/* preallocated blocks don't need to be run through journal_mark_freed */
static void reiserfs_free_prealloc_block (struct reiserfs_transaction_handle *th,
struct inode *inode, b_blocknr_t block) {
RFALSE(!th->t_super, "vs-4060: trying to free block on nonexistent device");
RFALSE(is_reusable (th->t_super, block, 1) == 0, "vs-4070: can not free such block");
BUG_ON (!th->t_trans_id);
_reiserfs_free_block(th, inode, block, 1) ;
}
void reiserfs_free_block (struct reiserfs_transaction_handle *th,
unsigned long block) {
struct super_block * s = th->t_super;
RFALSE(!s, "vs-4061: trying to free block on nonexistent device");
RFALSE(is_reusable (s, block, 1) == 0, "vs-4071: can not free such block");
/* mark it before we clear it, just in case */
journal_mark_freed(th, s, block) ;
_reiserfs_free_block(th, block) ;
}
/* Replace delimiting key of buffers S[h] and R[h] by the given key.*/
static void replace_rkey(struct tree_balance *tb, int h, struct item_head *key)
{
RFALSE(tb->R[h] == NULL || tb->CFR[h] == NULL,
"R[h](%p) and CFR[h](%p) must exist in replace_rkey",
tb->R[h], tb->CFR[h]);
RFALSE(B_NR_ITEMS(tb->R[h]) == 0,
"R[h] can not be empty if it exists (item number=%d)",
B_NR_ITEMS(tb->R[h]));
memcpy(B_N_PDELIM_KEY(tb->CFR[h], tb->rkey[h]), key, KEY_SIZE);
do_balance_mark_internal_dirty(tb, tb->CFR[h], 0);
}
/* Insert n_src'th key of buffer src before n_dest'th key of buffer dest. */
static void internal_insert_key(struct buffer_info *dest_bi,
/* insert key before key with n_dest number */
int dest_position_before,
struct buffer_head *src, int src_position)
{
struct buffer_head *dest = dest_bi->bi_bh;
int nr;
struct block_head *blkh;
struct reiserfs_key *key;
RFALSE(dest == NULL || src == NULL,
"source(%p) or dest(%p) buffer is 0", src, dest);
RFALSE(dest_position_before < 0 || src_position < 0,
"source(%d) or dest(%d) key number less than 0",
src_position, dest_position_before);
RFALSE(dest_position_before > B_NR_ITEMS(dest) ||
src_position >= B_NR_ITEMS(src),
"invalid position in dest (%d (key number %d)) or in src (%d (key number %d))",
dest_position_before, B_NR_ITEMS(dest),
src_position, B_NR_ITEMS(src));
RFALSE(B_FREE_SPACE(dest) < KEY_SIZE,
"no enough free space (%d) in dest buffer", B_FREE_SPACE(dest));
blkh = B_BLK_HEAD(dest);
nr = blkh_nr_item(blkh);
/* prepare space for inserting key */
key = internal_key(dest, dest_position_before);
memmove(key + 1, key,
(nr - dest_position_before) * KEY_SIZE + (nr + 1) * DC_SIZE);
/* insert key */
memcpy(key, internal_key(src, src_position), KEY_SIZE);
/* Change dirt, free space, item number fields. */
set_blkh_nr_item(blkh, blkh_nr_item(blkh) + 1);
set_blkh_free_space(blkh, blkh_free_space(blkh) - KEY_SIZE);
do_balance_mark_internal_dirty(dest_bi->tb, dest, 0);
if (dest_bi->bi_parent) {
struct disk_child *t_dc;
t_dc = B_N_CHILD(dest_bi->bi_parent, dest_bi->bi_position);
put_dc_size(t_dc, dc_size(t_dc) + KEY_SIZE);
do_balance_mark_internal_dirty(dest_bi->tb, dest_bi->bi_parent,
0);
}
}
void reiserfs_free_block (struct reiserfs_transaction_handle *th,
struct inode *inode, b_blocknr_t block,
int for_unformatted)
{
struct super_block * s = th->t_super;
BUG_ON (!th->t_trans_id);
RFALSE(!s, "vs-4061: trying to free block on nonexistent device");
RFALSE(is_reusable (s, block, 1) == 0, "vs-4071: can not free such block");
/* mark it before we clear it, just in case */
journal_mark_freed(th, s, block) ;
_reiserfs_free_block(th, inode, block, for_unformatted) ;
}
/* Insert d_key'th (delimiting) key from buffer cfr to head of dest.
* Copy n node pointers and n - 1 items from buffer src to buffer dest.
* Replace d_key'th key in buffer cfr.
* Delete n items and node pointers from buffer src.
*/
static void internal_shift_right(int mode, /* INTERNAL_FROM_S_TO_R | INTERNAL_FROM_L_TO_S */
struct tree_balance *tb,
int h, int pointer_amount)
{
struct buffer_info dest_bi, src_bi;
struct buffer_head *cf;
int d_key_position;
int nr;
internal_define_dest_src_infos(mode, tb, h, &dest_bi, &src_bi,
&d_key_position, &cf);
nr = B_NR_ITEMS(src_bi.bi_bh);
if (pointer_amount > 0) {
/* insert delimiting key from common father of dest and src to dest node into position 0 */
internal_insert_key(&dest_bi, 0, cf, d_key_position);
if (nr == pointer_amount - 1) {
RFALSE(src_bi.bi_bh != PATH_H_PBUFFER(tb->tb_path, h) /*tb->S[h] */ ||
dest_bi.bi_bh != tb->R[h],
"src (%p) must be == tb->S[h](%p) when it disappears",
src_bi.bi_bh, PATH_H_PBUFFER(tb->tb_path, h));
/* when S[h] disappers replace left delemiting key as well */
if (tb->CFL[h])
replace_key(tb, cf, d_key_position, tb->CFL[h],
tb->lkey[h]);
} else
replace_key(tb, cf, d_key_position, src_bi.bi_bh,
nr - pointer_amount);
}
/* last parameter is del_parameter */
internal_move_pointers_items(&dest_bi, &src_bi, LAST_TO_FIRST,
pointer_amount, 0);
}
/* Does the buffer contain a disk block which is in the tree. */
inline int B_IS_IN_TREE(const struct buffer_head *bh)
{
RFALSE(B_LEVEL(bh) > MAX_HEIGHT,
"PAP-1010: block (%b) has too big level (%z)", bh, bh);
return (B_LEVEL(bh) != FREE_LEVEL);
}
/* Get delimiting key of the buffer by looking for it in the buffers in the path, starting from the bottom
of the path, and going upwards. We must check the path's validity at each step. If the key is not in
the path, there is no delimiting key in the tree (buffer is first or last buffer in tree), and in this
case we return a special key, either MIN_KEY or MAX_KEY. */
static inline const struct reiserfs_key *get_lkey(const struct treepath *chk_path,
const struct super_block *sb)
{
int position, path_offset = chk_path->path_length;
struct buffer_head *parent;
RFALSE(path_offset < FIRST_PATH_ELEMENT_OFFSET,
"PAP-5010: invalid offset in the path");
/* While not higher in path than first element. */
while (path_offset-- > FIRST_PATH_ELEMENT_OFFSET) {
RFALSE(!buffer_uptodate
(PATH_OFFSET_PBUFFER(chk_path, path_offset)),
"PAP-5020: parent is not uptodate");
/* Parent at the path is not in the tree now. */
if (!B_IS_IN_TREE
(parent =
PATH_OFFSET_PBUFFER(chk_path, path_offset)))
return &MAX_KEY;
/* Check whether position in the parent is correct. */
if ((position =
PATH_OFFSET_POSITION(chk_path,
path_offset)) >
B_NR_ITEMS(parent))
return &MAX_KEY;
/* Check whether parent at the path really points to the child. */
if (B_N_CHILD_NUM(parent, position) !=
PATH_OFFSET_PBUFFER(chk_path,
path_offset + 1)->b_blocknr)
return &MAX_KEY;
/* Return delimiting key if position in the parent is not equal to zero. */
if (position)
return B_N_PDELIM_KEY(parent, position - 1);
}
/* Return MIN_KEY if we are in the root of the buffer tree. */
if (PATH_OFFSET_PBUFFER(chk_path, FIRST_PATH_ELEMENT_OFFSET)->
b_blocknr == SB_ROOT_BLOCK(sb))
return &MIN_KEY;
return &MAX_KEY;
}
/* Drop the reference to each buffer in a path */
void pathrelse(struct treepath *search_path)
{
int path_offset = search_path->path_length;
RFALSE(path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
"PAP-5090: invalid path offset");
while (path_offset > ILLEGAL_PATH_ELEMENT_OFFSET)
brelse(PATH_OFFSET_PBUFFER(search_path, path_offset--));
search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
}
/* This works by looking at the left and right delimiting keys for the buffer in the last path_element in
the path. These delimiting keys are stored at least one level above that buffer in the tree. If the
buffer is the first or last node in the tree order then one of the delimiting keys may be absent, and in
this case get_lkey and get_rkey return a special key which is MIN_KEY or MAX_KEY. */
static inline int key_in_buffer(struct treepath *chk_path, /* Path which should be checked. */
const struct cpu_key *key, /* Key which should be checked. */
struct super_block *sb
)
{
RFALSE(!key || chk_path->path_length < FIRST_PATH_ELEMENT_OFFSET
|| chk_path->path_length > MAX_HEIGHT,
"PAP-5050: pointer to the key(%p) is NULL or invalid path length(%d)",
key, chk_path->path_length);
RFALSE(!PATH_PLAST_BUFFER(chk_path)->b_bdev,
"PAP-5060: device must not be NODEV");
if (comp_keys(get_lkey(chk_path, sb), key) == 1)
/* left delimiting key is bigger, that the key we look for */
return 0;
/* if ( comp_keys(key, get_rkey(chk_path, sb)) != -1 ) */
if (comp_keys(get_rkey(chk_path, sb), key) != 1)
/* key must be less than right delimitiing key */
return 0;
return 1;
}
/* I wonder if it would be less modest
now that we use journaling. -Hans */
void reiserfs_free_block (struct reiserfs_transaction_handle *th, unsigned long block)
{
struct super_block * s = th->t_super;
struct reiserfs_super_block * rs;
struct buffer_head * sbh;
struct buffer_head ** apbh;
int nr, offset;
RFALSE(!s, "vs-4060: trying to free block on nonexistent device");
RFALSE(is_reusable (s, block, 1) == 0, "vs-4070: can not free such block");
PROC_INFO_INC( s, free_block );
rs = SB_DISK_SUPER_BLOCK (s);
sbh = SB_BUFFER_WITH_SB (s);
apbh = SB_AP_BITMAP (s);
get_bit_address (s, block, &nr, &offset);
/* mark it before we clear it, just in case */
journal_mark_freed(th, s, block) ;
reiserfs_prepare_for_journal(s, apbh[nr], 1 ) ;
/* clear bit for the given block in bit map */
if (!reiserfs_test_and_clear_le_bit (offset, apbh[nr]->b_data)) {
reiserfs_warning ("vs-4080: reiserfs_free_block: "
"free_block (%04x:%lu)[dev:blocknr]: bit already cleared\n",
s->s_dev, block);
}
journal_mark_dirty (th, s, apbh[nr]);
reiserfs_prepare_for_journal(s, sbh, 1) ;
/* update super block */
set_sb_free_blocks( rs, sb_free_blocks(rs) + 1 );
journal_mark_dirty (th, s, sbh);
s->s_dirt = 1;
}
/* Replace delimiting key of buffers L[h] and S[h] by the given key.*/
static void replace_lkey(struct tree_balance *tb, int h, struct item_head *key)
{
RFALSE(tb->L[h] == NULL || tb->CFL[h] == NULL,
"L[h](%p) and CFL[h](%p) must exist in replace_lkey",
tb->L[h], tb->CFL[h]);
if (B_NR_ITEMS(PATH_H_PBUFFER(tb->tb_path, h)) == 0)
return;
memcpy(B_N_PDELIM_KEY(tb->CFL[h], tb->lkey[h]), key, KEY_SIZE);
do_balance_mark_internal_dirty(tb, tb->CFL[h], 0);
}
/* Unreserve @blocks amount of blocks in fs pointed by @sb */
void reiserfs_release_claimed_blocks(
struct super_block *sb, /* super block of
filesystem where
blocks should be
reserved */
int blocks /* How much to unreserve */
)
{
/* Fast case, if unreservation is zero - exit immediately. */
if ( !blocks )
return;
sb->u.reiserfs_sb.reserved_blocks -= blocks;
RFALSE( sb->u.reiserfs_sb.reserved_blocks < 0, "amount of blocks reserved became zero?");
}
/* Drop the reference to each buffer in a path and restore
* dirty bits clean when preparing the buffer for the log.
