aboutsummaryrefslogtreecommitdiffstats
path: root/fs/reiserfs
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2012-03-21 16:36:41 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2012-03-21 16:36:41 -0400
commite2a0883e4071237d09b604a342c28b96b44a04b3 (patch)
treeaa56f4d376b5eb1c32358c19c2669c2a94e0e1fd /fs/reiserfs
parent3a990a52f9f25f45469e272017a31e7a3fda60ed (diff)
parent07c0c5d8b8c122b2f2df9ee574ac3083daefc981 (diff)
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull vfs pile 1 from Al Viro: "This is _not_ all; in particular, Miklos' and Jan's stuff is not there yet." * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (64 commits) ext4: initialization of ext4_li_mtx needs to be done earlier debugfs-related mode_t whack-a-mole hfsplus: add an ioctl to bless files hfsplus: change finder_info to u32 hfsplus: initialise userflags qnx4: new helper - try_extent() qnx4: get rid of qnx4_bread/qnx4_getblk take removal of PF_FORKNOEXEC to flush_old_exec() trim includes in inode.c um: uml_dup_mmap() relies on ->mmap_sem being held, but activate_mm() doesn't hold it um: embed ->stub_pages[] into mmu_context gadgetfs: list_for_each_safe() misuse ocfs2: fix leaks on failure exits in module_init ecryptfs: make register_filesystem() the last potential failure exit ntfs: forgets to unregister sysctls on register_filesystem() failure logfs: missing cleanup on register_filesystem() failure jfs: mising cleanup on register_filesystem() failure make configfs_pin_fs() return root dentry on success configfs: configfs_create_dir() has parent dentry in dentry->d_parent configfs: sanitize configfs_create() ...
Diffstat (limited to 'fs/reiserfs')
-rw-r--r--fs/reiserfs/acl.h76
-rw-r--r--fs/reiserfs/bitmap.c4
-rw-r--r--fs/reiserfs/dir.c2
-rw-r--r--fs/reiserfs/do_balan.c2
-rw-r--r--fs/reiserfs/file.c6
-rw-r--r--fs/reiserfs/fix_node.c2
-rw-r--r--fs/reiserfs/hashes.c2
-rw-r--r--fs/reiserfs/ibalance.c2
-rw-r--r--fs/reiserfs/inode.c6
-rw-r--r--fs/reiserfs/ioctl.c2
-rw-r--r--fs/reiserfs/item_ops.c2
-rw-r--r--fs/reiserfs/journal.c2
-rw-r--r--fs/reiserfs/lbalance.c2
-rw-r--r--fs/reiserfs/lock.c2
-rw-r--r--fs/reiserfs/namei.c6
-rw-r--r--fs/reiserfs/objectid.c3
-rw-r--r--fs/reiserfs/prints.c4
-rw-r--r--fs/reiserfs/procfs.c3
-rw-r--r--fs/reiserfs/reiserfs.h2922
-rw-r--r--fs/reiserfs/resize.c3
-rw-r--r--fs/reiserfs/stree.c2
-rw-r--r--fs/reiserfs/super.c12
-rw-r--r--fs/reiserfs/tail_conversion.c2
-rw-r--r--fs/reiserfs/xattr.c6
-rw-r--r--fs/reiserfs/xattr.h122
-rw-r--r--fs/reiserfs/xattr_acl.c6
-rw-r--r--fs/reiserfs/xattr_security.c4
-rw-r--r--fs/reiserfs/xattr_trusted.c4
-rw-r--r--fs/reiserfs/xattr_user.c4
29 files changed, 3164 insertions, 51 deletions
diff --git a/fs/reiserfs/acl.h b/fs/reiserfs/acl.h
new file mode 100644
index 000000000000..f096b80e73d8
--- /dev/null
+++ b/fs/reiserfs/acl.h
@@ -0,0 +1,76 @@
1#include <linux/init.h>
2#include <linux/posix_acl.h>
3
4#define REISERFS_ACL_VERSION 0x0001
5
6typedef struct {
7 __le16 e_tag;
8 __le16 e_perm;
9 __le32 e_id;
10} reiserfs_acl_entry;
11
12typedef struct {
13 __le16 e_tag;
14 __le16 e_perm;
15} reiserfs_acl_entry_short;
16
17typedef struct {
18 __le32 a_version;
19} reiserfs_acl_header;
20
21static inline size_t reiserfs_acl_size(int count)
22{
23 if (count <= 4) {
24 return sizeof(reiserfs_acl_header) +
25 count * sizeof(reiserfs_acl_entry_short);
26 } else {
27 return sizeof(reiserfs_acl_header) +
28 4 * sizeof(reiserfs_acl_entry_short) +
29 (count - 4) * sizeof(reiserfs_acl_entry);
30 }
31}
32
33static inline int reiserfs_acl_count(size_t size)
34{
35 ssize_t s;
36 size -= sizeof(reiserfs_acl_header);
37 s = size - 4 * sizeof(reiserfs_acl_entry_short);
38 if (s < 0) {
39 if (size % sizeof(reiserfs_acl_entry_short))
40 return -1;
41 return size / sizeof(reiserfs_acl_entry_short);
42 } else {
43 if (s % sizeof(reiserfs_acl_entry))
44 return -1;
45 return s / sizeof(reiserfs_acl_entry) + 4;
46 }
47}
48
49#ifdef CONFIG_REISERFS_FS_POSIX_ACL
50struct posix_acl *reiserfs_get_acl(struct inode *inode, int type);
51int reiserfs_acl_chmod(struct inode *inode);
52int reiserfs_inherit_default_acl(struct reiserfs_transaction_handle *th,
53 struct inode *dir, struct dentry *dentry,
54 struct inode *inode);
55int reiserfs_cache_default_acl(struct inode *dir);
56extern const struct xattr_handler reiserfs_posix_acl_default_handler;
57extern const struct xattr_handler reiserfs_posix_acl_access_handler;
58
59#else
60
61#define reiserfs_cache_default_acl(inode) 0
62#define reiserfs_get_acl NULL
63
64static inline int reiserfs_acl_chmod(struct inode *inode)
65{
66 return 0;
67}
68
69static inline int
70reiserfs_inherit_default_acl(struct reiserfs_transaction_handle *th,
71 const struct inode *dir, struct dentry *dentry,
72 struct inode *inode)
73{
74 return 0;
75}
76#endif
diff --git a/fs/reiserfs/bitmap.c b/fs/reiserfs/bitmap.c
index 70de42f09f1d..4c0c7d163d15 100644
--- a/fs/reiserfs/bitmap.c
+++ b/fs/reiserfs/bitmap.c
@@ -4,14 +4,12 @@
4/* Reiserfs block (de)allocator, bitmap-based. */ 4/* Reiserfs block (de)allocator, bitmap-based. */
5 5
6#include <linux/time.h> 6#include <linux/time.h>
7#include <linux/reiserfs_fs.h> 7#include "reiserfs.h"
8#include <linux/errno.h> 8#include <linux/errno.h>
9#include <linux/buffer_head.h> 9#include <linux/buffer_head.h>
10#include <linux/kernel.h> 10#include <linux/kernel.h>
11#include <linux/pagemap.h> 11#include <linux/pagemap.h>
12#include <linux/vmalloc.h> 12#include <linux/vmalloc.h>
13#include <linux/reiserfs_fs_sb.h>
14#include <linux/reiserfs_fs_i.h>
15#include <linux/quotaops.h> 13#include <linux/quotaops.h>
16#include <linux/seq_file.h> 14#include <linux/seq_file.h>
17 15
diff --git a/fs/reiserfs/dir.c b/fs/reiserfs/dir.c
index 133e9355dc6f..66c53b642a88 100644
--- a/fs/reiserfs/dir.c
+++ b/fs/reiserfs/dir.c
@@ -5,7 +5,7 @@
5#include <linux/string.h> 5#include <linux/string.h>
6#include <linux/errno.h> 6#include <linux/errno.h>
7#include <linux/fs.h> 7#include <linux/fs.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9#include <linux/stat.h> 9#include <linux/stat.h>
10#include <linux/buffer_head.h> 10#include <linux/buffer_head.h>
11#include <linux/slab.h> 11#include <linux/slab.h>
diff --git a/fs/reiserfs/do_balan.c b/fs/reiserfs/do_balan.c
index 60c080440661..2b7882b508db 100644
--- a/fs/reiserfs/do_balan.c
+++ b/fs/reiserfs/do_balan.c
@@ -17,7 +17,7 @@
17 17
18#include <asm/uaccess.h> 18#include <asm/uaccess.h>
19#include <linux/time.h> 19#include <linux/time.h>
20#include <linux/reiserfs_fs.h> 20#include "reiserfs.h"
21#include <linux/buffer_head.h> 21#include <linux/buffer_head.h>
22#include <linux/kernel.h> 22#include <linux/kernel.h>
23 23
diff --git a/fs/reiserfs/file.c b/fs/reiserfs/file.c
index ace635053a36..8375c922c0d5 100644
--- a/fs/reiserfs/file.c
+++ b/fs/reiserfs/file.c
@@ -3,9 +3,9 @@
3 */ 3 */
4 4
5#include <linux/time.h> 5#include <linux/time.h>
6#include <linux/reiserfs_fs.h> 6#include "reiserfs.h"
7#include <linux/reiserfs_acl.h> 7#include "acl.h"
8#include <linux/reiserfs_xattr.h> 8#include "xattr.h"
9#include <asm/uaccess.h> 9#include <asm/uaccess.h>
10#include <linux/pagemap.h> 10#include <linux/pagemap.h>
11#include <linux/swap.h> 11#include <linux/swap.h>
diff --git a/fs/reiserfs/fix_node.c b/fs/reiserfs/fix_node.c
index 1e4250bc3a6f..430e0658704c 100644
--- a/fs/reiserfs/fix_node.c
+++ b/fs/reiserfs/fix_node.c
@@ -37,7 +37,7 @@
37#include <linux/time.h> 37#include <linux/time.h>
38#include <linux/slab.h> 38#include <linux/slab.h>
39#include <linux/string.h> 39#include <linux/string.h>
40#include <linux/reiserfs_fs.h> 40#include "reiserfs.h"
41#include <linux/buffer_head.h> 41#include <linux/buffer_head.h>
42 42
43/* To make any changes in the tree we find a node, that contains item 43/* To make any changes in the tree we find a node, that contains item
diff --git a/fs/reiserfs/hashes.c b/fs/reiserfs/hashes.c
index 6471c670743e..91b0cc1242a2 100644
--- a/fs/reiserfs/hashes.c
+++ b/fs/reiserfs/hashes.c
@@ -19,7 +19,7 @@
19// 19//
20 20
21#include <linux/kernel.h> 21#include <linux/kernel.h>
22#include <linux/reiserfs_fs.h> 22#include "reiserfs.h"
23#include <asm/types.h> 23#include <asm/types.h>
24 24
25#define DELTA 0x9E3779B9 25#define DELTA 0x9E3779B9
diff --git a/fs/reiserfs/ibalance.c b/fs/reiserfs/ibalance.c
index 2074fd95046b..e1978fd895f5 100644
--- a/fs/reiserfs/ibalance.c
+++ b/fs/reiserfs/ibalance.c
@@ -5,7 +5,7 @@
5#include <asm/uaccess.h> 5#include <asm/uaccess.h>
6#include <linux/string.h> 6#include <linux/string.h>
7#include <linux/time.h> 7#include <linux/time.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9#include <linux/buffer_head.h> 9#include <linux/buffer_head.h>
10 10
11/* this is one and only function that is used outside (do_balance.c) */ 11/* this is one and only function that is used outside (do_balance.c) */
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index 9e8cd5acd79c..494c315c7417 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -4,9 +4,9 @@
4 4
5#include <linux/time.h> 5#include <linux/time.h>
6#include <linux/fs.h> 6#include <linux/fs.h>
7#include <linux/reiserfs_fs.h> 7#include "reiserfs.h"
8#include <linux/reiserfs_acl.h> 8#include "acl.h"
9#include <linux/reiserfs_xattr.h> 9#include "xattr.h"
10#include <linux/exportfs.h> 10#include <linux/exportfs.h>
11#include <linux/pagemap.h> 11#include <linux/pagemap.h>
12#include <linux/highmem.h> 12#include <linux/highmem.h>
diff --git a/fs/reiserfs/ioctl.c b/fs/reiserfs/ioctl.c
index 950e3d1b5c9e..0c2185042d5f 100644
--- a/fs/reiserfs/ioctl.c
+++ b/fs/reiserfs/ioctl.c
@@ -5,7 +5,7 @@
5#include <linux/capability.h> 5#include <linux/capability.h>
6#include <linux/fs.h> 6#include <linux/fs.h>
7#include <linux/mount.h> 7#include <linux/mount.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9#include <linux/time.h> 9#include <linux/time.h>
10#include <asm/uaccess.h> 10#include <asm/uaccess.h>
11#include <linux/pagemap.h> 11#include <linux/pagemap.h>
diff --git a/fs/reiserfs/item_ops.c b/fs/reiserfs/item_ops.c
index 72cb1cc51b87..ee382ef3d300 100644
--- a/fs/reiserfs/item_ops.c
+++ b/fs/reiserfs/item_ops.c
@@ -3,7 +3,7 @@
3 */ 3 */
4 4
5#include <linux/time.h> 5#include <linux/time.h>
6#include <linux/reiserfs_fs.h> 6#include "reiserfs.h"
7 7
8// this contains item handlers for old item types: sd, direct, 8// this contains item handlers for old item types: sd, direct,
9// indirect, directory 9// indirect, directory
diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c
index c3cf54fd4de3..cf9f4de00a95 100644
--- a/fs/reiserfs/journal.c
+++ b/fs/reiserfs/journal.c
@@ -37,7 +37,7 @@
37#include <linux/time.h> 37#include <linux/time.h>
38#include <linux/semaphore.h> 38#include <linux/semaphore.h>
39#include <linux/vmalloc.h> 39#include <linux/vmalloc.h>
40#include <linux/reiserfs_fs.h> 40#include "reiserfs.h"
41#include <linux/kernel.h> 41#include <linux/kernel.h>
42#include <linux/errno.h> 42#include <linux/errno.h>
43#include <linux/fcntl.h> 43#include <linux/fcntl.h>
diff --git a/fs/reiserfs/lbalance.c b/fs/reiserfs/lbalance.c
index b43d01556313..79e5a8b4c226 100644
--- a/fs/reiserfs/lbalance.c
+++ b/fs/reiserfs/lbalance.c
@@ -5,7 +5,7 @@
5#include <asm/uaccess.h> 5#include <asm/uaccess.h>
6#include <linux/string.h> 6#include <linux/string.h>
7#include <linux/time.h> 7#include <linux/time.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9#include <linux/buffer_head.h> 9#include <linux/buffer_head.h>
10 10
11/* these are used in do_balance.c */ 11/* these are used in do_balance.c */
diff --git a/fs/reiserfs/lock.c b/fs/reiserfs/lock.c
index 7df1ce48203a..d735bc8470e3 100644
--- a/fs/reiserfs/lock.c
+++ b/fs/reiserfs/lock.c
@@ -1,4 +1,4 @@
1#include <linux/reiserfs_fs.h> 1#include "reiserfs.h"
2#include <linux/mutex.h> 2#include <linux/mutex.h>
3 3
4/* 4/*
diff --git a/fs/reiserfs/namei.c b/fs/reiserfs/namei.c
index 146378865239..84e8a69cee9d 100644
--- a/fs/reiserfs/namei.c
+++ b/fs/reiserfs/namei.c
@@ -14,9 +14,9 @@
14#include <linux/time.h> 14#include <linux/time.h>
15#include <linux/bitops.h> 15#include <linux/bitops.h>
16#include <linux/slab.h> 16#include <linux/slab.h>
17#include <linux/reiserfs_fs.h> 17#include "reiserfs.h"
18#include <linux/reiserfs_acl.h> 18#include "acl.h"
19#include <linux/reiserfs_xattr.h> 19#include "xattr.h"
20#include <linux/quotaops.h> 20#include <linux/quotaops.h>
21 21
22#define INC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) { inc_nlink(i); if (i->i_nlink >= REISERFS_LINK_MAX) set_nlink(i, 1); } 22#define INC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) { inc_nlink(i); if (i->i_nlink >= REISERFS_LINK_MAX) set_nlink(i, 1); }
diff --git a/fs/reiserfs/objectid.c b/fs/reiserfs/objectid.c
index 3a6de810bd61..f732d6a5251d 100644
--- a/fs/reiserfs/objectid.c
+++ b/fs/reiserfs/objectid.c
@@ -5,8 +5,7 @@
5#include <linux/string.h> 5#include <linux/string.h>
6#include <linux/random.h> 6#include <linux/random.h>
7#include <linux/time.h> 7#include <linux/time.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9#include <linux/reiserfs_fs_sb.h>
10 9
11// find where objectid map starts 10// find where objectid map starts
12#define objectid_map(s,rs) (old_format_only (s) ? \ 11#define objectid_map(s,rs) (old_format_only (s) ? \
diff --git a/fs/reiserfs/prints.c b/fs/reiserfs/prints.c
index 45de98b59466..c0b1112ab7e3 100644
--- a/fs/reiserfs/prints.c
+++ b/fs/reiserfs/prints.c
@@ -4,7 +4,7 @@
4 4
5#include <linux/time.h> 5#include <linux/time.h>
6#include <linux/fs.h> 6#include <linux/fs.h>
7#include <linux/reiserfs_fs.h> 7#include "reiserfs.h"
8#include <linux/string.h> 8#include <linux/string.h>
9#include <linux/buffer_head.h> 9#include <linux/buffer_head.h>
10 10
@@ -329,7 +329,7 @@ void reiserfs_debug(struct super_block *s, int level, const char *fmt, ...)
329 Numbering scheme for panic used by Vladimir and Anatoly( Hans completely ignores this scheme, and considers it 329 Numbering scheme for panic used by Vladimir and Anatoly( Hans completely ignores this scheme, and considers it
330 pointless complexity): 330 pointless complexity):
331 331
332 panics in reiserfs_fs.h have numbers from 1000 to 1999 332 panics in reiserfs.h have numbers from 1000 to 1999
333 super.c 2000 to 2999 333 super.c 2000 to 2999
334 preserve.c (unused) 3000 to 3999 334 preserve.c (unused) 3000 to 3999
335 bitmap.c 4000 to 4999 335 bitmap.c 4000 to 4999
diff --git a/fs/reiserfs/procfs.c b/fs/reiserfs/procfs.c
index 7a9981196c1c..2c1ade692cc8 100644
--- a/fs/reiserfs/procfs.c
+++ b/fs/reiserfs/procfs.c
@@ -12,8 +12,7 @@
12#include <linux/time.h> 12#include <linux/time.h>
13#include <linux/seq_file.h> 13#include <linux/seq_file.h>
14#include <asm/uaccess.h> 14#include <asm/uaccess.h>
15#include <linux/reiserfs_fs.h> 15#include "reiserfs.h"
16#include <linux/reiserfs_fs_sb.h>
17#include <linux/init.h> 16#include <linux/init.h>
18#include <linux/proc_fs.h> 17#include <linux/proc_fs.h>
19 18
diff --git a/fs/reiserfs/reiserfs.h b/fs/reiserfs/reiserfs.h
new file mode 100644
index 000000000000..445d768eea44
--- /dev/null
+++ b/fs/reiserfs/reiserfs.h
@@ -0,0 +1,2922 @@
1/*
2 * Copyright 1996, 1997, 1998 Hans Reiser, see reiserfs/README for licensing and copyright details
3 */
4
5#include <linux/reiserfs_fs.h>
6
7#include <linux/slab.h>
8#include <linux/interrupt.h>
9#include <linux/sched.h>
10#include <linux/workqueue.h>
11#include <asm/unaligned.h>
12#include <linux/bitops.h>
13#include <linux/proc_fs.h>
14#include <linux/buffer_head.h>
15
16/* the 32 bit compat definitions with int argument */
17#define REISERFS_IOC32_UNPACK _IOW(0xCD, 1, int)
18#define REISERFS_IOC32_GETFLAGS FS_IOC32_GETFLAGS
19#define REISERFS_IOC32_SETFLAGS FS_IOC32_SETFLAGS
20#define REISERFS_IOC32_GETVERSION FS_IOC32_GETVERSION
21#define REISERFS_IOC32_SETVERSION FS_IOC32_SETVERSION
22
23struct reiserfs_journal_list;
24
25/** bitmasks for i_flags field in reiserfs-specific part of inode */
26typedef enum {
27 /** this says what format of key do all items (but stat data) of
28 an object have. If this is set, that format is 3.6 otherwise
29 - 3.5 */
30 i_item_key_version_mask = 0x0001,
31 /** If this is unset, object has 3.5 stat data, otherwise, it has
32 3.6 stat data with 64bit size, 32bit nlink etc. */
33 i_stat_data_version_mask = 0x0002,
34 /** file might need tail packing on close */
35 i_pack_on_close_mask = 0x0004,
36 /** don't pack tail of file */
37 i_nopack_mask = 0x0008,
38 /** If those is set, "safe link" was created for this file during
39 truncate or unlink. Safe link is used to avoid leakage of disk
40 space on crash with some files open, but unlinked. */
41 i_link_saved_unlink_mask = 0x0010,
42 i_link_saved_truncate_mask = 0x0020,
43 i_has_xattr_dir = 0x0040,
44 i_data_log = 0x0080,
45} reiserfs_inode_flags;
46
47struct reiserfs_inode_info {
48 __u32 i_key[4]; /* key is still 4 32 bit integers */
49 /** transient inode flags that are never stored on disk. Bitmasks
50 for this field are defined above. */
51 __u32 i_flags;
52
53 __u32 i_first_direct_byte; // offset of first byte stored in direct item.
