-12-14 15:11:09 +0800 | [diff] [blame] | 946 | extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc, |
| 947 | u64 snap); |
| 948 | extern void ceph_put_snapid_map(struct ceph_mds_client* mdsc, |
| 949 | struct ceph_snapid_map *sm); |
| 950 | extern void ceph_trim_snapid_map(struct ceph_mds_client *mdsc); |
| 951 | extern void ceph_cleanup_snapid_map(struct ceph_mds_client *mdsc); |
Jeff Layton | 631ed4b | 2021-10-14 11:10:47 -0400 | [diff] [blame] | 952 | void ceph_umount_begin(struct super_block *sb); |
Yan, Zheng | 75c9627 | 2017-12-14 15:11:09 +0800 | [diff] [blame] | 953 | |
| 954 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 955 | /* |
| 956 | * a cap_snap is "pending" if it is still awaiting an in-progress |
| 957 | * sync write (that may/may not still update size, mtime, etc.). |
| 958 | */ |
| 959 | static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci) |
| 960 | { |
| 961 | return !list_empty(&ci->i_cap_snaps) && |
Yan, Zheng | 8605609 | 2015-05-01 16:57:16 +0800 | [diff] [blame] | 962 | list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap, |
| 963 | ci_item)->writing; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 964 | } |
| 965 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 966 | /* inode.c */ |
Jeff Layton | 966c716 | 2019-12-05 09:09:25 -0500 | [diff] [blame] | 967 | struct ceph_mds_reply_info_in; |
| 968 | struct ceph_mds_reply_dirfrag; |
| 969 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 970 | extern const struct inode_operations ceph_file_iops; |
| 971 | |
| 972 | extern struct inode *ceph_alloc_inode(struct super_block *sb); |
Yan, Zheng | 87bc5b8 | 2019-06-02 09:45:38 +0800 | [diff] [blame] | 973 | extern void ceph_evict_inode(struct inode *inode); |
Al Viro | cfa6d41 | 2019-04-10 15:18:50 -0400 | [diff] [blame] | 974 | extern void ceph_free_inode(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 975 | |
| 976 | extern struct inode *ceph_get_inode(struct super_block *sb, |
| 977 | struct ceph_vino vino); |
| 978 | extern struct inode *ceph_get_snapdir(struct inode *parent); |
| 979 | extern int ceph_fill_file_size(struct inode *inode, int issued, |
| 980 | u32 truncate_seq, u64 truncate_size, u64 size); |
| 981 | extern void ceph_fill_file_time(struct inode *inode, int issued, |
Arnd Bergmann | 9bbeab4 | 2018-07-13 22:18:36 +0200 | [diff] [blame] | 982 | u64 time_warp_seq, struct timespec64 *ctime, |
| 983 | struct timespec64 *mtime, |
| 984 | struct timespec64 *atime); |
Jeff Layton | 966c716 | 2019-12-05 09:09:25 -0500 | [diff] [blame] | 985 | extern int ceph_fill_inode(struct inode *inode, struct page *locked_page, |
| 986 | struct ceph_mds_reply_info_in *iinfo, |
| 987 | struct ceph_mds_reply_dirfrag *dirinfo, |
| 988 | struct ceph_mds_session *session, int cap_fmode, |
| 989 | struct ceph_cap_reservation *caps_reservation); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 990 | extern int ceph_fill_trace(struct super_block *sb, |
Jeff Layton | f5a03b0 | 2017-01-31 11:06:13 -0500 | [diff] [blame] | 991 | struct ceph_mds_request *req); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 992 | extern int ceph_readdir_prepopulate(struct ceph_mds_request *req, |
| 993 | struct ceph_mds_session *session); |
| 994 | |
| 995 | extern int ceph_inode_holds_cap(struct inode *inode, int mask); |
| 996 | |
Yan, Zheng | efb0ca7 | 2017-05-22 12:03:32 +0800 | [diff] [blame] | 997 | extern bool ceph_inode_set_size(struct inode *inode, loff_t size); |
Yan, Zheng | b415bf4 | 2013-07-02 12:40:19 +0800 | [diff] [blame] | 998 | extern void __ceph_do_pending_vmtruncate(struct inode *inode); |
Jeff Layton | 64f28c6 | 2020-10-09 14:24:34 -0400 | [diff] [blame] | 999 | |
