/*
* We play buffer_head aliasing tricks to write data/metadata blocks to
* the journal without copying their contents, but for journal
* descriptor blocks we do need to generate bona fide buffers.
*
* After the caller of journal_get_descriptor_buffer() has finished modifying
* the buffer's contents they really should run flush_dcache_page(bh->b_page).
* But we don't bother doing that, so there will be coherency problems with
* mmaps of blockdevs which hold live JBD-controlled filesystems.
*/
struct journal_head *journal_get_descriptor_buffer(journal_t *journal)
{
struct buffer_head *bh;
unsigned long blocknr;
int err;
err = journal_next_log_block(journal, &blocknr);
if (err)
return NULL;
bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize);
lock_buffer(bh);
memset(bh->b_data, 0, journal->j_blocksize);
set_buffer_uptodate(bh);
unlock_buffer(bh);
BUFFER_TRACE(bh, "return this buffer");
return journal_add_journal_head(bh);
}
/**
* nilfs_forget_buffer - discard dirty state
* @inode: owner inode of the buffer
* @bh: buffer head of the buffer to be discarded
*/
void nilfs_forget_buffer(struct buffer_head *bh)
{
struct page *page = bh->b_page;
const unsigned long clear_bits =
(1 << BH_Uptodate | 1 << BH_Dirty | 1 << BH_Mapped |
1 << BH_Async_Write | 1 << BH_NILFS_Volatile |
1 << BH_NILFS_Checked | 1 << BH_NILFS_Redirected);
lock_buffer(bh);
set_mask_bits(&bh->b_state, clear_bits, 0);
if (nilfs_page_buffers_clean(page))
__nilfs_clear_page_dirty(page);
bh->b_blocknr = -1;
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
unlock_buffer(bh);
brelse(bh);
}
static void copy_sample_data(running_machine &machine, const INT16 *data, int bytes_to_copy)
{
void *buffer1, *buffer2 = (void *)NULL;
long length1, length2;
int cur_bytes;
// attempt to lock the stream buffer
if (lock_buffer(machine, stream_buffer_in, bytes_to_copy, &buffer1, &length1, &buffer2, &length2) < 0)
{
buffer_underflows++;
return;
}
// adjust the input pointer
stream_buffer_in += bytes_to_copy;
if (stream_buffer_in >= stream_buffer_size)
{
stream_buffer_in -= stream_buffer_size;
stream_loop = 1;
if (LOG_SOUND)
fprintf(sound_log, "stream_loop set to 1 (stream_buffer_in looped)\n");
}
// copy the first chunk
cur_bytes = (bytes_to_copy > length1) ? length1 : bytes_to_copy;
att_memcpy(buffer1, data, cur_bytes);
// adjust for the number of bytes
bytes_to_copy -= cur_bytes;
data = (INT16 *)((UINT8 *)data + cur_bytes);
// copy the second chunk
if (bytes_to_copy != 0)
{
cur_bytes = (bytes_to_copy > length2) ? length2 : bytes_to_copy;
att_memcpy(buffer2, data, cur_bytes);
}
// unlock
unlock_buffer();
}
/**
* nilfs_forget_buffer - discard dirty state
* @inode: owner inode of the buffer
* @bh: buffer head of the buffer to be discarded
*/
void nilfs_forget_buffer(struct buffer_head *bh)
{
struct page *page = bh->b_page;
lock_buffer(bh);
clear_buffer_nilfs_volatile(bh);
clear_buffer_nilfs_checked(bh);
clear_buffer_nilfs_redirected(bh);
clear_buffer_dirty(bh);
if (nilfs_page_buffers_clean(page))
__nilfs_clear_page_dirty(page);
clear_buffer_uptodate(bh);
clear_buffer_mapped(bh);
bh->b_blocknr = -1;
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
unlock_buffer(bh);
brelse(bh);
}
void gfs2_meta_wipe(struct gfs2_inode *ip, u64 bstart, u32 blen)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct buffer_head *bh;
while (blen) {
bh = gfs2_getbuf(ip->i_gl, bstart, NO_CREATE);
if (bh) {
lock_buffer(bh);
gfs2_log_lock(sdp);
gfs2_remove_from_journal(bh, current->journal_info, 1);
gfs2_log_unlock(sdp);
unlock_buffer(bh);
brelse(bh);
}
bstart++;
blen--;
}
}
int nilfs_mark_inode_dirty(struct inode *inode)
{
struct nilfs_sb_info *sbi = NILFS_SB(inode->i_sb);
struct buffer_head *ibh;
int err;
err = nilfs_load_inode_block(sbi, inode, &ibh);
if (unlikely(err)) {
nilfs_warning(inode->i_sb, __func__,
"failed to reget inode block.\n");
return err;
}
lock_buffer(ibh);
nilfs_update_inode(inode, ibh);
unlock_buffer(ibh);
nilfs_mdt_mark_buffer_dirty(ibh);
nilfs_mdt_mark_dirty(sbi->s_ifile);
brelse(ibh);
return 0;
}
static void gfs2_discard(struct gfs2_sbd *sdp, struct buffer_head *bh)
{
struct gfs2_bufdata *bd;
lock_buffer(bh);
gfs2_log_lock(sdp);
clear_buffer_dirty(bh);
bd = bh->b_private;
if (bd) {
if (!list_empty(&bd->bd_le.le_list) && !buffer_pinned(bh))
list_del_init(&bd->bd_le.le_list);
else
gfs2_remove_from_journal(bh, current->journal_info, 0);
}
bh->b_bdev = NULL;
clear_buffer_mapped(bh);
clear_buffer_req(bh);
clear_buffer_new(bh);
gfs2_log_unlock(sdp);
unlock_buffer(bh);
}
void reiserfs_unmap_buffer(struct buffer_head *bh)
{
lock_buffer(bh);
if (buffer_journaled(bh) || buffer_journal_dirty(bh)) {
BUG();
}
clear_buffer_dirty(bh);
if ((!list_empty(&bh->b_assoc_buffers) || bh->b_private) && bh->b_page) {
struct inode *inode = bh->b_page->mapping->host;
struct reiserfs_journal *j = SB_JOURNAL(inode->i_sb);
spin_lock(&j->j_dirty_buffers_lock);
list_del_init(&bh->b_assoc_buffers);
reiserfs_free_jh(bh);
spin_unlock(&j->j_dirty_buffers_lock);
}
clear_buffer_mapped(bh);
clear_buffer_req(bh);
clear_buffer_new(bh);
bh->b_bdev = NULL;
unlock_buffer(bh);
}
static void discard_buffer(struct gfs2_sbd *sdp, struct buffer_head *bh)
{
struct gfs2_bufdata *bd;
gfs2_log_lock(sdp);
bd = bh->b_private;
if (bd) {
bd->bd_bh = NULL;
bh->b_private = NULL;
}