54
55 /* copy of persistent inode flags read from sd_attrs. */
56 __u32 i_attrs;
57
58 int i_prealloc_block; /* first unused block of a sequence of unused blocks */
59 int i_prealloc_count; /* length of that sequence */
60 struct list_head i_prealloc_list; /* per-transaction list of inodes which
61 * have preallocated blocks */
62
63 unsigned new_packing_locality:1; /* new_packig_locality is created; new blocks
64 * for the contents of this directory should be
65 * displaced */
66
67 /* we use these for fsync or O_SYNC to decide which transaction
68 ** needs to be committed in order for this inode to be properly
69 ** flushed */
70 unsigned int i_trans_id;
71 struct reiserfs_journal_list *i_jl;
72 atomic_t openers;
73 struct mutex tailpack;
74#ifdef CONFIG_REISERFS_FS_XATTR
75 struct rw_semaphore i_xattr_sem;
76#endif
77 struct inode vfs_inode;
78};
79
80typedef enum {
81 reiserfs_attrs_cleared = 0x00000001,
82} reiserfs_super_block_flags;
83
84/* struct reiserfs_super_block accessors/mutators
85 * since this is a disk structure, it will always be in
86 * little endian format. */
87#define sb_block_count(sbp) (le32_to_cpu((sbp)->s_v1.s_block_count))
88#define set_sb_block_count(sbp,v) ((sbp)->s_v1.s_block_count = cpu_to_le32(v))
89#define sb_free_blocks(sbp) (le32_to_cpu((sbp)->s_v1.s_free_blocks))
90#define set_sb_free_blocks(sbp,v) ((sbp)->s_v1.s_free_blocks = cpu_to_le32(v))
91#define sb_root_block(sbp) (le32_to_cpu((sbp)->s_v1.s_root_block))
92#define set_sb_root_block(sbp,v) ((sbp)->s_v1.s_root_block = cpu_to_le32(v))
93
94#define sb_jp_journal_1st_block(sbp) \
95 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_1st_block))
96#define set_sb_jp_journal_1st_block(sbp,v) \
97 ((sbp)->s_v1.s_journal.jp_journal_1st_block = cpu_to_le32(v))
98#define sb_jp_journal_dev(sbp) \
99 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_dev))
100#define set_sb_jp_journal_dev(sbp,v) \
101 ((sbp)->s_v1.s_journal.jp_journal_dev = cpu_to_le32(v))
102#define sb_jp_journal_size(sbp) \
103 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_size))
104#define set_sb_jp_journal_size(sbp,v) \
105 ((sbp)->s_v1.s_journal.jp_journal_size = cpu_to_le32(v))
106#define sb_jp_journal_trans_max(sbp) \
107 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_trans_max))
108#define set_sb_jp_journal_trans_max(sbp,v) \
109 ((sbp)->s_v1.s_journal.jp_journal_trans_max = cpu_to_le32(v))
110#define sb_jp_journal_magic(sbp) \
111 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_magic))
112#define set_sb_jp_journal_magic(sbp,v) \
113 ((sbp)->s_v1.s_journal.jp_journal_magic = cpu_to_le32(v))
114#define sb_jp_journal_max_batch(sbp) \
115 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_max_batch))
116#define set_sb_jp_journal_max_batch(sbp,v) \
117 ((sbp)->s_v1.s_journal.jp_journal_max_batch = cpu_to_le32(v))
118#define sb_jp_jourmal_max_commit_age(sbp) \
119 (le32_to_cpu((sbp)->s_v1.s_journal.jp_journal_max_commit_age))
120#define set_sb_jp_journal_max_commit_age(sbp,v) \
121 ((sbp)->s_v1.s_journal.jp_journal_max_commit_age = cpu_to_le32(v))
122
123#define sb_blocksize(sbp) (le16_to_cpu((sbp)->s_v1.s_blocksize))
124#define set_sb_blocksize(sbp,v) ((sbp)->s_v1.s_blocksize = cpu_to_le16(v))
125#define sb_oid_maxsize(sbp) (le16_to_cpu((sbp)->s_v1.s_oid_maxsize))
126#define set_sb_oid_maxsize(sbp,v) ((sbp)->s_v1.s_oid_maxsize = cpu_to_le16(v))
127#define sb_oid_cursize(sbp) (le16_to_cpu((sbp)->s_v1.s_oid_cursize))
128#define set_sb_oid_cursize(sbp,v) ((sbp)->s_v1.s_oid_cursize = cpu_to_le16(v))
129#define sb_umount_state(sbp) (le16_to_cpu((sbp)->s_v1.s_umount_state))
130#define set_sb_umount_state(sbp,v) ((sbp)->s_v1.s_umount_state = cpu_to_le16(v))
131#define sb_fs_state(sbp) (le16_to_cpu((sbp)->s_v1.s_fs_state))
132#define set_sb_fs_state(sbp,v) ((sbp)->s_v1.s_fs_state = cpu_to_le16(v))
133#define sb_hash_function_code(sbp) \
134 (le32_to_cpu((sbp)->s_v1.s_hash_function_code))
135#define set_sb_hash_function_code(sbp,v) \
136 ((sbp)->s_v1.s_hash_function_code = cpu_to_le32(v))
137#define sb_tree_height(sbp) (le16_to_cpu((sbp)->s_v1.s_tree_height))
138#define set_sb_tree_height(sbp,v) ((sbp)->s_v1.s_tree_height = cpu_to_le16(v))
139#define sb_bmap_nr(sbp) (le16_to_cpu((sbp)->s_v1.s_bmap_nr))
140#define set_sb_bmap_nr(sbp,v) ((sbp)->s_v1.s_bmap_nr = cpu_to_le16(v))
141#define sb_version(sbp) (le16_to_cpu((sbp)->s_v1.s_version))
142#define set_sb_version(sbp,v) ((sbp)->s_v1.s_version = cpu_to_le16(v))
143
144#define sb_mnt_count(sbp) (le16_to_cpu((sbp)->s_mnt_count))
145#define set_sb_mnt_count(sbp, v) ((sbp)->s_mnt_count = cpu_to_le16(v))
146
147#define sb_reserved_for_journal(sbp) \
148 (le16_to_cpu((sbp)->s_v1.s_reserved_for_journal))
149#define set_sb_reserved_for_journal(sbp,v) \
150 ((sbp)->s_v1.s_reserved_for_journal = cpu_to_le16(v))
151
152/* LOGGING -- */
153
154/* These all interelate for performance.
155**
156** If the journal block count is smaller than n transactions, you lose speed.
157** I don't know what n is yet, I'm guessing 8-16.
158**
159** typical transaction size depends on the application, how often fsync is
160** called, and how many metadata blocks you dirty in a 30 second period.
161** The more small files (<16k) you use, the larger your transactions will
162** be.
163**
164** If your journal fills faster than dirty buffers get flushed to disk, it must flush them before allowing the journal
165** to wrap, which slows things down. If you need high speed meta data updates, the journal should be big enough
166** to prevent wrapping before dirty meta blocks get to disk.
167**
168** If the batch max is smaller than the transaction max, you'll waste space at the end of the journal
169** because journal_end sets the next transaction to start at 0 if the next transaction has any chance of wrapping.
170**
171** The large the batch max age, the better the speed, and the more meta data changes you'll lose after a crash.
172**
173*/
174
175/* don't mess with these for a while */
176 /* we have a node size define somewhere in reiserfs_fs.h. -Hans */
177#define JOURNAL_BLOCK_SIZE 4096 /* BUG gotta get rid of this */
178#define JOURNAL_MAX_CNODE 1500 /* max cnodes to allocate. */
179#define JOURNAL_HASH_SIZE 8192
180#define JOURNAL_NUM_BITMAPS 5 /* number of copies of the bitmaps to have floating. Must be >= 2 */
181
182/* One of these for every block in every transaction
183** Each one is in two hash tables. First, a hash of the current transaction, and after journal_end, a
184** hash of all the in memory transactions.
185** next and prev are used by the current transaction (journal_hash).
186** hnext and hprev are used by journal_list_hash. If a block is in more than one transaction, the journal_list_hash
187** links it in multiple times. This allows flush_journal_list to remove just the cnode belonging
188** to a given transaction.
189*/
190struct reiserfs_journal_cnode {
191 struct buffer_head *bh; /* real buffer head */
192 struct super_block *sb; /* dev of real buffer head */
193 __u32 blocknr; /* block number of real buffer head, == 0 when buffer on disk */
194 unsigned long state;
195 struct reiserfs_journal_list *jlist; /* journal list this cnode lives in */
196 struct reiserfs_journal_cnode *next; /* next in transaction list */
197 struct reiserfs_journal_cnode *prev; /* prev in transaction list */
198 struct reiserfs_journal_cnode *hprev; /* prev in hash list */
199 struct reiserfs_journal_cnode *hnext; /* next in hash list */
200};
201
202struct reiserfs_bitmap_node {
203 int id;
204 char *data;
205 struct list_head list;
206};
207
208struct reiserfs_list_bitmap {
209 struct reiserfs_journal_list *journal_list;
210 struct reiserfs_bitmap_node **bitmaps;
211};
212
213/*
214** one of these for each transaction. The most important part here is the j_realblock.
215** this list of cnodes is used to hash all the blocks in all the commits, to mark all the
216** real buffer heads dirty once all the commits hit the disk,
217** and to make sure every real block in a transaction is on disk before allowing the log area
218** to be overwritten */
219struct reiserfs_journal_list {
220 unsigned long j_start;
221 unsigned long j_state;
222 unsigned long j_len;
223 atomic_t j_nonzerolen;
224 atomic_t j_commit_left;
225 atomic_t j_older_commits_done; /* all commits older than this on disk */
226 struct mutex j_commit_mutex;
227 unsigned int j_trans_id;
228 time_t j_timestamp;
229 struct reiserfs_list_bitmap *j_list_bitmap;
230 struct buffer_head *j_commit_bh; /* commit buffer head */
231 struct reiserfs_journal_cnode *j_realblock;
232 struct reiserfs_journal_cnode *j_freedlist; /* list of buffers that were freed during this trans. free each of these on flush */
233 /* time ordered list of all active transactions */
234 struct list_head j_list;
235
236 /* time ordered list of all transactions we haven't tried to flush yet */
237 struct list_head j_working_list;
238
239 /* list of tail conversion targets in need of flush before commit */
240 struct list_head j_tail_bh_list;
241 /* list of data=ordered buffers in need of flush before commit */
242 struct list_head j_bh_list;
243 int j_refcount;
244};
245
246struct reiserfs_journal {
247 struct buffer_head **j_ap_blocks; /* journal blocks on disk */
248 struct reiserfs_journal_cnode *j_last; /* newest journal block */
249 struct reiserfs_journal_cnode *j_first; /* oldest journal block. start here for traverse */
250
251 struct block_device *j_dev_bd;
252 fmode_t j_dev_mode;
253 int j_1st_reserved_block; /* first block on s_dev of reserved area journal */
254
255 unsigned long j_state;
256 unsigned int j_trans_id;
257 unsigned long j_mount_id;
258 unsigned long j_start; /* start of current waiting commit (index into j_ap_blocks) */
259 unsigned long j_len; /* length of current waiting commit */
260 unsigned long j_len_alloc; /* number of buffers requested by journal_begin() */
261 atomic_t j_wcount; /* count of writers for current commit */
262 unsigned long j_bcount; /* batch count. allows turning X transactions into 1 */
263 unsigned long j_first_unflushed_offset; /* first unflushed transactions offset */
264 unsigned j_last_flush_trans_id; /* last fully flushed journal timestamp */
265 struct buffer_head *j_header_bh;
266
267 time_t j_trans_start_time; /* time this transaction started */
268 struct mutex j_mutex;
269 struct mutex j_flush_mutex;
270 wait_queue_head_t j_join_wait; /* wait for current transaction to finish before starting new one */
271 atomic_t j_jlock; /* lock for j_join_wait */
272 int j_list_bitmap_index; /* number of next list bitmap to use */
273 int j_must_wait; /* no more journal begins allowed. MUST sleep on j_join_wait */
274 int j_next_full_flush; /* next journal_end will flush all journal list */
275 int j_next_async_flush; /* next journal_end will flush all async commits */
276
277 int j_cnode_used; /* number of cnodes on the used list */
278 int j_cnode_free; /* number of cnodes on the free list */
279
280 unsigned int j_trans_max; /* max number of blocks in a transaction. */
281 unsigned int j_max_batch; /* max number of blocks to batch into a trans */
282 unsigned int j_max_commit_age; /* in seconds, how old can an async commit be */
283 unsigned int j_max_trans_age; /* in seconds, how old can a transaction be */
284 unsigned int j_default_max_commit_age; /* the default for the max commit age */
285
286 struct reiserfs_journal_cnode *j_cnode_free_list;
287 struct reiserfs_journal_cnode *j_cnode_free_orig; /* orig pointer returned from vmalloc */
288
289 struct reiserfs_journal_list *j_current_jl;
290 int j_free_bitmap_nodes;
291 int j_used_bitmap_nodes;
292
293 int j_num_lists; /* total number of active transactions */
294 int j_num_work_lists; /* number that need attention from kreiserfsd */
295
296 /* debugging to make sure things are flushed in order */
297 unsigned int j_last_flush_id;
298
299 /* debugging to make sure things are committed in order */
300 unsigned int j_last_commit_id;
301
302 struct list_head j_bitmap_nodes;
303 struct list_head j_dirty_buffers;
304 spinlock_t j_dirty_buffers_lock; /* protects j_dirty_buffers */
305
306 /* list of all active transactions */
307 struct list_head j_journal_list;
308 /* lists that haven't been touched by writeback attempts */
309 struct list_head j_working_list;
310
311 struct reiserfs_list_bitmap j_list_bitmap[JOURNAL_NUM_BITMAPS]; /* array of bitmaps to record the deleted blocks */
312 struct reiserfs_journal_cnode *j_hash_table[JOURNAL_HASH_SIZE]; /* hash table for real buffer heads in current trans */
313 struct reiserfs_journal_cnode *j_list_hash_table[JOURNAL_HASH_SIZE]; /* hash table for all the real buffer heads in all
314 the transactions */
315 struct list_head j_prealloc_list; /* list of inodes which have preallocated blocks */
316 int j_persistent_trans;
317 unsigned long j_max_trans_size;
318 unsigned long j_max_batch_size;
319
320 int j_errno;
321
322 /* when flushing ordered buffers, throttle new ordered writers */
323 struct delayed_work j_work;
324 struct super_block *j_work_sb;
325 atomic_t j_async_throttle;
326};
327
328enum journal_state_bits {
329 J_WRITERS_BLOCKED = 1, /* set when new writers not allowed */
330 J_WRITERS_QUEUED, /* set when log is full due to too many writers */
331 J_ABORTED, /* set when log is aborted */
332};
333
334#define JOURNAL_DESC_MAGIC "ReIsErLB" /* ick. magic string to find desc blocks in the journal */
335
336typedef __u32(*hashf_t) (const signed char *, int);
337
338struct reiserfs_bitmap_info {
339 __u32 free_count;
340};
341
342struct proc_dir_entry;
343
344#if defined( CONFIG_PROC_FS ) && defined( CONFIG_REISERFS_PROC_INFO )
345typedef unsigned long int stat_cnt_t;
346typedef struct reiserfs_proc_info_data {
347 spinlock_t lock;
348 int exiting;
349 int max_hash_collisions;
350
351 stat_cnt_t breads;
352 stat_cnt_t bread_miss;
353 stat_cnt_t search_by_key;
354 stat_cnt_t search_by_key_fs_changed;
355 stat_cnt_t search_by_key_restarted;
356
357 stat_cnt_t insert_item_restarted;
358 stat_cnt_t paste_into_item_restarted;
359 stat_cnt_t cut_from_item_restarted;
360 stat_cnt_t delete_solid_item_restarted;
361 stat_cnt_t delete_item_restarted;
362
363 stat_cnt_t leaked_oid;
364 stat_cnt_t leaves_removable;
365
366 /* balances per level. Use explicit 5 as MAX_HEIGHT is not visible yet. */
367 stat_cnt_t balance_at[5]; /* XXX */
368 /* sbk == search_by_key */
369 stat_cnt_t sbk_read_at[5]; /* XXX */
370 stat_cnt_t sbk_fs_changed[5];
371 stat_cnt_t sbk_restarted[5];
372 stat_cnt_t items_at[5]; /* XXX */
373 stat_cnt_t free_at[5]; /* XXX */
374 stat_cnt_t can_node_be_removed[5]; /* XXX */
375 long int lnum[5]; /* XXX */
376 long int rnum[5]; /* XXX */
377 long int lbytes[5]; /* XXX */
378 long int rbytes[5]; /* XXX */
379 stat_cnt_t get_neighbors[5];
380 stat_cnt_t get_neighbors_restart[5];
381 stat_cnt_t need_l_neighbor[5];
382 stat_cnt_t need_r_neighbor[5];
383
384 stat_cnt_t free_block;
385 struct __scan_bitmap_stats {
386 stat_cnt_t call;
387 stat_cnt_t wait;
388 stat_cnt_t bmap;
389 stat_cnt_t retry;
390 stat_cnt_t in_journal_hint;
391 stat_cnt_t in_journal_nohint;
392 stat_cnt_t stolen;
393 } scan_bitmap;
394 struct __journal_stats {
395 stat_cnt_t in_journal;
396 stat_cnt_t in_journal_bitmap;
397 stat_cnt_t in_journal_reusable;
398 stat_cnt_t lock_journal;
399 stat_cnt_t lock_journal_wait;
400 stat_cnt_t journal_being;
401 stat_cnt_t journal_relock_writers;
402 stat_cnt_t journal_relock_wcount;
403 stat_cnt_t mark_dirty;
404 stat_cnt_t mark_dirty_already;
405 stat_cnt_t mark_dirty_notjournal;
406 stat_cnt_t restore_prepared;
407 stat_cnt_t prepare;
408 stat_cnt_t prepare_retry;
409 } journal;
410} reiserfs_proc_info_data_t;
411#else
412typedef struct reiserfs_proc_info_data {
413} reiserfs_proc_info_data_t;
414#endif
415
416/* reiserfs union of in-core super block data */
417struct reiserfs_sb_info {
418 struct buffer_head *s_sbh; /* Buffer containing the super block */
419 /* both the comment and the choice of
420 name are unclear for s_rs -Hans */
421 struct reiserfs_super_block *s_rs; /* Pointer to the super block in the buffer */
422 struct reiserfs_bitmap_info *s_ap_bitmap;
423 struct reiserfs_journal *s_journal; /* pointer to journal information */
424 unsigned short s_mount_state; /* reiserfs state (valid, invalid) */
425
426 /* Serialize writers access, replace the old bkl */
427 struct mutex lock;
428 /* Owner of the lock (can be recursive) */
429 struct task_struct *lock_owner;
430 /* Depth of the lock, start from -1 like the bkl */
431 int lock_depth;
432
433 /* Comment? -Hans */
434 void (*end_io_handler) (struct buffer_head *, int);
435 hashf_t s_hash_function; /* pointer to function which is used
436 to sort names in directory. Set on
437 mount */
438 unsigned long s_mount_opt; /* reiserfs's mount options are set
439 here (currently - NOTAIL, NOLOG,
440 REPLAYONLY) */
441
442 struct { /* This is a structure that describes block allocator options */
443 unsigned long bits; /* Bitfield for enable/disable kind of options */
444 unsigned long large_file_size; /* size started from which we consider file to be a large one(in blocks) */
445 int border; /* percentage of disk, border takes */
446 int preallocmin; /* Minimal file size (in blocks) starting from which we do preallocations */
447 int preallocsize; /* Number of blocks we try to prealloc when file
448 reaches preallocmin size (in blocks) or
449 prealloc_list is empty. */
450 } s_alloc_options;
451
452 /* Comment? -Hans */
453 wait_queue_head_t s_wait;
454 /* To be obsoleted soon by per buffer seals.. -Hans */
455 atomic_t s_generation_counter; // increased by one every time the
456 // tree gets re-balanced
457 unsigned long s_properties; /* File system properties. Currently holds
458 on-disk FS format */
459
460 /* session statistics */
461 int s_disk_reads;
462 int s_disk_writes;
463 int s_fix_nodes;
464 int s_do_balance;
465 int s_unneeded_left_neighbor;
466 int s_good_search_by_key_reada;
467 int s_bmaps;
468 int s_bmaps_without_search;
469 int s_direct2indirect;
470 int s_indirect2direct;
471 /* set up when it's ok for reiserfs_read_inode2() to read from
472 disk inode with nlink==0. Currently this is only used during
473 finish_unfinished() processing at mount time */
474 int s_is_unlinked_ok;
475 reiserfs_proc_info_data_t s_proc_info_data;
476 struct proc_dir_entry *procdir;
477 int reserved_blocks; /* amount of blocks reserved for further allocations */
478 spinlock_t bitmap_lock; /* this lock on now only used to protect reserved_blocks variable */
479 struct dentry *priv_root; /* root of /.reiserfs_priv */
480 struct dentry *xattr_root; /* root of /.reiserfs_priv/xattrs */
481 int j_errno;
482#ifdef CONFIG_QUOTA
483 char *s_qf_names[MAXQUOTAS];
484 int s_jquota_fmt;
485#endif
486 char *s_jdev; /* Stored jdev for mount option showing */
487#ifdef CONFIG_REISERFS_CHECK
488
489 struct tree_balance *cur_tb; /*
490 * Detects whether more than one
491 * copy of tb exists per superblock
492 * as a means of checking whether
493 * do_balance is executing concurrently
494 * against another tree reader/writer
495 * on a same mount point.
496 */
497#endif
498};
499
500/* Definitions of reiserfs on-disk properties: */
501#define REISERFS_3_5 0
502#define REISERFS_3_6 1
503#define REISERFS_OLD_FORMAT 2
504
505enum reiserfs_mount_options {
506/* Mount options */
507 REISERFS_LARGETAIL, /* large tails will be created in a session */
508 REISERFS_SMALLTAIL, /* small (for files less than block size) tails will be created in a session */
509 REPLAYONLY, /* replay journal and return 0. Use by fsck */
510 REISERFS_CONVERT, /* -o conv: causes conversion of old
511 format super block to the new
512 format. If not specified - old
513 partition will be dealt with in a
514 manner of 3.5.x */
515
516/* -o hash={tea, rupasov, r5, detect} is meant for properly mounting
517** reiserfs disks from 3.5.19 or earlier. 99% of the time, this option
518** is not required. If the normal autodection code can't determine which
519** hash to use (because both hashes had the same value for a file)
520** use this option to force a specific hash. It won't allow you to override
521** the existing hash on the FS, so if you have a tea hash disk, and mount
522** with -o hash=rupasov, the mount will fail.
