aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2010-08-07 13:10:55 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2010-08-07 13:10:55 -0700
commita57f9a3e811cf1246b394f0cc667c6bc5a52e099 (patch)
tree488b4dd7cd061e7e0e059acb8443967e21051aab
parent09dc942c2a767e2d298f1cc9294bc19c7d7208c5 (diff)
parent89c0fd014d34d409a7b196667c2b9a4813b6c968 (diff)
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2: (45 commits) nilfs2: reject filesystem with unsupported block size nilfs2: avoid rec_len overflow with 64KB block size nilfs2: simplify nilfs_get_page function nilfs2: reject incompatible filesystem nilfs2: add feature set fields to super block nilfs2: clarify byte offset in super block format nilfs2: apply read-ahead for nilfs_btree_lookup_contig nilfs2: introduce check flag to btree node buffer nilfs2: add btree get block function with readahead option nilfs2: add read ahead mode to nilfs_btnode_submit_block nilfs2: fix buffer head leak in nilfs_btnode_submit_block nilfs2: eliminate inline keywords in btree implementation nilfs2: get maximum number of child nodes from bmap object nilfs2: reduce repetitive calculation of max number of child nodes nilfs2: optimize calculation of min/max number of btree node children nilfs2: remove redundant pointer checks in bmap lookup functions nilfs2: get rid of nilfs_bmap_union nilfs2: unify bmap set_target_v operations nilfs2: get rid of nilfs_btree uses nilfs2: get rid of nilfs_direct uses ...
-rw-r--r--Documentation/filesystems/nilfs2.txt12
-rw-r--r--fs/nilfs2/bmap.c6
-rw-r--r--fs/nilfs2/bmap.h16
-rw-r--r--fs/nilfs2/bmap_union.h42
-rw-r--r--fs/nilfs2/btnode.c23
-rw-r--r--fs/nilfs2/btnode.h4
-rw-r--r--fs/nilfs2/btree.c914
-rw-r--r--fs/nilfs2/btree.h12
-rw-r--r--fs/nilfs2/dir.c33
-rw-r--r--fs/nilfs2/direct.c96
-rw-r--r--fs/nilfs2/direct.h11
-rw-r--r--fs/nilfs2/gcinode.c17
-rw-r--r--fs/nilfs2/mdt.c1
-rw-r--r--fs/nilfs2/nilfs.h22
-rw-r--r--fs/nilfs2/page.c5
-rw-r--r--fs/nilfs2/page.h2
-rw-r--r--fs/nilfs2/recovery.c348
-rw-r--r--fs/nilfs2/segbuf.h24
-rw-r--r--fs/nilfs2/segment.c19
-rw-r--r--fs/nilfs2/segment.h10
-rw-r--r--fs/nilfs2/super.c333
-rw-r--r--fs/nilfs2/the_nilfs.c161
-rw-r--r--fs/nilfs2/the_nilfs.h23
-rw-r--r--include/linux/nilfs2_fs.h65
24 files changed, 1294 insertions, 905 deletions
diff --git a/Documentation/filesystems/nilfs2.txt b/Documentation/filesystems/nilfs2.txt
index d3e7673995e..d5c0cef38a7 100644
--- a/Documentation/filesystems/nilfs2.txt
+++ b/Documentation/filesystems/nilfs2.txt
@@ -49,7 +49,10 @@ Mount options
NILFS2 supports the following mount options:
(*) == default
-nobarrier Disables barriers.
+barrier(*) This enables/disables the use of write barriers. This
+nobarrier requires an IO stack which can support barriers, and
+ if nilfs gets an error on a barrier write, it will
+ disable again with a warning.
errors=continue Keep going on a filesystem error.
errors=remount-ro(*) Remount the filesystem read-only on an error.
errors=panic Panic and halt the machine if an error occurs.
@@ -74,9 +77,10 @@ norecovery Disable recovery of the filesystem on mount.
This disables every write access on the device for
read-only mounts or snapshots. This option will fail
for r/w mounts on an unclean volume.
-discard Issue discard/TRIM commands to the underlying block
- device when blocks are freed. This is useful for SSD
- devices and sparse/thinly-provisioned LUNs.
+discard This enables/disables the use of discard/TRIM commands.
+nodiscard(*) The discard/TRIM commands are sent to the underlying
+ block device when blocks are freed. This is useful
+ for SSD devices and sparse/thinly-provisioned LUNs.
NILFS2 usage
============
diff --git a/fs/nilfs2/bmap.c b/fs/nilfs2/bmap.c
index effdbdbe6c1..3dbdc1d356b 100644
--- a/fs/nilfs2/bmap.c
+++ b/fs/nilfs2/bmap.c
@@ -26,6 +26,8 @@
#include "nilfs.h"
#include "bmap.h"
#include "sb.h"
+#include "btree.h"
+#include "direct.h"
#include "btnode.h"
#include "mdt.h"
#include "dat.h"
@@ -533,7 +535,7 @@ void nilfs_bmap_init_gc(struct nilfs_bmap *bmap)
void nilfs_bmap_init_gcdat(struct nilfs_bmap *gcbmap, struct nilfs_bmap *bmap)
{
- memcpy(gcbmap, bmap, sizeof(union nilfs_bmap_union));
+ memcpy(gcbmap, bmap, sizeof(*bmap));
init_rwsem(&gcbmap->b_sem);
lockdep_set_class(&bmap->b_sem, &nilfs_bmap_dat_lock_key);
gcbmap->b_inode = &NILFS_BMAP_I(gcbmap)->vfs_inode;
@@ -541,7 +543,7 @@ void nilfs_bmap_init_gcdat(struct nilfs_bmap *gcbmap, struct nilfs_bmap *bmap)
void nilfs_bmap_commit_gcdat(struct nilfs_bmap *gcbmap, struct nilfs_bmap *bmap)
{
- memcpy(bmap, gcbmap, sizeof(union nilfs_bmap_union));
+ memcpy(bmap, gcbmap, sizeof(*bmap));
init_rwsem(&bmap->b_sem);
lockdep_set_class(&bmap->b_sem, &nilfs_bmap_dat_lock_key);
bmap->b_inode = &NILFS_BMAP_I(bmap)->vfs_inode;
diff --git a/fs/nilfs2/bmap.h b/fs/nilfs2/bmap.h
index 9980d7dbab9..a20569b1992 100644
--- a/fs/nilfs2/bmap.h
+++ b/fs/nilfs2/bmap.h
@@ -32,11 +32,6 @@
#define NILFS_BMAP_INVALID_PTR 0
-#define nilfs_bmap_dkey_to_key(dkey) le64_to_cpu(dkey)
-#define nilfs_bmap_key_to_dkey(key) cpu_to_le64(key)
-#define nilfs_bmap_dptr_to_ptr(dptr) le64_to_cpu(dptr)
-#define nilfs_bmap_ptr_to_dptr(ptr) cpu_to_le64(ptr)
-
#define nilfs_bmap_keydiff_abs(diff) ((diff) < 0 ? -(diff) : (diff))
@@ -71,7 +66,7 @@ struct nilfs_bmap_operations {
int (*bop_delete)(struct nilfs_bmap *, __u64);
void (*bop_clear)(struct nilfs_bmap *);
- int (*bop_propagate)(const struct nilfs_bmap *, struct buffer_head *);
+ int (*bop_propagate)(struct nilfs_bmap *, struct buffer_head *);
void (*bop_lookup_dirty_buffers)(struct nilfs_bmap *,
struct list_head *);
@@ -110,6 +105,7 @@ static inline int nilfs_bmap_is_new_ptr(unsigned long ptr)
* @b_last_allocated_ptr: last allocated ptr for data block
* @b_ptr_type: pointer type
* @b_state: state
+ * @b_nchildren_per_block: maximum number of child nodes for non-root nodes
*/
struct nilfs_bmap {
union {
@@ -123,6 +119,7 @@ struct nilfs_bmap {
__u64 b_last_allocated_ptr;
int b_ptr_type;
int b_state;
+ __u16 b_nchildren_per_block;
};
/* pointer type */
@@ -224,6 +221,13 @@ static inline void nilfs_bmap_abort_end_ptr(struct nilfs_bmap *bmap,
nilfs_dat_abort_end(dat, &req->bpr_req);
}
+static inline void nilfs_bmap_set_target_v(struct nilfs_bmap *bmap, __u64 key,
+ __u64 ptr)
+{
+ bmap->b_last_allocated_key = key;
+ bmap->b_last_allocated_ptr = ptr;
+}
+
__u64 nilfs_bmap_data_get_key(const struct nilfs_bmap *,
const struct buffer_head *);
diff --git a/fs/nilfs2/bmap_union.h b/fs/nilfs2/bmap_union.h
deleted file mode 100644
index d41509bff47..00000000000
--- a/fs/nilfs2/bmap_union.h
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- * bmap_union.h - NILFS block mapping.
- *
- * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
- *
- * Written by Koji Sato <koji@osrg.net>.
- */
-
-#ifndef _NILFS_BMAP_UNION_H
-#define _NILFS_BMAP_UNION_H
-
-#include "bmap.h"
-#include "direct.h"
-#include "btree.h"
-
-/**
- * nilfs_bmap_union -
- * @bi_bmap: bmap structure
- * @bi_btree: direct map structure
- * @bi_direct: B-tree structure
- */
-union nilfs_bmap_union {
- struct nilfs_bmap bi_bmap;
- struct nilfs_direct bi_direct;
- struct nilfs_btree bi_btree;
-};
-
-#endif /* _NILFS_BMAP_UNION_H */
diff --git a/fs/nilfs2/btnode.c b/fs/nilfs2/btnode.c
index 447ce47a330..f78ab1044d1 100644
--- a/fs/nilfs2/btnode.c
+++ b/fs/nilfs2/btnode.c
@@ -96,10 +96,12 @@ nilfs_btnode_create_block(struct address_space *btnc, __u64 blocknr)
}
int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
- sector_t pblocknr, struct buffer_head **pbh)
+ sector_t pblocknr, int mode,
+ struct buffer_head **pbh, sector_t *submit_ptr)
{
struct buffer_head *bh;
struct inode *inode = NILFS_BTNC_I(btnc);
+ struct page *page;
int err;
bh = nilfs_grab_buffer(inode, btnc, blocknr, 1 << BH_NILFS_Node);
@@ -107,6 +109,7 @@ int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
return -ENOMEM;
err = -EEXIST; /* internal code */
+ page = bh->b_page;
if (buffer_uptodate(bh) || buffer_dirty(bh))
goto found;
@@ -125,7 +128,16 @@ int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
}
}
}
- lock_buffer(bh);
+
+ if (mode == READA) {
+ if (pblocknr != *submit_ptr + 1 || !trylock_buffer(bh)) {
+ err = -EBUSY; /* internal code */
+ brelse(bh);
+ goto out_locked;
+ }
+ } else { /* mode == READ */
+ lock_buffer(bh);
+ }
if (buffer_uptodate(bh)) {
unlock_buffer(bh);
err = -EEXIST; /* internal code */
@@ -136,15 +148,16 @@ int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
bh->b_blocknr = pblocknr; /* set block address for read */
bh->b_end_io = end_buffer_read_sync;
get_bh(bh);
- submit_bh(READ, bh);
+ submit_bh(mode, bh);
bh->b_blocknr = blocknr; /* set back to the given block address */
+ *submit_ptr = pblocknr;
err = 0;
found:
*pbh = bh;
out_locked:
- unlock_page(bh->b_page);
- page_cache_release(bh->b_page);
+ unlock_page(page);
+ page_cache_release(page);
return err;
}
diff --git a/fs/nilfs2/btnode.h b/fs/nilfs2/btnode.h
index 07da83f0771..79037494f1e 100644
--- a/fs/nilfs2/btnode.h
+++ b/fs/nilfs2/btnode.h
@@ -42,8 +42,8 @@ void nilfs_btnode_cache_init(struct address_space *, struct backing_dev_info *);
void nilfs_btnode_cache_clear(struct address_space *);
struct buffer_head *nilfs_btnode_create_block(struct address_space *btnc,
__u64 blocknr);
-int nilfs_btnode_submit_block(struct address_space *, __u64, sector_t,
- struct buffer_head **);
+int nilfs_btnode_submit_block(struct address_space *, __u64, sector_t, int,
+ struct buffer_head **, sector_t *);
void nilfs_btnode_delete(struct buffer_head *);
int nilfs_btnode_prepare_change_key(struct address_space *,
struct nilfs_btnode_chkey_ctxt *);
diff --git a/fs/nilfs2/btree.c b/fs/nilfs2/btree.c
index b27a342c5af..300c2bc00c3 100644
--- a/fs/nilfs2/btree.c
+++ b/fs/nilfs2/btree.c
@@ -66,30 +66,10 @@ static void nilfs_btree_free_path(struct nilfs_btree_path *path)
/*
* B-tree node operations
*/
-static int nilfs_btree_get_block(const struct nilfs_btree *btree, __u64 ptr,
- struct buffer_head **bhp)
-{
- struct address_space *btnc =
- &NILFS_BMAP_I((struct nilfs_bmap *)btree)->i_btnode_cache;
- int err;
-
- err = nilfs_btnode_submit_block(btnc, ptr, 0, bhp);
- if (err)
- return err == -EEXIST ? 0 : err;
-
- wait_on_buffer(*bhp);
- if (!buffer_uptodate(*bhp)) {
- brelse(*bhp);
- return -EIO;
- }
- return 0;
-}
-
-static int nilfs_btree_get_new_block(const struct nilfs_btree *btree,
+static int nilfs_btree_get_new_block(const struct nilfs_bmap *btree,
__u64 ptr, struct buffer_head **bhp)
{
- struct address_space *btnc =
- &NILFS_BMAP_I((struct nilfs_bmap *)btree)->i_btnode_cache;
+ struct address_space *btnc = &NILFS_BMAP_I(btree)->i_btnode_cache;
struct buffer_head *bh;
bh = nilfs_btnode_create_block(btnc, ptr);
@@ -101,71 +81,55 @@ static int nilfs_btree_get_new_block(const struct nilfs_btree *btree,
return 0;
}
-static inline int
-nilfs_btree_node_get_flags(const struct nilfs_btree_node *node)
+static int nilfs_btree_node_get_flags(const struct nilfs_btree_node *node)
{
return node->bn_flags;
}
-static inline void
+static void
nilfs_btree_node_set_flags(struct nilfs_btree_node *node, int flags)
{
node->bn_flags = flags;
}
-static inline int nilfs_btree_node_root(const struct nilfs_btree_node *node)
+static int nilfs_btree_node_root(const struct nilfs_btree_node *node)
{
return nilfs_btree_node_get_flags(node) & NILFS_BTREE_NODE_ROOT;
}
-static inline int
-nilfs_btree_node_get_level(const struct nilfs_btree_node *node)
+static int nilfs_btree_node_get_level(const struct nilfs_btree_node *node)
{
return node->bn_level;
}
-static inline void
+static void
nilfs_btree_node_set_level(struct nilfs_btree_node *node, int level)
{
node->bn_level = level;
}
-static inline int
-nilfs_btree_node_get_nchildren(const struct nilfs_btree_node *node)
+static int nilfs_btree_node_get_nchildren(const struct nilfs_btree_node *node)
{
return le16_to_cpu(node->bn_nchildren);
}
-static inline void
+static void
nilfs_btree_node_set_nchildren(struct nilfs_btree_node *node, int nchildren)
{
node->bn_nchildren = cpu_to_le16(nchildren);
}
-static inline int nilfs_btree_node_size(const struct nilfs_btree *btree)
+static int nilfs_btree_node_size(const struct nilfs_bmap *btree)
{
- return 1 << btree->bt_bmap.b_inode->i_blkbits;
+ return 1 << btree->b_inode->i_blkbits;
}
-static inline int
-nilfs_btree_node_nchildren_min(const struct nilfs_btree_node *node,
- const struct nilfs_btree *btree)
+static int nilfs_btree_nchildren_per_block(const struct nilfs_bmap *btree)
{
- return nilfs_btree_node_root(node) ?
- NILFS_BTREE_ROOT_NCHILDREN_MIN :
- NILFS_BTREE_NODE_NCHILDREN_MIN(nilfs_btree_node_size(btree));
+ return btree->b_nchildren_per_block;
}
-static inline int
-nilfs_btree_node_nchildren_max(const struct nilfs_btree_node *node,
- const struct nilfs_btree *btree)
-{
- return nilfs_btree_node_root(node) ?
- NILFS_BTREE_ROOT_NCHILDREN_MAX :
- NILFS_BTREE_NODE_NCHILDREN_MAX(nilfs_btree_node_size(btree));
-}
-
-static inline __le64 *
+static __le64 *
nilfs_btree_node_dkeys(const struct nilfs_btree_node *node)
{
return (__le64 *)((char *)(node + 1) +
@@ -173,45 +137,40 @@ nilfs_btree_node_dkeys(const struct nilfs_btree_node *node)
0 : NILFS_BTREE_NODE_EXTRA_PAD_SIZE));
}
-static inline __le64 *
-nilfs_btree_node_dptrs(const struct nilfs_btree_node *node,
- const struct nilfs_btree *btree)
+static __le64 *
+nilfs_btree_node_dptrs(const struct nilfs_btree_node *node, int ncmax)
{
- return (__le64 *)(nilfs_btree_node_dkeys(node) +
- nilfs_btree_node_nchildren_max(node, btree));
+ return (__le64 *)(nilfs_btree_node_dkeys(node) + ncmax);
}
-static inline __u64
+static __u64
nilfs_btree_node_get_key(const struct nilfs_btree_node *node, int index)
{
- return nilfs_bmap_dkey_to_key(*(nilfs_btree_node_dkeys(node) + index));
+ return le64_to_cpu(*(nilfs_btree_node_dkeys(node) + index));
}
-static inline void
+static void
nilfs_btree_node_set_key(struct nilfs_btree_node *node, int index, __u64 key)
{
- *(nilfs_btree_node_dkeys(node) + index) = nilfs_bmap_key_to_dkey(key);
+ *(nilfs_btree_node_dkeys(node) + index) = cpu_to_le64(key);
}
-static inline __u64
-nilfs_btree_node_get_ptr(const struct nilfs_btree *btree,
- const struct nilfs_btree_node *node, int index)
+static __u64
+nilfs_btree_node_get_ptr(const struct nilfs_btree_node *node, int index,
+ int ncmax)
{
- return nilfs_bmap_dptr_to_ptr(*(nilfs_btree_node_dptrs(node, btree) +
- index));
+ return le64_to_cpu(*(nilfs_btree_node_dptrs(node, ncmax) + index));
}
-static inline void
-nilfs_btree_node_set_ptr(struct nilfs_btree *btree,
- struct nilfs_btree_node *node, int index, __u64 ptr)
+static void
+nilfs_btree_node_set_ptr(struct nilfs_btree_node *node, int index, __u64 ptr,
+ int ncmax)
{
- *(nilfs_btree_node_dptrs(node, btree) + index) =
- nilfs_bmap_ptr_to_dptr(ptr);
+ *(nilfs_btree_node_dptrs(node, ncmax) + index) = cpu_to_le64(ptr);
}
-static void nilfs_btree_node_init(struct nilfs_btree *btree,
- struct nilfs_btree_node *node,
- int flags, int level, int nchildren,
+static void nilfs_btree_node_init(struct nilfs_btree_node *node, int flags,
+ int level, int nchildren, int ncmax,
const __u64 *keys, const __u64 *ptrs)
{
__le64 *dkeys;
@@ -223,29 +182,28 @@ static void nilfs_btree_node_init(struct nilfs_btree *btree,
nilfs_btree_node_set_nchildren(node, nchildren);
dkeys = nilfs_btree_node_dkeys(node);
- dptrs = nilfs_btree_node_dptrs(node, btree);
+ dptrs = nilfs_btree_node_dptrs(node, ncmax);
for (i = 0; i < nchildren; i++) {
- dkeys[i] = nilfs_bmap_key_to_dkey(keys[i]);
- dptrs[i] = nilfs_bmap_ptr_to_dptr(ptrs[i]);
+ dkeys[i] = cpu_to_le64(keys[i]);
+ dptrs[i] = cpu_to_le64(ptrs[i]);
}
}
/* Assume the buffer heads corresponding to left and right are locked. */
-static void nilfs_btree_node_move_left(struct nilfs_btree *btree,
- struct nilfs_btree_node *left,
+static void nilfs_btree_node_move_left(struct nilfs_btree_node *left,
struct nilfs_btree_node *right,
- int n)
+ int n, int lncmax, int rncmax)
{
__le64 *ldkeys, *rdkeys;
__le64 *ldptrs, *rdptrs;
int lnchildren, rnchildren;
ldkeys = nilfs_btree_node_dkeys(left);
- ldptrs = nilfs_btree_node_dptrs(left, btree);
+ ldptrs = nilfs_btree_node_dptrs(left, lncmax);
lnchildren = nilfs_btree_node_get_nchildren(left);
rdkeys = nilfs_btree_node_dkeys(right);
- rdptrs = nilfs_btree_node_dptrs(right, btree);
+ rdptrs = nilfs_btree_node_dptrs(right, rncmax);
rnchildren = nilfs_btree_node_get_nchildren(right);
memcpy(ldkeys + lnchildren, rdkeys, n * sizeof(*rdkeys));
@@ -260,21 +218,20 @@ static void nilfs_btree_node_move_left(struct nilfs_btree *btree,
}
/* Assume that the buffer heads corresponding to left and right are locked. */
-static void nilfs_btree_node_move_right(struct nilfs_btree *btree,
- struct nilfs_btree_node *left,
+static void nilfs_btree_node_move_right(struct nilfs_btree_node *left,
struct nilfs_btree_node *right,
- int n)
+ int n, int lncmax, int rncmax)
{
__le64 *ldkeys, *rdkeys;
__le64 *ldptrs, *rdptrs;
int lnchildren, rnchildren;
ldkeys = nilfs_btree_node_dkeys(left);
- ldptrs = nilfs_btree_node_dptrs(left, btree);
+ ldptrs = nilfs_btree_node_dptrs(left, lncmax);
lnchildren = nilfs_btree_node_get_nchildren(left);
rdkeys = nilfs_btree_node_dkeys(right);
- rdptrs = nilfs_btree_node_dptrs(right, btree);
+ rdptrs = nilfs_btree_node_dptrs(right, rncmax);
rnchildren = nilfs_btree_node_get_nchildren(right);
memmove(rdkeys + n, rdkeys, rnchildren * sizeof(*rdkeys));
@@ -289,16 +246,15 @@ static void nilfs_btree_node_move_right(struct nilfs_btree *btree,
}
/* Assume that the buffer head corresponding to node is locked. */
-static void nilfs_btree_node_insert(struct nilfs_btree *btree,
- struct nilfs_btree_node *node,
- __u64 key, __u64 ptr, int index)
+static void nilfs_btree_node_insert(struct nilfs_btree_node *node, int index,
+ __u64 key, __u64 ptr, int ncmax)
{
__le64 *dkeys;
__le64 *dptrs;
int nchildren;
dkeys = nilfs_btree_node_dkeys(node);
- dptrs = nilfs_btree_node_dptrs(node, btree);
+ dptrs = nilfs_btree_node_dptrs(node, ncmax);
nchildren = nilfs_btree_node_get_nchildren(node);
if (index < nchildren) {
memmove(dkeys + index + 1, dkeys + index,
@@ -306,16 +262,15 @@ static void nilfs_btree_node_insert(struct nilfs_btree *btree,
memmove(dptrs + index + 1, dptrs + index,
(nchildren - index) * sizeof(*dptrs));
}
- dkeys[index] = nilfs_bmap_key_to_dkey(key);
- dptrs[index] = nilfs_bmap_ptr_to_dptr(ptr);
+ dkeys[index] = cpu_to_le64(key);
+ dptrs[index] = cpu_to_le64(ptr);
nchildren++;
nilfs_btree_node_set_nchildren(node, nchildren);
}
/* Assume that the buffer head corresponding to node is locked. */
-static void nilfs_btree_node_delete(struct nilfs_btree *btree,
- struct nilfs_btree_node *node,
- __u64 *keyp, __u64 *ptrp, int index)
+static void nilfs_btree_node_delete(struct nilfs_btree_node *node, int index,
+ __u64 *keyp, __u64 *ptrp, int ncmax)
{
__u64 key;
__u64 ptr;
@@ -324,9 +279,9 @@ static void nilfs_btree_node_delete(struct nilfs_btree *btree,
int nchildren;
dkeys = nilfs_btree_node_dkeys(node);
- dptrs = nilfs_btree_node_dptrs(node, btree);
- key = nilfs_bmap_dkey_to_key(dkeys[index]);
- ptr = nilfs_bmap_dptr_to_ptr(dptrs[index]);
+ dptrs = nilfs_btree_node_dptrs(node, ncmax);
+ key = le64_to_cpu(dkeys[index]);
+ ptr = le64_to_cpu(dptrs[index]);
nchildren = nilfs_btree_node_get_nchildren(node);
if (keyp != NULL)
*keyp = key;
@@ -382,40 +337,92 @@ static int nilfs_btree_node_lookup(const struct nilfs_btree_node *node,
return s == 0;
}
-static inline struct nilfs_btree_node *
-nilfs_btree_get_root(const struct nilfs_btree *btree)
+/**
+ * nilfs_btree_node_broken - verify consistency of btree node
+ * @node: btree node block to be examined
+ * @size: node size (in bytes)
+ * @blocknr: block number
+ *
+ * Return Value: If node is broken, 1 is returned. Otherwise, 0 is returned.