gfs2_log_unlock(sdp);
lock_buffer(bh);
clear_buffer_dirty(bh);
bh->b_bdev = NULL;
clear_buffer_mapped(bh);
clear_buffer_req(bh);
clear_buffer_new(bh);
clear_buffer_delay(bh);
unlock_buffer(bh);
}
void add_packet(packet_buffer* pb, msg* m) {
lock_buffer(pb);
if (m->pdu->seq <= pb->largest_seq) {
unlock_buffer(pb);
debug(I, "Packet has already been buffered.\n");
return;
}
if (m->pdu->seq > pb->largest_seq)
pb->largest_seq = m->pdu->seq;
if ((m->pdu->seq % pb->size == pb->oldest_seq % pb->size)
&& (m->pdu->seq != pb->oldest_seq)){
if((pb->buffer)[pb->oldest_seq % pb->size]){
free((pb->buffer)[pb->oldest_seq % pb->size]->pdu);
free((pb->buffer)[pb->oldest_seq % pb->size]);
}
pb->oldest_seq++;
}
(pb->buffer)[m->pdu->seq % pb->size] = m;
unlock_buffer(pb);
}
static int alloc_branch(struct inode *inode,
int num,
int *offsets,
Indirect *branch)
{
int n = 0;
int i;
int parent = minix_new_block(inode);
printk("parent new block %d\n",parent);
branch[0].key = cpu_to_block(parent);
if (parent) for (n = 1; n < num; n++) {
struct buffer_head *bh;
/* Allocate the next block */
int nr = minix_new_block(inode);
printk("loop new block %d",nr);
if (!nr)
break;
branch[n].key = cpu_to_block(nr);
bh = sb_getblk(inode->i_sb, parent);
lock_buffer(bh);
memset(bh->b_data, 0, bh->b_size);
branch[n].bh = bh;
branch[n].p = (block_t*) bh->b_data + offsets[n];
*branch[n].p = branch[n].key;
set_buffer_uptodate(bh);
unlock_buffer(bh);
mark_buffer_dirty_inode(bh, inode);
parent = nr;
}
if (n == num)
return 0;
/* Allocation failed, free what we already allocated */
for (i = 1; i < n; i++)
bforget(branch[i].bh);
for (i = 0; i < n; i++)
minix_free_block(inode, block_to_cpu(branch[i].key));
return -ENOSPC;
}
/**
* nilfs_clear_dirty_page - discard dirty page
* @page: dirty page that will be discarded
* @silent: suppress [true] or print [false] warning messages
*/
void nilfs_clear_dirty_page(struct page *page, bool silent)
{
struct inode *inode = page->mapping->host;
struct super_block *sb = inode->i_sb;
BUG_ON(!PageLocked(page));
if (!silent) {
nilfs_warning(sb, __func__,
"discard page: offset %lld, ino %lu",
page_offset(page), inode->i_ino);
}
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
if (page_has_buffers(page)) {
struct buffer_head *bh, *head;
bh = head = page_buffers(page);
do {
lock_buffer(bh);
if (!silent) {
nilfs_warning(sb, __func__,
"discard block %llu, size %zu",
(u64)bh->b_blocknr, bh->b_size);
}
clear_buffer_async_write(bh);
clear_buffer_dirty(bh);
clear_buffer_nilfs_volatile(bh);
clear_buffer_nilfs_checked(bh);
clear_buffer_nilfs_redirected(bh);
clear_buffer_uptodate(bh);
clear_buffer_mapped(bh);
unlock_buffer(bh);
} while (bh = bh->b_this_page, bh != head);
}
__nilfs_clear_page_dirty(page);
}
/*
* Write the MMP block using WRITE_SYNC to try to get the block on-disk
* faster.
*/
static int write_mmp_block(struct super_block *sb, struct buffer_head *bh)
{
struct mmp_struct *mmp = (struct mmp_struct *)(bh->b_data);
/*
* We protect against freezing so that we don't create dirty buffers
* on frozen filesystem.
*/
sb_start_write(sb);
ext4_mmp_csum_set(sb, mmp);
mark_buffer_dirty(bh);
lock_buffer(bh);
bh->b_end_io = end_buffer_write_sync;
get_bh(bh);
submit_bh(WRITE_SYNC | REQ_META | REQ_PRIO, bh);
wait_on_buffer(bh);
sb_end_write(sb);
if (unlikely(!buffer_uptodate(bh)))
return 1;
return 0;
}
static int __f2fs_commit_super(struct f2fs_sb_info *sbi, int block)
{
struct f2fs_super_block *super = F2FS_RAW_SUPER(sbi);
struct buffer_head *bh;
int err;
bh = sb_getblk(sbi->sb, block);
if (!bh)
return -EIO;
lock_buffer(bh);
memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
set_buffer_uptodate(bh);
set_buffer_dirty(bh);
unlock_buffer(bh);
/* it's rare case, we can do fua all the time */
err = __sync_dirty_buffer(bh, WRITE_FLUSH_FUA);
brelse(bh);
return err;
}
/*
* Read the MMP block. It _must_ be read from disk and hence we clear the
* uptodate flag on the buffer.
*/
static int read_mmp_block(struct super_block *sb, struct buffer_head **bh,
ext4_fsblk_t mmp_block)
{
struct mmp_struct *mmp;
if (*bh)
clear_buffer_uptodate(*bh);
/* This would be sb_bread(sb, mmp_block), except we need to be sure
* that the MD RAID device cache has been bypassed, and that the read
* is not blocked in the elevator. */
if (!*bh)
*bh = sb_getblk(sb, mmp_block);
if (*bh) {
get_bh(*bh);
lock_buffer(*bh);
(*bh)->b_end_io = end_buffer_read_sync;
#ifdef FEATURE_STORAGE_META_LOG
set_metadata_rw_status((*bh)->b_bdev->bd_disk->first_minor, WAIT_READ_CNT);
#endif
submit_bh(READ_SYNC, *bh);
wait_on_buffer(*bh);
if (!buffer_uptodate(*bh)) {
brelse(*bh);
*bh = NULL;
}
}
if (!*bh) {
ext4_warning(sb, "Error while reading MMP block %llu",
mmp_block);
return -EIO;
}
mmp = (struct mmp_struct *)((*bh)->b_data);
if (le32_to_cpu(mmp->mmp_magic) != EXT4_MMP_MAGIC)
return -EINVAL;
return 0;
}
int sys_sync()
{
extern void sync_inode();
struct buffer *bh;
sync_inode();
for (bh = buffer_table; bh < buffer_table + NR_BUFFER; bh++) {
lock_buffer(bh);
if (bh->b_flag & B_DIRTY){
/*
* buffer shold be unlock in write_block.
*/
rw_block(WRITE_BUF, bh);
irq_lock();
while(bh->b_lock.pid)
sleep_on(&(bh->b_lock.wait));
irq_unlock();
}
unlock_buffer(bh);
}
return 0;
}
/*
* Read the MMP block. It _must_ be read from disk and hence we clear the
* uptodate flag on the buffer.