523*/
524 FORCE_TEA_HASH, /* try to force tea hash on mount */
525 FORCE_RUPASOV_HASH, /* try to force rupasov hash on mount */
526 FORCE_R5_HASH, /* try to force rupasov hash on mount */
527 FORCE_HASH_DETECT, /* try to detect hash function on mount */
528
529 REISERFS_DATA_LOG,
530 REISERFS_DATA_ORDERED,
531 REISERFS_DATA_WRITEBACK,
532
533/* used for testing experimental features, makes benchmarking new
534 features with and without more convenient, should never be used by
535 users in any code shipped to users (ideally) */
536
537 REISERFS_NO_BORDER,
538 REISERFS_NO_UNHASHED_RELOCATION,
539 REISERFS_HASHED_RELOCATION,
540 REISERFS_ATTRS,
541 REISERFS_XATTRS_USER,
542 REISERFS_POSIXACL,
543 REISERFS_EXPOSE_PRIVROOT,
544 REISERFS_BARRIER_NONE,
545 REISERFS_BARRIER_FLUSH,
546
547 /* Actions on error */
548 REISERFS_ERROR_PANIC,
549 REISERFS_ERROR_RO,
550 REISERFS_ERROR_CONTINUE,
551
552 REISERFS_USRQUOTA, /* User quota option specified */
553 REISERFS_GRPQUOTA, /* Group quota option specified */
554
555 REISERFS_TEST1,
556 REISERFS_TEST2,
557 REISERFS_TEST3,
558 REISERFS_TEST4,
559 REISERFS_UNSUPPORTED_OPT,
560};
561
562#define reiserfs_r5_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_R5_HASH))
563#define reiserfs_rupasov_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_RUPASOV_HASH))
564#define reiserfs_tea_hash(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_TEA_HASH))
565#define reiserfs_hash_detect(s) (REISERFS_SB(s)->s_mount_opt & (1 << FORCE_HASH_DETECT))
566#define reiserfs_no_border(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_NO_BORDER))
567#define reiserfs_no_unhashed_relocation(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_NO_UNHASHED_RELOCATION))
568#define reiserfs_hashed_relocation(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_HASHED_RELOCATION))
569#define reiserfs_test4(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_TEST4))
570
571#define have_large_tails(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_LARGETAIL))
572#define have_small_tails(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_SMALLTAIL))
573#define replay_only(s) (REISERFS_SB(s)->s_mount_opt & (1 << REPLAYONLY))
574#define reiserfs_attrs(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ATTRS))
575#define old_format_only(s) (REISERFS_SB(s)->s_properties & (1 << REISERFS_3_5))
576#define convert_reiserfs(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_CONVERT))
577#define reiserfs_data_log(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_LOG))
578#define reiserfs_data_ordered(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_ORDERED))
579#define reiserfs_data_writeback(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_DATA_WRITEBACK))
580#define reiserfs_xattrs_user(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_XATTRS_USER))
581#define reiserfs_posixacl(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_POSIXACL))
582#define reiserfs_expose_privroot(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_EXPOSE_PRIVROOT))
583#define reiserfs_xattrs_optional(s) (reiserfs_xattrs_user(s) || reiserfs_posixacl(s))
584#define reiserfs_barrier_none(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_BARRIER_NONE))
585#define reiserfs_barrier_flush(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_BARRIER_FLUSH))
586
587#define reiserfs_error_panic(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ERROR_PANIC))
588#define reiserfs_error_ro(s) (REISERFS_SB(s)->s_mount_opt & (1 << REISERFS_ERROR_RO))
589
590void reiserfs_file_buffer(struct buffer_head *bh, int list);
591extern struct file_system_type reiserfs_fs_type;
592int reiserfs_resize(struct super_block *, unsigned long);
593
594#define CARRY_ON 0
595#define SCHEDULE_OCCURRED 1
596
597#define SB_BUFFER_WITH_SB(s) (REISERFS_SB(s)->s_sbh)
598#define SB_JOURNAL(s) (REISERFS_SB(s)->s_journal)
599#define SB_JOURNAL_1st_RESERVED_BLOCK(s) (SB_JOURNAL(s)->j_1st_reserved_block)
600#define SB_JOURNAL_LEN_FREE(s) (SB_JOURNAL(s)->j_journal_len_free)
601#define SB_AP_BITMAP(s) (REISERFS_SB(s)->s_ap_bitmap)
602
603#define SB_DISK_JOURNAL_HEAD(s) (SB_JOURNAL(s)->j_header_bh->)
604
605/* A safe version of the "bdevname", which returns the "s_id" field of
606 * a superblock or else "Null superblock" if the super block is NULL.
607 */
608static inline char *reiserfs_bdevname(struct super_block *s)
609{
610 return (s == NULL) ? "Null superblock" : s->s_id;
611}
612
613#define reiserfs_is_journal_aborted(journal) (unlikely (__reiserfs_is_journal_aborted (journal)))
614static inline int __reiserfs_is_journal_aborted(struct reiserfs_journal
615 *journal)
616{
617 return test_bit(J_ABORTED, &journal->j_state);
618}
619
620/*
621 * Locking primitives. The write lock is a per superblock
622 * special mutex that has properties close to the Big Kernel Lock
623 * which was used in the previous locking scheme.
624 */
625void reiserfs_write_lock(struct super_block *s);
626void reiserfs_write_unlock(struct super_block *s);
627int reiserfs_write_lock_once(struct super_block *s);
628void reiserfs_write_unlock_once(struct super_block *s, int lock_depth);
629
630#ifdef CONFIG_REISERFS_CHECK
631void reiserfs_lock_check_recursive(struct super_block *s);
632#else
633static inline void reiserfs_lock_check_recursive(struct super_block *s) { }
634#endif
635
636/*
637 * Several mutexes depend on the write lock.
638 * However sometimes we want to relax the write lock while we hold
639 * these mutexes, according to the release/reacquire on schedule()
640 * properties of the Bkl that were used.
641 * Reiserfs performances and locking were based on this scheme.
642 * Now that the write lock is a mutex and not the bkl anymore, doing so
643 * may result in a deadlock:
644 *
645 * A acquire write_lock
646 * A acquire j_commit_mutex
647 * A release write_lock and wait for something
648 * B acquire write_lock
649 * B can't acquire j_commit_mutex and sleep
650 * A can't acquire write lock anymore
651 * deadlock
652 *
653 * What we do here is avoiding such deadlock by playing the same game
654 * than the Bkl: if we can't acquire a mutex that depends on the write lock,
655 * we release the write lock, wait a bit and then retry.
656 *
657 * The mutexes concerned by this hack are:
658 * - The commit mutex of a journal list
659 * - The flush mutex
660 * - The journal lock
661 * - The inode mutex
662 */
663static inline void reiserfs_mutex_lock_safe(struct mutex *m,
664 struct super_block *s)
665{
666 reiserfs_lock_check_recursive(s);
667 reiserfs_write_unlock(s);
668 mutex_lock(m);
669 reiserfs_write_lock(s);
670}
671
672static inline void
673reiserfs_mutex_lock_nested_safe(struct mutex *m, unsigned int subclass,
674 struct super_block *s)
675{
676 reiserfs_lock_check_recursive(s);
677 reiserfs_write_unlock(s);
678 mutex_lock_nested(m, subclass);
679 reiserfs_write_lock(s);
680}
681
682static inline void
683reiserfs_down_read_safe(struct rw_semaphore *sem, struct super_block *s)
684{
685 reiserfs_lock_check_recursive(s);
686 reiserfs_write_unlock(s);
687 down_read(sem);
688 reiserfs_write_lock(s);
689}
690
691/*
692 * When we schedule, we usually want to also release the write lock,
693 * according to the previous bkl based locking scheme of reiserfs.
694 */
695static inline void reiserfs_cond_resched(struct super_block *s)
696{
697 if (need_resched()) {
698 reiserfs_write_unlock(s);
699 schedule();
700 reiserfs_write_lock(s);
701 }
702}
703
704struct fid;
705
706/* in reading the #defines, it may help to understand that they employ
707 the following abbreviations:
708
709 B = Buffer
710 I = Item header
711 H = Height within the tree (should be changed to LEV)
712 N = Number of the item in the node
713 STAT = stat data
714 DEH = Directory Entry Header
715 EC = Entry Count
716 E = Entry number
717 UL = Unsigned Long
718 BLKH = BLocK Header
719 UNFM = UNForMatted node
720 DC = Disk Child
721 P = Path
722
723 These #defines are named by concatenating these abbreviations,
724 where first comes the arguments, and last comes the return value,
725 of the macro.
726
727*/
728
729#define USE_INODE_GENERATION_COUNTER
730
731#define REISERFS_PREALLOCATE
732#define DISPLACE_NEW_PACKING_LOCALITIES
733#define PREALLOCATION_SIZE 9
734
735/* n must be power of 2 */
736#define _ROUND_UP(x,n) (((x)+(n)-1u) & ~((n)-1u))
737
738// to be ok for alpha and others we have to align structures to 8 byte
739// boundary.
740// FIXME: do not change 4 by anything else: there is code which relies on that
741#define ROUND_UP(x) _ROUND_UP(x,8LL)
742
743/* debug levels. Right now, CONFIG_REISERFS_CHECK means print all debug
744** messages.
745*/
746#define REISERFS_DEBUG_CODE 5 /* extra messages to help find/debug errors */
747
748void __reiserfs_warning(struct super_block *s, const char *id,
749 const char *func, const char *fmt, ...);
750#define reiserfs_warning(s, id, fmt, args...) \
751 __reiserfs_warning(s, id, __func__, fmt, ##args)
752/* assertions handling */
753
754/** always check a condition and panic if it's false. */
755#define __RASSERT(cond, scond, format, args...) \
756do { \
757 if (!(cond)) \
758 reiserfs_panic(NULL, "assertion failure", "(" #cond ") at " \
759 __FILE__ ":%i:%s: " format "\n", \
760 in_interrupt() ? -1 : task_pid_nr(current), \
761 __LINE__, __func__ , ##args); \
762} while (0)
763
764#define RASSERT(cond, format, args...) __RASSERT(cond, #cond, format, ##args)
765
766#if defined( CONFIG_REISERFS_CHECK )
767#define RFALSE(cond, format, args...) __RASSERT(!(cond), "!(" #cond ")", format, ##args)
768#else
769#define RFALSE( cond, format, args... ) do {;} while( 0 )
770#endif
771
772#define CONSTF __attribute_const__
773/*
774 * Disk Data Structures
775 */
776
777/***************************************************************************/
778/* SUPER BLOCK */
779/***************************************************************************/
780
781/*
782 * Structure of super block on disk, a version of which in RAM is often accessed as REISERFS_SB(s)->s_rs
783 * the version in RAM is part of a larger structure containing fields never written to disk.
784 */
785#define UNSET_HASH 0 // read_super will guess about, what hash names
786 // in directories were sorted with
787#define TEA_HASH 1
788#define YURA_HASH 2
789#define R5_HASH 3
790#define DEFAULT_HASH R5_HASH
791
792struct journal_params {
793 __le32 jp_journal_1st_block; /* where does journal start from on its
794 * device */
795 __le32 jp_journal_dev; /* journal device st_rdev */
796 __le32 jp_journal_size; /* size of the journal */
797 __le32 jp_journal_trans_max; /* max number of blocks in a transaction. */
798 __le32 jp_journal_magic; /* random value made on fs creation (this
799 * was sb_journal_block_count) */
800 __le32 jp_journal_max_batch; /* max number of blocks to batch into a
801 * trans */
802 __le32 jp_journal_max_commit_age; /* in seconds, how old can an async
803 * commit be */
804 __le32 jp_journal_max_trans_age; /* in seconds, how old can a transaction
805 * be */
806};
807
808/* this is the super from 3.5.X, where X >= 10 */
809struct reiserfs_super_block_v1 {
810 __le32 s_block_count; /* blocks count */
811 __le32 s_free_blocks; /* free blocks count */
812 __le32 s_root_block; /* root block number */
813 struct journal_params s_journal;
814 __le16 s_blocksize; /* block size */
815 __le16 s_oid_maxsize; /* max size of object id array, see
816 * get_objectid() commentary */
817 __le16 s_oid_cursize; /* current size of object id array */
818 __le16 s_umount_state; /* this is set to 1 when filesystem was
819 * umounted, to 2 - when not */
820 char s_magic[10]; /* reiserfs magic string indicates that
821 * file system is reiserfs:
822 * "ReIsErFs" or "ReIsEr2Fs" or "ReIsEr3Fs" */
823 __le16 s_fs_state; /* it is set to used by fsck to mark which
824 * phase of rebuilding is done */
825 __le32 s_hash_function_code; /* indicate, what hash function is being use
826 * to sort names in a directory*/
827 __le16 s_tree_height; /* height of disk tree */
828 __le16 s_bmap_nr; /* amount of bitmap blocks needed to address
829 * each block of file system */
830 __le16 s_version; /* this field is only reliable on filesystem
831 * with non-standard journal */
832 __le16 s_reserved_for_journal; /* size in blocks of journal area on main
833 * device, we need to keep after
834 * making fs with non-standard journal */
835} __attribute__ ((__packed__));
836
837#define SB_SIZE_V1 (sizeof(struct reiserfs_super_block_v1))
838
839/* this is the on disk super block */
840struct reiserfs_super_block {
841 struct reiserfs_super_block_v1 s_v1;
842 __le32 s_inode_generation;
843 __le32 s_flags; /* Right now used only by inode-attributes, if enabled */
844 unsigned char s_uuid[16]; /* filesystem unique identifier */
845 unsigned char s_label[16]; /* filesystem volume label */
846 __le16 s_mnt_count; /* Count of mounts since last fsck */
847 __le16 s_max_mnt_count; /* Maximum mounts before check */
848 __le32 s_lastcheck; /* Timestamp of last fsck */
849 __le32 s_check_interval; /* Interval between checks */
850 char s_unused[76]; /* zero filled by mkreiserfs and
851 * reiserfs_convert_objectid_map_v1()
852 * so any additions must be updated
853 * there as well. */
854} __attribute__ ((__packed__));
855
856#define SB_SIZE (sizeof(struct reiserfs_super_block))
857
858#define REISERFS_VERSION_1 0
859#define REISERFS_VERSION_2 2
860
861// on-disk super block fields converted to cpu form
862#define SB_DISK_SUPER_BLOCK(s) (REISERFS_SB(s)->s_rs)
863#define SB_V1_DISK_SUPER_BLOCK(s) (&(SB_DISK_SUPER_BLOCK(s)->s_v1))
864#define SB_BLOCKSIZE(s) \
865 le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_blocksize))
866#define SB_BLOCK_COUNT(s) \
867 le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_block_count))
868#define SB_FREE_BLOCKS(s) \
869 le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks))
870#define SB_REISERFS_MAGIC(s) \
871 (SB_V1_DISK_SUPER_BLOCK(s)->s_magic)
872#define SB_ROOT_BLOCK(s) \
873 le32_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_root_block))
874#define SB_TREE_HEIGHT(s) \
875 le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height))
876#define SB_REISERFS_STATE(s) \
877 le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state))
878#define SB_VERSION(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_version))
879#define SB_BMAP_NR(s) le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_bmap_nr))
880
881#define PUT_SB_BLOCK_COUNT(s, val) \
882 do { SB_V1_DISK_SUPER_BLOCK(s)->s_block_count = cpu_to_le32(val); } while (0)
883#define PUT_SB_FREE_BLOCKS(s, val) \
884 do { SB_V1_DISK_SUPER_BLOCK(s)->s_free_blocks = cpu_to_le32(val); } while (0)
885#define PUT_SB_ROOT_BLOCK(s, val) \
886 do { SB_V1_DISK_SUPER_BLOCK(s)->s_root_block = cpu_to_le32(val); } while (0)
887#define PUT_SB_TREE_HEIGHT(s, val) \
888 do { SB_V1_DISK_SUPER_BLOCK(s)->s_tree_height = cpu_to_le16(val); } while (0)
889#define PUT_SB_REISERFS_STATE(s, val) \
890 do { SB_V1_DISK_SUPER_BLOCK(s)->s_umount_state = cpu_to_le16(val); } while (0)
891#define PUT_SB_VERSION(s, val) \
892 do { SB_V1_DISK_SUPER_BLOCK(s)->s_version = cpu_to_le16(val); } while (0)
893#define PUT_SB_BMAP_NR(s, val) \
894 do { SB_V1_DISK_SUPER_BLOCK(s)->s_bmap_nr = cpu_to_le16 (val); } while (0)
895
896#define SB_ONDISK_JP(s) (&SB_V1_DISK_SUPER_BLOCK(s)->s_journal)
897#define SB_ONDISK_JOURNAL_SIZE(s) \
898 le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_size))
899#define SB_ONDISK_JOURNAL_1st_BLOCK(s) \
900 le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_1st_block))
901#define SB_ONDISK_JOURNAL_DEVICE(s) \
902 le32_to_cpu ((SB_ONDISK_JP(s)->jp_journal_dev))
903#define SB_ONDISK_RESERVED_FOR_JOURNAL(s) \
904 le16_to_cpu ((SB_V1_DISK_SUPER_BLOCK(s)->s_reserved_for_journal))
905
906#define is_block_in_log_or_reserved_area(s, block) \
907 block >= SB_JOURNAL_1st_RESERVED_BLOCK(s) \
908 && block < SB_JOURNAL_1st_RESERVED_BLOCK(s) + \
909 ((!is_reiserfs_jr(SB_DISK_SUPER_BLOCK(s)) ? \
910 SB_ONDISK_JOURNAL_SIZE(s) + 1 : SB_ONDISK_RESERVED_FOR_JOURNAL(s)))
911
912int is_reiserfs_3_5(struct reiserfs_super_block *rs);
913int is_reiserfs_3_6(struct reiserfs_super_block *rs);
914int is_reiserfs_jr(struct reiserfs_super_block *rs);
915
916/* ReiserFS leaves the first 64k unused, so that partition labels have
917 enough space. If someone wants to write a fancy bootloader that
918 needs more than 64k, let us know, and this will be increased in size.
919 This number must be larger than than the largest block size on any
920 platform, or code will break. -Hans */
921#define REISERFS_DISK_OFFSET_IN_BYTES (64 * 1024)
922#define REISERFS_FIRST_BLOCK unused_define
923#define REISERFS_JOURNAL_OFFSET_IN_BYTES REISERFS_DISK_OFFSET_IN_BYTES
924
925/* the spot for the super in versions 3.5 - 3.5.10 (inclusive) */
926#define REISERFS_OLD_DISK_OFFSET_IN_BYTES (8 * 1024)
927
928/* reiserfs internal error code (used by search_by_key and fix_nodes)) */
929#define CARRY_ON 0
930#define REPEAT_SEARCH -1
931#define IO_ERROR -2
932#define NO_DISK_SPACE -3
933#define NO_BALANCING_NEEDED (-4)
934#define NO_MORE_UNUSED_CONTIGUOUS_BLOCKS (-5)
935#define QUOTA_EXCEEDED -6
936
937typedef __u32 b_blocknr_t;
938typedef __le32 unp_t;
939
940struct unfm_nodeinfo {
941 unp_t unfm_nodenum;
942 unsigned short unfm_freespace;
943};
944
945/* there are two formats of keys: 3.5 and 3.6
946 */
947#define KEY_FORMAT_3_5 0
948#define KEY_FORMAT_3_6 1
949
950/* there are two stat datas */
951#define STAT_DATA_V1 0
952#define STAT_DATA_V2 1
953
954static inline struct reiserfs_inode_info *REISERFS_I(const struct inode *inode)
955{
956 return container_of(inode, struct reiserfs_inode_info, vfs_inode);
957}
958
959static inline struct reiserfs_sb_info *REISERFS_SB(const struct super_block *sb)
960{
961 return sb->s_fs_info;
962}
963
964/* Don't trust REISERFS_SB(sb)->s_bmap_nr, it's a u16
965 * which overflows on large file systems. */
966static inline __u32 reiserfs_bmap_count(struct super_block *sb)
967{
968 return (SB_BLOCK_COUNT(sb) - 1) / (sb->s_blocksize * 8) + 1;
969}
970
971static inline int bmap_would_wrap(unsigned bmap_nr)
972{
973 return bmap_nr > ((1LL << 16) - 1);
974}
975
976/** this says about version of key of all items (but stat data) the
977 object consists of */
978#define get_inode_item_key_version( inode ) \
979 ((REISERFS_I(inode)->i_flags & i_item_key_version_mask) ? KEY_FORMAT_3_6 : KEY_FORMAT_3_5)
980
981#define set_inode_item_key_version( inode, version ) \
982 ({ if((version)==KEY_FORMAT_3_6) \
983 REISERFS_I(inode)->i_flags |= i_item_key_version_mask; \
984 else \
985 REISERFS_I(inode)->i_flags &= ~i_item_key_version_mask; })
986
987#define get_inode_sd_version(inode) \
988 ((REISERFS_I(inode)->i_flags & i_stat_data_version_mask) ? STAT_DATA_V2 : STAT_DATA_V1)
989
990#define set_inode_sd_version(inode, version) \
991 ({ if((version)==STAT_DATA_V2) \
992 REISERFS_I(inode)->i_flags |= i_stat_data_version_mask; \
993 else \
994 REISERFS_I(inode)->i_flags &= ~i_stat_data_version_mask; })
995
996/* This is an aggressive tail suppression policy, I am hoping it
997 improves our benchmarks. The principle behind it is that percentage
998 space saving is what matters, not absolute space saving. This is
999 non-intuitive, but it helps to understand it if you consider that the
1000 cost to access 4 blocks is not much more than the cost to access 1
1001 block, if you have to do a seek and rotate. A tail risks a
1002 non-linear disk access that is significant as a percentage of total
1003 time cost for a 4 block file and saves an amount of space that is
1004 less significant as a percentage of space, or so goes the hypothesis.
