static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb,
struct ocfs2_alloc_context *ac,
int type,
u32 slot,
u64 *last_alloc_group,
int flags)
{
int status;
u32 bits_wanted = ac->ac_bits_wanted;
struct inode *alloc_inode;
struct buffer_head *bh = NULL;
struct ocfs2_dinode *fe;
u32 free_bits;
alloc_inode = ocfs2_get_system_file_inode(osb, type, slot);
if (!alloc_inode) {
mlog_errno(-EINVAL);
return -EINVAL;
}
mutex_lock(&alloc_inode->i_mutex);
status = ocfs2_inode_lock(alloc_inode, &bh, 1);
if (status < 0) {
mutex_unlock(&alloc_inode->i_mutex);
iput(alloc_inode);
mlog_errno(status);
return status;
}
ac->ac_inode = alloc_inode;
ac->ac_alloc_slot = slot;
fe = (struct ocfs2_dinode *) bh->b_data;
/* The bh was validated by the inode read inside
* ocfs2_inode_lock(). Any corruption is a code bug. */
BUG_ON(!OCFS2_IS_VALID_DINODE(fe));
if (!(fe->i_flags & cpu_to_le32(OCFS2_CHAIN_FL))) {
status = ocfs2_error(alloc_inode->i_sb,
"Invalid chain allocator %llu\n",
(unsigned long long)le64_to_cpu(fe->i_blkno));
goto bail;
}
free_bits = le32_to_cpu(fe->id1.bitmap1.i_total) -
le32_to_cpu(fe->id1.bitmap1.i_used);
if (bits_wanted > free_bits) {
/* cluster bitmap never grows */
if (ocfs2_is_cluster_bitmap(alloc_inode)) {
trace_ocfs2_reserve_suballoc_bits_nospc(bits_wanted,
free_bits);
status = -ENOSPC;
goto bail;
}
if (!(flags & ALLOC_NEW_GROUP)) {
trace_ocfs2_reserve_suballoc_bits_no_new_group(
slot, bits_wanted, free_bits);
status = -ENOSPC;
goto bail;
}
status = ocfs2_block_group_alloc(osb, alloc_inode, bh,
ac->ac_max_block,
last_alloc_group, flags);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
atomic_inc(&osb->alloc_stats.bg_extends);
/* You should never ask for this much metadata */
BUG_ON(bits_wanted >
(le32_to_cpu(fe->id1.bitmap1.i_total)
- le32_to_cpu(fe->id1.bitmap1.i_used)));
}
get_bh(bh);
ac->ac_bh = bh;
bail:
brelse(bh);
if (status)
mlog_errno(status);
return status;
}
/*
* Initialize the ACLs of a new inode. If parent directory has default ACL,
* then clone to new inode. Called from ocfs2_mknod.
*/
int ocfs2_init_acl(handle_t *handle,
struct inode *inode,
struct inode *dir,
struct buffer_head *di_bh,
struct buffer_head *dir_bh,
struct ocfs2_alloc_context *meta_ac,
struct ocfs2_alloc_context *data_ac)
{
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct posix_acl *acl = NULL;
int ret = 0, ret2;
mode_t mode;
if (!S_ISLNK(inode->i_mode)) {
if (osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) {
acl = ocfs2_get_acl_nolock(dir, ACL_TYPE_DEFAULT,
dir_bh);
if (IS_ERR(acl))
return PTR_ERR(acl);
}
if (!acl) {
mode = inode->i_mode & ~current_umask();
ret = ocfs2_acl_set_mode(inode, di_bh, handle, mode);
if (ret) {
mlog_errno(ret);
goto cleanup;
}
}
}
if ((osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL) && acl) {
struct posix_acl *clone;
if (S_ISDIR(inode->i_mode)) {
ret = ocfs2_set_acl(handle, inode, di_bh,
ACL_TYPE_DEFAULT, acl,
meta_ac, data_ac);
if (ret)
goto cleanup;
}
clone = posix_acl_clone(acl, GFP_NOFS);
ret = -ENOMEM;
if (!clone)
goto cleanup;
mode = inode->i_mode;
ret = posix_acl_create_masq(clone, &mode);
if (ret >= 0) {
ret2 = ocfs2_acl_set_mode(inode, di_bh, handle, mode);
if (ret2) {
mlog_errno(ret2);
ret = ret2;
goto cleanup;
}
if (ret > 0) {
ret = ocfs2_set_acl(handle, inode,
di_bh, ACL_TYPE_ACCESS,
clone, meta_ac, data_ac);
}
}
posix_acl_release(clone);
}
cleanup:
posix_acl_release(acl);
return ret;
}
/* Add a new group descriptor to global_bitmap. */
int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
{
int ret;
handle_t *handle;
struct buffer_head *main_bm_bh = NULL;
struct inode *main_bm_inode = NULL;
struct ocfs2_dinode *fe = NULL;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct buffer_head *group_bh = NULL;
struct ocfs2_group_desc *group = NULL;
struct ocfs2_chain_list *cl;
struct ocfs2_chain_rec *cr;
u16 cl_bpc;
mlog_entry_void();
if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
return -EROFS;
main_bm_inode = ocfs2_get_system_file_inode(osb,
GLOBAL_BITMAP_SYSTEM_INODE,
OCFS2_INVALID_SLOT);
if (!main_bm_inode) {
ret = -EINVAL;
mlog_errno(ret);
goto out;
}
mutex_lock(&main_bm_inode->i_mutex);
ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
if (ret < 0) {
mlog_errno(ret);
goto out_mutex;
}
fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
ocfs2_group_bitmap_size(osb->sb) * 8) {
mlog(ML_ERROR, "The disk is too old and small."
" Force to do offline resize.");
ret = -EINVAL;
goto out_unlock;
}
ret = ocfs2_read_blocks_sync(osb, input->group, 1, &group_bh);
if (ret < 0) {
mlog(ML_ERROR, "Can't read the group descriptor # %llu "
"from the device.", (unsigned long long)input->group);
goto out_unlock;
}
ocfs2_set_new_buffer_uptodate(inode, group_bh);
ret = ocfs2_verify_group_and_input(main_bm_inode, fe, input, group_bh);
if (ret) {
mlog_errno(ret);
goto out_unlock;
}
mlog(0, "Add a new group %llu in chain = %u, length = %u\n",
(unsigned long long)input->group, input->chain, input->clusters);
handle = ocfs2_start_trans(osb, OCFS2_GROUP_ADD_CREDITS);
if (IS_ERR(handle)) {
mlog_errno(PTR_ERR(handle));
ret = -EINVAL;
goto out_unlock;
}
cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
cl = &fe->id2.i_chain;
cr = &cl->cl_recs[input->chain];
ret = ocfs2_journal_access_gd(handle, main_bm_inode, group_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (ret < 0) {
mlog_errno(ret);
goto out_commit;
}
group = (struct ocfs2_group_desc *)group_bh->b_data;
group->bg_next_group = cr->c_blkno;
ret = ocfs2_journal_dirty(handle, group_bh);
if (ret < 0) {
mlog_errno(ret);
goto out_commit;
}
ret = ocfs2_journal_access_di(handle, main_bm_inode, main_bm_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (ret < 0) {
mlog_errno(ret);
goto out_commit;
}
if (input->chain == le16_to_cpu(cl->cl_next_free_rec)) {
le16_add_cpu(&cl->cl_next_free_rec, 1);
memset(cr, 0, sizeof(struct ocfs2_chain_rec));
}
cr->c_blkno = cpu_to_le64(input->group);
le32_add_cpu(&cr->c_total, input->clusters * cl_bpc);
le32_add_cpu(&cr->c_free, input->frees * cl_bpc);
le32_add_cpu(&fe->id1.bitmap1.i_total, input->clusters *cl_bpc);
le32_add_cpu(&fe->id1.bitmap1.i_used,
(input->clusters - input->frees) * cl_bpc);
le32_add_cpu(&fe->i_clusters, input->clusters);
ocfs2_journal_dirty(handle, main_bm_bh);
spin_lock(&OCFS2_I(main_bm_inode)->ip_lock);
OCFS2_I(main_bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
le64_add_cpu(&fe->i_size, input->clusters << osb->s_clustersize_bits);
spin_unlock(&OCFS2_I(main_bm_inode)->ip_lock);
i_size_write(main_bm_inode, le64_to_cpu(fe->i_size));
ocfs2_update_super_and_backups(main_bm_inode, input->clusters);
out_commit:
ocfs2_commit_trans(osb, handle);
out_unlock:
brelse(group_bh);
brelse(main_bm_bh);
ocfs2_inode_unlock(main_bm_inode, 1);
out_mutex:
mutex_unlock(&main_bm_inode->i_mutex);
iput(main_bm_inode);
out:
mlog_exit_void();
return ret;
}
static int ocfs2_remove_inode(struct inode *inode,
struct buffer_head *di_bh,
struct inode *orphan_dir_inode,
struct buffer_head *orphan_dir_bh)
{
int status;
struct inode *inode_alloc_inode = NULL;
struct buffer_head *inode_alloc_bh = NULL;
handle_t *handle;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
inode_alloc_inode =
ocfs2_get_system_file_inode(osb, INODE_ALLOC_SYSTEM_INODE,
le16_to_cpu(di->i_suballoc_slot));
if (!inode_alloc_inode) {
status = -EEXIST;
mlog_errno(status);
goto bail;
}
mutex_lock(&inode_alloc_inode->i_mutex);
status = ocfs2_inode_lock(inode_alloc_inode, &inode_alloc_bh, 1);
if (status < 0) {
mutex_unlock(&inode_alloc_inode->i_mutex);
mlog_errno(status);
goto bail;
}
handle = ocfs2_start_trans(osb, OCFS2_DELETE_INODE_CREDITS +
ocfs2_quota_trans_credits(inode->i_sb));
if (IS_ERR(handle)) {
status = PTR_ERR(handle);
mlog_errno(status);
goto bail_unlock;
}
if (!(OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
status = ocfs2_orphan_del(osb, handle, orphan_dir_inode, inode,
orphan_dir_bh);
if (status < 0) {
mlog_errno(status);
goto bail_commit;
}
}
/* set the inodes dtime */
status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto bail_commit;
}
di->i_dtime = cpu_to_le64(CURRENT_TIME.tv_sec);
di->i_flags &= cpu_to_le32(~(OCFS2_VALID_FL | OCFS2_ORPHANED_FL));
ocfs2_journal_dirty(handle, di_bh);
ocfs2_remove_from_cache(INODE_CACHE(inode), di_bh);
dquot_free_inode(inode);
status = ocfs2_free_dinode(handle, inode_alloc_inode,
inode_alloc_bh, di);
if (status < 0)
mlog_errno(status);
bail_commit:
ocfs2_commit_trans(osb, handle);
bail_unlock:
ocfs2_inode_unlock(inode_alloc_inode, 1);
mutex_unlock(&inode_alloc_inode->i_mutex);
brelse(inode_alloc_bh);
bail:
iput(inode_alloc_inode);
return status;
}
/* Query the cluster to determine whether we should wipe an inode from
* disk or not.