*/
static int read_mmp_block(struct super_block *sb, struct buffer_head **bh,
ext4_fsblk_t mmp_block)
{
struct mmp_struct *mmp;
if (*bh)
clear_buffer_uptodate(*bh);
/* This would be sb_bread(sb, mmp_block), except we need to be sure
* that the MD RAID device cache has been bypassed, and that the read
* is not blocked in the elevator. */
if (!*bh)
*bh = sb_getblk(sb, mmp_block);
if (!*bh)
return -ENOMEM;
if (*bh) {
get_bh(*bh);
lock_buffer(*bh);
(*bh)->b_end_io = end_buffer_read_sync;
submit_bh(READ_SYNC | REQ_META | REQ_PRIO, *bh);
wait_on_buffer(*bh);
if (!buffer_uptodate(*bh)) {
brelse(*bh);
*bh = NULL;
}
}
if (unlikely(!*bh)) {
ext4_warning(sb, "Error while reading MMP block %llu",
mmp_block);
return -EIO;
}
mmp = (struct mmp_struct *)((*bh)->b_data);
if (le32_to_cpu(mmp->mmp_magic) != EXT4_MMP_MAGIC ||
!ext4_mmp_csum_verify(sb, mmp))
return -EINVAL;
return 0;
}
void nilfs_clear_dirty_pages(struct address_space *mapping)
{
struct pagevec pvec;
unsigned int i;
pgoff_t index = 0;
pagevec_init(&pvec, 0);
while (pagevec_lookup_tag(&pvec, mapping, &index, PAGECACHE_TAG_DIRTY,
PAGEVEC_SIZE)) {
for (i = 0; i < pagevec_count(&pvec); i++) {
struct page *page = pvec.pages[i];
struct buffer_head *bh, *head;
lock_page(page);
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
bh = head = page_buffers(page);
do {
lock_buffer(bh);
clear_buffer_async_write(bh);
clear_buffer_dirty(bh);
clear_buffer_nilfs_volatile(bh);
clear_buffer_nilfs_checked(bh);
clear_buffer_nilfs_redirected(bh);
clear_buffer_uptodate(bh);
clear_buffer_mapped(bh);
unlock_buffer(bh);
bh = bh->b_this_page;
} while (bh != head);
__nilfs_clear_page_dirty(page);
unlock_page(page);
}
pagevec_release(&pvec);
cond_resched();
}
}
void journal_invalidatepage(journal_t *journal,
struct page *page,
unsigned long offset)
{
struct buffer_head *head, *bh, *next;
unsigned int curr_off = 0;
int may_free = 1;
if (!PageLocked(page))
BUG();
if (!page_has_buffers(page))
return;
/* We will potentially be playing with lists other than just the
* data lists (especially for journaled data mode), so be
* cautious in our locking. */
head = bh = page_buffers(page);
do {
unsigned int next_off = curr_off + bh->b_size;
next = bh->b_this_page;
if (offset <= curr_off) {
/* This block is wholly outside the truncation point */
lock_buffer(bh);
may_free &= journal_unmap_buffer(journal, bh);
unlock_buffer(bh);
}
curr_off = next_off;
bh = next;
} while (bh != head);
if (!offset) {
if (may_free && try_to_free_buffers(page))
J_ASSERT(!page_has_buffers(page));
}
}
/**
* nilfs_clear_dirty_page - discard dirty page
* @page: dirty page that will be discarded
* @silent: suppress [true] or print [false] warning messages
*/
void nilfs_clear_dirty_page(struct page *page, bool silent)
{
struct inode *inode = page->mapping->host;
struct super_block *sb = inode->i_sb;
BUG_ON(!PageLocked(page));
if (!silent) {
nilfs_warning(sb, __func__,
"discard page: offset %lld, ino %lu",
page_offset(page), inode->i_ino);
}
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
if (page_has_buffers(page)) {
struct buffer_head *bh, *head;
const unsigned long clear_bits =
(1 << BH_Uptodate | 1 << BH_Dirty | 1 << BH_Mapped |
1 << BH_Async_Write | 1 << BH_NILFS_Volatile |
1 << BH_NILFS_Checked | 1 << BH_NILFS_Redirected);
bh = head = page_buffers(page);
do {
lock_buffer(bh);
if (!silent) {
nilfs_warning(sb, __func__,
"discard block %llu, size %zu",
(u64)bh->b_blocknr, bh->b_size);
}
set_mask_bits(&bh->b_state, clear_bits, 0);
unlock_buffer(bh);
} while (bh = bh->b_this_page, bh != head);
}
__nilfs_clear_page_dirty(page);
}
/* stolen from fs/buffer.c */
void reiserfs_unmap_buffer(struct buffer_head *bh) {
lock_buffer(bh) ;
if (buffer_journaled(bh) || buffer_journal_dirty(bh)) {
BUG() ;
}
clear_buffer_dirty(bh) ;
/* Remove the buffer from whatever list it belongs to. We are mostly
interested in removing it from per-sb j_dirty_buffers list, to avoid
BUG() on attempt to write not mapped buffer */
if ( (!list_empty(&bh->b_assoc_buffers) || bh->b_private) && bh->b_page) {
struct inode *inode = bh->b_page->mapping->host;
struct reiserfs_journal *j = SB_JOURNAL(inode->i_sb);
spin_lock(&j->j_dirty_buffers_lock);
list_del_init(&bh->b_assoc_buffers);
reiserfs_free_jh(bh);
spin_unlock(&j->j_dirty_buffers_lock);
}
clear_buffer_mapped(bh) ;
clear_buffer_req(bh) ;
clear_buffer_new(bh);
bh->b_bdev = NULL;
unlock_buffer(bh) ;
}
/**
* nilfs_clear_dirty_page - discard dirty page
* @page: dirty page that will be discarded
* @silent: suppress [true] or print [false] warning messages
*/
void nilfs_clear_dirty_page(struct page *page, bool silent)
{
struct inode *inode = page->mapping->host;
struct super_block *sb = inode->i_sb;
BUG_ON(!PageLocked(page));
if (!silent)
nilfs_msg(sb, KERN_WARNING,
"discard dirty page: offset=%lld, ino=%lu",
page_offset(page), inode->i_ino);
ClearPageUptodate(page);
ClearPageMappedToDisk(page);
if (page_has_buffers(page)) {
struct buffer_head *bh, *head;
const unsigned long clear_bits =
(BIT(BH_Uptodate) | BIT(BH_Dirty) | BIT(BH_Mapped) |
BIT(BH_Async_Write) | BIT(BH_NILFS_Volatile) |
BIT(BH_NILFS_Checked) | BIT(BH_NILFS_Redirected));
bh = head = page_buffers(page);
do {
lock_buffer(bh);
if (!silent)
nilfs_msg(sb, KERN_WARNING,
"discard dirty block: blocknr=%llu, size=%zu",
(u64)bh->b_blocknr, bh->b_size);
set_mask_bits(&bh->b_state, clear_bits, 0);
unlock_buffer(bh);
} while (bh = bh->b_this_page, bh != head);
}
__nilfs_clear_page_dirty(page);
}
/*
* Write super block and backups doesn't need to collaborate with journal,
* so we don't need to lock ip_io_mutex and ci doesn't need to bea passed
* into this function.
*/
int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
struct buffer_head *bh)
{
int ret = 0;
struct ocfs2_dinode *di = (struct ocfs2_dinode *)bh->b_data;
BUG_ON(buffer_jbd(bh));
ocfs2_check_super_or_backup(osb->sb, bh->b_blocknr);
if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb)) {
ret = -EROFS;
mlog_errno(ret);
goto out;
}
lock_buffer(bh);
set_buffer_uptodate(bh);
/* remove from dirty list before I/O. */
clear_buffer_dirty(bh);
get_bh(bh); /* for end_buffer_write_sync() */
bh->b_end_io = end_buffer_write_sync;
ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &di->i_check);
submit_bh(REQ_OP_WRITE, 0, bh);
wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
ret = -EIO;
mlog_errno(ret);
}
out:
return ret;
}
/*
* Write super block and backups doesn't need to collaborate with journal,
* so we don't need to lock ip_io_mutex and inode doesn't need to bea passed
* into this function.