* This version should only be called from fix_nodes() */
void pathrelse_and_restore(struct super_block *sb,
struct treepath *search_path)
{
int path_offset = search_path->path_length;
RFALSE(path_offset < ILLEGAL_PATH_ELEMENT_OFFSET,
"clm-4000: invalid path offset");
while (path_offset > ILLEGAL_PATH_ELEMENT_OFFSET) {
struct buffer_head *bh;
bh = PATH_OFFSET_PBUFFER(search_path, path_offset--);
reiserfs_restore_prepared_buffer(sb, bh);
brelse(bh);
}
search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET;
}
/* Unreserve @blocks amount of blocks in fs pointed by @sb */
void reiserfs_release_claimed_blocks(
struct super_block *sb, /* super block of
filesystem where
blocks should be
reserved */
int blocks /* How much to unreserve */
)
{
/* Fast case, if unreservation is zero - exit immediately. */
if ( !blocks )
return;
spin_lock(&REISERFS_SB(sb)->bitmap_lock);
REISERFS_SB(sb)->reserved_blocks -= blocks;
spin_unlock(&REISERFS_SB(sb)->bitmap_lock);
RFALSE( REISERFS_SB(sb)->reserved_blocks < 0, "amount of blocks reserved became zero?");
}
/* returns 1 if it finds an indirect item and gets valid hint info
* from it, otherwise 0
*/
static int get_left_neighbor(reiserfs_blocknr_hint_t *hint)
{
struct path * path;
struct buffer_head * bh;
struct item_head * ih;
int pos_in_item;
__u32 * item;
int ret = 0;
if (!hint->path) /* reiserfs code can call this function w/o pointer to path
* structure supplied; then we rely on supplied search_start */
return 0;
path = hint->path;
bh = get_last_bh(path);
RFALSE( !bh, "green-4002: Illegal path specified to get_left_neighbor");
ih = get_ih(path);
pos_in_item = path->pos_in_item;
item = get_item (path);
hint->search_start = bh->b_blocknr;
if (!hint->formatted_node && is_indirect_le_ih (ih)) {
/* for indirect item: go to left and look for the first non-hole entry
in the indirect item */
if (pos_in_item == I_UNFM_NUM (ih))
pos_in_item--;
// pos_in_item = I_UNFM_NUM (ih) - 1;
while (pos_in_item >= 0) {
int t=get_block_num(item,pos_in_item);
if (t) {
hint->search_start = t;
ret = 1;
break;
}
pos_in_item --;
}
}
/* does result value fit into specified region? */
return ret;
}
/* it searches for a window of zero bits with given minimum and maximum lengths in one bitmap
* block; */
static int scan_bitmap_block (struct reiserfs_transaction_handle *th,
int bmap_n, int *beg, int boundary, int min, int max, int unfm)
{
struct super_block *s = th->t_super;
struct reiserfs_bitmap_info *bi=&SB_AP_BITMAP(s)[bmap_n];
int end, next;
int org = *beg;
BUG_ON (!th->t_trans_id);
RFALSE(bmap_n >= SB_BMAP_NR (s), "Bitmap %d is out of range (0..%d)",bmap_n, SB_BMAP_NR (s) - 1);
PROC_INFO_INC( s, scan_bitmap.bmap );
/* this is unclear and lacks comments, explain how journal bitmaps
work here for the reader. Convey a sense of the design here. What
is a window? */
/* - I mean `a window of zero bits' as in description of this function - Zam. */
if ( !bi ) {
reiserfs_warning (s, "NULL bitmap info pointer for bitmap %d", bmap_n);
return 0;
}
if (buffer_locked (bi->bh)) {
PROC_INFO_INC( s, scan_bitmap.wait );
__wait_on_buffer (bi->bh);
}
while (1) {
cont:
if (bi->free_count < min)
return 0; // No free blocks in this bitmap
/* search for a first zero bit -- beggining of a window */
*beg = reiserfs_find_next_zero_le_bit
((unsigned long*)(bi->bh->b_data), boundary, *beg);
if (*beg + min > boundary) {
/* search for a zero bit fails or the rest of bitmap block
* cannot contain a zero window of minimum size */
return 0;
}
if (unfm && is_block_in_journal(s,bmap_n, *beg, beg))
continue;
/* first zero bit found; we check next bits */
for (end = *beg + 1;; end ++) {
if (end >= *beg + max || end >= boundary || reiserfs_test_le_bit (end, bi->bh->b_data)) {
next = end;
break;
}
/* finding the other end of zero bit window requires looking into journal structures (in
* case of searching for free blocks for unformatted nodes) */
if (unfm && is_block_in_journal(s, bmap_n, end, &next))
break;
}
/* now (*beg) points to beginning of zero bits window,
* (end) points to one bit after the window end */
if (end - *beg >= min) { /* it seems we have found window of proper size */
int i;
reiserfs_prepare_for_journal (s, bi->bh, 1);
/* try to set all blocks used checking are they still free */
for (i = *beg; i < end; i++) {
/* It seems that we should not check in journal again. */
if (reiserfs_test_and_set_le_bit (i, bi->bh->b_data)) {
/* bit was set by another process
* while we slept in prepare_for_journal() */
PROC_INFO_INC( s, scan_bitmap.stolen );
if (i >= *beg + min) {
/* we can continue with smaller set of allocated blocks,
* if length of this set is more or equal to `min' */
end = i;
break;
}
/* otherwise we clear all bit were set ... */
while (--i >= *beg)
reiserfs_test_and_clear_le_bit (i, bi->bh->b_data);
reiserfs_restore_prepared_buffer (s, bi->bh);
*beg = org;
/* ... and search again in current block from beginning */
goto cont;
}
}
bi->free_count -= (end - *beg);
journal_mark_dirty (th, s, bi->bh);
/* free block count calculation */
reiserfs_prepare_for_journal (s, SB_BUFFER_WITH_SB(s), 1);
PUT_SB_FREE_BLOCKS(s, SB_FREE_BLOCKS(s) - (end - *beg));
journal_mark_dirty (th, s, SB_BUFFER_WITH_SB(s));
return end - (*beg);
} else {
*beg = next;
}
}
}
int balance_internal(struct tree_balance *tb, /* tree_balance structure */
int h, /* level of the tree */
int child_pos, struct item_head *insert_key, /* key for insertion on higher level */
struct buffer_head **insert_ptr /* node for insertion on higher level */
)
/* if inserting/pasting
{
child_pos is the position of the node-pointer in S[h] that *
pointed to S[h-1] before balancing of the h-1 level; *
this means that new pointers and items must be inserted AFTER *
child_pos
}
else
{
it is the position of the leftmost pointer that must be deleted (together with
its corresponding key to the left of the pointer)
as a result of the previous level's balancing.