1005 -Hans */
1006#define STORE_TAIL_IN_UNFM_S1(n_file_size,n_tail_size,n_block_size) \
1007(\
1008 (!(n_tail_size)) || \
1009 (((n_tail_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) || \
1010 ( (n_file_size) >= (n_block_size) * 4 ) || \
1011 ( ( (n_file_size) >= (n_block_size) * 3 ) && \
1012 ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/4) ) || \
1013 ( ( (n_file_size) >= (n_block_size) * 2 ) && \
1014 ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/2) ) || \
1015 ( ( (n_file_size) >= (n_block_size) ) && \
1016 ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size) * 3)/4) ) ) \
1017)
1018
1019/* Another strategy for tails, this one means only create a tail if all the
1020 file would fit into one DIRECT item.
1021 Primary intention for this one is to increase performance by decreasing
1022 seeking.
1023*/
1024#define STORE_TAIL_IN_UNFM_S2(n_file_size,n_tail_size,n_block_size) \
1025(\
1026 (!(n_tail_size)) || \
1027 (((n_file_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) ) \
1028)
1029
1030/*
1031 * values for s_umount_state field
1032 */
1033#define REISERFS_VALID_FS 1
1034#define REISERFS_ERROR_FS 2
1035
1036//
1037// there are 5 item types currently
1038//
1039#define TYPE_STAT_DATA 0
1040#define TYPE_INDIRECT 1
1041#define TYPE_DIRECT 2
1042#define TYPE_DIRENTRY 3
1043#define TYPE_MAXTYPE 3
1044#define TYPE_ANY 15 // FIXME: comment is required
1045
1046/***************************************************************************/
1047/* KEY & ITEM HEAD */
1048/***************************************************************************/
1049
1050//
1051// directories use this key as well as old files
1052//
1053struct offset_v1 {
1054 __le32 k_offset;
1055 __le32 k_uniqueness;
1056} __attribute__ ((__packed__));
1057
1058struct offset_v2 {
1059 __le64 v;
1060} __attribute__ ((__packed__));
1061
1062static inline __u16 offset_v2_k_type(const struct offset_v2 *v2)
1063{
1064 __u8 type = le64_to_cpu(v2->v) >> 60;
1065 return (type <= TYPE_MAXTYPE) ? type : TYPE_ANY;
1066}
1067
1068static inline void set_offset_v2_k_type(struct offset_v2 *v2, int type)
1069{
1070 v2->v =
1071 (v2->v & cpu_to_le64(~0ULL >> 4)) | cpu_to_le64((__u64) type << 60);
1072}
1073
1074static inline loff_t offset_v2_k_offset(const struct offset_v2 *v2)
1075{
1076 return le64_to_cpu(v2->v) & (~0ULL >> 4);
1077}
1078
1079static inline void set_offset_v2_k_offset(struct offset_v2 *v2, loff_t offset)
1080{
1081 offset &= (~0ULL >> 4);
1082 v2->v = (v2->v & cpu_to_le64(15ULL << 60)) | cpu_to_le64(offset);
1083}
1084
1085/* Key of an item determines its location in the S+tree, and
1086 is composed of 4 components */
1087struct reiserfs_key {
1088 __le32 k_dir_id; /* packing locality: by default parent
1089 directory object id */
1090 __le32 k_objectid; /* object identifier */
1091 union {
1092 struct offset_v1 k_offset_v1;
1093 struct offset_v2 k_offset_v2;
1094 } __attribute__ ((__packed__)) u;
1095} __attribute__ ((__packed__));
1096
1097struct in_core_key {
1098 __u32 k_dir_id; /* packing locality: by default parent
1099 directory object id */
1100 __u32 k_objectid; /* object identifier */
1101 __u64 k_offset;
1102 __u8 k_type;
1103};
1104
1105struct cpu_key {
1106 struct in_core_key on_disk_key;
1107 int version;
1108 int key_length; /* 3 in all cases but direct2indirect and
1109 indirect2direct conversion */
1110};
1111
1112/* Our function for comparing keys can compare keys of different
1113 lengths. It takes as a parameter the length of the keys it is to
1114 compare. These defines are used in determining what is to be passed
1115 to it as that parameter. */
1116#define REISERFS_FULL_KEY_LEN 4
1117#define REISERFS_SHORT_KEY_LEN 2
1118
1119/* The result of the key compare */
1120#define FIRST_GREATER 1
1121#define SECOND_GREATER -1
1122#define KEYS_IDENTICAL 0
1123#define KEY_FOUND 1
1124#define KEY_NOT_FOUND 0
1125
1126#define KEY_SIZE (sizeof(struct reiserfs_key))
1127#define SHORT_KEY_SIZE (sizeof (__u32) + sizeof (__u32))
1128
1129/* return values for search_by_key and clones */
1130#define ITEM_FOUND 1
1131#define ITEM_NOT_FOUND 0
1132#define ENTRY_FOUND 1
1133#define ENTRY_NOT_FOUND 0
1134#define DIRECTORY_NOT_FOUND -1
1135#define REGULAR_FILE_FOUND -2
1136#define DIRECTORY_FOUND -3
1137#define BYTE_FOUND 1
1138#define BYTE_NOT_FOUND 0
1139#define FILE_NOT_FOUND -1
1140
1141#define POSITION_FOUND 1
1142#define POSITION_NOT_FOUND 0
1143
1144// return values for reiserfs_find_entry and search_by_entry_key
1145#define NAME_FOUND 1
1146#define NAME_NOT_FOUND 0
1147#define GOTO_PREVIOUS_ITEM 2
1148#define NAME_FOUND_INVISIBLE 3
1149
1150/* Everything in the filesystem is stored as a set of items. The
1151 item head contains the key of the item, its free space (for
1152 indirect items) and specifies the location of the item itself
1153 within the block. */
1154
1155struct item_head {
1156 /* Everything in the tree is found by searching for it based on
1157 * its key.*/
1158 struct reiserfs_key ih_key;
1159 union {
1160 /* The free space in the last unformatted node of an
1161 indirect item if this is an indirect item. This
1162 equals 0xFFFF iff this is a direct item or stat data
1163 item. Note that the key, not this field, is used to
1164 determine the item type, and thus which field this
1165 union contains. */
1166 __le16 ih_free_space_reserved;
1167 /* Iff this is a directory item, this field equals the
1168 number of directory entries in the directory item. */
1169 __le16 ih_entry_count;
1170 } __attribute__ ((__packed__)) u;
1171 __le16 ih_item_len; /* total size of the item body */
1172 __le16 ih_item_location; /* an offset to the item body
1173 * within the block */
1174 __le16 ih_version; /* 0 for all old items, 2 for new
1175 ones. Highest bit is set by fsck
1176 temporary, cleaned after all
1177 done */
1178} __attribute__ ((__packed__));
1179/* size of item header */
1180#define IH_SIZE (sizeof(struct item_head))
1181
1182#define ih_free_space(ih) le16_to_cpu((ih)->u.ih_free_space_reserved)
1183#define ih_version(ih) le16_to_cpu((ih)->ih_version)
1184#define ih_entry_count(ih) le16_to_cpu((ih)->u.ih_entry_count)
1185#define ih_location(ih) le16_to_cpu((ih)->ih_item_location)
1186#define ih_item_len(ih) le16_to_cpu((ih)->ih_item_len)
1187
1188#define put_ih_free_space(ih, val) do { (ih)->u.ih_free_space_reserved = cpu_to_le16(val); } while(0)
1189#define put_ih_version(ih, val) do { (ih)->ih_version = cpu_to_le16(val); } while (0)
1190#define put_ih_entry_count(ih, val) do { (ih)->u.ih_entry_count = cpu_to_le16(val); } while (0)
1191#define put_ih_location(ih, val) do { (ih)->ih_item_location = cpu_to_le16(val); } while (0)
1192#define put_ih_item_len(ih, val) do { (ih)->ih_item_len = cpu_to_le16(val); } while (0)
1193
1194#define unreachable_item(ih) (ih_version(ih) & (1 << 15))
1195
1196#define get_ih_free_space(ih) (ih_version (ih) == KEY_FORMAT_3_6 ? 0 : ih_free_space (ih))
1197#define set_ih_free_space(ih,val) put_ih_free_space((ih), ((ih_version(ih) == KEY_FORMAT_3_6) ? 0 : (val)))
1198
1199/* these operate on indirect items, where you've got an array of ints
1200** at a possibly unaligned location. These are a noop on ia32
1201**
1202** p is the array of __u32, i is the index into the array, v is the value
1203** to store there.
1204*/
1205#define get_block_num(p, i) get_unaligned_le32((p) + (i))
1206#define put_block_num(p, i, v) put_unaligned_le32((v), (p) + (i))
1207
1208//
1209// in old version uniqueness field shows key type
1210//
1211#define V1_SD_UNIQUENESS 0
1212#define V1_INDIRECT_UNIQUENESS 0xfffffffe
1213#define V1_DIRECT_UNIQUENESS 0xffffffff
1214#define V1_DIRENTRY_UNIQUENESS 500
1215#define V1_ANY_UNIQUENESS 555 // FIXME: comment is required
1216
1217//
1218// here are conversion routines
1219//
1220static inline int uniqueness2type(__u32 uniqueness) CONSTF;
1221static inline int uniqueness2type(__u32 uniqueness)
1222{
1223 switch ((int)uniqueness) {
1224 case V1_SD_UNIQUENESS:
1225 return TYPE_STAT_DATA;
1226 case V1_INDIRECT_UNIQUENESS:
1227 return TYPE_INDIRECT;
1228 case V1_DIRECT_UNIQUENESS:
1229 return TYPE_DIRECT;
1230 case V1_DIRENTRY_UNIQUENESS:
1231 return TYPE_DIRENTRY;
1232 case V1_ANY_UNIQUENESS:
1233 default:
1234 return TYPE_ANY;
1235 }
1236}
1237
1238static inline __u32 type2uniqueness(int type) CONSTF;
1239static inline __u32 type2uniqueness(int type)
1240{
1241 switch (type) {
1242 case TYPE_STAT_DATA:
1243 return V1_SD_UNIQUENESS;
1244 case TYPE_INDIRECT:
1245 return V1_INDIRECT_UNIQUENESS;
1246 case TYPE_DIRECT:
1247 return V1_DIRECT_UNIQUENESS;
1248 case TYPE_DIRENTRY:
1249 return V1_DIRENTRY_UNIQUENESS;
1250 case TYPE_ANY:
1251 default:
1252 return V1_ANY_UNIQUENESS;
1253 }
1254}
1255
1256//
1257// key is pointer to on disk key which is stored in le, result is cpu,
1258// there is no way to get version of object from key, so, provide
1259// version to these defines
1260//
1261static inline loff_t le_key_k_offset(int version,
1262 const struct reiserfs_key *key)
1263{
1264 return (version == KEY_FORMAT_3_5) ?
1265 le32_to_cpu(key->u.k_offset_v1.k_offset) :
1266 offset_v2_k_offset(&(key->u.k_offset_v2));
1267}
1268
1269static inline loff_t le_ih_k_offset(const struct item_head *ih)
1270{
1271 return le_key_k_offset(ih_version(ih), &(ih->ih_key));
1272}
1273
1274static inline loff_t le_key_k_type(int version, const struct reiserfs_key *key)
1275{
1276 return (version == KEY_FORMAT_3_5) ?
1277 uniqueness2type(le32_to_cpu(key->u.k_offset_v1.k_uniqueness)) :
1278 offset_v2_k_type(&(key->u.k_offset_v2));
1279}
1280
1281static inline loff_t le_ih_k_type(const struct item_head *ih)
1282{
1283 return le_key_k_type(ih_version(ih), &(ih->ih_key));
1284}
1285
1286static inline void set_le_key_k_offset(int version, struct reiserfs_key *key,
1287 loff_t offset)
1288{
1289 (version == KEY_FORMAT_3_5) ? (void)(key->u.k_offset_v1.k_offset = cpu_to_le32(offset)) : /* jdm check */
1290 (void)(set_offset_v2_k_offset(&(key->u.k_offset_v2), offset));
1291}
1292
1293static inline void set_le_ih_k_offset(struct item_head *ih, loff_t offset)
1294{
1295 set_le_key_k_offset(ih_version(ih), &(ih->ih_key), offset);
1296}
1297
1298static inline void set_le_key_k_type(int version, struct reiserfs_key *key,
1299 int type)
1300{
1301 (version == KEY_FORMAT_3_5) ?
1302 (void)(key->u.k_offset_v1.k_uniqueness =
1303 cpu_to_le32(type2uniqueness(type)))
1304 : (void)(set_offset_v2_k_type(&(key->u.k_offset_v2), type));
1305}
1306
1307static inline void set_le_ih_k_type(struct item_head *ih, int type)
1308{
1309 set_le_key_k_type(ih_version(ih), &(ih->ih_key), type);
1310}
1311
1312static inline int is_direntry_le_key(int version, struct reiserfs_key *key)
1313{
1314 return le_key_k_type(version, key) == TYPE_DIRENTRY;
1315}
1316
1317static inline int is_direct_le_key(int version, struct reiserfs_key *key)
1318{
1319 return le_key_k_type(version, key) == TYPE_DIRECT;
1320}
1321
1322static inline int is_indirect_le_key(int version, struct reiserfs_key *key)
1323{
1324 return le_key_k_type(version, key) == TYPE_INDIRECT;
1325}
1326
1327static inline int is_statdata_le_key(int version, struct reiserfs_key *key)
1328{
1329 return le_key_k_type(version, key) == TYPE_STAT_DATA;
1330}
1331
1332//
1333// item header has version.
1334//
1335static inline int is_direntry_le_ih(struct item_head *ih)
1336{
1337 return is_direntry_le_key(ih_version(ih), &ih->ih_key);
1338}
1339
1340static inline int is_direct_le_ih(struct item_head *ih)
1341{
1342 return is_direct_le_key(ih_version(ih), &ih->ih_key);
1343}
1344
1345static inline int is_indirect_le_ih(struct item_head *ih)
1346{
1347 return is_indirect_le_key(ih_version(ih), &ih->ih_key);
1348}
1349
1350static inline int is_statdata_le_ih(struct item_head *ih)
1351{
1352 return is_statdata_le_key(ih_version(ih), &ih->ih_key);
1353}
1354
1355//
1356// key is pointer to cpu key, result is cpu
1357//
1358static inline loff_t cpu_key_k_offset(const struct cpu_key *key)
1359{
1360 return key->on_disk_key.k_offset;
1361}
1362
1363static inline loff_t cpu_key_k_type(const struct cpu_key *key)
1364{
1365 return key->on_disk_key.k_type;
1366}
1367
1368static inline void set_cpu_key_k_offset(struct cpu_key *key, loff_t offset)
1369{
1370 key->on_disk_key.k_offset = offset;
1371}
1372
1373static inline void set_cpu_key_k_type(struct cpu_key *key, int type)
1374{
1375 key->on_disk_key.k_type = type;
1376}
1377
1378static inline void cpu_key_k_offset_dec(struct cpu_key *key)
1379{
1380 key->on_disk_key.k_offset--;
1381}
1382
1383#define is_direntry_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRENTRY)
1384#define is_direct_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRECT)
1385#define is_indirect_cpu_key(key) (cpu_key_k_type (key) == TYPE_INDIRECT)
1386#define is_statdata_cpu_key(key) (cpu_key_k_type (key) == TYPE_STAT_DATA)
1387
1388/* are these used ? */
1389#define is_direntry_cpu_ih(ih) (is_direntry_cpu_key (&((ih)->ih_key)))
1390#define is_direct_cpu_ih(ih) (is_direct_cpu_key (&((ih)->ih_key)))
1391#define is_indirect_cpu_ih(ih) (is_indirect_cpu_key (&((ih)->ih_key)))
1392#define is_statdata_cpu_ih(ih) (is_statdata_cpu_key (&((ih)->ih_key)))
1393
1394#define I_K_KEY_IN_ITEM(ih, key, n_blocksize) \
1395 (!COMP_SHORT_KEYS(ih, key) && \
1396 I_OFF_BYTE_IN_ITEM(ih, k_offset(key), n_blocksize))
1397
1398/* maximal length of item */
1399#define MAX_ITEM_LEN(block_size) (block_size - BLKH_SIZE - IH_SIZE)
1400#define MIN_ITEM_LEN 1
1401
1402/* object identifier for root dir */
1403#define REISERFS_ROOT_OBJECTID 2
1404#define REISERFS_ROOT_PARENT_OBJECTID 1
1405
1406extern struct reiserfs_key root_key;
1407
1408/*
1409 * Picture represents a leaf of the S+tree
1410 * ______________________________________________________
1411 * | | Array of | | |
1412 * |Block | Object-Item | F r e e | Objects- |
1413 * | head | Headers | S p a c e | Items |
1414 * |______|_______________|___________________|___________|
1415 */
1416
1417/* Header of a disk block. More precisely, header of a formatted leaf
1418 or internal node, and not the header of an unformatted node. */
1419struct block_head {
1420 __le16 blk_level; /* Level of a block in the tree. */
1421 __le16 blk_nr_item; /* Number of keys/items in a block. */
1422 __le16 blk_free_space; /* Block free space in bytes. */
1423 __le16 blk_reserved;
1424 /* dump this in v4/planA */
1425 struct reiserfs_key blk_right_delim_key; /* kept only for compatibility */
1426};
1427
1428#define BLKH_SIZE (sizeof(struct block_head))
1429#define blkh_level(p_blkh) (le16_to_cpu((p_blkh)->blk_level))
1430#define blkh_nr_item(p_blkh) (le16_to_cpu((p_blkh)->blk_nr_item))
1431#define blkh_free_space(p_blkh) (le16_to_cpu((p_blkh)->blk_free_space))
1432#define blkh_reserved(p_blkh) (le16_to_cpu((p_blkh)->blk_reserved))
1433#define set_blkh_level(p_blkh,val) ((p_blkh)->blk_level = cpu_to_le16(val))
1434#define set_blkh_nr_item(p_blkh,val) ((p_blkh)->blk_nr_item = cpu_to_le16(val))
1435#define set_blkh_free_space(p_blkh,val) ((p_blkh)->blk_free_space = cpu_to_le16(val))
1436#define set_blkh_reserved(p_blkh,val) ((p_blkh)->blk_reserved = cpu_to_le16(val))
1437#define blkh_right_delim_key(p_blkh) ((p_blkh)->blk_right_delim_key)
1438#define set_blkh_right_delim_key(p_blkh,val) ((p_blkh)->blk_right_delim_key = val)
1439
1440/*
1441 * values for blk_level field of the struct block_head
1442 */
1443
1444#define FREE_LEVEL 0 /* when node gets removed from the tree its
1445 blk_level is set to FREE_LEVEL. It is then
1446 used to see whether the node is still in the
1447 tree */
1448
1449#define DISK_LEAF_NODE_LEVEL 1 /* Leaf node level. */
1450
1451/* Given the buffer head of a formatted node, resolve to the block head of that node. */
1452#define B_BLK_HEAD(bh) ((struct block_head *)((bh)->b_data))
1453/* Number of items that are in buffer. */
1454#define B_NR_ITEMS(bh) (blkh_nr_item(B_BLK_HEAD(bh)))
1455#define B_LEVEL(bh) (blkh_level(B_BLK_HEAD(bh)))
1456#define B_FREE_SPACE(bh) (blkh_free_space(B_BLK_HEAD(bh)))
1457
1458#define PUT_B_NR_ITEMS(bh, val) do { set_blkh_nr_item(B_BLK_HEAD(bh), val); } while (0)
1459#define PUT_B_LEVEL(bh, val) do { set_blkh_level(B_BLK_HEAD(bh), val); } while (0)
1460#define PUT_B_FREE_SPACE(bh, val) do { set_blkh_free_space(B_BLK_HEAD(bh), val); } while (0)
1461
1462/* Get right delimiting key. -- little endian */
1463#define B_PRIGHT_DELIM_KEY(bh) (&(blk_right_delim_key(B_BLK_HEAD(bh))))
1464
1465/* Does the buffer contain a disk leaf. */
1466#define B_IS_ITEMS_LEVEL(bh) (B_LEVEL(bh) == DISK_LEAF_NODE_LEVEL)
1467
1468/* Does the buffer contain a disk internal node */
1469#define B_IS_KEYS_LEVEL(bh) (B_LEVEL(bh) > DISK_LEAF_NODE_LEVEL \
1470 && B_LEVEL(bh) <= MAX_HEIGHT)
1471
1472/***************************************************************************/
1473/* STAT DATA */
1474/***************************************************************************/
1475
1476//
1477// old stat data is 32 bytes long. We are going to distinguish new one by
1478// different size
1479//
1480struct stat_data_v1 {
1481 __le16 sd_mode; /* file type, permissions */
1482 __le16 sd_nlink; /* number of hard links */
1483 __le16 sd_uid; /* owner */
1484 __le16 sd_gid; /* group */
1485 __le32 sd_size; /* file size */
1486 __le32 sd_atime; /* time of last access */
1487 __le32 sd_mtime; /* time file was last modified */
1488 __le32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
1489 union {
1490 __le32 sd_rdev;
1491 __le32 sd_blocks; /* number of blocks file uses */
1492 } __attribute__ ((__packed__)) u;
1493 __le32 sd_first_direct_byte; /* first byte of file which is stored
1494 in a direct item: except that if it
1495 equals 1 it is a symlink and if it
1496 equals ~(__u32)0 there is no
1497 direct item. The existence of this
1498 field really grates on me. Let's
1499 replace it with a macro based on
1500 sd_size and our tail suppression
1501 policy. Someday. -Hans */
1502} __attribute__ ((__packed__));
1503
1504#define SD_V1_SIZE (sizeof(struct stat_data_v1))
1505#define stat_data_v1(ih) (ih_version (ih) == KEY_FORMAT_3_5)
1506#define sd_v1_mode(sdp) (le16_to_cpu((sdp)->sd_mode))
1507#define set_sd_v1_mode(sdp,v) ((sdp)->sd_mode = cpu_to_le16(v))
1508#define sd_v1_nlink(sdp) (le16_to_cpu((sdp)->sd_nlink))
1509#define set_sd_v1_nlink(sdp,v) ((sdp)->sd_nlink = cpu_to_le16(v))
1510#define sd_v1_uid(sdp) (le16_to_cpu((sdp)->sd_uid))
1511#define set_sd_v1_uid(sdp,v) ((sdp)->sd_uid = cpu_to_le16(v))
1512#define sd_v1_gid(sdp) (le16_to_cpu((sdp)->sd_gid))
1513#define set_sd_v1_gid(sdp,v) ((sdp)->sd_gid = cpu_to_le16(v))
1514#define sd_v1_size(sdp) (le32_to_cpu((sdp)->sd_size))