Jeff Layton | 64f28c6 | 2020-10-09 14:24:34 -0400 | [diff] [blame] | 1000 | void ceph_queue_inode_work(struct inode *inode, int work_bit); |
| 1001 | |
| 1002 | static inline void ceph_queue_vmtruncate(struct inode *inode) |
| 1003 | { |
| 1004 | ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE); |
| 1005 | } |
| 1006 | |
| 1007 | static inline void ceph_queue_invalidate(struct inode *inode) |
| 1008 | { |
| 1009 | ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES); |
| 1010 | } |
| 1011 | |
| 1012 | static inline void ceph_queue_writeback(struct inode *inode) |
| 1013 | { |
| 1014 | ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK); |
| 1015 | } |
| 1016 | |
Jeff Layton | a8810cd | 2020-12-10 14:39:26 -0500 | [diff] [blame] | 1017 | static inline void ceph_queue_check_caps(struct inode *inode) |
| 1018 | { |
| 1019 | ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS); |
| 1020 | } |
| 1021 | |
| 1022 | static inline void ceph_queue_flush_snaps(struct inode *inode) |
| 1023 | { |
| 1024 | ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS); |
| 1025 | } |
| 1026 | |
Yan, Zheng | 01deead | 2014-11-14 21:56:29 +0800 | [diff] [blame] | 1027 | extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page, |
| 1028 | int mask, bool force); |
| 1029 | static inline int ceph_do_getattr(struct inode *inode, int mask, bool force) |
| 1030 | { |
| 1031 | return __ceph_do_getattr(inode, NULL, mask, force); |
| 1032 | } |
Christian Brauner | 549c729 | 2021-01-21 14:19:43 +0100 | [diff] [blame] | 1033 | extern int ceph_permission(struct user_namespace *mnt_userns, |
| 1034 | struct inode *inode, int mask); |
Andreas Gruenbacher | a26fecc | 2016-04-14 00:30:16 +0200 | [diff] [blame] | 1035 | extern int __ceph_setattr(struct inode *inode, struct iattr *attr); |
Christian Brauner | 549c729 | 2021-01-21 14:19:43 +0100 | [diff] [blame] | 1036 | extern int ceph_setattr(struct user_namespace *mnt_userns, |
| 1037 | struct dentry *dentry, struct iattr *attr); |
| 1038 | extern int ceph_getattr(struct user_namespace *mnt_userns, |
| 1039 | const struct path *path, struct kstat *stat, |
David Howells | a528d35 | 2017-01-31 16:46:22 +0000 | [diff] [blame] | 1040 | u32 request_mask, unsigned int flags); |
Jeff Layton | 5d6451b | 2021-08-31 13:39:13 -0400 | [diff] [blame] | 1041 | void ceph_inode_shutdown(struct inode *inode); |
| 1042 | |
| 1043 | static inline bool ceph_inode_is_shutdown(struct inode *inode) |
| 1044 | { |
| 1045 | unsigned long flags = READ_ONCE(ceph_inode(inode)->i_ceph_flags); |
| 1046 | struct ceph_fs_client *fsc = ceph_inode_to_client(inode); |
| 1047 | int state = READ_ONCE(fsc->mount_state); |
| 1048 | |
| 1049 | return (flags & CEPH_I_SHUTDOWN) || state >= CEPH_MOUNT_SHUTDOWN; |
| 1050 | } |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1051 | |
| 1052 | /* xattr.c */ |
Andreas Gruenbacher | a26fecc | 2016-04-14 00:30:16 +0200 | [diff] [blame] | 1053 | int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int); |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1054 | ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1055 | extern ssize_t ceph_listxattr(struct dentry *, char *, size_t); |
Luis Henriques | 12fe3dd | 2019-07-19 15:32:21 +0100 | [diff] [blame] | 1056 | extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1057 | extern void __ceph_destroy_xattrs(struct ceph_inode_info *ci); |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 1058 | extern const struct xattr_handler *ceph_xattr_handlers[]; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1059 | |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1060 | struct ceph_acl_sec_ctx { |
| 1061 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
| 1062 | void *default_acl; |