}
*/
{
struct buffer_head *tbSh = PATH_H_PBUFFER(tb->tb_path, h);
struct buffer_info bi;
int order; /* we return this: it is 0 if there is no S[h], else it is tb->S[h]->b_item_order */
int insert_num, n, k;
struct buffer_head *S_new;
struct item_head new_insert_key;
struct buffer_head *new_insert_ptr = NULL;
struct item_head *new_insert_key_addr = insert_key;
RFALSE(h < 1, "h (%d) can not be < 1 on internal level", h);
PROC_INFO_INC(tb->tb_sb, balance_at[h]);
order =
(tbSh) ? PATH_H_POSITION(tb->tb_path,
h + 1) /*tb->S[h]->b_item_order */ : 0;
/* Using insert_size[h] calculate the number insert_num of items
that must be inserted to or deleted from S[h]. */
insert_num = tb->insert_size[h] / ((int)(KEY_SIZE + DC_SIZE));
/* Check whether insert_num is proper * */
RFALSE(insert_num < -2 || insert_num > 2,
"incorrect number of items inserted to the internal node (%d)",
insert_num);
RFALSE(h > 1 && (insert_num > 1 || insert_num < -1),
"incorrect number of items (%d) inserted to the internal node on a level (h=%d) higher than last internal level",
insert_num, h);
/* Make balance in case insert_num < 0 */
if (insert_num < 0) {
balance_internal_when_delete(tb, h, child_pos);
return order;
}
k = 0;
if (tb->lnum[h] > 0) {
/* shift lnum[h] items from S[h] to the left neighbor L[h].
check how many of new items fall into L[h] or CFL[h] after
shifting */
n = B_NR_ITEMS(tb->L[h]); /* number of items in L[h] */
if (tb->lnum[h] <= child_pos) {
/* new items don't fall into L[h] or CFL[h] */
internal_shift_left(INTERNAL_SHIFT_FROM_S_TO_L, tb, h,
tb->lnum[h]);
/*internal_shift_left (tb->L[h],tb->CFL[h],tb->lkey[h],tbSh,tb->lnum[h]); */
child_pos -= tb->lnum[h];
} else if (tb->lnum[h] > child_pos + insert_num) {
/* all new items fall into L[h] */
internal_shift_left(INTERNAL_SHIFT_FROM_S_TO_L, tb, h,
tb->lnum[h] - insert_num);
/* internal_shift_left(tb->L[h],tb->CFL[h],tb->lkey[h],tbSh,
tb->lnum[h]-insert_num);
*/
/* insert insert_num keys and node-pointers into L[h] */
bi.tb = tb;
bi.bi_bh = tb->L[h];
bi.bi_parent = tb->FL[h];
bi.bi_position = get_left_neighbor_position(tb, h);
internal_insert_childs(&bi,
/*tb->L[h], tb->S[h-1]->b_next */
n + child_pos + 1,
insert_num, insert_key,
insert_ptr);
insert_num = 0;
} else {
struct disk_child *dc;
/* some items fall into L[h] or CFL[h], but some don't fall */
internal_shift1_left(tb, h, child_pos + 1);
/* calculate number of new items that fall into L[h] */
k = tb->lnum[h] - child_pos - 1;
bi.tb = tb;
bi.bi_bh = tb->L[h];
bi.bi_parent = tb->FL[h];
bi.bi_position = get_left_neighbor_position(tb, h);
internal_insert_childs(&bi,
/*tb->L[h], tb->S[h-1]->b_next, */
n + child_pos + 1, k,
insert_key, insert_ptr);
replace_lkey(tb, h, insert_key + k);
/* replace the first node-ptr in S[h] by node-ptr to insert_ptr[k] */
dc = B_N_CHILD(tbSh, 0);
put_dc_size(dc,
MAX_CHILD_SIZE(insert_ptr[k]) -
B_FREE_SPACE(insert_ptr[k]));
put_dc_block_number(dc, insert_ptr[k]->b_blocknr);
do_balance_mark_internal_dirty(tb, tbSh, 0);
k++;
insert_key += k;
insert_ptr += k;
insert_num -= k;
child_pos = 0;
}
}
/* tb->lnum[h] > 0 */
if (tb->rnum[h] > 0) {
/*shift rnum[h] items from S[h] to the right neighbor R[h] */
/* check how many of new items fall into R or CFR after shifting */
n = B_NR_ITEMS(tbSh); /* number of items in S[h] */
if (n - tb->rnum[h] >= child_pos)
/* new items fall into S[h] */
/*internal_shift_right(tb,h,tbSh,tb->CFR[h],tb->rkey[h],tb->R[h],tb->rnum[h]); */
internal_shift_right(INTERNAL_SHIFT_FROM_S_TO_R, tb, h,
tb->rnum[h]);
else if (n + insert_num - tb->rnum[h] < child_pos) {
/* all new items fall into R[h] */
/*internal_shift_right(tb,h,tbSh,tb->CFR[h],tb->rkey[h],tb->R[h],
tb->rnum[h] - insert_num); */
internal_shift_right(INTERNAL_SHIFT_FROM_S_TO_R, tb, h,
tb->rnum[h] - insert_num);
/* insert insert_num keys and node-pointers into R[h] */
bi.tb = tb;
bi.bi_bh = tb->R[h];
bi.bi_parent = tb->FR[h];
bi.bi_position = get_right_neighbor_position(tb, h);
internal_insert_childs(&bi,
/*tb->R[h],tb->S[h-1]->b_next */
child_pos - n - insert_num +
tb->rnum[h] - 1,
insert_num, insert_key,
insert_ptr);
insert_num = 0;
} else {
struct disk_child *dc;
/* one of the items falls into CFR[h] */
internal_shift1_right(tb, h, n - child_pos + 1);
/* calculate number of new items that fall into R[h] */
k = tb->rnum[h] - n + child_pos - 1;
bi.tb = tb;
bi.bi_bh = tb->R[h];
bi.bi_parent = tb->FR[h];
bi.bi_position = get_right_neighbor_position(tb, h);
internal_insert_childs(&bi,
/*tb->R[h], tb->R[h]->b_child, */
0, k, insert_key + 1,
insert_ptr + 1);
replace_rkey(tb, h, insert_key + insert_num - k - 1);
/* replace the first node-ptr in R[h] by node-ptr insert_ptr[insert_num-k-1] */
dc = B_N_CHILD(tb->R[h], 0);
put_dc_size(dc,
MAX_CHILD_SIZE(insert_ptr
[insert_num - k - 1]) -
B_FREE_SPACE(insert_ptr
[insert_num - k - 1]));
put_dc_block_number(dc,
insert_ptr[insert_num - k -
1]->b_blocknr);
do_balance_mark_internal_dirty(tb, tb->R[h], 0);
insert_num -= (k + 1);
}
}
/** Fill new node that appears instead of S[h] **/
RFALSE(tb->blknum[h] > 2, "blknum can not be > 2 for internal level");
RFALSE(tb->blknum[h] < 0, "blknum can not be < 0");
if (!tb->blknum[h]) { /* node S[h] is empty now */
RFALSE(!tbSh, "S[h] is equal NULL");
/* do what is needed for buffer thrown from tree */
reiserfs_invalidate_buffer(tb, tbSh);
return order;
}
if (!tbSh) {
/* create new root */
struct disk_child *dc;
struct buffer_head *tbSh_1 = PATH_H_PBUFFER(tb->tb_path, h - 1);
struct block_head *blkh;
if (tb->blknum[h] != 1)
reiserfs_panic(NULL, "ibalance-3", "One new node "
"required for creating the new root");
/* S[h] = empty buffer from the list FEB. */
tbSh = get_FEB(tb);
blkh = B_BLK_HEAD(tbSh);
set_blkh_level(blkh, h + 1);
/* Put the unique node-pointer to S[h] that points to S[h-1]. */
dc = B_N_CHILD(tbSh, 0);
put_dc_block_number(dc, tbSh_1->b_blocknr);
put_dc_size(dc,
(MAX_CHILD_SIZE(tbSh_1) - B_FREE_SPACE(tbSh_1)));
tb->insert_size[h] -= DC_SIZE;
set_blkh_free_space(blkh, blkh_free_space(blkh) - DC_SIZE);
do_balance_mark_internal_dirty(tb, tbSh, 0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(tbSh);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
/* put new root into path structure */
PATH_OFFSET_PBUFFER(tb->tb_path, ILLEGAL_PATH_ELEMENT_OFFSET) =
tbSh;
/* Change root in structure super block. */
PUT_SB_ROOT_BLOCK(tb->tb_sb, tbSh->b_blocknr);
PUT_SB_TREE_HEIGHT(tb->tb_sb, SB_TREE_HEIGHT(tb->tb_sb) + 1);
do_balance_mark_sb_dirty(tb, REISERFS_SB(tb->tb_sb)->s_sbh, 1);
}
if (tb->blknum[h] == 2) {
int snum;
struct buffer_info dest_bi, src_bi;
/* S_new = free buffer from list FEB */
S_new = get_FEB(tb);
set_blkh_level(B_BLK_HEAD(S_new), h + 1);
dest_bi.tb = tb;
dest_bi.bi_bh = S_new;
dest_bi.bi_parent = NULL;
dest_bi.bi_position = 0;
src_bi.tb = tb;
src_bi.bi_bh = tbSh;
src_bi.bi_parent = PATH_H_PPARENT(tb->tb_path, h);
src_bi.bi_position = PATH_H_POSITION(tb->tb_path, h + 1);
n = B_NR_ITEMS(tbSh); /* number of items in S[h] */
snum = (insert_num + n + 1) / 2;
if (n - snum >= child_pos) {
/* new items don't fall into S_new */
/* store the delimiting key for the next level */
/* new_insert_key = (n - snum)'th key in S[h] */
memcpy(&new_insert_key, B_N_PDELIM_KEY(tbSh, n - snum),
KEY_SIZE);
/* last parameter is del_par */
internal_move_pointers_items(&dest_bi, &src_bi,
LAST_TO_FIRST, snum, 0);
/* internal_move_pointers_items(S_new, tbSh, LAST_TO_FIRST, snum, 0); */
} else if (n + insert_num - snum < child_pos) {
/* all new items fall into S_new */
/* store the delimiting key for the next level */
/* new_insert_key = (n + insert_item - snum)'th key in S[h] */
memcpy(&new_insert_key,
B_N_PDELIM_KEY(tbSh, n + insert_num - snum),
KEY_SIZE);
/* last parameter is del_par */
internal_move_pointers_items(&dest_bi, &src_bi,
LAST_TO_FIRST,
snum - insert_num, 0);
/* internal_move_pointers_items(S_new,tbSh,1,snum - insert_num,0); */
/* insert insert_num keys and node-pointers into S_new */
internal_insert_childs(&dest_bi,
/*S_new,tb->S[h-1]->b_next, */
child_pos - n - insert_num +
snum - 1,
insert_num, insert_key,
insert_ptr);
insert_num = 0;
} else {
struct disk_child *dc;
/* some items fall into S_new, but some don't fall */
/* last parameter is del_par */
internal_move_pointers_items(&dest_bi, &src_bi,
LAST_TO_FIRST,
n - child_pos + 1, 1);
/* internal_move_pointers_items(S_new,tbSh,1,n - child_pos + 1,1); */
/* calculate number of new items that fall into S_new */
k = snum - n + child_pos - 1;