*/
int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
struct buffer_head *bh)
{
int ret = 0;
mlog_entry_void();
BUG_ON(buffer_jbd(bh));
ocfs2_check_super_or_backup(osb->sb, bh->b_blocknr);
if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb)) {
ret = -EROFS;
goto out;
}
lock_buffer(bh);
set_buffer_uptodate(bh);
/* remove from dirty list before I/O. */
clear_buffer_dirty(bh);
get_bh(bh); /* for end_buffer_write_sync() */
bh->b_end_io = end_buffer_write_sync;
submit_bh(WRITE, bh);
wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
ret = -EIO;
put_bh(bh);
}
out:
mlog_exit(ret);
return ret;
}
int nilfs_btnode_submit_block(struct address_space *btnc, __u64 blocknr,
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);
if (unlikely(!bh))
return -ENOMEM;
err = -EEXIST;
page = bh->b_page;
if (buffer_uptodate(bh) || buffer_dirty(bh))
goto found;
if (pblocknr == 0) {
pblocknr = blocknr;
if (inode->i_ino != NILFS_DAT_INO) {
struct the_nilfs *nilfs = inode->i_sb->s_fs_info;
err = nilfs_dat_translate(nilfs->ns_dat, blocknr,
&pblocknr);
if (unlikely(err)) {
brelse(bh);
goto out_locked;
}
}
}
if (mode == READA) {
if (pblocknr != *submit_ptr + 1 || !trylock_buffer(bh)) {
err = -EBUSY;
brelse(bh);
goto out_locked;
}
} else {
lock_buffer(bh);
}
if (buffer_uptodate(bh)) {
unlock_buffer(bh);
err = -EEXIST;
goto found;
}
set_buffer_mapped(bh);
bh->b_bdev = inode->i_sb->s_bdev;
bh->b_blocknr = pblocknr;
bh->b_end_io = end_buffer_read_sync;
get_bh(bh);
submit_bh(mode, bh);
bh->b_blocknr = blocknr;
*submit_ptr = pblocknr;
err = 0;
found:
*pbh = bh;
out_locked:
unlock_page(page);
page_cache_release(page);
return err;
}
/**
* ext4_read_block_bitmap()
* @sb: super block
* @block_group: given block group
*
* 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 *
ext4_read_block_bitmap(struct super_block *sb, ext4_group_t block_group)
{
struct ext4_group_desc *desc;
struct buffer_head *bh = NULL;
ext4_fsblk_t bitmap_blk;
desc = ext4_get_group_desc(sb, block_group, NULL);
if (!desc)
return NULL;
bitmap_blk = ext4_block_bitmap(sb, desc);
bh = sb_getblk(sb, bitmap_blk);
if (unlikely(!bh)) {
ext4_error(sb, __func__,
"Cannot read block bitmap - "
"block_group = %u, block_bitmap = %llu",
block_group, bitmap_blk);
return NULL;
}
if (bitmap_uptodate(bh))
return bh;
lock_buffer(bh);
if (bitmap_uptodate(bh)) {
unlock_buffer(bh);
return bh;
}
ext4_lock_group(sb, block_group);
if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT)) {
ext4_init_block_bitmap(sb, bh, block_group, desc);
set_bitmap_uptodate(bh);
set_buffer_uptodate(bh);
ext4_unlock_group(sb, block_group);
unlock_buffer(bh);
return bh;
}
ext4_unlock_group(sb, block_group);
if (buffer_uptodate(bh)) {
/*
* if not uninit if bh is uptodate,
* bitmap is also uptodate
*/
set_bitmap_uptodate(bh);
unlock_buffer(bh);
return bh;
}
/*
* submit the buffer_head for read. We can
* safely mark the bitmap as uptodate now.
* We do it here so the bitmap uptodate bit
* get set with buffer lock held.
*/
set_bitmap_uptodate(bh);
if (bh_submit_read(bh) < 0) {
put_bh(bh);
ext4_error(sb, __func__,
"Cannot read block bitmap - "
"block_group = %u, block_bitmap = %llu",
block_group, bitmap_blk);
return NULL;
}
ext4_valid_block_bitmap(sb, desc, block_group, bh);
/*
* file system mounted not to panic on error,
* continue with corrupt bitmap
*/
return bh;
}
int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
struct buffer_head *bhs[], int flags,
int (*validate)(struct super_block *sb,
struct buffer_head *bh))
{
int status = 0;
int i, ignore_cache = 0;
struct buffer_head *bh;
mlog_entry("(inode=%p, block=(%llu), nr=(%d), flags=%d)\n",
inode, (unsigned long long)block, nr, flags);
BUG_ON(!inode);
BUG_ON((flags & OCFS2_BH_READAHEAD) &&
(flags & OCFS2_BH_IGNORE_CACHE));
if (bhs == NULL) {
status = -EINVAL;
mlog_errno(status);
goto bail;
}
if (nr < 0) {
mlog(ML_ERROR, "asked to read %d blocks!\n", nr);
status = -EINVAL;
mlog_errno(status);
goto bail;
}
if (nr == 0) {
mlog(ML_BH_IO, "No buffers will be read!\n");
status = 0;
goto bail;
}
mutex_lock(&OCFS2_I(inode)->ip_io_mutex);
for (i = 0 ; i < nr ; i++) {
if (bhs[i] == NULL) {
bhs[i] = sb_getblk(inode->i_sb, block++);
if (bhs[i] == NULL) {
mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
status = -EIO;
mlog_errno(status);
goto bail;
}
}
bh = bhs[i];
ignore_cache = (flags & OCFS2_BH_IGNORE_CACHE);
/* There are three read-ahead cases here which we need to
* be concerned with. All three assume a buffer has
* previously been submitted with OCFS2_BH_READAHEAD
* and it hasn't yet completed I/O.
*
* 1) The current request is sync to disk. This rarely
* happens these days, and never when performance
* matters - the code can just wait on the buffer
* lock and re-submit.
*
* 2) The current request is cached, but not
* readahead. ocfs2_buffer_uptodate() will return
* false anyway, so we'll wind up waiting on the
* buffer lock to do I/O. We re-check the request
* with after getting the lock to avoid a re-submit.
*
* 3) The current request is readahead (and so must
* also be a caching one). We short circuit if the
* buffer is locked (under I/O) and if it's in the
* uptodate cache. The re-check from #2 catches the
* case that the previous read-ahead completes just
* before our is-it-in-flight check.
*/
if (!ignore_cache && !ocfs2_buffer_uptodate(inode, bh)) {
mlog(ML_UPTODATE,
"bh (%llu), inode %llu not uptodate\n",
(unsigned long long)bh->b_blocknr,
(unsigned long long)OCFS2_I(inode)->ip_blkno);
/* We're using ignore_cache here to say
* "go to disk" */
ignore_cache = 1;
}
if (buffer_jbd(bh)) {
if (ignore_cache)
mlog(ML_BH_IO, "trying to sync read a jbd "
"managed bh (blocknr = %llu)\n",
(unsigned long long)bh->b_blocknr);
continue;
}
if (ignore_cache) {
if (buffer_dirty(bh)) {
/* This should probably be a BUG, or
* at least return an error. */
mlog(ML_BH_IO, "asking me to sync read a dirty "
"buffer! (blocknr = %llu)\n",
(unsigned long long)bh->b_blocknr);
continue;
}
/* A read-ahead request was made - if the
* buffer is already under read-ahead from a
* previously submitted request than we are
* done here. */
if ((flags & OCFS2_BH_READAHEAD)
&& ocfs2_buffer_read_ahead(inode, bh))
continue;
lock_buffer(bh);
if (buffer_jbd(bh)) {
#ifdef CATCH_BH_JBD_RACES
mlog(ML_ERROR, "block %llu had the JBD bit set "
"while I was in lock_buffer!",
(unsigned long long)bh->b_blocknr);
BUG();
#else
unlock_buffer(bh);
continue;
#endif
}
/* Re-check ocfs2_buffer_uptodate() as a
* previously read-ahead buffer may have
* completed I/O while we were waiting for the
* buffer lock. */
if (!(flags & OCFS2_BH_IGNORE_CACHE)
&& !(flags & OCFS2_BH_READAHEAD)
&& ocfs2_buffer_uptodate(inode, bh)) {
unlock_buffer(bh);
continue;
}
clear_buffer_uptodate(bh);
get_bh(bh); /* for end_buffer_read_sync() */
if (validate)
set_buffer_needs_validate(bh);
bh->b_end_io = end_buffer_read_sync;
submit_bh(READ, bh);
continue;
}
}
status = 0;
for (i = (nr - 1); i >= 0; i--) {
bh = bhs[i];
if (!(flags & OCFS2_BH_READAHEAD)) {
/* We know this can't have changed as we hold the
* inode sem. Avoid doing any work on the bh if the
* journal has it. */
if (!buffer_jbd(bh))
wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
/* Status won't be cleared from here on out,
* so we can safely record this and loop back
* to cleanup the other buffers. Don't need to
* remove the clustered uptodate information
* for this bh as it's not marked locally
* uptodate. */
status = -EIO;
put_bh(bh);
bhs[i] = NULL;
continue;
}