| 1063 | void *acl; |
| 1064 | #endif |
Yan, Zheng | ac6713c | 2019-05-26 16:27:56 +0800 | [diff] [blame] | 1065 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
| 1066 | void *sec_ctx; |
| 1067 | u32 sec_ctxlen; |
| 1068 | #endif |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1069 | struct ceph_pagelist *pagelist; |
| 1070 | }; |
| 1071 | |
Yan, Zheng | 315f240 | 2016-03-07 10:34:50 +0800 | [diff] [blame] | 1072 | #ifdef CONFIG_SECURITY |
| 1073 | extern bool ceph_security_xattr_deadlock(struct inode *in); |
| 1074 | extern bool ceph_security_xattr_wanted(struct inode *in); |
| 1075 | #else |
| 1076 | static inline bool ceph_security_xattr_deadlock(struct inode *in) |
| 1077 | { |
| 1078 | return false; |
| 1079 | } |
| 1080 | static inline bool ceph_security_xattr_wanted(struct inode *in) |
| 1081 | { |
| 1082 | return false; |
| 1083 | } |
| 1084 | #endif |
| 1085 | |
Yan, Zheng | ac6713c | 2019-05-26 16:27:56 +0800 | [diff] [blame] | 1086 | #ifdef CONFIG_CEPH_FS_SECURITY_LABEL |
| 1087 | extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, |
| 1088 | struct ceph_acl_sec_ctx *ctx); |
Jeff Layton | 668959a | 2019-08-06 09:07:51 -0400 | [diff] [blame] | 1089 | static inline void ceph_security_invalidate_secctx(struct inode *inode) |
| 1090 | { |
| 1091 | security_inode_invalidate_secctx(inode); |
| 1092 | } |
Yan, Zheng | ac6713c | 2019-05-26 16:27:56 +0800 | [diff] [blame] | 1093 | #else |
| 1094 | static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode, |
| 1095 | struct ceph_acl_sec_ctx *ctx) |
| 1096 | { |
| 1097 | return 0; |
| 1098 | } |
| 1099 | static inline void ceph_security_invalidate_secctx(struct inode *inode) |
| 1100 | { |
| 1101 | } |
| 1102 | #endif |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1103 | |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1104 | void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx); |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1105 | |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1106 | /* acl.c */ |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1107 | #ifdef CONFIG_CEPH_FS_POSIX_ACL |
| 1108 | |
Miklos Szeredi | 0cad624 | 2021-08-18 22:08:24 +0200 | [diff] [blame] | 1109 | struct posix_acl *ceph_get_acl(struct inode *, int, bool); |
Christian Brauner | 549c729 | 2021-01-21 14:19:43 +0100 | [diff] [blame] | 1110 | int ceph_set_acl(struct user_namespace *mnt_userns, |
| 1111 | struct inode *inode, struct posix_acl *acl, int type); |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 1112 | int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1113 | struct ceph_acl_sec_ctx *as_ctx); |
| 1114 | void ceph_init_inode_acls(struct inode *inode, |
| 1115 | struct ceph_acl_sec_ctx *as_ctx); |
Guangliang Zhao | c969d9b | 2014-02-16 08:35:52 -0800 | [diff] [blame] | 1116 | |
| 1117 | static inline void ceph_forget_all_cached_acls(struct inode *inode) |
| 1118 | { |
| 1119 | forget_all_cached_acls(inode); |
| 1120 | } |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1121 | |
| 1122 | #else |
| 1123 | |
| 1124 | #define ceph_get_acl NULL |
Sage Weil | 72466d0 | 2014-01-29 06:22:25 -0800 | [diff] [blame] | 1125 | #define ceph_set_acl NULL |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1126 | |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 1127 | static inline int ceph_pre_init_acls(struct inode *dir, umode_t *mode, |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1128 | struct ceph_acl_sec_ctx *as_ctx) |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1129 | { |
| 1130 | return 0; |
| 1131 | } |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 1132 | static inline void ceph_init_inode_acls(struct inode *inode, |
