suballoc.c

linux 内核源代码

	int status;
	u64 bg_blkno;

	BUG_ON(!ac);
	BUG_ON(ac->ac_bits_wanted < (ac->ac_bits_given + bits_wanted));
	BUG_ON(ac->ac_which != OCFS2_AC_USE_META);

	status = ocfs2_claim_suballoc_bits(osb,
					   ac,
					   handle,
					   bits_wanted,
					   1,
					   suballoc_bit_start,
					   num_bits,
					   &bg_blkno);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	atomic_inc(&osb->alloc_stats.bg_allocs);

	*blkno_start = bg_blkno + (u64) *suballoc_bit_start;
	ac->ac_bits_given += (*num_bits);
	status = 0;
bail:
	mlog_exit(status);
	return status;
}

int ocfs2_claim_new_inode(struct ocfs2_super *osb,
			  handle_t *handle,
			  struct ocfs2_alloc_context *ac,
			  u16 *suballoc_bit,
			  u64 *fe_blkno)
{
	int status;
	unsigned int num_bits;
	u64 bg_blkno;

	mlog_entry_void();

	BUG_ON(!ac);
	BUG_ON(ac->ac_bits_given != 0);
	BUG_ON(ac->ac_bits_wanted != 1);
	BUG_ON(ac->ac_which != OCFS2_AC_USE_INODE);

	status = ocfs2_claim_suballoc_bits(osb,
					   ac,
					   handle,
					   1,
					   1,
					   suballoc_bit,
					   &num_bits,
					   &bg_blkno);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	atomic_inc(&osb->alloc_stats.bg_allocs);

	BUG_ON(num_bits != 1);

	*fe_blkno = bg_blkno + (u64) (*suballoc_bit);
	ac->ac_bits_given++;
	status = 0;
bail:
	mlog_exit(status);
	return status;
}

/* translate a group desc. blkno and it's bitmap offset into
 * disk cluster offset. */
static inline u32 ocfs2_desc_bitmap_to_cluster_off(struct inode *inode,
						   u64 bg_blkno,
						   u16 bg_bit_off)
{
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	u32 cluster = 0;

	BUG_ON(!ocfs2_is_cluster_bitmap(inode));

	if (bg_blkno != osb->first_cluster_group_blkno)
		cluster = ocfs2_blocks_to_clusters(inode->i_sb, bg_blkno);
	cluster += (u32) bg_bit_off;
	return cluster;
}

/* given a cluster offset, calculate which block group it belongs to
 * and return that block offset. */
static inline u64 ocfs2_which_cluster_group(struct inode *inode,
					    u32 cluster)
{
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	u32 group_no;

	BUG_ON(!ocfs2_is_cluster_bitmap(inode));

	group_no = cluster / osb->bitmap_cpg;
	if (!group_no)
		return osb->first_cluster_group_blkno;
	return ocfs2_clusters_to_blocks(inode->i_sb,
					group_no * osb->bitmap_cpg);
}

/* given the block number of a cluster start, calculate which cluster
 * group and descriptor bitmap offset that corresponds to. */
static inline void ocfs2_block_to_cluster_group(struct inode *inode,
						u64 data_blkno,
						u64 *bg_blkno,
						u16 *bg_bit_off)
{
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	u32 data_cluster = ocfs2_blocks_to_clusters(osb->sb, data_blkno);

	BUG_ON(!ocfs2_is_cluster_bitmap(inode));

	*bg_blkno = ocfs2_which_cluster_group(inode,
					      data_cluster);

	if (*bg_blkno == osb->first_cluster_group_blkno)
		*bg_bit_off = (u16) data_cluster;
	else
		*bg_bit_off = (u16) ocfs2_blocks_to_clusters(osb->sb,
							     data_blkno - *bg_blkno);
}

/*
 * 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 max_clusters,
			   u32 *cluster_start,
			   u32 *num_clusters)
{
	int status;
	unsigned int bits_wanted = max_clusters;
	u64 bg_blkno = 0;
	u16 bg_bit_off;

	mlog_entry_void();

	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;
}

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)
{
	unsigned int bits_wanted = ac->ac_bits_wanted - ac->ac_bits_given;

	return __ocfs2_claim_clusters(osb, handle, ac, min_clusters,
				      bits_wanted, cluster_start, num_clusters);
}

static inline int ocfs2_block_group_clear_bits(handle_t *handle,
					       struct inode *alloc_inode,
					       struct ocfs2_group_desc *bg,
					       struct buffer_head *group_bh,
					       unsigned int bit_off,
					       unsigned int num_bits)
{
	int status;
	unsigned int tmp;
	int journal_type = OCFS2_JOURNAL_ACCESS_WRITE;
	struct ocfs2_group_desc *undo_bg = NULL;

	mlog_entry_void();

	if (!OCFS2_IS_VALID_GROUP_DESC(bg)) {
		OCFS2_RO_ON_INVALID_GROUP_DESC(alloc_inode->i_sb, bg);
		status = -EIO;
		goto bail;
	}

	mlog(0, "off = %u, num = %u\n", bit_off, num_bits);

	if (ocfs2_is_cluster_bitmap(alloc_inode))
		journal_type = OCFS2_JOURNAL_ACCESS_UNDO;

	status = ocfs2_journal_access(handle, alloc_inode, group_bh,
				      journal_type);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	if (ocfs2_is_cluster_bitmap(alloc_inode))
		undo_bg = (struct ocfs2_group_desc *) bh2jh(group_bh)->b_committed_data;

	tmp = num_bits;
	while(tmp--) {
		ocfs2_clear_bit((bit_off + tmp),
				(unsigned long *) bg->bg_bitmap);
		if (ocfs2_is_cluster_bitmap(alloc_inode))
			ocfs2_set_bit(bit_off + tmp,
				      (unsigned long *) undo_bg->bg_bitmap);
	}
	le16_add_cpu(&bg->bg_free_bits_count, num_bits);

	status = ocfs2_journal_dirty(handle, group_bh);
	if (status < 0)
		mlog_errno(status);
bail:
	return status;
}