internal_insert_childs(&dest_bi, /*S_new, */ 0, k,
insert_key + 1, insert_ptr + 1);
/* new_insert_key = insert_key[insert_num - k - 1] */
memcpy(&new_insert_key, insert_key + insert_num - k - 1,
KEY_SIZE);
/* replace first node-ptr in S_new by node-ptr to insert_ptr[insert_num-k-1] */
dc = B_N_CHILD(S_new, 0);
put_dc_size(dc,
(MAX_CHILD_SIZE
(insert_ptr[insert_num - k - 1]) -
B_FREE_SPACE(insert_ptr
[insert_num - k - 1])));
put_dc_block_number(dc,
insert_ptr[insert_num - k -
1]->b_blocknr);
do_balance_mark_internal_dirty(tb, S_new, 0);
insert_num -= (k + 1);
}
/* new_insert_ptr = node_pointer to S_new */
new_insert_ptr = S_new;
RFALSE(!buffer_journaled(S_new) || buffer_journal_dirty(S_new)
|| buffer_dirty(S_new), "cm-00001: bad S_new (%b)",
S_new);
// S_new is released in unfix_nodes
}
n = B_NR_ITEMS(tbSh); /*number of items in S[h] */
if (0 <= child_pos && child_pos <= n && insert_num > 0) {
bi.tb = tb;
bi.bi_bh = tbSh;
bi.bi_parent = PATH_H_PPARENT(tb->tb_path, h);
bi.bi_position = PATH_H_POSITION(tb->tb_path, h + 1);
internal_insert_childs(&bi, /*tbSh, */
/* ( tb->S[h-1]->b_parent == tb->S[h] ) ? tb->S[h-1]->b_next : tb->S[h]->b_child->b_next, */
child_pos, insert_num, insert_key,
insert_ptr);
}
memcpy(new_insert_key_addr, &new_insert_key, KEY_SIZE);
insert_ptr[0] = new_insert_ptr;
return order;
}
/* makes object identifier unused */
void reiserfs_release_objectid(struct reiserfs_transaction_handle *th,
__u32 objectid_to_release)
{
struct super_block *s = th->t_super;
struct reiserfs_super_block *rs = SB_DISK_SUPER_BLOCK(s);
__le32 *map = objectid_map(s, rs);
int i = 0;
BUG_ON(!th->t_trans_id);
//return;
check_objectid_map(s, map);
reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1);
journal_mark_dirty(th, s, SB_BUFFER_WITH_SB(s));
/* start at the beginning of the objectid map (i = 0) and go to
the end of it (i = disk_sb->s_oid_cursize). Linear search is
what we use, though it is possible that binary search would be
more efficient after performing lots of deletions (which is
when oids is large.) We only check even i's. */
while (i < sb_oid_cursize(rs)) {
if (objectid_to_release == le32_to_cpu(map[i])) {
/* This incrementation unallocates the objectid. */
//map[i]++;
le32_add_cpu(&map[i], 1);
/* Did we unallocate the last member of an odd sequence, and can shrink oids? */
if (map[i] == map[i + 1]) {
/* shrink objectid map */
memmove(map + i, map + i + 2,
(sb_oid_cursize(rs) - i -
2) * sizeof(__u32));
//disk_sb->s_oid_cursize -= 2;
set_sb_oid_cursize(rs, sb_oid_cursize(rs) - 2);
RFALSE(sb_oid_cursize(rs) < 2 ||
sb_oid_cursize(rs) > sb_oid_maxsize(rs),
"vs-15005: objectid map corrupted cur_size == %d (max == %d)",
sb_oid_cursize(rs), sb_oid_maxsize(rs));
}
return;
}
if (objectid_to_release > le32_to_cpu(map[i]) &&
objectid_to_release < le32_to_cpu(map[i + 1])) {
/* size of objectid map is not changed */
if (objectid_to_release + 1 == le32_to_cpu(map[i + 1])) {
//objectid_map[i+1]--;
le32_add_cpu(&map[i + 1], -1);
return;
}
/* JDM comparing two little-endian values for equality -- safe */
if (sb_oid_cursize(rs) == sb_oid_maxsize(rs)) {
/* objectid map must be expanded, but there is no space */
PROC_INFO_INC(s, leaked_oid);
return;
}
/* expand the objectid map */
memmove(map + i + 3, map + i + 1,
(sb_oid_cursize(rs) - i - 1) * sizeof(__u32));
map[i + 1] = cpu_to_le32(objectid_to_release);
map[i + 2] = cpu_to_le32(objectid_to_release + 1);
set_sb_oid_cursize(rs, sb_oid_cursize(rs) + 2);
return;
}
i += 2;
}
reiserfs_error(s, "vs-15011", "tried to free free object id (%lu)",
(long unsigned)objectid_to_release);
}
/* Delete insert_num node pointers together with their left items
* and balance current node.*/
static void balance_internal_when_delete(struct tree_balance *tb,
int h, int child_pos)
{
int insert_num;
int n;
struct buffer_head *tbSh = PATH_H_PBUFFER(tb->tb_path, h);
struct buffer_info bi;
insert_num = tb->insert_size[h] / ((int)(DC_SIZE + KEY_SIZE));
/* delete child-node-pointer(s) together with their left item(s) */
bi.tb = tb;
bi.bi_bh = tbSh;
bi.bi_parent = PATH_H_PPARENT(tb->tb_path, h);
bi.bi_position = PATH_H_POSITION(tb->tb_path, h + 1);
internal_delete_childs(&bi, child_pos, -insert_num);
RFALSE(tb->blknum[h] > 1,
"tb->blknum[%d]=%d when insert_size < 0", h, tb->blknum[h]);
n = B_NR_ITEMS(tbSh);
if (tb->lnum[h] == 0 && tb->rnum[h] == 0) {
if (tb->blknum[h] == 0) {
/* node S[h] (root of the tree) is empty now */
struct buffer_head *new_root;
RFALSE(n
|| B_FREE_SPACE(tbSh) !=
MAX_CHILD_SIZE(tbSh) - DC_SIZE,
"buffer must have only 0 keys (%d)", n);
RFALSE(bi.bi_parent, "root has parent (%p)",
bi.bi_parent);
/* choose a new root */
if (!tb->L[h - 1] || !B_NR_ITEMS(tb->L[h - 1]))
new_root = tb->R[h - 1];
else
new_root = tb->L[h - 1];
/* switch super block's tree root block number to the new value */
PUT_SB_ROOT_BLOCK(tb->tb_sb, new_root->b_blocknr);
//REISERFS_SB(tb->tb_sb)->s_rs->s_tree_height --;
PUT_SB_TREE_HEIGHT(tb->tb_sb,
SB_TREE_HEIGHT(tb->tb_sb) - 1);
do_balance_mark_sb_dirty(tb,
REISERFS_SB(tb->tb_sb)->s_sbh,
1);
/*&&&&&&&&&&&&&&&&&&&&&& */
if (h > 1)
/* use check_internal if new root is an internal node */
check_internal(new_root);
/*&&&&&&&&&&&&&&&&&&&&&& */
/* do what is needed for buffer thrown from tree */
reiserfs_invalidate_buffer(tb, tbSh);
return;
}
return;
}
if (tb->L[h] && tb->lnum[h] == -B_NR_ITEMS(tb->L[h]) - 1) { /* join S[h] with L[h] */
RFALSE(tb->rnum[h] != 0,
"invalid tb->rnum[%d]==%d when joining S[h] with L[h]",
h, tb->rnum[h]);
internal_shift_left(INTERNAL_SHIFT_FROM_S_TO_L, tb, h, n + 1);
reiserfs_invalidate_buffer(tb, tbSh);
return;
}
if (tb->R[h] && tb->rnum[h] == -B_NR_ITEMS(tb->R[h]) - 1) { /* join S[h] with R[h] */
RFALSE(tb->lnum[h] != 0,
"invalid tb->lnum[%d]==%d when joining S[h] with R[h]",
h, tb->lnum[h]);
internal_shift_right(INTERNAL_SHIFT_FROM_S_TO_R, tb, h, n + 1);
reiserfs_invalidate_buffer(tb, tbSh);
return;
}
if (tb->lnum[h] < 0) { /* borrow from left neighbor L[h] */
RFALSE(tb->rnum[h] != 0,
"wrong tb->rnum[%d]==%d when borrow from L[h]", h,
tb->rnum[h]);
/*internal_shift_right (tb, h, tb->L[h], tb->CFL[h], tb->lkey[h], tb->S[h], -tb->lnum[h]); */
internal_shift_right(INTERNAL_SHIFT_FROM_L_TO_S, tb, h,
-tb->lnum[h]);
return;
}
if (tb->rnum[h] < 0) { /* borrow from right neighbor R[h] */
RFALSE(tb->lnum[h] != 0,
"invalid tb->lnum[%d]==%d when borrow from R[h]",
h, tb->lnum[h]);
internal_shift_left(INTERNAL_SHIFT_FROM_R_TO_S, tb, h, -tb->rnum[h]); /*tb->S[h], tb->CFR[h], tb->rkey[h], tb->R[h], -tb->rnum[h]); */
return;
}
if (tb->lnum[h] > 0) { /* split S[h] into two parts and put them into neighbors */
RFALSE(tb->rnum[h] == 0 || tb->lnum[h] + tb->rnum[h] != n + 1,
"invalid tb->lnum[%d]==%d or tb->rnum[%d]==%d when S[h](item number == %d) is split between them",
h, tb->lnum[h], h, tb->rnum[h], n);
internal_shift_left(INTERNAL_SHIFT_FROM_S_TO_L, tb, h, tb->lnum[h]); /*tb->L[h], tb->CFL[h], tb->lkey[h], tb->S[h], tb->lnum[h]); */
internal_shift_right(INTERNAL_SHIFT_FROM_S_TO_R, tb, h,
tb->rnum[h]);
reiserfs_invalidate_buffer(tb, tbSh);
return;
}
reiserfs_panic(tb->tb_sb, "ibalance-2",
"unexpected tb->lnum[%d]==%d or tb->rnum[%d]==%d",
h, tb->lnum[h], h, tb->rnum[h]);
}
int reiserfs_check_path(struct treepath *p)
{
RFALSE(p->path_length != ILLEGAL_PATH_ELEMENT_OFFSET,
"path not properly relsed");
return 0;
}
/* copy cpy_num node pointers and cpy_num - 1 items from buffer src to buffer dest
* last_first == FIRST_TO_LAST means, that we copy first items from src to tail of dest
* last_first == LAST_TO_FIRST means, that we copy last items from src to head of dest
*/
static void internal_copy_pointers_items(struct buffer_info *dest_bi,
struct buffer_head *src,
int last_first, int cpy_num)
{
/* ATTENTION! Number of node pointers in DEST is equal to number of items in DEST *
* as delimiting key have already inserted to buffer dest.*/
struct buffer_head *dest = dest_bi->bi_bh;
int nr_dest, nr_src;
int dest_order, src_order;
struct block_head *blkh;
struct reiserfs_key *key;
struct disk_child *dc;
nr_src = B_NR_ITEMS(src);
RFALSE(dest == NULL || src == NULL,
"src (%p) or dest (%p) buffer is 0", src, dest);
RFALSE(last_first != FIRST_TO_LAST && last_first != LAST_TO_FIRST,
"invalid last_first parameter (%d)", last_first);
RFALSE(nr_src < cpy_num - 1,
"no so many items (%d) in src (%d)", cpy_num, nr_src);
RFALSE(cpy_num < 0, "cpy_num less than 0 (%d)", cpy_num);
RFALSE(cpy_num - 1 + B_NR_ITEMS(dest) > (int)MAX_NR_KEY(dest),