Yan, Zheng | 5c31e92 | 2019-05-26 15:35:39 +0800 | [diff] [blame] | 1133 | struct ceph_acl_sec_ctx *as_ctx) |
Yan, Zheng | b1ee94a | 2014-09-16 20:35:17 +0800 | [diff] [blame] | 1134 | { |
| 1135 | } |
Guangliang Zhao | 7221fe4 | 2013-11-11 15:18:03 +0800 | [diff] [blame] | 1136 | static inline int ceph_acl_chmod(struct dentry *dentry, struct inode *inode) |
| 1137 | { |
| 1138 | return 0; |
| 1139 | } |
| 1140 | |
| 1141 | static inline void ceph_forget_all_cached_acls(struct inode *inode) |
| 1142 | { |
| 1143 | } |
| 1144 | |
| 1145 | #endif |
| 1146 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1147 | /* caps.c */ |
| 1148 | extern const char *ceph_cap_string(int c); |
| 1149 | extern void ceph_handle_caps(struct ceph_mds_session *session, |
| 1150 | struct ceph_msg *msg); |
Yan, Zheng | d9df278 | 2014-04-18 09:57:11 +0800 | [diff] [blame] | 1151 | extern struct ceph_cap *ceph_get_cap(struct ceph_mds_client *mdsc, |
| 1152 | struct ceph_cap_reservation *ctx); |
| 1153 | extern void ceph_add_cap(struct inode *inode, |
| 1154 | struct ceph_mds_session *session, u64 cap_id, |
Yan, Zheng | 135e671 | 2020-03-05 20:21:02 +0800 | [diff] [blame] | 1155 | unsigned issued, unsigned wanted, |
Yan, Zheng | d9df278 | 2014-04-18 09:57:11 +0800 | [diff] [blame] | 1156 | unsigned cap, unsigned seq, u64 realmino, int flags, |
| 1157 | struct ceph_cap **new_cap); |
Yan, Zheng | a096b09 | 2013-09-22 10:15:58 +0800 | [diff] [blame] | 1158 | extern void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
Xiubo Li | a76d0a9 | 2021-08-25 21:45:45 +0800 | [diff] [blame] | 1159 | extern void ceph_remove_cap(struct ceph_cap *cap, bool queue_release); |
Yan, Zheng | d6e4781 | 2019-05-23 11:01:37 +0800 | [diff] [blame] | 1160 | extern void __ceph_remove_caps(struct ceph_inode_info *ci); |
Yehuda Sadeh | 3715166 | 2010-06-17 16:16:12 -0700 | [diff] [blame] | 1161 | extern void ceph_put_cap(struct ceph_mds_client *mdsc, |
| 1162 | struct ceph_cap *cap); |
Yan, Zheng | 9215aee | 2013-11-30 12:47:41 +0800 | [diff] [blame] | 1163 | extern int ceph_is_any_caps(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1164 | |
Stephen Rothwell | f1a3d57 | 2010-01-18 11:53:08 +1100 | [diff] [blame] | 1165 | extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc); |
Josef Bacik | 02c24a8 | 2011-07-16 20:44:56 -0400 | [diff] [blame] | 1166 | extern int ceph_fsync(struct file *file, loff_t start, loff_t end, |
| 1167 | int datasync); |
Yan, Zheng | e548e9b | 2015-06-10 15:17:56 +0800 | [diff] [blame] | 1168 | extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, |
| 1169 | struct ceph_mds_session *session); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1170 | extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, |
| 1171 | struct ceph_mds_session *session); |
Jeff Layton | e8a4d26 | 2020-02-25 11:49:53 -0800 | [diff] [blame] | 1172 | void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, |
| 1173 | struct ceph_inode_info *ci); |
Greg Farnum | 2bc5025 | 2010-06-30 12:44:34 -0700 | [diff] [blame] | 1174 | extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, |
| 1175 | int mds); |
Jeff Layton | 40dcf75 | 2020-01-14 09:23:49 -0500 | [diff] [blame] | 1176 | extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, |
| 1177 | bool snap_rwsem_locked); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1178 | extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps); |
| 1179 | extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had); |
Jeff Layton | a8810cd | 2020-12-10 14:39:26 -0500 | [diff] [blame] | 1180 | extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had); |
Xiubo Li | e64f44a | 2020-05-27 09:09:27 -0400 | [diff] [blame] | 1181 | extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, |