*
* Requires the inode to have the cluster lock. */
static int ocfs2_query_inode_wipe(struct inode *inode,
struct buffer_head *di_bh,
int *wipe)
{
int status = 0, reason = 0;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
struct ocfs2_dinode *di;
*wipe = 0;
trace_ocfs2_query_inode_wipe_begin((unsigned long long)oi->ip_blkno,
inode->i_nlink);
/* While we were waiting for the cluster lock in
* ocfs2_delete_inode, another node might have asked to delete
* the inode. Recheck our flags to catch this. */
if (!ocfs2_inode_is_valid_to_delete(inode)) {
reason = 1;
goto bail;
}
/* Now that we have an up to date inode, we can double check
* the link count. */
if (inode->i_nlink)
goto bail;
/* Do some basic inode verification... */
di = (struct ocfs2_dinode *) di_bh->b_data;
if (!(di->i_flags & cpu_to_le32(OCFS2_ORPHANED_FL)) &&
!(oi->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
/*
* Inodes in the orphan dir must have ORPHANED_FL. The only
* inodes that come back out of the orphan dir are reflink
* targets. A reflink target may be moved out of the orphan
* dir between the time we scan the directory and the time we
* process it. This would lead to HAS_REFCOUNT_FL being set but
* ORPHANED_FL not.
*/
if (di->i_dyn_features & cpu_to_le16(OCFS2_HAS_REFCOUNT_FL)) {
reason = 2;
goto bail;
}
/* for lack of a better error? */
status = -EEXIST;
mlog(ML_ERROR,
"Inode %llu (on-disk %llu) not orphaned! "
"Disk flags 0x%x, inode flags 0x%x\n",
(unsigned long long)oi->ip_blkno,
(unsigned long long)le64_to_cpu(di->i_blkno),
le32_to_cpu(di->i_flags), oi->ip_flags);
goto bail;
}
/* has someone already deleted us?! baaad... */
if (di->i_dtime) {
status = -EEXIST;
mlog_errno(status);
goto bail;
}
/*
* This is how ocfs2 determines whether an inode is still live
* within the cluster. Every node takes a shared read lock on
* the inode open lock in ocfs2_read_locked_inode(). When we
* get to ->delete_inode(), each node tries to convert it's
* lock to an exclusive. Trylocks are serialized by the inode
* meta data lock. If the upconvert succeeds, we know the inode
* is no longer live and can be deleted.
*
* Though we call this with the meta data lock held, the
* trylock keeps us from ABBA deadlock.
*/
status = ocfs2_try_open_lock(inode, 1);
if (status == -EAGAIN) {
status = 0;
reason = 3;
goto bail;
}
if (status < 0) {
mlog_errno(status);
goto bail;
}
*wipe = 1;
trace_ocfs2_query_inode_wipe_succ(le16_to_cpu(di->i_orphaned_slot));
bail:
trace_ocfs2_query_inode_wipe_end(status, reason);
return status;
}
/*
* sync the local alloc to main bitmap.
*
* assumes you've already locked the main bitmap -- the bitmap inode
* passed is used for caching.
*/
static int ocfs2_sync_local_to_main(struct ocfs2_super *osb,
handle_t *handle,
struct ocfs2_dinode *alloc,
struct inode *main_bm_inode,
struct buffer_head *main_bm_bh)
{
int status = 0;
int bit_off, left, count, start;
u64 la_start_blk;
u64 blkno;
void *bitmap;
struct ocfs2_local_alloc *la = OCFS2_LOCAL_ALLOC(alloc);
trace_ocfs2_sync_local_to_main(
le32_to_cpu(alloc->id1.bitmap1.i_total),
le32_to_cpu(alloc->id1.bitmap1.i_used));
if (!alloc->id1.bitmap1.i_total) {
goto bail;
}
if (le32_to_cpu(alloc->id1.bitmap1.i_used) ==
le32_to_cpu(alloc->id1.bitmap1.i_total)) {
goto bail;
}
la_start_blk = ocfs2_clusters_to_blocks(osb->sb,
le32_to_cpu(la->la_bm_off));
bitmap = la->la_bitmap;
start = count = bit_off = 0;
left = le32_to_cpu(alloc->id1.bitmap1.i_total);
while ((bit_off = ocfs2_find_next_zero_bit(bitmap, left, start))
!= -1) {
if ((bit_off < left) && (bit_off == start)) {
count++;
start++;
continue;
}
if (count) {
blkno = la_start_blk +
ocfs2_clusters_to_blocks(osb->sb,
start - count);
trace_ocfs2_sync_local_to_main_free(
count, start - count,
(unsigned long long)la_start_blk,
(unsigned long long)blkno);
status = ocfs2_release_clusters(handle,
main_bm_inode,
main_bm_bh, blkno,
count);
if (status < 0) {
mlog_errno(status);
goto bail;
}
}
if (bit_off >= left)
break;
count = 1;
start = bit_off + 1;
}
bail:
if (status)
mlog_errno(status);
return status;
}
struct inode *ocfs2_iget(struct ocfs2_super *osb, u64 blkno, unsigned flags,
int sysfile_type)
{
struct inode *inode = NULL;
struct super_block *sb = osb->sb;
struct ocfs2_find_inode_args args;
journal_t *journal = OCFS2_SB(sb)->journal->j_journal;
trace_ocfs2_iget_begin((unsigned long long)blkno, flags,
sysfile_type);
/* Ok. By now we've either got the offsets passed to us by the
* caller, or we just pulled them off the bh. Lets do some
* sanity checks to make sure they're OK. */
if (blkno == 0) {
inode = ERR_PTR(-EINVAL);
mlog_errno(PTR_ERR(inode));
goto bail;
}
args.fi_blkno = blkno;
args.fi_flags = flags;
args.fi_ino = ino_from_blkno(sb, blkno);
args.fi_sysfile_type = sysfile_type;
inode = iget5_locked(sb, args.fi_ino, ocfs2_find_actor,
ocfs2_init_locked_inode, &args);
/* inode was *not* in the inode cache. 2.6.x requires
* us to do our own read_inode call and unlock it
* afterwards. */
if (inode == NULL) {
inode = ERR_PTR(-ENOMEM);
mlog_errno(PTR_ERR(inode));
goto bail;
}
trace_ocfs2_iget5_locked(inode->i_state);
if (inode->i_state & I_NEW) {
ocfs2_read_locked_inode(inode, &args);
unlock_new_inode(inode);
}
if (is_bad_inode(inode)) {
iput(inode);
inode = ERR_PTR(-ESTALE);
goto bail;
}
/*
* Set transaction id's of transactions that have to be committed
* to finish f[data]sync. We set them to currently running transaction
* as we cannot be sure that the inode or some of its metadata isn't
* part of the transaction - the inode could have been reclaimed and
* now it is reread from disk.
*/
if (journal) {
transaction_t *transaction;
tid_t tid;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
read_lock(&journal->j_state_lock);
if (journal->j_running_transaction)
transaction = journal->j_running_transaction;
else
transaction = journal->j_committing_transaction;
if (transaction)
tid = transaction->t_tid;
else
tid = journal->j_commit_sequence;
read_unlock(&journal->j_state_lock);
oi->i_sync_tid = tid;
oi->i_datasync_tid = tid;
}
bail:
if (!IS_ERR(inode)) {
trace_ocfs2_iget_end(inode,
(unsigned long long)OCFS2_I(inode)->ip_blkno);
}
return inode;
}
/*
* expects the suballoc inode to already be locked.
*/
static int ocfs2_free_suballoc_bits(handle_t *handle,
struct inode *alloc_inode,
struct buffer_head *alloc_bh,
unsigned int start_bit,
u64 bg_blkno,
unsigned int count)
{
int status = 0;
u32 tmp_used;
struct ocfs2_super *osb = OCFS2_SB(alloc_inode->i_sb);
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) alloc_bh->b_data;
struct ocfs2_chain_list *cl = &fe->id2.i_chain;
struct buffer_head *group_bh = NULL;
struct ocfs2_group_desc *group;
mlog_entry_void();
if (!OCFS2_IS_VALID_DINODE(fe)) {
OCFS2_RO_ON_INVALID_DINODE(alloc_inode->i_sb, fe);
status = -EIO;
goto bail;
}
BUG_ON((count + start_bit) > ocfs2_bits_per_group(cl));
mlog(0, "%llu: freeing %u bits from group %llu, starting at %u\n",
(unsigned long long)OCFS2_I(alloc_inode)->ip_blkno, count,
(unsigned long long)bg_blkno, start_bit);
status = ocfs2_read_block(osb, bg_blkno, &group_bh, OCFS2_BH_CACHED,
alloc_inode);
if (status < 0) {
mlog_errno(status);
goto bail;
}
group = (struct ocfs2_group_desc *) group_bh->b_data;
status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, group);
if (status) {
mlog_errno(status);
goto bail;
}
BUG_ON((count + start_bit) > le16_to_cpu(group->bg_bits));
status = ocfs2_block_group_clear_bits(handle, alloc_inode,
group, group_bh,
start_bit, count);
if (status < 0) {
mlog_errno(status);
goto bail;
}
status = ocfs2_journal_access(handle, alloc_inode, alloc_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
le32_add_cpu(&cl->cl_recs[le16_to_cpu(group->bg_chain)].c_free,
count);
tmp_used = le32_to_cpu(fe->id1.bitmap1.i_used);
fe->id1.bitmap1.i_used = cpu_to_le32(tmp_used - count);
status = ocfs2_journal_dirty(handle, alloc_bh);
if (status < 0) {
mlog_errno(status);
goto bail;
}
bail:
if (group_bh)
brelse(group_bh);
mlog_exit(status);
return status;
}
/*
* We expect the block group allocator to already be locked.