"cpy_num (%d) + item number in dest (%d) can not be > MAX_NR_KEY(%d)",
cpy_num, B_NR_ITEMS(dest), MAX_NR_KEY(dest));
if (cpy_num == 0)
return;
/* coping */
blkh = B_BLK_HEAD(dest);
nr_dest = blkh_nr_item(blkh);
/*dest_order = (last_first == LAST_TO_FIRST) ? 0 : nr_dest; */
/*src_order = (last_first == LAST_TO_FIRST) ? (nr_src - cpy_num + 1) : 0; */
(last_first == LAST_TO_FIRST) ? (dest_order = 0, src_order =
nr_src - cpy_num + 1) : (dest_order =
nr_dest,
src_order =
0);
/* prepare space for cpy_num pointers */
dc = B_N_CHILD(dest, dest_order);
memmove(dc + cpy_num, dc, (nr_dest - dest_order) * DC_SIZE);
/* insert pointers */
memcpy(dc, B_N_CHILD(src, src_order), DC_SIZE * cpy_num);
/* prepare space for cpy_num - 1 item headers */
key = B_N_PDELIM_KEY(dest, dest_order);
memmove(key + cpy_num - 1, key,
KEY_SIZE * (nr_dest - dest_order) + DC_SIZE * (nr_dest +
cpy_num));
/* insert headers */
memcpy(key, B_N_PDELIM_KEY(src, src_order), KEY_SIZE * (cpy_num - 1));
/* sizes, item number */
set_blkh_nr_item(blkh, blkh_nr_item(blkh) + (cpy_num - 1));
set_blkh_free_space(blkh,
blkh_free_space(blkh) - (KEY_SIZE * (cpy_num - 1) +
DC_SIZE * cpy_num));
do_balance_mark_internal_dirty(dest_bi->tb, dest, 0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(dest);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
if (dest_bi->bi_parent) {
struct disk_child *t_dc;
t_dc = B_N_CHILD(dest_bi->bi_parent, dest_bi->bi_position);
put_dc_size(t_dc,
dc_size(t_dc) + (KEY_SIZE * (cpy_num - 1) +
DC_SIZE * cpy_num));
do_balance_mark_internal_dirty(dest_bi->tb, dest_bi->bi_parent,
0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(dest_bi->bi_parent);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
}
}
/* Delete del_num items and node pointers from buffer cur starting from *
* the first_i'th item and first_p'th pointers respectively. */
static void internal_delete_pointers_items(struct buffer_info *cur_bi,
int first_p,
int first_i, int del_num)
{
struct buffer_head *cur = cur_bi->bi_bh;
int nr;
struct block_head *blkh;
struct reiserfs_key *key;
struct disk_child *dc;
RFALSE(cur == NULL, "buffer is 0");
RFALSE(del_num < 0,
"negative number of items (%d) can not be deleted", del_num);
RFALSE(first_p < 0 || first_p + del_num > B_NR_ITEMS(cur) + 1
|| first_i < 0,
"first pointer order (%d) < 0 or "
"no so many pointers (%d), only (%d) or "
"first key order %d < 0", first_p, first_p + del_num,
B_NR_ITEMS(cur) + 1, first_i);
if (del_num == 0)
return;
blkh = B_BLK_HEAD(cur);
nr = blkh_nr_item(blkh);
if (first_p == 0 && del_num == nr + 1) {
RFALSE(first_i != 0,
"1st deleted key must have order 0, not %d", first_i);
make_empty_node(cur_bi);
return;
}
RFALSE(first_i + del_num > B_NR_ITEMS(cur),
"first_i = %d del_num = %d "
"no so many keys (%d) in the node (%b)(%z)",
first_i, del_num, first_i + del_num, cur, cur);
/* deleting */
dc = B_N_CHILD(cur, first_p);
memmove(dc, dc + del_num, (nr + 1 - first_p - del_num) * DC_SIZE);
key = B_N_PDELIM_KEY(cur, first_i);
memmove(key, key + del_num,
(nr - first_i - del_num) * KEY_SIZE + (nr + 1 -
del_num) * DC_SIZE);
/* sizes, item number */
set_blkh_nr_item(blkh, blkh_nr_item(blkh) - del_num);
set_blkh_free_space(blkh,
blkh_free_space(blkh) +
(del_num * (KEY_SIZE + DC_SIZE)));
do_balance_mark_internal_dirty(cur_bi->tb, cur, 0);
/*&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(cur);
/*&&&&&&&&&&&&&&&&&&&&&&& */
if (cur_bi->bi_parent) {
struct disk_child *t_dc;
t_dc = B_N_CHILD(cur_bi->bi_parent, cur_bi->bi_position);
put_dc_size(t_dc,
dc_size(t_dc) - (del_num * (KEY_SIZE + DC_SIZE)));
do_balance_mark_internal_dirty(cur_bi->tb, cur_bi->bi_parent,
0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(cur_bi->bi_parent);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
}
}
/* Insert count node pointers into buffer cur before position to + 1.
* Insert count items into buffer cur before position to.
* Items and node pointers are specified by inserted and bh respectively.
*/
static void internal_insert_childs(struct buffer_info *cur_bi,
int to, int count,
struct item_head *inserted,
struct buffer_head **bh)
{
struct buffer_head *cur = cur_bi->bi_bh;
struct block_head *blkh;
int nr;
struct reiserfs_key *ih;
struct disk_child new_dc[2];
struct disk_child *dc;
int i;
if (count <= 0)
return;
blkh = B_BLK_HEAD(cur);
nr = blkh_nr_item(blkh);
RFALSE(count > 2, "too many children (%d) are to be inserted", count);
RFALSE(B_FREE_SPACE(cur) < count * (KEY_SIZE + DC_SIZE),
"no enough free space (%d), needed %d bytes",
B_FREE_SPACE(cur), count * (KEY_SIZE + DC_SIZE));
/* prepare space for count disk_child */
dc = B_N_CHILD(cur, to + 1);
memmove(dc + count, dc, (nr + 1 - (to + 1)) * DC_SIZE);
/* copy to_be_insert disk children */
for (i = 0; i < count; i++) {
put_dc_size(&(new_dc[i]),
MAX_CHILD_SIZE(bh[i]) - B_FREE_SPACE(bh[i]));
put_dc_block_number(&(new_dc[i]), bh[i]->b_blocknr);
}
memcpy(dc, new_dc, DC_SIZE * count);
/* prepare space for count items */
ih = B_N_PDELIM_KEY(cur, ((to == -1) ? 0 : to));
memmove(ih + count, ih,
(nr - to) * KEY_SIZE + (nr + 1 + count) * DC_SIZE);
/* copy item headers (keys) */
memcpy(ih, inserted, KEY_SIZE);
if (count > 1)
memcpy(ih + 1, inserted + 1, KEY_SIZE);
/* sizes, item number */
set_blkh_nr_item(blkh, blkh_nr_item(blkh) + count);
set_blkh_free_space(blkh,
blkh_free_space(blkh) - count * (DC_SIZE +
KEY_SIZE));
do_balance_mark_internal_dirty(cur_bi->tb, cur, 0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(cur);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
if (cur_bi->bi_parent) {
struct disk_child *t_dc =
B_N_CHILD(cur_bi->bi_parent, cur_bi->bi_position);
put_dc_size(t_dc,
dc_size(t_dc) + (count * (DC_SIZE + KEY_SIZE)));
do_balance_mark_internal_dirty(cur_bi->tb, cur_bi->bi_parent,
0);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
check_internal(cur_bi->bi_parent);
/*&&&&&&&&&&&&&&&&&&&&&&&& */
}
}
/* Allocates blocks for a file to fulfil write request.
Maps all unmapped but prepared pages from the list.
Updates metadata with newly allocated blocknumbers as needed */
int reiserfs_allocate_blocks_for_region(
struct reiserfs_transaction_handle *th,
struct inode *inode, /* Inode we work with */
loff_t pos, /* Writing position */
int num_pages, /* number of pages write going
to touch */
int write_bytes, /* amount of bytes to write */
struct page **prepared_pages, /* array of
prepared pages
*/
int blocks_to_allocate /* Amount of blocks we
need to allocate to
fit the data into file
*/
)
{
struct cpu_key key; // cpu key of item that we are going to deal with
struct item_head *ih; // pointer to item head that we are going to deal with
struct buffer_head *bh; // Buffer head that contains items that we are going to deal with
__u32 * item; // pointer to item we are going to deal with
INITIALIZE_PATH(path); // path to item, that we are going to deal with.
b_blocknr_t *allocated_blocks; // Pointer to a place where allocated blocknumbers would be stored.
reiserfs_blocknr_hint_t hint; // hint structure for block allocator.
size_t res; // return value of various functions that we call.
int curr_block; // current block used to keep track of unmapped blocks.
int i; // loop counter
int itempos; // position in item
unsigned int from = (pos & (PAGE_CACHE_SIZE - 1)); // writing position in
// first page
unsigned int to = ((pos + write_bytes - 1) & (PAGE_CACHE_SIZE - 1)) + 1; /* last modified byte offset in last page */
__u64 hole_size ; // amount of blocks for a file hole, if it needed to be created.
int modifying_this_item = 0; // Flag for items traversal code to keep track
// of the fact that we already prepared
// current block for journal
int will_prealloc = 0;
RFALSE(!blocks_to_allocate, "green-9004: tried to allocate zero blocks?");
/* only preallocate if this is a small write */
if (REISERFS_I(inode)->i_prealloc_count ||
(!(write_bytes & (inode->i_sb->s_blocksize -1)) &&
blocks_to_allocate <
REISERFS_SB(inode->i_sb)->s_alloc_options.preallocsize))
will_prealloc = REISERFS_SB(inode->i_sb)->s_alloc_options.preallocsize;
allocated_blocks = kmalloc((blocks_to_allocate + will_prealloc) *
sizeof(b_blocknr_t), GFP_NOFS);
/* First we compose a key to point at the writing position, we want to do
that outside of any locking region. */
make_cpu_key (&key, inode, pos+1, TYPE_ANY, 3/*key length*/);
/* If we came here, it means we absolutely need to open a transaction,
since we need to allocate some blocks */
reiserfs_write_lock(inode->i_sb); // Journaling stuff and we need that.