if (buffer_needs_validate(bh)) {
/* We never set NeedsValidate if the
* buffer was held by the journal, so
* that better not have changed */
BUG_ON(buffer_jbd(bh));
clear_buffer_needs_validate(bh);
status = validate(inode->i_sb, bh);
if (status) {
put_bh(bh);
bhs[i] = NULL;
continue;
}
}
}
/* Always set the buffer in the cache, even if it was
* a forced read, or read-ahead which hasn't yet
* completed. */
ocfs2_set_buffer_uptodate(inode, bh);
}
mutex_unlock(&OCFS2_I(inode)->ip_io_mutex);
mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n",
(unsigned long long)block, nr,
((flags & OCFS2_BH_IGNORE_CACHE) || ignore_cache) ? "no" : "yes",
flags);
bail:
mlog_exit(status);
return status;
}
int ocfs2_read_blocks_sync(struct ocfs2_super *osb, u64 block,
unsigned int nr, struct buffer_head *bhs[])
{
int status = 0;
unsigned int i;
struct buffer_head *bh;
if (!nr) {
mlog(ML_BH_IO, "No buffers will be read!\n");
goto bail;
}
for (i = 0 ; i < nr ; i++) {
if (bhs[i] == NULL) {
bhs[i] = sb_getblk(osb->sb, block++);
if (bhs[i] == NULL) {
status = -EIO;
mlog_errno(status);
goto bail;
}
}
bh = bhs[i];
if (buffer_jbd(bh)) {
mlog(ML_BH_IO,
"trying to sync read a jbd "
"managed bh (blocknr = %llu), skipping\n",
(unsigned long long)bh->b_blocknr);
continue;
}
if (buffer_dirty(bh)) {
/* This should probably be a BUG, or
* at least return an error. */
mlog(ML_ERROR,
"trying to sync read a dirty "
"buffer! (blocknr = %llu), skipping\n",
(unsigned long long)bh->b_blocknr);
continue;
}
lock_buffer(bh);
if (buffer_jbd(bh)) {
mlog(ML_ERROR,
"block %llu had the JBD bit set "
"while I was in lock_buffer!",
(unsigned long long)bh->b_blocknr);
BUG();
}
clear_buffer_uptodate(bh);
get_bh(bh); /* for end_buffer_read_sync() */
bh->b_end_io = end_buffer_read_sync;
submit_bh(READ, bh);
}
for (i = nr; i > 0; i--) {
bh = bhs[i - 1];
/* No need to wait on the buffer if it's managed by JBD. */
if (!buffer_jbd(bh))
wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
/* Status won't be cleared from here on out,
* so we can safely record this and loop back
* to cleanup the other buffers. */
status = -EIO;
put_bh(bh);
bhs[i - 1] = NULL;
}
}
bail:
return status;
}
static int do_one_pass(journal_t *journal,
struct recovery_info *info, enum passtype pass)
{
unsigned int first_commit_ID, next_commit_ID;
unsigned long next_log_block;
int err, success = 0;
journal_superblock_t * sb;
journal_header_t * tmp;
struct buffer_head * bh;
unsigned int sequence;
int blocktype;
/* Precompute the maximum metadata descriptors in a descriptor block */
int MAX_BLOCKS_PER_DESC;
MAX_BLOCKS_PER_DESC = ((journal->j_blocksize-sizeof(journal_header_t))
/ sizeof(journal_block_tag_t));
/*
* First thing is to establish what we expect to find in the log
* (in terms of transaction IDs), and where (in terms of log
* block offsets): query the superblock.
*/
sb = journal->j_superblock;
next_commit_ID = be32_to_cpu(sb->s_sequence);
next_log_block = be32_to_cpu(sb->s_start);
first_commit_ID = next_commit_ID;
if (pass == PASS_SCAN)
info->start_transaction = first_commit_ID;
jbd_debug(1, "Starting recovery pass %d\n", pass);
/*
* Now we walk through the log, transaction by transaction,
* making sure that each transaction has a commit block in the
* expected place. Each complete transaction gets replayed back
* into the main filesystem.
*/
while (1) {
int flags;
char * tagp;
journal_block_tag_t * tag;
struct buffer_head * obh;
struct buffer_head * nbh;
cond_resched(); /* We're under lock_kernel() */
/* If we already know where to stop the log traversal,
* check right now that we haven't gone past the end of
* the log. */
if (pass != PASS_SCAN)
if (tid_geq(next_commit_ID, info->end_transaction))
break;
jbd_debug(2, "Scanning for sequence ID %u at %lu/%lu\n",
next_commit_ID, next_log_block, journal->j_last);
/* Skip over each chunk of the transaction looking
* either the next descriptor block or the final commit
* record. */
jbd_debug(3, "JBD: checking block %ld\n", next_log_block);
err = jread(&bh, journal, next_log_block);
if (err)
goto failed;
next_log_block++;
wrap(journal, next_log_block);
/* What kind of buffer is it?
*
* If it is a descriptor block, check that it has the
* expected sequence number. Otherwise, we're all done
* here. */
tmp = (journal_header_t *)bh->b_data;
if (tmp->h_magic != cpu_to_be32(JFS_MAGIC_NUMBER)) {
brelse(bh);
break;
}
blocktype = be32_to_cpu(tmp->h_blocktype);
sequence = be32_to_cpu(tmp->h_sequence);
jbd_debug(3, "Found magic %d, sequence %d\n",
blocktype, sequence);
if (sequence != next_commit_ID) {
brelse(bh);
break;
}
/* OK, we have a valid descriptor block which matches
* all of the sequence number checks. What are we going
* to do with it? That depends on the pass... */
switch(blocktype) {
case JFS_DESCRIPTOR_BLOCK:
/* If it is a valid descriptor block, replay it
* in pass REPLAY; otherwise, just skip over the
* blocks it describes. */
if (pass != PASS_REPLAY) {
next_log_block +=
count_tags(bh, journal->j_blocksize);
wrap(journal, next_log_block);
brelse(bh);
continue;
}
/* A descriptor block: we can now write all of
* the data blocks. Yay, useful work is finally
* getting done here! */
tagp = &bh->b_data[sizeof(journal_header_t)];
while ((tagp - bh->b_data +sizeof(journal_block_tag_t))
<= journal->j_blocksize) {
unsigned long io_block;
tag = (journal_block_tag_t *) tagp;
flags = be32_to_cpu(tag->t_flags);
io_block = next_log_block++;
wrap(journal, next_log_block);
err = jread(&obh, journal, io_block);
if (err) {
/* Recover what we can, but
* report failure at the end. */
success = err;
printk (KERN_ERR
"JBD: IO error %d recovering "
"block %ld in log\n",
err, io_block);
} else {
unsigned long blocknr;
J_ASSERT(obh != NULL);
blocknr = be32_to_cpu(tag->t_blocknr);
/* If the block has been
* revoked, then we're all done
* here. */
if (journal_test_revoke
(journal, blocknr,
next_commit_ID)) {
brelse(obh);
++info->nr_revoke_hits;
goto skip_write;
}
/* Find a buffer for the new
* data being restored */
nbh = __getblk(journal->j_fs_dev,
blocknr,
journal->j_blocksize);
if (nbh == NULL) {
printk(KERN_ERR
"JBD: Out of memory "
"during recovery.\n");
err = -ENOMEM;
brelse(bh);
brelse(obh);
goto failed;
}
lock_buffer(nbh);
memcpy(nbh->b_data, obh->b_data,
journal->j_blocksize);
if (flags & JFS_FLAG_ESCAPE) {
*((__be32 *)bh->b_data) =
cpu_to_be32(JFS_MAGIC_NUMBER);
}
BUFFER_TRACE(nbh, "marking dirty");
set_buffer_uptodate(nbh);
mark_buffer_dirty(nbh);
BUFFER_TRACE(nbh, "marking uptodate");
++info->nr_replays;
/* ll_rw_block(WRITE, 1, &nbh); */
unlock_buffer(nbh);
brelse(obh);
brelse(nbh);
}
skip_write:
tagp += sizeof(journal_block_tag_t);
if (!(flags & JFS_FLAG_SAME_UUID))
tagp += 16;
if (flags & JFS_FLAG_LAST_TAG)
break;
}
brelse(bh);
continue;
case JFS_COMMIT_BLOCK:
/* Found an expected commit block: not much to
* do other than move on to the next sequence
* number. */
brelse(bh);
next_commit_ID++;
continue;
case JFS_REVOKE_BLOCK:
/* If we aren't in the REVOKE pass, then we can
* just skip over this block. */
if (pass != PASS_REVOKE) {
brelse(bh);
continue;
}
err = scan_revoke_records(journal, bh,
next_commit_ID, info);
brelse(bh);
if (err)
goto failed;
continue;
default:
jbd_debug(3, "Unrecognised magic %d, end of scan.\n",
blocktype);
goto done;
}
}
done:
/*
* We broke out of the log scan loop: either we came to the
* known end of the log or we found an unexpected block in the
* log. If the latter happened, then we know that the "current"
* transaction marks the end of the valid log.