*/
static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
struct inode *alloc_inode,
struct buffer_head *bh)
{
int status, credits;
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bh->b_data;
struct ocfs2_chain_list *cl;
struct ocfs2_alloc_context *ac = NULL;
handle_t *handle = NULL;
u32 bit_off, num_bits;
u16 alloc_rec;
u64 bg_blkno;
struct buffer_head *bg_bh = NULL;
struct ocfs2_group_desc *bg;
BUG_ON(ocfs2_is_cluster_bitmap(alloc_inode));
mlog_entry_void();
cl = &fe->id2.i_chain;
status = ocfs2_reserve_clusters(osb,
le16_to_cpu(cl->cl_cpg),
&ac);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
credits = ocfs2_calc_group_alloc_credits(osb->sb,
le16_to_cpu(cl->cl_cpg));
handle = ocfs2_start_trans(osb, credits);
if (IS_ERR(handle)) {
status = PTR_ERR(handle);
handle = NULL;
mlog_errno(status);
goto bail;
}
status = ocfs2_claim_clusters(osb,
handle,
ac,
le16_to_cpu(cl->cl_cpg),
&bit_off,
&num_bits);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
alloc_rec = ocfs2_find_smallest_chain(cl);
/* setup the group */
bg_blkno = ocfs2_clusters_to_blocks(osb->sb, bit_off);
mlog(0, "new descriptor, record %u, at block %llu\n",
alloc_rec, (unsigned long long)bg_blkno);
bg_bh = sb_getblk(osb->sb, bg_blkno);
if (!bg_bh) {
status = -EIO;
mlog_errno(status);
goto bail;
}
ocfs2_set_new_buffer_uptodate(alloc_inode, bg_bh);
status = ocfs2_block_group_fill(handle,
alloc_inode,
bg_bh,
bg_blkno,
alloc_rec,
cl);
if (status < 0) {
mlog_errno(status);
goto bail;
}
bg = (struct ocfs2_group_desc *) bg_bh->b_data;
status = ocfs2_journal_access(handle, alloc_inode,
bh, OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
le32_add_cpu(&cl->cl_recs[alloc_rec].c_free,
le16_to_cpu(bg->bg_free_bits_count));
le32_add_cpu(&cl->cl_recs[alloc_rec].c_total, le16_to_cpu(bg->bg_bits));
cl->cl_recs[alloc_rec].c_blkno = cpu_to_le64(bg_blkno);
if (le16_to_cpu(cl->cl_next_free_rec) < le16_to_cpu(cl->cl_count))
le16_add_cpu(&cl->cl_next_free_rec, 1);
le32_add_cpu(&fe->id1.bitmap1.i_used, le16_to_cpu(bg->bg_bits) -
le16_to_cpu(bg->bg_free_bits_count));
le32_add_cpu(&fe->id1.bitmap1.i_total, le16_to_cpu(bg->bg_bits));
le32_add_cpu(&fe->i_clusters, le16_to_cpu(cl->cl_cpg));
status = ocfs2_journal_dirty(handle, bh);
if (status < 0) {
mlog_errno(status);
goto bail;
}
spin_lock(&OCFS2_I(alloc_inode)->ip_lock);
OCFS2_I(alloc_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
fe->i_size = cpu_to_le64(ocfs2_clusters_to_bytes(alloc_inode->i_sb,
le32_to_cpu(fe->i_clusters)));
spin_unlock(&OCFS2_I(alloc_inode)->ip_lock);
i_size_write(alloc_inode, le64_to_cpu(fe->i_size));
alloc_inode->i_blocks = ocfs2_inode_sector_count(alloc_inode);
status = 0;
bail:
if (handle)
ocfs2_commit_trans(osb, handle);
if (ac)
ocfs2_free_alloc_context(ac);
if (bg_bh)
brelse(bg_bh);
mlog_exit(status);
return status;
}
/* will give out up to bits_wanted contiguous bits. */
static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
struct ocfs2_alloc_context *ac,
handle_t *handle,
u32 bits_wanted,
u32 min_bits,
u16 *bit_off,
unsigned int *num_bits,
u64 *bg_blkno)
{
int status;
u16 victim, i;
u16 bits_left = 0;
u64 hint_blkno = ac->ac_last_group;
struct ocfs2_chain_list *cl;
struct ocfs2_dinode *fe;
mlog_entry_void();
BUG_ON(ac->ac_bits_given >= ac->ac_bits_wanted);
BUG_ON(bits_wanted > (ac->ac_bits_wanted - ac->ac_bits_given));
BUG_ON(!ac->ac_bh);
fe = (struct ocfs2_dinode *) ac->ac_bh->b_data;
if (!OCFS2_IS_VALID_DINODE(fe)) {
OCFS2_RO_ON_INVALID_DINODE(osb->sb, fe);
status = -EIO;
goto bail;
}
if (le32_to_cpu(fe->id1.bitmap1.i_used) >=
le32_to_cpu(fe->id1.bitmap1.i_total)) {
ocfs2_error(osb->sb, "Chain allocator dinode %llu has %u used "
"bits but only %u total.",
(unsigned long long)le64_to_cpu(fe->i_blkno),
le32_to_cpu(fe->id1.bitmap1.i_used),
le32_to_cpu(fe->id1.bitmap1.i_total));
status = -EIO;
goto bail;
}
if (hint_blkno) {
/* Attempt to short-circuit the usual search mechanism
* by jumping straight to the most recently used
* allocation group. This helps us mantain some
* contiguousness across allocations. */
status = ocfs2_search_one_group(ac, handle, bits_wanted,
min_bits, bit_off, num_bits,
hint_blkno, &bits_left);
if (!status) {
/* Be careful to update *bg_blkno here as the
* caller is expecting it to be filled in, and
* ocfs2_search_one_group() won't do that for
* us. */
*bg_blkno = hint_blkno;
goto set_hint;
}
if (status < 0 && status != -ENOSPC) {
mlog_errno(status);
goto bail;
}
}
cl = (struct ocfs2_chain_list *) &fe->id2.i_chain;
victim = ocfs2_find_victim_chain(cl);
ac->ac_chain = victim;
ac->ac_allow_chain_relink = 1;
status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits, bit_off,
num_bits, bg_blkno, &bits_left);
if (!status)
goto set_hint;
if (status < 0 && status != -ENOSPC) {
mlog_errno(status);
goto bail;
}
mlog(0, "Search of victim chain %u came up with nothing, "
"trying all chains now.\n", victim);
/* If we didn't pick a good victim, then just default to
* searching each chain in order. Don't allow chain relinking
* because we only calculate enough journal credits for one
* relink per alloc. */
ac->ac_allow_chain_relink = 0;
for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i ++) {
if (i == victim)
continue;
if (!cl->cl_recs[i].c_free)
continue;
ac->ac_chain = i;
status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits,
bit_off, num_bits, bg_blkno,
&bits_left);
if (!status)
break;
if (status < 0 && status != -ENOSPC) {
mlog_errno(status);
goto bail;
}
}
set_hint:
if (status != -ENOSPC) {
/* If the next search of this group is not likely to
* yield a suitable extent, then we reset the last
* group hint so as to not waste a disk read */
if (bits_left < min_bits)
ac->ac_last_group = 0;
else
ac->ac_last_group = *bg_blkno;
}
bail:
mlog_exit(status);
return status;
}
/*
* min_bits - minimum contiguous chunk from this total allocation we
* can handle. set to what we asked for originally for a full
* contig. allocation, set to '1' to indicate we can deal with extents
* of any size.
*/
int ocfs2_claim_clusters(struct ocfs2_super *osb,
handle_t *handle,
struct ocfs2_alloc_context *ac,
u32 min_clusters,
u32 *cluster_start,
u32 *num_clusters)
{
int status;
unsigned int bits_wanted = ac->ac_bits_wanted - ac->ac_bits_given;
u64 bg_blkno = 0;
u16 bg_bit_off;
mlog_entry_void();
BUG_ON(!ac);
BUG_ON(ac->ac_bits_given >= ac->ac_bits_wanted);
BUG_ON(ac->ac_which != OCFS2_AC_USE_LOCAL
&& ac->ac_which != OCFS2_AC_USE_MAIN);
if (ac->ac_which == OCFS2_AC_USE_LOCAL) {
status = ocfs2_claim_local_alloc_bits(osb,
handle,
ac,
bits_wanted,
cluster_start,
num_clusters);
if (!status)
atomic_inc(&osb->alloc_stats.local_data);
} else {
if (min_clusters > (osb->bitmap_cpg - 1)) {
/* The only paths asking for contiguousness
* should know about this already. */
mlog(ML_ERROR, "minimum allocation requested exceeds "
"group bitmap size!");
status = -ENOSPC;
goto bail;
}
/* clamp the current request down to a realistic size. */
if (bits_wanted > (osb->bitmap_cpg - 1))
bits_wanted = osb->bitmap_cpg - 1;
status = ocfs2_claim_suballoc_bits(osb,
ac,
handle,
bits_wanted,
min_clusters,
&bg_bit_off,
num_clusters,
&bg_blkno);
if (!status) {
*cluster_start =
ocfs2_desc_bitmap_to_cluster_off(ac->ac_inode,
bg_blkno,
bg_bit_off);
atomic_inc(&osb->alloc_stats.bitmap_data);
}
}
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
ac->ac_bits_given += *num_clusters;
bail:
mlog_exit(status);
return status;
}
static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
handle_t *handle,
u32 bits_wanted,
u32 min_bits,
u16 *bit_off,
unsigned int *num_bits,
u64 *bg_blkno,
u16 *bits_left)
{
int status;
u16 chain, tmp_bits;
u32 tmp_used;
u64 next_group;
struct inode *alloc_inode = ac->ac_inode;
struct buffer_head *group_bh = NULL;
struct buffer_head *prev_group_bh = NULL;
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) ac->ac_bh->b_data;
struct ocfs2_chain_list *cl = (struct ocfs2_chain_list *) &fe->id2.i_chain;
struct ocfs2_group_desc *bg;
chain = ac->ac_chain;
mlog(0, "trying to alloc %u bits from chain %u, inode %llu\n",
bits_wanted, chain,
(unsigned long long)OCFS2_I(alloc_inode)->ip_blkno);
status = ocfs2_read_block(OCFS2_SB(alloc_inode->i_sb),
le64_to_cpu(cl->cl_recs[chain].c_blkno),
&group_bh, OCFS2_BH_CACHED, alloc_inode);
if (status < 0) {
mlog_errno(status);
goto bail;
}
bg = (struct ocfs2_group_desc *) group_bh->b_data;
status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, bg);
if (status) {
mlog_errno(status);
goto bail;
}
status = -ENOSPC;
/* for now, the chain search is a bit simplistic. We just use
* the 1st group with any empty bits. */
while ((status = ac->ac_group_search(alloc_inode, group_bh,
bits_wanted, min_bits, bit_off,
&tmp_bits)) == -ENOSPC) {
if (!bg->bg_next_group)
break;
if (prev_group_bh) {
brelse(prev_group_bh);
prev_group_bh = NULL;
}
next_group = le64_to_cpu(bg->bg_next_group);
prev_group_bh = group_bh;
group_bh = NULL;
status = ocfs2_read_block(OCFS2_SB(alloc_inode->i_sb),
next_group, &group_bh,
OCFS2_BH_CACHED, alloc_inode);
if (status < 0) {
mlog_errno(status);
goto bail;
}
bg = (struct ocfs2_group_desc *) group_bh->b_data;
status = ocfs2_check_group_descriptor(alloc_inode->i_sb, fe, bg);
if (status) {
mlog_errno(status);
goto bail;
}
}
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
mlog(0, "alloc succeeds: we give %u bits from block group %llu\n",
tmp_bits, (unsigned long long)le64_to_cpu(bg->bg_blkno));
*num_bits = tmp_bits;
BUG_ON(*num_bits == 0);
/*
* Keep track of previous block descriptor read. When
* we find a target, if we have read more than X
* number of descriptors, and the target is reasonably
* empty, relink him to top of his chain.