| 1182 | int had); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1183 | extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, |
| 1184 | struct ceph_snap_context *snapc); |
Xiubo Li | a6d37cc | 2021-08-25 21:45:43 +0800 | [diff] [blame] | 1185 | extern void __ceph_remove_capsnap(struct inode *inode, |
| 1186 | struct ceph_cap_snap *capsnap, |
| 1187 | bool *wake_ci, bool *wake_mdsc); |
| 1188 | extern void ceph_remove_capsnap(struct inode *inode, |
| 1189 | struct ceph_cap_snap *capsnap, |
| 1190 | bool *wake_ci, bool *wake_mdsc); |
Yan, Zheng | ed9b430 | 2016-07-05 21:08:07 +0800 | [diff] [blame] | 1191 | extern void ceph_flush_snaps(struct ceph_inode_info *ci, |
| 1192 | struct ceph_mds_session **psession); |
Yan, Zheng | efb0ca7 | 2017-05-22 12:03:32 +0800 | [diff] [blame] | 1193 | extern bool __ceph_should_report_size(struct ceph_inode_info *ci); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1194 | extern void ceph_check_caps(struct ceph_inode_info *ci, int flags, |
| 1195 | struct ceph_mds_session *session); |
Luis Henriques | bf2ba43 | 2021-07-06 14:52:41 +0100 | [diff] [blame] | 1196 | extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc); |
Sage Weil | afcdaea | 2009-10-14 14:27:38 -0700 | [diff] [blame] | 1197 | extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc); |
Zhi Zhang | 6ef0bc6 | 2018-01-24 21:24:33 +0800 | [diff] [blame] | 1198 | extern int ceph_drop_caps_for_unlink(struct inode *inode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1199 | extern int ceph_encode_inode_release(void **p, struct inode *inode, |
| 1200 | int mds, int drop, int unless, int force); |
| 1201 | extern int ceph_encode_dentry_release(void **p, struct dentry *dn, |
Jeff Layton | ca6c8ae | 2016-12-15 08:37:59 -0500 | [diff] [blame] | 1202 | struct inode *dir, |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1203 | int mds, int drop, int unless); |
| 1204 | |
Yan, Zheng | 5e3ded1 | 2019-07-25 20:16:43 +0800 | [diff] [blame] | 1205 | extern int ceph_get_caps(struct file *filp, int need, int want, |
Jeff Layton | e72968e | 2021-04-05 12:19:35 -0400 | [diff] [blame] | 1206 | loff_t endoff, int *got); |
Yan, Zheng | 5e3ded1 | 2019-07-25 20:16:43 +0800 | [diff] [blame] | 1207 | extern int ceph_try_get_caps(struct inode *inode, |
Luis Henriques | 2ee9dd9 | 2018-10-15 16:45:57 +0100 | [diff] [blame] | 1208 | int need, int want, bool nonblock, int *got); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1209 | |
| 1210 | /* for counting open files by mode */ |
Yan, Zheng | 719a251 | 2020-03-05 20:21:00 +0800 | [diff] [blame] | 1211 | extern void ceph_get_fmode(struct ceph_inode_info *ci, int mode, int count); |
| 1212 | extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode, int count); |
| 1213 | extern void __ceph_touch_fmode(struct ceph_inode_info *ci, |
| 1214 | struct ceph_mds_client *mdsc, int fmode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1215 | |
| 1216 | /* addr.c */ |
| 1217 | extern const struct address_space_operations ceph_aops; |
| 1218 | extern int ceph_mmap(struct file *file, struct vm_area_struct *vma); |
Yan, Zheng | 10183a6 | 2015-04-27 15:33:28 +0800 | [diff] [blame] | 1219 | extern int ceph_uninline_data(struct file *filp, struct page *locked_page); |
Yan, Zheng | 5e3ded1 | 2019-07-25 20:16:43 +0800 | [diff] [blame] | 1220 | extern int ceph_pool_perm_check(struct inode *inode, int need); |
Yan, Zheng | 10183a6 | 2015-04-27 15:33:28 +0800 | [diff] [blame] | 1221 | extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc); |
Jeff Layton | 36e6da9 | 2021-09-02 13:06:57 -0400 | [diff] [blame] | 1222 | int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1223 | |