+ */
+static int nilfs_btree_node_broken(const struct nilfs_btree_node *node,
+ size_t size, sector_t blocknr)
{
- return (struct nilfs_btree_node *)btree->bt_bmap.b_u.u_data;
+ int level, flags, nchildren;
+ int ret = 0;
+
+ level = nilfs_btree_node_get_level(node);
+ flags = nilfs_btree_node_get_flags(node);
+ nchildren = nilfs_btree_node_get_nchildren(node);
+
+ if (unlikely(level < NILFS_BTREE_LEVEL_NODE_MIN ||
+ level >= NILFS_BTREE_LEVEL_MAX ||
+ (flags & NILFS_BTREE_NODE_ROOT) ||
+ nchildren < 0 ||
+ nchildren > NILFS_BTREE_NODE_NCHILDREN_MAX(size))) {
+ printk(KERN_CRIT "NILFS: bad btree node (blocknr=%llu): "
+ "level = %d, flags = 0x%x, nchildren = %d\n",
+ (unsigned long long)blocknr, level, flags, nchildren);
+ ret = 1;
+ }
+ return ret;
}
-static inline struct nilfs_btree_node *
+int nilfs_btree_broken_node_block(struct buffer_head *bh)
+{
+ int ret;
+
+ if (buffer_nilfs_checked(bh))
+ return 0;
+
+ ret = nilfs_btree_node_broken((struct nilfs_btree_node *)bh->b_data,
+ bh->b_size, bh->b_blocknr);
+ if (likely(!ret))
+ set_buffer_nilfs_checked(bh);
+ return ret;
+}
+
+static struct nilfs_btree_node *
+nilfs_btree_get_root(const struct nilfs_bmap *btree)
+{
+ return (struct nilfs_btree_node *)btree->b_u.u_data;
+}
+
+static struct nilfs_btree_node *
nilfs_btree_get_nonroot_node(const struct nilfs_btree_path *path, int level)
{
return (struct nilfs_btree_node *)path[level].bp_bh->b_data;
}
-static inline struct nilfs_btree_node *
+static struct nilfs_btree_node *
nilfs_btree_get_sib_node(const struct nilfs_btree_path *path, int level)
{
return (struct nilfs_btree_node *)path[level].bp_sib_bh->b_data;
}
-static inline int nilfs_btree_height(const struct nilfs_btree *btree)
+static int nilfs_btree_height(const struct nilfs_bmap *btree)
{
return nilfs_btree_node_get_level(nilfs_btree_get_root(btree)) + 1;
}
-static inline struct nilfs_btree_node *
-nilfs_btree_get_node(const struct nilfs_btree *btree,
+static struct nilfs_btree_node *
+nilfs_btree_get_node(const struct nilfs_bmap *btree,
const struct nilfs_btree_path *path,
- int level)
+ int level, int *ncmaxp)
{
- return (level == nilfs_btree_height(btree) - 1) ?
- nilfs_btree_get_root(btree) :
- nilfs_btree_get_nonroot_node(path, level);
+ struct nilfs_btree_node *node;
+
+ if (level == nilfs_btree_height(btree) - 1) {
+ node = nilfs_btree_get_root(btree);
+ *ncmaxp = NILFS_BTREE_ROOT_NCHILDREN_MAX;
+ } else {
+ node = nilfs_btree_get_nonroot_node(path, level);
+ *ncmaxp = nilfs_btree_nchildren_per_block(btree);
+ }
+ return node;
}
-static inline int
+static int
nilfs_btree_bad_node(struct nilfs_btree_node *node, int level)
{
if (unlikely(nilfs_btree_node_get_level(node) != level)) {
@@ -427,13 +434,83 @@ nilfs_btree_bad_node(struct nilfs_btree_node *node, int level)
return 0;
}
-static int nilfs_btree_do_lookup(const struct nilfs_btree *btree,
+struct nilfs_btree_readahead_info {
+ struct nilfs_btree_node *node; /* parent node */
+ int max_ra_blocks; /* max nof blocks to read ahead */
+ int index; /* current index on the parent node */
+ int ncmax; /* nof children in the parent node */
+};
+
+static int __nilfs_btree_get_block(const struct nilfs_bmap *btree, __u64 ptr,
+ struct buffer_head **bhp,
+ const struct nilfs_btree_readahead_info *ra)
+{
+ struct address_space *btnc = &NILFS_BMAP_I(btree)->i_btnode_cache;
+ struct buffer_head *bh, *ra_bh;
+ sector_t submit_ptr = 0;
+ int ret;
+
+ ret = nilfs_btnode_submit_block(btnc, ptr, 0, READ, &bh, &submit_ptr);
+ if (ret) {
+ if (ret != -EEXIST)
+ return ret;
+ goto out_check;
+ }
+
+ if (ra) {
+ int i, n;
+ __u64 ptr2;
+
+ /* read ahead sibling nodes */
+ for (n = ra->max_ra_blocks, i = ra->index + 1;
+ n > 0 && i < ra->ncmax; n--, i++) {
+ ptr2 = nilfs_btree_node_get_ptr(ra->node, i, ra->ncmax);
+
+ ret = nilfs_btnode_submit_block(btnc, ptr2, 0, READA,
+ &ra_bh, &submit_ptr);
+ if (likely(!ret || ret == -EEXIST))
+ brelse(ra_bh);
+ else if (ret != -EBUSY)
+ break;
+ if (!buffer_locked(bh))
+ goto out_no_wait;
+ }
+ }
+
+ wait_on_buffer(bh);
+
+ out_no_wait:
+ if (!buffer_uptodate(bh)) {
+ brelse(bh);
+ return -EIO;
+ }
+
+ out_check:
+ if (nilfs_btree_broken_node_block(bh)) {
+ clear_buffer_uptodate(bh);
+ brelse(bh);
+ return -EINVAL;
+ }
+
+ *bhp = bh;
+ return 0;
+}
+
+static int nilfs_btree_get_block(const struct nilfs_bmap *btree, __u64 ptr,
+ struct buffer_head **bhp)
+{
+ return __nilfs_btree_get_block(btree, ptr, bhp, NULL);
+}
+
+static int nilfs_btree_do_lookup(const struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
- __u64 key, __u64 *ptrp, int minlevel)
+ __u64 key, __u64 *ptrp, int minlevel,
+ int readahead)
{
struct nilfs_btree_node *node;
+ struct nilfs_btree_readahead_info p, *ra;
__u64 ptr;
- int level, index, found, ret;
+ int level, index, found, ncmax, ret;
node = nilfs_btree_get_root(btree);
level = nilfs_btree_node_get_level(node);
@@ -441,14 +518,27 @@ static int nilfs_btree_do_lookup(const struct nilfs_btree *btree,
return -ENOENT;
found = nilfs_btree_node_lookup(node, key, &index);
- ptr = nilfs_btree_node_get_ptr(btree, node, index);
+ ptr = nilfs_btree_node_get_ptr(node, index,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
path[level].bp_bh = NULL;
path[level].bp_index = index;
- for (level--; level >= minlevel; level--) {
- ret = nilfs_btree_get_block(btree, ptr, &path[level].bp_bh);
+ ncmax = nilfs_btree_nchildren_per_block(btree);
+
+ while (--level >= minlevel) {
+ ra = NULL;
+ if (level == NILFS_BTREE_LEVEL_NODE_MIN && readahead) {
+ p.node = nilfs_btree_get_node(btree, path, level + 1,
+ &p.ncmax);
+ p.index = index;
+ p.max_ra_blocks = 7;
+ ra = &p;
+ }
+ ret = __nilfs_btree_get_block(btree, ptr, &path[level].bp_bh,
+ ra);
if (ret < 0)
return ret;
+
node = nilfs_btree_get_nonroot_node(path, level);
if (nilfs_btree_bad_node(node, level))
return -EINVAL;
@@ -456,9 +546,9 @@ static int nilfs_btree_do_lookup(const struct nilfs_btree *btree,
found = nilfs_btree_node_lookup(node, key, &index);
else
index = 0;
- if (index < nilfs_btree_node_nchildren_max(node, btree))
- ptr = nilfs_btree_node_get_ptr(btree, node, index);
- else {
+ if (index < ncmax) {
+ ptr = nilfs_btree_node_get_ptr(node, index, ncmax);
+ } else {
WARN_ON(found || level != NILFS_BTREE_LEVEL_NODE_MIN);
/* insert */
ptr = NILFS_BMAP_INVALID_PTR;
@@ -474,22 +564,24 @@ static int nilfs_btree_do_lookup(const struct nilfs_btree *btree,
return 0;
}
-static int nilfs_btree_do_lookup_last(const struct nilfs_btree *btree,
+static int nilfs_btree_do_lookup_last(const struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
__u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node;
__u64 ptr;
- int index, level, ret;
+ int index, level, ncmax, ret;
node = nilfs_btree_get_root(btree);
index = nilfs_btree_node_get_nchildren(node) - 1;
if (index < 0)
return -ENOENT;
level = nilfs_btree_node_get_level(node);
- ptr = nilfs_btree_node_get_ptr(btree, node, index);
+ ptr = nilfs_btree_node_get_ptr(node, index,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
path[level].bp_bh = NULL;
path[level].bp_index = index;
+ ncmax = nilfs_btree_nchildren_per_block(btree);
for (level--; level > 0; level--) {
ret = nilfs_btree_get_block(btree, ptr, &path[level].bp_bh);
@@ -499,7 +591,7 @@ static int nilfs_btree_do_lookup_last(const struct nilfs_btree *btree,
if (nilfs_btree_bad_node(node, level))
return -EINVAL;
index = nilfs_btree_node_get_nchildren(node) - 1;
- ptr = nilfs_btree_node_get_ptr(btree, node, index);
+ ptr = nilfs_btree_node_get_ptr(node, index, ncmax);
path[level].bp_index = index;
}
@@ -511,51 +603,45 @@ static int nilfs_btree_do_lookup_last(const struct nilfs_btree *btree,
return 0;
}
-static int nilfs_btree_lookup(const struct nilfs_bmap *bmap,
+static int nilfs_btree_lookup(const struct nilfs_bmap *btree,
__u64 key, int level, __u64 *ptrp)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
- __u64 ptr;
int ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
- ret = nilfs_btree_do_lookup(btree, path, key, &ptr, level);
-
- if (ptrp != NULL)
- *ptrp = ptr;
+ ret = nilfs_btree_do_lookup(btree, path, key, ptrp, level, 0);
nilfs_btree_free_path(path);
return ret;
}
-static int nilfs_btree_lookup_contig(const struct nilfs_bmap *bmap,
+static int nilfs_btree_lookup_contig(const struct nilfs_bmap *btree,
__u64 key, __u64 *ptrp, unsigned maxblocks)
{
- struct nilfs_btree *btree = (struct nilfs_btree *)bmap;
struct nilfs_btree_path *path;
struct nilfs_btree_node *node;
struct inode *dat = NULL;
__u64 ptr, ptr2;
sector_t blocknr;
int level = NILFS_BTREE_LEVEL_NODE_MIN;
- int ret, cnt, index, maxlevel;
+ int ret, cnt, index, maxlevel, ncmax;
+ struct nilfs_btree_readahead_info p;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
- ret = nilfs_btree_do_lookup(btree, path, key, &ptr, level);
+ ret = nilfs_btree_do_lookup(btree, path, key, &ptr, level, 1);
if (ret < 0)
goto out;
- if (NILFS_BMAP_USE_VBN(bmap)) {
- dat = nilfs_bmap_get_dat(bmap);
+ if (NILFS_BMAP_USE_VBN(btree)) {
+ dat = nilfs_bmap_get_dat(btree);
ret = nilfs_dat_translate(dat, ptr, &blocknr);
if (ret < 0)
goto out;
@@ -566,14 +652,14 @@ static int nilfs_btree_lookup_contig(const struct nilfs_bmap *bmap,
goto end;
maxlevel = nilfs_btree_height(btree) - 1;
- node = nilfs_btree_get_node(btree, path, level);
+ node = nilfs_btree_get_node(btree, path, level, &ncmax);
index = path[level].bp_index + 1;
for (;;) {
while (index < nilfs_btree_node_get_nchildren(node)) {
if (nilfs_btree_node_get_key(node, index) !=
key + cnt)
goto end;
- ptr2 = nilfs_btree_node_get_ptr(btree, node, index);
+ ptr2 = nilfs_btree_node_get_ptr(node, index, ncmax);
if (dat) {
ret = nilfs_dat_translate(dat, ptr2, &blocknr);
if (ret < 0)
@@ -589,20 +675,24 @@ static int nilfs_btree_lookup_contig(const struct nilfs_bmap *bmap,
break;
/* look-up right sibling node */
- node = nilfs_btree_get_node(btree, path, level + 1);
- index = path[level + 1].bp_index + 1;
- if (index >= nilfs_btree_node_get_nchildren(node) ||
- nilfs_btree_node_get_key(node, index) != key + cnt)
+ p.node = nilfs_btree_get_node(btree, path, level + 1, &p.ncmax);
+ p.index = path[level + 1].bp_index + 1;
+ p.max_ra_blocks = 7;
+ if (p.index >= nilfs_btree_node_get_nchildren(p.node) ||
+ nilfs_btree_node_get_key(p.node, p.index) != key + cnt)
break;
- ptr2 = nilfs_btree_node_get_ptr(btree, node, index);
- path[level + 1].bp_index = index;
+ ptr2 = nilfs_btree_node_get_ptr(p.node, p.index, p.ncmax);
+ path[level + 1].bp_index = p.index;
brelse(path[level].bp_bh);
path[level].bp_bh = NULL;
- ret = nilfs_btree_get_block(btree, ptr2, &path[level].bp_bh);
+
+ ret = __nilfs_btree_get_block(btree, ptr2, &path[level].bp_bh,
+ &p);
if (ret < 0)
goto out;
node = nilfs_btree_get_nonroot_node(path, level);
+ ncmax = nilfs_btree_nchildren_per_block(btree);
index = 0;
path[level].bp_index = index;
}
@@ -614,7 +704,7 @@ static int nilfs_btree_lookup_contig(const struct nilfs_bmap *bmap,
return ret;
}
-static void nilfs_btree_promote_key(struct nilfs_btree *btree,
+static void nilfs_btree_promote_key(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 key)
{
@@ -636,16 +726,18 @@ static void nilfs_btree_promote_key(struct nilfs_btree *btree,
}
}
-static void nilfs_btree_do_insert(struct nilfs_btree *btree,
+static void nilfs_btree_do_insert(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node;
+ int ncblk;
if (level < nilfs_btree_height(btree) - 1) {
node = nilfs_btree_get_nonroot_node(path, level);
- nilfs_btree_node_insert(btree, node, *keyp, *ptrp,
- path[level].bp_index);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
+ nilfs_btree_node_insert(node, path[level].bp_index,
+ *keyp, *ptrp, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -655,22 +747,24 @@ static void nilfs_btree_do_insert(struct nilfs_btree *btree,
0));
} else {
node = nilfs_btree_get_root(btree);
- nilfs_btree_node_insert(btree, node, *keyp, *ptrp,
- path[level].bp_index);
+ nilfs_btree_node_insert(node, path[level].bp_index,
+ *keyp, *ptrp,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
}
}
-static void nilfs_btree_carry_left(struct nilfs_btree *btree,
+static void nilfs_btree_carry_left(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *left;
- int nchildren, lnchildren, n, move;
+ int nchildren, lnchildren, n, move, ncblk;
node = nilfs_btree_get_nonroot_node(path, level);
left = nilfs_btree_get_sib_node(path, level);
nchildren = nilfs_btree_node_get_nchildren(node);
lnchildren = nilfs_btree_node_get_nchildren(left);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
move = 0;
n = (nchildren + lnchildren + 1) / 2 - lnchildren;
@@ -680,7 +774,7 @@ static void nilfs_btree_carry_left(struct nilfs_btree *btree,
move = 1;
}
- nilfs_btree_node_move_left(btree, left, node, n);
+ nilfs_btree_node_move_left(left, node, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -705,17 +799,18 @@ static void nilfs_btree_carry_left(struct nilfs_btree *btree,
nilfs_btree_do_insert(btree, path, level, keyp, ptrp);
}
-static void nilfs_btree_carry_right(struct nilfs_btree *btree,
+static void nilfs_btree_carry_right(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *right;
- int nchildren, rnchildren, n, move;
+ int nchildren, rnchildren, n, move, ncblk;
node = nilfs_btree_get_nonroot_node(path, level);
right = nilfs_btree_get_sib_node(path, level);
nchildren = nilfs_btree_node_get_nchildren(node);
rnchildren = nilfs_btree_node_get_nchildren(right);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
move = 0;
n = (nchildren + rnchildren + 1) / 2 - rnchildren;
@@ -725,7 +820,7 @@ static void nilfs_btree_carry_right(struct nilfs_btree *btree,
move = 1;
}
- nilfs_btree_node_move_right(btree, node, right, n);
+ nilfs_btree_node_move_right(node, right, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -751,18 +846,19 @@ static void nilfs_btree_carry_right(struct nilfs_btree *btree,
nilfs_btree_do_insert(btree, path, level, keyp, ptrp);
}
-static void nilfs_btree_split(struct nilfs_btree *btree,
+static void nilfs_btree_split(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *right;
__u64 newkey;
__u64 newptr;
- int nchildren, n, move;
+ int nchildren, n, move, ncblk;
node = nilfs_btree_get_nonroot_node(path, level);
right = nilfs_btree_get_sib_node(path, level);
nchildren = nilfs_btree_node_get_nchildren(node);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
move = 0;
n = (nchildren + 1) / 2;
@@ -771,7 +867,7 @@ static void nilfs_btree_split(struct nilfs_btree *btree,
move = 1;
}
- nilfs_btree_node_move_right(btree, node, right, n);
+ nilfs_btree_node_move_right(node, right, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -783,8 +879,8 @@ static void nilfs_btree_split(struct nilfs_btree *btree,
if (move) {
path[level].bp_index -= nilfs_btree_node_get_nchildren(node);
- nilfs_btree_node_insert(btree, right, *keyp, *ptrp,
- path[level].bp_index);
+ nilfs_btree_node_insert(right, path[level].bp_index,
+ *keyp, *ptrp, ncblk);
*keyp = nilfs_btree_node_get_key(right, 0);
*ptrp = path[level].bp_newreq.bpr_ptr;
@@ -805,19 +901,21 @@ static void nilfs_btree_split(struct nilfs_btree *btree,
path[level + 1].bp_index++;
}
-static void nilfs_btree_grow(struct nilfs_btree *btree,
+static void nilfs_btree_grow(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *root, *child;
- int n;
+ int n, ncblk;
root = nilfs_btree_get_root(btree);
child = nilfs_btree_get_sib_node(path, level);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
n = nilfs_btree_node_get_nchildren(root);
- nilfs_btree_node_move_right(btree, root, child, n);
+ nilfs_btree_node_move_right(root, child, n,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX, ncblk);
nilfs_btree_node_set_level(root, level + 1);
if (!buffer_dirty(path[level].bp_sib_bh))
@@ -832,11 +930,11 @@ static void nilfs_btree_grow(struct nilfs_btree *btree,
*ptrp = path[level].bp_newreq.bpr_ptr;
}
-static __u64 nilfs_btree_find_near(const struct nilfs_btree *btree,
+static __u64 nilfs_btree_find_near(const struct nilfs_bmap *btree,
const struct nilfs_btree_path *path)
{
struct nilfs_btree_node *node;
- int level;
+ int level, ncmax;
if (path == NULL)
return NILFS_BMAP_INVALID_PTR;
@@ -844,29 +942,30 @@ static __u64 nilfs_btree_find_near(const struct nilfs_btree *btree,
/* left sibling */
level = NILFS_BTREE_LEVEL_NODE_MIN;
if (path[level].bp_index > 0) {
- node = nilfs_btree_get_node(btree, path, level);
- return nilfs_btree_node_get_ptr(btree, node,
- path[level].bp_index - 1);
+ node = nilfs_btree_get_node(btree, path, level, &ncmax);
+ return nilfs_btree_node_get_ptr(node,
+ path[level].bp_index - 1,
+ ncmax);
}
/* parent */
level = NILFS_BTREE_LEVEL_NODE_MIN + 1;
if (level <= nilfs_btree_height(btree) - 1) {
- node = nilfs_btree_get_node(btree, path, level);
- return nilfs_btree_node_get_ptr(btree, node,
- path[level].bp_index);
+ node = nilfs_btree_get_node(btree, path, level, &ncmax);
+ return nilfs_btree_node_get_ptr(node, path[level].bp_index,
+ ncmax);
}
return NILFS_BMAP_INVALID_PTR;
}
-static __u64 nilfs_btree_find_target_v(const struct nilfs_btree *btree,
+static __u64 nilfs_btree_find_target_v(const struct nilfs_bmap *btree,
const struct nilfs_btree_path *path,
__u64 key)
{
__u64 ptr;
- ptr = nilfs_bmap_find_target_seq(&btree->bt_bmap, key);
+ ptr = nilfs_bmap_find_target_seq(btree, key);
if (ptr != NILFS_BMAP_INVALID_PTR)
/* sequential access */
return ptr;
@@ -877,17 +976,10 @@ static __u64 nilfs_btree_find_target_v(const struct nilfs_btree *btree,
return ptr;
}
/* block group */
- return nilfs_bmap_find_target_in_group(&btree->bt_bmap);
-}
-
-static void nilfs_btree_set_target_v(struct nilfs_btree *btree, __u64 key,
- __u64 ptr)
-{
- btree->bt_bmap.b_last_allocated_key = key;
- btree->bt_bmap.b_last_allocated_ptr = ptr;
+ return nilfs_bmap_find_target_in_group(btree);
}
-static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
+static int nilfs_btree_prepare_insert(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int *levelp, __u64 key, __u64 ptr,
struct nilfs_bmap_stats *stats)
@@ -895,79 +987,78 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
struct buffer_head *bh;
struct nilfs_btree_node *node, *parent, *sib;
__u64 sibptr;
- int pindex, level, ret;
+ int pindex, level, ncmax, ncblk, ret;
struct inode *dat = NULL;
stats->bs_nblocks = 0;
level = NILFS_BTREE_LEVEL_DATA;
/* allocate a new ptr for data block */
- if (NILFS_BMAP_USE_VBN(&btree->bt_bmap)) {
+ if (NILFS_BMAP_USE_VBN(btree)) {
path[level].bp_newreq.bpr_ptr =
nilfs_btree_find_target_v(btree, path, key);
- dat = nilfs_bmap_get_dat(&btree->bt_bmap);
+ dat = nilfs_bmap_get_dat(btree);
}
- ret = nilfs_bmap_prepare_alloc_ptr(&btree->bt_bmap,
- &path[level].bp_newreq, dat);
+ ret = nilfs_bmap_prepare_alloc_ptr(btree, &path[level].bp_newreq, dat);
if (ret < 0)
goto err_out_data;
+ ncblk = nilfs_btree_nchildren_per_block(btree);
+
for (level = NILFS_BTREE_LEVEL_NODE_MIN;
level < nilfs_btree_height(btree) - 1;
level++) {
node = nilfs_btree_get_nonroot_node(path, level);
- if (nilfs_btree_node_get_nchildren(node) <
- nilfs_btree_node_nchildren_max(node, btree)) {
+ if (nilfs_btree_node_get_nchildren(node) < ncblk) {
path[level].bp_op = nilfs_btree_do_insert;
stats->bs_nblocks++;
goto out;
}
- parent = nilfs_btree_get_node(btree, path, level + 1);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
pindex = path[level + 1].bp_index;
/* left sibling */