*/
if (pass == PASS_SCAN)
info->end_transaction = next_commit_ID;
else {
/* It's really bad news if different passes end up at
* different places (but possible due to IO errors). */
if (info->end_transaction != next_commit_ID) {
printk (KERN_ERR "JBD: recovery pass %d ended at "
"transaction %u, expected %u\n",
pass, next_commit_ID, info->end_transaction);
if (!success)
success = -EIO;
}
}
return success;
failed:
return err;
}
/*
* journal_commit_transaction
*
* The primary function for committing a transaction to the log. This
* function is called by the journal thread to begin a complete commit.
*/
void journal_commit_transaction(journal_t *journal)
{
transaction_t *commit_transaction;
struct journal_head *jh, *new_jh, *descriptor;
struct buffer_head **wbuf = journal->j_wbuf;
int bufs;
int flags;
int err;
unsigned int blocknr;
ktime_t start_time;
u64 commit_time;
char *tagp = NULL;
journal_header_t *header;
journal_block_tag_t *tag = NULL;
int space_left = 0;
int first_tag = 0;
int tag_flag;
int i;
struct blk_plug plug;
/*
* First job: lock down the current transaction and wait for
* all outstanding updates to complete.
*/
/* Do we need to erase the effects of a prior journal_flush? */
if (journal->j_flags & JFS_FLUSHED) {
jbd_debug(3, "super block updated\n");
journal_update_superblock(journal, 1);
} else {
jbd_debug(3, "superblock not updated\n");
}
J_ASSERT(journal->j_running_transaction != NULL);
J_ASSERT(journal->j_committing_transaction == NULL);
commit_transaction = journal->j_running_transaction;
J_ASSERT(commit_transaction->t_state == T_RUNNING);
trace_jbd_start_commit(journal, commit_transaction);
jbd_debug(1, "JBD: starting commit of transaction %d\n",
commit_transaction->t_tid);
spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_LOCKED;
trace_jbd_commit_locking(journal, commit_transaction);
spin_lock(&commit_transaction->t_handle_lock);
while (commit_transaction->t_updates) {
DEFINE_WAIT(wait);
prepare_to_wait(&journal->j_wait_updates, &wait,
TASK_UNINTERRUPTIBLE);
if (commit_transaction->t_updates) {
spin_unlock(&commit_transaction->t_handle_lock);
spin_unlock(&journal->j_state_lock);
schedule();
spin_lock(&journal->j_state_lock);
spin_lock(&commit_transaction->t_handle_lock);
}
finish_wait(&journal->j_wait_updates, &wait);
}
spin_unlock(&commit_transaction->t_handle_lock);
J_ASSERT (commit_transaction->t_outstanding_credits <=
journal->j_max_transaction_buffers);
/*
* First thing we are allowed to do is to discard any remaining
* BJ_Reserved buffers. Note, it is _not_ permissible to assume
* that there are no such buffers: if a large filesystem
* operation like a truncate needs to split itself over multiple
* transactions, then it may try to do a journal_restart() while
* there are still BJ_Reserved buffers outstanding. These must
* be released cleanly from the current transaction.
*
* In this case, the filesystem must still reserve write access
* again before modifying the buffer in the new transaction, but
* we do not require it to remember exactly which old buffers it
* has reserved. This is consistent with the existing behaviour
* that multiple journal_get_write_access() calls to the same
* buffer are perfectly permissible.
*/
while (commit_transaction->t_reserved_list) {
jh = commit_transaction->t_reserved_list;
JBUFFER_TRACE(jh, "reserved, unused: refile");
/*
* A journal_get_undo_access()+journal_release_buffer() may
* leave undo-committed data.
*/
if (jh->b_committed_data) {
struct buffer_head *bh = jh2bh(jh);
jbd_lock_bh_state(bh);
jbd_free(jh->b_committed_data, bh->b_size);
jh->b_committed_data = NULL;
jbd_unlock_bh_state(bh);
}
journal_refile_buffer(journal, jh);
}
/*
* Now try to drop any written-back buffers from the journal's
* checkpoint lists. We do this *before* commit because it potentially
* frees some memory
*/
spin_lock(&journal->j_list_lock);
__journal_clean_checkpoint_list(journal);
spin_unlock(&journal->j_list_lock);
jbd_debug (3, "JBD: commit phase 1\n");
/*
* Clear revoked flag to reflect there is no revoked buffers
* in the next transaction which is going to be started.
*/
journal_clear_buffer_revoked_flags(journal);
/*
* Switch to a new revoke table.
*/
journal_switch_revoke_table(journal);
trace_jbd_commit_flushing(journal, commit_transaction);
commit_transaction->t_state = T_FLUSH;
journal->j_committing_transaction = commit_transaction;
journal->j_running_transaction = NULL;
start_time = ktime_get();
commit_transaction->t_log_start = journal->j_head;
wake_up(&journal->j_wait_transaction_locked);
spin_unlock(&journal->j_state_lock);
jbd_debug (3, "JBD: commit phase 2\n");
/*
* Now start flushing things to disk, in the order they appear
* on the transaction lists. Data blocks go first.
*/
blk_start_plug(&plug);
err = journal_submit_data_buffers(journal, commit_transaction,
WRITE_SYNC);
blk_finish_plug(&plug);
/*
* Wait for all previously submitted IO to complete.
*/
spin_lock(&journal->j_list_lock);
while (commit_transaction->t_locked_list) {
struct buffer_head *bh;
jh = commit_transaction->t_locked_list->b_tprev;
bh = jh2bh(jh);
get_bh(bh);
if (buffer_locked(bh)) {
spin_unlock(&journal->j_list_lock);
wait_on_buffer(bh);
spin_lock(&journal->j_list_lock);
}
if (unlikely(!buffer_uptodate(bh))) {
if (!trylock_page(bh->b_page)) {
spin_unlock(&journal->j_list_lock);
lock_page(bh->b_page);
spin_lock(&journal->j_list_lock);
}
if (bh->b_page->mapping)
set_bit(AS_EIO, &bh->b_page->mapping->flags);
unlock_page(bh->b_page);
SetPageError(bh->b_page);
err = -EIO;
}
if (!inverted_lock(journal, bh)) {
put_bh(bh);
spin_lock(&journal->j_list_lock);
continue;
}
if (buffer_jbd(bh) && bh2jh(bh) == jh &&
jh->b_transaction == commit_transaction &&
jh->b_jlist == BJ_Locked)
__journal_unfile_buffer(jh);
jbd_unlock_bh_state(bh);
release_data_buffer(bh);
cond_resched_lock(&journal->j_list_lock);
}
spin_unlock(&journal->j_list_lock);
if (err) {
char b[BDEVNAME_SIZE];
printk(KERN_WARNING
"JBD: Detected IO errors while flushing file data "
"on %s\n", bdevname(journal->j_fs_dev, b));
if (journal->j_flags & JFS_ABORT_ON_SYNCDATA_ERR)
journal_abort(journal, err);
err = 0;
}
blk_start_plug(&plug);
journal_write_revoke_records(journal, commit_transaction, WRITE_SYNC);
/*
* If we found any dirty or locked buffers, then we should have
* looped back up to the write_out_data label. If there weren't
* any then journal_clean_data_list should have wiped the list
* clean by now, so check that it is in fact empty.