1515#define set_sd_v1_size(sdp,v) ((sdp)->sd_size = cpu_to_le32(v))
1516#define sd_v1_atime(sdp) (le32_to_cpu((sdp)->sd_atime))
1517#define set_sd_v1_atime(sdp,v) ((sdp)->sd_atime = cpu_to_le32(v))
1518#define sd_v1_mtime(sdp) (le32_to_cpu((sdp)->sd_mtime))
1519#define set_sd_v1_mtime(sdp,v) ((sdp)->sd_mtime = cpu_to_le32(v))
1520#define sd_v1_ctime(sdp) (le32_to_cpu((sdp)->sd_ctime))
1521#define set_sd_v1_ctime(sdp,v) ((sdp)->sd_ctime = cpu_to_le32(v))
1522#define sd_v1_rdev(sdp) (le32_to_cpu((sdp)->u.sd_rdev))
1523#define set_sd_v1_rdev(sdp,v) ((sdp)->u.sd_rdev = cpu_to_le32(v))
1524#define sd_v1_blocks(sdp) (le32_to_cpu((sdp)->u.sd_blocks))
1525#define set_sd_v1_blocks(sdp,v) ((sdp)->u.sd_blocks = cpu_to_le32(v))
1526#define sd_v1_first_direct_byte(sdp) \
1527 (le32_to_cpu((sdp)->sd_first_direct_byte))
1528#define set_sd_v1_first_direct_byte(sdp,v) \
1529 ((sdp)->sd_first_direct_byte = cpu_to_le32(v))
1530
1531/* inode flags stored in sd_attrs (nee sd_reserved) */
1532
1533/* we want common flags to have the same values as in ext2,
1534 so chattr(1) will work without problems */
1535#define REISERFS_IMMUTABLE_FL FS_IMMUTABLE_FL
1536#define REISERFS_APPEND_FL FS_APPEND_FL
1537#define REISERFS_SYNC_FL FS_SYNC_FL
1538#define REISERFS_NOATIME_FL FS_NOATIME_FL
1539#define REISERFS_NODUMP_FL FS_NODUMP_FL
1540#define REISERFS_SECRM_FL FS_SECRM_FL
1541#define REISERFS_UNRM_FL FS_UNRM_FL
1542#define REISERFS_COMPR_FL FS_COMPR_FL
1543#define REISERFS_NOTAIL_FL FS_NOTAIL_FL
1544
1545/* persistent flags that file inherits from the parent directory */
1546#define REISERFS_INHERIT_MASK ( REISERFS_IMMUTABLE_FL | \
1547 REISERFS_SYNC_FL | \
1548 REISERFS_NOATIME_FL | \
1549 REISERFS_NODUMP_FL | \
1550 REISERFS_SECRM_FL | \
1551 REISERFS_COMPR_FL | \
1552 REISERFS_NOTAIL_FL )
1553
1554/* Stat Data on disk (reiserfs version of UFS disk inode minus the
1555 address blocks) */
1556struct stat_data {
1557 __le16 sd_mode; /* file type, permissions */
1558 __le16 sd_attrs; /* persistent inode flags */
1559 __le32 sd_nlink; /* number of hard links */
1560 __le64 sd_size; /* file size */
1561 __le32 sd_uid; /* owner */
1562 __le32 sd_gid; /* group */
1563 __le32 sd_atime; /* time of last access */
1564 __le32 sd_mtime; /* time file was last modified */
1565 __le32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */
1566 __le32 sd_blocks;
1567 union {
1568 __le32 sd_rdev;
1569 __le32 sd_generation;
1570 //__le32 sd_first_direct_byte;
1571 /* first byte of file which is stored in a
1572 direct item: except that if it equals 1
1573 it is a symlink and if it equals
1574 ~(__u32)0 there is no direct item. The
1575 existence of this field really grates
1576 on me. Let's replace it with a macro
1577 based on sd_size and our tail
1578 suppression policy? */
1579 } __attribute__ ((__packed__)) u;
1580} __attribute__ ((__packed__));
1581//
1582// this is 44 bytes long
1583//
1584#define SD_SIZE (sizeof(struct stat_data))
1585#define SD_V2_SIZE SD_SIZE
1586#define stat_data_v2(ih) (ih_version (ih) == KEY_FORMAT_3_6)
1587#define sd_v2_mode(sdp) (le16_to_cpu((sdp)->sd_mode))
1588#define set_sd_v2_mode(sdp,v) ((sdp)->sd_mode = cpu_to_le16(v))
1589/* sd_reserved */
1590/* set_sd_reserved */
1591#define sd_v2_nlink(sdp) (le32_to_cpu((sdp)->sd_nlink))
1592#define set_sd_v2_nlink(sdp,v) ((sdp)->sd_nlink = cpu_to_le32(v))
1593#define sd_v2_size(sdp) (le64_to_cpu((sdp)->sd_size))
1594#define set_sd_v2_size(sdp,v) ((sdp)->sd_size = cpu_to_le64(v))
1595#define sd_v2_uid(sdp) (le32_to_cpu((sdp)->sd_uid))
1596#define set_sd_v2_uid(sdp,v) ((sdp)->sd_uid = cpu_to_le32(v))
1597#define sd_v2_gid(sdp) (le32_to_cpu((sdp)->sd_gid))
1598#define set_sd_v2_gid(sdp,v) ((sdp)->sd_gid = cpu_to_le32(v))
1599#define sd_v2_atime(sdp) (le32_to_cpu((sdp)->sd_atime))
1600#define set_sd_v2_atime(sdp,v) ((sdp)->sd_atime = cpu_to_le32(v))
1601#define sd_v2_mtime(sdp) (le32_to_cpu((sdp)->sd_mtime))
1602#define set_sd_v2_mtime(sdp,v) ((sdp)->sd_mtime = cpu_to_le32(v))
1603#define sd_v2_ctime(sdp) (le32_to_cpu((sdp)->sd_ctime))
1604#define set_sd_v2_ctime(sdp,v) ((sdp)->sd_ctime = cpu_to_le32(v))
1605#define sd_v2_blocks(sdp) (le32_to_cpu((sdp)->sd_blocks))
1606#define set_sd_v2_blocks(sdp,v) ((sdp)->sd_blocks = cpu_to_le32(v))
1607#define sd_v2_rdev(sdp) (le32_to_cpu((sdp)->u.sd_rdev))
1608#define set_sd_v2_rdev(sdp,v) ((sdp)->u.sd_rdev = cpu_to_le32(v))
1609#define sd_v2_generation(sdp) (le32_to_cpu((sdp)->u.sd_generation))
1610#define set_sd_v2_generation(sdp,v) ((sdp)->u.sd_generation = cpu_to_le32(v))
1611#define sd_v2_attrs(sdp) (le16_to_cpu((sdp)->sd_attrs))
1612#define set_sd_v2_attrs(sdp,v) ((sdp)->sd_attrs = cpu_to_le16(v))
1613
1614/***************************************************************************/
1615/* DIRECTORY STRUCTURE */
1616/***************************************************************************/
1617/*
1618 Picture represents the structure of directory items
1619 ________________________________________________
1620 | Array of | | | | | |
1621 | directory |N-1| N-2 | .... | 1st |0th|
1622 | entry headers | | | | | |
1623 |_______________|___|_____|________|_______|___|
1624 <---- directory entries ------>
1625
1626 First directory item has k_offset component 1. We store "." and ".."
1627 in one item, always, we never split "." and ".." into differing
1628 items. This makes, among other things, the code for removing
1629 directories simpler. */
1630#define SD_OFFSET 0
1631#define SD_UNIQUENESS 0
1632#define DOT_OFFSET 1
1633#define DOT_DOT_OFFSET 2
1634#define DIRENTRY_UNIQUENESS 500
1635
1636/* */
1637#define FIRST_ITEM_OFFSET 1
1638
1639/*
1640 Q: How to get key of object pointed to by entry from entry?
1641
1642 A: Each directory entry has its header. This header has deh_dir_id and deh_objectid fields, those are key
1643 of object, entry points to */
1644
1645/* NOT IMPLEMENTED:
1646 Directory will someday contain stat data of object */
1647
1648struct reiserfs_de_head {
1649 __le32 deh_offset; /* third component of the directory entry key */
1650 __le32 deh_dir_id; /* objectid of the parent directory of the object, that is referenced
1651 by directory entry */
1652 __le32 deh_objectid; /* objectid of the object, that is referenced by directory entry */
1653 __le16 deh_location; /* offset of name in the whole item */
1654 __le16 deh_state; /* whether 1) entry contains stat data (for future), and 2) whether
1655 entry is hidden (unlinked) */
1656} __attribute__ ((__packed__));
1657#define DEH_SIZE sizeof(struct reiserfs_de_head)
1658#define deh_offset(p_deh) (le32_to_cpu((p_deh)->deh_offset))
1659#define deh_dir_id(p_deh) (le32_to_cpu((p_deh)->deh_dir_id))
1660#define deh_objectid(p_deh) (le32_to_cpu((p_deh)->deh_objectid))
1661#define deh_location(p_deh) (le16_to_cpu((p_deh)->deh_location))
1662#define deh_state(p_deh) (le16_to_cpu((p_deh)->deh_state))
1663
1664#define put_deh_offset(p_deh,v) ((p_deh)->deh_offset = cpu_to_le32((v)))
1665#define put_deh_dir_id(p_deh,v) ((p_deh)->deh_dir_id = cpu_to_le32((v)))
1666#define put_deh_objectid(p_deh,v) ((p_deh)->deh_objectid = cpu_to_le32((v)))
1667#define put_deh_location(p_deh,v) ((p_deh)->deh_location = cpu_to_le16((v)))
1668#define put_deh_state(p_deh,v) ((p_deh)->deh_state = cpu_to_le16((v)))
1669
1670/* empty directory contains two entries "." and ".." and their headers */
1671#define EMPTY_DIR_SIZE \
1672(DEH_SIZE * 2 + ROUND_UP (strlen (".")) + ROUND_UP (strlen ("..")))
1673
1674/* old format directories have this size when empty */
1675#define EMPTY_DIR_SIZE_V1 (DEH_SIZE * 2 + 3)
1676
1677#define DEH_Statdata 0 /* not used now */
1678#define DEH_Visible 2
1679
1680/* 64 bit systems (and the S/390) need to be aligned explicitly -jdm */
1681#if BITS_PER_LONG == 64 || defined(__s390__) || defined(__hppa__)
1682# define ADDR_UNALIGNED_BITS (3)
1683#endif
1684
1685/* These are only used to manipulate deh_state.
1686 * Because of this, we'll use the ext2_ bit routines,
1687 * since they are little endian */
1688#ifdef ADDR_UNALIGNED_BITS
1689
1690# define aligned_address(addr) ((void *)((long)(addr) & ~((1UL << ADDR_UNALIGNED_BITS) - 1)))
1691# define unaligned_offset(addr) (((int)((long)(addr) & ((1 << ADDR_UNALIGNED_BITS) - 1))) << 3)
1692
1693# define set_bit_unaligned(nr, addr) \
1694 __test_and_set_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))
1695# define clear_bit_unaligned(nr, addr) \
1696 __test_and_clear_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))
1697# define test_bit_unaligned(nr, addr) \
1698 test_bit_le((nr) + unaligned_offset(addr), aligned_address(addr))
1699
1700#else
1701
1702# define set_bit_unaligned(nr, addr) __test_and_set_bit_le(nr, addr)
1703# define clear_bit_unaligned(nr, addr) __test_and_clear_bit_le(nr, addr)
1704# define test_bit_unaligned(nr, addr) test_bit_le(nr, addr)
1705
1706#endif
1707
1708#define mark_de_with_sd(deh) set_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
1709#define mark_de_without_sd(deh) clear_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
1710#define mark_de_visible(deh) set_bit_unaligned (DEH_Visible, &((deh)->deh_state))
1711#define mark_de_hidden(deh) clear_bit_unaligned (DEH_Visible, &((deh)->deh_state))
1712
1713#define de_with_sd(deh) test_bit_unaligned (DEH_Statdata, &((deh)->deh_state))
1714#define de_visible(deh) test_bit_unaligned (DEH_Visible, &((deh)->deh_state))
1715#define de_hidden(deh) !test_bit_unaligned (DEH_Visible, &((deh)->deh_state))
1716
1717extern void make_empty_dir_item_v1(char *body, __le32 dirid, __le32 objid,
1718 __le32 par_dirid, __le32 par_objid);
1719extern void make_empty_dir_item(char *body, __le32 dirid, __le32 objid,
1720 __le32 par_dirid, __le32 par_objid);
1721
1722/* array of the entry headers */
1723 /* get item body */
1724#define B_I_PITEM(bh,ih) ( (bh)->b_data + ih_location(ih) )
1725#define B_I_DEH(bh,ih) ((struct reiserfs_de_head *)(B_I_PITEM(bh,ih)))
1726
1727/* length of the directory entry in directory item. This define
1728 calculates length of i-th directory entry using directory entry
1729 locations from dir entry head. When it calculates length of 0-th
1730 directory entry, it uses length of whole item in place of entry
1731 location of the non-existent following entry in the calculation.
1732 See picture above.*/
1733/*
1734#define I_DEH_N_ENTRY_LENGTH(ih,deh,i) \
1735((i) ? (deh_location((deh)-1) - deh_location((deh))) : (ih_item_len((ih)) - deh_location((deh))))
1736*/
1737static inline int entry_length(const struct buffer_head *bh,
1738 const struct item_head *ih, int pos_in_item)
1739{
1740 struct reiserfs_de_head *deh;
1741
1742 deh = B_I_DEH(bh, ih) + pos_in_item;
1743 if (pos_in_item)
1744 return deh_location(deh - 1) - deh_location(deh);
1745
1746 return ih_item_len(ih) - deh_location(deh);
1747}
1748
1749/* number of entries in the directory item, depends on ENTRY_COUNT being at the start of directory dynamic data. */
1750#define I_ENTRY_COUNT(ih) (ih_entry_count((ih)))
1751
1752/* name by bh, ih and entry_num */
1753#define B_I_E_NAME(bh,ih,entry_num) ((char *)(bh->b_data + ih_location(ih) + deh_location(B_I_DEH(bh,ih)+(entry_num))))
1754
1755// two entries per block (at least)
1756#define REISERFS_MAX_NAME(block_size) 255
1757
1758/* this structure is used for operations on directory entries. It is
1759 not a disk structure. */
1760/* When reiserfs_find_entry or search_by_entry_key find directory
1761 entry, they return filled reiserfs_dir_entry structure */
1762struct reiserfs_dir_entry {
1763 struct buffer_head *de_bh;
1764 int de_item_num;
1765 struct item_head *de_ih;
1766 int de_entry_num;
1767 struct reiserfs_de_head *de_deh;
1768 int de_entrylen;
1769 int de_namelen;
1770 char *de_name;
1771 unsigned long *de_gen_number_bit_string;
1772
1773 __u32 de_dir_id;
1774 __u32 de_objectid;
1775
1776 struct cpu_key de_entry_key;
1777};
1778
1779/* these defines are useful when a particular member of a reiserfs_dir_entry is needed */
1780
1781/* pointer to file name, stored in entry */
1782#define B_I_DEH_ENTRY_FILE_NAME(bh,ih,deh) (B_I_PITEM (bh, ih) + deh_location(deh))
1783
1784/* length of name */
1785#define I_DEH_N_ENTRY_FILE_NAME_LENGTH(ih,deh,entry_num) \
1786(I_DEH_N_ENTRY_LENGTH (ih, deh, entry_num) - (de_with_sd (deh) ? SD_SIZE : 0))
1787
1788/* hash value occupies bits from 7 up to 30 */
1789#define GET_HASH_VALUE(offset) ((offset) & 0x7fffff80LL)
1790/* generation number occupies 7 bits starting from 0 up to 6 */
1791#define GET_GENERATION_NUMBER(offset) ((offset) & 0x7fLL)
1792#define MAX_GENERATION_NUMBER 127
1793
1794#define SET_GENERATION_NUMBER(offset,gen_number) (GET_HASH_VALUE(offset)|(gen_number))
1795
1796/*
1797 * Picture represents an internal node of the reiserfs tree
1798 * ______________________________________________________
1799 * | | Array of | Array of | Free |
1800 * |block | keys | pointers | space |
1801 * | head | N | N+1 | |
1802 * |______|_______________|___________________|___________|
1803 */
1804
1805/***************************************************************************/
1806/* DISK CHILD */
1807/***************************************************************************/
1808/* Disk child pointer: The pointer from an internal node of the tree
1809 to a node that is on disk. */
1810struct disk_child {
1811 __le32 dc_block_number; /* Disk child's block number. */
1812 __le16 dc_size; /* Disk child's used space. */
1813 __le16 dc_reserved;
1814};
1815
1816#define DC_SIZE (sizeof(struct disk_child))
1817#define dc_block_number(dc_p) (le32_to_cpu((dc_p)->dc_block_number))
1818#define dc_size(dc_p) (le16_to_cpu((dc_p)->dc_size))
1819#define put_dc_block_number(dc_p, val) do { (dc_p)->dc_block_number = cpu_to_le32(val); } while(0)
1820#define put_dc_size(dc_p, val) do { (dc_p)->dc_size = cpu_to_le16(val); } while(0)
1821
1822/* Get disk child by buffer header and position in the tree node. */
1823#define B_N_CHILD(bh, n_pos) ((struct disk_child *)\
1824((bh)->b_data + BLKH_SIZE + B_NR_ITEMS(bh) * KEY_SIZE + DC_SIZE * (n_pos)))
1825
1826/* Get disk child number by buffer header and position in the tree node. */
1827#define B_N_CHILD_NUM(bh, n_pos) (dc_block_number(B_N_CHILD(bh, n_pos)))
1828#define PUT_B_N_CHILD_NUM(bh, n_pos, val) \
1829 (put_dc_block_number(B_N_CHILD(bh, n_pos), val))
1830
1831 /* maximal value of field child_size in structure disk_child */
1832 /* child size is the combined size of all items and their headers */
1833#define MAX_CHILD_SIZE(bh) ((int)( (bh)->b_size - BLKH_SIZE ))
1834
1835/* amount of used space in buffer (not including block head) */
1836#define B_CHILD_SIZE(cur) (MAX_CHILD_SIZE(cur)-(B_FREE_SPACE(cur)))
1837
1838/* max and min number of keys in internal node */
1839#define MAX_NR_KEY(bh) ( (MAX_CHILD_SIZE(bh)-DC_SIZE)/(KEY_SIZE+DC_SIZE) )
1840#define MIN_NR_KEY(bh) (MAX_NR_KEY(bh)/2)
1841
1842/***************************************************************************/
1843/* PATH STRUCTURES AND DEFINES */
1844/***************************************************************************/
1845
1846/* Search_by_key fills up the path from the root to the leaf as it descends the tree looking for the
1847 key. It uses reiserfs_bread to try to find buffers in the cache given their block number. If it
1848 does not find them in the cache it reads them from disk. For each node search_by_key finds using
1849 reiserfs_bread it then uses bin_search to look through that node. bin_search will find the
1850 position of the block_number of the next node if it is looking through an internal node. If it
1851 is looking through a leaf node bin_search will find the position of the item which has key either
1852 equal to given key, or which is the maximal key less than the given key. */
1853
1854struct path_element {
1855 struct buffer_head *pe_buffer; /* Pointer to the buffer at the path in the tree. */
1856 int pe_position; /* Position in the tree node which is placed in the */
1857 /* buffer above. */
1858};
1859
1860#define MAX_HEIGHT 5 /* maximal height of a tree. don't change this without changing JOURNAL_PER_BALANCE_CNT */
1861#define EXTENDED_MAX_HEIGHT 7 /* Must be equals MAX_HEIGHT + FIRST_PATH_ELEMENT_OFFSET */
1862#define FIRST_PATH_ELEMENT_OFFSET 2 /* Must be equal to at least 2. */
1863
1864#define ILLEGAL_PATH_ELEMENT_OFFSET 1 /* Must be equal to FIRST_PATH_ELEMENT_OFFSET - 1 */
1865#define MAX_FEB_SIZE 6 /* this MUST be MAX_HEIGHT + 1. See about FEB below */
1866
1867/* We need to keep track of who the ancestors of nodes are. When we
1868 perform a search we record which nodes were visited while
1869 descending the tree looking for the node we searched for. This list
1870 of nodes is called the path. This information is used while
1871 performing balancing. Note that this path information may become
1872 invalid, and this means we must check it when using it to see if it
1873 is still valid. You'll need to read search_by_key and the comments
1874 in it, especially about decrement_counters_in_path(), to understand
1875 this structure.
1876
1877Paths make the code so much harder to work with and debug.... An
1878enormous number of bugs are due to them, and trying to write or modify
1879code that uses them just makes my head hurt. They are based on an
1880excessive effort to avoid disturbing the precious VFS code.:-( The
1881gods only know how we are going to SMP the code that uses them.
1882znodes are the way! */
1883
1884#define PATH_READA 0x1 /* do read ahead */
1885#define PATH_READA_BACK 0x2 /* read backwards */
1886
1887struct treepath {
1888 int path_length; /* Length of the array above. */
1889 int reada;
1890 struct path_element path_elements[EXTENDED_MAX_HEIGHT]; /* Array of the path elements. */
1891 int pos_in_item;
1892};
1893
1894#define pos_in_item(path) ((path)->pos_in_item)
1895
1896#define INITIALIZE_PATH(var) \
1897struct treepath var = {.path_length = ILLEGAL_PATH_ELEMENT_OFFSET, .reada = 0,}
1898
1899/* Get path element by path and path position. */
1900#define PATH_OFFSET_PELEMENT(path, n_offset) ((path)->path_elements + (n_offset))
1901
1902/* Get buffer header at the path by path and path position. */
1903#define PATH_OFFSET_PBUFFER(path, n_offset) (PATH_OFFSET_PELEMENT(path, n_offset)->pe_buffer)
1904
1905/* Get position in the element at the path by path and path position. */
1906#define PATH_OFFSET_POSITION(path, n_offset) (PATH_OFFSET_PELEMENT(path, n_offset)->pe_position)
1907
1908#define PATH_PLAST_BUFFER(path) (PATH_OFFSET_PBUFFER((path), (path)->path_length))
1909 /* you know, to the person who didn't
1910 write this the macro name does not
1911 at first suggest what it does.