if (pindex > 0) {
- sibptr = nilfs_btree_node_get_ptr(btree, parent,
- pindex - 1);
+ sibptr = nilfs_btree_node_get_ptr(parent, pindex - 1,
+ ncmax);
ret = nilfs_btree_get_block(btree, sibptr, &bh);
if (ret < 0)
goto err_out_child_node;
sib = (struct nilfs_btree_node *)bh->b_data;
- if (nilfs_btree_node_get_nchildren(sib) <
- nilfs_btree_node_nchildren_max(sib, btree)) {
+ if (nilfs_btree_node_get_nchildren(sib) < ncblk) {
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_carry_left;
stats->bs_nblocks++;
goto out;
- } else
+ } else {
brelse(bh);
+ }
}
/* right sibling */
- if (pindex <
- nilfs_btree_node_get_nchildren(parent) - 1) {
- sibptr = nilfs_btree_node_get_ptr(btree, parent,
- pindex + 1);
+ if (pindex < nilfs_btree_node_get_nchildren(parent) - 1) {
+ sibptr = nilfs_btree_node_get_ptr(parent, pindex + 1,
+ ncmax);
ret = nilfs_btree_get_block(btree, sibptr, &bh);
if (ret < 0)
goto err_out_child_node;
sib = (struct nilfs_btree_node *)bh->b_data;
- if (nilfs_btree_node_get_nchildren(sib) <
- nilfs_btree_node_nchildren_max(sib, btree)) {
+ if (nilfs_btree_node_get_nchildren(sib) < ncblk) {
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_carry_right;
stats->bs_nblocks++;
goto out;
- } else
+ } else {
brelse(bh);
+ }
}
/* split */
path[level].bp_newreq.bpr_ptr =
path[level - 1].bp_newreq.bpr_ptr + 1;
- ret = nilfs_bmap_prepare_alloc_ptr(&btree->bt_bmap,
+ ret = nilfs_bmap_prepare_alloc_ptr(btree,
&path[level].bp_newreq, dat);
if (ret < 0)
goto err_out_child_node;
@@ -979,9 +1070,8 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
stats->bs_nblocks++;
- nilfs_btree_node_init(btree,
- (struct nilfs_btree_node *)bh->b_data,
- 0, level, 0, NULL, NULL);
+ sib = (struct nilfs_btree_node *)bh->b_data;
+ nilfs_btree_node_init(sib, 0, level, 0, ncblk, NULL, NULL);
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_split;
}
@@ -989,7 +1079,7 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
/* root */
node = nilfs_btree_get_root(btree);
if (nilfs_btree_node_get_nchildren(node) <
- nilfs_btree_node_nchildren_max(node, btree)) {
+ NILFS_BTREE_ROOT_NCHILDREN_MAX) {
path[level].bp_op = nilfs_btree_do_insert;
stats->bs_nblocks++;
goto out;
@@ -997,8 +1087,7 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
/* grow */
path[level].bp_newreq.bpr_ptr = path[level - 1].bp_newreq.bpr_ptr + 1;
- ret = nilfs_bmap_prepare_alloc_ptr(&btree->bt_bmap,
- &path[level].bp_newreq, dat);
+ ret = nilfs_bmap_prepare_alloc_ptr(btree, &path[level].bp_newreq, dat);
if (ret < 0)
goto err_out_child_node;
ret = nilfs_btree_get_new_block(btree, path[level].bp_newreq.bpr_ptr,
@@ -1006,8 +1095,8 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
if (ret < 0)
goto err_out_curr_node;
- nilfs_btree_node_init(btree, (struct nilfs_btree_node *)bh->b_data,
- 0, level, 0, NULL, NULL);
+ nilfs_btree_node_init((struct nilfs_btree_node *)bh->b_data,
+ 0, level, 0, ncblk, NULL, NULL);
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_grow;
@@ -1024,25 +1113,22 @@ static int nilfs_btree_prepare_insert(struct nilfs_btree *btree,
/* error */
err_out_curr_node:
- nilfs_bmap_abort_alloc_ptr(&btree->bt_bmap, &path[level].bp_newreq,
- dat);
+ nilfs_bmap_abort_alloc_ptr(btree, &path[level].bp_newreq, dat);
err_out_child_node:
for (level--; level > NILFS_BTREE_LEVEL_DATA; level--) {
nilfs_btnode_delete(path[level].bp_sib_bh);
- nilfs_bmap_abort_alloc_ptr(&btree->bt_bmap,
- &path[level].bp_newreq, dat);
+ nilfs_bmap_abort_alloc_ptr(btree, &path[level].bp_newreq, dat);
}
- nilfs_bmap_abort_alloc_ptr(&btree->bt_bmap, &path[level].bp_newreq,
- dat);
+ nilfs_bmap_abort_alloc_ptr(btree, &path[level].bp_newreq, dat);
err_out_data:
*levelp = level;
stats->bs_nblocks = 0;
return ret;
}
-static void nilfs_btree_commit_insert(struct nilfs_btree *btree,
+static void nilfs_btree_commit_insert(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int maxlevel, __u64 key, __u64 ptr)
{
@@ -1051,35 +1137,33 @@ static void nilfs_btree_commit_insert(struct nilfs_btree *btree,
set_buffer_nilfs_volatile((struct buffer_head *)((unsigned long)ptr));
ptr = path[NILFS_BTREE_LEVEL_DATA].bp_newreq.bpr_ptr;
- if (NILFS_BMAP_USE_VBN(&btree->bt_bmap)) {
- nilfs_btree_set_target_v(btree, key, ptr);
- dat = nilfs_bmap_get_dat(&btree->bt_bmap);
+ if (NILFS_BMAP_USE_VBN(btree)) {
+ nilfs_bmap_set_target_v(btree, key, ptr);
+ dat = nilfs_bmap_get_dat(btree);
}
for (level = NILFS_BTREE_LEVEL_NODE_MIN; level <= maxlevel; level++) {
- nilfs_bmap_commit_alloc_ptr(&btree->bt_bmap,
+ nilfs_bmap_commit_alloc_ptr(btree,
&path[level - 1].bp_newreq, dat);
path[level].bp_op(btree, path, level, &key, &ptr);
}
- if (!nilfs_bmap_dirty(&btree->bt_bmap))
- nilfs_bmap_set_dirty(&btree->bt_bmap);
+ if (!nilfs_bmap_dirty(btree))
+ nilfs_bmap_set_dirty(btree);
}
-static int nilfs_btree_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
+static int nilfs_btree_insert(struct nilfs_bmap *btree, __u64 key, __u64 ptr)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
struct nilfs_bmap_stats stats;
int level, ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
ret = nilfs_btree_do_lookup(btree, path, key, NULL,
- NILFS_BTREE_LEVEL_NODE_MIN);
+ NILFS_BTREE_LEVEL_NODE_MIN, 0);
if (ret != -ENOENT) {
if (ret == 0)
ret = -EEXIST;
@@ -1090,23 +1174,25 @@ static int nilfs_btree_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
if (ret < 0)
goto out;
nilfs_btree_commit_insert(btree, path, level, key, ptr);
- nilfs_bmap_add_blocks(bmap, stats.bs_nblocks);
+ nilfs_bmap_add_blocks(btree, stats.bs_nblocks);
out:
nilfs_btree_free_path(path);
return ret;
}
-static void nilfs_btree_do_delete(struct nilfs_btree *btree,
+static void nilfs_btree_do_delete(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node;
+ int ncblk;
if (level < nilfs_btree_height(btree) - 1) {
node = nilfs_btree_get_nonroot_node(path, level);
- nilfs_btree_node_delete(btree, node, keyp, ptrp,
- path[level].bp_index);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
+ nilfs_btree_node_delete(node, path[level].bp_index,
+ keyp, ptrp, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
if (path[level].bp_index == 0)
@@ -1114,17 +1200,18 @@ static void nilfs_btree_do_delete(struct nilfs_btree *btree,
nilfs_btree_node_get_key(node, 0));
} else {
node = nilfs_btree_get_root(btree);
- nilfs_btree_node_delete(btree, node, keyp, ptrp,
- path[level].bp_index);
+ nilfs_btree_node_delete(node, path[level].bp_index,
+ keyp, ptrp,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
}
}
-static void nilfs_btree_borrow_left(struct nilfs_btree *btree,
+static void nilfs_btree_borrow_left(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *left;
- int nchildren, lnchildren, n;
+ int nchildren, lnchildren, n, ncblk;
nilfs_btree_do_delete(btree, path, level, keyp, ptrp);
@@ -1132,10 +1219,11 @@ static void nilfs_btree_borrow_left(struct nilfs_btree *btree,
left = nilfs_btree_get_sib_node(path, level);
nchildren = nilfs_btree_node_get_nchildren(node);
lnchildren = nilfs_btree_node_get_nchildren(left);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
n = (nchildren + lnchildren) / 2 - nchildren;
- nilfs_btree_node_move_right(btree, left, node, n);
+ nilfs_btree_node_move_right(left, node, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -1150,12 +1238,12 @@ static void nilfs_btree_borrow_left(struct nilfs_btree *btree,
path[level].bp_index += n;
}
-static void nilfs_btree_borrow_right(struct nilfs_btree *btree,
+static void nilfs_btree_borrow_right(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *right;
- int nchildren, rnchildren, n;
+ int nchildren, rnchildren, n, ncblk;
nilfs_btree_do_delete(btree, path, level, keyp, ptrp);
@@ -1163,10 +1251,11 @@ static void nilfs_btree_borrow_right(struct nilfs_btree *btree,
right = nilfs_btree_get_sib_node(path, level);
nchildren = nilfs_btree_node_get_nchildren(node);
rnchildren = nilfs_btree_node_get_nchildren(right);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
n = (nchildren + rnchildren) / 2 - nchildren;
- nilfs_btree_node_move_left(btree, node, right, n);
+ nilfs_btree_node_move_left(node, right, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -1182,21 +1271,22 @@ static void nilfs_btree_borrow_right(struct nilfs_btree *btree,
path[level].bp_sib_bh = NULL;
}
-static void nilfs_btree_concat_left(struct nilfs_btree *btree,
+static void nilfs_btree_concat_left(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *left;
- int n;
+ int n, ncblk;
nilfs_btree_do_delete(btree, path, level, keyp, ptrp);
node = nilfs_btree_get_nonroot_node(path, level);
left = nilfs_btree_get_sib_node(path, level);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
n = nilfs_btree_node_get_nchildren(node);
- nilfs_btree_node_move_left(btree, left, node, n);
+ nilfs_btree_node_move_left(left, node, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_sib_bh))
nilfs_btnode_mark_dirty(path[level].bp_sib_bh);
@@ -1207,21 +1297,22 @@ static void nilfs_btree_concat_left(struct nilfs_btree *btree,
path[level].bp_index += nilfs_btree_node_get_nchildren(left);
}
-static void nilfs_btree_concat_right(struct nilfs_btree *btree,
+static void nilfs_btree_concat_right(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *node, *right;
- int n;
+ int n, ncblk;
nilfs_btree_do_delete(btree, path, level, keyp, ptrp);
node = nilfs_btree_get_nonroot_node(path, level);
right = nilfs_btree_get_sib_node(path, level);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
n = nilfs_btree_node_get_nchildren(right);
- nilfs_btree_node_move_left(btree, node, right, n);
+ nilfs_btree_node_move_left(node, right, n, ncblk, ncblk);
if (!buffer_dirty(path[level].bp_bh))
nilfs_btnode_mark_dirty(path[level].bp_bh);
@@ -1231,29 +1322,32 @@ static void nilfs_btree_concat_right(struct nilfs_btree *btree,
path[level + 1].bp_index++;
}
-static void nilfs_btree_shrink(struct nilfs_btree *btree,
+static void nilfs_btree_shrink(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, __u64 *keyp, __u64 *ptrp)
{
struct nilfs_btree_node *root, *child;
- int n;
+ int n, ncblk;
nilfs_btree_do_delete(btree, path, level, keyp, ptrp);
root = nilfs_btree_get_root(btree);
child = nilfs_btree_get_nonroot_node(path, level);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
- nilfs_btree_node_delete(btree, root, NULL, NULL, 0);
+ nilfs_btree_node_delete(root, 0, NULL, NULL,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
nilfs_btree_node_set_level(root, level);
n = nilfs_btree_node_get_nchildren(child);
- nilfs_btree_node_move_left(btree, root, child, n);
+ nilfs_btree_node_move_left(root, child, n,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX, ncblk);
nilfs_btnode_delete(path[level].bp_bh);
path[level].bp_bh = NULL;
}
-static int nilfs_btree_prepare_delete(struct nilfs_btree *btree,
+static int nilfs_btree_prepare_delete(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int *levelp,
struct nilfs_bmap_stats *stats,
@@ -1262,42 +1356,43 @@ static int nilfs_btree_prepare_delete(struct nilfs_btree *btree,
struct buffer_head *bh;
struct nilfs_btree_node *node, *parent, *sib;
__u64 sibptr;
- int pindex, level, ret;
+ int pindex, level, ncmin, ncmax, ncblk, ret;
ret = 0;
stats->bs_nblocks = 0;
+ ncmin = NILFS_BTREE_NODE_NCHILDREN_MIN(nilfs_btree_node_size(btree));
+ ncblk = nilfs_btree_nchildren_per_block(btree);
+
for (level = NILFS_BTREE_LEVEL_NODE_MIN;
level < nilfs_btree_height(btree) - 1;
level++) {
node = nilfs_btree_get_nonroot_node(path, level);
path[level].bp_oldreq.bpr_ptr =
- nilfs_btree_node_get_ptr(btree, node,
- path[level].bp_index);
- ret = nilfs_bmap_prepare_end_ptr(&btree->bt_bmap,
+ nilfs_btree_node_get_ptr(node, path[level].bp_index,
+ ncblk);
+ ret = nilfs_bmap_prepare_end_ptr(btree,
&path[level].bp_oldreq, dat);
if (ret < 0)
goto err_out_child_node;
- if (nilfs_btree_node_get_nchildren(node) >
- nilfs_btree_node_nchildren_min(node, btree)) {
+ if (nilfs_btree_node_get_nchildren(node) > ncmin) {
path[level].bp_op = nilfs_btree_do_delete;
stats->bs_nblocks++;
goto out;
}
- parent = nilfs_btree_get_node(btree, path, level + 1);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
pindex = path[level + 1].bp_index;
if (pindex > 0) {
/* left sibling */
- sibptr = nilfs_btree_node_get_ptr(btree, parent,
- pindex - 1);
+ sibptr = nilfs_btree_node_get_ptr(parent, pindex - 1,
+ ncmax);
ret = nilfs_btree_get_block(btree, sibptr, &bh);
if (ret < 0)
goto err_out_curr_node;
sib = (struct nilfs_btree_node *)bh->b_data;
- if (nilfs_btree_node_get_nchildren(sib) >
- nilfs_btree_node_nchildren_min(sib, btree)) {
+ if (nilfs_btree_node_get_nchildren(sib) > ncmin) {
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_borrow_left;
stats->bs_nblocks++;
@@ -1311,14 +1406,13 @@ static int nilfs_btree_prepare_delete(struct nilfs_btree *btree,
} else if (pindex <
nilfs_btree_node_get_nchildren(parent) - 1) {
/* right sibling */
- sibptr = nilfs_btree_node_get_ptr(btree, parent,
- pindex + 1);
+ sibptr = nilfs_btree_node_get_ptr(parent, pindex + 1,
+ ncmax);
ret = nilfs_btree_get_block(btree, sibptr, &bh);
if (ret < 0)
goto err_out_curr_node;
sib = (struct nilfs_btree_node *)bh->b_data;
- if (nilfs_btree_node_get_nchildren(sib) >
- nilfs_btree_node_nchildren_min(sib, btree)) {
+ if (nilfs_btree_node_get_nchildren(sib) > ncmin) {
path[level].bp_sib_bh = bh;
path[level].bp_op = nilfs_btree_borrow_right;
stats->bs_nblocks++;
@@ -1349,10 +1443,10 @@ static int nilfs_btree_prepare_delete(struct nilfs_btree *btree,
node = nilfs_btree_get_root(btree);
path[level].bp_oldreq.bpr_ptr =
- nilfs_btree_node_get_ptr(btree, node, path[level].bp_index);
+ nilfs_btree_node_get_ptr(node, path[level].bp_index,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
- ret = nilfs_bmap_prepare_end_ptr(&btree->bt_bmap,
- &path[level].bp_oldreq, dat);
+ ret = nilfs_bmap_prepare_end_ptr(btree, &path[level].bp_oldreq, dat);
if (ret < 0)
goto err_out_child_node;
@@ -1367,75 +1461,68 @@ static int nilfs_btree_prepare_delete(struct nilfs_btree *btree,
/* error */
err_out_curr_node:
- nilfs_bmap_abort_end_ptr(&btree->bt_bmap, &path[level].bp_oldreq, dat);
+ nilfs_bmap_abort_end_ptr(btree, &path[level].bp_oldreq, dat);
err_out_child_node:
for (level--; level >= NILFS_BTREE_LEVEL_NODE_MIN; level--) {
brelse(path[level].bp_sib_bh);
- nilfs_bmap_abort_end_ptr(&btree->bt_bmap,
- &path[level].bp_oldreq, dat);
+ nilfs_bmap_abort_end_ptr(btree, &path[level].bp_oldreq, dat);
}
*levelp = level;
stats->bs_nblocks = 0;
return ret;
}
-static void nilfs_btree_commit_delete(struct nilfs_btree *btree,
+static void nilfs_btree_commit_delete(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int maxlevel, struct inode *dat)
{
int level;
for (level = NILFS_BTREE_LEVEL_NODE_MIN; level <= maxlevel; level++) {
- nilfs_bmap_commit_end_ptr(&btree->bt_bmap,
- &path[level].bp_oldreq, dat);
+ nilfs_bmap_commit_end_ptr(btree, &path[level].bp_oldreq, dat);
path[level].bp_op(btree, path, level, NULL, NULL);
}
- if (!nilfs_bmap_dirty(&btree->bt_bmap))
- nilfs_bmap_set_dirty(&btree->bt_bmap);
+ if (!nilfs_bmap_dirty(btree))
+ nilfs_bmap_set_dirty(btree);
}
-static int nilfs_btree_delete(struct nilfs_bmap *bmap, __u64 key)
+static int nilfs_btree_delete(struct nilfs_bmap *btree, __u64 key)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
struct nilfs_bmap_stats stats;
struct inode *dat;
int level, ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
ret = nilfs_btree_do_lookup(btree, path, key, NULL,
- NILFS_BTREE_LEVEL_NODE_MIN);
+ NILFS_BTREE_LEVEL_NODE_MIN, 0);
if (ret < 0)
goto out;
- dat = NILFS_BMAP_USE_VBN(&btree->bt_bmap) ?
- nilfs_bmap_get_dat(&btree->bt_bmap) : NULL;
+ dat = NILFS_BMAP_USE_VBN(btree) ? nilfs_bmap_get_dat(btree) : NULL;
ret = nilfs_btree_prepare_delete(btree, path, &level, &stats, dat);
if (ret < 0)
goto out;
nilfs_btree_commit_delete(btree, path, level, dat);
- nilfs_bmap_sub_blocks(bmap, stats.bs_nblocks);
+ nilfs_bmap_sub_blocks(btree, stats.bs_nblocks);
out:
nilfs_btree_free_path(path);
return ret;
}
-static int nilfs_btree_last_key(const struct nilfs_bmap *bmap, __u64 *keyp)
+static int nilfs_btree_last_key(const struct nilfs_bmap *btree, __u64 *keyp)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
int ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
@@ -1447,16 +1534,14 @@ static int nilfs_btree_last_key(const struct nilfs_bmap *bmap, __u64 *keyp)
return ret;
}
-static int nilfs_btree_check_delete(struct nilfs_bmap *bmap, __u64 key)
+static int nilfs_btree_check_delete(struct nilfs_bmap *btree, __u64 key)
{
struct buffer_head *bh;
- struct nilfs_btree *btree;
struct nilfs_btree_node *root, *node;
__u64 maxkey, nextmaxkey;
__u64 ptr;
int nchildren, ret;
- btree = (struct nilfs_btree *)bmap;
root = nilfs_btree_get_root(btree);
switch (nilfs_btree_height(btree)) {
case 2:
@@ -1467,7 +1552,8 @@ static int nilfs_btree_check_delete(struct nilfs_bmap *bmap, __u64 key)
nchildren = nilfs_btree_node_get_nchildren(root);
if (nchildren > 1)
return 0;
- ptr = nilfs_btree_node_get_ptr(btree, root, nchildren - 1);
+ ptr = nilfs_btree_node_get_ptr(root, nchildren - 1,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
ret = nilfs_btree_get_block(btree, ptr, &bh);
if (ret < 0)
return ret;
@@ -1487,32 +1573,33 @@ static int nilfs_btree_check_delete(struct nilfs_bmap *bmap, __u64 key)
return (maxkey == key) && (nextmaxkey < NILFS_BMAP_LARGE_LOW);
}
-static int nilfs_btree_gather_data(struct nilfs_bmap *bmap,
+static int nilfs_btree_gather_data(struct nilfs_bmap *btree,
__u64 *keys, __u64 *ptrs, int nitems)
{
struct buffer_head *bh;
- struct nilfs_btree *btree;
struct nilfs_btree_node *node, *root;
__le64 *dkeys;
__le64 *dptrs;
__u64 ptr;
- int nchildren, i, ret;
+ int nchildren, ncmax, i, ret;
- btree = (struct nilfs_btree *)bmap;
root = nilfs_btree_get_root(btree);
switch (nilfs_btree_height(btree)) {
case 2:
bh = NULL;
node = root;
+ ncmax = NILFS_BTREE_ROOT_NCHILDREN_MAX;
break;
case 3:
nchildren = nilfs_btree_node_get_nchildren(root);
WARN_ON(nchildren > 1);
- ptr = nilfs_btree_node_get_ptr(btree, root, nchildren - 1);
+ ptr = nilfs_btree_node_get_ptr(root, nchildren - 1,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
ret = nilfs_btree_get_block(btree, ptr, &bh);
if (ret < 0)
return ret;
node = (struct nilfs_btree_node *)bh->b_data;
+ ncmax = nilfs_btree_nchildren_per_block(btree);
break;
default:
node = NULL;
@@ -1523,10 +1610,10 @@ static int nilfs_btree_gather_data(struct nilfs_bmap *bmap,
if (nchildren < nitems)
nitems = nchildren;
dkeys = nilfs_btree_node_dkeys(node);
- dptrs = nilfs_btree_node_dptrs(node, btree);
+ dptrs = nilfs_btree_node_dptrs(node, ncmax);
for (i = 0; i < nitems; i++) {
- keys[i] = nilfs_bmap_dkey_to_key(dkeys[i]);
- ptrs[i] = nilfs_bmap_dptr_to_ptr(dptrs[i]);
+ keys[i] = le64_to_cpu(dkeys[i]);
+ ptrs[i] = le64_to_cpu(dptrs[i]);
}
if (bh != NULL)
@@ -1536,14 +1623,13 @@ static int nilfs_btree_gather_data(struct nilfs_bmap *bmap,
}
static int
-nilfs_btree_prepare_convert_and_insert(struct nilfs_bmap *bmap, __u64 key,
+nilfs_btree_prepare_convert_and_insert(struct nilfs_bmap *btree, __u64 key,
union nilfs_bmap_ptr_req *dreq,
union nilfs_bmap_ptr_req *nreq,
struct buffer_head **bhp,
struct nilfs_bmap_stats *stats)
{
struct buffer_head *bh;
- struct nilfs_btree *btree = (struct nilfs_btree *)bmap;
struct inode *dat = NULL;
int ret;
