#include <linux/capability.h>
#include <linux/fs.h>
#include <linux/jbd2.h>
-#include <linux/ext4_fs.h>
-#include <linux/ext4_jbd2.h>
#include <linux/quotaops.h>
#include <linux/buffer_head.h>
-
+#include "ext4.h"
+#include "ext4_jbd2.h"
#include "group.h"
+
/*
* balloc.c contains the blocks allocation and deallocation routines
*/
* Calculate the block group number and offset, given a block number
*/
void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
- unsigned long *blockgrpp, ext4_grpblk_t *offsetp)
+ ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp)
{
struct ext4_super_block *es = EXT4_SB(sb)->s_es;
ext4_grpblk_t offset;
/* Initializes an uninitialized block bitmap if given, and returns the
* number of blocks free in the group. */
unsigned ext4_init_block_bitmap(struct super_block *sb, struct buffer_head *bh,
- int block_group, struct ext4_group_desc *gdp)
+ ext4_group_t block_group, struct ext4_group_desc *gdp)
{
- unsigned long start;
int bit, bit_max;
unsigned free_blocks, group_blocks;
struct ext4_sb_info *sbi = EXT4_SB(sb);
/* If checksum is bad mark all blocks used to prevent allocation
* essentially implementing a per-group read-only flag. */
if (!ext4_group_desc_csum_verify(sbi, block_group, gdp)) {
- ext4_error(sb, __FUNCTION__,
- "Checksum bad for group %u\n", block_group);
+ ext4_error(sb, __func__,
+ "Checksum bad for group %lu\n", block_group);
gdp->bg_free_blocks_count = 0;
gdp->bg_free_inodes_count = 0;
gdp->bg_itable_unused = 0;
free_blocks = group_blocks - bit_max;
if (bh) {
+ ext4_fsblk_t start;
+
for (bit = 0; bit < bit_max; bit++)
ext4_set_bit(bit, bh->b_data);
- start = block_group * EXT4_BLOCKS_PER_GROUP(sb) +
- le32_to_cpu(sbi->s_es->s_first_data_block);
+ start = ext4_group_first_block_no(sb, block_group);
/* Set bits for block and inode bitmaps, and inode table */
ext4_set_bit(ext4_block_bitmap(sb, gdp) - start, bh->b_data);
* group descriptor
*/
struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
- unsigned int block_group,
+ ext4_group_t block_group,
struct buffer_head ** bh)
{
unsigned long group_desc;
if (block_group >= sbi->s_groups_count) {
ext4_error (sb, "ext4_get_group_desc",
"block_group >= groups_count - "
- "block_group = %d, groups_count = %lu",
+ "block_group = %lu, groups_count = %lu",
block_group, sbi->s_groups_count);
return NULL;
if (!sbi->s_group_desc[group_desc]) {
ext4_error (sb, "ext4_get_group_desc",
"Group descriptor not loaded - "
- "block_group = %d, group_desc = %lu, desc = %lu",
+ "block_group = %lu, group_desc = %lu, desc = %lu",
block_group, group_desc, offset);
return NULL;
}
return desc;
}
-static inline int
-block_in_use(ext4_fsblk_t block, struct super_block *sb, unsigned char *map)
+static int ext4_valid_block_bitmap(struct super_block *sb,
+ struct ext4_group_desc *desc,
+ unsigned int block_group,
+ struct buffer_head *bh)
{
ext4_grpblk_t offset;
+ ext4_grpblk_t next_zero_bit;
+ ext4_fsblk_t bitmap_blk;
+ ext4_fsblk_t group_first_block;
- ext4_get_group_no_and_offset(sb, block, NULL, &offset);
- return ext4_test_bit (offset, map);
-}
+ if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG)) {
+ /* with FLEX_BG, the inode/block bitmaps and itable
+ * blocks may not be in the group at all
+ * so the bitmap validation will be skipped for those groups
+ * or it has to also read the block group where the bitmaps
+ * are located to verify they are set.