| 1224 | /* file.c */ |
| 1225 | extern const struct file_operations ceph_file_fops; |
Alex Elder | 9e0eb85 | 2013-02-06 13:11:38 -0600 | [diff] [blame] | 1226 | |
Yan, Zheng | 719a251 | 2020-03-05 20:21:00 +0800 | [diff] [blame] | 1227 | extern int ceph_renew_caps(struct inode *inode, int fmode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1228 | extern int ceph_open(struct inode *inode, struct file *file); |
Sage Weil | 5ef50c3 | 2012-07-31 11:27:36 -0700 | [diff] [blame] | 1229 | extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, |
Al Viro | 44907d7 | 2018-06-08 13:32:02 -0400 | [diff] [blame] | 1230 | struct file *file, unsigned flags, umode_t mode); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1231 | extern int ceph_release(struct inode *inode, struct file *filp); |
Yan, Zheng | 31c542a | 2014-11-14 21:41:55 +0800 | [diff] [blame] | 1232 | extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page, |
| 1233 | char *data, size_t len); |
Ilya Dryomov | 55f2a04 | 2017-02-13 14:44:19 +0100 | [diff] [blame] | 1234 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1235 | /* dir.c */ |
| 1236 | extern const struct file_operations ceph_dir_fops; |
Yan, Zheng | 38c48b5 | 2015-01-14 13:46:04 +0800 | [diff] [blame] | 1237 | extern const struct file_operations ceph_snapdir_fops; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1238 | extern const struct inode_operations ceph_dir_iops; |
Yan, Zheng | 38c48b5 | 2015-01-14 13:46:04 +0800 | [diff] [blame] | 1239 | extern const struct inode_operations ceph_snapdir_iops; |
Al Viro | 18fc8ab | 2016-10-28 21:52:50 -0400 | [diff] [blame] | 1240 | extern const struct dentry_operations ceph_dentry_ops; |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1241 | |
Yan, Zheng | f3c4ebe | 2016-04-29 11:27:30 +0800 | [diff] [blame] | 1242 | extern loff_t ceph_make_fpos(unsigned high, unsigned off, bool hash_order); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1243 | extern int ceph_handle_notrace_create(struct inode *dir, struct dentry *dentry); |
Jeff Layton | aa60cfc | 2021-03-01 08:01:54 -0500 | [diff] [blame] | 1244 | extern struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req, |
Jeff Layton | 7a971e2 | 2021-06-02 12:46:07 -0400 | [diff] [blame] | 1245 | struct dentry *dentry); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1246 | extern struct dentry *ceph_finish_lookup(struct ceph_mds_request *req, |
| 1247 | struct dentry *dentry, int err); |
| 1248 | |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 1249 | extern void __ceph_dentry_lease_touch(struct ceph_dentry_info *di); |
| 1250 | extern void __ceph_dentry_dir_lease_touch(struct ceph_dentry_info *di); |
Sage Weil | 81a6cf2 | 2010-05-14 09:35:38 -0700 | [diff] [blame] | 1251 | extern void ceph_invalidate_dentry_lease(struct dentry *dentry); |
Yan, Zheng | 37c4efc | 2019-01-31 16:55:51 +0800 | [diff] [blame] | 1252 | extern int ceph_trim_dentries(struct ceph_mds_client *mdsc); |
Sage Weil | e5f86dc | 2011-07-26 11:30:55 -0700 | [diff] [blame] | 1253 | extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn); |
Yan, Zheng | fdd4e15 | 2015-06-16 20:48:56 +0800 | [diff] [blame] | 1254 | extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1255 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1256 | /* ioctl.c */ |
| 1257 | extern long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
| 1258 | |
| 1259 | /* export.c */ |
| 1260 | extern const struct export_operations ceph_export_ops; |
Luis Henriques | 3886274 | 2019-03-21 10:20:09 +0000 | [diff] [blame] | 1261 | struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino); |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1262 | |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 1263 | /* locks.c */ |