@@ -1551,12 +1637,12 @@ nilfs_btree_prepare_convert_and_insert(struct nilfs_bmap *bmap, __u64 key,
/* for data */
/* cannot find near ptr */
- if (NILFS_BMAP_USE_VBN(bmap)) {
+ if (NILFS_BMAP_USE_VBN(btree)) {
dreq->bpr_ptr = nilfs_btree_find_target_v(btree, NULL, key);
- dat = nilfs_bmap_get_dat(bmap);
+ dat = nilfs_bmap_get_dat(btree);
}
- ret = nilfs_bmap_prepare_alloc_ptr(bmap, dreq, dat);
+ ret = nilfs_bmap_prepare_alloc_ptr(btree, dreq, dat);
if (ret < 0)
return ret;
@@ -1564,7 +1650,7 @@ nilfs_btree_prepare_convert_and_insert(struct nilfs_bmap *bmap, __u64 key,
stats->bs_nblocks++;
if (nreq != NULL) {
nreq->bpr_ptr = dreq->bpr_ptr + 1;
- ret = nilfs_bmap_prepare_alloc_ptr(bmap, nreq, dat);
+ ret = nilfs_bmap_prepare_alloc_ptr(btree, nreq, dat);
if (ret < 0)
goto err_out_dreq;
@@ -1581,16 +1667,16 @@ nilfs_btree_prepare_convert_and_insert(struct nilfs_bmap *bmap, __u64 key,
/* error */
err_out_nreq:
- nilfs_bmap_abort_alloc_ptr(bmap, nreq, dat);
+ nilfs_bmap_abort_alloc_ptr(btree, nreq, dat);
err_out_dreq:
- nilfs_bmap_abort_alloc_ptr(bmap, dreq, dat);
+ nilfs_bmap_abort_alloc_ptr(btree, dreq, dat);
stats->bs_nblocks = 0;
return ret;
}
static void
-nilfs_btree_commit_convert_and_insert(struct nilfs_bmap *bmap,
+nilfs_btree_commit_convert_and_insert(struct nilfs_bmap *btree,
__u64 key, __u64 ptr,
const __u64 *keys, const __u64 *ptrs,
int n,
@@ -1598,57 +1684,59 @@ nilfs_btree_commit_convert_and_insert(struct nilfs_bmap *bmap,
union nilfs_bmap_ptr_req *nreq,
struct buffer_head *bh)
{
- struct nilfs_btree *btree = (struct nilfs_btree *)bmap;
struct nilfs_btree_node *node;
struct inode *dat;
__u64 tmpptr;
+ int ncblk;
/* free resources */
- if (bmap->b_ops->bop_clear != NULL)
- bmap->b_ops->bop_clear(bmap);
+ if (btree->b_ops->bop_clear != NULL)
+ btree->b_ops->bop_clear(btree);
/* ptr must be a pointer to a buffer head. */
set_buffer_nilfs_volatile((struct buffer_head *)((unsigned long)ptr));
/* convert and insert */
- dat = NILFS_BMAP_USE_VBN(bmap) ? nilfs_bmap_get_dat(bmap) : NULL;
- nilfs_btree_init(bmap);
+ dat = NILFS_BMAP_USE_VBN(btree) ? nilfs_bmap_get_dat(btree) : NULL;
+ nilfs_btree_init(btree);
if (nreq != NULL) {
- nilfs_bmap_commit_alloc_ptr(bmap, dreq, dat);
- nilfs_bmap_commit_alloc_ptr(bmap, nreq, dat);
+ nilfs_bmap_commit_alloc_ptr(btree, dreq, dat);
+ nilfs_bmap_commit_alloc_ptr(btree, nreq, dat);
/* create child node at level 1 */
node = (struct nilfs_btree_node *)bh->b_data;
- nilfs_btree_node_init(btree, node, 0, 1, n, keys, ptrs);
- nilfs_btree_node_insert(btree, node,
- key, dreq->bpr_ptr, n);
+ ncblk = nilfs_btree_nchildren_per_block(btree);
+ nilfs_btree_node_init(node, 0, 1, n, ncblk, keys, ptrs);
+ nilfs_btree_node_insert(node, n, key, dreq->bpr_ptr, ncblk);
if (!buffer_dirty(bh))
nilfs_btnode_mark_dirty(bh);
- if (!nilfs_bmap_dirty(bmap))
- nilfs_bmap_set_dirty(bmap);
+ if (!nilfs_bmap_dirty(btree))
+ nilfs_bmap_set_dirty(btree);
brelse(bh);
/* create root node at level 2 */
node = nilfs_btree_get_root(btree);
tmpptr = nreq->bpr_ptr;
- nilfs_btree_node_init(btree, node, NILFS_BTREE_NODE_ROOT,
- 2, 1, &keys[0], &tmpptr);
+ nilfs_btree_node_init(node, NILFS_BTREE_NODE_ROOT, 2, 1,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX,
+ &keys[0], &tmpptr);
} else {
- nilfs_bmap_commit_alloc_ptr(bmap, dreq, dat);
+ nilfs_bmap_commit_alloc_ptr(btree, dreq, dat);
/* create root node at level 1 */
node = nilfs_btree_get_root(btree);
- nilfs_btree_node_init(btree, node, NILFS_BTREE_NODE_ROOT,
- 1, n, keys, ptrs);
- nilfs_btree_node_insert(btree, node,
- key, dreq->bpr_ptr, n);
- if (!nilfs_bmap_dirty(bmap))
- nilfs_bmap_set_dirty(bmap);
+ nilfs_btree_node_init(node, NILFS_BTREE_NODE_ROOT, 1, n,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX,
+ keys, ptrs);
+ nilfs_btree_node_insert(node, n, key, dreq->bpr_ptr,
+ NILFS_BTREE_ROOT_NCHILDREN_MAX);
+ if (!nilfs_bmap_dirty(btree))
+ nilfs_bmap_set_dirty(btree);
}
- if (NILFS_BMAP_USE_VBN(bmap))
- nilfs_btree_set_target_v(btree, key, dreq->bpr_ptr);
+ if (NILFS_BMAP_USE_VBN(btree))
+ nilfs_bmap_set_target_v(btree, key, dreq->bpr_ptr);
}
/**
@@ -1660,7 +1748,7 @@ nilfs_btree_commit_convert_and_insert(struct nilfs_bmap *bmap,
* @ptrs:
* @n:
*/
-int nilfs_btree_convert_and_insert(struct nilfs_bmap *bmap,
+int nilfs_btree_convert_and_insert(struct nilfs_bmap *btree,
__u64 key, __u64 ptr,
const __u64 *keys, const __u64 *ptrs, int n)
{
@@ -1673,7 +1761,7 @@ int nilfs_btree_convert_and_insert(struct nilfs_bmap *bmap,
di = &dreq;
ni = NULL;
} else if ((n + 1) <= NILFS_BTREE_NODE_NCHILDREN_MAX(
- 1 << bmap->b_inode->i_blkbits)) {
+ 1 << btree->b_inode->i_blkbits)) {
di = &dreq;
ni = &nreq;
} else {
@@ -1682,17 +1770,17 @@ int nilfs_btree_convert_and_insert(struct nilfs_bmap *bmap,
BUG();
}
- ret = nilfs_btree_prepare_convert_and_insert(bmap, key, di, ni, &bh,
+ ret = nilfs_btree_prepare_convert_and_insert(btree, key, di, ni, &bh,
&stats);
if (ret < 0)
return ret;
- nilfs_btree_commit_convert_and_insert(bmap, key, ptr, keys, ptrs, n,
+ nilfs_btree_commit_convert_and_insert(btree, key, ptr, keys, ptrs, n,
di, ni, bh);
- nilfs_bmap_add_blocks(bmap, stats.bs_nblocks);
+ nilfs_bmap_add_blocks(btree, stats.bs_nblocks);
return 0;
}
-static int nilfs_btree_propagate_p(struct nilfs_btree *btree,
+static int nilfs_btree_propagate_p(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level,
struct buffer_head *bh)
@@ -1704,17 +1792,17 @@ static int nilfs_btree_propagate_p(struct nilfs_btree *btree,
return 0;
}
-static int nilfs_btree_prepare_update_v(struct nilfs_btree *btree,
+static int nilfs_btree_prepare_update_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, struct inode *dat)
{
struct nilfs_btree_node *parent;
- int ret;
+ int ncmax, ret;
- parent = nilfs_btree_get_node(btree, path, level + 1);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
path[level].bp_oldreq.bpr_ptr =
- nilfs_btree_node_get_ptr(btree, parent,
- path[level + 1].bp_index);
+ nilfs_btree_node_get_ptr(parent, path[level + 1].bp_index,
+ ncmax);
path[level].bp_newreq.bpr_ptr = path[level].bp_oldreq.bpr_ptr + 1;
ret = nilfs_dat_prepare_update(dat, &path[level].bp_oldreq.bpr_req,
&path[level].bp_newreq.bpr_req);
@@ -1726,7 +1814,7 @@ static int nilfs_btree_prepare_update_v(struct nilfs_btree *btree,
path[level].bp_ctxt.newkey = path[level].bp_newreq.bpr_ptr;
path[level].bp_ctxt.bh = path[level].bp_bh;
ret = nilfs_btnode_prepare_change_key(
- &NILFS_BMAP_I(&btree->bt_bmap)->i_btnode_cache,
+ &NILFS_BMAP_I(btree)->i_btnode_cache,
&path[level].bp_ctxt);
if (ret < 0) {
nilfs_dat_abort_update(dat,
@@ -1739,30 +1827,31 @@ static int nilfs_btree_prepare_update_v(struct nilfs_btree *btree,
return 0;
}
-static void nilfs_btree_commit_update_v(struct nilfs_btree *btree,
+static void nilfs_btree_commit_update_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, struct inode *dat)
{
struct nilfs_btree_node *parent;
+ int ncmax;
nilfs_dat_commit_update(dat, &path[level].bp_oldreq.bpr_req,
&path[level].bp_newreq.bpr_req,
- btree->bt_bmap.b_ptr_type == NILFS_BMAP_PTR_VS);
+ btree->b_ptr_type == NILFS_BMAP_PTR_VS);
if (buffer_nilfs_node(path[level].bp_bh)) {
nilfs_btnode_commit_change_key(
- &NILFS_BMAP_I(&btree->bt_bmap)->i_btnode_cache,
+ &NILFS_BMAP_I(btree)->i_btnode_cache,
&path[level].bp_ctxt);
path[level].bp_bh = path[level].bp_ctxt.bh;
}
set_buffer_nilfs_volatile(path[level].bp_bh);
- parent = nilfs_btree_get_node(btree, path, level + 1);
- nilfs_btree_node_set_ptr(btree, parent, path[level + 1].bp_index,
- path[level].bp_newreq.bpr_ptr);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
+ nilfs_btree_node_set_ptr(parent, path[level + 1].bp_index,
+ path[level].bp_newreq.bpr_ptr, ncmax);
}
-static void nilfs_btree_abort_update_v(struct nilfs_btree *btree,
+static void nilfs_btree_abort_update_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, struct inode *dat)
{
@@ -1770,11 +1859,11 @@ static void nilfs_btree_abort_update_v(struct nilfs_btree *btree,
&path[level].bp_newreq.bpr_req);
if (buffer_nilfs_node(path[level].bp_bh))
nilfs_btnode_abort_change_key(
- &NILFS_BMAP_I(&btree->bt_bmap)->i_btnode_cache,
+ &NILFS_BMAP_I(btree)->i_btnode_cache,
&path[level].bp_ctxt);
}
-static int nilfs_btree_prepare_propagate_v(struct nilfs_btree *btree,
+static int nilfs_btree_prepare_propagate_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int minlevel, int *maxlevelp,
struct inode *dat)
@@ -1809,7 +1898,7 @@ static int nilfs_btree_prepare_propagate_v(struct nilfs_btree *btree,
return ret;
}
-static void nilfs_btree_commit_propagate_v(struct nilfs_btree *btree,
+static void nilfs_btree_commit_propagate_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int minlevel, int maxlevel,
struct buffer_head *bh,
@@ -1824,14 +1913,15 @@ static void nilfs_btree_commit_propagate_v(struct nilfs_btree *btree,
nilfs_btree_commit_update_v(btree, path, level, dat);
}
-static int nilfs_btree_propagate_v(struct nilfs_btree *btree,
+static int nilfs_btree_propagate_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level, struct buffer_head *bh)
{
int maxlevel = 0, ret;
struct nilfs_btree_node *parent;
- struct inode *dat = nilfs_bmap_get_dat(&btree->bt_bmap);
+ struct inode *dat = nilfs_bmap_get_dat(btree);
__u64 ptr;
+ int ncmax;
get_bh(bh);
path[level].bp_bh = bh;
@@ -1841,9 +1931,10 @@ static int nilfs_btree_propagate_v(struct nilfs_btree *btree,
goto out;
if (buffer_nilfs_volatile(path[level].bp_bh)) {
- parent = nilfs_btree_get_node(btree, path, level + 1);
- ptr = nilfs_btree_node_get_ptr(btree, parent,
- path[level + 1].bp_index);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
+ ptr = nilfs_btree_node_get_ptr(parent,
+ path[level + 1].bp_index,
+ ncmax);
ret = nilfs_dat_mark_dirty(dat, ptr);
if (ret < 0)
goto out;
@@ -1857,10 +1948,9 @@ static int nilfs_btree_propagate_v(struct nilfs_btree *btree,
return ret;
}
-static int nilfs_btree_propagate(const struct nilfs_bmap *bmap,
+static int nilfs_btree_propagate(struct nilfs_bmap *btree,
struct buffer_head *bh)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
struct nilfs_btree_node *node;
__u64 key;
@@ -1868,7 +1958,6 @@ static int nilfs_btree_propagate(const struct nilfs_bmap *bmap,
WARN_ON(!buffer_dirty(bh));
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
@@ -1878,11 +1967,11 @@ static int nilfs_btree_propagate(const struct nilfs_bmap *bmap,
key = nilfs_btree_node_get_key(node, 0);
level = nilfs_btree_node_get_level(node);
} else {
- key = nilfs_bmap_data_get_key(bmap, bh);
+ key = nilfs_bmap_data_get_key(btree, bh);
level = NILFS_BTREE_LEVEL_DATA;
}
- ret = nilfs_btree_do_lookup(btree, path, key, NULL, level + 1);
+ ret = nilfs_btree_do_lookup(btree, path, key, NULL, level + 1, 0);
if (ret < 0) {
if (unlikely(ret == -ENOENT))
printk(KERN_CRIT "%s: key = %llu, level == %d\n",
@@ -1890,7 +1979,7 @@ static int nilfs_btree_propagate(const struct nilfs_bmap *bmap,
goto out;
}
- ret = NILFS_BMAP_USE_VBN(bmap) ?
+ ret = NILFS_BMAP_USE_VBN(btree) ?
nilfs_btree_propagate_v(btree, path, level, bh) :
nilfs_btree_propagate_p(btree, path, level, bh);
@@ -1900,13 +1989,13 @@ static int nilfs_btree_propagate(const struct nilfs_bmap *bmap,
return ret;
}
-static int nilfs_btree_propagate_gc(const struct nilfs_bmap *bmap,
+static int nilfs_btree_propagate_gc(struct nilfs_bmap *btree,
struct buffer_head *bh)
{
- return nilfs_dat_mark_dirty(nilfs_bmap_get_dat(bmap), bh->b_blocknr);
+ return nilfs_dat_mark_dirty(nilfs_bmap_get_dat(btree), bh->b_blocknr);
}
-static void nilfs_btree_add_dirty_buffer(struct nilfs_btree *btree,
+static void nilfs_btree_add_dirty_buffer(struct nilfs_bmap *btree,
struct list_head *lists,
struct buffer_head *bh)
{
@@ -1920,6 +2009,18 @@ static void nilfs_btree_add_dirty_buffer(struct nilfs_btree *btree,
node = (struct nilfs_btree_node *)bh->b_data;
key = nilfs_btree_node_get_key(node, 0);
level = nilfs_btree_node_get_level(node);
+ if (level < NILFS_BTREE_LEVEL_NODE_MIN ||
+ level >= NILFS_BTREE_LEVEL_MAX) {
+ dump_stack();
+ printk(KERN_WARNING
+ "%s: invalid btree level: %d (key=%llu, ino=%lu, "
+ "blocknr=%llu)\n",
+ __func__, level, (unsigned long long)key,
+ NILFS_BMAP_I(btree)->vfs_inode.i_ino,
+ (unsigned long long)bh->b_blocknr);
+ return;
+ }
+
list_for_each(head, &lists[level]) {
cbh = list_entry(head, struct buffer_head, b_assoc_buffers);
cnode = (struct nilfs_btree_node *)cbh->b_data;
@@ -1930,11 +2031,10 @@ static void nilfs_btree_add_dirty_buffer(struct nilfs_btree *btree,
list_add_tail(&bh->b_assoc_buffers, head);
}
-static void nilfs_btree_lookup_dirty_buffers(struct nilfs_bmap *bmap,
+static void nilfs_btree_lookup_dirty_buffers(struct nilfs_bmap *btree,
struct list_head *listp)
{
- struct nilfs_btree *btree = (struct nilfs_btree *)bmap;
- struct address_space *btcache = &NILFS_BMAP_I(bmap)->i_btnode_cache;
+ struct address_space *btcache = &NILFS_BMAP_I(btree)->i_btnode_cache;
struct list_head lists[NILFS_BTREE_LEVEL_MAX];
struct pagevec pvec;
struct buffer_head *bh, *head;
@@ -1968,7 +2068,7 @@ static void nilfs_btree_lookup_dirty_buffers(struct nilfs_bmap *bmap,
list_splice_tail(&lists[level], listp);
}
-static int nilfs_btree_assign_p(struct nilfs_btree *btree,
+static int nilfs_btree_assign_p(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level,
struct buffer_head **bh,
@@ -1978,38 +2078,38 @@ static int nilfs_btree_assign_p(struct nilfs_btree *btree,
struct nilfs_btree_node *parent;
__u64 key;
__u64 ptr;
- int ret;
+ int ncmax, ret;
- parent = nilfs_btree_get_node(btree, path, level + 1);
- ptr = nilfs_btree_node_get_ptr(btree, parent,
- path[level + 1].bp_index);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
+ ptr = nilfs_btree_node_get_ptr(parent, path[level + 1].bp_index,
+ ncmax);
if (buffer_nilfs_node(*bh)) {
path[level].bp_ctxt.oldkey = ptr;
path[level].bp_ctxt.newkey = blocknr;
path[level].bp_ctxt.bh = *bh;
ret = nilfs_btnode_prepare_change_key(
- &NILFS_BMAP_I(&btree->bt_bmap)->i_btnode_cache,
+ &NILFS_BMAP_I(btree)->i_btnode_cache,
&path[level].bp_ctxt);
if (ret < 0)
return ret;
nilfs_btnode_commit_change_key(
- &NILFS_BMAP_I(&btree->bt_bmap)->i_btnode_cache,
+ &NILFS_BMAP_I(btree)->i_btnode_cache,
&path[level].bp_ctxt);
*bh = path[level].bp_ctxt.bh;
}
- nilfs_btree_node_set_ptr(btree, parent,
- path[level + 1].bp_index, blocknr);
+ nilfs_btree_node_set_ptr(parent, path[level + 1].bp_index, blocknr,
+ ncmax);
key = nilfs_btree_node_get_key(parent, path[level + 1].bp_index);
/* on-disk format */
- binfo->bi_dat.bi_blkoff = nilfs_bmap_key_to_dkey(key);
+ binfo->bi_dat.bi_blkoff = cpu_to_le64(key);
binfo->bi_dat.bi_level = level;
return 0;
}
-static int nilfs_btree_assign_v(struct nilfs_btree *btree,
+static int nilfs_btree_assign_v(struct nilfs_bmap *btree,
struct nilfs_btree_path *path,
int level,
struct buffer_head **bh,
@@ -2017,15 +2117,15 @@ static int nilfs_btree_assign_v(struct nilfs_btree *btree,
union nilfs_binfo *binfo)
{
struct nilfs_btree_node *parent;
- struct inode *dat = nilfs_bmap_get_dat(&btree->bt_bmap);
+ struct inode *dat = nilfs_bmap_get_dat(btree);
__u64 key;
__u64 ptr;
union nilfs_bmap_ptr_req req;
- int ret;
+ int ncmax, ret;
- parent = nilfs_btree_get_node(btree, path, level + 1);
- ptr = nilfs_btree_node_get_ptr(btree, parent,
- path[level + 1].bp_index);
+ parent = nilfs_btree_get_node(btree, path, level + 1, &ncmax);
+ ptr = nilfs_btree_node_get_ptr(parent, path[level + 1].bp_index,
+ ncmax);
req.bpr_ptr = ptr;
ret = nilfs_dat_prepare_start(dat, &req.bpr_req);
if (ret < 0)
@@ -2034,24 +2134,22 @@ static int nilfs_btree_assign_v(struct nilfs_btree *btree,
key = nilfs_btree_node_get_key(parent, path[level + 1].bp_index);
/* on-disk format */
- binfo->bi_v.bi_vblocknr = nilfs_bmap_ptr_to_dptr(ptr);
- binfo->bi_v.bi_blkoff = nilfs_bmap_key_to_dkey(key);
+ binfo->bi_v.bi_vblocknr = cpu_to_le64(ptr);
+ binfo->bi_v.bi_blkoff = cpu_to_le64(key);
return 0;
}
-static int nilfs_btree_assign(struct nilfs_bmap *bmap,
+static int nilfs_btree_assign(struct nilfs_bmap *btree,
struct buffer_head **bh,
sector_t blocknr,
union nilfs_binfo *binfo)
{
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
struct nilfs_btree_node *node;
__u64 key;
int level, ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
@@ -2061,17 +2159,17 @@ static int nilfs_btree_assign(struct nilfs_bmap *bmap,
key = nilfs_btree_node_get_key(node, 0);
level = nilfs_btree_node_get_level(node);
} else {
- key = nilfs_bmap_data_get_key(bmap, *bh);
+ key = nilfs_bmap_data_get_key(btree, *bh);
level = NILFS_BTREE_LEVEL_DATA;
}
- ret = nilfs_btree_do_lookup(btree, path, key, NULL, level + 1);
+ ret = nilfs_btree_do_lookup(btree, path, key, NULL, level + 1, 0);
if (ret < 0) {
WARN_ON(ret == -ENOENT);
goto out;
}
- ret = NILFS_BMAP_USE_VBN(bmap) ?
+ ret = NILFS_BMAP_USE_VBN(btree) ?
nilfs_btree_assign_v(btree, path, level, bh, blocknr, binfo) :
nilfs_btree_assign_p(btree, path, level, bh, blocknr, binfo);
@@ -2081,7 +2179,7 @@ static int nilfs_btree_assign(struct nilfs_bmap *bmap,
return ret;
}
-static int nilfs_btree_assign_gc(struct nilfs_bmap *bmap,
+static int nilfs_btree_assign_gc(struct nilfs_bmap *btree,
struct buffer_head **bh,
sector_t blocknr,
union nilfs_binfo *binfo)
@@ -2090,7 +2188,7 @@ static int nilfs_btree_assign_gc(struct nilfs_bmap *bmap,
__u64 key;
int ret;
- ret = nilfs_dat_move(nilfs_bmap_get_dat(bmap), (*bh)->b_blocknr,
+ ret = nilfs_dat_move(nilfs_bmap_get_dat(btree), (*bh)->b_blocknr,
blocknr);
if (ret < 0)
return ret;
@@ -2099,29 +2197,27 @@ static int nilfs_btree_assign_gc(struct nilfs_bmap *bmap,
node = (struct nilfs_btree_node *)(*bh)->b_data;
key = nilfs_btree_node_get_key(node, 0);
} else
- key = nilfs_bmap_data_get_key(bmap, *bh);
+ key = nilfs_bmap_data_get_key(btree, *bh);
/* on-disk format */
binfo->bi_v.bi_vblocknr = cpu_to_le64((*bh)->b_blocknr);
- binfo->bi_v.bi_blkoff = nilfs_bmap_key_to_dkey(key);
+ binfo->bi_v.bi_blkoff = cpu_to_le64(key);
return 0;
}
-static int nilfs_btree_mark(struct nilfs_bmap *bmap, __u64 key, int level)
+static int nilfs_btree_mark(struct nilfs_bmap *btree, __u64 key, int level)
{
struct buffer_head *bh;
- struct nilfs_btree *btree;
struct nilfs_btree_path *path;
__u64 ptr;
int ret;
- btree = (struct nilfs_btree *)bmap;
path = nilfs_btree_alloc_path();
if (path == NULL)
return -ENOMEM;
- ret = nilfs_btree_do_lookup(btree, path, key, &ptr, level + 1);
+ ret = nilfs_btree_do_lookup(btree, path, key, &ptr, level + 1, 0);
if (ret < 0) {
WARN_ON(ret == -ENOENT);
goto out;
@@ -2135,8 +2231,8 @@ static int nilfs_btree_mark(struct nilfs_bmap *bmap, __u64 key, int level)
if (!buffer_dirty(bh))
nilfs_btnode_mark_dirty(bh);
brelse(bh);
- if (!nilfs_bmap_dirty(&btree->bt_bmap))
- nilfs_bmap_set_dirty(&btree->bt_bmap);
+ if (!nilfs_bmap_dirty(btree))
+ nilfs_bmap_set_dirty(btree);
out:
nilfs_btree_free_path(path);
@@ -2186,10 +2282,14 @@ static const struct nilfs_bmap_operations nilfs_btree_ops_gc = {
int nilfs_btree_init(struct nilfs_bmap *bmap)
{
bmap->b_ops = &nilfs_btree_ops;
+ bmap->b_nchildren_per_block =
+ NILFS_BTREE_NODE_NCHILDREN_MAX(nilfs_btree_node_size(bmap));
return 0;
}
void nilfs_btree_init_gc(struct nilfs_bmap *bmap)
{
bmap->b_ops = &nilfs_btree_ops_gc;
+ bmap->b_nchildren_per_block =
+ NILFS_BTREE_NODE_NCHILDREN_MAX(nilfs_btree_node_size(bmap));
}
diff --git a/fs/nilfs2/btree.h b/fs/nilfs2/btree.h
index 43c8c5b541f..22c02e35b6e 100644
--- a/fs/nilfs2/btree.h
+++ b/fs/nilfs2/btree.h
@@ -31,14 +31,6 @@