*
* We've read 0 extra blocks and only send one more to
* the transaction, yet the next guy to search has a
* much easier time.
*
* Do this *after* figuring out how many bits we're taking out
* of our target group.
*/
if (ac->ac_allow_chain_relink &&
(prev_group_bh) &&
(ocfs2_block_group_reasonably_empty(bg, *num_bits))) {
status = ocfs2_relink_block_group(handle, alloc_inode,
ac->ac_bh, group_bh,
prev_group_bh, chain);
if (status < 0) {
mlog_errno(status);
goto bail;
}
}
/* Ok, claim our bits now: set the info on dinode, chainlist
* and then the group */
status = ocfs2_journal_access(handle,
alloc_inode,
ac->ac_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
tmp_used = le32_to_cpu(fe->id1.bitmap1.i_used);
fe->id1.bitmap1.i_used = cpu_to_le32(*num_bits + tmp_used);
le32_add_cpu(&cl->cl_recs[chain].c_free, -(*num_bits));
status = ocfs2_journal_dirty(handle,
ac->ac_bh);
if (status < 0) {
mlog_errno(status);
goto bail;
}
status = ocfs2_block_group_set_bits(handle,
alloc_inode,
bg,
group_bh,
*bit_off,
*num_bits);
if (status < 0) {
mlog_errno(status);
goto bail;
}
mlog(0, "Allocated %u bits from suballocator %llu\n", *num_bits,
(unsigned long long)le64_to_cpu(fe->i_blkno));
*bg_blkno = le64_to_cpu(bg->bg_blkno);
*bits_left = le16_to_cpu(bg->bg_free_bits_count);
bail:
if (group_bh)
brelse(group_bh);
if (prev_group_bh)
brelse(prev_group_bh);
mlog_exit(status);
return status;
}
static int ocfs2_map_slot_buffers(struct ocfs2_super *osb,
struct ocfs2_slot_info *si)
{
int status = 0;
u64 blkno;
unsigned long long blocks, bytes;
unsigned int i;
struct buffer_head *bh;
status = ocfs2_slot_map_physical_size(osb, si->si_inode, &bytes);
if (status)
goto bail;
blocks = ocfs2_blocks_for_bytes(si->si_inode->i_sb, bytes);
BUG_ON(blocks > UINT_MAX);
si->si_blocks = blocks;
if (!si->si_blocks)
goto bail;
if (si->si_extended)
si->si_slots_per_block =
(osb->sb->s_blocksize /
sizeof(struct ocfs2_extended_slot));
else
si->si_slots_per_block = osb->sb->s_blocksize / sizeof(__le16);
/* The size checks above should ensure this */
BUG_ON((osb->max_slots / si->si_slots_per_block) > blocks);
mlog(0, "Slot map needs %u buffers for %llu bytes\n",
si->si_blocks, bytes);
si->si_bh = kzalloc(sizeof(struct buffer_head *) * si->si_blocks,
GFP_KERNEL);
if (!si->si_bh) {
status = -ENOMEM;
mlog_errno(status);
goto bail;
}
for (i = 0; i < si->si_blocks; i++) {
status = ocfs2_extent_map_get_blocks(si->si_inode, i,
&blkno, NULL, NULL);
if (status < 0) {
mlog_errno(status);
goto bail;
}
mlog(0, "Reading slot map block %u at %llu\n", i,
(unsigned long long)blkno);
bh = NULL; /* Acquire a fresh bh */
status = ocfs2_read_blocks(INODE_CACHE(si->si_inode), blkno,
1, &bh, OCFS2_BH_IGNORE_CACHE, NULL);
if (status < 0) {
mlog_errno(status);
goto bail;
}
si->si_bh[i] = bh;
}
bail:
return status;
}
int ocfs2_reserve_new_metadata_blocks(struct ocfs2_super *osb,
int blocks,
struct ocfs2_alloc_context **ac)
{
int status;
int slot = ocfs2_get_meta_steal_slot(osb);
*ac = kzalloc(sizeof(struct ocfs2_alloc_context), GFP_KERNEL);
if (!(*ac)) {
status = -ENOMEM;
mlog_errno(status);
goto bail;
}
(*ac)->ac_bits_wanted = blocks;
(*ac)->ac_which = OCFS2_AC_USE_META;
(*ac)->ac_group_search = ocfs2_block_group_search;
if (slot != OCFS2_INVALID_SLOT &&
atomic_read(&osb->s_num_meta_stolen) < OCFS2_MAX_TO_STEAL)
goto extent_steal;
atomic_set(&osb->s_num_meta_stolen, 0);
status = ocfs2_reserve_suballoc_bits(osb, (*ac),
EXTENT_ALLOC_SYSTEM_INODE,
(u32)osb->slot_num, NULL,
ALLOC_GROUPS_FROM_GLOBAL|ALLOC_NEW_GROUP);
if (status >= 0) {
status = 0;
if (slot != OCFS2_INVALID_SLOT)
ocfs2_init_meta_steal_slot(osb);
goto bail;
} else if (status < 0 && status != -ENOSPC) {
mlog_errno(status);
goto bail;
}
ocfs2_free_ac_resource(*ac);
extent_steal:
status = ocfs2_steal_meta(osb, *ac);
atomic_inc(&osb->s_num_meta_stolen);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
status = 0;
bail:
if ((status < 0) && *ac) {
ocfs2_free_alloc_context(*ac);
*ac = NULL;
}
if (status)
mlog_errno(status);
return status;
}
/*
* We want to free the bitmap bits outside of any recovery context as
* we'll need a cluster lock to do so, but we must clear the local
* alloc before giving up the recovered nodes journal. To solve this,
* we kmalloc a copy of the local alloc before it's change for the
* caller to process with ocfs2_complete_local_alloc_recovery
*/
int ocfs2_begin_local_alloc_recovery(struct ocfs2_super *osb,
int slot_num,
struct ocfs2_dinode **alloc_copy)
{
int status = 0;
struct buffer_head *alloc_bh = NULL;
struct inode *inode = NULL;
struct ocfs2_dinode *alloc;
trace_ocfs2_begin_local_alloc_recovery(slot_num);
*alloc_copy = NULL;
inode = ocfs2_get_system_file_inode(osb,
LOCAL_ALLOC_SYSTEM_INODE,
slot_num);
if (!inode) {
status = -EINVAL;
mlog_errno(status);
goto bail;
}
mutex_lock(&inode->i_mutex);
status = ocfs2_read_inode_block_full(inode, &alloc_bh,
OCFS2_BH_IGNORE_CACHE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
*alloc_copy = kmalloc(alloc_bh->b_size, GFP_KERNEL);
if (!(*alloc_copy)) {
status = -ENOMEM;
goto bail;
}
memcpy((*alloc_copy), alloc_bh->b_data, alloc_bh->b_size);
alloc = (struct ocfs2_dinode *) alloc_bh->b_data;
ocfs2_clear_local_alloc(alloc);
ocfs2_compute_meta_ecc(osb->sb, alloc_bh->b_data, &alloc->i_check);
status = ocfs2_write_block(osb, alloc_bh, INODE_CACHE(inode));
if (status < 0)
mlog_errno(status);
bail:
if (status < 0) {
kfree(*alloc_copy);
*alloc_copy = NULL;
}
brelse(alloc_bh);
if (inode) {
mutex_unlock(&inode->i_mutex);
iput(inode);
}
if (status)
mlog_errno(status);
return status;
}
static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb,
struct ocfs2_alloc_context *ac,
int type,
u32 slot)
{
int status;
u32 bits_wanted = ac->ac_bits_wanted;
struct inode *alloc_inode;
struct buffer_head *bh = NULL;
struct ocfs2_dinode *fe;
u32 free_bits;
mlog_entry_void();
alloc_inode = ocfs2_get_system_file_inode(osb, type, slot);
if (!alloc_inode) {
mlog_errno(-EINVAL);
return -EINVAL;
}
mutex_lock(&alloc_inode->i_mutex);
status = ocfs2_meta_lock(alloc_inode, &bh, 1);
if (status < 0) {
mutex_unlock(&alloc_inode->i_mutex);
iput(alloc_inode);
mlog_errno(status);
return status;
}
ac->ac_inode = alloc_inode;
fe = (struct ocfs2_dinode *) bh->b_data;
if (!OCFS2_IS_VALID_DINODE(fe)) {
OCFS2_RO_ON_INVALID_DINODE(alloc_inode->i_sb, fe);
status = -EIO;
goto bail;
}
if (!(fe->i_flags & cpu_to_le32(OCFS2_CHAIN_FL))) {
ocfs2_error(alloc_inode->i_sb, "Invalid chain allocator %llu",
(unsigned long long)le64_to_cpu(fe->i_blkno));
status = -EIO;
goto bail;
}
free_bits = le32_to_cpu(fe->id1.bitmap1.i_total) -
le32_to_cpu(fe->id1.bitmap1.i_used);
if (bits_wanted > free_bits) {
/* cluster bitmap never grows */
if (ocfs2_is_cluster_bitmap(alloc_inode)) {
mlog(0, "Disk Full: wanted=%u, free_bits=%u\n",
bits_wanted, free_bits);
status = -ENOSPC;
goto bail;
}
status = ocfs2_block_group_alloc(osb, alloc_inode, bh);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
atomic_inc(&osb->alloc_stats.bg_extends);
/* You should never ask for this much metadata */
BUG_ON(bits_wanted >
(le32_to_cpu(fe->id1.bitmap1.i_total)
- le32_to_cpu(fe->id1.bitmap1.i_used)));
}
get_bh(bh);
ac->ac_bh = bh;
bail:
if (bh)
brelse(bh);
mlog_exit(status);
return status;
}
/*
* make sure we've got at least bits_wanted contiguous bits in the
* local alloc. You lose them when you drop i_mutex.