+ */
+ return 1;
+ }
+ group_first_block = ext4_group_first_block_no(sb, block_group);
+
+ /* check whether block bitmap block number is set */
+ bitmap_blk = ext4_block_bitmap(sb, desc);
+ offset = bitmap_blk - group_first_block;
+ if (!ext4_test_bit(offset, bh->b_data))
+ /* bad block bitmap */
+ goto err_out;
+
+ /* check whether the inode bitmap block number is set */
+ bitmap_blk = ext4_inode_bitmap(sb, desc);
+ offset = bitmap_blk - group_first_block;
+ if (!ext4_test_bit(offset, bh->b_data))
+ /* bad block bitmap */
+ goto err_out;
+
+ /* check whether the inode table block number is set */
+ bitmap_blk = ext4_inode_table(sb, desc);
+ offset = bitmap_blk - group_first_block;
+ next_zero_bit = ext4_find_next_zero_bit(bh->b_data,
+ offset + EXT4_SB(sb)->s_itb_per_group,
+ offset);
+ if (next_zero_bit >= offset + EXT4_SB(sb)->s_itb_per_group)
+ /* good bitmap for inode tables */
+ return 1;
+err_out:
+ ext4_error(sb, __func__,
+ "Invalid block bitmap - "
+ "block_group = %d, block = %llu",
+ block_group, bitmap_blk);
+ return 0;
+}
/**
* read_block_bitmap()
* @sb: super block
* @block_group: given block group
*
- * Read the bitmap for a given block_group, reading into the specified
- * slot in the superblock's bitmap cache.
+ * Read the bitmap for a given block_group,and validate the
+ * bits for block/inode/inode tables are set in the bitmaps
*
* Return buffer_head on success or NULL in case of failure.
*/
struct buffer_head *
-read_block_bitmap(struct super_block *sb, unsigned int block_group)
+read_block_bitmap(struct super_block *sb, ext4_group_t block_group)
{
- int i;
struct ext4_group_desc * desc;
struct buffer_head * bh = NULL;
ext4_fsblk_t bitmap_blk;
if (!desc)
return NULL;
bitmap_blk = ext4_block_bitmap(sb, desc);
- if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) {
- bh = sb_getblk(sb, bitmap_blk);
- if (!buffer_uptodate(bh)) {
- lock_buffer(bh);
- if (!buffer_uptodate(bh)) {
- ext4_init_block_bitmap(sb, bh, block_group,
- desc);
- set_buffer_uptodate(bh);
- }
- unlock_buffer(bh);
- }
- } else {
- bh = sb_bread(sb, bitmap_blk);
- }
- if (!bh)
- ext4_error (sb, __FUNCTION__,
+ bh = sb_getblk(sb, bitmap_blk);
+ if (unlikely(!bh)) {
+ ext4_error(sb, __func__,
"Cannot read block bitmap - "
"block_group = %d, block_bitmap = %llu",
- block_group, bitmap_blk);
-
- /* check whether block bitmap block number is set */
- if (!block_in_use(bitmap_blk, sb, bh->b_data)) {
- /* bad block bitmap */
- goto error_out;
+ (int)block_group, (unsigned long long)bitmap_blk);
+ return NULL;
}
+ if (bh_uptodate_or_lock(bh))
+ return bh;
- /* check whether the inode bitmap block number is set */
- bitmap_blk = ext4_inode_bitmap(sb, desc);
- if (!block_in_use(bitmap_blk, sb, bh->b_data)) {
- /* bad block bitmap */
- goto error_out;
+ if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) {