#include "bmap.h"
/**
- * struct nilfs_btree - B-tree structure
- * @bt_bmap: bmap base structure
- */
-struct nilfs_btree {
- struct nilfs_bmap bt_bmap;
-};
-
-/**
* struct nilfs_btree_path - A path on which B-tree operations are executed
* @bp_bh: buffer head of node block
* @bp_sib_bh: buffer head of sibling node block
@@ -54,7 +46,7 @@ struct nilfs_btree_path {
union nilfs_bmap_ptr_req bp_oldreq;
union nilfs_bmap_ptr_req bp_newreq;
struct nilfs_btnode_chkey_ctxt bp_ctxt;
- void (*bp_op)(struct nilfs_btree *, struct nilfs_btree_path *,
+ void (*bp_op)(struct nilfs_bmap *, struct nilfs_btree_path *,
int, __u64 *, __u64 *);
};
@@ -80,4 +72,6 @@ int nilfs_btree_convert_and_insert(struct nilfs_bmap *, __u64, __u64,
const __u64 *, const __u64 *, int);
void nilfs_btree_init_gc(struct nilfs_bmap *);
+int nilfs_btree_broken_node_block(struct buffer_head *bh);
+
#endif /* _NILFS_BTREE_H */
diff --git a/fs/nilfs2/dir.c b/fs/nilfs2/dir.c
index 85c89dfc71f..b60277b4446 100644
--- a/fs/nilfs2/dir.c
+++ b/fs/nilfs2/dir.c
@@ -141,7 +141,7 @@ static void nilfs_check_page(struct page *page)
}
for (offs = 0; offs <= limit - NILFS_DIR_REC_LEN(1); offs += rec_len) {
p = (struct nilfs_dir_entry *)(kaddr + offs);
- rec_len = le16_to_cpu(p->rec_len);
+ rec_len = nilfs_rec_len_from_disk(p->rec_len);
if (rec_len < NILFS_DIR_REC_LEN(1))
goto Eshort;
@@ -199,13 +199,10 @@ fail:
static struct page *nilfs_get_page(struct inode *dir, unsigned long n)
{
struct address_space *mapping = dir->i_mapping;
- struct page *page = read_cache_page(mapping, n,
- (filler_t *)mapping->a_ops->readpage, NULL);
+ struct page *page = read_mapping_page(mapping, n, NULL);
+
if (!IS_ERR(page)) {
- wait_on_page_locked(page);
kmap(page);
- if (!PageUptodate(page))
- goto fail;
if (!PageChecked(page))
nilfs_check_page(page);
if (PageError(page))
@@ -238,7 +235,8 @@ nilfs_match(int len, const unsigned char *name, struct nilfs_dir_entry *de)
*/
static struct nilfs_dir_entry *nilfs_next_entry(struct nilfs_dir_entry *p)
{
- return (struct nilfs_dir_entry *)((char *)p + le16_to_cpu(p->rec_len));
+ return (struct nilfs_dir_entry *)((char *)p +
+ nilfs_rec_len_from_disk(p->rec_len));
}
static unsigned char
@@ -329,7 +327,7 @@ static int nilfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
goto success;
}
}
- filp->f_pos += le16_to_cpu(de->rec_len);
+ filp->f_pos += nilfs_rec_len_from_disk(de->rec_len);
}
nilfs_put_page(page);
}
@@ -444,7 +442,7 @@ void nilfs_set_link(struct inode *dir, struct nilfs_dir_entry *de,
struct page *page, struct inode *inode)
{
unsigned from = (char *) de - (char *) page_address(page);
- unsigned to = from + le16_to_cpu(de->rec_len);
+ unsigned to = from + nilfs_rec_len_from_disk(de->rec_len);
struct address_space *mapping = page->mapping;
int err;
@@ -500,7 +498,7 @@ int nilfs_add_link(struct dentry *dentry, struct inode *inode)
/* We hit i_size */
name_len = 0;
rec_len = chunk_size;
- de->rec_len = cpu_to_le16(chunk_size);
+ de->rec_len = nilfs_rec_len_to_disk(chunk_size);
de->inode = 0;
goto got_it;
}
@@ -514,7 +512,7 @@ int nilfs_add_link(struct dentry *dentry, struct inode *inode)
if (nilfs_match(namelen, name, de))
goto out_unlock;
name_len = NILFS_DIR_REC_LEN(de->name_len);
- rec_len = le16_to_cpu(de->rec_len);
+ rec_len = nilfs_rec_len_from_disk(de->rec_len);
if (!de->inode && rec_len >= reclen)
goto got_it;
if (rec_len >= name_len + reclen)
@@ -537,8 +535,8 @@ got_it:
struct nilfs_dir_entry *de1;
de1 = (struct nilfs_dir_entry *)((char *)de + name_len);
- de1->rec_len = cpu_to_le16(rec_len - name_len);
- de->rec_len = cpu_to_le16(name_len);
+ de1->rec_len = nilfs_rec_len_to_disk(rec_len - name_len);
+ de->rec_len = nilfs_rec_len_to_disk(name_len);
de = de1;
}
de->name_len = namelen;
@@ -569,7 +567,8 @@ int nilfs_delete_entry(struct nilfs_dir_entry *dir, struct page *page)
struct inode *inode = mapping->host;
char *kaddr = page_address(page);
unsigned from = ((char *)dir - kaddr) & ~(nilfs_chunk_size(inode) - 1);
- unsigned to = ((char *)dir - kaddr) + le16_to_cpu(dir->rec_len);
+ unsigned to = ((char *)dir - kaddr) +
+ nilfs_rec_len_from_disk(dir->rec_len);
struct nilfs_dir_entry *pde = NULL;
struct nilfs_dir_entry *de = (struct nilfs_dir_entry *)(kaddr + from);
int err;
@@ -590,7 +589,7 @@ int nilfs_delete_entry(struct nilfs_dir_entry *dir, struct page *page)
err = nilfs_prepare_chunk(page, mapping, from, to);
BUG_ON(err);
if (pde)
- pde->rec_len = cpu_to_le16(to - from);
+ pde->rec_len = nilfs_rec_len_to_disk(to - from);
dir->inode = 0;
nilfs_commit_chunk(page, mapping, from, to);
inode->i_ctime = inode->i_mtime = CURRENT_TIME;
@@ -624,14 +623,14 @@ int nilfs_make_empty(struct inode *inode, struct inode *parent)
memset(kaddr, 0, chunk_size);
de = (struct nilfs_dir_entry *)kaddr;
de->name_len = 1;
- de->rec_len = cpu_to_le16(NILFS_DIR_REC_LEN(1));
+ de->rec_len = nilfs_rec_len_to_disk(NILFS_DIR_REC_LEN(1));
memcpy(de->name, ".\0\0", 4);
de->inode = cpu_to_le64(inode->i_ino);
nilfs_set_de_type(de, inode);
de = (struct nilfs_dir_entry *)(kaddr + NILFS_DIR_REC_LEN(1));
de->name_len = 2;
- de->rec_len = cpu_to_le16(chunk_size - NILFS_DIR_REC_LEN(1));
+ de->rec_len = nilfs_rec_len_to_disk(chunk_size - NILFS_DIR_REC_LEN(1));
de->inode = cpu_to_le64(parent->i_ino);
memcpy(de->name, "..\0", 4);
nilfs_set_de_type(de, inode);
diff --git a/fs/nilfs2/direct.c b/fs/nilfs2/direct.c
index 236753df5cd..324d80c5751 100644
--- a/fs/nilfs2/direct.c
+++ b/fs/nilfs2/direct.c
@@ -27,47 +27,43 @@
#include "alloc.h"
#include "dat.h"
-static inline __le64 *nilfs_direct_dptrs(const struct nilfs_direct *direct)
+static inline __le64 *nilfs_direct_dptrs(const struct nilfs_bmap *direct)
{
return (__le64 *)
- ((struct nilfs_direct_node *)direct->d_bmap.b_u.u_data + 1);
+ ((struct nilfs_direct_node *)direct->b_u.u_data + 1);
}
static inline __u64
-nilfs_direct_get_ptr(const struct nilfs_direct *direct, __u64 key)
+nilfs_direct_get_ptr(const struct nilfs_bmap *direct, __u64 key)
{
- return nilfs_bmap_dptr_to_ptr(*(nilfs_direct_dptrs(direct) + key));
+ return le64_to_cpu(*(nilfs_direct_dptrs(direct) + key));
}
-static inline void nilfs_direct_set_ptr(struct nilfs_direct *direct,
+static inline void nilfs_direct_set_ptr(struct nilfs_bmap *direct,
__u64 key, __u64 ptr)
{
- *(nilfs_direct_dptrs(direct) + key) = nilfs_bmap_ptr_to_dptr(ptr);
+ *(nilfs_direct_dptrs(direct) + key) = cpu_to_le64(ptr);
}
-static int nilfs_direct_lookup(const struct nilfs_bmap *bmap,
+static int nilfs_direct_lookup(const struct nilfs_bmap *direct,
__u64 key, int level, __u64 *ptrp)
{
- struct nilfs_direct *direct;
__u64 ptr;
- direct = (struct nilfs_direct *)bmap; /* XXX: use macro for level 1 */
if (key > NILFS_DIRECT_KEY_MAX || level != 1)
return -ENOENT;
ptr = nilfs_direct_get_ptr(direct, key);
if (ptr == NILFS_BMAP_INVALID_PTR)
return -ENOENT;
- if (ptrp != NULL)
- *ptrp = ptr;
+ *ptrp = ptr;
return 0;
}
-static int nilfs_direct_lookup_contig(const struct nilfs_bmap *bmap,
+static int nilfs_direct_lookup_contig(const struct nilfs_bmap *direct,
__u64 key, __u64 *ptrp,
unsigned maxblocks)
{
- struct nilfs_direct *direct = (struct nilfs_direct *)bmap;
struct inode *dat = NULL;
__u64 ptr, ptr2;
sector_t blocknr;
@@ -79,8 +75,8 @@ static int nilfs_direct_lookup_contig(const struct nilfs_bmap *bmap,
if (ptr == NILFS_BMAP_INVALID_PTR)
return -ENOENT;
- if (NILFS_BMAP_USE_VBN(bmap)) {
- dat = nilfs_bmap_get_dat(bmap);
+ if (NILFS_BMAP_USE_VBN(direct)) {
+ dat = nilfs_bmap_get_dat(direct);
ret = nilfs_dat_translate(dat, ptr, &blocknr);
if (ret < 0)
return ret;
@@ -106,29 +102,21 @@ static int nilfs_direct_lookup_contig(const struct nilfs_bmap *bmap,
}
static __u64
-nilfs_direct_find_target_v(const struct nilfs_direct *direct, __u64 key)
+nilfs_direct_find_target_v(const struct nilfs_bmap *direct, __u64 key)
{
__u64 ptr;
- ptr = nilfs_bmap_find_target_seq(&direct->d_bmap, key);
+ ptr = nilfs_bmap_find_target_seq(direct, key);
if (ptr != NILFS_BMAP_INVALID_PTR)
/* sequential access */
return ptr;
else
/* block group */
- return nilfs_bmap_find_target_in_group(&direct->d_bmap);
-}
-
-static void nilfs_direct_set_target_v(struct nilfs_direct *direct,
- __u64 key, __u64 ptr)
-{
- direct->d_bmap.b_last_allocated_key = key;
- direct->d_bmap.b_last_allocated_ptr = ptr;
+ return nilfs_bmap_find_target_in_group(direct);
}
static int nilfs_direct_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
{
- struct nilfs_direct *direct = (struct nilfs_direct *)bmap;
union nilfs_bmap_ptr_req req;
struct inode *dat = NULL;
struct buffer_head *bh;
@@ -136,11 +124,11 @@ static int nilfs_direct_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
if (key > NILFS_DIRECT_KEY_MAX)
return -ENOENT;
- if (nilfs_direct_get_ptr(direct, key) != NILFS_BMAP_INVALID_PTR)
+ if (nilfs_direct_get_ptr(bmap, key) != NILFS_BMAP_INVALID_PTR)
return -EEXIST;
if (NILFS_BMAP_USE_VBN(bmap)) {
- req.bpr_ptr = nilfs_direct_find_target_v(direct, key);
+ req.bpr_ptr = nilfs_direct_find_target_v(bmap, key);
dat = nilfs_bmap_get_dat(bmap);
}
ret = nilfs_bmap_prepare_alloc_ptr(bmap, &req, dat);
@@ -150,13 +138,13 @@ static int nilfs_direct_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
set_buffer_nilfs_volatile(bh);
nilfs_bmap_commit_alloc_ptr(bmap, &req, dat);
- nilfs_direct_set_ptr(direct, key, req.bpr_ptr);
+ nilfs_direct_set_ptr(bmap, key, req.bpr_ptr);
if (!nilfs_bmap_dirty(bmap))
nilfs_bmap_set_dirty(bmap);
if (NILFS_BMAP_USE_VBN(bmap))
- nilfs_direct_set_target_v(direct, key, req.bpr_ptr);
+ nilfs_bmap_set_target_v(bmap, key, req.bpr_ptr);
nilfs_bmap_add_blocks(bmap, 1);
}
@@ -165,33 +153,30 @@ static int nilfs_direct_insert(struct nilfs_bmap *bmap, __u64 key, __u64 ptr)
static int nilfs_direct_delete(struct nilfs_bmap *bmap, __u64 key)
{
- struct nilfs_direct *direct = (struct nilfs_direct *)bmap;
union nilfs_bmap_ptr_req req;
struct inode *dat;
int ret;
if (key > NILFS_DIRECT_KEY_MAX ||
- nilfs_direct_get_ptr(direct, key) == NILFS_BMAP_INVALID_PTR)
+ nilfs_direct_get_ptr(bmap, key) == NILFS_BMAP_INVALID_PTR)
return -ENOENT;
dat = NILFS_BMAP_USE_VBN(bmap) ? nilfs_bmap_get_dat(bmap) : NULL;
- req.bpr_ptr = nilfs_direct_get_ptr(direct, key);
+ req.bpr_ptr = nilfs_direct_get_ptr(bmap, key);
ret = nilfs_bmap_prepare_end_ptr(bmap, &req, dat);
if (!ret) {
nilfs_bmap_commit_end_ptr(bmap, &req, dat);
- nilfs_direct_set_ptr(direct, key, NILFS_BMAP_INVALID_PTR);
+ nilfs_direct_set_ptr(bmap, key, NILFS_BMAP_INVALID_PTR);
nilfs_bmap_sub_blocks(bmap, 1);
}
return ret;
}
-static int nilfs_direct_last_key(const struct nilfs_bmap *bmap, __u64 *keyp)
+static int nilfs_direct_last_key(const struct nilfs_bmap *direct, __u64 *keyp)
{
- struct nilfs_direct *direct;
__u64 key, lastkey;
- direct = (struct nilfs_direct *)bmap;
lastkey = NILFS_DIRECT_KEY_MAX + 1;
for (key = NILFS_DIRECT_KEY_MIN; key <= NILFS_DIRECT_KEY_MAX; key++)
if (nilfs_direct_get_ptr(direct, key) !=
@@ -211,15 +196,13 @@ static int nilfs_direct_check_insert(const struct nilfs_bmap *bmap, __u64 key)
return key > NILFS_DIRECT_KEY_MAX;
}
-static int nilfs_direct_gather_data(struct nilfs_bmap *bmap,
+static int nilfs_direct_gather_data(struct nilfs_bmap *direct,
__u64 *keys, __u64 *ptrs, int nitems)
{
- struct nilfs_direct *direct;
__u64 key;
__u64 ptr;
int n;
- direct = (struct nilfs_direct *)bmap;
if (nitems > NILFS_DIRECT_NBLOCKS)
nitems = NILFS_DIRECT_NBLOCKS;
n = 0;
@@ -237,7 +220,6 @@ static int nilfs_direct_gather_data(struct nilfs_bmap *bmap,
int nilfs_direct_delete_and_convert(struct nilfs_bmap *bmap,
__u64 key, __u64 *keys, __u64 *ptrs, int n)
{
- struct nilfs_direct *direct;
__le64 *dptrs;
int ret, i, j;
@@ -253,12 +235,11 @@ int nilfs_direct_delete_and_convert(struct nilfs_bmap *bmap,
bmap->b_ops->bop_clear(bmap);
/* convert */
- direct = (struct nilfs_direct *)bmap;
- dptrs = nilfs_direct_dptrs(direct);
+ dptrs = nilfs_direct_dptrs(bmap);
for (i = 0, j = 0; i < NILFS_DIRECT_NBLOCKS; i++) {
if ((j < n) && (i == keys[j])) {
dptrs[i] = (i != key) ?
- nilfs_bmap_ptr_to_dptr(ptrs[j]) :
+ cpu_to_le64(ptrs[j]) :
NILFS_BMAP_INVALID_PTR;
j++;
} else
@@ -269,10 +250,9 @@ int nilfs_direct_delete_and_convert(struct nilfs_bmap *bmap,
return 0;
}
-static int nilfs_direct_propagate(const struct nilfs_bmap *bmap,
+static int nilfs_direct_propagate(struct nilfs_bmap *bmap,
struct buffer_head *bh)
{
- struct nilfs_direct *direct = (struct nilfs_direct *)bmap;
struct nilfs_palloc_req oldreq, newreq;
struct inode *dat;
__u64 key;
@@ -284,7 +264,7 @@ static int nilfs_direct_propagate(const struct nilfs_bmap *bmap,
dat = nilfs_bmap_get_dat(bmap);
key = nilfs_bmap_data_get_key(bmap, bh);
- ptr = nilfs_direct_get_ptr(direct, key);
+ ptr = nilfs_direct_get_ptr(bmap, key);
if (!buffer_nilfs_volatile(bh)) {
oldreq.pr_entry_nr = ptr;
newreq.pr_entry_nr = ptr;
@@ -294,20 +274,20 @@ static int nilfs_direct_propagate(const struct nilfs_bmap *bmap,
nilfs_dat_commit_update(dat, &oldreq, &newreq,
bmap->b_ptr_type == NILFS_BMAP_PTR_VS);
set_buffer_nilfs_volatile(bh);
- nilfs_direct_set_ptr(direct, key, newreq.pr_entry_nr);
+ nilfs_direct_set_ptr(bmap, key, newreq.pr_entry_nr);
} else
ret = nilfs_dat_mark_dirty(dat, ptr);
return ret;
}
-static int nilfs_direct_assign_v(struct nilfs_direct *direct,
+static int nilfs_direct_assign_v(struct nilfs_bmap *direct,
__u64 key, __u64 ptr,
struct buffer_head **bh,
sector_t blocknr,
union nilfs_binfo *binfo)
{
- struct inode *dat = nilfs_bmap_get_dat(&direct->d_bmap);
+ struct inode *dat = nilfs_bmap_get_dat(direct);
union nilfs_bmap_ptr_req req;
int ret;
@@ -315,13 +295,13 @@ static int nilfs_direct_assign_v(struct nilfs_direct *direct,
ret = nilfs_dat_prepare_start(dat, &req.bpr_req);
if (!ret) {
nilfs_dat_commit_start(dat, &req.bpr_req, blocknr);
- binfo->bi_v.bi_vblocknr = nilfs_bmap_ptr_to_dptr(ptr);
- binfo->bi_v.bi_blkoff = nilfs_bmap_key_to_dkey(key);
+ binfo->bi_v.bi_vblocknr = cpu_to_le64(ptr);
+ binfo->bi_v.bi_blkoff = cpu_to_le64(key);
}
return ret;
}
-static int nilfs_direct_assign_p(struct nilfs_direct *direct,
+static int nilfs_direct_assign_p(struct nilfs_bmap *direct,
__u64 key, __u64 ptr,
struct buffer_head **bh,
sector_t blocknr,
@@ -329,7 +309,7 @@ static int nilfs_direct_assign_p(struct nilfs_direct *direct,
{
nilfs_direct_set_ptr(direct, key, blocknr);
- binfo->bi_dat.bi_blkoff = nilfs_bmap_key_to_dkey(key);
+ binfo->bi_dat.bi_blkoff = cpu_to_le64(key);
binfo->bi_dat.bi_level = 0;
return 0;
@@ -340,18 +320,16 @@ static int nilfs_direct_assign(struct nilfs_bmap *bmap,
sector_t blocknr,
union nilfs_binfo *binfo)
{
- struct nilfs_direct *direct;
__u64 key;
__u64 ptr;
- direct = (struct nilfs_direct *)bmap;
key = nilfs_bmap_data_get_key(bmap, *bh);
if (unlikely(key > NILFS_DIRECT_KEY_MAX)) {
printk(KERN_CRIT "%s: invalid key: %llu\n", __func__,
(unsigned long long)key);
return -EINVAL;
}
- ptr = nilfs_direct_get_ptr(direct, key);
+ ptr = nilfs_direct_get_ptr(bmap, key);
if (unlikely(ptr == NILFS_BMAP_INVALID_PTR)) {
printk(KERN_CRIT "%s: invalid pointer: %llu\n", __func__,
(unsigned long long)ptr);
@@ -359,8 +337,8 @@ static int nilfs_direct_assign(struct nilfs_bmap *bmap,
}
return NILFS_BMAP_USE_VBN(bmap) ?
- nilfs_direct_assign_v(direct, key, ptr, bh, blocknr, binfo) :
- nilfs_direct_assign_p(direct, key, ptr, bh, blocknr, binfo);
+ nilfs_direct_assign_v(bmap, key, ptr, bh, blocknr, binfo) :
+ nilfs_direct_assign_p(bmap, key, ptr, bh, blocknr, binfo);
}
static const struct nilfs_bmap_operations nilfs_direct_ops = {
diff --git a/fs/nilfs2/direct.h b/fs/nilfs2/direct.h
index a5ffd66e25d..dc643de20a2 100644
--- a/fs/nilfs2/direct.h
+++ b/fs/nilfs2/direct.h
@@ -28,8 +28,6 @@
#include "bmap.h"
-struct nilfs_direct;
-
/**
* struct nilfs_direct_node - direct node
* @dn_flags: flags
@@ -40,15 +38,6 @@ struct nilfs_direct_node {
__u8 pad[7];
};
-/**
- * struct nilfs_direct - direct mapping
- * @d_bmap: bmap structure
- */
-struct nilfs_direct {
- struct nilfs_bmap d_bmap;
-};
-
-
#define NILFS_DIRECT_NBLOCKS (NILFS_BMAP_SIZE / sizeof(__le64) - 1)
#define NILFS_DIRECT_KEY_MIN 0
#define NILFS_DIRECT_KEY_MAX (NILFS_DIRECT_NBLOCKS - 1)
diff --git a/fs/nilfs2/gcinode.c b/fs/nilfs2/gcinode.c
index 145f03cd7d3..bed3a783129 100644
--- a/fs/nilfs2/gcinode.c
+++ b/fs/nilfs2/gcinode.c
@@ -48,6 +48,8 @@
#include <linux/slab.h>
#include <linux/swap.h>
#include "nilfs.h"
+#include "btree.h"
+#include "btnode.h"
#include "page.h"
#include "mdt.h"
#include "dat.h"
@@ -149,8 +151,10 @@ int nilfs_gccache_submit_read_data(struct inode *inode, sector_t blkoff,
int nilfs_gccache_submit_read_node(struct inode *inode, sector_t pbn,
__u64 vbn, struct buffer_head **out_bh)
{
- int ret = nilfs_btnode_submit_block(&NILFS_I(inode)->i_btnode_cache,
- vbn ? : pbn, pbn, out_bh);
+ int ret;
+
+ ret = nilfs_btnode_submit_block(&NILFS_I(inode)->i_btnode_cache,
+ vbn ? : pbn, pbn, READ, out_bh, &pbn);
if (ret == -EEXIST) /* internal code (cache hit) */
ret = 0;
return ret;
@@ -164,10 +168,15 @@ int nilfs_gccache_wait_and_mark_dirty(struct buffer_head *bh)
if (buffer_dirty(bh))
return -EEXIST;
- if (buffer_nilfs_node(bh))
+ if (buffer_nilfs_node(bh)) {
+ if (nilfs_btree_broken_node_block(bh)) {
+ clear_buffer_uptodate(bh);
+ return -EIO;
+ }
nilfs_btnode_mark_dirty(bh);
- else
+ } else {
nilfs_mdt_mark_buffer_dirty(bh);
+ }
return 0;
}
diff --git a/fs/nilfs2/mdt.c b/fs/nilfs2/mdt.c
index 024be8c35bb..d01aff4957d 100644
--- a/fs/nilfs2/mdt.c
+++ b/fs/nilfs2/mdt.c
@@ -28,6 +28,7 @@
#include <linux/swap.h>
#include <linux/slab.h>
#include "nilfs.h"
+#include "btnode.h"
#include "segment.h"
#include "page.h"
#include "mdt.h"
diff --git a/fs/nilfs2/nilfs.h b/fs/nilfs2/nilfs.h
index 47d6d792812..0842d775b3e 100644
--- a/fs/nilfs2/nilfs.h
+++ b/fs/nilfs2/nilfs.h
@@ -32,7 +32,6 @@
#include "the_nilfs.h"
#include "sb.h"
#include "bmap.h"
-#include "bmap_union.h"
/*
* nilfs inode data in memory
@@ -41,7 +40,7 @@ struct nilfs_inode_info {
__u32 i_flags;
unsigned long i_state; /* Dynamic state flags */
struct nilfs_bmap *i_bmap;
- union nilfs_bmap_union i_bmap_union;
+ struct nilfs_bmap i_bmap_data;
__u64 i_xattr; /* sector_t ??? */
__u32 i_dir_start_lookup;
__u64 i_cno; /* check point number for GC inode */
@@ -71,9 +70,7 @@ static inline struct nilfs_inode_info *NILFS_I(const struct inode *inode)
static inline struct nilfs_inode_info *
NILFS_BMAP_I(const struct nilfs_bmap *bmap)
{
- return container_of((union nilfs_bmap_union *)bmap,
- struct nilfs_inode_info,
- i_bmap_union);
+ return container_of(bmap, struct nilfs_inode_info, i_bmap_data);
}
static inline struct inode *NILFS_BTNC_I(struct address_space *btnc)
@@ -107,6 +104,14 @@ enum {
};
/*
+ * commit flags for nilfs_commit_super and nilfs_sync_super
+ */
+enum {
+ NILFS_SB_COMMIT = 0, /* Commit a super block alternately */
+ NILFS_SB_COMMIT_ALL /* Commit both super blocks */
+};
+
+/*
* Macros to check inode numbers
*/
#define NILFS_MDT_INO_BITS \
@@ -270,7 +275,14 @@ extern struct nilfs_super_block *
nilfs_read_super_block(struct super_block *, u64, int, struct buffer_head **);
extern int nilfs_store_magic_and_option(struct super_block *,
struct nilfs_super_block *, char *);
+extern int nilfs_check_feature_compatibility(struct super_block *,
+ struct nilfs_super_block *);
+extern void nilfs_set_log_cursor(struct nilfs_super_block *,
+ struct the_nilfs *);
+extern struct nilfs_super_block **nilfs_prepare_super(struct nilfs_sb_info *,
+ int flip);
extern int nilfs_commit_super(struct nilfs_sb_info *, int);