*
* We will add ourselves to the transaction passed in, but may start
* our own in order to shift windows.
*/
int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
u32 bits_wanted,
struct ocfs2_alloc_context *ac)
{
int status;
struct ocfs2_dinode *alloc;
struct inode *local_alloc_inode;
unsigned int free_bits;
BUG_ON(!ac);
local_alloc_inode =
ocfs2_get_system_file_inode(osb,
LOCAL_ALLOC_SYSTEM_INODE,
osb->slot_num);
if (!local_alloc_inode) {
status = -ENOENT;
mlog_errno(status);
goto bail;
}
mutex_lock(&local_alloc_inode->i_mutex);
/*
* We must double check state and allocator bits because
* another process may have changed them while holding i_mutex.
*/
spin_lock(&osb->osb_lock);
if (!ocfs2_la_state_enabled(osb) ||
(bits_wanted > osb->local_alloc_bits)) {
spin_unlock(&osb->osb_lock);
status = -ENOSPC;
goto bail;
}
spin_unlock(&osb->osb_lock);
alloc = (struct ocfs2_dinode *) osb->local_alloc_bh->b_data;
#ifdef CONFIG_OCFS2_DEBUG_FS
if (le32_to_cpu(alloc->id1.bitmap1.i_used) !=
ocfs2_local_alloc_count_bits(alloc)) {
ocfs2_error(osb->sb, "local alloc inode %llu says it has "
"%u free bits, but a count shows %u",
(unsigned long long)le64_to_cpu(alloc->i_blkno),
le32_to_cpu(alloc->id1.bitmap1.i_used),
ocfs2_local_alloc_count_bits(alloc));
status = -EIO;
goto bail;
}
#endif
free_bits = le32_to_cpu(alloc->id1.bitmap1.i_total) -
le32_to_cpu(alloc->id1.bitmap1.i_used);
if (bits_wanted > free_bits) {
/* uhoh, window change time. */
status =
ocfs2_local_alloc_slide_window(osb, local_alloc_inode);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
/*
* Under certain conditions, the window slide code
* might have reduced the number of bits available or
* disabled the the local alloc entirely. Re-check
* here and return -ENOSPC if necessary.
*/
status = -ENOSPC;
if (!ocfs2_la_state_enabled(osb))
goto bail;
free_bits = le32_to_cpu(alloc->id1.bitmap1.i_total) -
le32_to_cpu(alloc->id1.bitmap1.i_used);
if (bits_wanted > free_bits)
goto bail;
}
ac->ac_inode = local_alloc_inode;
/* We should never use localalloc from another slot */
ac->ac_alloc_slot = osb->slot_num;
ac->ac_which = OCFS2_AC_USE_LOCAL;
get_bh(osb->local_alloc_bh);
ac->ac_bh = osb->local_alloc_bh;
status = 0;
bail:
if (status < 0 && local_alloc_inode) {
mutex_unlock(&local_alloc_inode->i_mutex);
iput(local_alloc_inode);
}
trace_ocfs2_reserve_local_alloc_bits(
(unsigned long long)ac->ac_max_block,
bits_wanted, osb->slot_num, status);
if (status)
mlog_errno(status);
return status;
}
static int ocfs2_relink_block_group(handle_t *handle,
struct inode *alloc_inode,
struct buffer_head *fe_bh,
struct buffer_head *bg_bh,
struct buffer_head *prev_bg_bh,
u16 chain)
{
int status;
/* there is a really tiny chance the journal calls could fail,
* but we wouldn't want inconsistent blocks in *any* case. */
u64 fe_ptr, bg_ptr, prev_bg_ptr;
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) fe_bh->b_data;
struct ocfs2_group_desc *bg = (struct ocfs2_group_desc *) bg_bh->b_data;
struct ocfs2_group_desc *prev_bg = (struct ocfs2_group_desc *) prev_bg_bh->b_data;
if (!OCFS2_IS_VALID_DINODE(fe)) {
OCFS2_RO_ON_INVALID_DINODE(alloc_inode->i_sb, fe);
status = -EIO;
goto out;
}
if (!OCFS2_IS_VALID_GROUP_DESC(bg)) {
OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, bg);
status = -EIO;
goto out;
}
if (!OCFS2_IS_VALID_GROUP_DESC(prev_bg)) {
OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, prev_bg);
status = -EIO;
goto out;
}
mlog(0, "Suballoc %llu, chain %u, move group %llu to top, prev = %llu\n",
(unsigned long long)le64_to_cpu(fe->i_blkno), chain,
(unsigned long long)le64_to_cpu(bg->bg_blkno),
(unsigned long long)le64_to_cpu(prev_bg->bg_blkno));
fe_ptr = le64_to_cpu(fe->id2.i_chain.cl_recs[chain].c_blkno);
bg_ptr = le64_to_cpu(bg->bg_next_group);
prev_bg_ptr = le64_to_cpu(prev_bg->bg_next_group);
status = ocfs2_journal_access(handle, alloc_inode, prev_bg_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
prev_bg->bg_next_group = bg->bg_next_group;
status = ocfs2_journal_dirty(handle, prev_bg_bh);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
status = ocfs2_journal_access(handle, alloc_inode, bg_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
bg->bg_next_group = fe->id2.i_chain.cl_recs[chain].c_blkno;
status = ocfs2_journal_dirty(handle, bg_bh);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
status = ocfs2_journal_access(handle, alloc_inode, fe_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
fe->id2.i_chain.cl_recs[chain].c_blkno = bg->bg_blkno;
status = ocfs2_journal_dirty(handle, fe_bh);
if (status < 0) {
mlog_errno(status);
goto out_rollback;
}
status = 0;
out_rollback:
if (status < 0) {
fe->id2.i_chain.cl_recs[chain].c_blkno = cpu_to_le64(fe_ptr);
bg->bg_next_group = cpu_to_le64(bg_ptr);
prev_bg->bg_next_group = cpu_to_le64(prev_bg_ptr);
}
out:
mlog_exit(status);
return status;
}
static void ocfs2_clear_inode(struct inode *inode)
{
int status;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
clear_inode(inode);
trace_ocfs2_clear_inode((unsigned long long)oi->ip_blkno,
inode->i_nlink);
mlog_bug_on_msg(OCFS2_SB(inode->i_sb) == NULL,
"Inode=%lu\n", inode->i_ino);
dquot_drop(inode);
/* To preven remote deletes we hold open lock before, now it
* is time to unlock PR and EX open locks. */
ocfs2_open_unlock(inode);
/* Do these before all the other work so that we don't bounce
* the downconvert thread while waiting to destroy the locks. */
ocfs2_mark_lockres_freeing(&oi->ip_rw_lockres);
ocfs2_mark_lockres_freeing(&oi->ip_inode_lockres);
ocfs2_mark_lockres_freeing(&oi->ip_open_lockres);
ocfs2_resv_discard(&OCFS2_SB(inode->i_sb)->osb_la_resmap,
&oi->ip_la_data_resv);
ocfs2_resv_init_once(&oi->ip_la_data_resv);
/* We very well may get a clear_inode before all an inodes
* metadata has hit disk. Of course, we can't drop any cluster
* locks until the journal has finished with it. The only
* exception here are successfully wiped inodes - their
* metadata can now be considered to be part of the system
* inodes from which it came. */
if (!(OCFS2_I(inode)->ip_flags & OCFS2_INODE_DELETED))
ocfs2_checkpoint_inode(inode);
mlog_bug_on_msg(!list_empty(&oi->ip_io_markers),
"Clear inode of %llu, inode has io markers\n",
(unsigned long long)oi->ip_blkno);
ocfs2_extent_map_trunc(inode, 0);
status = ocfs2_drop_inode_locks(inode);
if (status < 0)
mlog_errno(status);
ocfs2_lock_res_free(&oi->ip_rw_lockres);
ocfs2_lock_res_free(&oi->ip_inode_lockres);
ocfs2_lock_res_free(&oi->ip_open_lockres);
ocfs2_metadata_cache_exit(INODE_CACHE(inode));
mlog_bug_on_msg(INODE_CACHE(inode)->ci_num_cached,
"Clear inode of %llu, inode has %u cache items\n",
(unsigned long long)oi->ip_blkno,
INODE_CACHE(inode)->ci_num_cached);
mlog_bug_on_msg(!(INODE_CACHE(inode)->ci_flags & OCFS2_CACHE_FL_INLINE),
"Clear inode of %llu, inode has a bad flag\n",
(unsigned long long)oi->ip_blkno);
mlog_bug_on_msg(spin_is_locked(&oi->ip_lock),
"Clear inode of %llu, inode is locked\n",
(unsigned long long)oi->ip_blkno);
mlog_bug_on_msg(!mutex_trylock(&oi->ip_io_mutex),
"Clear inode of %llu, io_mutex is locked\n",
(unsigned long long)oi->ip_blkno);
mutex_unlock(&oi->ip_io_mutex);
/*
* down_trylock() returns 0, down_write_trylock() returns 1
* kernel 1, world 0
*/
mlog_bug_on_msg(!down_write_trylock(&oi->ip_alloc_sem),
"Clear inode of %llu, alloc_sem is locked\n",
(unsigned long long)oi->ip_blkno);
up_write(&oi->ip_alloc_sem);
mlog_bug_on_msg(oi->ip_open_count,
"Clear inode of %llu has open count %d\n",
(unsigned long long)oi->ip_blkno, oi->ip_open_count);
/* Clear all other flags. */
oi->ip_flags = 0;
oi->ip_dir_start_lookup = 0;
oi->ip_blkno = 0ULL;
/*
* ip_jinode is used to track txns against this inode. We ensure that
* the journal is flushed before journal shutdown. Thus it is safe to
* have inodes get cleaned up after journal shutdown.