journal_begin(th, inode->i_sb, JOURNAL_PER_BALANCE_CNT * 3 + 1); // Wish I know if this number enough
reiserfs_update_inode_transaction(inode) ;
/* Look for the in-tree position of our write, need path for block allocator */
res = search_for_position_by_key(inode->i_sb, &key, &path);
if ( res == IO_ERROR ) {
res = -EIO;
goto error_exit;
}
/* Allocate blocks */
/* First fill in "hint" structure for block allocator */
hint.th = th; // transaction handle.
hint.path = &path; // Path, so that block allocator can determine packing locality or whatever it needs to determine.
hint.inode = inode; // Inode is needed by block allocator too.
hint.search_start = 0; // We have no hint on where to search free blocks for block allocator.
hint.key = key.on_disk_key; // on disk key of file.
hint.block = inode->i_blocks>>(inode->i_sb->s_blocksize_bits-9); // Number of disk blocks this file occupies already.
hint.formatted_node = 0; // We are allocating blocks for unformatted node.
hint.preallocate = will_prealloc;
/* Call block allocator to allocate blocks */
res = reiserfs_allocate_blocknrs(&hint, allocated_blocks, blocks_to_allocate, blocks_to_allocate);
if ( res != CARRY_ON ) {
if ( res == NO_DISK_SPACE ) {
/* We flush the transaction in case of no space. This way some
blocks might become free */
SB_JOURNAL(inode->i_sb)->j_must_wait = 1;
restart_transaction(th, inode, &path);
/* We might have scheduled, so search again */
res = search_for_position_by_key(inode->i_sb, &key, &path);
if ( res == IO_ERROR ) {
res = -EIO;
goto error_exit;
}
/* update changed info for hint structure. */
res = reiserfs_allocate_blocknrs(&hint, allocated_blocks, blocks_to_allocate, blocks_to_allocate);
if ( res != CARRY_ON ) {
res = -ENOSPC;
pathrelse(&path);
goto error_exit;
}
} else {
res = -ENOSPC;
pathrelse(&path);
goto error_exit;
}
}
#ifdef __BIG_ENDIAN
// Too bad, I have not found any way to convert a given region from
// cpu format to little endian format
{
int i;
for ( i = 0; i < blocks_to_allocate ; i++)
allocated_blocks[i]=cpu_to_le32(allocated_blocks[i]);
}
#endif
/* Blocks allocating well might have scheduled and tree might have changed,
let's search the tree again */
/* find where in the tree our write should go */
res = search_for_position_by_key(inode->i_sb, &key, &path);
if ( res == IO_ERROR ) {
res = -EIO;
goto error_exit_free_blocks;
}
bh = get_last_bh( &path ); // Get a bufferhead for last element in path.
ih = get_ih( &path ); // Get a pointer to last item head in path.
item = get_item( &path ); // Get a pointer to last item in path
/* Let's see what we have found */
if ( res != POSITION_FOUND ) { /* position not found, this means that we
might need to append file with holes
first */
// Since we are writing past the file's end, we need to find out if
// there is a hole that needs to be inserted before our writing
// position, and how many blocks it is going to cover (we need to
// populate pointers to file blocks representing the hole with zeros)
{
int item_offset = 1;
/*
* if ih is stat data, its offset is 0 and we don't want to
* add 1 to pos in the hole_size calculation
*/
if (is_statdata_le_ih(ih))
item_offset = 0;
hole_size = (pos + item_offset -
(le_key_k_offset( get_inode_item_key_version(inode),
&(ih->ih_key)) +
op_bytes_number(ih, inode->i_sb->s_blocksize))) >>
inode->i_sb->s_blocksize_bits;
}
/* path points to first direct item of the file regarless of how many of
them are there */
int direct2indirect (struct reiserfs_transaction_handle *th, struct inode * inode,
struct path * path, struct buffer_head * unbh,
loff_t tail_offset)
{
struct super_block * sb = inode->i_sb;
struct buffer_head *up_to_date_bh ;
struct item_head * p_le_ih = PATH_PITEM_HEAD (path);
unsigned long total_tail = 0 ;
struct cpu_key end_key; /* Key to search for the last byte of the
converted item. */
struct item_head ind_ih; /* new indirect item to be inserted or
key of unfm pointer to be pasted */
int n_blk_size,
n_retval; /* returned value for reiserfs_insert_item and clones */
struct unfm_nodeinfo unfm_ptr; /* Handle on an unformatted node
that will be inserted in the
tree. */
sb->u.reiserfs_sb.s_direct2indirect ++;
n_blk_size = sb->s_blocksize;
/* and key to search for append or insert pointer to the new
unformatted node. */
copy_item_head (&ind_ih, p_le_ih);
set_le_ih_k_offset (&ind_ih, tail_offset);
set_le_ih_k_type (&ind_ih, TYPE_INDIRECT);
/* Set the key to search for the place for new unfm pointer */
make_cpu_key (&end_key, inode, tail_offset, TYPE_INDIRECT, 4);
// FIXME: we could avoid this
if ( search_for_position_by_key (sb, &end_key, path) == POSITION_FOUND ) {
reiserfs_warning ("PAP-14030: direct2indirect: "
"pasted or inserted byte exists in the tree %K. "
"Use fsck to repair.\n", &end_key);
pathrelse(path);
return -EIO;
}
p_le_ih = PATH_PITEM_HEAD (path);
unfm_ptr.unfm_nodenum = cpu_to_le32 (unbh->b_blocknr);
unfm_ptr.unfm_freespace = 0; // ???
if ( is_statdata_le_ih (p_le_ih) ) {
/* Insert new indirect item. */
set_ih_free_space (&ind_ih, 0); /* delete at nearest future */
put_ih_item_len( &ind_ih, UNFM_P_SIZE );
PATH_LAST_POSITION (path)++;
n_retval = reiserfs_insert_item (th, path, &end_key, &ind_ih,
(char *)&unfm_ptr);
} else {
/* Paste into last indirect item of an object. */
n_retval = reiserfs_paste_into_item(th, path, &end_key,
(char *)&unfm_ptr, UNFM_P_SIZE);
}
if ( n_retval ) {
return n_retval;
}
// note: from here there are two keys which have matching first
// three key components. They only differ by the fourth one.
/* Set the key to search for the direct items of the file */
make_cpu_key (&end_key, inode, max_reiserfs_offset (inode), TYPE_DIRECT, 4);
/* Move bytes from the direct items to the new unformatted node
and delete them. */
while (1) {
int tail_size;
/* end_key.k_offset is set so, that we will always have found
last item of the file */
if ( search_for_position_by_key (sb, &end_key, path) == POSITION_FOUND )
reiserfs_panic (sb, "PAP-14050: direct2indirect: "
"direct item (%K) not found", &end_key);
p_le_ih = PATH_PITEM_HEAD (path);
RFALSE( !is_direct_le_ih (p_le_ih),
"vs-14055: direct item expected(%K), found %h",
&end_key, p_le_ih);
tail_size = (le_ih_k_offset (p_le_ih) & (n_blk_size - 1))
+ ih_item_len(p_le_ih) - 1;
/* we only send the unbh pointer if the buffer is not up to date.
** this avoids overwriting good data from writepage() with old data
** from the disk or buffer cache
** Special case: unbh->b_page will be NULL if we are coming through
** DIRECT_IO handler here.
*/
if ( !unbh->b_page || buffer_uptodate(unbh) || Page_Uptodate(unbh->b_page)) {
up_to_date_bh = NULL ;
} else {
up_to_date_bh = unbh ;
}
n_retval = reiserfs_delete_item (th, path, &end_key, inode,
up_to_date_bh) ;
total_tail += n_retval ;
if (tail_size == n_retval)
// done: file does not have direct items anymore
break;
}
/* if we've copied bytes from disk into the page, we need to zero
** out the unused part of the block (it was not up to date before)
** the page is still kmapped (by whoever called reiserfs_get_block)
*/
if (up_to_date_bh) {
unsigned pgoff = (tail_offset + total_tail - 1) & (PAGE_CACHE_SIZE - 1);
memset(page_address(unbh->b_page) + pgoff, 0, n_blk_size - total_tail) ;
}
inode->u.reiserfs_i.i_first_direct_byte = U32_MAX;
reiserfs_update_tail_transaction(inode);
return 0;
}
/*
* process, that is going to call fix_nodes/do_balance must hold only
* one path. If it holds 2 or more, it can get into endless waiting in
* get_empty_nodes or its clones
*/
static int reiserfs_rename(struct inode *old_dir, struct dentry *old_dentry,
struct inode *new_dir, struct dentry *new_dentry)
{
int retval;
INITIALIZE_PATH(old_entry_path);
INITIALIZE_PATH(new_entry_path);
INITIALIZE_PATH(dot_dot_entry_path);
struct item_head new_entry_ih, old_entry_ih, dot_dot_ih;
struct reiserfs_dir_entry old_de, new_de, dot_dot_de;
struct inode *old_inode, *new_dentry_inode;
struct reiserfs_transaction_handle th;
int jbegin_count;
umode_t old_inode_mode;
unsigned long savelink = 1;
struct timespec ctime;
/* three balancings: (1) old name removal, (2) new name insertion
and (3) maybe "save" link insertion
stat data updates: (1) old directory,
(2) new directory and (3) maybe old object stat data (when it is
directory) and (4) maybe stat data of object to which new entry
pointed initially and (5) maybe block containing ".." of
renamed directory
quota updates: two parent directories */
jbegin_count =
JOURNAL_PER_BALANCE_CNT * 3 + 5 +
4 * REISERFS_QUOTA_TRANS_BLOCKS(old_dir->i_sb);
dquot_initialize(old_dir);
dquot_initialize(new_dir);
old_inode = old_dentry->d_inode;
new_dentry_inode = new_dentry->d_inode;
// make sure, that oldname still exists and points to an object we
// are going to rename
old_de.de_gen_number_bit_string = NULL;
reiserfs_write_lock(old_dir->i_sb);
retval =
reiserfs_find_entry(old_dir, old_dentry->d_name.name,
old_dentry->d_name.len, &old_entry_path,
&old_de);
pathrelse(&old_entry_path);
if (retval == IO_ERROR) {
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
if (retval != NAME_FOUND || old_de.de_objectid != old_inode->i_ino) {
reiserfs_write_unlock(old_dir->i_sb);
return -ENOENT;
}
old_inode_mode = old_inode->i_mode;
if (S_ISDIR(old_inode_mode)) {
// make sure, that directory being renamed has correct ".."