+extern int nilfs_cleanup_super(struct nilfs_sb_info *);
extern int nilfs_attach_checkpoint(struct nilfs_sb_info *, __u64);
extern void nilfs_detach_checkpoint(struct nilfs_sb_info *);
diff --git a/fs/nilfs2/page.c b/fs/nilfs2/page.c
index 8de3e1e4813..aab11db2cb0 100644
--- a/fs/nilfs2/page.c
+++ b/fs/nilfs2/page.c
@@ -37,7 +37,8 @@
#define NILFS_BUFFER_INHERENT_BITS \
((1UL << BH_Uptodate) | (1UL << BH_Mapped) | (1UL << BH_NILFS_Node) | \
- (1UL << BH_NILFS_Volatile) | (1UL << BH_NILFS_Allocated))
+ (1UL << BH_NILFS_Volatile) | (1UL << BH_NILFS_Allocated) | \
+ (1UL << BH_NILFS_Checked))
static struct buffer_head *
__nilfs_get_page_block(struct page *page, unsigned long block, pgoff_t index,
@@ -129,6 +130,7 @@ void nilfs_forget_buffer(struct buffer_head *bh)
lock_buffer(bh);
clear_buffer_nilfs_volatile(bh);
+ clear_buffer_nilfs_checked(bh);
clear_buffer_dirty(bh);
if (nilfs_page_buffers_clean(page))
__nilfs_clear_page_dirty(page);
@@ -480,6 +482,7 @@ void nilfs_clear_dirty_pages(struct address_space *mapping)
lock_buffer(bh);
clear_buffer_dirty(bh);
clear_buffer_nilfs_volatile(bh);
+ clear_buffer_nilfs_checked(bh);
clear_buffer_uptodate(bh);
clear_buffer_mapped(bh);
unlock_buffer(bh);
diff --git a/fs/nilfs2/page.h b/fs/nilfs2/page.h
index 8abca4d1c1f..f53d8da41ed 100644
--- a/fs/nilfs2/page.h
+++ b/fs/nilfs2/page.h
@@ -34,11 +34,13 @@ enum {
BH_NILFS_Allocated = BH_PrivateStart,
BH_NILFS_Node,
BH_NILFS_Volatile,
+ BH_NILFS_Checked,
};
BUFFER_FNS(NILFS_Allocated, nilfs_allocated) /* nilfs private buffers */
BUFFER_FNS(NILFS_Node, nilfs_node) /* nilfs node buffers */
BUFFER_FNS(NILFS_Volatile, nilfs_volatile)
+BUFFER_FNS(NILFS_Checked, nilfs_checked) /* buffer is verified */
void nilfs_mark_buffer_dirty(struct buffer_head *bh);
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index bae2a516b4e..83e3d8c61a0 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -91,27 +91,9 @@ static int nilfs_warn_segment_error(int err)
return -EINVAL;
}
-static void store_segsum_info(struct nilfs_segsum_info *ssi,
- struct nilfs_segment_summary *sum,
- unsigned int blocksize)
-{
- ssi->flags = le16_to_cpu(sum->ss_flags);
- ssi->seg_seq = le64_to_cpu(sum->ss_seq);
- ssi->ctime = le64_to_cpu(sum->ss_create);
- ssi->next = le64_to_cpu(sum->ss_next);
- ssi->nblocks = le32_to_cpu(sum->ss_nblocks);
- ssi->nfinfo = le32_to_cpu(sum->ss_nfinfo);
- ssi->sumbytes = le32_to_cpu(sum->ss_sumbytes);
-
- ssi->nsumblk = DIV_ROUND_UP(ssi->sumbytes, blocksize);
- ssi->nfileblk = ssi->nblocks - ssi->nsumblk - !!NILFS_SEG_HAS_SR(ssi);
-
- /* need to verify ->ss_bytes field if read ->ss_cno */
-}
-
/**
- * calc_crc_cont - check CRC of blocks continuously
- * @sbi: nilfs_sb_info
+ * nilfs_compute_checksum - compute checksum of blocks continuously
+ * @nilfs: nilfs object
* @bhs: buffer head of start block
* @sum: place to store result
* @offset: offset bytes in the first block
@@ -119,23 +101,25 @@ static void store_segsum_info(struct nilfs_segsum_info *ssi,
* @start: DBN of start block
* @nblock: number of blocks to be checked
*/
-static int calc_crc_cont(struct nilfs_sb_info *sbi, struct buffer_head *bhs,
- u32 *sum, unsigned long offset, u64 check_bytes,
- sector_t start, unsigned long nblock)
+static int nilfs_compute_checksum(struct the_nilfs *nilfs,
+ struct buffer_head *bhs, u32 *sum,
+ unsigned long offset, u64 check_bytes,
+ sector_t start, unsigned long nblock)
{
- unsigned long blocksize = sbi->s_super->s_blocksize;
+ unsigned int blocksize = nilfs->ns_blocksize;
unsigned long size;
u32 crc;
BUG_ON(offset >= blocksize);
check_bytes -= offset;
size = min_t(u64, check_bytes, blocksize - offset);
- crc = crc32_le(sbi->s_nilfs->ns_crc_seed,
+ crc = crc32_le(nilfs->ns_crc_seed,
(unsigned char *)bhs->b_data + offset, size);
if (--nblock > 0) {
do {
- struct buffer_head *bh
- = sb_bread(sbi->s_super, ++start);
+ struct buffer_head *bh;
+
+ bh = __bread(nilfs->ns_bdev, ++start, blocksize);
if (!bh)
return -EIO;
check_bytes -= size;
@@ -150,12 +134,12 @@ static int calc_crc_cont(struct nilfs_sb_info *sbi, struct buffer_head *bhs,
/**
* nilfs_read_super_root_block - read super root block
- * @sb: super_block
+ * @nilfs: nilfs object
* @sr_block: disk block number of the super root block
* @pbh: address of a buffer_head pointer to return super root buffer
* @check: CRC check flag
*/
-int nilfs_read_super_root_block(struct super_block *sb, sector_t sr_block,
+int nilfs_read_super_root_block(struct the_nilfs *nilfs, sector_t sr_block,
struct buffer_head **pbh, int check)
{
struct buffer_head *bh_sr;
@@ -164,7 +148,7 @@ int nilfs_read_super_root_block(struct super_block *sb, sector_t sr_block,
int ret;
*pbh = NULL;
- bh_sr = sb_bread(sb, sr_block);
+ bh_sr = __bread(nilfs->ns_bdev, sr_block, nilfs->ns_blocksize);
if (unlikely(!bh_sr)) {
ret = NILFS_SEG_FAIL_IO;
goto failed;
@@ -174,12 +158,13 @@ int nilfs_read_super_root_block(struct super_block *sb, sector_t sr_block,
if (check) {
unsigned bytes = le16_to_cpu(sr->sr_bytes);
- if (bytes == 0 || bytes > sb->s_blocksize) {
+ if (bytes == 0 || bytes > nilfs->ns_blocksize) {
ret = NILFS_SEG_FAIL_CHECKSUM_SUPER_ROOT;
goto failed_bh;
}
- if (calc_crc_cont(NILFS_SB(sb), bh_sr, &crc,
- sizeof(sr->sr_sum), bytes, sr_block, 1)) {
+ if (nilfs_compute_checksum(
+ nilfs, bh_sr, &crc, sizeof(sr->sr_sum), bytes,
+ sr_block, 1)) {
ret = NILFS_SEG_FAIL_IO;
goto failed_bh;
}
@@ -199,64 +184,76 @@ int nilfs_read_super_root_block(struct super_block *sb, sector_t sr_block,
}
/**
- * load_segment_summary - read segment summary of the specified partial segment
- * @sbi: nilfs_sb_info
- * @pseg_start: start disk block number of partial segment
- * @seg_seq: sequence number requested
- * @ssi: pointer to nilfs_segsum_info struct to store information
+ * nilfs_read_log_header - read summary header of the specified log
+ * @nilfs: nilfs object
+ * @start_blocknr: start block number of the log
+ * @sum: pointer to return segment summary structure
*/
-static int
-load_segment_summary(struct nilfs_sb_info *sbi, sector_t pseg_start,
- u64 seg_seq, struct nilfs_segsum_info *ssi)
+static struct buffer_head *
+nilfs_read_log_header(struct the_nilfs *nilfs, sector_t start_blocknr,
+ struct nilfs_segment_summary **sum)
{
struct buffer_head *bh_sum;
- struct nilfs_segment_summary *sum;
+
+ bh_sum = __bread(nilfs->ns_bdev, start_blocknr, nilfs->ns_blocksize);
+ if (bh_sum)
+ *sum = (struct nilfs_segment_summary *)bh_sum->b_data;
+ return bh_sum;
+}
+
+/**
+ * nilfs_validate_log - verify consistency of log
+ * @nilfs: nilfs object
+ * @seg_seq: sequence number of segment
+ * @bh_sum: buffer head of summary block
+ * @sum: segment summary struct
+ */
+static int nilfs_validate_log(struct the_nilfs *nilfs, u64 seg_seq,
+ struct buffer_head *bh_sum,
+ struct nilfs_segment_summary *sum)
+{
unsigned long nblock;
u32 crc;
- int ret = NILFS_SEG_FAIL_IO;
+ int ret;
- bh_sum = sb_bread(sbi->s_super, pseg_start);
- if (!bh_sum)
+ ret = NILFS_SEG_FAIL_MAGIC;
+ if (le32_to_cpu(sum->ss_magic) != NILFS_SEGSUM_MAGIC)
goto out;
- sum = (struct nilfs_segment_summary *)bh_sum->b_data;
-
- /* Check consistency of segment summary */
- if (le32_to_cpu(sum->ss_magic) != NILFS_SEGSUM_MAGIC) {
- ret = NILFS_SEG_FAIL_MAGIC;
- goto failed;
- }
- store_segsum_info(ssi, sum, sbi->s_super->s_blocksize);
- if (seg_seq != ssi->seg_seq) {
- ret = NILFS_SEG_FAIL_SEQ;
- goto failed;
- }
+ ret = NILFS_SEG_FAIL_SEQ;
+ if (le64_to_cpu(sum->ss_seq) != seg_seq)
+ goto out;
- nblock = ssi->nblocks;
- if (unlikely(nblock == 0 ||
- nblock > sbi->s_nilfs->ns_blocks_per_segment)) {
+ nblock = le32_to_cpu(sum->ss_nblocks);
+ ret = NILFS_SEG_FAIL_CONSISTENCY;
+ if (unlikely(nblock == 0 || nblock > nilfs->ns_blocks_per_segment))
/* This limits the number of blocks read in the CRC check */
- ret = NILFS_SEG_FAIL_CONSISTENCY;
- goto failed;
- }
- if (calc_crc_cont(sbi, bh_sum, &crc, sizeof(sum->ss_datasum),
- ((u64)nblock << sbi->s_super->s_blocksize_bits),
- pseg_start, nblock)) {
- ret = NILFS_SEG_FAIL_IO;
- goto failed;
- }
- if (crc == le32_to_cpu(sum->ss_datasum))
- ret = 0;
- else
- ret = NILFS_SEG_FAIL_CHECKSUM_FULL;
- failed:
- brelse(bh_sum);
- out:
+ goto out;
+
+ ret = NILFS_SEG_FAIL_IO;
+ if (nilfs_compute_checksum(nilfs, bh_sum, &crc, sizeof(sum->ss_datasum),
+ ((u64)nblock << nilfs->ns_blocksize_bits),
+ bh_sum->b_blocknr, nblock))
+ goto out;
+
+ ret = NILFS_SEG_FAIL_CHECKSUM_FULL;
+ if (crc != le32_to_cpu(sum->ss_datasum))
+ goto out;
+ ret = 0;
+out:
return ret;
}
-static void *segsum_get(struct super_block *sb, struct buffer_head **pbh,
- unsigned int *offset, unsigned int bytes)
+/**
+ * nilfs_read_summary_info - read an item on summary blocks of a log
+ * @nilfs: nilfs object
+ * @pbh: the current buffer head on summary blocks [in, out]
+ * @offset: the current byte offset on summary blocks [in, out]
+ * @bytes: byte size of the item to be read
+ */
+static void *nilfs_read_summary_info(struct the_nilfs *nilfs,
+ struct buffer_head **pbh,
+ unsigned int *offset, unsigned int bytes)
{
void *ptr;
sector_t blocknr;
@@ -265,7 +262,8 @@ static void *segsum_get(struct super_block *sb, struct buffer_head **pbh,
if (bytes > (*pbh)->b_size - *offset) {
blocknr = (*pbh)->b_blocknr;
brelse(*pbh);
- *pbh = sb_bread(sb, blocknr + 1);
+ *pbh = __bread(nilfs->ns_bdev, blocknr + 1,
+ nilfs->ns_blocksize);
if (unlikely(!*pbh))
return NULL;
*offset = 0;
@@ -275,9 +273,18 @@ static void *segsum_get(struct super_block *sb, struct buffer_head **pbh,
return ptr;
}
-static void segsum_skip(struct super_block *sb, struct buffer_head **pbh,
- unsigned int *offset, unsigned int bytes,
- unsigned long count)
+/**
+ * nilfs_skip_summary_info - skip items on summary blocks of a log
+ * @nilfs: nilfs object
+ * @pbh: the current buffer head on summary blocks [in, out]
+ * @offset: the current byte offset on summary blocks [in, out]
+ * @bytes: byte size of the item to be skipped
+ * @count: number of items to be skipped
+ */
+static void nilfs_skip_summary_info(struct the_nilfs *nilfs,
+ struct buffer_head **pbh,
+ unsigned int *offset, unsigned int bytes,
+ unsigned long count)
{
unsigned int rest_item_in_current_block
= ((*pbh)->b_size - *offset) / bytes;
@@ -294,36 +301,46 @@ static void segsum_skip(struct super_block *sb, struct buffer_head **pbh,
*offset = bytes * (count - (bcnt - 1) * nitem_per_block);
brelse(*pbh);
- *pbh = sb_bread(sb, blocknr + bcnt);
+ *pbh = __bread(nilfs->ns_bdev, blocknr + bcnt,
+ nilfs->ns_blocksize);
}
}
-static int
-collect_blocks_from_segsum(struct nilfs_sb_info *sbi, sector_t sum_blocknr,
- struct nilfs_segsum_info *ssi,
- struct list_head *head)
+/**
+ * nilfs_scan_dsync_log - get block information of a log written for data sync
+ * @nilfs: nilfs object
+ * @start_blocknr: start block number of the log
+ * @sum: log summary information
+ * @head: list head to add nilfs_recovery_block struct
+ */
+static int nilfs_scan_dsync_log(struct the_nilfs *nilfs, sector_t start_blocknr,
+ struct nilfs_segment_summary *sum,
+ struct list_head *head)
{
struct buffer_head *bh;
unsigned int offset;
- unsigned long nfinfo = ssi->nfinfo;
- sector_t blocknr = sum_blocknr + ssi->nsumblk;
+ u32 nfinfo, sumbytes;
+ sector_t blocknr;
ino_t ino;
int err = -EIO;
+ nfinfo = le32_to_cpu(sum->ss_nfinfo);
if (!nfinfo)
return 0;
- bh = sb_bread(sbi->s_super, sum_blocknr);
+ sumbytes = le32_to_cpu(sum->ss_sumbytes);
+ blocknr = start_blocknr + DIV_ROUND_UP(sumbytes, nilfs->ns_blocksize);
+ bh = __bread(nilfs->ns_bdev, start_blocknr, nilfs->ns_blocksize);
if (unlikely(!bh))
goto out;
- offset = le16_to_cpu(
- ((struct nilfs_segment_summary *)bh->b_data)->ss_bytes);
+ offset = le16_to_cpu(sum->ss_bytes);
for (;;) {
unsigned long nblocks, ndatablk, nnodeblk;
struct nilfs_finfo *finfo;
- finfo = segsum_get(sbi->s_super, &bh, &offset, sizeof(*finfo));
+ finfo = nilfs_read_summary_info(nilfs, &bh, &offset,
+ sizeof(*finfo));
if (unlikely(!finfo))
goto out;
@@ -336,8 +353,8 @@ collect_blocks_from_segsum(struct nilfs_sb_info *sbi, sector_t sum_blocknr,
struct nilfs_recovery_block *rb;
struct nilfs_binfo_v *binfo;
- binfo = segsum_get(sbi->s_super, &bh, &offset,
- sizeof(*binfo));
+ binfo = nilfs_read_summary_info(nilfs, &bh, &offset,
+ sizeof(*binfo));
if (unlikely(!binfo))
goto out;
@@ -355,9 +372,9 @@ collect_blocks_from_segsum(struct nilfs_sb_info *sbi, sector_t sum_blocknr,
}
if (--nfinfo == 0)
break;
- blocknr += nnodeblk; /* always 0 for the data sync segments */
- segsum_skip(sbi->s_super, &bh, &offset, sizeof(__le64),
- nnodeblk);
+ blocknr += nnodeblk; /* always 0 for data sync logs */
+ nilfs_skip_summary_info(nilfs, &bh, &offset, sizeof(__le64),
+ nnodeblk);
if (unlikely(!bh))
goto out;
}
@@ -467,14 +484,14 @@ static int nilfs_prepare_segment_for_recovery(struct the_nilfs *nilfs,
return err;
}
-static int nilfs_recovery_copy_block(struct nilfs_sb_info *sbi,
+static int nilfs_recovery_copy_block(struct the_nilfs *nilfs,
struct nilfs_recovery_block *rb,
struct page *page)
{
struct buffer_head *bh_org;
void *kaddr;
- bh_org = sb_bread(sbi->s_super, rb->blocknr);
+ bh_org = __bread(nilfs->ns_bdev, rb->blocknr, nilfs->ns_blocksize);
if (unlikely(!bh_org))
return -EIO;
@@ -485,13 +502,14 @@ static int nilfs_recovery_copy_block(struct nilfs_sb_info *sbi,
return 0;
}
-static int recover_dsync_blocks(struct nilfs_sb_info *sbi,
- struct list_head *head,
- unsigned long *nr_salvaged_blocks)
+static int nilfs_recover_dsync_blocks(struct the_nilfs *nilfs,
+ struct nilfs_sb_info *sbi,
+ struct list_head *head,
+ unsigned long *nr_salvaged_blocks)
{
struct inode *inode;
struct nilfs_recovery_block *rb, *n;
- unsigned blocksize = sbi->s_super->s_blocksize;
+ unsigned blocksize = nilfs->ns_blocksize;
struct page *page;
loff_t pos;
int err = 0, err2 = 0;
@@ -511,7 +529,7 @@ static int recover_dsync_blocks(struct nilfs_sb_info *sbi,
if (unlikely(err))
goto failed_inode;
- err = nilfs_recovery_copy_block(sbi, rb, page);
+ err = nilfs_recovery_copy_block(nilfs, rb, page);
if (unlikely(err))
goto failed_page;
@@ -551,18 +569,20 @@ static int recover_dsync_blocks(struct nilfs_sb_info *sbi,
/**
* nilfs_do_roll_forward - salvage logical segments newer than the latest
* checkpoint
+ * @nilfs: nilfs object
* @sbi: nilfs_sb_info
- * @nilfs: the_nilfs
* @ri: pointer to a nilfs_recovery_info
*/
static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
struct nilfs_sb_info *sbi,
struct nilfs_recovery_info *ri)
{
- struct nilfs_segsum_info ssi;
+ struct buffer_head *bh_sum = NULL;
+ struct nilfs_segment_summary *sum;
sector_t pseg_start;
sector_t seg_start, seg_end; /* Starting/ending DBN of full segment */
unsigned long nsalvaged_blocks = 0;
+ unsigned int flags;
u64 seg_seq;
__u64 segnum, nextnum = 0;
int empty_seg = 0;
@@ -581,8 +601,14 @@ static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
nilfs_get_segment_range(nilfs, segnum, &seg_start, &seg_end);
while (segnum != ri->ri_segnum || pseg_start <= ri->ri_pseg_start) {
+ brelse(bh_sum);
+ bh_sum = nilfs_read_log_header(nilfs, pseg_start, &sum);
+ if (!bh_sum) {
+ err = -EIO;
+ goto failed;
+ }
- ret = load_segment_summary(sbi, pseg_start, seg_seq, &ssi);
+ ret = nilfs_validate_log(nilfs, seg_seq, bh_sum, sum);
if (ret) {
if (ret == NILFS_SEG_FAIL_IO) {
err = -EIO;
@@ -590,33 +616,38 @@ static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
}
goto strayed;
}
- if (unlikely(NILFS_SEG_HAS_SR(&ssi)))
+
+ flags = le16_to_cpu(sum->ss_flags);
+ if (flags & NILFS_SS_SR)
goto confused;
/* Found a valid partial segment; do recovery actions */
- nextnum = nilfs_get_segnum_of_block(nilfs, ssi.next);
+ nextnum = nilfs_get_segnum_of_block(nilfs,
+ le64_to_cpu(sum->ss_next));
empty_seg = 0;
- nilfs->ns_ctime = ssi.ctime;
- if (!(ssi.flags & NILFS_SS_GC))
- nilfs->ns_nongc_ctime = ssi.ctime;
+ nilfs->ns_ctime = le64_to_cpu(sum->ss_create);
+ if (!(flags & NILFS_SS_GC))
+ nilfs->ns_nongc_ctime = nilfs->ns_ctime;
switch (state) {
case RF_INIT_ST:
- if (!NILFS_SEG_LOGBGN(&ssi) || !NILFS_SEG_DSYNC(&ssi))
+ if (!(flags & NILFS_SS_LOGBGN) ||
+ !(flags & NILFS_SS_SYNDT))
goto try_next_pseg;
state = RF_DSYNC_ST;
/* Fall through */
case RF_DSYNC_ST:
- if (!NILFS_SEG_DSYNC(&ssi))
+ if (!(flags & NILFS_SS_SYNDT))
goto confused;
- err = collect_blocks_from_segsum(
- sbi, pseg_start, &ssi, &dsync_blocks);
+ err = nilfs_scan_dsync_log(nilfs, pseg_start, sum,
+ &dsync_blocks);
if (unlikely(err))
goto failed;
- if (NILFS_SEG_LOGEND(&ssi)) {
- err = recover_dsync_blocks(
- sbi, &dsync_blocks, &nsalvaged_blocks);
+ if (flags & NILFS_SS_LOGEND) {
+ err = nilfs_recover_dsync_blocks(
+ nilfs, sbi, &dsync_blocks,
+ &nsalvaged_blocks);
if (unlikely(err))
goto failed;
state = RF_INIT_ST;
@@ -627,7 +658,7 @@ static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
try_next_pseg:
if (pseg_start == ri->ri_lsegs_end)
break;
- pseg_start += ssi.nblocks;
+ pseg_start += le32_to_cpu(sum->ss_nblocks);
if (pseg_start < seg_end)
continue;
goto feed_segment;
@@ -652,8 +683,9 @@ static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
ri->ri_need_recovery = NILFS_RECOVERY_ROLLFORWARD_DONE;
}
out:
+ brelse(bh_sum);
dispose_recovery_list(&dsync_blocks);
- nilfs_detach_writer(sbi->s_nilfs, sbi);
+ nilfs_detach_writer(nilfs, sbi);
return err;
confused:
@@ -667,7 +699,6 @@ static int nilfs_do_roll_forward(struct the_nilfs *nilfs,
}
static void nilfs_finish_roll_forward(struct the_nilfs *nilfs,
- struct nilfs_sb_info *sbi,
struct nilfs_recovery_info *ri)
{
struct buffer_head *bh;
@@ -677,7 +708,7 @@ static void nilfs_finish_roll_forward(struct the_nilfs *nilfs,
nilfs_get_segnum_of_block(nilfs, ri->ri_super_root))
return;
- bh = sb_getblk(sbi->s_super, ri->ri_lsegs_start);
+ bh = __getblk(nilfs->ns_bdev, ri->ri_lsegs_start, nilfs->ns_blocksize);
BUG_ON(!bh);
memset(bh->b_data, 0, bh->b_size);
set_buffer_dirty(bh);
@@ -690,9 +721,8 @@ static void nilfs_finish_roll_forward(struct the_nilfs *nilfs,
}
/**
- * nilfs_recover_logical_segments - salvage logical segments written after
- * the latest super root
- * @nilfs: the_nilfs
+ * nilfs_salvage_orphan_logs - salvage logs written after the latest checkpoint
+ * @nilfs: nilfs object
* @sbi: nilfs_sb_info
* @ri: pointer to a nilfs_recovery_info struct to store search results.
*
@@ -709,9 +739,9 @@ static void nilfs_finish_roll_forward(struct the_nilfs *nilfs,
*
* %-ENOMEM - Insufficient memory available.
*/
-int nilfs_recover_logical_segments(struct the_nilfs *nilfs,
- struct nilfs_sb_info *sbi,
- struct nilfs_recovery_info *ri)
+int nilfs_salvage_orphan_logs(struct the_nilfs *nilfs,
+ struct nilfs_sb_info *sbi,
+ struct nilfs_recovery_info *ri)
{
int err;
@@ -751,7 +781,7 @@ int nilfs_recover_logical_segments(struct the_nilfs *nilfs,
goto failed;
}
- nilfs_finish_roll_forward(nilfs, sbi, ri);
+ nilfs_finish_roll_forward(nilfs, ri);
}
failed:
@@ -762,7 +792,6 @@ int nilfs_recover_logical_segments(struct the_nilfs *nilfs,
/**
* nilfs_search_super_root - search the latest valid super root
* @nilfs: the_nilfs
- * @sbi: nilfs_sb_info
* @ri: pointer to a nilfs_recovery_info struct to store search results.
*
* nilfs_search_super_root() looks for the latest super-root from a partial
@@ -775,14 +804,19 @@ int nilfs_recover_logical_segments(struct the_nilfs *nilfs,
* %-EINVAL - No valid segment found
*
* %-EIO - I/O error
+ *
+ * %-ENOMEM - Insufficient memory available.