1912 Maybe POSITION_FROM_PATH_END? Or
1913 maybe we should just focus on
1914 dumping paths... -Hans */
1915#define PATH_LAST_POSITION(path) (PATH_OFFSET_POSITION((path), (path)->path_length))
1916
1917#define PATH_PITEM_HEAD(path) B_N_PITEM_HEAD(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION(path))
1918
1919/* in do_balance leaf has h == 0 in contrast with path structure,
1920 where root has level == 0. That is why we need these defines */
1921#define PATH_H_PBUFFER(path, h) PATH_OFFSET_PBUFFER (path, path->path_length - (h)) /* tb->S[h] */
1922#define PATH_H_PPARENT(path, h) PATH_H_PBUFFER (path, (h) + 1) /* tb->F[h] or tb->S[0]->b_parent */
1923#define PATH_H_POSITION(path, h) PATH_OFFSET_POSITION (path, path->path_length - (h))
1924#define PATH_H_B_ITEM_ORDER(path, h) PATH_H_POSITION(path, h + 1) /* tb->S[h]->b_item_order */
1925
1926#define PATH_H_PATH_OFFSET(path, n_h) ((path)->path_length - (n_h))
1927
1928#define get_last_bh(path) PATH_PLAST_BUFFER(path)
1929#define get_ih(path) PATH_PITEM_HEAD(path)
1930#define get_item_pos(path) PATH_LAST_POSITION(path)
1931#define get_item(path) ((void *)B_N_PITEM(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION (path)))
1932#define item_moved(ih,path) comp_items(ih, path)
1933#define path_changed(ih,path) comp_items (ih, path)
1934
1935/***************************************************************************/
1936/* MISC */
1937/***************************************************************************/
1938
1939/* Size of pointer to the unformatted node. */
1940#define UNFM_P_SIZE (sizeof(unp_t))
1941#define UNFM_P_SHIFT 2
1942
1943// in in-core inode key is stored on le form
1944#define INODE_PKEY(inode) ((struct reiserfs_key *)(REISERFS_I(inode)->i_key))
1945
1946#define MAX_UL_INT 0xffffffff
1947#define MAX_INT 0x7ffffff
1948#define MAX_US_INT 0xffff
1949
1950// reiserfs version 2 has max offset 60 bits. Version 1 - 32 bit offset
1951#define U32_MAX (~(__u32)0)
1952
1953static inline loff_t max_reiserfs_offset(struct inode *inode)
1954{
1955 if (get_inode_item_key_version(inode) == KEY_FORMAT_3_5)
1956 return (loff_t) U32_MAX;
1957
1958 return (loff_t) ((~(__u64) 0) >> 4);
1959}
1960
1961/*#define MAX_KEY_UNIQUENESS MAX_UL_INT*/
1962#define MAX_KEY_OBJECTID MAX_UL_INT
1963
1964#define MAX_B_NUM MAX_UL_INT
1965#define MAX_FC_NUM MAX_US_INT
1966
1967/* the purpose is to detect overflow of an unsigned short */
1968#define REISERFS_LINK_MAX (MAX_US_INT - 1000)
1969
1970/* The following defines are used in reiserfs_insert_item and reiserfs_append_item */
1971#define REISERFS_KERNEL_MEM 0 /* reiserfs kernel memory mode */
1972#define REISERFS_USER_MEM 1 /* reiserfs user memory mode */
1973
1974#define fs_generation(s) (REISERFS_SB(s)->s_generation_counter)
1975#define get_generation(s) atomic_read (&fs_generation(s))
1976#define FILESYSTEM_CHANGED_TB(tb) (get_generation((tb)->tb_sb) != (tb)->fs_gen)
1977#define __fs_changed(gen,s) (gen != get_generation (s))
1978#define fs_changed(gen,s) \
1979({ \
1980 reiserfs_cond_resched(s); \
1981 __fs_changed(gen, s); \
1982})
1983
1984/***************************************************************************/
1985/* FIXATE NODES */
1986/***************************************************************************/
1987
1988#define VI_TYPE_LEFT_MERGEABLE 1
1989#define VI_TYPE_RIGHT_MERGEABLE 2
1990
1991/* To make any changes in the tree we always first find node, that
1992 contains item to be changed/deleted or place to insert a new
1993 item. We call this node S. To do balancing we need to decide what
1994 we will shift to left/right neighbor, or to a new node, where new
1995 item will be etc. To make this analysis simpler we build virtual
1996 node. Virtual node is an array of items, that will replace items of
1997 node S. (For instance if we are going to delete an item, virtual
1998 node does not contain it). Virtual node keeps information about
1999 item sizes and types, mergeability of first and last items, sizes
2000 of all entries in directory item. We use this array of items when
2001 calculating what we can shift to neighbors and how many nodes we
2002 have to have if we do not any shiftings, if we shift to left/right
2003 neighbor or to both. */
2004struct virtual_item {
2005 int vi_index; // index in the array of item operations
2006 unsigned short vi_type; // left/right mergeability
2007 unsigned short vi_item_len; /* length of item that it will have after balancing */
2008 struct item_head *vi_ih;
2009 const char *vi_item; // body of item (old or new)
2010 const void *vi_new_data; // 0 always but paste mode
2011 void *vi_uarea; // item specific area
2012};
2013
2014struct virtual_node {
2015 char *vn_free_ptr; /* this is a pointer to the free space in the buffer */
2016 unsigned short vn_nr_item; /* number of items in virtual node */
2017 short vn_size; /* size of node , that node would have if it has unlimited size and no balancing is performed */
2018 short vn_mode; /* mode of balancing (paste, insert, delete, cut) */
2019 short vn_affected_item_num;
2020 short vn_pos_in_item;
2021 struct item_head *vn_ins_ih; /* item header of inserted item, 0 for other modes */
2022 const void *vn_data;
2023 struct virtual_item *vn_vi; /* array of items (including a new one, excluding item to be deleted) */
2024};
2025
2026/* used by directory items when creating virtual nodes */
2027struct direntry_uarea {
2028 int flags;
2029 __u16 entry_count;
2030 __u16 entry_sizes[1];
2031} __attribute__ ((__packed__));
2032
2033/***************************************************************************/
2034/* TREE BALANCE */
2035/***************************************************************************/
2036
2037/* This temporary structure is used in tree balance algorithms, and
2038 constructed as we go to the extent that its various parts are
2039 needed. It contains arrays of nodes that can potentially be
2040 involved in the balancing of node S, and parameters that define how
2041 each of the nodes must be balanced. Note that in these algorithms
2042 for balancing the worst case is to need to balance the current node
2043 S and the left and right neighbors and all of their parents plus
2044 create a new node. We implement S1 balancing for the leaf nodes
2045 and S0 balancing for the internal nodes (S1 and S0 are defined in
2046 our papers.)*/
2047
2048#define MAX_FREE_BLOCK 7 /* size of the array of buffers to free at end of do_balance */
2049
2050/* maximum number of FEB blocknrs on a single level */
2051#define MAX_AMOUNT_NEEDED 2
2052
2053/* someday somebody will prefix every field in this struct with tb_ */
2054struct tree_balance {
2055 int tb_mode;
2056 int need_balance_dirty;
2057 struct super_block *tb_sb;
2058 struct reiserfs_transaction_handle *transaction_handle;
2059 struct treepath *tb_path;
2060 struct buffer_head *L[MAX_HEIGHT]; /* array of left neighbors of nodes in the path */
2061 struct buffer_head *R[MAX_HEIGHT]; /* array of right neighbors of nodes in the path */
2062 struct buffer_head *FL[MAX_HEIGHT]; /* array of fathers of the left neighbors */
2063 struct buffer_head *FR[MAX_HEIGHT]; /* array of fathers of the right neighbors */
2064 struct buffer_head *CFL[MAX_HEIGHT]; /* array of common parents of center node and its left neighbor */
2065 struct buffer_head *CFR[MAX_HEIGHT]; /* array of common parents of center node and its right neighbor */
2066
2067 struct buffer_head *FEB[MAX_FEB_SIZE]; /* array of empty buffers. Number of buffers in array equals
2068 cur_blknum. */
2069 struct buffer_head *used[MAX_FEB_SIZE];
2070 struct buffer_head *thrown[MAX_FEB_SIZE];
2071 int lnum[MAX_HEIGHT]; /* array of number of items which must be
2072 shifted to the left in order to balance the
2073 current node; for leaves includes item that
2074 will be partially shifted; for internal
2075 nodes, it is the number of child pointers
2076 rather than items. It includes the new item
2077 being created. The code sometimes subtracts
2078 one to get the number of wholly shifted
2079 items for other purposes. */
2080 int rnum[MAX_HEIGHT]; /* substitute right for left in comment above */
2081 int lkey[MAX_HEIGHT]; /* array indexed by height h mapping the key delimiting L[h] and
2082 S[h] to its item number within the node CFL[h] */
2083 int rkey[MAX_HEIGHT]; /* substitute r for l in comment above */
2084 int insert_size[MAX_HEIGHT]; /* the number of bytes by we are trying to add or remove from
2085 S[h]. A negative value means removing. */
2086 int blknum[MAX_HEIGHT]; /* number of nodes that will replace node S[h] after
2087 balancing on the level h of the tree. If 0 then S is
2088 being deleted, if 1 then S is remaining and no new nodes
2089 are being created, if 2 or 3 then 1 or 2 new nodes is
2090 being created */
2091
2092 /* fields that are used only for balancing leaves of the tree */
2093 int cur_blknum; /* number of empty blocks having been already allocated */
2094 int s0num; /* number of items that fall into left most node when S[0] splits */
2095 int s1num; /* number of items that fall into first new node when S[0] splits */
2096 int s2num; /* number of items that fall into second new node when S[0] splits */
2097 int lbytes; /* number of bytes which can flow to the left neighbor from the left */
2098 /* most liquid item that cannot be shifted from S[0] entirely */
2099 /* if -1 then nothing will be partially shifted */
2100 int rbytes; /* number of bytes which will flow to the right neighbor from the right */
2101 /* most liquid item that cannot be shifted from S[0] entirely */
2102 /* if -1 then nothing will be partially shifted */
2103 int s1bytes; /* number of bytes which flow to the first new node when S[0] splits */
2104 /* note: if S[0] splits into 3 nodes, then items do not need to be cut */
2105 int s2bytes;
2106 struct buffer_head *buf_to_free[MAX_FREE_BLOCK]; /* buffers which are to be freed after do_balance finishes by unfix_nodes */
2107 char *vn_buf; /* kmalloced memory. Used to create
2108 virtual node and keep map of
2109 dirtied bitmap blocks */
2110 int vn_buf_size; /* size of the vn_buf */
2111 struct virtual_node *tb_vn; /* VN starts after bitmap of bitmap blocks */
2112
2113 int fs_gen; /* saved value of `reiserfs_generation' counter
2114 see FILESYSTEM_CHANGED() macro in reiserfs_fs.h */
2115#ifdef DISPLACE_NEW_PACKING_LOCALITIES
2116 struct in_core_key key; /* key pointer, to pass to block allocator or
2117 another low-level subsystem */
2118#endif
2119};
2120
2121/* These are modes of balancing */
2122
2123/* When inserting an item. */
2124#define M_INSERT 'i'
2125/* When inserting into (directories only) or appending onto an already
2126 existent item. */
2127#define M_PASTE 'p'
2128/* When deleting an item. */
2129#define M_DELETE 'd'
2130/* When truncating an item or removing an entry from a (directory) item. */
2131#define M_CUT 'c'
2132
2133/* used when balancing on leaf level skipped (in reiserfsck) */
2134#define M_INTERNAL 'n'
2135
2136/* When further balancing is not needed, then do_balance does not need
2137 to be called. */
2138#define M_SKIP_BALANCING 's'
2139#define M_CONVERT 'v'
2140
2141/* modes of leaf_move_items */
2142#define LEAF_FROM_S_TO_L 0
2143#define LEAF_FROM_S_TO_R 1
2144#define LEAF_FROM_R_TO_L 2
2145#define LEAF_FROM_L_TO_R 3
2146#define LEAF_FROM_S_TO_SNEW 4
2147
2148#define FIRST_TO_LAST 0
2149#define LAST_TO_FIRST 1
2150
2151/* used in do_balance for passing parent of node information that has
2152 been gotten from tb struct */
2153struct buffer_info {
2154 struct tree_balance *tb;
2155 struct buffer_head *bi_bh;
2156 struct buffer_head *bi_parent;
2157 int bi_position;
2158};
2159
2160static inline struct super_block *sb_from_tb(struct tree_balance *tb)
2161{
2162 return tb ? tb->tb_sb : NULL;
2163}
2164
2165static inline struct super_block *sb_from_bi(struct buffer_info *bi)
2166{
2167 return bi ? sb_from_tb(bi->tb) : NULL;
2168}
2169
2170/* there are 4 types of items: stat data, directory item, indirect, direct.
2171+-------------------+------------+--------------+------------+
2172| | k_offset | k_uniqueness | mergeable? |
2173+-------------------+------------+--------------+------------+
2174| stat data | 0 | 0 | no |
2175+-------------------+------------+--------------+------------+
2176| 1st directory item| DOT_OFFSET |DIRENTRY_UNIQUENESS| no |
2177| non 1st directory | hash value | | yes |
2178| item | | | |
2179+-------------------+------------+--------------+------------+
2180| indirect item | offset + 1 |TYPE_INDIRECT | if this is not the first indirect item of the object
2181+-------------------+------------+--------------+------------+
2182| direct item | offset + 1 |TYPE_DIRECT | if not this is not the first direct item of the object
2183+-------------------+------------+--------------+------------+
2184*/
2185
2186struct item_operations {
2187 int (*bytes_number) (struct item_head * ih, int block_size);
2188 void (*decrement_key) (struct cpu_key *);
2189 int (*is_left_mergeable) (struct reiserfs_key * ih,
2190 unsigned long bsize);
2191 void (*print_item) (struct item_head *, char *item);
2192 void (*check_item) (struct item_head *, char *item);
2193
2194 int (*create_vi) (struct virtual_node * vn, struct virtual_item * vi,
2195 int is_affected, int insert_size);
2196 int (*check_left) (struct virtual_item * vi, int free,
2197 int start_skip, int end_skip);
2198 int (*check_right) (struct virtual_item * vi, int free);
2199 int (*part_size) (struct virtual_item * vi, int from, int to);
2200 int (*unit_num) (struct virtual_item * vi);
2201 void (*print_vi) (struct virtual_item * vi);
2202};
2203
2204extern struct item_operations *item_ops[TYPE_ANY + 1];
2205
2206#define op_bytes_number(ih,bsize) item_ops[le_ih_k_type (ih)]->bytes_number (ih, bsize)
2207#define op_is_left_mergeable(key,bsize) item_ops[le_key_k_type (le_key_version (key), key)]->is_left_mergeable (key, bsize)
2208#define op_print_item(ih,item) item_ops[le_ih_k_type (ih)]->print_item (ih, item)
2209#define op_check_item(ih,item) item_ops[le_ih_k_type (ih)]->check_item (ih, item)
2210#define op_create_vi(vn,vi,is_affected,insert_size) item_ops[le_ih_k_type ((vi)->vi_ih)]->create_vi (vn,vi,is_affected,insert_size)
2211#define op_check_left(vi,free,start_skip,end_skip) item_ops[(vi)->vi_index]->check_left (vi, free, start_skip, end_skip)
2212#define op_check_right(vi,free) item_ops[(vi)->vi_index]->check_right (vi, free)
2213#define op_part_size(vi,from,to) item_ops[(vi)->vi_index]->part_size (vi, from, to)
2214#define op_unit_num(vi) item_ops[(vi)->vi_index]->unit_num (vi)
2215#define op_print_vi(vi) item_ops[(vi)->vi_index]->print_vi (vi)
2216
2217#define COMP_SHORT_KEYS comp_short_keys
2218
2219/* number of blocks pointed to by the indirect item */
2220#define I_UNFM_NUM(ih) (ih_item_len(ih) / UNFM_P_SIZE)
2221
2222/* the used space within the unformatted node corresponding to pos within the item pointed to by ih */
2223#define I_POS_UNFM_SIZE(ih,pos,size) (((pos) == I_UNFM_NUM(ih) - 1 ) ? (size) - ih_free_space(ih) : (size))
2224
2225/* number of bytes contained by the direct item or the unformatted nodes the indirect item points to */
2226
2227/* get the item header */
2228#define B_N_PITEM_HEAD(bh,item_num) ( (struct item_head * )((bh)->b_data + BLKH_SIZE) + (item_num) )
2229
2230/* get key */
2231#define B_N_PDELIM_KEY(bh,item_num) ( (struct reiserfs_key * )((bh)->b_data + BLKH_SIZE) + (item_num) )
2232
2233/* get the key */
2234#define B_N_PKEY(bh,item_num) ( &(B_N_PITEM_HEAD(bh,item_num)->ih_key) )
2235
2236/* get item body */
2237#define B_N_PITEM(bh,item_num) ( (bh)->b_data + ih_location(B_N_PITEM_HEAD((bh),(item_num))))
2238
2239/* get the stat data by the buffer header and the item order */
2240#define B_N_STAT_DATA(bh,nr) \
2241( (struct stat_data *)((bh)->b_data + ih_location(B_N_PITEM_HEAD((bh),(nr))) ) )
2242
2243 /* following defines use reiserfs buffer header and item header */
2244
2245/* get stat-data */
2246#define B_I_STAT_DATA(bh, ih) ( (struct stat_data * )((bh)->b_data + ih_location(ih)) )
2247
2248// this is 3976 for size==4096
2249#define MAX_DIRECT_ITEM_LEN(size) ((size) - BLKH_SIZE - 2*IH_SIZE - SD_SIZE - UNFM_P_SIZE)
2250
2251/* indirect items consist of entries which contain blocknrs, pos
2252 indicates which entry, and B_I_POS_UNFM_POINTER resolves to the
2253 blocknr contained by the entry pos points to */
2254#define B_I_POS_UNFM_POINTER(bh,ih,pos) le32_to_cpu(*(((unp_t *)B_I_PITEM(bh,ih)) + (pos)))
2255#define PUT_B_I_POS_UNFM_POINTER(bh,ih,pos, val) do {*(((unp_t *)B_I_PITEM(bh,ih)) + (pos)) = cpu_to_le32(val); } while (0)
2256
2257struct reiserfs_iget_args {
2258 __u32 objectid;
2259 __u32 dirid;
2260};
2261
2262/***************************************************************************/
2263/* FUNCTION DECLARATIONS */
2264/***************************************************************************/
2265
2266#define get_journal_desc_magic(bh) (bh->b_data + bh->b_size - 12)
2267
2268#define journal_trans_half(blocksize) \
2269 ((blocksize - sizeof (struct reiserfs_journal_desc) + sizeof (__u32) - 12) / sizeof (__u32))
2270
2271/* journal.c see journal.c for all the comments here */
2272
2273/* first block written in a commit. */
2274struct reiserfs_journal_desc {
2275 __le32 j_trans_id; /* id of commit */
2276 __le32 j_len; /* length of commit. len +1 is the commit block */
2277 __le32 j_mount_id; /* mount id of this trans */
2278 __le32 j_realblock[1]; /* real locations for each block */
2279};
2280
2281#define get_desc_trans_id(d) le32_to_cpu((d)->j_trans_id)
2282#define get_desc_trans_len(d) le32_to_cpu((d)->j_len)
2283#define get_desc_mount_id(d) le32_to_cpu((d)->j_mount_id)
2284
2285#define set_desc_trans_id(d,val) do { (d)->j_trans_id = cpu_to_le32 (val); } while (0)
2286#define set_desc_trans_len(d,val) do { (d)->j_len = cpu_to_le32 (val); } while (0)
2287#define set_desc_mount_id(d,val) do { (d)->j_mount_id = cpu_to_le32 (val); } while (0)
2288
2289/* last block written in a commit */
2290struct reiserfs_journal_commit {
2291 __le32 j_trans_id; /* must match j_trans_id from the desc block */
2292 __le32 j_len; /* ditto */
2293 __le32 j_realblock[1]; /* real locations for each block */
2294};
2295
2296#define get_commit_trans_id(c) le32_to_cpu((c)->j_trans_id)
2297#define get_commit_trans_len(c) le32_to_cpu((c)->j_len)
2298#define get_commit_mount_id(c) le32_to_cpu((c)->j_mount_id)
2299
2300#define set_commit_trans_id(c,val) do { (c)->j_trans_id = cpu_to_le32 (val); } while (0)
2301#define set_commit_trans_len(c,val) do { (c)->j_len = cpu_to_le32 (val); } while (0)
2302
2303/* this header block gets written whenever a transaction is considered fully flushed, and is more recent than the
2304** last fully flushed transaction. fully flushed means all the log blocks and all the real blocks are on disk,
2305** and this transaction does not need to be replayed.