Yan, Zheng | eb13e83 | 2014-03-09 23:16:40 +0800 | [diff] [blame] | 1264 | extern __init void ceph_flock_init(void); |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 1265 | extern int ceph_lock(struct file *file, int cmd, struct file_lock *fl); |
| 1266 | extern int ceph_flock(struct file *file, int cmd, struct file_lock *fl); |
| 1267 | extern void ceph_count_locks(struct inode *inode, int *p_num, int *f_num); |
Jim Schutt | 39be95e | 2013-05-15 13:03:35 -0500 | [diff] [blame] | 1268 | extern int ceph_encode_locks_to_buffer(struct inode *inode, |
| 1269 | struct ceph_filelock *flocks, |
| 1270 | int num_fcntl_locks, |
| 1271 | int num_flock_locks); |
| 1272 | extern int ceph_locks_to_pagelist(struct ceph_filelock *flocks, |
| 1273 | struct ceph_pagelist *pagelist, |
| 1274 | int num_fcntl_locks, int num_flock_locks); |
Greg Farnum | 40819f6 | 2010-08-02 15:34:23 -0700 | [diff] [blame] | 1275 | |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 1276 | /* debugfs.c */ |
Greg Kroah-Hartman | 1a829ff | 2019-06-12 16:55:38 +0200 | [diff] [blame] | 1277 | extern void ceph_fs_debugfs_init(struct ceph_fs_client *client); |
Yehuda Sadeh | 3d14c5d | 2010-04-06 15:14:15 -0700 | [diff] [blame] | 1278 | extern void ceph_fs_debugfs_cleanup(struct ceph_fs_client *client); |
| 1279 | |
Luis Henriques | fb18a57 | 2018-01-05 10:47:18 +0000 | [diff] [blame] | 1280 | /* quota.c */ |
Luis Henriques | d557c48 | 2018-01-12 17:19:29 +0000 | [diff] [blame] | 1281 | static inline bool __ceph_has_any_quota(struct ceph_inode_info *ci) |
| 1282 | { |
| 1283 | return ci->i_max_files || ci->i_max_bytes; |
| 1284 | } |
| 1285 | |
| 1286 | extern void ceph_adjust_quota_realms_count(struct inode *inode, bool inc); |
| 1287 | |
| 1288 | static inline void __ceph_update_quota(struct ceph_inode_info *ci, |
| 1289 | u64 max_bytes, u64 max_files) |
| 1290 | { |
| 1291 | bool had_quota, has_quota; |
| 1292 | had_quota = __ceph_has_any_quota(ci); |
| 1293 | ci->i_max_bytes = max_bytes; |
| 1294 | ci->i_max_files = max_files; |
| 1295 | has_quota = __ceph_has_any_quota(ci); |
| 1296 | |
| 1297 | if (had_quota != has_quota) |
| 1298 | ceph_adjust_quota_realms_count(&ci->vfs_inode, has_quota); |
| 1299 | } |
| 1300 | |
Luis Henriques | fb18a57 | 2018-01-05 10:47:18 +0000 | [diff] [blame] | 1301 | extern void ceph_handle_quota(struct ceph_mds_client *mdsc, |
| 1302 | struct ceph_mds_session *session, |
| 1303 | struct ceph_msg *msg); |
Luis Henriques | b7a2921 | 2018-01-05 10:47:19 +0000 | [diff] [blame] | 1304 | extern bool ceph_quota_is_max_files_exceeded(struct inode *inode); |
Luis Henriques | 6646ea1 | 2020-11-12 15:23:21 +0000 | [diff] [blame] | 1305 | extern bool ceph_quota_is_same_realm(struct inode *old, struct inode *new); |
Luis Henriques | 2b83845 | 2018-01-05 10:47:21 +0000 | [diff] [blame] | 1306 | extern bool ceph_quota_is_max_bytes_exceeded(struct inode *inode, |
| 1307 | loff_t newlen); |
Luis Henriques | 1ab302a | 2018-01-05 10:47:22 +0000 | [diff] [blame] | 1308 | extern bool ceph_quota_is_max_bytes_approaching(struct inode *inode, |
| 1309 | loff_t newlen); |
Luis Henriques | 9122eed | 2018-01-31 10:53:13 +0000 | [diff] [blame] | 1310 | extern bool ceph_quota_update_statfs(struct ceph_fs_client *fsc, |
| 1311 | struct kstatfs *buf); |
Luis Henriques | 0c44a8e | 2019-03-21 10:20:10 +0000 | [diff] [blame] | 1312 | extern void ceph_cleanup_quotarealms_inodes(struct ceph_mds_client *mdsc); |
Luis Henriques | fb18a57 | 2018-01-05 10:47:18 +0000 | [diff] [blame] | 1313 | |
Sage Weil | de57606 | 2009-10-06 11:31:07 -0700 | [diff] [blame] | 1314 | #endif /* _FS_CEPH_SUPER_H */ |