*/
-int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
+int nilfs_search_super_root(struct the_nilfs *nilfs,
struct nilfs_recovery_info *ri)
{
- struct nilfs_segsum_info ssi;
+ struct buffer_head *bh_sum = NULL;
+ struct nilfs_segment_summary *sum;
sector_t pseg_start, pseg_end, sr_pseg_start = 0;
sector_t seg_start, seg_end; /* range of full segment (block number) */
sector_t b, end;
+ unsigned long nblocks;
+ unsigned int flags;
u64 seg_seq;
__u64 segnum, nextnum = 0;
__u64 cno;
@@ -801,17 +835,24 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
/* Read ahead segment */
b = seg_start;
while (b <= seg_end)
- sb_breadahead(sbi->s_super, b++);
+ __breadahead(nilfs->ns_bdev, b++, nilfs->ns_blocksize);
for (;;) {
- /* Load segment summary */
- ret = load_segment_summary(sbi, pseg_start, seg_seq, &ssi);
+ brelse(bh_sum);
+ ret = NILFS_SEG_FAIL_IO;
+ bh_sum = nilfs_read_log_header(nilfs, pseg_start, &sum);
+ if (!bh_sum)
+ goto failed;
+
+ ret = nilfs_validate_log(nilfs, seg_seq, bh_sum, sum);
if (ret) {
if (ret == NILFS_SEG_FAIL_IO)
goto failed;
goto strayed;
}
- pseg_end = pseg_start + ssi.nblocks - 1;
+
+ nblocks = le32_to_cpu(sum->ss_nblocks);
+ pseg_end = pseg_start + nblocks - 1;
if (unlikely(pseg_end > seg_end)) {
ret = NILFS_SEG_FAIL_CONSISTENCY;
goto strayed;
@@ -821,11 +862,13 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
ri->ri_pseg_start = pseg_start;
ri->ri_seq = seg_seq;
ri->ri_segnum = segnum;
- nextnum = nilfs_get_segnum_of_block(nilfs, ssi.next);
+ nextnum = nilfs_get_segnum_of_block(nilfs,
+ le64_to_cpu(sum->ss_next));
ri->ri_nextnum = nextnum;
empty_seg = 0;
- if (!NILFS_SEG_HAS_SR(&ssi) && !scan_newer) {
+ flags = le16_to_cpu(sum->ss_flags);
+ if (!(flags & NILFS_SS_SR) && !scan_newer) {
/* This will never happen because a superblock
(last_segment) always points to a pseg
having a super root. */
@@ -836,14 +879,15 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
if (pseg_start == seg_start) {
nilfs_get_segment_range(nilfs, nextnum, &b, &end);
while (b <= end)
- sb_breadahead(sbi->s_super, b++);
+ __breadahead(nilfs->ns_bdev, b++,
+ nilfs->ns_blocksize);
}
- if (!NILFS_SEG_HAS_SR(&ssi)) {
- if (!ri->ri_lsegs_start && NILFS_SEG_LOGBGN(&ssi)) {
+ if (!(flags & NILFS_SS_SR)) {
+ if (!ri->ri_lsegs_start && (flags & NILFS_SS_LOGBGN)) {
ri->ri_lsegs_start = pseg_start;
ri->ri_lsegs_start_seq = seg_seq;
}
- if (NILFS_SEG_LOGEND(&ssi))
+ if (flags & NILFS_SS_LOGEND)
ri->ri_lsegs_end = pseg_start;
goto try_next_pseg;
}
@@ -854,12 +898,12 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
ri->ri_lsegs_start = ri->ri_lsegs_end = 0;
nilfs_dispose_segment_list(&segments);
- nilfs->ns_pseg_offset = (sr_pseg_start = pseg_start)
- + ssi.nblocks - seg_start;
+ sr_pseg_start = pseg_start;
+ nilfs->ns_pseg_offset = pseg_start + nblocks - seg_start;
nilfs->ns_seg_seq = seg_seq;
nilfs->ns_segnum = segnum;
nilfs->ns_cno = cno; /* nilfs->ns_cno = ri->ri_cno + 1 */
- nilfs->ns_ctime = ssi.ctime;
+ nilfs->ns_ctime = le64_to_cpu(sum->ss_create);
nilfs->ns_nextnum = nextnum;
if (scan_newer)
@@ -870,15 +914,9 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
scan_newer = 1;
}
- /* reset region for roll-forward */
- pseg_start += ssi.nblocks;
- if (pseg_start < seg_end)
- continue;
- goto feed_segment;
-
try_next_pseg:
/* Standing on a course, or met an inconsistent state */
- pseg_start += ssi.nblocks;
+ pseg_start += nblocks;
if (pseg_start < seg_end)
continue;
goto feed_segment;
@@ -909,6 +947,7 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
super_root_found:
/* Updating pointers relating to the latest checkpoint */
+ brelse(bh_sum);
list_splice_tail(&segments, &ri->ri_used_segments);
nilfs->ns_last_pseg = sr_pseg_start;
nilfs->ns_last_seq = nilfs->ns_seg_seq;
@@ -916,6 +955,7 @@ int nilfs_search_super_root(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi,
return 0;
failed:
+ brelse(bh_sum);
nilfs_dispose_segment_list(&segments);
return (ret < 0) ? ret : nilfs_warn_segment_error(ret);
}
diff --git a/fs/nilfs2/segbuf.h b/fs/nilfs2/segbuf.h
index 85fbb66455e..b04f08cc239 100644
--- a/fs/nilfs2/segbuf.h
+++ b/fs/nilfs2/segbuf.h
@@ -54,17 +54,6 @@ struct nilfs_segsum_info {
sector_t next;
};
-/* macro for the flags */
-#define NILFS_SEG_HAS_SR(sum) ((sum)->flags & NILFS_SS_SR)
-#define NILFS_SEG_LOGBGN(sum) ((sum)->flags & NILFS_SS_LOGBGN)
-#define NILFS_SEG_LOGEND(sum) ((sum)->flags & NILFS_SS_LOGEND)
-#define NILFS_SEG_DSYNC(sum) ((sum)->flags & NILFS_SS_SYNDT)
-#define NILFS_SEG_SIMPLEX(sum) \
- (((sum)->flags & (NILFS_SS_LOGBGN | NILFS_SS_LOGEND)) == \
- (NILFS_SS_LOGBGN | NILFS_SS_LOGEND))
-
-#define NILFS_SEG_EMPTY(sum) ((sum)->nblocks == (sum)->nsumblk)
-
/**
* struct nilfs_segment_buffer - Segment buffer
* @sb_super: back pointer to a superblock struct
@@ -141,6 +130,19 @@ int nilfs_segbuf_extend_payload(struct nilfs_segment_buffer *,
struct buffer_head **);
void nilfs_segbuf_fill_in_segsum(struct nilfs_segment_buffer *);
+static inline int nilfs_segbuf_simplex(struct nilfs_segment_buffer *segbuf)
+{
+ unsigned int flags = segbuf->sb_sum.flags;
+
+ return (flags & (NILFS_SS_LOGBGN | NILFS_SS_LOGEND)) ==
+ (NILFS_SS_LOGBGN | NILFS_SS_LOGEND);
+}
+
+static inline int nilfs_segbuf_empty(struct nilfs_segment_buffer *segbuf)
+{
+ return segbuf->sb_sum.nblocks == segbuf->sb_sum.nsumblk;
+}
+
static inline void
nilfs_segbuf_add_segsum_buffer(struct nilfs_segment_buffer *segbuf,
struct buffer_head *bh)
diff --git a/fs/nilfs2/segment.c b/fs/nilfs2/segment.c
index c9201649cc4..9fd051a33c4 100644
--- a/fs/nilfs2/segment.c
+++ b/fs/nilfs2/segment.c
@@ -1914,12 +1914,12 @@ static void nilfs_segctor_complete_write(struct nilfs_sc_info *sci)
}
}
- if (!NILFS_SEG_SIMPLEX(&segbuf->sb_sum)) {
- if (NILFS_SEG_LOGBGN(&segbuf->sb_sum)) {
+ if (!nilfs_segbuf_simplex(segbuf)) {
+ if (segbuf->sb_sum.flags & NILFS_SS_LOGBGN) {
set_bit(NILFS_SC_UNCLOSED, &sci->sc_flags);
sci->sc_lseg_stime = jiffies;
}
- if (NILFS_SEG_LOGEND(&segbuf->sb_sum))
+ if (segbuf->sb_sum.flags & NILFS_SS_LOGEND)
clear_bit(NILFS_SC_UNCLOSED, &sci->sc_flags);
}
}
@@ -1951,7 +1951,6 @@ static void nilfs_segctor_complete_write(struct nilfs_sc_info *sci)
if (update_sr) {
nilfs_set_last_segment(nilfs, segbuf->sb_pseg_start,
segbuf->sb_sum.seg_seq, nilfs->ns_cno++);
- set_nilfs_sb_dirty(nilfs);
clear_bit(NILFS_SC_HAVE_DELTA, &sci->sc_flags);
clear_bit(NILFS_SC_DIRTY, &sci->sc_flags);
@@ -2082,7 +2081,7 @@ static int nilfs_segctor_do_construct(struct nilfs_sc_info *sci, int mode)
/* Avoid empty segment */
if (sci->sc_stage.scnt == NILFS_ST_DONE &&
- NILFS_SEG_EMPTY(&sci->sc_curseg->sb_sum)) {
+ nilfs_segbuf_empty(sci->sc_curseg)) {
nilfs_segctor_abort_construction(sci, nilfs, 1);
goto out;
}
@@ -2408,6 +2407,7 @@ static int nilfs_segctor_construct(struct nilfs_sc_info *sci, int mode)
{
struct nilfs_sb_info *sbi = sci->sc_sbi;
struct the_nilfs *nilfs = sbi->s_nilfs;
+ struct nilfs_super_block **sbp;
int err = 0;
nilfs_segctor_accept(sci);
@@ -2423,8 +2423,13 @@ static int nilfs_segctor_construct(struct nilfs_sc_info *sci, int mode)
if (test_bit(NILFS_SC_SUPER_ROOT, &sci->sc_flags) &&
nilfs_discontinued(nilfs)) {
down_write(&nilfs->ns_sem);
- err = nilfs_commit_super(
- sbi, nilfs_altsb_need_update(nilfs));
+ err = -EIO;
+ sbp = nilfs_prepare_super(sbi,
+ nilfs_sb_will_flip(nilfs));
+ if (likely(sbp)) {
+ nilfs_set_log_cursor(sbp[0], nilfs);
+ err = nilfs_commit_super(sbi, NILFS_SB_COMMIT);
+ }
up_write(&nilfs->ns_sem);
}
}
diff --git a/fs/nilfs2/segment.h b/fs/nilfs2/segment.h
index 01e20dbb217..17c487bd815 100644
--- a/fs/nilfs2/segment.h
+++ b/fs/nilfs2/segment.h
@@ -234,13 +234,13 @@ extern int nilfs_attach_segment_constructor(struct nilfs_sb_info *);
extern void nilfs_detach_segment_constructor(struct nilfs_sb_info *);
/* recovery.c */
-extern int nilfs_read_super_root_block(struct super_block *, sector_t,
+extern int nilfs_read_super_root_block(struct the_nilfs *, sector_t,
struct buffer_head **, int);
-extern int nilfs_search_super_root(struct the_nilfs *, struct nilfs_sb_info *,
+extern int nilfs_search_super_root(struct the_nilfs *,
struct nilfs_recovery_info *);
-extern int nilfs_recover_logical_segments(struct the_nilfs *,
- struct nilfs_sb_info *,
- struct nilfs_recovery_info *);
+extern int nilfs_salvage_orphan_logs(struct the_nilfs *,
+ struct nilfs_sb_info *,
+ struct nilfs_recovery_info *);
extern void nilfs_dispose_segment_list(struct list_head *);
#endif /* _NILFS_SEGMENT_H */
diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
index 414ef68931c..26078b3407c 100644
--- a/fs/nilfs2/super.c
+++ b/fs/nilfs2/super.c
@@ -55,6 +55,8 @@
#include "nilfs.h"
#include "mdt.h"
#include "alloc.h"
+#include "btree.h"
+#include "btnode.h"
#include "page.h"
#include "cpfile.h"
#include "ifile.h"
@@ -74,6 +76,25 @@ struct kmem_cache *nilfs_btree_path_cache;
static int nilfs_remount(struct super_block *sb, int *flags, char *data);
+static void nilfs_set_error(struct nilfs_sb_info *sbi)
+{
+ struct the_nilfs *nilfs = sbi->s_nilfs;
+ struct nilfs_super_block **sbp;
+
+ down_write(&nilfs->ns_sem);
+ if (!(nilfs->ns_mount_state & NILFS_ERROR_FS)) {
+ nilfs->ns_mount_state |= NILFS_ERROR_FS;
+ sbp = nilfs_prepare_super(sbi, 0);
+ if (likely(sbp)) {
+ sbp[0]->s_state |= cpu_to_le16(NILFS_ERROR_FS);
+ if (sbp[1])
+ sbp[1]->s_state |= cpu_to_le16(NILFS_ERROR_FS);
+ nilfs_commit_super(sbi, NILFS_SB_COMMIT_ALL);
+ }
+ }
+ up_write(&nilfs->ns_sem);
+}
+
/**
* nilfs_error() - report failure condition on a filesystem
*
@@ -99,16 +120,7 @@ void nilfs_error(struct super_block *sb, const char *function,
va_end(args);
if (!(sb->s_flags & MS_RDONLY)) {
- struct the_nilfs *nilfs = sbi->s_nilfs;
-
- down_write(&nilfs->ns_sem);
- if (!(nilfs->ns_mount_state & NILFS_ERROR_FS)) {
- nilfs->ns_mount_state |= NILFS_ERROR_FS;
- nilfs->ns_sbp[0]->s_state |=
- cpu_to_le16(NILFS_ERROR_FS);
- nilfs_commit_super(sbi, 1);
- }
- up_write(&nilfs->ns_sem);
+ nilfs_set_error(sbi);
if (nilfs_test_opt(sbi, ERRORS_RO)) {
printk(KERN_CRIT "Remounting filesystem read-only\n");
@@ -176,7 +188,7 @@ static void nilfs_clear_inode(struct inode *inode)
nilfs_btnode_cache_clear(&ii->i_btnode_cache);
}
-static int nilfs_sync_super(struct nilfs_sb_info *sbi, int dupsb)
+static int nilfs_sync_super(struct nilfs_sb_info *sbi, int flag)
{
struct the_nilfs *nilfs = sbi->s_nilfs;
int err;
@@ -202,12 +214,20 @@ static int nilfs_sync_super(struct nilfs_sb_info *sbi, int dupsb)
printk(KERN_ERR
"NILFS: unable to write superblock (err=%d)\n", err);
if (err == -EIO && nilfs->ns_sbh[1]) {
+ /*
+ * sbp[0] points to newer log than sbp[1],
+ * so copy sbp[0] to sbp[1] to take over sbp[0].
+ */
+ memcpy(nilfs->ns_sbp[1], nilfs->ns_sbp[0],
+ nilfs->ns_sbsize);
nilfs_fall_back_super_block(nilfs);
goto retry;
}
} else {
struct nilfs_super_block *sbp = nilfs->ns_sbp[0];
+ nilfs->ns_sbwcount++;
+
/*
* The latest segment becomes trailable from the position
* written in superblock.
@@ -216,66 +236,122 @@ static int nilfs_sync_super(struct nilfs_sb_info *sbi, int dupsb)
/* update GC protection for recent segments */
if (nilfs->ns_sbh[1]) {
- sbp = NULL;
- if (dupsb) {
+ if (flag == NILFS_SB_COMMIT_ALL) {
set_buffer_dirty(nilfs->ns_sbh[1]);
- if (!sync_dirty_buffer(nilfs->ns_sbh[1]))
- sbp = nilfs->ns_sbp[1];
+ if (sync_dirty_buffer(nilfs->ns_sbh[1]) < 0)
+ goto out;
}
+ if (le64_to_cpu(nilfs->ns_sbp[1]->s_last_cno) <
+ le64_to_cpu(nilfs->ns_sbp[0]->s_last_cno))
+ sbp = nilfs->ns_sbp[1];
}
- if (sbp) {
- spin_lock(&nilfs->ns_last_segment_lock);
- nilfs->ns_prot_seq = le64_to_cpu(sbp->s_last_seq);
- spin_unlock(&nilfs->ns_last_segment_lock);
- }
- }
+ spin_lock(&nilfs->ns_last_segment_lock);
+ nilfs->ns_prot_seq = le64_to_cpu(sbp->s_last_seq);
+ spin_unlock(&nilfs->ns_last_segment_lock);
+ }
+ out:
return err;
}
-int nilfs_commit_super(struct nilfs_sb_info *sbi, int dupsb)
+void nilfs_set_log_cursor(struct nilfs_super_block *sbp,
+ struct the_nilfs *nilfs)
+{
+ sector_t nfreeblocks;
+
+ /* nilfs->ns_sem must be locked by the caller. */
+ nilfs_count_free_blocks(nilfs, &nfreeblocks);
+ sbp->s_free_blocks_count = cpu_to_le64(nfreeblocks);
+
+ spin_lock(&nilfs->ns_last_segment_lock);
+ sbp->s_last_seq = cpu_to_le64(nilfs->ns_last_seq);
+ sbp->s_last_pseg = cpu_to_le64(nilfs->ns_last_pseg);
+ sbp->s_last_cno = cpu_to_le64(nilfs->ns_last_cno);
+ spin_unlock(&nilfs->ns_last_segment_lock);
+}
+
+struct nilfs_super_block **nilfs_prepare_super(struct nilfs_sb_info *sbi,
+ int flip)
{
struct the_nilfs *nilfs = sbi->s_nilfs;
struct nilfs_super_block **sbp = nilfs->ns_sbp;
- sector_t nfreeblocks;
- time_t t;
- int err;
- /* nilfs->sem must be locked by the caller. */
+ /* nilfs->ns_sem must be locked by the caller. */
if (sbp[0]->s_magic != cpu_to_le16(NILFS_SUPER_MAGIC)) {
- if (sbp[1] && sbp[1]->s_magic == cpu_to_le16(NILFS_SUPER_MAGIC))
- nilfs_swap_super_block(nilfs);
- else {
+ if (sbp[1] &&
+ sbp[1]->s_magic == cpu_to_le16(NILFS_SUPER_MAGIC)) {
+ memcpy(sbp[0], sbp[1], nilfs->ns_sbsize);
+ } else {
printk(KERN_CRIT "NILFS: superblock broke on dev %s\n",
sbi->s_super->s_id);
- return -EIO;
+ return NULL;
}
+ } else if (sbp[1] &&
+ sbp[1]->s_magic != cpu_to_le16(NILFS_SUPER_MAGIC)) {
+ memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
}
- err = nilfs_count_free_blocks(nilfs, &nfreeblocks);
- if (unlikely(err)) {
- printk(KERN_ERR "NILFS: failed to count free blocks\n");
- return err;
- }
- spin_lock(&nilfs->ns_last_segment_lock);
- sbp[0]->s_last_seq = cpu_to_le64(nilfs->ns_last_seq);
- sbp[0]->s_last_pseg = cpu_to_le64(nilfs->ns_last_pseg);
- sbp[0]->s_last_cno = cpu_to_le64(nilfs->ns_last_cno);
- spin_unlock(&nilfs->ns_last_segment_lock);
+ if (flip && sbp[1])
+ nilfs_swap_super_block(nilfs);
+
+ return sbp;
+}
+
+int nilfs_commit_super(struct nilfs_sb_info *sbi, int flag)
+{
+ struct the_nilfs *nilfs = sbi->s_nilfs;
+ struct nilfs_super_block **sbp = nilfs->ns_sbp;
+ time_t t;
+
+ /* nilfs->ns_sem must be locked by the caller. */
t = get_seconds();
- nilfs->ns_sbwtime[0] = t;
- sbp[0]->s_free_blocks_count = cpu_to_le64(nfreeblocks);
+ nilfs->ns_sbwtime = t;
sbp[0]->s_wtime = cpu_to_le64(t);
sbp[0]->s_sum = 0;
sbp[0]->s_sum = cpu_to_le32(crc32_le(nilfs->ns_crc_seed,
(unsigned char *)sbp[0],
nilfs->ns_sbsize));
- if (dupsb && sbp[1]) {
- memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
- nilfs->ns_sbwtime[1] = t;
+ if (flag == NILFS_SB_COMMIT_ALL && sbp[1]) {
+ sbp[1]->s_wtime = sbp[0]->s_wtime;
+ sbp[1]->s_sum = 0;
+ sbp[1]->s_sum = cpu_to_le32(crc32_le(nilfs->ns_crc_seed,
+ (unsigned char *)sbp[1],
+ nilfs->ns_sbsize));
}
clear_nilfs_sb_dirty(nilfs);
- return nilfs_sync_super(sbi, dupsb);
+ return nilfs_sync_super(sbi, flag);
+}
+
+/**
+ * nilfs_cleanup_super() - write filesystem state for cleanup
+ * @sbi: nilfs_sb_info to be unmounted or degraded to read-only
+ *
+ * This function restores state flags in the on-disk super block.
+ * This will set "clean" flag (i.e. NILFS_VALID_FS) unless the
+ * filesystem was not clean previously.
+ */
+int nilfs_cleanup_super(struct nilfs_sb_info *sbi)
+{
+ struct nilfs_super_block **sbp;
+ int flag = NILFS_SB_COMMIT;
+ int ret = -EIO;
+
+ sbp = nilfs_prepare_super(sbi, 0);
+ if (sbp) {
+ sbp[0]->s_state = cpu_to_le16(sbi->s_nilfs->ns_mount_state);
+ nilfs_set_log_cursor(sbp[0], sbi->s_nilfs);
+ if (sbp[1] && sbp[0]->s_last_cno == sbp[1]->s_last_cno) {
+ /*
+ * make the "clean" flag also to the opposite
+ * super block if both super blocks point to
+ * the same checkpoint.
+ */
+ sbp[1]->s_state = sbp[0]->s_state;
+ flag = NILFS_SB_COMMIT_ALL;
+ }
+ ret = nilfs_commit_super(sbi, flag);
+ }
+ return ret;
}
static void nilfs_put_super(struct super_block *sb)
@@ -289,8 +365,7 @@ static void nilfs_put_super(struct super_block *sb)
if (!(sb->s_flags & MS_RDONLY)) {
down_write(&nilfs->ns_sem);
- nilfs->ns_sbp[0]->s_state = cpu_to_le16(nilfs->ns_mount_state);
- nilfs_commit_super(sbi, 1);
+ nilfs_cleanup_super(sbi);
up_write(&nilfs->ns_sem);
}
down_write(&nilfs->ns_super_sem);
@@ -311,6 +386,7 @@ static int nilfs_sync_fs(struct super_block *sb, int wait)
{
struct nilfs_sb_info *sbi = NILFS_SB(sb);
struct the_nilfs *nilfs = sbi->s_nilfs;
+ struct nilfs_super_block **sbp;
int err = 0;
/* This function is called when super block should be written back */
@@ -318,8 +394,13 @@ static int nilfs_sync_fs(struct super_block *sb, int wait)
err = nilfs_construct_segment(sb);
down_write(&nilfs->ns_sem);
- if (nilfs_sb_dirty(nilfs))
- nilfs_commit_super(sbi, 1);
+ if (nilfs_sb_dirty(nilfs)) {
+ sbp = nilfs_prepare_super(sbi, nilfs_sb_will_flip(nilfs));
+ if (likely(sbp)) {
+ nilfs_set_log_cursor(sbp[0], nilfs);
+ nilfs_commit_super(sbi, NILFS_SB_COMMIT);
+ }
+ }
up_write(&nilfs->ns_sem);
return err;
@@ -442,20 +523,20 @@ static int nilfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
struct nilfs_sb_info *sbi = NILFS_SB(sb);
if (!nilfs_test_opt(sbi, BARRIER))
- seq_printf(seq, ",nobarrier");
+ seq_puts(seq, ",nobarrier");
if (nilfs_test_opt(sbi, SNAPSHOT))
seq_printf(seq, ",cp=%llu",
(unsigned long long int)sbi->s_snapshot_cno);
if (nilfs_test_opt(sbi, ERRORS_PANIC))
- seq_printf(seq, ",errors=panic");
+ seq_puts(seq, ",errors=panic");
if (nilfs_test_opt(sbi, ERRORS_CONT))
- seq_printf(seq, ",errors=continue");
+ seq_puts(seq, ",errors=continue");
if (nilfs_test_opt(sbi, STRICT_ORDER))
- seq_printf(seq, ",order=strict");
+ seq_puts(seq, ",order=strict");
if (nilfs_test_opt(sbi, NORECOVERY))
- seq_printf(seq, ",norecovery");
+ seq_puts(seq, ",norecovery");
if (nilfs_test_opt(sbi, DISCARD))
- seq_printf(seq, ",discard");
+ seq_puts(seq, ",discard");
return 0;
}
@@ -524,23 +605,25 @@ static const struct export_operations nilfs_export_ops = {
enum {
Opt_err_cont, Opt_err_panic, Opt_err_ro,
- Opt_nobarrier, Opt_snapshot, Opt_order, Opt_norecovery,
- Opt_discard, Opt_err,
+ Opt_barrier, Opt_nobarrier, Opt_snapshot, Opt_order, Opt_norecovery,
+ Opt_discard, Opt_nodiscard, Opt_err,
};
static match_table_t tokens = {
{Opt_err_cont, "errors=continue"},
{Opt_err_panic, "errors=panic"},
{Opt_err_ro, "errors=remount-ro"},
+ {Opt_barrier, "barrier"},
{Opt_nobarrier, "nobarrier"},
{Opt_snapshot, "cp=%u"},
{Opt_order, "order=%s"},
{Opt_norecovery, "norecovery"},
{Opt_discard, "discard"},
+ {Opt_nodiscard, "nodiscard"},
{Opt_err, NULL}
};
-static int parse_options(char *options, struct super_block *sb)
+static int parse_options(char *options, struct super_block *sb, int is_remount)
{
struct nilfs_sb_info *sbi = NILFS_SB(sb);
char *p;
@@ -557,6 +640,9 @@ static int parse_options(char *options, struct super_block *sb)
token = match_token(p, tokens, args);
switch (token) {
+ case Opt_barrier:
+ nilfs_set_opt(sbi, BARRIER);
+ break;
case Opt_nobarrier:
nilfs_clear_opt(sbi, BARRIER);
break;
@@ -582,8 +668,26 @@ static int parse_options(char *options, struct super_block *sb)
case Opt_snapshot:
if (match_int(&args[0], &option) || option <= 0)
return 0;
- if (!(sb->s_flags & MS_RDONLY))
+ if (is_remount) {
+ if (!nilfs_test_opt(sbi, SNAPSHOT)) {
+ printk(KERN_ERR
+ "NILFS: cannot change regular "
+ "mount to snapshot.\n");
+ return 0;
+ } else if (option != sbi->s_snapshot_cno) {
+ printk(KERN_ERR
+ "NILFS: cannot remount to a "
+ "different snapshot.\n");
+ return 0;
+ }
+ break;
+ }
+ if (!(sb->s_flags & MS_RDONLY)) {
+ printk(KERN_ERR "NILFS: cannot mount snapshot "
+ "read/write. A read-only option is "
+ "required.\n");
return 0;
+ }
sbi->s_snapshot_cno = option;
nilfs_set_opt(sbi, SNAPSHOT);
break;
@@ -593,6 +697,9 @@ static int parse_options(char *options, struct super_block *sb)
case Opt_discard:
nilfs_set_opt(sbi, DISCARD);
break;
+ case Opt_nodiscard:
+ nilfs_clear_opt(sbi, DISCARD);
+ break;
default:
printk(KERN_ERR
"NILFS: Unrecognized mount option \"%s\"\n", p);
@@ -613,11 +720,18 @@ nilfs_set_default_options(struct nilfs_sb_info *sbi,
static int nilfs_setup_super(struct nilfs_sb_info *sbi)
{
struct the_nilfs *nilfs = sbi->s_nilfs;
- struct nilfs_super_block *sbp = nilfs->ns_sbp[0];
- int max_mnt_count = le16_to_cpu(sbp->s_max_mnt_count);
- int mnt_count = le16_to_cpu(sbp->s_mnt_count);
+ struct nilfs_super_block **sbp;
+ int max_mnt_count;
+ int mnt_count;
+
+ /* nilfs->ns_sem must be locked by the caller. */
+ sbp = nilfs_prepare_super(sbi, 0);
+ if (!sbp)
+ return -EIO;
+
+ max_mnt_count = le16_to_cpu(sbp[0]->s_max_mnt_count);
+ mnt_count = le16_to_cpu(sbp[0]->s_mnt_count);
- /* nilfs->sem must be locked by the caller. */
if (nilfs->ns_mount_state & NILFS_ERROR_FS) {
printk(KERN_WARNING
"NILFS warning: mounting fs with errors\n");
@@ -628,12 +742,15 @@ static int nilfs_setup_super(struct nilfs_sb_info *sbi)
#endif
}
if (!max_mnt_count)
- sbp->s_max_mnt_count = cpu_to_le16(NILFS_DFL_MAX_MNT_COUNT);
-
- sbp->s_mnt_count = cpu_to_le16(mnt_count + 1);
- sbp->s_state = cpu_to_le16(le16_to_cpu(sbp->s_state) & ~NILFS_VALID_FS);
- sbp->s_mtime = cpu_to_le64(get_seconds());
- return nilfs_commit_super(sbi, 1);
+ sbp[0]->s_max_mnt_count = cpu_to_le16(NILFS_DFL_MAX_MNT_COUNT);
+
+ sbp[0]->s_mnt_count = cpu_to_le16(mnt_count + 1);
+ sbp[0]->s_state =
+ cpu_to_le16(le16_to_cpu(sbp[0]->s_state) & ~NILFS_VALID_FS);
+ sbp[0]->s_mtime = cpu_to_le64(get_seconds());
+ /* synchronize sbp[1] with sbp[0] */
+ memcpy(sbp[1], sbp[0], nilfs->ns_sbsize);
+ return nilfs_commit_super(sbi, NILFS_SB_COMMIT_ALL);
}
struct nilfs_super_block *nilfs_read_super_block(struct super_block *sb,
@@ -670,7 +787,31 @@ int nilfs_store_magic_and_option(struct super_block *sb,
sbi->s_interval = le32_to_cpu(sbp->s_c_interval);
sbi->s_watermark = le32_to_cpu(sbp->s_c_block_max);
- return !parse_options(data, sb) ? -EINVAL : 0 ;
+ return !parse_options(data, sb, 0) ? -EINVAL : 0 ;
+}
+
+int nilfs_check_feature_compatibility(struct super_block *sb,
+ struct nilfs_super_block *sbp)
+{
+ __u64 features;
+
+ features = le64_to_cpu(sbp->s_feature_incompat) &
+ ~NILFS_FEATURE_INCOMPAT_SUPP;
+ if (features) {
+ printk(KERN_ERR "NILFS: couldn't mount because of unsupported "
+ "optional features (%llx)\n",
+ (unsigned long long)features);
+ return -EINVAL;
+ }
+ features = le64_to_cpu(sbp->s_feature_compat_ro) &
+ ~NILFS_FEATURE_COMPAT_RO_SUPP;
+ if (!(sb->s_flags & MS_RDONLY) && features) {
+ printk(KERN_ERR "NILFS: couldn't mount RDWR because of "
+ "unsupported optional features (%llx)\n",
+ (unsigned long long)features);
+ return -EINVAL;
+ }
+ return 0;
}
/**
@@ -819,7 +960,6 @@ nilfs_fill_super(struct super_block *sb, void *data, int silent,
static int nilfs_remount(struct super_block *sb, int *flags, char *data)
{
struct nilfs_sb_info *sbi = NILFS_SB(sb);
- struct nilfs_super_block *sbp;
struct the_nilfs *nilfs = sbi->s_nilfs;
unsigned long old_sb_flags;
struct nilfs_mount_options old_opts;
@@ -833,32 +973,17 @@ static int nilfs_remount(struct super_block *sb, int *flags, char *data)
old_opts.snapshot_cno = sbi->s_snapshot_cno;
was_snapshot = nilfs_test_opt(sbi, SNAPSHOT);
- if (!parse_options(data, sb)) {
+ if (!parse_options(data, sb, 1)) {
err = -EINVAL;
goto restore_opts;
}
sb->s_flags = (sb->s_flags & ~MS_POSIXACL);
err = -EINVAL;
- if (was_snapshot) {
- if (!(*flags & MS_RDONLY)) {
- printk(KERN_ERR "NILFS (device %s): cannot remount "
- "snapshot read/write.\n",
- sb->s_id);
- goto restore_opts;
- } else if (sbi->s_snapshot_cno != old_opts.snapshot_cno) {
- printk(KERN_ERR "NILFS (device %s): cannot "
- "remount to a different snapshot.\n",
- sb->s_id);
- goto restore_opts;
- }
- } else {
- if (nilfs_test_opt(sbi, SNAPSHOT)) {
- printk(KERN_ERR "NILFS (device %s): cannot change "
- "a regular mount to a snapshot.\n",
- sb->s_id);
- goto restore_opts;
- }
+ if (was_snapshot && !(*flags & MS_RDONLY)) {
+ printk(KERN_ERR "NILFS (device %s): cannot remount snapshot "
+ "read/write.\n", sb->s_id);
+ goto restore_opts;
}
if (!nilfs_valid_fs(nilfs)) {
@@ -880,19 +1005,29 @@ static int nilfs_remount(struct super_block *sb, int *flags, char *data)
* the RDONLY flag and then mark the partition as valid again.