+ ext4_init_block_bitmap(sb, bh, block_group, desc);
+ set_buffer_uptodate(bh);
+ unlock_buffer(bh);
+ return bh;
}
- /* check whether the inode table block number is set */
- bitmap_blk = ext4_inode_table(sb, desc);
- for (i = 0; i < EXT4_SB(sb)->s_itb_per_group; i++, bitmap_blk++) {
- if (!block_in_use(bitmap_blk, sb, bh->b_data)) {
- /* bad block bitmap */
- goto error_out;
- }
+ if (bh_submit_read(bh) < 0) {
+ put_bh(bh);
+ ext4_error(sb, __func__,
+ "Cannot read block bitmap - "
+ "block_group = %d, block_bitmap = %llu",
+ (int)block_group, (unsigned long long)bitmap_blk);
+ return NULL;
+ }
+ if (!ext4_valid_block_bitmap(sb, desc, block_group, bh)) {
+ put_bh(bh);
+ return NULL;
}
return bh;
-
-error_out:
- brelse(bh);
- ext4_error(sb, __FUNCTION__,
- "Invalid block bitmap - "
- "block_group = %d, block = %llu",
- block_group, bitmap_blk);
- return NULL;
-
}
/*
* The reservation window structure operations
BUG();
}
#define rsv_window_dump(root, verbose) \
- __rsv_window_dump((root), (verbose), __FUNCTION__)
+ __rsv_window_dump((root), (verbose), __func__)
#else
#define rsv_window_dump(root, verbose) do {} while (0)
#endif
*/
static int
goal_in_my_reservation(struct ext4_reserve_window *rsv, ext4_grpblk_t grp_goal,
- unsigned int group, struct super_block * sb)
+ ext4_group_t group, struct super_block *sb)
{
ext4_fsblk_t group_first_block, group_last_block;
* when setting the reservation window size through ioctl before the file
* is open for write (needs block allocation).
*
- * Needs truncate_mutex protection prior to call this function.
+ * Needs down_write(i_data_sem) protection prior to call this function.
*/
void ext4_init_block_alloc_info(struct inode *inode)
{
struct ext4_reserve_window_node *rsv;
spinlock_t *rsv_lock = &EXT4_SB(inode->i_sb)->s_rsv_window_lock;
+ ext4_mb_discard_inode_preallocations(inode);
+
if (!block_i)
return;
{
struct buffer_head *bitmap_bh = NULL;
struct buffer_head *gd_bh;
- unsigned long block_group;
+ ext4_group_t block_group;
ext4_grpblk_t bit;
unsigned long i;
unsigned long overflow;
in_range(ext4_inode_bitmap(sb, desc), block, count) ||
in_range(block, ext4_inode_table(sb, desc), sbi->s_itb_per_group) ||
in_range(block + count - 1, ext4_inode_table(sb, desc),
- sbi->s_itb_per_group))
+ sbi->s_itb_per_group)) {
ext4_error (sb, "ext4_free_blocks",
"Freeing blocks in system zones - "
"Block = %llu, count = %lu",
block, count);
+ goto error_return;
+ }
/*
* We are about to start releasing blocks in the bitmap,
if (!ext4_clear_bit_atomic(sb_bgl_lock(sbi, block_group),
bit + i, bitmap_bh->b_data)) {
jbd_unlock_bh_state(bitmap_bh);
- ext4_error(sb, __FUNCTION__,
+ ext4_error(sb, __func__,
"bit already cleared for block %llu",
(ext4_fsblk_t)(block + i));
jbd_lock_bh_state(bitmap_bh);
jbd_unlock_bh_state(bitmap_bh);
spin_lock(sb_bgl_lock(sbi, block_group));