/*
 * expects the suballoc inode to already be locked.
 */
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;
}

int ocfs2_free_dinode(handle_t *handle,
		      struct inode *inode_alloc_inode,
		      struct buffer_head *inode_alloc_bh,
		      struct ocfs2_dinode *di)
{
	u64 blk = le64_to_cpu(di->i_blkno);
	u16 bit = le16_to_cpu(di->i_suballoc_bit);
	u64 bg_blkno = ocfs2_which_suballoc_group(blk, bit);

	return ocfs2_free_suballoc_bits(handle, inode_alloc_inode,
					inode_alloc_bh, bit, bg_blkno, 1);
}

int ocfs2_free_clusters(handle_t *handle,
		       struct inode *bitmap_inode,
		       struct buffer_head *bitmap_bh,
		       u64 start_blk,
		       unsigned int num_clusters)
{
	int status;
	u16 bg_start_bit;
	u64 bg_blkno;
	struct ocfs2_dinode *fe;

	/* You can't ever have a contiguous set of clusters
	 * bigger than a block group bitmap so we never have to worry
	 * about looping on them. */

	mlog_entry_void();

	/* This is expensive. We can safely remove once this stuff has
	 * gotten tested really well. */
	BUG_ON(start_blk != ocfs2_clusters_to_blocks(bitmap_inode->i_sb, ocfs2_blocks_to_clusters(bitmap_inode->i_sb, start_blk)));

	fe = (struct ocfs2_dinode *) bitmap_bh->b_data;

	ocfs2_block_to_cluster_group(bitmap_inode, start_blk, &bg_blkno,
				     &bg_start_bit);

	mlog(0, "want to free %u clusters starting at block %llu\n",
	     num_clusters, (unsigned long long)start_blk);
	mlog(0, "bg_blkno = %llu, bg_start_bit = %u\n",
	     (unsigned long long)bg_blkno, bg_start_bit);

	status = ocfs2_free_suballoc_bits(handle, bitmap_inode, bitmap_bh,
					  bg_start_bit, bg_blkno,
					  num_clusters);
	if (status < 0)
		mlog_errno(status);

	mlog_exit(status);
	return status;
}

static inline void ocfs2_debug_bg(struct ocfs2_group_desc *bg)
{
	printk("Block Group:\n");
	printk("bg_signature:       %s\n", bg->bg_signature);
	printk("bg_size:            %u\n", bg->bg_size);
	printk("bg_bits:            %u\n", bg->bg_bits);
	printk("bg_free_bits_count: %u\n", bg->bg_free_bits_count);
	printk("bg_chain:           %u\n", bg->bg_chain);
	printk("bg_generation:      %u\n", le32_to_cpu(bg->bg_generation));
	printk("bg_next_group:      %llu\n",
	       (unsigned long long)bg->bg_next_group);
	printk("bg_parent_dinode:   %llu\n",
	       (unsigned long long)bg->bg_parent_dinode);
	printk("bg_blkno:           %llu\n",
	       (unsigned long long)bg->bg_blkno);
}

static inline void ocfs2_debug_suballoc_inode(struct ocfs2_dinode *fe)
{
	int i;

	printk("Suballoc Inode %llu:\n", (unsigned long long)fe->i_blkno);
	printk("i_signature:                  %s\n", fe->i_signature);
	printk("i_size:                       %llu\n",
	       (unsigned long long)fe->i_size);
	printk("i_clusters:                   %u\n", fe->i_clusters);
	printk("i_generation:                 %u\n",
	       le32_to_cpu(fe->i_generation));
	printk("id1.bitmap1.i_used:           %u\n",
	       le32_to_cpu(fe->id1.bitmap1.i_used));
	printk("id1.bitmap1.i_total:          %u\n",
	       le32_to_cpu(fe->id1.bitmap1.i_total));
	printk("id2.i_chain.cl_cpg:           %u\n", fe->id2.i_chain.cl_cpg);
	printk("id2.i_chain.cl_bpc:           %u\n", fe->id2.i_chain.cl_bpc);
	printk("id2.i_chain.cl_count:         %u\n", fe->id2.i_chain.cl_count);
	printk("id2.i_chain.cl_next_free_rec: %u\n",
	       fe->id2.i_chain.cl_next_free_rec);
	for(i = 0; i < fe->id2.i_chain.cl_next_free_rec; i++) {
		printk("fe->id2.i_chain.cl_recs[%d].c_free:  %u\n", i,
		       fe->id2.i_chain.cl_recs[i].c_free);
		printk("fe->id2.i_chain.cl_recs[%d].c_total: %u\n", i,
		       fe->id2.i_chain.cl_recs[i].c_total);
		printk("fe->id2.i_chain.cl_recs[%d].c_blkno: %llu\n", i,
		       (unsigned long long)fe->id2.i_chain.cl_recs[i].c_blkno);
	}
}