// and that its new parent directory has not too many links
// already
if (new_dentry_inode) {
if (!reiserfs_empty_dir(new_dentry_inode)) {
reiserfs_write_unlock(old_dir->i_sb);
return -ENOTEMPTY;
}
}
/* directory is renamed, its parent directory will be changed,
** so find ".." entry
*/
dot_dot_de.de_gen_number_bit_string = NULL;
retval =
reiserfs_find_entry(old_inode, "..", 2, &dot_dot_entry_path,
&dot_dot_de);
pathrelse(&dot_dot_entry_path);
if (retval != NAME_FOUND) {
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
/* inode number of .. must equal old_dir->i_ino */
if (dot_dot_de.de_objectid != old_dir->i_ino) {
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
}
retval = journal_begin(&th, old_dir->i_sb, jbegin_count);
if (retval) {
reiserfs_write_unlock(old_dir->i_sb);
return retval;
}
/* add new entry (or find the existing one) */
retval =
reiserfs_add_entry(&th, new_dir, new_dentry->d_name.name,
new_dentry->d_name.len, old_inode, 0);
if (retval == -EEXIST) {
if (!new_dentry_inode) {
reiserfs_panic(old_dir->i_sb, "vs-7050",
"new entry is found, new inode == 0");
}
} else if (retval) {
int err = journal_end(&th, old_dir->i_sb, jbegin_count);
reiserfs_write_unlock(old_dir->i_sb);
return err ? err : retval;
}
reiserfs_update_inode_transaction(old_dir);
reiserfs_update_inode_transaction(new_dir);
/* this makes it so an fsync on an open fd for the old name will
** commit the rename operation
*/
reiserfs_update_inode_transaction(old_inode);
if (new_dentry_inode)
reiserfs_update_inode_transaction(new_dentry_inode);
while (1) {
// look for old name using corresponding entry key (found by reiserfs_find_entry)
if ((retval =
search_by_entry_key(new_dir->i_sb, &old_de.de_entry_key,
&old_entry_path,
&old_de)) != NAME_FOUND) {
pathrelse(&old_entry_path);
journal_end(&th, old_dir->i_sb, jbegin_count);
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
copy_item_head(&old_entry_ih, get_ih(&old_entry_path));
reiserfs_prepare_for_journal(old_inode->i_sb, old_de.de_bh, 1);
// look for new name by reiserfs_find_entry
new_de.de_gen_number_bit_string = NULL;
retval =
reiserfs_find_entry(new_dir, new_dentry->d_name.name,
new_dentry->d_name.len, &new_entry_path,
&new_de);
// reiserfs_add_entry should not return IO_ERROR, because it is called with essentially same parameters from
// reiserfs_add_entry above, and we'll catch any i/o errors before we get here.
if (retval != NAME_FOUND_INVISIBLE && retval != NAME_FOUND) {
pathrelse(&new_entry_path);
pathrelse(&old_entry_path);
journal_end(&th, old_dir->i_sb, jbegin_count);
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
copy_item_head(&new_entry_ih, get_ih(&new_entry_path));
reiserfs_prepare_for_journal(old_inode->i_sb, new_de.de_bh, 1);
if (S_ISDIR(old_inode->i_mode)) {
if ((retval =
search_by_entry_key(new_dir->i_sb,
&dot_dot_de.de_entry_key,
&dot_dot_entry_path,
&dot_dot_de)) != NAME_FOUND) {
pathrelse(&dot_dot_entry_path);
pathrelse(&new_entry_path);
pathrelse(&old_entry_path);
journal_end(&th, old_dir->i_sb, jbegin_count);
reiserfs_write_unlock(old_dir->i_sb);
return -EIO;
}
copy_item_head(&dot_dot_ih,
get_ih(&dot_dot_entry_path));
// node containing ".." gets into transaction
reiserfs_prepare_for_journal(old_inode->i_sb,
dot_dot_de.de_bh, 1);
}
/* we should check seals here, not do
this stuff, yes? Then, having
gathered everything into RAM we
should lock the buffers, yes? -Hans */
/* probably. our rename needs to hold more
** than one path at once. The seals would
** have to be written to deal with multi-path
** issues -chris
*/
/* sanity checking before doing the rename - avoid races many
** of the above checks could have scheduled. We have to be
** sure our items haven't been shifted by another process.
*/
if (item_moved(&new_entry_ih, &new_entry_path) ||
!entry_points_to_object(new_dentry->d_name.name,
new_dentry->d_name.len,
&new_de, new_dentry_inode) ||
item_moved(&old_entry_ih, &old_entry_path) ||
!entry_points_to_object(old_dentry->d_name.name,
old_dentry->d_name.len,
&old_de, old_inode)) {
reiserfs_restore_prepared_buffer(old_inode->i_sb,
new_de.de_bh);
reiserfs_restore_prepared_buffer(old_inode->i_sb,
old_de.de_bh);
if (S_ISDIR(old_inode_mode))
reiserfs_restore_prepared_buffer(old_inode->
i_sb,
dot_dot_de.
de_bh);
continue;
}
if (S_ISDIR(old_inode_mode)) {
if (item_moved(&dot_dot_ih, &dot_dot_entry_path) ||
!entry_points_to_object("..", 2, &dot_dot_de,
old_dir)) {
reiserfs_restore_prepared_buffer(old_inode->
i_sb,
old_de.de_bh);
reiserfs_restore_prepared_buffer(old_inode->
i_sb,
new_de.de_bh);
reiserfs_restore_prepared_buffer(old_inode->
i_sb,
dot_dot_de.
de_bh);
continue;
}
}
RFALSE(S_ISDIR(old_inode_mode) &&
!buffer_journal_prepared(dot_dot_de.de_bh), "");
break;
}
/* ok, all the changes can be done in one fell swoop when we
have claimed all the buffers needed. */
mark_de_visible(new_de.de_deh + new_de.de_entry_num);
set_ino_in_dir_entry(&new_de, INODE_PKEY(old_inode));
journal_mark_dirty(&th, old_dir->i_sb, new_de.de_bh);
mark_de_hidden(old_de.de_deh + old_de.de_entry_num);
journal_mark_dirty(&th, old_dir->i_sb, old_de.de_bh);
ctime = CURRENT_TIME_SEC;
old_dir->i_ctime = old_dir->i_mtime = ctime;
new_dir->i_ctime = new_dir->i_mtime = ctime;
/* thanks to Alex Adriaanse <*****@*****.**> for patch which adds ctime update of
renamed object */
old_inode->i_ctime = ctime;
if (new_dentry_inode) {
// adjust link number of the victim
if (S_ISDIR(new_dentry_inode->i_mode)) {
clear_nlink(new_dentry_inode);
} else {
drop_nlink(new_dentry_inode);
}
new_dentry_inode->i_ctime = ctime;
savelink = new_dentry_inode->i_nlink;
}
if (S_ISDIR(old_inode_mode)) {
/* adjust ".." of renamed directory */
set_ino_in_dir_entry(&dot_dot_de, INODE_PKEY(new_dir));
journal_mark_dirty(&th, new_dir->i_sb, dot_dot_de.de_bh);
if (!new_dentry_inode)
/* there (in new_dir) was no directory, so it got new link
(".." of renamed directory) */
INC_DIR_INODE_NLINK(new_dir);
/* old directory lost one link - ".. " of renamed directory */
DEC_DIR_INODE_NLINK(old_dir);
}
// looks like in 2.3.99pre3 brelse is atomic. so we can use pathrelse
pathrelse(&new_entry_path);
pathrelse(&dot_dot_entry_path);
// FIXME: this reiserfs_cut_from_item's return value may screw up
// anybody, but it will panic if will not be able to find the
// entry. This needs one more clean up
if (reiserfs_cut_from_item
(&th, &old_entry_path, &(old_de.de_entry_key), old_dir, NULL,
0) < 0)
reiserfs_error(old_dir->i_sb, "vs-7060",
"couldn't not cut old name. Fsck later?");
old_dir->i_size -= DEH_SIZE + old_de.de_entrylen;
reiserfs_update_sd(&th, old_dir);
reiserfs_update_sd(&th, new_dir);
reiserfs_update_sd(&th, old_inode);
if (new_dentry_inode) {
if (savelink == 0)
add_save_link(&th, new_dentry_inode,
0 /* not truncate */ );
reiserfs_update_sd(&th, new_dentry_inode);
}
retval = journal_end(&th, old_dir->i_sb, jbegin_count);
reiserfs_write_unlock(old_dir->i_sb);
return retval;
}
int reiserfs_readdir_dentry(struct dentry *dentry, void *dirent,
filldir_t filldir, loff_t *pos)
{
struct inode *inode = dentry->d_inode;
struct cpu_key pos_key;
INITIALIZE_PATH(path_to_entry);
struct buffer_head *bh;
int item_num, entry_num;
const struct reiserfs_key *rkey;
struct item_head *ih, tmp_ih;
int search_res;
char *local_buf;
loff_t next_pos;
char small_buf[32];
struct reiserfs_dir_entry de;
int ret = 0;
reiserfs_write_lock(inode->i_sb);
reiserfs_check_lock_depth(inode->i_sb, "readdir");
make_cpu_key(&pos_key, inode, *pos ?: DOT_OFFSET, TYPE_DIRENTRY, 3);
next_pos = cpu_key_k_offset(&pos_key);
path_to_entry.reada = PATH_READA;
while (1) {
research:
search_res =
search_by_entry_key(inode->i_sb, &pos_key, &path_to_entry,
&de);
if (search_res == IO_ERROR) {
ret = -EIO;
goto out;
}
entry_num = de.de_entry_num;
bh = de.de_bh;
item_num = de.de_item_num;
ih = de.de_ih;
store_ih(&tmp_ih, ih);
RFALSE(COMP_SHORT_KEYS(&(ih->ih_key), &pos_key),
"vs-9000: found item %h does not match to dir we readdir %K",
ih, &pos_key);
RFALSE(item_num > B_NR_ITEMS(bh) - 1,