- desc->bg_free_blocks_count =
- cpu_to_le16(le16_to_cpu(desc->bg_free_blocks_count) +
- group_freed);
+ le16_add_cpu(&desc->bg_free_blocks_count, group_freed);
desc->bg_checksum = ext4_group_desc_csum(sbi, block_group, desc);
spin_unlock(sb_bgl_lock(sbi, block_group));
percpu_counter_add(&sbi->s_freeblocks_counter, count);
* @inode: inode
* @block: start physical block to free
* @count: number of blocks to count
+ * @metadata: Are these metadata blocks
*/
void ext4_free_blocks(handle_t *handle, struct inode *inode,
- ext4_fsblk_t block, unsigned long count)
+ ext4_fsblk_t block, unsigned long count,
+ int metadata)
{
struct super_block * sb;
unsigned long dquot_freed_blocks;
+ /* this isn't the right place to decide whether block is metadata
+ * inode.c/extents.c knows better, but for safety ... */
+ if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode) ||
+ ext4_should_journal_data(inode))
+ metadata = 1;
+
sb = inode->i_sb;
- if (!sb) {
- printk ("ext4_free_blocks: nonexistent device");
- return;
- }
- ext4_free_blocks_sb(handle, sb, block, count, &dquot_freed_blocks);
+
+ if (!test_opt(sb, MBALLOC) || !EXT4_SB(sb)->s_group_info)
+ ext4_free_blocks_sb(handle, sb, block, count,
+ &dquot_freed_blocks);
+ else
+ ext4_mb_free_blocks(handle, inode, block, count,
+ metadata, &dquot_freed_blocks);
if (dquot_freed_blocks)
DQUOT_FREE_BLOCK(inode, dquot_freed_blocks);
return;
* ext4_journal_release_buffer(), else we'll run out of credits.
*/
static ext4_grpblk_t
-ext4_try_to_allocate(struct super_block *sb, handle_t *handle, int group,
- struct buffer_head *bitmap_bh, ext4_grpblk_t grp_goal,
- unsigned long *count, struct ext4_reserve_window *my_rsv)
+ext4_try_to_allocate(struct super_block *sb, handle_t *handle,
+ ext4_group_t group, struct buffer_head *bitmap_bh,
+ ext4_grpblk_t grp_goal, unsigned long *count,
+ struct ext4_reserve_window *my_rsv)
{
ext4_fsblk_t group_first_block;
ext4_grpblk_t start, end;
*/
static int alloc_new_reservation(struct ext4_reserve_window_node *my_rsv,
ext4_grpblk_t grp_goal, struct super_block *sb,
- unsigned int group, struct buffer_head *bitmap_bh)
+ ext4_group_t group, struct buffer_head *bitmap_bh)
{
struct ext4_reserve_window_node *search_head;
ext4_fsblk_t group_first_block, group_end_block, start_block;
*/
static ext4_grpblk_t
ext4_try_to_allocate_with_rsv(struct super_block *sb, handle_t *handle,
- unsigned int group, struct buffer_head *bitmap_bh,
+ ext4_group_t group, struct buffer_head *bitmap_bh,
ext4_grpblk_t grp_goal,
struct ext4_reserve_window_node * my_rsv,
unsigned long *count, int *errp)
}
/**
- * ext4_new_blocks() -- core block(s) allocation function
+ * ext4_new_blocks_old() -- core block(s) allocation function
* @handle: handle to this transaction
* @inode: file inode
* @goal: given target block(filesystem wide)
* any specific goal block.