2306*/
2307struct reiserfs_journal_header {
2308 __le32 j_last_flush_trans_id; /* id of last fully flushed transaction */
2309 __le32 j_first_unflushed_offset; /* offset in the log of where to start replay after a crash */
2310 __le32 j_mount_id;
2311 /* 12 */ struct journal_params jh_journal;
2312};
2313
2314/* biggest tunable defines are right here */
2315#define JOURNAL_BLOCK_COUNT 8192 /* number of blocks in the journal */
2316#define JOURNAL_TRANS_MAX_DEFAULT 1024 /* biggest possible single transaction, don't change for now (8/3/99) */
2317#define JOURNAL_TRANS_MIN_DEFAULT 256
2318#define JOURNAL_MAX_BATCH_DEFAULT 900 /* max blocks to batch into one transaction, don't make this any bigger than 900 */
2319#define JOURNAL_MIN_RATIO 2
2320#define JOURNAL_MAX_COMMIT_AGE 30
2321#define JOURNAL_MAX_TRANS_AGE 30
2322#define JOURNAL_PER_BALANCE_CNT (3 * (MAX_HEIGHT-2) + 9)
2323#define JOURNAL_BLOCKS_PER_OBJECT(sb) (JOURNAL_PER_BALANCE_CNT * 3 + \
2324 2 * (REISERFS_QUOTA_INIT_BLOCKS(sb) + \
2325 REISERFS_QUOTA_TRANS_BLOCKS(sb)))
2326
2327#ifdef CONFIG_QUOTA
2328#define REISERFS_QUOTA_OPTS ((1 << REISERFS_USRQUOTA) | (1 << REISERFS_GRPQUOTA))
2329/* We need to update data and inode (atime) */
2330#define REISERFS_QUOTA_TRANS_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & REISERFS_QUOTA_OPTS ? 2 : 0)
2331/* 1 balancing, 1 bitmap, 1 data per write + stat data update */
2332#define REISERFS_QUOTA_INIT_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & REISERFS_QUOTA_OPTS ? \
2333(DQUOT_INIT_ALLOC*(JOURNAL_PER_BALANCE_CNT+2)+DQUOT_INIT_REWRITE+1) : 0)
2334/* same as with INIT */
2335#define REISERFS_QUOTA_DEL_BLOCKS(s) (REISERFS_SB(s)->s_mount_opt & REISERFS_QUOTA_OPTS ? \
2336(DQUOT_DEL_ALLOC*(JOURNAL_PER_BALANCE_CNT+2)+DQUOT_DEL_REWRITE+1) : 0)
2337#else
2338#define REISERFS_QUOTA_TRANS_BLOCKS(s) 0
2339#define REISERFS_QUOTA_INIT_BLOCKS(s) 0
2340#define REISERFS_QUOTA_DEL_BLOCKS(s) 0
2341#endif
2342
2343/* both of these can be as low as 1, or as high as you want. The min is the
2344** number of 4k bitmap nodes preallocated on mount. New nodes are allocated
2345** as needed, and released when transactions are committed. On release, if
2346** the current number of nodes is > max, the node is freed, otherwise,
2347** it is put on a free list for faster use later.
2348*/
2349#define REISERFS_MIN_BITMAP_NODES 10
2350#define REISERFS_MAX_BITMAP_NODES 100
2351
2352#define JBH_HASH_SHIFT 13 /* these are based on journal hash size of 8192 */
2353#define JBH_HASH_MASK 8191
2354
2355#define _jhashfn(sb,block) \
2356 (((unsigned long)sb>>L1_CACHE_SHIFT) ^ \
2357 (((block)<<(JBH_HASH_SHIFT - 6)) ^ ((block) >> 13) ^ ((block) << (JBH_HASH_SHIFT - 12))))
2358#define journal_hash(t,sb,block) ((t)[_jhashfn((sb),(block)) & JBH_HASH_MASK])
2359
2360// We need these to make journal.c code more readable
2361#define journal_find_get_block(s, block) __find_get_block(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)
2362#define journal_getblk(s, block) __getblk(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)
2363#define journal_bread(s, block) __bread(SB_JOURNAL(s)->j_dev_bd, block, s->s_blocksize)
2364
2365enum reiserfs_bh_state_bits {
2366 BH_JDirty = BH_PrivateStart, /* buffer is in current transaction */
2367 BH_JDirty_wait,
2368 BH_JNew, /* disk block was taken off free list before
2369 * being in a finished transaction, or
2370 * written to disk. Can be reused immed. */
2371 BH_JPrepared,
2372 BH_JRestore_dirty,
2373 BH_JTest, // debugging only will go away
2374};
2375
2376BUFFER_FNS(JDirty, journaled);
2377TAS_BUFFER_FNS(JDirty, journaled);
2378BUFFER_FNS(JDirty_wait, journal_dirty);
2379TAS_BUFFER_FNS(JDirty_wait, journal_dirty);
2380BUFFER_FNS(JNew, journal_new);
2381TAS_BUFFER_FNS(JNew, journal_new);
2382BUFFER_FNS(JPrepared, journal_prepared);
2383TAS_BUFFER_FNS(JPrepared, journal_prepared);
2384BUFFER_FNS(JRestore_dirty, journal_restore_dirty);
2385TAS_BUFFER_FNS(JRestore_dirty, journal_restore_dirty);
2386BUFFER_FNS(JTest, journal_test);
2387TAS_BUFFER_FNS(JTest, journal_test);
2388
2389/*
2390** transaction handle which is passed around for all journal calls
2391*/
2392struct reiserfs_transaction_handle {
2393 struct super_block *t_super; /* super for this FS when journal_begin was
2394 called. saves calls to reiserfs_get_super
2395 also used by nested transactions to make
2396 sure they are nesting on the right FS
2397 _must_ be first in the handle
2398 */
2399 int t_refcount;
2400 int t_blocks_logged; /* number of blocks this writer has logged */
2401 int t_blocks_allocated; /* number of blocks this writer allocated */
2402 unsigned int t_trans_id; /* sanity check, equals the current trans id */
2403 void *t_handle_save; /* save existing current->journal_info */
2404 unsigned displace_new_blocks:1; /* if new block allocation occurres, that block
2405 should be displaced from others */
2406 struct list_head t_list;
2407};
2408
2409/* used to keep track of ordered and tail writes, attached to the buffer
2410 * head through b_journal_head.
2411 */
2412struct reiserfs_jh {
2413 struct reiserfs_journal_list *jl;
2414 struct buffer_head *bh;
2415 struct list_head list;
2416};
2417
2418void reiserfs_free_jh(struct buffer_head *bh);
2419int reiserfs_add_tail_list(struct inode *inode, struct buffer_head *bh);
2420int reiserfs_add_ordered_list(struct inode *inode, struct buffer_head *bh);
2421int journal_mark_dirty(struct reiserfs_transaction_handle *,
2422 struct super_block *, struct buffer_head *bh);
2423
2424static inline int reiserfs_file_data_log(struct inode *inode)
2425{
2426 if (reiserfs_data_log(inode->i_sb) ||
2427 (REISERFS_I(inode)->i_flags & i_data_log))
2428 return 1;
2429 return 0;
2430}
2431
2432static inline int reiserfs_transaction_running(struct super_block *s)
2433{
2434 struct reiserfs_transaction_handle *th = current->journal_info;
2435 if (th && th->t_super == s)
2436 return 1;
2437 if (th && th->t_super == NULL)
2438 BUG();
2439 return 0;
2440}
2441
2442static inline int reiserfs_transaction_free_space(struct reiserfs_transaction_handle *th)
2443{
2444 return th->t_blocks_allocated - th->t_blocks_logged;
2445}
2446
2447struct reiserfs_transaction_handle *reiserfs_persistent_transaction(struct
2448 super_block
2449 *,
2450 int count);
2451int reiserfs_end_persistent_transaction(struct reiserfs_transaction_handle *);
2452int reiserfs_commit_page(struct inode *inode, struct page *page,
2453 unsigned from, unsigned to);
2454int reiserfs_flush_old_commits(struct super_block *);
2455int reiserfs_commit_for_inode(struct inode *);
2456int reiserfs_inode_needs_commit(struct inode *);
2457void reiserfs_update_inode_transaction(struct inode *);
2458void reiserfs_wait_on_write_block(struct super_block *s);
2459void reiserfs_block_writes(struct reiserfs_transaction_handle *th);
2460void reiserfs_allow_writes(struct super_block *s);
2461void reiserfs_check_lock_depth(struct super_block *s, char *caller);
2462int reiserfs_prepare_for_journal(struct super_block *, struct buffer_head *bh,
2463 int wait);
2464void reiserfs_restore_prepared_buffer(struct super_block *,
2465 struct buffer_head *bh);
2466int journal_init(struct super_block *, const char *j_dev_name, int old_format,
2467 unsigned int);
2468int journal_release(struct reiserfs_transaction_handle *, struct super_block *);
2469int journal_release_error(struct reiserfs_transaction_handle *,
2470 struct super_block *);
2471int journal_end(struct reiserfs_transaction_handle *, struct super_block *,
2472 unsigned long);
2473int journal_end_sync(struct reiserfs_transaction_handle *, struct super_block *,
2474 unsigned long);
2475int journal_mark_freed(struct reiserfs_transaction_handle *,
2476 struct super_block *, b_blocknr_t blocknr);
2477int journal_transaction_should_end(struct reiserfs_transaction_handle *, int);
2478int reiserfs_in_journal(struct super_block *sb, unsigned int bmap_nr,
2479 int bit_nr, int searchall, b_blocknr_t *next);
2480int journal_begin(struct reiserfs_transaction_handle *,
2481 struct super_block *sb, unsigned long);
2482int journal_join_abort(struct reiserfs_transaction_handle *,
2483 struct super_block *sb, unsigned long);
2484void reiserfs_abort_journal(struct super_block *sb, int errno);
2485void reiserfs_abort(struct super_block *sb, int errno, const char *fmt, ...);
2486int reiserfs_allocate_list_bitmaps(struct super_block *s,
2487 struct reiserfs_list_bitmap *, unsigned int);
2488
2489void add_save_link(struct reiserfs_transaction_handle *th,
2490 struct inode *inode, int truncate);
2491int remove_save_link(struct inode *inode, int truncate);
2492
2493/* objectid.c */
2494__u32 reiserfs_get_unused_objectid(struct reiserfs_transaction_handle *th);
2495void reiserfs_release_objectid(struct reiserfs_transaction_handle *th,
2496 __u32 objectid_to_release);
2497int reiserfs_convert_objectid_map_v1(struct super_block *);
2498
2499/* stree.c */
2500int B_IS_IN_TREE(const struct buffer_head *);
2501extern void copy_item_head(struct item_head *to,
2502 const struct item_head *from);
2503
2504// first key is in cpu form, second - le
2505extern int comp_short_keys(const struct reiserfs_key *le_key,
2506 const struct cpu_key *cpu_key);
2507extern void le_key2cpu_key(struct cpu_key *to, const struct reiserfs_key *from);
2508
2509// both are in le form
2510extern int comp_le_keys(const struct reiserfs_key *,
2511 const struct reiserfs_key *);
2512extern int comp_short_le_keys(const struct reiserfs_key *,
2513 const struct reiserfs_key *);
2514
2515//
2516// get key version from on disk key - kludge
2517//
2518static inline int le_key_version(const struct reiserfs_key *key)
2519{
2520 int type;
2521
2522 type = offset_v2_k_type(&(key->u.k_offset_v2));
2523 if (type != TYPE_DIRECT && type != TYPE_INDIRECT
2524 && type != TYPE_DIRENTRY)
2525 return KEY_FORMAT_3_5;
2526
2527 return KEY_FORMAT_3_6;
2528
2529}
2530
2531static inline void copy_key(struct reiserfs_key *to,
2532 const struct reiserfs_key *from)
2533{
2534 memcpy(to, from, KEY_SIZE);
2535}
2536
2537int comp_items(const struct item_head *stored_ih, const struct treepath *path);
2538const struct reiserfs_key *get_rkey(const struct treepath *chk_path,
2539 const struct super_block *sb);
2540int search_by_key(struct super_block *, const struct cpu_key *,
2541 struct treepath *, int);
2542#define search_item(s,key,path) search_by_key (s, key, path, DISK_LEAF_NODE_LEVEL)
2543int search_for_position_by_key(struct super_block *sb,
2544 const struct cpu_key *cpu_key,
2545 struct treepath *search_path);
2546extern void decrement_bcount(struct buffer_head *bh);
2547void decrement_counters_in_path(struct treepath *search_path);
2548void pathrelse(struct treepath *search_path);
2549int reiserfs_check_path(struct treepath *p);
2550void pathrelse_and_restore(struct super_block *s, struct treepath *search_path);
2551
2552int reiserfs_insert_item(struct reiserfs_transaction_handle *th,
2553 struct treepath *path,
2554 const struct cpu_key *key,
2555 struct item_head *ih,
2556 struct inode *inode, const char *body);
2557
2558int reiserfs_paste_into_item(struct reiserfs_transaction_handle *th,
2559 struct treepath *path,
2560 const struct cpu_key *key,
2561 struct inode *inode,
2562 const char *body, int paste_size);
2563
2564int reiserfs_cut_from_item(struct reiserfs_transaction_handle *th,
2565 struct treepath *path,
2566 struct cpu_key *key,
2567 struct inode *inode,
2568 struct page *page, loff_t new_file_size);
2569
2570int reiserfs_delete_item(struct reiserfs_transaction_handle *th,
2571 struct treepath *path,
2572 const struct cpu_key *key,
2573 struct inode *inode, struct buffer_head *un_bh);
2574
2575void reiserfs_delete_solid_item(struct reiserfs_transaction_handle *th,
2576 struct inode *inode, struct reiserfs_key *key);
2577int reiserfs_delete_object(struct reiserfs_transaction_handle *th,
2578 struct inode *inode);
2579int reiserfs_do_truncate(struct reiserfs_transaction_handle *th,
2580 struct inode *inode, struct page *,
2581 int update_timestamps);
2582
2583#define i_block_size(inode) ((inode)->i_sb->s_blocksize)
2584#define file_size(inode) ((inode)->i_size)
2585#define tail_size(inode) (file_size (inode) & (i_block_size (inode) - 1))
2586
2587#define tail_has_to_be_packed(inode) (have_large_tails ((inode)->i_sb)?\
2588!STORE_TAIL_IN_UNFM_S1(file_size (inode), tail_size(inode), inode->i_sb->s_blocksize):have_small_tails ((inode)->i_sb)?!STORE_TAIL_IN_UNFM_S2(file_size (inode), tail_size(inode), inode->i_sb->s_blocksize):0 )
2589
2590void padd_item(char *item, int total_length, int length);
2591
2592/* inode.c */
2593/* args for the create parameter of reiserfs_get_block */
2594#define GET_BLOCK_NO_CREATE 0 /* don't create new blocks or convert tails */
2595#define GET_BLOCK_CREATE 1 /* add anything you need to find block */
2596#define GET_BLOCK_NO_HOLE 2 /* return -ENOENT for file holes */
2597#define GET_BLOCK_READ_DIRECT 4 /* read the tail if indirect item not found */
2598#define GET_BLOCK_NO_IMUX 8 /* i_mutex is not held, don't preallocate */
2599#define GET_BLOCK_NO_DANGLE 16 /* don't leave any transactions running */
2600
2601void reiserfs_read_locked_inode(struct inode *inode,
2602 struct reiserfs_iget_args *args);
2603int reiserfs_find_actor(struct inode *inode, void *p);
2604int reiserfs_init_locked_inode(struct inode *inode, void *p);
2605void reiserfs_evict_inode(struct inode *inode);
2606int reiserfs_write_inode(struct inode *inode, struct writeback_control *wbc);
2607int reiserfs_get_block(struct inode *inode, sector_t block,
2608 struct buffer_head *bh_result, int create);
2609struct dentry *reiserfs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2610 int fh_len, int fh_type);
2611struct dentry *reiserfs_fh_to_parent(struct super_block *sb, struct fid *fid,
2612 int fh_len, int fh_type);
2613int reiserfs_encode_fh(struct dentry *dentry, __u32 * data, int *lenp,
2614 int connectable);
2615
2616int reiserfs_truncate_file(struct inode *, int update_timestamps);
2617void make_cpu_key(struct cpu_key *cpu_key, struct inode *inode, loff_t offset,
2618 int type, int key_length);
2619void make_le_item_head(struct item_head *ih, const struct cpu_key *key,
2620 int version,
2621 loff_t offset, int type, int length, int entry_count);
2622struct inode *reiserfs_iget(struct super_block *s, const struct cpu_key *key);
2623
2624struct reiserfs_security_handle;
2625int reiserfs_new_inode(struct reiserfs_transaction_handle *th,
2626 struct inode *dir, umode_t mode,
2627 const char *symname, loff_t i_size,
2628 struct dentry *dentry, struct inode *inode,
2629 struct reiserfs_security_handle *security);
2630
2631void reiserfs_update_sd_size(struct reiserfs_transaction_handle *th,
2632 struct inode *inode, loff_t size);
2633
2634static inline void reiserfs_update_sd(struct reiserfs_transaction_handle *th,
2635 struct inode *inode)
2636{
2637 reiserfs_update_sd_size(th, inode, inode->i_size);
2638}
2639
2640void sd_attrs_to_i_attrs(__u16 sd_attrs, struct inode *inode);
2641void i_attrs_to_sd_attrs(struct inode *inode, __u16 * sd_attrs);
2642int reiserfs_setattr(struct dentry *dentry, struct iattr *attr);
2643
2644int __reiserfs_write_begin(struct page *page, unsigned from, unsigned len);
2645
2646/* namei.c */
2647void set_de_name_and_namelen(struct reiserfs_dir_entry *de);
2648int search_by_entry_key(struct super_block *sb, const struct cpu_key *key,
2649 struct treepath *path, struct reiserfs_dir_entry *de);
2650struct dentry *reiserfs_get_parent(struct dentry *);
2651
2652#ifdef CONFIG_REISERFS_PROC_INFO
2653int reiserfs_proc_info_init(struct super_block *sb);
2654int reiserfs_proc_info_done(struct super_block *sb);
2655int reiserfs_proc_info_global_init(void);
2656int reiserfs_proc_info_global_done(void);
2657
2658#define PROC_EXP( e ) e
2659
2660#define __PINFO( sb ) REISERFS_SB(sb) -> s_proc_info_data
2661#define PROC_INFO_MAX( sb, field, value ) \
2662 __PINFO( sb ).field = \
2663 max( REISERFS_SB( sb ) -> s_proc_info_data.field, value )
2664#define PROC_INFO_INC( sb, field ) ( ++ ( __PINFO( sb ).field ) )
2665#define PROC_INFO_ADD( sb, field, val ) ( __PINFO( sb ).field += ( val ) )
2666#define PROC_INFO_BH_STAT( sb, bh, level ) \
2667 PROC_INFO_INC( sb, sbk_read_at[ ( level ) ] ); \
2668 PROC_INFO_ADD( sb, free_at[ ( level ) ], B_FREE_SPACE( bh ) ); \
2669 PROC_INFO_ADD( sb, items_at[ ( level ) ], B_NR_ITEMS( bh ) )
2670#else
2671static inline int reiserfs_proc_info_init(struct super_block *sb)
2672{
2673 return 0;
2674}
2675
2676static inline int reiserfs_proc_info_done(struct super_block *sb)
2677{
2678 return 0;
2679}
2680
2681static inline int reiserfs_proc_info_global_init(void)
2682{
2683 return 0;
2684}
2685
2686static inline int reiserfs_proc_info_global_done(void)
2687{
2688 return 0;
2689}
2690
2691#define PROC_EXP( e )
2692#define VOID_V ( ( void ) 0 )
2693#define PROC_INFO_MAX( sb, field, value ) VOID_V
2694#define PROC_INFO_INC( sb, field ) VOID_V
2695#define PROC_INFO_ADD( sb, field, val ) VOID_V
2696#define PROC_INFO_BH_STAT(sb, bh, n_node_level) VOID_V
2697#endif
2698
2699/* dir.c */
2700extern const struct inode_operations reiserfs_dir_inode_operations;
2701extern const struct inode_operations reiserfs_symlink_inode_operations;
2702extern const struct inode_operations reiserfs_special_inode_operations;
2703extern const struct file_operations reiserfs_dir_operations;
2704int reiserfs_readdir_dentry(struct dentry *, void *, filldir_t, loff_t *);
2705
2706/* tail_conversion.c */
2707int direct2indirect(struct reiserfs_transaction_handle *, struct inode *,
2708 struct treepath *, struct buffer_head *, loff_t);
2709int indirect2direct(struct reiserfs_transaction_handle *, struct inode *,
2710 struct page *, struct treepath *, const struct cpu_key *,
2711 loff_t, char *);
2712void reiserfs_unmap_buffer(struct buffer_head *);
2713
2714/* file.c */
2715extern const struct inode_operations reiserfs_file_inode_operations;
2716extern const struct file_operations reiserfs_file_operations;
2717extern const struct address_space_operations reiserfs_address_space_operations;
2718
2719/* fix_nodes.c */
2720
2721int fix_nodes(int n_op_mode, struct tree_balance *tb,
2722 struct item_head *ins_ih, const void *);
2723void unfix_nodes(struct tree_balance *);
2724
2725/* prints.c */
2726void __reiserfs_panic(struct super_block *s, const char *id,
2727 const char *function, const char *fmt, ...)