*/
down_write(&nilfs->ns_sem);
- sbp = nilfs->ns_sbp[0];
- if (!(sbp->s_state & le16_to_cpu(NILFS_VALID_FS)) &&
- (nilfs->ns_mount_state & NILFS_VALID_FS))
- sbp->s_state = cpu_to_le16(nilfs->ns_mount_state);
- sbp->s_mtime = cpu_to_le64(get_seconds());
- nilfs_commit_super(sbi, 1);
+ nilfs_cleanup_super(sbi);
up_write(&nilfs->ns_sem);
} else {
+ __u64 features;
+
/*
* Mounting a RDONLY partition read-write, so reread and
* store the current valid flag. (It may have been changed
* by fsck since we originally mounted the partition.)
*/
+ down_read(&nilfs->ns_sem);
+ features = le64_to_cpu(nilfs->ns_sbp[0]->s_feature_compat_ro) &
+ ~NILFS_FEATURE_COMPAT_RO_SUPP;
+ up_read(&nilfs->ns_sem);
+ if (features) {
+ printk(KERN_WARNING "NILFS (device %s): couldn't "
+ "remount RDWR because of unsupported optional "
+ "features (%llx)\n",
+ sb->s_id, (unsigned long long)features);
+ err = -EROFS;
+ goto restore_opts;
+ }
+
sb->s_flags &= ~MS_RDONLY;
err = nilfs_attach_segment_constructor(sbi);
@@ -1119,7 +1254,7 @@ static void nilfs_inode_init_once(void *obj)
init_rwsem(&ii->xattr_sem);
#endif
nilfs_btnode_cache_init_once(&ii->i_btnode_cache);
- ii->i_bmap = (struct nilfs_bmap *)&ii->i_bmap_union;
+ ii->i_bmap = &ii->i_bmap_data;
inode_init_once(&ii->vfs_inode);
}
diff --git a/fs/nilfs2/the_nilfs.c b/fs/nilfs2/the_nilfs.c
index 8c1097327ab..37de1f062d8 100644
--- a/fs/nilfs2/the_nilfs.c
+++ b/fs/nilfs2/the_nilfs.c
@@ -38,6 +38,8 @@
static LIST_HEAD(nilfs_objects);
static DEFINE_SPINLOCK(nilfs_lock);
+static int nilfs_valid_sb(struct nilfs_super_block *sbp);
+
void nilfs_set_last_segment(struct the_nilfs *nilfs,
sector_t start_blocknr, u64 seq, __u64 cno)
{
@@ -45,6 +47,16 @@ void nilfs_set_last_segment(struct the_nilfs *nilfs,
nilfs->ns_last_pseg = start_blocknr;
nilfs->ns_last_seq = seq;
nilfs->ns_last_cno = cno;
+
+ if (!nilfs_sb_dirty(nilfs)) {
+ if (nilfs->ns_prev_seq == nilfs->ns_last_seq)
+ goto stay_cursor;
+
+ set_nilfs_sb_dirty(nilfs);
+ }
+ nilfs->ns_prev_seq = nilfs->ns_last_seq;
+
+ stay_cursor:
spin_unlock(&nilfs->ns_last_segment_lock);
}
@@ -159,8 +171,7 @@ void put_nilfs(struct the_nilfs *nilfs)
kfree(nilfs);
}
-static int nilfs_load_super_root(struct the_nilfs *nilfs,
- struct nilfs_sb_info *sbi, sector_t sr_block)
+static int nilfs_load_super_root(struct the_nilfs *nilfs, sector_t sr_block)
{
struct buffer_head *bh_sr;
struct nilfs_super_root *raw_sr;
@@ -169,7 +180,7 @@ static int nilfs_load_super_root(struct the_nilfs *nilfs,
unsigned inode_size;
int err;
- err = nilfs_read_super_root_block(sbi->s_super, sr_block, &bh_sr, 1);
+ err = nilfs_read_super_root_block(nilfs, sr_block, &bh_sr, 1);
if (unlikely(err))
return err;
@@ -248,6 +259,37 @@ static void nilfs_clear_recovery_info(struct nilfs_recovery_info *ri)
}
/**
+ * nilfs_store_log_cursor - load log cursor from a super block
+ * @nilfs: nilfs object
+ * @sbp: buffer storing super block to be read
+ *
+ * nilfs_store_log_cursor() reads the last position of the log
+ * containing a super root from a given super block, and initializes
+ * relevant information on the nilfs object preparatory for log
+ * scanning and recovery.
+ */
+static int nilfs_store_log_cursor(struct the_nilfs *nilfs,
+ struct nilfs_super_block *sbp)
+{
+ int ret = 0;
+
+ nilfs->ns_last_pseg = le64_to_cpu(sbp->s_last_pseg);
+ nilfs->ns_last_cno = le64_to_cpu(sbp->s_last_cno);
+ nilfs->ns_last_seq = le64_to_cpu(sbp->s_last_seq);
+
+ nilfs->ns_prev_seq = nilfs->ns_last_seq;
+ nilfs->ns_seg_seq = nilfs->ns_last_seq;
+ nilfs->ns_segnum =
+ nilfs_get_segnum_of_block(nilfs, nilfs->ns_last_pseg);
+ nilfs->ns_cno = nilfs->ns_last_cno + 1;
+ if (nilfs->ns_segnum >= nilfs->ns_nsegments) {
+ printk(KERN_ERR "NILFS invalid last segment number.\n");
+ ret = -EINVAL;
+ }
+ return ret;
+}
+
+/**
* load_nilfs - load and recover the nilfs
* @nilfs: the_nilfs structure to be released
* @sbi: nilfs_sb_info used to recover past segment
@@ -285,13 +327,55 @@ int load_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi)
nilfs_init_recovery_info(&ri);
- err = nilfs_search_super_root(nilfs, sbi, &ri);
+ err = nilfs_search_super_root(nilfs, &ri);
if (unlikely(err)) {
- printk(KERN_ERR "NILFS: error searching super root.\n");
- goto failed;
+ struct nilfs_super_block **sbp = nilfs->ns_sbp;
+ int blocksize;
+
+ if (err != -EINVAL)
+ goto scan_error;
+
+ if (!nilfs_valid_sb(sbp[1])) {
+ printk(KERN_WARNING
+ "NILFS warning: unable to fall back to spare"
+ "super block\n");
+ goto scan_error;
+ }
+ printk(KERN_INFO
+ "NILFS: try rollback from an earlier position\n");
+
+ /*
+ * restore super block with its spare and reconfigure
+ * relevant states of the nilfs object.
+ */
+ memcpy(sbp[0], sbp[1], nilfs->ns_sbsize);
+ nilfs->ns_crc_seed = le32_to_cpu(sbp[0]->s_crc_seed);
+ nilfs->ns_sbwtime = le64_to_cpu(sbp[0]->s_wtime);
+
+ /* verify consistency between two super blocks */
+ blocksize = BLOCK_SIZE << le32_to_cpu(sbp[0]->s_log_block_size);
+ if (blocksize != nilfs->ns_blocksize) {
+ printk(KERN_WARNING
+ "NILFS warning: blocksize differs between "
+ "two super blocks (%d != %d)\n",
+ blocksize, nilfs->ns_blocksize);
+ goto scan_error;
+ }
+
+ err = nilfs_store_log_cursor(nilfs, sbp[0]);
+ if (err)
+ goto scan_error;
+
+ /* drop clean flag to allow roll-forward and recovery */
+ nilfs->ns_mount_state &= ~NILFS_VALID_FS;
+ valid_fs = 0;
+
+ err = nilfs_search_super_root(nilfs, &ri);
+ if (err)
+ goto scan_error;
}
- err = nilfs_load_super_root(nilfs, sbi, ri.ri_super_root);
+ err = nilfs_load_super_root(nilfs, ri.ri_super_root);
if (unlikely(err)) {
printk(KERN_ERR "NILFS: error loading super root.\n");
goto failed;
@@ -301,11 +385,23 @@ int load_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi)
goto skip_recovery;
if (s_flags & MS_RDONLY) {
+ __u64 features;
+
if (nilfs_test_opt(sbi, NORECOVERY)) {
printk(KERN_INFO "NILFS: norecovery option specified. "
"skipping roll-forward recovery\n");
goto skip_recovery;
}
+ features = le64_to_cpu(nilfs->ns_sbp[0]->s_feature_compat_ro) &
+ ~NILFS_FEATURE_COMPAT_RO_SUPP;
+ if (features) {
+ printk(KERN_ERR "NILFS: couldn't proceed with "
+ "recovery because of unsupported optional "
+ "features (%llx)\n",
+ (unsigned long long)features);
+ err = -EROFS;
+ goto failed_unload;
+ }
if (really_read_only) {
printk(KERN_ERR "NILFS: write access "
"unavailable, cannot proceed.\n");
@@ -320,14 +416,13 @@ int load_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi)
goto failed_unload;
}
- err = nilfs_recover_logical_segments(nilfs, sbi, &ri);
+ err = nilfs_salvage_orphan_logs(nilfs, sbi, &ri);
if (err)
goto failed_unload;
down_write(&nilfs->ns_sem);
- nilfs->ns_mount_state |= NILFS_VALID_FS;
- nilfs->ns_sbp[0]->s_state = cpu_to_le16(nilfs->ns_mount_state);
- err = nilfs_commit_super(sbi, 1);
+ nilfs->ns_mount_state |= NILFS_VALID_FS; /* set "clean" flag */
+ err = nilfs_cleanup_super(sbi);
up_write(&nilfs->ns_sem);
if (err) {
@@ -343,6 +438,10 @@ int load_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi)
sbi->s_super->s_flags = s_flags;
return 0;
+ scan_error:
+ printk(KERN_ERR "NILFS: error searching super root.\n");
+ goto failed;
+
failed_unload:
nilfs_mdt_destroy(nilfs->ns_cpfile);
nilfs_mdt_destroy(nilfs->ns_sufile);
@@ -515,8 +614,8 @@ static int nilfs_load_super_block(struct the_nilfs *nilfs,
nilfs_swap_super_block(nilfs);
}
- nilfs->ns_sbwtime[0] = le64_to_cpu(sbp[0]->s_wtime);
- nilfs->ns_sbwtime[1] = valid[!swp] ? le64_to_cpu(sbp[1]->s_wtime) : 0;
+ nilfs->ns_sbwcount = 0;
+ nilfs->ns_sbwtime = le64_to_cpu(sbp[0]->s_wtime);
nilfs->ns_prot_seq = le64_to_cpu(sbp[valid[1] & !swp]->s_last_seq);
*sbpp = sbp[0];
return 0;
@@ -557,6 +656,10 @@ int init_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi, char *data)
if (err)
goto out;
+ err = nilfs_check_feature_compatibility(sb, sbp);
+ if (err)
+ goto out;
+
blocksize = BLOCK_SIZE << le32_to_cpu(sbp->s_log_block_size);
if (sb->s_blocksize != blocksize &&
!sb_set_blocksize(sb, blocksize)) {
@@ -568,7 +671,7 @@ int init_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi, char *data)
goto out;
}
- blocksize = sb_min_blocksize(sb, BLOCK_SIZE);
+ blocksize = sb_min_blocksize(sb, NILFS_MIN_BLOCK_SIZE);
if (!blocksize) {
printk(KERN_ERR "NILFS: unable to set blocksize\n");
err = -EINVAL;
@@ -582,7 +685,18 @@ int init_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi, char *data)
if (err)
goto failed_sbh;
+ err = nilfs_check_feature_compatibility(sb, sbp);
+ if (err)
+ goto failed_sbh;
+
blocksize = BLOCK_SIZE << le32_to_cpu(sbp->s_log_block_size);
+ if (blocksize < NILFS_MIN_BLOCK_SIZE ||
+ blocksize > NILFS_MAX_BLOCK_SIZE) {
+ printk(KERN_ERR "NILFS: couldn't mount because of unsupported "
+ "filesystem blocksize %d\n", blocksize);
+ err = -EINVAL;
+ goto failed_sbh;
+ }
if (sb->s_blocksize != blocksize) {
int hw_blocksize = bdev_logical_block_size(sb->s_bdev);
@@ -604,6 +718,7 @@ int init_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi, char *data)
when reloading fails. */
}
nilfs->ns_blocksize_bits = sb->s_blocksize_bits;
+ nilfs->ns_blocksize = blocksize;
err = nilfs_store_disk_layout(nilfs, sbp);
if (err)
@@ -616,23 +731,9 @@ int init_nilfs(struct the_nilfs *nilfs, struct nilfs_sb_info *sbi, char *data)
bdi = nilfs->ns_bdev->bd_inode->i_mapping->backing_dev_info;
nilfs->ns_bdi = bdi ? : &default_backing_dev_info;
- /* Finding last segment */
- nilfs->ns_last_pseg = le64_to_cpu(sbp->s_last_pseg);
- nilfs->ns_last_cno = le64_to_cpu(sbp->s_last_cno);
- nilfs->ns_last_seq = le64_to_cpu(sbp->s_last_seq);
-
- nilfs->ns_seg_seq = nilfs->ns_last_seq;
- nilfs->ns_segnum =
- nilfs_get_segnum_of_block(nilfs, nilfs->ns_last_pseg);
- nilfs->ns_cno = nilfs->ns_last_cno + 1;
- if (nilfs->ns_segnum >= nilfs->ns_nsegments) {
- printk(KERN_ERR "NILFS invalid last segment number.\n");
- err = -EINVAL;
+ err = nilfs_store_log_cursor(nilfs, sbp);
+ if (err)
goto failed_sbh;
- }
- /* Dummy values */
- nilfs->ns_free_segments_count =
- nilfs->ns_nsegments - (nilfs->ns_segnum + 1);
/* Initialize gcinode cache */
err = nilfs_init_gccache(nilfs);
diff --git a/fs/nilfs2/the_nilfs.h b/fs/nilfs2/the_nilfs.h
index 1ab97453369..f785a7b0ab9 100644
--- a/fs/nilfs2/the_nilfs.h
+++ b/fs/nilfs2/the_nilfs.h
@@ -57,7 +57,8 @@ enum {
* @ns_current: back pointer to current mount
* @ns_sbh: buffer heads of on-disk super blocks
* @ns_sbp: pointers to super block data
- * @ns_sbwtime: previous write time of super blocks
+ * @ns_sbwtime: previous write time of super block
+ * @ns_sbwcount: write count of super block
* @ns_sbsize: size of valid data in super block
* @ns_supers: list of nilfs super block structs
* @ns_seg_seq: segment sequence counter
@@ -73,7 +74,7 @@ enum {
* @ns_last_seq: sequence value of the latest segment
* @ns_last_cno: checkpoint number of the latest segment
* @ns_prot_seq: least sequence number of segments which must not be reclaimed
- * @ns_free_segments_count: counter of free segments
+ * @ns_prev_seq: base sequence number used to decide if advance log cursor
* @ns_segctor_sem: segment constructor semaphore
* @ns_dat: DAT file inode
* @ns_cpfile: checkpoint file inode
@@ -82,6 +83,7 @@ enum {
* @ns_gc_inodes: dummy inodes to keep live blocks
* @ns_gc_inodes_h: hash list to keep dummy inode holding live blocks
* @ns_blocksize_bits: bit length of block size
+ * @ns_blocksize: block size
* @ns_nsegments: number of segments in filesystem
* @ns_blocks_per_segment: number of blocks per segment
* @ns_r_segments_percentage: reserved segments percentage
@@ -119,7 +121,8 @@ struct the_nilfs {
*/
struct buffer_head *ns_sbh[2];
struct nilfs_super_block *ns_sbp[2];
- time_t ns_sbwtime[2];
+ time_t ns_sbwtime;
+ unsigned ns_sbwcount;
unsigned ns_sbsize;
unsigned ns_mount_state;
@@ -149,7 +152,7 @@ struct the_nilfs {
u64 ns_last_seq;
__u64 ns_last_cno;
u64 ns_prot_seq;
- unsigned long ns_free_segments_count;
+ u64 ns_prev_seq;
struct rw_semaphore ns_segctor_sem;
@@ -168,6 +171,7 @@ struct the_nilfs {
/* Disk layout information (static) */
unsigned int ns_blocksize_bits;
+ unsigned int ns_blocksize;
unsigned long ns_nsegments;
unsigned long ns_blocks_per_segment;
unsigned long ns_r_segments_percentage;
@@ -203,20 +207,17 @@ THE_NILFS_FNS(SB_DIRTY, sb_dirty)
/* Minimum interval of periodical update of superblocks (in seconds) */
#define NILFS_SB_FREQ 10
-#define NILFS_ALTSB_FREQ 60 /* spare superblock */
static inline int nilfs_sb_need_update(struct the_nilfs *nilfs)
{
u64 t = get_seconds();
- return t < nilfs->ns_sbwtime[0] ||
- t > nilfs->ns_sbwtime[0] + NILFS_SB_FREQ;
+ return t < nilfs->ns_sbwtime || t > nilfs->ns_sbwtime + NILFS_SB_FREQ;
}
-static inline int nilfs_altsb_need_update(struct the_nilfs *nilfs)
+static inline int nilfs_sb_will_flip(struct the_nilfs *nilfs)
{
- u64 t = get_seconds();
- struct nilfs_super_block **sbp = nilfs->ns_sbp;
- return sbp[1] && t > nilfs->ns_sbwtime[1] + NILFS_ALTSB_FREQ;
+ int flip_bits = nilfs->ns_sbwcount & 0x0FL;
+ return (flip_bits != 0x08 && flip_bits != 0x0F);
}
void nilfs_set_last_segment(struct the_nilfs *, sector_t, u64, __u64);
diff --git a/include/linux/nilfs2_fs.h b/include/linux/nilfs2_fs.h
index 8c2c6116e78..f5487b6f91e 100644
--- a/include/linux/nilfs2_fs.h
+++ b/include/linux/nilfs2_fs.h
@@ -160,7 +160,7 @@ struct nilfs_super_root {
* struct nilfs_super_block - structure of super block on disk
*/
struct nilfs_super_block {
- __le32 s_rev_level; /* Revision level */
+/*00*/ __le32 s_rev_level; /* Revision level */
__le16 s_minor_rev_level; /* minor revision level */
__le16 s_magic; /* Magic signature */
@@ -169,50 +169,53 @@ struct nilfs_super_block {
is excluded. */
__le16 s_flags; /* flags */
__le32 s_crc_seed; /* Seed value of CRC calculation */
- __le32 s_sum; /* Check sum of super block */
+/*10*/ __le32 s_sum; /* Check sum of super block */
__le32 s_log_block_size; /* Block size represented as follows
blocksize =
1 << (s_log_block_size + 10) */
__le64 s_nsegments; /* Number of segments in filesystem */
- __le64 s_dev_size; /* block device size in bytes */
+/*20*/ __le64 s_dev_size; /* block device size in bytes */
__le64 s_first_data_block; /* 1st seg disk block number */
- __le32 s_blocks_per_segment; /* number of blocks per full segment */
+/*30*/ __le32 s_blocks_per_segment; /* number of blocks per full segment */
__le32 s_r_segments_percentage; /* Reserved segments percentage */
__le64 s_last_cno; /* Last checkpoint number */
- __le64 s_last_pseg; /* disk block addr pseg written last */
+/*40*/ __le64 s_last_pseg; /* disk block addr pseg written last */
__le64 s_last_seq; /* seq. number of seg written last */
- __le64 s_free_blocks_count; /* Free blocks count */
+/*50*/ __le64 s_free_blocks_count; /* Free blocks count */
__le64 s_ctime; /* Creation time (execution time of
newfs) */
- __le64 s_mtime; /* Mount time */
+/*60*/ __le64 s_mtime; /* Mount time */
__le64 s_wtime; /* Write time */
- __le16 s_mnt_count; /* Mount count */
+/*70*/ __le16 s_mnt_count; /* Mount count */
__le16 s_max_mnt_count; /* Maximal mount count */
__le16 s_state; /* File system state */
__le16 s_errors; /* Behaviour when detecting errors */
__le64 s_lastcheck; /* time of last check */
- __le32 s_checkinterval; /* max. time between checks */
+/*80*/ __le32 s_checkinterval; /* max. time between checks */
__le32 s_creator_os; /* OS */
__le16 s_def_resuid; /* Default uid for reserved blocks */
__le16 s_def_resgid; /* Default gid for reserved blocks */
__le32 s_first_ino; /* First non-reserved inode */
- __le16 s_inode_size; /* Size of an inode */
+/*90*/ __le16 s_inode_size; /* Size of an inode */
__le16 s_dat_entry_size; /* Size of a dat entry */
__le16 s_checkpoint_size; /* Size of a checkpoint */
__le16 s_segment_usage_size; /* Size of a segment usage */
- __u8 s_uuid[16]; /* 128-bit uuid for volume */
- char s_volume_name[80]; /* volume name */
+/*98*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
+/*A8*/ char s_volume_name[80]; /* volume name */
- __le32 s_c_interval; /* Commit interval of segment */
+/*F8*/ __le32 s_c_interval; /* Commit interval of segment */
__le32 s_c_block_max; /* Threshold of data amount for
the segment construction */
- __u32 s_reserved[192]; /* padding to the end of the block */
+/*100*/ __le64 s_feature_compat; /* Compatible feature set */
+ __le64 s_feature_compat_ro; /* Read-only compatible feature set */
+ __le64 s_feature_incompat; /* Incompatible feature set */
+ __u32 s_reserved[186]; /* padding to the end of the block */
};
/*
@@ -228,6 +231,16 @@ struct nilfs_super_block {
#define NILFS_MINOR_REV 0 /* minor revision */
/*
+ * Feature set definitions
+ *
+ * If there is a bit set in the incompatible feature set that the kernel
+ * doesn't know about, it should refuse to mount the filesystem.
+ */
+#define NILFS_FEATURE_COMPAT_SUPP 0ULL
+#define NILFS_FEATURE_COMPAT_RO_SUPP 0ULL
+#define NILFS_FEATURE_INCOMPAT_SUPP 0ULL
+
+/*
* Bytes count of super_block for CRC-calculation
*/
#define NILFS_SB_BYTES \
@@ -274,6 +287,12 @@ struct nilfs_super_block {
#define NILFS_NAME_LEN 255
/*
+ * Block size limitations
+ */
+#define NILFS_MIN_BLOCK_SIZE 1024
+#define NILFS_MAX_BLOCK_SIZE 65536
+
+/*
* The new version of the directory entry. Since V0 structures are
* stored in intel byte order, and the name_len field could never be
* bigger than 255 chars, it's safe to reclaim the extra byte for the
@@ -313,7 +332,25 @@ enum {
#define NILFS_DIR_ROUND (NILFS_DIR_PAD - 1)
#define NILFS_DIR_REC_LEN(name_len) (((name_len) + 12 + NILFS_DIR_ROUND) & \
~NILFS_DIR_ROUND)
+#define NILFS_MAX_REC_LEN ((1<<16)-1)
+static inline unsigned nilfs_rec_len_from_disk(__le16 dlen)
+{
+ unsigned len = le16_to_cpu(dlen);
+
+ if (len == NILFS_MAX_REC_LEN)
+ return 1 << 16;
+ return len;
+}
+
+static inline __le16 nilfs_rec_len_to_disk(unsigned len)
+{
+ if (len == (1 << 16))
+ return cpu_to_le16(NILFS_MAX_REC_LEN);
+ else if (len > (1 << 16))
+ BUG();
+ return cpu_to_le16(len);
+}
/**
* struct nilfs_finfo - file information