*
*/
-ext4_fsblk_t ext4_new_blocks(handle_t *handle, struct inode *inode,
+ext4_fsblk_t ext4_new_blocks_old(handle_t *handle, struct inode *inode,
ext4_fsblk_t goal, unsigned long *count, int *errp)
{
struct buffer_head *bitmap_bh = NULL;
struct buffer_head *gdp_bh;
- unsigned long group_no;
- int goal_group;
+ ext4_group_t group_no;
+ ext4_group_t goal_group;
ext4_grpblk_t grp_target_blk; /* blockgroup relative goal block */
ext4_grpblk_t grp_alloc_blk; /* blockgroup-relative allocated block*/
ext4_fsblk_t ret_block; /* filesyetem-wide allocated block */
- int bgi; /* blockgroup iteration index */
+ ext4_group_t bgi; /* blockgroup iteration index */
int fatal = 0, err;
int performed_allocation = 0;
ext4_grpblk_t free_blocks; /* number of free blocks in a group */
struct ext4_reserve_window_node *my_rsv = NULL;
struct ext4_block_alloc_info *block_i;
unsigned short windowsz = 0;
-#ifdef EXT4FS_DEBUG
- static int goal_hits, goal_attempts;
-#endif
- unsigned long ngroups;
+ ext4_group_t ngroups;
unsigned long num = *count;
*errp = -ENOSPC;
sbi = EXT4_SB(sb);
es = EXT4_SB(sb)->s_es;
- ext4_debug("goal=%lu.\n", goal);
+ ext4_debug("goal=%llu.\n", goal);
/*
* Allocate a block from reservation only when
* filesystem is mounted with reservation(default,-o reservation), and
/*
* Now search the rest of the groups. We assume that
- * i and gdp correctly point to the last group visited.
+ * group_no and gdp correctly point to the last group visited.
*/
for (bgi = 0; bgi < ngroups; bgi++) {
group_no++;
allocated:
- ext4_debug("using block group %d(%d)\n",
+ ext4_debug("using block group %lu(%d)\n",
group_no, gdp->bg_free_blocks_count);
BUFFER_TRACE(gdp_bh, "get_write_access");
in_range(ret_block, ext4_inode_table(sb, gdp),
EXT4_SB(sb)->s_itb_per_group) ||
in_range(ret_block + num - 1, ext4_inode_table(sb, gdp),
- EXT4_SB(sb)->s_itb_per_group))
+ EXT4_SB(sb)->s_itb_per_group)) {
ext4_error(sb, "ext4_new_block",
"Allocating block in system zone - "
"blocks from %llu, length %lu",
ret_block, num);
+ goto out;
+ }
performed_allocation = 1;
if (ext4_test_bit(grp_alloc_blk+i,
bh2jh(bitmap_bh)->b_committed_data)) {
printk("%s: block was unexpectedly set in "
- "b_committed_data\n", __FUNCTION__);
+ "b_committed_data\n", __func__);
}
}
}
* list of some description. We don't know in advance whether
* the caller wants to use it as metadata or data.
*/
- ext4_debug("allocating block %lu. Goal hits %d of %d.\n",
- ret_block, goal_hits, goal_attempts);
-
spin_lock(sb_bgl_lock(sbi, group_no));
if (gdp->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))
gdp->bg_flags &= cpu_to_le16(~EXT4_BG_BLOCK_UNINIT);
- gdp->bg_free_blocks_count =
- cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count)-num);
+ le16_add_cpu(&gdp->bg_free_blocks_count, -num);
gdp->bg_checksum = ext4_group_desc_csum(sbi, group_no, gdp);
spin_unlock(sb_bgl_lock(sbi, group_no));
percpu_counter_sub(&sbi->s_freeblocks_counter, num);
}
ext4_fsblk_t ext4_new_block(handle_t *handle, struct inode *inode,
- ext4_fsblk_t goal, int *errp)
+ ext4_fsblk_t goal, int *errp)
{
- unsigned long count = 1;
+ struct ext4_allocation_request ar;
+ ext4_fsblk_t ret;
+
+ if (!test_opt(inode->i_sb, MBALLOC)) {
+ unsigned long count = 1;
+ ret = ext4_new_blocks_old(handle, inode, goal, &count, errp);
+ return ret;
+ }
- return ext4_new_blocks(handle, inode, goal, &count, errp);