2728 __attribute__ ((noreturn));
2729#define reiserfs_panic(s, id, fmt, args...) \
2730 __reiserfs_panic(s, id, __func__, fmt, ##args)
2731void __reiserfs_error(struct super_block *s, const char *id,
2732 const char *function, const char *fmt, ...);
2733#define reiserfs_error(s, id, fmt, args...) \
2734 __reiserfs_error(s, id, __func__, fmt, ##args)
2735void reiserfs_info(struct super_block *s, const char *fmt, ...);
2736void reiserfs_debug(struct super_block *s, int level, const char *fmt, ...);
2737void print_indirect_item(struct buffer_head *bh, int item_num);
2738void store_print_tb(struct tree_balance *tb);
2739void print_cur_tb(char *mes);
2740void print_de(struct reiserfs_dir_entry *de);
2741void print_bi(struct buffer_info *bi, char *mes);
2742#define PRINT_LEAF_ITEMS 1 /* print all items */
2743#define PRINT_DIRECTORY_ITEMS 2 /* print directory items */
2744#define PRINT_DIRECT_ITEMS 4 /* print contents of direct items */
2745void print_block(struct buffer_head *bh, ...);
2746void print_bmap(struct super_block *s, int silent);
2747void print_bmap_block(int i, char *data, int size, int silent);
2748/*void print_super_block (struct super_block * s, char * mes);*/
2749void print_objectid_map(struct super_block *s);
2750void print_block_head(struct buffer_head *bh, char *mes);
2751void check_leaf(struct buffer_head *bh);
2752void check_internal(struct buffer_head *bh);
2753void print_statistics(struct super_block *s);
2754char *reiserfs_hashname(int code);
2755
2756/* lbalance.c */
2757int leaf_move_items(int shift_mode, struct tree_balance *tb, int mov_num,
2758 int mov_bytes, struct buffer_head *Snew);
2759int leaf_shift_left(struct tree_balance *tb, int shift_num, int shift_bytes);
2760int leaf_shift_right(struct tree_balance *tb, int shift_num, int shift_bytes);
2761void leaf_delete_items(struct buffer_info *cur_bi, int last_first, int first,
2762 int del_num, int del_bytes);
2763void leaf_insert_into_buf(struct buffer_info *bi, int before,
2764 struct item_head *inserted_item_ih,
2765 const char *inserted_item_body, int zeros_number);
2766void leaf_paste_in_buffer(struct buffer_info *bi, int pasted_item_num,
2767 int pos_in_item, int paste_size, const char *body,
2768 int zeros_number);
2769void leaf_cut_from_buffer(struct buffer_info *bi, int cut_item_num,
2770 int pos_in_item, int cut_size);
2771void leaf_paste_entries(struct buffer_info *bi, int item_num, int before,
2772 int new_entry_count, struct reiserfs_de_head *new_dehs,
2773 const char *records, int paste_size);
2774/* ibalance.c */
2775int balance_internal(struct tree_balance *, int, int, struct item_head *,
2776 struct buffer_head **);
2777
2778/* do_balance.c */
2779void do_balance_mark_leaf_dirty(struct tree_balance *tb,
2780 struct buffer_head *bh, int flag);
2781#define do_balance_mark_internal_dirty do_balance_mark_leaf_dirty
2782#define do_balance_mark_sb_dirty do_balance_mark_leaf_dirty
2783
2784void do_balance(struct tree_balance *tb, struct item_head *ih,
2785 const char *body, int flag);
2786void reiserfs_invalidate_buffer(struct tree_balance *tb,
2787 struct buffer_head *bh);
2788
2789int get_left_neighbor_position(struct tree_balance *tb, int h);
2790int get_right_neighbor_position(struct tree_balance *tb, int h);
2791void replace_key(struct tree_balance *tb, struct buffer_head *, int,
2792 struct buffer_head *, int);
2793void make_empty_node(struct buffer_info *);
2794struct buffer_head *get_FEB(struct tree_balance *);
2795
2796/* bitmap.c */
2797
2798/* structure contains hints for block allocator, and it is a container for
2799 * arguments, such as node, search path, transaction_handle, etc. */
2800struct __reiserfs_blocknr_hint {
2801 struct inode *inode; /* inode passed to allocator, if we allocate unf. nodes */
2802 sector_t block; /* file offset, in blocks */
2803 struct in_core_key key;
2804 struct treepath *path; /* search path, used by allocator to deternine search_start by
2805 * various ways */
2806 struct reiserfs_transaction_handle *th; /* transaction handle is needed to log super blocks and
2807 * bitmap blocks changes */
2808 b_blocknr_t beg, end;
2809 b_blocknr_t search_start; /* a field used to transfer search start value (block number)
2810 * between different block allocator procedures
2811 * (determine_search_start() and others) */
2812 int prealloc_size; /* is set in determine_prealloc_size() function, used by underlayed
2813 * function that do actual allocation */
2814
2815 unsigned formatted_node:1; /* the allocator uses different polices for getting disk space for
2816 * formatted/unformatted blocks with/without preallocation */
2817 unsigned preallocate:1;
2818};
2819
2820typedef struct __reiserfs_blocknr_hint reiserfs_blocknr_hint_t;
2821
2822int reiserfs_parse_alloc_options(struct super_block *, char *);
2823void reiserfs_init_alloc_options(struct super_block *s);
2824
2825/*
2826 * given a directory, this will tell you what packing locality
2827 * to use for a new object underneat it. The locality is returned
2828 * in disk byte order (le).
2829 */
2830__le32 reiserfs_choose_packing(struct inode *dir);
2831
2832int reiserfs_init_bitmap_cache(struct super_block *sb);
2833void reiserfs_free_bitmap_cache(struct super_block *sb);
2834void reiserfs_cache_bitmap_metadata(struct super_block *sb, struct buffer_head *bh, struct reiserfs_bitmap_info *info);
2835struct buffer_head *reiserfs_read_bitmap_block(struct super_block *sb, unsigned int bitmap);
2836int is_reusable(struct super_block *s, b_blocknr_t block, int bit_value);
2837void reiserfs_free_block(struct reiserfs_transaction_handle *th, struct inode *,
2838 b_blocknr_t, int for_unformatted);
2839int reiserfs_allocate_blocknrs(reiserfs_blocknr_hint_t *, b_blocknr_t *, int,
2840 int);
2841static inline int reiserfs_new_form_blocknrs(struct tree_balance *tb,
2842 b_blocknr_t * new_blocknrs,
2843 int amount_needed)
2844{
2845 reiserfs_blocknr_hint_t hint = {
2846 .th = tb->transaction_handle,
2847 .path = tb->tb_path,
2848 .inode = NULL,
2849 .key = tb->key,
2850 .block = 0,
2851 .formatted_node = 1
2852 };
2853 return reiserfs_allocate_blocknrs(&hint, new_blocknrs, amount_needed,
2854 0);
2855}
2856
2857static inline int reiserfs_new_unf_blocknrs(struct reiserfs_transaction_handle
2858 *th, struct inode *inode,
2859 b_blocknr_t * new_blocknrs,
2860 struct treepath *path,
2861 sector_t block)
2862{
2863 reiserfs_blocknr_hint_t hint = {
2864 .th = th,
2865 .path = path,
2866 .inode = inode,
2867 .block = block,
2868 .formatted_node = 0,
2869 .preallocate = 0
2870 };
2871 return reiserfs_allocate_blocknrs(&hint, new_blocknrs, 1, 0);
2872}
2873
2874#ifdef REISERFS_PREALLOCATE
2875static inline int reiserfs_new_unf_blocknrs2(struct reiserfs_transaction_handle
2876 *th, struct inode *inode,
2877 b_blocknr_t * new_blocknrs,
2878 struct treepath *path,
2879 sector_t block)
2880{
2881 reiserfs_blocknr_hint_t hint = {
2882 .th = th,
2883 .path = path,
2884 .inode = inode,
2885 .block = block,
2886 .formatted_node = 0,
2887 .preallocate = 1
2888 };
2889 return reiserfs_allocate_blocknrs(&hint, new_blocknrs, 1, 0);
2890}
2891
2892void reiserfs_discard_prealloc(struct reiserfs_transaction_handle *th,
2893 struct inode *inode);
2894void reiserfs_discard_all_prealloc(struct reiserfs_transaction_handle *th);
2895#endif
2896
2897/* hashes.c */
2898__u32 keyed_hash(const signed char *msg, int len);
2899__u32 yura_hash(const signed char *msg, int len);
2900__u32 r5_hash(const signed char *msg, int len);
2901
2902#define reiserfs_set_le_bit __set_bit_le
2903#define reiserfs_test_and_set_le_bit __test_and_set_bit_le
2904#define reiserfs_clear_le_bit __clear_bit_le
2905#define reiserfs_test_and_clear_le_bit __test_and_clear_bit_le
2906#define reiserfs_test_le_bit test_bit_le
2907#define reiserfs_find_next_zero_le_bit find_next_zero_bit_le
2908
2909/* sometimes reiserfs_truncate may require to allocate few new blocks
2910 to perform indirect2direct conversion. People probably used to
2911 think, that truncate should work without problems on a filesystem
2912 without free disk space. They may complain that they can not
2913 truncate due to lack of free disk space. This spare space allows us
2914 to not worry about it. 500 is probably too much, but it should be
2915 absolutely safe */
2916#define SPARE_SPACE 500
2917
2918/* prototypes from ioctl.c */
2919long reiserfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg);
2920long reiserfs_compat_ioctl(struct file *filp,
2921 unsigned int cmd, unsigned long arg);
2922int reiserfs_unpack(struct inode *inode, struct file *filp);
diff --git a/fs/reiserfs/resize.c b/fs/reiserfs/resize.c
index 7483279b482d..9a17f63c3fd7 100644
--- a/fs/reiserfs/resize.c
+++ b/fs/reiserfs/resize.c
@@ -13,8 +13,7 @@
13#include <linux/vmalloc.h> 13#include <linux/vmalloc.h>
14#include <linux/string.h> 14#include <linux/string.h>
15#include <linux/errno.h> 15#include <linux/errno.h>
16#include <linux/reiserfs_fs.h> 16#include "reiserfs.h"
17#include <linux/reiserfs_fs_sb.h>
18#include <linux/buffer_head.h> 17#include <linux/buffer_head.h>
19 18
20int reiserfs_resize(struct super_block *s, unsigned long block_count_new) 19int reiserfs_resize(struct super_block *s, unsigned long block_count_new)
diff --git a/fs/reiserfs/stree.c b/fs/reiserfs/stree.c
index 77df82f9e70a..f8afa4b162b8 100644
--- a/fs/reiserfs/stree.c
+++ b/fs/reiserfs/stree.c
@@ -51,7 +51,7 @@
51#include <linux/time.h> 51#include <linux/time.h>
52#include <linux/string.h> 52#include <linux/string.h>
53#include <linux/pagemap.h> 53#include <linux/pagemap.h>
54#include <linux/reiserfs_fs.h> 54#include "reiserfs.h"
55#include <linux/buffer_head.h> 55#include <linux/buffer_head.h>
56#include <linux/quotaops.h> 56#include <linux/quotaops.h>
57 57
diff --git a/fs/reiserfs/super.c b/fs/reiserfs/super.c
index e12d8b97cd4d..8b7616ef06d8 100644
--- a/fs/reiserfs/super.c
+++ b/fs/reiserfs/super.c
@@ -16,9 +16,9 @@
16#include <linux/vmalloc.h> 16#include <linux/vmalloc.h>
17#include <linux/time.h> 17#include <linux/time.h>
18#include <asm/uaccess.h> 18#include <asm/uaccess.h>
19#include <linux/reiserfs_fs.h> 19#include "reiserfs.h"
20#include <linux/reiserfs_acl.h> 20#include "acl.h"
21#include <linux/reiserfs_xattr.h> 21#include "xattr.h"
22#include <linux/init.h> 22#include <linux/init.h>
23#include <linux/blkdev.h> 23#include <linux/blkdev.h>
24#include <linux/buffer_head.h> 24#include <linux/buffer_head.h>
@@ -1874,11 +1874,9 @@ static int reiserfs_fill_super(struct super_block *s, void *data, int silent)
1874 unlock_new_inode(root_inode); 1874 unlock_new_inode(root_inode);
1875 } 1875 }
1876 1876
1877 s->s_root = d_alloc_root(root_inode); 1877 s->s_root = d_make_root(root_inode);
1878 if (!s->s_root) { 1878 if (!s->s_root)
1879 iput(root_inode);
1880 goto error; 1879 goto error;
1881 }
1882 // define and initialize hash function 1880 // define and initialize hash function
1883 sbi->s_hash_function = hash_function(s); 1881 sbi->s_hash_function = hash_function(s);
1884 if (sbi->s_hash_function == NULL) { 1882 if (sbi->s_hash_function == NULL) {
diff --git a/fs/reiserfs/tail_conversion.c b/fs/reiserfs/tail_conversion.c
index 8f546bd473b8..5e2624d12f70 100644
--- a/fs/reiserfs/tail_conversion.c
+++ b/fs/reiserfs/tail_conversion.c
@@ -5,7 +5,7 @@
5#include <linux/time.h> 5#include <linux/time.h>
6#include <linux/pagemap.h> 6#include <linux/pagemap.h>
7#include <linux/buffer_head.h> 7#include <linux/buffer_head.h>
8#include <linux/reiserfs_fs.h> 8#include "reiserfs.h"
9 9
10/* access to tail : when one is going to read tail it must make sure, that is not running. 10/* access to tail : when one is going to read tail it must make sure, that is not running.
11 direct2indirect and indirect2direct can not run concurrently */ 11 direct2indirect and indirect2direct can not run concurrently */
diff --git a/fs/reiserfs/xattr.c b/fs/reiserfs/xattr.c
index c24deda8a8bc..46fc1c20a6b1 100644
--- a/fs/reiserfs/xattr.c
+++ b/fs/reiserfs/xattr.c
@@ -33,7 +33,7 @@
33 * The xattrs themselves are protected by the xattr_sem. 33 * The xattrs themselves are protected by the xattr_sem.
34 */ 34 */
35 35
36#include <linux/reiserfs_fs.h> 36#include "reiserfs.h"
37#include <linux/capability.h> 37#include <linux/capability.h>
38#include <linux/dcache.h> 38#include <linux/dcache.h>
39#include <linux/namei.h> 39#include <linux/namei.h>
@@ -43,8 +43,8 @@
43#include <linux/file.h> 43#include <linux/file.h>
44#include <linux/pagemap.h> 44#include <linux/pagemap.h>
45#include <linux/xattr.h> 45#include <linux/xattr.h>
46#include <linux/reiserfs_xattr.h> 46#include "xattr.h"
47#include <linux/reiserfs_acl.h> 47#include "acl.h"
48#include <asm/uaccess.h> 48#include <asm/uaccess.h>
49#include <net/checksum.h> 49#include <net/checksum.h>
50#include <linux/stat.h> 50#include <linux/stat.h>
diff --git a/fs/reiserfs/xattr.h b/fs/reiserfs/xattr.h
new file mode 100644
index 000000000000..f59626c5d33b
--- /dev/null
+++ b/fs/reiserfs/xattr.h
@@ -0,0 +1,122 @@
1#include <linux/reiserfs_xattr.h>
2#include <linux/init.h>
3#include <linux/list.h>
4#include <linux/rwsem.h>
5
6struct inode;
7struct dentry;
8struct iattr;
9struct super_block;
10struct nameidata;
11
12int reiserfs_xattr_register_handlers(void) __init;
13void reiserfs_xattr_unregister_handlers(void);
14int reiserfs_xattr_init(struct super_block *sb, int mount_flags);
15int reiserfs_lookup_privroot(struct super_block *sb);
16int reiserfs_delete_xattrs(struct inode *inode);
17int reiserfs_chown_xattrs(struct inode *inode, struct iattr *attrs);
18int reiserfs_permission(struct inode *inode, int mask);
19
20#ifdef CONFIG_REISERFS_FS_XATTR
21#define has_xattr_dir(inode) (REISERFS_I(inode)->i_flags & i_has_xattr_dir)
22ssize_t reiserfs_getxattr(struct dentry *dentry, const char *name,
23 void *buffer, size_t size);
24int reiserfs_setxattr(struct dentry *dentry, const char *name,
25 const void *value, size_t size, int flags);
26ssize_t reiserfs_listxattr(struct dentry *dentry, char *buffer, size_t size);
27int reiserfs_removexattr(struct dentry *dentry, const char *name);
28
29int reiserfs_xattr_get(struct inode *, const char *, void *, size_t);
30int reiserfs_xattr_set(struct inode *, const char *, const void *, size_t, int);
31int reiserfs_xattr_set_handle(struct reiserfs_transaction_handle *,
32 struct inode *, const char *, const void *,
33 size_t, int);
34
35extern const struct xattr_handler reiserfs_xattr_user_handler;
36extern const struct xattr_handler reiserfs_xattr_trusted_handler;
37extern const struct xattr_handler reiserfs_xattr_security_handler;
38#ifdef CONFIG_REISERFS_FS_SECURITY
39int reiserfs_security_init(struct inode *dir, struct inode *inode,
40 const struct qstr *qstr,
41 struct reiserfs_security_handle *sec);
42int reiserfs_security_write(struct reiserfs_transaction_handle *th,
43 struct inode *inode,
44 struct reiserfs_security_handle *sec);
45void reiserfs_security_free(struct reiserfs_security_handle *sec);
46#endif
47
48static inline int reiserfs_xattrs_initialized(struct super_block *sb)
49{
50 return REISERFS_SB(sb)->priv_root != NULL;
51}
52
53#define xattr_size(size) ((size) + sizeof(struct reiserfs_xattr_header))
54static inline loff_t reiserfs_xattr_nblocks(struct inode *inode, loff_t size)
55{
56 loff_t ret = 0;
57 if (reiserfs_file_data_log(inode)) {
58 ret = _ROUND_UP(xattr_size(size), inode->i_sb->s_blocksize);
59 ret >>= inode->i_sb->s_blocksize_bits;
60 }
61 return ret;
62}
63
64/* We may have to create up to 3 objects: xattr root, xattr dir, xattr file.
65 * Let's try to be smart about it.
66 * xattr root: We cache it. If it's not cached, we may need to create it.
67 * xattr dir: If anything has been loaded for this inode, we can set a flag
68 * saying so.
69 * xattr file: Since we don't cache xattrs, we can't tell. We always include
70 * blocks for it.
71 *
72 * However, since root and dir can be created between calls - YOU MUST SAVE
73 * THIS VALUE.
74 */
75static inline size_t reiserfs_xattr_jcreate_nblocks(struct inode *inode)
76{
77 size_t nblocks = JOURNAL_BLOCKS_PER_OBJECT(inode->i_sb);
78
79 if ((REISERFS_I(inode)->i_flags & i_has_xattr_dir) == 0) {
80 nblocks += JOURNAL_BLOCKS_PER_OBJECT(inode->i_sb);
81 if (!REISERFS_SB(inode->i_sb)->xattr_root->d_inode)
82 nblocks += JOURNAL_BLOCKS_PER_OBJECT(inode->i_sb);
83 }
84
85 return nblocks;
86}
87
88static inline void reiserfs_init_xattr_rwsem(struct inode *inode)
89{
90 init_rwsem(&REISERFS_I(inode)->i_xattr_sem);
91}
92
93#else
94
95#define reiserfs_getxattr NULL
96#define reiserfs_setxattr NULL
97#define reiserfs_listxattr NULL
98#define reiserfs_removexattr NULL
99
100static inline void reiserfs_init_xattr_rwsem(struct inode *inode)
101{
102}
103#endif /* CONFIG_REISERFS_FS_XATTR */
104
105#ifndef CONFIG_REISERFS_FS_SECURITY
106static inline int reiserfs_security_init(struct inode *dir,
107 struct inode *inode,
108 const struct qstr *qstr,
109 struct reiserfs_security_handle *sec)
110{
111 return 0;
112}
113static inline int
114reiserfs_security_write(struct reiserfs_transaction_handle *th,
115 struct inode *inode,
116 struct reiserfs_security_handle *sec)
117{
118 return 0;
119}
120static inline void reiserfs_security_free(struct reiserfs_security_handle *sec)
121{}
122#endif
diff --git a/fs/reiserfs/xattr_acl.c b/fs/reiserfs/xattr_acl.c
index 6da0396e5052..44474f9b990d 100644
--- a/fs/reiserfs/xattr_acl.c
+++ b/fs/reiserfs/xattr_acl.c
@@ -1,14 +1,14 @@
1#include <linux/capability.h> 1#include <linux/capability.h>
2#include <linux/fs.h> 2#include <linux/fs.h>
3#include <linux/posix_acl.h> 3#include <linux/posix_acl.h>
4#include <linux/reiserfs_fs.h> 4#include "reiserfs.h"
5#include <linux/errno.h> 5#include <linux/errno.h>
6#include <linux/pagemap.h> 6#include <linux/pagemap.h>
7#include <linux/xattr.h> 7#include <linux/xattr.h>
8#include <linux/slab.h> 8#include <linux/slab.h>
9#include <linux/posix_acl_xattr.h> 9#include <linux/posix_acl_xattr.h>
10#include <linux/reiserfs_xattr.h> 10#include "xattr.h"
11#include <linux/reiserfs_acl.h> 11#include "acl.h"
12#include <asm/uaccess.h> 12#include <asm/uaccess.h>
13 13
14static int reiserfs_set_acl(struct reiserfs_transaction_handle *th, 14static int reiserfs_set_acl(struct reiserfs_transaction_handle *th,
diff --git a/fs/reiserfs/xattr_security.c b/fs/reiserfs/xattr_security.c
index 534668fa41be..800a3cef6f62 100644
--- a/fs/reiserfs/xattr_security.c
+++ b/fs/reiserfs/xattr_security.c
@@ -1,10 +1,10 @@
1#include <linux/reiserfs_fs.h> 1#include "reiserfs.h"
2#include <linux/errno.h> 2#include <linux/errno.h>
3#include <linux/fs.h> 3#include <linux/fs.h>
4#include <linux/pagemap.h> 4#include <linux/pagemap.h>
5#include <linux/xattr.h> 5#include <linux/xattr.h>
6#include <linux/slab.h> 6#include <linux/slab.h>
7#include <linux/reiserfs_xattr.h> 7#include "xattr.h"
8#include <linux/security.h> 8#include <linux/security.h>
9#include <asm/uaccess.h> 9#include <asm/uaccess.h>
10 10
diff --git a/fs/reiserfs/xattr_trusted.c b/fs/reiserfs/xattr_trusted.c
index 9883736ce3ec..a0035719f66b 100644
--- a/fs/reiserfs/xattr_trusted.c
+++ b/fs/reiserfs/xattr_trusted.c
@@ -1,10 +1,10 @@
1#include <linux/reiserfs_fs.h> 1#include "reiserfs.h"
2#include <linux/capability.h> 2#include <linux/capability.h>
3#include <linux/errno.h> 3#include <linux/errno.h>
4#include <linux/fs.h> 4#include <linux/fs.h>
5#include <linux/pagemap.h> 5#include <linux/pagemap.h>
6#include <linux/xattr.h> 6#include <linux/xattr.h>
7#include <linux/reiserfs_xattr.h> 7#include "xattr.h"
8#include <asm/uaccess.h> 8#include <asm/uaccess.h>
9 9
10static int 10static int
diff --git a/fs/reiserfs/xattr_user.c b/fs/reiserfs/xattr_user.c
index 45ae1a00013a..8667491ae7c3 100644
--- a/fs/reiserfs/xattr_user.c
+++ b/fs/reiserfs/xattr_user.c
@@ -1,9 +1,9 @@
1#include <linux/reiserfs_fs.h> 1#include "reiserfs.h"
2#include <linux/errno.h> 2#include <linux/errno.h>
3#include <linux/fs.h> 3#include <linux/fs.h>
4#include <linux/pagemap.h> 4#include <linux/pagemap.h>
5#include <linux/xattr.h> 5#include <linux/xattr.h>
6#include <linux/reiserfs_xattr.h> 6#include "xattr.h"
7#include <asm/uaccess.h> 7#include <asm/uaccess.h>
8 8
9static int 9static int