*/
jbd2_journal_release_jbd_inode(OCFS2_SB(inode->i_sb)->journal->j_journal,
&oi->ip_jinode);
}
int ocfs2_read_blocks(struct inode *inode, u64 block, int nr,
struct buffer_head *bhs[], int flags)
{
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;
}
/* XXX: Can we ever get this and *not* have the cached
* flag set? */
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() */
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;
}
}
/* 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;
}
static int ocfs2_read_locked_inode(struct inode *inode,
struct ocfs2_find_inode_args *args)
{
struct super_block *sb;
struct ocfs2_super *osb;
struct ocfs2_dinode *fe;
struct buffer_head *bh = NULL;
int status, can_lock;
u32 generation = 0;
status = -EINVAL;
if (inode == NULL || inode->i_sb == NULL) {
mlog(ML_ERROR, "bad inode\n");
return status;
}
sb = inode->i_sb;
osb = OCFS2_SB(sb);
if (!args) {
mlog(ML_ERROR, "bad inode args\n");
make_bad_inode(inode);
return status;
}
/*
* To improve performance of cold-cache inode stats, we take
* the cluster lock here if possible.
*
* Generally, OCFS2 never trusts the contents of an inode
* unless it's holding a cluster lock, so taking it here isn't
* a correctness issue as much as it is a performance
* improvement.
*
* There are three times when taking the lock is not a good idea:
*
* 1) During startup, before we have initialized the DLM.
*
* 2) If we are reading certain system files which never get
* cluster locks (local alloc, truncate log).
*
* 3) If the process doing the iget() is responsible for
* orphan dir recovery. We're holding the orphan dir lock and
* can get into a deadlock with another process on another
* node in ->delete_inode().
*
* #1 and #2 can be simply solved by never taking the lock
* here for system files (which are the only type we read
* during mount). It's a heavier approach, but our main
* concern is user-accessible files anyway.
*
* #3 works itself out because we'll eventually take the
* cluster lock before trusting anything anyway.
*/
can_lock = !(args->fi_flags & OCFS2_FI_FLAG_SYSFILE)
&& !(args->fi_flags & OCFS2_FI_FLAG_ORPHAN_RECOVERY)
&& !ocfs2_mount_local(osb);
trace_ocfs2_read_locked_inode(
(unsigned long long)OCFS2_I(inode)->ip_blkno, can_lock);
/*
* To maintain backwards compatibility with older versions of
* ocfs2-tools, we still store the generation value for system
* files. The only ones that actually matter to userspace are
* the journals, but it's easier and inexpensive to just flag
* all system files similarly.
*/
if (args->fi_flags & OCFS2_FI_FLAG_SYSFILE)
generation = osb->fs_generation;
ocfs2_inode_lock_res_init(&OCFS2_I(inode)->ip_inode_lockres,
OCFS2_LOCK_TYPE_META,
generation, inode);
ocfs2_inode_lock_res_init(&OCFS2_I(inode)->ip_open_lockres,
OCFS2_LOCK_TYPE_OPEN,
0, inode);
if (can_lock) {
status = ocfs2_open_lock(inode);
if (status) {
make_bad_inode(inode);
mlog_errno(status);
return status;
}
status = ocfs2_inode_lock(inode, NULL, 0);
if (status) {
make_bad_inode(inode);
mlog_errno(status);
return status;
}
}
if (args->fi_flags & OCFS2_FI_FLAG_ORPHAN_RECOVERY) {
status = ocfs2_try_open_lock(inode, 0);
if (status) {
make_bad_inode(inode);
return status;
}
}
if (can_lock) {
status = ocfs2_read_inode_block_full(inode, &bh,
OCFS2_BH_IGNORE_CACHE);
} else {
status = ocfs2_read_blocks_sync(osb, args->fi_blkno, 1, &bh);
/*
* If buffer is in jbd, then its checksum may not have been
* computed as yet.
*/
if (!status && !buffer_jbd(bh))
status = ocfs2_validate_inode_block(osb->sb, bh);
}
if (status < 0) {
mlog_errno(status);
goto bail;
}
status = -EINVAL;
fe = (struct ocfs2_dinode *) bh->b_data;
/*
* This is a code bug. Right now the caller needs to
* understand whether it is asking for a system file inode or
* not so the proper lock names can be built.
*/
mlog_bug_on_msg(!!(fe->i_flags & cpu_to_le32(OCFS2_SYSTEM_FL)) !=
!!(args->fi_flags & OCFS2_FI_FLAG_SYSFILE),
"Inode %llu: system file state is ambigous\n",
(unsigned long long)args->fi_blkno);
if (S_ISCHR(le16_to_cpu(fe->i_mode)) ||
S_ISBLK(le16_to_cpu(fe->i_mode)))
inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
ocfs2_populate_inode(inode, fe, 0);
BUG_ON(args->fi_blkno != le64_to_cpu(fe->i_blkno));
status = 0;
bail:
if (can_lock)
ocfs2_inode_unlock(inode, 0);
if (status < 0)
make_bad_inode(inode);
if (args && bh)
brelse(bh);
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_ERROR,
"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];
if (buffer_jbd(bh)) {
mlog(ML_ERROR,
"the journal got the buffer while it was "
"locked for io! (blocknr = %llu)\n",
(unsigned long long)bh->b_blocknr);
BUG();
}
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 ocfs2_wipe_inode(struct inode *inode,
struct buffer_head *di_bh)
{
int status, orphaned_slot = -1;
struct inode *orphan_dir_inode = NULL;
struct buffer_head *orphan_dir_bh = NULL;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
if (!(OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
orphaned_slot = le16_to_cpu(di->i_orphaned_slot);
status = ocfs2_check_orphan_recovery_state(osb, orphaned_slot);
if (status)
return status;
orphan_dir_inode = ocfs2_get_system_file_inode(osb,
ORPHAN_DIR_SYSTEM_INODE,
orphaned_slot);
if (!orphan_dir_inode) {
status = -EEXIST;
mlog_errno(status);
goto bail;
}
/* Lock the orphan dir. The lock will be held for the entire
* delete_inode operation. We do this now to avoid races with
* recovery completion on other nodes. */
mutex_lock(&orphan_dir_inode->i_mutex);
status = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1);
if (status < 0) {
mutex_unlock(&orphan_dir_inode->i_mutex);
mlog_errno(status);
goto bail;
}
}
/* we do this while holding the orphan dir lock because we
* don't want recovery being run from another node to try an
* inode delete underneath us -- this will result in two nodes
* truncating the same file! */
status = ocfs2_truncate_for_delete(osb, inode, di_bh);
if (status < 0) {
mlog_errno(status);
goto bail_unlock_dir;
}
/* Remove any dir index tree */
if (S_ISDIR(inode->i_mode)) {
status = ocfs2_dx_dir_truncate(inode, di_bh);
if (status) {
mlog_errno(status);
goto bail_unlock_dir;
}
}
/*Free extended attribute resources associated with this inode.*/
status = ocfs2_xattr_remove(inode, di_bh);
if (status < 0) {
mlog_errno(status);
goto bail_unlock_dir;
}
status = ocfs2_remove_refcount_tree(inode, di_bh);
if (status < 0) {
mlog_errno(status);
goto bail_unlock_dir;
}
status = ocfs2_remove_inode(inode, di_bh, orphan_dir_inode,
orphan_dir_bh);
if (status < 0)
mlog_errno(status);
bail_unlock_dir:
if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)
return status;
ocfs2_inode_unlock(orphan_dir_inode, 1);
mutex_unlock(&orphan_dir_inode->i_mutex);
brelse(orphan_dir_bh);
bail:
iput(orphan_dir_inode);
ocfs2_signal_wipe_completion(osb, orphaned_slot);
return status;
}
static int __ocfs2_page_mkwrite(struct inode *inode, struct buffer_head *di_bh,
struct page *page)
{
int ret;
struct address_space *mapping = inode->i_mapping;
loff_t pos = page_offset(page);
unsigned int len = PAGE_CACHE_SIZE;
pgoff_t last_index;
struct page *locked_page = NULL;
void *fsdata;
loff_t size = i_size_read(inode);
/*
* Another node might have truncated while we were waiting on
* cluster locks.
*/
last_index = size >> PAGE_CACHE_SHIFT;
if (page->index > last_index) {
ret = -EINVAL;
goto out;
}
/*
* The i_size check above doesn't catch the case where nodes
* truncated and then re-extended the file. We'll re-check the
* page mapping after taking the page lock inside of
* ocfs2_write_begin_nolock().
*/
if (!PageUptodate(page) || page->mapping != inode->i_mapping) {
/*
* the page has been umapped in ocfs2_data_downconvert_worker.
* So return 0 here and let VFS retry.
*/
ret = 0;
goto out;
}
/*
* Call ocfs2_write_begin() and ocfs2_write_end() to take
* advantage of the allocation code there. We pass a write
* length of the whole page (chopped to i_size) to make sure
* the whole thing is allocated.
*
* Since we know the page is up to date, we don't have to
* worry about ocfs2_write_begin() skipping some buffer reads
* because the "write" would invalidate their data.