+ memset(&ar, 0, sizeof(ar));
+ ar.inode = inode;
+ ar.goal = goal;
+ ar.len = 1;
+ ret = ext4_mb_new_blocks(handle, &ar, errp);
+ return ret;
}
+ext4_fsblk_t ext4_new_blocks(handle_t *handle, struct inode *inode,
+ ext4_fsblk_t goal, unsigned long *count, int *errp)
+{
+ struct ext4_allocation_request ar;
+ ext4_fsblk_t ret;
+
+ if (!test_opt(inode->i_sb, MBALLOC)) {
+ ret = ext4_new_blocks_old(handle, inode, goal, count, errp);
+ return ret;
+ }
+
+ memset(&ar, 0, sizeof(ar));
+ ar.inode = inode;
+ ar.goal = goal;
+ ar.len = *count;
+ ret = ext4_mb_new_blocks(handle, &ar, errp);
+ *count = ar.len;
+ return ret;
+}
+
+
/**
* ext4_count_free_blocks() -- count filesystem free blocks
* @sb: superblock
{
ext4_fsblk_t desc_count;
struct ext4_group_desc *gdp;
- int i;
- unsigned long ngroups = EXT4_SB(sb)->s_groups_count;
+ ext4_group_t i;
+ ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
#ifdef EXT4FS_DEBUG
struct ext4_super_block *es;
ext4_fsblk_t bitmap_count;
continue;
x = ext4_count_free(bitmap_bh, sb->s_blocksize);
- printk("group %d: stored = %d, counted = %lu\n",
+ printk(KERN_DEBUG "group %lu: stored = %d, counted = %lu\n",
i, le16_to_cpu(gdp->bg_free_blocks_count), x);
bitmap_count += x;
}
brelse(bitmap_bh);
printk("ext4_count_free_blocks: stored = %llu"
", computed = %llu, %llu\n",
- EXT4_FREE_BLOCKS_COUNT(es),
+ ext4_free_blocks_count(es),
desc_count, bitmap_count);
return bitmap_count;
#else
#endif
}
-static inline int test_root(int a, int b)
+static inline int test_root(ext4_group_t a, int b)
{
int num = b;
return num == a;
}
-static int ext4_group_sparse(int group)
+static int ext4_group_sparse(ext4_group_t group)
{
if (group <= 1)
return 1;
* Return the number of blocks used by the superblock (primary or backup)
* in this group. Currently this will be only 0 or 1.
*/
-int ext4_bg_has_super(struct super_block *sb, int group)
+int ext4_bg_has_super(struct super_block *sb, ext4_group_t group)
{
if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER) &&
return 1;
}
-static unsigned long ext4_bg_num_gdb_meta(struct super_block *sb, int group)
+static unsigned long ext4_bg_num_gdb_meta(struct super_block *sb,
+ ext4_group_t group)
{
unsigned long metagroup = group / EXT4_DESC_PER_BLOCK(sb);
- unsigned long first = metagroup * EXT4_DESC_PER_BLOCK(sb);
- unsigned long last = first + EXT4_DESC_PER_BLOCK(sb) - 1;
+ ext4_group_t first = metagroup * EXT4_DESC_PER_BLOCK(sb);
+ ext4_group_t last = first + EXT4_DESC_PER_BLOCK(sb) - 1;
if (group == first || group == first + 1 || group == last)
return 1;
return 0;
}
-static unsigned long ext4_bg_num_gdb_nometa(struct super_block *sb, int group)
+static unsigned long ext4_bg_num_gdb_nometa(struct super_block *sb,
+ ext4_group_t group)
{
- if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
- EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER) &&
- !ext4_group_sparse(group))
- return 0;
- return EXT4_SB(sb)->s_gdb_count;
+ return ext4_bg_has_super(sb, group) ? EXT4_SB(sb)->s_gdb_count : 0;
}
/**
* (primary or backup) in this group. In the future there may be a
* different number of descriptor blocks in each group.
*/
-unsigned long ext4_bg_num_gdb(struct super_block *sb, int group)
+unsigned long ext4_bg_num_gdb(struct super_block *sb, ext4_group_t group)
{
unsigned long first_meta_bg =
le32_to_cpu(EXT4_SB(sb)->s_es->s_first_meta_bg);