*/
J_ASSERT (commit_transaction->t_sync_datalist == NULL);
jbd_debug (3, "JBD: commit phase 3\n");
/*
* Way to go: we have now written out all of the data for a
* transaction! Now comes the tricky part: we need to write out
* metadata. Loop over the transaction's entire buffer list:
*/
spin_lock(&journal->j_state_lock);
commit_transaction->t_state = T_COMMIT;
spin_unlock(&journal->j_state_lock);
trace_jbd_commit_logging(journal, commit_transaction);
J_ASSERT(commit_transaction->t_nr_buffers <=
commit_transaction->t_outstanding_credits);
descriptor = NULL;
bufs = 0;
while (commit_transaction->t_buffers) {
/* Find the next buffer to be journaled... */
jh = commit_transaction->t_buffers;
/* If we're in abort mode, we just un-journal the buffer and
release it. */
if (is_journal_aborted(journal)) {
clear_buffer_jbddirty(jh2bh(jh));
JBUFFER_TRACE(jh, "journal is aborting: refile");
journal_refile_buffer(journal, jh);
/* If that was the last one, we need to clean up
* any descriptor buffers which may have been
* already allocated, even if we are now
* aborting. */
if (!commit_transaction->t_buffers)
goto start_journal_io;
continue;
}
/* Make sure we have a descriptor block in which to
record the metadata buffer. */
if (!descriptor) {
struct buffer_head *bh;
J_ASSERT (bufs == 0);
jbd_debug(4, "JBD: get descriptor\n");
descriptor = journal_get_descriptor_buffer(journal);
if (!descriptor) {
journal_abort(journal, -EIO);
continue;
}
bh = jh2bh(descriptor);
jbd_debug(4, "JBD: got buffer %llu (%p)\n",
(unsigned long long)bh->b_blocknr, bh->b_data);
header = (journal_header_t *)&bh->b_data[0];
header->h_magic = cpu_to_be32(JFS_MAGIC_NUMBER);
header->h_blocktype = cpu_to_be32(JFS_DESCRIPTOR_BLOCK);
header->h_sequence = cpu_to_be32(commit_transaction->t_tid);
tagp = &bh->b_data[sizeof(journal_header_t)];
space_left = bh->b_size - sizeof(journal_header_t);
first_tag = 1;
set_buffer_jwrite(bh);
set_buffer_dirty(bh);
wbuf[bufs++] = bh;
/* Record it so that we can wait for IO
completion later */
BUFFER_TRACE(bh, "ph3: file as descriptor");
journal_file_buffer(descriptor, commit_transaction,
BJ_LogCtl);
}
/* Where is the buffer to be written? */
err = journal_next_log_block(journal, &blocknr);
/* If the block mapping failed, just abandon the buffer
and repeat this loop: we'll fall into the
refile-on-abort condition above. */
if (err) {
journal_abort(journal, err);
continue;
}
/*
* start_this_handle() uses t_outstanding_credits to determine
* the free space in the log, but this counter is changed
* by journal_next_log_block() also.
*/
commit_transaction->t_outstanding_credits--;
/* Bump b_count to prevent truncate from stumbling over
the shadowed buffer! @@@ This can go if we ever get
rid of the BJ_IO/BJ_Shadow pairing of buffers. */
get_bh(jh2bh(jh));
/* Make a temporary IO buffer with which to write it out
(this will requeue both the metadata buffer and the
temporary IO buffer). new_bh goes on BJ_IO*/
set_buffer_jwrite(jh2bh(jh));
/*
* akpm: journal_write_metadata_buffer() sets
* new_bh->b_transaction to commit_transaction.
* We need to clean this up before we release new_bh
* (which is of type BJ_IO)
*/
JBUFFER_TRACE(jh, "ph3: write metadata");
flags = journal_write_metadata_buffer(commit_transaction,
jh, &new_jh, blocknr);
set_buffer_jwrite(jh2bh(new_jh));
wbuf[bufs++] = jh2bh(new_jh);
/* Record the new block's tag in the current descriptor
buffer */
tag_flag = 0;
if (flags & 1)
tag_flag |= JFS_FLAG_ESCAPE;
if (!first_tag)
tag_flag |= JFS_FLAG_SAME_UUID;
tag = (journal_block_tag_t *) tagp;
tag->t_blocknr = cpu_to_be32(jh2bh(jh)->b_blocknr);
tag->t_flags = cpu_to_be32(tag_flag);
tagp += sizeof(journal_block_tag_t);
space_left -= sizeof(journal_block_tag_t);
if (first_tag) {
memcpy (tagp, journal->j_uuid, 16);
tagp += 16;
space_left -= 16;
first_tag = 0;
}
/* If there's no more to do, or if the descriptor is full,
let the IO rip! */
if (bufs == journal->j_wbufsize ||
commit_transaction->t_buffers == NULL ||
space_left < sizeof(journal_block_tag_t) + 16) {
jbd_debug(4, "JBD: Submit %d IOs\n", bufs);
/* Write an end-of-descriptor marker before
submitting the IOs. "tag" still points to
the last tag we set up. */
tag->t_flags |= cpu_to_be32(JFS_FLAG_LAST_TAG);
start_journal_io:
for (i = 0; i < bufs; i++) {
struct buffer_head *bh = wbuf[i];
lock_buffer(bh);
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
bh->b_end_io = journal_end_buffer_io_sync;
submit_bh(WRITE_SYNC, bh);
}
cond_resched();
/* Force a new descriptor to be generated next
time round the loop. */
descriptor = NULL;
bufs = 0;
}
}
blk_finish_plug(&plug);
/* Lo and behold: we have just managed to send a transaction to
the log. Before we can commit it, wait for the IO so far to
complete. Control buffers being written are on the
transaction's t_log_list queue, and metadata buffers are on
the t_iobuf_list queue.
Wait for the buffers in reverse order. That way we are
less likely to be woken up until all IOs have completed, and
so we incur less scheduling load.
*/
jbd_debug(3, "JBD: commit phase 4\n");
/*
* akpm: these are BJ_IO, and j_list_lock is not needed.
* See __journal_try_to_free_buffer.
*/
wait_for_iobuf:
while (commit_transaction->t_iobuf_list != NULL) {
struct buffer_head *bh;
jh = commit_transaction->t_iobuf_list->b_tprev;
bh = jh2bh(jh);
if (buffer_locked(bh)) {
wait_on_buffer(bh);
goto wait_for_iobuf;
}
if (cond_resched())
goto wait_for_iobuf;
if (unlikely(!buffer_uptodate(bh)))
err = -EIO;
clear_buffer_jwrite(bh);
JBUFFER_TRACE(jh, "ph4: unfile after journal write");
journal_unfile_buffer(journal, jh);
/*
* ->t_iobuf_list should contain only dummy buffer_heads
* which were created by journal_write_metadata_buffer().
*/
BUFFER_TRACE(bh, "dumping temporary bh");
journal_put_journal_head(jh);
__brelse(bh);
J_ASSERT_BH(bh, atomic_read(&bh->b_count) == 0);
free_buffer_head(bh);
/* We also have to unlock and free the corresponding
shadowed buffer */
jh = commit_transaction->t_shadow_list->b_tprev;
bh = jh2bh(jh);
clear_buffer_jwrite(bh);
J_ASSERT_BH(bh, buffer_jbddirty(bh));
/* The metadata is now released for reuse, but we need
to remember it against this transaction so that when
we finally commit, we can do any checkpointing
required. */
JBUFFER_TRACE(jh, "file as BJ_Forget");
journal_file_buffer(jh, commit_transaction, BJ_Forget);
/*
* Wake up any transactions which were waiting for this
* IO to complete. The barrier must be here so that changes
* by journal_file_buffer() take effect before wake_up_bit()
* does the waitqueue check.