*/
if (page->index == last_index)
len = size & ~PAGE_CACHE_MASK;
ret = ocfs2_write_begin_nolock(mapping, pos, len, 0, &locked_page,
&fsdata, di_bh, page);
if (ret) {
if (ret != -ENOSPC)
mlog_errno(ret);
goto out;
}
ret = ocfs2_write_end_nolock(mapping, pos, len, len, locked_page,
fsdata);
if (ret < 0) {
mlog_errno(ret);
goto out;
}
BUG_ON(ret != len);
ret = 0;
out:
return ret;
}
static void ocfs2_delete_inode(struct inode *inode)
{
int wipe, status;
sigset_t oldset;
struct buffer_head *di_bh = NULL;
trace_ocfs2_delete_inode(inode->i_ino,
(unsigned long long)OCFS2_I(inode)->ip_blkno,
is_bad_inode(inode));
/* When we fail in read_inode() we mark inode as bad. The second test
* catches the case when inode allocation fails before allocating
* a block for inode. */
if (is_bad_inode(inode) || !OCFS2_I(inode)->ip_blkno)
goto bail;
dquot_initialize(inode);
if (!ocfs2_inode_is_valid_to_delete(inode)) {
/* It's probably not necessary to truncate_inode_pages
* here but we do it for safety anyway (it will most
* likely be a no-op anyway) */
ocfs2_cleanup_delete_inode(inode, 0);
goto bail;
}
/* We want to block signals in delete_inode as the lock and
* messaging paths may return us -ERESTARTSYS. Which would
* cause us to exit early, resulting in inodes being orphaned
* forever. */
ocfs2_block_signals(&oldset);
/*
* Synchronize us against ocfs2_get_dentry. We take this in
* shared mode so that all nodes can still concurrently
* process deletes.
*/
status = ocfs2_nfs_sync_lock(OCFS2_SB(inode->i_sb), 0);
if (status < 0) {
mlog(ML_ERROR, "getting nfs sync lock(PR) failed %d\n", status);
ocfs2_cleanup_delete_inode(inode, 0);
goto bail_unblock;
}
/* Lock down the inode. This gives us an up to date view of
* it's metadata (for verification), and allows us to
* serialize delete_inode on multiple nodes.
*
* Even though we might be doing a truncate, we don't take the
* allocation lock here as it won't be needed - nobody will
* have the file open.
*/
status = ocfs2_inode_lock(inode, &di_bh, 1);
if (status < 0) {
if (status != -ENOENT)
mlog_errno(status);
ocfs2_cleanup_delete_inode(inode, 0);
goto bail_unlock_nfs_sync;
}
/* Query the cluster. This will be the final decision made
* before we go ahead and wipe the inode. */
status = ocfs2_query_inode_wipe(inode, di_bh, &wipe);
if (!wipe || status < 0) {
/* Error and remote inode busy both mean we won't be
* removing the inode, so they take almost the same
* path. */
if (status < 0)
mlog_errno(status);
/* Someone in the cluster has disallowed a wipe of
* this inode, or it was never completely
* orphaned. Write out the pages and exit now. */
ocfs2_cleanup_delete_inode(inode, 1);
goto bail_unlock_inode;
}
ocfs2_cleanup_delete_inode(inode, 0);
status = ocfs2_wipe_inode(inode, di_bh);
if (status < 0) {
if (status != -EDEADLK)
mlog_errno(status);
goto bail_unlock_inode;
}
/*
* Mark the inode as successfully deleted.
*
* This is important for ocfs2_clear_inode() as it will check
* this flag and skip any checkpointing work
*
* ocfs2_stuff_meta_lvb() also uses this flag to invalidate
* the LVB for other nodes.
*/
OCFS2_I(inode)->ip_flags |= OCFS2_INODE_DELETED;
bail_unlock_inode:
ocfs2_inode_unlock(inode, 1);
brelse(di_bh);
bail_unlock_nfs_sync:
ocfs2_nfs_sync_unlock(OCFS2_SB(inode->i_sb), 0);
bail_unblock:
ocfs2_unblock_signals(&oldset);
bail:
return;
}
int ocfs2_load_local_alloc(struct ocfs2_super *osb)
{
int status = 0;
struct ocfs2_dinode *alloc = NULL;
struct buffer_head *alloc_bh = NULL;
u32 num_used;
struct inode *inode = NULL;
struct ocfs2_local_alloc *la;
if (osb->local_alloc_bits == 0)
goto bail;
if (osb->local_alloc_bits >= osb->bitmap_cpg) {
mlog(ML_NOTICE, "Requested local alloc window %d is larger "
"than max possible %u. Using defaults.\n",
osb->local_alloc_bits, (osb->bitmap_cpg - 1));
osb->local_alloc_bits =
ocfs2_megabytes_to_clusters(osb->sb,
ocfs2_la_default_mb(osb));
}
/* read the alloc off disk */
inode = ocfs2_get_system_file_inode(osb, LOCAL_ALLOC_SYSTEM_INODE,
osb->slot_num);
if (!inode) {
status = -EINVAL;
mlog_errno(status);
goto bail;
}
status = ocfs2_read_inode_block_full(inode, &alloc_bh,
OCFS2_BH_IGNORE_CACHE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
alloc = (struct ocfs2_dinode *) alloc_bh->b_data;
la = OCFS2_LOCAL_ALLOC(alloc);
if (!(le32_to_cpu(alloc->i_flags) &
(OCFS2_LOCAL_ALLOC_FL|OCFS2_BITMAP_FL))) {
mlog(ML_ERROR, "Invalid local alloc inode, %llu\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno);
status = -EINVAL;
goto bail;
}
if ((la->la_size == 0) ||
(le16_to_cpu(la->la_size) > ocfs2_local_alloc_size(inode->i_sb))) {
mlog(ML_ERROR, "Local alloc size is invalid (la_size = %u)\n",
le16_to_cpu(la->la_size));
status = -EINVAL;
goto bail;
}
/* do a little verification. */
num_used = ocfs2_local_alloc_count_bits(alloc);
/* hopefully the local alloc has always been recovered before
* we load it. */
if (num_used
|| alloc->id1.bitmap1.i_used
|| alloc->id1.bitmap1.i_total
|| la->la_bm_off)
mlog(ML_ERROR, "Local alloc hasn't been recovered!\n"
"found = %u, set = %u, taken = %u, off = %u\n",
num_used, le32_to_cpu(alloc->id1.bitmap1.i_used),
le32_to_cpu(alloc->id1.bitmap1.i_total),
OCFS2_LOCAL_ALLOC(alloc)->la_bm_off);
osb->local_alloc_bh = alloc_bh;
osb->local_alloc_state = OCFS2_LA_ENABLED;
bail:
if (status < 0)
brelse(alloc_bh);
if (inode)
iput(inode);
trace_ocfs2_load_local_alloc(osb->local_alloc_bits);
if (status)
mlog_errno(status);
return status;
}
/*
* Extend the filesystem to the new number of clusters specified. This entry
* point is only used to extend the current filesystem to the end of the last
* existing group.
*/
int ocfs2_group_extend(struct inode * inode, int new_clusters)
{
int ret;
handle_t *handle;
struct buffer_head *main_bm_bh = NULL;
struct buffer_head *group_bh = NULL;
struct inode *main_bm_inode = NULL;
struct ocfs2_dinode *fe = NULL;
struct ocfs2_group_desc *group = NULL;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
u16 cl_bpc;
u32 first_new_cluster;
u64 lgd_blkno;
mlog_entry_void();
if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
return -EROFS;
if (new_clusters < 0)
return -EINVAL;
else if (new_clusters == 0)
return 0;
main_bm_inode = ocfs2_get_system_file_inode(osb,
GLOBAL_BITMAP_SYSTEM_INODE,
OCFS2_INVALID_SLOT);
if (!main_bm_inode) {
ret = -EINVAL;
mlog_errno(ret);
goto out;
}
mutex_lock(&main_bm_inode->i_mutex);
ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
if (ret < 0) {
mlog_errno(ret);
goto out_mutex;
}
fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
/* main_bm_bh is validated by inode read inside ocfs2_inode_lock(),
* so any corruption is a code bug. */
BUG_ON(!OCFS2_IS_VALID_DINODE(fe));
if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
ocfs2_group_bitmap_size(osb->sb) * 8) {
mlog(ML_ERROR, "The disk is too old and small. "
"Force to do offline resize.");
ret = -EINVAL;
goto out_unlock;
}
first_new_cluster = le32_to_cpu(fe->i_clusters);
lgd_blkno = ocfs2_which_cluster_group(main_bm_inode,
first_new_cluster - 1);
ret = ocfs2_read_group_descriptor(main_bm_inode, fe, lgd_blkno,
&group_bh);
if (ret < 0) {
mlog_errno(ret);
goto out_unlock;
}
group = (struct ocfs2_group_desc *)group_bh->b_data;
cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
if (le16_to_cpu(group->bg_bits) / cl_bpc + new_clusters >
le16_to_cpu(fe->id2.i_chain.cl_cpg)) {
ret = -EINVAL;
goto out_unlock;
}
mlog(0, "extend the last group at %llu, new clusters = %d\n",
(unsigned long long)le64_to_cpu(group->bg_blkno), new_clusters);
handle = ocfs2_start_trans(osb, OCFS2_GROUP_EXTEND_CREDITS);
if (IS_ERR(handle)) {
mlog_errno(PTR_ERR(handle));
ret = -EINVAL;
goto out_unlock;
}
/* update the last group descriptor and inode. */
ret = ocfs2_update_last_group_and_inode(handle, main_bm_inode,
main_bm_bh, group_bh,
first_new_cluster,
new_clusters);
if (ret) {
mlog_errno(ret);
goto out_commit;
}
ocfs2_update_super_and_backups(main_bm_inode, new_clusters);
out_commit:
ocfs2_commit_trans(osb, handle);
out_unlock:
brelse(group_bh);
brelse(main_bm_bh);
ocfs2_inode_unlock(main_bm_inode, 1);
out_mutex:
mutex_unlock(&main_bm_inode->i_mutex);
iput(main_bm_inode);
out:
mlog_exit_void();
return ret;
}
/*
* return any unused bits to the bitmap and write out a clean
* local_alloc.