"vs-9005 item_num == %d, item amount == %d",
item_num, B_NR_ITEMS(bh));
RFALSE(I_ENTRY_COUNT(ih) < entry_num,
"vs-9010: entry number is too big %d (%d)",
entry_num, I_ENTRY_COUNT(ih));
if (search_res == POSITION_FOUND
|| entry_num < I_ENTRY_COUNT(ih)) {
struct reiserfs_de_head *deh =
B_I_DEH(bh, ih) + entry_num;
for (; entry_num < I_ENTRY_COUNT(ih);
entry_num++, deh++) {
int d_reclen;
char *d_name;
off_t d_off;
ino_t d_ino;
if (!de_visible(deh))
continue;
d_reclen = entry_length(bh, ih, entry_num);
d_name = B_I_DEH_ENTRY_FILE_NAME(bh, ih, deh);
if (d_reclen <= 0 ||
d_name + d_reclen > bh->b_data + bh->b_size) {
pathrelse(&path_to_entry);
ret = -EIO;
goto out;
}
if (!d_name[d_reclen - 1])
d_reclen = strlen(d_name);
if (d_reclen >
REISERFS_MAX_NAME(inode->i_sb->
s_blocksize)) {
continue;
}
if (is_privroot_deh(dentry, deh))
continue;
d_off = deh_offset(deh);
*pos = d_off;
d_ino = deh_objectid(deh);
if (d_reclen <= 32) {
local_buf = small_buf;
} else {
local_buf = kmalloc(d_reclen,
GFP_NOFS);
if (!local_buf) {
pathrelse(&path_to_entry);
ret = -ENOMEM;
goto out;
}
if (item_moved(&tmp_ih, &path_to_entry)) {
kfree(local_buf);
goto research;
}
}
memcpy(local_buf, d_name, d_reclen);
reiserfs_write_unlock(inode->i_sb);
if (filldir
(dirent, local_buf, d_reclen, d_off, d_ino,
DT_UNKNOWN) < 0) {
reiserfs_write_lock(inode->i_sb);
if (local_buf != small_buf) {
kfree(local_buf);
}
goto end;
}
reiserfs_write_lock(inode->i_sb);
if (local_buf != small_buf) {
kfree(local_buf);
}
next_pos = deh_offset(deh) + 1;
if (item_moved(&tmp_ih, &path_to_entry)) {
goto research;
}
}
}
if (item_num != B_NR_ITEMS(bh) - 1)
goto end;
rkey = get_rkey(&path_to_entry, inode->i_sb);
if (!comp_le_keys(rkey, &MIN_KEY)) {
set_cpu_key_k_offset(&pos_key, next_pos);
continue;
}
if (COMP_SHORT_KEYS(rkey, &pos_key)) {
goto end;
}
set_cpu_key_k_offset(&pos_key,
le_key_k_offset(KEY_FORMAT_3_5, rkey));
}
end:
*pos = next_pos;
pathrelse(&path_to_entry);
reiserfs_check_path(&path_to_entry);
out:
reiserfs_write_unlock(inode->i_sb);
return ret;
}
static int reiserfs_readdir (struct file * filp, void * dirent, filldir_t filldir)
{
struct inode *inode = filp->f_dentry->d_inode;
struct cpu_key pos_key; /* key of current position in the directory (key of directory entry) */
INITIALIZE_PATH (path_to_entry);
struct buffer_head * bh;
int item_num, entry_num;
const struct key * rkey;
struct item_head * ih, tmp_ih;
int search_res;
char * local_buf;
loff_t next_pos;
char small_buf[32] ; /* avoid kmalloc if we can */
struct reiserfs_dir_entry de;
reiserfs_check_lock_depth("readdir") ;
/* form key for search the next directory entry using f_pos field of
file structure */
make_cpu_key (&pos_key, inode, (filp->f_pos) ? (filp->f_pos) : DOT_OFFSET,
TYPE_DIRENTRY, 3);
next_pos = cpu_key_k_offset (&pos_key);
/* reiserfs_warning ("reiserfs_readdir 1: f_pos = %Ld\n", filp->f_pos);*/
while (1) {
research:
/* search the directory item, containing entry with specified key */
search_res = search_by_entry_key (inode->i_sb, &pos_key, &path_to_entry, &de);
if (search_res == IO_ERROR) {
// FIXME: we could just skip part of directory which could
// not be read
return -EIO;
}
entry_num = de.de_entry_num;
bh = de.de_bh;
item_num = de.de_item_num;
ih = de.de_ih;
store_ih (&tmp_ih, ih);
/* we must have found item, that is item of this directory, */
RFALSE( COMP_SHORT_KEYS (&(ih->ih_key), &pos_key),
"vs-9000: found item %h does not match to dir we readdir %K",
ih, &pos_key);
RFALSE( item_num > B_NR_ITEMS (bh) - 1,
"vs-9005 item_num == %d, item amount == %d",
item_num, B_NR_ITEMS (bh));
/* and entry must be not more than number of entries in the item */
RFALSE( I_ENTRY_COUNT (ih) < entry_num,
"vs-9010: entry number is too big %d (%d)",
entry_num, I_ENTRY_COUNT (ih));
if (search_res == POSITION_FOUND || entry_num < I_ENTRY_COUNT (ih)) {
/* go through all entries in the directory item beginning from the entry, that has been found */
struct reiserfs_de_head * deh = B_I_DEH (bh, ih) + entry_num;
for (; entry_num < I_ENTRY_COUNT (ih); entry_num ++, deh ++) {
int d_reclen;
char * d_name;
off_t d_off;
ino_t d_ino;
if (!de_visible (deh))
/* it is hidden entry */
continue;
d_reclen = entry_length (bh, ih, entry_num);
d_name = B_I_DEH_ENTRY_FILE_NAME (bh, ih, deh);
if (!d_name[d_reclen - 1])
d_reclen = strlen (d_name);
if (d_reclen > REISERFS_MAX_NAME_LEN(inode->i_sb->s_blocksize)){
/* too big to send back to VFS */
continue ;
}
d_off = deh_offset (deh);
filp->f_pos = d_off ;
d_ino = deh_objectid (deh);
if (d_reclen <= 32) {
local_buf = small_buf ;
} else {
local_buf = reiserfs_kmalloc(d_reclen, GFP_NOFS, inode->i_sb) ;
if (!local_buf) {
pathrelse (&path_to_entry);
return -ENOMEM ;
}
if (item_moved (&tmp_ih, &path_to_entry)) {
reiserfs_kfree(local_buf, d_reclen, inode->i_sb) ;
goto research;
}
}
// Note, that we copy name to user space via temporary
// buffer (local_buf) because filldir will block if
// user space buffer is swapped out. At that time
// entry can move to somewhere else
memcpy (local_buf, d_name, d_reclen);
if (filldir (dirent, local_buf, d_reclen, d_off, d_ino,
DT_UNKNOWN) < 0) {
if (local_buf != small_buf) {
reiserfs_kfree(local_buf, d_reclen, inode->i_sb) ;
}
goto end;
}
if (local_buf != small_buf) {
reiserfs_kfree(local_buf, d_reclen, inode->i_sb) ;
}
// next entry should be looked for with such offset
next_pos = deh_offset (deh) + 1;
if (item_moved (&tmp_ih, &path_to_entry)) {
goto research;
}
} /* for */
}
if (item_num != B_NR_ITEMS (bh) - 1)
// end of directory has been reached
goto end;
/* item we went through is last item of node. Using right
delimiting key check is it directory end */
rkey = get_rkey (&path_to_entry, inode->i_sb);
if (! comp_le_keys (rkey, &MIN_KEY)) {
/* set pos_key to key, that is the smallest and greater
that key of the last entry in the item */
set_cpu_key_k_offset (&pos_key, next_pos);
continue;
}
if ( COMP_SHORT_KEYS (rkey, &pos_key)) {
// end of directory has been reached
goto end;
}
/* directory continues in the right neighboring block */
set_cpu_key_k_offset (&pos_key, le_key_k_offset (KEY_FORMAT_3_5, rkey));
} /* while */
end:
// FIXME: ext2_readdir does not reset f_pos
filp->f_pos = next_pos;
pathrelse (&path_to_entry);
reiserfs_check_path(&path_to_entry) ;
UPDATE_ATIME(inode) ;
return 0;
}
static int linear_search_in_dir_item(struct cpu_key *key,
struct reiserfs_dir_entry *de,
const char *name, int namelen)
{
struct reiserfs_de_head *deh = de->de_deh;
int retval;
int i;
i = de->de_entry_num;
if (i == I_ENTRY_COUNT(de->de_ih) ||
GET_HASH_VALUE(deh_offset(deh + i)) !=
GET_HASH_VALUE(cpu_key_k_offset(key))) {
i--;
}
RFALSE(de->de_deh != B_I_DEH(de->de_bh, de->de_ih),
"vs-7010: array of entry headers not found");
deh += i;
for (; i >= 0; i--, deh--) {
if (GET_HASH_VALUE(deh_offset(deh)) !=
GET_HASH_VALUE(cpu_key_k_offset(key))) {
// hash value does not match, no need to check whole name
return NAME_NOT_FOUND;
}
/* mark, that this generation number is used */
if (de->de_gen_number_bit_string)
set_bit(GET_GENERATION_NUMBER(deh_offset(deh)),
de->de_gen_number_bit_string);
// calculate pointer to name and namelen
de->de_entry_num = i;
set_de_name_and_namelen(de);
if ((retval =
reiserfs_match(de, name, namelen)) != NAME_NOT_FOUND) {
// de's de_name, de_namelen, de_recordlen are set. Fill the rest:
// key of pointed object
set_de_object_key(de);
store_de_entry_key(de);
// retval can be NAME_FOUND or NAME_FOUND_INVISIBLE
return retval;
}
}
if (GET_GENERATION_NUMBER(le_ih_k_offset(de->de_ih)) == 0)
/* we have reached left most entry in the node. In common we
have to go to the left neighbor, but if generation counter
is 0 already, we know for sure, that there is no name with
the same hash value */
// FIXME: this work correctly only because hash value can not
// be 0. Btw, in case of Yura's hash it is probably possible,
// so, this is a bug
return NAME_NOT_FOUND;
RFALSE(de->de_item_num,
"vs-7015: two diritems of the same directory in one node?");
return GOTO_PREVIOUS_ITEM;
}