*/
smp_mb();
wake_up_bit(&bh->b_state, BH_Unshadow);
JBUFFER_TRACE(jh, "brelse shadowed buffer");
__brelse(bh);
}
J_ASSERT (commit_transaction->t_shadow_list == NULL);
jbd_debug(3, "JBD: commit phase 5\n");
/* Here we wait for the revoke record and descriptor record buffers */
wait_for_ctlbuf:
while (commit_transaction->t_log_list != NULL) {
struct buffer_head *bh;
jh = commit_transaction->t_log_list->b_tprev;
bh = jh2bh(jh);
if (buffer_locked(bh)) {
wait_on_buffer(bh);
goto wait_for_ctlbuf;
}
if (cond_resched())
goto wait_for_ctlbuf;
if (unlikely(!buffer_uptodate(bh)))
err = -EIO;
BUFFER_TRACE(bh, "ph5: control buffer writeout done: unfile");
clear_buffer_jwrite(bh);
journal_unfile_buffer(journal, jh);
journal_put_journal_head(jh);
__brelse(bh); /* One for getblk */
/* AKPM: bforget here */
}
if (err)
journal_abort(journal, err);
jbd_debug(3, "JBD: commit phase 6\n");
/* All metadata is written, now write commit record and do cleanup */
spin_lock(&journal->j_state_lock);
J_ASSERT(commit_transaction->t_state == T_COMMIT);
commit_transaction->t_state = T_COMMIT_RECORD;
spin_unlock(&journal->j_state_lock);
if (journal_write_commit_record(journal, commit_transaction))
err = -EIO;
if (err)
journal_abort(journal, err);
/* End of a transaction! Finally, we can do checkpoint
processing: any buffers committed as a result of this
transaction can be removed from any checkpoint list it was on
before. */
jbd_debug(3, "JBD: commit phase 7\n");
J_ASSERT(commit_transaction->t_sync_datalist == NULL);
J_ASSERT(commit_transaction->t_buffers == NULL);
J_ASSERT(commit_transaction->t_checkpoint_list == NULL);
J_ASSERT(commit_transaction->t_iobuf_list == NULL);
J_ASSERT(commit_transaction->t_shadow_list == NULL);
J_ASSERT(commit_transaction->t_log_list == NULL);
restart_loop:
/*
* As there are other places (journal_unmap_buffer()) adding buffers
* to this list we have to be careful and hold the j_list_lock.
*/
spin_lock(&journal->j_list_lock);
while (commit_transaction->t_forget) {
transaction_t *cp_transaction;
struct buffer_head *bh;
int try_to_free = 0;
jh = commit_transaction->t_forget;
spin_unlock(&journal->j_list_lock);
bh = jh2bh(jh);
/*
* Get a reference so that bh cannot be freed before we are
* done with it.
*/
get_bh(bh);
jbd_lock_bh_state(bh);
J_ASSERT_JH(jh, jh->b_transaction == commit_transaction ||
jh->b_transaction == journal->j_running_transaction);
/*
* If there is undo-protected committed data against
* this buffer, then we can remove it now. If it is a
* buffer needing such protection, the old frozen_data
* field now points to a committed version of the
* buffer, so rotate that field to the new committed
* data.
*
* Otherwise, we can just throw away the frozen data now.
*/
if (jh->b_committed_data) {
jbd_free(jh->b_committed_data, bh->b_size);
jh->b_committed_data = NULL;
if (jh->b_frozen_data) {
jh->b_committed_data = jh->b_frozen_data;
jh->b_frozen_data = NULL;
}
} else if (jh->b_frozen_data) {
jbd_free(jh->b_frozen_data, bh->b_size);
jh->b_frozen_data = NULL;
}
spin_lock(&journal->j_list_lock);
cp_transaction = jh->b_cp_transaction;
if (cp_transaction) {
JBUFFER_TRACE(jh, "remove from old cp transaction");
__journal_remove_checkpoint(jh);
}
/* Only re-checkpoint the buffer_head if it is marked
* dirty. If the buffer was added to the BJ_Forget list
* by journal_forget, it may no longer be dirty and
* there's no point in keeping a checkpoint record for
* it. */
/* A buffer which has been freed while still being
* journaled by a previous transaction may end up still
* being dirty here, but we want to avoid writing back
* that buffer in the future after the "add to orphan"
* operation been committed, That's not only a performance
* gain, it also stops aliasing problems if the buffer is
* left behind for writeback and gets reallocated for another
* use in a different page. */
if (buffer_freed(bh) && !jh->b_next_transaction) {
clear_buffer_freed(bh);
clear_buffer_jbddirty(bh);
}
if (buffer_jbddirty(bh)) {
JBUFFER_TRACE(jh, "add to new checkpointing trans");
__journal_insert_checkpoint(jh, commit_transaction);
if (is_journal_aborted(journal))
clear_buffer_jbddirty(bh);
} else {
J_ASSERT_BH(bh, !buffer_dirty(bh));
/*
* The buffer on BJ_Forget list and not jbddirty means
* it has been freed by this transaction and hence it
* could not have been reallocated until this
* transaction has committed. *BUT* it could be
* reallocated once we have written all the data to
* disk and before we process the buffer on BJ_Forget
* list.
*/
if (!jh->b_next_transaction)
try_to_free = 1;
}
JBUFFER_TRACE(jh, "refile or unfile freed buffer");
__journal_refile_buffer(jh);
jbd_unlock_bh_state(bh);
if (try_to_free)
release_buffer_page(bh);
else
__brelse(bh);
cond_resched_lock(&journal->j_list_lock);
}
spin_unlock(&journal->j_list_lock);
/*
* This is a bit sleazy. We use j_list_lock to protect transition
* of a transaction into T_FINISHED state and calling
* __journal_drop_transaction(). Otherwise we could race with
* other checkpointing code processing the transaction...
*/
spin_lock(&journal->j_state_lock);
spin_lock(&journal->j_list_lock);
/*
* Now recheck if some buffers did not get attached to the transaction
* while the lock was dropped...
*/
if (commit_transaction->t_forget) {
spin_unlock(&journal->j_list_lock);
spin_unlock(&journal->j_state_lock);
goto restart_loop;
}
/* Done with this transaction! */
jbd_debug(3, "JBD: commit phase 8\n");
J_ASSERT(commit_transaction->t_state == T_COMMIT_RECORD);
commit_transaction->t_state = T_FINISHED;
J_ASSERT(commit_transaction == journal->j_committing_transaction);
journal->j_commit_sequence = commit_transaction->t_tid;
journal->j_committing_transaction = NULL;
commit_time = ktime_to_ns(ktime_sub(ktime_get(), start_time));
/*
* weight the commit time higher than the average time so we don't
* react too strongly to vast changes in commit time
*/
if (likely(journal->j_average_commit_time))
journal->j_average_commit_time = (commit_time*3 +
journal->j_average_commit_time) / 4;
else
journal->j_average_commit_time = commit_time;
spin_unlock(&journal->j_state_lock);
if (commit_transaction->t_checkpoint_list == NULL &&
commit_transaction->t_checkpoint_io_list == NULL) {
__journal_drop_transaction(journal, commit_transaction);
} else {
if (journal->j_checkpoint_transactions == NULL) {
journal->j_checkpoint_transactions = commit_transaction;
commit_transaction->t_cpnext = commit_transaction;
commit_transaction->t_cpprev = commit_transaction;
} else {
commit_transaction->t_cpnext =
journal->j_checkpoint_transactions;
commit_transaction->t_cpprev =
commit_transaction->t_cpnext->t_cpprev;
commit_transaction->t_cpnext->t_cpprev =
commit_transaction;
commit_transaction->t_cpprev->t_cpnext =
commit_transaction;
}
}
spin_unlock(&journal->j_list_lock);
trace_jbd_end_commit(journal, commit_transaction);
jbd_debug(1, "JBD: commit %d complete, head %d\n",
journal->j_commit_sequence, journal->j_tail_sequence);
wake_up(&journal->j_wait_done_commit);
}