*
* local_alloc_bh is optional. If not passed, we will simply use the
* one off osb. If you do pass it however, be warned that it *will* be
* returned brelse'd and NULL'd out.*/
void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
{
int status;
handle_t *handle;
struct inode *local_alloc_inode = NULL;
struct buffer_head *bh = NULL;
struct buffer_head *main_bm_bh = NULL;
struct inode *main_bm_inode = NULL;
struct ocfs2_dinode *alloc_copy = NULL;
struct ocfs2_dinode *alloc = NULL;
cancel_delayed_work(&osb->la_enable_wq);
flush_workqueue(ocfs2_wq);
if (osb->local_alloc_state == OCFS2_LA_UNUSED)
goto out;
local_alloc_inode =
ocfs2_get_system_file_inode(osb,
LOCAL_ALLOC_SYSTEM_INODE,
osb->slot_num);
if (!local_alloc_inode) {
status = -ENOENT;
mlog_errno(status);
goto out;
}
osb->local_alloc_state = OCFS2_LA_DISABLED;
ocfs2_resmap_uninit(&osb->osb_la_resmap);
main_bm_inode = ocfs2_get_system_file_inode(osb,
GLOBAL_BITMAP_SYSTEM_INODE,
OCFS2_INVALID_SLOT);
if (!main_bm_inode) {
status = -EINVAL;
mlog_errno(status);
goto out;
}
mutex_lock(&main_bm_inode->i_mutex);
status = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
if (status < 0) {
mlog_errno(status);
goto out_mutex;
}
/* WINDOW_MOVE_CREDITS is a bit heavy... */
handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS);
if (IS_ERR(handle)) {
mlog_errno(PTR_ERR(handle));
handle = NULL;
goto out_unlock;
}
bh = osb->local_alloc_bh;
alloc = (struct ocfs2_dinode *) bh->b_data;
alloc_copy = kmalloc(bh->b_size, GFP_NOFS);
if (!alloc_copy) {
status = -ENOMEM;
goto out_commit;
}
memcpy(alloc_copy, alloc, bh->b_size);
status = ocfs2_journal_access_di(handle, INODE_CACHE(local_alloc_inode),
bh, OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto out_commit;
}
ocfs2_clear_local_alloc(alloc);
ocfs2_journal_dirty(handle, bh);
brelse(bh);
osb->local_alloc_bh = NULL;
osb->local_alloc_state = OCFS2_LA_UNUSED;
status = ocfs2_sync_local_to_main(osb, handle, alloc_copy,
main_bm_inode, main_bm_bh);
if (status < 0)
mlog_errno(status);
out_commit:
ocfs2_commit_trans(osb, handle);
out_unlock:
brelse(main_bm_bh);
ocfs2_inode_unlock(main_bm_inode, 1);
out_mutex:
mutex_unlock(&main_bm_inode->i_mutex);
iput(main_bm_inode);
out:
if (local_alloc_inode)
iput(local_alloc_inode);
kfree(alloc_copy);
}
static int ocfs2_update_last_group_and_inode(handle_t *handle,
struct inode *bm_inode,
struct buffer_head *bm_bh,
struct buffer_head *group_bh,
u32 first_new_cluster,
int new_clusters)
{
int ret = 0;
struct ocfs2_super *osb = OCFS2_SB(bm_inode->i_sb);
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bm_bh->b_data;
struct ocfs2_chain_list *cl = &fe->id2.i_chain;
struct ocfs2_chain_rec *cr;
struct ocfs2_group_desc *group;
u16 chain, num_bits, backups = 0;
u16 cl_bpc = le16_to_cpu(cl->cl_bpc);
u16 cl_cpg = le16_to_cpu(cl->cl_cpg);
mlog_entry("(new_clusters=%d, first_new_cluster = %u)\n",
new_clusters, first_new_cluster);
ret = ocfs2_journal_access_gd(handle, bm_inode, group_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (ret < 0) {
mlog_errno(ret);
goto out;
}
group = (struct ocfs2_group_desc *)group_bh->b_data;
/* update the group first. */
num_bits = new_clusters * cl_bpc;
le16_add_cpu(&group->bg_bits, num_bits);
le16_add_cpu(&group->bg_free_bits_count, num_bits);
/*
* check whether there are some new backup superblocks exist in
* this group and update the group bitmap accordingly.
*/
if (OCFS2_HAS_COMPAT_FEATURE(osb->sb,
OCFS2_FEATURE_COMPAT_BACKUP_SB)) {
backups = ocfs2_calc_new_backup_super(bm_inode,
group,
new_clusters,
first_new_cluster,
cl_cpg, 1);
le16_add_cpu(&group->bg_free_bits_count, -1 * backups);
}
ret = ocfs2_journal_dirty(handle, group_bh);
if (ret < 0) {
mlog_errno(ret);
goto out_rollback;
}
/* update the inode accordingly. */
ret = ocfs2_journal_access_di(handle, bm_inode, bm_bh,
OCFS2_JOURNAL_ACCESS_WRITE);
if (ret < 0) {
mlog_errno(ret);
goto out_rollback;
}
chain = le16_to_cpu(group->bg_chain);
cr = (&cl->cl_recs[chain]);
le32_add_cpu(&cr->c_total, num_bits);
le32_add_cpu(&cr->c_free, num_bits);
le32_add_cpu(&fe->id1.bitmap1.i_total, num_bits);
le32_add_cpu(&fe->i_clusters, new_clusters);
if (backups) {
le32_add_cpu(&cr->c_free, -1 * backups);
le32_add_cpu(&fe->id1.bitmap1.i_used, backups);
}
spin_lock(&OCFS2_I(bm_inode)->ip_lock);
OCFS2_I(bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
le64_add_cpu(&fe->i_size, new_clusters << osb->s_clustersize_bits);
spin_unlock(&OCFS2_I(bm_inode)->ip_lock);
i_size_write(bm_inode, le64_to_cpu(fe->i_size));
ocfs2_journal_dirty(handle, bm_bh);
out_rollback:
if (ret < 0) {
ocfs2_calc_new_backup_super(bm_inode,
group,
new_clusters,
first_new_cluster,
cl_cpg, 0);
le16_add_cpu(&group->bg_free_bits_count, backups);
le16_add_cpu(&group->bg_bits, -1 * num_bits);
le16_add_cpu(&group->bg_free_bits_count, -1 * num_bits);
}
out:
mlog_exit(ret);
return ret;
}
/*
* We expect the block group allocator to already be locked.
*/
static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
struct inode *alloc_inode,
struct buffer_head *bh,
u64 max_block,
u64 *last_alloc_group,
int flags)
{
int status, credits;
struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bh->b_data;
struct ocfs2_chain_list *cl;
struct ocfs2_alloc_context *ac = NULL;
handle_t *handle = NULL;
u16 alloc_rec;
struct buffer_head *bg_bh = NULL;
struct ocfs2_group_desc *bg;
BUG_ON(ocfs2_is_cluster_bitmap(alloc_inode));
cl = &fe->id2.i_chain;
status = ocfs2_reserve_clusters_with_limit(osb,
le16_to_cpu(cl->cl_cpg),
max_block, flags, &ac);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
credits = ocfs2_calc_group_alloc_credits(osb->sb,
le16_to_cpu(cl->cl_cpg));
handle = ocfs2_start_trans(osb, credits);
if (IS_ERR(handle)) {
status = PTR_ERR(handle);
handle = NULL;
mlog_errno(status);
goto bail;
}
if (last_alloc_group && *last_alloc_group != 0) {
trace_ocfs2_block_group_alloc(
(unsigned long long)*last_alloc_group);
ac->ac_last_group = *last_alloc_group;
}
bg_bh = ocfs2_block_group_alloc_contig(osb, handle, alloc_inode,
ac, cl);
if (IS_ERR(bg_bh) && (PTR_ERR(bg_bh) == -ENOSPC))
bg_bh = ocfs2_block_group_alloc_discontig(handle,
alloc_inode,
ac, cl);
if (IS_ERR(bg_bh)) {
status = PTR_ERR(bg_bh);
bg_bh = NULL;
if (status != -ENOSPC)
mlog_errno(status);
goto bail;
}
bg = (struct ocfs2_group_desc *) bg_bh->b_data;
status = ocfs2_journal_access_di(handle, INODE_CACHE(alloc_inode),
bh, OCFS2_JOURNAL_ACCESS_WRITE);
if (status < 0) {
mlog_errno(status);
goto bail;
}
alloc_rec = le16_to_cpu(bg->bg_chain);
le32_add_cpu(&cl->cl_recs[alloc_rec].c_free,
le16_to_cpu(bg->bg_free_bits_count));
le32_add_cpu(&cl->cl_recs[alloc_rec].c_total,
le16_to_cpu(bg->bg_bits));
cl->cl_recs[alloc_rec].c_blkno = bg->bg_blkno;
if (le16_to_cpu(cl->cl_next_free_rec) < le16_to_cpu(cl->cl_count))
le16_add_cpu(&cl->cl_next_free_rec, 1);
le32_add_cpu(&fe->id1.bitmap1.i_used, le16_to_cpu(bg->bg_bits) -
le16_to_cpu(bg->bg_free_bits_count));
le32_add_cpu(&fe->id1.bitmap1.i_total, le16_to_cpu(bg->bg_bits));
le32_add_cpu(&fe->i_clusters, le16_to_cpu(cl->cl_cpg));
ocfs2_journal_dirty(handle, bh);
spin_lock(&OCFS2_I(alloc_inode)->ip_lock);
OCFS2_I(alloc_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
fe->i_size = cpu_to_le64(ocfs2_clusters_to_bytes(alloc_inode->i_sb,
le32_to_cpu(fe->i_clusters)));
spin_unlock(&OCFS2_I(alloc_inode)->ip_lock);
i_size_write(alloc_inode, le64_to_cpu(fe->i_size));
alloc_inode->i_blocks = ocfs2_inode_sector_count(alloc_inode);
ocfs2_update_inode_fsync_trans(handle, alloc_inode, 0);
status = 0;
/* save the new last alloc group so that the caller can cache it. */
if (last_alloc_group)
*last_alloc_group = ac->ac_last_group;
bail:
if (handle)
ocfs2_commit_trans(osb, handle);
if (ac)
ocfs2_free_alloc_context(ac);
brelse(bg_bh);
if (status)
mlog_errno(status);
return status;
}
